From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (qmail 29996 invoked by alias); 19 Jul 2010 21:53:59 -0000 Received: (qmail 29985 invoked by uid 22791); 19 Jul 2010 21:53:58 -0000 X-SWARE-Spam-Status: No, hits=-1.8 required=5.0 tests=AWL,BAYES_00,T_RP_MATCHES_RCVD X-Spam-Check-By: sourceware.org Received: from mail.codesourcery.com (HELO mail.codesourcery.com) (38.113.113.100) by sourceware.org (qpsmtpd/0.43rc1) with ESMTP; Mon, 19 Jul 2010 21:53:53 +0000 Received: (qmail 17568 invoked from network); 19 Jul 2010 21:53:51 -0000 Received: from unknown (HELO ?84.152.189.214?) (bernds@127.0.0.2) by mail.codesourcery.com with ESMTPA; 19 Jul 2010 21:53:51 -0000 Message-ID: <4C44C948.3050801@codesourcery.com> Date: Mon, 19 Jul 2010 21:53:00 -0000 From: Bernd Schmidt User-Agent: Mozilla/5.0 (X11; U; Linux i686; en-US; rv:1.9.1.10) Gecko/20100625 Thunderbird/3.0.5 MIME-Version: 1.0 To: gcc-patches@gcc.gnu.org, rdsandiford@googlemail.com Subject: Re: Extend widening_mul pass to handle fixed-point types References: <87fwzhro8i.fsf@firetop.home> <4C436835.20307@codesourcery.com> <87oce3s4kv.fsf@firetop.home> In-Reply-To: <87oce3s4kv.fsf@firetop.home> Content-Type: text/plain; charset=ISO-8859-1 Content-Transfer-Encoding: 7bit Mailing-List: contact gcc-patches-help@gcc.gnu.org; run by ezmlm Precedence: bulk List-Id: List-Archive: List-Post: List-Help: Sender: gcc-patches-owner@gcc.gnu.org X-SW-Source: 2010-07/txt/msg01534.txt.bz2 On 07/19/2010 08:34 PM, Richard Sandiford wrote: > I did wonder about splitting the stuff out into a separate function. > The advantage of not doing is that it avoids duplicate work in cases > where we visit the macc first, and where the WIDEN_MULT_EXPR _can_ > be used. I thought that would be quite a common case. Not sure about that - normally I'd expect the multiply and add to appear roughly in the order they're executed. > Here's a version that splits it out and drops the two-at-once thing. > Only lightly tested, but does it look better? That's pretty much what I had in mind. Ok if it passes testing. Bernd