Hello Rich,

09 Oct 99 09:43, Rich Wonneberger wrote to George White:

RW> If I had 1 Primary partition, and the rest logical partitions in an
RW> extended partition, I couldnt have more then 4, 2 GIG partitions??

It's quite simple.

Short answer:

An extended partition counts as a primary partition. 4 primary partitions on
one harddisk is the maximum, with extended partitions counting as primary
partitions because an extended partition takes the place of one primary
partition in the primary partition table.

An extended partition can "contain" one or more logical partitions.

So with one primary partition and the rest as logical partitions in one
'extended' partition, how many primary partition place holders are used and
how many are still free?


Long (programmers) answer:

All partition tables have room for only 4 pointers. Those pointers are all
zero's if not used, and if the pointer is valid and in use that pointer point
to either a partition or a next partition table.

The primary partition table is in the first physical sector on the harddisk
and therefore there can be only one primary partition table on one harddisk
(each harddisk has its own primary partition table).

The primary partition table can contain one or more pointers, each pointing to 
one primary partition OR one secondary partition table. The secondary
partition table can have one or more pointers, each pointing to a logical
partition or a next partition table.

There are two possibilities: each secondary (or next) partition table is
filled completely with pointers (each pointing to either a logical partition
or a next partition table), or each secondary (or next) partition table has
only two pointers: one pointing to a logical partition and the other pointing
to a next partition table.

The minimum size for each partition is one cylinder on the harddisk,
for each partition table, one cylinder on the harddisk is needed,
(for both whatever the size of one cylinder).

So most efficient use of harddisk space with logical partitions defined is to
use three pointers in each partition table to point to a partition, and the
4th pointer to point to a next partition table so that logical partitions can
always be created if needed and if there are harddisks cylinders free.

But the 'standard' way is to define one logical/next partition table for each
logical partition, one pointer pointing to that logical partition and the
other pointer pointing to a next partition table if there are more logical
partitions.

Logic dictates (=I think) that there is nothing to prevent you to define more
than one 'extended' partition (=pointer to a next/logical partition table) in
the same primary partition table, but there is no advantage doing so compared
to one 'extended' partition in one primary partition table and the
disadvantage is you can define one primary partition less for each 'extended'
partition, IMHO.

Why there is only room for 4 pointers in each partition table, IMHO, because
that had to fit in 256 bytes: half of the size of one sector, the second half
of the first physical sector on the harddisk, because the first half of that
sector is used as bootcode.

Now, this is purely my humble opinion, this whole construction of linked
partition tables with each partition table with room for only 4 pointers is a
kludge, but the IBM compatible pc was (certainly not in the beginning of those 
pc's) not designed to use harddisks, and I guess this is still a legacy.

I hope you see now the reason why one harddisk is limited to 4 primary
partitions.

  Greetings   -=Eddy=-        email: eddy.thilleman@net.hcc.nl

... Win3.0 = UAEs, Win3.1 = GPFs, Win95 = graceful GPFs
--- GoldED/2 3.0.1