[PATCH] Btrfs: don't leak transaction in btrfs_sync_file()

[PATCH] Btrfs: don't leak transaction in btrfs_sync_file()

[Filipe David Borba Manana]

In btrfs_sync_file(), if the call to btrfs_log_dentry_safe() returns
a negative error (for e.g. -ENOMEM via btrfs_log_inode()), we would
return without ending/freeing the transaction.
---
 fs/btrfs/file.c |    4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)

diff --git a/fs/btrfs/file.c b/fs/btrfs/file.c
index 5ba87b0..d12107e 100644
--- a/fs/btrfs/file.c
+++ b/fs/btrfs/file.c
@@ -1859,8 +1859,8 @@ int btrfs_sync_file(struct file *file, loff_t start, loff_t end, int datasync)
 
 	ret = btrfs_log_dentry_safe(trans, root, dentry);
 	if (ret < 0) {
-		mutex_unlock(&inode->i_mutex);
-		goto out;
+		/* Fallthrough and commit/free transaction. */
+		ret = 1;
 	}
 
 	/* we've logged all the items and now have a consistent
-- 
1.7.9.5

[PATCH v2] Btrfs: don't leak transaction in btrfs_sync_file()

[Filipe David Borba Manana]

In btrfs_sync_file(), if the call to btrfs_log_dentry_safe() returns
a negative error (for e.g. -ENOMEM via btrfs_log_inode()), we would
return without ending/freeing the transaction.

Signed-off-by: Filipe David Borba Manana <fdmanana@gmail.com>
---

V2: If btrfs_log_dentry_safe() returns error, don't fall through because
    that will override the final return value, and can make us return
    success (0) instead of an error.

 fs/btrfs/file.c |    1 +
 1 file changed, 1 insertion(+)

diff --git a/fs/btrfs/file.c b/fs/btrfs/file.c
index 5ba87b0..8c305f5 100644
--- a/fs/btrfs/file.c
+++ b/fs/btrfs/file.c
@@ -1860,6 +1860,7 @@ int btrfs_sync_file(struct file *file, loff_t start, loff_t end, int datasync)
 	ret = btrfs_log_dentry_safe(trans, root, dentry);
 	if (ret < 0) {
 		mutex_unlock(&inode->i_mutex);
+		btrfs_end_transaction(trans, root);
 		goto out;
 	}
 
-- 
1.7.9.5