spa_impl.h revision 290757
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, 2015 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 */ 29 30#ifndef _SYS_SPA_IMPL_H 31#define _SYS_SPA_IMPL_H 32 33#include <sys/spa.h> 34#include <sys/vdev.h> 35#include <sys/metaslab.h> 36#include <sys/dmu.h> 37#include <sys/dsl_pool.h> 38#include <sys/uberblock_impl.h> 39#include <sys/zfs_context.h> 40#include <sys/avl.h> 41#include <sys/refcount.h> 42#include <sys/bplist.h> 43#include <sys/bpobj.h> 44#include <sys/zfeature.h> 45#include <zfeature_common.h> 46 47#ifdef __cplusplus 48extern "C" { 49#endif 50 51typedef struct spa_error_entry { 52 zbookmark_phys_t se_bookmark; 53 char *se_name; 54 avl_node_t se_avl; 55} spa_error_entry_t; 56 57typedef struct spa_history_phys { 58 uint64_t sh_pool_create_len; /* ending offset of zpool create */ 59 uint64_t sh_phys_max_off; /* physical EOF */ 60 uint64_t sh_bof; /* logical BOF */ 61 uint64_t sh_eof; /* logical EOF */ 62 uint64_t sh_records_lost; /* num of records overwritten */ 63} spa_history_phys_t; 64 65struct spa_aux_vdev { 66 uint64_t sav_object; /* MOS object for device list */ 67 nvlist_t *sav_config; /* cached device config */ 68 vdev_t **sav_vdevs; /* devices */ 69 int sav_count; /* number devices */ 70 boolean_t sav_sync; /* sync the device list */ 71 nvlist_t **sav_pending; /* pending device additions */ 72 uint_t sav_npending; /* # pending devices */ 73}; 74 75typedef struct spa_config_lock { 76 kmutex_t scl_lock; 77 kthread_t *scl_writer; 78 int scl_write_wanted; 79 kcondvar_t scl_cv; 80 refcount_t scl_count; 81} spa_config_lock_t; 82 83typedef struct spa_config_dirent { 84 list_node_t scd_link; 85 char *scd_path; 86} spa_config_dirent_t; 87 88typedef enum zio_taskq_type { 89 ZIO_TASKQ_ISSUE = 0, 90 ZIO_TASKQ_ISSUE_HIGH, 91 ZIO_TASKQ_INTERRUPT, 92 ZIO_TASKQ_INTERRUPT_HIGH, 93 ZIO_TASKQ_TYPES 94} zio_taskq_type_t; 95 96/* 97 * State machine for the zpool-poolname process. The states transitions 98 * are done as follows: 99 * 100 * From To Routine 101 * PROC_NONE -> PROC_CREATED spa_activate() 102 * PROC_CREATED -> PROC_ACTIVE spa_thread() 103 * PROC_ACTIVE -> PROC_DEACTIVATE spa_deactivate() 104 * PROC_DEACTIVATE -> PROC_GONE spa_thread() 105 * PROC_GONE -> PROC_NONE spa_deactivate() 106 */ 107typedef enum spa_proc_state { 108 SPA_PROC_NONE, /* spa_proc = &p0, no process created */ 109 SPA_PROC_CREATED, /* spa_activate() has proc, is waiting */ 110 SPA_PROC_ACTIVE, /* taskqs created, spa_proc set */ 111 SPA_PROC_DEACTIVATE, /* spa_deactivate() requests process exit */ 112 SPA_PROC_GONE /* spa_thread() is exiting, spa_proc = &p0 */ 113} spa_proc_state_t; 114 115typedef struct spa_taskqs { 116 uint_t stqs_count; 117 taskq_t **stqs_taskq; 118} spa_taskqs_t; 119 120struct spa { 121 /* 122 * Fields protected by spa_namespace_lock. 123 */ 124 char spa_name[MAXNAMELEN]; /* pool name */ 125 char *spa_comment; /* comment */ 126 avl_node_t spa_avl; /* node in spa_namespace_avl */ 127 nvlist_t *spa_config; /* last synced config */ 128 nvlist_t *spa_config_syncing; /* currently syncing config */ 129 nvlist_t *spa_config_splitting; /* config for splitting */ 130 nvlist_t *spa_load_info; /* info and errors from load */ 131 uint64_t spa_config_txg; /* txg of last config change */ 132 int spa_sync_pass; /* iterate-to-convergence */ 133 pool_state_t spa_state; /* pool state */ 134 int spa_inject_ref; /* injection references */ 135 uint8_t spa_sync_on; /* sync threads are running */ 136 spa_load_state_t spa_load_state; /* current load operation */ 137 uint64_t spa_import_flags; /* import specific flags */ 138 spa_taskqs_t spa_zio_taskq[ZIO_TYPES][ZIO_TASKQ_TYPES]; 139 dsl_pool_t *spa_dsl_pool; 140 boolean_t spa_is_initializing; /* true while opening pool */ 141 metaslab_class_t *spa_normal_class; /* normal data class */ 142 metaslab_class_t *spa_log_class; /* intent log data class */ 143 uint64_t spa_first_txg; /* first txg after spa_open() */ 144 uint64_t spa_final_txg; /* txg of export/destroy */ 145 uint64_t spa_freeze_txg; /* freeze pool at this txg */ 146 uint64_t spa_load_max_txg; /* best initial ub_txg */ 147 uint64_t spa_claim_max_txg; /* highest claimed birth txg */ 148 timespec_t spa_loaded_ts; /* 1st successful open time */ 149 objset_t *spa_meta_objset; /* copy of dp->dp_meta_objset */ 150 kmutex_t spa_evicting_os_lock; /* Evicting objset list lock */ 151 list_t spa_evicting_os_list; /* Objsets being evicted. */ 152 kcondvar_t spa_evicting_os_cv; /* Objset Eviction Completion */ 153 txg_list_t spa_vdev_txg_list; /* per-txg dirty vdev list */ 154 vdev_t *spa_root_vdev; /* top-level vdev container */ 155 int spa_min_ashift; /* of vdevs in normal class */ 156 int spa_max_ashift; /* of vdevs in normal class */ 157 uint64_t spa_config_guid; /* config pool guid */ 158 uint64_t spa_load_guid; /* spa_load initialized guid */ 159 uint64_t spa_last_synced_guid; /* last synced guid */ 160 list_t spa_config_dirty_list; /* vdevs with dirty config */ 161 list_t spa_state_dirty_list; /* vdevs with dirty state */ 162 spa_aux_vdev_t spa_spares; /* hot spares */ 163 spa_aux_vdev_t spa_l2cache; /* L2ARC cache devices */ 164 nvlist_t *spa_label_features; /* Features for reading MOS */ 165 uint64_t spa_config_object; /* MOS object for pool config */ 166 uint64_t spa_config_generation; /* config generation number */ 167 uint64_t spa_syncing_txg; /* txg currently syncing */ 168 bpobj_t spa_deferred_bpobj; /* deferred-free bplist */ 169 bplist_t spa_free_bplist[TXG_SIZE]; /* bplist of stuff to free */ 170 zio_cksum_salt_t spa_cksum_salt; /* secret salt for cksum */ 171 /* checksum context templates */ 172 kmutex_t spa_cksum_tmpls_lock; 173 void *spa_cksum_tmpls[ZIO_CHECKSUM_FUNCTIONS]; 174 uberblock_t spa_ubsync; /* last synced uberblock */ 175 uberblock_t spa_uberblock; /* current uberblock */ 176 boolean_t spa_extreme_rewind; /* rewind past deferred frees */ 177 uint64_t spa_last_io; /* lbolt of last non-scan I/O */ 178 kmutex_t spa_scrub_lock; /* resilver/scrub lock */ 179 uint64_t spa_scrub_inflight; /* in-flight scrub I/Os */ 180 kcondvar_t spa_scrub_io_cv; /* scrub I/O completion */ 181 uint8_t spa_scrub_active; /* active or suspended? */ 182 uint8_t spa_scrub_type; /* type of scrub we're doing */ 183 uint8_t spa_scrub_finished; /* indicator to rotate logs */ 184 uint8_t spa_scrub_started; /* started since last boot */ 185 uint8_t spa_scrub_reopen; /* scrub doing vdev_reopen */ 186 uint64_t spa_scan_pass_start; /* start time per pass/reboot */ 187 uint64_t spa_scan_pass_exam; /* examined bytes per pass */ 188 kmutex_t spa_async_lock; /* protect async state */ 189 kthread_t *spa_async_thread; /* thread doing async task */ 190 kthread_t *spa_async_thread_vd; /* thread doing vd async task */ 191 int spa_async_suspended; /* async tasks suspended */ 192 kcondvar_t spa_async_cv; /* wait for thread_exit() */ 193 uint16_t spa_async_tasks; /* async task mask */ 194 char *spa_root; /* alternate root directory */ 195 uint64_t spa_ena; /* spa-wide ereport ENA */ 196 int spa_last_open_failed; /* error if last open failed */ 197 uint64_t spa_last_ubsync_txg; /* "best" uberblock txg */ 198 uint64_t spa_last_ubsync_txg_ts; /* timestamp from that ub */ 199 uint64_t spa_load_txg; /* ub txg that loaded */ 200 uint64_t spa_load_txg_ts; /* timestamp from that ub */ 201 uint64_t spa_load_meta_errors; /* verify metadata err count */ 202 uint64_t spa_load_data_errors; /* verify data err count */ 203 uint64_t spa_verify_min_txg; /* start txg of verify scrub */ 204 kmutex_t spa_errlog_lock; /* error log lock */ 205 uint64_t spa_errlog_last; /* last error log object */ 206 uint64_t spa_errlog_scrub; /* scrub error log object */ 207 kmutex_t spa_errlist_lock; /* error list/ereport lock */ 208 avl_tree_t spa_errlist_last; /* last error list */ 209 avl_tree_t spa_errlist_scrub; /* scrub error list */ 210 uint64_t spa_deflate; /* should we deflate? */ 211 uint64_t spa_history; /* history object */ 212 kmutex_t spa_history_lock; /* history lock */ 213 vdev_t *spa_pending_vdev; /* pending vdev additions */ 214 kmutex_t spa_props_lock; /* property lock */ 215 uint64_t spa_pool_props_object; /* object for properties */ 216 uint64_t spa_bootfs; /* default boot filesystem */ 217 uint64_t spa_failmode; /* failure mode for the pool */ 218 uint64_t spa_delegation; /* delegation on/off */ 219 list_t spa_config_list; /* previous cache file(s) */ 220 /* per-CPU array of root of async I/O: */ 221 zio_t **spa_async_zio_root; 222 zio_t *spa_suspend_zio_root; /* root of all suspended I/O */ 223 kmutex_t spa_suspend_lock; /* protects suspend_zio_root */ 224 kcondvar_t spa_suspend_cv; /* notification of resume */ 225 uint8_t spa_suspended; /* pool is suspended */ 226 uint8_t spa_claiming; /* pool is doing zil_claim() */ 227 boolean_t spa_debug; /* debug enabled? */ 228 boolean_t spa_is_root; /* pool is root */ 229 int spa_minref; /* num refs when first opened */ 230 int spa_mode; /* FREAD | FWRITE */ 231 spa_log_state_t spa_log_state; /* log state */ 232 uint64_t spa_autoexpand; /* lun expansion on/off */ 233 ddt_t *spa_ddt[ZIO_CHECKSUM_FUNCTIONS]; /* in-core DDTs */ 234 uint64_t spa_ddt_stat_object; /* DDT statistics */ 235 uint64_t spa_dedup_ditto; /* dedup ditto threshold */ 236 uint64_t spa_dedup_checksum; /* default dedup checksum */ 237 uint64_t spa_dspace; /* dspace in normal class */ 238 kmutex_t spa_vdev_top_lock; /* dueling offline/remove */ 239 kmutex_t spa_proc_lock; /* protects spa_proc* */ 240 kcondvar_t spa_proc_cv; /* spa_proc_state transitions */ 241 spa_proc_state_t spa_proc_state; /* see definition */ 242 struct proc *spa_proc; /* "zpool-poolname" process */ 243 uint64_t spa_did; /* if procp != p0, did of t1 */ 244 kthread_t *spa_trim_thread; /* thread sending TRIM I/Os */ 245 kmutex_t spa_trim_lock; /* protects spa_trim_cv */ 246 kcondvar_t spa_trim_cv; /* used to notify TRIM thread */ 247 boolean_t spa_autoreplace; /* autoreplace set in open */ 248 int spa_vdev_locks; /* locks grabbed */ 249 uint64_t spa_creation_version; /* version at pool creation */ 250 uint64_t spa_prev_software_version; /* See ub_software_version */ 251 uint64_t spa_feat_for_write_obj; /* required to write to pool */ 252 uint64_t spa_feat_for_read_obj; /* required to read from pool */ 253 uint64_t spa_feat_desc_obj; /* Feature descriptions */ 254 uint64_t spa_feat_enabled_txg_obj; /* Feature enabled txg */ 255 /* cache feature refcounts */ 256 uint64_t spa_feat_refcount_cache[SPA_FEATURES]; 257#ifdef illumos 258 cyclic_id_t spa_deadman_cycid; /* cyclic id */ 259#else /* FreeBSD */ 260#ifdef _KERNEL 261 struct callout spa_deadman_cycid; /* callout id */ 262#endif 263#endif /* illumos */ 264 uint64_t spa_deadman_calls; /* number of deadman calls */ 265 hrtime_t spa_sync_starttime; /* starting time fo spa_sync */ 266 uint64_t spa_deadman_synctime; /* deadman expiration timer */ 267#ifdef illumos 268 /* 269 * spa_iokstat_lock protects spa_iokstat and 270 * spa_queue_stats[]. 271 */ 272 kmutex_t spa_iokstat_lock; 273 struct kstat *spa_iokstat; /* kstat of io to this pool */ 274 struct { 275 int spa_active; 276 int spa_queued; 277 } spa_queue_stats[ZIO_PRIORITY_NUM_QUEUEABLE]; 278#endif 279 hrtime_t spa_ccw_fail_time; /* Conf cache write fail time */ 280 281 /* 282 * spa_refcount & spa_config_lock must be the last elements 283 * because refcount_t changes size based on compilation options. 284 * In order for the MDB module to function correctly, the other 285 * fields must remain in the same location. 286 */ 287 spa_config_lock_t spa_config_lock[SCL_LOCKS]; /* config changes */ 288 refcount_t spa_refcount; /* number of opens */ 289#ifndef sun 290 boolean_t spa_splitting_newspa; /* creating new spa in split */ 291#endif 292}; 293 294extern const char *spa_config_path; 295 296extern void spa_taskq_dispatch_ent(spa_t *spa, zio_type_t t, zio_taskq_type_t q, 297 task_func_t *func, void *arg, uint_t flags, taskq_ent_t *ent); 298 299#ifdef __cplusplus 300} 301#endif 302 303#endif /* _SYS_SPA_IMPL_H */ 304