From mboxrd@z Thu Jan  1 00:00:00 1970
Return-Path: <binutils-return-46446-listarch-binutils=sources.redhat.com@sourceware.org>
Received: (qmail 18509 invoked by alias); 15 Jun 2006 18:58:34 -0000
Received: (qmail 18501 invoked by uid 22791); 15 Jun 2006 18:58:33 -0000
X-Spam-Check-By: sourceware.org
Received: from potter.codesourcery.com (HELO mail.codesourcery.com) (65.74.133.4)     by sourceware.org (qpsmtpd/0.31) with ESMTP; Thu, 15 Jun 2006 18:58:31 +0000
Received: (qmail 32582 invoked from network); 15 Jun 2006 18:58:29 -0000
Received: from unknown (HELO 81-178-237-90.dsl.pipex.com) (paul@127.0.0.2)   by mail.codesourcery.com with ESMTPA; 15 Jun 2006 18:58:29 -0000
From: Paul Brook <paul@codesourcery.com>
To: "H. J. Lu" <hjl@lucon.org>
Subject: Re: PATCH: Add --alt-nops=short|long to x86/x86-64 assemblers
Date: Thu, 15 Jun 2006 19:02:00 -0000
User-Agent: KMail/1.9.1
Cc: binutils@sourceware.org,  Nick Clifton <nickc@redhat.com>
References: <20060613190254.GB21884@lucon.org> <200606151841.24590.paul@codesourcery.com> <20060615184510.GA9090@lucon.org>
In-Reply-To: <20060615184510.GA9090@lucon.org>
MIME-Version: 1.0
Content-Type: text/plain;   charset="iso-8859-1"
Content-Transfer-Encoding: 7bit
Content-Disposition: inline
Message-Id: <200606151958.27628.paul@codesourcery.com>
Mailing-List: contact binutils-help@sourceware.org; run by ezmlm
Precedence: bulk
List-Subscribe: <mailto:binutils-subscribe@sourceware.org>
List-Archive: <http://sourceware.org/ml/binutils/>
List-Post: <mailto:binutils@sourceware.org>
List-Help: <mailto:binutils-help@sourceware.org>, <http://sourceware.org/ml/#faqs>
Sender: binutils-owner@sourceware.org
X-SW-Source: 2006-06/txt/msg00255.txt.bz2


> in assembler. The reasons we haven't run into any serious problems are
>
> 1. By default, cpu_arch_flags is set to accept everything.
> 2. .arch directive isn't used much.
>
> So my -mtune=CPU switch will optimize for CPU by generating instruction
> for CPU if instruction set of CPU is available.

But you said these instructions only existed on ppro or later. You can't have 
mtune= effect the choice of instruction unless you also implement -march=. 

Consider gcc -march=386 -mtune=686. IIUC you're proposing that gcc doesn't 
emit .arch, and doesn't pass through -march=, so gas defaults to allowing all 
instructions. gas then sees -mtune= and generates instructions that don't 
work on 386.

Paul