riscy
[Top] [All Lists]

Re: Ping

To: riscy@sunsite.unc.edu
Subject: Re: Ping
From: Tommy Thorn <tthorn@daimi.aau.dk>
Date: Thu, 16 Dec 1993 16:34:54 +0100
In-reply-to: <9312161504.AA03906@resi.waldorf-gmbh.de>
Reply-to: Tommy.Thorn@daimi.aau.dk
Being not much of a HW guy, but wanting to do at least something,
I've been studing the MMU and exception handling of the r4000 and
r4200 carefully. I even consider extending the SPIM r2000/r3000
simulator to cover MIPS III (the ISA of r4x00). I've also been
playing with cross development tools.

I'm a bit worried about porting Linux without some sort of
coordination with Linus. From what I hear, the (ongoing) m68k
port hasn't tried to isolate machine dependent parts. Shouldn't
we/someone contact Linus and the kernel channel (I'm on it)?

Now for an old mail:
Daniel Veillard writes:

 >  from all I've read concerning the riscWS/EISA chipset we
 > plan to use, and information I got from the net, it seems that
 > the future board will be very similar to the Magnum 4000 machine:

stuff deleted

 >  This kind of compatibility may be interesting for a fast
 > port of Linux if we were able to get access to the source of other
 > operating system for the Magnum (may help for Device drivers
 > and interface with CPU/MMU). I have also heard that Plan 9
 > distributed operating system have been ported to the Magnum R4000
 > and that free source licence is available for Universities.
 > Getting sources of other OS running on Misp Magnum 4000 may help
 > developping Linux and that's something we can do before the board
 > is available.
 > 
 >  Furthermore, staying compatible with the Magnum products may be
 > a good way to offer binary compatibility and run commercial products
 > on Linux.
 > 
 >   - Am I completely wrong ?
 >   - Is there any other sources available publicly which may
 > help porting Linux ?
 >   - Any comments ?

Even getting access to a commercial unix for the MIPS is very
dangeous. Don't even think about it. We would risk a lawsuit.

Every bit of code in Linux must be *absolutely* free (in the GPL
sense).

There are other publicly sources which we could study: NetBSD
is ported to MIPS, I think. Mach is for sure, but double check
copyright and licence.

I really don't think we would gain much though, as the hard part is
always getting good HW docs. This should not be a problem with riscy.

Offering binary compatibility might be doable and possible worthwhile.

What I would much rather see is the software supplied the ARC 100
design, if any. I would like to know if we get the CPU from reset,
or if there is a boot PROM we could/should build on?

I hope I can afford riscy.

/Tommy

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