> I have a beta version of the firmware with support for ATAPI boot devices,
> ISO filesystesm and ext2 filesystems. This is should be available from the
> Chalice WWW site as soon as I am happy with it.

This would be the ideal solution. I've got ample device-programming facilities
at work.

> > Next thing I am going to try is swapping the Pentium motherboard back in
> > and using it to install wd0 from the CATS CD. Presumably the CATS firmware
> > will at least attempt to boot a netbsd kernel from that.
> 
> The firmware will boot from FFS based ATA devices ok.

Strangely enough, it didn't. Is there something about an i386-disklabelled
ATA device that the CATS firmware does not like?

I used NetBSD/i386 (booted from wd1) to install from the CATS installation CD
(CATS.19980805) onto wd0. Prior to running the CATS install script, I copied
it and a 'disklabel -rt wd0 > disktab' into /tmp so that the script could be
run from there and mount /cdrom without errors. The script itself executed
fine except for Exec Format errors during the pkg_add phase, which by itself
should not prevent me from getting a bootable system.

So now I have a chassis with two ATA drives, wd1 is still my i386 boot disk
and wd0 has been freshly installed by the CATS install script running on i386.
When I swap the CATS board into that chassis, it still does not want to boot.
The error is "Input/output error" or something like that, and it says this
for both wd0a:/netbsd (arm32) and wd1a:/netbsd (i386). So I suspect that there
is something about the i386 disklabel that the CATS firmware does not like,
because it doesn't appear to even get to the part where it checks the magic
number of /netbsd.

Since the CATS firmware reads the ffs directly, I expected the boot blocks
to not matter. That leaves the disklabel itself, I think.

Has anyone else tried this?

Todd Whitesel
toddpw @ wrs.com