[jaclaz]

wsxedcrfv, on 23 February 2010 - 11:35 PM, said:

Just for the record, I'm not trying to be a pain here.

It is possible that you are failing in your goal.

wsxedcrfv, on 23 February 2010 - 11:35 PM, said:

I'm just asking questions. If I had a drive larger than 500 gb, I'd try fdisk and format myself - and post the results.

Sure.

wsxedcrfv, on 23 February 2010 - 11:35 PM, said:

Maybe nobody can answer these, but I really would like to know why Microsoft said these things:

- Fdisk update is not designed for 48-bit LBA
- Fdisk update is not supported on drives larger than 128 gb
- Fdisk update is limited to 512 gb "by design" (contradicts previous two points)
- Win-Me Setup boot disk can be used to partition drives larger than 512 gb (how?)

Can anyone here post a reality check on the last point?

Can you post WHERE Microsoft said the listed 4 points?

Here they are numbered:

• - Fdisk update is not designed for 48-bit LBA
  
• - Fdisk update is not supported on drives larger than 128 gb
  
• - Fdisk update is limited to 512 gb "by design" (contradicts previous two points)
  
• - Win-Me Setup boot disk can be used to partition drives larger than 512 gb (how?)

The MS kb references should be these:
http://support.micro...kb/139579/en-us
http://support.micro...kb/239113/en-us
http://support.micro...kb/239119/en-us
http://support.micro...kb/245213/en-us
http://support.micro...kb/263044/en-us
http://support.micro...b/263045/EN-US/
http://support.micro...kb/280737/en-us < #3?
http://support.micro...kb/327202/en-us < #1 & #2?

I am missing where #4 may come from.

To clear some points (or maybe further confusing you ) read here:
http://thestarman.pc...m/mbr/FDISK.htm



jaclaz

[rloew]

wsxedcrfv, on 23 February 2010 - 11:35 PM, said:

Just for the record, I'm not trying to be a pain here. I'm just asking questions. If I had a drive larger than 500 gb, I'd try fdisk and format myself - and post the results.

Maybe nobody can answer these, but I really would like to know why Microsoft said these things:

- Fdisk update is not designed for 48-bit LBA
- Fdisk update is not supported on drives larger than 128 gb
- Fdisk update is limited to 512 gb "by design" (contradicts previous two points)
- Win-Me Setup boot disk can be used to partition drives larger than 512 gb (how?)

Can anyone here post a reality check on the last point?

I don't work for Microsoft, so I cannot give an official answer, but I gave my interpretation of why Microsoft said items #1 thru #3 in my last post.
FDISK cannot provide support for 48-Bit LBA limitations elsewhere (BIOS and/or ESDI_506.PDR) even if it can handle larger drives, such as USB.

[dencorso]

I thank jaclaz for the comprehensive list of KBs relating to FORMAT and FDISK.
I also thank both jaclaz and RLoew for their considerations about this matter.
I'd like to add that, after rereading those KBs, and reading for the first time KB327202 (great catch jaclaz!), I think RLoew pinpointed perfectly the point of MS's statements: both fdisk and format relay on the support of the BIOS for 48-bit LBA, they don't provide it themselves. So, provided there is support by the BIOS, fdisk can create partitions up to 512 GB in size, not more, and that's "by design". And they point you to Symantec's gdisk, if you want to create larger partitions...
The Ranish Partition Manager and RLoew's RFDISK and RFORMAT are other viable alternatives for DOS and Win 9x/ME. And MS does not state which is the maximum partition size that FORMAT can format correctly, despite showing wrong numbers during the operation. But what's known at present permits the inference it's also around, if not exactly, 512 GB.

@jaclaz: BTW, #4 seems to be one of the possible readings of the second paragraph of the resolution section of KB280737, but I don't think it's what was meant to be read from it...

[jaclaz]

dencorso, on 24 February 2010 - 09:56 PM, said:

@jaclaz: BTW, #4 seems to be one of the possible readings of the second paragraph of the resolution section of KB280737, but I don't think it's what was meant to be read from it...

Sure, but I don't see the connection with Me (in the sense of ME only).

Sometimes (read, as you prefer as "often" or "always") MS KB/docs are written in a "cryptic" way, without providing the "real information" or doing so in such a way that each single word need to be weighted, or without a clear explanation, as an example:
http://neosmart.net/...or-not-the-mbr/
http://technet.micro.../cc749177(WS.10).aspx
but it doesn't seem to me like this is the case.

As I see it, what MS says is:

• DO NOT USE FDISK (latest version for 98/Me) to create partitions larger than 512 Gb, as the utility was NOT designed to do this
  
• ANYTHING above 128 Gb results in "a suffusion of yellow" UNLESS the particular PC BIOS supports 48-LBA addressing

Whilst #2 is without any doubt true, and also verified, #1 needs experimentations, as an example, it is possible that partitions bigger than that can be created but they result, once FORMAT is used on them, in incorrect data in the bootsector, like wrong number of FAT tables, wrong number of hidden sectors, wrong something that may go unnoticed until you actually try and check EVERY aspect of the created partition/filesystem.

And again this problem may be something that it takes 5 seconds to fix with a hex/disk editor or something that needs a more complex patch.

jaclaz

This post has been edited by jaclaz: 25 February 2010 - 03:51 AM

[rloew]

I ran some tests on the following versions:

98SE FDISK Original
98SE FDISK Q263044 Update
ME FDISK Original
98SE FORMAT original
ME FORMAT Original

FDISK itself is limited to ~512GiB. Above 512GiB it wraps around, so a 600GiB drive would only be partitioned to 88GiB.

FORMAT will work with Partititons up to ~1TB. Above 1TB a divide error occurs.

All versions behaved similarly.

This post has been edited by rloew: 26 February 2010 - 12:15 AM

[dencorso]

Thanks a lot, RLoew! You rock!

I think your tests settle the OP questions.

Yet, as you seem to be the only one of us to have conditions to perform such tests, I'd ask you to perform one more test with FORMAT (let's say just the Win ME version), at your convenience, of course, to settle also the question left open by Marius '95 original thread (link)... Marius '95 used a RAID 0 from two WD5000AAJB, that should have a full capacity of 1,000,215,724,032 bytes = 931.52 GiB, considering the official value of 976,773,168 sectors per drive, found in WD document 2879-001146. He was unable to format it, but it could be ascribed to partitioning probles previous to the actual formatting... So my question is: can FORMAT actually format 1 TB exactly (= 931.32 GiB)?

[submix8c]

So... it seems (ref. 48-bit support in BIOS) -

1 - another partitioning method other than FDISK needs used
?or is it NO-Partitioning-Software>~512GiB if NO BIOS SUPPORT?
according to jaclaz, only FDISK is affected (above query...) so...
?No BIOS Support: Max=128 / BIOS Support: Max=512 Only FDISK?...
(...and still need the MBR correctly created?)

2 - FORMAT will format them, up to ~1TB only if BIOS Support (according to both)

Correct?

[dencorso]

Well... yes. DOS relies on BIOS. If the BIOS doesn't support 48-bit LBA one can:
(I) update the BIOS from the manufacturer (if available), or using an unofficial modded BIOS (if available),
for free or (II) update the BIOS from eSupport.com (if available) not for free, or (III) use RLoew's DDO (not for free), or (IV) use an add-on HDD controller (say, from Promise), instead of the motherboard's ports, or (V) dump the motherboard and move-on to another one (probably used) that supports 48-bit LBA (this is probably the most work-intensive alternative), or (VI) decide one can live with the 137 GB limit.

[rloew]

dencorso, on 26 February 2010 - 06:02 PM, said:

Thanks a lot, RLoew! You rock!

I think your tests settle the OP questions.

Yet, as you seem to be the only one of us to have conditions to perform such tests, I'd ask you to perform one more test with FORMAT (let's say just the Win ME version), at your convenience, of course, to settle also the question left open by Marius '95 original thread (link)... Marius '95 used a RAID 0 from two WD5000AAJB, that should have a full capacity of 1,000,215,724,032 bytes = 931.52 GiB, considering the official value of 976,773,168 sectors per drive, found in WD document 2879-001146. He was unable to format it, but it could be ascribed to partitioning probles previous to the actual formatting... So my question is: can FORMAT actually format 1 TB exactly (= 931.32 GiB)?

I ran the tests using a 2TB Hard Drive so I was not limited to 1TB. I tested the original Windows 98SE FORMAT using 1018.83GiB = 1,093,962,207,744 Bytes which is more than Marius' Raid System.

Quote

1 - another partitioning method other than FDISK needs used
?or is it NO-Partitioning-Software>~512GiB if NO BIOS SUPPORT?
according to jaclaz, only FDISK is affected (above query...) so...
?No BIOS Support: Max=128 / BIOS Support: Max=512 Only FDISK?...
(...and still need the MBR correctly created?)

BIOS support, if run from DOS, or Windows support, if run from Windows, is mandatory for any Partitioner using Motherboard INT 13 Calls.
The only exception is if the Partitioner supports raw Disk I/O itself, such as my RFDISK, or if you use a PCI card that has it's own BIOS.

Without 48-Bit LBA support, FDISK and other Partitioners will correctly create the MBR but any Logical Partition or Next Extended Partition Record will be placed in the first 128GiB even if the Partitioner thinks it is placing it above the limit. Reading back the record is affected in the same way, so everything will seem OK. You can reload the Partitioner and it will show you exactly what Partitions you thought you created.

If you use such a Partitioned Disk in the orignal system, everything will seem fine until data written in one 128GiB Block overwrites data in another 128Gib Block.

If you move the Hard Drive to a 48-Bit LBA aware setup, the Partitions starting above 128GiB will disappear.

[submix8c]

@rloew - Gotcha - check (confirmed by above, and thought so...)!

So only alternative is to use a "good" DDO to exceed 128gb on non-48-bit-BIOS which "simulates" 48-bit for want of a better term (please don't argue semantics, I know what DDO's do and how - ARGHH!).

[rloew]

submix8c, on 28 February 2010 - 10:32 AM, said:

@rloew - Gotcha - check (confirmed by above, and thought so...)!

So only alternative is to use a "good" DDO to exceed 128gb on non-48-bit-BIOS which "simulates" 48-bit for want of a better term (please don't argue semantics, I know what DDO's do and how - ARGHH!).

I would use "provide support for" rather than "simulate".
I think DDO's got such a bad reputation because some early versions protected their own code by shifting the sectors on the entire Hard Drive. This made the Hard Drive unreadable if the DDO wasn't active, and all data was lost if the DDO was damaged.

My BOOTMAN DDO and other DDO's I have written for other purposes, do not alter the Disk Layout, so they are safe.

[dencorso]

OK. Good. Now, please, if I may suggest something, do partition it to a single primary partiton with the free Ranish Partition Manager (use v. 240), and use your single > 650 GiB patition to test FORMAT.COM and SCANDISK.EXE...
from all we know they should both work OK.

BTW, a 750 GB Barracuda 7200.12, according to Seagate, has 750,156,374,016 bytes (= 698.6 GiB) , at least (from converting the published number of Guaranteed Sectors), so your BIOS is seeing the full HDD.

[Usher]

wsxedcrfv, on 04 March 2010 - 06:36 PM, said:

This 750 gb SATA drive was connected to my win-98se system via a Silicon Image SIL 3512 SATARaid PCI controller card.

The name is not SIL. It is SiI as Silicon Image

wsxedcrfv, on 04 March 2010 - 06:36 PM, said:

I tested 3 versions of fdisk:
(...)
All 3 versions reported that my new drive had a capacity of 5773 mb.
(...)
I'm not exactly sure of the math that fdisk is using (or misusing) to arrive at 5.77 gb for a 698 gb drive.

The size reported by fdisk is in MiB (use UPPER case when needed) and depends on CHS/LBA settings reported by the controller BIOS. Show your numbers and we can do the math.

[dencorso]

wsxedcrfv, on 08 March 2010 - 11:35 PM, said:

What I think I'll do next is see if DOS will format the drive, and I'll check to see if the drive is operating in SATA mode or legacy IDE mode (I'm not sure). Is there a way to know from within Windows if the drive is being accessed using the SATA driver, or ESDI_506.pdr? In device manager, drive properties, driver, it says "Provider: (Standard disk drives), Date: 4-23-1999. No driver files are required or have been loaded for this device."

Great! And yes, the Ranish PM has this strange behaviour of defaulting to 16 kiB clusters... But perhaps FORMAT.COM will change the cluster size, on reformatting. And if it doesn't, do give a try to the famous /Z switch.
Well, look unser SCSI, on the device manager. You probably have an entry for yor HDD there, and in its properties you'll see the SATA driver listed.
Please do run the Win ME DOS SCANDISK before and after using FORMAT.COM.

[rloew]

FORMAT.COM may be relying on the existing Cluster size when it reformats. 16KiB Clusters is not the recommended Cluster size.
Another possibility is that it is using the truncated value of 60GB to decide the Cluster size.

[dencorso]

Yes, there are. I'd like to know how the DOS programs SCANDISK.EXE (from Win ME or from BHDD31E) and, in case you have it, NDD.EXE (from Norton 2002) behave with your 45 million cluster 750 GB partition. After testing this, you could change the cluster size manually in the BPB of the partition boot record and try FORMAT.COM again: if RLoew is right and FORMAT.COM uses the value present in the BPB, when it finds one, this should cause it to use 32 kiB sectors.