reproducible xfs/vmap oops

[Christoph Hellwig]

On Tue, Feb 03, 2009 at 11:05:15AM -0500, Christoph Hellwig wrote:
> On Wed, Feb 04, 2009 at 03:03:13AM +1100, Nick Piggin wrote:
> > Hmm, seems similar to the other one Dave saw. I suppose it could
> > be due to bad parameters being passed to vmap API, but quite
> > likely it could be a bug in vmap as well.
> > 
> > Any chance you can print out va_start and va_end from va and tmp
> > before going bug? (or if you like, I can try reproduce it with
> > xfsqa if there is a URL for it -- a quick google didn't help me).
> 
> It's at http://git.kernel.org/?p=fs/xfs/xfstests-dev.git
> 
> Right now I'm doing two verification runs with the current codebase
> and just that patch reverted as a second verification.  After that
> I can do a patch with the debug printks.

With that patch reverted it completes three complete xfsqa runs without
problems.  Now running with the debug printks instead.