Path: utzoo!attcan!uunet!mcsun!hp4nl!star.cs.vu.nl!sreiz
From: sreiz@cs.vu.nl (=Steven Reiz)
Newsgroups: comp.os.minix
Subject: AMIGAMINIX USERS: READ THIS!
Keywords: amiga minix bug fix
Message-ID: <8009@star.cs.vu.nl>
Date: 22 Oct 90 14:50:18 GMT
Sender: news@cs.vu.nl
Lines: 52

Sorry for the loud subject line, but I posted this last week and got 0
(yes zero) responses to it, so i'm reposting it, slightly modified.

Recent amigas (A500s and A2000s with a 1MB agnus and A3000s) can have
(chip) memory in the $080000 - $100000 range and I just discovered that
minix has a bit of a problem with that... 

In fact on those machines it will add the screen memory and the disk buffers
to the free memory list. This means that if you execute a small minix program,
like ls, it will trash the disk buffers, and if you execute a large program,
or a program with a large data segment, like diskcopy, it will trash the
diskbuffers and the screen memory.

Indeed some people have reported exactly those problems: disk errors when
using various small programs and diskcopy messing up the screen.
If you had minix trashing your floppies too, please try the quick&dirty fix
that is explained below. It will let minix think that your machine does
not have memory at $080000 - $0c0000:

Change your BOOT:s/startup-sequence, which should currently be:

BOOT:c/assign devs: BOOT:devs
BOOT:c/setmap m_usa1
BOOT:Minix -s 0 -e $200

to:

BOOT:c/assign devs: BOOT:devs
BOOT:c/setmap m_usa1
BOOT:Minix -s 0 -e $200 -n $080000

Note that I assume here that you have 'installed' the BOOT disk.

Please let me know if this fix works or not!
You can email your findings to amiga@cs.vu.nl or sreiz@cs.vu.nl.
If you do not receive email back from me within two days saying that
I received your mail, could you please post your findings to this newsgroup?
I hope to prevent tens (hundreds? thousands?) of postings saying
'yes, it worked' or 'no, it !@#$% didn't work for me!' in this way
that are totally of no interest to most readers of this newsgroup.

I will summarize to the net as soon as possible.
Unfortunately the code that causes the bug is part of the minix kernel, not
the loader, so a proper fix will require a new version of minix.img.
When I am sure how this bugfix works out I will recompile minix.img and
post it.

Please let us know of any other problems that you have with minix as well!

Regards,

                        Steven Reiz