In article <200404261822.UAA0000078331@zel459.zel.kfa-juelich.de>,
Matthias Drochner <M.Drochner@fz-juelich.de> wrote:
>
>wileyc@rezrov.net said:
>> it's interesting that it broke after the siginfo stuff was committed.
>
>These siginfo changes only repaired cases where the previous code
>did _definitely_ panic. So, if I didn't botch something, things
>shouldn't have gotten worse.
>
>> Nobody has actually tested other mips3 platforms?
>
>I've tested these changes on a pmax (R3000+3010). While the si_code
>isn't correct for fp errors, the box didn't crash in any of the tests
>I tried.
>
>I don't know what a "cache error" means exactly, but shouldn't
>an error caused by invalid memory accesses look diffently?
>So it might be rather platform specific...

The siginfo changes prolly need more stack; are you sure we are not blowing
the kernel stack? Just guessing.

christos