From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S1750901AbWIJLVY (ORCPT ); Sun, 10 Sep 2006 07:21:24 -0400 Received: (majordomo@vger.kernel.org) by vger.kernel.org id S1750898AbWIJLVY (ORCPT ); Sun, 10 Sep 2006 07:21:24 -0400 Received: from pentafluge.infradead.org ([213.146.154.40]:60093 "EHLO pentafluge.infradead.org") by vger.kernel.org with ESMTP id S1750875AbWIJLVX (ORCPT ); Sun, 10 Sep 2006 07:21:23 -0400 Subject: Re: [PATCH RFC]: New termios take 2 From: David Woodhouse To: Alan Cox Cc: linux-kernel@vger.kernel.org, linux-arch@vger.kernel.org In-Reply-To: <1157886908.22571.11.camel@localhost.localdomain> References: <1157472883.9018.79.camel@localhost.localdomain> <1157885180.2977.133.camel@pmac.infradead.org> <1157886908.22571.11.camel@localhost.localdomain> Content-Type: text/plain Date: Sun, 10 Sep 2006 12:20:40 +0100 Message-Id: <1157887240.2977.147.camel@pmac.infradead.org> Mime-Version: 1.0 X-Mailer: Evolution 2.6.3 (2.6.3-1.fc5.5.dwmw2.1) Content-Transfer-Encoding: 7bit X-SRS-Rewrite: SMTP reverse-path rewritten from by pentafluge.infradead.org See http://www.infradead.org/rpr.html Sender: linux-kernel-owner@vger.kernel.org X-Mailing-List: linux-kernel@vger.kernel.org On Sun, 2006-09-10 at 12:15 +0100, Alan Cox wrote: > Kernel headers are not intended for user space. In this case the struct > termios presented by glibc already differs from the termios presented by > the kernel so the problem doesn't arise at all. Please note that we are moving away from the mindless repetition of that phrase, and moving towards a system where we actually _mean_ it. If you really don't want asm/term{bit,io}s.h to be visible in userspace, then the way to express that is to provide a patch to include/asm-generic/Kbuild which removes them from the export. I don't think I agree -- I think these files _do_ provide part of the kernel<->user ABI and should be kept. You're right that userspace in _general_ shouldn't be touching them -- as I said, it's only really the C libraries that are important here. As long as _they_ get it right when built against the headers with your changes, we're fine. But I don't think it's realistic to suggest that C libraries should be built without access to our asm/term{bit,io}s.h at all. However, I'm only really responsible for the new export _mechanism_ -- I'm not going to impose policy except when people like Andi do stupid things and sneakily send private patches to undo fixes I've already made. So if you want to unexport those headers and make sure the C libraries work like that, please do go ahead -- but don't just _say_ it, _do_ it -- and please do make sure that at _least_ glibc still builds afterwards. -- dwmw2