This is the mail archive of the
gcc-bugs@gcc.gnu.org
mailing list for the GCC project.
Re: Problem with gcc
- To: Michael Falck <mf at bocad dot com>
- Subject: Re: Problem with gcc
- From: Martin Kahlert <martin dot kahlert at mchp dot siemens dot de>
- Date: Fri, 5 Nov 1999 09:02:30 +0100
- Cc: egcs-bugs at egcs dot cygnus dot com
- References: <199911041220.JAA20570@sleipnir.valparaiso.cl> <382286DF.F0BFA0D6@bocad.com>
- Reply-To: martin dot kahlert at mchp dot siemens dot de
Quoting Michael Falck (mf@bocad.com):
> Horst von Brand wrote:
> >
> > Michael Falck <mf@bocad.com> said:
> > > we have some little trouble with gcc. During rebuild of our sources
> > > we get sometimes the following message:
> > > /usr1/17500/src/boclib/helfer/iabsrl.c
> > > gcc: Internal compiler error in `execute', at gcc.c:2327
> > > Please submit a full bug report.
> > > The binary seems to be ok. Compiling this file alone, doesn't produce
> > > this effect. A change of the hardware doesn't cure it. This
> > > effect often appears on short files, but from rebuild to rebuild
> > > different files seem to be affected.
> >
> > That points directly to hardware failures. They might not be readily
> > apparent, gcc is an excellent memory tester.
> > --
> > Horst von Brand vonbrand@sleipnir.valparaiso.cl
> > Casilla 9G, Viņa del Mar, Chile +56 32 672616
>
> As I mentioned before, we already tested this on other hardware with the
> same
> effect. Hardware does not explain, why this effect always happens at the
> same
> place in gcc (execute:2327), why only small files are affected (up to
> 100
> of effective code) longer ones compile without such problems, why the
> code
> is usable, why absoft compilers don't have that kind of problem and why
> that system is
> working fine below other heavy duty tasks. I accept that hardware may be
> indicated
> because the affected files compile ok if done one by one, and in two
> succeding
> batch runs different files may be affected.
>
> So for me several answers are possible:
> - problem solved with one of the next releases :-)
> - an explanation what happens in the compiler that
> makes visible, why the code is usable and
> gives us hints where to continue with
> investigations :-)
> - unusable short shuts, which means unable to help :-(
> please be confirmed, that we already invested time
> in that problem and checked those easy things before
> we issue such a bug report.
I once got very similar failures. It turned out, it was a NFS Problem.
If you compile on Linux-2.2.(don't know the version) and compile on
a NFS mounted filesystem exported by a Solaris 2.5 machine, you will
run into trouble.
Perhaps, it's something like that.
Bye,
Martin.
--
The early bird gets the worm. If you want something else for
breakfast, get up later.