-
Andrew Morton authored
ext3_sync_fs will start a commit and will wait on that commit. This means that on its return, all journalled file data has been dirtied and exposed to sync_inodes_sb(). Which is sufficient to fix the umount data loss problem.
012af46c