Path: utzoo!attcan!uunet!mcvax!hp4nl!botter!star.cs.vu.nl!ast
From: ast@cs.vu.nl (Andy Tanenbaum)
Newsgroups: comp.os.minix
Subject: PS/2  [was: mail to ast bounces]
Message-ID: <1870@ast.cs.vu.nl>
Date: 7 Jan 89 20:54:43 GMT
References: <4391@pitt.UUCP>
Reply-To: ast@cs.vu.nl (Andy Tanenbaum)
Organization: VU Informatica, Amsterdam
Lines: 62

In article <4391@pitt.UUCP> crew@vax.cs.pittsburgh.edu.UUCP (Al Crew) writes:
>Mail to you [ast] has been bouncing.  
Mail from me to you has also been bouncing.  My path is:
   ... uunet!mcvax!cs.vu.nl!ast
Please send me mail with your exact path from uunet (or some other major node).

>4) I plan to modify floppy.c to handle 1.44MB drives. (You are not
>   working on that are you?)
So far we can't get the floppy to work at all.  The current state is as
follows.  When the floppy driver is called for the first time on a Model 50
or higher, it (correctly) sees that it is not calibrated, and issues a
calibrate command.  Then it waits for the interrupt.  No interrupt
ever happens and the whole thing hangs forever.  The suspicion is  that
interrupts are not being initialized properly, or something microchannelish
is not being set up right.  DMA may also be a problem later on, but the
recalibrate doesn't use it.  If anyone has any ideas what to do, PLEASE
post them.  On the Model 30, everything works fine, so the problem apparently
relates to some difference between the 30 and the 50. It isn't the CPU difference
because MINIX works fine on plain old ATs, which have an 80286, just like the
Model 50.  If we can get recalibrate to work, we can start thinking about
1.44MB drives.  The Model 50 uses a NEC 765 chip, just like the PC, so 
setting up 1.44MB is simply a matter of adding a new table entry.  If anyone
knows the values to include, please post them.

>5) Warm reboot of Minix on the PS/2 Model 30-286 hangs immediately
>   after I type '='.  None of your print statements appeared.
No idea.  Does cold boot work ok?

>6) This semester I plan to work on an "operating system debugger"
>   for Minix.  Does anything like that exist?
Not that I know of.  My guess is that it won't help much.  Debugging
operating systems is not like debugging user code.  The problem with the
floppy is undoubtedly that you have to stuff some random value into some
equally random I/O port to make a Model 50 work, and the documentation
says nary a word about it.  I don't think a debugger will help there.
What is needed is an inspired guess done by someone who knows the Model 50.
(or other PS/2 model > 50).  Get out your crystal ball.

>7) Do mkfs prototype files exist for the V1.3 distribution.  
Not exactly.  The distribution was built in a somewhat funny way.  As long
as you have all the files and the checksums agree with the list I posted,
I wouldn't worry about which file goes on which floppy disk.
I would suggest making a diskette for each directory (kernel, mm, fs, etc.)
if you have 360K, or group related things together if you have larger ones.
In any event, be sure to keep a write protected 1.3 around to use as the
base when future updates are posted.

I have been corresponding by email with several people about trying to
get the PS/2 stuff working.  This has been frustrating due to flaky mail.
In particular, the only person who has a working MINIX system in the same
room as a PS/2, Steve Ackerman of uvm, is unreachable.  I think we
should conduct the PS/2 discussion in the newsgroup instead of by mail,
first because it is more reliable, and second because there are probably
other people around with a PS/2 who might be interested and/or have good
suggestions.  And unfortunately, I don't think we will make any progress
until somebody hits upon the answer.

One issue that is unresolved is the planar control register (port 65 on the
model 30).  Is there an analogous thing on the model 50, are there more of
them, what should be done with them, etc.  Suggestions are welcome.

Andy Tanenbaum (ast@cs.vu.nl)