Subject: Re: Omron Luna/88k?
To: None <jma7+@andrew.cmu.edu, netbsd-ports@NetBSD.ORG>
From: Jason Brazile <jason@sunra.csci.unt.edu>
List: netbsd-ports
Date: 05/15/1997 10:07:15
> Has anyone tried porting NetBSD to the Omron Luna/88k?  

For a while I was playing with a port to a Data General Aviion 310
that was loaned to me.  I got a NetBSD/i386 -> 88k gcc cross compiler
to work at least well enough to run some standalone programs using
prom routines.  I also managed to compile and link a diskless netbsd
kernel with everything stubbed out. But then I moved a year or so
ago and had to give the machine back.

At the time, I was looking at the omron/luna mach stuff to help me
with my work (I think I used a modification of its kernel printf
implementation) but to be honest probably nothing I have would be
very useful to you. I noticed that Jeffry Friedl whose name appeared
in those sources has recently reappeared as he wrote the O Reilly
book on Regular Expressions.  It should be easy to find his email
address and ask him questions about that code if you need to.

I also noticed that recently someone at Motorola has contributed
a mvme88k port to OpenBSD and presumably the source is available
via their anon CVS server.

> all I have is 88100/88200 specific documentation.
> (The machines I have are running Mach 2.5, 2.6, and 3.0)

This should be a big help if you have access to machines that are
running code that you have the source to.

> Has there been any work done on any 88k machines?  I saw an old post in the
> archives about this, but no evidence that this got very far.

At least my attempt didn't get very far.

> Also, is there any work being done with multiprocessor support in NetBSD?

It will be hard enough aiming for the moving target that is netbsd-current.
I would certainly put off thinking about multiprocessor support until much
later.

> Whats the best way to go about doing an initial port to a new machine
> like this?

For every port its different based on what resources you have available
to you. I think it wouldn't hurt to start out with trying to get a 
cross compilation enviroment, a kernel printf, and then trying to 
build a diskless kernel with panic() stubs. If you get that far then
you deserve to get advice from someone better than me.

Jason