From mboxrd@z Thu Jan 1 00:00:00 1970 From: Jeff Garzik Subject: Re: [PATCH 2.6.7] new NVIDIA libata SATA driver Date: Wed, 16 Jun 2004 21:47:16 -0400 Sender: linux-ide-owner@vger.kernel.org Message-ID: <40D0F824.8010108@pobox.com> References: Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii; format=flowed Content-Transfer-Encoding: 7bit Return-path: Received: from parcelfarce.linux.theplanet.co.uk ([195.92.249.252]:48583 "EHLO www.linux.org.uk") by vger.kernel.org with ESMTP id S266346AbUFQBrj (ORCPT ); Wed, 16 Jun 2004 21:47:39 -0400 In-Reply-To: List-Id: linux-ide@vger.kernel.org To: Andrew Chew Cc: Bartlomiej Zolnierkiewicz , linux-kernel@vger.kernel.org, linux-ide@vger.kernel.org Andrew Chew wrote: >>From: Jeff Garzik [mailto:jgarzik@pobox.com] > > >>If silicon isn't available yet, let's just remove those PCI >>IDs. That way we can ensure that these are libata only, >>without two drivers sharing the same PCI ids. > > > If that's the case, I'd rather the CK804 and MCP04 SATA device IDs be > added to sata_nv, since we want distributions to support these SATA > controllers when silicon does become available. OK >>>>Removing IDs from amd74xx.c is a bad idea, >>>>it breaks boot on systems already using these IDs. >>> >>>I assume these systems will be able to boot using the libata >>>subsystem. Is that a bad assumption? >> >>They can, but consider a system that uses 2.6.7 (IDE driver) >>then boots into 2.6.8 (libata driver): the drives move from >>/dev/hdX to /dev/sdX. That breaks stuff not using LABEL= in >>bootloader config and fstab. > > > That's true. I kinda chalk this up as an inevitable kernel upgrade > issue (they'll be getting support for NVIDIA SATA under libata with > eventual device hotplug support, at the cost of some system > reconfiguration). Is there a good solution? None really, other than trying to make sure we only add new SATA hardware support under libata. Most users with Red Hat or SuSE or whatever don't know enough other than "I booted the new Fedora 2.6.x kernel, and it panic'd when it tried to mount root filesystem" Your patch represented the _ideal_, but unfortunately once added to the IDE driver, those PCI ids should not be removed in the middle of a stable kernel series. If both drivers were /dev/sdX, or /dev/hdX, moving the PCI ids to another driver wouldn't be as big of a deal. Overall... no worries. I have to deal with this issue for a couple other chipsets, so I'll handle them all at the same time. Jeff