?  search:
summary | log | graphiclog1 | graphiclog2 | commit | commitdiff | tree | refs | edit | fork
blame | history | raw | HEAD
Btrfs: add inodes before dropping the extent lock in find_all_leafs
[linux-2.6/btrfs-unstable.git] / fs / btrfs / backref.c
blob3f75895c919bcc3b80ae63ab1fca42dcf335f95b
1 /*
2 * Copyright (C) 2011 STRATO. 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 "ctree.h"
20 #include "disk-io.h"
21 #include "backref.h"
22 #include "ulist.h"
23 #include "transaction.h"
24 #include "delayed-ref.h"
25 #include "locking.h"
26
27 struct extent_inode_elem {
28 u64 inum;
29 u64 offset;
30 struct extent_inode_elem *next;
31 };
32
33 static int check_extent_in_eb(struct btrfs_key *key, struct extent_buffer *eb,
34 struct btrfs_file_extent_item *fi,
35 u64 extent_item_pos,
36 struct extent_inode_elem **eie)
37 {
38 u64 data_offset;
39 u64 data_len;
40 struct extent_inode_elem *e;
41
42 data_offset = btrfs_file_extent_offset(eb, fi);
43 data_len = btrfs_file_extent_num_bytes(eb, fi);
44
45 if (extent_item_pos < data_offset ||
46 extent_item_pos >= data_offset + data_len)
47 return 1;
48
49 e = kmalloc(sizeof(*e), GFP_NOFS);
50 if (!e)
51 return -ENOMEM;
52
53 e->next = *eie;
54 e->inum = key->objectid;
55 e->offset = key->offset + (extent_item_pos - data_offset);
56 *eie = e;
57
58 return 0;
59 }
60
61 static int find_extent_in_eb(struct extent_buffer *eb, u64 wanted_disk_byte,
62 u64 extent_item_pos,
63 struct extent_inode_elem **eie)
64 {
65 u64 disk_byte;
66 struct btrfs_key key;
67 struct btrfs_file_extent_item *fi;
68 int slot;
69 int nritems;
70 int extent_type;
71 int ret;
72
73 /*
74 * from the shared data ref, we only have the leaf but we need
75 * the key. thus, we must look into all items and see that we
76 * find one (some) with a reference to our extent item.
77 */
78 nritems = btrfs_header_nritems(eb);
79 for (slot = 0; slot < nritems; ++slot) {
80 btrfs_item_key_to_cpu(eb, &key, slot);
81 if (key.type != BTRFS_EXTENT_DATA_KEY)
82 continue;
83 fi = btrfs_item_ptr(eb, slot, struct btrfs_file_extent_item);
84 extent_type = btrfs_file_extent_type(eb, fi);
85 if (extent_type == BTRFS_FILE_EXTENT_INLINE)
86 continue;
87 /* don't skip BTRFS_FILE_EXTENT_PREALLOC, we can handle that */
88 disk_byte = btrfs_file_extent_disk_bytenr(eb, fi);
89 if (disk_byte != wanted_disk_byte)
90 continue;
91
92 ret = check_extent_in_eb(&key, eb, fi, extent_item_pos, eie);
93 if (ret < 0)
94 return ret;
95 }
96
97 return 0;
98 }
99
100 /*
101 * this structure records all encountered refs on the way up to the root
102 */
103 struct __prelim_ref {
104 struct list_head list;
105 u64 root_id;
106 struct btrfs_key key_for_search;
107 int level;
108 int count;
109 struct extent_inode_elem *inode_list;
110 u64 parent;
111 u64 wanted_disk_byte;
112 };
113
114 /*
115 * the rules for all callers of this function are:
116 * - obtaining the parent is the goal
117 * - if you add a key, you must know that it is a correct key
118 * - if you cannot add the parent or a correct key, then we will look into the
119 * block later to set a correct key
120 *
121 * delayed refs
122 * ============
123 * backref type | shared | indirect | shared | indirect
124 * information | tree | tree | data | data
125 * --------------------+--------+----------+--------+----------
126 * parent logical | y | - | - | -
127 * key to resolve | - | y | y | y
128 * tree block logical | - | - | - | -
129 * root for resolving | y | y | y | y
130 *
131 * - column 1: we've the parent -> done
132 * - column 2, 3, 4: we use the key to find the parent
133 *
134 * on disk refs (inline or keyed)
135 * ==============================
136 * backref type | shared | indirect | shared | indirect
137 * information | tree | tree | data | data
138 * --------------------+--------+----------+--------+----------
139 * parent logical | y | - | y | -
140 * key to resolve | - | - | - | y
141 * tree block logical | y | y | y | y
142 * root for resolving | - | y | y | y
143 *
144 * - column 1, 3: we've the parent -> done
145 * - column 2: we take the first key from the block to find the parent
146 * (see __add_missing_keys)
147 * - column 4: we use the key to find the parent
148 *
149 * additional information that's available but not required to find the parent
150 * block might help in merging entries to gain some speed.
151 */
152
153 static int __add_prelim_ref(struct list_head *head, u64 root_id,
154 struct btrfs_key *key, int level,
155 u64 parent, u64 wanted_disk_byte, int count)
156 {
157 struct __prelim_ref *ref;
158
159 /* in case we're adding delayed refs, we're holding the refs spinlock */
160 ref = kmalloc(sizeof(*ref), GFP_ATOMIC);
161 if (!ref)
162 return -ENOMEM;
163
164 ref->root_id = root_id;
165 if (key)
166 ref->key_for_search = *key;
167 else
168 memset(&ref->key_for_search, 0, sizeof(ref->key_for_search));
169
170 ref->inode_list = NULL;
171 ref->level = level;
172 ref->count = count;
173 ref->parent = parent;
174 ref->wanted_disk_byte = wanted_disk_byte;
175 list_add_tail(&ref->list, head);
176
177 return 0;
178 }
179
180 static int add_all_parents(struct btrfs_root *root, struct btrfs_path *path,
181 struct ulist *parents, int level,
182 struct btrfs_key *key, u64 wanted_disk_byte,
183 const u64 *extent_item_pos)
184 {
185 int ret;
186 int slot = path->slots[level];
187 struct extent_buffer *eb = path->nodes[level];
188 struct btrfs_file_extent_item *fi;
189 struct extent_inode_elem *eie = NULL;
190 u64 disk_byte;
191 u64 wanted_objectid = key->objectid;
192
193 add_parent:
194 if (level == 0 && extent_item_pos) {
195 fi = btrfs_item_ptr(eb, slot, struct btrfs_file_extent_item);
196 ret = check_extent_in_eb(key, eb, fi, *extent_item_pos, &eie);
197 if (ret < 0)
198 return ret;
199 }
200 ret = ulist_add(parents, eb->start, (unsigned long)eie, GFP_NOFS);
201 if (ret < 0)
202 return ret;
203
204 if (level != 0)
205 return 0;
206
207 /*
208 * if the current leaf is full with EXTENT_DATA items, we must
209 * check the next one if that holds a reference as well.
210 * ref->count cannot be used to skip this check.
211 * repeat this until we don't find any additional EXTENT_DATA items.
212 */
213 while (1) {
214 eie = NULL;
215 ret = btrfs_next_leaf(root, path);
216 if (ret < 0)
217 return ret;
218 if (ret)
219 return 0;
220
221 eb = path->nodes[0];
222 for (slot = 0; slot < btrfs_header_nritems(eb); ++slot) {
223 btrfs_item_key_to_cpu(eb, key, slot);
224 if (key->objectid != wanted_objectid ||
225 key->type != BTRFS_EXTENT_DATA_KEY)
226 return 0;
227 fi = btrfs_item_ptr(eb, slot,
228 struct btrfs_file_extent_item);
229 disk_byte = btrfs_file_extent_disk_bytenr(eb, fi);
230 if (disk_byte == wanted_disk_byte)
231 goto add_parent;
232 }
233 }
234
235 return 0;
236 }
237
238 /*
239 * resolve an indirect backref in the form (root_id, key, level)
240 * to a logical address
241 */
242 static int __resolve_indirect_ref(struct btrfs_fs_info *fs_info,
243 int search_commit_root,
244 u64 time_seq,
245 struct __prelim_ref *ref,
246 struct ulist *parents,
247 const u64 *extent_item_pos)
248 {
249 struct btrfs_path *path;
250 struct btrfs_root *root;
251 struct btrfs_key root_key;
252 struct btrfs_key key = {0};
253 struct extent_buffer *eb;
254 int ret = 0;
255 int root_level;
256 int level = ref->level;
257
258 path = btrfs_alloc_path();
259 if (!path)
260 return -ENOMEM;
261 path->search_commit_root = !!search_commit_root;
262
263 root_key.objectid = ref->root_id;
264 root_key.type = BTRFS_ROOT_ITEM_KEY;
265 root_key.offset = (u64)-1;
266 root = btrfs_read_fs_root_no_name(fs_info, &root_key);
267 if (IS_ERR(root)) {
268 ret = PTR_ERR(root);
269 goto out;
270 }
271
272 rcu_read_lock();
273 root_level = btrfs_header_level(root->node);
274 rcu_read_unlock();
275
276 if (root_level + 1 == level)
277 goto out;
278
279 path->lowest_level = level;
280 ret = btrfs_search_old_slot(root, &ref->key_for_search, path, time_seq);
281 pr_debug("search slot in root %llu (level %d, ref count %d) returned "
282 "%d for key (%llu %u %llu)\n",
283 (unsigned long long)ref->root_id, level, ref->count, ret,
284 (unsigned long long)ref->key_for_search.objectid,
285 ref->key_for_search.type,
286 (unsigned long long)ref->key_for_search.offset);
287 if (ret < 0)
288 goto out;
289
290 eb = path->nodes[level];
291 if (!eb) {
292 WARN_ON(1);
293 ret = 1;
294 goto out;
295 }
296
297 if (level == 0) {
298 if (ret == 1 && path->slots[0] >= btrfs_header_nritems(eb)) {
299 ret = btrfs_next_leaf(root, path);
300 if (ret)
301 goto out;
302 eb = path->nodes[0];
303 }
304
305 btrfs_item_key_to_cpu(eb, &key, path->slots[0]);
306 }
307
308 ret = add_all_parents(root, path, parents, level, &key,
309 ref->wanted_disk_byte, extent_item_pos);
310 out:
311 btrfs_free_path(path);
312 return ret;
313 }
314
315 /*
316 * resolve all indirect backrefs from the list
317 */
318 static int __resolve_indirect_refs(struct btrfs_fs_info *fs_info,
319 int search_commit_root, u64 time_seq,
320 struct list_head *head,
321 const u64 *extent_item_pos)
322 {
323 int err;
324 int ret = 0;
325 struct __prelim_ref *ref;
326 struct __prelim_ref *ref_safe;
327 struct __prelim_ref *new_ref;
328 struct ulist *parents;
329 struct ulist_node *node;
330 struct ulist_iterator uiter;
331
332 parents = ulist_alloc(GFP_NOFS);
333 if (!parents)
334 return -ENOMEM;
335
336 /*
337 * _safe allows us to insert directly after the current item without
338 * iterating over the newly inserted items.
339 * we're also allowed to re-assign ref during iteration.
340 */
341 list_for_each_entry_safe(ref, ref_safe, head, list) {
342 if (ref->parent) /* already direct */
343 continue;
344 if (ref->count == 0)
345 continue;
346 err = __resolve_indirect_ref(fs_info, search_commit_root,
347 time_seq, ref, parents,
348 extent_item_pos);
349 if (err) {
350 if (ret == 0)
351 ret = err;
352 continue;
353 }
354
355 /* we put the first parent into the ref at hand */
356 ULIST_ITER_INIT(&uiter);
357 node = ulist_next(parents, &uiter);
358 ref->parent = node ? node->val : 0;
359 ref->inode_list =
360 node ? (struct extent_inode_elem *)node->aux : 0;
361
362 /* additional parents require new refs being added here */
363 while ((node = ulist_next(parents, &uiter))) {
364 new_ref = kmalloc(sizeof(*new_ref), GFP_NOFS);
365 if (!new_ref) {
366 ret = -ENOMEM;
367 break;
368 }
369 memcpy(new_ref, ref, sizeof(*ref));
370 new_ref->parent = node->val;
371 new_ref->inode_list =
372 (struct extent_inode_elem *)node->aux;
373 list_add(&new_ref->list, &ref->list);
374 }
375 ulist_reinit(parents);
376 }
377
378 ulist_free(parents);
379 return ret;
380 }
381
382 static inline int ref_for_same_block(struct __prelim_ref *ref1,
383 struct __prelim_ref *ref2)
384 {
385 if (ref1->level != ref2->level)
386 return 0;
387 if (ref1->root_id != ref2->root_id)
388 return 0;
389 if (ref1->key_for_search.type != ref2->key_for_search.type)
390 return 0;
391 if (ref1->key_for_search.objectid != ref2->key_for_search.objectid)
392 return 0;
393 if (ref1->key_for_search.offset != ref2->key_for_search.offset)
394 return 0;
395 if (ref1->parent != ref2->parent)
396 return 0;
397
398 return 1;
399 }
400
401 /*
402 * read tree blocks and add keys where required.
403 */
404 static int __add_missing_keys(struct btrfs_fs_info *fs_info,
405 struct list_head *head)
406 {
407 struct list_head *pos;
408 struct extent_buffer *eb;
409
410 list_for_each(pos, head) {
411 struct __prelim_ref *ref;
412 ref = list_entry(pos, struct __prelim_ref, list);
413
414 if (ref->parent)
415 continue;
416 if (ref->key_for_search.type)
417 continue;
418 BUG_ON(!ref->wanted_disk_byte);
419 eb = read_tree_block(fs_info->tree_root, ref->wanted_disk_byte,
420 fs_info->tree_root->leafsize, 0);
421 BUG_ON(!eb);
422 btrfs_tree_read_lock(eb);
423 if (btrfs_header_level(eb) == 0)
424 btrfs_item_key_to_cpu(eb, &ref->key_for_search, 0);
425 else
426 btrfs_node_key_to_cpu(eb, &ref->key_for_search, 0);
427 btrfs_tree_read_unlock(eb);
428 free_extent_buffer(eb);
429 }
430 return 0;
431 }
432
433 /*
434 * merge two lists of backrefs and adjust counts accordingly
435 *
436 * mode = 1: merge identical keys, if key is set
437 * FIXME: if we add more keys in __add_prelim_ref, we can merge more here.
438 * additionally, we could even add a key range for the blocks we
439 * looked into to merge even more (-> replace unresolved refs by those
440 * having a parent).
441 * mode = 2: merge identical parents
442 */
443 static int __merge_refs(struct list_head *head, int mode)
444 {
445 struct list_head *pos1;
446
447 list_for_each(pos1, head) {
448 struct list_head *n2;
449 struct list_head *pos2;
450 struct __prelim_ref *ref1;
451
452 ref1 = list_entry(pos1, struct __prelim_ref, list);
453
454 for (pos2 = pos1->next, n2 = pos2->next; pos2 != head;
455 pos2 = n2, n2 = pos2->next) {
456 struct __prelim_ref *ref2;
457 struct __prelim_ref *xchg;
458
459 ref2 = list_entry(pos2, struct __prelim_ref, list);
460
461 if (mode == 1) {
462 if (!ref_for_same_block(ref1, ref2))
463 continue;
464 if (!ref1->parent && ref2->parent) {
465 xchg = ref1;
466 ref1 = ref2;
467 ref2 = xchg;
468 }
469 ref1->count += ref2->count;
470 } else {
471 if (ref1->parent != ref2->parent)
472 continue;
473 ref1->count += ref2->count;
474 }
475 list_del(&ref2->list);
476 kfree(ref2);
477 }
478
479 }
480 return 0;
481 }
482
483 /*
484 * add all currently queued delayed refs from this head whose seq nr is
485 * smaller or equal that seq to the list
486 */
487 static int __add_delayed_refs(struct btrfs_delayed_ref_head *head, u64 seq,
488 struct list_head *prefs)
489 {
490 struct btrfs_delayed_extent_op *extent_op = head->extent_op;
491 struct rb_node *n = &head->node.rb_node;
492 struct btrfs_key key;
493 struct btrfs_key op_key = {0};
494 int sgn;
495 int ret = 0;
496
497 if (extent_op && extent_op->update_key)
498 btrfs_disk_key_to_cpu(&op_key, &extent_op->key);
499
500 while ((n = rb_prev(n))) {
501 struct btrfs_delayed_ref_node *node;
502 node = rb_entry(n, struct btrfs_delayed_ref_node,
503 rb_node);
504 if (node->bytenr != head->node.bytenr)
505 break;
506 WARN_ON(node->is_head);
507
508 if (node->seq > seq)
509 continue;
510
511 switch (node->action) {
512 case BTRFS_ADD_DELAYED_EXTENT:
513 case BTRFS_UPDATE_DELAYED_HEAD:
514 WARN_ON(1);
515 continue;
516 case BTRFS_ADD_DELAYED_REF:
517 sgn = 1;
518 break;
519 case BTRFS_DROP_DELAYED_REF:
520 sgn = -1;
521 break;
522 default:
523 BUG_ON(1);
524 }
525 switch (node->type) {
526 case BTRFS_TREE_BLOCK_REF_KEY: {
527 struct btrfs_delayed_tree_ref *ref;
528
529 ref = btrfs_delayed_node_to_tree_ref(node);
530 ret = __add_prelim_ref(prefs, ref->root, &op_key,
531 ref->level + 1, 0, node->bytenr,
532 node->ref_mod * sgn);
533 break;
534 }
535 case BTRFS_SHARED_BLOCK_REF_KEY: {
536 struct btrfs_delayed_tree_ref *ref;
537
538 ref = btrfs_delayed_node_to_tree_ref(node);
539 ret = __add_prelim_ref(prefs, ref->root, NULL,
540 ref->level + 1, ref->parent,
541 node->bytenr,
542 node->ref_mod * sgn);
543 break;
544 }
545 case BTRFS_EXTENT_DATA_REF_KEY: {
546 struct btrfs_delayed_data_ref *ref;
547 ref = btrfs_delayed_node_to_data_ref(node);
548
549 key.objectid = ref->objectid;
550 key.type = BTRFS_EXTENT_DATA_KEY;
551 key.offset = ref->offset;
552 ret = __add_prelim_ref(prefs, ref->root, &key, 0, 0,
553 node->bytenr,
554 node->ref_mod * sgn);
555 break;
556 }
557 case BTRFS_SHARED_DATA_REF_KEY: {
558 struct btrfs_delayed_data_ref *ref;
559
560 ref = btrfs_delayed_node_to_data_ref(node);
561
562 key.objectid = ref->objectid;
563 key.type = BTRFS_EXTENT_DATA_KEY;
564 key.offset = ref->offset;
565 ret = __add_prelim_ref(prefs, ref->root, &key, 0,
566 ref->parent, node->bytenr,
567 node->ref_mod * sgn);
568 break;
569 }
570 default:
571 WARN_ON(1);
572 }
573 BUG_ON(ret);
574 }
575
576 return 0;
577 }
578
579 /*
580 * add all inline backrefs for bytenr to the list
581 */
582 static int __add_inline_refs(struct btrfs_fs_info *fs_info,
583 struct btrfs_path *path, u64 bytenr,
584 int *info_level, struct list_head *prefs)
585 {
586 int ret = 0;
587 int slot;
588 struct extent_buffer *leaf;
589 struct btrfs_key key;
590 unsigned long ptr;
591 unsigned long end;
592 struct btrfs_extent_item *ei;
593 u64 flags;
594 u64 item_size;
595
596 /*
597 * enumerate all inline refs
598 */
599 leaf = path->nodes[0];
600 slot = path->slots[0];
601
602 item_size = btrfs_item_size_nr(leaf, slot);
603 BUG_ON(item_size < sizeof(*ei));
604
605 ei = btrfs_item_ptr(leaf, slot, struct btrfs_extent_item);
606 flags = btrfs_extent_flags(leaf, ei);
607
608 ptr = (unsigned long)(ei + 1);
609 end = (unsigned long)ei + item_size;
610
611 if (flags & BTRFS_EXTENT_FLAG_TREE_BLOCK) {
612 struct btrfs_tree_block_info *info;
613
614 info = (struct btrfs_tree_block_info *)ptr;
615 *info_level = btrfs_tree_block_level(leaf, info);
616 ptr += sizeof(struct btrfs_tree_block_info);
617 BUG_ON(ptr > end);
618 } else {
619 BUG_ON(!(flags & BTRFS_EXTENT_FLAG_DATA));
620 }
621
622 while (ptr < end) {
623 struct btrfs_extent_inline_ref *iref;
624 u64 offset;
625 int type;
626
627 iref = (struct btrfs_extent_inline_ref *)ptr;
628 type = btrfs_extent_inline_ref_type(leaf, iref);
629 offset = btrfs_extent_inline_ref_offset(leaf, iref);
630
631 switch (type) {
632 case BTRFS_SHARED_BLOCK_REF_KEY:
633 ret = __add_prelim_ref(prefs, 0, NULL,
634 *info_level + 1, offset,
635 bytenr, 1);
636 break;
637 case BTRFS_SHARED_DATA_REF_KEY: {
638 struct btrfs_shared_data_ref *sdref;
639 int count;
640
641 sdref = (struct btrfs_shared_data_ref *)(iref + 1);
642 count = btrfs_shared_data_ref_count(leaf, sdref);
643 ret = __add_prelim_ref(prefs, 0, NULL, 0, offset,
644 bytenr, count);
645 break;
646 }
647 case BTRFS_TREE_BLOCK_REF_KEY:
648 ret = __add_prelim_ref(prefs, offset, NULL,
649 *info_level + 1, 0,
650 bytenr, 1);
651 break;
652 case BTRFS_EXTENT_DATA_REF_KEY: {
653 struct btrfs_extent_data_ref *dref;
654 int count;
655 u64 root;
656
657 dref = (struct btrfs_extent_data_ref *)(&iref->offset);
658 count = btrfs_extent_data_ref_count(leaf, dref);
659 key.objectid = btrfs_extent_data_ref_objectid(leaf,
660 dref);
661 key.type = BTRFS_EXTENT_DATA_KEY;
662 key.offset = btrfs_extent_data_ref_offset(leaf, dref);
663 root = btrfs_extent_data_ref_root(leaf, dref);
664 ret = __add_prelim_ref(prefs, root, &key, 0, 0,
665 bytenr, count);
666 break;
667 }
668 default:
669 WARN_ON(1);
670 }
671 BUG_ON(ret);
672 ptr += btrfs_extent_inline_ref_size(type);
673 }
674
675 return 0;
676 }
677
678 /*
679 * add all non-inline backrefs for bytenr to the list
680 */
681 static int __add_keyed_refs(struct btrfs_fs_info *fs_info,
682 struct btrfs_path *path, u64 bytenr,
683 int info_level, struct list_head *prefs)
684 {
685 struct btrfs_root *extent_root = fs_info->extent_root;
686 int ret;
687 int slot;
688 struct extent_buffer *leaf;
689 struct btrfs_key key;
690
691 while (1) {
692 ret = btrfs_next_item(extent_root, path);
693 if (ret < 0)
694 break;
695 if (ret) {
696 ret = 0;
697 break;
698 }
699
700 slot = path->slots[0];
701 leaf = path->nodes[0];
702 btrfs_item_key_to_cpu(leaf, &key, slot);
703
704 if (key.objectid != bytenr)
705 break;
706 if (key.type < BTRFS_TREE_BLOCK_REF_KEY)
707 continue;
708 if (key.type > BTRFS_SHARED_DATA_REF_KEY)
709 break;
710
711 switch (key.type) {
712 case BTRFS_SHARED_BLOCK_REF_KEY:
713 ret = __add_prelim_ref(prefs, 0, NULL,
714 info_level + 1, key.offset,
715 bytenr, 1);
716 break;
717 case BTRFS_SHARED_DATA_REF_KEY: {
718 struct btrfs_shared_data_ref *sdref;
719 int count;
720
721 sdref = btrfs_item_ptr(leaf, slot,
722 struct btrfs_shared_data_ref);
723 count = btrfs_shared_data_ref_count(leaf, sdref);
724 ret = __add_prelim_ref(prefs, 0, NULL, 0, key.offset,
725 bytenr, count);
726 break;
727 }
728 case BTRFS_TREE_BLOCK_REF_KEY:
729 ret = __add_prelim_ref(prefs, key.offset, NULL,
730 info_level + 1, 0,
731 bytenr, 1);
732 break;
733 case BTRFS_EXTENT_DATA_REF_KEY: {
734 struct btrfs_extent_data_ref *dref;
735 int count;
736 u64 root;
737
738 dref = btrfs_item_ptr(leaf, slot,
739 struct btrfs_extent_data_ref);
740 count = btrfs_extent_data_ref_count(leaf, dref);
741 key.objectid = btrfs_extent_data_ref_objectid(leaf,
742 dref);
743 key.type = BTRFS_EXTENT_DATA_KEY;
744 key.offset = btrfs_extent_data_ref_offset(leaf, dref);
745 root = btrfs_extent_data_ref_root(leaf, dref);
746 ret = __add_prelim_ref(prefs, root, &key, 0, 0,
747 bytenr, count);
748 break;
749 }
750 default:
751 WARN_ON(1);
752 }
753 BUG_ON(ret);
754 }
755
756 return ret;
757 }
758
759 /*
760 * this adds all existing backrefs (inline backrefs, backrefs and delayed
761 * refs) for the given bytenr to the refs list, merges duplicates and resolves
762 * indirect refs to their parent bytenr.
763 * When roots are found, they're added to the roots list
764 *
765 * FIXME some caching might speed things up
766 */
767 static int find_parent_nodes(struct btrfs_trans_handle *trans,
768 struct btrfs_fs_info *fs_info, u64 bytenr,
769 u64 delayed_ref_seq, u64 time_seq,
770 struct ulist *refs, struct ulist *roots,
771 const u64 *extent_item_pos)
772 {
773 struct btrfs_key key;
774 struct btrfs_path *path;
775 struct btrfs_delayed_ref_root *delayed_refs = NULL;
776 struct btrfs_delayed_ref_head *head;
777 int info_level = 0;
778 int ret;
779 int search_commit_root = (trans == BTRFS_BACKREF_SEARCH_COMMIT_ROOT);
780 struct list_head prefs_delayed;
781 struct list_head prefs;
782 struct __prelim_ref *ref;
783
784 INIT_LIST_HEAD(&prefs);
785 INIT_LIST_HEAD(&prefs_delayed);
786
787 key.objectid = bytenr;
788 key.type = BTRFS_EXTENT_ITEM_KEY;
789 key.offset = (u64)-1;
790
791 path = btrfs_alloc_path();
792 if (!path)
793 return -ENOMEM;
794 path->search_commit_root = !!search_commit_root;
795
796 /*
797 * grab both a lock on the path and a lock on the delayed ref head.
798 * We need both to get a consistent picture of how the refs look
799 * at a specified point in time
800 */
801 again:
802 head = NULL;
803
804 ret = btrfs_search_slot(trans, fs_info->extent_root, &key, path, 0, 0);
805 if (ret < 0)
806 goto out;
807 BUG_ON(ret == 0);
808
809 if (trans != BTRFS_BACKREF_SEARCH_COMMIT_ROOT) {
810 /*
811 * look if there are updates for this ref queued and lock the
812 * head
813 */
814 delayed_refs = &trans->transaction->delayed_refs;
815 spin_lock(&delayed_refs->lock);
816 head = btrfs_find_delayed_ref_head(trans, bytenr);
817 if (head) {
818 if (!mutex_trylock(&head->mutex)) {
819 atomic_inc(&head->node.refs);
820 spin_unlock(&delayed_refs->lock);
821
822 btrfs_release_path(path);
823
824 /*
825 * Mutex was contended, block until it's
826 * released and try again
827 */
828 mutex_lock(&head->mutex);
829 mutex_unlock(&head->mutex);
830 btrfs_put_delayed_ref(&head->node);
831 goto again;
832 }
833 ret = __add_delayed_refs(head, delayed_ref_seq,
834 &prefs_delayed);
835 if (ret) {
836 spin_unlock(&delayed_refs->lock);
837 goto out;
838 }
839 }
840 spin_unlock(&delayed_refs->lock);
841 }
842
843 if (path->slots[0]) {
844 struct extent_buffer *leaf;
845 int slot;
846
847 path->slots[0]--;
848 leaf = path->nodes[0];
849 slot = path->slots[0];
850 btrfs_item_key_to_cpu(leaf, &key, slot);
851 if (key.objectid == bytenr &&
852 key.type == BTRFS_EXTENT_ITEM_KEY) {
853 ret = __add_inline_refs(fs_info, path, bytenr,
854 &info_level, &prefs);
855 if (ret)
856 goto out;
857 ret = __add_keyed_refs(fs_info, path, bytenr,
858 info_level, &prefs);
859 if (ret)
860 goto out;
861 }
862 }
863 btrfs_release_path(path);
864
865 list_splice_init(&prefs_delayed, &prefs);
866
867 ret = __add_missing_keys(fs_info, &prefs);
868 if (ret)
869 goto out;
870
871 ret = __merge_refs(&prefs, 1);
872 if (ret)
873 goto out;
874
875 ret = __resolve_indirect_refs(fs_info, search_commit_root, time_seq,
876 &prefs, extent_item_pos);
877 if (ret)
878 goto out;
879
880 ret = __merge_refs(&prefs, 2);
881 if (ret)
882 goto out;
883
884 while (!list_empty(&prefs)) {
885 ref = list_first_entry(&prefs, struct __prelim_ref, list);
886 list_del(&ref->list);
887 if (ref->count < 0)
888 WARN_ON(1);
889 if (ref->count && ref->root_id && ref->parent == 0) {
890 /* no parent == root of tree */
891 ret = ulist_add(roots, ref->root_id, 0, GFP_NOFS);
892 BUG_ON(ret < 0);
893 }
894 if (ref->count && ref->parent) {
895 struct extent_inode_elem *eie = NULL;
896 if (extent_item_pos && !ref->inode_list) {
897 u32 bsz;
898 struct extent_buffer *eb;
899 bsz = btrfs_level_size(fs_info->extent_root,
900 info_level);
901 eb = read_tree_block(fs_info->extent_root,
902 ref->parent, bsz, 0);
903 BUG_ON(!eb);
904 ret = find_extent_in_eb(eb, bytenr,
905 *extent_item_pos, &eie);
906 ref->inode_list = eie;
907 free_extent_buffer(eb);
908 }
909 ret = ulist_add_merge(refs, ref->parent,
910 (unsigned long)ref->inode_list,
911 (unsigned long *)&eie, GFP_NOFS);
912 if (!ret && extent_item_pos) {
913 /*
914 * we've recorded that parent, so we must extend
915 * its inode list here
916 */
917 BUG_ON(!eie);
918 while (eie->next)
919 eie = eie->next;
920 eie->next = ref->inode_list;
921 }
922 BUG_ON(ret < 0);
923 }
924 kfree(ref);
925 }
926
927 out:
928 if (head)
929 mutex_unlock(&head->mutex);
930 btrfs_free_path(path);
931 while (!list_empty(&prefs)) {
932 ref = list_first_entry(&prefs, struct __prelim_ref, list);
933 list_del(&ref->list);
934 kfree(ref);
935 }
936 while (!list_empty(&prefs_delayed)) {
937 ref = list_first_entry(&prefs_delayed, struct __prelim_ref,
938 list);
939 list_del(&ref->list);
940 kfree(ref);
941 }
942
943 return ret;
944 }
945
946 static void free_leaf_list(struct ulist *blocks)
947 {
948 struct ulist_node *node = NULL;
949 struct extent_inode_elem *eie;
950 struct extent_inode_elem *eie_next;
951 struct ulist_iterator uiter;
952
953 ULIST_ITER_INIT(&uiter);
954 while ((node = ulist_next(blocks, &uiter))) {
955 if (!node->aux)
956 continue;
957 eie = (struct extent_inode_elem *)node->aux;
958 for (; eie; eie = eie_next) {
959 eie_next = eie->next;
960 kfree(eie);
961 }
962 node->aux = 0;
963 }
964
965 ulist_free(blocks);
966 }
967
968 /*
969 * Finds all leafs with a reference to the specified combination of bytenr and
970 * offset. key_list_head will point to a list of corresponding keys (caller must
971 * free each list element). The leafs will be stored in the leafs ulist, which
972 * must be freed with ulist_free.
973 *
974 * returns 0 on success, <0 on error
975 */
976 static int btrfs_find_all_leafs(struct btrfs_trans_handle *trans,
977 struct btrfs_fs_info *fs_info, u64 bytenr,
978 u64 delayed_ref_seq, u64 time_seq,
979 struct ulist **leafs,
980 const u64 *extent_item_pos)
981 {
982 struct ulist *tmp;
983 int ret;
984
985 tmp = ulist_alloc(GFP_NOFS);
986 if (!tmp)
987 return -ENOMEM;
988 *leafs = ulist_alloc(GFP_NOFS);
989 if (!*leafs) {
990 ulist_free(tmp);
991 return -ENOMEM;
992 }
993
994 ret = find_parent_nodes(trans, fs_info, bytenr, delayed_ref_seq,
995 time_seq, *leafs, tmp, extent_item_pos);
996 ulist_free(tmp);
997
998 if (ret < 0 && ret != -ENOENT) {
999 free_leaf_list(*leafs);
1000 return ret;
1001 }
1002
1003 return 0;
1004 }
1005
1006 /*
1007 * walk all backrefs for a given extent to find all roots that reference this
1008 * extent. Walking a backref means finding all extents that reference this
1009 * extent and in turn walk the backrefs of those, too. Naturally this is a
1010 * recursive process, but here it is implemented in an iterative fashion: We
1011 * find all referencing extents for the extent in question and put them on a
1012 * list. In turn, we find all referencing extents for those, further appending
1013 * to the list. The way we iterate the list allows adding more elements after
1014 * the current while iterating. The process stops when we reach the end of the
1015 * list. Found roots are added to the roots list.
1016 *
1017 * returns 0 on success, < 0 on error.
1018 */
1019 int btrfs_find_all_roots(struct btrfs_trans_handle *trans,
1020 struct btrfs_fs_info *fs_info, u64 bytenr,
1021 u64 delayed_ref_seq, u64 time_seq,
1022 struct ulist **roots)
1023 {
1024 struct ulist *tmp;
1025 struct ulist_node *node = NULL;
1026 struct ulist_iterator uiter;
1027 int ret;
1028
1029 tmp = ulist_alloc(GFP_NOFS);
1030 if (!tmp)
1031 return -ENOMEM;
1032 *roots = ulist_alloc(GFP_NOFS);
1033 if (!*roots) {
1034 ulist_free(tmp);
1035 return -ENOMEM;
1036 }
1037
1038 ULIST_ITER_INIT(&uiter);
1039 while (1) {
1040 ret = find_parent_nodes(trans, fs_info, bytenr, delayed_ref_seq,
1041 time_seq, tmp, *roots, NULL);
1042 if (ret < 0 && ret != -ENOENT) {
1043 ulist_free(tmp);
1044 ulist_free(*roots);
1045 return ret;
1046 }
1047 node = ulist_next(tmp, &uiter);
1048 if (!node)
1049 break;
1050 bytenr = node->val;
1051 }
1052
1053 ulist_free(tmp);
1054 return 0;
1055 }
1056
1057
1058 static int __inode_info(u64 inum, u64 ioff, u8 key_type,
1059 struct btrfs_root *fs_root, struct btrfs_path *path,
1060 struct btrfs_key *found_key)
1061 {
1062 int ret;
1063 struct btrfs_key key;
1064 struct extent_buffer *eb;
1065
1066 key.type = key_type;
1067 key.objectid = inum;
1068 key.offset = ioff;
1069
1070 ret = btrfs_search_slot(NULL, fs_root, &key, path, 0, 0);
1071 if (ret < 0)
1072 return ret;
1073
1074 eb = path->nodes[0];
1075 if (ret && path->slots[0] >= btrfs_header_nritems(eb)) {
1076 ret = btrfs_next_leaf(fs_root, path);
1077 if (ret)
1078 return ret;
1079 eb = path->nodes[0];
1080 }
1081
1082 btrfs_item_key_to_cpu(eb, found_key, path->slots[0]);
1083 if (found_key->type != key.type || found_key->objectid != key.objectid)
1084 return 1;
1085
1086 return 0;
1087 }
1088
1089 /*
1090 * this makes the path point to (inum INODE_ITEM ioff)
1091 */
1092 int inode_item_info(u64 inum, u64 ioff, struct btrfs_root *fs_root,
1093 struct btrfs_path *path)
1094 {
1095 struct btrfs_key key;
1096 return __inode_info(inum, ioff, BTRFS_INODE_ITEM_KEY, fs_root, path,
1097 &key);
1098 }
1099
1100 static int inode_ref_info(u64 inum, u64 ioff, struct btrfs_root *fs_root,
1101 struct btrfs_path *path,
1102 struct btrfs_key *found_key)
1103 {
1104 return __inode_info(inum, ioff, BTRFS_INODE_REF_KEY, fs_root, path,
1105 found_key);
1106 }
1107
1108 /*
1109 * this iterates to turn a btrfs_inode_ref into a full filesystem path. elements
1110 * of the path are separated by '/' and the path is guaranteed to be
1111 * 0-terminated. the path is only given within the current file system.
1112 * Therefore, it never starts with a '/'. the caller is responsible to provide
1113 * "size" bytes in "dest". the dest buffer will be filled backwards. finally,
1114 * the start point of the resulting string is returned. this pointer is within
1115 * dest, normally.
1116 * in case the path buffer would overflow, the pointer is decremented further
1117 * as if output was written to the buffer, though no more output is actually
1118 * generated. that way, the caller can determine how much space would be
1119 * required for the path to fit into the buffer. in that case, the returned
1120 * value will be smaller than dest. callers must check this!
1121 */
1122 static char *iref_to_path(struct btrfs_root *fs_root, struct btrfs_path *path,
1123 struct btrfs_inode_ref *iref,
1124 struct extent_buffer *eb_in, u64 parent,
1125 char *dest, u32 size)
1126 {
1127 u32 len;
1128 int slot;
1129 u64 next_inum;
1130 int ret;
1131 s64 bytes_left = size - 1;
1132 struct extent_buffer *eb = eb_in;
1133 struct btrfs_key found_key;
1134 int leave_spinning = path->leave_spinning;
1135
1136 if (bytes_left >= 0)
1137 dest[bytes_left] = '\0';
1138
1139 path->leave_spinning = 1;
1140 while (1) {
1141 len = btrfs_inode_ref_name_len(eb, iref);
1142 bytes_left -= len;
1143 if (bytes_left >= 0)
1144 read_extent_buffer(eb, dest + bytes_left,
1145 (unsigned long)(iref + 1), len);
1146 if (eb != eb_in) {
1147 btrfs_tree_read_unlock_blocking(eb);
1148 free_extent_buffer(eb);
1149 }
1150 ret = inode_ref_info(parent, 0, fs_root, path, &found_key);
1151 if (ret > 0)
1152 ret = -ENOENT;
1153 if (ret)
1154 break;
1155 next_inum = found_key.offset;
1156
1157 /* regular exit ahead */
1158 if (parent == next_inum)
1159 break;
1160
1161 slot = path->slots[0];
1162 eb = path->nodes[0];
1163 /* make sure we can use eb after releasing the path */
1164 if (eb != eb_in) {
1165 atomic_inc(&eb->refs);
1166 btrfs_tree_read_lock(eb);
1167 btrfs_set_lock_blocking_rw(eb, BTRFS_READ_LOCK);
1168 }
1169 btrfs_release_path(path);
1170
1171 iref = btrfs_item_ptr(eb, slot, struct btrfs_inode_ref);
1172 parent = next_inum;
1173 --bytes_left;
1174 if (bytes_left >= 0)
1175 dest[bytes_left] = '/';
1176 }
1177
1178 btrfs_release_path(path);
1179 path->leave_spinning = leave_spinning;
1180
1181 if (ret)
1182 return ERR_PTR(ret);
1183
1184 return dest + bytes_left;
1185 }
1186
1187 /*
1188 * this makes the path point to (logical EXTENT_ITEM *)
1189 * returns BTRFS_EXTENT_FLAG_DATA for data, BTRFS_EXTENT_FLAG_TREE_BLOCK for
1190 * tree blocks and <0 on error.
1191 */
1192 int extent_from_logical(struct btrfs_fs_info *fs_info, u64 logical,
1193 struct btrfs_path *path, struct btrfs_key *found_key)
1194 {
1195 int ret;
1196 u64 flags;
1197 u32 item_size;
1198 struct extent_buffer *eb;
1199 struct btrfs_extent_item *ei;
1200 struct btrfs_key key;
1201
1202 key.type = BTRFS_EXTENT_ITEM_KEY;
1203 key.objectid = logical;
1204 key.offset = (u64)-1;
1205
1206 ret = btrfs_search_slot(NULL, fs_info->extent_root, &key, path, 0, 0);
1207 if (ret < 0)
1208 return ret;
1209 ret = btrfs_previous_item(fs_info->extent_root, path,
1210 0, BTRFS_EXTENT_ITEM_KEY);
1211 if (ret < 0)
1212 return ret;
1213
1214 btrfs_item_key_to_cpu(path->nodes[0], found_key, path->slots[0]);
1215 if (found_key->type != BTRFS_EXTENT_ITEM_KEY ||
1216 found_key->objectid > logical ||
1217 found_key->objectid + found_key->offset <= logical) {
1218 pr_debug("logical %llu is not within any extent\n",
1219 (unsigned long long)logical);
1220 return -ENOENT;
1221 }
1222
1223 eb = path->nodes[0];
1224 item_size = btrfs_item_size_nr(eb, path->slots[0]);
1225 BUG_ON(item_size < sizeof(*ei));
1226
1227 ei = btrfs_item_ptr(eb, path->slots[0], struct btrfs_extent_item);
1228 flags = btrfs_extent_flags(eb, ei);
1229
1230 pr_debug("logical %llu is at position %llu within the extent (%llu "
1231 "EXTENT_ITEM %llu) flags %#llx size %u\n",
1232 (unsigned long long)logical,
1233 (unsigned long long)(logical - found_key->objectid),
1234 (unsigned long long)found_key->objectid,
1235 (unsigned long long)found_key->offset,
1236 (unsigned long long)flags, item_size);
1237 if (flags & BTRFS_EXTENT_FLAG_TREE_BLOCK)
1238 return BTRFS_EXTENT_FLAG_TREE_BLOCK;
1239 if (flags & BTRFS_EXTENT_FLAG_DATA)
1240 return BTRFS_EXTENT_FLAG_DATA;
1241
1242 return -EIO;
1243 }
1244
1245 /*
1246 * helper function to iterate extent inline refs. ptr must point to a 0 value
1247 * for the first call and may be modified. it is used to track state.
1248 * if more refs exist, 0 is returned and the next call to
1249 * __get_extent_inline_ref must pass the modified ptr parameter to get the
1250 * next ref. after the last ref was processed, 1 is returned.
1251 * returns <0 on error
1252 */
1253 static int __get_extent_inline_ref(unsigned long *ptr, struct extent_buffer *eb,
1254 struct btrfs_extent_item *ei, u32 item_size,
1255 struct btrfs_extent_inline_ref **out_eiref,
1256 int *out_type)
1257 {
1258 unsigned long end;
1259 u64 flags;
1260 struct btrfs_tree_block_info *info;
1261
1262 if (!*ptr) {
1263 /* first call */
1264 flags = btrfs_extent_flags(eb, ei);
1265 if (flags & BTRFS_EXTENT_FLAG_TREE_BLOCK) {
1266 info = (struct btrfs_tree_block_info *)(ei + 1);
1267 *out_eiref =
1268 (struct btrfs_extent_inline_ref *)(info + 1);
1269 } else {
1270 *out_eiref = (struct btrfs_extent_inline_ref *)(ei + 1);
1271 }
1272 *ptr = (unsigned long)*out_eiref;
1273 if ((void *)*ptr >= (void *)ei + item_size)
1274 return -ENOENT;
1275 }
1276
1277 end = (unsigned long)ei + item_size;
1278 *out_eiref = (struct btrfs_extent_inline_ref *)*ptr;
1279 *out_type = btrfs_extent_inline_ref_type(eb, *out_eiref);
1280
1281 *ptr += btrfs_extent_inline_ref_size(*out_type);
1282 WARN_ON(*ptr > end);
1283 if (*ptr == end)
1284 return 1; /* last */
1285
1286 return 0;
1287 }
1288
1289 /*
1290 * reads the tree block backref for an extent. tree level and root are returned
1291 * through out_level and out_root. ptr must point to a 0 value for the first
1292 * call and may be modified (see __get_extent_inline_ref comment).
1293 * returns 0 if data was provided, 1 if there was no more data to provide or
1294 * <0 on error.
1295 */
1296 int tree_backref_for_extent(unsigned long *ptr, struct extent_buffer *eb,
1297 struct btrfs_extent_item *ei, u32 item_size,
1298 u64 *out_root, u8 *out_level)
1299 {
1300 int ret;
1301 int type;
1302 struct btrfs_tree_block_info *info;
1303 struct btrfs_extent_inline_ref *eiref;
1304
1305 if (*ptr == (unsigned long)-1)
1306 return 1;
1307
1308 while (1) {
1309 ret = __get_extent_inline_ref(ptr, eb, ei, item_size,
1310 &eiref, &type);
1311 if (ret < 0)
1312 return ret;
1313
1314 if (type == BTRFS_TREE_BLOCK_REF_KEY ||
1315 type == BTRFS_SHARED_BLOCK_REF_KEY)
1316 break;
1317
1318 if (ret == 1)
1319 return 1;
1320 }
1321
1322 /* we can treat both ref types equally here */
1323 info = (struct btrfs_tree_block_info *)(ei + 1);
1324 *out_root = btrfs_extent_inline_ref_offset(eb, eiref);
1325 *out_level = btrfs_tree_block_level(eb, info);
1326
1327 if (ret == 1)
1328 *ptr = (unsigned long)-1;
1329
1330 return 0;
1331 }
1332
1333 static int iterate_leaf_refs(struct extent_inode_elem *inode_list,
1334 u64 root, u64 extent_item_objectid,
1335 iterate_extent_inodes_t *iterate, void *ctx)
1336 {
1337 struct extent_inode_elem *eie;
1338 int ret = 0;
1339
1340 for (eie = inode_list; eie; eie = eie->next) {
1341 pr_debug("ref for %llu resolved, key (%llu EXTEND_DATA %llu), "
1342 "root %llu\n", extent_item_objectid,
1343 eie->inum, eie->offset, root);
1344 ret = iterate(eie->inum, eie->offset, root, ctx);
1345 if (ret) {
1346 pr_debug("stopping iteration for %llu due to ret=%d\n",
1347 extent_item_objectid, ret);
1348 break;
1349 }
1350 }
1351
1352 return ret;
1353 }
1354
1355 /*
1356 * calls iterate() for every inode that references the extent identified by
1357 * the given parameters.
1358 * when the iterator function returns a non-zero value, iteration stops.
1359 */
1360 int iterate_extent_inodes(struct btrfs_fs_info *fs_info,
1361 u64 extent_item_objectid, u64 extent_item_pos,
1362 int search_commit_root,
1363 iterate_extent_inodes_t *iterate, void *ctx)
1364 {
1365 int ret;
1366 struct list_head data_refs = LIST_HEAD_INIT(data_refs);
1367 struct list_head shared_refs = LIST_HEAD_INIT(shared_refs);
1368 struct btrfs_trans_handle *trans;
1369 struct ulist *refs = NULL;
1370 struct ulist *roots = NULL;
1371 struct ulist_node *ref_node = NULL;
1372 struct ulist_node *root_node = NULL;
1373 struct seq_list seq_elem = {};
1374 struct seq_list tree_mod_seq_elem = {};
1375 struct ulist_iterator ref_uiter;
1376 struct ulist_iterator root_uiter;
1377 struct btrfs_delayed_ref_root *delayed_refs = NULL;
1378
1379 pr_debug("resolving all inodes for extent %llu\n",
1380 extent_item_objectid);
1381
1382 if (search_commit_root) {
1383 trans = BTRFS_BACKREF_SEARCH_COMMIT_ROOT;
1384 } else {
1385 trans = btrfs_join_transaction(fs_info->extent_root);
1386 if (IS_ERR(trans))
1387 return PTR_ERR(trans);
1388
1389 delayed_refs = &trans->transaction->delayed_refs;
1390 spin_lock(&delayed_refs->lock);
1391 btrfs_get_delayed_seq(delayed_refs, &seq_elem);
1392 spin_unlock(&delayed_refs->lock);
1393 btrfs_get_tree_mod_seq(fs_info, &tree_mod_seq_elem);
1394 }
1395
1396 ret = btrfs_find_all_leafs(trans, fs_info, extent_item_objectid,
1397 seq_elem.seq, tree_mod_seq_elem.seq, &refs,
1398 &extent_item_pos);
1399 if (ret)
1400 goto out;
1401
1402 ULIST_ITER_INIT(&ref_uiter);
1403 while (!ret && (ref_node = ulist_next(refs, &ref_uiter))) {
1404 ret = btrfs_find_all_roots(trans, fs_info, ref_node->val,
1405 seq_elem.seq,
1406 tree_mod_seq_elem.seq, &roots);
1407 if (ret)
1408 break;
1409 ULIST_ITER_INIT(&root_uiter);
1410 while (!ret && (root_node = ulist_next(roots, &root_uiter))) {
1411 pr_debug("root %llu references leaf %llu, data list "
1412 "%#lx\n", root_node->val, ref_node->val,
1413 ref_node->aux);
1414 ret = iterate_leaf_refs(
1415 (struct extent_inode_elem *)ref_node->aux,
1416 root_node->val, extent_item_objectid,
1417 iterate, ctx);
1418 }
1419 ulist_free(roots);
1420 roots = NULL;
1421 }
1422
1423 free_leaf_list(refs);
1424 ulist_free(roots);
1425 out:
1426 if (!search_commit_root) {
1427 btrfs_put_tree_mod_seq(fs_info, &tree_mod_seq_elem);
1428 btrfs_put_delayed_seq(delayed_refs, &seq_elem);
1429 btrfs_end_transaction(trans, fs_info->extent_root);
1430 }
1431
1432 return ret;
1433 }
1434
1435 int iterate_inodes_from_logical(u64 logical, struct btrfs_fs_info *fs_info,
1436 struct btrfs_path *path,
1437 iterate_extent_inodes_t *iterate, void *ctx)
1438 {
1439 int ret;
1440 u64 extent_item_pos;
1441 struct btrfs_key found_key;
1442 int search_commit_root = path->search_commit_root;
1443
1444 ret = extent_from_logical(fs_info, logical, path,
1445 &found_key);
1446 btrfs_release_path(path);
1447 if (ret & BTRFS_EXTENT_FLAG_TREE_BLOCK)
1448 ret = -EINVAL;
1449 if (ret < 0)
1450 return ret;
1451
1452 extent_item_pos = logical - found_key.objectid;
1453 ret = iterate_extent_inodes(fs_info, found_key.objectid,
1454 extent_item_pos, search_commit_root,
1455 iterate, ctx);
1456
1457 return ret;
1458 }
1459
1460 static int iterate_irefs(u64 inum, struct btrfs_root *fs_root,
1461 struct btrfs_path *path,
1462 iterate_irefs_t *iterate, void *ctx)
1463 {
1464 int ret = 0;
1465 int slot;
1466 u32 cur;
1467 u32 len;
1468 u32 name_len;
1469 u64 parent = 0;
1470 int found = 0;
1471 struct extent_buffer *eb;
1472 struct btrfs_item *item;
1473 struct btrfs_inode_ref *iref;
1474 struct btrfs_key found_key;
1475
1476 while (!ret) {
1477 path->leave_spinning = 1;
1478 ret = inode_ref_info(inum, parent ? parent+1 : 0, fs_root, path,
1479 &found_key);
1480 if (ret < 0)
1481 break;
1482 if (ret) {
1483 ret = found ? 0 : -ENOENT;
1484 break;
1485 }
1486 ++found;
1487
1488 parent = found_key.offset;
1489 slot = path->slots[0];
1490 eb = path->nodes[0];
1491 /* make sure we can use eb after releasing the path */
1492 atomic_inc(&eb->refs);
1493 btrfs_tree_read_lock(eb);
1494 btrfs_set_lock_blocking_rw(eb, BTRFS_READ_LOCK);
1495 btrfs_release_path(path);
1496
1497 item = btrfs_item_nr(eb, slot);
1498 iref = btrfs_item_ptr(eb, slot, struct btrfs_inode_ref);
1499
1500 for (cur = 0; cur < btrfs_item_size(eb, item); cur += len) {
1501 name_len = btrfs_inode_ref_name_len(eb, iref);
1502 /* path must be released before calling iterate()! */
1503 pr_debug("following ref at offset %u for inode %llu in "
1504 "tree %llu\n", cur,
1505 (unsigned long long)found_key.objectid,
1506 (unsigned long long)fs_root->objectid);
1507 ret = iterate(parent, iref, eb, ctx);
1508 if (ret)
1509 break;
1510 len = sizeof(*iref) + name_len;
1511 iref = (struct btrfs_inode_ref *)((char *)iref + len);
1512 }
1513 btrfs_tree_read_unlock_blocking(eb);
1514 free_extent_buffer(eb);
1515 }
1516
1517 btrfs_release_path(path);
1518
1519 return ret;
1520 }
1521
1522 /*
1523 * returns 0 if the path could be dumped (probably truncated)
1524 * returns <0 in case of an error
1525 */
1526 static int inode_to_path(u64 inum, struct btrfs_inode_ref *iref,
1527 struct extent_buffer *eb, void *ctx)
1528 {
1529 struct inode_fs_paths *ipath = ctx;
1530 char *fspath;
1531 char *fspath_min;
1532 int i = ipath->fspath->elem_cnt;
1533 const int s_ptr = sizeof(char *);
1534 u32 bytes_left;
1535
1536 bytes_left = ipath->fspath->bytes_left > s_ptr ?
1537 ipath->fspath->bytes_left - s_ptr : 0;
1538
1539 fspath_min = (char *)ipath->fspath->val + (i + 1) * s_ptr;
1540 fspath = iref_to_path(ipath->fs_root, ipath->btrfs_path, iref, eb,
1541 inum, fspath_min, bytes_left);
1542 if (IS_ERR(fspath))
1543 return PTR_ERR(fspath);
1544
1545 if (fspath > fspath_min) {
1546 pr_debug("path resolved: %s\n", fspath);
1547 ipath->fspath->val[i] = (u64)(unsigned long)fspath;
1548 ++ipath->fspath->elem_cnt;
1549 ipath->fspath->bytes_left = fspath - fspath_min;
1550 } else {
1551 pr_debug("missed path, not enough space. missing bytes: %lu, "
1552 "constructed so far: %s\n",
1553 (unsigned long)(fspath_min - fspath), fspath_min);
1554 ++ipath->fspath->elem_missed;
1555 ipath->fspath->bytes_missing += fspath_min - fspath;
1556 ipath->fspath->bytes_left = 0;
1557 }
1558
1559 return 0;
1560 }
1561
1562 /*
1563 * this dumps all file system paths to the inode into the ipath struct, provided
1564 * is has been created large enough. each path is zero-terminated and accessed
1565 * from ipath->fspath->val[i].
1566 * when it returns, there are ipath->fspath->elem_cnt number of paths available
1567 * in ipath->fspath->val[]. when the allocated space wasn't sufficient, the
1568 * number of missed paths in recored in ipath->fspath->elem_missed, otherwise,
1569 * it's zero. ipath->fspath->bytes_missing holds the number of bytes that would
1570 * have been needed to return all paths.
1571 */
1572 int paths_from_inode(u64 inum, struct inode_fs_paths *ipath)
1573 {
1574 return iterate_irefs(inum, ipath->fs_root, ipath->btrfs_path,
1575 inode_to_path, ipath);
1576 }
1577
1578 struct btrfs_data_container *init_data_container(u32 total_bytes)
1579 {
1580 struct btrfs_data_container *data;
1581 size_t alloc_bytes;
1582
1583 alloc_bytes = max_t(size_t, total_bytes, sizeof(*data));
1584 data = kmalloc(alloc_bytes, GFP_NOFS);
1585 if (!data)
1586 return ERR_PTR(-ENOMEM);
1587
1588 if (total_bytes >= sizeof(*data)) {
1589 data->bytes_left = total_bytes - sizeof(*data);
1590 data->bytes_missing = 0;
1591 } else {
1592 data->bytes_missing = sizeof(*data) - total_bytes;
1593 data->bytes_left = 0;
1594 }
1595
1596 data->elem_cnt = 0;
1597 data->elem_missed = 0;
1598
1599 return data;
1600 }
1601
1602 /*
1603 * allocates space to return multiple file system paths for an inode.
1604 * total_bytes to allocate are passed, note that space usable for actual path
1605 * information will be total_bytes - sizeof(struct inode_fs_paths).
1606 * the returned pointer must be freed with free_ipath() in the end.
1607 */
1608 struct inode_fs_paths *init_ipath(s32 total_bytes, struct btrfs_root *fs_root,
1609 struct btrfs_path *path)
1610 {
1611 struct inode_fs_paths *ifp;
1612 struct btrfs_data_container *fspath;
1613
1614 fspath = init_data_container(total_bytes);
1615 if (IS_ERR(fspath))
1616 return (void *)fspath;
1617
1618 ifp = kmalloc(sizeof(*ifp), GFP_NOFS);
1619 if (!ifp) {
1620 kfree(fspath);
1621 return ERR_PTR(-ENOMEM);
1622 }
1623
1624 ifp->btrfs_path = path;
1625 ifp->fspath = fspath;
1626 ifp->fs_root = fs_root;
1627
1628 return ifp;
1629 }
1630
1631 void free_ipath(struct inode_fs_paths *ipath)
1632 {
1633 if (!ipath)
1634 return;
1635 kfree(ipath->fspath);
1636 kfree(ipath);
1637 }
(deprecated) Btrfs devel tree