From mboxrd@z Thu Jan  1 00:00:00 1970
Return-Path: <gcc-return-106690-listarch-gcc=gcc.gnu.org@gcc.gnu.org>
Received: (qmail 31045 invoked by alias); 16 Dec 2004 14:14:42 -0000
Mailing-List: contact gcc-help@gcc.gnu.org; run by ezmlm
Precedence: bulk
List-Archive: <http://gcc.gnu.org/ml/gcc/>
List-Post: <mailto:gcc@gcc.gnu.org>
List-Help: <http://gcc.gnu.org/ml/>
Sender: gcc-owner@gcc.gnu.org
Received: (qmail 29797 invoked from network); 16 Dec 2004 14:13:19 -0000
Received: from unknown (HELO NUTMEG.CAM.ARTIMI.COM) (217.40.111.177)
  by sourceware.org with SMTP; 16 Dec 2004 14:13:19 -0000
Received: from mace ([192.168.1.25]) by NUTMEG.CAM.ARTIMI.COM with Microsoft SMTPSVC(6.0.3790.211);
	 Thu, 16 Dec 2004 14:11:34 +0000
From: "Dave Korn" <dave.korn@artimi.com>
To: "'Alexandre Oliva'" <aoliva@redhat.com>,
	"'Richard Henderson'" <rth@redhat.com>
Cc: "'Bernd Schmidt'" <bernds_cb1@t-online.de>,
	<gcc@gcc.gnu.org>,
	<gcc-patches@gcc.gnu.org>
Subject: RE: [RFC] avoiding too narrow register classes in reload
Date: Thu, 16 Dec 2004 14:14:00 -0000
MIME-Version: 1.0
Content-Type: text/plain;
	charset="US-ASCII"
Content-Transfer-Encoding: 7bit
In-Reply-To: <or6533s4uc.fsf@livre.redhat.lsd.ic.unicamp.br>
Message-ID: <NUTMEGVtiQJYI02PrKl000007d9@NUTMEG.CAM.ARTIMI.COM>
X-OriginalArrivalTime: 16 Dec 2004 14:11:34.0562 (UTC) FILETIME=[2641F020:01C4E379]
X-SW-Source: 2004-12/txt/msg00652.txt.bz2

> -----Original Message-----
> From: gcc-owner On Behalf Of Alexandre Oliva
> Sent: 15 December 2004 20:07

> On Dec 15, 2004, Richard Henderson wrote:
> 

> > Ok, except,

> >> -		  if (reg_class_size[this_alternative[i]] == 1
> >> +		  if (reg_class_size[this_alternative[i]] > 0
> >> +		      && SMALL_REGISTER_CLASS_P (this_alternative[i])
> 
> > reg_class_size check redundant with SMALL_REGISTER_CLASS_P.
> 
> I thought I'd leave it in for, like, documentation purposes, just in
> case someone reworks SMALL_REGISTER_CLASS_P such that it can match
> even for 0-sized classes, but fine.  


  As long as it's not in a critical inner loop, I always feel that a couple of
memory accesses and a test-and-branch are a small price to pay for having code
that clearly self-documents the intent behind it.  But hey, no big deal :)


    cheers, 
      DaveK
-- 
Can't think of a witty .sigline today....