V/AT 2.4 w/DTC 7287 controller (long)

[Bill Kennedy]

In article <589 at ssbn.WLK.COM>, I wrote:
> I need some help getting V/AT 2.4 installed.  The installation
> instructions say to use the DTC BIOS to initialize the RLL drive
> and I did that.  The instructions are silent about any mapping
> options so I took the defaults.

For lack of reasonable install instructions these should suffice.
The system is a 12MHz no name clone with the subject DTC RLL controller,
Microscience HH-1120 half height 122Mb drive, 4Mb RAM on the motherboard,
AMI BIOS.  The controller, drive, and BIOS are new, $169, $725, and $20
respectively from a very reliable dealer in Santa Clara, CA.  I mention
that so that anyone interested can contact them.  Storage System
Engineering Service (408) 727-6040, ask for Eric.  I'm just a satisfied
customer, no other affiliation.

> In particular I can not get the fdisk to work.  Regardless of
> whether or not I have a DOS partition or how many UNIX partitions
> I set up, fdisk always claims that the partition table is wrong and
> clears it.  DOS fdisk seems to be able to use the one set up by
> Microport but Microport can't.

This was because of a shortcoming in the installation documentation.
I tried the Microport suggestion and achieved the above result, followed
my nose, ditto, got some suggestions from my earlier posting, THANKS!,
ditto.  Some of the install instructions are correct, some are not.

You must, indeed, use the DTC supplied drive definition program.  Just
like the doc says, select the non-standard drive and set it up for
1314 cylinders, 7 heads, no write precomp, 26 sectors per track, but
*DO* select mapping and use mapping option 2.  That will let the
controller spoof fdisk into thinking that the drive has 467 cylinders,
16 heads, and 32 sectors per track.  Then, make the selection permanent,
reboot, and go back into c800:5 to do the low level format and verify.
You need not enter bad track data unless you want to, that will make
more sense later.

Now you can boot V/AT from the first floppy and fdisk -s.  I set up a
DOS partition from cylinder 1-128 and UNIX from 129 to 466.  Make the
UNIX partition active, perform the bad track scan and make a note of
the cylinders and tracks reported.  In my case there were only two bad
spots on the drive, the bad track scan found one of them, I don't know
what became of the other, I fear I will find out when I least want to.
If you write down what the bad track scan finds you can save some time
later by typing them in.  They will not be the same as the list that
comes with the drive because of controller mapping.

After the fdisk I did an installit and when divvy came up I took the
number of blocks in /usr > 130,000 and moved them to /root (documented
problem with big file systems).  Proceed with the installation and
pay attention to the sequence called out on page 1-3 of the instructions.
My system would not boot after disk 1 was loaded, it said there was no
boot on the hard disk.  I booted from the floppy and did fdisk -s again
and it worked.

I wanted a max size DOS partition because often the machine runs native
DOS.  I could not find any way to coax DOS into formatting the partition
on cylinders 1-128 even though fdisk knew it was there.  I deleted the
DOS partition, created a new one with -> DOS <- fdisk, it will start on
cylinder 0.  Now you can format it and boot from it.  Use DOS fdisk again
to change the active partition from DOS to Non-DOS and you're done.

When you want to boot DOS just type dos in response to the boot prompt,
or wait for the floppy light to go out and press enter to boot V/AT.

Some notes regarding Microport's instructions.  They suggest using the
spare sector option.  What that does is make it so that there are 25
"live" sectors on the track and one spare.  When the verify takes place
and a bad sector is found the track is reformatted with the spare sector
now allocated and the bad sector as the spare.  The Microport doc (at
least the way I read it).  Suggests that the controller will correct
errors on the fly and use the spare sector to do it.  That controller is
pretty smart, but I don't think it's that smart.  If you use the spare
sector option you will lose one sector on each track regardless of
whether or not there's a bad spot.  That was not acceptable to me.

The doc is silent about mapping options.  Each one I tried resulted in
fdisk trashing the partition table and divvy failing with an "invalid
end partition record unit 0".  The first mapping option makes the drive
look like 917 cylinders, 15 heads, 17 sectors per track.  That's cute
but it won't work.  The second says that it's for drives with more than
1024 cylinders but <= 250Mb, that's the one that worked for me.  The third
is for > 1024 cylinders, > 250Mb.  I didn't try that.

Letting DOS start on cylinder 0 was the other zinger.  The V/AT fdisk will
not let you do that but the one for DOS will.  Oh, I was using 3.30a, I
guess it just insists on being on cylinder 0.

Sorry for the length, but I (showing my ignorance) spent nearly ten hours
stomping through minefields before I found the correct combination.  If
I can save someone some hours, it's worth it.  Many many thanks to those
who sent me email suggestions.  It kept me going on it.
-- 
Bill Kennedy  usenet      {killer,att,cs.utexas.edu,sun!daver}!ssbn!bill
              internet    bill at ssbn.WLK.COM