Yes, since you're open to a reformat that's definitely the easier way to go. The shell swap might've been helpful if you didn't want that, because it'd be too much stuff to reinstall or whatever.

elszgensa wrote on 2025-11-15, 22:05:

Yes, since you're open to a reformat that's definitely the easier way to go. The shell swap might've been helpful if you didn't want that, because it'd be too much stuff to reinstall or whatever.

Thankfully I don't have an awful lot installed or set up, so it won't require a lot of backtracking. I'll post updates after certain milestones/tests.

Well, I've reformatted and now have a fresh install of Windows on. Again. However, I seem to have another issue. Or rather, another problem involving the bloody Gotek FlashFloppy.

As I can't install the USB drivers yet, I don't have another way of getting my stuff on the PC. Not everything anyway. Just been trying to get GHOST back on so I can make a backup of the new install. It's too big to fit onto a regular floppy disk, so decided to disconnect the physical floppy drive for the time being and re-connect the Gotek. I made an image of it and put it on my pen drive, but now the Gotek isn't being recognised properly again. The LCD screen lights up and I'm able to scroll through everything, but the green light doesn't come on during boot/floppy seek. So I imagine I won't be able to access it in DOS. And it isn't recognised in Windows either. Not with it set to drive A or B. 🙁

Checklist:

- Gotek connected at the end of the floppy cable, red stripe on the cable lining up with pin 1
- BIOS is set to use A: as a 1.44MB drive
- "Boot up floppy seek" is enabled in BIOS. If the drive doesn't respond at boot, BIOS will stop with an error message.
- Jumpers on the Gotek are set properly ("S0" for it to be the first floppy drive, and also found this:

JC sets the interface mode to IBM PC compatible. It just changes the behaviour of pins 2 & 34 at the floppy header. Many need this; and equally many don't.

does that do anything?

this article also mentions that if your Gotek is a newer unit with no JC jumper, it needs "interface=ibmpc" in the ff.cfg config file.

asdf53 wrote on 2025-11-16, 06:01:

Checklist: […]
Show full quote

Checklist:

- Gotek connected at the end of the floppy cable, red stripe on the cable lining up with pin 1
- BIOS is set to use A: as a 1.44MB drive
- "Boot up floppy seek" is enabled in BIOS. If the drive doesn't respond at boot, BIOS will stop with an error message.
- Jumpers on the Gotek are set properly ("S0" for it to be the first floppy drive, and also found this:

JC sets the interface mode to IBM PC compatible. It just changes the behaviour of pins 2 & 34 at the floppy header. Many need this; and equally many don't.

does that do anything?

this article also mentions that if your Gotek is a newer unit with no JC jumper, it needs "interface=ibmpc" in the ff.cfg config file.

I'll check the cable as it could be wrong. My dual cable had to be replaced with a single one at short notice, so it could be something I've overlooked. Thanks. 😀

The rest, I believe, all check out. I've tried both A at 1.44MB, B, and both. Floppy Seek has been enabled and disabled, too. It always comes up with an error even if it's enabled. It did do even when it was working. I just figured it was because a physical one wasn't yet hooked up. I had to have both enabled (A and B) before in order to use the Gotek in Windows. If I only had B enabled, it would only give me a 5 1/2 drive in Windows and I couldn't access the Gotek. With A and B, it showed a 3 1/2 and I could.

The creator suggested I have it at JC in order to get it working in DOS, which it was before.

I'll look at removing FF.cfg and the CFG file next and try a factory reset. See if that sorts it. Failing that, I'm going to put a CD-R together with only the essential stuff on.

Drive B: is only relevant when you are using a double header floppy cable with a drive connected to the middle connector. That becomes drive B:. The drive at the end of the cable is always drive A:. This is physically enforced through the twist in the ribbon cable.

If you have no physical drive B: connected, you shouldn't ever enable drive B: in the BIOS. It just adds to the confusion, and it will always give you a drive not found error.

asdf53 wrote on 2025-11-16, 12:48:

Drive B: is only relevant when you are using a double header floppy cable with a drive connected to the middle connector. That becomes drive B:. The drive at the end of the cable is always drive A:. This is physically enforced through the twist in the ribbon cable.

If you have no physical drive B: connected, you shouldn't ever enable drive B: in the BIOS. It just adds to the confusion, and it will always give you a drive not found error.

I see. But that should be fine if the middle of the dual cable is hooked to the Gotek, right? That's how I had it set up before.

Yes, that's fine. The Gotek would then become drive B: in the BIOS, and if you had another drive at the end of the cable, that would become drive A:. You just have to tell the BIOS what's present, it will never detect drives by itself, nor does Windows.

It also says this about the Gotek on the Github page:

IBM PC compatibles have a non-Shugart interface which must be explicitly configured in FlashFloppy: […]
Show full quote

IBM PC compatibles have a non-Shugart interface which must be explicitly configured in FlashFloppy:

IBM-PC interface mode must be configured
Strap jumper JC at the rear of the Gotek; or
Specify via FF.CFG: interface = ibmpc
Strap select-line jumper S1 at the rear of the Gotek
S0, S2, MO should all be left open

So what I said above was wrong, don't set the S0 jumper.

Okay. It still wasn't detecting the Gotek with the physical floppy drive the last I tried. Even hooked to the middle cable. Unless I overlooked something. This was right up until my dual cable broke. Just waiting for another one to arrive.

At any rate, I burnt some essential drivers and software to a CD-R, so I've made an image of Windows as it is. I guess now it's just a matter of trying to exit to DOS, exit back into Windows and see if the error re-surfaces, and then try one thing at a time.

I am wondering if the error is linked to the Audigy 2 drivers I used before. The WDM ones. Maybe now that I'll be using the VXD ones, it will avoid any weird issues. We'll see.

Boom. Straight away. Quit to DOS, quit back to Windows - XIT error. 🤣

DustyShinigami wrote on 2025-11-16, 13:25:

Boom. Straight away. Quit to DOS, quit back to Windows - XIT error. 🤣

So this time you didn't install anything? No USB driver, no software, nothing? Your Windows 98 install CD is unmodified?

As long as your system is relatively clean, here's another idea: Run filemon, but this time, don't filter for "xit", let it show everything. Run filemon as a service through the registry as before to launch it early, and when the error pops up, save the filemon log. Since it records all activity sequentially, you could scroll to the part where "xit" appears and then look at what it did just before that. It might give a clue by looking at the path/folder it's trying to access before.

Also:

elszgensa wrote on 2025-11-15, 19:22:

You could change your shell from explorer to something else, like good old progman.exe, or something third party like bb4win. You could even boot straight into Olly and go from there... Anyways, the idea is to hopefully get the system up without starting explorer.exe at all. At this point you could, for example, open and prepare Filemon and Olly and leisurely launch a debug-attached explorer instead of having to time anything. Or if you still get the message when the new shell starts, at least you know it's not specifically explorer's fault.

That's a great idea actually. I just tried it (changing shell=explorer.exe to taskman.exe in c:\Windows\system.ini) and it would at least allow to run explorer.exe independently. The only problem is that I couldn't figure out how to make explorer launch the complete shell environment after that, it will only open a standalone explorer window. Is there a trick to make it load everything? In any case it'd still be interesting to try if the explorer window alone is enough to trigger the error.

asdf53 wrote on 2025-11-16, 14:12:

So this time you didn't install anything? No USB driver, no software, nothing? Your Windows 98 install CD is unmodified? […]
Show full quote

DustyShinigami wrote on 2025-11-16, 13:25:

Boom. Straight away. Quit to DOS, quit back to Windows - XIT error. 🤣

So this time you didn't install anything? No USB driver, no software, nothing? Your Windows 98 install CD is unmodified?

As long as your system is relatively clean, here's another idea: Run filemon, but this time, don't filter for "xit", let it show everything. Run filemon as a service through the registry as before to launch it early, and when the error pops up, save the filemon log. Since it records all activity sequentially, you could scroll to the part where "xit" appears and then look at what it did just before that. It might give a clue by looking at the path/folder it's trying to access before.

Also:

elszgensa wrote on 2025-11-15, 19:22:

You could change your shell from explorer to something else, like good old progman.exe, or something third party like bb4win. You could even boot straight into Olly and go from there... Anyways, the idea is to hopefully get the system up without starting explorer.exe at all. At this point you could, for example, open and prepare Filemon and Olly and leisurely launch a debug-attached explorer instead of having to time anything. Or if you still get the message when the new shell starts, at least you know it's not specifically explorer's fault.

That's a great idea actually. I just tried it (changing shell=explorer.exe to taskman.exe in c:\Windows\system.ini) and it would at least allow to run explorer.exe independently. The only problem is that I couldn't figure out how to make explorer launch the complete shell environment after that, it will only open a standalone explorer window. Is there a trick to make it load everything? In any case it'd still be interesting to try if the explorer window alone is enough to trigger the error.

Nope. Absolutely nothing. All I did was copy and paste the Ghost files into a DOSPro folder on C so I could make an image of its current state. The USB adapter was unplugged, too. And the Windows 98 disc is an original unmodified CD bought many years ago. But good idea about calling FileMon now. There won't be a lot it'll load. 😀

As to the second option, so long as I can easily switch it back to default again, it's worth a shot after trying FileMon.

DustyShinigami:

Can you please tell us a bit about the hardware side of things? Specifically what CPU, and, if it's multi/hyperthreading-capable, whether you turned that off in the BIOS or not. I am of course aware the 9x series doesn't know how to use actual multicore/-threading, and this is totally just a stab in the dark, but I'm thinking maybe the CPU does some internal rescheduling and ends up clobbering a string buffer? "xit" is awfully close to "exit". It's also just three characters so random matches like this are not unlikely... I dunno. An entirely fresh install with zero customisations (other than having Ghost copied onto the disk) doing this, absolutely stumps me.

Speaking of Ghost. So you installed Windows, did a reboot (probably multiple, judging by your Gotek issues, but whatever), copied over Ghost, ran that (or not? didn't actually say). At what point through this process did the error pop up? -- Actually, re-reading your post now it sounds like you didn't check for the error at all before introducing Ghost. You also mention it not fitting on a floppy - I remember that the version I had did fit. That was a DOS one meant to be run from a standalone boot floppy. If your Ghost is Windows based, and/or if it incorporates any kind of installation process, would you mind starting over and checking before running that?

And a question about your Windows CD. Sounds like a pressed disc, which is a good start, but just to make sure - is it a Microsoft-made one (including OEM copies), or by a bigger OEM doing their own media (meaning we can't entirely rule out customisations made by them)?

asdf53 wrote on 2025-11-16, 14:12:

The only problem is that I couldn't figure out how to make explorer launch the complete shell environment after that, it will only open a standalone explorer window. Is there a trick to make it load everything?

Beats me. As far as I can recall, I've always had the inverse issue, with the first instance of Explorer wanting to behave as a shell no matter what - I distinctly remember having to use a third party file manager because of that. At least that was my experience on XP, and tbh I only ever really swapped shells on that (to bb4win) as well as back on 3.x (to Calmira). On 9x it was limited to short stints like "progman still works, that's cool to see, now let's go back".

An update on this, and I have absolutely no idea why or how, but the error appears to have stopped. Usually it kept popping up if I exited from DOS back into Windows. To begin with, it did do it, but now it's stopped. So not sure if I installed something prior to playing around with installing DOS games...? Or maybe my SB Live! Value drivers did something...?