Re: xfs oops (CVS-2004-05-15_05:00_UTC)

[Krzysztof Rusocki <kszysiu@xxxxxxxxxxxxxxxxxxxx>]

On Thu, Jun 24, 2004 at 01:14:41PM +0100, Christoph Hellwig wrote:
> > - current CVS without 1.1371.750.18 _and_ undelay changes - no problems
> >   (tested 10 hours)

Ugh, this ^^^ seems to be no longer accurate. I'm guessing that
these changes are not neccessarily to blame then (wild guess).

> > - current CVS without 1.1371.750.18, 1.1587.5.6 1.1722.10.100 and undelay
> >   changes - no problems (tested 14 hours) (*)

I'll test (*) again and vanilla 2.6.4 (as Nathan suggested)...

>  
> +     ASSERT(list_empty(&pb->pb_list));
>       ASSERT(list_empty(&bp->pb_hash_list));
> +     ASSERT(!(bp->pb_flags & PBF_DELWRI));

For the record: I changed these assertions to BUG_ON(! )s
(as in Nathan's patch) since I do not have CONFIG_XFS_DEBUG
(afaics these assertions are not enabled until CONFIG_XFS_DEBUG
is set).

But... no, these BUG_ONs were not triggered, and oops did happen.


Krzysztof