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 http://www.opensolaris.org/os/licensing.
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, 2018 by Delphix. All rights reserved.
24 * Copyright 2011 Nexenta Systems, Inc. All rights reserved.
25 * Copyright 2013 Martin Matuska <mm@FreeBSD.org>. All rights reserved.
26 * Copyright (c) 2014 Spectra Logic Corporation, All rights reserved.
27 * Copyright 2013 Saso Kiselkov. All rights reserved.
28 * Copyright (c) 2017 Datto Inc.
31 #ifndef _SYS_SPA_IMPL_H
32 #define _SYS_SPA_IMPL_H
35 #include <sys/spa_checkpoint.h>
37 #include <sys/vdev_removal.h>
38 #include <sys/metaslab.h>
40 #include <sys/dsl_pool.h>
41 #include <sys/uberblock_impl.h>
42 #include <sys/zfs_context.h>
44 #include <sys/refcount.h>
45 #include <sys/bplist.h>
46 #include <sys/bpobj.h>
47 #include <sys/zfeature.h>
49 #include <zfeature_common.h>
55 typedef struct spa_error_entry {
56 zbookmark_phys_t se_bookmark;
61 typedef struct spa_history_phys {
62 uint64_t sh_pool_create_len; /* ending offset of zpool create */
63 uint64_t sh_phys_max_off; /* physical EOF */
64 uint64_t sh_bof; /* logical BOF */
65 uint64_t sh_eof; /* logical EOF */
66 uint64_t sh_records_lost; /* num of records overwritten */
70 * All members must be uint64_t, for byteswap purposes.
72 typedef struct spa_removing_phys {
73 uint64_t sr_state; /* dsl_scan_state_t */
76 * The vdev ID that we most recently attempted to remove,
77 * or -1 if no removal has been attempted.
79 uint64_t sr_removing_vdev;
82 * The vdev ID that we most recently successfully removed,
83 * or -1 if no devices have been removed.
85 uint64_t sr_prev_indirect_vdev;
87 uint64_t sr_start_time;
91 * Note that we can not use the space map's or indirect mapping's
92 * accounting as a substitute for these values, because we need to
93 * count frees of not-yet-copied data as though it did the copy.
94 * Otherwise, we could get into a situation where copied > to_copy,
95 * or we complete before copied == to_copy.
97 uint64_t sr_to_copy; /* bytes that need to be copied */
98 uint64_t sr_copied; /* bytes that have been copied or freed */
99 } spa_removing_phys_t;
102 * This struct is stored as an entry in the DMU_POOL_DIRECTORY_OBJECT
103 * (with key DMU_POOL_CONDENSING_INDIRECT). It is present if a condense
104 * of an indirect vdev's mapping object is in progress.
106 typedef struct spa_condensing_indirect_phys {
108 * The vdev ID of the indirect vdev whose indirect mapping is
114 * The vdev's old obsolete spacemap. This spacemap's contents are
115 * being integrated into the new mapping.
117 uint64_t scip_prev_obsolete_sm_object;
120 * The new mapping object that is being created.
122 uint64_t scip_next_mapping_object;
123 } spa_condensing_indirect_phys_t;
125 struct spa_aux_vdev {
126 uint64_t sav_object; /* MOS object for device list */
127 nvlist_t *sav_config; /* cached device config */
128 vdev_t **sav_vdevs; /* devices */
129 int sav_count; /* number devices */
130 boolean_t sav_sync; /* sync the device list */
131 nvlist_t **sav_pending; /* pending device additions */
132 uint_t sav_npending; /* # pending devices */
135 typedef struct spa_config_lock {
137 kthread_t *scl_writer;
138 int scl_write_wanted;
140 refcount_t scl_count;
143 typedef struct spa_config_dirent {
144 list_node_t scd_link;
146 } spa_config_dirent_t;
148 typedef enum zio_taskq_type {
150 ZIO_TASKQ_ISSUE_HIGH,
152 ZIO_TASKQ_INTERRUPT_HIGH,
157 * State machine for the zpool-poolname process. The states transitions
158 * are done as follows:
161 * PROC_NONE -> PROC_CREATED spa_activate()
162 * PROC_CREATED -> PROC_ACTIVE spa_thread()
163 * PROC_ACTIVE -> PROC_DEACTIVATE spa_deactivate()
164 * PROC_DEACTIVATE -> PROC_GONE spa_thread()
165 * PROC_GONE -> PROC_NONE spa_deactivate()
167 typedef enum spa_proc_state {
168 SPA_PROC_NONE, /* spa_proc = &p0, no process created */
169 SPA_PROC_CREATED, /* spa_activate() has proc, is waiting */
170 SPA_PROC_ACTIVE, /* taskqs created, spa_proc set */
171 SPA_PROC_DEACTIVATE, /* spa_deactivate() requests process exit */
172 SPA_PROC_GONE /* spa_thread() is exiting, spa_proc = &p0 */
175 typedef struct spa_taskqs {
177 taskq_t **stqs_taskq;
180 typedef enum spa_all_vdev_zap_action {
182 AVZ_ACTION_DESTROY, /* Destroy all per-vdev ZAPs and the AVZ. */
183 AVZ_ACTION_REBUILD, /* Populate the new AVZ, see spa_avz_rebuild */
184 AVZ_ACTION_INITIALIZE
187 typedef enum spa_config_source {
188 SPA_CONFIG_SRC_NONE = 0,
189 SPA_CONFIG_SRC_SCAN, /* scan of path (default: /dev/dsk) */
190 SPA_CONFIG_SRC_CACHEFILE, /* any cachefile */
191 SPA_CONFIG_SRC_TRYIMPORT, /* returned from call to tryimport */
192 SPA_CONFIG_SRC_SPLIT, /* new pool in a pool split */
193 SPA_CONFIG_SRC_MOS /* MOS, but not always from right txg */
194 } spa_config_source_t;
198 * Fields protected by spa_namespace_lock.
200 char spa_name[ZFS_MAX_DATASET_NAME_LEN]; /* pool name */
201 char *spa_comment; /* comment */
202 avl_node_t spa_avl; /* node in spa_namespace_avl */
203 nvlist_t *spa_config; /* last synced config */
204 nvlist_t *spa_config_syncing; /* currently syncing config */
205 nvlist_t *spa_config_splitting; /* config for splitting */
206 nvlist_t *spa_load_info; /* info and errors from load */
207 uint64_t spa_config_txg; /* txg of last config change */
208 int spa_sync_pass; /* iterate-to-convergence */
209 pool_state_t spa_state; /* pool state */
210 int spa_inject_ref; /* injection references */
211 uint8_t spa_sync_on; /* sync threads are running */
212 spa_load_state_t spa_load_state; /* current load operation */
213 boolean_t spa_indirect_vdevs_loaded; /* mappings loaded? */
214 boolean_t spa_trust_config; /* do we trust vdev tree? */
215 spa_config_source_t spa_config_source; /* where config comes from? */
216 uint64_t spa_import_flags; /* import specific flags */
217 spa_taskqs_t spa_zio_taskq[ZIO_TYPES][ZIO_TASKQ_TYPES];
218 dsl_pool_t *spa_dsl_pool;
219 boolean_t spa_is_initializing; /* true while opening pool */
220 metaslab_class_t *spa_normal_class; /* normal data class */
221 metaslab_class_t *spa_log_class; /* intent log data class */
222 uint64_t spa_first_txg; /* first txg after spa_open() */
223 uint64_t spa_final_txg; /* txg of export/destroy */
224 uint64_t spa_freeze_txg; /* freeze pool at this txg */
225 uint64_t spa_load_max_txg; /* best initial ub_txg */
226 uint64_t spa_claim_max_txg; /* highest claimed birth txg */
227 timespec_t spa_loaded_ts; /* 1st successful open time */
228 objset_t *spa_meta_objset; /* copy of dp->dp_meta_objset */
229 kmutex_t spa_evicting_os_lock; /* Evicting objset list lock */
230 list_t spa_evicting_os_list; /* Objsets being evicted. */
231 kcondvar_t spa_evicting_os_cv; /* Objset Eviction Completion */
232 txg_list_t spa_vdev_txg_list; /* per-txg dirty vdev list */
233 vdev_t *spa_root_vdev; /* top-level vdev container */
234 int spa_min_ashift; /* of vdevs in normal class */
235 int spa_max_ashift; /* of vdevs in normal class */
236 uint64_t spa_config_guid; /* config pool guid */
237 uint64_t spa_load_guid; /* spa_load initialized guid */
238 uint64_t spa_last_synced_guid; /* last synced guid */
239 list_t spa_config_dirty_list; /* vdevs with dirty config */
240 list_t spa_state_dirty_list; /* vdevs with dirty state */
241 kmutex_t spa_alloc_lock;
242 avl_tree_t spa_alloc_tree;
243 spa_aux_vdev_t spa_spares; /* hot spares */
244 spa_aux_vdev_t spa_l2cache; /* L2ARC cache devices */
245 nvlist_t *spa_label_features; /* Features for reading MOS */
246 uint64_t spa_config_object; /* MOS object for pool config */
247 uint64_t spa_config_generation; /* config generation number */
248 uint64_t spa_syncing_txg; /* txg currently syncing */
249 bpobj_t spa_deferred_bpobj; /* deferred-free bplist */
250 bplist_t spa_free_bplist[TXG_SIZE]; /* bplist of stuff to free */
251 zio_cksum_salt_t spa_cksum_salt; /* secret salt for cksum */
252 /* checksum context templates */
253 kmutex_t spa_cksum_tmpls_lock;
254 void *spa_cksum_tmpls[ZIO_CHECKSUM_FUNCTIONS];
255 uberblock_t spa_ubsync; /* last synced uberblock */
256 uberblock_t spa_uberblock; /* current uberblock */
257 boolean_t spa_extreme_rewind; /* rewind past deferred frees */
258 uint64_t spa_last_io; /* lbolt of last non-scan I/O */
259 kmutex_t spa_scrub_lock; /* resilver/scrub lock */
260 uint64_t spa_scrub_inflight; /* in-flight scrub bytes */
261 uint64_t spa_load_verify_ios; /* in-flight verifications IOs */
262 kcondvar_t spa_scrub_io_cv; /* scrub I/O completion */
263 uint8_t spa_scrub_active; /* active or suspended? */
264 uint8_t spa_scrub_type; /* type of scrub we're doing */
265 uint8_t spa_scrub_finished; /* indicator to rotate logs */
266 uint8_t spa_scrub_started; /* started since last boot */
267 uint8_t spa_scrub_reopen; /* scrub doing vdev_reopen */
268 uint64_t spa_scan_pass_start; /* start time per pass/reboot */
269 uint64_t spa_scan_pass_scrub_pause; /* scrub pause time */
270 uint64_t spa_scan_pass_scrub_spent_paused; /* total paused */
271 uint64_t spa_scan_pass_exam; /* examined bytes per pass */
272 uint64_t spa_scan_pass_issued; /* issued bytes per pass */
273 kmutex_t spa_async_lock; /* protect async state */
274 kthread_t *spa_async_thread; /* thread doing async task */
275 kthread_t *spa_async_thread_vd; /* thread doing vd async task */
276 int spa_async_suspended; /* async tasks suspended */
277 kcondvar_t spa_async_cv; /* wait for thread_exit() */
278 uint16_t spa_async_tasks; /* async task mask */
279 uint64_t spa_missing_tvds; /* unopenable tvds on load */
280 uint64_t spa_missing_tvds_allowed; /* allow loading spa? */
282 spa_removing_phys_t spa_removing_phys;
283 spa_vdev_removal_t *spa_vdev_removal;
285 spa_condensing_indirect_phys_t spa_condensing_indirect_phys;
286 spa_condensing_indirect_t *spa_condensing_indirect;
287 zthr_t *spa_condense_zthr; /* zthr doing condense. */
289 uint64_t spa_checkpoint_txg; /* the txg of the checkpoint */
290 spa_checkpoint_info_t spa_checkpoint_info; /* checkpoint accounting */
291 zthr_t *spa_checkpoint_discard_zthr;
293 char *spa_root; /* alternate root directory */
294 uint64_t spa_ena; /* spa-wide ereport ENA */
295 int spa_last_open_failed; /* error if last open failed */
296 uint64_t spa_last_ubsync_txg; /* "best" uberblock txg */
297 uint64_t spa_last_ubsync_txg_ts; /* timestamp from that ub */
298 uint64_t spa_load_txg; /* ub txg that loaded */
299 uint64_t spa_load_txg_ts; /* timestamp from that ub */
300 uint64_t spa_load_meta_errors; /* verify metadata err count */
301 uint64_t spa_load_data_errors; /* verify data err count */
302 uint64_t spa_verify_min_txg; /* start txg of verify scrub */
303 kmutex_t spa_errlog_lock; /* error log lock */
304 uint64_t spa_errlog_last; /* last error log object */
305 uint64_t spa_errlog_scrub; /* scrub error log object */
306 kmutex_t spa_errlist_lock; /* error list/ereport lock */
307 avl_tree_t spa_errlist_last; /* last error list */
308 avl_tree_t spa_errlist_scrub; /* scrub error list */
309 uint64_t spa_deflate; /* should we deflate? */
310 uint64_t spa_history; /* history object */
311 kmutex_t spa_history_lock; /* history lock */
312 vdev_t *spa_pending_vdev; /* pending vdev additions */
313 kmutex_t spa_props_lock; /* property lock */
314 uint64_t spa_pool_props_object; /* object for properties */
315 uint64_t spa_bootfs; /* default boot filesystem */
316 uint64_t spa_failmode; /* failure mode for the pool */
317 uint64_t spa_delegation; /* delegation on/off */
318 list_t spa_config_list; /* previous cache file(s) */
319 /* per-CPU array of root of async I/O: */
320 zio_t **spa_async_zio_root;
321 zio_t *spa_suspend_zio_root; /* root of all suspended I/O */
322 zio_t *spa_txg_zio[TXG_SIZE]; /* spa_sync() waits for this */
323 kmutex_t spa_suspend_lock; /* protects suspend_zio_root */
324 kcondvar_t spa_suspend_cv; /* notification of resume */
325 uint8_t spa_suspended; /* pool is suspended */
326 uint8_t spa_claiming; /* pool is doing zil_claim() */
327 boolean_t spa_debug; /* debug enabled? */
328 boolean_t spa_is_root; /* pool is root */
329 int spa_minref; /* num refs when first opened */
330 int spa_mode; /* FREAD | FWRITE */
331 spa_log_state_t spa_log_state; /* log state */
332 uint64_t spa_autoexpand; /* lun expansion on/off */
333 uint64_t spa_bootsize; /* efi system partition size */
334 ddt_t *spa_ddt[ZIO_CHECKSUM_FUNCTIONS]; /* in-core DDTs */
335 uint64_t spa_ddt_stat_object; /* DDT statistics */
336 uint64_t spa_dedup_ditto; /* dedup ditto threshold */
337 uint64_t spa_dedup_checksum; /* default dedup checksum */
338 uint64_t spa_dspace; /* dspace in normal class */
339 kmutex_t spa_vdev_top_lock; /* dueling offline/remove */
340 kmutex_t spa_proc_lock; /* protects spa_proc* */
341 kcondvar_t spa_proc_cv; /* spa_proc_state transitions */
342 spa_proc_state_t spa_proc_state; /* see definition */
343 struct proc *spa_proc; /* "zpool-poolname" process */
344 uint64_t spa_did; /* if procp != p0, did of t1 */
345 kthread_t *spa_trim_thread; /* thread sending TRIM I/Os */
346 kmutex_t spa_trim_lock; /* protects spa_trim_cv */
347 kcondvar_t spa_trim_cv; /* used to notify TRIM thread */
348 boolean_t spa_autoreplace; /* autoreplace set in open */
349 int spa_vdev_locks; /* locks grabbed */
350 uint64_t spa_creation_version; /* version at pool creation */
351 uint64_t spa_prev_software_version; /* See ub_software_version */
352 uint64_t spa_feat_for_write_obj; /* required to write to pool */
353 uint64_t spa_feat_for_read_obj; /* required to read from pool */
354 uint64_t spa_feat_desc_obj; /* Feature descriptions */
355 uint64_t spa_feat_enabled_txg_obj; /* Feature enabled txg */
356 /* cache feature refcounts */
357 uint64_t spa_feat_refcount_cache[SPA_FEATURES];
359 cyclic_id_t spa_deadman_cycid; /* cyclic id */
362 struct callout spa_deadman_cycid; /* callout id */
363 struct task spa_deadman_task;
366 uint64_t spa_deadman_calls; /* number of deadman calls */
367 hrtime_t spa_sync_starttime; /* starting time fo spa_sync */
368 uint64_t spa_deadman_synctime; /* deadman expiration timer */
369 uint64_t spa_all_vdev_zaps; /* ZAP of per-vd ZAP obj #s */
370 spa_avz_action_t spa_avz_action; /* destroy/rebuild AVZ? */
374 * spa_iokstat_lock protects spa_iokstat and
377 kmutex_t spa_iokstat_lock;
378 struct kstat *spa_iokstat; /* kstat of io to this pool */
382 } spa_queue_stats[ZIO_PRIORITY_NUM_QUEUEABLE];
384 hrtime_t spa_ccw_fail_time; /* Conf cache write fail time */
387 * spa_refcount & spa_config_lock must be the last elements
388 * because refcount_t changes size based on compilation options.
389 * In order for the MDB module to function correctly, the other
390 * fields must remain in the same location.
392 spa_config_lock_t spa_config_lock[SCL_LOCKS]; /* config changes */
393 refcount_t spa_refcount; /* number of opens */
395 boolean_t spa_splitting_newspa; /* creating new spa in split */
399 extern const char *spa_config_path;
401 extern void spa_taskq_dispatch_ent(spa_t *spa, zio_type_t t, zio_taskq_type_t q,
402 task_func_t *func, void *arg, uint_t flags, taskq_ent_t *ent);
403 extern void spa_load_spares(spa_t *spa);
404 extern void spa_load_l2cache(spa_t *spa);
410 #endif /* _SYS_SPA_IMPL_H */