linux-mips-fnet
[Top] [All Lists]

Re: DECstation status

To: linux-mips@fnet.fr
Subject: Re: DECstation status
From: Matt Messier <marius@prilnari.com>
Date: Mon, 25 Sep 1995 10:25:27 -0400
In-reply-to: <199509251339.PAA08033@bandsept.univ-evry.fr>
On Mon, 25 Sep 1995, Stoned Elipot wrote:

> >>>>> "Matt" == Matt Messier <marius@prilnari.com> writes:
> 
> Matt> Does anybody out there with a DEC 3100 or 5000 have the ability
> Matt> to test this stuff?  It is a little bit difficult for me at the
> Matt> moment because I need to swap drives around all of the time to
> Matt> test things. I'm quite willing to work on writing the code if
> Matt> someone can test it easily for me.  Once things get a bit
> Matt> further along and we know that it is working somewhat reasonably
> Matt> well, I'll be a bit more willing to spend half an hour at a shot
> Matt> to test a change :-)
> 
> I got an unplugged personnal DEC 5000 station with Ultrix 4.4 on its
> disks at about 50cm on my right hand, can it be of any help ?? :)

Quite possibly :-)  The only problem is that once you run this, you'll lose
your Ultrix bootloader.  Also, my program to install the bootloader is written
for Linux/i386, but should work without much difficulty under Ultrix.

The machine that I am working with has a barebones install of Ultrix 4.2 on
it.  I don't have a compiler, or much of anything for it for that matter.  My
primary means of getting things onto that machine is to nfs mount from my 486
Linux box.  Once I start testing, I lose that ability and have to start
swapping the drive back and forth.

At any rate, I have determined that our cross development tools are
producing binaries with valid headers.  I haven't been able to get anything
to execute in application mode, but I haven't tried real hard either.

I am at work right now so I don't have time to put the stuff together for
testing elsewhere, but I will do so tonight.  By the way, where should I
ftp this stuff to?  What shall it be called, and should it be separated from
Milo?  Personally I think the answer to the Milo question is yes.  The text
and data segments need to fit within 15360 bytes.  All that needs to happen in
the loader (at least for now) is to load the kernel and jump to it's entry
point.  It also needs to properly handle prom restarts, etc.  Fairly straight
forward, but it also needs to have knowledge of the filesystem so that it can
find the kernel.  What I have now does little more than print some messages
out to the console via the firmware callback vector.

Matt Messier
The Prilnari Project
Lincoln, RI, USA

<Prev in Thread] Current Thread [Next in Thread>