Btrfs: Integrate metadata reservation with start_transaction
Besides simplify the code, this change makes sure all metadata
reservation for normal metadata operations are released after
committing transaction.
Changes since V1:
Add code that check if unlink and rmdir will free space.
Add ENOSPC handling for clone ioctl.
Signed-off-by: Yan Zheng <zheng.yan@oracle.com>
Signed-off-by: Chris Mason <chris.mason@oracle.com>
diff --git a/fs/btrfs/ctree.h b/fs/btrfs/ctree.h
index 7d24796..e0aa9fb 100644
--- a/fs/btrfs/ctree.h
+++ b/fs/btrfs/ctree.h
@@ -34,6 +34,7 @@
struct btrfs_trans_handle;
struct btrfs_transaction;
+struct btrfs_pending_snapshot;
extern struct kmem_cache *btrfs_trans_handle_cachep;
extern struct kmem_cache *btrfs_transaction_cachep;
extern struct kmem_cache *btrfs_bit_radix_cachep;
@@ -970,6 +971,7 @@
int do_barriers;
int closing;
int log_root_recovering;
+ int enospc_unlink;
u64 total_pinned;
@@ -1995,6 +1997,9 @@
int btrfs_run_delayed_refs(struct btrfs_trans_handle *trans,
struct btrfs_root *root, unsigned long count);
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);
int btrfs_pin_extent(struct btrfs_root *root,
u64 bytenr, u64 num, int reserved);
int btrfs_drop_leaf_ref(struct btrfs_trans_handle *trans,
@@ -2075,8 +2080,6 @@
void btrfs_set_inode_space_info(struct btrfs_root *root, struct inode *ionde);
void btrfs_clear_space_info_full(struct btrfs_fs_info *info);
-int btrfs_reserve_metadata_space(struct btrfs_root *root, int num_items);
-int btrfs_unreserve_metadata_space(struct btrfs_root *root, int num_items);
int btrfs_unreserve_metadata_for_delalloc(struct btrfs_root *root,
struct inode *inode, int num_items);
int btrfs_reserve_metadata_for_delalloc(struct btrfs_root *root,
@@ -2089,6 +2092,13 @@
u64 bytes);
void btrfs_delalloc_free_space(struct btrfs_root *root, struct inode *inode,
u64 bytes);
+int btrfs_trans_reserve_metadata(struct btrfs_trans_handle *trans,
+ struct btrfs_root *root,
+ int num_items, int *retries);
+void btrfs_trans_release_metadata(struct btrfs_trans_handle *trans,
+ struct btrfs_root *root);
+int btrfs_snap_reserve_metadata(struct btrfs_trans_handle *trans,
+ struct btrfs_pending_snapshot *pending);
void btrfs_init_block_rsv(struct btrfs_block_rsv *rsv);
struct btrfs_block_rsv *btrfs_alloc_block_rsv(struct btrfs_root *root);
void btrfs_free_block_rsv(struct btrfs_root *root,
@@ -2296,6 +2306,12 @@
struct btrfs_root *root,
const char *name, int name_len,
u64 inode_objectid, u64 ref_objectid, u64 *index);
+struct btrfs_inode_ref *
+btrfs_lookup_inode_ref(struct btrfs_trans_handle *trans,
+ struct btrfs_root *root,
+ struct btrfs_path *path,
+ const char *name, int name_len,
+ u64 inode_objectid, u64 ref_objectid, int mod);
int btrfs_insert_empty_inode(struct btrfs_trans_handle *trans,
struct btrfs_root *root,
struct btrfs_path *path, u64 objectid);
diff --git a/fs/btrfs/delayed-ref.c b/fs/btrfs/delayed-ref.c
index 902ce50..e807b14 100644
--- a/fs/btrfs/delayed-ref.c
+++ b/fs/btrfs/delayed-ref.c
@@ -319,107 +319,6 @@
}
/*
- * helper function to lookup reference count and flags of extent.
- *
- * 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
- * node may also store the extent flags to set. This way you can check
- * to see what the reference count and extent flags would be if all of
- * the delayed refs are not processed.
- */
-int btrfs_lookup_extent_info(struct btrfs_trans_handle *trans,
- struct btrfs_root *root, u64 bytenr,
- u64 num_bytes, u64 *refs, u64 *flags)
-{
- struct btrfs_delayed_ref_node *ref;
- struct btrfs_delayed_ref_head *head;
- struct btrfs_delayed_ref_root *delayed_refs;
- struct btrfs_path *path;
- struct btrfs_extent_item *ei;
- struct extent_buffer *leaf;
- struct btrfs_key key;
- u32 item_size;
- u64 num_refs;
- u64 extent_flags;
- int ret;
-
- path = btrfs_alloc_path();
- if (!path)
- return -ENOMEM;
-
- key.objectid = bytenr;
- key.type = BTRFS_EXTENT_ITEM_KEY;
- key.offset = num_bytes;
- delayed_refs = &trans->transaction->delayed_refs;
-again:
- ret = btrfs_search_slot(trans, root->fs_info->extent_root,
- &key, path, 0, 0);
- if (ret < 0)
- goto out;
-
- if (ret == 0) {
- leaf = path->nodes[0];
- item_size = btrfs_item_size_nr(leaf, path->slots[0]);
- if (item_size >= sizeof(*ei)) {
- ei = btrfs_item_ptr(leaf, path->slots[0],
- struct btrfs_extent_item);
- num_refs = btrfs_extent_refs(leaf, ei);
- extent_flags = btrfs_extent_flags(leaf, ei);
- } else {
-#ifdef BTRFS_COMPAT_EXTENT_TREE_V0
- struct btrfs_extent_item_v0 *ei0;
- BUG_ON(item_size != sizeof(*ei0));
- ei0 = btrfs_item_ptr(leaf, path->slots[0],
- struct btrfs_extent_item_v0);
- num_refs = btrfs_extent_refs_v0(leaf, ei0);
- /* FIXME: this isn't correct for data */
- extent_flags = BTRFS_BLOCK_FLAG_FULL_BACKREF;
-#else
- BUG();
-#endif
- }
- BUG_ON(num_refs == 0);
- } else {
- num_refs = 0;
- extent_flags = 0;
- ret = 0;
- }
-
- spin_lock(&delayed_refs->lock);
- ref = find_ref_head(&delayed_refs->root, bytenr, NULL);
- if (ref) {
- head = btrfs_delayed_node_to_head(ref);
- if (!mutex_trylock(&head->mutex)) {
- atomic_inc(&ref->refs);
- spin_unlock(&delayed_refs->lock);
-
- btrfs_release_path(root->fs_info->extent_root, path);
-
- mutex_lock(&head->mutex);
- mutex_unlock(&head->mutex);
- btrfs_put_delayed_ref(ref);
- goto again;
- }
- if (head->extent_op && head->extent_op->update_flags)
- extent_flags |= head->extent_op->flags_to_set;
- else
- BUG_ON(num_refs == 0);
-
- num_refs += ref->ref_mod;
- mutex_unlock(&head->mutex);
- }
- WARN_ON(num_refs == 0);
- if (refs)
- *refs = num_refs;
- if (flags)
- *flags = extent_flags;
-out:
- spin_unlock(&delayed_refs->lock);
- btrfs_free_path(path);
- return ret;
-}
-
-/*
* helper function to update an extent delayed ref in the
* rbtree. existing and update must both have the same
* bytenr and parent
diff --git a/fs/btrfs/delayed-ref.h b/fs/btrfs/delayed-ref.h
index f6fc67d..50e3cf9 100644
--- a/fs/btrfs/delayed-ref.h
+++ b/fs/btrfs/delayed-ref.h
@@ -167,9 +167,6 @@
struct btrfs_delayed_ref_head *
btrfs_find_delayed_ref_head(struct btrfs_trans_handle *trans, u64 bytenr);
int btrfs_delayed_ref_pending(struct btrfs_trans_handle *trans, u64 bytenr);
-int btrfs_lookup_extent_info(struct btrfs_trans_handle *trans,
- struct btrfs_root *root, u64 bytenr,
- u64 num_bytes, u64 *refs, u64 *flags);
int btrfs_update_delayed_ref(struct btrfs_trans_handle *trans,
u64 bytenr, u64 num_bytes, u64 orig_parent,
u64 parent, u64 orig_ref_root, u64 ref_root,
diff --git a/fs/btrfs/disk-io.c b/fs/btrfs/disk-io.c
index 574594c..054b447 100644
--- a/fs/btrfs/disk-io.c
+++ b/fs/btrfs/disk-io.c
@@ -1522,7 +1522,7 @@
goto sleep;
}
mutex_unlock(&root->fs_info->trans_mutex);
- trans = btrfs_start_transaction(root, 1);
+ trans = btrfs_join_transaction(root, 1);
ret = btrfs_commit_transaction(trans, root);
sleep:
@@ -2409,11 +2409,11 @@
down_write(&root->fs_info->cleanup_work_sem);
up_write(&root->fs_info->cleanup_work_sem);
- trans = btrfs_start_transaction(root, 1);
+ trans = btrfs_join_transaction(root, 1);
ret = btrfs_commit_transaction(trans, root);
BUG_ON(ret);
/* run commit again to drop the original snapshot */
- trans = btrfs_start_transaction(root, 1);
+ trans = btrfs_join_transaction(root, 1);
btrfs_commit_transaction(trans, root);
ret = btrfs_write_and_wait_transaction(NULL, root);
BUG_ON(ret);
diff --git a/fs/btrfs/extent-tree.c b/fs/btrfs/extent-tree.c
index 3367278..657df6e 100644
--- a/fs/btrfs/extent-tree.c
+++ b/fs/btrfs/extent-tree.c
@@ -616,6 +616,113 @@
}
/*
+ * helper function to lookup reference count and flags of extent.
+ *
+ * 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
+ * node may also store the extent flags to set. This way you can check
+ * to see what the reference count and extent flags would be if all of
+ * the delayed refs are not processed.
+ */
+int btrfs_lookup_extent_info(struct btrfs_trans_handle *trans,
+ struct btrfs_root *root, u64 bytenr,
+ u64 num_bytes, u64 *refs, u64 *flags)
+{
+ struct btrfs_delayed_ref_head *head;
+ struct btrfs_delayed_ref_root *delayed_refs;
+ struct btrfs_path *path;
+ struct btrfs_extent_item *ei;
+ struct extent_buffer *leaf;
+ struct btrfs_key key;
+ u32 item_size;
+ u64 num_refs;
+ u64 extent_flags;
+ int ret;
+
+ path = btrfs_alloc_path();
+ if (!path)
+ return -ENOMEM;
+
+ key.objectid = bytenr;
+ key.type = BTRFS_EXTENT_ITEM_KEY;
+ key.offset = num_bytes;
+ if (!trans) {
+ path->skip_locking = 1;
+ path->search_commit_root = 1;
+ }
+again:
+ ret = btrfs_search_slot(trans, root->fs_info->extent_root,
+ &key, path, 0, 0);
+ if (ret < 0)
+ goto out_free;
+
+ if (ret == 0) {
+ leaf = path->nodes[0];
+ item_size = btrfs_item_size_nr(leaf, path->slots[0]);
+ if (item_size >= sizeof(*ei)) {
+ ei = btrfs_item_ptr(leaf, path->slots[0],
+ struct btrfs_extent_item);
+ num_refs = btrfs_extent_refs(leaf, ei);
+ extent_flags = btrfs_extent_flags(leaf, ei);
+ } else {
+#ifdef BTRFS_COMPAT_EXTENT_TREE_V0
+ struct btrfs_extent_item_v0 *ei0;
+ BUG_ON(item_size != sizeof(*ei0));
+ ei0 = btrfs_item_ptr(leaf, path->slots[0],
+ struct btrfs_extent_item_v0);
+ num_refs = btrfs_extent_refs_v0(leaf, ei0);
+ /* FIXME: this isn't correct for data */
+ extent_flags = BTRFS_BLOCK_FLAG_FULL_BACKREF;
+#else
+ BUG();
+#endif
+ }
+ BUG_ON(num_refs == 0);
+ } else {
+ num_refs = 0;
+ extent_flags = 0;
+ ret = 0;
+ }
+
+ if (!trans)
+ goto out;
+
+ delayed_refs = &trans->transaction->delayed_refs;
+ spin_lock(&delayed_refs->lock);
+ head = btrfs_find_delayed_ref_head(trans, bytenr);
+ if (head) {
+ if (!mutex_trylock(&head->mutex)) {
+ atomic_inc(&head->node.refs);
+ spin_unlock(&delayed_refs->lock);
+
+ btrfs_release_path(root->fs_info->extent_root, path);
+
+ mutex_lock(&head->mutex);
+ mutex_unlock(&head->mutex);
+ btrfs_put_delayed_ref(&head->node);
+ goto again;
+ }
+ if (head->extent_op && head->extent_op->update_flags)
+ extent_flags |= head->extent_op->flags_to_set;
+ else
+ BUG_ON(num_refs == 0);
+
+ num_refs += head->node.ref_mod;
+ mutex_unlock(&head->mutex);
+ }
+ spin_unlock(&delayed_refs->lock);
+out:
+ WARN_ON(num_refs == 0);
+ if (refs)
+ *refs = num_refs;
+ if (flags)
+ *flags = extent_flags;
+out_free:
+ btrfs_free_path(path);
+ return ret;
+}
+
+/*
* Back reference rules. Back refs have three main goals:
*
* 1) differentiate between all holders of references to an extent so that
@@ -2949,113 +3056,6 @@
}
/*
- * unreserve num_items number of items worth of metadata space. This needs to
- * be paired with btrfs_reserve_metadata_space.
- *
- * NOTE: if you have the option, run this _AFTER_ you do a
- * btrfs_end_transaction, since btrfs_end_transaction will run delayed ref
- * oprations which will result in more used metadata, so we want to make sure we
- * can do that without issue.
- */
-int btrfs_unreserve_metadata_space(struct btrfs_root *root, int num_items)
-{
- struct btrfs_fs_info *info = root->fs_info;
- struct btrfs_space_info *meta_sinfo;
- u64 num_bytes;
- u64 alloc_target;
- bool bug = false;
-
- /* get the space info for where the metadata will live */
- alloc_target = btrfs_get_alloc_profile(root, 0);
- meta_sinfo = __find_space_info(info, alloc_target);
-
- num_bytes = calculate_bytes_needed(root, num_items);
-
- spin_lock(&meta_sinfo->lock);
- if (meta_sinfo->bytes_may_use < num_bytes) {
- bug = true;
- meta_sinfo->bytes_may_use = 0;
- } else {
- meta_sinfo->bytes_may_use -= num_bytes;
- }
- spin_unlock(&meta_sinfo->lock);
-
- BUG_ON(bug);
-
- return 0;
-}
-
-/*
- * Reserve some metadata space for use. We'll calculate the worste case number
- * of bytes that would be needed to modify num_items number of items. If we
- * have space, fantastic, if not, you get -ENOSPC. Please call
- * btrfs_unreserve_metadata_space when you are done for the _SAME_ number of
- * items you reserved, since whatever metadata you needed should have already
- * been allocated.
- *
- * This will commit the transaction to make more space if we don't have enough
- * metadata space. THe only time we don't do this is if we're reserving space
- * inside of a transaction, then we will just return -ENOSPC and it is the
- * callers responsibility to handle it properly.
- */
-int btrfs_reserve_metadata_space(struct btrfs_root *root, int num_items)
-{
- struct btrfs_fs_info *info = root->fs_info;
- struct btrfs_space_info *meta_sinfo;
- u64 num_bytes;
- u64 used;
- u64 alloc_target;
- int retries = 0;
-
- /* get the space info for where the metadata will live */
- alloc_target = btrfs_get_alloc_profile(root, 0);
- meta_sinfo = __find_space_info(info, alloc_target);
-
- num_bytes = calculate_bytes_needed(root, num_items);
-again:
- spin_lock(&meta_sinfo->lock);
-
- if (unlikely(!meta_sinfo->bytes_root))
- meta_sinfo->bytes_root = calculate_bytes_needed(root, 6);
-
- if (!retries)
- meta_sinfo->bytes_may_use += num_bytes;
-
- used = meta_sinfo->bytes_used + meta_sinfo->bytes_reserved +
- meta_sinfo->bytes_pinned + meta_sinfo->bytes_readonly +
- meta_sinfo->bytes_super + meta_sinfo->bytes_root +
- meta_sinfo->bytes_may_use + meta_sinfo->bytes_delalloc;
-
- if (used > meta_sinfo->total_bytes) {
- retries++;
- if (retries == 1) {
- if (maybe_allocate_chunk(NULL, root, meta_sinfo,
- num_bytes))
- goto again;
- retries++;
- } else {
- spin_unlock(&meta_sinfo->lock);
- }
-
- if (retries == 2) {
- shrink_delalloc(NULL, root, meta_sinfo, num_bytes);
- goto again;
- }
- spin_lock(&meta_sinfo->lock);
- meta_sinfo->bytes_may_use -= num_bytes;
- spin_unlock(&meta_sinfo->lock);
-
- dump_space_info(meta_sinfo, 0, 0);
- return -ENOSPC;
- }
-
- check_force_delalloc(meta_sinfo);
- spin_unlock(&meta_sinfo->lock);
-
- return 0;
-}
-
-/*
* This will check the space that the inode allocates from to make sure we have
* enough space for bytes.
*/
@@ -3095,9 +3095,9 @@
spin_unlock(&data_sinfo->lock);
alloc:
alloc_target = btrfs_get_alloc_profile(root, 1);
- trans = btrfs_start_transaction(root, 1);
- if (!trans)
- return -ENOMEM;
+ trans = btrfs_join_transaction(root, 1);
+ if (IS_ERR(trans))
+ return PTR_ERR(trans);
ret = do_chunk_alloc(trans, root->fs_info->extent_root,
bytes + 2 * 1024 * 1024,
@@ -3118,8 +3118,8 @@
if (!committed && !root->fs_info->open_ioctl_trans) {
committed = 1;
trans = btrfs_join_transaction(root, 1);
- if (!trans)
- return -ENOMEM;
+ if (IS_ERR(trans))
+ return PTR_ERR(trans);
ret = btrfs_commit_transaction(trans, root);
if (ret)
return ret;
@@ -3701,6 +3701,59 @@
fs_info->chunk_root->block_rsv = &fs_info->chunk_block_rsv;
}
+static u64 calc_trans_metadata_size(struct btrfs_root *root, int num_items)
+{
+ return (root->leafsize + root->nodesize * (BTRFS_MAX_LEVEL - 1)) *
+ 3 * num_items;
+}
+
+int btrfs_trans_reserve_metadata(struct btrfs_trans_handle *trans,
+ struct btrfs_root *root,
+ int num_items, int *retries)
+{
+ u64 num_bytes;
+ int ret;
+
+ if (num_items == 0 || root->fs_info->chunk_root == root)
+ return 0;
+
+ num_bytes = calc_trans_metadata_size(root, num_items);
+ ret = btrfs_block_rsv_add(trans, root, &root->fs_info->trans_block_rsv,
+ num_bytes, retries);
+ if (!ret) {
+ trans->bytes_reserved += num_bytes;
+ trans->block_rsv = &root->fs_info->trans_block_rsv;
+ }
+ return ret;
+}
+
+void btrfs_trans_release_metadata(struct btrfs_trans_handle *trans,
+ struct btrfs_root *root)
+{
+ if (!trans->bytes_reserved)
+ return;
+
+ BUG_ON(trans->block_rsv != &root->fs_info->trans_block_rsv);
+ btrfs_block_rsv_release(root, trans->block_rsv,
+ trans->bytes_reserved);
+ trans->bytes_reserved = 0;
+}
+
+int btrfs_snap_reserve_metadata(struct btrfs_trans_handle *trans,
+ struct btrfs_pending_snapshot *pending)
+{
+ struct btrfs_root *root = pending->root;
+ struct btrfs_block_rsv *src_rsv = get_block_rsv(trans, root);
+ struct btrfs_block_rsv *dst_rsv = &pending->block_rsv;
+ /*
+ * two for root back/forward refs, two for directory entries
+ * and one for root of the snapshot.
+ */
+ u64 num_bytes = calc_trans_metadata_size(root, 5);
+ dst_rsv->space_info = src_rsv->space_info;
+ return block_rsv_migrate_bytes(src_rsv, dst_rsv, num_bytes);
+}
+
static int update_block_group(struct btrfs_trans_handle *trans,
struct btrfs_root *root,
u64 bytenr, u64 num_bytes, int alloc)
@@ -5824,7 +5877,7 @@
wc = kzalloc(sizeof(*wc), GFP_NOFS);
BUG_ON(!wc);
- trans = btrfs_start_transaction(tree_root, 1);
+ trans = btrfs_start_transaction(tree_root, 0);
if (btrfs_disk_key_objectid(&root_item->drop_progress) == 0) {
level = btrfs_header_level(root->node);
@@ -5920,7 +5973,9 @@
BUG_ON(ret);
btrfs_end_transaction(trans, tree_root);
- trans = btrfs_start_transaction(tree_root, 1);
+ trans = btrfs_start_transaction(tree_root, 0);
+ if (IS_ERR(trans))
+ return PTR_ERR(trans);
} else {
unsigned long update;
update = trans->delayed_ref_updates;
diff --git a/fs/btrfs/file.c b/fs/btrfs/file.c
index 29ff749..41e09e2 100644
--- a/fs/btrfs/file.c
+++ b/fs/btrfs/file.c
@@ -126,8 +126,7 @@
end_of_last_block = start_pos + num_bytes - 1;
err = btrfs_set_extent_delalloc(inode, start_pos, end_of_last_block,
NULL);
- if (err)
- return err;
+ BUG_ON(err);
for (i = 0; i < num_pages; i++) {
struct page *p = pages[i];
@@ -142,7 +141,7 @@
* at this time.
*/
}
- return err;
+ return 0;
}
/*
@@ -1008,7 +1007,7 @@
num_written = err;
if ((file->f_flags & O_DSYNC) || IS_SYNC(inode)) {
- trans = btrfs_start_transaction(root, 1);
+ trans = btrfs_start_transaction(root, 0);
ret = btrfs_log_dentry_safe(trans, root,
file->f_dentry);
if (ret == 0) {
@@ -1104,9 +1103,9 @@
if (file && file->private_data)
btrfs_ioctl_trans_end(file);
- trans = btrfs_start_transaction(root, 1);
- if (!trans) {
- ret = -ENOMEM;
+ trans = btrfs_start_transaction(root, 0);
+ if (IS_ERR(trans)) {
+ ret = PTR_ERR(trans);
goto out;
}
diff --git a/fs/btrfs/inode-item.c b/fs/btrfs/inode-item.c
index 72ce3c1..64f1150 100644
--- a/fs/btrfs/inode-item.c
+++ b/fs/btrfs/inode-item.c
@@ -49,6 +49,33 @@
return 0;
}
+struct btrfs_inode_ref *
+btrfs_lookup_inode_ref(struct btrfs_trans_handle *trans,
+ struct btrfs_root *root,
+ struct btrfs_path *path,
+ const char *name, int name_len,
+ u64 inode_objectid, u64 ref_objectid, int mod)
+{
+ struct btrfs_key key;
+ struct btrfs_inode_ref *ref;
+ int ins_len = mod < 0 ? -1 : 0;
+ int cow = mod != 0;
+ int ret;
+
+ key.objectid = inode_objectid;
+ key.type = BTRFS_INODE_REF_KEY;
+ key.offset = ref_objectid;
+
+ ret = btrfs_search_slot(trans, root, &key, path, ins_len, cow);
+ if (ret < 0)
+ return ERR_PTR(ret);
+ if (ret > 0)
+ return NULL;
+ if (!find_name_in_backref(path, name, name_len, &ref))
+ return NULL;
+ return ref;
+}
+
int btrfs_del_inode_ref(struct btrfs_trans_handle *trans,
struct btrfs_root *root,
const char *name, int name_len,
diff --git a/fs/btrfs/inode.c b/fs/btrfs/inode.c
index f444250..c4b0fd1 100644
--- a/fs/btrfs/inode.c
+++ b/fs/btrfs/inode.c
@@ -2135,7 +2135,7 @@
* do a destroy_inode
*/
if (is_bad_inode(inode)) {
- trans = btrfs_start_transaction(root, 1);
+ trans = btrfs_start_transaction(root, 0);
btrfs_orphan_del(trans, inode);
btrfs_end_transaction(trans, root);
iput(inode);
@@ -2478,29 +2478,201 @@
return ret;
}
+/* helper to check if there is any shared block in the path */
+static int check_path_shared(struct btrfs_root *root,
+ struct btrfs_path *path)
+{
+ struct extent_buffer *eb;
+ int level;
+ int ret;
+ u64 refs;
+
+ for (level = 0; level < BTRFS_MAX_LEVEL; level++) {
+ if (!path->nodes[level])
+ break;
+ eb = path->nodes[level];
+ if (!btrfs_block_can_be_shared(root, eb))
+ continue;
+ ret = btrfs_lookup_extent_info(NULL, root, eb->start, eb->len,
+ &refs, NULL);
+ if (refs > 1)
+ return 1;
+ }
+ return 0;
+}
+
+/*
+ * helper to start transaction for unlink and rmdir.
+ *
+ * unlink and rmdir are special in btrfs, they do not always free space.
+ * so in enospc case, we should make sure they will free space before
+ * allowing them to use the global metadata reservation.
+ */
+static struct btrfs_trans_handle *__unlink_start_trans(struct inode *dir,
+ struct dentry *dentry)
+{
+ struct btrfs_trans_handle *trans;
+ struct btrfs_root *root = BTRFS_I(dir)->root;
+ struct btrfs_path *path;
+ struct btrfs_inode_ref *ref;
+ struct btrfs_dir_item *di;
+ struct inode *inode = dentry->d_inode;
+ u64 index;
+ int check_link = 1;
+ int err = -ENOSPC;
+ int ret;
+
+ trans = btrfs_start_transaction(root, 10);
+ if (!IS_ERR(trans) || PTR_ERR(trans) != -ENOSPC)
+ return trans;
+
+ if (inode->i_ino == BTRFS_EMPTY_SUBVOL_DIR_OBJECTID)
+ return ERR_PTR(-ENOSPC);
+
+ /* check if there is someone else holds reference */
+ if (S_ISDIR(inode->i_mode) && atomic_read(&inode->i_count) > 1)
+ return ERR_PTR(-ENOSPC);
+
+ if (atomic_read(&inode->i_count) > 2)
+ return ERR_PTR(-ENOSPC);
+
+ if (xchg(&root->fs_info->enospc_unlink, 1))
+ return ERR_PTR(-ENOSPC);
+
+ path = btrfs_alloc_path();
+ if (!path) {
+ root->fs_info->enospc_unlink = 0;
+ return ERR_PTR(-ENOMEM);
+ }
+
+ trans = btrfs_start_transaction(root, 0);
+ if (IS_ERR(trans)) {
+ btrfs_free_path(path);
+ root->fs_info->enospc_unlink = 0;
+ return trans;
+ }
+
+ path->skip_locking = 1;
+ path->search_commit_root = 1;
+
+ ret = btrfs_lookup_inode(trans, root, path,
+ &BTRFS_I(dir)->location, 0);
+ if (ret < 0) {
+ err = ret;
+ goto out;
+ }
+ if (ret == 0) {
+ if (check_path_shared(root, path))
+ goto out;
+ } else {
+ check_link = 0;
+ }
+ btrfs_release_path(root, path);
+
+ ret = btrfs_lookup_inode(trans, root, path,
+ &BTRFS_I(inode)->location, 0);
+ if (ret < 0) {
+ err = ret;
+ goto out;
+ }
+ if (ret == 0) {
+ if (check_path_shared(root, path))
+ goto out;
+ } else {
+ check_link = 0;
+ }
+ btrfs_release_path(root, path);
+
+ if (ret == 0 && S_ISREG(inode->i_mode)) {
+ ret = btrfs_lookup_file_extent(trans, root, path,
+ inode->i_ino, (u64)-1, 0);
+ if (ret < 0) {
+ err = ret;
+ goto out;
+ }
+ BUG_ON(ret == 0);
+ if (check_path_shared(root, path))
+ goto out;
+ btrfs_release_path(root, path);
+ }
+
+ if (!check_link) {
+ err = 0;
+ goto out;
+ }
+
+ di = btrfs_lookup_dir_item(trans, root, path, dir->i_ino,
+ dentry->d_name.name, dentry->d_name.len, 0);
+ if (IS_ERR(di)) {
+ err = PTR_ERR(di);
+ goto out;
+ }
+ if (di) {
+ if (check_path_shared(root, path))
+ goto out;
+ } else {
+ err = 0;
+ goto out;
+ }
+ btrfs_release_path(root, path);
+
+ ref = btrfs_lookup_inode_ref(trans, root, path,
+ dentry->d_name.name, dentry->d_name.len,
+ inode->i_ino, dir->i_ino, 0);
+ if (IS_ERR(ref)) {
+ err = PTR_ERR(ref);
+ goto out;
+ }
+ BUG_ON(!ref);
+ if (check_path_shared(root, path))
+ goto out;
+ index = btrfs_inode_ref_index(path->nodes[0], ref);
+ btrfs_release_path(root, path);
+
+ di = btrfs_lookup_dir_index_item(trans, root, path, dir->i_ino, index,
+ dentry->d_name.name, dentry->d_name.len, 0);
+ if (IS_ERR(di)) {
+ err = PTR_ERR(di);
+ goto out;
+ }
+ BUG_ON(ret == -ENOENT);
+ if (check_path_shared(root, path))
+ goto out;
+
+ err = 0;
+out:
+ btrfs_free_path(path);
+ if (err) {
+ btrfs_end_transaction(trans, root);
+ root->fs_info->enospc_unlink = 0;
+ return ERR_PTR(err);
+ }
+
+ trans->block_rsv = &root->fs_info->global_block_rsv;
+ return trans;
+}
+
+static void __unlink_end_trans(struct btrfs_trans_handle *trans,
+ struct btrfs_root *root)
+{
+ if (trans->block_rsv == &root->fs_info->global_block_rsv) {
+ BUG_ON(!root->fs_info->enospc_unlink);
+ root->fs_info->enospc_unlink = 0;
+ }
+ btrfs_end_transaction_throttle(trans, root);
+}
+
static int btrfs_unlink(struct inode *dir, struct dentry *dentry)
{
- struct btrfs_root *root;
+ struct btrfs_root *root = BTRFS_I(dir)->root;
struct btrfs_trans_handle *trans;
struct inode *inode = dentry->d_inode;
int ret;
unsigned long nr = 0;
- root = BTRFS_I(dir)->root;
-
- /*
- * 5 items for unlink inode
- * 1 for orphan
- */
- ret = btrfs_reserve_metadata_space(root, 6);
- if (ret)
- return ret;
-
- trans = btrfs_start_transaction(root, 1);
- if (IS_ERR(trans)) {
- btrfs_unreserve_metadata_space(root, 6);
+ trans = __unlink_start_trans(dir, dentry);
+ if (IS_ERR(trans))
return PTR_ERR(trans);
- }
btrfs_set_trans_block_group(trans, dir);
@@ -2508,14 +2680,15 @@
ret = btrfs_unlink_inode(trans, root, dir, dentry->d_inode,
dentry->d_name.name, dentry->d_name.len);
+ BUG_ON(ret);
- if (inode->i_nlink == 0)
+ if (inode->i_nlink == 0) {
ret = btrfs_orphan_add(trans, inode);
+ BUG_ON(ret);
+ }
nr = trans->blocks_used;
-
- btrfs_end_transaction_throttle(trans, root);
- btrfs_unreserve_metadata_space(root, 6);
+ __unlink_end_trans(trans, root);
btrfs_btree_balance_dirty(root, nr);
return ret;
}
@@ -2587,7 +2760,6 @@
{
struct inode *inode = dentry->d_inode;
int err = 0;
- int ret;
struct btrfs_root *root = BTRFS_I(dir)->root;
struct btrfs_trans_handle *trans;
unsigned long nr = 0;
@@ -2596,15 +2768,9 @@
inode->i_ino == BTRFS_FIRST_FREE_OBJECTID)
return -ENOTEMPTY;
- ret = btrfs_reserve_metadata_space(root, 5);
- if (ret)
- return ret;
-
- trans = btrfs_start_transaction(root, 1);
- if (IS_ERR(trans)) {
- btrfs_unreserve_metadata_space(root, 5);
+ trans = __unlink_start_trans(dir, dentry);
+ if (IS_ERR(trans))
return PTR_ERR(trans);
- }
btrfs_set_trans_block_group(trans, dir);
@@ -2627,12 +2793,9 @@
btrfs_i_size_write(inode, 0);
out:
nr = trans->blocks_used;
- ret = btrfs_end_transaction_throttle(trans, root);
- btrfs_unreserve_metadata_space(root, 5);
+ __unlink_end_trans(trans, root);
btrfs_btree_balance_dirty(root, nr);
- if (ret && !err)
- err = ret;
return err;
}
@@ -3145,7 +3308,7 @@
struct btrfs_trans_handle *trans;
struct btrfs_root *root = BTRFS_I(inode)->root;
struct extent_io_tree *io_tree = &BTRFS_I(inode)->io_tree;
- struct extent_map *em;
+ struct extent_map *em = NULL;
struct extent_state *cached_state = NULL;
u64 mask = root->sectorsize - 1;
u64 hole_start = (inode->i_size + mask) & ~mask;
@@ -3183,11 +3346,11 @@
u64 hint_byte = 0;
hole_size = last_byte - cur_offset;
- err = btrfs_reserve_metadata_space(root, 2);
- if (err)
+ trans = btrfs_start_transaction(root, 2);
+ if (IS_ERR(trans)) {
+ err = PTR_ERR(trans);
break;
-
- trans = btrfs_start_transaction(root, 1);
+ }
btrfs_set_trans_block_group(trans, inode);
err = btrfs_drop_extents(trans, inode, cur_offset,
@@ -3205,14 +3368,15 @@
last_byte - 1, 0);
btrfs_end_transaction(trans, root);
- btrfs_unreserve_metadata_space(root, 2);
}
free_extent_map(em);
+ em = NULL;
cur_offset = last_byte;
if (cur_offset >= block_end)
break;
}
+ free_extent_map(em);
unlock_extent_cached(io_tree, hole_start, block_end - 1, &cached_state,
GFP_NOFS);
return err;
@@ -3239,10 +3403,6 @@
}
}
- ret = btrfs_reserve_metadata_space(root, 1);
- if (ret)
- return ret;
-
trans = btrfs_start_transaction(root, 1);
btrfs_set_trans_block_group(trans, inode);
@@ -3251,7 +3411,6 @@
nr = trans->blocks_used;
btrfs_end_transaction(trans, root);
- btrfs_unreserve_metadata_space(root, 1);
btrfs_btree_balance_dirty(root, nr);
if (attr->ia_size > inode->i_size) {
@@ -4223,26 +4382,21 @@
if (!new_valid_dev(rdev))
return -EINVAL;
+ err = btrfs_find_free_objectid(NULL, root, dir->i_ino, &objectid);
+ if (err)
+ return err;
+
/*
* 2 for inode item and ref
* 2 for dir items
* 1 for xattr if selinux is on
*/
- err = btrfs_reserve_metadata_space(root, 5);
- if (err)
- return err;
+ trans = btrfs_start_transaction(root, 5);
+ if (IS_ERR(trans))
+ return PTR_ERR(trans);
- trans = btrfs_start_transaction(root, 1);
- if (!trans)
- goto fail;
btrfs_set_trans_block_group(trans, dir);
- err = btrfs_find_free_objectid(trans, root, dir->i_ino, &objectid);
- if (err) {
- err = -ENOSPC;
- goto out_unlock;
- }
-
inode = btrfs_new_inode(trans, root, dir, dentry->d_name.name,
dentry->d_name.len,
dentry->d_parent->d_inode->i_ino, objectid,
@@ -4271,13 +4425,11 @@
out_unlock:
nr = trans->blocks_used;
btrfs_end_transaction_throttle(trans, root);
-fail:
- btrfs_unreserve_metadata_space(root, 5);
+ btrfs_btree_balance_dirty(root, nr);
if (drop_inode) {
inode_dec_link_count(inode);
iput(inode);
}
- btrfs_btree_balance_dirty(root, nr);
return err;
}
@@ -4287,32 +4439,26 @@
struct btrfs_trans_handle *trans;
struct btrfs_root *root = BTRFS_I(dir)->root;
struct inode *inode = NULL;
- int err;
int drop_inode = 0;
+ int err;
unsigned long nr = 0;
u64 objectid;
u64 index = 0;
+ err = btrfs_find_free_objectid(NULL, root, dir->i_ino, &objectid);
+ if (err)
+ return err;
/*
* 2 for inode item and ref
* 2 for dir items
* 1 for xattr if selinux is on
*/
- err = btrfs_reserve_metadata_space(root, 5);
- if (err)
- return err;
+ trans = btrfs_start_transaction(root, 5);
+ if (IS_ERR(trans))
+ return PTR_ERR(trans);
- trans = btrfs_start_transaction(root, 1);
- if (!trans)
- goto fail;
btrfs_set_trans_block_group(trans, dir);
- err = btrfs_find_free_objectid(trans, root, dir->i_ino, &objectid);
- if (err) {
- err = -ENOSPC;
- goto out_unlock;
- }
-
inode = btrfs_new_inode(trans, root, dir, dentry->d_name.name,
dentry->d_name.len,
dentry->d_parent->d_inode->i_ino,
@@ -4344,8 +4490,6 @@
out_unlock:
nr = trans->blocks_used;
btrfs_end_transaction_throttle(trans, root);
-fail:
- btrfs_unreserve_metadata_space(root, 5);
if (drop_inode) {
inode_dec_link_count(inode);
iput(inode);
@@ -4372,21 +4516,21 @@
if (root->objectid != BTRFS_I(inode)->root->objectid)
return -EPERM;
- /*
- * 1 item for inode ref
- * 2 items for dir items
- */
- err = btrfs_reserve_metadata_space(root, 3);
- if (err)
- return err;
-
btrfs_inc_nlink(inode);
err = btrfs_set_inode_index(dir, &index);
if (err)
goto fail;
- trans = btrfs_start_transaction(root, 1);
+ /*
+ * 1 item for inode ref
+ * 2 items for dir items
+ */
+ trans = btrfs_start_transaction(root, 3);
+ if (IS_ERR(trans)) {
+ err = PTR_ERR(trans);
+ goto fail;
+ }
btrfs_set_trans_block_group(trans, dir);
atomic_inc(&inode->i_count);
@@ -4405,7 +4549,6 @@
nr = trans->blocks_used;
btrfs_end_transaction_throttle(trans, root);
fail:
- btrfs_unreserve_metadata_space(root, 3);
if (drop_inode) {
inode_dec_link_count(inode);
iput(inode);
@@ -4425,28 +4568,20 @@
u64 index = 0;
unsigned long nr = 1;
+ err = btrfs_find_free_objectid(NULL, root, dir->i_ino, &objectid);
+ if (err)
+ return err;
+
/*
* 2 items for inode and ref
* 2 items for dir items
* 1 for xattr if selinux is on
*/
- err = btrfs_reserve_metadata_space(root, 5);
- if (err)
- return err;
-
- trans = btrfs_start_transaction(root, 1);
- if (!trans) {
- err = -ENOMEM;
- goto out_unlock;
- }
+ trans = btrfs_start_transaction(root, 5);
+ if (IS_ERR(trans))
+ return PTR_ERR(trans);
btrfs_set_trans_block_group(trans, dir);
- err = btrfs_find_free_objectid(trans, root, dir->i_ino, &objectid);
- if (err) {
- err = -ENOSPC;
- goto out_fail;
- }
-
inode = btrfs_new_inode(trans, root, dir, dentry->d_name.name,
dentry->d_name.len,
dentry->d_parent->d_inode->i_ino, objectid,
@@ -4486,9 +4621,6 @@
out_fail:
nr = trans->blocks_used;
btrfs_end_transaction_throttle(trans, root);
-
-out_unlock:
- btrfs_unreserve_metadata_space(root, 5);
if (drop_on_err)
iput(inode);
btrfs_btree_balance_dirty(root, nr);
@@ -5426,19 +5558,6 @@
if (S_ISDIR(old_inode->i_mode) && new_inode &&
new_inode->i_size > BTRFS_EMPTY_DIR_SIZE)
return -ENOTEMPTY;
-
- /*
- * We want to reserve the absolute worst case amount of items. So if
- * both inodes are subvols and we need to unlink them then that would
- * require 4 item modifications, but if they are both normal inodes it
- * would require 5 item modifications, so we'll assume their normal
- * inodes. So 5 * 2 is 10, plus 1 for the new link, so 11 total items
- * should cover the worst case number of items we'll modify.
- */
- ret = btrfs_reserve_metadata_space(root, 11);
- if (ret)
- return ret;
-
/*
* we're using rename to replace one file with another.
* and the replacement file is large. Start IO on it now so
@@ -5451,8 +5570,18 @@
/* close the racy window with snapshot create/destroy ioctl */
if (old_inode->i_ino == BTRFS_FIRST_FREE_OBJECTID)
down_read(&root->fs_info->subvol_sem);
+ /*
+ * We want to reserve the absolute worst case amount of items. So if
+ * both inodes are subvols and we need to unlink them then that would
+ * require 4 item modifications, but if they are both normal inodes it
+ * would require 5 item modifications, so we'll assume their normal
+ * inodes. So 5 * 2 is 10, plus 1 for the new link, so 11 total items
+ * should cover the worst case number of items we'll modify.
+ */
+ trans = btrfs_start_transaction(root, 20);
+ if (IS_ERR(trans))
+ return PTR_ERR(trans);
- trans = btrfs_start_transaction(root, 1);
btrfs_set_trans_block_group(trans, new_dir);
if (dest != root)
@@ -5551,7 +5680,6 @@
if (old_inode->i_ino == BTRFS_FIRST_FREE_OBJECTID)
up_read(&root->fs_info->subvol_sem);
- btrfs_unreserve_metadata_space(root, 11);
return ret;
}
@@ -5658,26 +5786,20 @@
if (name_len > BTRFS_MAX_INLINE_DATA_SIZE(root))
return -ENAMETOOLONG;
+ err = btrfs_find_free_objectid(NULL, root, dir->i_ino, &objectid);
+ if (err)
+ return err;
/*
* 2 items for inode item and ref
* 2 items for dir items
* 1 item for xattr if selinux is on
*/
- err = btrfs_reserve_metadata_space(root, 5);
- if (err)
- return err;
+ trans = btrfs_start_transaction(root, 5);
+ if (IS_ERR(trans))
+ return PTR_ERR(trans);
- trans = btrfs_start_transaction(root, 1);
- if (!trans)
- goto out_fail;
btrfs_set_trans_block_group(trans, dir);
- err = btrfs_find_free_objectid(trans, root, dir->i_ino, &objectid);
- if (err) {
- err = -ENOSPC;
- goto out_unlock;
- }
-
inode = btrfs_new_inode(trans, root, dir, dentry->d_name.name,
dentry->d_name.len,
dentry->d_parent->d_inode->i_ino, objectid,
@@ -5749,8 +5871,6 @@
out_unlock:
nr = trans->blocks_used;
btrfs_end_transaction_throttle(trans, root);
-out_fail:
- btrfs_unreserve_metadata_space(root, 5);
if (drop_inode) {
inode_dec_link_count(inode);
iput(inode);
@@ -5771,21 +5891,18 @@
u64 i_size;
while (num_bytes > 0) {
- trans = btrfs_start_transaction(root, 1);
+ trans = btrfs_start_transaction(root, 3);
+ if (IS_ERR(trans)) {
+ ret = PTR_ERR(trans);
+ break;
+ }
ret = btrfs_reserve_extent(trans, root, num_bytes,
root->sectorsize, 0, alloc_hint,
(u64)-1, &ins, 1);
if (ret) {
- WARN_ON(1);
- goto stop_trans;
- }
-
- ret = btrfs_reserve_metadata_space(root, 3);
- if (ret) {
- btrfs_free_reserved_extent(root, ins.objectid,
- ins.offset);
- goto stop_trans;
+ btrfs_end_transaction(trans, root);
+ break;
}
ret = insert_reserved_file_extent(trans, inode,
@@ -5819,14 +5936,8 @@
BUG_ON(ret);
btrfs_end_transaction(trans, root);
- btrfs_unreserve_metadata_space(root, 3);
}
return ret;
-
-stop_trans:
- btrfs_end_transaction(trans, root);
- return ret;
-
}
static long btrfs_fallocate(struct inode *inode, int mode,
diff --git a/fs/btrfs/ioctl.c b/fs/btrfs/ioctl.c
index 97a9783..3066da4 100644
--- a/fs/btrfs/ioctl.c
+++ b/fs/btrfs/ioctl.c
@@ -239,23 +239,19 @@
u64 new_dirid = BTRFS_FIRST_FREE_OBJECTID;
u64 index = 0;
+ ret = btrfs_find_free_objectid(NULL, root->fs_info->tree_root,
+ 0, &objectid);
+ if (ret)
+ return ret;
/*
* 1 - inode item
* 2 - refs
* 1 - root item
* 2 - dir items
*/
- ret = btrfs_reserve_metadata_space(root, 6);
- if (ret)
- return ret;
-
- trans = btrfs_start_transaction(root, 1);
- BUG_ON(!trans);
-
- ret = btrfs_find_free_objectid(trans, root->fs_info->tree_root,
- 0, &objectid);
- if (ret)
- goto fail;
+ trans = btrfs_start_transaction(root, 6);
+ if (IS_ERR(trans))
+ return PTR_ERR(trans);
leaf = btrfs_alloc_free_block(trans, root, root->leafsize,
0, objectid, NULL, 0, 0, 0);
@@ -345,13 +341,10 @@
err = btrfs_commit_transaction(trans, root);
if (err && !ret)
ret = err;
-
- btrfs_unreserve_metadata_space(root, 6);
return ret;
}
-static int create_snapshot(struct btrfs_root *root, struct dentry *dentry,
- char *name, int namelen)
+static int create_snapshot(struct btrfs_root *root, struct dentry *dentry)
{
struct inode *inode;
struct btrfs_pending_snapshot *pending_snapshot;
@@ -361,40 +354,33 @@
if (!root->ref_cows)
return -EINVAL;
- /*
- * 1 - inode item
- * 2 - refs
- * 1 - root item
- * 2 - dir items
- */
- ret = btrfs_reserve_metadata_space(root, 6);
+ pending_snapshot = kzalloc(sizeof(*pending_snapshot), GFP_NOFS);
+ if (!pending_snapshot)
+ return -ENOMEM;
+
+ btrfs_init_block_rsv(&pending_snapshot->block_rsv);
+ pending_snapshot->dentry = dentry;
+ pending_snapshot->root = root;
+
+ trans = btrfs_start_transaction(root->fs_info->extent_root, 5);
+ if (IS_ERR(trans)) {
+ ret = PTR_ERR(trans);
+ goto fail;
+ }
+
+ ret = btrfs_snap_reserve_metadata(trans, pending_snapshot);
+ BUG_ON(ret);
+
+ list_add(&pending_snapshot->list,
+ &trans->transaction->pending_snapshots);
+ ret = btrfs_commit_transaction(trans, root->fs_info->extent_root);
+ BUG_ON(ret);
+
+ ret = pending_snapshot->error;
if (ret)
goto fail;
- pending_snapshot = kzalloc(sizeof(*pending_snapshot), GFP_NOFS);
- if (!pending_snapshot) {
- ret = -ENOMEM;
- btrfs_unreserve_metadata_space(root, 6);
- goto fail;
- }
- pending_snapshot->name = kmalloc(namelen + 1, GFP_NOFS);
- if (!pending_snapshot->name) {
- ret = -ENOMEM;
- kfree(pending_snapshot);
- btrfs_unreserve_metadata_space(root, 6);
- goto fail;
- }
- memcpy(pending_snapshot->name, name, namelen);
- pending_snapshot->name[namelen] = '\0';
- pending_snapshot->dentry = dentry;
- trans = btrfs_start_transaction(root, 1);
- BUG_ON(!trans);
- pending_snapshot->root = root;
- list_add(&pending_snapshot->list,
- &trans->transaction->pending_snapshots);
- ret = btrfs_commit_transaction(trans, root);
- BUG_ON(ret);
- btrfs_unreserve_metadata_space(root, 6);
+ btrfs_orphan_cleanup(pending_snapshot->snap);
inode = btrfs_lookup_dentry(dentry->d_parent->d_inode, dentry);
if (IS_ERR(inode)) {
@@ -405,6 +391,7 @@
d_instantiate(dentry, inode);
ret = 0;
fail:
+ kfree(pending_snapshot);
return ret;
}
@@ -456,8 +443,7 @@
goto out_up_read;
if (snap_src) {
- error = create_snapshot(snap_src, dentry,
- name, namelen);
+ error = create_snapshot(snap_src, dentry);
} else {
error = create_subvol(BTRFS_I(dir)->root, dentry,
name, namelen);
@@ -811,7 +797,7 @@
device->name, (unsigned long long)new_size);
if (new_size > old_size) {
- trans = btrfs_start_transaction(root, 1);
+ trans = btrfs_start_transaction(root, 0);
ret = btrfs_grow_device(trans, device, new_size);
btrfs_commit_transaction(trans, root);
} else {
@@ -1300,7 +1286,13 @@
if (err)
goto out_up_write;
- trans = btrfs_start_transaction(root, 1);
+ trans = btrfs_start_transaction(root, 0);
+ if (IS_ERR(trans)) {
+ err = PTR_ERR(trans);
+ goto out;
+ }
+ trans->block_rsv = &root->fs_info->global_block_rsv;
+
ret = btrfs_unlink_subvol(trans, root, dir,
dest->root_key.objectid,
dentry->d_name.name,
@@ -1550,12 +1542,6 @@
btrfs_wait_ordered_range(src, off, off+len);
}
- trans = btrfs_start_transaction(root, 1);
- BUG_ON(!trans);
-
- /* punch hole in destination first */
- btrfs_drop_extents(trans, inode, off, off + len, &hint_byte, 1);
-
/* clone data */
key.objectid = src->i_ino;
key.type = BTRFS_EXTENT_DATA_KEY;
@@ -1566,7 +1552,7 @@
* note the key will change type as we walk through the
* tree.
*/
- ret = btrfs_search_slot(trans, root, &key, path, 0, 0);
+ ret = btrfs_search_slot(NULL, root, &key, path, 0, 0);
if (ret < 0)
goto out;
@@ -1629,12 +1615,31 @@
new_key.objectid = inode->i_ino;
new_key.offset = key.offset + destoff - off;
+ trans = btrfs_start_transaction(root, 1);
+ if (IS_ERR(trans)) {
+ ret = PTR_ERR(trans);
+ goto out;
+ }
+
if (type == BTRFS_FILE_EXTENT_REG ||
type == BTRFS_FILE_EXTENT_PREALLOC) {
+ if (off > key.offset) {
+ datao += off - key.offset;
+ datal -= off - key.offset;
+ }
+
+ if (key.offset + datal > off + len)
+ datal = off + len - key.offset;
+
+ ret = btrfs_drop_extents(trans, inode,
+ new_key.offset,
+ new_key.offset + datal,
+ &hint_byte, 1);
+ BUG_ON(ret);
+
ret = btrfs_insert_empty_item(trans, root, path,
&new_key, size);
- if (ret)
- goto out;
+ BUG_ON(ret);
leaf = path->nodes[0];
slot = path->slots[0];
@@ -1645,14 +1650,6 @@
extent = btrfs_item_ptr(leaf, slot,
struct btrfs_file_extent_item);
- if (off > key.offset) {
- datao += off - key.offset;
- datal -= off - key.offset;
- }
-
- if (key.offset + datal > off + len)
- datal = off + len - key.offset;
-
/* disko == 0 means it's a hole */
if (!disko)
datao = 0;
@@ -1683,14 +1680,21 @@
if (comp && (skip || trim)) {
ret = -EINVAL;
+ btrfs_end_transaction(trans, root);
goto out;
}
size -= skip + trim;
datal -= skip + trim;
+
+ ret = btrfs_drop_extents(trans, inode,
+ new_key.offset,
+ new_key.offset + datal,
+ &hint_byte, 1);
+ BUG_ON(ret);
+
ret = btrfs_insert_empty_item(trans, root, path,
&new_key, size);
- if (ret)
- goto out;
+ BUG_ON(ret);
if (skip) {
u32 start =
@@ -1708,8 +1712,17 @@
}
btrfs_mark_buffer_dirty(leaf);
- }
+ btrfs_release_path(root, path);
+ inode->i_mtime = inode->i_ctime = CURRENT_TIME;
+ if (new_key.offset + datal > inode->i_size)
+ btrfs_i_size_write(inode,
+ new_key.offset + datal);
+ BTRFS_I(inode)->flags = BTRFS_I(src)->flags;
+ ret = btrfs_update_inode(trans, root, inode);
+ BUG_ON(ret);
+ btrfs_end_transaction(trans, root);
+ }
next:
btrfs_release_path(root, path);
key.offset++;
@@ -1717,17 +1730,7 @@
ret = 0;
out:
btrfs_release_path(root, path);
- if (ret == 0) {
- inode->i_mtime = inode->i_ctime = CURRENT_TIME;
- if (destoff + olen > inode->i_size)
- btrfs_i_size_write(inode, destoff + olen);
- BTRFS_I(inode)->flags = BTRFS_I(src)->flags;
- ret = btrfs_update_inode(trans, root, inode);
- }
- btrfs_end_transaction(trans, root);
unlock_extent(&BTRFS_I(src)->io_tree, off, off+len, GFP_NOFS);
- if (ret)
- vmtruncate(inode, 0);
out_unlock:
mutex_unlock(&src->i_mutex);
mutex_unlock(&inode->i_mutex);
diff --git a/fs/btrfs/relocation.c b/fs/btrfs/relocation.c
index d565b45..145a468 100644
--- a/fs/btrfs/relocation.c
+++ b/fs/btrfs/relocation.c
@@ -1649,7 +1649,7 @@
}
if (level == 0 && rc->stage == UPDATE_DATA_PTRS) {
- trans = btrfs_start_transaction(root, 1);
+ trans = btrfs_start_transaction(root, 0);
leaf = path->nodes[0];
btrfs_item_key_to_cpu(leaf, &key, 0);
@@ -1675,7 +1675,7 @@
while (1) {
leaf = NULL;
replaced = 0;
- trans = btrfs_start_transaction(root, 1);
+ trans = btrfs_start_transaction(root, 0);
max_level = level;
ret = walk_down_reloc_tree(reloc_root, path, &level);
@@ -1803,7 +1803,7 @@
merge_reloc_root(async->rc, root);
- trans = btrfs_start_transaction(root, 1);
+ trans = btrfs_start_transaction(root, 0);
btrfs_update_reloc_root(trans, root);
btrfs_end_transaction(trans, root);
}
@@ -3297,11 +3297,11 @@
rc->create_reloc_root = 1;
set_reloc_control(rc);
- trans = btrfs_start_transaction(rc->extent_root, 1);
+ trans = btrfs_join_transaction(rc->extent_root, 1);
btrfs_commit_transaction(trans, rc->extent_root);
while (1) {
- trans = btrfs_start_transaction(rc->extent_root, 1);
+ trans = btrfs_start_transaction(rc->extent_root, 0);
ret = find_next_extent(trans, rc, path);
if (ret < 0)
@@ -3411,7 +3411,7 @@
smp_mb();
if (rc->extents_found > 0) {
- trans = btrfs_start_transaction(rc->extent_root, 1);
+ trans = btrfs_join_transaction(rc->extent_root, 1);
btrfs_commit_transaction(trans, rc->extent_root);
}
@@ -3420,7 +3420,7 @@
unset_reloc_control(rc);
/* get rid of pinned extents */
- trans = btrfs_start_transaction(rc->extent_root, 1);
+ trans = btrfs_join_transaction(rc->extent_root, 1);
btrfs_commit_transaction(trans, rc->extent_root);
return err;
@@ -3475,7 +3475,7 @@
if (IS_ERR(root))
return ERR_CAST(root);
- trans = btrfs_start_transaction(root, 1);
+ trans = btrfs_start_transaction(root, 6);
BUG_ON(!trans);
err = btrfs_find_free_objectid(trans, root, objectid, &objectid);
@@ -3619,7 +3619,7 @@
struct btrfs_trans_handle *trans;
int ret;
- trans = btrfs_start_transaction(root->fs_info->tree_root, 1);
+ trans = btrfs_start_transaction(root->fs_info->tree_root, 0);
memset(&root->root_item.drop_progress, 0,
sizeof(root->root_item.drop_progress));
diff --git a/fs/btrfs/super.c b/fs/btrfs/super.c
index 55265c24..38d91c7 100644
--- a/fs/btrfs/super.c
+++ b/fs/btrfs/super.c
@@ -498,7 +498,7 @@
btrfs_start_delalloc_inodes(root, 0);
btrfs_wait_ordered_extents(root, 0, 0);
- trans = btrfs_start_transaction(root, 1);
+ trans = btrfs_start_transaction(root, 0);
ret = btrfs_commit_transaction(trans, root);
return ret;
}
diff --git a/fs/btrfs/transaction.c b/fs/btrfs/transaction.c
index 21ad37c..2616491 100644
--- a/fs/btrfs/transaction.c
+++ b/fs/btrfs/transaction.c
@@ -165,53 +165,89 @@
TRANS_USERSPACE,
};
-static struct btrfs_trans_handle *start_transaction(struct btrfs_root *root,
- int num_blocks, int type)
+static int may_wait_transaction(struct btrfs_root *root, int type)
{
- struct btrfs_trans_handle *h =
- kmem_cache_alloc(btrfs_trans_handle_cachep, GFP_NOFS);
- int ret;
-
- mutex_lock(&root->fs_info->trans_mutex);
if (!root->fs_info->log_root_recovering &&
((type == TRANS_START && !root->fs_info->open_ioctl_trans) ||
type == TRANS_USERSPACE))
+ return 1;
+ return 0;
+}
+
+static struct btrfs_trans_handle *start_transaction(struct btrfs_root *root,
+ u64 num_items, int type)
+{
+ struct btrfs_trans_handle *h;
+ struct btrfs_transaction *cur_trans;
+ int retries = 0;
+ int ret;
+again:
+ h = kmem_cache_alloc(btrfs_trans_handle_cachep, GFP_NOFS);
+ if (!h)
+ return ERR_PTR(-ENOMEM);
+
+ mutex_lock(&root->fs_info->trans_mutex);
+ if (may_wait_transaction(root, type))
wait_current_trans(root);
+
ret = join_transaction(root);
BUG_ON(ret);
- h->transid = root->fs_info->running_transaction->transid;
- h->transaction = root->fs_info->running_transaction;
- h->blocks_reserved = num_blocks;
+ cur_trans = root->fs_info->running_transaction;
+ cur_trans->use_count++;
+ mutex_unlock(&root->fs_info->trans_mutex);
+
+ h->transid = cur_trans->transid;
+ h->transaction = cur_trans;
h->blocks_used = 0;
h->block_group = 0;
+ h->bytes_reserved = 0;
h->delayed_ref_updates = 0;
h->block_rsv = NULL;
- if (!current->journal_info && type != TRANS_USERSPACE)
- current->journal_info = h;
+ smp_mb();
+ if (cur_trans->blocked && may_wait_transaction(root, type)) {
+ btrfs_commit_transaction(h, root);
+ goto again;
+ }
- root->fs_info->running_transaction->use_count++;
+ if (num_items > 0) {
+ ret = btrfs_trans_reserve_metadata(h, root, num_items,
+ &retries);
+ if (ret == -EAGAIN) {
+ btrfs_commit_transaction(h, root);
+ goto again;
+ }
+ if (ret < 0) {
+ btrfs_end_transaction(h, root);
+ return ERR_PTR(ret);
+ }
+ }
+
+ mutex_lock(&root->fs_info->trans_mutex);
record_root_in_trans(h, root);
mutex_unlock(&root->fs_info->trans_mutex);
+
+ if (!current->journal_info && type != TRANS_USERSPACE)
+ current->journal_info = h;
return h;
}
struct btrfs_trans_handle *btrfs_start_transaction(struct btrfs_root *root,
- int num_blocks)
+ int num_items)
{
- return start_transaction(root, num_blocks, TRANS_START);
+ return start_transaction(root, num_items, TRANS_START);
}
struct btrfs_trans_handle *btrfs_join_transaction(struct btrfs_root *root,
int num_blocks)
{
- return start_transaction(root, num_blocks, TRANS_JOIN);
+ return start_transaction(root, 0, TRANS_JOIN);
}
struct btrfs_trans_handle *btrfs_start_ioctl_transaction(struct btrfs_root *r,
int num_blocks)
{
- return start_transaction(r, num_blocks, TRANS_USERSPACE);
+ return start_transaction(r, 0, TRANS_USERSPACE);
}
/* wait for a transaction commit to be fully complete */
@@ -312,6 +348,8 @@
count++;
}
+ btrfs_trans_release_metadata(trans, root);
+
mutex_lock(&info->trans_mutex);
cur_trans = info->running_transaction;
WARN_ON(cur_trans != trans->transaction);
@@ -757,47 +795,49 @@
struct btrfs_root *root = pending->root;
struct btrfs_root *parent_root;
struct inode *parent_inode;
+ struct dentry *dentry;
struct extent_buffer *tmp;
struct extent_buffer *old;
int ret;
- u64 objectid;
- int namelen;
u64 index = 0;
-
- parent_inode = pending->dentry->d_parent->d_inode;
- parent_root = BTRFS_I(parent_inode)->root;
+ u64 objectid;
new_root_item = kmalloc(sizeof(*new_root_item), GFP_NOFS);
if (!new_root_item) {
- ret = -ENOMEM;
+ pending->error = -ENOMEM;
goto fail;
}
+
ret = btrfs_find_free_objectid(trans, tree_root, 0, &objectid);
- if (ret)
+ if (ret) {
+ pending->error = ret;
goto fail;
+ }
key.objectid = objectid;
- /* record when the snapshot was created in key.offset */
- key.offset = trans->transid;
- btrfs_set_key_type(&key, BTRFS_ROOT_ITEM_KEY);
+ key.offset = (u64)-1;
+ key.type = BTRFS_ROOT_ITEM_KEY;
- memcpy(&pending->root_key, &key, sizeof(key));
- pending->root_key.offset = (u64)-1;
+ trans->block_rsv = &pending->block_rsv;
+ dentry = pending->dentry;
+ parent_inode = dentry->d_parent->d_inode;
+ parent_root = BTRFS_I(parent_inode)->root;
record_root_in_trans(trans, parent_root);
+
/*
* insert the directory item
*/
- namelen = strlen(pending->name);
ret = btrfs_set_inode_index(parent_inode, &index);
BUG_ON(ret);
ret = btrfs_insert_dir_item(trans, parent_root,
- pending->name, namelen,
- parent_inode->i_ino,
- &pending->root_key, BTRFS_FT_DIR, index);
+ dentry->d_name.name, dentry->d_name.len,
+ parent_inode->i_ino, &key,
+ BTRFS_FT_DIR, index);
BUG_ON(ret);
- btrfs_i_size_write(parent_inode, parent_inode->i_size + namelen * 2);
+ btrfs_i_size_write(parent_inode, parent_inode->i_size +
+ dentry->d_name.len * 2);
ret = btrfs_update_inode(trans, parent_root, parent_inode);
BUG_ON(ret);
@@ -814,22 +854,29 @@
free_extent_buffer(old);
btrfs_set_root_node(new_root_item, tmp);
- ret = btrfs_insert_root(trans, root->fs_info->tree_root, &key,
- new_root_item);
- BUG_ON(ret);
+ /* record when the snapshot was created in key.offset */
+ key.offset = trans->transid;
+ ret = btrfs_insert_root(trans, tree_root, &key, new_root_item);
btrfs_tree_unlock(tmp);
free_extent_buffer(tmp);
-
- ret = btrfs_add_root_ref(trans, parent_root->fs_info->tree_root,
- pending->root_key.objectid,
- parent_root->root_key.objectid,
- parent_inode->i_ino, index, pending->name,
- namelen);
BUG_ON(ret);
+ /*
+ * insert root back/forward references
+ */
+ ret = btrfs_add_root_ref(trans, tree_root, objectid,
+ parent_root->root_key.objectid,
+ parent_inode->i_ino, index,
+ dentry->d_name.name, dentry->d_name.len);
+ BUG_ON(ret);
+
+ key.offset = (u64)-1;
+ pending->snap = btrfs_read_fs_root_no_name(root->fs_info, &key);
+ BUG_ON(IS_ERR(pending->snap));
fail:
kfree(new_root_item);
- return ret;
+ btrfs_block_rsv_release(root, &pending->block_rsv, (u64)-1);
+ return 0;
}
/*
@@ -898,6 +945,8 @@
ret = btrfs_run_delayed_refs(trans, root, 0);
BUG_ON(ret);
+ btrfs_trans_release_metadata(trans, root);
+
cur_trans = trans->transaction;
/*
* set the flushing flag so procs in this transaction have to
diff --git a/fs/btrfs/transaction.h b/fs/btrfs/transaction.h
index 56e728d..14b3841 100644
--- a/fs/btrfs/transaction.h
+++ b/fs/btrfs/transaction.h
@@ -57,8 +57,11 @@
struct btrfs_pending_snapshot {
struct dentry *dentry;
struct btrfs_root *root;
- char *name;
- struct btrfs_key root_key;
+ struct btrfs_root *snap;
+ /* block reservation for the operation */
+ struct btrfs_block_rsv block_rsv;
+ /* extra metadata reseration for relocation */
+ int error;
struct list_head list;
};
@@ -85,11 +88,11 @@
int btrfs_end_transaction(struct btrfs_trans_handle *trans,
struct btrfs_root *root);
struct btrfs_trans_handle *btrfs_start_transaction(struct btrfs_root *root,
- int num_blocks);
+ int num_items);
struct btrfs_trans_handle *btrfs_join_transaction(struct btrfs_root *root,
- int num_blocks);
+ int num_blocks);
struct btrfs_trans_handle *btrfs_start_ioctl_transaction(struct btrfs_root *r,
- int num_blocks);
+ int num_blocks);
int btrfs_write_and_wait_transaction(struct btrfs_trans_handle *trans,
struct btrfs_root *root);
int btrfs_commit_tree_roots(struct btrfs_trans_handle *trans,
diff --git a/fs/btrfs/volumes.c b/fs/btrfs/volumes.c
index 8db7b14..d6e3af8 100644
--- a/fs/btrfs/volumes.c
+++ b/fs/btrfs/volumes.c
@@ -1097,7 +1097,7 @@
if (!path)
return -ENOMEM;
- trans = btrfs_start_transaction(root, 1);
+ trans = btrfs_start_transaction(root, 0);
key.objectid = BTRFS_DEV_ITEMS_OBJECTID;
key.type = BTRFS_DEV_ITEM_KEY;
key.offset = device->devid;
@@ -1486,7 +1486,7 @@
goto error;
}
- trans = btrfs_start_transaction(root, 1);
+ trans = btrfs_start_transaction(root, 0);
lock_chunks(root);
device->barriers = 1;
@@ -1751,9 +1751,10 @@
/* step one, relocate all the extents inside this chunk */
ret = btrfs_relocate_block_group(extent_root, chunk_offset);
- BUG_ON(ret);
+ if (ret)
+ return ret;
- trans = btrfs_start_transaction(root, 1);
+ trans = btrfs_start_transaction(root, 0);
BUG_ON(!trans);
lock_chunks(root);
@@ -1925,7 +1926,7 @@
break;
BUG_ON(ret);
- trans = btrfs_start_transaction(dev_root, 1);
+ trans = btrfs_start_transaction(dev_root, 0);
BUG_ON(!trans);
ret = btrfs_grow_device(trans, device, old_size);
@@ -2094,11 +2095,7 @@
}
/* Shrinking succeeded, else we would be at "done". */
- trans = btrfs_start_transaction(root, 1);
- if (!trans) {
- ret = -ENOMEM;
- goto done;
- }
+ trans = btrfs_start_transaction(root, 0);
lock_chunks(root);
device->disk_total_bytes = new_size;
diff --git a/fs/btrfs/xattr.c b/fs/btrfs/xattr.c
index 193b58f..007fae5 100644
--- a/fs/btrfs/xattr.c
+++ b/fs/btrfs/xattr.c
@@ -154,15 +154,10 @@
if (trans)
return do_setxattr(trans, inode, name, value, size, flags);
- ret = btrfs_reserve_metadata_space(root, 2);
- if (ret)
- return ret;
+ trans = btrfs_start_transaction(root, 2);
+ if (IS_ERR(trans))
+ return PTR_ERR(trans);
- trans = btrfs_start_transaction(root, 1);
- if (!trans) {
- ret = -ENOMEM;
- goto out;
- }
btrfs_set_trans_block_group(trans, inode);
ret = do_setxattr(trans, inode, name, value, size, flags);
@@ -174,7 +169,6 @@
BUG_ON(ret);
out:
btrfs_end_transaction_throttle(trans, root);
- btrfs_unreserve_metadata_space(root, 2);
return ret;
}