In article <16756.9291.969917.346310@mogli.lke.rad.klinikum.rwth-aachen.de>,
Markus W Kilbinger <kilbi@rad.rwth-aachen.de> wrote:
>>>>>> "Christos" == Christos Zoulas <christos@zoulas.com> writes:
>
>    Christos> No, this is a kernel bug. The kernel traps the floating
>    Christos> point operation and then has a bug handling it. The null
>    Christos> pointer bug is in the kernel.
>
>Ok, I guess then it's justified to send-pr... (to: port-cobalt?
>port-mips?)

I think port-mips, but mention you only tested it on the cobalt port.

>    Christos> With mips1, probably the compiler does not generate
>    Christos> floating point code and does the equivalent of
>    Christos> soft-float so that the buggy code does not get invoked.
>
>Is there a general problem not to use 'adequate' mips optimization on
>the cobalt port?
>

I don't think so; looks like a bug in the trap handler.

christos