Path: utzoo!attcan!utgpu!jarvis.csri.toronto.edu!mailrus!cornell!uw-beaver!fluke!ssc-vax!bcsaic!paula
From: paula@bcsaic.UUCP (Paul Allen)
Newsgroups: comp.os.minix
Subject: Re: 386 Minix?
Message-ID: <15541@bcsaic.UUCP>
Date: 5 Oct 89 20:56:30 GMT
References: <1588@ncrcce.StPaul.NCR.COM>
Reply-To: paula@bcsaic.UUCP (Paul Allen)
Organization: Boeing Computer Services AI Center, Seattle
Lines: 43

In article <1588@ncrcce.StPaul.NCR.COM> rogers@ncrcce.StPaul.NCR.COM (Bob Rogers) writes:
>I'm considering buying a PC and am trying to decide if the extra money
>for a 386 is worth it.  Is there currently a version of Minix for the
>386/386sx or is one in the works?  

IMHO, what you get with the 386 is the potential to get away from
64K segments.  You might be able to get a 286 for a bit less than
a 386, but not much less.  There's a company here in Bellevue
selling 25MHz 386's for $1600 with Hercules monochrome, 1Mb, 65M
RLL disk, and serial/parallel ports.  I bought one with 4Mb, a
16-bit VGA and color monitor, and extra serial port for $2700
including the 8% tax.  Minix runs *real* nice with a 3Mb ramdisk!

There's no 386 Minix today, but there will be one.
I'm tinkering in my spare time, and I know Bruce Evans has put a
lot of thought and time into a real 32-bit 386-based Minix.  There
are no doubt other hackers out there working on stuff.

Since I reached a sort-of milestone last night, perhaps this is a
good time to post a status report.  I have GNU gcc (1.35), gas (1.34),
and ld (May 1, 89) running on my Sun and generating 386 a.out files.
Last night, for the first time, I successfully linked two 386 object
files to produce an executable.  (Of course, it's a BSD-style a.out
and I have no place to run it, but hey, what do you want?  :-) )
I've merged the disassembler portion of Bruce Evans' 386 debugger
into the GNU objdump program.  From eye-balling the disassembled
a.out file, the only problem I see is that the loader appears to
be screwing up the resolution of addresses in calls to external
functions.  It's probably a byte-order problem.  (*All* of the
problems so far have been byte-order related.  Gcc and gas work fine
in a cross-development environment, but ld assumes that it's running
on the target machine.  :-( )  I may get a chance to look at it this 
evening.

The next step is to build a translator to go from the current BSD
a.out format to something more useful and start moving little test
programs over to my 386 box.

Paul Allen
-- 
------------------------------------------------------------------------
Paul L. Allen                       | pallen@atc.boeing.com
Boeing Advanced Technology Center   | ...!uw-beaver!bcsaic!pallen