Index: head/sys/cddl/contrib/opensolaris/uts/common/fs/zfs/dmu_object.c =================================================================== --- head/sys/cddl/contrib/opensolaris/uts/common/fs/zfs/dmu_object.c (revision 353555) +++ head/sys/cddl/contrib/opensolaris/uts/common/fs/zfs/dmu_object.c (revision 353556) @@ -1,444 +1,444 @@ /* * CDDL HEADER START * * The contents of this file are subject to the terms of the * Common Development and Distribution License (the "License"). * You may not use this file except in compliance with the License. * * You can obtain a copy of the license at usr/src/OPENSOLARIS.LICENSE * or http://www.opensolaris.org/os/licensing. * See the License for the specific language governing permissions * and limitations under the License. * * When distributing Covered Code, include this CDDL HEADER in each * file and include the License file at usr/src/OPENSOLARIS.LICENSE. * If applicable, add the following below this CDDL HEADER, with the * fields enclosed by brackets "[]" replaced with your own identifying * information: Portions Copyright [yyyy] [name of copyright owner] * * CDDL HEADER END */ /* * Copyright (c) 2005, 2010, Oracle and/or its affiliates. All rights reserved. * Copyright (c) 2013, 2017 by Delphix. All rights reserved. * Copyright 2014 HybridCluster. All rights reserved. */ #include #include #include #include #include #include #include /* * Each of the concurrent object allocators will grab * 2^dmu_object_alloc_chunk_shift dnode slots at a time. The default is to * grab 128 slots, which is 4 blocks worth. This was experimentally * determined to be the lowest value that eliminates the measurable effect * of lock contention from this code path. */ int dmu_object_alloc_chunk_shift = 7; static uint64_t dmu_object_alloc_impl(objset_t *os, dmu_object_type_t ot, int blocksize, int indirect_blockshift, dmu_object_type_t bonustype, int bonuslen, int dnodesize, dmu_tx_t *tx) { uint64_t object; uint64_t L1_dnode_count = DNODES_PER_BLOCK << (DMU_META_DNODE(os)->dn_indblkshift - SPA_BLKPTRSHIFT); dnode_t *dn = NULL; int dn_slots = dnodesize >> DNODE_SHIFT; boolean_t restarted = B_FALSE; uint64_t *cpuobj = &os->os_obj_next_percpu[CPU_SEQID % os->os_obj_next_percpu_len]; int dnodes_per_chunk = 1 << dmu_object_alloc_chunk_shift; int error; if (dn_slots == 0) { dn_slots = DNODE_MIN_SLOTS; } else { ASSERT3S(dn_slots, >=, DNODE_MIN_SLOTS); ASSERT3S(dn_slots, <=, DNODE_MAX_SLOTS); } /* * The "chunk" of dnodes that is assigned to a CPU-specific * allocator needs to be at least one block's worth, to avoid * lock contention on the dbuf. It can be at most one L1 block's * worth, so that the "rescan after polishing off a L1's worth" * logic below will be sure to kick in. */ if (dnodes_per_chunk < DNODES_PER_BLOCK) dnodes_per_chunk = DNODES_PER_BLOCK; if (dnodes_per_chunk > L1_dnode_count) dnodes_per_chunk = L1_dnode_count; #ifdef __FreeBSD__ object = atomic_load_64(cpuobj); #else object = *cpuobj; #endif for (;;) { /* * If we finished a chunk of dnodes, get a new one from * the global allocator. */ if ((P2PHASE(object, dnodes_per_chunk) == 0) || (P2PHASE(object + dn_slots - 1, dnodes_per_chunk) < dn_slots)) { DNODE_STAT_BUMP(dnode_alloc_next_chunk); mutex_enter(&os->os_obj_lock); ASSERT0(P2PHASE(os->os_obj_next_chunk, dnodes_per_chunk)); object = os->os_obj_next_chunk; /* * Each time we polish off a L1 bp worth of dnodes * (2^12 objects), move to another L1 bp that's * still reasonably sparse (at most 1/4 full). Look * from the beginning at most once per txg. If we * still can't allocate from that L1 block, search * for an empty L0 block, which will quickly skip * to the end of the metadnode if the no nearby L0 * blocks are empty. This fallback avoids a * pathology where full dnode blocks containing * large dnodes appear sparse because they have a * low blk_fill, leading to many failed allocation * attempts. In the long term a better mechanism to * search for sparse metadnode regions, such as * spacemaps, could be implemented. * * os_scan_dnodes is set during txg sync if enough * objects have been freed since the previous * rescan to justify backfilling again. * * Note that dmu_traverse depends on the behavior * that we use multiple blocks of the dnode object * before going back to reuse objects. Any change * to this algorithm should preserve that property * or find another solution to the issues described * in traverse_visitbp. */ if (P2PHASE(object, L1_dnode_count) == 0) { uint64_t offset; uint64_t blkfill; int minlvl; if (os->os_rescan_dnodes) { offset = 0; os->os_rescan_dnodes = B_FALSE; } else { offset = object << DNODE_SHIFT; } blkfill = restarted ? 1 : DNODES_PER_BLOCK >> 2; minlvl = restarted ? 1 : 2; restarted = B_TRUE; error = dnode_next_offset(DMU_META_DNODE(os), DNODE_FIND_HOLE, &offset, minlvl, blkfill, 0); if (error == 0) { object = offset >> DNODE_SHIFT; } } /* * Note: if "restarted", we may find a L0 that * is not suitably aligned. */ os->os_obj_next_chunk = P2ALIGN(object, dnodes_per_chunk) + dnodes_per_chunk; (void) atomic_swap_64(cpuobj, object); mutex_exit(&os->os_obj_lock); } /* * The value of (*cpuobj) before adding dn_slots is the object * ID assigned to us. The value afterwards is the object ID * assigned to whoever wants to do an allocation next. */ object = atomic_add_64_nv(cpuobj, dn_slots) - dn_slots; /* * XXX We should check for an i/o error here and return * up to our caller. Actually we should pre-read it in * dmu_tx_assign(), but there is currently no mechanism * to do so. */ error = dnode_hold_impl(os, object, DNODE_MUST_BE_FREE, dn_slots, FTAG, &dn); if (error == 0) { rw_enter(&dn->dn_struct_rwlock, RW_WRITER); /* * Another thread could have allocated it; check * again now that we have the struct lock. */ if (dn->dn_type == DMU_OT_NONE) { dnode_allocate(dn, ot, blocksize, 0, bonustype, bonuslen, dn_slots, tx); rw_exit(&dn->dn_struct_rwlock); dmu_tx_add_new_object(tx, dn); dnode_rele(dn, FTAG); return (object); } rw_exit(&dn->dn_struct_rwlock); dnode_rele(dn, FTAG); DNODE_STAT_BUMP(dnode_alloc_race); } /* * Skip to next known valid starting point on error. This * is the start of the next block of dnodes. */ if (dmu_object_next(os, &object, B_TRUE, 0) != 0) { object = P2ROUNDUP(object + 1, DNODES_PER_BLOCK); DNODE_STAT_BUMP(dnode_alloc_next_block); } (void) atomic_swap_64(cpuobj, object); } } uint64_t dmu_object_alloc(objset_t *os, dmu_object_type_t ot, int blocksize, dmu_object_type_t bonustype, int bonuslen, dmu_tx_t *tx) { return (dmu_object_alloc_impl(os, ot, blocksize, 0, bonustype, bonuslen, 0, tx)); } uint64_t dmu_object_alloc_ibs(objset_t *os, dmu_object_type_t ot, int blocksize, int indirect_blockshift, dmu_object_type_t bonustype, int bonuslen, dmu_tx_t *tx) { return (dmu_object_alloc_impl(os, ot, blocksize, indirect_blockshift, bonustype, bonuslen, 0, tx)); } uint64_t dmu_object_alloc_dnsize(objset_t *os, dmu_object_type_t ot, int blocksize, dmu_object_type_t bonustype, int bonuslen, int dnodesize, dmu_tx_t *tx) { return (dmu_object_alloc_impl(os, ot, blocksize, 0, bonustype, bonuslen, dnodesize, tx)); } int dmu_object_claim(objset_t *os, uint64_t object, dmu_object_type_t ot, int blocksize, dmu_object_type_t bonustype, int bonuslen, dmu_tx_t *tx) { return (dmu_object_claim_dnsize(os, object, ot, blocksize, bonustype, bonuslen, 0, tx)); } int dmu_object_claim_dnsize(objset_t *os, uint64_t object, dmu_object_type_t ot, int blocksize, dmu_object_type_t bonustype, int bonuslen, int dnodesize, dmu_tx_t *tx) { dnode_t *dn; int dn_slots = dnodesize >> DNODE_SHIFT; int err; if (dn_slots == 0) dn_slots = DNODE_MIN_SLOTS; ASSERT3S(dn_slots, >=, DNODE_MIN_SLOTS); ASSERT3S(dn_slots, <=, DNODE_MAX_SLOTS); if (object == DMU_META_DNODE_OBJECT && !dmu_tx_private_ok(tx)) return (SET_ERROR(EBADF)); err = dnode_hold_impl(os, object, DNODE_MUST_BE_FREE, dn_slots, FTAG, &dn); if (err) return (err); dnode_allocate(dn, ot, blocksize, 0, bonustype, bonuslen, dn_slots, tx); dmu_tx_add_new_object(tx, dn); dnode_rele(dn, FTAG); return (0); } int dmu_object_reclaim(objset_t *os, uint64_t object, dmu_object_type_t ot, int blocksize, dmu_object_type_t bonustype, int bonuslen, dmu_tx_t *tx) { return (dmu_object_reclaim_dnsize(os, object, ot, blocksize, bonustype, - bonuslen, 0, tx)); + bonuslen, DNODE_MIN_SIZE, tx)); } int dmu_object_reclaim_dnsize(objset_t *os, uint64_t object, dmu_object_type_t ot, int blocksize, dmu_object_type_t bonustype, int bonuslen, int dnodesize, dmu_tx_t *tx) { dnode_t *dn; int dn_slots = dnodesize >> DNODE_SHIFT; int err; if (dn_slots == 0) dn_slots = DNODE_MIN_SLOTS; if (object == DMU_META_DNODE_OBJECT) return (SET_ERROR(EBADF)); err = dnode_hold_impl(os, object, DNODE_MUST_BE_ALLOCATED, 0, FTAG, &dn); if (err) return (err); dnode_reallocate(dn, ot, blocksize, bonustype, bonuslen, dn_slots, tx); dnode_rele(dn, FTAG); return (err); } int dmu_object_free(objset_t *os, uint64_t object, dmu_tx_t *tx) { dnode_t *dn; int err; ASSERT(object != DMU_META_DNODE_OBJECT || dmu_tx_private_ok(tx)); err = dnode_hold_impl(os, object, DNODE_MUST_BE_ALLOCATED, 0, FTAG, &dn); if (err) return (err); ASSERT(dn->dn_type != DMU_OT_NONE); /* * If we don't create this free range, we'll leak indirect blocks when * we get to freeing the dnode in syncing context. */ dnode_free_range(dn, 0, DMU_OBJECT_END, tx); dnode_free(dn, tx); dnode_rele(dn, FTAG); return (0); } /* * Return (in *objectp) the next object which is allocated (or a hole) * after *object, taking into account only objects that may have been modified * after the specified txg. */ int dmu_object_next(objset_t *os, uint64_t *objectp, boolean_t hole, uint64_t txg) { uint64_t offset; uint64_t start_obj; struct dsl_dataset *ds = os->os_dsl_dataset; int error; if (*objectp == 0) { start_obj = 1; } else if (ds && ds->ds_feature_inuse[SPA_FEATURE_LARGE_DNODE]) { uint64_t i = *objectp + 1; uint64_t last_obj = *objectp | (DNODES_PER_BLOCK - 1); dmu_object_info_t doi; /* * Scan through the remaining meta dnode block. The contents * of each slot in the block are known so it can be quickly * checked. If the block is exhausted without a match then * hand off to dnode_next_offset() for further scanning. */ while (i <= last_obj) { error = dmu_object_info(os, i, &doi); if (error == ENOENT) { if (hole) { *objectp = i; return (0); } else { i++; } } else if (error == EEXIST) { i++; } else if (error == 0) { if (hole) { i += doi.doi_dnodesize >> DNODE_SHIFT; } else { *objectp = i; return (0); } } else { return (error); } } start_obj = i; } else { start_obj = *objectp + 1; } offset = start_obj << DNODE_SHIFT; error = dnode_next_offset(DMU_META_DNODE(os), (hole ? DNODE_FIND_HOLE : 0), &offset, 0, DNODES_PER_BLOCK, txg); *objectp = offset >> DNODE_SHIFT; return (error); } /* * Turn this object from old_type into DMU_OTN_ZAP_METADATA, and bump the * refcount on SPA_FEATURE_EXTENSIBLE_DATASET. * * Only for use from syncing context, on MOS objects. */ void dmu_object_zapify(objset_t *mos, uint64_t object, dmu_object_type_t old_type, dmu_tx_t *tx) { dnode_t *dn; ASSERT(dmu_tx_is_syncing(tx)); VERIFY0(dnode_hold(mos, object, FTAG, &dn)); if (dn->dn_type == DMU_OTN_ZAP_METADATA) { dnode_rele(dn, FTAG); return; } ASSERT3U(dn->dn_type, ==, old_type); ASSERT0(dn->dn_maxblkid); /* * We must initialize the ZAP data before changing the type, * so that concurrent calls to *_is_zapified() can determine if * the object has been completely zapified by checking the type. */ mzap_create_impl(mos, object, 0, 0, tx); dn->dn_next_type[tx->tx_txg & TXG_MASK] = dn->dn_type = DMU_OTN_ZAP_METADATA; dnode_setdirty(dn, tx); dnode_rele(dn, FTAG); spa_feature_incr(dmu_objset_spa(mos), SPA_FEATURE_EXTENSIBLE_DATASET, tx); } void dmu_object_free_zapified(objset_t *mos, uint64_t object, dmu_tx_t *tx) { dnode_t *dn; dmu_object_type_t t; ASSERT(dmu_tx_is_syncing(tx)); VERIFY0(dnode_hold(mos, object, FTAG, &dn)); t = dn->dn_type; dnode_rele(dn, FTAG); if (t == DMU_OTN_ZAP_METADATA) { spa_feature_decr(dmu_objset_spa(mos), SPA_FEATURE_EXTENSIBLE_DATASET, tx); } VERIFY0(dmu_object_free(mos, object, tx)); } Index: head/sys/cddl/contrib/opensolaris/uts/common/fs/zfs/dmu_send.c =================================================================== --- head/sys/cddl/contrib/opensolaris/uts/common/fs/zfs/dmu_send.c (revision 353555) +++ head/sys/cddl/contrib/opensolaris/uts/common/fs/zfs/dmu_send.c (revision 353556) @@ -1,3518 +1,3518 @@ /* * CDDL HEADER START * * The contents of this file are subject to the terms of the * Common Development and Distribution License (the "License"). * You may not use this file except in compliance with the License. * * You can obtain a copy of the license at usr/src/OPENSOLARIS.LICENSE * or http://www.opensolaris.org/os/licensing. * See the License for the specific language governing permissions * and limitations under the License. * * When distributing Covered Code, include this CDDL HEADER in each * file and include the License file at usr/src/OPENSOLARIS.LICENSE. * If applicable, add the following below this CDDL HEADER, with the * fields enclosed by brackets "[]" replaced with your own identifying * information: Portions Copyright [yyyy] [name of copyright owner] * * CDDL HEADER END */ /* * Copyright (c) 2005, 2010, Oracle and/or its affiliates. All rights reserved. * Copyright 2011 Nexenta Systems, Inc. All rights reserved. * Copyright (c) 2011, 2015 by Delphix. All rights reserved. * Copyright (c) 2014, Joyent, Inc. All rights reserved. * Copyright (c) 2012, Martin Matuska . All rights reserved. * Copyright 2014 HybridCluster. All rights reserved. * Copyright 2016 RackTop Systems. * Copyright (c) 2014 Integros [integros.com] */ #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #ifdef __FreeBSD__ #undef dump_write #define dump_write dmu_dump_write #endif /* Set this tunable to TRUE to replace corrupt data with 0x2f5baddb10c */ int zfs_send_corrupt_data = B_FALSE; int zfs_send_queue_length = 16 * 1024 * 1024; int zfs_recv_queue_length = 16 * 1024 * 1024; /* Set this tunable to FALSE to disable setting of DRR_FLAG_FREERECORDS */ int zfs_send_set_freerecords_bit = B_TRUE; #ifdef _KERNEL TUNABLE_INT("vfs.zfs.send_set_freerecords_bit", &zfs_send_set_freerecords_bit); #endif static char *dmu_recv_tag = "dmu_recv_tag"; const char *recv_clone_name = "%recv"; /* * Use this to override the recordsize calculation for fast zfs send estimates. */ uint64_t zfs_override_estimate_recordsize = 0; #define BP_SPAN(datablkszsec, indblkshift, level) \ (((uint64_t)datablkszsec) << (SPA_MINBLOCKSHIFT + \ (level) * (indblkshift - SPA_BLKPTRSHIFT))) static void byteswap_record(dmu_replay_record_t *drr); struct send_thread_arg { bqueue_t q; dsl_dataset_t *ds; /* Dataset to traverse */ uint64_t fromtxg; /* Traverse from this txg */ int flags; /* flags to pass to traverse_dataset */ int error_code; boolean_t cancel; zbookmark_phys_t resume; }; struct send_block_record { boolean_t eos_marker; /* Marks the end of the stream */ blkptr_t bp; zbookmark_phys_t zb; uint8_t indblkshift; uint16_t datablkszsec; bqueue_node_t ln; }; static int dump_bytes(dmu_sendarg_t *dsp, void *buf, int len) { dsl_dataset_t *ds = dmu_objset_ds(dsp->dsa_os); struct uio auio; struct iovec aiov; /* * The code does not rely on this (len being a multiple of 8). We keep * this assertion because of the corresponding assertion in * receive_read(). Keeping this assertion ensures that we do not * inadvertently break backwards compatibility (causing the assertion * in receive_read() to trigger on old software). * * Removing the assertions could be rolled into a new feature that uses * data that isn't 8-byte aligned; if the assertions were removed, a * feature flag would have to be added. */ ASSERT0(len % 8); aiov.iov_base = buf; aiov.iov_len = len; auio.uio_iov = &aiov; auio.uio_iovcnt = 1; auio.uio_resid = len; auio.uio_segflg = UIO_SYSSPACE; auio.uio_rw = UIO_WRITE; auio.uio_offset = (off_t)-1; auio.uio_td = dsp->dsa_td; #ifdef _KERNEL if (dsp->dsa_fp->f_type == DTYPE_VNODE) bwillwrite(); dsp->dsa_err = fo_write(dsp->dsa_fp, &auio, dsp->dsa_td->td_ucred, 0, dsp->dsa_td); #else fprintf(stderr, "%s: returning EOPNOTSUPP\n", __func__); dsp->dsa_err = EOPNOTSUPP; #endif mutex_enter(&ds->ds_sendstream_lock); *dsp->dsa_off += len; mutex_exit(&ds->ds_sendstream_lock); return (dsp->dsa_err); } /* * For all record types except BEGIN, fill in the checksum (overlaid in * drr_u.drr_checksum.drr_checksum). The checksum verifies everything * up to the start of the checksum itself. */ static int dump_record(dmu_sendarg_t *dsp, void *payload, int payload_len) { ASSERT3U(offsetof(dmu_replay_record_t, drr_u.drr_checksum.drr_checksum), ==, sizeof (dmu_replay_record_t) - sizeof (zio_cksum_t)); (void) fletcher_4_incremental_native(dsp->dsa_drr, offsetof(dmu_replay_record_t, drr_u.drr_checksum.drr_checksum), &dsp->dsa_zc); if (dsp->dsa_drr->drr_type == DRR_BEGIN) { dsp->dsa_sent_begin = B_TRUE; } else { ASSERT(ZIO_CHECKSUM_IS_ZERO(&dsp->dsa_drr->drr_u. drr_checksum.drr_checksum)); dsp->dsa_drr->drr_u.drr_checksum.drr_checksum = dsp->dsa_zc; } if (dsp->dsa_drr->drr_type == DRR_END) { dsp->dsa_sent_end = B_TRUE; } (void) fletcher_4_incremental_native(&dsp->dsa_drr-> drr_u.drr_checksum.drr_checksum, sizeof (zio_cksum_t), &dsp->dsa_zc); if (dump_bytes(dsp, dsp->dsa_drr, sizeof (dmu_replay_record_t)) != 0) return (SET_ERROR(EINTR)); if (payload_len != 0) { (void) fletcher_4_incremental_native(payload, payload_len, &dsp->dsa_zc); if (dump_bytes(dsp, payload, payload_len) != 0) return (SET_ERROR(EINTR)); } return (0); } /* * Fill in the drr_free struct, or perform aggregation if the previous record is * also a free record, and the two are adjacent. * * Note that we send free records even for a full send, because we want to be * able to receive a full send as a clone, which requires a list of all the free * and freeobject records that were generated on the source. */ static int dump_free(dmu_sendarg_t *dsp, uint64_t object, uint64_t offset, uint64_t length) { struct drr_free *drrf = &(dsp->dsa_drr->drr_u.drr_free); /* * When we receive a free record, dbuf_free_range() assumes * that the receiving system doesn't have any dbufs in the range * being freed. This is always true because there is a one-record * constraint: we only send one WRITE record for any given * object,offset. We know that the one-record constraint is * true because we always send data in increasing order by * object,offset. * * If the increasing-order constraint ever changes, we should find * another way to assert that the one-record constraint is still * satisfied. */ ASSERT(object > dsp->dsa_last_data_object || (object == dsp->dsa_last_data_object && offset > dsp->dsa_last_data_offset)); if (length != -1ULL && offset + length < offset) length = -1ULL; /* * If there is a pending op, but it's not PENDING_FREE, push it out, * since free block aggregation can only be done for blocks of the * same type (i.e., DRR_FREE records can only be aggregated with * other DRR_FREE records. DRR_FREEOBJECTS records can only be * aggregated with other DRR_FREEOBJECTS records. */ if (dsp->dsa_pending_op != PENDING_NONE && dsp->dsa_pending_op != PENDING_FREE) { if (dump_record(dsp, NULL, 0) != 0) return (SET_ERROR(EINTR)); dsp->dsa_pending_op = PENDING_NONE; } if (dsp->dsa_pending_op == PENDING_FREE) { /* * There should never be a PENDING_FREE if length is -1 * (because dump_dnode is the only place where this * function is called with a -1, and only after flushing * any pending record). */ ASSERT(length != -1ULL); /* * Check to see whether this free block can be aggregated * with pending one. */ if (drrf->drr_object == object && drrf->drr_offset + drrf->drr_length == offset) { drrf->drr_length += length; return (0); } else { /* not a continuation. Push out pending record */ if (dump_record(dsp, NULL, 0) != 0) return (SET_ERROR(EINTR)); dsp->dsa_pending_op = PENDING_NONE; } } /* create a FREE record and make it pending */ bzero(dsp->dsa_drr, sizeof (dmu_replay_record_t)); dsp->dsa_drr->drr_type = DRR_FREE; drrf->drr_object = object; drrf->drr_offset = offset; drrf->drr_length = length; drrf->drr_toguid = dsp->dsa_toguid; if (length == -1ULL) { if (dump_record(dsp, NULL, 0) != 0) return (SET_ERROR(EINTR)); } else { dsp->dsa_pending_op = PENDING_FREE; } return (0); } static int dump_write(dmu_sendarg_t *dsp, dmu_object_type_t type, uint64_t object, uint64_t offset, int lsize, int psize, const blkptr_t *bp, void *data) { uint64_t payload_size; struct drr_write *drrw = &(dsp->dsa_drr->drr_u.drr_write); /* * We send data in increasing object, offset order. * See comment in dump_free() for details. */ ASSERT(object > dsp->dsa_last_data_object || (object == dsp->dsa_last_data_object && offset > dsp->dsa_last_data_offset)); dsp->dsa_last_data_object = object; dsp->dsa_last_data_offset = offset + lsize - 1; /* * If there is any kind of pending aggregation (currently either * a grouping of free objects or free blocks), push it out to * the stream, since aggregation can't be done across operations * of different types. */ if (dsp->dsa_pending_op != PENDING_NONE) { if (dump_record(dsp, NULL, 0) != 0) return (SET_ERROR(EINTR)); dsp->dsa_pending_op = PENDING_NONE; } /* write a WRITE record */ bzero(dsp->dsa_drr, sizeof (dmu_replay_record_t)); dsp->dsa_drr->drr_type = DRR_WRITE; drrw->drr_object = object; drrw->drr_type = type; drrw->drr_offset = offset; drrw->drr_toguid = dsp->dsa_toguid; drrw->drr_logical_size = lsize; /* only set the compression fields if the buf is compressed */ if (lsize != psize) { ASSERT(dsp->dsa_featureflags & DMU_BACKUP_FEATURE_COMPRESSED); ASSERT(!BP_IS_EMBEDDED(bp)); ASSERT(!BP_SHOULD_BYTESWAP(bp)); ASSERT(!DMU_OT_IS_METADATA(BP_GET_TYPE(bp))); ASSERT3U(BP_GET_COMPRESS(bp), !=, ZIO_COMPRESS_OFF); ASSERT3S(psize, >, 0); ASSERT3S(lsize, >=, psize); drrw->drr_compressiontype = BP_GET_COMPRESS(bp); drrw->drr_compressed_size = psize; payload_size = drrw->drr_compressed_size; } else { payload_size = drrw->drr_logical_size; } if (bp == NULL || BP_IS_EMBEDDED(bp)) { /* * There's no pre-computed checksum for partial-block * writes or embedded BP's, so (like * fletcher4-checkummed blocks) userland will have to * compute a dedup-capable checksum itself. */ drrw->drr_checksumtype = ZIO_CHECKSUM_OFF; } else { drrw->drr_checksumtype = BP_GET_CHECKSUM(bp); if (zio_checksum_table[drrw->drr_checksumtype].ci_flags & ZCHECKSUM_FLAG_DEDUP) drrw->drr_checksumflags |= DRR_CHECKSUM_DEDUP; DDK_SET_LSIZE(&drrw->drr_key, BP_GET_LSIZE(bp)); DDK_SET_PSIZE(&drrw->drr_key, BP_GET_PSIZE(bp)); DDK_SET_COMPRESS(&drrw->drr_key, BP_GET_COMPRESS(bp)); drrw->drr_key.ddk_cksum = bp->blk_cksum; } if (dump_record(dsp, data, payload_size) != 0) return (SET_ERROR(EINTR)); return (0); } static int dump_write_embedded(dmu_sendarg_t *dsp, uint64_t object, uint64_t offset, int blksz, const blkptr_t *bp) { char buf[BPE_PAYLOAD_SIZE]; struct drr_write_embedded *drrw = &(dsp->dsa_drr->drr_u.drr_write_embedded); if (dsp->dsa_pending_op != PENDING_NONE) { if (dump_record(dsp, NULL, 0) != 0) return (EINTR); dsp->dsa_pending_op = PENDING_NONE; } ASSERT(BP_IS_EMBEDDED(bp)); bzero(dsp->dsa_drr, sizeof (dmu_replay_record_t)); dsp->dsa_drr->drr_type = DRR_WRITE_EMBEDDED; drrw->drr_object = object; drrw->drr_offset = offset; drrw->drr_length = blksz; drrw->drr_toguid = dsp->dsa_toguid; drrw->drr_compression = BP_GET_COMPRESS(bp); drrw->drr_etype = BPE_GET_ETYPE(bp); drrw->drr_lsize = BPE_GET_LSIZE(bp); drrw->drr_psize = BPE_GET_PSIZE(bp); decode_embedded_bp_compressed(bp, buf); if (dump_record(dsp, buf, P2ROUNDUP(drrw->drr_psize, 8)) != 0) return (EINTR); return (0); } static int dump_spill(dmu_sendarg_t *dsp, uint64_t object, int blksz, void *data) { struct drr_spill *drrs = &(dsp->dsa_drr->drr_u.drr_spill); if (dsp->dsa_pending_op != PENDING_NONE) { if (dump_record(dsp, NULL, 0) != 0) return (SET_ERROR(EINTR)); dsp->dsa_pending_op = PENDING_NONE; } /* write a SPILL record */ bzero(dsp->dsa_drr, sizeof (dmu_replay_record_t)); dsp->dsa_drr->drr_type = DRR_SPILL; drrs->drr_object = object; drrs->drr_length = blksz; drrs->drr_toguid = dsp->dsa_toguid; if (dump_record(dsp, data, blksz) != 0) return (SET_ERROR(EINTR)); return (0); } static int dump_freeobjects(dmu_sendarg_t *dsp, uint64_t firstobj, uint64_t numobjs) { struct drr_freeobjects *drrfo = &(dsp->dsa_drr->drr_u.drr_freeobjects); /* * If there is a pending op, but it's not PENDING_FREEOBJECTS, * push it out, since free block aggregation can only be done for * blocks of the same type (i.e., DRR_FREE records can only be * aggregated with other DRR_FREE records. DRR_FREEOBJECTS records * can only be aggregated with other DRR_FREEOBJECTS records. */ if (dsp->dsa_pending_op != PENDING_NONE && dsp->dsa_pending_op != PENDING_FREEOBJECTS) { if (dump_record(dsp, NULL, 0) != 0) return (SET_ERROR(EINTR)); dsp->dsa_pending_op = PENDING_NONE; } if (dsp->dsa_pending_op == PENDING_FREEOBJECTS) { /* * See whether this free object array can be aggregated * with pending one */ if (drrfo->drr_firstobj + drrfo->drr_numobjs == firstobj) { drrfo->drr_numobjs += numobjs; return (0); } else { /* can't be aggregated. Push out pending record */ if (dump_record(dsp, NULL, 0) != 0) return (SET_ERROR(EINTR)); dsp->dsa_pending_op = PENDING_NONE; } } /* write a FREEOBJECTS record */ bzero(dsp->dsa_drr, sizeof (dmu_replay_record_t)); dsp->dsa_drr->drr_type = DRR_FREEOBJECTS; drrfo->drr_firstobj = firstobj; drrfo->drr_numobjs = numobjs; drrfo->drr_toguid = dsp->dsa_toguid; dsp->dsa_pending_op = PENDING_FREEOBJECTS; return (0); } static int dump_dnode(dmu_sendarg_t *dsp, uint64_t object, dnode_phys_t *dnp) { struct drr_object *drro = &(dsp->dsa_drr->drr_u.drr_object); if (object < dsp->dsa_resume_object) { /* * Note: when resuming, we will visit all the dnodes in * the block of dnodes that we are resuming from. In * this case it's unnecessary to send the dnodes prior to * the one we are resuming from. We should be at most one * block's worth of dnodes behind the resume point. */ ASSERT3U(dsp->dsa_resume_object - object, <, 1 << (DNODE_BLOCK_SHIFT - DNODE_SHIFT)); return (0); } if (dnp == NULL || dnp->dn_type == DMU_OT_NONE) return (dump_freeobjects(dsp, object, 1)); if (dsp->dsa_pending_op != PENDING_NONE) { if (dump_record(dsp, NULL, 0) != 0) return (SET_ERROR(EINTR)); dsp->dsa_pending_op = PENDING_NONE; } /* write an OBJECT record */ bzero(dsp->dsa_drr, sizeof (dmu_replay_record_t)); dsp->dsa_drr->drr_type = DRR_OBJECT; drro->drr_object = object; drro->drr_type = dnp->dn_type; drro->drr_bonustype = dnp->dn_bonustype; drro->drr_blksz = dnp->dn_datablkszsec << SPA_MINBLOCKSHIFT; drro->drr_bonuslen = dnp->dn_bonuslen; drro->drr_dn_slots = dnp->dn_extra_slots + 1; drro->drr_checksumtype = dnp->dn_checksum; drro->drr_compress = dnp->dn_compress; drro->drr_toguid = dsp->dsa_toguid; if (!(dsp->dsa_featureflags & DMU_BACKUP_FEATURE_LARGE_BLOCKS) && drro->drr_blksz > SPA_OLD_MAXBLOCKSIZE) drro->drr_blksz = SPA_OLD_MAXBLOCKSIZE; if (dump_record(dsp, DN_BONUS(dnp), P2ROUNDUP(dnp->dn_bonuslen, 8)) != 0) { return (SET_ERROR(EINTR)); } /* Free anything past the end of the file. */ if (dump_free(dsp, object, (dnp->dn_maxblkid + 1) * (dnp->dn_datablkszsec << SPA_MINBLOCKSHIFT), -1ULL) != 0) return (SET_ERROR(EINTR)); if (dsp->dsa_err != 0) return (SET_ERROR(EINTR)); return (0); } static boolean_t backup_do_embed(dmu_sendarg_t *dsp, const blkptr_t *bp) { if (!BP_IS_EMBEDDED(bp)) return (B_FALSE); /* * Compression function must be legacy, or explicitly enabled. */ if ((BP_GET_COMPRESS(bp) >= ZIO_COMPRESS_LEGACY_FUNCTIONS && !(dsp->dsa_featureflags & DMU_BACKUP_FEATURE_LZ4))) return (B_FALSE); /* * Embed type must be explicitly enabled. */ switch (BPE_GET_ETYPE(bp)) { case BP_EMBEDDED_TYPE_DATA: if (dsp->dsa_featureflags & DMU_BACKUP_FEATURE_EMBED_DATA) return (B_TRUE); break; default: return (B_FALSE); } return (B_FALSE); } /* * This is the callback function to traverse_dataset that acts as the worker * thread for dmu_send_impl. */ /*ARGSUSED*/ static int send_cb(spa_t *spa, zilog_t *zilog, const blkptr_t *bp, const zbookmark_phys_t *zb, const struct dnode_phys *dnp, void *arg) { struct send_thread_arg *sta = arg; struct send_block_record *record; uint64_t record_size; int err = 0; ASSERT(zb->zb_object == DMU_META_DNODE_OBJECT || zb->zb_object >= sta->resume.zb_object); if (sta->cancel) return (SET_ERROR(EINTR)); if (bp == NULL) { ASSERT3U(zb->zb_level, ==, ZB_DNODE_LEVEL); return (0); } else if (zb->zb_level < 0) { return (0); } record = kmem_zalloc(sizeof (struct send_block_record), KM_SLEEP); record->eos_marker = B_FALSE; record->bp = *bp; record->zb = *zb; record->indblkshift = dnp->dn_indblkshift; record->datablkszsec = dnp->dn_datablkszsec; record_size = dnp->dn_datablkszsec << SPA_MINBLOCKSHIFT; bqueue_enqueue(&sta->q, record, record_size); return (err); } /* * This function kicks off the traverse_dataset. It also handles setting the * error code of the thread in case something goes wrong, and pushes the End of * Stream record when the traverse_dataset call has finished. If there is no * dataset to traverse, the thread immediately pushes End of Stream marker. */ static void send_traverse_thread(void *arg) { struct send_thread_arg *st_arg = arg; int err; struct send_block_record *data; if (st_arg->ds != NULL) { err = traverse_dataset_resume(st_arg->ds, st_arg->fromtxg, &st_arg->resume, st_arg->flags, send_cb, st_arg); if (err != EINTR) st_arg->error_code = err; } data = kmem_zalloc(sizeof (*data), KM_SLEEP); data->eos_marker = B_TRUE; bqueue_enqueue(&st_arg->q, data, 1); thread_exit(); } /* * This function actually handles figuring out what kind of record needs to be * dumped, reading the data (which has hopefully been prefetched), and calling * the appropriate helper function. */ static int do_dump(dmu_sendarg_t *dsa, struct send_block_record *data) { dsl_dataset_t *ds = dmu_objset_ds(dsa->dsa_os); const blkptr_t *bp = &data->bp; const zbookmark_phys_t *zb = &data->zb; uint8_t indblkshift = data->indblkshift; uint16_t dblkszsec = data->datablkszsec; spa_t *spa = ds->ds_dir->dd_pool->dp_spa; dmu_object_type_t type = bp ? BP_GET_TYPE(bp) : DMU_OT_NONE; int err = 0; ASSERT3U(zb->zb_level, >=, 0); ASSERT(zb->zb_object == DMU_META_DNODE_OBJECT || zb->zb_object >= dsa->dsa_resume_object); if (zb->zb_object != DMU_META_DNODE_OBJECT && DMU_OBJECT_IS_SPECIAL(zb->zb_object)) { return (0); } else if (BP_IS_HOLE(bp) && zb->zb_object == DMU_META_DNODE_OBJECT) { uint64_t span = BP_SPAN(dblkszsec, indblkshift, zb->zb_level); uint64_t dnobj = (zb->zb_blkid * span) >> DNODE_SHIFT; err = dump_freeobjects(dsa, dnobj, span >> DNODE_SHIFT); } else if (BP_IS_HOLE(bp)) { uint64_t span = BP_SPAN(dblkszsec, indblkshift, zb->zb_level); uint64_t offset = zb->zb_blkid * span; err = dump_free(dsa, zb->zb_object, offset, span); } else if (zb->zb_level > 0 || type == DMU_OT_OBJSET) { return (0); } else if (type == DMU_OT_DNODE) { int epb = BP_GET_LSIZE(bp) >> DNODE_SHIFT; arc_flags_t aflags = ARC_FLAG_WAIT; arc_buf_t *abuf; ASSERT0(zb->zb_level); if (arc_read(NULL, spa, bp, arc_getbuf_func, &abuf, ZIO_PRIORITY_ASYNC_READ, ZIO_FLAG_CANFAIL, &aflags, zb) != 0) return (SET_ERROR(EIO)); dnode_phys_t *blk = abuf->b_data; uint64_t dnobj = zb->zb_blkid * epb; for (int i = 0; i < epb; i += blk[i].dn_extra_slots + 1) { err = dump_dnode(dsa, dnobj + i, blk + i); if (err != 0) break; } arc_buf_destroy(abuf, &abuf); } else if (type == DMU_OT_SA) { arc_flags_t aflags = ARC_FLAG_WAIT; arc_buf_t *abuf; int blksz = BP_GET_LSIZE(bp); if (arc_read(NULL, spa, bp, arc_getbuf_func, &abuf, ZIO_PRIORITY_ASYNC_READ, ZIO_FLAG_CANFAIL, &aflags, zb) != 0) return (SET_ERROR(EIO)); err = dump_spill(dsa, zb->zb_object, blksz, abuf->b_data); arc_buf_destroy(abuf, &abuf); } else if (backup_do_embed(dsa, bp)) { /* it's an embedded level-0 block of a regular object */ int blksz = dblkszsec << SPA_MINBLOCKSHIFT; ASSERT0(zb->zb_level); err = dump_write_embedded(dsa, zb->zb_object, zb->zb_blkid * blksz, blksz, bp); } else { /* it's a level-0 block of a regular object */ arc_flags_t aflags = ARC_FLAG_WAIT; arc_buf_t *abuf; int blksz = dblkszsec << SPA_MINBLOCKSHIFT; uint64_t offset; /* * If we have large blocks stored on disk but the send flags * don't allow us to send large blocks, we split the data from * the arc buf into chunks. */ boolean_t split_large_blocks = blksz > SPA_OLD_MAXBLOCKSIZE && !(dsa->dsa_featureflags & DMU_BACKUP_FEATURE_LARGE_BLOCKS); /* * We should only request compressed data from the ARC if all * the following are true: * - stream compression was requested * - we aren't splitting large blocks into smaller chunks * - the data won't need to be byteswapped before sending * - this isn't an embedded block * - this isn't metadata (if receiving on a different endian * system it can be byteswapped more easily) */ boolean_t request_compressed = (dsa->dsa_featureflags & DMU_BACKUP_FEATURE_COMPRESSED) && !split_large_blocks && !BP_SHOULD_BYTESWAP(bp) && !BP_IS_EMBEDDED(bp) && !DMU_OT_IS_METADATA(BP_GET_TYPE(bp)); ASSERT0(zb->zb_level); ASSERT(zb->zb_object > dsa->dsa_resume_object || (zb->zb_object == dsa->dsa_resume_object && zb->zb_blkid * blksz >= dsa->dsa_resume_offset)); ASSERT0(zb->zb_level); ASSERT(zb->zb_object > dsa->dsa_resume_object || (zb->zb_object == dsa->dsa_resume_object && zb->zb_blkid * blksz >= dsa->dsa_resume_offset)); ASSERT3U(blksz, ==, BP_GET_LSIZE(bp)); enum zio_flag zioflags = ZIO_FLAG_CANFAIL; if (request_compressed) zioflags |= ZIO_FLAG_RAW; if (arc_read(NULL, spa, bp, arc_getbuf_func, &abuf, ZIO_PRIORITY_ASYNC_READ, zioflags, &aflags, zb) != 0) { if (zfs_send_corrupt_data) { /* Send a block filled with 0x"zfs badd bloc" */ abuf = arc_alloc_buf(spa, &abuf, ARC_BUFC_DATA, blksz); uint64_t *ptr; for (ptr = abuf->b_data; (char *)ptr < (char *)abuf->b_data + blksz; ptr++) *ptr = 0x2f5baddb10cULL; } else { return (SET_ERROR(EIO)); } } offset = zb->zb_blkid * blksz; if (split_large_blocks) { ASSERT3U(arc_get_compression(abuf), ==, ZIO_COMPRESS_OFF); char *buf = abuf->b_data; while (blksz > 0 && err == 0) { int n = MIN(blksz, SPA_OLD_MAXBLOCKSIZE); err = dump_write(dsa, type, zb->zb_object, offset, n, n, NULL, buf); offset += n; buf += n; blksz -= n; } } else { err = dump_write(dsa, type, zb->zb_object, offset, blksz, arc_buf_size(abuf), bp, abuf->b_data); } arc_buf_destroy(abuf, &abuf); } ASSERT(err == 0 || err == EINTR); return (err); } /* * Pop the new data off the queue, and free the old data. */ static struct send_block_record * get_next_record(bqueue_t *bq, struct send_block_record *data) { struct send_block_record *tmp = bqueue_dequeue(bq); kmem_free(data, sizeof (*data)); return (tmp); } /* * Actually do the bulk of the work in a zfs send. * * Note: Releases dp using the specified tag. */ static int dmu_send_impl(void *tag, dsl_pool_t *dp, dsl_dataset_t *to_ds, zfs_bookmark_phys_t *ancestor_zb, boolean_t is_clone, boolean_t embedok, boolean_t large_block_ok, boolean_t compressok, int outfd, uint64_t resumeobj, uint64_t resumeoff, #ifdef illumos vnode_t *vp, offset_t *off) #else struct file *fp, offset_t *off) #endif { objset_t *os; dmu_replay_record_t *drr; dmu_sendarg_t *dsp; int err; uint64_t fromtxg = 0; uint64_t featureflags = 0; struct send_thread_arg to_arg = { 0 }; err = dmu_objset_from_ds(to_ds, &os); if (err != 0) { dsl_pool_rele(dp, tag); return (err); } drr = kmem_zalloc(sizeof (dmu_replay_record_t), KM_SLEEP); drr->drr_type = DRR_BEGIN; drr->drr_u.drr_begin.drr_magic = DMU_BACKUP_MAGIC; DMU_SET_STREAM_HDRTYPE(drr->drr_u.drr_begin.drr_versioninfo, DMU_SUBSTREAM); #ifdef _KERNEL if (dmu_objset_type(os) == DMU_OST_ZFS) { uint64_t version; if (zfs_get_zplprop(os, ZFS_PROP_VERSION, &version) != 0) { kmem_free(drr, sizeof (dmu_replay_record_t)); dsl_pool_rele(dp, tag); return (SET_ERROR(EINVAL)); } if (version >= ZPL_VERSION_SA) { featureflags |= DMU_BACKUP_FEATURE_SA_SPILL; } } #endif if (large_block_ok && to_ds->ds_feature_inuse[SPA_FEATURE_LARGE_BLOCKS]) featureflags |= DMU_BACKUP_FEATURE_LARGE_BLOCKS; if (to_ds->ds_feature_inuse[SPA_FEATURE_LARGE_DNODE]) featureflags |= DMU_BACKUP_FEATURE_LARGE_DNODE; if (embedok && spa_feature_is_active(dp->dp_spa, SPA_FEATURE_EMBEDDED_DATA)) { featureflags |= DMU_BACKUP_FEATURE_EMBED_DATA; if (spa_feature_is_active(dp->dp_spa, SPA_FEATURE_LZ4_COMPRESS)) featureflags |= DMU_BACKUP_FEATURE_LZ4; } if (compressok) { featureflags |= DMU_BACKUP_FEATURE_COMPRESSED; } if ((featureflags & (DMU_BACKUP_FEATURE_EMBED_DATA | DMU_BACKUP_FEATURE_COMPRESSED)) != 0 && spa_feature_is_active(dp->dp_spa, SPA_FEATURE_LZ4_COMPRESS)) { featureflags |= DMU_BACKUP_FEATURE_LZ4; } if (resumeobj != 0 || resumeoff != 0) { featureflags |= DMU_BACKUP_FEATURE_RESUMING; } DMU_SET_FEATUREFLAGS(drr->drr_u.drr_begin.drr_versioninfo, featureflags); drr->drr_u.drr_begin.drr_creation_time = dsl_dataset_phys(to_ds)->ds_creation_time; drr->drr_u.drr_begin.drr_type = dmu_objset_type(os); if (is_clone) drr->drr_u.drr_begin.drr_flags |= DRR_FLAG_CLONE; drr->drr_u.drr_begin.drr_toguid = dsl_dataset_phys(to_ds)->ds_guid; if (dsl_dataset_phys(to_ds)->ds_flags & DS_FLAG_CI_DATASET) drr->drr_u.drr_begin.drr_flags |= DRR_FLAG_CI_DATA; if (zfs_send_set_freerecords_bit) drr->drr_u.drr_begin.drr_flags |= DRR_FLAG_FREERECORDS; if (ancestor_zb != NULL) { drr->drr_u.drr_begin.drr_fromguid = ancestor_zb->zbm_guid; fromtxg = ancestor_zb->zbm_creation_txg; } dsl_dataset_name(to_ds, drr->drr_u.drr_begin.drr_toname); if (!to_ds->ds_is_snapshot) { (void) strlcat(drr->drr_u.drr_begin.drr_toname, "@--head--", sizeof (drr->drr_u.drr_begin.drr_toname)); } dsp = kmem_zalloc(sizeof (dmu_sendarg_t), KM_SLEEP); dsp->dsa_drr = drr; dsp->dsa_outfd = outfd; dsp->dsa_proc = curproc; dsp->dsa_td = curthread; dsp->dsa_fp = fp; dsp->dsa_os = os; dsp->dsa_off = off; dsp->dsa_toguid = dsl_dataset_phys(to_ds)->ds_guid; dsp->dsa_pending_op = PENDING_NONE; dsp->dsa_featureflags = featureflags; dsp->dsa_resume_object = resumeobj; dsp->dsa_resume_offset = resumeoff; mutex_enter(&to_ds->ds_sendstream_lock); list_insert_head(&to_ds->ds_sendstreams, dsp); mutex_exit(&to_ds->ds_sendstream_lock); dsl_dataset_long_hold(to_ds, FTAG); dsl_pool_rele(dp, tag); void *payload = NULL; size_t payload_len = 0; if (resumeobj != 0 || resumeoff != 0) { dmu_object_info_t to_doi; err = dmu_object_info(os, resumeobj, &to_doi); if (err != 0) goto out; SET_BOOKMARK(&to_arg.resume, to_ds->ds_object, resumeobj, 0, resumeoff / to_doi.doi_data_block_size); nvlist_t *nvl = fnvlist_alloc(); fnvlist_add_uint64(nvl, "resume_object", resumeobj); fnvlist_add_uint64(nvl, "resume_offset", resumeoff); payload = fnvlist_pack(nvl, &payload_len); drr->drr_payloadlen = payload_len; fnvlist_free(nvl); } err = dump_record(dsp, payload, payload_len); fnvlist_pack_free(payload, payload_len); if (err != 0) { err = dsp->dsa_err; goto out; } err = bqueue_init(&to_arg.q, zfs_send_queue_length, offsetof(struct send_block_record, ln)); to_arg.error_code = 0; to_arg.cancel = B_FALSE; to_arg.ds = to_ds; to_arg.fromtxg = fromtxg; to_arg.flags = TRAVERSE_PRE | TRAVERSE_PREFETCH; (void) thread_create(NULL, 0, send_traverse_thread, &to_arg, 0, &p0, TS_RUN, minclsyspri); struct send_block_record *to_data; to_data = bqueue_dequeue(&to_arg.q); while (!to_data->eos_marker && err == 0) { err = do_dump(dsp, to_data); to_data = get_next_record(&to_arg.q, to_data); if (issig(JUSTLOOKING) && issig(FORREAL)) err = EINTR; } if (err != 0) { to_arg.cancel = B_TRUE; while (!to_data->eos_marker) { to_data = get_next_record(&to_arg.q, to_data); } } kmem_free(to_data, sizeof (*to_data)); bqueue_destroy(&to_arg.q); if (err == 0 && to_arg.error_code != 0) err = to_arg.error_code; if (err != 0) goto out; if (dsp->dsa_pending_op != PENDING_NONE) if (dump_record(dsp, NULL, 0) != 0) err = SET_ERROR(EINTR); if (err != 0) { if (err == EINTR && dsp->dsa_err != 0) err = dsp->dsa_err; goto out; } bzero(drr, sizeof (dmu_replay_record_t)); drr->drr_type = DRR_END; drr->drr_u.drr_end.drr_checksum = dsp->dsa_zc; drr->drr_u.drr_end.drr_toguid = dsp->dsa_toguid; if (dump_record(dsp, NULL, 0) != 0) err = dsp->dsa_err; out: mutex_enter(&to_ds->ds_sendstream_lock); list_remove(&to_ds->ds_sendstreams, dsp); mutex_exit(&to_ds->ds_sendstream_lock); VERIFY(err != 0 || (dsp->dsa_sent_begin && dsp->dsa_sent_end)); kmem_free(drr, sizeof (dmu_replay_record_t)); kmem_free(dsp, sizeof (dmu_sendarg_t)); dsl_dataset_long_rele(to_ds, FTAG); return (err); } int dmu_send_obj(const char *pool, uint64_t tosnap, uint64_t fromsnap, boolean_t embedok, boolean_t large_block_ok, boolean_t compressok, #ifdef illumos int outfd, vnode_t *vp, offset_t *off) #else int outfd, struct file *fp, offset_t *off) #endif { dsl_pool_t *dp; dsl_dataset_t *ds; dsl_dataset_t *fromds = NULL; int err; err = dsl_pool_hold(pool, FTAG, &dp); if (err != 0) return (err); err = dsl_dataset_hold_obj(dp, tosnap, FTAG, &ds); if (err != 0) { dsl_pool_rele(dp, FTAG); return (err); } if (fromsnap != 0) { zfs_bookmark_phys_t zb; boolean_t is_clone; err = dsl_dataset_hold_obj(dp, fromsnap, FTAG, &fromds); if (err != 0) { dsl_dataset_rele(ds, FTAG); dsl_pool_rele(dp, FTAG); return (err); } if (!dsl_dataset_is_before(ds, fromds, 0)) err = SET_ERROR(EXDEV); zb.zbm_creation_time = dsl_dataset_phys(fromds)->ds_creation_time; zb.zbm_creation_txg = dsl_dataset_phys(fromds)->ds_creation_txg; zb.zbm_guid = dsl_dataset_phys(fromds)->ds_guid; is_clone = (fromds->ds_dir != ds->ds_dir); dsl_dataset_rele(fromds, FTAG); err = dmu_send_impl(FTAG, dp, ds, &zb, is_clone, embedok, large_block_ok, compressok, outfd, 0, 0, fp, off); } else { err = dmu_send_impl(FTAG, dp, ds, NULL, B_FALSE, embedok, large_block_ok, compressok, outfd, 0, 0, fp, off); } dsl_dataset_rele(ds, FTAG); return (err); } int dmu_send(const char *tosnap, const char *fromsnap, boolean_t embedok, boolean_t large_block_ok, boolean_t compressok, int outfd, uint64_t resumeobj, uint64_t resumeoff, #ifdef illumos vnode_t *vp, offset_t *off) #else struct file *fp, offset_t *off) #endif { dsl_pool_t *dp; dsl_dataset_t *ds; int err; boolean_t owned = B_FALSE; if (fromsnap != NULL && strpbrk(fromsnap, "@#") == NULL) return (SET_ERROR(EINVAL)); err = dsl_pool_hold(tosnap, FTAG, &dp); if (err != 0) return (err); if (strchr(tosnap, '@') == NULL && spa_writeable(dp->dp_spa)) { /* * We are sending a filesystem or volume. Ensure * that it doesn't change by owning the dataset. */ err = dsl_dataset_own(dp, tosnap, FTAG, &ds); owned = B_TRUE; } else { err = dsl_dataset_hold(dp, tosnap, FTAG, &ds); } if (err != 0) { dsl_pool_rele(dp, FTAG); return (err); } if (fromsnap != NULL) { zfs_bookmark_phys_t zb; boolean_t is_clone = B_FALSE; int fsnamelen = strchr(tosnap, '@') - tosnap; /* * If the fromsnap is in a different filesystem, then * mark the send stream as a clone. */ if (strncmp(tosnap, fromsnap, fsnamelen) != 0 || (fromsnap[fsnamelen] != '@' && fromsnap[fsnamelen] != '#')) { is_clone = B_TRUE; } if (strchr(fromsnap, '@')) { dsl_dataset_t *fromds; err = dsl_dataset_hold(dp, fromsnap, FTAG, &fromds); if (err == 0) { if (!dsl_dataset_is_before(ds, fromds, 0)) err = SET_ERROR(EXDEV); zb.zbm_creation_time = dsl_dataset_phys(fromds)->ds_creation_time; zb.zbm_creation_txg = dsl_dataset_phys(fromds)->ds_creation_txg; zb.zbm_guid = dsl_dataset_phys(fromds)->ds_guid; is_clone = (ds->ds_dir != fromds->ds_dir); dsl_dataset_rele(fromds, FTAG); } } else { err = dsl_bookmark_lookup(dp, fromsnap, ds, &zb); } if (err != 0) { dsl_dataset_rele(ds, FTAG); dsl_pool_rele(dp, FTAG); return (err); } err = dmu_send_impl(FTAG, dp, ds, &zb, is_clone, embedok, large_block_ok, compressok, outfd, resumeobj, resumeoff, fp, off); } else { err = dmu_send_impl(FTAG, dp, ds, NULL, B_FALSE, embedok, large_block_ok, compressok, outfd, resumeobj, resumeoff, fp, off); } if (owned) dsl_dataset_disown(ds, FTAG); else dsl_dataset_rele(ds, FTAG); return (err); } static int dmu_adjust_send_estimate_for_indirects(dsl_dataset_t *ds, uint64_t uncompressed, uint64_t compressed, boolean_t stream_compressed, uint64_t *sizep) { int err = 0; uint64_t size; /* * Assume that space (both on-disk and in-stream) is dominated by * data. We will adjust for indirect blocks and the copies property, * but ignore per-object space used (eg, dnodes and DRR_OBJECT records). */ uint64_t recordsize; uint64_t record_count; objset_t *os; VERIFY0(dmu_objset_from_ds(ds, &os)); /* Assume all (uncompressed) blocks are recordsize. */ if (zfs_override_estimate_recordsize != 0) { recordsize = zfs_override_estimate_recordsize; } else if (os->os_phys->os_type == DMU_OST_ZVOL) { err = dsl_prop_get_int_ds(ds, zfs_prop_to_name(ZFS_PROP_VOLBLOCKSIZE), &recordsize); } else { err = dsl_prop_get_int_ds(ds, zfs_prop_to_name(ZFS_PROP_RECORDSIZE), &recordsize); } if (err != 0) return (err); record_count = uncompressed / recordsize; /* * If we're estimating a send size for a compressed stream, use the * compressed data size to estimate the stream size. Otherwise, use the * uncompressed data size. */ size = stream_compressed ? compressed : uncompressed; /* * Subtract out approximate space used by indirect blocks. * Assume most space is used by data blocks (non-indirect, non-dnode). * Assume no ditto blocks or internal fragmentation. * * Therefore, space used by indirect blocks is sizeof(blkptr_t) per * block. */ size -= record_count * sizeof (blkptr_t); /* Add in the space for the record associated with each block. */ size += record_count * sizeof (dmu_replay_record_t); *sizep = size; return (0); } int dmu_send_estimate(dsl_dataset_t *ds, dsl_dataset_t *fromds, boolean_t stream_compressed, uint64_t *sizep) { dsl_pool_t *dp = ds->ds_dir->dd_pool; int err; uint64_t uncomp, comp; ASSERT(dsl_pool_config_held(dp)); /* tosnap must be a snapshot */ if (!ds->ds_is_snapshot) return (SET_ERROR(EINVAL)); /* fromsnap, if provided, must be a snapshot */ if (fromds != NULL && !fromds->ds_is_snapshot) return (SET_ERROR(EINVAL)); /* * fromsnap must be an earlier snapshot from the same fs as tosnap, * or the origin's fs. */ if (fromds != NULL && !dsl_dataset_is_before(ds, fromds, 0)) return (SET_ERROR(EXDEV)); /* Get compressed and uncompressed size estimates of changed data. */ if (fromds == NULL) { uncomp = dsl_dataset_phys(ds)->ds_uncompressed_bytes; comp = dsl_dataset_phys(ds)->ds_compressed_bytes; } else { uint64_t used; err = dsl_dataset_space_written(fromds, ds, &used, &comp, &uncomp); if (err != 0) return (err); } err = dmu_adjust_send_estimate_for_indirects(ds, uncomp, comp, stream_compressed, sizep); /* * Add the size of the BEGIN and END records to the estimate. */ *sizep += 2 * sizeof (dmu_replay_record_t); return (err); } struct calculate_send_arg { uint64_t uncompressed; uint64_t compressed; }; /* * Simple callback used to traverse the blocks of a snapshot and sum their * uncompressed and compressed sizes. */ /* ARGSUSED */ static int dmu_calculate_send_traversal(spa_t *spa, zilog_t *zilog, const blkptr_t *bp, const zbookmark_phys_t *zb, const dnode_phys_t *dnp, void *arg) { struct calculate_send_arg *space = arg; if (bp != NULL && !BP_IS_HOLE(bp)) { space->uncompressed += BP_GET_UCSIZE(bp); space->compressed += BP_GET_PSIZE(bp); } return (0); } /* * Given a desination snapshot and a TXG, calculate the approximate size of a * send stream sent from that TXG. from_txg may be zero, indicating that the * whole snapshot will be sent. */ int dmu_send_estimate_from_txg(dsl_dataset_t *ds, uint64_t from_txg, boolean_t stream_compressed, uint64_t *sizep) { dsl_pool_t *dp = ds->ds_dir->dd_pool; int err; struct calculate_send_arg size = { 0 }; ASSERT(dsl_pool_config_held(dp)); /* tosnap must be a snapshot */ if (!ds->ds_is_snapshot) return (SET_ERROR(EINVAL)); /* verify that from_txg is before the provided snapshot was taken */ if (from_txg >= dsl_dataset_phys(ds)->ds_creation_txg) { return (SET_ERROR(EXDEV)); } /* * traverse the blocks of the snapshot with birth times after * from_txg, summing their uncompressed size */ err = traverse_dataset(ds, from_txg, TRAVERSE_POST, dmu_calculate_send_traversal, &size); if (err) return (err); err = dmu_adjust_send_estimate_for_indirects(ds, size.uncompressed, size.compressed, stream_compressed, sizep); return (err); } typedef struct dmu_recv_begin_arg { const char *drba_origin; dmu_recv_cookie_t *drba_cookie; cred_t *drba_cred; uint64_t drba_snapobj; } dmu_recv_begin_arg_t; static int recv_begin_check_existing_impl(dmu_recv_begin_arg_t *drba, dsl_dataset_t *ds, uint64_t fromguid) { uint64_t val; int error; dsl_pool_t *dp = ds->ds_dir->dd_pool; /* Temporary clone name must not exist. */ error = zap_lookup(dp->dp_meta_objset, dsl_dir_phys(ds->ds_dir)->dd_child_dir_zapobj, recv_clone_name, 8, 1, &val); if (error != ENOENT) return (error == 0 ? SET_ERROR(EBUSY) : error); /* Resume state must not be set. */ if (dsl_dataset_has_resume_receive_state(ds)) return (SET_ERROR(EBUSY)); /* New snapshot name must not exist. */ error = zap_lookup(dp->dp_meta_objset, dsl_dataset_phys(ds)->ds_snapnames_zapobj, drba->drba_cookie->drc_tosnap, 8, 1, &val); if (error != ENOENT) return (error == 0 ? SET_ERROR(EEXIST) : error); /* * Check snapshot limit before receiving. We'll recheck again at the * end, but might as well abort before receiving if we're already over * the limit. * * Note that we do not check the file system limit with * dsl_dir_fscount_check because the temporary %clones don't count * against that limit. */ error = dsl_fs_ss_limit_check(ds->ds_dir, 1, ZFS_PROP_SNAPSHOT_LIMIT, NULL, drba->drba_cred); if (error != 0) return (error); if (fromguid != 0) { dsl_dataset_t *snap; uint64_t obj = dsl_dataset_phys(ds)->ds_prev_snap_obj; /* Find snapshot in this dir that matches fromguid. */ while (obj != 0) { error = dsl_dataset_hold_obj(dp, obj, FTAG, &snap); if (error != 0) return (SET_ERROR(ENODEV)); if (snap->ds_dir != ds->ds_dir) { dsl_dataset_rele(snap, FTAG); return (SET_ERROR(ENODEV)); } if (dsl_dataset_phys(snap)->ds_guid == fromguid) break; obj = dsl_dataset_phys(snap)->ds_prev_snap_obj; dsl_dataset_rele(snap, FTAG); } if (obj == 0) return (SET_ERROR(ENODEV)); if (drba->drba_cookie->drc_force) { drba->drba_snapobj = obj; } else { /* * If we are not forcing, there must be no * changes since fromsnap. */ if (dsl_dataset_modified_since_snap(ds, snap)) { dsl_dataset_rele(snap, FTAG); return (SET_ERROR(ETXTBSY)); } drba->drba_snapobj = ds->ds_prev->ds_object; } dsl_dataset_rele(snap, FTAG); } else { /* if full, then must be forced */ if (!drba->drba_cookie->drc_force) return (SET_ERROR(EEXIST)); /* start from $ORIGIN@$ORIGIN, if supported */ drba->drba_snapobj = dp->dp_origin_snap != NULL ? dp->dp_origin_snap->ds_object : 0; } return (0); } static int dmu_recv_begin_check(void *arg, dmu_tx_t *tx) { dmu_recv_begin_arg_t *drba = arg; dsl_pool_t *dp = dmu_tx_pool(tx); struct drr_begin *drrb = drba->drba_cookie->drc_drrb; uint64_t fromguid = drrb->drr_fromguid; int flags = drrb->drr_flags; int error; uint64_t featureflags = DMU_GET_FEATUREFLAGS(drrb->drr_versioninfo); dsl_dataset_t *ds; const char *tofs = drba->drba_cookie->drc_tofs; /* already checked */ ASSERT3U(drrb->drr_magic, ==, DMU_BACKUP_MAGIC); ASSERT(!(featureflags & DMU_BACKUP_FEATURE_RESUMING)); if (DMU_GET_STREAM_HDRTYPE(drrb->drr_versioninfo) == DMU_COMPOUNDSTREAM || drrb->drr_type >= DMU_OST_NUMTYPES || ((flags & DRR_FLAG_CLONE) && drba->drba_origin == NULL)) return (SET_ERROR(EINVAL)); /* Verify pool version supports SA if SA_SPILL feature set */ if ((featureflags & DMU_BACKUP_FEATURE_SA_SPILL) && spa_version(dp->dp_spa) < SPA_VERSION_SA) return (SET_ERROR(ENOTSUP)); if (drba->drba_cookie->drc_resumable && !spa_feature_is_enabled(dp->dp_spa, SPA_FEATURE_EXTENSIBLE_DATASET)) return (SET_ERROR(ENOTSUP)); /* * The receiving code doesn't know how to translate a WRITE_EMBEDDED * record to a plain WRITE record, so the pool must have the * EMBEDDED_DATA feature enabled if the stream has WRITE_EMBEDDED * records. Same with WRITE_EMBEDDED records that use LZ4 compression. */ if ((featureflags & DMU_BACKUP_FEATURE_EMBED_DATA) && !spa_feature_is_enabled(dp->dp_spa, SPA_FEATURE_EMBEDDED_DATA)) return (SET_ERROR(ENOTSUP)); if ((featureflags & DMU_BACKUP_FEATURE_LZ4) && !spa_feature_is_enabled(dp->dp_spa, SPA_FEATURE_LZ4_COMPRESS)) return (SET_ERROR(ENOTSUP)); /* * The receiving code doesn't know how to translate large blocks * to smaller ones, so the pool must have the LARGE_BLOCKS * feature enabled if the stream has LARGE_BLOCKS. Same with * large dnodes. */ if ((featureflags & DMU_BACKUP_FEATURE_LARGE_BLOCKS) && !spa_feature_is_enabled(dp->dp_spa, SPA_FEATURE_LARGE_BLOCKS)) return (SET_ERROR(ENOTSUP)); if ((featureflags & DMU_BACKUP_FEATURE_LARGE_DNODE) && !spa_feature_is_enabled(dp->dp_spa, SPA_FEATURE_LARGE_DNODE)) return (SET_ERROR(ENOTSUP)); error = dsl_dataset_hold(dp, tofs, FTAG, &ds); if (error == 0) { /* target fs already exists; recv into temp clone */ /* Can't recv a clone into an existing fs */ if (flags & DRR_FLAG_CLONE || drba->drba_origin) { dsl_dataset_rele(ds, FTAG); return (SET_ERROR(EINVAL)); } error = recv_begin_check_existing_impl(drba, ds, fromguid); dsl_dataset_rele(ds, FTAG); } else if (error == ENOENT) { /* target fs does not exist; must be a full backup or clone */ char buf[ZFS_MAX_DATASET_NAME_LEN]; /* * If it's a non-clone incremental, we are missing the * target fs, so fail the recv. */ if (fromguid != 0 && !(flags & DRR_FLAG_CLONE || drba->drba_origin)) return (SET_ERROR(ENOENT)); /* * If we're receiving a full send as a clone, and it doesn't * contain all the necessary free records and freeobject * records, reject it. */ if (fromguid == 0 && drba->drba_origin && !(flags & DRR_FLAG_FREERECORDS)) return (SET_ERROR(EINVAL)); /* Open the parent of tofs */ ASSERT3U(strlen(tofs), <, sizeof (buf)); (void) strlcpy(buf, tofs, strrchr(tofs, '/') - tofs + 1); error = dsl_dataset_hold(dp, buf, FTAG, &ds); if (error != 0) return (error); /* * Check filesystem and snapshot limits before receiving. We'll * recheck snapshot limits again at the end (we create the * filesystems and increment those counts during begin_sync). */ error = dsl_fs_ss_limit_check(ds->ds_dir, 1, ZFS_PROP_FILESYSTEM_LIMIT, NULL, drba->drba_cred); if (error != 0) { dsl_dataset_rele(ds, FTAG); return (error); } error = dsl_fs_ss_limit_check(ds->ds_dir, 1, ZFS_PROP_SNAPSHOT_LIMIT, NULL, drba->drba_cred); if (error != 0) { dsl_dataset_rele(ds, FTAG); return (error); } if (drba->drba_origin != NULL) { dsl_dataset_t *origin; error = dsl_dataset_hold(dp, drba->drba_origin, FTAG, &origin); if (error != 0) { dsl_dataset_rele(ds, FTAG); return (error); } if (!origin->ds_is_snapshot) { dsl_dataset_rele(origin, FTAG); dsl_dataset_rele(ds, FTAG); return (SET_ERROR(EINVAL)); } if (dsl_dataset_phys(origin)->ds_guid != fromguid && fromguid != 0) { dsl_dataset_rele(origin, FTAG); dsl_dataset_rele(ds, FTAG); return (SET_ERROR(ENODEV)); } dsl_dataset_rele(origin, FTAG); } dsl_dataset_rele(ds, FTAG); error = 0; } return (error); } static void dmu_recv_begin_sync(void *arg, dmu_tx_t *tx) { dmu_recv_begin_arg_t *drba = arg; dsl_pool_t *dp = dmu_tx_pool(tx); objset_t *mos = dp->dp_meta_objset; struct drr_begin *drrb = drba->drba_cookie->drc_drrb; const char *tofs = drba->drba_cookie->drc_tofs; dsl_dataset_t *ds, *newds; uint64_t dsobj; int error; uint64_t crflags = 0; if (drrb->drr_flags & DRR_FLAG_CI_DATA) crflags |= DS_FLAG_CI_DATASET; error = dsl_dataset_hold(dp, tofs, FTAG, &ds); if (error == 0) { /* create temporary clone */ dsl_dataset_t *snap = NULL; if (drba->drba_snapobj != 0) { VERIFY0(dsl_dataset_hold_obj(dp, drba->drba_snapobj, FTAG, &snap)); } dsobj = dsl_dataset_create_sync(ds->ds_dir, recv_clone_name, snap, crflags, drba->drba_cred, tx); if (drba->drba_snapobj != 0) dsl_dataset_rele(snap, FTAG); dsl_dataset_rele(ds, FTAG); } else { dsl_dir_t *dd; const char *tail; dsl_dataset_t *origin = NULL; VERIFY0(dsl_dir_hold(dp, tofs, FTAG, &dd, &tail)); if (drba->drba_origin != NULL) { VERIFY0(dsl_dataset_hold(dp, drba->drba_origin, FTAG, &origin)); } /* Create new dataset. */ dsobj = dsl_dataset_create_sync(dd, strrchr(tofs, '/') + 1, origin, crflags, drba->drba_cred, tx); if (origin != NULL) dsl_dataset_rele(origin, FTAG); dsl_dir_rele(dd, FTAG); drba->drba_cookie->drc_newfs = B_TRUE; } VERIFY0(dsl_dataset_own_obj(dp, dsobj, dmu_recv_tag, &newds)); if (drba->drba_cookie->drc_resumable) { dsl_dataset_zapify(newds, tx); if (drrb->drr_fromguid != 0) { VERIFY0(zap_add(mos, dsobj, DS_FIELD_RESUME_FROMGUID, 8, 1, &drrb->drr_fromguid, tx)); } VERIFY0(zap_add(mos, dsobj, DS_FIELD_RESUME_TOGUID, 8, 1, &drrb->drr_toguid, tx)); VERIFY0(zap_add(mos, dsobj, DS_FIELD_RESUME_TONAME, 1, strlen(drrb->drr_toname) + 1, drrb->drr_toname, tx)); uint64_t one = 1; uint64_t zero = 0; VERIFY0(zap_add(mos, dsobj, DS_FIELD_RESUME_OBJECT, 8, 1, &one, tx)); VERIFY0(zap_add(mos, dsobj, DS_FIELD_RESUME_OFFSET, 8, 1, &zero, tx)); VERIFY0(zap_add(mos, dsobj, DS_FIELD_RESUME_BYTES, 8, 1, &zero, tx)); if (DMU_GET_FEATUREFLAGS(drrb->drr_versioninfo) & DMU_BACKUP_FEATURE_LARGE_BLOCKS) { VERIFY0(zap_add(mos, dsobj, DS_FIELD_RESUME_LARGEBLOCK, 8, 1, &one, tx)); } if (DMU_GET_FEATUREFLAGS(drrb->drr_versioninfo) & DMU_BACKUP_FEATURE_EMBED_DATA) { VERIFY0(zap_add(mos, dsobj, DS_FIELD_RESUME_EMBEDOK, 8, 1, &one, tx)); } if (DMU_GET_FEATUREFLAGS(drrb->drr_versioninfo) & DMU_BACKUP_FEATURE_COMPRESSED) { VERIFY0(zap_add(mos, dsobj, DS_FIELD_RESUME_COMPRESSOK, 8, 1, &one, tx)); } } dmu_buf_will_dirty(newds->ds_dbuf, tx); dsl_dataset_phys(newds)->ds_flags |= DS_FLAG_INCONSISTENT; /* * If we actually created a non-clone, we need to create the * objset in our new dataset. */ rrw_enter(&newds->ds_bp_rwlock, RW_READER, FTAG); if (BP_IS_HOLE(dsl_dataset_get_blkptr(newds))) { (void) dmu_objset_create_impl(dp->dp_spa, newds, dsl_dataset_get_blkptr(newds), drrb->drr_type, tx); } rrw_exit(&newds->ds_bp_rwlock, FTAG); drba->drba_cookie->drc_ds = newds; spa_history_log_internal_ds(newds, "receive", tx, ""); } static int dmu_recv_resume_begin_check(void *arg, dmu_tx_t *tx) { dmu_recv_begin_arg_t *drba = arg; dsl_pool_t *dp = dmu_tx_pool(tx); struct drr_begin *drrb = drba->drba_cookie->drc_drrb; int error; uint64_t featureflags = DMU_GET_FEATUREFLAGS(drrb->drr_versioninfo); dsl_dataset_t *ds; const char *tofs = drba->drba_cookie->drc_tofs; /* 6 extra bytes for /%recv */ char recvname[ZFS_MAX_DATASET_NAME_LEN + 6]; /* already checked */ ASSERT3U(drrb->drr_magic, ==, DMU_BACKUP_MAGIC); ASSERT(featureflags & DMU_BACKUP_FEATURE_RESUMING); if (DMU_GET_STREAM_HDRTYPE(drrb->drr_versioninfo) == DMU_COMPOUNDSTREAM || drrb->drr_type >= DMU_OST_NUMTYPES) return (SET_ERROR(EINVAL)); /* Verify pool version supports SA if SA_SPILL feature set */ if ((featureflags & DMU_BACKUP_FEATURE_SA_SPILL) && spa_version(dp->dp_spa) < SPA_VERSION_SA) return (SET_ERROR(ENOTSUP)); /* * The receiving code doesn't know how to translate a WRITE_EMBEDDED * record to a plain WRITE record, so the pool must have the * EMBEDDED_DATA feature enabled if the stream has WRITE_EMBEDDED * records. Same with WRITE_EMBEDDED records that use LZ4 compression. */ if ((featureflags & DMU_BACKUP_FEATURE_EMBED_DATA) && !spa_feature_is_enabled(dp->dp_spa, SPA_FEATURE_EMBEDDED_DATA)) return (SET_ERROR(ENOTSUP)); if ((featureflags & DMU_BACKUP_FEATURE_LZ4) && !spa_feature_is_enabled(dp->dp_spa, SPA_FEATURE_LZ4_COMPRESS)) return (SET_ERROR(ENOTSUP)); /* * The receiving code doesn't know how to translate large blocks * to smaller ones, so the pool must have the LARGE_BLOCKS * feature enabled if the stream has LARGE_BLOCKS. Same with * large dnodes. */ if ((featureflags & DMU_BACKUP_FEATURE_LARGE_BLOCKS) && !spa_feature_is_enabled(dp->dp_spa, SPA_FEATURE_LARGE_BLOCKS)) return (SET_ERROR(ENOTSUP)); if ((featureflags & DMU_BACKUP_FEATURE_LARGE_DNODE) && !spa_feature_is_enabled(dp->dp_spa, SPA_FEATURE_LARGE_DNODE)) return (SET_ERROR(ENOTSUP)); (void) snprintf(recvname, sizeof (recvname), "%s/%s", tofs, recv_clone_name); if (dsl_dataset_hold(dp, recvname, FTAG, &ds) != 0) { /* %recv does not exist; continue in tofs */ error = dsl_dataset_hold(dp, tofs, FTAG, &ds); if (error != 0) return (error); } /* check that ds is marked inconsistent */ if (!DS_IS_INCONSISTENT(ds)) { dsl_dataset_rele(ds, FTAG); return (SET_ERROR(EINVAL)); } /* check that there is resuming data, and that the toguid matches */ if (!dsl_dataset_is_zapified(ds)) { dsl_dataset_rele(ds, FTAG); return (SET_ERROR(EINVAL)); } uint64_t val; error = zap_lookup(dp->dp_meta_objset, ds->ds_object, DS_FIELD_RESUME_TOGUID, sizeof (val), 1, &val); if (error != 0 || drrb->drr_toguid != val) { dsl_dataset_rele(ds, FTAG); return (SET_ERROR(EINVAL)); } /* * Check if the receive is still running. If so, it will be owned. * Note that nothing else can own the dataset (e.g. after the receive * fails) because it will be marked inconsistent. */ if (dsl_dataset_has_owner(ds)) { dsl_dataset_rele(ds, FTAG); return (SET_ERROR(EBUSY)); } /* There should not be any snapshots of this fs yet. */ if (ds->ds_prev != NULL && ds->ds_prev->ds_dir == ds->ds_dir) { dsl_dataset_rele(ds, FTAG); return (SET_ERROR(EINVAL)); } /* * Note: resume point will be checked when we process the first WRITE * record. */ /* check that the origin matches */ val = 0; (void) zap_lookup(dp->dp_meta_objset, ds->ds_object, DS_FIELD_RESUME_FROMGUID, sizeof (val), 1, &val); if (drrb->drr_fromguid != val) { dsl_dataset_rele(ds, FTAG); return (SET_ERROR(EINVAL)); } dsl_dataset_rele(ds, FTAG); return (0); } static void dmu_recv_resume_begin_sync(void *arg, dmu_tx_t *tx) { dmu_recv_begin_arg_t *drba = arg; dsl_pool_t *dp = dmu_tx_pool(tx); const char *tofs = drba->drba_cookie->drc_tofs; dsl_dataset_t *ds; uint64_t dsobj; /* 6 extra bytes for /%recv */ char recvname[ZFS_MAX_DATASET_NAME_LEN + 6]; (void) snprintf(recvname, sizeof (recvname), "%s/%s", tofs, recv_clone_name); if (dsl_dataset_hold(dp, recvname, FTAG, &ds) != 0) { /* %recv does not exist; continue in tofs */ VERIFY0(dsl_dataset_hold(dp, tofs, FTAG, &ds)); drba->drba_cookie->drc_newfs = B_TRUE; } /* clear the inconsistent flag so that we can own it */ ASSERT(DS_IS_INCONSISTENT(ds)); dmu_buf_will_dirty(ds->ds_dbuf, tx); dsl_dataset_phys(ds)->ds_flags &= ~DS_FLAG_INCONSISTENT; dsobj = ds->ds_object; dsl_dataset_rele(ds, FTAG); VERIFY0(dsl_dataset_own_obj(dp, dsobj, dmu_recv_tag, &ds)); dmu_buf_will_dirty(ds->ds_dbuf, tx); dsl_dataset_phys(ds)->ds_flags |= DS_FLAG_INCONSISTENT; rrw_enter(&ds->ds_bp_rwlock, RW_READER, FTAG); ASSERT(!BP_IS_HOLE(dsl_dataset_get_blkptr(ds))); rrw_exit(&ds->ds_bp_rwlock, FTAG); drba->drba_cookie->drc_ds = ds; spa_history_log_internal_ds(ds, "resume receive", tx, ""); } /* * NB: callers *MUST* call dmu_recv_stream() if dmu_recv_begin() * succeeds; otherwise we will leak the holds on the datasets. */ int dmu_recv_begin(char *tofs, char *tosnap, dmu_replay_record_t *drr_begin, boolean_t force, boolean_t resumable, char *origin, dmu_recv_cookie_t *drc) { dmu_recv_begin_arg_t drba = { 0 }; bzero(drc, sizeof (dmu_recv_cookie_t)); drc->drc_drr_begin = drr_begin; drc->drc_drrb = &drr_begin->drr_u.drr_begin; drc->drc_tosnap = tosnap; drc->drc_tofs = tofs; drc->drc_force = force; drc->drc_resumable = resumable; drc->drc_cred = CRED(); drc->drc_clone = (origin != NULL); if (drc->drc_drrb->drr_magic == BSWAP_64(DMU_BACKUP_MAGIC)) { drc->drc_byteswap = B_TRUE; (void) fletcher_4_incremental_byteswap(drr_begin, sizeof (dmu_replay_record_t), &drc->drc_cksum); byteswap_record(drr_begin); } else if (drc->drc_drrb->drr_magic == DMU_BACKUP_MAGIC) { (void) fletcher_4_incremental_native(drr_begin, sizeof (dmu_replay_record_t), &drc->drc_cksum); } else { return (SET_ERROR(EINVAL)); } drba.drba_origin = origin; drba.drba_cookie = drc; drba.drba_cred = CRED(); if (DMU_GET_FEATUREFLAGS(drc->drc_drrb->drr_versioninfo) & DMU_BACKUP_FEATURE_RESUMING) { return (dsl_sync_task(tofs, dmu_recv_resume_begin_check, dmu_recv_resume_begin_sync, &drba, 5, ZFS_SPACE_CHECK_NORMAL)); } else { return (dsl_sync_task(tofs, dmu_recv_begin_check, dmu_recv_begin_sync, &drba, 5, ZFS_SPACE_CHECK_NORMAL)); } } struct receive_record_arg { dmu_replay_record_t header; void *payload; /* Pointer to a buffer containing the payload */ /* * If the record is a write, pointer to the arc_buf_t containing the * payload. */ arc_buf_t *write_buf; int payload_size; uint64_t bytes_read; /* bytes read from stream when record created */ boolean_t eos_marker; /* Marks the end of the stream */ bqueue_node_t node; }; struct receive_writer_arg { objset_t *os; boolean_t byteswap; bqueue_t q; /* * These three args are used to signal to the main thread that we're * done. */ kmutex_t mutex; kcondvar_t cv; boolean_t done; int err; /* A map from guid to dataset to help handle dedup'd streams. */ avl_tree_t *guid_to_ds_map; boolean_t resumable; uint64_t last_object; uint64_t last_offset; uint64_t max_object; /* highest object ID referenced in stream */ uint64_t bytes_read; /* bytes read when current record created */ }; struct objlist { list_t list; /* List of struct receive_objnode. */ /* * Last object looked up. Used to assert that objects are being looked * up in ascending order. */ uint64_t last_lookup; }; struct receive_objnode { list_node_t node; uint64_t object; }; struct receive_arg { objset_t *os; kthread_t *td; struct file *fp; uint64_t voff; /* The current offset in the stream */ uint64_t bytes_read; /* * A record that has had its payload read in, but hasn't yet been handed * off to the worker thread. */ struct receive_record_arg *rrd; /* A record that has had its header read in, but not its payload. */ struct receive_record_arg *next_rrd; zio_cksum_t cksum; zio_cksum_t prev_cksum; int err; boolean_t byteswap; /* Sorted list of objects not to issue prefetches for. */ struct objlist ignore_objlist; }; typedef struct guid_map_entry { uint64_t guid; dsl_dataset_t *gme_ds; avl_node_t avlnode; } guid_map_entry_t; static int guid_compare(const void *arg1, const void *arg2) { const guid_map_entry_t *gmep1 = (const guid_map_entry_t *)arg1; const guid_map_entry_t *gmep2 = (const guid_map_entry_t *)arg2; return (AVL_CMP(gmep1->guid, gmep2->guid)); } static void free_guid_map_onexit(void *arg) { avl_tree_t *ca = arg; void *cookie = NULL; guid_map_entry_t *gmep; while ((gmep = avl_destroy_nodes(ca, &cookie)) != NULL) { dsl_dataset_long_rele(gmep->gme_ds, gmep); dsl_dataset_rele(gmep->gme_ds, gmep); kmem_free(gmep, sizeof (guid_map_entry_t)); } avl_destroy(ca); kmem_free(ca, sizeof (avl_tree_t)); } static int restore_bytes(struct receive_arg *ra, void *buf, int len, off_t off, ssize_t *resid) { struct uio auio; struct iovec aiov; int error; aiov.iov_base = buf; aiov.iov_len = len; auio.uio_iov = &aiov; auio.uio_iovcnt = 1; auio.uio_resid = len; auio.uio_segflg = UIO_SYSSPACE; auio.uio_rw = UIO_READ; auio.uio_offset = off; auio.uio_td = ra->td; #ifdef _KERNEL error = fo_read(ra->fp, &auio, ra->td->td_ucred, FOF_OFFSET, ra->td); #else fprintf(stderr, "%s: returning EOPNOTSUPP\n", __func__); error = EOPNOTSUPP; #endif *resid = auio.uio_resid; return (error); } static int receive_read(struct receive_arg *ra, int len, void *buf) { int done = 0; /* * The code doesn't rely on this (lengths being multiples of 8). See * comment in dump_bytes. */ ASSERT0(len % 8); while (done < len) { ssize_t resid; ra->err = restore_bytes(ra, buf + done, len - done, ra->voff, &resid); if (resid == len - done) { /* * Note: ECKSUM indicates that the receive * was interrupted and can potentially be resumed. */ ra->err = SET_ERROR(ECKSUM); } ra->voff += len - done - resid; done = len - resid; if (ra->err != 0) return (ra->err); } ra->bytes_read += len; ASSERT3U(done, ==, len); return (0); } noinline static void byteswap_record(dmu_replay_record_t *drr) { #define DO64(X) (drr->drr_u.X = BSWAP_64(drr->drr_u.X)) #define DO32(X) (drr->drr_u.X = BSWAP_32(drr->drr_u.X)) drr->drr_type = BSWAP_32(drr->drr_type); drr->drr_payloadlen = BSWAP_32(drr->drr_payloadlen); switch (drr->drr_type) { case DRR_BEGIN: DO64(drr_begin.drr_magic); DO64(drr_begin.drr_versioninfo); DO64(drr_begin.drr_creation_time); DO32(drr_begin.drr_type); DO32(drr_begin.drr_flags); DO64(drr_begin.drr_toguid); DO64(drr_begin.drr_fromguid); break; case DRR_OBJECT: DO64(drr_object.drr_object); DO32(drr_object.drr_type); DO32(drr_object.drr_bonustype); DO32(drr_object.drr_blksz); DO32(drr_object.drr_bonuslen); DO64(drr_object.drr_toguid); break; case DRR_FREEOBJECTS: DO64(drr_freeobjects.drr_firstobj); DO64(drr_freeobjects.drr_numobjs); DO64(drr_freeobjects.drr_toguid); break; case DRR_WRITE: DO64(drr_write.drr_object); DO32(drr_write.drr_type); DO64(drr_write.drr_offset); DO64(drr_write.drr_logical_size); DO64(drr_write.drr_toguid); ZIO_CHECKSUM_BSWAP(&drr->drr_u.drr_write.drr_key.ddk_cksum); DO64(drr_write.drr_key.ddk_prop); DO64(drr_write.drr_compressed_size); break; case DRR_WRITE_BYREF: DO64(drr_write_byref.drr_object); DO64(drr_write_byref.drr_offset); DO64(drr_write_byref.drr_length); DO64(drr_write_byref.drr_toguid); DO64(drr_write_byref.drr_refguid); DO64(drr_write_byref.drr_refobject); DO64(drr_write_byref.drr_refoffset); ZIO_CHECKSUM_BSWAP(&drr->drr_u.drr_write_byref. drr_key.ddk_cksum); DO64(drr_write_byref.drr_key.ddk_prop); break; case DRR_WRITE_EMBEDDED: DO64(drr_write_embedded.drr_object); DO64(drr_write_embedded.drr_offset); DO64(drr_write_embedded.drr_length); DO64(drr_write_embedded.drr_toguid); DO32(drr_write_embedded.drr_lsize); DO32(drr_write_embedded.drr_psize); break; case DRR_FREE: DO64(drr_free.drr_object); DO64(drr_free.drr_offset); DO64(drr_free.drr_length); DO64(drr_free.drr_toguid); break; case DRR_SPILL: DO64(drr_spill.drr_object); DO64(drr_spill.drr_length); DO64(drr_spill.drr_toguid); break; case DRR_END: DO64(drr_end.drr_toguid); ZIO_CHECKSUM_BSWAP(&drr->drr_u.drr_end.drr_checksum); break; } if (drr->drr_type != DRR_BEGIN) { ZIO_CHECKSUM_BSWAP(&drr->drr_u.drr_checksum.drr_checksum); } #undef DO64 #undef DO32 } static inline uint8_t deduce_nblkptr(dmu_object_type_t bonus_type, uint64_t bonus_size) { if (bonus_type == DMU_OT_SA) { return (1); } else { return (1 + ((DN_OLD_MAX_BONUSLEN - MIN(DN_OLD_MAX_BONUSLEN, bonus_size)) >> SPA_BLKPTRSHIFT)); } } static void save_resume_state(struct receive_writer_arg *rwa, uint64_t object, uint64_t offset, dmu_tx_t *tx) { int txgoff = dmu_tx_get_txg(tx) & TXG_MASK; if (!rwa->resumable) return; /* * We use ds_resume_bytes[] != 0 to indicate that we need to * update this on disk, so it must not be 0. */ ASSERT(rwa->bytes_read != 0); /* * We only resume from write records, which have a valid * (non-meta-dnode) object number. */ ASSERT(object != 0); /* * For resuming to work correctly, we must receive records in order, * sorted by object,offset. This is checked by the callers, but * assert it here for good measure. */ ASSERT3U(object, >=, rwa->os->os_dsl_dataset->ds_resume_object[txgoff]); ASSERT(object != rwa->os->os_dsl_dataset->ds_resume_object[txgoff] || offset >= rwa->os->os_dsl_dataset->ds_resume_offset[txgoff]); ASSERT3U(rwa->bytes_read, >=, rwa->os->os_dsl_dataset->ds_resume_bytes[txgoff]); rwa->os->os_dsl_dataset->ds_resume_object[txgoff] = object; rwa->os->os_dsl_dataset->ds_resume_offset[txgoff] = offset; rwa->os->os_dsl_dataset->ds_resume_bytes[txgoff] = rwa->bytes_read; } noinline static int receive_object(struct receive_writer_arg *rwa, struct drr_object *drro, void *data) { dmu_object_info_t doi; dmu_tx_t *tx; uint64_t object; int err; uint8_t dn_slots = drro->drr_dn_slots != 0 ? drro->drr_dn_slots : DNODE_MIN_SLOTS; if (drro->drr_type == DMU_OT_NONE || !DMU_OT_IS_VALID(drro->drr_type) || !DMU_OT_IS_VALID(drro->drr_bonustype) || drro->drr_checksumtype >= ZIO_CHECKSUM_FUNCTIONS || drro->drr_compress >= ZIO_COMPRESS_FUNCTIONS || P2PHASE(drro->drr_blksz, SPA_MINBLOCKSIZE) || drro->drr_blksz < SPA_MINBLOCKSIZE || drro->drr_blksz > spa_maxblocksize(dmu_objset_spa(rwa->os)) || drro->drr_bonuslen > DN_BONUS_SIZE(spa_maxdnodesize(dmu_objset_spa(rwa->os))) || dn_slots > (spa_maxdnodesize(dmu_objset_spa(rwa->os)) >> DNODE_SHIFT)) { return (SET_ERROR(EINVAL)); } err = dmu_object_info(rwa->os, drro->drr_object, &doi); if (err != 0 && err != ENOENT && err != EEXIST) return (SET_ERROR(EINVAL)); if (drro->drr_object > rwa->max_object) rwa->max_object = drro->drr_object; /* * If we are losing blkptrs or changing the block size this must * be a new file instance. We must clear out the previous file * contents before we can change this type of metadata in the dnode. */ if (err == 0) { int nblkptr; object = drro->drr_object; nblkptr = deduce_nblkptr(drro->drr_bonustype, drro->drr_bonuslen); if (drro->drr_blksz != doi.doi_data_block_size || nblkptr < doi.doi_nblkptr || dn_slots != doi.doi_dnodesize >> DNODE_SHIFT) { err = dmu_free_long_range(rwa->os, drro->drr_object, 0, DMU_OBJECT_END); if (err != 0) return (SET_ERROR(EINVAL)); } } else if (err == EEXIST) { /* * The object requested is currently an interior slot of a * multi-slot dnode. This will be resolved when the next txg * is synced out, since the send stream will have told us * to free this slot when we freed the associated dnode * earlier in the stream. */ txg_wait_synced(dmu_objset_pool(rwa->os), 0); object = drro->drr_object; } else { /* object is free and we are about to allocate a new one */ object = DMU_NEW_OBJECT; } /* * If this is a multi-slot dnode there is a chance that this * object will expand into a slot that is already used by * another object from the previous snapshot. We must free * these objects before we attempt to allocate the new dnode. */ if (dn_slots > 1) { boolean_t need_sync = B_FALSE; for (uint64_t slot = drro->drr_object + 1; slot < drro->drr_object + dn_slots; slot++) { dmu_object_info_t slot_doi; err = dmu_object_info(rwa->os, slot, &slot_doi); if (err == ENOENT || err == EEXIST) continue; else if (err != 0) return (err); err = dmu_free_long_object(rwa->os, slot); if (err != 0) return (err); need_sync = B_TRUE; } if (need_sync) txg_wait_synced(dmu_objset_pool(rwa->os), 0); } tx = dmu_tx_create(rwa->os); dmu_tx_hold_bonus(tx, object); err = dmu_tx_assign(tx, TXG_WAIT); if (err != 0) { dmu_tx_abort(tx); return (err); } if (object == DMU_NEW_OBJECT) { /* currently free, want to be allocated */ err = dmu_object_claim_dnsize(rwa->os, drro->drr_object, drro->drr_type, drro->drr_blksz, drro->drr_bonustype, drro->drr_bonuslen, dn_slots << DNODE_SHIFT, tx); } else if (drro->drr_type != doi.doi_type || drro->drr_blksz != doi.doi_data_block_size || drro->drr_bonustype != doi.doi_bonus_type || - drro->drr_bonuslen != doi.doi_bonus_size) { + drro->drr_bonuslen != doi.doi_bonus_size || + drro->drr_dn_slots != (doi.doi_dnodesize >> DNODE_SHIFT)) { /* currently allocated, but with different properties */ - err = dmu_object_reclaim(rwa->os, drro->drr_object, + err = dmu_object_reclaim_dnsize(rwa->os, drro->drr_object, drro->drr_type, drro->drr_blksz, - drro->drr_bonustype, drro->drr_bonuslen, tx); + drro->drr_bonustype, drro->drr_bonuslen, + drro->drr_dn_slots << DNODE_SHIFT, tx); } if (err != 0) { dmu_tx_commit(tx); return (SET_ERROR(EINVAL)); } dmu_object_set_checksum(rwa->os, drro->drr_object, drro->drr_checksumtype, tx); dmu_object_set_compress(rwa->os, drro->drr_object, drro->drr_compress, tx); if (data != NULL) { dmu_buf_t *db; VERIFY0(dmu_bonus_hold(rwa->os, drro->drr_object, FTAG, &db)); dmu_buf_will_dirty(db, tx); ASSERT3U(db->db_size, >=, drro->drr_bonuslen); bcopy(data, db->db_data, drro->drr_bonuslen); if (rwa->byteswap) { dmu_object_byteswap_t byteswap = DMU_OT_BYTESWAP(drro->drr_bonustype); dmu_ot_byteswap[byteswap].ob_func(db->db_data, drro->drr_bonuslen); } dmu_buf_rele(db, FTAG); } dmu_tx_commit(tx); return (0); } /* ARGSUSED */ noinline static int receive_freeobjects(struct receive_writer_arg *rwa, struct drr_freeobjects *drrfo) { uint64_t obj; int next_err = 0; if (drrfo->drr_firstobj + drrfo->drr_numobjs < drrfo->drr_firstobj) return (SET_ERROR(EINVAL)); for (obj = drrfo->drr_firstobj == 0 ? 1 : drrfo->drr_firstobj; obj < drrfo->drr_firstobj + drrfo->drr_numobjs && next_err == 0; next_err = dmu_object_next(rwa->os, &obj, FALSE, 0)) { dmu_object_info_t doi; int err; err = dmu_object_info(rwa->os, obj, NULL); - if (err == ENOENT) { - obj++; + if (err == ENOENT) continue; - } else if (err != 0) { + else if (err != 0) return (err); - } err = dmu_free_long_object(rwa->os, obj); if (err != 0) return (err); if (obj > rwa->max_object) rwa->max_object = obj; } if (next_err != ESRCH) return (next_err); return (0); } noinline static int receive_write(struct receive_writer_arg *rwa, struct drr_write *drrw, arc_buf_t *abuf) { dmu_tx_t *tx; int err; if (drrw->drr_offset + drrw->drr_logical_size < drrw->drr_offset || !DMU_OT_IS_VALID(drrw->drr_type)) return (SET_ERROR(EINVAL)); /* * For resuming to work, records must be in increasing order * by (object, offset). */ if (drrw->drr_object < rwa->last_object || (drrw->drr_object == rwa->last_object && drrw->drr_offset < rwa->last_offset)) { return (SET_ERROR(EINVAL)); } rwa->last_object = drrw->drr_object; rwa->last_offset = drrw->drr_offset; if (rwa->last_object > rwa->max_object) rwa->max_object = rwa->last_object; if (dmu_object_info(rwa->os, drrw->drr_object, NULL) != 0) return (SET_ERROR(EINVAL)); tx = dmu_tx_create(rwa->os); dmu_tx_hold_write(tx, drrw->drr_object, drrw->drr_offset, drrw->drr_logical_size); err = dmu_tx_assign(tx, TXG_WAIT); if (err != 0) { dmu_tx_abort(tx); return (err); } if (rwa->byteswap) { dmu_object_byteswap_t byteswap = DMU_OT_BYTESWAP(drrw->drr_type); dmu_ot_byteswap[byteswap].ob_func(abuf->b_data, DRR_WRITE_PAYLOAD_SIZE(drrw)); } /* use the bonus buf to look up the dnode in dmu_assign_arcbuf */ dmu_buf_t *bonus; if (dmu_bonus_hold(rwa->os, drrw->drr_object, FTAG, &bonus) != 0) return (SET_ERROR(EINVAL)); dmu_assign_arcbuf(bonus, drrw->drr_offset, abuf, tx); /* * Note: If the receive fails, we want the resume stream to start * with the same record that we last successfully received (as opposed * to the next record), so that we can verify that we are * resuming from the correct location. */ save_resume_state(rwa, drrw->drr_object, drrw->drr_offset, tx); dmu_tx_commit(tx); dmu_buf_rele(bonus, FTAG); return (0); } /* * Handle a DRR_WRITE_BYREF record. This record is used in dedup'ed * streams to refer to a copy of the data that is already on the * system because it came in earlier in the stream. This function * finds the earlier copy of the data, and uses that copy instead of * data from the stream to fulfill this write. */ static int receive_write_byref(struct receive_writer_arg *rwa, struct drr_write_byref *drrwbr) { dmu_tx_t *tx; int err; guid_map_entry_t gmesrch; guid_map_entry_t *gmep; avl_index_t where; objset_t *ref_os = NULL; dmu_buf_t *dbp; if (drrwbr->drr_offset + drrwbr->drr_length < drrwbr->drr_offset) return (SET_ERROR(EINVAL)); /* * If the GUID of the referenced dataset is different from the * GUID of the target dataset, find the referenced dataset. */ if (drrwbr->drr_toguid != drrwbr->drr_refguid) { gmesrch.guid = drrwbr->drr_refguid; if ((gmep = avl_find(rwa->guid_to_ds_map, &gmesrch, &where)) == NULL) { return (SET_ERROR(EINVAL)); } if (dmu_objset_from_ds(gmep->gme_ds, &ref_os)) return (SET_ERROR(EINVAL)); } else { ref_os = rwa->os; } if (drrwbr->drr_object > rwa->max_object) rwa->max_object = drrwbr->drr_object; err = dmu_buf_hold(ref_os, drrwbr->drr_refobject, drrwbr->drr_refoffset, FTAG, &dbp, DMU_READ_PREFETCH); if (err != 0) return (err); tx = dmu_tx_create(rwa->os); dmu_tx_hold_write(tx, drrwbr->drr_object, drrwbr->drr_offset, drrwbr->drr_length); err = dmu_tx_assign(tx, TXG_WAIT); if (err != 0) { dmu_tx_abort(tx); return (err); } dmu_write(rwa->os, drrwbr->drr_object, drrwbr->drr_offset, drrwbr->drr_length, dbp->db_data, tx); dmu_buf_rele(dbp, FTAG); /* See comment in restore_write. */ save_resume_state(rwa, drrwbr->drr_object, drrwbr->drr_offset, tx); dmu_tx_commit(tx); return (0); } static int receive_write_embedded(struct receive_writer_arg *rwa, struct drr_write_embedded *drrwe, void *data) { dmu_tx_t *tx; int err; if (drrwe->drr_offset + drrwe->drr_length < drrwe->drr_offset) return (EINVAL); if (drrwe->drr_psize > BPE_PAYLOAD_SIZE) return (EINVAL); if (drrwe->drr_etype >= NUM_BP_EMBEDDED_TYPES) return (EINVAL); if (drrwe->drr_compression >= ZIO_COMPRESS_FUNCTIONS) return (EINVAL); if (drrwe->drr_object > rwa->max_object) rwa->max_object = drrwe->drr_object; tx = dmu_tx_create(rwa->os); dmu_tx_hold_write(tx, drrwe->drr_object, drrwe->drr_offset, drrwe->drr_length); err = dmu_tx_assign(tx, TXG_WAIT); if (err != 0) { dmu_tx_abort(tx); return (err); } dmu_write_embedded(rwa->os, drrwe->drr_object, drrwe->drr_offset, data, drrwe->drr_etype, drrwe->drr_compression, drrwe->drr_lsize, drrwe->drr_psize, rwa->byteswap ^ ZFS_HOST_BYTEORDER, tx); /* See comment in restore_write. */ save_resume_state(rwa, drrwe->drr_object, drrwe->drr_offset, tx); dmu_tx_commit(tx); return (0); } static int receive_spill(struct receive_writer_arg *rwa, struct drr_spill *drrs, void *data) { dmu_tx_t *tx; dmu_buf_t *db, *db_spill; int err; if (drrs->drr_length < SPA_MINBLOCKSIZE || drrs->drr_length > spa_maxblocksize(dmu_objset_spa(rwa->os))) return (SET_ERROR(EINVAL)); if (dmu_object_info(rwa->os, drrs->drr_object, NULL) != 0) return (SET_ERROR(EINVAL)); if (drrs->drr_object > rwa->max_object) rwa->max_object = drrs->drr_object; VERIFY0(dmu_bonus_hold(rwa->os, drrs->drr_object, FTAG, &db)); if ((err = dmu_spill_hold_by_bonus(db, FTAG, &db_spill)) != 0) { dmu_buf_rele(db, FTAG); return (err); } tx = dmu_tx_create(rwa->os); dmu_tx_hold_spill(tx, db->db_object); err = dmu_tx_assign(tx, TXG_WAIT); if (err != 0) { dmu_buf_rele(db, FTAG); dmu_buf_rele(db_spill, FTAG); dmu_tx_abort(tx); return (err); } dmu_buf_will_dirty(db_spill, tx); if (db_spill->db_size < drrs->drr_length) VERIFY(0 == dbuf_spill_set_blksz(db_spill, drrs->drr_length, tx)); bcopy(data, db_spill->db_data, drrs->drr_length); dmu_buf_rele(db, FTAG); dmu_buf_rele(db_spill, FTAG); dmu_tx_commit(tx); return (0); } /* ARGSUSED */ noinline static int receive_free(struct receive_writer_arg *rwa, struct drr_free *drrf) { int err; if (drrf->drr_length != -1ULL && drrf->drr_offset + drrf->drr_length < drrf->drr_offset) return (SET_ERROR(EINVAL)); if (dmu_object_info(rwa->os, drrf->drr_object, NULL) != 0) return (SET_ERROR(EINVAL)); if (drrf->drr_object > rwa->max_object) rwa->max_object = drrf->drr_object; err = dmu_free_long_range(rwa->os, drrf->drr_object, drrf->drr_offset, drrf->drr_length); return (err); } /* used to destroy the drc_ds on error */ static void dmu_recv_cleanup_ds(dmu_recv_cookie_t *drc) { if (drc->drc_resumable) { /* wait for our resume state to be written to disk */ txg_wait_synced(drc->drc_ds->ds_dir->dd_pool, 0); dsl_dataset_disown(drc->drc_ds, dmu_recv_tag); } else { char name[ZFS_MAX_DATASET_NAME_LEN]; dsl_dataset_name(drc->drc_ds, name); dsl_dataset_disown(drc->drc_ds, dmu_recv_tag); (void) dsl_destroy_head(name); } } static void receive_cksum(struct receive_arg *ra, int len, void *buf) { if (ra->byteswap) { (void) fletcher_4_incremental_byteswap(buf, len, &ra->cksum); } else { (void) fletcher_4_incremental_native(buf, len, &ra->cksum); } } /* * Read the payload into a buffer of size len, and update the current record's * payload field. * Allocate ra->next_rrd and read the next record's header into * ra->next_rrd->header. * Verify checksum of payload and next record. */ static int receive_read_payload_and_next_header(struct receive_arg *ra, int len, void *buf) { int err; if (len != 0) { ASSERT3U(len, <=, SPA_MAXBLOCKSIZE); err = receive_read(ra, len, buf); if (err != 0) return (err); receive_cksum(ra, len, buf); /* note: rrd is NULL when reading the begin record's payload */ if (ra->rrd != NULL) { ra->rrd->payload = buf; ra->rrd->payload_size = len; ra->rrd->bytes_read = ra->bytes_read; } } ra->prev_cksum = ra->cksum; ra->next_rrd = kmem_zalloc(sizeof (*ra->next_rrd), KM_SLEEP); err = receive_read(ra, sizeof (ra->next_rrd->header), &ra->next_rrd->header); ra->next_rrd->bytes_read = ra->bytes_read; if (err != 0) { kmem_free(ra->next_rrd, sizeof (*ra->next_rrd)); ra->next_rrd = NULL; return (err); } if (ra->next_rrd->header.drr_type == DRR_BEGIN) { kmem_free(ra->next_rrd, sizeof (*ra->next_rrd)); ra->next_rrd = NULL; return (SET_ERROR(EINVAL)); } /* * Note: checksum is of everything up to but not including the * checksum itself. */ ASSERT3U(offsetof(dmu_replay_record_t, drr_u.drr_checksum.drr_checksum), ==, sizeof (dmu_replay_record_t) - sizeof (zio_cksum_t)); receive_cksum(ra, offsetof(dmu_replay_record_t, drr_u.drr_checksum.drr_checksum), &ra->next_rrd->header); zio_cksum_t cksum_orig = ra->next_rrd->header.drr_u.drr_checksum.drr_checksum; zio_cksum_t *cksump = &ra->next_rrd->header.drr_u.drr_checksum.drr_checksum; if (ra->byteswap) byteswap_record(&ra->next_rrd->header); if ((!ZIO_CHECKSUM_IS_ZERO(cksump)) && !ZIO_CHECKSUM_EQUAL(ra->cksum, *cksump)) { kmem_free(ra->next_rrd, sizeof (*ra->next_rrd)); ra->next_rrd = NULL; return (SET_ERROR(ECKSUM)); } receive_cksum(ra, sizeof (cksum_orig), &cksum_orig); return (0); } static void objlist_create(struct objlist *list) { list_create(&list->list, sizeof (struct receive_objnode), offsetof(struct receive_objnode, node)); list->last_lookup = 0; } static void objlist_destroy(struct objlist *list) { for (struct receive_objnode *n = list_remove_head(&list->list); n != NULL; n = list_remove_head(&list->list)) { kmem_free(n, sizeof (*n)); } list_destroy(&list->list); } /* * This function looks through the objlist to see if the specified object number * is contained in the objlist. In the process, it will remove all object * numbers in the list that are smaller than the specified object number. Thus, * any lookup of an object number smaller than a previously looked up object * number will always return false; therefore, all lookups should be done in * ascending order. */ static boolean_t objlist_exists(struct objlist *list, uint64_t object) { struct receive_objnode *node = list_head(&list->list); ASSERT3U(object, >=, list->last_lookup); list->last_lookup = object; while (node != NULL && node->object < object) { VERIFY3P(node, ==, list_remove_head(&list->list)); kmem_free(node, sizeof (*node)); node = list_head(&list->list); } return (node != NULL && node->object == object); } /* * The objlist is a list of object numbers stored in ascending order. However, * the insertion of new object numbers does not seek out the correct location to * store a new object number; instead, it appends it to the list for simplicity. * Thus, any users must take care to only insert new object numbers in ascending * order. */ static void objlist_insert(struct objlist *list, uint64_t object) { struct receive_objnode *node = kmem_zalloc(sizeof (*node), KM_SLEEP); node->object = object; #ifdef ZFS_DEBUG struct receive_objnode *last_object = list_tail(&list->list); uint64_t last_objnum = (last_object != NULL ? last_object->object : 0); ASSERT3U(node->object, >, last_objnum); #endif list_insert_tail(&list->list, node); } /* * Issue the prefetch reads for any necessary indirect blocks. * * We use the object ignore list to tell us whether or not to issue prefetches * for a given object. We do this for both correctness (in case the blocksize * of an object has changed) and performance (if the object doesn't exist, don't * needlessly try to issue prefetches). We also trim the list as we go through * the stream to prevent it from growing to an unbounded size. * * The object numbers within will always be in sorted order, and any write * records we see will also be in sorted order, but they're not sorted with * respect to each other (i.e. we can get several object records before * receiving each object's write records). As a result, once we've reached a * given object number, we can safely remove any reference to lower object * numbers in the ignore list. In practice, we receive up to 32 object records * before receiving write records, so the list can have up to 32 nodes in it. */ /* ARGSUSED */ static void receive_read_prefetch(struct receive_arg *ra, uint64_t object, uint64_t offset, uint64_t length) { if (!objlist_exists(&ra->ignore_objlist, object)) { dmu_prefetch(ra->os, object, 1, offset, length, ZIO_PRIORITY_SYNC_READ); } } /* * Read records off the stream, issuing any necessary prefetches. */ static int receive_read_record(struct receive_arg *ra) { int err; switch (ra->rrd->header.drr_type) { case DRR_OBJECT: { struct drr_object *drro = &ra->rrd->header.drr_u.drr_object; uint32_t size = P2ROUNDUP(drro->drr_bonuslen, 8); void *buf = kmem_zalloc(size, KM_SLEEP); dmu_object_info_t doi; err = receive_read_payload_and_next_header(ra, size, buf); if (err != 0) { kmem_free(buf, size); return (err); } err = dmu_object_info(ra->os, drro->drr_object, &doi); /* * See receive_read_prefetch for an explanation why we're * storing this object in the ignore_obj_list. */ if (err == ENOENT || (err == 0 && doi.doi_data_block_size != drro->drr_blksz)) { objlist_insert(&ra->ignore_objlist, drro->drr_object); err = 0; } return (err); } case DRR_FREEOBJECTS: { err = receive_read_payload_and_next_header(ra, 0, NULL); return (err); } case DRR_WRITE: { struct drr_write *drrw = &ra->rrd->header.drr_u.drr_write; arc_buf_t *abuf; boolean_t is_meta = DMU_OT_IS_METADATA(drrw->drr_type); if (DRR_WRITE_COMPRESSED(drrw)) { ASSERT3U(drrw->drr_compressed_size, >, 0); ASSERT3U(drrw->drr_logical_size, >=, drrw->drr_compressed_size); ASSERT(!is_meta); abuf = arc_loan_compressed_buf( dmu_objset_spa(ra->os), drrw->drr_compressed_size, drrw->drr_logical_size, drrw->drr_compressiontype); } else { abuf = arc_loan_buf(dmu_objset_spa(ra->os), is_meta, drrw->drr_logical_size); } err = receive_read_payload_and_next_header(ra, DRR_WRITE_PAYLOAD_SIZE(drrw), abuf->b_data); if (err != 0) { dmu_return_arcbuf(abuf); return (err); } ra->rrd->write_buf = abuf; receive_read_prefetch(ra, drrw->drr_object, drrw->drr_offset, drrw->drr_logical_size); return (err); } case DRR_WRITE_BYREF: { struct drr_write_byref *drrwb = &ra->rrd->header.drr_u.drr_write_byref; err = receive_read_payload_and_next_header(ra, 0, NULL); receive_read_prefetch(ra, drrwb->drr_object, drrwb->drr_offset, drrwb->drr_length); return (err); } case DRR_WRITE_EMBEDDED: { struct drr_write_embedded *drrwe = &ra->rrd->header.drr_u.drr_write_embedded; uint32_t size = P2ROUNDUP(drrwe->drr_psize, 8); void *buf = kmem_zalloc(size, KM_SLEEP); err = receive_read_payload_and_next_header(ra, size, buf); if (err != 0) { kmem_free(buf, size); return (err); } receive_read_prefetch(ra, drrwe->drr_object, drrwe->drr_offset, drrwe->drr_length); return (err); } case DRR_FREE: { /* * It might be beneficial to prefetch indirect blocks here, but * we don't really have the data to decide for sure. */ err = receive_read_payload_and_next_header(ra, 0, NULL); return (err); } case DRR_END: { struct drr_end *drre = &ra->rrd->header.drr_u.drr_end; if (!ZIO_CHECKSUM_EQUAL(ra->prev_cksum, drre->drr_checksum)) return (SET_ERROR(ECKSUM)); return (0); } case DRR_SPILL: { struct drr_spill *drrs = &ra->rrd->header.drr_u.drr_spill; void *buf = kmem_zalloc(drrs->drr_length, KM_SLEEP); err = receive_read_payload_and_next_header(ra, drrs->drr_length, buf); if (err != 0) kmem_free(buf, drrs->drr_length); return (err); } default: return (SET_ERROR(EINVAL)); } } /* * Commit the records to the pool. */ static int receive_process_record(struct receive_writer_arg *rwa, struct receive_record_arg *rrd) { int err; /* Processing in order, therefore bytes_read should be increasing. */ ASSERT3U(rrd->bytes_read, >=, rwa->bytes_read); rwa->bytes_read = rrd->bytes_read; switch (rrd->header.drr_type) { case DRR_OBJECT: { struct drr_object *drro = &rrd->header.drr_u.drr_object; err = receive_object(rwa, drro, rrd->payload); kmem_free(rrd->payload, rrd->payload_size); rrd->payload = NULL; return (err); } case DRR_FREEOBJECTS: { struct drr_freeobjects *drrfo = &rrd->header.drr_u.drr_freeobjects; return (receive_freeobjects(rwa, drrfo)); } case DRR_WRITE: { struct drr_write *drrw = &rrd->header.drr_u.drr_write; err = receive_write(rwa, drrw, rrd->write_buf); /* if receive_write() is successful, it consumes the arc_buf */ if (err != 0) dmu_return_arcbuf(rrd->write_buf); rrd->write_buf = NULL; rrd->payload = NULL; return (err); } case DRR_WRITE_BYREF: { struct drr_write_byref *drrwbr = &rrd->header.drr_u.drr_write_byref; return (receive_write_byref(rwa, drrwbr)); } case DRR_WRITE_EMBEDDED: { struct drr_write_embedded *drrwe = &rrd->header.drr_u.drr_write_embedded; err = receive_write_embedded(rwa, drrwe, rrd->payload); kmem_free(rrd->payload, rrd->payload_size); rrd->payload = NULL; return (err); } case DRR_FREE: { struct drr_free *drrf = &rrd->header.drr_u.drr_free; return (receive_free(rwa, drrf)); } case DRR_SPILL: { struct drr_spill *drrs = &rrd->header.drr_u.drr_spill; err = receive_spill(rwa, drrs, rrd->payload); kmem_free(rrd->payload, rrd->payload_size); rrd->payload = NULL; return (err); } default: return (SET_ERROR(EINVAL)); } } /* * dmu_recv_stream's worker thread; pull records off the queue, and then call * receive_process_record When we're done, signal the main thread and exit. */ static void receive_writer_thread(void *arg) { struct receive_writer_arg *rwa = arg; struct receive_record_arg *rrd; for (rrd = bqueue_dequeue(&rwa->q); !rrd->eos_marker; rrd = bqueue_dequeue(&rwa->q)) { /* * If there's an error, the main thread will stop putting things * on the queue, but we need to clear everything in it before we * can exit. */ if (rwa->err == 0) { rwa->err = receive_process_record(rwa, rrd); } else if (rrd->write_buf != NULL) { dmu_return_arcbuf(rrd->write_buf); rrd->write_buf = NULL; rrd->payload = NULL; } else if (rrd->payload != NULL) { kmem_free(rrd->payload, rrd->payload_size); rrd->payload = NULL; } kmem_free(rrd, sizeof (*rrd)); } kmem_free(rrd, sizeof (*rrd)); mutex_enter(&rwa->mutex); rwa->done = B_TRUE; cv_signal(&rwa->cv); mutex_exit(&rwa->mutex); thread_exit(); } static int resume_check(struct receive_arg *ra, nvlist_t *begin_nvl) { uint64_t val; objset_t *mos = dmu_objset_pool(ra->os)->dp_meta_objset; uint64_t dsobj = dmu_objset_id(ra->os); uint64_t resume_obj, resume_off; if (nvlist_lookup_uint64(begin_nvl, "resume_object", &resume_obj) != 0 || nvlist_lookup_uint64(begin_nvl, "resume_offset", &resume_off) != 0) { return (SET_ERROR(EINVAL)); } VERIFY0(zap_lookup(mos, dsobj, DS_FIELD_RESUME_OBJECT, sizeof (val), 1, &val)); if (resume_obj != val) return (SET_ERROR(EINVAL)); VERIFY0(zap_lookup(mos, dsobj, DS_FIELD_RESUME_OFFSET, sizeof (val), 1, &val)); if (resume_off != val) return (SET_ERROR(EINVAL)); return (0); } /* * Read in the stream's records, one by one, and apply them to the pool. There * are two threads involved; the thread that calls this function will spin up a * worker thread, read the records off the stream one by one, and issue * prefetches for any necessary indirect blocks. It will then push the records * onto an internal blocking queue. The worker thread will pull the records off * the queue, and actually write the data into the DMU. This way, the worker * thread doesn't have to wait for reads to complete, since everything it needs * (the indirect blocks) will be prefetched. * * NB: callers *must* call dmu_recv_end() if this succeeds. */ int dmu_recv_stream(dmu_recv_cookie_t *drc, struct file *fp, offset_t *voffp, int cleanup_fd, uint64_t *action_handlep) { int err = 0; struct receive_arg ra = { 0 }; struct receive_writer_arg rwa = { 0 }; int featureflags; nvlist_t *begin_nvl = NULL; ra.byteswap = drc->drc_byteswap; ra.cksum = drc->drc_cksum; ra.td = curthread; ra.fp = fp; ra.voff = *voffp; if (dsl_dataset_is_zapified(drc->drc_ds)) { (void) zap_lookup(drc->drc_ds->ds_dir->dd_pool->dp_meta_objset, drc->drc_ds->ds_object, DS_FIELD_RESUME_BYTES, sizeof (ra.bytes_read), 1, &ra.bytes_read); } objlist_create(&ra.ignore_objlist); /* these were verified in dmu_recv_begin */ ASSERT3U(DMU_GET_STREAM_HDRTYPE(drc->drc_drrb->drr_versioninfo), ==, DMU_SUBSTREAM); ASSERT3U(drc->drc_drrb->drr_type, <, DMU_OST_NUMTYPES); /* * Open the objset we are modifying. */ VERIFY0(dmu_objset_from_ds(drc->drc_ds, &ra.os)); ASSERT(dsl_dataset_phys(drc->drc_ds)->ds_flags & DS_FLAG_INCONSISTENT); featureflags = DMU_GET_FEATUREFLAGS(drc->drc_drrb->drr_versioninfo); /* if this stream is dedup'ed, set up the avl tree for guid mapping */ if (featureflags & DMU_BACKUP_FEATURE_DEDUP) { minor_t minor; if (cleanup_fd == -1) { ra.err = SET_ERROR(EBADF); goto out; } ra.err = zfs_onexit_fd_hold(cleanup_fd, &minor); if (ra.err != 0) { cleanup_fd = -1; goto out; } if (*action_handlep == 0) { rwa.guid_to_ds_map = kmem_alloc(sizeof (avl_tree_t), KM_SLEEP); avl_create(rwa.guid_to_ds_map, guid_compare, sizeof (guid_map_entry_t), offsetof(guid_map_entry_t, avlnode)); err = zfs_onexit_add_cb(minor, free_guid_map_onexit, rwa.guid_to_ds_map, action_handlep); if (ra.err != 0) goto out; } else { err = zfs_onexit_cb_data(minor, *action_handlep, (void **)&rwa.guid_to_ds_map); if (ra.err != 0) goto out; } drc->drc_guid_to_ds_map = rwa.guid_to_ds_map; } uint32_t payloadlen = drc->drc_drr_begin->drr_payloadlen; void *payload = NULL; if (payloadlen != 0) payload = kmem_alloc(payloadlen, KM_SLEEP); err = receive_read_payload_and_next_header(&ra, payloadlen, payload); if (err != 0) { if (payloadlen != 0) kmem_free(payload, payloadlen); goto out; } if (payloadlen != 0) { err = nvlist_unpack(payload, payloadlen, &begin_nvl, KM_SLEEP); kmem_free(payload, payloadlen); if (err != 0) goto out; } if (featureflags & DMU_BACKUP_FEATURE_RESUMING) { err = resume_check(&ra, begin_nvl); if (err != 0) goto out; } (void) bqueue_init(&rwa.q, zfs_recv_queue_length, offsetof(struct receive_record_arg, node)); cv_init(&rwa.cv, NULL, CV_DEFAULT, NULL); mutex_init(&rwa.mutex, NULL, MUTEX_DEFAULT, NULL); rwa.os = ra.os; rwa.byteswap = drc->drc_byteswap; rwa.resumable = drc->drc_resumable; (void) thread_create(NULL, 0, receive_writer_thread, &rwa, 0, &p0, TS_RUN, minclsyspri); /* * We're reading rwa.err without locks, which is safe since we are the * only reader, and the worker thread is the only writer. It's ok if we * miss a write for an iteration or two of the loop, since the writer * thread will keep freeing records we send it until we send it an eos * marker. * * We can leave this loop in 3 ways: First, if rwa.err is * non-zero. In that case, the writer thread will free the rrd we just * pushed. Second, if we're interrupted; in that case, either it's the * first loop and ra.rrd was never allocated, or it's later, and ra.rrd * has been handed off to the writer thread who will free it. Finally, * if receive_read_record fails or we're at the end of the stream, then * we free ra.rrd and exit. */ while (rwa.err == 0) { if (issig(JUSTLOOKING) && issig(FORREAL)) { err = SET_ERROR(EINTR); break; } ASSERT3P(ra.rrd, ==, NULL); ra.rrd = ra.next_rrd; ra.next_rrd = NULL; /* Allocates and loads header into ra.next_rrd */ err = receive_read_record(&ra); if (ra.rrd->header.drr_type == DRR_END || err != 0) { kmem_free(ra.rrd, sizeof (*ra.rrd)); ra.rrd = NULL; break; } bqueue_enqueue(&rwa.q, ra.rrd, sizeof (struct receive_record_arg) + ra.rrd->payload_size); ra.rrd = NULL; } if (ra.next_rrd == NULL) ra.next_rrd = kmem_zalloc(sizeof (*ra.next_rrd), KM_SLEEP); ra.next_rrd->eos_marker = B_TRUE; bqueue_enqueue(&rwa.q, ra.next_rrd, 1); mutex_enter(&rwa.mutex); while (!rwa.done) { cv_wait(&rwa.cv, &rwa.mutex); } mutex_exit(&rwa.mutex); /* * If we are receiving a full stream as a clone, all object IDs which * are greater than the maximum ID referenced in the stream are * by definition unused and must be freed. Note that it's possible that * we've resumed this send and the first record we received was the END * record. In that case, max_object would be 0, but we shouldn't start * freeing all objects from there; instead we should start from the * resumeobj. */ if (drc->drc_clone && drc->drc_drrb->drr_fromguid == 0) { uint64_t obj; if (nvlist_lookup_uint64(begin_nvl, "resume_object", &obj) != 0) obj = 0; if (rwa.max_object > obj) obj = rwa.max_object; obj++; int free_err = 0; int next_err = 0; while (next_err == 0) { free_err = dmu_free_long_object(rwa.os, obj); if (free_err != 0 && free_err != ENOENT) break; next_err = dmu_object_next(rwa.os, &obj, FALSE, 0); } if (err == 0) { if (free_err != 0 && free_err != ENOENT) err = free_err; else if (next_err != ESRCH) err = next_err; } } cv_destroy(&rwa.cv); mutex_destroy(&rwa.mutex); bqueue_destroy(&rwa.q); if (err == 0) err = rwa.err; out: nvlist_free(begin_nvl); if ((featureflags & DMU_BACKUP_FEATURE_DEDUP) && (cleanup_fd != -1)) zfs_onexit_fd_rele(cleanup_fd); if (err != 0) { /* * Clean up references. If receive is not resumable, * destroy what we created, so we don't leave it in * the inconsistent state. */ dmu_recv_cleanup_ds(drc); } *voffp = ra.voff; objlist_destroy(&ra.ignore_objlist); return (err); } static int dmu_recv_end_check(void *arg, dmu_tx_t *tx) { dmu_recv_cookie_t *drc = arg; dsl_pool_t *dp = dmu_tx_pool(tx); int error; ASSERT3P(drc->drc_ds->ds_owner, ==, dmu_recv_tag); if (!drc->drc_newfs) { dsl_dataset_t *origin_head; error = dsl_dataset_hold(dp, drc->drc_tofs, FTAG, &origin_head); if (error != 0) return (error); if (drc->drc_force) { /* * We will destroy any snapshots in tofs (i.e. before * origin_head) that are after the origin (which is * the snap before drc_ds, because drc_ds can not * have any snaps of its own). */ uint64_t obj; obj = dsl_dataset_phys(origin_head)->ds_prev_snap_obj; while (obj != dsl_dataset_phys(drc->drc_ds)->ds_prev_snap_obj) { dsl_dataset_t *snap; error = dsl_dataset_hold_obj(dp, obj, FTAG, &snap); if (error != 0) break; if (snap->ds_dir != origin_head->ds_dir) error = SET_ERROR(EINVAL); if (error == 0) { error = dsl_destroy_snapshot_check_impl( snap, B_FALSE); } obj = dsl_dataset_phys(snap)->ds_prev_snap_obj; dsl_dataset_rele(snap, FTAG); if (error != 0) break; } if (error != 0) { dsl_dataset_rele(origin_head, FTAG); return (error); } } error = dsl_dataset_clone_swap_check_impl(drc->drc_ds, origin_head, drc->drc_force, drc->drc_owner, tx); if (error != 0) { dsl_dataset_rele(origin_head, FTAG); return (error); } error = dsl_dataset_snapshot_check_impl(origin_head, drc->drc_tosnap, tx, B_TRUE, 1, drc->drc_cred); dsl_dataset_rele(origin_head, FTAG); if (error != 0) return (error); error = dsl_destroy_head_check_impl(drc->drc_ds, 1); } else { error = dsl_dataset_snapshot_check_impl(drc->drc_ds, drc->drc_tosnap, tx, B_TRUE, 1, drc->drc_cred); } return (error); } static void dmu_recv_end_sync(void *arg, dmu_tx_t *tx) { dmu_recv_cookie_t *drc = arg; dsl_pool_t *dp = dmu_tx_pool(tx); spa_history_log_internal_ds(drc->drc_ds, "finish receiving", tx, "snap=%s", drc->drc_tosnap); if (!drc->drc_newfs) { dsl_dataset_t *origin_head; VERIFY0(dsl_dataset_hold(dp, drc->drc_tofs, FTAG, &origin_head)); if (drc->drc_force) { /* * Destroy any snapshots of drc_tofs (origin_head) * after the origin (the snap before drc_ds). */ uint64_t obj; obj = dsl_dataset_phys(origin_head)->ds_prev_snap_obj; while (obj != dsl_dataset_phys(drc->drc_ds)->ds_prev_snap_obj) { dsl_dataset_t *snap; VERIFY0(dsl_dataset_hold_obj(dp, obj, FTAG, &snap)); ASSERT3P(snap->ds_dir, ==, origin_head->ds_dir); obj = dsl_dataset_phys(snap)->ds_prev_snap_obj; dsl_destroy_snapshot_sync_impl(snap, B_FALSE, tx); dsl_dataset_rele(snap, FTAG); } } VERIFY3P(drc->drc_ds->ds_prev, ==, origin_head->ds_prev); dsl_dataset_clone_swap_sync_impl(drc->drc_ds, origin_head, tx); dsl_dataset_snapshot_sync_impl(origin_head, drc->drc_tosnap, tx); /* set snapshot's creation time and guid */ dmu_buf_will_dirty(origin_head->ds_prev->ds_dbuf, tx); dsl_dataset_phys(origin_head->ds_prev)->ds_creation_time = drc->drc_drrb->drr_creation_time; dsl_dataset_phys(origin_head->ds_prev)->ds_guid = drc->drc_drrb->drr_toguid; dsl_dataset_phys(origin_head->ds_prev)->ds_flags &= ~DS_FLAG_INCONSISTENT; dmu_buf_will_dirty(origin_head->ds_dbuf, tx); dsl_dataset_phys(origin_head)->ds_flags &= ~DS_FLAG_INCONSISTENT; drc->drc_newsnapobj = dsl_dataset_phys(origin_head)->ds_prev_snap_obj; dsl_dataset_rele(origin_head, FTAG); dsl_destroy_head_sync_impl(drc->drc_ds, tx); if (drc->drc_owner != NULL) VERIFY3P(origin_head->ds_owner, ==, drc->drc_owner); } else { dsl_dataset_t *ds = drc->drc_ds; dsl_dataset_snapshot_sync_impl(ds, drc->drc_tosnap, tx); /* set snapshot's creation time and guid */ dmu_buf_will_dirty(ds->ds_prev->ds_dbuf, tx); dsl_dataset_phys(ds->ds_prev)->ds_creation_time = drc->drc_drrb->drr_creation_time; dsl_dataset_phys(ds->ds_prev)->ds_guid = drc->drc_drrb->drr_toguid; dsl_dataset_phys(ds->ds_prev)->ds_flags &= ~DS_FLAG_INCONSISTENT; dmu_buf_will_dirty(ds->ds_dbuf, tx); dsl_dataset_phys(ds)->ds_flags &= ~DS_FLAG_INCONSISTENT; if (dsl_dataset_has_resume_receive_state(ds)) { (void) zap_remove(dp->dp_meta_objset, ds->ds_object, DS_FIELD_RESUME_FROMGUID, tx); (void) zap_remove(dp->dp_meta_objset, ds->ds_object, DS_FIELD_RESUME_OBJECT, tx); (void) zap_remove(dp->dp_meta_objset, ds->ds_object, DS_FIELD_RESUME_OFFSET, tx); (void) zap_remove(dp->dp_meta_objset, ds->ds_object, DS_FIELD_RESUME_BYTES, tx); (void) zap_remove(dp->dp_meta_objset, ds->ds_object, DS_FIELD_RESUME_TOGUID, tx); (void) zap_remove(dp->dp_meta_objset, ds->ds_object, DS_FIELD_RESUME_TONAME, tx); } drc->drc_newsnapobj = dsl_dataset_phys(drc->drc_ds)->ds_prev_snap_obj; } /* * Release the hold from dmu_recv_begin. This must be done before * we return to open context, so that when we free the dataset's dnode, * we can evict its bonus buffer. */ dsl_dataset_disown(drc->drc_ds, dmu_recv_tag); drc->drc_ds = NULL; } static int add_ds_to_guidmap(const char *name, avl_tree_t *guid_map, uint64_t snapobj) { dsl_pool_t *dp; dsl_dataset_t *snapds; guid_map_entry_t *gmep; int err; ASSERT(guid_map != NULL); err = dsl_pool_hold(name, FTAG, &dp); if (err != 0) return (err); gmep = kmem_alloc(sizeof (*gmep), KM_SLEEP); err = dsl_dataset_hold_obj(dp, snapobj, gmep, &snapds); if (err == 0) { gmep->guid = dsl_dataset_phys(snapds)->ds_guid; gmep->gme_ds = snapds; avl_add(guid_map, gmep); dsl_dataset_long_hold(snapds, gmep); } else kmem_free(gmep, sizeof (*gmep)); dsl_pool_rele(dp, FTAG); return (err); } static int dmu_recv_end_modified_blocks = 3; static int dmu_recv_existing_end(dmu_recv_cookie_t *drc) { #ifdef _KERNEL /* * We will be destroying the ds; make sure its origin is unmounted if * necessary. */ char name[ZFS_MAX_DATASET_NAME_LEN]; dsl_dataset_name(drc->drc_ds, name); zfs_destroy_unmount_origin(name); #endif return (dsl_sync_task(drc->drc_tofs, dmu_recv_end_check, dmu_recv_end_sync, drc, dmu_recv_end_modified_blocks, ZFS_SPACE_CHECK_NORMAL)); } static int dmu_recv_new_end(dmu_recv_cookie_t *drc) { return (dsl_sync_task(drc->drc_tofs, dmu_recv_end_check, dmu_recv_end_sync, drc, dmu_recv_end_modified_blocks, ZFS_SPACE_CHECK_NORMAL)); } int dmu_recv_end(dmu_recv_cookie_t *drc, void *owner) { int error; drc->drc_owner = owner; if (drc->drc_newfs) error = dmu_recv_new_end(drc); else error = dmu_recv_existing_end(drc); if (error != 0) { dmu_recv_cleanup_ds(drc); } else if (drc->drc_guid_to_ds_map != NULL) { (void) add_ds_to_guidmap(drc->drc_tofs, drc->drc_guid_to_ds_map, drc->drc_newsnapobj); } return (error); } /* * Return TRUE if this objset is currently being received into. */ boolean_t dmu_objset_is_receiving(objset_t *os) { return (os->os_dsl_dataset != NULL && os->os_dsl_dataset->ds_owner == dmu_recv_tag); } Index: head/sys/cddl/contrib/opensolaris/uts/common/fs/zfs/dnode.c =================================================================== --- head/sys/cddl/contrib/opensolaris/uts/common/fs/zfs/dnode.c (revision 353555) +++ head/sys/cddl/contrib/opensolaris/uts/common/fs/zfs/dnode.c (revision 353556) @@ -1,2409 +1,2408 @@ /* * CDDL HEADER START * * The contents of this file are subject to the terms of the * Common Development and Distribution License (the "License"). * You may not use this file except in compliance with the License. * * You can obtain a copy of the license at usr/src/OPENSOLARIS.LICENSE * or http://www.opensolaris.org/os/licensing. * See the License for the specific language governing permissions * and limitations under the License. * * When distributing Covered Code, include this CDDL HEADER in each * file and include the License file at usr/src/OPENSOLARIS.LICENSE. * If applicable, add the following below this CDDL HEADER, with the * fields enclosed by brackets "[]" replaced with your own identifying * information: Portions Copyright [yyyy] [name of copyright owner] * * CDDL HEADER END */ /* * Copyright (c) 2005, 2010, Oracle and/or its affiliates. All rights reserved. * Copyright (c) 2012, 2017 by Delphix. All rights reserved. * Copyright (c) 2014 Spectra Logic Corporation, All rights reserved. * Copyright (c) 2014 Integros [integros.com] * Copyright 2017 RackTop Systems. */ #include #include #include #include #include #include #include #include #include #include #include #include #include dnode_stats_t dnode_stats = { { "dnode_hold_dbuf_hold", KSTAT_DATA_UINT64 }, { "dnode_hold_dbuf_read", KSTAT_DATA_UINT64 }, { "dnode_hold_alloc_hits", KSTAT_DATA_UINT64 }, { "dnode_hold_alloc_misses", KSTAT_DATA_UINT64 }, { "dnode_hold_alloc_interior", KSTAT_DATA_UINT64 }, { "dnode_hold_alloc_lock_retry", KSTAT_DATA_UINT64 }, { "dnode_hold_alloc_lock_misses", KSTAT_DATA_UINT64 }, { "dnode_hold_alloc_type_none", KSTAT_DATA_UINT64 }, { "dnode_hold_free_hits", KSTAT_DATA_UINT64 }, { "dnode_hold_free_misses", KSTAT_DATA_UINT64 }, { "dnode_hold_free_lock_misses", KSTAT_DATA_UINT64 }, { "dnode_hold_free_lock_retry", KSTAT_DATA_UINT64 }, { "dnode_hold_free_overflow", KSTAT_DATA_UINT64 }, { "dnode_hold_free_refcount", KSTAT_DATA_UINT64 }, { "dnode_hold_free_txg", KSTAT_DATA_UINT64 }, { "dnode_free_interior_lock_retry", KSTAT_DATA_UINT64 }, { "dnode_allocate", KSTAT_DATA_UINT64 }, { "dnode_reallocate", KSTAT_DATA_UINT64 }, { "dnode_buf_evict", KSTAT_DATA_UINT64 }, { "dnode_alloc_next_chunk", KSTAT_DATA_UINT64 }, { "dnode_alloc_race", KSTAT_DATA_UINT64 }, { "dnode_alloc_next_block", KSTAT_DATA_UINT64 }, { "dnode_move_invalid", KSTAT_DATA_UINT64 }, { "dnode_move_recheck1", KSTAT_DATA_UINT64 }, { "dnode_move_recheck2", KSTAT_DATA_UINT64 }, { "dnode_move_special", KSTAT_DATA_UINT64 }, { "dnode_move_handle", KSTAT_DATA_UINT64 }, { "dnode_move_rwlock", KSTAT_DATA_UINT64 }, { "dnode_move_active", KSTAT_DATA_UINT64 }, }; static kstat_t *dnode_ksp; static kmem_cache_t *dnode_cache; static dnode_phys_t dnode_phys_zero; int zfs_default_bs = SPA_MINBLOCKSHIFT; int zfs_default_ibs = DN_MAX_INDBLKSHIFT; SYSCTL_DECL(_vfs_zfs); SYSCTL_INT(_vfs_zfs, OID_AUTO, default_bs, CTLFLAG_RWTUN, &zfs_default_bs, 0, "Default dnode block shift"); SYSCTL_INT(_vfs_zfs, OID_AUTO, default_ibs, CTLFLAG_RWTUN, &zfs_default_ibs, 0, "Default dnode indirect block shift"); #ifdef illumos #ifdef _KERNEL static kmem_cbrc_t dnode_move(void *, void *, size_t, void *); #endif /* _KERNEL */ #endif static int dbuf_compare(const void *x1, const void *x2) { const dmu_buf_impl_t *d1 = x1; const dmu_buf_impl_t *d2 = x2; int cmp = AVL_CMP(d1->db_level, d2->db_level); if (likely(cmp)) return (cmp); cmp = AVL_CMP(d1->db_blkid, d2->db_blkid); if (likely(cmp)) return (cmp); if (d1->db_state == DB_SEARCH) { ASSERT3S(d2->db_state, !=, DB_SEARCH); return (-1); } else if (d2->db_state == DB_SEARCH) { ASSERT3S(d1->db_state, !=, DB_SEARCH); return (1); } return (AVL_PCMP(d1, d2)); } /* ARGSUSED */ static int dnode_cons(void *arg, void *unused, int kmflag) { dnode_t *dn = arg; int i; rw_init(&dn->dn_struct_rwlock, NULL, RW_DEFAULT, NULL); mutex_init(&dn->dn_mtx, NULL, MUTEX_DEFAULT, NULL); mutex_init(&dn->dn_dbufs_mtx, NULL, MUTEX_DEFAULT, NULL); cv_init(&dn->dn_notxholds, NULL, CV_DEFAULT, NULL); /* * Every dbuf has a reference, and dropping a tracked reference is * O(number of references), so don't track dn_holds. */ refcount_create_untracked(&dn->dn_holds); refcount_create(&dn->dn_tx_holds); list_link_init(&dn->dn_link); bzero(&dn->dn_next_nblkptr[0], sizeof (dn->dn_next_nblkptr)); bzero(&dn->dn_next_nlevels[0], sizeof (dn->dn_next_nlevels)); bzero(&dn->dn_next_indblkshift[0], sizeof (dn->dn_next_indblkshift)); bzero(&dn->dn_next_bonustype[0], sizeof (dn->dn_next_bonustype)); bzero(&dn->dn_rm_spillblk[0], sizeof (dn->dn_rm_spillblk)); bzero(&dn->dn_next_bonuslen[0], sizeof (dn->dn_next_bonuslen)); bzero(&dn->dn_next_blksz[0], sizeof (dn->dn_next_blksz)); for (i = 0; i < TXG_SIZE; i++) { list_link_init(&dn->dn_dirty_link[i]); dn->dn_free_ranges[i] = NULL; list_create(&dn->dn_dirty_records[i], sizeof (dbuf_dirty_record_t), offsetof(dbuf_dirty_record_t, dr_dirty_node)); } dn->dn_allocated_txg = 0; dn->dn_free_txg = 0; dn->dn_assigned_txg = 0; dn->dn_dirtyctx = 0; dn->dn_dirtyctx_firstset = NULL; dn->dn_bonus = NULL; dn->dn_have_spill = B_FALSE; dn->dn_zio = NULL; dn->dn_oldused = 0; dn->dn_oldflags = 0; dn->dn_olduid = 0; dn->dn_oldgid = 0; dn->dn_newuid = 0; dn->dn_newgid = 0; dn->dn_id_flags = 0; dn->dn_dbufs_count = 0; avl_create(&dn->dn_dbufs, dbuf_compare, sizeof (dmu_buf_impl_t), offsetof(dmu_buf_impl_t, db_link)); dn->dn_moved = 0; POINTER_INVALIDATE(&dn->dn_objset); return (0); } /* ARGSUSED */ static void dnode_dest(void *arg, void *unused) { int i; dnode_t *dn = arg; rw_destroy(&dn->dn_struct_rwlock); mutex_destroy(&dn->dn_mtx); mutex_destroy(&dn->dn_dbufs_mtx); cv_destroy(&dn->dn_notxholds); refcount_destroy(&dn->dn_holds); refcount_destroy(&dn->dn_tx_holds); ASSERT(!list_link_active(&dn->dn_link)); for (i = 0; i < TXG_SIZE; i++) { ASSERT(!list_link_active(&dn->dn_dirty_link[i])); ASSERT3P(dn->dn_free_ranges[i], ==, NULL); list_destroy(&dn->dn_dirty_records[i]); ASSERT0(dn->dn_next_nblkptr[i]); ASSERT0(dn->dn_next_nlevels[i]); ASSERT0(dn->dn_next_indblkshift[i]); ASSERT0(dn->dn_next_bonustype[i]); ASSERT0(dn->dn_rm_spillblk[i]); ASSERT0(dn->dn_next_bonuslen[i]); ASSERT0(dn->dn_next_blksz[i]); } ASSERT0(dn->dn_allocated_txg); ASSERT0(dn->dn_free_txg); ASSERT0(dn->dn_assigned_txg); ASSERT0(dn->dn_dirtyctx); ASSERT3P(dn->dn_dirtyctx_firstset, ==, NULL); ASSERT3P(dn->dn_bonus, ==, NULL); ASSERT(!dn->dn_have_spill); ASSERT3P(dn->dn_zio, ==, NULL); ASSERT0(dn->dn_oldused); ASSERT0(dn->dn_oldflags); ASSERT0(dn->dn_olduid); ASSERT0(dn->dn_oldgid); ASSERT0(dn->dn_newuid); ASSERT0(dn->dn_newgid); ASSERT0(dn->dn_id_flags); ASSERT0(dn->dn_dbufs_count); avl_destroy(&dn->dn_dbufs); } void dnode_init(void) { ASSERT(dnode_cache == NULL); dnode_cache = kmem_cache_create("dnode_t", sizeof (dnode_t), 0, dnode_cons, dnode_dest, NULL, NULL, NULL, 0); #ifdef _KERNEL kmem_cache_set_move(dnode_cache, dnode_move); dnode_ksp = kstat_create("zfs", 0, "dnodestats", "misc", KSTAT_TYPE_NAMED, sizeof (dnode_stats) / sizeof (kstat_named_t), KSTAT_FLAG_VIRTUAL); if (dnode_ksp != NULL) { dnode_ksp->ks_data = &dnode_stats; kstat_install(dnode_ksp); } #endif /* _KERNEL */ } void dnode_fini(void) { if (dnode_ksp != NULL) { kstat_delete(dnode_ksp); dnode_ksp = NULL; } kmem_cache_destroy(dnode_cache); dnode_cache = NULL; } #ifdef ZFS_DEBUG void dnode_verify(dnode_t *dn) { int drop_struct_lock = FALSE; ASSERT(dn->dn_phys); ASSERT(dn->dn_objset); ASSERT(dn->dn_handle->dnh_dnode == dn); ASSERT(DMU_OT_IS_VALID(dn->dn_phys->dn_type)); if (!(zfs_flags & ZFS_DEBUG_DNODE_VERIFY)) return; if (!RW_WRITE_HELD(&dn->dn_struct_rwlock)) { rw_enter(&dn->dn_struct_rwlock, RW_READER); drop_struct_lock = TRUE; } if (dn->dn_phys->dn_type != DMU_OT_NONE || dn->dn_allocated_txg != 0) { int i; int max_bonuslen = DN_SLOTS_TO_BONUSLEN(dn->dn_num_slots); ASSERT3U(dn->dn_indblkshift, >=, 0); ASSERT3U(dn->dn_indblkshift, <=, SPA_MAXBLOCKSHIFT); if (dn->dn_datablkshift) { ASSERT3U(dn->dn_datablkshift, >=, SPA_MINBLOCKSHIFT); ASSERT3U(dn->dn_datablkshift, <=, SPA_MAXBLOCKSHIFT); ASSERT3U(1<dn_datablkshift, ==, dn->dn_datablksz); } ASSERT3U(dn->dn_nlevels, <=, 30); ASSERT(DMU_OT_IS_VALID(dn->dn_type)); ASSERT3U(dn->dn_nblkptr, >=, 1); ASSERT3U(dn->dn_nblkptr, <=, DN_MAX_NBLKPTR); ASSERT3U(dn->dn_bonuslen, <=, max_bonuslen); ASSERT3U(dn->dn_datablksz, ==, dn->dn_datablkszsec << SPA_MINBLOCKSHIFT); ASSERT3U(ISP2(dn->dn_datablksz), ==, dn->dn_datablkshift != 0); ASSERT3U((dn->dn_nblkptr - 1) * sizeof (blkptr_t) + dn->dn_bonuslen, <=, max_bonuslen); for (i = 0; i < TXG_SIZE; i++) { ASSERT3U(dn->dn_next_nlevels[i], <=, dn->dn_nlevels); } } if (dn->dn_phys->dn_type != DMU_OT_NONE) ASSERT3U(dn->dn_phys->dn_nlevels, <=, dn->dn_nlevels); ASSERT(DMU_OBJECT_IS_SPECIAL(dn->dn_object) || dn->dn_dbuf != NULL); if (dn->dn_dbuf != NULL) { ASSERT3P(dn->dn_phys, ==, (dnode_phys_t *)dn->dn_dbuf->db.db_data + (dn->dn_object % (dn->dn_dbuf->db.db_size >> DNODE_SHIFT))); } if (drop_struct_lock) rw_exit(&dn->dn_struct_rwlock); } #endif void dnode_byteswap(dnode_phys_t *dnp) { uint64_t *buf64 = (void*)&dnp->dn_blkptr; int i; if (dnp->dn_type == DMU_OT_NONE) { bzero(dnp, sizeof (dnode_phys_t)); return; } dnp->dn_datablkszsec = BSWAP_16(dnp->dn_datablkszsec); dnp->dn_bonuslen = BSWAP_16(dnp->dn_bonuslen); dnp->dn_extra_slots = BSWAP_8(dnp->dn_extra_slots); dnp->dn_maxblkid = BSWAP_64(dnp->dn_maxblkid); dnp->dn_used = BSWAP_64(dnp->dn_used); /* * dn_nblkptr is only one byte, so it's OK to read it in either * byte order. We can't read dn_bouslen. */ ASSERT(dnp->dn_indblkshift <= SPA_MAXBLOCKSHIFT); ASSERT(dnp->dn_nblkptr <= DN_MAX_NBLKPTR); for (i = 0; i < dnp->dn_nblkptr * sizeof (blkptr_t)/8; i++) buf64[i] = BSWAP_64(buf64[i]); /* * OK to check dn_bonuslen for zero, because it won't matter if * we have the wrong byte order. This is necessary because the * dnode dnode is smaller than a regular dnode. */ if (dnp->dn_bonuslen != 0) { /* * Note that the bonus length calculated here may be * longer than the actual bonus buffer. This is because * we always put the bonus buffer after the last block * pointer (instead of packing it against the end of the * dnode buffer). */ int off = (dnp->dn_nblkptr-1) * sizeof (blkptr_t); int slots = dnp->dn_extra_slots + 1; size_t len = DN_SLOTS_TO_BONUSLEN(slots) - off; ASSERT(DMU_OT_IS_VALID(dnp->dn_bonustype)); dmu_object_byteswap_t byteswap = DMU_OT_BYTESWAP(dnp->dn_bonustype); dmu_ot_byteswap[byteswap].ob_func(dnp->dn_bonus + off, len); } /* Swap SPILL block if we have one */ if (dnp->dn_flags & DNODE_FLAG_SPILL_BLKPTR) byteswap_uint64_array(DN_SPILL_BLKPTR(dnp), sizeof (blkptr_t)); } void dnode_buf_byteswap(void *vbuf, size_t size) { int i = 0; ASSERT3U(sizeof (dnode_phys_t), ==, (1<dn_type != DMU_OT_NONE) i += dnp->dn_extra_slots * DNODE_MIN_SIZE; } } void dnode_setbonuslen(dnode_t *dn, int newsize, dmu_tx_t *tx) { ASSERT3U(refcount_count(&dn->dn_holds), >=, 1); dnode_setdirty(dn, tx); rw_enter(&dn->dn_struct_rwlock, RW_WRITER); ASSERT3U(newsize, <=, DN_SLOTS_TO_BONUSLEN(dn->dn_num_slots) - (dn->dn_nblkptr-1) * sizeof (blkptr_t)); dn->dn_bonuslen = newsize; if (newsize == 0) dn->dn_next_bonuslen[tx->tx_txg & TXG_MASK] = DN_ZERO_BONUSLEN; else dn->dn_next_bonuslen[tx->tx_txg & TXG_MASK] = dn->dn_bonuslen; rw_exit(&dn->dn_struct_rwlock); } void dnode_setbonus_type(dnode_t *dn, dmu_object_type_t newtype, dmu_tx_t *tx) { ASSERT3U(refcount_count(&dn->dn_holds), >=, 1); dnode_setdirty(dn, tx); rw_enter(&dn->dn_struct_rwlock, RW_WRITER); dn->dn_bonustype = newtype; dn->dn_next_bonustype[tx->tx_txg & TXG_MASK] = dn->dn_bonustype; rw_exit(&dn->dn_struct_rwlock); } void dnode_rm_spill(dnode_t *dn, dmu_tx_t *tx) { ASSERT3U(refcount_count(&dn->dn_holds), >=, 1); ASSERT(RW_WRITE_HELD(&dn->dn_struct_rwlock)); dnode_setdirty(dn, tx); dn->dn_rm_spillblk[tx->tx_txg&TXG_MASK] = DN_KILL_SPILLBLK; dn->dn_have_spill = B_FALSE; } static void dnode_setdblksz(dnode_t *dn, int size) { ASSERT0(P2PHASE(size, SPA_MINBLOCKSIZE)); ASSERT3U(size, <=, SPA_MAXBLOCKSIZE); ASSERT3U(size, >=, SPA_MINBLOCKSIZE); ASSERT3U(size >> SPA_MINBLOCKSHIFT, <, 1<<(sizeof (dn->dn_phys->dn_datablkszsec) * 8)); dn->dn_datablksz = size; dn->dn_datablkszsec = size >> SPA_MINBLOCKSHIFT; dn->dn_datablkshift = ISP2(size) ? highbit64(size - 1) : 0; } static dnode_t * dnode_create(objset_t *os, dnode_phys_t *dnp, dmu_buf_impl_t *db, uint64_t object, dnode_handle_t *dnh) { dnode_t *dn; dn = kmem_cache_alloc(dnode_cache, KM_SLEEP); #ifdef _KERNEL ASSERT(!POINTER_IS_VALID(dn->dn_objset)); #endif /* _KERNEL */ dn->dn_moved = 0; /* * Defer setting dn_objset until the dnode is ready to be a candidate * for the dnode_move() callback. */ dn->dn_object = object; dn->dn_dbuf = db; dn->dn_handle = dnh; dn->dn_phys = dnp; if (dnp->dn_datablkszsec) { dnode_setdblksz(dn, dnp->dn_datablkszsec << SPA_MINBLOCKSHIFT); } else { dn->dn_datablksz = 0; dn->dn_datablkszsec = 0; dn->dn_datablkshift = 0; } dn->dn_indblkshift = dnp->dn_indblkshift; dn->dn_nlevels = dnp->dn_nlevels; dn->dn_type = dnp->dn_type; dn->dn_nblkptr = dnp->dn_nblkptr; dn->dn_checksum = dnp->dn_checksum; dn->dn_compress = dnp->dn_compress; dn->dn_bonustype = dnp->dn_bonustype; dn->dn_bonuslen = dnp->dn_bonuslen; dn->dn_num_slots = dnp->dn_extra_slots + 1; dn->dn_maxblkid = dnp->dn_maxblkid; dn->dn_have_spill = ((dnp->dn_flags & DNODE_FLAG_SPILL_BLKPTR) != 0); dn->dn_id_flags = 0; dmu_zfetch_init(&dn->dn_zfetch, dn); ASSERT(DMU_OT_IS_VALID(dn->dn_phys->dn_type)); ASSERT(zrl_is_locked(&dnh->dnh_zrlock)); ASSERT(!DN_SLOT_IS_PTR(dnh->dnh_dnode)); mutex_enter(&os->os_lock); /* * Exclude special dnodes from os_dnodes so an empty os_dnodes * signifies that the special dnodes have no references from * their children (the entries in os_dnodes). This allows * dnode_destroy() to easily determine if the last child has * been removed and then complete eviction of the objset. */ if (!DMU_OBJECT_IS_SPECIAL(object)) list_insert_head(&os->os_dnodes, dn); membar_producer(); /* * Everything else must be valid before assigning dn_objset * makes the dnode eligible for dnode_move(). */ dn->dn_objset = os; dnh->dnh_dnode = dn; mutex_exit(&os->os_lock); arc_space_consume(sizeof (dnode_t), ARC_SPACE_DNODE); return (dn); } /* * Caller must be holding the dnode handle, which is released upon return. */ static void dnode_destroy(dnode_t *dn) { objset_t *os = dn->dn_objset; boolean_t complete_os_eviction = B_FALSE; ASSERT((dn->dn_id_flags & DN_ID_NEW_EXIST) == 0); mutex_enter(&os->os_lock); POINTER_INVALIDATE(&dn->dn_objset); if (!DMU_OBJECT_IS_SPECIAL(dn->dn_object)) { list_remove(&os->os_dnodes, dn); complete_os_eviction = list_is_empty(&os->os_dnodes) && list_link_active(&os->os_evicting_node); } mutex_exit(&os->os_lock); /* the dnode can no longer move, so we can release the handle */ if (!zrl_is_locked(&dn->dn_handle->dnh_zrlock)) zrl_remove(&dn->dn_handle->dnh_zrlock); dn->dn_allocated_txg = 0; dn->dn_free_txg = 0; dn->dn_assigned_txg = 0; dn->dn_dirtyctx = 0; if (dn->dn_dirtyctx_firstset != NULL) { kmem_free(dn->dn_dirtyctx_firstset, 1); dn->dn_dirtyctx_firstset = NULL; } if (dn->dn_bonus != NULL) { mutex_enter(&dn->dn_bonus->db_mtx); dbuf_destroy(dn->dn_bonus); dn->dn_bonus = NULL; } dn->dn_zio = NULL; dn->dn_have_spill = B_FALSE; dn->dn_oldused = 0; dn->dn_oldflags = 0; dn->dn_olduid = 0; dn->dn_oldgid = 0; dn->dn_newuid = 0; dn->dn_newgid = 0; dn->dn_id_flags = 0; dmu_zfetch_fini(&dn->dn_zfetch); kmem_cache_free(dnode_cache, dn); arc_space_return(sizeof (dnode_t), ARC_SPACE_DNODE); if (complete_os_eviction) dmu_objset_evict_done(os); } void dnode_allocate(dnode_t *dn, dmu_object_type_t ot, int blocksize, int ibs, dmu_object_type_t bonustype, int bonuslen, int dn_slots, dmu_tx_t *tx) { int i; ASSERT3U(dn_slots, >, 0); ASSERT3U(dn_slots << DNODE_SHIFT, <=, spa_maxdnodesize(dmu_objset_spa(dn->dn_objset))); ASSERT3U(blocksize, <=, spa_maxblocksize(dmu_objset_spa(dn->dn_objset))); if (blocksize == 0) blocksize = 1 << zfs_default_bs; else blocksize = P2ROUNDUP(blocksize, SPA_MINBLOCKSIZE); if (ibs == 0) ibs = zfs_default_ibs; ibs = MIN(MAX(ibs, DN_MIN_INDBLKSHIFT), DN_MAX_INDBLKSHIFT); dprintf("os=%p obj=%" PRIu64 " txg=%" PRIu64 " blocksize=%d ibs=%d dn_slots=%d\n", dn->dn_objset, dn->dn_object, tx->tx_txg, blocksize, ibs, dn_slots); DNODE_STAT_BUMP(dnode_allocate); ASSERT(dn->dn_type == DMU_OT_NONE); ASSERT(bcmp(dn->dn_phys, &dnode_phys_zero, sizeof (dnode_phys_t)) == 0); ASSERT(dn->dn_phys->dn_type == DMU_OT_NONE); ASSERT(ot != DMU_OT_NONE); ASSERT(DMU_OT_IS_VALID(ot)); ASSERT((bonustype == DMU_OT_NONE && bonuslen == 0) || (bonustype == DMU_OT_SA && bonuslen == 0) || (bonustype != DMU_OT_NONE && bonuslen != 0)); ASSERT(DMU_OT_IS_VALID(bonustype)); ASSERT3U(bonuslen, <=, DN_SLOTS_TO_BONUSLEN(dn_slots)); ASSERT(dn->dn_type == DMU_OT_NONE); ASSERT0(dn->dn_maxblkid); ASSERT0(dn->dn_allocated_txg); ASSERT0(dn->dn_assigned_txg); ASSERT(refcount_is_zero(&dn->dn_tx_holds)); ASSERT3U(refcount_count(&dn->dn_holds), <=, 1); ASSERT(avl_is_empty(&dn->dn_dbufs)); for (i = 0; i < TXG_SIZE; i++) { ASSERT0(dn->dn_next_nblkptr[i]); ASSERT0(dn->dn_next_nlevels[i]); ASSERT0(dn->dn_next_indblkshift[i]); ASSERT0(dn->dn_next_bonuslen[i]); ASSERT0(dn->dn_next_bonustype[i]); ASSERT0(dn->dn_rm_spillblk[i]); ASSERT0(dn->dn_next_blksz[i]); ASSERT(!list_link_active(&dn->dn_dirty_link[i])); ASSERT3P(list_head(&dn->dn_dirty_records[i]), ==, NULL); ASSERT3P(dn->dn_free_ranges[i], ==, NULL); } dn->dn_type = ot; dnode_setdblksz(dn, blocksize); dn->dn_indblkshift = ibs; dn->dn_nlevels = 1; dn->dn_num_slots = dn_slots; if (bonustype == DMU_OT_SA) /* Maximize bonus space for SA */ dn->dn_nblkptr = 1; else { dn->dn_nblkptr = MIN(DN_MAX_NBLKPTR, 1 + ((DN_SLOTS_TO_BONUSLEN(dn_slots) - bonuslen) >> SPA_BLKPTRSHIFT)); } dn->dn_bonustype = bonustype; dn->dn_bonuslen = bonuslen; dn->dn_checksum = ZIO_CHECKSUM_INHERIT; dn->dn_compress = ZIO_COMPRESS_INHERIT; dn->dn_dirtyctx = 0; dn->dn_free_txg = 0; if (dn->dn_dirtyctx_firstset) { kmem_free(dn->dn_dirtyctx_firstset, 1); dn->dn_dirtyctx_firstset = NULL; } dn->dn_allocated_txg = tx->tx_txg; dn->dn_id_flags = 0; dnode_setdirty(dn, tx); dn->dn_next_indblkshift[tx->tx_txg & TXG_MASK] = ibs; dn->dn_next_bonuslen[tx->tx_txg & TXG_MASK] = dn->dn_bonuslen; dn->dn_next_bonustype[tx->tx_txg & TXG_MASK] = dn->dn_bonustype; dn->dn_next_blksz[tx->tx_txg & TXG_MASK] = dn->dn_datablksz; } void dnode_reallocate(dnode_t *dn, dmu_object_type_t ot, int blocksize, dmu_object_type_t bonustype, int bonuslen, int dn_slots, dmu_tx_t *tx) { int nblkptr; ASSERT3U(blocksize, >=, SPA_MINBLOCKSIZE); ASSERT3U(blocksize, <=, spa_maxblocksize(dmu_objset_spa(dn->dn_objset))); ASSERT0(blocksize % SPA_MINBLOCKSIZE); ASSERT(dn->dn_object != DMU_META_DNODE_OBJECT || dmu_tx_private_ok(tx)); ASSERT(tx->tx_txg != 0); ASSERT((bonustype == DMU_OT_NONE && bonuslen == 0) || (bonustype != DMU_OT_NONE && bonuslen != 0) || (bonustype == DMU_OT_SA && bonuslen == 0)); ASSERT(DMU_OT_IS_VALID(bonustype)); ASSERT3U(bonuslen, <=, DN_BONUS_SIZE(spa_maxdnodesize(dmu_objset_spa(dn->dn_objset)))); - - dn_slots = dn_slots > 0 ? dn_slots : DNODE_MIN_SLOTS; + ASSERT3U(bonuslen, <=, DN_BONUS_SIZE(dn_slots << DNODE_SHIFT)); dnode_free_interior_slots(dn); DNODE_STAT_BUMP(dnode_reallocate); /* clean up any unreferenced dbufs */ dnode_evict_dbufs(dn); dn->dn_id_flags = 0; rw_enter(&dn->dn_struct_rwlock, RW_WRITER); dnode_setdirty(dn, tx); if (dn->dn_datablksz != blocksize) { /* change blocksize */ ASSERT(dn->dn_maxblkid == 0 && (BP_IS_HOLE(&dn->dn_phys->dn_blkptr[0]) || dnode_block_freed(dn, 0))); dnode_setdblksz(dn, blocksize); dn->dn_next_blksz[tx->tx_txg&TXG_MASK] = blocksize; } if (dn->dn_bonuslen != bonuslen) dn->dn_next_bonuslen[tx->tx_txg&TXG_MASK] = bonuslen; if (bonustype == DMU_OT_SA) /* Maximize bonus space for SA */ nblkptr = 1; else nblkptr = MIN(DN_MAX_NBLKPTR, 1 + ((DN_SLOTS_TO_BONUSLEN(dn_slots) - bonuslen) >> SPA_BLKPTRSHIFT)); if (dn->dn_bonustype != bonustype) dn->dn_next_bonustype[tx->tx_txg&TXG_MASK] = bonustype; if (dn->dn_nblkptr != nblkptr) dn->dn_next_nblkptr[tx->tx_txg&TXG_MASK] = nblkptr; if (dn->dn_phys->dn_flags & DNODE_FLAG_SPILL_BLKPTR) { dbuf_rm_spill(dn, tx); dnode_rm_spill(dn, tx); } rw_exit(&dn->dn_struct_rwlock); /* change type */ dn->dn_type = ot; /* change bonus size and type */ mutex_enter(&dn->dn_mtx); dn->dn_bonustype = bonustype; dn->dn_bonuslen = bonuslen; dn->dn_num_slots = dn_slots; dn->dn_nblkptr = nblkptr; dn->dn_checksum = ZIO_CHECKSUM_INHERIT; dn->dn_compress = ZIO_COMPRESS_INHERIT; ASSERT3U(dn->dn_nblkptr, <=, DN_MAX_NBLKPTR); /* fix up the bonus db_size */ if (dn->dn_bonus) { dn->dn_bonus->db.db_size = DN_SLOTS_TO_BONUSLEN(dn->dn_num_slots) - (dn->dn_nblkptr - 1) * sizeof (blkptr_t); ASSERT(dn->dn_bonuslen <= dn->dn_bonus->db.db_size); } dn->dn_allocated_txg = tx->tx_txg; mutex_exit(&dn->dn_mtx); } #ifdef _KERNEL static void dnode_move_impl(dnode_t *odn, dnode_t *ndn) { int i; ASSERT(!RW_LOCK_HELD(&odn->dn_struct_rwlock)); ASSERT(MUTEX_NOT_HELD(&odn->dn_mtx)); ASSERT(MUTEX_NOT_HELD(&odn->dn_dbufs_mtx)); ASSERT(!RW_LOCK_HELD(&odn->dn_zfetch.zf_rwlock)); /* Copy fields. */ ndn->dn_objset = odn->dn_objset; ndn->dn_object = odn->dn_object; ndn->dn_dbuf = odn->dn_dbuf; ndn->dn_handle = odn->dn_handle; ndn->dn_phys = odn->dn_phys; ndn->dn_type = odn->dn_type; ndn->dn_bonuslen = odn->dn_bonuslen; ndn->dn_bonustype = odn->dn_bonustype; ndn->dn_nblkptr = odn->dn_nblkptr; ndn->dn_checksum = odn->dn_checksum; ndn->dn_compress = odn->dn_compress; ndn->dn_nlevels = odn->dn_nlevels; ndn->dn_indblkshift = odn->dn_indblkshift; ndn->dn_datablkshift = odn->dn_datablkshift; ndn->dn_datablkszsec = odn->dn_datablkszsec; ndn->dn_datablksz = odn->dn_datablksz; ndn->dn_maxblkid = odn->dn_maxblkid; ndn->dn_num_slots = odn->dn_num_slots; bcopy(&odn->dn_next_type[0], &ndn->dn_next_type[0], sizeof (odn->dn_next_type)); bcopy(&odn->dn_next_nblkptr[0], &ndn->dn_next_nblkptr[0], sizeof (odn->dn_next_nblkptr)); bcopy(&odn->dn_next_nlevels[0], &ndn->dn_next_nlevels[0], sizeof (odn->dn_next_nlevels)); bcopy(&odn->dn_next_indblkshift[0], &ndn->dn_next_indblkshift[0], sizeof (odn->dn_next_indblkshift)); bcopy(&odn->dn_next_bonustype[0], &ndn->dn_next_bonustype[0], sizeof (odn->dn_next_bonustype)); bcopy(&odn->dn_rm_spillblk[0], &ndn->dn_rm_spillblk[0], sizeof (odn->dn_rm_spillblk)); bcopy(&odn->dn_next_bonuslen[0], &ndn->dn_next_bonuslen[0], sizeof (odn->dn_next_bonuslen)); bcopy(&odn->dn_next_blksz[0], &ndn->dn_next_blksz[0], sizeof (odn->dn_next_blksz)); for (i = 0; i < TXG_SIZE; i++) { list_move_tail(&ndn->dn_dirty_records[i], &odn->dn_dirty_records[i]); } bcopy(&odn->dn_free_ranges[0], &ndn->dn_free_ranges[0], sizeof (odn->dn_free_ranges)); ndn->dn_allocated_txg = odn->dn_allocated_txg; ndn->dn_free_txg = odn->dn_free_txg; ndn->dn_assigned_txg = odn->dn_assigned_txg; ndn->dn_dirtyctx = odn->dn_dirtyctx; ndn->dn_dirtyctx_firstset = odn->dn_dirtyctx_firstset; ASSERT(refcount_count(&odn->dn_tx_holds) == 0); refcount_transfer(&ndn->dn_holds, &odn->dn_holds); ASSERT(avl_is_empty(&ndn->dn_dbufs)); avl_swap(&ndn->dn_dbufs, &odn->dn_dbufs); ndn->dn_dbufs_count = odn->dn_dbufs_count; ndn->dn_bonus = odn->dn_bonus; ndn->dn_have_spill = odn->dn_have_spill; ndn->dn_zio = odn->dn_zio; ndn->dn_oldused = odn->dn_oldused; ndn->dn_oldflags = odn->dn_oldflags; ndn->dn_olduid = odn->dn_olduid; ndn->dn_oldgid = odn->dn_oldgid; ndn->dn_newuid = odn->dn_newuid; ndn->dn_newgid = odn->dn_newgid; ndn->dn_id_flags = odn->dn_id_flags; dmu_zfetch_init(&ndn->dn_zfetch, NULL); list_move_tail(&ndn->dn_zfetch.zf_stream, &odn->dn_zfetch.zf_stream); ndn->dn_zfetch.zf_dnode = odn->dn_zfetch.zf_dnode; /* * Update back pointers. Updating the handle fixes the back pointer of * every descendant dbuf as well as the bonus dbuf. */ ASSERT(ndn->dn_handle->dnh_dnode == odn); ndn->dn_handle->dnh_dnode = ndn; if (ndn->dn_zfetch.zf_dnode == odn) { ndn->dn_zfetch.zf_dnode = ndn; } /* * Invalidate the original dnode by clearing all of its back pointers. */ odn->dn_dbuf = NULL; odn->dn_handle = NULL; avl_create(&odn->dn_dbufs, dbuf_compare, sizeof (dmu_buf_impl_t), offsetof(dmu_buf_impl_t, db_link)); odn->dn_dbufs_count = 0; odn->dn_bonus = NULL; odn->dn_zfetch.zf_dnode = NULL; /* * Set the low bit of the objset pointer to ensure that dnode_move() * recognizes the dnode as invalid in any subsequent callback. */ POINTER_INVALIDATE(&odn->dn_objset); /* * Satisfy the destructor. */ for (i = 0; i < TXG_SIZE; i++) { list_create(&odn->dn_dirty_records[i], sizeof (dbuf_dirty_record_t), offsetof(dbuf_dirty_record_t, dr_dirty_node)); odn->dn_free_ranges[i] = NULL; odn->dn_next_nlevels[i] = 0; odn->dn_next_indblkshift[i] = 0; odn->dn_next_bonustype[i] = 0; odn->dn_rm_spillblk[i] = 0; odn->dn_next_bonuslen[i] = 0; odn->dn_next_blksz[i] = 0; } odn->dn_allocated_txg = 0; odn->dn_free_txg = 0; odn->dn_assigned_txg = 0; odn->dn_dirtyctx = 0; odn->dn_dirtyctx_firstset = NULL; odn->dn_have_spill = B_FALSE; odn->dn_zio = NULL; odn->dn_oldused = 0; odn->dn_oldflags = 0; odn->dn_olduid = 0; odn->dn_oldgid = 0; odn->dn_newuid = 0; odn->dn_newgid = 0; odn->dn_id_flags = 0; /* * Mark the dnode. */ ndn->dn_moved = 1; odn->dn_moved = (uint8_t)-1; } #ifdef illumos /*ARGSUSED*/ static kmem_cbrc_t dnode_move(void *buf, void *newbuf, size_t size, void *arg) { dnode_t *odn = buf, *ndn = newbuf; objset_t *os; int64_t refcount; uint32_t dbufs; /* * The dnode is on the objset's list of known dnodes if the objset * pointer is valid. We set the low bit of the objset pointer when * freeing the dnode to invalidate it, and the memory patterns written * by kmem (baddcafe and deadbeef) set at least one of the two low bits. * A newly created dnode sets the objset pointer last of all to indicate * that the dnode is known and in a valid state to be moved by this * function. */ os = odn->dn_objset; if (!POINTER_IS_VALID(os)) { DNODE_STAT_BUMP(dnode_move_invalid); return (KMEM_CBRC_DONT_KNOW); } /* * Ensure that the objset does not go away during the move. */ rw_enter(&os_lock, RW_WRITER); if (os != odn->dn_objset) { rw_exit(&os_lock); DNODE_STAT_BUMP(dnode_move_recheck1); return (KMEM_CBRC_DONT_KNOW); } /* * If the dnode is still valid, then so is the objset. We know that no * valid objset can be freed while we hold os_lock, so we can safely * ensure that the objset remains in use. */ mutex_enter(&os->os_lock); /* * Recheck the objset pointer in case the dnode was removed just before * acquiring the lock. */ if (os != odn->dn_objset) { mutex_exit(&os->os_lock); rw_exit(&os_lock); DNODE_STAT_BUMP(dnode_move_recheck2); return (KMEM_CBRC_DONT_KNOW); } /* * At this point we know that as long as we hold os->os_lock, the dnode * cannot be freed and fields within the dnode can be safely accessed. * The objset listing this dnode cannot go away as long as this dnode is * on its list. */ rw_exit(&os_lock); if (DMU_OBJECT_IS_SPECIAL(odn->dn_object)) { mutex_exit(&os->os_lock); DNODE_STAT_BUMP(dnode_move_special); return (KMEM_CBRC_NO); } ASSERT(odn->dn_dbuf != NULL); /* only "special" dnodes have no parent */ /* * Lock the dnode handle to prevent the dnode from obtaining any new * holds. This also prevents the descendant dbufs and the bonus dbuf * from accessing the dnode, so that we can discount their holds. The * handle is safe to access because we know that while the dnode cannot * go away, neither can its handle. Once we hold dnh_zrlock, we can * safely move any dnode referenced only by dbufs. */ if (!zrl_tryenter(&odn->dn_handle->dnh_zrlock)) { mutex_exit(&os->os_lock); DNODE_STAT_BUMP(dnode_move_handle); return (KMEM_CBRC_LATER); } /* * Ensure a consistent view of the dnode's holds and the dnode's dbufs. * We need to guarantee that there is a hold for every dbuf in order to * determine whether the dnode is actively referenced. Falsely matching * a dbuf to an active hold would lead to an unsafe move. It's possible * that a thread already having an active dnode hold is about to add a * dbuf, and we can't compare hold and dbuf counts while the add is in * progress. */ if (!rw_tryenter(&odn->dn_struct_rwlock, RW_WRITER)) { zrl_exit(&odn->dn_handle->dnh_zrlock); mutex_exit(&os->os_lock); DNODE_STAT_BUMP(dnode_move_rwlock); return (KMEM_CBRC_LATER); } /* * A dbuf may be removed (evicted) without an active dnode hold. In that * case, the dbuf count is decremented under the handle lock before the * dbuf's hold is released. This order ensures that if we count the hold * after the dbuf is removed but before its hold is released, we will * treat the unmatched hold as active and exit safely. If we count the * hold before the dbuf is removed, the hold is discounted, and the * removal is blocked until the move completes. */ refcount = refcount_count(&odn->dn_holds); ASSERT(refcount >= 0); dbufs = odn->dn_dbufs_count; /* We can't have more dbufs than dnode holds. */ ASSERT3U(dbufs, <=, refcount); DTRACE_PROBE3(dnode__move, dnode_t *, odn, int64_t, refcount, uint32_t, dbufs); if (refcount > dbufs) { rw_exit(&odn->dn_struct_rwlock); zrl_exit(&odn->dn_handle->dnh_zrlock); mutex_exit(&os->os_lock); DNODE_STAT_BUMP(dnode_move_active); return (KMEM_CBRC_LATER); } rw_exit(&odn->dn_struct_rwlock); /* * At this point we know that anyone with a hold on the dnode is not * actively referencing it. The dnode is known and in a valid state to * move. We're holding the locks needed to execute the critical section. */ dnode_move_impl(odn, ndn); list_link_replace(&odn->dn_link, &ndn->dn_link); /* If the dnode was safe to move, the refcount cannot have changed. */ ASSERT(refcount == refcount_count(&ndn->dn_holds)); ASSERT(dbufs == ndn->dn_dbufs_count); zrl_exit(&ndn->dn_handle->dnh_zrlock); /* handle has moved */ mutex_exit(&os->os_lock); return (KMEM_CBRC_YES); } #endif /* illumos */ #endif /* _KERNEL */ static void dnode_slots_hold(dnode_children_t *children, int idx, int slots) { ASSERT3S(idx + slots, <=, DNODES_PER_BLOCK); for (int i = idx; i < idx + slots; i++) { dnode_handle_t *dnh = &children->dnc_children[i]; zrl_add(&dnh->dnh_zrlock); } } static void dnode_slots_rele(dnode_children_t *children, int idx, int slots) { ASSERT3S(idx + slots, <=, DNODES_PER_BLOCK); for (int i = idx; i < idx + slots; i++) { dnode_handle_t *dnh = &children->dnc_children[i]; if (zrl_is_locked(&dnh->dnh_zrlock)) zrl_exit(&dnh->dnh_zrlock); else zrl_remove(&dnh->dnh_zrlock); } } static int dnode_slots_tryenter(dnode_children_t *children, int idx, int slots) { ASSERT3S(idx + slots, <=, DNODES_PER_BLOCK); for (int i = idx; i < idx + slots; i++) { dnode_handle_t *dnh = &children->dnc_children[i]; if (!zrl_tryenter(&dnh->dnh_zrlock)) { for (int j = idx; j < i; j++) { dnh = &children->dnc_children[j]; zrl_exit(&dnh->dnh_zrlock); } return (0); } } return (1); } static void dnode_set_slots(dnode_children_t *children, int idx, int slots, void *ptr) { ASSERT3S(idx + slots, <=, DNODES_PER_BLOCK); for (int i = idx; i < idx + slots; i++) { dnode_handle_t *dnh = &children->dnc_children[i]; dnh->dnh_dnode = ptr; } } static boolean_t dnode_check_slots_free(dnode_children_t *children, int idx, int slots) { ASSERT3S(idx + slots, <=, DNODES_PER_BLOCK); for (int i = idx; i < idx + slots; i++) { dnode_handle_t *dnh = &children->dnc_children[i]; dnode_t *dn = dnh->dnh_dnode; if (dn == DN_SLOT_FREE) { continue; } else if (DN_SLOT_IS_PTR(dn)) { mutex_enter(&dn->dn_mtx); dmu_object_type_t type = dn->dn_type; mutex_exit(&dn->dn_mtx); if (type != DMU_OT_NONE) return (B_FALSE); continue; } else { return (B_FALSE); } return (B_FALSE); } return (B_TRUE); } static void dnode_reclaim_slots(dnode_children_t *children, int idx, int slots) { ASSERT3S(idx + slots, <=, DNODES_PER_BLOCK); for (int i = idx; i < idx + slots; i++) { dnode_handle_t *dnh = &children->dnc_children[i]; ASSERT(zrl_is_locked(&dnh->dnh_zrlock)); if (DN_SLOT_IS_PTR(dnh->dnh_dnode)) { ASSERT3S(dnh->dnh_dnode->dn_type, ==, DMU_OT_NONE); dnode_destroy(dnh->dnh_dnode); dnh->dnh_dnode = DN_SLOT_FREE; } } } void dnode_free_interior_slots(dnode_t *dn) { dnode_children_t *children = dmu_buf_get_user(&dn->dn_dbuf->db); int epb = dn->dn_dbuf->db.db_size >> DNODE_SHIFT; int idx = (dn->dn_object & (epb - 1)) + 1; int slots = dn->dn_num_slots - 1; if (slots == 0) return; ASSERT3S(idx + slots, <=, DNODES_PER_BLOCK); while (!dnode_slots_tryenter(children, idx, slots)) DNODE_STAT_BUMP(dnode_free_interior_lock_retry); dnode_set_slots(children, idx, slots, DN_SLOT_FREE); dnode_slots_rele(children, idx, slots); } void dnode_special_close(dnode_handle_t *dnh) { dnode_t *dn = dnh->dnh_dnode; /* * Wait for final references to the dnode to clear. This can * only happen if the arc is asynchronously evicting state that * has a hold on this dnode while we are trying to evict this * dnode. */ while (refcount_count(&dn->dn_holds) > 0) delay(1); ASSERT(dn->dn_dbuf == NULL || dmu_buf_get_user(&dn->dn_dbuf->db) == NULL); zrl_add(&dnh->dnh_zrlock); dnode_destroy(dn); /* implicit zrl_remove() */ zrl_destroy(&dnh->dnh_zrlock); dnh->dnh_dnode = NULL; } void dnode_special_open(objset_t *os, dnode_phys_t *dnp, uint64_t object, dnode_handle_t *dnh) { dnode_t *dn; zrl_init(&dnh->dnh_zrlock); zrl_tryenter(&dnh->dnh_zrlock); dn = dnode_create(os, dnp, NULL, object, dnh); DNODE_VERIFY(dn); zrl_exit(&dnh->dnh_zrlock); } static void dnode_buf_evict_async(void *dbu) { dnode_children_t *dnc = dbu; DNODE_STAT_BUMP(dnode_buf_evict); for (int i = 0; i < dnc->dnc_count; i++) { dnode_handle_t *dnh = &dnc->dnc_children[i]; dnode_t *dn; /* * The dnode handle lock guards against the dnode moving to * another valid address, so there is no need here to guard * against changes to or from NULL. */ if (!DN_SLOT_IS_PTR(dnh->dnh_dnode)) { zrl_destroy(&dnh->dnh_zrlock); dnh->dnh_dnode = DN_SLOT_UNINIT; continue; } zrl_add(&dnh->dnh_zrlock); dn = dnh->dnh_dnode; /* * If there are holds on this dnode, then there should * be holds on the dnode's containing dbuf as well; thus * it wouldn't be eligible for eviction and this function * would not have been called. */ ASSERT(refcount_is_zero(&dn->dn_holds)); ASSERT(refcount_is_zero(&dn->dn_tx_holds)); dnode_destroy(dn); /* implicit zrl_remove() for first slot */ zrl_destroy(&dnh->dnh_zrlock); dnh->dnh_dnode = DN_SLOT_UNINIT; } kmem_free(dnc, sizeof (dnode_children_t) + dnc->dnc_count * sizeof (dnode_handle_t)); } /* * When the DNODE_MUST_BE_FREE flag is set, the "slots" parameter is used * to ensure the hole at the specified object offset is large enough to * hold the dnode being created. The slots parameter is also used to ensure * a dnode does not span multiple dnode blocks. In both of these cases, if * a failure occurs, ENOSPC is returned. Keep in mind, these failure cases * are only possible when using DNODE_MUST_BE_FREE. * * If the DNODE_MUST_BE_ALLOCATED flag is set, "slots" must be 0. * dnode_hold_impl() will check if the requested dnode is already consumed * as an extra dnode slot by an large dnode, in which case it returns * ENOENT. * * errors: * EINVAL - invalid object number or flags. * ENOSPC - hole too small to fulfill "slots" request (DNODE_MUST_BE_FREE) * EEXIST - Refers to an allocated dnode (DNODE_MUST_BE_FREE) * - Refers to a freeing dnode (DNODE_MUST_BE_FREE) * - Refers to an interior dnode slot (DNODE_MUST_BE_ALLOCATED) * ENOENT - The requested dnode is not allocated (DNODE_MUST_BE_ALLOCATED) * - The requested dnode is being freed (DNODE_MUST_BE_ALLOCATED) * EIO - i/o error error when reading the meta dnode dbuf. * succeeds even for free dnodes. */ int dnode_hold_impl(objset_t *os, uint64_t object, int flag, int slots, void *tag, dnode_t **dnp) { int epb, idx, err, i; int drop_struct_lock = FALSE; int type; uint64_t blk; dnode_t *mdn, *dn; dmu_buf_impl_t *db; dnode_children_t *dnc; dnode_phys_t *dn_block; dnode_phys_t *dn_block_begin; dnode_handle_t *dnh; ASSERT(!(flag & DNODE_MUST_BE_ALLOCATED) || (slots == 0)); ASSERT(!(flag & DNODE_MUST_BE_FREE) || (slots > 0)); /* * If you are holding the spa config lock as writer, you shouldn't * be asking the DMU to do *anything* unless it's the root pool * which may require us to read from the root filesystem while * holding some (not all) of the locks as writer. */ ASSERT(spa_config_held(os->os_spa, SCL_ALL, RW_WRITER) == 0 || (spa_is_root(os->os_spa) && spa_config_held(os->os_spa, SCL_STATE, RW_WRITER))); ASSERT((flag & DNODE_MUST_BE_ALLOCATED) || (flag & DNODE_MUST_BE_FREE)); if (object == DMU_USERUSED_OBJECT || object == DMU_GROUPUSED_OBJECT) { dn = (object == DMU_USERUSED_OBJECT) ? DMU_USERUSED_DNODE(os) : DMU_GROUPUSED_DNODE(os); if (dn == NULL) return (SET_ERROR(ENOENT)); type = dn->dn_type; if ((flag & DNODE_MUST_BE_ALLOCATED) && type == DMU_OT_NONE) return (SET_ERROR(ENOENT)); if ((flag & DNODE_MUST_BE_FREE) && type != DMU_OT_NONE) return (SET_ERROR(EEXIST)); DNODE_VERIFY(dn); (void) refcount_add(&dn->dn_holds, tag); *dnp = dn; return (0); } if (object == 0 || object >= DN_MAX_OBJECT) return (SET_ERROR(EINVAL)); mdn = DMU_META_DNODE(os); ASSERT(mdn->dn_object == DMU_META_DNODE_OBJECT); DNODE_VERIFY(mdn); if (!RW_WRITE_HELD(&mdn->dn_struct_rwlock)) { rw_enter(&mdn->dn_struct_rwlock, RW_READER); drop_struct_lock = TRUE; } blk = dbuf_whichblock(mdn, 0, object * sizeof (dnode_phys_t)); db = dbuf_hold(mdn, blk, FTAG); if (drop_struct_lock) rw_exit(&mdn->dn_struct_rwlock); if (db == NULL) { DNODE_STAT_BUMP(dnode_hold_dbuf_hold); return (SET_ERROR(EIO)); } err = dbuf_read(db, NULL, DB_RF_CANFAIL); if (err) { DNODE_STAT_BUMP(dnode_hold_dbuf_read); dbuf_rele(db, FTAG); return (err); } ASSERT3U(db->db.db_size, >=, 1<db.db_size >> DNODE_SHIFT; idx = object & (epb - 1); dn_block = (dnode_phys_t *)db->db.db_data; ASSERT(DB_DNODE(db)->dn_type == DMU_OT_DNODE); dnc = dmu_buf_get_user(&db->db); dnh = NULL; if (dnc == NULL) { dnode_children_t *winner; int skip = 0; dnc = kmem_zalloc(sizeof (dnode_children_t) + epb * sizeof (dnode_handle_t), KM_SLEEP); dnc->dnc_count = epb; dnh = &dnc->dnc_children[0]; /* Initialize dnode slot status from dnode_phys_t */ for (int i = 0; i < epb; i++) { zrl_init(&dnh[i].dnh_zrlock); if (skip) { skip--; continue; } if (dn_block[i].dn_type != DMU_OT_NONE) { int interior = dn_block[i].dn_extra_slots; dnode_set_slots(dnc, i, 1, DN_SLOT_ALLOCATED); dnode_set_slots(dnc, i + 1, interior, DN_SLOT_INTERIOR); skip = interior; } else { dnh[i].dnh_dnode = DN_SLOT_FREE; skip = 0; } } dmu_buf_init_user(&dnc->dnc_dbu, NULL, dnode_buf_evict_async, NULL); winner = dmu_buf_set_user(&db->db, &dnc->dnc_dbu); if (winner != NULL) { for (int i = 0; i < epb; i++) zrl_destroy(&dnh[i].dnh_zrlock); kmem_free(dnc, sizeof (dnode_children_t) + epb * sizeof (dnode_handle_t)); dnc = winner; } } ASSERT(dnc->dnc_count == epb); dn = DN_SLOT_UNINIT; if (flag & DNODE_MUST_BE_ALLOCATED) { slots = 1; while (dn == DN_SLOT_UNINIT) { dnode_slots_hold(dnc, idx, slots); dnh = &dnc->dnc_children[idx]; if (DN_SLOT_IS_PTR(dnh->dnh_dnode)) { dn = dnh->dnh_dnode; break; } else if (dnh->dnh_dnode == DN_SLOT_INTERIOR) { DNODE_STAT_BUMP(dnode_hold_alloc_interior); dnode_slots_rele(dnc, idx, slots); dbuf_rele(db, FTAG); return (SET_ERROR(EEXIST)); } else if (dnh->dnh_dnode != DN_SLOT_ALLOCATED) { DNODE_STAT_BUMP(dnode_hold_alloc_misses); dnode_slots_rele(dnc, idx, slots); dbuf_rele(db, FTAG); return (SET_ERROR(ENOENT)); } dnode_slots_rele(dnc, idx, slots); if (!dnode_slots_tryenter(dnc, idx, slots)) { DNODE_STAT_BUMP(dnode_hold_alloc_lock_retry); continue; } /* * Someone else won the race and called dnode_create() * after we checked DN_SLOT_IS_PTR() above but before * we acquired the lock. */ if (DN_SLOT_IS_PTR(dnh->dnh_dnode)) { DNODE_STAT_BUMP(dnode_hold_alloc_lock_misses); dn = dnh->dnh_dnode; } else { dn = dnode_create(os, dn_block + idx, db, object, dnh); } } mutex_enter(&dn->dn_mtx); if (dn->dn_type == DMU_OT_NONE || dn->dn_free_txg != 0) { DNODE_STAT_BUMP(dnode_hold_alloc_type_none); mutex_exit(&dn->dn_mtx); dnode_slots_rele(dnc, idx, slots); dbuf_rele(db, FTAG); return (SET_ERROR(ENOENT)); } DNODE_STAT_BUMP(dnode_hold_alloc_hits); } else if (flag & DNODE_MUST_BE_FREE) { if (idx + slots - 1 >= DNODES_PER_BLOCK) { DNODE_STAT_BUMP(dnode_hold_free_overflow); dbuf_rele(db, FTAG); return (SET_ERROR(ENOSPC)); } while (dn == DN_SLOT_UNINIT) { dnode_slots_hold(dnc, idx, slots); if (!dnode_check_slots_free(dnc, idx, slots)) { DNODE_STAT_BUMP(dnode_hold_free_misses); dnode_slots_rele(dnc, idx, slots); dbuf_rele(db, FTAG); return (SET_ERROR(ENOSPC)); } dnode_slots_rele(dnc, idx, slots); if (!dnode_slots_tryenter(dnc, idx, slots)) { DNODE_STAT_BUMP(dnode_hold_free_lock_retry); continue; } if (!dnode_check_slots_free(dnc, idx, slots)) { DNODE_STAT_BUMP(dnode_hold_free_lock_misses); dnode_slots_rele(dnc, idx, slots); dbuf_rele(db, FTAG); return (SET_ERROR(ENOSPC)); } /* * Allocated but otherwise free dnodes which would * be in the interior of a multi-slot dnodes need * to be freed. Single slot dnodes can be safely * re-purposed as a performance optimization. */ if (slots > 1) dnode_reclaim_slots(dnc, idx + 1, slots - 1); dnh = &dnc->dnc_children[idx]; if (DN_SLOT_IS_PTR(dnh->dnh_dnode)) { dn = dnh->dnh_dnode; } else { dn = dnode_create(os, dn_block + idx, db, object, dnh); } } mutex_enter(&dn->dn_mtx); if (!refcount_is_zero(&dn->dn_holds) || dn->dn_free_txg) { DNODE_STAT_BUMP(dnode_hold_free_refcount); mutex_exit(&dn->dn_mtx); dnode_slots_rele(dnc, idx, slots); dbuf_rele(db, FTAG); return (SET_ERROR(EEXIST)); } dnode_set_slots(dnc, idx + 1, slots - 1, DN_SLOT_INTERIOR); DNODE_STAT_BUMP(dnode_hold_free_hits); } else { dbuf_rele(db, FTAG); return (SET_ERROR(EINVAL)); } if (dn->dn_free_txg) { DNODE_STAT_BUMP(dnode_hold_free_txg); type = dn->dn_type; mutex_exit(&dn->dn_mtx); dnode_slots_rele(dnc, idx, slots); dbuf_rele(db, FTAG); return (SET_ERROR((flag & DNODE_MUST_BE_ALLOCATED) ? ENOENT : EEXIST)); } if (refcount_add(&dn->dn_holds, tag) == 1) dbuf_add_ref(db, dnh); mutex_exit(&dn->dn_mtx); /* Now we can rely on the hold to prevent the dnode from moving. */ dnode_slots_rele(dnc, idx, slots); DNODE_VERIFY(dn); ASSERT3P(dn->dn_dbuf, ==, db); ASSERT3U(dn->dn_object, ==, object); dbuf_rele(db, FTAG); *dnp = dn; return (0); } /* * Return held dnode if the object is allocated, NULL if not. */ int dnode_hold(objset_t *os, uint64_t object, void *tag, dnode_t **dnp) { return (dnode_hold_impl(os, object, DNODE_MUST_BE_ALLOCATED, 0, tag, dnp)); } /* * Can only add a reference if there is already at least one * reference on the dnode. Returns FALSE if unable to add a * new reference. */ boolean_t dnode_add_ref(dnode_t *dn, void *tag) { mutex_enter(&dn->dn_mtx); if (refcount_is_zero(&dn->dn_holds)) { mutex_exit(&dn->dn_mtx); return (FALSE); } VERIFY(1 < refcount_add(&dn->dn_holds, tag)); mutex_exit(&dn->dn_mtx); return (TRUE); } void dnode_rele(dnode_t *dn, void *tag) { mutex_enter(&dn->dn_mtx); dnode_rele_and_unlock(dn, tag, B_FALSE); } void dnode_rele_and_unlock(dnode_t *dn, void *tag, boolean_t evicting) { uint64_t refs; /* Get while the hold prevents the dnode from moving. */ dmu_buf_impl_t *db = dn->dn_dbuf; dnode_handle_t *dnh = dn->dn_handle; refs = refcount_remove(&dn->dn_holds, tag); mutex_exit(&dn->dn_mtx); /* * It's unsafe to release the last hold on a dnode by dnode_rele() or * indirectly by dbuf_rele() while relying on the dnode handle to * prevent the dnode from moving, since releasing the last hold could * result in the dnode's parent dbuf evicting its dnode handles. For * that reason anyone calling dnode_rele() or dbuf_rele() without some * other direct or indirect hold on the dnode must first drop the dnode * handle. */ ASSERT(refs > 0 || dnh->dnh_zrlock.zr_owner != curthread); /* NOTE: the DNODE_DNODE does not have a dn_dbuf */ if (refs == 0 && db != NULL) { /* * Another thread could add a hold to the dnode handle in * dnode_hold_impl() while holding the parent dbuf. Since the * hold on the parent dbuf prevents the handle from being * destroyed, the hold on the handle is OK. We can't yet assert * that the handle has zero references, but that will be * asserted anyway when the handle gets destroyed. */ mutex_enter(&db->db_mtx); dbuf_rele_and_unlock(db, dnh, evicting); } } void dnode_setdirty(dnode_t *dn, dmu_tx_t *tx) { objset_t *os = dn->dn_objset; uint64_t txg = tx->tx_txg; if (DMU_OBJECT_IS_SPECIAL(dn->dn_object)) { dsl_dataset_dirty(os->os_dsl_dataset, tx); return; } DNODE_VERIFY(dn); #ifdef ZFS_DEBUG mutex_enter(&dn->dn_mtx); ASSERT(dn->dn_phys->dn_type || dn->dn_allocated_txg); ASSERT(dn->dn_free_txg == 0 || dn->dn_free_txg >= txg); mutex_exit(&dn->dn_mtx); #endif /* * Determine old uid/gid when necessary */ dmu_objset_userquota_get_ids(dn, B_TRUE, tx); multilist_t *dirtylist = os->os_dirty_dnodes[txg & TXG_MASK]; multilist_sublist_t *mls = multilist_sublist_lock_obj(dirtylist, dn); /* * If we are already marked dirty, we're done. */ if (list_link_active(&dn->dn_dirty_link[txg & TXG_MASK])) { multilist_sublist_unlock(mls); return; } ASSERT(!refcount_is_zero(&dn->dn_holds) || !avl_is_empty(&dn->dn_dbufs)); ASSERT(dn->dn_datablksz != 0); ASSERT0(dn->dn_next_bonuslen[txg&TXG_MASK]); ASSERT0(dn->dn_next_blksz[txg&TXG_MASK]); ASSERT0(dn->dn_next_bonustype[txg&TXG_MASK]); dprintf_ds(os->os_dsl_dataset, "obj=%llu txg=%llu\n", dn->dn_object, txg); multilist_sublist_insert_head(mls, dn); multilist_sublist_unlock(mls); /* * The dnode maintains a hold on its containing dbuf as * long as there are holds on it. Each instantiated child * dbuf maintains a hold on the dnode. When the last child * drops its hold, the dnode will drop its hold on the * containing dbuf. We add a "dirty hold" here so that the * dnode will hang around after we finish processing its * children. */ VERIFY(dnode_add_ref(dn, (void *)(uintptr_t)tx->tx_txg)); (void) dbuf_dirty(dn->dn_dbuf, tx); dsl_dataset_dirty(os->os_dsl_dataset, tx); } void dnode_free(dnode_t *dn, dmu_tx_t *tx) { mutex_enter(&dn->dn_mtx); if (dn->dn_type == DMU_OT_NONE || dn->dn_free_txg) { mutex_exit(&dn->dn_mtx); return; } dn->dn_free_txg = tx->tx_txg; mutex_exit(&dn->dn_mtx); dnode_setdirty(dn, tx); } /* * Try to change the block size for the indicated dnode. This can only * succeed if there are no blocks allocated or dirty beyond first block */ int dnode_set_blksz(dnode_t *dn, uint64_t size, int ibs, dmu_tx_t *tx) { dmu_buf_impl_t *db; int err; ASSERT3U(size, <=, spa_maxblocksize(dmu_objset_spa(dn->dn_objset))); if (size == 0) size = SPA_MINBLOCKSIZE; else size = P2ROUNDUP(size, SPA_MINBLOCKSIZE); if (ibs == dn->dn_indblkshift) ibs = 0; if (size >> SPA_MINBLOCKSHIFT == dn->dn_datablkszsec && ibs == 0) return (0); rw_enter(&dn->dn_struct_rwlock, RW_WRITER); /* Check for any allocated blocks beyond the first */ if (dn->dn_maxblkid != 0) goto fail; mutex_enter(&dn->dn_dbufs_mtx); for (db = avl_first(&dn->dn_dbufs); db != NULL; db = AVL_NEXT(&dn->dn_dbufs, db)) { if (db->db_blkid != 0 && db->db_blkid != DMU_BONUS_BLKID && db->db_blkid != DMU_SPILL_BLKID) { mutex_exit(&dn->dn_dbufs_mtx); goto fail; } } mutex_exit(&dn->dn_dbufs_mtx); if (ibs && dn->dn_nlevels != 1) goto fail; /* resize the old block */ err = dbuf_hold_impl(dn, 0, 0, TRUE, FALSE, FTAG, &db); if (err == 0) dbuf_new_size(db, size, tx); else if (err != ENOENT) goto fail; dnode_setdblksz(dn, size); dnode_setdirty(dn, tx); dn->dn_next_blksz[tx->tx_txg&TXG_MASK] = size; if (ibs) { dn->dn_indblkshift = ibs; dn->dn_next_indblkshift[tx->tx_txg&TXG_MASK] = ibs; } /* rele after we have fixed the blocksize in the dnode */ if (db) dbuf_rele(db, FTAG); rw_exit(&dn->dn_struct_rwlock); return (0); fail: rw_exit(&dn->dn_struct_rwlock); return (SET_ERROR(ENOTSUP)); } /* read-holding callers must not rely on the lock being continuously held */ void dnode_new_blkid(dnode_t *dn, uint64_t blkid, dmu_tx_t *tx, boolean_t have_read) { uint64_t txgoff = tx->tx_txg & TXG_MASK; int epbs, new_nlevels; uint64_t sz; ASSERT(blkid != DMU_BONUS_BLKID); ASSERT(have_read ? RW_READ_HELD(&dn->dn_struct_rwlock) : RW_WRITE_HELD(&dn->dn_struct_rwlock)); /* * if we have a read-lock, check to see if we need to do any work * before upgrading to a write-lock. */ if (have_read) { if (blkid <= dn->dn_maxblkid) return; if (!rw_tryupgrade(&dn->dn_struct_rwlock)) { rw_exit(&dn->dn_struct_rwlock); rw_enter(&dn->dn_struct_rwlock, RW_WRITER); } } if (blkid <= dn->dn_maxblkid) goto out; dn->dn_maxblkid = blkid; /* * Compute the number of levels necessary to support the new maxblkid. */ new_nlevels = 1; epbs = dn->dn_indblkshift - SPA_BLKPTRSHIFT; for (sz = dn->dn_nblkptr; sz <= blkid && sz >= dn->dn_nblkptr; sz <<= epbs) new_nlevels++; if (new_nlevels > dn->dn_nlevels) { int old_nlevels = dn->dn_nlevels; dmu_buf_impl_t *db; list_t *list; dbuf_dirty_record_t *new, *dr, *dr_next; dn->dn_nlevels = new_nlevels; ASSERT3U(new_nlevels, >, dn->dn_next_nlevels[txgoff]); dn->dn_next_nlevels[txgoff] = new_nlevels; /* dirty the left indirects */ db = dbuf_hold_level(dn, old_nlevels, 0, FTAG); ASSERT(db != NULL); new = dbuf_dirty(db, tx); dbuf_rele(db, FTAG); /* transfer the dirty records to the new indirect */ mutex_enter(&dn->dn_mtx); mutex_enter(&new->dt.di.dr_mtx); list = &dn->dn_dirty_records[txgoff]; for (dr = list_head(list); dr; dr = dr_next) { dr_next = list_next(&dn->dn_dirty_records[txgoff], dr); if (dr->dr_dbuf->db_level != new_nlevels-1 && dr->dr_dbuf->db_blkid != DMU_BONUS_BLKID && dr->dr_dbuf->db_blkid != DMU_SPILL_BLKID) { ASSERT(dr->dr_dbuf->db_level == old_nlevels-1); list_remove(&dn->dn_dirty_records[txgoff], dr); list_insert_tail(&new->dt.di.dr_children, dr); dr->dr_parent = new; } } mutex_exit(&new->dt.di.dr_mtx); mutex_exit(&dn->dn_mtx); } out: if (have_read) rw_downgrade(&dn->dn_struct_rwlock); } static void dnode_dirty_l1(dnode_t *dn, uint64_t l1blkid, dmu_tx_t *tx) { dmu_buf_impl_t *db = dbuf_hold_level(dn, 1, l1blkid, FTAG); if (db != NULL) { dmu_buf_will_dirty(&db->db, tx); dbuf_rele(db, FTAG); } } /* * Dirty all the in-core level-1 dbufs in the range specified by start_blkid * and end_blkid. */ static void dnode_dirty_l1range(dnode_t *dn, uint64_t start_blkid, uint64_t end_blkid, dmu_tx_t *tx) { dmu_buf_impl_t db_search; dmu_buf_impl_t *db; avl_index_t where; mutex_enter(&dn->dn_dbufs_mtx); db_search.db_level = 1; db_search.db_blkid = start_blkid + 1; db_search.db_state = DB_SEARCH; for (;;) { db = avl_find(&dn->dn_dbufs, &db_search, &where); if (db == NULL) db = avl_nearest(&dn->dn_dbufs, where, AVL_AFTER); if (db == NULL || db->db_level != 1 || db->db_blkid >= end_blkid) { break; } /* * Setup the next blkid we want to search for. */ db_search.db_blkid = db->db_blkid + 1; ASSERT3U(db->db_blkid, >=, start_blkid); /* * If the dbuf transitions to DB_EVICTING while we're trying * to dirty it, then we will be unable to discover it in * the dbuf hash table. This will result in a call to * dbuf_create() which needs to acquire the dn_dbufs_mtx * lock. To avoid a deadlock, we drop the lock before * dirtying the level-1 dbuf. */ mutex_exit(&dn->dn_dbufs_mtx); dnode_dirty_l1(dn, db->db_blkid, tx); mutex_enter(&dn->dn_dbufs_mtx); } #ifdef ZFS_DEBUG /* * Walk all the in-core level-1 dbufs and verify they have been dirtied. */ db_search.db_level = 1; db_search.db_blkid = start_blkid + 1; db_search.db_state = DB_SEARCH; db = avl_find(&dn->dn_dbufs, &db_search, &where); if (db == NULL) db = avl_nearest(&dn->dn_dbufs, where, AVL_AFTER); for (; db != NULL; db = AVL_NEXT(&dn->dn_dbufs, db)) { if (db->db_level != 1 || db->db_blkid >= end_blkid) break; ASSERT(db->db_dirtycnt > 0); } #endif mutex_exit(&dn->dn_dbufs_mtx); } void dnode_free_range(dnode_t *dn, uint64_t off, uint64_t len, dmu_tx_t *tx) { dmu_buf_impl_t *db; uint64_t blkoff, blkid, nblks; int blksz, blkshift, head, tail; int trunc = FALSE; int epbs; rw_enter(&dn->dn_struct_rwlock, RW_WRITER); blksz = dn->dn_datablksz; blkshift = dn->dn_datablkshift; epbs = dn->dn_indblkshift - SPA_BLKPTRSHIFT; if (len == DMU_OBJECT_END) { len = UINT64_MAX - off; trunc = TRUE; } /* * First, block align the region to free: */ if (ISP2(blksz)) { head = P2NPHASE(off, blksz); blkoff = P2PHASE(off, blksz); if ((off >> blkshift) > dn->dn_maxblkid) goto out; } else { ASSERT(dn->dn_maxblkid == 0); if (off == 0 && len >= blksz) { /* * Freeing the whole block; fast-track this request. */ blkid = 0; nblks = 1; if (dn->dn_nlevels > 1) dnode_dirty_l1(dn, 0, tx); goto done; } else if (off >= blksz) { /* Freeing past end-of-data */ goto out; } else { /* Freeing part of the block. */ head = blksz - off; ASSERT3U(head, >, 0); } blkoff = off; } /* zero out any partial block data at the start of the range */ if (head) { ASSERT3U(blkoff + head, ==, blksz); if (len < head) head = len; if (dbuf_hold_impl(dn, 0, dbuf_whichblock(dn, 0, off), TRUE, FALSE, FTAG, &db) == 0) { caddr_t data; /* don't dirty if it isn't on disk and isn't dirty */ if (db->db_last_dirty || (db->db_blkptr && !BP_IS_HOLE(db->db_blkptr))) { rw_exit(&dn->dn_struct_rwlock); dmu_buf_will_dirty(&db->db, tx); rw_enter(&dn->dn_struct_rwlock, RW_WRITER); data = db->db.db_data; bzero(data + blkoff, head); } dbuf_rele(db, FTAG); } off += head; len -= head; } /* If the range was less than one block, we're done */ if (len == 0) goto out; /* If the remaining range is past end of file, we're done */ if ((off >> blkshift) > dn->dn_maxblkid) goto out; ASSERT(ISP2(blksz)); if (trunc) tail = 0; else tail = P2PHASE(len, blksz); ASSERT0(P2PHASE(off, blksz)); /* zero out any partial block data at the end of the range */ if (tail) { if (len < tail) tail = len; if (dbuf_hold_impl(dn, 0, dbuf_whichblock(dn, 0, off+len), TRUE, FALSE, FTAG, &db) == 0) { /* don't dirty if not on disk and not dirty */ if (db->db_last_dirty || (db->db_blkptr && !BP_IS_HOLE(db->db_blkptr))) { rw_exit(&dn->dn_struct_rwlock); dmu_buf_will_dirty(&db->db, tx); rw_enter(&dn->dn_struct_rwlock, RW_WRITER); bzero(db->db.db_data, tail); } dbuf_rele(db, FTAG); } len -= tail; } /* If the range did not include a full block, we are done */ if (len == 0) goto out; ASSERT(IS_P2ALIGNED(off, blksz)); ASSERT(trunc || IS_P2ALIGNED(len, blksz)); blkid = off >> blkshift; nblks = len >> blkshift; if (trunc) nblks += 1; /* * Dirty all the indirect blocks in this range. Note that only * the first and last indirect blocks can actually be written * (if they were partially freed) -- they must be dirtied, even if * they do not exist on disk yet. The interior blocks will * be freed by free_children(), so they will not actually be written. * Even though these interior blocks will not be written, we * dirty them for two reasons: * * - It ensures that the indirect blocks remain in memory until * syncing context. (They have already been prefetched by * dmu_tx_hold_free(), so we don't have to worry about reading * them serially here.) * * - The dirty space accounting will put pressure on the txg sync * mechanism to begin syncing, and to delay transactions if there * is a large amount of freeing. Even though these indirect * blocks will not be written, we could need to write the same * amount of space if we copy the freed BPs into deadlists. */ if (dn->dn_nlevels > 1) { uint64_t first, last; first = blkid >> epbs; dnode_dirty_l1(dn, first, tx); if (trunc) last = dn->dn_maxblkid >> epbs; else last = (blkid + nblks - 1) >> epbs; if (last != first) dnode_dirty_l1(dn, last, tx); dnode_dirty_l1range(dn, first, last, tx); int shift = dn->dn_datablkshift + dn->dn_indblkshift - SPA_BLKPTRSHIFT; for (uint64_t i = first + 1; i < last; i++) { /* * Set i to the blockid of the next non-hole * level-1 indirect block at or after i. Note * that dnode_next_offset() operates in terms of * level-0-equivalent bytes. */ uint64_t ibyte = i << shift; int err = dnode_next_offset(dn, DNODE_FIND_HAVELOCK, &ibyte, 2, 1, 0); i = ibyte >> shift; if (i >= last) break; /* * Normally we should not see an error, either * from dnode_next_offset() or dbuf_hold_level() * (except for ESRCH from dnode_next_offset). * If there is an i/o error, then when we read * this block in syncing context, it will use * ZIO_FLAG_MUSTSUCCEED, and thus hang/panic according * to the "failmode" property. dnode_next_offset() * doesn't have a flag to indicate MUSTSUCCEED. */ if (err != 0) break; dnode_dirty_l1(dn, i, tx); } } done: /* * Add this range to the dnode range list. * We will finish up this free operation in the syncing phase. */ mutex_enter(&dn->dn_mtx); int txgoff = tx->tx_txg & TXG_MASK; if (dn->dn_free_ranges[txgoff] == NULL) { dn->dn_free_ranges[txgoff] = range_tree_create(NULL, NULL); } range_tree_clear(dn->dn_free_ranges[txgoff], blkid, nblks); range_tree_add(dn->dn_free_ranges[txgoff], blkid, nblks); dprintf_dnode(dn, "blkid=%llu nblks=%llu txg=%llu\n", blkid, nblks, tx->tx_txg); mutex_exit(&dn->dn_mtx); dbuf_free_range(dn, blkid, blkid + nblks - 1, tx); dnode_setdirty(dn, tx); out: rw_exit(&dn->dn_struct_rwlock); } static boolean_t dnode_spill_freed(dnode_t *dn) { int i; mutex_enter(&dn->dn_mtx); for (i = 0; i < TXG_SIZE; i++) { if (dn->dn_rm_spillblk[i] == DN_KILL_SPILLBLK) break; } mutex_exit(&dn->dn_mtx); return (i < TXG_SIZE); } /* return TRUE if this blkid was freed in a recent txg, or FALSE if it wasn't */ uint64_t dnode_block_freed(dnode_t *dn, uint64_t blkid) { void *dp = spa_get_dsl(dn->dn_objset->os_spa); int i; if (blkid == DMU_BONUS_BLKID) return (FALSE); /* * If we're in the process of opening the pool, dp will not be * set yet, but there shouldn't be anything dirty. */ if (dp == NULL) return (FALSE); if (dn->dn_free_txg) return (TRUE); if (blkid == DMU_SPILL_BLKID) return (dnode_spill_freed(dn)); mutex_enter(&dn->dn_mtx); for (i = 0; i < TXG_SIZE; i++) { if (dn->dn_free_ranges[i] != NULL && range_tree_contains(dn->dn_free_ranges[i], blkid, 1)) break; } mutex_exit(&dn->dn_mtx); return (i < TXG_SIZE); } /* call from syncing context when we actually write/free space for this dnode */ void dnode_diduse_space(dnode_t *dn, int64_t delta) { uint64_t space; dprintf_dnode(dn, "dn=%p dnp=%p used=%llu delta=%lld\n", dn, dn->dn_phys, (u_longlong_t)dn->dn_phys->dn_used, (longlong_t)delta); mutex_enter(&dn->dn_mtx); space = DN_USED_BYTES(dn->dn_phys); if (delta > 0) { ASSERT3U(space + delta, >=, space); /* no overflow */ } else { ASSERT3U(space, >=, -delta); /* no underflow */ } space += delta; if (spa_version(dn->dn_objset->os_spa) < SPA_VERSION_DNODE_BYTES) { ASSERT((dn->dn_phys->dn_flags & DNODE_FLAG_USED_BYTES) == 0); ASSERT0(P2PHASE(space, 1<dn_phys->dn_used = space >> DEV_BSHIFT; } else { dn->dn_phys->dn_used = space; dn->dn_phys->dn_flags |= DNODE_FLAG_USED_BYTES; } mutex_exit(&dn->dn_mtx); } /* * Scans a block at the indicated "level" looking for a hole or data, * depending on 'flags'. * * If level > 0, then we are scanning an indirect block looking at its * pointers. If level == 0, then we are looking at a block of dnodes. * * If we don't find what we are looking for in the block, we return ESRCH. * Otherwise, return with *offset pointing to the beginning (if searching * forwards) or end (if searching backwards) of the range covered by the * block pointer we matched on (or dnode). * * The basic search algorithm used below by dnode_next_offset() is to * use this function to search up the block tree (widen the search) until * we find something (i.e., we don't return ESRCH) and then search back * down the tree (narrow the search) until we reach our original search * level. */ static int dnode_next_offset_level(dnode_t *dn, int flags, uint64_t *offset, int lvl, uint64_t blkfill, uint64_t txg) { dmu_buf_impl_t *db = NULL; void *data = NULL; uint64_t epbs = dn->dn_phys->dn_indblkshift - SPA_BLKPTRSHIFT; uint64_t epb = 1ULL << epbs; uint64_t minfill, maxfill; boolean_t hole; int i, inc, error, span; dprintf("probing object %llu offset %llx level %d of %u\n", dn->dn_object, *offset, lvl, dn->dn_phys->dn_nlevels); hole = ((flags & DNODE_FIND_HOLE) != 0); inc = (flags & DNODE_FIND_BACKWARDS) ? -1 : 1; ASSERT(txg == 0 || !hole); if (lvl == dn->dn_phys->dn_nlevels) { error = 0; epb = dn->dn_phys->dn_nblkptr; data = dn->dn_phys->dn_blkptr; } else { uint64_t blkid = dbuf_whichblock(dn, lvl, *offset); error = dbuf_hold_impl(dn, lvl, blkid, TRUE, FALSE, FTAG, &db); if (error) { if (error != ENOENT) return (error); if (hole) return (0); /* * This can only happen when we are searching up * the block tree for data. We don't really need to * adjust the offset, as we will just end up looking * at the pointer to this block in its parent, and its * going to be unallocated, so we will skip over it. */ return (SET_ERROR(ESRCH)); } error = dbuf_read(db, NULL, DB_RF_CANFAIL | DB_RF_HAVESTRUCT); if (error) { dbuf_rele(db, FTAG); return (error); } data = db->db.db_data; } if (db != NULL && txg != 0 && (db->db_blkptr == NULL || db->db_blkptr->blk_birth <= txg || BP_IS_HOLE(db->db_blkptr))) { /* * This can only happen when we are searching up the tree * and these conditions mean that we need to keep climbing. */ error = SET_ERROR(ESRCH); } else if (lvl == 0) { dnode_phys_t *dnp = data; ASSERT(dn->dn_type == DMU_OT_DNODE); ASSERT(!(flags & DNODE_FIND_BACKWARDS)); for (i = (*offset >> DNODE_SHIFT) & (blkfill - 1); i < blkfill; i += dnp[i].dn_extra_slots + 1) { if ((dnp[i].dn_type == DMU_OT_NONE) == hole) break; } if (i == blkfill) error = SET_ERROR(ESRCH); *offset = (*offset & ~(DNODE_BLOCK_SIZE - 1)) + (i << DNODE_SHIFT); } else { blkptr_t *bp = data; uint64_t start = *offset; span = (lvl - 1) * epbs + dn->dn_datablkshift; minfill = 0; maxfill = blkfill << ((lvl - 1) * epbs); if (hole) maxfill--; else minfill++; *offset = *offset >> span; for (i = BF64_GET(*offset, 0, epbs); i >= 0 && i < epb; i += inc) { if (BP_GET_FILL(&bp[i]) >= minfill && BP_GET_FILL(&bp[i]) <= maxfill && (hole || bp[i].blk_birth > txg)) break; if (inc > 0 || *offset > 0) *offset += inc; } *offset = *offset << span; if (inc < 0) { /* traversing backwards; position offset at the end */ ASSERT3U(*offset, <=, start); *offset = MIN(*offset + (1ULL << span) - 1, start); } else if (*offset < start) { *offset = start; } if (i < 0 || i >= epb) error = SET_ERROR(ESRCH); } if (db) dbuf_rele(db, FTAG); return (error); } /* * Find the next hole, data, or sparse region at or after *offset. * The value 'blkfill' tells us how many items we expect to find * in an L0 data block; this value is 1 for normal objects, * DNODES_PER_BLOCK for the meta dnode, and some fraction of * DNODES_PER_BLOCK when searching for sparse regions thereof. * * Examples: * * dnode_next_offset(dn, flags, offset, 1, 1, 0); * Finds the next/previous hole/data in a file. * Used in dmu_offset_next(). * * dnode_next_offset(mdn, flags, offset, 0, DNODES_PER_BLOCK, txg); * Finds the next free/allocated dnode an objset's meta-dnode. * Only finds objects that have new contents since txg (ie. * bonus buffer changes and content removal are ignored). * Used in dmu_object_next(). * * dnode_next_offset(mdn, DNODE_FIND_HOLE, offset, 2, DNODES_PER_BLOCK >> 2, 0); * Finds the next L2 meta-dnode bp that's at most 1/4 full. * Used in dmu_object_alloc(). */ int dnode_next_offset(dnode_t *dn, int flags, uint64_t *offset, int minlvl, uint64_t blkfill, uint64_t txg) { uint64_t initial_offset = *offset; int lvl, maxlvl; int error = 0; if (!(flags & DNODE_FIND_HAVELOCK)) rw_enter(&dn->dn_struct_rwlock, RW_READER); if (dn->dn_phys->dn_nlevels == 0) { error = SET_ERROR(ESRCH); goto out; } if (dn->dn_datablkshift == 0) { if (*offset < dn->dn_datablksz) { if (flags & DNODE_FIND_HOLE) *offset = dn->dn_datablksz; } else { error = SET_ERROR(ESRCH); } goto out; } maxlvl = dn->dn_phys->dn_nlevels; for (lvl = minlvl; lvl <= maxlvl; lvl++) { error = dnode_next_offset_level(dn, flags, offset, lvl, blkfill, txg); if (error != ESRCH) break; } while (error == 0 && --lvl >= minlvl) { error = dnode_next_offset_level(dn, flags, offset, lvl, blkfill, txg); } /* * There's always a "virtual hole" at the end of the object, even * if all BP's which physically exist are non-holes. */ if ((flags & DNODE_FIND_HOLE) && error == ESRCH && txg == 0 && minlvl == 1 && blkfill == 1 && !(flags & DNODE_FIND_BACKWARDS)) { error = 0; } if (error == 0 && (flags & DNODE_FIND_BACKWARDS ? initial_offset < *offset : initial_offset > *offset)) error = SET_ERROR(ESRCH); out: if (!(flags & DNODE_FIND_HAVELOCK)) rw_exit(&dn->dn_struct_rwlock); return (error); } Index: head/sys/cddl/contrib/opensolaris =================================================================== --- head/sys/cddl/contrib/opensolaris (revision 353555) +++ head/sys/cddl/contrib/opensolaris (revision 353556) Property changes on: head/sys/cddl/contrib/opensolaris ___________________________________________________________________ Modified: svn:mergeinfo ## -0,0 +0,1 ## Merged /vendor-sys/illumos/dist:r353551