Hello,
I am running kernel 2.6.4 on XScale/ARM with BusyBox. Quite frequently
after shutdown I have a number of orphaned inodes on the ext3
filesystem. (I have yet to figure out the precise cause of the orphans,
but I believe it has to do with umount failing at shutdown due to a bug
in BusyBox init that does not properly kill processes.)
It is my understanding the orphaned inode list should be replayed on the
next mount (after the journal is recovered), however this often fails
due to a corrupt orphaned inode list (as reported by e2fsck). In extreme
cases, then, I end up with a "clean" filesystem, but e2fsck -f fixes
tens of thousands of orphaned inodes.
What causes the corruption in the orphaned inode list itself? Is this
why the orphaned inode list is not being cleaned up on mount? Is this at
all related to the number of ext3 fixes noted in the kernel changelogs
since 2.6.4?
Thanks,
Mark
--
----------------------
Mark S. Reibert, Ph.D.
mark at reibert.com
----------------------
