From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (qmail 28797 invoked by alias); 4 Jan 2005 04:04:47 -0000 Mailing-List: contact gcc-bugs-help@gcc.gnu.org; run by ezmlm Precedence: bulk List-Archive: List-Post: List-Help: Sender: gcc-bugs-owner@gcc.gnu.org Received: (qmail 28656 invoked by alias); 4 Jan 2005 04:04:38 -0000 Date: Tue, 04 Jan 2005 04:04:00 -0000 Message-ID: <20050104040438.28655.qmail@sourceware.org> From: "law at redhat dot com" To: gcc-bugs@gcc.gnu.org In-Reply-To: <20041216155140.19038.dje@gcc.gnu.org> References: <20041216155140.19038.dje@gcc.gnu.org> Reply-To: gcc-bugzilla@gcc.gnu.org Subject: [Bug tree-optimization/19038] [4.0 Regression] out-of ssa causing loops to have more than one BB X-Bugzilla-Reason: CC X-SW-Source: 2005-01/txt/msg00307.txt.bz2 List-Id: ------- Additional Comments From law at redhat dot com 2005-01-04 04:04 ------- Subject: Re: [4.0 Regression] out-of ssa causing loops to have more than one BB On Thu, 2004-12-30 at 21:51 +0000, pinskia at gcc dot gnu dot org wrote: > ------- Additional Comments From pinskia at gcc dot gnu dot org 2004-12-30 21:50 ------- > (In reply to comment #34) > Actually it is because we are placing statements in the loop's latch. > It is done in create_new_iv. So just to be clear, it's not my change that is causing a regression. The IV code creates situations which prevent my change from having any kind of impact because the loop backedge is already split by the IV code (and thus the loop backedge is no longer critical and my out-of-ssa code does nothing). Right? Jeff ps. It seems to me that the IV code could use a trick similar to what I did to the out-of-ssa code. The only significant complication would be that the IV code would have to verify that the code it wants to insert is safe on both path (loop backedge and loop exit). -- http://gcc.gnu.org/bugzilla/show_bug.cgi?id=19038