Andre Noll wrote:
> Hi
> 
> MATLAB triggers the following bug on both of our new 16-way opteron
> machines (64G Ram): The same kernel is running with no problems on a
> bunch of smaller (8-way, 4-way, max 32G Ram) cluster nodes.
> 
> Any hints?
> Andre
> 
> 
> ----------- [cut here ] --------- [please bite here ] ---------
> Kernel BUG at ...aid0/home/maan/scm/stable/linux-2.6.18.y/mm/rmap.c:522

Ah, this old thing. I hope it is repeatable?

What we really want is the bit before this, the "Eeek! page_mapcount went
negative" part.

It is also nice if we can work out where the page actually came from. The
following attached patch should help out a bit with that, if you could
run with it?

Thanks a lot for reporting, this is very useful.

Nick

-- 
SUSE Labs, Novell Inc.