From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (qmail 25195 invoked by alias); 28 Feb 2007 02:25:13 -0000 Received: (qmail 24922 invoked by uid 48); 28 Feb 2007 02:24:59 -0000 Date: Wed, 28 Feb 2007 02:25:00 -0000 Message-ID: <20070228022459.24921.qmail@sourceware.org> X-Bugzilla-Reason: CC References: Subject: [Bug rtl-optimization/30987] [4.3 Regression] Problem while compiling gcc for score In-Reply-To: Reply-To: gcc-bugzilla@gcc.gnu.org To: gcc-bugs@gcc.gnu.org From: "ian at airs dot com" Mailing-List: contact gcc-bugs-help@gcc.gnu.org; run by ezmlm Precedence: bulk List-Id: List-Archive: List-Post: List-Help: Sender: gcc-bugs-owner@gcc.gnu.org X-SW-Source: 2007-02/txt/msg03007.txt.bz2 ------- Comment #2 from ian at airs dot com 2007-02-28 02:24 ------- The problem is that the bitset_c insn comes before the iorsi3 insn in the MD files. It matches the ior:SI insn with one added clobber. Unfortunately, the added clobber is of a hard reg, and so this is rejected by insn_invalid_p in recog.c. I don't understand how bitset_c will ever be used in practice. Nothing generates it explicitly. Typical ior:SI insns get passed to recog_memoized which does not add clobbers. I can fix the problem easily enough by moving (include "misc.md") and (include "mac.md") to the end of score.md. I'm not sure how to really test that fix. -- http://gcc.gnu.org/bugzilla/show_bug.cgi?id=30987