Path: utzoo!utgpu!news-server.csri.toronto.edu!rpi!zaphod.mps.ohio-state.edu!wuarchive!uunet!mcsun!ukc!cam-cl!news
From: maj@cl.cam.ac.uk (Martyn Johnson)
Newsgroups: comp.unix.ultrix
Subject: Re:  CD-rom > TK50?
Summary: Almost possible...
Keywords: TK50 CDROM Ultrix
Message-ID: <1991Jun5.111537.1885@cl.cam.ac.uk>
Date: 5 Jun 91 11:15:37 GMT
References: <91155.133326JIW2@psuvm.psu.edu>
Sender: news@cl.cam.ac.uk (The news facility)
Organization: U of Cambridge Computer Lab, UK
Lines: 59

>  Here's a problem I've been working on for quite a while, with no
> success:  Does anyone know how to produce setld-installable TK50's
> from the files on a CD-rom?

I have put a certain amount of thought into this problem, though since I 
do not have any Ultrix CD-ROMs I have not been able to try anything out.

>  The files on the CD-rom are, as far as I am able to tell,
> identical to what one gets on disk by doing setld -x /dev/rmt0h
> against a setld-compatible tape.

This means that it should be possible to use the documented "gentapes" 
utility to generate a tape.  For a layered product, all of the useful 
information on the tape exists in the disc directory, and it is merely a 
matter of putting it back to tape in the right format.  Note that the 
order of the files on the tape is important, and is implied by the MTLOC 
records in the control files. Check out section 6.3 of the "Guide to 
Preparing Software for Disribution on ULTRIX Systems" for a script to 
read these and make a suitable /etc/kitcap.

As I say, all this should be fine for a layered product tape - by which I 
mean a non-bootable TK50 to be read by setld. The problem arises with the 
bootable tape.  The bootstrap files on the front of the tape are not read 
in by "setld -x" (you've already booted, haven't you) and they are 
presumably not on the CDROM (you booted from the CDROM, didn't you?).

What are these magic files?  Well, there are three:

The use of the first two seem to be architecture dependent. On the VAX, 
file 1 is a bootstrap and file 2 is a standalone kernel.  On RISC, file 1 
appears to be a bootstrap and kernel combined, and file 2 is a dummy.

For both architectures, file 3 is a dump (i.e. /etc/dump format) of the 
initial root filesystem for use by the installation script (this gets 
copied to file ROOT by /etc/setld -x, so this should be on the CDROM).

I think, therefore, that the problem of making a bootable tape reduces to 
that of getting the bootstrap files (2 for VAX, 1 for RISC).  Whether 
they are lurking on the CDROM somewhere is anybody's guess. The VAX 
kernel is probably there (because the disc bootstrap probably needs it 
too), but I doubt that the others are present.

It strikes me that it would be a very modest amount of work for Digital 
to provide the necessary files and a suitable /etc/kitcap file, so that 
the TK50 kit can be made with gentapes. Perhaps it is already there and 
just needs documenting.  Could somebody in Digital Ultrix engineering 
comment? It is not the sort of thing I would expect a CSC to be able to 
help with.

This is IMPORTANT.  We are being encouraged to switch to CDROM, but it 
would be absurd to buy a CDROM drive for every machine. Having an extra 
MDDS contract just to get a bootstrap file seems equally ludicrous - 
there are better things to spend money on. Digital have stated that
the right to use software is bought by paying for the licence, not the
media.  Therefore it seems sensible to allow customers to convert kits
from any media type to any other.

Martyn Johnson      maj@cl.cam.ac.uk
University of Cambridge Computer Lab