Path: sparky!uunet!psinntp!iscnvx!leadsv!daver!fifi.pcs.COM!...@fifi.pcs.COM
From: j...@fifi.pcs.COM (Jordan K. Hubbard)
Newsgroups: comp.sys.nsc.32k
Subject: LINUX
Message-ID: <9201101332.AA00383@fifi.pcs.com>
Date: 10 Jan 92 13:32:35 GMT
Sender: n...@daver.bungi.com (Network News)
Organization: Association for the Prevention of Polar Bears and Kangaroos
Lines: 11
Has anyone had a close look at LINIX from the Finns? It's AT&T code free,
supports rudimentary POSIX standards and is reasonably V.3 compatable.
It also comes with full source and less restrictive redistribution rights
than Minix (basically, if it ran on the PC532, it would be freely
distributable). There are several announcements pertaining to it on
comp.os.mach, comp.os.minix, etc.
Bruce? :-)
Jordan
Newsgroups: comp.sys.nsc.32k
Path: sparky!uunet!UB.com!daver!cs.wwu.EDU!p...@cs.wwu.EDU
From: p...@cs.wwu.EDU (Phil Nelson)
Subject: LINUX
Message-ID: <9201101819.AA00302@strawberry.cs.wwu.edu>
Sender: n...@daver.bungi.com (Network News)
Reply-To: p...@cs.wwu.edu
Organization: Association for the Prevention of Polar Bears and Kangaroos
References: <9201101332.AA00383@fifi.pcs.com>
Date: Fri, 10 Jan 1992 18:19:01 GMT
>Has anyone had a close look at LINIX from the Finns? It's AT&T code free,
>supports rudimentary POSIX standards and is reasonably V.3 compatable.
Yes, I have. I started looking at version 0.02.... It would take MANY
hours to port to the pc532. Linus did not make any effort to localize
the machine specific parts. In fact, the sources have many "asm(...)"
statements in them, including the library. (I just looked at parts of
it last night.) And the file system has many of the "problems" that
minix has . . . 14 character names, 64 Meg maximum sized partitions ...
This is not to say we should ignore it. It has some things that Minix
is missing, ... multi threaded file system, real paging to/from disk,
ptys, POSIX job control, ...
I did think it would be fun to play with, but it is moving so fast
that it would be hard to keep up with during any porting efforts. It is
possible that some of the pc532 specific code from minix could be used
in Linux (SCSI driver(s), some of the rs232 code, ....)
I would love to see it working on a pc532. We could then have a completely
copyable (source and binaries) *UX for the pc532!
But then I think about the fact that Mach 3.0 is almost working on the
pc532 and that the Hurd is under way. But then again, linux is likely
to be much smaller than mach/hurd. Who knows?
Isn't this FUN, all these different possibilities?
--Phil
Newsgroups: comp.sys.nsc.32k
Path: sparky!uunet!UB.com!daver!mthvax.cs.miami.EDU!ya...@mthvax.cs.miami.EDU
From: ya...@mthvax.cs.miami.EDU (Yanek Martinson)
Subject: Re: LINUX
Message-ID: <9201110122.AA19149@mthvax.cs.miami.edu>
Sender: n...@daver.bungi.com (Network News)
Organization: Association for the Prevention of Polar Bears and Kangaroos
References: <9201101332.AA00383@fifi.pcs.com>
Date: Sat, 11 Jan 1992 01:21:59 GMT
> Has anyone had a close look at LINIX from the Finns? It's AT&T code free,
It is LINUX not LINIX. From name of the author, Linus Torvalds.
I am using it on my 386, it is very good. I am typing from it right
now.
> than Minix (basically, if it ran on the PC532, it would be freely
There is no way to make it run on PC532 it needs a 80386 or 80486
processor.
Path: sparky!uunet!psinntp!iscnvx!leadsv!daver!fifi.pcs.COM!...@fifi.pcs.COM
From: j...@fifi.pcs.COM (Jordan K. Hubbard)
Newsgroups: comp.sys.nsc.32k
Subject: re: LINUX
Message-ID: <9201131811.AA00965@fifi.pcs.com>
Date: 13 Jan 92 18:11:55 GMT
Sender: n...@daver.bungi.com (Network News)
Organization: Association for the Prevention of Polar Bears and Kangaroos
Lines: 29
>hours to port to the pc532. Linus did not make any effort to localize
>the machine specific parts. In fact, the sources have many "asm(...)"
>statements in them, including the library. (I just looked at parts of
Yes, I know. Sorry I didn't make this clear in my first posting.
It's pretty obvious that the VM stuff is going to have to be
completely restructured, but then again I suppose that is what
generally happens when a kernel wants to suddenly start running on
more than on architecture (the early Unices were very PDP dependant).
It would be a two stage process: First identify the offending 386
dependancies then encapsulate them into a more general mechanism (and
write pc532 support using that mechanism). It would be a lot of work,
yes.
As Phil says, too, Mach/Hurd are coming along fairly well and definately
make more allowances for differing architectures. I find it odd that the
author of LINUX would have the skill to write his own kernel from scratch
but would be so shortsighted as to imbed 386 dependencies into the
kernel at random. A quick read of Tannenbaum would have been enough to
discourage that. Oh well.
I thought it merely interesting food for thought. I still have no idea
when the UX server or a suitable replacement for Mach is going to be AT&T
free. I have the second network release tape of 4.4 BSD here and I haven't
had a chance to look at what it's missing yet. We definately need to jump
start the PC532 world with something *soon* (and saying that, I guess I'll
maybe turn mine back on tonite and work on the Forth some!).
Jordan
Path: sparky!uunet!caen!zaphod.mps.ohio-state.edu!mips!daver!
hplwbc.hpl.hp.COM!culbe...@hplwbc.hpl.hp.COM
From: culbe...@hplwbc.hpl.hp.COM (Bruce Culbertson)
Newsgroups: comp.sys.nsc.32k
Subject: re: LINUX
Message-ID: <9201142021.AA04193@hplwbc.hpl.hp.com>
Date: 14 Jan 92 20:21:01 GMT
Sender: n...@daver.bungi.com (Network News)
Organization: Association for the Prevention of Polar Bears and Kangaroos
Lines: 33
Jordan writes:
> I find it odd that the author of LINUX would have the skill to write his
> own kernel from scratch but would be so shortsighted as to imbed 386
> dependencies into the kernel at random. A quick read of Tannenbaum
> would have been enough to discourage that. Oh well.
I am sympathetic with Linus. I seriously doubt that he is unaware
of the virtues and methods of portability.
The time any of us can devote to projects which we intend to give away
is very limited. To have any chance of success when we undertake a
large not-for-profit project, we have to narrow the scope as much as
possible. "Runs on 386's" is a much narrower scope than "runs on
386's but is easily ported to other architectures".
If we don't plan to make money on a project, then we need some other
motivation; usually, the motivation is fun. If Linus doesn't find
portability fun, we can't expect him to devote too much time to it.
Finally, making OS's portable across architectures is extremely
difficult. Many of the best OS authors have made decisions which made
porting difficult, even when portability was one of their top
priorities. Mach is a good example. Mach makes assumptions about the
ability to alias several virtual addresses onto the same physical
address. The popular workstation architectures which use virtually
addressed caches cannot efficiently implement the aliasing Mach needs,
even though they are well suited to BSD and SysV. Minix is another good
example. To take advantage of the 32532's MMU, I had to significantly
overhaul Minix's memory management code. My changes went far beyond the
intended machine dependent routines and I believe that was unavoidable.
Bruce
Path: sparky!uunet!ukma!wupost!spool.mu.edu!mips!daver!zen.cs.hut.fi!
...@zen.cs.hut.fi
From: a...@zen.cs.hut.fi (Ari Lemmke)
Newsgroups: comp.sys.nsc.32k
Subject: Re: LINUX
Message-ID: <199201160213.AA23147@zen.cs.hut.fi>
Date: 16 Jan 92 02:13:32 GMT
Article-I.D.: zen.199201160213.AA23147
Sender: n...@daver.bungi.com (Network News)
Organization: Association for the Prevention of Polar Bears and Kangaroos
Lines: 44
>It is LINUX not LINIX. From name of the author, Linus Torvalds.
He wanted to change the name ;-) but I created FTP site
and mailing list based on that name .. so one can
see he's resposible of it (har har). And Linux was better
name than the other.
>> than Minix (basically, if it ran on the PC532, it would be freely
>There is no way to make it run on PC532 it needs a 80386 or 80486
>processor.
I think he meant portig it to the PC532.
It is possible to port it, but it will be hard, because
Linus was using all 386 features to do it (because
Linux was his exercise ;-)
I asked Linus to hack direct asm's away .. so he made
them as macros .. which made Linux little bit more
portable.
But if you check the code, you should check newest ones,
i.e. 0.12 (still alfa). Available at nic.funet.fi:/pub/OS/Linux
and tsx-11.mit.edu:/pub/linux (tsx-11 is much better for
the US sites).
If you have good ideas about Linux, or would like it to be
more portable, please send the ideas/diffs to Linus, so
he can make them available to the next release and you
don't have to do the porting again (i.e. when the next
release comes out). Linus has dead-lines (about
one week from requesting patches) until he starts to make
the next version.
Mailing list requests: linux-activists-requ...@niksula.hut.fi
Quite many articles and questions etc. asked on list,
so if you don't have big-enough mailbox don't subscribe.
I'd love to run Linux with my PC532 (it might take a long
while until we get Mach/Bsd to it).
arl
Path: sparky!uunet!dg-rtp!psinntp!rpi!think.com!mips!pacbell.com!UB.com!
daver!manatee.UUCP!jo...@manatee.UUCP
From: jo...@manatee.UUCP (john connin)
Newsgroups: comp.sys.nsc.32k
Subject: re: LINUX
Message-ID: <9201160030.AA25156@manatee.UUCP>
Date: 16 Jan 92 05:30:27 GMT
References: <9201142021.AA04193@hplwbc.hpl.hp.com>
Sender: n...@daver.bungi.com (Network News)
Organization: Association for the Prevention of Polar Bears and Kangaroos
Lines: 44
>
> Jordan writes:
>
> > I find it odd that the author of LINUX would have the skill to write his
> > own kernel from scratch but would be so shortsighted as to imbed 386
> > dependencies into the kernel at random. A quick read of Tannenbaum
> > would have been enough to discourage that. Oh well.
>
Bruce C writes
> I am sympathetic with Linus. I seriously doubt that he is unaware
> of the virtues and methods of portability.
>
> The time any of us can devote to projects which we intend to give away
> is very limited. To have any chance of success when we undertake a
> large not-for-profit project, we have to narrow the scope as much as
> possible. "Runs on 386's" is a much narrower scope than "runs on
> 386's but is easily ported to other architectures".
[ stuff deleted ]
I have to agree with Bruce -- and of course he should know !!
Some time back I took a quick peek at Linux and was quite impressed.
In my eye the code was elegant in that it was clean (particularily
considering it was alpha 0.10+), and leveraged off of the 386 hardware
capabilities in a lean insightful manner.
I really think it would be "fun" to port Linux to the PC532 and track/
contribute to Linuxs' development. However, unfortunately like most, if
not all of us, I simply do not have the time -- hell, I don't even have
my video board programmed yet !!
On the other hand, if there was some reasonable way to attack a BSD
port I would sorely be tempted to try and make the time available.
humm.. I wonder if porting BSDI's BSD/386 would be reasonable ??
best regards,
johnc
--
John Connin: manatee Orlando, Florida
UUCP: {uunet,ge-dab,ucf-cs}!tarpit!tous!manatee!johnc
Path: sparky!uunet!think.com!mips!daver!fifi.pcs.COM!...@fifi.pcs.COM
From: j...@fifi.pcs.COM (Jordan K. Hubbard)
Newsgroups: comp.sys.nsc.32k
Subject: whoooo.. now I've done it.
Message-ID: <9201161745.AA01936@fifi.pcs.com>
Date: 16 Jan 92 17:45:52 GMT
Sender: n...@daver.bungi.com (Network News)
Organization: Association for the Prevention of Polar Bears and Kangaroos
Lines: 43
>Some time back I took a quick peek at Linux and was quite impressed.
>In my eye the code was elegant in that it was clean (particularily
>considering it was alpha 0.10+), and leveraged off of the 386 hardware
>capabilities in a lean insightful manner.
Ok, I guess I should apologise for the tone of my original criticism -
it was overly judgemental (I can here the jury yelling "And just how
many operating systems have *you* written, wise guy??"). I still
stand on my point that one can be very system dependent yet still take
care to at least *isolate* the dependencies in such a way that they're
obvious. As someone else said, Linus has already been pursueded into
using macros for some of the asm stuff (I don't know where or how yet,
since 0.12 is not released).
In any case, I think it's sort of moot since most of us seem to be
drawn to some varient of BSD, and it looks the CSRG is moving very
quickly to a free'd release. As someone mentioned, BSDI has been
advertising a 386 version with full source for $1000 (apropos: Is this
really available yet? Has someone actually ordered it?) which means
that there can't be all *that* much missing. I know Guy Harris has
been working on some Sparc release recently, and I suspect it's based
on the networking II tape as well.
Perhaps we should all ask Bruce, as our uncontested PC532 OS expert,
what *his* opinion is.
Myself, I'm still fighting to bring up Mach on my strange and weird
486 box so that I can start cross developing for the PC532. This
seemingly simple task has turned into a 7 month odyssey of misery and
woe for both myself and Gary, one it seems we're finally on the brink
of solving only after resorting to the age old expedient of throwing
lots of money at the problem. I got Mach to boot single user last
night and hope to get the C compiler working tonite (yay!).
Whether Mach or BSD 4.x, however, much of the code is the same. The
more collective BSD expertise we develop in this group, the better
poised we'll be to take advantage of HURD, 4.4 BSD or a free UX server
(not imminently likely). I think this lineage holds the most promise.
Now if there was only some solution on the horizon for X. I guess the
frame buffer project has finally gone Tango Uniform, eh?
Jordan
Path: sparky!uunet!ukma!wupost!sdd.hp.com!mips!daver!mthvax.cs.miami.EDU!
ya...@mthvax.cs.miami.EDU
From: ya...@mthvax.cs.miami.EDU (Yanek Martinson)
Newsgroups: comp.sys.nsc.32k
Subject: Linux 0.12
Message-ID: <9201162211.AA05996@mthvax.cs.miami.edu>
Date: 16 Jan 92 22:11:01 GMT
References: <9201161745.AA01936@fifi.pcs.com>
Sender: n...@daver.bungi.com (Network News)
Organization: Association for the Prevention of Polar Bears and Kangaroos
Lines: 5
> using macros for some of the asm stuff (I don't know where or how yet,
> since 0.12 is not released).
Linux 0.12 IS released, available at the usual ftp sites
(tsx-11.mit.edu and nic.funet.fi)
Path: sparky!uunet!mcsun!news.funet.fi!funic!nic!vinsci
From: vin...@nic.funet.fi (Leonard Norrgard)
Newsgroups: alt.os.linux
Subject: Anyone porting Linux to m68k?
Message-ID: <VINSCI.92Jan21181059@nic.nic.funet.fi>
Date: 21 Jan 92 17:10:59 GMT
Sender: vin...@nic.funet.fi (Leonard Norrgard)
Distribution: alt
Organization: Soft Service, Inc.
Lines: 5
Well, the subject line says it all: is anyone working on a port of
Linux to m68k processors? I know it's supposed to be non-trivial,
but it can't be impossible...
-- Leonard
Newsgroups: alt.os.linux
Path: sparky!uunet!zaphod.mps.ohio-state.edu!van-bc!ubc-cs!alberta!martin
From: mar...@cs.UAlberta.CA (Tim Martin; FSO; Soil Sciences)
Subject: Re: Anyone porting Linux to m68k?
Message-ID: <martin.696018485@menaik>
Sender: n...@cs.UAlberta.CA (News Administrator)
Nntp-Posting-Host: menaik.cs.ualberta.ca
Organization: University of Alberta, Edmonton, Canada
References: <VINSCI.92Jan21181059@nic.nic.funet.fi>
Distribution: alt
Date: Tue, 21 Jan 1992 18:28:05 GMT
vin...@nic.funet.fi (Leonard Norrgard) writes:
>Well, the subject line says it all: is anyone working on a port of
>Linux to m68k processors? I know it's supposed to be non-trivial,
>but it can't be impossible...
Might be easier to simply run Linux under the PC-Soft 386 emulation
software. (Assuming the m68k is a Mac box.) I wonder if that would
work?
Tim.
-------------------------------------------------------------
Tim Martin *
Soil Science * These opinions are my own:
University of Alberta * My employer has none!
mar...@cs.ualberta.ca *
-------------------------------------------------------------