Subject: Re: Some progress with Multia - but still not done
To: NetBSD/Alpha <port-alpha@netbsd.org>
From: Jack Twilley <jmt@twilley.org>
List: port-alpha
Date: 01/30/2002 10:01:46
>>>>> "Roland" == R C Dowdeswell <elric@mabelode.imrryr.org> writes:

Jack> I'm using V3.8-3 of the firmware, the netboot from the online
Jack> form with broken firmware set to yes, and a set of NFS mounts
Jack> constructed as per the NetBSD documentation on setting up a
Jack> diskless workstation.

Roland> Try it with Broken Firmware set to `no'.  That's what works
Roland> for my Multia which has V3.8-3.

I tried that, and it didn't work.  I've since gone with a local root
and swap, and mounting /usr.  Much better.

Jack> Four times alpha sends out DHCP:DISCOVER messages.  Not TFTP
Jack> messages for "netbsd", but DHCP messages.  duchess responds with
Jack> all the appropriate information, including a root path.  After
Jack> four attempts, alpha complains that it cannot get the kernel.

Jack> Am I missing something here?  I've looked at the documentation,
Jack> but there seems to be no way to make this thing mount the root
Jack> path and "just do it".  Are there flags that I should supply?
Jack> Help?

Roland> So, to see if I can help with the procedure, this is what
Roland> `should' happen:

Roland> 1.  multia SRM does a bootp request, server answers.

Yep.  This works as advertised.

Roland> 2.  multia SRM loads bootloader via TFTP.

Different firmware has different definitions for -file, some load
-file as the first stage, some load it as the second stage.

Roland> 3.  multia bootloader does a bootp request, server answers
Roland> 	(this is where your multia is failing)

I agree.  And it's silly for it to do a bootp request anyway.  It
should be doing a TFTP request for the file it knows it needs.  *sigh*

[...]

Roland> 8.  everything works...

A mythical state. :-)

Jack.
-- 
Jack Twilley
jmt at twilley dot org
http colon slash slash www dot twilley dot org slash tilde jmt slash