update: UTek (V) filesystems and boot blocks

From: Jules Richardson <julesrichardsonuk_at_yahoo.co.uk>
Date: Tue Jan 25 07:21:08 2005

This Tek's almost out of disk space, so I got to working out how to add
a second drive to it... I found an 'hdsetup' program on the disk, and
also a definition file for the partition table of a 300MB drive (whether
that's the same drive as the failing one that came with the machine, I
don't know).

Seems like a drive *always* has 9 partitions, although some may be left
empty in the raw table on the drive and padded with zeros. Partition
entry 0 is a root partition, entry 1 is swap, entries 2-5 are 'user'
partitions, 6 corresponds to 'whole disk', and 7-8 I'm guessing are
something to do with bootstrap (3MB and 2.5MB, possibly a data partition
containing a mini kernel and filesystem drivers to boot the main kernel,
and a swap partition??)

hdsetup won't let me create a root partition, or types 7 & 8 (fair
enough given that it's a post-install program)

Looking at the raw disk image I've saved, I've worked out where the
partition table is within the image (it's at offset 0x280). Most of the
stuff preceding it is zeros, except for a few flags and one place that
seems to be total size of the disk in blocks *including partition table

I'm now wondering if I can clone the OS onto a bigger drive without
install media :-) If I copy partition table area and the 'boot'
partitions (7 & 8 - which happen to be at the start of the disk) then it
might work. I'll have to manually edit the table (and disk size field)
on the new drive, leaving the boot partitions as they are. Then boot the
Tek with the new disk as an auxiliary, format the root partition on the
new disk, and copy everything across via dump/restore to the new drive.

*if* partitions 7 & 8 really are to do with bootstrap, and they contain
filesystem drivers to find and boot the main kernel, and I leave them in
the same place on the new drive as the old, then in theory it could
work... (sorry, I'm getting quite into hacking here ;)

I've hosed two 1.2GB drives already using Tek's hdsetup prog
though :-( I have a feeling it doesn't cope with drives having more
than 1024 cylinders for some reason - when it formats the drive (which
it says it needs to do before it'll let me partition it) it all works
fine on the Tek until the next boot, at which point the drive vanishes.
Putting it into a PC at that point shows it to be broken too, it doesn't
even respond to inquiry at initialise time. Grrr! The Tek must have
trashed the drive firmware somehow. At least with the previous OS
cloning theory I don't need to use hdsetup at all.

On the 'mounting under Linux' front, I've had no luck so far. Seems that
ufs is the fstype I want under Linux, which in turn covers about 8
different ffs types. Unfortunately none of them seem to be compatible
with Tek's idea of ffs... (it's possible I'm pulling the wrong blocks
out of the raw image though - maybe the partition table's relative to
cylinder 1 rather than 0 say, to compensate for boot area... I'll have
to try that)

Phew. I'm having fun anyway - I miss this kind of messing around with
old hardware :-)


Received on Tue Jan 25 2005 - 07:21:08 GMT

This archive was generated by hypermail 2.3.0 : Fri Oct 10 2014 - 23:37:45 BST