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