1 // SPDX-License-Identifier: GPL-2.0
5 #include "btree_journal_iter.h"
6 #include "journal_io.h"
8 #include <linux/sort.h>
11 * For managing keys we read from the journal: until journal replay works normal
12 * btree lookups need to be able to find and return keys from the journal where
13 * they overwrite what's in the btree, so we have a special iterator and
14 * operations for the regular btree iter code to use:
17 static int __journal_key_cmp(enum btree_id l_btree_id,
20 const struct journal_key *r)
22 return (cmp_int(l_btree_id, r->btree_id) ?:
23 cmp_int(l_level, r->level) ?:
24 bpos_cmp(l_pos, r->k->k.p));
27 static int journal_key_cmp(const struct journal_key *l, const struct journal_key *r)
29 return __journal_key_cmp(l->btree_id, l->level, l->k->k.p, r);
32 static inline size_t idx_to_pos(struct journal_keys *keys, size_t idx)
34 size_t gap_size = keys->size - keys->nr;
41 static inline struct journal_key *idx_to_key(struct journal_keys *keys, size_t idx)
43 return keys->d + idx_to_pos(keys, idx);
46 static size_t __bch2_journal_key_search(struct journal_keys *keys,
47 enum btree_id id, unsigned level,
50 size_t l = 0, r = keys->nr, m;
53 m = l + ((r - l) >> 1);
54 if (__journal_key_cmp(id, level, pos, idx_to_key(keys, m)) > 0)
60 BUG_ON(l < keys->nr &&
61 __journal_key_cmp(id, level, pos, idx_to_key(keys, l)) > 0);
64 __journal_key_cmp(id, level, pos, idx_to_key(keys, l - 1)) <= 0);
69 static size_t bch2_journal_key_search(struct journal_keys *keys,
70 enum btree_id id, unsigned level,
73 return idx_to_pos(keys, __bch2_journal_key_search(keys, id, level, pos));
76 struct bkey_i *bch2_journal_keys_peek_upto(struct bch_fs *c, enum btree_id btree_id,
77 unsigned level, struct bpos pos,
78 struct bpos end_pos, size_t *idx)
80 struct journal_keys *keys = &c->journal_keys;
82 struct journal_key *k;
85 *idx = __bch2_journal_key_search(keys, btree_id, level, pos);
87 while ((k = *idx < keys->nr ? idx_to_key(keys, *idx) : NULL)) {
88 if (__journal_key_cmp(btree_id, level, end_pos, k) < 0)
91 if (__journal_key_cmp(btree_id, level, pos, k) <= 0 &&
106 struct bkey_i *bch2_journal_keys_peek_slot(struct bch_fs *c, enum btree_id btree_id,
107 unsigned level, struct bpos pos)
111 return bch2_journal_keys_peek_upto(c, btree_id, level, pos, pos, &idx);
114 static void journal_iters_fix(struct bch_fs *c)
116 struct journal_keys *keys = &c->journal_keys;
117 /* The key we just inserted is immediately before the gap: */
118 size_t gap_end = keys->gap + (keys->size - keys->nr);
119 struct btree_and_journal_iter *iter;
122 * If an iterator points one after the key we just inserted, decrement
123 * the iterator so it points at the key we just inserted - if the
124 * decrement was unnecessary, bch2_btree_and_journal_iter_peek() will
127 list_for_each_entry(iter, &c->journal_iters, journal.list)
128 if (iter->journal.idx == gap_end)
129 iter->journal.idx = keys->gap - 1;
132 static void journal_iters_move_gap(struct bch_fs *c, size_t old_gap, size_t new_gap)
134 struct journal_keys *keys = &c->journal_keys;
135 struct journal_iter *iter;
136 size_t gap_size = keys->size - keys->nr;
138 list_for_each_entry(iter, &c->journal_iters, list) {
139 if (iter->idx > old_gap)
140 iter->idx -= gap_size;
141 if (iter->idx >= new_gap)
142 iter->idx += gap_size;
146 int bch2_journal_key_insert_take(struct bch_fs *c, enum btree_id id,
147 unsigned level, struct bkey_i *k)
149 struct journal_key n = {
155 * Ensure these keys are done last by journal replay, to unblock
158 .journal_seq = U32_MAX,
160 struct journal_keys *keys = &c->journal_keys;
161 size_t idx = bch2_journal_key_search(keys, id, level, k->k.p);
163 BUG_ON(test_bit(BCH_FS_RW, &c->flags));
165 if (idx < keys->size &&
166 journal_key_cmp(&n, &keys->d[idx]) == 0) {
167 if (keys->d[idx].allocated)
168 kfree(keys->d[idx].k);
174 idx -= keys->size - keys->nr;
176 if (keys->nr == keys->size) {
177 struct journal_keys new_keys = {
179 .size = max_t(size_t, keys->size, 8) * 2,
182 new_keys.d = kvmalloc_array(new_keys.size, sizeof(new_keys.d[0]), GFP_KERNEL);
184 bch_err(c, "%s: error allocating new key array (size %zu)",
185 __func__, new_keys.size);
186 return -BCH_ERR_ENOMEM_journal_key_insert;
189 /* Since @keys was full, there was no gap: */
190 memcpy(new_keys.d, keys->d, sizeof(keys->d[0]) * keys->nr);
194 /* And now the gap is at the end: */
195 keys->gap = keys->nr;
198 journal_iters_move_gap(c, keys->gap, idx);
200 move_gap(keys->d, keys->nr, keys->size, keys->gap, idx);
204 keys->d[keys->gap++] = n;
206 journal_iters_fix(c);
212 * Can only be used from the recovery thread while we're still RO - can't be
213 * used once we've got RW, as journal_keys is at that point used by multiple
216 int bch2_journal_key_insert(struct bch_fs *c, enum btree_id id,
217 unsigned level, struct bkey_i *k)
222 n = kmalloc(bkey_bytes(&k->k), GFP_KERNEL);
224 return -BCH_ERR_ENOMEM_journal_key_insert;
227 ret = bch2_journal_key_insert_take(c, id, level, n);
233 int bch2_journal_key_delete(struct bch_fs *c, enum btree_id id,
234 unsigned level, struct bpos pos)
236 struct bkey_i whiteout;
238 bkey_init(&whiteout.k);
241 return bch2_journal_key_insert(c, id, level, &whiteout);
244 void bch2_journal_key_overwritten(struct bch_fs *c, enum btree_id btree,
245 unsigned level, struct bpos pos)
247 struct journal_keys *keys = &c->journal_keys;
248 size_t idx = bch2_journal_key_search(keys, btree, level, pos);
250 if (idx < keys->size &&
251 keys->d[idx].btree_id == btree &&
252 keys->d[idx].level == level &&
253 bpos_eq(keys->d[idx].k->k.p, pos))
254 keys->d[idx].overwritten = true;
257 static void bch2_journal_iter_advance(struct journal_iter *iter)
259 if (iter->idx < iter->keys->size) {
261 if (iter->idx == iter->keys->gap)
262 iter->idx += iter->keys->size - iter->keys->nr;
266 static struct bkey_s_c bch2_journal_iter_peek(struct journal_iter *iter)
268 struct journal_key *k = iter->keys->d + iter->idx;
270 while (k < iter->keys->d + iter->keys->size &&
271 k->btree_id == iter->btree_id &&
272 k->level == iter->level) {
274 return bkey_i_to_s_c(k->k);
276 bch2_journal_iter_advance(iter);
277 k = iter->keys->d + iter->idx;
280 return bkey_s_c_null;
283 static void bch2_journal_iter_exit(struct journal_iter *iter)
285 list_del(&iter->list);
288 static void bch2_journal_iter_init(struct bch_fs *c,
289 struct journal_iter *iter,
290 enum btree_id id, unsigned level,
295 iter->keys = &c->journal_keys;
296 iter->idx = bch2_journal_key_search(&c->journal_keys, id, level, pos);
299 static struct bkey_s_c bch2_journal_iter_peek_btree(struct btree_and_journal_iter *iter)
301 return bch2_btree_node_iter_peek_unpack(&iter->node_iter,
302 iter->b, &iter->unpacked);
305 static void bch2_journal_iter_advance_btree(struct btree_and_journal_iter *iter)
307 bch2_btree_node_iter_advance(&iter->node_iter, iter->b);
310 void bch2_btree_and_journal_iter_advance(struct btree_and_journal_iter *iter)
312 if (bpos_eq(iter->pos, SPOS_MAX))
315 iter->pos = bpos_successor(iter->pos);
318 struct bkey_s_c bch2_btree_and_journal_iter_peek(struct btree_and_journal_iter *iter)
320 struct bkey_s_c btree_k, journal_k, ret;
323 return bkey_s_c_null;
325 while ((btree_k = bch2_journal_iter_peek_btree(iter)).k &&
326 bpos_lt(btree_k.k->p, iter->pos))
327 bch2_journal_iter_advance_btree(iter);
329 while ((journal_k = bch2_journal_iter_peek(&iter->journal)).k &&
330 bpos_lt(journal_k.k->p, iter->pos))
331 bch2_journal_iter_advance(&iter->journal);
334 (!btree_k.k || bpos_le(journal_k.k->p, btree_k.k->p))
338 if (ret.k && iter->b && bpos_gt(ret.k->p, iter->b->data->max_key))
342 iter->pos = ret.k->p;
343 if (bkey_deleted(ret.k)) {
344 bch2_btree_and_journal_iter_advance(iter);
348 iter->pos = SPOS_MAX;
355 void bch2_btree_and_journal_iter_exit(struct btree_and_journal_iter *iter)
357 bch2_journal_iter_exit(&iter->journal);
360 void __bch2_btree_and_journal_iter_init_node_iter(struct btree_and_journal_iter *iter,
363 struct btree_node_iter node_iter,
366 memset(iter, 0, sizeof(*iter));
369 iter->node_iter = node_iter;
370 bch2_journal_iter_init(c, &iter->journal, b->c.btree_id, b->c.level, pos);
371 INIT_LIST_HEAD(&iter->journal.list);
372 iter->pos = b->data->min_key;
373 iter->at_end = false;
377 * this version is used by btree_gc before filesystem has gone RW and
378 * multithreaded, so uses the journal_iters list:
380 void bch2_btree_and_journal_iter_init_node_iter(struct btree_and_journal_iter *iter,
384 struct btree_node_iter node_iter;
386 bch2_btree_node_iter_init_from_start(&node_iter, b);
387 __bch2_btree_and_journal_iter_init_node_iter(iter, c, b, node_iter, b->data->min_key);
388 list_add(&iter->journal.list, &c->journal_iters);
391 /* sort and dedup all keys in the journal: */
393 void bch2_journal_entries_free(struct bch_fs *c)
395 struct journal_replay **i;
396 struct genradix_iter iter;
398 genradix_for_each(&c->journal_entries, iter, i)
400 kvpfree(*i, offsetof(struct journal_replay, j) +
401 vstruct_bytes(&(*i)->j));
402 genradix_free(&c->journal_entries);
406 * When keys compare equal, oldest compares first:
408 static int journal_sort_key_cmp(const void *_l, const void *_r)
410 const struct journal_key *l = _l;
411 const struct journal_key *r = _r;
413 return journal_key_cmp(l, r) ?:
414 cmp_int(l->journal_seq, r->journal_seq) ?:
415 cmp_int(l->journal_offset, r->journal_offset);
418 void bch2_journal_keys_free(struct journal_keys *keys)
420 struct journal_key *i;
422 move_gap(keys->d, keys->nr, keys->size, keys->gap, keys->nr);
423 keys->gap = keys->nr;
425 for (i = keys->d; i < keys->d + keys->nr; i++)
431 keys->nr = keys->gap = keys->size = 0;
434 static void __journal_keys_sort(struct journal_keys *keys)
436 struct journal_key *src, *dst;
438 sort(keys->d, keys->nr, sizeof(keys->d[0]), journal_sort_key_cmp, NULL);
441 while (src < keys->d + keys->nr) {
442 while (src + 1 < keys->d + keys->nr &&
443 src[0].btree_id == src[1].btree_id &&
444 src[0].level == src[1].level &&
445 bpos_eq(src[0].k->k.p, src[1].k->k.p))
451 keys->nr = dst - keys->d;
454 int bch2_journal_keys_sort(struct bch_fs *c)
456 struct genradix_iter iter;
457 struct journal_replay *i, **_i;
458 struct jset_entry *entry;
460 struct journal_keys *keys = &c->journal_keys;
461 size_t nr_keys = 0, nr_read = 0;
463 genradix_for_each(&c->journal_entries, iter, _i) {
469 for_each_jset_key(k, entry, &i->j)
476 keys->size = roundup_pow_of_two(nr_keys);
478 keys->d = kvmalloc_array(keys->size, sizeof(keys->d[0]), GFP_KERNEL);
480 bch_err(c, "Failed to allocate buffer for sorted journal keys (%zu keys); trying slowpath",
485 keys->d = kvmalloc_array(keys->size, sizeof(keys->d[0]), GFP_KERNEL);
486 } while (!keys->d && keys->size > nr_keys / 8);
489 bch_err(c, "Failed to allocate %zu size buffer for sorted journal keys; exiting",
491 return -BCH_ERR_ENOMEM_journal_keys_sort;
495 genradix_for_each(&c->journal_entries, iter, _i) {
503 for_each_jset_key(k, entry, &i->j) {
504 if (keys->nr == keys->size) {
505 __journal_keys_sort(keys);
507 if (keys->nr > keys->size * 7 / 8) {
508 bch_err(c, "Too many journal keys for slowpath; have %zu compacted, buf size %zu, processed %zu/%zu",
509 keys->nr, keys->size, nr_read, nr_keys);
510 return -BCH_ERR_ENOMEM_journal_keys_sort;
514 keys->d[keys->nr++] = (struct journal_key) {
515 .btree_id = entry->btree_id,
516 .level = entry->level,
518 .journal_seq = le64_to_cpu(i->j.seq),
519 .journal_offset = k->_data - i->j._data,
526 __journal_keys_sort(keys);
527 keys->gap = keys->nr;
529 bch_verbose(c, "Journal keys: %zu read, %zu after sorting and compacting", nr_keys, keys->nr);