Subject: Re: Install problem with 2.0
To: None <firstname.lastname@example.org, port-mac68k@NetBSD.org>
From: Gene ENonymous <email@example.com>
Date: 12/15/2004 14:18:43
> From: "Bruce O'Neel" <firstname.lastname@example.org>
> Date: Wed, 15 Dec 2004 06:24:14 +0000
> I've installed a number of LC040 systems now and there seems to
> be some oddities when you choose Install vs Update.
> Everything is fine until the edit the partition table step.
> The disk label seems to be written ok, but, a newfs is not done
> and the partitions are not mounted.
> I can work around it by choosing 'none' for the method to load
> the sets from, run /bin/sh, newfs /dev/sd0a manually,
> mount /dev/sd0a /targetroot, exit the shell, and redo the
> 'Install NetBSD' steps. fsck complains about the disk not being
> clean but that's because it was mounted.
I'm probably missunderstanding, but when I try to do a fresh
'Install NetBSD' it always tries to create new partitions.
> Do others see this? There is always the chance that the softfloat
> patches have broken something.
I don't think this has to do with soft float. I've had the same
issues on a Centris w/40M ram and a full (w f/p) 68040.
My required steps post disklabel (250M /, 3.5G /usr, 150M /home, 250M swap)
1) as you said, newfs - /, /usr and /home
2) for me the easiest way to mount with /bin/sh was to create a dummy /etc/fstab
and I just mounted on /tmproot
$ cat > /etc/fstab
/dev/sd1a /tmproot ffs rw 1 1
3) mount / (/dev/sd1a) on /tmproot
4) cd in /tmproot and mkdir home, etc, usr and tmp
5) make a real fstab in /tmproot/etc/fstab
$ cat > /tmproot/etc/fstab
/dev/sd1a / ffs rw 1 1
/dev/sd1b none swap sw 0 0
/dev/sd1g /usr ffs rw 1 2
/dev/sd1d /home ffs rw 1 2
6) sync; umount all
7) exit /bin/sh or just start sysinst if you are already out
8) then choose update or install new. The fstab and the mount points
let it think that there is an existing good installation.