Here are a couple of other things I would try under gdb:

- set breakpoints at exit and _exit, then run and see where they're being
called from
- if that doesn't work, set a breakpoint where the "Exiting" message is
printed, then use 'next' and 'single-step' from there until you get the exit
code message (tedious, but still probably faster than using valgrind)

Steve

On Sat, Apr 19, 2008 at 12:24 AM, Gabriel Michael Black <
[EMAIL PROTECTED]> wrote:

>    This might be some sort of exit code if the standard library detects
> heap corruption (bad memory management) or something similar. Your
> simulation will take a lot longer, but if you run m5 under valgrind it
> might tell you something useful. There's an error suppression file in
> util that should help get rid of errors from when valgrind is just
> confused by the python interpreter.
>
> Gabe
>
> Shoaib Akram wrote:
> > There is no error message else , Exiting at cycle xxx because all
> threads reached max insts. ABORTED. And under gdb, after the
> Exiting...message ,it says, program exited with code 03000.
> >
> > For the same configuration, sometimes it happens if I change workload.
> Sometimes, for same workload and changing system to having L3 cache.
> >
> > I am running separate applications on multiple cores.
> >
> > ---- Original message ----
> >
> >> Date: Sat, 19 Apr 2008 01:31:23 -0400
> >> From: Ali Saidi <[EMAIL PROTECTED]>
> >> Subject: Re: [m5-users] m5 Aborted,code 03000
> >> To: M5 users mailing list <[email protected]>
> >>
> >> Dumping the statistics to a file is one of the last things m5 does, so
> >> if M5 terminated abnormally you wouldn't seen any statistics. Do you
> >> have an exact error message?  I don't know of any case where we end
> >> with an error code other than 0,1, or 3 so 03000 seems a bit strange.
> >>
> >> Ali
> >>
> >>
> >> On Apr 19, 2008, at 1:06 AM, Shoaib Akram wrote:
> >>
> >>
> >>> Often times (for some workloads and configurations), my simulations
> >>> end with message Aborted but the benchmarks seems to be working. No
> >>> statistics are collected though. Under gdb it says program ended
> >>> with code 03000.
> >>> _______________________________________________
> >>> m5-users mailing list
> >>> [email protected]
> >>> http://m5sim.org/cgi-bin/mailman/listinfo/m5-users
> >>>
> >>>
> >> _______________________________________________
> >> m5-users mailing list
> >> [email protected]
> >> http://m5sim.org/cgi-bin/mailman/listinfo/m5-users
> >>
> > _______________________________________________
> > m5-users mailing list
> > [email protected]
> > http://m5sim.org/cgi-bin/mailman/listinfo/m5-users
> >
>
> _______________________________________________
> m5-users mailing list
> [email protected]
> http://m5sim.org/cgi-bin/mailman/listinfo/m5-users
>
_______________________________________________
m5-users mailing list
[email protected]
http://m5sim.org/cgi-bin/mailman/listinfo/m5-users

Reply via email to