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.
19 #include <linux/sched.h>
20 #include <linux/crc32c.h>
21 #include <linux/pagemap.h>
25 #include "print-tree.h"
26 #include "transaction.h"
29 #define BLOCK_GROUP_DATA EXTENT_WRITEBACK
30 #define BLOCK_GROUP_METADATA EXTENT_UPTODATE
31 #define BLOCK_GROUP_SYSTEM EXTENT_NEW
33 #define BLOCK_GROUP_DIRTY EXTENT_DIRTY
35 static int finish_current_insert(struct btrfs_trans_handle *trans, struct
36 btrfs_root *extent_root);
37 static int del_pending_extents(struct btrfs_trans_handle *trans, struct
38 btrfs_root *extent_root);
39 int btrfs_make_block_group(struct btrfs_trans_handle *trans,
40 struct btrfs_root *root, u64 bytes_used,
41 u64 type, u64 chunk_tree, u64 chunk_objectid,
45 static int cache_block_group(struct btrfs_root *root,
46 struct btrfs_block_group_cache *block_group)
48 struct btrfs_path *path;
51 struct extent_buffer *leaf;
52 struct extent_io_tree *free_space_cache;
62 root = root->fs_info->extent_root;
63 free_space_cache = &root->fs_info->free_space_cache;
65 if (block_group->cached)
68 path = btrfs_alloc_path();
73 first_free = block_group->key.objectid;
74 key.objectid = block_group->key.objectid;
76 btrfs_set_key_type(&key, BTRFS_EXTENT_ITEM_KEY);
77 ret = btrfs_search_slot(NULL, root, &key, path, 0, 0);
80 ret = btrfs_previous_item(root, path, 0, BTRFS_EXTENT_ITEM_KEY);
84 leaf = path->nodes[0];
85 btrfs_item_key_to_cpu(leaf, &key, path->slots[0]);
86 if (key.objectid + key.offset > first_free)
87 first_free = key.objectid + key.offset;
90 leaf = path->nodes[0];
91 slot = path->slots[0];
92 if (slot >= btrfs_header_nritems(leaf)) {
93 ret = btrfs_next_leaf(root, path);
102 btrfs_item_key_to_cpu(leaf, &key, slot);
103 if (key.objectid < block_group->key.objectid) {
106 if (key.objectid >= block_group->key.objectid +
107 block_group->key.offset) {
111 if (btrfs_key_type(&key) == BTRFS_EXTENT_ITEM_KEY) {
116 if (key.objectid > last) {
117 hole_size = key.objectid - last;
118 set_extent_dirty(free_space_cache, last,
119 last + hole_size - 1,
122 last = key.objectid + key.offset;
130 if (block_group->key.objectid +
131 block_group->key.offset > last) {
132 hole_size = block_group->key.objectid +
133 block_group->key.offset - last;
134 set_extent_dirty(free_space_cache, last,
135 last + hole_size - 1, GFP_NOFS);
137 block_group->cached = 1;
139 btrfs_free_path(path);
143 struct btrfs_block_group_cache *btrfs_lookup_block_group(struct
147 struct extent_io_tree *block_group_cache;
148 struct btrfs_block_group_cache *block_group = NULL;
154 block_group_cache = &info->block_group_cache;
155 ret = find_first_extent_bit(block_group_cache,
156 bytenr, &start, &end,
157 BLOCK_GROUP_DATA | BLOCK_GROUP_METADATA |
162 ret = get_state_private(block_group_cache, start, &ptr);
166 block_group = (struct btrfs_block_group_cache *)(unsigned long)ptr;
167 if (block_group->key.objectid <= bytenr && bytenr <
168 block_group->key.objectid + block_group->key.offset)
173 static int block_group_bits(struct btrfs_block_group_cache *cache, u64 bits)
175 return (cache->flags & bits);
178 static int noinline find_search_start(struct btrfs_root *root,
179 struct btrfs_block_group_cache **cache_ret,
180 u64 *start_ret, int num, int data)
183 struct btrfs_block_group_cache *cache = *cache_ret;
184 struct extent_io_tree *free_space_cache;
190 u64 search_start = *start_ret;
195 total_fs_bytes = btrfs_super_total_bytes(&root->fs_info->super_copy);
196 free_space_cache = &root->fs_info->free_space_cache;
199 ret = cache_block_group(root, cache);
203 last = max(search_start, cache->key.objectid);
204 if (!block_group_bits(cache, data)) {
209 ret = find_first_extent_bit(&root->fs_info->free_space_cache,
210 last, &start, &end, EXTENT_DIRTY);
217 start = max(last, start);
219 if (last - start < num) {
220 if (last == cache->key.objectid + cache->key.offset)
224 if (start + num > cache->key.objectid + cache->key.offset)
226 if (start + num > total_fs_bytes)
232 cache = btrfs_lookup_block_group(root->fs_info, search_start);
234 printk("Unable to find block group for %Lu\n", search_start);
240 last = cache->key.objectid + cache->key.offset;
242 cache = btrfs_lookup_block_group(root->fs_info, last);
243 if (!cache || cache->key.objectid >= total_fs_bytes) {
252 if (cache_miss && !cache->cached) {
253 cache_block_group(root, cache);
255 cache = btrfs_lookup_block_group(root->fs_info, last);
257 cache = btrfs_find_block_group(root, cache, last, data, 0);
265 static u64 div_factor(u64 num, int factor)
274 static int block_group_state_bits(u64 flags)
277 if (flags & BTRFS_BLOCK_GROUP_DATA)
278 bits |= BLOCK_GROUP_DATA;
279 if (flags & BTRFS_BLOCK_GROUP_METADATA)
280 bits |= BLOCK_GROUP_METADATA;
281 if (flags & BTRFS_BLOCK_GROUP_SYSTEM)
282 bits |= BLOCK_GROUP_SYSTEM;
286 struct btrfs_block_group_cache *btrfs_find_block_group(struct btrfs_root *root,
287 struct btrfs_block_group_cache
288 *hint, u64 search_start,
291 struct btrfs_block_group_cache *cache;
292 struct extent_io_tree *block_group_cache;
293 struct btrfs_block_group_cache *found_group = NULL;
294 struct btrfs_fs_info *info = root->fs_info;
308 block_group_cache = &info->block_group_cache;
309 total_fs_bytes = btrfs_super_total_bytes(&root->fs_info->super_copy);
314 bit = block_group_state_bits(data);
316 if (search_start && search_start < total_fs_bytes) {
317 struct btrfs_block_group_cache *shint;
318 shint = btrfs_lookup_block_group(info, search_start);
319 if (shint && block_group_bits(shint, data)) {
320 used = btrfs_block_group_used(&shint->item);
321 if (used + shint->pinned <
322 div_factor(shint->key.offset, factor)) {
327 if (hint && block_group_bits(hint, data) &&
328 hint->key.objectid < total_fs_bytes) {
329 used = btrfs_block_group_used(&hint->item);
330 if (used + hint->pinned <
331 div_factor(hint->key.offset, factor)) {
334 last = hint->key.objectid + hint->key.offset;
338 hint_last = max(hint->key.objectid, search_start);
340 hint_last = search_start;
342 if (hint_last >= total_fs_bytes)
343 hint_last = search_start;
348 ret = find_first_extent_bit(block_group_cache, last,
353 ret = get_state_private(block_group_cache, start, &ptr);
357 cache = (struct btrfs_block_group_cache *)(unsigned long)ptr;
358 last = cache->key.objectid + cache->key.offset;
359 used = btrfs_block_group_used(&cache->item);
361 if (cache->key.objectid > total_fs_bytes)
365 free_check = cache->key.offset;
367 free_check = div_factor(cache->key.offset, factor);
369 if (used + cache->pinned < free_check) {
374 printk("failed on cache %Lu used %Lu total %Lu\n",
375 cache->key.objectid, used, cache->key.offset);
388 static u64 hash_extent_ref(u64 root_objectid, u64 ref_generation,
389 u64 owner, u64 owner_offset)
391 u32 high_crc = ~(u32)0;
392 u32 low_crc = ~(u32)0;
395 lenum = cpu_to_le64(root_objectid);
396 high_crc = crc32c(high_crc, &lenum, sizeof(lenum));
397 lenum = cpu_to_le64(ref_generation);
398 low_crc = crc32c(low_crc, &lenum, sizeof(lenum));
399 if (owner >= BTRFS_FIRST_FREE_OBJECTID) {
400 lenum = cpu_to_le64(owner);
401 low_crc = crc32c(low_crc, &lenum, sizeof(lenum));
402 lenum = cpu_to_le64(owner_offset);
403 low_crc = crc32c(low_crc, &lenum, sizeof(lenum));
405 return ((u64)high_crc << 32) | (u64)low_crc;
408 static int match_extent_ref(struct extent_buffer *leaf,
409 struct btrfs_extent_ref *disk_ref,
410 struct btrfs_extent_ref *cpu_ref)
415 if (cpu_ref->objectid)
416 len = sizeof(*cpu_ref);
418 len = 2 * sizeof(u64);
419 ret = memcmp_extent_buffer(leaf, cpu_ref, (unsigned long)disk_ref,
424 static int noinline lookup_extent_backref(struct btrfs_trans_handle *trans,
425 struct btrfs_root *root,
426 struct btrfs_path *path, u64 bytenr,
428 u64 ref_generation, u64 owner,
429 u64 owner_offset, int del)
432 struct btrfs_key key;
433 struct btrfs_key found_key;
434 struct btrfs_extent_ref ref;
435 struct extent_buffer *leaf;
436 struct btrfs_extent_ref *disk_ref;
440 btrfs_set_stack_ref_root(&ref, root_objectid);
441 btrfs_set_stack_ref_generation(&ref, ref_generation);
442 btrfs_set_stack_ref_objectid(&ref, owner);
443 btrfs_set_stack_ref_offset(&ref, owner_offset);
445 hash = hash_extent_ref(root_objectid, ref_generation, owner,
448 key.objectid = bytenr;
449 key.type = BTRFS_EXTENT_REF_KEY;
452 ret = btrfs_search_slot(trans, root, &key, path,
456 leaf = path->nodes[0];
458 u32 nritems = btrfs_header_nritems(leaf);
459 if (path->slots[0] >= nritems) {
460 ret2 = btrfs_next_leaf(root, path);
463 leaf = path->nodes[0];
465 btrfs_item_key_to_cpu(leaf, &found_key, path->slots[0]);
466 if (found_key.objectid != bytenr ||
467 found_key.type != BTRFS_EXTENT_REF_KEY)
469 key.offset = found_key.offset;
471 btrfs_release_path(root, path);
475 disk_ref = btrfs_item_ptr(path->nodes[0],
477 struct btrfs_extent_ref);
478 if (match_extent_ref(path->nodes[0], disk_ref, &ref)) {
482 btrfs_item_key_to_cpu(leaf, &found_key, path->slots[0]);
483 key.offset = found_key.offset + 1;
484 btrfs_release_path(root, path);
491 * Back reference rules. Back refs have three main goals:
493 * 1) differentiate between all holders of references to an extent so that
494 * when a reference is dropped we can make sure it was a valid reference
495 * before freeing the extent.
497 * 2) Provide enough information to quickly find the holders of an extent
498 * if we notice a given block is corrupted or bad.
500 * 3) Make it easy to migrate blocks for FS shrinking or storage pool
501 * maintenance. This is actually the same as #2, but with a slightly
502 * different use case.
504 * File extents can be referenced by:
506 * - multiple snapshots, subvolumes, or different generations in one subvol
507 * - different files inside a single subvolume (in theory, not implemented yet)
508 * - different offsets inside a file (bookend extents in file.c)
510 * The extent ref structure has fields for:
512 * - Objectid of the subvolume root
513 * - Generation number of the tree holding the reference
514 * - objectid of the file holding the reference
515 * - offset in the file corresponding to the key holding the reference
517 * When a file extent is allocated the fields are filled in:
518 * (root_key.objectid, trans->transid, inode objectid, offset in file)
520 * When a leaf is cow'd new references are added for every file extent found
521 * in the leaf. It looks the same as the create case, but trans->transid
522 * will be different when the block is cow'd.
524 * (root_key.objectid, trans->transid, inode objectid, offset in file)
526 * When a file extent is removed either during snapshot deletion or file
527 * truncation, the corresponding back reference is found
530 * (btrfs_header_owner(leaf), btrfs_header_generation(leaf),
531 * inode objectid, offset in file)
533 * Btree extents can be referenced by:
535 * - Different subvolumes
536 * - Different generations of the same subvolume
538 * Storing sufficient information for a full reverse mapping of a btree
539 * block would require storing the lowest key of the block in the backref,
540 * and it would require updating that lowest key either before write out or
541 * every time it changed. Instead, the objectid of the lowest key is stored
542 * along with the level of the tree block. This provides a hint
543 * about where in the btree the block can be found. Searches through the
544 * btree only need to look for a pointer to that block, so they stop one
545 * level higher than the level recorded in the backref.
547 * Some btrees do not do reference counting on their extents. These
548 * include the extent tree and the tree of tree roots. Backrefs for these
549 * trees always have a generation of zero.
551 * When a tree block is created, back references are inserted:
553 * (root->root_key.objectid, trans->transid or zero, level, lowest_key_objectid)
555 * When a tree block is cow'd in a reference counted root,
556 * new back references are added for all the blocks it points to.
557 * These are of the form (trans->transid will have increased since creation):
559 * (root->root_key.objectid, trans->transid, level, lowest_key_objectid)
561 * Because the lowest_key_objectid and the level are just hints
562 * they are not used when backrefs are deleted. When a backref is deleted:
564 * if backref was for a tree root:
565 * root_objectid = root->root_key.objectid
567 * root_objectid = btrfs_header_owner(parent)
569 * (root_objectid, btrfs_header_generation(parent) or zero, 0, 0)
571 * Back Reference Key hashing:
573 * Back references have four fields, each 64 bits long. Unfortunately,
574 * This is hashed into a single 64 bit number and placed into the key offset.
575 * The key objectid corresponds to the first byte in the extent, and the
576 * key type is set to BTRFS_EXTENT_REF_KEY
578 int btrfs_insert_extent_backref(struct btrfs_trans_handle *trans,
579 struct btrfs_root *root,
580 struct btrfs_path *path, u64 bytenr,
581 u64 root_objectid, u64 ref_generation,
582 u64 owner, u64 owner_offset)
585 struct btrfs_key key;
586 struct btrfs_extent_ref ref;
587 struct btrfs_extent_ref *disk_ref;
590 btrfs_set_stack_ref_root(&ref, root_objectid);
591 btrfs_set_stack_ref_generation(&ref, ref_generation);
592 btrfs_set_stack_ref_objectid(&ref, owner);
593 btrfs_set_stack_ref_offset(&ref, owner_offset);
595 hash = hash_extent_ref(root_objectid, ref_generation, owner,
598 key.objectid = bytenr;
599 key.type = BTRFS_EXTENT_REF_KEY;
601 ret = btrfs_insert_empty_item(trans, root, path, &key, sizeof(ref));
602 while (ret == -EEXIST) {
603 disk_ref = btrfs_item_ptr(path->nodes[0], path->slots[0],
604 struct btrfs_extent_ref);
605 if (match_extent_ref(path->nodes[0], disk_ref, &ref))
608 btrfs_release_path(root, path);
609 ret = btrfs_insert_empty_item(trans, root, path, &key,
614 disk_ref = btrfs_item_ptr(path->nodes[0], path->slots[0],
615 struct btrfs_extent_ref);
616 write_extent_buffer(path->nodes[0], &ref, (unsigned long)disk_ref,
618 btrfs_mark_buffer_dirty(path->nodes[0]);
620 btrfs_release_path(root, path);
624 int btrfs_inc_extent_ref(struct btrfs_trans_handle *trans,
625 struct btrfs_root *root,
626 u64 bytenr, u64 num_bytes,
627 u64 root_objectid, u64 ref_generation,
628 u64 owner, u64 owner_offset)
630 struct btrfs_path *path;
632 struct btrfs_key key;
633 struct extent_buffer *l;
634 struct btrfs_extent_item *item;
637 WARN_ON(num_bytes < root->sectorsize);
638 path = btrfs_alloc_path();
643 key.objectid = bytenr;
644 btrfs_set_key_type(&key, BTRFS_EXTENT_ITEM_KEY);
645 key.offset = num_bytes;
646 ret = btrfs_search_slot(trans, root->fs_info->extent_root, &key, path,
655 item = btrfs_item_ptr(l, path->slots[0], struct btrfs_extent_item);
656 refs = btrfs_extent_refs(l, item);
657 btrfs_set_extent_refs(l, item, refs + 1);
658 btrfs_mark_buffer_dirty(path->nodes[0]);
660 btrfs_release_path(root->fs_info->extent_root, path);
663 ret = btrfs_insert_extent_backref(trans, root->fs_info->extent_root,
664 path, bytenr, root_objectid,
665 ref_generation, owner, owner_offset);
667 finish_current_insert(trans, root->fs_info->extent_root);
668 del_pending_extents(trans, root->fs_info->extent_root);
670 btrfs_free_path(path);
674 int btrfs_extent_post_op(struct btrfs_trans_handle *trans,
675 struct btrfs_root *root)
677 finish_current_insert(trans, root->fs_info->extent_root);
678 del_pending_extents(trans, root->fs_info->extent_root);
682 static int lookup_extent_ref(struct btrfs_trans_handle *trans,
683 struct btrfs_root *root, u64 bytenr,
684 u64 num_bytes, u32 *refs)
686 struct btrfs_path *path;
688 struct btrfs_key key;
689 struct extent_buffer *l;
690 struct btrfs_extent_item *item;
692 WARN_ON(num_bytes < root->sectorsize);
693 path = btrfs_alloc_path();
695 key.objectid = bytenr;
696 key.offset = num_bytes;
697 btrfs_set_key_type(&key, BTRFS_EXTENT_ITEM_KEY);
698 ret = btrfs_search_slot(trans, root->fs_info->extent_root, &key, path,
703 btrfs_print_leaf(root, path->nodes[0]);
704 printk("failed to find block number %Lu\n", bytenr);
708 item = btrfs_item_ptr(l, path->slots[0], struct btrfs_extent_item);
709 *refs = btrfs_extent_refs(l, item);
711 btrfs_free_path(path);
715 u32 btrfs_count_snapshots_in_path(struct btrfs_root *root,
716 struct btrfs_path *count_path,
719 struct btrfs_root *extent_root = root->fs_info->extent_root;
720 struct btrfs_path *path;
723 u64 root_objectid = root->root_key.objectid;
728 struct btrfs_key key;
729 struct btrfs_key found_key;
730 struct extent_buffer *l;
731 struct btrfs_extent_item *item;
732 struct btrfs_extent_ref *ref_item;
735 path = btrfs_alloc_path();
738 bytenr = first_extent;
740 bytenr = count_path->nodes[level]->start;
743 key.objectid = bytenr;
746 btrfs_set_key_type(&key, BTRFS_EXTENT_ITEM_KEY);
747 ret = btrfs_search_slot(NULL, extent_root, &key, path, 0, 0);
753 btrfs_item_key_to_cpu(l, &found_key, path->slots[0]);
755 if (found_key.objectid != bytenr ||
756 found_key.type != BTRFS_EXTENT_ITEM_KEY) {
760 item = btrfs_item_ptr(l, path->slots[0], struct btrfs_extent_item);
763 nritems = btrfs_header_nritems(l);
764 if (path->slots[0] >= nritems) {
765 ret = btrfs_next_leaf(extent_root, path);
770 btrfs_item_key_to_cpu(l, &found_key, path->slots[0]);
771 if (found_key.objectid != bytenr)
774 if (found_key.type != BTRFS_EXTENT_REF_KEY) {
780 ref_item = btrfs_item_ptr(l, path->slots[0],
781 struct btrfs_extent_ref);
782 found_objectid = btrfs_ref_root(l, ref_item);
784 if (found_objectid != root_objectid) {
791 if (cur_count == 0) {
795 if (level >= 0 && root->node == count_path->nodes[level])
798 btrfs_release_path(root, path);
802 btrfs_free_path(path);
805 int btrfs_inc_root_ref(struct btrfs_trans_handle *trans,
806 struct btrfs_root *root, u64 owner_objectid)
812 struct btrfs_disk_key disk_key;
814 level = btrfs_header_level(root->node);
815 generation = trans->transid;
816 nritems = btrfs_header_nritems(root->node);
819 btrfs_item_key(root->node, &disk_key, 0);
821 btrfs_node_key(root->node, &disk_key, 0);
822 key_objectid = btrfs_disk_key_objectid(&disk_key);
826 return btrfs_inc_extent_ref(trans, root, root->node->start,
827 root->node->len, owner_objectid,
828 generation, level, key_objectid);
831 int btrfs_inc_ref(struct btrfs_trans_handle *trans, struct btrfs_root *root,
832 struct extent_buffer *buf)
836 struct btrfs_key key;
837 struct btrfs_file_extent_item *fi;
846 level = btrfs_header_level(buf);
847 nritems = btrfs_header_nritems(buf);
848 for (i = 0; i < nritems; i++) {
851 btrfs_item_key_to_cpu(buf, &key, i);
852 if (btrfs_key_type(&key) != BTRFS_EXTENT_DATA_KEY)
854 fi = btrfs_item_ptr(buf, i,
855 struct btrfs_file_extent_item);
856 if (btrfs_file_extent_type(buf, fi) ==
857 BTRFS_FILE_EXTENT_INLINE)
859 disk_bytenr = btrfs_file_extent_disk_bytenr(buf, fi);
860 if (disk_bytenr == 0)
862 ret = btrfs_inc_extent_ref(trans, root, disk_bytenr,
863 btrfs_file_extent_disk_num_bytes(buf, fi),
864 root->root_key.objectid, trans->transid,
865 key.objectid, key.offset);
871 bytenr = btrfs_node_blockptr(buf, i);
872 btrfs_node_key_to_cpu(buf, &key, i);
873 ret = btrfs_inc_extent_ref(trans, root, bytenr,
874 btrfs_level_size(root, level - 1),
875 root->root_key.objectid,
877 level - 1, key.objectid);
888 for (i =0; i < faili; i++) {
891 btrfs_item_key_to_cpu(buf, &key, i);
892 if (btrfs_key_type(&key) != BTRFS_EXTENT_DATA_KEY)
894 fi = btrfs_item_ptr(buf, i,
895 struct btrfs_file_extent_item);
896 if (btrfs_file_extent_type(buf, fi) ==
897 BTRFS_FILE_EXTENT_INLINE)
899 disk_bytenr = btrfs_file_extent_disk_bytenr(buf, fi);
900 if (disk_bytenr == 0)
902 err = btrfs_free_extent(trans, root, disk_bytenr,
903 btrfs_file_extent_disk_num_bytes(buf,
907 bytenr = btrfs_node_blockptr(buf, i);
908 err = btrfs_free_extent(trans, root, bytenr,
909 btrfs_level_size(root, level - 1), 0);
917 static int write_one_cache_group(struct btrfs_trans_handle *trans,
918 struct btrfs_root *root,
919 struct btrfs_path *path,
920 struct btrfs_block_group_cache *cache)
924 struct btrfs_root *extent_root = root->fs_info->extent_root;
926 struct extent_buffer *leaf;
928 ret = btrfs_search_slot(trans, extent_root, &cache->key, path, 0, 1);
933 leaf = path->nodes[0];
934 bi = btrfs_item_ptr_offset(leaf, path->slots[0]);
935 write_extent_buffer(leaf, &cache->item, bi, sizeof(cache->item));
936 btrfs_mark_buffer_dirty(leaf);
937 btrfs_release_path(extent_root, path);
939 finish_current_insert(trans, extent_root);
940 pending_ret = del_pending_extents(trans, extent_root);
949 int btrfs_write_dirty_block_groups(struct btrfs_trans_handle *trans,
950 struct btrfs_root *root)
952 struct extent_io_tree *block_group_cache;
953 struct btrfs_block_group_cache *cache;
957 struct btrfs_path *path;
963 block_group_cache = &root->fs_info->block_group_cache;
964 path = btrfs_alloc_path();
969 ret = find_first_extent_bit(block_group_cache, last,
970 &start, &end, BLOCK_GROUP_DIRTY);
975 ret = get_state_private(block_group_cache, start, &ptr);
979 cache = (struct btrfs_block_group_cache *)(unsigned long)ptr;
980 err = write_one_cache_group(trans, root,
983 * if we fail to write the cache group, we want
984 * to keep it marked dirty in hopes that a later
991 clear_extent_bits(block_group_cache, start, end,
992 BLOCK_GROUP_DIRTY, GFP_NOFS);
994 btrfs_free_path(path);
998 static struct btrfs_space_info *__find_space_info(struct btrfs_fs_info *info,
1001 struct list_head *head = &info->space_info;
1002 struct list_head *cur;
1003 struct btrfs_space_info *found;
1004 list_for_each(cur, head) {
1005 found = list_entry(cur, struct btrfs_space_info, list);
1006 if (found->flags == flags)
1013 static int do_chunk_alloc(struct btrfs_trans_handle *trans,
1014 struct btrfs_root *extent_root, u64 alloc_bytes,
1017 struct btrfs_space_info *space_info;
1023 space_info = __find_space_info(extent_root->fs_info, flags);
1024 BUG_ON(!space_info);
1026 if (space_info->full)
1029 thresh = div_factor(space_info->total_bytes, 7);
1030 if ((space_info->bytes_used + space_info->bytes_pinned + alloc_bytes) <
1034 ret = btrfs_alloc_chunk(trans, extent_root, &start, &num_bytes, flags);
1035 if (ret == -ENOSPC) {
1036 printk("space info full %Lu\n", flags);
1037 space_info->full = 1;
1043 ret = btrfs_make_block_group(trans, extent_root, 0, flags,
1044 extent_root->fs_info->chunk_root->root_key.objectid,
1050 static int update_block_group(struct btrfs_trans_handle *trans,
1051 struct btrfs_root *root,
1052 u64 bytenr, u64 num_bytes, int alloc,
1055 struct btrfs_block_group_cache *cache;
1056 struct btrfs_fs_info *info = root->fs_info;
1057 u64 total = num_bytes;
1064 cache = btrfs_lookup_block_group(info, bytenr);
1068 byte_in_group = bytenr - cache->key.objectid;
1069 WARN_ON(byte_in_group > cache->key.offset);
1070 start = cache->key.objectid;
1071 end = start + cache->key.offset - 1;
1072 set_extent_bits(&info->block_group_cache, start, end,
1073 BLOCK_GROUP_DIRTY, GFP_NOFS);
1075 old_val = btrfs_block_group_used(&cache->item);
1076 num_bytes = min(total, cache->key.offset - byte_in_group);
1078 old_val += num_bytes;
1079 cache->space_info->bytes_used += num_bytes;
1081 old_val -= num_bytes;
1082 cache->space_info->bytes_used -= num_bytes;
1084 set_extent_dirty(&info->free_space_cache,
1085 bytenr, bytenr + num_bytes - 1,
1089 btrfs_set_block_group_used(&cache->item, old_val);
1091 bytenr += num_bytes;
1096 static int update_pinned_extents(struct btrfs_root *root,
1097 u64 bytenr, u64 num, int pin)
1100 struct btrfs_block_group_cache *cache;
1101 struct btrfs_fs_info *fs_info = root->fs_info;
1104 set_extent_dirty(&fs_info->pinned_extents,
1105 bytenr, bytenr + num - 1, GFP_NOFS);
1107 clear_extent_dirty(&fs_info->pinned_extents,
1108 bytenr, bytenr + num - 1, GFP_NOFS);
1111 cache = btrfs_lookup_block_group(fs_info, bytenr);
1113 len = min(num, cache->key.offset -
1114 (bytenr - cache->key.objectid));
1116 cache->pinned += len;
1117 cache->space_info->bytes_pinned += len;
1118 fs_info->total_pinned += len;
1120 cache->pinned -= len;
1121 cache->space_info->bytes_pinned -= len;
1122 fs_info->total_pinned -= len;
1130 int btrfs_copy_pinned(struct btrfs_root *root, struct extent_io_tree *copy)
1135 struct extent_io_tree *pinned_extents = &root->fs_info->pinned_extents;
1139 ret = find_first_extent_bit(pinned_extents, last,
1140 &start, &end, EXTENT_DIRTY);
1143 set_extent_dirty(copy, start, end, GFP_NOFS);
1149 int btrfs_finish_extent_commit(struct btrfs_trans_handle *trans,
1150 struct btrfs_root *root,
1151 struct extent_io_tree *unpin)
1156 struct extent_io_tree *free_space_cache;
1157 free_space_cache = &root->fs_info->free_space_cache;
1160 ret = find_first_extent_bit(unpin, 0, &start, &end,
1164 update_pinned_extents(root, start, end + 1 - start, 0);
1165 clear_extent_dirty(unpin, start, end, GFP_NOFS);
1166 set_extent_dirty(free_space_cache, start, end, GFP_NOFS);
1171 static int finish_current_insert(struct btrfs_trans_handle *trans,
1172 struct btrfs_root *extent_root)
1176 struct btrfs_fs_info *info = extent_root->fs_info;
1177 struct extent_buffer *eb;
1178 struct btrfs_path *path;
1179 struct btrfs_key ins;
1180 struct btrfs_disk_key first;
1181 struct btrfs_extent_item extent_item;
1186 btrfs_set_stack_extent_refs(&extent_item, 1);
1187 btrfs_set_key_type(&ins, BTRFS_EXTENT_ITEM_KEY);
1188 path = btrfs_alloc_path();
1191 ret = find_first_extent_bit(&info->extent_ins, 0, &start,
1192 &end, EXTENT_LOCKED);
1196 ins.objectid = start;
1197 ins.offset = end + 1 - start;
1198 err = btrfs_insert_item(trans, extent_root, &ins,
1199 &extent_item, sizeof(extent_item));
1200 clear_extent_bits(&info->extent_ins, start, end, EXTENT_LOCKED,
1202 eb = read_tree_block(extent_root, ins.objectid, ins.offset);
1203 level = btrfs_header_level(eb);
1205 btrfs_item_key(eb, &first, 0);
1207 btrfs_node_key(eb, &first, 0);
1209 err = btrfs_insert_extent_backref(trans, extent_root, path,
1210 start, extent_root->root_key.objectid,
1212 btrfs_disk_key_objectid(&first));
1214 free_extent_buffer(eb);
1216 btrfs_free_path(path);
1220 static int pin_down_bytes(struct btrfs_root *root, u64 bytenr, u32 num_bytes,
1224 struct extent_buffer *buf;
1227 buf = btrfs_find_tree_block(root, bytenr, num_bytes);
1229 if (btrfs_buffer_uptodate(buf)) {
1231 root->fs_info->running_transaction->transid;
1232 u64 header_transid =
1233 btrfs_header_generation(buf);
1234 if (header_transid == transid) {
1235 clean_tree_block(NULL, root, buf);
1236 free_extent_buffer(buf);
1240 free_extent_buffer(buf);
1242 update_pinned_extents(root, bytenr, num_bytes, 1);
1244 set_extent_bits(&root->fs_info->pending_del,
1245 bytenr, bytenr + num_bytes - 1,
1246 EXTENT_LOCKED, GFP_NOFS);
1253 * remove an extent from the root, returns 0 on success
1255 static int __free_extent(struct btrfs_trans_handle *trans, struct btrfs_root
1256 *root, u64 bytenr, u64 num_bytes,
1257 u64 root_objectid, u64 ref_generation,
1258 u64 owner_objectid, u64 owner_offset, int pin,
1261 struct btrfs_path *path;
1262 struct btrfs_key key;
1263 struct btrfs_fs_info *info = root->fs_info;
1264 struct btrfs_root *extent_root = info->extent_root;
1265 struct extent_buffer *leaf;
1267 int extent_slot = 0;
1268 int found_extent = 0;
1270 struct btrfs_extent_item *ei;
1273 key.objectid = bytenr;
1274 btrfs_set_key_type(&key, BTRFS_EXTENT_ITEM_KEY);
1275 key.offset = num_bytes;
1276 path = btrfs_alloc_path();
1281 ret = lookup_extent_backref(trans, extent_root, path,
1282 bytenr, root_objectid,
1284 owner_objectid, owner_offset, 1);
1286 struct btrfs_key found_key;
1287 extent_slot = path->slots[0];
1288 while(extent_slot > 0) {
1290 btrfs_item_key_to_cpu(path->nodes[0], &found_key,
1292 if (found_key.objectid != bytenr)
1294 if (found_key.type == BTRFS_EXTENT_ITEM_KEY &&
1295 found_key.offset == num_bytes) {
1299 if (path->slots[0] - extent_slot > 5)
1303 ret = btrfs_del_item(trans, extent_root, path);
1305 btrfs_print_leaf(extent_root, path->nodes[0]);
1307 printk("Unable to find ref byte nr %Lu root %Lu "
1308 " gen %Lu owner %Lu offset %Lu\n", bytenr,
1309 root_objectid, ref_generation, owner_objectid,
1312 if (!found_extent) {
1313 btrfs_release_path(extent_root, path);
1314 ret = btrfs_search_slot(trans, extent_root, &key, path, -1, 1);
1318 extent_slot = path->slots[0];
1321 leaf = path->nodes[0];
1322 ei = btrfs_item_ptr(leaf, extent_slot,
1323 struct btrfs_extent_item);
1324 refs = btrfs_extent_refs(leaf, ei);
1327 btrfs_set_extent_refs(leaf, ei, refs);
1329 btrfs_mark_buffer_dirty(leaf);
1331 if (refs == 0 && found_extent && path->slots[0] == extent_slot + 1) {
1332 /* if the back ref and the extent are next to each other
1333 * they get deleted below in one shot
1335 path->slots[0] = extent_slot;
1337 } else if (found_extent) {
1338 /* otherwise delete the extent back ref */
1339 ret = btrfs_del_item(trans, extent_root, path);
1341 /* if refs are 0, we need to setup the path for deletion */
1343 btrfs_release_path(extent_root, path);
1344 ret = btrfs_search_slot(trans, extent_root, &key, path,
1357 ret = pin_down_bytes(root, bytenr, num_bytes, 0);
1363 /* block accounting for super block */
1364 super_used = btrfs_super_bytes_used(&info->super_copy);
1365 btrfs_set_super_bytes_used(&info->super_copy,
1366 super_used - num_bytes);
1368 /* block accounting for root item */
1369 root_used = btrfs_root_used(&root->root_item);
1370 btrfs_set_root_used(&root->root_item,
1371 root_used - num_bytes);
1372 ret = btrfs_del_items(trans, extent_root, path, path->slots[0],
1377 ret = update_block_group(trans, root, bytenr, num_bytes, 0,
1381 btrfs_free_path(path);
1382 finish_current_insert(trans, extent_root);
1387 * find all the blocks marked as pending in the radix tree and remove
1388 * them from the extent map
1390 static int del_pending_extents(struct btrfs_trans_handle *trans, struct
1391 btrfs_root *extent_root)
1397 struct extent_io_tree *pending_del;
1398 struct extent_io_tree *pinned_extents;
1400 pending_del = &extent_root->fs_info->pending_del;
1401 pinned_extents = &extent_root->fs_info->pinned_extents;
1404 ret = find_first_extent_bit(pending_del, 0, &start, &end,
1408 update_pinned_extents(extent_root, start, end + 1 - start, 1);
1409 clear_extent_bits(pending_del, start, end, EXTENT_LOCKED,
1411 ret = __free_extent(trans, extent_root,
1412 start, end + 1 - start,
1413 extent_root->root_key.objectid,
1422 * remove an extent from the root, returns 0 on success
1424 int btrfs_free_extent(struct btrfs_trans_handle *trans, struct btrfs_root
1425 *root, u64 bytenr, u64 num_bytes,
1426 u64 root_objectid, u64 ref_generation,
1427 u64 owner_objectid, u64 owner_offset, int pin)
1429 struct btrfs_root *extent_root = root->fs_info->extent_root;
1433 WARN_ON(num_bytes < root->sectorsize);
1434 if (!root->ref_cows)
1437 if (root == extent_root) {
1438 pin_down_bytes(root, bytenr, num_bytes, 1);
1441 ret = __free_extent(trans, root, bytenr, num_bytes, root_objectid,
1442 ref_generation, owner_objectid, owner_offset,
1444 pending_ret = del_pending_extents(trans, root->fs_info->extent_root);
1445 return ret ? ret : pending_ret;
1448 static u64 stripe_align(struct btrfs_root *root, u64 val)
1450 u64 mask = ((u64)root->stripesize - 1);
1451 u64 ret = (val + mask) & ~mask;
1456 * walks the btree of allocated extents and find a hole of a given size.
1457 * The key ins is changed to record the hole:
1458 * ins->objectid == block start
1459 * ins->flags = BTRFS_EXTENT_ITEM_KEY
1460 * ins->offset == number of blocks
1461 * Any available blocks before search_start are skipped.
1463 static int noinline find_free_extent(struct btrfs_trans_handle *trans,
1464 struct btrfs_root *orig_root,
1465 u64 num_bytes, u64 empty_size,
1466 u64 search_start, u64 search_end,
1467 u64 hint_byte, struct btrfs_key *ins,
1468 u64 exclude_start, u64 exclude_nr,
1472 u64 orig_search_start = search_start;
1473 struct btrfs_root * root = orig_root->fs_info->extent_root;
1474 struct btrfs_fs_info *info = root->fs_info;
1475 u64 total_needed = num_bytes;
1476 struct btrfs_block_group_cache *block_group;
1480 WARN_ON(num_bytes < root->sectorsize);
1481 btrfs_set_key_type(ins, BTRFS_EXTENT_ITEM_KEY);
1483 if (search_end == (u64)-1)
1484 search_end = btrfs_super_total_bytes(&info->super_copy);
1487 block_group = btrfs_lookup_block_group(info, hint_byte);
1489 hint_byte = search_start;
1490 block_group = btrfs_find_block_group(root, block_group,
1491 hint_byte, data, 1);
1493 block_group = btrfs_find_block_group(root,
1495 search_start, data, 1);
1498 total_needed += empty_size;
1502 block_group = btrfs_lookup_block_group(info, search_start);
1504 block_group = btrfs_lookup_block_group(info,
1507 ret = find_search_start(root, &block_group, &search_start,
1508 total_needed, data);
1512 search_start = stripe_align(root, search_start);
1513 ins->objectid = search_start;
1514 ins->offset = num_bytes;
1516 if (ins->objectid + num_bytes >= search_end)
1519 if (ins->objectid + num_bytes >
1520 block_group->key.objectid + block_group->key.offset) {
1521 search_start = block_group->key.objectid +
1522 block_group->key.offset;
1526 if (test_range_bit(&info->extent_ins, ins->objectid,
1527 ins->objectid + num_bytes -1, EXTENT_LOCKED, 0)) {
1528 search_start = ins->objectid + num_bytes;
1532 if (test_range_bit(&info->pinned_extents, ins->objectid,
1533 ins->objectid + num_bytes -1, EXTENT_DIRTY, 0)) {
1534 search_start = ins->objectid + num_bytes;
1538 if (exclude_nr > 0 && (ins->objectid + num_bytes > exclude_start &&
1539 ins->objectid < exclude_start + exclude_nr)) {
1540 search_start = exclude_start + exclude_nr;
1544 if (!(data & BTRFS_BLOCK_GROUP_DATA)) {
1545 block_group = btrfs_lookup_block_group(info, ins->objectid);
1547 trans->block_group = block_group;
1549 ins->offset = num_bytes;
1553 if (search_start + num_bytes >= search_end) {
1555 search_start = orig_search_start;
1562 total_needed -= empty_size;
1567 block_group = btrfs_lookup_block_group(info, search_start);
1569 block_group = btrfs_find_block_group(root, block_group,
1570 search_start, data, 0);
1577 * finds a free extent and does all the dirty work required for allocation
1578 * returns the key for the extent through ins, and a tree buffer for
1579 * the first block of the extent through buf.
1581 * returns 0 if everything worked, non-zero otherwise.
1583 int btrfs_alloc_extent(struct btrfs_trans_handle *trans,
1584 struct btrfs_root *root,
1585 u64 num_bytes, u64 root_objectid, u64 ref_generation,
1586 u64 owner, u64 owner_offset,
1587 u64 empty_size, u64 hint_byte,
1588 u64 search_end, struct btrfs_key *ins, int data)
1594 u64 search_start = 0;
1597 struct btrfs_fs_info *info = root->fs_info;
1598 struct btrfs_root *extent_root = info->extent_root;
1599 struct btrfs_extent_item *extent_item;
1600 struct btrfs_extent_ref *ref;
1601 struct btrfs_path *path;
1602 struct btrfs_key keys[2];
1605 data = BTRFS_BLOCK_GROUP_DATA;
1606 } else if (root == root->fs_info->chunk_root) {
1607 data = BTRFS_BLOCK_GROUP_SYSTEM;
1609 data = BTRFS_BLOCK_GROUP_METADATA;
1612 if (root->ref_cows) {
1613 if (data != BTRFS_BLOCK_GROUP_METADATA) {
1614 ret = do_chunk_alloc(trans, root->fs_info->extent_root,
1616 BTRFS_BLOCK_GROUP_METADATA);
1619 ret = do_chunk_alloc(trans, root->fs_info->extent_root,
1624 new_hint = max(hint_byte, root->fs_info->alloc_start);
1625 if (new_hint < btrfs_super_total_bytes(&info->super_copy))
1626 hint_byte = new_hint;
1628 WARN_ON(num_bytes < root->sectorsize);
1629 ret = find_free_extent(trans, root, num_bytes, empty_size,
1630 search_start, search_end, hint_byte, ins,
1631 trans->alloc_exclude_start,
1632 trans->alloc_exclude_nr, data);
1637 /* block accounting for super block */
1638 super_used = btrfs_super_bytes_used(&info->super_copy);
1639 btrfs_set_super_bytes_used(&info->super_copy, super_used + num_bytes);
1641 /* block accounting for root item */
1642 root_used = btrfs_root_used(&root->root_item);
1643 btrfs_set_root_used(&root->root_item, root_used + num_bytes);
1645 clear_extent_dirty(&root->fs_info->free_space_cache,
1646 ins->objectid, ins->objectid + ins->offset - 1,
1649 if (root == extent_root) {
1650 set_extent_bits(&root->fs_info->extent_ins, ins->objectid,
1651 ins->objectid + ins->offset - 1,
1652 EXTENT_LOCKED, GFP_NOFS);
1656 WARN_ON(trans->alloc_exclude_nr);
1657 trans->alloc_exclude_start = ins->objectid;
1658 trans->alloc_exclude_nr = ins->offset;
1660 memcpy(&keys[0], ins, sizeof(*ins));
1661 keys[1].offset = hash_extent_ref(root_objectid, ref_generation,
1662 owner, owner_offset);
1663 keys[1].objectid = ins->objectid;
1664 keys[1].type = BTRFS_EXTENT_REF_KEY;
1665 sizes[0] = sizeof(*extent_item);
1666 sizes[1] = sizeof(*ref);
1668 path = btrfs_alloc_path();
1671 ret = btrfs_insert_empty_items(trans, extent_root, path, keys,
1675 extent_item = btrfs_item_ptr(path->nodes[0], path->slots[0],
1676 struct btrfs_extent_item);
1677 btrfs_set_extent_refs(path->nodes[0], extent_item, 1);
1678 ref = btrfs_item_ptr(path->nodes[0], path->slots[0] + 1,
1679 struct btrfs_extent_ref);
1681 btrfs_set_ref_root(path->nodes[0], ref, root_objectid);
1682 btrfs_set_ref_generation(path->nodes[0], ref, ref_generation);
1683 btrfs_set_ref_objectid(path->nodes[0], ref, owner);
1684 btrfs_set_ref_offset(path->nodes[0], ref, owner_offset);
1686 btrfs_mark_buffer_dirty(path->nodes[0]);
1688 trans->alloc_exclude_start = 0;
1689 trans->alloc_exclude_nr = 0;
1690 btrfs_free_path(path);
1691 finish_current_insert(trans, extent_root);
1692 pending_ret = del_pending_extents(trans, extent_root);
1702 ret = update_block_group(trans, root, ins->objectid, ins->offset, 1, 0);
1704 printk("update block group failed for %Lu %Lu\n",
1705 ins->objectid, ins->offset);
1712 * helper function to allocate a block for a given tree
1713 * returns the tree buffer or NULL.
1715 struct extent_buffer *btrfs_alloc_free_block(struct btrfs_trans_handle *trans,
1716 struct btrfs_root *root,
1718 u64 root_objectid, u64 hint,
1724 ref_generation = trans->transid;
1729 return __btrfs_alloc_free_block(trans, root, blocksize, root_objectid,
1730 ref_generation, 0, 0, hint, empty_size);
1734 * helper function to allocate a block for a given tree
1735 * returns the tree buffer or NULL.
1737 struct extent_buffer *__btrfs_alloc_free_block(struct btrfs_trans_handle *trans,
1738 struct btrfs_root *root,
1747 struct btrfs_key ins;
1749 struct extent_buffer *buf;
1751 ret = btrfs_alloc_extent(trans, root, blocksize,
1752 root_objectid, ref_generation,
1753 level, first_objectid, empty_size, hint,
1757 return ERR_PTR(ret);
1759 buf = btrfs_find_create_tree_block(root, ins.objectid, blocksize);
1761 btrfs_free_extent(trans, root, ins.objectid, blocksize,
1762 root->root_key.objectid, ref_generation,
1764 return ERR_PTR(-ENOMEM);
1766 btrfs_set_header_generation(buf, trans->transid);
1767 clean_tree_block(trans, root, buf);
1768 wait_on_tree_block_writeback(root, buf);
1769 btrfs_set_buffer_uptodate(buf);
1771 if (PageDirty(buf->first_page)) {
1772 printk("page %lu dirty\n", buf->first_page->index);
1776 set_extent_dirty(&trans->transaction->dirty_pages, buf->start,
1777 buf->start + buf->len - 1, GFP_NOFS);
1778 set_extent_bits(&BTRFS_I(root->fs_info->btree_inode)->io_tree,
1779 buf->start, buf->start + buf->len - 1,
1780 EXTENT_CSUM, GFP_NOFS);
1781 buf->flags |= EXTENT_CSUM;
1782 if (!btrfs_test_opt(root, SSD))
1783 btrfs_set_buffer_defrag(buf);
1784 trans->blocks_used++;
1788 static int noinline drop_leaf_ref(struct btrfs_trans_handle *trans,
1789 struct btrfs_root *root,
1790 struct extent_buffer *leaf)
1793 u64 leaf_generation;
1794 struct btrfs_key key;
1795 struct btrfs_file_extent_item *fi;
1800 BUG_ON(!btrfs_is_leaf(leaf));
1801 nritems = btrfs_header_nritems(leaf);
1802 leaf_owner = btrfs_header_owner(leaf);
1803 leaf_generation = btrfs_header_generation(leaf);
1805 for (i = 0; i < nritems; i++) {
1808 btrfs_item_key_to_cpu(leaf, &key, i);
1809 if (btrfs_key_type(&key) != BTRFS_EXTENT_DATA_KEY)
1811 fi = btrfs_item_ptr(leaf, i, struct btrfs_file_extent_item);
1812 if (btrfs_file_extent_type(leaf, fi) ==
1813 BTRFS_FILE_EXTENT_INLINE)
1816 * FIXME make sure to insert a trans record that
1817 * repeats the snapshot del on crash
1819 disk_bytenr = btrfs_file_extent_disk_bytenr(leaf, fi);
1820 if (disk_bytenr == 0)
1822 ret = btrfs_free_extent(trans, root, disk_bytenr,
1823 btrfs_file_extent_disk_num_bytes(leaf, fi),
1824 leaf_owner, leaf_generation,
1825 key.objectid, key.offset, 0);
1831 static void noinline reada_walk_down(struct btrfs_root *root,
1832 struct extent_buffer *node,
1845 nritems = btrfs_header_nritems(node);
1846 level = btrfs_header_level(node);
1850 for (i = slot; i < nritems && skipped < 32; i++) {
1851 bytenr = btrfs_node_blockptr(node, i);
1852 if (last && ((bytenr > last && bytenr - last > 32 * 1024) ||
1853 (last > bytenr && last - bytenr > 32 * 1024))) {
1857 blocksize = btrfs_level_size(root, level - 1);
1859 ret = lookup_extent_ref(NULL, root, bytenr,
1867 mutex_unlock(&root->fs_info->fs_mutex);
1868 ret = readahead_tree_block(root, bytenr, blocksize);
1869 last = bytenr + blocksize;
1871 mutex_lock(&root->fs_info->fs_mutex);
1878 * helper function for drop_snapshot, this walks down the tree dropping ref
1879 * counts as it goes.
1881 static int noinline walk_down_tree(struct btrfs_trans_handle *trans,
1882 struct btrfs_root *root,
1883 struct btrfs_path *path, int *level)
1888 struct extent_buffer *next;
1889 struct extent_buffer *cur;
1890 struct extent_buffer *parent;
1895 WARN_ON(*level < 0);
1896 WARN_ON(*level >= BTRFS_MAX_LEVEL);
1897 ret = lookup_extent_ref(trans, root,
1898 path->nodes[*level]->start,
1899 path->nodes[*level]->len, &refs);
1905 * walk down to the last node level and free all the leaves
1907 while(*level >= 0) {
1908 WARN_ON(*level < 0);
1909 WARN_ON(*level >= BTRFS_MAX_LEVEL);
1910 cur = path->nodes[*level];
1912 if (btrfs_header_level(cur) != *level)
1915 if (path->slots[*level] >=
1916 btrfs_header_nritems(cur))
1919 ret = drop_leaf_ref(trans, root, cur);
1923 bytenr = btrfs_node_blockptr(cur, path->slots[*level]);
1924 blocksize = btrfs_level_size(root, *level - 1);
1925 ret = lookup_extent_ref(trans, root, bytenr, blocksize, &refs);
1928 parent = path->nodes[*level];
1929 root_owner = btrfs_header_owner(parent);
1930 root_gen = btrfs_header_generation(parent);
1931 path->slots[*level]++;
1932 ret = btrfs_free_extent(trans, root, bytenr,
1933 blocksize, root_owner,
1938 next = btrfs_find_tree_block(root, bytenr, blocksize);
1939 if (!next || !btrfs_buffer_uptodate(next)) {
1940 free_extent_buffer(next);
1941 reada_walk_down(root, cur, path->slots[*level]);
1942 mutex_unlock(&root->fs_info->fs_mutex);
1943 next = read_tree_block(root, bytenr, blocksize);
1944 mutex_lock(&root->fs_info->fs_mutex);
1946 /* we dropped the lock, check one more time */
1947 ret = lookup_extent_ref(trans, root, bytenr,
1951 parent = path->nodes[*level];
1952 root_owner = btrfs_header_owner(parent);
1953 root_gen = btrfs_header_generation(parent);
1955 path->slots[*level]++;
1956 free_extent_buffer(next);
1957 ret = btrfs_free_extent(trans, root, bytenr,
1965 WARN_ON(*level <= 0);
1966 if (path->nodes[*level-1])
1967 free_extent_buffer(path->nodes[*level-1]);
1968 path->nodes[*level-1] = next;
1969 *level = btrfs_header_level(next);
1970 path->slots[*level] = 0;
1973 WARN_ON(*level < 0);
1974 WARN_ON(*level >= BTRFS_MAX_LEVEL);
1976 if (path->nodes[*level] == root->node) {
1977 root_owner = root->root_key.objectid;
1978 parent = path->nodes[*level];
1980 parent = path->nodes[*level + 1];
1981 root_owner = btrfs_header_owner(parent);
1984 root_gen = btrfs_header_generation(parent);
1985 ret = btrfs_free_extent(trans, root, path->nodes[*level]->start,
1986 path->nodes[*level]->len,
1987 root_owner, root_gen, 0, 0, 1);
1988 free_extent_buffer(path->nodes[*level]);
1989 path->nodes[*level] = NULL;
1996 * helper for dropping snapshots. This walks back up the tree in the path
1997 * to find the first node higher up where we haven't yet gone through
2000 static int noinline walk_up_tree(struct btrfs_trans_handle *trans,
2001 struct btrfs_root *root,
2002 struct btrfs_path *path, int *level)
2006 struct btrfs_root_item *root_item = &root->root_item;
2011 for(i = *level; i < BTRFS_MAX_LEVEL - 1 && path->nodes[i]; i++) {
2012 slot = path->slots[i];
2013 if (slot < btrfs_header_nritems(path->nodes[i]) - 1) {
2014 struct extent_buffer *node;
2015 struct btrfs_disk_key disk_key;
2016 node = path->nodes[i];
2019 WARN_ON(*level == 0);
2020 btrfs_node_key(node, &disk_key, path->slots[i]);
2021 memcpy(&root_item->drop_progress,
2022 &disk_key, sizeof(disk_key));
2023 root_item->drop_level = i;
2026 if (path->nodes[*level] == root->node) {
2027 root_owner = root->root_key.objectid;
2029 btrfs_header_generation(path->nodes[*level]);
2031 struct extent_buffer *node;
2032 node = path->nodes[*level + 1];
2033 root_owner = btrfs_header_owner(node);
2034 root_gen = btrfs_header_generation(node);
2036 ret = btrfs_free_extent(trans, root,
2037 path->nodes[*level]->start,
2038 path->nodes[*level]->len,
2039 root_owner, root_gen, 0, 0, 1);
2041 free_extent_buffer(path->nodes[*level]);
2042 path->nodes[*level] = NULL;
2050 * drop the reference count on the tree rooted at 'snap'. This traverses
2051 * the tree freeing any blocks that have a ref count of zero after being
2054 int btrfs_drop_snapshot(struct btrfs_trans_handle *trans, struct btrfs_root
2060 struct btrfs_path *path;
2063 struct btrfs_root_item *root_item = &root->root_item;
2065 path = btrfs_alloc_path();
2068 level = btrfs_header_level(root->node);
2070 if (btrfs_disk_key_objectid(&root_item->drop_progress) == 0) {
2071 path->nodes[level] = root->node;
2072 extent_buffer_get(root->node);
2073 path->slots[level] = 0;
2075 struct btrfs_key key;
2076 struct btrfs_disk_key found_key;
2077 struct extent_buffer *node;
2079 btrfs_disk_key_to_cpu(&key, &root_item->drop_progress);
2080 level = root_item->drop_level;
2081 path->lowest_level = level;
2082 wret = btrfs_search_slot(NULL, root, &key, path, 0, 0);
2087 node = path->nodes[level];
2088 btrfs_node_key(node, &found_key, path->slots[level]);
2089 WARN_ON(memcmp(&found_key, &root_item->drop_progress,
2090 sizeof(found_key)));
2093 wret = walk_down_tree(trans, root, path, &level);
2099 wret = walk_up_tree(trans, root, path, &level);
2107 for (i = 0; i <= orig_level; i++) {
2108 if (path->nodes[i]) {
2109 free_extent_buffer(path->nodes[i]);
2110 path->nodes[i] = NULL;
2114 btrfs_free_path(path);
2118 int btrfs_free_block_groups(struct btrfs_fs_info *info)
2125 ret = find_first_extent_bit(&info->block_group_cache, 0,
2126 &start, &end, (unsigned int)-1);
2129 ret = get_state_private(&info->block_group_cache, start, &ptr);
2131 kfree((void *)(unsigned long)ptr);
2132 clear_extent_bits(&info->block_group_cache, start,
2133 end, (unsigned int)-1, GFP_NOFS);
2136 ret = find_first_extent_bit(&info->free_space_cache, 0,
2137 &start, &end, EXTENT_DIRTY);
2140 clear_extent_dirty(&info->free_space_cache, start,
2146 static int noinline relocate_inode_pages(struct inode *inode, u64 start,
2152 u64 existing_delalloc;
2153 unsigned long last_index;
2156 struct extent_io_tree *io_tree = &BTRFS_I(inode)->io_tree;
2157 struct file_ra_state *ra;
2159 ra = kzalloc(sizeof(*ra), GFP_NOFS);
2161 mutex_lock(&inode->i_mutex);
2162 i = start >> PAGE_CACHE_SHIFT;
2163 last_index = (start + len - 1) >> PAGE_CACHE_SHIFT;
2165 file_ra_state_init(ra, inode->i_mapping);
2166 btrfs_force_ra(inode->i_mapping, ra, NULL, i, last_index);
2169 for (; i <= last_index; i++) {
2170 page = grab_cache_page(inode->i_mapping, i);
2173 if (!PageUptodate(page)) {
2174 btrfs_readpage(NULL, page);
2176 if (!PageUptodate(page)) {
2178 page_cache_release(page);
2182 page_start = (u64)page->index << PAGE_CACHE_SHIFT;
2183 page_end = page_start + PAGE_CACHE_SIZE - 1;
2185 lock_extent(io_tree, page_start, page_end, GFP_NOFS);
2187 delalloc_start = page_start;
2188 existing_delalloc = count_range_bits(io_tree,
2189 &delalloc_start, page_end,
2190 PAGE_CACHE_SIZE, EXTENT_DELALLOC);
2192 set_extent_delalloc(io_tree, page_start,
2193 page_end, GFP_NOFS);
2195 unlock_extent(io_tree, page_start, page_end, GFP_NOFS);
2196 set_page_dirty(page);
2198 page_cache_release(page);
2202 mutex_unlock(&inode->i_mutex);
2207 * note, this releases the path
2209 static int noinline relocate_one_reference(struct btrfs_root *extent_root,
2210 struct btrfs_path *path,
2211 struct btrfs_key *extent_key)
2213 struct inode *inode;
2214 struct btrfs_root *found_root;
2215 struct btrfs_key *root_location;
2216 struct btrfs_extent_ref *ref;
2223 ref = btrfs_item_ptr(path->nodes[0], path->slots[0],
2224 struct btrfs_extent_ref);
2225 ref_root = btrfs_ref_root(path->nodes[0], ref);
2226 ref_gen = btrfs_ref_generation(path->nodes[0], ref);
2227 ref_objectid = btrfs_ref_objectid(path->nodes[0], ref);
2228 ref_offset = btrfs_ref_offset(path->nodes[0], ref);
2229 btrfs_release_path(extent_root, path);
2231 root_location = kmalloc(sizeof(*root_location), GFP_NOFS);
2232 root_location->objectid = ref_root;
2234 root_location->offset = 0;
2236 root_location->offset = (u64)-1;
2237 root_location->type = BTRFS_ROOT_ITEM_KEY;
2239 found_root = btrfs_read_fs_root_no_name(extent_root->fs_info,
2241 BUG_ON(!found_root);
2242 kfree(root_location);
2244 if (ref_objectid >= BTRFS_FIRST_FREE_OBJECTID) {
2245 mutex_unlock(&extent_root->fs_info->fs_mutex);
2246 inode = btrfs_iget_locked(extent_root->fs_info->sb,
2247 ref_objectid, found_root);
2248 if (inode->i_state & I_NEW) {
2249 /* the inode and parent dir are two different roots */
2250 BTRFS_I(inode)->root = found_root;
2251 BTRFS_I(inode)->location.objectid = ref_objectid;
2252 BTRFS_I(inode)->location.type = BTRFS_INODE_ITEM_KEY;
2253 BTRFS_I(inode)->location.offset = 0;
2254 btrfs_read_locked_inode(inode);
2255 unlock_new_inode(inode);
2258 /* this can happen if the reference is not against
2259 * the latest version of the tree root
2261 if (is_bad_inode(inode)) {
2262 mutex_lock(&extent_root->fs_info->fs_mutex);
2265 relocate_inode_pages(inode, ref_offset, extent_key->offset);
2266 /* FIXME, data=ordered will help get rid of this */
2267 filemap_fdatawrite(inode->i_mapping);
2269 mutex_lock(&extent_root->fs_info->fs_mutex);
2271 struct btrfs_trans_handle *trans;
2272 struct btrfs_key found_key;
2273 struct extent_buffer *eb;
2277 trans = btrfs_start_transaction(found_root, 1);
2278 eb = read_tree_block(found_root, extent_key->objectid,
2279 extent_key->offset);
2280 level = btrfs_header_level(eb);
2283 btrfs_item_key_to_cpu(eb, &found_key, 0);
2285 btrfs_node_key_to_cpu(eb, &found_key, 0);
2287 free_extent_buffer(eb);
2289 path->lowest_level = level;
2291 ret = btrfs_search_slot(trans, found_root, &found_key, path,
2293 path->lowest_level = 0;
2294 for (i = level; i < BTRFS_MAX_LEVEL; i++) {
2295 if (!path->nodes[i])
2297 free_extent_buffer(path->nodes[i]);
2298 path->nodes[i] = NULL;
2300 btrfs_release_path(found_root, path);
2301 btrfs_end_transaction(trans, found_root);
2308 static int noinline relocate_one_extent(struct btrfs_root *extent_root,
2309 struct btrfs_path *path,
2310 struct btrfs_key *extent_key)
2312 struct btrfs_key key;
2313 struct btrfs_key found_key;
2314 struct extent_buffer *leaf;
2319 key.objectid = extent_key->objectid;
2320 key.type = BTRFS_EXTENT_REF_KEY;
2324 ret = btrfs_search_slot(NULL, extent_root, &key, path, 0, 0);
2330 leaf = path->nodes[0];
2331 nritems = btrfs_header_nritems(leaf);
2332 if (path->slots[0] == nritems)
2335 btrfs_item_key_to_cpu(leaf, &found_key, path->slots[0]);
2336 if (found_key.objectid != extent_key->objectid)
2339 if (found_key.type != BTRFS_EXTENT_REF_KEY)
2342 key.offset = found_key.offset + 1;
2343 item_size = btrfs_item_size_nr(leaf, path->slots[0]);
2345 ret = relocate_one_reference(extent_root, path, extent_key);
2351 btrfs_release_path(extent_root, path);
2355 int btrfs_shrink_extent_tree(struct btrfs_root *root, u64 new_size)
2357 struct btrfs_trans_handle *trans;
2358 struct btrfs_root *tree_root = root->fs_info->tree_root;
2359 struct btrfs_path *path;
2362 struct btrfs_fs_info *info = root->fs_info;
2363 struct extent_io_tree *block_group_cache;
2364 struct btrfs_key key;
2365 struct btrfs_key found_key;
2366 struct extent_buffer *leaf;
2371 btrfs_set_super_total_bytes(&info->super_copy, new_size);
2372 clear_extent_dirty(&info->free_space_cache, new_size, (u64)-1,
2374 block_group_cache = &info->block_group_cache;
2375 path = btrfs_alloc_path();
2376 root = root->fs_info->extent_root;
2381 key.objectid = new_size;
2384 cur_byte = key.objectid;
2386 ret = btrfs_search_slot(NULL, root, &key, path, 0, 0);
2390 ret = btrfs_previous_item(root, path, 0, BTRFS_EXTENT_ITEM_KEY);
2394 leaf = path->nodes[0];
2395 btrfs_item_key_to_cpu(leaf, &found_key, path->slots[0]);
2396 if (found_key.objectid + found_key.offset > new_size) {
2397 cur_byte = found_key.objectid;
2398 key.objectid = cur_byte;
2401 btrfs_release_path(root, path);
2404 ret = btrfs_search_slot(NULL, root, &key, path, 0, 0);
2408 leaf = path->nodes[0];
2409 nritems = btrfs_header_nritems(leaf);
2411 if (path->slots[0] >= nritems) {
2412 ret = btrfs_next_leaf(root, path);
2419 leaf = path->nodes[0];
2420 nritems = btrfs_header_nritems(leaf);
2423 btrfs_item_key_to_cpu(leaf, &found_key, path->slots[0]);
2425 if (progress && need_resched()) {
2426 memcpy(&key, &found_key, sizeof(key));
2427 mutex_unlock(&root->fs_info->fs_mutex);
2429 mutex_lock(&root->fs_info->fs_mutex);
2430 btrfs_release_path(root, path);
2431 btrfs_search_slot(NULL, root, &key, path, 0, 0);
2437 if (btrfs_key_type(&found_key) != BTRFS_EXTENT_ITEM_KEY ||
2438 found_key.objectid + found_key.offset <= cur_byte) {
2444 cur_byte = found_key.objectid + found_key.offset;
2445 key.objectid = cur_byte;
2446 btrfs_release_path(root, path);
2447 ret = relocate_one_extent(root, path, &found_key);
2450 btrfs_release_path(root, path);
2452 if (total_found > 0) {
2453 trans = btrfs_start_transaction(tree_root, 1);
2454 btrfs_commit_transaction(trans, tree_root);
2456 mutex_unlock(&root->fs_info->fs_mutex);
2457 btrfs_clean_old_snapshots(tree_root);
2458 mutex_lock(&root->fs_info->fs_mutex);
2460 trans = btrfs_start_transaction(tree_root, 1);
2461 btrfs_commit_transaction(trans, tree_root);
2465 trans = btrfs_start_transaction(root, 1);
2466 key.objectid = new_size;
2472 ret = btrfs_search_slot(trans, root, &key, path, -1, 1);
2476 leaf = path->nodes[0];
2477 nritems = btrfs_header_nritems(leaf);
2479 if (path->slots[0] >= nritems) {
2480 ret = btrfs_next_leaf(root, path);
2487 leaf = path->nodes[0];
2488 btrfs_item_key_to_cpu(leaf, &found_key, path->slots[0]);
2491 * btrfs_next_leaf doesn't cow buffers, we have to
2492 * do the search again
2494 memcpy(&key, &found_key, sizeof(key));
2495 btrfs_release_path(root, path);
2499 btrfs_item_key_to_cpu(leaf, &found_key, path->slots[0]);
2500 if (btrfs_key_type(&found_key) != BTRFS_BLOCK_GROUP_ITEM_KEY) {
2501 printk("shrinker found key %Lu %u %Lu\n",
2502 found_key.objectid, found_key.type,
2507 ret = get_state_private(&info->block_group_cache,
2508 found_key.objectid, &ptr);
2510 kfree((void *)(unsigned long)ptr);
2512 clear_extent_bits(&info->block_group_cache, found_key.objectid,
2513 found_key.objectid + found_key.offset - 1,
2514 (unsigned int)-1, GFP_NOFS);
2516 key.objectid = found_key.objectid + 1;
2517 btrfs_del_item(trans, root, path);
2518 btrfs_release_path(root, path);
2520 if (need_resched()) {
2521 mutex_unlock(&root->fs_info->fs_mutex);
2523 mutex_lock(&root->fs_info->fs_mutex);
2526 clear_extent_dirty(&info->free_space_cache, new_size, (u64)-1,
2528 btrfs_commit_transaction(trans, root);
2530 btrfs_free_path(path);
2534 int btrfs_grow_extent_tree(struct btrfs_trans_handle *trans,
2535 struct btrfs_root *root, u64 new_size)
2537 btrfs_set_super_total_bytes(&root->fs_info->super_copy, new_size);
2541 int find_first_block_group(struct btrfs_root *root, struct btrfs_path *path,
2542 struct btrfs_key *key)
2545 struct btrfs_key found_key;
2546 struct extent_buffer *leaf;
2549 ret = btrfs_search_slot(NULL, root, key, path, 0, 0);
2553 slot = path->slots[0];
2554 leaf = path->nodes[0];
2555 if (slot >= btrfs_header_nritems(leaf)) {
2556 ret = btrfs_next_leaf(root, path);
2563 btrfs_item_key_to_cpu(leaf, &found_key, slot);
2565 if (found_key.objectid >= key->objectid &&
2566 found_key.type == BTRFS_BLOCK_GROUP_ITEM_KEY)
2575 static int update_space_info(struct btrfs_fs_info *info, u64 flags,
2576 u64 total_bytes, u64 bytes_used,
2577 struct btrfs_space_info **space_info)
2579 struct btrfs_space_info *found;
2581 found = __find_space_info(info, flags);
2583 found->total_bytes += total_bytes;
2584 found->bytes_used += bytes_used;
2585 WARN_ON(found->total_bytes < found->bytes_used);
2586 *space_info = found;
2589 found = kmalloc(sizeof(*found), GFP_NOFS);
2593 list_add(&found->list, &info->space_info);
2594 found->flags = flags;
2595 found->total_bytes = total_bytes;
2596 found->bytes_used = bytes_used;
2597 found->bytes_pinned = 0;
2599 *space_info = found;
2603 int btrfs_read_block_groups(struct btrfs_root *root)
2605 struct btrfs_path *path;
2608 struct btrfs_block_group_cache *cache;
2609 struct btrfs_fs_info *info = root->fs_info;
2610 struct btrfs_space_info *space_info;
2611 struct extent_io_tree *block_group_cache;
2612 struct btrfs_key key;
2613 struct btrfs_key found_key;
2614 struct extent_buffer *leaf;
2616 block_group_cache = &info->block_group_cache;
2617 root = info->extent_root;
2620 btrfs_set_key_type(&key, BTRFS_BLOCK_GROUP_ITEM_KEY);
2621 path = btrfs_alloc_path();
2626 ret = find_first_block_group(root, path, &key);
2634 leaf = path->nodes[0];
2635 btrfs_item_key_to_cpu(leaf, &found_key, path->slots[0]);
2636 cache = kmalloc(sizeof(*cache), GFP_NOFS);
2642 read_extent_buffer(leaf, &cache->item,
2643 btrfs_item_ptr_offset(leaf, path->slots[0]),
2644 sizeof(cache->item));
2645 memcpy(&cache->key, &found_key, sizeof(found_key));
2649 key.objectid = found_key.objectid + found_key.offset;
2650 btrfs_release_path(root, path);
2651 cache->flags = btrfs_block_group_flags(&cache->item);
2653 if (cache->flags & BTRFS_BLOCK_GROUP_DATA) {
2654 bit = BLOCK_GROUP_DATA;
2655 } else if (cache->flags & BTRFS_BLOCK_GROUP_SYSTEM) {
2656 bit = BLOCK_GROUP_SYSTEM;
2657 } else if (cache->flags & BTRFS_BLOCK_GROUP_METADATA) {
2658 bit = BLOCK_GROUP_METADATA;
2661 ret = update_space_info(info, cache->flags, found_key.offset,
2662 btrfs_block_group_used(&cache->item),
2665 cache->space_info = space_info;
2667 /* use EXTENT_LOCKED to prevent merging */
2668 set_extent_bits(block_group_cache, found_key.objectid,
2669 found_key.objectid + found_key.offset - 1,
2670 bit | EXTENT_LOCKED, GFP_NOFS);
2671 set_state_private(block_group_cache, found_key.objectid,
2672 (unsigned long)cache);
2675 btrfs_super_total_bytes(&info->super_copy))
2680 btrfs_free_path(path);
2684 int btrfs_make_block_group(struct btrfs_trans_handle *trans,
2685 struct btrfs_root *root, u64 bytes_used,
2686 u64 type, u64 chunk_tree, u64 chunk_objectid,
2691 struct btrfs_root *extent_root;
2692 struct btrfs_block_group_cache *cache;
2693 struct extent_io_tree *block_group_cache;
2695 extent_root = root->fs_info->extent_root;
2696 block_group_cache = &root->fs_info->block_group_cache;
2698 cache = kmalloc(sizeof(*cache), GFP_NOFS);
2700 cache->key.objectid = chunk_objectid;
2701 cache->key.offset = size;
2704 btrfs_set_key_type(&cache->key, BTRFS_BLOCK_GROUP_ITEM_KEY);
2705 memset(&cache->item, 0, sizeof(cache->item));
2706 btrfs_set_block_group_used(&cache->item, bytes_used);
2707 btrfs_set_block_group_chunk_tree(&cache->item, chunk_tree);
2708 btrfs_set_block_group_chunk_objectid(&cache->item, chunk_objectid);
2709 cache->flags = type;
2710 btrfs_set_block_group_flags(&cache->item, type);
2712 ret = update_space_info(root->fs_info, cache->flags, size, bytes_used,
2713 &cache->space_info);
2716 if (type & BTRFS_BLOCK_GROUP_DATA) {
2717 bit = BLOCK_GROUP_DATA;
2718 } else if (type & BTRFS_BLOCK_GROUP_SYSTEM) {
2719 bit = BLOCK_GROUP_SYSTEM;
2720 } else if (type & BTRFS_BLOCK_GROUP_METADATA) {
2721 bit = BLOCK_GROUP_METADATA;
2723 set_extent_bits(block_group_cache, chunk_objectid,
2724 chunk_objectid + size - 1,
2725 bit | EXTENT_LOCKED, GFP_NOFS);
2726 set_state_private(block_group_cache, chunk_objectid,
2727 (unsigned long)cache);
2729 ret = btrfs_insert_item(trans, extent_root, &cache->key, &cache->item,
2730 sizeof(cache->item));
2733 finish_current_insert(trans, extent_root);
2734 ret = del_pending_extents(trans, extent_root);