Btrfs: don't log the inode in file_write while growing the file
authorChris Mason <chris.mason@oracle.com>
Sun, 28 Jun 2009 01:06:22 +0000 (21:06 -0400)
committerChris Mason <chris.mason@oracle.com>
Thu, 2 Jul 2009 17:41:16 +0000 (13:41 -0400)
fs/btrfs/file.c

index 126477e..7c3cd24 100644 (file)
@@ -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: