From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (qmail 10188 invoked by alias); 7 Mar 2012 18:42:36 -0000 Received: (qmail 10178 invoked by uid 22791); 7 Mar 2012 18:42:35 -0000 X-SWARE-Spam-Status: No, hits=2.9 required=5.0 tests=AWL,BAYES_50,FREEMAIL_FROM,RCVD_IN_DNSWL_NONE,T_RP_MATCHES_RCVD X-Spam-Check-By: sourceware.org Received: from blu0-omc4-s19.blu0.hotmail.com (HELO blu0-omc4-s19.blu0.hotmail.com) (65.55.111.158) by sourceware.org (qpsmtpd/0.43rc1) with ESMTP; Wed, 07 Mar 2012 18:42:23 +0000 Received: from BLU124-W24 ([65.55.111.135]) by blu0-omc4-s19.blu0.hotmail.com with Microsoft SMTPSVC(6.0.3790.4675); Wed, 7 Mar 2012 10:42:21 -0800 Message-ID: From: Peng Zhao To: gcc gcc , Subject: questions about dependence analysis in CG Date: Wed, 07 Mar 2012 18:42:00 -0000 In-Reply-To: References: Content-Type: text/plain; charset="iso-8859-1" Content-Transfer-Encoding: quoted-printable MIME-Version: 1.0 X-IsSubscribed: yes Mailing-List: contact gcc-help@gcc.gnu.org; run by ezmlm Precedence: bulk List-Id: List-Archive: List-Post: List-Help: Sender: gcc-owner@gcc.gnu.org X-SW-Source: 2012-03/txt/msg00087.txt.bz2 Hello, I am a little confused by the code in sched-deps.c.=20 1. what is the purpose of the reg_note and ds_t? I see the function dk_to_d= s, and the comment in sched-int.h, "Dependence on instruction can be of mul= tiple types (e.g. true and output). This fields enhance REG_NOTE_KIND information of th= e dependence.". What is enhanced from REG_DEP_TRUE to DEP_TRUE? in struct _dep, I only find the dependence that is related with REG_NOTE (o= r registers). Even ds_t only as DEP_TRUE/DEP_OUTPUT/DEP_ANTI. Does this mea= n that gcc doesn't differentiate reg dependence and memory dependence? Where can I find information about memory dependences such as a write and r= ead on elementa[100] ? Sometimes meory dependence should be handled differe= nly with reg dependece in CG, such as the latency between the producer and = the consumer. 2. What different purpose are haifa_note_mem_dep and haifa_note_dep for? thanks=20=09=09=20=09=20=20=20=09=09=20=20