diff --git a/include/sys/bptree.h b/include/sys/bptree.h index 971507211875..a533cb949021 100644 --- a/include/sys/bptree.h +++ b/include/sys/bptree.h @@ -1,64 +1,65 @@ /* * 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) 2012 by Delphix. All rights reserved. + * Copyright (c) 2013 by Delphix. All rights reserved. */ #ifndef _SYS_BPTREE_H #define _SYS_BPTREE_H #include #include #ifdef __cplusplus extern "C" { #endif typedef struct bptree_phys { uint64_t bt_begin; uint64_t bt_end; uint64_t bt_bytes; uint64_t bt_comp; uint64_t bt_uncomp; } bptree_phys_t; typedef struct bptree_entry_phys { blkptr_t be_bp; uint64_t be_birth_txg; /* only delete blocks born after this txg */ zbookmark_t be_zb; /* holds traversal resume point if needed */ } bptree_entry_phys_t; typedef int bptree_itor_t(void *arg, const blkptr_t *bp, dmu_tx_t *tx); uint64_t bptree_alloc(objset_t *os, dmu_tx_t *tx); int bptree_free(objset_t *os, uint64_t obj, dmu_tx_t *tx); +boolean_t bptree_is_empty(objset_t *os, uint64_t obj); void bptree_add(objset_t *os, uint64_t obj, blkptr_t *bp, uint64_t birth_txg, uint64_t bytes, uint64_t comp, uint64_t uncomp, dmu_tx_t *tx); int bptree_iterate(objset_t *os, uint64_t obj, boolean_t free, bptree_itor_t func, void *arg, dmu_tx_t *tx); #ifdef __cplusplus } #endif #endif /* _SYS_BPTREE_H */ diff --git a/include/sys/dmu.h b/include/sys/dmu.h index 89a0e5bd7a93..aa3e8f25aa74 100644 --- a/include/sys/dmu.h +++ b/include/sys/dmu.h @@ -1,833 +1,832 @@ /* * 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) 2011, 2014 by Delphix. All rights reserved. * Copyright 2011 Nexenta Systems, Inc. All rights reserved. * Copyright (c) 2012, Joyent, Inc. All rights reserved. */ /* Portions Copyright 2010 Robert Milkowski */ #ifndef _SYS_DMU_H #define _SYS_DMU_H /* * This file describes the interface that the DMU provides for its * consumers. * * The DMU also interacts with the SPA. That interface is described in * dmu_spa.h. */ #include #include #include #include #include #include #include #ifdef __cplusplus extern "C" { #endif struct page; struct vnode; struct spa; struct zilog; struct zio; struct blkptr; struct zap_cursor; struct dsl_dataset; struct dsl_pool; struct dnode; struct drr_begin; struct drr_end; struct zbookmark; struct spa; struct nvlist; struct arc_buf; struct zio_prop; struct sa_handle; typedef struct objset objset_t; typedef struct dmu_tx dmu_tx_t; typedef struct dsl_dir dsl_dir_t; typedef enum dmu_object_byteswap { DMU_BSWAP_UINT8, DMU_BSWAP_UINT16, DMU_BSWAP_UINT32, DMU_BSWAP_UINT64, DMU_BSWAP_ZAP, DMU_BSWAP_DNODE, DMU_BSWAP_OBJSET, DMU_BSWAP_ZNODE, DMU_BSWAP_OLDACL, DMU_BSWAP_ACL, /* * Allocating a new byteswap type number makes the on-disk format * incompatible with any other format that uses the same number. * * Data can usually be structured to work with one of the * DMU_BSWAP_UINT* or DMU_BSWAP_ZAP types. */ DMU_BSWAP_NUMFUNCS } dmu_object_byteswap_t; #define DMU_OT_NEWTYPE 0x80 #define DMU_OT_METADATA 0x40 #define DMU_OT_BYTESWAP_MASK 0x3f /* * Defines a uint8_t object type. Object types specify if the data * in the object is metadata (boolean) and how to byteswap the data * (dmu_object_byteswap_t). */ #define DMU_OT(byteswap, metadata) \ (DMU_OT_NEWTYPE | \ ((metadata) ? DMU_OT_METADATA : 0) | \ ((byteswap) & DMU_OT_BYTESWAP_MASK)) #define DMU_OT_IS_VALID(ot) (((ot) & DMU_OT_NEWTYPE) ? \ ((ot) & DMU_OT_BYTESWAP_MASK) < DMU_BSWAP_NUMFUNCS : \ (ot) < DMU_OT_NUMTYPES) #define DMU_OT_IS_METADATA(ot) (((ot) & DMU_OT_NEWTYPE) ? \ ((ot) & DMU_OT_METADATA) : \ dmu_ot[(int)(ot)].ot_metadata) /* * These object types use bp_fill != 1 for their L0 bp's. Therefore they can't * have their data embedded (i.e. use a BP_IS_EMBEDDED() bp), because bp_fill * is repurposed for embedded BPs. */ #define DMU_OT_HAS_FILL(ot) \ ((ot) == DMU_OT_DNODE || (ot) == DMU_OT_OBJSET) #define DMU_OT_BYTESWAP(ot) (((ot) & DMU_OT_NEWTYPE) ? \ ((ot) & DMU_OT_BYTESWAP_MASK) : \ dmu_ot[(int)(ot)].ot_byteswap) typedef enum dmu_object_type { DMU_OT_NONE, /* general: */ DMU_OT_OBJECT_DIRECTORY, /* ZAP */ DMU_OT_OBJECT_ARRAY, /* UINT64 */ DMU_OT_PACKED_NVLIST, /* UINT8 (XDR by nvlist_pack/unpack) */ DMU_OT_PACKED_NVLIST_SIZE, /* UINT64 */ DMU_OT_BPOBJ, /* UINT64 */ DMU_OT_BPOBJ_HDR, /* UINT64 */ /* spa: */ DMU_OT_SPACE_MAP_HEADER, /* UINT64 */ DMU_OT_SPACE_MAP, /* UINT64 */ /* zil: */ DMU_OT_INTENT_LOG, /* UINT64 */ /* dmu: */ DMU_OT_DNODE, /* DNODE */ DMU_OT_OBJSET, /* OBJSET */ /* dsl: */ DMU_OT_DSL_DIR, /* UINT64 */ DMU_OT_DSL_DIR_CHILD_MAP, /* ZAP */ DMU_OT_DSL_DS_SNAP_MAP, /* ZAP */ DMU_OT_DSL_PROPS, /* ZAP */ DMU_OT_DSL_DATASET, /* UINT64 */ /* zpl: */ DMU_OT_ZNODE, /* ZNODE */ DMU_OT_OLDACL, /* Old ACL */ DMU_OT_PLAIN_FILE_CONTENTS, /* UINT8 */ DMU_OT_DIRECTORY_CONTENTS, /* ZAP */ DMU_OT_MASTER_NODE, /* ZAP */ DMU_OT_UNLINKED_SET, /* ZAP */ /* zvol: */ DMU_OT_ZVOL, /* UINT8 */ DMU_OT_ZVOL_PROP, /* ZAP */ /* other; for testing only! */ DMU_OT_PLAIN_OTHER, /* UINT8 */ DMU_OT_UINT64_OTHER, /* UINT64 */ DMU_OT_ZAP_OTHER, /* ZAP */ /* new object types: */ DMU_OT_ERROR_LOG, /* ZAP */ DMU_OT_SPA_HISTORY, /* UINT8 */ DMU_OT_SPA_HISTORY_OFFSETS, /* spa_his_phys_t */ DMU_OT_POOL_PROPS, /* ZAP */ DMU_OT_DSL_PERMS, /* ZAP */ DMU_OT_ACL, /* ACL */ DMU_OT_SYSACL, /* SYSACL */ DMU_OT_FUID, /* FUID table (Packed NVLIST UINT8) */ DMU_OT_FUID_SIZE, /* FUID table size UINT64 */ DMU_OT_NEXT_CLONES, /* ZAP */ DMU_OT_SCAN_QUEUE, /* ZAP */ DMU_OT_USERGROUP_USED, /* ZAP */ DMU_OT_USERGROUP_QUOTA, /* ZAP */ DMU_OT_USERREFS, /* ZAP */ DMU_OT_DDT_ZAP, /* ZAP */ DMU_OT_DDT_STATS, /* ZAP */ DMU_OT_SA, /* System attr */ DMU_OT_SA_MASTER_NODE, /* ZAP */ DMU_OT_SA_ATTR_REGISTRATION, /* ZAP */ DMU_OT_SA_ATTR_LAYOUTS, /* ZAP */ DMU_OT_SCAN_XLATE, /* ZAP */ DMU_OT_DEDUP, /* fake dedup BP from ddt_bp_create() */ DMU_OT_DEADLIST, /* ZAP */ DMU_OT_DEADLIST_HDR, /* UINT64 */ DMU_OT_DSL_CLONES, /* ZAP */ DMU_OT_BPOBJ_SUBOBJ, /* UINT64 */ /* * Do not allocate new object types here. Doing so makes the on-disk * format incompatible with any other format that uses the same object * type number. * * When creating an object which does not have one of the above types * use the DMU_OTN_* type with the correct byteswap and metadata * values. * * The DMU_OTN_* types do not have entries in the dmu_ot table, * use the DMU_OT_IS_METDATA() and DMU_OT_BYTESWAP() macros instead * of indexing into dmu_ot directly (this works for both DMU_OT_* types * and DMU_OTN_* types). */ DMU_OT_NUMTYPES, /* * Names for valid types declared with DMU_OT(). */ DMU_OTN_UINT8_DATA = DMU_OT(DMU_BSWAP_UINT8, B_FALSE), DMU_OTN_UINT8_METADATA = DMU_OT(DMU_BSWAP_UINT8, B_TRUE), DMU_OTN_UINT16_DATA = DMU_OT(DMU_BSWAP_UINT16, B_FALSE), DMU_OTN_UINT16_METADATA = DMU_OT(DMU_BSWAP_UINT16, B_TRUE), DMU_OTN_UINT32_DATA = DMU_OT(DMU_BSWAP_UINT32, B_FALSE), DMU_OTN_UINT32_METADATA = DMU_OT(DMU_BSWAP_UINT32, B_TRUE), DMU_OTN_UINT64_DATA = DMU_OT(DMU_BSWAP_UINT64, B_FALSE), DMU_OTN_UINT64_METADATA = DMU_OT(DMU_BSWAP_UINT64, B_TRUE), DMU_OTN_ZAP_DATA = DMU_OT(DMU_BSWAP_ZAP, B_FALSE), DMU_OTN_ZAP_METADATA = DMU_OT(DMU_BSWAP_ZAP, B_TRUE), } dmu_object_type_t; typedef enum txg_how { TXG_WAIT = 1, TXG_NOWAIT, TXG_WAITED, } txg_how_t; void byteswap_uint64_array(void *buf, size_t size); void byteswap_uint32_array(void *buf, size_t size); void byteswap_uint16_array(void *buf, size_t size); void byteswap_uint8_array(void *buf, size_t size); void zap_byteswap(void *buf, size_t size); void zfs_oldacl_byteswap(void *buf, size_t size); void zfs_acl_byteswap(void *buf, size_t size); void zfs_znode_byteswap(void *buf, size_t size); #define DS_FIND_SNAPSHOTS (1<<0) #define DS_FIND_CHILDREN (1<<1) /* * The maximum number of bytes that can be accessed as part of one * operation, including metadata. */ #define DMU_MAX_ACCESS (10<<20) /* 10MB */ #define DMU_MAX_DELETEBLKCNT (20480) /* ~5MB of indirect blocks */ #define DMU_USERUSED_OBJECT (-1ULL) #define DMU_GROUPUSED_OBJECT (-2ULL) -#define DMU_DEADLIST_OBJECT (-3ULL) /* * artificial blkids for bonus buffer and spill blocks */ #define DMU_BONUS_BLKID (-1ULL) #define DMU_SPILL_BLKID (-2ULL) /* * Public routines to create, destroy, open, and close objsets. */ int dmu_objset_hold(const char *name, void *tag, objset_t **osp); int dmu_objset_own(const char *name, dmu_objset_type_t type, boolean_t readonly, void *tag, objset_t **osp); void dmu_objset_rele(objset_t *os, void *tag); void dmu_objset_disown(objset_t *os, void *tag); int dmu_objset_open_ds(struct dsl_dataset *ds, objset_t **osp); void dmu_objset_evict_dbufs(objset_t *os); int dmu_objset_create(const char *name, dmu_objset_type_t type, uint64_t flags, void (*func)(objset_t *os, void *arg, cred_t *cr, dmu_tx_t *tx), void *arg); int dmu_objset_clone(const char *name, const char *origin); int dsl_destroy_snapshots_nvl(struct nvlist *snaps, boolean_t defer, struct nvlist *errlist); int dmu_objset_snapshot_one(const char *fsname, const char *snapname); int dmu_objset_snapshot_tmp(const char *, const char *, int); int dmu_objset_find(char *name, int func(const char *, void *), void *arg, int flags); void dmu_objset_byteswap(void *buf, size_t size); int dsl_dataset_rename_snapshot(const char *fsname, const char *oldsnapname, const char *newsnapname, boolean_t recursive); typedef struct dmu_buf { uint64_t db_object; /* object that this buffer is part of */ uint64_t db_offset; /* byte offset in this object */ uint64_t db_size; /* size of buffer in bytes */ void *db_data; /* data in buffer */ } dmu_buf_t; typedef void dmu_buf_evict_func_t(struct dmu_buf *db, void *user_ptr); /* * The names of zap entries in the DIRECTORY_OBJECT of the MOS. */ #define DMU_POOL_DIRECTORY_OBJECT 1 #define DMU_POOL_CONFIG "config" #define DMU_POOL_FEATURES_FOR_WRITE "features_for_write" #define DMU_POOL_FEATURES_FOR_READ "features_for_read" #define DMU_POOL_FEATURE_DESCRIPTIONS "feature_descriptions" #define DMU_POOL_FEATURE_ENABLED_TXG "feature_enabled_txg" #define DMU_POOL_ROOT_DATASET "root_dataset" #define DMU_POOL_SYNC_BPOBJ "sync_bplist" #define DMU_POOL_ERRLOG_SCRUB "errlog_scrub" #define DMU_POOL_ERRLOG_LAST "errlog_last" #define DMU_POOL_SPARES "spares" #define DMU_POOL_DEFLATE "deflate" #define DMU_POOL_HISTORY "history" #define DMU_POOL_PROPS "pool_props" #define DMU_POOL_L2CACHE "l2cache" #define DMU_POOL_TMP_USERREFS "tmp_userrefs" #define DMU_POOL_DDT "DDT-%s-%s-%s" #define DMU_POOL_DDT_STATS "DDT-statistics" #define DMU_POOL_CREATION_VERSION "creation_version" #define DMU_POOL_SCAN "scan" #define DMU_POOL_FREE_BPOBJ "free_bpobj" #define DMU_POOL_BPTREE_OBJ "bptree_obj" #define DMU_POOL_EMPTY_BPOBJ "empty_bpobj" /* * Allocate an object from this objset. The range of object numbers * available is (0, DN_MAX_OBJECT). Object 0 is the meta-dnode. * * The transaction must be assigned to a txg. The newly allocated * object will be "held" in the transaction (ie. you can modify the * newly allocated object in this transaction). * * dmu_object_alloc() chooses an object and returns it in *objectp. * * dmu_object_claim() allocates a specific object number. If that * number is already allocated, it fails and returns EEXIST. * * Return 0 on success, or ENOSPC or EEXIST as specified above. */ uint64_t dmu_object_alloc(objset_t *os, dmu_object_type_t ot, int blocksize, dmu_object_type_t bonus_type, int bonus_len, dmu_tx_t *tx); int dmu_object_claim(objset_t *os, uint64_t object, dmu_object_type_t ot, int blocksize, dmu_object_type_t bonus_type, int bonus_len, dmu_tx_t *tx); int dmu_object_reclaim(objset_t *os, uint64_t object, dmu_object_type_t ot, int blocksize, dmu_object_type_t bonustype, int bonuslen); /* * Free an object from this objset. * * The object's data will be freed as well (ie. you don't need to call * dmu_free(object, 0, -1, tx)). * * The object need not be held in the transaction. * * If there are any holds on this object's buffers (via dmu_buf_hold()), * or tx holds on the object (via dmu_tx_hold_object()), you can not * free it; it fails and returns EBUSY. * * If the object is not allocated, it fails and returns ENOENT. * * Return 0 on success, or EBUSY or ENOENT as specified above. */ int dmu_object_free(objset_t *os, uint64_t object, dmu_tx_t *tx); /* * Find the next allocated or free object. * * The objectp parameter is in-out. It will be updated to be the next * object which is allocated. Ignore objects which have not been * modified since txg. * * XXX Can only be called on a objset with no dirty data. * * Returns 0 on success, or ENOENT if there are no more objects. */ int dmu_object_next(objset_t *os, uint64_t *objectp, boolean_t hole, uint64_t txg); /* * Set the data blocksize for an object. * * The object cannot have any blocks allcated beyond the first. If * the first block is allocated already, the new size must be greater * than the current block size. If these conditions are not met, * ENOTSUP will be returned. * * Returns 0 on success, or EBUSY if there are any holds on the object * contents, or ENOTSUP as described above. */ int dmu_object_set_blocksize(objset_t *os, uint64_t object, uint64_t size, int ibs, dmu_tx_t *tx); /* * Set the checksum property on a dnode. The new checksum algorithm will * apply to all newly written blocks; existing blocks will not be affected. */ void dmu_object_set_checksum(objset_t *os, uint64_t object, uint8_t checksum, dmu_tx_t *tx); /* * Set the compress property on a dnode. The new compression algorithm will * apply to all newly written blocks; existing blocks will not be affected. */ void dmu_object_set_compress(objset_t *os, uint64_t object, uint8_t compress, dmu_tx_t *tx); void dmu_write_embedded(objset_t *os, uint64_t object, uint64_t offset, void *data, uint8_t etype, uint8_t comp, int uncompressed_size, int compressed_size, int byteorder, dmu_tx_t *tx); /* * Decide how to write a block: checksum, compression, number of copies, etc. */ #define WP_NOFILL 0x1 #define WP_DMU_SYNC 0x2 #define WP_SPILL 0x4 void dmu_write_policy(objset_t *os, struct dnode *dn, int level, int wp, struct zio_prop *zp); /* * The bonus data is accessed more or less like a regular buffer. * You must dmu_bonus_hold() to get the buffer, which will give you a * dmu_buf_t with db_offset==-1ULL, and db_size = the size of the bonus * data. As with any normal buffer, you must call dmu_buf_read() to * read db_data, dmu_buf_will_dirty() before modifying it, and the * object must be held in an assigned transaction before calling * dmu_buf_will_dirty. You may use dmu_buf_set_user() on the bonus * buffer as well. You must release what you hold with dmu_buf_rele(). * * Returns ENOENT, EIO, or 0. */ int dmu_bonus_hold(objset_t *os, uint64_t object, void *tag, dmu_buf_t **); int dmu_bonus_max(void); int dmu_set_bonus(dmu_buf_t *, int, dmu_tx_t *); int dmu_set_bonustype(dmu_buf_t *, dmu_object_type_t, dmu_tx_t *); dmu_object_type_t dmu_get_bonustype(dmu_buf_t *); int dmu_rm_spill(objset_t *, uint64_t, dmu_tx_t *); /* * Special spill buffer support used by "SA" framework */ int dmu_spill_hold_by_bonus(dmu_buf_t *bonus, void *tag, dmu_buf_t **dbp); int dmu_spill_hold_by_dnode(struct dnode *dn, uint32_t flags, void *tag, dmu_buf_t **dbp); int dmu_spill_hold_existing(dmu_buf_t *bonus, void *tag, dmu_buf_t **dbp); /* * Obtain the DMU buffer from the specified object which contains the * specified offset. dmu_buf_hold() puts a "hold" on the buffer, so * that it will remain in memory. You must release the hold with * dmu_buf_rele(). You must not access the dmu_buf_t after releasing * what you hold. You must have a hold on any dmu_buf_t* you pass to the DMU. * * You must call dmu_buf_read, dmu_buf_will_dirty, or dmu_buf_will_fill * on the returned buffer before reading or writing the buffer's * db_data. The comments for those routines describe what particular * operations are valid after calling them. * * The object number must be a valid, allocated object number. */ int dmu_buf_hold(objset_t *os, uint64_t object, uint64_t offset, void *tag, dmu_buf_t **, int flags); void dmu_buf_add_ref(dmu_buf_t *db, void* tag); void dmu_buf_rele(dmu_buf_t *db, void *tag); uint64_t dmu_buf_refcount(dmu_buf_t *db); /* * dmu_buf_hold_array holds the DMU buffers which contain all bytes in a * range of an object. A pointer to an array of dmu_buf_t*'s is * returned (in *dbpp). * * dmu_buf_rele_array releases the hold on an array of dmu_buf_t*'s, and * frees the array. The hold on the array of buffers MUST be released * with dmu_buf_rele_array. You can NOT release the hold on each buffer * individually with dmu_buf_rele. */ int dmu_buf_hold_array_by_bonus(dmu_buf_t *db, uint64_t offset, uint64_t length, int read, void *tag, int *numbufsp, dmu_buf_t ***dbpp); void dmu_buf_rele_array(dmu_buf_t **, int numbufs, void *tag); /* * Returns NULL on success, or the existing user ptr if it's already * been set. * * user_ptr is for use by the user and can be obtained via dmu_buf_get_user(). * * user_data_ptr_ptr should be NULL, or a pointer to a pointer which * will be set to db->db_data when you are allowed to access it. Note * that db->db_data (the pointer) can change when you do dmu_buf_read(), * dmu_buf_tryupgrade(), dmu_buf_will_dirty(), or dmu_buf_will_fill(). * *user_data_ptr_ptr will be set to the new value when it changes. * * If non-NULL, pageout func will be called when this buffer is being * excised from the cache, so that you can clean up the data structure * pointed to by user_ptr. * * dmu_evict_user() will call the pageout func for all buffers in a * objset with a given pageout func. */ void *dmu_buf_set_user(dmu_buf_t *db, void *user_ptr, void *user_data_ptr_ptr, dmu_buf_evict_func_t *pageout_func); /* * set_user_ie is the same as set_user, but request immediate eviction * when hold count goes to zero. */ void *dmu_buf_set_user_ie(dmu_buf_t *db, void *user_ptr, void *user_data_ptr_ptr, dmu_buf_evict_func_t *pageout_func); void *dmu_buf_update_user(dmu_buf_t *db_fake, void *old_user_ptr, void *user_ptr, void *user_data_ptr_ptr, dmu_buf_evict_func_t *pageout_func); void dmu_evict_user(objset_t *os, dmu_buf_evict_func_t *func); /* * Returns the user_ptr set with dmu_buf_set_user(), or NULL if not set. */ void *dmu_buf_get_user(dmu_buf_t *db); /* * Returns the blkptr associated with this dbuf, or NULL if not set. */ struct blkptr *dmu_buf_get_blkptr(dmu_buf_t *db); /* * Indicate that you are going to modify the buffer's data (db_data). * * The transaction (tx) must be assigned to a txg (ie. you've called * dmu_tx_assign()). The buffer's object must be held in the tx * (ie. you've called dmu_tx_hold_object(tx, db->db_object)). */ void dmu_buf_will_dirty(dmu_buf_t *db, dmu_tx_t *tx); /* * Tells if the given dbuf is freeable. */ boolean_t dmu_buf_freeable(dmu_buf_t *); /* * You must create a transaction, then hold the objects which you will * (or might) modify as part of this transaction. Then you must assign * the transaction to a transaction group. Once the transaction has * been assigned, you can modify buffers which belong to held objects as * part of this transaction. You can't modify buffers before the * transaction has been assigned; you can't modify buffers which don't * belong to objects which this transaction holds; you can't hold * objects once the transaction has been assigned. You may hold an * object which you are going to free (with dmu_object_free()), but you * don't have to. * * You can abort the transaction before it has been assigned. * * Note that you may hold buffers (with dmu_buf_hold) at any time, * regardless of transaction state. */ #define DMU_NEW_OBJECT (-1ULL) #define DMU_OBJECT_END (-1ULL) dmu_tx_t *dmu_tx_create(objset_t *os); void dmu_tx_hold_write(dmu_tx_t *tx, uint64_t object, uint64_t off, int len); void dmu_tx_hold_free(dmu_tx_t *tx, uint64_t object, uint64_t off, uint64_t len); void dmu_tx_hold_zap(dmu_tx_t *tx, uint64_t object, int add, const char *name); void dmu_tx_hold_bonus(dmu_tx_t *tx, uint64_t object); void dmu_tx_hold_spill(dmu_tx_t *tx, uint64_t object); void dmu_tx_hold_sa(dmu_tx_t *tx, struct sa_handle *hdl, boolean_t may_grow); void dmu_tx_hold_sa_create(dmu_tx_t *tx, int total_size); void dmu_tx_abort(dmu_tx_t *tx); int dmu_tx_assign(dmu_tx_t *tx, enum txg_how txg_how); void dmu_tx_wait(dmu_tx_t *tx); void dmu_tx_commit(dmu_tx_t *tx); /* * To register a commit callback, dmu_tx_callback_register() must be called. * * dcb_data is a pointer to caller private data that is passed on as a * callback parameter. The caller is responsible for properly allocating and * freeing it. * * When registering a callback, the transaction must be already created, but * it cannot be committed or aborted. It can be assigned to a txg or not. * * The callback will be called after the transaction has been safely written * to stable storage and will also be called if the dmu_tx is aborted. * If there is any error which prevents the transaction from being committed to * disk, the callback will be called with a value of error != 0. */ typedef void dmu_tx_callback_func_t(void *dcb_data, int error); void dmu_tx_callback_register(dmu_tx_t *tx, dmu_tx_callback_func_t *dcb_func, void *dcb_data); /* * Free up the data blocks for a defined range of a file. If size is * -1, the range from offset to end-of-file is freed. */ int dmu_free_range(objset_t *os, uint64_t object, uint64_t offset, uint64_t size, dmu_tx_t *tx); int dmu_free_long_range(objset_t *os, uint64_t object, uint64_t offset, uint64_t size); int dmu_free_long_object(objset_t *os, uint64_t object); /* * Convenience functions. * * Canfail routines will return 0 on success, or an errno if there is a * nonrecoverable I/O error. */ #define DMU_READ_PREFETCH 0 /* prefetch */ #define DMU_READ_NO_PREFETCH 1 /* don't prefetch */ int dmu_read(objset_t *os, uint64_t object, uint64_t offset, uint64_t size, void *buf, uint32_t flags); void dmu_write(objset_t *os, uint64_t object, uint64_t offset, uint64_t size, const void *buf, dmu_tx_t *tx); void dmu_prealloc(objset_t *os, uint64_t object, uint64_t offset, uint64_t size, dmu_tx_t *tx); #ifdef _KERNEL #include int dmu_read_req(objset_t *os, uint64_t object, struct request *req); int dmu_write_req(objset_t *os, uint64_t object, struct request *req, dmu_tx_t *tx); int dmu_read_uio(objset_t *os, uint64_t object, struct uio *uio, uint64_t size); int dmu_write_uio(objset_t *os, uint64_t object, struct uio *uio, uint64_t size, dmu_tx_t *tx); int dmu_write_uio_dbuf(dmu_buf_t *zdb, struct uio *uio, uint64_t size, dmu_tx_t *tx); #endif struct arc_buf *dmu_request_arcbuf(dmu_buf_t *handle, int size); void dmu_return_arcbuf(struct arc_buf *buf); void dmu_assign_arcbuf(dmu_buf_t *handle, uint64_t offset, struct arc_buf *buf, dmu_tx_t *tx); int dmu_xuio_init(struct xuio *uio, int niov); void dmu_xuio_fini(struct xuio *uio); int dmu_xuio_add(struct xuio *uio, struct arc_buf *abuf, offset_t off, size_t n); int dmu_xuio_cnt(struct xuio *uio); struct arc_buf *dmu_xuio_arcbuf(struct xuio *uio, int i); void dmu_xuio_clear(struct xuio *uio, int i); void xuio_stat_wbuf_copied(void); void xuio_stat_wbuf_nocopy(void); extern int zfs_prefetch_disable; /* * Asynchronously try to read in the data. */ void dmu_prefetch(objset_t *os, uint64_t object, uint64_t offset, uint64_t len); typedef struct dmu_object_info { /* All sizes are in bytes unless otherwise indicated. */ uint32_t doi_data_block_size; uint32_t doi_metadata_block_size; dmu_object_type_t doi_type; dmu_object_type_t doi_bonus_type; uint64_t doi_bonus_size; uint8_t doi_indirection; /* 2 = dnode->indirect->data */ uint8_t doi_checksum; uint8_t doi_compress; uint8_t doi_pad[5]; uint64_t doi_physical_blocks_512; /* data + metadata, 512b blks */ uint64_t doi_max_offset; uint64_t doi_fill_count; /* number of non-empty blocks */ } dmu_object_info_t; typedef void (*const arc_byteswap_func_t)(void *buf, size_t size); typedef struct dmu_object_type_info { dmu_object_byteswap_t ot_byteswap; boolean_t ot_metadata; char *ot_name; } dmu_object_type_info_t; typedef const struct dmu_object_byteswap_info { arc_byteswap_func_t ob_func; char *ob_name; } dmu_object_byteswap_info_t; extern const dmu_object_type_info_t dmu_ot[DMU_OT_NUMTYPES]; extern const dmu_object_byteswap_info_t dmu_ot_byteswap[DMU_BSWAP_NUMFUNCS]; /* * Get information on a DMU object. * * Return 0 on success or ENOENT if object is not allocated. * * If doi is NULL, just indicates whether the object exists. */ int dmu_object_info(objset_t *os, uint64_t object, dmu_object_info_t *doi); void __dmu_object_info_from_dnode(struct dnode *dn, dmu_object_info_t *doi); /* Like dmu_object_info, but faster if you have a held dnode in hand. */ void dmu_object_info_from_dnode(struct dnode *dn, dmu_object_info_t *doi); /* Like dmu_object_info, but faster if you have a held dbuf in hand. */ void dmu_object_info_from_db(dmu_buf_t *db, dmu_object_info_t *doi); /* * Like dmu_object_info_from_db, but faster still when you only care about * the size. This is specifically optimized for zfs_getattr(). */ void dmu_object_size_from_db(dmu_buf_t *db, uint32_t *blksize, u_longlong_t *nblk512); typedef struct dmu_objset_stats { uint64_t dds_num_clones; /* number of clones of this */ uint64_t dds_creation_txg; uint64_t dds_guid; dmu_objset_type_t dds_type; uint8_t dds_is_snapshot; uint8_t dds_inconsistent; char dds_origin[MAXNAMELEN]; } dmu_objset_stats_t; /* * Get stats on a dataset. */ void dmu_objset_fast_stat(objset_t *os, dmu_objset_stats_t *stat); /* * Add entries to the nvlist for all the objset's properties. See * zfs_prop_table[] and zfs(1m) for details on the properties. */ void dmu_objset_stats(objset_t *os, struct nvlist *nv); /* * Get the space usage statistics for statvfs(). * * refdbytes is the amount of space "referenced" by this objset. * availbytes is the amount of space available to this objset, taking * into account quotas & reservations, assuming that no other objsets * use the space first. These values correspond to the 'referenced' and * 'available' properties, described in the zfs(1m) manpage. * * usedobjs and availobjs are the number of objects currently allocated, * and available. */ void dmu_objset_space(objset_t *os, uint64_t *refdbytesp, uint64_t *availbytesp, uint64_t *usedobjsp, uint64_t *availobjsp); /* * The fsid_guid is a 56-bit ID that can change to avoid collisions. * (Contrast with the ds_guid which is a 64-bit ID that will never * change, so there is a small probability that it will collide.) */ uint64_t dmu_objset_fsid_guid(objset_t *os); /* * Get the [cm]time for an objset's snapshot dir */ timestruc_t dmu_objset_snap_cmtime(objset_t *os); int dmu_objset_is_snapshot(objset_t *os); extern struct spa *dmu_objset_spa(objset_t *os); extern struct zilog *dmu_objset_zil(objset_t *os); extern struct dsl_pool *dmu_objset_pool(objset_t *os); extern struct dsl_dataset *dmu_objset_ds(objset_t *os); extern void dmu_objset_name(objset_t *os, char *buf); extern dmu_objset_type_t dmu_objset_type(objset_t *os); extern uint64_t dmu_objset_id(objset_t *os); extern zfs_sync_type_t dmu_objset_syncprop(objset_t *os); extern zfs_logbias_op_t dmu_objset_logbias(objset_t *os); extern int dmu_snapshot_list_next(objset_t *os, int namelen, char *name, uint64_t *id, uint64_t *offp, boolean_t *case_conflict); extern int dmu_snapshot_lookup(objset_t *os, const char *name, uint64_t *val); extern int dmu_snapshot_realname(objset_t *os, char *name, char *real, int maxlen, boolean_t *conflict); extern int dmu_dir_list_next(objset_t *os, int namelen, char *name, uint64_t *idp, uint64_t *offp); typedef int objset_used_cb_t(dmu_object_type_t bonustype, void *bonus, uint64_t *userp, uint64_t *groupp); extern void dmu_objset_register_type(dmu_objset_type_t ost, objset_used_cb_t *cb); extern void dmu_objset_set_user(objset_t *os, void *user_ptr); extern void *dmu_objset_get_user(objset_t *os); /* * Return the txg number for the given assigned transaction. */ uint64_t dmu_tx_get_txg(dmu_tx_t *tx); /* * Synchronous write. * If a parent zio is provided this function initiates a write on the * provided buffer as a child of the parent zio. * In the absence of a parent zio, the write is completed synchronously. * At write completion, blk is filled with the bp of the written block. * Note that while the data covered by this function will be on stable * storage when the write completes this new data does not become a * permanent part of the file until the associated transaction commits. */ /* * {zfs,zvol,ztest}_get_done() args */ typedef struct zgd { struct zilog *zgd_zilog; struct blkptr *zgd_bp; dmu_buf_t *zgd_db; struct rl *zgd_rl; void *zgd_private; } zgd_t; typedef void dmu_sync_cb_t(zgd_t *arg, int error); int dmu_sync(struct zio *zio, uint64_t txg, dmu_sync_cb_t *done, zgd_t *zgd); /* * Find the next hole or data block in file starting at *off * Return found offset in *off. Return ESRCH for end of file. */ int dmu_offset_next(objset_t *os, uint64_t object, boolean_t hole, uint64_t *off); /* * Initial setup and final teardown. */ extern void dmu_init(void); extern void dmu_fini(void); typedef void (*dmu_traverse_cb_t)(objset_t *os, void *arg, struct blkptr *bp, uint64_t object, uint64_t offset, int len); void dmu_traverse_objset(objset_t *os, uint64_t txg_start, dmu_traverse_cb_t cb, void *arg); int dmu_diff(const char *tosnap_name, const char *fromsnap_name, struct vnode *vp, offset_t *offp); /* CRC64 table */ #define ZFS_CRC64_POLY 0xC96C5795D7870F42ULL /* ECMA-182, reflected form */ extern uint64_t zfs_crc64_table[256]; extern int zfs_mdcomp_disable; #ifdef __cplusplus } #endif #endif /* _SYS_DMU_H */ diff --git a/include/sys/dsl_dir.h b/include/sys/dsl_dir.h index d69d47696f23..3aa77523293b 100644 --- a/include/sys/dsl_dir.h +++ b/include/sys/dsl_dir.h @@ -1,166 +1,167 @@ /* * 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 by Delphix. All rights reserved. */ #ifndef _SYS_DSL_DIR_H #define _SYS_DSL_DIR_H #include #include #include #include #include #ifdef __cplusplus extern "C" { #endif struct dsl_dataset; typedef enum dd_used { DD_USED_HEAD, DD_USED_SNAP, DD_USED_CHILD, DD_USED_CHILD_RSRV, DD_USED_REFRSRV, DD_USED_NUM } dd_used_t; #define DD_FLAG_USED_BREAKDOWN (1<<0) typedef struct dsl_dir_phys { uint64_t dd_creation_time; /* not actually used */ uint64_t dd_head_dataset_obj; uint64_t dd_parent_obj; uint64_t dd_origin_obj; uint64_t dd_child_dir_zapobj; /* * how much space our children are accounting for; for leaf * datasets, == physical space used by fs + snaps */ uint64_t dd_used_bytes; uint64_t dd_compressed_bytes; uint64_t dd_uncompressed_bytes; /* Administrative quota setting */ uint64_t dd_quota; /* Administrative reservation setting */ uint64_t dd_reserved; uint64_t dd_props_zapobj; uint64_t dd_deleg_zapobj; /* dataset delegation permissions */ uint64_t dd_flags; uint64_t dd_used_breakdown[DD_USED_NUM]; uint64_t dd_clones; /* dsl_dir objects */ uint64_t dd_pad[13]; /* pad out to 256 bytes for good measure */ } dsl_dir_phys_t; struct dsl_dir { /* These are immutable; no lock needed: */ uint64_t dd_object; dsl_dir_phys_t *dd_phys; dmu_buf_t *dd_dbuf; dsl_pool_t *dd_pool; /* protected by lock on pool's dp_dirty_dirs list */ txg_node_t dd_dirty_link; /* protected by dp_config_rwlock */ dsl_dir_t *dd_parent; /* Protected by dd_lock */ kmutex_t dd_lock; list_t dd_prop_cbs; /* list of dsl_prop_cb_record_t's */ timestruc_t dd_snap_cmtime; /* last time snapshot namespace changed */ uint64_t dd_origin_txg; /* gross estimate of space used by in-flight tx's */ uint64_t dd_tempreserved[TXG_SIZE]; /* amount of space we expect to write; == amount of dirty data */ int64_t dd_space_towrite[TXG_SIZE]; /* protected by dd_lock; keep at end of struct for better locality */ char dd_myname[MAXNAMELEN]; }; void dsl_dir_rele(dsl_dir_t *dd, void *tag); int dsl_dir_hold(dsl_pool_t *dp, const char *name, void *tag, dsl_dir_t **, const char **tail); int dsl_dir_hold_obj(dsl_pool_t *dp, uint64_t ddobj, const char *tail, void *tag, dsl_dir_t **); void dsl_dir_name(dsl_dir_t *dd, char *buf); int dsl_dir_namelen(dsl_dir_t *dd); uint64_t dsl_dir_create_sync(dsl_pool_t *dp, dsl_dir_t *pds, const char *name, dmu_tx_t *tx); void dsl_dir_stats(dsl_dir_t *dd, nvlist_t *nv); uint64_t dsl_dir_space_available(dsl_dir_t *dd, dsl_dir_t *ancestor, int64_t delta, int ondiskonly); void dsl_dir_dirty(dsl_dir_t *dd, dmu_tx_t *tx); void dsl_dir_sync(dsl_dir_t *dd, dmu_tx_t *tx); int dsl_dir_tempreserve_space(dsl_dir_t *dd, uint64_t mem, uint64_t asize, uint64_t fsize, uint64_t usize, void **tr_cookiep, dmu_tx_t *tx); void dsl_dir_tempreserve_clear(void *tr_cookie, dmu_tx_t *tx); void dsl_dir_willuse_space(dsl_dir_t *dd, int64_t space, dmu_tx_t *tx); void dsl_dir_diduse_space(dsl_dir_t *dd, dd_used_t type, int64_t used, int64_t compressed, int64_t uncompressed, dmu_tx_t *tx); void dsl_dir_transfer_space(dsl_dir_t *dd, int64_t delta, dd_used_t oldtype, dd_used_t newtype, dmu_tx_t *tx); int dsl_dir_set_quota(const char *ddname, zprop_source_t source, uint64_t quota); int dsl_dir_set_reservation(const char *ddname, zprop_source_t source, uint64_t reservation); int dsl_dir_rename(const char *oldname, const char *newname); int dsl_dir_transfer_possible(dsl_dir_t *sdd, dsl_dir_t *tdd, uint64_t space); boolean_t dsl_dir_is_clone(dsl_dir_t *dd); void dsl_dir_new_refreservation(dsl_dir_t *dd, struct dsl_dataset *ds, uint64_t reservation, cred_t *cr, dmu_tx_t *tx); void dsl_dir_snap_cmtime_update(dsl_dir_t *dd); timestruc_t dsl_dir_snap_cmtime(dsl_dir_t *dd); void dsl_dir_set_reservation_sync_impl(dsl_dir_t *dd, uint64_t value, dmu_tx_t *tx); /* internal reserved dir name */ #define MOS_DIR_NAME "$MOS" #define ORIGIN_DIR_NAME "$ORIGIN" #define XLATION_DIR_NAME "$XLATION" #define FREE_DIR_NAME "$FREE" +#define LEAK_DIR_NAME "$LEAK" #ifdef ZFS_DEBUG #define dprintf_dd(dd, fmt, ...) do { \ if (zfs_flags & ZFS_DEBUG_DPRINTF) { \ char *__ds_name = kmem_alloc(MAXNAMELEN + strlen(MOS_DIR_NAME) + 1, \ KM_PUSHPAGE); \ dsl_dir_name(dd, __ds_name); \ dprintf("dd=%s " fmt, __ds_name, __VA_ARGS__); \ kmem_free(__ds_name, MAXNAMELEN + strlen(MOS_DIR_NAME) + 1); \ } \ _NOTE(CONSTCOND) } while (0) #else #define dprintf_dd(dd, fmt, ...) #endif #ifdef __cplusplus } #endif #endif /* _SYS_DSL_DIR_H */ diff --git a/include/sys/dsl_pool.h b/include/sys/dsl_pool.h index d5bad8dc1919..34dc65ba40ea 100644 --- a/include/sys/dsl_pool.h +++ b/include/sys/dsl_pool.h @@ -1,176 +1,177 @@ /* * 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 by Delphix. All rights reserved. */ #ifndef _SYS_DSL_POOL_H #define _SYS_DSL_POOL_H #include #include #include #include #include #include #include #include #include #include #include #ifdef __cplusplus extern "C" { #endif extern int zfs_txg_synctime_ms; struct objset; struct dsl_dir; struct dsl_dataset; struct dsl_pool; struct dmu_tx; struct dsl_scan; extern unsigned long zfs_dirty_data_max; extern unsigned long zfs_dirty_data_max_max; extern unsigned long zfs_dirty_data_sync; extern int zfs_dirty_data_max_percent; extern int zfs_dirty_data_max_max_percent; extern int zfs_delay_min_dirty_percent; extern unsigned long zfs_delay_scale; /* These macros are for indexing into the zfs_all_blkstats_t. */ #define DMU_OT_DEFERRED DMU_OT_NONE #define DMU_OT_OTHER DMU_OT_NUMTYPES /* place holder for DMU_OT() types */ #define DMU_OT_TOTAL (DMU_OT_NUMTYPES + 1) typedef struct zfs_blkstat { uint64_t zb_count; uint64_t zb_asize; uint64_t zb_lsize; uint64_t zb_psize; uint64_t zb_gangs; uint64_t zb_ditto_2_of_2_samevdev; uint64_t zb_ditto_2_of_3_samevdev; uint64_t zb_ditto_3_of_3_samevdev; } zfs_blkstat_t; typedef struct zfs_all_blkstats { zfs_blkstat_t zab_type[DN_MAX_LEVELS + 1][DMU_OT_TOTAL + 1]; } zfs_all_blkstats_t; typedef struct dsl_pool { /* Immutable */ spa_t *dp_spa; struct objset *dp_meta_objset; struct dsl_dir *dp_root_dir; struct dsl_dir *dp_mos_dir; struct dsl_dir *dp_free_dir; + struct dsl_dir *dp_leak_dir; struct dsl_dataset *dp_origin_snap; uint64_t dp_root_dir_obj; struct taskq *dp_iput_taskq; /* No lock needed - sync context only */ blkptr_t dp_meta_rootbp; uint64_t dp_tmp_userrefs_obj; bpobj_t dp_free_bpobj; uint64_t dp_bptree_obj; uint64_t dp_empty_bpobj; struct dsl_scan *dp_scan; /* Uses dp_lock */ kmutex_t dp_lock; kcondvar_t dp_spaceavail_cv; uint64_t dp_dirty_pertxg[TXG_SIZE]; uint64_t dp_dirty_total; uint64_t dp_mos_used_delta; uint64_t dp_mos_compressed_delta; uint64_t dp_mos_uncompressed_delta; /* * Time of most recently scheduled (furthest in the future) * wakeup for delayed transactions. */ hrtime_t dp_last_wakeup; /* Has its own locking */ tx_state_t dp_tx; txg_list_t dp_dirty_datasets; txg_list_t dp_dirty_zilogs; txg_list_t dp_dirty_dirs; txg_list_t dp_sync_tasks; /* * Protects administrative changes (properties, namespace) * * It is only held for write in syncing context. Therefore * syncing context does not need to ever have it for read, since * nobody else could possibly have it for write. */ rrwlock_t dp_config_rwlock; zfs_all_blkstats_t *dp_blkstats; } dsl_pool_t; int dsl_pool_init(spa_t *spa, uint64_t txg, dsl_pool_t **dpp); int dsl_pool_open(dsl_pool_t *dp); void dsl_pool_close(dsl_pool_t *dp); dsl_pool_t *dsl_pool_create(spa_t *spa, nvlist_t *zplprops, uint64_t txg); void dsl_pool_sync(dsl_pool_t *dp, uint64_t txg); void dsl_pool_sync_done(dsl_pool_t *dp, uint64_t txg); int dsl_pool_sync_context(dsl_pool_t *dp); uint64_t dsl_pool_adjustedsize(dsl_pool_t *dp, boolean_t netfree); uint64_t dsl_pool_adjustedfree(dsl_pool_t *dp, boolean_t netfree); void dsl_pool_dirty_space(dsl_pool_t *dp, int64_t space, dmu_tx_t *tx); void dsl_pool_undirty_space(dsl_pool_t *dp, int64_t space, uint64_t txg); void dsl_free(dsl_pool_t *dp, uint64_t txg, const blkptr_t *bpp); void dsl_free_sync(zio_t *pio, dsl_pool_t *dp, uint64_t txg, const blkptr_t *bpp); void dsl_pool_create_origin(dsl_pool_t *dp, dmu_tx_t *tx); void dsl_pool_upgrade_clones(dsl_pool_t *dp, dmu_tx_t *tx); void dsl_pool_upgrade_dir_clones(dsl_pool_t *dp, dmu_tx_t *tx); void dsl_pool_mos_diduse_space(dsl_pool_t *dp, int64_t used, int64_t comp, int64_t uncomp); boolean_t dsl_pool_need_dirty_delay(dsl_pool_t *dp); void dsl_pool_config_enter(dsl_pool_t *dp, void *tag); void dsl_pool_config_exit(dsl_pool_t *dp, void *tag); boolean_t dsl_pool_config_held(dsl_pool_t *dp); taskq_t *dsl_pool_iput_taskq(dsl_pool_t *dp); int dsl_pool_user_hold(dsl_pool_t *dp, uint64_t dsobj, const char *tag, uint64_t now, dmu_tx_t *tx); int dsl_pool_user_release(dsl_pool_t *dp, uint64_t dsobj, const char *tag, dmu_tx_t *tx); void dsl_pool_clean_tmp_userrefs(dsl_pool_t *dp); int dsl_pool_open_special_dir(dsl_pool_t *dp, const char *name, dsl_dir_t **); int dsl_pool_hold(const char *name, void *tag, dsl_pool_t **dp); void dsl_pool_rele(dsl_pool_t *dp, void *tag); #ifdef __cplusplus } #endif #endif /* _SYS_DSL_POOL_H */ diff --git a/include/sys/dsl_scan.h b/include/sys/dsl_scan.h index bcb85d67d38e..de6a7d17a3ef 100644 --- a/include/sys/dsl_scan.h +++ b/include/sys/dsl_scan.h @@ -1,146 +1,147 @@ /* * 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) 2010, Oracle and/or its affiliates. All rights reserved. * Copyright (c) 2013 by Delphix. All rights reserved. */ #ifndef _SYS_DSL_SCAN_H #define _SYS_DSL_SCAN_H #include #include #include #include #ifdef __cplusplus extern "C" { #endif struct objset; struct dsl_dir; struct dsl_dataset; struct dsl_pool; struct dmu_tx; /* * All members of this structure must be uint64_t, for byteswap * purposes. */ typedef struct dsl_scan_phys { uint64_t scn_func; /* pool_scan_func_t */ uint64_t scn_state; /* dsl_scan_state_t */ uint64_t scn_queue_obj; uint64_t scn_min_txg; uint64_t scn_max_txg; uint64_t scn_cur_min_txg; uint64_t scn_cur_max_txg; uint64_t scn_start_time; uint64_t scn_end_time; uint64_t scn_to_examine; /* total bytes to be scanned */ uint64_t scn_examined; /* bytes scanned so far */ uint64_t scn_to_process; uint64_t scn_processed; uint64_t scn_errors; /* scan I/O error count */ uint64_t scn_ddt_class_max; ddt_bookmark_t scn_ddt_bookmark; zbookmark_t scn_bookmark; uint64_t scn_flags; /* dsl_scan_flags_t */ } dsl_scan_phys_t; #define SCAN_PHYS_NUMINTS (sizeof (dsl_scan_phys_t) / sizeof (uint64_t)) typedef enum dsl_scan_flags { DSF_VISIT_DS_AGAIN = 1<<0, } dsl_scan_flags_t; #define DSL_SCAN_FLAGS_MASK (DSF_VISIT_DS_AGAIN) /* * Every pool will have one dsl_scan_t and this structure will contain * in-memory information about the scan and a pointer to the on-disk * representation (i.e. dsl_scan_phys_t). Most of the state of the scan * is contained on-disk to allow the scan to resume in the event of a reboot * or panic. This structure maintains information about the behavior of a * running scan, some caching information, and how it should traverse the pool. * * The following members of this structure direct the behavior of the scan: * * scn_pausing - a scan that cannot be completed in a single txg or * has exceeded its allotted time will need to pause. * When this flag is set the scanner will stop traversing * the pool and write out the current state to disk. * * scn_restart_txg - directs the scanner to either restart or start a * a scan at the specified txg value. * * scn_done_txg - when a scan completes its traversal it will set * the completion txg to the next txg. This is necessary * to ensure that any blocks that were freed during * the scan but have not yet been processed (i.e deferred * frees) are accounted for. * * This structure also maintains information about deferred frees which are * a special kind of traversal. Deferred free can exist in either a bptree or * a bpobj structure. The scn_is_bptree flag will indicate the type of * deferred free that is in progress. If the deferred free is part of an * asynchronous destroy then the scn_async_destroying flag will be set. */ typedef struct dsl_scan { struct dsl_pool *scn_dp; boolean_t scn_pausing; uint64_t scn_restart_txg; uint64_t scn_done_txg; uint64_t scn_sync_start_time; zio_t *scn_zio_root; /* for freeing blocks */ boolean_t scn_is_bptree; boolean_t scn_async_destroying; + boolean_t scn_async_stalled; /* for debugging / information */ uint64_t scn_visited_this_txg; dsl_scan_phys_t scn_phys; } dsl_scan_t; int dsl_scan_init(struct dsl_pool *dp, uint64_t txg); void dsl_scan_fini(struct dsl_pool *dp); void dsl_scan_sync(struct dsl_pool *, dmu_tx_t *); int dsl_scan_cancel(struct dsl_pool *); int dsl_scan(struct dsl_pool *, pool_scan_func_t); void dsl_resilver_restart(struct dsl_pool *, uint64_t txg); boolean_t dsl_scan_resilvering(struct dsl_pool *dp); boolean_t dsl_dataset_unstable(struct dsl_dataset *ds); void dsl_scan_ddt_entry(dsl_scan_t *scn, enum zio_checksum checksum, ddt_entry_t *dde, dmu_tx_t *tx); void dsl_scan_ds_destroyed(struct dsl_dataset *ds, struct dmu_tx *tx); void dsl_scan_ds_snapshotted(struct dsl_dataset *ds, struct dmu_tx *tx); void dsl_scan_ds_clone_swapped(struct dsl_dataset *ds1, struct dsl_dataset *ds2, struct dmu_tx *tx); boolean_t dsl_scan_active(dsl_scan_t *scn); #ifdef __cplusplus } #endif #endif /* _SYS_DSL_SCAN_H */ diff --git a/include/sys/fs/zfs.h b/include/sys/fs/zfs.h index 5371e12efa7e..227d8b2fbd88 100644 --- a/include/sys/fs/zfs.h +++ b/include/sys/fs/zfs.h @@ -1,982 +1,983 @@ /* * 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) 2011, 2014 by Delphix. All rights reserved. * Copyright 2011 Nexenta Systems, Inc. All rights reserved. * Copyright (c) 2012, Joyent, Inc. All rights reserved. */ /* Portions Copyright 2010 Robert Milkowski */ #ifndef _SYS_FS_ZFS_H #define _SYS_FS_ZFS_H #include #ifdef __cplusplus extern "C" { #endif /* * Types and constants shared between userland and the kernel. */ /* * Each dataset can be one of the following types. These constants can be * combined into masks that can be passed to various functions. */ typedef enum { ZFS_TYPE_FILESYSTEM = (1 << 0), ZFS_TYPE_SNAPSHOT = (1 << 1), ZFS_TYPE_VOLUME = (1 << 2), ZFS_TYPE_POOL = (1 << 3), ZFS_TYPE_BOOKMARK = (1 << 4) } zfs_type_t; typedef enum dmu_objset_type { DMU_OST_NONE, DMU_OST_META, DMU_OST_ZFS, DMU_OST_ZVOL, DMU_OST_OTHER, /* For testing only! */ DMU_OST_ANY, /* Be careful! */ DMU_OST_NUMTYPES } dmu_objset_type_t; #define ZFS_TYPE_DATASET \ (ZFS_TYPE_FILESYSTEM | ZFS_TYPE_VOLUME | ZFS_TYPE_SNAPSHOT) #define ZAP_MAXNAMELEN 256 #define ZAP_MAXVALUELEN (1024 * 8) #define ZAP_OLDMAXVALUELEN 1024 /* * Dataset properties are identified by these constants and must be added to * the end of this list to ensure that external consumers are not affected * by the change. If you make any changes to this list, be sure to update * the property table in module/zcommon/zfs_prop.c. */ typedef enum { ZFS_PROP_TYPE, ZFS_PROP_CREATION, ZFS_PROP_USED, ZFS_PROP_AVAILABLE, ZFS_PROP_REFERENCED, ZFS_PROP_COMPRESSRATIO, ZFS_PROP_MOUNTED, ZFS_PROP_ORIGIN, ZFS_PROP_QUOTA, ZFS_PROP_RESERVATION, ZFS_PROP_VOLSIZE, ZFS_PROP_VOLBLOCKSIZE, ZFS_PROP_RECORDSIZE, ZFS_PROP_MOUNTPOINT, ZFS_PROP_SHARENFS, ZFS_PROP_CHECKSUM, ZFS_PROP_COMPRESSION, ZFS_PROP_ATIME, ZFS_PROP_DEVICES, ZFS_PROP_EXEC, ZFS_PROP_SETUID, ZFS_PROP_READONLY, ZFS_PROP_ZONED, ZFS_PROP_SNAPDIR, ZFS_PROP_PRIVATE, /* not exposed to user, temporary */ ZFS_PROP_ACLINHERIT, ZFS_PROP_CREATETXG, /* not exposed to the user */ ZFS_PROP_NAME, /* not exposed to the user */ ZFS_PROP_CANMOUNT, ZFS_PROP_ISCSIOPTIONS, /* not exposed to the user */ ZFS_PROP_XATTR, ZFS_PROP_NUMCLONES, /* not exposed to the user */ ZFS_PROP_COPIES, ZFS_PROP_VERSION, ZFS_PROP_UTF8ONLY, ZFS_PROP_NORMALIZE, ZFS_PROP_CASE, ZFS_PROP_VSCAN, ZFS_PROP_NBMAND, ZFS_PROP_SHARESMB, ZFS_PROP_REFQUOTA, ZFS_PROP_REFRESERVATION, ZFS_PROP_GUID, ZFS_PROP_PRIMARYCACHE, ZFS_PROP_SECONDARYCACHE, ZFS_PROP_USEDSNAP, ZFS_PROP_USEDDS, ZFS_PROP_USEDCHILD, ZFS_PROP_USEDREFRESERV, ZFS_PROP_USERACCOUNTING, /* not exposed to the user */ ZFS_PROP_STMF_SHAREINFO, /* not exposed to the user */ ZFS_PROP_DEFER_DESTROY, ZFS_PROP_USERREFS, ZFS_PROP_LOGBIAS, ZFS_PROP_UNIQUE, /* not exposed to the user */ ZFS_PROP_OBJSETID, /* not exposed to the user */ ZFS_PROP_DEDUP, ZFS_PROP_MLSLABEL, ZFS_PROP_SYNC, ZFS_PROP_REFRATIO, ZFS_PROP_WRITTEN, ZFS_PROP_CLONES, ZFS_PROP_LOGICALUSED, ZFS_PROP_LOGICALREFERENCED, ZFS_PROP_INCONSISTENT, /* not exposed to the user */ ZFS_PROP_SNAPDEV, ZFS_PROP_ACLTYPE, ZFS_PROP_SELINUX_CONTEXT, ZFS_PROP_SELINUX_FSCONTEXT, ZFS_PROP_SELINUX_DEFCONTEXT, ZFS_PROP_SELINUX_ROOTCONTEXT, ZFS_PROP_RELATIME, ZFS_PROP_REDUNDANT_METADATA, ZFS_NUM_PROPS } zfs_prop_t; typedef enum { ZFS_PROP_USERUSED, ZFS_PROP_USERQUOTA, ZFS_PROP_GROUPUSED, ZFS_PROP_GROUPQUOTA, ZFS_NUM_USERQUOTA_PROPS } zfs_userquota_prop_t; extern const char *zfs_userquota_prop_prefixes[ZFS_NUM_USERQUOTA_PROPS]; /* * Pool properties are identified by these constants and must be added to the * end of this list to ensure that external consumers are not affected * by the change. If you make any changes to this list, be sure to update * the property table in module/zcommon/zpool_prop.c. */ typedef enum { ZPOOL_PROP_NAME, ZPOOL_PROP_SIZE, ZPOOL_PROP_CAPACITY, ZPOOL_PROP_ALTROOT, ZPOOL_PROP_HEALTH, ZPOOL_PROP_GUID, ZPOOL_PROP_VERSION, ZPOOL_PROP_BOOTFS, ZPOOL_PROP_DELEGATION, ZPOOL_PROP_AUTOREPLACE, ZPOOL_PROP_CACHEFILE, ZPOOL_PROP_FAILUREMODE, ZPOOL_PROP_LISTSNAPS, ZPOOL_PROP_AUTOEXPAND, ZPOOL_PROP_DEDUPDITTO, ZPOOL_PROP_DEDUPRATIO, ZPOOL_PROP_FREE, ZPOOL_PROP_ALLOCATED, ZPOOL_PROP_READONLY, ZPOOL_PROP_ASHIFT, ZPOOL_PROP_COMMENT, ZPOOL_PROP_EXPANDSZ, ZPOOL_PROP_FREEING, + ZPOOL_PROP_LEAKED, ZPOOL_NUM_PROPS } zpool_prop_t; /* Small enough to not hog a whole line of printout in zpool(1M). */ #define ZPROP_MAX_COMMENT 32 #define ZPROP_CONT -2 #define ZPROP_INVAL -1 #define ZPROP_VALUE "value" #define ZPROP_SOURCE "source" typedef enum { ZPROP_SRC_NONE = 0x1, ZPROP_SRC_DEFAULT = 0x2, ZPROP_SRC_TEMPORARY = 0x4, ZPROP_SRC_LOCAL = 0x8, ZPROP_SRC_INHERITED = 0x10, ZPROP_SRC_RECEIVED = 0x20 } zprop_source_t; #define ZPROP_SRC_ALL 0x3f #define ZPROP_SOURCE_VAL_RECVD "$recvd" #define ZPROP_N_MORE_ERRORS "N_MORE_ERRORS" /* * Dataset flag implemented as a special entry in the props zap object * indicating that the dataset has received properties on or after * SPA_VERSION_RECVD_PROPS. The first such receive blows away local properties * just as it did in earlier versions, and thereafter, local properties are * preserved. */ #define ZPROP_HAS_RECVD "$hasrecvd" typedef enum { ZPROP_ERR_NOCLEAR = 0x1, /* failure to clear existing props */ ZPROP_ERR_NORESTORE = 0x2 /* failure to restore props on error */ } zprop_errflags_t; typedef int (*zprop_func)(int, void *); /* * Properties to be set on the root file system of a new pool * are stuffed into their own nvlist, which is then included in * the properties nvlist with the pool properties. */ #define ZPOOL_ROOTFS_PROPS "root-props-nvl" /* * Dataset property functions shared between libzfs and kernel. */ const char *zfs_prop_default_string(zfs_prop_t); uint64_t zfs_prop_default_numeric(zfs_prop_t); boolean_t zfs_prop_readonly(zfs_prop_t); boolean_t zfs_prop_inheritable(zfs_prop_t); boolean_t zfs_prop_setonce(zfs_prop_t); const char *zfs_prop_to_name(zfs_prop_t); zfs_prop_t zfs_name_to_prop(const char *); boolean_t zfs_prop_user(const char *); boolean_t zfs_prop_userquota(const char *); boolean_t zfs_prop_written(const char *); int zfs_prop_index_to_string(zfs_prop_t, uint64_t, const char **); int zfs_prop_string_to_index(zfs_prop_t, const char *, uint64_t *); uint64_t zfs_prop_random_value(zfs_prop_t, uint64_t seed); boolean_t zfs_prop_valid_for_type(int, zfs_type_t, boolean_t); /* * Pool property functions shared between libzfs and kernel. */ zpool_prop_t zpool_name_to_prop(const char *); const char *zpool_prop_to_name(zpool_prop_t); const char *zpool_prop_default_string(zpool_prop_t); uint64_t zpool_prop_default_numeric(zpool_prop_t); boolean_t zpool_prop_readonly(zpool_prop_t); boolean_t zpool_prop_feature(const char *); boolean_t zpool_prop_unsupported(const char *); int zpool_prop_index_to_string(zpool_prop_t, uint64_t, const char **); int zpool_prop_string_to_index(zpool_prop_t, const char *, uint64_t *); uint64_t zpool_prop_random_value(zpool_prop_t, uint64_t seed); /* * Definitions for the Delegation. */ typedef enum { ZFS_DELEG_WHO_UNKNOWN = 0, ZFS_DELEG_USER = 'u', ZFS_DELEG_USER_SETS = 'U', ZFS_DELEG_GROUP = 'g', ZFS_DELEG_GROUP_SETS = 'G', ZFS_DELEG_EVERYONE = 'e', ZFS_DELEG_EVERYONE_SETS = 'E', ZFS_DELEG_CREATE = 'c', ZFS_DELEG_CREATE_SETS = 'C', ZFS_DELEG_NAMED_SET = 's', ZFS_DELEG_NAMED_SET_SETS = 'S' } zfs_deleg_who_type_t; typedef enum { ZFS_DELEG_NONE = 0, ZFS_DELEG_PERM_LOCAL = 1, ZFS_DELEG_PERM_DESCENDENT = 2, ZFS_DELEG_PERM_LOCALDESCENDENT = 3, ZFS_DELEG_PERM_CREATE = 4 } zfs_deleg_inherit_t; #define ZFS_DELEG_PERM_UID "uid" #define ZFS_DELEG_PERM_GID "gid" #define ZFS_DELEG_PERM_GROUPS "groups" #define ZFS_MLSLABEL_DEFAULT "none" #define ZFS_SMB_ACL_SRC "src" #define ZFS_SMB_ACL_TARGET "target" typedef enum { ZFS_CANMOUNT_OFF = 0, ZFS_CANMOUNT_ON = 1, ZFS_CANMOUNT_NOAUTO = 2 } zfs_canmount_type_t; typedef enum { ZFS_LOGBIAS_LATENCY = 0, ZFS_LOGBIAS_THROUGHPUT = 1 } zfs_logbias_op_t; typedef enum zfs_share_op { ZFS_SHARE_NFS = 0, ZFS_UNSHARE_NFS = 1, ZFS_SHARE_SMB = 2, ZFS_UNSHARE_SMB = 3 } zfs_share_op_t; typedef enum zfs_smb_acl_op { ZFS_SMB_ACL_ADD, ZFS_SMB_ACL_REMOVE, ZFS_SMB_ACL_RENAME, ZFS_SMB_ACL_PURGE } zfs_smb_acl_op_t; typedef enum zfs_cache_type { ZFS_CACHE_NONE = 0, ZFS_CACHE_METADATA = 1, ZFS_CACHE_ALL = 2 } zfs_cache_type_t; typedef enum { ZFS_SYNC_STANDARD = 0, ZFS_SYNC_ALWAYS = 1, ZFS_SYNC_DISABLED = 2 } zfs_sync_type_t; typedef enum { ZFS_XATTR_OFF = 0, ZFS_XATTR_DIR = 1, ZFS_XATTR_SA = 2 } zfs_xattr_type_t; typedef enum { ZFS_REDUNDANT_METADATA_ALL, ZFS_REDUNDANT_METADATA_MOST } zfs_redundant_metadata_type_t; /* * On-disk version number. */ #define SPA_VERSION_1 1ULL #define SPA_VERSION_2 2ULL #define SPA_VERSION_3 3ULL #define SPA_VERSION_4 4ULL #define SPA_VERSION_5 5ULL #define SPA_VERSION_6 6ULL #define SPA_VERSION_7 7ULL #define SPA_VERSION_8 8ULL #define SPA_VERSION_9 9ULL #define SPA_VERSION_10 10ULL #define SPA_VERSION_11 11ULL #define SPA_VERSION_12 12ULL #define SPA_VERSION_13 13ULL #define SPA_VERSION_14 14ULL #define SPA_VERSION_15 15ULL #define SPA_VERSION_16 16ULL #define SPA_VERSION_17 17ULL #define SPA_VERSION_18 18ULL #define SPA_VERSION_19 19ULL #define SPA_VERSION_20 20ULL #define SPA_VERSION_21 21ULL #define SPA_VERSION_22 22ULL #define SPA_VERSION_23 23ULL #define SPA_VERSION_24 24ULL #define SPA_VERSION_25 25ULL #define SPA_VERSION_26 26ULL #define SPA_VERSION_27 27ULL #define SPA_VERSION_28 28ULL #define SPA_VERSION_5000 5000ULL /* * When bumping up SPA_VERSION, make sure GRUB ZFS understands the on-disk * format change. Go to usr/src/grub/grub-0.97/stage2/{zfs-include/, fsys_zfs*}, * and do the appropriate changes. Also bump the version number in * usr/src/grub/capability. */ #define SPA_VERSION SPA_VERSION_5000 #define SPA_VERSION_STRING "5000" /* * Symbolic names for the changes that caused a SPA_VERSION switch. * Used in the code when checking for presence or absence of a feature. * Feel free to define multiple symbolic names for each version if there * were multiple changes to on-disk structures during that version. * * NOTE: When checking the current SPA_VERSION in your code, be sure * to use spa_version() since it reports the version of the * last synced uberblock. Checking the in-flight version can * be dangerous in some cases. */ #define SPA_VERSION_INITIAL SPA_VERSION_1 #define SPA_VERSION_DITTO_BLOCKS SPA_VERSION_2 #define SPA_VERSION_SPARES SPA_VERSION_3 #define SPA_VERSION_RAIDZ2 SPA_VERSION_3 #define SPA_VERSION_BPOBJ_ACCOUNT SPA_VERSION_3 #define SPA_VERSION_RAIDZ_DEFLATE SPA_VERSION_3 #define SPA_VERSION_DNODE_BYTES SPA_VERSION_3 #define SPA_VERSION_ZPOOL_HISTORY SPA_VERSION_4 #define SPA_VERSION_GZIP_COMPRESSION SPA_VERSION_5 #define SPA_VERSION_BOOTFS SPA_VERSION_6 #define SPA_VERSION_SLOGS SPA_VERSION_7 #define SPA_VERSION_DELEGATED_PERMS SPA_VERSION_8 #define SPA_VERSION_FUID SPA_VERSION_9 #define SPA_VERSION_REFRESERVATION SPA_VERSION_9 #define SPA_VERSION_REFQUOTA SPA_VERSION_9 #define SPA_VERSION_UNIQUE_ACCURATE SPA_VERSION_9 #define SPA_VERSION_L2CACHE SPA_VERSION_10 #define SPA_VERSION_NEXT_CLONES SPA_VERSION_11 #define SPA_VERSION_ORIGIN SPA_VERSION_11 #define SPA_VERSION_DSL_SCRUB SPA_VERSION_11 #define SPA_VERSION_SNAP_PROPS SPA_VERSION_12 #define SPA_VERSION_USED_BREAKDOWN SPA_VERSION_13 #define SPA_VERSION_PASSTHROUGH_X SPA_VERSION_14 #define SPA_VERSION_USERSPACE SPA_VERSION_15 #define SPA_VERSION_STMF_PROP SPA_VERSION_16 #define SPA_VERSION_RAIDZ3 SPA_VERSION_17 #define SPA_VERSION_USERREFS SPA_VERSION_18 #define SPA_VERSION_HOLES SPA_VERSION_19 #define SPA_VERSION_ZLE_COMPRESSION SPA_VERSION_20 #define SPA_VERSION_DEDUP SPA_VERSION_21 #define SPA_VERSION_RECVD_PROPS SPA_VERSION_22 #define SPA_VERSION_SLIM_ZIL SPA_VERSION_23 #define SPA_VERSION_SA SPA_VERSION_24 #define SPA_VERSION_SCAN SPA_VERSION_25 #define SPA_VERSION_DIR_CLONES SPA_VERSION_26 #define SPA_VERSION_DEADLISTS SPA_VERSION_26 #define SPA_VERSION_FAST_SNAP SPA_VERSION_27 #define SPA_VERSION_MULTI_REPLACE SPA_VERSION_28 #define SPA_VERSION_BEFORE_FEATURES SPA_VERSION_28 #define SPA_VERSION_FEATURES SPA_VERSION_5000 #define SPA_VERSION_IS_SUPPORTED(v) \ (((v) >= SPA_VERSION_INITIAL && (v) <= SPA_VERSION_BEFORE_FEATURES) || \ ((v) >= SPA_VERSION_FEATURES && (v) <= SPA_VERSION)) /* * ZPL version - rev'd whenever an incompatible on-disk format change * occurs. This is independent of SPA/DMU/ZAP versioning. You must * also update the version_table[] and help message in zfs_prop.c. * * When changing, be sure to teach GRUB how to read the new format! * See usr/src/grub/grub-0.97/stage2/{zfs-include/,fsys_zfs*} */ #define ZPL_VERSION_1 1ULL #define ZPL_VERSION_2 2ULL #define ZPL_VERSION_3 3ULL #define ZPL_VERSION_4 4ULL #define ZPL_VERSION_5 5ULL #define ZPL_VERSION ZPL_VERSION_5 #define ZPL_VERSION_STRING "5" #define ZPL_VERSION_INITIAL ZPL_VERSION_1 #define ZPL_VERSION_DIRENT_TYPE ZPL_VERSION_2 #define ZPL_VERSION_FUID ZPL_VERSION_3 #define ZPL_VERSION_NORMALIZATION ZPL_VERSION_3 #define ZPL_VERSION_SYSATTR ZPL_VERSION_3 #define ZPL_VERSION_USERSPACE ZPL_VERSION_4 #define ZPL_VERSION_SA ZPL_VERSION_5 /* Rewind request information */ #define ZPOOL_NO_REWIND 1 /* No policy - default behavior */ #define ZPOOL_NEVER_REWIND 2 /* Do not search for best txg or rewind */ #define ZPOOL_TRY_REWIND 4 /* Search for best txg, but do not rewind */ #define ZPOOL_DO_REWIND 8 /* Rewind to best txg w/in deferred frees */ #define ZPOOL_EXTREME_REWIND 16 /* Allow extreme measures to find best txg */ #define ZPOOL_REWIND_MASK 28 /* All the possible rewind bits */ #define ZPOOL_REWIND_POLICIES 31 /* All the possible policy bits */ typedef struct zpool_rewind_policy { uint32_t zrp_request; /* rewind behavior requested */ uint64_t zrp_maxmeta; /* max acceptable meta-data errors */ uint64_t zrp_maxdata; /* max acceptable data errors */ uint64_t zrp_txg; /* specific txg to load */ } zpool_rewind_policy_t; /* * The following are configuration names used in the nvlist describing a pool's * configuration. */ #define ZPOOL_CONFIG_VERSION "version" #define ZPOOL_CONFIG_POOL_NAME "name" #define ZPOOL_CONFIG_POOL_STATE "state" #define ZPOOL_CONFIG_POOL_TXG "txg" #define ZPOOL_CONFIG_POOL_GUID "pool_guid" #define ZPOOL_CONFIG_CREATE_TXG "create_txg" #define ZPOOL_CONFIG_TOP_GUID "top_guid" #define ZPOOL_CONFIG_VDEV_TREE "vdev_tree" #define ZPOOL_CONFIG_TYPE "type" #define ZPOOL_CONFIG_CHILDREN "children" #define ZPOOL_CONFIG_ID "id" #define ZPOOL_CONFIG_GUID "guid" #define ZPOOL_CONFIG_PATH "path" #define ZPOOL_CONFIG_DEVID "devid" #define ZPOOL_CONFIG_METASLAB_ARRAY "metaslab_array" #define ZPOOL_CONFIG_METASLAB_SHIFT "metaslab_shift" #define ZPOOL_CONFIG_ASHIFT "ashift" #define ZPOOL_CONFIG_ASIZE "asize" #define ZPOOL_CONFIG_DTL "DTL" #define ZPOOL_CONFIG_SCAN_STATS "scan_stats" /* not stored on disk */ #define ZPOOL_CONFIG_VDEV_STATS "vdev_stats" /* not stored on disk */ #define ZPOOL_CONFIG_WHOLE_DISK "whole_disk" #define ZPOOL_CONFIG_ERRCOUNT "error_count" #define ZPOOL_CONFIG_NOT_PRESENT "not_present" #define ZPOOL_CONFIG_SPARES "spares" #define ZPOOL_CONFIG_IS_SPARE "is_spare" #define ZPOOL_CONFIG_NPARITY "nparity" #define ZPOOL_CONFIG_HOSTID "hostid" #define ZPOOL_CONFIG_HOSTNAME "hostname" #define ZPOOL_CONFIG_LOADED_TIME "initial_load_time" #define ZPOOL_CONFIG_UNSPARE "unspare" #define ZPOOL_CONFIG_PHYS_PATH "phys_path" #define ZPOOL_CONFIG_IS_LOG "is_log" #define ZPOOL_CONFIG_L2CACHE "l2cache" #define ZPOOL_CONFIG_HOLE_ARRAY "hole_array" #define ZPOOL_CONFIG_VDEV_CHILDREN "vdev_children" #define ZPOOL_CONFIG_IS_HOLE "is_hole" #define ZPOOL_CONFIG_DDT_HISTOGRAM "ddt_histogram" #define ZPOOL_CONFIG_DDT_OBJ_STATS "ddt_object_stats" #define ZPOOL_CONFIG_DDT_STATS "ddt_stats" #define ZPOOL_CONFIG_SPLIT "splitcfg" #define ZPOOL_CONFIG_ORIG_GUID "orig_guid" #define ZPOOL_CONFIG_SPLIT_GUID "split_guid" #define ZPOOL_CONFIG_SPLIT_LIST "guid_list" #define ZPOOL_CONFIG_REMOVING "removing" #define ZPOOL_CONFIG_RESILVER_TXG "resilver_txg" #define ZPOOL_CONFIG_COMMENT "comment" #define ZPOOL_CONFIG_SUSPENDED "suspended" /* not stored on disk */ #define ZPOOL_CONFIG_TIMESTAMP "timestamp" /* not stored on disk */ #define ZPOOL_CONFIG_BOOTFS "bootfs" /* not stored on disk */ #define ZPOOL_CONFIG_MISSING_DEVICES "missing_vdevs" /* not stored on disk */ #define ZPOOL_CONFIG_LOAD_INFO "load_info" /* not stored on disk */ #define ZPOOL_CONFIG_REWIND_INFO "rewind_info" /* not stored on disk */ #define ZPOOL_CONFIG_UNSUP_FEAT "unsup_feat" /* not stored on disk */ #define ZPOOL_CONFIG_ENABLED_FEAT "enabled_feat" /* not stored on disk */ #define ZPOOL_CONFIG_CAN_RDONLY "can_rdonly" /* not stored on disk */ #define ZPOOL_CONFIG_FEATURES_FOR_READ "features_for_read" #define ZPOOL_CONFIG_FEATURE_STATS "feature_stats" /* not stored on disk */ #define ZPOOL_CONFIG_ERRATA "errata" /* not stored on disk */ /* * The persistent vdev state is stored as separate values rather than a single * 'vdev_state' entry. This is because a device can be in multiple states, such * as offline and degraded. */ #define ZPOOL_CONFIG_OFFLINE "offline" #define ZPOOL_CONFIG_FAULTED "faulted" #define ZPOOL_CONFIG_DEGRADED "degraded" #define ZPOOL_CONFIG_REMOVED "removed" #define ZPOOL_CONFIG_FRU "fru" #define ZPOOL_CONFIG_AUX_STATE "aux_state" /* Rewind policy parameters */ #define ZPOOL_REWIND_POLICY "rewind-policy" #define ZPOOL_REWIND_REQUEST "rewind-request" #define ZPOOL_REWIND_REQUEST_TXG "rewind-request-txg" #define ZPOOL_REWIND_META_THRESH "rewind-meta-thresh" #define ZPOOL_REWIND_DATA_THRESH "rewind-data-thresh" /* Rewind data discovered */ #define ZPOOL_CONFIG_LOAD_TIME "rewind_txg_ts" #define ZPOOL_CONFIG_LOAD_DATA_ERRORS "verify_data_errors" #define ZPOOL_CONFIG_REWIND_TIME "seconds_of_rewind" #define VDEV_TYPE_ROOT "root" #define VDEV_TYPE_MIRROR "mirror" #define VDEV_TYPE_REPLACING "replacing" #define VDEV_TYPE_RAIDZ "raidz" #define VDEV_TYPE_DISK "disk" #define VDEV_TYPE_FILE "file" #define VDEV_TYPE_MISSING "missing" #define VDEV_TYPE_HOLE "hole" #define VDEV_TYPE_SPARE "spare" #define VDEV_TYPE_LOG "log" #define VDEV_TYPE_L2CACHE "l2cache" /* * This is needed in userland to report the minimum necessary device size. */ #define SPA_MINDEVSIZE (64ULL << 20) /* * The location of the pool configuration repository, shared between kernel and * userland. */ #define ZPOOL_CACHE "/etc/zfs/zpool.cache" /* * vdev states are ordered from least to most healthy. * A vdev that's CANT_OPEN or below is considered unusable. */ typedef enum vdev_state { VDEV_STATE_UNKNOWN = 0, /* Uninitialized vdev */ VDEV_STATE_CLOSED, /* Not currently open */ VDEV_STATE_OFFLINE, /* Not allowed to open */ VDEV_STATE_REMOVED, /* Explicitly removed from system */ VDEV_STATE_CANT_OPEN, /* Tried to open, but failed */ VDEV_STATE_FAULTED, /* External request to fault device */ VDEV_STATE_DEGRADED, /* Replicated vdev with unhealthy kids */ VDEV_STATE_HEALTHY /* Presumed good */ } vdev_state_t; #define VDEV_STATE_ONLINE VDEV_STATE_HEALTHY /* * vdev aux states. When a vdev is in the CANT_OPEN state, the aux field * of the vdev stats structure uses these constants to distinguish why. */ typedef enum vdev_aux { VDEV_AUX_NONE, /* no error */ VDEV_AUX_OPEN_FAILED, /* ldi_open_*() or vn_open() failed */ VDEV_AUX_CORRUPT_DATA, /* bad label or disk contents */ VDEV_AUX_NO_REPLICAS, /* insufficient number of replicas */ VDEV_AUX_BAD_GUID_SUM, /* vdev guid sum doesn't match */ VDEV_AUX_TOO_SMALL, /* vdev size is too small */ VDEV_AUX_BAD_LABEL, /* the label is OK but invalid */ VDEV_AUX_VERSION_NEWER, /* on-disk version is too new */ VDEV_AUX_VERSION_OLDER, /* on-disk version is too old */ VDEV_AUX_UNSUP_FEAT, /* unsupported features */ VDEV_AUX_SPARED, /* hot spare used in another pool */ VDEV_AUX_ERR_EXCEEDED, /* too many errors */ VDEV_AUX_IO_FAILURE, /* experienced I/O failure */ VDEV_AUX_BAD_LOG, /* cannot read log chain(s) */ VDEV_AUX_EXTERNAL, /* external diagnosis */ VDEV_AUX_SPLIT_POOL /* vdev was split off into another pool */ } vdev_aux_t; /* * pool state. The following states are written to disk as part of the normal * SPA lifecycle: ACTIVE, EXPORTED, DESTROYED, SPARE, L2CACHE. The remaining * states are software abstractions used at various levels to communicate * pool state. */ typedef enum pool_state { POOL_STATE_ACTIVE = 0, /* In active use */ POOL_STATE_EXPORTED, /* Explicitly exported */ POOL_STATE_DESTROYED, /* Explicitly destroyed */ POOL_STATE_SPARE, /* Reserved for hot spare use */ POOL_STATE_L2CACHE, /* Level 2 ARC device */ POOL_STATE_UNINITIALIZED, /* Internal spa_t state */ POOL_STATE_UNAVAIL, /* Internal libzfs state */ POOL_STATE_POTENTIALLY_ACTIVE /* Internal libzfs state */ } pool_state_t; /* * Scan Functions. */ typedef enum pool_scan_func { POOL_SCAN_NONE, POOL_SCAN_SCRUB, POOL_SCAN_RESILVER, POOL_SCAN_FUNCS } pool_scan_func_t; /* * ZIO types. Needed to interpret vdev statistics below. */ typedef enum zio_type { ZIO_TYPE_NULL = 0, ZIO_TYPE_READ, ZIO_TYPE_WRITE, ZIO_TYPE_FREE, ZIO_TYPE_CLAIM, ZIO_TYPE_IOCTL, ZIO_TYPES } zio_type_t; /* * Pool statistics. Note: all fields should be 64-bit because this * is passed between kernel and userland as an nvlist uint64 array. */ typedef struct pool_scan_stat { /* values stored on disk */ uint64_t pss_func; /* pool_scan_func_t */ uint64_t pss_state; /* dsl_scan_state_t */ uint64_t pss_start_time; /* scan start time */ uint64_t pss_end_time; /* scan end time */ uint64_t pss_to_examine; /* total bytes to scan */ uint64_t pss_examined; /* total examined bytes */ uint64_t pss_to_process; /* total bytes to process */ uint64_t pss_processed; /* total processed bytes */ uint64_t pss_errors; /* scan errors */ /* values not stored on disk */ uint64_t pss_pass_exam; /* examined bytes per scan pass */ uint64_t pss_pass_start; /* start time of a scan pass */ } pool_scan_stat_t; typedef enum dsl_scan_state { DSS_NONE, DSS_SCANNING, DSS_FINISHED, DSS_CANCELED, DSS_NUM_STATES } dsl_scan_state_t; /* * Errata described by http://zfsonlinux.org/msg/ZFS-8000-ER. The ordering * of this enum must be maintained to ensure the errata identifiers map to * the correct documentation. New errata may only be appended to the list * and must contain corresponding documentation at the above link. */ typedef enum zpool_errata { ZPOOL_ERRATA_NONE, ZPOOL_ERRATA_ZOL_2094_SCRUB, ZPOOL_ERRATA_ZOL_2094_ASYNC_DESTROY, } zpool_errata_t; /* * Vdev statistics. Note: all fields should be 64-bit because this * is passed between kernel and userland as an nvlist uint64 array. */ typedef struct vdev_stat { hrtime_t vs_timestamp; /* time since vdev load */ uint64_t vs_state; /* vdev state */ uint64_t vs_aux; /* see vdev_aux_t */ uint64_t vs_alloc; /* space allocated */ uint64_t vs_space; /* total capacity */ uint64_t vs_dspace; /* deflated capacity */ uint64_t vs_rsize; /* replaceable dev size */ uint64_t vs_esize; /* expandable dev size */ uint64_t vs_ops[ZIO_TYPES]; /* operation count */ uint64_t vs_bytes[ZIO_TYPES]; /* bytes read/written */ uint64_t vs_read_errors; /* read errors */ uint64_t vs_write_errors; /* write errors */ uint64_t vs_checksum_errors; /* checksum errors */ uint64_t vs_self_healed; /* self-healed bytes */ uint64_t vs_scan_removing; /* removing? */ uint64_t vs_scan_processed; /* scan processed bytes */ } vdev_stat_t; /* * DDT statistics. Note: all fields should be 64-bit because this * is passed between kernel and userland as an nvlist uint64 array. */ typedef struct ddt_object { uint64_t ddo_count; /* number of elments in ddt */ uint64_t ddo_dspace; /* size of ddt on disk */ uint64_t ddo_mspace; /* size of ddt in-core */ } ddt_object_t; typedef struct ddt_stat { uint64_t dds_blocks; /* blocks */ uint64_t dds_lsize; /* logical size */ uint64_t dds_psize; /* physical size */ uint64_t dds_dsize; /* deflated allocated size */ uint64_t dds_ref_blocks; /* referenced blocks */ uint64_t dds_ref_lsize; /* referenced lsize * refcnt */ uint64_t dds_ref_psize; /* referenced psize * refcnt */ uint64_t dds_ref_dsize; /* referenced dsize * refcnt */ } ddt_stat_t; typedef struct ddt_histogram { ddt_stat_t ddh_stat[64]; /* power-of-two histogram buckets */ } ddt_histogram_t; #define ZVOL_DRIVER "zvol" #define ZFS_DRIVER "zfs" #define ZFS_DEV "/dev/zfs" /* general zvol path */ #define ZVOL_DIR "/dev" #define ZVOL_MAJOR 230 #define ZVOL_MINOR_BITS 4 #define ZVOL_MINOR_MASK ((1U << ZVOL_MINOR_BITS) - 1) #define ZVOL_MINORS (1 << 4) #define ZVOL_DEV_NAME "zd" #define ZVOL_PROP_NAME "name" #define ZVOL_DEFAULT_BLOCKSIZE 8192 /* * /dev/zfs ioctl numbers. */ typedef enum zfs_ioc { /* * Illumos - 70/128 numbers reserved. */ ZFS_IOC_FIRST = ('Z' << 8), ZFS_IOC = ZFS_IOC_FIRST, ZFS_IOC_POOL_CREATE = ZFS_IOC_FIRST, ZFS_IOC_POOL_DESTROY, ZFS_IOC_POOL_IMPORT, ZFS_IOC_POOL_EXPORT, ZFS_IOC_POOL_CONFIGS, ZFS_IOC_POOL_STATS, ZFS_IOC_POOL_TRYIMPORT, ZFS_IOC_POOL_SCAN, ZFS_IOC_POOL_FREEZE, ZFS_IOC_POOL_UPGRADE, ZFS_IOC_POOL_GET_HISTORY, ZFS_IOC_VDEV_ADD, ZFS_IOC_VDEV_REMOVE, ZFS_IOC_VDEV_SET_STATE, ZFS_IOC_VDEV_ATTACH, ZFS_IOC_VDEV_DETACH, ZFS_IOC_VDEV_SETPATH, ZFS_IOC_VDEV_SETFRU, ZFS_IOC_OBJSET_STATS, ZFS_IOC_OBJSET_ZPLPROPS, ZFS_IOC_DATASET_LIST_NEXT, ZFS_IOC_SNAPSHOT_LIST_NEXT, ZFS_IOC_SET_PROP, ZFS_IOC_CREATE, ZFS_IOC_DESTROY, ZFS_IOC_ROLLBACK, ZFS_IOC_RENAME, ZFS_IOC_RECV, ZFS_IOC_SEND, ZFS_IOC_INJECT_FAULT, ZFS_IOC_CLEAR_FAULT, ZFS_IOC_INJECT_LIST_NEXT, ZFS_IOC_ERROR_LOG, ZFS_IOC_CLEAR, ZFS_IOC_PROMOTE, ZFS_IOC_SNAPSHOT, ZFS_IOC_DSOBJ_TO_DSNAME, ZFS_IOC_OBJ_TO_PATH, ZFS_IOC_POOL_SET_PROPS, ZFS_IOC_POOL_GET_PROPS, ZFS_IOC_SET_FSACL, ZFS_IOC_GET_FSACL, ZFS_IOC_SHARE, ZFS_IOC_INHERIT_PROP, ZFS_IOC_SMB_ACL, ZFS_IOC_USERSPACE_ONE, ZFS_IOC_USERSPACE_MANY, ZFS_IOC_USERSPACE_UPGRADE, ZFS_IOC_HOLD, ZFS_IOC_RELEASE, ZFS_IOC_GET_HOLDS, ZFS_IOC_OBJSET_RECVD_PROPS, ZFS_IOC_VDEV_SPLIT, ZFS_IOC_NEXT_OBJ, ZFS_IOC_DIFF, ZFS_IOC_TMP_SNAPSHOT, ZFS_IOC_OBJ_TO_STATS, ZFS_IOC_SPACE_WRITTEN, ZFS_IOC_SPACE_SNAPS, ZFS_IOC_DESTROY_SNAPS, ZFS_IOC_POOL_REGUID, ZFS_IOC_POOL_REOPEN, ZFS_IOC_SEND_PROGRESS, ZFS_IOC_LOG_HISTORY, ZFS_IOC_SEND_NEW, ZFS_IOC_SEND_SPACE, ZFS_IOC_CLONE, ZFS_IOC_BOOKMARK, ZFS_IOC_GET_BOOKMARKS, ZFS_IOC_DESTROY_BOOKMARKS, /* * Linux - 3/64 numbers reserved. */ ZFS_IOC_LINUX = ('Z' << 8) + 0x80, ZFS_IOC_EVENTS_NEXT, ZFS_IOC_EVENTS_CLEAR, ZFS_IOC_EVENTS_SEEK, /* * FreeBSD - 1/64 numbers reserved. */ ZFS_IOC_FREEBSD = ('Z' << 8) + 0xC0, ZFS_IOC_LAST } zfs_ioc_t; /* * zvol ioctl to get dataset name */ #define BLKZNAME _IOR(0x12, 125, char[ZFS_MAXNAMELEN]) /* * Internal SPA load state. Used by FMA diagnosis engine. */ typedef enum { SPA_LOAD_NONE, /* no load in progress */ SPA_LOAD_OPEN, /* normal open */ SPA_LOAD_IMPORT, /* import in progress */ SPA_LOAD_TRYIMPORT, /* tryimport in progress */ SPA_LOAD_RECOVER, /* recovery requested */ SPA_LOAD_ERROR /* load failed */ } spa_load_state_t; /* * Bookmark name values. */ #define ZPOOL_ERR_LIST "error list" #define ZPOOL_ERR_DATASET "dataset" #define ZPOOL_ERR_OBJECT "object" #define HIS_MAX_RECORD_LEN (MAXPATHLEN + MAXPATHLEN + 1) /* * The following are names used in the nvlist describing * the pool's history log. */ #define ZPOOL_HIST_RECORD "history record" #define ZPOOL_HIST_TIME "history time" #define ZPOOL_HIST_CMD "history command" #define ZPOOL_HIST_WHO "history who" #define ZPOOL_HIST_ZONE "history zone" #define ZPOOL_HIST_HOST "history hostname" #define ZPOOL_HIST_TXG "history txg" #define ZPOOL_HIST_INT_EVENT "history internal event" #define ZPOOL_HIST_INT_STR "history internal str" #define ZPOOL_HIST_INT_NAME "internal_name" #define ZPOOL_HIST_IOCTL "ioctl" #define ZPOOL_HIST_INPUT_NVL "in_nvl" #define ZPOOL_HIST_OUTPUT_NVL "out_nvl" #define ZPOOL_HIST_DSNAME "dsname" #define ZPOOL_HIST_DSID "dsid" /* * Flags for ZFS_IOC_VDEV_SET_STATE */ #define ZFS_ONLINE_CHECKREMOVE 0x1 #define ZFS_ONLINE_UNSPARE 0x2 #define ZFS_ONLINE_FORCEFAULT 0x4 #define ZFS_ONLINE_EXPAND 0x8 #define ZFS_OFFLINE_TEMPORARY 0x1 /* * Flags for ZFS_IOC_POOL_IMPORT */ #define ZFS_IMPORT_NORMAL 0x0 #define ZFS_IMPORT_VERBATIM 0x1 #define ZFS_IMPORT_ANY_HOST 0x2 #define ZFS_IMPORT_MISSING_LOG 0x4 #define ZFS_IMPORT_ONLY 0x8 #define ZFS_IMPORT_TEMP_NAME 0x10 /* * Sysevent payload members. ZFS will generate the following sysevents with the * given payloads: * * ESC_ZFS_RESILVER_START * ESC_ZFS_RESILVER_END * ESC_ZFS_POOL_DESTROY * ESC_ZFS_POOL_REGUID * * ZFS_EV_POOL_NAME DATA_TYPE_STRING * ZFS_EV_POOL_GUID DATA_TYPE_UINT64 * * ESC_ZFS_VDEV_REMOVE * ESC_ZFS_VDEV_CLEAR * ESC_ZFS_VDEV_CHECK * * ZFS_EV_POOL_NAME DATA_TYPE_STRING * ZFS_EV_POOL_GUID DATA_TYPE_UINT64 * ZFS_EV_VDEV_PATH DATA_TYPE_STRING (optional) * ZFS_EV_VDEV_GUID DATA_TYPE_UINT64 */ #define ZFS_EV_POOL_NAME "pool_name" #define ZFS_EV_POOL_GUID "pool_guid" #define ZFS_EV_VDEV_PATH "vdev_path" #define ZFS_EV_VDEV_GUID "vdev_guid" #ifdef __cplusplus } #endif #endif /* _SYS_FS_ZFS_H */ diff --git a/include/sys/zfs_debug.h b/include/sys/zfs_debug.h index e51207955555..829b37a4674f 100644 --- a/include/sys/zfs_debug.h +++ b/include/sys/zfs_debug.h @@ -1,109 +1,110 @@ /* * 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 by Delphix. All rights reserved. */ #ifndef _SYS_ZFS_DEBUG_H #define _SYS_ZFS_DEBUG_H #ifdef __cplusplus extern "C" { #endif #ifndef TRUE #define TRUE 1 #endif #ifndef FALSE #define FALSE 0 #endif /* * ZFS debugging - Always enabled for user space builds. */ #if !defined(ZFS_DEBUG) && !defined(_KERNEL) #define ZFS_DEBUG #endif extern int zfs_flags; extern int zfs_recover; +extern int zfs_free_leak_on_eio; #define ZFS_DEBUG_DPRINTF (1<<0) #define ZFS_DEBUG_DBUF_VERIFY (1<<1) #define ZFS_DEBUG_DNODE_VERIFY (1<<2) #define ZFS_DEBUG_SNAPNAMES (1<<3) #define ZFS_DEBUG_MODIFY (1<<4) #define ZFS_DEBUG_SPA (1<<5) #define ZFS_DEBUG_ZIO_FREE (1<<6) /* * Always log zfs debug messages to the spl debug subsystem as SS_USER1. * When the SPL is configured with debugging enabled these messages will * appear in the internal spl debug log, otherwise they are a no-op. */ #if defined(_KERNEL) #include #define dprintf(...) \ if (zfs_flags & ZFS_DEBUG_DPRINTF) \ __SDEBUG(NULL, SS_USER1, SD_DPRINTF, __VA_ARGS__) /* * When zfs is running is user space the debugging is always enabled. * The messages will be printed using the __dprintf() function and * filtered based on the zfs_flags variable. */ #else #define dprintf(...) \ if (zfs_flags & ZFS_DEBUG_DPRINTF) \ __dprintf(__FILE__, __func__, __LINE__, __VA_ARGS__) #endif /* _KERNEL */ extern void zfs_panic_recover(const char *fmt, ...); typedef struct zfs_dbgmsg { list_node_t zdm_node; time_t zdm_timestamp; char zdm_msg[1]; /* variable length allocation */ } zfs_dbgmsg_t; extern void zfs_dbgmsg_init(void); extern void zfs_dbgmsg_fini(void); #if defined(_KERNEL) && defined(__linux__) #define zfs_dbgmsg(...) dprintf(__VA_ARGS__) #else extern void zfs_dbgmsg(const char *fmt, ...); extern void zfs_dbgmsg_print(const char *tag); #endif #ifndef _KERNEL extern int dprintf_find_string(const char *string); #endif #ifdef __cplusplus } #endif #endif /* _SYS_ZFS_DEBUG_H */ diff --git a/lib/libzfs/libzfs_pool.c b/lib/libzfs/libzfs_pool.c index a5d2c28affe4..fce8fa96d2ce 100644 --- a/lib/libzfs/libzfs_pool.c +++ b/lib/libzfs/libzfs_pool.c @@ -1,4211 +1,4212 @@ /* * 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) 2013 by Delphix. All rights reserved. */ #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include "zfs_namecheck.h" #include "zfs_prop.h" #include "libzfs_impl.h" #include "zfs_comutil.h" #include "zfeature_common.h" static int read_efi_label(nvlist_t *config, diskaddr_t *sb); typedef struct prop_flags { int create:1; /* Validate property on creation */ int import:1; /* Validate property on import */ } prop_flags_t; /* * ==================================================================== * zpool property functions * ==================================================================== */ static int zpool_get_all_props(zpool_handle_t *zhp) { zfs_cmd_t zc = {"\0"}; libzfs_handle_t *hdl = zhp->zpool_hdl; (void) strlcpy(zc.zc_name, zhp->zpool_name, sizeof (zc.zc_name)); if (zcmd_alloc_dst_nvlist(hdl, &zc, 0) != 0) return (-1); while (ioctl(hdl->libzfs_fd, ZFS_IOC_POOL_GET_PROPS, &zc) != 0) { if (errno == ENOMEM) { if (zcmd_expand_dst_nvlist(hdl, &zc) != 0) { zcmd_free_nvlists(&zc); return (-1); } } else { zcmd_free_nvlists(&zc); return (-1); } } if (zcmd_read_dst_nvlist(hdl, &zc, &zhp->zpool_props) != 0) { zcmd_free_nvlists(&zc); return (-1); } zcmd_free_nvlists(&zc); return (0); } static int zpool_props_refresh(zpool_handle_t *zhp) { nvlist_t *old_props; old_props = zhp->zpool_props; if (zpool_get_all_props(zhp) != 0) return (-1); nvlist_free(old_props); return (0); } static char * zpool_get_prop_string(zpool_handle_t *zhp, zpool_prop_t prop, zprop_source_t *src) { nvlist_t *nv, *nvl; uint64_t ival; char *value; zprop_source_t source; nvl = zhp->zpool_props; if (nvlist_lookup_nvlist(nvl, zpool_prop_to_name(prop), &nv) == 0) { verify(nvlist_lookup_uint64(nv, ZPROP_SOURCE, &ival) == 0); source = ival; verify(nvlist_lookup_string(nv, ZPROP_VALUE, &value) == 0); } else { source = ZPROP_SRC_DEFAULT; if ((value = (char *)zpool_prop_default_string(prop)) == NULL) value = "-"; } if (src) *src = source; return (value); } uint64_t zpool_get_prop_int(zpool_handle_t *zhp, zpool_prop_t prop, zprop_source_t *src) { nvlist_t *nv, *nvl; uint64_t value; zprop_source_t source; if (zhp->zpool_props == NULL && zpool_get_all_props(zhp)) { /* * zpool_get_all_props() has most likely failed because * the pool is faulted, but if all we need is the top level * vdev's guid then get it from the zhp config nvlist. */ if ((prop == ZPOOL_PROP_GUID) && (nvlist_lookup_nvlist(zhp->zpool_config, ZPOOL_CONFIG_VDEV_TREE, &nv) == 0) && (nvlist_lookup_uint64(nv, ZPOOL_CONFIG_GUID, &value) == 0)) { return (value); } return (zpool_prop_default_numeric(prop)); } nvl = zhp->zpool_props; if (nvlist_lookup_nvlist(nvl, zpool_prop_to_name(prop), &nv) == 0) { verify(nvlist_lookup_uint64(nv, ZPROP_SOURCE, &value) == 0); source = value; verify(nvlist_lookup_uint64(nv, ZPROP_VALUE, &value) == 0); } else { source = ZPROP_SRC_DEFAULT; value = zpool_prop_default_numeric(prop); } if (src) *src = source; return (value); } /* * Map VDEV STATE to printed strings. */ char * zpool_state_to_name(vdev_state_t state, vdev_aux_t aux) { switch (state) { default: break; case VDEV_STATE_CLOSED: case VDEV_STATE_OFFLINE: return (gettext("OFFLINE")); case VDEV_STATE_REMOVED: return (gettext("REMOVED")); case VDEV_STATE_CANT_OPEN: if (aux == VDEV_AUX_CORRUPT_DATA || aux == VDEV_AUX_BAD_LOG) return (gettext("FAULTED")); else if (aux == VDEV_AUX_SPLIT_POOL) return (gettext("SPLIT")); else return (gettext("UNAVAIL")); case VDEV_STATE_FAULTED: return (gettext("FAULTED")); case VDEV_STATE_DEGRADED: return (gettext("DEGRADED")); case VDEV_STATE_HEALTHY: return (gettext("ONLINE")); } return (gettext("UNKNOWN")); } /* * Map POOL STATE to printed strings. */ const char * zpool_pool_state_to_name(pool_state_t state) { switch (state) { default: break; case POOL_STATE_ACTIVE: return (gettext("ACTIVE")); case POOL_STATE_EXPORTED: return (gettext("EXPORTED")); case POOL_STATE_DESTROYED: return (gettext("DESTROYED")); case POOL_STATE_SPARE: return (gettext("SPARE")); case POOL_STATE_L2CACHE: return (gettext("L2CACHE")); case POOL_STATE_UNINITIALIZED: return (gettext("UNINITIALIZED")); case POOL_STATE_UNAVAIL: return (gettext("UNAVAIL")); case POOL_STATE_POTENTIALLY_ACTIVE: return (gettext("POTENTIALLY_ACTIVE")); } return (gettext("UNKNOWN")); } /* * API compatibility wrapper around zpool_get_prop_literal */ int zpool_get_prop(zpool_handle_t *zhp, zpool_prop_t prop, char *buf, size_t len, zprop_source_t *srctype) { return (zpool_get_prop_literal(zhp, prop, buf, len, srctype, B_FALSE)); } /* * Get a zpool property value for 'prop' and return the value in * a pre-allocated buffer. */ int zpool_get_prop_literal(zpool_handle_t *zhp, zpool_prop_t prop, char *buf, size_t len, zprop_source_t *srctype, boolean_t literal) { uint64_t intval; const char *strval; zprop_source_t src = ZPROP_SRC_NONE; nvlist_t *nvroot; vdev_stat_t *vs; uint_t vsc; if (zpool_get_state(zhp) == POOL_STATE_UNAVAIL) { switch (prop) { case ZPOOL_PROP_NAME: (void) strlcpy(buf, zpool_get_name(zhp), len); break; case ZPOOL_PROP_HEALTH: (void) strlcpy(buf, "FAULTED", len); break; case ZPOOL_PROP_GUID: intval = zpool_get_prop_int(zhp, prop, &src); (void) snprintf(buf, len, "%llu", (u_longlong_t)intval); break; case ZPOOL_PROP_ALTROOT: case ZPOOL_PROP_CACHEFILE: case ZPOOL_PROP_COMMENT: if (zhp->zpool_props != NULL || zpool_get_all_props(zhp) == 0) { (void) strlcpy(buf, zpool_get_prop_string(zhp, prop, &src), len); if (srctype != NULL) *srctype = src; return (0); } /* FALLTHROUGH */ default: (void) strlcpy(buf, "-", len); break; } if (srctype != NULL) *srctype = src; return (0); } if (zhp->zpool_props == NULL && zpool_get_all_props(zhp) && prop != ZPOOL_PROP_NAME) return (-1); switch (zpool_prop_get_type(prop)) { case PROP_TYPE_STRING: (void) strlcpy(buf, zpool_get_prop_string(zhp, prop, &src), len); break; case PROP_TYPE_NUMBER: intval = zpool_get_prop_int(zhp, prop, &src); switch (prop) { case ZPOOL_PROP_SIZE: case ZPOOL_PROP_ALLOCATED: case ZPOOL_PROP_FREE: case ZPOOL_PROP_FREEING: + case ZPOOL_PROP_LEAKED: case ZPOOL_PROP_EXPANDSZ: case ZPOOL_PROP_ASHIFT: if (literal) (void) snprintf(buf, len, "%llu", (u_longlong_t)intval); else (void) zfs_nicenum(intval, buf, len); break; case ZPOOL_PROP_CAPACITY: (void) snprintf(buf, len, "%llu%%", (u_longlong_t)intval); break; case ZPOOL_PROP_DEDUPRATIO: (void) snprintf(buf, len, "%llu.%02llux", (u_longlong_t)(intval / 100), (u_longlong_t)(intval % 100)); break; case ZPOOL_PROP_HEALTH: verify(nvlist_lookup_nvlist(zpool_get_config(zhp, NULL), ZPOOL_CONFIG_VDEV_TREE, &nvroot) == 0); verify(nvlist_lookup_uint64_array(nvroot, ZPOOL_CONFIG_VDEV_STATS, (uint64_t **)&vs, &vsc) == 0); (void) strlcpy(buf, zpool_state_to_name(intval, vs->vs_aux), len); break; case ZPOOL_PROP_VERSION: if (intval >= SPA_VERSION_FEATURES) { (void) snprintf(buf, len, "-"); break; } /* FALLTHROUGH */ default: (void) snprintf(buf, len, "%llu", (u_longlong_t)intval); } break; case PROP_TYPE_INDEX: intval = zpool_get_prop_int(zhp, prop, &src); if (zpool_prop_index_to_string(prop, intval, &strval) != 0) return (-1); (void) strlcpy(buf, strval, len); break; default: abort(); } if (srctype) *srctype = src; return (0); } /* * Check if the bootfs name has the same pool name as it is set to. * Assuming bootfs is a valid dataset name. */ static boolean_t bootfs_name_valid(const char *pool, char *bootfs) { int len = strlen(pool); if (!zfs_name_valid(bootfs, ZFS_TYPE_FILESYSTEM|ZFS_TYPE_SNAPSHOT)) return (B_FALSE); if (strncmp(pool, bootfs, len) == 0 && (bootfs[len] == '/' || bootfs[len] == '\0')) return (B_TRUE); return (B_FALSE); } #if defined(__sun__) || defined(__sun) /* * Inspect the configuration to determine if any of the devices contain * an EFI label. */ static boolean_t pool_uses_efi(nvlist_t *config) { nvlist_t **child; uint_t c, children; if (nvlist_lookup_nvlist_array(config, ZPOOL_CONFIG_CHILDREN, &child, &children) != 0) return (read_efi_label(config, NULL) >= 0); for (c = 0; c < children; c++) { if (pool_uses_efi(child[c])) return (B_TRUE); } return (B_FALSE); } #endif boolean_t zpool_is_bootable(zpool_handle_t *zhp) { char bootfs[ZPOOL_MAXNAMELEN]; return (zpool_get_prop(zhp, ZPOOL_PROP_BOOTFS, bootfs, sizeof (bootfs), NULL) == 0 && strncmp(bootfs, "-", sizeof (bootfs)) != 0); } /* * Given an nvlist of zpool properties to be set, validate that they are * correct, and parse any numeric properties (index, boolean, etc) if they are * specified as strings. */ static nvlist_t * zpool_valid_proplist(libzfs_handle_t *hdl, const char *poolname, nvlist_t *props, uint64_t version, prop_flags_t flags, char *errbuf) { nvpair_t *elem; nvlist_t *retprops; zpool_prop_t prop; char *strval; uint64_t intval; char *slash, *check; struct stat64 statbuf; zpool_handle_t *zhp; nvlist_t *nvroot; if (nvlist_alloc(&retprops, NV_UNIQUE_NAME, 0) != 0) { (void) no_memory(hdl); return (NULL); } elem = NULL; while ((elem = nvlist_next_nvpair(props, elem)) != NULL) { const char *propname = nvpair_name(elem); prop = zpool_name_to_prop(propname); if (prop == ZPROP_INVAL && zpool_prop_feature(propname)) { int err; char *fname = strchr(propname, '@') + 1; err = zfeature_lookup_name(fname, NULL); if (err != 0) { ASSERT3U(err, ==, ENOENT); zfs_error_aux(hdl, dgettext(TEXT_DOMAIN, "invalid feature '%s'"), fname); (void) zfs_error(hdl, EZFS_BADPROP, errbuf); goto error; } if (nvpair_type(elem) != DATA_TYPE_STRING) { zfs_error_aux(hdl, dgettext(TEXT_DOMAIN, "'%s' must be a string"), propname); (void) zfs_error(hdl, EZFS_BADPROP, errbuf); goto error; } (void) nvpair_value_string(elem, &strval); if (strcmp(strval, ZFS_FEATURE_ENABLED) != 0) { zfs_error_aux(hdl, dgettext(TEXT_DOMAIN, "property '%s' can only be set to " "'enabled'"), propname); (void) zfs_error(hdl, EZFS_BADPROP, errbuf); goto error; } if (nvlist_add_uint64(retprops, propname, 0) != 0) { (void) no_memory(hdl); goto error; } continue; } /* * Make sure this property is valid and applies to this type. */ if (prop == ZPROP_INVAL) { zfs_error_aux(hdl, dgettext(TEXT_DOMAIN, "invalid property '%s'"), propname); (void) zfs_error(hdl, EZFS_BADPROP, errbuf); goto error; } if (zpool_prop_readonly(prop)) { zfs_error_aux(hdl, dgettext(TEXT_DOMAIN, "'%s' " "is readonly"), propname); (void) zfs_error(hdl, EZFS_PROPREADONLY, errbuf); goto error; } if (zprop_parse_value(hdl, elem, prop, ZFS_TYPE_POOL, retprops, &strval, &intval, errbuf) != 0) goto error; /* * Perform additional checking for specific properties. */ switch (prop) { default: break; case ZPOOL_PROP_VERSION: if (intval < version || !SPA_VERSION_IS_SUPPORTED(intval)) { zfs_error_aux(hdl, dgettext(TEXT_DOMAIN, "property '%s' number %d is invalid."), propname, intval); (void) zfs_error(hdl, EZFS_BADVERSION, errbuf); goto error; } break; case ZPOOL_PROP_ASHIFT: if (!flags.create) { zfs_error_aux(hdl, dgettext(TEXT_DOMAIN, "property '%s' can only be set at " "creation time"), propname); (void) zfs_error(hdl, EZFS_BADPROP, errbuf); goto error; } if (intval != 0 && (intval < 9 || intval > 13)) { zfs_error_aux(hdl, dgettext(TEXT_DOMAIN, "property '%s' number %d is invalid."), propname, intval); (void) zfs_error(hdl, EZFS_BADPROP, errbuf); goto error; } break; case ZPOOL_PROP_BOOTFS: if (flags.create || flags.import) { zfs_error_aux(hdl, dgettext(TEXT_DOMAIN, "property '%s' cannot be set at creation " "or import time"), propname); (void) zfs_error(hdl, EZFS_BADPROP, errbuf); goto error; } if (version < SPA_VERSION_BOOTFS) { zfs_error_aux(hdl, dgettext(TEXT_DOMAIN, "pool must be upgraded to support " "'%s' property"), propname); (void) zfs_error(hdl, EZFS_BADVERSION, errbuf); goto error; } /* * bootfs property value has to be a dataset name and * the dataset has to be in the same pool as it sets to. */ if (strval[0] != '\0' && !bootfs_name_valid(poolname, strval)) { zfs_error_aux(hdl, dgettext(TEXT_DOMAIN, "'%s' " "is an invalid name"), strval); (void) zfs_error(hdl, EZFS_INVALIDNAME, errbuf); goto error; } if ((zhp = zpool_open_canfail(hdl, poolname)) == NULL) { zfs_error_aux(hdl, dgettext(TEXT_DOMAIN, "could not open pool '%s'"), poolname); (void) zfs_error(hdl, EZFS_OPENFAILED, errbuf); goto error; } verify(nvlist_lookup_nvlist(zpool_get_config(zhp, NULL), ZPOOL_CONFIG_VDEV_TREE, &nvroot) == 0); #if defined(__sun__) || defined(__sun) /* * bootfs property cannot be set on a disk which has * been EFI labeled. */ if (pool_uses_efi(nvroot)) { zfs_error_aux(hdl, dgettext(TEXT_DOMAIN, "property '%s' not supported on " "EFI labeled devices"), propname); (void) zfs_error(hdl, EZFS_POOL_NOTSUP, errbuf); zpool_close(zhp); goto error; } #endif zpool_close(zhp); break; case ZPOOL_PROP_ALTROOT: if (!flags.create && !flags.import) { zfs_error_aux(hdl, dgettext(TEXT_DOMAIN, "property '%s' can only be set during pool " "creation or import"), propname); (void) zfs_error(hdl, EZFS_BADPROP, errbuf); goto error; } if (strval[0] != '/') { zfs_error_aux(hdl, dgettext(TEXT_DOMAIN, "bad alternate root '%s'"), strval); (void) zfs_error(hdl, EZFS_BADPATH, errbuf); goto error; } break; case ZPOOL_PROP_CACHEFILE: if (strval[0] == '\0') break; if (strcmp(strval, "none") == 0) break; if (strval[0] != '/') { zfs_error_aux(hdl, dgettext(TEXT_DOMAIN, "property '%s' must be empty, an " "absolute path, or 'none'"), propname); (void) zfs_error(hdl, EZFS_BADPATH, errbuf); goto error; } slash = strrchr(strval, '/'); if (slash[1] == '\0' || strcmp(slash, "/.") == 0 || strcmp(slash, "/..") == 0) { zfs_error_aux(hdl, dgettext(TEXT_DOMAIN, "'%s' is not a valid file"), strval); (void) zfs_error(hdl, EZFS_BADPATH, errbuf); goto error; } *slash = '\0'; if (strval[0] != '\0' && (stat64(strval, &statbuf) != 0 || !S_ISDIR(statbuf.st_mode))) { zfs_error_aux(hdl, dgettext(TEXT_DOMAIN, "'%s' is not a valid directory"), strval); (void) zfs_error(hdl, EZFS_BADPATH, errbuf); goto error; } *slash = '/'; break; case ZPOOL_PROP_COMMENT: for (check = strval; *check != '\0'; check++) { if (!isprint(*check)) { zfs_error_aux(hdl, dgettext(TEXT_DOMAIN, "comment may only have printable " "characters")); (void) zfs_error(hdl, EZFS_BADPROP, errbuf); goto error; } } if (strlen(strval) > ZPROP_MAX_COMMENT) { zfs_error_aux(hdl, dgettext(TEXT_DOMAIN, "comment must not exceed %d characters"), ZPROP_MAX_COMMENT); (void) zfs_error(hdl, EZFS_BADPROP, errbuf); goto error; } break; case ZPOOL_PROP_READONLY: if (!flags.import) { zfs_error_aux(hdl, dgettext(TEXT_DOMAIN, "property '%s' can only be set at " "import time"), propname); (void) zfs_error(hdl, EZFS_BADPROP, errbuf); goto error; } break; } } return (retprops); error: nvlist_free(retprops); return (NULL); } /* * Set zpool property : propname=propval. */ int zpool_set_prop(zpool_handle_t *zhp, const char *propname, const char *propval) { zfs_cmd_t zc = {"\0"}; int ret = -1; char errbuf[1024]; nvlist_t *nvl = NULL; nvlist_t *realprops; uint64_t version; prop_flags_t flags = { 0 }; (void) snprintf(errbuf, sizeof (errbuf), dgettext(TEXT_DOMAIN, "cannot set property for '%s'"), zhp->zpool_name); if (nvlist_alloc(&nvl, NV_UNIQUE_NAME, 0) != 0) return (no_memory(zhp->zpool_hdl)); if (nvlist_add_string(nvl, propname, propval) != 0) { nvlist_free(nvl); return (no_memory(zhp->zpool_hdl)); } version = zpool_get_prop_int(zhp, ZPOOL_PROP_VERSION, NULL); if ((realprops = zpool_valid_proplist(zhp->zpool_hdl, zhp->zpool_name, nvl, version, flags, errbuf)) == NULL) { nvlist_free(nvl); return (-1); } nvlist_free(nvl); nvl = realprops; /* * Execute the corresponding ioctl() to set this property. */ (void) strlcpy(zc.zc_name, zhp->zpool_name, sizeof (zc.zc_name)); if (zcmd_write_src_nvlist(zhp->zpool_hdl, &zc, nvl) != 0) { nvlist_free(nvl); return (-1); } ret = zfs_ioctl(zhp->zpool_hdl, ZFS_IOC_POOL_SET_PROPS, &zc); zcmd_free_nvlists(&zc); nvlist_free(nvl); if (ret) (void) zpool_standard_error(zhp->zpool_hdl, errno, errbuf); else (void) zpool_props_refresh(zhp); return (ret); } int zpool_expand_proplist(zpool_handle_t *zhp, zprop_list_t **plp) { libzfs_handle_t *hdl = zhp->zpool_hdl; zprop_list_t *entry; char buf[ZFS_MAXPROPLEN]; nvlist_t *features = NULL; nvpair_t *nvp; zprop_list_t **last; boolean_t firstexpand = (NULL == *plp); int i; if (zprop_expand_list(hdl, plp, ZFS_TYPE_POOL) != 0) return (-1); last = plp; while (*last != NULL) last = &(*last)->pl_next; if ((*plp)->pl_all) features = zpool_get_features(zhp); if ((*plp)->pl_all && firstexpand) { for (i = 0; i < SPA_FEATURES; i++) { zprop_list_t *entry = zfs_alloc(hdl, sizeof (zprop_list_t)); entry->pl_prop = ZPROP_INVAL; entry->pl_user_prop = zfs_asprintf(hdl, "feature@%s", spa_feature_table[i].fi_uname); entry->pl_width = strlen(entry->pl_user_prop); entry->pl_all = B_TRUE; *last = entry; last = &entry->pl_next; } } /* add any unsupported features */ for (nvp = nvlist_next_nvpair(features, NULL); nvp != NULL; nvp = nvlist_next_nvpair(features, nvp)) { char *propname; boolean_t found; zprop_list_t *entry; if (zfeature_is_supported(nvpair_name(nvp))) continue; propname = zfs_asprintf(hdl, "unsupported@%s", nvpair_name(nvp)); /* * Before adding the property to the list make sure that no * other pool already added the same property. */ found = B_FALSE; entry = *plp; while (entry != NULL) { if (entry->pl_user_prop != NULL && strcmp(propname, entry->pl_user_prop) == 0) { found = B_TRUE; break; } entry = entry->pl_next; } if (found) { free(propname); continue; } entry = zfs_alloc(hdl, sizeof (zprop_list_t)); entry->pl_prop = ZPROP_INVAL; entry->pl_user_prop = propname; entry->pl_width = strlen(entry->pl_user_prop); entry->pl_all = B_TRUE; *last = entry; last = &entry->pl_next; } for (entry = *plp; entry != NULL; entry = entry->pl_next) { if (entry->pl_fixed) continue; if (entry->pl_prop != ZPROP_INVAL && zpool_get_prop(zhp, entry->pl_prop, buf, sizeof (buf), NULL) == 0) { if (strlen(buf) > entry->pl_width) entry->pl_width = strlen(buf); } } return (0); } /* * Get the state for the given feature on the given ZFS pool. */ int zpool_prop_get_feature(zpool_handle_t *zhp, const char *propname, char *buf, size_t len) { uint64_t refcount; boolean_t found = B_FALSE; nvlist_t *features = zpool_get_features(zhp); boolean_t supported; const char *feature = strchr(propname, '@') + 1; supported = zpool_prop_feature(propname); ASSERT(supported || zpool_prop_unsupported(propname)); /* * Convert from feature name to feature guid. This conversion is * unecessary for unsupported@... properties because they already * use guids. */ if (supported) { int ret; spa_feature_t fid; ret = zfeature_lookup_name(feature, &fid); if (ret != 0) { (void) strlcpy(buf, "-", len); return (ENOTSUP); } feature = spa_feature_table[fid].fi_guid; } if (nvlist_lookup_uint64(features, feature, &refcount) == 0) found = B_TRUE; if (supported) { if (!found) { (void) strlcpy(buf, ZFS_FEATURE_DISABLED, len); } else { if (refcount == 0) (void) strlcpy(buf, ZFS_FEATURE_ENABLED, len); else (void) strlcpy(buf, ZFS_FEATURE_ACTIVE, len); } } else { if (found) { if (refcount == 0) { (void) strcpy(buf, ZFS_UNSUPPORTED_INACTIVE); } else { (void) strcpy(buf, ZFS_UNSUPPORTED_READONLY); } } else { (void) strlcpy(buf, "-", len); return (ENOTSUP); } } return (0); } /* * Don't start the slice at the default block of 34; many storage * devices will use a stripe width of 128k, other vendors prefer a 1m * alignment. It is best to play it safe and ensure a 1m alignment * given 512B blocks. When the block size is larger by a power of 2 * we will still be 1m aligned. Some devices are sensitive to the * partition ending alignment as well. */ #define NEW_START_BLOCK 2048 #define PARTITION_END_ALIGNMENT 2048 /* * Validate the given pool name, optionally putting an extended error message in * 'buf'. */ boolean_t zpool_name_valid(libzfs_handle_t *hdl, boolean_t isopen, const char *pool) { namecheck_err_t why; char what; int ret; ret = pool_namecheck(pool, &why, &what); /* * The rules for reserved pool names were extended at a later point. * But we need to support users with existing pools that may now be * invalid. So we only check for this expanded set of names during a * create (or import), and only in userland. */ if (ret == 0 && !isopen && (strncmp(pool, "mirror", 6) == 0 || strncmp(pool, "raidz", 5) == 0 || strncmp(pool, "spare", 5) == 0 || strcmp(pool, "log") == 0)) { if (hdl != NULL) zfs_error_aux(hdl, dgettext(TEXT_DOMAIN, "name is reserved")); return (B_FALSE); } if (ret != 0) { if (hdl != NULL) { switch (why) { case NAME_ERR_TOOLONG: zfs_error_aux(hdl, dgettext(TEXT_DOMAIN, "name is too long")); break; case NAME_ERR_INVALCHAR: zfs_error_aux(hdl, dgettext(TEXT_DOMAIN, "invalid character " "'%c' in pool name"), what); break; case NAME_ERR_NOLETTER: zfs_error_aux(hdl, dgettext(TEXT_DOMAIN, "name must begin with a letter")); break; case NAME_ERR_RESERVED: zfs_error_aux(hdl, dgettext(TEXT_DOMAIN, "name is reserved")); break; case NAME_ERR_DISKLIKE: zfs_error_aux(hdl, dgettext(TEXT_DOMAIN, "pool name is reserved")); break; case NAME_ERR_LEADING_SLASH: zfs_error_aux(hdl, dgettext(TEXT_DOMAIN, "leading slash in name")); break; case NAME_ERR_EMPTY_COMPONENT: zfs_error_aux(hdl, dgettext(TEXT_DOMAIN, "empty component in name")); break; case NAME_ERR_TRAILING_SLASH: zfs_error_aux(hdl, dgettext(TEXT_DOMAIN, "trailing slash in name")); break; case NAME_ERR_MULTIPLE_AT: zfs_error_aux(hdl, dgettext(TEXT_DOMAIN, "multiple '@' delimiters in name")); break; case NAME_ERR_NO_AT: zfs_error_aux(hdl, dgettext(TEXT_DOMAIN, "permission set is missing '@'")); break; } } return (B_FALSE); } return (B_TRUE); } /* * Open a handle to the given pool, even if the pool is currently in the FAULTED * state. */ zpool_handle_t * zpool_open_canfail(libzfs_handle_t *hdl, const char *pool) { zpool_handle_t *zhp; boolean_t missing; /* * Make sure the pool name is valid. */ if (!zpool_name_valid(hdl, B_TRUE, pool)) { (void) zfs_error_fmt(hdl, EZFS_INVALIDNAME, dgettext(TEXT_DOMAIN, "cannot open '%s'"), pool); return (NULL); } if ((zhp = zfs_alloc(hdl, sizeof (zpool_handle_t))) == NULL) return (NULL); zhp->zpool_hdl = hdl; (void) strlcpy(zhp->zpool_name, pool, sizeof (zhp->zpool_name)); if (zpool_refresh_stats(zhp, &missing) != 0) { zpool_close(zhp); return (NULL); } if (missing) { zfs_error_aux(hdl, dgettext(TEXT_DOMAIN, "no such pool")); (void) zfs_error_fmt(hdl, EZFS_NOENT, dgettext(TEXT_DOMAIN, "cannot open '%s'"), pool); zpool_close(zhp); return (NULL); } return (zhp); } /* * Like the above, but silent on error. Used when iterating over pools (because * the configuration cache may be out of date). */ int zpool_open_silent(libzfs_handle_t *hdl, const char *pool, zpool_handle_t **ret) { zpool_handle_t *zhp; boolean_t missing; if ((zhp = zfs_alloc(hdl, sizeof (zpool_handle_t))) == NULL) return (-1); zhp->zpool_hdl = hdl; (void) strlcpy(zhp->zpool_name, pool, sizeof (zhp->zpool_name)); if (zpool_refresh_stats(zhp, &missing) != 0) { zpool_close(zhp); return (-1); } if (missing) { zpool_close(zhp); *ret = NULL; return (0); } *ret = zhp; return (0); } /* * Similar to zpool_open_canfail(), but refuses to open pools in the faulted * state. */ zpool_handle_t * zpool_open(libzfs_handle_t *hdl, const char *pool) { zpool_handle_t *zhp; if ((zhp = zpool_open_canfail(hdl, pool)) == NULL) return (NULL); if (zhp->zpool_state == POOL_STATE_UNAVAIL) { (void) zfs_error_fmt(hdl, EZFS_POOLUNAVAIL, dgettext(TEXT_DOMAIN, "cannot open '%s'"), zhp->zpool_name); zpool_close(zhp); return (NULL); } return (zhp); } /* * Close the handle. Simply frees the memory associated with the handle. */ void zpool_close(zpool_handle_t *zhp) { if (zhp->zpool_config) nvlist_free(zhp->zpool_config); if (zhp->zpool_old_config) nvlist_free(zhp->zpool_old_config); if (zhp->zpool_props) nvlist_free(zhp->zpool_props); free(zhp); } /* * Return the name of the pool. */ const char * zpool_get_name(zpool_handle_t *zhp) { return (zhp->zpool_name); } /* * Return the state of the pool (ACTIVE or UNAVAILABLE) */ int zpool_get_state(zpool_handle_t *zhp) { return (zhp->zpool_state); } /* * Create the named pool, using the provided vdev list. It is assumed * that the consumer has already validated the contents of the nvlist, so we * don't have to worry about error semantics. */ int zpool_create(libzfs_handle_t *hdl, const char *pool, nvlist_t *nvroot, nvlist_t *props, nvlist_t *fsprops) { zfs_cmd_t zc = {"\0"}; nvlist_t *zc_fsprops = NULL; nvlist_t *zc_props = NULL; char msg[1024]; int ret = -1; (void) snprintf(msg, sizeof (msg), dgettext(TEXT_DOMAIN, "cannot create '%s'"), pool); if (!zpool_name_valid(hdl, B_FALSE, pool)) return (zfs_error(hdl, EZFS_INVALIDNAME, msg)); if (zcmd_write_conf_nvlist(hdl, &zc, nvroot) != 0) return (-1); if (props) { prop_flags_t flags = { .create = B_TRUE, .import = B_FALSE }; if ((zc_props = zpool_valid_proplist(hdl, pool, props, SPA_VERSION_1, flags, msg)) == NULL) { goto create_failed; } } if (fsprops) { uint64_t zoned; char *zonestr; zoned = ((nvlist_lookup_string(fsprops, zfs_prop_to_name(ZFS_PROP_ZONED), &zonestr) == 0) && strcmp(zonestr, "on") == 0); if ((zc_fsprops = zfs_valid_proplist(hdl, ZFS_TYPE_FILESYSTEM, fsprops, zoned, NULL, msg)) == NULL) { goto create_failed; } if (!zc_props && (nvlist_alloc(&zc_props, NV_UNIQUE_NAME, 0) != 0)) { goto create_failed; } if (nvlist_add_nvlist(zc_props, ZPOOL_ROOTFS_PROPS, zc_fsprops) != 0) { goto create_failed; } } if (zc_props && zcmd_write_src_nvlist(hdl, &zc, zc_props) != 0) goto create_failed; (void) strlcpy(zc.zc_name, pool, sizeof (zc.zc_name)); if ((ret = zfs_ioctl(hdl, ZFS_IOC_POOL_CREATE, &zc)) != 0) { zcmd_free_nvlists(&zc); nvlist_free(zc_props); nvlist_free(zc_fsprops); switch (errno) { case EBUSY: /* * This can happen if the user has specified the same * device multiple times. We can't reliably detect this * until we try to add it and see we already have a * label. This can also happen under if the device is * part of an active md or lvm device. */ zfs_error_aux(hdl, dgettext(TEXT_DOMAIN, "one or more vdevs refer to the same device, or " "one of\nthe devices is part of an active md or " "lvm device")); return (zfs_error(hdl, EZFS_BADDEV, msg)); case EOVERFLOW: /* * This occurs when one of the devices is below * SPA_MINDEVSIZE. Unfortunately, we can't detect which * device was the problem device since there's no * reliable way to determine device size from userland. */ { char buf[64]; zfs_nicenum(SPA_MINDEVSIZE, buf, sizeof (buf)); zfs_error_aux(hdl, dgettext(TEXT_DOMAIN, "one or more devices is less than the " "minimum size (%s)"), buf); } return (zfs_error(hdl, EZFS_BADDEV, msg)); case ENOSPC: zfs_error_aux(hdl, dgettext(TEXT_DOMAIN, "one or more devices is out of space")); return (zfs_error(hdl, EZFS_BADDEV, msg)); case ENOTBLK: zfs_error_aux(hdl, dgettext(TEXT_DOMAIN, "cache device must be a disk or disk slice")); return (zfs_error(hdl, EZFS_BADDEV, msg)); default: return (zpool_standard_error(hdl, errno, msg)); } } create_failed: zcmd_free_nvlists(&zc); nvlist_free(zc_props); nvlist_free(zc_fsprops); return (ret); } /* * Destroy the given pool. It is up to the caller to ensure that there are no * datasets left in the pool. */ int zpool_destroy(zpool_handle_t *zhp, const char *log_str) { zfs_cmd_t zc = {"\0"}; zfs_handle_t *zfp = NULL; libzfs_handle_t *hdl = zhp->zpool_hdl; char msg[1024]; if (zhp->zpool_state == POOL_STATE_ACTIVE && (zfp = zfs_open(hdl, zhp->zpool_name, ZFS_TYPE_FILESYSTEM)) == NULL) return (-1); (void) strlcpy(zc.zc_name, zhp->zpool_name, sizeof (zc.zc_name)); zc.zc_history = (uint64_t)(uintptr_t)log_str; if (zfs_ioctl(hdl, ZFS_IOC_POOL_DESTROY, &zc) != 0) { (void) snprintf(msg, sizeof (msg), dgettext(TEXT_DOMAIN, "cannot destroy '%s'"), zhp->zpool_name); if (errno == EROFS) { zfs_error_aux(hdl, dgettext(TEXT_DOMAIN, "one or more devices is read only")); (void) zfs_error(hdl, EZFS_BADDEV, msg); } else { (void) zpool_standard_error(hdl, errno, msg); } if (zfp) zfs_close(zfp); return (-1); } if (zfp) { remove_mountpoint(zfp); zfs_close(zfp); } return (0); } /* * Add the given vdevs to the pool. The caller must have already performed the * necessary verification to ensure that the vdev specification is well-formed. */ int zpool_add(zpool_handle_t *zhp, nvlist_t *nvroot) { zfs_cmd_t zc = {"\0"}; int ret; libzfs_handle_t *hdl = zhp->zpool_hdl; char msg[1024]; nvlist_t **spares, **l2cache; uint_t nspares, nl2cache; (void) snprintf(msg, sizeof (msg), dgettext(TEXT_DOMAIN, "cannot add to '%s'"), zhp->zpool_name); if (zpool_get_prop_int(zhp, ZPOOL_PROP_VERSION, NULL) < SPA_VERSION_SPARES && nvlist_lookup_nvlist_array(nvroot, ZPOOL_CONFIG_SPARES, &spares, &nspares) == 0) { zfs_error_aux(hdl, dgettext(TEXT_DOMAIN, "pool must be " "upgraded to add hot spares")); return (zfs_error(hdl, EZFS_BADVERSION, msg)); } #if defined(__sun__) || defined(__sun) if (zpool_is_bootable(zhp) && nvlist_lookup_nvlist_array(nvroot, ZPOOL_CONFIG_SPARES, &spares, &nspares) == 0) { uint64_t s; for (s = 0; s < nspares; s++) { char *path; if (nvlist_lookup_string(spares[s], ZPOOL_CONFIG_PATH, &path) == 0 && pool_uses_efi(spares[s])) { zfs_error_aux(hdl, dgettext(TEXT_DOMAIN, "device '%s' contains an EFI label and " "cannot be used on root pools."), zpool_vdev_name(hdl, NULL, spares[s], B_FALSE)); return (zfs_error(hdl, EZFS_POOL_NOTSUP, msg)); } } } #endif if (zpool_get_prop_int(zhp, ZPOOL_PROP_VERSION, NULL) < SPA_VERSION_L2CACHE && nvlist_lookup_nvlist_array(nvroot, ZPOOL_CONFIG_L2CACHE, &l2cache, &nl2cache) == 0) { zfs_error_aux(hdl, dgettext(TEXT_DOMAIN, "pool must be " "upgraded to add cache devices")); return (zfs_error(hdl, EZFS_BADVERSION, msg)); } if (zcmd_write_conf_nvlist(hdl, &zc, nvroot) != 0) return (-1); (void) strlcpy(zc.zc_name, zhp->zpool_name, sizeof (zc.zc_name)); if (zfs_ioctl(hdl, ZFS_IOC_VDEV_ADD, &zc) != 0) { switch (errno) { case EBUSY: /* * This can happen if the user has specified the same * device multiple times. We can't reliably detect this * until we try to add it and see we already have a * label. */ zfs_error_aux(hdl, dgettext(TEXT_DOMAIN, "one or more vdevs refer to the same device")); (void) zfs_error(hdl, EZFS_BADDEV, msg); break; case EOVERFLOW: /* * This occurrs when one of the devices is below * SPA_MINDEVSIZE. Unfortunately, we can't detect which * device was the problem device since there's no * reliable way to determine device size from userland. */ { char buf[64]; zfs_nicenum(SPA_MINDEVSIZE, buf, sizeof (buf)); zfs_error_aux(hdl, dgettext(TEXT_DOMAIN, "device is less than the minimum " "size (%s)"), buf); } (void) zfs_error(hdl, EZFS_BADDEV, msg); break; case ENOTSUP: zfs_error_aux(hdl, dgettext(TEXT_DOMAIN, "pool must be upgraded to add these vdevs")); (void) zfs_error(hdl, EZFS_BADVERSION, msg); break; case ENOTBLK: zfs_error_aux(hdl, dgettext(TEXT_DOMAIN, "cache device must be a disk or disk slice")); (void) zfs_error(hdl, EZFS_BADDEV, msg); break; default: (void) zpool_standard_error(hdl, errno, msg); } ret = -1; } else { ret = 0; } zcmd_free_nvlists(&zc); return (ret); } /* * Exports the pool from the system. The caller must ensure that there are no * mounted datasets in the pool. */ static int zpool_export_common(zpool_handle_t *zhp, boolean_t force, boolean_t hardforce, const char *log_str) { zfs_cmd_t zc = {"\0"}; char msg[1024]; (void) snprintf(msg, sizeof (msg), dgettext(TEXT_DOMAIN, "cannot export '%s'"), zhp->zpool_name); (void) strlcpy(zc.zc_name, zhp->zpool_name, sizeof (zc.zc_name)); zc.zc_cookie = force; zc.zc_guid = hardforce; zc.zc_history = (uint64_t)(uintptr_t)log_str; if (zfs_ioctl(zhp->zpool_hdl, ZFS_IOC_POOL_EXPORT, &zc) != 0) { switch (errno) { case EXDEV: zfs_error_aux(zhp->zpool_hdl, dgettext(TEXT_DOMAIN, "use '-f' to override the following errors:\n" "'%s' has an active shared spare which could be" " used by other pools once '%s' is exported."), zhp->zpool_name, zhp->zpool_name); return (zfs_error(zhp->zpool_hdl, EZFS_ACTIVE_SPARE, msg)); default: return (zpool_standard_error_fmt(zhp->zpool_hdl, errno, msg)); } } return (0); } int zpool_export(zpool_handle_t *zhp, boolean_t force, const char *log_str) { return (zpool_export_common(zhp, force, B_FALSE, log_str)); } int zpool_export_force(zpool_handle_t *zhp, const char *log_str) { return (zpool_export_common(zhp, B_TRUE, B_TRUE, log_str)); } static void zpool_rewind_exclaim(libzfs_handle_t *hdl, const char *name, boolean_t dryrun, nvlist_t *config) { nvlist_t *nv = NULL; uint64_t rewindto; int64_t loss = -1; struct tm t; char timestr[128]; if (!hdl->libzfs_printerr || config == NULL) return; if (nvlist_lookup_nvlist(config, ZPOOL_CONFIG_LOAD_INFO, &nv) != 0 || nvlist_lookup_nvlist(nv, ZPOOL_CONFIG_REWIND_INFO, &nv) != 0) { return; } if (nvlist_lookup_uint64(nv, ZPOOL_CONFIG_LOAD_TIME, &rewindto) != 0) return; (void) nvlist_lookup_int64(nv, ZPOOL_CONFIG_REWIND_TIME, &loss); if (localtime_r((time_t *)&rewindto, &t) != NULL && strftime(timestr, 128, "%c", &t) != 0) { if (dryrun) { (void) printf(dgettext(TEXT_DOMAIN, "Would be able to return %s " "to its state as of %s.\n"), name, timestr); } else { (void) printf(dgettext(TEXT_DOMAIN, "Pool %s returned to its state as of %s.\n"), name, timestr); } if (loss > 120) { (void) printf(dgettext(TEXT_DOMAIN, "%s approximately %lld "), dryrun ? "Would discard" : "Discarded", ((longlong_t)loss + 30) / 60); (void) printf(dgettext(TEXT_DOMAIN, "minutes of transactions.\n")); } else if (loss > 0) { (void) printf(dgettext(TEXT_DOMAIN, "%s approximately %lld "), dryrun ? "Would discard" : "Discarded", (longlong_t)loss); (void) printf(dgettext(TEXT_DOMAIN, "seconds of transactions.\n")); } } } void zpool_explain_recover(libzfs_handle_t *hdl, const char *name, int reason, nvlist_t *config) { nvlist_t *nv = NULL; int64_t loss = -1; uint64_t edata = UINT64_MAX; uint64_t rewindto; struct tm t; char timestr[128]; if (!hdl->libzfs_printerr) return; if (reason >= 0) (void) printf(dgettext(TEXT_DOMAIN, "action: ")); else (void) printf(dgettext(TEXT_DOMAIN, "\t")); /* All attempted rewinds failed if ZPOOL_CONFIG_LOAD_TIME missing */ if (nvlist_lookup_nvlist(config, ZPOOL_CONFIG_LOAD_INFO, &nv) != 0 || nvlist_lookup_nvlist(nv, ZPOOL_CONFIG_REWIND_INFO, &nv) != 0 || nvlist_lookup_uint64(nv, ZPOOL_CONFIG_LOAD_TIME, &rewindto) != 0) goto no_info; (void) nvlist_lookup_int64(nv, ZPOOL_CONFIG_REWIND_TIME, &loss); (void) nvlist_lookup_uint64(nv, ZPOOL_CONFIG_LOAD_DATA_ERRORS, &edata); (void) printf(dgettext(TEXT_DOMAIN, "Recovery is possible, but will result in some data loss.\n")); if (localtime_r((time_t *)&rewindto, &t) != NULL && strftime(timestr, 128, "%c", &t) != 0) { (void) printf(dgettext(TEXT_DOMAIN, "\tReturning the pool to its state as of %s\n" "\tshould correct the problem. "), timestr); } else { (void) printf(dgettext(TEXT_DOMAIN, "\tReverting the pool to an earlier state " "should correct the problem.\n\t")); } if (loss > 120) { (void) printf(dgettext(TEXT_DOMAIN, "Approximately %lld minutes of data\n" "\tmust be discarded, irreversibly. "), ((longlong_t)loss + 30) / 60); } else if (loss > 0) { (void) printf(dgettext(TEXT_DOMAIN, "Approximately %lld seconds of data\n" "\tmust be discarded, irreversibly. "), (longlong_t)loss); } if (edata != 0 && edata != UINT64_MAX) { if (edata == 1) { (void) printf(dgettext(TEXT_DOMAIN, "After rewind, at least\n" "\tone persistent user-data error will remain. ")); } else { (void) printf(dgettext(TEXT_DOMAIN, "After rewind, several\n" "\tpersistent user-data errors will remain. ")); } } (void) printf(dgettext(TEXT_DOMAIN, "Recovery can be attempted\n\tby executing 'zpool %s -F %s'. "), reason >= 0 ? "clear" : "import", name); (void) printf(dgettext(TEXT_DOMAIN, "A scrub of the pool\n" "\tis strongly recommended after recovery.\n")); return; no_info: (void) printf(dgettext(TEXT_DOMAIN, "Destroy and re-create the pool from\n\ta backup source.\n")); } /* * zpool_import() is a contracted interface. Should be kept the same * if possible. * * Applications should use zpool_import_props() to import a pool with * new properties value to be set. */ int zpool_import(libzfs_handle_t *hdl, nvlist_t *config, const char *newname, char *altroot) { nvlist_t *props = NULL; int ret; if (altroot != NULL) { if (nvlist_alloc(&props, NV_UNIQUE_NAME, 0) != 0) { return (zfs_error_fmt(hdl, EZFS_NOMEM, dgettext(TEXT_DOMAIN, "cannot import '%s'"), newname)); } if (nvlist_add_string(props, zpool_prop_to_name(ZPOOL_PROP_ALTROOT), altroot) != 0 || nvlist_add_string(props, zpool_prop_to_name(ZPOOL_PROP_CACHEFILE), "none") != 0) { nvlist_free(props); return (zfs_error_fmt(hdl, EZFS_NOMEM, dgettext(TEXT_DOMAIN, "cannot import '%s'"), newname)); } } ret = zpool_import_props(hdl, config, newname, props, ZFS_IMPORT_NORMAL); if (props) nvlist_free(props); return (ret); } static void print_vdev_tree(libzfs_handle_t *hdl, const char *name, nvlist_t *nv, int indent) { nvlist_t **child; uint_t c, children; char *vname; uint64_t is_log = 0; (void) nvlist_lookup_uint64(nv, ZPOOL_CONFIG_IS_LOG, &is_log); if (name != NULL) (void) printf("\t%*s%s%s\n", indent, "", name, is_log ? " [log]" : ""); if (nvlist_lookup_nvlist_array(nv, ZPOOL_CONFIG_CHILDREN, &child, &children) != 0) return; for (c = 0; c < children; c++) { vname = zpool_vdev_name(hdl, NULL, child[c], B_TRUE); print_vdev_tree(hdl, vname, child[c], indent + 2); free(vname); } } void zpool_print_unsup_feat(nvlist_t *config) { nvlist_t *nvinfo, *unsup_feat; nvpair_t *nvp; verify(nvlist_lookup_nvlist(config, ZPOOL_CONFIG_LOAD_INFO, &nvinfo) == 0); verify(nvlist_lookup_nvlist(nvinfo, ZPOOL_CONFIG_UNSUP_FEAT, &unsup_feat) == 0); for (nvp = nvlist_next_nvpair(unsup_feat, NULL); nvp != NULL; nvp = nvlist_next_nvpair(unsup_feat, nvp)) { char *desc; verify(nvpair_type(nvp) == DATA_TYPE_STRING); verify(nvpair_value_string(nvp, &desc) == 0); if (strlen(desc) > 0) (void) printf("\t%s (%s)\n", nvpair_name(nvp), desc); else (void) printf("\t%s\n", nvpair_name(nvp)); } } /* * Import the given pool using the known configuration and a list of * properties to be set. The configuration should have come from * zpool_find_import(). The 'newname' parameters control whether the pool * is imported with a different name. */ int zpool_import_props(libzfs_handle_t *hdl, nvlist_t *config, const char *newname, nvlist_t *props, int flags) { zfs_cmd_t zc = {"\0"}; zpool_rewind_policy_t policy; nvlist_t *nv = NULL; nvlist_t *nvinfo = NULL; nvlist_t *missing = NULL; char *thename; char *origname; int ret; int error = 0; char errbuf[1024]; verify(nvlist_lookup_string(config, ZPOOL_CONFIG_POOL_NAME, &origname) == 0); (void) snprintf(errbuf, sizeof (errbuf), dgettext(TEXT_DOMAIN, "cannot import pool '%s'"), origname); if (newname != NULL) { if (!zpool_name_valid(hdl, B_FALSE, newname)) return (zfs_error_fmt(hdl, EZFS_INVALIDNAME, dgettext(TEXT_DOMAIN, "cannot import '%s'"), newname)); thename = (char *)newname; } else { thename = origname; } if (props) { uint64_t version; prop_flags_t flags = { .create = B_FALSE, .import = B_TRUE }; verify(nvlist_lookup_uint64(config, ZPOOL_CONFIG_VERSION, &version) == 0); if ((props = zpool_valid_proplist(hdl, origname, props, version, flags, errbuf)) == NULL) { return (-1); } else if (zcmd_write_src_nvlist(hdl, &zc, props) != 0) { nvlist_free(props); return (-1); } } (void) strlcpy(zc.zc_name, thename, sizeof (zc.zc_name)); verify(nvlist_lookup_uint64(config, ZPOOL_CONFIG_POOL_GUID, &zc.zc_guid) == 0); if (zcmd_write_conf_nvlist(hdl, &zc, config) != 0) { nvlist_free(props); return (-1); } if (zcmd_alloc_dst_nvlist(hdl, &zc, zc.zc_nvlist_conf_size * 2) != 0) { nvlist_free(props); return (-1); } zc.zc_cookie = flags; while ((ret = zfs_ioctl(hdl, ZFS_IOC_POOL_IMPORT, &zc)) != 0 && errno == ENOMEM) { if (zcmd_expand_dst_nvlist(hdl, &zc) != 0) { zcmd_free_nvlists(&zc); return (-1); } } if (ret != 0) error = errno; (void) zcmd_read_dst_nvlist(hdl, &zc, &nv); zpool_get_rewind_policy(config, &policy); if (error) { char desc[1024]; /* * Dry-run failed, but we print out what success * looks like if we found a best txg */ if (policy.zrp_request & ZPOOL_TRY_REWIND) { zpool_rewind_exclaim(hdl, newname ? origname : thename, B_TRUE, nv); nvlist_free(nv); return (-1); } if (newname == NULL) (void) snprintf(desc, sizeof (desc), dgettext(TEXT_DOMAIN, "cannot import '%s'"), thename); else (void) snprintf(desc, sizeof (desc), dgettext(TEXT_DOMAIN, "cannot import '%s' as '%s'"), origname, thename); switch (error) { case ENOTSUP: if (nv != NULL && nvlist_lookup_nvlist(nv, ZPOOL_CONFIG_LOAD_INFO, &nvinfo) == 0 && nvlist_exists(nvinfo, ZPOOL_CONFIG_UNSUP_FEAT)) { (void) printf(dgettext(TEXT_DOMAIN, "This " "pool uses the following feature(s) not " "supported by this system:\n")); zpool_print_unsup_feat(nv); if (nvlist_exists(nvinfo, ZPOOL_CONFIG_CAN_RDONLY)) { (void) printf(dgettext(TEXT_DOMAIN, "All unsupported features are only " "required for writing to the pool." "\nThe pool can be imported using " "'-o readonly=on'.\n")); } } /* * Unsupported version. */ (void) zfs_error(hdl, EZFS_BADVERSION, desc); break; case EINVAL: (void) zfs_error(hdl, EZFS_INVALCONFIG, desc); break; case EROFS: zfs_error_aux(hdl, dgettext(TEXT_DOMAIN, "one or more devices is read only")); (void) zfs_error(hdl, EZFS_BADDEV, desc); break; case ENXIO: if (nv && nvlist_lookup_nvlist(nv, ZPOOL_CONFIG_LOAD_INFO, &nvinfo) == 0 && nvlist_lookup_nvlist(nvinfo, ZPOOL_CONFIG_MISSING_DEVICES, &missing) == 0) { (void) printf(dgettext(TEXT_DOMAIN, "The devices below are missing, use " "'-m' to import the pool anyway:\n")); print_vdev_tree(hdl, NULL, missing, 2); (void) printf("\n"); } (void) zpool_standard_error(hdl, error, desc); break; case EEXIST: (void) zpool_standard_error(hdl, error, desc); break; case EBUSY: zfs_error_aux(hdl, dgettext(TEXT_DOMAIN, "one or more devices are already in use\n")); (void) zfs_error(hdl, EZFS_BADDEV, desc); break; default: (void) zpool_standard_error(hdl, error, desc); zpool_explain_recover(hdl, newname ? origname : thename, -error, nv); break; } nvlist_free(nv); ret = -1; } else { zpool_handle_t *zhp; /* * This should never fail, but play it safe anyway. */ if (zpool_open_silent(hdl, thename, &zhp) != 0) ret = -1; else if (zhp != NULL) zpool_close(zhp); if (policy.zrp_request & (ZPOOL_DO_REWIND | ZPOOL_TRY_REWIND)) { zpool_rewind_exclaim(hdl, newname ? origname : thename, ((policy.zrp_request & ZPOOL_TRY_REWIND) != 0), nv); } nvlist_free(nv); return (0); } zcmd_free_nvlists(&zc); nvlist_free(props); return (ret); } /* * Scan the pool. */ int zpool_scan(zpool_handle_t *zhp, pool_scan_func_t func) { zfs_cmd_t zc = {"\0"}; char msg[1024]; libzfs_handle_t *hdl = zhp->zpool_hdl; (void) strlcpy(zc.zc_name, zhp->zpool_name, sizeof (zc.zc_name)); zc.zc_cookie = func; if (zfs_ioctl(hdl, ZFS_IOC_POOL_SCAN, &zc) == 0 || (errno == ENOENT && func != POOL_SCAN_NONE)) return (0); if (func == POOL_SCAN_SCRUB) { (void) snprintf(msg, sizeof (msg), dgettext(TEXT_DOMAIN, "cannot scrub %s"), zc.zc_name); } else if (func == POOL_SCAN_NONE) { (void) snprintf(msg, sizeof (msg), dgettext(TEXT_DOMAIN, "cannot cancel scrubbing %s"), zc.zc_name); } else { assert(!"unexpected result"); } if (errno == EBUSY) { nvlist_t *nvroot; pool_scan_stat_t *ps = NULL; uint_t psc; verify(nvlist_lookup_nvlist(zhp->zpool_config, ZPOOL_CONFIG_VDEV_TREE, &nvroot) == 0); (void) nvlist_lookup_uint64_array(nvroot, ZPOOL_CONFIG_SCAN_STATS, (uint64_t **)&ps, &psc); if (ps && ps->pss_func == POOL_SCAN_SCRUB) return (zfs_error(hdl, EZFS_SCRUBBING, msg)); else return (zfs_error(hdl, EZFS_RESILVERING, msg)); } else if (errno == ENOENT) { return (zfs_error(hdl, EZFS_NO_SCRUB, msg)); } else { return (zpool_standard_error(hdl, errno, msg)); } } /* * Find a vdev that matches the search criteria specified. We use the * the nvpair name to determine how we should look for the device. * 'avail_spare' is set to TRUE if the provided guid refers to an AVAIL * spare; but FALSE if its an INUSE spare. */ static nvlist_t * vdev_to_nvlist_iter(nvlist_t *nv, nvlist_t *search, boolean_t *avail_spare, boolean_t *l2cache, boolean_t *log) { uint_t c, children; nvlist_t **child; nvlist_t *ret; uint64_t is_log; char *srchkey; nvpair_t *pair = nvlist_next_nvpair(search, NULL); /* Nothing to look for */ if (search == NULL || pair == NULL) return (NULL); /* Obtain the key we will use to search */ srchkey = nvpair_name(pair); switch (nvpair_type(pair)) { case DATA_TYPE_UINT64: if (strcmp(srchkey, ZPOOL_CONFIG_GUID) == 0) { uint64_t srchval, theguid; verify(nvpair_value_uint64(pair, &srchval) == 0); verify(nvlist_lookup_uint64(nv, ZPOOL_CONFIG_GUID, &theguid) == 0); if (theguid == srchval) return (nv); } break; case DATA_TYPE_STRING: { char *srchval, *val; verify(nvpair_value_string(pair, &srchval) == 0); if (nvlist_lookup_string(nv, srchkey, &val) != 0) break; /* * Search for the requested value. Special cases: * * - ZPOOL_CONFIG_PATH for whole disk entries. These end in * "-part1", or "p1". The suffix is hidden from the user, * but included in the string, so this matches around it. * - ZPOOL_CONFIG_PATH for short names zfs_strcmp_shortname() * is used to check all possible expanded paths. * - looking for a top-level vdev name (i.e. ZPOOL_CONFIG_TYPE). * * Otherwise, all other searches are simple string compares. */ if (strcmp(srchkey, ZPOOL_CONFIG_PATH) == 0) { uint64_t wholedisk = 0; (void) nvlist_lookup_uint64(nv, ZPOOL_CONFIG_WHOLE_DISK, &wholedisk); if (zfs_strcmp_pathname(srchval, val, wholedisk) == 0) return (nv); } else if (strcmp(srchkey, ZPOOL_CONFIG_TYPE) == 0 && val) { char *type, *idx, *end, *p; uint64_t id, vdev_id; /* * Determine our vdev type, keeping in mind * that the srchval is composed of a type and * vdev id pair (i.e. mirror-4). */ if ((type = strdup(srchval)) == NULL) return (NULL); if ((p = strrchr(type, '-')) == NULL) { free(type); break; } idx = p + 1; *p = '\0'; /* * If the types don't match then keep looking. */ if (strncmp(val, type, strlen(val)) != 0) { free(type); break; } verify(strncmp(type, VDEV_TYPE_RAIDZ, strlen(VDEV_TYPE_RAIDZ)) == 0 || strncmp(type, VDEV_TYPE_MIRROR, strlen(VDEV_TYPE_MIRROR)) == 0); verify(nvlist_lookup_uint64(nv, ZPOOL_CONFIG_ID, &id) == 0); errno = 0; vdev_id = strtoull(idx, &end, 10); free(type); if (errno != 0) return (NULL); /* * Now verify that we have the correct vdev id. */ if (vdev_id == id) return (nv); } /* * Common case */ if (strcmp(srchval, val) == 0) return (nv); break; } default: break; } if (nvlist_lookup_nvlist_array(nv, ZPOOL_CONFIG_CHILDREN, &child, &children) != 0) return (NULL); for (c = 0; c < children; c++) { if ((ret = vdev_to_nvlist_iter(child[c], search, avail_spare, l2cache, NULL)) != NULL) { /* * The 'is_log' value is only set for the toplevel * vdev, not the leaf vdevs. So we always lookup the * log device from the root of the vdev tree (where * 'log' is non-NULL). */ if (log != NULL && nvlist_lookup_uint64(child[c], ZPOOL_CONFIG_IS_LOG, &is_log) == 0 && is_log) { *log = B_TRUE; } return (ret); } } if (nvlist_lookup_nvlist_array(nv, ZPOOL_CONFIG_SPARES, &child, &children) == 0) { for (c = 0; c < children; c++) { if ((ret = vdev_to_nvlist_iter(child[c], search, avail_spare, l2cache, NULL)) != NULL) { *avail_spare = B_TRUE; return (ret); } } } if (nvlist_lookup_nvlist_array(nv, ZPOOL_CONFIG_L2CACHE, &child, &children) == 0) { for (c = 0; c < children; c++) { if ((ret = vdev_to_nvlist_iter(child[c], search, avail_spare, l2cache, NULL)) != NULL) { *l2cache = B_TRUE; return (ret); } } } return (NULL); } /* * Given a physical path (minus the "/devices" prefix), find the * associated vdev. */ nvlist_t * zpool_find_vdev_by_physpath(zpool_handle_t *zhp, const char *ppath, boolean_t *avail_spare, boolean_t *l2cache, boolean_t *log) { nvlist_t *search, *nvroot, *ret; verify(nvlist_alloc(&search, NV_UNIQUE_NAME, KM_SLEEP) == 0); verify(nvlist_add_string(search, ZPOOL_CONFIG_PHYS_PATH, ppath) == 0); verify(nvlist_lookup_nvlist(zhp->zpool_config, ZPOOL_CONFIG_VDEV_TREE, &nvroot) == 0); *avail_spare = B_FALSE; *l2cache = B_FALSE; if (log != NULL) *log = B_FALSE; ret = vdev_to_nvlist_iter(nvroot, search, avail_spare, l2cache, log); nvlist_free(search); return (ret); } /* * Determine if we have an "interior" top-level vdev (i.e mirror/raidz). */ boolean_t zpool_vdev_is_interior(const char *name) { if (strncmp(name, VDEV_TYPE_RAIDZ, strlen(VDEV_TYPE_RAIDZ)) == 0 || strncmp(name, VDEV_TYPE_MIRROR, strlen(VDEV_TYPE_MIRROR)) == 0) return (B_TRUE); return (B_FALSE); } nvlist_t * zpool_find_vdev(zpool_handle_t *zhp, const char *path, boolean_t *avail_spare, boolean_t *l2cache, boolean_t *log) { char *end; nvlist_t *nvroot, *search, *ret; uint64_t guid; verify(nvlist_alloc(&search, NV_UNIQUE_NAME, KM_SLEEP) == 0); guid = strtoull(path, &end, 0); if (guid != 0 && *end == '\0') { verify(nvlist_add_uint64(search, ZPOOL_CONFIG_GUID, guid) == 0); } else if (zpool_vdev_is_interior(path)) { verify(nvlist_add_string(search, ZPOOL_CONFIG_TYPE, path) == 0); } else { verify(nvlist_add_string(search, ZPOOL_CONFIG_PATH, path) == 0); } verify(nvlist_lookup_nvlist(zhp->zpool_config, ZPOOL_CONFIG_VDEV_TREE, &nvroot) == 0); *avail_spare = B_FALSE; *l2cache = B_FALSE; if (log != NULL) *log = B_FALSE; ret = vdev_to_nvlist_iter(nvroot, search, avail_spare, l2cache, log); nvlist_free(search); return (ret); } static int vdev_online(nvlist_t *nv) { uint64_t ival; if (nvlist_lookup_uint64(nv, ZPOOL_CONFIG_OFFLINE, &ival) == 0 || nvlist_lookup_uint64(nv, ZPOOL_CONFIG_FAULTED, &ival) == 0 || nvlist_lookup_uint64(nv, ZPOOL_CONFIG_REMOVED, &ival) == 0) return (0); return (1); } /* * Helper function for zpool_get_physpaths(). */ static int vdev_get_one_physpath(nvlist_t *config, char *physpath, size_t physpath_size, size_t *bytes_written) { size_t bytes_left, pos, rsz; char *tmppath; const char *format; if (nvlist_lookup_string(config, ZPOOL_CONFIG_PHYS_PATH, &tmppath) != 0) return (EZFS_NODEVICE); pos = *bytes_written; bytes_left = physpath_size - pos; format = (pos == 0) ? "%s" : " %s"; rsz = snprintf(physpath + pos, bytes_left, format, tmppath); *bytes_written += rsz; if (rsz >= bytes_left) { /* if physpath was not copied properly, clear it */ if (bytes_left != 0) { physpath[pos] = 0; } return (EZFS_NOSPC); } return (0); } static int vdev_get_physpaths(nvlist_t *nv, char *physpath, size_t phypath_size, size_t *rsz, boolean_t is_spare) { char *type; int ret; if (nvlist_lookup_string(nv, ZPOOL_CONFIG_TYPE, &type) != 0) return (EZFS_INVALCONFIG); if (strcmp(type, VDEV_TYPE_DISK) == 0) { /* * An active spare device has ZPOOL_CONFIG_IS_SPARE set. * For a spare vdev, we only want to boot from the active * spare device. */ if (is_spare) { uint64_t spare = 0; (void) nvlist_lookup_uint64(nv, ZPOOL_CONFIG_IS_SPARE, &spare); if (!spare) return (EZFS_INVALCONFIG); } if (vdev_online(nv)) { if ((ret = vdev_get_one_physpath(nv, physpath, phypath_size, rsz)) != 0) return (ret); } } else if (strcmp(type, VDEV_TYPE_MIRROR) == 0 || strcmp(type, VDEV_TYPE_REPLACING) == 0 || (is_spare = (strcmp(type, VDEV_TYPE_SPARE) == 0))) { nvlist_t **child; uint_t count; int i, ret; if (nvlist_lookup_nvlist_array(nv, ZPOOL_CONFIG_CHILDREN, &child, &count) != 0) return (EZFS_INVALCONFIG); for (i = 0; i < count; i++) { ret = vdev_get_physpaths(child[i], physpath, phypath_size, rsz, is_spare); if (ret == EZFS_NOSPC) return (ret); } } return (EZFS_POOL_INVALARG); } /* * Get phys_path for a root pool config. * Return 0 on success; non-zero on failure. */ static int zpool_get_config_physpath(nvlist_t *config, char *physpath, size_t phypath_size) { size_t rsz; nvlist_t *vdev_root; nvlist_t **child; uint_t count; char *type; rsz = 0; if (nvlist_lookup_nvlist(config, ZPOOL_CONFIG_VDEV_TREE, &vdev_root) != 0) return (EZFS_INVALCONFIG); if (nvlist_lookup_string(vdev_root, ZPOOL_CONFIG_TYPE, &type) != 0 || nvlist_lookup_nvlist_array(vdev_root, ZPOOL_CONFIG_CHILDREN, &child, &count) != 0) return (EZFS_INVALCONFIG); #if defined(__sun__) || defined(__sun) /* * root pool can not have EFI labeled disks and can only have * a single top-level vdev. */ if (strcmp(type, VDEV_TYPE_ROOT) != 0 || count != 1 || pool_uses_efi(vdev_root)) return (EZFS_POOL_INVALARG); #endif (void) vdev_get_physpaths(child[0], physpath, phypath_size, &rsz, B_FALSE); /* No online devices */ if (rsz == 0) return (EZFS_NODEVICE); return (0); } /* * Get phys_path for a root pool * Return 0 on success; non-zero on failure. */ int zpool_get_physpath(zpool_handle_t *zhp, char *physpath, size_t phypath_size) { return (zpool_get_config_physpath(zhp->zpool_config, physpath, phypath_size)); } /* * If the device has being dynamically expanded then we need to relabel * the disk to use the new unallocated space. */ static int zpool_relabel_disk(libzfs_handle_t *hdl, const char *path, const char *msg) { int fd, error; if ((fd = open(path, O_RDWR|O_DIRECT)) < 0) { zfs_error_aux(hdl, dgettext(TEXT_DOMAIN, "cannot " "relabel '%s': unable to open device: %d"), path, errno); return (zfs_error(hdl, EZFS_OPENFAILED, msg)); } /* * It's possible that we might encounter an error if the device * does not have any unallocated space left. If so, we simply * ignore that error and continue on. * * Also, we don't call efi_rescan() - that would just return EBUSY. * The module will do it for us in vdev_disk_open(). */ error = efi_use_whole_disk(fd); (void) close(fd); if (error && error != VT_ENOSPC) { zfs_error_aux(hdl, dgettext(TEXT_DOMAIN, "cannot " "relabel '%s': unable to read disk capacity"), path); return (zfs_error(hdl, EZFS_NOCAP, msg)); } return (0); } /* * Bring the specified vdev online. The 'flags' parameter is a set of the * ZFS_ONLINE_* flags. */ int zpool_vdev_online(zpool_handle_t *zhp, const char *path, int flags, vdev_state_t *newstate) { zfs_cmd_t zc = {"\0"}; char msg[1024]; nvlist_t *tgt; boolean_t avail_spare, l2cache, islog; libzfs_handle_t *hdl = zhp->zpool_hdl; int error; if (flags & ZFS_ONLINE_EXPAND) { (void) snprintf(msg, sizeof (msg), dgettext(TEXT_DOMAIN, "cannot expand %s"), path); } else { (void) snprintf(msg, sizeof (msg), dgettext(TEXT_DOMAIN, "cannot online %s"), path); } (void) strlcpy(zc.zc_name, zhp->zpool_name, sizeof (zc.zc_name)); if ((tgt = zpool_find_vdev(zhp, path, &avail_spare, &l2cache, &islog)) == NULL) return (zfs_error(hdl, EZFS_NODEVICE, msg)); verify(nvlist_lookup_uint64(tgt, ZPOOL_CONFIG_GUID, &zc.zc_guid) == 0); if (avail_spare) return (zfs_error(hdl, EZFS_ISSPARE, msg)); if (flags & ZFS_ONLINE_EXPAND || zpool_get_prop_int(zhp, ZPOOL_PROP_AUTOEXPAND, NULL)) { uint64_t wholedisk = 0; (void) nvlist_lookup_uint64(tgt, ZPOOL_CONFIG_WHOLE_DISK, &wholedisk); /* * XXX - L2ARC 1.0 devices can't support expansion. */ if (l2cache) { zfs_error_aux(hdl, dgettext(TEXT_DOMAIN, "cannot expand cache devices")); return (zfs_error(hdl, EZFS_VDEVNOTSUP, msg)); } if (wholedisk) { const char *fullpath = path; char buf[MAXPATHLEN]; if (path[0] != '/') { error = zfs_resolve_shortname(path, buf, sizeof (buf)); if (error != 0) return (zfs_error(hdl, EZFS_NODEVICE, msg)); fullpath = buf; } error = zpool_relabel_disk(hdl, fullpath, msg); if (error != 0) return (error); } } zc.zc_cookie = VDEV_STATE_ONLINE; zc.zc_obj = flags; if (zfs_ioctl(hdl, ZFS_IOC_VDEV_SET_STATE, &zc) != 0) { if (errno == EINVAL) { zfs_error_aux(hdl, dgettext(TEXT_DOMAIN, "was split " "from this pool into a new one. Use '%s' " "instead"), "zpool detach"); return (zfs_error(hdl, EZFS_POSTSPLIT_ONLINE, msg)); } return (zpool_standard_error(hdl, errno, msg)); } *newstate = zc.zc_cookie; return (0); } /* * Take the specified vdev offline */ int zpool_vdev_offline(zpool_handle_t *zhp, const char *path, boolean_t istmp) { zfs_cmd_t zc = {"\0"}; char msg[1024]; nvlist_t *tgt; boolean_t avail_spare, l2cache; libzfs_handle_t *hdl = zhp->zpool_hdl; (void) snprintf(msg, sizeof (msg), dgettext(TEXT_DOMAIN, "cannot offline %s"), path); (void) strlcpy(zc.zc_name, zhp->zpool_name, sizeof (zc.zc_name)); if ((tgt = zpool_find_vdev(zhp, path, &avail_spare, &l2cache, NULL)) == NULL) return (zfs_error(hdl, EZFS_NODEVICE, msg)); verify(nvlist_lookup_uint64(tgt, ZPOOL_CONFIG_GUID, &zc.zc_guid) == 0); if (avail_spare) return (zfs_error(hdl, EZFS_ISSPARE, msg)); zc.zc_cookie = VDEV_STATE_OFFLINE; zc.zc_obj = istmp ? ZFS_OFFLINE_TEMPORARY : 0; if (zfs_ioctl(hdl, ZFS_IOC_VDEV_SET_STATE, &zc) == 0) return (0); switch (errno) { case EBUSY: /* * There are no other replicas of this device. */ return (zfs_error(hdl, EZFS_NOREPLICAS, msg)); case EEXIST: /* * The log device has unplayed logs */ return (zfs_error(hdl, EZFS_UNPLAYED_LOGS, msg)); default: return (zpool_standard_error(hdl, errno, msg)); } } /* * Mark the given vdev faulted. */ int zpool_vdev_fault(zpool_handle_t *zhp, uint64_t guid, vdev_aux_t aux) { zfs_cmd_t zc = {"\0"}; char msg[1024]; libzfs_handle_t *hdl = zhp->zpool_hdl; (void) snprintf(msg, sizeof (msg), dgettext(TEXT_DOMAIN, "cannot fault %llu"), (u_longlong_t)guid); (void) strlcpy(zc.zc_name, zhp->zpool_name, sizeof (zc.zc_name)); zc.zc_guid = guid; zc.zc_cookie = VDEV_STATE_FAULTED; zc.zc_obj = aux; if (ioctl(hdl->libzfs_fd, ZFS_IOC_VDEV_SET_STATE, &zc) == 0) return (0); switch (errno) { case EBUSY: /* * There are no other replicas of this device. */ return (zfs_error(hdl, EZFS_NOREPLICAS, msg)); default: return (zpool_standard_error(hdl, errno, msg)); } } /* * Mark the given vdev degraded. */ int zpool_vdev_degrade(zpool_handle_t *zhp, uint64_t guid, vdev_aux_t aux) { zfs_cmd_t zc = {"\0"}; char msg[1024]; libzfs_handle_t *hdl = zhp->zpool_hdl; (void) snprintf(msg, sizeof (msg), dgettext(TEXT_DOMAIN, "cannot degrade %llu"), (u_longlong_t)guid); (void) strlcpy(zc.zc_name, zhp->zpool_name, sizeof (zc.zc_name)); zc.zc_guid = guid; zc.zc_cookie = VDEV_STATE_DEGRADED; zc.zc_obj = aux; if (ioctl(hdl->libzfs_fd, ZFS_IOC_VDEV_SET_STATE, &zc) == 0) return (0); return (zpool_standard_error(hdl, errno, msg)); } /* * Returns TRUE if the given nvlist is a vdev that was originally swapped in as * a hot spare. */ static boolean_t is_replacing_spare(nvlist_t *search, nvlist_t *tgt, int which) { nvlist_t **child; uint_t c, children; char *type; if (nvlist_lookup_nvlist_array(search, ZPOOL_CONFIG_CHILDREN, &child, &children) == 0) { verify(nvlist_lookup_string(search, ZPOOL_CONFIG_TYPE, &type) == 0); if (strcmp(type, VDEV_TYPE_SPARE) == 0 && children == 2 && child[which] == tgt) return (B_TRUE); for (c = 0; c < children; c++) if (is_replacing_spare(child[c], tgt, which)) return (B_TRUE); } return (B_FALSE); } /* * Attach new_disk (fully described by nvroot) to old_disk. * If 'replacing' is specified, the new disk will replace the old one. */ int zpool_vdev_attach(zpool_handle_t *zhp, const char *old_disk, const char *new_disk, nvlist_t *nvroot, int replacing) { zfs_cmd_t zc = {"\0"}; char msg[1024]; int ret; nvlist_t *tgt; boolean_t avail_spare, l2cache, islog; uint64_t val; char *newname; nvlist_t **child; uint_t children; nvlist_t *config_root; libzfs_handle_t *hdl = zhp->zpool_hdl; boolean_t rootpool = zpool_is_bootable(zhp); if (replacing) (void) snprintf(msg, sizeof (msg), dgettext(TEXT_DOMAIN, "cannot replace %s with %s"), old_disk, new_disk); else (void) snprintf(msg, sizeof (msg), dgettext(TEXT_DOMAIN, "cannot attach %s to %s"), new_disk, old_disk); #if defined(__sun__) || defined(__sun) /* * If this is a root pool, make sure that we're not attaching an * EFI labeled device. */ if (rootpool && pool_uses_efi(nvroot)) { zfs_error_aux(hdl, dgettext(TEXT_DOMAIN, "EFI labeled devices are not supported on root pools.")); return (zfs_error(hdl, EZFS_POOL_NOTSUP, msg)); } #endif (void) strlcpy(zc.zc_name, zhp->zpool_name, sizeof (zc.zc_name)); if ((tgt = zpool_find_vdev(zhp, old_disk, &avail_spare, &l2cache, &islog)) == 0) return (zfs_error(hdl, EZFS_NODEVICE, msg)); if (avail_spare) return (zfs_error(hdl, EZFS_ISSPARE, msg)); if (l2cache) return (zfs_error(hdl, EZFS_ISL2CACHE, msg)); verify(nvlist_lookup_uint64(tgt, ZPOOL_CONFIG_GUID, &zc.zc_guid) == 0); zc.zc_cookie = replacing; if (nvlist_lookup_nvlist_array(nvroot, ZPOOL_CONFIG_CHILDREN, &child, &children) != 0 || children != 1) { zfs_error_aux(hdl, dgettext(TEXT_DOMAIN, "new device must be a single disk")); return (zfs_error(hdl, EZFS_INVALCONFIG, msg)); } verify(nvlist_lookup_nvlist(zpool_get_config(zhp, NULL), ZPOOL_CONFIG_VDEV_TREE, &config_root) == 0); if ((newname = zpool_vdev_name(NULL, NULL, child[0], B_FALSE)) == NULL) return (-1); /* * If the target is a hot spare that has been swapped in, we can only * replace it with another hot spare. */ if (replacing && nvlist_lookup_uint64(tgt, ZPOOL_CONFIG_IS_SPARE, &val) == 0 && (zpool_find_vdev(zhp, newname, &avail_spare, &l2cache, NULL) == NULL || !avail_spare) && is_replacing_spare(config_root, tgt, 1)) { zfs_error_aux(hdl, dgettext(TEXT_DOMAIN, "can only be replaced by another hot spare")); free(newname); return (zfs_error(hdl, EZFS_BADTARGET, msg)); } free(newname); if (zcmd_write_conf_nvlist(hdl, &zc, nvroot) != 0) return (-1); ret = zfs_ioctl(hdl, ZFS_IOC_VDEV_ATTACH, &zc); zcmd_free_nvlists(&zc); if (ret == 0) { if (rootpool) { /* * XXX need a better way to prevent user from * booting up a half-baked vdev. */ (void) fprintf(stderr, dgettext(TEXT_DOMAIN, "Make " "sure to wait until resilver is done " "before rebooting.\n")); } return (0); } switch (errno) { case ENOTSUP: /* * Can't attach to or replace this type of vdev. */ if (replacing) { uint64_t version = zpool_get_prop_int(zhp, ZPOOL_PROP_VERSION, NULL); if (islog) zfs_error_aux(hdl, dgettext(TEXT_DOMAIN, "cannot replace a log with a spare")); else if (version >= SPA_VERSION_MULTI_REPLACE) zfs_error_aux(hdl, dgettext(TEXT_DOMAIN, "already in replacing/spare config; wait " "for completion or use 'zpool detach'")); else zfs_error_aux(hdl, dgettext(TEXT_DOMAIN, "cannot replace a replacing device")); } else { zfs_error_aux(hdl, dgettext(TEXT_DOMAIN, "can only attach to mirrors and top-level " "disks")); } (void) zfs_error(hdl, EZFS_BADTARGET, msg); break; case EINVAL: /* * The new device must be a single disk. */ zfs_error_aux(hdl, dgettext(TEXT_DOMAIN, "new device must be a single disk")); (void) zfs_error(hdl, EZFS_INVALCONFIG, msg); break; case EBUSY: zfs_error_aux(hdl, dgettext(TEXT_DOMAIN, "%s is busy"), new_disk); (void) zfs_error(hdl, EZFS_BADDEV, msg); break; case EOVERFLOW: /* * The new device is too small. */ zfs_error_aux(hdl, dgettext(TEXT_DOMAIN, "device is too small")); (void) zfs_error(hdl, EZFS_BADDEV, msg); break; case EDOM: /* * The new device has a different optimal sector size. */ zfs_error_aux(hdl, dgettext(TEXT_DOMAIN, "new device has a different optimal sector size; use the " "option '-o ashift=N' to override the optimal size")); (void) zfs_error(hdl, EZFS_BADDEV, msg); break; case ENAMETOOLONG: /* * The resulting top-level vdev spec won't fit in the label. */ (void) zfs_error(hdl, EZFS_DEVOVERFLOW, msg); break; default: (void) zpool_standard_error(hdl, errno, msg); } return (-1); } /* * Detach the specified device. */ int zpool_vdev_detach(zpool_handle_t *zhp, const char *path) { zfs_cmd_t zc = {"\0"}; char msg[1024]; nvlist_t *tgt; boolean_t avail_spare, l2cache; libzfs_handle_t *hdl = zhp->zpool_hdl; (void) snprintf(msg, sizeof (msg), dgettext(TEXT_DOMAIN, "cannot detach %s"), path); (void) strlcpy(zc.zc_name, zhp->zpool_name, sizeof (zc.zc_name)); if ((tgt = zpool_find_vdev(zhp, path, &avail_spare, &l2cache, NULL)) == 0) return (zfs_error(hdl, EZFS_NODEVICE, msg)); if (avail_spare) return (zfs_error(hdl, EZFS_ISSPARE, msg)); if (l2cache) return (zfs_error(hdl, EZFS_ISL2CACHE, msg)); verify(nvlist_lookup_uint64(tgt, ZPOOL_CONFIG_GUID, &zc.zc_guid) == 0); if (zfs_ioctl(hdl, ZFS_IOC_VDEV_DETACH, &zc) == 0) return (0); switch (errno) { case ENOTSUP: /* * Can't detach from this type of vdev. */ zfs_error_aux(hdl, dgettext(TEXT_DOMAIN, "only " "applicable to mirror and replacing vdevs")); (void) zfs_error(hdl, EZFS_BADTARGET, msg); break; case EBUSY: /* * There are no other replicas of this device. */ (void) zfs_error(hdl, EZFS_NOREPLICAS, msg); break; default: (void) zpool_standard_error(hdl, errno, msg); } return (-1); } /* * Find a mirror vdev in the source nvlist. * * The mchild array contains a list of disks in one of the top-level mirrors * of the source pool. The schild array contains a list of disks that the * user specified on the command line. We loop over the mchild array to * see if any entry in the schild array matches. * * If a disk in the mchild array is found in the schild array, we return * the index of that entry. Otherwise we return -1. */ static int find_vdev_entry(zpool_handle_t *zhp, nvlist_t **mchild, uint_t mchildren, nvlist_t **schild, uint_t schildren) { uint_t mc; for (mc = 0; mc < mchildren; mc++) { uint_t sc; char *mpath = zpool_vdev_name(zhp->zpool_hdl, zhp, mchild[mc], B_FALSE); for (sc = 0; sc < schildren; sc++) { char *spath = zpool_vdev_name(zhp->zpool_hdl, zhp, schild[sc], B_FALSE); boolean_t result = (strcmp(mpath, spath) == 0); free(spath); if (result) { free(mpath); return (mc); } } free(mpath); } return (-1); } /* * Split a mirror pool. If newroot points to null, then a new nvlist * is generated and it is the responsibility of the caller to free it. */ int zpool_vdev_split(zpool_handle_t *zhp, char *newname, nvlist_t **newroot, nvlist_t *props, splitflags_t flags) { zfs_cmd_t zc = {"\0"}; char msg[1024]; nvlist_t *tree, *config, **child, **newchild, *newconfig = NULL; nvlist_t **varray = NULL, *zc_props = NULL; uint_t c, children, newchildren, lastlog = 0, vcount, found = 0; libzfs_handle_t *hdl = zhp->zpool_hdl; uint64_t vers; boolean_t freelist = B_FALSE, memory_err = B_TRUE; int retval = 0; (void) snprintf(msg, sizeof (msg), dgettext(TEXT_DOMAIN, "Unable to split %s"), zhp->zpool_name); if (!zpool_name_valid(hdl, B_FALSE, newname)) return (zfs_error(hdl, EZFS_INVALIDNAME, msg)); if ((config = zpool_get_config(zhp, NULL)) == NULL) { (void) fprintf(stderr, gettext("Internal error: unable to " "retrieve pool configuration\n")); return (-1); } verify(nvlist_lookup_nvlist(config, ZPOOL_CONFIG_VDEV_TREE, &tree) == 0); verify(nvlist_lookup_uint64(config, ZPOOL_CONFIG_VERSION, &vers) == 0); if (props) { prop_flags_t flags = { .create = B_FALSE, .import = B_TRUE }; if ((zc_props = zpool_valid_proplist(hdl, zhp->zpool_name, props, vers, flags, msg)) == NULL) return (-1); } if (nvlist_lookup_nvlist_array(tree, ZPOOL_CONFIG_CHILDREN, &child, &children) != 0) { zfs_error_aux(hdl, dgettext(TEXT_DOMAIN, "Source pool is missing vdev tree")); if (zc_props) nvlist_free(zc_props); return (-1); } varray = zfs_alloc(hdl, children * sizeof (nvlist_t *)); vcount = 0; if (*newroot == NULL || nvlist_lookup_nvlist_array(*newroot, ZPOOL_CONFIG_CHILDREN, &newchild, &newchildren) != 0) newchildren = 0; for (c = 0; c < children; c++) { uint64_t is_log = B_FALSE, is_hole = B_FALSE; char *type; nvlist_t **mchild, *vdev; uint_t mchildren; int entry; /* * Unlike cache & spares, slogs are stored in the * ZPOOL_CONFIG_CHILDREN array. We filter them out here. */ (void) nvlist_lookup_uint64(child[c], ZPOOL_CONFIG_IS_LOG, &is_log); (void) nvlist_lookup_uint64(child[c], ZPOOL_CONFIG_IS_HOLE, &is_hole); if (is_log || is_hole) { /* * Create a hole vdev and put it in the config. */ if (nvlist_alloc(&vdev, NV_UNIQUE_NAME, 0) != 0) goto out; if (nvlist_add_string(vdev, ZPOOL_CONFIG_TYPE, VDEV_TYPE_HOLE) != 0) goto out; if (nvlist_add_uint64(vdev, ZPOOL_CONFIG_IS_HOLE, 1) != 0) goto out; if (lastlog == 0) lastlog = vcount; varray[vcount++] = vdev; continue; } lastlog = 0; verify(nvlist_lookup_string(child[c], ZPOOL_CONFIG_TYPE, &type) == 0); if (strcmp(type, VDEV_TYPE_MIRROR) != 0) { zfs_error_aux(hdl, dgettext(TEXT_DOMAIN, "Source pool must be composed only of mirrors\n")); retval = zfs_error(hdl, EZFS_INVALCONFIG, msg); goto out; } verify(nvlist_lookup_nvlist_array(child[c], ZPOOL_CONFIG_CHILDREN, &mchild, &mchildren) == 0); /* find or add an entry for this top-level vdev */ if (newchildren > 0 && (entry = find_vdev_entry(zhp, mchild, mchildren, newchild, newchildren)) >= 0) { /* We found a disk that the user specified. */ vdev = mchild[entry]; ++found; } else { /* User didn't specify a disk for this vdev. */ vdev = mchild[mchildren - 1]; } if (nvlist_dup(vdev, &varray[vcount++], 0) != 0) goto out; } /* did we find every disk the user specified? */ if (found != newchildren) { zfs_error_aux(hdl, dgettext(TEXT_DOMAIN, "Device list must " "include at most one disk from each mirror")); retval = zfs_error(hdl, EZFS_INVALCONFIG, msg); goto out; } /* Prepare the nvlist for populating. */ if (*newroot == NULL) { if (nvlist_alloc(newroot, NV_UNIQUE_NAME, 0) != 0) goto out; freelist = B_TRUE; if (nvlist_add_string(*newroot, ZPOOL_CONFIG_TYPE, VDEV_TYPE_ROOT) != 0) goto out; } else { verify(nvlist_remove_all(*newroot, ZPOOL_CONFIG_CHILDREN) == 0); } /* Add all the children we found */ if (nvlist_add_nvlist_array(*newroot, ZPOOL_CONFIG_CHILDREN, varray, lastlog == 0 ? vcount : lastlog) != 0) goto out; /* * If we're just doing a dry run, exit now with success. */ if (flags.dryrun) { memory_err = B_FALSE; freelist = B_FALSE; goto out; } /* now build up the config list & call the ioctl */ if (nvlist_alloc(&newconfig, NV_UNIQUE_NAME, 0) != 0) goto out; if (nvlist_add_nvlist(newconfig, ZPOOL_CONFIG_VDEV_TREE, *newroot) != 0 || nvlist_add_string(newconfig, ZPOOL_CONFIG_POOL_NAME, newname) != 0 || nvlist_add_uint64(newconfig, ZPOOL_CONFIG_VERSION, vers) != 0) goto out; /* * The new pool is automatically part of the namespace unless we * explicitly export it. */ if (!flags.import) zc.zc_cookie = ZPOOL_EXPORT_AFTER_SPLIT; (void) strlcpy(zc.zc_name, zhp->zpool_name, sizeof (zc.zc_name)); (void) strlcpy(zc.zc_string, newname, sizeof (zc.zc_string)); if (zcmd_write_conf_nvlist(hdl, &zc, newconfig) != 0) goto out; if (zc_props != NULL && zcmd_write_src_nvlist(hdl, &zc, zc_props) != 0) goto out; if (zfs_ioctl(hdl, ZFS_IOC_VDEV_SPLIT, &zc) != 0) { retval = zpool_standard_error(hdl, errno, msg); goto out; } freelist = B_FALSE; memory_err = B_FALSE; out: if (varray != NULL) { int v; for (v = 0; v < vcount; v++) nvlist_free(varray[v]); free(varray); } zcmd_free_nvlists(&zc); if (zc_props) nvlist_free(zc_props); if (newconfig) nvlist_free(newconfig); if (freelist) { nvlist_free(*newroot); *newroot = NULL; } if (retval != 0) return (retval); if (memory_err) return (no_memory(hdl)); return (0); } /* * Remove the given device. Currently, this is supported only for hot spares * and level 2 cache devices. */ int zpool_vdev_remove(zpool_handle_t *zhp, const char *path) { zfs_cmd_t zc = {"\0"}; char msg[1024]; nvlist_t *tgt; boolean_t avail_spare, l2cache, islog; libzfs_handle_t *hdl = zhp->zpool_hdl; uint64_t version; (void) snprintf(msg, sizeof (msg), dgettext(TEXT_DOMAIN, "cannot remove %s"), path); (void) strlcpy(zc.zc_name, zhp->zpool_name, sizeof (zc.zc_name)); if ((tgt = zpool_find_vdev(zhp, path, &avail_spare, &l2cache, &islog)) == 0) return (zfs_error(hdl, EZFS_NODEVICE, msg)); /* * XXX - this should just go away. */ if (!avail_spare && !l2cache && !islog) { zfs_error_aux(hdl, dgettext(TEXT_DOMAIN, "only inactive hot spares, cache, top-level, " "or log devices can be removed")); return (zfs_error(hdl, EZFS_NODEVICE, msg)); } version = zpool_get_prop_int(zhp, ZPOOL_PROP_VERSION, NULL); if (islog && version < SPA_VERSION_HOLES) { zfs_error_aux(hdl, dgettext(TEXT_DOMAIN, "pool must be upgrade to support log removal")); return (zfs_error(hdl, EZFS_BADVERSION, msg)); } verify(nvlist_lookup_uint64(tgt, ZPOOL_CONFIG_GUID, &zc.zc_guid) == 0); if (zfs_ioctl(hdl, ZFS_IOC_VDEV_REMOVE, &zc) == 0) return (0); return (zpool_standard_error(hdl, errno, msg)); } /* * Clear the errors for the pool, or the particular device if specified. */ int zpool_clear(zpool_handle_t *zhp, const char *path, nvlist_t *rewindnvl) { zfs_cmd_t zc = {"\0"}; char msg[1024]; nvlist_t *tgt; zpool_rewind_policy_t policy; boolean_t avail_spare, l2cache; libzfs_handle_t *hdl = zhp->zpool_hdl; nvlist_t *nvi = NULL; int error; if (path) (void) snprintf(msg, sizeof (msg), dgettext(TEXT_DOMAIN, "cannot clear errors for %s"), path); else (void) snprintf(msg, sizeof (msg), dgettext(TEXT_DOMAIN, "cannot clear errors for %s"), zhp->zpool_name); (void) strlcpy(zc.zc_name, zhp->zpool_name, sizeof (zc.zc_name)); if (path) { if ((tgt = zpool_find_vdev(zhp, path, &avail_spare, &l2cache, NULL)) == 0) return (zfs_error(hdl, EZFS_NODEVICE, msg)); /* * Don't allow error clearing for hot spares. Do allow * error clearing for l2cache devices. */ if (avail_spare) return (zfs_error(hdl, EZFS_ISSPARE, msg)); verify(nvlist_lookup_uint64(tgt, ZPOOL_CONFIG_GUID, &zc.zc_guid) == 0); } zpool_get_rewind_policy(rewindnvl, &policy); zc.zc_cookie = policy.zrp_request; if (zcmd_alloc_dst_nvlist(hdl, &zc, zhp->zpool_config_size * 2) != 0) return (-1); if (zcmd_write_src_nvlist(hdl, &zc, rewindnvl) != 0) return (-1); while ((error = zfs_ioctl(hdl, ZFS_IOC_CLEAR, &zc)) != 0 && errno == ENOMEM) { if (zcmd_expand_dst_nvlist(hdl, &zc) != 0) { zcmd_free_nvlists(&zc); return (-1); } } if (!error || ((policy.zrp_request & ZPOOL_TRY_REWIND) && errno != EPERM && errno != EACCES)) { if (policy.zrp_request & (ZPOOL_DO_REWIND | ZPOOL_TRY_REWIND)) { (void) zcmd_read_dst_nvlist(hdl, &zc, &nvi); zpool_rewind_exclaim(hdl, zc.zc_name, ((policy.zrp_request & ZPOOL_TRY_REWIND) != 0), nvi); nvlist_free(nvi); } zcmd_free_nvlists(&zc); return (0); } zcmd_free_nvlists(&zc); return (zpool_standard_error(hdl, errno, msg)); } /* * Similar to zpool_clear(), but takes a GUID (used by fmd). */ int zpool_vdev_clear(zpool_handle_t *zhp, uint64_t guid) { zfs_cmd_t zc = {"\0"}; char msg[1024]; libzfs_handle_t *hdl = zhp->zpool_hdl; (void) snprintf(msg, sizeof (msg), dgettext(TEXT_DOMAIN, "cannot clear errors for %llx"), (u_longlong_t)guid); (void) strlcpy(zc.zc_name, zhp->zpool_name, sizeof (zc.zc_name)); zc.zc_guid = guid; zc.zc_cookie = ZPOOL_NO_REWIND; if (ioctl(hdl->libzfs_fd, ZFS_IOC_CLEAR, &zc) == 0) return (0); return (zpool_standard_error(hdl, errno, msg)); } /* * Change the GUID for a pool. */ int zpool_reguid(zpool_handle_t *zhp) { char msg[1024]; libzfs_handle_t *hdl = zhp->zpool_hdl; zfs_cmd_t zc = {"\0"}; (void) snprintf(msg, sizeof (msg), dgettext(TEXT_DOMAIN, "cannot reguid '%s'"), zhp->zpool_name); (void) strlcpy(zc.zc_name, zhp->zpool_name, sizeof (zc.zc_name)); if (zfs_ioctl(hdl, ZFS_IOC_POOL_REGUID, &zc) == 0) return (0); return (zpool_standard_error(hdl, errno, msg)); } /* * Reopen the pool. */ int zpool_reopen(zpool_handle_t *zhp) { zfs_cmd_t zc = {"\0"}; char msg[1024]; libzfs_handle_t *hdl = zhp->zpool_hdl; (void) snprintf(msg, sizeof (msg), dgettext(TEXT_DOMAIN, "cannot reopen '%s'"), zhp->zpool_name); (void) strlcpy(zc.zc_name, zhp->zpool_name, sizeof (zc.zc_name)); if (zfs_ioctl(hdl, ZFS_IOC_POOL_REOPEN, &zc) == 0) return (0); return (zpool_standard_error(hdl, errno, msg)); } /* * Convert from a devid string to a path. */ static char * devid_to_path(char *devid_str) { ddi_devid_t devid; char *minor; char *path; devid_nmlist_t *list = NULL; int ret; if (devid_str_decode(devid_str, &devid, &minor) != 0) return (NULL); ret = devid_deviceid_to_nmlist("/dev", devid, minor, &list); devid_str_free(minor); devid_free(devid); if (ret != 0) return (NULL); if ((path = strdup(list[0].devname)) == NULL) return (NULL); devid_free_nmlist(list); return (path); } /* * Convert from a path to a devid string. */ static char * path_to_devid(const char *path) { int fd; ddi_devid_t devid; char *minor, *ret; if ((fd = open(path, O_RDONLY)) < 0) return (NULL); minor = NULL; ret = NULL; if (devid_get(fd, &devid) == 0) { if (devid_get_minor_name(fd, &minor) == 0) ret = devid_str_encode(devid, minor); if (minor != NULL) devid_str_free(minor); devid_free(devid); } (void) close(fd); return (ret); } /* * Issue the necessary ioctl() to update the stored path value for the vdev. We * ignore any failure here, since a common case is for an unprivileged user to * type 'zpool status', and we'll display the correct information anyway. */ static void set_path(zpool_handle_t *zhp, nvlist_t *nv, const char *path) { zfs_cmd_t zc = {"\0"}; (void) strncpy(zc.zc_name, zhp->zpool_name, sizeof (zc.zc_name)); (void) strncpy(zc.zc_value, path, sizeof (zc.zc_value)); verify(nvlist_lookup_uint64(nv, ZPOOL_CONFIG_GUID, &zc.zc_guid) == 0); (void) ioctl(zhp->zpool_hdl->libzfs_fd, ZFS_IOC_VDEV_SETPATH, &zc); } /* * Remove partition suffix from a vdev path. Partition suffixes may take three * forms: "-partX", "pX", or "X", where X is a string of digits. The second * case only occurs when the suffix is preceded by a digit, i.e. "md0p0" The * third case only occurs when preceded by a string matching the regular * expression "^[hs]d[a-z]+", i.e. a scsi or ide disk. */ static char * strip_partition(libzfs_handle_t *hdl, char *path) { char *tmp = zfs_strdup(hdl, path); char *part = NULL, *d = NULL; if ((part = strstr(tmp, "-part")) && part != tmp) { d = part + 5; } else if ((part = strrchr(tmp, 'p')) && part > tmp + 1 && isdigit(*(part-1))) { d = part + 1; } else if ((tmp[0] == 'h' || tmp[0] == 's') && tmp[1] == 'd') { for (d = &tmp[2]; isalpha(*d); part = ++d); } if (part && d && *d != '\0') { for (; isdigit(*d); d++); if (*d == '\0') *part = '\0'; } return (tmp); } #define PATH_BUF_LEN 64 /* * Given a vdev, return the name to display in iostat. If the vdev has a path, * we use that, stripping off any leading "/dev/dsk/"; if not, we use the type. * We also check if this is a whole disk, in which case we strip off the * trailing 's0' slice name. * * This routine is also responsible for identifying when disks have been * reconfigured in a new location. The kernel will have opened the device by * devid, but the path will still refer to the old location. To catch this, we * first do a path -> devid translation (which is fast for the common case). If * the devid matches, we're done. If not, we do a reverse devid -> path * translation and issue the appropriate ioctl() to update the path of the vdev. * If 'zhp' is NULL, then this is an exported pool, and we don't need to do any * of these checks. */ char * zpool_vdev_name(libzfs_handle_t *hdl, zpool_handle_t *zhp, nvlist_t *nv, boolean_t verbose) { char *path, *devid, *type; uint64_t value; char buf[PATH_BUF_LEN]; char tmpbuf[PATH_BUF_LEN]; vdev_stat_t *vs; uint_t vsc; if (nvlist_lookup_uint64(nv, ZPOOL_CONFIG_NOT_PRESENT, &value) == 0) { verify(nvlist_lookup_uint64(nv, ZPOOL_CONFIG_GUID, &value) == 0); (void) snprintf(buf, sizeof (buf), "%llu", (u_longlong_t)value); path = buf; } else if (nvlist_lookup_string(nv, ZPOOL_CONFIG_PATH, &path) == 0) { /* * If the device is dead (faulted, offline, etc) then don't * bother opening it. Otherwise we may be forcing the user to * open a misbehaving device, which can have undesirable * effects. */ if ((nvlist_lookup_uint64_array(nv, ZPOOL_CONFIG_VDEV_STATS, (uint64_t **)&vs, &vsc) != 0 || vs->vs_state >= VDEV_STATE_DEGRADED) && zhp != NULL && nvlist_lookup_string(nv, ZPOOL_CONFIG_DEVID, &devid) == 0) { /* * Determine if the current path is correct. */ char *newdevid = path_to_devid(path); if (newdevid == NULL || strcmp(devid, newdevid) != 0) { char *newpath; if ((newpath = devid_to_path(devid)) != NULL) { /* * Update the path appropriately. */ set_path(zhp, nv, newpath); if (nvlist_add_string(nv, ZPOOL_CONFIG_PATH, newpath) == 0) verify(nvlist_lookup_string(nv, ZPOOL_CONFIG_PATH, &path) == 0); free(newpath); } } if (newdevid) devid_str_free(newdevid); } /* * For a block device only use the name. */ verify(nvlist_lookup_string(nv, ZPOOL_CONFIG_TYPE, &type) == 0); if (strcmp(type, VDEV_TYPE_DISK) == 0) { path = strrchr(path, '/'); path++; } /* * Remove the partition from the path it this is a whole disk. */ if (nvlist_lookup_uint64(nv, ZPOOL_CONFIG_WHOLE_DISK, &value) == 0 && value) { return (strip_partition(hdl, path)); } } else { verify(nvlist_lookup_string(nv, ZPOOL_CONFIG_TYPE, &path) == 0); /* * If it's a raidz device, we need to stick in the parity level. */ if (strcmp(path, VDEV_TYPE_RAIDZ) == 0) { verify(nvlist_lookup_uint64(nv, ZPOOL_CONFIG_NPARITY, &value) == 0); (void) snprintf(buf, sizeof (buf), "%s%llu", path, (u_longlong_t)value); path = buf; } /* * We identify each top-level vdev by using a * naming convention. */ if (verbose) { uint64_t id; verify(nvlist_lookup_uint64(nv, ZPOOL_CONFIG_ID, &id) == 0); (void) snprintf(tmpbuf, sizeof (tmpbuf), "%s-%llu", path, (u_longlong_t)id); path = tmpbuf; } } return (zfs_strdup(hdl, path)); } static int zbookmark_compare(const void *a, const void *b) { return (memcmp(a, b, sizeof (zbookmark_t))); } /* * Retrieve the persistent error log, uniquify the members, and return to the * caller. */ int zpool_get_errlog(zpool_handle_t *zhp, nvlist_t **nverrlistp) { zfs_cmd_t zc = {"\0"}; uint64_t count; zbookmark_t *zb = NULL; int i; /* * Retrieve the raw error list from the kernel. If the number of errors * has increased, allocate more space and continue until we get the * entire list. */ verify(nvlist_lookup_uint64(zhp->zpool_config, ZPOOL_CONFIG_ERRCOUNT, &count) == 0); if (count == 0) return (0); if ((zc.zc_nvlist_dst = (uintptr_t)zfs_alloc(zhp->zpool_hdl, count * sizeof (zbookmark_t))) == (uintptr_t)NULL) return (-1); zc.zc_nvlist_dst_size = count; (void) strcpy(zc.zc_name, zhp->zpool_name); for (;;) { if (ioctl(zhp->zpool_hdl->libzfs_fd, ZFS_IOC_ERROR_LOG, &zc) != 0) { free((void *)(uintptr_t)zc.zc_nvlist_dst); if (errno == ENOMEM) { count = zc.zc_nvlist_dst_size; if ((zc.zc_nvlist_dst = (uintptr_t) zfs_alloc(zhp->zpool_hdl, count * sizeof (zbookmark_t))) == (uintptr_t)NULL) return (-1); } else { return (-1); } } else { break; } } /* * Sort the resulting bookmarks. This is a little confusing due to the * implementation of ZFS_IOC_ERROR_LOG. The bookmarks are copied last * to first, and 'zc_nvlist_dst_size' indicates the number of boomarks * _not_ copied as part of the process. So we point the start of our * array appropriate and decrement the total number of elements. */ zb = ((zbookmark_t *)(uintptr_t)zc.zc_nvlist_dst) + zc.zc_nvlist_dst_size; count -= zc.zc_nvlist_dst_size; qsort(zb, count, sizeof (zbookmark_t), zbookmark_compare); verify(nvlist_alloc(nverrlistp, 0, KM_SLEEP) == 0); /* * Fill in the nverrlistp with nvlist's of dataset and object numbers. */ for (i = 0; i < count; i++) { nvlist_t *nv; /* ignoring zb_blkid and zb_level for now */ if (i > 0 && zb[i-1].zb_objset == zb[i].zb_objset && zb[i-1].zb_object == zb[i].zb_object) continue; if (nvlist_alloc(&nv, NV_UNIQUE_NAME, KM_SLEEP) != 0) goto nomem; if (nvlist_add_uint64(nv, ZPOOL_ERR_DATASET, zb[i].zb_objset) != 0) { nvlist_free(nv); goto nomem; } if (nvlist_add_uint64(nv, ZPOOL_ERR_OBJECT, zb[i].zb_object) != 0) { nvlist_free(nv); goto nomem; } if (nvlist_add_nvlist(*nverrlistp, "ejk", nv) != 0) { nvlist_free(nv); goto nomem; } nvlist_free(nv); } free((void *)(uintptr_t)zc.zc_nvlist_dst); return (0); nomem: free((void *)(uintptr_t)zc.zc_nvlist_dst); return (no_memory(zhp->zpool_hdl)); } /* * Upgrade a ZFS pool to the latest on-disk version. */ int zpool_upgrade(zpool_handle_t *zhp, uint64_t new_version) { zfs_cmd_t zc = {"\0"}; libzfs_handle_t *hdl = zhp->zpool_hdl; (void) strcpy(zc.zc_name, zhp->zpool_name); zc.zc_cookie = new_version; if (zfs_ioctl(hdl, ZFS_IOC_POOL_UPGRADE, &zc) != 0) return (zpool_standard_error_fmt(hdl, errno, dgettext(TEXT_DOMAIN, "cannot upgrade '%s'"), zhp->zpool_name)); return (0); } void zfs_save_arguments(int argc, char **argv, char *string, int len) { int i; (void) strlcpy(string, basename(argv[0]), len); for (i = 1; i < argc; i++) { (void) strlcat(string, " ", len); (void) strlcat(string, argv[i], len); } } int zpool_log_history(libzfs_handle_t *hdl, const char *message) { zfs_cmd_t zc = {"\0"}; nvlist_t *args; int err; args = fnvlist_alloc(); fnvlist_add_string(args, "message", message); err = zcmd_write_src_nvlist(hdl, &zc, args); if (err == 0) err = ioctl(hdl->libzfs_fd, ZFS_IOC_LOG_HISTORY, &zc); nvlist_free(args); zcmd_free_nvlists(&zc); return (err); } /* * Perform ioctl to get some command history of a pool. * * 'buf' is the buffer to fill up to 'len' bytes. 'off' is the * logical offset of the history buffer to start reading from. * * Upon return, 'off' is the next logical offset to read from and * 'len' is the actual amount of bytes read into 'buf'. */ static int get_history(zpool_handle_t *zhp, char *buf, uint64_t *off, uint64_t *len) { zfs_cmd_t zc = {"\0"}; libzfs_handle_t *hdl = zhp->zpool_hdl; (void) strlcpy(zc.zc_name, zhp->zpool_name, sizeof (zc.zc_name)); zc.zc_history = (uint64_t)(uintptr_t)buf; zc.zc_history_len = *len; zc.zc_history_offset = *off; if (ioctl(hdl->libzfs_fd, ZFS_IOC_POOL_GET_HISTORY, &zc) != 0) { switch (errno) { case EPERM: return (zfs_error_fmt(hdl, EZFS_PERM, dgettext(TEXT_DOMAIN, "cannot show history for pool '%s'"), zhp->zpool_name)); case ENOENT: return (zfs_error_fmt(hdl, EZFS_NOHISTORY, dgettext(TEXT_DOMAIN, "cannot get history for pool " "'%s'"), zhp->zpool_name)); case ENOTSUP: return (zfs_error_fmt(hdl, EZFS_BADVERSION, dgettext(TEXT_DOMAIN, "cannot get history for pool " "'%s', pool must be upgraded"), zhp->zpool_name)); default: return (zpool_standard_error_fmt(hdl, errno, dgettext(TEXT_DOMAIN, "cannot get history for '%s'"), zhp->zpool_name)); } } *len = zc.zc_history_len; *off = zc.zc_history_offset; return (0); } /* * Process the buffer of nvlists, unpacking and storing each nvlist record * into 'records'. 'leftover' is set to the number of bytes that weren't * processed as there wasn't a complete record. */ int zpool_history_unpack(char *buf, uint64_t bytes_read, uint64_t *leftover, nvlist_t ***records, uint_t *numrecords) { uint64_t reclen; nvlist_t *nv; int i; while (bytes_read > sizeof (reclen)) { /* get length of packed record (stored as little endian) */ for (i = 0, reclen = 0; i < sizeof (reclen); i++) reclen += (uint64_t)(((uchar_t *)buf)[i]) << (8*i); if (bytes_read < sizeof (reclen) + reclen) break; /* unpack record */ if (nvlist_unpack(buf + sizeof (reclen), reclen, &nv, 0) != 0) return (ENOMEM); bytes_read -= sizeof (reclen) + reclen; buf += sizeof (reclen) + reclen; /* add record to nvlist array */ (*numrecords)++; if (ISP2(*numrecords + 1)) { *records = realloc(*records, *numrecords * 2 * sizeof (nvlist_t *)); } (*records)[*numrecords - 1] = nv; } *leftover = bytes_read; return (0); } #define HIS_BUF_LEN (128*1024) /* * Retrieve the command history of a pool. */ int zpool_get_history(zpool_handle_t *zhp, nvlist_t **nvhisp) { char buf[HIS_BUF_LEN]; uint64_t off = 0; nvlist_t **records = NULL; uint_t numrecords = 0; int err, i; do { uint64_t bytes_read = sizeof (buf); uint64_t leftover; if ((err = get_history(zhp, buf, &off, &bytes_read)) != 0) break; /* if nothing else was read in, we're at EOF, just return */ if (!bytes_read) break; if ((err = zpool_history_unpack(buf, bytes_read, &leftover, &records, &numrecords)) != 0) break; off -= leftover; /* CONSTCOND */ } while (1); if (!err) { verify(nvlist_alloc(nvhisp, NV_UNIQUE_NAME, 0) == 0); verify(nvlist_add_nvlist_array(*nvhisp, ZPOOL_HIST_RECORD, records, numrecords) == 0); } for (i = 0; i < numrecords; i++) nvlist_free(records[i]); free(records); return (err); } /* * Retrieve the next event given the passed 'zevent_fd' file descriptor. * If there is a new event available 'nvp' will contain a newly allocated * nvlist and 'dropped' will be set to the number of missed events since * the last call to this function. When 'nvp' is set to NULL it indicates * no new events are available. In either case the function returns 0 and * it is up to the caller to free 'nvp'. In the case of a fatal error the * function will return a non-zero value. When the function is called in * blocking mode (the default, unless the ZEVENT_NONBLOCK flag is passed), * it will not return until a new event is available. */ int zpool_events_next(libzfs_handle_t *hdl, nvlist_t **nvp, int *dropped, unsigned flags, int zevent_fd) { zfs_cmd_t zc = {"\0"}; int error = 0; *nvp = NULL; *dropped = 0; zc.zc_cleanup_fd = zevent_fd; if (flags & ZEVENT_NONBLOCK) zc.zc_guid = ZEVENT_NONBLOCK; if (zcmd_alloc_dst_nvlist(hdl, &zc, ZEVENT_SIZE) != 0) return (-1); retry: if (zfs_ioctl(hdl, ZFS_IOC_EVENTS_NEXT, &zc) != 0) { switch (errno) { case ESHUTDOWN: error = zfs_error_fmt(hdl, EZFS_POOLUNAVAIL, dgettext(TEXT_DOMAIN, "zfs shutdown")); goto out; case ENOENT: /* Blocking error case should not occur */ if (!(flags & ZEVENT_NONBLOCK)) error = zpool_standard_error_fmt(hdl, errno, dgettext(TEXT_DOMAIN, "cannot get event")); goto out; case ENOMEM: if (zcmd_expand_dst_nvlist(hdl, &zc) != 0) { error = zfs_error_fmt(hdl, EZFS_NOMEM, dgettext(TEXT_DOMAIN, "cannot get event")); goto out; } else { goto retry; } default: error = zpool_standard_error_fmt(hdl, errno, dgettext(TEXT_DOMAIN, "cannot get event")); goto out; } } error = zcmd_read_dst_nvlist(hdl, &zc, nvp); if (error != 0) goto out; *dropped = (int)zc.zc_cookie; out: zcmd_free_nvlists(&zc); return (error); } /* * Clear all events. */ int zpool_events_clear(libzfs_handle_t *hdl, int *count) { zfs_cmd_t zc = {"\0"}; char msg[1024]; (void) snprintf(msg, sizeof (msg), dgettext(TEXT_DOMAIN, "cannot clear events")); if (zfs_ioctl(hdl, ZFS_IOC_EVENTS_CLEAR, &zc) != 0) return (zpool_standard_error_fmt(hdl, errno, msg)); if (count != NULL) *count = (int)zc.zc_cookie; /* # of events cleared */ return (0); } /* * Seek to a specific EID, ZEVENT_SEEK_START, or ZEVENT_SEEK_END for * the passed zevent_fd file handle. On success zero is returned, * otherwise -1 is returned and hdl->libzfs_error is set to the errno. */ int zpool_events_seek(libzfs_handle_t *hdl, uint64_t eid, int zevent_fd) { zfs_cmd_t zc = {"\0"}; int error = 0; zc.zc_guid = eid; zc.zc_cleanup_fd = zevent_fd; if (zfs_ioctl(hdl, ZFS_IOC_EVENTS_SEEK, &zc) != 0) { switch (errno) { case ENOENT: error = zfs_error_fmt(hdl, EZFS_NOENT, dgettext(TEXT_DOMAIN, "cannot get event")); break; case ENOMEM: error = zfs_error_fmt(hdl, EZFS_NOMEM, dgettext(TEXT_DOMAIN, "cannot get event")); break; default: error = zpool_standard_error_fmt(hdl, errno, dgettext(TEXT_DOMAIN, "cannot get event")); break; } } return (error); } void zpool_obj_to_path(zpool_handle_t *zhp, uint64_t dsobj, uint64_t obj, char *pathname, size_t len) { zfs_cmd_t zc = {"\0"}; boolean_t mounted = B_FALSE; char *mntpnt = NULL; char dsname[MAXNAMELEN]; if (dsobj == 0) { /* special case for the MOS */ (void) snprintf(pathname, len, ":<0x%llx>", (longlong_t)obj); return; } /* get the dataset's name */ (void) strlcpy(zc.zc_name, zhp->zpool_name, sizeof (zc.zc_name)); zc.zc_obj = dsobj; if (ioctl(zhp->zpool_hdl->libzfs_fd, ZFS_IOC_DSOBJ_TO_DSNAME, &zc) != 0) { /* just write out a path of two object numbers */ (void) snprintf(pathname, len, "<0x%llx>:<0x%llx>", (longlong_t)dsobj, (longlong_t)obj); return; } (void) strlcpy(dsname, zc.zc_value, sizeof (dsname)); /* find out if the dataset is mounted */ mounted = is_mounted(zhp->zpool_hdl, dsname, &mntpnt); /* get the corrupted object's path */ (void) strlcpy(zc.zc_name, dsname, sizeof (zc.zc_name)); zc.zc_obj = obj; if (ioctl(zhp->zpool_hdl->libzfs_fd, ZFS_IOC_OBJ_TO_PATH, &zc) == 0) { if (mounted) { (void) snprintf(pathname, len, "%s%s", mntpnt, zc.zc_value); } else { (void) snprintf(pathname, len, "%s:%s", dsname, zc.zc_value); } } else { (void) snprintf(pathname, len, "%s:<0x%llx>", dsname, (longlong_t)obj); } free(mntpnt); } /* * Read the EFI label from the config, if a label does not exist then * pass back the error to the caller. If the caller has passed a non-NULL * diskaddr argument then we set it to the starting address of the EFI * partition. */ static int read_efi_label(nvlist_t *config, diskaddr_t *sb) { char *path; int fd; char diskname[MAXPATHLEN]; int err = -1; if (nvlist_lookup_string(config, ZPOOL_CONFIG_PATH, &path) != 0) return (err); (void) snprintf(diskname, sizeof (diskname), "%s%s", DISK_ROOT, strrchr(path, '/')); if ((fd = open(diskname, O_RDWR|O_DIRECT)) >= 0) { struct dk_gpt *vtoc; if ((err = efi_alloc_and_read(fd, &vtoc)) >= 0) { if (sb != NULL) *sb = vtoc->efi_parts[0].p_start; efi_free(vtoc); } (void) close(fd); } return (err); } /* * determine where a partition starts on a disk in the current * configuration */ static diskaddr_t find_start_block(nvlist_t *config) { nvlist_t **child; uint_t c, children; diskaddr_t sb = MAXOFFSET_T; uint64_t wholedisk; if (nvlist_lookup_nvlist_array(config, ZPOOL_CONFIG_CHILDREN, &child, &children) != 0) { if (nvlist_lookup_uint64(config, ZPOOL_CONFIG_WHOLE_DISK, &wholedisk) != 0 || !wholedisk) { return (MAXOFFSET_T); } if (read_efi_label(config, &sb) < 0) sb = MAXOFFSET_T; return (sb); } for (c = 0; c < children; c++) { sb = find_start_block(child[c]); if (sb != MAXOFFSET_T) { return (sb); } } return (MAXOFFSET_T); } int zpool_label_disk_wait(char *path, int timeout) { struct stat64 statbuf; int i; /* * Wait timeout miliseconds for a newly created device to be available * from the given path. There is a small window when a /dev/ device * will exist and the udev link will not, so we must wait for the * symlink. Depending on the udev rules this may take a few seconds. */ for (i = 0; i < timeout; i++) { usleep(1000); errno = 0; if ((stat64(path, &statbuf) == 0) && (errno == 0)) return (0); } return (ENOENT); } int zpool_label_disk_check(char *path) { struct dk_gpt *vtoc; int fd, err; if ((fd = open(path, O_RDWR|O_DIRECT)) < 0) return (errno); if ((err = efi_alloc_and_read(fd, &vtoc)) != 0) { (void) close(fd); return (err); } if (vtoc->efi_flags & EFI_GPT_PRIMARY_CORRUPT) { efi_free(vtoc); (void) close(fd); return (EIDRM); } efi_free(vtoc); (void) close(fd); return (0); } /* * Label an individual disk. The name provided is the short name, * stripped of any leading /dev path. */ int zpool_label_disk(libzfs_handle_t *hdl, zpool_handle_t *zhp, char *name) { char path[MAXPATHLEN]; struct dk_gpt *vtoc; int rval, fd; size_t resv = EFI_MIN_RESV_SIZE; uint64_t slice_size; diskaddr_t start_block; char errbuf[1024]; /* prepare an error message just in case */ (void) snprintf(errbuf, sizeof (errbuf), dgettext(TEXT_DOMAIN, "cannot label '%s'"), name); if (zhp) { nvlist_t *nvroot; #if defined(__sun__) || defined(__sun) if (zpool_is_bootable(zhp)) { zfs_error_aux(hdl, dgettext(TEXT_DOMAIN, "EFI labeled devices are not supported on root " "pools.")); return (zfs_error(hdl, EZFS_POOL_NOTSUP, errbuf)); } #endif verify(nvlist_lookup_nvlist(zhp->zpool_config, ZPOOL_CONFIG_VDEV_TREE, &nvroot) == 0); if (zhp->zpool_start_block == 0) start_block = find_start_block(nvroot); else start_block = zhp->zpool_start_block; zhp->zpool_start_block = start_block; } else { /* new pool */ start_block = NEW_START_BLOCK; } (void) snprintf(path, sizeof (path), "%s/%s", DISK_ROOT, name); if ((fd = open(path, O_RDWR|O_DIRECT)) < 0) { /* * This shouldn't happen. We've long since verified that this * is a valid device. */ zfs_error_aux(hdl, dgettext(TEXT_DOMAIN, "cannot " "label '%s': unable to open device: %d"), path, errno); return (zfs_error(hdl, EZFS_OPENFAILED, errbuf)); } if (efi_alloc_and_init(fd, EFI_NUMPAR, &vtoc) != 0) { /* * The only way this can fail is if we run out of memory, or we * were unable to read the disk's capacity */ if (errno == ENOMEM) (void) no_memory(hdl); (void) close(fd); zfs_error_aux(hdl, dgettext(TEXT_DOMAIN, "cannot " "label '%s': unable to read disk capacity"), path); return (zfs_error(hdl, EZFS_NOCAP, errbuf)); } slice_size = vtoc->efi_last_u_lba + 1; slice_size -= EFI_MIN_RESV_SIZE; if (start_block == MAXOFFSET_T) start_block = NEW_START_BLOCK; slice_size -= start_block; slice_size = P2ALIGN(slice_size, PARTITION_END_ALIGNMENT); vtoc->efi_parts[0].p_start = start_block; vtoc->efi_parts[0].p_size = slice_size; /* * Why we use V_USR: V_BACKUP confuses users, and is considered * disposable by some EFI utilities (since EFI doesn't have a backup * slice). V_UNASSIGNED is supposed to be used only for zero size * partitions, and efi_write() will fail if we use it. V_ROOT, V_BOOT, * etc. were all pretty specific. V_USR is as close to reality as we * can get, in the absence of V_OTHER. */ vtoc->efi_parts[0].p_tag = V_USR; (void) strcpy(vtoc->efi_parts[0].p_name, "zfs"); vtoc->efi_parts[8].p_start = slice_size + start_block; vtoc->efi_parts[8].p_size = resv; vtoc->efi_parts[8].p_tag = V_RESERVED; if ((rval = efi_write(fd, vtoc)) != 0 || (rval = efi_rescan(fd)) != 0) { /* * Some block drivers (like pcata) may not support EFI * GPT labels. Print out a helpful error message dir- * ecting the user to manually label the disk and give * a specific slice. */ (void) close(fd); efi_free(vtoc); zfs_error_aux(hdl, dgettext(TEXT_DOMAIN, "try using " "parted(8) and then provide a specific slice: %d"), rval); return (zfs_error(hdl, EZFS_LABELFAILED, errbuf)); } (void) close(fd); efi_free(vtoc); /* Wait for the first expected partition to appear. */ (void) snprintf(path, sizeof (path), "%s/%s", DISK_ROOT, name); (void) zfs_append_partition(path, MAXPATHLEN); rval = zpool_label_disk_wait(path, 3000); if (rval) { zfs_error_aux(hdl, dgettext(TEXT_DOMAIN, "failed to " "detect device partitions on '%s': %d"), path, rval); return (zfs_error(hdl, EZFS_LABELFAILED, errbuf)); } /* We can't be to paranoid. Read the label back and verify it. */ (void) snprintf(path, sizeof (path), "%s/%s", DISK_ROOT, name); rval = zpool_label_disk_check(path); if (rval) { zfs_error_aux(hdl, dgettext(TEXT_DOMAIN, "freshly written " "EFI label on '%s' is damaged. Ensure\nthis device " "is not in in use, and is functioning properly: %d"), path, rval); return (zfs_error(hdl, EZFS_LABELFAILED, errbuf)); } return (0); } diff --git a/man/man5/zfs-module-parameters.5 b/man/man5/zfs-module-parameters.5 index d3aa2e6b043b..2f87f114c50b 100644 --- a/man/man5/zfs-module-parameters.5 +++ b/man/man5/zfs-module-parameters.5 @@ -1,1419 +1,1456 @@ '\" te .\" Copyright (c) 2013 by Turbo Fredriksson . All rights reserved. .\" 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] .TH ZFS-MODULE-PARAMETERS 5 "Nov 16, 2013" .SH NAME zfs\-module\-parameters \- ZFS module parameters .SH DESCRIPTION .sp .LP Description of the different parameters to the ZFS module. .SS "Module parameters" .sp .LP .sp .ne 2 .na \fBl2arc_feed_again\fR (int) .ad .RS 12n Turbo L2ARC warmup .sp Use \fB1\fR for yes (default) and \fB0\fR to disable. .RE .sp .ne 2 .na \fBl2arc_feed_min_ms\fR (ulong) .ad .RS 12n Min feed interval in milliseconds .sp Default value: \fB200\fR. .RE .sp .ne 2 .na \fBl2arc_feed_secs\fR (ulong) .ad .RS 12n Seconds between L2ARC writing .sp Default value: \fB1\fR. .RE .sp .ne 2 .na \fBl2arc_headroom\fR (ulong) .ad .RS 12n Number of max device writes to precache .sp Default value: \fB2\fR. .RE .sp .ne 2 .na \fBl2arc_headroom_boost\fR (ulong) .ad .RS 12n Compressed l2arc_headroom multiplier .sp Default value: \fB200\fR. .RE .sp .ne 2 .na \fBl2arc_nocompress\fR (int) .ad .RS 12n Skip compressing L2ARC buffers .sp Use \fB1\fR for yes and \fB0\fR for no (default). .RE .sp .ne 2 .na \fBl2arc_noprefetch\fR (int) .ad .RS 12n Skip caching prefetched buffers .sp Use \fB1\fR for yes (default) and \fB0\fR to disable. .RE .sp .ne 2 .na \fBl2arc_norw\fR (int) .ad .RS 12n No reads during writes .sp Use \fB1\fR for yes and \fB0\fR for no (default). .RE .sp .ne 2 .na \fBl2arc_write_boost\fR (ulong) .ad .RS 12n Extra write bytes during device warmup .sp Default value: \fB8,388,608\fR. .RE .sp .ne 2 .na \fBl2arc_write_max\fR (ulong) .ad .RS 12n Max write bytes per interval .sp Default value: \fB8,388,608\fR. .RE .sp .ne 2 .na \fBmetaslab_debug_load\fR (int) .ad .RS 12n Load all metaslabs during pool import. .sp Use \fB1\fR for yes and \fB0\fR for no (default). .RE .sp .ne 2 .na \fBmetaslab_debug_unload\fR (int) .ad .RS 12n Prevent metaslabs from being unloaded. .sp Use \fB1\fR for yes and \fB0\fR for no (default). .RE .sp .ne 2 .na \fBspa_config_path\fR (charp) .ad .RS 12n SPA config file .sp Default value: \fB/etc/zfs/zpool.cache\fR. .RE .sp .ne 2 .na \fBspa_asize_inflation\fR (int) .ad .RS 12n Multiplication factor used to estimate actual disk consumption from the size of data being written. The default value is a worst case estimate, but lower values may be valid for a given pool depending on its configuration. Pool administrators who understand the factors involved may wish to specify a more realistic inflation factor, particularly if they operate close to quota or capacity limits. .sp Default value: 24 .RE .sp .ne 2 .na \fBzfetch_array_rd_sz\fR (ulong) .ad .RS 12n If prefetching is enabled, disable prefetching for reads larger than this size. .sp Default value: \fB1,048,576\fR. .RE .sp .ne 2 .na \fBzfetch_block_cap\fR (uint) .ad .RS 12n Max number of blocks to prefetch at a time .sp Default value: \fB256\fR. .RE .sp .ne 2 .na \fBzfetch_max_streams\fR (uint) .ad .RS 12n Max number of streams per zfetch (prefetch streams per file). .sp Default value: \fB8\fR. .RE .sp .ne 2 .na \fBzfetch_min_sec_reap\fR (uint) .ad .RS 12n Min time before an active prefetch stream can be reclaimed .sp Default value: \fB2\fR. .RE .sp .ne 2 .na \fBzfs_arc_grow_retry\fR (int) .ad .RS 12n Seconds before growing arc size .sp Default value: \fB5\fR. .RE .sp .ne 2 .na \fBzfs_arc_max\fR (ulong) .ad .RS 12n Max arc size .sp Default value: \fB0\fR. .RE .sp .ne 2 .na \fBzfs_arc_memory_throttle_disable\fR (int) .ad .RS 12n Disable memory throttle .sp Use \fB1\fR for yes (default) and \fB0\fR to disable. .RE .sp .ne 2 .na \fBzfs_arc_meta_limit\fR (ulong) .ad .RS 12n Meta limit for arc size .sp Default value: \fB0\fR. .RE .sp .ne 2 .na \fBzfs_arc_meta_prune\fR (int) .ad .RS 12n Bytes of meta data to prune .sp Default value: \fB1,048,576\fR. .RE .sp .ne 2 .na \fBzfs_arc_min\fR (ulong) .ad .RS 12n Min arc size .sp Default value: \fB100\fR. .RE .sp .ne 2 .na \fBzfs_arc_min_prefetch_lifespan\fR (int) .ad .RS 12n Min life of prefetch block .sp Default value: \fB100\fR. .RE .sp .ne 2 .na \fBzfs_arc_p_aggressive_disable\fR (int) .ad .RS 12n Disable aggressive arc_p growth .sp Use \fB1\fR for yes (default) and \fB0\fR to disable. .RE .sp .ne 2 .na \fBzfs_arc_p_dampener_disable\fR (int) .ad .RS 12n Disable arc_p adapt dampener .sp Use \fB1\fR for yes (default) and \fB0\fR to disable. .RE .sp .ne 2 .na \fBzfs_arc_shrink_shift\fR (int) .ad .RS 12n log2(fraction of arc to reclaim) .sp Default value: \fB5\fR. .RE .sp .ne 2 .na \fBzfs_autoimport_disable\fR (int) .ad .RS 12n Disable pool import at module load by ignoring the cache file (typically \fB/etc/zfs/zpool.cache\fR). .sp Use \fB1\fR for yes and \fB0\fR for no (default). .RE .sp .ne 2 .na \fBzfs_dbuf_state_index\fR (int) .ad .RS 12n Calculate arc header index .sp Default value: \fB0\fR. .RE .sp .ne 2 .na \fBzfs_deadman_enabled\fR (int) .ad .RS 12n Enable deadman timer .sp Use \fB1\fR for yes (default) and \fB0\fR to disable. .RE .sp .ne 2 .na \fBzfs_deadman_synctime_ms\fR (ulong) .ad .RS 12n Expiration time in milliseconds. This value has two meanings. First it is used to determine when the spa_deadman() logic should fire. By default the spa_deadman() will fire if spa_sync() has not completed in 1000 seconds. Secondly, the value determines if an I/O is considered "hung". Any I/O that has not completed in zfs_deadman_synctime_ms is considered "hung" resulting in a zevent being logged. .sp Default value: \fB1,000,000\fR. .RE .sp .ne 2 .na \fBzfs_dedup_prefetch\fR (int) .ad .RS 12n Enable prefetching dedup-ed blks .sp Use \fB1\fR for yes (default) and \fB0\fR to disable. .RE .sp .ne 2 .na \fBzfs_delay_min_dirty_percent\fR (int) .ad .RS 12n Start to delay each transaction once there is this amount of dirty data, expressed as a percentage of \fBzfs_dirty_data_max\fR. This value should be >= zfs_vdev_async_write_active_max_dirty_percent. See the section "ZFS TRANSACTION DELAY". .sp Default value: \fB60\fR. .RE .sp .ne 2 .na \fBzfs_delay_scale\fR (int) .ad .RS 12n This controls how quickly the transaction delay approaches infinity. Larger values cause longer delays for a given amount of dirty data. .sp For the smoothest delay, this value should be about 1 billion divided by the maximum number of operations per second. This will smoothly handle between 10x and 1/10th this number. .sp See the section "ZFS TRANSACTION DELAY". .sp Note: \fBzfs_delay_scale\fR * \fBzfs_dirty_data_max\fR must be < 2^64. .sp Default value: \fB500,000\fR. .RE .sp .ne 2 .na \fBzfs_dirty_data_max\fR (int) .ad .RS 12n Determines the dirty space limit in bytes. Once this limit is exceeded, new writes are halted until space frees up. This parameter takes precedence over \fBzfs_dirty_data_max_percent\fR. See the section "ZFS TRANSACTION DELAY". .sp Default value: 10 percent of all memory, capped at \fBzfs_dirty_data_max_max\fR. .RE .sp .ne 2 .na \fBzfs_dirty_data_max_max\fR (int) .ad .RS 12n Maximum allowable value of \fBzfs_dirty_data_max\fR, expressed in bytes. This limit is only enforced at module load time, and will be ignored if \fBzfs_dirty_data_max\fR is later changed. This parameter takes precedence over \fBzfs_dirty_data_max_max_percent\fR. See the section "ZFS TRANSACTION DELAY". .sp Default value: 25% of physical RAM. .RE .sp .ne 2 .na \fBzfs_dirty_data_max_max_percent\fR (int) .ad .RS 12n Maximum allowable value of \fBzfs_dirty_data_max\fR, expressed as a percentage of physical RAM. This limit is only enforced at module load time, and will be ignored if \fBzfs_dirty_data_max\fR is later changed. The parameter \fBzfs_dirty_data_max_max\fR takes precedence over this one. See the section "ZFS TRANSACTION DELAY". .sp Default value: 25 .RE .sp .ne 2 .na \fBzfs_dirty_data_max_percent\fR (int) .ad .RS 12n Determines the dirty space limit, expressed as a percentage of all memory. Once this limit is exceeded, new writes are halted until space frees up. The parameter \fBzfs_dirty_data_max\fR takes precedence over this one. See the section "ZFS TRANSACTION DELAY". .sp Default value: 10%, subject to \fBzfs_dirty_data_max_max\fR. .RE .sp .ne 2 .na \fBzfs_dirty_data_sync\fR (int) .ad .RS 12n Start syncing out a transaction group if there is at least this much dirty data. .sp Default value: \fB67,108,864\fR. .RE .sp .ne 2 .na \fBzfs_vdev_async_read_max_active\fR (int) .ad .RS 12n Maxium asynchronous read I/Os active to each device. See the section "ZFS I/O SCHEDULER". .sp Default value: \fB3\fR. .RE .sp .ne 2 .na \fBzfs_vdev_async_read_min_active\fR (int) .ad .RS 12n Minimum asynchronous read I/Os active to each device. See the section "ZFS I/O SCHEDULER". .sp Default value: \fB1\fR. .RE .sp .ne 2 .na \fBzfs_vdev_async_write_active_max_dirty_percent\fR (int) .ad .RS 12n When the pool has more than \fBzfs_vdev_async_write_active_max_dirty_percent\fR dirty data, use \fBzfs_vdev_async_write_max_active\fR to limit active async writes. If the dirty data is between min and max, the active I/O limit is linearly interpolated. See the section "ZFS I/O SCHEDULER". .sp Default value: \fB60\fR. .RE .sp .ne 2 .na \fBzfs_vdev_async_write_active_min_dirty_percent\fR (int) .ad .RS 12n When the pool has less than \fBzfs_vdev_async_write_active_min_dirty_percent\fR dirty data, use \fBzfs_vdev_async_write_min_active\fR to limit active async writes. If the dirty data is between min and max, the active I/O limit is linearly interpolated. See the section "ZFS I/O SCHEDULER". .sp Default value: \fB30\fR. .RE .sp .ne 2 .na \fBzfs_vdev_async_write_max_active\fR (int) .ad .RS 12n Maxium asynchronous write I/Os active to each device. See the section "ZFS I/O SCHEDULER". .sp Default value: \fB10\fR. .RE .sp .ne 2 .na \fBzfs_vdev_async_write_min_active\fR (int) .ad .RS 12n Minimum asynchronous write I/Os active to each device. See the section "ZFS I/O SCHEDULER". .sp Default value: \fB1\fR. .RE .sp .ne 2 .na \fBzfs_vdev_max_active\fR (int) .ad .RS 12n The maximum number of I/Os active to each device. Ideally, this will be >= the sum of each queue's max_active. It must be at least the sum of each queue's min_active. See the section "ZFS I/O SCHEDULER". .sp Default value: \fB1,000\fR. .RE .sp .ne 2 .na \fBzfs_vdev_scrub_max_active\fR (int) .ad .RS 12n Maxium scrub I/Os active to each device. See the section "ZFS I/O SCHEDULER". .sp Default value: \fB2\fR. .RE .sp .ne 2 .na \fBzfs_vdev_scrub_min_active\fR (int) .ad .RS 12n Minimum scrub I/Os active to each device. See the section "ZFS I/O SCHEDULER". .sp Default value: \fB1\fR. .RE .sp .ne 2 .na \fBzfs_vdev_sync_read_max_active\fR (int) .ad .RS 12n Maxium synchronous read I/Os active to each device. See the section "ZFS I/O SCHEDULER". .sp Default value: \fB10\fR. .RE .sp .ne 2 .na \fBzfs_vdev_sync_read_min_active\fR (int) .ad .RS 12n Minimum synchronous read I/Os active to each device. See the section "ZFS I/O SCHEDULER". .sp Default value: \fB10\fR. .RE .sp .ne 2 .na \fBzfs_vdev_sync_write_max_active\fR (int) .ad .RS 12n Maxium synchronous write I/Os active to each device. See the section "ZFS I/O SCHEDULER". .sp Default value: \fB10\fR. .RE .sp .ne 2 .na \fBzfs_vdev_sync_write_min_active\fR (int) .ad .RS 12n Minimum synchronous write I/Os active to each device. See the section "ZFS I/O SCHEDULER". .sp Default value: \fB10\fR. .RE .sp .ne 2 .na \fBzfs_disable_dup_eviction\fR (int) .ad .RS 12n Disable duplicate buffer eviction .sp Use \fB1\fR for yes and \fB0\fR for no (default). .RE .sp .ne 2 .na \fBzfs_expire_snapshot\fR (int) .ad .RS 12n Seconds to expire .zfs/snapshot .sp Default value: \fB300\fR. .RE .sp .ne 2 .na \fBzfs_flags\fR (int) .ad .RS 12n Set additional debugging flags .sp Default value: \fB1\fR. .RE +.sp +.ne 2 +.na +\fBzfs_free_leak_on_eio\fR (int) +.ad +.RS 12n +If destroy encounters an EIO while reading metadata (e.g. indirect +blocks), space referenced by the missing metadata can not be freed. +Normally this causes the background destroy to become "stalled", as +it is unable to make forward progress. While in this stalled state, +all remaining space to free from the error-encountering filesystem is +"temporarily leaked". Set this flag to cause it to ignore the EIO, +permanently leak the space from indirect blocks that can not be read, +and continue to free everything else that it can. + +The default, "stalling" behavior is useful if the storage partially +fails (i.e. some but not all i/os fail), and then later recovers. In +this case, we will be able to continue pool operations while it is +partially failed, and when it recovers, we can continue to free the +space, with no leaks. However, note that this case is actually +fairly rare. + +Typically pools either (a) fail completely (but perhaps temporarily, +e.g. a top-level vdev going offline), or (b) have localized, +permanent errors (e.g. disk returns the wrong data due to bit flip or +firmware bug). In case (a), this setting does not matter because the +pool will be suspended and the sync thread will not be able to make +forward progress regardless. In case (b), because the error is +permanent, the best we can do is leak the minimum amount of space, +which is what setting this flag will do. Therefore, it is reasonable +for this flag to normally be set, but we chose the more conservative +approach of not setting it, so that there is no possibility of +leaking space in the "partial temporary" failure case. +.sp +Default value: \fB0\fR. +.RE + .sp .ne 2 .na \fBzfs_free_min_time_ms\fR (int) .ad .RS 12n Min millisecs to free per txg .sp Default value: \fB1,000\fR. .RE .sp .ne 2 .na \fBzfs_immediate_write_sz\fR (long) .ad .RS 12n Largest data block to write to zil .sp Default value: \fB32,768\fR. .RE .sp .ne 2 .na \fBzfs_mdcomp_disable\fR (int) .ad .RS 12n Disable meta data compression .sp Use \fB1\fR for yes and \fB0\fR for no (default). .RE .sp .ne 2 .na \fBzfs_mg_noalloc_threshold\fR (int) .ad .RS 12n Defines a threshold at which metaslab groups should be eligible for allocations. The value is expressed as a percentage of free space beyond which a metaslab group is always eligible for allocations. If a metaslab group's free space is less than or equal to the the threshold, the allocator will avoid allocating to that group unless all groups in the pool have reached the threshold. Once all groups have reached the threshold, all groups are allowed to accept allocations. The default value of 0 disables the feature and causes all metaslab groups to be eligible for allocations. This parameter allows to deal with pools having heavily imbalanced vdevs such as would be the case when a new vdev has been added. Setting the threshold to a non-zero percentage will stop allocations from being made to vdevs that aren't filled to the specified percentage and allow lesser filled vdevs to acquire more allocations than they otherwise would under the old \fBzfs_mg_alloc_failures\fR facility. .sp Default value: \fB0\fR. .RE .sp .ne 2 .na \fBzfs_no_scrub_io\fR (int) .ad .RS 12n Set for no scrub I/O .sp Use \fB1\fR for yes and \fB0\fR for no (default). .RE .sp .ne 2 .na \fBzfs_no_scrub_prefetch\fR (int) .ad .RS 12n Set for no scrub prefetching .sp Use \fB1\fR for yes and \fB0\fR for no (default). .RE .sp .ne 2 .na \fBzfs_nocacheflush\fR (int) .ad .RS 12n Disable cache flushes .sp Use \fB1\fR for yes and \fB0\fR for no (default). .RE .sp .ne 2 .na \fBzfs_nopwrite_enabled\fR (int) .ad .RS 12n Enable NOP writes .sp Use \fB1\fR for yes (default) and \fB0\fR to disable. .RE .sp .ne 2 .na \fBzfs_pd_blks_max\fR (int) .ad .RS 12n Max number of blocks to prefetch .sp Default value: \fB100\fR. .RE .sp .ne 2 .na \fBzfs_prefetch_disable\fR (int) .ad .RS 12n Disable all ZFS prefetching .sp Use \fB1\fR for yes and \fB0\fR for no (default). .RE .sp .ne 2 .na \fBzfs_read_chunk_size\fR (long) .ad .RS 12n Bytes to read per chunk .sp Default value: \fB1,048,576\fR. .RE .sp .ne 2 .na \fBzfs_read_history\fR (int) .ad .RS 12n Historic statistics for the last N reads .sp Default value: \fB0\fR. .RE .sp .ne 2 .na \fBzfs_read_history_hits\fR (int) .ad .RS 12n Include cache hits in read history .sp Use \fB1\fR for yes and \fB0\fR for no (default). .RE .sp .ne 2 .na \fBzfs_recover\fR (int) .ad .RS 12n Set to attempt to recover from fatal errors. This should only be used as a last resort, as it typically results in leaked space, or worse. .sp Use \fB1\fR for yes and \fB0\fR for no (default). .RE .sp .ne 2 .na \fBzfs_resilver_delay\fR (int) .ad .RS 12n Number of ticks to delay prior to issuing a resilver I/O operation when a non-resilver or non-scrub I/O operation has occurred within the past \fBzfs_scan_idle\fR ticks. .sp Default value: \fB2\fR. .RE .sp .ne 2 .na \fBzfs_resilver_min_time_ms\fR (int) .ad .RS 12n Min millisecs to resilver per txg .sp Default value: \fB3,000\fR. .RE .sp .ne 2 .na \fBzfs_scan_idle\fR (int) .ad .RS 12n Idle window in clock ticks. During a scrub or a resilver, if a non-scrub or non-resilver I/O operation has occurred during this window, the next scrub or resilver operation is delayed by, respectively \fBzfs_scrub_delay\fR or \fBzfs_resilver_delay\fR ticks. .sp Default value: \fB50\fR. .RE .sp .ne 2 .na \fBzfs_scan_min_time_ms\fR (int) .ad .RS 12n Min millisecs to scrub per txg .sp Default value: \fB1,000\fR. .RE .sp .ne 2 .na \fBzfs_scrub_delay\fR (int) .ad .RS 12n Number of ticks to delay prior to issuing a scrub I/O operation when a non-scrub or non-resilver I/O operation has occurred within the past \fBzfs_scan_idle\fR ticks. .sp Default value: \fB4\fR. .RE .sp .ne 2 .na \fBzfs_send_corrupt_data\fR (int) .ad .RS 12n Allow to send corrupt data (ignore read/checksum errors when sending data) .sp Use \fB1\fR for yes and \fB0\fR for no (default). .RE .sp .ne 2 .na \fBzfs_sync_pass_deferred_free\fR (int) .ad .RS 12n Defer frees starting in this pass .sp Default value: \fB2\fR. .RE .sp .ne 2 .na \fBzfs_sync_pass_dont_compress\fR (int) .ad .RS 12n Don't compress starting in this pass .sp Default value: \fB5\fR. .RE .sp .ne 2 .na \fBzfs_sync_pass_rewrite\fR (int) .ad .RS 12n Rewrite new bps starting in this pass .sp Default value: \fB2\fR. .RE .sp .ne 2 .na \fBzfs_top_maxinflight\fR (int) .ad .RS 12n Max I/Os per top-level vdev during scrub or resilver operations. .sp Default value: \fB32\fR. .RE .sp .ne 2 .na \fBzfs_txg_history\fR (int) .ad .RS 12n Historic statistics for the last N txgs .sp Default value: \fB0\fR. .RE .sp .ne 2 .na \fBzfs_txg_timeout\fR (int) .ad .RS 12n Max seconds worth of delta per txg .sp Default value: \fB5\fR. .RE .sp .ne 2 .na \fBzfs_vdev_aggregation_limit\fR (int) .ad .RS 12n Max vdev I/O aggregation size .sp Default value: \fB131,072\fR. .RE .sp .ne 2 .na \fBzfs_vdev_cache_bshift\fR (int) .ad .RS 12n Shift size to inflate reads too .sp Default value: \fB16\fR. .RE .sp .ne 2 .na \fBzfs_vdev_cache_max\fR (int) .ad .RS 12n Inflate reads small than max .RE .sp .ne 2 .na \fBzfs_vdev_cache_size\fR (int) .ad .RS 12n Total size of the per-disk cache .sp Default value: \fB0\fR. .RE .sp .ne 2 .na \fBzfs_vdev_mirror_switch_us\fR (int) .ad .RS 12n Switch mirrors every N usecs .sp Default value: \fB10,000\fR. .RE .sp .ne 2 .na \fBzfs_vdev_read_gap_limit\fR (int) .ad .RS 12n Aggregate read I/O over gap .sp Default value: \fB32,768\fR. .RE .sp .ne 2 .na \fBzfs_vdev_scheduler\fR (charp) .ad .RS 12n I/O scheduler .sp Default value: \fBnoop\fR. .RE .sp .ne 2 .na \fBzfs_vdev_write_gap_limit\fR (int) .ad .RS 12n Aggregate write I/O over gap .sp Default value: \fB4,096\fR. .RE .sp .ne 2 .na \fBzfs_zevent_cols\fR (int) .ad .RS 12n Max event column width .sp Default value: \fB80\fR. .RE .sp .ne 2 .na \fBzfs_zevent_console\fR (int) .ad .RS 12n Log events to the console .sp Use \fB1\fR for yes and \fB0\fR for no (default). .RE .sp .ne 2 .na \fBzfs_zevent_len_max\fR (int) .ad .RS 12n Max event queue length .sp Default value: \fB0\fR. .RE .sp .ne 2 .na \fBzil_replay_disable\fR (int) .ad .RS 12n Disable intent logging replay .sp Use \fB1\fR for yes and \fB0\fR for no (default). .RE .sp .ne 2 .na \fBzil_slog_limit\fR (ulong) .ad .RS 12n Max commit bytes to separate log device .sp Default value: \fB1,048,576\fR. .RE .sp .ne 2 .na \fBzio_bulk_flags\fR (int) .ad .RS 12n Additional flags to pass to bulk buffers .sp Default value: \fB0\fR. .RE .sp .ne 2 .na \fBzio_delay_max\fR (int) .ad .RS 12n Max zio millisec delay before posting event .sp Default value: \fB30,000\fR. .RE .sp .ne 2 .na \fBzio_injection_enabled\fR (int) .ad .RS 12n Enable fault injection .sp Use \fB1\fR for yes and \fB0\fR for no (default). .RE .sp .ne 2 .na \fBzio_requeue_io_start_cut_in_line\fR (int) .ad .RS 12n Prioritize requeued I/O .sp Default value: \fB0\fR. .RE .sp .ne 2 .na \fBzvol_inhibit_dev\fR (uint) .ad .RS 12n Do not create zvol device nodes .sp Use \fB1\fR for yes and \fB0\fR for no (default). .RE .sp .ne 2 .na \fBzvol_major\fR (uint) .ad .RS 12n Major number for zvol device .sp Default value: \fB230\fR. .RE .sp .ne 2 .na \fBzvol_max_discard_blocks\fR (ulong) .ad .RS 12n Max number of blocks to discard at once .sp Default value: \fB16,384\fR. .RE .sp .ne 2 .na \fBzvol_threads\fR (uint) .ad .RS 12n Number of threads for zvol device .sp Default value: \fB32\fR. .RE .SH ZFS I/O SCHEDULER ZFS issues I/O operations to leaf vdevs to satisfy and complete I/Os. The I/O scheduler determines when and in what order those operations are issued. The I/O scheduler divides operations into five I/O classes prioritized in the following order: sync read, sync write, async read, async write, and scrub/resilver. Each queue defines the minimum and maximum number of concurrent operations that may be issued to the device. In addition, the device has an aggregate maximum, \fBzfs_vdev_max_active\fR. Note that the sum of the per-queue minimums must not exceed the aggregate maximum. If the sum of the per-queue maximums exceeds the aggregate maximum, then the number of active I/Os may reach \fBzfs_vdev_max_active\fR, in which case no further I/Os will be issued regardless of whether all per-queue minimums have been met. .sp For many physical devices, throughput increases with the number of concurrent operations, but latency typically suffers. Further, physical devices typically have a limit at which more concurrent operations have no effect on throughput or can actually cause it to decrease. .sp The scheduler selects the next operation to issue by first looking for an I/O class whose minimum has not been satisfied. Once all are satisfied and the aggregate maximum has not been hit, the scheduler looks for classes whose maximum has not been satisfied. Iteration through the I/O classes is done in the order specified above. No further operations are issued if the aggregate maximum number of concurrent operations has been hit or if there are no operations queued for an I/O class that has not hit its maximum. Every time an I/O is queued or an operation completes, the I/O scheduler looks for new operations to issue. .sp In general, smaller max_active's will lead to lower latency of synchronous operations. Larger max_active's may lead to higher overall throughput, depending on underlying storage. .sp The ratio of the queues' max_actives determines the balance of performance between reads, writes, and scrubs. E.g., increasing \fBzfs_vdev_scrub_max_active\fR will cause the scrub or resilver to complete more quickly, but reads and writes to have higher latency and lower throughput. .sp All I/O classes have a fixed maximum number of outstanding operations except for the async write class. Asynchronous writes represent the data that is committed to stable storage during the syncing stage for transaction groups. Transaction groups enter the syncing state periodically so the number of queued async writes will quickly burst up and then bleed down to zero. Rather than servicing them as quickly as possible, the I/O scheduler changes the maximum number of active async write I/Os according to the amount of dirty data in the pool. Since both throughput and latency typically increase with the number of concurrent operations issued to physical devices, reducing the burstiness in the number of concurrent operations also stabilizes the response time of operations from other -- and in particular synchronous -- queues. In broad strokes, the I/O scheduler will issue more concurrent operations from the async write queue as there's more dirty data in the pool. .sp Async Writes .sp The number of concurrent operations issued for the async write I/O class follows a piece-wise linear function defined by a few adjustable points. .nf | o---------| <-- zfs_vdev_async_write_max_active ^ | /^ | | | / | | active | / | | I/O | / | | count | / | | | / | | |-------o | | <-- zfs_vdev_async_write_min_active 0|_______^______|_________| 0% | | 100% of zfs_dirty_data_max | | | `-- zfs_vdev_async_write_active_max_dirty_percent `--------- zfs_vdev_async_write_active_min_dirty_percent .fi Until the amount of dirty data exceeds a minimum percentage of the dirty data allowed in the pool, the I/O scheduler will limit the number of concurrent operations to the minimum. As that threshold is crossed, the number of concurrent operations issued increases linearly to the maximum at the specified maximum percentage of the dirty data allowed in the pool. .sp Ideally, the amount of dirty data on a busy pool will stay in the sloped part of the function between \fBzfs_vdev_async_write_active_min_dirty_percent\fR and \fBzfs_vdev_async_write_active_max_dirty_percent\fR. If it exceeds the maximum percentage, this indicates that the rate of incoming data is greater than the rate that the backend storage can handle. In this case, we must further throttle incoming writes, as described in the next section. .SH ZFS TRANSACTION DELAY We delay transactions when we've determined that the backend storage isn't able to accommodate the rate of incoming writes. .sp If there is already a transaction waiting, we delay relative to when that transaction will finish waiting. This way the calculated delay time is independent of the number of threads concurrently executing transactions. .sp If we are the only waiter, wait relative to when the transaction started, rather than the current time. This credits the transaction for "time already served", e.g. reading indirect blocks. .sp The minimum time for a transaction to take is calculated as: .nf min_time = zfs_delay_scale * (dirty - min) / (max - dirty) min_time is then capped at 100 milliseconds. .fi .sp The delay has two degrees of freedom that can be adjusted via tunables. The percentage of dirty data at which we start to delay is defined by \fBzfs_delay_min_dirty_percent\fR. This should typically be at or above \fBzfs_vdev_async_write_active_max_dirty_percent\fR so that we only start to delay after writing at full speed has failed to keep up with the incoming write rate. The scale of the curve is defined by \fBzfs_delay_scale\fR. Roughly speaking, this variable determines the amount of delay at the midpoint of the curve. .sp .nf delay 10ms +-------------------------------------------------------------*+ | *| 9ms + *+ | *| 8ms + *+ | * | 7ms + * + | * | 6ms + * + | * | 5ms + * + | * | 4ms + * + | * | 3ms + * + | * | 2ms + (midpoint) * + | | ** | 1ms + v *** + | zfs_delay_scale ----------> ******** | 0 +-------------------------------------*********----------------+ 0% <- zfs_dirty_data_max -> 100% .fi .sp Note that since the delay is added to the outstanding time remaining on the most recent transaction, the delay is effectively the inverse of IOPS. Here the midpoint of 500us translates to 2000 IOPS. The shape of the curve was chosen such that small changes in the amount of accumulated dirty data in the first 3/4 of the curve yield relatively small differences in the amount of delay. .sp The effects can be easier to understand when the amount of delay is represented on a log scale: .sp .nf delay 100ms +-------------------------------------------------------------++ + + | | + *+ 10ms + *+ + ** + | (midpoint) ** | + | ** + 1ms + v **** + + zfs_delay_scale ----------> ***** + | **** | + **** + 100us + ** + + * + | * | + * + 10us + * + + + | | + + +--------------------------------------------------------------+ 0% <- zfs_dirty_data_max -> 100% .fi .sp Note here that only as the amount of dirty data approaches its limit does the delay start to increase rapidly. The goal of a properly tuned system should be to keep the amount of dirty data out of that range by first ensuring that the appropriate limits are set for the I/O scheduler to reach optimal throughput on the backend storage, and then by changing the value of \fBzfs_delay_scale\fR to increase the steepness of the curve. diff --git a/module/zcommon/zpool_prop.c b/module/zcommon/zpool_prop.c index 1173fc0c9cf2..6775c09d3473 100644 --- a/module/zcommon/zpool_prop.c +++ b/module/zcommon/zpool_prop.c @@ -1,252 +1,254 @@ /* * 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) 2007, 2010, Oracle and/or its affiliates. All rights reserved. * Copyright 2011 Nexenta Systems, Inc. All rights reserved. * Copyright (c) 2012 by Delphix. All rights reserved. */ #include #include #include #include #include #include "zfs_prop.h" #if defined(_KERNEL) #include #else #include #include #include #endif static zprop_desc_t zpool_prop_table[ZPOOL_NUM_PROPS]; zprop_desc_t * zpool_prop_get_table(void) { return (zpool_prop_table); } void zpool_prop_init(void) { static zprop_index_t boolean_table[] = { { "off", 0}, { "on", 1}, { NULL } }; static zprop_index_t failuremode_table[] = { { "wait", ZIO_FAILURE_MODE_WAIT }, { "continue", ZIO_FAILURE_MODE_CONTINUE }, { "panic", ZIO_FAILURE_MODE_PANIC }, { NULL } }; /* string properties */ zprop_register_string(ZPOOL_PROP_ALTROOT, "altroot", NULL, PROP_DEFAULT, ZFS_TYPE_POOL, "", "ALTROOT"); zprop_register_string(ZPOOL_PROP_BOOTFS, "bootfs", NULL, PROP_DEFAULT, ZFS_TYPE_POOL, "", "BOOTFS"); zprop_register_string(ZPOOL_PROP_CACHEFILE, "cachefile", NULL, PROP_DEFAULT, ZFS_TYPE_POOL, " | none", "CACHEFILE"); zprop_register_string(ZPOOL_PROP_COMMENT, "comment", NULL, PROP_DEFAULT, ZFS_TYPE_POOL, "", "COMMENT"); /* readonly number properties */ zprop_register_number(ZPOOL_PROP_SIZE, "size", 0, PROP_READONLY, ZFS_TYPE_POOL, "", "SIZE"); zprop_register_number(ZPOOL_PROP_FREE, "free", 0, PROP_READONLY, ZFS_TYPE_POOL, "", "FREE"); zprop_register_number(ZPOOL_PROP_FREEING, "freeing", 0, PROP_READONLY, ZFS_TYPE_POOL, "", "FREEING"); + zprop_register_number(ZPOOL_PROP_LEAKED, "leaked", 0, PROP_READONLY, + ZFS_TYPE_POOL, "", "LEAKED"); zprop_register_number(ZPOOL_PROP_ALLOCATED, "allocated", 0, PROP_READONLY, ZFS_TYPE_POOL, "", "ALLOC"); zprop_register_number(ZPOOL_PROP_EXPANDSZ, "expandsize", 0, PROP_READONLY, ZFS_TYPE_POOL, "", "EXPANDSZ"); zprop_register_number(ZPOOL_PROP_CAPACITY, "capacity", 0, PROP_READONLY, ZFS_TYPE_POOL, "", "CAP"); zprop_register_number(ZPOOL_PROP_GUID, "guid", 0, PROP_READONLY, ZFS_TYPE_POOL, "", "GUID"); zprop_register_number(ZPOOL_PROP_HEALTH, "health", 0, PROP_READONLY, ZFS_TYPE_POOL, "", "HEALTH"); zprop_register_number(ZPOOL_PROP_DEDUPRATIO, "dedupratio", 0, PROP_READONLY, ZFS_TYPE_POOL, "<1.00x or higher if deduped>", "DEDUP"); /* readonly onetime number properties */ zprop_register_number(ZPOOL_PROP_ASHIFT, "ashift", 0, PROP_ONETIME, ZFS_TYPE_POOL, "", "ASHIFT"); /* default number properties */ zprop_register_number(ZPOOL_PROP_VERSION, "version", SPA_VERSION, PROP_DEFAULT, ZFS_TYPE_POOL, "", "VERSION"); zprop_register_number(ZPOOL_PROP_DEDUPDITTO, "dedupditto", 0, PROP_DEFAULT, ZFS_TYPE_POOL, "", "DEDUPDITTO"); /* default index (boolean) properties */ zprop_register_index(ZPOOL_PROP_DELEGATION, "delegation", 1, PROP_DEFAULT, ZFS_TYPE_POOL, "on | off", "DELEGATION", boolean_table); zprop_register_index(ZPOOL_PROP_AUTOREPLACE, "autoreplace", 0, PROP_DEFAULT, ZFS_TYPE_POOL, "on | off", "REPLACE", boolean_table); zprop_register_index(ZPOOL_PROP_LISTSNAPS, "listsnapshots", 0, PROP_DEFAULT, ZFS_TYPE_POOL, "on | off", "LISTSNAPS", boolean_table); zprop_register_index(ZPOOL_PROP_AUTOEXPAND, "autoexpand", 0, PROP_DEFAULT, ZFS_TYPE_POOL, "on | off", "EXPAND", boolean_table); zprop_register_index(ZPOOL_PROP_READONLY, "readonly", 0, PROP_DEFAULT, ZFS_TYPE_POOL, "on | off", "RDONLY", boolean_table); /* default index properties */ zprop_register_index(ZPOOL_PROP_FAILUREMODE, "failmode", ZIO_FAILURE_MODE_WAIT, PROP_DEFAULT, ZFS_TYPE_POOL, "wait | continue | panic", "FAILMODE", failuremode_table); /* hidden properties */ zprop_register_hidden(ZPOOL_PROP_NAME, "name", PROP_TYPE_STRING, PROP_READONLY, ZFS_TYPE_POOL, "NAME"); } /* * Given a property name and its type, returns the corresponding property ID. */ zpool_prop_t zpool_name_to_prop(const char *propname) { return (zprop_name_to_prop(propname, ZFS_TYPE_POOL)); } /* * Given a pool property ID, returns the corresponding name. * Assuming the pool propety ID is valid. */ const char * zpool_prop_to_name(zpool_prop_t prop) { return (zpool_prop_table[prop].pd_name); } zprop_type_t zpool_prop_get_type(zpool_prop_t prop) { return (zpool_prop_table[prop].pd_proptype); } boolean_t zpool_prop_readonly(zpool_prop_t prop) { return (zpool_prop_table[prop].pd_attr == PROP_READONLY); } const char * zpool_prop_default_string(zpool_prop_t prop) { return (zpool_prop_table[prop].pd_strdefault); } uint64_t zpool_prop_default_numeric(zpool_prop_t prop) { return (zpool_prop_table[prop].pd_numdefault); } /* * Returns true if this is a valid feature@ property. */ boolean_t zpool_prop_feature(const char *name) { static const char *prefix = "feature@"; return (strncmp(name, prefix, strlen(prefix)) == 0); } /* * Returns true if this is a valid unsupported@ property. */ boolean_t zpool_prop_unsupported(const char *name) { static const char *prefix = "unsupported@"; return (strncmp(name, prefix, strlen(prefix)) == 0); } int zpool_prop_string_to_index(zpool_prop_t prop, const char *string, uint64_t *index) { return (zprop_string_to_index(prop, string, index, ZFS_TYPE_POOL)); } int zpool_prop_index_to_string(zpool_prop_t prop, uint64_t index, const char **string) { return (zprop_index_to_string(prop, index, string, ZFS_TYPE_POOL)); } uint64_t zpool_prop_random_value(zpool_prop_t prop, uint64_t seed) { return (zprop_random_value(prop, seed, ZFS_TYPE_POOL)); } #ifndef _KERNEL const char * zpool_prop_values(zpool_prop_t prop) { return (zpool_prop_table[prop].pd_values); } const char * zpool_prop_column_name(zpool_prop_t prop) { return (zpool_prop_table[prop].pd_colname); } boolean_t zpool_prop_align_right(zpool_prop_t prop) { return (zpool_prop_table[prop].pd_rightalign); } #endif #if defined(_KERNEL) && defined(HAVE_SPL) /* zpool property functions */ EXPORT_SYMBOL(zpool_prop_init); EXPORT_SYMBOL(zpool_prop_get_type); EXPORT_SYMBOL(zpool_prop_get_table); /* Pool property functions shared between libzfs and kernel. */ EXPORT_SYMBOL(zpool_name_to_prop); EXPORT_SYMBOL(zpool_prop_to_name); EXPORT_SYMBOL(zpool_prop_default_string); EXPORT_SYMBOL(zpool_prop_default_numeric); EXPORT_SYMBOL(zpool_prop_readonly); EXPORT_SYMBOL(zpool_prop_feature); EXPORT_SYMBOL(zpool_prop_unsupported); EXPORT_SYMBOL(zpool_prop_index_to_string); EXPORT_SYMBOL(zpool_prop_string_to_index); #endif diff --git a/module/zfs/bptree.c b/module/zfs/bptree.c index 83f365864dec..cbe8d1caaed4 100644 --- a/module/zfs/bptree.c +++ b/module/zfs/bptree.c @@ -1,236 +1,301 @@ /* * 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) 2013 by Delphix. All rights reserved. */ #include #include #include #include #include #include #include #include #include #include #include #include /* * A bptree is a queue of root block pointers from destroyed datasets. When a * dataset is destroyed its root block pointer is put on the end of the pool's * bptree queue so the dataset's blocks can be freed asynchronously by * dsl_scan_sync. This allows the delete operation to finish without traversing * all the dataset's blocks. * * Note that while bt_begin and bt_end are only ever incremented in this code, * they are effectively reset to 0 every time the entire bptree is freed because * the bptree's object is destroyed and re-created. */ struct bptree_args { bptree_phys_t *ba_phys; /* data in bonus buffer, dirtied if freeing */ boolean_t ba_free; /* true if freeing during traversal */ bptree_itor_t *ba_func; /* function to call for each blockpointer */ void *ba_arg; /* caller supplied argument to ba_func */ dmu_tx_t *ba_tx; /* caller supplied tx, NULL if not freeing */ } bptree_args_t; uint64_t bptree_alloc(objset_t *os, dmu_tx_t *tx) { uint64_t obj; dmu_buf_t *db; bptree_phys_t *bt; obj = dmu_object_alloc(os, DMU_OTN_UINT64_METADATA, SPA_MAXBLOCKSIZE, DMU_OTN_UINT64_METADATA, sizeof (bptree_phys_t), tx); /* * Bonus buffer contents are already initialized to 0, but for * readability we make it explicit. */ VERIFY3U(0, ==, dmu_bonus_hold(os, obj, FTAG, &db)); dmu_buf_will_dirty(db, tx); bt = db->db_data; bt->bt_begin = 0; bt->bt_end = 0; bt->bt_bytes = 0; bt->bt_comp = 0; bt->bt_uncomp = 0; dmu_buf_rele(db, FTAG); return (obj); } int bptree_free(objset_t *os, uint64_t obj, dmu_tx_t *tx) { dmu_buf_t *db; bptree_phys_t *bt; VERIFY3U(0, ==, dmu_bonus_hold(os, obj, FTAG, &db)); bt = db->db_data; ASSERT3U(bt->bt_begin, ==, bt->bt_end); ASSERT0(bt->bt_bytes); ASSERT0(bt->bt_comp); ASSERT0(bt->bt_uncomp); dmu_buf_rele(db, FTAG); return (dmu_object_free(os, obj, tx)); } +boolean_t +bptree_is_empty(objset_t *os, uint64_t obj) +{ + dmu_buf_t *db; + bptree_phys_t *bt; + boolean_t rv; + + VERIFY0(dmu_bonus_hold(os, obj, FTAG, &db)); + bt = db->db_data; + rv = (bt->bt_begin == bt->bt_end); + dmu_buf_rele(db, FTAG); + return (rv); +} + void bptree_add(objset_t *os, uint64_t obj, blkptr_t *bp, uint64_t birth_txg, uint64_t bytes, uint64_t comp, uint64_t uncomp, dmu_tx_t *tx) { dmu_buf_t *db; bptree_phys_t *bt; - bptree_entry_phys_t bte; + bptree_entry_phys_t *bte; /* * bptree objects are in the pool mos, therefore they can only be * modified in syncing context. Furthermore, this is only modified * by the sync thread, so no locking is necessary. */ ASSERT(dmu_tx_is_syncing(tx)); VERIFY3U(0, ==, dmu_bonus_hold(os, obj, FTAG, &db)); bt = db->db_data; - bte.be_birth_txg = birth_txg; - bte.be_bp = *bp; - bzero(&bte.be_zb, sizeof (bte.be_zb)); - dmu_write(os, obj, bt->bt_end * sizeof (bte), sizeof (bte), &bte, tx); + bte = kmem_zalloc(sizeof (*bte), KM_PUSHPAGE); + bte->be_birth_txg = birth_txg; + bte->be_bp = *bp; + dmu_write(os, obj, bt->bt_end * sizeof (*bte), sizeof (*bte), bte, tx); + kmem_free(bte, sizeof (*bte)); dmu_buf_will_dirty(db, tx); bt->bt_end++; bt->bt_bytes += bytes; bt->bt_comp += comp; bt->bt_uncomp += uncomp; dmu_buf_rele(db, FTAG); } /* ARGSUSED */ static int bptree_visit_cb(spa_t *spa, zilog_t *zilog, const blkptr_t *bp, const zbookmark_t *zb, const dnode_phys_t *dnp, void *arg) { int err; struct bptree_args *ba = arg; if (BP_IS_HOLE(bp)) return (0); err = ba->ba_func(ba->ba_arg, bp, ba->ba_tx); if (err == 0 && ba->ba_free) { ba->ba_phys->bt_bytes -= bp_get_dsize_sync(spa, bp); ba->ba_phys->bt_comp -= BP_GET_PSIZE(bp); ba->ba_phys->bt_uncomp -= BP_GET_UCSIZE(bp); } return (err); } +/* + * If "free" is set: + * - It is assumed that "func" will be freeing the block pointers. + * - If "func" returns nonzero, the bookmark will be remembered and + * iteration will be restarted from this point on next invocation. + * - If an i/o error is encountered (e.g. "func" returns EIO or ECKSUM), + * bptree_iterate will remember the bookmark, continue traversing + * any additional entries, and return 0. + * + * If "free" is not set, traversal will stop and return an error if + * an i/o error is encountered. + * + * In either case, if zfs_free_leak_on_eio is set, i/o errors will be + * ignored and traversal will continue (i.e. TRAVERSE_HARD will be passed to + * traverse_dataset_destroyed()). + */ int bptree_iterate(objset_t *os, uint64_t obj, boolean_t free, bptree_itor_t func, void *arg, dmu_tx_t *tx) { + boolean_t ioerr = B_FALSE; int err; uint64_t i; dmu_buf_t *db; struct bptree_args ba; ASSERT(!free || dmu_tx_is_syncing(tx)); err = dmu_bonus_hold(os, obj, FTAG, &db); if (err != 0) return (err); if (free) dmu_buf_will_dirty(db, tx); ba.ba_phys = db->db_data; ba.ba_free = free; ba.ba_func = func; ba.ba_arg = arg; ba.ba_tx = tx; err = 0; for (i = ba.ba_phys->bt_begin; i < ba.ba_phys->bt_end; i++) { bptree_entry_phys_t bte; int flags = TRAVERSE_PREFETCH_METADATA | TRAVERSE_POST; - ASSERT(!free || i == ba.ba_phys->bt_begin); - err = dmu_read(os, obj, i * sizeof (bte), sizeof (bte), &bte, DMU_READ_NO_PREFETCH); if (err != 0) break; - if (zfs_recover) + if (zfs_free_leak_on_eio) flags |= TRAVERSE_HARD; + zfs_dbgmsg("bptree index %d: traversing from min_txg=%lld " + "bookmark %lld/%lld/%lld/%lld", + i, (longlong_t)bte.be_birth_txg, + (longlong_t)bte.be_zb.zb_objset, + (longlong_t)bte.be_zb.zb_object, + (longlong_t)bte.be_zb.zb_level, + (longlong_t)bte.be_zb.zb_blkid); err = traverse_dataset_destroyed(os->os_spa, &bte.be_bp, bte.be_birth_txg, &bte.be_zb, flags, bptree_visit_cb, &ba); if (free) { - if (err == ERESTART) { + /* + * The callback has freed the visited block pointers. + * Record our traversal progress on disk, either by + * updating this record's bookmark, or by logically + * removing this record by advancing bt_begin. + */ + if (err != 0) { /* save bookmark for future resume */ ASSERT3U(bte.be_zb.zb_objset, ==, ZB_DESTROYED_OBJSET); ASSERT0(bte.be_zb.zb_level); dmu_write(os, obj, i * sizeof (bte), sizeof (bte), &bte, tx); - break; - } - if (err != 0) { + if (err == EIO || err == ECKSUM || + err == ENXIO) { + /* + * Skip the rest of this tree and + * continue on to the next entry. + */ + err = 0; + ioerr = B_TRUE; + } else { + break; + } + } else if (ioerr) { /* - * We can not properly handle an i/o - * error, because the traversal code - * does not know how to resume from an - * arbitrary bookmark. + * This entry is finished, but there were + * i/o errors on previous entries, so we + * can't adjust bt_begin. Set this entry's + * be_birth_txg such that it will be + * treated as a no-op in future traversals. */ - zfs_panic_recover("error %u from " - "traverse_dataset_destroyed()", err); + bte.be_birth_txg = UINT64_MAX; + dmu_write(os, obj, i * sizeof (bte), + sizeof (bte), &bte, tx); } - ba.ba_phys->bt_begin++; - (void) dmu_free_range(os, obj, - i * sizeof (bte), sizeof (bte), tx); + if (!ioerr) { + ba.ba_phys->bt_begin++; + (void) dmu_free_range(os, obj, + i * sizeof (bte), sizeof (bte), tx); + } + } else if (err != 0) { + break; } } - ASSERT(!free || err != 0 || ba.ba_phys->bt_begin == ba.ba_phys->bt_end); + ASSERT(!free || err != 0 || ioerr || + ba.ba_phys->bt_begin == ba.ba_phys->bt_end); /* if all blocks are free there should be no used space */ if (ba.ba_phys->bt_begin == ba.ba_phys->bt_end) { + if (zfs_free_leak_on_eio) { + ba.ba_phys->bt_bytes = 0; + ba.ba_phys->bt_comp = 0; + ba.ba_phys->bt_uncomp = 0; + } + ASSERT0(ba.ba_phys->bt_bytes); ASSERT0(ba.ba_phys->bt_comp); ASSERT0(ba.ba_phys->bt_uncomp); } dmu_buf_rele(db, FTAG); return (err); } diff --git a/module/zfs/dmu_traverse.c b/module/zfs/dmu_traverse.c index e086e2487e8a..fe352469b012 100644 --- a/module/zfs/dmu_traverse.c +++ b/module/zfs/dmu_traverse.c @@ -1,680 +1,661 @@ /* * 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 by Delphix. All rights reserved. */ #include #include #include #include #include #include #include #include #include #include #include #include #include #include int zfs_pd_blks_max = 100; typedef struct prefetch_data { kmutex_t pd_mtx; kcondvar_t pd_cv; int pd_blks_max; int pd_blks_fetched; int pd_flags; boolean_t pd_cancel; boolean_t pd_exited; } prefetch_data_t; typedef struct traverse_data { spa_t *td_spa; uint64_t td_objset; blkptr_t *td_rootbp; uint64_t td_min_txg; zbookmark_t *td_resume; int td_flags; prefetch_data_t *td_pfd; + boolean_t td_paused; blkptr_cb_t *td_func; void *td_arg; } traverse_data_t; -#define TD_HARD(td) (td->td_flags & TRAVERSE_HARD) - static int traverse_dnode(traverse_data_t *td, const dnode_phys_t *dnp, uint64_t objset, uint64_t object); static void prefetch_dnode_metadata(traverse_data_t *td, const dnode_phys_t *, uint64_t objset, uint64_t object); static int traverse_zil_block(zilog_t *zilog, blkptr_t *bp, void *arg, uint64_t claim_txg) { traverse_data_t *td = arg; zbookmark_t zb; if (BP_IS_HOLE(bp)) return (0); if (claim_txg == 0 && bp->blk_birth >= spa_first_txg(td->td_spa)) return (0); SET_BOOKMARK(&zb, td->td_objset, ZB_ZIL_OBJECT, ZB_ZIL_LEVEL, bp->blk_cksum.zc_word[ZIL_ZC_SEQ]); (void) td->td_func(td->td_spa, zilog, bp, &zb, NULL, td->td_arg); return (0); } static int traverse_zil_record(zilog_t *zilog, lr_t *lrc, void *arg, uint64_t claim_txg) { traverse_data_t *td = arg; if (lrc->lrc_txtype == TX_WRITE) { lr_write_t *lr = (lr_write_t *)lrc; blkptr_t *bp = &lr->lr_blkptr; zbookmark_t zb; if (BP_IS_HOLE(bp)) return (0); if (claim_txg == 0 || bp->blk_birth < claim_txg) return (0); SET_BOOKMARK(&zb, td->td_objset, lr->lr_foid, ZB_ZIL_LEVEL, lr->lr_offset / BP_GET_LSIZE(bp)); (void) td->td_func(td->td_spa, zilog, bp, &zb, NULL, td->td_arg); } return (0); } static void traverse_zil(traverse_data_t *td, zil_header_t *zh) { uint64_t claim_txg = zh->zh_claim_txg; zilog_t *zilog; /* * We only want to visit blocks that have been claimed but not yet * replayed; plus, in read-only mode, blocks that are already stable. */ if (claim_txg == 0 && spa_writeable(td->td_spa)) return; zilog = zil_alloc(spa_get_dsl(td->td_spa)->dp_meta_objset, zh); (void) zil_parse(zilog, traverse_zil_block, traverse_zil_record, td, claim_txg); zil_free(zilog); } typedef enum resume_skip { RESUME_SKIP_ALL, RESUME_SKIP_NONE, RESUME_SKIP_CHILDREN } resume_skip_t; /* * Returns RESUME_SKIP_ALL if td indicates that we are resuming a traversal and * the block indicated by zb does not need to be visited at all. Returns * RESUME_SKIP_CHILDREN if we are resuming a post traversal and we reach the * resume point. This indicates that this block should be visited but not its * children (since they must have been visited in a previous traversal). * Otherwise returns RESUME_SKIP_NONE. */ static resume_skip_t resume_skip_check(traverse_data_t *td, const dnode_phys_t *dnp, const zbookmark_t *zb) { if (td->td_resume != NULL && !ZB_IS_ZERO(td->td_resume)) { /* * If we already visited this bp & everything below, * don't bother doing it again. */ if (zbookmark_is_before(dnp, zb, td->td_resume)) return (RESUME_SKIP_ALL); /* * If we found the block we're trying to resume from, zero * the bookmark out to indicate that we have resumed. */ - ASSERT3U(zb->zb_object, <=, td->td_resume->zb_object); if (bcmp(zb, td->td_resume, sizeof (*zb)) == 0) { bzero(td->td_resume, sizeof (*zb)); if (td->td_flags & TRAVERSE_POST) return (RESUME_SKIP_CHILDREN); } } return (RESUME_SKIP_NONE); } -static void -traverse_pause(traverse_data_t *td, const zbookmark_t *zb) -{ - ASSERT(td->td_resume != NULL); - ASSERT0(zb->zb_level); - bcopy(zb, td->td_resume, sizeof (*td->td_resume)); -} - static void traverse_prefetch_metadata(traverse_data_t *td, const blkptr_t *bp, const zbookmark_t *zb) { uint32_t flags = ARC_NOWAIT | ARC_PREFETCH; if (!(td->td_flags & TRAVERSE_PREFETCH_METADATA)) return; /* * If we are in the process of resuming, don't prefetch, because * some children will not be needed (and in fact may have already * been freed). */ if (td->td_resume != NULL && !ZB_IS_ZERO(td->td_resume)) return; if (BP_IS_HOLE(bp) || bp->blk_birth <= td->td_min_txg) return; if (BP_GET_LEVEL(bp) == 0 && BP_GET_TYPE(bp) != DMU_OT_DNODE) return; (void) arc_read(NULL, td->td_spa, bp, NULL, NULL, ZIO_PRIORITY_ASYNC_READ, ZIO_FLAG_CANFAIL, &flags, zb); } static int traverse_visitbp(traverse_data_t *td, const dnode_phys_t *dnp, const blkptr_t *bp, const zbookmark_t *zb) { - int err = 0, lasterr = 0; + int err = 0; arc_buf_t *buf = NULL; - boolean_t pause = B_FALSE; switch (resume_skip_check(td, dnp, zb)) { case RESUME_SKIP_ALL: return (0); case RESUME_SKIP_CHILDREN: goto post; case RESUME_SKIP_NONE: break; default: ASSERT(0); } if (bp->blk_birth == 0) { if (spa_feature_is_active(td->td_spa, SPA_FEATURE_HOLE_BIRTH)) { /* * Since this block has a birth time of 0 it must be a * hole created before the SPA_FEATURE_HOLE_BIRTH * feature was enabled. If SPA_FEATURE_HOLE_BIRTH * was enabled before the min_txg for this traveral we * know the hole must have been created before the * min_txg for this traveral, so we can skip it. If * SPA_FEATURE_HOLE_BIRTH was enabled after the min_txg * for this traveral we cannot tell if the hole was * created before or after the min_txg for this * traversal, so we cannot skip it. */ uint64_t hole_birth_enabled_txg; VERIFY(spa_feature_enabled_txg(td->td_spa, SPA_FEATURE_HOLE_BIRTH, &hole_birth_enabled_txg)); if (hole_birth_enabled_txg < td->td_min_txg) return (0); } } else if (bp->blk_birth <= td->td_min_txg) { return (0); } if (BP_IS_HOLE(bp)) { err = td->td_func(td->td_spa, NULL, bp, zb, dnp, td->td_arg); - return (err); + if (err != 0) + goto post; + return (0); } if (td->td_pfd && !td->td_pfd->pd_exited && ((td->td_pfd->pd_flags & TRAVERSE_PREFETCH_DATA) || BP_GET_TYPE(bp) == DMU_OT_DNODE || BP_GET_LEVEL(bp) > 0)) { mutex_enter(&td->td_pfd->pd_mtx); ASSERT(td->td_pfd->pd_blks_fetched >= 0); while (td->td_pfd->pd_blks_fetched == 0 && !td->td_pfd->pd_exited) cv_wait(&td->td_pfd->pd_cv, &td->td_pfd->pd_mtx); td->td_pfd->pd_blks_fetched--; cv_broadcast(&td->td_pfd->pd_cv); mutex_exit(&td->td_pfd->pd_mtx); } if (td->td_flags & TRAVERSE_PRE) { err = td->td_func(td->td_spa, NULL, bp, zb, dnp, td->td_arg); if (err == TRAVERSE_VISIT_NO_CHILDREN) return (0); - if (err == ERESTART) - pause = B_TRUE; /* handle pausing at a common point */ if (err != 0) goto post; } if (BP_GET_LEVEL(bp) > 0) { uint32_t flags = ARC_WAIT; int32_t i; int32_t epb = BP_GET_LSIZE(bp) >> SPA_BLKPTRSHIFT; zbookmark_t *czb; err = arc_read(NULL, td->td_spa, bp, arc_getbuf_func, &buf, ZIO_PRIORITY_ASYNC_READ, ZIO_FLAG_CANFAIL, &flags, zb); if (err != 0) - return (err); + goto post; czb = kmem_alloc(sizeof (zbookmark_t), KM_PUSHPAGE); for (i = 0; i < epb; i++) { SET_BOOKMARK(czb, zb->zb_objset, zb->zb_object, zb->zb_level - 1, zb->zb_blkid * epb + i); traverse_prefetch_metadata(td, &((blkptr_t *)buf->b_data)[i], czb); } /* recursively visitbp() blocks below this */ for (i = 0; i < epb; i++) { SET_BOOKMARK(czb, zb->zb_objset, zb->zb_object, zb->zb_level - 1, zb->zb_blkid * epb + i); err = traverse_visitbp(td, dnp, &((blkptr_t *)buf->b_data)[i], czb); - if (err != 0) { - if (!TD_HARD(td)) - break; - lasterr = err; - } + if (err != 0) + break; } kmem_free(czb, sizeof (zbookmark_t)); } else if (BP_GET_TYPE(bp) == DMU_OT_DNODE) { uint32_t flags = ARC_WAIT; int32_t i; int32_t epb = BP_GET_LSIZE(bp) >> DNODE_SHIFT; err = arc_read(NULL, td->td_spa, bp, arc_getbuf_func, &buf, ZIO_PRIORITY_ASYNC_READ, ZIO_FLAG_CANFAIL, &flags, zb); if (err != 0) - return (err); + goto post; dnp = buf->b_data; for (i = 0; i < epb; i++) { prefetch_dnode_metadata(td, &dnp[i], zb->zb_objset, zb->zb_blkid * epb + i); } /* recursively visitbp() blocks below this */ for (i = 0; i < epb; i++) { err = traverse_dnode(td, &dnp[i], zb->zb_objset, zb->zb_blkid * epb + i); - if (err != 0) { - if (!TD_HARD(td)) - break; - lasterr = err; - } + if (err != 0) + break; } } else if (BP_GET_TYPE(bp) == DMU_OT_OBJSET) { uint32_t flags = ARC_WAIT; objset_phys_t *osp; dnode_phys_t *dnp; err = arc_read(NULL, td->td_spa, bp, arc_getbuf_func, &buf, ZIO_PRIORITY_ASYNC_READ, ZIO_FLAG_CANFAIL, &flags, zb); if (err != 0) - return (err); + goto post; osp = buf->b_data; dnp = &osp->os_meta_dnode; prefetch_dnode_metadata(td, dnp, zb->zb_objset, DMU_META_DNODE_OBJECT); if (arc_buf_size(buf) >= sizeof (objset_phys_t)) { prefetch_dnode_metadata(td, &osp->os_groupused_dnode, zb->zb_objset, DMU_GROUPUSED_OBJECT); prefetch_dnode_metadata(td, &osp->os_userused_dnode, zb->zb_objset, DMU_USERUSED_OBJECT); } err = traverse_dnode(td, dnp, zb->zb_objset, DMU_META_DNODE_OBJECT); - if (err && TD_HARD(td)) { - lasterr = err; - err = 0; - } if (err == 0 && arc_buf_size(buf) >= sizeof (objset_phys_t)) { dnp = &osp->os_groupused_dnode; err = traverse_dnode(td, dnp, zb->zb_objset, DMU_GROUPUSED_OBJECT); } - if (err && TD_HARD(td)) { - lasterr = err; - err = 0; - } if (err == 0 && arc_buf_size(buf) >= sizeof (objset_phys_t)) { dnp = &osp->os_userused_dnode; err = traverse_dnode(td, dnp, zb->zb_objset, DMU_USERUSED_OBJECT); } } if (buf) (void) arc_buf_remove_ref(buf, &buf); post: - if (err == 0 && (td->td_flags & TRAVERSE_POST)) { + if (err == 0 && (td->td_flags & TRAVERSE_POST)) err = td->td_func(td->td_spa, NULL, bp, zb, dnp, td->td_arg); - if (err == ERESTART) - pause = B_TRUE; + + if ((td->td_flags & TRAVERSE_HARD) && (err == EIO || err == ECKSUM)) { + /* + * Ignore this disk error as requested by the HARD flag, + * and continue traversal. + */ + err = 0; } - if (pause && td->td_resume != NULL) { - ASSERT3U(err, ==, ERESTART); - ASSERT(!TD_HARD(td)); - traverse_pause(td, zb); + /* + * If we are stopping here, set td_resume. + */ + if (td->td_resume != NULL && err != 0 && !td->td_paused) { + td->td_resume->zb_objset = zb->zb_objset; + td->td_resume->zb_object = zb->zb_object; + td->td_resume->zb_level = 0; + /* + * If we have stopped on an indirect block (e.g. due to + * i/o error), we have not visited anything below it. + * Set the bookmark to the first level-0 block that we need + * to visit. This way, the resuming code does not need to + * deal with resuming from indirect blocks. + */ + td->td_resume->zb_blkid = zb->zb_blkid << + (zb->zb_level * (dnp->dn_indblkshift - SPA_BLKPTRSHIFT)); + td->td_paused = B_TRUE; } - return (err != 0 ? err : lasterr); + return (err); } static void prefetch_dnode_metadata(traverse_data_t *td, const dnode_phys_t *dnp, uint64_t objset, uint64_t object) { int j; zbookmark_t czb; for (j = 0; j < dnp->dn_nblkptr; j++) { SET_BOOKMARK(&czb, objset, object, dnp->dn_nlevels - 1, j); traverse_prefetch_metadata(td, &dnp->dn_blkptr[j], &czb); } if (dnp->dn_flags & DNODE_FLAG_SPILL_BLKPTR) { SET_BOOKMARK(&czb, objset, object, 0, DMU_SPILL_BLKID); traverse_prefetch_metadata(td, &dnp->dn_spill, &czb); } } static int traverse_dnode(traverse_data_t *td, const dnode_phys_t *dnp, uint64_t objset, uint64_t object) { - int j, err = 0, lasterr = 0; + int j, err = 0; zbookmark_t czb; for (j = 0; j < dnp->dn_nblkptr; j++) { SET_BOOKMARK(&czb, objset, object, dnp->dn_nlevels - 1, j); err = traverse_visitbp(td, dnp, &dnp->dn_blkptr[j], &czb); - if (err != 0) { - if (!TD_HARD(td)) - break; - lasterr = err; - } + if (err != 0) + break; } if (dnp->dn_flags & DNODE_FLAG_SPILL_BLKPTR) { SET_BOOKMARK(&czb, objset, object, 0, DMU_SPILL_BLKID); err = traverse_visitbp(td, dnp, &dnp->dn_spill, &czb); - if (err != 0) { - if (!TD_HARD(td)) - return (err); - lasterr = err; - } } - return (err != 0 ? err : lasterr); + return (err); } /* ARGSUSED */ static int traverse_prefetcher(spa_t *spa, zilog_t *zilog, const blkptr_t *bp, const zbookmark_t *zb, const dnode_phys_t *dnp, void *arg) { prefetch_data_t *pfd = arg; uint32_t aflags = ARC_NOWAIT | ARC_PREFETCH; ASSERT(pfd->pd_blks_fetched >= 0); if (pfd->pd_cancel) return (SET_ERROR(EINTR)); if (BP_IS_HOLE(bp) || BP_IS_EMBEDDED(bp) || !((pfd->pd_flags & TRAVERSE_PREFETCH_DATA) || BP_GET_TYPE(bp) == DMU_OT_DNODE || BP_GET_LEVEL(bp) > 0) || BP_GET_TYPE(bp) == DMU_OT_INTENT_LOG) return (0); mutex_enter(&pfd->pd_mtx); while (!pfd->pd_cancel && pfd->pd_blks_fetched >= pfd->pd_blks_max) cv_wait(&pfd->pd_cv, &pfd->pd_mtx); pfd->pd_blks_fetched++; cv_broadcast(&pfd->pd_cv); mutex_exit(&pfd->pd_mtx); (void) arc_read(NULL, spa, bp, NULL, NULL, ZIO_PRIORITY_ASYNC_READ, ZIO_FLAG_CANFAIL | ZIO_FLAG_SPECULATIVE, &aflags, zb); return (0); } static void traverse_prefetch_thread(void *arg) { traverse_data_t *td_main = arg; traverse_data_t td = *td_main; zbookmark_t czb; td.td_func = traverse_prefetcher; td.td_arg = td_main->td_pfd; td.td_pfd = NULL; SET_BOOKMARK(&czb, td.td_objset, ZB_ROOT_OBJECT, ZB_ROOT_LEVEL, ZB_ROOT_BLKID); (void) traverse_visitbp(&td, NULL, td.td_rootbp, &czb); mutex_enter(&td_main->td_pfd->pd_mtx); td_main->td_pfd->pd_exited = B_TRUE; cv_broadcast(&td_main->td_pfd->pd_cv); mutex_exit(&td_main->td_pfd->pd_mtx); } /* * NB: dataset must not be changing on-disk (eg, is a snapshot or we are * in syncing context). */ static int traverse_impl(spa_t *spa, dsl_dataset_t *ds, uint64_t objset, blkptr_t *rootbp, uint64_t txg_start, zbookmark_t *resume, int flags, blkptr_cb_t func, void *arg) { traverse_data_t *td; prefetch_data_t *pd; zbookmark_t *czb; int err; ASSERT(ds == NULL || objset == ds->ds_object); ASSERT(!(flags & TRAVERSE_PRE) || !(flags & TRAVERSE_POST)); /* * The data prefetching mechanism (the prefetch thread) is incompatible * with resuming from a bookmark. */ ASSERT(resume == NULL || !(flags & TRAVERSE_PREFETCH_DATA)); td = kmem_alloc(sizeof (traverse_data_t), KM_PUSHPAGE); pd = kmem_zalloc(sizeof (prefetch_data_t), KM_PUSHPAGE); czb = kmem_alloc(sizeof (zbookmark_t), KM_PUSHPAGE); td->td_spa = spa; td->td_objset = objset; td->td_rootbp = rootbp; td->td_min_txg = txg_start; td->td_resume = resume; td->td_func = func; td->td_arg = arg; td->td_pfd = pd; td->td_flags = flags; + td->td_paused = B_FALSE; pd->pd_blks_max = zfs_pd_blks_max; pd->pd_flags = flags; mutex_init(&pd->pd_mtx, NULL, MUTEX_DEFAULT, NULL); cv_init(&pd->pd_cv, NULL, CV_DEFAULT, NULL); SET_BOOKMARK(czb, td->td_objset, ZB_ROOT_OBJECT, ZB_ROOT_LEVEL, ZB_ROOT_BLKID); /* See comment on ZIL traversal in dsl_scan_visitds. */ if (ds != NULL && !dsl_dataset_is_snapshot(ds) && !BP_IS_HOLE(rootbp)) { uint32_t flags = ARC_WAIT; objset_phys_t *osp; arc_buf_t *buf; err = arc_read(NULL, td->td_spa, rootbp, arc_getbuf_func, &buf, ZIO_PRIORITY_ASYNC_READ, ZIO_FLAG_CANFAIL, &flags, czb); if (err != 0) return (err); osp = buf->b_data; traverse_zil(td, &osp->os_zil_header); (void) arc_buf_remove_ref(buf, &buf); } if (!(flags & TRAVERSE_PREFETCH_DATA) || 0 == taskq_dispatch(system_taskq, traverse_prefetch_thread, td, TQ_NOQUEUE)) pd->pd_exited = B_TRUE; err = traverse_visitbp(td, NULL, rootbp, czb); mutex_enter(&pd->pd_mtx); pd->pd_cancel = B_TRUE; cv_broadcast(&pd->pd_cv); while (!pd->pd_exited) cv_wait(&pd->pd_cv, &pd->pd_mtx); mutex_exit(&pd->pd_mtx); mutex_destroy(&pd->pd_mtx); cv_destroy(&pd->pd_cv); kmem_free(czb, sizeof (zbookmark_t)); kmem_free(pd, sizeof (struct prefetch_data)); kmem_free(td, sizeof (struct traverse_data)); return (err); } /* * NB: dataset must not be changing on-disk (eg, is a snapshot or we are * in syncing context). */ int traverse_dataset(dsl_dataset_t *ds, uint64_t txg_start, int flags, blkptr_cb_t func, void *arg) { return (traverse_impl(ds->ds_dir->dd_pool->dp_spa, ds, ds->ds_object, &ds->ds_phys->ds_bp, txg_start, NULL, flags, func, arg)); } int traverse_dataset_destroyed(spa_t *spa, blkptr_t *blkptr, uint64_t txg_start, zbookmark_t *resume, int flags, blkptr_cb_t func, void *arg) { return (traverse_impl(spa, NULL, ZB_DESTROYED_OBJSET, blkptr, txg_start, resume, flags, func, arg)); } /* * NB: pool must not be changing on-disk (eg, from zdb or sync context). */ int traverse_pool(spa_t *spa, uint64_t txg_start, int flags, blkptr_cb_t func, void *arg) { - int err, lasterr = 0; + int err; uint64_t obj; dsl_pool_t *dp = spa_get_dsl(spa); objset_t *mos = dp->dp_meta_objset; boolean_t hard = (flags & TRAVERSE_HARD); /* visit the MOS */ err = traverse_impl(spa, NULL, 0, spa_get_rootblkptr(spa), txg_start, NULL, flags, func, arg); if (err != 0) return (err); /* visit each dataset */ - for (obj = 1; err == 0 || (err != ESRCH && hard); + for (obj = 1; err == 0; err = dmu_object_next(mos, &obj, FALSE, txg_start)) { dmu_object_info_t doi; err = dmu_object_info(mos, obj, &doi); if (err != 0) { - if (!hard) - return (err); - lasterr = err; - continue; + if (hard) + continue; + break; } if (doi.doi_bonus_type == DMU_OT_DSL_DATASET) { dsl_dataset_t *ds; uint64_t txg = txg_start; dsl_pool_config_enter(dp, FTAG); err = dsl_dataset_hold_obj(dp, obj, FTAG, &ds); dsl_pool_config_exit(dp, FTAG); if (err != 0) { - if (!hard) - return (err); - lasterr = err; - continue; + if (hard) + continue; + break; } if (ds->ds_phys->ds_prev_snap_txg > txg) txg = ds->ds_phys->ds_prev_snap_txg; err = traverse_dataset(ds, txg, flags, func, arg); dsl_dataset_rele(ds, FTAG); - if (err != 0) { - if (!hard) - return (err); - lasterr = err; - } + if (err != 0) + break; } } if (err == ESRCH) err = 0; - return (err != 0 ? err : lasterr); + return (err); } #if defined(_KERNEL) && defined(HAVE_SPL) EXPORT_SYMBOL(traverse_dataset); EXPORT_SYMBOL(traverse_pool); module_param(zfs_pd_blks_max, int, 0644); MODULE_PARM_DESC(zfs_pd_blks_max, "Max number of blocks to prefetch"); #endif diff --git a/module/zfs/dsl_pool.c b/module/zfs/dsl_pool.c index 91476cc0b6b7..f1ab29c6214c 100644 --- a/module/zfs/dsl_pool.c +++ b/module/zfs/dsl_pool.c @@ -1,1075 +1,1084 @@ /* * 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 by Delphix. All rights reserved. * Copyright (c) 2013 Steven Hartland. All rights reserved. */ #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include /* * ZFS Write Throttle * ------------------ * * ZFS must limit the rate of incoming writes to the rate at which it is able * to sync data modifications to the backend storage. Throttling by too much * creates an artificial limit; throttling by too little can only be sustained * for short periods and would lead to highly lumpy performance. On a per-pool * basis, ZFS tracks the amount of modified (dirty) data. As operations change * data, the amount of dirty data increases; as ZFS syncs out data, the amount * of dirty data decreases. When the amount of dirty data exceeds a * predetermined threshold further modifications are blocked until the amount * of dirty data decreases (as data is synced out). * * The limit on dirty data is tunable, and should be adjusted according to * both the IO capacity and available memory of the system. The larger the * window, the more ZFS is able to aggregate and amortize metadata (and data) * changes. However, memory is a limited resource, and allowing for more dirty * data comes at the cost of keeping other useful data in memory (for example * ZFS data cached by the ARC). * * Implementation * * As buffers are modified dsl_pool_willuse_space() increments both the per- * txg (dp_dirty_pertxg[]) and poolwide (dp_dirty_total) accounting of * dirty space used; dsl_pool_dirty_space() decrements those values as data * is synced out from dsl_pool_sync(). While only the poolwide value is * relevant, the per-txg value is useful for debugging. The tunable * zfs_dirty_data_max determines the dirty space limit. Once that value is * exceeded, new writes are halted until space frees up. * * The zfs_dirty_data_sync tunable dictates the threshold at which we * ensure that there is a txg syncing (see the comment in txg.c for a full * description of transaction group stages). * * The IO scheduler uses both the dirty space limit and current amount of * dirty data as inputs. Those values affect the number of concurrent IOs ZFS * issues. See the comment in vdev_queue.c for details of the IO scheduler. * * The delay is also calculated based on the amount of dirty data. See the * comment above dmu_tx_delay() for details. */ /* * zfs_dirty_data_max will be set to zfs_dirty_data_max_percent% of all memory, * capped at zfs_dirty_data_max_max. It can also be overridden with a module * parameter. */ unsigned long zfs_dirty_data_max = 0; unsigned long zfs_dirty_data_max_max = 0; int zfs_dirty_data_max_percent = 10; int zfs_dirty_data_max_max_percent = 25; /* * If there is at least this much dirty data, push out a txg. */ unsigned long zfs_dirty_data_sync = 64 * 1024 * 1024; /* * Once there is this amount of dirty data, the dmu_tx_delay() will kick in * and delay each transaction. * This value should be >= zfs_vdev_async_write_active_max_dirty_percent. */ int zfs_delay_min_dirty_percent = 60; /* * This controls how quickly the delay approaches infinity. * Larger values cause it to delay more for a given amount of dirty data. * Therefore larger values will cause there to be less dirty data for a * given throughput. * * For the smoothest delay, this value should be about 1 billion divided * by the maximum number of operations per second. This will smoothly * handle between 10x and 1/10th this number. * * Note: zfs_delay_scale * zfs_dirty_data_max must be < 2^64, due to the * multiply in dmu_tx_delay(). */ unsigned long zfs_delay_scale = 1000 * 1000 * 1000 / 2000; hrtime_t zfs_throttle_delay = MSEC2NSEC(10); hrtime_t zfs_throttle_resolution = MSEC2NSEC(10); int dsl_pool_open_special_dir(dsl_pool_t *dp, const char *name, dsl_dir_t **ddp) { uint64_t obj; int err; err = zap_lookup(dp->dp_meta_objset, dp->dp_root_dir->dd_phys->dd_child_dir_zapobj, name, sizeof (obj), 1, &obj); if (err) return (err); return (dsl_dir_hold_obj(dp, obj, name, dp, ddp)); } static dsl_pool_t * dsl_pool_open_impl(spa_t *spa, uint64_t txg) { dsl_pool_t *dp; blkptr_t *bp = spa_get_rootblkptr(spa); dp = kmem_zalloc(sizeof (dsl_pool_t), KM_SLEEP); dp->dp_spa = spa; dp->dp_meta_rootbp = *bp; rrw_init(&dp->dp_config_rwlock, B_TRUE); txg_init(dp, txg); txg_list_create(&dp->dp_dirty_datasets, offsetof(dsl_dataset_t, ds_dirty_link)); txg_list_create(&dp->dp_dirty_zilogs, offsetof(zilog_t, zl_dirty_link)); txg_list_create(&dp->dp_dirty_dirs, offsetof(dsl_dir_t, dd_dirty_link)); txg_list_create(&dp->dp_sync_tasks, offsetof(dsl_sync_task_t, dst_node)); mutex_init(&dp->dp_lock, NULL, MUTEX_DEFAULT, NULL); cv_init(&dp->dp_spaceavail_cv, NULL, CV_DEFAULT, NULL); dp->dp_iput_taskq = taskq_create("zfs_iput_taskq", 1, minclsyspri, 1, 4, 0); return (dp); } int dsl_pool_init(spa_t *spa, uint64_t txg, dsl_pool_t **dpp) { int err; dsl_pool_t *dp = dsl_pool_open_impl(spa, txg); err = dmu_objset_open_impl(spa, NULL, &dp->dp_meta_rootbp, &dp->dp_meta_objset); if (err != 0) dsl_pool_close(dp); else *dpp = dp; return (err); } int dsl_pool_open(dsl_pool_t *dp) { int err; dsl_dir_t *dd; dsl_dataset_t *ds; uint64_t obj; rrw_enter(&dp->dp_config_rwlock, RW_WRITER, FTAG); err = zap_lookup(dp->dp_meta_objset, DMU_POOL_DIRECTORY_OBJECT, DMU_POOL_ROOT_DATASET, sizeof (uint64_t), 1, &dp->dp_root_dir_obj); if (err) goto out; err = dsl_dir_hold_obj(dp, dp->dp_root_dir_obj, NULL, dp, &dp->dp_root_dir); if (err) goto out; err = dsl_pool_open_special_dir(dp, MOS_DIR_NAME, &dp->dp_mos_dir); if (err) goto out; if (spa_version(dp->dp_spa) >= SPA_VERSION_ORIGIN) { err = dsl_pool_open_special_dir(dp, ORIGIN_DIR_NAME, &dd); if (err) goto out; err = dsl_dataset_hold_obj(dp, dd->dd_phys->dd_head_dataset_obj, FTAG, &ds); if (err == 0) { err = dsl_dataset_hold_obj(dp, ds->ds_phys->ds_prev_snap_obj, dp, &dp->dp_origin_snap); dsl_dataset_rele(ds, FTAG); } dsl_dir_rele(dd, dp); if (err) goto out; } if (spa_version(dp->dp_spa) >= SPA_VERSION_DEADLISTS) { err = dsl_pool_open_special_dir(dp, FREE_DIR_NAME, &dp->dp_free_dir); if (err) goto out; err = zap_lookup(dp->dp_meta_objset, DMU_POOL_DIRECTORY_OBJECT, DMU_POOL_FREE_BPOBJ, sizeof (uint64_t), 1, &obj); if (err) goto out; VERIFY0(bpobj_open(&dp->dp_free_bpobj, dp->dp_meta_objset, obj)); } + /* + * Note: errors ignored, because the leak dir will not exist if we + * have not encountered a leak yet. + */ + (void) dsl_pool_open_special_dir(dp, LEAK_DIR_NAME, + &dp->dp_leak_dir); + if (spa_feature_is_active(dp->dp_spa, SPA_FEATURE_ASYNC_DESTROY)) { err = zap_lookup(dp->dp_meta_objset, DMU_POOL_DIRECTORY_OBJECT, DMU_POOL_BPTREE_OBJ, sizeof (uint64_t), 1, &dp->dp_bptree_obj); if (err != 0) goto out; } if (spa_feature_is_active(dp->dp_spa, SPA_FEATURE_EMPTY_BPOBJ)) { err = zap_lookup(dp->dp_meta_objset, DMU_POOL_DIRECTORY_OBJECT, DMU_POOL_EMPTY_BPOBJ, sizeof (uint64_t), 1, &dp->dp_empty_bpobj); if (err != 0) goto out; } err = zap_lookup(dp->dp_meta_objset, DMU_POOL_DIRECTORY_OBJECT, DMU_POOL_TMP_USERREFS, sizeof (uint64_t), 1, &dp->dp_tmp_userrefs_obj); if (err == ENOENT) err = 0; if (err) goto out; err = dsl_scan_init(dp, dp->dp_tx.tx_open_txg); out: rrw_exit(&dp->dp_config_rwlock, FTAG); return (err); } void dsl_pool_close(dsl_pool_t *dp) { /* * Drop our references from dsl_pool_open(). * * Since we held the origin_snap from "syncing" context (which * includes pool-opening context), it actually only got a "ref" * and not a hold, so just drop that here. */ if (dp->dp_origin_snap) dsl_dataset_rele(dp->dp_origin_snap, dp); if (dp->dp_mos_dir) dsl_dir_rele(dp->dp_mos_dir, dp); if (dp->dp_free_dir) dsl_dir_rele(dp->dp_free_dir, dp); + if (dp->dp_leak_dir) + dsl_dir_rele(dp->dp_leak_dir, dp); if (dp->dp_root_dir) dsl_dir_rele(dp->dp_root_dir, dp); bpobj_close(&dp->dp_free_bpobj); /* undo the dmu_objset_open_impl(mos) from dsl_pool_open() */ if (dp->dp_meta_objset) dmu_objset_evict(dp->dp_meta_objset); txg_list_destroy(&dp->dp_dirty_datasets); txg_list_destroy(&dp->dp_dirty_zilogs); txg_list_destroy(&dp->dp_sync_tasks); txg_list_destroy(&dp->dp_dirty_dirs); arc_flush(dp->dp_spa); txg_fini(dp); dsl_scan_fini(dp); rrw_destroy(&dp->dp_config_rwlock); mutex_destroy(&dp->dp_lock); taskq_destroy(dp->dp_iput_taskq); if (dp->dp_blkstats) kmem_free(dp->dp_blkstats, sizeof (zfs_all_blkstats_t)); kmem_free(dp, sizeof (dsl_pool_t)); } dsl_pool_t * dsl_pool_create(spa_t *spa, nvlist_t *zplprops, uint64_t txg) { int err; dsl_pool_t *dp = dsl_pool_open_impl(spa, txg); dmu_tx_t *tx = dmu_tx_create_assigned(dp, txg); objset_t *os; dsl_dataset_t *ds; uint64_t obj; rrw_enter(&dp->dp_config_rwlock, RW_WRITER, FTAG); /* create and open the MOS (meta-objset) */ dp->dp_meta_objset = dmu_objset_create_impl(spa, NULL, &dp->dp_meta_rootbp, DMU_OST_META, tx); /* create the pool directory */ err = zap_create_claim(dp->dp_meta_objset, DMU_POOL_DIRECTORY_OBJECT, DMU_OT_OBJECT_DIRECTORY, DMU_OT_NONE, 0, tx); ASSERT0(err); /* Initialize scan structures */ VERIFY0(dsl_scan_init(dp, txg)); /* create and open the root dir */ dp->dp_root_dir_obj = dsl_dir_create_sync(dp, NULL, NULL, tx); VERIFY0(dsl_dir_hold_obj(dp, dp->dp_root_dir_obj, NULL, dp, &dp->dp_root_dir)); /* create and open the meta-objset dir */ (void) dsl_dir_create_sync(dp, dp->dp_root_dir, MOS_DIR_NAME, tx); VERIFY0(dsl_pool_open_special_dir(dp, MOS_DIR_NAME, &dp->dp_mos_dir)); if (spa_version(spa) >= SPA_VERSION_DEADLISTS) { /* create and open the free dir */ (void) dsl_dir_create_sync(dp, dp->dp_root_dir, FREE_DIR_NAME, tx); VERIFY0(dsl_pool_open_special_dir(dp, FREE_DIR_NAME, &dp->dp_free_dir)); /* create and open the free_bplist */ obj = bpobj_alloc(dp->dp_meta_objset, SPA_MAXBLOCKSIZE, tx); VERIFY(zap_add(dp->dp_meta_objset, DMU_POOL_DIRECTORY_OBJECT, DMU_POOL_FREE_BPOBJ, sizeof (uint64_t), 1, &obj, tx) == 0); VERIFY0(bpobj_open(&dp->dp_free_bpobj, dp->dp_meta_objset, obj)); } if (spa_version(spa) >= SPA_VERSION_DSL_SCRUB) dsl_pool_create_origin(dp, tx); /* create the root dataset */ obj = dsl_dataset_create_sync_dd(dp->dp_root_dir, NULL, 0, tx); /* create the root objset */ VERIFY0(dsl_dataset_hold_obj(dp, obj, FTAG, &ds)); VERIFY(NULL != (os = dmu_objset_create_impl(dp->dp_spa, ds, dsl_dataset_get_blkptr(ds), DMU_OST_ZFS, tx))); #ifdef _KERNEL zfs_create_fs(os, kcred, zplprops, tx); #endif dsl_dataset_rele(ds, FTAG); dmu_tx_commit(tx); rrw_exit(&dp->dp_config_rwlock, FTAG); return (dp); } /* * Account for the meta-objset space in its placeholder dsl_dir. */ void dsl_pool_mos_diduse_space(dsl_pool_t *dp, int64_t used, int64_t comp, int64_t uncomp) { ASSERT3U(comp, ==, uncomp); /* it's all metadata */ mutex_enter(&dp->dp_lock); dp->dp_mos_used_delta += used; dp->dp_mos_compressed_delta += comp; dp->dp_mos_uncompressed_delta += uncomp; mutex_exit(&dp->dp_lock); } static int deadlist_enqueue_cb(void *arg, const blkptr_t *bp, dmu_tx_t *tx) { dsl_deadlist_t *dl = arg; dsl_deadlist_insert(dl, bp, tx); return (0); } static void dsl_pool_sync_mos(dsl_pool_t *dp, dmu_tx_t *tx) { zio_t *zio = zio_root(dp->dp_spa, NULL, NULL, ZIO_FLAG_MUSTSUCCEED); dmu_objset_sync(dp->dp_meta_objset, zio, tx); VERIFY0(zio_wait(zio)); dprintf_bp(&dp->dp_meta_rootbp, "meta objset rootbp is %s", ""); spa_set_rootblkptr(dp->dp_spa, &dp->dp_meta_rootbp); } static void dsl_pool_dirty_delta(dsl_pool_t *dp, int64_t delta) { ASSERT(MUTEX_HELD(&dp->dp_lock)); if (delta < 0) ASSERT3U(-delta, <=, dp->dp_dirty_total); dp->dp_dirty_total += delta; /* * Note: we signal even when increasing dp_dirty_total. * This ensures forward progress -- each thread wakes the next waiter. */ if (dp->dp_dirty_total <= zfs_dirty_data_max) cv_signal(&dp->dp_spaceavail_cv); } void dsl_pool_sync(dsl_pool_t *dp, uint64_t txg) { zio_t *zio; dmu_tx_t *tx; dsl_dir_t *dd; dsl_dataset_t *ds; objset_t *mos = dp->dp_meta_objset; list_t synced_datasets; list_create(&synced_datasets, sizeof (dsl_dataset_t), offsetof(dsl_dataset_t, ds_synced_link)); tx = dmu_tx_create_assigned(dp, txg); /* * Write out all dirty blocks of dirty datasets. */ zio = zio_root(dp->dp_spa, NULL, NULL, ZIO_FLAG_MUSTSUCCEED); while ((ds = txg_list_remove(&dp->dp_dirty_datasets, txg)) != NULL) { /* * We must not sync any non-MOS datasets twice, because * we may have taken a snapshot of them. However, we * may sync newly-created datasets on pass 2. */ ASSERT(!list_link_active(&ds->ds_synced_link)); list_insert_tail(&synced_datasets, ds); dsl_dataset_sync(ds, zio, tx); } VERIFY0(zio_wait(zio)); /* * We have written all of the accounted dirty data, so our * dp_space_towrite should now be zero. However, some seldom-used * code paths do not adhere to this (e.g. dbuf_undirty(), also * rounding error in dbuf_write_physdone). * Shore up the accounting of any dirtied space now. */ dsl_pool_undirty_space(dp, dp->dp_dirty_pertxg[txg & TXG_MASK], txg); /* * After the data blocks have been written (ensured by the zio_wait() * above), update the user/group space accounting. */ for (ds = list_head(&synced_datasets); ds != NULL; ds = list_next(&synced_datasets, ds)) { dmu_objset_do_userquota_updates(ds->ds_objset, tx); } /* * Sync the datasets again to push out the changes due to * userspace updates. This must be done before we process the * sync tasks, so that any snapshots will have the correct * user accounting information (and we won't get confused * about which blocks are part of the snapshot). */ zio = zio_root(dp->dp_spa, NULL, NULL, ZIO_FLAG_MUSTSUCCEED); while ((ds = txg_list_remove(&dp->dp_dirty_datasets, txg)) != NULL) { ASSERT(list_link_active(&ds->ds_synced_link)); dmu_buf_rele(ds->ds_dbuf, ds); dsl_dataset_sync(ds, zio, tx); } VERIFY0(zio_wait(zio)); /* * Now that the datasets have been completely synced, we can * clean up our in-memory structures accumulated while syncing: * * - move dead blocks from the pending deadlist to the on-disk deadlist * - release hold from dsl_dataset_dirty() */ while ((ds = list_remove_head(&synced_datasets)) != NULL) { ASSERTV(objset_t *os = ds->ds_objset); bplist_iterate(&ds->ds_pending_deadlist, deadlist_enqueue_cb, &ds->ds_deadlist, tx); ASSERT(!dmu_objset_is_dirty(os, txg)); dmu_buf_rele(ds->ds_dbuf, ds); } while ((dd = txg_list_remove(&dp->dp_dirty_dirs, txg)) != NULL) { dsl_dir_sync(dd, tx); } /* * The MOS's space is accounted for in the pool/$MOS * (dp_mos_dir). We can't modify the mos while we're syncing * it, so we remember the deltas and apply them here. */ if (dp->dp_mos_used_delta != 0 || dp->dp_mos_compressed_delta != 0 || dp->dp_mos_uncompressed_delta != 0) { dsl_dir_diduse_space(dp->dp_mos_dir, DD_USED_HEAD, dp->dp_mos_used_delta, dp->dp_mos_compressed_delta, dp->dp_mos_uncompressed_delta, tx); dp->dp_mos_used_delta = 0; dp->dp_mos_compressed_delta = 0; dp->dp_mos_uncompressed_delta = 0; } if (list_head(&mos->os_dirty_dnodes[txg & TXG_MASK]) != NULL || list_head(&mos->os_free_dnodes[txg & TXG_MASK]) != NULL) { dsl_pool_sync_mos(dp, tx); } /* * If we modify a dataset in the same txg that we want to destroy it, * its dsl_dir's dd_dbuf will be dirty, and thus have a hold on it. * dsl_dir_destroy_check() will fail if there are unexpected holds. * Therefore, we want to sync the MOS (thus syncing the dd_dbuf * and clearing the hold on it) before we process the sync_tasks. * The MOS data dirtied by the sync_tasks will be synced on the next * pass. */ if (!txg_list_empty(&dp->dp_sync_tasks, txg)) { dsl_sync_task_t *dst; /* * No more sync tasks should have been added while we * were syncing. */ ASSERT3U(spa_sync_pass(dp->dp_spa), ==, 1); while ((dst = txg_list_remove(&dp->dp_sync_tasks, txg)) != NULL) dsl_sync_task_sync(dst, tx); } dmu_tx_commit(tx); DTRACE_PROBE2(dsl_pool_sync__done, dsl_pool_t *dp, dp, uint64_t, txg); } void dsl_pool_sync_done(dsl_pool_t *dp, uint64_t txg) { zilog_t *zilog; while ((zilog = txg_list_remove(&dp->dp_dirty_zilogs, txg))) { dsl_dataset_t *ds = dmu_objset_ds(zilog->zl_os); zil_clean(zilog, txg); ASSERT(!dmu_objset_is_dirty(zilog->zl_os, txg)); dmu_buf_rele(ds->ds_dbuf, zilog); } ASSERT(!dmu_objset_is_dirty(dp->dp_meta_objset, txg)); } /* * TRUE if the current thread is the tx_sync_thread or if we * are being called from SPA context during pool initialization. */ int dsl_pool_sync_context(dsl_pool_t *dp) { return (curthread == dp->dp_tx.tx_sync_thread || spa_is_initializing(dp->dp_spa)); } uint64_t dsl_pool_adjustedsize(dsl_pool_t *dp, boolean_t netfree) { uint64_t space, resv; /* * Reserve about 1.6% (1/64), or at least 32MB, for allocation * efficiency. * XXX The intent log is not accounted for, so it must fit * within this slop. * * If we're trying to assess whether it's OK to do a free, * cut the reservation in half to allow forward progress * (e.g. make it possible to rm(1) files from a full pool). */ space = spa_get_dspace(dp->dp_spa); resv = MAX(space >> 6, SPA_MINDEVSIZE >> 1); if (netfree) resv >>= 1; return (space - resv); } boolean_t dsl_pool_need_dirty_delay(dsl_pool_t *dp) { uint64_t delay_min_bytes = zfs_dirty_data_max * zfs_delay_min_dirty_percent / 100; boolean_t rv; mutex_enter(&dp->dp_lock); if (dp->dp_dirty_total > zfs_dirty_data_sync) txg_kick(dp); rv = (dp->dp_dirty_total > delay_min_bytes); mutex_exit(&dp->dp_lock); return (rv); } void dsl_pool_dirty_space(dsl_pool_t *dp, int64_t space, dmu_tx_t *tx) { if (space > 0) { mutex_enter(&dp->dp_lock); dp->dp_dirty_pertxg[tx->tx_txg & TXG_MASK] += space; dsl_pool_dirty_delta(dp, space); mutex_exit(&dp->dp_lock); } } void dsl_pool_undirty_space(dsl_pool_t *dp, int64_t space, uint64_t txg) { ASSERT3S(space, >=, 0); if (space == 0) return; mutex_enter(&dp->dp_lock); if (dp->dp_dirty_pertxg[txg & TXG_MASK] < space) { /* XXX writing something we didn't dirty? */ space = dp->dp_dirty_pertxg[txg & TXG_MASK]; } ASSERT3U(dp->dp_dirty_pertxg[txg & TXG_MASK], >=, space); dp->dp_dirty_pertxg[txg & TXG_MASK] -= space; ASSERT3U(dp->dp_dirty_total, >=, space); dsl_pool_dirty_delta(dp, -space); mutex_exit(&dp->dp_lock); } /* ARGSUSED */ static int upgrade_clones_cb(dsl_pool_t *dp, dsl_dataset_t *hds, void *arg) { dmu_tx_t *tx = arg; dsl_dataset_t *ds, *prev = NULL; int err; err = dsl_dataset_hold_obj(dp, hds->ds_object, FTAG, &ds); if (err) return (err); while (ds->ds_phys->ds_prev_snap_obj != 0) { err = dsl_dataset_hold_obj(dp, ds->ds_phys->ds_prev_snap_obj, FTAG, &prev); if (err) { dsl_dataset_rele(ds, FTAG); return (err); } if (prev->ds_phys->ds_next_snap_obj != ds->ds_object) break; dsl_dataset_rele(ds, FTAG); ds = prev; prev = NULL; } if (prev == NULL) { prev = dp->dp_origin_snap; /* * The $ORIGIN can't have any data, or the accounting * will be wrong. */ ASSERT0(prev->ds_phys->ds_bp.blk_birth); /* The origin doesn't get attached to itself */ if (ds->ds_object == prev->ds_object) { dsl_dataset_rele(ds, FTAG); return (0); } dmu_buf_will_dirty(ds->ds_dbuf, tx); ds->ds_phys->ds_prev_snap_obj = prev->ds_object; ds->ds_phys->ds_prev_snap_txg = prev->ds_phys->ds_creation_txg; dmu_buf_will_dirty(ds->ds_dir->dd_dbuf, tx); ds->ds_dir->dd_phys->dd_origin_obj = prev->ds_object; dmu_buf_will_dirty(prev->ds_dbuf, tx); prev->ds_phys->ds_num_children++; if (ds->ds_phys->ds_next_snap_obj == 0) { ASSERT(ds->ds_prev == NULL); VERIFY0(dsl_dataset_hold_obj(dp, ds->ds_phys->ds_prev_snap_obj, ds, &ds->ds_prev)); } } ASSERT3U(ds->ds_dir->dd_phys->dd_origin_obj, ==, prev->ds_object); ASSERT3U(ds->ds_phys->ds_prev_snap_obj, ==, prev->ds_object); if (prev->ds_phys->ds_next_clones_obj == 0) { dmu_buf_will_dirty(prev->ds_dbuf, tx); prev->ds_phys->ds_next_clones_obj = zap_create(dp->dp_meta_objset, DMU_OT_NEXT_CLONES, DMU_OT_NONE, 0, tx); } VERIFY0(zap_add_int(dp->dp_meta_objset, prev->ds_phys->ds_next_clones_obj, ds->ds_object, tx)); dsl_dataset_rele(ds, FTAG); if (prev != dp->dp_origin_snap) dsl_dataset_rele(prev, FTAG); return (0); } void dsl_pool_upgrade_clones(dsl_pool_t *dp, dmu_tx_t *tx) { ASSERT(dmu_tx_is_syncing(tx)); ASSERT(dp->dp_origin_snap != NULL); VERIFY0(dmu_objset_find_dp(dp, dp->dp_root_dir_obj, upgrade_clones_cb, tx, DS_FIND_CHILDREN)); } /* ARGSUSED */ static int upgrade_dir_clones_cb(dsl_pool_t *dp, dsl_dataset_t *ds, void *arg) { dmu_tx_t *tx = arg; objset_t *mos = dp->dp_meta_objset; if (ds->ds_dir->dd_phys->dd_origin_obj != 0) { dsl_dataset_t *origin; VERIFY0(dsl_dataset_hold_obj(dp, ds->ds_dir->dd_phys->dd_origin_obj, FTAG, &origin)); if (origin->ds_dir->dd_phys->dd_clones == 0) { dmu_buf_will_dirty(origin->ds_dir->dd_dbuf, tx); origin->ds_dir->dd_phys->dd_clones = zap_create(mos, DMU_OT_DSL_CLONES, DMU_OT_NONE, 0, tx); } VERIFY0(zap_add_int(dp->dp_meta_objset, origin->ds_dir->dd_phys->dd_clones, ds->ds_object, tx)); dsl_dataset_rele(origin, FTAG); } return (0); } void dsl_pool_upgrade_dir_clones(dsl_pool_t *dp, dmu_tx_t *tx) { uint64_t obj; ASSERT(dmu_tx_is_syncing(tx)); (void) dsl_dir_create_sync(dp, dp->dp_root_dir, FREE_DIR_NAME, tx); VERIFY0(dsl_pool_open_special_dir(dp, FREE_DIR_NAME, &dp->dp_free_dir)); /* * We can't use bpobj_alloc(), because spa_version() still * returns the old version, and we need a new-version bpobj with * subobj support. So call dmu_object_alloc() directly. */ obj = dmu_object_alloc(dp->dp_meta_objset, DMU_OT_BPOBJ, SPA_MAXBLOCKSIZE, DMU_OT_BPOBJ_HDR, sizeof (bpobj_phys_t), tx); VERIFY0(zap_add(dp->dp_meta_objset, DMU_POOL_DIRECTORY_OBJECT, DMU_POOL_FREE_BPOBJ, sizeof (uint64_t), 1, &obj, tx)); VERIFY0(bpobj_open(&dp->dp_free_bpobj, dp->dp_meta_objset, obj)); VERIFY0(dmu_objset_find_dp(dp, dp->dp_root_dir_obj, upgrade_dir_clones_cb, tx, DS_FIND_CHILDREN)); } void dsl_pool_create_origin(dsl_pool_t *dp, dmu_tx_t *tx) { uint64_t dsobj; dsl_dataset_t *ds; ASSERT(dmu_tx_is_syncing(tx)); ASSERT(dp->dp_origin_snap == NULL); ASSERT(rrw_held(&dp->dp_config_rwlock, RW_WRITER)); /* create the origin dir, ds, & snap-ds */ dsobj = dsl_dataset_create_sync(dp->dp_root_dir, ORIGIN_DIR_NAME, NULL, 0, kcred, tx); VERIFY0(dsl_dataset_hold_obj(dp, dsobj, FTAG, &ds)); dsl_dataset_snapshot_sync_impl(ds, ORIGIN_DIR_NAME, tx); VERIFY0(dsl_dataset_hold_obj(dp, ds->ds_phys->ds_prev_snap_obj, dp, &dp->dp_origin_snap)); dsl_dataset_rele(ds, FTAG); } taskq_t * dsl_pool_iput_taskq(dsl_pool_t *dp) { return (dp->dp_iput_taskq); } /* * Walk through the pool-wide zap object of temporary snapshot user holds * and release them. */ void dsl_pool_clean_tmp_userrefs(dsl_pool_t *dp) { zap_attribute_t za; zap_cursor_t zc; objset_t *mos = dp->dp_meta_objset; uint64_t zapobj = dp->dp_tmp_userrefs_obj; nvlist_t *holds; if (zapobj == 0) return; ASSERT(spa_version(dp->dp_spa) >= SPA_VERSION_USERREFS); holds = fnvlist_alloc(); for (zap_cursor_init(&zc, mos, zapobj); zap_cursor_retrieve(&zc, &za) == 0; zap_cursor_advance(&zc)) { char *htag; nvlist_t *tags; htag = strchr(za.za_name, '-'); *htag = '\0'; ++htag; if (nvlist_lookup_nvlist(holds, za.za_name, &tags) != 0) { tags = fnvlist_alloc(); fnvlist_add_boolean(tags, htag); fnvlist_add_nvlist(holds, za.za_name, tags); fnvlist_free(tags); } else { fnvlist_add_boolean(tags, htag); } } dsl_dataset_user_release_tmp(dp, holds); fnvlist_free(holds); zap_cursor_fini(&zc); } /* * Create the pool-wide zap object for storing temporary snapshot holds. */ void dsl_pool_user_hold_create_obj(dsl_pool_t *dp, dmu_tx_t *tx) { objset_t *mos = dp->dp_meta_objset; ASSERT(dp->dp_tmp_userrefs_obj == 0); ASSERT(dmu_tx_is_syncing(tx)); dp->dp_tmp_userrefs_obj = zap_create_link(mos, DMU_OT_USERREFS, DMU_POOL_DIRECTORY_OBJECT, DMU_POOL_TMP_USERREFS, tx); } static int dsl_pool_user_hold_rele_impl(dsl_pool_t *dp, uint64_t dsobj, const char *tag, uint64_t now, dmu_tx_t *tx, boolean_t holding) { objset_t *mos = dp->dp_meta_objset; uint64_t zapobj = dp->dp_tmp_userrefs_obj; char *name; int error; ASSERT(spa_version(dp->dp_spa) >= SPA_VERSION_USERREFS); ASSERT(dmu_tx_is_syncing(tx)); /* * If the pool was created prior to SPA_VERSION_USERREFS, the * zap object for temporary holds might not exist yet. */ if (zapobj == 0) { if (holding) { dsl_pool_user_hold_create_obj(dp, tx); zapobj = dp->dp_tmp_userrefs_obj; } else { return (SET_ERROR(ENOENT)); } } name = kmem_asprintf("%llx-%s", (u_longlong_t)dsobj, tag); if (holding) error = zap_add(mos, zapobj, name, 8, 1, &now, tx); else error = zap_remove(mos, zapobj, name, tx); strfree(name); return (error); } /* * Add a temporary hold for the given dataset object and tag. */ int dsl_pool_user_hold(dsl_pool_t *dp, uint64_t dsobj, const char *tag, uint64_t now, dmu_tx_t *tx) { return (dsl_pool_user_hold_rele_impl(dp, dsobj, tag, now, tx, B_TRUE)); } /* * Release a temporary hold for the given dataset object and tag. */ int dsl_pool_user_release(dsl_pool_t *dp, uint64_t dsobj, const char *tag, dmu_tx_t *tx) { return (dsl_pool_user_hold_rele_impl(dp, dsobj, tag, 0, tx, B_FALSE)); } /* * DSL Pool Configuration Lock * * The dp_config_rwlock protects against changes to DSL state (e.g. dataset * creation / destruction / rename / property setting). It must be held for * read to hold a dataset or dsl_dir. I.e. you must call * dsl_pool_config_enter() or dsl_pool_hold() before calling * dsl_{dataset,dir}_hold{_obj}. In most circumstances, the dp_config_rwlock * must be held continuously until all datasets and dsl_dirs are released. * * The only exception to this rule is that if a "long hold" is placed on * a dataset, then the dp_config_rwlock may be dropped while the dataset * is still held. The long hold will prevent the dataset from being * destroyed -- the destroy will fail with EBUSY. A long hold can be * obtained by calling dsl_dataset_long_hold(), or by "owning" a dataset * (by calling dsl_{dataset,objset}_{try}own{_obj}). * * Legitimate long-holders (including owners) should be long-running, cancelable * tasks that should cause "zfs destroy" to fail. This includes DMU * consumers (i.e. a ZPL filesystem being mounted or ZVOL being open), * "zfs send", and "zfs diff". There are several other long-holders whose * uses are suboptimal (e.g. "zfs promote", and zil_suspend()). * * The usual formula for long-holding would be: * dsl_pool_hold() * dsl_dataset_hold() * ... perform checks ... * dsl_dataset_long_hold() * dsl_pool_rele() * ... perform long-running task ... * dsl_dataset_long_rele() * dsl_dataset_rele() * * Note that when the long hold is released, the dataset is still held but * the pool is not held. The dataset may change arbitrarily during this time * (e.g. it could be destroyed). Therefore you shouldn't do anything to the * dataset except release it. * * User-initiated operations (e.g. ioctls, zfs_ioc_*()) are either read-only * or modifying operations. * * Modifying operations should generally use dsl_sync_task(). The synctask * infrastructure enforces proper locking strategy with respect to the * dp_config_rwlock. See the comment above dsl_sync_task() for details. * * Read-only operations will manually hold the pool, then the dataset, obtain * information from the dataset, then release the pool and dataset. * dmu_objset_{hold,rele}() are convenience routines that also do the pool * hold/rele. */ int dsl_pool_hold(const char *name, void *tag, dsl_pool_t **dp) { spa_t *spa; int error; error = spa_open(name, &spa, tag); if (error == 0) { *dp = spa_get_dsl(spa); dsl_pool_config_enter(*dp, tag); } return (error); } void dsl_pool_rele(dsl_pool_t *dp, void *tag) { dsl_pool_config_exit(dp, tag); spa_close(dp->dp_spa, tag); } void dsl_pool_config_enter(dsl_pool_t *dp, void *tag) { /* * We use a "reentrant" reader-writer lock, but not reentrantly. * * The rrwlock can (with the track_all flag) track all reading threads, * which is very useful for debugging which code path failed to release * the lock, and for verifying that the *current* thread does hold * the lock. * * (Unlike a rwlock, which knows that N threads hold it for * read, but not *which* threads, so rw_held(RW_READER) returns TRUE * if any thread holds it for read, even if this thread doesn't). */ ASSERT(!rrw_held(&dp->dp_config_rwlock, RW_READER)); rrw_enter(&dp->dp_config_rwlock, RW_READER, tag); } void dsl_pool_config_exit(dsl_pool_t *dp, void *tag) { rrw_exit(&dp->dp_config_rwlock, tag); } boolean_t dsl_pool_config_held(dsl_pool_t *dp) { return (RRW_LOCK_HELD(&dp->dp_config_rwlock)); } #if defined(_KERNEL) && defined(HAVE_SPL) EXPORT_SYMBOL(dsl_pool_config_enter); EXPORT_SYMBOL(dsl_pool_config_exit); /* zfs_dirty_data_max_percent only applied at module load in arc_init(). */ module_param(zfs_dirty_data_max_percent, int, 0444); MODULE_PARM_DESC(zfs_dirty_data_max_percent, "percent of ram can be dirty"); /* zfs_dirty_data_max_max_percent only applied at module load in arc_init(). */ module_param(zfs_dirty_data_max_max_percent, int, 0444); MODULE_PARM_DESC(zfs_dirty_data_max_max_percent, "zfs_dirty_data_max upper bound as % of RAM"); module_param(zfs_delay_min_dirty_percent, int, 0644); MODULE_PARM_DESC(zfs_delay_min_dirty_percent, "transaction delay threshold"); module_param(zfs_dirty_data_max, ulong, 0644); MODULE_PARM_DESC(zfs_dirty_data_max, "determines the dirty space limit"); /* zfs_dirty_data_max_max only applied at module load in arc_init(). */ module_param(zfs_dirty_data_max_max, ulong, 0444); MODULE_PARM_DESC(zfs_dirty_data_max_max, "zfs_dirty_data_max upper bound in bytes"); module_param(zfs_dirty_data_sync, ulong, 0644); MODULE_PARM_DESC(zfs_dirty_data_sync, "sync txg when this much dirty data"); module_param(zfs_delay_scale, ulong, 0644); MODULE_PARM_DESC(zfs_delay_scale, "how quickly delay approaches infinity"); #endif diff --git a/module/zfs/dsl_scan.c b/module/zfs/dsl_scan.c index f03f1f54b450..0f0243b310a0 100644 --- a/module/zfs/dsl_scan.c +++ b/module/zfs/dsl_scan.c @@ -1,1832 +1,1880 @@ /* * 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) 2008, 2010, Oracle and/or its affiliates. All rights reserved. * Copyright (c) 2013 by Delphix. All rights reserved. */ #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #ifdef _KERNEL #include #endif typedef int (scan_cb_t)(dsl_pool_t *, const blkptr_t *, const zbookmark_t *); static scan_cb_t dsl_scan_scrub_cb; static void dsl_scan_cancel_sync(void *, dmu_tx_t *); static void dsl_scan_sync_state(dsl_scan_t *, dmu_tx_t *tx); int zfs_top_maxinflight = 32; /* maximum I/Os per top-level */ int zfs_resilver_delay = 2; /* number of ticks to delay resilver */ int zfs_scrub_delay = 4; /* number of ticks to delay scrub */ int zfs_scan_idle = 50; /* idle window in clock ticks */ int zfs_scan_min_time_ms = 1000; /* min millisecs to scrub per txg */ int zfs_free_min_time_ms = 1000; /* min millisecs to free per txg */ int zfs_resilver_min_time_ms = 3000; /* min millisecs to resilver per txg */ int zfs_no_scrub_io = B_FALSE; /* set to disable scrub i/o */ -int zfs_no_scrub_prefetch = B_FALSE; /* set to disable srub prefetching */ +int zfs_no_scrub_prefetch = B_FALSE; /* set to disable scrub prefetch */ enum ddt_class zfs_scrub_ddt_class_max = DDT_CLASS_DUPLICATE; int dsl_scan_delay_completion = B_FALSE; /* set to delay scan completion */ #define DSL_SCAN_IS_SCRUB_RESILVER(scn) \ ((scn)->scn_phys.scn_func == POOL_SCAN_SCRUB || \ (scn)->scn_phys.scn_func == POOL_SCAN_RESILVER) /* the order has to match pool_scan_type */ static scan_cb_t *scan_funcs[POOL_SCAN_FUNCS] = { NULL, dsl_scan_scrub_cb, /* POOL_SCAN_SCRUB */ dsl_scan_scrub_cb, /* POOL_SCAN_RESILVER */ }; int dsl_scan_init(dsl_pool_t *dp, uint64_t txg) { int err; dsl_scan_t *scn; spa_t *spa = dp->dp_spa; uint64_t f; scn = dp->dp_scan = kmem_zalloc(sizeof (dsl_scan_t), KM_SLEEP); scn->scn_dp = dp; /* * It's possible that we're resuming a scan after a reboot so * make sure that the scan_async_destroying flag is initialized * appropriately. */ ASSERT(!scn->scn_async_destroying); scn->scn_async_destroying = spa_feature_is_active(dp->dp_spa, SPA_FEATURE_ASYNC_DESTROY); err = zap_lookup(dp->dp_meta_objset, DMU_POOL_DIRECTORY_OBJECT, "scrub_func", sizeof (uint64_t), 1, &f); if (err == 0) { /* * There was an old-style scrub in progress. Restart a * new-style scrub from the beginning. */ scn->scn_restart_txg = txg; zfs_dbgmsg("old-style scrub was in progress; " "restarting new-style scrub in txg %llu", scn->scn_restart_txg); /* * Load the queue obj from the old location so that it * can be freed by dsl_scan_done(). */ (void) zap_lookup(dp->dp_meta_objset, DMU_POOL_DIRECTORY_OBJECT, "scrub_queue", sizeof (uint64_t), 1, &scn->scn_phys.scn_queue_obj); } else { err = zap_lookup(dp->dp_meta_objset, DMU_POOL_DIRECTORY_OBJECT, DMU_POOL_SCAN, sizeof (uint64_t), SCAN_PHYS_NUMINTS, &scn->scn_phys); /* * Detect if the pool contains the signature of #2094. If it * does properly update the scn->scn_phys structure and notify * the administrator by setting an errata for the pool. */ if (err == EOVERFLOW) { uint64_t zaptmp[SCAN_PHYS_NUMINTS + 1]; VERIFY3S(SCAN_PHYS_NUMINTS, ==, 24); VERIFY3S(offsetof(dsl_scan_phys_t, scn_flags), ==, (23 * sizeof (uint64_t))); err = zap_lookup(dp->dp_meta_objset, DMU_POOL_DIRECTORY_OBJECT, DMU_POOL_SCAN, sizeof (uint64_t), SCAN_PHYS_NUMINTS + 1, &zaptmp); if (err == 0) { uint64_t overflow = zaptmp[SCAN_PHYS_NUMINTS]; if (overflow & ~DSL_SCAN_FLAGS_MASK || scn->scn_async_destroying) { spa->spa_errata = ZPOOL_ERRATA_ZOL_2094_ASYNC_DESTROY; return (EOVERFLOW); } bcopy(zaptmp, &scn->scn_phys, SCAN_PHYS_NUMINTS * sizeof (uint64_t)); scn->scn_phys.scn_flags = overflow; /* Required scrub already in progress. */ if (scn->scn_phys.scn_state == DSS_FINISHED || scn->scn_phys.scn_state == DSS_CANCELED) spa->spa_errata = ZPOOL_ERRATA_ZOL_2094_SCRUB; } } if (err == ENOENT) return (0); else if (err) return (err); if (scn->scn_phys.scn_state == DSS_SCANNING && spa_prev_software_version(dp->dp_spa) < SPA_VERSION_SCAN) { /* * A new-type scrub was in progress on an old * pool, and the pool was accessed by old * software. Restart from the beginning, since * the old software may have changed the pool in * the meantime. */ scn->scn_restart_txg = txg; zfs_dbgmsg("new-style scrub was modified " "by old software; restarting in txg %llu", scn->scn_restart_txg); } } spa_scan_stat_init(spa); return (0); } void dsl_scan_fini(dsl_pool_t *dp) { if (dp->dp_scan) { kmem_free(dp->dp_scan, sizeof (dsl_scan_t)); dp->dp_scan = NULL; } } /* ARGSUSED */ static int dsl_scan_setup_check(void *arg, dmu_tx_t *tx) { dsl_scan_t *scn = dmu_tx_pool(tx)->dp_scan; if (scn->scn_phys.scn_state == DSS_SCANNING) return (SET_ERROR(EBUSY)); return (0); } static void dsl_scan_setup_sync(void *arg, dmu_tx_t *tx) { dsl_scan_t *scn = dmu_tx_pool(tx)->dp_scan; pool_scan_func_t *funcp = arg; dmu_object_type_t ot = 0; dsl_pool_t *dp = scn->scn_dp; spa_t *spa = dp->dp_spa; ASSERT(scn->scn_phys.scn_state != DSS_SCANNING); ASSERT(*funcp > POOL_SCAN_NONE && *funcp < POOL_SCAN_FUNCS); bzero(&scn->scn_phys, sizeof (scn->scn_phys)); scn->scn_phys.scn_func = *funcp; scn->scn_phys.scn_state = DSS_SCANNING; scn->scn_phys.scn_min_txg = 0; scn->scn_phys.scn_max_txg = tx->tx_txg; scn->scn_phys.scn_ddt_class_max = DDT_CLASSES - 1; /* the entire DDT */ scn->scn_phys.scn_start_time = gethrestime_sec(); scn->scn_phys.scn_errors = 0; scn->scn_phys.scn_to_examine = spa->spa_root_vdev->vdev_stat.vs_alloc; scn->scn_restart_txg = 0; scn->scn_done_txg = 0; spa_scan_stat_init(spa); if (DSL_SCAN_IS_SCRUB_RESILVER(scn)) { scn->scn_phys.scn_ddt_class_max = zfs_scrub_ddt_class_max; /* rewrite all disk labels */ vdev_config_dirty(spa->spa_root_vdev); if (vdev_resilver_needed(spa->spa_root_vdev, &scn->scn_phys.scn_min_txg, &scn->scn_phys.scn_max_txg)) { spa_event_notify(spa, NULL, FM_EREPORT_ZFS_RESILVER_START); } else { spa_event_notify(spa, NULL, FM_EREPORT_ZFS_SCRUB_START); } spa->spa_scrub_started = B_TRUE; /* * If this is an incremental scrub, limit the DDT scrub phase * to just the auto-ditto class (for correctness); the rest * of the scrub should go faster using top-down pruning. */ if (scn->scn_phys.scn_min_txg > TXG_INITIAL) scn->scn_phys.scn_ddt_class_max = DDT_CLASS_DITTO; } /* back to the generic stuff */ if (dp->dp_blkstats == NULL) { dp->dp_blkstats = kmem_alloc(sizeof (zfs_all_blkstats_t), KM_PUSHPAGE | KM_NODEBUG); } bzero(dp->dp_blkstats, sizeof (zfs_all_blkstats_t)); if (spa_version(spa) < SPA_VERSION_DSL_SCRUB) ot = DMU_OT_ZAP_OTHER; scn->scn_phys.scn_queue_obj = zap_create(dp->dp_meta_objset, ot ? ot : DMU_OT_SCAN_QUEUE, DMU_OT_NONE, 0, tx); dsl_scan_sync_state(scn, tx); spa_history_log_internal(spa, "scan setup", tx, "func=%u mintxg=%llu maxtxg=%llu", *funcp, scn->scn_phys.scn_min_txg, scn->scn_phys.scn_max_txg); } /* ARGSUSED */ static void dsl_scan_done(dsl_scan_t *scn, boolean_t complete, dmu_tx_t *tx) { static const char *old_names[] = { "scrub_bookmark", "scrub_ddt_bookmark", "scrub_ddt_class_max", "scrub_queue", "scrub_min_txg", "scrub_max_txg", "scrub_func", "scrub_errors", NULL }; dsl_pool_t *dp = scn->scn_dp; spa_t *spa = dp->dp_spa; int i; /* Remove any remnants of an old-style scrub. */ for (i = 0; old_names[i]; i++) { (void) zap_remove(dp->dp_meta_objset, DMU_POOL_DIRECTORY_OBJECT, old_names[i], tx); } if (scn->scn_phys.scn_queue_obj != 0) { VERIFY(0 == dmu_object_free(dp->dp_meta_objset, scn->scn_phys.scn_queue_obj, tx)); scn->scn_phys.scn_queue_obj = 0; } /* * If we were "restarted" from a stopped state, don't bother * with anything else. */ if (scn->scn_phys.scn_state != DSS_SCANNING) return; if (complete) scn->scn_phys.scn_state = DSS_FINISHED; else scn->scn_phys.scn_state = DSS_CANCELED; spa_history_log_internal(spa, "scan done", tx, "complete=%u", complete); if (DSL_SCAN_IS_SCRUB_RESILVER(scn)) { mutex_enter(&spa->spa_scrub_lock); while (spa->spa_scrub_inflight > 0) { cv_wait(&spa->spa_scrub_io_cv, &spa->spa_scrub_lock); } mutex_exit(&spa->spa_scrub_lock); spa->spa_scrub_started = B_FALSE; spa->spa_scrub_active = B_FALSE; /* * If the scrub/resilver completed, update all DTLs to * reflect this. Whether it succeeded or not, vacate * all temporary scrub DTLs. */ vdev_dtl_reassess(spa->spa_root_vdev, tx->tx_txg, complete ? scn->scn_phys.scn_max_txg : 0, B_TRUE); if (complete) { spa_event_notify(spa, NULL, scn->scn_phys.scn_min_txg ? FM_EREPORT_ZFS_RESILVER_FINISH : FM_EREPORT_ZFS_SCRUB_FINISH); } spa_errlog_rotate(spa); /* * We may have finished replacing a device. * Let the async thread assess this and handle the detach. */ spa_async_request(spa, SPA_ASYNC_RESILVER_DONE); } scn->scn_phys.scn_end_time = gethrestime_sec(); if (spa->spa_errata == ZPOOL_ERRATA_ZOL_2094_SCRUB) spa->spa_errata = 0; } /* ARGSUSED */ static int dsl_scan_cancel_check(void *arg, dmu_tx_t *tx) { dsl_scan_t *scn = dmu_tx_pool(tx)->dp_scan; if (scn->scn_phys.scn_state != DSS_SCANNING) return (SET_ERROR(ENOENT)); return (0); } /* ARGSUSED */ static void dsl_scan_cancel_sync(void *arg, dmu_tx_t *tx) { dsl_scan_t *scn = dmu_tx_pool(tx)->dp_scan; dsl_scan_done(scn, B_FALSE, tx); dsl_scan_sync_state(scn, tx); } int dsl_scan_cancel(dsl_pool_t *dp) { return (dsl_sync_task(spa_name(dp->dp_spa), dsl_scan_cancel_check, dsl_scan_cancel_sync, NULL, 3)); } static void dsl_scan_visitbp(blkptr_t *bp, const zbookmark_t *zb, dnode_phys_t *dnp, arc_buf_t *pbuf, dsl_dataset_t *ds, dsl_scan_t *scn, dmu_objset_type_t ostype, dmu_tx_t *tx); inline __attribute__((always_inline)) static void dsl_scan_visitdnode( dsl_scan_t *, dsl_dataset_t *ds, dmu_objset_type_t ostype, dnode_phys_t *dnp, arc_buf_t *buf, uint64_t object, dmu_tx_t *tx); void dsl_free(dsl_pool_t *dp, uint64_t txg, const blkptr_t *bp) { zio_free(dp->dp_spa, txg, bp); } void dsl_free_sync(zio_t *pio, dsl_pool_t *dp, uint64_t txg, const blkptr_t *bpp) { ASSERT(dsl_pool_sync_context(dp)); zio_nowait(zio_free_sync(pio, dp->dp_spa, txg, bpp, pio->io_flags)); } static uint64_t dsl_scan_ds_maxtxg(dsl_dataset_t *ds) { uint64_t smt = ds->ds_dir->dd_pool->dp_scan->scn_phys.scn_max_txg; if (dsl_dataset_is_snapshot(ds)) return (MIN(smt, ds->ds_phys->ds_creation_txg)); return (smt); } static void dsl_scan_sync_state(dsl_scan_t *scn, dmu_tx_t *tx) { VERIFY0(zap_update(scn->scn_dp->dp_meta_objset, DMU_POOL_DIRECTORY_OBJECT, DMU_POOL_SCAN, sizeof (uint64_t), SCAN_PHYS_NUMINTS, &scn->scn_phys, tx)); } static boolean_t dsl_scan_check_pause(dsl_scan_t *scn, const zbookmark_t *zb) { uint64_t elapsed_nanosecs; int mintime; /* we never skip user/group accounting objects */ if (zb && (int64_t)zb->zb_object < 0) return (B_FALSE); if (scn->scn_pausing) return (B_TRUE); /* we're already pausing */ if (!ZB_IS_ZERO(&scn->scn_phys.scn_bookmark)) return (B_FALSE); /* we're resuming */ /* We only know how to resume from level-0 blocks. */ if (zb && zb->zb_level != 0) return (B_FALSE); mintime = (scn->scn_phys.scn_func == POOL_SCAN_RESILVER) ? zfs_resilver_min_time_ms : zfs_scan_min_time_ms; elapsed_nanosecs = gethrtime() - scn->scn_sync_start_time; if (elapsed_nanosecs / NANOSEC > zfs_txg_timeout || (NSEC2MSEC(elapsed_nanosecs) > mintime && txg_sync_waiting(scn->scn_dp)) || spa_shutting_down(scn->scn_dp->dp_spa)) { if (zb) { dprintf("pausing at bookmark %llx/%llx/%llx/%llx\n", (longlong_t)zb->zb_objset, (longlong_t)zb->zb_object, (longlong_t)zb->zb_level, (longlong_t)zb->zb_blkid); scn->scn_phys.scn_bookmark = *zb; } dprintf("pausing at DDT bookmark %llx/%llx/%llx/%llx\n", (longlong_t)scn->scn_phys.scn_ddt_bookmark.ddb_class, (longlong_t)scn->scn_phys.scn_ddt_bookmark.ddb_type, (longlong_t)scn->scn_phys.scn_ddt_bookmark.ddb_checksum, (longlong_t)scn->scn_phys.scn_ddt_bookmark.ddb_cursor); scn->scn_pausing = B_TRUE; return (B_TRUE); } return (B_FALSE); } typedef struct zil_scan_arg { dsl_pool_t *zsa_dp; zil_header_t *zsa_zh; } zil_scan_arg_t; /* ARGSUSED */ static int dsl_scan_zil_block(zilog_t *zilog, blkptr_t *bp, void *arg, uint64_t claim_txg) { zil_scan_arg_t *zsa = arg; dsl_pool_t *dp = zsa->zsa_dp; dsl_scan_t *scn = dp->dp_scan; zil_header_t *zh = zsa->zsa_zh; zbookmark_t zb; if (BP_IS_HOLE(bp) || bp->blk_birth <= scn->scn_phys.scn_cur_min_txg) return (0); /* * One block ("stubby") can be allocated a long time ago; we * want to visit that one because it has been allocated * (on-disk) even if it hasn't been claimed (even though for * scrub there's nothing to do to it). */ if (claim_txg == 0 && bp->blk_birth >= spa_first_txg(dp->dp_spa)) return (0); SET_BOOKMARK(&zb, zh->zh_log.blk_cksum.zc_word[ZIL_ZC_OBJSET], ZB_ZIL_OBJECT, ZB_ZIL_LEVEL, bp->blk_cksum.zc_word[ZIL_ZC_SEQ]); VERIFY(0 == scan_funcs[scn->scn_phys.scn_func](dp, bp, &zb)); return (0); } /* ARGSUSED */ static int dsl_scan_zil_record(zilog_t *zilog, lr_t *lrc, void *arg, uint64_t claim_txg) { if (lrc->lrc_txtype == TX_WRITE) { zil_scan_arg_t *zsa = arg; dsl_pool_t *dp = zsa->zsa_dp; dsl_scan_t *scn = dp->dp_scan; zil_header_t *zh = zsa->zsa_zh; lr_write_t *lr = (lr_write_t *)lrc; blkptr_t *bp = &lr->lr_blkptr; zbookmark_t zb; if (BP_IS_HOLE(bp) || bp->blk_birth <= scn->scn_phys.scn_cur_min_txg) return (0); /* * birth can be < claim_txg if this record's txg is * already txg sync'ed (but this log block contains * other records that are not synced) */ if (claim_txg == 0 || bp->blk_birth < claim_txg) return (0); SET_BOOKMARK(&zb, zh->zh_log.blk_cksum.zc_word[ZIL_ZC_OBJSET], lr->lr_foid, ZB_ZIL_LEVEL, lr->lr_offset / BP_GET_LSIZE(bp)); VERIFY(0 == scan_funcs[scn->scn_phys.scn_func](dp, bp, &zb)); } return (0); } static void dsl_scan_zil(dsl_pool_t *dp, zil_header_t *zh) { uint64_t claim_txg = zh->zh_claim_txg; zil_scan_arg_t zsa = { dp, zh }; zilog_t *zilog; /* * We only want to visit blocks that have been claimed but not yet * replayed (or, in read-only mode, blocks that *would* be claimed). */ if (claim_txg == 0 && spa_writeable(dp->dp_spa)) return; zilog = zil_alloc(dp->dp_meta_objset, zh); (void) zil_parse(zilog, dsl_scan_zil_block, dsl_scan_zil_record, &zsa, claim_txg); zil_free(zilog); } /* ARGSUSED */ static void dsl_scan_prefetch(dsl_scan_t *scn, arc_buf_t *buf, blkptr_t *bp, uint64_t objset, uint64_t object, uint64_t blkid) { zbookmark_t czb; uint32_t flags = ARC_NOWAIT | ARC_PREFETCH; if (zfs_no_scrub_prefetch) return; if (BP_IS_HOLE(bp) || bp->blk_birth <= scn->scn_phys.scn_min_txg || (BP_GET_LEVEL(bp) == 0 && BP_GET_TYPE(bp) != DMU_OT_DNODE)) return; SET_BOOKMARK(&czb, objset, object, BP_GET_LEVEL(bp), blkid); (void) arc_read(scn->scn_zio_root, scn->scn_dp->dp_spa, bp, NULL, NULL, ZIO_PRIORITY_ASYNC_READ, ZIO_FLAG_CANFAIL | ZIO_FLAG_SCAN_THREAD, &flags, &czb); } static boolean_t dsl_scan_check_resume(dsl_scan_t *scn, const dnode_phys_t *dnp, const zbookmark_t *zb) { /* * We never skip over user/group accounting objects (obj<0) */ if (!ZB_IS_ZERO(&scn->scn_phys.scn_bookmark) && (int64_t)zb->zb_object >= 0) { /* * If we already visited this bp & everything below (in * a prior txg sync), don't bother doing it again. */ if (zbookmark_is_before(dnp, zb, &scn->scn_phys.scn_bookmark)) return (B_TRUE); /* * If we found the block we're trying to resume from, or * we went past it to a different object, zero it out to * indicate that it's OK to start checking for pausing * again. */ if (bcmp(zb, &scn->scn_phys.scn_bookmark, sizeof (*zb)) == 0 || zb->zb_object > scn->scn_phys.scn_bookmark.zb_object) { dprintf("resuming at %llx/%llx/%llx/%llx\n", (longlong_t)zb->zb_objset, (longlong_t)zb->zb_object, (longlong_t)zb->zb_level, (longlong_t)zb->zb_blkid); bzero(&scn->scn_phys.scn_bookmark, sizeof (*zb)); } } return (B_FALSE); } /* * Return nonzero on i/o error. * Return new buf to write out in *bufp. */ inline __attribute__((always_inline)) static int dsl_scan_recurse(dsl_scan_t *scn, dsl_dataset_t *ds, dmu_objset_type_t ostype, dnode_phys_t *dnp, const blkptr_t *bp, const zbookmark_t *zb, dmu_tx_t *tx, arc_buf_t **bufp) { dsl_pool_t *dp = scn->scn_dp; int zio_flags = ZIO_FLAG_CANFAIL | ZIO_FLAG_SCAN_THREAD; int err; if (BP_GET_LEVEL(bp) > 0) { uint32_t flags = ARC_WAIT; int i; blkptr_t *cbp; int epb = BP_GET_LSIZE(bp) >> SPA_BLKPTRSHIFT; err = arc_read(NULL, dp->dp_spa, bp, arc_getbuf_func, bufp, ZIO_PRIORITY_ASYNC_READ, zio_flags, &flags, zb); if (err) { scn->scn_phys.scn_errors++; return (err); } for (i = 0, cbp = (*bufp)->b_data; i < epb; i++, cbp++) { dsl_scan_prefetch(scn, *bufp, cbp, zb->zb_objset, zb->zb_object, zb->zb_blkid * epb + i); } for (i = 0, cbp = (*bufp)->b_data; i < epb; i++, cbp++) { zbookmark_t czb; SET_BOOKMARK(&czb, zb->zb_objset, zb->zb_object, zb->zb_level - 1, zb->zb_blkid * epb + i); dsl_scan_visitbp(cbp, &czb, dnp, *bufp, ds, scn, ostype, tx); } } else if (BP_GET_TYPE(bp) == DMU_OT_USERGROUP_USED) { uint32_t flags = ARC_WAIT; err = arc_read(NULL, dp->dp_spa, bp, arc_getbuf_func, bufp, ZIO_PRIORITY_ASYNC_READ, zio_flags, &flags, zb); if (err) { scn->scn_phys.scn_errors++; return (err); } } else if (BP_GET_TYPE(bp) == DMU_OT_DNODE) { uint32_t flags = ARC_WAIT; dnode_phys_t *cdnp; int i, j; int epb = BP_GET_LSIZE(bp) >> DNODE_SHIFT; err = arc_read(NULL, dp->dp_spa, bp, arc_getbuf_func, bufp, ZIO_PRIORITY_ASYNC_READ, zio_flags, &flags, zb); if (err) { scn->scn_phys.scn_errors++; return (err); } for (i = 0, cdnp = (*bufp)->b_data; i < epb; i++, cdnp++) { for (j = 0; j < cdnp->dn_nblkptr; j++) { blkptr_t *cbp = &cdnp->dn_blkptr[j]; dsl_scan_prefetch(scn, *bufp, cbp, zb->zb_objset, zb->zb_blkid * epb + i, j); } } for (i = 0, cdnp = (*bufp)->b_data; i < epb; i++, cdnp++) { dsl_scan_visitdnode(scn, ds, ostype, cdnp, *bufp, zb->zb_blkid * epb + i, tx); } } else if (BP_GET_TYPE(bp) == DMU_OT_OBJSET) { uint32_t flags = ARC_WAIT; objset_phys_t *osp; err = arc_read(NULL, dp->dp_spa, bp, arc_getbuf_func, bufp, ZIO_PRIORITY_ASYNC_READ, zio_flags, &flags, zb); if (err) { scn->scn_phys.scn_errors++; return (err); } osp = (*bufp)->b_data; dsl_scan_visitdnode(scn, ds, osp->os_type, &osp->os_meta_dnode, *bufp, DMU_META_DNODE_OBJECT, tx); if (OBJSET_BUF_HAS_USERUSED(*bufp)) { /* * We also always visit user/group accounting * objects, and never skip them, even if we are * pausing. This is necessary so that the space * deltas from this txg get integrated. */ dsl_scan_visitdnode(scn, ds, osp->os_type, &osp->os_groupused_dnode, *bufp, DMU_GROUPUSED_OBJECT, tx); dsl_scan_visitdnode(scn, ds, osp->os_type, &osp->os_userused_dnode, *bufp, DMU_USERUSED_OBJECT, tx); } } return (0); } inline __attribute__((always_inline)) static void dsl_scan_visitdnode(dsl_scan_t *scn, dsl_dataset_t *ds, dmu_objset_type_t ostype, dnode_phys_t *dnp, arc_buf_t *buf, uint64_t object, dmu_tx_t *tx) { int j; for (j = 0; j < dnp->dn_nblkptr; j++) { zbookmark_t czb; SET_BOOKMARK(&czb, ds ? ds->ds_object : 0, object, dnp->dn_nlevels - 1, j); dsl_scan_visitbp(&dnp->dn_blkptr[j], &czb, dnp, buf, ds, scn, ostype, tx); } if (dnp->dn_flags & DNODE_FLAG_SPILL_BLKPTR) { zbookmark_t czb; SET_BOOKMARK(&czb, ds ? ds->ds_object : 0, object, 0, DMU_SPILL_BLKID); dsl_scan_visitbp(&dnp->dn_spill, &czb, dnp, buf, ds, scn, ostype, tx); } } /* * The arguments are in this order because mdb can only print the * first 5; we want them to be useful. */ static void dsl_scan_visitbp(blkptr_t *bp, const zbookmark_t *zb, dnode_phys_t *dnp, arc_buf_t *pbuf, dsl_dataset_t *ds, dsl_scan_t *scn, dmu_objset_type_t ostype, dmu_tx_t *tx) { dsl_pool_t *dp = scn->scn_dp; arc_buf_t *buf = NULL; blkptr_t *bp_toread; bp_toread = kmem_alloc(sizeof (blkptr_t), KM_PUSHPAGE); *bp_toread = *bp; /* ASSERT(pbuf == NULL || arc_released(pbuf)); */ if (dsl_scan_check_pause(scn, zb)) goto out; if (dsl_scan_check_resume(scn, dnp, zb)) goto out; if (BP_IS_HOLE(bp)) goto out; scn->scn_visited_this_txg++; /* * This debugging is commented out to conserve stack space. This * function is called recursively and the debugging addes several * bytes to the stack for each call. It can be commented back in * if required to debug an issue in dsl_scan_visitbp(). * * dprintf_bp(bp, * "visiting ds=%p/%llu zb=%llx/%llx/%llx/%llx buf=%p bp=%p", * ds, ds ? ds->ds_object : 0, * zb->zb_objset, zb->zb_object, zb->zb_level, zb->zb_blkid, * pbuf, bp); */ if (bp->blk_birth <= scn->scn_phys.scn_cur_min_txg) goto out; if (dsl_scan_recurse(scn, ds, ostype, dnp, bp_toread, zb, tx, &buf) != 0) goto out; /* * If dsl_scan_ddt() has aready visited this block, it will have * already done any translations or scrubbing, so don't call the * callback again. */ if (ddt_class_contains(dp->dp_spa, scn->scn_phys.scn_ddt_class_max, bp)) { ASSERT(buf == NULL); goto out; } /* * If this block is from the future (after cur_max_txg), then we * are doing this on behalf of a deleted snapshot, and we will * revisit the future block on the next pass of this dataset. * Don't scan it now unless we need to because something * under it was modified. */ if (BP_PHYSICAL_BIRTH(bp) <= scn->scn_phys.scn_cur_max_txg) { scan_funcs[scn->scn_phys.scn_func](dp, bp, zb); } if (buf) (void) arc_buf_remove_ref(buf, &buf); out: kmem_free(bp_toread, sizeof (blkptr_t)); } static void dsl_scan_visit_rootbp(dsl_scan_t *scn, dsl_dataset_t *ds, blkptr_t *bp, dmu_tx_t *tx) { zbookmark_t zb; SET_BOOKMARK(&zb, ds ? ds->ds_object : DMU_META_OBJSET, ZB_ROOT_OBJECT, ZB_ROOT_LEVEL, ZB_ROOT_BLKID); dsl_scan_visitbp(bp, &zb, NULL, NULL, ds, scn, DMU_OST_NONE, tx); dprintf_ds(ds, "finished scan%s", ""); } void dsl_scan_ds_destroyed(dsl_dataset_t *ds, dmu_tx_t *tx) { dsl_pool_t *dp = ds->ds_dir->dd_pool; dsl_scan_t *scn = dp->dp_scan; uint64_t mintxg; if (scn->scn_phys.scn_state != DSS_SCANNING) return; if (scn->scn_phys.scn_bookmark.zb_objset == ds->ds_object) { if (dsl_dataset_is_snapshot(ds)) { /* Note, scn_cur_{min,max}_txg stays the same. */ scn->scn_phys.scn_bookmark.zb_objset = ds->ds_phys->ds_next_snap_obj; zfs_dbgmsg("destroying ds %llu; currently traversing; " "reset zb_objset to %llu", (u_longlong_t)ds->ds_object, (u_longlong_t)ds->ds_phys->ds_next_snap_obj); scn->scn_phys.scn_flags |= DSF_VISIT_DS_AGAIN; } else { SET_BOOKMARK(&scn->scn_phys.scn_bookmark, ZB_DESTROYED_OBJSET, 0, 0, 0); zfs_dbgmsg("destroying ds %llu; currently traversing; " "reset bookmark to -1,0,0,0", (u_longlong_t)ds->ds_object); } } else if (zap_lookup_int_key(dp->dp_meta_objset, scn->scn_phys.scn_queue_obj, ds->ds_object, &mintxg) == 0) { ASSERT3U(ds->ds_phys->ds_num_children, <=, 1); VERIFY3U(0, ==, zap_remove_int(dp->dp_meta_objset, scn->scn_phys.scn_queue_obj, ds->ds_object, tx)); if (dsl_dataset_is_snapshot(ds)) { /* * We keep the same mintxg; it could be > * ds_creation_txg if the previous snapshot was * deleted too. */ VERIFY(zap_add_int_key(dp->dp_meta_objset, scn->scn_phys.scn_queue_obj, ds->ds_phys->ds_next_snap_obj, mintxg, tx) == 0); zfs_dbgmsg("destroying ds %llu; in queue; " "replacing with %llu", (u_longlong_t)ds->ds_object, (u_longlong_t)ds->ds_phys->ds_next_snap_obj); } else { zfs_dbgmsg("destroying ds %llu; in queue; removing", (u_longlong_t)ds->ds_object); } } else { zfs_dbgmsg("destroying ds %llu; ignoring", (u_longlong_t)ds->ds_object); } /* * dsl_scan_sync() should be called after this, and should sync * out our changed state, but just to be safe, do it here. */ dsl_scan_sync_state(scn, tx); } void dsl_scan_ds_snapshotted(dsl_dataset_t *ds, dmu_tx_t *tx) { dsl_pool_t *dp = ds->ds_dir->dd_pool; dsl_scan_t *scn = dp->dp_scan; uint64_t mintxg; if (scn->scn_phys.scn_state != DSS_SCANNING) return; ASSERT(ds->ds_phys->ds_prev_snap_obj != 0); if (scn->scn_phys.scn_bookmark.zb_objset == ds->ds_object) { scn->scn_phys.scn_bookmark.zb_objset = ds->ds_phys->ds_prev_snap_obj; zfs_dbgmsg("snapshotting ds %llu; currently traversing; " "reset zb_objset to %llu", (u_longlong_t)ds->ds_object, (u_longlong_t)ds->ds_phys->ds_prev_snap_obj); } else if (zap_lookup_int_key(dp->dp_meta_objset, scn->scn_phys.scn_queue_obj, ds->ds_object, &mintxg) == 0) { VERIFY3U(0, ==, zap_remove_int(dp->dp_meta_objset, scn->scn_phys.scn_queue_obj, ds->ds_object, tx)); VERIFY(zap_add_int_key(dp->dp_meta_objset, scn->scn_phys.scn_queue_obj, ds->ds_phys->ds_prev_snap_obj, mintxg, tx) == 0); zfs_dbgmsg("snapshotting ds %llu; in queue; " "replacing with %llu", (u_longlong_t)ds->ds_object, (u_longlong_t)ds->ds_phys->ds_prev_snap_obj); } dsl_scan_sync_state(scn, tx); } void dsl_scan_ds_clone_swapped(dsl_dataset_t *ds1, dsl_dataset_t *ds2, dmu_tx_t *tx) { dsl_pool_t *dp = ds1->ds_dir->dd_pool; dsl_scan_t *scn = dp->dp_scan; uint64_t mintxg; if (scn->scn_phys.scn_state != DSS_SCANNING) return; if (scn->scn_phys.scn_bookmark.zb_objset == ds1->ds_object) { scn->scn_phys.scn_bookmark.zb_objset = ds2->ds_object; zfs_dbgmsg("clone_swap ds %llu; currently traversing; " "reset zb_objset to %llu", (u_longlong_t)ds1->ds_object, (u_longlong_t)ds2->ds_object); } else if (scn->scn_phys.scn_bookmark.zb_objset == ds2->ds_object) { scn->scn_phys.scn_bookmark.zb_objset = ds1->ds_object; zfs_dbgmsg("clone_swap ds %llu; currently traversing; " "reset zb_objset to %llu", (u_longlong_t)ds2->ds_object, (u_longlong_t)ds1->ds_object); } if (zap_lookup_int_key(dp->dp_meta_objset, scn->scn_phys.scn_queue_obj, ds1->ds_object, &mintxg) == 0) { int err; ASSERT3U(mintxg, ==, ds1->ds_phys->ds_prev_snap_txg); ASSERT3U(mintxg, ==, ds2->ds_phys->ds_prev_snap_txg); VERIFY3U(0, ==, zap_remove_int(dp->dp_meta_objset, scn->scn_phys.scn_queue_obj, ds1->ds_object, tx)); err = zap_add_int_key(dp->dp_meta_objset, scn->scn_phys.scn_queue_obj, ds2->ds_object, mintxg, tx); VERIFY(err == 0 || err == EEXIST); if (err == EEXIST) { /* Both were there to begin with */ VERIFY(0 == zap_add_int_key(dp->dp_meta_objset, scn->scn_phys.scn_queue_obj, ds1->ds_object, mintxg, tx)); } zfs_dbgmsg("clone_swap ds %llu; in queue; " "replacing with %llu", (u_longlong_t)ds1->ds_object, (u_longlong_t)ds2->ds_object); } else if (zap_lookup_int_key(dp->dp_meta_objset, scn->scn_phys.scn_queue_obj, ds2->ds_object, &mintxg) == 0) { ASSERT3U(mintxg, ==, ds1->ds_phys->ds_prev_snap_txg); ASSERT3U(mintxg, ==, ds2->ds_phys->ds_prev_snap_txg); VERIFY3U(0, ==, zap_remove_int(dp->dp_meta_objset, scn->scn_phys.scn_queue_obj, ds2->ds_object, tx)); VERIFY(0 == zap_add_int_key(dp->dp_meta_objset, scn->scn_phys.scn_queue_obj, ds1->ds_object, mintxg, tx)); zfs_dbgmsg("clone_swap ds %llu; in queue; " "replacing with %llu", (u_longlong_t)ds2->ds_object, (u_longlong_t)ds1->ds_object); } dsl_scan_sync_state(scn, tx); } struct enqueue_clones_arg { dmu_tx_t *tx; uint64_t originobj; }; /* ARGSUSED */ static int enqueue_clones_cb(dsl_pool_t *dp, dsl_dataset_t *hds, void *arg) { struct enqueue_clones_arg *eca = arg; dsl_dataset_t *ds; int err; dsl_scan_t *scn = dp->dp_scan; if (hds->ds_dir->dd_phys->dd_origin_obj != eca->originobj) return (0); err = dsl_dataset_hold_obj(dp, hds->ds_object, FTAG, &ds); if (err) return (err); while (ds->ds_phys->ds_prev_snap_obj != eca->originobj) { dsl_dataset_t *prev; err = dsl_dataset_hold_obj(dp, ds->ds_phys->ds_prev_snap_obj, FTAG, &prev); dsl_dataset_rele(ds, FTAG); if (err) return (err); ds = prev; } VERIFY(zap_add_int_key(dp->dp_meta_objset, scn->scn_phys.scn_queue_obj, ds->ds_object, ds->ds_phys->ds_prev_snap_txg, eca->tx) == 0); dsl_dataset_rele(ds, FTAG); return (0); } static void dsl_scan_visitds(dsl_scan_t *scn, uint64_t dsobj, dmu_tx_t *tx) { dsl_pool_t *dp = scn->scn_dp; dsl_dataset_t *ds; objset_t *os; char *dsname; VERIFY3U(0, ==, dsl_dataset_hold_obj(dp, dsobj, FTAG, &ds)); if (dmu_objset_from_ds(ds, &os)) goto out; /* * Only the ZIL in the head (non-snapshot) is valid. Even though * snapshots can have ZIL block pointers (which may be the same * BP as in the head), they must be ignored. So we traverse the * ZIL here, rather than in scan_recurse(), because the regular * snapshot block-sharing rules don't apply to it. */ if (DSL_SCAN_IS_SCRUB_RESILVER(scn) && !dsl_dataset_is_snapshot(ds)) dsl_scan_zil(dp, &os->os_zil_header); /* * Iterate over the bps in this ds. */ dmu_buf_will_dirty(ds->ds_dbuf, tx); dsl_scan_visit_rootbp(scn, ds, &ds->ds_phys->ds_bp, tx); dsname = kmem_alloc(ZFS_MAXNAMELEN, KM_PUSHPAGE); dsl_dataset_name(ds, dsname); zfs_dbgmsg("scanned dataset %llu (%s) with min=%llu max=%llu; " "pausing=%u", (longlong_t)dsobj, dsname, (longlong_t)scn->scn_phys.scn_cur_min_txg, (longlong_t)scn->scn_phys.scn_cur_max_txg, (int)scn->scn_pausing); kmem_free(dsname, ZFS_MAXNAMELEN); if (scn->scn_pausing) goto out; /* * We've finished this pass over this dataset. */ /* * If we did not completely visit this dataset, do another pass. */ if (scn->scn_phys.scn_flags & DSF_VISIT_DS_AGAIN) { zfs_dbgmsg("incomplete pass; visiting again"); scn->scn_phys.scn_flags &= ~DSF_VISIT_DS_AGAIN; VERIFY(zap_add_int_key(dp->dp_meta_objset, scn->scn_phys.scn_queue_obj, ds->ds_object, scn->scn_phys.scn_cur_max_txg, tx) == 0); goto out; } /* * Add descendent datasets to work queue. */ if (ds->ds_phys->ds_next_snap_obj != 0) { VERIFY(zap_add_int_key(dp->dp_meta_objset, scn->scn_phys.scn_queue_obj, ds->ds_phys->ds_next_snap_obj, ds->ds_phys->ds_creation_txg, tx) == 0); } if (ds->ds_phys->ds_num_children > 1) { boolean_t usenext = B_FALSE; if (ds->ds_phys->ds_next_clones_obj != 0) { uint64_t count; /* * A bug in a previous version of the code could * cause upgrade_clones_cb() to not set * ds_next_snap_obj when it should, leading to a * missing entry. Therefore we can only use the * next_clones_obj when its count is correct. */ int err = zap_count(dp->dp_meta_objset, ds->ds_phys->ds_next_clones_obj, &count); if (err == 0 && count == ds->ds_phys->ds_num_children - 1) usenext = B_TRUE; } if (usenext) { VERIFY0(zap_join_key(dp->dp_meta_objset, ds->ds_phys->ds_next_clones_obj, scn->scn_phys.scn_queue_obj, ds->ds_phys->ds_creation_txg, tx)); } else { struct enqueue_clones_arg eca; eca.tx = tx; eca.originobj = ds->ds_object; VERIFY0(dmu_objset_find_dp(dp, dp->dp_root_dir_obj, enqueue_clones_cb, &eca, DS_FIND_CHILDREN)); } } out: dsl_dataset_rele(ds, FTAG); } /* ARGSUSED */ static int enqueue_cb(dsl_pool_t *dp, dsl_dataset_t *hds, void *arg) { dmu_tx_t *tx = arg; dsl_dataset_t *ds; int err; dsl_scan_t *scn = dp->dp_scan; err = dsl_dataset_hold_obj(dp, hds->ds_object, FTAG, &ds); if (err) return (err); while (ds->ds_phys->ds_prev_snap_obj != 0) { dsl_dataset_t *prev; err = dsl_dataset_hold_obj(dp, ds->ds_phys->ds_prev_snap_obj, FTAG, &prev); if (err) { dsl_dataset_rele(ds, FTAG); return (err); } /* * If this is a clone, we don't need to worry about it for now. */ if (prev->ds_phys->ds_next_snap_obj != ds->ds_object) { dsl_dataset_rele(ds, FTAG); dsl_dataset_rele(prev, FTAG); return (0); } dsl_dataset_rele(ds, FTAG); ds = prev; } VERIFY(zap_add_int_key(dp->dp_meta_objset, scn->scn_phys.scn_queue_obj, ds->ds_object, ds->ds_phys->ds_prev_snap_txg, tx) == 0); dsl_dataset_rele(ds, FTAG); return (0); } /* * Scrub/dedup interaction. * * If there are N references to a deduped block, we don't want to scrub it * N times -- ideally, we should scrub it exactly once. * * We leverage the fact that the dde's replication class (enum ddt_class) * is ordered from highest replication class (DDT_CLASS_DITTO) to lowest * (DDT_CLASS_UNIQUE) so that we may walk the DDT in that order. * * To prevent excess scrubbing, the scrub begins by walking the DDT * to find all blocks with refcnt > 1, and scrubs each of these once. * Since there are two replication classes which contain blocks with * refcnt > 1, we scrub the highest replication class (DDT_CLASS_DITTO) first. * Finally the top-down scrub begins, only visiting blocks with refcnt == 1. * * There would be nothing more to say if a block's refcnt couldn't change * during a scrub, but of course it can so we must account for changes * in a block's replication class. * * Here's an example of what can occur: * * If a block has refcnt > 1 during the DDT scrub phase, but has refcnt == 1 * when visited during the top-down scrub phase, it will be scrubbed twice. * This negates our scrub optimization, but is otherwise harmless. * * If a block has refcnt == 1 during the DDT scrub phase, but has refcnt > 1 * on each visit during the top-down scrub phase, it will never be scrubbed. * To catch this, ddt_sync_entry() notifies the scrub code whenever a block's * reference class transitions to a higher level (i.e DDT_CLASS_UNIQUE to * DDT_CLASS_DUPLICATE); if it transitions from refcnt == 1 to refcnt > 1 * while a scrub is in progress, it scrubs the block right then. */ static void dsl_scan_ddt(dsl_scan_t *scn, dmu_tx_t *tx) { ddt_bookmark_t *ddb = &scn->scn_phys.scn_ddt_bookmark; ddt_entry_t dde; int error; uint64_t n = 0; bzero(&dde, sizeof (ddt_entry_t)); while ((error = ddt_walk(scn->scn_dp->dp_spa, ddb, &dde)) == 0) { ddt_t *ddt; if (ddb->ddb_class > scn->scn_phys.scn_ddt_class_max) break; dprintf("visiting ddb=%llu/%llu/%llu/%llx\n", (longlong_t)ddb->ddb_class, (longlong_t)ddb->ddb_type, (longlong_t)ddb->ddb_checksum, (longlong_t)ddb->ddb_cursor); /* There should be no pending changes to the dedup table */ ddt = scn->scn_dp->dp_spa->spa_ddt[ddb->ddb_checksum]; ASSERT(avl_first(&ddt->ddt_tree) == NULL); dsl_scan_ddt_entry(scn, ddb->ddb_checksum, &dde, tx); n++; if (dsl_scan_check_pause(scn, NULL)) break; } zfs_dbgmsg("scanned %llu ddt entries with class_max = %u; pausing=%u", (longlong_t)n, (int)scn->scn_phys.scn_ddt_class_max, (int)scn->scn_pausing); ASSERT(error == 0 || error == ENOENT); ASSERT(error != ENOENT || ddb->ddb_class > scn->scn_phys.scn_ddt_class_max); } /* ARGSUSED */ void dsl_scan_ddt_entry(dsl_scan_t *scn, enum zio_checksum checksum, ddt_entry_t *dde, dmu_tx_t *tx) { const ddt_key_t *ddk = &dde->dde_key; ddt_phys_t *ddp = dde->dde_phys; blkptr_t bp; zbookmark_t zb = { 0 }; int p; if (scn->scn_phys.scn_state != DSS_SCANNING) return; for (p = 0; p < DDT_PHYS_TYPES; p++, ddp++) { if (ddp->ddp_phys_birth == 0 || ddp->ddp_phys_birth > scn->scn_phys.scn_max_txg) continue; ddt_bp_create(checksum, ddk, ddp, &bp); scn->scn_visited_this_txg++; scan_funcs[scn->scn_phys.scn_func](scn->scn_dp, &bp, &zb); } } static void dsl_scan_visit(dsl_scan_t *scn, dmu_tx_t *tx) { dsl_pool_t *dp = scn->scn_dp; zap_cursor_t *zc; zap_attribute_t *za; if (scn->scn_phys.scn_ddt_bookmark.ddb_class <= scn->scn_phys.scn_ddt_class_max) { scn->scn_phys.scn_cur_min_txg = scn->scn_phys.scn_min_txg; scn->scn_phys.scn_cur_max_txg = scn->scn_phys.scn_max_txg; dsl_scan_ddt(scn, tx); if (scn->scn_pausing) return; } if (scn->scn_phys.scn_bookmark.zb_objset == DMU_META_OBJSET) { /* First do the MOS & ORIGIN */ scn->scn_phys.scn_cur_min_txg = scn->scn_phys.scn_min_txg; scn->scn_phys.scn_cur_max_txg = scn->scn_phys.scn_max_txg; dsl_scan_visit_rootbp(scn, NULL, &dp->dp_meta_rootbp, tx); spa_set_rootblkptr(dp->dp_spa, &dp->dp_meta_rootbp); if (scn->scn_pausing) return; if (spa_version(dp->dp_spa) < SPA_VERSION_DSL_SCRUB) { VERIFY0(dmu_objset_find_dp(dp, dp->dp_root_dir_obj, enqueue_cb, tx, DS_FIND_CHILDREN)); } else { dsl_scan_visitds(scn, dp->dp_origin_snap->ds_object, tx); } ASSERT(!scn->scn_pausing); } else if (scn->scn_phys.scn_bookmark.zb_objset != ZB_DESTROYED_OBJSET) { /* * If we were paused, continue from here. Note if the * ds we were paused on was deleted, the zb_objset may * be -1, so we will skip this and find a new objset * below. */ dsl_scan_visitds(scn, scn->scn_phys.scn_bookmark.zb_objset, tx); if (scn->scn_pausing) return; } /* * In case we were paused right at the end of the ds, zero the * bookmark so we don't think that we're still trying to resume. */ bzero(&scn->scn_phys.scn_bookmark, sizeof (zbookmark_t)); zc = kmem_alloc(sizeof (zap_cursor_t), KM_PUSHPAGE); za = kmem_alloc(sizeof (zap_attribute_t), KM_PUSHPAGE); /* keep pulling things out of the zap-object-as-queue */ while (zap_cursor_init(zc, dp->dp_meta_objset, scn->scn_phys.scn_queue_obj), zap_cursor_retrieve(zc, za) == 0) { dsl_dataset_t *ds; uint64_t dsobj; dsobj = strtonum(za->za_name, NULL); VERIFY3U(0, ==, zap_remove_int(dp->dp_meta_objset, scn->scn_phys.scn_queue_obj, dsobj, tx)); /* Set up min/max txg */ VERIFY3U(0, ==, dsl_dataset_hold_obj(dp, dsobj, FTAG, &ds)); if (za->za_first_integer != 0) { scn->scn_phys.scn_cur_min_txg = MAX(scn->scn_phys.scn_min_txg, za->za_first_integer); } else { scn->scn_phys.scn_cur_min_txg = MAX(scn->scn_phys.scn_min_txg, ds->ds_phys->ds_prev_snap_txg); } scn->scn_phys.scn_cur_max_txg = dsl_scan_ds_maxtxg(ds); dsl_dataset_rele(ds, FTAG); dsl_scan_visitds(scn, dsobj, tx); zap_cursor_fini(zc); if (scn->scn_pausing) goto out; } zap_cursor_fini(zc); out: kmem_free(za, sizeof (zap_attribute_t)); kmem_free(zc, sizeof (zap_cursor_t)); } static boolean_t dsl_scan_free_should_pause(dsl_scan_t *scn) { uint64_t elapsed_nanosecs; if (zfs_recover) return (B_FALSE); elapsed_nanosecs = gethrtime() - scn->scn_sync_start_time; return (elapsed_nanosecs / NANOSEC > zfs_txg_timeout || (NSEC2MSEC(elapsed_nanosecs) > zfs_free_min_time_ms && txg_sync_waiting(scn->scn_dp)) || spa_shutting_down(scn->scn_dp->dp_spa)); } static int dsl_scan_free_block_cb(void *arg, const blkptr_t *bp, dmu_tx_t *tx) { dsl_scan_t *scn = arg; if (!scn->scn_is_bptree || (BP_GET_LEVEL(bp) == 0 && BP_GET_TYPE(bp) != DMU_OT_OBJSET)) { if (dsl_scan_free_should_pause(scn)) return (SET_ERROR(ERESTART)); } zio_nowait(zio_free_sync(scn->scn_zio_root, scn->scn_dp->dp_spa, dmu_tx_get_txg(tx), bp, 0)); dsl_dir_diduse_space(tx->tx_pool->dp_free_dir, DD_USED_HEAD, -bp_get_dsize_sync(scn->scn_dp->dp_spa, bp), -BP_GET_PSIZE(bp), -BP_GET_UCSIZE(bp), tx); scn->scn_visited_this_txg++; return (0); } boolean_t dsl_scan_active(dsl_scan_t *scn) { spa_t *spa = scn->scn_dp->dp_spa; uint64_t used = 0, comp, uncomp; if (spa->spa_load_state != SPA_LOAD_NONE) return (B_FALSE); if (spa_shutting_down(spa)) return (B_FALSE); if (scn->scn_phys.scn_state == DSS_SCANNING || - scn->scn_async_destroying) + (scn->scn_async_destroying && !scn->scn_async_stalled)) return (B_TRUE); if (spa_version(scn->scn_dp->dp_spa) >= SPA_VERSION_DEADLISTS) { (void) bpobj_space(&scn->scn_dp->dp_free_bpobj, &used, &comp, &uncomp); } return (used != 0); } void dsl_scan_sync(dsl_pool_t *dp, dmu_tx_t *tx) { dsl_scan_t *scn = dp->dp_scan; spa_t *spa = dp->dp_spa; - int err; + int err = 0; /* * Check for scn_restart_txg before checking spa_load_state, so * that we can restart an old-style scan while the pool is being * imported (see dsl_scan_init). */ if (scn->scn_restart_txg != 0 && scn->scn_restart_txg <= tx->tx_txg) { pool_scan_func_t func = POOL_SCAN_SCRUB; dsl_scan_done(scn, B_FALSE, tx); if (vdev_resilver_needed(spa->spa_root_vdev, NULL, NULL)) func = POOL_SCAN_RESILVER; zfs_dbgmsg("restarting scan func=%u txg=%llu", func, tx->tx_txg); dsl_scan_setup_sync(&func, tx); } - if (!dsl_scan_active(scn) || + /* + * If the scan is inactive due to a stalled async destroy, try again. + */ + if ((!scn->scn_async_stalled && !dsl_scan_active(scn)) || spa_sync_pass(dp->dp_spa) > 1) return; scn->scn_visited_this_txg = 0; scn->scn_pausing = B_FALSE; scn->scn_sync_start_time = gethrtime(); spa->spa_scrub_active = B_TRUE; /* - * First process the free list. If we pause the free, don't do - * any scanning. This ensures that there is no free list when - * we are scanning, so the scan code doesn't have to worry about - * traversing it. + * First process the async destroys. If we pause, don't do + * any scrubbing or resilvering. This ensures that there are no + * async destroys while we are scanning, so the scan code doesn't + * have to worry about traversing it. It is also faster to free the + * blocks than to scrub them. */ if (spa_version(dp->dp_spa) >= SPA_VERSION_DEADLISTS) { scn->scn_is_bptree = B_FALSE; scn->scn_zio_root = zio_root(dp->dp_spa, NULL, NULL, ZIO_FLAG_MUSTSUCCEED); err = bpobj_iterate(&dp->dp_free_bpobj, dsl_scan_free_block_cb, scn, tx); VERIFY3U(0, ==, zio_wait(scn->scn_zio_root)); - if (err == 0 && spa_feature_is_active(spa, - SPA_FEATURE_ASYNC_DESTROY)) { - ASSERT(scn->scn_async_destroying); - scn->scn_is_bptree = B_TRUE; - scn->scn_zio_root = zio_root(dp->dp_spa, NULL, - NULL, ZIO_FLAG_MUSTSUCCEED); - err = bptree_iterate(dp->dp_meta_objset, - dp->dp_bptree_obj, B_TRUE, dsl_scan_free_block_cb, - scn, tx); - VERIFY0(zio_wait(scn->scn_zio_root)); + if (err != 0 && err != ERESTART) + zfs_panic_recover("error %u from bpobj_iterate()", err); + } - if (err == 0) { - /* finished; deactivate async destroy feature */ - spa_feature_decr(spa, SPA_FEATURE_ASYNC_DESTROY, - tx); - ASSERT(!spa_feature_is_active(spa, - SPA_FEATURE_ASYNC_DESTROY)); - VERIFY0(zap_remove(dp->dp_meta_objset, - DMU_POOL_DIRECTORY_OBJECT, - DMU_POOL_BPTREE_OBJ, tx)); - VERIFY0(bptree_free(dp->dp_meta_objset, - dp->dp_bptree_obj, tx)); - dp->dp_bptree_obj = 0; - scn->scn_async_destroying = B_FALSE; - } + if (err == 0 && spa_feature_is_active(spa, SPA_FEATURE_ASYNC_DESTROY)) { + ASSERT(scn->scn_async_destroying); + scn->scn_is_bptree = B_TRUE; + scn->scn_zio_root = zio_root(dp->dp_spa, NULL, + NULL, ZIO_FLAG_MUSTSUCCEED); + err = bptree_iterate(dp->dp_meta_objset, + dp->dp_bptree_obj, B_TRUE, dsl_scan_free_block_cb, scn, tx); + VERIFY0(zio_wait(scn->scn_zio_root)); + + if (err == EIO || err == ECKSUM) { + err = 0; + } else if (err != 0 && err != ERESTART) { + zfs_panic_recover("error %u from " + "traverse_dataset_destroyed()", err); } - if (scn->scn_visited_this_txg) { - zfs_dbgmsg("freed %llu blocks in %llums from " - "free_bpobj/bptree txg %llu", - (longlong_t)scn->scn_visited_this_txg, - (longlong_t) - NSEC2MSEC(gethrtime() - scn->scn_sync_start_time), - (longlong_t)tx->tx_txg); - scn->scn_visited_this_txg = 0; - /* - * Re-sync the ddt so that we can further modify - * it when doing bprewrite. - */ - ddt_sync(spa, tx->tx_txg); + + /* + * If we didn't make progress, mark the async destroy as + * stalled, so that we will not initiate a spa_sync() on + * its behalf. + */ + scn->scn_async_stalled = (scn->scn_visited_this_txg == 0); + + if (bptree_is_empty(dp->dp_meta_objset, dp->dp_bptree_obj)) { + /* finished; deactivate async destroy feature */ + spa_feature_decr(spa, SPA_FEATURE_ASYNC_DESTROY, tx); + ASSERT(!spa_feature_is_active(spa, + SPA_FEATURE_ASYNC_DESTROY)); + VERIFY0(zap_remove(dp->dp_meta_objset, + DMU_POOL_DIRECTORY_OBJECT, + DMU_POOL_BPTREE_OBJ, tx)); + VERIFY0(bptree_free(dp->dp_meta_objset, + dp->dp_bptree_obj, tx)); + dp->dp_bptree_obj = 0; + scn->scn_async_destroying = B_FALSE; } - if (err == ERESTART) - return; + } + if (scn->scn_visited_this_txg) { + zfs_dbgmsg("freed %llu blocks in %llums from " + "free_bpobj/bptree txg %llu; err=%u", + (longlong_t)scn->scn_visited_this_txg, + (longlong_t) + NSEC2MSEC(gethrtime() - scn->scn_sync_start_time), + (longlong_t)tx->tx_txg, err); + scn->scn_visited_this_txg = 0; + + /* + * Write out changes to the DDT that may be required as a + * result of the blocks freed. This ensures that the DDT + * is clean when a scrub/resilver runs. + */ + ddt_sync(spa, tx->tx_txg); + } + if (err != 0) + return; + if (!scn->scn_async_destroying && zfs_free_leak_on_eio && + (dp->dp_free_dir->dd_phys->dd_used_bytes != 0 || + dp->dp_free_dir->dd_phys->dd_compressed_bytes != 0 || + dp->dp_free_dir->dd_phys->dd_uncompressed_bytes != 0)) { + /* + * We have finished background destroying, but there is still + * some space left in the dp_free_dir. Transfer this leaked + * space to the dp_leak_dir. + */ + if (dp->dp_leak_dir == NULL) { + rrw_enter(&dp->dp_config_rwlock, RW_WRITER, FTAG); + (void) dsl_dir_create_sync(dp, dp->dp_root_dir, + LEAK_DIR_NAME, tx); + VERIFY0(dsl_pool_open_special_dir(dp, + LEAK_DIR_NAME, &dp->dp_leak_dir)); + rrw_exit(&dp->dp_config_rwlock, FTAG); + } + dsl_dir_diduse_space(dp->dp_leak_dir, DD_USED_HEAD, + dp->dp_free_dir->dd_phys->dd_used_bytes, + dp->dp_free_dir->dd_phys->dd_compressed_bytes, + dp->dp_free_dir->dd_phys->dd_uncompressed_bytes, tx); + dsl_dir_diduse_space(dp->dp_free_dir, DD_USED_HEAD, + -dp->dp_free_dir->dd_phys->dd_used_bytes, + -dp->dp_free_dir->dd_phys->dd_compressed_bytes, + -dp->dp_free_dir->dd_phys->dd_uncompressed_bytes, tx); + } + if (!scn->scn_async_destroying) { /* finished; verify that space accounting went to zero */ ASSERT0(dp->dp_free_dir->dd_phys->dd_used_bytes); ASSERT0(dp->dp_free_dir->dd_phys->dd_compressed_bytes); ASSERT0(dp->dp_free_dir->dd_phys->dd_uncompressed_bytes); } if (scn->scn_phys.scn_state != DSS_SCANNING) return; if (scn->scn_done_txg == tx->tx_txg) { ASSERT(!scn->scn_pausing); /* finished with scan. */ zfs_dbgmsg("txg %llu scan complete", tx->tx_txg); dsl_scan_done(scn, B_TRUE, tx); ASSERT3U(spa->spa_scrub_inflight, ==, 0); dsl_scan_sync_state(scn, tx); return; } if (scn->scn_phys.scn_ddt_bookmark.ddb_class <= scn->scn_phys.scn_ddt_class_max) { zfs_dbgmsg("doing scan sync txg %llu; " "ddt bm=%llu/%llu/%llu/%llx", (longlong_t)tx->tx_txg, (longlong_t)scn->scn_phys.scn_ddt_bookmark.ddb_class, (longlong_t)scn->scn_phys.scn_ddt_bookmark.ddb_type, (longlong_t)scn->scn_phys.scn_ddt_bookmark.ddb_checksum, (longlong_t)scn->scn_phys.scn_ddt_bookmark.ddb_cursor); ASSERT(scn->scn_phys.scn_bookmark.zb_objset == 0); ASSERT(scn->scn_phys.scn_bookmark.zb_object == 0); ASSERT(scn->scn_phys.scn_bookmark.zb_level == 0); ASSERT(scn->scn_phys.scn_bookmark.zb_blkid == 0); } else { zfs_dbgmsg("doing scan sync txg %llu; bm=%llu/%llu/%llu/%llu", (longlong_t)tx->tx_txg, (longlong_t)scn->scn_phys.scn_bookmark.zb_objset, (longlong_t)scn->scn_phys.scn_bookmark.zb_object, (longlong_t)scn->scn_phys.scn_bookmark.zb_level, (longlong_t)scn->scn_phys.scn_bookmark.zb_blkid); } scn->scn_zio_root = zio_root(dp->dp_spa, NULL, NULL, ZIO_FLAG_CANFAIL); dsl_pool_config_enter(dp, FTAG); dsl_scan_visit(scn, tx); dsl_pool_config_exit(dp, FTAG); (void) zio_wait(scn->scn_zio_root); scn->scn_zio_root = NULL; zfs_dbgmsg("visited %llu blocks in %llums", (longlong_t)scn->scn_visited_this_txg, (longlong_t)NSEC2MSEC(gethrtime() - scn->scn_sync_start_time)); if (!scn->scn_pausing) { scn->scn_done_txg = tx->tx_txg + 1; zfs_dbgmsg("txg %llu traversal complete, waiting till txg %llu", tx->tx_txg, scn->scn_done_txg); } if (DSL_SCAN_IS_SCRUB_RESILVER(scn)) { mutex_enter(&spa->spa_scrub_lock); while (spa->spa_scrub_inflight > 0) { cv_wait(&spa->spa_scrub_io_cv, &spa->spa_scrub_lock); } mutex_exit(&spa->spa_scrub_lock); } dsl_scan_sync_state(scn, tx); } /* * This will start a new scan, or restart an existing one. */ void dsl_resilver_restart(dsl_pool_t *dp, uint64_t txg) { if (txg == 0) { dmu_tx_t *tx; tx = dmu_tx_create_dd(dp->dp_mos_dir); VERIFY(0 == dmu_tx_assign(tx, TXG_WAIT)); txg = dmu_tx_get_txg(tx); dp->dp_scan->scn_restart_txg = txg; dmu_tx_commit(tx); } else { dp->dp_scan->scn_restart_txg = txg; } zfs_dbgmsg("restarting resilver txg=%llu", txg); } boolean_t dsl_scan_resilvering(dsl_pool_t *dp) { return (dp->dp_scan->scn_phys.scn_state == DSS_SCANNING && dp->dp_scan->scn_phys.scn_func == POOL_SCAN_RESILVER); } /* * scrub consumers */ static void count_block(zfs_all_blkstats_t *zab, const blkptr_t *bp) { int i; /* * If we resume after a reboot, zab will be NULL; don't record * incomplete stats in that case. */ if (zab == NULL) return; for (i = 0; i < 4; i++) { int l = (i < 2) ? BP_GET_LEVEL(bp) : DN_MAX_LEVELS; int t = (i & 1) ? BP_GET_TYPE(bp) : DMU_OT_TOTAL; int equal; zfs_blkstat_t *zb; if (t & DMU_OT_NEWTYPE) t = DMU_OT_OTHER; zb = &zab->zab_type[l][t]; zb->zb_count++; zb->zb_asize += BP_GET_ASIZE(bp); zb->zb_lsize += BP_GET_LSIZE(bp); zb->zb_psize += BP_GET_PSIZE(bp); zb->zb_gangs += BP_COUNT_GANG(bp); switch (BP_GET_NDVAS(bp)) { case 2: if (DVA_GET_VDEV(&bp->blk_dva[0]) == DVA_GET_VDEV(&bp->blk_dva[1])) zb->zb_ditto_2_of_2_samevdev++; break; case 3: equal = (DVA_GET_VDEV(&bp->blk_dva[0]) == DVA_GET_VDEV(&bp->blk_dva[1])) + (DVA_GET_VDEV(&bp->blk_dva[0]) == DVA_GET_VDEV(&bp->blk_dva[2])) + (DVA_GET_VDEV(&bp->blk_dva[1]) == DVA_GET_VDEV(&bp->blk_dva[2])); if (equal == 1) zb->zb_ditto_2_of_3_samevdev++; else if (equal == 3) zb->zb_ditto_3_of_3_samevdev++; break; } } } static void dsl_scan_scrub_done(zio_t *zio) { spa_t *spa = zio->io_spa; zio_data_buf_free(zio->io_data, zio->io_size); mutex_enter(&spa->spa_scrub_lock); spa->spa_scrub_inflight--; cv_broadcast(&spa->spa_scrub_io_cv); if (zio->io_error && (zio->io_error != ECKSUM || !(zio->io_flags & ZIO_FLAG_SPECULATIVE))) { spa->spa_dsl_pool->dp_scan->scn_phys.scn_errors++; } mutex_exit(&spa->spa_scrub_lock); } static int dsl_scan_scrub_cb(dsl_pool_t *dp, const blkptr_t *bp, const zbookmark_t *zb) { dsl_scan_t *scn = dp->dp_scan; size_t size = BP_GET_PSIZE(bp); spa_t *spa = dp->dp_spa; uint64_t phys_birth = BP_PHYSICAL_BIRTH(bp); boolean_t needs_io = B_FALSE; int zio_flags = ZIO_FLAG_SCAN_THREAD | ZIO_FLAG_RAW | ZIO_FLAG_CANFAIL; int scan_delay = 0; int d; if (phys_birth <= scn->scn_phys.scn_min_txg || phys_birth >= scn->scn_phys.scn_max_txg) return (0); count_block(dp->dp_blkstats, bp); if (BP_IS_EMBEDDED(bp)) return (0); ASSERT(DSL_SCAN_IS_SCRUB_RESILVER(scn)); if (scn->scn_phys.scn_func == POOL_SCAN_SCRUB) { zio_flags |= ZIO_FLAG_SCRUB; needs_io = B_TRUE; scan_delay = zfs_scrub_delay; } else { ASSERT3U(scn->scn_phys.scn_func, ==, POOL_SCAN_RESILVER); zio_flags |= ZIO_FLAG_RESILVER; needs_io = B_FALSE; scan_delay = zfs_resilver_delay; } /* If it's an intent log block, failure is expected. */ if (zb->zb_level == ZB_ZIL_LEVEL) zio_flags |= ZIO_FLAG_SPECULATIVE; for (d = 0; d < BP_GET_NDVAS(bp); d++) { vdev_t *vd = vdev_lookup_top(spa, DVA_GET_VDEV(&bp->blk_dva[d])); /* * Keep track of how much data we've examined so that * zpool(1M) status can make useful progress reports. */ scn->scn_phys.scn_examined += DVA_GET_ASIZE(&bp->blk_dva[d]); spa->spa_scan_pass_exam += DVA_GET_ASIZE(&bp->blk_dva[d]); /* if it's a resilver, this may not be in the target range */ if (!needs_io) { if (DVA_GET_GANG(&bp->blk_dva[d])) { /* * Gang members may be spread across multiple * vdevs, so the best estimate we have is the * scrub range, which has already been checked. * XXX -- it would be better to change our * allocation policy to ensure that all * gang members reside on the same vdev. */ needs_io = B_TRUE; } else { needs_io = vdev_dtl_contains(vd, DTL_PARTIAL, phys_birth, 1); } } } if (needs_io && !zfs_no_scrub_io) { vdev_t *rvd = spa->spa_root_vdev; uint64_t maxinflight = rvd->vdev_children * zfs_top_maxinflight; void *data = zio_data_buf_alloc(size); mutex_enter(&spa->spa_scrub_lock); while (spa->spa_scrub_inflight >= maxinflight) cv_wait(&spa->spa_scrub_io_cv, &spa->spa_scrub_lock); spa->spa_scrub_inflight++; mutex_exit(&spa->spa_scrub_lock); /* * If we're seeing recent (zfs_scan_idle) "important" I/Os * then throttle our workload to limit the impact of a scan. */ if (ddi_get_lbolt64() - spa->spa_last_io <= zfs_scan_idle) delay(scan_delay); zio_nowait(zio_read(NULL, spa, bp, data, size, dsl_scan_scrub_done, NULL, ZIO_PRIORITY_SCRUB, zio_flags, zb)); } /* do not relocate this block */ return (0); } int dsl_scan(dsl_pool_t *dp, pool_scan_func_t func) { spa_t *spa = dp->dp_spa; /* * Purge all vdev caches and probe all devices. We do this here * rather than in sync context because this requires a writer lock * on the spa_config lock, which we can't do from sync context. The * spa_scrub_reopen flag indicates that vdev_open() should not * attempt to start another scrub. */ spa_vdev_state_enter(spa, SCL_NONE); spa->spa_scrub_reopen = B_TRUE; vdev_reopen(spa->spa_root_vdev); spa->spa_scrub_reopen = B_FALSE; (void) spa_vdev_state_exit(spa, NULL, 0); return (dsl_sync_task(spa_name(spa), dsl_scan_setup_check, dsl_scan_setup_sync, &func, 0)); } #if defined(_KERNEL) && defined(HAVE_SPL) module_param(zfs_top_maxinflight, int, 0644); MODULE_PARM_DESC(zfs_top_maxinflight, "Max I/Os per top-level"); module_param(zfs_resilver_delay, int, 0644); MODULE_PARM_DESC(zfs_resilver_delay, "Number of ticks to delay resilver"); module_param(zfs_scrub_delay, int, 0644); MODULE_PARM_DESC(zfs_scrub_delay, "Number of ticks to delay scrub"); module_param(zfs_scan_idle, int, 0644); MODULE_PARM_DESC(zfs_scan_idle, "Idle window in clock ticks"); module_param(zfs_scan_min_time_ms, int, 0644); MODULE_PARM_DESC(zfs_scan_min_time_ms, "Min millisecs to scrub per txg"); module_param(zfs_free_min_time_ms, int, 0644); MODULE_PARM_DESC(zfs_free_min_time_ms, "Min millisecs to free per txg"); module_param(zfs_resilver_min_time_ms, int, 0644); MODULE_PARM_DESC(zfs_resilver_min_time_ms, "Min millisecs to resilver per txg"); module_param(zfs_no_scrub_io, int, 0644); MODULE_PARM_DESC(zfs_no_scrub_io, "Set to disable scrub I/O"); module_param(zfs_no_scrub_prefetch, int, 0644); MODULE_PARM_DESC(zfs_no_scrub_prefetch, "Set to disable scrub prefetching"); #endif diff --git a/module/zfs/spa.c b/module/zfs/spa.c index b9fa45f8299e..e4b71ea72d4f 100644 --- a/module/zfs/spa.c +++ b/module/zfs/spa.c @@ -1,6606 +1,6612 @@ /* * 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 by Delphix. All rights reserved. * Copyright 2013 Nexenta Systems, Inc. All rights reserved. */ /* * SPA: Storage Pool Allocator * * This file contains all the routines used when modifying on-disk SPA state. * This includes opening, importing, destroying, exporting a pool, and syncing a * pool. */ #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 #include #include #include #include #include #include #include #include #ifdef _KERNEL #include #include #include #include #include #include #endif /* _KERNEL */ #include "zfs_prop.h" #include "zfs_comutil.h" typedef enum zti_modes { ZTI_MODE_FIXED, /* value is # of threads (min 1) */ ZTI_MODE_BATCH, /* cpu-intensive; value is ignored */ ZTI_MODE_NULL, /* don't create a taskq */ ZTI_NMODES } zti_modes_t; #define ZTI_P(n, q) { ZTI_MODE_FIXED, (n), (q) } #define ZTI_PCT(n) { ZTI_MODE_ONLINE_PERCENT, (n), 1 } #define ZTI_BATCH { ZTI_MODE_BATCH, 0, 1 } #define ZTI_NULL { ZTI_MODE_NULL, 0, 0 } #define ZTI_N(n) ZTI_P(n, 1) #define ZTI_ONE ZTI_N(1) typedef struct zio_taskq_info { zti_modes_t zti_mode; uint_t zti_value; uint_t zti_count; } zio_taskq_info_t; static const char *const zio_taskq_types[ZIO_TASKQ_TYPES] = { "iss", "iss_h", "int", "int_h" }; /* * This table defines the taskq settings for each ZFS I/O type. When * initializing a pool, we use this table to create an appropriately sized * taskq. Some operations are low volume and therefore have a small, static * number of threads assigned to their taskqs using the ZTI_N(#) or ZTI_ONE * macros. Other operations process a large amount of data; the ZTI_BATCH * macro causes us to create a taskq oriented for throughput. Some operations * are so high frequency and short-lived that the taskq itself can become a a * point of lock contention. The ZTI_P(#, #) macro indicates that we need an * additional degree of parallelism specified by the number of threads per- * taskq and the number of taskqs; when dispatching an event in this case, the * particular taskq is chosen at random. * * The different taskq priorities are to handle the different contexts (issue * and interrupt) and then to reserve threads for ZIO_PRIORITY_NOW I/Os that * need to be handled with minimum delay. */ const zio_taskq_info_t zio_taskqs[ZIO_TYPES][ZIO_TASKQ_TYPES] = { /* ISSUE ISSUE_HIGH INTR INTR_HIGH */ { ZTI_ONE, ZTI_NULL, ZTI_ONE, ZTI_NULL }, /* NULL */ { ZTI_N(8), ZTI_NULL, ZTI_BATCH, ZTI_NULL }, /* READ */ { ZTI_BATCH, ZTI_N(5), ZTI_N(16), ZTI_N(5) }, /* WRITE */ { ZTI_P(4, 8), ZTI_NULL, ZTI_ONE, ZTI_NULL }, /* FREE */ { ZTI_ONE, ZTI_NULL, ZTI_ONE, ZTI_NULL }, /* CLAIM */ { ZTI_ONE, ZTI_NULL, ZTI_ONE, ZTI_NULL }, /* IOCTL */ }; static void spa_sync_version(void *arg, dmu_tx_t *tx); static void spa_sync_props(void *arg, dmu_tx_t *tx); static boolean_t spa_has_active_shared_spare(spa_t *spa); static inline int spa_load_impl(spa_t *spa, uint64_t, nvlist_t *config, spa_load_state_t state, spa_import_type_t type, boolean_t mosconfig, char **ereport); static void spa_vdev_resilver_done(spa_t *spa); uint_t zio_taskq_batch_pct = 75; /* 1 thread per cpu in pset */ id_t zio_taskq_psrset_bind = PS_NONE; boolean_t zio_taskq_sysdc = B_TRUE; /* use SDC scheduling class */ uint_t zio_taskq_basedc = 80; /* base duty cycle */ boolean_t spa_create_process = B_TRUE; /* no process ==> no sysdc */ /* * This (illegal) pool name is used when temporarily importing a spa_t in order * to get the vdev stats associated with the imported devices. */ #define TRYIMPORT_NAME "$import" /* * ========================================================================== * SPA properties routines * ========================================================================== */ /* * Add a (source=src, propname=propval) list to an nvlist. */ static void spa_prop_add_list(nvlist_t *nvl, zpool_prop_t prop, char *strval, uint64_t intval, zprop_source_t src) { const char *propname = zpool_prop_to_name(prop); nvlist_t *propval; VERIFY(nvlist_alloc(&propval, NV_UNIQUE_NAME, KM_PUSHPAGE) == 0); VERIFY(nvlist_add_uint64(propval, ZPROP_SOURCE, src) == 0); if (strval != NULL) VERIFY(nvlist_add_string(propval, ZPROP_VALUE, strval) == 0); else VERIFY(nvlist_add_uint64(propval, ZPROP_VALUE, intval) == 0); VERIFY(nvlist_add_nvlist(nvl, propname, propval) == 0); nvlist_free(propval); } /* * Get property values from the spa configuration. */ static void spa_prop_get_config(spa_t *spa, nvlist_t **nvp) { vdev_t *rvd = spa->spa_root_vdev; dsl_pool_t *pool = spa->spa_dsl_pool; uint64_t size; uint64_t alloc; uint64_t space; uint64_t cap, version; zprop_source_t src = ZPROP_SRC_NONE; spa_config_dirent_t *dp; int c; ASSERT(MUTEX_HELD(&spa->spa_props_lock)); if (rvd != NULL) { alloc = metaslab_class_get_alloc(spa_normal_class(spa)); size = metaslab_class_get_space(spa_normal_class(spa)); spa_prop_add_list(*nvp, ZPOOL_PROP_NAME, spa_name(spa), 0, src); spa_prop_add_list(*nvp, ZPOOL_PROP_SIZE, NULL, size, src); spa_prop_add_list(*nvp, ZPOOL_PROP_ALLOCATED, NULL, alloc, src); spa_prop_add_list(*nvp, ZPOOL_PROP_FREE, NULL, size - alloc, src); space = 0; for (c = 0; c < rvd->vdev_children; c++) { vdev_t *tvd = rvd->vdev_child[c]; space += tvd->vdev_max_asize - tvd->vdev_asize; } spa_prop_add_list(*nvp, ZPOOL_PROP_EXPANDSZ, NULL, space, src); spa_prop_add_list(*nvp, ZPOOL_PROP_READONLY, NULL, (spa_mode(spa) == FREAD), src); cap = (size == 0) ? 0 : (alloc * 100 / size); spa_prop_add_list(*nvp, ZPOOL_PROP_CAPACITY, NULL, cap, src); spa_prop_add_list(*nvp, ZPOOL_PROP_DEDUPRATIO, NULL, ddt_get_pool_dedup_ratio(spa), src); spa_prop_add_list(*nvp, ZPOOL_PROP_HEALTH, NULL, rvd->vdev_state, src); version = spa_version(spa); if (version == zpool_prop_default_numeric(ZPOOL_PROP_VERSION)) src = ZPROP_SRC_DEFAULT; else src = ZPROP_SRC_LOCAL; spa_prop_add_list(*nvp, ZPOOL_PROP_VERSION, NULL, version, src); } if (pool != NULL) { - dsl_dir_t *freedir = pool->dp_free_dir; - /* * The $FREE directory was introduced in SPA_VERSION_DEADLISTS, * when opening pools before this version freedir will be NULL. */ - if (freedir != NULL) { + if (pool->dp_free_dir != NULL) { spa_prop_add_list(*nvp, ZPOOL_PROP_FREEING, NULL, - freedir->dd_phys->dd_used_bytes, src); + pool->dp_free_dir->dd_phys->dd_used_bytes, src); } else { spa_prop_add_list(*nvp, ZPOOL_PROP_FREEING, NULL, 0, src); } + + if (pool->dp_leak_dir != NULL) { + spa_prop_add_list(*nvp, ZPOOL_PROP_LEAKED, NULL, + pool->dp_leak_dir->dd_phys->dd_used_bytes, src); + } else { + spa_prop_add_list(*nvp, ZPOOL_PROP_LEAKED, + NULL, 0, src); + } } spa_prop_add_list(*nvp, ZPOOL_PROP_GUID, NULL, spa_guid(spa), src); if (spa->spa_comment != NULL) { spa_prop_add_list(*nvp, ZPOOL_PROP_COMMENT, spa->spa_comment, 0, ZPROP_SRC_LOCAL); } if (spa->spa_root != NULL) spa_prop_add_list(*nvp, ZPOOL_PROP_ALTROOT, spa->spa_root, 0, ZPROP_SRC_LOCAL); if ((dp = list_head(&spa->spa_config_list)) != NULL) { if (dp->scd_path == NULL) { spa_prop_add_list(*nvp, ZPOOL_PROP_CACHEFILE, "none", 0, ZPROP_SRC_LOCAL); } else if (strcmp(dp->scd_path, spa_config_path) != 0) { spa_prop_add_list(*nvp, ZPOOL_PROP_CACHEFILE, dp->scd_path, 0, ZPROP_SRC_LOCAL); } } } /* * Get zpool property values. */ int spa_prop_get(spa_t *spa, nvlist_t **nvp) { objset_t *mos = spa->spa_meta_objset; zap_cursor_t zc; zap_attribute_t za; int err; err = nvlist_alloc(nvp, NV_UNIQUE_NAME, KM_PUSHPAGE); if (err) return (err); mutex_enter(&spa->spa_props_lock); /* * Get properties from the spa config. */ spa_prop_get_config(spa, nvp); /* If no pool property object, no more prop to get. */ if (mos == NULL || spa->spa_pool_props_object == 0) { mutex_exit(&spa->spa_props_lock); goto out; } /* * Get properties from the MOS pool property object. */ for (zap_cursor_init(&zc, mos, spa->spa_pool_props_object); (err = zap_cursor_retrieve(&zc, &za)) == 0; zap_cursor_advance(&zc)) { uint64_t intval = 0; char *strval = NULL; zprop_source_t src = ZPROP_SRC_DEFAULT; zpool_prop_t prop; if ((prop = zpool_name_to_prop(za.za_name)) == ZPROP_INVAL) continue; switch (za.za_integer_length) { case 8: /* integer property */ if (za.za_first_integer != zpool_prop_default_numeric(prop)) src = ZPROP_SRC_LOCAL; if (prop == ZPOOL_PROP_BOOTFS) { dsl_pool_t *dp; dsl_dataset_t *ds = NULL; dp = spa_get_dsl(spa); dsl_pool_config_enter(dp, FTAG); if ((err = dsl_dataset_hold_obj(dp, za.za_first_integer, FTAG, &ds))) { dsl_pool_config_exit(dp, FTAG); break; } strval = kmem_alloc( MAXNAMELEN + strlen(MOS_DIR_NAME) + 1, KM_PUSHPAGE); dsl_dataset_name(ds, strval); dsl_dataset_rele(ds, FTAG); dsl_pool_config_exit(dp, FTAG); } else { strval = NULL; intval = za.za_first_integer; } spa_prop_add_list(*nvp, prop, strval, intval, src); if (strval != NULL) kmem_free(strval, MAXNAMELEN + strlen(MOS_DIR_NAME) + 1); break; case 1: /* string property */ strval = kmem_alloc(za.za_num_integers, KM_PUSHPAGE); err = zap_lookup(mos, spa->spa_pool_props_object, za.za_name, 1, za.za_num_integers, strval); if (err) { kmem_free(strval, za.za_num_integers); break; } spa_prop_add_list(*nvp, prop, strval, 0, src); kmem_free(strval, za.za_num_integers); break; default: break; } } zap_cursor_fini(&zc); mutex_exit(&spa->spa_props_lock); out: if (err && err != ENOENT) { nvlist_free(*nvp); *nvp = NULL; return (err); } return (0); } /* * Validate the given pool properties nvlist and modify the list * for the property values to be set. */ static int spa_prop_validate(spa_t *spa, nvlist_t *props) { nvpair_t *elem; int error = 0, reset_bootfs = 0; uint64_t objnum = 0; boolean_t has_feature = B_FALSE; elem = NULL; while ((elem = nvlist_next_nvpair(props, elem)) != NULL) { uint64_t intval; char *strval, *slash, *check, *fname; const char *propname = nvpair_name(elem); zpool_prop_t prop = zpool_name_to_prop(propname); switch ((int)prop) { case ZPROP_INVAL: if (!zpool_prop_feature(propname)) { error = SET_ERROR(EINVAL); break; } /* * Sanitize the input. */ if (nvpair_type(elem) != DATA_TYPE_UINT64) { error = SET_ERROR(EINVAL); break; } if (nvpair_value_uint64(elem, &intval) != 0) { error = SET_ERROR(EINVAL); break; } if (intval != 0) { error = SET_ERROR(EINVAL); break; } fname = strchr(propname, '@') + 1; if (zfeature_lookup_name(fname, NULL) != 0) { error = SET_ERROR(EINVAL); break; } has_feature = B_TRUE; break; case ZPOOL_PROP_VERSION: error = nvpair_value_uint64(elem, &intval); if (!error && (intval < spa_version(spa) || intval > SPA_VERSION_BEFORE_FEATURES || has_feature)) error = SET_ERROR(EINVAL); break; case ZPOOL_PROP_DELEGATION: case ZPOOL_PROP_AUTOREPLACE: case ZPOOL_PROP_LISTSNAPS: case ZPOOL_PROP_AUTOEXPAND: error = nvpair_value_uint64(elem, &intval); if (!error && intval > 1) error = SET_ERROR(EINVAL); break; case ZPOOL_PROP_BOOTFS: /* * If the pool version is less than SPA_VERSION_BOOTFS, * or the pool is still being created (version == 0), * the bootfs property cannot be set. */ if (spa_version(spa) < SPA_VERSION_BOOTFS) { error = SET_ERROR(ENOTSUP); break; } /* * Make sure the vdev config is bootable */ if (!vdev_is_bootable(spa->spa_root_vdev)) { error = SET_ERROR(ENOTSUP); break; } reset_bootfs = 1; error = nvpair_value_string(elem, &strval); if (!error) { objset_t *os; uint64_t compress; if (strval == NULL || strval[0] == '\0') { objnum = zpool_prop_default_numeric( ZPOOL_PROP_BOOTFS); break; } error = dmu_objset_hold(strval, FTAG, &os); if (error) break; /* Must be ZPL and not gzip compressed. */ if (dmu_objset_type(os) != DMU_OST_ZFS) { error = SET_ERROR(ENOTSUP); } else if ((error = dsl_prop_get_int_ds(dmu_objset_ds(os), zfs_prop_to_name(ZFS_PROP_COMPRESSION), &compress)) == 0 && !BOOTFS_COMPRESS_VALID(compress)) { error = SET_ERROR(ENOTSUP); } else { objnum = dmu_objset_id(os); } dmu_objset_rele(os, FTAG); } break; case ZPOOL_PROP_FAILUREMODE: error = nvpair_value_uint64(elem, &intval); if (!error && (intval < ZIO_FAILURE_MODE_WAIT || intval > ZIO_FAILURE_MODE_PANIC)) error = SET_ERROR(EINVAL); /* * This is a special case which only occurs when * the pool has completely failed. This allows * the user to change the in-core failmode property * without syncing it out to disk (I/Os might * currently be blocked). We do this by returning * EIO to the caller (spa_prop_set) to trick it * into thinking we encountered a property validation * error. */ if (!error && spa_suspended(spa)) { spa->spa_failmode = intval; error = SET_ERROR(EIO); } break; case ZPOOL_PROP_CACHEFILE: if ((error = nvpair_value_string(elem, &strval)) != 0) break; if (strval[0] == '\0') break; if (strcmp(strval, "none") == 0) break; if (strval[0] != '/') { error = SET_ERROR(EINVAL); break; } slash = strrchr(strval, '/'); ASSERT(slash != NULL); if (slash[1] == '\0' || strcmp(slash, "/.") == 0 || strcmp(slash, "/..") == 0) error = SET_ERROR(EINVAL); break; case ZPOOL_PROP_COMMENT: if ((error = nvpair_value_string(elem, &strval)) != 0) break; for (check = strval; *check != '\0'; check++) { if (!isprint(*check)) { error = SET_ERROR(EINVAL); break; } check++; } if (strlen(strval) > ZPROP_MAX_COMMENT) error = SET_ERROR(E2BIG); break; case ZPOOL_PROP_DEDUPDITTO: if (spa_version(spa) < SPA_VERSION_DEDUP) error = SET_ERROR(ENOTSUP); else error = nvpair_value_uint64(elem, &intval); if (error == 0 && intval != 0 && intval < ZIO_DEDUPDITTO_MIN) error = SET_ERROR(EINVAL); break; default: break; } if (error) break; } if (!error && reset_bootfs) { error = nvlist_remove(props, zpool_prop_to_name(ZPOOL_PROP_BOOTFS), DATA_TYPE_STRING); if (!error) { error = nvlist_add_uint64(props, zpool_prop_to_name(ZPOOL_PROP_BOOTFS), objnum); } } return (error); } void spa_configfile_set(spa_t *spa, nvlist_t *nvp, boolean_t need_sync) { char *cachefile; spa_config_dirent_t *dp; if (nvlist_lookup_string(nvp, zpool_prop_to_name(ZPOOL_PROP_CACHEFILE), &cachefile) != 0) return; dp = kmem_alloc(sizeof (spa_config_dirent_t), KM_PUSHPAGE); if (cachefile[0] == '\0') dp->scd_path = spa_strdup(spa_config_path); else if (strcmp(cachefile, "none") == 0) dp->scd_path = NULL; else dp->scd_path = spa_strdup(cachefile); list_insert_head(&spa->spa_config_list, dp); if (need_sync) spa_async_request(spa, SPA_ASYNC_CONFIG_UPDATE); } int spa_prop_set(spa_t *spa, nvlist_t *nvp) { int error; nvpair_t *elem = NULL; boolean_t need_sync = B_FALSE; if ((error = spa_prop_validate(spa, nvp)) != 0) return (error); while ((elem = nvlist_next_nvpair(nvp, elem)) != NULL) { zpool_prop_t prop = zpool_name_to_prop(nvpair_name(elem)); if (prop == ZPOOL_PROP_CACHEFILE || prop == ZPOOL_PROP_ALTROOT || prop == ZPOOL_PROP_READONLY) continue; if (prop == ZPOOL_PROP_VERSION || prop == ZPROP_INVAL) { uint64_t ver; if (prop == ZPOOL_PROP_VERSION) { VERIFY(nvpair_value_uint64(elem, &ver) == 0); } else { ASSERT(zpool_prop_feature(nvpair_name(elem))); ver = SPA_VERSION_FEATURES; need_sync = B_TRUE; } /* Save time if the version is already set. */ if (ver == spa_version(spa)) continue; /* * In addition to the pool directory object, we might * create the pool properties object, the features for * read object, the features for write object, or the * feature descriptions object. */ error = dsl_sync_task(spa->spa_name, NULL, spa_sync_version, &ver, 6); if (error) return (error); continue; } need_sync = B_TRUE; break; } if (need_sync) { return (dsl_sync_task(spa->spa_name, NULL, spa_sync_props, nvp, 6)); } return (0); } /* * If the bootfs property value is dsobj, clear it. */ void spa_prop_clear_bootfs(spa_t *spa, uint64_t dsobj, dmu_tx_t *tx) { if (spa->spa_bootfs == dsobj && spa->spa_pool_props_object != 0) { VERIFY(zap_remove(spa->spa_meta_objset, spa->spa_pool_props_object, zpool_prop_to_name(ZPOOL_PROP_BOOTFS), tx) == 0); spa->spa_bootfs = 0; } } /*ARGSUSED*/ static int spa_change_guid_check(void *arg, dmu_tx_t *tx) { spa_t *spa = dmu_tx_pool(tx)->dp_spa; vdev_t *rvd = spa->spa_root_vdev; uint64_t vdev_state; ASSERTV(uint64_t *newguid = arg); spa_config_enter(spa, SCL_STATE, FTAG, RW_READER); vdev_state = rvd->vdev_state; spa_config_exit(spa, SCL_STATE, FTAG); if (vdev_state != VDEV_STATE_HEALTHY) return (SET_ERROR(ENXIO)); ASSERT3U(spa_guid(spa), !=, *newguid); return (0); } static void spa_change_guid_sync(void *arg, dmu_tx_t *tx) { uint64_t *newguid = arg; spa_t *spa = dmu_tx_pool(tx)->dp_spa; uint64_t oldguid; vdev_t *rvd = spa->spa_root_vdev; oldguid = spa_guid(spa); spa_config_enter(spa, SCL_STATE, FTAG, RW_READER); rvd->vdev_guid = *newguid; rvd->vdev_guid_sum += (*newguid - oldguid); vdev_config_dirty(rvd); spa_config_exit(spa, SCL_STATE, FTAG); spa_history_log_internal(spa, "guid change", tx, "old=%llu new=%llu", oldguid, *newguid); } /* * Change the GUID for the pool. This is done so that we can later * re-import a pool built from a clone of our own vdevs. We will modify * the root vdev's guid, our own pool guid, and then mark all of our * vdevs dirty. Note that we must make sure that all our vdevs are * online when we do this, or else any vdevs that weren't present * would be orphaned from our pool. We are also going to issue a * sysevent to update any watchers. */ int spa_change_guid(spa_t *spa) { int error; uint64_t guid; mutex_enter(&spa->spa_vdev_top_lock); mutex_enter(&spa_namespace_lock); guid = spa_generate_guid(NULL); error = dsl_sync_task(spa->spa_name, spa_change_guid_check, spa_change_guid_sync, &guid, 5); if (error == 0) { spa_config_sync(spa, B_FALSE, B_TRUE); spa_event_notify(spa, NULL, FM_EREPORT_ZFS_POOL_REGUID); } mutex_exit(&spa_namespace_lock); mutex_exit(&spa->spa_vdev_top_lock); return (error); } /* * ========================================================================== * SPA state manipulation (open/create/destroy/import/export) * ========================================================================== */ static int spa_error_entry_compare(const void *a, const void *b) { spa_error_entry_t *sa = (spa_error_entry_t *)a; spa_error_entry_t *sb = (spa_error_entry_t *)b; int ret; ret = bcmp(&sa->se_bookmark, &sb->se_bookmark, sizeof (zbookmark_t)); if (ret < 0) return (-1); else if (ret > 0) return (1); else return (0); } /* * Utility function which retrieves copies of the current logs and * re-initializes them in the process. */ void spa_get_errlists(spa_t *spa, avl_tree_t *last, avl_tree_t *scrub) { ASSERT(MUTEX_HELD(&spa->spa_errlist_lock)); bcopy(&spa->spa_errlist_last, last, sizeof (avl_tree_t)); bcopy(&spa->spa_errlist_scrub, scrub, sizeof (avl_tree_t)); avl_create(&spa->spa_errlist_scrub, spa_error_entry_compare, sizeof (spa_error_entry_t), offsetof(spa_error_entry_t, se_avl)); avl_create(&spa->spa_errlist_last, spa_error_entry_compare, sizeof (spa_error_entry_t), offsetof(spa_error_entry_t, se_avl)); } static void spa_taskqs_init(spa_t *spa, zio_type_t t, zio_taskq_type_t q) { const zio_taskq_info_t *ztip = &zio_taskqs[t][q]; enum zti_modes mode = ztip->zti_mode; uint_t value = ztip->zti_value; uint_t count = ztip->zti_count; spa_taskqs_t *tqs = &spa->spa_zio_taskq[t][q]; char name[32]; uint_t i, flags = 0; boolean_t batch = B_FALSE; if (mode == ZTI_MODE_NULL) { tqs->stqs_count = 0; tqs->stqs_taskq = NULL; return; } ASSERT3U(count, >, 0); tqs->stqs_count = count; tqs->stqs_taskq = kmem_alloc(count * sizeof (taskq_t *), KM_SLEEP); switch (mode) { case ZTI_MODE_FIXED: ASSERT3U(value, >=, 1); value = MAX(value, 1); break; case ZTI_MODE_BATCH: batch = B_TRUE; flags |= TASKQ_THREADS_CPU_PCT; value = zio_taskq_batch_pct; break; default: panic("unrecognized mode for %s_%s taskq (%u:%u) in " "spa_activate()", zio_type_name[t], zio_taskq_types[q], mode, value); break; } for (i = 0; i < count; i++) { taskq_t *tq; if (count > 1) { (void) snprintf(name, sizeof (name), "%s_%s_%u", zio_type_name[t], zio_taskq_types[q], i); } else { (void) snprintf(name, sizeof (name), "%s_%s", zio_type_name[t], zio_taskq_types[q]); } if (zio_taskq_sysdc && spa->spa_proc != &p0) { if (batch) flags |= TASKQ_DC_BATCH; tq = taskq_create_sysdc(name, value, 50, INT_MAX, spa->spa_proc, zio_taskq_basedc, flags); } else { pri_t pri = maxclsyspri; /* * The write issue taskq can be extremely CPU * intensive. Run it at slightly lower priority * than the other taskqs. */ if (t == ZIO_TYPE_WRITE && q == ZIO_TASKQ_ISSUE) pri--; tq = taskq_create_proc(name, value, pri, 50, INT_MAX, spa->spa_proc, flags); } tqs->stqs_taskq[i] = tq; } } static void spa_taskqs_fini(spa_t *spa, zio_type_t t, zio_taskq_type_t q) { spa_taskqs_t *tqs = &spa->spa_zio_taskq[t][q]; uint_t i; if (tqs->stqs_taskq == NULL) { ASSERT3U(tqs->stqs_count, ==, 0); return; } for (i = 0; i < tqs->stqs_count; i++) { ASSERT3P(tqs->stqs_taskq[i], !=, NULL); taskq_destroy(tqs->stqs_taskq[i]); } kmem_free(tqs->stqs_taskq, tqs->stqs_count * sizeof (taskq_t *)); tqs->stqs_taskq = NULL; } /* * Dispatch a task to the appropriate taskq for the ZFS I/O type and priority. * Note that a type may have multiple discrete taskqs to avoid lock contention * on the taskq itself. In that case we choose which taskq at random by using * the low bits of gethrtime(). */ void spa_taskq_dispatch_ent(spa_t *spa, zio_type_t t, zio_taskq_type_t q, task_func_t *func, void *arg, uint_t flags, taskq_ent_t *ent) { spa_taskqs_t *tqs = &spa->spa_zio_taskq[t][q]; taskq_t *tq; ASSERT3P(tqs->stqs_taskq, !=, NULL); ASSERT3U(tqs->stqs_count, !=, 0); if (tqs->stqs_count == 1) { tq = tqs->stqs_taskq[0]; } else { tq = tqs->stqs_taskq[((uint64_t)gethrtime()) % tqs->stqs_count]; } taskq_dispatch_ent(tq, func, arg, flags, ent); } /* * Same as spa_taskq_dispatch_ent() but block on the task until completion. */ void spa_taskq_dispatch_sync(spa_t *spa, zio_type_t t, zio_taskq_type_t q, task_func_t *func, void *arg, uint_t flags) { spa_taskqs_t *tqs = &spa->spa_zio_taskq[t][q]; taskq_t *tq; taskqid_t id; ASSERT3P(tqs->stqs_taskq, !=, NULL); ASSERT3U(tqs->stqs_count, !=, 0); if (tqs->stqs_count == 1) { tq = tqs->stqs_taskq[0]; } else { tq = tqs->stqs_taskq[((uint64_t)gethrtime()) % tqs->stqs_count]; } id = taskq_dispatch(tq, func, arg, flags); if (id) taskq_wait_id(tq, id); } static void spa_create_zio_taskqs(spa_t *spa) { int t, q; for (t = 0; t < ZIO_TYPES; t++) { for (q = 0; q < ZIO_TASKQ_TYPES; q++) { spa_taskqs_init(spa, t, q); } } } #if defined(_KERNEL) && defined(HAVE_SPA_THREAD) static void spa_thread(void *arg) { callb_cpr_t cprinfo; spa_t *spa = arg; user_t *pu = PTOU(curproc); CALLB_CPR_INIT(&cprinfo, &spa->spa_proc_lock, callb_generic_cpr, spa->spa_name); ASSERT(curproc != &p0); (void) snprintf(pu->u_psargs, sizeof (pu->u_psargs), "zpool-%s", spa->spa_name); (void) strlcpy(pu->u_comm, pu->u_psargs, sizeof (pu->u_comm)); /* bind this thread to the requested psrset */ if (zio_taskq_psrset_bind != PS_NONE) { pool_lock(); mutex_enter(&cpu_lock); mutex_enter(&pidlock); mutex_enter(&curproc->p_lock); if (cpupart_bind_thread(curthread, zio_taskq_psrset_bind, 0, NULL, NULL) == 0) { curthread->t_bind_pset = zio_taskq_psrset_bind; } else { cmn_err(CE_WARN, "Couldn't bind process for zfs pool \"%s\" to " "pset %d\n", spa->spa_name, zio_taskq_psrset_bind); } mutex_exit(&curproc->p_lock); mutex_exit(&pidlock); mutex_exit(&cpu_lock); pool_unlock(); } if (zio_taskq_sysdc) { sysdc_thread_enter(curthread, 100, 0); } spa->spa_proc = curproc; spa->spa_did = curthread->t_did; spa_create_zio_taskqs(spa); mutex_enter(&spa->spa_proc_lock); ASSERT(spa->spa_proc_state == SPA_PROC_CREATED); spa->spa_proc_state = SPA_PROC_ACTIVE; cv_broadcast(&spa->spa_proc_cv); CALLB_CPR_SAFE_BEGIN(&cprinfo); while (spa->spa_proc_state == SPA_PROC_ACTIVE) cv_wait(&spa->spa_proc_cv, &spa->spa_proc_lock); CALLB_CPR_SAFE_END(&cprinfo, &spa->spa_proc_lock); ASSERT(spa->spa_proc_state == SPA_PROC_DEACTIVATE); spa->spa_proc_state = SPA_PROC_GONE; spa->spa_proc = &p0; cv_broadcast(&spa->spa_proc_cv); CALLB_CPR_EXIT(&cprinfo); /* drops spa_proc_lock */ mutex_enter(&curproc->p_lock); lwp_exit(); } #endif /* * Activate an uninitialized pool. */ static void spa_activate(spa_t *spa, int mode) { ASSERT(spa->spa_state == POOL_STATE_UNINITIALIZED); spa->spa_state = POOL_STATE_ACTIVE; spa->spa_mode = mode; spa->spa_normal_class = metaslab_class_create(spa, zfs_metaslab_ops); spa->spa_log_class = metaslab_class_create(spa, zfs_metaslab_ops); /* Try to create a covering process */ mutex_enter(&spa->spa_proc_lock); ASSERT(spa->spa_proc_state == SPA_PROC_NONE); ASSERT(spa->spa_proc == &p0); spa->spa_did = 0; #ifdef HAVE_SPA_THREAD /* Only create a process if we're going to be around a while. */ if (spa_create_process && strcmp(spa->spa_name, TRYIMPORT_NAME) != 0) { if (newproc(spa_thread, (caddr_t)spa, syscid, maxclsyspri, NULL, 0) == 0) { spa->spa_proc_state = SPA_PROC_CREATED; while (spa->spa_proc_state == SPA_PROC_CREATED) { cv_wait(&spa->spa_proc_cv, &spa->spa_proc_lock); } ASSERT(spa->spa_proc_state == SPA_PROC_ACTIVE); ASSERT(spa->spa_proc != &p0); ASSERT(spa->spa_did != 0); } else { #ifdef _KERNEL cmn_err(CE_WARN, "Couldn't create process for zfs pool \"%s\"\n", spa->spa_name); #endif } } #endif /* HAVE_SPA_THREAD */ mutex_exit(&spa->spa_proc_lock); /* If we didn't create a process, we need to create our taskqs. */ if (spa->spa_proc == &p0) { spa_create_zio_taskqs(spa); } list_create(&spa->spa_config_dirty_list, sizeof (vdev_t), offsetof(vdev_t, vdev_config_dirty_node)); list_create(&spa->spa_state_dirty_list, sizeof (vdev_t), offsetof(vdev_t, vdev_state_dirty_node)); txg_list_create(&spa->spa_vdev_txg_list, offsetof(struct vdev, vdev_txg_node)); avl_create(&spa->spa_errlist_scrub, spa_error_entry_compare, sizeof (spa_error_entry_t), offsetof(spa_error_entry_t, se_avl)); avl_create(&spa->spa_errlist_last, spa_error_entry_compare, sizeof (spa_error_entry_t), offsetof(spa_error_entry_t, se_avl)); } /* * Opposite of spa_activate(). */ static void spa_deactivate(spa_t *spa) { int t, q; ASSERT(spa->spa_sync_on == B_FALSE); ASSERT(spa->spa_dsl_pool == NULL); ASSERT(spa->spa_root_vdev == NULL); ASSERT(spa->spa_async_zio_root == NULL); ASSERT(spa->spa_state != POOL_STATE_UNINITIALIZED); txg_list_destroy(&spa->spa_vdev_txg_list); list_destroy(&spa->spa_config_dirty_list); list_destroy(&spa->spa_state_dirty_list); taskq_cancel_id(system_taskq, spa->spa_deadman_tqid); for (t = 0; t < ZIO_TYPES; t++) { for (q = 0; q < ZIO_TASKQ_TYPES; q++) { spa_taskqs_fini(spa, t, q); } } metaslab_class_destroy(spa->spa_normal_class); spa->spa_normal_class = NULL; metaslab_class_destroy(spa->spa_log_class); spa->spa_log_class = NULL; /* * If this was part of an import or the open otherwise failed, we may * still have errors left in the queues. Empty them just in case. */ spa_errlog_drain(spa); avl_destroy(&spa->spa_errlist_scrub); avl_destroy(&spa->spa_errlist_last); spa->spa_state = POOL_STATE_UNINITIALIZED; mutex_enter(&spa->spa_proc_lock); if (spa->spa_proc_state != SPA_PROC_NONE) { ASSERT(spa->spa_proc_state == SPA_PROC_ACTIVE); spa->spa_proc_state = SPA_PROC_DEACTIVATE; cv_broadcast(&spa->spa_proc_cv); while (spa->spa_proc_state == SPA_PROC_DEACTIVATE) { ASSERT(spa->spa_proc != &p0); cv_wait(&spa->spa_proc_cv, &spa->spa_proc_lock); } ASSERT(spa->spa_proc_state == SPA_PROC_GONE); spa->spa_proc_state = SPA_PROC_NONE; } ASSERT(spa->spa_proc == &p0); mutex_exit(&spa->spa_proc_lock); /* * We want to make sure spa_thread() has actually exited the ZFS * module, so that the module can't be unloaded out from underneath * it. */ if (spa->spa_did != 0) { thread_join(spa->spa_did); spa->spa_did = 0; } } /* * Verify a pool configuration, and construct the vdev tree appropriately. This * will create all the necessary vdevs in the appropriate layout, with each vdev * in the CLOSED state. This will prep the pool before open/creation/import. * All vdev validation is done by the vdev_alloc() routine. */ static int spa_config_parse(spa_t *spa, vdev_t **vdp, nvlist_t *nv, vdev_t *parent, uint_t id, int atype) { nvlist_t **child; uint_t children; int error; int c; if ((error = vdev_alloc(spa, vdp, nv, parent, id, atype)) != 0) return (error); if ((*vdp)->vdev_ops->vdev_op_leaf) return (0); error = nvlist_lookup_nvlist_array(nv, ZPOOL_CONFIG_CHILDREN, &child, &children); if (error == ENOENT) return (0); if (error) { vdev_free(*vdp); *vdp = NULL; return (SET_ERROR(EINVAL)); } for (c = 0; c < children; c++) { vdev_t *vd; if ((error = spa_config_parse(spa, &vd, child[c], *vdp, c, atype)) != 0) { vdev_free(*vdp); *vdp = NULL; return (error); } } ASSERT(*vdp != NULL); return (0); } /* * Opposite of spa_load(). */ static void spa_unload(spa_t *spa) { int i; ASSERT(MUTEX_HELD(&spa_namespace_lock)); /* * Stop async tasks. */ spa_async_suspend(spa); /* * Stop syncing. */ if (spa->spa_sync_on) { txg_sync_stop(spa->spa_dsl_pool); spa->spa_sync_on = B_FALSE; } /* * Wait for any outstanding async I/O to complete. */ if (spa->spa_async_zio_root != NULL) { (void) zio_wait(spa->spa_async_zio_root); spa->spa_async_zio_root = NULL; } bpobj_close(&spa->spa_deferred_bpobj); spa_config_enter(spa, SCL_ALL, FTAG, RW_WRITER); /* * Close all vdevs. */ if (spa->spa_root_vdev) vdev_free(spa->spa_root_vdev); ASSERT(spa->spa_root_vdev == NULL); /* * Close the dsl pool. */ if (spa->spa_dsl_pool) { dsl_pool_close(spa->spa_dsl_pool); spa->spa_dsl_pool = NULL; spa->spa_meta_objset = NULL; } ddt_unload(spa); /* * Drop and purge level 2 cache */ spa_l2cache_drop(spa); for (i = 0; i < spa->spa_spares.sav_count; i++) vdev_free(spa->spa_spares.sav_vdevs[i]); if (spa->spa_spares.sav_vdevs) { kmem_free(spa->spa_spares.sav_vdevs, spa->spa_spares.sav_count * sizeof (void *)); spa->spa_spares.sav_vdevs = NULL; } if (spa->spa_spares.sav_config) { nvlist_free(spa->spa_spares.sav_config); spa->spa_spares.sav_config = NULL; } spa->spa_spares.sav_count = 0; for (i = 0; i < spa->spa_l2cache.sav_count; i++) { vdev_clear_stats(spa->spa_l2cache.sav_vdevs[i]); vdev_free(spa->spa_l2cache.sav_vdevs[i]); } if (spa->spa_l2cache.sav_vdevs) { kmem_free(spa->spa_l2cache.sav_vdevs, spa->spa_l2cache.sav_count * sizeof (void *)); spa->spa_l2cache.sav_vdevs = NULL; } if (spa->spa_l2cache.sav_config) { nvlist_free(spa->spa_l2cache.sav_config); spa->spa_l2cache.sav_config = NULL; } spa->spa_l2cache.sav_count = 0; spa->spa_async_suspended = 0; if (spa->spa_comment != NULL) { spa_strfree(spa->spa_comment); spa->spa_comment = NULL; } spa_config_exit(spa, SCL_ALL, FTAG); } /* * Load (or re-load) the current list of vdevs describing the active spares for * this pool. When this is called, we have some form of basic information in * 'spa_spares.sav_config'. We parse this into vdevs, try to open them, and * then re-generate a more complete list including status information. */ static void spa_load_spares(spa_t *spa) { nvlist_t **spares; uint_t nspares; int i; vdev_t *vd, *tvd; ASSERT(spa_config_held(spa, SCL_ALL, RW_WRITER) == SCL_ALL); /* * First, close and free any existing spare vdevs. */ for (i = 0; i < spa->spa_spares.sav_count; i++) { vd = spa->spa_spares.sav_vdevs[i]; /* Undo the call to spa_activate() below */ if ((tvd = spa_lookup_by_guid(spa, vd->vdev_guid, B_FALSE)) != NULL && tvd->vdev_isspare) spa_spare_remove(tvd); vdev_close(vd); vdev_free(vd); } if (spa->spa_spares.sav_vdevs) kmem_free(spa->spa_spares.sav_vdevs, spa->spa_spares.sav_count * sizeof (void *)); if (spa->spa_spares.sav_config == NULL) nspares = 0; else VERIFY(nvlist_lookup_nvlist_array(spa->spa_spares.sav_config, ZPOOL_CONFIG_SPARES, &spares, &nspares) == 0); spa->spa_spares.sav_count = (int)nspares; spa->spa_spares.sav_vdevs = NULL; if (nspares == 0) return; /* * Construct the array of vdevs, opening them to get status in the * process. For each spare, there is potentially two different vdev_t * structures associated with it: one in the list of spares (used only * for basic validation purposes) and one in the active vdev * configuration (if it's spared in). During this phase we open and * validate each vdev on the spare list. If the vdev also exists in the * active configuration, then we also mark this vdev as an active spare. */ spa->spa_spares.sav_vdevs = kmem_zalloc(nspares * sizeof (void *), KM_PUSHPAGE); for (i = 0; i < spa->spa_spares.sav_count; i++) { VERIFY(spa_config_parse(spa, &vd, spares[i], NULL, 0, VDEV_ALLOC_SPARE) == 0); ASSERT(vd != NULL); spa->spa_spares.sav_vdevs[i] = vd; if ((tvd = spa_lookup_by_guid(spa, vd->vdev_guid, B_FALSE)) != NULL) { if (!tvd->vdev_isspare) spa_spare_add(tvd); /* * We only mark the spare active if we were successfully * able to load the vdev. Otherwise, importing a pool * with a bad active spare would result in strange * behavior, because multiple pool would think the spare * is actively in use. * * There is a vulnerability here to an equally bizarre * circumstance, where a dead active spare is later * brought back to life (onlined or otherwise). Given * the rarity of this scenario, and the extra complexity * it adds, we ignore the possibility. */ if (!vdev_is_dead(tvd)) spa_spare_activate(tvd); } vd->vdev_top = vd; vd->vdev_aux = &spa->spa_spares; if (vdev_open(vd) != 0) continue; if (vdev_validate_aux(vd) == 0) spa_spare_add(vd); } /* * Recompute the stashed list of spares, with status information * this time. */ VERIFY(nvlist_remove(spa->spa_spares.sav_config, ZPOOL_CONFIG_SPARES, DATA_TYPE_NVLIST_ARRAY) == 0); spares = kmem_alloc(spa->spa_spares.sav_count * sizeof (void *), KM_PUSHPAGE); for (i = 0; i < spa->spa_spares.sav_count; i++) spares[i] = vdev_config_generate(spa, spa->spa_spares.sav_vdevs[i], B_TRUE, VDEV_CONFIG_SPARE); VERIFY(nvlist_add_nvlist_array(spa->spa_spares.sav_config, ZPOOL_CONFIG_SPARES, spares, spa->spa_spares.sav_count) == 0); for (i = 0; i < spa->spa_spares.sav_count; i++) nvlist_free(spares[i]); kmem_free(spares, spa->spa_spares.sav_count * sizeof (void *)); } /* * Load (or re-load) the current list of vdevs describing the active l2cache for * this pool. When this is called, we have some form of basic information in * 'spa_l2cache.sav_config'. We parse this into vdevs, try to open them, and * then re-generate a more complete list including status information. * Devices which are already active have their details maintained, and are * not re-opened. */ static void spa_load_l2cache(spa_t *spa) { nvlist_t **l2cache; uint_t nl2cache; int i, j, oldnvdevs; uint64_t guid; vdev_t *vd, **oldvdevs, **newvdevs; spa_aux_vdev_t *sav = &spa->spa_l2cache; ASSERT(spa_config_held(spa, SCL_ALL, RW_WRITER) == SCL_ALL); if (sav->sav_config != NULL) { VERIFY(nvlist_lookup_nvlist_array(sav->sav_config, ZPOOL_CONFIG_L2CACHE, &l2cache, &nl2cache) == 0); newvdevs = kmem_alloc(nl2cache * sizeof (void *), KM_PUSHPAGE); } else { nl2cache = 0; newvdevs = NULL; } oldvdevs = sav->sav_vdevs; oldnvdevs = sav->sav_count; sav->sav_vdevs = NULL; sav->sav_count = 0; /* * Process new nvlist of vdevs. */ for (i = 0; i < nl2cache; i++) { VERIFY(nvlist_lookup_uint64(l2cache[i], ZPOOL_CONFIG_GUID, &guid) == 0); newvdevs[i] = NULL; for (j = 0; j < oldnvdevs; j++) { vd = oldvdevs[j]; if (vd != NULL && guid == vd->vdev_guid) { /* * Retain previous vdev for add/remove ops. */ newvdevs[i] = vd; oldvdevs[j] = NULL; break; } } if (newvdevs[i] == NULL) { /* * Create new vdev */ VERIFY(spa_config_parse(spa, &vd, l2cache[i], NULL, 0, VDEV_ALLOC_L2CACHE) == 0); ASSERT(vd != NULL); newvdevs[i] = vd; /* * Commit this vdev as an l2cache device, * even if it fails to open. */ spa_l2cache_add(vd); vd->vdev_top = vd; vd->vdev_aux = sav; spa_l2cache_activate(vd); if (vdev_open(vd) != 0) continue; (void) vdev_validate_aux(vd); if (!vdev_is_dead(vd)) l2arc_add_vdev(spa, vd); } } /* * Purge vdevs that were dropped */ for (i = 0; i < oldnvdevs; i++) { uint64_t pool; vd = oldvdevs[i]; if (vd != NULL) { ASSERT(vd->vdev_isl2cache); if (spa_l2cache_exists(vd->vdev_guid, &pool) && pool != 0ULL && l2arc_vdev_present(vd)) l2arc_remove_vdev(vd); vdev_clear_stats(vd); vdev_free(vd); } } if (oldvdevs) kmem_free(oldvdevs, oldnvdevs * sizeof (void *)); if (sav->sav_config == NULL) goto out; sav->sav_vdevs = newvdevs; sav->sav_count = (int)nl2cache; /* * Recompute the stashed list of l2cache devices, with status * information this time. */ VERIFY(nvlist_remove(sav->sav_config, ZPOOL_CONFIG_L2CACHE, DATA_TYPE_NVLIST_ARRAY) == 0); l2cache = kmem_alloc(sav->sav_count * sizeof (void *), KM_PUSHPAGE); for (i = 0; i < sav->sav_count; i++) l2cache[i] = vdev_config_generate(spa, sav->sav_vdevs[i], B_TRUE, VDEV_CONFIG_L2CACHE); VERIFY(nvlist_add_nvlist_array(sav->sav_config, ZPOOL_CONFIG_L2CACHE, l2cache, sav->sav_count) == 0); out: for (i = 0; i < sav->sav_count; i++) nvlist_free(l2cache[i]); if (sav->sav_count) kmem_free(l2cache, sav->sav_count * sizeof (void *)); } static int load_nvlist(spa_t *spa, uint64_t obj, nvlist_t **value) { dmu_buf_t *db; char *packed = NULL; size_t nvsize = 0; int error; *value = NULL; error = dmu_bonus_hold(spa->spa_meta_objset, obj, FTAG, &db); if (error) return (error); nvsize = *(uint64_t *)db->db_data; dmu_buf_rele(db, FTAG); packed = kmem_alloc(nvsize, KM_PUSHPAGE | KM_NODEBUG); error = dmu_read(spa->spa_meta_objset, obj, 0, nvsize, packed, DMU_READ_PREFETCH); if (error == 0) error = nvlist_unpack(packed, nvsize, value, 0); kmem_free(packed, nvsize); return (error); } /* * Checks to see if the given vdev could not be opened, in which case we post a * sysevent to notify the autoreplace code that the device has been removed. */ static void spa_check_removed(vdev_t *vd) { int c; for (c = 0; c < vd->vdev_children; c++) spa_check_removed(vd->vdev_child[c]); if (vd->vdev_ops->vdev_op_leaf && vdev_is_dead(vd) && !vd->vdev_ishole) { zfs_ereport_post(FM_EREPORT_RESOURCE_AUTOREPLACE, vd->vdev_spa, vd, NULL, 0, 0); spa_event_notify(vd->vdev_spa, vd, FM_EREPORT_ZFS_DEVICE_CHECK); } } /* * Validate the current config against the MOS config */ static boolean_t spa_config_valid(spa_t *spa, nvlist_t *config) { vdev_t *mrvd, *rvd = spa->spa_root_vdev; nvlist_t *nv; int c, i; VERIFY(nvlist_lookup_nvlist(config, ZPOOL_CONFIG_VDEV_TREE, &nv) == 0); spa_config_enter(spa, SCL_ALL, FTAG, RW_WRITER); VERIFY(spa_config_parse(spa, &mrvd, nv, NULL, 0, VDEV_ALLOC_LOAD) == 0); ASSERT3U(rvd->vdev_children, ==, mrvd->vdev_children); /* * If we're doing a normal import, then build up any additional * diagnostic information about missing devices in this config. * We'll pass this up to the user for further processing. */ if (!(spa->spa_import_flags & ZFS_IMPORT_MISSING_LOG)) { nvlist_t **child, *nv; uint64_t idx = 0; child = kmem_alloc(rvd->vdev_children * sizeof (nvlist_t **), KM_PUSHPAGE); VERIFY(nvlist_alloc(&nv, NV_UNIQUE_NAME, KM_PUSHPAGE) == 0); for (c = 0; c < rvd->vdev_children; c++) { vdev_t *tvd = rvd->vdev_child[c]; vdev_t *mtvd = mrvd->vdev_child[c]; if (tvd->vdev_ops == &vdev_missing_ops && mtvd->vdev_ops != &vdev_missing_ops && mtvd->vdev_islog) child[idx++] = vdev_config_generate(spa, mtvd, B_FALSE, 0); } if (idx) { VERIFY(nvlist_add_nvlist_array(nv, ZPOOL_CONFIG_CHILDREN, child, idx) == 0); VERIFY(nvlist_add_nvlist(spa->spa_load_info, ZPOOL_CONFIG_MISSING_DEVICES, nv) == 0); for (i = 0; i < idx; i++) nvlist_free(child[i]); } nvlist_free(nv); kmem_free(child, rvd->vdev_children * sizeof (char **)); } /* * Compare the root vdev tree with the information we have * from the MOS config (mrvd). Check each top-level vdev * with the corresponding MOS config top-level (mtvd). */ for (c = 0; c < rvd->vdev_children; c++) { vdev_t *tvd = rvd->vdev_child[c]; vdev_t *mtvd = mrvd->vdev_child[c]; /* * Resolve any "missing" vdevs in the current configuration. * If we find that the MOS config has more accurate information * about the top-level vdev then use that vdev instead. */ if (tvd->vdev_ops == &vdev_missing_ops && mtvd->vdev_ops != &vdev_missing_ops) { if (!(spa->spa_import_flags & ZFS_IMPORT_MISSING_LOG)) continue; /* * Device specific actions. */ if (mtvd->vdev_islog) { spa_set_log_state(spa, SPA_LOG_CLEAR); } else { /* * XXX - once we have 'readonly' pool * support we should be able to handle * missing data devices by transitioning * the pool to readonly. */ continue; } /* * Swap the missing vdev with the data we were * able to obtain from the MOS config. */ vdev_remove_child(rvd, tvd); vdev_remove_child(mrvd, mtvd); vdev_add_child(rvd, mtvd); vdev_add_child(mrvd, tvd); spa_config_exit(spa, SCL_ALL, FTAG); vdev_load(mtvd); spa_config_enter(spa, SCL_ALL, FTAG, RW_WRITER); vdev_reopen(rvd); } else if (mtvd->vdev_islog) { /* * Load the slog device's state from the MOS config * since it's possible that the label does not * contain the most up-to-date information. */ vdev_load_log_state(tvd, mtvd); vdev_reopen(tvd); } } vdev_free(mrvd); spa_config_exit(spa, SCL_ALL, FTAG); /* * Ensure we were able to validate the config. */ return (rvd->vdev_guid_sum == spa->spa_uberblock.ub_guid_sum); } /* * Check for missing log devices */ static boolean_t spa_check_logs(spa_t *spa) { boolean_t rv = B_FALSE; switch (spa->spa_log_state) { default: break; case SPA_LOG_MISSING: /* need to recheck in case slog has been restored */ case SPA_LOG_UNKNOWN: rv = (dmu_objset_find(spa->spa_name, zil_check_log_chain, NULL, DS_FIND_CHILDREN) != 0); if (rv) spa_set_log_state(spa, SPA_LOG_MISSING); break; } return (rv); } static boolean_t spa_passivate_log(spa_t *spa) { vdev_t *rvd = spa->spa_root_vdev; boolean_t slog_found = B_FALSE; int c; ASSERT(spa_config_held(spa, SCL_ALLOC, RW_WRITER)); if (!spa_has_slogs(spa)) return (B_FALSE); for (c = 0; c < rvd->vdev_children; c++) { vdev_t *tvd = rvd->vdev_child[c]; metaslab_group_t *mg = tvd->vdev_mg; if (tvd->vdev_islog) { metaslab_group_passivate(mg); slog_found = B_TRUE; } } return (slog_found); } static void spa_activate_log(spa_t *spa) { vdev_t *rvd = spa->spa_root_vdev; int c; ASSERT(spa_config_held(spa, SCL_ALLOC, RW_WRITER)); for (c = 0; c < rvd->vdev_children; c++) { vdev_t *tvd = rvd->vdev_child[c]; metaslab_group_t *mg = tvd->vdev_mg; if (tvd->vdev_islog) metaslab_group_activate(mg); } } int spa_offline_log(spa_t *spa) { int error; error = dmu_objset_find(spa_name(spa), zil_vdev_offline, NULL, DS_FIND_CHILDREN); if (error == 0) { /* * We successfully offlined the log device, sync out the * current txg so that the "stubby" block can be removed * by zil_sync(). */ txg_wait_synced(spa->spa_dsl_pool, 0); } return (error); } static void spa_aux_check_removed(spa_aux_vdev_t *sav) { int i; for (i = 0; i < sav->sav_count; i++) spa_check_removed(sav->sav_vdevs[i]); } void spa_claim_notify(zio_t *zio) { spa_t *spa = zio->io_spa; if (zio->io_error) return; mutex_enter(&spa->spa_props_lock); /* any mutex will do */ if (spa->spa_claim_max_txg < zio->io_bp->blk_birth) spa->spa_claim_max_txg = zio->io_bp->blk_birth; mutex_exit(&spa->spa_props_lock); } typedef struct spa_load_error { uint64_t sle_meta_count; uint64_t sle_data_count; } spa_load_error_t; static void spa_load_verify_done(zio_t *zio) { blkptr_t *bp = zio->io_bp; spa_load_error_t *sle = zio->io_private; dmu_object_type_t type = BP_GET_TYPE(bp); int error = zio->io_error; if (error) { if ((BP_GET_LEVEL(bp) != 0 || DMU_OT_IS_METADATA(type)) && type != DMU_OT_INTENT_LOG) atomic_add_64(&sle->sle_meta_count, 1); else atomic_add_64(&sle->sle_data_count, 1); } zio_data_buf_free(zio->io_data, zio->io_size); } /*ARGSUSED*/ static int spa_load_verify_cb(spa_t *spa, zilog_t *zilog, const blkptr_t *bp, const zbookmark_t *zb, const dnode_phys_t *dnp, void *arg) { if (!BP_IS_HOLE(bp) && !BP_IS_EMBEDDED(bp)) { zio_t *rio = arg; size_t size = BP_GET_PSIZE(bp); void *data = zio_data_buf_alloc(size); zio_nowait(zio_read(rio, spa, bp, data, size, spa_load_verify_done, rio->io_private, ZIO_PRIORITY_SCRUB, ZIO_FLAG_SPECULATIVE | ZIO_FLAG_CANFAIL | ZIO_FLAG_SCRUB | ZIO_FLAG_RAW, zb)); } return (0); } static int spa_load_verify(spa_t *spa) { zio_t *rio; spa_load_error_t sle = { 0 }; zpool_rewind_policy_t policy; boolean_t verify_ok = B_FALSE; int error; zpool_get_rewind_policy(spa->spa_config, &policy); if (policy.zrp_request & ZPOOL_NEVER_REWIND) return (0); rio = zio_root(spa, NULL, &sle, ZIO_FLAG_CANFAIL | ZIO_FLAG_SPECULATIVE); error = traverse_pool(spa, spa->spa_verify_min_txg, TRAVERSE_PRE | TRAVERSE_PREFETCH, spa_load_verify_cb, rio); (void) zio_wait(rio); spa->spa_load_meta_errors = sle.sle_meta_count; spa->spa_load_data_errors = sle.sle_data_count; if (!error && sle.sle_meta_count <= policy.zrp_maxmeta && sle.sle_data_count <= policy.zrp_maxdata) { int64_t loss = 0; verify_ok = B_TRUE; spa->spa_load_txg = spa->spa_uberblock.ub_txg; spa->spa_load_txg_ts = spa->spa_uberblock.ub_timestamp; loss = spa->spa_last_ubsync_txg_ts - spa->spa_load_txg_ts; VERIFY(nvlist_add_uint64(spa->spa_load_info, ZPOOL_CONFIG_LOAD_TIME, spa->spa_load_txg_ts) == 0); VERIFY(nvlist_add_int64(spa->spa_load_info, ZPOOL_CONFIG_REWIND_TIME, loss) == 0); VERIFY(nvlist_add_uint64(spa->spa_load_info, ZPOOL_CONFIG_LOAD_DATA_ERRORS, sle.sle_data_count) == 0); } else { spa->spa_load_max_txg = spa->spa_uberblock.ub_txg; } if (error) { if (error != ENXIO && error != EIO) error = SET_ERROR(EIO); return (error); } return (verify_ok ? 0 : EIO); } /* * Find a value in the pool props object. */ static void spa_prop_find(spa_t *spa, zpool_prop_t prop, uint64_t *val) { (void) zap_lookup(spa->spa_meta_objset, spa->spa_pool_props_object, zpool_prop_to_name(prop), sizeof (uint64_t), 1, val); } /* * Find a value in the pool directory object. */ static int spa_dir_prop(spa_t *spa, const char *name, uint64_t *val) { return (zap_lookup(spa->spa_meta_objset, DMU_POOL_DIRECTORY_OBJECT, name, sizeof (uint64_t), 1, val)); } static int spa_vdev_err(vdev_t *vdev, vdev_aux_t aux, int err) { vdev_set_state(vdev, B_TRUE, VDEV_STATE_CANT_OPEN, aux); return (err); } /* * Fix up config after a partly-completed split. This is done with the * ZPOOL_CONFIG_SPLIT nvlist. Both the splitting pool and the split-off * pool have that entry in their config, but only the splitting one contains * a list of all the guids of the vdevs that are being split off. * * This function determines what to do with that list: either rejoin * all the disks to the pool, or complete the splitting process. To attempt * the rejoin, each disk that is offlined is marked online again, and * we do a reopen() call. If the vdev label for every disk that was * marked online indicates it was successfully split off (VDEV_AUX_SPLIT_POOL) * then we call vdev_split() on each disk, and complete the split. * * Otherwise we leave the config alone, with all the vdevs in place in * the original pool. */ static void spa_try_repair(spa_t *spa, nvlist_t *config) { uint_t extracted; uint64_t *glist; uint_t i, gcount; nvlist_t *nvl; vdev_t **vd; boolean_t attempt_reopen; if (nvlist_lookup_nvlist(config, ZPOOL_CONFIG_SPLIT, &nvl) != 0) return; /* check that the config is complete */ if (nvlist_lookup_uint64_array(nvl, ZPOOL_CONFIG_SPLIT_LIST, &glist, &gcount) != 0) return; vd = kmem_zalloc(gcount * sizeof (vdev_t *), KM_PUSHPAGE); /* attempt to online all the vdevs & validate */ attempt_reopen = B_TRUE; for (i = 0; i < gcount; i++) { if (glist[i] == 0) /* vdev is hole */ continue; vd[i] = spa_lookup_by_guid(spa, glist[i], B_FALSE); if (vd[i] == NULL) { /* * Don't bother attempting to reopen the disks; * just do the split. */ attempt_reopen = B_FALSE; } else { /* attempt to re-online it */ vd[i]->vdev_offline = B_FALSE; } } if (attempt_reopen) { vdev_reopen(spa->spa_root_vdev); /* check each device to see what state it's in */ for (extracted = 0, i = 0; i < gcount; i++) { if (vd[i] != NULL && vd[i]->vdev_stat.vs_aux != VDEV_AUX_SPLIT_POOL) break; ++extracted; } } /* * If every disk has been moved to the new pool, or if we never * even attempted to look at them, then we split them off for * good. */ if (!attempt_reopen || gcount == extracted) { for (i = 0; i < gcount; i++) if (vd[i] != NULL) vdev_split(vd[i]); vdev_reopen(spa->spa_root_vdev); } kmem_free(vd, gcount * sizeof (vdev_t *)); } static int spa_load(spa_t *spa, spa_load_state_t state, spa_import_type_t type, boolean_t mosconfig) { nvlist_t *config = spa->spa_config; char *ereport = FM_EREPORT_ZFS_POOL; char *comment; int error; uint64_t pool_guid; nvlist_t *nvl; if (nvlist_lookup_uint64(config, ZPOOL_CONFIG_POOL_GUID, &pool_guid)) return (SET_ERROR(EINVAL)); ASSERT(spa->spa_comment == NULL); if (nvlist_lookup_string(config, ZPOOL_CONFIG_COMMENT, &comment) == 0) spa->spa_comment = spa_strdup(comment); /* * Versioning wasn't explicitly added to the label until later, so if * it's not present treat it as the initial version. */ if (nvlist_lookup_uint64(config, ZPOOL_CONFIG_VERSION, &spa->spa_ubsync.ub_version) != 0) spa->spa_ubsync.ub_version = SPA_VERSION_INITIAL; (void) nvlist_lookup_uint64(config, ZPOOL_CONFIG_POOL_TXG, &spa->spa_config_txg); if ((state == SPA_LOAD_IMPORT || state == SPA_LOAD_TRYIMPORT) && spa_guid_exists(pool_guid, 0)) { error = SET_ERROR(EEXIST); } else { spa->spa_config_guid = pool_guid; if (nvlist_lookup_nvlist(config, ZPOOL_CONFIG_SPLIT, &nvl) == 0) { VERIFY(nvlist_dup(nvl, &spa->spa_config_splitting, KM_PUSHPAGE) == 0); } nvlist_free(spa->spa_load_info); spa->spa_load_info = fnvlist_alloc(); gethrestime(&spa->spa_loaded_ts); error = spa_load_impl(spa, pool_guid, config, state, type, mosconfig, &ereport); } spa->spa_minref = refcount_count(&spa->spa_refcount); if (error) { if (error != EEXIST) { spa->spa_loaded_ts.tv_sec = 0; spa->spa_loaded_ts.tv_nsec = 0; } if (error != EBADF) { zfs_ereport_post(ereport, spa, NULL, NULL, 0, 0); } } spa->spa_load_state = error ? SPA_LOAD_ERROR : SPA_LOAD_NONE; spa->spa_ena = 0; return (error); } /* * Load an existing storage pool, using the pool's builtin spa_config as a * source of configuration information. */ __attribute__((always_inline)) static inline int spa_load_impl(spa_t *spa, uint64_t pool_guid, nvlist_t *config, spa_load_state_t state, spa_import_type_t type, boolean_t mosconfig, char **ereport) { int error = 0; nvlist_t *nvroot = NULL; nvlist_t *label; vdev_t *rvd; uberblock_t *ub = &spa->spa_uberblock; uint64_t children, config_cache_txg = spa->spa_config_txg; int orig_mode = spa->spa_mode; int parse; uint64_t obj; boolean_t missing_feat_write = B_FALSE; /* * If this is an untrusted config, access the pool in read-only mode. * This prevents things like resilvering recently removed devices. */ if (!mosconfig) spa->spa_mode = FREAD; ASSERT(MUTEX_HELD(&spa_namespace_lock)); spa->spa_load_state = state; if (nvlist_lookup_nvlist(config, ZPOOL_CONFIG_VDEV_TREE, &nvroot)) return (SET_ERROR(EINVAL)); parse = (type == SPA_IMPORT_EXISTING ? VDEV_ALLOC_LOAD : VDEV_ALLOC_SPLIT); /* * Create "The Godfather" zio to hold all async IOs */ spa->spa_async_zio_root = zio_root(spa, NULL, NULL, ZIO_FLAG_CANFAIL | ZIO_FLAG_SPECULATIVE | ZIO_FLAG_GODFATHER); /* * Parse the configuration into a vdev tree. We explicitly set the * value that will be returned by spa_version() since parsing the * configuration requires knowing the version number. */ spa_config_enter(spa, SCL_ALL, FTAG, RW_WRITER); error = spa_config_parse(spa, &rvd, nvroot, NULL, 0, parse); spa_config_exit(spa, SCL_ALL, FTAG); if (error != 0) return (error); ASSERT(spa->spa_root_vdev == rvd); if (type != SPA_IMPORT_ASSEMBLE) { ASSERT(spa_guid(spa) == pool_guid); } /* * Try to open all vdevs, loading each label in the process. */ spa_config_enter(spa, SCL_ALL, FTAG, RW_WRITER); error = vdev_open(rvd); spa_config_exit(spa, SCL_ALL, FTAG); if (error != 0) return (error); /* * We need to validate the vdev labels against the configuration that * we have in hand, which is dependent on the setting of mosconfig. If * mosconfig is true then we're validating the vdev labels based on * that config. Otherwise, we're validating against the cached config * (zpool.cache) that was read when we loaded the zfs module, and then * later we will recursively call spa_load() and validate against * the vdev config. * * If we're assembling a new pool that's been split off from an * existing pool, the labels haven't yet been updated so we skip * validation for now. */ if (type != SPA_IMPORT_ASSEMBLE) { spa_config_enter(spa, SCL_ALL, FTAG, RW_WRITER); error = vdev_validate(rvd, mosconfig); spa_config_exit(spa, SCL_ALL, FTAG); if (error != 0) return (error); if (rvd->vdev_state <= VDEV_STATE_CANT_OPEN) return (SET_ERROR(ENXIO)); } /* * Find the best uberblock. */ vdev_uberblock_load(rvd, ub, &label); /* * If we weren't able to find a single valid uberblock, return failure. */ if (ub->ub_txg == 0) { nvlist_free(label); return (spa_vdev_err(rvd, VDEV_AUX_CORRUPT_DATA, ENXIO)); } /* * If the pool has an unsupported version we can't open it. */ if (!SPA_VERSION_IS_SUPPORTED(ub->ub_version)) { nvlist_free(label); return (spa_vdev_err(rvd, VDEV_AUX_VERSION_NEWER, ENOTSUP)); } if (ub->ub_version >= SPA_VERSION_FEATURES) { nvlist_t *features; /* * If we weren't able to find what's necessary for reading the * MOS in the label, return failure. */ if (label == NULL || nvlist_lookup_nvlist(label, ZPOOL_CONFIG_FEATURES_FOR_READ, &features) != 0) { nvlist_free(label); return (spa_vdev_err(rvd, VDEV_AUX_CORRUPT_DATA, ENXIO)); } /* * Update our in-core representation with the definitive values * from the label. */ nvlist_free(spa->spa_label_features); VERIFY(nvlist_dup(features, &spa->spa_label_features, 0) == 0); } nvlist_free(label); /* * Look through entries in the label nvlist's features_for_read. If * there is a feature listed there which we don't understand then we * cannot open a pool. */ if (ub->ub_version >= SPA_VERSION_FEATURES) { nvlist_t *unsup_feat; nvpair_t *nvp; VERIFY(nvlist_alloc(&unsup_feat, NV_UNIQUE_NAME, KM_SLEEP) == 0); for (nvp = nvlist_next_nvpair(spa->spa_label_features, NULL); nvp != NULL; nvp = nvlist_next_nvpair(spa->spa_label_features, nvp)) { if (!zfeature_is_supported(nvpair_name(nvp))) { VERIFY(nvlist_add_string(unsup_feat, nvpair_name(nvp), "") == 0); } } if (!nvlist_empty(unsup_feat)) { VERIFY(nvlist_add_nvlist(spa->spa_load_info, ZPOOL_CONFIG_UNSUP_FEAT, unsup_feat) == 0); nvlist_free(unsup_feat); return (spa_vdev_err(rvd, VDEV_AUX_UNSUP_FEAT, ENOTSUP)); } nvlist_free(unsup_feat); } /* * If the vdev guid sum doesn't match the uberblock, we have an * incomplete configuration. We first check to see if the pool * is aware of the complete config (i.e ZPOOL_CONFIG_VDEV_CHILDREN). * If it is, defer the vdev_guid_sum check till later so we * can handle missing vdevs. */ if (nvlist_lookup_uint64(config, ZPOOL_CONFIG_VDEV_CHILDREN, &children) != 0 && mosconfig && type != SPA_IMPORT_ASSEMBLE && rvd->vdev_guid_sum != ub->ub_guid_sum) return (spa_vdev_err(rvd, VDEV_AUX_BAD_GUID_SUM, ENXIO)); if (type != SPA_IMPORT_ASSEMBLE && spa->spa_config_splitting) { spa_config_enter(spa, SCL_ALL, FTAG, RW_WRITER); spa_try_repair(spa, config); spa_config_exit(spa, SCL_ALL, FTAG); nvlist_free(spa->spa_config_splitting); spa->spa_config_splitting = NULL; } /* * Initialize internal SPA structures. */ spa->spa_state = POOL_STATE_ACTIVE; spa->spa_ubsync = spa->spa_uberblock; spa->spa_verify_min_txg = spa->spa_extreme_rewind ? TXG_INITIAL - 1 : spa_last_synced_txg(spa) - TXG_DEFER_SIZE - 1; spa->spa_first_txg = spa->spa_last_ubsync_txg ? spa->spa_last_ubsync_txg : spa_last_synced_txg(spa) + 1; spa->spa_claim_max_txg = spa->spa_first_txg; spa->spa_prev_software_version = ub->ub_software_version; error = dsl_pool_init(spa, spa->spa_first_txg, &spa->spa_dsl_pool); if (error) return (spa_vdev_err(rvd, VDEV_AUX_CORRUPT_DATA, EIO)); spa->spa_meta_objset = spa->spa_dsl_pool->dp_meta_objset; if (spa_dir_prop(spa, DMU_POOL_CONFIG, &spa->spa_config_object) != 0) return (spa_vdev_err(rvd, VDEV_AUX_CORRUPT_DATA, EIO)); if (spa_version(spa) >= SPA_VERSION_FEATURES) { boolean_t missing_feat_read = B_FALSE; nvlist_t *unsup_feat, *enabled_feat; spa_feature_t i; if (spa_dir_prop(spa, DMU_POOL_FEATURES_FOR_READ, &spa->spa_feat_for_read_obj) != 0) { return (spa_vdev_err(rvd, VDEV_AUX_CORRUPT_DATA, EIO)); } if (spa_dir_prop(spa, DMU_POOL_FEATURES_FOR_WRITE, &spa->spa_feat_for_write_obj) != 0) { return (spa_vdev_err(rvd, VDEV_AUX_CORRUPT_DATA, EIO)); } if (spa_dir_prop(spa, DMU_POOL_FEATURE_DESCRIPTIONS, &spa->spa_feat_desc_obj) != 0) { return (spa_vdev_err(rvd, VDEV_AUX_CORRUPT_DATA, EIO)); } enabled_feat = fnvlist_alloc(); unsup_feat = fnvlist_alloc(); if (!spa_features_check(spa, B_FALSE, unsup_feat, enabled_feat)) missing_feat_read = B_TRUE; if (spa_writeable(spa) || state == SPA_LOAD_TRYIMPORT) { if (!spa_features_check(spa, B_TRUE, unsup_feat, enabled_feat)) { missing_feat_write = B_TRUE; } } fnvlist_add_nvlist(spa->spa_load_info, ZPOOL_CONFIG_ENABLED_FEAT, enabled_feat); if (!nvlist_empty(unsup_feat)) { fnvlist_add_nvlist(spa->spa_load_info, ZPOOL_CONFIG_UNSUP_FEAT, unsup_feat); } fnvlist_free(enabled_feat); fnvlist_free(unsup_feat); if (!missing_feat_read) { fnvlist_add_boolean(spa->spa_load_info, ZPOOL_CONFIG_CAN_RDONLY); } /* * If the state is SPA_LOAD_TRYIMPORT, our objective is * twofold: to determine whether the pool is available for * import in read-write mode and (if it is not) whether the * pool is available for import in read-only mode. If the pool * is available for import in read-write mode, it is displayed * as available in userland; if it is not available for import * in read-only mode, it is displayed as unavailable in * userland. If the pool is available for import in read-only * mode but not read-write mode, it is displayed as unavailable * in userland with a special note that the pool is actually * available for open in read-only mode. * * As a result, if the state is SPA_LOAD_TRYIMPORT and we are * missing a feature for write, we must first determine whether * the pool can be opened read-only before returning to * userland in order to know whether to display the * abovementioned note. */ if (missing_feat_read || (missing_feat_write && spa_writeable(spa))) { return (spa_vdev_err(rvd, VDEV_AUX_UNSUP_FEAT, ENOTSUP)); } /* * Load refcounts for ZFS features from disk into an in-memory * cache during SPA initialization. */ for (i = 0; i < SPA_FEATURES; i++) { uint64_t refcount; error = feature_get_refcount_from_disk(spa, &spa_feature_table[i], &refcount); if (error == 0) { spa->spa_feat_refcount_cache[i] = refcount; } else if (error == ENOTSUP) { spa->spa_feat_refcount_cache[i] = SPA_FEATURE_DISABLED; } else { return (spa_vdev_err(rvd, VDEV_AUX_CORRUPT_DATA, EIO)); } } } if (spa_feature_is_active(spa, SPA_FEATURE_ENABLED_TXG)) { if (spa_dir_prop(spa, DMU_POOL_FEATURE_ENABLED_TXG, &spa->spa_feat_enabled_txg_obj) != 0) return (spa_vdev_err(rvd, VDEV_AUX_CORRUPT_DATA, EIO)); } spa->spa_is_initializing = B_TRUE; error = dsl_pool_open(spa->spa_dsl_pool); spa->spa_is_initializing = B_FALSE; if (error != 0) return (spa_vdev_err(rvd, VDEV_AUX_CORRUPT_DATA, EIO)); if (!mosconfig) { uint64_t hostid; nvlist_t *policy = NULL, *nvconfig; if (load_nvlist(spa, spa->spa_config_object, &nvconfig) != 0) return (spa_vdev_err(rvd, VDEV_AUX_CORRUPT_DATA, EIO)); if (!spa_is_root(spa) && nvlist_lookup_uint64(nvconfig, ZPOOL_CONFIG_HOSTID, &hostid) == 0) { char *hostname; unsigned long myhostid = 0; VERIFY(nvlist_lookup_string(nvconfig, ZPOOL_CONFIG_HOSTNAME, &hostname) == 0); #ifdef _KERNEL myhostid = zone_get_hostid(NULL); #else /* _KERNEL */ /* * We're emulating the system's hostid in userland, so * we can't use zone_get_hostid(). */ (void) ddi_strtoul(hw_serial, NULL, 10, &myhostid); #endif /* _KERNEL */ if (hostid != 0 && myhostid != 0 && hostid != myhostid) { nvlist_free(nvconfig); cmn_err(CE_WARN, "pool '%s' could not be " "loaded as it was last accessed by another " "system (host: %s hostid: 0x%lx). See: " "http://zfsonlinux.org/msg/ZFS-8000-EY", spa_name(spa), hostname, (unsigned long)hostid); return (SET_ERROR(EBADF)); } } if (nvlist_lookup_nvlist(spa->spa_config, ZPOOL_REWIND_POLICY, &policy) == 0) VERIFY(nvlist_add_nvlist(nvconfig, ZPOOL_REWIND_POLICY, policy) == 0); spa_config_set(spa, nvconfig); spa_unload(spa); spa_deactivate(spa); spa_activate(spa, orig_mode); return (spa_load(spa, state, SPA_IMPORT_EXISTING, B_TRUE)); } if (spa_dir_prop(spa, DMU_POOL_SYNC_BPOBJ, &obj) != 0) return (spa_vdev_err(rvd, VDEV_AUX_CORRUPT_DATA, EIO)); error = bpobj_open(&spa->spa_deferred_bpobj, spa->spa_meta_objset, obj); if (error != 0) return (spa_vdev_err(rvd, VDEV_AUX_CORRUPT_DATA, EIO)); /* * Load the bit that tells us to use the new accounting function * (raid-z deflation). If we have an older pool, this will not * be present. */ error = spa_dir_prop(spa, DMU_POOL_DEFLATE, &spa->spa_deflate); if (error != 0 && error != ENOENT) return (spa_vdev_err(rvd, VDEV_AUX_CORRUPT_DATA, EIO)); error = spa_dir_prop(spa, DMU_POOL_CREATION_VERSION, &spa->spa_creation_version); if (error != 0 && error != ENOENT) return (spa_vdev_err(rvd, VDEV_AUX_CORRUPT_DATA, EIO)); /* * Load the persistent error log. If we have an older pool, this will * not be present. */ error = spa_dir_prop(spa, DMU_POOL_ERRLOG_LAST, &spa->spa_errlog_last); if (error != 0 && error != ENOENT) return (spa_vdev_err(rvd, VDEV_AUX_CORRUPT_DATA, EIO)); error = spa_dir_prop(spa, DMU_POOL_ERRLOG_SCRUB, &spa->spa_errlog_scrub); if (error != 0 && error != ENOENT) return (spa_vdev_err(rvd, VDEV_AUX_CORRUPT_DATA, EIO)); /* * Load the history object. If we have an older pool, this * will not be present. */ error = spa_dir_prop(spa, DMU_POOL_HISTORY, &spa->spa_history); if (error != 0 && error != ENOENT) return (spa_vdev_err(rvd, VDEV_AUX_CORRUPT_DATA, EIO)); /* * If we're assembling the pool from the split-off vdevs of * an existing pool, we don't want to attach the spares & cache * devices. */ /* * Load any hot spares for this pool. */ error = spa_dir_prop(spa, DMU_POOL_SPARES, &spa->spa_spares.sav_object); if (error != 0 && error != ENOENT) return (spa_vdev_err(rvd, VDEV_AUX_CORRUPT_DATA, EIO)); if (error == 0 && type != SPA_IMPORT_ASSEMBLE) { ASSERT(spa_version(spa) >= SPA_VERSION_SPARES); if (load_nvlist(spa, spa->spa_spares.sav_object, &spa->spa_spares.sav_config) != 0) return (spa_vdev_err(rvd, VDEV_AUX_CORRUPT_DATA, EIO)); spa_config_enter(spa, SCL_ALL, FTAG, RW_WRITER); spa_load_spares(spa); spa_config_exit(spa, SCL_ALL, FTAG); } else if (error == 0) { spa->spa_spares.sav_sync = B_TRUE; } /* * Load any level 2 ARC devices for this pool. */ error = spa_dir_prop(spa, DMU_POOL_L2CACHE, &spa->spa_l2cache.sav_object); if (error != 0 && error != ENOENT) return (spa_vdev_err(rvd, VDEV_AUX_CORRUPT_DATA, EIO)); if (error == 0 && type != SPA_IMPORT_ASSEMBLE) { ASSERT(spa_version(spa) >= SPA_VERSION_L2CACHE); if (load_nvlist(spa, spa->spa_l2cache.sav_object, &spa->spa_l2cache.sav_config) != 0) return (spa_vdev_err(rvd, VDEV_AUX_CORRUPT_DATA, EIO)); spa_config_enter(spa, SCL_ALL, FTAG, RW_WRITER); spa_load_l2cache(spa); spa_config_exit(spa, SCL_ALL, FTAG); } else if (error == 0) { spa->spa_l2cache.sav_sync = B_TRUE; } spa->spa_delegation = zpool_prop_default_numeric(ZPOOL_PROP_DELEGATION); error = spa_dir_prop(spa, DMU_POOL_PROPS, &spa->spa_pool_props_object); if (error && error != ENOENT) return (spa_vdev_err(rvd, VDEV_AUX_CORRUPT_DATA, EIO)); if (error == 0) { uint64_t autoreplace = 0; spa_prop_find(spa, ZPOOL_PROP_BOOTFS, &spa->spa_bootfs); spa_prop_find(spa, ZPOOL_PROP_AUTOREPLACE, &autoreplace); spa_prop_find(spa, ZPOOL_PROP_DELEGATION, &spa->spa_delegation); spa_prop_find(spa, ZPOOL_PROP_FAILUREMODE, &spa->spa_failmode); spa_prop_find(spa, ZPOOL_PROP_AUTOEXPAND, &spa->spa_autoexpand); spa_prop_find(spa, ZPOOL_PROP_DEDUPDITTO, &spa->spa_dedup_ditto); spa->spa_autoreplace = (autoreplace != 0); } /* * If the 'autoreplace' property is set, then post a resource notifying * the ZFS DE that it should not issue any faults for unopenable * devices. We also iterate over the vdevs, and post a sysevent for any * unopenable vdevs so that the normal autoreplace handler can take * over. */ if (spa->spa_autoreplace && state != SPA_LOAD_TRYIMPORT) { spa_check_removed(spa->spa_root_vdev); /* * For the import case, this is done in spa_import(), because * at this point we're using the spare definitions from * the MOS config, not necessarily from the userland config. */ if (state != SPA_LOAD_IMPORT) { spa_aux_check_removed(&spa->spa_spares); spa_aux_check_removed(&spa->spa_l2cache); } } /* * Load the vdev state for all toplevel vdevs. */ vdev_load(rvd); /* * Propagate the leaf DTLs we just loaded all the way up the tree. */ spa_config_enter(spa, SCL_ALL, FTAG, RW_WRITER); vdev_dtl_reassess(rvd, 0, 0, B_FALSE); spa_config_exit(spa, SCL_ALL, FTAG); /* * Load the DDTs (dedup tables). */ error = ddt_load(spa); if (error != 0) return (spa_vdev_err(rvd, VDEV_AUX_CORRUPT_DATA, EIO)); spa_update_dspace(spa); /* * Validate the config, using the MOS config to fill in any * information which might be missing. If we fail to validate * the config then declare the pool unfit for use. If we're * assembling a pool from a split, the log is not transferred * over. */ if (type != SPA_IMPORT_ASSEMBLE) { nvlist_t *nvconfig; if (load_nvlist(spa, spa->spa_config_object, &nvconfig) != 0) return (spa_vdev_err(rvd, VDEV_AUX_CORRUPT_DATA, EIO)); if (!spa_config_valid(spa, nvconfig)) { nvlist_free(nvconfig); return (spa_vdev_err(rvd, VDEV_AUX_BAD_GUID_SUM, ENXIO)); } nvlist_free(nvconfig); /* * Now that we've validated the config, check the state of the * root vdev. If it can't be opened, it indicates one or * more toplevel vdevs are faulted. */ if (rvd->vdev_state <= VDEV_STATE_CANT_OPEN) return (SET_ERROR(ENXIO)); if (spa_check_logs(spa)) { *ereport = FM_EREPORT_ZFS_LOG_REPLAY; return (spa_vdev_err(rvd, VDEV_AUX_BAD_LOG, ENXIO)); } } if (missing_feat_write) { ASSERT(state == SPA_LOAD_TRYIMPORT); /* * At this point, we know that we can open the pool in * read-only mode but not read-write mode. We now have enough * information and can return to userland. */ return (spa_vdev_err(rvd, VDEV_AUX_UNSUP_FEAT, ENOTSUP)); } /* * We've successfully opened the pool, verify that we're ready * to start pushing transactions. */ if (state != SPA_LOAD_TRYIMPORT) { if ((error = spa_load_verify(spa))) return (spa_vdev_err(rvd, VDEV_AUX_CORRUPT_DATA, error)); } if (spa_writeable(spa) && (state == SPA_LOAD_RECOVER || spa->spa_load_max_txg == UINT64_MAX)) { dmu_tx_t *tx; int need_update = B_FALSE; int c; ASSERT(state != SPA_LOAD_TRYIMPORT); /* * Claim log blocks that haven't been committed yet. * This must all happen in a single txg. * Note: spa_claim_max_txg is updated by spa_claim_notify(), * invoked from zil_claim_log_block()'s i/o done callback. * Price of rollback is that we abandon the log. */ spa->spa_claiming = B_TRUE; tx = dmu_tx_create_assigned(spa_get_dsl(spa), spa_first_txg(spa)); (void) dmu_objset_find(spa_name(spa), zil_claim, tx, DS_FIND_CHILDREN); dmu_tx_commit(tx); spa->spa_claiming = B_FALSE; spa_set_log_state(spa, SPA_LOG_GOOD); spa->spa_sync_on = B_TRUE; txg_sync_start(spa->spa_dsl_pool); /* * Wait for all claims to sync. We sync up to the highest * claimed log block birth time so that claimed log blocks * don't appear to be from the future. spa_claim_max_txg * will have been set for us by either zil_check_log_chain() * (invoked from spa_check_logs()) or zil_claim() above. */ txg_wait_synced(spa->spa_dsl_pool, spa->spa_claim_max_txg); /* * If the config cache is stale, or we have uninitialized * metaslabs (see spa_vdev_add()), then update the config. * * If this is a verbatim import, trust the current * in-core spa_config and update the disk labels. */ if (config_cache_txg != spa->spa_config_txg || state == SPA_LOAD_IMPORT || state == SPA_LOAD_RECOVER || (spa->spa_import_flags & ZFS_IMPORT_VERBATIM)) need_update = B_TRUE; for (c = 0; c < rvd->vdev_children; c++) if (rvd->vdev_child[c]->vdev_ms_array == 0) need_update = B_TRUE; /* * Update the config cache asychronously in case we're the * root pool, in which case the config cache isn't writable yet. */ if (need_update) spa_async_request(spa, SPA_ASYNC_CONFIG_UPDATE); /* * Check all DTLs to see if anything needs resilvering. */ if (!dsl_scan_resilvering(spa->spa_dsl_pool) && vdev_resilver_needed(rvd, NULL, NULL)) spa_async_request(spa, SPA_ASYNC_RESILVER); /* * Log the fact that we booted up (so that we can detect if * we rebooted in the middle of an operation). */ spa_history_log_version(spa, "open"); /* * Delete any inconsistent datasets. */ (void) dmu_objset_find(spa_name(spa), dsl_destroy_inconsistent, NULL, DS_FIND_CHILDREN); /* * Clean up any stale temporary dataset userrefs. */ dsl_pool_clean_tmp_userrefs(spa->spa_dsl_pool); } return (0); } static int spa_load_retry(spa_t *spa, spa_load_state_t state, int mosconfig) { int mode = spa->spa_mode; spa_unload(spa); spa_deactivate(spa); spa->spa_load_max_txg--; spa_activate(spa, mode); spa_async_suspend(spa); return (spa_load(spa, state, SPA_IMPORT_EXISTING, mosconfig)); } /* * If spa_load() fails this function will try loading prior txg's. If * 'state' is SPA_LOAD_RECOVER and one of these loads succeeds the pool * will be rewound to that txg. If 'state' is not SPA_LOAD_RECOVER this * function will not rewind the pool and will return the same error as * spa_load(). */ static int spa_load_best(spa_t *spa, spa_load_state_t state, int mosconfig, uint64_t max_request, int rewind_flags) { nvlist_t *loadinfo = NULL; nvlist_t *config = NULL; int load_error, rewind_error; uint64_t safe_rewind_txg; uint64_t min_txg; if (spa->spa_load_txg && state == SPA_LOAD_RECOVER) { spa->spa_load_max_txg = spa->spa_load_txg; spa_set_log_state(spa, SPA_LOG_CLEAR); } else { spa->spa_load_max_txg = max_request; } load_error = rewind_error = spa_load(spa, state, SPA_IMPORT_EXISTING, mosconfig); if (load_error == 0) return (0); if (spa->spa_root_vdev != NULL) config = spa_config_generate(spa, NULL, -1ULL, B_TRUE); spa->spa_last_ubsync_txg = spa->spa_uberblock.ub_txg; spa->spa_last_ubsync_txg_ts = spa->spa_uberblock.ub_timestamp; if (rewind_flags & ZPOOL_NEVER_REWIND) { nvlist_free(config); return (load_error); } if (state == SPA_LOAD_RECOVER) { /* Price of rolling back is discarding txgs, including log */ spa_set_log_state(spa, SPA_LOG_CLEAR); } else { /* * If we aren't rolling back save the load info from our first * import attempt so that we can restore it after attempting * to rewind. */ loadinfo = spa->spa_load_info; spa->spa_load_info = fnvlist_alloc(); } spa->spa_load_max_txg = spa->spa_last_ubsync_txg; safe_rewind_txg = spa->spa_last_ubsync_txg - TXG_DEFER_SIZE; min_txg = (rewind_flags & ZPOOL_EXTREME_REWIND) ? TXG_INITIAL : safe_rewind_txg; /* * Continue as long as we're finding errors, we're still within * the acceptable rewind range, and we're still finding uberblocks */ while (rewind_error && spa->spa_uberblock.ub_txg >= min_txg && spa->spa_uberblock.ub_txg <= spa->spa_load_max_txg) { if (spa->spa_load_max_txg < safe_rewind_txg) spa->spa_extreme_rewind = B_TRUE; rewind_error = spa_load_retry(spa, state, mosconfig); } spa->spa_extreme_rewind = B_FALSE; spa->spa_load_max_txg = UINT64_MAX; if (config && (rewind_error || state != SPA_LOAD_RECOVER)) spa_config_set(spa, config); if (state == SPA_LOAD_RECOVER) { ASSERT3P(loadinfo, ==, NULL); return (rewind_error); } else { /* Store the rewind info as part of the initial load info */ fnvlist_add_nvlist(loadinfo, ZPOOL_CONFIG_REWIND_INFO, spa->spa_load_info); /* Restore the initial load info */ fnvlist_free(spa->spa_load_info); spa->spa_load_info = loadinfo; return (load_error); } } /* * Pool Open/Import * * The import case is identical to an open except that the configuration is sent * down from userland, instead of grabbed from the configuration cache. For the * case of an open, the pool configuration will exist in the * POOL_STATE_UNINITIALIZED state. * * The stats information (gen/count/ustats) is used to gather vdev statistics at * the same time open the pool, without having to keep around the spa_t in some * ambiguous state. */ static int spa_open_common(const char *pool, spa_t **spapp, void *tag, nvlist_t *nvpolicy, nvlist_t **config) { spa_t *spa; spa_load_state_t state = SPA_LOAD_OPEN; int error; int locked = B_FALSE; int firstopen = B_FALSE; *spapp = NULL; /* * As disgusting as this is, we need to support recursive calls to this * function because dsl_dir_open() is called during spa_load(), and ends * up calling spa_open() again. The real fix is to figure out how to * avoid dsl_dir_open() calling this in the first place. */ if (mutex_owner(&spa_namespace_lock) != curthread) { mutex_enter(&spa_namespace_lock); locked = B_TRUE; } if ((spa = spa_lookup(pool)) == NULL) { if (locked) mutex_exit(&spa_namespace_lock); return (SET_ERROR(ENOENT)); } if (spa->spa_state == POOL_STATE_UNINITIALIZED) { zpool_rewind_policy_t policy; firstopen = B_TRUE; zpool_get_rewind_policy(nvpolicy ? nvpolicy : spa->spa_config, &policy); if (policy.zrp_request & ZPOOL_DO_REWIND) state = SPA_LOAD_RECOVER; spa_activate(spa, spa_mode_global); if (state != SPA_LOAD_RECOVER) spa->spa_last_ubsync_txg = spa->spa_load_txg = 0; error = spa_load_best(spa, state, B_FALSE, policy.zrp_txg, policy.zrp_request); if (error == EBADF) { /* * If vdev_validate() returns failure (indicated by * EBADF), it indicates that one of the vdevs indicates * that the pool has been exported or destroyed. If * this is the case, the config cache is out of sync and * we should remove the pool from the namespace. */ spa_unload(spa); spa_deactivate(spa); spa_config_sync(spa, B_TRUE, B_TRUE); spa_remove(spa); if (locked) mutex_exit(&spa_namespace_lock); return (SET_ERROR(ENOENT)); } if (error) { /* * We can't open the pool, but we still have useful * information: the state of each vdev after the * attempted vdev_open(). Return this to the user. */ if (config != NULL && spa->spa_config) { VERIFY(nvlist_dup(spa->spa_config, config, KM_PUSHPAGE) == 0); VERIFY(nvlist_add_nvlist(*config, ZPOOL_CONFIG_LOAD_INFO, spa->spa_load_info) == 0); } spa_unload(spa); spa_deactivate(spa); spa->spa_last_open_failed = error; if (locked) mutex_exit(&spa_namespace_lock); *spapp = NULL; return (error); } } spa_open_ref(spa, tag); if (config != NULL) *config = spa_config_generate(spa, NULL, -1ULL, B_TRUE); /* * If we've recovered the pool, pass back any information we * gathered while doing the load. */ if (state == SPA_LOAD_RECOVER) { VERIFY(nvlist_add_nvlist(*config, ZPOOL_CONFIG_LOAD_INFO, spa->spa_load_info) == 0); } if (locked) { spa->spa_last_open_failed = 0; spa->spa_last_ubsync_txg = 0; spa->spa_load_txg = 0; mutex_exit(&spa_namespace_lock); } #ifdef _KERNEL if (firstopen) zvol_create_minors(spa->spa_name); #endif *spapp = spa; return (0); } int spa_open_rewind(const char *name, spa_t **spapp, void *tag, nvlist_t *policy, nvlist_t **config) { return (spa_open_common(name, spapp, tag, policy, config)); } int spa_open(const char *name, spa_t **spapp, void *tag) { return (spa_open_common(name, spapp, tag, NULL, NULL)); } /* * Lookup the given spa_t, incrementing the inject count in the process, * preventing it from being exported or destroyed. */ spa_t * spa_inject_addref(char *name) { spa_t *spa; mutex_enter(&spa_namespace_lock); if ((spa = spa_lookup(name)) == NULL) { mutex_exit(&spa_namespace_lock); return (NULL); } spa->spa_inject_ref++; mutex_exit(&spa_namespace_lock); return (spa); } void spa_inject_delref(spa_t *spa) { mutex_enter(&spa_namespace_lock); spa->spa_inject_ref--; mutex_exit(&spa_namespace_lock); } /* * Add spares device information to the nvlist. */ static void spa_add_spares(spa_t *spa, nvlist_t *config) { nvlist_t **spares; uint_t i, nspares; nvlist_t *nvroot; uint64_t guid; vdev_stat_t *vs; uint_t vsc; uint64_t pool; ASSERT(spa_config_held(spa, SCL_CONFIG, RW_READER)); if (spa->spa_spares.sav_count == 0) return; VERIFY(nvlist_lookup_nvlist(config, ZPOOL_CONFIG_VDEV_TREE, &nvroot) == 0); VERIFY(nvlist_lookup_nvlist_array(spa->spa_spares.sav_config, ZPOOL_CONFIG_SPARES, &spares, &nspares) == 0); if (nspares != 0) { VERIFY(nvlist_add_nvlist_array(nvroot, ZPOOL_CONFIG_SPARES, spares, nspares) == 0); VERIFY(nvlist_lookup_nvlist_array(nvroot, ZPOOL_CONFIG_SPARES, &spares, &nspares) == 0); /* * Go through and find any spares which have since been * repurposed as an active spare. If this is the case, update * their status appropriately. */ for (i = 0; i < nspares; i++) { VERIFY(nvlist_lookup_uint64(spares[i], ZPOOL_CONFIG_GUID, &guid) == 0); if (spa_spare_exists(guid, &pool, NULL) && pool != 0ULL) { VERIFY(nvlist_lookup_uint64_array( spares[i], ZPOOL_CONFIG_VDEV_STATS, (uint64_t **)&vs, &vsc) == 0); vs->vs_state = VDEV_STATE_CANT_OPEN; vs->vs_aux = VDEV_AUX_SPARED; } } } } /* * Add l2cache device information to the nvlist, including vdev stats. */ static void spa_add_l2cache(spa_t *spa, nvlist_t *config) { nvlist_t **l2cache; uint_t i, j, nl2cache; nvlist_t *nvroot; uint64_t guid; vdev_t *vd; vdev_stat_t *vs; uint_t vsc; ASSERT(spa_config_held(spa, SCL_CONFIG, RW_READER)); if (spa->spa_l2cache.sav_count == 0) return; VERIFY(nvlist_lookup_nvlist(config, ZPOOL_CONFIG_VDEV_TREE, &nvroot) == 0); VERIFY(nvlist_lookup_nvlist_array(spa->spa_l2cache.sav_config, ZPOOL_CONFIG_L2CACHE, &l2cache, &nl2cache) == 0); if (nl2cache != 0) { VERIFY(nvlist_add_nvlist_array(nvroot, ZPOOL_CONFIG_L2CACHE, l2cache, nl2cache) == 0); VERIFY(nvlist_lookup_nvlist_array(nvroot, ZPOOL_CONFIG_L2CACHE, &l2cache, &nl2cache) == 0); /* * Update level 2 cache device stats. */ for (i = 0; i < nl2cache; i++) { VERIFY(nvlist_lookup_uint64(l2cache[i], ZPOOL_CONFIG_GUID, &guid) == 0); vd = NULL; for (j = 0; j < spa->spa_l2cache.sav_count; j++) { if (guid == spa->spa_l2cache.sav_vdevs[j]->vdev_guid) { vd = spa->spa_l2cache.sav_vdevs[j]; break; } } ASSERT(vd != NULL); VERIFY(nvlist_lookup_uint64_array(l2cache[i], ZPOOL_CONFIG_VDEV_STATS, (uint64_t **)&vs, &vsc) == 0); vdev_get_stats(vd, vs); } } } static void spa_add_feature_stats(spa_t *spa, nvlist_t *config) { nvlist_t *features; zap_cursor_t zc; zap_attribute_t za; ASSERT(spa_config_held(spa, SCL_CONFIG, RW_READER)); VERIFY(nvlist_alloc(&features, NV_UNIQUE_NAME, KM_SLEEP) == 0); if (spa->spa_feat_for_read_obj != 0) { for (zap_cursor_init(&zc, spa->spa_meta_objset, spa->spa_feat_for_read_obj); zap_cursor_retrieve(&zc, &za) == 0; zap_cursor_advance(&zc)) { ASSERT(za.za_integer_length == sizeof (uint64_t) && za.za_num_integers == 1); VERIFY3U(0, ==, nvlist_add_uint64(features, za.za_name, za.za_first_integer)); } zap_cursor_fini(&zc); } if (spa->spa_feat_for_write_obj != 0) { for (zap_cursor_init(&zc, spa->spa_meta_objset, spa->spa_feat_for_write_obj); zap_cursor_retrieve(&zc, &za) == 0; zap_cursor_advance(&zc)) { ASSERT(za.za_integer_length == sizeof (uint64_t) && za.za_num_integers == 1); VERIFY3U(0, ==, nvlist_add_uint64(features, za.za_name, za.za_first_integer)); } zap_cursor_fini(&zc); } VERIFY(nvlist_add_nvlist(config, ZPOOL_CONFIG_FEATURE_STATS, features) == 0); nvlist_free(features); } int spa_get_stats(const char *name, nvlist_t **config, char *altroot, size_t buflen) { int error; spa_t *spa; *config = NULL; error = spa_open_common(name, &spa, FTAG, NULL, config); if (spa != NULL) { /* * This still leaves a window of inconsistency where the spares * or l2cache devices could change and the config would be * self-inconsistent. */ spa_config_enter(spa, SCL_CONFIG, FTAG, RW_READER); if (*config != NULL) { uint64_t loadtimes[2]; loadtimes[0] = spa->spa_loaded_ts.tv_sec; loadtimes[1] = spa->spa_loaded_ts.tv_nsec; VERIFY(nvlist_add_uint64_array(*config, ZPOOL_CONFIG_LOADED_TIME, loadtimes, 2) == 0); VERIFY(nvlist_add_uint64(*config, ZPOOL_CONFIG_ERRCOUNT, spa_get_errlog_size(spa)) == 0); if (spa_suspended(spa)) VERIFY(nvlist_add_uint64(*config, ZPOOL_CONFIG_SUSPENDED, spa->spa_failmode) == 0); spa_add_spares(spa, *config); spa_add_l2cache(spa, *config); spa_add_feature_stats(spa, *config); } } /* * We want to get the alternate root even for faulted pools, so we cheat * and call spa_lookup() directly. */ if (altroot) { if (spa == NULL) { mutex_enter(&spa_namespace_lock); spa = spa_lookup(name); if (spa) spa_altroot(spa, altroot, buflen); else altroot[0] = '\0'; spa = NULL; mutex_exit(&spa_namespace_lock); } else { spa_altroot(spa, altroot, buflen); } } if (spa != NULL) { spa_config_exit(spa, SCL_CONFIG, FTAG); spa_close(spa, FTAG); } return (error); } /* * Validate that the auxiliary device array is well formed. We must have an * array of nvlists, each which describes a valid leaf vdev. If this is an * import (mode is VDEV_ALLOC_SPARE), then we allow corrupted spares to be * specified, as long as they are well-formed. */ static int spa_validate_aux_devs(spa_t *spa, nvlist_t *nvroot, uint64_t crtxg, int mode, spa_aux_vdev_t *sav, const char *config, uint64_t version, vdev_labeltype_t label) { nvlist_t **dev; uint_t i, ndev; vdev_t *vd; int error; ASSERT(spa_config_held(spa, SCL_ALL, RW_WRITER) == SCL_ALL); /* * It's acceptable to have no devs specified. */ if (nvlist_lookup_nvlist_array(nvroot, config, &dev, &ndev) != 0) return (0); if (ndev == 0) return (SET_ERROR(EINVAL)); /* * Make sure the pool is formatted with a version that supports this * device type. */ if (spa_version(spa) < version) return (SET_ERROR(ENOTSUP)); /* * Set the pending device list so we correctly handle device in-use * checking. */ sav->sav_pending = dev; sav->sav_npending = ndev; for (i = 0; i < ndev; i++) { if ((error = spa_config_parse(spa, &vd, dev[i], NULL, 0, mode)) != 0) goto out; if (!vd->vdev_ops->vdev_op_leaf) { vdev_free(vd); error = SET_ERROR(EINVAL); goto out; } /* * The L2ARC currently only supports disk devices in * kernel context. For user-level testing, we allow it. */ #ifdef _KERNEL if ((strcmp(config, ZPOOL_CONFIG_L2CACHE) == 0) && strcmp(vd->vdev_ops->vdev_op_type, VDEV_TYPE_DISK) != 0) { error = SET_ERROR(ENOTBLK); vdev_free(vd); goto out; } #endif vd->vdev_top = vd; if ((error = vdev_open(vd)) == 0 && (error = vdev_label_init(vd, crtxg, label)) == 0) { VERIFY(nvlist_add_uint64(dev[i], ZPOOL_CONFIG_GUID, vd->vdev_guid) == 0); } vdev_free(vd); if (error && (mode != VDEV_ALLOC_SPARE && mode != VDEV_ALLOC_L2CACHE)) goto out; else error = 0; } out: sav->sav_pending = NULL; sav->sav_npending = 0; return (error); } static int spa_validate_aux(spa_t *spa, nvlist_t *nvroot, uint64_t crtxg, int mode) { int error; ASSERT(spa_config_held(spa, SCL_ALL, RW_WRITER) == SCL_ALL); if ((error = spa_validate_aux_devs(spa, nvroot, crtxg, mode, &spa->spa_spares, ZPOOL_CONFIG_SPARES, SPA_VERSION_SPARES, VDEV_LABEL_SPARE)) != 0) { return (error); } return (spa_validate_aux_devs(spa, nvroot, crtxg, mode, &spa->spa_l2cache, ZPOOL_CONFIG_L2CACHE, SPA_VERSION_L2CACHE, VDEV_LABEL_L2CACHE)); } static void spa_set_aux_vdevs(spa_aux_vdev_t *sav, nvlist_t **devs, int ndevs, const char *config) { int i; if (sav->sav_config != NULL) { nvlist_t **olddevs; uint_t oldndevs; nvlist_t **newdevs; /* * Generate new dev list by concatentating with the * current dev list. */ VERIFY(nvlist_lookup_nvlist_array(sav->sav_config, config, &olddevs, &oldndevs) == 0); newdevs = kmem_alloc(sizeof (void *) * (ndevs + oldndevs), KM_PUSHPAGE); for (i = 0; i < oldndevs; i++) VERIFY(nvlist_dup(olddevs[i], &newdevs[i], KM_PUSHPAGE) == 0); for (i = 0; i < ndevs; i++) VERIFY(nvlist_dup(devs[i], &newdevs[i + oldndevs], KM_PUSHPAGE) == 0); VERIFY(nvlist_remove(sav->sav_config, config, DATA_TYPE_NVLIST_ARRAY) == 0); VERIFY(nvlist_add_nvlist_array(sav->sav_config, config, newdevs, ndevs + oldndevs) == 0); for (i = 0; i < oldndevs + ndevs; i++) nvlist_free(newdevs[i]); kmem_free(newdevs, (oldndevs + ndevs) * sizeof (void *)); } else { /* * Generate a new dev list. */ VERIFY(nvlist_alloc(&sav->sav_config, NV_UNIQUE_NAME, KM_PUSHPAGE) == 0); VERIFY(nvlist_add_nvlist_array(sav->sav_config, config, devs, ndevs) == 0); } } /* * Stop and drop level 2 ARC devices */ void spa_l2cache_drop(spa_t *spa) { vdev_t *vd; int i; spa_aux_vdev_t *sav = &spa->spa_l2cache; for (i = 0; i < sav->sav_count; i++) { uint64_t pool; vd = sav->sav_vdevs[i]; ASSERT(vd != NULL); if (spa_l2cache_exists(vd->vdev_guid, &pool) && pool != 0ULL && l2arc_vdev_present(vd)) l2arc_remove_vdev(vd); } } /* * Pool Creation */ int spa_create(const char *pool, nvlist_t *nvroot, nvlist_t *props, nvlist_t *zplprops) { spa_t *spa; char *altroot = NULL; vdev_t *rvd; dsl_pool_t *dp; dmu_tx_t *tx; int error = 0; uint64_t txg = TXG_INITIAL; nvlist_t **spares, **l2cache; uint_t nspares, nl2cache; uint64_t version, obj; boolean_t has_features; nvpair_t *elem; int c; /* * If this pool already exists, return failure. */ mutex_enter(&spa_namespace_lock); if (spa_lookup(pool) != NULL) { mutex_exit(&spa_namespace_lock); return (SET_ERROR(EEXIST)); } /* * Allocate a new spa_t structure. */ (void) nvlist_lookup_string(props, zpool_prop_to_name(ZPOOL_PROP_ALTROOT), &altroot); spa = spa_add(pool, NULL, altroot); spa_activate(spa, spa_mode_global); if (props && (error = spa_prop_validate(spa, props))) { spa_deactivate(spa); spa_remove(spa); mutex_exit(&spa_namespace_lock); return (error); } has_features = B_FALSE; for (elem = nvlist_next_nvpair(props, NULL); elem != NULL; elem = nvlist_next_nvpair(props, elem)) { if (zpool_prop_feature(nvpair_name(elem))) has_features = B_TRUE; } if (has_features || nvlist_lookup_uint64(props, zpool_prop_to_name(ZPOOL_PROP_VERSION), &version) != 0) { version = SPA_VERSION; } ASSERT(SPA_VERSION_IS_SUPPORTED(version)); spa->spa_first_txg = txg; spa->spa_uberblock.ub_txg = txg - 1; spa->spa_uberblock.ub_version = version; spa->spa_ubsync = spa->spa_uberblock; /* * Create "The Godfather" zio to hold all async IOs */ spa->spa_async_zio_root = zio_root(spa, NULL, NULL, ZIO_FLAG_CANFAIL | ZIO_FLAG_SPECULATIVE | ZIO_FLAG_GODFATHER); /* * Create the root vdev. */ spa_config_enter(spa, SCL_ALL, FTAG, RW_WRITER); error = spa_config_parse(spa, &rvd, nvroot, NULL, 0, VDEV_ALLOC_ADD); ASSERT(error != 0 || rvd != NULL); ASSERT(error != 0 || spa->spa_root_vdev == rvd); if (error == 0 && !zfs_allocatable_devs(nvroot)) error = SET_ERROR(EINVAL); if (error == 0 && (error = vdev_create(rvd, txg, B_FALSE)) == 0 && (error = spa_validate_aux(spa, nvroot, txg, VDEV_ALLOC_ADD)) == 0) { for (c = 0; c < rvd->vdev_children; c++) { vdev_metaslab_set_size(rvd->vdev_child[c]); vdev_expand(rvd->vdev_child[c], txg); } } spa_config_exit(spa, SCL_ALL, FTAG); if (error != 0) { spa_unload(spa); spa_deactivate(spa); spa_remove(spa); mutex_exit(&spa_namespace_lock); return (error); } /* * Get the list of spares, if specified. */ if (nvlist_lookup_nvlist_array(nvroot, ZPOOL_CONFIG_SPARES, &spares, &nspares) == 0) { VERIFY(nvlist_alloc(&spa->spa_spares.sav_config, NV_UNIQUE_NAME, KM_PUSHPAGE) == 0); VERIFY(nvlist_add_nvlist_array(spa->spa_spares.sav_config, ZPOOL_CONFIG_SPARES, spares, nspares) == 0); spa_config_enter(spa, SCL_ALL, FTAG, RW_WRITER); spa_load_spares(spa); spa_config_exit(spa, SCL_ALL, FTAG); spa->spa_spares.sav_sync = B_TRUE; } /* * Get the list of level 2 cache devices, if specified. */ if (nvlist_lookup_nvlist_array(nvroot, ZPOOL_CONFIG_L2CACHE, &l2cache, &nl2cache) == 0) { VERIFY(nvlist_alloc(&spa->spa_l2cache.sav_config, NV_UNIQUE_NAME, KM_PUSHPAGE) == 0); VERIFY(nvlist_add_nvlist_array(spa->spa_l2cache.sav_config, ZPOOL_CONFIG_L2CACHE, l2cache, nl2cache) == 0); spa_config_enter(spa, SCL_ALL, FTAG, RW_WRITER); spa_load_l2cache(spa); spa_config_exit(spa, SCL_ALL, FTAG); spa->spa_l2cache.sav_sync = B_TRUE; } spa->spa_is_initializing = B_TRUE; spa->spa_dsl_pool = dp = dsl_pool_create(spa, zplprops, txg); spa->spa_meta_objset = dp->dp_meta_objset; spa->spa_is_initializing = B_FALSE; /* * Create DDTs (dedup tables). */ ddt_create(spa); spa_update_dspace(spa); tx = dmu_tx_create_assigned(dp, txg); /* * Create the pool config object. */ spa->spa_config_object = dmu_object_alloc(spa->spa_meta_objset, DMU_OT_PACKED_NVLIST, SPA_CONFIG_BLOCKSIZE, DMU_OT_PACKED_NVLIST_SIZE, sizeof (uint64_t), tx); if (zap_add(spa->spa_meta_objset, DMU_POOL_DIRECTORY_OBJECT, DMU_POOL_CONFIG, sizeof (uint64_t), 1, &spa->spa_config_object, tx) != 0) { cmn_err(CE_PANIC, "failed to add pool config"); } if (spa_version(spa) >= SPA_VERSION_FEATURES) spa_feature_create_zap_objects(spa, tx); if (zap_add(spa->spa_meta_objset, DMU_POOL_DIRECTORY_OBJECT, DMU_POOL_CREATION_VERSION, sizeof (uint64_t), 1, &version, tx) != 0) { cmn_err(CE_PANIC, "failed to add pool version"); } /* Newly created pools with the right version are always deflated. */ if (version >= SPA_VERSION_RAIDZ_DEFLATE) { spa->spa_deflate = TRUE; if (zap_add(spa->spa_meta_objset, DMU_POOL_DIRECTORY_OBJECT, DMU_POOL_DEFLATE, sizeof (uint64_t), 1, &spa->spa_deflate, tx) != 0) { cmn_err(CE_PANIC, "failed to add deflate"); } } /* * Create the deferred-free bpobj. Turn off compression * because sync-to-convergence takes longer if the blocksize * keeps changing. */ obj = bpobj_alloc(spa->spa_meta_objset, 1 << 14, tx); dmu_object_set_compress(spa->spa_meta_objset, obj, ZIO_COMPRESS_OFF, tx); if (zap_add(spa->spa_meta_objset, DMU_POOL_DIRECTORY_OBJECT, DMU_POOL_SYNC_BPOBJ, sizeof (uint64_t), 1, &obj, tx) != 0) { cmn_err(CE_PANIC, "failed to add bpobj"); } VERIFY3U(0, ==, bpobj_open(&spa->spa_deferred_bpobj, spa->spa_meta_objset, obj)); /* * Create the pool's history object. */ if (version >= SPA_VERSION_ZPOOL_HISTORY) spa_history_create_obj(spa, tx); /* * Set pool properties. */ spa->spa_bootfs = zpool_prop_default_numeric(ZPOOL_PROP_BOOTFS); spa->spa_delegation = zpool_prop_default_numeric(ZPOOL_PROP_DELEGATION); spa->spa_failmode = zpool_prop_default_numeric(ZPOOL_PROP_FAILUREMODE); spa->spa_autoexpand = zpool_prop_default_numeric(ZPOOL_PROP_AUTOEXPAND); if (props != NULL) { spa_configfile_set(spa, props, B_FALSE); spa_sync_props(props, tx); } dmu_tx_commit(tx); spa->spa_sync_on = B_TRUE; txg_sync_start(spa->spa_dsl_pool); /* * We explicitly wait for the first transaction to complete so that our * bean counters are appropriately updated. */ txg_wait_synced(spa->spa_dsl_pool, txg); spa_config_sync(spa, B_FALSE, B_TRUE); spa_history_log_version(spa, "create"); spa->spa_minref = refcount_count(&spa->spa_refcount); mutex_exit(&spa_namespace_lock); return (0); } #ifdef _KERNEL /* * Get the root pool information from the root disk, then import the root pool * during the system boot up time. */ extern int vdev_disk_read_rootlabel(char *, char *, nvlist_t **); static nvlist_t * spa_generate_rootconf(char *devpath, char *devid, uint64_t *guid) { nvlist_t *config; nvlist_t *nvtop, *nvroot; uint64_t pgid; if (vdev_disk_read_rootlabel(devpath, devid, &config) != 0) return (NULL); /* * Add this top-level vdev to the child array. */ VERIFY(nvlist_lookup_nvlist(config, ZPOOL_CONFIG_VDEV_TREE, &nvtop) == 0); VERIFY(nvlist_lookup_uint64(config, ZPOOL_CONFIG_POOL_GUID, &pgid) == 0); VERIFY(nvlist_lookup_uint64(config, ZPOOL_CONFIG_GUID, guid) == 0); /* * Put this pool's top-level vdevs into a root vdev. */ VERIFY(nvlist_alloc(&nvroot, NV_UNIQUE_NAME, KM_PUSHPAGE) == 0); VERIFY(nvlist_add_string(nvroot, ZPOOL_CONFIG_TYPE, VDEV_TYPE_ROOT) == 0); VERIFY(nvlist_add_uint64(nvroot, ZPOOL_CONFIG_ID, 0ULL) == 0); VERIFY(nvlist_add_uint64(nvroot, ZPOOL_CONFIG_GUID, pgid) == 0); VERIFY(nvlist_add_nvlist_array(nvroot, ZPOOL_CONFIG_CHILDREN, &nvtop, 1) == 0); /* * Replace the existing vdev_tree with the new root vdev in * this pool's configuration (remove the old, add the new). */ VERIFY(nvlist_add_nvlist(config, ZPOOL_CONFIG_VDEV_TREE, nvroot) == 0); nvlist_free(nvroot); return (config); } /* * Walk the vdev tree and see if we can find a device with "better" * configuration. A configuration is "better" if the label on that * device has a more recent txg. */ static void spa_alt_rootvdev(vdev_t *vd, vdev_t **avd, uint64_t *txg) { int c; for (c = 0; c < vd->vdev_children; c++) spa_alt_rootvdev(vd->vdev_child[c], avd, txg); if (vd->vdev_ops->vdev_op_leaf) { nvlist_t *label; uint64_t label_txg; if (vdev_disk_read_rootlabel(vd->vdev_physpath, vd->vdev_devid, &label) != 0) return; VERIFY(nvlist_lookup_uint64(label, ZPOOL_CONFIG_POOL_TXG, &label_txg) == 0); /* * Do we have a better boot device? */ if (label_txg > *txg) { *txg = label_txg; *avd = vd; } nvlist_free(label); } } /* * Import a root pool. * * For x86. devpath_list will consist of devid and/or physpath name of * the vdev (e.g. "id1,sd@SSEAGATE..." or "/pci@1f,0/ide@d/disk@0,0:a"). * The GRUB "findroot" command will return the vdev we should boot. * * For Sparc, devpath_list consists the physpath name of the booting device * no matter the rootpool is a single device pool or a mirrored pool. * e.g. * "/pci@1f,0/ide@d/disk@0,0:a" */ int spa_import_rootpool(char *devpath, char *devid) { spa_t *spa; vdev_t *rvd, *bvd, *avd = NULL; nvlist_t *config, *nvtop; uint64_t guid, txg; char *pname; int error; /* * Read the label from the boot device and generate a configuration. */ config = spa_generate_rootconf(devpath, devid, &guid); #if defined(_OBP) && defined(_KERNEL) if (config == NULL) { if (strstr(devpath, "/iscsi/ssd") != NULL) { /* iscsi boot */ get_iscsi_bootpath_phy(devpath); config = spa_generate_rootconf(devpath, devid, &guid); } } #endif if (config == NULL) { cmn_err(CE_NOTE, "Cannot read the pool label from '%s'", devpath); return (SET_ERROR(EIO)); } VERIFY(nvlist_lookup_string(config, ZPOOL_CONFIG_POOL_NAME, &pname) == 0); VERIFY(nvlist_lookup_uint64(config, ZPOOL_CONFIG_POOL_TXG, &txg) == 0); mutex_enter(&spa_namespace_lock); if ((spa = spa_lookup(pname)) != NULL) { /* * Remove the existing root pool from the namespace so that we * can replace it with the correct config we just read in. */ spa_remove(spa); } spa = spa_add(pname, config, NULL); spa->spa_is_root = B_TRUE; spa->spa_import_flags = ZFS_IMPORT_VERBATIM; /* * Build up a vdev tree based on the boot device's label config. */ VERIFY(nvlist_lookup_nvlist(config, ZPOOL_CONFIG_VDEV_TREE, &nvtop) == 0); spa_config_enter(spa, SCL_ALL, FTAG, RW_WRITER); error = spa_config_parse(spa, &rvd, nvtop, NULL, 0, VDEV_ALLOC_ROOTPOOL); spa_config_exit(spa, SCL_ALL, FTAG); if (error) { mutex_exit(&spa_namespace_lock); nvlist_free(config); cmn_err(CE_NOTE, "Can not parse the config for pool '%s'", pname); return (error); } /* * Get the boot vdev. */ if ((bvd = vdev_lookup_by_guid(rvd, guid)) == NULL) { cmn_err(CE_NOTE, "Can not find the boot vdev for guid %llu", (u_longlong_t)guid); error = SET_ERROR(ENOENT); goto out; } /* * Determine if there is a better boot device. */ avd = bvd; spa_alt_rootvdev(rvd, &avd, &txg); if (avd != bvd) { cmn_err(CE_NOTE, "The boot device is 'degraded'. Please " "try booting from '%s'", avd->vdev_path); error = SET_ERROR(EINVAL); goto out; } /* * If the boot device is part of a spare vdev then ensure that * we're booting off the active spare. */ if (bvd->vdev_parent->vdev_ops == &vdev_spare_ops && !bvd->vdev_isspare) { cmn_err(CE_NOTE, "The boot device is currently spared. Please " "try booting from '%s'", bvd->vdev_parent-> vdev_child[bvd->vdev_parent->vdev_children - 1]->vdev_path); error = SET_ERROR(EINVAL); goto out; } error = 0; out: spa_config_enter(spa, SCL_ALL, FTAG, RW_WRITER); vdev_free(rvd); spa_config_exit(spa, SCL_ALL, FTAG); mutex_exit(&spa_namespace_lock); nvlist_free(config); return (error); } #endif /* * Import a non-root pool into the system. */ int spa_import(char *pool, nvlist_t *config, nvlist_t *props, uint64_t flags) { spa_t *spa; char *altroot = NULL; spa_load_state_t state = SPA_LOAD_IMPORT; zpool_rewind_policy_t policy; uint64_t mode = spa_mode_global; uint64_t readonly = B_FALSE; int error; nvlist_t *nvroot; nvlist_t **spares, **l2cache; uint_t nspares, nl2cache; /* * If a pool with this name exists, return failure. */ mutex_enter(&spa_namespace_lock); if (spa_lookup(pool) != NULL) { mutex_exit(&spa_namespace_lock); return (SET_ERROR(EEXIST)); } /* * Create and initialize the spa structure. */ (void) nvlist_lookup_string(props, zpool_prop_to_name(ZPOOL_PROP_ALTROOT), &altroot); (void) nvlist_lookup_uint64(props, zpool_prop_to_name(ZPOOL_PROP_READONLY), &readonly); if (readonly) mode = FREAD; spa = spa_add(pool, config, altroot); spa->spa_import_flags = flags; /* * Verbatim import - Take a pool and insert it into the namespace * as if it had been loaded at boot. */ if (spa->spa_import_flags & ZFS_IMPORT_VERBATIM) { if (props != NULL) spa_configfile_set(spa, props, B_FALSE); spa_config_sync(spa, B_FALSE, B_TRUE); mutex_exit(&spa_namespace_lock); return (0); } spa_activate(spa, mode); /* * Don't start async tasks until we know everything is healthy. */ spa_async_suspend(spa); zpool_get_rewind_policy(config, &policy); if (policy.zrp_request & ZPOOL_DO_REWIND) state = SPA_LOAD_RECOVER; /* * Pass off the heavy lifting to spa_load(). Pass TRUE for mosconfig * because the user-supplied config is actually the one to trust when * doing an import. */ if (state != SPA_LOAD_RECOVER) spa->spa_last_ubsync_txg = spa->spa_load_txg = 0; error = spa_load_best(spa, state, B_TRUE, policy.zrp_txg, policy.zrp_request); /* * Propagate anything learned while loading the pool and pass it * back to caller (i.e. rewind info, missing devices, etc). */ VERIFY(nvlist_add_nvlist(config, ZPOOL_CONFIG_LOAD_INFO, spa->spa_load_info) == 0); spa_config_enter(spa, SCL_ALL, FTAG, RW_WRITER); /* * Toss any existing sparelist, as it doesn't have any validity * anymore, and conflicts with spa_has_spare(). */ if (spa->spa_spares.sav_config) { nvlist_free(spa->spa_spares.sav_config); spa->spa_spares.sav_config = NULL; spa_load_spares(spa); } if (spa->spa_l2cache.sav_config) { nvlist_free(spa->spa_l2cache.sav_config); spa->spa_l2cache.sav_config = NULL; spa_load_l2cache(spa); } VERIFY(nvlist_lookup_nvlist(config, ZPOOL_CONFIG_VDEV_TREE, &nvroot) == 0); if (error == 0) error = spa_validate_aux(spa, nvroot, -1ULL, VDEV_ALLOC_SPARE); if (error == 0) error = spa_validate_aux(spa, nvroot, -1ULL, VDEV_ALLOC_L2CACHE); spa_config_exit(spa, SCL_ALL, FTAG); if (props != NULL) spa_configfile_set(spa, props, B_FALSE); if (error != 0 || (props && spa_writeable(spa) && (error = spa_prop_set(spa, props)))) { spa_unload(spa); spa_deactivate(spa); spa_remove(spa); mutex_exit(&spa_namespace_lock); return (error); } spa_async_resume(spa); /* * Override any spares and level 2 cache devices as specified by * the user, as these may have correct device names/devids, etc. */ if (nvlist_lookup_nvlist_array(nvroot, ZPOOL_CONFIG_SPARES, &spares, &nspares) == 0) { if (spa->spa_spares.sav_config) VERIFY(nvlist_remove(spa->spa_spares.sav_config, ZPOOL_CONFIG_SPARES, DATA_TYPE_NVLIST_ARRAY) == 0); else VERIFY(nvlist_alloc(&spa->spa_spares.sav_config, NV_UNIQUE_NAME, KM_PUSHPAGE) == 0); VERIFY(nvlist_add_nvlist_array(spa->spa_spares.sav_config, ZPOOL_CONFIG_SPARES, spares, nspares) == 0); spa_config_enter(spa, SCL_ALL, FTAG, RW_WRITER); spa_load_spares(spa); spa_config_exit(spa, SCL_ALL, FTAG); spa->spa_spares.sav_sync = B_TRUE; } if (nvlist_lookup_nvlist_array(nvroot, ZPOOL_CONFIG_L2CACHE, &l2cache, &nl2cache) == 0) { if (spa->spa_l2cache.sav_config) VERIFY(nvlist_remove(spa->spa_l2cache.sav_config, ZPOOL_CONFIG_L2CACHE, DATA_TYPE_NVLIST_ARRAY) == 0); else VERIFY(nvlist_alloc(&spa->spa_l2cache.sav_config, NV_UNIQUE_NAME, KM_PUSHPAGE) == 0); VERIFY(nvlist_add_nvlist_array(spa->spa_l2cache.sav_config, ZPOOL_CONFIG_L2CACHE, l2cache, nl2cache) == 0); spa_config_enter(spa, SCL_ALL, FTAG, RW_WRITER); spa_load_l2cache(spa); spa_config_exit(spa, SCL_ALL, FTAG); spa->spa_l2cache.sav_sync = B_TRUE; } /* * Check for any removed devices. */ if (spa->spa_autoreplace) { spa_aux_check_removed(&spa->spa_spares); spa_aux_check_removed(&spa->spa_l2cache); } if (spa_writeable(spa)) { /* * Update the config cache to include the newly-imported pool. */ spa_config_update(spa, SPA_CONFIG_UPDATE_POOL); } /* * It's possible that the pool was expanded while it was exported. * We kick off an async task to handle this for us. */ spa_async_request(spa, SPA_ASYNC_AUTOEXPAND); mutex_exit(&spa_namespace_lock); spa_history_log_version(spa, "import"); #ifdef _KERNEL zvol_create_minors(pool); #endif return (0); } nvlist_t * spa_tryimport(nvlist_t *tryconfig) { nvlist_t *config = NULL; char *poolname; spa_t *spa; uint64_t state; int error; if (nvlist_lookup_string(tryconfig, ZPOOL_CONFIG_POOL_NAME, &poolname)) return (NULL); if (nvlist_lookup_uint64(tryconfig, ZPOOL_CONFIG_POOL_STATE, &state)) return (NULL); /* * Create and initialize the spa structure. */ mutex_enter(&spa_namespace_lock); spa = spa_add(TRYIMPORT_NAME, tryconfig, NULL); spa_activate(spa, FREAD); /* * Pass off the heavy lifting to spa_load(). * Pass TRUE for mosconfig because the user-supplied config * is actually the one to trust when doing an import. */ error = spa_load(spa, SPA_LOAD_TRYIMPORT, SPA_IMPORT_EXISTING, B_TRUE); /* * If 'tryconfig' was at least parsable, return the current config. */ if (spa->spa_root_vdev != NULL) { config = spa_config_generate(spa, NULL, -1ULL, B_TRUE); VERIFY(nvlist_add_string(config, ZPOOL_CONFIG_POOL_NAME, poolname) == 0); VERIFY(nvlist_add_uint64(config, ZPOOL_CONFIG_POOL_STATE, state) == 0); VERIFY(nvlist_add_uint64(config, ZPOOL_CONFIG_TIMESTAMP, spa->spa_uberblock.ub_timestamp) == 0); VERIFY(nvlist_add_nvlist(config, ZPOOL_CONFIG_LOAD_INFO, spa->spa_load_info) == 0); VERIFY(nvlist_add_uint64(config, ZPOOL_CONFIG_ERRATA, spa->spa_errata) == 0); /* * If the bootfs property exists on this pool then we * copy it out so that external consumers can tell which * pools are bootable. */ if ((!error || error == EEXIST) && spa->spa_bootfs) { char *tmpname = kmem_alloc(MAXPATHLEN, KM_PUSHPAGE); /* * We have to play games with the name since the * pool was opened as TRYIMPORT_NAME. */ if (dsl_dsobj_to_dsname(spa_name(spa), spa->spa_bootfs, tmpname) == 0) { char *cp; char *dsname; dsname = kmem_alloc(MAXPATHLEN, KM_PUSHPAGE); cp = strchr(tmpname, '/'); if (cp == NULL) { (void) strlcpy(dsname, tmpname, MAXPATHLEN); } else { (void) snprintf(dsname, MAXPATHLEN, "%s/%s", poolname, ++cp); } VERIFY(nvlist_add_string(config, ZPOOL_CONFIG_BOOTFS, dsname) == 0); kmem_free(dsname, MAXPATHLEN); } kmem_free(tmpname, MAXPATHLEN); } /* * Add the list of hot spares and level 2 cache devices. */ spa_config_enter(spa, SCL_CONFIG, FTAG, RW_READER); spa_add_spares(spa, config); spa_add_l2cache(spa, config); spa_config_exit(spa, SCL_CONFIG, FTAG); } spa_unload(spa); spa_deactivate(spa); spa_remove(spa); mutex_exit(&spa_namespace_lock); return (config); } /* * Pool export/destroy * * The act of destroying or exporting a pool is very simple. We make sure there * is no more pending I/O and any references to the pool are gone. Then, we * update the pool state and sync all the labels to disk, removing the * configuration from the cache afterwards. If the 'hardforce' flag is set, then * we don't sync the labels or remove the configuration cache. */ static int spa_export_common(char *pool, int new_state, nvlist_t **oldconfig, boolean_t force, boolean_t hardforce) { spa_t *spa; if (oldconfig) *oldconfig = NULL; if (!(spa_mode_global & FWRITE)) return (SET_ERROR(EROFS)); mutex_enter(&spa_namespace_lock); if ((spa = spa_lookup(pool)) == NULL) { mutex_exit(&spa_namespace_lock); return (SET_ERROR(ENOENT)); } /* * Put a hold on the pool, drop the namespace lock, stop async tasks, * reacquire the namespace lock, and see if we can export. */ spa_open_ref(spa, FTAG); mutex_exit(&spa_namespace_lock); spa_async_suspend(spa); mutex_enter(&spa_namespace_lock); spa_close(spa, FTAG); /* * The pool will be in core if it's openable, * in which case we can modify its state. */ if (spa->spa_state != POOL_STATE_UNINITIALIZED && spa->spa_sync_on) { /* * Objsets may be open only because they're dirty, so we * have to force it to sync before checking spa_refcnt. */ txg_wait_synced(spa->spa_dsl_pool, 0); /* * A pool cannot be exported or destroyed if there are active * references. If we are resetting a pool, allow references by * fault injection handlers. */ if (!spa_refcount_zero(spa) || (spa->spa_inject_ref != 0 && new_state != POOL_STATE_UNINITIALIZED)) { spa_async_resume(spa); mutex_exit(&spa_namespace_lock); return (SET_ERROR(EBUSY)); } /* * A pool cannot be exported if it has an active shared spare. * This is to prevent other pools stealing the active spare * from an exported pool. At user's own will, such pool can * be forcedly exported. */ if (!force && new_state == POOL_STATE_EXPORTED && spa_has_active_shared_spare(spa)) { spa_async_resume(spa); mutex_exit(&spa_namespace_lock); return (SET_ERROR(EXDEV)); } /* * We want this to be reflected on every label, * so mark them all dirty. spa_unload() will do the * final sync that pushes these changes out. */ if (new_state != POOL_STATE_UNINITIALIZED && !hardforce) { spa_config_enter(spa, SCL_ALL, FTAG, RW_WRITER); spa->spa_state = new_state; spa->spa_final_txg = spa_last_synced_txg(spa) + TXG_DEFER_SIZE + 1; vdev_config_dirty(spa->spa_root_vdev); spa_config_exit(spa, SCL_ALL, FTAG); } } spa_event_notify(spa, NULL, FM_EREPORT_ZFS_POOL_DESTROY); if (spa->spa_state != POOL_STATE_UNINITIALIZED) { spa_unload(spa); spa_deactivate(spa); } if (oldconfig && spa->spa_config) VERIFY(nvlist_dup(spa->spa_config, oldconfig, 0) == 0); if (new_state != POOL_STATE_UNINITIALIZED) { if (!hardforce) spa_config_sync(spa, B_TRUE, B_TRUE); spa_remove(spa); } mutex_exit(&spa_namespace_lock); return (0); } /* * Destroy a storage pool. */ int spa_destroy(char *pool) { return (spa_export_common(pool, POOL_STATE_DESTROYED, NULL, B_FALSE, B_FALSE)); } /* * Export a storage pool. */ int spa_export(char *pool, nvlist_t **oldconfig, boolean_t force, boolean_t hardforce) { return (spa_export_common(pool, POOL_STATE_EXPORTED, oldconfig, force, hardforce)); } /* * Similar to spa_export(), this unloads the spa_t without actually removing it * from the namespace in any way. */ int spa_reset(char *pool) { return (spa_export_common(pool, POOL_STATE_UNINITIALIZED, NULL, B_FALSE, B_FALSE)); } /* * ========================================================================== * Device manipulation * ========================================================================== */ /* * Add a device to a storage pool. */ int spa_vdev_add(spa_t *spa, nvlist_t *nvroot) { uint64_t txg, id; int error; vdev_t *rvd = spa->spa_root_vdev; vdev_t *vd, *tvd; nvlist_t **spares, **l2cache; uint_t nspares, nl2cache; int c; ASSERT(spa_writeable(spa)); txg = spa_vdev_enter(spa); if ((error = spa_config_parse(spa, &vd, nvroot, NULL, 0, VDEV_ALLOC_ADD)) != 0) return (spa_vdev_exit(spa, NULL, txg, error)); spa->spa_pending_vdev = vd; /* spa_vdev_exit() will clear this */ if (nvlist_lookup_nvlist_array(nvroot, ZPOOL_CONFIG_SPARES, &spares, &nspares) != 0) nspares = 0; if (nvlist_lookup_nvlist_array(nvroot, ZPOOL_CONFIG_L2CACHE, &l2cache, &nl2cache) != 0) nl2cache = 0; if (vd->vdev_children == 0 && nspares == 0 && nl2cache == 0) return (spa_vdev_exit(spa, vd, txg, EINVAL)); if (vd->vdev_children != 0 && (error = vdev_create(vd, txg, B_FALSE)) != 0) return (spa_vdev_exit(spa, vd, txg, error)); /* * We must validate the spares and l2cache devices after checking the * children. Otherwise, vdev_inuse() will blindly overwrite the spare. */ if ((error = spa_validate_aux(spa, nvroot, txg, VDEV_ALLOC_ADD)) != 0) return (spa_vdev_exit(spa, vd, txg, error)); /* * Transfer each new top-level vdev from vd to rvd. */ for (c = 0; c < vd->vdev_children; c++) { /* * Set the vdev id to the first hole, if one exists. */ for (id = 0; id < rvd->vdev_children; id++) { if (rvd->vdev_child[id]->vdev_ishole) { vdev_free(rvd->vdev_child[id]); break; } } tvd = vd->vdev_child[c]; vdev_remove_child(vd, tvd); tvd->vdev_id = id; vdev_add_child(rvd, tvd); vdev_config_dirty(tvd); } if (nspares != 0) { spa_set_aux_vdevs(&spa->spa_spares, spares, nspares, ZPOOL_CONFIG_SPARES); spa_load_spares(spa); spa->spa_spares.sav_sync = B_TRUE; } if (nl2cache != 0) { spa_set_aux_vdevs(&spa->spa_l2cache, l2cache, nl2cache, ZPOOL_CONFIG_L2CACHE); spa_load_l2cache(spa); spa->spa_l2cache.sav_sync = B_TRUE; } /* * We have to be careful when adding new vdevs to an existing pool. * If other threads start allocating from these vdevs before we * sync the config cache, and we lose power, then upon reboot we may * fail to open the pool because there are DVAs that the config cache * can't translate. Therefore, we first add the vdevs without * initializing metaslabs; sync the config cache (via spa_vdev_exit()); * and then let spa_config_update() initialize the new metaslabs. * * spa_load() checks for added-but-not-initialized vdevs, so that * if we lose power at any point in this sequence, the remaining * steps will be completed the next time we load the pool. */ (void) spa_vdev_exit(spa, vd, txg, 0); mutex_enter(&spa_namespace_lock); spa_config_update(spa, SPA_CONFIG_UPDATE_POOL); mutex_exit(&spa_namespace_lock); return (0); } /* * Attach a device to a mirror. The arguments are the path to any device * in the mirror, and the nvroot for the new device. If the path specifies * a device that is not mirrored, we automatically insert the mirror vdev. * * If 'replacing' is specified, the new device is intended to replace the * existing device; in this case the two devices are made into their own * mirror using the 'replacing' vdev, which is functionally identical to * the mirror vdev (it actually reuses all the same ops) but has a few * extra rules: you can't attach to it after it's been created, and upon * completion of resilvering, the first disk (the one being replaced) * is automatically detached. */ int spa_vdev_attach(spa_t *spa, uint64_t guid, nvlist_t *nvroot, int replacing) { uint64_t txg, dtl_max_txg; vdev_t *oldvd, *newvd, *newrootvd, *pvd, *tvd; vdev_ops_t *pvops; char *oldvdpath, *newvdpath; int newvd_isspare; int error; ASSERTV(vdev_t *rvd = spa->spa_root_vdev); ASSERT(spa_writeable(spa)); txg = spa_vdev_enter(spa); oldvd = spa_lookup_by_guid(spa, guid, B_FALSE); if (oldvd == NULL) return (spa_vdev_exit(spa, NULL, txg, ENODEV)); if (!oldvd->vdev_ops->vdev_op_leaf) return (spa_vdev_exit(spa, NULL, txg, ENOTSUP)); pvd = oldvd->vdev_parent; if ((error = spa_config_parse(spa, &newrootvd, nvroot, NULL, 0, VDEV_ALLOC_ATTACH)) != 0) return (spa_vdev_exit(spa, NULL, txg, EINVAL)); if (newrootvd->vdev_children != 1) return (spa_vdev_exit(spa, newrootvd, txg, EINVAL)); newvd = newrootvd->vdev_child[0]; if (!newvd->vdev_ops->vdev_op_leaf) return (spa_vdev_exit(spa, newrootvd, txg, EINVAL)); if ((error = vdev_create(newrootvd, txg, replacing)) != 0) return (spa_vdev_exit(spa, newrootvd, txg, error)); /* * Spares can't replace logs */ if (oldvd->vdev_top->vdev_islog && newvd->vdev_isspare) return (spa_vdev_exit(spa, newrootvd, txg, ENOTSUP)); if (!replacing) { /* * For attach, the only allowable parent is a mirror or the root * vdev. */ if (pvd->vdev_ops != &vdev_mirror_ops && pvd->vdev_ops != &vdev_root_ops) return (spa_vdev_exit(spa, newrootvd, txg, ENOTSUP)); pvops = &vdev_mirror_ops; } else { /* * Active hot spares can only be replaced by inactive hot * spares. */ if (pvd->vdev_ops == &vdev_spare_ops && oldvd->vdev_isspare && !spa_has_spare(spa, newvd->vdev_guid)) return (spa_vdev_exit(spa, newrootvd, txg, ENOTSUP)); /* * If the source is a hot spare, and the parent isn't already a * spare, then we want to create a new hot spare. Otherwise, we * want to create a replacing vdev. The user is not allowed to * attach to a spared vdev child unless the 'isspare' state is * the same (spare replaces spare, non-spare replaces * non-spare). */ if (pvd->vdev_ops == &vdev_replacing_ops && spa_version(spa) < SPA_VERSION_MULTI_REPLACE) { return (spa_vdev_exit(spa, newrootvd, txg, ENOTSUP)); } else if (pvd->vdev_ops == &vdev_spare_ops && newvd->vdev_isspare != oldvd->vdev_isspare) { return (spa_vdev_exit(spa, newrootvd, txg, ENOTSUP)); } if (newvd->vdev_isspare) pvops = &vdev_spare_ops; else pvops = &vdev_replacing_ops; } /* * Make sure the new device is big enough. */ if (newvd->vdev_asize < vdev_get_min_asize(oldvd)) return (spa_vdev_exit(spa, newrootvd, txg, EOVERFLOW)); /* * The new device cannot have a higher alignment requirement * than the top-level vdev. */ if (newvd->vdev_ashift > oldvd->vdev_top->vdev_ashift) return (spa_vdev_exit(spa, newrootvd, txg, EDOM)); /* * If this is an in-place replacement, update oldvd's path and devid * to make it distinguishable from newvd, and unopenable from now on. */ if (strcmp(oldvd->vdev_path, newvd->vdev_path) == 0) { spa_strfree(oldvd->vdev_path); oldvd->vdev_path = kmem_alloc(strlen(newvd->vdev_path) + 5, KM_PUSHPAGE); (void) sprintf(oldvd->vdev_path, "%s/%s", newvd->vdev_path, "old"); if (oldvd->vdev_devid != NULL) { spa_strfree(oldvd->vdev_devid); oldvd->vdev_devid = NULL; } } /* mark the device being resilvered */ newvd->vdev_resilver_txg = txg; /* * If the parent is not a mirror, or if we're replacing, insert the new * mirror/replacing/spare vdev above oldvd. */ if (pvd->vdev_ops != pvops) pvd = vdev_add_parent(oldvd, pvops); ASSERT(pvd->vdev_top->vdev_parent == rvd); ASSERT(pvd->vdev_ops == pvops); ASSERT(oldvd->vdev_parent == pvd); /* * Extract the new device from its root and add it to pvd. */ vdev_remove_child(newrootvd, newvd); newvd->vdev_id = pvd->vdev_children; newvd->vdev_crtxg = oldvd->vdev_crtxg; vdev_add_child(pvd, newvd); tvd = newvd->vdev_top; ASSERT(pvd->vdev_top == tvd); ASSERT(tvd->vdev_parent == rvd); vdev_config_dirty(tvd); /* * Set newvd's DTL to [TXG_INITIAL, dtl_max_txg) so that we account * for any dmu_sync-ed blocks. It will propagate upward when * spa_vdev_exit() calls vdev_dtl_reassess(). */ dtl_max_txg = txg + TXG_CONCURRENT_STATES; vdev_dtl_dirty(newvd, DTL_MISSING, TXG_INITIAL, dtl_max_txg - TXG_INITIAL); if (newvd->vdev_isspare) { spa_spare_activate(newvd); spa_event_notify(spa, newvd, FM_EREPORT_ZFS_DEVICE_SPARE); } oldvdpath = spa_strdup(oldvd->vdev_path); newvdpath = spa_strdup(newvd->vdev_path); newvd_isspare = newvd->vdev_isspare; /* * Mark newvd's DTL dirty in this txg. */ vdev_dirty(tvd, VDD_DTL, newvd, txg); /* * Schedule the resilver to restart in the future. We do this to * ensure that dmu_sync-ed blocks have been stitched into the * respective datasets. */ dsl_resilver_restart(spa->spa_dsl_pool, dtl_max_txg); /* * Commit the config */ (void) spa_vdev_exit(spa, newrootvd, dtl_max_txg, 0); spa_history_log_internal(spa, "vdev attach", NULL, "%s vdev=%s %s vdev=%s", replacing && newvd_isspare ? "spare in" : replacing ? "replace" : "attach", newvdpath, replacing ? "for" : "to", oldvdpath); spa_strfree(oldvdpath); spa_strfree(newvdpath); if (spa->spa_bootfs) spa_event_notify(spa, newvd, FM_EREPORT_ZFS_BOOTFS_VDEV_ATTACH); return (0); } /* * Detach a device from a mirror or replacing vdev. * * If 'replace_done' is specified, only detach if the parent * is a replacing vdev. */ int spa_vdev_detach(spa_t *spa, uint64_t guid, uint64_t pguid, int replace_done) { uint64_t txg; int error; vdev_t *vd, *pvd, *cvd, *tvd; boolean_t unspare = B_FALSE; uint64_t unspare_guid = 0; char *vdpath; int c, t; ASSERTV(vdev_t *rvd = spa->spa_root_vdev); ASSERT(spa_writeable(spa)); txg = spa_vdev_enter(spa); vd = spa_lookup_by_guid(spa, guid, B_FALSE); if (vd == NULL) return (spa_vdev_exit(spa, NULL, txg, ENODEV)); if (!vd->vdev_ops->vdev_op_leaf) return (spa_vdev_exit(spa, NULL, txg, ENOTSUP)); pvd = vd->vdev_parent; /* * If the parent/child relationship is not as expected, don't do it. * Consider M(A,R(B,C)) -- that is, a mirror of A with a replacing * vdev that's replacing B with C. The user's intent in replacing * is to go from M(A,B) to M(A,C). If the user decides to cancel * the replace by detaching C, the expected behavior is to end up * M(A,B). But suppose that right after deciding to detach C, * the replacement of B completes. We would have M(A,C), and then * ask to detach C, which would leave us with just A -- not what * the user wanted. To prevent this, we make sure that the * parent/child relationship hasn't changed -- in this example, * that C's parent is still the replacing vdev R. */ if (pvd->vdev_guid != pguid && pguid != 0) return (spa_vdev_exit(spa, NULL, txg, EBUSY)); /* * Only 'replacing' or 'spare' vdevs can be replaced. */ if (replace_done && pvd->vdev_ops != &vdev_replacing_ops && pvd->vdev_ops != &vdev_spare_ops) return (spa_vdev_exit(spa, NULL, txg, ENOTSUP)); ASSERT(pvd->vdev_ops != &vdev_spare_ops || spa_version(spa) >= SPA_VERSION_SPARES); /* * Only mirror, replacing, and spare vdevs support detach. */ if (pvd->vdev_ops != &vdev_replacing_ops && pvd->vdev_ops != &vdev_mirror_ops && pvd->vdev_ops != &vdev_spare_ops) return (spa_vdev_exit(spa, NULL, txg, ENOTSUP)); /* * If this device has the only valid copy of some data, * we cannot safely detach it. */ if (vdev_dtl_required(vd)) return (spa_vdev_exit(spa, NULL, txg, EBUSY)); ASSERT(pvd->vdev_children >= 2); /* * If we are detaching the second disk from a replacing vdev, then * check to see if we changed the original vdev's path to have "/old" * at the end in spa_vdev_attach(). If so, undo that change now. */ if (pvd->vdev_ops == &vdev_replacing_ops && vd->vdev_id > 0 && vd->vdev_path != NULL) { size_t len = strlen(vd->vdev_path); for (c = 0; c < pvd->vdev_children; c++) { cvd = pvd->vdev_child[c]; if (cvd == vd || cvd->vdev_path == NULL) continue; if (strncmp(cvd->vdev_path, vd->vdev_path, len) == 0 && strcmp(cvd->vdev_path + len, "/old") == 0) { spa_strfree(cvd->vdev_path); cvd->vdev_path = spa_strdup(vd->vdev_path); break; } } } /* * If we are detaching the original disk from a spare, then it implies * that the spare should become a real disk, and be removed from the * active spare list for the pool. */ if (pvd->vdev_ops == &vdev_spare_ops && vd->vdev_id == 0 && pvd->vdev_child[pvd->vdev_children - 1]->vdev_isspare) unspare = B_TRUE; /* * Erase the disk labels so the disk can be used for other things. * This must be done after all other error cases are handled, * but before we disembowel vd (so we can still do I/O to it). * But if we can't do it, don't treat the error as fatal -- * it may be that the unwritability of the disk is the reason * it's being detached! */ error = vdev_label_init(vd, 0, VDEV_LABEL_REMOVE); /* * Remove vd from its parent and compact the parent's children. */ vdev_remove_child(pvd, vd); vdev_compact_children(pvd); /* * Remember one of the remaining children so we can get tvd below. */ cvd = pvd->vdev_child[pvd->vdev_children - 1]; /* * If we need to remove the remaining child from the list of hot spares, * do it now, marking the vdev as no longer a spare in the process. * We must do this before vdev_remove_parent(), because that can * change the GUID if it creates a new toplevel GUID. For a similar * reason, we must remove the spare now, in the same txg as the detach; * otherwise someone could attach a new sibling, change the GUID, and * the subsequent attempt to spa_vdev_remove(unspare_guid) would fail. */ if (unspare) { ASSERT(cvd->vdev_isspare); spa_spare_remove(cvd); unspare_guid = cvd->vdev_guid; (void) spa_vdev_remove(spa, unspare_guid, B_TRUE); cvd->vdev_unspare = B_TRUE; } /* * If the parent mirror/replacing vdev only has one child, * the parent is no longer needed. Remove it from the tree. */ if (pvd->vdev_children == 1) { if (pvd->vdev_ops == &vdev_spare_ops) cvd->vdev_unspare = B_FALSE; vdev_remove_parent(cvd); } /* * We don't set tvd until now because the parent we just removed * may have been the previous top-level vdev. */ tvd = cvd->vdev_top; ASSERT(tvd->vdev_parent == rvd); /* * Reevaluate the parent vdev state. */ vdev_propagate_state(cvd); /* * If the 'autoexpand' property is set on the pool then automatically * try to expand the size of the pool. For example if the device we * just detached was smaller than the others, it may be possible to * add metaslabs (i.e. grow the pool). We need to reopen the vdev * first so that we can obtain the updated sizes of the leaf vdevs. */ if (spa->spa_autoexpand) { vdev_reopen(tvd); vdev_expand(tvd, txg); } vdev_config_dirty(tvd); /* * Mark vd's DTL as dirty in this txg. vdev_dtl_sync() will see that * vd->vdev_detached is set and free vd's DTL object in syncing context. * But first make sure we're not on any *other* txg's DTL list, to * prevent vd from being accessed after it's freed. */ vdpath = spa_strdup(vd->vdev_path); for (t = 0; t < TXG_SIZE; t++) (void) txg_list_remove_this(&tvd->vdev_dtl_list, vd, t); vd->vdev_detached = B_TRUE; vdev_dirty(tvd, VDD_DTL, vd, txg); spa_event_notify(spa, vd, FM_EREPORT_ZFS_DEVICE_REMOVE); /* hang on to the spa before we release the lock */ spa_open_ref(spa, FTAG); error = spa_vdev_exit(spa, vd, txg, 0); spa_history_log_internal(spa, "detach", NULL, "vdev=%s", vdpath); spa_strfree(vdpath); /* * If this was the removal of the original device in a hot spare vdev, * then we want to go through and remove the device from the hot spare * list of every other pool. */ if (unspare) { spa_t *altspa = NULL; mutex_enter(&spa_namespace_lock); while ((altspa = spa_next(altspa)) != NULL) { if (altspa->spa_state != POOL_STATE_ACTIVE || altspa == spa) continue; spa_open_ref(altspa, FTAG); mutex_exit(&spa_namespace_lock); (void) spa_vdev_remove(altspa, unspare_guid, B_TRUE); mutex_enter(&spa_namespace_lock); spa_close(altspa, FTAG); } mutex_exit(&spa_namespace_lock); /* search the rest of the vdevs for spares to remove */ spa_vdev_resilver_done(spa); } /* all done with the spa; OK to release */ mutex_enter(&spa_namespace_lock); spa_close(spa, FTAG); mutex_exit(&spa_namespace_lock); return (error); } /* * Split a set of devices from their mirrors, and create a new pool from them. */ int spa_vdev_split_mirror(spa_t *spa, char *newname, nvlist_t *config, nvlist_t *props, boolean_t exp) { int error = 0; uint64_t txg, *glist; spa_t *newspa; uint_t c, children, lastlog; nvlist_t **child, *nvl, *tmp; dmu_tx_t *tx; char *altroot = NULL; vdev_t *rvd, **vml = NULL; /* vdev modify list */ boolean_t activate_slog; ASSERT(spa_writeable(spa)); txg = spa_vdev_enter(spa); /* clear the log and flush everything up to now */ activate_slog = spa_passivate_log(spa); (void) spa_vdev_config_exit(spa, NULL, txg, 0, FTAG); error = spa_offline_log(spa); txg = spa_vdev_config_enter(spa); if (activate_slog) spa_activate_log(spa); if (error != 0) return (spa_vdev_exit(spa, NULL, txg, error)); /* check new spa name before going any further */ if (spa_lookup(newname) != NULL) return (spa_vdev_exit(spa, NULL, txg, EEXIST)); /* * scan through all the children to ensure they're all mirrors */ if (nvlist_lookup_nvlist(config, ZPOOL_CONFIG_VDEV_TREE, &nvl) != 0 || nvlist_lookup_nvlist_array(nvl, ZPOOL_CONFIG_CHILDREN, &child, &children) != 0) return (spa_vdev_exit(spa, NULL, txg, EINVAL)); /* first, check to ensure we've got the right child count */ rvd = spa->spa_root_vdev; lastlog = 0; for (c = 0; c < rvd->vdev_children; c++) { vdev_t *vd = rvd->vdev_child[c]; /* don't count the holes & logs as children */ if (vd->vdev_islog || vd->vdev_ishole) { if (lastlog == 0) lastlog = c; continue; } lastlog = 0; } if (children != (lastlog != 0 ? lastlog : rvd->vdev_children)) return (spa_vdev_exit(spa, NULL, txg, EINVAL)); /* next, ensure no spare or cache devices are part of the split */ if (nvlist_lookup_nvlist(nvl, ZPOOL_CONFIG_SPARES, &tmp) == 0 || nvlist_lookup_nvlist(nvl, ZPOOL_CONFIG_L2CACHE, &tmp) == 0) return (spa_vdev_exit(spa, NULL, txg, EINVAL)); vml = kmem_zalloc(children * sizeof (vdev_t *), KM_PUSHPAGE); glist = kmem_zalloc(children * sizeof (uint64_t), KM_PUSHPAGE); /* then, loop over each vdev and validate it */ for (c = 0; c < children; c++) { uint64_t is_hole = 0; (void) nvlist_lookup_uint64(child[c], ZPOOL_CONFIG_IS_HOLE, &is_hole); if (is_hole != 0) { if (spa->spa_root_vdev->vdev_child[c]->vdev_ishole || spa->spa_root_vdev->vdev_child[c]->vdev_islog) { continue; } else { error = SET_ERROR(EINVAL); break; } } /* which disk is going to be split? */ if (nvlist_lookup_uint64(child[c], ZPOOL_CONFIG_GUID, &glist[c]) != 0) { error = SET_ERROR(EINVAL); break; } /* look it up in the spa */ vml[c] = spa_lookup_by_guid(spa, glist[c], B_FALSE); if (vml[c] == NULL) { error = SET_ERROR(ENODEV); break; } /* make sure there's nothing stopping the split */ if (vml[c]->vdev_parent->vdev_ops != &vdev_mirror_ops || vml[c]->vdev_islog || vml[c]->vdev_ishole || vml[c]->vdev_isspare || vml[c]->vdev_isl2cache || !vdev_writeable(vml[c]) || vml[c]->vdev_children != 0 || vml[c]->vdev_state != VDEV_STATE_HEALTHY || c != spa->spa_root_vdev->vdev_child[c]->vdev_id) { error = SET_ERROR(EINVAL); break; } if (vdev_dtl_required(vml[c])) { error = SET_ERROR(EBUSY); break; } /* we need certain info from the top level */ VERIFY(nvlist_add_uint64(child[c], ZPOOL_CONFIG_METASLAB_ARRAY, vml[c]->vdev_top->vdev_ms_array) == 0); VERIFY(nvlist_add_uint64(child[c], ZPOOL_CONFIG_METASLAB_SHIFT, vml[c]->vdev_top->vdev_ms_shift) == 0); VERIFY(nvlist_add_uint64(child[c], ZPOOL_CONFIG_ASIZE, vml[c]->vdev_top->vdev_asize) == 0); VERIFY(nvlist_add_uint64(child[c], ZPOOL_CONFIG_ASHIFT, vml[c]->vdev_top->vdev_ashift) == 0); } if (error != 0) { kmem_free(vml, children * sizeof (vdev_t *)); kmem_free(glist, children * sizeof (uint64_t)); return (spa_vdev_exit(spa, NULL, txg, error)); } /* stop writers from using the disks */ for (c = 0; c < children; c++) { if (vml[c] != NULL) vml[c]->vdev_offline = B_TRUE; } vdev_reopen(spa->spa_root_vdev); /* * Temporarily record the splitting vdevs in the spa config. This * will disappear once the config is regenerated. */ VERIFY(nvlist_alloc(&nvl, NV_UNIQUE_NAME, KM_PUSHPAGE) == 0); VERIFY(nvlist_add_uint64_array(nvl, ZPOOL_CONFIG_SPLIT_LIST, glist, children) == 0); kmem_free(glist, children * sizeof (uint64_t)); mutex_enter(&spa->spa_props_lock); VERIFY(nvlist_add_nvlist(spa->spa_config, ZPOOL_CONFIG_SPLIT, nvl) == 0); mutex_exit(&spa->spa_props_lock); spa->spa_config_splitting = nvl; vdev_config_dirty(spa->spa_root_vdev); /* configure and create the new pool */ VERIFY(nvlist_add_string(config, ZPOOL_CONFIG_POOL_NAME, newname) == 0); VERIFY(nvlist_add_uint64(config, ZPOOL_CONFIG_POOL_STATE, exp ? POOL_STATE_EXPORTED : POOL_STATE_ACTIVE) == 0); VERIFY(nvlist_add_uint64(config, ZPOOL_CONFIG_VERSION, spa_version(spa)) == 0); VERIFY(nvlist_add_uint64(config, ZPOOL_CONFIG_POOL_TXG, spa->spa_config_txg) == 0); VERIFY(nvlist_add_uint64(config, ZPOOL_CONFIG_POOL_GUID, spa_generate_guid(NULL)) == 0); (void) nvlist_lookup_string(props, zpool_prop_to_name(ZPOOL_PROP_ALTROOT), &altroot); /* add the new pool to the namespace */ newspa = spa_add(newname, config, altroot); newspa->spa_config_txg = spa->spa_config_txg; spa_set_log_state(newspa, SPA_LOG_CLEAR); /* release the spa config lock, retaining the namespace lock */ spa_vdev_config_exit(spa, NULL, txg, 0, FTAG); if (zio_injection_enabled) zio_handle_panic_injection(spa, FTAG, 1); spa_activate(newspa, spa_mode_global); spa_async_suspend(newspa); /* create the new pool from the disks of the original pool */ error = spa_load(newspa, SPA_LOAD_IMPORT, SPA_IMPORT_ASSEMBLE, B_TRUE); if (error) goto out; /* if that worked, generate a real config for the new pool */ if (newspa->spa_root_vdev != NULL) { VERIFY(nvlist_alloc(&newspa->spa_config_splitting, NV_UNIQUE_NAME, KM_PUSHPAGE) == 0); VERIFY(nvlist_add_uint64(newspa->spa_config_splitting, ZPOOL_CONFIG_SPLIT_GUID, spa_guid(spa)) == 0); spa_config_set(newspa, spa_config_generate(newspa, NULL, -1ULL, B_TRUE)); } /* set the props */ if (props != NULL) { spa_configfile_set(newspa, props, B_FALSE); error = spa_prop_set(newspa, props); if (error) goto out; } /* flush everything */ txg = spa_vdev_config_enter(newspa); vdev_config_dirty(newspa->spa_root_vdev); (void) spa_vdev_config_exit(newspa, NULL, txg, 0, FTAG); if (zio_injection_enabled) zio_handle_panic_injection(spa, FTAG, 2); spa_async_resume(newspa); /* finally, update the original pool's config */ txg = spa_vdev_config_enter(spa); tx = dmu_tx_create_dd(spa_get_dsl(spa)->dp_mos_dir); error = dmu_tx_assign(tx, TXG_WAIT); if (error != 0) dmu_tx_abort(tx); for (c = 0; c < children; c++) { if (vml[c] != NULL) { vdev_split(vml[c]); if (error == 0) spa_history_log_internal(spa, "detach", tx, "vdev=%s", vml[c]->vdev_path); vdev_free(vml[c]); } } vdev_config_dirty(spa->spa_root_vdev); spa->spa_config_splitting = NULL; nvlist_free(nvl); if (error == 0) dmu_tx_commit(tx); (void) spa_vdev_exit(spa, NULL, txg, 0); if (zio_injection_enabled) zio_handle_panic_injection(spa, FTAG, 3); /* split is complete; log a history record */ spa_history_log_internal(newspa, "split", NULL, "from pool %s", spa_name(spa)); kmem_free(vml, children * sizeof (vdev_t *)); /* if we're not going to mount the filesystems in userland, export */ if (exp) error = spa_export_common(newname, POOL_STATE_EXPORTED, NULL, B_FALSE, B_FALSE); return (error); out: spa_unload(newspa); spa_deactivate(newspa); spa_remove(newspa); txg = spa_vdev_config_enter(spa); /* re-online all offlined disks */ for (c = 0; c < children; c++) { if (vml[c] != NULL) vml[c]->vdev_offline = B_FALSE; } vdev_reopen(spa->spa_root_vdev); nvlist_free(spa->spa_config_splitting); spa->spa_config_splitting = NULL; (void) spa_vdev_exit(spa, NULL, txg, error); kmem_free(vml, children * sizeof (vdev_t *)); return (error); } static nvlist_t * spa_nvlist_lookup_by_guid(nvlist_t **nvpp, int count, uint64_t target_guid) { int i; for (i = 0; i < count; i++) { uint64_t guid; VERIFY(nvlist_lookup_uint64(nvpp[i], ZPOOL_CONFIG_GUID, &guid) == 0); if (guid == target_guid) return (nvpp[i]); } return (NULL); } static void spa_vdev_remove_aux(nvlist_t *config, char *name, nvlist_t **dev, int count, nvlist_t *dev_to_remove) { nvlist_t **newdev = NULL; int i, j; if (count > 1) newdev = kmem_alloc((count - 1) * sizeof (void *), KM_PUSHPAGE); for (i = 0, j = 0; i < count; i++) { if (dev[i] == dev_to_remove) continue; VERIFY(nvlist_dup(dev[i], &newdev[j++], KM_PUSHPAGE) == 0); } VERIFY(nvlist_remove(config, name, DATA_TYPE_NVLIST_ARRAY) == 0); VERIFY(nvlist_add_nvlist_array(config, name, newdev, count - 1) == 0); for (i = 0; i < count - 1; i++) nvlist_free(newdev[i]); if (count > 1) kmem_free(newdev, (count - 1) * sizeof (void *)); } /* * Evacuate the device. */ static int spa_vdev_remove_evacuate(spa_t *spa, vdev_t *vd) { uint64_t txg; int error = 0; ASSERT(MUTEX_HELD(&spa_namespace_lock)); ASSERT(spa_config_held(spa, SCL_ALL, RW_WRITER) == 0); ASSERT(vd == vd->vdev_top); /* * Evacuate the device. We don't hold the config lock as writer * since we need to do I/O but we do keep the * spa_namespace_lock held. Once this completes the device * should no longer have any blocks allocated on it. */ if (vd->vdev_islog) { if (vd->vdev_stat.vs_alloc != 0) error = spa_offline_log(spa); } else { error = SET_ERROR(ENOTSUP); } if (error) return (error); /* * The evacuation succeeded. Remove any remaining MOS metadata * associated with this vdev, and wait for these changes to sync. */ ASSERT0(vd->vdev_stat.vs_alloc); txg = spa_vdev_config_enter(spa); vd->vdev_removing = B_TRUE; vdev_dirty_leaves(vd, VDD_DTL, txg); vdev_config_dirty(vd); spa_vdev_config_exit(spa, NULL, txg, 0, FTAG); return (0); } /* * Complete the removal by cleaning up the namespace. */ static void spa_vdev_remove_from_namespace(spa_t *spa, vdev_t *vd) { vdev_t *rvd = spa->spa_root_vdev; uint64_t id = vd->vdev_id; boolean_t last_vdev = (id == (rvd->vdev_children - 1)); ASSERT(MUTEX_HELD(&spa_namespace_lock)); ASSERT(spa_config_held(spa, SCL_ALL, RW_WRITER) == SCL_ALL); ASSERT(vd == vd->vdev_top); /* * Only remove any devices which are empty. */ if (vd->vdev_stat.vs_alloc != 0) return; (void) vdev_label_init(vd, 0, VDEV_LABEL_REMOVE); if (list_link_active(&vd->vdev_state_dirty_node)) vdev_state_clean(vd); if (list_link_active(&vd->vdev_config_dirty_node)) vdev_config_clean(vd); vdev_free(vd); if (last_vdev) { vdev_compact_children(rvd); } else { vd = vdev_alloc_common(spa, id, 0, &vdev_hole_ops); vdev_add_child(rvd, vd); } vdev_config_dirty(rvd); /* * Reassess the health of our root vdev. */ vdev_reopen(rvd); } /* * Remove a device from the pool - * * Removing a device from the vdev namespace requires several steps * and can take a significant amount of time. As a result we use * the spa_vdev_config_[enter/exit] functions which allow us to * grab and release the spa_config_lock while still holding the namespace * lock. During each step the configuration is synced out. * * Currently, this supports removing only hot spares, slogs, and level 2 ARC * devices. */ int spa_vdev_remove(spa_t *spa, uint64_t guid, boolean_t unspare) { vdev_t *vd; metaslab_group_t *mg; nvlist_t **spares, **l2cache, *nv; uint64_t txg = 0; uint_t nspares, nl2cache; int error = 0; boolean_t locked = MUTEX_HELD(&spa_namespace_lock); ASSERT(spa_writeable(spa)); if (!locked) txg = spa_vdev_enter(spa); vd = spa_lookup_by_guid(spa, guid, B_FALSE); if (spa->spa_spares.sav_vdevs != NULL && nvlist_lookup_nvlist_array(spa->spa_spares.sav_config, ZPOOL_CONFIG_SPARES, &spares, &nspares) == 0 && (nv = spa_nvlist_lookup_by_guid(spares, nspares, guid)) != NULL) { /* * Only remove the hot spare if it's not currently in use * in this pool. */ if (vd == NULL || unspare) { spa_vdev_remove_aux(spa->spa_spares.sav_config, ZPOOL_CONFIG_SPARES, spares, nspares, nv); spa_load_spares(spa); spa->spa_spares.sav_sync = B_TRUE; } else { error = SET_ERROR(EBUSY); } } else if (spa->spa_l2cache.sav_vdevs != NULL && nvlist_lookup_nvlist_array(spa->spa_l2cache.sav_config, ZPOOL_CONFIG_L2CACHE, &l2cache, &nl2cache) == 0 && (nv = spa_nvlist_lookup_by_guid(l2cache, nl2cache, guid)) != NULL) { /* * Cache devices can always be removed. */ spa_vdev_remove_aux(spa->spa_l2cache.sav_config, ZPOOL_CONFIG_L2CACHE, l2cache, nl2cache, nv); spa_load_l2cache(spa); spa->spa_l2cache.sav_sync = B_TRUE; } else if (vd != NULL && vd->vdev_islog) { ASSERT(!locked); ASSERT(vd == vd->vdev_top); mg = vd->vdev_mg; /* * Stop allocating from this vdev. */ metaslab_group_passivate(mg); /* * Wait for the youngest allocations and frees to sync, * and then wait for the deferral of those frees to finish. */ spa_vdev_config_exit(spa, NULL, txg + TXG_CONCURRENT_STATES + TXG_DEFER_SIZE, 0, FTAG); /* * Attempt to evacuate the vdev. */ error = spa_vdev_remove_evacuate(spa, vd); txg = spa_vdev_config_enter(spa); /* * If we couldn't evacuate the vdev, unwind. */ if (error) { metaslab_group_activate(mg); return (spa_vdev_exit(spa, NULL, txg, error)); } /* * Clean up the vdev namespace. */ spa_vdev_remove_from_namespace(spa, vd); } else if (vd != NULL) { /* * Normal vdevs cannot be removed (yet). */ error = SET_ERROR(ENOTSUP); } else { /* * There is no vdev of any kind with the specified guid. */ error = SET_ERROR(ENOENT); } if (!locked) return (spa_vdev_exit(spa, NULL, txg, error)); return (error); } /* * Find any device that's done replacing, or a vdev marked 'unspare' that's * currently spared, so we can detach it. */ static vdev_t * spa_vdev_resilver_done_hunt(vdev_t *vd) { vdev_t *newvd, *oldvd; int c; for (c = 0; c < vd->vdev_children; c++) { oldvd = spa_vdev_resilver_done_hunt(vd->vdev_child[c]); if (oldvd != NULL) return (oldvd); } /* * Check for a completed replacement. We always consider the first * vdev in the list to be the oldest vdev, and the last one to be * the newest (see spa_vdev_attach() for how that works). In * the case where the newest vdev is faulted, we will not automatically * remove it after a resilver completes. This is OK as it will require * user intervention to determine which disk the admin wishes to keep. */ if (vd->vdev_ops == &vdev_replacing_ops) { ASSERT(vd->vdev_children > 1); newvd = vd->vdev_child[vd->vdev_children - 1]; oldvd = vd->vdev_child[0]; if (vdev_dtl_empty(newvd, DTL_MISSING) && vdev_dtl_empty(newvd, DTL_OUTAGE) && !vdev_dtl_required(oldvd)) return (oldvd); } /* * Check for a completed resilver with the 'unspare' flag set. */ if (vd->vdev_ops == &vdev_spare_ops) { vdev_t *first = vd->vdev_child[0]; vdev_t *last = vd->vdev_child[vd->vdev_children - 1]; if (last->vdev_unspare) { oldvd = first; newvd = last; } else if (first->vdev_unspare) { oldvd = last; newvd = first; } else { oldvd = NULL; } if (oldvd != NULL && vdev_dtl_empty(newvd, DTL_MISSING) && vdev_dtl_empty(newvd, DTL_OUTAGE) && !vdev_dtl_required(oldvd)) return (oldvd); /* * If there are more than two spares attached to a disk, * and those spares are not required, then we want to * attempt to free them up now so that they can be used * by other pools. Once we're back down to a single * disk+spare, we stop removing them. */ if (vd->vdev_children > 2) { newvd = vd->vdev_child[1]; if (newvd->vdev_isspare && last->vdev_isspare && vdev_dtl_empty(last, DTL_MISSING) && vdev_dtl_empty(last, DTL_OUTAGE) && !vdev_dtl_required(newvd)) return (newvd); } } return (NULL); } static void spa_vdev_resilver_done(spa_t *spa) { vdev_t *vd, *pvd, *ppvd; uint64_t guid, sguid, pguid, ppguid; spa_config_enter(spa, SCL_ALL, FTAG, RW_WRITER); while ((vd = spa_vdev_resilver_done_hunt(spa->spa_root_vdev)) != NULL) { pvd = vd->vdev_parent; ppvd = pvd->vdev_parent; guid = vd->vdev_guid; pguid = pvd->vdev_guid; ppguid = ppvd->vdev_guid; sguid = 0; /* * If we have just finished replacing a hot spared device, then * we need to detach the parent's first child (the original hot * spare) as well. */ if (ppvd->vdev_ops == &vdev_spare_ops && pvd->vdev_id == 0 && ppvd->vdev_children == 2) { ASSERT(pvd->vdev_ops == &vdev_replacing_ops); sguid = ppvd->vdev_child[1]->vdev_guid; } ASSERT(vd->vdev_resilver_txg == 0 || !vdev_dtl_required(vd)); spa_config_exit(spa, SCL_ALL, FTAG); if (spa_vdev_detach(spa, guid, pguid, B_TRUE) != 0) return; if (sguid && spa_vdev_detach(spa, sguid, ppguid, B_TRUE) != 0) return; spa_config_enter(spa, SCL_ALL, FTAG, RW_WRITER); } spa_config_exit(spa, SCL_ALL, FTAG); } /* * Update the stored path or FRU for this vdev. */ int spa_vdev_set_common(spa_t *spa, uint64_t guid, const char *value, boolean_t ispath) { vdev_t *vd; boolean_t sync = B_FALSE; ASSERT(spa_writeable(spa)); spa_vdev_state_enter(spa, SCL_ALL); if ((vd = spa_lookup_by_guid(spa, guid, B_TRUE)) == NULL) return (spa_vdev_state_exit(spa, NULL, ENOENT)); if (!vd->vdev_ops->vdev_op_leaf) return (spa_vdev_state_exit(spa, NULL, ENOTSUP)); if (ispath) { if (strcmp(value, vd->vdev_path) != 0) { spa_strfree(vd->vdev_path); vd->vdev_path = spa_strdup(value); sync = B_TRUE; } } else { if (vd->vdev_fru == NULL) { vd->vdev_fru = spa_strdup(value); sync = B_TRUE; } else if (strcmp(value, vd->vdev_fru) != 0) { spa_strfree(vd->vdev_fru); vd->vdev_fru = spa_strdup(value); sync = B_TRUE; } } return (spa_vdev_state_exit(spa, sync ? vd : NULL, 0)); } int spa_vdev_setpath(spa_t *spa, uint64_t guid, const char *newpath) { return (spa_vdev_set_common(spa, guid, newpath, B_TRUE)); } int spa_vdev_setfru(spa_t *spa, uint64_t guid, const char *newfru) { return (spa_vdev_set_common(spa, guid, newfru, B_FALSE)); } /* * ========================================================================== * SPA Scanning * ========================================================================== */ int spa_scan_stop(spa_t *spa) { ASSERT(spa_config_held(spa, SCL_ALL, RW_WRITER) == 0); if (dsl_scan_resilvering(spa->spa_dsl_pool)) return (SET_ERROR(EBUSY)); return (dsl_scan_cancel(spa->spa_dsl_pool)); } int spa_scan(spa_t *spa, pool_scan_func_t func) { ASSERT(spa_config_held(spa, SCL_ALL, RW_WRITER) == 0); if (func >= POOL_SCAN_FUNCS || func == POOL_SCAN_NONE) return (SET_ERROR(ENOTSUP)); /* * If a resilver was requested, but there is no DTL on a * writeable leaf device, we have nothing to do. */ if (func == POOL_SCAN_RESILVER && !vdev_resilver_needed(spa->spa_root_vdev, NULL, NULL)) { spa_async_request(spa, SPA_ASYNC_RESILVER_DONE); return (0); } return (dsl_scan(spa->spa_dsl_pool, func)); } /* * ========================================================================== * SPA async task processing * ========================================================================== */ static void spa_async_remove(spa_t *spa, vdev_t *vd) { int c; if (vd->vdev_remove_wanted) { vd->vdev_remove_wanted = B_FALSE; vd->vdev_delayed_close = B_FALSE; vdev_set_state(vd, B_FALSE, VDEV_STATE_REMOVED, VDEV_AUX_NONE); /* * We want to clear the stats, but we don't want to do a full * vdev_clear() as that will cause us to throw away * degraded/faulted state as well as attempt to reopen the * device, all of which is a waste. */ vd->vdev_stat.vs_read_errors = 0; vd->vdev_stat.vs_write_errors = 0; vd->vdev_stat.vs_checksum_errors = 0; vdev_state_dirty(vd->vdev_top); } for (c = 0; c < vd->vdev_children; c++) spa_async_remove(spa, vd->vdev_child[c]); } static void spa_async_probe(spa_t *spa, vdev_t *vd) { int c; if (vd->vdev_probe_wanted) { vd->vdev_probe_wanted = B_FALSE; vdev_reopen(vd); /* vdev_open() does the actual probe */ } for (c = 0; c < vd->vdev_children; c++) spa_async_probe(spa, vd->vdev_child[c]); } static void spa_async_autoexpand(spa_t *spa, vdev_t *vd) { int c; if (!spa->spa_autoexpand) return; for (c = 0; c < vd->vdev_children; c++) { vdev_t *cvd = vd->vdev_child[c]; spa_async_autoexpand(spa, cvd); } if (!vd->vdev_ops->vdev_op_leaf || vd->vdev_physpath == NULL) return; spa_event_notify(vd->vdev_spa, vd, FM_EREPORT_ZFS_DEVICE_AUTOEXPAND); } static void spa_async_thread(spa_t *spa) { int tasks, i; ASSERT(spa->spa_sync_on); mutex_enter(&spa->spa_async_lock); tasks = spa->spa_async_tasks; spa->spa_async_tasks = 0; mutex_exit(&spa->spa_async_lock); /* * See if the config needs to be updated. */ if (tasks & SPA_ASYNC_CONFIG_UPDATE) { uint64_t old_space, new_space; mutex_enter(&spa_namespace_lock); old_space = metaslab_class_get_space(spa_normal_class(spa)); spa_config_update(spa, SPA_CONFIG_UPDATE_POOL); new_space = metaslab_class_get_space(spa_normal_class(spa)); mutex_exit(&spa_namespace_lock); /* * If the pool grew as a result of the config update, * then log an internal history event. */ if (new_space != old_space) { spa_history_log_internal(spa, "vdev online", NULL, "pool '%s' size: %llu(+%llu)", spa_name(spa), new_space, new_space - old_space); } } /* * See if any devices need to be marked REMOVED. */ if (tasks & SPA_ASYNC_REMOVE) { spa_vdev_state_enter(spa, SCL_NONE); spa_async_remove(spa, spa->spa_root_vdev); for (i = 0; i < spa->spa_l2cache.sav_count; i++) spa_async_remove(spa, spa->spa_l2cache.sav_vdevs[i]); for (i = 0; i < spa->spa_spares.sav_count; i++) spa_async_remove(spa, spa->spa_spares.sav_vdevs[i]); (void) spa_vdev_state_exit(spa, NULL, 0); } if ((tasks & SPA_ASYNC_AUTOEXPAND) && !spa_suspended(spa)) { spa_config_enter(spa, SCL_CONFIG, FTAG, RW_READER); spa_async_autoexpand(spa, spa->spa_root_vdev); spa_config_exit(spa, SCL_CONFIG, FTAG); } /* * See if any devices need to be probed. */ if (tasks & SPA_ASYNC_PROBE) { spa_vdev_state_enter(spa, SCL_NONE); spa_async_probe(spa, spa->spa_root_vdev); (void) spa_vdev_state_exit(spa, NULL, 0); } /* * If any devices are done replacing, detach them. */ if (tasks & SPA_ASYNC_RESILVER_DONE) spa_vdev_resilver_done(spa); /* * Kick off a resilver. */ if (tasks & SPA_ASYNC_RESILVER) dsl_resilver_restart(spa->spa_dsl_pool, 0); /* * Let the world know that we're done. */ mutex_enter(&spa->spa_async_lock); spa->spa_async_thread = NULL; cv_broadcast(&spa->spa_async_cv); mutex_exit(&spa->spa_async_lock); thread_exit(); } void spa_async_suspend(spa_t *spa) { mutex_enter(&spa->spa_async_lock); spa->spa_async_suspended++; while (spa->spa_async_thread != NULL) cv_wait(&spa->spa_async_cv, &spa->spa_async_lock); mutex_exit(&spa->spa_async_lock); } void spa_async_resume(spa_t *spa) { mutex_enter(&spa->spa_async_lock); ASSERT(spa->spa_async_suspended != 0); spa->spa_async_suspended--; mutex_exit(&spa->spa_async_lock); } static void spa_async_dispatch(spa_t *spa) { mutex_enter(&spa->spa_async_lock); if (spa->spa_async_tasks && !spa->spa_async_suspended && spa->spa_async_thread == NULL && rootdir != NULL && !vn_is_readonly(rootdir)) spa->spa_async_thread = thread_create(NULL, 0, spa_async_thread, spa, 0, &p0, TS_RUN, maxclsyspri); mutex_exit(&spa->spa_async_lock); } void spa_async_request(spa_t *spa, int task) { zfs_dbgmsg("spa=%s async request task=%u", spa->spa_name, task); mutex_enter(&spa->spa_async_lock); spa->spa_async_tasks |= task; mutex_exit(&spa->spa_async_lock); } /* * ========================================================================== * SPA syncing routines * ========================================================================== */ static int bpobj_enqueue_cb(void *arg, const blkptr_t *bp, dmu_tx_t *tx) { bpobj_t *bpo = arg; bpobj_enqueue(bpo, bp, tx); return (0); } static int spa_free_sync_cb(void *arg, const blkptr_t *bp, dmu_tx_t *tx) { zio_t *zio = arg; zio_nowait(zio_free_sync(zio, zio->io_spa, dmu_tx_get_txg(tx), bp, zio->io_flags)); return (0); } /* * Note: this simple function is not inlined to make it easier to dtrace the * amount of time spent syncing frees. */ static void spa_sync_frees(spa_t *spa, bplist_t *bpl, dmu_tx_t *tx) { zio_t *zio = zio_root(spa, NULL, NULL, 0); bplist_iterate(bpl, spa_free_sync_cb, zio, tx); VERIFY(zio_wait(zio) == 0); } /* * Note: this simple function is not inlined to make it easier to dtrace the * amount of time spent syncing deferred frees. */ static void spa_sync_deferred_frees(spa_t *spa, dmu_tx_t *tx) { zio_t *zio = zio_root(spa, NULL, NULL, 0); VERIFY3U(bpobj_iterate(&spa->spa_deferred_bpobj, spa_free_sync_cb, zio, tx), ==, 0); VERIFY0(zio_wait(zio)); } static void spa_sync_nvlist(spa_t *spa, uint64_t obj, nvlist_t *nv, dmu_tx_t *tx) { char *packed = NULL; size_t bufsize; size_t nvsize = 0; dmu_buf_t *db; VERIFY(nvlist_size(nv, &nvsize, NV_ENCODE_XDR) == 0); /* * Write full (SPA_CONFIG_BLOCKSIZE) blocks of configuration * information. This avoids the dmu_buf_will_dirty() path and * saves us a pre-read to get data we don't actually care about. */ bufsize = P2ROUNDUP((uint64_t)nvsize, SPA_CONFIG_BLOCKSIZE); packed = vmem_alloc(bufsize, KM_PUSHPAGE); VERIFY(nvlist_pack(nv, &packed, &nvsize, NV_ENCODE_XDR, KM_PUSHPAGE) == 0); bzero(packed + nvsize, bufsize - nvsize); dmu_write(spa->spa_meta_objset, obj, 0, bufsize, packed, tx); vmem_free(packed, bufsize); VERIFY(0 == dmu_bonus_hold(spa->spa_meta_objset, obj, FTAG, &db)); dmu_buf_will_dirty(db, tx); *(uint64_t *)db->db_data = nvsize; dmu_buf_rele(db, FTAG); } static void spa_sync_aux_dev(spa_t *spa, spa_aux_vdev_t *sav, dmu_tx_t *tx, const char *config, const char *entry) { nvlist_t *nvroot; nvlist_t **list; int i; if (!sav->sav_sync) return; /* * Update the MOS nvlist describing the list of available devices. * spa_validate_aux() will have already made sure this nvlist is * valid and the vdevs are labeled appropriately. */ if (sav->sav_object == 0) { sav->sav_object = dmu_object_alloc(spa->spa_meta_objset, DMU_OT_PACKED_NVLIST, 1 << 14, DMU_OT_PACKED_NVLIST_SIZE, sizeof (uint64_t), tx); VERIFY(zap_update(spa->spa_meta_objset, DMU_POOL_DIRECTORY_OBJECT, entry, sizeof (uint64_t), 1, &sav->sav_object, tx) == 0); } VERIFY(nvlist_alloc(&nvroot, NV_UNIQUE_NAME, KM_PUSHPAGE) == 0); if (sav->sav_count == 0) { VERIFY(nvlist_add_nvlist_array(nvroot, config, NULL, 0) == 0); } else { list = kmem_alloc(sav->sav_count*sizeof (void *), KM_PUSHPAGE); for (i = 0; i < sav->sav_count; i++) list[i] = vdev_config_generate(spa, sav->sav_vdevs[i], B_FALSE, VDEV_CONFIG_L2CACHE); VERIFY(nvlist_add_nvlist_array(nvroot, config, list, sav->sav_count) == 0); for (i = 0; i < sav->sav_count; i++) nvlist_free(list[i]); kmem_free(list, sav->sav_count * sizeof (void *)); } spa_sync_nvlist(spa, sav->sav_object, nvroot, tx); nvlist_free(nvroot); sav->sav_sync = B_FALSE; } static void spa_sync_config_object(spa_t *spa, dmu_tx_t *tx) { nvlist_t *config; if (list_is_empty(&spa->spa_config_dirty_list)) return; spa_config_enter(spa, SCL_STATE, FTAG, RW_READER); config = spa_config_generate(spa, spa->spa_root_vdev, dmu_tx_get_txg(tx), B_FALSE); /* * If we're upgrading the spa version then make sure that * the config object gets updated with the correct version. */ if (spa->spa_ubsync.ub_version < spa->spa_uberblock.ub_version) fnvlist_add_uint64(config, ZPOOL_CONFIG_VERSION, spa->spa_uberblock.ub_version); spa_config_exit(spa, SCL_STATE, FTAG); if (spa->spa_config_syncing) nvlist_free(spa->spa_config_syncing); spa->spa_config_syncing = config; spa_sync_nvlist(spa, spa->spa_config_object, config, tx); } static void spa_sync_version(void *arg, dmu_tx_t *tx) { uint64_t *versionp = arg; uint64_t version = *versionp; spa_t *spa = dmu_tx_pool(tx)->dp_spa; /* * Setting the version is special cased when first creating the pool. */ ASSERT(tx->tx_txg != TXG_INITIAL); ASSERT(SPA_VERSION_IS_SUPPORTED(version)); ASSERT(version >= spa_version(spa)); spa->spa_uberblock.ub_version = version; vdev_config_dirty(spa->spa_root_vdev); spa_history_log_internal(spa, "set", tx, "version=%lld", version); } /* * Set zpool properties. */ static void spa_sync_props(void *arg, dmu_tx_t *tx) { nvlist_t *nvp = arg; spa_t *spa = dmu_tx_pool(tx)->dp_spa; objset_t *mos = spa->spa_meta_objset; nvpair_t *elem = NULL; mutex_enter(&spa->spa_props_lock); while ((elem = nvlist_next_nvpair(nvp, elem))) { uint64_t intval; char *strval, *fname; zpool_prop_t prop; const char *propname; zprop_type_t proptype; spa_feature_t fid; prop = zpool_name_to_prop(nvpair_name(elem)); switch ((int)prop) { case ZPROP_INVAL: /* * We checked this earlier in spa_prop_validate(). */ ASSERT(zpool_prop_feature(nvpair_name(elem))); fname = strchr(nvpair_name(elem), '@') + 1; VERIFY0(zfeature_lookup_name(fname, &fid)); spa_feature_enable(spa, fid, tx); spa_history_log_internal(spa, "set", tx, "%s=enabled", nvpair_name(elem)); break; case ZPOOL_PROP_VERSION: intval = fnvpair_value_uint64(elem); /* * The version is synced seperatly before other * properties and should be correct by now. */ ASSERT3U(spa_version(spa), >=, intval); break; case ZPOOL_PROP_ALTROOT: /* * 'altroot' is a non-persistent property. It should * have been set temporarily at creation or import time. */ ASSERT(spa->spa_root != NULL); break; case ZPOOL_PROP_READONLY: case ZPOOL_PROP_CACHEFILE: /* * 'readonly' and 'cachefile' are also non-persisitent * properties. */ break; case ZPOOL_PROP_COMMENT: strval = fnvpair_value_string(elem); if (spa->spa_comment != NULL) spa_strfree(spa->spa_comment); spa->spa_comment = spa_strdup(strval); /* * We need to dirty the configuration on all the vdevs * so that their labels get updated. It's unnecessary * to do this for pool creation since the vdev's * configuratoin has already been dirtied. */ if (tx->tx_txg != TXG_INITIAL) vdev_config_dirty(spa->spa_root_vdev); spa_history_log_internal(spa, "set", tx, "%s=%s", nvpair_name(elem), strval); break; default: /* * Set pool property values in the poolprops mos object. */ if (spa->spa_pool_props_object == 0) { spa->spa_pool_props_object = zap_create_link(mos, DMU_OT_POOL_PROPS, DMU_POOL_DIRECTORY_OBJECT, DMU_POOL_PROPS, tx); } /* normalize the property name */ propname = zpool_prop_to_name(prop); proptype = zpool_prop_get_type(prop); if (nvpair_type(elem) == DATA_TYPE_STRING) { ASSERT(proptype == PROP_TYPE_STRING); strval = fnvpair_value_string(elem); VERIFY0(zap_update(mos, spa->spa_pool_props_object, propname, 1, strlen(strval) + 1, strval, tx)); spa_history_log_internal(spa, "set", tx, "%s=%s", nvpair_name(elem), strval); } else if (nvpair_type(elem) == DATA_TYPE_UINT64) { intval = fnvpair_value_uint64(elem); if (proptype == PROP_TYPE_INDEX) { const char *unused; VERIFY0(zpool_prop_index_to_string( prop, intval, &unused)); } VERIFY0(zap_update(mos, spa->spa_pool_props_object, propname, 8, 1, &intval, tx)); spa_history_log_internal(spa, "set", tx, "%s=%lld", nvpair_name(elem), intval); } else { ASSERT(0); /* not allowed */ } switch (prop) { case ZPOOL_PROP_DELEGATION: spa->spa_delegation = intval; break; case ZPOOL_PROP_BOOTFS: spa->spa_bootfs = intval; break; case ZPOOL_PROP_FAILUREMODE: spa->spa_failmode = intval; break; case ZPOOL_PROP_AUTOEXPAND: spa->spa_autoexpand = intval; if (tx->tx_txg != TXG_INITIAL) spa_async_request(spa, SPA_ASYNC_AUTOEXPAND); break; case ZPOOL_PROP_DEDUPDITTO: spa->spa_dedup_ditto = intval; break; default: break; } } } mutex_exit(&spa->spa_props_lock); } /* * Perform one-time upgrade on-disk changes. spa_version() does not * reflect the new version this txg, so there must be no changes this * txg to anything that the upgrade code depends on after it executes. * Therefore this must be called after dsl_pool_sync() does the sync * tasks. */ static void spa_sync_upgrades(spa_t *spa, dmu_tx_t *tx) { dsl_pool_t *dp = spa->spa_dsl_pool; ASSERT(spa->spa_sync_pass == 1); rrw_enter(&dp->dp_config_rwlock, RW_WRITER, FTAG); if (spa->spa_ubsync.ub_version < SPA_VERSION_ORIGIN && spa->spa_uberblock.ub_version >= SPA_VERSION_ORIGIN) { dsl_pool_create_origin(dp, tx); /* Keeping the origin open increases spa_minref */ spa->spa_minref += 3; } if (spa->spa_ubsync.ub_version < SPA_VERSION_NEXT_CLONES && spa->spa_uberblock.ub_version >= SPA_VERSION_NEXT_CLONES) { dsl_pool_upgrade_clones(dp, tx); } if (spa->spa_ubsync.ub_version < SPA_VERSION_DIR_CLONES && spa->spa_uberblock.ub_version >= SPA_VERSION_DIR_CLONES) { dsl_pool_upgrade_dir_clones(dp, tx); /* Keeping the freedir open increases spa_minref */ spa->spa_minref += 3; } if (spa->spa_ubsync.ub_version < SPA_VERSION_FEATURES && spa->spa_uberblock.ub_version >= SPA_VERSION_FEATURES) { spa_feature_create_zap_objects(spa, tx); } rrw_exit(&dp->dp_config_rwlock, FTAG); } /* * Sync the specified transaction group. New blocks may be dirtied as * part of the process, so we iterate until it converges. */ void spa_sync(spa_t *spa, uint64_t txg) { dsl_pool_t *dp = spa->spa_dsl_pool; objset_t *mos = spa->spa_meta_objset; bplist_t *free_bpl = &spa->spa_free_bplist[txg & TXG_MASK]; vdev_t *rvd = spa->spa_root_vdev; vdev_t *vd; dmu_tx_t *tx; int error; int c; VERIFY(spa_writeable(spa)); /* * Lock out configuration changes. */ spa_config_enter(spa, SCL_CONFIG, FTAG, RW_READER); spa->spa_syncing_txg = txg; spa->spa_sync_pass = 0; /* * If there are any pending vdev state changes, convert them * into config changes that go out with this transaction group. */ spa_config_enter(spa, SCL_STATE, FTAG, RW_READER); while (list_head(&spa->spa_state_dirty_list) != NULL) { /* * We need the write lock here because, for aux vdevs, * calling vdev_config_dirty() modifies sav_config. * This is ugly and will become unnecessary when we * eliminate the aux vdev wart by integrating all vdevs * into the root vdev tree. */ spa_config_exit(spa, SCL_CONFIG | SCL_STATE, FTAG); spa_config_enter(spa, SCL_CONFIG | SCL_STATE, FTAG, RW_WRITER); while ((vd = list_head(&spa->spa_state_dirty_list)) != NULL) { vdev_state_clean(vd); vdev_config_dirty(vd); } spa_config_exit(spa, SCL_CONFIG | SCL_STATE, FTAG); spa_config_enter(spa, SCL_CONFIG | SCL_STATE, FTAG, RW_READER); } spa_config_exit(spa, SCL_STATE, FTAG); tx = dmu_tx_create_assigned(dp, txg); spa->spa_sync_starttime = gethrtime(); taskq_cancel_id(system_taskq, spa->spa_deadman_tqid); spa->spa_deadman_tqid = taskq_dispatch_delay(system_taskq, spa_deadman, spa, TQ_PUSHPAGE, ddi_get_lbolt() + NSEC_TO_TICK(spa->spa_deadman_synctime)); /* * If we are upgrading to SPA_VERSION_RAIDZ_DEFLATE this txg, * set spa_deflate if we have no raid-z vdevs. */ if (spa->spa_ubsync.ub_version < SPA_VERSION_RAIDZ_DEFLATE && spa->spa_uberblock.ub_version >= SPA_VERSION_RAIDZ_DEFLATE) { int i; for (i = 0; i < rvd->vdev_children; i++) { vd = rvd->vdev_child[i]; if (vd->vdev_deflate_ratio != SPA_MINBLOCKSIZE) break; } if (i == rvd->vdev_children) { spa->spa_deflate = TRUE; VERIFY(0 == zap_add(spa->spa_meta_objset, DMU_POOL_DIRECTORY_OBJECT, DMU_POOL_DEFLATE, sizeof (uint64_t), 1, &spa->spa_deflate, tx)); } } /* * If anything has changed in this txg, or if someone is waiting * for this txg to sync (eg, spa_vdev_remove()), push the * deferred frees from the previous txg. If not, leave them * alone so that we don't generate work on an otherwise idle * system. */ if (!txg_list_empty(&dp->dp_dirty_datasets, txg) || !txg_list_empty(&dp->dp_dirty_dirs, txg) || !txg_list_empty(&dp->dp_sync_tasks, txg) || ((dsl_scan_active(dp->dp_scan) || txg_sync_waiting(dp)) && !spa_shutting_down(spa))) { spa_sync_deferred_frees(spa, tx); } /* * Iterate to convergence. */ do { int pass = ++spa->spa_sync_pass; spa_sync_config_object(spa, tx); spa_sync_aux_dev(spa, &spa->spa_spares, tx, ZPOOL_CONFIG_SPARES, DMU_POOL_SPARES); spa_sync_aux_dev(spa, &spa->spa_l2cache, tx, ZPOOL_CONFIG_L2CACHE, DMU_POOL_L2CACHE); spa_errlog_sync(spa, txg); dsl_pool_sync(dp, txg); if (pass < zfs_sync_pass_deferred_free) { spa_sync_frees(spa, free_bpl, tx); } else { bplist_iterate(free_bpl, bpobj_enqueue_cb, &spa->spa_deferred_bpobj, tx); } ddt_sync(spa, txg); dsl_scan_sync(dp, tx); while ((vd = txg_list_remove(&spa->spa_vdev_txg_list, txg))) vdev_sync(vd, txg); if (pass == 1) spa_sync_upgrades(spa, tx); } while (dmu_objset_is_dirty(mos, txg)); /* * Rewrite the vdev configuration (which includes the uberblock) * to commit the transaction group. * * If there are no dirty vdevs, we sync the uberblock to a few * random top-level vdevs that are known to be visible in the * config cache (see spa_vdev_add() for a complete description). * If there *are* dirty vdevs, sync the uberblock to all vdevs. */ for (;;) { /* * We hold SCL_STATE to prevent vdev open/close/etc. * while we're attempting to write the vdev labels. */ spa_config_enter(spa, SCL_STATE, FTAG, RW_READER); if (list_is_empty(&spa->spa_config_dirty_list)) { vdev_t *svd[SPA_DVAS_PER_BP]; int svdcount = 0; int children = rvd->vdev_children; int c0 = spa_get_random(children); for (c = 0; c < children; c++) { vd = rvd->vdev_child[(c0 + c) % children]; if (vd->vdev_ms_array == 0 || vd->vdev_islog) continue; svd[svdcount++] = vd; if (svdcount == SPA_DVAS_PER_BP) break; } error = vdev_config_sync(svd, svdcount, txg, B_FALSE); if (error != 0) error = vdev_config_sync(svd, svdcount, txg, B_TRUE); } else { error = vdev_config_sync(rvd->vdev_child, rvd->vdev_children, txg, B_FALSE); if (error != 0) error = vdev_config_sync(rvd->vdev_child, rvd->vdev_children, txg, B_TRUE); } if (error == 0) spa->spa_last_synced_guid = rvd->vdev_guid; spa_config_exit(spa, SCL_STATE, FTAG); if (error == 0) break; zio_suspend(spa, NULL); zio_resume_wait(spa); } dmu_tx_commit(tx); taskq_cancel_id(system_taskq, spa->spa_deadman_tqid); spa->spa_deadman_tqid = 0; /* * Clear the dirty config list. */ while ((vd = list_head(&spa->spa_config_dirty_list)) != NULL) vdev_config_clean(vd); /* * Now that the new config has synced transactionally, * let it become visible to the config cache. */ if (spa->spa_config_syncing != NULL) { spa_config_set(spa, spa->spa_config_syncing); spa->spa_config_txg = txg; spa->spa_config_syncing = NULL; } spa->spa_ubsync = spa->spa_uberblock; dsl_pool_sync_done(dp, txg); /* * Update usable space statistics. */ while ((vd = txg_list_remove(&spa->spa_vdev_txg_list, TXG_CLEAN(txg)))) vdev_sync_done(vd, txg); spa_update_dspace(spa); /* * It had better be the case that we didn't dirty anything * since vdev_config_sync(). */ ASSERT(txg_list_empty(&dp->dp_dirty_datasets, txg)); ASSERT(txg_list_empty(&dp->dp_dirty_dirs, txg)); ASSERT(txg_list_empty(&spa->spa_vdev_txg_list, txg)); spa->spa_sync_pass = 0; spa_config_exit(spa, SCL_CONFIG, FTAG); spa_handle_ignored_writes(spa); /* * If any async tasks have been requested, kick them off. */ spa_async_dispatch(spa); } /* * Sync all pools. We don't want to hold the namespace lock across these * operations, so we take a reference on the spa_t and drop the lock during the * sync. */ void spa_sync_allpools(void) { spa_t *spa = NULL; mutex_enter(&spa_namespace_lock); while ((spa = spa_next(spa)) != NULL) { if (spa_state(spa) != POOL_STATE_ACTIVE || !spa_writeable(spa) || spa_suspended(spa)) continue; spa_open_ref(spa, FTAG); mutex_exit(&spa_namespace_lock); txg_wait_synced(spa_get_dsl(spa), 0); mutex_enter(&spa_namespace_lock); spa_close(spa, FTAG); } mutex_exit(&spa_namespace_lock); } /* * ========================================================================== * Miscellaneous routines * ========================================================================== */ /* * Remove all pools in the system. */ void spa_evict_all(void) { spa_t *spa; /* * Remove all cached state. All pools should be closed now, * so every spa in the AVL tree should be unreferenced. */ mutex_enter(&spa_namespace_lock); while ((spa = spa_next(NULL)) != NULL) { /* * Stop async tasks. The async thread may need to detach * a device that's been replaced, which requires grabbing * spa_namespace_lock, so we must drop it here. */ spa_open_ref(spa, FTAG); mutex_exit(&spa_namespace_lock); spa_async_suspend(spa); mutex_enter(&spa_namespace_lock); spa_close(spa, FTAG); if (spa->spa_state != POOL_STATE_UNINITIALIZED) { spa_unload(spa); spa_deactivate(spa); } spa_remove(spa); } mutex_exit(&spa_namespace_lock); } vdev_t * spa_lookup_by_guid(spa_t *spa, uint64_t guid, boolean_t aux) { vdev_t *vd; int i; if ((vd = vdev_lookup_by_guid(spa->spa_root_vdev, guid)) != NULL) return (vd); if (aux) { for (i = 0; i < spa->spa_l2cache.sav_count; i++) { vd = spa->spa_l2cache.sav_vdevs[i]; if (vd->vdev_guid == guid) return (vd); } for (i = 0; i < spa->spa_spares.sav_count; i++) { vd = spa->spa_spares.sav_vdevs[i]; if (vd->vdev_guid == guid) return (vd); } } return (NULL); } void spa_upgrade(spa_t *spa, uint64_t version) { ASSERT(spa_writeable(spa)); spa_config_enter(spa, SCL_ALL, FTAG, RW_WRITER); /* * This should only be called for a non-faulted pool, and since a * future version would result in an unopenable pool, this shouldn't be * possible. */ ASSERT(SPA_VERSION_IS_SUPPORTED(spa->spa_uberblock.ub_version)); ASSERT3U(version, >=, spa->spa_uberblock.ub_version); spa->spa_uberblock.ub_version = version; vdev_config_dirty(spa->spa_root_vdev); spa_config_exit(spa, SCL_ALL, FTAG); txg_wait_synced(spa_get_dsl(spa), 0); } boolean_t spa_has_spare(spa_t *spa, uint64_t guid) { int i; uint64_t spareguid; spa_aux_vdev_t *sav = &spa->spa_spares; for (i = 0; i < sav->sav_count; i++) if (sav->sav_vdevs[i]->vdev_guid == guid) return (B_TRUE); for (i = 0; i < sav->sav_npending; i++) { if (nvlist_lookup_uint64(sav->sav_pending[i], ZPOOL_CONFIG_GUID, &spareguid) == 0 && spareguid == guid) return (B_TRUE); } return (B_FALSE); } /* * Check if a pool has an active shared spare device. * Note: reference count of an active spare is 2, as a spare and as a replace */ static boolean_t spa_has_active_shared_spare(spa_t *spa) { int i, refcnt; uint64_t pool; spa_aux_vdev_t *sav = &spa->spa_spares; for (i = 0; i < sav->sav_count; i++) { if (spa_spare_exists(sav->sav_vdevs[i]->vdev_guid, &pool, &refcnt) && pool != 0ULL && pool == spa_guid(spa) && refcnt > 2) return (B_TRUE); } return (B_FALSE); } /* * Post a FM_EREPORT_ZFS_* event from sys/fm/fs/zfs.h. The payload will be * filled in from the spa and (optionally) the vdev. This doesn't do anything * in the userland libzpool, as we don't want consumers to misinterpret ztest * or zdb as real changes. */ void spa_event_notify(spa_t *spa, vdev_t *vd, const char *name) { #ifdef _KERNEL zfs_ereport_post(name, spa, vd, NULL, 0, 0); #endif } #if defined(_KERNEL) && defined(HAVE_SPL) /* state manipulation functions */ EXPORT_SYMBOL(spa_open); EXPORT_SYMBOL(spa_open_rewind); EXPORT_SYMBOL(spa_get_stats); EXPORT_SYMBOL(spa_create); EXPORT_SYMBOL(spa_import_rootpool); EXPORT_SYMBOL(spa_import); EXPORT_SYMBOL(spa_tryimport); EXPORT_SYMBOL(spa_destroy); EXPORT_SYMBOL(spa_export); EXPORT_SYMBOL(spa_reset); EXPORT_SYMBOL(spa_async_request); EXPORT_SYMBOL(spa_async_suspend); EXPORT_SYMBOL(spa_async_resume); EXPORT_SYMBOL(spa_inject_addref); EXPORT_SYMBOL(spa_inject_delref); EXPORT_SYMBOL(spa_scan_stat_init); EXPORT_SYMBOL(spa_scan_get_stats); /* device maniion */ EXPORT_SYMBOL(spa_vdev_add); EXPORT_SYMBOL(spa_vdev_attach); EXPORT_SYMBOL(spa_vdev_detach); EXPORT_SYMBOL(spa_vdev_remove); EXPORT_SYMBOL(spa_vdev_setpath); EXPORT_SYMBOL(spa_vdev_setfru); EXPORT_SYMBOL(spa_vdev_split_mirror); /* spare statech is global across all pools) */ EXPORT_SYMBOL(spa_spare_add); EXPORT_SYMBOL(spa_spare_remove); EXPORT_SYMBOL(spa_spare_exists); EXPORT_SYMBOL(spa_spare_activate); /* L2ARC statech is global across all pools) */ EXPORT_SYMBOL(spa_l2cache_add); EXPORT_SYMBOL(spa_l2cache_remove); EXPORT_SYMBOL(spa_l2cache_exists); EXPORT_SYMBOL(spa_l2cache_activate); EXPORT_SYMBOL(spa_l2cache_drop); /* scanning */ EXPORT_SYMBOL(spa_scan); EXPORT_SYMBOL(spa_scan_stop); /* spa syncing */ EXPORT_SYMBOL(spa_sync); /* only for DMU use */ EXPORT_SYMBOL(spa_sync_allpools); /* properties */ EXPORT_SYMBOL(spa_prop_set); EXPORT_SYMBOL(spa_prop_get); EXPORT_SYMBOL(spa_prop_clear_bootfs); /* asynchronous event notification */ EXPORT_SYMBOL(spa_event_notify); #endif diff --git a/module/zfs/zfs_debug.c b/module/zfs/zfs_debug.c index 4f612e16ba8b..47b7834f58c5 100644 --- a/module/zfs/zfs_debug.c +++ b/module/zfs/zfs_debug.c @@ -1,166 +1,201 @@ /* * 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) 2010, Oracle and/or its affiliates. All rights reserved. * Copyright (c) 2013 by Delphix. All rights reserved. */ #include #if !defined(_KERNEL) || !defined(__linux__) list_t zfs_dbgmsgs; int zfs_dbgmsg_size; kmutex_t zfs_dbgmsgs_lock; -int zfs_dbgmsg_maxsize = 1<<20; /* 1MB */ +int zfs_dbgmsg_maxsize = 4<<20; /* 4MB */ #endif /* * Enable various debugging features. */ int zfs_flags = 0; /* * zfs_recover can be set to nonzero to attempt to recover from * otherwise-fatal errors, typically caused by on-disk corruption. When * set, calls to zfs_panic_recover() will turn into warning messages. * This should only be used as a last resort, as it typically results * in leaked space, or worse. */ -int zfs_recover = 0; +int zfs_recover = B_FALSE; + +/* + * If destroy encounters an EIO while reading metadata (e.g. indirect + * blocks), space referenced by the missing metadata can not be freed. + * Normally this causes the background destroy to become "stalled", as + * it is unable to make forward progress. While in this stalled state, + * all remaining space to free from the error-encountering filesystem is + * "temporarily leaked". Set this flag to cause it to ignore the EIO, + * permanently leak the space from indirect blocks that can not be read, + * and continue to free everything else that it can. + * + * The default, "stalling" behavior is useful if the storage partially + * fails (i.e. some but not all i/os fail), and then later recovers. In + * this case, we will be able to continue pool operations while it is + * partially failed, and when it recovers, we can continue to free the + * space, with no leaks. However, note that this case is actually + * fairly rare. + * + * Typically pools either (a) fail completely (but perhaps temporarily, + * e.g. a top-level vdev going offline), or (b) have localized, + * permanent errors (e.g. disk returns the wrong data due to bit flip or + * firmware bug). In case (a), this setting does not matter because the + * pool will be suspended and the sync thread will not be able to make + * forward progress regardless. In case (b), because the error is + * permanent, the best we can do is leak the minimum amount of space, + * which is what setting this flag will do. Therefore, it is reasonable + * for this flag to normally be set, but we chose the more conservative + * approach of not setting it, so that there is no possibility of + * leaking space in the "partial temporary" failure case. + */ +int zfs_free_leak_on_eio = B_FALSE; void zfs_panic_recover(const char *fmt, ...) { va_list adx; va_start(adx, fmt); vcmn_err(zfs_recover ? CE_WARN : CE_PANIC, fmt, adx); va_end(adx); } /* * Debug logging is enabled by default for production kernel builds. * The overhead for this is negligible and the logs can be valuable when * debugging. For non-production user space builds all debugging except * logging is enabled since performance is no longer a concern. */ void zfs_dbgmsg_init(void) { #if !defined(_KERNEL) || !defined(__linux__) list_create(&zfs_dbgmsgs, sizeof (zfs_dbgmsg_t), offsetof(zfs_dbgmsg_t, zdm_node)); mutex_init(&zfs_dbgmsgs_lock, NULL, MUTEX_DEFAULT, NULL); #endif if (zfs_flags == 0) { #if defined(_KERNEL) zfs_flags = ZFS_DEBUG_DPRINTF; spl_debug_set_mask(spl_debug_get_mask() | SD_DPRINTF); spl_debug_set_subsys(spl_debug_get_subsys() | SS_USER1); #else zfs_flags = ~ZFS_DEBUG_DPRINTF; #endif /* _KERNEL */ } } void zfs_dbgmsg_fini(void) { #if !defined(_KERNEL) || !defined(__linux__) zfs_dbgmsg_t *zdm; while ((zdm = list_remove_head(&zfs_dbgmsgs)) != NULL) { int size = sizeof (zfs_dbgmsg_t) + strlen(zdm->zdm_msg); kmem_free(zdm, size); zfs_dbgmsg_size -= size; } mutex_destroy(&zfs_dbgmsgs_lock); ASSERT0(zfs_dbgmsg_size); #endif } #if !defined(_KERNEL) || !defined(__linux__) /* * Print these messages by running: * echo ::zfs_dbgmsg | mdb -k * * Monitor these messages by running: * dtrace -q -n 'zfs-dbgmsg{printf("%s\n", stringof(arg0))}' */ void zfs_dbgmsg(const char *fmt, ...) { int size; va_list adx; zfs_dbgmsg_t *zdm; va_start(adx, fmt); size = vsnprintf(NULL, 0, fmt, adx); va_end(adx); /* * There is one byte of string in sizeof (zfs_dbgmsg_t), used * for the terminating null. */ zdm = kmem_alloc(sizeof (zfs_dbgmsg_t) + size, KM_SLEEP); zdm->zdm_timestamp = gethrestime_sec(); va_start(adx, fmt); (void) vsnprintf(zdm->zdm_msg, size + 1, fmt, adx); va_end(adx); DTRACE_PROBE1(zfs__dbgmsg, char *, zdm->zdm_msg); mutex_enter(&zfs_dbgmsgs_lock); list_insert_tail(&zfs_dbgmsgs, zdm); zfs_dbgmsg_size += sizeof (zfs_dbgmsg_t) + size; while (zfs_dbgmsg_size > zfs_dbgmsg_maxsize) { zdm = list_remove_head(&zfs_dbgmsgs); size = sizeof (zfs_dbgmsg_t) + strlen(zdm->zdm_msg); kmem_free(zdm, size); zfs_dbgmsg_size -= size; } mutex_exit(&zfs_dbgmsgs_lock); } void zfs_dbgmsg_print(const char *tag) { zfs_dbgmsg_t *zdm; (void) printf("ZFS_DBGMSG(%s):\n", tag); mutex_enter(&zfs_dbgmsgs_lock); for (zdm = list_head(&zfs_dbgmsgs); zdm; zdm = list_next(&zfs_dbgmsgs, zdm)) (void) printf("%s\n", zdm->zdm_msg); mutex_exit(&zfs_dbgmsgs_lock); } #endif #if defined(_KERNEL) module_param(zfs_flags, int, 0644); MODULE_PARM_DESC(zfs_flags, "Set additional debugging flags"); module_param(zfs_recover, int, 0644); MODULE_PARM_DESC(zfs_recover, "Set to attempt to recover from fatal errors"); + +module_param(zfs_free_leak_on_eio, int, 0644); +MODULE_PARM_DESC(zfs_free_leak_on_eio, + "Set to ignore IO errors during free and permanently leak the space"); #endif /* _KERNEL */ diff --git a/module/zfs/zio.c b/module/zfs/zio.c index ad97ef5dbd02..58d4550fbf79 100644 --- a/module/zfs/zio.c +++ b/module/zfs/zio.c @@ -1,3422 +1,3415 @@ /* * 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) 2011, 2014 by Delphix. All rights reserved. * Copyright (c) 2011 Nexenta Systems, Inc. All rights reserved. */ #include #include #include #include #include #include #include #include #include #include #include #include #include #include /* * ========================================================================== * I/O type descriptions * ========================================================================== */ const char *zio_type_name[ZIO_TYPES] = { "z_null", "z_rd", "z_wr", "z_fr", "z_cl", "z_ioctl" }; /* * ========================================================================== * I/O kmem caches * ========================================================================== */ kmem_cache_t *zio_cache; kmem_cache_t *zio_link_cache; kmem_cache_t *zio_vdev_cache; kmem_cache_t *zio_buf_cache[SPA_MAXBLOCKSIZE >> SPA_MINBLOCKSHIFT]; kmem_cache_t *zio_data_buf_cache[SPA_MAXBLOCKSIZE >> SPA_MINBLOCKSHIFT]; int zio_bulk_flags = 0; int zio_delay_max = ZIO_DELAY_MAX; /* * The following actions directly effect the spa's sync-to-convergence logic. * The values below define the sync pass when we start performing the action. * Care should be taken when changing these values as they directly impact * spa_sync() performance. Tuning these values may introduce subtle performance * pathologies and should only be done in the context of performance analysis. * These tunables will eventually be removed and replaced with #defines once * enough analysis has been done to determine optimal values. * * The 'zfs_sync_pass_deferred_free' pass must be greater than 1 to ensure that * regular blocks are not deferred. */ int zfs_sync_pass_deferred_free = 2; /* defer frees starting in this pass */ int zfs_sync_pass_dont_compress = 5; /* don't compress starting in this pass */ int zfs_sync_pass_rewrite = 2; /* rewrite new bps starting in this pass */ /* * An allocating zio is one that either currently has the DVA allocate * stage set or will have it later in its lifetime. */ #define IO_IS_ALLOCATING(zio) ((zio)->io_orig_pipeline & ZIO_STAGE_DVA_ALLOCATE) int zio_requeue_io_start_cut_in_line = 1; #ifdef ZFS_DEBUG int zio_buf_debug_limit = 16384; #else int zio_buf_debug_limit = 0; #endif static inline void __zio_execute(zio_t *zio); static int zio_cons(void *arg, void *unused, int kmflag) { zio_t *zio = arg; bzero(zio, sizeof (zio_t)); mutex_init(&zio->io_lock, NULL, MUTEX_DEFAULT, NULL); cv_init(&zio->io_cv, NULL, CV_DEFAULT, NULL); list_create(&zio->io_parent_list, sizeof (zio_link_t), offsetof(zio_link_t, zl_parent_node)); list_create(&zio->io_child_list, sizeof (zio_link_t), offsetof(zio_link_t, zl_child_node)); return (0); } static void zio_dest(void *arg, void *unused) { zio_t *zio = arg; mutex_destroy(&zio->io_lock); cv_destroy(&zio->io_cv); list_destroy(&zio->io_parent_list); list_destroy(&zio->io_child_list); } void zio_init(void) { size_t c; vmem_t *data_alloc_arena = NULL; zio_cache = kmem_cache_create("zio_cache", sizeof (zio_t), 0, zio_cons, zio_dest, NULL, NULL, NULL, 0); zio_link_cache = kmem_cache_create("zio_link_cache", sizeof (zio_link_t), 0, NULL, NULL, NULL, NULL, NULL, 0); zio_vdev_cache = kmem_cache_create("zio_vdev_cache", sizeof (vdev_io_t), PAGESIZE, NULL, NULL, NULL, NULL, NULL, 0); /* * For small buffers, we want a cache for each multiple of * SPA_MINBLOCKSIZE. For medium-size buffers, we want a cache * for each quarter-power of 2. For large buffers, we want * a cache for each multiple of PAGESIZE. */ for (c = 0; c < SPA_MAXBLOCKSIZE >> SPA_MINBLOCKSHIFT; c++) { size_t size = (c + 1) << SPA_MINBLOCKSHIFT; size_t p2 = size; size_t align = 0; while (p2 & (p2 - 1)) p2 &= p2 - 1; #ifndef _KERNEL /* * If we are using watchpoints, put each buffer on its own page, * to eliminate the performance overhead of trapping to the * kernel when modifying a non-watched buffer that shares the * page with a watched buffer. */ if (arc_watch && !IS_P2ALIGNED(size, PAGESIZE)) continue; #endif if (size <= 4 * SPA_MINBLOCKSIZE) { align = SPA_MINBLOCKSIZE; } else if (IS_P2ALIGNED(size, PAGESIZE)) { align = PAGESIZE; } else if (IS_P2ALIGNED(size, p2 >> 2)) { align = p2 >> 2; } if (align != 0) { char name[36]; int flags = zio_bulk_flags; (void) sprintf(name, "zio_buf_%lu", (ulong_t)size); zio_buf_cache[c] = kmem_cache_create(name, size, align, NULL, NULL, NULL, NULL, NULL, flags); (void) sprintf(name, "zio_data_buf_%lu", (ulong_t)size); zio_data_buf_cache[c] = kmem_cache_create(name, size, align, NULL, NULL, NULL, NULL, data_alloc_arena, flags); } } while (--c != 0) { ASSERT(zio_buf_cache[c] != NULL); if (zio_buf_cache[c - 1] == NULL) zio_buf_cache[c - 1] = zio_buf_cache[c]; ASSERT(zio_data_buf_cache[c] != NULL); if (zio_data_buf_cache[c - 1] == NULL) zio_data_buf_cache[c - 1] = zio_data_buf_cache[c]; } zio_inject_init(); lz4_init(); } void zio_fini(void) { size_t c; kmem_cache_t *last_cache = NULL; kmem_cache_t *last_data_cache = NULL; for (c = 0; c < SPA_MAXBLOCKSIZE >> SPA_MINBLOCKSHIFT; c++) { if (zio_buf_cache[c] != last_cache) { last_cache = zio_buf_cache[c]; kmem_cache_destroy(zio_buf_cache[c]); } zio_buf_cache[c] = NULL; if (zio_data_buf_cache[c] != last_data_cache) { last_data_cache = zio_data_buf_cache[c]; kmem_cache_destroy(zio_data_buf_cache[c]); } zio_data_buf_cache[c] = NULL; } kmem_cache_destroy(zio_vdev_cache); kmem_cache_destroy(zio_link_cache); kmem_cache_destroy(zio_cache); zio_inject_fini(); lz4_fini(); } /* * ========================================================================== * Allocate and free I/O buffers * ========================================================================== */ /* * Use zio_buf_alloc to allocate ZFS metadata. This data will appear in a * crashdump if the kernel panics, so use it judiciously. Obviously, it's * useful to inspect ZFS metadata, but if possible, we should avoid keeping * excess / transient data in-core during a crashdump. */ void * zio_buf_alloc(size_t size) { size_t c = (size - 1) >> SPA_MINBLOCKSHIFT; ASSERT3U(c, <, SPA_MAXBLOCKSIZE >> SPA_MINBLOCKSHIFT); return (kmem_cache_alloc(zio_buf_cache[c], KM_PUSHPAGE | KM_NODEBUG)); } /* * Use zio_data_buf_alloc to allocate data. The data will not appear in a * crashdump if the kernel panics. This exists so that we will limit the amount * of ZFS data that shows up in a kernel crashdump. (Thus reducing the amount * of kernel heap dumped to disk when the kernel panics) */ void * zio_data_buf_alloc(size_t size) { size_t c = (size - 1) >> SPA_MINBLOCKSHIFT; ASSERT(c < SPA_MAXBLOCKSIZE >> SPA_MINBLOCKSHIFT); return (kmem_cache_alloc(zio_data_buf_cache[c], KM_PUSHPAGE | KM_NODEBUG)); } void zio_buf_free(void *buf, size_t size) { size_t c = (size - 1) >> SPA_MINBLOCKSHIFT; ASSERT(c < SPA_MAXBLOCKSIZE >> SPA_MINBLOCKSHIFT); kmem_cache_free(zio_buf_cache[c], buf); } void zio_data_buf_free(void *buf, size_t size) { size_t c = (size - 1) >> SPA_MINBLOCKSHIFT; ASSERT(c < SPA_MAXBLOCKSIZE >> SPA_MINBLOCKSHIFT); kmem_cache_free(zio_data_buf_cache[c], buf); } /* * Dedicated I/O buffers to ensure that memory fragmentation never prevents * or significantly delays the issuing of a zio. These buffers are used * to aggregate I/O and could be used for raidz stripes. */ void * zio_vdev_alloc(void) { return (kmem_cache_alloc(zio_vdev_cache, KM_PUSHPAGE)); } void zio_vdev_free(void *buf) { kmem_cache_free(zio_vdev_cache, buf); } /* * ========================================================================== * Push and pop I/O transform buffers * ========================================================================== */ static void zio_push_transform(zio_t *zio, void *data, uint64_t size, uint64_t bufsize, zio_transform_func_t *transform) { zio_transform_t *zt = kmem_alloc(sizeof (zio_transform_t), KM_PUSHPAGE); zt->zt_orig_data = zio->io_data; zt->zt_orig_size = zio->io_size; zt->zt_bufsize = bufsize; zt->zt_transform = transform; zt->zt_next = zio->io_transform_stack; zio->io_transform_stack = zt; zio->io_data = data; zio->io_size = size; } static void zio_pop_transforms(zio_t *zio) { zio_transform_t *zt; while ((zt = zio->io_transform_stack) != NULL) { if (zt->zt_transform != NULL) zt->zt_transform(zio, zt->zt_orig_data, zt->zt_orig_size); if (zt->zt_bufsize != 0) zio_buf_free(zio->io_data, zt->zt_bufsize); zio->io_data = zt->zt_orig_data; zio->io_size = zt->zt_orig_size; zio->io_transform_stack = zt->zt_next; kmem_free(zt, sizeof (zio_transform_t)); } } /* * ========================================================================== * I/O transform callbacks for subblocks and decompression * ========================================================================== */ static void zio_subblock(zio_t *zio, void *data, uint64_t size) { ASSERT(zio->io_size > size); if (zio->io_type == ZIO_TYPE_READ) bcopy(zio->io_data, data, size); } static void zio_decompress(zio_t *zio, void *data, uint64_t size) { if (zio->io_error == 0 && zio_decompress_data(BP_GET_COMPRESS(zio->io_bp), zio->io_data, data, zio->io_size, size) != 0) zio->io_error = SET_ERROR(EIO); } /* * ========================================================================== * I/O parent/child relationships and pipeline interlocks * ========================================================================== */ /* * NOTE - Callers to zio_walk_parents() and zio_walk_children must * continue calling these functions until they return NULL. * Otherwise, the next caller will pick up the list walk in * some indeterminate state. (Otherwise every caller would * have to pass in a cookie to keep the state represented by * io_walk_link, which gets annoying.) */ zio_t * zio_walk_parents(zio_t *cio) { zio_link_t *zl = cio->io_walk_link; list_t *pl = &cio->io_parent_list; zl = (zl == NULL) ? list_head(pl) : list_next(pl, zl); cio->io_walk_link = zl; if (zl == NULL) return (NULL); ASSERT(zl->zl_child == cio); return (zl->zl_parent); } zio_t * zio_walk_children(zio_t *pio) { zio_link_t *zl = pio->io_walk_link; list_t *cl = &pio->io_child_list; zl = (zl == NULL) ? list_head(cl) : list_next(cl, zl); pio->io_walk_link = zl; if (zl == NULL) return (NULL); ASSERT(zl->zl_parent == pio); return (zl->zl_child); } zio_t * zio_unique_parent(zio_t *cio) { zio_t *pio = zio_walk_parents(cio); VERIFY(zio_walk_parents(cio) == NULL); return (pio); } void zio_add_child(zio_t *pio, zio_t *cio) { zio_link_t *zl = kmem_cache_alloc(zio_link_cache, KM_PUSHPAGE); int w; /* * Logical I/Os can have logical, gang, or vdev children. * Gang I/Os can have gang or vdev children. * Vdev I/Os can only have vdev children. * The following ASSERT captures all of these constraints. */ ASSERT(cio->io_child_type <= pio->io_child_type); zl->zl_parent = pio; zl->zl_child = cio; mutex_enter(&cio->io_lock); mutex_enter(&pio->io_lock); ASSERT(pio->io_state[ZIO_WAIT_DONE] == 0); for (w = 0; w < ZIO_WAIT_TYPES; w++) pio->io_children[cio->io_child_type][w] += !cio->io_state[w]; list_insert_head(&pio->io_child_list, zl); list_insert_head(&cio->io_parent_list, zl); pio->io_child_count++; cio->io_parent_count++; mutex_exit(&pio->io_lock); mutex_exit(&cio->io_lock); } static void zio_remove_child(zio_t *pio, zio_t *cio, zio_link_t *zl) { ASSERT(zl->zl_parent == pio); ASSERT(zl->zl_child == cio); mutex_enter(&cio->io_lock); mutex_enter(&pio->io_lock); list_remove(&pio->io_child_list, zl); list_remove(&cio->io_parent_list, zl); pio->io_child_count--; cio->io_parent_count--; mutex_exit(&pio->io_lock); mutex_exit(&cio->io_lock); kmem_cache_free(zio_link_cache, zl); } static boolean_t zio_wait_for_children(zio_t *zio, enum zio_child child, enum zio_wait_type wait) { uint64_t *countp = &zio->io_children[child][wait]; boolean_t waiting = B_FALSE; mutex_enter(&zio->io_lock); ASSERT(zio->io_stall == NULL); if (*countp != 0) { zio->io_stage >>= 1; zio->io_stall = countp; waiting = B_TRUE; } mutex_exit(&zio->io_lock); return (waiting); } __attribute__((always_inline)) static inline void zio_notify_parent(zio_t *pio, zio_t *zio, enum zio_wait_type wait) { uint64_t *countp = &pio->io_children[zio->io_child_type][wait]; int *errorp = &pio->io_child_error[zio->io_child_type]; mutex_enter(&pio->io_lock); if (zio->io_error && !(zio->io_flags & ZIO_FLAG_DONT_PROPAGATE)) *errorp = zio_worst_error(*errorp, zio->io_error); pio->io_reexecute |= zio->io_reexecute; ASSERT3U(*countp, >, 0); (*countp)--; if (*countp == 0 && pio->io_stall == countp) { pio->io_stall = NULL; mutex_exit(&pio->io_lock); __zio_execute(pio); } else { mutex_exit(&pio->io_lock); } } static void zio_inherit_child_errors(zio_t *zio, enum zio_child c) { if (zio->io_child_error[c] != 0 && zio->io_error == 0) zio->io_error = zio->io_child_error[c]; } /* * ========================================================================== * Create the various types of I/O (read, write, free, etc) * ========================================================================== */ static zio_t * zio_create(zio_t *pio, spa_t *spa, uint64_t txg, const blkptr_t *bp, void *data, uint64_t size, zio_done_func_t *done, void *private, zio_type_t type, zio_priority_t priority, enum zio_flag flags, vdev_t *vd, uint64_t offset, const zbookmark_t *zb, enum zio_stage stage, enum zio_stage pipeline) { zio_t *zio; ASSERT3U(size, <=, SPA_MAXBLOCKSIZE); ASSERT(P2PHASE(size, SPA_MINBLOCKSIZE) == 0); ASSERT(P2PHASE(offset, SPA_MINBLOCKSIZE) == 0); ASSERT(!vd || spa_config_held(spa, SCL_STATE_ALL, RW_READER)); ASSERT(!bp || !(flags & ZIO_FLAG_CONFIG_WRITER)); ASSERT(vd || stage == ZIO_STAGE_OPEN); zio = kmem_cache_alloc(zio_cache, KM_PUSHPAGE); if (vd != NULL) zio->io_child_type = ZIO_CHILD_VDEV; else if (flags & ZIO_FLAG_GANG_CHILD) zio->io_child_type = ZIO_CHILD_GANG; else if (flags & ZIO_FLAG_DDT_CHILD) zio->io_child_type = ZIO_CHILD_DDT; else zio->io_child_type = ZIO_CHILD_LOGICAL; if (bp != NULL) { zio->io_logical = NULL; zio->io_bp = (blkptr_t *)bp; zio->io_bp_copy = *bp; zio->io_bp_orig = *bp; if (type != ZIO_TYPE_WRITE || zio->io_child_type == ZIO_CHILD_DDT) zio->io_bp = &zio->io_bp_copy; /* so caller can free */ if (zio->io_child_type == ZIO_CHILD_LOGICAL) zio->io_logical = zio; if (zio->io_child_type > ZIO_CHILD_GANG && BP_IS_GANG(bp)) pipeline |= ZIO_GANG_STAGES; } else { zio->io_logical = NULL; zio->io_bp = NULL; bzero(&zio->io_bp_copy, sizeof (blkptr_t)); bzero(&zio->io_bp_orig, sizeof (blkptr_t)); } zio->io_spa = spa; zio->io_txg = txg; zio->io_ready = NULL; zio->io_physdone = NULL; zio->io_done = done; zio->io_private = private; zio->io_prev_space_delta = 0; zio->io_type = type; zio->io_priority = priority; zio->io_vd = vd; zio->io_vsd = NULL; zio->io_vsd_ops = NULL; zio->io_offset = offset; zio->io_timestamp = 0; zio->io_delta = 0; zio->io_delay = 0; zio->io_orig_data = zio->io_data = data; zio->io_orig_size = zio->io_size = size; zio->io_orig_flags = zio->io_flags = flags; zio->io_orig_stage = zio->io_stage = stage; zio->io_orig_pipeline = zio->io_pipeline = pipeline; bzero(&zio->io_prop, sizeof (zio_prop_t)); zio->io_cmd = 0; zio->io_reexecute = 0; zio->io_bp_override = NULL; zio->io_walk_link = NULL; zio->io_transform_stack = NULL; zio->io_error = 0; zio->io_child_count = 0; zio->io_phys_children = 0; zio->io_parent_count = 0; zio->io_stall = NULL; zio->io_gang_leader = NULL; zio->io_gang_tree = NULL; zio->io_executor = NULL; zio->io_waiter = NULL; zio->io_cksum_report = NULL; zio->io_ena = 0; bzero(zio->io_child_error, sizeof (int) * ZIO_CHILD_TYPES); bzero(zio->io_children, sizeof (uint64_t) * ZIO_CHILD_TYPES * ZIO_WAIT_TYPES); bzero(&zio->io_bookmark, sizeof (zbookmark_t)); zio->io_state[ZIO_WAIT_READY] = (stage >= ZIO_STAGE_READY); zio->io_state[ZIO_WAIT_DONE] = (stage >= ZIO_STAGE_DONE); if (zb != NULL) zio->io_bookmark = *zb; if (pio != NULL) { if (zio->io_logical == NULL) zio->io_logical = pio->io_logical; if (zio->io_child_type == ZIO_CHILD_GANG) zio->io_gang_leader = pio->io_gang_leader; zio_add_child(pio, zio); } taskq_init_ent(&zio->io_tqent); return (zio); } static void zio_destroy(zio_t *zio) { kmem_cache_free(zio_cache, zio); } zio_t * zio_null(zio_t *pio, spa_t *spa, vdev_t *vd, zio_done_func_t *done, void *private, enum zio_flag flags) { zio_t *zio; zio = zio_create(pio, spa, 0, NULL, NULL, 0, done, private, ZIO_TYPE_NULL, ZIO_PRIORITY_NOW, flags, vd, 0, NULL, ZIO_STAGE_OPEN, ZIO_INTERLOCK_PIPELINE); return (zio); } zio_t * zio_root(spa_t *spa, zio_done_func_t *done, void *private, enum zio_flag flags) { return (zio_null(NULL, spa, NULL, done, private, flags)); } zio_t * zio_read(zio_t *pio, spa_t *spa, const blkptr_t *bp, void *data, uint64_t size, zio_done_func_t *done, void *private, zio_priority_t priority, enum zio_flag flags, const zbookmark_t *zb) { zio_t *zio; zio = zio_create(pio, spa, BP_PHYSICAL_BIRTH(bp), bp, data, size, done, private, ZIO_TYPE_READ, priority, flags, NULL, 0, zb, ZIO_STAGE_OPEN, (flags & ZIO_FLAG_DDT_CHILD) ? ZIO_DDT_CHILD_READ_PIPELINE : ZIO_READ_PIPELINE); return (zio); } zio_t * zio_write(zio_t *pio, spa_t *spa, uint64_t txg, blkptr_t *bp, void *data, uint64_t size, const zio_prop_t *zp, zio_done_func_t *ready, zio_done_func_t *physdone, zio_done_func_t *done, void *private, zio_priority_t priority, enum zio_flag flags, const zbookmark_t *zb) { zio_t *zio; ASSERT(zp->zp_checksum >= ZIO_CHECKSUM_OFF && zp->zp_checksum < ZIO_CHECKSUM_FUNCTIONS && zp->zp_compress >= ZIO_COMPRESS_OFF && zp->zp_compress < ZIO_COMPRESS_FUNCTIONS && DMU_OT_IS_VALID(zp->zp_type) && zp->zp_level < 32 && zp->zp_copies > 0 && zp->zp_copies <= spa_max_replication(spa)); zio = zio_create(pio, spa, txg, bp, data, size, done, private, ZIO_TYPE_WRITE, priority, flags, NULL, 0, zb, ZIO_STAGE_OPEN, (flags & ZIO_FLAG_DDT_CHILD) ? ZIO_DDT_CHILD_WRITE_PIPELINE : ZIO_WRITE_PIPELINE); zio->io_ready = ready; zio->io_physdone = physdone; zio->io_prop = *zp; /* * Data can be NULL if we are going to call zio_write_override() to * provide the already-allocated BP. But we may need the data to * verify a dedup hit (if requested). In this case, don't try to * dedup (just take the already-allocated BP verbatim). */ if (data == NULL && zio->io_prop.zp_dedup_verify) { zio->io_prop.zp_dedup = zio->io_prop.zp_dedup_verify = B_FALSE; } return (zio); } zio_t * zio_rewrite(zio_t *pio, spa_t *spa, uint64_t txg, blkptr_t *bp, void *data, uint64_t size, zio_done_func_t *done, void *private, zio_priority_t priority, enum zio_flag flags, zbookmark_t *zb) { zio_t *zio; zio = zio_create(pio, spa, txg, bp, data, size, done, private, ZIO_TYPE_WRITE, priority, flags, NULL, 0, zb, ZIO_STAGE_OPEN, ZIO_REWRITE_PIPELINE); return (zio); } void zio_write_override(zio_t *zio, blkptr_t *bp, int copies, boolean_t nopwrite) { ASSERT(zio->io_type == ZIO_TYPE_WRITE); ASSERT(zio->io_child_type == ZIO_CHILD_LOGICAL); ASSERT(zio->io_stage == ZIO_STAGE_OPEN); ASSERT(zio->io_txg == spa_syncing_txg(zio->io_spa)); /* * We must reset the io_prop to match the values that existed * when the bp was first written by dmu_sync() keeping in mind * that nopwrite and dedup are mutually exclusive. */ zio->io_prop.zp_dedup = nopwrite ? B_FALSE : zio->io_prop.zp_dedup; zio->io_prop.zp_nopwrite = nopwrite; zio->io_prop.zp_copies = copies; zio->io_bp_override = bp; } void zio_free(spa_t *spa, uint64_t txg, const blkptr_t *bp) { /* * The check for EMBEDDED is a performance optimization. We * process the free here (by ignoring it) rather than * putting it on the list and then processing it in zio_free_sync(). */ if (BP_IS_EMBEDDED(bp)) return; metaslab_check_free(spa, bp); /* * Frees that are for the currently-syncing txg, are not going to be * deferred, and which will not need to do a read (i.e. not GANG or * DEDUP), can be processed immediately. Otherwise, put them on the * in-memory list for later processing. */ if (BP_IS_GANG(bp) || BP_GET_DEDUP(bp) || txg != spa->spa_syncing_txg || spa_sync_pass(spa) >= zfs_sync_pass_deferred_free) { bplist_append(&spa->spa_free_bplist[txg & TXG_MASK], bp); } else { VERIFY0(zio_wait(zio_free_sync(NULL, spa, txg, bp, 0))); } } zio_t * zio_free_sync(zio_t *pio, spa_t *spa, uint64_t txg, const blkptr_t *bp, enum zio_flag flags) { zio_t *zio; enum zio_stage stage = ZIO_FREE_PIPELINE; ASSERT(!BP_IS_HOLE(bp)); ASSERT(spa_syncing_txg(spa) == txg); ASSERT(spa_sync_pass(spa) < zfs_sync_pass_deferred_free); if (BP_IS_EMBEDDED(bp)) return (zio_null(pio, spa, NULL, NULL, NULL, 0)); metaslab_check_free(spa, bp); arc_freed(spa, bp); /* * GANG and DEDUP blocks can induce a read (for the gang block header, * or the DDT), so issue them asynchronously so that this thread is * not tied up. */ if (BP_IS_GANG(bp) || BP_GET_DEDUP(bp)) stage |= ZIO_STAGE_ISSUE_ASYNC; zio = zio_create(pio, spa, txg, bp, NULL, BP_GET_PSIZE(bp), NULL, NULL, ZIO_TYPE_FREE, ZIO_PRIORITY_NOW, flags, NULL, 0, NULL, ZIO_STAGE_OPEN, stage); return (zio); } zio_t * zio_claim(zio_t *pio, spa_t *spa, uint64_t txg, const blkptr_t *bp, zio_done_func_t *done, void *private, enum zio_flag flags) { zio_t *zio; dprintf_bp(bp, "claiming in txg %llu", txg); if (BP_IS_EMBEDDED(bp)) return (zio_null(pio, spa, NULL, NULL, NULL, 0)); /* * A claim is an allocation of a specific block. Claims are needed * to support immediate writes in the intent log. The issue is that * immediate writes contain committed data, but in a txg that was * *not* committed. Upon opening the pool after an unclean shutdown, * the intent log claims all blocks that contain immediate write data * so that the SPA knows they're in use. * * All claims *must* be resolved in the first txg -- before the SPA * starts allocating blocks -- so that nothing is allocated twice. * If txg == 0 we just verify that the block is claimable. */ ASSERT3U(spa->spa_uberblock.ub_rootbp.blk_birth, <, spa_first_txg(spa)); ASSERT(txg == spa_first_txg(spa) || txg == 0); ASSERT(!BP_GET_DEDUP(bp) || !spa_writeable(spa)); /* zdb(1M) */ zio = zio_create(pio, spa, txg, bp, NULL, BP_GET_PSIZE(bp), done, private, ZIO_TYPE_CLAIM, ZIO_PRIORITY_NOW, flags, NULL, 0, NULL, ZIO_STAGE_OPEN, ZIO_CLAIM_PIPELINE); return (zio); } zio_t * zio_ioctl(zio_t *pio, spa_t *spa, vdev_t *vd, int cmd, zio_done_func_t *done, void *private, enum zio_flag flags) { zio_t *zio; int c; if (vd->vdev_children == 0) { zio = zio_create(pio, spa, 0, NULL, NULL, 0, done, private, ZIO_TYPE_IOCTL, ZIO_PRIORITY_NOW, flags, vd, 0, NULL, ZIO_STAGE_OPEN, ZIO_IOCTL_PIPELINE); zio->io_cmd = cmd; } else { zio = zio_null(pio, spa, NULL, NULL, NULL, flags); for (c = 0; c < vd->vdev_children; c++) zio_nowait(zio_ioctl(zio, spa, vd->vdev_child[c], cmd, done, private, flags)); } return (zio); } zio_t * zio_read_phys(zio_t *pio, vdev_t *vd, uint64_t offset, uint64_t size, void *data, int checksum, zio_done_func_t *done, void *private, zio_priority_t priority, enum zio_flag flags, boolean_t labels) { zio_t *zio; ASSERT(vd->vdev_children == 0); ASSERT(!labels || offset + size <= VDEV_LABEL_START_SIZE || offset >= vd->vdev_psize - VDEV_LABEL_END_SIZE); ASSERT3U(offset + size, <=, vd->vdev_psize); zio = zio_create(pio, vd->vdev_spa, 0, NULL, data, size, done, private, ZIO_TYPE_READ, priority, flags, vd, offset, NULL, ZIO_STAGE_OPEN, ZIO_READ_PHYS_PIPELINE); zio->io_prop.zp_checksum = checksum; return (zio); } zio_t * zio_write_phys(zio_t *pio, vdev_t *vd, uint64_t offset, uint64_t size, void *data, int checksum, zio_done_func_t *done, void *private, zio_priority_t priority, enum zio_flag flags, boolean_t labels) { zio_t *zio; ASSERT(vd->vdev_children == 0); ASSERT(!labels || offset + size <= VDEV_LABEL_START_SIZE || offset >= vd->vdev_psize - VDEV_LABEL_END_SIZE); ASSERT3U(offset + size, <=, vd->vdev_psize); zio = zio_create(pio, vd->vdev_spa, 0, NULL, data, size, done, private, ZIO_TYPE_WRITE, priority, flags, vd, offset, NULL, ZIO_STAGE_OPEN, ZIO_WRITE_PHYS_PIPELINE); zio->io_prop.zp_checksum = checksum; if (zio_checksum_table[checksum].ci_eck) { /* * zec checksums are necessarily destructive -- they modify * the end of the write buffer to hold the verifier/checksum. * Therefore, we must make a local copy in case the data is * being written to multiple places in parallel. */ void *wbuf = zio_buf_alloc(size); bcopy(data, wbuf, size); zio_push_transform(zio, wbuf, size, size, NULL); } return (zio); } /* * Create a child I/O to do some work for us. */ zio_t * zio_vdev_child_io(zio_t *pio, blkptr_t *bp, vdev_t *vd, uint64_t offset, void *data, uint64_t size, int type, zio_priority_t priority, enum zio_flag flags, zio_done_func_t *done, void *private) { enum zio_stage pipeline = ZIO_VDEV_CHILD_PIPELINE; zio_t *zio; ASSERT(vd->vdev_parent == (pio->io_vd ? pio->io_vd : pio->io_spa->spa_root_vdev)); if (type == ZIO_TYPE_READ && bp != NULL) { /* * If we have the bp, then the child should perform the * checksum and the parent need not. This pushes error * detection as close to the leaves as possible and * eliminates redundant checksums in the interior nodes. */ pipeline |= ZIO_STAGE_CHECKSUM_VERIFY; pio->io_pipeline &= ~ZIO_STAGE_CHECKSUM_VERIFY; } if (vd->vdev_children == 0) offset += VDEV_LABEL_START_SIZE; flags |= ZIO_VDEV_CHILD_FLAGS(pio) | ZIO_FLAG_DONT_PROPAGATE; /* * If we've decided to do a repair, the write is not speculative -- * even if the original read was. */ if (flags & ZIO_FLAG_IO_REPAIR) flags &= ~ZIO_FLAG_SPECULATIVE; zio = zio_create(pio, pio->io_spa, pio->io_txg, bp, data, size, done, private, type, priority, flags, vd, offset, &pio->io_bookmark, ZIO_STAGE_VDEV_IO_START >> 1, pipeline); zio->io_physdone = pio->io_physdone; if (vd->vdev_ops->vdev_op_leaf && zio->io_logical != NULL) zio->io_logical->io_phys_children++; return (zio); } zio_t * zio_vdev_delegated_io(vdev_t *vd, uint64_t offset, void *data, uint64_t size, int type, zio_priority_t priority, enum zio_flag flags, zio_done_func_t *done, void *private) { zio_t *zio; ASSERT(vd->vdev_ops->vdev_op_leaf); zio = zio_create(NULL, vd->vdev_spa, 0, NULL, data, size, done, private, type, priority, flags | ZIO_FLAG_CANFAIL | ZIO_FLAG_DONT_RETRY | ZIO_FLAG_DELEGATED, vd, offset, NULL, ZIO_STAGE_VDEV_IO_START >> 1, ZIO_VDEV_CHILD_PIPELINE); return (zio); } void zio_flush(zio_t *zio, vdev_t *vd) { zio_nowait(zio_ioctl(zio, zio->io_spa, vd, DKIOCFLUSHWRITECACHE, NULL, NULL, ZIO_FLAG_CANFAIL | ZIO_FLAG_DONT_PROPAGATE | ZIO_FLAG_DONT_RETRY)); } void zio_shrink(zio_t *zio, uint64_t size) { ASSERT(zio->io_executor == NULL); ASSERT(zio->io_orig_size == zio->io_size); ASSERT(size <= zio->io_size); /* * We don't shrink for raidz because of problems with the * reconstruction when reading back less than the block size. * Note, BP_IS_RAIDZ() assumes no compression. */ ASSERT(BP_GET_COMPRESS(zio->io_bp) == ZIO_COMPRESS_OFF); if (!BP_IS_RAIDZ(zio->io_bp)) zio->io_orig_size = zio->io_size = size; } /* * ========================================================================== * Prepare to read and write logical blocks * ========================================================================== */ static int zio_read_bp_init(zio_t *zio) { blkptr_t *bp = zio->io_bp; if (BP_GET_COMPRESS(bp) != ZIO_COMPRESS_OFF && zio->io_child_type == ZIO_CHILD_LOGICAL && !(zio->io_flags & ZIO_FLAG_RAW)) { uint64_t psize = BP_IS_EMBEDDED(bp) ? BPE_GET_PSIZE(bp) : BP_GET_PSIZE(bp); void *cbuf = zio_buf_alloc(psize); zio_push_transform(zio, cbuf, psize, psize, zio_decompress); } if (BP_IS_EMBEDDED(bp) && BPE_GET_ETYPE(bp) == BP_EMBEDDED_TYPE_DATA) { zio->io_pipeline = ZIO_INTERLOCK_PIPELINE; decode_embedded_bp_compressed(bp, zio->io_data); } else { ASSERT(!BP_IS_EMBEDDED(bp)); } if (!DMU_OT_IS_METADATA(BP_GET_TYPE(bp)) && BP_GET_LEVEL(bp) == 0) zio->io_flags |= ZIO_FLAG_DONT_CACHE; if (BP_GET_TYPE(bp) == DMU_OT_DDT_ZAP) zio->io_flags |= ZIO_FLAG_DONT_CACHE; if (BP_GET_DEDUP(bp) && zio->io_child_type == ZIO_CHILD_LOGICAL) zio->io_pipeline = ZIO_DDT_READ_PIPELINE; return (ZIO_PIPELINE_CONTINUE); } static int zio_write_bp_init(zio_t *zio) { spa_t *spa = zio->io_spa; zio_prop_t *zp = &zio->io_prop; enum zio_compress compress = zp->zp_compress; blkptr_t *bp = zio->io_bp; uint64_t lsize = zio->io_size; uint64_t psize = lsize; int pass = 1; /* * If our children haven't all reached the ready stage, * wait for them and then repeat this pipeline stage. */ if (zio_wait_for_children(zio, ZIO_CHILD_GANG, ZIO_WAIT_READY) || zio_wait_for_children(zio, ZIO_CHILD_LOGICAL, ZIO_WAIT_READY)) return (ZIO_PIPELINE_STOP); if (!IO_IS_ALLOCATING(zio)) return (ZIO_PIPELINE_CONTINUE); ASSERT(zio->io_child_type != ZIO_CHILD_DDT); if (zio->io_bp_override) { ASSERT(bp->blk_birth != zio->io_txg); ASSERT(BP_GET_DEDUP(zio->io_bp_override) == 0); *bp = *zio->io_bp_override; zio->io_pipeline = ZIO_INTERLOCK_PIPELINE; if (BP_IS_EMBEDDED(bp)) return (ZIO_PIPELINE_CONTINUE); /* * If we've been overridden and nopwrite is set then * set the flag accordingly to indicate that a nopwrite * has already occurred. */ if (!BP_IS_HOLE(bp) && zp->zp_nopwrite) { ASSERT(!zp->zp_dedup); zio->io_flags |= ZIO_FLAG_NOPWRITE; return (ZIO_PIPELINE_CONTINUE); } ASSERT(!zp->zp_nopwrite); if (BP_IS_HOLE(bp) || !zp->zp_dedup) return (ZIO_PIPELINE_CONTINUE); ASSERT(zio_checksum_table[zp->zp_checksum].ci_dedup || zp->zp_dedup_verify); if (BP_GET_CHECKSUM(bp) == zp->zp_checksum) { BP_SET_DEDUP(bp, 1); zio->io_pipeline |= ZIO_STAGE_DDT_WRITE; return (ZIO_PIPELINE_CONTINUE); } zio->io_bp_override = NULL; BP_ZERO(bp); } if (!BP_IS_HOLE(bp) && bp->blk_birth == zio->io_txg) { /* * We're rewriting an existing block, which means we're * working on behalf of spa_sync(). For spa_sync() to * converge, it must eventually be the case that we don't * have to allocate new blocks. But compression changes * the blocksize, which forces a reallocate, and makes * convergence take longer. Therefore, after the first * few passes, stop compressing to ensure convergence. */ pass = spa_sync_pass(spa); ASSERT(zio->io_txg == spa_syncing_txg(spa)); ASSERT(zio->io_child_type == ZIO_CHILD_LOGICAL); ASSERT(!BP_GET_DEDUP(bp)); if (pass >= zfs_sync_pass_dont_compress) compress = ZIO_COMPRESS_OFF; /* Make sure someone doesn't change their mind on overwrites */ ASSERT(BP_IS_EMBEDDED(bp) || MIN(zp->zp_copies + BP_IS_GANG(bp), spa_max_replication(spa)) == BP_GET_NDVAS(bp)); } if (compress != ZIO_COMPRESS_OFF) { void *cbuf = zio_buf_alloc(lsize); psize = zio_compress_data(compress, zio->io_data, cbuf, lsize); if (psize == 0 || psize == lsize) { compress = ZIO_COMPRESS_OFF; zio_buf_free(cbuf, lsize); } else if (!zp->zp_dedup && psize <= BPE_PAYLOAD_SIZE && zp->zp_level == 0 && !DMU_OT_HAS_FILL(zp->zp_type) && spa_feature_is_enabled(spa, SPA_FEATURE_EMBEDDED_DATA)) { encode_embedded_bp_compressed(bp, cbuf, compress, lsize, psize); BPE_SET_ETYPE(bp, BP_EMBEDDED_TYPE_DATA); BP_SET_TYPE(bp, zio->io_prop.zp_type); BP_SET_LEVEL(bp, zio->io_prop.zp_level); zio_buf_free(cbuf, lsize); bp->blk_birth = zio->io_txg; zio->io_pipeline = ZIO_INTERLOCK_PIPELINE; ASSERT(spa_feature_is_active(spa, SPA_FEATURE_EMBEDDED_DATA)); return (ZIO_PIPELINE_CONTINUE); } else { /* * Round up compressed size to MINBLOCKSIZE and * zero the tail. */ size_t rounded = P2ROUNDUP(psize, (size_t)SPA_MINBLOCKSIZE); if (rounded > psize) { bzero((char *)cbuf + psize, rounded - psize); psize = rounded; } if (psize == lsize) { compress = ZIO_COMPRESS_OFF; zio_buf_free(cbuf, lsize); } else { zio_push_transform(zio, cbuf, psize, lsize, NULL); } } } /* * The final pass of spa_sync() must be all rewrites, but the first * few passes offer a trade-off: allocating blocks defers convergence, * but newly allocated blocks are sequential, so they can be written * to disk faster. Therefore, we allow the first few passes of * spa_sync() to allocate new blocks, but force rewrites after that. * There should only be a handful of blocks after pass 1 in any case. */ if (!BP_IS_HOLE(bp) && bp->blk_birth == zio->io_txg && BP_GET_PSIZE(bp) == psize && pass >= zfs_sync_pass_rewrite) { enum zio_stage gang_stages = zio->io_pipeline & ZIO_GANG_STAGES; ASSERT(psize != 0); zio->io_pipeline = ZIO_REWRITE_PIPELINE | gang_stages; zio->io_flags |= ZIO_FLAG_IO_REWRITE; } else { BP_ZERO(bp); zio->io_pipeline = ZIO_WRITE_PIPELINE; } if (psize == 0) { if (zio->io_bp_orig.blk_birth != 0 && spa_feature_is_active(spa, SPA_FEATURE_HOLE_BIRTH)) { BP_SET_LSIZE(bp, lsize); BP_SET_TYPE(bp, zp->zp_type); BP_SET_LEVEL(bp, zp->zp_level); BP_SET_BIRTH(bp, zio->io_txg, 0); } zio->io_pipeline = ZIO_INTERLOCK_PIPELINE; } else { ASSERT(zp->zp_checksum != ZIO_CHECKSUM_GANG_HEADER); BP_SET_LSIZE(bp, lsize); BP_SET_TYPE(bp, zp->zp_type); BP_SET_LEVEL(bp, zp->zp_level); BP_SET_PSIZE(bp, psize); BP_SET_COMPRESS(bp, compress); BP_SET_CHECKSUM(bp, zp->zp_checksum); BP_SET_DEDUP(bp, zp->zp_dedup); BP_SET_BYTEORDER(bp, ZFS_HOST_BYTEORDER); if (zp->zp_dedup) { ASSERT(zio->io_child_type == ZIO_CHILD_LOGICAL); ASSERT(!(zio->io_flags & ZIO_FLAG_IO_REWRITE)); zio->io_pipeline = ZIO_DDT_WRITE_PIPELINE; } if (zp->zp_nopwrite) { ASSERT(zio->io_child_type == ZIO_CHILD_LOGICAL); ASSERT(!(zio->io_flags & ZIO_FLAG_IO_REWRITE)); zio->io_pipeline |= ZIO_STAGE_NOP_WRITE; } } return (ZIO_PIPELINE_CONTINUE); } static int zio_free_bp_init(zio_t *zio) { blkptr_t *bp = zio->io_bp; if (zio->io_child_type == ZIO_CHILD_LOGICAL) { if (BP_GET_DEDUP(bp)) zio->io_pipeline = ZIO_DDT_FREE_PIPELINE; } return (ZIO_PIPELINE_CONTINUE); } /* * ========================================================================== * Execute the I/O pipeline * ========================================================================== */ static void zio_taskq_dispatch(zio_t *zio, zio_taskq_type_t q, boolean_t cutinline) { spa_t *spa = zio->io_spa; zio_type_t t = zio->io_type; int flags = (cutinline ? TQ_FRONT : 0); /* * If we're a config writer or a probe, the normal issue and * interrupt threads may all be blocked waiting for the config lock. * In this case, select the otherwise-unused taskq for ZIO_TYPE_NULL. */ if (zio->io_flags & (ZIO_FLAG_CONFIG_WRITER | ZIO_FLAG_PROBE)) t = ZIO_TYPE_NULL; /* * A similar issue exists for the L2ARC write thread until L2ARC 2.0. */ if (t == ZIO_TYPE_WRITE && zio->io_vd && zio->io_vd->vdev_aux) t = ZIO_TYPE_NULL; /* * If this is a high priority I/O, then use the high priority taskq if * available. */ if (zio->io_priority == ZIO_PRIORITY_NOW && spa->spa_zio_taskq[t][q + 1].stqs_count != 0) q++; ASSERT3U(q, <, ZIO_TASKQ_TYPES); /* * NB: We are assuming that the zio can only be dispatched * to a single taskq at a time. It would be a grievous error * to dispatch the zio to another taskq at the same time. */ ASSERT(taskq_empty_ent(&zio->io_tqent)); spa_taskq_dispatch_ent(spa, t, q, (task_func_t *)zio_execute, zio, flags, &zio->io_tqent); } static boolean_t zio_taskq_member(zio_t *zio, zio_taskq_type_t q) { kthread_t *executor = zio->io_executor; spa_t *spa = zio->io_spa; zio_type_t t; for (t = 0; t < ZIO_TYPES; t++) { spa_taskqs_t *tqs = &spa->spa_zio_taskq[t][q]; uint_t i; for (i = 0; i < tqs->stqs_count; i++) { if (taskq_member(tqs->stqs_taskq[i], executor)) return (B_TRUE); } } return (B_FALSE); } static int zio_issue_async(zio_t *zio) { zio_taskq_dispatch(zio, ZIO_TASKQ_ISSUE, B_FALSE); return (ZIO_PIPELINE_STOP); } void zio_interrupt(zio_t *zio) { zio_taskq_dispatch(zio, ZIO_TASKQ_INTERRUPT, B_FALSE); } /* * Execute the I/O pipeline until one of the following occurs: * (1) the I/O completes; (2) the pipeline stalls waiting for * dependent child I/Os; (3) the I/O issues, so we're waiting * for an I/O completion interrupt; (4) the I/O is delegated by * vdev-level caching or aggregation; (5) the I/O is deferred * due to vdev-level queueing; (6) the I/O is handed off to * another thread. In all cases, the pipeline stops whenever * there's no CPU work; it never burns a thread in cv_wait_io(). * * There's no locking on io_stage because there's no legitimate way * for multiple threads to be attempting to process the same I/O. */ static zio_pipe_stage_t *zio_pipeline[]; /* * zio_execute() is a wrapper around the static function * __zio_execute() so that we can force __zio_execute() to be * inlined. This reduces stack overhead which is important * because __zio_execute() is called recursively in several zio * code paths. zio_execute() itself cannot be inlined because * it is externally visible. */ void zio_execute(zio_t *zio) { __zio_execute(zio); } __attribute__((always_inline)) static inline void __zio_execute(zio_t *zio) { zio->io_executor = curthread; while (zio->io_stage < ZIO_STAGE_DONE) { enum zio_stage pipeline = zio->io_pipeline; enum zio_stage stage = zio->io_stage; dsl_pool_t *dp; boolean_t cut; int rv; ASSERT(!MUTEX_HELD(&zio->io_lock)); ASSERT(ISP2(stage)); ASSERT(zio->io_stall == NULL); do { stage <<= 1; } while ((stage & pipeline) == 0); ASSERT(stage <= ZIO_STAGE_DONE); dp = spa_get_dsl(zio->io_spa); cut = (stage == ZIO_STAGE_VDEV_IO_START) ? zio_requeue_io_start_cut_in_line : B_FALSE; /* * If we are in interrupt context and this pipeline stage * will grab a config lock that is held across I/O, * or may wait for an I/O that needs an interrupt thread * to complete, issue async to avoid deadlock. * * For VDEV_IO_START, we cut in line so that the io will * be sent to disk promptly. */ if ((stage & ZIO_BLOCKING_STAGES) && zio->io_vd == NULL && zio_taskq_member(zio, ZIO_TASKQ_INTERRUPT)) { zio_taskq_dispatch(zio, ZIO_TASKQ_ISSUE, cut); return; } /* * If we executing in the context of the tx_sync_thread, * or we are performing pool initialization outside of a * zio_taskq[ZIO_TASKQ_ISSUE|ZIO_TASKQ_ISSUE_HIGH] context. * Then issue the zio asynchronously to minimize stack usage * for these deep call paths. */ if ((dp && curthread == dp->dp_tx.tx_sync_thread) || (dp && spa_is_initializing(dp->dp_spa) && !zio_taskq_member(zio, ZIO_TASKQ_ISSUE) && !zio_taskq_member(zio, ZIO_TASKQ_ISSUE_HIGH))) { zio_taskq_dispatch(zio, ZIO_TASKQ_ISSUE, cut); return; } zio->io_stage = stage; rv = zio_pipeline[highbit64(stage) - 1](zio); if (rv == ZIO_PIPELINE_STOP) return; ASSERT(rv == ZIO_PIPELINE_CONTINUE); } } /* * ========================================================================== * Initiate I/O, either sync or async * ========================================================================== */ int zio_wait(zio_t *zio) { int error; ASSERT(zio->io_stage == ZIO_STAGE_OPEN); ASSERT(zio->io_executor == NULL); zio->io_waiter = curthread; __zio_execute(zio); mutex_enter(&zio->io_lock); while (zio->io_executor != NULL) cv_wait_io(&zio->io_cv, &zio->io_lock); mutex_exit(&zio->io_lock); error = zio->io_error; zio_destroy(zio); return (error); } void zio_nowait(zio_t *zio) { ASSERT(zio->io_executor == NULL); if (zio->io_child_type == ZIO_CHILD_LOGICAL && zio_unique_parent(zio) == NULL) { /* * This is a logical async I/O with no parent to wait for it. * We add it to the spa_async_root_zio "Godfather" I/O which * will ensure they complete prior to unloading the pool. */ spa_t *spa = zio->io_spa; zio_add_child(spa->spa_async_zio_root, zio); } __zio_execute(zio); } /* * ========================================================================== * Reexecute or suspend/resume failed I/O * ========================================================================== */ static void zio_reexecute(zio_t *pio) { zio_t *cio, *cio_next; int c, w; ASSERT(pio->io_child_type == ZIO_CHILD_LOGICAL); ASSERT(pio->io_orig_stage == ZIO_STAGE_OPEN); ASSERT(pio->io_gang_leader == NULL); ASSERT(pio->io_gang_tree == NULL); pio->io_flags = pio->io_orig_flags; pio->io_stage = pio->io_orig_stage; pio->io_pipeline = pio->io_orig_pipeline; pio->io_reexecute = 0; pio->io_flags |= ZIO_FLAG_REEXECUTED; pio->io_error = 0; for (w = 0; w < ZIO_WAIT_TYPES; w++) pio->io_state[w] = 0; for (c = 0; c < ZIO_CHILD_TYPES; c++) pio->io_child_error[c] = 0; if (IO_IS_ALLOCATING(pio)) BP_ZERO(pio->io_bp); /* * As we reexecute pio's children, new children could be created. * New children go to the head of pio's io_child_list, however, * so we will (correctly) not reexecute them. The key is that * the remainder of pio's io_child_list, from 'cio_next' onward, * cannot be affected by any side effects of reexecuting 'cio'. */ for (cio = zio_walk_children(pio); cio != NULL; cio = cio_next) { cio_next = zio_walk_children(pio); mutex_enter(&pio->io_lock); for (w = 0; w < ZIO_WAIT_TYPES; w++) pio->io_children[cio->io_child_type][w]++; mutex_exit(&pio->io_lock); zio_reexecute(cio); } /* * Now that all children have been reexecuted, execute the parent. * We don't reexecute "The Godfather" I/O here as it's the * responsibility of the caller to wait on him. */ if (!(pio->io_flags & ZIO_FLAG_GODFATHER)) __zio_execute(pio); } void zio_suspend(spa_t *spa, zio_t *zio) { if (spa_get_failmode(spa) == ZIO_FAILURE_MODE_PANIC) fm_panic("Pool '%s' has encountered an uncorrectable I/O " "failure and the failure mode property for this pool " "is set to panic.", spa_name(spa)); cmn_err(CE_WARN, "Pool '%s' has encountered an uncorrectable I/O " "failure and has been suspended.\n", spa_name(spa)); zfs_ereport_post(FM_EREPORT_ZFS_IO_FAILURE, spa, NULL, NULL, 0, 0); mutex_enter(&spa->spa_suspend_lock); if (spa->spa_suspend_zio_root == NULL) spa->spa_suspend_zio_root = zio_root(spa, NULL, NULL, ZIO_FLAG_CANFAIL | ZIO_FLAG_SPECULATIVE | ZIO_FLAG_GODFATHER); spa->spa_suspended = B_TRUE; if (zio != NULL) { ASSERT(!(zio->io_flags & ZIO_FLAG_GODFATHER)); ASSERT(zio != spa->spa_suspend_zio_root); ASSERT(zio->io_child_type == ZIO_CHILD_LOGICAL); ASSERT(zio_unique_parent(zio) == NULL); ASSERT(zio->io_stage == ZIO_STAGE_DONE); zio_add_child(spa->spa_suspend_zio_root, zio); } mutex_exit(&spa->spa_suspend_lock); } int zio_resume(spa_t *spa) { zio_t *pio; /* * Reexecute all previously suspended i/o. */ mutex_enter(&spa->spa_suspend_lock); spa->spa_suspended = B_FALSE; cv_broadcast(&spa->spa_suspend_cv); pio = spa->spa_suspend_zio_root; spa->spa_suspend_zio_root = NULL; mutex_exit(&spa->spa_suspend_lock); if (pio == NULL) return (0); zio_reexecute(pio); return (zio_wait(pio)); } void zio_resume_wait(spa_t *spa) { mutex_enter(&spa->spa_suspend_lock); while (spa_suspended(spa)) cv_wait(&spa->spa_suspend_cv, &spa->spa_suspend_lock); mutex_exit(&spa->spa_suspend_lock); } /* * ========================================================================== * Gang blocks. * * A gang block is a collection of small blocks that looks to the DMU * like one large block. When zio_dva_allocate() cannot find a block * of the requested size, due to either severe fragmentation or the pool * being nearly full, it calls zio_write_gang_block() to construct the * block from smaller fragments. * * A gang block consists of a gang header (zio_gbh_phys_t) and up to * three (SPA_GBH_NBLKPTRS) gang members. The gang header is just like * an indirect block: it's an array of block pointers. It consumes * only one sector and hence is allocatable regardless of fragmentation. * The gang header's bps point to its gang members, which hold the data. * * Gang blocks are self-checksumming, using the bp's * as the verifier to ensure uniqueness of the SHA256 checksum. * Critically, the gang block bp's blk_cksum is the checksum of the data, * not the gang header. This ensures that data block signatures (needed for * deduplication) are independent of how the block is physically stored. * * Gang blocks can be nested: a gang member may itself be a gang block. * Thus every gang block is a tree in which root and all interior nodes are * gang headers, and the leaves are normal blocks that contain user data. * The root of the gang tree is called the gang leader. * * To perform any operation (read, rewrite, free, claim) on a gang block, * zio_gang_assemble() first assembles the gang tree (minus data leaves) * in the io_gang_tree field of the original logical i/o by recursively * reading the gang leader and all gang headers below it. This yields * an in-core tree containing the contents of every gang header and the * bps for every constituent of the gang block. * * With the gang tree now assembled, zio_gang_issue() just walks the gang tree * and invokes a callback on each bp. To free a gang block, zio_gang_issue() * calls zio_free_gang() -- a trivial wrapper around zio_free() -- for each bp. * zio_claim_gang() provides a similarly trivial wrapper for zio_claim(). * zio_read_gang() is a wrapper around zio_read() that omits reading gang * headers, since we already have those in io_gang_tree. zio_rewrite_gang() * performs a zio_rewrite() of the data or, for gang headers, a zio_rewrite() * of the gang header plus zio_checksum_compute() of the data to update the * gang header's blk_cksum as described above. * * The two-phase assemble/issue model solves the problem of partial failure -- * what if you'd freed part of a gang block but then couldn't read the * gang header for another part? Assembling the entire gang tree first * ensures that all the necessary gang header I/O has succeeded before * starting the actual work of free, claim, or write. Once the gang tree * is assembled, free and claim are in-memory operations that cannot fail. * * In the event that a gang write fails, zio_dva_unallocate() walks the * gang tree to immediately free (i.e. insert back into the space map) * everything we've allocated. This ensures that we don't get ENOSPC * errors during repeated suspend/resume cycles due to a flaky device. * * Gang rewrites only happen during sync-to-convergence. If we can't assemble * the gang tree, we won't modify the block, so we can safely defer the free * (knowing that the block is still intact). If we *can* assemble the gang * tree, then even if some of the rewrites fail, zio_dva_unallocate() will free * each constituent bp and we can allocate a new block on the next sync pass. * * In all cases, the gang tree allows complete recovery from partial failure. * ========================================================================== */ static zio_t * zio_read_gang(zio_t *pio, blkptr_t *bp, zio_gang_node_t *gn, void *data) { if (gn != NULL) return (pio); return (zio_read(pio, pio->io_spa, bp, data, BP_GET_PSIZE(bp), NULL, NULL, pio->io_priority, ZIO_GANG_CHILD_FLAGS(pio), &pio->io_bookmark)); } zio_t * zio_rewrite_gang(zio_t *pio, blkptr_t *bp, zio_gang_node_t *gn, void *data) { zio_t *zio; if (gn != NULL) { zio = zio_rewrite(pio, pio->io_spa, pio->io_txg, bp, gn->gn_gbh, SPA_GANGBLOCKSIZE, NULL, NULL, pio->io_priority, ZIO_GANG_CHILD_FLAGS(pio), &pio->io_bookmark); /* * As we rewrite each gang header, the pipeline will compute * a new gang block header checksum for it; but no one will * compute a new data checksum, so we do that here. The one * exception is the gang leader: the pipeline already computed * its data checksum because that stage precedes gang assembly. * (Presently, nothing actually uses interior data checksums; * this is just good hygiene.) */ if (gn != pio->io_gang_leader->io_gang_tree) { zio_checksum_compute(zio, BP_GET_CHECKSUM(bp), data, BP_GET_PSIZE(bp)); } /* * If we are here to damage data for testing purposes, * leave the GBH alone so that we can detect the damage. */ if (pio->io_gang_leader->io_flags & ZIO_FLAG_INDUCE_DAMAGE) zio->io_pipeline &= ~ZIO_VDEV_IO_STAGES; } else { zio = zio_rewrite(pio, pio->io_spa, pio->io_txg, bp, data, BP_GET_PSIZE(bp), NULL, NULL, pio->io_priority, ZIO_GANG_CHILD_FLAGS(pio), &pio->io_bookmark); } return (zio); } /* ARGSUSED */ zio_t * zio_free_gang(zio_t *pio, blkptr_t *bp, zio_gang_node_t *gn, void *data) { return (zio_free_sync(pio, pio->io_spa, pio->io_txg, bp, ZIO_GANG_CHILD_FLAGS(pio))); } /* ARGSUSED */ zio_t * zio_claim_gang(zio_t *pio, blkptr_t *bp, zio_gang_node_t *gn, void *data) { return (zio_claim(pio, pio->io_spa, pio->io_txg, bp, NULL, NULL, ZIO_GANG_CHILD_FLAGS(pio))); } static zio_gang_issue_func_t *zio_gang_issue_func[ZIO_TYPES] = { NULL, zio_read_gang, zio_rewrite_gang, zio_free_gang, zio_claim_gang, NULL }; static void zio_gang_tree_assemble_done(zio_t *zio); static zio_gang_node_t * zio_gang_node_alloc(zio_gang_node_t **gnpp) { zio_gang_node_t *gn; ASSERT(*gnpp == NULL); gn = kmem_zalloc(sizeof (*gn), KM_PUSHPAGE); gn->gn_gbh = zio_buf_alloc(SPA_GANGBLOCKSIZE); *gnpp = gn; return (gn); } static void zio_gang_node_free(zio_gang_node_t **gnpp) { zio_gang_node_t *gn = *gnpp; int g; for (g = 0; g < SPA_GBH_NBLKPTRS; g++) ASSERT(gn->gn_child[g] == NULL); zio_buf_free(gn->gn_gbh, SPA_GANGBLOCKSIZE); kmem_free(gn, sizeof (*gn)); *gnpp = NULL; } static void zio_gang_tree_free(zio_gang_node_t **gnpp) { zio_gang_node_t *gn = *gnpp; int g; if (gn == NULL) return; for (g = 0; g < SPA_GBH_NBLKPTRS; g++) zio_gang_tree_free(&gn->gn_child[g]); zio_gang_node_free(gnpp); } static void zio_gang_tree_assemble(zio_t *gio, blkptr_t *bp, zio_gang_node_t **gnpp) { zio_gang_node_t *gn = zio_gang_node_alloc(gnpp); ASSERT(gio->io_gang_leader == gio); ASSERT(BP_IS_GANG(bp)); zio_nowait(zio_read(gio, gio->io_spa, bp, gn->gn_gbh, SPA_GANGBLOCKSIZE, zio_gang_tree_assemble_done, gn, gio->io_priority, ZIO_GANG_CHILD_FLAGS(gio), &gio->io_bookmark)); } static void zio_gang_tree_assemble_done(zio_t *zio) { zio_t *gio = zio->io_gang_leader; zio_gang_node_t *gn = zio->io_private; blkptr_t *bp = zio->io_bp; int g; ASSERT(gio == zio_unique_parent(zio)); ASSERT(zio->io_child_count == 0); if (zio->io_error) return; if (BP_SHOULD_BYTESWAP(bp)) byteswap_uint64_array(zio->io_data, zio->io_size); ASSERT(zio->io_data == gn->gn_gbh); ASSERT(zio->io_size == SPA_GANGBLOCKSIZE); ASSERT(gn->gn_gbh->zg_tail.zec_magic == ZEC_MAGIC); for (g = 0; g < SPA_GBH_NBLKPTRS; g++) { blkptr_t *gbp = &gn->gn_gbh->zg_blkptr[g]; if (!BP_IS_GANG(gbp)) continue; zio_gang_tree_assemble(gio, gbp, &gn->gn_child[g]); } } static void zio_gang_tree_issue(zio_t *pio, zio_gang_node_t *gn, blkptr_t *bp, void *data) { zio_t *gio = pio->io_gang_leader; zio_t *zio; int g; ASSERT(BP_IS_GANG(bp) == !!gn); ASSERT(BP_GET_CHECKSUM(bp) == BP_GET_CHECKSUM(gio->io_bp)); ASSERT(BP_GET_LSIZE(bp) == BP_GET_PSIZE(bp) || gn == gio->io_gang_tree); /* * If you're a gang header, your data is in gn->gn_gbh. * If you're a gang member, your data is in 'data' and gn == NULL. */ zio = zio_gang_issue_func[gio->io_type](pio, bp, gn, data); if (gn != NULL) { ASSERT(gn->gn_gbh->zg_tail.zec_magic == ZEC_MAGIC); for (g = 0; g < SPA_GBH_NBLKPTRS; g++) { blkptr_t *gbp = &gn->gn_gbh->zg_blkptr[g]; if (BP_IS_HOLE(gbp)) continue; zio_gang_tree_issue(zio, gn->gn_child[g], gbp, data); data = (char *)data + BP_GET_PSIZE(gbp); } } if (gn == gio->io_gang_tree) ASSERT3P((char *)gio->io_data + gio->io_size, ==, data); if (zio != pio) zio_nowait(zio); } static int zio_gang_assemble(zio_t *zio) { blkptr_t *bp = zio->io_bp; ASSERT(BP_IS_GANG(bp) && zio->io_gang_leader == NULL); ASSERT(zio->io_child_type > ZIO_CHILD_GANG); zio->io_gang_leader = zio; zio_gang_tree_assemble(zio, bp, &zio->io_gang_tree); return (ZIO_PIPELINE_CONTINUE); } static int zio_gang_issue(zio_t *zio) { blkptr_t *bp = zio->io_bp; if (zio_wait_for_children(zio, ZIO_CHILD_GANG, ZIO_WAIT_DONE)) return (ZIO_PIPELINE_STOP); ASSERT(BP_IS_GANG(bp) && zio->io_gang_leader == zio); ASSERT(zio->io_child_type > ZIO_CHILD_GANG); if (zio->io_child_error[ZIO_CHILD_GANG] == 0) zio_gang_tree_issue(zio, zio->io_gang_tree, bp, zio->io_data); else zio_gang_tree_free(&zio->io_gang_tree); zio->io_pipeline = ZIO_INTERLOCK_PIPELINE; return (ZIO_PIPELINE_CONTINUE); } static void zio_write_gang_member_ready(zio_t *zio) { zio_t *pio = zio_unique_parent(zio); dva_t *cdva = zio->io_bp->blk_dva; dva_t *pdva = pio->io_bp->blk_dva; uint64_t asize; int d; ASSERTV(zio_t *gio = zio->io_gang_leader); if (BP_IS_HOLE(zio->io_bp)) return; ASSERT(BP_IS_HOLE(&zio->io_bp_orig)); ASSERT(zio->io_child_type == ZIO_CHILD_GANG); ASSERT3U(zio->io_prop.zp_copies, ==, gio->io_prop.zp_copies); ASSERT3U(zio->io_prop.zp_copies, <=, BP_GET_NDVAS(zio->io_bp)); ASSERT3U(pio->io_prop.zp_copies, <=, BP_GET_NDVAS(pio->io_bp)); ASSERT3U(BP_GET_NDVAS(zio->io_bp), <=, BP_GET_NDVAS(pio->io_bp)); mutex_enter(&pio->io_lock); for (d = 0; d < BP_GET_NDVAS(zio->io_bp); d++) { ASSERT(DVA_GET_GANG(&pdva[d])); asize = DVA_GET_ASIZE(&pdva[d]); asize += DVA_GET_ASIZE(&cdva[d]); DVA_SET_ASIZE(&pdva[d], asize); } mutex_exit(&pio->io_lock); } static int zio_write_gang_block(zio_t *pio) { spa_t *spa = pio->io_spa; blkptr_t *bp = pio->io_bp; zio_t *gio = pio->io_gang_leader; zio_t *zio; zio_gang_node_t *gn, **gnpp; zio_gbh_phys_t *gbh; uint64_t txg = pio->io_txg; uint64_t resid = pio->io_size; uint64_t lsize; int copies = gio->io_prop.zp_copies; int gbh_copies = MIN(copies + 1, spa_max_replication(spa)); zio_prop_t zp; int g, error; error = metaslab_alloc(spa, spa_normal_class(spa), SPA_GANGBLOCKSIZE, bp, gbh_copies, txg, pio == gio ? NULL : gio->io_bp, METASLAB_HINTBP_FAVOR | METASLAB_GANG_HEADER); if (error) { pio->io_error = error; return (ZIO_PIPELINE_CONTINUE); } if (pio == gio) { gnpp = &gio->io_gang_tree; } else { gnpp = pio->io_private; ASSERT(pio->io_ready == zio_write_gang_member_ready); } gn = zio_gang_node_alloc(gnpp); gbh = gn->gn_gbh; bzero(gbh, SPA_GANGBLOCKSIZE); /* * Create the gang header. */ zio = zio_rewrite(pio, spa, txg, bp, gbh, SPA_GANGBLOCKSIZE, NULL, NULL, pio->io_priority, ZIO_GANG_CHILD_FLAGS(pio), &pio->io_bookmark); /* * Create and nowait the gang children. */ for (g = 0; resid != 0; resid -= lsize, g++) { lsize = P2ROUNDUP(resid / (SPA_GBH_NBLKPTRS - g), SPA_MINBLOCKSIZE); ASSERT(lsize >= SPA_MINBLOCKSIZE && lsize <= resid); zp.zp_checksum = gio->io_prop.zp_checksum; zp.zp_compress = ZIO_COMPRESS_OFF; zp.zp_type = DMU_OT_NONE; zp.zp_level = 0; zp.zp_copies = gio->io_prop.zp_copies; zp.zp_dedup = B_FALSE; zp.zp_dedup_verify = B_FALSE; zp.zp_nopwrite = B_FALSE; zio_nowait(zio_write(zio, spa, txg, &gbh->zg_blkptr[g], (char *)pio->io_data + (pio->io_size - resid), lsize, &zp, zio_write_gang_member_ready, NULL, NULL, &gn->gn_child[g], pio->io_priority, ZIO_GANG_CHILD_FLAGS(pio), &pio->io_bookmark)); } /* * Set pio's pipeline to just wait for zio to finish. */ pio->io_pipeline = ZIO_INTERLOCK_PIPELINE; /* * We didn't allocate this bp, so make sure it doesn't get unmarked. */ pio->io_flags &= ~ZIO_FLAG_FASTWRITE; zio_nowait(zio); return (ZIO_PIPELINE_CONTINUE); } /* * The zio_nop_write stage in the pipeline determines if allocating * a new bp is necessary. By leveraging a cryptographically secure checksum, * such as SHA256, we can compare the checksums of the new data and the old * to determine if allocating a new block is required. The nopwrite * feature can handle writes in either syncing or open context (i.e. zil * writes) and as a result is mutually exclusive with dedup. */ static int zio_nop_write(zio_t *zio) { blkptr_t *bp = zio->io_bp; blkptr_t *bp_orig = &zio->io_bp_orig; zio_prop_t *zp = &zio->io_prop; ASSERT(BP_GET_LEVEL(bp) == 0); ASSERT(!(zio->io_flags & ZIO_FLAG_IO_REWRITE)); ASSERT(zp->zp_nopwrite); ASSERT(!zp->zp_dedup); ASSERT(zio->io_bp_override == NULL); ASSERT(IO_IS_ALLOCATING(zio)); /* * Check to see if the original bp and the new bp have matching * characteristics (i.e. same checksum, compression algorithms, etc). * If they don't then just continue with the pipeline which will * allocate a new bp. */ if (BP_IS_HOLE(bp_orig) || !zio_checksum_table[BP_GET_CHECKSUM(bp)].ci_dedup || BP_GET_CHECKSUM(bp) != BP_GET_CHECKSUM(bp_orig) || BP_GET_COMPRESS(bp) != BP_GET_COMPRESS(bp_orig) || BP_GET_DEDUP(bp) != BP_GET_DEDUP(bp_orig) || zp->zp_copies != BP_GET_NDVAS(bp_orig)) return (ZIO_PIPELINE_CONTINUE); /* * If the checksums match then reset the pipeline so that we * avoid allocating a new bp and issuing any I/O. */ if (ZIO_CHECKSUM_EQUAL(bp->blk_cksum, bp_orig->blk_cksum)) { ASSERT(zio_checksum_table[zp->zp_checksum].ci_dedup); ASSERT3U(BP_GET_PSIZE(bp), ==, BP_GET_PSIZE(bp_orig)); ASSERT3U(BP_GET_LSIZE(bp), ==, BP_GET_LSIZE(bp_orig)); ASSERT(zp->zp_compress != ZIO_COMPRESS_OFF); ASSERT(bcmp(&bp->blk_prop, &bp_orig->blk_prop, sizeof (uint64_t)) == 0); *bp = *bp_orig; zio->io_pipeline = ZIO_INTERLOCK_PIPELINE; zio->io_flags |= ZIO_FLAG_NOPWRITE; } return (ZIO_PIPELINE_CONTINUE); } /* * ========================================================================== * Dedup * ========================================================================== */ static void zio_ddt_child_read_done(zio_t *zio) { blkptr_t *bp = zio->io_bp; ddt_entry_t *dde = zio->io_private; ddt_phys_t *ddp; zio_t *pio = zio_unique_parent(zio); mutex_enter(&pio->io_lock); ddp = ddt_phys_select(dde, bp); if (zio->io_error == 0) ddt_phys_clear(ddp); /* this ddp doesn't need repair */ if (zio->io_error == 0 && dde->dde_repair_data == NULL) dde->dde_repair_data = zio->io_data; else zio_buf_free(zio->io_data, zio->io_size); mutex_exit(&pio->io_lock); } static int zio_ddt_read_start(zio_t *zio) { blkptr_t *bp = zio->io_bp; int p; ASSERT(BP_GET_DEDUP(bp)); ASSERT(BP_GET_PSIZE(bp) == zio->io_size); ASSERT(zio->io_child_type == ZIO_CHILD_LOGICAL); if (zio->io_child_error[ZIO_CHILD_DDT]) { ddt_t *ddt = ddt_select(zio->io_spa, bp); ddt_entry_t *dde = ddt_repair_start(ddt, bp); ddt_phys_t *ddp = dde->dde_phys; ddt_phys_t *ddp_self = ddt_phys_select(dde, bp); blkptr_t blk; ASSERT(zio->io_vsd == NULL); zio->io_vsd = dde; if (ddp_self == NULL) return (ZIO_PIPELINE_CONTINUE); for (p = 0; p < DDT_PHYS_TYPES; p++, ddp++) { if (ddp->ddp_phys_birth == 0 || ddp == ddp_self) continue; ddt_bp_create(ddt->ddt_checksum, &dde->dde_key, ddp, &blk); zio_nowait(zio_read(zio, zio->io_spa, &blk, zio_buf_alloc(zio->io_size), zio->io_size, zio_ddt_child_read_done, dde, zio->io_priority, ZIO_DDT_CHILD_FLAGS(zio) | ZIO_FLAG_DONT_PROPAGATE, &zio->io_bookmark)); } return (ZIO_PIPELINE_CONTINUE); } zio_nowait(zio_read(zio, zio->io_spa, bp, zio->io_data, zio->io_size, NULL, NULL, zio->io_priority, ZIO_DDT_CHILD_FLAGS(zio), &zio->io_bookmark)); return (ZIO_PIPELINE_CONTINUE); } static int zio_ddt_read_done(zio_t *zio) { blkptr_t *bp = zio->io_bp; if (zio_wait_for_children(zio, ZIO_CHILD_DDT, ZIO_WAIT_DONE)) return (ZIO_PIPELINE_STOP); ASSERT(BP_GET_DEDUP(bp)); ASSERT(BP_GET_PSIZE(bp) == zio->io_size); ASSERT(zio->io_child_type == ZIO_CHILD_LOGICAL); if (zio->io_child_error[ZIO_CHILD_DDT]) { ddt_t *ddt = ddt_select(zio->io_spa, bp); ddt_entry_t *dde = zio->io_vsd; if (ddt == NULL) { ASSERT(spa_load_state(zio->io_spa) != SPA_LOAD_NONE); return (ZIO_PIPELINE_CONTINUE); } if (dde == NULL) { zio->io_stage = ZIO_STAGE_DDT_READ_START >> 1; zio_taskq_dispatch(zio, ZIO_TASKQ_ISSUE, B_FALSE); return (ZIO_PIPELINE_STOP); } if (dde->dde_repair_data != NULL) { bcopy(dde->dde_repair_data, zio->io_data, zio->io_size); zio->io_child_error[ZIO_CHILD_DDT] = 0; } ddt_repair_done(ddt, dde); zio->io_vsd = NULL; } ASSERT(zio->io_vsd == NULL); return (ZIO_PIPELINE_CONTINUE); } static boolean_t zio_ddt_collision(zio_t *zio, ddt_t *ddt, ddt_entry_t *dde) { spa_t *spa = zio->io_spa; int p; /* * Note: we compare the original data, not the transformed data, * because when zio->io_bp is an override bp, we will not have * pushed the I/O transforms. That's an important optimization * because otherwise we'd compress/encrypt all dmu_sync() data twice. */ for (p = DDT_PHYS_SINGLE; p <= DDT_PHYS_TRIPLE; p++) { zio_t *lio = dde->dde_lead_zio[p]; if (lio != NULL) { return (lio->io_orig_size != zio->io_orig_size || bcmp(zio->io_orig_data, lio->io_orig_data, zio->io_orig_size) != 0); } } for (p = DDT_PHYS_SINGLE; p <= DDT_PHYS_TRIPLE; p++) { ddt_phys_t *ddp = &dde->dde_phys[p]; if (ddp->ddp_phys_birth != 0) { arc_buf_t *abuf = NULL; uint32_t aflags = ARC_WAIT; blkptr_t blk = *zio->io_bp; int error; ddt_bp_fill(ddp, &blk, ddp->ddp_phys_birth); ddt_exit(ddt); error = arc_read(NULL, spa, &blk, arc_getbuf_func, &abuf, ZIO_PRIORITY_SYNC_READ, ZIO_FLAG_CANFAIL | ZIO_FLAG_SPECULATIVE, &aflags, &zio->io_bookmark); if (error == 0) { if (arc_buf_size(abuf) != zio->io_orig_size || bcmp(abuf->b_data, zio->io_orig_data, zio->io_orig_size) != 0) error = SET_ERROR(EEXIST); VERIFY(arc_buf_remove_ref(abuf, &abuf)); } ddt_enter(ddt); return (error != 0); } } return (B_FALSE); } static void zio_ddt_child_write_ready(zio_t *zio) { int p = zio->io_prop.zp_copies; ddt_t *ddt = ddt_select(zio->io_spa, zio->io_bp); ddt_entry_t *dde = zio->io_private; ddt_phys_t *ddp = &dde->dde_phys[p]; zio_t *pio; if (zio->io_error) return; ddt_enter(ddt); ASSERT(dde->dde_lead_zio[p] == zio); ddt_phys_fill(ddp, zio->io_bp); while ((pio = zio_walk_parents(zio)) != NULL) ddt_bp_fill(ddp, pio->io_bp, zio->io_txg); ddt_exit(ddt); } static void zio_ddt_child_write_done(zio_t *zio) { int p = zio->io_prop.zp_copies; ddt_t *ddt = ddt_select(zio->io_spa, zio->io_bp); ddt_entry_t *dde = zio->io_private; ddt_phys_t *ddp = &dde->dde_phys[p]; ddt_enter(ddt); ASSERT(ddp->ddp_refcnt == 0); ASSERT(dde->dde_lead_zio[p] == zio); dde->dde_lead_zio[p] = NULL; if (zio->io_error == 0) { while (zio_walk_parents(zio) != NULL) ddt_phys_addref(ddp); } else { ddt_phys_clear(ddp); } ddt_exit(ddt); } static void zio_ddt_ditto_write_done(zio_t *zio) { int p = DDT_PHYS_DITTO; blkptr_t *bp = zio->io_bp; ddt_t *ddt = ddt_select(zio->io_spa, bp); ddt_entry_t *dde = zio->io_private; ddt_phys_t *ddp = &dde->dde_phys[p]; ddt_key_t *ddk = &dde->dde_key; ASSERTV(zio_prop_t *zp = &zio->io_prop); ddt_enter(ddt); ASSERT(ddp->ddp_refcnt == 0); ASSERT(dde->dde_lead_zio[p] == zio); dde->dde_lead_zio[p] = NULL; if (zio->io_error == 0) { ASSERT(ZIO_CHECKSUM_EQUAL(bp->blk_cksum, ddk->ddk_cksum)); ASSERT(zp->zp_copies < SPA_DVAS_PER_BP); ASSERT(zp->zp_copies == BP_GET_NDVAS(bp) - BP_IS_GANG(bp)); if (ddp->ddp_phys_birth != 0) ddt_phys_free(ddt, ddk, ddp, zio->io_txg); ddt_phys_fill(ddp, bp); } ddt_exit(ddt); } static int zio_ddt_write(zio_t *zio) { spa_t *spa = zio->io_spa; blkptr_t *bp = zio->io_bp; uint64_t txg = zio->io_txg; zio_prop_t *zp = &zio->io_prop; int p = zp->zp_copies; int ditto_copies; zio_t *cio = NULL; zio_t *dio = NULL; ddt_t *ddt = ddt_select(spa, bp); ddt_entry_t *dde; ddt_phys_t *ddp; ASSERT(BP_GET_DEDUP(bp)); ASSERT(BP_GET_CHECKSUM(bp) == zp->zp_checksum); ASSERT(BP_IS_HOLE(bp) || zio->io_bp_override); ddt_enter(ddt); dde = ddt_lookup(ddt, bp, B_TRUE); ddp = &dde->dde_phys[p]; if (zp->zp_dedup_verify && zio_ddt_collision(zio, ddt, dde)) { /* * If we're using a weak checksum, upgrade to a strong checksum * and try again. If we're already using a strong checksum, * we can't resolve it, so just convert to an ordinary write. * (And automatically e-mail a paper to Nature?) */ if (!zio_checksum_table[zp->zp_checksum].ci_dedup) { zp->zp_checksum = spa_dedup_checksum(spa); zio_pop_transforms(zio); zio->io_stage = ZIO_STAGE_OPEN; BP_ZERO(bp); } else { zp->zp_dedup = B_FALSE; } zio->io_pipeline = ZIO_WRITE_PIPELINE; ddt_exit(ddt); return (ZIO_PIPELINE_CONTINUE); } ditto_copies = ddt_ditto_copies_needed(ddt, dde, ddp); ASSERT(ditto_copies < SPA_DVAS_PER_BP); if (ditto_copies > ddt_ditto_copies_present(dde) && dde->dde_lead_zio[DDT_PHYS_DITTO] == NULL) { zio_prop_t czp = *zp; czp.zp_copies = ditto_copies; /* * If we arrived here with an override bp, we won't have run * the transform stack, so we won't have the data we need to * generate a child i/o. So, toss the override bp and restart. * This is safe, because using the override bp is just an * optimization; and it's rare, so the cost doesn't matter. */ if (zio->io_bp_override) { zio_pop_transforms(zio); zio->io_stage = ZIO_STAGE_OPEN; zio->io_pipeline = ZIO_WRITE_PIPELINE; zio->io_bp_override = NULL; BP_ZERO(bp); ddt_exit(ddt); return (ZIO_PIPELINE_CONTINUE); } dio = zio_write(zio, spa, txg, bp, zio->io_orig_data, zio->io_orig_size, &czp, NULL, NULL, zio_ddt_ditto_write_done, dde, zio->io_priority, ZIO_DDT_CHILD_FLAGS(zio), &zio->io_bookmark); zio_push_transform(dio, zio->io_data, zio->io_size, 0, NULL); dde->dde_lead_zio[DDT_PHYS_DITTO] = dio; } if (ddp->ddp_phys_birth != 0 || dde->dde_lead_zio[p] != NULL) { if (ddp->ddp_phys_birth != 0) ddt_bp_fill(ddp, bp, txg); if (dde->dde_lead_zio[p] != NULL) zio_add_child(zio, dde->dde_lead_zio[p]); else ddt_phys_addref(ddp); } else if (zio->io_bp_override) { ASSERT(bp->blk_birth == txg); ASSERT(BP_EQUAL(bp, zio->io_bp_override)); ddt_phys_fill(ddp, bp); ddt_phys_addref(ddp); } else { cio = zio_write(zio, spa, txg, bp, zio->io_orig_data, zio->io_orig_size, zp, zio_ddt_child_write_ready, NULL, zio_ddt_child_write_done, dde, zio->io_priority, ZIO_DDT_CHILD_FLAGS(zio), &zio->io_bookmark); zio_push_transform(cio, zio->io_data, zio->io_size, 0, NULL); dde->dde_lead_zio[p] = cio; } ddt_exit(ddt); if (cio) zio_nowait(cio); if (dio) zio_nowait(dio); return (ZIO_PIPELINE_CONTINUE); } ddt_entry_t *freedde; /* for debugging */ static int zio_ddt_free(zio_t *zio) { spa_t *spa = zio->io_spa; blkptr_t *bp = zio->io_bp; ddt_t *ddt = ddt_select(spa, bp); ddt_entry_t *dde; ddt_phys_t *ddp; ASSERT(BP_GET_DEDUP(bp)); ASSERT(zio->io_child_type == ZIO_CHILD_LOGICAL); ddt_enter(ddt); freedde = dde = ddt_lookup(ddt, bp, B_TRUE); if (dde) { ddp = ddt_phys_select(dde, bp); if (ddp) ddt_phys_decref(ddp); } ddt_exit(ddt); return (ZIO_PIPELINE_CONTINUE); } /* * ========================================================================== * Allocate and free blocks * ========================================================================== */ static int zio_dva_allocate(zio_t *zio) { spa_t *spa = zio->io_spa; metaslab_class_t *mc = spa_normal_class(spa); blkptr_t *bp = zio->io_bp; int error; int flags = 0; if (zio->io_gang_leader == NULL) { ASSERT(zio->io_child_type > ZIO_CHILD_GANG); zio->io_gang_leader = zio; } ASSERT(BP_IS_HOLE(bp)); ASSERT0(BP_GET_NDVAS(bp)); ASSERT3U(zio->io_prop.zp_copies, >, 0); ASSERT3U(zio->io_prop.zp_copies, <=, spa_max_replication(spa)); ASSERT3U(zio->io_size, ==, BP_GET_PSIZE(bp)); /* * The dump device does not support gang blocks so allocation on * behalf of the dump device (i.e. ZIO_FLAG_NODATA) must avoid * the "fast" gang feature. */ flags |= (zio->io_flags & ZIO_FLAG_NODATA) ? METASLAB_GANG_AVOID : 0; flags |= (zio->io_flags & ZIO_FLAG_GANG_CHILD) ? METASLAB_GANG_CHILD : 0; flags |= (zio->io_flags & ZIO_FLAG_FASTWRITE) ? METASLAB_FASTWRITE : 0; error = metaslab_alloc(spa, mc, zio->io_size, bp, zio->io_prop.zp_copies, zio->io_txg, NULL, flags); if (error) { spa_dbgmsg(spa, "%s: metaslab allocation failure: zio %p, " "size %llu, error %d", spa_name(spa), zio, zio->io_size, error); if (error == ENOSPC && zio->io_size > SPA_MINBLOCKSIZE) return (zio_write_gang_block(zio)); zio->io_error = error; } return (ZIO_PIPELINE_CONTINUE); } static int zio_dva_free(zio_t *zio) { metaslab_free(zio->io_spa, zio->io_bp, zio->io_txg, B_FALSE); return (ZIO_PIPELINE_CONTINUE); } static int zio_dva_claim(zio_t *zio) { int error; error = metaslab_claim(zio->io_spa, zio->io_bp, zio->io_txg); if (error) zio->io_error = error; return (ZIO_PIPELINE_CONTINUE); } /* * Undo an allocation. This is used by zio_done() when an I/O fails * and we want to give back the block we just allocated. * This handles both normal blocks and gang blocks. */ static void zio_dva_unallocate(zio_t *zio, zio_gang_node_t *gn, blkptr_t *bp) { int g; ASSERT(bp->blk_birth == zio->io_txg || BP_IS_HOLE(bp)); ASSERT(zio->io_bp_override == NULL); if (!BP_IS_HOLE(bp)) metaslab_free(zio->io_spa, bp, bp->blk_birth, B_TRUE); if (gn != NULL) { for (g = 0; g < SPA_GBH_NBLKPTRS; g++) { zio_dva_unallocate(zio, gn->gn_child[g], &gn->gn_gbh->zg_blkptr[g]); } } } /* * Try to allocate an intent log block. Return 0 on success, errno on failure. */ int zio_alloc_zil(spa_t *spa, uint64_t txg, blkptr_t *new_bp, uint64_t size, boolean_t use_slog) { int error = 1; ASSERT(txg > spa_syncing_txg(spa)); /* * ZIL blocks are always contiguous (i.e. not gang blocks) so we * set the METASLAB_GANG_AVOID flag so that they don't "fast gang" * when allocating them. */ if (use_slog) { error = metaslab_alloc(spa, spa_log_class(spa), size, new_bp, 1, txg, NULL, METASLAB_FASTWRITE | METASLAB_GANG_AVOID); } if (error) { error = metaslab_alloc(spa, spa_normal_class(spa), size, new_bp, 1, txg, NULL, METASLAB_FASTWRITE); } if (error == 0) { BP_SET_LSIZE(new_bp, size); BP_SET_PSIZE(new_bp, size); BP_SET_COMPRESS(new_bp, ZIO_COMPRESS_OFF); BP_SET_CHECKSUM(new_bp, spa_version(spa) >= SPA_VERSION_SLIM_ZIL ? ZIO_CHECKSUM_ZILOG2 : ZIO_CHECKSUM_ZILOG); BP_SET_TYPE(new_bp, DMU_OT_INTENT_LOG); BP_SET_LEVEL(new_bp, 0); BP_SET_DEDUP(new_bp, 0); BP_SET_BYTEORDER(new_bp, ZFS_HOST_BYTEORDER); } return (error); } /* * Free an intent log block. */ void zio_free_zil(spa_t *spa, uint64_t txg, blkptr_t *bp) { ASSERT(BP_GET_TYPE(bp) == DMU_OT_INTENT_LOG); ASSERT(!BP_IS_GANG(bp)); zio_free(spa, txg, bp); } /* * ========================================================================== * Read and write to physical devices * ========================================================================== */ static int zio_vdev_io_start(zio_t *zio) { vdev_t *vd = zio->io_vd; uint64_t align; spa_t *spa = zio->io_spa; ASSERT(zio->io_error == 0); ASSERT(zio->io_child_error[ZIO_CHILD_VDEV] == 0); if (vd == NULL) { if (!(zio->io_flags & ZIO_FLAG_CONFIG_WRITER)) spa_config_enter(spa, SCL_ZIO, zio, RW_READER); /* * The mirror_ops handle multiple DVAs in a single BP. */ return (vdev_mirror_ops.vdev_op_io_start(zio)); } /* * We keep track of time-sensitive I/Os so that the scan thread * can quickly react to certain workloads. In particular, we care * about non-scrubbing, top-level reads and writes with the following * characteristics: * - synchronous writes of user data to non-slog devices * - any reads of user data * When these conditions are met, adjust the timestamp of spa_last_io * which allows the scan thread to adjust its workload accordingly. */ if (!(zio->io_flags & ZIO_FLAG_SCAN_THREAD) && zio->io_bp != NULL && vd == vd->vdev_top && !vd->vdev_islog && zio->io_bookmark.zb_objset != DMU_META_OBJSET && zio->io_txg != spa_syncing_txg(spa)) { uint64_t old = spa->spa_last_io; uint64_t new = ddi_get_lbolt64(); if (old != new) (void) atomic_cas_64(&spa->spa_last_io, old, new); } align = 1ULL << vd->vdev_top->vdev_ashift; if (P2PHASE(zio->io_size, align) != 0) { uint64_t asize = P2ROUNDUP(zio->io_size, align); char *abuf = zio_buf_alloc(asize); ASSERT(vd == vd->vdev_top); if (zio->io_type == ZIO_TYPE_WRITE) { bcopy(zio->io_data, abuf, zio->io_size); bzero(abuf + zio->io_size, asize - zio->io_size); } zio_push_transform(zio, abuf, asize, asize, zio_subblock); } ASSERT(P2PHASE(zio->io_offset, align) == 0); ASSERT(P2PHASE(zio->io_size, align) == 0); VERIFY(zio->io_type != ZIO_TYPE_WRITE || spa_writeable(spa)); /* * If this is a repair I/O, and there's no self-healing involved -- * that is, we're just resilvering what we expect to resilver -- * then don't do the I/O unless zio's txg is actually in vd's DTL. * This prevents spurious resilvering with nested replication. * For example, given a mirror of mirrors, (A+B)+(C+D), if only * A is out of date, we'll read from C+D, then use the data to * resilver A+B -- but we don't actually want to resilver B, just A. * The top-level mirror has no way to know this, so instead we just * discard unnecessary repairs as we work our way down the vdev tree. * The same logic applies to any form of nested replication: * ditto + mirror, RAID-Z + replacing, etc. This covers them all. */ if ((zio->io_flags & ZIO_FLAG_IO_REPAIR) && !(zio->io_flags & ZIO_FLAG_SELF_HEAL) && zio->io_txg != 0 && /* not a delegated i/o */ !vdev_dtl_contains(vd, DTL_PARTIAL, zio->io_txg, 1)) { ASSERT(zio->io_type == ZIO_TYPE_WRITE); zio_vdev_io_bypass(zio); return (ZIO_PIPELINE_CONTINUE); } if (vd->vdev_ops->vdev_op_leaf && (zio->io_type == ZIO_TYPE_READ || zio->io_type == ZIO_TYPE_WRITE)) { if (zio->io_type == ZIO_TYPE_READ && vdev_cache_read(zio)) return (ZIO_PIPELINE_CONTINUE); if ((zio = vdev_queue_io(zio)) == NULL) return (ZIO_PIPELINE_STOP); if (!vdev_accessible(vd, zio)) { zio->io_error = SET_ERROR(ENXIO); zio_interrupt(zio); return (ZIO_PIPELINE_STOP); } } return (vd->vdev_ops->vdev_op_io_start(zio)); } static int zio_vdev_io_done(zio_t *zio) { vdev_t *vd = zio->io_vd; vdev_ops_t *ops = vd ? vd->vdev_ops : &vdev_mirror_ops; boolean_t unexpected_error = B_FALSE; if (zio_wait_for_children(zio, ZIO_CHILD_VDEV, ZIO_WAIT_DONE)) return (ZIO_PIPELINE_STOP); ASSERT(zio->io_type == ZIO_TYPE_READ || zio->io_type == ZIO_TYPE_WRITE); if (vd != NULL && vd->vdev_ops->vdev_op_leaf) { vdev_queue_io_done(zio); if (zio->io_type == ZIO_TYPE_WRITE) vdev_cache_write(zio); if (zio_injection_enabled && zio->io_error == 0) zio->io_error = zio_handle_device_injection(vd, zio, EIO); if (zio_injection_enabled && zio->io_error == 0) zio->io_error = zio_handle_label_injection(zio, EIO); if (zio->io_error) { if (!vdev_accessible(vd, zio)) { zio->io_error = SET_ERROR(ENXIO); } else { unexpected_error = B_TRUE; } } } ops->vdev_op_io_done(zio); if (unexpected_error) VERIFY(vdev_probe(vd, zio) == NULL); return (ZIO_PIPELINE_CONTINUE); } /* * For non-raidz ZIOs, we can just copy aside the bad data read from the * disk, and use that to finish the checksum ereport later. */ static void zio_vsd_default_cksum_finish(zio_cksum_report_t *zcr, const void *good_buf) { /* no processing needed */ zfs_ereport_finish_checksum(zcr, good_buf, zcr->zcr_cbdata, B_FALSE); } /*ARGSUSED*/ void zio_vsd_default_cksum_report(zio_t *zio, zio_cksum_report_t *zcr, void *ignored) { void *buf = zio_buf_alloc(zio->io_size); bcopy(zio->io_data, buf, zio->io_size); zcr->zcr_cbinfo = zio->io_size; zcr->zcr_cbdata = buf; zcr->zcr_finish = zio_vsd_default_cksum_finish; zcr->zcr_free = zio_buf_free; } static int zio_vdev_io_assess(zio_t *zio) { vdev_t *vd = zio->io_vd; if (zio_wait_for_children(zio, ZIO_CHILD_VDEV, ZIO_WAIT_DONE)) return (ZIO_PIPELINE_STOP); if (vd == NULL && !(zio->io_flags & ZIO_FLAG_CONFIG_WRITER)) spa_config_exit(zio->io_spa, SCL_ZIO, zio); if (zio->io_vsd != NULL) { zio->io_vsd_ops->vsd_free(zio); zio->io_vsd = NULL; } if (zio_injection_enabled && zio->io_error == 0) zio->io_error = zio_handle_fault_injection(zio, EIO); /* * If the I/O failed, determine whether we should attempt to retry it. * * On retry, we cut in line in the issue queue, since we don't want * compression/checksumming/etc. work to prevent our (cheap) IO reissue. */ if (zio->io_error && vd == NULL && !(zio->io_flags & (ZIO_FLAG_DONT_RETRY | ZIO_FLAG_IO_RETRY))) { ASSERT(!(zio->io_flags & ZIO_FLAG_DONT_QUEUE)); /* not a leaf */ ASSERT(!(zio->io_flags & ZIO_FLAG_IO_BYPASS)); /* not a leaf */ zio->io_error = 0; zio->io_flags |= ZIO_FLAG_IO_RETRY | ZIO_FLAG_DONT_CACHE | ZIO_FLAG_DONT_AGGREGATE; zio->io_stage = ZIO_STAGE_VDEV_IO_START >> 1; zio_taskq_dispatch(zio, ZIO_TASKQ_ISSUE, zio_requeue_io_start_cut_in_line); return (ZIO_PIPELINE_STOP); } /* * If we got an error on a leaf device, convert it to ENXIO * if the device is not accessible at all. */ if (zio->io_error && vd != NULL && vd->vdev_ops->vdev_op_leaf && !vdev_accessible(vd, zio)) zio->io_error = SET_ERROR(ENXIO); /* * If we can't write to an interior vdev (mirror or RAID-Z), * set vdev_cant_write so that we stop trying to allocate from it. */ if (zio->io_error == ENXIO && zio->io_type == ZIO_TYPE_WRITE && vd != NULL && !vd->vdev_ops->vdev_op_leaf) { vd->vdev_cant_write = B_TRUE; } if (zio->io_error) zio->io_pipeline = ZIO_INTERLOCK_PIPELINE; if (vd != NULL && vd->vdev_ops->vdev_op_leaf && zio->io_physdone != NULL) { ASSERT(!(zio->io_flags & ZIO_FLAG_DELEGATED)); ASSERT(zio->io_child_type == ZIO_CHILD_VDEV); zio->io_physdone(zio->io_logical); } return (ZIO_PIPELINE_CONTINUE); } void zio_vdev_io_reissue(zio_t *zio) { ASSERT(zio->io_stage == ZIO_STAGE_VDEV_IO_START); ASSERT(zio->io_error == 0); zio->io_stage >>= 1; } void zio_vdev_io_redone(zio_t *zio) { ASSERT(zio->io_stage == ZIO_STAGE_VDEV_IO_DONE); zio->io_stage >>= 1; } void zio_vdev_io_bypass(zio_t *zio) { ASSERT(zio->io_stage == ZIO_STAGE_VDEV_IO_START); ASSERT(zio->io_error == 0); zio->io_flags |= ZIO_FLAG_IO_BYPASS; zio->io_stage = ZIO_STAGE_VDEV_IO_ASSESS >> 1; } /* * ========================================================================== * Generate and verify checksums * ========================================================================== */ static int zio_checksum_generate(zio_t *zio) { blkptr_t *bp = zio->io_bp; enum zio_checksum checksum; if (bp == NULL) { /* * This is zio_write_phys(). * We're either generating a label checksum, or none at all. */ checksum = zio->io_prop.zp_checksum; if (checksum == ZIO_CHECKSUM_OFF) return (ZIO_PIPELINE_CONTINUE); ASSERT(checksum == ZIO_CHECKSUM_LABEL); } else { if (BP_IS_GANG(bp) && zio->io_child_type == ZIO_CHILD_GANG) { ASSERT(!IO_IS_ALLOCATING(zio)); checksum = ZIO_CHECKSUM_GANG_HEADER; } else { checksum = BP_GET_CHECKSUM(bp); } } zio_checksum_compute(zio, checksum, zio->io_data, zio->io_size); return (ZIO_PIPELINE_CONTINUE); } static int zio_checksum_verify(zio_t *zio) { zio_bad_cksum_t info; blkptr_t *bp = zio->io_bp; int error; ASSERT(zio->io_vd != NULL); if (bp == NULL) { /* * This is zio_read_phys(). * We're either verifying a label checksum, or nothing at all. */ if (zio->io_prop.zp_checksum == ZIO_CHECKSUM_OFF) return (ZIO_PIPELINE_CONTINUE); ASSERT(zio->io_prop.zp_checksum == ZIO_CHECKSUM_LABEL); } if ((error = zio_checksum_error(zio, &info)) != 0) { zio->io_error = error; if (!(zio->io_flags & ZIO_FLAG_SPECULATIVE)) { zfs_ereport_start_checksum(zio->io_spa, zio->io_vd, zio, zio->io_offset, zio->io_size, NULL, &info); } } return (ZIO_PIPELINE_CONTINUE); } /* * Called by RAID-Z to ensure we don't compute the checksum twice. */ void zio_checksum_verified(zio_t *zio) { zio->io_pipeline &= ~ZIO_STAGE_CHECKSUM_VERIFY; } /* * ========================================================================== * Error rank. Error are ranked in the order 0, ENXIO, ECKSUM, EIO, other. * An error of 0 indicates success. ENXIO indicates whole-device failure, * which may be transient (e.g. unplugged) or permament. ECKSUM and EIO * indicate errors that are specific to one I/O, and most likely permanent. * Any other error is presumed to be worse because we weren't expecting it. * ========================================================================== */ int zio_worst_error(int e1, int e2) { static int zio_error_rank[] = { 0, ENXIO, ECKSUM, EIO }; int r1, r2; for (r1 = 0; r1 < sizeof (zio_error_rank) / sizeof (int); r1++) if (e1 == zio_error_rank[r1]) break; for (r2 = 0; r2 < sizeof (zio_error_rank) / sizeof (int); r2++) if (e2 == zio_error_rank[r2]) break; return (r1 > r2 ? e1 : e2); } /* * ========================================================================== * I/O completion * ========================================================================== */ static int zio_ready(zio_t *zio) { blkptr_t *bp = zio->io_bp; zio_t *pio, *pio_next; if (zio_wait_for_children(zio, ZIO_CHILD_GANG, ZIO_WAIT_READY) || zio_wait_for_children(zio, ZIO_CHILD_DDT, ZIO_WAIT_READY)) return (ZIO_PIPELINE_STOP); if (zio->io_ready) { ASSERT(IO_IS_ALLOCATING(zio)); ASSERT(bp->blk_birth == zio->io_txg || BP_IS_HOLE(bp) || (zio->io_flags & ZIO_FLAG_NOPWRITE)); ASSERT(zio->io_children[ZIO_CHILD_GANG][ZIO_WAIT_READY] == 0); zio->io_ready(zio); } if (bp != NULL && bp != &zio->io_bp_copy) zio->io_bp_copy = *bp; if (zio->io_error) zio->io_pipeline = ZIO_INTERLOCK_PIPELINE; mutex_enter(&zio->io_lock); zio->io_state[ZIO_WAIT_READY] = 1; pio = zio_walk_parents(zio); mutex_exit(&zio->io_lock); /* * As we notify zio's parents, new parents could be added. * New parents go to the head of zio's io_parent_list, however, * so we will (correctly) not notify them. The remainder of zio's * io_parent_list, from 'pio_next' onward, cannot change because * all parents must wait for us to be done before they can be done. */ for (; pio != NULL; pio = pio_next) { pio_next = zio_walk_parents(zio); zio_notify_parent(pio, zio, ZIO_WAIT_READY); } if (zio->io_flags & ZIO_FLAG_NODATA) { if (BP_IS_GANG(bp)) { zio->io_flags &= ~ZIO_FLAG_NODATA; } else { ASSERT((uintptr_t)zio->io_data < SPA_MAXBLOCKSIZE); zio->io_pipeline &= ~ZIO_VDEV_IO_STAGES; } } if (zio_injection_enabled && zio->io_spa->spa_syncing_txg == zio->io_txg) zio_handle_ignored_writes(zio); return (ZIO_PIPELINE_CONTINUE); } static int zio_done(zio_t *zio) { zio_t *pio, *pio_next; int c, w; /* * If our children haven't all completed, * wait for them and then repeat this pipeline stage. */ if (zio_wait_for_children(zio, ZIO_CHILD_VDEV, ZIO_WAIT_DONE) || zio_wait_for_children(zio, ZIO_CHILD_GANG, ZIO_WAIT_DONE) || zio_wait_for_children(zio, ZIO_CHILD_DDT, ZIO_WAIT_DONE) || zio_wait_for_children(zio, ZIO_CHILD_LOGICAL, ZIO_WAIT_DONE)) return (ZIO_PIPELINE_STOP); for (c = 0; c < ZIO_CHILD_TYPES; c++) for (w = 0; w < ZIO_WAIT_TYPES; w++) ASSERT(zio->io_children[c][w] == 0); if (zio->io_bp != NULL && !BP_IS_EMBEDDED(zio->io_bp)) { ASSERT(zio->io_bp->blk_pad[0] == 0); ASSERT(zio->io_bp->blk_pad[1] == 0); ASSERT(bcmp(zio->io_bp, &zio->io_bp_copy, sizeof (blkptr_t)) == 0 || (zio->io_bp == zio_unique_parent(zio)->io_bp)); if (zio->io_type == ZIO_TYPE_WRITE && !BP_IS_HOLE(zio->io_bp) && zio->io_bp_override == NULL && !(zio->io_flags & ZIO_FLAG_IO_REPAIR)) { ASSERT(!BP_SHOULD_BYTESWAP(zio->io_bp)); ASSERT3U(zio->io_prop.zp_copies, <=, BP_GET_NDVAS(zio->io_bp)); ASSERT(BP_COUNT_GANG(zio->io_bp) == 0 || (BP_COUNT_GANG(zio->io_bp) == BP_GET_NDVAS(zio->io_bp))); } if (zio->io_flags & ZIO_FLAG_NOPWRITE) VERIFY(BP_EQUAL(zio->io_bp, &zio->io_bp_orig)); } /* * If there were child vdev/gang/ddt errors, they apply to us now. */ zio_inherit_child_errors(zio, ZIO_CHILD_VDEV); zio_inherit_child_errors(zio, ZIO_CHILD_GANG); zio_inherit_child_errors(zio, ZIO_CHILD_DDT); /* * If the I/O on the transformed data was successful, generate any * checksum reports now while we still have the transformed data. */ if (zio->io_error == 0) { while (zio->io_cksum_report != NULL) { zio_cksum_report_t *zcr = zio->io_cksum_report; uint64_t align = zcr->zcr_align; uint64_t asize = P2ROUNDUP(zio->io_size, align); char *abuf = zio->io_data; if (asize != zio->io_size) { abuf = zio_buf_alloc(asize); bcopy(zio->io_data, abuf, zio->io_size); bzero(abuf+zio->io_size, asize-zio->io_size); } zio->io_cksum_report = zcr->zcr_next; zcr->zcr_next = NULL; zcr->zcr_finish(zcr, abuf); zfs_ereport_free_checksum(zcr); if (asize != zio->io_size) zio_buf_free(abuf, asize); } } zio_pop_transforms(zio); /* note: may set zio->io_error */ vdev_stat_update(zio, zio->io_size); /* * If this I/O is attached to a particular vdev is slow, exceeding * 30 seconds to complete, post an error described the I/O delay. * We ignore these errors if the device is currently unavailable. */ if (zio->io_delay >= MSEC_TO_TICK(zio_delay_max)) { if (zio->io_vd != NULL && !vdev_is_dead(zio->io_vd)) zfs_ereport_post(FM_EREPORT_ZFS_DELAY, zio->io_spa, zio->io_vd, zio, 0, 0); } if (zio->io_error) { /* * If this I/O is attached to a particular vdev, * generate an error message describing the I/O failure * at the block level. We ignore these errors if the * device is currently unavailable. */ if (zio->io_error != ECKSUM && zio->io_vd != NULL && !vdev_is_dead(zio->io_vd)) zfs_ereport_post(FM_EREPORT_ZFS_IO, zio->io_spa, zio->io_vd, zio, 0, 0); if ((zio->io_error == EIO || !(zio->io_flags & (ZIO_FLAG_SPECULATIVE | ZIO_FLAG_DONT_PROPAGATE))) && zio == zio->io_logical) { /* * For logical I/O requests, tell the SPA to log the * error and generate a logical data ereport. */ spa_log_error(zio->io_spa, zio); zfs_ereport_post(FM_EREPORT_ZFS_DATA, zio->io_spa, NULL, zio, 0, 0); } } if (zio->io_error && zio == zio->io_logical) { /* * Determine whether zio should be reexecuted. This will * propagate all the way to the root via zio_notify_parent(). */ ASSERT(zio->io_vd == NULL && zio->io_bp != NULL); ASSERT(zio->io_child_type == ZIO_CHILD_LOGICAL); if (IO_IS_ALLOCATING(zio) && !(zio->io_flags & ZIO_FLAG_CANFAIL)) { if (zio->io_error != ENOSPC) zio->io_reexecute |= ZIO_REEXECUTE_NOW; else zio->io_reexecute |= ZIO_REEXECUTE_SUSPEND; } if ((zio->io_type == ZIO_TYPE_READ || zio->io_type == ZIO_TYPE_FREE) && !(zio->io_flags & ZIO_FLAG_SCAN_THREAD) && zio->io_error == ENXIO && spa_load_state(zio->io_spa) == SPA_LOAD_NONE && spa_get_failmode(zio->io_spa) != ZIO_FAILURE_MODE_CONTINUE) zio->io_reexecute |= ZIO_REEXECUTE_SUSPEND; if (!(zio->io_flags & ZIO_FLAG_CANFAIL) && !zio->io_reexecute) zio->io_reexecute |= ZIO_REEXECUTE_SUSPEND; /* * Here is a possibly good place to attempt to do * either combinatorial reconstruction or error correction * based on checksums. It also might be a good place * to send out preliminary ereports before we suspend * processing. */ } /* * If there were logical child errors, they apply to us now. * We defer this until now to avoid conflating logical child * errors with errors that happened to the zio itself when * updating vdev stats and reporting FMA events above. */ zio_inherit_child_errors(zio, ZIO_CHILD_LOGICAL); if ((zio->io_error || zio->io_reexecute) && IO_IS_ALLOCATING(zio) && zio->io_gang_leader == zio && !(zio->io_flags & (ZIO_FLAG_IO_REWRITE | ZIO_FLAG_NOPWRITE))) zio_dva_unallocate(zio, zio->io_gang_tree, zio->io_bp); zio_gang_tree_free(&zio->io_gang_tree); /* * Godfather I/Os should never suspend. */ if ((zio->io_flags & ZIO_FLAG_GODFATHER) && (zio->io_reexecute & ZIO_REEXECUTE_SUSPEND)) zio->io_reexecute = 0; if (zio->io_reexecute) { /* * This is a logical I/O that wants to reexecute. * * Reexecute is top-down. When an i/o fails, if it's not * the root, it simply notifies its parent and sticks around. * The parent, seeing that it still has children in zio_done(), * does the same. This percolates all the way up to the root. * The root i/o will reexecute or suspend the entire tree. * * This approach ensures that zio_reexecute() honors * all the original i/o dependency relationships, e.g. * parents not executing until children are ready. */ ASSERT(zio->io_child_type == ZIO_CHILD_LOGICAL); zio->io_gang_leader = NULL; mutex_enter(&zio->io_lock); zio->io_state[ZIO_WAIT_DONE] = 1; mutex_exit(&zio->io_lock); /* * "The Godfather" I/O monitors its children but is * not a true parent to them. It will track them through * the pipeline but severs its ties whenever they get into * trouble (e.g. suspended). This allows "The Godfather" * I/O to return status without blocking. */ for (pio = zio_walk_parents(zio); pio != NULL; pio = pio_next) { zio_link_t *zl = zio->io_walk_link; pio_next = zio_walk_parents(zio); if ((pio->io_flags & ZIO_FLAG_GODFATHER) && (zio->io_reexecute & ZIO_REEXECUTE_SUSPEND)) { zio_remove_child(pio, zio, zl); zio_notify_parent(pio, zio, ZIO_WAIT_DONE); } } if ((pio = zio_unique_parent(zio)) != NULL) { /* * We're not a root i/o, so there's nothing to do * but notify our parent. Don't propagate errors * upward since we haven't permanently failed yet. */ ASSERT(!(zio->io_flags & ZIO_FLAG_GODFATHER)); zio->io_flags |= ZIO_FLAG_DONT_PROPAGATE; zio_notify_parent(pio, zio, ZIO_WAIT_DONE); } else if (zio->io_reexecute & ZIO_REEXECUTE_SUSPEND) { /* * We'd fail again if we reexecuted now, so suspend * until conditions improve (e.g. device comes online). */ zio_suspend(zio->io_spa, zio); } else { /* * Reexecution is potentially a huge amount of work. * Hand it off to the otherwise-unused claim taskq. */ ASSERT(taskq_empty_ent(&zio->io_tqent)); spa_taskq_dispatch_ent(zio->io_spa, ZIO_TYPE_CLAIM, ZIO_TASKQ_ISSUE, (task_func_t *)zio_reexecute, zio, 0, &zio->io_tqent); } return (ZIO_PIPELINE_STOP); } ASSERT(zio->io_child_count == 0); ASSERT(zio->io_reexecute == 0); ASSERT(zio->io_error == 0 || (zio->io_flags & ZIO_FLAG_CANFAIL)); /* * Report any checksum errors, since the I/O is complete. */ while (zio->io_cksum_report != NULL) { zio_cksum_report_t *zcr = zio->io_cksum_report; zio->io_cksum_report = zcr->zcr_next; zcr->zcr_next = NULL; zcr->zcr_finish(zcr, NULL); zfs_ereport_free_checksum(zcr); } if (zio->io_flags & ZIO_FLAG_FASTWRITE && zio->io_bp && !BP_IS_HOLE(zio->io_bp) && !BP_IS_EMBEDDED(zio->io_bp) && !(zio->io_flags & ZIO_FLAG_NOPWRITE)) { metaslab_fastwrite_unmark(zio->io_spa, zio->io_bp); } /* * It is the responsibility of the done callback to ensure that this * particular zio is no longer discoverable for adoption, and as * such, cannot acquire any new parents. */ if (zio->io_done) zio->io_done(zio); mutex_enter(&zio->io_lock); zio->io_state[ZIO_WAIT_DONE] = 1; mutex_exit(&zio->io_lock); for (pio = zio_walk_parents(zio); pio != NULL; pio = pio_next) { zio_link_t *zl = zio->io_walk_link; pio_next = zio_walk_parents(zio); zio_remove_child(pio, zio, zl); zio_notify_parent(pio, zio, ZIO_WAIT_DONE); } if (zio->io_waiter != NULL) { mutex_enter(&zio->io_lock); zio->io_executor = NULL; cv_broadcast(&zio->io_cv); mutex_exit(&zio->io_lock); } else { zio_destroy(zio); } return (ZIO_PIPELINE_STOP); } /* * ========================================================================== * I/O pipeline definition * ========================================================================== */ static zio_pipe_stage_t *zio_pipeline[] = { NULL, zio_read_bp_init, zio_free_bp_init, zio_issue_async, zio_write_bp_init, zio_checksum_generate, zio_nop_write, zio_ddt_read_start, zio_ddt_read_done, zio_ddt_write, zio_ddt_free, zio_gang_assemble, zio_gang_issue, zio_dva_allocate, zio_dva_free, zio_dva_claim, zio_ready, zio_vdev_io_start, zio_vdev_io_done, zio_vdev_io_assess, zio_checksum_verify, zio_done }; /* dnp is the dnode for zb1->zb_object */ boolean_t zbookmark_is_before(const dnode_phys_t *dnp, const zbookmark_t *zb1, const zbookmark_t *zb2) { uint64_t zb1nextL0, zb2thisobj; ASSERT(zb1->zb_objset == zb2->zb_objset); ASSERT(zb2->zb_level == 0); - /* - * A bookmark in the deadlist is considered to be after - * everything else. - */ - if (zb2->zb_object == DMU_DEADLIST_OBJECT) - return (B_TRUE); - /* The objset_phys_t isn't before anything. */ if (dnp == NULL) return (B_FALSE); zb1nextL0 = (zb1->zb_blkid + 1) << ((zb1->zb_level) * (dnp->dn_indblkshift - SPA_BLKPTRSHIFT)); zb2thisobj = zb2->zb_object ? zb2->zb_object : zb2->zb_blkid << (DNODE_BLOCK_SHIFT - DNODE_SHIFT); if (zb1->zb_object == DMU_META_DNODE_OBJECT) { uint64_t nextobj = zb1nextL0 * (dnp->dn_datablkszsec << SPA_MINBLOCKSHIFT) >> DNODE_SHIFT; return (nextobj <= zb2thisobj); } if (zb1->zb_object < zb2thisobj) return (B_TRUE); if (zb1->zb_object > zb2thisobj) return (B_FALSE); if (zb2->zb_object == DMU_META_DNODE_OBJECT) return (B_FALSE); return (zb1nextL0 <= zb2->zb_blkid); } #if defined(_KERNEL) && defined(HAVE_SPL) /* Fault injection */ EXPORT_SYMBOL(zio_injection_enabled); EXPORT_SYMBOL(zio_inject_fault); EXPORT_SYMBOL(zio_inject_list_next); EXPORT_SYMBOL(zio_clear_fault); EXPORT_SYMBOL(zio_handle_fault_injection); EXPORT_SYMBOL(zio_handle_device_injection); EXPORT_SYMBOL(zio_handle_label_injection); EXPORT_SYMBOL(zio_type_name); module_param(zio_bulk_flags, int, 0644); MODULE_PARM_DESC(zio_bulk_flags, "Additional flags to pass to bulk buffers"); module_param(zio_delay_max, int, 0644); MODULE_PARM_DESC(zio_delay_max, "Max zio millisec delay before posting event"); module_param(zio_requeue_io_start_cut_in_line, int, 0644); MODULE_PARM_DESC(zio_requeue_io_start_cut_in_line, "Prioritize requeued I/O"); module_param(zfs_sync_pass_deferred_free, int, 0644); MODULE_PARM_DESC(zfs_sync_pass_deferred_free, "Defer frees starting in this pass"); module_param(zfs_sync_pass_dont_compress, int, 0644); MODULE_PARM_DESC(zfs_sync_pass_dont_compress, "Don't compress starting in this pass"); module_param(zfs_sync_pass_rewrite, int, 0644); MODULE_PARM_DESC(zfs_sync_pass_rewrite, "Rewrite new bps starting in this pass"); #endif