[regression] stack overflow in xfs_buf_iodone_callbacks

[Christoph Hellwig]

On Thu, Jun 21, 2012 at 07:18:03PM +1000, Dave Chinner wrote:
> Folks,
> 
> I just had a stack overflow in the delayed write buffer error
> handling with a shut down filesystem:

I've looked at this a bit more, and it seems the effect really can't
be an XFS shutdown.  We never do the shutdown check inside
xfs_buf_iorequest.  The issue obviously is real, but could it be that
you had an actual persistent I/O error on the underlying device?