From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S1757124AbXJOXCl (ORCPT ); Mon, 15 Oct 2007 19:02:41 -0400 Received: (majordomo@vger.kernel.org) by vger.kernel.org id S1753866AbXJOXCd (ORCPT ); Mon, 15 Oct 2007 19:02:33 -0400 Received: from www.tglx.de ([62.245.132.106]:42085 "EHLO www.tglx.de" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1752973AbXJOXCc (ORCPT ); Mon, 15 Oct 2007 19:02:32 -0400 Date: Tue, 16 Oct 2007 01:01:53 +0200 (CEST) From: Thomas Gleixner To: Arnd Bergmann cc: linux-kernel@vger.kernel.org, Andi Kleen , Andrew Morton Subject: Re: asm-x86/* exported headers using CONFIG_X86_32 In-Reply-To: <200710160054.00138.arnd@arndb.de> Message-ID: References: <200710160054.00138.arnd@arndb.de> MIME-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII Sender: linux-kernel-owner@vger.kernel.org X-Mailing-List: linux-kernel@vger.kernel.org On Tue, 16 Oct 2007, Arnd Bergmann wrote: > While looking through the new header files, I noticed lots of occurences > of #ifdef CONFIG_X86_32 in headers files exported for glibc. > > This is fundamentally broken because user applications including them > do not know about any CONFIG_* symbols, and if they did, those > would incorrectly describe the ABI. > > I guess in most cases, the headers that are interesting to user space > can simply be merged without any such #ifdef remaining, but those > that are still needed should be converted to use #ifdef __x86_64__, > which is set by the compiler. > This is how the other architectures avoid this particular problem. Yup, they slipped through. I have fixups in my pile already. Will send them tomorrow when I'm actually awake. tglx