From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S1756246Ab0FELjq (ORCPT ); Sat, 5 Jun 2010 07:39:46 -0400 Received: from cantor.suse.de ([195.135.220.2]:57324 "EHLO mx1.suse.de" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1751940Ab0FELjp (ORCPT ); Sat, 5 Jun 2010 07:39:45 -0400 Date: Sat, 5 Jun 2010 21:39:30 +1000 From: Nick Piggin To: Christoph Hellwig Cc: Al Viro , linux-kernel@vger.kernel.org, linux-fsdevel@vger.kernel.org, Alan Cox Subject: Re: [PATCH, RFC] tty: stop abusing file->f_u.fu_list Message-ID: <20100605113930.GH26335@laptop> References: <20100604064307.737085373@suse.de> <20100604072618.227745945@suse.de> <20100604183934.GA2309@infradead.org> MIME-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Disposition: inline In-Reply-To: <20100604183934.GA2309@infradead.org> User-Agent: Mutt/1.5.20 (2009-06-14) Sender: linux-kernel-owner@vger.kernel.org List-ID: X-Mailing-List: linux-kernel@vger.kernel.org On Fri, Jun 04, 2010 at 02:39:34PM -0400, Christoph Hellwig wrote: > The ttry code currently abuses the file anchor for the per-sb file list > to track instances of a given tty. But there's no good reason for > that, we can just install a proxy object in file->private that gets > added to the list and points to the tty and keep the list away from > VFS internals. Well thanks for this. Yes it is an obviously nicer way to do it, so tty doesn't have to know what vfs uses files list for. > Note that I've just if 0'd the selinux mess poking into it. While we > could trivially port it to the new code by making the tty_private > structure public this code is just too revolting to be kept around. > It would never have been there anyway if a person with some amount of > clue had ever reviewed the selinux code. And no, it's not just the > tty portion, the rest of that function is just as bad. Why is it a mess? Just because of the conceptual nastiness of checking a tty object via a random one of its inodes? How would be a better way to do this? I think for a first pass, a simple conversion for all code would be good for me because then it stops blocking the scaling patch. (and it's more bisectable).