Newsgroups: comp.os.minix
Path: sserve!newshost.anu.edu.au!harbinger.cc.monash.edu.au!simtel!news.sprintlink.net!howland.reston.ans.net!news.nic.surfnet.nl!sun4nl!cs.vu.nl!kjb
From: kjb@cs.vu.nl (Kees J. Bot)
Subject: Re: bios vs ps HD drivers
Nntp-Posting-Host: hornet.cs.vu.nl
References: <radugaDB6HyH.761@netcom.com> <DB8Iny.9DJ.0.-s@cs.vu.nl> <radugaDBAC4M.C6u@netcom.com>
Sender: news@cs.vu.nl
Organization: Fac. Wiskunde & Informatica, VU, Amsterdam
Date: Thu, 6 Jul 1995 12:03:40 GMT
Message-ID: <DBAMu5.1Jw.0.-s@cs.vu.nl>
Lines: 116

raduga@netcom.com writes:

>Kees J. Bot (kjb@cs.vu.nl) wrote:

>: I probably wrecked the I/O code.

>If you disable it altogether, its hard to tell 
>if its really wrecked or not. :)  

I only disabled writing, so it should read ok.  If you try to write to
the device then it will print something like:

	ps-hd0: write at 285/2/10:1024 not done

So you should be able to at least mount a device.

>: I rearranged both the PS and ESDI drivers without being able to test
>: them.  Luckily someone lend me a PS/2 to play with, so I could fix 1.7
>: and the ESDI driver up.  I learned alot about the interrupt controller.

>The ESDI driver does not recognise my drive.  I'm assuming this is
>expected behavior. Do you know which PS/2 systems use the ESDI
>and which use the PS type drives?

I've only seen one PS/2 in my life, with a Dutch keyboard, aargh.

>: The PS driver is still untested, because I couldn't find someone with
>: that kind of drive.  Until now.  :-)

>WOW. does this mean that i'm the only minix user of a ps-type PS/2 system
>left in existence?  
>(note, i've been fending off requests and polite suggestions from 
> family members and concerned friends, that i junk this machine 
> solely based on its age.  I dont find that argument compelling. :)

You can now claim you have a historic artifact.  And it also heats your
house in winter, and keeps the door open in the summer.

>SERIOUSLY though, is there any way I can help you test the driver
>more thoroughly? (you could maybe share bits of test code, 
>or tell me how to get diagnostics that would give you the info
>you need to fix it :) 

There is one test you can make, see if /dev/hd0 is accessible.  Run
'part' on it.  Good partition table or junk?  Does

	cp /dev/hd0 /dev/null

do something (stop it with DEL), does it immediately return, or does it
return with an error?

I'm gone from July 6 to 12, so no patches for some time.

>: The BIOS driver has the advantage that it is more reliable, because it
>: uses the code of the manufacturer.  On old interleaved drives it is just
>: as fast as a Minix driver.  I have been thinking about sending the XT
>: driver to /dev/null for these reasons.  Any reason why I should not?

>I assume when you say that its a REAL mode as oposed to a PROTECTED
>mode driver, that it will only function in the former.
>Can it be adapted to use in PROTECTED mode, or is it just too 
>much kludge and work, to even bother?

Actually, I have just managed to get the BIOS driver working in protected
mode, but it requires a 386.  A 286 can't return to real mode, unless
you hit it with a hammer.

>lastly, when trying to make my system bootable from the HD,
>i find installboot -d nearly always dies with "not enough 
>space on /dev/foo for 'boot'" even when its a brand-new
>partition with an empty new-made fs on it.

There must be a file 'boot' in the file system itself.  Installboot now
tries to attach 'boot' outside the file system.  It is best to copy ROOT
into your new file system before running installboot.

>after trying *several* times wiht utter frustration,
>i think im gonna give the auto-installation package a shot.
>Is there a way to do an auto-install with the TINYROOT
>and USR disks? looks like im gonna hae to use the real mode
>kernel, at least until i can get a working PS driver :(

You can boot from TINYROOT and then proceed as described in usage(8).
Use ROOT when it wants to make a root file system.

Note that the only advantages of a manual installation over an
automatic installation is that you get to choose the root and scratch
partition sizes yourself, instead of the defaults.

>when booting from TINYROOT and mounting /dev/hd2 
>with a full /usr and /src install, (manually)
>i run out of memory midway through the compile.

The shell of bin (/usr/bin/ash) is too big to leave enough memory for
compilations.  You can change the shell in /etc/passwd, or you can type
'exec sh' before doing anything memory intensive.

>it suggested I chmem some of the compiler utils
>(no luck) with the added frustration that 
>after an aborted make I had to remake from the beginning.
>Yes, i gave it -F on the cflags, no help.

Don't chmem anything, 'ncem' needs to be chmem'd to the max.  I will
change the message about chmem in the cc.c driver to something more
useful.

>is there any place I can send formal bug reports?

You happen to be talking to the Minix 1.7.0 bug registration office.

>Or do all the Minix developers read c.os.minix anyway
>and repeat posting would be a waste of breath :)

You also happen to be talking to the entire Minix development group.  :-)
--
								Kees.
