Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id ; Tue, 19 Mar 2002 06:16:35 -0500 Received: (majordomo@vger.kernel.org) by vger.kernel.org id ; Tue, 19 Mar 2002 06:16:25 -0500 Received: from mail49-s.fg.online.no ([148.122.161.49]:12421 "EHLO mail49.fg.online.no") by vger.kernel.org with ESMTP id ; Tue, 19 Mar 2002 06:16:14 -0500 To: Andrew Morton Cc: Colin Leroy , linux-kernel@vger.kernel.org Subject: Re: question about 2.4.18 and ext3 In-Reply-To: <20020318180158.2886dd4a.colin@colino.net> <3C96510A.24CDE6BC@zip.com.au> From: hakon@cyberglobe.net (=?iso-8859-1?q?H=E5kon?= Alstadheim) Date: 19 Mar 2002 08:57:51 +0100 Message-ID: Lines: 55 User-Agent: Gnus/5.0808 (Gnus v5.8.8) Emacs/20.7 MIME-Version: 1.0 Content-Type: text/plain; charset=iso-8859-1 Content-Transfer-Encoding: 8bit Sender: linux-kernel-owner@vger.kernel.org X-Mailing-List: linux-kernel@vger.kernel.org Andrew Morton writes: > Colin Leroy wrote: > > > > Hello all, > > > > I really hope I'm not asking a FAQ, i looked in the archives since 15 Feb > > and didn't see anything about this. > > > > I upgraded from 2.2.20 to 2.4.18 on my powerbook two weeks ago, and > > compiled ext3 in the kernel in order to quietly crash :) > > > > However, I had about a dozen strange crashes, sometimes when the computer > > woke up from sleep, sometimes when launching a program : every visible > > soft died, then X, then blackscreen, and the computer didn't even answer > > pings. So I reset the computer and here, each time, yaboot (ppc equivalent > > of lilo) told me that "cannot load image". Booting and fscking from a > > rescue CD showed that superblock was corrupt. > > It may be a yaboot/ext3 incompatibility. Your version of yaboot > may not know how to mount a needs-recovery ext3 filesystem. > There are some words on this at > http://www.zip.com.au/~akpm/linux/ext3/ext3-usage.html > > I am told that yaboot 1.3.5 and later will do the right thing. > What version are you using? One way to work around this is to keep your kernels in /boot and have /boot be a separate ext2 filesystem that you normally mount ro. That way it will not need to be recovered after a crash. During install of a new kernel you will of course need to do "/bin/mount -o remount,rw /boot" and then afterwards "/bin/mount -o remount,ro /boot" . Your /etc/fstab will then look something like: [...] /dev/hde3 / ext3 # your usual parameters here /dev/hda6 /boot ext2 ro,exec 1 2 [...] Remember to make sure you know which devices are / and /boot/ respectively, and also which device holds the bootsector. Make sure you know which one to give to your bootloader where. I run grub-install like this: /usr/local/sbin/grub-install /dev/hda --root-directory=/boot Which tells grub to install its loader into the MBR of my first hd, and then look for the kernel and the second stage boot-loader on /dev/hda6 (which is mounted as /boot/). I seem to remember that grub copied the files it needed into /boot/boot/grub/ itself. It looks funny, but I left it at that. -- H?kon Alstadheim, hjemmepappa. - To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to majordomo@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html Please read the FAQ at http://www.tux.org/lkml/