Subject: mopd success (sort of :-)
To: None <port-pmax@NetBSD.ORG>
From: Simon Burge <simonb@NetBSD.ORG>
List: port-pmax
Date: 08/18/1999 01:07:20
[[ cc'd to port-vax as they use mopd there (I think) ]]

Just a quick note before I get some much-needed sleep.  I've just
successfully booted the smallnet.ecoff kernel via a greatly hacked
mopd running of another NetBSD server.  Here's what syslog had to

	Aug 18 00:44:38 mona mopd[1568]: mopd 2.5.3 started.
	Aug 18 00:44:38 mona mopd[1568]: Initialized le0
	Aug 18 00:44:56 mona mopd[1568]: 8:0:2b:1c:27:4c (1) Do you have 08002b1c274c? (Yes)
	Aug 18 00:44:56 mona mopd[1568]: 0:0:0:0:0:0 Send me 08002b1c274c
	Aug 18 00:45:56 mona mopd[1568]: 8:0:2b:1c:27:4c Send me 08002b1c274c
	Aug 18 00:46:00 mona mopd[1568]: 8:0:2b:1c:27:4c Load completed

At the moment the modified mopd only runs on another NetBSD/pmax machine
because it relies on the contents of some mips-specific ecoff header
files.  It looks like I'm going to have to put a lot of pmax-specific
ecoff knowledge directly in the mop code - but it also should be
possible to add ELF knowledge too.

mopd works on loading images from the /tftpboot/mop directory, but the
files in there can be symlinks.  Hopefully soon you'll be able to have:

	/tftpboot/mop/08002b123456.SYS -> /dlboot0/client.root/netbsd

where "/dlboot0/client.root" is the NFS root directory of the client
and "netbsd" is your normal ELF kernel, thus getting rid of a separate
copy (in a different format!) of your kernel just for booting.

I've also fixed a bug in the mop code so you can now run moptrace and
watch an Ultrix/mips MOP boot without it dumping core because of a
buffer overflow.

I still have plans of one day getting real two-stage network boots using
TFTP and NFS working, but this will at least reduce the need of that.