xref: /openbmc/linux/drivers/md/bcache/extents.c (revision 2359ccdd)
1 // SPDX-License-Identifier: GPL-2.0
2 /*
3  * Copyright (C) 2010 Kent Overstreet <kent.overstreet@gmail.com>
4  *
5  * Uses a block device as cache for other block devices; optimized for SSDs.
6  * All allocation is done in buckets, which should match the erase block size
7  * of the device.
8  *
9  * Buckets containing cached data are kept on a heap sorted by priority;
10  * bucket priority is increased on cache hit, and periodically all the buckets
11  * on the heap have their priority scaled down. This currently is just used as
12  * an LRU but in the future should allow for more intelligent heuristics.
13  *
14  * Buckets have an 8 bit counter; freeing is accomplished by incrementing the
15  * counter. Garbage collection is used to remove stale pointers.
16  *
17  * Indexing is done via a btree; nodes are not necessarily fully sorted, rather
18  * as keys are inserted we only sort the pages that have not yet been written.
19  * When garbage collection is run, we resort the entire node.
20  *
21  * All configuration is done via sysfs; see Documentation/bcache.txt.
22  */
23 
24 #include "bcache.h"
25 #include "btree.h"
26 #include "debug.h"
27 #include "extents.h"
28 #include "writeback.h"
29 
30 static void sort_key_next(struct btree_iter *iter,
31 			  struct btree_iter_set *i)
32 {
33 	i->k = bkey_next(i->k);
34 
35 	if (i->k == i->end)
36 		*i = iter->data[--iter->used];
37 }
38 
39 static bool bch_key_sort_cmp(struct btree_iter_set l,
40 			     struct btree_iter_set r)
41 {
42 	int64_t c = bkey_cmp(l.k, r.k);
43 
44 	return c ? c > 0 : l.k < r.k;
45 }
46 
47 static bool __ptr_invalid(struct cache_set *c, const struct bkey *k)
48 {
49 	unsigned i;
50 
51 	for (i = 0; i < KEY_PTRS(k); i++)
52 		if (ptr_available(c, k, i)) {
53 			struct cache *ca = PTR_CACHE(c, k, i);
54 			size_t bucket = PTR_BUCKET_NR(c, k, i);
55 			size_t r = bucket_remainder(c, PTR_OFFSET(k, i));
56 
57 			if (KEY_SIZE(k) + r > c->sb.bucket_size ||
58 			    bucket <  ca->sb.first_bucket ||
59 			    bucket >= ca->sb.nbuckets)
60 				return true;
61 		}
62 
63 	return false;
64 }
65 
66 /* Common among btree and extent ptrs */
67 
68 static const char *bch_ptr_status(struct cache_set *c, const struct bkey *k)
69 {
70 	unsigned i;
71 
72 	for (i = 0; i < KEY_PTRS(k); i++)
73 		if (ptr_available(c, k, i)) {
74 			struct cache *ca = PTR_CACHE(c, k, i);
75 			size_t bucket = PTR_BUCKET_NR(c, k, i);
76 			size_t r = bucket_remainder(c, PTR_OFFSET(k, i));
77 
78 			if (KEY_SIZE(k) + r > c->sb.bucket_size)
79 				return "bad, length too big";
80 			if (bucket <  ca->sb.first_bucket)
81 				return "bad, short offset";
82 			if (bucket >= ca->sb.nbuckets)
83 				return "bad, offset past end of device";
84 			if (ptr_stale(c, k, i))
85 				return "stale";
86 		}
87 
88 	if (!bkey_cmp(k, &ZERO_KEY))
89 		return "bad, null key";
90 	if (!KEY_PTRS(k))
91 		return "bad, no pointers";
92 	if (!KEY_SIZE(k))
93 		return "zeroed key";
94 	return "";
95 }
96 
97 void bch_extent_to_text(char *buf, size_t size, const struct bkey *k)
98 {
99 	unsigned i = 0;
100 	char *out = buf, *end = buf + size;
101 
102 #define p(...)	(out += scnprintf(out, end - out, __VA_ARGS__))
103 
104 	p("%llu:%llu len %llu -> [", KEY_INODE(k), KEY_START(k), KEY_SIZE(k));
105 
106 	for (i = 0; i < KEY_PTRS(k); i++) {
107 		if (i)
108 			p(", ");
109 
110 		if (PTR_DEV(k, i) == PTR_CHECK_DEV)
111 			p("check dev");
112 		else
113 			p("%llu:%llu gen %llu", PTR_DEV(k, i),
114 			  PTR_OFFSET(k, i), PTR_GEN(k, i));
115 	}
116 
117 	p("]");
118 
119 	if (KEY_DIRTY(k))
120 		p(" dirty");
121 	if (KEY_CSUM(k))
122 		p(" cs%llu %llx", KEY_CSUM(k), k->ptr[1]);
123 #undef p
124 }
125 
126 static void bch_bkey_dump(struct btree_keys *keys, const struct bkey *k)
127 {
128 	struct btree *b = container_of(keys, struct btree, keys);
129 	unsigned j;
130 	char buf[80];
131 
132 	bch_extent_to_text(buf, sizeof(buf), k);
133 	printk(" %s", buf);
134 
135 	for (j = 0; j < KEY_PTRS(k); j++) {
136 		size_t n = PTR_BUCKET_NR(b->c, k, j);
137 		printk(" bucket %zu", n);
138 
139 		if (n >= b->c->sb.first_bucket && n < b->c->sb.nbuckets)
140 			printk(" prio %i",
141 			       PTR_BUCKET(b->c, k, j)->prio);
142 	}
143 
144 	printk(" %s\n", bch_ptr_status(b->c, k));
145 }
146 
147 /* Btree ptrs */
148 
149 bool __bch_btree_ptr_invalid(struct cache_set *c, const struct bkey *k)
150 {
151 	char buf[80];
152 
153 	if (!KEY_PTRS(k) || !KEY_SIZE(k) || KEY_DIRTY(k))
154 		goto bad;
155 
156 	if (__ptr_invalid(c, k))
157 		goto bad;
158 
159 	return false;
160 bad:
161 	bch_extent_to_text(buf, sizeof(buf), k);
162 	cache_bug(c, "spotted btree ptr %s: %s", buf, bch_ptr_status(c, k));
163 	return true;
164 }
165 
166 static bool bch_btree_ptr_invalid(struct btree_keys *bk, const struct bkey *k)
167 {
168 	struct btree *b = container_of(bk, struct btree, keys);
169 	return __bch_btree_ptr_invalid(b->c, k);
170 }
171 
172 static bool btree_ptr_bad_expensive(struct btree *b, const struct bkey *k)
173 {
174 	unsigned i;
175 	char buf[80];
176 	struct bucket *g;
177 
178 	if (mutex_trylock(&b->c->bucket_lock)) {
179 		for (i = 0; i < KEY_PTRS(k); i++)
180 			if (ptr_available(b->c, k, i)) {
181 				g = PTR_BUCKET(b->c, k, i);
182 
183 				if (KEY_DIRTY(k) ||
184 				    g->prio != BTREE_PRIO ||
185 				    (b->c->gc_mark_valid &&
186 				     GC_MARK(g) != GC_MARK_METADATA))
187 					goto err;
188 			}
189 
190 		mutex_unlock(&b->c->bucket_lock);
191 	}
192 
193 	return false;
194 err:
195 	mutex_unlock(&b->c->bucket_lock);
196 	bch_extent_to_text(buf, sizeof(buf), k);
197 	btree_bug(b,
198 "inconsistent btree pointer %s: bucket %zi pin %i prio %i gen %i last_gc %i mark %llu",
199 		  buf, PTR_BUCKET_NR(b->c, k, i), atomic_read(&g->pin),
200 		  g->prio, g->gen, g->last_gc, GC_MARK(g));
201 	return true;
202 }
203 
204 static bool bch_btree_ptr_bad(struct btree_keys *bk, const struct bkey *k)
205 {
206 	struct btree *b = container_of(bk, struct btree, keys);
207 	unsigned i;
208 
209 	if (!bkey_cmp(k, &ZERO_KEY) ||
210 	    !KEY_PTRS(k) ||
211 	    bch_ptr_invalid(bk, k))
212 		return true;
213 
214 	for (i = 0; i < KEY_PTRS(k); i++)
215 		if (!ptr_available(b->c, k, i) ||
216 		    ptr_stale(b->c, k, i))
217 			return true;
218 
219 	if (expensive_debug_checks(b->c) &&
220 	    btree_ptr_bad_expensive(b, k))
221 		return true;
222 
223 	return false;
224 }
225 
226 static bool bch_btree_ptr_insert_fixup(struct btree_keys *bk,
227 				       struct bkey *insert,
228 				       struct btree_iter *iter,
229 				       struct bkey *replace_key)
230 {
231 	struct btree *b = container_of(bk, struct btree, keys);
232 
233 	if (!KEY_OFFSET(insert))
234 		btree_current_write(b)->prio_blocked++;
235 
236 	return false;
237 }
238 
239 const struct btree_keys_ops bch_btree_keys_ops = {
240 	.sort_cmp	= bch_key_sort_cmp,
241 	.insert_fixup	= bch_btree_ptr_insert_fixup,
242 	.key_invalid	= bch_btree_ptr_invalid,
243 	.key_bad	= bch_btree_ptr_bad,
244 	.key_to_text	= bch_extent_to_text,
245 	.key_dump	= bch_bkey_dump,
246 };
247 
248 /* Extents */
249 
250 /*
251  * Returns true if l > r - unless l == r, in which case returns true if l is
252  * older than r.
253  *
254  * Necessary for btree_sort_fixup() - if there are multiple keys that compare
255  * equal in different sets, we have to process them newest to oldest.
256  */
257 static bool bch_extent_sort_cmp(struct btree_iter_set l,
258 				struct btree_iter_set r)
259 {
260 	int64_t c = bkey_cmp(&START_KEY(l.k), &START_KEY(r.k));
261 
262 	return c ? c > 0 : l.k < r.k;
263 }
264 
265 static struct bkey *bch_extent_sort_fixup(struct btree_iter *iter,
266 					  struct bkey *tmp)
267 {
268 	while (iter->used > 1) {
269 		struct btree_iter_set *top = iter->data, *i = top + 1;
270 
271 		if (iter->used > 2 &&
272 		    bch_extent_sort_cmp(i[0], i[1]))
273 			i++;
274 
275 		if (bkey_cmp(top->k, &START_KEY(i->k)) <= 0)
276 			break;
277 
278 		if (!KEY_SIZE(i->k)) {
279 			sort_key_next(iter, i);
280 			heap_sift(iter, i - top, bch_extent_sort_cmp);
281 			continue;
282 		}
283 
284 		if (top->k > i->k) {
285 			if (bkey_cmp(top->k, i->k) >= 0)
286 				sort_key_next(iter, i);
287 			else
288 				bch_cut_front(top->k, i->k);
289 
290 			heap_sift(iter, i - top, bch_extent_sort_cmp);
291 		} else {
292 			/* can't happen because of comparison func */
293 			BUG_ON(!bkey_cmp(&START_KEY(top->k), &START_KEY(i->k)));
294 
295 			if (bkey_cmp(i->k, top->k) < 0) {
296 				bkey_copy(tmp, top->k);
297 
298 				bch_cut_back(&START_KEY(i->k), tmp);
299 				bch_cut_front(i->k, top->k);
300 				heap_sift(iter, 0, bch_extent_sort_cmp);
301 
302 				return tmp;
303 			} else {
304 				bch_cut_back(&START_KEY(i->k), top->k);
305 			}
306 		}
307 	}
308 
309 	return NULL;
310 }
311 
312 static void bch_subtract_dirty(struct bkey *k,
313 			   struct cache_set *c,
314 			   uint64_t offset,
315 			   int sectors)
316 {
317 	if (KEY_DIRTY(k))
318 		bcache_dev_sectors_dirty_add(c, KEY_INODE(k),
319 					     offset, -sectors);
320 }
321 
322 static bool bch_extent_insert_fixup(struct btree_keys *b,
323 				    struct bkey *insert,
324 				    struct btree_iter *iter,
325 				    struct bkey *replace_key)
326 {
327 	struct cache_set *c = container_of(b, struct btree, keys)->c;
328 
329 	uint64_t old_offset;
330 	unsigned old_size, sectors_found = 0;
331 
332 	BUG_ON(!KEY_OFFSET(insert));
333 	BUG_ON(!KEY_SIZE(insert));
334 
335 	while (1) {
336 		struct bkey *k = bch_btree_iter_next(iter);
337 		if (!k)
338 			break;
339 
340 		if (bkey_cmp(&START_KEY(k), insert) >= 0) {
341 			if (KEY_SIZE(k))
342 				break;
343 			else
344 				continue;
345 		}
346 
347 		if (bkey_cmp(k, &START_KEY(insert)) <= 0)
348 			continue;
349 
350 		old_offset = KEY_START(k);
351 		old_size = KEY_SIZE(k);
352 
353 		/*
354 		 * We might overlap with 0 size extents; we can't skip these
355 		 * because if they're in the set we're inserting to we have to
356 		 * adjust them so they don't overlap with the key we're
357 		 * inserting. But we don't want to check them for replace
358 		 * operations.
359 		 */
360 
361 		if (replace_key && KEY_SIZE(k)) {
362 			/*
363 			 * k might have been split since we inserted/found the
364 			 * key we're replacing
365 			 */
366 			unsigned i;
367 			uint64_t offset = KEY_START(k) -
368 				KEY_START(replace_key);
369 
370 			/* But it must be a subset of the replace key */
371 			if (KEY_START(k) < KEY_START(replace_key) ||
372 			    KEY_OFFSET(k) > KEY_OFFSET(replace_key))
373 				goto check_failed;
374 
375 			/* We didn't find a key that we were supposed to */
376 			if (KEY_START(k) > KEY_START(insert) + sectors_found)
377 				goto check_failed;
378 
379 			if (!bch_bkey_equal_header(k, replace_key))
380 				goto check_failed;
381 
382 			/* skip past gen */
383 			offset <<= 8;
384 
385 			BUG_ON(!KEY_PTRS(replace_key));
386 
387 			for (i = 0; i < KEY_PTRS(replace_key); i++)
388 				if (k->ptr[i] != replace_key->ptr[i] + offset)
389 					goto check_failed;
390 
391 			sectors_found = KEY_OFFSET(k) - KEY_START(insert);
392 		}
393 
394 		if (bkey_cmp(insert, k) < 0 &&
395 		    bkey_cmp(&START_KEY(insert), &START_KEY(k)) > 0) {
396 			/*
397 			 * We overlapped in the middle of an existing key: that
398 			 * means we have to split the old key. But we have to do
399 			 * slightly different things depending on whether the
400 			 * old key has been written out yet.
401 			 */
402 
403 			struct bkey *top;
404 
405 			bch_subtract_dirty(k, c, KEY_START(insert),
406 				       KEY_SIZE(insert));
407 
408 			if (bkey_written(b, k)) {
409 				/*
410 				 * We insert a new key to cover the top of the
411 				 * old key, and the old key is modified in place
412 				 * to represent the bottom split.
413 				 *
414 				 * It's completely arbitrary whether the new key
415 				 * is the top or the bottom, but it has to match
416 				 * up with what btree_sort_fixup() does - it
417 				 * doesn't check for this kind of overlap, it
418 				 * depends on us inserting a new key for the top
419 				 * here.
420 				 */
421 				top = bch_bset_search(b, bset_tree_last(b),
422 						      insert);
423 				bch_bset_insert(b, top, k);
424 			} else {
425 				BKEY_PADDED(key) temp;
426 				bkey_copy(&temp.key, k);
427 				bch_bset_insert(b, k, &temp.key);
428 				top = bkey_next(k);
429 			}
430 
431 			bch_cut_front(insert, top);
432 			bch_cut_back(&START_KEY(insert), k);
433 			bch_bset_fix_invalidated_key(b, k);
434 			goto out;
435 		}
436 
437 		if (bkey_cmp(insert, k) < 0) {
438 			bch_cut_front(insert, k);
439 		} else {
440 			if (bkey_cmp(&START_KEY(insert), &START_KEY(k)) > 0)
441 				old_offset = KEY_START(insert);
442 
443 			if (bkey_written(b, k) &&
444 			    bkey_cmp(&START_KEY(insert), &START_KEY(k)) <= 0) {
445 				/*
446 				 * Completely overwrote, so we don't have to
447 				 * invalidate the binary search tree
448 				 */
449 				bch_cut_front(k, k);
450 			} else {
451 				__bch_cut_back(&START_KEY(insert), k);
452 				bch_bset_fix_invalidated_key(b, k);
453 			}
454 		}
455 
456 		bch_subtract_dirty(k, c, old_offset, old_size - KEY_SIZE(k));
457 	}
458 
459 check_failed:
460 	if (replace_key) {
461 		if (!sectors_found) {
462 			return true;
463 		} else if (sectors_found < KEY_SIZE(insert)) {
464 			SET_KEY_OFFSET(insert, KEY_OFFSET(insert) -
465 				       (KEY_SIZE(insert) - sectors_found));
466 			SET_KEY_SIZE(insert, sectors_found);
467 		}
468 	}
469 out:
470 	if (KEY_DIRTY(insert))
471 		bcache_dev_sectors_dirty_add(c, KEY_INODE(insert),
472 					     KEY_START(insert),
473 					     KEY_SIZE(insert));
474 
475 	return false;
476 }
477 
478 bool __bch_extent_invalid(struct cache_set *c, const struct bkey *k)
479 {
480 	char buf[80];
481 
482 	if (!KEY_SIZE(k))
483 		return true;
484 
485 	if (KEY_SIZE(k) > KEY_OFFSET(k))
486 		goto bad;
487 
488 	if (__ptr_invalid(c, k))
489 		goto bad;
490 
491 	return false;
492 bad:
493 	bch_extent_to_text(buf, sizeof(buf), k);
494 	cache_bug(c, "spotted extent %s: %s", buf, bch_ptr_status(c, k));
495 	return true;
496 }
497 
498 static bool bch_extent_invalid(struct btree_keys *bk, const struct bkey *k)
499 {
500 	struct btree *b = container_of(bk, struct btree, keys);
501 	return __bch_extent_invalid(b->c, k);
502 }
503 
504 static bool bch_extent_bad_expensive(struct btree *b, const struct bkey *k,
505 				     unsigned ptr)
506 {
507 	struct bucket *g = PTR_BUCKET(b->c, k, ptr);
508 	char buf[80];
509 
510 	if (mutex_trylock(&b->c->bucket_lock)) {
511 		if (b->c->gc_mark_valid &&
512 		    (!GC_MARK(g) ||
513 		     GC_MARK(g) == GC_MARK_METADATA ||
514 		     (GC_MARK(g) != GC_MARK_DIRTY && KEY_DIRTY(k))))
515 			goto err;
516 
517 		if (g->prio == BTREE_PRIO)
518 			goto err;
519 
520 		mutex_unlock(&b->c->bucket_lock);
521 	}
522 
523 	return false;
524 err:
525 	mutex_unlock(&b->c->bucket_lock);
526 	bch_extent_to_text(buf, sizeof(buf), k);
527 	btree_bug(b,
528 "inconsistent extent pointer %s:\nbucket %zu pin %i prio %i gen %i last_gc %i mark %llu",
529 		  buf, PTR_BUCKET_NR(b->c, k, ptr), atomic_read(&g->pin),
530 		  g->prio, g->gen, g->last_gc, GC_MARK(g));
531 	return true;
532 }
533 
534 static bool bch_extent_bad(struct btree_keys *bk, const struct bkey *k)
535 {
536 	struct btree *b = container_of(bk, struct btree, keys);
537 	unsigned i, stale;
538 
539 	if (!KEY_PTRS(k) ||
540 	    bch_extent_invalid(bk, k))
541 		return true;
542 
543 	for (i = 0; i < KEY_PTRS(k); i++)
544 		if (!ptr_available(b->c, k, i))
545 			return true;
546 
547 	if (!expensive_debug_checks(b->c) && KEY_DIRTY(k))
548 		return false;
549 
550 	for (i = 0; i < KEY_PTRS(k); i++) {
551 		stale = ptr_stale(b->c, k, i);
552 
553 		btree_bug_on(stale > 96, b,
554 			     "key too stale: %i, need_gc %u",
555 			     stale, b->c->need_gc);
556 
557 		btree_bug_on(stale && KEY_DIRTY(k) && KEY_SIZE(k),
558 			     b, "stale dirty pointer");
559 
560 		if (stale)
561 			return true;
562 
563 		if (expensive_debug_checks(b->c) &&
564 		    bch_extent_bad_expensive(b, k, i))
565 			return true;
566 	}
567 
568 	return false;
569 }
570 
571 static uint64_t merge_chksums(struct bkey *l, struct bkey *r)
572 {
573 	return (l->ptr[KEY_PTRS(l)] + r->ptr[KEY_PTRS(r)]) &
574 		~((uint64_t)1 << 63);
575 }
576 
577 static bool bch_extent_merge(struct btree_keys *bk, struct bkey *l, struct bkey *r)
578 {
579 	struct btree *b = container_of(bk, struct btree, keys);
580 	unsigned i;
581 
582 	if (key_merging_disabled(b->c))
583 		return false;
584 
585 	for (i = 0; i < KEY_PTRS(l); i++)
586 		if (l->ptr[i] + MAKE_PTR(0, KEY_SIZE(l), 0) != r->ptr[i] ||
587 		    PTR_BUCKET_NR(b->c, l, i) != PTR_BUCKET_NR(b->c, r, i))
588 			return false;
589 
590 	/* Keys with no pointers aren't restricted to one bucket and could
591 	 * overflow KEY_SIZE
592 	 */
593 	if (KEY_SIZE(l) + KEY_SIZE(r) > USHRT_MAX) {
594 		SET_KEY_OFFSET(l, KEY_OFFSET(l) + USHRT_MAX - KEY_SIZE(l));
595 		SET_KEY_SIZE(l, USHRT_MAX);
596 
597 		bch_cut_front(l, r);
598 		return false;
599 	}
600 
601 	if (KEY_CSUM(l)) {
602 		if (KEY_CSUM(r))
603 			l->ptr[KEY_PTRS(l)] = merge_chksums(l, r);
604 		else
605 			SET_KEY_CSUM(l, 0);
606 	}
607 
608 	SET_KEY_OFFSET(l, KEY_OFFSET(l) + KEY_SIZE(r));
609 	SET_KEY_SIZE(l, KEY_SIZE(l) + KEY_SIZE(r));
610 
611 	return true;
612 }
613 
614 const struct btree_keys_ops bch_extent_keys_ops = {
615 	.sort_cmp	= bch_extent_sort_cmp,
616 	.sort_fixup	= bch_extent_sort_fixup,
617 	.insert_fixup	= bch_extent_insert_fixup,
618 	.key_invalid	= bch_extent_invalid,
619 	.key_bad	= bch_extent_bad,
620 	.key_merge	= bch_extent_merge,
621 	.key_to_text	= bch_extent_to_text,
622 	.key_dump	= bch_bkey_dump,
623 	.is_extents	= true,
624 };
625