From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (qmail 23577 invoked by alias); 9 Sep 2005 12:19:29 -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 23357 invoked by uid 48); 9 Sep 2005 12:19:20 -0000 Date: Fri, 09 Sep 2005 12:19:00 -0000 Message-ID: <20050909121920.23356.qmail@sourceware.org> From: "rguenth at gcc dot gnu dot org" To: gcc-bugs@gcc.gnu.org In-Reply-To: <20050901005347.23665.kst@mib.org> References: <20050901005347.23665.kst@mib.org> Reply-To: gcc-bugzilla@gcc.gnu.org Subject: [Bug c/23665] [3.4 Regression] ICE in convert_move with -O3 (ok at -O2) X-Bugzilla-Reason: CC X-SW-Source: 2005-09/txt/msg01109.txt.bz2 List-Id: ------- Additional Comments From rguenth at gcc dot gnu dot org 2005-09-09 12:19 ------- Problem of the C frontend using the first decl with int arg and the inliner expecting to have a double in unit-at-a-time mode, -O1 -funit-at-a-time: inline double func(); int main(void) { double d = func(42); return 0; } inline double func(double arg) { return arg * 2.0; } src/tests/pr23665.c: In function `main': src/tests/pr23665.c:10: internal compiler error: in convert_move, at expr.c:563 Please submit a full bug report, with preprocessed source if appropriate. See for instructions. We're probably confusing the two decls. I guess we will not fix this for 3.4. -- http://gcc.gnu.org/bugzilla/show_bug.cgi?id=23665