This is the mail archive of the
gcc-bugs@gcc.gnu.org
mailing list for the GCC project.
Re: a long-standing reload bug
- To: Zack Weinberg <zack at rabi dot columbia dot edu>
- Subject: Re: a long-standing reload bug
- From: Jeffrey A Law <law at cygnus dot com>
- Date: Fri, 30 Oct 1998 00:55:47 -0700
- cc: egcs-bugs at cygnus dot com
- Reply-To: law at cygnus dot com
In message <199810300749.CAA03544@blastula.phys.columbia.edu>you write:
> On Thu, 29 Oct 1998 23:23:44 -0700, Jeffrey A Law wrote:
> >
> > In message <199810292203.RAA00552@blastula.phys.columbia.edu>you write:
> > >
> > > Now that Bernd's reload changes are all in, it's appropriate to bring
> > > this up again. egcs/CVS of last night (around 10pm EDT) gets an ICE
> > > on the appended code when compiled with any optimization:
> > >
> > > ../../../egcs/gcc/reload.c:4403: Internal compiler error in function
> > > find_reloads_toplev
> > >
> > > The CVS mainline has been doing this since shortly after 1.1 (I
> > > think). It doesn't happen with 1.0.3. I don't have 1.1 around to
> > > test. Platform is x86. The code comes from glibc's math tests.
> > >
> >This is a known problem. We have two choices right now.
> >
> >Generate incorrect code for some cases
> >
> >Abort for some cases
> >
> >Until someone can fix the problem correctly we will continue to abort inst
> ead
> >of silently generating incorrect code.
>
> Ok. Does 1.1 have this problem? We're coming in on a release and it
> would be bad if we had an interaction bug in the public headers.
It has one of the two bugs, I don't remember which. THere are no plans to
change the behavior for 1.1 for this matter. This is not a critical problem.
jeff