From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S964954AbWBGDdi (ORCPT ); Mon, 6 Feb 2006 22:33:38 -0500 Received: (majordomo@vger.kernel.org) by vger.kernel.org id S964957AbWBGDdi (ORCPT ); Mon, 6 Feb 2006 22:33:38 -0500 Received: from rwcrmhc13.comcast.net ([216.148.227.153]:4858 "EHLO rwcrmhc13.comcast.net") by vger.kernel.org with ESMTP id S964954AbWBGDdh (ORCPT ); Mon, 6 Feb 2006 22:33:37 -0500 Message-ID: <43E8150E.9030801@comcast.net> Date: Mon, 06 Feb 2006 22:33:34 -0500 From: Ed Sweetman User-Agent: Debian Thunderbird 1.0.7 (X11/20051019) X-Accept-Language: en-us, en MIME-Version: 1.0 To: Matthew Garrett CC: Andrew Morton , alan@lxorguk.ukuu.org.uk, harald.dunkel@t-online.de, linux-kernel@vger.kernel.org, rdunlap@xenotime.net Subject: Re: 2.6.16-rc1-mm2 pata driver confusion + tsc sync issues References: <43E3D103.70505@comcast.net> <43E7A4C0.4020209@t-online.de> <1139255800.10437.51.camel@localhost.localdomain> <43E805D4.5010602@comcast.net> <43E7F73E.2070004@comcast.net> <43E7F73E.2070004@comcast.net> <20060206173520.43412664.akpm@osdl.org> In-Reply-To: Content-Type: text/plain; charset=ISO-8859-1; format=flowed Content-Transfer-Encoding: 7bit Sender: linux-kernel-owner@vger.kernel.org X-Mailing-List: linux-kernel@vger.kernel.org Matthew Garrett wrote: >Andrew Morton wrote: > > > >>That's bad. >> >> > >Given that libata goes through the scsi layer at the moment, shifting >from the traditional PATA drivers to the libata ones is going to result >in a shift from hdfoo to sdbar. We're not really looking forward to this >from the distribution point of view, though I think the same thing >happened in the past when shifting from the ancient SATA drivers to the >libata ones. > > > Well, the badness mentioned above is the swapping of what gets loaded first from the testing branch to the upstream patches. In mm, sata gets loaded before pata in libata land. In alan cox's patches it's the reverse. This results in different device names for the same config when switching between mm and release, which is bad, but is a problem that can be overcome with the use of labels instead of device names. Perhaps from a distribution standpoint, moving to a label method of describing what gets mounted where would be best, rather than worrying about scsi naming schemes or ide ones. Just think of the fun of a system with multiple usb storage devices and such. I'm just not sure if grub and the kernel "root=" parameter can handle it.