following up a message from Roy J. Tellason to JIM HOLSONBACK:

 RJT> JIM HOLSONBACK wrote in a message to ROY J. TELLASON:

 JH> Hello, Roy.

 JH> Awhile back, we were talking here about your problem HDD.  I've
 JH> been meaning to write you about this, but waited long, and now I
 JH> can't find the old msgs.  Part of my delay was that I wanted to do
 JH> some research, but that is still aways off for me due to other
 JH> obligations, and I want to go ahead and at least get the subject
 JH> out there for discussion and maybe someone else can fill us in.

 JH> OK, from memory - - you had reported autodetected CHS for your
 JH> 6.4GB or so HDD.  I looked at those, compared to the CHS reported
 JH> here on this 1.6GB  HDD

 RJT> CHS?  I seem to recall there being three choices for mode in
 RJT> there,  "NORMAL",  "LARGE",  and
"LBA".  I chose LBA in all cases,
 RJT> on pretty much everything here.

That turns out to be the case.

 JH> My 1.6GB autodetected at 787/64/63 , and multiplying those by 512
 JH> bytes per sector yields  1.62 GB +/-.  BIR you reported yours had
 JH> autodetected at about the same, except for one of the parameters
 JH> about double mine (probably heads), and at 512 bytes per sector
 JH> that would point to a HDD of about 3.2GB. That is why I had asked
 JH> you whether your HPFS partition may have used 1024 bytes per sector
 JH> for its formatting.

 RJT> I don't recall now what it was,  and would have to reboot the
 RJT> machine to find out...

 JH> So anyway, I wonder if you have that 6+GB drive attached to a
 JH> mainboard which can properly recognize it.

 RJT> Yeah,  I believe so.

Yep.

 JH> In larger sizes, I think the autodetect chs x 512 falls apart, but
 JH> no time right now to look up and cite the reasons I'm saying that. 

 JH> Info contributions eagerly sought.

 RJT> Let me save this,  reboot,  and I'll look.

Ok,  here's what I saw:

                  Size  Cyls  Heads Secs

Primary Master    6473  787   255   63
Secondary Master  1263  612   64    63
Primary Slave     3249  787   128   63

If that third one looks familiar,  it's the one that was in the pizza box.  

So,  no question that the system recognizes the drive.  Multiplying 787
times 255 times 63 times 512 bytes/sector gets me 6,473,295,360.  Close
enough to 6.4G for me...

The problem,  as I see it at this point in time,  isn't how the system is
seeing the drive.  It's the partition table.  Which is *maybe* caused by
that board's interpretation of the drive geometry,  perhaps.  I wish I'd
written down the prior setup before that MB wasn't functional any more...

Something I read recently mentioned a "signature" of two bytes
consisting of hex 55AA,  or maybe that was the other way around?  Anyhow, 
I happened to notice this floppy sitting here on my desk that had Notron
Diskedit on it,  so I did a bit of poking around with it and found that in
a number of places.  And on looking in each sector that contained that
signature,  I found "OS2" in a couple of places,  in one place
also found "HPFS",  "OS2KRNL",  "OS2LDR", 
and (apparently a label) "OS2 Boot".  That sure as heck seems to
me to be the boot sector for the extended partition.  It also appears to be
on a cylinder boundary,  I think,  if I can figure out how to translate
what they're calling out in that program to an actual physical location on
the drive.  That one was at "Cyl 11",  "Side 16", 
"Sec 1",  "Offset 176,778".  That last one might be
particularly significant -- the first thing on the drive is the OS/2 Boot
Manager,  which is still working just fine and gave me a choice of dos or
os/2 when I rebooted just now.  That typically takes one cylinder.  Then
the next partition is one primary,  my dos boot partition,  which is around
80M or so.  Let's see,  176,778 sectors times 512 bytes is 90,510,336
bytes,  or divided by 1024 is 88,389 MB.  That sounds about right to me...

Now if I can just figure out how to interpret exactly where the system
*thinks* that the partitions are set,  maybe I'll get somewhere.  One of
these days I expect I'll boot with something linuxish,  and see what _its_
fdisk has to say about it.  That program doesn't insist on cylinder
boundaries,  etc. and will let you get seriously weird if you want,  though
my intent is to just see what's happening there.

---