Here's the situation -- kdb build using XFS cvs tree from 9/5/02. 11 drive
RAID5 with spare setup, created an lvm volume on the RAID. Failed a drive
in the RAID5 during the sync, then created an xfs filesystem on the volume.
When mounting the volume, the init processes oopses.
In kdb I found the oops was inside kmem_cache_alloc(). Somehow, the
slabs_free structure inside of names_cache's kmem_cache_t had become
corrupted. The other list heads in the structure seemed fine.
Any idea what could possibly cause this?
Attached is the back trace of init, of the mount, of the raid5 recovery
daemon, dissasembly of the relevant part of kmem_cache_alloc, and the
kmem_cache_t structure.
Dale Stephenson
steph@xxxxxxxxxxxxxx
oops.txt
Description: Binary data
|