[Mono-devel-list] Valgrind and mono

[Jaroslaw Kowalski]

I get a lot of errors like this (Fedora Core 1 using 2.4.22-1.2174.ntpl
kernel and glibc-2.3.2-101.4) with valgrind-2.0.0:

==23970== Conditional jump or move depends on uninitialised value(s)
==23970==    at 0x2160FD: GC_push_all_eager (mark.c:1469)
==23970==    by 0x217015: GC_push_current_stack (mark_rts.c:488)
==23970==    by 0x21B9DC: GC_generic_push_regs (mach_dep.c:450)
==23970==    by 0x217130: GC_push_roots (mark_rts.c:628)
==23970==
==23970== Conditional jump or move depends on uninitialised value(s)
==23970==    at 0x216102: GC_push_all_eager (mark.c:1469)
==23970==    by 0x217015: GC_push_current_stack (mark_rts.c:488)
==23970==    by 0x21B9DC: GC_generic_push_regs (mach_dep.c:450)
==23970==    by 0x217130: GC_push_roots (mark_rts.c:628)
==23970==
==23970== valgrind's libpthread.so: KLUDGED call to: __libc_current_sigrtmin
==23970== Warning: bad signal number -1 in __NR_sigaction.

Does it mean I've found a problem with mono? gcc? libgc? How should I report
it?

Jarek

----- Original Message ----- 
From: "Dick Porter" <dick at ximian.com>
To: <mono-devel-list at lists.ximian.com>
Sent: Tuesday, May 18, 2004 3:32 PM
Subject: Re: [Mono-devel-list] Valgrind and mono


> On Tue, 2004-05-18 at 13:53, Jaroslaw Kowalski wrote:
> > Hi guys!
> >
> > Does valgrind (memory debugging tool for Linux) run with mono?
> >
> > On valgrind's page http://valgrind.kde.org/ there's a note that
> > valgrind is used by Mono. I have a highly unstable mono setup right
> > now (ASP.NET doesn't work for me in a predictable manner and I suspect
> > race conditions and/or memory corruption), so I've tried to run
> > valgrind (both 2.0.0 and 2.1.2) on mod-mono-server.exe, but it always
> > crashed. Most errors reported seem to be GC- and pthread-related.
> >
> > Can someone provide a step-by-step recipe for running mono with
> > valgrind ?
>
> It works out of the box for me, on rh9 with latest cvs mono and
> valgrind-2.1.1.
>
> You might need to a) make sure mono's configure finds the
> valgrind/memcheck.h file so the jit doesn't try and call a couple of
> pthread functions that valgrind doesn't emulate, and b) up the pthread
> stack size in coregrind/vg_include.h.
>
> - Dick
>
>
> _______________________________________________
> Mono-devel-list mailing list
> Mono-devel-list at lists.ximian.com
> http://lists.ximian.com/mailman/listinfo/mono-devel-list
>