>
> --45Z9DzgjV8m4Oswq
> Content-Type: text/plain; charset=iso-8859-1
> Content-Disposition: inline
> Content-Transfer-Encoding: 8bit
>
> Hello,
>
> After a couple of assertion failure-free months, here's a
> reproducible one. I can trigger this by compiling the kernel tree
> on a 500M XFS partition, on a Celeron 300 MHz 128M IDE machine.
> Attempting to unmount the filesystem after the assertion has been
> triggered hangs the umount process:
> sys_oldumount->sys_umount->do_umount->shrink_dcache_sb->iput
> ->linvfs_put_inode->vn_put->xfs_inactive->xfs_inactive_free_eofblocks
> ->xfs_ilock->xfs_ilock_ra->mraccessf->lock_wait *hang*
>
> xfs_repair 1.2.1 reveals (-n) many errors. I have the damaged
> filesystem handy for poking around with xfs_db.
>
> Kernel is 2.4.3-XFS 2001/04/04 15:42 CEST.
>
> Peter
>
We had a few days this week between good trees, I broke it on Monday or
Tuesday and the fix for this was checked in on the 5th.
The repair output would probably disappear after a mount of the
filesystem. Repair does not take into account replaying the log, only
mounting the filesystem would do this. The thread which tripped the assert
did so holding locks which are what was preventing the unmount.
Steve
|