RE: UP2000 crashes -- update

["Greg Lindahl" <glindahl hpti com>]

> > Basically, setup_sigcontext is getting a pointer with illegal alignment.
>
> Impossible.

I'm happy to be wrong if we find the bug ;-)

With 2.2.13 and 2.2.10, the code gives those exception messages and the
machine claims its out of fds. With 2.2.5-22, it gives the exception
messages and doesn't run out of fds. The code may be
running out of swap space. The code doesn't die with a SEGV under any of
these kernels, which is what's supposed to happen to programs that run out
of stack space -- and I think I've observed segvs under 2.2.5.

Perhaps that kernel routine doesn't check to see that the stack is
exhausted? Or does that exception always mean unaligned access? Or is it
just any exception and it guesses that it means unaligned access?

-- g