This is the mail archive of the
gcc-patches@gcc.gnu.org
mailing list for the GCC project.
Re: patch to alias.c checked in
- To: Bernd Schmidt <bernds at redhat dot com>
- Subject: Re: patch to alias.c checked in
- From: Franz Sirl <Franz dot Sirl-kernel at lauterbach dot com>
- Date: Fri, 19 Jan 2001 14:47:26 +0100
- Cc: Joern Rennecke <amylaar at cambridge dot redhat dot com>,<gcc-patches at gcc dot gnu dot org>
- References: <5.0.2.1.2.20010119134425.03294ea0@mail.lauterbach.com>
At 14:09 2001-01-19, Bernd Schmidt wrote:
>On Fri, 19 Jan 2001, Franz Sirl wrote:
>
> > At 12:50 2001-01-19, Bernd Schmidt wrote:
> > >On Fri, 19 Jan 2001, Franz Sirl wrote:
> > > >
> > > > Was this exposed by Bernd's recent work on init_alias_analysis() to
> > > > recognize more stuff? Or is this an older bug? I'm asking cause I'm
> seeing
> > > > miscompilation of glibc-2.2.1 on PPC that is fixed if I back out Bernds
> > > patch.
> > >
> > >Testcase?
> >
> > Yes, sort of :-/. time_output() in locale/programs/ld-time.c gets
> > miscompiled, fixed either by -fno-schedule-insns or by backing out the big
> > hunk to init_alias_analysis in your alias.c patch on 3/4 Jan. The problem
> > is that I haven't been able to produce an executable testcase yet, as the
> > symptom is that time_output() writes a buggy LC_TIME file. And
> > time_output() is a rather lengthy routine with large differences in .sched
> > depending on your patch being in or out...
>
>Have you updated your CVS tree recently? I've already backed out most of
>that patch.
Yes, I saw that, but this only affected the REGNO<->ORIGINAL_REGNO changes,
what breaks things for me is:
(init_alias_analysis): Add more cases to detect known values.
And this hunk wasn't backed out yet. I made a diff between revision 1.110
and 1.112 of of alias.c, edited out all but the big hunk to
init_alias_analysis() and reverse applied it on my mainline tree. That
fixed things for me.
> > If the .i file is enough for you though, I can post it later today. If you
> > are testing glibc natively on PPC yourself, be sure to apply one of
> > Alexandres recent patches to reload1.c (fixes a glibc tst-random failure,
> > see the reload_cse_move2add thread) and my patch UNKNOWN handling patch for
> > comparison_dominates_p() (fixes various glibc math testcases).
>
>I have no access to a ppc system right now. You should probably post the .i
>file unless the problem is already fixed.
Ok, will send it when I'm home again.
Franz.