5 * SOME HIGH LEVEL CODE DOCUMENTATION:
7 * Bcache mostly works with cache sets, cache devices, and backing devices.
9 * Support for multiple cache devices hasn't quite been finished off yet, but
10 * it's about 95% plumbed through. A cache set and its cache devices is sort of
11 * like a md raid array and its component devices. Most of the code doesn't care
12 * about individual cache devices, the main abstraction is the cache set.
14 * Multiple cache devices is intended to give us the ability to mirror dirty
15 * cached data and metadata, without mirroring clean cached data.
17 * Backing devices are different, in that they have a lifetime independent of a
18 * cache set. When you register a newly formatted backing device it'll come up
19 * in passthrough mode, and then you can attach and detach a backing device from
20 * a cache set at runtime - while it's mounted and in use. Detaching implicitly
21 * invalidates any cached data for that backing device.
23 * A cache set can have multiple (many) backing devices attached to it.
25 * There's also flash only volumes - this is the reason for the distinction
26 * between struct cached_dev and struct bcache_device. A flash only volume
27 * works much like a bcache device that has a backing device, except the
28 * "cached" data is always dirty. The end result is that we get thin
29 * provisioning with very little additional code.
31 * Flash only volumes work but they're not production ready because the moving
32 * garbage collector needs more work. More on that later.
36 * Bcache is primarily designed for caching, which means that in normal
37 * operation all of our available space will be allocated. Thus, we need an
38 * efficient way of deleting things from the cache so we can write new things to
41 * To do this, we first divide the cache device up into buckets. A bucket is the
42 * unit of allocation; they're typically around 1 mb - anywhere from 128k to 2M+
45 * Each bucket has a 16 bit priority, and an 8 bit generation associated with
46 * it. The gens and priorities for all the buckets are stored contiguously and
47 * packed on disk (in a linked list of buckets - aside from the superblock, all
48 * of bcache's metadata is stored in buckets).
50 * The priority is used to implement an LRU. We reset a bucket's priority when
51 * we allocate it or on cache it, and every so often we decrement the priority
52 * of each bucket. It could be used to implement something more sophisticated,
53 * if anyone ever gets around to it.
55 * The generation is used for invalidating buckets. Each pointer also has an 8
56 * bit generation embedded in it; for a pointer to be considered valid, its gen
57 * must match the gen of the bucket it points into. Thus, to reuse a bucket all
58 * we have to do is increment its gen (and write its new gen to disk; we batch
61 * Bcache is entirely COW - we never write twice to a bucket, even buckets that
62 * contain metadata (including btree nodes).
66 * Bcache is in large part design around the btree.
68 * At a high level, the btree is just an index of key -> ptr tuples.
70 * Keys represent extents, and thus have a size field. Keys also have a variable
71 * number of pointers attached to them (potentially zero, which is handy for
72 * invalidating the cache).
74 * The key itself is an inode:offset pair. The inode number corresponds to a
75 * backing device or a flash only volume. The offset is the ending offset of the
76 * extent within the inode - not the starting offset; this makes lookups
77 * slightly more convenient.
79 * Pointers contain the cache device id, the offset on that device, and an 8 bit
80 * generation number. More on the gen later.
82 * Index lookups are not fully abstracted - cache lookups in particular are
83 * still somewhat mixed in with the btree code, but things are headed in that
86 * Updates are fairly well abstracted, though. There are two different ways of
87 * updating the btree; insert and replace.
89 * BTREE_INSERT will just take a list of keys and insert them into the btree -
90 * overwriting (possibly only partially) any extents they overlap with. This is
91 * used to update the index after a write.
93 * BTREE_REPLACE is really cmpxchg(); it inserts a key into the btree iff it is
94 * overwriting a key that matches another given key. This is used for inserting
95 * data into the cache after a cache miss, and for background writeback, and for
96 * the moving garbage collector.
98 * There is no "delete" operation; deleting things from the index is
99 * accomplished by either by invalidating pointers (by incrementing a bucket's
100 * gen) or by inserting a key with 0 pointers - which will overwrite anything
101 * previously present at that location in the index.
103 * This means that there are always stale/invalid keys in the btree. They're
104 * filtered out by the code that iterates through a btree node, and removed when
105 * a btree node is rewritten.
109 * Our unit of allocation is a bucket, and we we can't arbitrarily allocate and
110 * free smaller than a bucket - so, that's how big our btree nodes are.
112 * (If buckets are really big we'll only use part of the bucket for a btree node
113 * - no less than 1/4th - but a bucket still contains no more than a single
114 * btree node. I'd actually like to change this, but for now we rely on the
115 * bucket's gen for deleting btree nodes when we rewrite/split a node.)
117 * Anyways, btree nodes are big - big enough to be inefficient with a textbook
118 * btree implementation.
120 * The way this is solved is that btree nodes are internally log structured; we
121 * can append new keys to an existing btree node without rewriting it. This
122 * means each set of keys we write is sorted, but the node is not.
124 * We maintain this log structure in memory - keeping 1Mb of keys sorted would
125 * be expensive, and we have to distinguish between the keys we have written and
126 * the keys we haven't. So to do a lookup in a btree node, we have to search
127 * each sorted set. But we do merge written sets together lazily, so the cost of
128 * these extra searches is quite low (normally most of the keys in a btree node
129 * will be in one big set, and then there'll be one or two sets that are much
132 * This log structure makes bcache's btree more of a hybrid between a
133 * conventional btree and a compacting data structure, with some of the
134 * advantages of both.
136 * GARBAGE COLLECTION:
138 * We can't just invalidate any bucket - it might contain dirty data or
139 * metadata. If it once contained dirty data, other writes might overwrite it
140 * later, leaving no valid pointers into that bucket in the index.
142 * Thus, the primary purpose of garbage collection is to find buckets to reuse.
143 * It also counts how much valid data it each bucket currently contains, so that
144 * allocation can reuse buckets sooner when they've been mostly overwritten.
146 * It also does some things that are really internal to the btree
147 * implementation. If a btree node contains pointers that are stale by more than
148 * some threshold, it rewrites the btree node to avoid the bucket's generation
149 * wrapping around. It also merges adjacent btree nodes if they're empty enough.
153 * Bcache's journal is not necessary for consistency; we always strictly
154 * order metadata writes so that the btree and everything else is consistent on
155 * disk in the event of an unclean shutdown, and in fact bcache had writeback
156 * caching (with recovery from unclean shutdown) before journalling was
159 * Rather, the journal is purely a performance optimization; we can't complete a
160 * write until we've updated the index on disk, otherwise the cache would be
161 * inconsistent in the event of an unclean shutdown. This means that without the
162 * journal, on random write workloads we constantly have to update all the leaf
163 * nodes in the btree, and those writes will be mostly empty (appending at most
164 * a few keys each) - highly inefficient in terms of amount of metadata writes,
165 * and it puts more strain on the various btree resorting/compacting code.
167 * The journal is just a log of keys we've inserted; on startup we just reinsert
168 * all the keys in the open journal entries. That means that when we're updating
169 * a node in the btree, we can wait until a 4k block of keys fills up before
172 * For simplicity, we only journal updates to leaf nodes; updates to parent
173 * nodes are rare enough (since our leaf nodes are huge) that it wasn't worth
174 * the complexity to deal with journalling them (in particular, journal replay)
175 * - updates to non leaf nodes just happen synchronously (see btree_split()).
179 #define pr_fmt(fmt) "bcachefs: %s() " fmt "\n", __func__
181 #include <linux/bug.h>
182 #include <linux/bio.h>
183 #include <linux/closure.h>
184 #include <linux/kobject.h>
185 #include <linux/list.h>
186 #include <linux/mutex.h>
187 #include <linux/percpu-refcount.h>
188 #include <linux/percpu-rwsem.h>
189 #include <linux/rhashtable.h>
190 #include <linux/rwsem.h>
191 #include <linux/seqlock.h>
192 #include <linux/shrinker.h>
193 #include <linux/types.h>
194 #include <linux/workqueue.h>
195 #include <linux/zstd.h>
197 #include "bcachefs_format.h"
202 #include <linux/dynamic_fault.h>
204 #define bch2_fs_init_fault(name) \
205 dynamic_fault("bcachefs:bch_fs_init:" name)
206 #define bch2_meta_read_fault(name) \
207 dynamic_fault("bcachefs:meta:read:" name)
208 #define bch2_meta_write_fault(name) \
209 dynamic_fault("bcachefs:meta:write:" name)
212 #define bch2_fmt(_c, fmt) "bcachefs (%s): " fmt "\n", ((_c)->name)
214 #define bch2_fmt(_c, fmt) fmt "\n"
217 #define bch_info(c, fmt, ...) \
218 printk(KERN_INFO bch2_fmt(c, fmt), ##__VA_ARGS__)
219 #define bch_notice(c, fmt, ...) \
220 printk(KERN_NOTICE bch2_fmt(c, fmt), ##__VA_ARGS__)
221 #define bch_warn(c, fmt, ...) \
222 printk(KERN_WARNING bch2_fmt(c, fmt), ##__VA_ARGS__)
223 #define bch_err(c, fmt, ...) \
224 printk(KERN_ERR bch2_fmt(c, fmt), ##__VA_ARGS__)
226 #define bch_verbose(c, fmt, ...) \
228 if ((c)->opts.verbose_recovery) \
229 bch_info(c, fmt, ##__VA_ARGS__); \
232 #define pr_verbose_init(opts, fmt, ...) \
234 if (opt_get(opts, verbose_init)) \
235 pr_info(fmt, ##__VA_ARGS__); \
238 /* Parameters that are useful for debugging, but should always be compiled in: */
239 #define BCH_DEBUG_PARAMS_ALWAYS() \
240 BCH_DEBUG_PARAM(key_merging_disabled, \
241 "Disables merging of extents") \
242 BCH_DEBUG_PARAM(btree_gc_always_rewrite, \
243 "Causes mark and sweep to compact and rewrite every " \
244 "btree node it traverses") \
245 BCH_DEBUG_PARAM(btree_gc_rewrite_disabled, \
246 "Disables rewriting of btree nodes during mark and sweep")\
247 BCH_DEBUG_PARAM(btree_shrinker_disabled, \
248 "Disables the shrinker callback for the btree node cache")
250 /* Parameters that should only be compiled in in debug mode: */
251 #define BCH_DEBUG_PARAMS_DEBUG() \
252 BCH_DEBUG_PARAM(expensive_debug_checks, \
253 "Enables various runtime debugging checks that " \
254 "significantly affect performance") \
255 BCH_DEBUG_PARAM(debug_check_bkeys, \
256 "Run bkey_debugcheck (primarily checking GC/allocation "\
257 "information) when iterating over keys") \
258 BCH_DEBUG_PARAM(verify_btree_ondisk, \
259 "Reread btree nodes at various points to verify the " \
260 "mergesort in the read path against modifications " \
262 BCH_DEBUG_PARAM(journal_seq_verify, \
263 "Store the journal sequence number in the version " \
264 "number of every btree key, and verify that btree " \
265 "update ordering is preserved during recovery") \
266 BCH_DEBUG_PARAM(inject_invalid_keys, \
267 "Store the journal sequence number in the version " \
268 "number of every btree key, and verify that btree " \
269 "update ordering is preserved during recovery") \
270 BCH_DEBUG_PARAM(test_alloc_startup, \
271 "Force allocator startup to use the slowpath where it" \
272 "can't find enough free buckets without invalidating" \
275 #define BCH_DEBUG_PARAMS_ALL() BCH_DEBUG_PARAMS_ALWAYS() BCH_DEBUG_PARAMS_DEBUG()
277 #ifdef CONFIG_BCACHEFS_DEBUG
278 #define BCH_DEBUG_PARAMS() BCH_DEBUG_PARAMS_ALL()
280 #define BCH_DEBUG_PARAMS() BCH_DEBUG_PARAMS_ALWAYS()
283 #define BCH_TIME_STATS() \
284 x(btree_node_mem_alloc) \
289 x(btree_lock_contended_read) \
290 x(btree_lock_contended_intent) \
291 x(btree_lock_contended_write) \
300 enum bch_time_stats {
301 #define x(name) BCH_TIME_##name,
307 #include "alloc_types.h"
308 #include "btree_types.h"
309 #include "buckets_types.h"
310 #include "clock_types.h"
311 #include "journal_types.h"
312 #include "keylist_types.h"
313 #include "quota_types.h"
314 #include "rebalance_types.h"
315 #include "replicas_types.h"
316 #include "super_types.h"
318 /* Number of nodes btree coalesce will try to coalesce at once */
319 #define GC_MERGE_NODES 4U
321 /* Maximum number of nodes we might need to allocate atomically: */
322 #define BTREE_RESERVE_MAX (BTREE_MAX_DEPTH + (BTREE_MAX_DEPTH - 1))
324 /* Size of the freelist we allocate btree nodes from: */
325 #define BTREE_NODE_RESERVE (BTREE_RESERVE_MAX * 4)
333 #define DEF_BTREE_ID(kwd, val, name) GC_PHASE_BTREE_##kwd,
334 DEFINE_BCH_BTREE_IDS()
337 GC_PHASE_PENDING_DELETE,
349 u64 sectors[2][BCH_DATA_NR];
354 struct percpu_ref ref;
355 struct completion ref_completion;
356 struct percpu_ref io_ref;
357 struct completion io_ref_completion;
363 * Cached version of this device's member info from superblock
364 * Committed by bch2_write_super() -> bch_fs_mi_update()
366 struct bch_member_cpu mi;
368 char name[BDEVNAME_SIZE];
370 struct bch_sb_handle disk_sb;
373 struct bch_devs_mask self;
375 /* biosets used in cloned bios for writing multiple replicas */
376 struct bio_set replica_set;
380 * Per-bucket arrays are protected by c->usage_lock, bucket_lock and
381 * gc_lock, for device resize - holding any is sufficient for access:
382 * Or rcu_read_lock(), but only for ptr_stale():
384 struct bucket_array __rcu *buckets;
385 unsigned long *buckets_dirty;
386 /* most out of date gen in the btree */
388 struct rw_semaphore bucket_lock;
390 struct bch_dev_usage __percpu *usage_percpu;
391 struct bch_dev_usage usage_cached;
394 struct task_struct __rcu *alloc_thread;
397 * free: Buckets that are ready to be used
399 * free_inc: Incoming buckets - these are buckets that currently have
400 * cached data in them, and we can't reuse them until after we write
401 * their new gen to disk. After prio_write() finishes writing the new
402 * gens/prios, they'll be moved to the free list (and possibly discarded
405 alloc_fifo free[RESERVE_NR];
407 spinlock_t freelist_lock;
409 u8 open_buckets_partial[OPEN_BUCKETS_COUNT];
410 unsigned open_buckets_partial_nr;
412 size_t fifo_last_bucket;
414 /* last calculated minimum prio */
415 u16 max_last_bucket_io[2];
417 size_t inc_gen_needs_gc;
418 size_t inc_gen_really_needs_gc;
419 bool allocator_blocked;
421 alloc_heap alloc_heap;
424 struct task_struct *copygc_thread;
425 copygc_heap copygc_heap;
426 struct bch_pd_controller copygc_pd;
427 struct write_point copygc_write_point;
428 u64 copygc_threshold;
430 atomic64_t rebalance_work;
432 struct journal_device journal;
434 struct work_struct io_error_work;
436 /* The rest of this all shows up in sysfs */
437 atomic64_t cur_latency[2];
438 struct time_stats io_latency[2];
440 #define CONGESTED_MAX 1024
444 struct io_count __percpu *io_done;
448 * Flag bits for what phase of startup/shutdown the cache set is at, how we're
449 * shutting down, etc.:
451 * BCH_FS_UNREGISTERING means we're not just shutting down, we're detaching
452 * all the backing devices first (their cached data gets invalidated, and they
453 * won't automatically reattach).
457 BCH_FS_ALLOC_READ_DONE,
458 BCH_FS_ALLOCATOR_STARTED,
459 BCH_FS_INITIAL_GC_DONE,
465 BCH_FS_WRITE_DISABLE_COMPLETE,
473 BCH_FS_FSCK_FIXED_ERRORS,
474 BCH_FS_FSCK_UNFIXED_ERRORS,
476 BCH_FS_REBUILD_REPLICAS,
477 BCH_FS_HOLD_BTREE_WRITES,
482 struct dentry *btree;
483 struct dentry *btree_format;
484 struct dentry *failed;
497 struct list_head list;
499 struct kobject internal;
500 struct kobject opts_dir;
501 struct kobject time_stats;
505 struct device *chardev;
506 struct super_block *vfs_sb;
509 /* ro/rw, add/remove devices: */
510 struct mutex state_lock;
511 enum bch_fs_state state;
513 /* Counts outstanding writes, for clean transition to read-only */
514 struct percpu_ref writes;
515 struct work_struct read_only_work;
517 struct bch_dev __rcu *devs[BCH_SB_MEMBERS_MAX];
519 struct bch_replicas_cpu __rcu *replicas;
520 struct bch_replicas_cpu __rcu *replicas_gc;
521 struct mutex replicas_gc_lock;
523 struct bch_disk_groups_cpu __rcu *disk_groups;
525 struct bch_opts opts;
527 /* Updated by bch2_sb_update():*/
532 u16 encoded_extent_max;
545 struct bch_sb_handle disk_sb;
547 unsigned short block_bits; /* ilog2(block_size) */
549 u16 btree_foreground_merge_threshold;
551 struct closure sb_write;
552 struct mutex sb_lock;
555 struct bio_set btree_bio;
557 struct btree_root btree_roots[BTREE_ID_NR];
558 bool btree_roots_dirty;
559 struct mutex btree_root_lock;
561 struct btree_cache btree_cache;
563 mempool_t btree_reserve_pool;
566 * Cache of allocated btree nodes - if we allocate a btree node and
567 * don't use it, if we free it that space can't be reused until going
568 * _all_ the way through the allocator (which exposes us to a livelock
569 * when allocating btree reserves fail halfway through) - instead, we
570 * can stick them here:
572 struct btree_alloc btree_reserve_cache[BTREE_NODE_RESERVE * 2];
573 unsigned btree_reserve_cache_nr;
574 struct mutex btree_reserve_cache_lock;
576 mempool_t btree_interior_update_pool;
577 struct list_head btree_interior_update_list;
578 struct mutex btree_interior_update_lock;
579 struct closure_waitlist btree_interior_update_wait;
581 mempool_t btree_iters_pool;
583 struct workqueue_struct *wq;
584 /* copygc needs its own workqueue for index updates.. */
585 struct workqueue_struct *copygc_wq;
588 struct delayed_work pd_controllers_update;
589 unsigned pd_controllers_update_seconds;
591 struct bch_devs_mask rw_devs[BCH_DATA_NR];
593 u64 capacity; /* sectors */
596 * When capacity _decreases_ (due to a disk being removed), we
597 * increment capacity_gen - this invalidates outstanding reservations
598 * and forces them to be revalidated
602 atomic64_t sectors_available;
604 struct bch_fs_usage __percpu *usage_percpu;
605 struct bch_fs_usage usage_cached;
606 struct percpu_rw_semaphore usage_lock;
608 struct closure_waitlist freelist_wait;
611 * When we invalidate buckets, we use both the priority and the amount
612 * of good data to determine which buckets to reuse first - to weight
613 * those together consistently we keep track of the smallest nonzero
614 * priority of any bucket.
616 struct bucket_clock bucket_clock[2];
618 struct io_clock io_clock[2];
621 spinlock_t freelist_lock;
622 u8 open_buckets_freelist;
623 u8 open_buckets_nr_free;
624 struct closure_waitlist open_buckets_wait;
625 struct open_bucket open_buckets[OPEN_BUCKETS_COUNT];
627 struct write_point btree_write_point;
628 struct write_point rebalance_write_point;
630 struct write_point write_points[WRITE_POINT_COUNT];
631 struct hlist_head write_points_hash[WRITE_POINT_COUNT];
632 struct mutex write_points_hash_lock;
634 /* GARBAGE COLLECTION */
635 struct task_struct *gc_thread;
637 unsigned long gc_count;
640 * Tracks GC's progress - everything in the range [ZERO_KEY..gc_cur_pos]
641 * has been marked by GC.
643 * gc_cur_phase is a superset of btree_ids (BTREE_ID_EXTENTS etc.)
645 * gc_cur_phase == GC_PHASE_DONE indicates that gc is finished/not
646 * currently running, and gc marks are currently valid
648 * Protected by gc_pos_lock. Only written to by GC thread, so GC thread
649 * can read without a lock.
651 seqcount_t gc_pos_lock;
652 struct gc_pos gc_pos;
655 * The allocation code needs gc_mark in struct bucket to be correct, but
656 * it's not while a gc is in progress.
658 struct rw_semaphore gc_lock;
661 struct bio_set bio_read;
662 struct bio_set bio_read_split;
663 struct bio_set bio_write;
664 struct mutex bio_bounce_pages_lock;
665 mempool_t bio_bounce_pages;
666 struct rhashtable promote_table;
668 mempool_t compression_bounce[2];
669 mempool_t compress_workspace[BCH_COMPRESSION_NR];
670 mempool_t decompress_workspace;
671 ZSTD_parameters zstd_params;
673 struct crypto_shash *sha256;
674 struct crypto_skcipher *chacha20;
675 struct crypto_shash *poly1305;
677 atomic64_t key_version;
680 struct bch_fs_rebalance rebalance;
682 /* VFS IO PATH - fs-io.c */
683 struct bio_set writepage_bioset;
684 struct bio_set dio_write_bioset;
685 struct bio_set dio_read_bioset;
687 struct bio_list btree_write_error_list;
688 struct work_struct btree_write_error_work;
689 spinlock_t btree_write_error_lock;
692 struct list_head fsck_errors;
693 struct mutex fsck_error_lock;
697 atomic_long_t nr_inodes;
700 struct bch_memquota_type quotas[QTYP_NR];
703 struct dentry *debug;
704 struct btree_debug btree_debug[BTREE_ID_NR];
705 #ifdef CONFIG_BCACHEFS_DEBUG
706 struct btree *verify_data;
707 struct btree_node *verify_ondisk;
708 struct mutex verify_lock;
711 u64 unused_inode_hint;
714 * A btree node on disk could have too many bsets for an iterator to fit
715 * on the stack - have to dynamically allocate them
719 mempool_t btree_bounce_pool;
721 struct journal journal;
723 u64 last_bucket_seq_cleanup;
725 /* The rest of this all shows up in sysfs */
726 atomic_long_t read_realloc_races;
727 atomic_long_t extent_migrate_done;
728 atomic_long_t extent_migrate_raced;
730 unsigned btree_gc_periodic:1;
731 unsigned copy_gc_enabled:1;
732 bool promote_whole_extents;
734 #define BCH_DEBUG_PARAM(name, description) bool name;
735 BCH_DEBUG_PARAMS_ALL()
736 #undef BCH_DEBUG_PARAM
738 struct time_stats times[BCH_TIME_STAT_NR];
741 static inline void bch2_set_ra_pages(struct bch_fs *c, unsigned ra_pages)
743 #ifndef NO_BCACHEFS_FS
745 c->vfs_sb->s_bdi->ra_pages = ra_pages;
749 static inline bool bch2_fs_running(struct bch_fs *c)
751 return c->state == BCH_FS_RO || c->state == BCH_FS_RW;
754 static inline unsigned bucket_bytes(const struct bch_dev *ca)
756 return ca->mi.bucket_size << 9;
759 static inline unsigned block_bytes(const struct bch_fs *c)
761 return c->opts.block_size << 9;
764 #endif /* _BCACHEFS_H */