]> git.proxmox.com Git - mirror_zfs.git/blob - include/sys/spa_impl.h
Refactor deadman set failmode to be cross platform
[mirror_zfs.git] / include / sys / spa_impl.h
1 /*
2 * CDDL HEADER START
3 *
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.
7 *
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.
12 *
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]
18 *
19 * CDDL HEADER END
20 */
21 /*
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 Datto Inc.
29 * Copyright (c) 2017, Intel Corporation.
30 */
31
32 #ifndef _SYS_SPA_IMPL_H
33 #define _SYS_SPA_IMPL_H
34
35 #include <sys/spa.h>
36 #include <sys/spa_checkpoint.h>
37 #include <sys/spa_log_spacemap.h>
38 #include <sys/vdev.h>
39 #include <sys/vdev_removal.h>
40 #include <sys/metaslab.h>
41 #include <sys/dmu.h>
42 #include <sys/dsl_pool.h>
43 #include <sys/uberblock_impl.h>
44 #include <sys/zfs_context.h>
45 #include <sys/avl.h>
46 #include <sys/refcount.h>
47 #include <sys/bplist.h>
48 #include <sys/bpobj.h>
49 #include <sys/dsl_crypt.h>
50 #include <sys/zfeature.h>
51 #include <sys/zthr.h>
52 #include <sys/dsl_deadlist.h>
53 #include <zfeature_common.h>
54
55 #ifdef __cplusplus
56 extern "C" {
57 #endif
58
59 typedef struct spa_error_entry {
60 zbookmark_phys_t se_bookmark;
61 char *se_name;
62 avl_node_t se_avl;
63 } spa_error_entry_t;
64
65 typedef struct spa_history_phys {
66 uint64_t sh_pool_create_len; /* ending offset of zpool create */
67 uint64_t sh_phys_max_off; /* physical EOF */
68 uint64_t sh_bof; /* logical BOF */
69 uint64_t sh_eof; /* logical EOF */
70 uint64_t sh_records_lost; /* num of records overwritten */
71 } spa_history_phys_t;
72
73 /*
74 * All members must be uint64_t, for byteswap purposes.
75 */
76 typedef struct spa_removing_phys {
77 uint64_t sr_state; /* dsl_scan_state_t */
78
79 /*
80 * The vdev ID that we most recently attempted to remove,
81 * or -1 if no removal has been attempted.
82 */
83 uint64_t sr_removing_vdev;
84
85 /*
86 * The vdev ID that we most recently successfully removed,
87 * or -1 if no devices have been removed.
88 */
89 uint64_t sr_prev_indirect_vdev;
90
91 uint64_t sr_start_time;
92 uint64_t sr_end_time;
93
94 /*
95 * Note that we can not use the space map's or indirect mapping's
96 * accounting as a substitute for these values, because we need to
97 * count frees of not-yet-copied data as though it did the copy.
98 * Otherwise, we could get into a situation where copied > to_copy,
99 * or we complete before copied == to_copy.
100 */
101 uint64_t sr_to_copy; /* bytes that need to be copied */
102 uint64_t sr_copied; /* bytes that have been copied or freed */
103 } spa_removing_phys_t;
104
105 /*
106 * This struct is stored as an entry in the DMU_POOL_DIRECTORY_OBJECT
107 * (with key DMU_POOL_CONDENSING_INDIRECT). It is present if a condense
108 * of an indirect vdev's mapping object is in progress.
109 */
110 typedef struct spa_condensing_indirect_phys {
111 /*
112 * The vdev ID of the indirect vdev whose indirect mapping is
113 * being condensed.
114 */
115 uint64_t scip_vdev;
116
117 /*
118 * The vdev's old obsolete spacemap. This spacemap's contents are
119 * being integrated into the new mapping.
120 */
121 uint64_t scip_prev_obsolete_sm_object;
122
123 /*
124 * The new mapping object that is being created.
125 */
126 uint64_t scip_next_mapping_object;
127 } spa_condensing_indirect_phys_t;
128
129 struct spa_aux_vdev {
130 uint64_t sav_object; /* MOS object for device list */
131 nvlist_t *sav_config; /* cached device config */
132 vdev_t **sav_vdevs; /* devices */
133 int sav_count; /* number devices */
134 boolean_t sav_sync; /* sync the device list */
135 nvlist_t **sav_pending; /* pending device additions */
136 uint_t sav_npending; /* # pending devices */
137 };
138
139 typedef struct spa_config_lock {
140 kmutex_t scl_lock;
141 kthread_t *scl_writer;
142 int scl_write_wanted;
143 kcondvar_t scl_cv;
144 zfs_refcount_t scl_count;
145 } spa_config_lock_t;
146
147 typedef struct spa_config_dirent {
148 list_node_t scd_link;
149 char *scd_path;
150 } spa_config_dirent_t;
151
152 typedef enum zio_taskq_type {
153 ZIO_TASKQ_ISSUE = 0,
154 ZIO_TASKQ_ISSUE_HIGH,
155 ZIO_TASKQ_INTERRUPT,
156 ZIO_TASKQ_INTERRUPT_HIGH,
157 ZIO_TASKQ_TYPES
158 } zio_taskq_type_t;
159
160 /*
161 * State machine for the zpool-poolname process. The states transitions
162 * are done as follows:
163 *
164 * From To Routine
165 * PROC_NONE -> PROC_CREATED spa_activate()
166 * PROC_CREATED -> PROC_ACTIVE spa_thread()
167 * PROC_ACTIVE -> PROC_DEACTIVATE spa_deactivate()
168 * PROC_DEACTIVATE -> PROC_GONE spa_thread()
169 * PROC_GONE -> PROC_NONE spa_deactivate()
170 */
171 typedef enum spa_proc_state {
172 SPA_PROC_NONE, /* spa_proc = &p0, no process created */
173 SPA_PROC_CREATED, /* spa_activate() has proc, is waiting */
174 SPA_PROC_ACTIVE, /* taskqs created, spa_proc set */
175 SPA_PROC_DEACTIVATE, /* spa_deactivate() requests process exit */
176 SPA_PROC_GONE /* spa_thread() is exiting, spa_proc = &p0 */
177 } spa_proc_state_t;
178
179 typedef struct spa_taskqs {
180 uint_t stqs_count;
181 taskq_t **stqs_taskq;
182 } spa_taskqs_t;
183
184 typedef enum spa_all_vdev_zap_action {
185 AVZ_ACTION_NONE = 0,
186 AVZ_ACTION_DESTROY, /* Destroy all per-vdev ZAPs and the AVZ. */
187 AVZ_ACTION_REBUILD, /* Populate the new AVZ, see spa_avz_rebuild */
188 AVZ_ACTION_INITIALIZE
189 } spa_avz_action_t;
190
191 typedef enum spa_config_source {
192 SPA_CONFIG_SRC_NONE = 0,
193 SPA_CONFIG_SRC_SCAN, /* scan of path (default: /dev/dsk) */
194 SPA_CONFIG_SRC_CACHEFILE, /* any cachefile */
195 SPA_CONFIG_SRC_TRYIMPORT, /* returned from call to tryimport */
196 SPA_CONFIG_SRC_SPLIT, /* new pool in a pool split */
197 SPA_CONFIG_SRC_MOS /* MOS, but not always from right txg */
198 } spa_config_source_t;
199
200 struct spa {
201 /*
202 * Fields protected by spa_namespace_lock.
203 */
204 char spa_name[ZFS_MAX_DATASET_NAME_LEN]; /* pool name */
205 char *spa_comment; /* comment */
206 avl_node_t spa_avl; /* node in spa_namespace_avl */
207 nvlist_t *spa_config; /* last synced config */
208 nvlist_t *spa_config_syncing; /* currently syncing config */
209 nvlist_t *spa_config_splitting; /* config for splitting */
210 nvlist_t *spa_load_info; /* info and errors from load */
211 uint64_t spa_config_txg; /* txg of last config change */
212 int spa_sync_pass; /* iterate-to-convergence */
213 pool_state_t spa_state; /* pool state */
214 int spa_inject_ref; /* injection references */
215 uint8_t spa_sync_on; /* sync threads are running */
216 spa_load_state_t spa_load_state; /* current load operation */
217 boolean_t spa_indirect_vdevs_loaded; /* mappings loaded? */
218 boolean_t spa_trust_config; /* do we trust vdev tree? */
219 spa_config_source_t spa_config_source; /* where config comes from? */
220 uint64_t spa_import_flags; /* import specific flags */
221 spa_taskqs_t spa_zio_taskq[ZIO_TYPES][ZIO_TASKQ_TYPES];
222 dsl_pool_t *spa_dsl_pool;
223 boolean_t spa_is_initializing; /* true while opening pool */
224 boolean_t spa_is_exporting; /* true while exporting pool */
225 metaslab_class_t *spa_normal_class; /* normal data class */
226 metaslab_class_t *spa_log_class; /* intent log data class */
227 metaslab_class_t *spa_special_class; /* special allocation class */
228 metaslab_class_t *spa_dedup_class; /* dedup allocation class */
229 uint64_t spa_first_txg; /* first txg after spa_open() */
230 uint64_t spa_final_txg; /* txg of export/destroy */
231 uint64_t spa_freeze_txg; /* freeze pool at this txg */
232 uint64_t spa_load_max_txg; /* best initial ub_txg */
233 uint64_t spa_claim_max_txg; /* highest claimed birth txg */
234 inode_timespec_t spa_loaded_ts; /* 1st successful open time */
235 objset_t *spa_meta_objset; /* copy of dp->dp_meta_objset */
236 kmutex_t spa_evicting_os_lock; /* Evicting objset list lock */
237 list_t spa_evicting_os_list; /* Objsets being evicted. */
238 kcondvar_t spa_evicting_os_cv; /* Objset Eviction Completion */
239 txg_list_t spa_vdev_txg_list; /* per-txg dirty vdev list */
240 vdev_t *spa_root_vdev; /* top-level vdev container */
241 int spa_min_ashift; /* of vdevs in normal class */
242 int spa_max_ashift; /* of vdevs in normal class */
243 uint64_t spa_config_guid; /* config pool guid */
244 uint64_t spa_load_guid; /* spa_load initialized guid */
245 uint64_t spa_last_synced_guid; /* last synced guid */
246 list_t spa_config_dirty_list; /* vdevs with dirty config */
247 list_t spa_state_dirty_list; /* vdevs with dirty state */
248 /*
249 * spa_alloc_locks and spa_alloc_trees are arrays, whose lengths are
250 * stored in spa_alloc_count. There is one tree and one lock for each
251 * allocator, to help improve allocation performance in write-heavy
252 * workloads.
253 */
254 kmutex_t *spa_alloc_locks;
255 avl_tree_t *spa_alloc_trees;
256 int spa_alloc_count;
257
258 spa_aux_vdev_t spa_spares; /* hot spares */
259 spa_aux_vdev_t spa_l2cache; /* L2ARC cache devices */
260 nvlist_t *spa_label_features; /* Features for reading MOS */
261 uint64_t spa_config_object; /* MOS object for pool config */
262 uint64_t spa_config_generation; /* config generation number */
263 uint64_t spa_syncing_txg; /* txg currently syncing */
264 bpobj_t spa_deferred_bpobj; /* deferred-free bplist */
265 bplist_t spa_free_bplist[TXG_SIZE]; /* bplist of stuff to free */
266 zio_cksum_salt_t spa_cksum_salt; /* secret salt for cksum */
267 /* checksum context templates */
268 kmutex_t spa_cksum_tmpls_lock;
269 void *spa_cksum_tmpls[ZIO_CHECKSUM_FUNCTIONS];
270 uberblock_t spa_ubsync; /* last synced uberblock */
271 uberblock_t spa_uberblock; /* current uberblock */
272 boolean_t spa_extreme_rewind; /* rewind past deferred frees */
273 kmutex_t spa_scrub_lock; /* resilver/scrub lock */
274 uint64_t spa_scrub_inflight; /* in-flight scrub bytes */
275
276 /* in-flight verification bytes */
277 uint64_t spa_load_verify_bytes;
278 kcondvar_t spa_scrub_io_cv; /* scrub I/O completion */
279 uint8_t spa_scrub_active; /* active or suspended? */
280 uint8_t spa_scrub_type; /* type of scrub we're doing */
281 uint8_t spa_scrub_finished; /* indicator to rotate logs */
282 uint8_t spa_scrub_started; /* started since last boot */
283 uint8_t spa_scrub_reopen; /* scrub doing vdev_reopen */
284 uint64_t spa_scan_pass_start; /* start time per pass/reboot */
285 uint64_t spa_scan_pass_scrub_pause; /* scrub pause time */
286 uint64_t spa_scan_pass_scrub_spent_paused; /* total paused */
287 uint64_t spa_scan_pass_exam; /* examined bytes per pass */
288 uint64_t spa_scan_pass_issued; /* issued bytes per pass */
289
290 /*
291 * We are in the middle of a resilver, and another resilver
292 * is needed once this one completes. This is set iff any
293 * vdev_resilver_deferred is set.
294 */
295 boolean_t spa_resilver_deferred;
296 kmutex_t spa_async_lock; /* protect async state */
297 kthread_t *spa_async_thread; /* thread doing async task */
298 int spa_async_suspended; /* async tasks suspended */
299 kcondvar_t spa_async_cv; /* wait for thread_exit() */
300 uint16_t spa_async_tasks; /* async task mask */
301 uint64_t spa_missing_tvds; /* unopenable tvds on load */
302 uint64_t spa_missing_tvds_allowed; /* allow loading spa? */
303
304 spa_removing_phys_t spa_removing_phys;
305 spa_vdev_removal_t *spa_vdev_removal;
306
307 spa_condensing_indirect_phys_t spa_condensing_indirect_phys;
308 spa_condensing_indirect_t *spa_condensing_indirect;
309 zthr_t *spa_condense_zthr; /* zthr doing condense. */
310
311 uint64_t spa_checkpoint_txg; /* the txg of the checkpoint */
312 spa_checkpoint_info_t spa_checkpoint_info; /* checkpoint accounting */
313 zthr_t *spa_checkpoint_discard_zthr;
314
315 space_map_t *spa_syncing_log_sm; /* current log space map */
316 avl_tree_t spa_sm_logs_by_txg;
317 kmutex_t spa_flushed_ms_lock; /* for metaslabs_by_flushed */
318 avl_tree_t spa_metaslabs_by_flushed;
319 spa_unflushed_stats_t spa_unflushed_stats;
320 list_t spa_log_summary;
321 uint64_t spa_log_flushall_txg;
322
323 zthr_t *spa_livelist_delete_zthr; /* deleting livelists */
324 zthr_t *spa_livelist_condense_zthr; /* condensing livelists */
325 uint64_t spa_livelists_to_delete; /* set of livelists to free */
326 livelist_condense_entry_t spa_to_condense; /* next to condense */
327
328 char *spa_root; /* alternate root directory */
329 uint64_t spa_ena; /* spa-wide ereport ENA */
330 int spa_last_open_failed; /* error if last open failed */
331 uint64_t spa_last_ubsync_txg; /* "best" uberblock txg */
332 uint64_t spa_last_ubsync_txg_ts; /* timestamp from that ub */
333 uint64_t spa_load_txg; /* ub txg that loaded */
334 uint64_t spa_load_txg_ts; /* timestamp from that ub */
335 uint64_t spa_load_meta_errors; /* verify metadata err count */
336 uint64_t spa_load_data_errors; /* verify data err count */
337 uint64_t spa_verify_min_txg; /* start txg of verify scrub */
338 kmutex_t spa_errlog_lock; /* error log lock */
339 uint64_t spa_errlog_last; /* last error log object */
340 uint64_t spa_errlog_scrub; /* scrub error log object */
341 kmutex_t spa_errlist_lock; /* error list/ereport lock */
342 avl_tree_t spa_errlist_last; /* last error list */
343 avl_tree_t spa_errlist_scrub; /* scrub error list */
344 uint64_t spa_deflate; /* should we deflate? */
345 uint64_t spa_history; /* history object */
346 kmutex_t spa_history_lock; /* history lock */
347 vdev_t *spa_pending_vdev; /* pending vdev additions */
348 kmutex_t spa_props_lock; /* property lock */
349 uint64_t spa_pool_props_object; /* object for properties */
350 uint64_t spa_bootfs; /* default boot filesystem */
351 uint64_t spa_failmode; /* failure mode for the pool */
352 uint64_t spa_deadman_failmode; /* failure mode for deadman */
353 uint64_t spa_delegation; /* delegation on/off */
354 list_t spa_config_list; /* previous cache file(s) */
355 /* per-CPU array of root of async I/O: */
356 zio_t **spa_async_zio_root;
357 zio_t *spa_suspend_zio_root; /* root of all suspended I/O */
358 zio_t *spa_txg_zio[TXG_SIZE]; /* spa_sync() waits for this */
359 kmutex_t spa_suspend_lock; /* protects suspend_zio_root */
360 kcondvar_t spa_suspend_cv; /* notification of resume */
361 zio_suspend_reason_t spa_suspended; /* pool is suspended */
362 uint8_t spa_claiming; /* pool is doing zil_claim() */
363 boolean_t spa_is_root; /* pool is root */
364 int spa_minref; /* num refs when first opened */
365 spa_mode_t spa_mode; /* SPA_MODE_{READ|WRITE} */
366 spa_log_state_t spa_log_state; /* log state */
367 uint64_t spa_autoexpand; /* lun expansion on/off */
368 ddt_t *spa_ddt[ZIO_CHECKSUM_FUNCTIONS]; /* in-core DDTs */
369 uint64_t spa_ddt_stat_object; /* DDT statistics */
370 uint64_t spa_dedup_dspace; /* Cache get_dedup_dspace() */
371 uint64_t spa_dedup_checksum; /* default dedup checksum */
372 uint64_t spa_dspace; /* dspace in normal class */
373 kmutex_t spa_vdev_top_lock; /* dueling offline/remove */
374 kmutex_t spa_proc_lock; /* protects spa_proc* */
375 kcondvar_t spa_proc_cv; /* spa_proc_state transitions */
376 spa_proc_state_t spa_proc_state; /* see definition */
377 proc_t *spa_proc; /* "zpool-poolname" process */
378 uint64_t spa_did; /* if procp != p0, did of t1 */
379 boolean_t spa_autoreplace; /* autoreplace set in open */
380 int spa_vdev_locks; /* locks grabbed */
381 uint64_t spa_creation_version; /* version at pool creation */
382 uint64_t spa_prev_software_version; /* See ub_software_version */
383 uint64_t spa_feat_for_write_obj; /* required to write to pool */
384 uint64_t spa_feat_for_read_obj; /* required to read from pool */
385 uint64_t spa_feat_desc_obj; /* Feature descriptions */
386 uint64_t spa_feat_enabled_txg_obj; /* Feature enabled txg */
387 kmutex_t spa_feat_stats_lock; /* protects spa_feat_stats */
388 nvlist_t *spa_feat_stats; /* Cache of enabled features */
389 /* cache feature refcounts */
390 uint64_t spa_feat_refcount_cache[SPA_FEATURES];
391 taskqid_t spa_deadman_tqid; /* Task id */
392 uint64_t spa_deadman_calls; /* number of deadman calls */
393 hrtime_t spa_sync_starttime; /* starting time of spa_sync */
394 uint64_t spa_deadman_synctime; /* deadman sync expiration */
395 uint64_t spa_deadman_ziotime; /* deadman zio expiration */
396 uint64_t spa_all_vdev_zaps; /* ZAP of per-vd ZAP obj #s */
397 spa_avz_action_t spa_avz_action; /* destroy/rebuild AVZ? */
398 uint64_t spa_autotrim; /* automatic background trim? */
399 uint64_t spa_errata; /* errata issues detected */
400 spa_stats_t spa_stats; /* assorted spa statistics */
401 spa_keystore_t spa_keystore; /* loaded crypto keys */
402
403 /* arc_memory_throttle() parameters during low memory condition */
404 uint64_t spa_lowmem_page_load; /* memory load during txg */
405 uint64_t spa_lowmem_last_txg; /* txg window start */
406
407 hrtime_t spa_ccw_fail_time; /* Conf cache write fail time */
408 taskq_t *spa_zvol_taskq; /* Taskq for minor management */
409 taskq_t *spa_prefetch_taskq; /* Taskq for prefetch threads */
410 uint64_t spa_multihost; /* multihost aware (mmp) */
411 mmp_thread_t spa_mmp; /* multihost mmp thread */
412 list_t spa_leaf_list; /* list of leaf vdevs */
413 uint64_t spa_leaf_list_gen; /* track leaf_list changes */
414 uint32_t spa_hostid; /* cached system hostid */
415
416 /* synchronization for threads in spa_wait */
417 kmutex_t spa_activities_lock;
418 kcondvar_t spa_activities_cv;
419 kcondvar_t spa_waiters_cv;
420 int spa_waiters; /* number of waiting threads */
421 boolean_t spa_waiters_cancel; /* waiters should return */
422
423 /*
424 * spa_refcount & spa_config_lock must be the last elements
425 * because zfs_refcount_t changes size based on compilation options.
426 * In order for the MDB module to function correctly, the other
427 * fields must remain in the same location.
428 */
429 spa_config_lock_t spa_config_lock[SCL_LOCKS]; /* config changes */
430 zfs_refcount_t spa_refcount; /* number of opens */
431
432 taskq_t *spa_upgrade_taskq; /* taskq for upgrade jobs */
433 };
434
435 extern char *spa_config_path;
436 extern char *zfs_deadman_failmode;
437 extern int spa_slop_shift;
438 extern void spa_taskq_dispatch_ent(spa_t *spa, zio_type_t t, zio_taskq_type_t q,
439 task_func_t *func, void *arg, uint_t flags, taskq_ent_t *ent);
440 extern void spa_taskq_dispatch_sync(spa_t *, zio_type_t t, zio_taskq_type_t q,
441 task_func_t *func, void *arg, uint_t flags);
442 extern void spa_load_spares(spa_t *spa);
443 extern void spa_load_l2cache(spa_t *spa);
444 extern sysevent_t *spa_event_create(spa_t *spa, vdev_t *vd, nvlist_t *hist_nvl,
445 const char *name);
446 extern void spa_event_post(sysevent_t *ev);
447 extern int param_set_deadman_failmode_common(const char *val);
448
449 #ifdef __cplusplus
450 }
451 #endif
452
453 #endif /* _SYS_SPA_IMPL_H */