]>
Commit | Line | Data |
---|---|---|
56bec294 CM |
1 | /* |
2 | * Copyright (C) 2009 Oracle. All rights reserved. | |
3 | * | |
4 | * This program is free software; you can redistribute it and/or | |
5 | * modify it under the terms of the GNU General Public | |
6 | * License v2 as published by the Free Software Foundation. | |
7 | * | |
8 | * This program is distributed in the hope that it will be useful, | |
9 | * but WITHOUT ANY WARRANTY; without even the implied warranty of | |
10 | * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU | |
11 | * General Public License for more details. | |
12 | * | |
13 | * You should have received a copy of the GNU General Public | |
14 | * License along with this program; if not, write to the | |
15 | * Free Software Foundation, Inc., 59 Temple Place - Suite 330, | |
16 | * Boston, MA 021110-1307, USA. | |
17 | */ | |
18 | ||
19 | #include <linux/sched.h> | |
5a0e3ad6 | 20 | #include <linux/slab.h> |
56bec294 | 21 | #include <linux/sort.h> |
56bec294 CM |
22 | #include "ctree.h" |
23 | #include "delayed-ref.h" | |
24 | #include "transaction.h" | |
25 | ||
78a6184a MX |
26 | struct kmem_cache *btrfs_delayed_ref_head_cachep; |
27 | struct kmem_cache *btrfs_delayed_tree_ref_cachep; | |
28 | struct kmem_cache *btrfs_delayed_data_ref_cachep; | |
29 | struct kmem_cache *btrfs_delayed_extent_op_cachep; | |
56bec294 CM |
30 | /* |
31 | * delayed back reference update tracking. For subvolume trees | |
32 | * we queue up extent allocations and backref maintenance for | |
33 | * delayed processing. This avoids deep call chains where we | |
34 | * add extents in the middle of btrfs_search_slot, and it allows | |
35 | * us to buffer up frequently modified backrefs in an rb tree instead | |
36 | * of hammering updates on the extent allocation tree. | |
56bec294 CM |
37 | */ |
38 | ||
39 | /* | |
5d4f98a2 YZ |
40 | * compare two delayed tree backrefs with same bytenr and type |
41 | */ | |
42 | static int comp_tree_refs(struct btrfs_delayed_tree_ref *ref2, | |
41b0fc42 | 43 | struct btrfs_delayed_tree_ref *ref1, int type) |
5d4f98a2 | 44 | { |
41b0fc42 JB |
45 | if (type == BTRFS_TREE_BLOCK_REF_KEY) { |
46 | if (ref1->root < ref2->root) | |
47 | return -1; | |
48 | if (ref1->root > ref2->root) | |
49 | return 1; | |
50 | } else { | |
51 | if (ref1->parent < ref2->parent) | |
52 | return -1; | |
53 | if (ref1->parent > ref2->parent) | |
54 | return 1; | |
55 | } | |
5d4f98a2 YZ |
56 | return 0; |
57 | } | |
58 | ||
59 | /* | |
60 | * compare two delayed data backrefs with same bytenr and type | |
56bec294 | 61 | */ |
5d4f98a2 YZ |
62 | static int comp_data_refs(struct btrfs_delayed_data_ref *ref2, |
63 | struct btrfs_delayed_data_ref *ref1) | |
56bec294 | 64 | { |
5d4f98a2 YZ |
65 | if (ref1->node.type == BTRFS_EXTENT_DATA_REF_KEY) { |
66 | if (ref1->root < ref2->root) | |
67 | return -1; | |
68 | if (ref1->root > ref2->root) | |
69 | return 1; | |
70 | if (ref1->objectid < ref2->objectid) | |
71 | return -1; | |
72 | if (ref1->objectid > ref2->objectid) | |
73 | return 1; | |
74 | if (ref1->offset < ref2->offset) | |
75 | return -1; | |
76 | if (ref1->offset > ref2->offset) | |
77 | return 1; | |
78 | } else { | |
79 | if (ref1->parent < ref2->parent) | |
80 | return -1; | |
81 | if (ref1->parent > ref2->parent) | |
82 | return 1; | |
83 | } | |
84 | return 0; | |
85 | } | |
86 | ||
87 | /* | |
88 | * entries in the rb tree are ordered by the byte number of the extent, | |
89 | * type of the delayed backrefs and content of delayed backrefs. | |
90 | */ | |
91 | static int comp_entry(struct btrfs_delayed_ref_node *ref2, | |
ae1e206b JB |
92 | struct btrfs_delayed_ref_node *ref1, |
93 | bool compare_seq) | |
5d4f98a2 YZ |
94 | { |
95 | if (ref1->bytenr < ref2->bytenr) | |
56bec294 | 96 | return -1; |
5d4f98a2 | 97 | if (ref1->bytenr > ref2->bytenr) |
56bec294 | 98 | return 1; |
5d4f98a2 YZ |
99 | if (ref1->is_head && ref2->is_head) |
100 | return 0; | |
101 | if (ref2->is_head) | |
56bec294 | 102 | return -1; |
5d4f98a2 | 103 | if (ref1->is_head) |
56bec294 | 104 | return 1; |
5d4f98a2 YZ |
105 | if (ref1->type < ref2->type) |
106 | return -1; | |
107 | if (ref1->type > ref2->type) | |
108 | return 1; | |
00f04b88 | 109 | /* merging of sequenced refs is not allowed */ |
ae1e206b JB |
110 | if (compare_seq) { |
111 | if (ref1->seq < ref2->seq) | |
112 | return -1; | |
113 | if (ref1->seq > ref2->seq) | |
114 | return 1; | |
115 | } | |
5d4f98a2 YZ |
116 | if (ref1->type == BTRFS_TREE_BLOCK_REF_KEY || |
117 | ref1->type == BTRFS_SHARED_BLOCK_REF_KEY) { | |
118 | return comp_tree_refs(btrfs_delayed_node_to_tree_ref(ref2), | |
41b0fc42 JB |
119 | btrfs_delayed_node_to_tree_ref(ref1), |
120 | ref1->type); | |
5d4f98a2 YZ |
121 | } else if (ref1->type == BTRFS_EXTENT_DATA_REF_KEY || |
122 | ref1->type == BTRFS_SHARED_DATA_REF_KEY) { | |
123 | return comp_data_refs(btrfs_delayed_node_to_data_ref(ref2), | |
124 | btrfs_delayed_node_to_data_ref(ref1)); | |
125 | } | |
126 | BUG(); | |
56bec294 CM |
127 | return 0; |
128 | } | |
129 | ||
130 | /* | |
131 | * insert a new ref into the rbtree. This returns any existing refs | |
132 | * for the same (bytenr,parent) tuple, or NULL if the new node was properly | |
133 | * inserted. | |
134 | */ | |
135 | static struct btrfs_delayed_ref_node *tree_insert(struct rb_root *root, | |
56bec294 CM |
136 | struct rb_node *node) |
137 | { | |
138 | struct rb_node **p = &root->rb_node; | |
139 | struct rb_node *parent_node = NULL; | |
140 | struct btrfs_delayed_ref_node *entry; | |
5d4f98a2 | 141 | struct btrfs_delayed_ref_node *ins; |
56bec294 CM |
142 | int cmp; |
143 | ||
5d4f98a2 | 144 | ins = rb_entry(node, struct btrfs_delayed_ref_node, rb_node); |
56bec294 CM |
145 | while (*p) { |
146 | parent_node = *p; | |
147 | entry = rb_entry(parent_node, struct btrfs_delayed_ref_node, | |
148 | rb_node); | |
149 | ||
ae1e206b | 150 | cmp = comp_entry(entry, ins, 1); |
56bec294 CM |
151 | if (cmp < 0) |
152 | p = &(*p)->rb_left; | |
153 | else if (cmp > 0) | |
154 | p = &(*p)->rb_right; | |
155 | else | |
156 | return entry; | |
157 | } | |
158 | ||
56bec294 CM |
159 | rb_link_node(node, parent_node, p); |
160 | rb_insert_color(node, root); | |
161 | return NULL; | |
162 | } | |
163 | ||
164 | /* | |
5d4f98a2 | 165 | * find an head entry based on bytenr. This returns the delayed ref |
d1270cd9 AJ |
166 | * head if it was able to find one, or NULL if nothing was in that spot. |
167 | * If return_bigger is given, the next bigger entry is returned if no exact | |
168 | * match is found. | |
56bec294 | 169 | */ |
5d4f98a2 YZ |
170 | static struct btrfs_delayed_ref_node *find_ref_head(struct rb_root *root, |
171 | u64 bytenr, | |
d1270cd9 AJ |
172 | struct btrfs_delayed_ref_node **last, |
173 | int return_bigger) | |
56bec294 | 174 | { |
d1270cd9 | 175 | struct rb_node *n; |
56bec294 | 176 | struct btrfs_delayed_ref_node *entry; |
d1270cd9 | 177 | int cmp = 0; |
56bec294 | 178 | |
d1270cd9 AJ |
179 | again: |
180 | n = root->rb_node; | |
181 | entry = NULL; | |
56bec294 CM |
182 | while (n) { |
183 | entry = rb_entry(n, struct btrfs_delayed_ref_node, rb_node); | |
184 | WARN_ON(!entry->in_tree); | |
c3e69d58 CM |
185 | if (last) |
186 | *last = entry; | |
56bec294 | 187 | |
5d4f98a2 YZ |
188 | if (bytenr < entry->bytenr) |
189 | cmp = -1; | |
190 | else if (bytenr > entry->bytenr) | |
191 | cmp = 1; | |
192 | else if (!btrfs_delayed_ref_is_head(entry)) | |
193 | cmp = 1; | |
194 | else | |
195 | cmp = 0; | |
196 | ||
56bec294 CM |
197 | if (cmp < 0) |
198 | n = n->rb_left; | |
199 | else if (cmp > 0) | |
200 | n = n->rb_right; | |
201 | else | |
202 | return entry; | |
203 | } | |
d1270cd9 AJ |
204 | if (entry && return_bigger) { |
205 | if (cmp > 0) { | |
206 | n = rb_next(&entry->rb_node); | |
207 | if (!n) | |
208 | n = rb_first(root); | |
209 | entry = rb_entry(n, struct btrfs_delayed_ref_node, | |
210 | rb_node); | |
211 | bytenr = entry->bytenr; | |
212 | return_bigger = 0; | |
213 | goto again; | |
214 | } | |
215 | return entry; | |
216 | } | |
56bec294 CM |
217 | return NULL; |
218 | } | |
219 | ||
c3e69d58 CM |
220 | int btrfs_delayed_ref_lock(struct btrfs_trans_handle *trans, |
221 | struct btrfs_delayed_ref_head *head) | |
56bec294 | 222 | { |
c3e69d58 CM |
223 | struct btrfs_delayed_ref_root *delayed_refs; |
224 | ||
225 | delayed_refs = &trans->transaction->delayed_refs; | |
226 | assert_spin_locked(&delayed_refs->lock); | |
227 | if (mutex_trylock(&head->mutex)) | |
228 | return 0; | |
229 | ||
230 | atomic_inc(&head->node.refs); | |
231 | spin_unlock(&delayed_refs->lock); | |
232 | ||
233 | mutex_lock(&head->mutex); | |
234 | spin_lock(&delayed_refs->lock); | |
235 | if (!head->node.in_tree) { | |
236 | mutex_unlock(&head->mutex); | |
237 | btrfs_put_delayed_ref(&head->node); | |
238 | return -EAGAIN; | |
239 | } | |
240 | btrfs_put_delayed_ref(&head->node); | |
241 | return 0; | |
242 | } | |
243 | ||
35a3621b | 244 | static inline void drop_delayed_ref(struct btrfs_trans_handle *trans, |
ae1e206b JB |
245 | struct btrfs_delayed_ref_root *delayed_refs, |
246 | struct btrfs_delayed_ref_node *ref) | |
247 | { | |
248 | rb_erase(&ref->rb_node, &delayed_refs->root); | |
249 | ref->in_tree = 0; | |
250 | btrfs_put_delayed_ref(ref); | |
251 | delayed_refs->num_entries--; | |
252 | if (trans->delayed_ref_updates) | |
253 | trans->delayed_ref_updates--; | |
254 | } | |
255 | ||
256 | static int merge_ref(struct btrfs_trans_handle *trans, | |
257 | struct btrfs_delayed_ref_root *delayed_refs, | |
258 | struct btrfs_delayed_ref_node *ref, u64 seq) | |
259 | { | |
260 | struct rb_node *node; | |
261 | int merged = 0; | |
262 | int mod = 0; | |
263 | int done = 0; | |
264 | ||
265 | node = rb_prev(&ref->rb_node); | |
266 | while (node) { | |
267 | struct btrfs_delayed_ref_node *next; | |
268 | ||
269 | next = rb_entry(node, struct btrfs_delayed_ref_node, rb_node); | |
270 | node = rb_prev(node); | |
271 | if (next->bytenr != ref->bytenr) | |
272 | break; | |
273 | if (seq && next->seq >= seq) | |
274 | break; | |
275 | if (comp_entry(ref, next, 0)) | |
276 | continue; | |
277 | ||
278 | if (ref->action == next->action) { | |
279 | mod = next->ref_mod; | |
280 | } else { | |
281 | if (ref->ref_mod < next->ref_mod) { | |
282 | struct btrfs_delayed_ref_node *tmp; | |
283 | ||
284 | tmp = ref; | |
285 | ref = next; | |
286 | next = tmp; | |
287 | done = 1; | |
288 | } | |
289 | mod = -next->ref_mod; | |
290 | } | |
291 | ||
292 | merged++; | |
293 | drop_delayed_ref(trans, delayed_refs, next); | |
294 | ref->ref_mod += mod; | |
295 | if (ref->ref_mod == 0) { | |
296 | drop_delayed_ref(trans, delayed_refs, ref); | |
297 | break; | |
298 | } else { | |
299 | /* | |
300 | * You can't have multiples of the same ref on a tree | |
301 | * block. | |
302 | */ | |
303 | WARN_ON(ref->type == BTRFS_TREE_BLOCK_REF_KEY || | |
304 | ref->type == BTRFS_SHARED_BLOCK_REF_KEY); | |
305 | } | |
306 | ||
307 | if (done) | |
308 | break; | |
309 | node = rb_prev(&ref->rb_node); | |
310 | } | |
311 | ||
312 | return merged; | |
313 | } | |
314 | ||
315 | void btrfs_merge_delayed_refs(struct btrfs_trans_handle *trans, | |
316 | struct btrfs_fs_info *fs_info, | |
317 | struct btrfs_delayed_ref_root *delayed_refs, | |
318 | struct btrfs_delayed_ref_head *head) | |
319 | { | |
320 | struct rb_node *node; | |
321 | u64 seq = 0; | |
322 | ||
323 | spin_lock(&fs_info->tree_mod_seq_lock); | |
324 | if (!list_empty(&fs_info->tree_mod_seq_list)) { | |
325 | struct seq_list *elem; | |
326 | ||
327 | elem = list_first_entry(&fs_info->tree_mod_seq_list, | |
328 | struct seq_list, list); | |
329 | seq = elem->seq; | |
330 | } | |
331 | spin_unlock(&fs_info->tree_mod_seq_lock); | |
332 | ||
333 | node = rb_prev(&head->node.rb_node); | |
334 | while (node) { | |
335 | struct btrfs_delayed_ref_node *ref; | |
336 | ||
337 | ref = rb_entry(node, struct btrfs_delayed_ref_node, | |
338 | rb_node); | |
339 | if (ref->bytenr != head->node.bytenr) | |
340 | break; | |
341 | ||
342 | /* We can't merge refs that are outside of our seq count */ | |
343 | if (seq && ref->seq >= seq) | |
344 | break; | |
345 | if (merge_ref(trans, delayed_refs, ref, seq)) | |
346 | node = rb_prev(&head->node.rb_node); | |
347 | else | |
348 | node = rb_prev(node); | |
349 | } | |
350 | } | |
351 | ||
097b8a7c JS |
352 | int btrfs_check_delayed_seq(struct btrfs_fs_info *fs_info, |
353 | struct btrfs_delayed_ref_root *delayed_refs, | |
00f04b88 AJ |
354 | u64 seq) |
355 | { | |
356 | struct seq_list *elem; | |
097b8a7c JS |
357 | int ret = 0; |
358 | ||
359 | spin_lock(&fs_info->tree_mod_seq_lock); | |
360 | if (!list_empty(&fs_info->tree_mod_seq_list)) { | |
361 | elem = list_first_entry(&fs_info->tree_mod_seq_list, | |
362 | struct seq_list, list); | |
363 | if (seq >= elem->seq) { | |
fc36ed7e JS |
364 | pr_debug("holding back delayed_ref %#x.%x, lowest is %#x.%x (%p)\n", |
365 | (u32)(seq >> 32), (u32)seq, | |
366 | (u32)(elem->seq >> 32), (u32)elem->seq, | |
367 | delayed_refs); | |
097b8a7c JS |
368 | ret = 1; |
369 | } | |
00f04b88 | 370 | } |
097b8a7c JS |
371 | |
372 | spin_unlock(&fs_info->tree_mod_seq_lock); | |
373 | return ret; | |
00f04b88 AJ |
374 | } |
375 | ||
c3e69d58 CM |
376 | int btrfs_find_ref_cluster(struct btrfs_trans_handle *trans, |
377 | struct list_head *cluster, u64 start) | |
378 | { | |
379 | int count = 0; | |
380 | struct btrfs_delayed_ref_root *delayed_refs; | |
56bec294 | 381 | struct rb_node *node; |
c3e69d58 | 382 | struct btrfs_delayed_ref_node *ref; |
56bec294 | 383 | struct btrfs_delayed_ref_head *head; |
56bec294 | 384 | |
c3e69d58 CM |
385 | delayed_refs = &trans->transaction->delayed_refs; |
386 | if (start == 0) { | |
387 | node = rb_first(&delayed_refs->root); | |
388 | } else { | |
389 | ref = NULL; | |
d1270cd9 | 390 | find_ref_head(&delayed_refs->root, start + 1, &ref, 1); |
c3e69d58 | 391 | if (ref) { |
c3e69d58 CM |
392 | node = &ref->rb_node; |
393 | } else | |
394 | node = rb_first(&delayed_refs->root); | |
395 | } | |
396 | again: | |
397 | while (node && count < 32) { | |
398 | ref = rb_entry(node, struct btrfs_delayed_ref_node, rb_node); | |
56bec294 CM |
399 | if (btrfs_delayed_ref_is_head(ref)) { |
400 | head = btrfs_delayed_node_to_head(ref); | |
c3e69d58 CM |
401 | if (list_empty(&head->cluster)) { |
402 | list_add_tail(&head->cluster, cluster); | |
403 | delayed_refs->run_delayed_start = | |
404 | head->node.bytenr; | |
405 | count++; | |
406 | ||
407 | WARN_ON(delayed_refs->num_heads_ready == 0); | |
408 | delayed_refs->num_heads_ready--; | |
409 | } else if (count) { | |
410 | /* the goal of the clustering is to find extents | |
411 | * that are likely to end up in the same extent | |
412 | * leaf on disk. So, we don't want them spread | |
413 | * all over the tree. Stop now if we've hit | |
414 | * a head that was already in use | |
415 | */ | |
56bec294 CM |
416 | break; |
417 | } | |
418 | } | |
c3e69d58 | 419 | node = rb_next(node); |
56bec294 | 420 | } |
c3e69d58 CM |
421 | if (count) { |
422 | return 0; | |
423 | } else if (start) { | |
424 | /* | |
425 | * we've gone to the end of the rbtree without finding any | |
426 | * clusters. start from the beginning and try again | |
427 | */ | |
428 | start = 0; | |
429 | node = rb_first(&delayed_refs->root); | |
430 | goto again; | |
431 | } | |
432 | return 1; | |
56bec294 CM |
433 | } |
434 | ||
093486c4 MX |
435 | void btrfs_release_ref_cluster(struct list_head *cluster) |
436 | { | |
437 | struct list_head *pos, *q; | |
438 | ||
439 | list_for_each_safe(pos, q, cluster) | |
440 | list_del_init(pos); | |
441 | } | |
442 | ||
56bec294 CM |
443 | /* |
444 | * helper function to update an extent delayed ref in the | |
445 | * rbtree. existing and update must both have the same | |
446 | * bytenr and parent | |
447 | * | |
448 | * This may free existing if the update cancels out whatever | |
449 | * operation it was doing. | |
450 | */ | |
451 | static noinline void | |
452 | update_existing_ref(struct btrfs_trans_handle *trans, | |
453 | struct btrfs_delayed_ref_root *delayed_refs, | |
454 | struct btrfs_delayed_ref_node *existing, | |
455 | struct btrfs_delayed_ref_node *update) | |
456 | { | |
5d4f98a2 | 457 | if (update->action != existing->action) { |
56bec294 CM |
458 | /* |
459 | * this is effectively undoing either an add or a | |
460 | * drop. We decrement the ref_mod, and if it goes | |
461 | * down to zero we just delete the entry without | |
462 | * every changing the extent allocation tree. | |
463 | */ | |
464 | existing->ref_mod--; | |
ae1e206b JB |
465 | if (existing->ref_mod == 0) |
466 | drop_delayed_ref(trans, delayed_refs, existing); | |
467 | else | |
5d4f98a2 YZ |
468 | WARN_ON(existing->type == BTRFS_TREE_BLOCK_REF_KEY || |
469 | existing->type == BTRFS_SHARED_BLOCK_REF_KEY); | |
56bec294 | 470 | } else { |
5d4f98a2 YZ |
471 | WARN_ON(existing->type == BTRFS_TREE_BLOCK_REF_KEY || |
472 | existing->type == BTRFS_SHARED_BLOCK_REF_KEY); | |
56bec294 CM |
473 | /* |
474 | * the action on the existing ref matches | |
475 | * the action on the ref we're trying to add. | |
476 | * Bump the ref_mod by one so the backref that | |
477 | * is eventually added/removed has the correct | |
478 | * reference count | |
479 | */ | |
480 | existing->ref_mod += update->ref_mod; | |
481 | } | |
482 | } | |
483 | ||
484 | /* | |
485 | * helper function to update the accounting in the head ref | |
486 | * existing and update must have the same bytenr | |
487 | */ | |
488 | static noinline void | |
489 | update_existing_head_ref(struct btrfs_delayed_ref_node *existing, | |
490 | struct btrfs_delayed_ref_node *update) | |
491 | { | |
492 | struct btrfs_delayed_ref_head *existing_ref; | |
493 | struct btrfs_delayed_ref_head *ref; | |
494 | ||
495 | existing_ref = btrfs_delayed_node_to_head(existing); | |
496 | ref = btrfs_delayed_node_to_head(update); | |
5d4f98a2 | 497 | BUG_ON(existing_ref->is_data != ref->is_data); |
56bec294 CM |
498 | |
499 | if (ref->must_insert_reserved) { | |
500 | /* if the extent was freed and then | |
501 | * reallocated before the delayed ref | |
502 | * entries were processed, we can end up | |
503 | * with an existing head ref without | |
504 | * the must_insert_reserved flag set. | |
505 | * Set it again here | |
506 | */ | |
507 | existing_ref->must_insert_reserved = ref->must_insert_reserved; | |
508 | ||
509 | /* | |
510 | * update the num_bytes so we make sure the accounting | |
511 | * is done correctly | |
512 | */ | |
513 | existing->num_bytes = update->num_bytes; | |
514 | ||
515 | } | |
516 | ||
5d4f98a2 YZ |
517 | if (ref->extent_op) { |
518 | if (!existing_ref->extent_op) { | |
519 | existing_ref->extent_op = ref->extent_op; | |
520 | } else { | |
521 | if (ref->extent_op->update_key) { | |
522 | memcpy(&existing_ref->extent_op->key, | |
523 | &ref->extent_op->key, | |
524 | sizeof(ref->extent_op->key)); | |
525 | existing_ref->extent_op->update_key = 1; | |
526 | } | |
527 | if (ref->extent_op->update_flags) { | |
528 | existing_ref->extent_op->flags_to_set |= | |
529 | ref->extent_op->flags_to_set; | |
530 | existing_ref->extent_op->update_flags = 1; | |
531 | } | |
78a6184a | 532 | btrfs_free_delayed_extent_op(ref->extent_op); |
5d4f98a2 YZ |
533 | } |
534 | } | |
56bec294 CM |
535 | /* |
536 | * update the reference mod on the head to reflect this new operation | |
537 | */ | |
538 | existing->ref_mod += update->ref_mod; | |
539 | } | |
540 | ||
541 | /* | |
5d4f98a2 | 542 | * helper function to actually insert a head node into the rbtree. |
56bec294 | 543 | * this does all the dirty work in terms of maintaining the correct |
5d4f98a2 | 544 | * overall modification count. |
56bec294 | 545 | */ |
143bede5 | 546 | static noinline void add_delayed_ref_head(struct btrfs_fs_info *fs_info, |
66d7e7f0 | 547 | struct btrfs_trans_handle *trans, |
5d4f98a2 YZ |
548 | struct btrfs_delayed_ref_node *ref, |
549 | u64 bytenr, u64 num_bytes, | |
550 | int action, int is_data) | |
56bec294 CM |
551 | { |
552 | struct btrfs_delayed_ref_node *existing; | |
c3e69d58 | 553 | struct btrfs_delayed_ref_head *head_ref = NULL; |
56bec294 CM |
554 | struct btrfs_delayed_ref_root *delayed_refs; |
555 | int count_mod = 1; | |
556 | int must_insert_reserved = 0; | |
557 | ||
558 | /* | |
559 | * the head node stores the sum of all the mods, so dropping a ref | |
560 | * should drop the sum in the head node by one. | |
561 | */ | |
5d4f98a2 YZ |
562 | if (action == BTRFS_UPDATE_DELAYED_HEAD) |
563 | count_mod = 0; | |
564 | else if (action == BTRFS_DROP_DELAYED_REF) | |
565 | count_mod = -1; | |
56bec294 CM |
566 | |
567 | /* | |
568 | * BTRFS_ADD_DELAYED_EXTENT means that we need to update | |
569 | * the reserved accounting when the extent is finally added, or | |
570 | * if a later modification deletes the delayed ref without ever | |
571 | * inserting the extent into the extent allocation tree. | |
572 | * ref->must_insert_reserved is the flag used to record | |
573 | * that accounting mods are required. | |
574 | * | |
575 | * Once we record must_insert_reserved, switch the action to | |
576 | * BTRFS_ADD_DELAYED_REF because other special casing is not required. | |
577 | */ | |
5d4f98a2 | 578 | if (action == BTRFS_ADD_DELAYED_EXTENT) |
56bec294 | 579 | must_insert_reserved = 1; |
5d4f98a2 | 580 | else |
56bec294 | 581 | must_insert_reserved = 0; |
56bec294 CM |
582 | |
583 | delayed_refs = &trans->transaction->delayed_refs; | |
584 | ||
585 | /* first set the basic ref node struct up */ | |
586 | atomic_set(&ref->refs, 1); | |
587 | ref->bytenr = bytenr; | |
5d4f98a2 | 588 | ref->num_bytes = num_bytes; |
56bec294 | 589 | ref->ref_mod = count_mod; |
5d4f98a2 YZ |
590 | ref->type = 0; |
591 | ref->action = 0; | |
592 | ref->is_head = 1; | |
56bec294 | 593 | ref->in_tree = 1; |
00f04b88 | 594 | ref->seq = 0; |
5d4f98a2 YZ |
595 | |
596 | head_ref = btrfs_delayed_node_to_head(ref); | |
597 | head_ref->must_insert_reserved = must_insert_reserved; | |
598 | head_ref->is_data = is_data; | |
599 | ||
600 | INIT_LIST_HEAD(&head_ref->cluster); | |
601 | mutex_init(&head_ref->mutex); | |
602 | ||
599c75ec | 603 | trace_add_delayed_ref_head(ref, head_ref, action); |
1abe9b8a | 604 | |
5d4f98a2 YZ |
605 | existing = tree_insert(&delayed_refs->root, &ref->rb_node); |
606 | ||
607 | if (existing) { | |
608 | update_existing_head_ref(existing, ref); | |
609 | /* | |
610 | * we've updated the existing ref, free the newly | |
611 | * allocated ref | |
612 | */ | |
78a6184a | 613 | kmem_cache_free(btrfs_delayed_ref_head_cachep, head_ref); |
5d4f98a2 YZ |
614 | } else { |
615 | delayed_refs->num_heads++; | |
616 | delayed_refs->num_heads_ready++; | |
617 | delayed_refs->num_entries++; | |
618 | trans->delayed_ref_updates++; | |
619 | } | |
5d4f98a2 YZ |
620 | } |
621 | ||
622 | /* | |
623 | * helper to insert a delayed tree ref into the rbtree. | |
624 | */ | |
143bede5 | 625 | static noinline void add_delayed_tree_ref(struct btrfs_fs_info *fs_info, |
66d7e7f0 | 626 | struct btrfs_trans_handle *trans, |
5d4f98a2 YZ |
627 | struct btrfs_delayed_ref_node *ref, |
628 | u64 bytenr, u64 num_bytes, u64 parent, | |
66d7e7f0 AJ |
629 | u64 ref_root, int level, int action, |
630 | int for_cow) | |
5d4f98a2 YZ |
631 | { |
632 | struct btrfs_delayed_ref_node *existing; | |
633 | struct btrfs_delayed_tree_ref *full_ref; | |
634 | struct btrfs_delayed_ref_root *delayed_refs; | |
00f04b88 | 635 | u64 seq = 0; |
5d4f98a2 YZ |
636 | |
637 | if (action == BTRFS_ADD_DELAYED_EXTENT) | |
638 | action = BTRFS_ADD_DELAYED_REF; | |
639 | ||
640 | delayed_refs = &trans->transaction->delayed_refs; | |
641 | ||
642 | /* first set the basic ref node struct up */ | |
643 | atomic_set(&ref->refs, 1); | |
644 | ref->bytenr = bytenr; | |
56bec294 | 645 | ref->num_bytes = num_bytes; |
5d4f98a2 YZ |
646 | ref->ref_mod = 1; |
647 | ref->action = action; | |
648 | ref->is_head = 0; | |
649 | ref->in_tree = 1; | |
56bec294 | 650 | |
546adb0d JS |
651 | if (need_ref_seq(for_cow, ref_root)) |
652 | seq = btrfs_get_tree_mod_seq(fs_info, &trans->delayed_ref_elem); | |
00f04b88 AJ |
653 | ref->seq = seq; |
654 | ||
5d4f98a2 | 655 | full_ref = btrfs_delayed_node_to_tree_ref(ref); |
eebe063b AJ |
656 | full_ref->parent = parent; |
657 | full_ref->root = ref_root; | |
658 | if (parent) | |
5d4f98a2 | 659 | ref->type = BTRFS_SHARED_BLOCK_REF_KEY; |
eebe063b | 660 | else |
5d4f98a2 | 661 | ref->type = BTRFS_TREE_BLOCK_REF_KEY; |
5d4f98a2 | 662 | full_ref->level = level; |
56bec294 | 663 | |
599c75ec | 664 | trace_add_delayed_tree_ref(ref, full_ref, action); |
1abe9b8a | 665 | |
5d4f98a2 | 666 | existing = tree_insert(&delayed_refs->root, &ref->rb_node); |
56bec294 CM |
667 | |
668 | if (existing) { | |
5d4f98a2 YZ |
669 | update_existing_ref(trans, delayed_refs, existing, ref); |
670 | /* | |
671 | * we've updated the existing ref, free the newly | |
672 | * allocated ref | |
673 | */ | |
78a6184a | 674 | kmem_cache_free(btrfs_delayed_tree_ref_cachep, full_ref); |
5d4f98a2 YZ |
675 | } else { |
676 | delayed_refs->num_entries++; | |
677 | trans->delayed_ref_updates++; | |
678 | } | |
5d4f98a2 YZ |
679 | } |
680 | ||
681 | /* | |
682 | * helper to insert a delayed data ref into the rbtree. | |
683 | */ | |
143bede5 | 684 | static noinline void add_delayed_data_ref(struct btrfs_fs_info *fs_info, |
66d7e7f0 | 685 | struct btrfs_trans_handle *trans, |
5d4f98a2 YZ |
686 | struct btrfs_delayed_ref_node *ref, |
687 | u64 bytenr, u64 num_bytes, u64 parent, | |
688 | u64 ref_root, u64 owner, u64 offset, | |
66d7e7f0 | 689 | int action, int for_cow) |
5d4f98a2 YZ |
690 | { |
691 | struct btrfs_delayed_ref_node *existing; | |
692 | struct btrfs_delayed_data_ref *full_ref; | |
693 | struct btrfs_delayed_ref_root *delayed_refs; | |
00f04b88 | 694 | u64 seq = 0; |
5d4f98a2 YZ |
695 | |
696 | if (action == BTRFS_ADD_DELAYED_EXTENT) | |
697 | action = BTRFS_ADD_DELAYED_REF; | |
698 | ||
699 | delayed_refs = &trans->transaction->delayed_refs; | |
700 | ||
701 | /* first set the basic ref node struct up */ | |
702 | atomic_set(&ref->refs, 1); | |
703 | ref->bytenr = bytenr; | |
704 | ref->num_bytes = num_bytes; | |
705 | ref->ref_mod = 1; | |
706 | ref->action = action; | |
707 | ref->is_head = 0; | |
708 | ref->in_tree = 1; | |
709 | ||
546adb0d JS |
710 | if (need_ref_seq(for_cow, ref_root)) |
711 | seq = btrfs_get_tree_mod_seq(fs_info, &trans->delayed_ref_elem); | |
00f04b88 AJ |
712 | ref->seq = seq; |
713 | ||
5d4f98a2 | 714 | full_ref = btrfs_delayed_node_to_data_ref(ref); |
eebe063b AJ |
715 | full_ref->parent = parent; |
716 | full_ref->root = ref_root; | |
717 | if (parent) | |
5d4f98a2 | 718 | ref->type = BTRFS_SHARED_DATA_REF_KEY; |
eebe063b | 719 | else |
5d4f98a2 | 720 | ref->type = BTRFS_EXTENT_DATA_REF_KEY; |
66d7e7f0 | 721 | |
5d4f98a2 YZ |
722 | full_ref->objectid = owner; |
723 | full_ref->offset = offset; | |
56bec294 | 724 | |
599c75ec | 725 | trace_add_delayed_data_ref(ref, full_ref, action); |
1abe9b8a | 726 | |
5d4f98a2 YZ |
727 | existing = tree_insert(&delayed_refs->root, &ref->rb_node); |
728 | ||
729 | if (existing) { | |
730 | update_existing_ref(trans, delayed_refs, existing, ref); | |
56bec294 CM |
731 | /* |
732 | * we've updated the existing ref, free the newly | |
733 | * allocated ref | |
734 | */ | |
78a6184a | 735 | kmem_cache_free(btrfs_delayed_data_ref_cachep, full_ref); |
56bec294 CM |
736 | } else { |
737 | delayed_refs->num_entries++; | |
738 | trans->delayed_ref_updates++; | |
739 | } | |
56bec294 CM |
740 | } |
741 | ||
742 | /* | |
5d4f98a2 | 743 | * add a delayed tree ref. This does all of the accounting required |
56bec294 CM |
744 | * to make sure the delayed ref is eventually processed before this |
745 | * transaction commits. | |
746 | */ | |
66d7e7f0 AJ |
747 | int btrfs_add_delayed_tree_ref(struct btrfs_fs_info *fs_info, |
748 | struct btrfs_trans_handle *trans, | |
5d4f98a2 YZ |
749 | u64 bytenr, u64 num_bytes, u64 parent, |
750 | u64 ref_root, int level, int action, | |
66d7e7f0 AJ |
751 | struct btrfs_delayed_extent_op *extent_op, |
752 | int for_cow) | |
56bec294 | 753 | { |
5d4f98a2 | 754 | struct btrfs_delayed_tree_ref *ref; |
56bec294 CM |
755 | struct btrfs_delayed_ref_head *head_ref; |
756 | struct btrfs_delayed_ref_root *delayed_refs; | |
56bec294 | 757 | |
5d4f98a2 | 758 | BUG_ON(extent_op && extent_op->is_data); |
78a6184a | 759 | ref = kmem_cache_alloc(btrfs_delayed_tree_ref_cachep, GFP_NOFS); |
56bec294 CM |
760 | if (!ref) |
761 | return -ENOMEM; | |
762 | ||
78a6184a | 763 | head_ref = kmem_cache_alloc(btrfs_delayed_ref_head_cachep, GFP_NOFS); |
5d4f98a2 | 764 | if (!head_ref) { |
78a6184a | 765 | kmem_cache_free(btrfs_delayed_tree_ref_cachep, ref); |
5d4f98a2 YZ |
766 | return -ENOMEM; |
767 | } | |
768 | ||
769 | head_ref->extent_op = extent_op; | |
770 | ||
771 | delayed_refs = &trans->transaction->delayed_refs; | |
772 | spin_lock(&delayed_refs->lock); | |
773 | ||
56bec294 | 774 | /* |
5d4f98a2 YZ |
775 | * insert both the head node and the new ref without dropping |
776 | * the spin lock | |
56bec294 | 777 | */ |
143bede5 | 778 | add_delayed_ref_head(fs_info, trans, &head_ref->node, bytenr, |
66d7e7f0 | 779 | num_bytes, action, 0); |
5d4f98a2 | 780 | |
143bede5 | 781 | add_delayed_tree_ref(fs_info, trans, &ref->node, bytenr, |
66d7e7f0 AJ |
782 | num_bytes, parent, ref_root, level, action, |
783 | for_cow); | |
5d4f98a2 | 784 | spin_unlock(&delayed_refs->lock); |
546adb0d JS |
785 | if (need_ref_seq(for_cow, ref_root)) |
786 | btrfs_qgroup_record_ref(trans, &ref->node, extent_op); | |
95a06077 | 787 | |
5d4f98a2 YZ |
788 | return 0; |
789 | } | |
790 | ||
791 | /* | |
792 | * add a delayed data ref. it's similar to btrfs_add_delayed_tree_ref. | |
793 | */ | |
66d7e7f0 AJ |
794 | int btrfs_add_delayed_data_ref(struct btrfs_fs_info *fs_info, |
795 | struct btrfs_trans_handle *trans, | |
5d4f98a2 YZ |
796 | u64 bytenr, u64 num_bytes, |
797 | u64 parent, u64 ref_root, | |
798 | u64 owner, u64 offset, int action, | |
66d7e7f0 AJ |
799 | struct btrfs_delayed_extent_op *extent_op, |
800 | int for_cow) | |
5d4f98a2 YZ |
801 | { |
802 | struct btrfs_delayed_data_ref *ref; | |
803 | struct btrfs_delayed_ref_head *head_ref; | |
804 | struct btrfs_delayed_ref_root *delayed_refs; | |
5d4f98a2 YZ |
805 | |
806 | BUG_ON(extent_op && !extent_op->is_data); | |
78a6184a | 807 | ref = kmem_cache_alloc(btrfs_delayed_data_ref_cachep, GFP_NOFS); |
5d4f98a2 YZ |
808 | if (!ref) |
809 | return -ENOMEM; | |
56bec294 | 810 | |
78a6184a | 811 | head_ref = kmem_cache_alloc(btrfs_delayed_ref_head_cachep, GFP_NOFS); |
56bec294 | 812 | if (!head_ref) { |
78a6184a | 813 | kmem_cache_free(btrfs_delayed_data_ref_cachep, ref); |
56bec294 CM |
814 | return -ENOMEM; |
815 | } | |
5d4f98a2 YZ |
816 | |
817 | head_ref->extent_op = extent_op; | |
818 | ||
56bec294 CM |
819 | delayed_refs = &trans->transaction->delayed_refs; |
820 | spin_lock(&delayed_refs->lock); | |
821 | ||
822 | /* | |
823 | * insert both the head node and the new ref without dropping | |
824 | * the spin lock | |
825 | */ | |
143bede5 | 826 | add_delayed_ref_head(fs_info, trans, &head_ref->node, bytenr, |
66d7e7f0 | 827 | num_bytes, action, 1); |
56bec294 | 828 | |
143bede5 | 829 | add_delayed_data_ref(fs_info, trans, &ref->node, bytenr, |
66d7e7f0 AJ |
830 | num_bytes, parent, ref_root, owner, offset, |
831 | action, for_cow); | |
5d4f98a2 | 832 | spin_unlock(&delayed_refs->lock); |
546adb0d JS |
833 | if (need_ref_seq(for_cow, ref_root)) |
834 | btrfs_qgroup_record_ref(trans, &ref->node, extent_op); | |
95a06077 | 835 | |
5d4f98a2 YZ |
836 | return 0; |
837 | } | |
838 | ||
66d7e7f0 AJ |
839 | int btrfs_add_delayed_extent_op(struct btrfs_fs_info *fs_info, |
840 | struct btrfs_trans_handle *trans, | |
5d4f98a2 YZ |
841 | u64 bytenr, u64 num_bytes, |
842 | struct btrfs_delayed_extent_op *extent_op) | |
843 | { | |
844 | struct btrfs_delayed_ref_head *head_ref; | |
845 | struct btrfs_delayed_ref_root *delayed_refs; | |
5d4f98a2 | 846 | |
78a6184a | 847 | head_ref = kmem_cache_alloc(btrfs_delayed_ref_head_cachep, GFP_NOFS); |
5d4f98a2 YZ |
848 | if (!head_ref) |
849 | return -ENOMEM; | |
850 | ||
851 | head_ref->extent_op = extent_op; | |
852 | ||
853 | delayed_refs = &trans->transaction->delayed_refs; | |
854 | spin_lock(&delayed_refs->lock); | |
855 | ||
143bede5 | 856 | add_delayed_ref_head(fs_info, trans, &head_ref->node, bytenr, |
5d4f98a2 YZ |
857 | num_bytes, BTRFS_UPDATE_DELAYED_HEAD, |
858 | extent_op->is_data); | |
5d4f98a2 | 859 | |
56bec294 CM |
860 | spin_unlock(&delayed_refs->lock); |
861 | return 0; | |
862 | } | |
863 | ||
1887be66 CM |
864 | /* |
865 | * this does a simple search for the head node for a given extent. | |
866 | * It must be called with the delayed ref spinlock held, and it returns | |
867 | * the head node if any where found, or NULL if not. | |
868 | */ | |
869 | struct btrfs_delayed_ref_head * | |
870 | btrfs_find_delayed_ref_head(struct btrfs_trans_handle *trans, u64 bytenr) | |
871 | { | |
872 | struct btrfs_delayed_ref_node *ref; | |
873 | struct btrfs_delayed_ref_root *delayed_refs; | |
874 | ||
875 | delayed_refs = &trans->transaction->delayed_refs; | |
d1270cd9 | 876 | ref = find_ref_head(&delayed_refs->root, bytenr, NULL, 0); |
1887be66 CM |
877 | if (ref) |
878 | return btrfs_delayed_node_to_head(ref); | |
879 | return NULL; | |
880 | } | |
78a6184a MX |
881 | |
882 | void btrfs_delayed_ref_exit(void) | |
883 | { | |
884 | if (btrfs_delayed_ref_head_cachep) | |
885 | kmem_cache_destroy(btrfs_delayed_ref_head_cachep); | |
886 | if (btrfs_delayed_tree_ref_cachep) | |
887 | kmem_cache_destroy(btrfs_delayed_tree_ref_cachep); | |
888 | if (btrfs_delayed_data_ref_cachep) | |
889 | kmem_cache_destroy(btrfs_delayed_data_ref_cachep); | |
890 | if (btrfs_delayed_extent_op_cachep) | |
891 | kmem_cache_destroy(btrfs_delayed_extent_op_cachep); | |
892 | } | |
893 | ||
894 | int btrfs_delayed_ref_init(void) | |
895 | { | |
896 | btrfs_delayed_ref_head_cachep = kmem_cache_create( | |
897 | "btrfs_delayed_ref_head", | |
898 | sizeof(struct btrfs_delayed_ref_head), 0, | |
899 | SLAB_RECLAIM_ACCOUNT | SLAB_MEM_SPREAD, NULL); | |
900 | if (!btrfs_delayed_ref_head_cachep) | |
901 | goto fail; | |
902 | ||
903 | btrfs_delayed_tree_ref_cachep = kmem_cache_create( | |
904 | "btrfs_delayed_tree_ref", | |
905 | sizeof(struct btrfs_delayed_tree_ref), 0, | |
906 | SLAB_RECLAIM_ACCOUNT | SLAB_MEM_SPREAD, NULL); | |
907 | if (!btrfs_delayed_tree_ref_cachep) | |
908 | goto fail; | |
909 | ||
910 | btrfs_delayed_data_ref_cachep = kmem_cache_create( | |
911 | "btrfs_delayed_data_ref", | |
912 | sizeof(struct btrfs_delayed_data_ref), 0, | |
913 | SLAB_RECLAIM_ACCOUNT | SLAB_MEM_SPREAD, NULL); | |
914 | if (!btrfs_delayed_data_ref_cachep) | |
915 | goto fail; | |
916 | ||
917 | btrfs_delayed_extent_op_cachep = kmem_cache_create( | |
918 | "btrfs_delayed_extent_op", | |
919 | sizeof(struct btrfs_delayed_extent_op), 0, | |
920 | SLAB_RECLAIM_ACCOUNT | SLAB_MEM_SPREAD, NULL); | |
921 | if (!btrfs_delayed_extent_op_cachep) | |
922 | goto fail; | |
923 | ||
924 | return 0; | |
925 | fail: | |
926 | btrfs_delayed_ref_exit(); | |
927 | return -ENOMEM; | |
928 | } |