summaryrefslogtreecommitdiffstats
path: root/fs/btrfs
diff options
context:
space:
mode:
Diffstat (limited to 'fs/btrfs')
-rw-r--r--fs/btrfs/acl.c44
-rw-r--r--fs/btrfs/async-thread.c2
-rw-r--r--fs/btrfs/btrfs_inode.h4
-rw-r--r--fs/btrfs/ctree.h5
-rw-r--r--fs/btrfs/extent-tree.c566
-rw-r--r--fs/btrfs/file.c5
-rw-r--r--fs/btrfs/inode.c43
-rw-r--r--fs/btrfs/ioctl.c6
-rw-r--r--fs/btrfs/relocation.c5
-rw-r--r--fs/btrfs/transaction.c4
10 files changed, 434 insertions, 250 deletions
diff --git a/fs/btrfs/acl.c b/fs/btrfs/acl.c
index 603972576f0..f128427b995 100644
--- a/fs/btrfs/acl.c
+++ b/fs/btrfs/acl.c
@@ -29,51 +29,28 @@
#ifdef CONFIG_FS_POSIX_ACL
-static void btrfs_update_cached_acl(struct inode *inode,
- struct posix_acl **p_acl,
- struct posix_acl *acl)
-{
- spin_lock(&inode->i_lock);
- if (*p_acl && *p_acl != BTRFS_ACL_NOT_CACHED)
- posix_acl_release(*p_acl);
- *p_acl = posix_acl_dup(acl);
- spin_unlock(&inode->i_lock);
-}
-
static struct posix_acl *btrfs_get_acl(struct inode *inode, int type)
{
int size;
const char *name;
char *value = NULL;
- struct posix_acl *acl = NULL, **p_acl;
+ struct posix_acl *acl;
+
+ acl = get_cached_acl(inode, type);
+ if (acl != ACL_NOT_CACHED)
+ return acl;
switch (type) {
case ACL_TYPE_ACCESS:
name = POSIX_ACL_XATTR_ACCESS;
- p_acl = &BTRFS_I(inode)->i_acl;
break;
case ACL_TYPE_DEFAULT:
name = POSIX_ACL_XATTR_DEFAULT;
- p_acl = &BTRFS_I(inode)->i_default_acl;
break;
default:
- return ERR_PTR(-EINVAL);
+ BUG();
}
- /* Handle the cached NULL acl case without locking */
- acl = ACCESS_ONCE(*p_acl);
- if (!acl)
- return acl;
-
- spin_lock(&inode->i_lock);
- acl = *p_acl;
- if (acl != BTRFS_ACL_NOT_CACHED)
- acl = posix_acl_dup(acl);
- spin_unlock(&inode->i_lock);
-
- if (acl != BTRFS_ACL_NOT_CACHED)
- return acl;
-
size = __btrfs_getxattr(inode, name, "", 0);
if (size > 0) {
value = kzalloc(size, GFP_NOFS);
@@ -82,13 +59,13 @@ static struct posix_acl *btrfs_get_acl(struct inode *inode, int type)
size = __btrfs_getxattr(inode, name, value, size);
if (size > 0) {
acl = posix_acl_from_xattr(value, size);
- btrfs_update_cached_acl(inode, p_acl, acl);
+ set_cached_acl(inode, type, acl);
}
kfree(value);
} else if (size == -ENOENT || size == -ENODATA || size == 0) {
/* FIXME, who returns -ENOENT? I think nobody */
acl = NULL;
- btrfs_update_cached_acl(inode, p_acl, acl);
+ set_cached_acl(inode, type, acl);
} else {
acl = ERR_PTR(-EIO);
}
@@ -121,7 +98,6 @@ static int btrfs_set_acl(struct inode *inode, struct posix_acl *acl, int type)
{
int ret, size = 0;
const char *name;
- struct posix_acl **p_acl;
char *value = NULL;
mode_t mode;
@@ -141,13 +117,11 @@ static int btrfs_set_acl(struct inode *inode, struct posix_acl *acl, int type)
ret = 0;
inode->i_mode = mode;
name = POSIX_ACL_XATTR_ACCESS;
- p_acl = &BTRFS_I(inode)->i_acl;
break;
case ACL_TYPE_DEFAULT:
if (!S_ISDIR(inode->i_mode))
return acl ? -EINVAL : 0;
name = POSIX_ACL_XATTR_DEFAULT;
- p_acl = &BTRFS_I(inode)->i_default_acl;
break;
default:
return -EINVAL;
@@ -172,7 +146,7 @@ out:
kfree(value);
if (!ret)
- btrfs_update_cached_acl(inode, p_acl, acl);
+ set_cached_acl(inode, type, acl);
return ret;
}
diff --git a/fs/btrfs/async-thread.c b/fs/btrfs/async-thread.c
index 7f88628a1a7..6e4f6c50a12 100644
--- a/fs/btrfs/async-thread.c
+++ b/fs/btrfs/async-thread.c
@@ -299,8 +299,8 @@ int btrfs_start_workers(struct btrfs_workers *workers, int num_workers)
"btrfs-%s-%d", workers->name,
workers->num_workers + i);
if (IS_ERR(worker->task)) {
- kfree(worker);
ret = PTR_ERR(worker->task);
+ kfree(worker);
goto fail;
}
diff --git a/fs/btrfs/btrfs_inode.h b/fs/btrfs/btrfs_inode.h
index acb4f351758..ea1ea0af8c0 100644
--- a/fs/btrfs/btrfs_inode.h
+++ b/fs/btrfs/btrfs_inode.h
@@ -53,10 +53,6 @@ struct btrfs_inode {
/* used to order data wrt metadata */
struct btrfs_ordered_inode_tree ordered_tree;
- /* standard acl pointers */
- struct posix_acl *i_acl;
- struct posix_acl *i_default_acl;
-
/* for keeping track of orphaned inodes */
struct list_head i_orphan;
diff --git a/fs/btrfs/ctree.h b/fs/btrfs/ctree.h
index 03441a99ea3..98a87383871 100644
--- a/fs/btrfs/ctree.h
+++ b/fs/btrfs/ctree.h
@@ -41,8 +41,6 @@ struct btrfs_ordered_sum;
#define BTRFS_MAGIC "_BHRfS_M"
-#define BTRFS_ACL_NOT_CACHED ((void *)-1)
-
#define BTRFS_MAX_LEVEL 8
#define BTRFS_COMPAT_EXTENT_TREE_V0
@@ -2076,8 +2074,7 @@ static inline int btrfs_insert_empty_item(struct btrfs_trans_handle *trans,
int btrfs_next_leaf(struct btrfs_root *root, struct btrfs_path *path);
int btrfs_prev_leaf(struct btrfs_root *root, struct btrfs_path *path);
int btrfs_leaf_free_space(struct btrfs_root *root, struct extent_buffer *leaf);
-int btrfs_drop_snapshot(struct btrfs_trans_handle *trans, struct btrfs_root
- *root);
+int btrfs_drop_snapshot(struct btrfs_root *root, int update_ref);
int btrfs_drop_subtree(struct btrfs_trans_handle *trans,
struct btrfs_root *root,
struct extent_buffer *node,
diff --git a/fs/btrfs/extent-tree.c b/fs/btrfs/extent-tree.c
index edc7d208c5c..a5aca3997d4 100644
--- a/fs/btrfs/extent-tree.c
+++ b/fs/btrfs/extent-tree.c
@@ -990,15 +990,13 @@ static inline int extent_ref_type(u64 parent, u64 owner)
return type;
}
-static int find_next_key(struct btrfs_path *path, struct btrfs_key *key)
+static int find_next_key(struct btrfs_path *path, int level,
+ struct btrfs_key *key)
{
- int level;
- BUG_ON(!path->keep_locks);
- for (level = 0; level < BTRFS_MAX_LEVEL; level++) {
+ for (; level < BTRFS_MAX_LEVEL; level++) {
if (!path->nodes[level])
break;
- btrfs_assert_tree_locked(path->nodes[level]);
if (path->slots[level] + 1 >=
btrfs_header_nritems(path->nodes[level]))
continue;
@@ -1158,7 +1156,8 @@ int lookup_inline_extent_backref(struct btrfs_trans_handle *trans,
* For simplicity, we just do not add new inline back
* ref if there is any kind of item for this block
*/
- if (find_next_key(path, &key) == 0 && key.objectid == bytenr &&
+ if (find_next_key(path, 0, &key) == 0 &&
+ key.objectid == bytenr &&
key.type < BTRFS_BLOCK_GROUP_ITEM_KEY) {
err = -EAGAIN;
goto out;
@@ -2697,7 +2696,7 @@ again:
printk(KERN_ERR "no space left, need %llu, %llu delalloc bytes"
", %llu bytes_used, %llu bytes_reserved, "
- "%llu bytes_pinned, %llu bytes_readonly, %llu may use"
+ "%llu bytes_pinned, %llu bytes_readonly, %llu may use "
"%llu total\n", (unsigned long long)bytes,
(unsigned long long)data_sinfo->bytes_delalloc,
(unsigned long long)data_sinfo->bytes_used,
@@ -4128,6 +4127,7 @@ struct extent_buffer *btrfs_alloc_free_block(struct btrfs_trans_handle *trans,
return buf;
}
+#if 0
int btrfs_drop_leaf_ref(struct btrfs_trans_handle *trans,
struct btrfs_root *root, struct extent_buffer *leaf)
{
@@ -4171,8 +4171,6 @@ int btrfs_drop_leaf_ref(struct btrfs_trans_handle *trans,
return 0;
}
-#if 0
-
static noinline int cache_drop_leaf_ref(struct btrfs_trans_handle *trans,
struct btrfs_root *root,
struct btrfs_leaf_ref *ref)
@@ -4553,262 +4551,471 @@ out:
}
#endif
+struct walk_control {
+ u64 refs[BTRFS_MAX_LEVEL];
+ u64 flags[BTRFS_MAX_LEVEL];
+ struct btrfs_key update_progress;
+ int stage;
+ int level;
+ int shared_level;
+ int update_ref;
+ int keep_locks;
+};
+
+#define DROP_REFERENCE 1
+#define UPDATE_BACKREF 2
+
/*
- * helper function for drop_subtree, this function is similar to
- * walk_down_tree. The main difference is that it checks reference
- * counts while tree blocks are locked.
+ * hepler to process tree block while walking down the tree.
+ *
+ * when wc->stage == DROP_REFERENCE, this function checks
+ * reference count of the block. if the block is shared and
+ * we need update back refs for the subtree rooted at the
+ * block, this function changes wc->stage to UPDATE_BACKREF
+ *
+ * when wc->stage == UPDATE_BACKREF, this function updates
+ * back refs for pointers in the block.
+ *
+ * NOTE: return value 1 means we should stop walking down.
*/
-static noinline int walk_down_tree(struct btrfs_trans_handle *trans,
+static noinline int walk_down_proc(struct btrfs_trans_handle *trans,
struct btrfs_root *root,
- struct btrfs_path *path, int *level)
+ struct btrfs_path *path,
+ struct walk_control *wc)
{
- struct extent_buffer *next;
- struct extent_buffer *cur;
- struct extent_buffer *parent;
- u64 bytenr;
- u64 ptr_gen;
- u64 refs;
- u64 flags;
- u32 blocksize;
+ int level = wc->level;
+ struct extent_buffer *eb = path->nodes[level];
+ struct btrfs_key key;
+ u64 flag = BTRFS_BLOCK_FLAG_FULL_BACKREF;
int ret;
- cur = path->nodes[*level];
- ret = btrfs_lookup_extent_info(trans, root, cur->start, cur->len,
- &refs, &flags);
- BUG_ON(ret);
- if (refs > 1)
- goto out;
+ if (wc->stage == UPDATE_BACKREF &&
+ btrfs_header_owner(eb) != root->root_key.objectid)
+ return 1;
- BUG_ON(!(flags & BTRFS_BLOCK_FLAG_FULL_BACKREF));
+ /*
+ * when reference count of tree block is 1, it won't increase
+ * again. once full backref flag is set, we never clear it.
+ */
+ if ((wc->stage == DROP_REFERENCE && wc->refs[level] != 1) ||
+ (wc->stage == UPDATE_BACKREF && !(wc->flags[level] & flag))) {
+ BUG_ON(!path->locks[level]);
+ ret = btrfs_lookup_extent_info(trans, root,
+ eb->start, eb->len,
+ &wc->refs[level],
+ &wc->flags[level]);
+ BUG_ON(ret);
+ BUG_ON(wc->refs[level] == 0);
+ }
- while (*level >= 0) {
- cur = path->nodes[*level];
- if (*level == 0) {
- ret = btrfs_drop_leaf_ref(trans, root, cur);
- BUG_ON(ret);
- clean_tree_block(trans, root, cur);
- break;
- }
- if (path->slots[*level] >= btrfs_header_nritems(cur)) {
- clean_tree_block(trans, root, cur);
- break;
+ if (wc->stage == DROP_REFERENCE &&
+ wc->update_ref && wc->refs[level] > 1) {
+ BUG_ON(eb == root->node);
+ BUG_ON(path->slots[level] > 0);
+ if (level == 0)
+ btrfs_item_key_to_cpu(eb, &key, path->slots[level]);
+ else
+ btrfs_node_key_to_cpu(eb, &key, path->slots[level]);
+ if (btrfs_header_owner(eb) == root->root_key.objectid &&
+ btrfs_comp_cpu_keys(&key, &wc->update_progress) >= 0) {
+ wc->stage = UPDATE_BACKREF;
+ wc->shared_level = level;
}
+ }
- bytenr = btrfs_node_blockptr(cur, path->slots[*level]);
- blocksize = btrfs_level_size(root, *level - 1);
- ptr_gen = btrfs_node_ptr_generation(cur, path->slots[*level]);
+ if (wc->stage == DROP_REFERENCE) {
+ if (wc->refs[level] > 1)
+ return 1;
- next = read_tree_block(root, bytenr, blocksize, ptr_gen);
- btrfs_tree_lock(next);
- btrfs_set_lock_blocking(next);
+ if (path->locks[level] && !wc->keep_locks) {
+ btrfs_tree_unlock(eb);
+ path->locks[level] = 0;
+ }
+ return 0;
+ }
- ret = btrfs_lookup_extent_info(trans, root, bytenr, blocksize,
- &refs, &flags);
+ /* wc->stage == UPDATE_BACKREF */
+ if (!(wc->flags[level] & flag)) {
+ BUG_ON(!path->locks[level]);
+ ret = btrfs_inc_ref(trans, root, eb, 1);
BUG_ON(ret);
- if (refs > 1) {
- parent = path->nodes[*level];
- ret = btrfs_free_extent(trans, root, bytenr,
- blocksize, parent->start,
- btrfs_header_owner(parent),
- *level - 1, 0);
+ ret = btrfs_dec_ref(trans, root, eb, 0);
+ BUG_ON(ret);
+ ret = btrfs_set_disk_extent_flags(trans, root, eb->start,
+ eb->len, flag, 0);
+ BUG_ON(ret);
+ wc->flags[level] |= flag;
+ }
+
+ /*
+ * the block is shared by multiple trees, so it's not good to
+ * keep the tree lock
+ */
+ if (path->locks[level] && level > 0) {
+ btrfs_tree_unlock(eb);
+ path->locks[level] = 0;
+ }
+ return 0;
+}
+
+/*
+ * hepler to process tree block while walking up the tree.
+ *
+ * when wc->stage == DROP_REFERENCE, this function drops
+ * reference count on the block.
+ *
+ * when wc->stage == UPDATE_BACKREF, this function changes
+ * wc->stage back to DROP_REFERENCE if we changed wc->stage
+ * to UPDATE_BACKREF previously while processing the block.
+ *
+ * NOTE: return value 1 means we should stop walking up.
+ */
+static noinline int walk_up_proc(struct btrfs_trans_handle *trans,
+ struct btrfs_root *root,
+ struct btrfs_path *path,
+ struct walk_control *wc)
+{
+ int ret = 0;
+ int level = wc->level;
+ struct extent_buffer *eb = path->nodes[level];
+ u64 parent = 0;
+
+ if (wc->stage == UPDATE_BACKREF) {
+ BUG_ON(wc->shared_level < level);
+ if (level < wc->shared_level)
+ goto out;
+
+ BUG_ON(wc->refs[level] <= 1);
+ ret = find_next_key(path, level + 1, &wc->update_progress);
+ if (ret > 0)
+ wc->update_ref = 0;
+
+ wc->stage = DROP_REFERENCE;
+ wc->shared_level = -1;
+ path->slots[level] = 0;
+
+ /*
+ * check reference count again if the block isn't locked.
+ * we should start walking down the tree again if reference
+ * count is one.
+ */
+ if (!path->locks[level]) {
+ BUG_ON(level == 0);
+ btrfs_tree_lock(eb);
+ btrfs_set_lock_blocking(eb);
+ path->locks[level] = 1;
+
+ ret = btrfs_lookup_extent_info(trans, root,
+ eb->start, eb->len,
+ &wc->refs[level],
+ &wc->flags[level]);
BUG_ON(ret);
- path->slots[*level]++;
- btrfs_tree_unlock(next);
- free_extent_buffer(next);
- continue;
+ BUG_ON(wc->refs[level] == 0);
+ if (wc->refs[level] == 1) {
+ btrfs_tree_unlock(eb);
+ path->locks[level] = 0;
+ return 1;
+ }
+ } else {
+ BUG_ON(level != 0);
}
+ }
- BUG_ON(!(flags & BTRFS_BLOCK_FLAG_FULL_BACKREF));
+ /* wc->stage == DROP_REFERENCE */
+ BUG_ON(wc->refs[level] > 1 && !path->locks[level]);
- *level = btrfs_header_level(next);
- path->nodes[*level] = next;
- path->slots[*level] = 0;
- path->locks[*level] = 1;
- cond_resched();
+ if (wc->refs[level] == 1) {
+ if (level == 0) {
+ if (wc->flags[level] & BTRFS_BLOCK_FLAG_FULL_BACKREF)
+ ret = btrfs_dec_ref(trans, root, eb, 1);
+ else
+ ret = btrfs_dec_ref(trans, root, eb, 0);
+ BUG_ON(ret);
+ }
+ /* make block locked assertion in clean_tree_block happy */
+ if (!path->locks[level] &&
+ btrfs_header_generation(eb) == trans->transid) {
+ btrfs_tree_lock(eb);
+ btrfs_set_lock_blocking(eb);
+ path->locks[level] = 1;
+ }
+ clean_tree_block(trans, root, eb);
+ }
+
+ if (eb == root->node) {
+ if (wc->flags[level] & BTRFS_BLOCK_FLAG_FULL_BACKREF)
+ parent = eb->start;
+ else
+ BUG_ON(root->root_key.objectid !=
+ btrfs_header_owner(eb));
+ } else {
+ if (wc->flags[level + 1] & BTRFS_BLOCK_FLAG_FULL_BACKREF)
+ parent = path->nodes[level + 1]->start;
+ else
+ BUG_ON(root->root_key.objectid !=
+ btrfs_header_owner(path->nodes[level + 1]));
}
-out:
- if (path->nodes[*level] == root->node)
- parent = path->nodes[*level];
- else
- parent = path->nodes[*level + 1];
- bytenr = path->nodes[*level]->start;
- blocksize = path->nodes[*level]->len;
- ret = btrfs_free_extent(trans, root, bytenr, blocksize, parent->start,
- btrfs_header_owner(parent), *level, 0);
+ ret = btrfs_free_extent(trans, root, eb->start, eb->len, parent,
+ root->root_key.objectid, level, 0);
BUG_ON(ret);
+out:
+ wc->refs[level] = 0;
+ wc->flags[level] = 0;
+ return ret;
+}
+
+static noinline int walk_down_tree(struct btrfs_trans_handle *trans,
+ struct btrfs_root *root,
+ struct btrfs_path *path,
+ struct walk_control *wc)
+{
+ struct extent_buffer *next;
+ struct extent_buffer *cur;
+ u64 bytenr;
+ u64 ptr_gen;
+ u32 blocksize;
+ int level = wc->level;
+ int ret;
+
+ while (level >= 0) {
+ cur = path->nodes[level];
+ BUG_ON(path->slots[level] >= btrfs_header_nritems(cur));
- if (path->locks[*level]) {
- btrfs_tree_unlock(path->nodes[*level]);
- path->locks[*level] = 0;
+ ret = walk_down_proc(trans, root, path, wc);
+ if (ret > 0)
+ break;
+
+ if (level == 0)
+ break;
+
+ bytenr = btrfs_node_blockptr(cur, path->slots[level]);
+ blocksize = btrfs_level_size(root, level - 1);
+ ptr_gen = btrfs_node_ptr_generation(cur, path->slots[level]);
+
+ next = read_tree_block(root, bytenr, blocksize, ptr_gen);
+ btrfs_tree_lock(next);
+ btrfs_set_lock_blocking(next);
+
+ level--;
+ BUG_ON(level != btrfs_header_level(next));
+ path->nodes[level] = next;
+ path->slots[level] = 0;
+ path->locks[level] = 1;
+ wc->level = level;
}
- free_extent_buffer(path->nodes[*level]);
- path->nodes[*level] = NULL;
- *level += 1;
- cond_resched();
return 0;
}
-/*
- * helper for dropping snapshots. This walks back up the tree in the path
- * to find the first node higher up where we haven't yet gone through
- * all the slots
- */
static noinline int walk_up_tree(struct btrfs_trans_handle *trans,
struct btrfs_root *root,
struct btrfs_path *path,
- int *level, int max_level)
+ struct walk_control *wc, int max_level)
{
- struct btrfs_root_item *root_item = &root->root_item;
- int i;
- int slot;
+ int level = wc->level;
int ret;
- for (i = *level; i < max_level && path->nodes[i]; i++) {
- slot = path->slots[i];
- if (slot + 1 < btrfs_header_nritems(path->nodes[i])) {
- /*
- * there is more work to do in this level.
- * Update the drop_progress marker to reflect
- * the work we've done so far, and then bump
- * the slot number
- */
- path->slots[i]++;
- WARN_ON(*level == 0);
- if (max_level == BTRFS_MAX_LEVEL) {
- btrfs_node_key(path->nodes[i],
- &root_item->drop_progress,
- path->slots[i]);
- root_item->drop_level = i;
- }
- *level = i;
+ path->slots[level] = btrfs_header_nritems(path->nodes[level]);
+ while (level < max_level && path->nodes[level]) {
+ wc->level = level;
+ if (path->slots[level] + 1 <
+ btrfs_header_nritems(path->nodes[level])) {
+ path->slots[level]++;
return 0;
} else {
- struct extent_buffer *parent;
-
- /*
- * this whole node is done, free our reference
- * on it and go up one level
- */
- if (path->nodes[*level] == root->node)
- parent = path->nodes[*level];
- else
- parent = path->nodes[*level + 1];
+ ret = walk_up_proc(trans, root, path, wc);
+ if (ret > 0)
+ return 0;
- clean_tree_block(trans, root, path->nodes[i]);
- ret = btrfs_free_extent(trans, root,
- path->nodes[i]->start,
- path->nodes[i]->len,
- parent->start,
- btrfs_header_owner(parent),
- *level, 0);
- BUG_ON(ret);
- if (path->locks[*level]) {
- btrfs_tree_unlock(path->nodes[i]);
- path->locks[i] = 0;
+ if (path->locks[level]) {
+ btrfs_tree_unlock(path->nodes[level]);
+ path->locks[level] = 0;
}
- free_extent_buffer(path->nodes[i]);
- path->nodes[i] = NULL;
- *level = i + 1;
+ free_extent_buffer(path->nodes[level]);
+ path->nodes[level] = NULL;
+ level++;
}
}
return 1;
}
/*
- * drop the reference count on the tree rooted at 'snap'. This traverses
- * the tree freeing any blocks that have a ref count of zero after being
- * decremented.
+ * drop a subvolume tree.
+ *
+ * this function traverses the tree freeing any blocks that only
+ * referenced by the tree.
+ *
+ * when a shared tree block is found. this function decreases its
+ * reference count by one. if update_ref is true, this function
+ * also make sure backrefs for the shared block and all lower level
+ * blocks are properly updated.
*/
-int btrfs_drop_snapshot(struct btrfs_trans_handle *trans, struct btrfs_root
- *root)
+int btrfs_drop_snapshot(struct btrfs_root *root, int update_ref)
{
- int ret = 0;
- int wret;
- int level;
struct btrfs_path *path;
- int update_count;
+ struct btrfs_trans_handle *trans;
+ struct btrfs_root *tree_root = root->fs_info->tree_root;
struct btrfs_root_item *root_item = &root->root_item;
+ struct walk_control *wc;
+ struct btrfs_key key;
+ int err = 0;
+ int ret;
+ int level;
path = btrfs_alloc_path();
BUG_ON(!path);
- level = btrfs_header_level(root->node);
+ wc = kzalloc(sizeof(*wc), GFP_NOFS);
+ BUG_ON(!wc);
+
+ trans = btrfs_start_transaction(tree_root, 1);
+
if (btrfs_disk_key_objectid(&root_item->drop_progress) == 0) {
+ level = btrfs_header_level(root->node);
path->nodes[level] = btrfs_lock_root_node(root);
btrfs_set_lock_blocking(path->nodes[level]);
path->slots[level] = 0;
path->locks[level] = 1;
+ memset(&wc->update_progress, 0,
+ sizeof(wc->update_progress));
} else {
- struct btrfs_key key;
- struct btrfs_disk_key found_key;
- struct extent_buffer *node;
-
btrfs_disk_key_to_cpu(&key, &root_item->drop_progress);
+ memcpy(&wc->update_progress, &key,
+ sizeof(wc->update_progress));
+
level = root_item->drop_level;
+ BUG_ON(level == 0);
path->lowest_level = level;
- wret = btrfs_search_slot(NULL, root, &key, path, 0, 0);
- if (wret < 0) {
- ret = wret;
+ ret = btrfs_search_slot(NULL, root, &key, path, 0, 0);
+ path->lowest_level = 0;
+ if (ret < 0) {
+ err = ret;
goto out;
}
- node = path->nodes[level];
- btrfs_node_key(node, &found_key, path->slots[level]);
- WARN_ON(memcmp(&found_key, &root_item->drop_progress,
- sizeof(found_key)));
+ btrfs_node_key_to_cpu(path->nodes[level], &key,
+ path->slots[level]);
+ WARN_ON(memcmp(&key, &wc->update_progress, sizeof(key)));
+
/*
* unlock our path, this is safe because only this
* function is allowed to delete this snapshot
*/
btrfs_unlock_up_safe(path, 0);
+
+ level = btrfs_header_level(root->node);
+ while (1) {
+ btrfs_tree_lock(path->nodes[level]);
+ btrfs_set_lock_blocking(path->nodes[level]);
+
+ ret = btrfs_lookup_extent_info(trans, root,
+ path->nodes[level]->start,
+ path->nodes[level]->len,
+ &wc->refs[level],
+ &wc->flags[level]);
+ BUG_ON(ret);
+ BUG_ON(wc->refs[level] == 0);
+
+ if (level == root_item->drop_level)
+ break;
+
+ btrfs_tree_unlock(path->nodes[level]);
+ WARN_ON(wc->refs[level] != 1);
+ level--;
+ }
}
+
+ wc->level = level;
+ wc->shared_level = -1;
+ wc->stage = DROP_REFERENCE;
+ wc->update_ref = update_ref;
+ wc->keep_locks = 0;
+
while (1) {
- unsigned long update;
- wret = walk_down_tree(trans, root, path, &level);
- if (wret > 0)
+ ret = walk_down_tree(trans, root, path, wc);
+ if (ret < 0) {
+ err = ret;
break;
- if (wret < 0)
- ret = wret;
+ }
- wret = walk_up_tree(trans, root, path, &level,
- BTRFS_MAX_LEVEL);
- if (wret > 0)
+ ret = walk_up_tree(trans, root, path, wc, BTRFS_MAX_LEVEL);
+ if (ret < 0) {
+ err = ret;
break;
- if (wret < 0)
- ret = wret;
- if (trans->transaction->in_commit ||
- trans->transaction->delayed_refs.flushing) {
- ret = -EAGAIN;
+ }
+
+ if (ret > 0) {
+ BUG_ON(wc->stage != DROP_REFERENCE);
break;
}
- for (update_count = 0; update_count < 16; update_count++) {
+
+ if (wc->stage == DROP_REFERENCE) {
+ level = wc->level;
+ btrfs_node_key(path->nodes[level],
+ &root_item->drop_progress,
+ path->slots[level]);
+ root_item->drop_level = level;
+ }
+
+ BUG_ON(wc->level == 0);
+ if (trans->transaction->in_commit ||
+ trans->transaction->delayed_refs.flushing) {
+ ret = btrfs_update_root(trans, tree_root,
+ &root->root_key,
+ root_item);
+ BUG_ON(ret);
+
+ btrfs_end_transaction(trans, tree_root);
+ trans = btrfs_start_transaction(tree_root, 1);
+ } else {
+ unsigned long update;
update = trans->delayed_ref_updates;
trans->delayed_ref_updates = 0;
if (update)
- btrfs_run_delayed_refs(trans, root, update);
- else
- break;
+ btrfs_run_delayed_refs(trans, tree_root,
+ update);
}
}
+ btrfs_release_path(root, path);
+ BUG_ON(err);
+
+ ret = btrfs_del_root(trans, tree_root, &root->root_key);
+ BUG_ON(ret);
+
+ free_extent_buffer(root->node);
+ free_extent_buffer(root->commit_root);
+ kfree(root);
out:
+ btrfs_end_transaction(trans, tree_root);
+ kfree(wc);
btrfs_free_path(path);
- return ret;
+ return err;
}
+/*
+ * drop subtree rooted at tree block 'node'.
+ *
+ * NOTE: this function will unlock and release tree block 'node'
+ */
int btrfs_drop_subtree(struct btrfs_trans_handle *trans,
struct btrfs_root *root,
struct extent_buffer *node,
struct extent_buffer *parent)
{
struct btrfs_path *path;
+ struct walk_control *wc;
int level;
int parent_level;
int ret = 0;
int wret;
+ BUG_ON(root->root_key.objectid != BTRFS_TREE_RELOC_OBJECTID);
+
path = btrfs_alloc_path();
BUG_ON(!path);
+ wc = kzalloc(sizeof(*wc), GFP_NOFS);
+ BUG_ON(!wc);
+
btrfs_assert_tree_locked(parent);
parent_level = btrfs_header_level(parent);
extent_buffer_get(parent);
@@ -4817,24 +5024,33 @@ int btrfs_drop_subtree(struct btrfs_trans_handle *trans,
btrfs_assert_tree_locked(node);
level = btrfs_header_level(node);
- extent_buffer_get(node);
path->nodes[level] = node;
path->slots[level] = 0;
+ path->locks[level] = 1;
+
+ wc->refs[parent_level] = 1;
+ wc->flags[parent_level] = BTRFS_BLOCK_FLAG_FULL_BACKREF;
+ wc->level = level;
+ wc->shared_level = -1;
+ wc->stage = DROP_REFERENCE;
+ wc->update_ref = 0;
+ wc->keep_locks = 1;
while (1) {
- wret = walk_down_tree(trans, root, path, &level);
- if (wret < 0)
+ wret = walk_down_tree(trans, root, path, wc);
+ if (wret < 0) {
ret = wret;
- if (wret != 0)
break;
+ }
- wret = walk_up_tree(trans, root, path, &level, parent_level);
+ wret = walk_up_tree(trans, root, path, wc, parent_level);
if (wret < 0)
ret = wret;
if (wret != 0)
break;
}
+ kfree(wc);
btrfs_free_path(path);
return ret;
}
diff --git a/fs/btrfs/file.c b/fs/btrfs/file.c
index 126477eaecf..7c3cd248d8d 100644
--- a/fs/btrfs/file.c
+++ b/fs/btrfs/file.c
@@ -151,7 +151,10 @@ static noinline int dirty_and_release_pages(struct btrfs_trans_handle *trans,
}
if (end_pos > isize) {
i_size_write(inode, end_pos);
- btrfs_update_inode(trans, root, inode);
+ /* we've only changed i_size in ram, and we haven't updated
+ * the disk i_size. There is no need to log the inode
+ * at this time.
+ */
}
err = btrfs_end_transaction(trans, root);
out_unlock:
diff --git a/fs/btrfs/inode.c b/fs/btrfs/inode.c
index 8612b3a0981..7ffa3d34ea1 100644
--- a/fs/btrfs/inode.c
+++ b/fs/btrfs/inode.c
@@ -2122,10 +2122,8 @@ static void btrfs_read_locked_inode(struct inode *inode)
* any xattrs or acls
*/
maybe_acls = acls_after_inode_item(leaf, path->slots[0], inode->i_ino);
- if (!maybe_acls) {
- BTRFS_I(inode)->i_acl = NULL;
- BTRFS_I(inode)->i_default_acl = NULL;
- }
+ if (!maybe_acls)
+ cache_no_acl(inode);
BTRFS_I(inode)->block_group = btrfs_find_block_group(root, 0,
alloc_group_block, 0);
@@ -3141,9 +3139,6 @@ static noinline void init_btrfs_i(struct inode *inode)
{
struct btrfs_inode *bi = BTRFS_I(inode);
- bi->i_acl = BTRFS_ACL_NOT_CACHED;
- bi->i_default_acl = BTRFS_ACL_NOT_CACHED;
-
bi->generation = 0;
bi->sequence = 0;
bi->last_trans = 0;
@@ -3585,12 +3580,6 @@ static struct inode *btrfs_new_inode(struct btrfs_trans_handle *trans,
owner = 1;
BTRFS_I(inode)->block_group =
btrfs_find_block_group(root, 0, alloc_hint, owner);
- if ((mode & S_IFREG)) {
- if (btrfs_test_opt(root, NODATASUM))
- BTRFS_I(inode)->flags |= BTRFS_INODE_NODATASUM;
- if (btrfs_test_opt(root, NODATACOW))
- BTRFS_I(inode)->flags |= BTRFS_INODE_NODATACOW;
- }
key[0].objectid = objectid;
btrfs_set_key_type(&key[0], BTRFS_INODE_ITEM_KEY);
@@ -3645,6 +3634,13 @@ static struct inode *btrfs_new_inode(struct btrfs_trans_handle *trans,
btrfs_inherit_iflags(inode, dir);
+ if ((mode & S_IFREG)) {
+ if (btrfs_test_opt(root, NODATASUM))
+ BTRFS_I(inode)->flags |= BTRFS_INODE_NODATASUM;
+ if (btrfs_test_opt(root, NODATACOW))
+ BTRFS_I(inode)->flags |= BTRFS_INODE_NODATACOW;
+ }
+
insert_inode_hash(inode);
inode_tree_add(inode);
return inode;
@@ -4640,8 +4636,6 @@ struct inode *btrfs_alloc_inode(struct super_block *sb)
ei->last_trans = 0;
ei->logged_trans = 0;
btrfs_ordered_inode_tree_init(&ei->ordered_tree);
- ei->i_acl = BTRFS_ACL_NOT_CACHED;
- ei->i_default_acl = BTRFS_ACL_NOT_CACHED;
INIT_LIST_HEAD(&ei->i_orphan);
INIT_LIST_HEAD(&ei->ordered_operations);
return &ei->vfs_inode;
@@ -4655,13 +4649,6 @@ void btrfs_destroy_inode(struct inode *inode)
WARN_ON(!list_empty(&inode->i_dentry));
WARN_ON(inode->i_data.nrpages);
- if (BTRFS_I(inode)->i_acl &&
- BTRFS_I(inode)->i_acl != BTRFS_ACL_NOT_CACHED)
- posix_acl_release(BTRFS_I(inode)->i_acl);
- if (BTRFS_I(inode)->i_default_acl &&
- BTRFS_I(inode)->i_default_acl != BTRFS_ACL_NOT_CACHED)
- posix_acl_release(BTRFS_I(inode)->i_default_acl);
-
/*
* Make sure we're properly removed from the ordered operation
* lists.
@@ -5096,6 +5083,7 @@ static long btrfs_fallocate(struct inode *inode, int mode,
u64 mask = BTRFS_I(inode)->root->sectorsize - 1;
struct extent_map *em;
struct btrfs_trans_handle *trans;
+ struct btrfs_root *root;
int ret;
alloc_start = offset & ~mask;
@@ -5114,6 +5102,13 @@ static long btrfs_fallocate(struct inode *inode, int mode,
goto out;
}
+ root = BTRFS_I(inode)->root;
+
+ ret = btrfs_check_data_free_space(root, inode,
+ alloc_end - alloc_start);
+ if (ret)
+ goto out;
+
locked_end = alloc_end - 1;
while (1) {
struct btrfs_ordered_extent *ordered;
@@ -5121,7 +5116,7 @@ static long btrfs_fallocate(struct inode *inode, int mode,
trans = btrfs_start_transaction(BTRFS_I(inode)->root, 1);
if (!trans) {
ret = -EIO;
- goto out;
+ goto out_free;
}
/* the extent lock is ordered inside the running
@@ -5182,6 +5177,8 @@ static long btrfs_fallocate(struct inode *inode, int mode,
GFP_NOFS);
btrfs_end_transaction(trans, BTRFS_I(inode)->root);
+out_free:
+ btrfs_free_reserved_data_space(root, inode, alloc_end - alloc_start);
out:
mutex_unlock(&inode->i_mutex);
return ret;
diff --git a/fs/btrfs/ioctl.c b/fs/btrfs/ioctl.c
index eff18f5b536..9f4db848db1 100644
--- a/fs/btrfs/ioctl.c
+++ b/fs/btrfs/ioctl.c
@@ -1028,7 +1028,8 @@ static long btrfs_ioctl_clone(struct file *file, unsigned long srcfd,
struct btrfs_file_extent_item);
comp = btrfs_file_extent_compression(leaf, extent);
type = btrfs_file_extent_type(leaf, extent);
- if (type == BTRFS_FILE_EXTENT_REG) {
+ if (type == BTRFS_FILE_EXTENT_REG ||
+ type == BTRFS_FILE_EXTENT_PREALLOC) {
disko = btrfs_file_extent_disk_bytenr(leaf,
extent);
diskl = btrfs_file_extent_disk_num_bytes(leaf,
@@ -1051,7 +1052,8 @@ static long btrfs_ioctl_clone(struct file *file, unsigned long srcfd,
new_key.objectid = inode->i_ino;
new_key.offset = key.offset + destoff - off;
- if (type == BTRFS_FILE_EXTENT_REG) {
+ if (type == BTRFS_FILE_EXTENT_REG ||
+ type == BTRFS_FILE_EXTENT_PREALLOC) {
ret = btrfs_insert_empty_item(trans, root, path,
&new_key, size);
if (ret)
diff --git a/fs/btrfs/relocation.c b/fs/btrfs/relocation.c
index b23dc209ae1..00839793477 100644
--- a/fs/btrfs/relocation.c
+++ b/fs/btrfs/relocation.c
@@ -1788,7 +1788,7 @@ static void merge_func(struct btrfs_work *work)
btrfs_end_transaction(trans, root);
}
- btrfs_drop_dead_root(reloc_root);
+ btrfs_drop_snapshot(reloc_root, 0);
if (atomic_dec_and_test(async->num_pending))
complete(async->done);
@@ -2075,9 +2075,6 @@ static int do_relocation(struct btrfs_trans_handle *trans,
ret = btrfs_drop_subtree(trans, root, eb, upper->eb);
BUG_ON(ret);
-
- btrfs_tree_unlock(eb);
- free_extent_buffer(eb);
}
if (!lowest) {
btrfs_tree_unlock(upper->eb);
diff --git a/fs/btrfs/transaction.c b/fs/btrfs/transaction.c
index 4e83457ea25..2dbf1c1f56e 100644
--- a/fs/btrfs/transaction.c
+++ b/fs/btrfs/transaction.c
@@ -593,6 +593,7 @@ int btrfs_defrag_root(struct btrfs_root *root, int cacheonly)
return 0;
}
+#if 0
/*
* when dropping snapshots, we generate a ton of delayed refs, and it makes
* sense not to join the transaction while it is trying to flush the current
@@ -681,6 +682,7 @@ int btrfs_drop_dead_root(struct btrfs_root *root)
btrfs_btree_balance_dirty(tree_root, nr);
return ret;
}
+#endif
/*
* new snapshots need to be created at a very specific time in the
@@ -1081,7 +1083,7 @@ int btrfs_clean_old_snapshots(struct btrfs_root *root)
while (!list_empty(&list)) {
root = list_entry(list.next, struct btrfs_root, root_list);
list_del_init(&root->root_list);
- btrfs_drop_dead_root(root);
+ btrfs_drop_snapshot(root, 0);
}
return 0;
}
generated by cgit v1.2.3-70-g09d2 (git 2.46.0) at 2024-11-18 09:04:59 +0000