From mboxrd@z Thu Jan  1 00:00:00 1970
Return-Path: <gcc-return-116301-listarch-gcc=gcc.gnu.org@gcc.gnu.org>
Received: (qmail 13081 invoked by alias); 24 Jun 2005 02:07:20 -0000
Mailing-List: contact gcc-help@gcc.gnu.org; run by ezmlm
Precedence: bulk
List-Archive: <http://gcc.gnu.org/ml/gcc/>
List-Post: <mailto:gcc@gcc.gnu.org>
List-Help: <http://gcc.gnu.org/ml/>
Sender: gcc-owner@gcc.gnu.org
Received: (qmail 12987 invoked by uid 22791); 24 Jun 2005 02:07:12 -0000
Received: from [216.208.38.107] (HELO OTTLS.pngxnet.com) (216.208.38.107)
    by sourceware.org (qpsmtpd/0.30-dev) with ESMTP; Fri, 24 Jun 2005 02:07:12 +0000
Received: from 56.254.255.10.in-addr.arpa ([10.255.254.56])
	by OTTLS.pngxnet.com (8.12.4/8.12.4) with ESMTP id j5O27BWV009524;
	Thu, 23 Jun 2005 22:07:11 -0400
Received: by greed.geoffk.org (Postfix, from userid 501)
	id BE113533B0F; Thu, 23 Jun 2005 14:08:20 -0400 (EDT)
To: Ian Lance Taylor <ian@airs.com>
Cc: gcc@gcc.gnu.org
Subject: Re: dead label use?
References: <EE6A5C4E-E376-11D9-9BDB-003065BDF310@apple.com>
	<m3oe9x94j4.fsf@gossamer.airs.com>
From: Geoffrey Keating <geoffk@geoffk.org>
Date: Fri, 24 Jun 2005 02:07:00 -0000
In-Reply-To: <m3oe9x94j4.fsf@gossamer.airs.com>
Message-ID: <m21x6tq73g.fsf@greed.apple.com>
User-Agent: Gnus/5.09 (Gnus v5.9.0) Emacs/21.2
MIME-Version: 1.0
Content-Type: text/plain; charset=us-ascii
X-SW-Source: 2005-06/txt/msg01012.txt.bz2

Ian Lance Taylor <ian@airs.com> writes:

> Mike Stump <mrs@apple.com> writes:
> 
> > Forgive me ignorance, is there is use for the use of the label below?
> > 
> >  From rs6000, though, certainly there are other examples of this sort
> > of thing in the md files:
> > 
> > (define_insn ""
> >    [(set (pc)
> >          (match_operand:P 0 "register_operand" "c,*l"))
> >     (use (label_ref (match_operand 1 "" "")))]
> >    ""
> >    "@
> >     bctr
> >     {br|blr}"
> >    [(set_attr "type" "jmpreg")])
> 
> This is an instantiation of a tablejump created via define_expand.
> Operand 1 is the jump table.  The insn needs to refer to it, or the
> table will be deleted as unreferenced.

The problem is that the table has to be referenced *from a jump_insn*,
I think.  Obviously if the table is really unreferenced it could be deleted.