From patchwork Tue Jul 20 18:38:53 2010
Content-Type: text/plain; charset="utf-8"
MIME-Version: 1.0
Content-Transfer-Encoding: 7bit
X-Patchwork-Submitter: Nathan Froyd <froydnj@codesourcery.com>
X-Patchwork-Id: 59352
Return-Path: 
 <gcc-patches-return-268597-incoming=patchwork.ozlabs.org@gcc.gnu.org>
X-Original-To: incoming@patchwork.ozlabs.org
Delivered-To: patchwork-incoming@bilbo.ozlabs.org
Received: from sourceware.org (server1.sourceware.org [209.132.180.131])
	by ozlabs.org (Postfix) with SMTP id F336FB6F0E
	for <incoming@patchwork.ozlabs.org>;
	Wed, 21 Jul 2010 04:39:02 +1000 (EST)
Received: (qmail 28657 invoked by alias); 20 Jul 2010 18:39:00 -0000
Received: (qmail 28646 invoked by uid 22791); 20 Jul 2010 18:38:59 -0000
X-SWARE-Spam-Status: No, hits=-1.7 required=5.0	tests=AWL, BAYES_00,
	T_RP_MATCHES_RCVD
X-Spam-Check-By: sourceware.org
Received: from mail.codesourcery.com (HELO mail.codesourcery.com)
	(38.113.113.100) by sourceware.org (qpsmtpd/0.43rc1) with
	ESMTP; Tue, 20 Jul 2010 18:38:55 +0000
Received: (qmail 20066 invoked from network); 20 Jul 2010 18:38:54 -0000
Received: from unknown (HELO localhost) (froydnj@127.0.0.2) by
	mail.codesourcery.com with ESMTPA; 20 Jul 2010 18:38:54 -0000
Date: Tue, 20 Jul 2010 11:38:53 -0700
From: Nathan Froyd <froydnj@codesourcery.com>
To: gcc-patches@gcc.gnu.org
Cc: dje.gcc@gmail.com
Subject: [rs6000,committed] fix abs_isel patterns
Message-ID: <20100720183853.GK26037@codesourcery.com>
MIME-Version: 1.0
Content-Disposition: inline
User-Agent: Mutt/1.5.17+20080114 (2008-01-14)
X-IsSubscribed: yes
Mailing-List: contact gcc-patches-help@gcc.gnu.org; run by ezmlm
Precedence: bulk
List-Id: <gcc-patches.gcc.gnu.org>
List-Unsubscribe: 
 <mailto:gcc-patches-unsubscribe-incoming=patchwork.ozlabs.org@gcc.gnu.org>
List-Archive: <http://gcc.gnu.org/ml/gcc-patches/>
List-Post: <mailto:gcc-patches@gcc.gnu.org>
List-Help: <mailto:gcc-patches-help@gcc.gnu.org>
Sender: gcc-patches-owner@gcc.gnu.org
Delivered-To: mailing list gcc-patches@gcc.gnu.org

While debugging other problems, I noticed that my isel patch:

http://gcc.gnu.org/ml/gcc-patches/2010-07/msg01298.html

was causing ICEs when compiling libgcc.  After making the isel patterns
only accept lt/eq/gt comparison codes, the *abs<mode>2_isel splitters
split to RTL that's not recognizable.  (This should have shown up when
compiling libgcc for powerpc-eabispe, but doesn't because the default
processor for that target isn't an isel-capable processor...patch
forthcoming.)

Fixed with the below patch, which reverses the sense of the condition in
the patterns, along with the operands.

Applied as obvious after testing on powerpc-eabispe.

-Nathan

	* config/rs6000/rs6000.md (abs<mode>2_isel, nabs<mode>2_isel):
	Reverse sense of if_then_else condition.

Index: config/rs6000/rs6000.md
===================================================================
--- config/rs6000/rs6000.md	(revision 162343)
+++ config/rs6000/rs6000.md	(working copy)
@@ -2105,10 +2105,10 @@ (define_insn_and_split "abs<mode>2_isel"
 	(compare:CC (match_dup 1)
 		    (const_int 0)))
    (set (match_dup 0)
-	(if_then_else:GPR (ge (match_dup 3)
+	(if_then_else:GPR (lt (match_dup 3)
 			      (const_int 0))
-			  (match_dup 1)
-			  (match_dup 2)))]
+			  (match_dup 2)
+			  (match_dup 1)))]
   "")
 
 (define_insn_and_split "nabs<mode>2_isel"
@@ -2124,10 +2124,10 @@ (define_insn_and_split "nabs<mode>2_isel
 	(compare:CC (match_dup 1)
 		    (const_int 0)))
    (set (match_dup 0)
-	(if_then_else:GPR (ge (match_dup 3)
+	(if_then_else:GPR (lt (match_dup 3)
 			      (const_int 0))
-			  (match_dup 2)
-			  (match_dup 1)))]
+			  (match_dup 1)
+			  (match_dup 2)))]
   "")
 
 (define_insn_and_split "abssi2_nopower"