wow a 6 or 8Mhz 386 ? ok kidding aside... I think you are in for an uphill battle with that Dell and a 512MB CF unless you use an XT-IDE bios. In 1988 biggest IDE drives were barely 40Mb iirc
How are you setting the C/H/S in that Dell ?

Interesting failure mode. Have you tried rebooting after creating some files, then listing them? Since the directory contents might be coming from cache instead of the card. That'd at least confirm things are properly broken, instead of only kinda-sorta...

Though, that still wouldn't explain fdisk being able to create a partition table...

uniracers wrote on 2024-02-11, 00:01:

Does anyone have any suggestions on what to try? I know I could try another CF card, but I don't want to spend money blindly trying to find one that works.

Hi there, yes I do. 🙂

Please try XTIDE BIOS, as an ROM on an ethernet card.

You can even use the floppy version for a test.
XUBDisk - floppy disk XTIDE Universal BIOS booter

The PC seems to be using Phoenix BIOS from the 80s, which has old HDD routines.

Just like with my Schneider Tower AT, which had a Schneider BIOS (based on Phoenix BIOS).

These old routines seem to have several issues with HDDs introduced after ATA-2 specification.

- Here's an article about the Windows 3.1 FastDisk driver, which also has trouble with it.
It gives an insight about the compatibility issues.
https://www.os2museum.com/wp/how-to-please-wdctrl/

I think it's because the BIOS isn't even IDE aware yet,
but merely uses the old WD1003 command set for MFM/RLL controllers (ST506/ST412 HDDs).

In my case, it was even worse.
Selecting a HDD type and adding a CF or DOM caused the BIOS to hang.
With using XTIDE Universal BIOS, the issue went away. It wasn't a hardware issue.

Here's a working copy of XTIDE Universal BIOS:
Re: IDE disk-on-module + 386

Good luck. 🙂

Edit:

Though, that still wouldn't explain fdisk being able to create a partition table...

I think that's because FDISK uses different BIOS function calls to access the HDD.
Or it reads the registers in a different way, to determine the drive capacity.
In comparison to MS-DOS itself, I mean.

There's a bit more information on it in the docs for the MS-DOS 5 HDD patch.
The file is available here: Re: Browsing a harddisk under DOS without using BIOS routines (INT13H)

Also note that not all CF modules support 8bit mode!

Thanks to all for the replies.

Horun wrote on 2024-02-11, 03:21:

wow a 6 or 8Mhz 386 ? ok kidding aside... I think you are in for an uphill battle with that Dell and a 512MB CF unless you use an XT-IDE bios. In 1988 biggest IDE drives were barely 40Mb iirc
How are you setting the C/H/S in that Dell ?

For the 256 MB card, I'm using 502 cylinders, 16 heads, and 63 sectors. This is what is reported by IDEINFO.

elszgensa wrote on 2024-02-11, 05:39:

Interesting failure mode. Have you tried rebooting after creating some files, then listing them? Since the directory contents might be coming from cache instead of the card. That'd at least confirm things are properly broken, instead of only kinda-sorta...

Though, that still wouldn't explain fdisk being able to create a partition table...

After a reboot, listing the directory still works.

Jo22 wrote on 2024-02-11, 07:45:

Hi there, yes I do. 🙂 […]
Show full quote

Hi there, yes I do. 🙂

Please try XTIDE BIOS, as an ROM on an ethernet card.

You can even use the floppy version for a test.
XUBDisk - floppy disk XTIDE Universal BIOS booter

The PC seems to be using Phoenix BIOS from the 80s, which has old HDD routines.

Just like with my Schneider Tower AT, which had a Schneider BIOS (based on Phoenix BIOS).

These old routines seem to have several issues with HDDs introduced after ATA-2 specification.

- Here's an article about the Windows 3.1 FastDisk driver, which also has trouble with it.
It gives an insight about the compatibility issues.
https://www.os2museum.com/wp/how-to-please-wdctrl/

I think it's because the BIOS isn't even IDE aware yet,
but merely uses the old WD1003 command set for MFM/RLL controllers (ST506/ST412 HDDs).

In my case, it was even worse.
Selecting a HDD type and adding a CF or DOM caused the BIOS to hang.
With using XTIDE Universal BIOS, the issue went away. It wasn't a hardware issue.

Here's a working copy of XTIDE Universal BIOS:
Re: IDE disk-on-module + 386

Good luck. 🙂

Edit:

Though, that still wouldn't explain fdisk being able to create a partition table...

I think that's because FDISK uses different BIOS function calls to access the HDD.
Or it reads the registers in a different way, to determine the drive capacity.
In comparison to MS-DOS itself, I mean.

There's a bit more information on it in the docs for the MS-DOS 5 HDD patch.
The file is available here: Re: Browsing a harddisk under DOS without using BIOS routines (INT13H)

Yes. it displays as "Phoenix 80386 ROM BIOS PLUS Version 1.10 A06". I will give this XUBDisk floppy booter a try when I get a chance.

wierd_w wrote on 2024-02-11, 07:54:

Also note that not all CF modules support 8bit mode!

What would be in 8bit mode, the XTIDE bios?

Is your IDE controller 16bit?

If it is, and you are using the AT versiin of XT-IDE, then disregard.

XT version of the bios will only try 8 bit modes, iirc.

This sounds like the bug in early AT BIOSes where IRQs get missed when reading multiple sectors because reading the status register (which clears pending IRQs) occurs at the wrong time

Yeah, I remember reading that as well.

'Losing half the word' was what it sounded like, and I immediately thought '8bit mode on CF that doesnt support it', but 'losing half the word due to mishandled IRQ' also fits.

jakethompson1 wrote on 2024-02-11, 17:20:

This sounds like the bug in early AT BIOSes where IRQs get missed when reading multiple sectors because reading the status register (which clears pending IRQs) occurs at the wrong time

Would Xtide Universal Bios fix that too?

Good news. Using XUBDisk, I was able to successfully boot from the Compact Flash card.

I wrote the XUBDisk image to a 1.44 mb floppy, and booted from that. As it was booting, I pressed the A key so that it would then boot from the floppy, and swapped the XUBDisk with my disk that has the MS-DOS files on it. I then went through the fdisk and format procedure, wrote a few files to the CF card, and confirmed that I was able to read them. Then, I tried letting it boot from the CF card, and it worked!

My next step will be to do some research on setting up the XTIDE bios on an ethernet card.

As far as whether the IDE controller is 8 bit or 16 bit, I'm not sure, but it is integrated into the motherboard.

Thanks @Jo22 for the suggestion and to everyone that responded.

douglar wrote on 2024-02-11, 18:44:

jakethompson1 wrote on 2024-02-11, 17:20:

This sounds like the bug in early AT BIOSes where IRQs get missed when reading multiple sectors because reading the status register (which clears pending IRQs) occurs at the wrong time

Would Xtide Universal Bios fix that too?

yes