Path: utzoo!utgpu!news-server.csri.toronto.edu!mailrus!wuarchive!zaphod.mps.ohio-state.edu!samsung!munnari.oz.au!bruce!cechew
From: cechew@bruce.cs.monash.OZ.AU (Earl Chew)
Newsgroups: comp.os.minix
Subject: Re: shoelace and PS/2 30
Message-ID: <2994@bruce.cs.monash.OZ.AU>
Date: 31 Aug 90 23:14:01 GMT
References: <29110@nigel.ee.udel.edu>
Organization: Monash Uni. Computer Science, Australia
Lines: 79

csc239@central1.lancaster.ac.uk (A. S. Lunn) writes:

>Over the last couple of days I've installed shoelace with minix 1.3
>system on a PS2 model 30 so that it boots off /dev/hd2 and takes a ram
>image from /dev/hd3. The fs has been modified to load the ram disk from
>/dev/hd3 without looking at the floppy first. I've run into two
>problems, one major and the another minor.

Sigh! It looks like it's time to work on ShoeLace again. To date I've only had
one major bug report to do with certain XT hard disk controllers and their
associated roms. Apparently they forget the hard disk parameters. There is a
fix but:

1. the person who did the fix didn't send it to me
2. I haven't got around to replicating it.

Regarding the problems with PS2/30s. I don't have access to one of these
beasts, so in some sense you're on your own. Remote debugging from the other
side of the planet is too painful.

Now grasping at straws:

1. Did the floppy drive work properly *before* the change to fs?

To summarise:

1. Using a hard disk boot floppy with winiboot works without a hitch.

2. Using winiboot directly off the hard disk works for MSDOS:
	a) Sector not found errors

   for Minix:
	a) floppy problems:
	      On mount:
		    Unrecoverable disk error on device 2/0, block 1 
	      Unexpected interrupt to vector 2
	b) CTL_ALT_DEL problems

>If I cat the first couple of blocks of /dev/hd0 onto a floppy and boot that
>the floppy drive works fine. 

This would suggest that something in the floppy isn't being initialised.

>If I boot into MS-DOS using winiboot the floppy  giving sector
>not found error. 

This is mystifying since the partition table boot code (winiboot or the
original) doesn't (to my knowledge) muck around with the floppy. Try
disassembling the original partition table boot code to see how it differs from
winiboot.

>My guess is that its something to do with the parameter table or

This does not seem likely since once Minix gets control, all floppy accesses
are done by the floppy driver the floppy parameter table is only used by the
BIOS.

>The other minor problem is when rebooting from minix with CTRL-ALT-DEL
>winiboot prints the partition table, waits for about half a second and
>then boot the default partition. With a cold boot it waits the fifteen
>seconds before booting the default partition. It also does not check is
>there is a bootable floppy in the drive. CTRL-ALT_DEL from MS-DOS works
>properly.

Which is the `it'? winiboot was not designed to check for the existence of a
bootable floppy. Sounds to me like there is something amiss with the warm boot
code in the BIOS. Hmmm... I think that this might have something to do with the
clock tick rate when rebooting from Minix. Perhaps the BIOS is not resetting
the clock frequency. I don't think that winiboot should be expected to fix
that! You might consider hacking the reboot code in keyboard.c to reset the
clock frequency.

I hope this helps a bit.

Earl
-- 
Earl Chew, Dept of Computer Science, Monash University, Australia 3168
EMAIL: cechew@bruce.cs.monash.edu.au PHONE: 03 5655447 FAX: 03 5655146
----------------------------------------------------------------------