2 * Copyright (C) 2007 Oracle. All rights reserved.
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.
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.
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.
18 #include <linux/sched.h>
19 #include <linux/pagemap.h>
20 #include <linux/writeback.h>
21 #include <linux/blkdev.h>
26 #include "print-tree.h"
27 #include "transaction.h"
30 #include "ref-cache.h"
32 #define PENDING_EXTENT_INSERT 0
33 #define PENDING_EXTENT_DELETE 1
34 #define PENDING_BACKREF_UPDATE 2
36 struct pending_extent_op {
47 static int finish_current_insert(struct btrfs_trans_handle *trans, struct
48 btrfs_root *extent_root);
49 static int del_pending_extents(struct btrfs_trans_handle *trans, struct
50 btrfs_root *extent_root);
51 static struct btrfs_block_group_cache *
52 __btrfs_find_block_group(struct btrfs_root *root,
53 struct btrfs_block_group_cache *hint,
54 u64 search_start, int data, int owner);
56 void maybe_lock_mutex(struct btrfs_root *root)
58 if (root != root->fs_info->extent_root &&
59 root != root->fs_info->chunk_root &&
60 root != root->fs_info->dev_root) {
61 mutex_lock(&root->fs_info->alloc_mutex);
65 void maybe_unlock_mutex(struct btrfs_root *root)
67 if (root != root->fs_info->extent_root &&
68 root != root->fs_info->chunk_root &&
69 root != root->fs_info->dev_root) {
70 mutex_unlock(&root->fs_info->alloc_mutex);
74 static int block_group_bits(struct btrfs_block_group_cache *cache, u64 bits)
76 return (cache->flags & bits) == bits;
80 * this adds the block group to the fs_info rb tree for the block group
83 int btrfs_add_block_group_cache(struct btrfs_fs_info *info,
84 struct btrfs_block_group_cache *block_group)
87 struct rb_node *parent = NULL;
88 struct btrfs_block_group_cache *cache;
90 spin_lock(&info->block_group_cache_lock);
91 p = &info->block_group_cache_tree.rb_node;
95 cache = rb_entry(parent, struct btrfs_block_group_cache,
97 if (block_group->key.objectid < cache->key.objectid) {
99 } else if (block_group->key.objectid > cache->key.objectid) {
102 spin_unlock(&info->block_group_cache_lock);
107 rb_link_node(&block_group->cache_node, parent, p);
108 rb_insert_color(&block_group->cache_node,
109 &info->block_group_cache_tree);
110 spin_unlock(&info->block_group_cache_lock);
116 * This will return the block group at or after bytenr if contains is 0, else
117 * it will return the block group that contains the bytenr
119 static struct btrfs_block_group_cache *
120 block_group_cache_tree_search(struct btrfs_fs_info *info, u64 bytenr,
123 struct btrfs_block_group_cache *cache, *ret = NULL;
127 spin_lock(&info->block_group_cache_lock);
128 n = info->block_group_cache_tree.rb_node;
131 cache = rb_entry(n, struct btrfs_block_group_cache,
133 end = cache->key.objectid + cache->key.offset - 1;
134 start = cache->key.objectid;
136 if (bytenr < start) {
137 if (!contains && (!ret || start < ret->key.objectid))
140 } else if (bytenr > start) {
141 if (contains && bytenr <= end) {
151 spin_unlock(&info->block_group_cache_lock);
157 * this is only called by cache_block_group, since we could have freed extents
158 * we need to check the pinned_extents for any extents that can't be used yet
159 * since their free space will be released as soon as the transaction commits.
161 static int add_new_free_space(struct btrfs_block_group_cache *block_group,
162 struct btrfs_fs_info *info, u64 start, u64 end)
164 u64 extent_start, extent_end, size;
167 while (start < end) {
168 ret = find_first_extent_bit(&info->pinned_extents, start,
169 &extent_start, &extent_end,
174 if (extent_start == start) {
175 start = extent_end + 1;
176 } else if (extent_start > start && extent_start < end) {
177 size = extent_start - start;
178 ret = btrfs_add_free_space(block_group, start, size);
180 start = extent_end + 1;
188 ret = btrfs_add_free_space(block_group, start, size);
195 static int cache_block_group(struct btrfs_root *root,
196 struct btrfs_block_group_cache *block_group)
198 struct btrfs_path *path;
200 struct btrfs_key key;
201 struct extent_buffer *leaf;
210 root = root->fs_info->extent_root;
212 if (block_group->cached)
215 path = btrfs_alloc_path();
221 * we get into deadlocks with paths held by callers of this function.
222 * since the alloc_mutex is protecting things right now, just
223 * skip the locking here
225 path->skip_locking = 1;
226 first_free = max_t(u64, block_group->key.objectid,
227 BTRFS_SUPER_INFO_OFFSET + BTRFS_SUPER_INFO_SIZE);
228 key.objectid = block_group->key.objectid;
230 btrfs_set_key_type(&key, BTRFS_EXTENT_ITEM_KEY);
231 ret = btrfs_search_slot(NULL, root, &key, path, 0, 0);
234 ret = btrfs_previous_item(root, path, 0, BTRFS_EXTENT_ITEM_KEY);
238 leaf = path->nodes[0];
239 btrfs_item_key_to_cpu(leaf, &key, path->slots[0]);
240 if (key.objectid + key.offset > first_free)
241 first_free = key.objectid + key.offset;
244 leaf = path->nodes[0];
245 slot = path->slots[0];
246 if (slot >= btrfs_header_nritems(leaf)) {
247 ret = btrfs_next_leaf(root, path);
255 btrfs_item_key_to_cpu(leaf, &key, slot);
256 if (key.objectid < block_group->key.objectid)
259 if (key.objectid >= block_group->key.objectid +
260 block_group->key.offset)
263 if (btrfs_key_type(&key) == BTRFS_EXTENT_ITEM_KEY) {
269 add_new_free_space(block_group, root->fs_info, last,
272 last = key.objectid + key.offset;
281 add_new_free_space(block_group, root->fs_info, last,
282 block_group->key.objectid +
283 block_group->key.offset);
285 block_group->cached = 1;
288 btrfs_free_path(path);
293 * return the block group that starts at or after bytenr
295 struct btrfs_block_group_cache *btrfs_lookup_first_block_group(struct
299 struct btrfs_block_group_cache *cache;
301 cache = block_group_cache_tree_search(info, bytenr, 0);
307 * return the block group that contains teh given bytenr
309 struct btrfs_block_group_cache *btrfs_lookup_block_group(struct
313 struct btrfs_block_group_cache *cache;
315 cache = block_group_cache_tree_search(info, bytenr, 1);
320 static struct btrfs_space_info *__find_space_info(struct btrfs_fs_info *info,
323 struct list_head *head = &info->space_info;
324 struct list_head *cur;
325 struct btrfs_space_info *found;
326 list_for_each(cur, head) {
327 found = list_entry(cur, struct btrfs_space_info, list);
328 if (found->flags == flags)
334 static u64 div_factor(u64 num, int factor)
343 static struct btrfs_block_group_cache *
344 __btrfs_find_block_group(struct btrfs_root *root,
345 struct btrfs_block_group_cache *hint,
346 u64 search_start, int data, int owner)
348 struct btrfs_block_group_cache *cache;
349 struct btrfs_block_group_cache *found_group = NULL;
350 struct btrfs_fs_info *info = root->fs_info;
358 if (data & BTRFS_BLOCK_GROUP_METADATA)
362 struct btrfs_block_group_cache *shint;
363 shint = btrfs_lookup_first_block_group(info, search_start);
364 if (shint && block_group_bits(shint, data) && !shint->ro) {
365 spin_lock(&shint->lock);
366 used = btrfs_block_group_used(&shint->item);
367 if (used + shint->pinned + shint->reserved <
368 div_factor(shint->key.offset, factor)) {
369 spin_unlock(&shint->lock);
372 spin_unlock(&shint->lock);
375 if (hint && !hint->ro && block_group_bits(hint, data)) {
376 spin_lock(&hint->lock);
377 used = btrfs_block_group_used(&hint->item);
378 if (used + hint->pinned + hint->reserved <
379 div_factor(hint->key.offset, factor)) {
380 spin_unlock(&hint->lock);
383 spin_unlock(&hint->lock);
384 last = hint->key.objectid + hint->key.offset;
387 last = max(hint->key.objectid, search_start);
393 cache = btrfs_lookup_first_block_group(root->fs_info, last);
397 spin_lock(&cache->lock);
398 last = cache->key.objectid + cache->key.offset;
399 used = btrfs_block_group_used(&cache->item);
401 if (!cache->ro && block_group_bits(cache, data)) {
402 free_check = div_factor(cache->key.offset, factor);
403 if (used + cache->pinned + cache->reserved <
406 spin_unlock(&cache->lock);
410 spin_unlock(&cache->lock);
418 if (!full_search && factor < 10) {
428 struct btrfs_block_group_cache *btrfs_find_block_group(struct btrfs_root *root,
429 struct btrfs_block_group_cache
430 *hint, u64 search_start,
434 struct btrfs_block_group_cache *ret;
435 ret = __btrfs_find_block_group(root, hint, search_start, data, owner);
439 /* simple helper to search for an existing extent at a given offset */
440 int btrfs_lookup_extent(struct btrfs_root *root, u64 start, u64 len)
443 struct btrfs_key key;
444 struct btrfs_path *path;
446 path = btrfs_alloc_path();
448 maybe_lock_mutex(root);
449 key.objectid = start;
451 btrfs_set_key_type(&key, BTRFS_EXTENT_ITEM_KEY);
452 ret = btrfs_search_slot(NULL, root->fs_info->extent_root, &key, path,
454 maybe_unlock_mutex(root);
455 btrfs_free_path(path);
460 * Back reference rules. Back refs have three main goals:
462 * 1) differentiate between all holders of references to an extent so that
463 * when a reference is dropped we can make sure it was a valid reference
464 * before freeing the extent.
466 * 2) Provide enough information to quickly find the holders of an extent
467 * if we notice a given block is corrupted or bad.
469 * 3) Make it easy to migrate blocks for FS shrinking or storage pool
470 * maintenance. This is actually the same as #2, but with a slightly
471 * different use case.
473 * File extents can be referenced by:
475 * - multiple snapshots, subvolumes, or different generations in one subvol
476 * - different files inside a single subvolume
477 * - different offsets inside a file (bookend extents in file.c)
479 * The extent ref structure has fields for:
481 * - Objectid of the subvolume root
482 * - Generation number of the tree holding the reference
483 * - objectid of the file holding the reference
484 * - number of references holding by parent node (alway 1 for tree blocks)
486 * Btree leaf may hold multiple references to a file extent. In most cases,
487 * these references are from same file and the corresponding offsets inside
488 * the file are close together.
490 * When a file extent is allocated the fields are filled in:
491 * (root_key.objectid, trans->transid, inode objectid, 1)
493 * When a leaf is cow'd new references are added for every file extent found
494 * in the leaf. It looks similar to the create case, but trans->transid will
495 * be different when the block is cow'd.
497 * (root_key.objectid, trans->transid, inode objectid,
498 * number of references in the leaf)
500 * When a file extent is removed either during snapshot deletion or
501 * file truncation, we find the corresponding back reference and check
502 * the following fields:
504 * (btrfs_header_owner(leaf), btrfs_header_generation(leaf),
507 * Btree extents can be referenced by:
509 * - Different subvolumes
510 * - Different generations of the same subvolume
512 * When a tree block is created, back references are inserted:
514 * (root->root_key.objectid, trans->transid, level, 1)
516 * When a tree block is cow'd, new back references are added for all the
517 * blocks it points to. If the tree block isn't in reference counted root,
518 * the old back references are removed. These new back references are of
519 * the form (trans->transid will have increased since creation):
521 * (root->root_key.objectid, trans->transid, level, 1)
523 * When a backref is in deleting, the following fields are checked:
525 * if backref was for a tree root:
526 * (btrfs_header_owner(itself), btrfs_header_generation(itself), level)
528 * (btrfs_header_owner(parent), btrfs_header_generation(parent), level)
530 * Back Reference Key composing:
532 * The key objectid corresponds to the first byte in the extent, the key
533 * type is set to BTRFS_EXTENT_REF_KEY, and the key offset is the first
534 * byte of parent extent. If a extent is tree root, the key offset is set
535 * to the key objectid.
538 static int noinline lookup_extent_backref(struct btrfs_trans_handle *trans,
539 struct btrfs_root *root,
540 struct btrfs_path *path,
541 u64 bytenr, u64 parent,
542 u64 ref_root, u64 ref_generation,
543 u64 owner_objectid, int del)
545 struct btrfs_key key;
546 struct btrfs_extent_ref *ref;
547 struct extent_buffer *leaf;
551 key.objectid = bytenr;
552 key.type = BTRFS_EXTENT_REF_KEY;
555 ret = btrfs_search_slot(trans, root, &key, path, del ? -1 : 0, 1);
563 leaf = path->nodes[0];
564 ref = btrfs_item_ptr(leaf, path->slots[0], struct btrfs_extent_ref);
565 ref_objectid = btrfs_ref_objectid(leaf, ref);
566 if (btrfs_ref_root(leaf, ref) != ref_root ||
567 btrfs_ref_generation(leaf, ref) != ref_generation ||
568 (ref_objectid != owner_objectid &&
569 ref_objectid != BTRFS_MULTIPLE_OBJECTIDS)) {
579 static int noinline insert_extent_backref(struct btrfs_trans_handle *trans,
580 struct btrfs_root *root,
581 struct btrfs_path *path,
582 u64 bytenr, u64 parent,
583 u64 ref_root, u64 ref_generation,
586 struct btrfs_key key;
587 struct extent_buffer *leaf;
588 struct btrfs_extent_ref *ref;
592 key.objectid = bytenr;
593 key.type = BTRFS_EXTENT_REF_KEY;
596 ret = btrfs_insert_empty_item(trans, root, path, &key, sizeof(*ref));
598 leaf = path->nodes[0];
599 ref = btrfs_item_ptr(leaf, path->slots[0],
600 struct btrfs_extent_ref);
601 btrfs_set_ref_root(leaf, ref, ref_root);
602 btrfs_set_ref_generation(leaf, ref, ref_generation);
603 btrfs_set_ref_objectid(leaf, ref, owner_objectid);
604 btrfs_set_ref_num_refs(leaf, ref, 1);
605 } else if (ret == -EEXIST) {
607 BUG_ON(owner_objectid < BTRFS_FIRST_FREE_OBJECTID);
608 leaf = path->nodes[0];
609 ref = btrfs_item_ptr(leaf, path->slots[0],
610 struct btrfs_extent_ref);
611 if (btrfs_ref_root(leaf, ref) != ref_root ||
612 btrfs_ref_generation(leaf, ref) != ref_generation) {
618 num_refs = btrfs_ref_num_refs(leaf, ref);
619 BUG_ON(num_refs == 0);
620 btrfs_set_ref_num_refs(leaf, ref, num_refs + 1);
622 existing_owner = btrfs_ref_objectid(leaf, ref);
623 if (existing_owner != owner_objectid &&
624 existing_owner != BTRFS_MULTIPLE_OBJECTIDS) {
625 btrfs_set_ref_objectid(leaf, ref,
626 BTRFS_MULTIPLE_OBJECTIDS);
632 btrfs_mark_buffer_dirty(path->nodes[0]);
634 btrfs_release_path(root, path);
638 static int noinline remove_extent_backref(struct btrfs_trans_handle *trans,
639 struct btrfs_root *root,
640 struct btrfs_path *path)
642 struct extent_buffer *leaf;
643 struct btrfs_extent_ref *ref;
647 leaf = path->nodes[0];
648 ref = btrfs_item_ptr(leaf, path->slots[0], struct btrfs_extent_ref);
649 num_refs = btrfs_ref_num_refs(leaf, ref);
650 BUG_ON(num_refs == 0);
653 ret = btrfs_del_item(trans, root, path);
655 btrfs_set_ref_num_refs(leaf, ref, num_refs);
656 btrfs_mark_buffer_dirty(leaf);
658 btrfs_release_path(root, path);
662 static int __btrfs_update_extent_ref(struct btrfs_trans_handle *trans,
663 struct btrfs_root *root, u64 bytenr,
664 u64 orig_parent, u64 parent,
665 u64 orig_root, u64 ref_root,
666 u64 orig_generation, u64 ref_generation,
670 struct btrfs_root *extent_root = root->fs_info->extent_root;
671 struct btrfs_path *path;
673 if (root == root->fs_info->extent_root) {
674 struct pending_extent_op *extent_op;
677 BUG_ON(owner_objectid >= BTRFS_MAX_LEVEL);
678 num_bytes = btrfs_level_size(root, (int)owner_objectid);
679 if (test_range_bit(&root->fs_info->extent_ins, bytenr,
680 bytenr + num_bytes - 1, EXTENT_LOCKED, 0)) {
682 ret = get_state_private(&root->fs_info->extent_ins,
685 extent_op = (struct pending_extent_op *)
687 BUG_ON(extent_op->parent != orig_parent);
688 BUG_ON(extent_op->generation != orig_generation);
689 extent_op->parent = parent;
690 extent_op->generation = ref_generation;
692 extent_op = kmalloc(sizeof(*extent_op), GFP_NOFS);
695 extent_op->type = PENDING_BACKREF_UPDATE;
696 extent_op->bytenr = bytenr;
697 extent_op->num_bytes = num_bytes;
698 extent_op->parent = parent;
699 extent_op->orig_parent = orig_parent;
700 extent_op->generation = ref_generation;
701 extent_op->orig_generation = orig_generation;
702 extent_op->level = (int)owner_objectid;
704 set_extent_bits(&root->fs_info->extent_ins,
705 bytenr, bytenr + num_bytes - 1,
706 EXTENT_LOCKED, GFP_NOFS);
707 set_state_private(&root->fs_info->extent_ins,
708 bytenr, (unsigned long)extent_op);
713 path = btrfs_alloc_path();
716 ret = lookup_extent_backref(trans, extent_root, path,
717 bytenr, orig_parent, orig_root,
718 orig_generation, owner_objectid, 1);
721 ret = remove_extent_backref(trans, extent_root, path);
724 ret = insert_extent_backref(trans, extent_root, path, bytenr,
725 parent, ref_root, ref_generation,
728 finish_current_insert(trans, extent_root);
729 del_pending_extents(trans, extent_root);
731 btrfs_free_path(path);
735 int btrfs_update_extent_ref(struct btrfs_trans_handle *trans,
736 struct btrfs_root *root, u64 bytenr,
737 u64 orig_parent, u64 parent,
738 u64 ref_root, u64 ref_generation,
742 if (ref_root == BTRFS_TREE_LOG_OBJECTID &&
743 owner_objectid < BTRFS_FIRST_FREE_OBJECTID)
745 maybe_lock_mutex(root);
746 ret = __btrfs_update_extent_ref(trans, root, bytenr, orig_parent,
747 parent, ref_root, ref_root,
748 ref_generation, ref_generation,
750 maybe_unlock_mutex(root);
754 static int __btrfs_inc_extent_ref(struct btrfs_trans_handle *trans,
755 struct btrfs_root *root, u64 bytenr,
756 u64 orig_parent, u64 parent,
757 u64 orig_root, u64 ref_root,
758 u64 orig_generation, u64 ref_generation,
761 struct btrfs_path *path;
763 struct btrfs_key key;
764 struct extent_buffer *l;
765 struct btrfs_extent_item *item;
768 path = btrfs_alloc_path();
773 key.objectid = bytenr;
774 key.type = BTRFS_EXTENT_ITEM_KEY;
775 key.offset = (u64)-1;
777 ret = btrfs_search_slot(trans, root->fs_info->extent_root, &key, path,
781 BUG_ON(ret == 0 || path->slots[0] == 0);
786 btrfs_item_key_to_cpu(l, &key, path->slots[0]);
787 BUG_ON(key.objectid != bytenr);
788 BUG_ON(key.type != BTRFS_EXTENT_ITEM_KEY);
790 item = btrfs_item_ptr(l, path->slots[0], struct btrfs_extent_item);
791 refs = btrfs_extent_refs(l, item);
792 btrfs_set_extent_refs(l, item, refs + 1);
793 btrfs_mark_buffer_dirty(path->nodes[0]);
795 btrfs_release_path(root->fs_info->extent_root, path);
798 ret = insert_extent_backref(trans, root->fs_info->extent_root,
799 path, bytenr, parent,
800 ref_root, ref_generation,
803 finish_current_insert(trans, root->fs_info->extent_root);
804 del_pending_extents(trans, root->fs_info->extent_root);
806 btrfs_free_path(path);
810 int btrfs_inc_extent_ref(struct btrfs_trans_handle *trans,
811 struct btrfs_root *root,
812 u64 bytenr, u64 num_bytes, u64 parent,
813 u64 ref_root, u64 ref_generation,
817 if (ref_root == BTRFS_TREE_LOG_OBJECTID &&
818 owner_objectid < BTRFS_FIRST_FREE_OBJECTID)
820 maybe_lock_mutex(root);
821 ret = __btrfs_inc_extent_ref(trans, root, bytenr, 0, parent,
822 0, ref_root, 0, ref_generation,
824 maybe_unlock_mutex(root);
828 int btrfs_extent_post_op(struct btrfs_trans_handle *trans,
829 struct btrfs_root *root)
831 finish_current_insert(trans, root->fs_info->extent_root);
832 del_pending_extents(trans, root->fs_info->extent_root);
836 int btrfs_lookup_extent_ref(struct btrfs_trans_handle *trans,
837 struct btrfs_root *root, u64 bytenr,
838 u64 num_bytes, u32 *refs)
840 struct btrfs_path *path;
842 struct btrfs_key key;
843 struct extent_buffer *l;
844 struct btrfs_extent_item *item;
846 WARN_ON(num_bytes < root->sectorsize);
847 path = btrfs_alloc_path();
849 key.objectid = bytenr;
850 key.offset = num_bytes;
851 btrfs_set_key_type(&key, BTRFS_EXTENT_ITEM_KEY);
852 ret = btrfs_search_slot(trans, root->fs_info->extent_root, &key, path,
857 btrfs_print_leaf(root, path->nodes[0]);
858 printk("failed to find block number %Lu\n", bytenr);
862 item = btrfs_item_ptr(l, path->slots[0], struct btrfs_extent_item);
863 *refs = btrfs_extent_refs(l, item);
865 btrfs_free_path(path);
869 static int get_reference_status(struct btrfs_root *root, u64 bytenr,
870 u64 parent_gen, u64 ref_objectid,
871 u64 *min_generation, u32 *ref_count)
873 struct btrfs_root *extent_root = root->fs_info->extent_root;
874 struct btrfs_path *path;
875 struct extent_buffer *leaf;
876 struct btrfs_extent_ref *ref_item;
877 struct btrfs_key key;
878 struct btrfs_key found_key;
879 u64 root_objectid = root->root_key.objectid;
884 key.objectid = bytenr;
885 key.offset = (u64)-1;
886 key.type = BTRFS_EXTENT_ITEM_KEY;
888 path = btrfs_alloc_path();
889 mutex_lock(&root->fs_info->alloc_mutex);
890 ret = btrfs_search_slot(NULL, extent_root, &key, path, 0, 0);
894 if (ret < 0 || path->slots[0] == 0)
898 leaf = path->nodes[0];
899 btrfs_item_key_to_cpu(leaf, &found_key, path->slots[0]);
901 if (found_key.objectid != bytenr ||
902 found_key.type != BTRFS_EXTENT_ITEM_KEY) {
908 *min_generation = (u64)-1;
911 leaf = path->nodes[0];
912 nritems = btrfs_header_nritems(leaf);
913 if (path->slots[0] >= nritems) {
914 ret = btrfs_next_leaf(extent_root, path);
921 btrfs_item_key_to_cpu(leaf, &found_key, path->slots[0]);
922 if (found_key.objectid != bytenr)
925 if (found_key.type != BTRFS_EXTENT_REF_KEY) {
930 ref_item = btrfs_item_ptr(leaf, path->slots[0],
931 struct btrfs_extent_ref);
932 ref_generation = btrfs_ref_generation(leaf, ref_item);
934 * For (parent_gen > 0 && parent_gen > ref_generation):
936 * we reach here through the oldest root, therefore
937 * all other reference from same snapshot should have
938 * a larger generation.
940 if ((root_objectid != btrfs_ref_root(leaf, ref_item)) ||
941 (parent_gen > 0 && parent_gen > ref_generation) ||
942 (ref_objectid >= BTRFS_FIRST_FREE_OBJECTID &&
943 ref_objectid != btrfs_ref_objectid(leaf, ref_item))) {
949 if (*min_generation > ref_generation)
950 *min_generation = ref_generation;
956 mutex_unlock(&root->fs_info->alloc_mutex);
957 btrfs_free_path(path);
961 int btrfs_cross_ref_exists(struct btrfs_trans_handle *trans,
962 struct btrfs_root *root,
963 struct btrfs_key *key, u64 bytenr)
965 struct btrfs_root *old_root;
966 struct btrfs_path *path = NULL;
967 struct extent_buffer *eb;
968 struct btrfs_file_extent_item *item;
976 BUG_ON(trans == NULL);
977 BUG_ON(key->type != BTRFS_EXTENT_DATA_KEY);
978 ret = get_reference_status(root, bytenr, 0, key->objectid,
979 &min_generation, &ref_count);
986 old_root = root->dirty_root->root;
987 ref_generation = old_root->root_key.offset;
989 /* all references are created in running transaction */
990 if (min_generation > ref_generation) {
995 path = btrfs_alloc_path();
1001 path->skip_locking = 1;
1002 /* if no item found, the extent is referenced by other snapshot */
1003 ret = btrfs_search_slot(NULL, old_root, key, path, 0, 0);
1007 eb = path->nodes[0];
1008 item = btrfs_item_ptr(eb, path->slots[0],
1009 struct btrfs_file_extent_item);
1010 if (btrfs_file_extent_type(eb, item) != BTRFS_FILE_EXTENT_REG ||
1011 btrfs_file_extent_disk_bytenr(eb, item) != bytenr) {
1016 for (level = BTRFS_MAX_LEVEL - 1; level >= -1; level--) {
1018 eb = path->nodes[level];
1021 extent_start = eb->start;
1023 extent_start = bytenr;
1025 ret = get_reference_status(root, extent_start, ref_generation,
1026 0, &min_generation, &ref_count);
1030 if (ref_count != 1) {
1035 ref_generation = btrfs_header_generation(eb);
1040 btrfs_free_path(path);
1044 int btrfs_cache_ref(struct btrfs_trans_handle *trans, struct btrfs_root *root,
1045 struct extent_buffer *buf, u32 nr_extents)
1047 struct btrfs_key key;
1048 struct btrfs_file_extent_item *fi;
1056 if (!root->ref_cows)
1059 if (root->root_key.objectid != BTRFS_TREE_RELOC_OBJECTID) {
1061 root_gen = root->root_key.offset;
1064 root_gen = trans->transid - 1;
1067 level = btrfs_header_level(buf);
1068 nritems = btrfs_header_nritems(buf);
1071 struct btrfs_leaf_ref *ref;
1072 struct btrfs_extent_info *info;
1074 ref = btrfs_alloc_leaf_ref(root, nr_extents);
1080 ref->root_gen = root_gen;
1081 ref->bytenr = buf->start;
1082 ref->owner = btrfs_header_owner(buf);
1083 ref->generation = btrfs_header_generation(buf);
1084 ref->nritems = nr_extents;
1085 info = ref->extents;
1087 for (i = 0; nr_extents > 0 && i < nritems; i++) {
1089 btrfs_item_key_to_cpu(buf, &key, i);
1090 if (btrfs_key_type(&key) != BTRFS_EXTENT_DATA_KEY)
1092 fi = btrfs_item_ptr(buf, i,
1093 struct btrfs_file_extent_item);
1094 if (btrfs_file_extent_type(buf, fi) ==
1095 BTRFS_FILE_EXTENT_INLINE)
1097 disk_bytenr = btrfs_file_extent_disk_bytenr(buf, fi);
1098 if (disk_bytenr == 0)
1101 info->bytenr = disk_bytenr;
1103 btrfs_file_extent_disk_num_bytes(buf, fi);
1104 info->objectid = key.objectid;
1105 info->offset = key.offset;
1109 ret = btrfs_add_leaf_ref(root, ref, shared);
1110 if (ret == -EEXIST && shared) {
1111 struct btrfs_leaf_ref *old;
1112 old = btrfs_lookup_leaf_ref(root, ref->bytenr);
1114 btrfs_remove_leaf_ref(root, old);
1115 btrfs_free_leaf_ref(root, old);
1116 ret = btrfs_add_leaf_ref(root, ref, shared);
1119 btrfs_free_leaf_ref(root, ref);
1125 int btrfs_inc_ref(struct btrfs_trans_handle *trans, struct btrfs_root *root,
1126 struct extent_buffer *orig_buf, struct extent_buffer *buf,
1133 u64 orig_generation;
1135 u32 nr_file_extents = 0;
1136 struct btrfs_key key;
1137 struct btrfs_file_extent_item *fi;
1142 int (*process_func)(struct btrfs_trans_handle *, struct btrfs_root *,
1143 u64, u64, u64, u64, u64, u64, u64, u64);
1145 ref_root = btrfs_header_owner(buf);
1146 ref_generation = btrfs_header_generation(buf);
1147 orig_root = btrfs_header_owner(orig_buf);
1148 orig_generation = btrfs_header_generation(orig_buf);
1150 nritems = btrfs_header_nritems(buf);
1151 level = btrfs_header_level(buf);
1153 if (root->ref_cows) {
1154 process_func = __btrfs_inc_extent_ref;
1157 root->root_key.objectid != BTRFS_TREE_LOG_OBJECTID)
1160 root->root_key.objectid == BTRFS_TREE_LOG_OBJECTID)
1162 process_func = __btrfs_update_extent_ref;
1165 for (i = 0; i < nritems; i++) {
1168 btrfs_item_key_to_cpu(buf, &key, i);
1169 if (btrfs_key_type(&key) != BTRFS_EXTENT_DATA_KEY)
1171 fi = btrfs_item_ptr(buf, i,
1172 struct btrfs_file_extent_item);
1173 if (btrfs_file_extent_type(buf, fi) ==
1174 BTRFS_FILE_EXTENT_INLINE)
1176 bytenr = btrfs_file_extent_disk_bytenr(buf, fi);
1182 maybe_lock_mutex(root);
1183 ret = process_func(trans, root, bytenr,
1184 orig_buf->start, buf->start,
1185 orig_root, ref_root,
1186 orig_generation, ref_generation,
1188 maybe_unlock_mutex(root);
1196 bytenr = btrfs_node_blockptr(buf, i);
1197 maybe_lock_mutex(root);
1198 ret = process_func(trans, root, bytenr,
1199 orig_buf->start, buf->start,
1200 orig_root, ref_root,
1201 orig_generation, ref_generation,
1203 maybe_unlock_mutex(root);
1214 *nr_extents = nr_file_extents;
1216 *nr_extents = nritems;
1224 int btrfs_update_ref(struct btrfs_trans_handle *trans,
1225 struct btrfs_root *root, struct extent_buffer *orig_buf,
1226 struct extent_buffer *buf, int start_slot, int nr)
1233 u64 orig_generation;
1234 struct btrfs_key key;
1235 struct btrfs_file_extent_item *fi;
1241 BUG_ON(start_slot < 0);
1242 BUG_ON(start_slot + nr > btrfs_header_nritems(buf));
1244 ref_root = btrfs_header_owner(buf);
1245 ref_generation = btrfs_header_generation(buf);
1246 orig_root = btrfs_header_owner(orig_buf);
1247 orig_generation = btrfs_header_generation(orig_buf);
1248 level = btrfs_header_level(buf);
1250 if (!root->ref_cows) {
1252 root->root_key.objectid != BTRFS_TREE_LOG_OBJECTID)
1255 root->root_key.objectid == BTRFS_TREE_LOG_OBJECTID)
1259 for (i = 0, slot = start_slot; i < nr; i++, slot++) {
1262 btrfs_item_key_to_cpu(buf, &key, slot);
1263 if (btrfs_key_type(&key) != BTRFS_EXTENT_DATA_KEY)
1265 fi = btrfs_item_ptr(buf, slot,
1266 struct btrfs_file_extent_item);
1267 if (btrfs_file_extent_type(buf, fi) ==
1268 BTRFS_FILE_EXTENT_INLINE)
1270 bytenr = btrfs_file_extent_disk_bytenr(buf, fi);
1273 maybe_lock_mutex(root);
1274 ret = __btrfs_update_extent_ref(trans, root, bytenr,
1275 orig_buf->start, buf->start,
1276 orig_root, ref_root,
1277 orig_generation, ref_generation,
1279 maybe_unlock_mutex(root);
1283 bytenr = btrfs_node_blockptr(buf, slot);
1284 maybe_lock_mutex(root);
1285 ret = __btrfs_update_extent_ref(trans, root, bytenr,
1286 orig_buf->start, buf->start,
1287 orig_root, ref_root,
1288 orig_generation, ref_generation,
1290 maybe_unlock_mutex(root);
1301 static int write_one_cache_group(struct btrfs_trans_handle *trans,
1302 struct btrfs_root *root,
1303 struct btrfs_path *path,
1304 struct btrfs_block_group_cache *cache)
1308 struct btrfs_root *extent_root = root->fs_info->extent_root;
1310 struct extent_buffer *leaf;
1312 ret = btrfs_search_slot(trans, extent_root, &cache->key, path, 0, 1);
1317 leaf = path->nodes[0];
1318 bi = btrfs_item_ptr_offset(leaf, path->slots[0]);
1319 write_extent_buffer(leaf, &cache->item, bi, sizeof(cache->item));
1320 btrfs_mark_buffer_dirty(leaf);
1321 btrfs_release_path(extent_root, path);
1323 finish_current_insert(trans, extent_root);
1324 pending_ret = del_pending_extents(trans, extent_root);
1333 int btrfs_write_dirty_block_groups(struct btrfs_trans_handle *trans,
1334 struct btrfs_root *root)
1336 struct btrfs_block_group_cache *cache, *entry;
1340 struct btrfs_path *path;
1343 path = btrfs_alloc_path();
1347 mutex_lock(&root->fs_info->alloc_mutex);
1350 spin_lock(&root->fs_info->block_group_cache_lock);
1351 for (n = rb_first(&root->fs_info->block_group_cache_tree);
1352 n; n = rb_next(n)) {
1353 entry = rb_entry(n, struct btrfs_block_group_cache,
1360 spin_unlock(&root->fs_info->block_group_cache_lock);
1366 last += cache->key.offset;
1368 err = write_one_cache_group(trans, root,
1371 * if we fail to write the cache group, we want
1372 * to keep it marked dirty in hopes that a later
1380 btrfs_free_path(path);
1381 mutex_unlock(&root->fs_info->alloc_mutex);
1385 static int update_space_info(struct btrfs_fs_info *info, u64 flags,
1386 u64 total_bytes, u64 bytes_used,
1387 struct btrfs_space_info **space_info)
1389 struct btrfs_space_info *found;
1391 found = __find_space_info(info, flags);
1393 found->total_bytes += total_bytes;
1394 found->bytes_used += bytes_used;
1396 *space_info = found;
1399 found = kmalloc(sizeof(*found), GFP_NOFS);
1403 list_add(&found->list, &info->space_info);
1404 INIT_LIST_HEAD(&found->block_groups);
1405 init_rwsem(&found->groups_sem);
1406 spin_lock_init(&found->lock);
1407 found->flags = flags;
1408 found->total_bytes = total_bytes;
1409 found->bytes_used = bytes_used;
1410 found->bytes_pinned = 0;
1411 found->bytes_reserved = 0;
1413 found->force_alloc = 0;
1414 *space_info = found;
1418 static void set_avail_alloc_bits(struct btrfs_fs_info *fs_info, u64 flags)
1420 u64 extra_flags = flags & (BTRFS_BLOCK_GROUP_RAID0 |
1421 BTRFS_BLOCK_GROUP_RAID1 |
1422 BTRFS_BLOCK_GROUP_RAID10 |
1423 BTRFS_BLOCK_GROUP_DUP);
1425 if (flags & BTRFS_BLOCK_GROUP_DATA)
1426 fs_info->avail_data_alloc_bits |= extra_flags;
1427 if (flags & BTRFS_BLOCK_GROUP_METADATA)
1428 fs_info->avail_metadata_alloc_bits |= extra_flags;
1429 if (flags & BTRFS_BLOCK_GROUP_SYSTEM)
1430 fs_info->avail_system_alloc_bits |= extra_flags;
1434 static u64 reduce_alloc_profile(struct btrfs_root *root, u64 flags)
1436 u64 num_devices = root->fs_info->fs_devices->num_devices;
1438 if (num_devices == 1)
1439 flags &= ~(BTRFS_BLOCK_GROUP_RAID1 | BTRFS_BLOCK_GROUP_RAID0);
1440 if (num_devices < 4)
1441 flags &= ~BTRFS_BLOCK_GROUP_RAID10;
1443 if ((flags & BTRFS_BLOCK_GROUP_DUP) &&
1444 (flags & (BTRFS_BLOCK_GROUP_RAID1 |
1445 BTRFS_BLOCK_GROUP_RAID10))) {
1446 flags &= ~BTRFS_BLOCK_GROUP_DUP;
1449 if ((flags & BTRFS_BLOCK_GROUP_RAID1) &&
1450 (flags & BTRFS_BLOCK_GROUP_RAID10)) {
1451 flags &= ~BTRFS_BLOCK_GROUP_RAID1;
1454 if ((flags & BTRFS_BLOCK_GROUP_RAID0) &&
1455 ((flags & BTRFS_BLOCK_GROUP_RAID1) |
1456 (flags & BTRFS_BLOCK_GROUP_RAID10) |
1457 (flags & BTRFS_BLOCK_GROUP_DUP)))
1458 flags &= ~BTRFS_BLOCK_GROUP_RAID0;
1462 static int do_chunk_alloc(struct btrfs_trans_handle *trans,
1463 struct btrfs_root *extent_root, u64 alloc_bytes,
1464 u64 flags, int force)
1466 struct btrfs_space_info *space_info;
1470 int ret = 0, waited = 0;
1472 flags = reduce_alloc_profile(extent_root, flags);
1474 space_info = __find_space_info(extent_root->fs_info, flags);
1476 ret = update_space_info(extent_root->fs_info, flags,
1480 BUG_ON(!space_info);
1482 if (space_info->force_alloc) {
1484 space_info->force_alloc = 0;
1486 if (space_info->full)
1489 thresh = div_factor(space_info->total_bytes, 6);
1491 (space_info->bytes_used + space_info->bytes_pinned +
1492 space_info->bytes_reserved + alloc_bytes) < thresh)
1495 while (!mutex_trylock(&extent_root->fs_info->chunk_mutex)) {
1498 mutex_unlock(&extent_root->fs_info->alloc_mutex);
1500 mutex_lock(&extent_root->fs_info->alloc_mutex);
1504 if (waited && space_info->full)
1507 ret = btrfs_alloc_chunk(trans, extent_root, &start, &num_bytes, flags);
1508 if (ret == -ENOSPC) {
1509 printk("space info full %Lu\n", flags);
1510 space_info->full = 1;
1515 ret = btrfs_make_block_group(trans, extent_root, 0, flags,
1516 BTRFS_FIRST_CHUNK_TREE_OBJECTID, start, num_bytes);
1520 mutex_unlock(&extent_root->fs_info->chunk_mutex);
1525 static int update_block_group(struct btrfs_trans_handle *trans,
1526 struct btrfs_root *root,
1527 u64 bytenr, u64 num_bytes, int alloc,
1530 struct btrfs_block_group_cache *cache;
1531 struct btrfs_fs_info *info = root->fs_info;
1532 u64 total = num_bytes;
1536 WARN_ON(!mutex_is_locked(&root->fs_info->alloc_mutex));
1538 cache = btrfs_lookup_block_group(info, bytenr);
1542 byte_in_group = bytenr - cache->key.objectid;
1543 WARN_ON(byte_in_group > cache->key.offset);
1545 spin_lock(&cache->lock);
1547 old_val = btrfs_block_group_used(&cache->item);
1548 num_bytes = min(total, cache->key.offset - byte_in_group);
1550 old_val += num_bytes;
1551 cache->space_info->bytes_used += num_bytes;
1552 btrfs_set_block_group_used(&cache->item, old_val);
1553 spin_unlock(&cache->lock);
1555 old_val -= num_bytes;
1556 cache->space_info->bytes_used -= num_bytes;
1557 btrfs_set_block_group_used(&cache->item, old_val);
1558 spin_unlock(&cache->lock);
1561 ret = btrfs_add_free_space(cache, bytenr,
1568 bytenr += num_bytes;
1573 static u64 first_logical_byte(struct btrfs_root *root, u64 search_start)
1575 struct btrfs_block_group_cache *cache;
1577 cache = btrfs_lookup_first_block_group(root->fs_info, search_start);
1581 return cache->key.objectid;
1584 int btrfs_update_pinned_extents(struct btrfs_root *root,
1585 u64 bytenr, u64 num, int pin)
1588 struct btrfs_block_group_cache *cache;
1589 struct btrfs_fs_info *fs_info = root->fs_info;
1591 WARN_ON(!mutex_is_locked(&root->fs_info->alloc_mutex));
1593 set_extent_dirty(&fs_info->pinned_extents,
1594 bytenr, bytenr + num - 1, GFP_NOFS);
1596 clear_extent_dirty(&fs_info->pinned_extents,
1597 bytenr, bytenr + num - 1, GFP_NOFS);
1600 cache = btrfs_lookup_block_group(fs_info, bytenr);
1602 len = min(num, cache->key.offset -
1603 (bytenr - cache->key.objectid));
1605 spin_lock(&cache->lock);
1606 cache->pinned += len;
1607 cache->space_info->bytes_pinned += len;
1608 spin_unlock(&cache->lock);
1609 fs_info->total_pinned += len;
1611 spin_lock(&cache->lock);
1612 cache->pinned -= len;
1613 cache->space_info->bytes_pinned -= len;
1614 spin_unlock(&cache->lock);
1615 fs_info->total_pinned -= len;
1623 static int update_reserved_extents(struct btrfs_root *root,
1624 u64 bytenr, u64 num, int reserve)
1627 struct btrfs_block_group_cache *cache;
1628 struct btrfs_fs_info *fs_info = root->fs_info;
1630 WARN_ON(!mutex_is_locked(&root->fs_info->alloc_mutex));
1632 cache = btrfs_lookup_block_group(fs_info, bytenr);
1634 len = min(num, cache->key.offset -
1635 (bytenr - cache->key.objectid));
1637 spin_lock(&cache->lock);
1638 cache->reserved += len;
1639 cache->space_info->bytes_reserved += len;
1640 spin_unlock(&cache->lock);
1642 spin_lock(&cache->lock);
1643 cache->reserved -= len;
1644 cache->space_info->bytes_reserved -= len;
1645 spin_unlock(&cache->lock);
1653 int btrfs_copy_pinned(struct btrfs_root *root, struct extent_io_tree *copy)
1658 struct extent_io_tree *pinned_extents = &root->fs_info->pinned_extents;
1662 ret = find_first_extent_bit(pinned_extents, last,
1663 &start, &end, EXTENT_DIRTY);
1666 set_extent_dirty(copy, start, end, GFP_NOFS);
1672 int btrfs_finish_extent_commit(struct btrfs_trans_handle *trans,
1673 struct btrfs_root *root,
1674 struct extent_io_tree *unpin)
1679 struct btrfs_block_group_cache *cache;
1681 mutex_lock(&root->fs_info->alloc_mutex);
1683 ret = find_first_extent_bit(unpin, 0, &start, &end,
1687 btrfs_update_pinned_extents(root, start, end + 1 - start, 0);
1688 clear_extent_dirty(unpin, start, end, GFP_NOFS);
1689 cache = btrfs_lookup_block_group(root->fs_info, start);
1691 btrfs_add_free_space(cache, start, end - start + 1);
1692 if (need_resched()) {
1693 mutex_unlock(&root->fs_info->alloc_mutex);
1695 mutex_lock(&root->fs_info->alloc_mutex);
1698 mutex_unlock(&root->fs_info->alloc_mutex);
1702 static int finish_current_insert(struct btrfs_trans_handle *trans,
1703 struct btrfs_root *extent_root)
1708 struct btrfs_fs_info *info = extent_root->fs_info;
1709 struct btrfs_path *path;
1710 struct btrfs_extent_ref *ref;
1711 struct pending_extent_op *extent_op;
1712 struct btrfs_key key;
1713 struct btrfs_extent_item extent_item;
1717 WARN_ON(!mutex_is_locked(&extent_root->fs_info->alloc_mutex));
1718 btrfs_set_stack_extent_refs(&extent_item, 1);
1719 path = btrfs_alloc_path();
1722 ret = find_first_extent_bit(&info->extent_ins, 0, &start,
1723 &end, EXTENT_LOCKED);
1727 ret = get_state_private(&info->extent_ins, start, &priv);
1729 extent_op = (struct pending_extent_op *)(unsigned long)priv;
1731 if (extent_op->type == PENDING_EXTENT_INSERT) {
1732 key.objectid = start;
1733 key.offset = end + 1 - start;
1734 key.type = BTRFS_EXTENT_ITEM_KEY;
1735 err = btrfs_insert_item(trans, extent_root, &key,
1736 &extent_item, sizeof(extent_item));
1739 clear_extent_bits(&info->extent_ins, start, end,
1740 EXTENT_LOCKED, GFP_NOFS);
1742 err = insert_extent_backref(trans, extent_root, path,
1743 start, extent_op->parent,
1744 extent_root->root_key.objectid,
1745 extent_op->generation,
1748 } else if (extent_op->type == PENDING_BACKREF_UPDATE) {
1749 err = lookup_extent_backref(trans, extent_root, path,
1750 start, extent_op->orig_parent,
1751 extent_root->root_key.objectid,
1752 extent_op->orig_generation,
1753 extent_op->level, 0);
1756 clear_extent_bits(&info->extent_ins, start, end,
1757 EXTENT_LOCKED, GFP_NOFS);
1759 key.objectid = start;
1760 key.offset = extent_op->parent;
1761 key.type = BTRFS_EXTENT_REF_KEY;
1762 err = btrfs_set_item_key_safe(trans, extent_root, path,
1765 ref = btrfs_item_ptr(path->nodes[0], path->slots[0],
1766 struct btrfs_extent_ref);
1767 btrfs_set_ref_generation(path->nodes[0], ref,
1768 extent_op->generation);
1769 btrfs_mark_buffer_dirty(path->nodes[0]);
1770 btrfs_release_path(extent_root, path);
1776 if (need_resched()) {
1777 mutex_unlock(&extent_root->fs_info->alloc_mutex);
1779 mutex_lock(&extent_root->fs_info->alloc_mutex);
1782 btrfs_free_path(path);
1786 static int pin_down_bytes(struct btrfs_trans_handle *trans,
1787 struct btrfs_root *root,
1788 u64 bytenr, u64 num_bytes, int is_data)
1791 struct extent_buffer *buf;
1793 WARN_ON(!mutex_is_locked(&root->fs_info->alloc_mutex));
1797 buf = btrfs_find_tree_block(root, bytenr, num_bytes);
1801 /* we can reuse a block if it hasn't been written
1802 * and it is from this transaction. We can't
1803 * reuse anything from the tree log root because
1804 * it has tiny sub-transactions.
1806 if (btrfs_buffer_uptodate(buf, 0) &&
1807 btrfs_try_tree_lock(buf)) {
1808 u64 header_owner = btrfs_header_owner(buf);
1809 u64 header_transid = btrfs_header_generation(buf);
1810 if (header_owner != BTRFS_TREE_LOG_OBJECTID &&
1811 header_owner != BTRFS_TREE_RELOC_OBJECTID &&
1812 header_transid == trans->transid &&
1813 !btrfs_header_flag(buf, BTRFS_HEADER_FLAG_WRITTEN)) {
1814 clean_tree_block(NULL, root, buf);
1815 btrfs_tree_unlock(buf);
1816 free_extent_buffer(buf);
1819 btrfs_tree_unlock(buf);
1821 free_extent_buffer(buf);
1823 btrfs_update_pinned_extents(root, bytenr, num_bytes, 1);
1830 * remove an extent from the root, returns 0 on success
1832 static int __free_extent(struct btrfs_trans_handle *trans,
1833 struct btrfs_root *root,
1834 u64 bytenr, u64 num_bytes, u64 parent,
1835 u64 root_objectid, u64 ref_generation,
1836 u64 owner_objectid, int pin, int mark_free)
1838 struct btrfs_path *path;
1839 struct btrfs_key key;
1840 struct btrfs_fs_info *info = root->fs_info;
1841 struct btrfs_root *extent_root = info->extent_root;
1842 struct extent_buffer *leaf;
1844 int extent_slot = 0;
1845 int found_extent = 0;
1847 struct btrfs_extent_item *ei;
1850 WARN_ON(!mutex_is_locked(&root->fs_info->alloc_mutex));
1851 key.objectid = bytenr;
1852 btrfs_set_key_type(&key, BTRFS_EXTENT_ITEM_KEY);
1853 key.offset = num_bytes;
1854 path = btrfs_alloc_path();
1859 ret = lookup_extent_backref(trans, extent_root, path,
1860 bytenr, parent, root_objectid,
1861 ref_generation, owner_objectid, 1);
1863 struct btrfs_key found_key;
1864 extent_slot = path->slots[0];
1865 while(extent_slot > 0) {
1867 btrfs_item_key_to_cpu(path->nodes[0], &found_key,
1869 if (found_key.objectid != bytenr)
1871 if (found_key.type == BTRFS_EXTENT_ITEM_KEY &&
1872 found_key.offset == num_bytes) {
1876 if (path->slots[0] - extent_slot > 5)
1879 if (!found_extent) {
1880 ret = remove_extent_backref(trans, extent_root, path);
1882 btrfs_release_path(extent_root, path);
1883 ret = btrfs_search_slot(trans, extent_root,
1886 extent_slot = path->slots[0];
1889 btrfs_print_leaf(extent_root, path->nodes[0]);
1891 printk("Unable to find ref byte nr %Lu root %Lu "
1892 "gen %Lu owner %Lu\n", bytenr,
1893 root_objectid, ref_generation, owner_objectid);
1896 leaf = path->nodes[0];
1897 ei = btrfs_item_ptr(leaf, extent_slot,
1898 struct btrfs_extent_item);
1899 refs = btrfs_extent_refs(leaf, ei);
1902 btrfs_set_extent_refs(leaf, ei, refs);
1904 btrfs_mark_buffer_dirty(leaf);
1906 if (refs == 0 && found_extent && path->slots[0] == extent_slot + 1) {
1907 struct btrfs_extent_ref *ref;
1908 ref = btrfs_item_ptr(leaf, path->slots[0],
1909 struct btrfs_extent_ref);
1910 BUG_ON(btrfs_ref_num_refs(leaf, ref) != 1);
1911 /* if the back ref and the extent are next to each other
1912 * they get deleted below in one shot
1914 path->slots[0] = extent_slot;
1916 } else if (found_extent) {
1917 /* otherwise delete the extent back ref */
1918 ret = remove_extent_backref(trans, extent_root, path);
1920 /* if refs are 0, we need to setup the path for deletion */
1922 btrfs_release_path(extent_root, path);
1923 ret = btrfs_search_slot(trans, extent_root, &key, path,
1932 #ifdef BIO_RW_DISCARD
1933 u64 map_length = num_bytes;
1934 struct btrfs_multi_bio *multi = NULL;
1938 ret = pin_down_bytes(trans, root, bytenr, num_bytes,
1939 owner_objectid >= BTRFS_FIRST_FREE_OBJECTID);
1945 /* block accounting for super block */
1946 spin_lock_irq(&info->delalloc_lock);
1947 super_used = btrfs_super_bytes_used(&info->super_copy);
1948 btrfs_set_super_bytes_used(&info->super_copy,
1949 super_used - num_bytes);
1950 spin_unlock_irq(&info->delalloc_lock);
1952 /* block accounting for root item */
1953 root_used = btrfs_root_used(&root->root_item);
1954 btrfs_set_root_used(&root->root_item,
1955 root_used - num_bytes);
1956 ret = btrfs_del_items(trans, extent_root, path, path->slots[0],
1959 ret = update_block_group(trans, root, bytenr, num_bytes, 0,
1963 #ifdef BIO_RW_DISCARD
1964 /* Tell the block device(s) that the sectors can be discarded */
1965 ret = btrfs_map_block(&root->fs_info->mapping_tree, READ,
1966 bytenr, &map_length, &multi, 0);
1968 struct btrfs_bio_stripe *stripe = multi->stripes;
1971 if (map_length > num_bytes)
1972 map_length = num_bytes;
1974 for (i = 0; i < multi->num_stripes; i++, stripe++) {
1975 blkdev_issue_discard(stripe->dev->bdev,
1976 stripe->physical >> 9,
1983 btrfs_free_path(path);
1984 finish_current_insert(trans, extent_root);
1989 * find all the blocks marked as pending in the radix tree and remove
1990 * them from the extent map
1992 static int del_pending_extents(struct btrfs_trans_handle *trans, struct
1993 btrfs_root *extent_root)
2001 struct extent_io_tree *pending_del;
2002 struct extent_io_tree *extent_ins;
2003 struct pending_extent_op *extent_op;
2005 WARN_ON(!mutex_is_locked(&extent_root->fs_info->alloc_mutex));
2006 extent_ins = &extent_root->fs_info->extent_ins;
2007 pending_del = &extent_root->fs_info->pending_del;
2010 ret = find_first_extent_bit(pending_del, 0, &start, &end,
2015 ret = get_state_private(pending_del, start, &priv);
2017 extent_op = (struct pending_extent_op *)(unsigned long)priv;
2019 clear_extent_bits(pending_del, start, end, EXTENT_LOCKED,
2022 ret = pin_down_bytes(trans, extent_root, start,
2023 end + 1 - start, 0);
2024 mark_free = ret > 0;
2025 if (!test_range_bit(extent_ins, start, end,
2026 EXTENT_LOCKED, 0)) {
2028 ret = __free_extent(trans, extent_root,
2029 start, end + 1 - start,
2030 extent_op->orig_parent,
2031 extent_root->root_key.objectid,
2032 extent_op->orig_generation,
2033 extent_op->level, 0, mark_free);
2037 ret = get_state_private(extent_ins, start, &priv);
2039 extent_op = (struct pending_extent_op *)
2040 (unsigned long)priv;
2042 clear_extent_bits(extent_ins, start, end,
2043 EXTENT_LOCKED, GFP_NOFS);
2045 if (extent_op->type == PENDING_BACKREF_UPDATE)
2048 ret = update_block_group(trans, extent_root, start,
2049 end + 1 - start, 0, mark_free);
2056 if (need_resched()) {
2057 mutex_unlock(&extent_root->fs_info->alloc_mutex);
2059 mutex_lock(&extent_root->fs_info->alloc_mutex);
2066 * remove an extent from the root, returns 0 on success
2068 static int __btrfs_free_extent(struct btrfs_trans_handle *trans,
2069 struct btrfs_root *root,
2070 u64 bytenr, u64 num_bytes, u64 parent,
2071 u64 root_objectid, u64 ref_generation,
2072 u64 owner_objectid, int pin)
2074 struct btrfs_root *extent_root = root->fs_info->extent_root;
2078 WARN_ON(num_bytes < root->sectorsize);
2079 if (root == extent_root) {
2080 struct pending_extent_op *extent_op;
2082 extent_op = kmalloc(sizeof(*extent_op), GFP_NOFS);
2085 extent_op->type = PENDING_EXTENT_DELETE;
2086 extent_op->bytenr = bytenr;
2087 extent_op->num_bytes = num_bytes;
2088 extent_op->parent = parent;
2089 extent_op->orig_parent = parent;
2090 extent_op->generation = ref_generation;
2091 extent_op->orig_generation = ref_generation;
2092 extent_op->level = (int)owner_objectid;
2094 set_extent_bits(&root->fs_info->pending_del,
2095 bytenr, bytenr + num_bytes - 1,
2096 EXTENT_LOCKED, GFP_NOFS);
2097 set_state_private(&root->fs_info->pending_del,
2098 bytenr, (unsigned long)extent_op);
2101 /* if metadata always pin */
2102 if (owner_objectid < BTRFS_FIRST_FREE_OBJECTID) {
2103 if (root->root_key.objectid == BTRFS_TREE_LOG_OBJECTID) {
2104 struct btrfs_block_group_cache *cache;
2106 /* btrfs_free_reserved_extent */
2107 cache = btrfs_lookup_block_group(root->fs_info, bytenr);
2109 btrfs_add_free_space(cache, bytenr, num_bytes);
2110 update_reserved_extents(root, bytenr, num_bytes, 0);
2116 /* if data pin when any transaction has committed this */
2117 if (ref_generation != trans->transid)
2120 ret = __free_extent(trans, root, bytenr, num_bytes, parent,
2121 root_objectid, ref_generation,
2122 owner_objectid, pin, pin == 0);
2124 finish_current_insert(trans, root->fs_info->extent_root);
2125 pending_ret = del_pending_extents(trans, root->fs_info->extent_root);
2126 return ret ? ret : pending_ret;
2129 int btrfs_free_extent(struct btrfs_trans_handle *trans,
2130 struct btrfs_root *root,
2131 u64 bytenr, u64 num_bytes, u64 parent,
2132 u64 root_objectid, u64 ref_generation,
2133 u64 owner_objectid, int pin)
2137 maybe_lock_mutex(root);
2138 ret = __btrfs_free_extent(trans, root, bytenr, num_bytes, parent,
2139 root_objectid, ref_generation,
2140 owner_objectid, pin);
2141 maybe_unlock_mutex(root);
2145 static u64 stripe_align(struct btrfs_root *root, u64 val)
2147 u64 mask = ((u64)root->stripesize - 1);
2148 u64 ret = (val + mask) & ~mask;
2153 * walks the btree of allocated extents and find a hole of a given size.
2154 * The key ins is changed to record the hole:
2155 * ins->objectid == block start
2156 * ins->flags = BTRFS_EXTENT_ITEM_KEY
2157 * ins->offset == number of blocks
2158 * Any available blocks before search_start are skipped.
2160 static int noinline find_free_extent(struct btrfs_trans_handle *trans,
2161 struct btrfs_root *orig_root,
2162 u64 num_bytes, u64 empty_size,
2163 u64 search_start, u64 search_end,
2164 u64 hint_byte, struct btrfs_key *ins,
2165 u64 exclude_start, u64 exclude_nr,
2169 struct btrfs_root * root = orig_root->fs_info->extent_root;
2170 u64 total_needed = num_bytes;
2171 u64 *last_ptr = NULL;
2172 struct btrfs_block_group_cache *block_group = NULL;
2173 int chunk_alloc_done = 0;
2174 int empty_cluster = 2 * 1024 * 1024;
2175 int allowed_chunk_alloc = 0;
2176 struct list_head *head = NULL, *cur = NULL;
2178 struct btrfs_space_info *space_info;
2180 WARN_ON(num_bytes < root->sectorsize);
2181 btrfs_set_key_type(ins, BTRFS_EXTENT_ITEM_KEY);
2185 if (orig_root->ref_cows || empty_size)
2186 allowed_chunk_alloc = 1;
2188 if (data & BTRFS_BLOCK_GROUP_METADATA) {
2189 last_ptr = &root->fs_info->last_alloc;
2190 empty_cluster = 256 * 1024;
2193 if ((data & BTRFS_BLOCK_GROUP_DATA) && btrfs_test_opt(root, SSD))
2194 last_ptr = &root->fs_info->last_data_alloc;
2198 hint_byte = *last_ptr;
2200 empty_size += empty_cluster;
2202 search_start = max(search_start, first_logical_byte(root, 0));
2203 search_start = max(search_start, hint_byte);
2204 total_needed += empty_size;
2206 block_group = btrfs_lookup_block_group(root->fs_info, search_start);
2207 space_info = __find_space_info(root->fs_info, data);
2209 down_read(&space_info->groups_sem);
2211 struct btrfs_free_space *free_space;
2213 * the only way this happens if our hint points to a block
2214 * group thats not of the proper type, while looping this
2215 * should never happen
2217 if (unlikely(!block_group_bits(block_group, data)))
2220 ret = cache_block_group(root, block_group);
2224 if (block_group->ro)
2227 free_space = btrfs_find_free_space(block_group, search_start,
2230 u64 start = block_group->key.objectid;
2231 u64 end = block_group->key.objectid +
2232 block_group->key.offset;
2234 search_start = stripe_align(root, free_space->offset);
2236 /* move on to the next group */
2237 if (search_start + num_bytes >= search_end)
2240 /* move on to the next group */
2241 if (search_start + num_bytes > end)
2244 if (exclude_nr > 0 &&
2245 (search_start + num_bytes > exclude_start &&
2246 search_start < exclude_start + exclude_nr)) {
2247 search_start = exclude_start + exclude_nr;
2249 * if search_start is still in this block group
2250 * then we just re-search this block group
2252 if (search_start >= start &&
2256 /* else we go to the next block group */
2260 ins->objectid = search_start;
2261 ins->offset = num_bytes;
2262 /* we are all good, lets return */
2267 * Here's how this works.
2268 * loop == 0: we were searching a block group via a hint
2269 * and didn't find anything, so we start at
2270 * the head of the block groups and keep searching
2271 * loop == 1: we're searching through all of the block groups
2272 * if we hit the head again we have searched
2273 * all of the block groups for this space and we
2274 * need to try and allocate, if we cant error out.
2275 * loop == 2: we allocated more space and are looping through
2276 * all of the block groups again.
2279 head = &space_info->block_groups;
2282 if (last_ptr && *last_ptr) {
2283 total_needed += empty_cluster;
2287 } else if (loop == 1 && cur == head) {
2288 if (allowed_chunk_alloc && !chunk_alloc_done) {
2289 up_read(&space_info->groups_sem);
2290 ret = do_chunk_alloc(trans, root, num_bytes +
2291 2 * 1024 * 1024, data, 1);
2294 down_read(&space_info->groups_sem);
2296 head = &space_info->block_groups;
2298 chunk_alloc_done = 1;
2299 } else if (!allowed_chunk_alloc) {
2300 space_info->force_alloc = 1;
2305 } else if (cur == head) {
2309 block_group = list_entry(cur, struct btrfs_block_group_cache,
2311 search_start = block_group->key.objectid;
2315 /* we found what we needed */
2316 if (ins->objectid) {
2317 if (!(data & BTRFS_BLOCK_GROUP_DATA))
2318 trans->block_group = block_group;
2321 *last_ptr = ins->objectid + ins->offset;
2327 up_read(&space_info->groups_sem);
2331 static void dump_space_info(struct btrfs_space_info *info, u64 bytes)
2333 struct btrfs_block_group_cache *cache;
2334 struct list_head *l;
2336 printk(KERN_INFO "space_info has %Lu free, is %sfull\n",
2337 info->total_bytes - info->bytes_used - info->bytes_pinned -
2338 info->bytes_reserved, (info->full) ? "" : "not ");
2340 down_read(&info->groups_sem);
2341 list_for_each(l, &info->block_groups) {
2342 cache = list_entry(l, struct btrfs_block_group_cache, list);
2343 spin_lock(&cache->lock);
2344 printk(KERN_INFO "block group %Lu has %Lu bytes, %Lu used "
2345 "%Lu pinned %Lu reserved\n",
2346 cache->key.objectid, cache->key.offset,
2347 btrfs_block_group_used(&cache->item),
2348 cache->pinned, cache->reserved);
2349 btrfs_dump_free_space(cache, bytes);
2350 spin_unlock(&cache->lock);
2352 up_read(&info->groups_sem);
2355 static int __btrfs_reserve_extent(struct btrfs_trans_handle *trans,
2356 struct btrfs_root *root,
2357 u64 num_bytes, u64 min_alloc_size,
2358 u64 empty_size, u64 hint_byte,
2359 u64 search_end, struct btrfs_key *ins,
2363 u64 search_start = 0;
2365 struct btrfs_fs_info *info = root->fs_info;
2366 struct btrfs_block_group_cache *cache;
2369 alloc_profile = info->avail_data_alloc_bits &
2370 info->data_alloc_profile;
2371 data = BTRFS_BLOCK_GROUP_DATA | alloc_profile;
2372 } else if (root == root->fs_info->chunk_root) {
2373 alloc_profile = info->avail_system_alloc_bits &
2374 info->system_alloc_profile;
2375 data = BTRFS_BLOCK_GROUP_SYSTEM | alloc_profile;
2377 alloc_profile = info->avail_metadata_alloc_bits &
2378 info->metadata_alloc_profile;
2379 data = BTRFS_BLOCK_GROUP_METADATA | alloc_profile;
2382 data = reduce_alloc_profile(root, data);
2384 * the only place that sets empty_size is btrfs_realloc_node, which
2385 * is not called recursively on allocations
2387 if (empty_size || root->ref_cows) {
2388 if (!(data & BTRFS_BLOCK_GROUP_METADATA)) {
2389 ret = do_chunk_alloc(trans, root->fs_info->extent_root,
2391 BTRFS_BLOCK_GROUP_METADATA |
2392 (info->metadata_alloc_profile &
2393 info->avail_metadata_alloc_bits), 0);
2395 ret = do_chunk_alloc(trans, root->fs_info->extent_root,
2396 num_bytes + 2 * 1024 * 1024, data, 0);
2399 WARN_ON(num_bytes < root->sectorsize);
2400 ret = find_free_extent(trans, root, num_bytes, empty_size,
2401 search_start, search_end, hint_byte, ins,
2402 trans->alloc_exclude_start,
2403 trans->alloc_exclude_nr, data);
2405 if (ret == -ENOSPC && num_bytes > min_alloc_size) {
2406 num_bytes = num_bytes >> 1;
2407 num_bytes = num_bytes & ~(root->sectorsize - 1);
2408 num_bytes = max(num_bytes, min_alloc_size);
2409 do_chunk_alloc(trans, root->fs_info->extent_root,
2410 num_bytes, data, 1);
2414 struct btrfs_space_info *sinfo;
2416 sinfo = __find_space_info(root->fs_info, data);
2417 printk("allocation failed flags %Lu, wanted %Lu\n",
2419 dump_space_info(sinfo, num_bytes);
2422 cache = btrfs_lookup_block_group(root->fs_info, ins->objectid);
2424 printk(KERN_ERR "Unable to find block group for %Lu\n", ins->objectid);
2428 ret = btrfs_remove_free_space(cache, ins->objectid, ins->offset);
2433 int btrfs_free_reserved_extent(struct btrfs_root *root, u64 start, u64 len)
2435 struct btrfs_block_group_cache *cache;
2437 maybe_lock_mutex(root);
2438 cache = btrfs_lookup_block_group(root->fs_info, start);
2440 printk(KERN_ERR "Unable to find block group for %Lu\n", start);
2441 maybe_unlock_mutex(root);
2444 btrfs_add_free_space(cache, start, len);
2445 update_reserved_extents(root, start, len, 0);
2446 maybe_unlock_mutex(root);
2450 int btrfs_reserve_extent(struct btrfs_trans_handle *trans,
2451 struct btrfs_root *root,
2452 u64 num_bytes, u64 min_alloc_size,
2453 u64 empty_size, u64 hint_byte,
2454 u64 search_end, struct btrfs_key *ins,
2458 maybe_lock_mutex(root);
2459 ret = __btrfs_reserve_extent(trans, root, num_bytes, min_alloc_size,
2460 empty_size, hint_byte, search_end, ins,
2462 update_reserved_extents(root, ins->objectid, ins->offset, 1);
2463 maybe_unlock_mutex(root);
2467 static int __btrfs_alloc_reserved_extent(struct btrfs_trans_handle *trans,
2468 struct btrfs_root *root, u64 parent,
2469 u64 root_objectid, u64 ref_generation,
2470 u64 owner, struct btrfs_key *ins)
2476 u64 num_bytes = ins->offset;
2478 struct btrfs_fs_info *info = root->fs_info;
2479 struct btrfs_root *extent_root = info->extent_root;
2480 struct btrfs_extent_item *extent_item;
2481 struct btrfs_extent_ref *ref;
2482 struct btrfs_path *path;
2483 struct btrfs_key keys[2];
2486 parent = ins->objectid;
2488 /* block accounting for super block */
2489 spin_lock_irq(&info->delalloc_lock);
2490 super_used = btrfs_super_bytes_used(&info->super_copy);
2491 btrfs_set_super_bytes_used(&info->super_copy, super_used + num_bytes);
2492 spin_unlock_irq(&info->delalloc_lock);
2494 /* block accounting for root item */
2495 root_used = btrfs_root_used(&root->root_item);
2496 btrfs_set_root_used(&root->root_item, root_used + num_bytes);
2498 if (root == extent_root) {
2499 struct pending_extent_op *extent_op;
2501 extent_op = kmalloc(sizeof(*extent_op), GFP_NOFS);
2504 extent_op->type = PENDING_EXTENT_INSERT;
2505 extent_op->bytenr = ins->objectid;
2506 extent_op->num_bytes = ins->offset;
2507 extent_op->parent = parent;
2508 extent_op->orig_parent = 0;
2509 extent_op->generation = ref_generation;
2510 extent_op->orig_generation = 0;
2511 extent_op->level = (int)owner;
2513 set_extent_bits(&root->fs_info->extent_ins, ins->objectid,
2514 ins->objectid + ins->offset - 1,
2515 EXTENT_LOCKED, GFP_NOFS);
2516 set_state_private(&root->fs_info->extent_ins,
2517 ins->objectid, (unsigned long)extent_op);
2521 memcpy(&keys[0], ins, sizeof(*ins));
2522 keys[1].objectid = ins->objectid;
2523 keys[1].type = BTRFS_EXTENT_REF_KEY;
2524 keys[1].offset = parent;
2525 sizes[0] = sizeof(*extent_item);
2526 sizes[1] = sizeof(*ref);
2528 path = btrfs_alloc_path();
2531 ret = btrfs_insert_empty_items(trans, extent_root, path, keys,
2535 extent_item = btrfs_item_ptr(path->nodes[0], path->slots[0],
2536 struct btrfs_extent_item);
2537 btrfs_set_extent_refs(path->nodes[0], extent_item, 1);
2538 ref = btrfs_item_ptr(path->nodes[0], path->slots[0] + 1,
2539 struct btrfs_extent_ref);
2541 btrfs_set_ref_root(path->nodes[0], ref, root_objectid);
2542 btrfs_set_ref_generation(path->nodes[0], ref, ref_generation);
2543 btrfs_set_ref_objectid(path->nodes[0], ref, owner);
2544 btrfs_set_ref_num_refs(path->nodes[0], ref, 1);
2546 btrfs_mark_buffer_dirty(path->nodes[0]);
2548 trans->alloc_exclude_start = 0;
2549 trans->alloc_exclude_nr = 0;
2550 btrfs_free_path(path);
2551 finish_current_insert(trans, extent_root);
2552 pending_ret = del_pending_extents(trans, extent_root);
2562 ret = update_block_group(trans, root, ins->objectid, ins->offset, 1, 0);
2564 printk("update block group failed for %Lu %Lu\n",
2565 ins->objectid, ins->offset);
2572 int btrfs_alloc_reserved_extent(struct btrfs_trans_handle *trans,
2573 struct btrfs_root *root, u64 parent,
2574 u64 root_objectid, u64 ref_generation,
2575 u64 owner, struct btrfs_key *ins)
2579 if (root_objectid == BTRFS_TREE_LOG_OBJECTID)
2581 maybe_lock_mutex(root);
2582 ret = __btrfs_alloc_reserved_extent(trans, root, parent, root_objectid,
2583 ref_generation, owner, ins);
2584 update_reserved_extents(root, ins->objectid, ins->offset, 0);
2585 maybe_unlock_mutex(root);
2590 * this is used by the tree logging recovery code. It records that
2591 * an extent has been allocated and makes sure to clear the free
2592 * space cache bits as well
2594 int btrfs_alloc_logged_extent(struct btrfs_trans_handle *trans,
2595 struct btrfs_root *root, u64 parent,
2596 u64 root_objectid, u64 ref_generation,
2597 u64 owner, struct btrfs_key *ins)
2600 struct btrfs_block_group_cache *block_group;
2602 maybe_lock_mutex(root);
2603 block_group = btrfs_lookup_block_group(root->fs_info, ins->objectid);
2604 cache_block_group(root, block_group);
2606 ret = btrfs_remove_free_space(block_group, ins->objectid, ins->offset);
2608 ret = __btrfs_alloc_reserved_extent(trans, root, parent, root_objectid,
2609 ref_generation, owner, ins);
2610 maybe_unlock_mutex(root);
2615 * finds a free extent and does all the dirty work required for allocation
2616 * returns the key for the extent through ins, and a tree buffer for
2617 * the first block of the extent through buf.
2619 * returns 0 if everything worked, non-zero otherwise.
2621 int btrfs_alloc_extent(struct btrfs_trans_handle *trans,
2622 struct btrfs_root *root,
2623 u64 num_bytes, u64 parent, u64 min_alloc_size,
2624 u64 root_objectid, u64 ref_generation,
2625 u64 owner_objectid, u64 empty_size, u64 hint_byte,
2626 u64 search_end, struct btrfs_key *ins, u64 data)
2630 maybe_lock_mutex(root);
2632 ret = __btrfs_reserve_extent(trans, root, num_bytes,
2633 min_alloc_size, empty_size, hint_byte,
2634 search_end, ins, data);
2636 if (root_objectid != BTRFS_TREE_LOG_OBJECTID) {
2637 ret = __btrfs_alloc_reserved_extent(trans, root, parent,
2638 root_objectid, ref_generation,
2639 owner_objectid, ins);
2643 update_reserved_extents(root, ins->objectid, ins->offset, 1);
2645 maybe_unlock_mutex(root);
2649 struct extent_buffer *btrfs_init_new_buffer(struct btrfs_trans_handle *trans,
2650 struct btrfs_root *root,
2651 u64 bytenr, u32 blocksize)
2653 struct extent_buffer *buf;
2655 buf = btrfs_find_create_tree_block(root, bytenr, blocksize);
2657 return ERR_PTR(-ENOMEM);
2658 btrfs_set_header_generation(buf, trans->transid);
2659 btrfs_tree_lock(buf);
2660 clean_tree_block(trans, root, buf);
2661 btrfs_set_buffer_uptodate(buf);
2662 if (root->root_key.objectid == BTRFS_TREE_LOG_OBJECTID) {
2663 set_extent_dirty(&root->dirty_log_pages, buf->start,
2664 buf->start + buf->len - 1, GFP_NOFS);
2666 set_extent_dirty(&trans->transaction->dirty_pages, buf->start,
2667 buf->start + buf->len - 1, GFP_NOFS);
2669 trans->blocks_used++;
2674 * helper function to allocate a block for a given tree
2675 * returns the tree buffer or NULL.
2677 struct extent_buffer *btrfs_alloc_free_block(struct btrfs_trans_handle *trans,
2678 struct btrfs_root *root,
2679 u32 blocksize, u64 parent,
2686 struct btrfs_key ins;
2688 struct extent_buffer *buf;
2690 ret = btrfs_alloc_extent(trans, root, blocksize, parent, blocksize,
2691 root_objectid, ref_generation, level,
2692 empty_size, hint, (u64)-1, &ins, 0);
2695 return ERR_PTR(ret);
2698 buf = btrfs_init_new_buffer(trans, root, ins.objectid, blocksize);
2702 int btrfs_drop_leaf_ref(struct btrfs_trans_handle *trans,
2703 struct btrfs_root *root, struct extent_buffer *leaf)
2706 u64 leaf_generation;
2707 struct btrfs_key key;
2708 struct btrfs_file_extent_item *fi;
2713 BUG_ON(!btrfs_is_leaf(leaf));
2714 nritems = btrfs_header_nritems(leaf);
2715 leaf_owner = btrfs_header_owner(leaf);
2716 leaf_generation = btrfs_header_generation(leaf);
2718 for (i = 0; i < nritems; i++) {
2722 btrfs_item_key_to_cpu(leaf, &key, i);
2723 if (btrfs_key_type(&key) != BTRFS_EXTENT_DATA_KEY)
2725 fi = btrfs_item_ptr(leaf, i, struct btrfs_file_extent_item);
2726 if (btrfs_file_extent_type(leaf, fi) ==
2727 BTRFS_FILE_EXTENT_INLINE)
2730 * FIXME make sure to insert a trans record that
2731 * repeats the snapshot del on crash
2733 disk_bytenr = btrfs_file_extent_disk_bytenr(leaf, fi);
2734 if (disk_bytenr == 0)
2737 mutex_lock(&root->fs_info->alloc_mutex);
2738 ret = __btrfs_free_extent(trans, root, disk_bytenr,
2739 btrfs_file_extent_disk_num_bytes(leaf, fi),
2740 leaf->start, leaf_owner, leaf_generation,
2742 mutex_unlock(&root->fs_info->alloc_mutex);
2745 atomic_inc(&root->fs_info->throttle_gen);
2746 wake_up(&root->fs_info->transaction_throttle);
2752 static int noinline cache_drop_leaf_ref(struct btrfs_trans_handle *trans,
2753 struct btrfs_root *root,
2754 struct btrfs_leaf_ref *ref)
2758 struct btrfs_extent_info *info = ref->extents;
2760 for (i = 0; i < ref->nritems; i++) {
2761 mutex_lock(&root->fs_info->alloc_mutex);
2762 ret = __btrfs_free_extent(trans, root, info->bytenr,
2763 info->num_bytes, ref->bytenr,
2764 ref->owner, ref->generation,
2766 mutex_unlock(&root->fs_info->alloc_mutex);
2768 atomic_inc(&root->fs_info->throttle_gen);
2769 wake_up(&root->fs_info->transaction_throttle);
2779 int drop_snap_lookup_refcount(struct btrfs_root *root, u64 start, u64 len,
2784 ret = btrfs_lookup_extent_ref(NULL, root, start, len, refs);
2787 #if 0 // some debugging code in case we see problems here
2788 /* if the refs count is one, it won't get increased again. But
2789 * if the ref count is > 1, someone may be decreasing it at
2790 * the same time we are.
2793 struct extent_buffer *eb = NULL;
2794 eb = btrfs_find_create_tree_block(root, start, len);
2796 btrfs_tree_lock(eb);
2798 mutex_lock(&root->fs_info->alloc_mutex);
2799 ret = lookup_extent_ref(NULL, root, start, len, refs);
2801 mutex_unlock(&root->fs_info->alloc_mutex);
2804 btrfs_tree_unlock(eb);
2805 free_extent_buffer(eb);
2808 printk("block %llu went down to one during drop_snap\n",
2809 (unsigned long long)start);
2820 * helper function for drop_snapshot, this walks down the tree dropping ref
2821 * counts as it goes.
2823 static int noinline walk_down_tree(struct btrfs_trans_handle *trans,
2824 struct btrfs_root *root,
2825 struct btrfs_path *path, int *level)
2831 struct extent_buffer *next;
2832 struct extent_buffer *cur;
2833 struct extent_buffer *parent;
2834 struct btrfs_leaf_ref *ref;
2839 WARN_ON(*level < 0);
2840 WARN_ON(*level >= BTRFS_MAX_LEVEL);
2841 ret = drop_snap_lookup_refcount(root, path->nodes[*level]->start,
2842 path->nodes[*level]->len, &refs);
2848 * walk down to the last node level and free all the leaves
2850 while(*level >= 0) {
2851 WARN_ON(*level < 0);
2852 WARN_ON(*level >= BTRFS_MAX_LEVEL);
2853 cur = path->nodes[*level];
2855 if (btrfs_header_level(cur) != *level)
2858 if (path->slots[*level] >=
2859 btrfs_header_nritems(cur))
2862 ret = btrfs_drop_leaf_ref(trans, root, cur);
2866 bytenr = btrfs_node_blockptr(cur, path->slots[*level]);
2867 ptr_gen = btrfs_node_ptr_generation(cur, path->slots[*level]);
2868 blocksize = btrfs_level_size(root, *level - 1);
2870 ret = drop_snap_lookup_refcount(root, bytenr, blocksize, &refs);
2873 parent = path->nodes[*level];
2874 root_owner = btrfs_header_owner(parent);
2875 root_gen = btrfs_header_generation(parent);
2876 path->slots[*level]++;
2878 mutex_lock(&root->fs_info->alloc_mutex);
2879 ret = __btrfs_free_extent(trans, root, bytenr,
2880 blocksize, parent->start,
2881 root_owner, root_gen,
2884 mutex_unlock(&root->fs_info->alloc_mutex);
2886 atomic_inc(&root->fs_info->throttle_gen);
2887 wake_up(&root->fs_info->transaction_throttle);
2893 * at this point, we have a single ref, and since the
2894 * only place referencing this extent is a dead root
2895 * the reference count should never go higher.
2896 * So, we don't need to check it again
2899 ref = btrfs_lookup_leaf_ref(root, bytenr);
2900 if (ref && ref->generation != ptr_gen) {
2901 btrfs_free_leaf_ref(root, ref);
2905 ret = cache_drop_leaf_ref(trans, root, ref);
2907 btrfs_remove_leaf_ref(root, ref);
2908 btrfs_free_leaf_ref(root, ref);
2912 if (printk_ratelimit()) {
2913 printk("leaf ref miss for bytenr %llu\n",
2914 (unsigned long long)bytenr);
2917 next = btrfs_find_tree_block(root, bytenr, blocksize);
2918 if (!next || !btrfs_buffer_uptodate(next, ptr_gen)) {
2919 free_extent_buffer(next);
2921 next = read_tree_block(root, bytenr, blocksize,
2926 * this is a debugging check and can go away
2927 * the ref should never go all the way down to 1
2930 ret = lookup_extent_ref(NULL, root, bytenr, blocksize,
2936 WARN_ON(*level <= 0);
2937 if (path->nodes[*level-1])
2938 free_extent_buffer(path->nodes[*level-1]);
2939 path->nodes[*level-1] = next;
2940 *level = btrfs_header_level(next);
2941 path->slots[*level] = 0;
2945 WARN_ON(*level < 0);
2946 WARN_ON(*level >= BTRFS_MAX_LEVEL);
2948 if (path->nodes[*level] == root->node) {
2949 parent = path->nodes[*level];
2950 bytenr = path->nodes[*level]->start;
2952 parent = path->nodes[*level + 1];
2953 bytenr = btrfs_node_blockptr(parent, path->slots[*level + 1]);
2956 blocksize = btrfs_level_size(root, *level);
2957 root_owner = btrfs_header_owner(parent);
2958 root_gen = btrfs_header_generation(parent);
2960 mutex_lock(&root->fs_info->alloc_mutex);
2961 ret = __btrfs_free_extent(trans, root, bytenr, blocksize,
2962 parent->start, root_owner, root_gen,
2964 mutex_unlock(&root->fs_info->alloc_mutex);
2965 free_extent_buffer(path->nodes[*level]);
2966 path->nodes[*level] = NULL;
2975 * helper function for drop_subtree, this function is similar to
2976 * walk_down_tree. The main difference is that it checks reference
2977 * counts while tree blocks are locked.
2979 static int noinline walk_down_subtree(struct btrfs_trans_handle *trans,
2980 struct btrfs_root *root,
2981 struct btrfs_path *path, int *level)
2983 struct extent_buffer *next;
2984 struct extent_buffer *cur;
2985 struct extent_buffer *parent;
2992 cur = path->nodes[*level];
2993 ret = btrfs_lookup_extent_ref(trans, root, cur->start, cur->len,
2999 while (*level >= 0) {
3000 cur = path->nodes[*level];
3002 ret = btrfs_drop_leaf_ref(trans, root, cur);
3004 clean_tree_block(trans, root, cur);
3007 if (path->slots[*level] >= btrfs_header_nritems(cur)) {
3008 clean_tree_block(trans, root, cur);
3012 bytenr = btrfs_node_blockptr(cur, path->slots[*level]);
3013 blocksize = btrfs_level_size(root, *level - 1);
3014 ptr_gen = btrfs_node_ptr_generation(cur, path->slots[*level]);
3016 next = read_tree_block(root, bytenr, blocksize, ptr_gen);
3017 btrfs_tree_lock(next);
3019 ret = btrfs_lookup_extent_ref(trans, root, bytenr, blocksize,
3023 parent = path->nodes[*level];
3024 ret = btrfs_free_extent(trans, root, bytenr,
3025 blocksize, parent->start,
3026 btrfs_header_owner(parent),
3027 btrfs_header_generation(parent),
3030 path->slots[*level]++;
3031 btrfs_tree_unlock(next);
3032 free_extent_buffer(next);
3036 *level = btrfs_header_level(next);
3037 path->nodes[*level] = next;
3038 path->slots[*level] = 0;
3039 path->locks[*level] = 1;
3043 parent = path->nodes[*level + 1];
3044 bytenr = path->nodes[*level]->start;
3045 blocksize = path->nodes[*level]->len;
3047 ret = btrfs_free_extent(trans, root, bytenr, blocksize,
3048 parent->start, btrfs_header_owner(parent),
3049 btrfs_header_generation(parent), *level, 1);
3052 if (path->locks[*level]) {
3053 btrfs_tree_unlock(path->nodes[*level]);
3054 path->locks[*level] = 0;
3056 free_extent_buffer(path->nodes[*level]);
3057 path->nodes[*level] = NULL;
3064 * helper for dropping snapshots. This walks back up the tree in the path
3065 * to find the first node higher up where we haven't yet gone through
3068 static int noinline walk_up_tree(struct btrfs_trans_handle *trans,
3069 struct btrfs_root *root,
3070 struct btrfs_path *path,
3071 int *level, int max_level)
3075 struct btrfs_root_item *root_item = &root->root_item;
3080 for (i = *level; i < max_level && path->nodes[i]; i++) {
3081 slot = path->slots[i];
3082 if (slot < btrfs_header_nritems(path->nodes[i]) - 1) {
3083 struct extent_buffer *node;
3084 struct btrfs_disk_key disk_key;
3085 node = path->nodes[i];
3088 WARN_ON(*level == 0);
3089 btrfs_node_key(node, &disk_key, path->slots[i]);
3090 memcpy(&root_item->drop_progress,
3091 &disk_key, sizeof(disk_key));
3092 root_item->drop_level = i;
3095 struct extent_buffer *parent;
3096 if (path->nodes[*level] == root->node)
3097 parent = path->nodes[*level];
3099 parent = path->nodes[*level + 1];
3101 root_owner = btrfs_header_owner(parent);
3102 root_gen = btrfs_header_generation(parent);
3104 clean_tree_block(trans, root, path->nodes[*level]);
3105 ret = btrfs_free_extent(trans, root,
3106 path->nodes[*level]->start,
3107 path->nodes[*level]->len,
3108 parent->start, root_owner,
3109 root_gen, *level, 1);
3111 if (path->locks[*level]) {
3112 btrfs_tree_unlock(path->nodes[*level]);
3113 path->locks[*level] = 0;
3115 free_extent_buffer(path->nodes[*level]);
3116 path->nodes[*level] = NULL;
3124 * drop the reference count on the tree rooted at 'snap'. This traverses
3125 * the tree freeing any blocks that have a ref count of zero after being
3128 int btrfs_drop_snapshot(struct btrfs_trans_handle *trans, struct btrfs_root
3134 struct btrfs_path *path;
3137 struct btrfs_root_item *root_item = &root->root_item;
3139 WARN_ON(!mutex_is_locked(&root->fs_info->drop_mutex));
3140 path = btrfs_alloc_path();
3143 level = btrfs_header_level(root->node);
3145 if (btrfs_disk_key_objectid(&root_item->drop_progress) == 0) {
3146 path->nodes[level] = root->node;
3147 extent_buffer_get(root->node);
3148 path->slots[level] = 0;
3150 struct btrfs_key key;
3151 struct btrfs_disk_key found_key;
3152 struct extent_buffer *node;
3154 btrfs_disk_key_to_cpu(&key, &root_item->drop_progress);
3155 level = root_item->drop_level;
3156 path->lowest_level = level;
3157 wret = btrfs_search_slot(NULL, root, &key, path, 0, 0);
3162 node = path->nodes[level];
3163 btrfs_node_key(node, &found_key, path->slots[level]);
3164 WARN_ON(memcmp(&found_key, &root_item->drop_progress,
3165 sizeof(found_key)));
3167 * unlock our path, this is safe because only this
3168 * function is allowed to delete this snapshot
3170 for (i = 0; i < BTRFS_MAX_LEVEL; i++) {
3171 if (path->nodes[i] && path->locks[i]) {
3173 btrfs_tree_unlock(path->nodes[i]);
3178 wret = walk_down_tree(trans, root, path, &level);
3184 wret = walk_up_tree(trans, root, path, &level,
3190 if (trans->transaction->in_commit) {
3194 atomic_inc(&root->fs_info->throttle_gen);
3195 wake_up(&root->fs_info->transaction_throttle);
3197 for (i = 0; i <= orig_level; i++) {
3198 if (path->nodes[i]) {
3199 free_extent_buffer(path->nodes[i]);
3200 path->nodes[i] = NULL;
3204 btrfs_free_path(path);
3208 int btrfs_drop_subtree(struct btrfs_trans_handle *trans,
3209 struct btrfs_root *root,
3210 struct extent_buffer *node,
3211 struct extent_buffer *parent)
3213 struct btrfs_path *path;
3219 path = btrfs_alloc_path();
3222 BUG_ON(!btrfs_tree_locked(parent));
3223 parent_level = btrfs_header_level(parent);
3224 extent_buffer_get(parent);
3225 path->nodes[parent_level] = parent;
3226 path->slots[parent_level] = btrfs_header_nritems(parent);
3228 BUG_ON(!btrfs_tree_locked(node));
3229 level = btrfs_header_level(node);
3230 extent_buffer_get(node);
3231 path->nodes[level] = node;
3232 path->slots[level] = 0;
3235 wret = walk_down_subtree(trans, root, path, &level);
3241 wret = walk_up_tree(trans, root, path, &level, parent_level);
3248 btrfs_free_path(path);
3252 static unsigned long calc_ra(unsigned long start, unsigned long last,
3255 return min(last, start + nr - 1);
3258 static int noinline relocate_inode_pages(struct inode *inode, u64 start,
3263 unsigned long first_index;
3264 unsigned long last_index;
3267 struct extent_io_tree *io_tree = &BTRFS_I(inode)->io_tree;
3268 struct file_ra_state *ra;
3269 struct btrfs_ordered_extent *ordered;
3270 unsigned int total_read = 0;
3271 unsigned int total_dirty = 0;
3274 ra = kzalloc(sizeof(*ra), GFP_NOFS);
3276 mutex_lock(&inode->i_mutex);
3277 first_index = start >> PAGE_CACHE_SHIFT;
3278 last_index = (start + len - 1) >> PAGE_CACHE_SHIFT;
3280 /* make sure the dirty trick played by the caller work */
3281 ret = invalidate_inode_pages2_range(inode->i_mapping,
3282 first_index, last_index);
3286 file_ra_state_init(ra, inode->i_mapping);
3288 for (i = first_index ; i <= last_index; i++) {
3289 if (total_read % ra->ra_pages == 0) {
3290 btrfs_force_ra(inode->i_mapping, ra, NULL, i,
3291 calc_ra(i, last_index, ra->ra_pages));
3295 if (((u64)i << PAGE_CACHE_SHIFT) > i_size_read(inode))
3297 page = grab_cache_page(inode->i_mapping, i);
3302 if (!PageUptodate(page)) {
3303 btrfs_readpage(NULL, page);
3305 if (!PageUptodate(page)) {
3307 page_cache_release(page);
3312 wait_on_page_writeback(page);
3314 page_start = (u64)page->index << PAGE_CACHE_SHIFT;
3315 page_end = page_start + PAGE_CACHE_SIZE - 1;
3316 lock_extent(io_tree, page_start, page_end, GFP_NOFS);
3318 ordered = btrfs_lookup_ordered_extent(inode, page_start);
3320 unlock_extent(io_tree, page_start, page_end, GFP_NOFS);
3322 page_cache_release(page);
3323 btrfs_start_ordered_extent(inode, ordered, 1);
3324 btrfs_put_ordered_extent(ordered);
3327 set_page_extent_mapped(page);
3329 btrfs_set_extent_delalloc(inode, page_start, page_end);
3330 if (i == first_index)
3331 set_extent_bits(io_tree, page_start, page_end,
3332 EXTENT_BOUNDARY, GFP_NOFS);
3334 set_page_dirty(page);
3337 unlock_extent(io_tree, page_start, page_end, GFP_NOFS);
3339 page_cache_release(page);
3344 mutex_unlock(&inode->i_mutex);
3345 balance_dirty_pages_ratelimited_nr(inode->i_mapping, total_dirty);
3349 static int noinline relocate_data_extent(struct inode *reloc_inode,
3350 struct btrfs_key *extent_key,
3353 struct btrfs_root *root = BTRFS_I(reloc_inode)->root;
3354 struct extent_map_tree *em_tree = &BTRFS_I(reloc_inode)->extent_tree;
3355 struct extent_map *em;
3357 em = alloc_extent_map(GFP_NOFS);
3358 BUG_ON(!em || IS_ERR(em));
3360 em->start = extent_key->objectid - offset;
3361 em->len = extent_key->offset;
3362 em->block_len = extent_key->offset;
3363 em->block_start = extent_key->objectid;
3364 em->bdev = root->fs_info->fs_devices->latest_bdev;
3365 set_bit(EXTENT_FLAG_PINNED, &em->flags);
3367 /* setup extent map to cheat btrfs_readpage */
3368 mutex_lock(&BTRFS_I(reloc_inode)->extent_mutex);
3371 spin_lock(&em_tree->lock);
3372 ret = add_extent_mapping(em_tree, em);
3373 spin_unlock(&em_tree->lock);
3374 if (ret != -EEXIST) {
3375 free_extent_map(em);
3378 btrfs_drop_extent_cache(reloc_inode, em->start,
3379 em->start + em->len - 1, 0);
3381 mutex_unlock(&BTRFS_I(reloc_inode)->extent_mutex);
3383 return relocate_inode_pages(reloc_inode, extent_key->objectid - offset,
3384 extent_key->offset);
3387 struct btrfs_ref_path {
3389 u64 nodes[BTRFS_MAX_LEVEL];
3391 u64 root_generation;
3398 struct btrfs_key node_keys[BTRFS_MAX_LEVEL];
3399 u64 new_nodes[BTRFS_MAX_LEVEL];
3402 struct disk_extent {
3413 static int is_cowonly_root(u64 root_objectid)
3415 if (root_objectid == BTRFS_ROOT_TREE_OBJECTID ||
3416 root_objectid == BTRFS_EXTENT_TREE_OBJECTID ||
3417 root_objectid == BTRFS_CHUNK_TREE_OBJECTID ||
3418 root_objectid == BTRFS_DEV_TREE_OBJECTID ||
3419 root_objectid == BTRFS_TREE_LOG_OBJECTID)
3424 static int noinline __next_ref_path(struct btrfs_trans_handle *trans,
3425 struct btrfs_root *extent_root,
3426 struct btrfs_ref_path *ref_path,
3429 struct extent_buffer *leaf;
3430 struct btrfs_path *path;
3431 struct btrfs_extent_ref *ref;
3432 struct btrfs_key key;
3433 struct btrfs_key found_key;
3439 path = btrfs_alloc_path();
3443 mutex_lock(&extent_root->fs_info->alloc_mutex);
3446 ref_path->lowest_level = -1;
3447 ref_path->current_level = -1;
3448 ref_path->shared_level = -1;
3452 level = ref_path->current_level - 1;
3453 while (level >= -1) {
3455 if (level < ref_path->lowest_level)
3459 bytenr = ref_path->nodes[level];
3461 bytenr = ref_path->extent_start;
3463 BUG_ON(bytenr == 0);
3465 parent = ref_path->nodes[level + 1];
3466 ref_path->nodes[level + 1] = 0;
3467 ref_path->current_level = level;
3468 BUG_ON(parent == 0);
3470 key.objectid = bytenr;
3471 key.offset = parent + 1;
3472 key.type = BTRFS_EXTENT_REF_KEY;
3474 ret = btrfs_search_slot(trans, extent_root, &key, path, 0, 0);
3479 leaf = path->nodes[0];
3480 nritems = btrfs_header_nritems(leaf);
3481 if (path->slots[0] >= nritems) {
3482 ret = btrfs_next_leaf(extent_root, path);
3487 leaf = path->nodes[0];
3490 btrfs_item_key_to_cpu(leaf, &found_key, path->slots[0]);
3491 if (found_key.objectid == bytenr &&
3492 found_key.type == BTRFS_EXTENT_REF_KEY) {
3493 if (level < ref_path->shared_level)
3494 ref_path->shared_level = level;
3499 btrfs_release_path(extent_root, path);
3500 if (need_resched()) {
3501 mutex_unlock(&extent_root->fs_info->alloc_mutex);
3503 mutex_lock(&extent_root->fs_info->alloc_mutex);
3506 /* reached lowest level */
3510 level = ref_path->current_level;
3511 while (level < BTRFS_MAX_LEVEL - 1) {
3514 bytenr = ref_path->nodes[level];
3516 bytenr = ref_path->extent_start;
3518 BUG_ON(bytenr == 0);
3520 key.objectid = bytenr;
3522 key.type = BTRFS_EXTENT_REF_KEY;
3524 ret = btrfs_search_slot(trans, extent_root, &key, path, 0, 0);
3528 leaf = path->nodes[0];
3529 nritems = btrfs_header_nritems(leaf);
3530 if (path->slots[0] >= nritems) {
3531 ret = btrfs_next_leaf(extent_root, path);
3535 /* the extent was freed by someone */
3536 if (ref_path->lowest_level == level)
3538 btrfs_release_path(extent_root, path);
3541 leaf = path->nodes[0];
3544 btrfs_item_key_to_cpu(leaf, &found_key, path->slots[0]);
3545 if (found_key.objectid != bytenr ||
3546 found_key.type != BTRFS_EXTENT_REF_KEY) {
3547 /* the extent was freed by someone */
3548 if (ref_path->lowest_level == level) {
3552 btrfs_release_path(extent_root, path);
3556 ref = btrfs_item_ptr(leaf, path->slots[0],
3557 struct btrfs_extent_ref);
3558 ref_objectid = btrfs_ref_objectid(leaf, ref);
3559 if (ref_objectid < BTRFS_FIRST_FREE_OBJECTID) {
3561 level = (int)ref_objectid;
3562 BUG_ON(level >= BTRFS_MAX_LEVEL);
3563 ref_path->lowest_level = level;
3564 ref_path->current_level = level;
3565 ref_path->nodes[level] = bytenr;
3567 WARN_ON(ref_objectid != level);
3570 WARN_ON(level != -1);
3574 if (ref_path->lowest_level == level) {
3575 ref_path->owner_objectid = ref_objectid;
3576 ref_path->num_refs = btrfs_ref_num_refs(leaf, ref);
3580 * the block is tree root or the block isn't in reference
3583 if (found_key.objectid == found_key.offset ||
3584 is_cowonly_root(btrfs_ref_root(leaf, ref))) {
3585 ref_path->root_objectid = btrfs_ref_root(leaf, ref);
3586 ref_path->root_generation =
3587 btrfs_ref_generation(leaf, ref);
3589 /* special reference from the tree log */
3590 ref_path->nodes[0] = found_key.offset;
3591 ref_path->current_level = 0;
3598 BUG_ON(ref_path->nodes[level] != 0);
3599 ref_path->nodes[level] = found_key.offset;
3600 ref_path->current_level = level;
3603 * the reference was created in the running transaction,
3604 * no need to continue walking up.
3606 if (btrfs_ref_generation(leaf, ref) == trans->transid) {
3607 ref_path->root_objectid = btrfs_ref_root(leaf, ref);
3608 ref_path->root_generation =
3609 btrfs_ref_generation(leaf, ref);
3614 btrfs_release_path(extent_root, path);
3615 if (need_resched()) {
3616 mutex_unlock(&extent_root->fs_info->alloc_mutex);
3618 mutex_lock(&extent_root->fs_info->alloc_mutex);
3621 /* reached max tree level, but no tree root found. */
3624 mutex_unlock(&extent_root->fs_info->alloc_mutex);
3625 btrfs_free_path(path);
3629 static int btrfs_first_ref_path(struct btrfs_trans_handle *trans,
3630 struct btrfs_root *extent_root,
3631 struct btrfs_ref_path *ref_path,
3634 memset(ref_path, 0, sizeof(*ref_path));
3635 ref_path->extent_start = extent_start;
3637 return __next_ref_path(trans, extent_root, ref_path, 1);
3640 static int btrfs_next_ref_path(struct btrfs_trans_handle *trans,
3641 struct btrfs_root *extent_root,
3642 struct btrfs_ref_path *ref_path)
3644 return __next_ref_path(trans, extent_root, ref_path, 0);
3647 static int noinline get_new_locations(struct inode *reloc_inode,
3648 struct btrfs_key *extent_key,
3649 u64 offset, int no_fragment,
3650 struct disk_extent **extents,
3653 struct btrfs_root *root = BTRFS_I(reloc_inode)->root;
3654 struct btrfs_path *path;
3655 struct btrfs_file_extent_item *fi;
3656 struct extent_buffer *leaf;
3657 struct disk_extent *exts = *extents;
3658 struct btrfs_key found_key;
3663 int max = *nr_extents;
3666 WARN_ON(!no_fragment && *extents);
3669 exts = kmalloc(sizeof(*exts) * max, GFP_NOFS);
3674 path = btrfs_alloc_path();
3677 cur_pos = extent_key->objectid - offset;
3678 last_byte = extent_key->objectid + extent_key->offset;
3679 ret = btrfs_lookup_file_extent(NULL, root, path, reloc_inode->i_ino,
3689 leaf = path->nodes[0];
3690 nritems = btrfs_header_nritems(leaf);
3691 if (path->slots[0] >= nritems) {
3692 ret = btrfs_next_leaf(root, path);
3697 leaf = path->nodes[0];
3700 btrfs_item_key_to_cpu(leaf, &found_key, path->slots[0]);
3701 if (found_key.offset != cur_pos ||
3702 found_key.type != BTRFS_EXTENT_DATA_KEY ||
3703 found_key.objectid != reloc_inode->i_ino)
3706 fi = btrfs_item_ptr(leaf, path->slots[0],
3707 struct btrfs_file_extent_item);
3708 if (btrfs_file_extent_type(leaf, fi) !=
3709 BTRFS_FILE_EXTENT_REG ||
3710 btrfs_file_extent_disk_bytenr(leaf, fi) == 0)
3714 struct disk_extent *old = exts;
3716 exts = kzalloc(sizeof(*exts) * max, GFP_NOFS);
3717 memcpy(exts, old, sizeof(*exts) * nr);
3718 if (old != *extents)
3722 exts[nr].disk_bytenr =
3723 btrfs_file_extent_disk_bytenr(leaf, fi);
3724 exts[nr].disk_num_bytes =
3725 btrfs_file_extent_disk_num_bytes(leaf, fi);
3726 exts[nr].offset = btrfs_file_extent_offset(leaf, fi);
3727 exts[nr].num_bytes = btrfs_file_extent_num_bytes(leaf, fi);
3728 exts[nr].ram_bytes = btrfs_file_extent_ram_bytes(leaf, fi);
3729 exts[nr].compression = btrfs_file_extent_compression(leaf, fi);
3730 exts[nr].encryption = btrfs_file_extent_encryption(leaf, fi);
3731 exts[nr].other_encoding = btrfs_file_extent_other_encoding(leaf,
3733 WARN_ON(exts[nr].offset > 0);
3734 WARN_ON(exts[nr].num_bytes != exts[nr].disk_num_bytes);
3736 cur_pos += exts[nr].num_bytes;
3739 if (cur_pos + offset >= last_byte)
3749 WARN_ON(cur_pos + offset > last_byte);
3750 if (cur_pos + offset < last_byte) {
3756 btrfs_free_path(path);
3758 if (exts != *extents)
3767 static int noinline replace_one_extent(struct btrfs_trans_handle *trans,
3768 struct btrfs_root *root,
3769 struct btrfs_path *path,
3770 struct btrfs_key *extent_key,
3771 struct btrfs_key *leaf_key,
3772 struct btrfs_ref_path *ref_path,
3773 struct disk_extent *new_extents,
3776 struct extent_buffer *leaf;
3777 struct btrfs_file_extent_item *fi;
3778 struct inode *inode = NULL;
3779 struct btrfs_key key;
3787 int extent_locked = 0;
3790 memcpy(&key, leaf_key, sizeof(key));
3791 first_pos = INT_LIMIT(loff_t) - extent_key->offset;
3792 if (ref_path->owner_objectid != BTRFS_MULTIPLE_OBJECTIDS) {
3793 if (key.objectid < ref_path->owner_objectid ||
3794 (key.objectid == ref_path->owner_objectid &&
3795 key.type < BTRFS_EXTENT_DATA_KEY)) {
3796 key.objectid = ref_path->owner_objectid;
3797 key.type = BTRFS_EXTENT_DATA_KEY;
3803 ret = btrfs_search_slot(trans, root, &key, path, 0, 1);
3807 leaf = path->nodes[0];
3808 nritems = btrfs_header_nritems(leaf);
3810 if (extent_locked && ret > 0) {
3812 * the file extent item was modified by someone
3813 * before the extent got locked.
3815 mutex_unlock(&BTRFS_I(inode)->extent_mutex);
3816 unlock_extent(&BTRFS_I(inode)->io_tree, lock_start,
3817 lock_end, GFP_NOFS);
3821 if (path->slots[0] >= nritems) {
3822 if (++nr_scaned > 2)
3825 BUG_ON(extent_locked);
3826 ret = btrfs_next_leaf(root, path);
3831 leaf = path->nodes[0];
3832 nritems = btrfs_header_nritems(leaf);
3835 btrfs_item_key_to_cpu(leaf, &key, path->slots[0]);
3837 if (ref_path->owner_objectid != BTRFS_MULTIPLE_OBJECTIDS) {
3838 if ((key.objectid > ref_path->owner_objectid) ||
3839 (key.objectid == ref_path->owner_objectid &&
3840 key.type > BTRFS_EXTENT_DATA_KEY) ||
3841 (key.offset >= first_pos + extent_key->offset))
3845 if (inode && key.objectid != inode->i_ino) {
3846 BUG_ON(extent_locked);
3847 btrfs_release_path(root, path);
3848 mutex_unlock(&inode->i_mutex);
3854 if (key.type != BTRFS_EXTENT_DATA_KEY) {
3859 fi = btrfs_item_ptr(leaf, path->slots[0],
3860 struct btrfs_file_extent_item);
3861 if ((btrfs_file_extent_type(leaf, fi) !=
3862 BTRFS_FILE_EXTENT_REG) ||
3863 (btrfs_file_extent_disk_bytenr(leaf, fi) !=
3864 extent_key->objectid)) {
3870 num_bytes = btrfs_file_extent_num_bytes(leaf, fi);
3871 ext_offset = btrfs_file_extent_offset(leaf, fi);
3873 if (first_pos > key.offset - ext_offset)
3874 first_pos = key.offset - ext_offset;
3876 if (!extent_locked) {
3877 lock_start = key.offset;
3878 lock_end = lock_start + num_bytes - 1;
3880 BUG_ON(lock_start != key.offset);
3881 BUG_ON(lock_end - lock_start + 1 < num_bytes);
3885 btrfs_release_path(root, path);
3887 inode = btrfs_iget_locked(root->fs_info->sb,
3888 key.objectid, root);
3889 if (inode->i_state & I_NEW) {
3890 BTRFS_I(inode)->root = root;
3891 BTRFS_I(inode)->location.objectid =
3893 BTRFS_I(inode)->location.type =
3894 BTRFS_INODE_ITEM_KEY;
3895 BTRFS_I(inode)->location.offset = 0;
3896 btrfs_read_locked_inode(inode);
3897 unlock_new_inode(inode);
3900 * some code call btrfs_commit_transaction while
3901 * holding the i_mutex, so we can't use mutex_lock
3904 if (is_bad_inode(inode) ||
3905 !mutex_trylock(&inode->i_mutex)) {
3908 key.offset = (u64)-1;
3913 if (!extent_locked) {
3914 struct btrfs_ordered_extent *ordered;
3916 btrfs_release_path(root, path);
3918 lock_extent(&BTRFS_I(inode)->io_tree, lock_start,
3919 lock_end, GFP_NOFS);
3920 ordered = btrfs_lookup_first_ordered_extent(inode,
3923 ordered->file_offset <= lock_end &&
3924 ordered->file_offset + ordered->len > lock_start) {
3925 unlock_extent(&BTRFS_I(inode)->io_tree,
3926 lock_start, lock_end, GFP_NOFS);
3927 btrfs_start_ordered_extent(inode, ordered, 1);
3928 btrfs_put_ordered_extent(ordered);
3929 key.offset += num_bytes;
3933 btrfs_put_ordered_extent(ordered);
3935 mutex_lock(&BTRFS_I(inode)->extent_mutex);
3940 if (nr_extents == 1) {
3941 /* update extent pointer in place */
3942 btrfs_set_file_extent_generation(leaf, fi,
3944 btrfs_set_file_extent_disk_bytenr(leaf, fi,
3945 new_extents[0].disk_bytenr);
3946 btrfs_set_file_extent_disk_num_bytes(leaf, fi,
3947 new_extents[0].disk_num_bytes);
3948 btrfs_set_file_extent_ram_bytes(leaf, fi,
3949 new_extents[0].ram_bytes);
3950 ext_offset += new_extents[0].offset;
3951 btrfs_set_file_extent_offset(leaf, fi, ext_offset);
3952 btrfs_mark_buffer_dirty(leaf);
3954 btrfs_drop_extent_cache(inode, key.offset,
3955 key.offset + num_bytes - 1, 0);
3957 ret = btrfs_inc_extent_ref(trans, root,
3958 new_extents[0].disk_bytenr,
3959 new_extents[0].disk_num_bytes,
3961 root->root_key.objectid,
3966 ret = btrfs_free_extent(trans, root,
3967 extent_key->objectid,
3970 btrfs_header_owner(leaf),
3971 btrfs_header_generation(leaf),
3975 btrfs_release_path(root, path);
3976 key.offset += num_bytes;
3982 * drop old extent pointer at first, then insert the
3983 * new pointers one bye one
3985 btrfs_release_path(root, path);
3986 ret = btrfs_drop_extents(trans, root, inode, key.offset,
3987 key.offset + num_bytes,
3988 key.offset, &alloc_hint);
3991 for (i = 0; i < nr_extents; i++) {
3992 if (ext_offset >= new_extents[i].num_bytes) {
3993 ext_offset -= new_extents[i].num_bytes;
3996 extent_len = min(new_extents[i].num_bytes -
3997 ext_offset, num_bytes);
3999 ret = btrfs_insert_empty_item(trans, root,
4004 leaf = path->nodes[0];
4005 fi = btrfs_item_ptr(leaf, path->slots[0],
4006 struct btrfs_file_extent_item);
4007 btrfs_set_file_extent_generation(leaf, fi,
4009 btrfs_set_file_extent_type(leaf, fi,
4010 BTRFS_FILE_EXTENT_REG);
4011 btrfs_set_file_extent_disk_bytenr(leaf, fi,
4012 new_extents[i].disk_bytenr);
4013 btrfs_set_file_extent_disk_num_bytes(leaf, fi,
4014 new_extents[i].disk_num_bytes);
4015 btrfs_set_file_extent_ram_bytes(leaf, fi,
4016 new_extents[i].ram_bytes);
4018 btrfs_set_file_extent_compression(leaf, fi,
4019 new_extents[i].compression);
4020 btrfs_set_file_extent_encryption(leaf, fi,
4021 new_extents[i].encryption);
4022 btrfs_set_file_extent_other_encoding(leaf, fi,
4023 new_extents[i].other_encoding);
4025 btrfs_set_file_extent_num_bytes(leaf, fi,
4027 ext_offset += new_extents[i].offset;
4028 btrfs_set_file_extent_offset(leaf, fi,
4030 btrfs_mark_buffer_dirty(leaf);
4032 btrfs_drop_extent_cache(inode, key.offset,
4033 key.offset + extent_len - 1, 0);
4035 ret = btrfs_inc_extent_ref(trans, root,
4036 new_extents[i].disk_bytenr,
4037 new_extents[i].disk_num_bytes,
4039 root->root_key.objectid,
4040 trans->transid, key.objectid);
4042 btrfs_release_path(root, path);
4044 inode_add_bytes(inode, extent_len);
4047 num_bytes -= extent_len;
4048 key.offset += extent_len;
4053 BUG_ON(i >= nr_extents);
4056 if (extent_locked) {
4057 mutex_unlock(&BTRFS_I(inode)->extent_mutex);
4058 unlock_extent(&BTRFS_I(inode)->io_tree, lock_start,
4059 lock_end, GFP_NOFS);
4063 if (ref_path->owner_objectid != BTRFS_MULTIPLE_OBJECTIDS &&
4064 key.offset >= first_pos + extent_key->offset)
4071 btrfs_release_path(root, path);
4073 mutex_unlock(&inode->i_mutex);
4074 if (extent_locked) {
4075 mutex_unlock(&BTRFS_I(inode)->extent_mutex);
4076 unlock_extent(&BTRFS_I(inode)->io_tree, lock_start,
4077 lock_end, GFP_NOFS);
4084 int btrfs_reloc_tree_cache_ref(struct btrfs_trans_handle *trans,
4085 struct btrfs_root *root,
4086 struct extent_buffer *buf, u64 orig_start)
4091 BUG_ON(btrfs_header_generation(buf) != trans->transid);
4092 BUG_ON(root->root_key.objectid != BTRFS_TREE_RELOC_OBJECTID);
4094 level = btrfs_header_level(buf);
4096 struct btrfs_leaf_ref *ref;
4097 struct btrfs_leaf_ref *orig_ref;
4099 orig_ref = btrfs_lookup_leaf_ref(root, orig_start);
4103 ref = btrfs_alloc_leaf_ref(root, orig_ref->nritems);
4105 btrfs_free_leaf_ref(root, orig_ref);
4109 ref->nritems = orig_ref->nritems;
4110 memcpy(ref->extents, orig_ref->extents,
4111 sizeof(ref->extents[0]) * ref->nritems);
4113 btrfs_free_leaf_ref(root, orig_ref);
4115 ref->root_gen = trans->transid;
4116 ref->bytenr = buf->start;
4117 ref->owner = btrfs_header_owner(buf);
4118 ref->generation = btrfs_header_generation(buf);
4119 ret = btrfs_add_leaf_ref(root, ref, 0);
4121 btrfs_free_leaf_ref(root, ref);
4126 static int noinline invalidate_extent_cache(struct btrfs_root *root,
4127 struct extent_buffer *leaf,
4128 struct btrfs_block_group_cache *group,
4129 struct btrfs_root *target_root)
4131 struct btrfs_key key;
4132 struct inode *inode = NULL;
4133 struct btrfs_file_extent_item *fi;
4135 u64 skip_objectid = 0;
4139 nritems = btrfs_header_nritems(leaf);
4140 for (i = 0; i < nritems; i++) {
4141 btrfs_item_key_to_cpu(leaf, &key, i);
4142 if (key.objectid == skip_objectid ||
4143 key.type != BTRFS_EXTENT_DATA_KEY)
4145 fi = btrfs_item_ptr(leaf, i, struct btrfs_file_extent_item);
4146 if (btrfs_file_extent_type(leaf, fi) ==
4147 BTRFS_FILE_EXTENT_INLINE)
4149 if (btrfs_file_extent_disk_bytenr(leaf, fi) == 0)
4151 if (!inode || inode->i_ino != key.objectid) {
4153 inode = btrfs_ilookup(target_root->fs_info->sb,
4154 key.objectid, target_root, 1);
4157 skip_objectid = key.objectid;
4160 num_bytes = btrfs_file_extent_num_bytes(leaf, fi);
4162 lock_extent(&BTRFS_I(inode)->io_tree, key.offset,
4163 key.offset + num_bytes - 1, GFP_NOFS);
4164 mutex_lock(&BTRFS_I(inode)->extent_mutex);
4165 btrfs_drop_extent_cache(inode, key.offset,
4166 key.offset + num_bytes - 1, 1);
4167 mutex_unlock(&BTRFS_I(inode)->extent_mutex);
4168 unlock_extent(&BTRFS_I(inode)->io_tree, key.offset,
4169 key.offset + num_bytes - 1, GFP_NOFS);
4176 static int noinline replace_extents_in_leaf(struct btrfs_trans_handle *trans,
4177 struct btrfs_root *root,
4178 struct extent_buffer *leaf,
4179 struct btrfs_block_group_cache *group,
4180 struct inode *reloc_inode)
4182 struct btrfs_key key;
4183 struct btrfs_key extent_key;
4184 struct btrfs_file_extent_item *fi;
4185 struct btrfs_leaf_ref *ref;
4186 struct disk_extent *new_extent;
4195 new_extent = kmalloc(sizeof(*new_extent), GFP_NOFS);
4196 BUG_ON(!new_extent);
4198 ref = btrfs_lookup_leaf_ref(root, leaf->start);
4202 nritems = btrfs_header_nritems(leaf);
4203 for (i = 0; i < nritems; i++) {
4204 btrfs_item_key_to_cpu(leaf, &key, i);
4205 if (btrfs_key_type(&key) != BTRFS_EXTENT_DATA_KEY)
4207 fi = btrfs_item_ptr(leaf, i, struct btrfs_file_extent_item);
4208 if (btrfs_file_extent_type(leaf, fi) ==
4209 BTRFS_FILE_EXTENT_INLINE)
4211 bytenr = btrfs_file_extent_disk_bytenr(leaf, fi);
4212 num_bytes = btrfs_file_extent_disk_num_bytes(leaf, fi);
4217 if (bytenr >= group->key.objectid + group->key.offset ||
4218 bytenr + num_bytes <= group->key.objectid)
4221 extent_key.objectid = bytenr;
4222 extent_key.offset = num_bytes;
4223 extent_key.type = BTRFS_EXTENT_ITEM_KEY;
4225 ret = get_new_locations(reloc_inode, &extent_key,
4226 group->key.objectid, 1,
4227 &new_extent, &nr_extent);
4232 BUG_ON(ref->extents[ext_index].bytenr != bytenr);
4233 BUG_ON(ref->extents[ext_index].num_bytes != num_bytes);
4234 ref->extents[ext_index].bytenr = new_extent->disk_bytenr;
4235 ref->extents[ext_index].num_bytes = new_extent->disk_num_bytes;
4237 btrfs_set_file_extent_generation(leaf, fi, trans->transid);
4238 btrfs_set_file_extent_ram_bytes(leaf, fi,
4239 new_extent->ram_bytes);
4240 btrfs_set_file_extent_disk_bytenr(leaf, fi,
4241 new_extent->disk_bytenr);
4242 btrfs_set_file_extent_disk_num_bytes(leaf, fi,
4243 new_extent->disk_num_bytes);
4244 new_extent->offset += btrfs_file_extent_offset(leaf, fi);
4245 btrfs_set_file_extent_offset(leaf, fi, new_extent->offset);
4246 btrfs_mark_buffer_dirty(leaf);
4248 ret = btrfs_inc_extent_ref(trans, root,
4249 new_extent->disk_bytenr,
4250 new_extent->disk_num_bytes,
4252 root->root_key.objectid,
4253 trans->transid, key.objectid);
4255 ret = btrfs_free_extent(trans, root,
4256 bytenr, num_bytes, leaf->start,
4257 btrfs_header_owner(leaf),
4258 btrfs_header_generation(leaf),
4264 BUG_ON(ext_index + 1 != ref->nritems);
4265 btrfs_free_leaf_ref(root, ref);
4269 int btrfs_free_reloc_root(struct btrfs_trans_handle *trans,
4270 struct btrfs_root *root)
4272 struct btrfs_root *reloc_root;
4275 if (root->reloc_root) {
4276 reloc_root = root->reloc_root;
4277 root->reloc_root = NULL;
4278 list_add(&reloc_root->dead_list,
4279 &root->fs_info->dead_reloc_roots);
4281 btrfs_set_root_bytenr(&reloc_root->root_item,
4282 reloc_root->node->start);
4283 btrfs_set_root_level(&root->root_item,
4284 btrfs_header_level(reloc_root->node));
4285 memset(&reloc_root->root_item.drop_progress, 0,
4286 sizeof(struct btrfs_disk_key));
4287 reloc_root->root_item.drop_level = 0;
4289 ret = btrfs_update_root(trans, root->fs_info->tree_root,
4290 &reloc_root->root_key,
4291 &reloc_root->root_item);
4297 int btrfs_drop_dead_reloc_roots(struct btrfs_root *root)
4299 struct btrfs_trans_handle *trans;
4300 struct btrfs_root *reloc_root;
4301 struct btrfs_root *prev_root = NULL;
4302 struct list_head dead_roots;
4306 INIT_LIST_HEAD(&dead_roots);
4307 list_splice_init(&root->fs_info->dead_reloc_roots, &dead_roots);
4309 while (!list_empty(&dead_roots)) {
4310 reloc_root = list_entry(dead_roots.prev,
4311 struct btrfs_root, dead_list);
4312 list_del_init(&reloc_root->dead_list);
4314 BUG_ON(reloc_root->commit_root != NULL);
4316 trans = btrfs_join_transaction(root, 1);
4319 mutex_lock(&root->fs_info->drop_mutex);
4320 ret = btrfs_drop_snapshot(trans, reloc_root);
4323 mutex_unlock(&root->fs_info->drop_mutex);
4325 nr = trans->blocks_used;
4326 ret = btrfs_end_transaction(trans, root);
4328 btrfs_btree_balance_dirty(root, nr);
4331 free_extent_buffer(reloc_root->node);
4333 ret = btrfs_del_root(trans, root->fs_info->tree_root,
4334 &reloc_root->root_key);
4336 mutex_unlock(&root->fs_info->drop_mutex);
4338 nr = trans->blocks_used;
4339 ret = btrfs_end_transaction(trans, root);
4341 btrfs_btree_balance_dirty(root, nr);
4344 prev_root = reloc_root;
4347 btrfs_remove_leaf_refs(prev_root, (u64)-1, 0);
4353 int btrfs_add_dead_reloc_root(struct btrfs_root *root)
4355 list_add(&root->dead_list, &root->fs_info->dead_reloc_roots);
4359 int btrfs_cleanup_reloc_trees(struct btrfs_root *root)
4361 struct btrfs_root *reloc_root;
4362 struct btrfs_trans_handle *trans;
4363 struct btrfs_key location;
4367 mutex_lock(&root->fs_info->tree_reloc_mutex);
4368 ret = btrfs_find_dead_roots(root, BTRFS_TREE_RELOC_OBJECTID, NULL);
4370 found = !list_empty(&root->fs_info->dead_reloc_roots);
4371 mutex_unlock(&root->fs_info->tree_reloc_mutex);
4374 trans = btrfs_start_transaction(root, 1);
4376 ret = btrfs_commit_transaction(trans, root);
4380 location.objectid = BTRFS_DATA_RELOC_TREE_OBJECTID;
4381 location.offset = (u64)-1;
4382 location.type = BTRFS_ROOT_ITEM_KEY;
4384 reloc_root = btrfs_read_fs_root_no_name(root->fs_info, &location);
4385 BUG_ON(!reloc_root);
4386 btrfs_orphan_cleanup(reloc_root);
4390 static int noinline init_reloc_tree(struct btrfs_trans_handle *trans,
4391 struct btrfs_root *root)
4393 struct btrfs_root *reloc_root;
4394 struct extent_buffer *eb;
4395 struct btrfs_root_item *root_item;
4396 struct btrfs_key root_key;
4399 BUG_ON(!root->ref_cows);
4400 if (root->reloc_root)
4403 root_item = kmalloc(sizeof(*root_item), GFP_NOFS);
4406 ret = btrfs_copy_root(trans, root, root->commit_root,
4407 &eb, BTRFS_TREE_RELOC_OBJECTID);
4410 root_key.objectid = BTRFS_TREE_RELOC_OBJECTID;
4411 root_key.offset = root->root_key.objectid;
4412 root_key.type = BTRFS_ROOT_ITEM_KEY;
4414 memcpy(root_item, &root->root_item, sizeof(root_item));
4415 btrfs_set_root_refs(root_item, 0);
4416 btrfs_set_root_bytenr(root_item, eb->start);
4417 btrfs_set_root_level(root_item, btrfs_header_level(eb));
4419 btrfs_tree_unlock(eb);
4420 free_extent_buffer(eb);
4422 ret = btrfs_insert_root(trans, root->fs_info->tree_root,
4423 &root_key, root_item);
4427 reloc_root = btrfs_read_fs_root_no_radix(root->fs_info->tree_root,
4429 BUG_ON(!reloc_root);
4430 reloc_root->last_trans = trans->transid;
4431 reloc_root->commit_root = NULL;
4432 reloc_root->ref_tree = &root->fs_info->reloc_ref_tree;
4434 root->reloc_root = reloc_root;
4439 * Core function of space balance.
4441 * The idea is using reloc trees to relocate tree blocks in reference
4442 * counted roots. There is one reloc tree for each subvol, and all
4443 * reloc trees share same root key objectid. Reloc trees are snapshots
4444 * of the latest committed roots of subvols (root->commit_root).
4446 * To relocate a tree block referenced by a subvol, there are two steps.
4447 * COW the block through subvol's reloc tree, then update block pointer
4448 * in the subvol to point to the new block. Since all reloc trees share
4449 * same root key objectid, doing special handing for tree blocks owned
4450 * by them is easy. Once a tree block has been COWed in one reloc tree,
4451 * we can use the resulting new block directly when the same block is
4452 * required to COW again through other reloc trees. By this way, relocated
4453 * tree blocks are shared between reloc trees, so they are also shared
4456 static int noinline relocate_one_path(struct btrfs_trans_handle *trans,
4457 struct btrfs_root *root,
4458 struct btrfs_path *path,
4459 struct btrfs_key *first_key,
4460 struct btrfs_ref_path *ref_path,
4461 struct btrfs_block_group_cache *group,
4462 struct inode *reloc_inode)
4464 struct btrfs_root *reloc_root;
4465 struct extent_buffer *eb = NULL;
4466 struct btrfs_key *keys;
4470 int lowest_level = 0;
4473 if (ref_path->owner_objectid < BTRFS_FIRST_FREE_OBJECTID)
4474 lowest_level = ref_path->owner_objectid;
4476 if (!root->ref_cows) {
4477 path->lowest_level = lowest_level;
4478 ret = btrfs_search_slot(trans, root, first_key, path, 0, 1);
4480 path->lowest_level = 0;
4481 btrfs_release_path(root, path);
4485 mutex_lock(&root->fs_info->tree_reloc_mutex);
4486 ret = init_reloc_tree(trans, root);
4488 reloc_root = root->reloc_root;
4490 shared_level = ref_path->shared_level;
4491 ref_path->shared_level = BTRFS_MAX_LEVEL - 1;
4493 keys = ref_path->node_keys;
4494 nodes = ref_path->new_nodes;
4495 memset(&keys[shared_level + 1], 0,
4496 sizeof(*keys) * (BTRFS_MAX_LEVEL - shared_level - 1));
4497 memset(&nodes[shared_level + 1], 0,
4498 sizeof(*nodes) * (BTRFS_MAX_LEVEL - shared_level - 1));
4500 if (nodes[lowest_level] == 0) {
4501 path->lowest_level = lowest_level;
4502 ret = btrfs_search_slot(trans, reloc_root, first_key, path,
4505 for (level = lowest_level; level < BTRFS_MAX_LEVEL; level++) {
4506 eb = path->nodes[level];
4507 if (!eb || eb == reloc_root->node)
4509 nodes[level] = eb->start;
4511 btrfs_item_key_to_cpu(eb, &keys[level], 0);
4513 btrfs_node_key_to_cpu(eb, &keys[level], 0);
4515 if (ref_path->owner_objectid >= BTRFS_FIRST_FREE_OBJECTID) {
4516 eb = path->nodes[0];
4517 ret = replace_extents_in_leaf(trans, reloc_root, eb,
4518 group, reloc_inode);
4521 btrfs_release_path(reloc_root, path);
4523 ret = btrfs_merge_path(trans, reloc_root, keys, nodes,
4529 * replace tree blocks in the fs tree with tree blocks in
4532 ret = btrfs_merge_path(trans, root, keys, nodes, lowest_level);
4535 if (ref_path->owner_objectid >= BTRFS_FIRST_FREE_OBJECTID) {
4536 ret = btrfs_search_slot(trans, reloc_root, first_key, path,
4539 extent_buffer_get(path->nodes[0]);
4540 eb = path->nodes[0];
4541 btrfs_release_path(reloc_root, path);
4542 ret = invalidate_extent_cache(reloc_root, eb, group, root);
4544 free_extent_buffer(eb);
4547 mutex_unlock(&root->fs_info->tree_reloc_mutex);
4548 path->lowest_level = 0;
4552 static int noinline relocate_tree_block(struct btrfs_trans_handle *trans,
4553 struct btrfs_root *root,
4554 struct btrfs_path *path,
4555 struct btrfs_key *first_key,
4556 struct btrfs_ref_path *ref_path)
4561 if (root == root->fs_info->extent_root ||
4562 root == root->fs_info->chunk_root ||
4563 root == root->fs_info->dev_root) {
4565 mutex_lock(&root->fs_info->alloc_mutex);
4568 ret = relocate_one_path(trans, root, path, first_key,
4569 ref_path, NULL, NULL);
4572 if (root == root->fs_info->extent_root)
4573 btrfs_extent_post_op(trans, root);
4575 mutex_unlock(&root->fs_info->alloc_mutex);
4580 static int noinline del_extent_zero(struct btrfs_trans_handle *trans,
4581 struct btrfs_root *extent_root,
4582 struct btrfs_path *path,
4583 struct btrfs_key *extent_key)
4587 mutex_lock(&extent_root->fs_info->alloc_mutex);
4588 ret = btrfs_search_slot(trans, extent_root, extent_key, path, -1, 1);
4591 ret = btrfs_del_item(trans, extent_root, path);
4593 btrfs_release_path(extent_root, path);
4594 mutex_unlock(&extent_root->fs_info->alloc_mutex);
4598 static struct btrfs_root noinline *read_ref_root(struct btrfs_fs_info *fs_info,
4599 struct btrfs_ref_path *ref_path)
4601 struct btrfs_key root_key;
4603 root_key.objectid = ref_path->root_objectid;
4604 root_key.type = BTRFS_ROOT_ITEM_KEY;
4605 if (is_cowonly_root(ref_path->root_objectid))
4606 root_key.offset = 0;
4608 root_key.offset = (u64)-1;
4610 return btrfs_read_fs_root_no_name(fs_info, &root_key);
4613 static int noinline relocate_one_extent(struct btrfs_root *extent_root,
4614 struct btrfs_path *path,
4615 struct btrfs_key *extent_key,
4616 struct btrfs_block_group_cache *group,
4617 struct inode *reloc_inode, int pass)
4619 struct btrfs_trans_handle *trans;
4620 struct btrfs_root *found_root;
4621 struct btrfs_ref_path *ref_path = NULL;
4622 struct disk_extent *new_extents = NULL;
4627 struct btrfs_key first_key;
4630 mutex_unlock(&extent_root->fs_info->alloc_mutex);
4632 trans = btrfs_start_transaction(extent_root, 1);
4635 if (extent_key->objectid == 0) {
4636 ret = del_extent_zero(trans, extent_root, path, extent_key);
4640 ref_path = kmalloc(sizeof(*ref_path), GFP_NOFS);
4646 for (loops = 0; ; loops++) {
4648 ret = btrfs_first_ref_path(trans, extent_root, ref_path,
4649 extent_key->objectid);
4651 ret = btrfs_next_ref_path(trans, extent_root, ref_path);
4658 if (ref_path->root_objectid == BTRFS_TREE_LOG_OBJECTID ||
4659 ref_path->root_objectid == BTRFS_TREE_RELOC_OBJECTID)
4662 found_root = read_ref_root(extent_root->fs_info, ref_path);
4663 BUG_ON(!found_root);
4665 * for reference counted tree, only process reference paths
4666 * rooted at the latest committed root.
4668 if (found_root->ref_cows &&
4669 ref_path->root_generation != found_root->root_key.offset)
4672 if (ref_path->owner_objectid >= BTRFS_FIRST_FREE_OBJECTID) {
4675 * copy data extents to new locations
4677 u64 group_start = group->key.objectid;
4678 ret = relocate_data_extent(reloc_inode,
4687 level = ref_path->owner_objectid;
4690 if (prev_block != ref_path->nodes[level]) {
4691 struct extent_buffer *eb;
4692 u64 block_start = ref_path->nodes[level];
4693 u64 block_size = btrfs_level_size(found_root, level);
4695 eb = read_tree_block(found_root, block_start,
4697 btrfs_tree_lock(eb);
4698 BUG_ON(level != btrfs_header_level(eb));
4701 btrfs_item_key_to_cpu(eb, &first_key, 0);
4703 btrfs_node_key_to_cpu(eb, &first_key, 0);
4705 btrfs_tree_unlock(eb);
4706 free_extent_buffer(eb);
4707 prev_block = block_start;
4710 if (ref_path->owner_objectid >= BTRFS_FIRST_FREE_OBJECTID &&
4713 * use fallback method to process the remaining
4717 u64 group_start = group->key.objectid;
4718 ret = get_new_locations(reloc_inode,
4726 btrfs_record_root_in_trans(found_root);
4727 ret = replace_one_extent(trans, found_root,
4729 &first_key, ref_path,
4730 new_extents, nr_extents);
4736 btrfs_record_root_in_trans(found_root);
4737 if (ref_path->owner_objectid < BTRFS_FIRST_FREE_OBJECTID) {
4738 ret = relocate_tree_block(trans, found_root, path,
4739 &first_key, ref_path);
4742 * try to update data extent references while
4743 * keeping metadata shared between snapshots.
4745 ret = relocate_one_path(trans, found_root, path,
4746 &first_key, ref_path,
4747 group, reloc_inode);
4754 btrfs_end_transaction(trans, extent_root);
4757 mutex_lock(&extent_root->fs_info->alloc_mutex);
4761 static u64 update_block_group_flags(struct btrfs_root *root, u64 flags)
4764 u64 stripped = BTRFS_BLOCK_GROUP_RAID0 |
4765 BTRFS_BLOCK_GROUP_RAID1 | BTRFS_BLOCK_GROUP_RAID10;
4767 num_devices = root->fs_info->fs_devices->num_devices;
4768 if (num_devices == 1) {
4769 stripped |= BTRFS_BLOCK_GROUP_DUP;
4770 stripped = flags & ~stripped;
4772 /* turn raid0 into single device chunks */
4773 if (flags & BTRFS_BLOCK_GROUP_RAID0)
4776 /* turn mirroring into duplication */
4777 if (flags & (BTRFS_BLOCK_GROUP_RAID1 |
4778 BTRFS_BLOCK_GROUP_RAID10))
4779 return stripped | BTRFS_BLOCK_GROUP_DUP;
4782 /* they already had raid on here, just return */
4783 if (flags & stripped)
4786 stripped |= BTRFS_BLOCK_GROUP_DUP;
4787 stripped = flags & ~stripped;
4789 /* switch duplicated blocks with raid1 */
4790 if (flags & BTRFS_BLOCK_GROUP_DUP)
4791 return stripped | BTRFS_BLOCK_GROUP_RAID1;
4793 /* turn single device chunks into raid0 */
4794 return stripped | BTRFS_BLOCK_GROUP_RAID0;
4799 int __alloc_chunk_for_shrink(struct btrfs_root *root,
4800 struct btrfs_block_group_cache *shrink_block_group,
4803 struct btrfs_trans_handle *trans;
4804 u64 new_alloc_flags;
4807 spin_lock(&shrink_block_group->lock);
4808 if (btrfs_block_group_used(&shrink_block_group->item) > 0) {
4809 spin_unlock(&shrink_block_group->lock);
4810 mutex_unlock(&root->fs_info->alloc_mutex);
4812 trans = btrfs_start_transaction(root, 1);
4813 mutex_lock(&root->fs_info->alloc_mutex);
4814 spin_lock(&shrink_block_group->lock);
4816 new_alloc_flags = update_block_group_flags(root,
4817 shrink_block_group->flags);
4818 if (new_alloc_flags != shrink_block_group->flags) {
4820 btrfs_block_group_used(&shrink_block_group->item);
4822 calc = shrink_block_group->key.offset;
4824 spin_unlock(&shrink_block_group->lock);
4826 do_chunk_alloc(trans, root->fs_info->extent_root,
4827 calc + 2 * 1024 * 1024, new_alloc_flags, force);
4829 mutex_unlock(&root->fs_info->alloc_mutex);
4830 btrfs_end_transaction(trans, root);
4831 mutex_lock(&root->fs_info->alloc_mutex);
4833 spin_unlock(&shrink_block_group->lock);
4837 static int __insert_orphan_inode(struct btrfs_trans_handle *trans,
4838 struct btrfs_root *root,
4839 u64 objectid, u64 size)
4841 struct btrfs_path *path;
4842 struct btrfs_inode_item *item;
4843 struct extent_buffer *leaf;
4846 path = btrfs_alloc_path();
4850 ret = btrfs_insert_empty_inode(trans, root, path, objectid);
4854 leaf = path->nodes[0];
4855 item = btrfs_item_ptr(leaf, path->slots[0], struct btrfs_inode_item);
4856 memset_extent_buffer(leaf, 0, (unsigned long)item, sizeof(*item));
4857 btrfs_set_inode_generation(leaf, item, 1);
4858 btrfs_set_inode_size(leaf, item, size);
4859 btrfs_set_inode_mode(leaf, item, S_IFREG | 0600);
4860 btrfs_set_inode_flags(leaf, item, BTRFS_INODE_NODATASUM);
4861 btrfs_mark_buffer_dirty(leaf);
4862 btrfs_release_path(root, path);
4864 btrfs_free_path(path);
4868 static struct inode noinline *create_reloc_inode(struct btrfs_fs_info *fs_info,
4869 struct btrfs_block_group_cache *group)
4871 struct inode *inode = NULL;
4872 struct btrfs_trans_handle *trans;
4873 struct btrfs_root *root;
4874 struct btrfs_key root_key;
4875 u64 objectid = BTRFS_FIRST_FREE_OBJECTID;
4878 root_key.objectid = BTRFS_DATA_RELOC_TREE_OBJECTID;
4879 root_key.type = BTRFS_ROOT_ITEM_KEY;
4880 root_key.offset = (u64)-1;
4881 root = btrfs_read_fs_root_no_name(fs_info, &root_key);
4883 return ERR_CAST(root);
4885 trans = btrfs_start_transaction(root, 1);
4888 err = btrfs_find_free_objectid(trans, root, objectid, &objectid);
4892 err = __insert_orphan_inode(trans, root, objectid, group->key.offset);
4895 err = btrfs_insert_file_extent(trans, root, objectid, 0, 0, 0,
4896 group->key.offset, 0, group->key.offset,
4900 inode = btrfs_iget_locked(root->fs_info->sb, objectid, root);
4901 if (inode->i_state & I_NEW) {
4902 BTRFS_I(inode)->root = root;
4903 BTRFS_I(inode)->location.objectid = objectid;
4904 BTRFS_I(inode)->location.type = BTRFS_INODE_ITEM_KEY;
4905 BTRFS_I(inode)->location.offset = 0;
4906 btrfs_read_locked_inode(inode);
4907 unlock_new_inode(inode);
4908 BUG_ON(is_bad_inode(inode));
4913 err = btrfs_orphan_add(trans, inode);
4915 btrfs_end_transaction(trans, root);
4919 inode = ERR_PTR(err);
4924 int btrfs_relocate_block_group(struct btrfs_root *root, u64 group_start)
4926 struct btrfs_trans_handle *trans;
4927 struct btrfs_path *path;
4928 struct btrfs_fs_info *info = root->fs_info;
4929 struct extent_buffer *leaf;
4930 struct inode *reloc_inode;
4931 struct btrfs_block_group_cache *block_group;
4932 struct btrfs_key key;
4940 root = root->fs_info->extent_root;
4942 block_group = btrfs_lookup_block_group(info, group_start);
4943 BUG_ON(!block_group);
4945 printk("btrfs relocating block group %llu flags %llu\n",
4946 (unsigned long long)block_group->key.objectid,
4947 (unsigned long long)block_group->flags);
4949 path = btrfs_alloc_path();
4952 reloc_inode = create_reloc_inode(info, block_group);
4953 BUG_ON(IS_ERR(reloc_inode));
4955 mutex_lock(&root->fs_info->alloc_mutex);
4957 __alloc_chunk_for_shrink(root, block_group, 1);
4958 block_group->ro = 1;
4959 block_group->space_info->total_bytes -= block_group->key.offset;
4961 mutex_unlock(&root->fs_info->alloc_mutex);
4963 btrfs_start_delalloc_inodes(info->tree_root);
4964 btrfs_wait_ordered_extents(info->tree_root, 0);
4968 key.objectid = block_group->key.objectid;
4971 cur_byte = key.objectid;
4973 trans = btrfs_start_transaction(info->tree_root, 1);
4974 btrfs_commit_transaction(trans, info->tree_root);
4976 mutex_lock(&root->fs_info->cleaner_mutex);
4977 btrfs_clean_old_snapshots(info->tree_root);
4978 btrfs_remove_leaf_refs(info->tree_root, (u64)-1, 1);
4979 mutex_unlock(&root->fs_info->cleaner_mutex);
4981 mutex_lock(&root->fs_info->alloc_mutex);
4984 ret = btrfs_search_slot(NULL, root, &key, path, 0, 0);
4988 leaf = path->nodes[0];
4989 nritems = btrfs_header_nritems(leaf);
4990 if (path->slots[0] >= nritems) {
4991 ret = btrfs_next_leaf(root, path);
4998 leaf = path->nodes[0];
4999 nritems = btrfs_header_nritems(leaf);
5002 btrfs_item_key_to_cpu(leaf, &key, path->slots[0]);
5004 if (key.objectid >= block_group->key.objectid +
5005 block_group->key.offset)
5008 if (progress && need_resched()) {
5009 btrfs_release_path(root, path);
5010 mutex_unlock(&root->fs_info->alloc_mutex);
5012 mutex_lock(&root->fs_info->alloc_mutex);
5018 if (btrfs_key_type(&key) != BTRFS_EXTENT_ITEM_KEY ||
5019 key.objectid + key.offset <= cur_byte) {
5025 cur_byte = key.objectid + key.offset;
5026 btrfs_release_path(root, path);
5028 __alloc_chunk_for_shrink(root, block_group, 0);
5029 ret = relocate_one_extent(root, path, &key, block_group,
5033 key.objectid = cur_byte;
5038 btrfs_release_path(root, path);
5039 mutex_unlock(&root->fs_info->alloc_mutex);
5042 btrfs_wait_ordered_range(reloc_inode, 0, (u64)-1);
5043 invalidate_mapping_pages(reloc_inode->i_mapping, 0, -1);
5044 WARN_ON(reloc_inode->i_mapping->nrpages);
5047 if (total_found > 0) {
5048 printk("btrfs found %llu extents in pass %d\n",
5049 (unsigned long long)total_found, pass);
5054 /* delete reloc_inode */
5057 /* unpin extents in this range */
5058 trans = btrfs_start_transaction(info->tree_root, 1);
5059 btrfs_commit_transaction(trans, info->tree_root);
5061 mutex_lock(&root->fs_info->alloc_mutex);
5063 spin_lock(&block_group->lock);
5064 WARN_ON(block_group->pinned > 0);
5065 WARN_ON(block_group->reserved > 0);
5066 WARN_ON(btrfs_block_group_used(&block_group->item) > 0);
5067 spin_unlock(&block_group->lock);
5070 mutex_unlock(&root->fs_info->alloc_mutex);
5071 btrfs_free_path(path);
5075 int find_first_block_group(struct btrfs_root *root, struct btrfs_path *path,
5076 struct btrfs_key *key)
5079 struct btrfs_key found_key;
5080 struct extent_buffer *leaf;
5083 ret = btrfs_search_slot(NULL, root, key, path, 0, 0);
5088 slot = path->slots[0];
5089 leaf = path->nodes[0];
5090 if (slot >= btrfs_header_nritems(leaf)) {
5091 ret = btrfs_next_leaf(root, path);
5098 btrfs_item_key_to_cpu(leaf, &found_key, slot);
5100 if (found_key.objectid >= key->objectid &&
5101 found_key.type == BTRFS_BLOCK_GROUP_ITEM_KEY) {
5112 int btrfs_free_block_groups(struct btrfs_fs_info *info)
5114 struct btrfs_block_group_cache *block_group;
5117 mutex_lock(&info->alloc_mutex);
5118 spin_lock(&info->block_group_cache_lock);
5119 while ((n = rb_last(&info->block_group_cache_tree)) != NULL) {
5120 block_group = rb_entry(n, struct btrfs_block_group_cache,
5123 spin_unlock(&info->block_group_cache_lock);
5124 btrfs_remove_free_space_cache(block_group);
5125 spin_lock(&info->block_group_cache_lock);
5127 rb_erase(&block_group->cache_node,
5128 &info->block_group_cache_tree);
5129 down_write(&block_group->space_info->groups_sem);
5130 list_del(&block_group->list);
5131 up_write(&block_group->space_info->groups_sem);
5134 spin_unlock(&info->block_group_cache_lock);
5135 mutex_unlock(&info->alloc_mutex);
5139 int btrfs_read_block_groups(struct btrfs_root *root)
5141 struct btrfs_path *path;
5143 struct btrfs_block_group_cache *cache;
5144 struct btrfs_fs_info *info = root->fs_info;
5145 struct btrfs_space_info *space_info;
5146 struct btrfs_key key;
5147 struct btrfs_key found_key;
5148 struct extent_buffer *leaf;
5150 root = info->extent_root;
5153 btrfs_set_key_type(&key, BTRFS_BLOCK_GROUP_ITEM_KEY);
5154 path = btrfs_alloc_path();
5158 mutex_lock(&root->fs_info->alloc_mutex);
5160 ret = find_first_block_group(root, path, &key);
5168 leaf = path->nodes[0];
5169 btrfs_item_key_to_cpu(leaf, &found_key, path->slots[0]);
5170 cache = kzalloc(sizeof(*cache), GFP_NOFS);
5176 spin_lock_init(&cache->lock);
5177 INIT_LIST_HEAD(&cache->list);
5178 read_extent_buffer(leaf, &cache->item,
5179 btrfs_item_ptr_offset(leaf, path->slots[0]),
5180 sizeof(cache->item));
5181 memcpy(&cache->key, &found_key, sizeof(found_key));
5183 key.objectid = found_key.objectid + found_key.offset;
5184 btrfs_release_path(root, path);
5185 cache->flags = btrfs_block_group_flags(&cache->item);
5187 ret = update_space_info(info, cache->flags, found_key.offset,
5188 btrfs_block_group_used(&cache->item),
5191 cache->space_info = space_info;
5192 down_write(&space_info->groups_sem);
5193 list_add_tail(&cache->list, &space_info->block_groups);
5194 up_write(&space_info->groups_sem);
5196 ret = btrfs_add_block_group_cache(root->fs_info, cache);
5199 set_avail_alloc_bits(root->fs_info, cache->flags);
5203 btrfs_free_path(path);
5204 mutex_unlock(&root->fs_info->alloc_mutex);
5208 int btrfs_make_block_group(struct btrfs_trans_handle *trans,
5209 struct btrfs_root *root, u64 bytes_used,
5210 u64 type, u64 chunk_objectid, u64 chunk_offset,
5214 struct btrfs_root *extent_root;
5215 struct btrfs_block_group_cache *cache;
5217 WARN_ON(!mutex_is_locked(&root->fs_info->alloc_mutex));
5218 extent_root = root->fs_info->extent_root;
5220 root->fs_info->last_trans_new_blockgroup = trans->transid;
5222 cache = kzalloc(sizeof(*cache), GFP_NOFS);
5226 cache->key.objectid = chunk_offset;
5227 cache->key.offset = size;
5228 spin_lock_init(&cache->lock);
5229 INIT_LIST_HEAD(&cache->list);
5230 btrfs_set_key_type(&cache->key, BTRFS_BLOCK_GROUP_ITEM_KEY);
5232 btrfs_set_block_group_used(&cache->item, bytes_used);
5233 btrfs_set_block_group_chunk_objectid(&cache->item, chunk_objectid);
5234 cache->flags = type;
5235 btrfs_set_block_group_flags(&cache->item, type);
5237 ret = update_space_info(root->fs_info, cache->flags, size, bytes_used,
5238 &cache->space_info);
5240 down_write(&cache->space_info->groups_sem);
5241 list_add_tail(&cache->list, &cache->space_info->block_groups);
5242 up_write(&cache->space_info->groups_sem);
5244 ret = btrfs_add_block_group_cache(root->fs_info, cache);
5247 ret = btrfs_insert_item(trans, extent_root, &cache->key, &cache->item,
5248 sizeof(cache->item));
5251 finish_current_insert(trans, extent_root);
5252 ret = del_pending_extents(trans, extent_root);
5254 set_avail_alloc_bits(extent_root->fs_info, type);
5259 int btrfs_remove_block_group(struct btrfs_trans_handle *trans,
5260 struct btrfs_root *root, u64 group_start)
5262 struct btrfs_path *path;
5263 struct btrfs_block_group_cache *block_group;
5264 struct btrfs_key key;
5267 BUG_ON(!mutex_is_locked(&root->fs_info->alloc_mutex));
5268 root = root->fs_info->extent_root;
5270 block_group = btrfs_lookup_block_group(root->fs_info, group_start);
5271 BUG_ON(!block_group);
5273 memcpy(&key, &block_group->key, sizeof(key));
5275 path = btrfs_alloc_path();
5278 btrfs_remove_free_space_cache(block_group);
5279 rb_erase(&block_group->cache_node,
5280 &root->fs_info->block_group_cache_tree);
5281 down_write(&block_group->space_info->groups_sem);
5282 list_del(&block_group->list);
5283 up_write(&block_group->space_info->groups_sem);
5286 memset(shrink_block_group, 0, sizeof(*shrink_block_group));
5287 kfree(shrink_block_group);
5290 ret = btrfs_search_slot(trans, root, &key, path, -1, 1);
5296 ret = btrfs_del_item(trans, root, path);
5298 btrfs_free_path(path);