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 | ||
26 | /* | |
27 | * delayed back reference update tracking. For subvolume trees | |
28 | * we queue up extent allocations and backref maintenance for | |
29 | * delayed processing. This avoids deep call chains where we | |
30 | * add extents in the middle of btrfs_search_slot, and it allows | |
31 | * us to buffer up frequently modified backrefs in an rb tree instead | |
32 | * of hammering updates on the extent allocation tree. | |
56bec294 CM |
33 | */ |
34 | ||
35 | /* | |
5d4f98a2 YZ |
36 | * compare two delayed tree backrefs with same bytenr and type |
37 | */ | |
38 | static int comp_tree_refs(struct btrfs_delayed_tree_ref *ref2, | |
39 | struct btrfs_delayed_tree_ref *ref1) | |
40 | { | |
ae1e206b JB |
41 | if (ref1->root < ref2->root) |
42 | return -1; | |
43 | if (ref1->root > ref2->root) | |
44 | return 1; | |
45 | if (ref1->parent < ref2->parent) | |
46 | return -1; | |
47 | if (ref1->parent > ref2->parent) | |
48 | return 1; | |
5d4f98a2 YZ |
49 | return 0; |
50 | } | |
51 | ||
52 | /* | |
53 | * compare two delayed data backrefs with same bytenr and type | |
56bec294 | 54 | */ |
5d4f98a2 YZ |
55 | static int comp_data_refs(struct btrfs_delayed_data_ref *ref2, |
56 | struct btrfs_delayed_data_ref *ref1) | |
56bec294 | 57 | { |
5d4f98a2 YZ |
58 | if (ref1->node.type == BTRFS_EXTENT_DATA_REF_KEY) { |
59 | if (ref1->root < ref2->root) | |
60 | return -1; | |
61 | if (ref1->root > ref2->root) | |
62 | return 1; | |
63 | if (ref1->objectid < ref2->objectid) | |
64 | return -1; | |
65 | if (ref1->objectid > ref2->objectid) | |
66 | return 1; | |
67 | if (ref1->offset < ref2->offset) | |
68 | return -1; | |
69 | if (ref1->offset > ref2->offset) | |
70 | return 1; | |
71 | } else { | |
72 | if (ref1->parent < ref2->parent) | |
73 | return -1; | |
74 | if (ref1->parent > ref2->parent) | |
75 | return 1; | |
76 | } | |
77 | return 0; | |
78 | } | |
79 | ||
80 | /* | |
81 | * entries in the rb tree are ordered by the byte number of the extent, | |
82 | * type of the delayed backrefs and content of delayed backrefs. | |
83 | */ | |
84 | static int comp_entry(struct btrfs_delayed_ref_node *ref2, | |
ae1e206b JB |
85 | struct btrfs_delayed_ref_node *ref1, |
86 | bool compare_seq) | |
5d4f98a2 YZ |
87 | { |
88 | if (ref1->bytenr < ref2->bytenr) | |
56bec294 | 89 | return -1; |
5d4f98a2 | 90 | if (ref1->bytenr > ref2->bytenr) |
56bec294 | 91 | return 1; |
5d4f98a2 YZ |
92 | if (ref1->is_head && ref2->is_head) |
93 | return 0; | |
94 | if (ref2->is_head) | |
56bec294 | 95 | return -1; |
5d4f98a2 | 96 | if (ref1->is_head) |
56bec294 | 97 | return 1; |
5d4f98a2 YZ |
98 | if (ref1->type < ref2->type) |
99 | return -1; | |
100 | if (ref1->type > ref2->type) | |
101 | return 1; | |
00f04b88 | 102 | /* merging of sequenced refs is not allowed */ |
ae1e206b JB |
103 | if (compare_seq) { |
104 | if (ref1->seq < ref2->seq) | |
105 | return -1; | |
106 | if (ref1->seq > ref2->seq) | |
107 | return 1; | |
108 | } | |
5d4f98a2 YZ |
109 | if (ref1->type == BTRFS_TREE_BLOCK_REF_KEY || |
110 | ref1->type == BTRFS_SHARED_BLOCK_REF_KEY) { | |
111 | return comp_tree_refs(btrfs_delayed_node_to_tree_ref(ref2), | |
112 | btrfs_delayed_node_to_tree_ref(ref1)); | |
113 | } else if (ref1->type == BTRFS_EXTENT_DATA_REF_KEY || | |
114 | ref1->type == BTRFS_SHARED_DATA_REF_KEY) { | |
115 | return comp_data_refs(btrfs_delayed_node_to_data_ref(ref2), | |
116 | btrfs_delayed_node_to_data_ref(ref1)); | |
117 | } | |
118 | BUG(); | |
56bec294 CM |
119 | return 0; |
120 | } | |
121 | ||
122 | /* | |
123 | * insert a new ref into the rbtree. This returns any existing refs | |
124 | * for the same (bytenr,parent) tuple, or NULL if the new node was properly | |
125 | * inserted. | |
126 | */ | |
127 | static struct btrfs_delayed_ref_node *tree_insert(struct rb_root *root, | |
56bec294 CM |
128 | struct rb_node *node) |
129 | { | |
130 | struct rb_node **p = &root->rb_node; | |
131 | struct rb_node *parent_node = NULL; | |
132 | struct btrfs_delayed_ref_node *entry; | |
5d4f98a2 | 133 | struct btrfs_delayed_ref_node *ins; |
56bec294 CM |
134 | int cmp; |
135 | ||
5d4f98a2 | 136 | ins = rb_entry(node, struct btrfs_delayed_ref_node, rb_node); |
56bec294 CM |
137 | while (*p) { |
138 | parent_node = *p; | |
139 | entry = rb_entry(parent_node, struct btrfs_delayed_ref_node, | |
140 | rb_node); | |
141 | ||
ae1e206b | 142 | cmp = comp_entry(entry, ins, 1); |
56bec294 CM |
143 | if (cmp < 0) |
144 | p = &(*p)->rb_left; | |
145 | else if (cmp > 0) | |
146 | p = &(*p)->rb_right; | |
147 | else | |
148 | return entry; | |
149 | } | |
150 | ||
56bec294 CM |
151 | rb_link_node(node, parent_node, p); |
152 | rb_insert_color(node, root); | |
153 | return NULL; | |
154 | } | |
155 | ||
156 | /* | |
5d4f98a2 | 157 | * find an head entry based on bytenr. This returns the delayed ref |
d1270cd9 AJ |
158 | * head if it was able to find one, or NULL if nothing was in that spot. |
159 | * If return_bigger is given, the next bigger entry is returned if no exact | |
160 | * match is found. | |
56bec294 | 161 | */ |
5d4f98a2 YZ |
162 | static struct btrfs_delayed_ref_node *find_ref_head(struct rb_root *root, |
163 | u64 bytenr, | |
d1270cd9 AJ |
164 | struct btrfs_delayed_ref_node **last, |
165 | int return_bigger) | |
56bec294 | 166 | { |
d1270cd9 | 167 | struct rb_node *n; |
56bec294 | 168 | struct btrfs_delayed_ref_node *entry; |
d1270cd9 | 169 | int cmp = 0; |
56bec294 | 170 | |
d1270cd9 AJ |
171 | again: |
172 | n = root->rb_node; | |
173 | entry = NULL; | |
56bec294 CM |
174 | while (n) { |
175 | entry = rb_entry(n, struct btrfs_delayed_ref_node, rb_node); | |
176 | WARN_ON(!entry->in_tree); | |
c3e69d58 CM |
177 | if (last) |
178 | *last = entry; | |
56bec294 | 179 | |
5d4f98a2 YZ |
180 | if (bytenr < entry->bytenr) |
181 | cmp = -1; | |
182 | else if (bytenr > entry->bytenr) | |
183 | cmp = 1; | |
184 | else if (!btrfs_delayed_ref_is_head(entry)) | |
185 | cmp = 1; | |
186 | else | |
187 | cmp = 0; | |
188 | ||
56bec294 CM |
189 | if (cmp < 0) |
190 | n = n->rb_left; | |
191 | else if (cmp > 0) | |
192 | n = n->rb_right; | |
193 | else | |
194 | return entry; | |
195 | } | |
d1270cd9 AJ |
196 | if (entry && return_bigger) { |
197 | if (cmp > 0) { | |
198 | n = rb_next(&entry->rb_node); | |
199 | if (!n) | |
200 | n = rb_first(root); | |
201 | entry = rb_entry(n, struct btrfs_delayed_ref_node, | |
202 | rb_node); | |
203 | bytenr = entry->bytenr; | |
204 | return_bigger = 0; | |
205 | goto again; | |
206 | } | |
207 | return entry; | |
208 | } | |
56bec294 CM |
209 | return NULL; |
210 | } | |
211 | ||
c3e69d58 CM |
212 | int btrfs_delayed_ref_lock(struct btrfs_trans_handle *trans, |
213 | struct btrfs_delayed_ref_head *head) | |
56bec294 | 214 | { |
c3e69d58 CM |
215 | struct btrfs_delayed_ref_root *delayed_refs; |
216 | ||
217 | delayed_refs = &trans->transaction->delayed_refs; | |
218 | assert_spin_locked(&delayed_refs->lock); | |
219 | if (mutex_trylock(&head->mutex)) | |
220 | return 0; | |
221 | ||
222 | atomic_inc(&head->node.refs); | |
223 | spin_unlock(&delayed_refs->lock); | |
224 | ||
225 | mutex_lock(&head->mutex); | |
226 | spin_lock(&delayed_refs->lock); | |
227 | if (!head->node.in_tree) { | |
228 | mutex_unlock(&head->mutex); | |
229 | btrfs_put_delayed_ref(&head->node); | |
230 | return -EAGAIN; | |
231 | } | |
232 | btrfs_put_delayed_ref(&head->node); | |
233 | return 0; | |
234 | } | |
235 | ||
ae1e206b JB |
236 | static void inline drop_delayed_ref(struct btrfs_trans_handle *trans, |
237 | struct btrfs_delayed_ref_root *delayed_refs, | |
238 | struct btrfs_delayed_ref_node *ref) | |
239 | { | |
240 | rb_erase(&ref->rb_node, &delayed_refs->root); | |
241 | ref->in_tree = 0; | |
242 | btrfs_put_delayed_ref(ref); | |
243 | delayed_refs->num_entries--; | |
244 | if (trans->delayed_ref_updates) | |
245 | trans->delayed_ref_updates--; | |
246 | } | |
247 | ||
248 | static int merge_ref(struct btrfs_trans_handle *trans, | |
249 | struct btrfs_delayed_ref_root *delayed_refs, | |
250 | struct btrfs_delayed_ref_node *ref, u64 seq) | |
251 | { | |
252 | struct rb_node *node; | |
253 | int merged = 0; | |
254 | int mod = 0; | |
255 | int done = 0; | |
256 | ||
257 | node = rb_prev(&ref->rb_node); | |
258 | while (node) { | |
259 | struct btrfs_delayed_ref_node *next; | |
260 | ||
261 | next = rb_entry(node, struct btrfs_delayed_ref_node, rb_node); | |
262 | node = rb_prev(node); | |
263 | if (next->bytenr != ref->bytenr) | |
264 | break; | |
265 | if (seq && next->seq >= seq) | |
266 | break; | |
267 | if (comp_entry(ref, next, 0)) | |
268 | continue; | |
269 | ||
270 | if (ref->action == next->action) { | |
271 | mod = next->ref_mod; | |
272 | } else { | |
273 | if (ref->ref_mod < next->ref_mod) { | |
274 | struct btrfs_delayed_ref_node *tmp; | |
275 | ||
276 | tmp = ref; | |
277 | ref = next; | |
278 | next = tmp; | |
279 | done = 1; | |
280 | } | |
281 | mod = -next->ref_mod; | |
282 | } | |
283 | ||
284 | merged++; | |
285 | drop_delayed_ref(trans, delayed_refs, next); | |
286 | ref->ref_mod += mod; | |
287 | if (ref->ref_mod == 0) { | |
288 | drop_delayed_ref(trans, delayed_refs, ref); | |
289 | break; | |
290 | } else { | |
291 | /* | |
292 | * You can't have multiples of the same ref on a tree | |
293 | * block. | |
294 | */ | |
295 | WARN_ON(ref->type == BTRFS_TREE_BLOCK_REF_KEY || | |
296 | ref->type == BTRFS_SHARED_BLOCK_REF_KEY); | |
297 | } | |
298 | ||
299 | if (done) | |
300 | break; | |
301 | node = rb_prev(&ref->rb_node); | |
302 | } | |
303 | ||
304 | return merged; | |
305 | } | |
306 | ||
307 | void btrfs_merge_delayed_refs(struct btrfs_trans_handle *trans, | |
308 | struct btrfs_fs_info *fs_info, | |
309 | struct btrfs_delayed_ref_root *delayed_refs, | |
310 | struct btrfs_delayed_ref_head *head) | |
311 | { | |
312 | struct rb_node *node; | |
313 | u64 seq = 0; | |
314 | ||
315 | spin_lock(&fs_info->tree_mod_seq_lock); | |
316 | if (!list_empty(&fs_info->tree_mod_seq_list)) { | |
317 | struct seq_list *elem; | |
318 | ||
319 | elem = list_first_entry(&fs_info->tree_mod_seq_list, | |
320 | struct seq_list, list); | |
321 | seq = elem->seq; | |
322 | } | |
323 | spin_unlock(&fs_info->tree_mod_seq_lock); | |
324 | ||
325 | node = rb_prev(&head->node.rb_node); | |
326 | while (node) { | |
327 | struct btrfs_delayed_ref_node *ref; | |
328 | ||
329 | ref = rb_entry(node, struct btrfs_delayed_ref_node, | |
330 | rb_node); | |
331 | if (ref->bytenr != head->node.bytenr) | |
332 | break; | |
333 | ||
334 | /* We can't merge refs that are outside of our seq count */ | |
335 | if (seq && ref->seq >= seq) | |
336 | break; | |
337 | if (merge_ref(trans, delayed_refs, ref, seq)) | |
338 | node = rb_prev(&head->node.rb_node); | |
339 | else | |
340 | node = rb_prev(node); | |
341 | } | |
342 | } | |
343 | ||
097b8a7c JS |
344 | int btrfs_check_delayed_seq(struct btrfs_fs_info *fs_info, |
345 | struct btrfs_delayed_ref_root *delayed_refs, | |
00f04b88 AJ |
346 | u64 seq) |
347 | { | |
348 | struct seq_list *elem; | |
097b8a7c JS |
349 | int ret = 0; |
350 | ||
351 | spin_lock(&fs_info->tree_mod_seq_lock); | |
352 | if (!list_empty(&fs_info->tree_mod_seq_list)) { | |
353 | elem = list_first_entry(&fs_info->tree_mod_seq_list, | |
354 | struct seq_list, list); | |
355 | if (seq >= elem->seq) { | |
356 | pr_debug("holding back delayed_ref %llu, lowest is " | |
357 | "%llu (%p)\n", seq, elem->seq, delayed_refs); | |
358 | ret = 1; | |
359 | } | |
00f04b88 | 360 | } |
097b8a7c JS |
361 | |
362 | spin_unlock(&fs_info->tree_mod_seq_lock); | |
363 | return ret; | |
00f04b88 AJ |
364 | } |
365 | ||
c3e69d58 CM |
366 | int btrfs_find_ref_cluster(struct btrfs_trans_handle *trans, |
367 | struct list_head *cluster, u64 start) | |
368 | { | |
369 | int count = 0; | |
370 | struct btrfs_delayed_ref_root *delayed_refs; | |
56bec294 | 371 | struct rb_node *node; |
c3e69d58 | 372 | struct btrfs_delayed_ref_node *ref; |
56bec294 | 373 | struct btrfs_delayed_ref_head *head; |
56bec294 | 374 | |
c3e69d58 CM |
375 | delayed_refs = &trans->transaction->delayed_refs; |
376 | if (start == 0) { | |
377 | node = rb_first(&delayed_refs->root); | |
378 | } else { | |
379 | ref = NULL; | |
d1270cd9 | 380 | find_ref_head(&delayed_refs->root, start + 1, &ref, 1); |
c3e69d58 | 381 | if (ref) { |
c3e69d58 CM |
382 | node = &ref->rb_node; |
383 | } else | |
384 | node = rb_first(&delayed_refs->root); | |
385 | } | |
386 | again: | |
387 | while (node && count < 32) { | |
388 | ref = rb_entry(node, struct btrfs_delayed_ref_node, rb_node); | |
56bec294 CM |
389 | if (btrfs_delayed_ref_is_head(ref)) { |
390 | head = btrfs_delayed_node_to_head(ref); | |
c3e69d58 CM |
391 | if (list_empty(&head->cluster)) { |
392 | list_add_tail(&head->cluster, cluster); | |
393 | delayed_refs->run_delayed_start = | |
394 | head->node.bytenr; | |
395 | count++; | |
396 | ||
397 | WARN_ON(delayed_refs->num_heads_ready == 0); | |
398 | delayed_refs->num_heads_ready--; | |
399 | } else if (count) { | |
400 | /* the goal of the clustering is to find extents | |
401 | * that are likely to end up in the same extent | |
402 | * leaf on disk. So, we don't want them spread | |
403 | * all over the tree. Stop now if we've hit | |
404 | * a head that was already in use | |
405 | */ | |
56bec294 CM |
406 | break; |
407 | } | |
408 | } | |
c3e69d58 | 409 | node = rb_next(node); |
56bec294 | 410 | } |
c3e69d58 CM |
411 | if (count) { |
412 | return 0; | |
413 | } else if (start) { | |
414 | /* | |
415 | * we've gone to the end of the rbtree without finding any | |
416 | * clusters. start from the beginning and try again | |
417 | */ | |
418 | start = 0; | |
419 | node = rb_first(&delayed_refs->root); | |
420 | goto again; | |
421 | } | |
422 | return 1; | |
56bec294 CM |
423 | } |
424 | ||
56bec294 CM |
425 | /* |
426 | * helper function to update an extent delayed ref in the | |
427 | * rbtree. existing and update must both have the same | |
428 | * bytenr and parent | |
429 | * | |
430 | * This may free existing if the update cancels out whatever | |
431 | * operation it was doing. | |
432 | */ | |
433 | static noinline void | |
434 | update_existing_ref(struct btrfs_trans_handle *trans, | |
435 | struct btrfs_delayed_ref_root *delayed_refs, | |
436 | struct btrfs_delayed_ref_node *existing, | |
437 | struct btrfs_delayed_ref_node *update) | |
438 | { | |
5d4f98a2 | 439 | if (update->action != existing->action) { |
56bec294 CM |
440 | /* |
441 | * this is effectively undoing either an add or a | |
442 | * drop. We decrement the ref_mod, and if it goes | |
443 | * down to zero we just delete the entry without | |
444 | * every changing the extent allocation tree. | |
445 | */ | |
446 | existing->ref_mod--; | |
ae1e206b JB |
447 | if (existing->ref_mod == 0) |
448 | drop_delayed_ref(trans, delayed_refs, existing); | |
449 | else | |
5d4f98a2 YZ |
450 | WARN_ON(existing->type == BTRFS_TREE_BLOCK_REF_KEY || |
451 | existing->type == BTRFS_SHARED_BLOCK_REF_KEY); | |
56bec294 | 452 | } else { |
5d4f98a2 YZ |
453 | WARN_ON(existing->type == BTRFS_TREE_BLOCK_REF_KEY || |
454 | existing->type == BTRFS_SHARED_BLOCK_REF_KEY); | |
56bec294 CM |
455 | /* |
456 | * the action on the existing ref matches | |
457 | * the action on the ref we're trying to add. | |
458 | * Bump the ref_mod by one so the backref that | |
459 | * is eventually added/removed has the correct | |
460 | * reference count | |
461 | */ | |
462 | existing->ref_mod += update->ref_mod; | |
463 | } | |
464 | } | |
465 | ||
466 | /* | |
467 | * helper function to update the accounting in the head ref | |
468 | * existing and update must have the same bytenr | |
469 | */ | |
470 | static noinline void | |
471 | update_existing_head_ref(struct btrfs_delayed_ref_node *existing, | |
472 | struct btrfs_delayed_ref_node *update) | |
473 | { | |
474 | struct btrfs_delayed_ref_head *existing_ref; | |
475 | struct btrfs_delayed_ref_head *ref; | |
476 | ||
477 | existing_ref = btrfs_delayed_node_to_head(existing); | |
478 | ref = btrfs_delayed_node_to_head(update); | |
5d4f98a2 | 479 | BUG_ON(existing_ref->is_data != ref->is_data); |
56bec294 CM |
480 | |
481 | if (ref->must_insert_reserved) { | |
482 | /* if the extent was freed and then | |
483 | * reallocated before the delayed ref | |
484 | * entries were processed, we can end up | |
485 | * with an existing head ref without | |
486 | * the must_insert_reserved flag set. | |
487 | * Set it again here | |
488 | */ | |
489 | existing_ref->must_insert_reserved = ref->must_insert_reserved; | |
490 | ||
491 | /* | |
492 | * update the num_bytes so we make sure the accounting | |
493 | * is done correctly | |
494 | */ | |
495 | existing->num_bytes = update->num_bytes; | |
496 | ||
497 | } | |
498 | ||
5d4f98a2 YZ |
499 | if (ref->extent_op) { |
500 | if (!existing_ref->extent_op) { | |
501 | existing_ref->extent_op = ref->extent_op; | |
502 | } else { | |
503 | if (ref->extent_op->update_key) { | |
504 | memcpy(&existing_ref->extent_op->key, | |
505 | &ref->extent_op->key, | |
506 | sizeof(ref->extent_op->key)); | |
507 | existing_ref->extent_op->update_key = 1; | |
508 | } | |
509 | if (ref->extent_op->update_flags) { | |
510 | existing_ref->extent_op->flags_to_set |= | |
511 | ref->extent_op->flags_to_set; | |
512 | existing_ref->extent_op->update_flags = 1; | |
513 | } | |
514 | kfree(ref->extent_op); | |
515 | } | |
516 | } | |
56bec294 CM |
517 | /* |
518 | * update the reference mod on the head to reflect this new operation | |
519 | */ | |
520 | existing->ref_mod += update->ref_mod; | |
521 | } | |
522 | ||
523 | /* | |
5d4f98a2 | 524 | * helper function to actually insert a head node into the rbtree. |
56bec294 | 525 | * this does all the dirty work in terms of maintaining the correct |
5d4f98a2 | 526 | * overall modification count. |
56bec294 | 527 | */ |
143bede5 | 528 | static noinline void add_delayed_ref_head(struct btrfs_fs_info *fs_info, |
66d7e7f0 | 529 | struct btrfs_trans_handle *trans, |
5d4f98a2 YZ |
530 | struct btrfs_delayed_ref_node *ref, |
531 | u64 bytenr, u64 num_bytes, | |
532 | int action, int is_data) | |
56bec294 CM |
533 | { |
534 | struct btrfs_delayed_ref_node *existing; | |
c3e69d58 | 535 | struct btrfs_delayed_ref_head *head_ref = NULL; |
56bec294 CM |
536 | struct btrfs_delayed_ref_root *delayed_refs; |
537 | int count_mod = 1; | |
538 | int must_insert_reserved = 0; | |
539 | ||
540 | /* | |
541 | * the head node stores the sum of all the mods, so dropping a ref | |
542 | * should drop the sum in the head node by one. | |
543 | */ | |
5d4f98a2 YZ |
544 | if (action == BTRFS_UPDATE_DELAYED_HEAD) |
545 | count_mod = 0; | |
546 | else if (action == BTRFS_DROP_DELAYED_REF) | |
547 | count_mod = -1; | |
56bec294 CM |
548 | |
549 | /* | |
550 | * BTRFS_ADD_DELAYED_EXTENT means that we need to update | |
551 | * the reserved accounting when the extent is finally added, or | |
552 | * if a later modification deletes the delayed ref without ever | |
553 | * inserting the extent into the extent allocation tree. | |
554 | * ref->must_insert_reserved is the flag used to record | |
555 | * that accounting mods are required. | |
556 | * | |
557 | * Once we record must_insert_reserved, switch the action to | |
558 | * BTRFS_ADD_DELAYED_REF because other special casing is not required. | |
559 | */ | |
5d4f98a2 | 560 | if (action == BTRFS_ADD_DELAYED_EXTENT) |
56bec294 | 561 | must_insert_reserved = 1; |
5d4f98a2 | 562 | else |
56bec294 | 563 | must_insert_reserved = 0; |
56bec294 CM |
564 | |
565 | delayed_refs = &trans->transaction->delayed_refs; | |
566 | ||
567 | /* first set the basic ref node struct up */ | |
568 | atomic_set(&ref->refs, 1); | |
569 | ref->bytenr = bytenr; | |
5d4f98a2 | 570 | ref->num_bytes = num_bytes; |
56bec294 | 571 | ref->ref_mod = count_mod; |
5d4f98a2 YZ |
572 | ref->type = 0; |
573 | ref->action = 0; | |
574 | ref->is_head = 1; | |
56bec294 | 575 | ref->in_tree = 1; |
00f04b88 | 576 | ref->seq = 0; |
5d4f98a2 YZ |
577 | |
578 | head_ref = btrfs_delayed_node_to_head(ref); | |
579 | head_ref->must_insert_reserved = must_insert_reserved; | |
580 | head_ref->is_data = is_data; | |
581 | ||
582 | INIT_LIST_HEAD(&head_ref->cluster); | |
583 | mutex_init(&head_ref->mutex); | |
584 | ||
1abe9b8a | 585 | trace_btrfs_delayed_ref_head(ref, head_ref, action); |
586 | ||
5d4f98a2 YZ |
587 | existing = tree_insert(&delayed_refs->root, &ref->rb_node); |
588 | ||
589 | if (existing) { | |
590 | update_existing_head_ref(existing, ref); | |
591 | /* | |
592 | * we've updated the existing ref, free the newly | |
593 | * allocated ref | |
594 | */ | |
cddcd800 | 595 | kfree(head_ref); |
5d4f98a2 YZ |
596 | } else { |
597 | delayed_refs->num_heads++; | |
598 | delayed_refs->num_heads_ready++; | |
599 | delayed_refs->num_entries++; | |
600 | trans->delayed_ref_updates++; | |
601 | } | |
5d4f98a2 YZ |
602 | } |
603 | ||
604 | /* | |
605 | * helper to insert a delayed tree ref into the rbtree. | |
606 | */ | |
143bede5 | 607 | static noinline void add_delayed_tree_ref(struct btrfs_fs_info *fs_info, |
66d7e7f0 | 608 | struct btrfs_trans_handle *trans, |
5d4f98a2 YZ |
609 | struct btrfs_delayed_ref_node *ref, |
610 | u64 bytenr, u64 num_bytes, u64 parent, | |
66d7e7f0 AJ |
611 | u64 ref_root, int level, int action, |
612 | int for_cow) | |
5d4f98a2 YZ |
613 | { |
614 | struct btrfs_delayed_ref_node *existing; | |
615 | struct btrfs_delayed_tree_ref *full_ref; | |
616 | struct btrfs_delayed_ref_root *delayed_refs; | |
00f04b88 | 617 | u64 seq = 0; |
5d4f98a2 YZ |
618 | |
619 | if (action == BTRFS_ADD_DELAYED_EXTENT) | |
620 | action = BTRFS_ADD_DELAYED_REF; | |
621 | ||
622 | delayed_refs = &trans->transaction->delayed_refs; | |
623 | ||
624 | /* first set the basic ref node struct up */ | |
625 | atomic_set(&ref->refs, 1); | |
626 | ref->bytenr = bytenr; | |
56bec294 | 627 | ref->num_bytes = num_bytes; |
5d4f98a2 YZ |
628 | ref->ref_mod = 1; |
629 | ref->action = action; | |
630 | ref->is_head = 0; | |
631 | ref->in_tree = 1; | |
56bec294 | 632 | |
546adb0d JS |
633 | if (need_ref_seq(for_cow, ref_root)) |
634 | seq = btrfs_get_tree_mod_seq(fs_info, &trans->delayed_ref_elem); | |
00f04b88 AJ |
635 | ref->seq = seq; |
636 | ||
5d4f98a2 | 637 | full_ref = btrfs_delayed_node_to_tree_ref(ref); |
eebe063b AJ |
638 | full_ref->parent = parent; |
639 | full_ref->root = ref_root; | |
640 | if (parent) | |
5d4f98a2 | 641 | ref->type = BTRFS_SHARED_BLOCK_REF_KEY; |
eebe063b | 642 | else |
5d4f98a2 | 643 | ref->type = BTRFS_TREE_BLOCK_REF_KEY; |
5d4f98a2 | 644 | full_ref->level = level; |
56bec294 | 645 | |
1abe9b8a | 646 | trace_btrfs_delayed_tree_ref(ref, full_ref, action); |
647 | ||
5d4f98a2 | 648 | existing = tree_insert(&delayed_refs->root, &ref->rb_node); |
56bec294 CM |
649 | |
650 | if (existing) { | |
5d4f98a2 YZ |
651 | update_existing_ref(trans, delayed_refs, existing, ref); |
652 | /* | |
653 | * we've updated the existing ref, free the newly | |
654 | * allocated ref | |
655 | */ | |
cddcd800 | 656 | kfree(full_ref); |
5d4f98a2 YZ |
657 | } else { |
658 | delayed_refs->num_entries++; | |
659 | trans->delayed_ref_updates++; | |
660 | } | |
5d4f98a2 YZ |
661 | } |
662 | ||
663 | /* | |
664 | * helper to insert a delayed data ref into the rbtree. | |
665 | */ | |
143bede5 | 666 | static noinline void add_delayed_data_ref(struct btrfs_fs_info *fs_info, |
66d7e7f0 | 667 | struct btrfs_trans_handle *trans, |
5d4f98a2 YZ |
668 | struct btrfs_delayed_ref_node *ref, |
669 | u64 bytenr, u64 num_bytes, u64 parent, | |
670 | u64 ref_root, u64 owner, u64 offset, | |
66d7e7f0 | 671 | int action, int for_cow) |
5d4f98a2 YZ |
672 | { |
673 | struct btrfs_delayed_ref_node *existing; | |
674 | struct btrfs_delayed_data_ref *full_ref; | |
675 | struct btrfs_delayed_ref_root *delayed_refs; | |
00f04b88 | 676 | u64 seq = 0; |
5d4f98a2 YZ |
677 | |
678 | if (action == BTRFS_ADD_DELAYED_EXTENT) | |
679 | action = BTRFS_ADD_DELAYED_REF; | |
680 | ||
681 | delayed_refs = &trans->transaction->delayed_refs; | |
682 | ||
683 | /* first set the basic ref node struct up */ | |
684 | atomic_set(&ref->refs, 1); | |
685 | ref->bytenr = bytenr; | |
686 | ref->num_bytes = num_bytes; | |
687 | ref->ref_mod = 1; | |
688 | ref->action = action; | |
689 | ref->is_head = 0; | |
690 | ref->in_tree = 1; | |
691 | ||
546adb0d JS |
692 | if (need_ref_seq(for_cow, ref_root)) |
693 | seq = btrfs_get_tree_mod_seq(fs_info, &trans->delayed_ref_elem); | |
00f04b88 AJ |
694 | ref->seq = seq; |
695 | ||
5d4f98a2 | 696 | full_ref = btrfs_delayed_node_to_data_ref(ref); |
eebe063b AJ |
697 | full_ref->parent = parent; |
698 | full_ref->root = ref_root; | |
699 | if (parent) | |
5d4f98a2 | 700 | ref->type = BTRFS_SHARED_DATA_REF_KEY; |
eebe063b | 701 | else |
5d4f98a2 | 702 | ref->type = BTRFS_EXTENT_DATA_REF_KEY; |
66d7e7f0 | 703 | |
5d4f98a2 YZ |
704 | full_ref->objectid = owner; |
705 | full_ref->offset = offset; | |
56bec294 | 706 | |
1abe9b8a | 707 | trace_btrfs_delayed_data_ref(ref, full_ref, action); |
708 | ||
5d4f98a2 YZ |
709 | existing = tree_insert(&delayed_refs->root, &ref->rb_node); |
710 | ||
711 | if (existing) { | |
712 | update_existing_ref(trans, delayed_refs, existing, ref); | |
56bec294 CM |
713 | /* |
714 | * we've updated the existing ref, free the newly | |
715 | * allocated ref | |
716 | */ | |
cddcd800 | 717 | kfree(full_ref); |
56bec294 CM |
718 | } else { |
719 | delayed_refs->num_entries++; | |
720 | trans->delayed_ref_updates++; | |
721 | } | |
56bec294 CM |
722 | } |
723 | ||
724 | /* | |
5d4f98a2 | 725 | * add a delayed tree ref. This does all of the accounting required |
56bec294 CM |
726 | * to make sure the delayed ref is eventually processed before this |
727 | * transaction commits. | |
728 | */ | |
66d7e7f0 AJ |
729 | int btrfs_add_delayed_tree_ref(struct btrfs_fs_info *fs_info, |
730 | struct btrfs_trans_handle *trans, | |
5d4f98a2 YZ |
731 | u64 bytenr, u64 num_bytes, u64 parent, |
732 | u64 ref_root, int level, int action, | |
66d7e7f0 AJ |
733 | struct btrfs_delayed_extent_op *extent_op, |
734 | int for_cow) | |
56bec294 | 735 | { |
5d4f98a2 | 736 | struct btrfs_delayed_tree_ref *ref; |
56bec294 CM |
737 | struct btrfs_delayed_ref_head *head_ref; |
738 | struct btrfs_delayed_ref_root *delayed_refs; | |
56bec294 | 739 | |
5d4f98a2 | 740 | BUG_ON(extent_op && extent_op->is_data); |
56bec294 CM |
741 | ref = kmalloc(sizeof(*ref), GFP_NOFS); |
742 | if (!ref) | |
743 | return -ENOMEM; | |
744 | ||
5d4f98a2 YZ |
745 | head_ref = kmalloc(sizeof(*head_ref), GFP_NOFS); |
746 | if (!head_ref) { | |
747 | kfree(ref); | |
748 | return -ENOMEM; | |
749 | } | |
750 | ||
751 | head_ref->extent_op = extent_op; | |
752 | ||
753 | delayed_refs = &trans->transaction->delayed_refs; | |
754 | spin_lock(&delayed_refs->lock); | |
755 | ||
56bec294 | 756 | /* |
5d4f98a2 YZ |
757 | * insert both the head node and the new ref without dropping |
758 | * the spin lock | |
56bec294 | 759 | */ |
143bede5 | 760 | add_delayed_ref_head(fs_info, trans, &head_ref->node, bytenr, |
66d7e7f0 | 761 | num_bytes, action, 0); |
5d4f98a2 | 762 | |
143bede5 | 763 | add_delayed_tree_ref(fs_info, trans, &ref->node, bytenr, |
66d7e7f0 AJ |
764 | num_bytes, parent, ref_root, level, action, |
765 | for_cow); | |
5d4f98a2 | 766 | spin_unlock(&delayed_refs->lock); |
546adb0d JS |
767 | if (need_ref_seq(for_cow, ref_root)) |
768 | btrfs_qgroup_record_ref(trans, &ref->node, extent_op); | |
95a06077 | 769 | |
5d4f98a2 YZ |
770 | return 0; |
771 | } | |
772 | ||
773 | /* | |
774 | * add a delayed data ref. it's similar to btrfs_add_delayed_tree_ref. | |
775 | */ | |
66d7e7f0 AJ |
776 | int btrfs_add_delayed_data_ref(struct btrfs_fs_info *fs_info, |
777 | struct btrfs_trans_handle *trans, | |
5d4f98a2 YZ |
778 | u64 bytenr, u64 num_bytes, |
779 | u64 parent, u64 ref_root, | |
780 | u64 owner, u64 offset, int action, | |
66d7e7f0 AJ |
781 | struct btrfs_delayed_extent_op *extent_op, |
782 | int for_cow) | |
5d4f98a2 YZ |
783 | { |
784 | struct btrfs_delayed_data_ref *ref; | |
785 | struct btrfs_delayed_ref_head *head_ref; | |
786 | struct btrfs_delayed_ref_root *delayed_refs; | |
5d4f98a2 YZ |
787 | |
788 | BUG_ON(extent_op && !extent_op->is_data); | |
789 | ref = kmalloc(sizeof(*ref), GFP_NOFS); | |
790 | if (!ref) | |
791 | return -ENOMEM; | |
56bec294 CM |
792 | |
793 | head_ref = kmalloc(sizeof(*head_ref), GFP_NOFS); | |
794 | if (!head_ref) { | |
795 | kfree(ref); | |
796 | return -ENOMEM; | |
797 | } | |
5d4f98a2 YZ |
798 | |
799 | head_ref->extent_op = extent_op; | |
800 | ||
56bec294 CM |
801 | delayed_refs = &trans->transaction->delayed_refs; |
802 | spin_lock(&delayed_refs->lock); | |
803 | ||
804 | /* | |
805 | * insert both the head node and the new ref without dropping | |
806 | * the spin lock | |
807 | */ | |
143bede5 | 808 | add_delayed_ref_head(fs_info, trans, &head_ref->node, bytenr, |
66d7e7f0 | 809 | num_bytes, action, 1); |
56bec294 | 810 | |
143bede5 | 811 | add_delayed_data_ref(fs_info, trans, &ref->node, bytenr, |
66d7e7f0 AJ |
812 | num_bytes, parent, ref_root, owner, offset, |
813 | action, for_cow); | |
5d4f98a2 | 814 | spin_unlock(&delayed_refs->lock); |
546adb0d JS |
815 | if (need_ref_seq(for_cow, ref_root)) |
816 | btrfs_qgroup_record_ref(trans, &ref->node, extent_op); | |
95a06077 | 817 | |
5d4f98a2 YZ |
818 | return 0; |
819 | } | |
820 | ||
66d7e7f0 AJ |
821 | int btrfs_add_delayed_extent_op(struct btrfs_fs_info *fs_info, |
822 | struct btrfs_trans_handle *trans, | |
5d4f98a2 YZ |
823 | u64 bytenr, u64 num_bytes, |
824 | struct btrfs_delayed_extent_op *extent_op) | |
825 | { | |
826 | struct btrfs_delayed_ref_head *head_ref; | |
827 | struct btrfs_delayed_ref_root *delayed_refs; | |
5d4f98a2 YZ |
828 | |
829 | head_ref = kmalloc(sizeof(*head_ref), GFP_NOFS); | |
830 | if (!head_ref) | |
831 | return -ENOMEM; | |
832 | ||
833 | head_ref->extent_op = extent_op; | |
834 | ||
835 | delayed_refs = &trans->transaction->delayed_refs; | |
836 | spin_lock(&delayed_refs->lock); | |
837 | ||
143bede5 | 838 | add_delayed_ref_head(fs_info, trans, &head_ref->node, bytenr, |
5d4f98a2 YZ |
839 | num_bytes, BTRFS_UPDATE_DELAYED_HEAD, |
840 | extent_op->is_data); | |
5d4f98a2 | 841 | |
56bec294 CM |
842 | spin_unlock(&delayed_refs->lock); |
843 | return 0; | |
844 | } | |
845 | ||
1887be66 CM |
846 | /* |
847 | * this does a simple search for the head node for a given extent. | |
848 | * It must be called with the delayed ref spinlock held, and it returns | |
849 | * the head node if any where found, or NULL if not. | |
850 | */ | |
851 | struct btrfs_delayed_ref_head * | |
852 | btrfs_find_delayed_ref_head(struct btrfs_trans_handle *trans, u64 bytenr) | |
853 | { | |
854 | struct btrfs_delayed_ref_node *ref; | |
855 | struct btrfs_delayed_ref_root *delayed_refs; | |
856 | ||
857 | delayed_refs = &trans->transaction->delayed_refs; | |
d1270cd9 | 858 | ref = find_ref_head(&delayed_refs->root, bytenr, NULL, 0); |
1887be66 CM |
859 | if (ref) |
860 | return btrfs_delayed_node_to_head(ref); | |
861 | return NULL; | |
862 | } |