Newsgroups: comp.os.minix
Path: euryale.cc.adfa.oz.au!newshost.anu.edu.au!harbinger.cc.monash.edu.au!aggedor.rmit.EDU.AU!goanna.cs.rmit.edu.au!munnari.oz.au!spool.mu.edu!usenet.eel.ufl.edu!news.mathworks.com!newshost.marcam.com!usc!crash!cwr
From: cwr@crash.cts.com (Will Rose)
Subject: Re: Minix success report and questions
Organization: CTS Network Services (CTSNET), San Diego, CA
Date: Mon, 4 Sep 1995 13:27:29 GMT
Message-ID: <DEDupu.5B0@crash.cts.com>
X-Newsreader: TIN [version 1.2 PL2]
References: <JCH.95Sep2180453@maddalena.dm.unibo.it>
Sender: news@crash.cts.com (news subsystem)
Nntp-Posting-Host: crash-i2.cts.com
Lines: 76

Juliusz Chroboczek (jch@dm.unibo.it) wrote:
:   I just installed Minix 1.7.0 beta on a 286/12, 1Mb RAM and 20Mb HD.
: It works great --- thank you AST & KJB!  The installation went without
: a problem, except that I could not load root into the ram disk (not
: enough core).  That was no problem, though, as I have two floppies.

:   After playing with it for a few days, I've got zillions of stupid
: questions to ask.  Here is a sample.

: o Why are the console and the other ttys in a single driver?  It seems
:   that they are handled completely differently.

A console is a stream of bytes, like any other tty.  The implementation
may vary a bit, tho...

: o Am I right in understanding that (a fixed number of) virtual
:   consoles would be trivial to implement?  Simply associate each
:   console with a different area of RAM, and implement a key that swaps
:   a virtual screen with the physical one, and changes the base
:   addresses of the screens.  Or am I missing something?

Get the 1.5 virtual console code from the archives - it wasn't that big.
I can't remember how it was implemented, but a lot of people liked it.

: o From both reading the source and experimenting (cp /dev/fd0 /dev/fd1),
:   I've got the feeling that FS is single-threaded which, with, as FS
:   calls the device drivers synchronously (i.e. FS sleeps until the
:   transfer is finished), often slows the system to a crawl.  Is there
:   any patch available to this?  Preferable with code that a Minix
:   newbie can understand.

That was a problem under 1.5, and probably still is, but with today's
larger buffers it's a lot less noticeable.

: o The Unices I know have two sets of devices for disks -- raw and
:   cooked.  Why doesn't Minix need that?

Well, it has hd0 for the disk and hd1...4 for the filesystems.
Is that what you mean?  I've usually seen cooked applied to tty
output, so I'm missing something.

: o Is there any way to get more then 64Kb+64Kb (on a 286) other than
:   forking for each segment of RAM?

Write a compiler that understands the more baroque Intel memory
models.  That was discussed a bit before the 386 came out, and 
the problem then became moot.

: o Are there any patches to get ptys (that's something I've never
:   really understood in Unix, and I'd love to get my hands on some
:   source)?  Process groups?  Job control?

Ptys are in tnet (for 1.5) and in some other places.  Process groups
I don't think so.  Don't mention the war, er... job control in this
group again.

: o Should I buy AST's book now, or wait for a new edition (this type of
:   book costs about twice as much here than in the States).

Depends how keen you are on books - it's a very good one, but the release
date of the next edition is not set.  And if it was, I wouldn't believe it.

: o Who are AST and KJB anyway?

The author and mainter of Minix, respectively.

:   And, well, perhaps c.o.minix could become a little more technical
: now.  I've been seeing installation questions all the time...

In the beginning was the deed; then came marmelade.

Will
cwr@crash.cts.com

(Sorry about the unwonted frivolity - I'm short of sleep).

