On Fri, Jul 4, 2008 at 11:34 AM, Hugo Vanwoerkom <hvw59601@care2.com> wrote:
Thanks Florian. But funny is that I can run it OK in another Sid partition
on this box that runs a recompiled 2.6.24 Debian i386 kernel.
This Sid partition runs a home rolled 2.6.25.9 and has the problem...
I don't see how it could be the kernel exposing this issue.
run audacious inside gdb and find out where it hits the illegal
instruction - the few times I've encountered this problem, it's been
because the binary was compiled with features the processor doesn't
support - SSE2 for instance. Some people assume that everyone in
i386-land have processors that can handle that stuff, and mine
couldn't and I got bit by the last nvidia driver update in lenny. Now
that I have an amd64, I don't have to worry about this, but that's
beside the point. :) For the last seven years though, I had a
pre-Palomino vintage AMD Athlon Thunderbird processor. No SSE2 on that
thing.
When you do a gdb, then run the program as usual and you'll see the
offending instruction by doing a disassemble $pc-32 $pc+32 which
allows you to see the context. It'll probably be a movaps or some
other instruction your cpu can't do.