Could you tell somehting more, why cpu_swithchto on popping ebx (or other registers for this matter close the spot in code) from stack would exist in such way as it does right now? Something manages to corrupt stack to this amount/something else?
On Sat, Jul 10, 2010 at 5:51 PM, Theo de Raadt <dera...@cvs.openbsd.org> wrote: >> On Sat, Jul 10, 2010 at 12:48 AM, Theo de Raadt <dera...@cvs.openbsd.org> >> wrote: >> > Can you please try a kernel with softraid disabled? >> > >> Yes, I will try. Any additional requests/specific procedure how it >> should be tried? > > I believe I finally understand the bug, and it is a softraid bug.
