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()).
178 #define pr_fmt(fmt) "bcache: %s() " fmt "\n", __func__
180 #include <linux/bcache.h>
181 #include <linux/bio.h>
182 #include <linux/kobject.h>
183 #include <linux/list.h>
184 #include <linux/mutex.h>
185 #include <linux/rbtree.h>
186 #include <linux/rwsem.h>
187 #include <linux/types.h>
188 #include <linux/workqueue.h>
198 uint8_t last_gc
; /* Most out of date gen in the btree */
204 * I'd use bitfields for these, but I don't trust the compiler not to screw me
205 * as multiple threads touch struct bucket without locking
208 BITMASK(GC_MARK
, struct bucket
, gc_mark
, 0, 2);
209 #define GC_MARK_RECLAIMABLE 0
210 #define GC_MARK_DIRTY 1
211 #define GC_MARK_METADATA 2
212 BITMASK(GC_SECTORS_USED
, struct bucket
, gc_mark
, 2, 14);
227 struct bkey last_scanned
;
231 * Beginning and end of range in rb tree - so that we can skip taking
232 * lock and checking the rb tree when we need to check for overlapping
240 #define KEYBUF_NR 500
241 DECLARE_ARRAY_ALLOCATOR(struct keybuf_key
, freelist
, KEYBUF_NR
);
244 struct bio_split_pool
{
245 struct bio_set
*bio_split
;
246 mempool_t
*bio_split_hook
;
249 struct bio_split_hook
{
251 struct bio_split_pool
*p
;
253 bio_end_io_t
*bi_end_io
;
257 struct bcache_device
{
264 #define BCACHEDEVNAME_SIZE 12
265 char name
[BCACHEDEVNAME_SIZE
];
267 struct gendisk
*disk
;
269 /* If nonzero, we're closing */
272 /* If nonzero, we're detaching/unregistering from cache set */
277 unsigned stripe_size
;
278 atomic_t
*stripe_sectors_dirty
;
279 unsigned long *full_dirty_stripes
;
281 unsigned long sectors_dirty_last
;
282 long sectors_dirty_derivative
;
284 mempool_t
*unaligned_bvec
;
285 struct bio_set
*bio_split
;
287 unsigned data_csum
:1;
289 int (*cache_miss
)(struct btree
*, struct search
*,
290 struct bio
*, unsigned);
291 int (*ioctl
) (struct bcache_device
*, fmode_t
, unsigned, unsigned long);
293 struct bio_split_pool bio_split_hook
;
297 /* Used to track sequential IO so it can be skipped */
298 struct hlist_node hash
;
299 struct list_head lru
;
301 unsigned long jiffies
;
307 struct list_head list
;
308 struct bcache_device disk
;
309 struct block_device
*bdev
;
313 struct bio_vec sb_bv
[1];
314 struct closure_with_waitlist sb_write
;
316 /* Refcount on the cache set. Always nonzero when we're caching. */
318 struct work_struct detach
;
321 * Device might not be running if it's dirty and the cache set hasn't
327 * Writes take a shared lock from start to finish; scanning for dirty
328 * data to refill the rb tree requires an exclusive lock.
330 struct rw_semaphore writeback_lock
;
333 * Nonzero, and writeback has a refcount (d->count), iff there is dirty
334 * data in the cache. Protected by writeback_lock; must have an
335 * shared lock to set and exclusive lock to clear.
339 struct bch_ratelimit writeback_rate
;
340 struct delayed_work writeback_rate_update
;
343 * Internal to the writeback code, so read_dirty() can keep track of
348 /* Limit number of writeback bios in flight */
349 struct semaphore in_flight
;
350 struct task_struct
*writeback_thread
;
352 struct keybuf writeback_keys
;
354 /* For tracking sequential IO */
355 #define RECENT_IO_BITS 7
356 #define RECENT_IO (1 << RECENT_IO_BITS)
357 struct io io
[RECENT_IO
];
358 struct hlist_head io_hash
[RECENT_IO
+ 1];
359 struct list_head io_lru
;
362 struct cache_accounting accounting
;
364 /* The rest of this all shows up in sysfs */
365 unsigned sequential_cutoff
;
370 unsigned partial_stripes_expensive
:1;
371 unsigned writeback_metadata
:1;
372 unsigned writeback_running
:1;
373 unsigned char writeback_percent
;
374 unsigned writeback_delay
;
376 int writeback_rate_change
;
377 int64_t writeback_rate_derivative
;
378 uint64_t writeback_rate_target
;
380 unsigned writeback_rate_update_seconds
;
381 unsigned writeback_rate_d_term
;
382 unsigned writeback_rate_p_term_inverse
;
383 unsigned writeback_rate_d_smooth
;
386 enum alloc_watermarks
{
395 struct cache_set
*set
;
398 struct bio_vec sb_bv
[1];
401 struct block_device
*bdev
;
403 unsigned watermark
[WATERMARK_MAX
];
405 struct task_struct
*alloc_thread
;
408 struct prio_set
*disk_buckets
;
411 * When allocating new buckets, prio_write() gets first dibs - since we
412 * may not be allocate at all without writing priorities and gens.
413 * prio_buckets[] contains the last buckets we wrote priorities to (so
414 * gc can mark them as metadata), prio_next[] contains the buckets
415 * allocated for the next prio write.
417 uint64_t *prio_buckets
;
418 uint64_t *prio_last_buckets
;
421 * free: Buckets that are ready to be used
423 * free_inc: Incoming buckets - these are buckets that currently have
424 * cached data in them, and we can't reuse them until after we write
425 * their new gen to disk. After prio_write() finishes writing the new
426 * gens/prios, they'll be moved to the free list (and possibly discarded
429 * unused: GC found nothing pointing into these buckets (possibly
430 * because all the data they contained was overwritten), so we only
431 * need to discard them before they can be moved to the free list.
433 DECLARE_FIFO(long, free
);
434 DECLARE_FIFO(long, free_inc
);
435 DECLARE_FIFO(long, unused
);
437 size_t fifo_last_bucket
;
439 /* Allocation stuff: */
440 struct bucket
*buckets
;
442 DECLARE_HEAP(struct bucket
*, heap
);
445 * max(gen - disk_gen) for all buckets. When it gets too big we have to
446 * call prio_write() to keep gens from wrapping.
448 uint8_t need_save_prio
;
449 unsigned gc_move_threshold
;
452 * If nonzero, we know we aren't going to find any buckets to invalidate
453 * until a gc finishes - otherwise we could pointlessly burn a ton of
456 unsigned invalidate_needs_gc
:1;
458 bool discard
; /* Get rid of? */
460 struct journal_device journal
;
462 /* The rest of this all shows up in sysfs */
463 #define IO_ERROR_SHIFT 20
467 atomic_long_t meta_sectors_written
;
468 atomic_long_t btree_sectors_written
;
469 atomic_long_t sectors_written
;
471 struct bio_split_pool bio_split_hook
;
479 uint64_t data
; /* sectors */
480 unsigned in_use
; /* percent */
484 * Flag bits, for how the cache set is shutting down, and what phase it's at:
486 * CACHE_SET_UNREGISTERING means we're not just shutting down, we're detaching
487 * all the backing devices first (their cached data gets invalidated, and they
488 * won't automatically reattach).
490 * CACHE_SET_STOPPING always gets set first when we're closing down a cache set;
491 * we'll continue to run normally for awhile with CACHE_SET_STOPPING set (i.e.
492 * flushing dirty data).
494 #define CACHE_SET_UNREGISTERING 0
495 #define CACHE_SET_STOPPING 1
500 struct list_head list
;
502 struct kobject internal
;
503 struct dentry
*debug
;
504 struct cache_accounting accounting
;
510 struct cache
*cache
[MAX_CACHES_PER_SET
];
511 struct cache
*cache_by_alloc
[MAX_CACHES_PER_SET
];
514 struct bcache_device
**devices
;
515 struct list_head cached_devs
;
516 uint64_t cached_dev_sectors
;
517 struct closure caching
;
519 struct closure_with_waitlist sb_write
;
523 struct bio_set
*bio_split
;
525 /* For the btree cache */
526 struct shrinker shrink
;
528 /* For the btree cache and anything allocation related */
529 struct mutex bucket_lock
;
531 /* log2(bucket_size), in sectors */
532 unsigned short bucket_bits
;
534 /* log2(block_size), in sectors */
535 unsigned short block_bits
;
538 * Default number of pages for a new btree node - may be less than a
541 unsigned btree_pages
;
544 * Lists of struct btrees; lru is the list for structs that have memory
545 * allocated for actual btree node, freed is for structs that do not.
547 * We never free a struct btree, except on shutdown - we just put it on
548 * the btree_cache_freed list and reuse it later. This simplifies the
549 * code, and it doesn't cost us much memory as the memory usage is
550 * dominated by buffers that hold the actual btree node data and those
551 * can be freed - and the number of struct btrees allocated is
552 * effectively bounded.
554 * btree_cache_freeable effectively is a small cache - we use it because
555 * high order page allocations can be rather expensive, and it's quite
556 * common to delete and allocate btree nodes in quick succession. It
557 * should never grow past ~2-3 nodes in practice.
559 struct list_head btree_cache
;
560 struct list_head btree_cache_freeable
;
561 struct list_head btree_cache_freed
;
563 /* Number of elements in btree_cache + btree_cache_freeable lists */
564 unsigned bucket_cache_used
;
567 * If we need to allocate memory for a new btree node and that
568 * allocation fails, we can cannibalize another node in the btree cache
569 * to satisfy the allocation. However, only one thread can be doing this
570 * at a time, for obvious reasons - try_harder and try_wait are
571 * basically a lock for this that we can wait on asynchronously. The
572 * btree_root() macro releases the lock when it returns.
574 struct task_struct
*try_harder
;
575 wait_queue_head_t try_wait
;
576 uint64_t try_harder_start
;
579 * When we free a btree node, we increment the gen of the bucket the
580 * node is in - but we can't rewrite the prios and gens until we
581 * finished whatever it is we were doing, otherwise after a crash the
582 * btree node would be freed but for say a split, we might not have the
583 * pointers to the new nodes inserted into the btree yet.
585 * This is a refcount that blocks prio_write() until the new keys are
588 atomic_t prio_blocked
;
589 wait_queue_head_t bucket_wait
;
592 * For any bio we don't skip we subtract the number of sectors from
593 * rescale; when it hits 0 we rescale all the bucket priorities.
597 * When we invalidate buckets, we use both the priority and the amount
598 * of good data to determine which buckets to reuse first - to weight
599 * those together consistently we keep track of the smallest nonzero
600 * priority of any bucket.
605 * max(gen - gc_gen) for all buckets. When it gets too big we have to gc
606 * to keep gens from wrapping around.
609 struct gc_stat gc_stats
;
612 struct task_struct
*gc_thread
;
613 /* Where in the btree gc currently is */
617 * The allocation code needs gc_mark in struct bucket to be correct, but
618 * it's not while a gc is in progress. Protected by bucket_lock.
622 /* Counts how many sectors bio_insert has added to the cache */
623 atomic_t sectors_to_gc
;
625 wait_queue_head_t moving_gc_wait
;
626 struct keybuf moving_gc_keys
;
627 /* Number of moving GC bios in flight */
628 struct semaphore moving_in_flight
;
632 #ifdef CONFIG_BCACHE_DEBUG
633 struct btree
*verify_data
;
634 struct mutex verify_lock
;
638 struct uuid_entry
*uuids
;
639 BKEY_PADDED(uuid_bucket
);
640 struct closure_with_waitlist uuid_write
;
643 * A btree node on disk could have too many bsets for an iterator to fit
644 * on the stack - have to dynamically allocate them
646 mempool_t
*fill_iter
;
649 * btree_sort() is a merge sort and requires temporary space - single
652 struct mutex sort_lock
;
654 unsigned sort_crit_factor
;
656 /* List of buckets we're currently writing data to */
657 struct list_head data_buckets
;
658 spinlock_t data_bucket_lock
;
660 struct journal journal
;
662 #define CONGESTED_MAX 1024
663 unsigned congested_last_us
;
666 /* The rest of this all shows up in sysfs */
667 unsigned congested_read_threshold_us
;
668 unsigned congested_write_threshold_us
;
670 struct time_stats sort_time
;
671 struct time_stats btree_gc_time
;
672 struct time_stats btree_split_time
;
673 struct time_stats btree_read_time
;
674 struct time_stats try_harder_time
;
676 atomic_long_t cache_read_races
;
677 atomic_long_t writeback_keys_done
;
678 atomic_long_t writeback_keys_failed
;
684 unsigned error_limit
;
685 unsigned error_decay
;
687 unsigned short journal_delay_ms
;
689 unsigned key_merging_disabled
:1;
690 unsigned expensive_debug_checks
:1;
691 unsigned gc_always_rewrite
:1;
692 unsigned shrinker_disabled
:1;
693 unsigned copy_gc_enabled
:1;
695 #define BUCKET_HASH_BITS 12
696 struct hlist_head bucket_hash
[1 << BUCKET_HASH_BITS
];
700 unsigned submit_time_us
;
705 * We only need pad = 3 here because we only ever carry around a
706 * single pointer - i.e. the pointer we're doing io to/from.
712 static inline unsigned local_clock_us(void)
714 return local_clock() >> 10;
717 #define BTREE_PRIO USHRT_MAX
718 #define INITIAL_PRIO 32768
720 #define btree_bytes(c) ((c)->btree_pages * PAGE_SIZE)
721 #define btree_blocks(b) \
722 ((unsigned) (KEY_SIZE(&b->key) >> (b)->c->block_bits))
724 #define btree_default_blocks(c) \
725 ((unsigned) ((PAGE_SECTORS * (c)->btree_pages) >> (c)->block_bits))
727 #define bucket_pages(c) ((c)->sb.bucket_size / PAGE_SECTORS)
728 #define bucket_bytes(c) ((c)->sb.bucket_size << 9)
729 #define block_bytes(c) ((c)->sb.block_size << 9)
731 #define __set_bytes(i, k) (sizeof(*(i)) + (k) * sizeof(uint64_t))
732 #define set_bytes(i) __set_bytes(i, i->keys)
734 #define __set_blocks(i, k, c) DIV_ROUND_UP(__set_bytes(i, k), block_bytes(c))
735 #define set_blocks(i, c) __set_blocks(i, (i)->keys, c)
737 #define node(i, j) ((struct bkey *) ((i)->d + (j)))
738 #define end(i) node(i, (i)->keys)
740 #define index(i, b) \
741 ((size_t) (((void *) i - (void *) (b)->sets[0].data) / \
744 #define btree_data_space(b) (PAGE_SIZE << (b)->page_order)
746 #define prios_per_bucket(c) \
747 ((bucket_bytes(c) - sizeof(struct prio_set)) / \
748 sizeof(struct bucket_disk))
749 #define prio_buckets(c) \
750 DIV_ROUND_UP((size_t) (c)->sb.nbuckets, prios_per_bucket(c))
752 static inline size_t sector_to_bucket(struct cache_set
*c
, sector_t s
)
754 return s
>> c
->bucket_bits
;
757 static inline sector_t
bucket_to_sector(struct cache_set
*c
, size_t b
)
759 return ((sector_t
) b
) << c
->bucket_bits
;
762 static inline sector_t
bucket_remainder(struct cache_set
*c
, sector_t s
)
764 return s
& (c
->sb
.bucket_size
- 1);
767 static inline struct cache
*PTR_CACHE(struct cache_set
*c
,
768 const struct bkey
*k
,
771 return c
->cache
[PTR_DEV(k
, ptr
)];
774 static inline size_t PTR_BUCKET_NR(struct cache_set
*c
,
775 const struct bkey
*k
,
778 return sector_to_bucket(c
, PTR_OFFSET(k
, ptr
));
781 static inline struct bucket
*PTR_BUCKET(struct cache_set
*c
,
782 const struct bkey
*k
,
785 return PTR_CACHE(c
, k
, ptr
)->buckets
+ PTR_BUCKET_NR(c
, k
, ptr
);
788 /* Btree key macros */
790 static inline void bkey_init(struct bkey
*k
)
796 * This is used for various on disk data structures - cache_sb, prio_set, bset,
797 * jset: The checksum is _always_ the first 8 bytes of these structs
799 #define csum_set(i) \
800 bch_crc64(((void *) (i)) + sizeof(uint64_t), \
801 ((void *) end(i)) - (((void *) (i)) + sizeof(uint64_t)))
803 /* Error handling macros */
805 #define btree_bug(b, ...) \
807 if (bch_cache_set_error((b)->c, __VA_ARGS__)) \
811 #define cache_bug(c, ...) \
813 if (bch_cache_set_error(c, __VA_ARGS__)) \
817 #define btree_bug_on(cond, b, ...) \
820 btree_bug(b, __VA_ARGS__); \
823 #define cache_bug_on(cond, c, ...) \
826 cache_bug(c, __VA_ARGS__); \
829 #define cache_set_err_on(cond, c, ...) \
832 bch_cache_set_error(c, __VA_ARGS__); \
837 #define for_each_cache(ca, cs, iter) \
838 for (iter = 0; ca = cs->cache[iter], iter < (cs)->sb.nr_in_set; iter++)
840 #define for_each_bucket(b, ca) \
841 for (b = (ca)->buckets + (ca)->sb.first_bucket; \
842 b < (ca)->buckets + (ca)->sb.nbuckets; b++)
844 static inline void cached_dev_put(struct cached_dev
*dc
)
846 if (atomic_dec_and_test(&dc
->count
))
847 schedule_work(&dc
->detach
);
850 static inline bool cached_dev_get(struct cached_dev
*dc
)
852 if (!atomic_inc_not_zero(&dc
->count
))
855 /* Paired with the mb in cached_dev_attach */
856 smp_mb__after_atomic_inc();
861 * bucket_gc_gen() returns the difference between the bucket's current gen and
862 * the oldest gen of any pointer into that bucket in the btree (last_gc).
864 * bucket_disk_gen() returns the difference between the current gen and the gen
865 * on disk; they're both used to make sure gens don't wrap around.
868 static inline uint8_t bucket_gc_gen(struct bucket
*b
)
870 return b
->gen
- b
->last_gc
;
873 static inline uint8_t bucket_disk_gen(struct bucket
*b
)
875 return b
->gen
- b
->disk_gen
;
878 #define BUCKET_GC_GEN_MAX 96U
879 #define BUCKET_DISK_GEN_MAX 64U
881 #define kobj_attribute_write(n, fn) \
882 static struct kobj_attribute ksysfs_##n = __ATTR(n, S_IWUSR, NULL, fn)
884 #define kobj_attribute_rw(n, show, store) \
885 static struct kobj_attribute ksysfs_##n = \
886 __ATTR(n, S_IWUSR|S_IRUSR, show, store)
888 static inline void wake_up_allocators(struct cache_set
*c
)
893 for_each_cache(ca
, c
, i
)
894 wake_up_process(ca
->alloc_thread
);
897 /* Forward declarations */
899 void bch_count_io_errors(struct cache
*, int, const char *);
900 void bch_bbio_count_io_errors(struct cache_set
*, struct bio
*,
902 void bch_bbio_endio(struct cache_set
*, struct bio
*, int, const char *);
903 void bch_bbio_free(struct bio
*, struct cache_set
*);
904 struct bio
*bch_bbio_alloc(struct cache_set
*);
906 struct bio
*bch_bio_split(struct bio
*, int, gfp_t
, struct bio_set
*);
907 void bch_generic_make_request(struct bio
*, struct bio_split_pool
*);
908 void __bch_submit_bbio(struct bio
*, struct cache_set
*);
909 void bch_submit_bbio(struct bio
*, struct cache_set
*, struct bkey
*, unsigned);
911 uint8_t bch_inc_gen(struct cache
*, struct bucket
*);
912 void bch_rescale_priorities(struct cache_set
*, int);
913 bool bch_bucket_add_unused(struct cache
*, struct bucket
*);
915 long bch_bucket_alloc(struct cache
*, unsigned, bool);
916 void bch_bucket_free(struct cache_set
*, struct bkey
*);
918 int __bch_bucket_alloc_set(struct cache_set
*, unsigned,
919 struct bkey
*, int, bool);
920 int bch_bucket_alloc_set(struct cache_set
*, unsigned,
921 struct bkey
*, int, bool);
922 bool bch_alloc_sectors(struct cache_set
*, struct bkey
*, unsigned,
923 unsigned, unsigned, bool);
926 bool bch_cache_set_error(struct cache_set
*, const char *, ...);
928 void bch_prio_write(struct cache
*);
929 void bch_write_bdev_super(struct cached_dev
*, struct closure
*);
931 extern struct workqueue_struct
*bcache_wq
;
932 extern const char * const bch_cache_modes
[];
933 extern struct mutex bch_register_lock
;
934 extern struct list_head bch_cache_sets
;
936 extern struct kobj_type bch_cached_dev_ktype
;
937 extern struct kobj_type bch_flash_dev_ktype
;
938 extern struct kobj_type bch_cache_set_ktype
;
939 extern struct kobj_type bch_cache_set_internal_ktype
;
940 extern struct kobj_type bch_cache_ktype
;
942 void bch_cached_dev_release(struct kobject
*);
943 void bch_flash_dev_release(struct kobject
*);
944 void bch_cache_set_release(struct kobject
*);
945 void bch_cache_release(struct kobject
*);
947 int bch_uuid_write(struct cache_set
*);
948 void bcache_write_super(struct cache_set
*);
950 int bch_flash_dev_create(struct cache_set
*c
, uint64_t size
);
952 int bch_cached_dev_attach(struct cached_dev
*, struct cache_set
*);
953 void bch_cached_dev_detach(struct cached_dev
*);
954 void bch_cached_dev_run(struct cached_dev
*);
955 void bcache_device_stop(struct bcache_device
*);
957 void bch_cache_set_unregister(struct cache_set
*);
958 void bch_cache_set_stop(struct cache_set
*);
960 struct cache_set
*bch_cache_set_alloc(struct cache_sb
*);
961 void bch_btree_cache_free(struct cache_set
*);
962 int bch_btree_cache_alloc(struct cache_set
*);
963 void bch_moving_init_cache_set(struct cache_set
*);
964 int bch_open_buckets_alloc(struct cache_set
*);
965 void bch_open_buckets_free(struct cache_set
*);
967 int bch_cache_allocator_start(struct cache
*ca
);
968 int bch_cache_allocator_init(struct cache
*ca
);
970 void bch_debug_exit(void);
971 int bch_debug_init(struct kobject
*);
972 void bch_request_exit(void);
973 int bch_request_init(void);
974 void bch_btree_exit(void);
975 int bch_btree_init(void);
977 #endif /* _BCACHE_H */