*/
clear_bit(BTRFS_INODE_NEEDS_FULL_SYNC,
&BTRFS_I(inode)->runtime_flags);
+ /*
+ * An ordered extent might have started before and completed
+ * already with io errors, in which case the inode was not
+ * updated and we end up here. So check the inode's mapping
+ * flags for any errors that might have happened while doing
+ * writeback of file data.
+ */
+ ret = btrfs_inode_check_errors(inode);
inode_unlock(inode);
goto out;
}