Well, my memory seems to be faulty. I don't think its a software issue after all. It doesn't matter when I hit CTRL-ALT-DEL, even if I hit it before POST, CTRL-ALT-DEL equals not going to boot. Been going through the BIOS options to see if I may have something set wrong in there (have the latest revision.) But not having much luck, nothing seems appropriate. Any ideas, or am I going to have to just hit the reset switch all the time?

Sounds dumb but did you tried with a different keyboard (maybe it is not sending the right sentence code or motherboard fail to interprete it, like data corruption in the signal that can halt the computer in its process) ? You can always simulate the ctrl alt del process with DEBUG under DOS (dont have the exact command in mind)... if it works that may indicate a problem between data send by the keyboard and data received by the motherboard.

EDIT : found it here http://www.robvanderwoude.com/debug.php

That works fine. All the software methods of rebooting the system work just fine, it's only the <CTRL><ALT><DEL> combo that gives me a problem. I don't have access to another keyboard, or I would try switching it out. The kicker is the way that key combination is designed to work. It is a hard wired signal that cannot be generated in software (why it is used, or it was created that way because, to trigger the logon screen in the early NT OS.) All the G= command does is start the BIOS startup sequence (what the computer does after POST,) which is why it triggers a warm reboot.

Doing web searches, <CTRL><ALT><DEL> broadcasts a pending shutdown alert then jumps to FFFF:0000. FFFF:0000 (regardless of how you address it, and the 8086/80286/80386 each had different methods,) is a pre-set start location that is usually just a JMP command to the start of the BIOS startup sequence (wherever it installs itself on any particular board.) It seems that the trouble comes from the "broadcasts a pending shutdown alert" portion as that is the only variable that is both present in all cases of crash, and missing in all cases where things go smoothly.

Thing is, I have no idea what is actually done when BIOS "broadcasts....etc..." How/Why would this cause an issue? Also, it is possible that the signal the KB sends that triggers the BIOS to do this could be causing the issue. It's probably either something that is absurdly simple or something I'll never be able to fix.

Thanks for clarifying things out about that reboot sequence 😊
Really hope that the simple one 😵

Unplug all HDDs and just use a Floppy. Does it have the same issue?

I had this issue often with PCI SATA controller cards on my Super Socket 7 boards. So in my case it has to do with storage and somehow gets stuck. I also don't know what the fix is as wasn't consistent.

Well, tracked down another keyboard (PS2 - as I am out of adapters.) Hooked it up and everything works fine. So, I hooked my main kb up directly (moving the adapter with it from the KVM,) and again it works fine. Attached the PS2 KB to the KVM and the problem came back. So, my KVM is the problem.... but how? Sitting on hold with StarTech now to see if they can help figure it out. The thing is, will this be fixed by just replacing the KVM, or will I have the problem with any KVM I get? I've been planning on getting a new KVM eventually anyway, but I'd sort of like to make sure the problem won't persist.

Wow that's a weird issue but awesome for tracking it down to the KVM!

I've been planning for ages to replace this KVM anyways, with one that either uses a safer function key or has a programmable function key. I use the CTRL key way to much to have a KVM that can only use that as the function key. This one will be moved to my 8-bit (ish) table. Where I have a C128D, TI-99/4A, A500, & Apple IIGS (eventually,) sharing a monitor & mouse only (they all have their own, proprietary keyboards - and adapters for the mouse.) The issue I have is.... What is causing the problem? Still can't track that down. There is one thing I could try (later when I'm not so out of it,) and that is to move the jumper enabling KB power on the P3V133 board. Tremors are a little bad right now, handling the tiny little jumper is well beyond me at the moment 🙁

Maybe the ps/2 extension cable between the computer and the kvm switch is the culprit (loss of signal or corrupted data due to a defect somewhere),or just that the signal send by the kvm is not strong enough for the motherboard ?
I know I extrapolate too much, but can it be due to some loops in the data transfered ? Making something like a packet collision on a network, driving the motherboard crazy when received 😵

It could be something like that, though I don't see how. From what I understand the signal from the KB just triggers an interrupt, it's a binary condition on/off. If the signal was too weak, it just wouldn't trigger the interrupt. This interrupt tells the BIOS (or the advanced OS for Windows & such,) that the user just hit the combo. At this point the KB is "supposed" to be out of the loop, the BIOS takes over, sends the alert which tells any properly programmed TSRs to do their end process cleanup. At this point it executes a jump to FFFF:0000, which initiates the BIOS startup sequence - thus a warm boot (no POST.) The signal on the KB is supposed to be hardwired, as is the interrupt. All you can do is change where the interrupt goes. The BIOS, during it's startup sequence, programs the interrupt to go to it's warm boot sequence (described above.) Later, an advanced OS (like Windows,) can change the interrupt target to whatever it needs (like the security sequence in Windows NT.)

The only thing that is making any sense to me is the CTRL = Function key aspect of my KVM. Maybe it's causing the hardwired signal to pulse or some such. If the MB is sensitive to this, "maybe" it could cause corruption on the interrupt line, or in a register, or something. Basically, when I hit the combo, even for just a second, the motherboard actually gets a series of interrupts all in a row instead of just a single constant signal.

There was something similar to this causing me problems with my A2000. My first KB adapter wouldn't allow me to ROM switch. The ROM switch board I installed in the Amiga uses the reboot signal to switch. By holding down the reboot combo for 2 seconds, it would switch ROMs and signal with a beep. Only, with my first adapter, it would just reboot like I had just tapped the combo instead of holding it down. My new adapter is a little more versatile, and it works just great. If the signal is pulsing, maybe the first adapter just sent the first signal and waited for the rest to clear up before it started converting again, while the second adapter just passes it through as is, or blends the pulsed signal into a constant.

FeedingDragon wrote:

the CTRL = Function key aspect of my KVM

I've totally missed that part before 😒 It effectively sounds like a big source of trouble here 😀
Glad that the computer is ok though 🤣