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, 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. 29 */ 30 31#ifndef _SYS_SPA_IMPL_H 32#define _SYS_SPA_IMPL_H 33 34#include <sys/spa.h> 35#include <sys/spa_checkpoint.h> 36#include <sys/vdev.h> 37#include <sys/vdev_removal.h> 38#include <sys/metaslab.h> 39#include <sys/dmu.h> 40#include <sys/dsl_pool.h> 41#include <sys/uberblock_impl.h> 42#include <sys/zfs_context.h> 43#include <sys/avl.h> 44#include <sys/refcount.h> 45#include <sys/bplist.h> 46#include <sys/bpobj.h> 47#include <sys/zfeature.h> 48#include <sys/zthr.h> 49#include <zfeature_common.h> 50 51#ifdef __cplusplus 52extern "C" { 53#endif 54 55typedef struct spa_error_entry { 56 zbookmark_phys_t se_bookmark; 57 char *se_name; 58 avl_node_t se_avl; 59} spa_error_entry_t; 60 61typedef 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 */ 67} spa_history_phys_t; 68 69/* 70 * All members must be uint64_t, for byteswap purposes. 71 */ 72typedef struct spa_removing_phys { 73 uint64_t sr_state; /* dsl_scan_state_t */ 74 75 /* 76 * The vdev ID that we most recently attempted to remove, 77 * or -1 if no removal has been attempted. 78 */ 79 uint64_t sr_removing_vdev; 80 81 /* 82 * The vdev ID that we most recently successfully removed, 83 * or -1 if no devices have been removed. 84 */ 85 uint64_t sr_prev_indirect_vdev; 86 87 uint64_t sr_start_time; 88 uint64_t sr_end_time; 89 90 /* 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. 96 */ 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; 100 101/* 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. 105 */ 106typedef struct spa_condensing_indirect_phys { 107 /* 108 * The vdev ID of the indirect vdev whose indirect mapping is 109 * being condensed. 110 */ 111 uint64_t scip_vdev; 112 113 /* 114 * The vdev's old obsolete spacemap. This spacemap's contents are 115 * being integrated into the new mapping. 116 */ 117 uint64_t scip_prev_obsolete_sm_object; 118 119 /* 120 * The new mapping object that is being created. 121 */ 122 uint64_t scip_next_mapping_object; 123} spa_condensing_indirect_phys_t; 124 125struct 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 */ 133}; 134 135typedef struct spa_config_lock { 136 kmutex_t scl_lock; 137 kthread_t *scl_writer; 138 int scl_write_wanted; 139 kcondvar_t scl_cv; 140 refcount_t scl_count; 141} spa_config_lock_t; 142 143typedef struct spa_config_dirent { 144 list_node_t scd_link; 145 char *scd_path; 146} spa_config_dirent_t; 147 148typedef enum zio_taskq_type { 149 ZIO_TASKQ_ISSUE = 0, 150 ZIO_TASKQ_ISSUE_HIGH, 151 ZIO_TASKQ_INTERRUPT, 152 ZIO_TASKQ_INTERRUPT_HIGH, 153 ZIO_TASKQ_TYPES 154} zio_taskq_type_t; 155 156/* 157 * State machine for the zpool-poolname process. The states transitions 158 * are done as follows: 159 * 160 * From To Routine 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() 166 */ 167typedef 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 */ 173} spa_proc_state_t; 174 175typedef struct spa_taskqs { 176 uint_t stqs_count; 177 taskq_t **stqs_taskq; 178} spa_taskqs_t; 179 180typedef enum spa_all_vdev_zap_action { 181 AVZ_ACTION_NONE = 0, 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 185} spa_avz_action_t; 186 187typedef 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; 195 196struct spa { 197 /* 198 * Fields protected by spa_namespace_lock. 199 */ 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 /* 242 * spa_alloc_locks and spa_alloc_trees are arrays, whose lengths are 243 * stored in spa_alloc_count. There is one tree and one lock for each 244 * allocator, to help improve allocation performance in write-heavy 245 * workloads. 246 */ 247 kmutex_t *spa_alloc_locks; 248 avl_tree_t *spa_alloc_trees; 249 int spa_alloc_count; 250 251 spa_aux_vdev_t spa_spares; /* hot spares */ 252 spa_aux_vdev_t spa_l2cache; /* L2ARC cache devices */ 253 nvlist_t *spa_label_features; /* Features for reading MOS */ 254 uint64_t spa_config_object; /* MOS object for pool config */ 255 uint64_t spa_config_generation; /* config generation number */ 256 uint64_t spa_syncing_txg; /* txg currently syncing */ 257 bpobj_t spa_deferred_bpobj; /* deferred-free bplist */ 258 bplist_t spa_free_bplist[TXG_SIZE]; /* bplist of stuff to free */ 259 zio_cksum_salt_t spa_cksum_salt; /* secret salt for cksum */ 260 /* checksum context templates */ 261 kmutex_t spa_cksum_tmpls_lock; 262 void *spa_cksum_tmpls[ZIO_CHECKSUM_FUNCTIONS]; 263 uberblock_t spa_ubsync; /* last synced uberblock */ 264 uberblock_t spa_uberblock; /* current uberblock */ 265 boolean_t spa_extreme_rewind; /* rewind past deferred frees */ 266 uint64_t spa_last_io; /* lbolt of last non-scan I/O */ 267 kmutex_t spa_scrub_lock; /* resilver/scrub lock */ 268 uint64_t spa_scrub_inflight; /* in-flight scrub bytes */ 269 uint64_t spa_load_verify_ios; /* in-flight verifications IOs */ 270 kcondvar_t spa_scrub_io_cv; /* scrub I/O completion */ 271 uint8_t spa_scrub_active; /* active or suspended? */ 272 uint8_t spa_scrub_type; /* type of scrub we're doing */ 273 uint8_t spa_scrub_finished; /* indicator to rotate logs */ 274 uint8_t spa_scrub_started; /* started since last boot */ 275 uint8_t spa_scrub_reopen; /* scrub doing vdev_reopen */ 276 uint64_t spa_scan_pass_start; /* start time per pass/reboot */ 277 uint64_t spa_scan_pass_scrub_pause; /* scrub pause time */ 278 uint64_t spa_scan_pass_scrub_spent_paused; /* total paused */ 279 uint64_t spa_scan_pass_exam; /* examined bytes per pass */ 280 uint64_t spa_scan_pass_issued; /* issued bytes per pass */ 281 kmutex_t spa_async_lock; /* protect async state */ 282 kthread_t *spa_async_thread; /* thread doing async task */ 283 kthread_t *spa_async_thread_vd; /* thread doing vd async task */ 284 int spa_async_suspended; /* async tasks suspended */ 285 kcondvar_t spa_async_cv; /* wait for thread_exit() */ 286 uint16_t spa_async_tasks; /* async task mask */ 287 uint64_t spa_missing_tvds; /* unopenable tvds on load */ 288 uint64_t spa_missing_tvds_allowed; /* allow loading spa? */ 289 290 spa_removing_phys_t spa_removing_phys; 291 spa_vdev_removal_t *spa_vdev_removal; 292 293 spa_condensing_indirect_phys_t spa_condensing_indirect_phys; 294 spa_condensing_indirect_t *spa_condensing_indirect; 295 zthr_t *spa_condense_zthr; /* zthr doing condense. */ 296 297 uint64_t spa_checkpoint_txg; /* the txg of the checkpoint */ 298 spa_checkpoint_info_t spa_checkpoint_info; /* checkpoint accounting */ 299 zthr_t *spa_checkpoint_discard_zthr; 300 301 char *spa_root; /* alternate root directory */ 302 uint64_t spa_ena; /* spa-wide ereport ENA */ 303 int spa_last_open_failed; /* error if last open failed */ 304 uint64_t spa_last_ubsync_txg; /* "best" uberblock txg */ 305 uint64_t spa_last_ubsync_txg_ts; /* timestamp from that ub */ 306 uint64_t spa_load_txg; /* ub txg that loaded */ 307 uint64_t spa_load_txg_ts; /* timestamp from that ub */ 308 uint64_t spa_load_meta_errors; /* verify metadata err count */ 309 uint64_t spa_load_data_errors; /* verify data err count */ 310 uint64_t spa_verify_min_txg; /* start txg of verify scrub */ 311 kmutex_t spa_errlog_lock; /* error log lock */ 312 uint64_t spa_errlog_last; /* last error log object */ 313 uint64_t spa_errlog_scrub; /* scrub error log object */ 314 kmutex_t spa_errlist_lock; /* error list/ereport lock */ 315 avl_tree_t spa_errlist_last; /* last error list */ 316 avl_tree_t spa_errlist_scrub; /* scrub error list */ 317 uint64_t spa_deflate; /* should we deflate? */ 318 uint64_t spa_history; /* history object */ 319 kmutex_t spa_history_lock; /* history lock */ 320 vdev_t *spa_pending_vdev; /* pending vdev additions */ 321 kmutex_t spa_props_lock; /* property lock */ 322 uint64_t spa_pool_props_object; /* object for properties */ 323 uint64_t spa_bootfs; /* default boot filesystem */ 324 uint64_t spa_failmode; /* failure mode for the pool */ 325 uint64_t spa_delegation; /* delegation on/off */ 326 list_t spa_config_list; /* previous cache file(s) */ 327 /* per-CPU array of root of async I/O: */ 328 zio_t **spa_async_zio_root; 329 zio_t *spa_suspend_zio_root; /* root of all suspended I/O */ 330 zio_t *spa_txg_zio[TXG_SIZE]; /* spa_sync() waits for this */ 331 kmutex_t spa_suspend_lock; /* protects suspend_zio_root */ 332 kcondvar_t spa_suspend_cv; /* notification of resume */ 333 uint8_t spa_suspended; /* pool is suspended */ 334 uint8_t spa_claiming; /* pool is doing zil_claim() */ 335 boolean_t spa_is_root; /* pool is root */ 336 int spa_minref; /* num refs when first opened */ 337 int spa_mode; /* FREAD | FWRITE */ 338 spa_log_state_t spa_log_state; /* log state */ 339 uint64_t spa_autoexpand; /* lun expansion on/off */ 340 uint64_t spa_bootsize; /* efi system partition size */ 341 ddt_t *spa_ddt[ZIO_CHECKSUM_FUNCTIONS]; /* in-core DDTs */ 342 uint64_t spa_ddt_stat_object; /* DDT statistics */ 343 uint64_t spa_dedup_ditto; /* dedup ditto threshold */ 344 uint64_t spa_dedup_checksum; /* default dedup checksum */ 345 uint64_t spa_dspace; /* dspace in normal class */ 346 kmutex_t spa_vdev_top_lock; /* dueling offline/remove */ 347 kmutex_t spa_proc_lock; /* protects spa_proc* */ 348 kcondvar_t spa_proc_cv; /* spa_proc_state transitions */ 349 spa_proc_state_t spa_proc_state; /* see definition */ 350 struct proc *spa_proc; /* "zpool-poolname" process */ 351 uint64_t spa_did; /* if procp != p0, did of t1 */ 352 kthread_t *spa_trim_thread; /* thread sending TRIM I/Os */ 353 kmutex_t spa_trim_lock; /* protects spa_trim_cv */ 354 kcondvar_t spa_trim_cv; /* used to notify TRIM thread */ 355 boolean_t spa_autoreplace; /* autoreplace set in open */ 356 int spa_vdev_locks; /* locks grabbed */ 357 uint64_t spa_creation_version; /* version at pool creation */ 358 uint64_t spa_prev_software_version; /* See ub_software_version */ 359 uint64_t spa_feat_for_write_obj; /* required to write to pool */ 360 uint64_t spa_feat_for_read_obj; /* required to read from pool */ 361 uint64_t spa_feat_desc_obj; /* Feature descriptions */ 362 uint64_t spa_feat_enabled_txg_obj; /* Feature enabled txg */ 363 kmutex_t spa_feat_stats_lock; /* protects spa_feat_stats */ 364 nvlist_t *spa_feat_stats; /* Cache of enabled features */ 365 /* cache feature refcounts */ 366 uint64_t spa_feat_refcount_cache[SPA_FEATURES]; 367#ifdef illumos 368 cyclic_id_t spa_deadman_cycid; /* cyclic id */ 369#else /* !illumos */ 370#ifdef _KERNEL 371 struct callout spa_deadman_cycid; /* callout id */ 372 struct task spa_deadman_task; 373#endif 374#endif /* illumos */ 375 uint64_t spa_deadman_calls; /* number of deadman calls */ 376 hrtime_t spa_sync_starttime; /* starting time fo spa_sync */ 377 uint64_t spa_deadman_synctime; /* deadman expiration timer */ 378 uint64_t spa_all_vdev_zaps; /* ZAP of per-vd ZAP obj #s */ 379 spa_avz_action_t spa_avz_action; /* destroy/rebuild AVZ? */ 380 381#ifdef illumos 382 /* 383 * spa_iokstat_lock protects spa_iokstat and 384 * spa_queue_stats[]. 385 */ 386 kmutex_t spa_iokstat_lock; 387 struct kstat *spa_iokstat; /* kstat of io to this pool */ 388 struct { 389 int spa_active; 390 int spa_queued; 391 } spa_queue_stats[ZIO_PRIORITY_NUM_QUEUEABLE]; 392#endif 393 /* arc_memory_throttle() parameters during low memory condition */ 394 uint64_t spa_lowmem_page_load; /* memory load during txg */ 395 uint64_t spa_lowmem_last_txg; /* txg window start */ 396 397 hrtime_t spa_ccw_fail_time; /* Conf cache write fail time */ 398 399 /* 400 * spa_refcount & spa_config_lock must be the last elements 401 * because refcount_t changes size based on compilation options. 402 * In order for the MDB module to function correctly, the other 403 * fields must remain in the same location. 404 */ 405 spa_config_lock_t spa_config_lock[SCL_LOCKS]; /* config changes */ 406 refcount_t spa_refcount; /* number of opens */ 407#ifndef illumos 408 boolean_t spa_splitting_newspa; /* creating new spa in split */ 409#endif 410}; 411 412extern const char *spa_config_path; 413 414extern void spa_taskq_dispatch_ent(spa_t *spa, zio_type_t t, zio_taskq_type_t q, 415 task_func_t *func, void *arg, uint_t flags, taskq_ent_t *ent); 416extern void spa_load_spares(spa_t *spa); 417extern void spa_load_l2cache(spa_t *spa); 418 419#ifdef __cplusplus 420} 421#endif 422 423#endif /* _SYS_SPA_IMPL_H */ 424