From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S934686AbZHEQuK (ORCPT ); Wed, 5 Aug 2009 12:50:10 -0400 Received: (majordomo@vger.kernel.org) by vger.kernel.org id S934517AbZHEQuJ (ORCPT ); Wed, 5 Aug 2009 12:50:09 -0400 Received: from cantor.suse.de ([195.135.220.2]:45687 "EHLO mx1.suse.de" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S934451AbZHEQuI (ORCPT ); Wed, 5 Aug 2009 12:50:08 -0400 Message-ID: <4A79B8CA.8050504@suse.de> Date: Thu, 06 Aug 2009 01:52:26 +0900 From: Tejun Heo User-Agent: Thunderbird 2.0.0.22 (X11/20090605) MIME-Version: 1.0 To: Alan Cox Cc: Greg KH , Al Viro , Takashi Iwai , Linux Kernel , cguthrie@mandriva.org Subject: Re: [PATCH 2/2] sound: make OSS device number claiming optional References: <4A79283E.7030202@kernel.org> <4A79296A.4090600@suse.de> <20090805101551.6ee053e5@lxorguk.ukuu.org.uk> <4A7951B0.10208@suse.de> <20090805110031.51424dbe@lxorguk.ukuu.org.uk> <4A796C94.5040301@suse.de> <20090805134823.4409e197@lxorguk.ukuu.org.uk> <4A799392.3000108@suse.de> <20090805152942.42e785f3@lxorguk.ukuu.org.uk> <4A79AD2A.4060308@suse.de> <20090805173346.130d80bb@lxorguk.ukuu.org.uk> In-Reply-To: <20090805173346.130d80bb@lxorguk.ukuu.org.uk> X-Enigmail-Version: 0.95.7 Content-Type: text/plain; charset=ISO-8859-1 Content-Transfer-Encoding: 7bit Sender: linux-kernel-owner@vger.kernel.org List-ID: X-Mailing-List: linux-kernel@vger.kernel.org Alan Cox wrote: >> I don't see how I would be able to achieve the latter with one liner. >> Can you please elaborate a little bit? > > It depends how much of a hurry you are in, but for the mainstream we can > do this > > 1. Make soundcore also issue the request_module() as if the char > minor was unclaimed. Mark the old one as obsolete > > 2. Wait a year while people adjust their scripts to trigger on the > char event instead (and fix the ordering bits for ALSA if the soundcore > docs are still valid on that) > > 3. Apply a patch which makes ALSA use char dev allocation directly > for its OSS devices and dumps out soundcore, remove soundcore and switch > any other code using it. It's quite difficult to think that those now mostly unused module alises are worth full year of waiting and careful coordination. I think we're chasing a non-issue here. If it really matters, I'll try to teach chrdev about alternative module alises. If you think the above is a good solution, how about the following? 1. Merge the weird switch thing and the extra standard chrdev module alias patch 2. Add to feature-removal that snd-slot/service-* are going away in a year along with the weird switches. This allows people who wish to try or switch in the meantime to do so. 3. After a year, drop module loading related code from sound_core along with the weird config option and kernel parameter. In the end, the only choice we have to make is whether to keep snd-slot/service-* aliases. If we're gonna (I don't see why tho), the cleanest way would be to teach chrdev about aliases. If not, the best way is to add a switch so that it can be phased out gradually. Thanks. -- tejun