partition count limitation under CE.NET

Hi,

I have done extensive research on the FAT16 file system. I have found no
evidence that there exists a limit (other than physical disk capacity) for
the number of partitions that FAT16 supports. That is, from a strict FAT16
design point of view, once the primary partition is created and defined, any
number of extended partitions, up to the maximum partition size limitation
of roughly 2 GB, can be created and accessed.

Granted, this assessment relies on the fact that there are typically several
places that C/H/S (cylinder/head/sector) get converted to LBA and
vice-versa. This depends on the ATA drive, BIOS, and OS.

Fom a CE.NET point of view, the implementation of the FSD/FSDM/PartManager
could limit the number of partitions based on some arbitrary value.

Although I did not find any such limitation while digging through the
FSD/FSDM/PartManager code, I could have missed it. Does anyone know of a
limitation to the number of partitions that the CE.NET file system subsystem
imposes? If so, what is it?

I have a few that I can think of (again no proof).

1. Since all file systems are mounted under the root, there could be a
limitation due to the maximum number of folders allowed off of the root.
This could be imposed by the storage manager or the FSDM.

2. Assuming that the Partition Manger enumerates all the links in the FAT
chain, my guess is that for CE.NET, there could be an upper limit somewhere
between 62 and 68 because of the so-called 137 MB limitation imposed by most
modern ATA devices. 137 MB is a result of using 28 bits for sector
addressing (note: ATA-6 defines 48 bits). So, depending on how big you make
the partition (somewhere between 2 and 2.2 GB) you'll end up with a number
between 62 and 68.

3. Hardware BIOS limitations. Certain BIOS implementations could not be
providing extended int 13 disk services. Int 13 is x86 based. I don't know
how other processors (specifically ARM) provide BIOS disk support. Any
additional information on this would be helpful.

Thanks,
Nick.

Jeff
Thu Sep 25 13:29:26 CDT 2003

Last time I checked, CE's FAT16 never made use of CHS addressing because LBA
is required by the ATA spec.

In any case, partitioning is orthogonal to choice of filesystem; it is
possible to create partitions much larger than FAT16 can reasonably address.
In fact, without using EFI (which CE wasn't using and probably still isn't)
there is a cap of 2TB on total disk size (32-bit LBA * 512B sectors) and
that can be split up into as many or as few partitions as you like. CE 3.0
(and to my knowledge, CE 4 also) ignores CHS portions of the partition
tables.

- Jeff.

"Ten" <nick.bartolotti@tentechnologies.com> wrote in message
news:n1ncb.1331$oa4.294@twister.nyroc.rr.com...
> Hi,
>
> I have done extensive research on the FAT16 file system. I have found no
> evidence that there exists a limit (other than physical disk capacity) for
> the number of partitions that FAT16 supports. That is, from a strict
FAT16
> design point of view, once the primary partition is created and defined,
any
> number of extended partitions, up to the maximum partition size limitation
> of roughly 2 GB, can be created and accessed.
>
> Granted, this assessment relies on the fact that there are typically
several
> places that C/H/S (cylinder/head/sector) get converted to LBA and
> vice-versa. This depends on the ATA drive, BIOS, and OS.
>
> Fom a CE.NET point of view, the implementation of the FSD/FSDM/PartManager
> could limit the number of partitions based on some arbitrary value.
>
> Although I did not find any such limitation while digging through the
> FSD/FSDM/PartManager code, I could have missed it. Does anyone know of a
> limitation to the number of partitions that the CE.NET file system
subsystem
> imposes? If so, what is it?
>
> I have a few that I can think of (again no proof).
>
> 1. Since all file systems are mounted under the root, there could be a
> limitation due to the maximum number of folders allowed off of the root.
> This could be imposed by the storage manager or the FSDM.
>
> 2. Assuming that the Partition Manger enumerates all the links in the FAT
> chain, my guess is that for CE.NET, there could be an upper limit
somewhere
> between 62 and 68 because of the so-called 137 MB limitation imposed by
most
> modern ATA devices. 137 MB is a result of using 28 bits for sector
> addressing (note: ATA-6 defines 48 bits). So, depending on how big you
make
> the partition (somewhere between 2 and 2.2 GB) you'll end up with a number
> between 62 and 68.
>
> 3. Hardware BIOS limitations. Certain BIOS implementations could not be
> providing extended int 13 disk services. Int 13 is x86 based. I don't
know
> how other processors (specifically ARM) provide BIOS disk support. Any
> additional information on this would be helpful.
>
> Thanks,
> Nick.
>
>

Ten
Fri Sep 26 07:12:47 CDT 2003

Jeff,

Thanks for the response.

A few thoughts arise from your comment.

I searched again and found in the documentation that CE uses LBA mode unless
particular settings in the registry force CHSMode. In the case where these
settings are not defined or 0, the driver uses the mode as specified by the
device in response to the IDENTIFY command. So, since most ATA drives use
LBA mode now, CE will address the drive in LBA mode. Amazing how a small
change in the search parameters can affect the results.

What is EFI? I have not heard that term before.

The 2 TB cap must utilize ATA-6 as the underlying technology for the storage
device. ATA-5 has 28 bits for sector addresses and ATA-6 increases to 48
bits.

Thanks,
Nick.






"Jeff Rosenfeld" <spamophobe@comcast.net> wrote in message
news:WsOdnQkwePOaqe6iU-KYuA@comcast.com...
> Last time I checked, CE's FAT16 never made use of CHS addressing because
LBA
> is required by the ATA spec.
>
> In any case, partitioning is orthogonal to choice of filesystem; it is
> possible to create partitions much larger than FAT16 can reasonably
address.
> In fact, without using EFI (which CE wasn't using and probably still
isn't)
> there is a cap of 2TB on total disk size (32-bit LBA * 512B sectors) and
> that can be split up into as many or as few partitions as you like. CE 3.0
> (and to my knowledge, CE 4 also) ignores CHS portions of the partition
> tables.
>
> - Jeff.
>
> "Ten" <nick.bartolotti@tentechnologies.com> wrote in message
> news:n1ncb.1331$oa4.294@twister.nyroc.rr.com...
> > Hi,
> >
> > I have done extensive research on the FAT16 file system. I have found
no
> > evidence that there exists a limit (other than physical disk capacity)
for
> > the number of partitions that FAT16 supports. That is, from a strict
> FAT16
> > design point of view, once the primary partition is created and defined,
> any
> > number of extended partitions, up to the maximum partition size
limitation
> > of roughly 2 GB, can be created and accessed.
> >
> > Granted, this assessment relies on the fact that there are typically
> several
> > places that C/H/S (cylinder/head/sector) get converted to LBA and
> > vice-versa. This depends on the ATA drive, BIOS, and OS.
> >
> > Fom a CE.NET point of view, the implementation of the
FSD/FSDM/PartManager
> > could limit the number of partitions based on some arbitrary value.
> >
> > Although I did not find any such limitation while digging through the
> > FSD/FSDM/PartManager code, I could have missed it. Does anyone know of
a
> > limitation to the number of partitions that the CE.NET file system
> subsystem
> > imposes? If so, what is it?
> >
> > I have a few that I can think of (again no proof).
> >
> > 1. Since all file systems are mounted under the root, there could be a
> > limitation due to the maximum number of folders allowed off of the root.
> > This could be imposed by the storage manager or the FSDM.
> >
> > 2. Assuming that the Partition Manger enumerates all the links in the
FAT
> > chain, my guess is that for CE.NET, there could be an upper limit
> somewhere
> > between 62 and 68 because of the so-called 137 MB limitation imposed by
> most
> > modern ATA devices. 137 MB is a result of using 28 bits for sector
> > addressing (note: ATA-6 defines 48 bits). So, depending on how big you
> make
> > the partition (somewhere between 2 and 2.2 GB) you'll end up with a
number
> > between 62 and 68.
> >
> > 3. Hardware BIOS limitations. Certain BIOS implementations could not be
> > providing extended int 13 disk services. Int 13 is x86 based. I don't
> know
> > how other processors (specifically ARM) provide BIOS disk support. Any
> > additional information on this would be helpful.
> >
> > Thanks,
> > Nick.
> >
> >
>
>

Jeff
Fri Sep 26 17:42:59 CDT 2003

EFI was some new partitioning format defined for use with really big disks a
few years ago. Maybe I am misremembering the acronym.

You're right about the 2TB; I was thinking of the specifiability in the
partition table rather than the disk's inherent addressability.

- Jeff.

"Ten" <nick.bartolotti@tentechnologies.com> wrote in message
news:33Wcb.9638$oa4.292@twister.nyroc.rr.com...
> Jeff,
>
> Thanks for the response.
>
> A few thoughts arise from your comment.
>
> I searched again and found in the documentation that CE uses LBA mode
unless
> particular settings in the registry force CHSMode. In the case where
these
> settings are not defined or 0, the driver uses the mode as specified by
the
> device in response to the IDENTIFY command. So, since most ATA drives use
> LBA mode now, CE will address the drive in LBA mode. Amazing how a small
> change in the search parameters can affect the results.
>
> What is EFI? I have not heard that term before.
>
> The 2 TB cap must utilize ATA-6 as the underlying technology for the
storage
> device. ATA-5 has 28 bits for sector addresses and ATA-6 increases to 48
> bits.
>
> Thanks,
> Nick.
>
>
>
>
>
>
> "Jeff Rosenfeld" <spamophobe@comcast.net> wrote in message
> news:WsOdnQkwePOaqe6iU-KYuA@comcast.com...
> > Last time I checked, CE's FAT16 never made use of CHS addressing because
> LBA
> > is required by the ATA spec.
> >
> > In any case, partitioning is orthogonal to choice of filesystem; it is
> > possible to create partitions much larger than FAT16 can reasonably
> address.
> > In fact, without using EFI (which CE wasn't using and probably still
> isn't)
> > there is a cap of 2TB on total disk size (32-bit LBA * 512B sectors) and
> > that can be split up into as many or as few partitions as you like. CE
3.0
> > (and to my knowledge, CE 4 also) ignores CHS portions of the partition
> > tables.
> >
> > - Jeff.
> >
> > "Ten" <nick.bartolotti@tentechnologies.com> wrote in message
> > news:n1ncb.1331$oa4.294@twister.nyroc.rr.com...
> > > Hi,
> > >
> > > I have done extensive research on the FAT16 file system. I have found
> no
> > > evidence that there exists a limit (other than physical disk capacity)
> for
> > > the number of partitions that FAT16 supports. That is, from a strict
> > FAT16
> > > design point of view, once the primary partition is created and
defined,
> > any
> > > number of extended partitions, up to the maximum partition size
> limitation
> > > of roughly 2 GB, can be created and accessed.
> > >
> > > Granted, this assessment relies on the fact that there are typically
> > several
> > > places that C/H/S (cylinder/head/sector) get converted to LBA and
> > > vice-versa. This depends on the ATA drive, BIOS, and OS.
> > >
> > > Fom a CE.NET point of view, the implementation of the
> FSD/FSDM/PartManager
> > > could limit the number of partitions based on some arbitrary value.
> > >
> > > Although I did not find any such limitation while digging through the
> > > FSD/FSDM/PartManager code, I could have missed it. Does anyone know
of
> a
> > > limitation to the number of partitions that the CE.NET file system
> > subsystem
> > > imposes? If so, what is it?
> > >
> > > I have a few that I can think of (again no proof).
> > >
> > > 1. Since all file systems are mounted under the root, there could be a
> > > limitation due to the maximum number of folders allowed off of the
root.
> > > This could be imposed by the storage manager or the FSDM.
> > >
> > > 2. Assuming that the Partition Manger enumerates all the links in the
> FAT
> > > chain, my guess is that for CE.NET, there could be an upper limit
> > somewhere
> > > between 62 and 68 because of the so-called 137 MB limitation imposed
by
> > most
> > > modern ATA devices. 137 MB is a result of using 28 bits for sector
> > > addressing (note: ATA-6 defines 48 bits). So, depending on how big
you
> > make
> > > the partition (somewhere between 2 and 2.2 GB) you'll end up with a
> number
> > > between 62 and 68.
> > >
> > > 3. Hardware BIOS limitations. Certain BIOS implementations could not
be
> > > providing extended int 13 disk services. Int 13 is x86 based. I
don't
> > know
> > > how other processors (specifically ARM) provide BIOS disk support.
Any
> > > additional information on this would be helpful.
> > >
> > > Thanks,
> > > Nick.
> > >
> > >
> >
> >
>
>

Steve
Sun Sep 28 23:51:19 CDT 2003

The maximum number of partitions is fixed in the MBR and has nothing to do
with FAT16/32 or any other file system. It's possible you could create your
own partition manager that used some other method of allocating partitions
then the normal MBR that could have an arbitrary number of partitions.

--
Steve Maillet (eMVP)
Entelechy Consulting
smaillet_AT_EntelechyConsulting_DOT_com