ZZ>
  >  JdeBP>   2^32 sectors, or 2 Terabytes.  This is the theoretical
JP>limit
  >  JdeBP>   of an HPFS partition.
  >
  > I'm fairly sure there are other limitations.
ZZ>

JP>  Everyone says that.  Everyone says that there is supposed to be
JP>  another limitation, but this is always an "I heard it from a friend
JP>  who knows someone who works for IBM" sort of rumour and I have never
JP>  been able to pin it down to an actual figure or a given cause.

JP>  One source gave it as 2^27 sectors (64Gb) the same as you did and
JP>  explained it as IBM not using the top 5 bits of the sector address.
JP>  I'd believe it if it wasn't for the odd number of bits.  2^24 or 2^28
JP>  would be believable.

I guess that is because Doug Azzarito himself says it. All the BOCA
people on IBMLink say that for HPFS, Doug is the lead.  Here is the
reason that I was given, as I finally found my notes. I was not stalling
you over in OS2DOS, just had to find them on tape as it was early last
year. Here is the note from my electronic journal of the event.

----------
Background:  This discussion was in specific relation to the question;
"does the HPFS cache use a fixed allocation per drive". posed to Doug
Azzarito IBM HPFS troubleshooter and resident Phoo bah

Answer:  It does not and here was his explanation.

1. HPFS uses a 32-bit logical sector number.
2. 5 bits are reserved for drive letters allowing the HPFS driver to
   support 32 HPFS drives.
3. The remaining 27 bits are used for sector numbers.  The logical
   sector number is used to look up data in the cache.

Speculation:  So it appears the 64GB figure is a limitation of the
caching scheme and really has nothing to do with addressability which
should allow 2199GB with 512 byte sectors.  Throw away the current
caching scheme and the limit is gone (?)  Maybe throw away OS2 drive
letters and stick with physical drive (?).  Seems that we have an
apparent DOS throwback of drives A..Z rounded up to the next binary
value = 32 drives leaving whatever was left for sector addressing.  Just
looks like left over "640K is enough for anybody" mentality when Gordon
(?)  and crew were creating this thing.  "64GB is enough for anybody"
IBM should get a numerologist to tell them that 6 & 4 are unlucky
numbers for them.

*EON*
------------

He's the guy with the code so I have been on the lookout for him to
re-verify that this is still the real deal.  Most people are more
concerned with the current gamma's right now but I have not forgot your
quest.  Considering the resistance to upping the cache size; I think
that we will see a new file system or a scratch rewrite of HPFS before
you find anyone to beat up on.  Rumor has it that a rewrite is occuring.
I have not seen Doug or Mary post on OS2BBS since the new year so they
and the rest of their group must be up to something.  Doug always pops
in unless he is totally swamped, likewise for Mary.

--Lynn

 * SLMR 2.1a * Hindsight is an exact science.

--- DB 1.39/004485