jfs: flush journal completely before releasing metadata inodes
commit1c8007b0769d37aa5fcb343b383b0af89ade2f71
authorDave Kleikamp <dave.kleikamp@oracle.com>
Mon, 1 Aug 2011 17:41:00 +0000 (1 12:41 -0500)
committerDave Kleikamp <dave.kleikamp@oracle.com>
Mon, 1 Aug 2011 17:41:00 +0000 (1 12:41 -0500)
treedc7eecc1229a417156edf353ef269e08bf8795f0
parent3da3f872aa175f59e20766ed30aaea67fd4fa7d1
jfs: flush journal completely before releasing metadata inodes

This fixes a race during unmount. We need to not only make sure that
the journal is completely written, but that the metadata changes make
it to disk before releasing ipimap and ipbmap.

Signed-off-by: Dave Kleikamp <dave.kleikamp@oracle.com>
fs/jfs/jfs_umount.c