#include "print-tree.h"
#include "transaction.h"
#include "crc32c.h"
+#include "volumes.h"
-#define BLOCK_GROUP_DATA EXTENT_WRITEBACK
+#define BLOCK_GROUP_DATA EXTENT_WRITEBACK
#define BLOCK_GROUP_METADATA EXTENT_UPTODATE
+#define BLOCK_GROUP_SYSTEM EXTENT_NEW
+
#define BLOCK_GROUP_DIRTY EXTENT_DIRTY
+#define PENDING_EXTENT_INSERT 0
+#define PENDING_EXTENT_DELETE 1
+#define PENDING_BACKREF_UPDATE 2
+
+struct pending_extent_op {
+ int type;
+ u64 bytenr;
+ u64 num_bytes;
+ u64 flags;
+ struct btrfs_disk_key key;
+ int level;
+};
+
+static int alloc_reserved_tree_block(struct btrfs_trans_handle *trans,
+ struct btrfs_root *root,
+ u64 root_objectid, u64 generation,
+ u64 flags, struct btrfs_disk_key *key,
+ int level, struct btrfs_key *ins);
+static int __free_extent(struct btrfs_trans_handle *trans,
+ struct btrfs_root *root,
+ u64 bytenr, u64 num_bytes, u64 parent,
+ u64 root_objectid, u64 owner_objectid,
+ u64 owner_offset, int refs_to_drop);
static int finish_current_insert(struct btrfs_trans_handle *trans, struct
btrfs_root *extent_root);
static int del_pending_extents(struct btrfs_trans_handle *trans, struct
btrfs_root *extent_root);
+static int remove_sb_from_cache(struct btrfs_root *root,
+ struct btrfs_block_group_cache *cache)
+{
+ u64 bytenr;
+ u64 *logical;
+ int stripe_len;
+ int i, nr, ret;
+ struct extent_io_tree *free_space_cache;
+
+ free_space_cache = &root->fs_info->free_space_cache;
+ for (i = 0; i < BTRFS_SUPER_MIRROR_MAX; i++) {
+ bytenr = btrfs_sb_offset(i);
+ ret = btrfs_rmap_block(&root->fs_info->mapping_tree,
+ cache->key.objectid, bytenr, 0,
+ &logical, &nr, &stripe_len);
+ BUG_ON(ret);
+ while (nr--) {
+ clear_extent_dirty(free_space_cache, logical[nr],
+ logical[nr] + stripe_len - 1, GFP_NOFS);
+ }
+ kfree(logical);
+ }
+ return 0;
+}
+
static int cache_block_group(struct btrfs_root *root,
struct btrfs_block_group_cache *block_group)
{
int ret;
struct btrfs_key key;
struct extent_buffer *leaf;
- struct extent_map_tree *free_space_cache;
+ struct extent_io_tree *free_space_cache;
int slot;
- u64 last = 0;
+ u64 last;
u64 hole_size;
- u64 first_free;
- int found = 0;
if (!block_group)
return 0;
return -ENOMEM;
path->reada = 2;
- first_free = block_group->key.objectid;
- key.objectid = block_group->key.objectid;
+ last = max_t(u64, block_group->key.objectid, BTRFS_SUPER_INFO_OFFSET);
+ key.objectid = last;
key.offset = 0;
-
btrfs_set_key_type(&key, BTRFS_EXTENT_ITEM_KEY);
ret = btrfs_search_slot(NULL, root, &key, path, 0, 0);
-
if (ret < 0)
- return ret;
-
- if (ret && path->slots[0] > 0)
- path->slots[0]--;
+ goto err;
while(1) {
leaf = path->nodes[0];
break;
}
}
-
btrfs_item_key_to_cpu(leaf, &key, slot);
if (key.objectid < block_group->key.objectid) {
- if (btrfs_key_type(&key) != BTRFS_EXTENT_REF_KEY &&
- key.objectid + key.offset > first_free)
- first_free = key.objectid + key.offset;
goto next;
}
-
if (key.objectid >= block_group->key.objectid +
block_group->key.offset) {
break;
}
if (btrfs_key_type(&key) == BTRFS_EXTENT_ITEM_KEY) {
- if (!found) {
- last = first_free;
- found = 1;
- }
if (key.objectid > last) {
hole_size = key.objectid - last;
set_extent_dirty(free_space_cache, last,
path->slots[0]++;
}
- if (!found)
- last = first_free;
if (block_group->key.objectid +
block_group->key.offset > last) {
hole_size = block_group->key.objectid +
set_extent_dirty(free_space_cache, last,
last + hole_size - 1, GFP_NOFS);
}
+ remove_sb_from_cache(root, block_group);
block_group->cached = 1;
err:
btrfs_free_path(path);
return 0;
}
+struct btrfs_block_group_cache *btrfs_lookup_first_block_group(struct
+ btrfs_fs_info *info,
+ u64 bytenr)
+{
+ struct extent_io_tree *block_group_cache;
+ struct btrfs_block_group_cache *block_group = NULL;
+ u64 ptr;
+ u64 start;
+ u64 end;
+ int ret;
+
+ bytenr = max_t(u64, bytenr,
+ BTRFS_SUPER_INFO_OFFSET + BTRFS_SUPER_INFO_SIZE);
+ block_group_cache = &info->block_group_cache;
+ ret = find_first_extent_bit(block_group_cache,
+ bytenr, &start, &end,
+ BLOCK_GROUP_DATA | BLOCK_GROUP_METADATA |
+ BLOCK_GROUP_SYSTEM);
+ if (ret) {
+ return NULL;
+ }
+ ret = get_state_private(block_group_cache, start, &ptr);
+ if (ret)
+ return NULL;
+
+ block_group = (struct btrfs_block_group_cache *)(unsigned long)ptr;
+ return block_group;
+}
+
struct btrfs_block_group_cache *btrfs_lookup_block_group(struct
btrfs_fs_info *info,
u64 bytenr)
{
- struct extent_map_tree *block_group_cache;
+ struct extent_io_tree *block_group_cache;
struct btrfs_block_group_cache *block_group = NULL;
u64 ptr;
u64 start;
block_group_cache = &info->block_group_cache;
ret = find_first_extent_bit(block_group_cache,
bytenr, &start, &end,
- BLOCK_GROUP_DATA | BLOCK_GROUP_METADATA);
+ BLOCK_GROUP_DATA | BLOCK_GROUP_METADATA |
+ BLOCK_GROUP_SYSTEM);
if (ret) {
return NULL;
}
return NULL;
}
-static u64 noinline find_search_start(struct btrfs_root *root,
+static int block_group_bits(struct btrfs_block_group_cache *cache, u64 bits)
+{
+ return (cache->flags & bits) == bits;
+}
+
+static int noinline find_search_start(struct btrfs_root *root,
struct btrfs_block_group_cache **cache_ret,
- u64 search_start, int num,
- int data, int full_scan)
+ u64 *start_ret, int num, int data)
{
int ret;
struct btrfs_block_group_cache *cache = *cache_ret;
u64 last;
u64 start = 0;
u64 end = 0;
- u64 cache_miss = 0;
+ u64 search_start = *start_ret;
int wrapped = 0;
if (!cache) {
goto out;
last = max(search_start, cache->key.objectid);
+ if (cache->ro || !block_group_bits(cache, data)) {
+ goto new_group;
+ }
while(1) {
ret = find_first_extent_bit(&root->fs_info->free_space_cache,
last, &start, &end, EXTENT_DIRTY);
if (ret) {
- if (!cache_miss)
- cache_miss = last;
goto new_group;
}
start = max(last, start);
last = end + 1;
if (last - start < num) {
- if (last == cache->key.objectid + cache->key.offset)
- cache_miss = start;
continue;
}
- if (data != BTRFS_BLOCK_GROUP_MIXED &&
- start + num > cache->key.objectid + cache->key.offset)
+ if (start + num > cache->key.objectid + cache->key.offset) {
goto new_group;
- return start;
+ }
+ *start_ret = start;
+ return 0;
}
out:
cache = btrfs_lookup_block_group(root->fs_info, search_start);
if (!cache) {
- printk("Unable to find block group for %Lu\n",
- search_start);
+ printk("Unable to find block group for %llu\n",
+ (unsigned long long)search_start);
WARN_ON(1);
- return search_start;
}
- return search_start;
+ return -ENOSPC;
new_group:
last = cache->key.objectid + cache->key.offset;
wrapped:
- cache = btrfs_lookup_block_group(root->fs_info, last);
+ cache = btrfs_lookup_first_block_group(root->fs_info, last);
if (!cache) {
no_cache:
if (!wrapped) {
wrapped = 1;
last = search_start;
- data = BTRFS_BLOCK_GROUP_MIXED;
goto wrapped;
}
goto out;
}
- if (cache_miss && !cache->cached) {
- cache_block_group(root, cache);
- last = cache_miss;
- cache = btrfs_lookup_block_group(root->fs_info, last);
- }
+ cache = btrfs_find_block_group(root, cache, last, data, 0);
cache = btrfs_find_block_group(root, cache, last, data, 0);
if (!cache)
goto no_cache;
+
*cache_ret = cache;
- cache_miss = 0;
goto again;
}
return num;
}
+static int block_group_state_bits(u64 flags)
+{
+ int bits = 0;
+ if (flags & BTRFS_BLOCK_GROUP_DATA)
+ bits |= BLOCK_GROUP_DATA;
+ if (flags & BTRFS_BLOCK_GROUP_METADATA)
+ bits |= BLOCK_GROUP_METADATA;
+ if (flags & BTRFS_BLOCK_GROUP_SYSTEM)
+ bits |= BLOCK_GROUP_SYSTEM;
+ return bits;
+}
+
struct btrfs_block_group_cache *btrfs_find_block_group(struct btrfs_root *root,
struct btrfs_block_group_cache
*hint, u64 search_start,
int data, int owner)
{
struct btrfs_block_group_cache *cache;
- struct extent_map_tree *block_group_cache;
+ struct extent_io_tree *block_group_cache;
struct btrfs_block_group_cache *found_group = NULL;
struct btrfs_fs_info *info = root->fs_info;
u64 used;
int bit;
int ret;
int full_search = 0;
- int factor = 8;
- int data_swap = 0;
+ int factor = 10;
block_group_cache = &info->block_group_cache;
if (!owner)
- factor = 8;
-
- if (data == BTRFS_BLOCK_GROUP_MIXED) {
- bit = BLOCK_GROUP_DATA | BLOCK_GROUP_METADATA;
factor = 10;
- } else if (data)
- bit = BLOCK_GROUP_DATA;
- else
- bit = BLOCK_GROUP_METADATA;
+
+ bit = block_group_state_bits(data);
if (search_start) {
struct btrfs_block_group_cache *shint;
shint = btrfs_lookup_block_group(info, search_start);
- if (shint && (shint->data == data ||
- shint->data == BTRFS_BLOCK_GROUP_MIXED)) {
+ if (shint && !shint->ro && block_group_bits(shint, data)) {
used = btrfs_block_group_used(&shint->item);
if (used + shint->pinned <
div_factor(shint->key.offset, factor)) {
}
}
}
- if (hint && (hint->data == data ||
- hint->data == BTRFS_BLOCK_GROUP_MIXED)) {
+ if (hint && !hint->ro && block_group_bits(hint, data)) {
used = btrfs_block_group_used(&hint->item);
if (used + hint->pinned <
div_factor(hint->key.offset, factor)) {
last = cache->key.objectid + cache->key.offset;
used = btrfs_block_group_used(&cache->item);
- if (full_search)
- free_check = cache->key.offset;
- else
- free_check = div_factor(cache->key.offset, factor);
- if (used + cache->pinned < free_check) {
- found_group = cache;
- goto found;
+ if (!cache->ro && block_group_bits(cache, data)) {
+ if (full_search)
+ free_check = cache->key.offset;
+ else
+ free_check = div_factor(cache->key.offset,
+ factor);
+
+ if (used + cache->pinned < free_check) {
+ found_group = cache;
+ goto found;
+ }
}
cond_resched();
}
full_search = 1;
goto again;
}
- if (!data_swap) {
- data_swap = 1;
- bit = BLOCK_GROUP_DATA | BLOCK_GROUP_METADATA;
- last = search_start;
- goto again;
- }
found:
return found_group;
}
-static u64 hash_extent_ref(u64 root_objectid, u64 ref_generation,
- u64 owner, u64 owner_offset)
-{
- u32 high_crc = ~(u32)0;
- u32 low_crc = ~(u32)0;
- __le64 lenum;
-
- lenum = cpu_to_le64(root_objectid);
- high_crc = crc32c(high_crc, &lenum, sizeof(lenum));
- lenum = cpu_to_le64(ref_generation);
- low_crc = crc32c(low_crc, &lenum, sizeof(lenum));
-
-#if 0
- lenum = cpu_to_le64(owner);
- low_crc = crc32c(low_crc, &lenum, sizeof(lenum));
- lenum = cpu_to_le64(owner_offset);
- low_crc = crc32c(low_crc, &lenum, sizeof(lenum));
-#endif
- return ((u64)high_crc << 32) | (u64)low_crc;
-}
-
-static int match_extent_ref(struct extent_buffer *leaf,
- struct btrfs_extent_ref *disk_ref,
- struct btrfs_extent_ref *cpu_ref)
-{
- int ret;
- int len;
-
- if (cpu_ref->objectid)
- len = sizeof(*cpu_ref);
- else
- len = 2 * sizeof(u64);
- ret = memcmp_extent_buffer(leaf, cpu_ref, (unsigned long)disk_ref,
- len);
- return ret == 0;
-}
-
-static int noinline lookup_extent_backref(struct btrfs_trans_handle *trans,
- struct btrfs_root *root,
- struct btrfs_path *path, u64 bytenr,
- u64 root_objectid,
- u64 ref_generation, u64 owner,
- u64 owner_offset, int del)
-{
- u64 hash;
- struct btrfs_key key;
- struct btrfs_key found_key;
- struct btrfs_extent_ref ref;
- struct extent_buffer *leaf;
- struct btrfs_extent_ref *disk_ref;
- int ret;
- int ret2;
-
- btrfs_set_stack_ref_root(&ref, root_objectid);
- btrfs_set_stack_ref_generation(&ref, ref_generation);
- btrfs_set_stack_ref_objectid(&ref, owner);
- btrfs_set_stack_ref_offset(&ref, owner_offset);
-
- hash = hash_extent_ref(root_objectid, ref_generation, owner,
- owner_offset);
- key.offset = hash;
- key.objectid = bytenr;
- key.type = BTRFS_EXTENT_REF_KEY;
-
- while (1) {
- ret = btrfs_search_slot(trans, root, &key, path,
- del ? -1 : 0, del);
- if (ret < 0)
- goto out;
- leaf = path->nodes[0];
- if (ret != 0) {
- u32 nritems = btrfs_header_nritems(leaf);
- if (path->slots[0] >= nritems) {
- ret2 = btrfs_next_leaf(root, path);
- if (ret2)
- goto out;
- leaf = path->nodes[0];
- }
- btrfs_item_key_to_cpu(leaf, &found_key, path->slots[0]);
- if (found_key.objectid != bytenr ||
- found_key.type != BTRFS_EXTENT_REF_KEY)
- goto out;
- key.offset = found_key.offset;
- if (del) {
- btrfs_release_path(root, path);
- continue;
- }
- }
- disk_ref = btrfs_item_ptr(path->nodes[0],
- path->slots[0],
- struct btrfs_extent_ref);
- if (match_extent_ref(path->nodes[0], disk_ref, &ref)) {
- ret = 0;
- goto out;
- }
- btrfs_item_key_to_cpu(leaf, &found_key, path->slots[0]);
- key.offset = found_key.offset + 1;
- btrfs_release_path(root, path);
- }
-out:
- return ret;
-}
-
/*
* Back reference rules. Back refs have three main goals:
*
* maintenance. This is actually the same as #2, but with a slightly
* different use case.
*
- * File extents can be referenced by:
+ * There are two kinds of back refs. The implicit back refs is optimized
+ * for pointers in non-shared tree blocks. For a given pointer in a block,
+ * back refs of this kind provide information about the block's owner tree
+ * and the pointer's key. These information allow us to find the block by
+ * b-tree searching. The full back refs is for pointers in tree blocks not
+ * referenced by their owner trees. The location of tree block is recorded
+ * in the back refs. Actually the full back refs is generic, and can be
+ * used in all cases the implicit back refs is used. The major shortcoming
+ * of the full back refs is its overhead. Every time a tree block gets
+ * COWed, we have to update back refs entry for all pointers in it.
*
- * - multiple snapshots, subvolumes, or different generations in one subvol
- * - different files inside a single subvolume (in theory, not implemented yet)
- * - different offsets inside a file (bookend extents in file.c)
+ * For a newly allocated tree block, we use implicit back refs for
+ * pointers in it. This means most tree related operations only involve
+ * implicit back refs. For a tree block created in old transaction, the
+ * only way to drop a reference to it is COW it. So we can detect the
+ * event that tree block loses its owner tree's reference and do the
+ * back refs conversion.
*
- * The extent ref structure has fields for:
+ * When a tree block is COW'd through a tree, there are four cases:
*
- * - Objectid of the subvolume root
- * - Generation number of the tree holding the reference
- * - objectid of the file holding the reference
- * - offset in the file corresponding to the key holding the reference
+ * The reference count of the block is one and the tree is the block's
+ * owner tree. Nothing to do in this case.
*
- * When a file extent is allocated the fields are filled in:
- * (root_key.objectid, trans->transid, inode objectid, offset in file)
+ * The reference count of the block is one and the tree is not the
+ * block's owner tree. In this case, full back refs is used for pointers
+ * in the block. Remove these full back refs, add implicit back refs for
+ * every pointers in the new block.
*
- * When a leaf is cow'd new references are added for every file extent found
- * in the leaf. It looks the same as the create case, but trans->transid
- * will be different when the block is cow'd.
+ * The reference count of the block is greater than one and the tree is
+ * the block's owner tree. In this case, implicit back refs is used for
+ * pointers in the block. Add full back refs for every pointers in the
+ * block, increase lower level extents' reference counts. The original
+ * implicit back refs are entailed to the new block.
*
- * (root_key.objectid, trans->transid, inode objectid, offset in file)
+ * The reference count of the block is greater than one and the tree is
+ * not the block's owner tree. Add implicit back refs for every pointer in
+ * the new block, increase lower level extents' reference count.
*
- * When a file extent is removed either during snapshot deletion or file
- * truncation, the corresponding back reference is found
- * by searching for:
+ * Back Reference Key composing:
*
- * (btrfs_header_owner(leaf), btrfs_header_generation(leaf),
- * inode objectid, offset in file)
+ * The key objectid corresponds to the first byte in the extent,
+ * The key type is used to differentiate between types of back refs.
+ * There are different meanings of the key offset for different types
+ * of back refs.
*
- * Btree extents can be referenced by:
+ * File extents can be referenced by:
*
- * - Different subvolumes
- * - Different generations of the same subvolume
+ * - multiple snapshots, subvolumes, or different generations in one subvol
+ * - different files inside a single subvolume
+ * - different offsets inside a file (bookend extents in file.c)
+ *
+ * The extent ref structure for the implicit back refs has fields for:
+ *
+ * - Objectid of the subvolume root
+ * - objectid of the file holding the reference
+ * - original offset in the file
+ * - how many bookend extents
*
- * Storing sufficient information for a full reverse mapping of a btree
- * block would require storing the lowest key of the block in the backref,
- * and it would require updating that lowest key either before write out or
- * every time it changed. Instead, the objectid of the lowest key is stored
- * along with the level of the tree block. This provides a hint
- * about where in the btree the block can be found. Searches through the
- * btree only need to look for a pointer to that block, so they stop one
- * level higher than the level recorded in the backref.
+ * The key offset for the implicit back refs is hash of the first
+ * three fields.
*
- * Some btrees do not do reference counting on their extents. These
- * include the extent tree and the tree of tree roots. Backrefs for these
- * trees always have a generation of zero.
+ * The extent ref structure for the full back refs has field for:
*
- * When a tree block is created, back references are inserted:
+ * - number of pointers in the tree leaf
*
- * (root->root_key.objectid, trans->transid or zero, level, lowest_key_objectid)
+ * The key offset for the implicit back refs is the first byte of
+ * the tree leaf
*
- * When a tree block is cow'd in a reference counted root,
- * new back references are added for all the blocks it points to.
- * These are of the form (trans->transid will have increased since creation):
+ * When a file extent is allocated, The implicit back refs is used.
+ * the fields are filled in:
*
- * (root->root_key.objectid, trans->transid, level, lowest_key_objectid)
+ * (root_key.objectid, inode objectid, offset in file, 1)
*
- * Because the lowest_key_objectid and the level are just hints
- * they are not used when backrefs are deleted. When a backref is deleted:
+ * When a file extent is removed file truncation, we find the
+ * corresponding implicit back refs and check the following fields:
*
- * if backref was for a tree root:
- * root_objectid = root->root_key.objectid
- * else
- * root_objectid = btrfs_header_owner(parent)
+ * (btrfs_header_owner(leaf), inode objectid, offset in file)
+ *
+ * Btree extents can be referenced by:
*
- * (root_objectid, btrfs_header_generation(parent) or zero, 0, 0)
+ * - Different subvolumes
*
- * Back Reference Key hashing:
+ * Both the implicit back refs and the full back refs for tree blocks
+ * only consist of key. The key offset for the implicit back refs is
+ * objectid of block's owner tree. The key offset for the full back refs
+ * is the first byte of parent block.
*
- * Back references have four fields, each 64 bits long. Unfortunately,
- * This is hashed into a single 64 bit number and placed into the key offset.
- * The key objectid corresponds to the first byte in the extent, and the
- * key type is set to BTRFS_EXTENT_REF_KEY
+ * When implicit back refs is used, information about the lowest key and
+ * level of the tree block are required. These information are stored in
+ * tree block info structure.
*/
-int btrfs_insert_extent_backref(struct btrfs_trans_handle *trans,
- struct btrfs_root *root,
- struct btrfs_path *path, u64 bytenr,
- u64 root_objectid, u64 ref_generation,
- u64 owner, u64 owner_offset)
+
+#ifdef BTRFS_COMPAT_EXTENT_TREE_V0
+static int convert_extent_item_v0(struct btrfs_trans_handle *trans,
+ struct btrfs_root *root,
+ struct btrfs_path *path,
+ u64 owner, u32 extra_size)
{
- u64 hash;
+ struct btrfs_extent_item *item;
+ struct btrfs_extent_item_v0 *ei0;
+ struct btrfs_extent_ref_v0 *ref0;
+ struct btrfs_tree_block_info *bi;
+ struct extent_buffer *leaf;
struct btrfs_key key;
- struct btrfs_extent_ref ref;
- struct btrfs_extent_ref *disk_ref;
+ struct btrfs_key found_key;
+ u32 new_size = sizeof(*item);
+ u64 refs;
int ret;
- btrfs_set_stack_ref_root(&ref, root_objectid);
- btrfs_set_stack_ref_generation(&ref, ref_generation);
- btrfs_set_stack_ref_objectid(&ref, owner);
- btrfs_set_stack_ref_offset(&ref, owner_offset);
+ leaf = path->nodes[0];
+ BUG_ON(btrfs_item_size_nr(leaf, path->slots[0]) != sizeof(*ei0));
- hash = hash_extent_ref(root_objectid, ref_generation, owner,
- owner_offset);
- key.offset = hash;
- key.objectid = bytenr;
- key.type = BTRFS_EXTENT_REF_KEY;
+ btrfs_item_key_to_cpu(leaf, &key, path->slots[0]);
+ ei0 = btrfs_item_ptr(leaf, path->slots[0],
+ struct btrfs_extent_item_v0);
+ refs = btrfs_extent_refs_v0(leaf, ei0);
- ret = btrfs_insert_empty_item(trans, root, path, &key, sizeof(ref));
- while (ret == -EEXIST) {
- disk_ref = btrfs_item_ptr(path->nodes[0], path->slots[0],
- struct btrfs_extent_ref);
- if (match_extent_ref(path->nodes[0], disk_ref, &ref))
- goto out;
- key.offset++;
- btrfs_release_path(root, path);
- ret = btrfs_insert_empty_item(trans, root, path, &key,
- sizeof(ref));
+ if (owner == (u64)-1) {
+ while (1) {
+ if (path->slots[0] >= btrfs_header_nritems(leaf)) {
+ ret = btrfs_next_leaf(root, path);
+ if (ret < 0)
+ return ret;
+ BUG_ON(ret > 0);
+ leaf = path->nodes[0];
+ }
+ btrfs_item_key_to_cpu(leaf, &found_key,
+ path->slots[0]);
+ BUG_ON(key.objectid != found_key.objectid);
+ if (found_key.type != BTRFS_EXTENT_REF_V0_KEY) {
+ path->slots[0]++;
+ continue;
+ }
+ ref0 = btrfs_item_ptr(leaf, path->slots[0],
+ struct btrfs_extent_ref_v0);
+ owner = btrfs_ref_objectid_v0(leaf, ref0);
+ break;
+ }
}
- if (ret)
- goto out;
- disk_ref = btrfs_item_ptr(path->nodes[0], path->slots[0],
- struct btrfs_extent_ref);
- write_extent_buffer(path->nodes[0], &ref, (unsigned long)disk_ref,
- sizeof(ref));
- btrfs_mark_buffer_dirty(path->nodes[0]);
-out:
btrfs_release_path(root, path);
- return ret;
-}
-
-int btrfs_inc_extent_ref(struct btrfs_trans_handle *trans,
- struct btrfs_root *root,
- u64 bytenr, u64 num_bytes,
- u64 root_objectid, u64 ref_generation,
- u64 owner, u64 owner_offset)
-{
- struct btrfs_path *path;
- int ret;
- struct btrfs_key key;
- struct extent_buffer *l;
- struct btrfs_extent_item *item;
- u32 refs;
- WARN_ON(num_bytes < root->sectorsize);
- path = btrfs_alloc_path();
- if (!path)
- return -ENOMEM;
+ if (owner < BTRFS_FIRST_FREE_OBJECTID)
+ new_size += sizeof(*bi);
- key.objectid = bytenr;
- btrfs_set_key_type(&key, BTRFS_EXTENT_ITEM_KEY);
- key.offset = num_bytes;
- ret = btrfs_search_slot(trans, root->fs_info->extent_root, &key, path,
- 0, 1);
+ new_size -= sizeof(*ei0);
+ ret = btrfs_search_slot(trans, root, &key, path, new_size, 1);
if (ret < 0)
return ret;
- if (ret != 0) {
- BUG();
- }
- BUG_ON(ret != 0);
- l = path->nodes[0];
- item = btrfs_item_ptr(l, path->slots[0], struct btrfs_extent_item);
- refs = btrfs_extent_refs(l, item);
- btrfs_set_extent_refs(l, item, refs + 1);
- btrfs_mark_buffer_dirty(path->nodes[0]);
-
- btrfs_release_path(root->fs_info->extent_root, path);
+ BUG_ON(ret);
- ret = btrfs_insert_extent_backref(trans, root->fs_info->extent_root,
- path, bytenr, root_objectid,
- ref_generation, owner, owner_offset);
+ ret = btrfs_extend_item(trans, root, path, new_size);
BUG_ON(ret);
- finish_current_insert(trans, root->fs_info->extent_root);
- del_pending_extents(trans, root->fs_info->extent_root);
- btrfs_free_path(path);
+ leaf = path->nodes[0];
+ item = btrfs_item_ptr(leaf, path->slots[0], struct btrfs_extent_item);
+ btrfs_set_extent_refs(leaf, item, refs);
+ /* FIXME: get real generation */
+ btrfs_set_extent_generation(leaf, item, 0);
+ if (owner < BTRFS_FIRST_FREE_OBJECTID) {
+ btrfs_set_extent_flags(leaf, item,
+ BTRFS_EXTENT_FLAG_TREE_BLOCK |
+ BTRFS_BLOCK_FLAG_FULL_BACKREF);
+ bi = (struct btrfs_tree_block_info *)(item + 1);
+ /* FIXME: get first key of the block */
+ memset_extent_buffer(leaf, 0, (unsigned long)bi, sizeof(*bi));
+ btrfs_set_tree_block_level(leaf, bi, (int)owner);
+ } else {
+ btrfs_set_extent_flags(leaf, item, BTRFS_EXTENT_FLAG_DATA);
+ }
+ btrfs_mark_buffer_dirty(leaf);
return 0;
}
+#endif
-int btrfs_extent_post_op(struct btrfs_trans_handle *trans,
- struct btrfs_root *root)
+static u64 hash_extent_data_ref(u64 root_objectid, u64 owner, u64 offset)
{
- finish_current_insert(trans, root->fs_info->extent_root);
- del_pending_extents(trans, root->fs_info->extent_root);
- return 0;
+ u32 high_crc = ~(u32)0;
+ u32 low_crc = ~(u32)0;
+ __le64 lenum;
+
+ lenum = cpu_to_le64(root_objectid);
+ high_crc = btrfs_crc32c(high_crc, &lenum, sizeof(lenum));
+ lenum = cpu_to_le64(owner);
+ low_crc = btrfs_crc32c(low_crc, &lenum, sizeof(lenum));
+ lenum = cpu_to_le64(offset);
+ low_crc = btrfs_crc32c(low_crc, &lenum, sizeof(lenum));
+
+ return ((u64)high_crc << 31) ^ (u64)low_crc;
}
-static int lookup_extent_ref(struct btrfs_trans_handle *trans,
- struct btrfs_root *root, u64 bytenr,
- u64 num_bytes, u32 *refs)
+static u64 hash_extent_data_ref_item(struct extent_buffer *leaf,
+ struct btrfs_extent_data_ref *ref)
{
- struct btrfs_path *path;
- int ret;
- struct btrfs_key key;
- struct extent_buffer *l;
- struct btrfs_extent_item *item;
+ return hash_extent_data_ref(btrfs_extent_data_ref_root(leaf, ref),
+ btrfs_extent_data_ref_objectid(leaf, ref),
+ btrfs_extent_data_ref_offset(leaf, ref));
+}
- WARN_ON(num_bytes < root->sectorsize);
- path = btrfs_alloc_path();
- key.objectid = bytenr;
- key.offset = num_bytes;
- btrfs_set_key_type(&key, BTRFS_EXTENT_ITEM_KEY);
- ret = btrfs_search_slot(trans, root->fs_info->extent_root, &key, path,
- 0, 0);
- if (ret < 0)
- goto out;
- if (ret != 0) {
- btrfs_print_leaf(root, path->nodes[0]);
- printk("failed to find block number %Lu\n", bytenr);
- BUG();
- }
- l = path->nodes[0];
- item = btrfs_item_ptr(l, path->slots[0], struct btrfs_extent_item);
- *refs = btrfs_extent_refs(l, item);
-out:
- btrfs_free_path(path);
- return 0;
+static int match_extent_data_ref(struct extent_buffer *leaf,
+ struct btrfs_extent_data_ref *ref,
+ u64 root_objectid, u64 owner, u64 offset)
+{
+ if (btrfs_extent_data_ref_root(leaf, ref) != root_objectid ||
+ btrfs_extent_data_ref_objectid(leaf, ref) != owner ||
+ btrfs_extent_data_ref_offset(leaf, ref) != offset)
+ return 0;
+ return 1;
}
-u32 btrfs_count_snapshots_in_path(struct btrfs_root *root,
- struct btrfs_path *count_path,
- u64 first_extent)
+static noinline int lookup_extent_data_ref(struct btrfs_trans_handle *trans,
+ struct btrfs_root *root,
+ struct btrfs_path *path,
+ u64 bytenr, u64 parent,
+ u64 root_objectid,
+ u64 owner, u64 offset)
{
- struct btrfs_root *extent_root = root->fs_info->extent_root;
- struct btrfs_path *path;
- u64 bytenr;
- u64 found_objectid;
- u64 root_objectid = root->root_key.objectid;
- u32 total_count = 0;
- u32 cur_count;
- u32 refs;
+ struct btrfs_key key;
+ struct btrfs_extent_data_ref *ref;
+ struct extent_buffer *leaf;
u32 nritems;
int ret;
- struct btrfs_key key;
- struct btrfs_key found_key;
- struct extent_buffer *l;
- struct btrfs_extent_item *item;
- struct btrfs_extent_ref *ref_item;
- int level = -1;
-
- path = btrfs_alloc_path();
-again:
- if (level == -1)
- bytenr = first_extent;
- else
- bytenr = count_path->nodes[level]->start;
+ int recow;
+ int err = -ENOENT;
- cur_count = 0;
key.objectid = bytenr;
- key.offset = 0;
-
- btrfs_set_key_type(&key, BTRFS_EXTENT_ITEM_KEY);
- ret = btrfs_search_slot(NULL, extent_root, &key, path, 0, 0);
- if (ret < 0)
- goto out;
- BUG_ON(ret == 0);
-
- l = path->nodes[0];
- btrfs_item_key_to_cpu(l, &found_key, path->slots[0]);
-
- if (found_key.objectid != bytenr ||
- found_key.type != BTRFS_EXTENT_ITEM_KEY) {
- goto out;
+ if (parent) {
+ key.type = BTRFS_SHARED_DATA_REF_KEY;
+ key.offset = parent;
+ } else {
+ key.type = BTRFS_EXTENT_DATA_REF_KEY;
+ key.offset = hash_extent_data_ref(root_objectid,
+ owner, offset);
+ }
+again:
+ recow = 0;
+ ret = btrfs_search_slot(trans, root, &key, path, -1, 1);
+ if (ret < 0) {
+ err = ret;
+ goto fail;
}
- item = btrfs_item_ptr(l, path->slots[0], struct btrfs_extent_item);
- refs = btrfs_extent_refs(l, item);
+ if (parent) {
+ if (!ret)
+ return 0;
+#ifdef BTRFS_COMPAT_EXTENT_TREE_V0
+ key.type = BTRFS_EXTENT_REF_V0_KEY;
+ btrfs_release_path(root, path);
+ ret = btrfs_search_slot(trans, root, &key, path, -1, 1);
+ if (ret < 0) {
+ err = ret;
+ goto fail;
+ }
+ if (!ret)
+ return 0;
+#endif
+ goto fail;
+ }
+
+ leaf = path->nodes[0];
+ nritems = btrfs_header_nritems(leaf);
while (1) {
- nritems = btrfs_header_nritems(l);
if (path->slots[0] >= nritems) {
- ret = btrfs_next_leaf(extent_root, path);
- if (ret == 0)
- continue;
+ ret = btrfs_next_leaf(root, path);
+ if (ret < 0)
+ err = ret;
+ if (ret)
+ goto fail;
+
+ leaf = path->nodes[0];
+ nritems = btrfs_header_nritems(leaf);
+ recow = 1;
+ }
+
+ btrfs_item_key_to_cpu(leaf, &key, path->slots[0]);
+ if (key.objectid != bytenr ||
+ key.type != BTRFS_EXTENT_DATA_REF_KEY)
+ goto fail;
+
+ ref = btrfs_item_ptr(leaf, path->slots[0],
+ struct btrfs_extent_data_ref);
+
+ if (match_extent_data_ref(leaf, ref, root_objectid,
+ owner, offset)) {
+ if (recow) {
+ btrfs_release_path(root, path);
+ goto again;
+ }
+ err = 0;
break;
}
- btrfs_item_key_to_cpu(l, &found_key, path->slots[0]);
- if (found_key.objectid != bytenr)
+ path->slots[0]++;
+ }
+fail:
+ return err;
+}
+
+static noinline int insert_extent_data_ref(struct btrfs_trans_handle *trans,
+ struct btrfs_root *root,
+ struct btrfs_path *path,
+ u64 bytenr, u64 parent,
+ u64 root_objectid, u64 owner,
+ u64 offset, int refs_to_add)
+{
+ struct btrfs_key key;
+ struct extent_buffer *leaf;
+ u32 size;
+ u32 num_refs;
+ int ret;
+
+ key.objectid = bytenr;
+ if (parent) {
+ key.type = BTRFS_SHARED_DATA_REF_KEY;
+ key.offset = parent;
+ size = sizeof(struct btrfs_shared_data_ref);
+ } else {
+ key.type = BTRFS_EXTENT_DATA_REF_KEY;
+ key.offset = hash_extent_data_ref(root_objectid,
+ owner, offset);
+ size = sizeof(struct btrfs_extent_data_ref);
+ }
+
+ ret = btrfs_insert_empty_item(trans, root, path, &key, size);
+ if (ret && ret != -EEXIST)
+ goto fail;
+
+ leaf = path->nodes[0];
+ if (parent) {
+ struct btrfs_shared_data_ref *ref;
+ ref = btrfs_item_ptr(leaf, path->slots[0],
+ struct btrfs_shared_data_ref);
+ if (ret == 0) {
+ btrfs_set_shared_data_ref_count(leaf, ref, refs_to_add);
+ } else {
+ num_refs = btrfs_shared_data_ref_count(leaf, ref);
+ num_refs += refs_to_add;
+ btrfs_set_shared_data_ref_count(leaf, ref, num_refs);
+ }
+ } else {
+ struct btrfs_extent_data_ref *ref;
+ while (ret == -EEXIST) {
+ ref = btrfs_item_ptr(leaf, path->slots[0],
+ struct btrfs_extent_data_ref);
+ if (match_extent_data_ref(leaf, ref, root_objectid,
+ owner, offset))
+ break;
+ btrfs_release_path(root, path);
+
+ key.offset++;
+ ret = btrfs_insert_empty_item(trans, root, path, &key,
+ size);
+ if (ret && ret != -EEXIST)
+ goto fail;
+
+ leaf = path->nodes[0];
+ }
+ ref = btrfs_item_ptr(leaf, path->slots[0],
+ struct btrfs_extent_data_ref);
+ if (ret == 0) {
+ btrfs_set_extent_data_ref_root(leaf, ref,
+ root_objectid);
+ btrfs_set_extent_data_ref_objectid(leaf, ref, owner);
+ btrfs_set_extent_data_ref_offset(leaf, ref, offset);
+ btrfs_set_extent_data_ref_count(leaf, ref, refs_to_add);
+ } else {
+ num_refs = btrfs_extent_data_ref_count(leaf, ref);
+ num_refs += refs_to_add;
+ btrfs_set_extent_data_ref_count(leaf, ref, num_refs);
+ }
+ }
+ btrfs_mark_buffer_dirty(leaf);
+ ret = 0;
+fail:
+ btrfs_release_path(root, path);
+ return ret;
+}
+
+static noinline int remove_extent_data_ref(struct btrfs_trans_handle *trans,
+ struct btrfs_root *root,
+ struct btrfs_path *path,
+ int refs_to_drop)
+{
+ struct btrfs_key key;
+ struct btrfs_extent_data_ref *ref1 = NULL;
+ struct btrfs_shared_data_ref *ref2 = NULL;
+ struct extent_buffer *leaf;
+ u32 num_refs = 0;
+ int ret = 0;
+
+ leaf = path->nodes[0];
+ btrfs_item_key_to_cpu(leaf, &key, path->slots[0]);
+
+ if (key.type == BTRFS_EXTENT_DATA_REF_KEY) {
+ ref1 = btrfs_item_ptr(leaf, path->slots[0],
+ struct btrfs_extent_data_ref);
+ num_refs = btrfs_extent_data_ref_count(leaf, ref1);
+ } else if (key.type == BTRFS_SHARED_DATA_REF_KEY) {
+ ref2 = btrfs_item_ptr(leaf, path->slots[0],
+ struct btrfs_shared_data_ref);
+ num_refs = btrfs_shared_data_ref_count(leaf, ref2);
+#ifdef BTRFS_COMPAT_EXTENT_TREE_V0
+ } else if (key.type == BTRFS_EXTENT_REF_V0_KEY) {
+ struct btrfs_extent_ref_v0 *ref0;
+ ref0 = btrfs_item_ptr(leaf, path->slots[0],
+ struct btrfs_extent_ref_v0);
+ num_refs = btrfs_ref_count_v0(leaf, ref0);
+#endif
+ } else {
+ BUG();
+ }
+
+ BUG_ON(num_refs < refs_to_drop);
+ num_refs -= refs_to_drop;
+
+ if (num_refs == 0) {
+ ret = btrfs_del_item(trans, root, path);
+ } else {
+ if (key.type == BTRFS_EXTENT_DATA_REF_KEY)
+ btrfs_set_extent_data_ref_count(leaf, ref1, num_refs);
+ else if (key.type == BTRFS_SHARED_DATA_REF_KEY)
+ btrfs_set_shared_data_ref_count(leaf, ref2, num_refs);
+#ifdef BTRFS_COMPAT_EXTENT_TREE_V0
+ else {
+ struct btrfs_extent_ref_v0 *ref0;
+ ref0 = btrfs_item_ptr(leaf, path->slots[0],
+ struct btrfs_extent_ref_v0);
+ btrfs_set_ref_count_v0(leaf, ref0, num_refs);
+ }
+#endif
+ btrfs_mark_buffer_dirty(leaf);
+ }
+ return ret;
+}
+
+static noinline u32 extent_data_ref_count(struct btrfs_root *root,
+ struct btrfs_path *path,
+ struct btrfs_extent_inline_ref *iref)
+{
+ struct btrfs_key key;
+ struct extent_buffer *leaf;
+ struct btrfs_extent_data_ref *ref1;
+ struct btrfs_shared_data_ref *ref2;
+ u32 num_refs = 0;
+
+ leaf = path->nodes[0];
+ btrfs_item_key_to_cpu(leaf, &key, path->slots[0]);
+ if (iref) {
+ if (btrfs_extent_inline_ref_type(leaf, iref) ==
+ BTRFS_EXTENT_DATA_REF_KEY) {
+ ref1 = (struct btrfs_extent_data_ref *)(&iref->offset);
+ num_refs = btrfs_extent_data_ref_count(leaf, ref1);
+ } else {
+ ref2 = (struct btrfs_shared_data_ref *)(iref + 1);
+ num_refs = btrfs_shared_data_ref_count(leaf, ref2);
+ }
+ } else if (key.type == BTRFS_EXTENT_DATA_REF_KEY) {
+ ref1 = btrfs_item_ptr(leaf, path->slots[0],
+ struct btrfs_extent_data_ref);
+ num_refs = btrfs_extent_data_ref_count(leaf, ref1);
+ } else if (key.type == BTRFS_SHARED_DATA_REF_KEY) {
+ ref2 = btrfs_item_ptr(leaf, path->slots[0],
+ struct btrfs_shared_data_ref);
+ num_refs = btrfs_shared_data_ref_count(leaf, ref2);
+#ifdef BTRFS_COMPAT_EXTENT_TREE_V0
+ } else if (key.type == BTRFS_EXTENT_REF_V0_KEY) {
+ struct btrfs_extent_ref_v0 *ref0;
+ ref0 = btrfs_item_ptr(leaf, path->slots[0],
+ struct btrfs_extent_ref_v0);
+ num_refs = btrfs_ref_count_v0(leaf, ref0);
+#endif
+ } else {
+ BUG();
+ }
+ return num_refs;
+}
+
+static noinline int lookup_tree_block_ref(struct btrfs_trans_handle *trans,
+ struct btrfs_root *root,
+ struct btrfs_path *path,
+ u64 bytenr, u64 parent,
+ u64 root_objectid)
+{
+ struct btrfs_key key;
+ int ret;
+
+ key.objectid = bytenr;
+ if (parent) {
+ key.type = BTRFS_SHARED_BLOCK_REF_KEY;
+ key.offset = parent;
+ } else {
+ key.type = BTRFS_TREE_BLOCK_REF_KEY;
+ key.offset = root_objectid;
+ }
+
+ ret = btrfs_search_slot(trans, root, &key, path, -1, 1);
+ if (ret > 0)
+ ret = -ENOENT;
+#ifdef BTRFS_COMPAT_EXTENT_TREE_V0
+ if (ret == -ENOENT && parent) {
+ btrfs_release_path(root, path);
+ key.type = BTRFS_EXTENT_REF_V0_KEY;
+ ret = btrfs_search_slot(trans, root, &key, path, -1, 1);
+ if (ret > 0)
+ ret = -ENOENT;
+ }
+#endif
+ return ret;
+}
+
+static noinline int insert_tree_block_ref(struct btrfs_trans_handle *trans,
+ struct btrfs_root *root,
+ struct btrfs_path *path,
+ u64 bytenr, u64 parent,
+ u64 root_objectid)
+{
+ struct btrfs_key key;
+ int ret;
+
+ key.objectid = bytenr;
+ if (parent) {
+ key.type = BTRFS_SHARED_BLOCK_REF_KEY;
+ key.offset = parent;
+ } else {
+ key.type = BTRFS_TREE_BLOCK_REF_KEY;
+ key.offset = root_objectid;
+ }
+
+ ret = btrfs_insert_empty_item(trans, root, path, &key, 0);
+
+ btrfs_release_path(root, path);
+ return ret;
+}
+
+static inline int extent_ref_type(u64 parent, u64 owner)
+{
+ if (owner < BTRFS_FIRST_FREE_OBJECTID) {
+ if (parent > 0)
+ return BTRFS_SHARED_BLOCK_REF_KEY;
+ else
+ return BTRFS_TREE_BLOCK_REF_KEY;
+ } else {
+ if (parent > 0)
+ return BTRFS_SHARED_DATA_REF_KEY;
+ else
+ return BTRFS_EXTENT_DATA_REF_KEY;
+ }
+}
+
+static int find_next_key(struct btrfs_path *path, struct btrfs_key *key)
+
+{
+ int level;
+ for (level = 0; level < BTRFS_MAX_LEVEL; level++) {
+ if (!path->nodes[level])
break;
- if (found_key.type != BTRFS_EXTENT_REF_KEY) {
- path->slots[0]++;
+ if (path->slots[level] + 1 >=
+ btrfs_header_nritems(path->nodes[level]))
continue;
+ if (level == 0)
+ btrfs_item_key_to_cpu(path->nodes[level], key,
+ path->slots[level] + 1);
+ else
+ btrfs_node_key_to_cpu(path->nodes[level], key,
+ path->slots[level] + 1);
+ return 0;
+ }
+ return 1;
+}
+
+static int lookup_inline_extent_backref(struct btrfs_trans_handle *trans,
+ struct btrfs_root *root,
+ struct btrfs_path *path,
+ struct btrfs_extent_inline_ref **ref_ret,
+ u64 bytenr, u64 num_bytes,
+ u64 parent, u64 root_objectid,
+ u64 owner, u64 offset, int insert)
+{
+ struct btrfs_key key;
+ struct extent_buffer *leaf;
+ struct btrfs_extent_item *ei;
+ struct btrfs_extent_inline_ref *iref;
+ u64 flags;
+ u32 item_size;
+ unsigned long ptr;
+ unsigned long end;
+ int extra_size;
+ int type;
+ int want;
+ int ret;
+ int err = 0;
+
+ key.objectid = bytenr;
+ key.type = BTRFS_EXTENT_ITEM_KEY;
+ key.offset = num_bytes;
+
+ want = extent_ref_type(parent, owner);
+ if (insert)
+ extra_size = btrfs_extent_inline_ref_size(want);
+ else
+ extra_size = -1;
+ ret = btrfs_search_slot(trans, root, &key, path, extra_size, 1);
+ if (ret < 0) {
+ err = ret;
+ goto out;
+ }
+ if (ret) {
+ printf("Failed to find [%llu, %u, %llu]\n", key.objectid, key.type, key.offset);
+ return -ENOENT;
+ }
+
+ BUG_ON(ret);
+
+ leaf = path->nodes[0];
+ item_size = btrfs_item_size_nr(leaf, path->slots[0]);
+#ifdef BTRFS_COMPAT_EXTENT_TREE_V0
+ if (item_size < sizeof(*ei)) {
+ if (!insert) {
+ err = -ENOENT;
+ goto out;
}
+ ret = convert_extent_item_v0(trans, root, path, owner,
+ extra_size);
+ if (ret < 0) {
+ err = ret;
+ goto out;
+ }
+ leaf = path->nodes[0];
+ item_size = btrfs_item_size_nr(leaf, path->slots[0]);
+ }
+#endif
+ if (item_size < sizeof(*ei)) {
+ printf("Size is %u, needs to be %u, slot %d\n",
+ (unsigned)item_size,
+ (unsigned)sizeof(*ei), path->slots[0]);
+ btrfs_print_leaf(root, leaf);
+ return -EINVAL;
+ }
+ BUG_ON(item_size < sizeof(*ei));
+
+ ei = btrfs_item_ptr(leaf, path->slots[0], struct btrfs_extent_item);
+ flags = btrfs_extent_flags(leaf, ei);
+
+ ptr = (unsigned long)(ei + 1);
+ end = (unsigned long)ei + item_size;
- cur_count++;
- ref_item = btrfs_item_ptr(l, path->slots[0],
- struct btrfs_extent_ref);
- found_objectid = btrfs_ref_root(l, ref_item);
+ if (flags & BTRFS_EXTENT_FLAG_TREE_BLOCK) {
+ ptr += sizeof(struct btrfs_tree_block_info);
+ BUG_ON(ptr > end);
+ } else {
+ if (!(flags & BTRFS_EXTENT_FLAG_DATA)) {
+ return -EIO;
+ }
+ }
- if (found_objectid != root_objectid) {
- total_count = 2;
+ err = -ENOENT;
+ while (1) {
+ if (ptr >= end) {
+ WARN_ON(ptr > end);
+ break;
+ }
+ iref = (struct btrfs_extent_inline_ref *)ptr;
+ type = btrfs_extent_inline_ref_type(leaf, iref);
+ if (want < type)
+ break;
+ if (want > type) {
+ ptr += btrfs_extent_inline_ref_size(type);
+ continue;
+ }
+
+ if (type == BTRFS_EXTENT_DATA_REF_KEY) {
+ struct btrfs_extent_data_ref *dref;
+ dref = (struct btrfs_extent_data_ref *)(&iref->offset);
+ if (match_extent_data_ref(leaf, dref, root_objectid,
+ owner, offset)) {
+ err = 0;
+ break;
+ }
+ if (hash_extent_data_ref_item(leaf, dref) <
+ hash_extent_data_ref(root_objectid, owner, offset))
+ break;
+ } else {
+ u64 ref_offset;
+ ref_offset = btrfs_extent_inline_ref_offset(leaf, iref);
+ if (parent > 0) {
+ if (parent == ref_offset) {
+ err = 0;
+ break;
+ }
+ if (ref_offset < parent)
+ break;
+ } else {
+ if (root_objectid == ref_offset) {
+ err = 0;
+ break;
+ }
+ if (ref_offset < root_objectid)
+ break;
+ }
+ }
+ ptr += btrfs_extent_inline_ref_size(type);
+ }
+ if (err == -ENOENT && insert) {
+ if (item_size + extra_size >=
+ BTRFS_MAX_EXTENT_ITEM_SIZE(root)) {
+ err = -EAGAIN;
+ goto out;
+ }
+ /*
+ * To add new inline back ref, we have to make sure
+ * there is no corresponding back ref item.
+ * For simplicity, we just do not add new inline back
+ * ref if there is any back ref item.
+ */
+ if (find_next_key(path, &key) == 0 && key.objectid == bytenr &&
+ key.type < BTRFS_BLOCK_GROUP_ITEM_KEY) {
+ err = -EAGAIN;
goto out;
}
- total_count = 1;
- path->slots[0]++;
}
- if (cur_count == 0) {
- total_count = 0;
- goto out;
+ *ref_ret = (struct btrfs_extent_inline_ref *)ptr;
+out:
+ return err;
+}
+
+static int setup_inline_extent_backref(struct btrfs_trans_handle *trans,
+ struct btrfs_root *root,
+ struct btrfs_path *path,
+ struct btrfs_extent_inline_ref *iref,
+ u64 parent, u64 root_objectid,
+ u64 owner, u64 offset, int refs_to_add)
+{
+ struct extent_buffer *leaf;
+ struct btrfs_extent_item *ei;
+ unsigned long ptr;
+ unsigned long end;
+ unsigned long item_offset;
+ u64 refs;
+ int size;
+ int type;
+ int ret;
+
+ leaf = path->nodes[0];
+ ei = btrfs_item_ptr(leaf, path->slots[0], struct btrfs_extent_item);
+ item_offset = (unsigned long)iref - (unsigned long)ei;
+
+ type = extent_ref_type(parent, owner);
+ size = btrfs_extent_inline_ref_size(type);
+
+ ret = btrfs_extend_item(trans, root, path, size);
+ BUG_ON(ret);
+
+ ei = btrfs_item_ptr(leaf, path->slots[0], struct btrfs_extent_item);
+ refs = btrfs_extent_refs(leaf, ei);
+ refs += refs_to_add;
+ btrfs_set_extent_refs(leaf, ei, refs);
+
+ ptr = (unsigned long)ei + item_offset;
+ end = (unsigned long)ei + btrfs_item_size_nr(leaf, path->slots[0]);
+ if (ptr < end - size)
+ memmove_extent_buffer(leaf, ptr + size, ptr,
+ end - size - ptr);
+
+ iref = (struct btrfs_extent_inline_ref *)ptr;
+ btrfs_set_extent_inline_ref_type(leaf, iref, type);
+ if (type == BTRFS_EXTENT_DATA_REF_KEY) {
+ struct btrfs_extent_data_ref *dref;
+ dref = (struct btrfs_extent_data_ref *)(&iref->offset);
+ btrfs_set_extent_data_ref_root(leaf, dref, root_objectid);
+ btrfs_set_extent_data_ref_objectid(leaf, dref, owner);
+ btrfs_set_extent_data_ref_offset(leaf, dref, offset);
+ btrfs_set_extent_data_ref_count(leaf, dref, refs_to_add);
+ } else if (type == BTRFS_SHARED_DATA_REF_KEY) {
+ struct btrfs_shared_data_ref *sref;
+ sref = (struct btrfs_shared_data_ref *)(iref + 1);
+ btrfs_set_shared_data_ref_count(leaf, sref, refs_to_add);
+ btrfs_set_extent_inline_ref_offset(leaf, iref, parent);
+ } else if (type == BTRFS_SHARED_BLOCK_REF_KEY) {
+ btrfs_set_extent_inline_ref_offset(leaf, iref, parent);
+ } else {
+ btrfs_set_extent_inline_ref_offset(leaf, iref, root_objectid);
}
- if (level >= 0 && root->node == count_path->nodes[level])
- goto out;
- level++;
+ btrfs_mark_buffer_dirty(leaf);
+ return 0;
+}
+
+static int lookup_extent_backref(struct btrfs_trans_handle *trans,
+ struct btrfs_root *root,
+ struct btrfs_path *path,
+ struct btrfs_extent_inline_ref **ref_ret,
+ u64 bytenr, u64 num_bytes, u64 parent,
+ u64 root_objectid, u64 owner, u64 offset)
+{
+ int ret;
+
+ ret = lookup_inline_extent_backref(trans, root, path, ref_ret,
+ bytenr, num_bytes, parent,
+ root_objectid, owner, offset, 0);
+ if (ret != -ENOENT)
+ return ret;
+
btrfs_release_path(root, path);
- goto again;
+ *ref_ret = NULL;
+ if (owner < BTRFS_FIRST_FREE_OBJECTID) {
+ ret = lookup_tree_block_ref(trans, root, path, bytenr, parent,
+ root_objectid);
+ } else {
+ ret = lookup_extent_data_ref(trans, root, path, bytenr, parent,
+ root_objectid, owner, offset);
+ }
+ return ret;
+}
+
+static int update_inline_extent_backref(struct btrfs_trans_handle *trans,
+ struct btrfs_root *root,
+ struct btrfs_path *path,
+ struct btrfs_extent_inline_ref *iref,
+ int refs_to_mod)
+{
+ struct extent_buffer *leaf;
+ struct btrfs_extent_item *ei;
+ struct btrfs_extent_data_ref *dref = NULL;
+ struct btrfs_shared_data_ref *sref = NULL;
+ unsigned long ptr;
+ unsigned long end;
+ u32 item_size;
+ int size;
+ int type;
+ int ret;
+ u64 refs;
+
+ leaf = path->nodes[0];
+ ei = btrfs_item_ptr(leaf, path->slots[0], struct btrfs_extent_item);
+ refs = btrfs_extent_refs(leaf, ei);
+ WARN_ON(refs_to_mod < 0 && refs + refs_to_mod <= 0);
+ refs += refs_to_mod;
+ btrfs_set_extent_refs(leaf, ei, refs);
+
+ type = btrfs_extent_inline_ref_type(leaf, iref);
+
+ if (type == BTRFS_EXTENT_DATA_REF_KEY) {
+ dref = (struct btrfs_extent_data_ref *)(&iref->offset);
+ refs = btrfs_extent_data_ref_count(leaf, dref);
+ } else if (type == BTRFS_SHARED_DATA_REF_KEY) {
+ sref = (struct btrfs_shared_data_ref *)(iref + 1);
+ refs = btrfs_shared_data_ref_count(leaf, sref);
+ } else {
+ refs = 1;
+ BUG_ON(refs_to_mod != -1);
+ }
+
+ BUG_ON(refs_to_mod < 0 && refs < -refs_to_mod);
+ refs += refs_to_mod;
+
+ if (refs > 0) {
+ if (type == BTRFS_EXTENT_DATA_REF_KEY)
+ btrfs_set_extent_data_ref_count(leaf, dref, refs);
+ else
+ btrfs_set_shared_data_ref_count(leaf, sref, refs);
+ } else {
+ size = btrfs_extent_inline_ref_size(type);
+ item_size = btrfs_item_size_nr(leaf, path->slots[0]);
+ ptr = (unsigned long)iref;
+ end = (unsigned long)ei + item_size;
+ if (ptr + size < end)
+ memmove_extent_buffer(leaf, ptr, ptr + size,
+ end - ptr - size);
+ item_size -= size;
+ ret = btrfs_truncate_item(trans, root, path, item_size, 1);
+ BUG_ON(ret);
+ }
+ btrfs_mark_buffer_dirty(leaf);
+ return 0;
+}
+
+static int insert_inline_extent_backref(struct btrfs_trans_handle *trans,
+ struct btrfs_root *root,
+ struct btrfs_path *path,
+ u64 bytenr, u64 num_bytes, u64 parent,
+ u64 root_objectid, u64 owner,
+ u64 offset, int refs_to_add)
+{
+ struct btrfs_extent_inline_ref *iref;
+ int ret;
+
+ ret = lookup_inline_extent_backref(trans, root, path, &iref,
+ bytenr, num_bytes, parent,
+ root_objectid, owner, offset, 1);
+ if (ret == 0) {
+ BUG_ON(owner < BTRFS_FIRST_FREE_OBJECTID);
+ ret = update_inline_extent_backref(trans, root, path, iref,
+ refs_to_add);
+ } else if (ret == -ENOENT) {
+ ret = setup_inline_extent_backref(trans, root, path, iref,
+ parent, root_objectid,
+ owner, offset, refs_to_add);
+ }
+ return ret;
+}
+
+static int insert_extent_backref(struct btrfs_trans_handle *trans,
+ struct btrfs_root *root,
+ struct btrfs_path *path,
+ u64 bytenr, u64 parent, u64 root_objectid,
+ u64 owner, u64 offset, int refs_to_add)
+{
+ int ret;
+
+ if (owner >= BTRFS_FIRST_FREE_OBJECTID) {
+ ret = insert_extent_data_ref(trans, root, path, bytenr,
+ parent, root_objectid,
+ owner, offset, refs_to_add);
+ } else {
+ BUG_ON(refs_to_add != 1);
+ ret = insert_tree_block_ref(trans, root, path, bytenr,
+ parent, root_objectid);
+ }
+ return ret;
+}
+
+static int remove_extent_backref(struct btrfs_trans_handle *trans,
+ struct btrfs_root *root,
+ struct btrfs_path *path,
+ struct btrfs_extent_inline_ref *iref,
+ int refs_to_drop, int is_data)
+{
+ int ret;
+
+ BUG_ON(!is_data && refs_to_drop != 1);
+ if (iref) {
+ ret = update_inline_extent_backref(trans, root, path, iref,
+ -refs_to_drop);
+ } else if (is_data) {
+ ret = remove_extent_data_ref(trans, root, path, refs_to_drop);
+ } else {
+ ret = btrfs_del_item(trans, root, path);
+ }
+ return ret;
+}
+
+int btrfs_inc_extent_ref(struct btrfs_trans_handle *trans,
+ struct btrfs_root *root,
+ u64 bytenr, u64 num_bytes, u64 parent,
+ u64 root_objectid, u64 owner, u64 offset)
+{
+ struct btrfs_path *path;
+ struct extent_buffer *leaf;
+ struct btrfs_extent_item *item;
+ u64 refs;
+ int ret;
+ int err = 0;
+
+ path = btrfs_alloc_path();
+ if (!path)
+ return -ENOMEM;
+
+ path->reada = 1;
+ path->leave_spinning = 1;
+
+ ret = insert_inline_extent_backref(trans, root->fs_info->extent_root,
+ path, bytenr, num_bytes, parent,
+ root_objectid, owner, offset, 1);
+ if (ret == 0)
+ goto out;
+
+ if (ret != -EAGAIN) {
+ err = ret;
+ goto out;
+ }
+
+ leaf = path->nodes[0];
+ item = btrfs_item_ptr(leaf, path->slots[0], struct btrfs_extent_item);
+ refs = btrfs_extent_refs(leaf, item);
+ btrfs_set_extent_refs(leaf, item, refs + 1);
+
+ btrfs_mark_buffer_dirty(leaf);
+ btrfs_release_path(root->fs_info->extent_root, path);
+
+ path->reada = 1;
+ path->leave_spinning = 1;
+
+ /* now insert the actual backref */
+ ret = insert_extent_backref(trans, root->fs_info->extent_root,
+ path, bytenr, parent, root_objectid,
+ owner, offset, 1);
+ if (ret)
+ err = ret;
out:
btrfs_free_path(path);
- return total_count;
+ finish_current_insert(trans, root->fs_info->extent_root);
+ del_pending_extents(trans, root->fs_info->extent_root);
+ BUG_ON(err);
+ return err;
+}
+int btrfs_extent_post_op(struct btrfs_trans_handle *trans,
+ struct btrfs_root *root)
+{
+ finish_current_insert(trans, root->fs_info->extent_root);
+ del_pending_extents(trans, root->fs_info->extent_root);
+ return 0;
}
-int btrfs_inc_root_ref(struct btrfs_trans_handle *trans,
- struct btrfs_root *root, u64 owner_objectid)
+int btrfs_lookup_extent_info(struct btrfs_trans_handle *trans,
+ struct btrfs_root *root, u64 bytenr,
+ u64 num_bytes, u64 *refs, u64 *flags)
{
- u64 generation;
- u64 key_objectid;
- u64 level;
- u32 nritems;
- struct btrfs_disk_key disk_key;
+ struct btrfs_path *path;
+ int ret;
+ struct btrfs_key key;
+ struct extent_buffer *l;
+ struct btrfs_extent_item *item;
+ u32 item_size;
+ u64 num_refs;
+ u64 extent_flags;
- level = btrfs_header_level(root->node);
- generation = trans->transid;
- nritems = btrfs_header_nritems(root->node);
- if (nritems > 0) {
- if (level == 0)
- btrfs_item_key(root->node, &disk_key, 0);
- else
- btrfs_node_key(root->node, &disk_key, 0);
- key_objectid = btrfs_disk_key_objectid(&disk_key);
+ WARN_ON(num_bytes < root->sectorsize);
+ path = btrfs_alloc_path();
+ path->reada = 1;
+ key.objectid = bytenr;
+ key.offset = num_bytes;
+ btrfs_set_key_type(&key, BTRFS_EXTENT_ITEM_KEY);
+ ret = btrfs_search_slot(trans, root->fs_info->extent_root, &key, path,
+ 0, 0);
+ if (ret < 0)
+ goto out;
+ if (ret != 0) {
+ ret = -EIO;
+ goto out;
+ }
+
+ l = path->nodes[0];
+ item_size = btrfs_item_size_nr(l, path->slots[0]);
+ if (item_size >= sizeof(*item)) {
+ item = btrfs_item_ptr(l, path->slots[0],
+ struct btrfs_extent_item);
+ num_refs = btrfs_extent_refs(l, item);
+ extent_flags = btrfs_extent_flags(l, item);
} else {
- key_objectid = 0;
+#ifdef BTRFS_COMPAT_EXTENT_TREE_V0
+ struct btrfs_extent_item_v0 *ei0;
+ BUG_ON(item_size != sizeof(*ei0));
+ ei0 = btrfs_item_ptr(l, path->slots[0],
+ struct btrfs_extent_item_v0);
+ num_refs = btrfs_extent_refs_v0(l, ei0);
+ /* FIXME: this isn't correct for data */
+ extent_flags = BTRFS_BLOCK_FLAG_FULL_BACKREF;
+#else
+ BUG();
+#endif
}
- return btrfs_inc_extent_ref(trans, root, root->node->start,
- root->node->len, owner_objectid,
- generation, level, key_objectid);
+ item = btrfs_item_ptr(l, path->slots[0], struct btrfs_extent_item);
+ if (refs)
+ *refs = num_refs;
+ if (flags)
+ *flags = extent_flags;
+out:
+ btrfs_free_path(path);
+ return 0;
}
-int btrfs_inc_ref(struct btrfs_trans_handle *trans, struct btrfs_root *root,
- struct extent_buffer *buf)
+int btrfs_set_block_flags(struct btrfs_trans_handle *trans,
+ struct btrfs_root *root,
+ u64 bytenr, u64 num_bytes, u64 flags)
+{
+ struct btrfs_path *path;
+ int ret;
+ struct btrfs_key key;
+ struct extent_buffer *l;
+ struct btrfs_extent_item *item;
+ u32 item_size;
+
+ WARN_ON(num_bytes < root->sectorsize);
+ path = btrfs_alloc_path();
+ path->reada = 1;
+ key.objectid = bytenr;
+ key.offset = num_bytes;
+ btrfs_set_key_type(&key, BTRFS_EXTENT_ITEM_KEY);
+ ret = btrfs_search_slot(trans, root->fs_info->extent_root, &key, path,
+ 0, 0);
+ if (ret < 0)
+ goto out;
+ if (ret != 0) {
+ btrfs_print_leaf(root, path->nodes[0]);
+ printk("failed to find block number %Lu\n",
+ (unsigned long long)bytenr);
+ BUG();
+ }
+ l = path->nodes[0];
+ item_size = btrfs_item_size_nr(l, path->slots[0]);
+#ifdef BTRFS_COMPAT_EXTENT_TREE_V0
+ if (item_size < sizeof(*item)) {
+ ret = convert_extent_item_v0(trans, root->fs_info->extent_root,
+ path, (u64)-1, 0);
+ if (ret < 0)
+ goto out;
+
+ l = path->nodes[0];
+ item_size = btrfs_item_size_nr(l, path->slots[0]);
+ }
+#endif
+ BUG_ON(item_size < sizeof(*item));
+ item = btrfs_item_ptr(l, path->slots[0], struct btrfs_extent_item);
+ flags |= btrfs_extent_flags(l, item);
+ btrfs_set_extent_flags(l, item, flags);
+out:
+ btrfs_free_path(path);
+ finish_current_insert(trans, root->fs_info->extent_root);
+ del_pending_extents(trans, root->fs_info->extent_root);
+ return ret;
+}
+
+static int __btrfs_mod_ref(struct btrfs_trans_handle *trans,
+ struct btrfs_root *root,
+ struct extent_buffer *buf,
+ int record_parent, int inc)
{
u64 bytenr;
+ u64 num_bytes;
+ u64 parent;
+ u64 ref_root;
u32 nritems;
struct btrfs_key key;
struct btrfs_file_extent_item *fi;
int i;
int level;
- int ret;
- int faili;
+ int ret = 0;
+ int (*process_func)(struct btrfs_trans_handle *trans,
+ struct btrfs_root *root,
+ u64, u64, u64, u64, u64, u64);
- if (!root->ref_cows)
+ ref_root = btrfs_header_owner(buf);
+ nritems = btrfs_header_nritems(buf);
+ level = btrfs_header_level(buf);
+
+ if (!root->ref_cows && level == 0)
return 0;
- level = btrfs_header_level(buf);
- nritems = btrfs_header_nritems(buf);
+ if (inc)
+ process_func = btrfs_inc_extent_ref;
+ else
+ process_func = btrfs_free_extent;
+
+ if (record_parent)
+ parent = buf->start;
+ else
+ parent = 0;
+
for (i = 0; i < nritems; i++) {
+ cond_resched();
if (level == 0) {
- u64 disk_bytenr;
btrfs_item_key_to_cpu(buf, &key, i);
if (btrfs_key_type(&key) != BTRFS_EXTENT_DATA_KEY)
continue;
if (btrfs_file_extent_type(buf, fi) ==
BTRFS_FILE_EXTENT_INLINE)
continue;
- disk_bytenr = btrfs_file_extent_disk_bytenr(buf, fi);
- if (disk_bytenr == 0)
+ bytenr = btrfs_file_extent_disk_bytenr(buf, fi);
+ if (bytenr == 0)
continue;
- ret = btrfs_inc_extent_ref(trans, root, disk_bytenr,
- btrfs_file_extent_disk_num_bytes(buf, fi),
- root->root_key.objectid, trans->transid,
- key.objectid, key.offset);
+
+ num_bytes = btrfs_file_extent_disk_num_bytes(buf, fi);
+ key.offset -= btrfs_file_extent_offset(buf, fi);
+ ret = process_func(trans, root, bytenr, num_bytes,
+ parent, ref_root, key.objectid,
+ key.offset);
if (ret) {
- faili = i;
+ WARN_ON(1);
goto fail;
}
} else {
bytenr = btrfs_node_blockptr(buf, i);
- btrfs_node_key_to_cpu(buf, &key, i);
- ret = btrfs_inc_extent_ref(trans, root, bytenr,
- btrfs_level_size(root, level - 1),
- root->root_key.objectid,
- trans->transid,
- level - 1, key.objectid);
+ num_bytes = btrfs_level_size(root, level - 1);
+ ret = process_func(trans, root, bytenr, num_bytes,
+ parent, ref_root, level - 1, 0);
if (ret) {
- faili = i;
+ WARN_ON(1);
goto fail;
}
}
return 0;
fail:
WARN_ON(1);
-#if 0
- for (i =0; i < faili; i++) {
- if (level == 0) {
- u64 disk_bytenr;
- btrfs_item_key_to_cpu(buf, &key, i);
- if (btrfs_key_type(&key) != BTRFS_EXTENT_DATA_KEY)
- continue;
- fi = btrfs_item_ptr(buf, i,
- struct btrfs_file_extent_item);
- if (btrfs_file_extent_type(buf, fi) ==
- BTRFS_FILE_EXTENT_INLINE)
- continue;
- disk_bytenr = btrfs_file_extent_disk_bytenr(buf, fi);
- if (disk_bytenr == 0)
- continue;
- err = btrfs_free_extent(trans, root, disk_bytenr,
- btrfs_file_extent_disk_num_bytes(buf,
- fi), 0);
- BUG_ON(err);
- } else {
- bytenr = btrfs_node_blockptr(buf, i);
- err = btrfs_free_extent(trans, root, bytenr,
- btrfs_level_size(root, level - 1), 0);
- BUG_ON(err);
- }
- }
-#endif
return ret;
}
+int btrfs_inc_ref(struct btrfs_trans_handle *trans, struct btrfs_root *root,
+ struct extent_buffer *buf, int record_parent)
+{
+ return __btrfs_mod_ref(trans, root, buf, record_parent, 1);
+}
+
+int btrfs_dec_ref(struct btrfs_trans_handle *trans, struct btrfs_root *root,
+ struct extent_buffer *buf, int record_parent)
+{
+ return __btrfs_mod_ref(trans, root, buf, record_parent, 0);
+}
+
static int write_one_cache_group(struct btrfs_trans_handle *trans,
struct btrfs_root *root,
struct btrfs_path *path,
int btrfs_write_dirty_block_groups(struct btrfs_trans_handle *trans,
struct btrfs_root *root)
{
- struct extent_map_tree *block_group_cache;
+ struct extent_io_tree *block_group_cache;
struct btrfs_block_group_cache *cache;
int ret;
- int err = 0;
- int werr = 0;
struct btrfs_path *path;
u64 last = 0;
u64 start;
while(1) {
ret = find_first_extent_bit(block_group_cache, last,
&start, &end, BLOCK_GROUP_DIRTY);
- if (ret)
- break;
+ if (ret) {
+ if (last == 0)
+ break;
+ last = 0;
+ continue;
+ }
+
+ last = end + 1;
+ ret = get_state_private(block_group_cache, start, &ptr);
+ BUG_ON(ret);
+
+ clear_extent_bits(block_group_cache, start, end,
+ BLOCK_GROUP_DIRTY, GFP_NOFS);
+
+ cache = (struct btrfs_block_group_cache *)(unsigned long)ptr;
+ ret = write_one_cache_group(trans, root, path, cache);
+ }
+ btrfs_free_path(path);
+ return 0;
+}
+
+static struct btrfs_space_info *__find_space_info(struct btrfs_fs_info *info,
+ u64 flags)
+{
+ struct list_head *head = &info->space_info;
+ struct list_head *cur;
+ struct btrfs_space_info *found;
+ list_for_each(cur, head) {
+ found = list_entry(cur, struct btrfs_space_info, list);
+ if (found->flags & flags)
+ return found;
+ }
+ return NULL;
+
+}
+
+static int update_space_info(struct btrfs_fs_info *info, u64 flags,
+ u64 total_bytes, u64 bytes_used,
+ struct btrfs_space_info **space_info)
+{
+ struct btrfs_space_info *found;
+
+ found = __find_space_info(info, flags);
+ if (found) {
+ found->total_bytes += total_bytes;
+ found->bytes_used += bytes_used;
+ if (found->total_bytes < found->bytes_used) {
+ fprintf(stderr, "warning, bad space info total_bytes "
+ "%llu used %llu\n",
+ (unsigned long long)found->total_bytes,
+ (unsigned long long)found->bytes_used);
+ }
+ *space_info = found;
+ return 0;
+ }
+ found = kmalloc(sizeof(*found), GFP_NOFS);
+ if (!found)
+ return -ENOMEM;
+
+ list_add(&found->list, &info->space_info);
+ found->flags = flags;
+ found->total_bytes = total_bytes;
+ found->bytes_used = bytes_used;
+ found->bytes_pinned = 0;
+ found->full = 0;
+ *space_info = found;
+ return 0;
+}
+
+
+static void set_avail_alloc_bits(struct btrfs_fs_info *fs_info, u64 flags)
+{
+ u64 extra_flags = flags & (BTRFS_BLOCK_GROUP_RAID0 |
+ BTRFS_BLOCK_GROUP_RAID1 |
+ BTRFS_BLOCK_GROUP_RAID10 |
+ BTRFS_BLOCK_GROUP_RAID5 |
+ BTRFS_BLOCK_GROUP_RAID6 |
+ BTRFS_BLOCK_GROUP_DUP);
+ if (extra_flags) {
+ if (flags & BTRFS_BLOCK_GROUP_DATA)
+ fs_info->avail_data_alloc_bits |= extra_flags;
+ if (flags & BTRFS_BLOCK_GROUP_METADATA)
+ fs_info->avail_metadata_alloc_bits |= extra_flags;
+ if (flags & BTRFS_BLOCK_GROUP_SYSTEM)
+ fs_info->avail_system_alloc_bits |= extra_flags;
+ }
+}
+
+static int do_chunk_alloc(struct btrfs_trans_handle *trans,
+ struct btrfs_root *extent_root, u64 alloc_bytes,
+ u64 flags)
+{
+ struct btrfs_space_info *space_info;
+ u64 thresh;
+ u64 start;
+ u64 num_bytes;
+ int ret;
+
+ space_info = __find_space_info(extent_root->fs_info, flags);
+ if (!space_info) {
+ ret = update_space_info(extent_root->fs_info, flags,
+ 0, 0, &space_info);
+ BUG_ON(ret);
+ }
+ BUG_ON(!space_info);
- last = end + 1;
- ret = get_state_private(block_group_cache, start, &ptr);
- if (ret)
- break;
+ if (space_info->full)
+ return 0;
- cache = (struct btrfs_block_group_cache *)(unsigned long)ptr;
- err = write_one_cache_group(trans, root,
- path, cache);
- /*
- * if we fail to write the cache group, we want
- * to keep it marked dirty in hopes that a later
- * write will work
- */
- if (err) {
- werr = err;
- continue;
- }
- clear_extent_bits(block_group_cache, start, end,
- BLOCK_GROUP_DIRTY, GFP_NOFS);
+ thresh = div_factor(space_info->total_bytes, 7);
+ if ((space_info->bytes_used + space_info->bytes_pinned + alloc_bytes) <
+ thresh)
+ return 0;
+
+ ret = btrfs_alloc_chunk(trans, extent_root, &start, &num_bytes,
+ space_info->flags);
+ if (ret == -ENOSPC) {
+ space_info->full = 1;
+ return 0;
}
- btrfs_free_path(path);
- return werr;
+
+ BUG_ON(ret);
+
+ ret = btrfs_make_block_group(trans, extent_root, 0, space_info->flags,
+ BTRFS_FIRST_CHUNK_TREE_OBJECTID, start, num_bytes);
+ BUG_ON(ret);
+ return 0;
}
static int update_block_group(struct btrfs_trans_handle *trans,
struct btrfs_root *root,
u64 bytenr, u64 num_bytes, int alloc,
- int mark_free, int data)
+ int mark_free)
{
struct btrfs_block_group_cache *cache;
struct btrfs_fs_info *info = root->fs_info;
u64 start;
u64 end;
+ /* block accounting for super block */
+ old_val = btrfs_super_bytes_used(info->super_copy);
+ if (alloc)
+ old_val += num_bytes;
+ else
+ old_val -= num_bytes;
+ btrfs_set_super_bytes_used(info->super_copy, old_val);
+
+ /* block accounting for root item */
+ old_val = btrfs_root_used(&root->root_item);
+ if (alloc)
+ old_val += num_bytes;
+ else
+ old_val -= num_bytes;
+ btrfs_set_root_used(&root->root_item, old_val);
+
while(total) {
cache = btrfs_lookup_block_group(info, bytenr);
if (!cache) {
old_val = btrfs_block_group_used(&cache->item);
num_bytes = min(total, cache->key.offset - byte_in_group);
+
if (alloc) {
- if (cache->data != data &&
- old_val < (cache->key.offset >> 1)) {
- int bit_to_clear;
- int bit_to_set;
- cache->data = data;
- if (data) {
- bit_to_clear = BLOCK_GROUP_METADATA;
- bit_to_set = BLOCK_GROUP_DATA;
- cache->item.flags &=
- ~BTRFS_BLOCK_GROUP_MIXED;
- cache->item.flags |=
- BTRFS_BLOCK_GROUP_DATA;
- } else {
- bit_to_clear = BLOCK_GROUP_DATA;
- bit_to_set = BLOCK_GROUP_METADATA;
- cache->item.flags &=
- ~BTRFS_BLOCK_GROUP_MIXED;
- cache->item.flags &=
- ~BTRFS_BLOCK_GROUP_DATA;
- }
- clear_extent_bits(&info->block_group_cache,
- start, end, bit_to_clear,
- GFP_NOFS);
- set_extent_bits(&info->block_group_cache,
- start, end, bit_to_set,
- GFP_NOFS);
- } else if (cache->data != data &&
- cache->data != BTRFS_BLOCK_GROUP_MIXED) {
- cache->data = BTRFS_BLOCK_GROUP_MIXED;
- set_extent_bits(&info->block_group_cache,
- start, end,
- BLOCK_GROUP_DATA |
- BLOCK_GROUP_METADATA,
- GFP_NOFS);
- }
old_val += num_bytes;
+ cache->space_info->bytes_used += num_bytes;
} else {
old_val -= num_bytes;
+ cache->space_info->bytes_used -= num_bytes;
if (mark_free) {
set_extent_dirty(&info->free_space_cache,
bytenr, bytenr + num_bytes - 1,
}
return 0;
}
+
static int update_pinned_extents(struct btrfs_root *root,
u64 bytenr, u64 num, int pin)
{
}
while (num > 0) {
cache = btrfs_lookup_block_group(fs_info, bytenr);
+ if (!cache) {
+ len = min((u64)root->sectorsize, num);
+ goto next;
+ }
WARN_ON(!cache);
len = min(num, cache->key.offset -
(bytenr - cache->key.objectid));
if (pin) {
cache->pinned += len;
+ cache->space_info->bytes_pinned += len;
fs_info->total_pinned += len;
} else {
cache->pinned -= len;
+ cache->space_info->bytes_pinned -= len;
fs_info->total_pinned -= len;
}
+next:
bytenr += len;
num -= len;
}
return 0;
}
-int btrfs_copy_pinned(struct btrfs_root *root, struct extent_map_tree *copy)
+int btrfs_copy_pinned(struct btrfs_root *root, struct extent_io_tree *copy)
{
u64 last = 0;
u64 start;
u64 end;
- struct extent_map_tree *pinned_extents = &root->fs_info->pinned_extents;
+ struct extent_io_tree *pinned_extents = &root->fs_info->pinned_extents;
int ret;
while(1) {
int btrfs_finish_extent_commit(struct btrfs_trans_handle *trans,
struct btrfs_root *root,
- struct extent_map_tree *unpin)
+ struct extent_io_tree *unpin)
{
u64 start;
u64 end;
int ret;
- struct extent_map_tree *free_space_cache;
+ struct extent_io_tree *free_space_cache;
free_space_cache = &root->fs_info->free_space_cache;
while(1) {
return 0;
}
+static int extent_root_pending_ops(struct btrfs_fs_info *info)
+{
+ u64 start;
+ u64 end;
+ int ret;
+
+ ret = find_first_extent_bit(&info->extent_ins, 0, &start,
+ &end, EXTENT_LOCKED);
+ if (!ret) {
+ ret = find_first_extent_bit(&info->pending_del, 0, &start, &end,
+ EXTENT_LOCKED);
+ }
+ return ret == 0;
+
+}
static int finish_current_insert(struct btrfs_trans_handle *trans,
struct btrfs_root *extent_root)
{
u64 start;
u64 end;
+ u64 priv;
struct btrfs_fs_info *info = extent_root->fs_info;
- struct extent_buffer *eb;
struct btrfs_path *path;
- struct btrfs_key ins;
- struct btrfs_disk_key first;
- struct btrfs_extent_item extent_item;
+ struct pending_extent_op *extent_op;
+ struct btrfs_key key;
int ret;
- int level;
- int err = 0;
- btrfs_set_stack_extent_refs(&extent_item, 1);
- btrfs_set_key_type(&ins, BTRFS_EXTENT_ITEM_KEY);
path = btrfs_alloc_path();
while(1) {
if (ret)
break;
- ins.objectid = start;
- ins.offset = end + 1 - start;
- err = btrfs_insert_item(trans, extent_root, &ins,
- &extent_item, sizeof(extent_item));
- clear_extent_bits(&info->extent_ins, start, end, EXTENT_LOCKED,
- GFP_NOFS);
- eb = read_tree_block(extent_root, ins.objectid, ins.offset);
- level = btrfs_header_level(eb);
- if (level == 0) {
- btrfs_item_key(eb, &first, 0);
+ ret = get_state_private(&info->extent_ins, start, &priv);
+ BUG_ON(ret);
+ extent_op = (struct pending_extent_op *)(unsigned long)priv;
+
+ if (extent_op->type == PENDING_EXTENT_INSERT) {
+ key.objectid = start;
+ key.offset = end + 1 - start;
+ key.type = BTRFS_EXTENT_ITEM_KEY;
+ ret = alloc_reserved_tree_block(trans, extent_root,
+ extent_root->root_key.objectid,
+ trans->transid,
+ extent_op->flags,
+ &extent_op->key,
+ extent_op->level, &key);
} else {
- btrfs_node_key(eb, &first, 0);
+ BUG_ON(1);
}
- err = btrfs_insert_extent_backref(trans, extent_root, path,
- start, extent_root->root_key.objectid,
- 0, level,
- btrfs_disk_key_objectid(&first));
- BUG_ON(err);
- free_extent_buffer(eb);
+
+ clear_extent_bits(&info->extent_ins, start, end, EXTENT_LOCKED,
+ GFP_NOFS);
+ kfree(extent_op);
}
btrfs_free_path(path);
return 0;
}
-static int pin_down_bytes(struct btrfs_root *root, u64 bytenr, u32 num_bytes,
- int pending)
+static int pin_down_bytes(struct btrfs_trans_handle *trans,
+ struct btrfs_root *root,
+ u64 bytenr, u64 num_bytes, int is_data)
{
int err = 0;
struct extent_buffer *buf;
- if (!pending) {
- buf = btrfs_find_tree_block(root, bytenr, num_bytes);
- if (buf) {
- if (btrfs_buffer_uptodate(buf)) {
- u64 transid =
- root->fs_info->running_transaction->transid;
- if (btrfs_header_generation(buf) == transid) {
- free_extent_buffer(buf);
- return 1;
- }
- }
+ if (is_data)
+ goto pinit;
+
+ buf = btrfs_find_tree_block(root, bytenr, num_bytes);
+ if (!buf)
+ goto pinit;
+
+ /* we can reuse a block if it hasn't been written
+ * and it is from this transaction. We can't
+ * reuse anything from the tree log root because
+ * it has tiny sub-transactions.
+ */
+ if (btrfs_buffer_uptodate(buf, 0)) {
+ u64 header_owner = btrfs_header_owner(buf);
+ u64 header_transid = btrfs_header_generation(buf);
+ if (header_owner != BTRFS_TREE_LOG_OBJECTID &&
+ header_transid == trans->transid &&
+ !btrfs_header_flag(buf, BTRFS_HEADER_FLAG_WRITTEN)) {
+ clean_tree_block(NULL, root, buf);
free_extent_buffer(buf);
+ return 1;
}
- update_pinned_extents(root, bytenr, num_bytes, 1);
- } else {
- set_extent_bits(&root->fs_info->pending_del,
- bytenr, bytenr + num_bytes - 1,
- EXTENT_LOCKED, GFP_NOFS);
}
+ free_extent_buffer(buf);
+pinit:
+ update_pinned_extents(root, bytenr, num_bytes, 1);
+
BUG_ON(err < 0);
return 0;
}
+void btrfs_pin_extent(struct btrfs_fs_info *fs_info,
+ u64 bytenr, u64 num_bytes)
+{
+ update_pinned_extents(fs_info->extent_root, bytenr, num_bytes, 1);
+}
+
/*
* remove an extent from the root, returns 0 on success
*/
-static int __free_extent(struct btrfs_trans_handle *trans, struct btrfs_root
- *root, u64 bytenr, u64 num_bytes,
- u64 root_objectid, u64 ref_generation,
- u64 owner_objectid, u64 owner_offset, int pin,
- int mark_free)
+static int __free_extent(struct btrfs_trans_handle *trans,
+ struct btrfs_root *root,
+ u64 bytenr, u64 num_bytes, u64 parent,
+ u64 root_objectid, u64 owner_objectid,
+ u64 owner_offset, int refs_to_drop)
{
- struct btrfs_path *path;
+
struct btrfs_key key;
- struct btrfs_fs_info *info = root->fs_info;
- struct btrfs_extent_ops *ops = info->extent_ops;
- struct btrfs_root *extent_root = info->extent_root;
+ struct btrfs_path *path;
+ struct btrfs_extent_ops *ops = root->fs_info->extent_ops;
+ struct btrfs_root *extent_root = root->fs_info->extent_root;
struct extent_buffer *leaf;
- int ret;
struct btrfs_extent_item *ei;
- u32 refs;
+ struct btrfs_extent_inline_ref *iref;
+ int ret;
+ int is_data;
+ int extent_slot = 0;
+ int found_extent = 0;
+ int num_to_del = 1;
+ u32 item_size;
+ u64 refs;
- key.objectid = bytenr;
- btrfs_set_key_type(&key, BTRFS_EXTENT_ITEM_KEY);
- key.offset = num_bytes;
+ if (root->fs_info->free_extent_hook) {
+ root->fs_info->free_extent_hook(trans, root, bytenr, num_bytes,
+ parent, root_objectid, owner_objectid,
+ owner_offset, refs_to_drop);
+ }
path = btrfs_alloc_path();
if (!path)
return -ENOMEM;
- ret = lookup_extent_backref(trans, extent_root, path,
- bytenr, root_objectid,
- ref_generation,
- owner_objectid, owner_offset, 1);
+ path->reada = 1;
+ path->leave_spinning = 1;
+
+ is_data = owner_objectid >= BTRFS_FIRST_FREE_OBJECTID;
+ BUG_ON(!is_data && refs_to_drop != 1);
+
+ ret = lookup_extent_backref(trans, extent_root, path, &iref,
+ bytenr, num_bytes, parent,
+ root_objectid, owner_objectid,
+ owner_offset);
if (ret == 0) {
- ret = btrfs_del_item(trans, extent_root, path);
+ extent_slot = path->slots[0];
+ while (extent_slot >= 0) {
+ btrfs_item_key_to_cpu(path->nodes[0], &key,
+ extent_slot);
+ if (key.objectid != bytenr)
+ break;
+ if (key.type == BTRFS_EXTENT_ITEM_KEY &&
+ key.offset == num_bytes) {
+ found_extent = 1;
+ break;
+ }
+ if (path->slots[0] - extent_slot > 5)
+ break;
+ extent_slot--;
+ }
+#ifdef BTRFS_COMPAT_EXTENT_TREE_V0
+ item_size = btrfs_item_size_nr(path->nodes[0], extent_slot);
+ if (found_extent && item_size < sizeof(*ei))
+ found_extent = 0;
+#endif
+ if (!found_extent) {
+ BUG_ON(iref);
+ ret = remove_extent_backref(trans, extent_root, path,
+ NULL, refs_to_drop,
+ is_data);
+ BUG_ON(ret);
+ btrfs_release_path(extent_root, path);
+ path->leave_spinning = 1;
+
+ key.objectid = bytenr;
+ key.type = BTRFS_EXTENT_ITEM_KEY;
+ key.offset = num_bytes;
+
+ ret = btrfs_search_slot(trans, extent_root,
+ &key, path, -1, 1);
+ if (ret) {
+ printk(KERN_ERR "umm, got %d back from search"
+ ", was looking for %llu\n", ret,
+ (unsigned long long)bytenr);
+ btrfs_print_leaf(extent_root, path->nodes[0]);
+ }
+ BUG_ON(ret);
+ extent_slot = path->slots[0];
+ }
} else {
- btrfs_print_leaf(extent_root, path->nodes[0]);
- WARN_ON(1);
- printk("Unable to find ref byte nr %Lu root %Lu "
- " gen %Lu owner %Lu offset %Lu\n", bytenr,
- root_objectid, ref_generation, owner_objectid,
- owner_offset);
+ printk(KERN_ERR "btrfs unable to find ref byte nr %llu "
+ "parent %llu root %llu owner %llu offset %llu\n",
+ (unsigned long long)bytenr,
+ (unsigned long long)parent,
+ (unsigned long long)root_objectid,
+ (unsigned long long)owner_objectid,
+ (unsigned long long)owner_offset);
+ ret = -EIO;
+ goto fail;
}
- btrfs_release_path(extent_root, path);
- ret = btrfs_search_slot(trans, extent_root, &key, path, -1, 1);
- if (ret < 0)
- return ret;
- BUG_ON(ret);
leaf = path->nodes[0];
- ei = btrfs_item_ptr(leaf, path->slots[0],
+ item_size = btrfs_item_size_nr(leaf, extent_slot);
+#ifdef BTRFS_COMPAT_EXTENT_TREE_V0
+ if (item_size < sizeof(*ei)) {
+ BUG_ON(found_extent || extent_slot != path->slots[0]);
+ ret = convert_extent_item_v0(trans, extent_root, path,
+ owner_objectid, 0);
+ BUG_ON(ret < 0);
+
+ btrfs_release_path(extent_root, path);
+ path->leave_spinning = 1;
+
+ key.objectid = bytenr;
+ key.type = BTRFS_EXTENT_ITEM_KEY;
+ key.offset = num_bytes;
+
+ ret = btrfs_search_slot(trans, extent_root, &key, path,
+ -1, 1);
+ if (ret) {
+ printk(KERN_ERR "umm, got %d back from search"
+ ", was looking for %llu\n", ret,
+ (unsigned long long)bytenr);
+ btrfs_print_leaf(extent_root, path->nodes[0]);
+ }
+ BUG_ON(ret);
+ extent_slot = path->slots[0];
+ leaf = path->nodes[0];
+ item_size = btrfs_item_size_nr(leaf, extent_slot);
+ }
+#endif
+ BUG_ON(item_size < sizeof(*ei));
+ ei = btrfs_item_ptr(leaf, extent_slot,
struct btrfs_extent_item);
+ if (owner_objectid < BTRFS_FIRST_FREE_OBJECTID) {
+ struct btrfs_tree_block_info *bi;
+ BUG_ON(item_size < sizeof(*ei) + sizeof(*bi));
+ bi = (struct btrfs_tree_block_info *)(ei + 1);
+ WARN_ON(owner_objectid != btrfs_tree_block_level(leaf, bi));
+ }
+
refs = btrfs_extent_refs(leaf, ei);
- BUG_ON(refs == 0);
- refs -= 1;
- btrfs_set_extent_refs(leaf, ei, refs);
- btrfs_mark_buffer_dirty(leaf);
+ BUG_ON(refs < refs_to_drop);
+ refs -= refs_to_drop;
+
+ if (refs > 0) {
+ /*
+ * In the case of inline back ref, reference count will
+ * be updated by remove_extent_backref
+ */
+ if (iref) {
+ BUG_ON(!found_extent);
+ } else {
+ btrfs_set_extent_refs(leaf, ei, refs);
+ btrfs_mark_buffer_dirty(leaf);
+ }
+ if (found_extent) {
+ ret = remove_extent_backref(trans, extent_root, path,
+ iref, refs_to_drop,
+ is_data);
+ BUG_ON(ret);
+ }
+ } else {
+ int mark_free = 0;
+ int pin = 1;
+
+ if (found_extent) {
+ BUG_ON(is_data && refs_to_drop !=
+ extent_data_ref_count(root, path, iref));
+ if (iref) {
+ BUG_ON(path->slots[0] != extent_slot);
+ } else {
+ BUG_ON(path->slots[0] != extent_slot + 1);
+ path->slots[0] = extent_slot;
+ num_to_del = 2;
+ }
+ }
- if (refs == 0) {
- u64 super_used;
- u64 root_used;
+ if (ops && ops->free_extent) {
+ ret = ops->free_extent(root, bytenr, num_bytes);
+ if (ret > 0) {
+ pin = 0;
+ mark_free = 0;
+ }
+ }
if (pin) {
- ret = pin_down_bytes(root, bytenr, num_bytes, 0);
+ ret = pin_down_bytes(trans, root, bytenr, num_bytes,
+ is_data);
if (ret > 0)
mark_free = 1;
BUG_ON(ret < 0);
}
- /* block accounting for super block */
- super_used = btrfs_super_bytes_used(&info->super_copy);
- btrfs_set_super_bytes_used(&info->super_copy,
- super_used - num_bytes);
-
- /* block accounting for root item */
- root_used = btrfs_root_used(&root->root_item);
- btrfs_set_root_used(&root->root_item,
- root_used - num_bytes);
- ret = btrfs_del_item(trans, extent_root, path);
- if (ret)
- return ret;
+ ret = btrfs_del_items(trans, extent_root, path, path->slots[0],
+ num_to_del);
+ BUG_ON(ret);
+ btrfs_release_path(extent_root, path);
- if (ops && ops->free_extent)
- ops->free_extent(root, bytenr, num_bytes);
+ if (is_data) {
+ ret = btrfs_del_csums(trans, root, bytenr, num_bytes);
+ BUG_ON(ret);
+ }
- ret = update_block_group(trans, root, bytenr, num_bytes, 0,
- mark_free, 0);
- BUG_ON(ret);
+ update_block_group(trans, root, bytenr, num_bytes, 0, mark_free);
}
+fail:
btrfs_free_path(path);
finish_current_insert(trans, extent_root);
return ret;
int err = 0;
u64 start;
u64 end;
- struct extent_map_tree *pending_del;
- struct extent_map_tree *pinned_extents;
+ u64 priv;
+ struct extent_io_tree *pending_del;
+ struct extent_io_tree *extent_ins;
+ struct pending_extent_op *extent_op;
+ extent_ins = &extent_root->fs_info->extent_ins;
pending_del = &extent_root->fs_info->pending_del;
- pinned_extents = &extent_root->fs_info->pinned_extents;
while(1) {
ret = find_first_extent_bit(pending_del, 0, &start, &end,
EXTENT_LOCKED);
if (ret)
break;
- update_pinned_extents(extent_root, start, end + 1 - start, 1);
+
+ ret = get_state_private(pending_del, start, &priv);
+ BUG_ON(ret);
+ extent_op = (struct pending_extent_op *)(unsigned long)priv;
+
clear_extent_bits(pending_del, start, end, EXTENT_LOCKED,
GFP_NOFS);
- ret = __free_extent(trans, extent_root,
- start, end + 1 - start,
- extent_root->root_key.objectid,
- 0, 0, 0, 0, 0);
+
+ if (!test_range_bit(extent_ins, start, end,
+ EXTENT_LOCKED, 0)) {
+ ret = __free_extent(trans, extent_root,
+ start, end + 1 - start, 0,
+ extent_root->root_key.objectid,
+ extent_op->level, 0, 1);
+ kfree(extent_op);
+ } else {
+ kfree(extent_op);
+ ret = get_state_private(extent_ins, start, &priv);
+ BUG_ON(ret);
+ extent_op = (struct pending_extent_op *)
+ (unsigned long)priv;
+
+ clear_extent_bits(extent_ins, start, end,
+ EXTENT_LOCKED, GFP_NOFS);
+
+ if (extent_op->type == PENDING_BACKREF_UPDATE)
+ BUG_ON(1);
+
+ kfree(extent_op);
+ }
if (ret)
err = ret;
}
/*
* remove an extent from the root, returns 0 on success
*/
-int btrfs_free_extent(struct btrfs_trans_handle *trans, struct btrfs_root
- *root, u64 bytenr, u64 num_bytes,
- u64 root_objectid, u64 ref_generation,
- u64 owner_objectid, u64 owner_offset, int pin)
+
+int btrfs_free_extent(struct btrfs_trans_handle *trans,
+ struct btrfs_root *root,
+ u64 bytenr, u64 num_bytes, u64 parent,
+ u64 root_objectid, u64 owner, u64 offset)
{
struct btrfs_root *extent_root = root->fs_info->extent_root;
int pending_ret;
int ret;
WARN_ON(num_bytes < root->sectorsize);
- if (!root->ref_cows)
- ref_generation = 0;
-
if (root == extent_root) {
- pin_down_bytes(root, bytenr, num_bytes, 1);
+ struct pending_extent_op *extent_op;
+
+ extent_op = kmalloc(sizeof(*extent_op), GFP_NOFS);
+ BUG_ON(!extent_op);
+
+ extent_op->type = PENDING_EXTENT_DELETE;
+ extent_op->bytenr = bytenr;
+ extent_op->num_bytes = num_bytes;
+ extent_op->level = (int)owner;
+
+ set_extent_bits(&root->fs_info->pending_del,
+ bytenr, bytenr + num_bytes - 1,
+ EXTENT_LOCKED, GFP_NOFS);
+ set_state_private(&root->fs_info->pending_del,
+ bytenr, (unsigned long)extent_op);
return 0;
}
- ret = __free_extent(trans, root, bytenr, num_bytes, root_objectid,
- ref_generation, owner_objectid, owner_offset,
- pin, pin == 0);
+ ret = __free_extent(trans, root, bytenr, num_bytes, parent,
+ root_objectid, owner, offset, 1);
pending_ret = del_pending_extents(trans, root->fs_info->extent_root);
return ret ? ret : pending_ret;
}
u64 exclude_start, u64 exclude_nr,
int data)
{
- struct btrfs_path *path;
- struct btrfs_key key;
- u64 hole_size = 0;
- u64 aligned;
int ret;
- int slot = 0;
- u64 last_byte = 0;
u64 orig_search_start = search_start;
- int start_found;
- struct extent_buffer *l;
struct btrfs_root * root = orig_root->fs_info->extent_root;
struct btrfs_fs_info *info = root->fs_info;
u64 total_needed = num_bytes;
- int level;
struct btrfs_block_group_cache *block_group;
int full_scan = 0;
int wrapped = 0;
- u64 cached_start;
WARN_ON(num_bytes < root->sectorsize);
btrfs_set_key_type(ins, BTRFS_EXTENT_ITEM_KEY);
- level = btrfs_header_level(root->node);
-
- if (num_bytes >= 32 * 1024 * 1024 && hint_byte) {
- data = BTRFS_BLOCK_GROUP_MIXED;
- }
+ search_start = stripe_align(root, search_start);
- if (search_end == (u64)-1)
- search_end = btrfs_super_total_bytes(&info->super_copy);
if (hint_byte) {
- block_group = btrfs_lookup_block_group(info, hint_byte);
+ block_group = btrfs_lookup_first_block_group(info, hint_byte);
if (!block_group)
hint_byte = search_start;
block_group = btrfs_find_block_group(root, block_group,
}
total_needed += empty_size;
- path = btrfs_alloc_path();
+
check_failed:
+ search_start = stripe_align(root, search_start);
if (!block_group) {
- block_group = btrfs_lookup_block_group(info, search_start);
+ block_group = btrfs_lookup_first_block_group(info,
+ search_start);
if (!block_group)
- block_group = btrfs_lookup_block_group(info,
+ block_group = btrfs_lookup_first_block_group(info,
orig_search_start);
}
- search_start = find_search_start(root, &block_group, search_start,
- total_needed, data, full_scan);
- search_start = stripe_align(root, search_start);
- cached_start = search_start;
- btrfs_init_path(path);
- ins->objectid = search_start;
- ins->offset = 0;
- start_found = 0;
- path->reada = 2;
-
- ret = btrfs_search_slot(trans, root, ins, path, 0, 0);
- if (ret < 0)
+ ret = find_search_start(root, &block_group, &search_start,
+ total_needed, data);
+ if (ret)
goto error;
- if (path->slots[0] > 0) {
- path->slots[0]--;
- }
-
- l = path->nodes[0];
- btrfs_item_key_to_cpu(l, &key, path->slots[0]);
-
- /*
- * walk backwards to find the first extent item key
- */
- while(btrfs_key_type(&key) != BTRFS_EXTENT_ITEM_KEY) {
- if (path->slots[0] == 0) {
- ret = btrfs_prev_leaf(root, path);
- if (ret != 0) {
- ret = btrfs_search_slot(trans, root, ins,
- path, 0, 0);
- if (ret < 0)
- goto error;
- if (path->slots[0] > 0)
- path->slots[0]--;
- break;
- }
- } else {
- path->slots[0]--;
- }
- l = path->nodes[0];
- btrfs_item_key_to_cpu(l, &key, path->slots[0]);
- }
- while (1) {
- l = path->nodes[0];
- slot = path->slots[0];
- if (slot >= btrfs_header_nritems(l)) {
- ret = btrfs_next_leaf(root, path);
- if (ret == 0)
- continue;
- if (ret < 0)
- goto error;
-
- search_start = max(search_start,
- block_group->key.objectid);
- if (!start_found) {
- aligned = stripe_align(root, search_start);
- ins->objectid = aligned;
- if (aligned >= search_end) {
- ret = -ENOSPC;
- goto error;
- }
- ins->offset = search_end - aligned;
- start_found = 1;
- goto check_pending;
- }
- ins->objectid = stripe_align(root,
- last_byte > search_start ?
- last_byte : search_start);
- if (search_end <= ins->objectid) {
- ret = -ENOSPC;
- goto error;
- }
- ins->offset = search_end - ins->objectid;
- BUG_ON(ins->objectid >= search_end);
- goto check_pending;
- }
- btrfs_item_key_to_cpu(l, &key, slot);
-
- if (key.objectid >= search_start && key.objectid > last_byte &&
- start_found) {
- if (last_byte < search_start)
- last_byte = search_start;
- aligned = stripe_align(root, last_byte);
- hole_size = key.objectid - aligned;
- if (key.objectid > aligned && hole_size >= num_bytes) {
- ins->objectid = aligned;
- ins->offset = hole_size;
- goto check_pending;
- }
- }
- if (btrfs_key_type(&key) != BTRFS_EXTENT_ITEM_KEY) {
- if (!start_found && btrfs_key_type(&key) ==
- BTRFS_BLOCK_GROUP_ITEM_KEY) {
- last_byte = key.objectid;
- start_found = 1;
- }
- goto next;
- }
-
-
- start_found = 1;
- last_byte = key.objectid + key.offset;
-
- if (!full_scan && data != BTRFS_BLOCK_GROUP_MIXED &&
- last_byte >= block_group->key.objectid +
- block_group->key.offset) {
- btrfs_release_path(root, path);
- search_start = block_group->key.objectid +
- block_group->key.offset;
- goto new_group;
- }
-next:
- path->slots[0]++;
- cond_resched();
- }
-check_pending:
- /* we have to make sure we didn't find an extent that has already
- * been allocated by the map tree or the original allocation
- */
- btrfs_release_path(root, path);
- BUG_ON(ins->objectid < search_start);
+ ins->objectid = search_start;
+ ins->offset = num_bytes;
- if (ins->objectid + num_bytes >= search_end)
- goto enospc;
- if (!full_scan && data != BTRFS_BLOCK_GROUP_MIXED &&
- ins->objectid + num_bytes > block_group->
- key.objectid + block_group->key.offset) {
+ if (ins->objectid + num_bytes >
+ block_group->key.objectid + block_group->key.offset) {
search_start = block_group->key.objectid +
block_group->key.offset;
goto new_group;
}
+
if (test_range_bit(&info->extent_ins, ins->objectid,
ins->objectid + num_bytes -1, EXTENT_LOCKED, 0)) {
search_start = ins->objectid + num_bytes;
goto new_group;
}
+
if (test_range_bit(&info->pinned_extents, ins->objectid,
ins->objectid + num_bytes -1, EXTENT_DIRTY, 0)) {
search_start = ins->objectid + num_bytes;
goto new_group;
}
+
if (exclude_nr > 0 && (ins->objectid + num_bytes > exclude_start &&
ins->objectid < exclude_start + exclude_nr)) {
search_start = exclude_start + exclude_nr;
goto new_group;
}
- if (!data) {
+
+ if (!(data & BTRFS_BLOCK_GROUP_DATA)) {
block_group = btrfs_lookup_block_group(info, ins->objectid);
if (block_group)
trans->block_group = block_group;
}
ins->offset = num_bytes;
- btrfs_free_path(path);
return 0;
new_group:
- if (search_start + num_bytes >= search_end) {
-enospc:
+ block_group = btrfs_lookup_first_block_group(info, search_start);
+ if (!block_group) {
search_start = orig_search_start;
if (full_scan) {
ret = -ENOSPC;
if (!full_scan)
total_needed -= empty_size;
full_scan = 1;
- data = BTRFS_BLOCK_GROUP_MIXED;
} else
wrapped = 1;
}
- block_group = btrfs_lookup_block_group(info, search_start);
cond_resched();
block_group = btrfs_find_block_group(root, block_group,
search_start, data, 0);
goto check_failed;
error:
- btrfs_release_path(root, path);
- btrfs_free_path(path);
return ret;
}
-/*
- * finds a free extent and does all the dirty work required for allocation
- * returns the key for the extent through ins, and a tree buffer for
- * the first block of the extent through buf.
- *
- * returns 0 if everything worked, non-zero otherwise.
- */
-int btrfs_alloc_extent(struct btrfs_trans_handle *trans,
- struct btrfs_root *root,
- u64 num_bytes, u64 root_objectid, u64 ref_generation,
- u64 owner, u64 owner_offset,
- u64 empty_size, u64 hint_byte,
- u64 search_end, struct btrfs_key *ins, int data)
+
+static int btrfs_reserve_extent(struct btrfs_trans_handle *trans,
+ struct btrfs_root *root,
+ u64 num_bytes, u64 empty_size,
+ u64 hint_byte, u64 search_end,
+ struct btrfs_key *ins, int data)
{
int ret;
- int pending_ret;
- u64 super_used, root_used;
u64 search_start = 0;
- /*
- u64 new_hint;
- */
+ u64 alloc_profile;
struct btrfs_fs_info *info = root->fs_info;
- struct btrfs_extent_ops *ops = info->extent_ops;
- struct btrfs_root *extent_root = info->extent_root;
- struct btrfs_extent_item extent_item;
- struct btrfs_path *path;
-
- btrfs_set_stack_extent_refs(&extent_item, 1);
- /*
- new_hint = max(hint_byte, root->fs_info->alloc_start);
- if (new_hint < btrfs_super_total_bytes(&info->super_copy))
- hint_byte = new_hint;
- */
- WARN_ON(num_bytes < root->sectorsize);
- if (ops && ops->alloc_extent) {
+ if (info->extent_ops) {
+ struct btrfs_extent_ops *ops = info->extent_ops;
ret = ops->alloc_extent(root, num_bytes, hint_byte, ins);
+ BUG_ON(ret);
+ goto found;
+ }
+
+ if (data) {
+ alloc_profile = info->avail_data_alloc_bits &
+ info->data_alloc_profile;
+ data = BTRFS_BLOCK_GROUP_DATA | alloc_profile;
+ } else if ((info->system_allocs > 0 || root == info->chunk_root) &&
+ info->system_allocs >= 0) {
+ alloc_profile = info->avail_system_alloc_bits &
+ info->system_alloc_profile;
+ data = BTRFS_BLOCK_GROUP_SYSTEM | alloc_profile;
} else {
- ret = find_free_extent(trans, root, num_bytes, empty_size,
- search_start, search_end, hint_byte,
- ins, trans->alloc_exclude_start,
- trans->alloc_exclude_nr, data);
+ alloc_profile = info->avail_metadata_alloc_bits &
+ info->metadata_alloc_profile;
+ data = BTRFS_BLOCK_GROUP_METADATA | alloc_profile;
}
- BUG_ON(ret);
- if (ret)
- return ret;
-
- /* block accounting for super block */
- super_used = btrfs_super_bytes_used(&info->super_copy);
- btrfs_set_super_bytes_used(&info->super_copy, super_used + num_bytes);
- /* block accounting for root item */
- root_used = btrfs_root_used(&root->root_item);
- btrfs_set_root_used(&root->root_item, root_used + num_bytes);
+ if (root->ref_cows) {
+ if (!(data & BTRFS_BLOCK_GROUP_METADATA)) {
+ ret = do_chunk_alloc(trans, root->fs_info->extent_root,
+ num_bytes,
+ BTRFS_BLOCK_GROUP_METADATA);
+ BUG_ON(ret);
+ }
+ ret = do_chunk_alloc(trans, root->fs_info->extent_root,
+ num_bytes + 2 * 1024 * 1024, data);
+ BUG_ON(ret);
+ }
+ WARN_ON(num_bytes < root->sectorsize);
+ ret = find_free_extent(trans, root, num_bytes, empty_size,
+ search_start, search_end, hint_byte, ins,
+ trans->alloc_exclude_start,
+ trans->alloc_exclude_nr, data);
+ BUG_ON(ret);
+found:
clear_extent_dirty(&root->fs_info->free_space_cache,
ins->objectid, ins->objectid + ins->offset - 1,
GFP_NOFS);
+ return ret;
+}
- if (root == extent_root) {
- set_extent_bits(&root->fs_info->extent_ins, ins->objectid,
- ins->objectid + ins->offset - 1,
- EXTENT_LOCKED, GFP_NOFS);
- WARN_ON(data == 1);
- goto update_block;
- }
-
- WARN_ON(trans->alloc_exclude_nr);
- trans->alloc_exclude_start = ins->objectid;
- trans->alloc_exclude_nr = ins->offset;
- ret = btrfs_insert_item(trans, extent_root, ins, &extent_item,
- sizeof(extent_item));
-
- trans->alloc_exclude_start = 0;
- trans->alloc_exclude_nr = 0;
- BUG_ON(ret);
+static int alloc_reserved_tree_block(struct btrfs_trans_handle *trans,
+ struct btrfs_root *root,
+ u64 root_objectid, u64 generation,
+ u64 flags, struct btrfs_disk_key *key,
+ int level, struct btrfs_key *ins)
+{
+ int ret;
+ struct btrfs_fs_info *fs_info = root->fs_info;
+ struct btrfs_extent_item *extent_item;
+ struct btrfs_tree_block_info *block_info;
+ struct btrfs_extent_inline_ref *iref;
+ struct btrfs_path *path;
+ struct extent_buffer *leaf;
+ u32 size = sizeof(*extent_item) + sizeof(*block_info) + sizeof(*iref);
path = btrfs_alloc_path();
BUG_ON(!path);
- ret = btrfs_insert_extent_backref(trans, extent_root, path,
- ins->objectid, root_objectid,
- ref_generation, owner, owner_offset);
+ path->leave_spinning = 1;
+ ret = btrfs_insert_empty_item(trans, fs_info->extent_root, path,
+ ins, size);
BUG_ON(ret);
- btrfs_free_path(path);
- finish_current_insert(trans, extent_root);
- pending_ret = del_pending_extents(trans, extent_root);
- if (ret) {
- return ret;
- }
- if (pending_ret) {
- return pending_ret;
- }
+ leaf = path->nodes[0];
+ extent_item = btrfs_item_ptr(leaf, path->slots[0],
+ struct btrfs_extent_item);
+ btrfs_set_extent_refs(leaf, extent_item, 1);
+ btrfs_set_extent_generation(leaf, extent_item, generation);
+ btrfs_set_extent_flags(leaf, extent_item,
+ flags | BTRFS_EXTENT_FLAG_TREE_BLOCK);
+ block_info = (struct btrfs_tree_block_info *)(extent_item + 1);
-update_block:
- ret = update_block_group(trans, root, ins->objectid, ins->offset, 1, 0,
- data);
- BUG_ON(ret);
+ btrfs_set_tree_block_key(leaf, block_info, key);
+ btrfs_set_tree_block_level(leaf, block_info, level);
+
+ iref = (struct btrfs_extent_inline_ref *)(block_info + 1);
+ btrfs_set_extent_inline_ref_type(leaf, iref, BTRFS_TREE_BLOCK_REF_KEY);
+ btrfs_set_extent_inline_ref_offset(leaf, iref, root_objectid);
+
+ btrfs_mark_buffer_dirty(leaf);
+ btrfs_free_path(path);
+
+ ret = update_block_group(trans, root, ins->objectid, ins->offset,
+ 1, 0);
return 0;
}
-/*
- * helper function to allocate a block for a given tree
- * returns the tree buffer or NULL.
- */
-struct extent_buffer *btrfs_alloc_free_block(struct btrfs_trans_handle *trans,
- struct btrfs_root *root,
- u32 blocksize,
- u64 root_objectid, u64 hint,
- u64 empty_size)
+static int alloc_tree_block(struct btrfs_trans_handle *trans,
+ struct btrfs_root *root, u64 num_bytes,
+ u64 root_objectid, u64 generation,
+ u64 flags, struct btrfs_disk_key *key,
+ int level, u64 empty_size, u64 hint_byte,
+ u64 search_end, struct btrfs_key *ins)
{
- u64 ref_generation;
+ int ret;
+ ret = btrfs_reserve_extent(trans, root, num_bytes, empty_size,
+ hint_byte, search_end, ins, 0);
+ BUG_ON(ret);
- if (root->ref_cows)
- ref_generation = trans->transid;
- else
- ref_generation = 0;
+ if (root_objectid == BTRFS_EXTENT_TREE_OBJECTID) {
+ struct pending_extent_op *extent_op;
+ extent_op = kmalloc(sizeof(*extent_op), GFP_NOFS);
+ BUG_ON(!extent_op);
- return __btrfs_alloc_free_block(trans, root, blocksize, root_objectid,
- ref_generation, 0, 0, hint, empty_size);
+ extent_op->type = PENDING_EXTENT_INSERT;
+ extent_op->bytenr = ins->objectid;
+ extent_op->num_bytes = ins->offset;
+ extent_op->level = level;
+ extent_op->flags = flags;
+ memcpy(&extent_op->key, key, sizeof(*key));
+
+ set_extent_bits(&root->fs_info->extent_ins, ins->objectid,
+ ins->objectid + ins->offset - 1,
+ EXTENT_LOCKED, GFP_NOFS);
+ set_state_private(&root->fs_info->extent_ins,
+ ins->objectid, (unsigned long)extent_op);
+ } else {
+ ret = alloc_reserved_tree_block(trans, root, root_objectid,
+ generation, flags,
+ key, level, ins);
+ finish_current_insert(trans, root->fs_info->extent_root);
+ del_pending_extents(trans, root->fs_info->extent_root);
+ }
+ return ret;
}
/*
* helper function to allocate a block for a given tree
* returns the tree buffer or NULL.
*/
-struct extent_buffer *__btrfs_alloc_free_block(struct btrfs_trans_handle *trans,
- struct btrfs_root *root,
- u32 blocksize,
- u64 root_objectid,
- u64 ref_generation,
- u64 first_objectid,
- int level,
- u64 hint,
- u64 empty_size)
+struct extent_buffer *btrfs_alloc_free_block(struct btrfs_trans_handle *trans,
+ struct btrfs_root *root,
+ u32 blocksize, u64 root_objectid,
+ struct btrfs_disk_key *key, int level,
+ u64 hint, u64 empty_size)
{
struct btrfs_key ins;
int ret;
struct extent_buffer *buf;
- ret = btrfs_alloc_extent(trans, root, blocksize,
- root_objectid, ref_generation,
- level, first_objectid, empty_size, hint,
- (u64)-1, &ins, 0);
+ ret = alloc_tree_block(trans, root, blocksize, root_objectid,
+ trans->transid, 0, key, level,
+ empty_size, hint, (u64)-1, &ins);
if (ret) {
BUG_ON(ret > 0);
return ERR_PTR(ret);
}
+
buf = btrfs_find_create_tree_block(root, ins.objectid, blocksize);
if (!buf) {
- btrfs_free_extent(trans, root, ins.objectid, blocksize,
- root->root_key.objectid, ref_generation,
- 0, 0, 0);
+ btrfs_free_extent(trans, root, ins.objectid, ins.offset,
+ 0, root->root_key.objectid, level, 0);
+ BUG_ON(1);
return ERR_PTR(-ENOMEM);
}
btrfs_set_buffer_uptodate(buf);
- /*
- set_extent_dirty(&trans->transaction->dirty_pages, buf->start,
- buf->start + buf->len - 1, GFP_NOFS);
- set_extent_bits(&BTRFS_I(root->fs_info->btree_inode)->extent_tree,
- buf->start, buf->start + buf->len - 1,
- EXTENT_CSUM, GFP_NOFS);
- buf->flags |= EXTENT_CSUM;
- btrfs_set_buffer_defrag(buf);
- */
trans->blocks_used++;
+
return buf;
}
+#if 0
+
static int noinline drop_leaf_ref(struct btrfs_trans_handle *trans,
struct btrfs_root *root,
struct extent_buffer *leaf)
continue;
ret = btrfs_free_extent(trans, root, disk_bytenr,
btrfs_file_extent_disk_num_bytes(leaf, fi),
- leaf_owner, leaf_generation,
- key.objectid, key.offset, 0);
+ leaf->start, leaf_owner, leaf_generation,
+ key.objectid, 0);
BUG_ON(ret);
}
return 0;
}
static void noinline reada_walk_down(struct btrfs_root *root,
- struct extent_buffer *node)
+ struct extent_buffer *node,
+ int slot)
{
- int i;
- u32 nritems;
u64 bytenr;
- int ret;
+ u64 last = 0;
+ u32 nritems;
u32 refs;
- int level;
u32 blocksize;
+ int ret;
+ int i;
+ int level;
+ int skipped = 0;
nritems = btrfs_header_nritems(node);
level = btrfs_header_level(node);
- for (i = 0; i < nritems; i++) {
+ if (level)
+ return;
+
+ for (i = slot; i < nritems && skipped < 32; i++) {
bytenr = btrfs_node_blockptr(node, i);
- blocksize = btrfs_level_size(root, level - 1);
- ret = lookup_extent_ref(NULL, root, bytenr, blocksize, &refs);
- BUG_ON(ret);
- if (refs != 1)
+ if (last && ((bytenr > last && bytenr - last > 32 * 1024) ||
+ (last > bytenr && last - bytenr > 32 * 1024))) {
+ skipped++;
continue;
+ }
+ blocksize = btrfs_level_size(root, level - 1);
+ if (i != slot) {
+ ret = btrfs_lookup_extent_ref(NULL, root, bytenr,
+ blocksize, &refs);
+ BUG_ON(ret);
+ if (refs != 1) {
+ skipped++;
+ continue;
+ }
+ }
mutex_unlock(&root->fs_info->fs_mutex);
- ret = readahead_tree_block(root, bytenr, blocksize);
+ ret = readahead_tree_block(root, bytenr, blocksize,
+ btrfs_node_ptr_generation(node, i));
+ last = bytenr + blocksize;
cond_resched();
mutex_lock(&root->fs_info->fs_mutex);
if (ret)
u64 root_owner;
u64 root_gen;
u64 bytenr;
+ u64 ptr_gen;
struct extent_buffer *next;
struct extent_buffer *cur;
struct extent_buffer *parent;
WARN_ON(*level < 0);
WARN_ON(*level >= BTRFS_MAX_LEVEL);
- ret = lookup_extent_ref(trans, root,
- path->nodes[*level]->start,
- path->nodes[*level]->len, &refs);
+ ret = btrfs_lookup_extent_ref(trans, root,
+ path->nodes[*level]->start,
+ path->nodes[*level]->len, &refs);
BUG_ON(ret);
if (refs > 1)
goto out;
WARN_ON(*level >= BTRFS_MAX_LEVEL);
cur = path->nodes[*level];
- if (*level > 0 && path->slots[*level] == 0)
- reada_walk_down(root, cur);
-
if (btrfs_header_level(cur) != *level)
WARN_ON(1);
break;
}
bytenr = btrfs_node_blockptr(cur, path->slots[*level]);
+ ptr_gen = btrfs_node_ptr_generation(cur, path->slots[*level]);
blocksize = btrfs_level_size(root, *level - 1);
- ret = lookup_extent_ref(trans, root, bytenr, blocksize, &refs);
+ ret = btrfs_lookup_extent_ref(trans, root, bytenr, blocksize,
+ &refs);
BUG_ON(ret);
if (refs != 1) {
parent = path->nodes[*level];
root_owner = btrfs_header_owner(parent);
root_gen = btrfs_header_generation(parent);
path->slots[*level]++;
- ret = btrfs_free_extent(trans, root, bytenr,
- blocksize, root_owner,
- root_gen, 0, 0, 1);
+ ret = btrfs_free_extent(trans, root, bytenr, blocksize,
+ parent->start, root_owner,
+ root_gen, *level - 1, 1);
BUG_ON(ret);
continue;
}
next = btrfs_find_tree_block(root, bytenr, blocksize);
- if (!next || !btrfs_buffer_uptodate(next)) {
+ if (!next || !btrfs_buffer_uptodate(next, ptr_gen)) {
free_extent_buffer(next);
+ reada_walk_down(root, cur, path->slots[*level]);
mutex_unlock(&root->fs_info->fs_mutex);
- next = read_tree_block(root, bytenr, blocksize);
+ next = read_tree_block(root, bytenr, blocksize,
+ ptr_gen);
mutex_lock(&root->fs_info->fs_mutex);
-
- /* we dropped the lock, check one more time */
- ret = lookup_extent_ref(trans, root, bytenr,
- blocksize, &refs);
- BUG_ON(ret);
- if (refs != 1) {
- parent = path->nodes[*level];
- root_owner = btrfs_header_owner(parent);
- root_gen = btrfs_header_generation(parent);
-
- path->slots[*level]++;
- free_extent_buffer(next);
- ret = btrfs_free_extent(trans, root, bytenr,
- blocksize,
- root_owner,
- root_gen, 0, 0, 1);
- BUG_ON(ret);
- continue;
- }
}
WARN_ON(*level <= 0);
if (path->nodes[*level-1])
root_gen = btrfs_header_generation(parent);
ret = btrfs_free_extent(trans, root, path->nodes[*level]->start,
- path->nodes[*level]->len,
- root_owner, root_gen, 0, 0, 1);
+ path->nodes[*level]->len, parent->start,
+ root_owner, root_gen, *level, 1);
free_extent_buffer(path->nodes[*level]);
path->nodes[*level] = NULL;
*level += 1;
root_item->drop_level = i;
return 0;
} else {
- if (path->nodes[*level] == root->node) {
- root_owner = root->root_key.objectid;
- root_gen =
- btrfs_header_generation(path->nodes[*level]);
- } else {
- struct extent_buffer *node;
- node = path->nodes[*level + 1];
- root_owner = btrfs_header_owner(node);
- root_gen = btrfs_header_generation(node);
- }
+ struct extent_buffer *parent;
+ if (path->nodes[*level] == root->node)
+ parent = path->nodes[*level];
+ else
+ parent = path->nodes[*level + 1];
+
+ root_owner = btrfs_header_owner(parent);
+ root_gen = btrfs_header_generation(parent);
ret = btrfs_free_extent(trans, root,
path->nodes[*level]->start,
path->nodes[*level]->len,
- root_owner, root_gen, 0, 0, 1);
+ parent->start, root_owner,
+ root_gen, *level, 1);
BUG_ON(ret);
free_extent_buffer(path->nodes[*level]);
path->nodes[*level] = NULL;
return ret;
}
+#endif
+
int btrfs_free_block_groups(struct btrfs_fs_info *info)
{
+ struct btrfs_space_info *sinfo;
u64 start;
u64 end;
u64 ptr;
int ret;
+
while(1) {
ret = find_first_extent_bit(&info->block_group_cache, 0,
&start, &end, (unsigned int)-1);
clear_extent_dirty(&info->free_space_cache, start,
end, GFP_NOFS);
}
+
+ while (!list_empty(&info->space_info)) {
+ sinfo = list_entry(info->space_info.next,
+ struct btrfs_space_info, list);
+ list_del_init(&sinfo->list);
+ kfree(sinfo);
+ }
return 0;
}
+int find_first_block_group(struct btrfs_root *root, struct btrfs_path *path,
+ struct btrfs_key *key)
+{
+ int ret;
+ struct btrfs_key found_key;
+ struct extent_buffer *leaf;
+ int slot;
+
+ ret = btrfs_search_slot(NULL, root, key, path, 0, 0);
+ if (ret < 0)
+ return ret;
+ while(1) {
+ slot = path->slots[0];
+ leaf = path->nodes[0];
+ if (slot >= btrfs_header_nritems(leaf)) {
+ ret = btrfs_next_leaf(root, path);
+ if (ret == 0)
+ continue;
+ if (ret < 0)
+ goto error;
+ break;
+ }
+ btrfs_item_key_to_cpu(leaf, &found_key, slot);
+
+ if (found_key.objectid >= key->objectid &&
+ found_key.type == BTRFS_BLOCK_GROUP_ITEM_KEY)
+ return 0;
+ path->slots[0]++;
+ }
+ ret = -ENOENT;
+error:
+ return ret;
+}
+
int btrfs_read_block_groups(struct btrfs_root *root)
{
struct btrfs_path *path;
int ret;
- int err = 0;
int bit;
struct btrfs_block_group_cache *cache;
struct btrfs_fs_info *info = root->fs_info;
- struct extent_map_tree *block_group_cache;
+ struct btrfs_space_info *space_info;
+ struct extent_io_tree *block_group_cache;
struct btrfs_key key;
struct btrfs_key found_key;
struct extent_buffer *leaf;
root = info->extent_root;
key.objectid = 0;
- key.offset = BTRFS_BLOCK_GROUP_SIZE;
+ key.offset = 0;
btrfs_set_key_type(&key, BTRFS_BLOCK_GROUP_ITEM_KEY);
-
path = btrfs_alloc_path();
if (!path)
return -ENOMEM;
while(1) {
- ret = btrfs_search_slot(NULL, info->extent_root,
- &key, path, 0, 0);
+ ret = find_first_block_group(root, path, &key);
+ if (ret > 0) {
+ ret = 0;
+ goto error;
+ }
if (ret != 0) {
- err = ret;
- break;
+ goto error;
}
leaf = path->nodes[0];
btrfs_item_key_to_cpu(leaf, &found_key, path->slots[0]);
- cache = kmalloc(sizeof(*cache), GFP_NOFS);
+ cache = kzalloc(sizeof(*cache), GFP_NOFS);
if (!cache) {
- err = -1;
+ ret = -ENOMEM;
break;
}
cache->pinned = 0;
key.objectid = found_key.objectid + found_key.offset;
btrfs_release_path(root, path);
-
- if (cache->item.flags & BTRFS_BLOCK_GROUP_MIXED) {
- bit = BLOCK_GROUP_DATA | BLOCK_GROUP_METADATA;
- cache->data = BTRFS_BLOCK_GROUP_MIXED;
- } else if (cache->item.flags & BTRFS_BLOCK_GROUP_DATA) {
+ cache->flags = btrfs_block_group_flags(&cache->item);
+ bit = 0;
+ if (cache->flags & BTRFS_BLOCK_GROUP_DATA) {
bit = BLOCK_GROUP_DATA;
- cache->data = BTRFS_BLOCK_GROUP_DATA;
- } else {
+ } else if (cache->flags & BTRFS_BLOCK_GROUP_SYSTEM) {
+ bit = BLOCK_GROUP_SYSTEM;
+ } else if (cache->flags & BTRFS_BLOCK_GROUP_METADATA) {
bit = BLOCK_GROUP_METADATA;
- cache->data = 0;
}
+ set_avail_alloc_bits(info, cache->flags);
+ if (btrfs_chunk_readonly(root, cache->key.objectid))
+ cache->ro = 1;
+
+ ret = update_space_info(info, cache->flags, found_key.offset,
+ btrfs_block_group_used(&cache->item),
+ &space_info);
+ BUG_ON(ret);
+ cache->space_info = space_info;
/* use EXTENT_LOCKED to prevent merging */
set_extent_bits(block_group_cache, found_key.objectid,
bit | EXTENT_LOCKED, GFP_NOFS);
set_state_private(block_group_cache, found_key.objectid,
(unsigned long)cache);
-
- if (key.objectid >=
- btrfs_super_total_bytes(&info->super_copy))
- break;
}
-
+ ret = 0;
+error:
btrfs_free_path(path);
- return 0;
+ return ret;
}
-static int btrfs_insert_block_group(struct btrfs_trans_handle *trans,
- struct btrfs_root *root,
- struct btrfs_key *key,
- struct btrfs_block_group_item *bi)
+int btrfs_make_block_group(struct btrfs_trans_handle *trans,
+ struct btrfs_root *root, u64 bytes_used,
+ u64 type, u64 chunk_objectid, u64 chunk_offset,
+ u64 size)
{
int ret;
- int pending_ret;
+ int bit = 0;
struct btrfs_root *extent_root;
+ struct btrfs_block_group_cache *cache;
+ struct extent_io_tree *block_group_cache;
extent_root = root->fs_info->extent_root;
- ret = btrfs_insert_item(trans, extent_root, key, bi, sizeof(*bi));
+ block_group_cache = &root->fs_info->block_group_cache;
+
+ cache = kzalloc(sizeof(*cache), GFP_NOFS);
+ BUG_ON(!cache);
+ cache->key.objectid = chunk_offset;
+ cache->key.offset = size;
+
+ btrfs_set_key_type(&cache->key, BTRFS_BLOCK_GROUP_ITEM_KEY);
+ btrfs_set_block_group_used(&cache->item, bytes_used);
+ btrfs_set_block_group_chunk_objectid(&cache->item, chunk_objectid);
+ cache->flags = type;
+ btrfs_set_block_group_flags(&cache->item, type);
+
+ ret = update_space_info(root->fs_info, cache->flags, size, bytes_used,
+ &cache->space_info);
+ BUG_ON(ret);
+
+ bit = block_group_state_bits(type);
+ set_extent_bits(block_group_cache, chunk_offset,
+ chunk_offset + size - 1,
+ bit | EXTENT_LOCKED, GFP_NOFS);
+
+ set_state_private(block_group_cache, chunk_offset,
+ (unsigned long)cache);
+ ret = btrfs_insert_item(trans, extent_root, &cache->key, &cache->item,
+ sizeof(cache->item));
+ BUG_ON(ret);
+
finish_current_insert(trans, extent_root);
- pending_ret = del_pending_extents(trans, extent_root);
- if (ret)
- return ret;
- if (pending_ret)
- return pending_ret;
+ ret = del_pending_extents(trans, extent_root);
+ set_avail_alloc_bits(extent_root->fs_info, type);
return 0;
}
+/*
+ * This is for converter use only.
+ *
+ * In that case, we don't know where are free blocks located.
+ * Therefore all block group cache entries must be setup properly
+ * before doing any block allocation.
+ */
int btrfs_make_block_groups(struct btrfs_trans_handle *trans,
struct btrfs_root *root)
{
- u64 group_size;
- u64 bytes_used;
u64 total_bytes;
u64 cur_start;
- u64 nr = 0;
+ u64 group_type;
+ u64 group_size;
+ u64 group_align;
+ u64 total_data = 0;
+ u64 total_metadata = 0;
+ u64 chunk_objectid;
int ret;
int bit;
struct btrfs_root *extent_root;
struct btrfs_block_group_cache *cache;
- struct extent_map_tree *block_group_cache;
+ struct extent_io_tree *block_group_cache;
extent_root = root->fs_info->extent_root;
block_group_cache = &root->fs_info->block_group_cache;
- group_size = BTRFS_BLOCK_GROUP_SIZE;
- bytes_used = btrfs_super_bytes_used(&root->fs_info->super_copy);
- total_bytes = btrfs_super_total_bytes(&root->fs_info->super_copy);
+ chunk_objectid = BTRFS_FIRST_CHUNK_TREE_OBJECTID;
+ total_bytes = btrfs_super_total_bytes(root->fs_info->super_copy);
+ group_align = 64 * root->sectorsize;
cur_start = 0;
while (cur_start < total_bytes) {
- cache = malloc(sizeof(*cache));
+ group_size = total_bytes / 12;
+ group_size = min_t(u64, group_size, total_bytes - cur_start);
+ if (cur_start == 0) {
+ bit = BLOCK_GROUP_SYSTEM;
+ group_type = BTRFS_BLOCK_GROUP_SYSTEM;
+ group_size /= 4;
+ group_size &= ~(group_align - 1);
+ group_size = max_t(u64, group_size, 8 * 1024 * 1024);
+ group_size = min_t(u64, group_size, 32 * 1024 * 1024);
+ } else {
+ group_size &= ~(group_align - 1);
+ if (total_data >= total_metadata * 2) {
+ group_type = BTRFS_BLOCK_GROUP_METADATA;
+ group_size = min_t(u64, group_size,
+ 1ULL * 1024 * 1024 * 1024);
+ total_metadata += group_size;
+ } else {
+ group_type = BTRFS_BLOCK_GROUP_DATA;
+ group_size = min_t(u64, group_size,
+ 5ULL * 1024 * 1024 * 1024);
+ total_data += group_size;
+ }
+ if ((total_bytes - cur_start) * 4 < group_size * 5)
+ group_size = total_bytes - cur_start;
+ }
+
+ cache = kzalloc(sizeof(*cache), GFP_NOFS);
BUG_ON(!cache);
+
cache->key.objectid = cur_start;
cache->key.offset = group_size;
btrfs_set_key_type(&cache->key, BTRFS_BLOCK_GROUP_ITEM_KEY);
- memset(&cache->item, 0, sizeof(cache->item));
- if (nr == 0)
- btrfs_set_block_group_used(&cache->item, bytes_used);
- if (nr++ % 3) {
- bit = BLOCK_GROUP_DATA;
- cache->data = 1;
- cache->item.flags |= BTRFS_BLOCK_GROUP_DATA;
- } else {
- bit = BLOCK_GROUP_METADATA;
- cache->data = 0;
- }
+
+ btrfs_set_block_group_used(&cache->item, 0);
+ btrfs_set_block_group_chunk_objectid(&cache->item,
+ chunk_objectid);
+ btrfs_set_block_group_flags(&cache->item, group_type);
+
+ cache->flags = group_type;
+
+ ret = update_space_info(root->fs_info, group_type, group_size,
+ 0, &cache->space_info);
+ BUG_ON(ret);
+ set_avail_alloc_bits(extent_root->fs_info, group_type);
set_extent_bits(block_group_cache, cur_start,
cur_start + group_size - 1,
while(cur_start < total_bytes) {
cache = btrfs_lookup_block_group(root->fs_info, cur_start);
BUG_ON(!cache);
- ret = btrfs_insert_block_group(trans, root, &cache->key,
- &cache->item);
+
+ ret = btrfs_insert_item(trans, extent_root, &cache->key, &cache->item,
+ sizeof(cache->item));
BUG_ON(ret);
- cur_start += group_size;
+
+ finish_current_insert(trans, extent_root);
+ ret = del_pending_extents(trans, extent_root);
+ BUG_ON(ret);
+
+ cur_start = cache->key.objectid + cache->key.offset;
}
return 0;
}
-u64 btrfs_hash_extent_ref(u64 root_objectid, u64 ref_generation,
- u64 owner, u64 owner_offset)
-{
- return hash_extent_ref(root_objectid, ref_generation,
- owner, owner_offset);
-}
-
int btrfs_update_block_group(struct btrfs_trans_handle *trans,
struct btrfs_root *root,
u64 bytenr, u64 num_bytes, int alloc,
- int mark_free, int data)
+ int mark_free)
{
return update_block_group(trans, root, bytenr, num_bytes,
- alloc, mark_free, data);
+ alloc, mark_free);
+}
+
+static int btrfs_count_extents_in_block_group(struct btrfs_root *root,
+ struct btrfs_path *path, u64 start,
+ u64 len,
+ u64 *total)
+{
+ struct btrfs_key key;
+ struct extent_buffer *leaf;
+ u64 bytes_used = 0;
+ int ret;
+ int slot;
+
+ key.offset = 0;
+ key.objectid = start;
+ btrfs_set_key_type(&key, BTRFS_EXTENT_ITEM_KEY);
+ ret = btrfs_search_slot(NULL, root->fs_info->extent_root,
+ &key, path, 0, 0);
+ if (ret < 0)
+ return ret;
+ while(1) {
+ leaf = path->nodes[0];
+ slot = path->slots[0];
+ if (slot >= btrfs_header_nritems(leaf)) {
+ ret = btrfs_next_leaf(root, path);
+ if (ret < 0)
+ return ret;
+ if (ret > 0)
+ break;
+ leaf = path->nodes[0];
+ slot = path->slots[0];
+ }
+ btrfs_item_key_to_cpu(leaf, &key, slot);
+ if (key.objectid > start + len)
+ break;
+ if (key.type == BTRFS_EXTENT_ITEM_KEY)
+ bytes_used += key.offset;
+ path->slots[0]++;
+ }
+ *total = bytes_used;
+ btrfs_release_path(root, path);
+ return 0;
+}
+
+int btrfs_check_block_accounting(struct btrfs_root *root)
+{
+ int ret;
+ u64 start = 0;
+ u64 bytes_used = 0;
+ struct btrfs_path path;
+ struct btrfs_block_group_cache *cache;
+ struct btrfs_fs_info *fs_info = root->fs_info;
+
+ btrfs_init_path(&path);
+
+ while(1) {
+ cache = btrfs_lookup_block_group(fs_info, start);
+ if (!cache)
+ break;
+
+ ret = btrfs_count_extents_in_block_group(root, &path,
+ cache->key.objectid,
+ cache->key.offset,
+ &bytes_used);
+
+ if (ret == 0) {
+ u64 on_disk = btrfs_block_group_used(&cache->item);
+ if (on_disk != bytes_used) {
+ fprintf(stderr, "bad block group accounting found %llu "
+ "expected %llu block group %llu\n",
+ (unsigned long long)bytes_used,
+ (unsigned long long)on_disk,
+ (unsigned long long)cache->key.objectid);
+ }
+ }
+ start = cache->key.objectid + cache->key.offset;
+
+ cache->space_info->bytes_used = 0;
+ }
+ return 0;
+}
+
+/*
+ * Fixup block accounting. The initial block accounting created by
+ * make_block_groups isn't accuracy in this case.
+ */
+int btrfs_fix_block_accounting(struct btrfs_trans_handle *trans,
+ struct btrfs_root *root)
+{
+ int ret;
+ int slot;
+ u64 start = 0;
+ u64 bytes_used = 0;
+ struct btrfs_path path;
+ struct btrfs_key key;
+ struct extent_buffer *leaf;
+ struct btrfs_block_group_cache *cache;
+ struct btrfs_fs_info *fs_info = root->fs_info;
+
+ root = root->fs_info->extent_root;
+
+ while(extent_root_pending_ops(fs_info)) {
+ ret = finish_current_insert(trans, root);
+ if (ret)
+ return ret;
+ ret = del_pending_extents(trans, root);
+ if (ret)
+ return ret;
+ }
+
+ while(1) {
+ cache = btrfs_lookup_first_block_group(fs_info, start);
+ if (!cache)
+ break;
+ start = cache->key.objectid + cache->key.offset;
+ btrfs_set_block_group_used(&cache->item, 0);
+ cache->space_info->bytes_used = 0;
+ set_extent_bits(&root->fs_info->block_group_cache,
+ cache->key.objectid,
+ cache->key.objectid + cache->key.offset -1,
+ BLOCK_GROUP_DIRTY, GFP_NOFS);
+ }
+
+ btrfs_init_path(&path);
+ key.offset = 0;
+ key.objectid = 0;
+ btrfs_set_key_type(&key, BTRFS_EXTENT_ITEM_KEY);
+ ret = btrfs_search_slot(trans, root->fs_info->extent_root,
+ &key, &path, 0, 0);
+ if (ret < 0)
+ return ret;
+ while(1) {
+ leaf = path.nodes[0];
+ slot = path.slots[0];
+ if (slot >= btrfs_header_nritems(leaf)) {
+ ret = btrfs_next_leaf(root, &path);
+ if (ret < 0)
+ return ret;
+ if (ret > 0)
+ break;
+ leaf = path.nodes[0];
+ slot = path.slots[0];
+ }
+ btrfs_item_key_to_cpu(leaf, &key, slot);
+ if (key.type == BTRFS_EXTENT_ITEM_KEY) {
+ bytes_used += key.offset;
+ ret = btrfs_update_block_group(trans, root,
+ key.objectid, key.offset, 1, 0);
+ BUG_ON(ret);
+ }
+ path.slots[0]++;
+ }
+ btrfs_set_super_bytes_used(root->fs_info->super_copy, bytes_used);
+ btrfs_release_path(root, &path);
+ return 0;
}