From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (qmail 16205 invoked by alias); 7 Aug 2007 15:03:04 -0000 Received: (qmail 16100 invoked by uid 22791); 7 Aug 2007 15:03:03 -0000 X-Spam-Check-By: sourceware.org Received: from mx1.redhat.com (HELO mx1.redhat.com) (66.187.233.31) by sourceware.org (qpsmtpd/0.31) with ESMTP; Tue, 07 Aug 2007 15:02:55 +0000 Received: from int-mx1.corp.redhat.com (int-mx1.corp.redhat.com [172.16.52.254]) by mx1.redhat.com (8.13.1/8.13.1) with ESMTP id l77F1eho015935; Tue, 7 Aug 2007 11:01:40 -0400 Received: from pobox.fab.redhat.com (pobox.fab.redhat.com [10.33.63.12]) by int-mx1.corp.redhat.com (8.13.1/8.13.1) with ESMTP id l77F1dx0003594; Tue, 7 Aug 2007 11:01:39 -0400 Received: from [10.33.6.4] (vpn-6-4.fab.redhat.com [10.33.6.4]) by pobox.fab.redhat.com (8.13.1/8.13.1) with ESMTP id l77F1c6N019434; Tue, 7 Aug 2007 11:01:38 -0400 Message-ID: <46B88950.8080000@redhat.com> Date: Tue, 07 Aug 2007 15:03:00 -0000 From: Nick Clifton User-Agent: Thunderbird 1.5.0.10 (X11/20070301) MIME-Version: 1.0 To: Nick Clifton , echristo@apple.com, gcc-patches@gcc.gnu.org, richard@codesourcery.com Subject: Re: RFA/RFC: GCC has an incorrect length for the ldc1 instruction References: <87hcnb8ybn.fsf@firetop.home> In-Reply-To: <87hcnb8ybn.fsf@firetop.home> Content-Type: text/plain; charset=ISO-8859-1; format=flowed Content-Transfer-Encoding: 7bit X-IsSubscribed: yes 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: 2007-08/txt/msg00436.txt.bz2 Hi Richard, > Unfortunately, this isn't quite right. DFmode moves still take > 2 instructions when the target is a GPR. For MIPS I, they will > also take 2 instructions when the target is an FPR. Darn, I was afraid that there might be a case like that. > I think instead that mips_fetch_insns should try to prove that > the move does not need to be split, and tell mips_address_insns > about it. Does the patch below work for you? It does. I also tested the patch on the slightly older sources that make up the toolchain that was sent to our customer (which compiles the test case with only one ldc1 instruction between a mfhi and a mult instruction) and it fixes that toolchain as well. Cheers Nick