From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (qmail 5196 invoked by alias); 29 Sep 2004 18:11:08 -0000 Mailing-List: contact java-prs-help@gcc.gnu.org; run by ezmlm Precedence: bulk List-Subscribe: List-Archive: List-Post: List-Help: , Sender: java-prs-owner@gcc.gnu.org Received: (qmail 5175 invoked by uid 48); 29 Sep 2004 18:11:07 -0000 Date: Wed, 29 Sep 2004 18:11:00 -0000 Message-ID: <20040929181107.5174.qmail@sourceware.org> From: "pinskia at gcc dot gnu dot org" To: java-prs@gcc.gnu.org In-Reply-To: <20040929175802.17731.tromey@gcc.gnu.org> References: <20040929175802.17731.tromey@gcc.gnu.org> Reply-To: gcc-bugzilla@gcc.gnu.org Subject: [Bug java/17731] sub-optimal code generated for left shift X-Bugzilla-Reason: CC X-SW-Source: 2004-q3/txt/msg00729.txt.bz2 List-Id: ------- Additional Comments From pinskia at gcc dot gnu dot org 2004-09-29 18:11 ------- on PPC: __ZN4test5shiftEii: rlwinm r5,r5,0,27,31 slw r3,r4,r5 blr Which is correct as r5 could be greater than 31 and being greater than 31 is undefined (4 bits are only used really but it could do something different than that). So if this were to be done correctly, we would need a target hook for the Java front-end. Note the only reason why I bring this up is because we just had a bootstrap failure where we depended on being anded by the machine and not in the code. -- http://gcc.gnu.org/bugzilla/show_bug.cgi?id=17731