?  search:
summary | log | graphiclog1 | graphiclog2 | commit | commitdiff | tree | refs | edit | fork
blame | history | raw | HEAD
powerpc/perf: Fix book3s kernel to userspace backtraces
[linux/fpc-iii.git] / fs / btrfs / backref.c
blobf55721ff938544c1b73d0d8cb057a52dba5b7884
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 ret = btrfs_next_old_leaf(root, path, time_seq);
255
256 while (!ret && count < total_refs) {
257 eb = path->nodes[0];
258 slot = path->slots[0];
259
260 btrfs_item_key_to_cpu(eb, &key, slot);
261
262 if (key.objectid != key_for_search->objectid ||
263 key.type != BTRFS_EXTENT_DATA_KEY)
264 break;
265
266 fi = btrfs_item_ptr(eb, slot, struct btrfs_file_extent_item);
267 disk_byte = btrfs_file_extent_disk_bytenr(eb, fi);
268
269 if (disk_byte == wanted_disk_byte) {
270 eie = NULL;
271 old = NULL;
272 count++;
273 if (extent_item_pos) {
274 ret = check_extent_in_eb(&key, eb, fi,
275 *extent_item_pos,
276 &eie);
277 if (ret < 0)
278 break;
279 }
280 if (ret > 0)
281 goto next;
282 ret = ulist_add_merge_ptr(parents, eb->start,
283 eie, (void **)&old, GFP_NOFS);
284 if (ret < 0)
285 break;
286 if (!ret && extent_item_pos) {
287 while (old->next)
288 old = old->next;
289 old->next = eie;
290 }
291 eie = NULL;
292 }
293 next:
294 ret = btrfs_next_old_item(root, path, time_seq);
295 }
296
297 if (ret > 0)
298 ret = 0;
299 else if (ret < 0)
300 free_inode_elem_list(eie);
301 return ret;
302 }
303
304 /*
305 * resolve an indirect backref in the form (root_id, key, level)
306 * to a logical address
307 */
308 static int __resolve_indirect_ref(struct btrfs_fs_info *fs_info,
309 struct btrfs_path *path, u64 time_seq,
310 struct __prelim_ref *ref,
311 struct ulist *parents,
312 const u64 *extent_item_pos, u64 total_refs)
313 {
314 struct btrfs_root *root;
315 struct btrfs_key root_key;
316 struct extent_buffer *eb;
317 int ret = 0;
318 int root_level;
319 int level = ref->level;
320 int index;
321
322 root_key.objectid = ref->root_id;
323 root_key.type = BTRFS_ROOT_ITEM_KEY;
324 root_key.offset = (u64)-1;
325
326 index = srcu_read_lock(&fs_info->subvol_srcu);
327
328 root = btrfs_read_fs_root_no_name(fs_info, &root_key);
329 if (IS_ERR(root)) {
330 srcu_read_unlock(&fs_info->subvol_srcu, index);
331 ret = PTR_ERR(root);
332 goto out;
333 }
334
335 if (path->search_commit_root)
336 root_level = btrfs_header_level(root->commit_root);
337 else
338 root_level = btrfs_old_root_level(root, time_seq);
339
340 if (root_level + 1 == level) {
341 srcu_read_unlock(&fs_info->subvol_srcu, index);
342 goto out;
343 }
344
345 path->lowest_level = level;
346 ret = btrfs_search_old_slot(root, &ref->key_for_search, path, time_seq);
347
348 /* root node has been locked, we can release @subvol_srcu safely here */
349 srcu_read_unlock(&fs_info->subvol_srcu, index);
350
351 pr_debug("search slot in root %llu (level %d, ref count %d) returned "
352 "%d for key (%llu %u %llu)\n",
353 ref->root_id, level, ref->count, ret,
354 ref->key_for_search.objectid, ref->key_for_search.type,
355 ref->key_for_search.offset);
356 if (ret < 0)
357 goto out;
358
359 eb = path->nodes[level];
360 while (!eb) {
361 if (WARN_ON(!level)) {
362 ret = 1;
363 goto out;
364 }
365 level--;
366 eb = path->nodes[level];
367 }
368
369 ret = add_all_parents(root, path, parents, ref, level, time_seq,
370 extent_item_pos, total_refs);
371 out:
372 path->lowest_level = 0;
373 btrfs_release_path(path);
374 return ret;
375 }
376
377 /*
378 * resolve all indirect backrefs from the list
379 */
380 static int __resolve_indirect_refs(struct btrfs_fs_info *fs_info,
381 struct btrfs_path *path, u64 time_seq,
382 struct list_head *head,
383 const u64 *extent_item_pos, u64 total_refs,
384 u64 root_objectid)
385 {
386 int err;
387 int ret = 0;
388 struct __prelim_ref *ref;
389 struct __prelim_ref *ref_safe;
390 struct __prelim_ref *new_ref;
391 struct ulist *parents;
392 struct ulist_node *node;
393 struct ulist_iterator uiter;
394
395 parents = ulist_alloc(GFP_NOFS);
396 if (!parents)
397 return -ENOMEM;
398
399 /*
400 * _safe allows us to insert directly after the current item without
401 * iterating over the newly inserted items.
402 * we're also allowed to re-assign ref during iteration.
403 */
404 list_for_each_entry_safe(ref, ref_safe, head, list) {
405 if (ref->parent) /* already direct */
406 continue;
407 if (ref->count == 0)
408 continue;
409 if (root_objectid && ref->root_id != root_objectid) {
410 ret = BACKREF_FOUND_SHARED;
411 goto out;
412 }
413 err = __resolve_indirect_ref(fs_info, path, time_seq, ref,
414 parents, extent_item_pos,
415 total_refs);
416 /*
417 * we can only tolerate ENOENT,otherwise,we should catch error
418 * and return directly.
419 */
420 if (err == -ENOENT) {
421 continue;
422 } else if (err) {
423 ret = err;
424 goto out;
425 }
426
427 /* we put the first parent into the ref at hand */
428 ULIST_ITER_INIT(&uiter);
429 node = ulist_next(parents, &uiter);
430 ref->parent = node ? node->val : 0;
431 ref->inode_list = node ?
432 (struct extent_inode_elem *)(uintptr_t)node->aux : NULL;
433
434 /* additional parents require new refs being added here */
435 while ((node = ulist_next(parents, &uiter))) {
436 new_ref = kmem_cache_alloc(btrfs_prelim_ref_cache,
437 GFP_NOFS);
438 if (!new_ref) {
439 ret = -ENOMEM;
440 goto out;
441 }
442 memcpy(new_ref, ref, sizeof(*ref));
443 new_ref->parent = node->val;
444 new_ref->inode_list = (struct extent_inode_elem *)
445 (uintptr_t)node->aux;
446 list_add(&new_ref->list, &ref->list);
447 }
448 ulist_reinit(parents);
449 }
450 out:
451 ulist_free(parents);
452 return ret;
453 }
454
455 static inline int ref_for_same_block(struct __prelim_ref *ref1,
456 struct __prelim_ref *ref2)
457 {
458 if (ref1->level != ref2->level)
459 return 0;
460 if (ref1->root_id != ref2->root_id)
461 return 0;
462 if (ref1->key_for_search.type != ref2->key_for_search.type)
463 return 0;
464 if (ref1->key_for_search.objectid != ref2->key_for_search.objectid)
465 return 0;
466 if (ref1->key_for_search.offset != ref2->key_for_search.offset)
467 return 0;
468 if (ref1->parent != ref2->parent)
469 return 0;
470
471 return 1;
472 }
473
474 /*
475 * read tree blocks and add keys where required.
476 */
477 static int __add_missing_keys(struct btrfs_fs_info *fs_info,
478 struct list_head *head)
479 {
480 struct list_head *pos;
481 struct extent_buffer *eb;
482
483 list_for_each(pos, head) {
484 struct __prelim_ref *ref;
485 ref = list_entry(pos, struct __prelim_ref, list);
486
487 if (ref->parent)
488 continue;
489 if (ref->key_for_search.type)
490 continue;
491 BUG_ON(!ref->wanted_disk_byte);
492 eb = read_tree_block(fs_info->tree_root, ref->wanted_disk_byte,
493 0);
494 if (!eb || !extent_buffer_uptodate(eb)) {
495 free_extent_buffer(eb);
496 return -EIO;
497 }
498 btrfs_tree_read_lock(eb);
499 if (btrfs_header_level(eb) == 0)
500 btrfs_item_key_to_cpu(eb, &ref->key_for_search, 0);
501 else
502 btrfs_node_key_to_cpu(eb, &ref->key_for_search, 0);
503 btrfs_tree_read_unlock(eb);
504 free_extent_buffer(eb);
505 }
506 return 0;
507 }
508
509 /*
510 * merge two lists of backrefs and adjust counts accordingly
511 *
512 * mode = 1: merge identical keys, if key is set
513 * FIXME: if we add more keys in __add_prelim_ref, we can merge more here.
514 * additionally, we could even add a key range for the blocks we
515 * looked into to merge even more (-> replace unresolved refs by those
516 * having a parent).
517 * mode = 2: merge identical parents
518 */
519 static void __merge_refs(struct list_head *head, int mode)
520 {
521 struct list_head *pos1;
522
523 list_for_each(pos1, head) {
524 struct list_head *n2;
525 struct list_head *pos2;
526 struct __prelim_ref *ref1;
527
528 ref1 = list_entry(pos1, struct __prelim_ref, list);
529
530 for (pos2 = pos1->next, n2 = pos2->next; pos2 != head;
531 pos2 = n2, n2 = pos2->next) {
532 struct __prelim_ref *ref2;
533 struct __prelim_ref *xchg;
534 struct extent_inode_elem *eie;
535
536 ref2 = list_entry(pos2, struct __prelim_ref, list);
537
538 if (mode == 1) {
539 if (!ref_for_same_block(ref1, ref2))
540 continue;
541 if (!ref1->parent && ref2->parent) {
542 xchg = ref1;
543 ref1 = ref2;
544 ref2 = xchg;
545 }
546 } else {
547 if (ref1->parent != ref2->parent)
548 continue;
549 }
550
551 eie = ref1->inode_list;
552 while (eie && eie->next)
553 eie = eie->next;
554 if (eie)
555 eie->next = ref2->inode_list;
556 else
557 ref1->inode_list = ref2->inode_list;
558 ref1->count += ref2->count;
559
560 list_del(&ref2->list);
561 kmem_cache_free(btrfs_prelim_ref_cache, ref2);
562 }
563
564 }
565 }
566
567 /*
568 * add all currently queued delayed refs from this head whose seq nr is
569 * smaller or equal that seq to the list
570 */
571 static int __add_delayed_refs(struct btrfs_delayed_ref_head *head, u64 seq,
572 struct list_head *prefs, u64 *total_refs,
573 u64 inum)
574 {
575 struct btrfs_delayed_extent_op *extent_op = head->extent_op;
576 struct rb_node *n = &head->node.rb_node;
577 struct btrfs_key key;
578 struct btrfs_key op_key = {0};
579 int sgn;
580 int ret = 0;
581
582 if (extent_op && extent_op->update_key)
583 btrfs_disk_key_to_cpu(&op_key, &extent_op->key);
584
585 spin_lock(&head->lock);
586 n = rb_first(&head->ref_root);
587 while (n) {
588 struct btrfs_delayed_ref_node *node;
589 node = rb_entry(n, struct btrfs_delayed_ref_node,
590 rb_node);
591 n = rb_next(n);
592 if (node->seq > seq)
593 continue;
594
595 switch (node->action) {
596 case BTRFS_ADD_DELAYED_EXTENT:
597 case BTRFS_UPDATE_DELAYED_HEAD:
598 WARN_ON(1);
599 continue;
600 case BTRFS_ADD_DELAYED_REF:
601 sgn = 1;
602 break;
603 case BTRFS_DROP_DELAYED_REF:
604 sgn = -1;
605 break;
606 default:
607 BUG_ON(1);
608 }
609 *total_refs += (node->ref_mod * sgn);
610 switch (node->type) {
611 case BTRFS_TREE_BLOCK_REF_KEY: {
612 struct btrfs_delayed_tree_ref *ref;
613
614 ref = btrfs_delayed_node_to_tree_ref(node);
615 ret = __add_prelim_ref(prefs, ref->root, &op_key,
616 ref->level + 1, 0, node->bytenr,
617 node->ref_mod * sgn, GFP_ATOMIC);
618 break;
619 }
620 case BTRFS_SHARED_BLOCK_REF_KEY: {
621 struct btrfs_delayed_tree_ref *ref;
622
623 ref = btrfs_delayed_node_to_tree_ref(node);
624 ret = __add_prelim_ref(prefs, ref->root, NULL,
625 ref->level + 1, ref->parent,
626 node->bytenr,
627 node->ref_mod * sgn, GFP_ATOMIC);
628 break;
629 }
630 case BTRFS_EXTENT_DATA_REF_KEY: {
631 struct btrfs_delayed_data_ref *ref;
632 ref = btrfs_delayed_node_to_data_ref(node);
633
634 key.objectid = ref->objectid;
635 key.type = BTRFS_EXTENT_DATA_KEY;
636 key.offset = ref->offset;
637
638 /*
639 * Found a inum that doesn't match our known inum, we
640 * know it's shared.
641 */
642 if (inum && ref->objectid != inum) {
643 ret = BACKREF_FOUND_SHARED;
644 break;
645 }
646
647 ret = __add_prelim_ref(prefs, ref->root, &key, 0, 0,
648 node->bytenr,
649 node->ref_mod * sgn, GFP_ATOMIC);
650 break;
651 }
652 case BTRFS_SHARED_DATA_REF_KEY: {
653 struct btrfs_delayed_data_ref *ref;
654
655 ref = btrfs_delayed_node_to_data_ref(node);
656
657 key.objectid = ref->objectid;
658 key.type = BTRFS_EXTENT_DATA_KEY;
659 key.offset = ref->offset;
660 ret = __add_prelim_ref(prefs, ref->root, &key, 0,
661 ref->parent, node->bytenr,
662 node->ref_mod * sgn, GFP_ATOMIC);
663 break;
664 }
665 default:
666 WARN_ON(1);
667 }
668 if (ret)
669 break;
670 }
671 spin_unlock(&head->lock);
672 return ret;
673 }
674
675 /*
676 * add all inline backrefs for bytenr to the list
677 */
678 static int __add_inline_refs(struct btrfs_fs_info *fs_info,
679 struct btrfs_path *path, u64 bytenr,
680 int *info_level, struct list_head *prefs,
681 u64 *total_refs, u64 inum)
682 {
683 int ret = 0;
684 int slot;
685 struct extent_buffer *leaf;
686 struct btrfs_key key;
687 struct btrfs_key found_key;
688 unsigned long ptr;
689 unsigned long end;
690 struct btrfs_extent_item *ei;
691 u64 flags;
692 u64 item_size;
693
694 /*
695 * enumerate all inline refs
696 */
697 leaf = path->nodes[0];
698 slot = path->slots[0];
699
700 item_size = btrfs_item_size_nr(leaf, slot);
701 BUG_ON(item_size < sizeof(*ei));
702
703 ei = btrfs_item_ptr(leaf, slot, struct btrfs_extent_item);
704 flags = btrfs_extent_flags(leaf, ei);
705 *total_refs += btrfs_extent_refs(leaf, ei);
706 btrfs_item_key_to_cpu(leaf, &found_key, slot);
707
708 ptr = (unsigned long)(ei + 1);
709 end = (unsigned long)ei + item_size;
710
711 if (found_key.type == BTRFS_EXTENT_ITEM_KEY &&
712 flags & BTRFS_EXTENT_FLAG_TREE_BLOCK) {
713 struct btrfs_tree_block_info *info;
714
715 info = (struct btrfs_tree_block_info *)ptr;
716 *info_level = btrfs_tree_block_level(leaf, info);
717 ptr += sizeof(struct btrfs_tree_block_info);
718 BUG_ON(ptr > end);
719 } else if (found_key.type == BTRFS_METADATA_ITEM_KEY) {
720 *info_level = found_key.offset;
721 } else {
722 BUG_ON(!(flags & BTRFS_EXTENT_FLAG_DATA));
723 }
724
725 while (ptr < end) {
726 struct btrfs_extent_inline_ref *iref;
727 u64 offset;
728 int type;
729
730 iref = (struct btrfs_extent_inline_ref *)ptr;
731 type = btrfs_extent_inline_ref_type(leaf, iref);
732 offset = btrfs_extent_inline_ref_offset(leaf, iref);
733
734 switch (type) {
735 case BTRFS_SHARED_BLOCK_REF_KEY:
736 ret = __add_prelim_ref(prefs, 0, NULL,
737 *info_level + 1, offset,
738 bytenr, 1, GFP_NOFS);
739 break;
740 case BTRFS_SHARED_DATA_REF_KEY: {
741 struct btrfs_shared_data_ref *sdref;
742 int count;
743
744 sdref = (struct btrfs_shared_data_ref *)(iref + 1);
745 count = btrfs_shared_data_ref_count(leaf, sdref);
746 ret = __add_prelim_ref(prefs, 0, NULL, 0, offset,
747 bytenr, count, GFP_NOFS);
748 break;
749 }
750 case BTRFS_TREE_BLOCK_REF_KEY:
751 ret = __add_prelim_ref(prefs, offset, NULL,
752 *info_level + 1, 0,
753 bytenr, 1, GFP_NOFS);
754 break;
755 case BTRFS_EXTENT_DATA_REF_KEY: {
756 struct btrfs_extent_data_ref *dref;
757 int count;
758 u64 root;
759
760 dref = (struct btrfs_extent_data_ref *)(&iref->offset);
761 count = btrfs_extent_data_ref_count(leaf, dref);
762 key.objectid = btrfs_extent_data_ref_objectid(leaf,
763 dref);
764 key.type = BTRFS_EXTENT_DATA_KEY;
765 key.offset = btrfs_extent_data_ref_offset(leaf, dref);
766
767 if (inum && key.objectid != inum) {
768 ret = BACKREF_FOUND_SHARED;
769 break;
770 }
771
772 root = btrfs_extent_data_ref_root(leaf, dref);
773 ret = __add_prelim_ref(prefs, root, &key, 0, 0,
774 bytenr, count, GFP_NOFS);
775 break;
776 }
777 default:
778 WARN_ON(1);
779 }
780 if (ret)
781 return ret;
782 ptr += btrfs_extent_inline_ref_size(type);
783 }
784
785 return 0;
786 }
787
788 /*
789 * add all non-inline backrefs for bytenr to the list
790 */
791 static int __add_keyed_refs(struct btrfs_fs_info *fs_info,
792 struct btrfs_path *path, u64 bytenr,
793 int info_level, struct list_head *prefs, u64 inum)
794 {
795 struct btrfs_root *extent_root = fs_info->extent_root;
796 int ret;
797 int slot;
798 struct extent_buffer *leaf;
799 struct btrfs_key key;
800
801 while (1) {
802 ret = btrfs_next_item(extent_root, path);
803 if (ret < 0)
804 break;
805 if (ret) {
806 ret = 0;
807 break;
808 }
809
810 slot = path->slots[0];
811 leaf = path->nodes[0];
812 btrfs_item_key_to_cpu(leaf, &key, slot);
813
814 if (key.objectid != bytenr)
815 break;
816 if (key.type < BTRFS_TREE_BLOCK_REF_KEY)
817 continue;
818 if (key.type > BTRFS_SHARED_DATA_REF_KEY)
819 break;
820
821 switch (key.type) {
822 case BTRFS_SHARED_BLOCK_REF_KEY:
823 ret = __add_prelim_ref(prefs, 0, NULL,
824 info_level + 1, key.offset,
825 bytenr, 1, GFP_NOFS);
826 break;
827 case BTRFS_SHARED_DATA_REF_KEY: {
828 struct btrfs_shared_data_ref *sdref;
829 int count;
830
831 sdref = btrfs_item_ptr(leaf, slot,
832 struct btrfs_shared_data_ref);
833 count = btrfs_shared_data_ref_count(leaf, sdref);
834 ret = __add_prelim_ref(prefs, 0, NULL, 0, key.offset,
835 bytenr, count, GFP_NOFS);
836 break;
837 }
838 case BTRFS_TREE_BLOCK_REF_KEY:
839 ret = __add_prelim_ref(prefs, key.offset, NULL,
840 info_level + 1, 0,
841 bytenr, 1, GFP_NOFS);
842 break;
843 case BTRFS_EXTENT_DATA_REF_KEY: {
844 struct btrfs_extent_data_ref *dref;
845 int count;
846 u64 root;
847
848 dref = btrfs_item_ptr(leaf, slot,
849 struct btrfs_extent_data_ref);
850 count = btrfs_extent_data_ref_count(leaf, dref);
851 key.objectid = btrfs_extent_data_ref_objectid(leaf,
852 dref);
853 key.type = BTRFS_EXTENT_DATA_KEY;
854 key.offset = btrfs_extent_data_ref_offset(leaf, dref);
855
856 if (inum && key.objectid != inum) {
857 ret = BACKREF_FOUND_SHARED;
858 break;
859 }
860
861 root = btrfs_extent_data_ref_root(leaf, dref);
862 ret = __add_prelim_ref(prefs, root, &key, 0, 0,
863 bytenr, count, GFP_NOFS);
864 break;
865 }
866 default:
867 WARN_ON(1);
868 }
869 if (ret)
870 return ret;
871
872 }
873
874 return ret;
875 }
876
877 /*
878 * this adds all existing backrefs (inline backrefs, backrefs and delayed
879 * refs) for the given bytenr to the refs list, merges duplicates and resolves
880 * indirect refs to their parent bytenr.
881 * When roots are found, they're added to the roots list
882 *
883 * FIXME some caching might speed things up
884 */
885 static int find_parent_nodes(struct btrfs_trans_handle *trans,
886 struct btrfs_fs_info *fs_info, u64 bytenr,
887 u64 time_seq, struct ulist *refs,
888 struct ulist *roots, const u64 *extent_item_pos,
889 u64 root_objectid, u64 inum)
890 {
891 struct btrfs_key key;
892 struct btrfs_path *path;
893 struct btrfs_delayed_ref_root *delayed_refs = NULL;
894 struct btrfs_delayed_ref_head *head;
895 int info_level = 0;
896 int ret;
897 struct list_head prefs_delayed;
898 struct list_head prefs;
899 struct __prelim_ref *ref;
900 struct extent_inode_elem *eie = NULL;
901 u64 total_refs = 0;
902
903 INIT_LIST_HEAD(&prefs);
904 INIT_LIST_HEAD(&prefs_delayed);
905
906 key.objectid = bytenr;
907 key.offset = (u64)-1;
908 if (btrfs_fs_incompat(fs_info, SKINNY_METADATA))
909 key.type = BTRFS_METADATA_ITEM_KEY;
910 else
911 key.type = BTRFS_EXTENT_ITEM_KEY;
912
913 path = btrfs_alloc_path();
914 if (!path)
915 return -ENOMEM;
916 if (!trans) {
917 path->search_commit_root = 1;
918 path->skip_locking = 1;
919 }
920
921 /*
922 * grab both a lock on the path and a lock on the delayed ref head.
923 * We need both to get a consistent picture of how the refs look
924 * at a specified point in time
925 */
926 again:
927 head = NULL;
928
929 ret = btrfs_search_slot(trans, fs_info->extent_root, &key, path, 0, 0);
930 if (ret < 0)
931 goto out;
932 BUG_ON(ret == 0);
933
934 #ifdef CONFIG_BTRFS_FS_RUN_SANITY_TESTS
935 if (trans && likely(trans->type != __TRANS_DUMMY)) {
936 #else
937 if (trans) {
938 #endif
939 /*
940 * look if there are updates for this ref queued and lock the
941 * head
942 */
943 delayed_refs = &trans->transaction->delayed_refs;
944 spin_lock(&delayed_refs->lock);
945 head = btrfs_find_delayed_ref_head(trans, bytenr);
946 if (head) {
947 if (!mutex_trylock(&head->mutex)) {
948 atomic_inc(&head->node.refs);
949 spin_unlock(&delayed_refs->lock);
950
951 btrfs_release_path(path);
952
953 /*
954 * Mutex was contended, block until it's
955 * released and try again
956 */
957 mutex_lock(&head->mutex);
958 mutex_unlock(&head->mutex);
959 btrfs_put_delayed_ref(&head->node);
960 goto again;
961 }
962 spin_unlock(&delayed_refs->lock);
963 ret = __add_delayed_refs(head, time_seq,
964 &prefs_delayed, &total_refs,
965 inum);
966 mutex_unlock(&head->mutex);
967 if (ret)
968 goto out;
969 } else {
970 spin_unlock(&delayed_refs->lock);
971 }
972 }
973
974 if (path->slots[0]) {
975 struct extent_buffer *leaf;
976 int slot;
977
978 path->slots[0]--;
979 leaf = path->nodes[0];
980 slot = path->slots[0];
981 btrfs_item_key_to_cpu(leaf, &key, slot);
982 if (key.objectid == bytenr &&
983 (key.type == BTRFS_EXTENT_ITEM_KEY ||
984 key.type == BTRFS_METADATA_ITEM_KEY)) {
985 ret = __add_inline_refs(fs_info, path, bytenr,
986 &info_level, &prefs,
987 &total_refs, inum);
988 if (ret)
989 goto out;
990 ret = __add_keyed_refs(fs_info, path, bytenr,
991 info_level, &prefs, inum);
992 if (ret)
993 goto out;
994 }
995 }
996 btrfs_release_path(path);
997
998 list_splice_init(&prefs_delayed, &prefs);
999
1000 ret = __add_missing_keys(fs_info, &prefs);
1001 if (ret)
1002 goto out;
1003
1004 __merge_refs(&prefs, 1);
1005
1006 ret = __resolve_indirect_refs(fs_info, path, time_seq, &prefs,
1007 extent_item_pos, total_refs,
1008 root_objectid);
1009 if (ret)
1010 goto out;
1011
1012 __merge_refs(&prefs, 2);
1013
1014 while (!list_empty(&prefs)) {
1015 ref = list_first_entry(&prefs, struct __prelim_ref, list);
1016 WARN_ON(ref->count < 0);
1017 if (roots && ref->count && ref->root_id && ref->parent == 0) {
1018 if (root_objectid && ref->root_id != root_objectid) {
1019 ret = BACKREF_FOUND_SHARED;
1020 goto out;
1021 }
1022
1023 /* no parent == root of tree */
1024 ret = ulist_add(roots, ref->root_id, 0, GFP_NOFS);
1025 if (ret < 0)
1026 goto out;
1027 }
1028 if (ref->count && ref->parent) {
1029 if (extent_item_pos && !ref->inode_list &&
1030 ref->level == 0) {
1031 struct extent_buffer *eb;
1032
1033 eb = read_tree_block(fs_info->extent_root,
1034 ref->parent, 0);
1035 if (!eb || !extent_buffer_uptodate(eb)) {
1036 free_extent_buffer(eb);
1037 ret = -EIO;
1038 goto out;
1039 }
1040 btrfs_tree_read_lock(eb);
1041 btrfs_set_lock_blocking_rw(eb, BTRFS_READ_LOCK);
1042 ret = find_extent_in_eb(eb, bytenr,
1043 *extent_item_pos, &eie);
1044 btrfs_tree_read_unlock_blocking(eb);
1045 free_extent_buffer(eb);
1046 if (ret < 0)
1047 goto out;
1048 ref->inode_list = eie;
1049 }
1050 ret = ulist_add_merge_ptr(refs, ref->parent,
1051 ref->inode_list,
1052 (void **)&eie, GFP_NOFS);
1053 if (ret < 0)
1054 goto out;
1055 if (!ret && extent_item_pos) {
1056 /*
1057 * we've recorded that parent, so we must extend
1058 * its inode list here
1059 */
1060 BUG_ON(!eie);
1061 while (eie->next)
1062 eie = eie->next;
1063 eie->next = ref->inode_list;
1064 }
1065 eie = NULL;
1066 }
1067 list_del(&ref->list);
1068 kmem_cache_free(btrfs_prelim_ref_cache, ref);
1069 }
1070
1071 out:
1072 btrfs_free_path(path);
1073 while (!list_empty(&prefs)) {
1074 ref = list_first_entry(&prefs, struct __prelim_ref, list);
1075 list_del(&ref->list);
1076 kmem_cache_free(btrfs_prelim_ref_cache, ref);
1077 }
1078 while (!list_empty(&prefs_delayed)) {
1079 ref = list_first_entry(&prefs_delayed, struct __prelim_ref,
1080 list);
1081 list_del(&ref->list);
1082 kmem_cache_free(btrfs_prelim_ref_cache, ref);
1083 }
1084 if (ret < 0)
1085 free_inode_elem_list(eie);
1086 return ret;
1087 }
1088
1089 static void free_leaf_list(struct ulist *blocks)
1090 {
1091 struct ulist_node *node = NULL;
1092 struct extent_inode_elem *eie;
1093 struct ulist_iterator uiter;
1094
1095 ULIST_ITER_INIT(&uiter);
1096 while ((node = ulist_next(blocks, &uiter))) {
1097 if (!node->aux)
1098 continue;
1099 eie = (struct extent_inode_elem *)(uintptr_t)node->aux;
1100 free_inode_elem_list(eie);
1101 node->aux = 0;
1102 }
1103
1104 ulist_free(blocks);
1105 }
1106
1107 /*
1108 * Finds all leafs with a reference to the specified combination of bytenr and
1109 * offset. key_list_head will point to a list of corresponding keys (caller must
1110 * free each list element). The leafs will be stored in the leafs ulist, which
1111 * must be freed with ulist_free.
1112 *
1113 * returns 0 on success, <0 on error
1114 */
1115 static int btrfs_find_all_leafs(struct btrfs_trans_handle *trans,
1116 struct btrfs_fs_info *fs_info, u64 bytenr,
1117 u64 time_seq, struct ulist **leafs,
1118 const u64 *extent_item_pos)
1119 {
1120 int ret;
1121
1122 *leafs = ulist_alloc(GFP_NOFS);
1123 if (!*leafs)
1124 return -ENOMEM;
1125
1126 ret = find_parent_nodes(trans, fs_info, bytenr,
1127 time_seq, *leafs, NULL, extent_item_pos, 0, 0);
1128 if (ret < 0 && ret != -ENOENT) {
1129 free_leaf_list(*leafs);
1130 return ret;
1131 }
1132
1133 return 0;
1134 }
1135
1136 /*
1137 * walk all backrefs for a given extent to find all roots that reference this
1138 * extent. Walking a backref means finding all extents that reference this
1139 * extent and in turn walk the backrefs of those, too. Naturally this is a
1140 * recursive process, but here it is implemented in an iterative fashion: We
1141 * find all referencing extents for the extent in question and put them on a
1142 * list. In turn, we find all referencing extents for those, further appending
1143 * to the list. The way we iterate the list allows adding more elements after
1144 * the current while iterating. The process stops when we reach the end of the
1145 * list. Found roots are added to the roots list.
1146 *
1147 * returns 0 on success, < 0 on error.
1148 */
1149 static int __btrfs_find_all_roots(struct btrfs_trans_handle *trans,
1150 struct btrfs_fs_info *fs_info, u64 bytenr,
1151 u64 time_seq, struct ulist **roots)
1152 {
1153 struct ulist *tmp;
1154 struct ulist_node *node = NULL;
1155 struct ulist_iterator uiter;
1156 int ret;
1157
1158 tmp = ulist_alloc(GFP_NOFS);
1159 if (!tmp)
1160 return -ENOMEM;
1161 *roots = ulist_alloc(GFP_NOFS);
1162 if (!*roots) {
1163 ulist_free(tmp);
1164 return -ENOMEM;
1165 }
1166
1167 ULIST_ITER_INIT(&uiter);
1168 while (1) {
1169 ret = find_parent_nodes(trans, fs_info, bytenr,
1170 time_seq, tmp, *roots, NULL, 0, 0);
1171 if (ret < 0 && ret != -ENOENT) {
1172 ulist_free(tmp);
1173 ulist_free(*roots);
1174 return ret;
1175 }
1176 node = ulist_next(tmp, &uiter);
1177 if (!node)
1178 break;
1179 bytenr = node->val;
1180 cond_resched();
1181 }
1182
1183 ulist_free(tmp);
1184 return 0;
1185 }
1186
1187 int btrfs_find_all_roots(struct btrfs_trans_handle *trans,
1188 struct btrfs_fs_info *fs_info, u64 bytenr,
1189 u64 time_seq, struct ulist **roots)
1190 {
1191 int ret;
1192
1193 if (!trans)
1194 down_read(&fs_info->commit_root_sem);
1195 ret = __btrfs_find_all_roots(trans, fs_info, bytenr, time_seq, roots);
1196 if (!trans)
1197 up_read(&fs_info->commit_root_sem);
1198 return ret;
1199 }
1200
1201 int btrfs_check_shared(struct btrfs_trans_handle *trans,
1202 struct btrfs_fs_info *fs_info, u64 root_objectid,
1203 u64 inum, u64 bytenr)
1204 {
1205 struct ulist *tmp = NULL;
1206 struct ulist *roots = NULL;
1207 struct ulist_iterator uiter;
1208 struct ulist_node *node;
1209 struct seq_list elem = {};
1210 int ret = 0;
1211
1212 tmp = ulist_alloc(GFP_NOFS);
1213 roots = ulist_alloc(GFP_NOFS);
1214 if (!tmp || !roots) {
1215 ulist_free(tmp);
1216 ulist_free(roots);
1217 return -ENOMEM;
1218 }
1219
1220 if (trans)
1221 btrfs_get_tree_mod_seq(fs_info, &elem);
1222 else
1223 down_read(&fs_info->commit_root_sem);
1224 ULIST_ITER_INIT(&uiter);
1225 while (1) {
1226 ret = find_parent_nodes(trans, fs_info, bytenr, elem.seq, tmp,
1227 roots, NULL, root_objectid, inum);
1228 if (ret == BACKREF_FOUND_SHARED) {
1229 ret = 1;
1230 break;
1231 }
1232 if (ret < 0 && ret != -ENOENT)
1233 break;
1234 node = ulist_next(tmp, &uiter);
1235 if (!node)
1236 break;
1237 bytenr = node->val;
1238 cond_resched();
1239 }
1240 if (trans)
1241 btrfs_put_tree_mod_seq(fs_info, &elem);
1242 else
1243 up_read(&fs_info->commit_root_sem);
1244 ulist_free(tmp);
1245 ulist_free(roots);
1246 return ret;
1247 }
1248
1249 int btrfs_find_one_extref(struct btrfs_root *root, u64 inode_objectid,
1250 u64 start_off, struct btrfs_path *path,
1251 struct btrfs_inode_extref **ret_extref,
1252 u64 *found_off)
1253 {
1254 int ret, slot;
1255 struct btrfs_key key;
1256 struct btrfs_key found_key;
1257 struct btrfs_inode_extref *extref;
1258 struct extent_buffer *leaf;
1259 unsigned long ptr;
1260
1261 key.objectid = inode_objectid;
1262 key.type = BTRFS_INODE_EXTREF_KEY;
1263 key.offset = start_off;
1264
1265 ret = btrfs_search_slot(NULL, root, &key, path, 0, 0);
1266 if (ret < 0)
1267 return ret;
1268
1269 while (1) {
1270 leaf = path->nodes[0];
1271 slot = path->slots[0];
1272 if (slot >= btrfs_header_nritems(leaf)) {
1273 /*
1274 * If the item at offset is not found,
1275 * btrfs_search_slot will point us to the slot
1276 * where it should be inserted. In our case
1277 * that will be the slot directly before the
1278 * next INODE_REF_KEY_V2 item. In the case
1279 * that we're pointing to the last slot in a
1280 * leaf, we must move one leaf over.
1281 */
1282 ret = btrfs_next_leaf(root, path);
1283 if (ret) {
1284 if (ret >= 1)
1285 ret = -ENOENT;
1286 break;
1287 }
1288 continue;
1289 }
1290
1291 btrfs_item_key_to_cpu(leaf, &found_key, slot);
1292
1293 /*
1294 * Check that we're still looking at an extended ref key for
1295 * this particular objectid. If we have different
1296 * objectid or type then there are no more to be found
1297 * in the tree and we can exit.
1298 */
1299 ret = -ENOENT;
1300 if (found_key.objectid != inode_objectid)
1301 break;
1302 if (found_key.type != BTRFS_INODE_EXTREF_KEY)
1303 break;
1304
1305 ret = 0;
1306 ptr = btrfs_item_ptr_offset(leaf, path->slots[0]);
1307 extref = (struct btrfs_inode_extref *)ptr;
1308 *ret_extref = extref;
1309 if (found_off)
1310 *found_off = found_key.offset;
1311 break;
1312 }
1313
1314 return ret;
1315 }
1316
1317 /*
1318 * this iterates to turn a name (from iref/extref) into a full filesystem path.
1319 * Elements of the path are separated by '/' and the path is guaranteed to be
1320 * 0-terminated. the path is only given within the current file system.
1321 * Therefore, it never starts with a '/'. the caller is responsible to provide
1322 * "size" bytes in "dest". the dest buffer will be filled backwards. finally,
1323 * the start point of the resulting string is returned. this pointer is within
1324 * dest, normally.
1325 * in case the path buffer would overflow, the pointer is decremented further
1326 * as if output was written to the buffer, though no more output is actually
1327 * generated. that way, the caller can determine how much space would be
1328 * required for the path to fit into the buffer. in that case, the returned
1329 * value will be smaller than dest. callers must check this!
1330 */
1331 char *btrfs_ref_to_path(struct btrfs_root *fs_root, struct btrfs_path *path,
1332 u32 name_len, unsigned long name_off,
1333 struct extent_buffer *eb_in, u64 parent,
1334 char *dest, u32 size)
1335 {
1336 int slot;
1337 u64 next_inum;
1338 int ret;
1339 s64 bytes_left = ((s64)size) - 1;
1340 struct extent_buffer *eb = eb_in;
1341 struct btrfs_key found_key;
1342 int leave_spinning = path->leave_spinning;
1343 struct btrfs_inode_ref *iref;
1344
1345 if (bytes_left >= 0)
1346 dest[bytes_left] = '\0';
1347
1348 path->leave_spinning = 1;
1349 while (1) {
1350 bytes_left -= name_len;
1351 if (bytes_left >= 0)
1352 read_extent_buffer(eb, dest + bytes_left,
1353 name_off, name_len);
1354 if (eb != eb_in) {
1355 btrfs_tree_read_unlock_blocking(eb);
1356 free_extent_buffer(eb);
1357 }
1358 ret = btrfs_find_item(fs_root, path, parent, 0,
1359 BTRFS_INODE_REF_KEY, &found_key);
1360 if (ret > 0)
1361 ret = -ENOENT;
1362 if (ret)
1363 break;
1364
1365 next_inum = found_key.offset;
1366
1367 /* regular exit ahead */
1368 if (parent == next_inum)
1369 break;
1370
1371 slot = path->slots[0];
1372 eb = path->nodes[0];
1373 /* make sure we can use eb after releasing the path */
1374 if (eb != eb_in) {
1375 atomic_inc(&eb->refs);
1376 btrfs_tree_read_lock(eb);
1377 btrfs_set_lock_blocking_rw(eb, BTRFS_READ_LOCK);
1378 }
1379 btrfs_release_path(path);
1380 iref = btrfs_item_ptr(eb, slot, struct btrfs_inode_ref);
1381
1382 name_len = btrfs_inode_ref_name_len(eb, iref);
1383 name_off = (unsigned long)(iref + 1);
1384
1385 parent = next_inum;
1386 --bytes_left;
1387 if (bytes_left >= 0)
1388 dest[bytes_left] = '/';
1389 }
1390
1391 btrfs_release_path(path);
1392 path->leave_spinning = leave_spinning;
1393
1394 if (ret)
1395 return ERR_PTR(ret);
1396
1397 return dest + bytes_left;
1398 }
1399
1400 /*
1401 * this makes the path point to (logical EXTENT_ITEM *)
1402 * returns BTRFS_EXTENT_FLAG_DATA for data, BTRFS_EXTENT_FLAG_TREE_BLOCK for
1403 * tree blocks and <0 on error.
1404 */
1405 int extent_from_logical(struct btrfs_fs_info *fs_info, u64 logical,
1406 struct btrfs_path *path, struct btrfs_key *found_key,
1407 u64 *flags_ret)
1408 {
1409 int ret;
1410 u64 flags;
1411 u64 size = 0;
1412 u32 item_size;
1413 struct extent_buffer *eb;
1414 struct btrfs_extent_item *ei;
1415 struct btrfs_key key;
1416
1417 if (btrfs_fs_incompat(fs_info, SKINNY_METADATA))
1418 key.type = BTRFS_METADATA_ITEM_KEY;
1419 else
1420 key.type = BTRFS_EXTENT_ITEM_KEY;
1421 key.objectid = logical;
1422 key.offset = (u64)-1;
1423
1424 ret = btrfs_search_slot(NULL, fs_info->extent_root, &key, path, 0, 0);
1425 if (ret < 0)
1426 return ret;
1427
1428 ret = btrfs_previous_extent_item(fs_info->extent_root, path, 0);
1429 if (ret) {
1430 if (ret > 0)
1431 ret = -ENOENT;
1432 return ret;
1433 }
1434 btrfs_item_key_to_cpu(path->nodes[0], found_key, path->slots[0]);
1435 if (found_key->type == BTRFS_METADATA_ITEM_KEY)
1436 size = fs_info->extent_root->nodesize;
1437 else if (found_key->type == BTRFS_EXTENT_ITEM_KEY)
1438 size = found_key->offset;
1439
1440 if (found_key->objectid > logical ||
1441 found_key->objectid + size <= logical) {
1442 pr_debug("logical %llu is not within any extent\n", logical);
1443 return -ENOENT;
1444 }
1445
1446 eb = path->nodes[0];
1447 item_size = btrfs_item_size_nr(eb, path->slots[0]);
1448 BUG_ON(item_size < sizeof(*ei));
1449
1450 ei = btrfs_item_ptr(eb, path->slots[0], struct btrfs_extent_item);
1451 flags = btrfs_extent_flags(eb, ei);
1452
1453 pr_debug("logical %llu is at position %llu within the extent (%llu "
1454 "EXTENT_ITEM %llu) flags %#llx size %u\n",
1455 logical, logical - found_key->objectid, found_key->objectid,
1456 found_key->offset, flags, item_size);
1457
1458 WARN_ON(!flags_ret);
1459 if (flags_ret) {
1460 if (flags & BTRFS_EXTENT_FLAG_TREE_BLOCK)
1461 *flags_ret = BTRFS_EXTENT_FLAG_TREE_BLOCK;
1462 else if (flags & BTRFS_EXTENT_FLAG_DATA)
1463 *flags_ret = BTRFS_EXTENT_FLAG_DATA;
1464 else
1465 BUG_ON(1);
1466 return 0;
1467 }
1468
1469 return -EIO;
1470 }
1471
1472 /*
1473 * helper function to iterate extent inline refs. ptr must point to a 0 value
1474 * for the first call and may be modified. it is used to track state.
1475 * if more refs exist, 0 is returned and the next call to
1476 * __get_extent_inline_ref must pass the modified ptr parameter to get the
1477 * next ref. after the last ref was processed, 1 is returned.
1478 * returns <0 on error
1479 */
1480 static int __get_extent_inline_ref(unsigned long *ptr, struct extent_buffer *eb,
1481 struct btrfs_key *key,
1482 struct btrfs_extent_item *ei, u32 item_size,
1483 struct btrfs_extent_inline_ref **out_eiref,
1484 int *out_type)
1485 {
1486 unsigned long end;
1487 u64 flags;
1488 struct btrfs_tree_block_info *info;
1489
1490 if (!*ptr) {
1491 /* first call */
1492 flags = btrfs_extent_flags(eb, ei);
1493 if (flags & BTRFS_EXTENT_FLAG_TREE_BLOCK) {
1494 if (key->type == BTRFS_METADATA_ITEM_KEY) {
1495 /* a skinny metadata extent */
1496 *out_eiref =
1497 (struct btrfs_extent_inline_ref *)(ei + 1);
1498 } else {
1499 WARN_ON(key->type != BTRFS_EXTENT_ITEM_KEY);
1500 info = (struct btrfs_tree_block_info *)(ei + 1);
1501 *out_eiref =
1502 (struct btrfs_extent_inline_ref *)(info + 1);
1503 }
1504 } else {
1505 *out_eiref = (struct btrfs_extent_inline_ref *)(ei + 1);
1506 }
1507 *ptr = (unsigned long)*out_eiref;
1508 if ((unsigned long)(*ptr) >= (unsigned long)ei + item_size)
1509 return -ENOENT;
1510 }
1511
1512 end = (unsigned long)ei + item_size;
1513 *out_eiref = (struct btrfs_extent_inline_ref *)(*ptr);
1514 *out_type = btrfs_extent_inline_ref_type(eb, *out_eiref);
1515
1516 *ptr += btrfs_extent_inline_ref_size(*out_type);
1517 WARN_ON(*ptr > end);
1518 if (*ptr == end)
1519 return 1; /* last */
1520
1521 return 0;
1522 }
1523
1524 /*
1525 * reads the tree block backref for an extent. tree level and root are returned
1526 * through out_level and out_root. ptr must point to a 0 value for the first
1527 * call and may be modified (see __get_extent_inline_ref comment).
1528 * returns 0 if data was provided, 1 if there was no more data to provide or
1529 * <0 on error.
1530 */
1531 int tree_backref_for_extent(unsigned long *ptr, struct extent_buffer *eb,
1532 struct btrfs_key *key, struct btrfs_extent_item *ei,
1533 u32 item_size, u64 *out_root, u8 *out_level)
1534 {
1535 int ret;
1536 int type;
1537 struct btrfs_extent_inline_ref *eiref;
1538
1539 if (*ptr == (unsigned long)-1)
1540 return 1;
1541
1542 while (1) {
1543 ret = __get_extent_inline_ref(ptr, eb, key, ei, item_size,
1544 &eiref, &type);
1545 if (ret < 0)
1546 return ret;
1547
1548 if (type == BTRFS_TREE_BLOCK_REF_KEY ||
1549 type == BTRFS_SHARED_BLOCK_REF_KEY)
1550 break;
1551
1552 if (ret == 1)
1553 return 1;
1554 }
1555
1556 /* we can treat both ref types equally here */
1557 *out_root = btrfs_extent_inline_ref_offset(eb, eiref);
1558
1559 if (key->type == BTRFS_EXTENT_ITEM_KEY) {
1560 struct btrfs_tree_block_info *info;
1561
1562 info = (struct btrfs_tree_block_info *)(ei + 1);
1563 *out_level = btrfs_tree_block_level(eb, info);
1564 } else {
1565 ASSERT(key->type == BTRFS_METADATA_ITEM_KEY);
1566 *out_level = (u8)key->offset;
1567 }
1568
1569 if (ret == 1)
1570 *ptr = (unsigned long)-1;
1571
1572 return 0;
1573 }
1574
1575 static int iterate_leaf_refs(struct extent_inode_elem *inode_list,
1576 u64 root, u64 extent_item_objectid,
1577 iterate_extent_inodes_t *iterate, void *ctx)
1578 {
1579 struct extent_inode_elem *eie;
1580 int ret = 0;
1581
1582 for (eie = inode_list; eie; eie = eie->next) {
1583 pr_debug("ref for %llu resolved, key (%llu EXTEND_DATA %llu), "
1584 "root %llu\n", extent_item_objectid,
1585 eie->inum, eie->offset, root);
1586 ret = iterate(eie->inum, eie->offset, root, ctx);
1587 if (ret) {
1588 pr_debug("stopping iteration for %llu due to ret=%d\n",
1589 extent_item_objectid, ret);
1590 break;
1591 }
1592 }
1593
1594 return ret;
1595 }
1596
1597 /*
1598 * calls iterate() for every inode that references the extent identified by
1599 * the given parameters.
1600 * when the iterator function returns a non-zero value, iteration stops.
1601 */
1602 int iterate_extent_inodes(struct btrfs_fs_info *fs_info,
1603 u64 extent_item_objectid, u64 extent_item_pos,
1604 int search_commit_root,
1605 iterate_extent_inodes_t *iterate, void *ctx)
1606 {
1607 int ret;
1608 struct btrfs_trans_handle *trans = NULL;
1609 struct ulist *refs = NULL;
1610 struct ulist *roots = NULL;
1611 struct ulist_node *ref_node = NULL;
1612 struct ulist_node *root_node = NULL;
1613 struct seq_list tree_mod_seq_elem = {};
1614 struct ulist_iterator ref_uiter;
1615 struct ulist_iterator root_uiter;
1616
1617 pr_debug("resolving all inodes for extent %llu\n",
1618 extent_item_objectid);
1619
1620 if (!search_commit_root) {
1621 trans = btrfs_join_transaction(fs_info->extent_root);
1622 if (IS_ERR(trans))
1623 return PTR_ERR(trans);
1624 btrfs_get_tree_mod_seq(fs_info, &tree_mod_seq_elem);
1625 } else {
1626 down_read(&fs_info->commit_root_sem);
1627 }
1628
1629 ret = btrfs_find_all_leafs(trans, fs_info, extent_item_objectid,
1630 tree_mod_seq_elem.seq, &refs,
1631 &extent_item_pos);
1632 if (ret)
1633 goto out;
1634
1635 ULIST_ITER_INIT(&ref_uiter);
1636 while (!ret && (ref_node = ulist_next(refs, &ref_uiter))) {
1637 ret = __btrfs_find_all_roots(trans, fs_info, ref_node->val,
1638 tree_mod_seq_elem.seq, &roots);
1639 if (ret)
1640 break;
1641 ULIST_ITER_INIT(&root_uiter);
1642 while (!ret && (root_node = ulist_next(roots, &root_uiter))) {
1643 pr_debug("root %llu references leaf %llu, data list "
1644 "%#llx\n", root_node->val, ref_node->val,
1645 ref_node->aux);
1646 ret = iterate_leaf_refs((struct extent_inode_elem *)
1647 (uintptr_t)ref_node->aux,
1648 root_node->val,
1649 extent_item_objectid,
1650 iterate, ctx);
1651 }
1652 ulist_free(roots);
1653 }
1654
1655 free_leaf_list(refs);
1656 out:
1657 if (!search_commit_root) {
1658 btrfs_put_tree_mod_seq(fs_info, &tree_mod_seq_elem);
1659 btrfs_end_transaction(trans, fs_info->extent_root);
1660 } else {
1661 up_read(&fs_info->commit_root_sem);
1662 }
1663
1664 return ret;
1665 }
1666
1667 int iterate_inodes_from_logical(u64 logical, struct btrfs_fs_info *fs_info,
1668 struct btrfs_path *path,
1669 iterate_extent_inodes_t *iterate, void *ctx)
1670 {
1671 int ret;
1672 u64 extent_item_pos;
1673 u64 flags = 0;
1674 struct btrfs_key found_key;
1675 int search_commit_root = path->search_commit_root;
1676
1677 ret = extent_from_logical(fs_info, logical, path, &found_key, &flags);
1678 btrfs_release_path(path);
1679 if (ret < 0)
1680 return ret;
1681 if (flags & BTRFS_EXTENT_FLAG_TREE_BLOCK)
1682 return -EINVAL;
1683
1684 extent_item_pos = logical - found_key.objectid;
1685 ret = iterate_extent_inodes(fs_info, found_key.objectid,
1686 extent_item_pos, search_commit_root,
1687 iterate, ctx);
1688
1689 return ret;
1690 }
1691
1692 typedef int (iterate_irefs_t)(u64 parent, u32 name_len, unsigned long name_off,
1693 struct extent_buffer *eb, void *ctx);
1694
1695 static int iterate_inode_refs(u64 inum, struct btrfs_root *fs_root,
1696 struct btrfs_path *path,
1697 iterate_irefs_t *iterate, void *ctx)
1698 {
1699 int ret = 0;
1700 int slot;
1701 u32 cur;
1702 u32 len;
1703 u32 name_len;
1704 u64 parent = 0;
1705 int found = 0;
1706 struct extent_buffer *eb;
1707 struct btrfs_item *item;
1708 struct btrfs_inode_ref *iref;
1709 struct btrfs_key found_key;
1710
1711 while (!ret) {
1712 ret = btrfs_find_item(fs_root, path, inum,
1713 parent ? parent + 1 : 0, BTRFS_INODE_REF_KEY,
1714 &found_key);
1715
1716 if (ret < 0)
1717 break;
1718 if (ret) {
1719 ret = found ? 0 : -ENOENT;
1720 break;
1721 }
1722 ++found;
1723
1724 parent = found_key.offset;
1725 slot = path->slots[0];
1726 eb = btrfs_clone_extent_buffer(path->nodes[0]);
1727 if (!eb) {
1728 ret = -ENOMEM;
1729 break;
1730 }
1731 extent_buffer_get(eb);
1732 btrfs_tree_read_lock(eb);
1733 btrfs_set_lock_blocking_rw(eb, BTRFS_READ_LOCK);
1734 btrfs_release_path(path);
1735
1736 item = btrfs_item_nr(slot);
1737 iref = btrfs_item_ptr(eb, slot, struct btrfs_inode_ref);
1738
1739 for (cur = 0; cur < btrfs_item_size(eb, item); cur += len) {
1740 name_len = btrfs_inode_ref_name_len(eb, iref);
1741 /* path must be released before calling iterate()! */
1742 pr_debug("following ref at offset %u for inode %llu in "
1743 "tree %llu\n", cur, found_key.objectid,
1744 fs_root->objectid);
1745 ret = iterate(parent, name_len,
1746 (unsigned long)(iref + 1), eb, ctx);
1747 if (ret)
1748 break;
1749 len = sizeof(*iref) + name_len;
1750 iref = (struct btrfs_inode_ref *)((char *)iref + len);
1751 }
1752 btrfs_tree_read_unlock_blocking(eb);
1753 free_extent_buffer(eb);
1754 }
1755
1756 btrfs_release_path(path);
1757
1758 return ret;
1759 }
1760
1761 static int iterate_inode_extrefs(u64 inum, struct btrfs_root *fs_root,
1762 struct btrfs_path *path,
1763 iterate_irefs_t *iterate, void *ctx)
1764 {
1765 int ret;
1766 int slot;
1767 u64 offset = 0;
1768 u64 parent;
1769 int found = 0;
1770 struct extent_buffer *eb;
1771 struct btrfs_inode_extref *extref;
1772 struct extent_buffer *leaf;
1773 u32 item_size;
1774 u32 cur_offset;
1775 unsigned long ptr;
1776
1777 while (1) {
1778 ret = btrfs_find_one_extref(fs_root, inum, offset, path, &extref,
1779 &offset);
1780 if (ret < 0)
1781 break;
1782 if (ret) {
1783 ret = found ? 0 : -ENOENT;
1784 break;
1785 }
1786 ++found;
1787
1788 slot = path->slots[0];
1789 eb = btrfs_clone_extent_buffer(path->nodes[0]);
1790 if (!eb) {
1791 ret = -ENOMEM;
1792 break;
1793 }
1794 extent_buffer_get(eb);
1795
1796 btrfs_tree_read_lock(eb);
1797 btrfs_set_lock_blocking_rw(eb, BTRFS_READ_LOCK);
1798 btrfs_release_path(path);
1799
1800 leaf = path->nodes[0];
1801 item_size = btrfs_item_size_nr(leaf, slot);
1802 ptr = btrfs_item_ptr_offset(leaf, slot);
1803 cur_offset = 0;
1804
1805 while (cur_offset < item_size) {
1806 u32 name_len;
1807
1808 extref = (struct btrfs_inode_extref *)(ptr + cur_offset);
1809 parent = btrfs_inode_extref_parent(eb, extref);
1810 name_len = btrfs_inode_extref_name_len(eb, extref);
1811 ret = iterate(parent, name_len,
1812 (unsigned long)&extref->name, eb, ctx);
1813 if (ret)
1814 break;
1815
1816 cur_offset += btrfs_inode_extref_name_len(leaf, extref);
1817 cur_offset += sizeof(*extref);
1818 }
1819 btrfs_tree_read_unlock_blocking(eb);
1820 free_extent_buffer(eb);
1821
1822 offset++;
1823 }
1824
1825 btrfs_release_path(path);
1826
1827 return ret;
1828 }
1829
1830 static int iterate_irefs(u64 inum, struct btrfs_root *fs_root,
1831 struct btrfs_path *path, iterate_irefs_t *iterate,
1832 void *ctx)
1833 {
1834 int ret;
1835 int found_refs = 0;
1836
1837 ret = iterate_inode_refs(inum, fs_root, path, iterate, ctx);
1838 if (!ret)
1839 ++found_refs;
1840 else if (ret != -ENOENT)
1841 return ret;
1842
1843 ret = iterate_inode_extrefs(inum, fs_root, path, iterate, ctx);
1844 if (ret == -ENOENT && found_refs)
1845 return 0;
1846
1847 return ret;
1848 }
1849
1850 /*
1851 * returns 0 if the path could be dumped (probably truncated)
1852 * returns <0 in case of an error
1853 */
1854 static int inode_to_path(u64 inum, u32 name_len, unsigned long name_off,
1855 struct extent_buffer *eb, void *ctx)
1856 {
1857 struct inode_fs_paths *ipath = ctx;
1858 char *fspath;
1859 char *fspath_min;
1860 int i = ipath->fspath->elem_cnt;
1861 const int s_ptr = sizeof(char *);
1862 u32 bytes_left;
1863
1864 bytes_left = ipath->fspath->bytes_left > s_ptr ?
1865 ipath->fspath->bytes_left - s_ptr : 0;
1866
1867 fspath_min = (char *)ipath->fspath->val + (i + 1) * s_ptr;
1868 fspath = btrfs_ref_to_path(ipath->fs_root, ipath->btrfs_path, name_len,
1869 name_off, eb, inum, fspath_min, bytes_left);
1870 if (IS_ERR(fspath))
1871 return PTR_ERR(fspath);
1872
1873 if (fspath > fspath_min) {
1874 ipath->fspath->val[i] = (u64)(unsigned long)fspath;
1875 ++ipath->fspath->elem_cnt;
1876 ipath->fspath->bytes_left = fspath - fspath_min;
1877 } else {
1878 ++ipath->fspath->elem_missed;
1879 ipath->fspath->bytes_missing += fspath_min - fspath;
1880 ipath->fspath->bytes_left = 0;
1881 }
1882
1883 return 0;
1884 }
1885
1886 /*
1887 * this dumps all file system paths to the inode into the ipath struct, provided
1888 * is has been created large enough. each path is zero-terminated and accessed
1889 * from ipath->fspath->val[i].
1890 * when it returns, there are ipath->fspath->elem_cnt number of paths available
1891 * in ipath->fspath->val[]. when the allocated space wasn't sufficient, the
1892 * number of missed paths in recored in ipath->fspath->elem_missed, otherwise,
1893 * it's zero. ipath->fspath->bytes_missing holds the number of bytes that would
1894 * have been needed to return all paths.
1895 */
1896 int paths_from_inode(u64 inum, struct inode_fs_paths *ipath)
1897 {
1898 return iterate_irefs(inum, ipath->fs_root, ipath->btrfs_path,
1899 inode_to_path, ipath);
1900 }
1901
1902 struct btrfs_data_container *init_data_container(u32 total_bytes)
1903 {
1904 struct btrfs_data_container *data;
1905 size_t alloc_bytes;
1906
1907 alloc_bytes = max_t(size_t, total_bytes, sizeof(*data));
1908 data = vmalloc(alloc_bytes);
1909 if (!data)
1910 return ERR_PTR(-ENOMEM);
1911
1912 if (total_bytes >= sizeof(*data)) {
1913 data->bytes_left = total_bytes - sizeof(*data);
1914 data->bytes_missing = 0;
1915 } else {
1916 data->bytes_missing = sizeof(*data) - total_bytes;
1917 data->bytes_left = 0;
1918 }
1919
1920 data->elem_cnt = 0;
1921 data->elem_missed = 0;
1922
1923 return data;
1924 }
1925
1926 /*
1927 * allocates space to return multiple file system paths for an inode.
1928 * total_bytes to allocate are passed, note that space usable for actual path
1929 * information will be total_bytes - sizeof(struct inode_fs_paths).
1930 * the returned pointer must be freed with free_ipath() in the end.
1931 */
1932 struct inode_fs_paths *init_ipath(s32 total_bytes, struct btrfs_root *fs_root,
1933 struct btrfs_path *path)
1934 {
1935 struct inode_fs_paths *ifp;
1936 struct btrfs_data_container *fspath;
1937
1938 fspath = init_data_container(total_bytes);
1939 if (IS_ERR(fspath))
1940 return (void *)fspath;
1941
1942 ifp = kmalloc(sizeof(*ifp), GFP_NOFS);
1943 if (!ifp) {
1944 kfree(fspath);
1945 return ERR_PTR(-ENOMEM);
1946 }
1947
1948 ifp->btrfs_path = path;
1949 ifp->fspath = fspath;
1950 ifp->fs_root = fs_root;
1951
1952 return ifp;
1953 }
1954
1955 void free_ipath(struct inode_fs_paths *ipath)
1956 {
1957 if (!ipath)
1958 return;
1959 vfree(ipath->fspath);
1960 kfree(ipath);
1961 }
Linux with added FPC-III support