Hmmmm
Some candidates.

[HKEY_LOCAL_MACHINE\DeadNT\ControlSet001\Services\Abiosdsk]
"Start"=dword:00000004
"Group"="Primary disk"

[HKEY_LOCAL_MACHINE\DeadNT\ControlSet001\Services\Atdisk]
"Start"=dword:00000004
"Group"="Primary disk"

Start=4 = disabled.

Load registry on one of the 2 drives
Change to start=0
put in OG machine
test

This Dell one is also a candidate
[HKEY_LOCAL_MACHINE\DeadNT\ControlSet001\Services\Delldsa]
"Start"=dword:00000004
"Group"="SCSI miniport"

It's very weird that it only has 4 set to 0

[HKEY_LOCAL_MACHINE\DeadNT\ControlSet001\Services\atapi] (cdrom)
[HKEY_LOCAL_MACHINE\DeadNT\ControlSet001\Services\Disk] (filesystem)
[HKEY_LOCAL_MACHINE\DeadNT\ControlSet001\Services\intlfxsr] (no idea)
[HKEY_LOCAL_MACHINE\DeadNT\ControlSet001\Services\PowerSwitch] (duh)

NT4's atapi.sys is the IDE port driver (also used for hard disks), the cdrom is called cdrom.sys. intlfxsr is a driver that enables CR4.OSFXSR to do fast context switches on CPUs supporting SSE instructions, abiosdsk is for disk controllers connectred to MCA running on an IBM machine with ABIOS firmware.
I'm not sure what atdisk is.

myne wrote on 2024-10-24, 14:59:

Hmmmm Some candidates. […]
Show full quote

Hmmmm
Some candidates.

[HKEY_LOCAL_MACHINE\DeadNT\ControlSet001\Services\Abiosdsk]
"Start"=dword:00000004
"Group"="Primary disk"
....

Sorry but my level of knowledge is not that high, so I'm a bit lost and I need to understand.
What are we looking at exactly? Are those registry keys telling the bootloader which device should be accessed at boot?

Also, how could the problem be in the registry file? AFAIK nothing could have possibly changed there since I just plugged the drive into another PC to clone from it.

I've copied all the content from the image of the original HDD's main partition to the the main partition of a virtualized NT4 install I had done previously, and it works. Heck, I even tried to recreate the same 2 partition layout of the original computer and that also seem to work, despite me using W11 disk mangement to patition the VHD. I'm so puzzled.
Granted, there might be some Virtualbox trickery going on that makes it work, but I'm still more inclined to believe the problem is outside of the main partitions otherwise I should have gotten the same "INACCESSIBLE_BOOT_DEVICE" error, no? After all the image of the disk was done after I had already plugged it into another PC, so it should carry the same errors.

Well if you're going to just copy files, make sure you keep the registry at least.
It's those files without extensions.
System etc

V0LDY wrote on 2024-10-24, 18:31:

Sorry but my level of knowledge is not that high, so I'm a bit lost and I need to understand. What are we looking at exactly? Ar […]
Show full quote

Sorry but my level of knowledge is not that high, so I'm a bit lost and I need to understand.
What are we looking at exactly? Are those registry keys telling the bootloader which device should be accessed at boot?

Also, how could the problem be in the registry file? AFAIK nothing could have possibly changed there since I just plugged the drive into another PC to clone from it.

I've copied all the content from the image of the original HDD's main partition to the the main partition of a virtualized NT4 install I had done previously, and it works. Heck, I even tried to recreate the same 2 partition layout of the original computer and that also seem to work, despite me using W11 disk mangement to patition the VHD. I'm so puzzled.
Granted, there might be some Virtualbox trickery going on that makes it work, but I'm still more inclined to believe the problem is outside of the main partitions otherwise I should have gotten the same "INACCESSIBLE_BOOT_DEVICE" error, no? After all the image of the disk was done after I had already plugged it into another PC, so it should carry the same errors.

Then it goes back to an issue I have pointed out elsewhere: The BIOS settings (and prolly a dead CMOS battery).

Because if nothing is damaged or changed in the original disk image, and it works in a virtual machine, then I don't know what the issue could be apart from settings on the machine itself, unless the fuckery that may have happened in the boot loader causes issues only on the OG machine.

Also, I've sent you a private message on facebook the other day, could you check it? (prolly went lost in "messages requests"). I didn't ask it publicly because they could be more private informations.

Then it goes back to an issue I have pointed out elsewhere: The BIOS settings (and prolly a dead CMOS battery).

Nope, BIOS is rock solid, even bringing the computer back home it didn't lose anything. I took a photo of each page just in case the battery dies on the return trip, but it seems to be mostly standard configuration.

Btw, I managed to bring back the system online complete with the 2 partitions doing what I did to the VM: I reinstalled Windows NT 4, updated it with the service pack, then I've connected it to a modern computer (this time with an USB adapter I just bought and not directly using an IDE cable), deleted all the files (at least the visible ones, see below*), copied back the files from the backup image and it seemed to boot just fine, the software that matters says it can't find a directory but I'm pretty sure that was on the network so it should work once it's connected again.
I haven't had the opportunity to see if it works with the machinery, but so far it seems promising.

*Here I've noticed something weird, mounting the image file I've created with Macrium I can't see some of the files needed for NT 4 bootloader to work (specifically ntldr and NTDETECT.COM) are missing. And I can't see those files either in WIndows even if I access the drives that are now working. Even enabling "hidden files" it seems that that some stuff is just hidden no matter what. Same goes for accessing the virtual drives, if I attach them to a Linux VM everything is displayed including ntlr etc, but not in Windows. Now that might be common knowledge, but I was still puzzled by that.
When I'm deliting stuff from the Working NT4 install it means I'm just deleting the "visible" files and replacing them with other "visible" files, not sure how many are't actually being displayed (and deleted/replaced), but it might actually be a good thing since it's booting.

And speaking of Macrium, there I might have made a mistake, I didn't check the advanced option and used the default "Intelligent sectory copy" instead of the "Make exact copy" which should be a 1 to 1 clone suitable for forensic examination etc. Now, I'm not sure if the files are present or not in the Macrium image (for sure I can't see them in Windows), but that might explain why the cloned disk didn't work. I say might because it's still suspiscious that the error was the exact same as the one I got from the original drive, so it's possible it occurred before the cloning and that it carried over to the image.

---

That said, I'd still relly like to know WTF happened to the original drive, I've now connected it to a linux pc and everything seems to be there (including all the files that might be missing from the image).
I'll now try and see if I can create a proper image to feed to the VM and see if it boots or if it will give me the same INACCESSIBLE_BOOT_DEVICE error.
Another thing I might to reinstall Windows 4.0 from scracth onto another drive and connect it again to the computer that seemingly made the original drive unreadable, just to see if it happens again.
My guess at this point is that somehow the MBR got fckd up, apart from that I really dunno.

So... Somehow ntldr was deleted from both images?!
Do you have a wayward virus scanner?
Check logs maybe?

Theory: drive inserted, virus scanner quarantined ntldr and Co, macrium did as expected but the files were gone.

If that's the case, just copying ntldr and co back to the original drive will fix it.

Ntfs is pretty good like that. You don't need special sectors. Just the files there. It's entirely possible to just format as Ntfs and copy across.

It is weird you got that error though. I would have expected "ntldr missing".

myne wrote on 2024-10-26, 13:47:

So... Somehow ntldr was deleted from both images?! Do you have a wayward virus scanner? Check logs maybe? […]
Show full quote

So... Somehow ntldr was deleted from both images?!
Do you have a wayward virus scanner?
Check logs maybe?

Theory: drive inserted, virus scanner quarantined ntldr and Co, macrium did as expected but the files were gone.

If that's the case, just copying ntldr and co back to the original drive will fix it.

Ntfs is pretty good like that. You don't need special sectors. Just the files there. It's entirely possible to just format as Ntfs and copy across.

It is weird you got that error though. I would have expected "ntldr missing".

I'm not sure ntldr is in the images or not. Macrium allows me to mount the image as a virtual drive on Windows, and from Windows I can't see it, but that might be Windows problem, not an issue with the image itself.
Right now I don't have any disk burned with the Macrium image so I can't check, I might see if I can restore the image to a VHD, in that case it should be visible if I then open the VHD on Linux.

Btw, I attached the original drive to a Linux system and all the files are there. As I said, in theory I never did anything to that drive!

___

Edit: ok, I've now restored the Macrium image to a VHD and the virtual machine f*ing worked.
Like, not even a fuzz, it just straight up worked.
I modified the 4Gb VHD my NT 4 virtual machine used, resizing it to 115Gb in order to fit both the partitions from the Macrium Image, I wiped it with DISKPART on Windows 11, then I just mounted the VHD and I told Macrium to restore the original disk image to that.
Detached the drive, attached it to the VM, ran it and it worked.
I'm just so confused.

Obviously once inside NT4 everything seems to be in place, includint ntldr etc otherwise it wouldn't have booted in the first splace.

I'll now try to restore it again to physical drive, maybe the "clean" command with DISKPART is the key? I'll let you know!

V0LDY wrote on 2024-10-26, 13:28:

Nope, BIOS is rock solid, even bringing the computer back home it didn't lose anything. I took a photo of each page just in case […]
Show full quote

Then it goes back to an issue I have pointed out elsewhere: The BIOS settings (and prolly a dead CMOS battery).

Nope, BIOS is rock solid, even bringing the computer back home it didn't lose anything. I took a photo of each page just in case the battery dies on the return trip, but it seems to be mostly standard configuration.

Btw, I managed to bring back the system online complete with the 2 partitions doing what I did to the VM: I reinstalled Windows NT 4, updated it with the service pack, then I've connected it to a modern computer (this time with an USB adapter I just bought and not directly using an IDE cable), deleted all the files (at least the visible ones, see below*), copied back the files from the backup image and it seemed to boot just fine, the software that matters says it can't find a directory but I'm pretty sure that was on the network so it should work once it's connected again.
I haven't had the opportunity to see if it works with the machinery, but so far it seems promising.

*Here I've noticed something weird, mounting the image file I've created with Macrium I can't see some of the files needed for NT 4 bootloader to work (specifically ntldr and NTDETECT.COM) are missing. And I can't see those files either in WIndows even if I access the drives that are now working. Even enabling "hidden files" it seems that that some stuff is just hidden no matter what. Same goes for accessing the virtual drives, if I attach them to a Linux VM everything is displayed including ntlr etc, but not in Windows. Now that might be common knowledge, but I was still puzzled by that.
When I'm deliting stuff from the Working NT4 install it means I'm just deleting the "visible" files and replacing them with other "visible" files, not sure how many are't actually being displayed (and deleted/replaced), but it might actually be a good thing since it's booting.

And speaking of Macrium, there I might have made a mistake, I didn't check the advanced option and used the default "Intelligent sectory copy" instead of the "Make exact copy" which should be a 1 to 1 clone suitable for forensic examination etc. Now, I'm not sure if the files are present or not in the Macrium image (for sure I can't see them in Windows), but that might explain why the cloned disk didn't work. I say might because it's still suspiscious that the error was the exact same as the one I got from the original drive, so it's possible it occurred before the cloning and that it carried over to the image.

---

That said, I'd still relly like to know WTF happened to the original drive, I've now connected it to a linux pc and everything seems to be there (including all the files that might be missing from the image).
I'll now try and see if I can create a proper image to feed to the VM and see if it boots or if it will give me the same INACCESSIBLE_BOOT_DEVICE error.
Another thing I might to reinstall Windows 4.0 from scracth onto another drive and connect it again to the computer that seemingly made the original drive unreadable, just to see if it happens again.
My guess at this point is that somehow the MBR got fckd up, apart from that I really dunno.

That's actually a relief.

But maybe something REALLY happened when you hooked it up directly to the modern machine.

myne wrote on 2024-10-26, 13:47:

Theory: drive inserted, virus scanner quarantined ntldr and Co, macrium did as expected but the files were gone.

That could be it! I actually didn't even consider that the antivirus on the modern PC could be the culprit. I'm so used to running windows defender by itself that I completely stopped thinking about the old kind of intrusive antivirus softwares.
Something really happened right in that moment between connecting the HDD to the new PC and starting the backup process.

I'll give an update tomorrow once I have the confirmation everything is working again (including the machinery), but apart from that I might have some interesting new developments I'll share for the curious ones.

Ok, I finally had the chance to test the computer with the machinery and it all worked fine.

What I also found out is that somehow, if I used another computer with an USB/IDE adapter to clone the very first image I've made to a drive it worked on ther old PC without any issue.
Granted, I've cleaned the drive with DISKPART before cloning, but it's so weird because that image was created with the computer that created all the non working drives.

Because yes, in the end I narrowed it down to the computer I used initially to clone the drive.
I had two different USB/IDE adapter I wanted to use initially to clone the old drive, one turned out to have a dead PSU, the other one had a capcacitor that literally exploded when I plugged it in.
I've then used an old PC I had lying around which had both IDE and SATA ports to create the image with Macrium, I just connected the old IDE drive to the IDE cable and cloned it on Windows 10 installed on the SATA SSD.

Well, apparently both the IDE drives (the original one and the newer with the cloned NT system on it) stop working on the 1998 computer if they are plugged into the "intermediate" computer I've used to create the image. The image itself was perfectly fine tho, and that's what I've used to restore the system, but somehow it just "breaks" the drives for the old PC.

Any idea for why that could happend? At this point I don't even need that old PC anymore since I have the new USB adapter, but I was curious to try and understand what could have caused all this trouble.
Would this have happened on any other computer just by plugging the drive directly into an IDE controller or was it my specific computer doing something weird? It feels strange to me that Windows would handle the drive differently (or make any modification to it) depending on how it is connected.