Subject: Re: Superblock Corruption?
To: Frank van der Linden <firstname.lastname@example.org>
From: Russell Neches <email@example.com>
Date: 04/15/2003 18:40:00
Frank van der Linden <firstname.lastname@example.org> writes:
> On Sun, Apr 13, 2003 at 11:55:40PM +1000, Charlie Allom wrote:
>> On Sun, Apr 13, 2003 at 04:45:13PM +0900, Curt Sampson wrote:
>> > On my Libretto L2 I'm using a November userland at the moment. When I
>> > boot up an April 11th kernel and mount all the filesystems read-only,
>> > and then reboot (using either the April or Novemember kernels), it tells
>> > me that the superblock is corrupt on all my filesystems, and I have to
>> > fsck by hand and restore the superblock from a backup superblock.
>> I filed a PR about this on my i386 + APPLE_UFS disk: kern/21083
> The issue Curt talks about is "old fsck compares too many superblock
> fields and incorrectly signals a bad superblock". The only thing you
> can do about that is install a new fsck. 1.6.1 has a forward compat
> patch, so that a 1.6.1 -> 2.0 upgrade will not have this problem.
> The Apple UFS problem looks to be something else.
> - Frank
Ahh. So it's a userland problem, then? That would stand to reason,
since I've gone through about 30 or 40 reboots, every one with a
manual fsck, and there's been no actual damage I can detect to the
filesystem. That certainly suggests a spurious filesystem integrity
Well, I guess I have to rebuild userland after all. I was hoping I
could get away without doing so. ^_^
This document is intended for outdoor use only.