Btrfs: add a incompatible format change for smaller metadata extent refs
We currently store the first key of the tree block inside the reference for the
tree block in the extent tree. This takes up quite a bit of space. Make a new
key type for metadata which holds the level as the offset and completely removes
storing the btrfs_tree_block_info inside the extent ref. This reduces the size
from 51 bytes to 33 bytes per extent reference for each tree block. In practice
this results in a 30-35% decrease in the size of our extent tree, which means we
COW less and can keep more of the extent tree in memory which makes our heavy
metadata operations go much faster. This is not an automatic format change, you
must enable it at mkfs time or with btrfstune. This patch deals with having
metadata stored as either the old format or the new format so it is easy to
convert. Thanks,
Signed-off-by: Josef Bacik <jbacik@fusionio.com>
diff --git a/fs/btrfs/ctree.c b/fs/btrfs/ctree.c
index ca9d8f1..fe032ab 100644
--- a/fs/btrfs/ctree.c
+++ b/fs/btrfs/ctree.c
@@ -867,7 +867,8 @@
if (btrfs_block_can_be_shared(root, buf)) {
ret = btrfs_lookup_extent_info(trans, root, buf->start,
- buf->len, &refs, &flags);
+ btrfs_header_level(buf), 1,
+ &refs, &flags);
if (ret)
return ret;
if (refs == 0) {
diff --git a/fs/btrfs/ctree.h b/fs/btrfs/ctree.h
index e2f14b5..efb2feb 100644
--- a/fs/btrfs/ctree.h
+++ b/fs/btrfs/ctree.h
@@ -509,6 +509,7 @@
#define BTRFS_FEATURE_INCOMPAT_EXTENDED_IREF (1ULL << 6)
#define BTRFS_FEATURE_INCOMPAT_RAID56 (1ULL << 7)
+#define BTRFS_FEATURE_INCOMPAT_SKINNY_METADATA (1ULL << 8)
#define BTRFS_FEATURE_COMPAT_SUPP 0ULL
#define BTRFS_FEATURE_COMPAT_RO_SUPP 0ULL
@@ -519,7 +520,8 @@
BTRFS_FEATURE_INCOMPAT_BIG_METADATA | \
BTRFS_FEATURE_INCOMPAT_COMPRESS_LZO | \
BTRFS_FEATURE_INCOMPAT_RAID56 | \
- BTRFS_FEATURE_INCOMPAT_EXTENDED_IREF)
+ BTRFS_FEATURE_INCOMPAT_EXTENDED_IREF | \
+ BTRFS_FEATURE_INCOMPAT_SKINNY_METADATA)
/*
* A leaf is full of items. offset and size tell us where to find
@@ -1809,6 +1811,12 @@
*/
#define BTRFS_EXTENT_ITEM_KEY 168
+/*
+ * The same as the BTRFS_EXTENT_ITEM_KEY, except it's metadata we already know
+ * the length, so we save the level in key->offset instead of the length.
+ */
+#define BTRFS_METADATA_ITEM_KEY 169
+
#define BTRFS_TREE_BLOCK_REF_KEY 176
#define BTRFS_EXTENT_DATA_REF_KEY 178
@@ -3006,7 +3014,7 @@
int btrfs_lookup_extent(struct btrfs_root *root, u64 start, u64 len);
int btrfs_lookup_extent_info(struct btrfs_trans_handle *trans,
struct btrfs_root *root, u64 bytenr,
- u64 num_bytes, u64 *refs, u64 *flags);
+ u64 offset, int metadata, u64 *refs, u64 *flags);
int btrfs_pin_extent(struct btrfs_root *root,
u64 bytenr, u64 num, int reserved);
int btrfs_pin_extent_for_log_replay(struct btrfs_root *root,
@@ -3669,6 +3677,16 @@
}
}
+#define btrfs_fs_incompat(fs_info, opt) \
+ __btrfs_fs_incompat((fs_info), BTRFS_FEATURE_INCOMPAT_##opt)
+
+static inline int __btrfs_fs_incompat(struct btrfs_fs_info *fs_info, u64 flag)
+{
+ struct btrfs_super_block *disk_super;
+ disk_super = fs_info->super_copy;
+ return !!(btrfs_super_incompat_flags(disk_super) & flag);
+}
+
/*
* Call btrfs_abort_transaction as early as possible when an error condition is
* detected, that way the exact line number is reported.
diff --git a/fs/btrfs/disk-io.c b/fs/btrfs/disk-io.c
index e0665488..f47754a 100644
--- a/fs/btrfs/disk-io.c
+++ b/fs/btrfs/disk-io.c
@@ -2290,6 +2290,9 @@
if (tree_root->fs_info->compress_type == BTRFS_COMPRESS_LZO)
features |= BTRFS_FEATURE_INCOMPAT_COMPRESS_LZO;
+ if (features & BTRFS_FEATURE_INCOMPAT_SKINNY_METADATA)
+ printk(KERN_ERR "btrfs: has skinny extents\n");
+
/*
* flag our filesystem as having big metadata blocks if
* they are bigger than the page size
diff --git a/fs/btrfs/extent-tree.c b/fs/btrfs/extent-tree.c
index 3d55123..7505856 100644
--- a/fs/btrfs/extent-tree.c
+++ b/fs/btrfs/extent-tree.c
@@ -442,11 +442,16 @@
block_group->key.offset)
break;
- if (key.type == BTRFS_EXTENT_ITEM_KEY) {
+ if (key.type == BTRFS_EXTENT_ITEM_KEY ||
+ key.type == BTRFS_METADATA_ITEM_KEY) {
total_found += add_new_free_space(block_group,
fs_info, last,
key.objectid);
- last = key.objectid + key.offset;
+ if (key.type == BTRFS_METADATA_ITEM_KEY)
+ last = key.objectid +
+ fs_info->tree_root->leafsize;
+ else
+ last = key.objectid + key.offset;
if (total_found > (1024 * 1024 * 2)) {
total_found = 0;
@@ -718,15 +723,21 @@
key.objectid = start;
key.offset = len;
- btrfs_set_key_type(&key, BTRFS_EXTENT_ITEM_KEY);
+ key.type = BTRFS_EXTENT_ITEM_KEY;
ret = btrfs_search_slot(NULL, root->fs_info->extent_root, &key, path,
0, 0);
+ if (ret > 0) {
+ btrfs_item_key_to_cpu(path->nodes[0], &key, path->slots[0]);
+ if (key.objectid == start &&
+ key.type == BTRFS_METADATA_ITEM_KEY)
+ ret = 0;
+ }
btrfs_free_path(path);
return ret;
}
/*
- * helper function to lookup reference count and flags of extent.
+ * helper function to lookup reference count and flags of a tree block.
*
* the head node for delayed ref is used to store the sum of all the
* reference count modifications queued up in the rbtree. the head
@@ -736,7 +747,7 @@
*/
int btrfs_lookup_extent_info(struct btrfs_trans_handle *trans,
struct btrfs_root *root, u64 bytenr,
- u64 num_bytes, u64 *refs, u64 *flags)
+ u64 offset, int metadata, u64 *refs, u64 *flags)
{
struct btrfs_delayed_ref_head *head;
struct btrfs_delayed_ref_root *delayed_refs;
@@ -749,13 +760,29 @@
u64 extent_flags;
int ret;
+ /*
+ * If we don't have skinny metadata, don't bother doing anything
+ * different
+ */
+ if (metadata && !btrfs_fs_incompat(root->fs_info, SKINNY_METADATA)) {
+ offset = root->leafsize;
+ metadata = 0;
+ }
+
path = btrfs_alloc_path();
if (!path)
return -ENOMEM;
- key.objectid = bytenr;
- key.type = BTRFS_EXTENT_ITEM_KEY;
- key.offset = num_bytes;
+ if (metadata) {
+ key.objectid = bytenr;
+ key.type = BTRFS_METADATA_ITEM_KEY;
+ key.offset = offset;
+ } else {
+ key.objectid = bytenr;
+ key.type = BTRFS_EXTENT_ITEM_KEY;
+ key.offset = offset;
+ }
+
if (!trans) {
path->skip_locking = 1;
path->search_commit_root = 1;
@@ -766,6 +793,13 @@
if (ret < 0)
goto out_free;
+ if (ret > 0 && metadata && key.type == BTRFS_METADATA_ITEM_KEY) {
+ key.type = BTRFS_EXTENT_ITEM_KEY;
+ key.offset = root->leafsize;
+ btrfs_release_path(path);
+ goto again;
+ }
+
if (ret == 0) {
leaf = path->nodes[0];
item_size = btrfs_item_size_nr(leaf, path->slots[0]);
@@ -1453,6 +1487,8 @@
int want;
int ret;
int err = 0;
+ bool skinny_metadata = btrfs_fs_incompat(root->fs_info,
+ SKINNY_METADATA);
key.objectid = bytenr;
key.type = BTRFS_EXTENT_ITEM_KEY;
@@ -1464,11 +1500,46 @@
path->keep_locks = 1;
} else
extra_size = -1;
+
+ /*
+ * Owner is our parent level, so we can just add one to get the level
+ * for the block we are interested in.
+ */
+ if (skinny_metadata && owner < BTRFS_FIRST_FREE_OBJECTID) {
+ key.type = BTRFS_METADATA_ITEM_KEY;
+ key.offset = owner;
+ }
+
+again:
ret = btrfs_search_slot(trans, root, &key, path, extra_size, 1);
if (ret < 0) {
err = ret;
goto out;
}
+
+ /*
+ * We may be a newly converted file system which still has the old fat
+ * extent entries for metadata, so try and see if we have one of those.
+ */
+ if (ret > 0 && skinny_metadata) {
+ skinny_metadata = false;
+ if (path->slots[0]) {
+ path->slots[0]--;
+ btrfs_item_key_to_cpu(path->nodes[0], &key,
+ path->slots[0]);
+ if (key.objectid == bytenr &&
+ key.type == BTRFS_EXTENT_ITEM_KEY &&
+ key.offset == num_bytes)
+ ret = 0;
+ }
+ if (ret) {
+ key.type = BTRFS_EXTENT_ITEM_KEY;
+ key.offset = num_bytes;
+ btrfs_release_path(path);
+ goto again;
+ }
+ }
+
if (ret && !insert) {
err = -ENOENT;
goto out;
@@ -1504,11 +1575,9 @@
ptr = (unsigned long)(ei + 1);
end = (unsigned long)ei + item_size;
- if (flags & BTRFS_EXTENT_FLAG_TREE_BLOCK) {
+ if (flags & BTRFS_EXTENT_FLAG_TREE_BLOCK && !skinny_metadata) {
ptr += sizeof(struct btrfs_tree_block_info);
BUG_ON(ptr > end);
- } else {
- BUG_ON(!(flags & BTRFS_EXTENT_FLAG_DATA));
}
err = -ENOENT;
@@ -1973,10 +2042,8 @@
ref_root = ref->root;
if (node->action == BTRFS_ADD_DELAYED_REF && insert_reserved) {
- if (extent_op) {
- BUG_ON(extent_op->update_key);
+ if (extent_op)
flags |= extent_op->flags_to_set;
- }
ret = alloc_reserved_file_extent(trans, root,
parent, ref_root, flags,
ref->objectid, ref->offset,
@@ -2029,18 +2096,33 @@
u32 item_size;
int ret;
int err = 0;
+ int metadata = (node->type == BTRFS_TREE_BLOCK_REF_KEY ||
+ node->type == BTRFS_SHARED_BLOCK_REF_KEY);
if (trans->aborted)
return 0;
+ if (metadata && !btrfs_fs_incompat(root->fs_info, SKINNY_METADATA))
+ metadata = 0;
+
path = btrfs_alloc_path();
if (!path)
return -ENOMEM;
key.objectid = node->bytenr;
- key.type = BTRFS_EXTENT_ITEM_KEY;
- key.offset = node->num_bytes;
+ if (metadata) {
+ struct btrfs_delayed_tree_ref *tree_ref;
+
+ tree_ref = btrfs_delayed_node_to_tree_ref(node);
+ key.type = BTRFS_METADATA_ITEM_KEY;
+ key.offset = tree_ref->level;
+ } else {
+ key.type = BTRFS_EXTENT_ITEM_KEY;
+ key.offset = node->num_bytes;
+ }
+
+again:
path->reada = 1;
path->leave_spinning = 1;
ret = btrfs_search_slot(trans, root->fs_info->extent_root, &key,
@@ -2050,6 +2132,14 @@
goto out;
}
if (ret > 0) {
+ if (metadata) {
+ btrfs_release_path(path);
+ metadata = 0;
+
+ key.offset = node->num_bytes;
+ key.type = BTRFS_EXTENT_ITEM_KEY;
+ goto again;
+ }
err = -EIO;
goto out;
}
@@ -2089,10 +2179,8 @@
struct btrfs_key ins;
u64 parent = 0;
u64 ref_root = 0;
-
- ins.objectid = node->bytenr;
- ins.offset = node->num_bytes;
- ins.type = BTRFS_EXTENT_ITEM_KEY;
+ bool skinny_metadata = btrfs_fs_incompat(root->fs_info,
+ SKINNY_METADATA);
ref = btrfs_delayed_node_to_tree_ref(node);
if (node->type == BTRFS_SHARED_BLOCK_REF_KEY)
@@ -2100,10 +2188,18 @@
else
ref_root = ref->root;
+ ins.objectid = node->bytenr;
+ if (skinny_metadata) {
+ ins.offset = ref->level;
+ ins.type = BTRFS_METADATA_ITEM_KEY;
+ } else {
+ ins.offset = node->num_bytes;
+ ins.type = BTRFS_EXTENT_ITEM_KEY;
+ }
+
BUG_ON(node->ref_mod != 1);
if (node->action == BTRFS_ADD_DELAYED_REF && insert_reserved) {
- BUG_ON(!extent_op || !extent_op->update_flags ||
- !extent_op->update_key);
+ BUG_ON(!extent_op || !extent_op->update_flags);
ret = alloc_reserved_tree_block(trans, root,
parent, ref_root,
extent_op->flags_to_set,
@@ -5312,6 +5408,8 @@
int num_to_del = 1;
u32 item_size;
u64 refs;
+ bool skinny_metadata = btrfs_fs_incompat(root->fs_info,
+ SKINNY_METADATA);
path = btrfs_alloc_path();
if (!path)
@@ -5323,6 +5421,9 @@
is_data = owner_objectid >= BTRFS_FIRST_FREE_OBJECTID;
BUG_ON(!is_data && refs_to_drop != 1);
+ if (is_data)
+ skinny_metadata = 0;
+
ret = lookup_extent_backref(trans, extent_root, path, &iref,
bytenr, num_bytes, parent,
root_objectid, owner_objectid,
@@ -5339,6 +5440,11 @@
found_extent = 1;
break;
}
+ if (key.type == BTRFS_METADATA_ITEM_KEY &&
+ key.offset == owner_objectid) {
+ found_extent = 1;
+ break;
+ }
if (path->slots[0] - extent_slot > 5)
break;
extent_slot--;
@@ -5364,8 +5470,36 @@
key.type = BTRFS_EXTENT_ITEM_KEY;
key.offset = num_bytes;
+ if (!is_data && skinny_metadata) {
+ key.type = BTRFS_METADATA_ITEM_KEY;
+ key.offset = owner_objectid;
+ }
+
ret = btrfs_search_slot(trans, extent_root,
&key, path, -1, 1);
+ if (ret > 0 && skinny_metadata && path->slots[0]) {
+ /*
+ * Couldn't find our skinny metadata item,
+ * see if we have ye olde extent item.
+ */
+ path->slots[0]--;
+ btrfs_item_key_to_cpu(path->nodes[0], &key,
+ path->slots[0]);
+ if (key.objectid == bytenr &&
+ key.type == BTRFS_EXTENT_ITEM_KEY &&
+ key.offset == num_bytes)
+ ret = 0;
+ }
+
+ if (ret > 0 && skinny_metadata) {
+ skinny_metadata = false;
+ key.type = BTRFS_EXTENT_ITEM_KEY;
+ key.offset = num_bytes;
+ btrfs_release_path(path);
+ 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,
@@ -5435,7 +5569,8 @@
BUG_ON(item_size < sizeof(*ei));
ei = btrfs_item_ptr(leaf, extent_slot,
struct btrfs_extent_item);
- if (owner_objectid < BTRFS_FIRST_FREE_OBJECTID) {
+ if (owner_objectid < BTRFS_FIRST_FREE_OBJECTID &&
+ key.type == BTRFS_EXTENT_ITEM_KEY) {
struct btrfs_tree_block_info *bi;
BUG_ON(item_size < sizeof(*ei) + sizeof(*bi));
bi = (struct btrfs_tree_block_info *)(ei + 1);
@@ -6349,7 +6484,12 @@
struct btrfs_extent_inline_ref *iref;
struct btrfs_path *path;
struct extent_buffer *leaf;
- u32 size = sizeof(*extent_item) + sizeof(*block_info) + sizeof(*iref);
+ u32 size = sizeof(*extent_item) + sizeof(*iref);
+ bool skinny_metadata = btrfs_fs_incompat(root->fs_info,
+ SKINNY_METADATA);
+
+ if (!skinny_metadata)
+ size += sizeof(*block_info);
path = btrfs_alloc_path();
if (!path)
@@ -6370,12 +6510,16 @@
btrfs_set_extent_generation(leaf, extent_item, trans->transid);
btrfs_set_extent_flags(leaf, extent_item,
flags | BTRFS_EXTENT_FLAG_TREE_BLOCK);
- block_info = (struct btrfs_tree_block_info *)(extent_item + 1);
- btrfs_set_tree_block_key(leaf, block_info, key);
- btrfs_set_tree_block_level(leaf, block_info, level);
+ if (skinny_metadata) {
+ iref = (struct btrfs_extent_inline_ref *)(extent_item + 1);
+ } else {
+ block_info = (struct btrfs_tree_block_info *)(extent_item + 1);
+ 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);
+ }
- iref = (struct btrfs_extent_inline_ref *)(block_info + 1);
if (parent > 0) {
BUG_ON(!(flags & BTRFS_BLOCK_FLAG_FULL_BACKREF));
btrfs_set_extent_inline_ref_type(leaf, iref,
@@ -6390,7 +6534,7 @@
btrfs_mark_buffer_dirty(leaf);
btrfs_free_path(path);
- ret = update_block_group(root, ins->objectid, ins->offset, 1);
+ ret = update_block_group(root, ins->objectid, root->leafsize, 1);
if (ret) { /* -ENOENT, logic error */
printk(KERN_ERR "btrfs update block group failed for %llu "
"%llu\n", (unsigned long long)ins->objectid,
@@ -6594,7 +6738,8 @@
struct extent_buffer *buf;
u64 flags = 0;
int ret;
-
+ bool skinny_metadata = btrfs_fs_incompat(root->fs_info,
+ SKINNY_METADATA);
block_rsv = use_block_rsv(trans, root, blocksize);
if (IS_ERR(block_rsv))
@@ -6627,7 +6772,10 @@
else
memset(&extent_op->key, 0, sizeof(extent_op->key));
extent_op->flags_to_set = flags;
- extent_op->update_key = 1;
+ if (skinny_metadata)
+ extent_op->update_key = 0;
+ else
+ extent_op->update_key = 1;
extent_op->update_flags = 1;
extent_op->is_data = 0;
@@ -6704,8 +6852,9 @@
continue;
/* We don't lock the tree block, it's OK to be racy here */
- ret = btrfs_lookup_extent_info(trans, root, bytenr, blocksize,
- &refs, &flags);
+ ret = btrfs_lookup_extent_info(trans, root, bytenr,
+ wc->level - 1, 1, &refs,
+ &flags);
/* We don't care about errors in readahead. */
if (ret < 0)
continue;
@@ -6772,7 +6921,7 @@
(wc->stage == UPDATE_BACKREF && !(wc->flags[level] & flag)))) {
BUG_ON(!path->locks[level]);
ret = btrfs_lookup_extent_info(trans, root,
- eb->start, eb->len,
+ eb->start, level, 1,
&wc->refs[level],
&wc->flags[level]);
BUG_ON(ret == -ENOMEM);
@@ -6870,7 +7019,7 @@
btrfs_tree_lock(next);
btrfs_set_lock_blocking(next);
- ret = btrfs_lookup_extent_info(trans, root, bytenr, blocksize,
+ ret = btrfs_lookup_extent_info(trans, root, bytenr, level - 1, 1,
&wc->refs[level - 1],
&wc->flags[level - 1]);
if (ret < 0) {
@@ -7001,7 +7150,7 @@
path->locks[level] = BTRFS_WRITE_LOCK_BLOCKING;
ret = btrfs_lookup_extent_info(trans, root,
- eb->start, eb->len,
+ eb->start, level, 1,
&wc->refs[level],
&wc->flags[level]);
if (ret < 0) {
@@ -7211,8 +7360,7 @@
ret = btrfs_lookup_extent_info(trans, root,
path->nodes[level]->start,
- path->nodes[level]->len,
- &wc->refs[level],
+ level, 1, &wc->refs[level],
&wc->flags[level]);
if (ret < 0) {
err = ret;
diff --git a/fs/btrfs/inode.c b/fs/btrfs/inode.c
index 24e8a35..c69145b 100644
--- a/fs/btrfs/inode.c
+++ b/fs/btrfs/inode.c
@@ -3660,7 +3660,7 @@
eb = path->nodes[level];
if (!btrfs_block_can_be_shared(root, eb))
continue;
- ret = btrfs_lookup_extent_info(NULL, root, eb->start, eb->len,
+ ret = btrfs_lookup_extent_info(NULL, root, eb->start, level, 1,
&refs, NULL);
if (refs > 1)
return 1;
diff --git a/fs/btrfs/relocation.c b/fs/btrfs/relocation.c
index b67171e..86f192f 100644
--- a/fs/btrfs/relocation.c
+++ b/fs/btrfs/relocation.c
@@ -619,10 +619,13 @@
int find_inline_backref(struct extent_buffer *leaf, int slot,
unsigned long *ptr, unsigned long *end)
{
+ struct btrfs_key key;
struct btrfs_extent_item *ei;
struct btrfs_tree_block_info *bi;
u32 item_size;
+ btrfs_item_key_to_cpu(leaf, &key, slot);
+
item_size = btrfs_item_size_nr(leaf, slot);
#ifdef BTRFS_COMPAT_EXTENT_TREE_V0
if (item_size < sizeof(*ei)) {
@@ -634,13 +637,18 @@
WARN_ON(!(btrfs_extent_flags(leaf, ei) &
BTRFS_EXTENT_FLAG_TREE_BLOCK));
- if (item_size <= sizeof(*ei) + sizeof(*bi)) {
+ if (key.type == BTRFS_EXTENT_ITEM_KEY &&
+ item_size <= sizeof(*ei) + sizeof(*bi)) {
WARN_ON(item_size < sizeof(*ei) + sizeof(*bi));
return 1;
}
- bi = (struct btrfs_tree_block_info *)(ei + 1);
- *ptr = (unsigned long)(bi + 1);
+ if (key.type == BTRFS_EXTENT_ITEM_KEY) {
+ bi = (struct btrfs_tree_block_info *)(ei + 1);
+ *ptr = (unsigned long)(bi + 1);
+ } else {
+ *ptr = (unsigned long)(ei + 1);
+ }
*end = (unsigned long)ei + item_size;
return 0;
}
@@ -708,7 +716,7 @@
end = 0;
ptr = 0;
key.objectid = cur->bytenr;
- key.type = BTRFS_EXTENT_ITEM_KEY;
+ key.type = BTRFS_METADATA_ITEM_KEY;
key.offset = (u64)-1;
path1->search_commit_root = 1;
@@ -766,7 +774,8 @@
break;
}
- if (key.type == BTRFS_EXTENT_ITEM_KEY) {
+ if (key.type == BTRFS_EXTENT_ITEM_KEY ||
+ key.type == BTRFS_METADATA_ITEM_KEY) {
ret = find_inline_backref(eb, path1->slots[0],
&ptr, &end);
if (ret)
@@ -2768,8 +2777,13 @@
struct tree_block *block)
{
BUG_ON(block->key_ready);
- readahead_tree_block(rc->extent_root, block->bytenr,
- block->key.objectid, block->key.offset);
+ if (block->key.type == BTRFS_METADATA_ITEM_KEY)
+ readahead_tree_block(rc->extent_root, block->bytenr,
+ block->key.objectid,
+ rc->extent_root->leafsize);
+ else
+ readahead_tree_block(rc->extent_root, block->bytenr,
+ block->key.objectid, block->key.offset);
return 0;
}
@@ -3176,12 +3190,17 @@
eb = path->nodes[0];
item_size = btrfs_item_size_nr(eb, path->slots[0]);
- if (item_size >= sizeof(*ei) + sizeof(*bi)) {
+ if (extent_key->type == BTRFS_METADATA_ITEM_KEY ||
+ item_size >= sizeof(*ei) + sizeof(*bi)) {
ei = btrfs_item_ptr(eb, path->slots[0],
struct btrfs_extent_item);
- bi = (struct btrfs_tree_block_info *)(ei + 1);
+ if (extent_key->type == BTRFS_EXTENT_ITEM_KEY) {
+ bi = (struct btrfs_tree_block_info *)(ei + 1);
+ level = btrfs_tree_block_level(eb, bi);
+ } else {
+ level = (int)extent_key->offset;
+ }
generation = btrfs_extent_generation(eb, ei);
- level = btrfs_tree_block_level(eb, bi);
} else {
#ifdef BTRFS_COMPAT_EXTENT_TREE_V0
u64 ref_owner;
@@ -3210,7 +3229,7 @@
return -ENOMEM;
block->bytenr = extent_key->objectid;
- block->key.objectid = extent_key->offset;
+ block->key.objectid = rc->extent_root->leafsize;
block->key.offset = generation;
block->level = level;
block->key_ready = 0;
@@ -3252,9 +3271,15 @@
ret = btrfs_search_slot(NULL, rc->extent_root, &key, path, 0, 0);
if (ret < 0)
goto out;
- BUG_ON(ret);
btrfs_item_key_to_cpu(path->nodes[0], &key, path->slots[0]);
+ if (ret > 0) {
+ if (key.objectid == bytenr &&
+ key.type == BTRFS_METADATA_ITEM_KEY)
+ ret = 0;
+ }
+ BUG_ON(ret);
+
ret = add_tree_block(rc, &key, path, blocks);
out:
btrfs_free_path(path);
@@ -3275,7 +3300,8 @@
return 1;
ret = btrfs_lookup_extent_info(NULL, rc->extent_root,
- eb->start, eb->len, NULL, &flags);
+ eb->start, btrfs_header_level(eb), 1,
+ NULL, &flags);
BUG_ON(ret);
if (flags & BTRFS_BLOCK_FLAG_FULL_BACKREF)
@@ -3644,12 +3670,25 @@
break;
}
- if (key.type != BTRFS_EXTENT_ITEM_KEY ||
+ if (key.type != BTRFS_EXTENT_ITEM_KEY &&
+ key.type != BTRFS_METADATA_ITEM_KEY) {
+ path->slots[0]++;
+ goto next;
+ }
+
+ if (key.type == BTRFS_EXTENT_ITEM_KEY &&
key.objectid + key.offset <= rc->search_start) {
path->slots[0]++;
goto next;
}
+ if (key.type == BTRFS_METADATA_ITEM_KEY &&
+ key.objectid + rc->extent_root->leafsize <=
+ rc->search_start) {
+ path->slots[0]++;
+ goto next;
+ }
+
ret = find_first_extent_bit(&rc->processed_blocks,
key.objectid, &start, &end,
EXTENT_DIRTY, NULL);
@@ -3658,7 +3697,11 @@
btrfs_release_path(path);
rc->search_start = end + 1;
} else {
- rc->search_start = key.objectid + key.offset;
+ if (key.type == BTRFS_EXTENT_ITEM_KEY)
+ rc->search_start = key.objectid + key.offset;
+ else
+ rc->search_start = key.objectid +
+ rc->extent_root->leafsize;
memcpy(extent_key, &key, sizeof(key));
return 0;
}
diff --git a/fs/btrfs/scrub.c b/fs/btrfs/scrub.c
index 3d29d60..28db5dc 100644
--- a/fs/btrfs/scrub.c
+++ b/fs/btrfs/scrub.c
@@ -2312,8 +2312,8 @@
key_start.type = BTRFS_EXTENT_ITEM_KEY;
key_start.offset = (u64)0;
key_end.objectid = base + offset + nstripes * increment;
- key_end.type = BTRFS_EXTENT_ITEM_KEY;
- key_end.offset = (u64)0;
+ key_end.type = BTRFS_METADATA_ITEM_KEY;
+ key_end.offset = (u64)-1;
reada1 = btrfs_reada_add(root, &key_start, &key_end);
key_start.objectid = BTRFS_EXTENT_CSUM_OBJECTID;
@@ -2401,6 +2401,7 @@
ret = btrfs_search_slot(NULL, root, &key, path, 0, 0);
if (ret < 0)
goto out;
+
if (ret > 0) {
ret = btrfs_previous_item(root, path, 0,
BTRFS_EXTENT_ITEM_KEY);
@@ -2418,6 +2419,8 @@
}
while (1) {
+ u64 bytes;
+
l = path->nodes[0];
slot = path->slots[0];
if (slot >= btrfs_header_nritems(l)) {
@@ -2431,14 +2434,21 @@
}
btrfs_item_key_to_cpu(l, &key, slot);
- if (key.objectid + key.offset <= logical)
+ if (key.type != BTRFS_EXTENT_ITEM_KEY &&
+ key.type != BTRFS_METADATA_ITEM_KEY)
+ goto next;
+
+ if (key.type == BTRFS_METADATA_ITEM_KEY)
+ bytes = root->leafsize;
+ else
+ bytes = key.offset;
+
+ if (key.objectid + bytes <= logical)
goto next;
if (key.objectid >= logical + map->stripe_len)
break;
- if (btrfs_key_type(&key) != BTRFS_EXTENT_ITEM_KEY)
- goto next;
extent = btrfs_item_ptr(l, slot,
struct btrfs_extent_item);
@@ -2459,18 +2469,18 @@
* trim extent to this stripe
*/
if (key.objectid < logical) {
- key.offset -= logical - key.objectid;
+ bytes -= logical - key.objectid;
key.objectid = logical;
}
- if (key.objectid + key.offset >
+ if (key.objectid + bytes >
logical + map->stripe_len) {
- key.offset = logical + map->stripe_len -
- key.objectid;
+ bytes = logical + map->stripe_len -
+ key.objectid;
}
extent_logical = key.objectid;
extent_physical = key.objectid - logical + physical;
- extent_len = key.offset;
+ extent_len = bytes;
extent_dev = scrub_dev;
extent_mirror_num = mirror_num;
if (is_dev_replace)