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]
23 * Copyright (c) 2017 by Delphix. All rights reserved.
27 * Storage Pool Checkpoint
29 * A storage pool checkpoint can be thought of as a pool-wide snapshot or
30 * a stable version of extreme rewind that guarantees no blocks from the
31 * checkpointed state will have been overwritten. It remembers the entire
32 * state of the storage pool (e.g. snapshots, dataset names, etc..) from the
33 * point that it was taken and the user can rewind back to that point even if
34 * they applied destructive operations on their datasets or even enabled new
35 * zpool on-disk features. If a pool has a checkpoint that is no longer
36 * needed, the user can discard it.
38 * == On disk data structures used ==
40 * - The pool has a new feature flag and a new entry in the MOS. The feature
41 * flag is set to active when we create the checkpoint and remains active
42 * until the checkpoint is fully discarded. The entry in the MOS config
43 * (DMU_POOL_ZPOOL_CHECKPOINT) is populated with the uberblock that
44 * references the state of the pool when we take the checkpoint. The entry
45 * remains populated until we start discarding the checkpoint or we rewind
48 * - Each vdev contains a vdev-wide space map while the pool has a checkpoint,
49 * which persists until the checkpoint is fully discarded. The space map
50 * contains entries that have been freed in the current state of the pool
51 * but we want to keep around in case we decide to rewind to the checkpoint.
52 * [see vdev_checkpoint_sm]
54 * - Each metaslab's ms_sm space map behaves the same as without the
55 * checkpoint, with the only exception being the scenario when we free
56 * blocks that belong to the checkpoint. In this case, these blocks remain
57 * ALLOCATED in the metaslab's space map and they are added as FREE in the
58 * vdev's checkpoint space map.
60 * - Each uberblock has a field (ub_checkpoint_txg) which holds the txg that
61 * the uberblock was checkpointed. For normal uberblocks this field is 0.
63 * == Overview of operations ==
65 * - To create a checkpoint, we first wait for the current TXG to be synced,
66 * so we can use the most recently synced uberblock (spa_ubsync) as the
67 * checkpointed uberblock. Then we use an early synctask to place that
68 * uberblock in MOS config, increment the feature flag for the checkpoint
69 * (marking it active), and setting spa_checkpoint_txg (see its use below)
70 * to the TXG of the checkpointed uberblock. We use an early synctask for
71 * the aforementioned operations to ensure that no blocks were dirtied
72 * between the current TXG and the TXG of the checkpointed uberblock
73 * (e.g the previous txg).
75 * - When a checkpoint exists, we need to ensure that the blocks that
76 * belong to the checkpoint are freed but never reused. This means that
77 * these blocks should never end up in the ms_allocatable or the ms_freeing
78 * trees of a metaslab. Therefore, whenever there is a checkpoint the new
79 * ms_checkpointing tree is used in addition to the aforementioned ones.
81 * Whenever a block is freed and we find out that it is referenced by the
82 * checkpoint (we find out by comparing its birth to spa_checkpoint_txg),
83 * we place it in the ms_checkpointing tree instead of the ms_freeingtree.
84 * This way, we divide the blocks that are being freed into checkpointed
85 * and not-checkpointed blocks.
87 * In order to persist these frees, we write the extents from the
88 * ms_freeingtree to the ms_sm as usual, and the extents from the
89 * ms_checkpointing tree to the vdev_checkpoint_sm. This way, these
90 * checkpointed extents will remain allocated in the metaslab's ms_sm space
91 * map, and therefore won't be reused [see metaslab_sync()]. In addition,
92 * when we discard the checkpoint, we can find the entries that have
93 * actually been freed in vdev_checkpoint_sm.
94 * [see spa_checkpoint_discard_thread_sync()]
96 * - To discard the checkpoint we use an early synctask to delete the
97 * checkpointed uberblock from the MOS config, set spa_checkpoint_txg to 0,
98 * and wakeup the discarding zthr thread (an open-context async thread).
99 * We use an early synctask to ensure that the operation happens before any
100 * new data end up in the checkpoint's data structures.
102 * Once the synctask is done and the discarding zthr is awake, we discard
103 * the checkpointed data over multiple TXGs by having the zthr prefetching
104 * entries from vdev_checkpoint_sm and then starting a synctask that places
105 * them as free blocks into their respective ms_allocatable and ms_sm
107 * [see spa_checkpoint_discard_thread()]
109 * When there are no entries left in the vdev_checkpoint_sm of all
110 * top-level vdevs, a final synctask runs that decrements the feature flag.
112 * - To rewind to the checkpoint, we first use the current uberblock and
113 * open the MOS so we can access the checkpointed uberblock from the MOS
114 * config. After we retrieve the checkpointed uberblock, we use it as the
115 * current uberblock for the pool by writing it to disk with an updated
116 * TXG, opening its version of the MOS, and moving on as usual from there.
117 * [see spa_ld_checkpoint_rewind()]
119 * An important note on rewinding to the checkpoint has to do with how we
120 * handle ZIL blocks. In the scenario of a rewind, we clear out any ZIL
121 * blocks that have not been claimed by the time we took the checkpoint
122 * as they should no longer be valid.
123 * [see comment in zil_claim()]
125 * == Miscellaneous information ==
127 * - In the hypothetical event that we take a checkpoint, remove a vdev,
128 * and attempt to rewind, the rewind would fail as the checkpointed
129 * uberblock would reference data in the removed device. For this reason
130 * and others of similar nature, we disallow the following operations that
131 * can change the config:
132 * vdev removal and attach/detach, mirror splitting, and pool reguid.
134 * - As most of the checkpoint logic is implemented in the SPA and doesn't
135 * distinguish datasets when it comes to space accounting, having a
136 * checkpoint can potentially break the boundaries set by dataset
140 #include <sys/dmu_tx.h>
141 #include <sys/dsl_dir.h>
142 #include <sys/dsl_synctask.h>
143 #include <sys/metaslab_impl.h>
145 #include <sys/spa_impl.h>
146 #include <sys/spa_checkpoint.h>
147 #include <sys/vdev_impl.h>
149 #include <sys/zfeature.h>
152 * The following parameter limits the amount of memory to be used for the
153 * prefetching of the checkpoint space map done on each vdev while
154 * discarding the checkpoint.
156 * The reason it exists is because top-level vdevs with long checkpoint
157 * space maps can potentially take up a lot of memory depending on the
158 * amount of checkpointed data that has been freed within them while
159 * the pool had a checkpoint.
161 unsigned long zfs_spa_discard_memory_limit = 16 * 1024 * 1024;
164 spa_checkpoint_get_stats(spa_t *spa, pool_checkpoint_stat_t *pcs)
166 if (!spa_feature_is_active(spa, SPA_FEATURE_POOL_CHECKPOINT))
167 return (SET_ERROR(ZFS_ERR_NO_CHECKPOINT));
169 bzero(pcs, sizeof (pool_checkpoint_stat_t));
171 int error = zap_contains(spa_meta_objset(spa),
172 DMU_POOL_DIRECTORY_OBJECT, DMU_POOL_ZPOOL_CHECKPOINT);
173 ASSERT(error == 0 || error == ENOENT);
176 pcs->pcs_state = CS_CHECKPOINT_DISCARDING;
178 pcs->pcs_state = CS_CHECKPOINT_EXISTS;
180 pcs->pcs_space = spa->spa_checkpoint_info.sci_dspace;
181 pcs->pcs_start_time = spa->spa_checkpoint_info.sci_timestamp;
187 spa_checkpoint_discard_complete_sync(void *arg, dmu_tx_t *tx)
191 spa->spa_checkpoint_info.sci_timestamp = 0;
193 spa_feature_decr(spa, SPA_FEATURE_POOL_CHECKPOINT, tx);
194 spa_notify_waiters(spa);
196 spa_history_log_internal(spa, "spa discard checkpoint", tx,
197 "finished discarding checkpointed state from the pool");
200 typedef struct spa_checkpoint_discard_sync_callback_arg {
203 uint64_t sdc_entry_limit;
204 } spa_checkpoint_discard_sync_callback_arg_t;
207 spa_checkpoint_discard_sync_callback(space_map_entry_t *sme, void *arg)
209 spa_checkpoint_discard_sync_callback_arg_t *sdc = arg;
210 vdev_t *vd = sdc->sdc_vd;
211 metaslab_t *ms = vd->vdev_ms[sme->sme_offset >> vd->vdev_ms_shift];
212 uint64_t end = sme->sme_offset + sme->sme_run;
214 if (sdc->sdc_entry_limit == 0)
218 * Since the space map is not condensed, we know that
219 * none of its entries is crossing the boundaries of
220 * its respective metaslab.
222 * That said, there is no fundamental requirement that
223 * the checkpoint's space map entries should not cross
224 * metaslab boundaries. So if needed we could add code
225 * that handles metaslab-crossing segments in the future.
227 VERIFY3U(sme->sme_type, ==, SM_FREE);
228 VERIFY3U(sme->sme_offset, >=, ms->ms_start);
229 VERIFY3U(end, <=, ms->ms_start + ms->ms_size);
232 * At this point we should not be processing any
233 * other frees concurrently, so the lock is technically
234 * unnecessary. We use the lock anyway though to
235 * potentially save ourselves from future headaches.
237 mutex_enter(&ms->ms_lock);
238 if (range_tree_is_empty(ms->ms_freeing))
239 vdev_dirty(vd, VDD_METASLAB, ms, sdc->sdc_txg);
240 range_tree_add(ms->ms_freeing, sme->sme_offset, sme->sme_run);
241 mutex_exit(&ms->ms_lock);
243 ASSERT3U(vd->vdev_spa->spa_checkpoint_info.sci_dspace, >=,
245 ASSERT3U(vd->vdev_stat.vs_checkpoint_space, >=, sme->sme_run);
247 vd->vdev_spa->spa_checkpoint_info.sci_dspace -= sme->sme_run;
248 vd->vdev_stat.vs_checkpoint_space -= sme->sme_run;
249 sdc->sdc_entry_limit--;
256 spa_checkpoint_accounting_verify(spa_t *spa)
258 vdev_t *rvd = spa->spa_root_vdev;
259 uint64_t ckpoint_sm_space_sum = 0;
260 uint64_t vs_ckpoint_space_sum = 0;
262 for (uint64_t c = 0; c < rvd->vdev_children; c++) {
263 vdev_t *vd = rvd->vdev_child[c];
265 if (vd->vdev_checkpoint_sm != NULL) {
266 ckpoint_sm_space_sum +=
267 -space_map_allocated(vd->vdev_checkpoint_sm);
268 vs_ckpoint_space_sum +=
269 vd->vdev_stat.vs_checkpoint_space;
270 ASSERT3U(ckpoint_sm_space_sum, ==,
271 vs_ckpoint_space_sum);
273 ASSERT0(vd->vdev_stat.vs_checkpoint_space);
276 ASSERT3U(spa->spa_checkpoint_info.sci_dspace, ==, ckpoint_sm_space_sum);
281 spa_checkpoint_discard_thread_sync(void *arg, dmu_tx_t *tx)
287 * The space map callback is applied only to non-debug entries.
288 * Because the number of debug entries is less or equal to the
289 * number of non-debug entries, we want to ensure that we only
290 * read what we prefetched from open-context.
292 * Thus, we set the maximum entries that the space map callback
293 * will be applied to be half the entries that could fit in the
294 * imposed memory limit.
296 * Note that since this is a conservative estimate we also
297 * assume the worst case scenario in our computation where each
300 uint64_t max_entry_limit =
301 (zfs_spa_discard_memory_limit / (2 * sizeof (uint64_t))) >> 1;
304 * Iterate from the end of the space map towards the beginning,
305 * placing its entries on ms_freeing and removing them from the
306 * space map. The iteration stops if one of the following
307 * conditions is true:
309 * 1] We reached the beginning of the space map. At this point
310 * the space map should be completely empty and
311 * space_map_incremental_destroy should have returned 0.
312 * The next step would be to free and close the space map
313 * and remove its entry from its vdev's top zap. This allows
314 * spa_checkpoint_discard_thread() to move on to the next vdev.
316 * 2] We reached the memory limit (amount of memory used to hold
317 * space map entries in memory) and space_map_incremental_destroy
318 * returned EINTR. This means that there are entries remaining
319 * in the space map that will be cleared in a future invocation
320 * of this function by spa_checkpoint_discard_thread().
322 spa_checkpoint_discard_sync_callback_arg_t sdc;
324 sdc.sdc_txg = tx->tx_txg;
325 sdc.sdc_entry_limit = max_entry_limit;
327 uint64_t words_before =
328 space_map_length(vd->vdev_checkpoint_sm) / sizeof (uint64_t);
330 error = space_map_incremental_destroy(vd->vdev_checkpoint_sm,
331 spa_checkpoint_discard_sync_callback, &sdc, tx);
333 uint64_t words_after =
334 space_map_length(vd->vdev_checkpoint_sm) / sizeof (uint64_t);
337 spa_checkpoint_accounting_verify(vd->vdev_spa);
340 zfs_dbgmsg("discarding checkpoint: txg %llu, vdev id %d, "
341 "deleted %llu words - %llu words are left",
342 tx->tx_txg, vd->vdev_id, (words_before - words_after),
345 if (error != EINTR) {
347 zfs_panic_recover("zfs: error %d was returned "
348 "while incrementally destroying the checkpoint "
349 "space map of vdev %llu\n",
352 ASSERT0(words_after);
353 ASSERT0(space_map_allocated(vd->vdev_checkpoint_sm));
354 ASSERT0(space_map_length(vd->vdev_checkpoint_sm));
356 space_map_free(vd->vdev_checkpoint_sm, tx);
357 space_map_close(vd->vdev_checkpoint_sm);
358 vd->vdev_checkpoint_sm = NULL;
360 VERIFY0(zap_remove(spa_meta_objset(vd->vdev_spa),
361 vd->vdev_top_zap, VDEV_TOP_ZAP_POOL_CHECKPOINT_SM, tx));
366 spa_checkpoint_discard_is_done(spa_t *spa)
368 vdev_t *rvd = spa->spa_root_vdev;
370 ASSERT(!spa_has_checkpoint(spa));
371 ASSERT(spa_feature_is_active(spa, SPA_FEATURE_POOL_CHECKPOINT));
373 for (uint64_t c = 0; c < rvd->vdev_children; c++) {
374 if (rvd->vdev_child[c]->vdev_checkpoint_sm != NULL)
376 ASSERT0(rvd->vdev_child[c]->vdev_stat.vs_checkpoint_space);
384 spa_checkpoint_discard_thread_check(void *arg, zthr_t *zthr)
388 if (!spa_feature_is_active(spa, SPA_FEATURE_POOL_CHECKPOINT))
391 if (spa_has_checkpoint(spa))
398 spa_checkpoint_discard_thread(void *arg, zthr_t *zthr)
401 vdev_t *rvd = spa->spa_root_vdev;
403 for (uint64_t c = 0; c < rvd->vdev_children; c++) {
404 vdev_t *vd = rvd->vdev_child[c];
406 while (vd->vdev_checkpoint_sm != NULL) {
407 space_map_t *checkpoint_sm = vd->vdev_checkpoint_sm;
411 if (zthr_iscancelled(zthr))
414 ASSERT3P(vd->vdev_ops, !=, &vdev_indirect_ops);
416 uint64_t size = MIN(space_map_length(checkpoint_sm),
417 zfs_spa_discard_memory_limit);
419 space_map_length(checkpoint_sm) - size;
422 * Ensure that the part of the space map that will
423 * be destroyed by the synctask, is prefetched in
424 * memory before the synctask runs.
426 int error = dmu_buf_hold_array_by_bonus(
427 checkpoint_sm->sm_dbuf, offset, size,
428 B_TRUE, FTAG, &numbufs, &dbp);
430 zfs_panic_recover("zfs: error %d was returned "
431 "while prefetching checkpoint space map "
432 "entries of vdev %llu\n",
436 VERIFY0(dsl_sync_task(spa->spa_name, NULL,
437 spa_checkpoint_discard_thread_sync, vd,
438 0, ZFS_SPACE_CHECK_NONE));
440 dmu_buf_rele_array(dbp, numbufs, FTAG);
444 VERIFY(spa_checkpoint_discard_is_done(spa));
445 VERIFY0(spa->spa_checkpoint_info.sci_dspace);
446 VERIFY0(dsl_sync_task(spa->spa_name, NULL,
447 spa_checkpoint_discard_complete_sync, spa,
448 0, ZFS_SPACE_CHECK_NONE));
454 spa_checkpoint_check(void *arg, dmu_tx_t *tx)
456 spa_t *spa = dmu_tx_pool(tx)->dp_spa;
458 if (!spa_feature_is_enabled(spa, SPA_FEATURE_POOL_CHECKPOINT))
459 return (SET_ERROR(ENOTSUP));
461 if (!spa_top_vdevs_spacemap_addressable(spa))
462 return (SET_ERROR(ZFS_ERR_VDEV_TOO_BIG));
464 if (spa->spa_removing_phys.sr_state == DSS_SCANNING)
465 return (SET_ERROR(ZFS_ERR_DEVRM_IN_PROGRESS));
467 if (spa->spa_checkpoint_txg != 0)
468 return (SET_ERROR(ZFS_ERR_CHECKPOINT_EXISTS));
470 if (spa_feature_is_active(spa, SPA_FEATURE_POOL_CHECKPOINT))
471 return (SET_ERROR(ZFS_ERR_DISCARDING_CHECKPOINT));
478 spa_checkpoint_sync(void *arg, dmu_tx_t *tx)
480 dsl_pool_t *dp = dmu_tx_pool(tx);
481 spa_t *spa = dp->dp_spa;
482 uberblock_t checkpoint = spa->spa_ubsync;
485 * At this point, there should not be a checkpoint in the MOS.
487 ASSERT3U(zap_contains(spa_meta_objset(spa), DMU_POOL_DIRECTORY_OBJECT,
488 DMU_POOL_ZPOOL_CHECKPOINT), ==, ENOENT);
490 ASSERT0(spa->spa_checkpoint_info.sci_timestamp);
491 ASSERT0(spa->spa_checkpoint_info.sci_dspace);
494 * Since the checkpointed uberblock is the one that just got synced
495 * (we use spa_ubsync), its txg must be equal to the txg number of
496 * the txg we are syncing, minus 1.
498 ASSERT3U(checkpoint.ub_txg, ==, spa->spa_syncing_txg - 1);
501 * Once the checkpoint is in place, we need to ensure that none of
502 * its blocks will be marked for reuse after it has been freed.
503 * When there is a checkpoint and a block is freed, we compare its
504 * birth txg to the txg of the checkpointed uberblock to see if the
505 * block is part of the checkpoint or not. Therefore, we have to set
506 * spa_checkpoint_txg before any frees happen in this txg (which is
507 * why this is done as an early_synctask as explained in the comment
508 * in spa_checkpoint()).
510 spa->spa_checkpoint_txg = checkpoint.ub_txg;
511 spa->spa_checkpoint_info.sci_timestamp = checkpoint.ub_timestamp;
513 checkpoint.ub_checkpoint_txg = checkpoint.ub_txg;
514 VERIFY0(zap_add(spa->spa_dsl_pool->dp_meta_objset,
515 DMU_POOL_DIRECTORY_OBJECT, DMU_POOL_ZPOOL_CHECKPOINT,
516 sizeof (uint64_t), sizeof (uberblock_t) / sizeof (uint64_t),
520 * Increment the feature refcount and thus activate the feature.
521 * Note that the feature will be deactivated when we've
522 * completely discarded all checkpointed state (both vdev
523 * space maps and uberblock).
525 spa_feature_incr(spa, SPA_FEATURE_POOL_CHECKPOINT, tx);
527 spa_history_log_internal(spa, "spa checkpoint", tx,
528 "checkpointed uberblock txg=%llu", (u_longlong_t)checkpoint.ub_txg);
532 * Create a checkpoint for the pool.
535 spa_checkpoint(const char *pool)
540 error = spa_open(pool, &spa, FTAG);
544 mutex_enter(&spa->spa_vdev_top_lock);
547 * Wait for current syncing txg to finish so the latest synced
548 * uberblock (spa_ubsync) has all the changes that we expect
549 * to see if we were to revert later to the checkpoint. In other
550 * words we want the checkpointed uberblock to include/reference
551 * all the changes that were pending at the time that we issued
552 * the checkpoint command.
554 txg_wait_synced(spa_get_dsl(spa), 0);
557 * As the checkpointed uberblock references blocks from the previous
558 * txg (spa_ubsync) we want to ensure that are not freeing any of
559 * these blocks in the same txg that the following synctask will
560 * run. Thus, we run it as an early synctask, so the dirty changes
561 * that are synced to disk afterwards during zios and other synctasks
562 * do not reuse checkpointed blocks.
564 error = dsl_early_sync_task(pool, spa_checkpoint_check,
565 spa_checkpoint_sync, NULL, 0, ZFS_SPACE_CHECK_NORMAL);
567 mutex_exit(&spa->spa_vdev_top_lock);
569 spa_close(spa, FTAG);
575 spa_checkpoint_discard_check(void *arg, dmu_tx_t *tx)
577 spa_t *spa = dmu_tx_pool(tx)->dp_spa;
579 if (!spa_feature_is_active(spa, SPA_FEATURE_POOL_CHECKPOINT))
580 return (SET_ERROR(ZFS_ERR_NO_CHECKPOINT));
582 if (spa->spa_checkpoint_txg == 0)
583 return (SET_ERROR(ZFS_ERR_DISCARDING_CHECKPOINT));
585 VERIFY0(zap_contains(spa_meta_objset(spa),
586 DMU_POOL_DIRECTORY_OBJECT, DMU_POOL_ZPOOL_CHECKPOINT));
593 spa_checkpoint_discard_sync(void *arg, dmu_tx_t *tx)
595 spa_t *spa = dmu_tx_pool(tx)->dp_spa;
597 VERIFY0(zap_remove(spa_meta_objset(spa), DMU_POOL_DIRECTORY_OBJECT,
598 DMU_POOL_ZPOOL_CHECKPOINT, tx));
600 spa->spa_checkpoint_txg = 0;
602 zthr_wakeup(spa->spa_checkpoint_discard_zthr);
604 spa_history_log_internal(spa, "spa discard checkpoint", tx,
605 "started discarding checkpointed state from the pool");
609 * Discard the checkpoint from a pool.
612 spa_checkpoint_discard(const char *pool)
615 * Similarly to spa_checkpoint(), we want our synctask to run
616 * before any pending dirty data are written to disk so they
617 * won't end up in the checkpoint's data structures (e.g.
618 * ms_checkpointing and vdev_checkpoint_sm) and re-create any
619 * space maps that the discarding open-context thread has
621 * [see spa_discard_checkpoint_sync and spa_discard_checkpoint_thread]
623 return (dsl_early_sync_task(pool, spa_checkpoint_discard_check,
624 spa_checkpoint_discard_sync, NULL, 0,
625 ZFS_SPACE_CHECK_DISCARD_CHECKPOINT));
628 EXPORT_SYMBOL(spa_checkpoint_get_stats);
629 EXPORT_SYMBOL(spa_checkpoint_discard_thread);
630 EXPORT_SYMBOL(spa_checkpoint_discard_thread_check);
633 ZFS_MODULE_PARAM(zfs_spa, zfs_spa_, discard_memory_limit, ULONG, ZMOD_RW,
634 "Limit for memory used in prefetching the checkpoint space map done "
635 "on each vdev while discarding the checkpoint");