?  search:
summary | log | graphiclog1 | graphiclog2 | commit | commitdiff | tree | refs | edit | fork
blame | history | raw | HEAD
Merge tag 'gpio-v3.13-3' of git://git.kernel.org/pub/scm/linux/kernel/git/linusw...
[linux-2.6.git] / drivers / md / bcache / btree.c
blob5e2765aadce174e9b7cffd479667a2c48bcf500f
1 /*
2 * Copyright (C) 2010 Kent Overstreet <kent.overstreet@gmail.com>
3 *
4 * Uses a block device as cache for other block devices; optimized for SSDs.
5 * All allocation is done in buckets, which should match the erase block size
6 * of the device.
7 *
8 * Buckets containing cached data are kept on a heap sorted by priority;
9 * bucket priority is increased on cache hit, and periodically all the buckets
10 * on the heap have their priority scaled down. This currently is just used as
11 * an LRU but in the future should allow for more intelligent heuristics.
12 *
13 * Buckets have an 8 bit counter; freeing is accomplished by incrementing the
14 * counter. Garbage collection is used to remove stale pointers.
15 *
16 * Indexing is done via a btree; nodes are not necessarily fully sorted, rather
17 * as keys are inserted we only sort the pages that have not yet been written.
18 * When garbage collection is run, we resort the entire node.
19 *
20 * All configuration is done via sysfs; see Documentation/bcache.txt.
21 */
22
23 #include "bcache.h"
24 #include "btree.h"
25 #include "debug.h"
26 #include "writeback.h"
27
28 #include <linux/slab.h>
29 #include <linux/bitops.h>
30 #include <linux/freezer.h>
31 #include <linux/hash.h>
32 #include <linux/kthread.h>
33 #include <linux/prefetch.h>
34 #include <linux/random.h>
35 #include <linux/rcupdate.h>
36 #include <trace/events/bcache.h>
37
38 /*
39 * Todo:
40 * register_bcache: Return errors out to userspace correctly
41 *
42 * Writeback: don't undirty key until after a cache flush
43 *
44 * Create an iterator for key pointers
45 *
46 * On btree write error, mark bucket such that it won't be freed from the cache
47 *
48 * Journalling:
49 * Check for bad keys in replay
50 * Propagate barriers
51 * Refcount journal entries in journal_replay
52 *
53 * Garbage collection:
54 * Finish incremental gc
55 * Gc should free old UUIDs, data for invalid UUIDs
56 *
57 * Provide a way to list backing device UUIDs we have data cached for, and
58 * probably how long it's been since we've seen them, and a way to invalidate
59 * dirty data for devices that will never be attached again
60 *
61 * Keep 1 min/5 min/15 min statistics of how busy a block device has been, so
62 * that based on that and how much dirty data we have we can keep writeback
63 * from being starved
64 *
65 * Add a tracepoint or somesuch to watch for writeback starvation
66 *
67 * When btree depth > 1 and splitting an interior node, we have to make sure
68 * alloc_bucket() cannot fail. This should be true but is not completely
69 * obvious.
70 *
71 * Make sure all allocations get charged to the root cgroup
72 *
73 * Plugging?
74 *
75 * If data write is less than hard sector size of ssd, round up offset in open
76 * bucket to the next whole sector
77 *
78 * Also lookup by cgroup in get_open_bucket()
79 *
80 * Superblock needs to be fleshed out for multiple cache devices
81 *
82 * Add a sysfs tunable for the number of writeback IOs in flight
83 *
84 * Add a sysfs tunable for the number of open data buckets
85 *
86 * IO tracking: Can we track when one process is doing io on behalf of another?
87 * IO tracking: Don't use just an average, weigh more recent stuff higher
88 *
89 * Test module load/unload
90 */
91
92 enum {
93 BTREE_INSERT_STATUS_INSERT,
94 BTREE_INSERT_STATUS_BACK_MERGE,
95 BTREE_INSERT_STATUS_OVERWROTE,
96 BTREE_INSERT_STATUS_FRONT_MERGE,
97 };
98
99 #define MAX_NEED_GC 64
100 #define MAX_SAVE_PRIO 72
101
102 #define PTR_DIRTY_BIT (((uint64_t) 1 << 36))
103
104 #define PTR_HASH(c, k) \
105 (((k)->ptr[0] >> c->bucket_bits) | PTR_GEN(k, 0))
106
107 static struct workqueue_struct *btree_io_wq;
108
109 static inline bool should_split(struct btree *b)
110 {
111 struct bset *i = write_block(b);
112 return b->written >= btree_blocks(b) ||
113 (b->written + __set_blocks(i, i->keys + 15, b->c)
114 > btree_blocks(b));
115 }
116
117 #define insert_lock(s, b) ((b)->level <= (s)->lock)
118
119 /*
120 * These macros are for recursing down the btree - they handle the details of
121 * locking and looking up nodes in the cache for you. They're best treated as
122 * mere syntax when reading code that uses them.
123 *
124 * op->lock determines whether we take a read or a write lock at a given depth.
125 * If you've got a read lock and find that you need a write lock (i.e. you're
126 * going to have to split), set op->lock and return -EINTR; btree_root() will
127 * call you again and you'll have the correct lock.
128 */
129
130 /**
131 * btree - recurse down the btree on a specified key
132 * @fn: function to call, which will be passed the child node
133 * @key: key to recurse on
134 * @b: parent btree node
135 * @op: pointer to struct btree_op
136 */
137 #define btree(fn, key, b, op, ...) \
138 ({ \
139 int _r, l = (b)->level - 1; \
140 bool _w = l <= (op)->lock; \
141 struct btree *_child = bch_btree_node_get((b)->c, key, l, _w); \
142 if (!IS_ERR(_child)) { \
143 _child->parent = (b); \
144 _r = bch_btree_ ## fn(_child, op, ##__VA_ARGS__); \
145 rw_unlock(_w, _child); \
146 } else \
147 _r = PTR_ERR(_child); \
148 _r; \
149 })
150
151 /**
152 * btree_root - call a function on the root of the btree
153 * @fn: function to call, which will be passed the child node
154 * @c: cache set
155 * @op: pointer to struct btree_op
156 */
157 #define btree_root(fn, c, op, ...) \
158 ({ \
159 int _r = -EINTR; \
160 do { \
161 struct btree *_b = (c)->root; \
162 bool _w = insert_lock(op, _b); \
163 rw_lock(_w, _b, _b->level); \
164 if (_b == (c)->root && \
165 _w == insert_lock(op, _b)) { \
166 _b->parent = NULL; \
167 _r = bch_btree_ ## fn(_b, op, ##__VA_ARGS__); \
168 } \
169 rw_unlock(_w, _b); \
170 bch_cannibalize_unlock(c); \
171 if (_r == -ENOSPC) { \
172 wait_event((c)->try_wait, \
173 !(c)->try_harder); \
174 _r = -EINTR; \
175 } \
176 } while (_r == -EINTR); \
177 \
178 _r; \
179 })
180
181 /* Btree key manipulation */
182
183 void bkey_put(struct cache_set *c, struct bkey *k)
184 {
185 unsigned i;
186
187 for (i = 0; i < KEY_PTRS(k); i++)
188 if (ptr_available(c, k, i))
189 atomic_dec_bug(&PTR_BUCKET(c, k, i)->pin);
190 }
191
192 /* Btree IO */
193
194 static uint64_t btree_csum_set(struct btree *b, struct bset *i)
195 {
196 uint64_t crc = b->key.ptr[0];
197 void *data = (void *) i + 8, *end = end(i);
198
199 crc = bch_crc64_update(crc, data, end - data);
200 return crc ^ 0xffffffffffffffffULL;
201 }
202
203 static void bch_btree_node_read_done(struct btree *b)
204 {
205 const char *err = "bad btree header";
206 struct bset *i = b->sets[0].data;
207 struct btree_iter *iter;
208
209 iter = mempool_alloc(b->c->fill_iter, GFP_NOWAIT);
210 iter->size = b->c->sb.bucket_size / b->c->sb.block_size;
211 iter->used = 0;
212
213 #ifdef CONFIG_BCACHE_DEBUG
214 iter->b = b;
215 #endif
216
217 if (!i->seq)
218 goto err;
219
220 for (;
221 b->written < btree_blocks(b) && i->seq == b->sets[0].data->seq;
222 i = write_block(b)) {
223 err = "unsupported bset version";
224 if (i->version > BCACHE_BSET_VERSION)
225 goto err;
226
227 err = "bad btree header";
228 if (b->written + set_blocks(i, b->c) > btree_blocks(b))
229 goto err;
230
231 err = "bad magic";
232 if (i->magic != bset_magic(&b->c->sb))
233 goto err;
234
235 err = "bad checksum";
236 switch (i->version) {
237 case 0:
238 if (i->csum != csum_set(i))
239 goto err;
240 break;
241 case BCACHE_BSET_VERSION:
242 if (i->csum != btree_csum_set(b, i))
243 goto err;
244 break;
245 }
246
247 err = "empty set";
248 if (i != b->sets[0].data && !i->keys)
249 goto err;
250
251 bch_btree_iter_push(iter, i->start, end(i));
252
253 b->written += set_blocks(i, b->c);
254 }
255
256 err = "corrupted btree";
257 for (i = write_block(b);
258 index(i, b) < btree_blocks(b);
259 i = ((void *) i) + block_bytes(b->c))
260 if (i->seq == b->sets[0].data->seq)
261 goto err;
262
263 bch_btree_sort_and_fix_extents(b, iter);
264
265 i = b->sets[0].data;
266 err = "short btree key";
267 if (b->sets[0].size &&
268 bkey_cmp(&b->key, &b->sets[0].end) < 0)
269 goto err;
270
271 if (b->written < btree_blocks(b))
272 bch_bset_init_next(b);
273 out:
274 mempool_free(iter, b->c->fill_iter);
275 return;
276 err:
277 set_btree_node_io_error(b);
278 bch_cache_set_error(b->c, "%s at bucket %zu, block %zu, %u keys",
279 err, PTR_BUCKET_NR(b->c, &b->key, 0),
280 index(i, b), i->keys);
281 goto out;
282 }
283
284 static void btree_node_read_endio(struct bio *bio, int error)
285 {
286 struct closure *cl = bio->bi_private;
287 closure_put(cl);
288 }
289
290 void bch_btree_node_read(struct btree *b)
291 {
292 uint64_t start_time = local_clock();
293 struct closure cl;
294 struct bio *bio;
295
296 trace_bcache_btree_read(b);
297
298 closure_init_stack(&cl);
299
300 bio = bch_bbio_alloc(b->c);
301 bio->bi_rw = REQ_META|READ_SYNC;
302 bio->bi_size = KEY_SIZE(&b->key) << 9;
303 bio->bi_end_io = btree_node_read_endio;
304 bio->bi_private = &cl;
305
306 bch_bio_map(bio, b->sets[0].data);
307
308 bch_submit_bbio(bio, b->c, &b->key, 0);
309 closure_sync(&cl);
310
311 if (!test_bit(BIO_UPTODATE, &bio->bi_flags))
312 set_btree_node_io_error(b);
313
314 bch_bbio_free(bio, b->c);
315
316 if (btree_node_io_error(b))
317 goto err;
318
319 bch_btree_node_read_done(b);
320 bch_time_stats_update(&b->c->btree_read_time, start_time);
321
322 return;
323 err:
324 bch_cache_set_error(b->c, "io error reading bucket %zu",
325 PTR_BUCKET_NR(b->c, &b->key, 0));
326 }
327
328 static void btree_complete_write(struct btree *b, struct btree_write *w)
329 {
330 if (w->prio_blocked &&
331 !atomic_sub_return(w->prio_blocked, &b->c->prio_blocked))
332 wake_up_allocators(b->c);
333
334 if (w->journal) {
335 atomic_dec_bug(w->journal);
336 __closure_wake_up(&b->c->journal.wait);
337 }
338
339 w->prio_blocked = 0;
340 w->journal = NULL;
341 }
342
343 static void __btree_node_write_done(struct closure *cl)
344 {
345 struct btree *b = container_of(cl, struct btree, io.cl);
346 struct btree_write *w = btree_prev_write(b);
347
348 bch_bbio_free(b->bio, b->c);
349 b->bio = NULL;
350 btree_complete_write(b, w);
351
352 if (btree_node_dirty(b))
353 queue_delayed_work(btree_io_wq, &b->work,
354 msecs_to_jiffies(30000));
355
356 closure_return(cl);
357 }
358
359 static void btree_node_write_done(struct closure *cl)
360 {
361 struct btree *b = container_of(cl, struct btree, io.cl);
362 struct bio_vec *bv;
363 int n;
364
365 __bio_for_each_segment(bv, b->bio, n, 0)
366 __free_page(bv->bv_page);
367
368 __btree_node_write_done(cl);
369 }
370
371 static void btree_node_write_endio(struct bio *bio, int error)
372 {
373 struct closure *cl = bio->bi_private;
374 struct btree *b = container_of(cl, struct btree, io.cl);
375
376 if (error)
377 set_btree_node_io_error(b);
378
379 bch_bbio_count_io_errors(b->c, bio, error, "writing btree");
380 closure_put(cl);
381 }
382
383 static void do_btree_node_write(struct btree *b)
384 {
385 struct closure *cl = &b->io.cl;
386 struct bset *i = b->sets[b->nsets].data;
387 BKEY_PADDED(key) k;
388
389 i->version = BCACHE_BSET_VERSION;
390 i->csum = btree_csum_set(b, i);
391
392 BUG_ON(b->bio);
393 b->bio = bch_bbio_alloc(b->c);
394
395 b->bio->bi_end_io = btree_node_write_endio;
396 b->bio->bi_private = cl;
397 b->bio->bi_rw = REQ_META|WRITE_SYNC|REQ_FUA;
398 b->bio->bi_size = set_blocks(i, b->c) * block_bytes(b->c);
399 bch_bio_map(b->bio, i);
400
401 /*
402 * If we're appending to a leaf node, we don't technically need FUA -
403 * this write just needs to be persisted before the next journal write,
404 * which will be marked FLUSH|FUA.
405 *
406 * Similarly if we're writing a new btree root - the pointer is going to
407 * be in the next journal entry.
408 *
409 * But if we're writing a new btree node (that isn't a root) or
410 * appending to a non leaf btree node, we need either FUA or a flush
411 * when we write the parent with the new pointer. FUA is cheaper than a
412 * flush, and writes appending to leaf nodes aren't blocking anything so
413 * just make all btree node writes FUA to keep things sane.
414 */
415
416 bkey_copy(&k.key, &b->key);
417 SET_PTR_OFFSET(&k.key, 0, PTR_OFFSET(&k.key, 0) + bset_offset(b, i));
418
419 if (!bio_alloc_pages(b->bio, GFP_NOIO)) {
420 int j;
421 struct bio_vec *bv;
422 void *base = (void *) ((unsigned long) i & ~(PAGE_SIZE - 1));
423
424 bio_for_each_segment(bv, b->bio, j)
425 memcpy(page_address(bv->bv_page),
426 base + j * PAGE_SIZE, PAGE_SIZE);
427
428 bch_submit_bbio(b->bio, b->c, &k.key, 0);
429
430 continue_at(cl, btree_node_write_done, NULL);
431 } else {
432 b->bio->bi_vcnt = 0;
433 bch_bio_map(b->bio, i);
434
435 bch_submit_bbio(b->bio, b->c, &k.key, 0);
436
437 closure_sync(cl);
438 __btree_node_write_done(cl);
439 }
440 }
441
442 void bch_btree_node_write(struct btree *b, struct closure *parent)
443 {
444 struct bset *i = b->sets[b->nsets].data;
445
446 trace_bcache_btree_write(b);
447
448 BUG_ON(current->bio_list);
449 BUG_ON(b->written >= btree_blocks(b));
450 BUG_ON(b->written && !i->keys);
451 BUG_ON(b->sets->data->seq != i->seq);
452 bch_check_keys(b, "writing");
453
454 cancel_delayed_work(&b->work);
455
456 /* If caller isn't waiting for write, parent refcount is cache set */
457 closure_lock(&b->io, parent ?: &b->c->cl);
458
459 clear_bit(BTREE_NODE_dirty, &b->flags);
460 change_bit(BTREE_NODE_write_idx, &b->flags);
461
462 do_btree_node_write(b);
463
464 b->written += set_blocks(i, b->c);
465 atomic_long_add(set_blocks(i, b->c) * b->c->sb.block_size,
466 &PTR_CACHE(b->c, &b->key, 0)->btree_sectors_written);
467
468 bch_btree_sort_lazy(b);
469
470 if (b->written < btree_blocks(b))
471 bch_bset_init_next(b);
472 }
473
474 static void bch_btree_node_write_sync(struct btree *b)
475 {
476 struct closure cl;
477
478 closure_init_stack(&cl);
479 bch_btree_node_write(b, &cl);
480 closure_sync(&cl);
481 }
482
483 static void btree_node_write_work(struct work_struct *w)
484 {
485 struct btree *b = container_of(to_delayed_work(w), struct btree, work);
486
487 rw_lock(true, b, b->level);
488
489 if (btree_node_dirty(b))
490 bch_btree_node_write(b, NULL);
491 rw_unlock(true, b);
492 }
493
494 static void bch_btree_leaf_dirty(struct btree *b, atomic_t *journal_ref)
495 {
496 struct bset *i = b->sets[b->nsets].data;
497 struct btree_write *w = btree_current_write(b);
498
499 BUG_ON(!b->written);
500 BUG_ON(!i->keys);
501
502 if (!btree_node_dirty(b))
503 queue_delayed_work(btree_io_wq, &b->work, 30 * HZ);
504
505 set_btree_node_dirty(b);
506
507 if (journal_ref) {
508 if (w->journal &&
509 journal_pin_cmp(b->c, w->journal, journal_ref)) {
510 atomic_dec_bug(w->journal);
511 w->journal = NULL;
512 }
513
514 if (!w->journal) {
515 w->journal = journal_ref;
516 atomic_inc(w->journal);
517 }
518 }
519
520 /* Force write if set is too big */
521 if (set_bytes(i) > PAGE_SIZE - 48 &&
522 !current->bio_list)
523 bch_btree_node_write(b, NULL);
524 }
525
526 /*
527 * Btree in memory cache - allocation/freeing
528 * mca -> memory cache
529 */
530
531 static void mca_reinit(struct btree *b)
532 {
533 unsigned i;
534
535 b->flags = 0;
536 b->written = 0;
537 b->nsets = 0;
538
539 for (i = 0; i < MAX_BSETS; i++)
540 b->sets[i].size = 0;
541 /*
542 * Second loop starts at 1 because b->sets[0]->data is the memory we
543 * allocated
544 */
545 for (i = 1; i < MAX_BSETS; i++)
546 b->sets[i].data = NULL;
547 }
548
549 #define mca_reserve(c) (((c->root && c->root->level) \
550 ? c->root->level : 1) * 8 + 16)
551 #define mca_can_free(c) \
552 max_t(int, 0, c->bucket_cache_used - mca_reserve(c))
553
554 static void mca_data_free(struct btree *b)
555 {
556 struct bset_tree *t = b->sets;
557 BUG_ON(!closure_is_unlocked(&b->io.cl));
558
559 if (bset_prev_bytes(b) < PAGE_SIZE)
560 kfree(t->prev);
561 else
562 free_pages((unsigned long) t->prev,
563 get_order(bset_prev_bytes(b)));
564
565 if (bset_tree_bytes(b) < PAGE_SIZE)
566 kfree(t->tree);
567 else
568 free_pages((unsigned long) t->tree,
569 get_order(bset_tree_bytes(b)));
570
571 free_pages((unsigned long) t->data, b->page_order);
572
573 t->prev = NULL;
574 t->tree = NULL;
575 t->data = NULL;
576 list_move(&b->list, &b->c->btree_cache_freed);
577 b->c->bucket_cache_used--;
578 }
579
580 static void mca_bucket_free(struct btree *b)
581 {
582 BUG_ON(btree_node_dirty(b));
583
584 b->key.ptr[0] = 0;
585 hlist_del_init_rcu(&b->hash);
586 list_move(&b->list, &b->c->btree_cache_freeable);
587 }
588
589 static unsigned btree_order(struct bkey *k)
590 {
591 return ilog2(KEY_SIZE(k) / PAGE_SECTORS ?: 1);
592 }
593
594 static void mca_data_alloc(struct btree *b, struct bkey *k, gfp_t gfp)
595 {
596 struct bset_tree *t = b->sets;
597 BUG_ON(t->data);
598
599 b->page_order = max_t(unsigned,
600 ilog2(b->c->btree_pages),
601 btree_order(k));
602
603 t->data = (void *) __get_free_pages(gfp, b->page_order);
604 if (!t->data)
605 goto err;
606
607 t->tree = bset_tree_bytes(b) < PAGE_SIZE
608 ? kmalloc(bset_tree_bytes(b), gfp)
609 : (void *) __get_free_pages(gfp, get_order(bset_tree_bytes(b)));
610 if (!t->tree)
611 goto err;
612
613 t->prev = bset_prev_bytes(b) < PAGE_SIZE
614 ? kmalloc(bset_prev_bytes(b), gfp)
615 : (void *) __get_free_pages(gfp, get_order(bset_prev_bytes(b)));
616 if (!t->prev)
617 goto err;
618
619 list_move(&b->list, &b->c->btree_cache);
620 b->c->bucket_cache_used++;
621 return;
622 err:
623 mca_data_free(b);
624 }
625
626 static struct btree *mca_bucket_alloc(struct cache_set *c,
627 struct bkey *k, gfp_t gfp)
628 {
629 struct btree *b = kzalloc(sizeof(struct btree), gfp);
630 if (!b)
631 return NULL;
632
633 init_rwsem(&b->lock);
634 lockdep_set_novalidate_class(&b->lock);
635 INIT_LIST_HEAD(&b->list);
636 INIT_DELAYED_WORK(&b->work, btree_node_write_work);
637 b->c = c;
638 closure_init_unlocked(&b->io);
639
640 mca_data_alloc(b, k, gfp);
641 return b;
642 }
643
644 static int mca_reap(struct btree *b, unsigned min_order, bool flush)
645 {
646 struct closure cl;
647
648 closure_init_stack(&cl);
649 lockdep_assert_held(&b->c->bucket_lock);
650
651 if (!down_write_trylock(&b->lock))
652 return -ENOMEM;
653
654 BUG_ON(btree_node_dirty(b) && !b->sets[0].data);
655
656 if (b->page_order < min_order ||
657 (!flush &&
658 (btree_node_dirty(b) ||
659 atomic_read(&b->io.cl.remaining) != -1))) {
660 rw_unlock(true, b);
661 return -ENOMEM;
662 }
663
664 if (btree_node_dirty(b))
665 bch_btree_node_write_sync(b);
666
667 /* wait for any in flight btree write */
668 closure_wait_event(&b->io.wait, &cl,
669 atomic_read(&b->io.cl.remaining) == -1);
670
671 return 0;
672 }
673
674 static unsigned long bch_mca_scan(struct shrinker *shrink,
675 struct shrink_control *sc)
676 {
677 struct cache_set *c = container_of(shrink, struct cache_set, shrink);
678 struct btree *b, *t;
679 unsigned long i, nr = sc->nr_to_scan;
680 unsigned long freed = 0;
681
682 if (c->shrinker_disabled)
683 return SHRINK_STOP;
684
685 if (c->try_harder)
686 return SHRINK_STOP;
687
688 /* Return -1 if we can't do anything right now */
689 if (sc->gfp_mask & __GFP_IO)
690 mutex_lock(&c->bucket_lock);
691 else if (!mutex_trylock(&c->bucket_lock))
692 return -1;
693
694 /*
695 * It's _really_ critical that we don't free too many btree nodes - we
696 * have to always leave ourselves a reserve. The reserve is how we
697 * guarantee that allocating memory for a new btree node can always
698 * succeed, so that inserting keys into the btree can always succeed and
699 * IO can always make forward progress:
700 */
701 nr /= c->btree_pages;
702 nr = min_t(unsigned long, nr, mca_can_free(c));
703
704 i = 0;
705 list_for_each_entry_safe(b, t, &c->btree_cache_freeable, list) {
706 if (freed >= nr)
707 break;
708
709 if (++i > 3 &&
710 !mca_reap(b, 0, false)) {
711 mca_data_free(b);
712 rw_unlock(true, b);
713 freed++;
714 }
715 }
716
717 /*
718 * Can happen right when we first start up, before we've read in any
719 * btree nodes
720 */
721 if (list_empty(&c->btree_cache))
722 goto out;
723
724 for (i = 0; (nr--) && i < c->bucket_cache_used; i++) {
725 b = list_first_entry(&c->btree_cache, struct btree, list);
726 list_rotate_left(&c->btree_cache);
727
728 if (!b->accessed &&
729 !mca_reap(b, 0, false)) {
730 mca_bucket_free(b);
731 mca_data_free(b);
732 rw_unlock(true, b);
733 freed++;
734 } else
735 b->accessed = 0;
736 }
737 out:
738 mutex_unlock(&c->bucket_lock);
739 return freed;
740 }
741
742 static unsigned long bch_mca_count(struct shrinker *shrink,
743 struct shrink_control *sc)
744 {
745 struct cache_set *c = container_of(shrink, struct cache_set, shrink);
746
747 if (c->shrinker_disabled)
748 return 0;
749
750 if (c->try_harder)
751 return 0;
752
753 return mca_can_free(c) * c->btree_pages;
754 }
755
756 void bch_btree_cache_free(struct cache_set *c)
757 {
758 struct btree *b;
759 struct closure cl;
760 closure_init_stack(&cl);
761
762 if (c->shrink.list.next)
763 unregister_shrinker(&c->shrink);
764
765 mutex_lock(&c->bucket_lock);
766
767 #ifdef CONFIG_BCACHE_DEBUG
768 if (c->verify_data)
769 list_move(&c->verify_data->list, &c->btree_cache);
770 #endif
771
772 list_splice(&c->btree_cache_freeable,
773 &c->btree_cache);
774
775 while (!list_empty(&c->btree_cache)) {
776 b = list_first_entry(&c->btree_cache, struct btree, list);
777
778 if (btree_node_dirty(b))
779 btree_complete_write(b, btree_current_write(b));
780 clear_bit(BTREE_NODE_dirty, &b->flags);
781
782 mca_data_free(b);
783 }
784
785 while (!list_empty(&c->btree_cache_freed)) {
786 b = list_first_entry(&c->btree_cache_freed,
787 struct btree, list);
788 list_del(&b->list);
789 cancel_delayed_work_sync(&b->work);
790 kfree(b);
791 }
792
793 mutex_unlock(&c->bucket_lock);
794 }
795
796 int bch_btree_cache_alloc(struct cache_set *c)
797 {
798 unsigned i;
799
800 for (i = 0; i < mca_reserve(c); i++)
801 if (!mca_bucket_alloc(c, &ZERO_KEY, GFP_KERNEL))
802 return -ENOMEM;
803
804 list_splice_init(&c->btree_cache,
805 &c->btree_cache_freeable);
806
807 #ifdef CONFIG_BCACHE_DEBUG
808 mutex_init(&c->verify_lock);
809
810 c->verify_data = mca_bucket_alloc(c, &ZERO_KEY, GFP_KERNEL);
811
812 if (c->verify_data &&
813 c->verify_data->sets[0].data)
814 list_del_init(&c->verify_data->list);
815 else
816 c->verify_data = NULL;
817 #endif
818
819 c->shrink.count_objects = bch_mca_count;
820 c->shrink.scan_objects = bch_mca_scan;
821 c->shrink.seeks = 4;
822 c->shrink.batch = c->btree_pages * 2;
823 register_shrinker(&c->shrink);
824
825 return 0;
826 }
827
828 /* Btree in memory cache - hash table */
829
830 static struct hlist_head *mca_hash(struct cache_set *c, struct bkey *k)
831 {
832 return &c->bucket_hash[hash_32(PTR_HASH(c, k), BUCKET_HASH_BITS)];
833 }
834
835 static struct btree *mca_find(struct cache_set *c, struct bkey *k)
836 {
837 struct btree *b;
838
839 rcu_read_lock();
840 hlist_for_each_entry_rcu(b, mca_hash(c, k), hash)
841 if (PTR_HASH(c, &b->key) == PTR_HASH(c, k))
842 goto out;
843 b = NULL;
844 out:
845 rcu_read_unlock();
846 return b;
847 }
848
849 static struct btree *mca_cannibalize(struct cache_set *c, struct bkey *k)
850 {
851 struct btree *b;
852
853 trace_bcache_btree_cache_cannibalize(c);
854
855 if (!c->try_harder) {
856 c->try_harder = current;
857 c->try_harder_start = local_clock();
858 } else if (c->try_harder != current)
859 return ERR_PTR(-ENOSPC);
860
861 list_for_each_entry_reverse(b, &c->btree_cache, list)
862 if (!mca_reap(b, btree_order(k), false))
863 return b;
864
865 list_for_each_entry_reverse(b, &c->btree_cache, list)
866 if (!mca_reap(b, btree_order(k), true))
867 return b;
868
869 return ERR_PTR(-ENOMEM);
870 }
871
872 /*
873 * We can only have one thread cannibalizing other cached btree nodes at a time,
874 * or we'll deadlock. We use an open coded mutex to ensure that, which a
875 * cannibalize_bucket() will take. This means every time we unlock the root of
876 * the btree, we need to release this lock if we have it held.
877 */
878 static void bch_cannibalize_unlock(struct cache_set *c)
879 {
880 if (c->try_harder == current) {
881 bch_time_stats_update(&c->try_harder_time, c->try_harder_start);
882 c->try_harder = NULL;
883 wake_up(&c->try_wait);
884 }
885 }
886
887 static struct btree *mca_alloc(struct cache_set *c, struct bkey *k, int level)
888 {
889 struct btree *b;
890
891 BUG_ON(current->bio_list);
892
893 lockdep_assert_held(&c->bucket_lock);
894
895 if (mca_find(c, k))
896 return NULL;
897
898 /* btree_free() doesn't free memory; it sticks the node on the end of
899 * the list. Check if there's any freed nodes there:
900 */
901 list_for_each_entry(b, &c->btree_cache_freeable, list)
902 if (!mca_reap(b, btree_order(k), false))
903 goto out;
904
905 /* We never free struct btree itself, just the memory that holds the on
906 * disk node. Check the freed list before allocating a new one:
907 */
908 list_for_each_entry(b, &c->btree_cache_freed, list)
909 if (!mca_reap(b, 0, false)) {
910 mca_data_alloc(b, k, __GFP_NOWARN|GFP_NOIO);
911 if (!b->sets[0].data)
912 goto err;
913 else
914 goto out;
915 }
916
917 b = mca_bucket_alloc(c, k, __GFP_NOWARN|GFP_NOIO);
918 if (!b)
919 goto err;
920
921 BUG_ON(!down_write_trylock(&b->lock));
922 if (!b->sets->data)
923 goto err;
924 out:
925 BUG_ON(!closure_is_unlocked(&b->io.cl));
926
927 bkey_copy(&b->key, k);
928 list_move(&b->list, &c->btree_cache);
929 hlist_del_init_rcu(&b->hash);
930 hlist_add_head_rcu(&b->hash, mca_hash(c, k));
931
932 lock_set_subclass(&b->lock.dep_map, level + 1, _THIS_IP_);
933 b->level = level;
934 b->parent = (void *) ~0UL;
935
936 mca_reinit(b);
937
938 return b;
939 err:
940 if (b)
941 rw_unlock(true, b);
942
943 b = mca_cannibalize(c, k);
944 if (!IS_ERR(b))
945 goto out;
946
947 return b;
948 }
949
950 /**
951 * bch_btree_node_get - find a btree node in the cache and lock it, reading it
952 * in from disk if necessary.
953 *
954 * If IO is necessary and running under generic_make_request, returns -EAGAIN.
955 *
956 * The btree node will have either a read or a write lock held, depending on
957 * level and op->lock.
958 */
959 struct btree *bch_btree_node_get(struct cache_set *c, struct bkey *k,
960 int level, bool write)
961 {
962 int i = 0;
963 struct btree *b;
964
965 BUG_ON(level < 0);
966 retry:
967 b = mca_find(c, k);
968
969 if (!b) {
970 if (current->bio_list)
971 return ERR_PTR(-EAGAIN);
972
973 mutex_lock(&c->bucket_lock);
974 b = mca_alloc(c, k, level);
975 mutex_unlock(&c->bucket_lock);
976
977 if (!b)
978 goto retry;
979 if (IS_ERR(b))
980 return b;
981
982 bch_btree_node_read(b);
983
984 if (!write)
985 downgrade_write(&b->lock);
986 } else {
987 rw_lock(write, b, level);
988 if (PTR_HASH(c, &b->key) != PTR_HASH(c, k)) {
989 rw_unlock(write, b);
990 goto retry;
991 }
992 BUG_ON(b->level != level);
993 }
994
995 b->accessed = 1;
996
997 for (; i <= b->nsets && b->sets[i].size; i++) {
998 prefetch(b->sets[i].tree);
999 prefetch(b->sets[i].data);
1000 }
1001
1002 for (; i <= b->nsets; i++)
1003 prefetch(b->sets[i].data);
1004
1005 if (btree_node_io_error(b)) {
1006 rw_unlock(write, b);
1007 return ERR_PTR(-EIO);
1008 }
1009
1010 BUG_ON(!b->written);
1011
1012 return b;
1013 }
1014
1015 static void btree_node_prefetch(struct cache_set *c, struct bkey *k, int level)
1016 {
1017 struct btree *b;
1018
1019 mutex_lock(&c->bucket_lock);
1020 b = mca_alloc(c, k, level);
1021 mutex_unlock(&c->bucket_lock);
1022
1023 if (!IS_ERR_OR_NULL(b)) {
1024 bch_btree_node_read(b);
1025 rw_unlock(true, b);
1026 }
1027 }
1028
1029 /* Btree alloc */
1030
1031 static void btree_node_free(struct btree *b)
1032 {
1033 unsigned i;
1034
1035 trace_bcache_btree_node_free(b);
1036
1037 BUG_ON(b == b->c->root);
1038
1039 if (btree_node_dirty(b))
1040 btree_complete_write(b, btree_current_write(b));
1041 clear_bit(BTREE_NODE_dirty, &b->flags);
1042
1043 cancel_delayed_work(&b->work);
1044
1045 mutex_lock(&b->c->bucket_lock);
1046
1047 for (i = 0; i < KEY_PTRS(&b->key); i++) {
1048 BUG_ON(atomic_read(&PTR_BUCKET(b->c, &b->key, i)->pin));
1049
1050 bch_inc_gen(PTR_CACHE(b->c, &b->key, i),
1051 PTR_BUCKET(b->c, &b->key, i));
1052 }
1053
1054 bch_bucket_free(b->c, &b->key);
1055 mca_bucket_free(b);
1056 mutex_unlock(&b->c->bucket_lock);
1057 }
1058
1059 struct btree *bch_btree_node_alloc(struct cache_set *c, int level, bool wait)
1060 {
1061 BKEY_PADDED(key) k;
1062 struct btree *b = ERR_PTR(-EAGAIN);
1063
1064 mutex_lock(&c->bucket_lock);
1065 retry:
1066 if (__bch_bucket_alloc_set(c, WATERMARK_METADATA, &k.key, 1, wait))
1067 goto err;
1068
1069 bkey_put(c, &k.key);
1070 SET_KEY_SIZE(&k.key, c->btree_pages * PAGE_SECTORS);
1071
1072 b = mca_alloc(c, &k.key, level);
1073 if (IS_ERR(b))
1074 goto err_free;
1075
1076 if (!b) {
1077 cache_bug(c,
1078 "Tried to allocate bucket that was in btree cache");
1079 goto retry;
1080 }
1081
1082 b->accessed = 1;
1083 bch_bset_init_next(b);
1084
1085 mutex_unlock(&c->bucket_lock);
1086
1087 trace_bcache_btree_node_alloc(b);
1088 return b;
1089 err_free:
1090 bch_bucket_free(c, &k.key);
1091 err:
1092 mutex_unlock(&c->bucket_lock);
1093
1094 trace_bcache_btree_node_alloc_fail(b);
1095 return b;
1096 }
1097
1098 static struct btree *btree_node_alloc_replacement(struct btree *b, bool wait)
1099 {
1100 struct btree *n = bch_btree_node_alloc(b->c, b->level, wait);
1101 if (!IS_ERR_OR_NULL(n))
1102 bch_btree_sort_into(b, n);
1103
1104 return n;
1105 }
1106
1107 static void make_btree_freeing_key(struct btree *b, struct bkey *k)
1108 {
1109 unsigned i;
1110
1111 bkey_copy(k, &b->key);
1112 bkey_copy_key(k, &ZERO_KEY);
1113
1114 for (i = 0; i < KEY_PTRS(k); i++) {
1115 uint8_t g = PTR_BUCKET(b->c, k, i)->gen + 1;
1116
1117 SET_PTR_GEN(k, i, g);
1118 }
1119
1120 atomic_inc(&b->c->prio_blocked);
1121 }
1122
1123 /* Garbage collection */
1124
1125 uint8_t __bch_btree_mark_key(struct cache_set *c, int level, struct bkey *k)
1126 {
1127 uint8_t stale = 0;
1128 unsigned i;
1129 struct bucket *g;
1130
1131 /*
1132 * ptr_invalid() can't return true for the keys that mark btree nodes as
1133 * freed, but since ptr_bad() returns true we'll never actually use them
1134 * for anything and thus we don't want mark their pointers here
1135 */
1136 if (!bkey_cmp(k, &ZERO_KEY))
1137 return stale;
1138
1139 for (i = 0; i < KEY_PTRS(k); i++) {
1140 if (!ptr_available(c, k, i))
1141 continue;
1142
1143 g = PTR_BUCKET(c, k, i);
1144
1145 if (gen_after(g->gc_gen, PTR_GEN(k, i)))
1146 g->gc_gen = PTR_GEN(k, i);
1147
1148 if (ptr_stale(c, k, i)) {
1149 stale = max(stale, ptr_stale(c, k, i));
1150 continue;
1151 }
1152
1153 cache_bug_on(GC_MARK(g) &&
1154 (GC_MARK(g) == GC_MARK_METADATA) != (level != 0),
1155 c, "inconsistent ptrs: mark = %llu, level = %i",
1156 GC_MARK(g), level);
1157
1158 if (level)
1159 SET_GC_MARK(g, GC_MARK_METADATA);
1160 else if (KEY_DIRTY(k))
1161 SET_GC_MARK(g, GC_MARK_DIRTY);
1162
1163 /* guard against overflow */
1164 SET_GC_SECTORS_USED(g, min_t(unsigned,
1165 GC_SECTORS_USED(g) + KEY_SIZE(k),
1166 (1 << 14) - 1));
1167
1168 BUG_ON(!GC_SECTORS_USED(g));
1169 }
1170
1171 return stale;
1172 }
1173
1174 #define btree_mark_key(b, k) __bch_btree_mark_key(b->c, b->level, k)
1175
1176 static bool btree_gc_mark_node(struct btree *b, struct gc_stat *gc)
1177 {
1178 uint8_t stale = 0;
1179 unsigned keys = 0, good_keys = 0;
1180 struct bkey *k;
1181 struct btree_iter iter;
1182 struct bset_tree *t;
1183
1184 gc->nodes++;
1185
1186 for_each_key_filter(b, k, &iter, bch_ptr_invalid) {
1187 stale = max(stale, btree_mark_key(b, k));
1188 keys++;
1189
1190 if (bch_ptr_bad(b, k))
1191 continue;
1192
1193 gc->key_bytes += bkey_u64s(k);
1194 gc->nkeys++;
1195 good_keys++;
1196
1197 gc->data += KEY_SIZE(k);
1198 }
1199
1200 for (t = b->sets; t <= &b->sets[b->nsets]; t++)
1201 btree_bug_on(t->size &&
1202 bset_written(b, t) &&
1203 bkey_cmp(&b->key, &t->end) < 0,
1204 b, "found short btree key in gc");
1205
1206 if (b->c->gc_always_rewrite)
1207 return true;
1208
1209 if (stale > 10)
1210 return true;
1211
1212 if ((keys - good_keys) * 2 > keys)
1213 return true;
1214
1215 return false;
1216 }
1217
1218 #define GC_MERGE_NODES 4U
1219
1220 struct gc_merge_info {
1221 struct btree *b;
1222 unsigned keys;
1223 };
1224
1225 static int bch_btree_insert_node(struct btree *, struct btree_op *,
1226 struct keylist *, atomic_t *, struct bkey *);
1227
1228 static int btree_gc_coalesce(struct btree *b, struct btree_op *op,
1229 struct keylist *keylist, struct gc_stat *gc,
1230 struct gc_merge_info *r)
1231 {
1232 unsigned i, nodes = 0, keys = 0, blocks;
1233 struct btree *new_nodes[GC_MERGE_NODES];
1234 struct closure cl;
1235 struct bkey *k;
1236
1237 memset(new_nodes, 0, sizeof(new_nodes));
1238 closure_init_stack(&cl);
1239
1240 while (nodes < GC_MERGE_NODES && !IS_ERR_OR_NULL(r[nodes].b))
1241 keys += r[nodes++].keys;
1242
1243 blocks = btree_default_blocks(b->c) * 2 / 3;
1244
1245 if (nodes < 2 ||
1246 __set_blocks(b->sets[0].data, keys, b->c) > blocks * (nodes - 1))
1247 return 0;
1248
1249 for (i = 0; i < nodes; i++) {
1250 new_nodes[i] = btree_node_alloc_replacement(r[i].b, false);
1251 if (IS_ERR_OR_NULL(new_nodes[i]))
1252 goto out_nocoalesce;
1253 }
1254
1255 for (i = nodes - 1; i > 0; --i) {
1256 struct bset *n1 = new_nodes[i]->sets->data;
1257 struct bset *n2 = new_nodes[i - 1]->sets->data;
1258 struct bkey *k, *last = NULL;
1259
1260 keys = 0;
1261
1262 if (i > 1) {
1263 for (k = n2->start;
1264 k < end(n2);
1265 k = bkey_next(k)) {
1266 if (__set_blocks(n1, n1->keys + keys +
1267 bkey_u64s(k), b->c) > blocks)
1268 break;
1269
1270 last = k;
1271 keys += bkey_u64s(k);
1272 }
1273 } else {
1274 /*
1275 * Last node we're not getting rid of - we're getting
1276 * rid of the node at r[0]. Have to try and fit all of
1277 * the remaining keys into this node; we can't ensure
1278 * they will always fit due to rounding and variable
1279 * length keys (shouldn't be possible in practice,
1280 * though)
1281 */
1282 if (__set_blocks(n1, n1->keys + n2->keys,
1283 b->c) > btree_blocks(new_nodes[i]))
1284 goto out_nocoalesce;
1285
1286 keys = n2->keys;
1287 /* Take the key of the node we're getting rid of */
1288 last = &r->b->key;
1289 }
1290
1291 BUG_ON(__set_blocks(n1, n1->keys + keys,
1292 b->c) > btree_blocks(new_nodes[i]));
1293
1294 if (last)
1295 bkey_copy_key(&new_nodes[i]->key, last);
1296
1297 memcpy(end(n1),
1298 n2->start,
1299 (void *) node(n2, keys) - (void *) n2->start);
1300
1301 n1->keys += keys;
1302 r[i].keys = n1->keys;
1303
1304 memmove(n2->start,
1305 node(n2, keys),
1306 (void *) end(n2) - (void *) node(n2, keys));
1307
1308 n2->keys -= keys;
1309
1310 if (bch_keylist_realloc(keylist,
1311 KEY_PTRS(&new_nodes[i]->key), b->c))
1312 goto out_nocoalesce;
1313
1314 bch_btree_node_write(new_nodes[i], &cl);
1315 bch_keylist_add(keylist, &new_nodes[i]->key);
1316 }
1317
1318 for (i = 0; i < nodes; i++) {
1319 if (bch_keylist_realloc(keylist, KEY_PTRS(&r[i].b->key), b->c))
1320 goto out_nocoalesce;
1321
1322 make_btree_freeing_key(r[i].b, keylist->top);
1323 bch_keylist_push(keylist);
1324 }
1325
1326 /* We emptied out this node */
1327 BUG_ON(new_nodes[0]->sets->data->keys);
1328 btree_node_free(new_nodes[0]);
1329 rw_unlock(true, new_nodes[0]);
1330
1331 closure_sync(&cl);
1332
1333 for (i = 0; i < nodes; i++) {
1334 btree_node_free(r[i].b);
1335 rw_unlock(true, r[i].b);
1336
1337 r[i].b = new_nodes[i];
1338 }
1339
1340 bch_btree_insert_node(b, op, keylist, NULL, NULL);
1341 BUG_ON(!bch_keylist_empty(keylist));
1342
1343 memmove(r, r + 1, sizeof(r[0]) * (nodes - 1));
1344 r[nodes - 1].b = ERR_PTR(-EINTR);
1345
1346 trace_bcache_btree_gc_coalesce(nodes);
1347 gc->nodes--;
1348
1349 /* Invalidated our iterator */
1350 return -EINTR;
1351
1352 out_nocoalesce:
1353 closure_sync(&cl);
1354
1355 while ((k = bch_keylist_pop(keylist)))
1356 if (!bkey_cmp(k, &ZERO_KEY))
1357 atomic_dec(&b->c->prio_blocked);
1358
1359 for (i = 0; i < nodes; i++)
1360 if (!IS_ERR_OR_NULL(new_nodes[i])) {
1361 btree_node_free(new_nodes[i]);
1362 rw_unlock(true, new_nodes[i]);
1363 }
1364 return 0;
1365 }
1366
1367 static unsigned btree_gc_count_keys(struct btree *b)
1368 {
1369 struct bkey *k;
1370 struct btree_iter iter;
1371 unsigned ret = 0;
1372
1373 for_each_key_filter(b, k, &iter, bch_ptr_bad)
1374 ret += bkey_u64s(k);
1375
1376 return ret;
1377 }
1378
1379 static int btree_gc_recurse(struct btree *b, struct btree_op *op,
1380 struct closure *writes, struct gc_stat *gc)
1381 {
1382 unsigned i;
1383 int ret = 0;
1384 bool should_rewrite;
1385 struct btree *n;
1386 struct bkey *k;
1387 struct keylist keys;
1388 struct btree_iter iter;
1389 struct gc_merge_info r[GC_MERGE_NODES];
1390 struct gc_merge_info *last = r + GC_MERGE_NODES - 1;
1391
1392 bch_keylist_init(&keys);
1393 bch_btree_iter_init(b, &iter, &b->c->gc_done);
1394
1395 for (i = 0; i < GC_MERGE_NODES; i++)
1396 r[i].b = ERR_PTR(-EINTR);
1397
1398 while (1) {
1399 k = bch_btree_iter_next_filter(&iter, b, bch_ptr_bad);
1400 if (k) {
1401 r->b = bch_btree_node_get(b->c, k, b->level - 1, true);
1402 if (IS_ERR(r->b)) {
1403 ret = PTR_ERR(r->b);
1404 break;
1405 }
1406
1407 r->keys = btree_gc_count_keys(r->b);
1408
1409 ret = btree_gc_coalesce(b, op, &keys, gc, r);
1410 if (ret)
1411 break;
1412 }
1413
1414 if (!last->b)
1415 break;
1416
1417 if (!IS_ERR(last->b)) {
1418 should_rewrite = btree_gc_mark_node(last->b, gc);
1419 if (should_rewrite) {
1420 n = btree_node_alloc_replacement(last->b,
1421 false);
1422
1423 if (!IS_ERR_OR_NULL(n)) {
1424 bch_btree_node_write_sync(n);
1425 bch_keylist_add(&keys, &n->key);
1426
1427 make_btree_freeing_key(last->b,
1428 keys.top);
1429 bch_keylist_push(&keys);
1430
1431 btree_node_free(last->b);
1432
1433 bch_btree_insert_node(b, op, &keys,
1434 NULL, NULL);
1435 BUG_ON(!bch_keylist_empty(&keys));
1436
1437 rw_unlock(true, last->b);
1438 last->b = n;
1439
1440 /* Invalidated our iterator */
1441 ret = -EINTR;
1442 break;
1443 }
1444 }
1445
1446 if (last->b->level) {
1447 ret = btree_gc_recurse(last->b, op, writes, gc);
1448 if (ret)
1449 break;
1450 }
1451
1452 bkey_copy_key(&b->c->gc_done, &last->b->key);
1453
1454 /*
1455 * Must flush leaf nodes before gc ends, since replace
1456 * operations aren't journalled
1457 */
1458 if (btree_node_dirty(last->b))
1459 bch_btree_node_write(last->b, writes);
1460 rw_unlock(true, last->b);
1461 }
1462
1463 memmove(r + 1, r, sizeof(r[0]) * (GC_MERGE_NODES - 1));
1464 r->b = NULL;
1465
1466 if (need_resched()) {
1467 ret = -EAGAIN;
1468 break;
1469 }
1470 }
1471
1472 for (i = 0; i < GC_MERGE_NODES; i++)
1473 if (!IS_ERR_OR_NULL(r[i].b)) {
1474 if (btree_node_dirty(r[i].b))
1475 bch_btree_node_write(r[i].b, writes);
1476 rw_unlock(true, r[i].b);
1477 }
1478
1479 bch_keylist_free(&keys);
1480
1481 return ret;
1482 }
1483
1484 static int bch_btree_gc_root(struct btree *b, struct btree_op *op,
1485 struct closure *writes, struct gc_stat *gc)
1486 {
1487 struct btree *n = NULL;
1488 int ret = 0;
1489 bool should_rewrite;
1490
1491 should_rewrite = btree_gc_mark_node(b, gc);
1492 if (should_rewrite) {
1493 n = btree_node_alloc_replacement(b, false);
1494
1495 if (!IS_ERR_OR_NULL(n)) {
1496 bch_btree_node_write_sync(n);
1497 bch_btree_set_root(n);
1498 btree_node_free(b);
1499 rw_unlock(true, n);
1500
1501 return -EINTR;
1502 }
1503 }
1504
1505 if (b->level) {
1506 ret = btree_gc_recurse(b, op, writes, gc);
1507 if (ret)
1508 return ret;
1509 }
1510
1511 bkey_copy_key(&b->c->gc_done, &b->key);
1512
1513 return ret;
1514 }
1515
1516 static void btree_gc_start(struct cache_set *c)
1517 {
1518 struct cache *ca;
1519 struct bucket *b;
1520 unsigned i;
1521
1522 if (!c->gc_mark_valid)
1523 return;
1524
1525 mutex_lock(&c->bucket_lock);
1526
1527 c->gc_mark_valid = 0;
1528 c->gc_done = ZERO_KEY;
1529
1530 for_each_cache(ca, c, i)
1531 for_each_bucket(b, ca) {
1532 b->gc_gen = b->gen;
1533 if (!atomic_read(&b->pin)) {
1534 SET_GC_MARK(b, GC_MARK_RECLAIMABLE);
1535 SET_GC_SECTORS_USED(b, 0);
1536 }
1537 }
1538
1539 mutex_unlock(&c->bucket_lock);
1540 }
1541
1542 size_t bch_btree_gc_finish(struct cache_set *c)
1543 {
1544 size_t available = 0;
1545 struct bucket *b;
1546 struct cache *ca;
1547 unsigned i;
1548
1549 mutex_lock(&c->bucket_lock);
1550
1551 set_gc_sectors(c);
1552 c->gc_mark_valid = 1;
1553 c->need_gc = 0;
1554
1555 if (c->root)
1556 for (i = 0; i < KEY_PTRS(&c->root->key); i++)
1557 SET_GC_MARK(PTR_BUCKET(c, &c->root->key, i),
1558 GC_MARK_METADATA);
1559
1560 for (i = 0; i < KEY_PTRS(&c->uuid_bucket); i++)
1561 SET_GC_MARK(PTR_BUCKET(c, &c->uuid_bucket, i),
1562 GC_MARK_METADATA);
1563
1564 for_each_cache(ca, c, i) {
1565 uint64_t *i;
1566
1567 ca->invalidate_needs_gc = 0;
1568
1569 for (i = ca->sb.d; i < ca->sb.d + ca->sb.keys; i++)
1570 SET_GC_MARK(ca->buckets + *i, GC_MARK_METADATA);
1571
1572 for (i = ca->prio_buckets;
1573 i < ca->prio_buckets + prio_buckets(ca) * 2; i++)
1574 SET_GC_MARK(ca->buckets + *i, GC_MARK_METADATA);
1575
1576 for_each_bucket(b, ca) {
1577 b->last_gc = b->gc_gen;
1578 c->need_gc = max(c->need_gc, bucket_gc_gen(b));
1579
1580 if (!atomic_read(&b->pin) &&
1581 GC_MARK(b) == GC_MARK_RECLAIMABLE) {
1582 available++;
1583 if (!GC_SECTORS_USED(b))
1584 bch_bucket_add_unused(ca, b);
1585 }
1586 }
1587 }
1588
1589 mutex_unlock(&c->bucket_lock);
1590 return available;
1591 }
1592
1593 static void bch_btree_gc(struct cache_set *c)
1594 {
1595 int ret;
1596 unsigned long available;
1597 struct gc_stat stats;
1598 struct closure writes;
1599 struct btree_op op;
1600 uint64_t start_time = local_clock();
1601
1602 trace_bcache_gc_start(c);
1603
1604 memset(&stats, 0, sizeof(struct gc_stat));
1605 closure_init_stack(&writes);
1606 bch_btree_op_init(&op, SHRT_MAX);
1607
1608 btree_gc_start(c);
1609
1610 do {
1611 ret = btree_root(gc_root, c, &op, &writes, &stats);
1612 closure_sync(&writes);
1613
1614 if (ret && ret != -EAGAIN)
1615 pr_warn("gc failed!");
1616 } while (ret);
1617
1618 available = bch_btree_gc_finish(c);
1619 wake_up_allocators(c);
1620
1621 bch_time_stats_update(&c->btree_gc_time, start_time);
1622
1623 stats.key_bytes *= sizeof(uint64_t);
1624 stats.data <<= 9;
1625 stats.in_use = (c->nbuckets - available) * 100 / c->nbuckets;
1626 memcpy(&c->gc_stats, &stats, sizeof(struct gc_stat));
1627
1628 trace_bcache_gc_end(c);
1629
1630 bch_moving_gc(c);
1631 }
1632
1633 static int bch_gc_thread(void *arg)
1634 {
1635 struct cache_set *c = arg;
1636 struct cache *ca;
1637 unsigned i;
1638
1639 while (1) {
1640 again:
1641 bch_btree_gc(c);
1642
1643 set_current_state(TASK_INTERRUPTIBLE);
1644 if (kthread_should_stop())
1645 break;
1646
1647 mutex_lock(&c->bucket_lock);
1648
1649 for_each_cache(ca, c, i)
1650 if (ca->invalidate_needs_gc) {
1651 mutex_unlock(&c->bucket_lock);
1652 set_current_state(TASK_RUNNING);
1653 goto again;
1654 }
1655
1656 mutex_unlock(&c->bucket_lock);
1657
1658 try_to_freeze();
1659 schedule();
1660 }
1661
1662 return 0;
1663 }
1664
1665 int bch_gc_thread_start(struct cache_set *c)
1666 {
1667 c->gc_thread = kthread_create(bch_gc_thread, c, "bcache_gc");
1668 if (IS_ERR(c->gc_thread))
1669 return PTR_ERR(c->gc_thread);
1670
1671 set_task_state(c->gc_thread, TASK_INTERRUPTIBLE);
1672 return 0;
1673 }
1674
1675 /* Initial partial gc */
1676
1677 static int bch_btree_check_recurse(struct btree *b, struct btree_op *op,
1678 unsigned long **seen)
1679 {
1680 int ret = 0;
1681 unsigned i;
1682 struct bkey *k, *p = NULL;
1683 struct bucket *g;
1684 struct btree_iter iter;
1685
1686 for_each_key_filter(b, k, &iter, bch_ptr_invalid) {
1687 for (i = 0; i < KEY_PTRS(k); i++) {
1688 if (!ptr_available(b->c, k, i))
1689 continue;
1690
1691 g = PTR_BUCKET(b->c, k, i);
1692
1693 if (!__test_and_set_bit(PTR_BUCKET_NR(b->c, k, i),
1694 seen[PTR_DEV(k, i)]) ||
1695 !ptr_stale(b->c, k, i)) {
1696 g->gen = PTR_GEN(k, i);
1697
1698 if (b->level)
1699 g->prio = BTREE_PRIO;
1700 else if (g->prio == BTREE_PRIO)
1701 g->prio = INITIAL_PRIO;
1702 }
1703 }
1704
1705 btree_mark_key(b, k);
1706 }
1707
1708 if (b->level) {
1709 bch_btree_iter_init(b, &iter, NULL);
1710
1711 do {
1712 k = bch_btree_iter_next_filter(&iter, b, bch_ptr_bad);
1713 if (k)
1714 btree_node_prefetch(b->c, k, b->level - 1);
1715
1716 if (p)
1717 ret = btree(check_recurse, p, b, op, seen);
1718
1719 p = k;
1720 } while (p && !ret);
1721 }
1722
1723 return 0;
1724 }
1725
1726 int bch_btree_check(struct cache_set *c)
1727 {
1728 int ret = -ENOMEM;
1729 unsigned i;
1730 unsigned long *seen[MAX_CACHES_PER_SET];
1731 struct btree_op op;
1732
1733 memset(seen, 0, sizeof(seen));
1734 bch_btree_op_init(&op, SHRT_MAX);
1735
1736 for (i = 0; c->cache[i]; i++) {
1737 size_t n = DIV_ROUND_UP(c->cache[i]->sb.nbuckets, 8);
1738 seen[i] = kmalloc(n, GFP_KERNEL);
1739 if (!seen[i])
1740 goto err;
1741
1742 /* Disables the seen array until prio_read() uses it too */
1743 memset(seen[i], 0xFF, n);
1744 }
1745
1746 ret = btree_root(check_recurse, c, &op, seen);
1747 err:
1748 for (i = 0; i < MAX_CACHES_PER_SET; i++)
1749 kfree(seen[i]);
1750 return ret;
1751 }
1752
1753 /* Btree insertion */
1754
1755 static void shift_keys(struct btree *b, struct bkey *where, struct bkey *insert)
1756 {
1757 struct bset *i = b->sets[b->nsets].data;
1758
1759 memmove((uint64_t *) where + bkey_u64s(insert),
1760 where,
1761 (void *) end(i) - (void *) where);
1762
1763 i->keys += bkey_u64s(insert);
1764 bkey_copy(where, insert);
1765 bch_bset_fix_lookup_table(b, where);
1766 }
1767
1768 static bool fix_overlapping_extents(struct btree *b, struct bkey *insert,
1769 struct btree_iter *iter,
1770 struct bkey *replace_key)
1771 {
1772 void subtract_dirty(struct bkey *k, uint64_t offset, int sectors)
1773 {
1774 if (KEY_DIRTY(k))
1775 bcache_dev_sectors_dirty_add(b->c, KEY_INODE(k),
1776 offset, -sectors);
1777 }
1778
1779 uint64_t old_offset;
1780 unsigned old_size, sectors_found = 0;
1781
1782 while (1) {
1783 struct bkey *k = bch_btree_iter_next(iter);
1784 if (!k ||
1785 bkey_cmp(&START_KEY(k), insert) >= 0)
1786 break;
1787
1788 if (bkey_cmp(k, &START_KEY(insert)) <= 0)
1789 continue;
1790
1791 old_offset = KEY_START(k);
1792 old_size = KEY_SIZE(k);
1793
1794 /*
1795 * We might overlap with 0 size extents; we can't skip these
1796 * because if they're in the set we're inserting to we have to
1797 * adjust them so they don't overlap with the key we're
1798 * inserting. But we don't want to check them for replace
1799 * operations.
1800 */
1801
1802 if (replace_key && KEY_SIZE(k)) {
1803 /*
1804 * k might have been split since we inserted/found the
1805 * key we're replacing
1806 */
1807 unsigned i;
1808 uint64_t offset = KEY_START(k) -
1809 KEY_START(replace_key);
1810
1811 /* But it must be a subset of the replace key */
1812 if (KEY_START(k) < KEY_START(replace_key) ||
1813 KEY_OFFSET(k) > KEY_OFFSET(replace_key))
1814 goto check_failed;
1815
1816 /* We didn't find a key that we were supposed to */
1817 if (KEY_START(k) > KEY_START(insert) + sectors_found)
1818 goto check_failed;
1819
1820 if (KEY_PTRS(replace_key) != KEY_PTRS(k))
1821 goto check_failed;
1822
1823 /* skip past gen */
1824 offset <<= 8;
1825
1826 BUG_ON(!KEY_PTRS(replace_key));
1827
1828 for (i = 0; i < KEY_PTRS(replace_key); i++)
1829 if (k->ptr[i] != replace_key->ptr[i] + offset)
1830 goto check_failed;
1831
1832 sectors_found = KEY_OFFSET(k) - KEY_START(insert);
1833 }
1834
1835 if (bkey_cmp(insert, k) < 0 &&
1836 bkey_cmp(&START_KEY(insert), &START_KEY(k)) > 0) {
1837 /*
1838 * We overlapped in the middle of an existing key: that
1839 * means we have to split the old key. But we have to do
1840 * slightly different things depending on whether the
1841 * old key has been written out yet.
1842 */
1843
1844 struct bkey *top;
1845
1846 subtract_dirty(k, KEY_START(insert), KEY_SIZE(insert));
1847
1848 if (bkey_written(b, k)) {
1849 /*
1850 * We insert a new key to cover the top of the
1851 * old key, and the old key is modified in place
1852 * to represent the bottom split.
1853 *
1854 * It's completely arbitrary whether the new key
1855 * is the top or the bottom, but it has to match
1856 * up with what btree_sort_fixup() does - it
1857 * doesn't check for this kind of overlap, it
1858 * depends on us inserting a new key for the top
1859 * here.
1860 */
1861 top = bch_bset_search(b, &b->sets[b->nsets],
1862 insert);
1863 shift_keys(b, top, k);
1864 } else {
1865 BKEY_PADDED(key) temp;
1866 bkey_copy(&temp.key, k);
1867 shift_keys(b, k, &temp.key);
1868 top = bkey_next(k);
1869 }
1870
1871 bch_cut_front(insert, top);
1872 bch_cut_back(&START_KEY(insert), k);
1873 bch_bset_fix_invalidated_key(b, k);
1874 return false;
1875 }
1876
1877 if (bkey_cmp(insert, k) < 0) {
1878 bch_cut_front(insert, k);
1879 } else {
1880 if (bkey_cmp(&START_KEY(insert), &START_KEY(k)) > 0)
1881 old_offset = KEY_START(insert);
1882
1883 if (bkey_written(b, k) &&
1884 bkey_cmp(&START_KEY(insert), &START_KEY(k)) <= 0) {
1885 /*
1886 * Completely overwrote, so we don't have to
1887 * invalidate the binary search tree
1888 */
1889 bch_cut_front(k, k);
1890 } else {
1891 __bch_cut_back(&START_KEY(insert), k);
1892 bch_bset_fix_invalidated_key(b, k);
1893 }
1894 }
1895
1896 subtract_dirty(k, old_offset, old_size - KEY_SIZE(k));
1897 }
1898
1899 check_failed:
1900 if (replace_key) {
1901 if (!sectors_found) {
1902 return true;
1903 } else if (sectors_found < KEY_SIZE(insert)) {
1904 SET_KEY_OFFSET(insert, KEY_OFFSET(insert) -
1905 (KEY_SIZE(insert) - sectors_found));
1906 SET_KEY_SIZE(insert, sectors_found);
1907 }
1908 }
1909
1910 return false;
1911 }
1912
1913 static bool btree_insert_key(struct btree *b, struct btree_op *op,
1914 struct bkey *k, struct bkey *replace_key)
1915 {
1916 struct bset *i = b->sets[b->nsets].data;
1917 struct bkey *m, *prev;
1918 unsigned status = BTREE_INSERT_STATUS_INSERT;
1919
1920 BUG_ON(bkey_cmp(k, &b->key) > 0);
1921 BUG_ON(b->level && !KEY_PTRS(k));
1922 BUG_ON(!b->level && !KEY_OFFSET(k));
1923
1924 if (!b->level) {
1925 struct btree_iter iter;
1926
1927 /*
1928 * bset_search() returns the first key that is strictly greater
1929 * than the search key - but for back merging, we want to find
1930 * the previous key.
1931 */
1932 prev = NULL;
1933 m = bch_btree_iter_init(b, &iter, PRECEDING_KEY(&START_KEY(k)));
1934
1935 if (fix_overlapping_extents(b, k, &iter, replace_key)) {
1936 op->insert_collision = true;
1937 return false;
1938 }
1939
1940 if (KEY_DIRTY(k))
1941 bcache_dev_sectors_dirty_add(b->c, KEY_INODE(k),
1942 KEY_START(k), KEY_SIZE(k));
1943
1944 while (m != end(i) &&
1945 bkey_cmp(k, &START_KEY(m)) > 0)
1946 prev = m, m = bkey_next(m);
1947
1948 if (key_merging_disabled(b->c))
1949 goto insert;
1950
1951 /* prev is in the tree, if we merge we're done */
1952 status = BTREE_INSERT_STATUS_BACK_MERGE;
1953 if (prev &&
1954 bch_bkey_try_merge(b, prev, k))
1955 goto merged;
1956
1957 status = BTREE_INSERT_STATUS_OVERWROTE;
1958 if (m != end(i) &&
1959 KEY_PTRS(m) == KEY_PTRS(k) && !KEY_SIZE(m))
1960 goto copy;
1961
1962 status = BTREE_INSERT_STATUS_FRONT_MERGE;
1963 if (m != end(i) &&
1964 bch_bkey_try_merge(b, k, m))
1965 goto copy;
1966 } else {
1967 BUG_ON(replace_key);
1968 m = bch_bset_search(b, &b->sets[b->nsets], k);
1969 }
1970
1971 insert: shift_keys(b, m, k);
1972 copy: bkey_copy(m, k);
1973 merged:
1974 bch_check_keys(b, "%u for %s", status,
1975 replace_key ? "replace" : "insert");
1976
1977 if (b->level && !KEY_OFFSET(k))
1978 btree_current_write(b)->prio_blocked++;
1979
1980 trace_bcache_btree_insert_key(b, k, replace_key != NULL, status);
1981
1982 return true;
1983 }
1984
1985 static bool bch_btree_insert_keys(struct btree *b, struct btree_op *op,
1986 struct keylist *insert_keys,
1987 struct bkey *replace_key)
1988 {
1989 bool ret = false;
1990 int oldsize = bch_count_data(b);
1991
1992 while (!bch_keylist_empty(insert_keys)) {
1993 struct bset *i = write_block(b);
1994 struct bkey *k = insert_keys->keys;
1995
1996 if (b->written + __set_blocks(i, i->keys + bkey_u64s(k), b->c)
1997 > btree_blocks(b))
1998 break;
1999
2000 if (bkey_cmp(k, &b->key) <= 0) {
2001 if (!b->level)
2002 bkey_put(b->c, k);
2003
2004 ret |= btree_insert_key(b, op, k, replace_key);
2005 bch_keylist_pop_front(insert_keys);
2006 } else if (bkey_cmp(&START_KEY(k), &b->key) < 0) {
2007 BKEY_PADDED(key) temp;
2008 bkey_copy(&temp.key, insert_keys->keys);
2009
2010 bch_cut_back(&b->key, &temp.key);
2011 bch_cut_front(&b->key, insert_keys->keys);
2012
2013 ret |= btree_insert_key(b, op, &temp.key, replace_key);
2014 break;
2015 } else {
2016 break;
2017 }
2018 }
2019
2020 BUG_ON(!bch_keylist_empty(insert_keys) && b->level);
2021
2022 BUG_ON(bch_count_data(b) < oldsize);
2023 return ret;
2024 }
2025
2026 static int btree_split(struct btree *b, struct btree_op *op,
2027 struct keylist *insert_keys,
2028 struct bkey *replace_key)
2029 {
2030 bool split;
2031 struct btree *n1, *n2 = NULL, *n3 = NULL;
2032 uint64_t start_time = local_clock();
2033 struct closure cl;
2034 struct keylist parent_keys;
2035
2036 closure_init_stack(&cl);
2037 bch_keylist_init(&parent_keys);
2038
2039 n1 = btree_node_alloc_replacement(b, true);
2040 if (IS_ERR(n1))
2041 goto err;
2042
2043 split = set_blocks(n1->sets[0].data, n1->c) > (btree_blocks(b) * 4) / 5;
2044
2045 if (split) {
2046 unsigned keys = 0;
2047
2048 trace_bcache_btree_node_split(b, n1->sets[0].data->keys);
2049
2050 n2 = bch_btree_node_alloc(b->c, b->level, true);
2051 if (IS_ERR(n2))
2052 goto err_free1;
2053
2054 if (!b->parent) {
2055 n3 = bch_btree_node_alloc(b->c, b->level + 1, true);
2056 if (IS_ERR(n3))
2057 goto err_free2;
2058 }
2059
2060 bch_btree_insert_keys(n1, op, insert_keys, replace_key);
2061
2062 /*
2063 * Has to be a linear search because we don't have an auxiliary
2064 * search tree yet
2065 */
2066
2067 while (keys < (n1->sets[0].data->keys * 3) / 5)
2068 keys += bkey_u64s(node(n1->sets[0].data, keys));
2069
2070 bkey_copy_key(&n1->key, node(n1->sets[0].data, keys));
2071 keys += bkey_u64s(node(n1->sets[0].data, keys));
2072
2073 n2->sets[0].data->keys = n1->sets[0].data->keys - keys;
2074 n1->sets[0].data->keys = keys;
2075
2076 memcpy(n2->sets[0].data->start,
2077 end(n1->sets[0].data),
2078 n2->sets[0].data->keys * sizeof(uint64_t));
2079
2080 bkey_copy_key(&n2->key, &b->key);
2081
2082 bch_keylist_add(&parent_keys, &n2->key);
2083 bch_btree_node_write(n2, &cl);
2084 rw_unlock(true, n2);
2085 } else {
2086 trace_bcache_btree_node_compact(b, n1->sets[0].data->keys);
2087
2088 bch_btree_insert_keys(n1, op, insert_keys, replace_key);
2089 }
2090
2091 bch_keylist_add(&parent_keys, &n1->key);
2092 bch_btree_node_write(n1, &cl);
2093
2094 if (n3) {
2095 /* Depth increases, make a new root */
2096 bkey_copy_key(&n3->key, &MAX_KEY);
2097 bch_btree_insert_keys(n3, op, &parent_keys, NULL);
2098 bch_btree_node_write(n3, &cl);
2099
2100 closure_sync(&cl);
2101 bch_btree_set_root(n3);
2102 rw_unlock(true, n3);
2103
2104 btree_node_free(b);
2105 } else if (!b->parent) {
2106 /* Root filled up but didn't need to be split */
2107 closure_sync(&cl);
2108 bch_btree_set_root(n1);
2109
2110 btree_node_free(b);
2111 } else {
2112 /* Split a non root node */
2113 closure_sync(&cl);
2114 make_btree_freeing_key(b, parent_keys.top);
2115 bch_keylist_push(&parent_keys);
2116
2117 btree_node_free(b);
2118
2119 bch_btree_insert_node(b->parent, op, &parent_keys, NULL, NULL);
2120 BUG_ON(!bch_keylist_empty(&parent_keys));
2121 }
2122
2123 rw_unlock(true, n1);
2124
2125 bch_time_stats_update(&b->c->btree_split_time, start_time);
2126
2127 return 0;
2128 err_free2:
2129 btree_node_free(n2);
2130 rw_unlock(true, n2);
2131 err_free1:
2132 btree_node_free(n1);
2133 rw_unlock(true, n1);
2134 err:
2135 if (n3 == ERR_PTR(-EAGAIN) ||
2136 n2 == ERR_PTR(-EAGAIN) ||
2137 n1 == ERR_PTR(-EAGAIN))
2138 return -EAGAIN;
2139
2140 pr_warn("couldn't split");
2141 return -ENOMEM;
2142 }
2143
2144 static int bch_btree_insert_node(struct btree *b, struct btree_op *op,
2145 struct keylist *insert_keys,
2146 atomic_t *journal_ref,
2147 struct bkey *replace_key)
2148 {
2149 BUG_ON(b->level && replace_key);
2150
2151 if (should_split(b)) {
2152 if (current->bio_list) {
2153 op->lock = b->c->root->level + 1;
2154 return -EAGAIN;
2155 } else if (op->lock <= b->c->root->level) {
2156 op->lock = b->c->root->level + 1;
2157 return -EINTR;
2158 } else {
2159 /* Invalidated all iterators */
2160 return btree_split(b, op, insert_keys, replace_key) ?:
2161 -EINTR;
2162 }
2163 } else {
2164 BUG_ON(write_block(b) != b->sets[b->nsets].data);
2165
2166 if (bch_btree_insert_keys(b, op, insert_keys, replace_key)) {
2167 if (!b->level)
2168 bch_btree_leaf_dirty(b, journal_ref);
2169 else
2170 bch_btree_node_write_sync(b);
2171 }
2172
2173 return 0;
2174 }
2175 }
2176
2177 int bch_btree_insert_check_key(struct btree *b, struct btree_op *op,
2178 struct bkey *check_key)
2179 {
2180 int ret = -EINTR;
2181 uint64_t btree_ptr = b->key.ptr[0];
2182 unsigned long seq = b->seq;
2183 struct keylist insert;
2184 bool upgrade = op->lock == -1;
2185
2186 bch_keylist_init(&insert);
2187
2188 if (upgrade) {
2189 rw_unlock(false, b);
2190 rw_lock(true, b, b->level);
2191
2192 if (b->key.ptr[0] != btree_ptr ||
2193 b->seq != seq + 1)
2194 goto out;
2195 }
2196
2197 SET_KEY_PTRS(check_key, 1);
2198 get_random_bytes(&check_key->ptr[0], sizeof(uint64_t));
2199
2200 SET_PTR_DEV(check_key, 0, PTR_CHECK_DEV);
2201
2202 bch_keylist_add(&insert, check_key);
2203
2204 ret = bch_btree_insert_node(b, op, &insert, NULL, NULL);
2205
2206 BUG_ON(!ret && !bch_keylist_empty(&insert));
2207 out:
2208 if (upgrade)
2209 downgrade_write(&b->lock);
2210 return ret;
2211 }
2212
2213 struct btree_insert_op {
2214 struct btree_op op;
2215 struct keylist *keys;
2216 atomic_t *journal_ref;
2217 struct bkey *replace_key;
2218 };
2219
2220 int btree_insert_fn(struct btree_op *b_op, struct btree *b)
2221 {
2222 struct btree_insert_op *op = container_of(b_op,
2223 struct btree_insert_op, op);
2224
2225 int ret = bch_btree_insert_node(b, &op->op, op->keys,
2226 op->journal_ref, op->replace_key);
2227 if (ret && !bch_keylist_empty(op->keys))
2228 return ret;
2229 else
2230 return MAP_DONE;
2231 }
2232
2233 int bch_btree_insert(struct cache_set *c, struct keylist *keys,
2234 atomic_t *journal_ref, struct bkey *replace_key)
2235 {
2236 struct btree_insert_op op;
2237 int ret = 0;
2238
2239 BUG_ON(current->bio_list);
2240 BUG_ON(bch_keylist_empty(keys));
2241
2242 bch_btree_op_init(&op.op, 0);
2243 op.keys = keys;
2244 op.journal_ref = journal_ref;
2245 op.replace_key = replace_key;
2246
2247 while (!ret && !bch_keylist_empty(keys)) {
2248 op.op.lock = 0;
2249 ret = bch_btree_map_leaf_nodes(&op.op, c,
2250 &START_KEY(keys->keys),
2251 btree_insert_fn);
2252 }
2253
2254 if (ret) {
2255 struct bkey *k;
2256
2257 pr_err("error %i", ret);
2258
2259 while ((k = bch_keylist_pop(keys)))
2260 bkey_put(c, k);
2261 } else if (op.op.insert_collision)
2262 ret = -ESRCH;
2263
2264 return ret;
2265 }
2266
2267 void bch_btree_set_root(struct btree *b)
2268 {
2269 unsigned i;
2270 struct closure cl;
2271
2272 closure_init_stack(&cl);
2273
2274 trace_bcache_btree_set_root(b);
2275
2276 BUG_ON(!b->written);
2277
2278 for (i = 0; i < KEY_PTRS(&b->key); i++)
2279 BUG_ON(PTR_BUCKET(b->c, &b->key, i)->prio != BTREE_PRIO);
2280
2281 mutex_lock(&b->c->bucket_lock);
2282 list_del_init(&b->list);
2283 mutex_unlock(&b->c->bucket_lock);
2284
2285 b->c->root = b;
2286
2287 bch_journal_meta(b->c, &cl);
2288 closure_sync(&cl);
2289 }
2290
2291 /* Map across nodes or keys */
2292
2293 static int bch_btree_map_nodes_recurse(struct btree *b, struct btree_op *op,
2294 struct bkey *from,
2295 btree_map_nodes_fn *fn, int flags)
2296 {
2297 int ret = MAP_CONTINUE;
2298
2299 if (b->level) {
2300 struct bkey *k;
2301 struct btree_iter iter;
2302
2303 bch_btree_iter_init(b, &iter, from);
2304
2305 while ((k = bch_btree_iter_next_filter(&iter, b,
2306 bch_ptr_bad))) {
2307 ret = btree(map_nodes_recurse, k, b,
2308 op, from, fn, flags);
2309 from = NULL;
2310
2311 if (ret != MAP_CONTINUE)
2312 return ret;
2313 }
2314 }
2315
2316 if (!b->level || flags == MAP_ALL_NODES)
2317 ret = fn(op, b);
2318
2319 return ret;
2320 }
2321
2322 int __bch_btree_map_nodes(struct btree_op *op, struct cache_set *c,
2323 struct bkey *from, btree_map_nodes_fn *fn, int flags)
2324 {
2325 return btree_root(map_nodes_recurse, c, op, from, fn, flags);
2326 }
2327
2328 static int bch_btree_map_keys_recurse(struct btree *b, struct btree_op *op,
2329 struct bkey *from, btree_map_keys_fn *fn,
2330 int flags)
2331 {
2332 int ret = MAP_CONTINUE;
2333 struct bkey *k;
2334 struct btree_iter iter;
2335
2336 bch_btree_iter_init(b, &iter, from);
2337
2338 while ((k = bch_btree_iter_next_filter(&iter, b, bch_ptr_bad))) {
2339 ret = !b->level
2340 ? fn(op, b, k)
2341 : btree(map_keys_recurse, k, b, op, from, fn, flags);
2342 from = NULL;
2343
2344 if (ret != MAP_CONTINUE)
2345 return ret;
2346 }
2347
2348 if (!b->level && (flags & MAP_END_KEY))
2349 ret = fn(op, b, &KEY(KEY_INODE(&b->key),
2350 KEY_OFFSET(&b->key), 0));
2351
2352 return ret;
2353 }
2354
2355 int bch_btree_map_keys(struct btree_op *op, struct cache_set *c,
2356 struct bkey *from, btree_map_keys_fn *fn, int flags)
2357 {
2358 return btree_root(map_keys_recurse, c, op, from, fn, flags);
2359 }
2360
2361 /* Keybuf code */
2362
2363 static inline int keybuf_cmp(struct keybuf_key *l, struct keybuf_key *r)
2364 {
2365 /* Overlapping keys compare equal */
2366 if (bkey_cmp(&l->key, &START_KEY(&r->key)) <= 0)
2367 return -1;
2368 if (bkey_cmp(&START_KEY(&l->key), &r->key) >= 0)
2369 return 1;
2370 return 0;
2371 }
2372
2373 static inline int keybuf_nonoverlapping_cmp(struct keybuf_key *l,
2374 struct keybuf_key *r)
2375 {
2376 return clamp_t(int64_t, bkey_cmp(&l->key, &r->key), -1, 1);
2377 }
2378
2379 struct refill {
2380 struct btree_op op;
2381 unsigned nr_found;
2382 struct keybuf *buf;
2383 struct bkey *end;
2384 keybuf_pred_fn *pred;
2385 };
2386
2387 static int refill_keybuf_fn(struct btree_op *op, struct btree *b,
2388 struct bkey *k)
2389 {
2390 struct refill *refill = container_of(op, struct refill, op);
2391 struct keybuf *buf = refill->buf;
2392 int ret = MAP_CONTINUE;
2393
2394 if (bkey_cmp(k, refill->end) >= 0) {
2395 ret = MAP_DONE;
2396 goto out;
2397 }
2398
2399 if (!KEY_SIZE(k)) /* end key */
2400 goto out;
2401
2402 if (refill->pred(buf, k)) {
2403 struct keybuf_key *w;
2404
2405 spin_lock(&buf->lock);
2406
2407 w = array_alloc(&buf->freelist);
2408 if (!w) {
2409 spin_unlock(&buf->lock);
2410 return MAP_DONE;
2411 }
2412
2413 w->private = NULL;
2414 bkey_copy(&w->key, k);
2415
2416 if (RB_INSERT(&buf->keys, w, node, keybuf_cmp))
2417 array_free(&buf->freelist, w);
2418 else
2419 refill->nr_found++;
2420
2421 if (array_freelist_empty(&buf->freelist))
2422 ret = MAP_DONE;
2423
2424 spin_unlock(&buf->lock);
2425 }
2426 out:
2427 buf->last_scanned = *k;
2428 return ret;
2429 }
2430
2431 void bch_refill_keybuf(struct cache_set *c, struct keybuf *buf,
2432 struct bkey *end, keybuf_pred_fn *pred)
2433 {
2434 struct bkey start = buf->last_scanned;
2435 struct refill refill;
2436
2437 cond_resched();
2438
2439 bch_btree_op_init(&refill.op, -1);
2440 refill.nr_found = 0;
2441 refill.buf = buf;
2442 refill.end = end;
2443 refill.pred = pred;
2444
2445 bch_btree_map_keys(&refill.op, c, &buf->last_scanned,
2446 refill_keybuf_fn, MAP_END_KEY);
2447
2448 trace_bcache_keyscan(refill.nr_found,
2449 KEY_INODE(&start), KEY_OFFSET(&start),
2450 KEY_INODE(&buf->last_scanned),
2451 KEY_OFFSET(&buf->last_scanned));
2452
2453 spin_lock(&buf->lock);
2454
2455 if (!RB_EMPTY_ROOT(&buf->keys)) {
2456 struct keybuf_key *w;
2457 w = RB_FIRST(&buf->keys, struct keybuf_key, node);
2458 buf->start = START_KEY(&w->key);
2459
2460 w = RB_LAST(&buf->keys, struct keybuf_key, node);
2461 buf->end = w->key;
2462 } else {
2463 buf->start = MAX_KEY;
2464 buf->end = MAX_KEY;
2465 }
2466
2467 spin_unlock(&buf->lock);
2468 }
2469
2470 static void __bch_keybuf_del(struct keybuf *buf, struct keybuf_key *w)
2471 {
2472 rb_erase(&w->node, &buf->keys);
2473 array_free(&buf->freelist, w);
2474 }
2475
2476 void bch_keybuf_del(struct keybuf *buf, struct keybuf_key *w)
2477 {
2478 spin_lock(&buf->lock);
2479 __bch_keybuf_del(buf, w);
2480 spin_unlock(&buf->lock);
2481 }
2482
2483 bool bch_keybuf_check_overlapping(struct keybuf *buf, struct bkey *start,
2484 struct bkey *end)
2485 {
2486 bool ret = false;
2487 struct keybuf_key *p, *w, s;
2488 s.key = *start;
2489
2490 if (bkey_cmp(end, &buf->start) <= 0 ||
2491 bkey_cmp(start, &buf->end) >= 0)
2492 return false;
2493
2494 spin_lock(&buf->lock);
2495 w = RB_GREATER(&buf->keys, s, node, keybuf_nonoverlapping_cmp);
2496
2497 while (w && bkey_cmp(&START_KEY(&w->key), end) < 0) {
2498 p = w;
2499 w = RB_NEXT(w, node);
2500
2501 if (p->private)
2502 ret = true;
2503 else
2504 __bch_keybuf_del(buf, p);
2505 }
2506
2507 spin_unlock(&buf->lock);
2508 return ret;
2509 }
2510
2511 struct keybuf_key *bch_keybuf_next(struct keybuf *buf)
2512 {
2513 struct keybuf_key *w;
2514 spin_lock(&buf->lock);
2515
2516 w = RB_FIRST(&buf->keys, struct keybuf_key, node);
2517
2518 while (w && w->private)
2519 w = RB_NEXT(w, node);
2520
2521 if (w)
2522 w->private = ERR_PTR(-EINTR);
2523
2524 spin_unlock(&buf->lock);
2525 return w;
2526 }
2527
2528 struct keybuf_key *bch_keybuf_next_rescan(struct cache_set *c,
2529 struct keybuf *buf,
2530 struct bkey *end,
2531 keybuf_pred_fn *pred)
2532 {
2533 struct keybuf_key *ret;
2534
2535 while (1) {
2536 ret = bch_keybuf_next(buf);
2537 if (ret)
2538 break;
2539
2540 if (bkey_cmp(&buf->last_scanned, end) >= 0) {
2541 pr_debug("scan finished");
2542 break;
2543 }
2544
2545 bch_refill_keybuf(c, buf, end, pred);
2546 }
2547
2548 return ret;
2549 }
2550
2551 void bch_keybuf_init(struct keybuf *buf)
2552 {
2553 buf->last_scanned = MAX_KEY;
2554 buf->keys = RB_ROOT;
2555
2556 spin_lock_init(&buf->lock);
2557 array_allocator_init(&buf->freelist);
2558 }
2559
2560 void bch_btree_exit(void)
2561 {
2562 if (btree_io_wq)
2563 destroy_workqueue(btree_io_wq);
2564 }
2565
2566 int __init bch_btree_init(void)
2567 {
2568 btree_io_wq = create_singlethread_workqueue("bch_btree_io");
2569 if (!btree_io_wq)
2570 return -ENOMEM;
2571
2572 return 0;
2573 }
Linus' kernel tree