From mboxrd@z Thu Jan  1 00:00:00 1970
Return-Path: <libc-hacker-return-5721-listarch-libc-hacker=sourceware.cygnus.com@sources.redhat.com>
Received: (qmail 14811 invoked by alias); 15 Oct 2002 22:43:06 -0000
Mailing-List: contact libc-hacker-help@sources.redhat.com; run by ezmlm
Precedence: bulk
List-Subscribe: <mailto:libc-hacker-subscribe@sources.redhat.com>
List-Archive: <http://sources.redhat.com/ml/libc-hacker/>
List-Post: <mailto:libc-hacker@sources.redhat.com>
List-Help: <mailto:libc-hacker-help@sources.redhat.com>, <http://sources.redhat.com/ml/#faqs>
Sender: libc-hacker-owner@sources.redhat.com
Received: (qmail 14792 invoked from network); 15 Oct 2002 22:43:06 -0000
Received: from unknown (HELO gateway.sf.frob.com) (64.163.212.31)
  by sources.redhat.com with SMTP; 15 Oct 2002 22:43:06 -0000
Received: from magilla.sf.frob.com (magilla.sf.frob.com [198.49.250.228])
	by gateway.sf.frob.com (Postfix) with ESMTP
	id 21982357E; Tue, 15 Oct 2002 15:43:05 -0700 (PDT)
Received: (from roland@localhost)
	by magilla.sf.frob.com (8.11.6/8.11.6) id g9FMh4s11189;
	Tue, 15 Oct 2002 15:43:04 -0700
Date: Tue, 15 Oct 2002 20:04:00 -0000
Message-Id: <200210152243.g9FMh4s11189@magilla.sf.frob.com>
MIME-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Transfer-Encoding: 7bit
From: Roland McGrath <roland@redhat.com>
To: Jakub Jelinek <jakub@redhat.com>
Cc: Ulrich Drepper <drepper@redhat.com>,
	Glibc hackers <libc-hacker@sources.redhat.com>
Subject: Re: [PATCH] Use __attribute__((tls_model(""))), add hidden aliases to TLS vars
In-Reply-To: Jakub Jelinek's message of  Wednesday, 16 October 2002 00:26:47 +0200 <20021016002647.K3451@sunsite.ms.mff.cuni.cz>
X-Zippy-Says: With YOU, I can be MYSELF..  We don't NEED Dan Rather..
X-SW-Source: 2002-10/txt/msg00060.txt.bz2

Cool, I'll put that in.  I was going to get to that myself but hadn't yet
installed the compiler that supports it.

It occurs to me you could use libc_hidden_proto instead of macros.
But that is uglier in the asm, and both ways do the same thing.

One thing that concerns me is if a program (erroneously) does 
"__thread int errno;" (sans extern) or suchlike.  Then the program
and libpthread will use its errno, but libc will only set __libc_errno.

It would be nice if this produced a link-time error like a multiple
definition with a non-shared object would.  But AFAIK no .so symbol will
ever produce a multiple definition error.  Perhaps we could do some hack in
the linker script.