?  search:
summary | log | graphiclog1 | graphiclog2 | commit | commitdiff | tree | refs | edit | fork
blame | history | raw | HEAD
dmaengine: imx-dma: fix slow path issue in prep_dma_cyclic
[linux/fpc-iii.git] / fs / btrfs / backref.c
blob8bc5e8ccb091852dd53f262eb556fe5578a419a3
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 struct extent_inode_elem {
29 u64 inum;
30 u64 offset;
31 struct extent_inode_elem *next;
32 };
33
34 static int check_extent_in_eb(struct btrfs_key *key, struct extent_buffer *eb,
35 struct btrfs_file_extent_item *fi,
36 u64 extent_item_pos,
37 struct extent_inode_elem **eie)
38 {
39 u64 offset = 0;
40 struct extent_inode_elem *e;
41
42 if (!btrfs_file_extent_compression(eb, fi) &&
43 !btrfs_file_extent_encryption(eb, fi) &&
44 !btrfs_file_extent_other_encoding(eb, fi)) {
45 u64 data_offset;
46 u64 data_len;
47
48 data_offset = btrfs_file_extent_offset(eb, fi);
49 data_len = btrfs_file_extent_num_bytes(eb, fi);
50
51 if (extent_item_pos < data_offset ||
52 extent_item_pos >= data_offset + data_len)
53 return 1;
54 offset = extent_item_pos - data_offset;
55 }
56
57 e = kmalloc(sizeof(*e), GFP_NOFS);
58 if (!e)
59 return -ENOMEM;
60
61 e->next = *eie;
62 e->inum = key->objectid;
63 e->offset = key->offset + offset;
64 *eie = e;
65
66 return 0;
67 }
68
69 static int find_extent_in_eb(struct extent_buffer *eb, u64 wanted_disk_byte,
70 u64 extent_item_pos,
71 struct extent_inode_elem **eie)
72 {
73 u64 disk_byte;
74 struct btrfs_key key;
75 struct btrfs_file_extent_item *fi;
76 int slot;
77 int nritems;
78 int extent_type;
79 int ret;
80
81 /*
82 * from the shared data ref, we only have the leaf but we need
83 * the key. thus, we must look into all items and see that we
84 * find one (some) with a reference to our extent item.
85 */
86 nritems = btrfs_header_nritems(eb);
87 for (slot = 0; slot < nritems; ++slot) {
88 btrfs_item_key_to_cpu(eb, &key, slot);
89 if (key.type != BTRFS_EXTENT_DATA_KEY)
90 continue;
91 fi = btrfs_item_ptr(eb, slot, struct btrfs_file_extent_item);
92 extent_type = btrfs_file_extent_type(eb, fi);
93 if (extent_type == BTRFS_FILE_EXTENT_INLINE)
94 continue;
95 /* don't skip BTRFS_FILE_EXTENT_PREALLOC, we can handle that */
96 disk_byte = btrfs_file_extent_disk_bytenr(eb, fi);
97 if (disk_byte != wanted_disk_byte)
98 continue;
99
100 ret = check_extent_in_eb(&key, eb, fi, extent_item_pos, eie);
101 if (ret < 0)
102 return ret;
103 }
104
105 return 0;
106 }
107
108 /*
109 * this structure records all encountered refs on the way up to the root
110 */
111 struct __prelim_ref {
112 struct list_head list;
113 u64 root_id;
114 struct btrfs_key key_for_search;
115 int level;
116 int count;
117 struct extent_inode_elem *inode_list;
118 u64 parent;
119 u64 wanted_disk_byte;
120 };
121
122 /*
123 * the rules for all callers of this function are:
124 * - obtaining the parent is the goal
125 * - if you add a key, you must know that it is a correct key
126 * - if you cannot add the parent or a correct key, then we will look into the
127 * block later to set a correct key
128 *
129 * delayed refs
130 * ============
131 * backref type | shared | indirect | shared | indirect
132 * information | tree | tree | data | data
133 * --------------------+--------+----------+--------+----------
134 * parent logical | y | - | - | -
135 * key to resolve | - | y | y | y
136 * tree block logical | - | - | - | -
137 * root for resolving | y | y | y | y
138 *
139 * - column 1: we've the parent -> done
140 * - column 2, 3, 4: we use the key to find the parent
141 *
142 * on disk refs (inline or keyed)
143 * ==============================
144 * backref type | shared | indirect | shared | indirect
145 * information | tree | tree | data | data
146 * --------------------+--------+----------+--------+----------
147 * parent logical | y | - | y | -
148 * key to resolve | - | - | - | y
149 * tree block logical | y | y | y | y
150 * root for resolving | - | y | y | y
151 *
152 * - column 1, 3: we've the parent -> done
153 * - column 2: we take the first key from the block to find the parent
154 * (see __add_missing_keys)
155 * - column 4: we use the key to find the parent
156 *
157 * additional information that's available but not required to find the parent
158 * block might help in merging entries to gain some speed.
159 */
160
161 static int __add_prelim_ref(struct list_head *head, u64 root_id,
162 struct btrfs_key *key, int level,
163 u64 parent, u64 wanted_disk_byte, int count)
164 {
165 struct __prelim_ref *ref;
166
167 /* in case we're adding delayed refs, we're holding the refs spinlock */
168 ref = kmalloc(sizeof(*ref), GFP_ATOMIC);
169 if (!ref)
170 return -ENOMEM;
171
172 ref->root_id = root_id;
173 if (key)
174 ref->key_for_search = *key;
175 else
176 memset(&ref->key_for_search, 0, sizeof(ref->key_for_search));
177
178 ref->inode_list = NULL;
179 ref->level = level;
180 ref->count = count;
181 ref->parent = parent;
182 ref->wanted_disk_byte = wanted_disk_byte;
183 list_add_tail(&ref->list, head);
184
185 return 0;
186 }
187
188 static int add_all_parents(struct btrfs_root *root, struct btrfs_path *path,
189 struct ulist *parents, int level,
190 struct btrfs_key *key_for_search, u64 time_seq,
191 u64 wanted_disk_byte,
192 const u64 *extent_item_pos)
193 {
194 int ret = 0;
195 int slot;
196 struct extent_buffer *eb;
197 struct btrfs_key key;
198 struct btrfs_file_extent_item *fi;
199 struct extent_inode_elem *eie = NULL, *old = NULL;
200 u64 disk_byte;
201
202 if (level != 0) {
203 eb = path->nodes[level];
204 ret = ulist_add(parents, eb->start, 0, GFP_NOFS);
205 if (ret < 0)
206 return ret;
207 return 0;
208 }
209
210 /*
211 * We normally enter this function with the path already pointing to
212 * the first item to check. But sometimes, we may enter it with
213 * slot==nritems. In that case, go to the next leaf before we continue.
214 */
215 if (path->slots[0] >= btrfs_header_nritems(path->nodes[0]))
216 ret = btrfs_next_old_leaf(root, path, time_seq);
217
218 while (!ret) {
219 eb = path->nodes[0];
220 slot = path->slots[0];
221
222 btrfs_item_key_to_cpu(eb, &key, slot);
223
224 if (key.objectid != key_for_search->objectid ||
225 key.type != BTRFS_EXTENT_DATA_KEY)
226 break;
227
228 fi = btrfs_item_ptr(eb, slot, struct btrfs_file_extent_item);
229 disk_byte = btrfs_file_extent_disk_bytenr(eb, fi);
230
231 if (disk_byte == wanted_disk_byte) {
232 eie = NULL;
233 old = NULL;
234 if (extent_item_pos) {
235 ret = check_extent_in_eb(&key, eb, fi,
236 *extent_item_pos,
237 &eie);
238 if (ret < 0)
239 break;
240 }
241 if (ret > 0)
242 goto next;
243 ret = ulist_add_merge(parents, eb->start,
244 (uintptr_t)eie,
245 (u64 *)&old, GFP_NOFS);
246 if (ret < 0)
247 break;
248 if (!ret && extent_item_pos) {
249 while (old->next)
250 old = old->next;
251 old->next = eie;
252 }
253 }
254 next:
255 ret = btrfs_next_old_item(root, path, time_seq);
256 }
257
258 if (ret > 0)
259 ret = 0;
260 return ret;
261 }
262
263 /*
264 * resolve an indirect backref in the form (root_id, key, level)
265 * to a logical address
266 */
267 static int __resolve_indirect_ref(struct btrfs_fs_info *fs_info,
268 struct btrfs_path *path, u64 time_seq,
269 struct __prelim_ref *ref,
270 struct ulist *parents,
271 const u64 *extent_item_pos)
272 {
273 struct btrfs_root *root;
274 struct btrfs_key root_key;
275 struct extent_buffer *eb;
276 int ret = 0;
277 int root_level;
278 int level = ref->level;
279
280 root_key.objectid = ref->root_id;
281 root_key.type = BTRFS_ROOT_ITEM_KEY;
282 root_key.offset = (u64)-1;
283 root = btrfs_read_fs_root_no_name(fs_info, &root_key);
284 if (IS_ERR(root)) {
285 ret = PTR_ERR(root);
286 goto out;
287 }
288
289 root_level = btrfs_old_root_level(root, time_seq);
290
291 if (root_level + 1 == level)
292 goto out;
293
294 path->lowest_level = level;
295 ret = btrfs_search_old_slot(root, &ref->key_for_search, path, time_seq);
296 pr_debug("search slot in root %llu (level %d, ref count %d) returned "
297 "%d for key (%llu %u %llu)\n",
298 (unsigned long long)ref->root_id, level, ref->count, ret,
299 (unsigned long long)ref->key_for_search.objectid,
300 ref->key_for_search.type,
301 (unsigned long long)ref->key_for_search.offset);
302 if (ret < 0)
303 goto out;
304
305 eb = path->nodes[level];
306 while (!eb) {
307 if (!level) {
308 WARN_ON(1);
309 ret = 1;
310 goto out;
311 }
312 level--;
313 eb = path->nodes[level];
314 }
315
316 ret = add_all_parents(root, path, parents, level, &ref->key_for_search,
317 time_seq, ref->wanted_disk_byte,
318 extent_item_pos);
319 out:
320 path->lowest_level = 0;
321 btrfs_release_path(path);
322 return ret;
323 }
324
325 /*
326 * resolve all indirect backrefs from the list
327 */
328 static int __resolve_indirect_refs(struct btrfs_fs_info *fs_info,
329 struct btrfs_path *path, u64 time_seq,
330 struct list_head *head,
331 const u64 *extent_item_pos)
332 {
333 int err;
334 int ret = 0;
335 struct __prelim_ref *ref;
336 struct __prelim_ref *ref_safe;
337 struct __prelim_ref *new_ref;
338 struct ulist *parents;
339 struct ulist_node *node;
340 struct ulist_iterator uiter;
341
342 parents = ulist_alloc(GFP_NOFS);
343 if (!parents)
344 return -ENOMEM;
345
346 /*
347 * _safe allows us to insert directly after the current item without
348 * iterating over the newly inserted items.
349 * we're also allowed to re-assign ref during iteration.
350 */
351 list_for_each_entry_safe(ref, ref_safe, head, list) {
352 if (ref->parent) /* already direct */
353 continue;
354 if (ref->count == 0)
355 continue;
356 err = __resolve_indirect_ref(fs_info, path, time_seq, ref,
357 parents, extent_item_pos);
358 if (err == -ENOMEM)
359 goto out;
360 if (err)
361 continue;
362
363 /* we put the first parent into the ref at hand */
364 ULIST_ITER_INIT(&uiter);
365 node = ulist_next(parents, &uiter);
366 ref->parent = node ? node->val : 0;
367 ref->inode_list = node ?
368 (struct extent_inode_elem *)(uintptr_t)node->aux : 0;
369
370 /* additional parents require new refs being added here */
371 while ((node = ulist_next(parents, &uiter))) {
372 new_ref = kmalloc(sizeof(*new_ref), GFP_NOFS);
373 if (!new_ref) {
374 ret = -ENOMEM;
375 goto out;
376 }
377 memcpy(new_ref, ref, sizeof(*ref));
378 new_ref->parent = node->val;
379 new_ref->inode_list = (struct extent_inode_elem *)
380 (uintptr_t)node->aux;
381 list_add(&new_ref->list, &ref->list);
382 }
383 ulist_reinit(parents);
384 }
385 out:
386 ulist_free(parents);
387 return ret;
388 }
389
390 static inline int ref_for_same_block(struct __prelim_ref *ref1,
391 struct __prelim_ref *ref2)
392 {
393 if (ref1->level != ref2->level)
394 return 0;
395 if (ref1->root_id != ref2->root_id)
396 return 0;
397 if (ref1->key_for_search.type != ref2->key_for_search.type)
398 return 0;
399 if (ref1->key_for_search.objectid != ref2->key_for_search.objectid)
400 return 0;
401 if (ref1->key_for_search.offset != ref2->key_for_search.offset)
402 return 0;
403 if (ref1->parent != ref2->parent)
404 return 0;
405
406 return 1;
407 }
408
409 /*
410 * read tree blocks and add keys where required.
411 */
412 static int __add_missing_keys(struct btrfs_fs_info *fs_info,
413 struct list_head *head)
414 {
415 struct list_head *pos;
416 struct extent_buffer *eb;
417
418 list_for_each(pos, head) {
419 struct __prelim_ref *ref;
420 ref = list_entry(pos, struct __prelim_ref, list);
421
422 if (ref->parent)
423 continue;
424 if (ref->key_for_search.type)
425 continue;
426 BUG_ON(!ref->wanted_disk_byte);
427 eb = read_tree_block(fs_info->tree_root, ref->wanted_disk_byte,
428 fs_info->tree_root->leafsize, 0);
429 if (!eb || !extent_buffer_uptodate(eb)) {
430 free_extent_buffer(eb);
431 return -EIO;
432 }
433 btrfs_tree_read_lock(eb);
434 if (btrfs_header_level(eb) == 0)
435 btrfs_item_key_to_cpu(eb, &ref->key_for_search, 0);
436 else
437 btrfs_node_key_to_cpu(eb, &ref->key_for_search, 0);
438 btrfs_tree_read_unlock(eb);
439 free_extent_buffer(eb);
440 }
441 return 0;
442 }
443
444 /*
445 * merge two lists of backrefs and adjust counts accordingly
446 *
447 * mode = 1: merge identical keys, if key is set
448 * FIXME: if we add more keys in __add_prelim_ref, we can merge more here.
449 * additionally, we could even add a key range for the blocks we
450 * looked into to merge even more (-> replace unresolved refs by those
451 * having a parent).
452 * mode = 2: merge identical parents
453 */
454 static void __merge_refs(struct list_head *head, int mode)
455 {
456 struct list_head *pos1;
457
458 list_for_each(pos1, head) {
459 struct list_head *n2;
460 struct list_head *pos2;
461 struct __prelim_ref *ref1;
462
463 ref1 = list_entry(pos1, struct __prelim_ref, list);
464
465 for (pos2 = pos1->next, n2 = pos2->next; pos2 != head;
466 pos2 = n2, n2 = pos2->next) {
467 struct __prelim_ref *ref2;
468 struct __prelim_ref *xchg;
469 struct extent_inode_elem *eie;
470
471 ref2 = list_entry(pos2, struct __prelim_ref, list);
472
473 if (mode == 1) {
474 if (!ref_for_same_block(ref1, ref2))
475 continue;
476 if (!ref1->parent && ref2->parent) {
477 xchg = ref1;
478 ref1 = ref2;
479 ref2 = xchg;
480 }
481 } else {
482 if (ref1->parent != ref2->parent)
483 continue;
484 }
485
486 eie = ref1->inode_list;
487 while (eie && eie->next)
488 eie = eie->next;
489 if (eie)
490 eie->next = ref2->inode_list;
491 else
492 ref1->inode_list = ref2->inode_list;
493 ref1->count += ref2->count;
494
495 list_del(&ref2->list);
496 kfree(ref2);
497 }
498
499 }
500 }
501
502 /*
503 * add all currently queued delayed refs from this head whose seq nr is
504 * smaller or equal that seq to the list
505 */
506 static int __add_delayed_refs(struct btrfs_delayed_ref_head *head, u64 seq,
507 struct list_head *prefs)
508 {
509 struct btrfs_delayed_extent_op *extent_op = head->extent_op;
510 struct rb_node *n = &head->node.rb_node;
511 struct btrfs_key key;
512 struct btrfs_key op_key = {0};
513 int sgn;
514 int ret = 0;
515
516 if (extent_op && extent_op->update_key)
517 btrfs_disk_key_to_cpu(&op_key, &extent_op->key);
518
519 while ((n = rb_prev(n))) {
520 struct btrfs_delayed_ref_node *node;
521 node = rb_entry(n, struct btrfs_delayed_ref_node,
522 rb_node);
523 if (node->bytenr != head->node.bytenr)
524 break;
525 WARN_ON(node->is_head);
526
527 if (node->seq > seq)
528 continue;
529
530 switch (node->action) {
531 case BTRFS_ADD_DELAYED_EXTENT:
532 case BTRFS_UPDATE_DELAYED_HEAD:
533 WARN_ON(1);
534 continue;
535 case BTRFS_ADD_DELAYED_REF:
536 sgn = 1;
537 break;
538 case BTRFS_DROP_DELAYED_REF:
539 sgn = -1;
540 break;
541 default:
542 BUG_ON(1);
543 }
544 switch (node->type) {
545 case BTRFS_TREE_BLOCK_REF_KEY: {
546 struct btrfs_delayed_tree_ref *ref;
547
548 ref = btrfs_delayed_node_to_tree_ref(node);
549 ret = __add_prelim_ref(prefs, ref->root, &op_key,
550 ref->level + 1, 0, node->bytenr,
551 node->ref_mod * sgn);
552 break;
553 }
554 case BTRFS_SHARED_BLOCK_REF_KEY: {
555 struct btrfs_delayed_tree_ref *ref;
556
557 ref = btrfs_delayed_node_to_tree_ref(node);
558 ret = __add_prelim_ref(prefs, ref->root, NULL,
559 ref->level + 1, ref->parent,
560 node->bytenr,
561 node->ref_mod * sgn);
562 break;
563 }
564 case BTRFS_EXTENT_DATA_REF_KEY: {
565 struct btrfs_delayed_data_ref *ref;
566 ref = btrfs_delayed_node_to_data_ref(node);
567
568 key.objectid = ref->objectid;
569 key.type = BTRFS_EXTENT_DATA_KEY;
570 key.offset = ref->offset;
571 ret = __add_prelim_ref(prefs, ref->root, &key, 0, 0,
572 node->bytenr,
573 node->ref_mod * sgn);
574 break;
575 }
576 case BTRFS_SHARED_DATA_REF_KEY: {
577 struct btrfs_delayed_data_ref *ref;
578
579 ref = btrfs_delayed_node_to_data_ref(node);
580
581 key.objectid = ref->objectid;
582 key.type = BTRFS_EXTENT_DATA_KEY;
583 key.offset = ref->offset;
584 ret = __add_prelim_ref(prefs, ref->root, &key, 0,
585 ref->parent, node->bytenr,
586 node->ref_mod * sgn);
587 break;
588 }
589 default:
590 WARN_ON(1);
591 }
592 if (ret)
593 return ret;
594 }
595
596 return 0;
597 }
598
599 /*
600 * add all inline backrefs for bytenr to the list
601 */
602 static int __add_inline_refs(struct btrfs_fs_info *fs_info,
603 struct btrfs_path *path, u64 bytenr,
604 int *info_level, struct list_head *prefs)
605 {
606 int ret = 0;
607 int slot;
608 struct extent_buffer *leaf;
609 struct btrfs_key key;
610 struct btrfs_key found_key;
611 unsigned long ptr;
612 unsigned long end;
613 struct btrfs_extent_item *ei;
614 u64 flags;
615 u64 item_size;
616
617 /*
618 * enumerate all inline refs
619 */
620 leaf = path->nodes[0];
621 slot = path->slots[0];
622
623 item_size = btrfs_item_size_nr(leaf, slot);
624 BUG_ON(item_size < sizeof(*ei));
625
626 ei = btrfs_item_ptr(leaf, slot, struct btrfs_extent_item);
627 flags = btrfs_extent_flags(leaf, ei);
628 btrfs_item_key_to_cpu(leaf, &found_key, slot);
629
630 ptr = (unsigned long)(ei + 1);
631 end = (unsigned long)ei + item_size;
632
633 if (found_key.type == BTRFS_EXTENT_ITEM_KEY &&
634 flags & BTRFS_EXTENT_FLAG_TREE_BLOCK) {
635 struct btrfs_tree_block_info *info;
636
637 info = (struct btrfs_tree_block_info *)ptr;
638 *info_level = btrfs_tree_block_level(leaf, info);
639 ptr += sizeof(struct btrfs_tree_block_info);
640 BUG_ON(ptr > end);
641 } else if (found_key.type == BTRFS_METADATA_ITEM_KEY) {
642 *info_level = found_key.offset;
643 } else {
644 BUG_ON(!(flags & BTRFS_EXTENT_FLAG_DATA));
645 }
646
647 while (ptr < end) {
648 struct btrfs_extent_inline_ref *iref;
649 u64 offset;
650 int type;
651
652 iref = (struct btrfs_extent_inline_ref *)ptr;
653 type = btrfs_extent_inline_ref_type(leaf, iref);
654 offset = btrfs_extent_inline_ref_offset(leaf, iref);
655
656 switch (type) {
657 case BTRFS_SHARED_BLOCK_REF_KEY:
658 ret = __add_prelim_ref(prefs, 0, NULL,
659 *info_level + 1, offset,
660 bytenr, 1);
661 break;
662 case BTRFS_SHARED_DATA_REF_KEY: {
663 struct btrfs_shared_data_ref *sdref;
664 int count;
665
666 sdref = (struct btrfs_shared_data_ref *)(iref + 1);
667 count = btrfs_shared_data_ref_count(leaf, sdref);
668 ret = __add_prelim_ref(prefs, 0, NULL, 0, offset,
669 bytenr, count);
670 break;
671 }
672 case BTRFS_TREE_BLOCK_REF_KEY:
673 ret = __add_prelim_ref(prefs, offset, NULL,
674 *info_level + 1, 0,
675 bytenr, 1);
676 break;
677 case BTRFS_EXTENT_DATA_REF_KEY: {
678 struct btrfs_extent_data_ref *dref;
679 int count;
680 u64 root;
681
682 dref = (struct btrfs_extent_data_ref *)(&iref->offset);
683 count = btrfs_extent_data_ref_count(leaf, dref);
684 key.objectid = btrfs_extent_data_ref_objectid(leaf,
685 dref);
686 key.type = BTRFS_EXTENT_DATA_KEY;
687 key.offset = btrfs_extent_data_ref_offset(leaf, dref);
688 root = btrfs_extent_data_ref_root(leaf, dref);
689 ret = __add_prelim_ref(prefs, root, &key, 0, 0,
690 bytenr, count);
691 break;
692 }
693 default:
694 WARN_ON(1);
695 }
696 if (ret)
697 return ret;
698 ptr += btrfs_extent_inline_ref_size(type);
699 }
700
701 return 0;
702 }
703
704 /*
705 * add all non-inline backrefs for bytenr to the list
706 */
707 static int __add_keyed_refs(struct btrfs_fs_info *fs_info,
708 struct btrfs_path *path, u64 bytenr,
709 int info_level, struct list_head *prefs)
710 {
711 struct btrfs_root *extent_root = fs_info->extent_root;
712 int ret;
713 int slot;
714 struct extent_buffer *leaf;
715 struct btrfs_key key;
716
717 while (1) {
718 ret = btrfs_next_item(extent_root, path);
719 if (ret < 0)
720 break;
721 if (ret) {
722 ret = 0;
723 break;
724 }
725
726 slot = path->slots[0];
727 leaf = path->nodes[0];
728 btrfs_item_key_to_cpu(leaf, &key, slot);
729
730 if (key.objectid != bytenr)
731 break;
732 if (key.type < BTRFS_TREE_BLOCK_REF_KEY)
733 continue;
734 if (key.type > BTRFS_SHARED_DATA_REF_KEY)
735 break;
736
737 switch (key.type) {
738 case BTRFS_SHARED_BLOCK_REF_KEY:
739 ret = __add_prelim_ref(prefs, 0, NULL,
740 info_level + 1, key.offset,
741 bytenr, 1);
742 break;
743 case BTRFS_SHARED_DATA_REF_KEY: {
744 struct btrfs_shared_data_ref *sdref;
745 int count;
746
747 sdref = btrfs_item_ptr(leaf, slot,
748 struct btrfs_shared_data_ref);
749 count = btrfs_shared_data_ref_count(leaf, sdref);
750 ret = __add_prelim_ref(prefs, 0, NULL, 0, key.offset,
751 bytenr, count);
752 break;
753 }
754 case BTRFS_TREE_BLOCK_REF_KEY:
755 ret = __add_prelim_ref(prefs, key.offset, NULL,
756 info_level + 1, 0,
757 bytenr, 1);
758 break;
759 case BTRFS_EXTENT_DATA_REF_KEY: {
760 struct btrfs_extent_data_ref *dref;
761 int count;
762 u64 root;
763
764 dref = btrfs_item_ptr(leaf, slot,
765 struct btrfs_extent_data_ref);
766 count = btrfs_extent_data_ref_count(leaf, dref);
767 key.objectid = btrfs_extent_data_ref_objectid(leaf,
768 dref);
769 key.type = BTRFS_EXTENT_DATA_KEY;
770 key.offset = btrfs_extent_data_ref_offset(leaf, dref);
771 root = btrfs_extent_data_ref_root(leaf, dref);
772 ret = __add_prelim_ref(prefs, root, &key, 0, 0,
773 bytenr, count);
774 break;
775 }
776 default:
777 WARN_ON(1);
778 }
779 if (ret)
780 return ret;
781
782 }
783
784 return ret;
785 }
786
787 /*
788 * this adds all existing backrefs (inline backrefs, backrefs and delayed
789 * refs) for the given bytenr to the refs list, merges duplicates and resolves
790 * indirect refs to their parent bytenr.
791 * When roots are found, they're added to the roots list
792 *
793 * FIXME some caching might speed things up
794 */
795 static int find_parent_nodes(struct btrfs_trans_handle *trans,
796 struct btrfs_fs_info *fs_info, u64 bytenr,
797 u64 time_seq, struct ulist *refs,
798 struct ulist *roots, const u64 *extent_item_pos)
799 {
800 struct btrfs_key key;
801 struct btrfs_path *path;
802 struct btrfs_delayed_ref_root *delayed_refs = NULL;
803 struct btrfs_delayed_ref_head *head;
804 int info_level = 0;
805 int ret;
806 struct list_head prefs_delayed;
807 struct list_head prefs;
808 struct __prelim_ref *ref;
809
810 INIT_LIST_HEAD(&prefs);
811 INIT_LIST_HEAD(&prefs_delayed);
812
813 key.objectid = bytenr;
814 key.offset = (u64)-1;
815 if (btrfs_fs_incompat(fs_info, SKINNY_METADATA))
816 key.type = BTRFS_METADATA_ITEM_KEY;
817 else
818 key.type = BTRFS_EXTENT_ITEM_KEY;
819
820 path = btrfs_alloc_path();
821 if (!path)
822 return -ENOMEM;
823 if (!trans)
824 path->search_commit_root = 1;
825
826 /*
827 * grab both a lock on the path and a lock on the delayed ref head.
828 * We need both to get a consistent picture of how the refs look
829 * at a specified point in time
830 */
831 again:
832 head = NULL;
833
834 ret = btrfs_search_slot(trans, fs_info->extent_root, &key, path, 0, 0);
835 if (ret < 0)
836 goto out;
837 BUG_ON(ret == 0);
838
839 if (trans) {
840 /*
841 * look if there are updates for this ref queued and lock the
842 * head
843 */
844 delayed_refs = &trans->transaction->delayed_refs;
845 spin_lock(&delayed_refs->lock);
846 head = btrfs_find_delayed_ref_head(trans, bytenr);
847 if (head) {
848 if (!mutex_trylock(&head->mutex)) {
849 atomic_inc(&head->node.refs);
850 spin_unlock(&delayed_refs->lock);
851
852 btrfs_release_path(path);
853
854 /*
855 * Mutex was contended, block until it's
856 * released and try again
857 */
858 mutex_lock(&head->mutex);
859 mutex_unlock(&head->mutex);
860 btrfs_put_delayed_ref(&head->node);
861 goto again;
862 }
863 ret = __add_delayed_refs(head, time_seq,
864 &prefs_delayed);
865 mutex_unlock(&head->mutex);
866 if (ret) {
867 spin_unlock(&delayed_refs->lock);
868 goto out;
869 }
870 }
871 spin_unlock(&delayed_refs->lock);
872 }
873
874 if (path->slots[0]) {
875 struct extent_buffer *leaf;
876 int slot;
877
878 path->slots[0]--;
879 leaf = path->nodes[0];
880 slot = path->slots[0];
881 btrfs_item_key_to_cpu(leaf, &key, slot);
882 if (key.objectid == bytenr &&
883 (key.type == BTRFS_EXTENT_ITEM_KEY ||
884 key.type == BTRFS_METADATA_ITEM_KEY)) {
885 ret = __add_inline_refs(fs_info, path, bytenr,
886 &info_level, &prefs);
887 if (ret)
888 goto out;
889 ret = __add_keyed_refs(fs_info, path, bytenr,
890 info_level, &prefs);
891 if (ret)
892 goto out;
893 }
894 }
895 btrfs_release_path(path);
896
897 list_splice_init(&prefs_delayed, &prefs);
898
899 ret = __add_missing_keys(fs_info, &prefs);
900 if (ret)
901 goto out;
902
903 __merge_refs(&prefs, 1);
904
905 ret = __resolve_indirect_refs(fs_info, path, time_seq, &prefs,
906 extent_item_pos);
907 if (ret)
908 goto out;
909
910 __merge_refs(&prefs, 2);
911
912 while (!list_empty(&prefs)) {
913 ref = list_first_entry(&prefs, struct __prelim_ref, list);
914 list_del(&ref->list);
915 WARN_ON(ref->count < 0);
916 if (ref->count && ref->root_id && ref->parent == 0) {
917 /* no parent == root of tree */
918 ret = ulist_add(roots, ref->root_id, 0, GFP_NOFS);
919 if (ret < 0)
920 goto out;
921 }
922 if (ref->count && ref->parent) {
923 struct extent_inode_elem *eie = NULL;
924 if (extent_item_pos && !ref->inode_list) {
925 u32 bsz;
926 struct extent_buffer *eb;
927 bsz = btrfs_level_size(fs_info->extent_root,
928 info_level);
929 eb = read_tree_block(fs_info->extent_root,
930 ref->parent, bsz, 0);
931 if (!eb || !extent_buffer_uptodate(eb)) {
932 free_extent_buffer(eb);
933 ret = -EIO;
934 goto out;
935 }
936 ret = find_extent_in_eb(eb, bytenr,
937 *extent_item_pos, &eie);
938 ref->inode_list = eie;
939 free_extent_buffer(eb);
940 }
941 ret = ulist_add_merge(refs, ref->parent,
942 (uintptr_t)ref->inode_list,
943 (u64 *)&eie, GFP_NOFS);
944 if (ret < 0)
945 goto out;
946 if (!ret && extent_item_pos) {
947 /*
948 * we've recorded that parent, so we must extend
949 * its inode list here
950 */
951 BUG_ON(!eie);
952 while (eie->next)
953 eie = eie->next;
954 eie->next = ref->inode_list;
955 }
956 }
957 kfree(ref);
958 }
959
960 out:
961 btrfs_free_path(path);
962 while (!list_empty(&prefs)) {
963 ref = list_first_entry(&prefs, struct __prelim_ref, list);
964 list_del(&ref->list);
965 kfree(ref);
966 }
967 while (!list_empty(&prefs_delayed)) {
968 ref = list_first_entry(&prefs_delayed, struct __prelim_ref,
969 list);
970 list_del(&ref->list);
971 kfree(ref);
972 }
973
974 return ret;
975 }
976
977 static void free_leaf_list(struct ulist *blocks)
978 {
979 struct ulist_node *node = NULL;
980 struct extent_inode_elem *eie;
981 struct extent_inode_elem *eie_next;
982 struct ulist_iterator uiter;
983
984 ULIST_ITER_INIT(&uiter);
985 while ((node = ulist_next(blocks, &uiter))) {
986 if (!node->aux)
987 continue;
988 eie = (struct extent_inode_elem *)(uintptr_t)node->aux;
989 for (; eie; eie = eie_next) {
990 eie_next = eie->next;
991 kfree(eie);
992 }
993 node->aux = 0;
994 }
995
996 ulist_free(blocks);
997 }
998
999 /*
1000 * Finds all leafs with a reference to the specified combination of bytenr and
1001 * offset. key_list_head will point to a list of corresponding keys (caller must
1002 * free each list element). The leafs will be stored in the leafs ulist, which
1003 * must be freed with ulist_free.
1004 *
1005 * returns 0 on success, <0 on error
1006 */
1007 static int btrfs_find_all_leafs(struct btrfs_trans_handle *trans,
1008 struct btrfs_fs_info *fs_info, u64 bytenr,
1009 u64 time_seq, struct ulist **leafs,
1010 const u64 *extent_item_pos)
1011 {
1012 struct ulist *tmp;
1013 int ret;
1014
1015 tmp = ulist_alloc(GFP_NOFS);
1016 if (!tmp)
1017 return -ENOMEM;
1018 *leafs = ulist_alloc(GFP_NOFS);
1019 if (!*leafs) {
1020 ulist_free(tmp);
1021 return -ENOMEM;
1022 }
1023
1024 ret = find_parent_nodes(trans, fs_info, bytenr,
1025 time_seq, *leafs, tmp, extent_item_pos);
1026 ulist_free(tmp);
1027
1028 if (ret < 0 && ret != -ENOENT) {
1029 free_leaf_list(*leafs);
1030 return ret;
1031 }
1032
1033 return 0;
1034 }
1035
1036 /*
1037 * walk all backrefs for a given extent to find all roots that reference this
1038 * extent. Walking a backref means finding all extents that reference this
1039 * extent and in turn walk the backrefs of those, too. Naturally this is a
1040 * recursive process, but here it is implemented in an iterative fashion: We
1041 * find all referencing extents for the extent in question and put them on a
1042 * list. In turn, we find all referencing extents for those, further appending
1043 * to the list. The way we iterate the list allows adding more elements after
1044 * the current while iterating. The process stops when we reach the end of the
1045 * list. Found roots are added to the roots list.
1046 *
1047 * returns 0 on success, < 0 on error.
1048 */
1049 int btrfs_find_all_roots(struct btrfs_trans_handle *trans,
1050 struct btrfs_fs_info *fs_info, u64 bytenr,
1051 u64 time_seq, struct ulist **roots)
1052 {
1053 struct ulist *tmp;
1054 struct ulist_node *node = NULL;
1055 struct ulist_iterator uiter;
1056 int ret;
1057
1058 tmp = ulist_alloc(GFP_NOFS);
1059 if (!tmp)
1060 return -ENOMEM;
1061 *roots = ulist_alloc(GFP_NOFS);
1062 if (!*roots) {
1063 ulist_free(tmp);
1064 return -ENOMEM;
1065 }
1066
1067 ULIST_ITER_INIT(&uiter);
1068 while (1) {
1069 ret = find_parent_nodes(trans, fs_info, bytenr,
1070 time_seq, tmp, *roots, NULL);
1071 if (ret < 0 && ret != -ENOENT) {
1072 ulist_free(tmp);
1073 ulist_free(*roots);
1074 return ret;
1075 }
1076 node = ulist_next(tmp, &uiter);
1077 if (!node)
1078 break;
1079 bytenr = node->val;
1080 }
1081
1082 ulist_free(tmp);
1083 return 0;
1084 }
1085
1086
1087 static int __inode_info(u64 inum, u64 ioff, u8 key_type,
1088 struct btrfs_root *fs_root, struct btrfs_path *path,
1089 struct btrfs_key *found_key)
1090 {
1091 int ret;
1092 struct btrfs_key key;
1093 struct extent_buffer *eb;
1094
1095 key.type = key_type;
1096 key.objectid = inum;
1097 key.offset = ioff;
1098
1099 ret = btrfs_search_slot(NULL, fs_root, &key, path, 0, 0);
1100 if (ret < 0)
1101 return ret;
1102
1103 eb = path->nodes[0];
1104 if (ret && path->slots[0] >= btrfs_header_nritems(eb)) {
1105 ret = btrfs_next_leaf(fs_root, path);
1106 if (ret)
1107 return ret;
1108 eb = path->nodes[0];
1109 }
1110
1111 btrfs_item_key_to_cpu(eb, found_key, path->slots[0]);
1112 if (found_key->type != key.type || found_key->objectid != key.objectid)
1113 return 1;
1114
1115 return 0;
1116 }
1117
1118 /*
1119 * this makes the path point to (inum INODE_ITEM ioff)
1120 */
1121 int inode_item_info(u64 inum, u64 ioff, struct btrfs_root *fs_root,
1122 struct btrfs_path *path)
1123 {
1124 struct btrfs_key key;
1125 return __inode_info(inum, ioff, BTRFS_INODE_ITEM_KEY, fs_root, path,
1126 &key);
1127 }
1128
1129 static int inode_ref_info(u64 inum, u64 ioff, struct btrfs_root *fs_root,
1130 struct btrfs_path *path,
1131 struct btrfs_key *found_key)
1132 {
1133 return __inode_info(inum, ioff, BTRFS_INODE_REF_KEY, fs_root, path,
1134 found_key);
1135 }
1136
1137 int btrfs_find_one_extref(struct btrfs_root *root, u64 inode_objectid,
1138 u64 start_off, struct btrfs_path *path,
1139 struct btrfs_inode_extref **ret_extref,
1140 u64 *found_off)
1141 {
1142 int ret, slot;
1143 struct btrfs_key key;
1144 struct btrfs_key found_key;
1145 struct btrfs_inode_extref *extref;
1146 struct extent_buffer *leaf;
1147 unsigned long ptr;
1148
1149 key.objectid = inode_objectid;
1150 btrfs_set_key_type(&key, BTRFS_INODE_EXTREF_KEY);
1151 key.offset = start_off;
1152
1153 ret = btrfs_search_slot(NULL, root, &key, path, 0, 0);
1154 if (ret < 0)
1155 return ret;
1156
1157 while (1) {
1158 leaf = path->nodes[0];
1159 slot = path->slots[0];
1160 if (slot >= btrfs_header_nritems(leaf)) {
1161 /*
1162 * If the item at offset is not found,
1163 * btrfs_search_slot will point us to the slot
1164 * where it should be inserted. In our case
1165 * that will be the slot directly before the
1166 * next INODE_REF_KEY_V2 item. In the case
1167 * that we're pointing to the last slot in a
1168 * leaf, we must move one leaf over.
1169 */
1170 ret = btrfs_next_leaf(root, path);
1171 if (ret) {
1172 if (ret >= 1)
1173 ret = -ENOENT;
1174 break;
1175 }
1176 continue;
1177 }
1178
1179 btrfs_item_key_to_cpu(leaf, &found_key, slot);
1180
1181 /*
1182 * Check that we're still looking at an extended ref key for
1183 * this particular objectid. If we have different
1184 * objectid or type then there are no more to be found
1185 * in the tree and we can exit.
1186 */
1187 ret = -ENOENT;
1188 if (found_key.objectid != inode_objectid)
1189 break;
1190 if (btrfs_key_type(&found_key) != BTRFS_INODE_EXTREF_KEY)
1191 break;
1192
1193 ret = 0;
1194 ptr = btrfs_item_ptr_offset(leaf, path->slots[0]);
1195 extref = (struct btrfs_inode_extref *)ptr;
1196 *ret_extref = extref;
1197 if (found_off)
1198 *found_off = found_key.offset;
1199 break;
1200 }
1201
1202 return ret;
1203 }
1204
1205 /*
1206 * this iterates to turn a name (from iref/extref) into a full filesystem path.
1207 * Elements of the path are separated by '/' and the path is guaranteed to be
1208 * 0-terminated. the path is only given within the current file system.
1209 * Therefore, it never starts with a '/'. the caller is responsible to provide
1210 * "size" bytes in "dest". the dest buffer will be filled backwards. finally,
1211 * the start point of the resulting string is returned. this pointer is within
1212 * dest, normally.
1213 * in case the path buffer would overflow, the pointer is decremented further
1214 * as if output was written to the buffer, though no more output is actually
1215 * generated. that way, the caller can determine how much space would be
1216 * required for the path to fit into the buffer. in that case, the returned
1217 * value will be smaller than dest. callers must check this!
1218 */
1219 char *btrfs_ref_to_path(struct btrfs_root *fs_root, struct btrfs_path *path,
1220 u32 name_len, unsigned long name_off,
1221 struct extent_buffer *eb_in, u64 parent,
1222 char *dest, u32 size)
1223 {
1224 int slot;
1225 u64 next_inum;
1226 int ret;
1227 s64 bytes_left = ((s64)size) - 1;
1228 struct extent_buffer *eb = eb_in;
1229 struct btrfs_key found_key;
1230 int leave_spinning = path->leave_spinning;
1231 struct btrfs_inode_ref *iref;
1232
1233 if (bytes_left >= 0)
1234 dest[bytes_left] = '\0';
1235
1236 path->leave_spinning = 1;
1237 while (1) {
1238 bytes_left -= name_len;
1239 if (bytes_left >= 0)
1240 read_extent_buffer(eb, dest + bytes_left,
1241 name_off, name_len);
1242 if (eb != eb_in) {
1243 btrfs_tree_read_unlock_blocking(eb);
1244 free_extent_buffer(eb);
1245 }
1246 ret = inode_ref_info(parent, 0, fs_root, path, &found_key);
1247 if (ret > 0)
1248 ret = -ENOENT;
1249 if (ret)
1250 break;
1251
1252 next_inum = found_key.offset;
1253
1254 /* regular exit ahead */
1255 if (parent == next_inum)
1256 break;
1257
1258 slot = path->slots[0];
1259 eb = path->nodes[0];
1260 /* make sure we can use eb after releasing the path */
1261 if (eb != eb_in) {
1262 atomic_inc(&eb->refs);
1263 btrfs_tree_read_lock(eb);
1264 btrfs_set_lock_blocking_rw(eb, BTRFS_READ_LOCK);
1265 }
1266 btrfs_release_path(path);
1267 iref = btrfs_item_ptr(eb, slot, struct btrfs_inode_ref);
1268
1269 name_len = btrfs_inode_ref_name_len(eb, iref);
1270 name_off = (unsigned long)(iref + 1);
1271
1272 parent = next_inum;
1273 --bytes_left;
1274 if (bytes_left >= 0)
1275 dest[bytes_left] = '/';
1276 }
1277
1278 btrfs_release_path(path);
1279 path->leave_spinning = leave_spinning;
1280
1281 if (ret)
1282 return ERR_PTR(ret);
1283
1284 return dest + bytes_left;
1285 }
1286
1287 /*
1288 * this makes the path point to (logical EXTENT_ITEM *)
1289 * returns BTRFS_EXTENT_FLAG_DATA for data, BTRFS_EXTENT_FLAG_TREE_BLOCK for
1290 * tree blocks and <0 on error.
1291 */
1292 int extent_from_logical(struct btrfs_fs_info *fs_info, u64 logical,
1293 struct btrfs_path *path, struct btrfs_key *found_key,
1294 u64 *flags_ret)
1295 {
1296 int ret;
1297 u64 flags;
1298 u64 size = 0;
1299 u32 item_size;
1300 struct extent_buffer *eb;
1301 struct btrfs_extent_item *ei;
1302 struct btrfs_key key;
1303
1304 if (btrfs_fs_incompat(fs_info, SKINNY_METADATA))
1305 key.type = BTRFS_METADATA_ITEM_KEY;
1306 else
1307 key.type = BTRFS_EXTENT_ITEM_KEY;
1308 key.objectid = logical;
1309 key.offset = (u64)-1;
1310
1311 ret = btrfs_search_slot(NULL, fs_info->extent_root, &key, path, 0, 0);
1312 if (ret < 0)
1313 return ret;
1314 ret = btrfs_previous_item(fs_info->extent_root, path,
1315 0, BTRFS_EXTENT_ITEM_KEY);
1316 if (ret < 0)
1317 return ret;
1318
1319 btrfs_item_key_to_cpu(path->nodes[0], found_key, path->slots[0]);
1320 if (found_key->type == BTRFS_METADATA_ITEM_KEY)
1321 size = fs_info->extent_root->leafsize;
1322 else if (found_key->type == BTRFS_EXTENT_ITEM_KEY)
1323 size = found_key->offset;
1324
1325 if ((found_key->type != BTRFS_EXTENT_ITEM_KEY &&
1326 found_key->type != BTRFS_METADATA_ITEM_KEY) ||
1327 found_key->objectid > logical ||
1328 found_key->objectid + size <= logical) {
1329 pr_debug("logical %llu is not within any extent\n",
1330 (unsigned long long)logical);
1331 return -ENOENT;
1332 }
1333
1334 eb = path->nodes[0];
1335 item_size = btrfs_item_size_nr(eb, path->slots[0]);
1336 BUG_ON(item_size < sizeof(*ei));
1337
1338 ei = btrfs_item_ptr(eb, path->slots[0], struct btrfs_extent_item);
1339 flags = btrfs_extent_flags(eb, ei);
1340
1341 pr_debug("logical %llu is at position %llu within the extent (%llu "
1342 "EXTENT_ITEM %llu) flags %#llx size %u\n",
1343 (unsigned long long)logical,
1344 (unsigned long long)(logical - found_key->objectid),
1345 (unsigned long long)found_key->objectid,
1346 (unsigned long long)found_key->offset,
1347 (unsigned long long)flags, item_size);
1348
1349 WARN_ON(!flags_ret);
1350 if (flags_ret) {
1351 if (flags & BTRFS_EXTENT_FLAG_TREE_BLOCK)
1352 *flags_ret = BTRFS_EXTENT_FLAG_TREE_BLOCK;
1353 else if (flags & BTRFS_EXTENT_FLAG_DATA)
1354 *flags_ret = BTRFS_EXTENT_FLAG_DATA;
1355 else
1356 BUG_ON(1);
1357 return 0;
1358 }
1359
1360 return -EIO;
1361 }
1362
1363 /*
1364 * helper function to iterate extent inline refs. ptr must point to a 0 value
1365 * for the first call and may be modified. it is used to track state.
1366 * if more refs exist, 0 is returned and the next call to
1367 * __get_extent_inline_ref must pass the modified ptr parameter to get the
1368 * next ref. after the last ref was processed, 1 is returned.
1369 * returns <0 on error
1370 */
1371 static int __get_extent_inline_ref(unsigned long *ptr, struct extent_buffer *eb,
1372 struct btrfs_extent_item *ei, u32 item_size,
1373 struct btrfs_extent_inline_ref **out_eiref,
1374 int *out_type)
1375 {
1376 unsigned long end;
1377 u64 flags;
1378 struct btrfs_tree_block_info *info;
1379
1380 if (!*ptr) {
1381 /* first call */
1382 flags = btrfs_extent_flags(eb, ei);
1383 if (flags & BTRFS_EXTENT_FLAG_TREE_BLOCK) {
1384 info = (struct btrfs_tree_block_info *)(ei + 1);
1385 *out_eiref =
1386 (struct btrfs_extent_inline_ref *)(info + 1);
1387 } else {
1388 *out_eiref = (struct btrfs_extent_inline_ref *)(ei + 1);
1389 }
1390 *ptr = (unsigned long)*out_eiref;
1391 if ((void *)*ptr >= (void *)ei + item_size)
1392 return -ENOENT;
1393 }
1394
1395 end = (unsigned long)ei + item_size;
1396 *out_eiref = (struct btrfs_extent_inline_ref *)*ptr;
1397 *out_type = btrfs_extent_inline_ref_type(eb, *out_eiref);
1398
1399 *ptr += btrfs_extent_inline_ref_size(*out_type);
1400 WARN_ON(*ptr > end);
1401 if (*ptr == end)
1402 return 1; /* last */
1403
1404 return 0;
1405 }
1406
1407 /*
1408 * reads the tree block backref for an extent. tree level and root are returned
1409 * through out_level and out_root. ptr must point to a 0 value for the first
1410 * call and may be modified (see __get_extent_inline_ref comment).
1411 * returns 0 if data was provided, 1 if there was no more data to provide or
1412 * <0 on error.
1413 */
1414 int tree_backref_for_extent(unsigned long *ptr, struct extent_buffer *eb,
1415 struct btrfs_extent_item *ei, u32 item_size,
1416 u64 *out_root, u8 *out_level)
1417 {
1418 int ret;
1419 int type;
1420 struct btrfs_tree_block_info *info;
1421 struct btrfs_extent_inline_ref *eiref;
1422
1423 if (*ptr == (unsigned long)-1)
1424 return 1;
1425
1426 while (1) {
1427 ret = __get_extent_inline_ref(ptr, eb, ei, item_size,
1428 &eiref, &type);
1429 if (ret < 0)
1430 return ret;
1431
1432 if (type == BTRFS_TREE_BLOCK_REF_KEY ||
1433 type == BTRFS_SHARED_BLOCK_REF_KEY)
1434 break;
1435
1436 if (ret == 1)
1437 return 1;
1438 }
1439
1440 /* we can treat both ref types equally here */
1441 info = (struct btrfs_tree_block_info *)(ei + 1);
1442 *out_root = btrfs_extent_inline_ref_offset(eb, eiref);
1443 *out_level = btrfs_tree_block_level(eb, info);
1444
1445 if (ret == 1)
1446 *ptr = (unsigned long)-1;
1447
1448 return 0;
1449 }
1450
1451 static int iterate_leaf_refs(struct extent_inode_elem *inode_list,
1452 u64 root, u64 extent_item_objectid,
1453 iterate_extent_inodes_t *iterate, void *ctx)
1454 {
1455 struct extent_inode_elem *eie;
1456 int ret = 0;
1457
1458 for (eie = inode_list; eie; eie = eie->next) {
1459 pr_debug("ref for %llu resolved, key (%llu EXTEND_DATA %llu), "
1460 "root %llu\n", extent_item_objectid,
1461 eie->inum, eie->offset, root);
1462 ret = iterate(eie->inum, eie->offset, root, ctx);
1463 if (ret) {
1464 pr_debug("stopping iteration for %llu due to ret=%d\n",
1465 extent_item_objectid, ret);
1466 break;
1467 }
1468 }
1469
1470 return ret;
1471 }
1472
1473 /*
1474 * calls iterate() for every inode that references the extent identified by
1475 * the given parameters.
1476 * when the iterator function returns a non-zero value, iteration stops.
1477 */
1478 int iterate_extent_inodes(struct btrfs_fs_info *fs_info,
1479 u64 extent_item_objectid, u64 extent_item_pos,
1480 int search_commit_root,
1481 iterate_extent_inodes_t *iterate, void *ctx)
1482 {
1483 int ret;
1484 struct btrfs_trans_handle *trans = NULL;
1485 struct ulist *refs = NULL;
1486 struct ulist *roots = NULL;
1487 struct ulist_node *ref_node = NULL;
1488 struct ulist_node *root_node = NULL;
1489 struct seq_list tree_mod_seq_elem = {};
1490 struct ulist_iterator ref_uiter;
1491 struct ulist_iterator root_uiter;
1492
1493 pr_debug("resolving all inodes for extent %llu\n",
1494 extent_item_objectid);
1495
1496 if (!search_commit_root) {
1497 trans = btrfs_join_transaction(fs_info->extent_root);
1498 if (IS_ERR(trans))
1499 return PTR_ERR(trans);
1500 btrfs_get_tree_mod_seq(fs_info, &tree_mod_seq_elem);
1501 }
1502
1503 ret = btrfs_find_all_leafs(trans, fs_info, extent_item_objectid,
1504 tree_mod_seq_elem.seq, &refs,
1505 &extent_item_pos);
1506 if (ret)
1507 goto out;
1508
1509 ULIST_ITER_INIT(&ref_uiter);
1510 while (!ret && (ref_node = ulist_next(refs, &ref_uiter))) {
1511 ret = btrfs_find_all_roots(trans, fs_info, ref_node->val,
1512 tree_mod_seq_elem.seq, &roots);
1513 if (ret)
1514 break;
1515 ULIST_ITER_INIT(&root_uiter);
1516 while (!ret && (root_node = ulist_next(roots, &root_uiter))) {
1517 pr_debug("root %llu references leaf %llu, data list "
1518 "%#llx\n", root_node->val, ref_node->val,
1519 (long long)ref_node->aux);
1520 ret = iterate_leaf_refs((struct extent_inode_elem *)
1521 (uintptr_t)ref_node->aux,
1522 root_node->val,
1523 extent_item_objectid,
1524 iterate, ctx);
1525 }
1526 ulist_free(roots);
1527 }
1528
1529 free_leaf_list(refs);
1530 out:
1531 if (!search_commit_root) {
1532 btrfs_put_tree_mod_seq(fs_info, &tree_mod_seq_elem);
1533 btrfs_end_transaction(trans, fs_info->extent_root);
1534 }
1535
1536 return ret;
1537 }
1538
1539 int iterate_inodes_from_logical(u64 logical, struct btrfs_fs_info *fs_info,
1540 struct btrfs_path *path,
1541 iterate_extent_inodes_t *iterate, void *ctx)
1542 {
1543 int ret;
1544 u64 extent_item_pos;
1545 u64 flags = 0;
1546 struct btrfs_key found_key;
1547 int search_commit_root = path->search_commit_root;
1548
1549 ret = extent_from_logical(fs_info, logical, path, &found_key, &flags);
1550 btrfs_release_path(path);
1551 if (ret < 0)
1552 return ret;
1553 if (flags & BTRFS_EXTENT_FLAG_TREE_BLOCK)
1554 return -EINVAL;
1555
1556 extent_item_pos = logical - found_key.objectid;
1557 ret = iterate_extent_inodes(fs_info, found_key.objectid,
1558 extent_item_pos, search_commit_root,
1559 iterate, ctx);
1560
1561 return ret;
1562 }
1563
1564 typedef int (iterate_irefs_t)(u64 parent, u32 name_len, unsigned long name_off,
1565 struct extent_buffer *eb, void *ctx);
1566
1567 static int iterate_inode_refs(u64 inum, struct btrfs_root *fs_root,
1568 struct btrfs_path *path,
1569 iterate_irefs_t *iterate, void *ctx)
1570 {
1571 int ret = 0;
1572 int slot;
1573 u32 cur;
1574 u32 len;
1575 u32 name_len;
1576 u64 parent = 0;
1577 int found = 0;
1578 struct extent_buffer *eb;
1579 struct btrfs_item *item;
1580 struct btrfs_inode_ref *iref;
1581 struct btrfs_key found_key;
1582
1583 while (!ret) {
1584 path->leave_spinning = 1;
1585 ret = inode_ref_info(inum, parent ? parent+1 : 0, fs_root, path,
1586 &found_key);
1587 if (ret < 0)
1588 break;
1589 if (ret) {
1590 ret = found ? 0 : -ENOENT;
1591 break;
1592 }
1593 ++found;
1594
1595 parent = found_key.offset;
1596 slot = path->slots[0];
1597 eb = path->nodes[0];
1598 /* make sure we can use eb after releasing the path */
1599 atomic_inc(&eb->refs);
1600 btrfs_tree_read_lock(eb);
1601 btrfs_set_lock_blocking_rw(eb, BTRFS_READ_LOCK);
1602 btrfs_release_path(path);
1603
1604 item = btrfs_item_nr(eb, slot);
1605 iref = btrfs_item_ptr(eb, slot, struct btrfs_inode_ref);
1606
1607 for (cur = 0; cur < btrfs_item_size(eb, item); cur += len) {
1608 name_len = btrfs_inode_ref_name_len(eb, iref);
1609 /* path must be released before calling iterate()! */
1610 pr_debug("following ref at offset %u for inode %llu in "
1611 "tree %llu\n", cur,
1612 (unsigned long long)found_key.objectid,
1613 (unsigned long long)fs_root->objectid);
1614 ret = iterate(parent, name_len,
1615 (unsigned long)(iref + 1), eb, ctx);
1616 if (ret)
1617 break;
1618 len = sizeof(*iref) + name_len;
1619 iref = (struct btrfs_inode_ref *)((char *)iref + len);
1620 }
1621 btrfs_tree_read_unlock_blocking(eb);
1622 free_extent_buffer(eb);
1623 }
1624
1625 btrfs_release_path(path);
1626
1627 return ret;
1628 }
1629
1630 static int iterate_inode_extrefs(u64 inum, struct btrfs_root *fs_root,
1631 struct btrfs_path *path,
1632 iterate_irefs_t *iterate, void *ctx)
1633 {
1634 int ret;
1635 int slot;
1636 u64 offset = 0;
1637 u64 parent;
1638 int found = 0;
1639 struct extent_buffer *eb;
1640 struct btrfs_inode_extref *extref;
1641 struct extent_buffer *leaf;
1642 u32 item_size;
1643 u32 cur_offset;
1644 unsigned long ptr;
1645
1646 while (1) {
1647 ret = btrfs_find_one_extref(fs_root, inum, offset, path, &extref,
1648 &offset);
1649 if (ret < 0)
1650 break;
1651 if (ret) {
1652 ret = found ? 0 : -ENOENT;
1653 break;
1654 }
1655 ++found;
1656
1657 slot = path->slots[0];
1658 eb = path->nodes[0];
1659 /* make sure we can use eb after releasing the path */
1660 atomic_inc(&eb->refs);
1661
1662 btrfs_tree_read_lock(eb);
1663 btrfs_set_lock_blocking_rw(eb, BTRFS_READ_LOCK);
1664 btrfs_release_path(path);
1665
1666 leaf = path->nodes[0];
1667 item_size = btrfs_item_size_nr(leaf, path->slots[0]);
1668 ptr = btrfs_item_ptr_offset(leaf, path->slots[0]);
1669 cur_offset = 0;
1670
1671 while (cur_offset < item_size) {
1672 u32 name_len;
1673
1674 extref = (struct btrfs_inode_extref *)(ptr + cur_offset);
1675 parent = btrfs_inode_extref_parent(eb, extref);
1676 name_len = btrfs_inode_extref_name_len(eb, extref);
1677 ret = iterate(parent, name_len,
1678 (unsigned long)&extref->name, eb, ctx);
1679 if (ret)
1680 break;
1681
1682 cur_offset += btrfs_inode_extref_name_len(leaf, extref);
1683 cur_offset += sizeof(*extref);
1684 }
1685 btrfs_tree_read_unlock_blocking(eb);
1686 free_extent_buffer(eb);
1687
1688 offset++;
1689 }
1690
1691 btrfs_release_path(path);
1692
1693 return ret;
1694 }
1695
1696 static int iterate_irefs(u64 inum, struct btrfs_root *fs_root,
1697 struct btrfs_path *path, iterate_irefs_t *iterate,
1698 void *ctx)
1699 {
1700 int ret;
1701 int found_refs = 0;
1702
1703 ret = iterate_inode_refs(inum, fs_root, path, iterate, ctx);
1704 if (!ret)
1705 ++found_refs;
1706 else if (ret != -ENOENT)
1707 return ret;
1708
1709 ret = iterate_inode_extrefs(inum, fs_root, path, iterate, ctx);
1710 if (ret == -ENOENT && found_refs)
1711 return 0;
1712
1713 return ret;
1714 }
1715
1716 /*
1717 * returns 0 if the path could be dumped (probably truncated)
1718 * returns <0 in case of an error
1719 */
1720 static int inode_to_path(u64 inum, u32 name_len, unsigned long name_off,
1721 struct extent_buffer *eb, void *ctx)
1722 {
1723 struct inode_fs_paths *ipath = ctx;
1724 char *fspath;
1725 char *fspath_min;
1726 int i = ipath->fspath->elem_cnt;
1727 const int s_ptr = sizeof(char *);
1728 u32 bytes_left;
1729
1730 bytes_left = ipath->fspath->bytes_left > s_ptr ?
1731 ipath->fspath->bytes_left - s_ptr : 0;
1732
1733 fspath_min = (char *)ipath->fspath->val + (i + 1) * s_ptr;
1734 fspath = btrfs_ref_to_path(ipath->fs_root, ipath->btrfs_path, name_len,
1735 name_off, eb, inum, fspath_min, bytes_left);
1736 if (IS_ERR(fspath))
1737 return PTR_ERR(fspath);
1738
1739 if (fspath > fspath_min) {
1740 ipath->fspath->val[i] = (u64)(unsigned long)fspath;
1741 ++ipath->fspath->elem_cnt;
1742 ipath->fspath->bytes_left = fspath - fspath_min;
1743 } else {
1744 ++ipath->fspath->elem_missed;
1745 ipath->fspath->bytes_missing += fspath_min - fspath;
1746 ipath->fspath->bytes_left = 0;
1747 }
1748
1749 return 0;
1750 }
1751
1752 /*
1753 * this dumps all file system paths to the inode into the ipath struct, provided
1754 * is has been created large enough. each path is zero-terminated and accessed
1755 * from ipath->fspath->val[i].
1756 * when it returns, there are ipath->fspath->elem_cnt number of paths available
1757 * in ipath->fspath->val[]. when the allocated space wasn't sufficient, the
1758 * number of missed paths in recored in ipath->fspath->elem_missed, otherwise,
1759 * it's zero. ipath->fspath->bytes_missing holds the number of bytes that would
1760 * have been needed to return all paths.
1761 */
1762 int paths_from_inode(u64 inum, struct inode_fs_paths *ipath)
1763 {
1764 return iterate_irefs(inum, ipath->fs_root, ipath->btrfs_path,
1765 inode_to_path, ipath);
1766 }
1767
1768 struct btrfs_data_container *init_data_container(u32 total_bytes)
1769 {
1770 struct btrfs_data_container *data;
1771 size_t alloc_bytes;
1772
1773 alloc_bytes = max_t(size_t, total_bytes, sizeof(*data));
1774 data = vmalloc(alloc_bytes);
1775 if (!data)
1776 return ERR_PTR(-ENOMEM);
1777
1778 if (total_bytes >= sizeof(*data)) {
1779 data->bytes_left = total_bytes - sizeof(*data);
1780 data->bytes_missing = 0;
1781 } else {
1782 data->bytes_missing = sizeof(*data) - total_bytes;
1783 data->bytes_left = 0;
1784 }
1785
1786 data->elem_cnt = 0;
1787 data->elem_missed = 0;
1788
1789 return data;
1790 }
1791
1792 /*
1793 * allocates space to return multiple file system paths for an inode.
1794 * total_bytes to allocate are passed, note that space usable for actual path
1795 * information will be total_bytes - sizeof(struct inode_fs_paths).
1796 * the returned pointer must be freed with free_ipath() in the end.
1797 */
1798 struct inode_fs_paths *init_ipath(s32 total_bytes, struct btrfs_root *fs_root,
1799 struct btrfs_path *path)
1800 {
1801 struct inode_fs_paths *ifp;
1802 struct btrfs_data_container *fspath;
1803
1804 fspath = init_data_container(total_bytes);
1805 if (IS_ERR(fspath))
1806 return (void *)fspath;
1807
1808 ifp = kmalloc(sizeof(*ifp), GFP_NOFS);
1809 if (!ifp) {
1810 kfree(fspath);
1811 return ERR_PTR(-ENOMEM);
1812 }
1813
1814 ifp->btrfs_path = path;
1815 ifp->fspath = fspath;
1816 ifp->fs_root = fs_root;
1817
1818 return ifp;
1819 }
1820
1821 void free_ipath(struct inode_fs_paths *ipath)
1822 {
1823 if (!ipath)
1824 return;
1825 vfree(ipath->fspath);
1826 kfree(ipath);
1827 }
Linux with added FPC-III support