?  search:
summary | log | graphiclog1 | graphiclog2 | commit | commitdiff | tree | refs | edit | fork
blame | history | raw | HEAD
ARM: rockchip: fix broken build
[linux/fpc-iii.git] / fs / btrfs / backref.c
blob802fabb30e1575338eec8ae53f157e9f2d7a6b56
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 <linux/vmalloc.h>
20 #include "ctree.h"
21 #include "disk-io.h"
22 #include "backref.h"
23 #include "ulist.h"
24 #include "transaction.h"
25 #include "delayed-ref.h"
26 #include "locking.h"
27
28 /* Just an arbitrary number so we can be sure this happened */
29 #define BACKREF_FOUND_SHARED 6
30
31 struct extent_inode_elem {
32 u64 inum;
33 u64 offset;
34 struct extent_inode_elem *next;
35 };
36
37 static int check_extent_in_eb(struct btrfs_key *key, struct extent_buffer *eb,
38 struct btrfs_file_extent_item *fi,
39 u64 extent_item_pos,
40 struct extent_inode_elem **eie)
41 {
42 u64 offset = 0;
43 struct extent_inode_elem *e;
44
45 if (!btrfs_file_extent_compression(eb, fi) &&
46 !btrfs_file_extent_encryption(eb, fi) &&
47 !btrfs_file_extent_other_encoding(eb, fi)) {
48 u64 data_offset;
49 u64 data_len;
50
51 data_offset = btrfs_file_extent_offset(eb, fi);
52 data_len = btrfs_file_extent_num_bytes(eb, fi);
53
54 if (extent_item_pos < data_offset ||
55 extent_item_pos >= data_offset + data_len)
56 return 1;
57 offset = extent_item_pos - data_offset;
58 }
59
60 e = kmalloc(sizeof(*e), GFP_NOFS);
61 if (!e)
62 return -ENOMEM;
63
64 e->next = *eie;
65 e->inum = key->objectid;
66 e->offset = key->offset + offset;
67 *eie = e;
68
69 return 0;
70 }
71
72 static void free_inode_elem_list(struct extent_inode_elem *eie)
73 {
74 struct extent_inode_elem *eie_next;
75
76 for (; eie; eie = eie_next) {
77 eie_next = eie->next;
78 kfree(eie);
79 }
80 }
81
82 static int find_extent_in_eb(struct extent_buffer *eb, u64 wanted_disk_byte,
83 u64 extent_item_pos,
84 struct extent_inode_elem **eie)
85 {
86 u64 disk_byte;
87 struct btrfs_key key;
88 struct btrfs_file_extent_item *fi;
89 int slot;
90 int nritems;
91 int extent_type;
92 int ret;
93
94 /*
95 * from the shared data ref, we only have the leaf but we need
96 * the key. thus, we must look into all items and see that we
97 * find one (some) with a reference to our extent item.
98 */
99 nritems = btrfs_header_nritems(eb);
100 for (slot = 0; slot < nritems; ++slot) {
101 btrfs_item_key_to_cpu(eb, &key, slot);
102 if (key.type != BTRFS_EXTENT_DATA_KEY)
103 continue;
104 fi = btrfs_item_ptr(eb, slot, struct btrfs_file_extent_item);
105 extent_type = btrfs_file_extent_type(eb, fi);
106 if (extent_type == BTRFS_FILE_EXTENT_INLINE)
107 continue;
108 /* don't skip BTRFS_FILE_EXTENT_PREALLOC, we can handle that */
109 disk_byte = btrfs_file_extent_disk_bytenr(eb, fi);
110 if (disk_byte != wanted_disk_byte)
111 continue;
112
113 ret = check_extent_in_eb(&key, eb, fi, extent_item_pos, eie);
114 if (ret < 0)
115 return ret;
116 }
117
118 return 0;
119 }
120
121 /*
122 * this structure records all encountered refs on the way up to the root
123 */
124 struct __prelim_ref {
125 struct list_head list;
126 u64 root_id;
127 struct btrfs_key key_for_search;
128 int level;
129 int count;
130 struct extent_inode_elem *inode_list;
131 u64 parent;
132 u64 wanted_disk_byte;
133 };
134
135 static struct kmem_cache *btrfs_prelim_ref_cache;
136
137 int __init btrfs_prelim_ref_init(void)
138 {
139 btrfs_prelim_ref_cache = kmem_cache_create("btrfs_prelim_ref",
140 sizeof(struct __prelim_ref),
141 0,
142 SLAB_RECLAIM_ACCOUNT | SLAB_MEM_SPREAD,
143 NULL);
144 if (!btrfs_prelim_ref_cache)
145 return -ENOMEM;
146 return 0;
147 }
148
149 void btrfs_prelim_ref_exit(void)
150 {
151 if (btrfs_prelim_ref_cache)
152 kmem_cache_destroy(btrfs_prelim_ref_cache);
153 }
154
155 /*
156 * the rules for all callers of this function are:
157 * - obtaining the parent is the goal
158 * - if you add a key, you must know that it is a correct key
159 * - if you cannot add the parent or a correct key, then we will look into the
160 * block later to set a correct key
161 *
162 * delayed refs
163 * ============
164 * backref type | shared | indirect | shared | indirect
165 * information | tree | tree | data | data
166 * --------------------+--------+----------+--------+----------
167 * parent logical | y | - | - | -
168 * key to resolve | - | y | y | y
169 * tree block logical | - | - | - | -
170 * root for resolving | y | y | y | y
171 *
172 * - column 1: we've the parent -> done
173 * - column 2, 3, 4: we use the key to find the parent
174 *
175 * on disk refs (inline or keyed)
176 * ==============================
177 * backref type | shared | indirect | shared | indirect
178 * information | tree | tree | data | data
179 * --------------------+--------+----------+--------+----------
180 * parent logical | y | - | y | -
181 * key to resolve | - | - | - | y
182 * tree block logical | y | y | y | y
183 * root for resolving | - | y | y | y
184 *
185 * - column 1, 3: we've the parent -> done
186 * - column 2: we take the first key from the block to find the parent
187 * (see __add_missing_keys)
188 * - column 4: we use the key to find the parent
189 *
190 * additional information that's available but not required to find the parent
191 * block might help in merging entries to gain some speed.
192 */
193
194 static int __add_prelim_ref(struct list_head *head, u64 root_id,
195 struct btrfs_key *key, int level,
196 u64 parent, u64 wanted_disk_byte, int count,
197 gfp_t gfp_mask)
198 {
199 struct __prelim_ref *ref;
200
201 if (root_id == BTRFS_DATA_RELOC_TREE_OBJECTID)
202 return 0;
203
204 ref = kmem_cache_alloc(btrfs_prelim_ref_cache, gfp_mask);
205 if (!ref)
206 return -ENOMEM;
207
208 ref->root_id = root_id;
209 if (key)
210 ref->key_for_search = *key;
211 else
212 memset(&ref->key_for_search, 0, sizeof(ref->key_for_search));
213
214 ref->inode_list = NULL;
215 ref->level = level;
216 ref->count = count;
217 ref->parent = parent;
218 ref->wanted_disk_byte = wanted_disk_byte;
219 list_add_tail(&ref->list, head);
220
221 return 0;
222 }
223
224 static int add_all_parents(struct btrfs_root *root, struct btrfs_path *path,
225 struct ulist *parents, struct __prelim_ref *ref,
226 int level, u64 time_seq, const u64 *extent_item_pos,
227 u64 total_refs)
228 {
229 int ret = 0;
230 int slot;
231 struct extent_buffer *eb;
232 struct btrfs_key key;
233 struct btrfs_key *key_for_search = &ref->key_for_search;
234 struct btrfs_file_extent_item *fi;
235 struct extent_inode_elem *eie = NULL, *old = NULL;
236 u64 disk_byte;
237 u64 wanted_disk_byte = ref->wanted_disk_byte;
238 u64 count = 0;
239
240 if (level != 0) {
241 eb = path->nodes[level];
242 ret = ulist_add(parents, eb->start, 0, GFP_NOFS);
243 if (ret < 0)
244 return ret;
245 return 0;
246 }
247
248 /*
249 * We normally enter this function with the path already pointing to
250 * the first item to check. But sometimes, we may enter it with
251 * slot==nritems. In that case, go to the next leaf before we continue.
252 */
253 if (path->slots[0] >= btrfs_header_nritems(path->nodes[0])) {
254 if (time_seq == (u64)-1)
255 ret = btrfs_next_leaf(root, path);
256 else
257 ret = btrfs_next_old_leaf(root, path, time_seq);
258 }
259
260 while (!ret && count < total_refs) {
261 eb = path->nodes[0];
262 slot = path->slots[0];
263
264 btrfs_item_key_to_cpu(eb, &key, slot);
265
266 if (key.objectid != key_for_search->objectid ||
267 key.type != BTRFS_EXTENT_DATA_KEY)
268 break;
269
270 fi = btrfs_item_ptr(eb, slot, struct btrfs_file_extent_item);
271 disk_byte = btrfs_file_extent_disk_bytenr(eb, fi);
272
273 if (disk_byte == wanted_disk_byte) {
274 eie = NULL;
275 old = NULL;
276 count++;
277 if (extent_item_pos) {
278 ret = check_extent_in_eb(&key, eb, fi,
279 *extent_item_pos,
280 &eie);
281 if (ret < 0)
282 break;
283 }
284 if (ret > 0)
285 goto next;
286 ret = ulist_add_merge_ptr(parents, eb->start,
287 eie, (void **)&old, GFP_NOFS);
288 if (ret < 0)
289 break;
290 if (!ret && extent_item_pos) {
291 while (old->next)
292 old = old->next;
293 old->next = eie;
294 }
295 eie = NULL;
296 }
297 next:
298 if (time_seq == (u64)-1)
299 ret = btrfs_next_item(root, path);
300 else
301 ret = btrfs_next_old_item(root, path, time_seq);
302 }
303
304 if (ret > 0)
305 ret = 0;
306 else if (ret < 0)
307 free_inode_elem_list(eie);
308 return ret;
309 }
310
311 /*
312 * resolve an indirect backref in the form (root_id, key, level)
313 * to a logical address
314 */
315 static int __resolve_indirect_ref(struct btrfs_fs_info *fs_info,
316 struct btrfs_path *path, u64 time_seq,
317 struct __prelim_ref *ref,
318 struct ulist *parents,
319 const u64 *extent_item_pos, u64 total_refs)
320 {
321 struct btrfs_root *root;
322 struct btrfs_key root_key;
323 struct extent_buffer *eb;
324 int ret = 0;
325 int root_level;
326 int level = ref->level;
327 int index;
328
329 root_key.objectid = ref->root_id;
330 root_key.type = BTRFS_ROOT_ITEM_KEY;
331 root_key.offset = (u64)-1;
332
333 index = srcu_read_lock(&fs_info->subvol_srcu);
334
335 root = btrfs_read_fs_root_no_name(fs_info, &root_key);
336 if (IS_ERR(root)) {
337 srcu_read_unlock(&fs_info->subvol_srcu, index);
338 ret = PTR_ERR(root);
339 goto out;
340 }
341
342 if (path->search_commit_root)
343 root_level = btrfs_header_level(root->commit_root);
344 else if (time_seq == (u64)-1)
345 root_level = btrfs_header_level(root->node);
346 else
347 root_level = btrfs_old_root_level(root, time_seq);
348
349 if (root_level + 1 == level) {
350 srcu_read_unlock(&fs_info->subvol_srcu, index);
351 goto out;
352 }
353
354 path->lowest_level = level;
355 if (time_seq == (u64)-1)
356 ret = btrfs_search_slot(NULL, root, &ref->key_for_search, path,
357 0, 0);
358 else
359 ret = btrfs_search_old_slot(root, &ref->key_for_search, path,
360 time_seq);
361
362 /* root node has been locked, we can release @subvol_srcu safely here */
363 srcu_read_unlock(&fs_info->subvol_srcu, index);
364
365 pr_debug("search slot in root %llu (level %d, ref count %d) returned "
366 "%d for key (%llu %u %llu)\n",
367 ref->root_id, level, ref->count, ret,
368 ref->key_for_search.objectid, ref->key_for_search.type,
369 ref->key_for_search.offset);
370 if (ret < 0)
371 goto out;
372
373 eb = path->nodes[level];
374 while (!eb) {
375 if (WARN_ON(!level)) {
376 ret = 1;
377 goto out;
378 }
379 level--;
380 eb = path->nodes[level];
381 }
382
383 ret = add_all_parents(root, path, parents, ref, level, time_seq,
384 extent_item_pos, total_refs);
385 out:
386 path->lowest_level = 0;
387 btrfs_release_path(path);
388 return ret;
389 }
390
391 /*
392 * resolve all indirect backrefs from the list
393 */
394 static int __resolve_indirect_refs(struct btrfs_fs_info *fs_info,
395 struct btrfs_path *path, u64 time_seq,
396 struct list_head *head,
397 const u64 *extent_item_pos, u64 total_refs,
398 u64 root_objectid)
399 {
400 int err;
401 int ret = 0;
402 struct __prelim_ref *ref;
403 struct __prelim_ref *ref_safe;
404 struct __prelim_ref *new_ref;
405 struct ulist *parents;
406 struct ulist_node *node;
407 struct ulist_iterator uiter;
408
409 parents = ulist_alloc(GFP_NOFS);
410 if (!parents)
411 return -ENOMEM;
412
413 /*
414 * _safe allows us to insert directly after the current item without
415 * iterating over the newly inserted items.
416 * we're also allowed to re-assign ref during iteration.
417 */
418 list_for_each_entry_safe(ref, ref_safe, head, list) {
419 if (ref->parent) /* already direct */
420 continue;
421 if (ref->count == 0)
422 continue;
423 if (root_objectid && ref->root_id != root_objectid) {
424 ret = BACKREF_FOUND_SHARED;
425 goto out;
426 }
427 err = __resolve_indirect_ref(fs_info, path, time_seq, ref,
428 parents, extent_item_pos,
429 total_refs);
430 /*
431 * we can only tolerate ENOENT,otherwise,we should catch error
432 * and return directly.
433 */
434 if (err == -ENOENT) {
435 continue;
436 } else if (err) {
437 ret = err;
438 goto out;
439 }
440
441 /* we put the first parent into the ref at hand */
442 ULIST_ITER_INIT(&uiter);
443 node = ulist_next(parents, &uiter);
444 ref->parent = node ? node->val : 0;
445 ref->inode_list = node ?
446 (struct extent_inode_elem *)(uintptr_t)node->aux : NULL;
447
448 /* additional parents require new refs being added here */
449 while ((node = ulist_next(parents, &uiter))) {
450 new_ref = kmem_cache_alloc(btrfs_prelim_ref_cache,
451 GFP_NOFS);
452 if (!new_ref) {
453 ret = -ENOMEM;
454 goto out;
455 }
456 memcpy(new_ref, ref, sizeof(*ref));
457 new_ref->parent = node->val;
458 new_ref->inode_list = (struct extent_inode_elem *)
459 (uintptr_t)node->aux;
460 list_add(&new_ref->list, &ref->list);
461 }
462 ulist_reinit(parents);
463 }
464 out:
465 ulist_free(parents);
466 return ret;
467 }
468
469 static inline int ref_for_same_block(struct __prelim_ref *ref1,
470 struct __prelim_ref *ref2)
471 {
472 if (ref1->level != ref2->level)
473 return 0;
474 if (ref1->root_id != ref2->root_id)
475 return 0;
476 if (ref1->key_for_search.type != ref2->key_for_search.type)
477 return 0;
478 if (ref1->key_for_search.objectid != ref2->key_for_search.objectid)
479 return 0;
480 if (ref1->key_for_search.offset != ref2->key_for_search.offset)
481 return 0;
482 if (ref1->parent != ref2->parent)
483 return 0;
484
485 return 1;
486 }
487
488 /*
489 * read tree blocks and add keys where required.
490 */
491 static int __add_missing_keys(struct btrfs_fs_info *fs_info,
492 struct list_head *head)
493 {
494 struct list_head *pos;
495 struct extent_buffer *eb;
496
497 list_for_each(pos, head) {
498 struct __prelim_ref *ref;
499 ref = list_entry(pos, struct __prelim_ref, list);
500
501 if (ref->parent)
502 continue;
503 if (ref->key_for_search.type)
504 continue;
505 BUG_ON(!ref->wanted_disk_byte);
506 eb = read_tree_block(fs_info->tree_root, ref->wanted_disk_byte,
507 0);
508 if (IS_ERR(eb)) {
509 return PTR_ERR(eb);
510 } else if (!extent_buffer_uptodate(eb)) {
511 free_extent_buffer(eb);
512 return -EIO;
513 }
514 btrfs_tree_read_lock(eb);
515 if (btrfs_header_level(eb) == 0)
516 btrfs_item_key_to_cpu(eb, &ref->key_for_search, 0);
517 else
518 btrfs_node_key_to_cpu(eb, &ref->key_for_search, 0);
519 btrfs_tree_read_unlock(eb);
520 free_extent_buffer(eb);
521 }
522 return 0;
523 }
524
525 /*
526 * merge backrefs and adjust counts accordingly
527 *
528 * mode = 1: merge identical keys, if key is set
529 * FIXME: if we add more keys in __add_prelim_ref, we can merge more here.
530 * additionally, we could even add a key range for the blocks we
531 * looked into to merge even more (-> replace unresolved refs by those
532 * having a parent).
533 * mode = 2: merge identical parents
534 */
535 static void __merge_refs(struct list_head *head, int mode)
536 {
537 struct list_head *pos1;
538
539 list_for_each(pos1, head) {
540 struct list_head *n2;
541 struct list_head *pos2;
542 struct __prelim_ref *ref1;
543
544 ref1 = list_entry(pos1, struct __prelim_ref, list);
545
546 for (pos2 = pos1->next, n2 = pos2->next; pos2 != head;
547 pos2 = n2, n2 = pos2->next) {
548 struct __prelim_ref *ref2;
549 struct __prelim_ref *xchg;
550 struct extent_inode_elem *eie;
551
552 ref2 = list_entry(pos2, struct __prelim_ref, list);
553
554 if (!ref_for_same_block(ref1, ref2))
555 continue;
556 if (mode == 1) {
557 if (!ref1->parent && ref2->parent) {
558 xchg = ref1;
559 ref1 = ref2;
560 ref2 = xchg;
561 }
562 } else {
563 if (ref1->parent != ref2->parent)
564 continue;
565 }
566
567 eie = ref1->inode_list;
568 while (eie && eie->next)
569 eie = eie->next;
570 if (eie)
571 eie->next = ref2->inode_list;
572 else
573 ref1->inode_list = ref2->inode_list;
574 ref1->count += ref2->count;
575
576 list_del(&ref2->list);
577 kmem_cache_free(btrfs_prelim_ref_cache, ref2);
578 }
579
580 }
581 }
582
583 /*
584 * add all currently queued delayed refs from this head whose seq nr is
585 * smaller or equal that seq to the list
586 */
587 static int __add_delayed_refs(struct btrfs_delayed_ref_head *head, u64 seq,
588 struct list_head *prefs, u64 *total_refs,
589 u64 inum)
590 {
591 struct btrfs_delayed_ref_node *node;
592 struct btrfs_delayed_extent_op *extent_op = head->extent_op;
593 struct btrfs_key key;
594 struct btrfs_key op_key = {0};
595 int sgn;
596 int ret = 0;
597
598 if (extent_op && extent_op->update_key)
599 btrfs_disk_key_to_cpu(&op_key, &extent_op->key);
600
601 spin_lock(&head->lock);
602 list_for_each_entry(node, &head->ref_list, list) {
603 if (node->seq > seq)
604 continue;
605
606 switch (node->action) {
607 case BTRFS_ADD_DELAYED_EXTENT:
608 case BTRFS_UPDATE_DELAYED_HEAD:
609 WARN_ON(1);
610 continue;
611 case BTRFS_ADD_DELAYED_REF:
612 sgn = 1;
613 break;
614 case BTRFS_DROP_DELAYED_REF:
615 sgn = -1;
616 break;
617 default:
618 BUG_ON(1);
619 }
620 *total_refs += (node->ref_mod * sgn);
621 switch (node->type) {
622 case BTRFS_TREE_BLOCK_REF_KEY: {
623 struct btrfs_delayed_tree_ref *ref;
624
625 ref = btrfs_delayed_node_to_tree_ref(node);
626 ret = __add_prelim_ref(prefs, ref->root, &op_key,
627 ref->level + 1, 0, node->bytenr,
628 node->ref_mod * sgn, GFP_ATOMIC);
629 break;
630 }
631 case BTRFS_SHARED_BLOCK_REF_KEY: {
632 struct btrfs_delayed_tree_ref *ref;
633
634 ref = btrfs_delayed_node_to_tree_ref(node);
635 ret = __add_prelim_ref(prefs, ref->root, NULL,
636 ref->level + 1, ref->parent,
637 node->bytenr,
638 node->ref_mod * sgn, GFP_ATOMIC);
639 break;
640 }
641 case BTRFS_EXTENT_DATA_REF_KEY: {
642 struct btrfs_delayed_data_ref *ref;
643 ref = btrfs_delayed_node_to_data_ref(node);
644
645 key.objectid = ref->objectid;
646 key.type = BTRFS_EXTENT_DATA_KEY;
647 key.offset = ref->offset;
648
649 /*
650 * Found a inum that doesn't match our known inum, we
651 * know it's shared.
652 */
653 if (inum && ref->objectid != inum) {
654 ret = BACKREF_FOUND_SHARED;
655 break;
656 }
657
658 ret = __add_prelim_ref(prefs, ref->root, &key, 0, 0,
659 node->bytenr,
660 node->ref_mod * sgn, GFP_ATOMIC);
661 break;
662 }
663 case BTRFS_SHARED_DATA_REF_KEY: {
664 struct btrfs_delayed_data_ref *ref;
665
666 ref = btrfs_delayed_node_to_data_ref(node);
667
668 key.objectid = ref->objectid;
669 key.type = BTRFS_EXTENT_DATA_KEY;
670 key.offset = ref->offset;
671 ret = __add_prelim_ref(prefs, ref->root, &key, 0,
672 ref->parent, node->bytenr,
673 node->ref_mod * sgn, GFP_ATOMIC);
674 break;
675 }
676 default:
677 WARN_ON(1);
678 }
679 if (ret)
680 break;
681 }
682 spin_unlock(&head->lock);
683 return ret;
684 }
685
686 /*
687 * add all inline backrefs for bytenr to the list
688 */
689 static int __add_inline_refs(struct btrfs_fs_info *fs_info,
690 struct btrfs_path *path, u64 bytenr,
691 int *info_level, struct list_head *prefs,
692 u64 *total_refs, u64 inum)
693 {
694 int ret = 0;
695 int slot;
696 struct extent_buffer *leaf;
697 struct btrfs_key key;
698 struct btrfs_key found_key;
699 unsigned long ptr;
700 unsigned long end;
701 struct btrfs_extent_item *ei;
702 u64 flags;
703 u64 item_size;
704
705 /*
706 * enumerate all inline refs
707 */
708 leaf = path->nodes[0];
709 slot = path->slots[0];
710
711 item_size = btrfs_item_size_nr(leaf, slot);
712 BUG_ON(item_size < sizeof(*ei));
713
714 ei = btrfs_item_ptr(leaf, slot, struct btrfs_extent_item);
715 flags = btrfs_extent_flags(leaf, ei);
716 *total_refs += btrfs_extent_refs(leaf, ei);
717 btrfs_item_key_to_cpu(leaf, &found_key, slot);
718
719 ptr = (unsigned long)(ei + 1);
720 end = (unsigned long)ei + item_size;
721
722 if (found_key.type == BTRFS_EXTENT_ITEM_KEY &&
723 flags & BTRFS_EXTENT_FLAG_TREE_BLOCK) {
724 struct btrfs_tree_block_info *info;
725
726 info = (struct btrfs_tree_block_info *)ptr;
727 *info_level = btrfs_tree_block_level(leaf, info);
728 ptr += sizeof(struct btrfs_tree_block_info);
729 BUG_ON(ptr > end);
730 } else if (found_key.type == BTRFS_METADATA_ITEM_KEY) {
731 *info_level = found_key.offset;
732 } else {
733 BUG_ON(!(flags & BTRFS_EXTENT_FLAG_DATA));
734 }
735
736 while (ptr < end) {
737 struct btrfs_extent_inline_ref *iref;
738 u64 offset;
739 int type;
740
741 iref = (struct btrfs_extent_inline_ref *)ptr;
742 type = btrfs_extent_inline_ref_type(leaf, iref);
743 offset = btrfs_extent_inline_ref_offset(leaf, iref);
744
745 switch (type) {
746 case BTRFS_SHARED_BLOCK_REF_KEY:
747 ret = __add_prelim_ref(prefs, 0, NULL,
748 *info_level + 1, offset,
749 bytenr, 1, GFP_NOFS);
750 break;
751 case BTRFS_SHARED_DATA_REF_KEY: {
752 struct btrfs_shared_data_ref *sdref;
753 int count;
754
755 sdref = (struct btrfs_shared_data_ref *)(iref + 1);
756 count = btrfs_shared_data_ref_count(leaf, sdref);
757 ret = __add_prelim_ref(prefs, 0, NULL, 0, offset,
758 bytenr, count, GFP_NOFS);
759 break;
760 }
761 case BTRFS_TREE_BLOCK_REF_KEY:
762 ret = __add_prelim_ref(prefs, offset, NULL,
763 *info_level + 1, 0,
764 bytenr, 1, GFP_NOFS);
765 break;
766 case BTRFS_EXTENT_DATA_REF_KEY: {
767 struct btrfs_extent_data_ref *dref;
768 int count;
769 u64 root;
770
771 dref = (struct btrfs_extent_data_ref *)(&iref->offset);
772 count = btrfs_extent_data_ref_count(leaf, dref);
773 key.objectid = btrfs_extent_data_ref_objectid(leaf,
774 dref);
775 key.type = BTRFS_EXTENT_DATA_KEY;
776 key.offset = btrfs_extent_data_ref_offset(leaf, dref);
777
778 if (inum && key.objectid != inum) {
779 ret = BACKREF_FOUND_SHARED;
780 break;
781 }
782
783 root = btrfs_extent_data_ref_root(leaf, dref);
784 ret = __add_prelim_ref(prefs, root, &key, 0, 0,
785 bytenr, count, GFP_NOFS);
786 break;
787 }
788 default:
789 WARN_ON(1);
790 }
791 if (ret)
792 return ret;
793 ptr += btrfs_extent_inline_ref_size(type);
794 }
795
796 return 0;
797 }
798
799 /*
800 * add all non-inline backrefs for bytenr to the list
801 */
802 static int __add_keyed_refs(struct btrfs_fs_info *fs_info,
803 struct btrfs_path *path, u64 bytenr,
804 int info_level, struct list_head *prefs, u64 inum)
805 {
806 struct btrfs_root *extent_root = fs_info->extent_root;
807 int ret;
808 int slot;
809 struct extent_buffer *leaf;
810 struct btrfs_key key;
811
812 while (1) {
813 ret = btrfs_next_item(extent_root, path);
814 if (ret < 0)
815 break;
816 if (ret) {
817 ret = 0;
818 break;
819 }
820
821 slot = path->slots[0];
822 leaf = path->nodes[0];
823 btrfs_item_key_to_cpu(leaf, &key, slot);
824
825 if (key.objectid != bytenr)
826 break;
827 if (key.type < BTRFS_TREE_BLOCK_REF_KEY)
828 continue;
829 if (key.type > BTRFS_SHARED_DATA_REF_KEY)
830 break;
831
832 switch (key.type) {
833 case BTRFS_SHARED_BLOCK_REF_KEY:
834 ret = __add_prelim_ref(prefs, 0, NULL,
835 info_level + 1, key.offset,
836 bytenr, 1, GFP_NOFS);
837 break;
838 case BTRFS_SHARED_DATA_REF_KEY: {
839 struct btrfs_shared_data_ref *sdref;
840 int count;
841
842 sdref = btrfs_item_ptr(leaf, slot,
843 struct btrfs_shared_data_ref);
844 count = btrfs_shared_data_ref_count(leaf, sdref);
845 ret = __add_prelim_ref(prefs, 0, NULL, 0, key.offset,
846 bytenr, count, GFP_NOFS);
847 break;
848 }
849 case BTRFS_TREE_BLOCK_REF_KEY:
850 ret = __add_prelim_ref(prefs, key.offset, NULL,
851 info_level + 1, 0,
852 bytenr, 1, GFP_NOFS);
853 break;
854 case BTRFS_EXTENT_DATA_REF_KEY: {
855 struct btrfs_extent_data_ref *dref;
856 int count;
857 u64 root;
858
859 dref = btrfs_item_ptr(leaf, slot,
860 struct btrfs_extent_data_ref);
861 count = btrfs_extent_data_ref_count(leaf, dref);
862 key.objectid = btrfs_extent_data_ref_objectid(leaf,
863 dref);
864 key.type = BTRFS_EXTENT_DATA_KEY;
865 key.offset = btrfs_extent_data_ref_offset(leaf, dref);
866
867 if (inum && key.objectid != inum) {
868 ret = BACKREF_FOUND_SHARED;
869 break;
870 }
871
872 root = btrfs_extent_data_ref_root(leaf, dref);
873 ret = __add_prelim_ref(prefs, root, &key, 0, 0,
874 bytenr, count, GFP_NOFS);
875 break;
876 }
877 default:
878 WARN_ON(1);
879 }
880 if (ret)
881 return ret;
882
883 }
884
885 return ret;
886 }
887
888 /*
889 * this adds all existing backrefs (inline backrefs, backrefs and delayed
890 * refs) for the given bytenr to the refs list, merges duplicates and resolves
891 * indirect refs to their parent bytenr.
892 * When roots are found, they're added to the roots list
893 *
894 * NOTE: This can return values > 0
895 *
896 * If time_seq is set to (u64)-1, it will not search delayed_refs, and behave
897 * much like trans == NULL case, the difference only lies in it will not
898 * commit root.
899 * The special case is for qgroup to search roots in commit_transaction().
900 *
901 * FIXME some caching might speed things up
902 */
903 static int find_parent_nodes(struct btrfs_trans_handle *trans,
904 struct btrfs_fs_info *fs_info, u64 bytenr,
905 u64 time_seq, struct ulist *refs,
906 struct ulist *roots, const u64 *extent_item_pos,
907 u64 root_objectid, u64 inum)
908 {
909 struct btrfs_key key;
910 struct btrfs_path *path;
911 struct btrfs_delayed_ref_root *delayed_refs = NULL;
912 struct btrfs_delayed_ref_head *head;
913 int info_level = 0;
914 int ret;
915 struct list_head prefs_delayed;
916 struct list_head prefs;
917 struct __prelim_ref *ref;
918 struct extent_inode_elem *eie = NULL;
919 u64 total_refs = 0;
920
921 INIT_LIST_HEAD(&prefs);
922 INIT_LIST_HEAD(&prefs_delayed);
923
924 key.objectid = bytenr;
925 key.offset = (u64)-1;
926 if (btrfs_fs_incompat(fs_info, SKINNY_METADATA))
927 key.type = BTRFS_METADATA_ITEM_KEY;
928 else
929 key.type = BTRFS_EXTENT_ITEM_KEY;
930
931 path = btrfs_alloc_path();
932 if (!path)
933 return -ENOMEM;
934 if (!trans) {
935 path->search_commit_root = 1;
936 path->skip_locking = 1;
937 }
938
939 if (time_seq == (u64)-1)
940 path->skip_locking = 1;
941
942 /*
943 * grab both a lock on the path and a lock on the delayed ref head.
944 * We need both to get a consistent picture of how the refs look
945 * at a specified point in time
946 */
947 again:
948 head = NULL;
949
950 ret = btrfs_search_slot(trans, fs_info->extent_root, &key, path, 0, 0);
951 if (ret < 0)
952 goto out;
953 BUG_ON(ret == 0);
954
955 #ifdef CONFIG_BTRFS_FS_RUN_SANITY_TESTS
956 if (trans && likely(trans->type != __TRANS_DUMMY) &&
957 time_seq != (u64)-1) {
958 #else
959 if (trans && time_seq != (u64)-1) {
960 #endif
961 /*
962 * look if there are updates for this ref queued and lock the
963 * head
964 */
965 delayed_refs = &trans->transaction->delayed_refs;
966 spin_lock(&delayed_refs->lock);
967 head = btrfs_find_delayed_ref_head(trans, bytenr);
968 if (head) {
969 if (!mutex_trylock(&head->mutex)) {
970 atomic_inc(&head->node.refs);
971 spin_unlock(&delayed_refs->lock);
972
973 btrfs_release_path(path);
974
975 /*
976 * Mutex was contended, block until it's
977 * released and try again
978 */
979 mutex_lock(&head->mutex);
980 mutex_unlock(&head->mutex);
981 btrfs_put_delayed_ref(&head->node);
982 goto again;
983 }
984 spin_unlock(&delayed_refs->lock);
985 ret = __add_delayed_refs(head, time_seq,
986 &prefs_delayed, &total_refs,
987 inum);
988 mutex_unlock(&head->mutex);
989 if (ret)
990 goto out;
991 } else {
992 spin_unlock(&delayed_refs->lock);
993 }
994 }
995
996 if (path->slots[0]) {
997 struct extent_buffer *leaf;
998 int slot;
999
1000 path->slots[0]--;
1001 leaf = path->nodes[0];
1002 slot = path->slots[0];
1003 btrfs_item_key_to_cpu(leaf, &key, slot);
1004 if (key.objectid == bytenr &&
1005 (key.type == BTRFS_EXTENT_ITEM_KEY ||
1006 key.type == BTRFS_METADATA_ITEM_KEY)) {
1007 ret = __add_inline_refs(fs_info, path, bytenr,
1008 &info_level, &prefs,
1009 &total_refs, inum);
1010 if (ret)
1011 goto out;
1012 ret = __add_keyed_refs(fs_info, path, bytenr,
1013 info_level, &prefs, inum);
1014 if (ret)
1015 goto out;
1016 }
1017 }
1018 btrfs_release_path(path);
1019
1020 list_splice_init(&prefs_delayed, &prefs);
1021
1022 ret = __add_missing_keys(fs_info, &prefs);
1023 if (ret)
1024 goto out;
1025
1026 __merge_refs(&prefs, 1);
1027
1028 ret = __resolve_indirect_refs(fs_info, path, time_seq, &prefs,
1029 extent_item_pos, total_refs,
1030 root_objectid);
1031 if (ret)
1032 goto out;
1033
1034 __merge_refs(&prefs, 2);
1035
1036 while (!list_empty(&prefs)) {
1037 ref = list_first_entry(&prefs, struct __prelim_ref, list);
1038 WARN_ON(ref->count < 0);
1039 if (roots && ref->count && ref->root_id && ref->parent == 0) {
1040 if (root_objectid && ref->root_id != root_objectid) {
1041 ret = BACKREF_FOUND_SHARED;
1042 goto out;
1043 }
1044
1045 /* no parent == root of tree */
1046 ret = ulist_add(roots, ref->root_id, 0, GFP_NOFS);
1047 if (ret < 0)
1048 goto out;
1049 }
1050 if (ref->count && ref->parent) {
1051 if (extent_item_pos && !ref->inode_list &&
1052 ref->level == 0) {
1053 struct extent_buffer *eb;
1054
1055 eb = read_tree_block(fs_info->extent_root,
1056 ref->parent, 0);
1057 if (IS_ERR(eb)) {
1058 ret = PTR_ERR(eb);
1059 goto out;
1060 } else if (!extent_buffer_uptodate(eb)) {
1061 free_extent_buffer(eb);
1062 ret = -EIO;
1063 goto out;
1064 }
1065 btrfs_tree_read_lock(eb);
1066 btrfs_set_lock_blocking_rw(eb, BTRFS_READ_LOCK);
1067 ret = find_extent_in_eb(eb, bytenr,
1068 *extent_item_pos, &eie);
1069 btrfs_tree_read_unlock_blocking(eb);
1070 free_extent_buffer(eb);
1071 if (ret < 0)
1072 goto out;
1073 ref->inode_list = eie;
1074 }
1075 ret = ulist_add_merge_ptr(refs, ref->parent,
1076 ref->inode_list,
1077 (void **)&eie, GFP_NOFS);
1078 if (ret < 0)
1079 goto out;
1080 if (!ret && extent_item_pos) {
1081 /*
1082 * we've recorded that parent, so we must extend
1083 * its inode list here
1084 */
1085 BUG_ON(!eie);
1086 while (eie->next)
1087 eie = eie->next;
1088 eie->next = ref->inode_list;
1089 }
1090 eie = NULL;
1091 }
1092 list_del(&ref->list);
1093 kmem_cache_free(btrfs_prelim_ref_cache, ref);
1094 }
1095
1096 out:
1097 btrfs_free_path(path);
1098 while (!list_empty(&prefs)) {
1099 ref = list_first_entry(&prefs, struct __prelim_ref, list);
1100 list_del(&ref->list);
1101 kmem_cache_free(btrfs_prelim_ref_cache, ref);
1102 }
1103 while (!list_empty(&prefs_delayed)) {
1104 ref = list_first_entry(&prefs_delayed, struct __prelim_ref,
1105 list);
1106 list_del(&ref->list);
1107 kmem_cache_free(btrfs_prelim_ref_cache, ref);
1108 }
1109 if (ret < 0)
1110 free_inode_elem_list(eie);
1111 return ret;
1112 }
1113
1114 static void free_leaf_list(struct ulist *blocks)
1115 {
1116 struct ulist_node *node = NULL;
1117 struct extent_inode_elem *eie;
1118 struct ulist_iterator uiter;
1119
1120 ULIST_ITER_INIT(&uiter);
1121 while ((node = ulist_next(blocks, &uiter))) {
1122 if (!node->aux)
1123 continue;
1124 eie = (struct extent_inode_elem *)(uintptr_t)node->aux;
1125 free_inode_elem_list(eie);
1126 node->aux = 0;
1127 }
1128
1129 ulist_free(blocks);
1130 }
1131
1132 /*
1133 * Finds all leafs with a reference to the specified combination of bytenr and
1134 * offset. key_list_head will point to a list of corresponding keys (caller must
1135 * free each list element). The leafs will be stored in the leafs ulist, which
1136 * must be freed with ulist_free.
1137 *
1138 * returns 0 on success, <0 on error
1139 */
1140 static int btrfs_find_all_leafs(struct btrfs_trans_handle *trans,
1141 struct btrfs_fs_info *fs_info, u64 bytenr,
1142 u64 time_seq, struct ulist **leafs,
1143 const u64 *extent_item_pos)
1144 {
1145 int ret;
1146
1147 *leafs = ulist_alloc(GFP_NOFS);
1148 if (!*leafs)
1149 return -ENOMEM;
1150
1151 ret = find_parent_nodes(trans, fs_info, bytenr,
1152 time_seq, *leafs, NULL, extent_item_pos, 0, 0);
1153 if (ret < 0 && ret != -ENOENT) {
1154 free_leaf_list(*leafs);
1155 return ret;
1156 }
1157
1158 return 0;
1159 }
1160
1161 /*
1162 * walk all backrefs for a given extent to find all roots that reference this
1163 * extent. Walking a backref means finding all extents that reference this
1164 * extent and in turn walk the backrefs of those, too. Naturally this is a
1165 * recursive process, but here it is implemented in an iterative fashion: We
1166 * find all referencing extents for the extent in question and put them on a
1167 * list. In turn, we find all referencing extents for those, further appending
1168 * to the list. The way we iterate the list allows adding more elements after
1169 * the current while iterating. The process stops when we reach the end of the
1170 * list. Found roots are added to the roots list.
1171 *
1172 * returns 0 on success, < 0 on error.
1173 */
1174 static int __btrfs_find_all_roots(struct btrfs_trans_handle *trans,
1175 struct btrfs_fs_info *fs_info, u64 bytenr,
1176 u64 time_seq, struct ulist **roots)
1177 {
1178 struct ulist *tmp;
1179 struct ulist_node *node = NULL;
1180 struct ulist_iterator uiter;
1181 int ret;
1182
1183 tmp = ulist_alloc(GFP_NOFS);
1184 if (!tmp)
1185 return -ENOMEM;
1186 *roots = ulist_alloc(GFP_NOFS);
1187 if (!*roots) {
1188 ulist_free(tmp);
1189 return -ENOMEM;
1190 }
1191
1192 ULIST_ITER_INIT(&uiter);
1193 while (1) {
1194 ret = find_parent_nodes(trans, fs_info, bytenr,
1195 time_seq, tmp, *roots, NULL, 0, 0);
1196 if (ret < 0 && ret != -ENOENT) {
1197 ulist_free(tmp);
1198 ulist_free(*roots);
1199 return ret;
1200 }
1201 node = ulist_next(tmp, &uiter);
1202 if (!node)
1203 break;
1204 bytenr = node->val;
1205 cond_resched();
1206 }
1207
1208 ulist_free(tmp);
1209 return 0;
1210 }
1211
1212 int btrfs_find_all_roots(struct btrfs_trans_handle *trans,
1213 struct btrfs_fs_info *fs_info, u64 bytenr,
1214 u64 time_seq, struct ulist **roots)
1215 {
1216 int ret;
1217
1218 if (!trans)
1219 down_read(&fs_info->commit_root_sem);
1220 ret = __btrfs_find_all_roots(trans, fs_info, bytenr, time_seq, roots);
1221 if (!trans)
1222 up_read(&fs_info->commit_root_sem);
1223 return ret;
1224 }
1225
1226 /**
1227 * btrfs_check_shared - tell us whether an extent is shared
1228 *
1229 * @trans: optional trans handle
1230 *
1231 * btrfs_check_shared uses the backref walking code but will short
1232 * circuit as soon as it finds a root or inode that doesn't match the
1233 * one passed in. This provides a significant performance benefit for
1234 * callers (such as fiemap) which want to know whether the extent is
1235 * shared but do not need a ref count.
1236 *
1237 * Return: 0 if extent is not shared, 1 if it is shared, < 0 on error.
1238 */
1239 int btrfs_check_shared(struct btrfs_trans_handle *trans,
1240 struct btrfs_fs_info *fs_info, u64 root_objectid,
1241 u64 inum, u64 bytenr)
1242 {
1243 struct ulist *tmp = NULL;
1244 struct ulist *roots = NULL;
1245 struct ulist_iterator uiter;
1246 struct ulist_node *node;
1247 struct seq_list elem = SEQ_LIST_INIT(elem);
1248 int ret = 0;
1249
1250 tmp = ulist_alloc(GFP_NOFS);
1251 roots = ulist_alloc(GFP_NOFS);
1252 if (!tmp || !roots) {
1253 ulist_free(tmp);
1254 ulist_free(roots);
1255 return -ENOMEM;
1256 }
1257
1258 if (trans)
1259 btrfs_get_tree_mod_seq(fs_info, &elem);
1260 else
1261 down_read(&fs_info->commit_root_sem);
1262 ULIST_ITER_INIT(&uiter);
1263 while (1) {
1264 ret = find_parent_nodes(trans, fs_info, bytenr, elem.seq, tmp,
1265 roots, NULL, root_objectid, inum);
1266 if (ret == BACKREF_FOUND_SHARED) {
1267 /* this is the only condition under which we return 1 */
1268 ret = 1;
1269 break;
1270 }
1271 if (ret < 0 && ret != -ENOENT)
1272 break;
1273 ret = 0;
1274 node = ulist_next(tmp, &uiter);
1275 if (!node)
1276 break;
1277 bytenr = node->val;
1278 cond_resched();
1279 }
1280 if (trans)
1281 btrfs_put_tree_mod_seq(fs_info, &elem);
1282 else
1283 up_read(&fs_info->commit_root_sem);
1284 ulist_free(tmp);
1285 ulist_free(roots);
1286 return ret;
1287 }
1288
1289 int btrfs_find_one_extref(struct btrfs_root *root, u64 inode_objectid,
1290 u64 start_off, struct btrfs_path *path,
1291 struct btrfs_inode_extref **ret_extref,
1292 u64 *found_off)
1293 {
1294 int ret, slot;
1295 struct btrfs_key key;
1296 struct btrfs_key found_key;
1297 struct btrfs_inode_extref *extref;
1298 struct extent_buffer *leaf;
1299 unsigned long ptr;
1300
1301 key.objectid = inode_objectid;
1302 key.type = BTRFS_INODE_EXTREF_KEY;
1303 key.offset = start_off;
1304
1305 ret = btrfs_search_slot(NULL, root, &key, path, 0, 0);
1306 if (ret < 0)
1307 return ret;
1308
1309 while (1) {
1310 leaf = path->nodes[0];
1311 slot = path->slots[0];
1312 if (slot >= btrfs_header_nritems(leaf)) {
1313 /*
1314 * If the item at offset is not found,
1315 * btrfs_search_slot will point us to the slot
1316 * where it should be inserted. In our case
1317 * that will be the slot directly before the
1318 * next INODE_REF_KEY_V2 item. In the case
1319 * that we're pointing to the last slot in a
1320 * leaf, we must move one leaf over.
1321 */
1322 ret = btrfs_next_leaf(root, path);
1323 if (ret) {
1324 if (ret >= 1)
1325 ret = -ENOENT;
1326 break;
1327 }
1328 continue;
1329 }
1330
1331 btrfs_item_key_to_cpu(leaf, &found_key, slot);
1332
1333 /*
1334 * Check that we're still looking at an extended ref key for
1335 * this particular objectid. If we have different
1336 * objectid or type then there are no more to be found
1337 * in the tree and we can exit.
1338 */
1339 ret = -ENOENT;
1340 if (found_key.objectid != inode_objectid)
1341 break;
1342 if (found_key.type != BTRFS_INODE_EXTREF_KEY)
1343 break;
1344
1345 ret = 0;
1346 ptr = btrfs_item_ptr_offset(leaf, path->slots[0]);
1347 extref = (struct btrfs_inode_extref *)ptr;
1348 *ret_extref = extref;
1349 if (found_off)
1350 *found_off = found_key.offset;
1351 break;
1352 }
1353
1354 return ret;
1355 }
1356
1357 /*
1358 * this iterates to turn a name (from iref/extref) into a full filesystem path.
1359 * Elements of the path are separated by '/' and the path is guaranteed to be
1360 * 0-terminated. the path is only given within the current file system.
1361 * Therefore, it never starts with a '/'. the caller is responsible to provide
1362 * "size" bytes in "dest". the dest buffer will be filled backwards. finally,
1363 * the start point of the resulting string is returned. this pointer is within
1364 * dest, normally.
1365 * in case the path buffer would overflow, the pointer is decremented further
1366 * as if output was written to the buffer, though no more output is actually
1367 * generated. that way, the caller can determine how much space would be
1368 * required for the path to fit into the buffer. in that case, the returned
1369 * value will be smaller than dest. callers must check this!
1370 */
1371 char *btrfs_ref_to_path(struct btrfs_root *fs_root, struct btrfs_path *path,
1372 u32 name_len, unsigned long name_off,
1373 struct extent_buffer *eb_in, u64 parent,
1374 char *dest, u32 size)
1375 {
1376 int slot;
1377 u64 next_inum;
1378 int ret;
1379 s64 bytes_left = ((s64)size) - 1;
1380 struct extent_buffer *eb = eb_in;
1381 struct btrfs_key found_key;
1382 int leave_spinning = path->leave_spinning;
1383 struct btrfs_inode_ref *iref;
1384
1385 if (bytes_left >= 0)
1386 dest[bytes_left] = '\0';
1387
1388 path->leave_spinning = 1;
1389 while (1) {
1390 bytes_left -= name_len;
1391 if (bytes_left >= 0)
1392 read_extent_buffer(eb, dest + bytes_left,
1393 name_off, name_len);
1394 if (eb != eb_in) {
1395 btrfs_tree_read_unlock_blocking(eb);
1396 free_extent_buffer(eb);
1397 }
1398 ret = btrfs_find_item(fs_root, path, parent, 0,
1399 BTRFS_INODE_REF_KEY, &found_key);
1400 if (ret > 0)
1401 ret = -ENOENT;
1402 if (ret)
1403 break;
1404
1405 next_inum = found_key.offset;
1406
1407 /* regular exit ahead */
1408 if (parent == next_inum)
1409 break;
1410
1411 slot = path->slots[0];
1412 eb = path->nodes[0];
1413 /* make sure we can use eb after releasing the path */
1414 if (eb != eb_in) {
1415 atomic_inc(&eb->refs);
1416 btrfs_tree_read_lock(eb);
1417 btrfs_set_lock_blocking_rw(eb, BTRFS_READ_LOCK);
1418 }
1419 btrfs_release_path(path);
1420 iref = btrfs_item_ptr(eb, slot, struct btrfs_inode_ref);
1421
1422 name_len = btrfs_inode_ref_name_len(eb, iref);
1423 name_off = (unsigned long)(iref + 1);
1424
1425 parent = next_inum;
1426 --bytes_left;
1427 if (bytes_left >= 0)
1428 dest[bytes_left] = '/';
1429 }
1430
1431 btrfs_release_path(path);
1432 path->leave_spinning = leave_spinning;
1433
1434 if (ret)
1435 return ERR_PTR(ret);
1436
1437 return dest + bytes_left;
1438 }
1439
1440 /*
1441 * this makes the path point to (logical EXTENT_ITEM *)
1442 * returns BTRFS_EXTENT_FLAG_DATA for data, BTRFS_EXTENT_FLAG_TREE_BLOCK for
1443 * tree blocks and <0 on error.
1444 */
1445 int extent_from_logical(struct btrfs_fs_info *fs_info, u64 logical,
1446 struct btrfs_path *path, struct btrfs_key *found_key,
1447 u64 *flags_ret)
1448 {
1449 int ret;
1450 u64 flags;
1451 u64 size = 0;
1452 u32 item_size;
1453 struct extent_buffer *eb;
1454 struct btrfs_extent_item *ei;
1455 struct btrfs_key key;
1456
1457 if (btrfs_fs_incompat(fs_info, SKINNY_METADATA))
1458 key.type = BTRFS_METADATA_ITEM_KEY;
1459 else
1460 key.type = BTRFS_EXTENT_ITEM_KEY;
1461 key.objectid = logical;
1462 key.offset = (u64)-1;
1463
1464 ret = btrfs_search_slot(NULL, fs_info->extent_root, &key, path, 0, 0);
1465 if (ret < 0)
1466 return ret;
1467
1468 ret = btrfs_previous_extent_item(fs_info->extent_root, path, 0);
1469 if (ret) {
1470 if (ret > 0)
1471 ret = -ENOENT;
1472 return ret;
1473 }
1474 btrfs_item_key_to_cpu(path->nodes[0], found_key, path->slots[0]);
1475 if (found_key->type == BTRFS_METADATA_ITEM_KEY)
1476 size = fs_info->extent_root->nodesize;
1477 else if (found_key->type == BTRFS_EXTENT_ITEM_KEY)
1478 size = found_key->offset;
1479
1480 if (found_key->objectid > logical ||
1481 found_key->objectid + size <= logical) {
1482 pr_debug("logical %llu is not within any extent\n", logical);
1483 return -ENOENT;
1484 }
1485
1486 eb = path->nodes[0];
1487 item_size = btrfs_item_size_nr(eb, path->slots[0]);
1488 BUG_ON(item_size < sizeof(*ei));
1489
1490 ei = btrfs_item_ptr(eb, path->slots[0], struct btrfs_extent_item);
1491 flags = btrfs_extent_flags(eb, ei);
1492
1493 pr_debug("logical %llu is at position %llu within the extent (%llu "
1494 "EXTENT_ITEM %llu) flags %#llx size %u\n",
1495 logical, logical - found_key->objectid, found_key->objectid,
1496 found_key->offset, flags, item_size);
1497
1498 WARN_ON(!flags_ret);
1499 if (flags_ret) {
1500 if (flags & BTRFS_EXTENT_FLAG_TREE_BLOCK)
1501 *flags_ret = BTRFS_EXTENT_FLAG_TREE_BLOCK;
1502 else if (flags & BTRFS_EXTENT_FLAG_DATA)
1503 *flags_ret = BTRFS_EXTENT_FLAG_DATA;
1504 else
1505 BUG_ON(1);
1506 return 0;
1507 }
1508
1509 return -EIO;
1510 }
1511
1512 /*
1513 * helper function to iterate extent inline refs. ptr must point to a 0 value
1514 * for the first call and may be modified. it is used to track state.
1515 * if more refs exist, 0 is returned and the next call to
1516 * __get_extent_inline_ref must pass the modified ptr parameter to get the
1517 * next ref. after the last ref was processed, 1 is returned.
1518 * returns <0 on error
1519 */
1520 static int __get_extent_inline_ref(unsigned long *ptr, struct extent_buffer *eb,
1521 struct btrfs_key *key,
1522 struct btrfs_extent_item *ei, u32 item_size,
1523 struct btrfs_extent_inline_ref **out_eiref,
1524 int *out_type)
1525 {
1526 unsigned long end;
1527 u64 flags;
1528 struct btrfs_tree_block_info *info;
1529
1530 if (!*ptr) {
1531 /* first call */
1532 flags = btrfs_extent_flags(eb, ei);
1533 if (flags & BTRFS_EXTENT_FLAG_TREE_BLOCK) {
1534 if (key->type == BTRFS_METADATA_ITEM_KEY) {
1535 /* a skinny metadata extent */
1536 *out_eiref =
1537 (struct btrfs_extent_inline_ref *)(ei + 1);
1538 } else {
1539 WARN_ON(key->type != BTRFS_EXTENT_ITEM_KEY);
1540 info = (struct btrfs_tree_block_info *)(ei + 1);
1541 *out_eiref =
1542 (struct btrfs_extent_inline_ref *)(info + 1);
1543 }
1544 } else {
1545 *out_eiref = (struct btrfs_extent_inline_ref *)(ei + 1);
1546 }
1547 *ptr = (unsigned long)*out_eiref;
1548 if ((unsigned long)(*ptr) >= (unsigned long)ei + item_size)
1549 return -ENOENT;
1550 }
1551
1552 end = (unsigned long)ei + item_size;
1553 *out_eiref = (struct btrfs_extent_inline_ref *)(*ptr);
1554 *out_type = btrfs_extent_inline_ref_type(eb, *out_eiref);
1555
1556 *ptr += btrfs_extent_inline_ref_size(*out_type);
1557 WARN_ON(*ptr > end);
1558 if (*ptr == end)
1559 return 1; /* last */
1560
1561 return 0;
1562 }
1563
1564 /*
1565 * reads the tree block backref for an extent. tree level and root are returned
1566 * through out_level and out_root. ptr must point to a 0 value for the first
1567 * call and may be modified (see __get_extent_inline_ref comment).
1568 * returns 0 if data was provided, 1 if there was no more data to provide or
1569 * <0 on error.
1570 */
1571 int tree_backref_for_extent(unsigned long *ptr, struct extent_buffer *eb,
1572 struct btrfs_key *key, struct btrfs_extent_item *ei,
1573 u32 item_size, u64 *out_root, u8 *out_level)
1574 {
1575 int ret;
1576 int type;
1577 struct btrfs_extent_inline_ref *eiref;
1578
1579 if (*ptr == (unsigned long)-1)
1580 return 1;
1581
1582 while (1) {
1583 ret = __get_extent_inline_ref(ptr, eb, key, ei, item_size,
1584 &eiref, &type);
1585 if (ret < 0)
1586 return ret;
1587
1588 if (type == BTRFS_TREE_BLOCK_REF_KEY ||
1589 type == BTRFS_SHARED_BLOCK_REF_KEY)
1590 break;
1591
1592 if (ret == 1)
1593 return 1;
1594 }
1595
1596 /* we can treat both ref types equally here */
1597 *out_root = btrfs_extent_inline_ref_offset(eb, eiref);
1598
1599 if (key->type == BTRFS_EXTENT_ITEM_KEY) {
1600 struct btrfs_tree_block_info *info;
1601
1602 info = (struct btrfs_tree_block_info *)(ei + 1);
1603 *out_level = btrfs_tree_block_level(eb, info);
1604 } else {
1605 ASSERT(key->type == BTRFS_METADATA_ITEM_KEY);
1606 *out_level = (u8)key->offset;
1607 }
1608
1609 if (ret == 1)
1610 *ptr = (unsigned long)-1;
1611
1612 return 0;
1613 }
1614
1615 static int iterate_leaf_refs(struct extent_inode_elem *inode_list,
1616 u64 root, u64 extent_item_objectid,
1617 iterate_extent_inodes_t *iterate, void *ctx)
1618 {
1619 struct extent_inode_elem *eie;
1620 int ret = 0;
1621
1622 for (eie = inode_list; eie; eie = eie->next) {
1623 pr_debug("ref for %llu resolved, key (%llu EXTEND_DATA %llu), "
1624 "root %llu\n", extent_item_objectid,
1625 eie->inum, eie->offset, root);
1626 ret = iterate(eie->inum, eie->offset, root, ctx);
1627 if (ret) {
1628 pr_debug("stopping iteration for %llu due to ret=%d\n",
1629 extent_item_objectid, ret);
1630 break;
1631 }
1632 }
1633
1634 return ret;
1635 }
1636
1637 /*
1638 * calls iterate() for every inode that references the extent identified by
1639 * the given parameters.
1640 * when the iterator function returns a non-zero value, iteration stops.
1641 */
1642 int iterate_extent_inodes(struct btrfs_fs_info *fs_info,
1643 u64 extent_item_objectid, u64 extent_item_pos,
1644 int search_commit_root,
1645 iterate_extent_inodes_t *iterate, void *ctx)
1646 {
1647 int ret;
1648 struct btrfs_trans_handle *trans = NULL;
1649 struct ulist *refs = NULL;
1650 struct ulist *roots = NULL;
1651 struct ulist_node *ref_node = NULL;
1652 struct ulist_node *root_node = NULL;
1653 struct seq_list tree_mod_seq_elem = SEQ_LIST_INIT(tree_mod_seq_elem);
1654 struct ulist_iterator ref_uiter;
1655 struct ulist_iterator root_uiter;
1656
1657 pr_debug("resolving all inodes for extent %llu\n",
1658 extent_item_objectid);
1659
1660 if (!search_commit_root) {
1661 trans = btrfs_join_transaction(fs_info->extent_root);
1662 if (IS_ERR(trans))
1663 return PTR_ERR(trans);
1664 btrfs_get_tree_mod_seq(fs_info, &tree_mod_seq_elem);
1665 } else {
1666 down_read(&fs_info->commit_root_sem);
1667 }
1668
1669 ret = btrfs_find_all_leafs(trans, fs_info, extent_item_objectid,
1670 tree_mod_seq_elem.seq, &refs,
1671 &extent_item_pos);
1672 if (ret)
1673 goto out;
1674
1675 ULIST_ITER_INIT(&ref_uiter);
1676 while (!ret && (ref_node = ulist_next(refs, &ref_uiter))) {
1677 ret = __btrfs_find_all_roots(trans, fs_info, ref_node->val,
1678 tree_mod_seq_elem.seq, &roots);
1679 if (ret)
1680 break;
1681 ULIST_ITER_INIT(&root_uiter);
1682 while (!ret && (root_node = ulist_next(roots, &root_uiter))) {
1683 pr_debug("root %llu references leaf %llu, data list "
1684 "%#llx\n", root_node->val, ref_node->val,
1685 ref_node->aux);
1686 ret = iterate_leaf_refs((struct extent_inode_elem *)
1687 (uintptr_t)ref_node->aux,
1688 root_node->val,
1689 extent_item_objectid,
1690 iterate, ctx);
1691 }
1692 ulist_free(roots);
1693 }
1694
1695 free_leaf_list(refs);
1696 out:
1697 if (!search_commit_root) {
1698 btrfs_put_tree_mod_seq(fs_info, &tree_mod_seq_elem);
1699 btrfs_end_transaction(trans, fs_info->extent_root);
1700 } else {
1701 up_read(&fs_info->commit_root_sem);
1702 }
1703
1704 return ret;
1705 }
1706
1707 int iterate_inodes_from_logical(u64 logical, struct btrfs_fs_info *fs_info,
1708 struct btrfs_path *path,
1709 iterate_extent_inodes_t *iterate, void *ctx)
1710 {
1711 int ret;
1712 u64 extent_item_pos;
1713 u64 flags = 0;
1714 struct btrfs_key found_key;
1715 int search_commit_root = path->search_commit_root;
1716
1717 ret = extent_from_logical(fs_info, logical, path, &found_key, &flags);
1718 btrfs_release_path(path);
1719 if (ret < 0)
1720 return ret;
1721 if (flags & BTRFS_EXTENT_FLAG_TREE_BLOCK)
1722 return -EINVAL;
1723
1724 extent_item_pos = logical - found_key.objectid;
1725 ret = iterate_extent_inodes(fs_info, found_key.objectid,
1726 extent_item_pos, search_commit_root,
1727 iterate, ctx);
1728
1729 return ret;
1730 }
1731
1732 typedef int (iterate_irefs_t)(u64 parent, u32 name_len, unsigned long name_off,
1733 struct extent_buffer *eb, void *ctx);
1734
1735 static int iterate_inode_refs(u64 inum, struct btrfs_root *fs_root,
1736 struct btrfs_path *path,
1737 iterate_irefs_t *iterate, void *ctx)
1738 {
1739 int ret = 0;
1740 int slot;
1741 u32 cur;
1742 u32 len;
1743 u32 name_len;
1744 u64 parent = 0;
1745 int found = 0;
1746 struct extent_buffer *eb;
1747 struct btrfs_item *item;
1748 struct btrfs_inode_ref *iref;
1749 struct btrfs_key found_key;
1750
1751 while (!ret) {
1752 ret = btrfs_find_item(fs_root, path, inum,
1753 parent ? parent + 1 : 0, BTRFS_INODE_REF_KEY,
1754 &found_key);
1755
1756 if (ret < 0)
1757 break;
1758 if (ret) {
1759 ret = found ? 0 : -ENOENT;
1760 break;
1761 }
1762 ++found;
1763
1764 parent = found_key.offset;
1765 slot = path->slots[0];
1766 eb = btrfs_clone_extent_buffer(path->nodes[0]);
1767 if (!eb) {
1768 ret = -ENOMEM;
1769 break;
1770 }
1771 extent_buffer_get(eb);
1772 btrfs_tree_read_lock(eb);
1773 btrfs_set_lock_blocking_rw(eb, BTRFS_READ_LOCK);
1774 btrfs_release_path(path);
1775
1776 item = btrfs_item_nr(slot);
1777 iref = btrfs_item_ptr(eb, slot, struct btrfs_inode_ref);
1778
1779 for (cur = 0; cur < btrfs_item_size(eb, item); cur += len) {
1780 name_len = btrfs_inode_ref_name_len(eb, iref);
1781 /* path must be released before calling iterate()! */
1782 pr_debug("following ref at offset %u for inode %llu in "
1783 "tree %llu\n", cur, found_key.objectid,
1784 fs_root->objectid);
1785 ret = iterate(parent, name_len,
1786 (unsigned long)(iref + 1), eb, ctx);
1787 if (ret)
1788 break;
1789 len = sizeof(*iref) + name_len;
1790 iref = (struct btrfs_inode_ref *)((char *)iref + len);
1791 }
1792 btrfs_tree_read_unlock_blocking(eb);
1793 free_extent_buffer(eb);
1794 }
1795
1796 btrfs_release_path(path);
1797
1798 return ret;
1799 }
1800
1801 static int iterate_inode_extrefs(u64 inum, struct btrfs_root *fs_root,
1802 struct btrfs_path *path,
1803 iterate_irefs_t *iterate, void *ctx)
1804 {
1805 int ret;
1806 int slot;
1807 u64 offset = 0;
1808 u64 parent;
1809 int found = 0;
1810 struct extent_buffer *eb;
1811 struct btrfs_inode_extref *extref;
1812 struct extent_buffer *leaf;
1813 u32 item_size;
1814 u32 cur_offset;
1815 unsigned long ptr;
1816
1817 while (1) {
1818 ret = btrfs_find_one_extref(fs_root, inum, offset, path, &extref,
1819 &offset);
1820 if (ret < 0)
1821 break;
1822 if (ret) {
1823 ret = found ? 0 : -ENOENT;
1824 break;
1825 }
1826 ++found;
1827
1828 slot = path->slots[0];
1829 eb = btrfs_clone_extent_buffer(path->nodes[0]);
1830 if (!eb) {
1831 ret = -ENOMEM;
1832 break;
1833 }
1834 extent_buffer_get(eb);
1835
1836 btrfs_tree_read_lock(eb);
1837 btrfs_set_lock_blocking_rw(eb, BTRFS_READ_LOCK);
1838 btrfs_release_path(path);
1839
1840 leaf = path->nodes[0];
1841 item_size = btrfs_item_size_nr(leaf, slot);
1842 ptr = btrfs_item_ptr_offset(leaf, slot);
1843 cur_offset = 0;
1844
1845 while (cur_offset < item_size) {
1846 u32 name_len;
1847
1848 extref = (struct btrfs_inode_extref *)(ptr + cur_offset);
1849 parent = btrfs_inode_extref_parent(eb, extref);
1850 name_len = btrfs_inode_extref_name_len(eb, extref);
1851 ret = iterate(parent, name_len,
1852 (unsigned long)&extref->name, eb, ctx);
1853 if (ret)
1854 break;
1855
1856 cur_offset += btrfs_inode_extref_name_len(leaf, extref);
1857 cur_offset += sizeof(*extref);
1858 }
1859 btrfs_tree_read_unlock_blocking(eb);
1860 free_extent_buffer(eb);
1861
1862 offset++;
1863 }
1864
1865 btrfs_release_path(path);
1866
1867 return ret;
1868 }
1869
1870 static int iterate_irefs(u64 inum, struct btrfs_root *fs_root,
1871 struct btrfs_path *path, iterate_irefs_t *iterate,
1872 void *ctx)
1873 {
1874 int ret;
1875 int found_refs = 0;
1876
1877 ret = iterate_inode_refs(inum, fs_root, path, iterate, ctx);
1878 if (!ret)
1879 ++found_refs;
1880 else if (ret != -ENOENT)
1881 return ret;
1882
1883 ret = iterate_inode_extrefs(inum, fs_root, path, iterate, ctx);
1884 if (ret == -ENOENT && found_refs)
1885 return 0;
1886
1887 return ret;
1888 }
1889
1890 /*
1891 * returns 0 if the path could be dumped (probably truncated)
1892 * returns <0 in case of an error
1893 */
1894 static int inode_to_path(u64 inum, u32 name_len, unsigned long name_off,
1895 struct extent_buffer *eb, void *ctx)
1896 {
1897 struct inode_fs_paths *ipath = ctx;
1898 char *fspath;
1899 char *fspath_min;
1900 int i = ipath->fspath->elem_cnt;
1901 const int s_ptr = sizeof(char *);
1902 u32 bytes_left;
1903
1904 bytes_left = ipath->fspath->bytes_left > s_ptr ?
1905 ipath->fspath->bytes_left - s_ptr : 0;
1906
1907 fspath_min = (char *)ipath->fspath->val + (i + 1) * s_ptr;
1908 fspath = btrfs_ref_to_path(ipath->fs_root, ipath->btrfs_path, name_len,
1909 name_off, eb, inum, fspath_min, bytes_left);
1910 if (IS_ERR(fspath))
1911 return PTR_ERR(fspath);
1912
1913 if (fspath > fspath_min) {
1914 ipath->fspath->val[i] = (u64)(unsigned long)fspath;
1915 ++ipath->fspath->elem_cnt;
1916 ipath->fspath->bytes_left = fspath - fspath_min;
1917 } else {
1918 ++ipath->fspath->elem_missed;
1919 ipath->fspath->bytes_missing += fspath_min - fspath;
1920 ipath->fspath->bytes_left = 0;
1921 }
1922
1923 return 0;
1924 }
1925
1926 /*
1927 * this dumps all file system paths to the inode into the ipath struct, provided
1928 * is has been created large enough. each path is zero-terminated and accessed
1929 * from ipath->fspath->val[i].
1930 * when it returns, there are ipath->fspath->elem_cnt number of paths available
1931 * in ipath->fspath->val[]. when the allocated space wasn't sufficient, the
1932 * number of missed paths in recored in ipath->fspath->elem_missed, otherwise,
1933 * it's zero. ipath->fspath->bytes_missing holds the number of bytes that would
1934 * have been needed to return all paths.
1935 */
1936 int paths_from_inode(u64 inum, struct inode_fs_paths *ipath)
1937 {
1938 return iterate_irefs(inum, ipath->fs_root, ipath->btrfs_path,
1939 inode_to_path, ipath);
1940 }
1941
1942 struct btrfs_data_container *init_data_container(u32 total_bytes)
1943 {
1944 struct btrfs_data_container *data;
1945 size_t alloc_bytes;
1946
1947 alloc_bytes = max_t(size_t, total_bytes, sizeof(*data));
1948 data = vmalloc(alloc_bytes);
1949 if (!data)
1950 return ERR_PTR(-ENOMEM);
1951
1952 if (total_bytes >= sizeof(*data)) {
1953 data->bytes_left = total_bytes - sizeof(*data);
1954 data->bytes_missing = 0;
1955 } else {
1956 data->bytes_missing = sizeof(*data) - total_bytes;
1957 data->bytes_left = 0;
1958 }
1959
1960 data->elem_cnt = 0;
1961 data->elem_missed = 0;
1962
1963 return data;
1964 }
1965
1966 /*
1967 * allocates space to return multiple file system paths for an inode.
1968 * total_bytes to allocate are passed, note that space usable for actual path
1969 * information will be total_bytes - sizeof(struct inode_fs_paths).
1970 * the returned pointer must be freed with free_ipath() in the end.
1971 */
1972 struct inode_fs_paths *init_ipath(s32 total_bytes, struct btrfs_root *fs_root,
1973 struct btrfs_path *path)
1974 {
1975 struct inode_fs_paths *ifp;
1976 struct btrfs_data_container *fspath;
1977
1978 fspath = init_data_container(total_bytes);
1979 if (IS_ERR(fspath))
1980 return (void *)fspath;
1981
1982 ifp = kmalloc(sizeof(*ifp), GFP_NOFS);
1983 if (!ifp) {
1984 kfree(fspath);
1985 return ERR_PTR(-ENOMEM);
1986 }
1987
1988 ifp->btrfs_path = path;
1989 ifp->fspath = fspath;
1990 ifp->fs_root = fs_root;
1991
1992 return ifp;
1993 }
1994
1995 void free_ipath(struct inode_fs_paths *ipath)
1996 {
1997 if (!ipath)
1998 return;
1999 vfree(ipath->fspath);
2000 kfree(ipath);
2001 }
Linux with added FPC-III support