At 11:34 AM +0000 4/11/05, Bruce O'Neel wrote:
>
>Do make sure that the drive is ID 0.  I spent several frustrating hours
>with this
>assuming that it was 0 when in fact it was 4.

It's not entirely related, but I don't think I ever posted the following
code to fix autoconf.c from dropping the boot device node when it isn't ID
0. I submitted it for review and checked on it several times, but it was
never committed.

http://www.dialectronics.com/NetBSD/code/autoconf.c

The problem is that in many cases getting the canonical path name will drop
any node and the current code in autoconf.c assumes that the node was 0. It
also has a tweak in it to report what the device boot path is as each
component is discovered. Doing this helped reveal the problem that the
obio.c code assumed there was only one interrupt controller (mac-io), when
this is not always true.

tim