From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (qmail 25791 invoked by alias); 9 Apr 2003 09:13:45 -0000 Mailing-List: contact gcc-help@gcc.gnu.org; run by ezmlm Precedence: bulk List-Archive: List-Post: List-Help: Sender: gcc-owner@gcc.gnu.org Received: (qmail 25781 invoked from network); 9 Apr 2003 09:13:44 -0000 Received: from unknown (HELO nikam.ms.mff.cuni.cz) (195.113.18.106) by sources.redhat.com with SMTP; 9 Apr 2003 09:13:44 -0000 Received: from camelot.ms.mff.cuni.cz (kampanus.ms.mff.cuni.cz [195.113.18.107]) by nikam.ms.mff.cuni.cz (Postfix) with SMTP id 760784DECF; Wed, 9 Apr 2003 11:13:44 +0200 (CEST) Received: by camelot.ms.mff.cuni.cz (sSMTP sendmail emulation); Wed, 9 Apr 2003 11:13:45 +0200 Date: Wed, 09 Apr 2003 11:25:00 -0000 From: Jan Hubicka To: Eric Botcazou Cc: Jan Hubicka , gcc@gcc.gnu.org Subject: Re: Reload bug Message-ID: <20030409091345.GG9102@kam.mff.cuni.cz> References: <200304081937.16859.ebotcazou@libertysurf.fr> <200304090909.30614.ebotcazou@libertysurf.fr> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Disposition: inline In-Reply-To: <200304090909.30614.ebotcazou@libertysurf.fr> User-Agent: Mutt/1.3.28i X-SW-Source: 2003-04/txt/msg00356.txt.bz2 > > Thu Oct 17 17:14:07 CEST 2002 Jan Hubicka > > > > PR opt/7630 > > * reload.c (reload_inner_reg_of_subreg): New argument output; > > (push_reload): Update call. > > This patch is really bogus, at least on the 3.2 branch: sure, GCC doesn't ICE > when compiling pr7630.c any more but now it silently produces wrong code! Urm. The use of SSE is completely broken idea here.. Will have to look why register allocator decided that. > > What you have done is to teach the reload pass to silently transform any > non-accessible part of a multi-word register into the low part. In the case the subreg is undable (ie it is in the SSE), or can this happen in the integer register too? > > Please revert on the 3.2 branch or consider applying the correction I posted > (which of course brings back the ICE for PR opt/7630). Or we can backport the CANNOT_CHANGE_MODE_PATCH. I would be happy about any alternative for 3.2 Honza > > Thanks. > > -- > Eric Botcazou