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 | #if defined(__NetBSD__) && defined(_KERNEL) |
48 | #include <sys/workqueue.h> |
49 | #endif |
50 | |
51 | #ifdef __cplusplus |
52 | extern "C" { |
53 | #endif |
54 | |
55 | typedef 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 | |
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 */ |
67 | } spa_history_phys_t; |
68 | |
69 | struct spa_aux_vdev { |
70 | uint64_t sav_object; /* MOS object for device list */ |
71 | nvlist_t *sav_config; /* cached device config */ |
72 | vdev_t **sav_vdevs; /* devices */ |
73 | int sav_count; /* number devices */ |
74 | boolean_t sav_sync; /* sync the device list */ |
75 | nvlist_t **sav_pending; /* pending device additions */ |
76 | uint_t sav_npending; /* # pending devices */ |
77 | }; |
78 | |
79 | typedef struct spa_config_lock { |
80 | kmutex_t scl_lock; |
81 | kthread_t *scl_writer; |
82 | int scl_write_wanted; |
83 | kcondvar_t scl_cv; |
84 | refcount_t scl_count; |
85 | } spa_config_lock_t; |
86 | |
87 | typedef struct spa_config_dirent { |
88 | list_node_t scd_link; |
89 | char *scd_path; |
90 | } spa_config_dirent_t; |
91 | |
92 | typedef enum zio_taskq_type { |
93 | ZIO_TASKQ_ISSUE = 0, |
94 | ZIO_TASKQ_ISSUE_HIGH, |
95 | ZIO_TASKQ_INTERRUPT, |
96 | ZIO_TASKQ_INTERRUPT_HIGH, |
97 | ZIO_TASKQ_TYPES |
98 | } zio_taskq_type_t; |
99 | |
100 | /* |
101 | * State machine for the zpool-poolname process. The states transitions |
102 | * are done as follows: |
103 | * |
104 | * From To Routine |
105 | * PROC_NONE -> PROC_CREATED spa_activate() |
106 | * PROC_CREATED -> PROC_ACTIVE spa_thread() |
107 | * PROC_ACTIVE -> PROC_DEACTIVATE spa_deactivate() |
108 | * PROC_DEACTIVATE -> PROC_GONE spa_thread() |
109 | * PROC_GONE -> PROC_NONE spa_deactivate() |
110 | */ |
111 | typedef enum spa_proc_state { |
112 | SPA_PROC_NONE, /* spa_proc = &p0, no process created */ |
113 | SPA_PROC_CREATED, /* spa_activate() has proc, is waiting */ |
114 | SPA_PROC_ACTIVE, /* taskqs created, spa_proc set */ |
115 | SPA_PROC_DEACTIVATE, /* spa_deactivate() requests process exit */ |
116 | SPA_PROC_GONE /* spa_thread() is exiting, spa_proc = &p0 */ |
117 | } spa_proc_state_t; |
118 | |
119 | typedef struct spa_taskqs { |
120 | uint_t stqs_count; |
121 | taskq_t **stqs_taskq; |
122 | } spa_taskqs_t; |
123 | |
124 | typedef enum spa_all_vdev_zap_action { |
125 | AVZ_ACTION_NONE = 0, |
126 | AVZ_ACTION_DESTROY, /* Destroy all per-vdev ZAPs and the AVZ. */ |
127 | AVZ_ACTION_REBUILD /* Populate the new AVZ, see spa_avz_rebuild */ |
128 | } spa_avz_action_t; |
129 | |
130 | struct spa { |
131 | /* |
132 | * Fields protected by spa_namespace_lock. |
133 | */ |
134 | char spa_name[ZFS_MAX_DATASET_NAME_LEN]; /* pool name */ |
135 | char *; /* comment */ |
136 | avl_node_t spa_avl; /* node in spa_namespace_avl */ |
137 | nvlist_t *spa_config; /* last synced config */ |
138 | nvlist_t *spa_config_syncing; /* currently syncing config */ |
139 | nvlist_t *spa_config_splitting; /* config for splitting */ |
140 | nvlist_t *spa_load_info; /* info and errors from load */ |
141 | uint64_t spa_config_txg; /* txg of last config change */ |
142 | int spa_sync_pass; /* iterate-to-convergence */ |
143 | pool_state_t spa_state; /* pool state */ |
144 | int spa_inject_ref; /* injection references */ |
145 | uint8_t spa_sync_on; /* sync threads are running */ |
146 | spa_load_state_t spa_load_state; /* current load operation */ |
147 | uint64_t spa_import_flags; /* import specific flags */ |
148 | spa_taskqs_t spa_zio_taskq[ZIO_TYPES][ZIO_TASKQ_TYPES]; |
149 | dsl_pool_t *spa_dsl_pool; |
150 | boolean_t spa_is_initializing; /* true while opening pool */ |
151 | metaslab_class_t *spa_normal_class; /* normal data class */ |
152 | metaslab_class_t *spa_log_class; /* intent log data class */ |
153 | uint64_t spa_first_txg; /* first txg after spa_open() */ |
154 | uint64_t spa_final_txg; /* txg of export/destroy */ |
155 | uint64_t spa_freeze_txg; /* freeze pool at this txg */ |
156 | uint64_t spa_load_max_txg; /* best initial ub_txg */ |
157 | uint64_t spa_claim_max_txg; /* highest claimed birth txg */ |
158 | timespec_t spa_loaded_ts; /* 1st successful open time */ |
159 | objset_t *spa_meta_objset; /* copy of dp->dp_meta_objset */ |
160 | kmutex_t spa_evicting_os_lock; /* Evicting objset list lock */ |
161 | list_t spa_evicting_os_list; /* Objsets being evicted. */ |
162 | kcondvar_t spa_evicting_os_cv; /* Objset Eviction Completion */ |
163 | txg_list_t spa_vdev_txg_list; /* per-txg dirty vdev list */ |
164 | vdev_t *spa_root_vdev; /* top-level vdev container */ |
165 | int spa_min_ashift; /* of vdevs in normal class */ |
166 | int spa_max_ashift; /* of vdevs in normal class */ |
167 | uint64_t spa_config_guid; /* config pool guid */ |
168 | uint64_t spa_load_guid; /* spa_load initialized guid */ |
169 | uint64_t spa_last_synced_guid; /* last synced guid */ |
170 | list_t spa_config_dirty_list; /* vdevs with dirty config */ |
171 | list_t spa_state_dirty_list; /* vdevs with dirty state */ |
172 | kmutex_t spa_alloc_lock; |
173 | avl_tree_t spa_alloc_tree; |
174 | spa_aux_vdev_t spa_spares; /* hot spares */ |
175 | spa_aux_vdev_t spa_l2cache; /* L2ARC cache devices */ |
176 | nvlist_t *spa_label_features; /* Features for reading MOS */ |
177 | uint64_t spa_config_object; /* MOS object for pool config */ |
178 | uint64_t spa_config_generation; /* config generation number */ |
179 | uint64_t spa_syncing_txg; /* txg currently syncing */ |
180 | bpobj_t spa_deferred_bpobj; /* deferred-free bplist */ |
181 | bplist_t spa_free_bplist[TXG_SIZE]; /* bplist of stuff to free */ |
182 | zio_cksum_salt_t spa_cksum_salt; /* secret salt for cksum */ |
183 | /* checksum context templates */ |
184 | kmutex_t spa_cksum_tmpls_lock; |
185 | void *spa_cksum_tmpls[ZIO_CHECKSUM_FUNCTIONS]; |
186 | uberblock_t spa_ubsync; /* last synced uberblock */ |
187 | uberblock_t spa_uberblock; /* current uberblock */ |
188 | boolean_t spa_extreme_rewind; /* rewind past deferred frees */ |
189 | uint64_t spa_last_io; /* lbolt of last non-scan I/O */ |
190 | kmutex_t spa_scrub_lock; /* resilver/scrub lock */ |
191 | uint64_t spa_scrub_inflight; /* in-flight scrub I/Os */ |
192 | kcondvar_t spa_scrub_io_cv; /* scrub I/O completion */ |
193 | uint8_t spa_scrub_active; /* active or suspended? */ |
194 | uint8_t spa_scrub_type; /* type of scrub we're doing */ |
195 | uint8_t spa_scrub_finished; /* indicator to rotate logs */ |
196 | uint8_t spa_scrub_started; /* started since last boot */ |
197 | uint8_t spa_scrub_reopen; /* scrub doing vdev_reopen */ |
198 | uint64_t spa_scan_pass_start; /* start time per pass/reboot */ |
199 | uint64_t spa_scan_pass_exam; /* examined bytes per pass */ |
200 | kmutex_t spa_async_lock; /* protect async state */ |
201 | kthread_t *spa_async_thread; /* thread doing async task */ |
202 | kthread_t *spa_async_thread_vd; /* thread doing vd async task */ |
203 | int spa_async_suspended; /* async tasks suspended */ |
204 | kcondvar_t spa_async_cv; /* wait for thread_exit() */ |
205 | uint16_t spa_async_tasks; /* async task mask */ |
206 | char *spa_root; /* alternate root directory */ |
207 | uint64_t spa_ena; /* spa-wide ereport ENA */ |
208 | int spa_last_open_failed; /* error if last open failed */ |
209 | uint64_t spa_last_ubsync_txg; /* "best" uberblock txg */ |
210 | uint64_t spa_last_ubsync_txg_ts; /* timestamp from that ub */ |
211 | uint64_t spa_load_txg; /* ub txg that loaded */ |
212 | uint64_t spa_load_txg_ts; /* timestamp from that ub */ |
213 | uint64_t spa_load_meta_errors; /* verify metadata err count */ |
214 | uint64_t spa_load_data_errors; /* verify data err count */ |
215 | uint64_t spa_verify_min_txg; /* start txg of verify scrub */ |
216 | kmutex_t spa_errlog_lock; /* error log lock */ |
217 | uint64_t spa_errlog_last; /* last error log object */ |
218 | uint64_t spa_errlog_scrub; /* scrub error log object */ |
219 | kmutex_t spa_errlist_lock; /* error list/ereport lock */ |
220 | avl_tree_t spa_errlist_last; /* last error list */ |
221 | avl_tree_t spa_errlist_scrub; /* scrub error list */ |
222 | uint64_t spa_deflate; /* should we deflate? */ |
223 | uint64_t spa_history; /* history object */ |
224 | kmutex_t spa_history_lock; /* history lock */ |
225 | vdev_t *spa_pending_vdev; /* pending vdev additions */ |
226 | kmutex_t spa_props_lock; /* property lock */ |
227 | uint64_t spa_pool_props_object; /* object for properties */ |
228 | uint64_t spa_bootfs; /* default boot filesystem */ |
229 | uint64_t spa_failmode; /* failure mode for the pool */ |
230 | uint64_t spa_delegation; /* delegation on/off */ |
231 | list_t spa_config_list; /* previous cache file(s) */ |
232 | /* per-CPU array of root of async I/O: */ |
233 | zio_t **spa_async_zio_root; |
234 | zio_t *spa_suspend_zio_root; /* root of all suspended I/O */ |
235 | kmutex_t spa_suspend_lock; /* protects suspend_zio_root */ |
236 | kcondvar_t spa_suspend_cv; /* notification of resume */ |
237 | uint8_t spa_suspended; /* pool is suspended */ |
238 | uint8_t spa_claiming; /* pool is doing zil_claim() */ |
239 | boolean_t spa_debug; /* debug enabled? */ |
240 | boolean_t spa_is_root; /* pool is root */ |
241 | int spa_minref; /* num refs when first opened */ |
242 | int spa_mode; /* FREAD | FWRITE */ |
243 | spa_log_state_t spa_log_state; /* log state */ |
244 | uint64_t spa_autoexpand; /* lun expansion on/off */ |
245 | ddt_t *spa_ddt[ZIO_CHECKSUM_FUNCTIONS]; /* in-core DDTs */ |
246 | uint64_t spa_ddt_stat_object; /* DDT statistics */ |
247 | uint64_t spa_dedup_ditto; /* dedup ditto threshold */ |
248 | uint64_t spa_dedup_checksum; /* default dedup checksum */ |
249 | uint64_t spa_dspace; /* dspace in normal class */ |
250 | kmutex_t spa_vdev_top_lock; /* dueling offline/remove */ |
251 | kmutex_t spa_proc_lock; /* protects spa_proc* */ |
252 | kcondvar_t spa_proc_cv; /* spa_proc_state transitions */ |
253 | spa_proc_state_t spa_proc_state; /* see definition */ |
254 | struct proc *spa_proc; /* "zpool-poolname" process */ |
255 | uint64_t spa_did; /* if procp != p0, did of t1 */ |
256 | kthread_t *spa_trim_thread; /* thread sending TRIM I/Os */ |
257 | kmutex_t spa_trim_lock; /* protects spa_trim_cv */ |
258 | kcondvar_t spa_trim_cv; /* used to notify TRIM thread */ |
259 | boolean_t spa_autoreplace; /* autoreplace set in open */ |
260 | int spa_vdev_locks; /* locks grabbed */ |
261 | uint64_t spa_creation_version; /* version at pool creation */ |
262 | uint64_t spa_prev_software_version; /* See ub_software_version */ |
263 | uint64_t spa_feat_for_write_obj; /* required to write to pool */ |
264 | uint64_t spa_feat_for_read_obj; /* required to read from pool */ |
265 | uint64_t spa_feat_desc_obj; /* Feature descriptions */ |
266 | uint64_t spa_feat_enabled_txg_obj; /* Feature enabled txg */ |
267 | /* cache feature refcounts */ |
268 | uint64_t spa_feat_refcount_cache[SPA_FEATURES]; |
269 | #ifdef illumos |
270 | cyclic_id_t spa_deadman_cycid; /* cyclic id */ |
271 | #endif /* illumos */ |
272 | #ifdef __FreeBSD__ |
273 | #ifdef _KERNEL |
274 | struct callout spa_deadman_cycid; /* callout id */ |
275 | struct task spa_deadman_task; |
276 | #endif |
277 | #endif /* __FreeBSD__ */ |
278 | #ifdef __NetBSD__ |
279 | #ifdef _KERNEL |
280 | struct callout spa_deadman_cycid; /* callout id */ |
281 | struct work spa_deadman_work; |
282 | #endif |
283 | #endif /* __NetBSD__ */ |
284 | uint64_t spa_deadman_calls; /* number of deadman calls */ |
285 | hrtime_t spa_sync_starttime; /* starting time fo spa_sync */ |
286 | uint64_t spa_deadman_synctime; /* deadman expiration timer */ |
287 | uint64_t spa_all_vdev_zaps; /* ZAP of per-vd ZAP obj #s */ |
288 | spa_avz_action_t spa_avz_action; /* destroy/rebuild AVZ? */ |
289 | |
290 | #ifdef illumos |
291 | /* |
292 | * spa_iokstat_lock protects spa_iokstat and |
293 | * spa_queue_stats[]. |
294 | */ |
295 | kmutex_t spa_iokstat_lock; |
296 | struct kstat *spa_iokstat; /* kstat of io to this pool */ |
297 | struct { |
298 | int spa_active; |
299 | int spa_queued; |
300 | } spa_queue_stats[ZIO_PRIORITY_NUM_QUEUEABLE]; |
301 | #endif |
302 | hrtime_t spa_ccw_fail_time; /* Conf cache write fail time */ |
303 | |
304 | /* |
305 | * spa_refcount & spa_config_lock must be the last elements |
306 | * because refcount_t changes size based on compilation options. |
307 | * In order for the MDB module to function correctly, the other |
308 | * fields must remain in the same location. |
309 | */ |
310 | spa_config_lock_t spa_config_lock[SCL_LOCKS]; /* config changes */ |
311 | refcount_t spa_refcount; /* number of opens */ |
312 | #ifndef illumos |
313 | boolean_t spa_splitting_newspa; /* creating new spa in split */ |
314 | #endif |
315 | }; |
316 | |
317 | extern const char *spa_config_path; |
318 | |
319 | extern void spa_taskq_dispatch_ent(spa_t *spa, zio_type_t t, zio_taskq_type_t q, |
320 | task_func_t *func, void *arg, uint_t flags, taskq_ent_t *ent); |
321 | |
322 | #ifdef __cplusplus |
323 | } |
324 | #endif |
325 | |
326 | #endif /* _SYS_SPA_IMPL_H */ |
327 | |