Hacker Newsnew | past | comments | ask | show | jobs | submitlogin

People give that one a lot of crap, but what it meant was just "press F1 after connecting the keyboard". In later BIOSes at least the error could usually be turned off, if you want to boot without a keyboard.


The thing is, it should have shown: Power off to reconnect Keyboard. At least up to PS/2-style connectors. Because you could short/blow out the onboard Super-IO doing that, while powered on.


Is that actually true? I'm not convinced IBM and other PC manufacturers made such a mistake, but I'm happy to be proven wrong. It sounds a lot like it could be some sort of "common wisdom", because some ports presumably were dangerous to hotplug, so laypeople were just told to generally be cautious.

FWIW, I never followed that advice for anything as far as I can remember (save for really obvious stuff like ISA or PCI cards, or in professional settings where I was following some procedure), and nothing ever blew up. Could have been lucky.

(Super-IO was very late, by the way, it was 8042s and derivative microcontrollers originally.)


Hey, regarding the really obvious stuff like PCI cards...

I once unplugged a NCR/Symbios Logic SCSI-controller from a running system.

And plugged it back in very carefully after a while.

While running graphically under NetBSD with 1.5GB Virtual Channel Memory. (Slightly better SD-RAM, only supported by some chipsets)

Monitor went ZONK, because NetBSD switched to text-mode console, which 21" Supersync barely supported, spewed endless alerts, after alerts.

After replugging something like found controller, resetting, found disk(s), resetting, alerts gone, switching back to X.

Magic.

Didn't even have filesystem errors, which I checked for after that stunt.

Which btw. didn't work with an Adaptec2940 fast and wide.

Nor Linux, in whichever config.

Some hot plugging, eh? :-)


PCI could be hot-pluggable actually. It requires a lot of fiddly bits though. Especially, you need to setup enough address space for the cards without knowing what they are.


I think this is actually the reason why you can only use 3 GB of 4GB on some older Thinkpads. Because docking, and such, they took care of that.


Heh, amazing. I love when stuff like this "works out".


I have fried a PS/2 port before by doing that. Once, in the 1990s, at work. That doesn't tell you how often it happens though because I haven't hot plugged in a PS/2 device since then (and now I don't use them much anymore).


Personally, I've been lucky. Though I have seen sparks while plugging while powered. Also I've got many broken systems where at least one port (of two) was broken, but could be compensated by 'Y-wire' where you'd put mouse and keyboard via that Y-wire into the remaining port. Which later became some sort of pseudo-standard, cost cutting, because only one port is cheaper? I also seem to remember advertising on the package of better boards, which had protection against that.


The smarter KVM switches, well, KV switches, would simulate a keyboard even if you're not connected to the server, but the dumber KV switches were dumb, so you'd have to select the rebooting server so it could see the keyboard. Then you'd adjust the bios so it doesn't require a keyboard to boot.

That's just how it was in the old days.




Consider applying for YC's Summer 2026 batch! Applications are open till May 4

Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: