8 GB barrier in XP and W2k for 20 GB Seagate HDD ST320420A

[Folkert Rienstra]

I now see that the disk does not support the "Host Protected Area
feature set", meaning that the disk size cannot be set using software.
My thought was that doing that could have had a chance to correct the
"Number of logical cylinders" value,

After the SET MAX ADDRESS or SET MAX ADDRESS EXT
command has been issued, the device shall report only the reduced
user address space in response to an IDENTIFY DEVICE command
Or i.o.w., only the LBA translation related words are affected.

which should be 16383.

It could be possible that the disk supports the Set Max Address command
anyway, but my Setsize tool cannot attempt to set the size if the disk re-
ports it is not supported.

Oh!?
The message " New size too small. Invalid new size" appears to say different.
Or maybe there was some truth in my initial comment that said:
"Sounds like bad command line interpretation/error reporting".

Also according to the Seagate documentation for the disk in vailpmb.pdf
(marked revision B) the Number of logical cylinders value should be 16383.

That is what it should be on all IDE drives larger than or equal to 8 GB.

Seagate says on the site that disk firmware for ATA disks is not available.

The only chance I can see to get the disk used at full size in Windows
2000/XP would be to find a driver version, maybe from the chip
manufacturer, which recognizes the disk correctly.

What makes you think that it is the driver?

 

[Folkert Rienstra]

'default'

current geometry (translation)

This is my theory too. Yes to point 3.

Which is of very limited significance unless you can confirm that 1) is true also.

Possible, or Sun just got the very first versions of the disk.


It is the motherboard chip set manufacturers which typically have the drivers.

Mostly when a new driver is needed and Microsoft isn't supplying it or when
a new chipset is introduced after the OS was released and no driver will be
available otherwise.

 

[Svend Olaf Mikkelsen]

After the SET MAX ADDRESS or SET MAX ADDRESS EXT
command has been issued, the device shall report only the reduced
user address space in response to an IDENTIFY DEVICE command

Or i.o.w., only the LBA translation related words are affected.

Nonsense objection.

Oh!?
The message " New size too small. Invalid new size" appears to say different.
Or maybe there was some truth in my initial comment that said:
"Sounds like bad command line interpretation/error reporting".

This is not the issue here.

That is what it should be on all IDE drives larger than or equal to 8 GB.

That was said already.

What makes you think that it is the driver?

Please get another attitude. Now. Think yourself.

 

[Folkert Rienstra]

Nonsense objection.

Incomprehensible answer, as usual, combined with creative snipping when
you are found out not to be familiar with the IDE spec, when you should be.

This is not the issue here.

The issue here is whether you are thruthfull or putting up a front. You were
caught on a blatant lie. If you lie so easily, what else can we not trust you on?

That was said already.

Yes, it was, wasn't it. Guess who said it, it wasn't you. You came up with that
only after I had mentioned it and your other attempts came up with nothing.

Please get another attitude.

Get another attitude yourself. And you better be a bit more thruthful.
It's obvious from other posts in this thread that you haven't got a
clue but aren't prepared to say so. You are just grasping at straws.
You asked him things to do and then didn't bother to comment and then
when asked to, still refused to do so.

Now. Think yourself.

I do think myself. Question was, do you?
Johan has named the Windows NT flavors as being the culprit, no mention
of exceptions for certain chipset drivers. Either the description is wrong
and it is actually driver dependable or it is not. So if you think driver,
you should have a good reason for it. I asked for your reasoning.

Given the evasive answer I will gather that you don't have any and this is
just you grasping at another straw.

 

[Alexander Grigoriev]

No, I don't work for MS. I've got to admit, though, that I'm an MS MVP in
DDK/SDK domain.
I work for a communication chipsets company in SoCal.

Do you mind that your and OP's time spent on this tiny piece of unusable
storage is vastly more expensive than buying a new bigger HD which will
work with no problem?

When you don't understand, kill it, right?
Btw, are you working for M$?

[snip]

 

[Svend Olaf Mikkelsen]

Incomprehensible answer, as usual, combined with creative snipping when
you are found out not to be familiar with the IDE spec, when you should be.

I did stop reading after this. You will have to stop attacking, and
help and discuss in stead.

 

[Folkert Rienstra]

No, I don't work for MS. I've got to admit, though, that I'm an MS MVP in
DDK/SDK domain.

Then I think you'd be the person able to comment on this idiotic working_in
_CHS_mode_upto_8GB thing but instead you try to revert attention away.
Yeah, I'd say that you work for M$ alright.

I work for a communication chipsets company in SoCal.

Do you mind that your and OP's time spent on this tiny piece of unusable
storage is vastly more expensive than buying a new bigger HD which will
work with no problem?

When you don't understand, kill it, right?
Btw, are you working for M$?

On 19 Jun 2004 13:11:01 -0700, (e-mail address removed) (JVP) wrote:

There are no lines with "Native sectors:" I'm afraid.

[snip]

 

[Svend Olaf Mikkelsen]

Nonsense objection.

The explanation that this is nonsense is that the Set Max Address
commands of course will set word 1.


Another issue is that the documentation says:

"2.2.3 Alternate capacity jumper
Some older computers may “hang” if their BIOS detects a hard drive
that has more than 4,092 cylinders at startup. To eliminate this
problem, the drive includes a capacity-limiting jumper that sets the
drive’s default translation geometry to 4,092 cylinders. This limits
the drive’s capacity to 2.1 Gbytes unless third-party software is
used."

I cannot see how third-party software would set the size.

 

[JVP]

Hi Svend,

It seems like my stupid disk problem still bothers you. ;-) In fact, I am also
still looking to it, too stubborn to let it rest.

Although I still believe that the standard Windows 2000/XP IDE drivers need to
see logical geometry 16383/16/63 before switching to LBA mode for going over the
8.4 GB barrier, I would like to have certainty about this and I don't have it
yet...

I checked again on the website of my motherboard (Gigabyte GA-6BXE) and they
only have drivers for Windows 95 and NT. I still have to see if I can do
something with those on Windows 2000/XP.

I also mailed Seagate support, without expecting to much ;-), and indeed, I
already got an answer back that they want the exact partnumber/serial number to
check whether the drive is standard or OEM. They do not support OEM:

From their standard reply:

STANDARD DISTRIBUTION = Drives sold through Retail Channel
OEM = Drives sold to ORIGINAL EQUIPMENT MANUFACTURERS
Example: (Dell, Gateway, IBM, Compaq, HP, Sun etc.)

When Seagate manufactures drives for OEM's, we design these drives to
their custom specifications. These design changes can cause the drive to
operate differently than standard distribution drives.

Since our OEM customers consider these design changes proprietary
information, Seagate retains no information on these drives.
OEM drives are supported by the OEM that ordered the drives.

I guess SUN will be very happy to support me for this issue...

The explanation that this is nonsense is that the Set Max Address
commands of course will set word 1.

Just to show off to Folkert that I have also read (part of) the ATA-5 manual ;-)
(Version markup: T13/1321D revision 3)

8.38.1.8 Description [of Set Max Address]
.... IDENTIFY DEVICE response words 1, 54, 57, 60, and 61 shall reflect the
maximum address set with this command."
....
After a successful SET MAX ADDRESS command using a new maximum LBA address the
content of all IDENTIFY DEVICE words shall comply with 6.2.1 in addition to the
following:
- The content of words (61:60) shall be equal to the new Maximum LBA address + 1.
If the device supports CHS addressing:
- The content of words 3, 6, 55, and 56 are unchanged.
- If the new content of words (61:60) is less than 16,514,064, then the content
of word 1 shall be equal to [(the new content of words (61:60)) ? [(the content
of word 3) * (the content of word 6)]] or 65,535, whichever is less.
- If the new content of words (61:60) is greater than or equal to 16,514,064,
then the content of word 1 shall be equal to 16,383.


From a previous post of Svend:

It could be possible that the disk supports the Set Max Address
command anyway, but my Setsize tool cannot attempt to set the size if
the disk reports it is not supported.

I read in the manual in the description of set max address:
"The contents of IDENTIFY DEVICE words and the max address shall not be changed
if a SET MAX ADDRESS command fails."

So would there be tools that do attempt to Set Max Address without checking?
Would Maxtor's set_geom.exe (see one of my earlier posts in this thread) be
worth trying?

Oops, just saw in the disk's manual that it does not support Set Max Address...

Another issue is that the documentation says:
"2.2.3 Alternate capacity jumper
Some older computers may ?hang? if their BIOS detects a hard drive
that has more than 4,092 cylinders at startup. To eliminate this
problem, the drive includes a capacity-limiting jumper that sets the
drive?s default translation geometry to 4,092 cylinders. This limits
the drive?s capacity to 2.1 Gbytes unless third-party software is
used."
I cannot see how third-party software would set the size.

I also saw this quote in the documentation, but interpreted "third-party
software" as DDO disk overlay sofware. Maybe this is something I should try:
set the AC jumper and use DDO...

Thanks for your time,

Johan.

 

[Svend Olaf Mikkelsen]

I checked again on the website of my motherboard (Gigabyte GA-6BXE) and they
only have drivers for Windows 95 and NT. I still have to see if I can do
something with those on Windows 2000/XP.

I also saw this quote in the documentation, but interpreted "third-party
software" as DDO disk overlay sofware. Maybe this is something I should try:
set the AC jumper and use DDO...

Since it is an Intel chipset it may be possible that the Intel
Application Accelerator can be used, but I do not know, and also do
not know if it would involve a risk to try.

The overlay software would have to send a command to the disk to set
the full size, or maybe the read/write commands for sectors after 2 GB
would just work, but then the software would have to have a way to
determine the full size and it could only be used with DOS.

 

[Folkert Rienstra]

Hi Svend,

It seems like my stupid disk problem still bothers you. ;-) In fact, I am also
still looking to it, too stubborn to let it rest.

Although I still believe that the standard Windows 2000/XP IDE drivers need to
see logical geometry 16383/16/63 before switching to LBA mode for going over the
8.4 GB barrier, I would like to have certainty about this and I don't have it yet...
[snip]

I guess SUN will be very happy to support me for this issue...

The explanation that this is nonsense is that the Set Max Address
commands of course will set word 1.

In ATA-5.

Just to show off to Folkert that I have also read (part of) the ATA-5 manual ;-)
(Version markup: T13/1321D revision 3)

8.38.1.8 Description [of Set Max Address]
... IDENTIFY DEVICE response words 1, 54, 57, 60, and 61 shall reflect the
maximum address set with this command."

Yes, turns out that I quoted my information from ATA-6, not ATA-5. My mistake.
The " ... or SET MAX ADDRESS EXT..." should have been the dead giveaway and
it did bother me for a second but apparently not enough to check on that deeper.
If I had, I would have noticed that all CHS related words are obsoleted in ATA-6
and thats why these words weren't mentioned. Please accept my apologies.

...
After a successful SET MAX ADDRESS command using a new maximum LBA address the
content of all IDENTIFY DEVICE words shall comply with 6.2.1 in addition to the
following:
- The content of words (61:60) shall be equal to the new Maximum LBA address + 1.
If the device supports CHS addressing:
- The content of words 3, 6, 55, and 56 are unchanged.
- If the new content of words (61:60) is less than 16,514,064, then the content
of word 1 shall be equal to [(the new content of words (61:60)) ? [(the content
of word 3) * (the content of word 6)]] or 65,535, whichever is less.
- If the new content of words (61:60) is greater than or equal to 16,514,064,
then the content of word 1 shall be equal to 16,383.


From a previous post of Svend:

It could be possible that the disk supports the Set Max Address
command anyway, but my Setsize tool cannot attempt to set the size if
the disk reports it is not supported.

It will do that fine if Svend decides that his tool will do that.
And judging from the error messages you got, it already does that.

I read in the manual in the description of set max address:
"The contents of IDENTIFY DEVICE words and the max address shall not be changed
if a SET MAX ADDRESS command fails."

So would there be tools that do attempt to Set Max Address without checking?

Svend's did, as you must have noticed.

Would Maxtor's set_geom.exe (see one of my earlier posts in this thread) be
worth trying?

Oops, just saw in the disk's manual that it does not support Set Max Address...



I also saw this quote in the documentation, but interpreted "third-party
software" as DDO disk overlay sofware. Maybe this is something I should try:
set the AC jumper and use DDO...

The DDO will have to use the SET MAX ADDRESS command. Guess what .....

 

[Alexander Grigoriev]

My impression is that MS developpers go to great length to achieve
compatibility with the drives and BIOS bugs. The goal is to provide the in
the driver the same translation as BIOS provides. If in some cases
16383/15/63 gives problems for switching to LBA with compatible layout, no
wonder that such configuration may be rejected by ATAPI.SYS as not fully
LBA-compatible.

What I suspect that 16383/15/63 geometry gets written to MBR and the picked
by BIOS and/or ATAPI.SYS. I saw in some cases that geometry desribed by MBR
gets precedence over what is reported by the drive itself.
I'd advice to clear MBR with disk editor, reboot and let BIOS autodetect the
drive.

No, I don't work for MS. I've got to admit, though, that I'm an MS MVP in
DDK/SDK domain.

Then I think you'd be the person able to comment on this idiotic working_in
_CHS_mode_upto_8GB thing but instead you try to revert attention away.
Yeah, I'd say that you work for M$ alright.

I work for a communication chipsets company in SoCal.

Do you mind that your and OP's time spent on this tiny piece of unusable
storage is vastly more expensive than buying a new bigger HD which will
work with no problem?

When you don't understand, kill it, right?
Btw, are you working for M$?

There are no lines with "Native sectors:" I'm afraid.

[snip]

 

[JVP]

I also saw this quote in the documentation, but interpreted "third-party
software" as DDO disk overlay sofware. Maybe this is something I should try:
set the AC jumper and use DDO...

I could solve my problem by using the Alternate Capacity jumper on the
drive. The drive then reports CHS 4092/16/63 --- even my strange OEM
drive does this (*) --- and this was sufficient to trick Windows XP
into LBA mode for large drives.

I can now use the full drive capacity under XP (with SP1), without
using DDO. It does not work for Windows 2000, however. The Windows
2000 drivers probably really expect CHS 16383/16/63 before recognizing
a large drive above 8.4GB.

(*) My drive is indeed an OEM drive, with special specifications that
are not conform to the ATA standard. It reports CHS 39535/16/63
instead of the standard CHS 16383/16/63 for drives above 8.4GB.
Seagate support told me that these parameters can not be changed on
their (OEM) ATA drives.

Thanks to the people taking part in this discussion (especially Svend)
for their time and helping me finding an explainable solution.

Johan.

 

[Folkert Rienstra]

I could solve my problem by using the Alternate Capacity jumper on the
drive. The drive then reports CHS 4092/16/63 --- even my strange OEM
drive does this (*) --- and this was sufficient to trick Windows XP
into LBA mode for large drives.

That is really-really strange as it should now recognize it as a 2GB size drive.
That's what that jumper is for, to reduce size.
It should do this for BIOS and OS alike.

Although it says:
"..... jumper that sets the drive’s default translation geometry to 4,092 cylinders."

which may indicate that it does this for words 1,3 and 6 only it also says:

"This limits the drive’s capacity to 2.1 Gbytes ....."

which suggests that that isn't all as capacity is decided by several more words.

Apparently WinNT flavors will look at this value trying to decide whether
it will make an error and if not so, ignore it altogether. Makes sense, right?

It would be interresting to know how the drive is seen, size-wise, by
BIOS, DOS, Svend's utilities (all size related words) and Windows.

I can now use the full drive capacity under XP (with SP1), without using DDO.

So obviously the default geometry and/or size reduction is informational and
doesn't affect the drive's internal command checking/size-error reporting.
Apparently the jumper setting is different from a Set Max Address as that
should reject addresses over the Set Max Address set limit with an IDNF error

It does not work for Windows 2000, however. The Windows 2000 drivers
probably really expect CHS 16383/16/63 before recognizing a large drive
above 8.4GB.

Or Win2k checks partition sizes against (reported) physical size whereas
XP accepts them and possibly overrides physical size with those values.
Or XP requests the native capacity and uses that instead of the override.
Problem with the latter is that this drive doesn't supposedly support that.

(*) My drive is indeed an OEM drive, with special specifications that
are not conform to the ATA standard. It reports CHS 39535/16/63
instead of the standard CHS 16383/16/63 for drives above 8.4GB.
Seagate support told me that these parameters can not be changed on
their (OEM) ATA drives.

Thanks to the people taking part in this discussion

(especially Svend)

Who thinks that CHS values over 8GB are quite OK.
I wonder whether we will now finally see new versions of his utilities
where these flaws will have been removed.

for their time and helping me finding an explainable solution.

Yup, I was only the first one to comment on that default geometry
value being out of range so I obviously had nothing to do with it.

 

[Folkert Rienstra]

My impression is that MS developpers go to great length to achieve
compatibility with the drives and BIOS bugs. The goal is to provide in
the driver the same translation as BIOS provides.

The MBR defines the CHS (L-CHS) that the bios will use towards the application.
The question then is, what translated CHS (P-CHS) the bios will use to address
the drive with. This could be drive CHS (default or current) or it could be LBA,
depending on what translation is selected in the bios setup, LBA-assist or Large.

If in some cases 16383/15/63 gives problems for switching to LBA with
compatible layout,

What 'layout', there is no 'layout' in LBA, nor CHS. Both are linear.
There is a problem only when drive and software use a different translation
between them (for which it is presumed not to be happening except for bugs).

The only issue is whether the selected translation allows the full use of the
capacity within 8GB or not. Since the drive is bigger than 8GB no incomplete
end cylinder is assumed and both Large (bit-shift translation) and LBA assist
address the full 8GB.

no wonder that such configuration may be rejected by ATAPI.SYS as not fully
LBA-compatible.

CHS NE LBA, so what 'compatible'?

Oh, and btw, why would this be fine for Win9x but not the NT flavors?

What I suspect that 16383/15/63 geometry gets written to MBR

Actually, the MBR values are in Int13 format (L-CHS). Int13 expects
1023/255/63 as a maximum. 16383/15/63 (P-CHS) certainly will *not* fit.

and the picked by BIOS and/or ATAPI.SYS.

I saw in some cases that geometry desribed by MBR
gets precedence over what is reported by the drive itself.

The drive CHS (P-CHS) is a different CHS than the BIOS CHS (L-CHS).
This has been so since the 16-bit CHS (known as 'Normal' in Bios setup)
was extended to 22-bit CHS, a lifetime ago? So, the drive does *not*
report any L-CHS unless it's smaller than 528MB or so.

I'd advice to clear MBR with disk editor, reboot and let BIOS autodetect
the drive.

Your advice stinks.

And you didn't answer the question.

And please, learn to post.

No, I don't work for MS. I've got to admit, though, that I'm an MS MVP in
DDK/SDK domain.

Then I think you'd be the person able to comment on this idiotic working_in
_CHS_mode_upto_8GB thing but instead you try to revert attention away.
Yeah, I'd say that you work for M$ alright.

I work for a communication chipsets company in SoCal.

Do you mind that your and OP's time spent on this tiny piece of unusable
storage is vastly more expensive than buying a new bigger HD which will
work with no problem?

When you don't understand, kill it, right?
Btw, are you working for M$?

There are no lines with "Native sectors:" I'm afraid.

[snip]