Subject: Re: 68060
To: None <amiga@NetBSD.ORG>
From: Michael L. Hitch <firstname.lastname@example.org>
Date: 02/19/1995 12:26:49
On Feb 19, 12:44pm, Stephen Champion wrote:
> > > Who's working on making NetBSD run on 68060? Who's going to?
> > And who's going to buy them the Trans-Warp(s) :^)
> The person(s) that do '060 support will be the first to have a
> meeting of several necessary prerequisites. They need the expertise, the
> '060 board, the time, and the desire. Since '060 boards have barely hit
> the market yet, it's virtually impossible to say who will do it - that
> person probably doesn't even know it yet, even if he has the time,
> expertise, and desire.
I probably have the expertise (after doing the 68040 - being the only
way I could get NetBSD on my system) and certainly the desire. Time
might be be something else though. There are several things I've done
in NetBSD that I need to do some more work on.
I haven't bothered learning anything about the '060 yet, as I haven't
had any need to. I doubt it would be all that difficult (for me at
Anybody want to provide me with a board? :-) :-) [I've spent enough
on '040 boards, A4000's, and disks already - I don't have any plans on
buying a '060 board any time soon.]
> However there are a few problems that may make this one takea
> while longer than some would like. From the little I've heard, the '060
> is not fully register and instruction compatible with any of the previous
> M680x0's. All portions of the OS that are done in assembly need to be
> looked at and checked for '060 compatibility. The parts done in C would
> be easy except for one thing - GCC 2.4.5 doesn't support the '060.
> Consequently, an '060 NetBSD would need to run with newer versions of
> GAS, GCC, libgcc, libg++, and probably other utilities as well. I'm not
> sure that even GCC 2.6.3 supports the '060, so this could be a problem.
> This should involve a compiler upgrade for all of NetBSD, to retain
> full compatibility.
I don't think the compiler or assembler is much of an issue - assuming
that the '060 is upward compatible to the '020/'030 for user mode
instructions. Dealing with '060-specific kernel mode instructions
shouldn't be much different that doing the '040-specific instructions.
The current kernel doesn't even require that gcc/gas be able to handle
the '040 (the kernel is only compiled with -mc68020). All the
'040-specific instructions are coded as data.
Michael L. Hitch INTERNET: email@example.com
Office of Systems and Computing Services
Montana State University Bozeman, MT USA