powerpc: Fix up dma_alloc_coherent() on platforms without cache coherency.
[deliverable/linux.git] / fs / btrfs / delayed-ref.c
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>
20 #include <linux/sort.h>
21 #include "ctree.h"
22 #include "delayed-ref.h"
23 #include "transaction.h"
24
25 /*
26 * delayed back reference update tracking. For subvolume trees
27 * we queue up extent allocations and backref maintenance for
28 * delayed processing. This avoids deep call chains where we
29 * add extents in the middle of btrfs_search_slot, and it allows
30 * us to buffer up frequently modified backrefs in an rb tree instead
31 * of hammering updates on the extent allocation tree.
32 *
33 * Right now this code is only used for reference counted trees, but
34 * the long term goal is to get rid of the similar code for delayed
35 * extent tree modifications.
36 */
37
38 /*
39 * entries in the rb tree are ordered by the byte number of the extent
40 * and by the byte number of the parent block.
41 */
42 static int comp_entry(struct btrfs_delayed_ref_node *ref,
43 u64 bytenr, u64 parent)
44 {
45 if (bytenr < ref->bytenr)
46 return -1;
47 if (bytenr > ref->bytenr)
48 return 1;
49 if (parent < ref->parent)
50 return -1;
51 if (parent > ref->parent)
52 return 1;
53 return 0;
54 }
55
56 /*
57 * insert a new ref into the rbtree. This returns any existing refs
58 * for the same (bytenr,parent) tuple, or NULL if the new node was properly
59 * inserted.
60 */
61 static struct btrfs_delayed_ref_node *tree_insert(struct rb_root *root,
62 u64 bytenr, u64 parent,
63 struct rb_node *node)
64 {
65 struct rb_node **p = &root->rb_node;
66 struct rb_node *parent_node = NULL;
67 struct btrfs_delayed_ref_node *entry;
68 int cmp;
69
70 while (*p) {
71 parent_node = *p;
72 entry = rb_entry(parent_node, struct btrfs_delayed_ref_node,
73 rb_node);
74
75 cmp = comp_entry(entry, bytenr, parent);
76 if (cmp < 0)
77 p = &(*p)->rb_left;
78 else if (cmp > 0)
79 p = &(*p)->rb_right;
80 else
81 return entry;
82 }
83
84 entry = rb_entry(node, struct btrfs_delayed_ref_node, rb_node);
85 rb_link_node(node, parent_node, p);
86 rb_insert_color(node, root);
87 return NULL;
88 }
89
90 /*
91 * find an entry based on (bytenr,parent). This returns the delayed
92 * ref if it was able to find one, or NULL if nothing was in that spot
93 */
94 static struct btrfs_delayed_ref_node *tree_search(struct rb_root *root,
95 u64 bytenr, u64 parent,
96 struct btrfs_delayed_ref_node **last)
97 {
98 struct rb_node *n = root->rb_node;
99 struct btrfs_delayed_ref_node *entry;
100 int cmp;
101
102 while (n) {
103 entry = rb_entry(n, struct btrfs_delayed_ref_node, rb_node);
104 WARN_ON(!entry->in_tree);
105 if (last)
106 *last = entry;
107
108 cmp = comp_entry(entry, bytenr, parent);
109 if (cmp < 0)
110 n = n->rb_left;
111 else if (cmp > 0)
112 n = n->rb_right;
113 else
114 return entry;
115 }
116 return NULL;
117 }
118
119 int btrfs_delayed_ref_lock(struct btrfs_trans_handle *trans,
120 struct btrfs_delayed_ref_head *head)
121 {
122 struct btrfs_delayed_ref_root *delayed_refs;
123
124 delayed_refs = &trans->transaction->delayed_refs;
125 assert_spin_locked(&delayed_refs->lock);
126 if (mutex_trylock(&head->mutex))
127 return 0;
128
129 atomic_inc(&head->node.refs);
130 spin_unlock(&delayed_refs->lock);
131
132 mutex_lock(&head->mutex);
133 spin_lock(&delayed_refs->lock);
134 if (!head->node.in_tree) {
135 mutex_unlock(&head->mutex);
136 btrfs_put_delayed_ref(&head->node);
137 return -EAGAIN;
138 }
139 btrfs_put_delayed_ref(&head->node);
140 return 0;
141 }
142
143 int btrfs_find_ref_cluster(struct btrfs_trans_handle *trans,
144 struct list_head *cluster, u64 start)
145 {
146 int count = 0;
147 struct btrfs_delayed_ref_root *delayed_refs;
148 struct rb_node *node;
149 struct btrfs_delayed_ref_node *ref;
150 struct btrfs_delayed_ref_head *head;
151
152 delayed_refs = &trans->transaction->delayed_refs;
153 if (start == 0) {
154 node = rb_first(&delayed_refs->root);
155 } else {
156 ref = NULL;
157 tree_search(&delayed_refs->root, start, (u64)-1, &ref);
158 if (ref) {
159 struct btrfs_delayed_ref_node *tmp;
160
161 node = rb_prev(&ref->rb_node);
162 while (node) {
163 tmp = rb_entry(node,
164 struct btrfs_delayed_ref_node,
165 rb_node);
166 if (tmp->bytenr < start)
167 break;
168 ref = tmp;
169 node = rb_prev(&ref->rb_node);
170 }
171 node = &ref->rb_node;
172 } else
173 node = rb_first(&delayed_refs->root);
174 }
175 again:
176 while (node && count < 32) {
177 ref = rb_entry(node, struct btrfs_delayed_ref_node, rb_node);
178 if (btrfs_delayed_ref_is_head(ref)) {
179 head = btrfs_delayed_node_to_head(ref);
180 if (list_empty(&head->cluster)) {
181 list_add_tail(&head->cluster, cluster);
182 delayed_refs->run_delayed_start =
183 head->node.bytenr;
184 count++;
185
186 WARN_ON(delayed_refs->num_heads_ready == 0);
187 delayed_refs->num_heads_ready--;
188 } else if (count) {
189 /* the goal of the clustering is to find extents
190 * that are likely to end up in the same extent
191 * leaf on disk. So, we don't want them spread
192 * all over the tree. Stop now if we've hit
193 * a head that was already in use
194 */
195 break;
196 }
197 }
198 node = rb_next(node);
199 }
200 if (count) {
201 return 0;
202 } else if (start) {
203 /*
204 * we've gone to the end of the rbtree without finding any
205 * clusters. start from the beginning and try again
206 */
207 start = 0;
208 node = rb_first(&delayed_refs->root);
209 goto again;
210 }
211 return 1;
212 }
213
214 /*
215 * This checks to see if there are any delayed refs in the
216 * btree for a given bytenr. It returns one if it finds any
217 * and zero otherwise.
218 *
219 * If it only finds a head node, it returns 0.
220 *
221 * The idea is to use this when deciding if you can safely delete an
222 * extent from the extent allocation tree. There may be a pending
223 * ref in the rbtree that adds or removes references, so as long as this
224 * returns one you need to leave the BTRFS_EXTENT_ITEM in the extent
225 * allocation tree.
226 */
227 int btrfs_delayed_ref_pending(struct btrfs_trans_handle *trans, u64 bytenr)
228 {
229 struct btrfs_delayed_ref_node *ref;
230 struct btrfs_delayed_ref_root *delayed_refs;
231 struct rb_node *prev_node;
232 int ret = 0;
233
234 delayed_refs = &trans->transaction->delayed_refs;
235 spin_lock(&delayed_refs->lock);
236
237 ref = tree_search(&delayed_refs->root, bytenr, (u64)-1, NULL);
238 if (ref) {
239 prev_node = rb_prev(&ref->rb_node);
240 if (!prev_node)
241 goto out;
242 ref = rb_entry(prev_node, struct btrfs_delayed_ref_node,
243 rb_node);
244 if (ref->bytenr == bytenr)
245 ret = 1;
246 }
247 out:
248 spin_unlock(&delayed_refs->lock);
249 return ret;
250 }
251
252 /*
253 * helper function to lookup reference count
254 *
255 * the head node for delayed ref is used to store the sum of all the
256 * reference count modifications queued up in the rbtree. This way you
257 * can check to see what the reference count would be if all of the
258 * delayed refs are processed.
259 */
260 int btrfs_lookup_extent_ref(struct btrfs_trans_handle *trans,
261 struct btrfs_root *root, u64 bytenr,
262 u64 num_bytes, u32 *refs)
263 {
264 struct btrfs_delayed_ref_node *ref;
265 struct btrfs_delayed_ref_head *head;
266 struct btrfs_delayed_ref_root *delayed_refs;
267 struct btrfs_path *path;
268 struct extent_buffer *leaf;
269 struct btrfs_extent_item *ei;
270 struct btrfs_key key;
271 u32 num_refs;
272 int ret;
273
274 path = btrfs_alloc_path();
275 if (!path)
276 return -ENOMEM;
277
278 key.objectid = bytenr;
279 key.type = BTRFS_EXTENT_ITEM_KEY;
280 key.offset = num_bytes;
281 delayed_refs = &trans->transaction->delayed_refs;
282 again:
283 ret = btrfs_search_slot(trans, root->fs_info->extent_root,
284 &key, path, 0, 0);
285 if (ret < 0)
286 goto out;
287
288 if (ret == 0) {
289 leaf = path->nodes[0];
290 ei = btrfs_item_ptr(leaf, path->slots[0],
291 struct btrfs_extent_item);
292 num_refs = btrfs_extent_refs(leaf, ei);
293 } else {
294 num_refs = 0;
295 ret = 0;
296 }
297
298 spin_lock(&delayed_refs->lock);
299 ref = tree_search(&delayed_refs->root, bytenr, (u64)-1, NULL);
300 if (ref) {
301 head = btrfs_delayed_node_to_head(ref);
302 if (mutex_trylock(&head->mutex)) {
303 num_refs += ref->ref_mod;
304 mutex_unlock(&head->mutex);
305 *refs = num_refs;
306 goto out;
307 }
308
309 atomic_inc(&ref->refs);
310 spin_unlock(&delayed_refs->lock);
311
312 btrfs_release_path(root->fs_info->extent_root, path);
313
314 mutex_lock(&head->mutex);
315 mutex_unlock(&head->mutex);
316 btrfs_put_delayed_ref(ref);
317 goto again;
318 } else {
319 *refs = num_refs;
320 }
321 out:
322 spin_unlock(&delayed_refs->lock);
323 btrfs_free_path(path);
324 return ret;
325 }
326
327 /*
328 * helper function to update an extent delayed ref in the
329 * rbtree. existing and update must both have the same
330 * bytenr and parent
331 *
332 * This may free existing if the update cancels out whatever
333 * operation it was doing.
334 */
335 static noinline void
336 update_existing_ref(struct btrfs_trans_handle *trans,
337 struct btrfs_delayed_ref_root *delayed_refs,
338 struct btrfs_delayed_ref_node *existing,
339 struct btrfs_delayed_ref_node *update)
340 {
341 struct btrfs_delayed_ref *existing_ref;
342 struct btrfs_delayed_ref *ref;
343
344 existing_ref = btrfs_delayed_node_to_ref(existing);
345 ref = btrfs_delayed_node_to_ref(update);
346
347 if (ref->pin)
348 existing_ref->pin = 1;
349
350 if (ref->action != existing_ref->action) {
351 /*
352 * this is effectively undoing either an add or a
353 * drop. We decrement the ref_mod, and if it goes
354 * down to zero we just delete the entry without
355 * every changing the extent allocation tree.
356 */
357 existing->ref_mod--;
358 if (existing->ref_mod == 0) {
359 rb_erase(&existing->rb_node,
360 &delayed_refs->root);
361 existing->in_tree = 0;
362 btrfs_put_delayed_ref(existing);
363 delayed_refs->num_entries--;
364 if (trans->delayed_ref_updates)
365 trans->delayed_ref_updates--;
366 }
367 } else {
368 if (existing_ref->action == BTRFS_ADD_DELAYED_REF) {
369 /* if we're adding refs, make sure all the
370 * details match up. The extent could
371 * have been totally freed and reallocated
372 * by a different owner before the delayed
373 * ref entries were removed.
374 */
375 existing_ref->owner_objectid = ref->owner_objectid;
376 existing_ref->generation = ref->generation;
377 existing_ref->root = ref->root;
378 existing->num_bytes = update->num_bytes;
379 }
380 /*
381 * the action on the existing ref matches
382 * the action on the ref we're trying to add.
383 * Bump the ref_mod by one so the backref that
384 * is eventually added/removed has the correct
385 * reference count
386 */
387 existing->ref_mod += update->ref_mod;
388 }
389 }
390
391 /*
392 * helper function to update the accounting in the head ref
393 * existing and update must have the same bytenr
394 */
395 static noinline void
396 update_existing_head_ref(struct btrfs_delayed_ref_node *existing,
397 struct btrfs_delayed_ref_node *update)
398 {
399 struct btrfs_delayed_ref_head *existing_ref;
400 struct btrfs_delayed_ref_head *ref;
401
402 existing_ref = btrfs_delayed_node_to_head(existing);
403 ref = btrfs_delayed_node_to_head(update);
404
405 if (ref->must_insert_reserved) {
406 /* if the extent was freed and then
407 * reallocated before the delayed ref
408 * entries were processed, we can end up
409 * with an existing head ref without
410 * the must_insert_reserved flag set.
411 * Set it again here
412 */
413 existing_ref->must_insert_reserved = ref->must_insert_reserved;
414
415 /*
416 * update the num_bytes so we make sure the accounting
417 * is done correctly
418 */
419 existing->num_bytes = update->num_bytes;
420
421 }
422
423 /*
424 * update the reference mod on the head to reflect this new operation
425 */
426 existing->ref_mod += update->ref_mod;
427 }
428
429 /*
430 * helper function to actually insert a delayed ref into the rbtree.
431 * this does all the dirty work in terms of maintaining the correct
432 * overall modification count in the head node and properly dealing
433 * with updating existing nodes as new modifications are queued.
434 */
435 static noinline int __btrfs_add_delayed_ref(struct btrfs_trans_handle *trans,
436 struct btrfs_delayed_ref_node *ref,
437 u64 bytenr, u64 num_bytes, u64 parent, u64 ref_root,
438 u64 ref_generation, u64 owner_objectid, int action,
439 int pin)
440 {
441 struct btrfs_delayed_ref_node *existing;
442 struct btrfs_delayed_ref *full_ref;
443 struct btrfs_delayed_ref_head *head_ref = NULL;
444 struct btrfs_delayed_ref_root *delayed_refs;
445 int count_mod = 1;
446 int must_insert_reserved = 0;
447
448 /*
449 * the head node stores the sum of all the mods, so dropping a ref
450 * should drop the sum in the head node by one.
451 */
452 if (parent == (u64)-1) {
453 if (action == BTRFS_DROP_DELAYED_REF)
454 count_mod = -1;
455 else if (action == BTRFS_UPDATE_DELAYED_HEAD)
456 count_mod = 0;
457 }
458
459 /*
460 * BTRFS_ADD_DELAYED_EXTENT means that we need to update
461 * the reserved accounting when the extent is finally added, or
462 * if a later modification deletes the delayed ref without ever
463 * inserting the extent into the extent allocation tree.
464 * ref->must_insert_reserved is the flag used to record
465 * that accounting mods are required.
466 *
467 * Once we record must_insert_reserved, switch the action to
468 * BTRFS_ADD_DELAYED_REF because other special casing is not required.
469 */
470 if (action == BTRFS_ADD_DELAYED_EXTENT) {
471 must_insert_reserved = 1;
472 action = BTRFS_ADD_DELAYED_REF;
473 } else {
474 must_insert_reserved = 0;
475 }
476
477
478 delayed_refs = &trans->transaction->delayed_refs;
479
480 /* first set the basic ref node struct up */
481 atomic_set(&ref->refs, 1);
482 ref->bytenr = bytenr;
483 ref->parent = parent;
484 ref->ref_mod = count_mod;
485 ref->in_tree = 1;
486 ref->num_bytes = num_bytes;
487
488 if (btrfs_delayed_ref_is_head(ref)) {
489 head_ref = btrfs_delayed_node_to_head(ref);
490 head_ref->must_insert_reserved = must_insert_reserved;
491 INIT_LIST_HEAD(&head_ref->cluster);
492 mutex_init(&head_ref->mutex);
493 } else {
494 full_ref = btrfs_delayed_node_to_ref(ref);
495 full_ref->root = ref_root;
496 full_ref->generation = ref_generation;
497 full_ref->owner_objectid = owner_objectid;
498 full_ref->pin = pin;
499 full_ref->action = action;
500 }
501
502 existing = tree_insert(&delayed_refs->root, bytenr,
503 parent, &ref->rb_node);
504
505 if (existing) {
506 if (btrfs_delayed_ref_is_head(ref))
507 update_existing_head_ref(existing, ref);
508 else
509 update_existing_ref(trans, delayed_refs, existing, ref);
510
511 /*
512 * we've updated the existing ref, free the newly
513 * allocated ref
514 */
515 kfree(ref);
516 } else {
517 if (btrfs_delayed_ref_is_head(ref)) {
518 delayed_refs->num_heads++;
519 delayed_refs->num_heads_ready++;
520 }
521 delayed_refs->num_entries++;
522 trans->delayed_ref_updates++;
523 }
524 return 0;
525 }
526
527 /*
528 * add a delayed ref to the tree. This does all of the accounting required
529 * to make sure the delayed ref is eventually processed before this
530 * transaction commits.
531 */
532 int btrfs_add_delayed_ref(struct btrfs_trans_handle *trans,
533 u64 bytenr, u64 num_bytes, u64 parent, u64 ref_root,
534 u64 ref_generation, u64 owner_objectid, int action,
535 int pin)
536 {
537 struct btrfs_delayed_ref *ref;
538 struct btrfs_delayed_ref_head *head_ref;
539 struct btrfs_delayed_ref_root *delayed_refs;
540 int ret;
541
542 ref = kmalloc(sizeof(*ref), GFP_NOFS);
543 if (!ref)
544 return -ENOMEM;
545
546 /*
547 * the parent = 0 case comes from cases where we don't actually
548 * know the parent yet. It will get updated later via a add/drop
549 * pair.
550 */
551 if (parent == 0)
552 parent = bytenr;
553
554 head_ref = kmalloc(sizeof(*head_ref), GFP_NOFS);
555 if (!head_ref) {
556 kfree(ref);
557 return -ENOMEM;
558 }
559 delayed_refs = &trans->transaction->delayed_refs;
560 spin_lock(&delayed_refs->lock);
561
562 /*
563 * insert both the head node and the new ref without dropping
564 * the spin lock
565 */
566 ret = __btrfs_add_delayed_ref(trans, &head_ref->node, bytenr, num_bytes,
567 (u64)-1, 0, 0, 0, action, pin);
568 BUG_ON(ret);
569
570 ret = __btrfs_add_delayed_ref(trans, &ref->node, bytenr, num_bytes,
571 parent, ref_root, ref_generation,
572 owner_objectid, action, pin);
573 BUG_ON(ret);
574 spin_unlock(&delayed_refs->lock);
575 return 0;
576 }
577
578 /*
579 * this does a simple search for the head node for a given extent.
580 * It must be called with the delayed ref spinlock held, and it returns
581 * the head node if any where found, or NULL if not.
582 */
583 struct btrfs_delayed_ref_head *
584 btrfs_find_delayed_ref_head(struct btrfs_trans_handle *trans, u64 bytenr)
585 {
586 struct btrfs_delayed_ref_node *ref;
587 struct btrfs_delayed_ref_root *delayed_refs;
588
589 delayed_refs = &trans->transaction->delayed_refs;
590 ref = tree_search(&delayed_refs->root, bytenr, (u64)-1, NULL);
591 if (ref)
592 return btrfs_delayed_node_to_head(ref);
593 return NULL;
594 }
595
596 /*
597 * add a delayed ref to the tree. This does all of the accounting required
598 * to make sure the delayed ref is eventually processed before this
599 * transaction commits.
600 *
601 * The main point of this call is to add and remove a backreference in a single
602 * shot, taking the lock only once, and only searching for the head node once.
603 *
604 * It is the same as doing a ref add and delete in two separate calls.
605 */
606 int btrfs_update_delayed_ref(struct btrfs_trans_handle *trans,
607 u64 bytenr, u64 num_bytes, u64 orig_parent,
608 u64 parent, u64 orig_ref_root, u64 ref_root,
609 u64 orig_ref_generation, u64 ref_generation,
610 u64 owner_objectid, int pin)
611 {
612 struct btrfs_delayed_ref *ref;
613 struct btrfs_delayed_ref *old_ref;
614 struct btrfs_delayed_ref_head *head_ref;
615 struct btrfs_delayed_ref_root *delayed_refs;
616 int ret;
617
618 ref = kmalloc(sizeof(*ref), GFP_NOFS);
619 if (!ref)
620 return -ENOMEM;
621
622 old_ref = kmalloc(sizeof(*old_ref), GFP_NOFS);
623 if (!old_ref) {
624 kfree(ref);
625 return -ENOMEM;
626 }
627
628 /*
629 * the parent = 0 case comes from cases where we don't actually
630 * know the parent yet. It will get updated later via a add/drop
631 * pair.
632 */
633 if (parent == 0)
634 parent = bytenr;
635 if (orig_parent == 0)
636 orig_parent = bytenr;
637
638 head_ref = kmalloc(sizeof(*head_ref), GFP_NOFS);
639 if (!head_ref) {
640 kfree(ref);
641 kfree(old_ref);
642 return -ENOMEM;
643 }
644 delayed_refs = &trans->transaction->delayed_refs;
645 spin_lock(&delayed_refs->lock);
646
647 /*
648 * insert both the head node and the new ref without dropping
649 * the spin lock
650 */
651 ret = __btrfs_add_delayed_ref(trans, &head_ref->node, bytenr, num_bytes,
652 (u64)-1, 0, 0, 0,
653 BTRFS_UPDATE_DELAYED_HEAD, 0);
654 BUG_ON(ret);
655
656 ret = __btrfs_add_delayed_ref(trans, &ref->node, bytenr, num_bytes,
657 parent, ref_root, ref_generation,
658 owner_objectid, BTRFS_ADD_DELAYED_REF, 0);
659 BUG_ON(ret);
660
661 ret = __btrfs_add_delayed_ref(trans, &old_ref->node, bytenr, num_bytes,
662 orig_parent, orig_ref_root,
663 orig_ref_generation, owner_objectid,
664 BTRFS_DROP_DELAYED_REF, pin);
665 BUG_ON(ret);
666 spin_unlock(&delayed_refs->lock);
667 return 0;
668 }
This page took 0.045857 seconds and 5 git commands to generate.