Subject: Re: Linux on PowerMac
To: Ken Nakata <kenn@romulus.rutgers.edu>
From: Nick STEPHEN <stephen@gr.osf.org>
List: port-mac68k
Date: 02/07/1996 15:41:50
>> In fact, we've ported the Mach kernel to the PowerMac and we've ported
>> Linux to Mach as an architecture-neutral pseudo machine. There are
>
>Hm, are the code available somewhere on the net?

Not yet, but it will be soon. We want to reach a little further down
the road before releasing it. See apple's page:

http://www.mklinux.apple.com

>BTW, may I ask you why you've chosen Linux over *BSD?  To me, it's an
>odd decision because Linux's strengths always seemed to lie in its
>tightly-coupled-ness with the Intel PC hardware (how many people have
>told us that Linux is faster than *BSD?  Yeah, I know more than half
>of those are groundless B.S.).  Then it seems to me that there is no
>reason that we should chose, for instance, ext2fs over ffs, a proven
>technology that has over 10 years of real use.  Also, the lightness
>Linux users often attribute is due to Linux kernel's somewhat overly
>simple scheduling algorithm, and it's gone if scheduling is done by
>Mach MK.  If it's the brand, you can tell me.  I won't get heart
>attack from hearing it ;-)

See the paper we presented at the FSF conference, which is under

http://www.gr.osf.org/mklinux

Basically, there are already x BSD-ish servers running on Mach, and
the vm structure of FreeBSD is basically taken from Mach these
days... we wanted to experiment with a different vm architecture and
show that it could run just as efficiently over Mach. It does.

>Oh yes, there IS Lites...  It would be much easier to recompile Lites
>on MK/PowerMac than to port Linux server to it.

We wanted to reuse the native linux sources to gain leverage from the
work being done on that source base. Someone else can recompile lites,
it would take a little work because of the minor differences to the
microkernel interface.

>BTW, again, speaking of SMP, are you planning to port this MK to
>DayStar's Genesis MP (four 604s clocked at 150MHz)?  It'd be a shame
>if this SMP-ready kernel weren't to run on that screamer...

No plans for now, we're already overloaded with work. Of course,
there's nothing stopping someone else from doing it once the sources
are out there, we'd be happy to merge in changes.

                [ Nick ]