This is the mail archive of the
gcc-patches@gcc.gnu.org
mailing list for the GCC project.
Re: reg-stack strikes back II
- To: Jan Hubicka <jh at suse dot cz>
- Subject: Re: reg-stack strikes back II
- From: Richard Henderson <rth at redhat dot com>
- Date: Thu, 7 Jun 2001 12:49:54 -0700
- Cc: gcc-patches at gcc dot gnu dot org
- References: <20010607165340.B20748@atrey.karlin.mff.cuni.cz>
On Thu, Jun 07, 2001 at 04:53:40PM +0200, Jan Hubicka wrote:
> Hi,
> unfortunately the previous patch didn't fix the problem.
> The split_all_insns already has code to deal with broken flowgraph, since it is
> used after reload, but it is rather humorous - it simply allows loops to drop
> off basic block to the end. That has potential to bring problem, as deleted
> instruction may be traversed.
>
> I've fixed it by making split_insn independent on basic block information
> if upd_info is 0.
> This fixes all problems I am seeing, but in case I will not get response
> until mips bootstrap finishes (i386/sparc already did), I will probably
> revert my last toplev.c patch and wait for approval with this one before
> installing it back.
>
> I apologize for so frequent breakage this day :(
> Really hope it will not repreat
>
> Honza
>
> Thu Jun 7 16:48:39 CEST 2001 Jan Hubicka <jh@suse.cz>
> * recog.c (split_insn): Break out from ...
> (split_all_insns): ... here; do not use basic block information
> when it is broken.
Ok.
r~