This is the mail archive of the
gcc-bugs@gcc.gnu.org
mailing list for the GCC project.
Re: Abort in redirect_edge_and_branch on vax at line 866
- From: Jan Hubicka <jh at suse dot cz>
- To: law at redhat dot com
- Cc: Jan Hubicka <jh at suse dot cz>, John David Anglin <dave at hiauly1 dot hia dot nrc dot ca>, gcc-bugs at gcc dot gnu dot org, rth at redhat dot com, gcc-patches at gcc dot gnu dot org
- Date: Thu, 10 Jan 2002 10:47:22 +0100
- Subject: Re: Abort in redirect_edge_and_branch on vax at line 866
- References: <20020109102533.GA4091@atrey.karlin.mff.cuni.cz> <22097.1010616954@porcupine.cygnus.com>
> > The attached patch should do the trick for you.
> > Bootstrapped/regtested i386. OK to install?
> >
> > Wed Jan 9 11:23:03 CET 2002 Jan Hubicka <jh@suse.cz>
> > * gcse.c (cprop_jump): Delete insn if simplified jump is no-op.
> Be VERY VERY careful here. Deleting the insn is likely to cause all
I tried to be - to my understanding all information gcse holds at that
point is the global information for each basic block. As I am not
making the global properties worse, I am just improving them, the datastructures
remain to be conservative approximation of the real thing making all other
changes valid.
I tought the point of not deleting the instruction has been to keep CFG
consistent, really, but possibly I do miss something important but even
second pass didn't show something wrong to me.
Honza
> kinds of problems with the datastructures that gcse.c uses. That was
> the entire point behind deferring deletion of jumps until after gcse.c
> was finished.
>
> I would like to see an analysis of the code before this patch is installed.
> jeff