4 * The contents of this file are subject to the terms of the
5 * Common Development and Distribution License (the "License").
6 * You may not use this file except in compliance with the License.
8 * You can obtain a copy of the license at usr/src/OPENSOLARIS.LICENSE
9 * or https://opensource.org/licenses/CDDL-1.0.
10 * See the License for the specific language governing permissions
11 * and limitations under the License.
13 * When distributing Covered Code, include this CDDL HEADER in each
14 * file and include the License file at usr/src/OPENSOLARIS.LICENSE.
15 * If applicable, add the following below this CDDL HEADER, with the
16 * fields enclosed by brackets "[]" replaced with your own identifying
17 * information: Portions Copyright [yyyy] [name of copyright owner]
22 * Copyright (c) 2005, 2010, Oracle and/or its affiliates. All rights reserved.
23 * Copyright (c) 2011, 2019 by Delphix. All rights reserved.
24 * Copyright 2011 Nexenta Systems, Inc. All rights reserved.
25 * Copyright (c) 2014 Spectra Logic Corporation, All rights reserved.
26 * Copyright 2013 Saso Kiselkov. All rights reserved.
27 * Copyright (c) 2016 Actifio, Inc. All rights reserved.
28 * Copyright (c) 2017, Intel Corporation.
29 * Copyright (c) 2019 Datto Inc.
32 #ifndef _SYS_SPA_IMPL_H
33 #define _SYS_SPA_IMPL_H
36 #include <sys/spa_checkpoint.h>
37 #include <sys/spa_log_spacemap.h>
39 #include <sys/vdev_rebuild.h>
40 #include <sys/vdev_removal.h>
41 #include <sys/metaslab.h>
43 #include <sys/dsl_pool.h>
44 #include <sys/uberblock_impl.h>
45 #include <sys/zfs_context.h>
47 #include <sys/zfs_refcount.h>
48 #include <sys/bplist.h>
49 #include <sys/bpobj.h>
50 #include <sys/dsl_crypt.h>
51 #include <sys/zfeature.h>
53 #include <sys/dsl_deadlist.h>
54 #include <zfeature_common.h>
60 typedef struct spa_alloc
{
63 } ____cacheline_aligned spa_alloc_t
;
65 typedef struct spa_error_entry
{
66 zbookmark_phys_t se_bookmark
;
69 zbookmark_err_phys_t se_zep
; /* not accounted in avl_find */
72 typedef struct spa_history_phys
{
73 uint64_t sh_pool_create_len
; /* ending offset of zpool create */
74 uint64_t sh_phys_max_off
; /* physical EOF */
75 uint64_t sh_bof
; /* logical BOF */
76 uint64_t sh_eof
; /* logical EOF */
77 uint64_t sh_records_lost
; /* num of records overwritten */
81 * All members must be uint64_t, for byteswap purposes.
83 typedef struct spa_removing_phys
{
84 uint64_t sr_state
; /* dsl_scan_state_t */
87 * The vdev ID that we most recently attempted to remove,
88 * or -1 if no removal has been attempted.
90 uint64_t sr_removing_vdev
;
93 * The vdev ID that we most recently successfully removed,
94 * or -1 if no devices have been removed.
96 uint64_t sr_prev_indirect_vdev
;
98 uint64_t sr_start_time
;
102 * Note that we can not use the space map's or indirect mapping's
103 * accounting as a substitute for these values, because we need to
104 * count frees of not-yet-copied data as though it did the copy.
105 * Otherwise, we could get into a situation where copied > to_copy,
106 * or we complete before copied == to_copy.
108 uint64_t sr_to_copy
; /* bytes that need to be copied */
109 uint64_t sr_copied
; /* bytes that have been copied or freed */
110 } spa_removing_phys_t
;
113 * This struct is stored as an entry in the DMU_POOL_DIRECTORY_OBJECT
114 * (with key DMU_POOL_CONDENSING_INDIRECT). It is present if a condense
115 * of an indirect vdev's mapping object is in progress.
117 typedef struct spa_condensing_indirect_phys
{
119 * The vdev ID of the indirect vdev whose indirect mapping is
125 * The vdev's old obsolete spacemap. This spacemap's contents are
126 * being integrated into the new mapping.
128 uint64_t scip_prev_obsolete_sm_object
;
131 * The new mapping object that is being created.
133 uint64_t scip_next_mapping_object
;
134 } spa_condensing_indirect_phys_t
;
136 struct spa_aux_vdev
{
137 uint64_t sav_object
; /* MOS object for device list */
138 nvlist_t
*sav_config
; /* cached device config */
139 vdev_t
**sav_vdevs
; /* devices */
140 int sav_count
; /* number devices */
141 boolean_t sav_sync
; /* sync the device list */
142 nvlist_t
**sav_pending
; /* pending device additions */
143 uint_t sav_npending
; /* # pending devices */
146 typedef struct spa_config_lock
{
148 kthread_t
*scl_writer
;
149 int scl_write_wanted
;
152 } ____cacheline_aligned spa_config_lock_t
;
154 typedef struct spa_config_dirent
{
155 list_node_t scd_link
;
157 } spa_config_dirent_t
;
159 typedef enum zio_taskq_type
{
161 ZIO_TASKQ_ISSUE_HIGH
,
163 ZIO_TASKQ_INTERRUPT_HIGH
,
168 * State machine for the zpool-poolname process. The states transitions
169 * are done as follows:
172 * PROC_NONE -> PROC_CREATED spa_activate()
173 * PROC_CREATED -> PROC_ACTIVE spa_thread()
174 * PROC_ACTIVE -> PROC_DEACTIVATE spa_deactivate()
175 * PROC_DEACTIVATE -> PROC_GONE spa_thread()
176 * PROC_GONE -> PROC_NONE spa_deactivate()
178 typedef enum spa_proc_state
{
179 SPA_PROC_NONE
, /* spa_proc = &p0, no process created */
180 SPA_PROC_CREATED
, /* spa_activate() has proc, is waiting */
181 SPA_PROC_ACTIVE
, /* taskqs created, spa_proc set */
182 SPA_PROC_DEACTIVATE
, /* spa_deactivate() requests process exit */
183 SPA_PROC_GONE
/* spa_thread() is exiting, spa_proc = &p0 */
186 typedef struct spa_taskqs
{
188 taskq_t
**stqs_taskq
;
191 typedef enum spa_all_vdev_zap_action
{
193 AVZ_ACTION_DESTROY
, /* Destroy all per-vdev ZAPs and the AVZ. */
194 AVZ_ACTION_REBUILD
, /* Populate the new AVZ, see spa_avz_rebuild */
195 AVZ_ACTION_INITIALIZE
198 typedef enum spa_config_source
{
199 SPA_CONFIG_SRC_NONE
= 0,
200 SPA_CONFIG_SRC_SCAN
, /* scan of path (default: /dev/dsk) */
201 SPA_CONFIG_SRC_CACHEFILE
, /* any cachefile */
202 SPA_CONFIG_SRC_TRYIMPORT
, /* returned from call to tryimport */
203 SPA_CONFIG_SRC_SPLIT
, /* new pool in a pool split */
204 SPA_CONFIG_SRC_MOS
/* MOS, but not always from right txg */
205 } spa_config_source_t
;
209 * Fields protected by spa_namespace_lock.
211 char spa_name
[ZFS_MAX_DATASET_NAME_LEN
]; /* pool name */
212 char *spa_comment
; /* comment */
213 avl_node_t spa_avl
; /* node in spa_namespace_avl */
214 nvlist_t
*spa_config
; /* last synced config */
215 nvlist_t
*spa_config_syncing
; /* currently syncing config */
216 nvlist_t
*spa_config_splitting
; /* config for splitting */
217 nvlist_t
*spa_load_info
; /* info and errors from load */
218 uint64_t spa_config_txg
; /* txg of last config change */
219 uint32_t spa_sync_pass
; /* iterate-to-convergence */
220 pool_state_t spa_state
; /* pool state */
221 int spa_inject_ref
; /* injection references */
222 uint8_t spa_sync_on
; /* sync threads are running */
223 spa_load_state_t spa_load_state
; /* current load operation */
224 boolean_t spa_indirect_vdevs_loaded
; /* mappings loaded? */
225 boolean_t spa_trust_config
; /* do we trust vdev tree? */
226 boolean_t spa_is_splitting
; /* in the middle of a split? */
227 spa_config_source_t spa_config_source
; /* where config comes from? */
228 uint64_t spa_import_flags
; /* import specific flags */
229 spa_taskqs_t spa_zio_taskq
[ZIO_TYPES
][ZIO_TASKQ_TYPES
];
230 dsl_pool_t
*spa_dsl_pool
;
231 boolean_t spa_is_initializing
; /* true while opening pool */
232 boolean_t spa_is_exporting
; /* true while exporting pool */
233 metaslab_class_t
*spa_normal_class
; /* normal data class */
234 metaslab_class_t
*spa_log_class
; /* intent log data class */
235 metaslab_class_t
*spa_embedded_log_class
; /* log on normal vdevs */
236 metaslab_class_t
*spa_special_class
; /* special allocation class */
237 metaslab_class_t
*spa_dedup_class
; /* dedup allocation class */
238 uint64_t spa_first_txg
; /* first txg after spa_open() */
239 uint64_t spa_final_txg
; /* txg of export/destroy */
240 uint64_t spa_freeze_txg
; /* freeze pool at this txg */
241 uint64_t spa_load_max_txg
; /* best initial ub_txg */
242 uint64_t spa_claim_max_txg
; /* highest claimed birth txg */
243 inode_timespec_t spa_loaded_ts
; /* 1st successful open time */
244 objset_t
*spa_meta_objset
; /* copy of dp->dp_meta_objset */
245 kmutex_t spa_evicting_os_lock
; /* Evicting objset list lock */
246 list_t spa_evicting_os_list
; /* Objsets being evicted. */
247 kcondvar_t spa_evicting_os_cv
; /* Objset Eviction Completion */
248 txg_list_t spa_vdev_txg_list
; /* per-txg dirty vdev list */
249 vdev_t
*spa_root_vdev
; /* top-level vdev container */
250 uint64_t spa_min_ashift
; /* of vdevs in normal class */
251 uint64_t spa_max_ashift
; /* of vdevs in normal class */
252 uint64_t spa_min_alloc
; /* of vdevs in normal class */
253 uint64_t spa_gcd_alloc
; /* of vdevs in normal class */
254 uint64_t spa_config_guid
; /* config pool guid */
255 uint64_t spa_load_guid
; /* spa_load initialized guid */
256 uint64_t spa_last_synced_guid
; /* last synced guid */
257 list_t spa_config_dirty_list
; /* vdevs with dirty config */
258 list_t spa_state_dirty_list
; /* vdevs with dirty state */
260 * spa_allocs is an array, whose lengths is stored in spa_alloc_count.
261 * There is one tree and one lock for each allocator, to help improve
262 * allocation performance in write-heavy workloads.
264 spa_alloc_t
*spa_allocs
;
267 spa_aux_vdev_t spa_spares
; /* hot spares */
268 spa_aux_vdev_t spa_l2cache
; /* L2ARC cache devices */
269 boolean_t spa_aux_sync_uber
; /* need to sync aux uber */
270 nvlist_t
*spa_label_features
; /* Features for reading MOS */
271 uint64_t spa_config_object
; /* MOS object for pool config */
272 uint64_t spa_config_generation
; /* config generation number */
273 uint64_t spa_syncing_txg
; /* txg currently syncing */
274 bpobj_t spa_deferred_bpobj
; /* deferred-free bplist */
275 bplist_t spa_free_bplist
[TXG_SIZE
]; /* bplist of stuff to free */
276 zio_cksum_salt_t spa_cksum_salt
; /* secret salt for cksum */
277 /* checksum context templates */
278 kmutex_t spa_cksum_tmpls_lock
;
279 void *spa_cksum_tmpls
[ZIO_CHECKSUM_FUNCTIONS
];
280 uberblock_t spa_ubsync
; /* last synced uberblock */
281 uberblock_t spa_uberblock
; /* current uberblock */
282 boolean_t spa_extreme_rewind
; /* rewind past deferred frees */
283 kmutex_t spa_scrub_lock
; /* resilver/scrub lock */
284 uint64_t spa_scrub_inflight
; /* in-flight scrub bytes */
286 /* in-flight verification bytes */
287 uint64_t spa_load_verify_bytes
;
288 kcondvar_t spa_scrub_io_cv
; /* scrub I/O completion */
289 uint8_t spa_scrub_active
; /* active or suspended? */
290 uint8_t spa_scrub_type
; /* type of scrub we're doing */
291 uint8_t spa_scrub_finished
; /* indicator to rotate logs */
292 uint8_t spa_scrub_started
; /* started since last boot */
293 uint8_t spa_scrub_reopen
; /* scrub doing vdev_reopen */
294 uint64_t spa_scan_pass_start
; /* start time per pass/reboot */
295 uint64_t spa_scan_pass_scrub_pause
; /* scrub pause time */
296 uint64_t spa_scan_pass_scrub_spent_paused
; /* total paused */
297 uint64_t spa_scan_pass_exam
; /* examined bytes per pass */
298 uint64_t spa_scan_pass_issued
; /* issued bytes per pass */
300 /* error scrub pause time in milliseconds */
301 uint64_t spa_scan_pass_errorscrub_pause
;
302 /* total error scrub paused time in milliseconds */
303 uint64_t spa_scan_pass_errorscrub_spent_paused
;
305 * We are in the middle of a resilver, and another resilver
306 * is needed once this one completes. This is set iff any
307 * vdev_resilver_deferred is set.
309 boolean_t spa_resilver_deferred
;
310 kmutex_t spa_async_lock
; /* protect async state */
311 kthread_t
*spa_async_thread
; /* thread doing async task */
312 int spa_async_suspended
; /* async tasks suspended */
313 kcondvar_t spa_async_cv
; /* wait for thread_exit() */
314 uint16_t spa_async_tasks
; /* async task mask */
315 uint64_t spa_missing_tvds
; /* unopenable tvds on load */
316 uint64_t spa_missing_tvds_allowed
; /* allow loading spa? */
318 uint64_t spa_nonallocating_dspace
;
319 spa_removing_phys_t spa_removing_phys
;
320 spa_vdev_removal_t
*spa_vdev_removal
;
322 spa_condensing_indirect_phys_t spa_condensing_indirect_phys
;
323 spa_condensing_indirect_t
*spa_condensing_indirect
;
324 zthr_t
*spa_condense_zthr
; /* zthr doing condense. */
326 uint64_t spa_checkpoint_txg
; /* the txg of the checkpoint */
327 spa_checkpoint_info_t spa_checkpoint_info
; /* checkpoint accounting */
328 zthr_t
*spa_checkpoint_discard_zthr
;
330 space_map_t
*spa_syncing_log_sm
; /* current log space map */
331 avl_tree_t spa_sm_logs_by_txg
;
332 kmutex_t spa_flushed_ms_lock
; /* for metaslabs_by_flushed */
333 avl_tree_t spa_metaslabs_by_flushed
;
334 spa_unflushed_stats_t spa_unflushed_stats
;
335 list_t spa_log_summary
;
336 uint64_t spa_log_flushall_txg
;
338 zthr_t
*spa_livelist_delete_zthr
; /* deleting livelists */
339 zthr_t
*spa_livelist_condense_zthr
; /* condensing livelists */
340 uint64_t spa_livelists_to_delete
; /* set of livelists to free */
341 livelist_condense_entry_t spa_to_condense
; /* next to condense */
343 char *spa_root
; /* alternate root directory */
344 uint64_t spa_ena
; /* spa-wide ereport ENA */
345 int spa_last_open_failed
; /* error if last open failed */
346 uint64_t spa_last_ubsync_txg
; /* "best" uberblock txg */
347 uint64_t spa_last_ubsync_txg_ts
; /* timestamp from that ub */
348 uint64_t spa_load_txg
; /* ub txg that loaded */
349 uint64_t spa_load_txg_ts
; /* timestamp from that ub */
350 uint64_t spa_load_meta_errors
; /* verify metadata err count */
351 uint64_t spa_load_data_errors
; /* verify data err count */
352 uint64_t spa_verify_min_txg
; /* start txg of verify scrub */
353 kmutex_t spa_errlog_lock
; /* error log lock */
354 uint64_t spa_errlog_last
; /* last error log object */
355 uint64_t spa_errlog_scrub
; /* scrub error log object */
356 kmutex_t spa_errlist_lock
; /* error list/ereport lock */
357 avl_tree_t spa_errlist_last
; /* last error list */
358 avl_tree_t spa_errlist_scrub
; /* scrub error list */
359 avl_tree_t spa_errlist_healed
; /* list of healed blocks */
360 uint64_t spa_deflate
; /* should we deflate? */
361 uint64_t spa_history
; /* history object */
362 kmutex_t spa_history_lock
; /* history lock */
363 vdev_t
*spa_pending_vdev
; /* pending vdev additions */
364 kmutex_t spa_props_lock
; /* property lock */
365 uint64_t spa_pool_props_object
; /* object for properties */
366 uint64_t spa_bootfs
; /* default boot filesystem */
367 uint64_t spa_failmode
; /* failure mode for the pool */
368 uint64_t spa_deadman_failmode
; /* failure mode for deadman */
369 uint64_t spa_delegation
; /* delegation on/off */
370 list_t spa_config_list
; /* previous cache file(s) */
371 /* per-CPU array of root of async I/O: */
372 zio_t
**spa_async_zio_root
;
373 zio_t
*spa_suspend_zio_root
; /* root of all suspended I/O */
374 zio_t
*spa_txg_zio
[TXG_SIZE
]; /* spa_sync() waits for this */
375 kmutex_t spa_suspend_lock
; /* protects suspend_zio_root */
376 kcondvar_t spa_suspend_cv
; /* notification of resume */
377 zio_suspend_reason_t spa_suspended
; /* pool is suspended */
378 uint8_t spa_claiming
; /* pool is doing zil_claim() */
379 boolean_t spa_is_root
; /* pool is root */
380 int spa_minref
; /* num refs when first opened */
381 spa_mode_t spa_mode
; /* SPA_MODE_{READ|WRITE} */
382 boolean_t spa_read_spacemaps
; /* spacemaps available if ro */
383 spa_log_state_t spa_log_state
; /* log state */
384 uint64_t spa_autoexpand
; /* lun expansion on/off */
385 ddt_t
*spa_ddt
[ZIO_CHECKSUM_FUNCTIONS
]; /* in-core DDTs */
386 uint64_t spa_ddt_stat_object
; /* DDT statistics */
387 uint64_t spa_dedup_dspace
; /* Cache get_dedup_dspace() */
388 uint64_t spa_dedup_checksum
; /* default dedup checksum */
389 uint64_t spa_dspace
; /* dspace in normal class */
390 struct brt
*spa_brt
; /* in-core BRT */
391 kmutex_t spa_vdev_top_lock
; /* dueling offline/remove */
392 kmutex_t spa_proc_lock
; /* protects spa_proc* */
393 kcondvar_t spa_proc_cv
; /* spa_proc_state transitions */
394 spa_proc_state_t spa_proc_state
; /* see definition */
395 proc_t
*spa_proc
; /* "zpool-poolname" process */
396 uintptr_t spa_did
; /* if procp != p0, did of t1 */
397 boolean_t spa_autoreplace
; /* autoreplace set in open */
398 int spa_vdev_locks
; /* locks grabbed */
399 uint64_t spa_creation_version
; /* version at pool creation */
400 uint64_t spa_prev_software_version
; /* See ub_software_version */
401 uint64_t spa_feat_for_write_obj
; /* required to write to pool */
402 uint64_t spa_feat_for_read_obj
; /* required to read from pool */
403 uint64_t spa_feat_desc_obj
; /* Feature descriptions */
404 uint64_t spa_feat_enabled_txg_obj
; /* Feature enabled txg */
405 kmutex_t spa_feat_stats_lock
; /* protects spa_feat_stats */
406 nvlist_t
*spa_feat_stats
; /* Cache of enabled features */
407 /* cache feature refcounts */
408 uint64_t spa_feat_refcount_cache
[SPA_FEATURES
];
409 taskqid_t spa_deadman_tqid
; /* Task id */
410 uint64_t spa_deadman_calls
; /* number of deadman calls */
411 hrtime_t spa_sync_starttime
; /* starting time of spa_sync */
412 uint64_t spa_deadman_synctime
; /* deadman sync expiration */
413 uint64_t spa_deadman_ziotime
; /* deadman zio expiration */
414 uint64_t spa_all_vdev_zaps
; /* ZAP of per-vd ZAP obj #s */
415 spa_avz_action_t spa_avz_action
; /* destroy/rebuild AVZ? */
416 uint64_t spa_autotrim
; /* automatic background trim? */
417 uint64_t spa_errata
; /* errata issues detected */
418 spa_stats_t spa_stats
; /* assorted spa statistics */
419 spa_keystore_t spa_keystore
; /* loaded crypto keys */
421 /* arc_memory_throttle() parameters during low memory condition */
422 uint64_t spa_lowmem_page_load
; /* memory load during txg */
423 uint64_t spa_lowmem_last_txg
; /* txg window start */
425 hrtime_t spa_ccw_fail_time
; /* Conf cache write fail time */
426 taskq_t
*spa_zvol_taskq
; /* Taskq for minor management */
427 taskq_t
*spa_metaslab_taskq
; /* Taskq for metaslab preload */
428 taskq_t
*spa_prefetch_taskq
; /* Taskq for prefetch threads */
429 taskq_t
*spa_upgrade_taskq
; /* Taskq for upgrade jobs */
430 uint64_t spa_multihost
; /* multihost aware (mmp) */
431 mmp_thread_t spa_mmp
; /* multihost mmp thread */
432 list_t spa_leaf_list
; /* list of leaf vdevs */
433 uint64_t spa_leaf_list_gen
; /* track leaf_list changes */
434 uint32_t spa_hostid
; /* cached system hostid */
436 /* synchronization for threads in spa_wait */
437 kmutex_t spa_activities_lock
;
438 kcondvar_t spa_activities_cv
;
439 kcondvar_t spa_waiters_cv
;
440 int spa_waiters
; /* number of waiting threads */
441 boolean_t spa_waiters_cancel
; /* waiters should return */
443 char *spa_compatibility
; /* compatibility file(s) */
446 * spa_refcount & spa_config_lock must be the last elements
447 * because zfs_refcount_t changes size based on compilation options.
448 * In order for the MDB module to function correctly, the other
449 * fields must remain in the same location.
451 spa_config_lock_t spa_config_lock
[SCL_LOCKS
]; /* config changes */
452 zfs_refcount_t spa_refcount
; /* number of opens */
455 extern char *spa_config_path
;
456 extern const char *zfs_deadman_failmode
;
457 extern uint_t spa_slop_shift
;
458 extern void spa_taskq_dispatch_ent(spa_t
*spa
, zio_type_t t
, zio_taskq_type_t q
,
459 task_func_t
*func
, void *arg
, uint_t flags
, taskq_ent_t
*ent
);
460 extern void spa_taskq_dispatch_sync(spa_t
*, zio_type_t t
, zio_taskq_type_t q
,
461 task_func_t
*func
, void *arg
, uint_t flags
);
462 extern void spa_load_spares(spa_t
*spa
);
463 extern void spa_load_l2cache(spa_t
*spa
);
464 extern sysevent_t
*spa_event_create(spa_t
*spa
, vdev_t
*vd
, nvlist_t
*hist_nvl
,
466 extern void spa_event_post(sysevent_t
*ev
);
467 extern int param_set_deadman_failmode_common(const char *val
);
468 extern void spa_set_deadman_synctime(hrtime_t ns
);
469 extern void spa_set_deadman_ziotime(hrtime_t ns
);
470 extern const char *spa_history_zone(void);
476 #endif /* _SYS_SPA_IMPL_H */