This is x86 assembler. (Actually, looking at the register names, it’s probably x86_64. On old school x86, they were named something like al, ah (8 bit), ax (16 bit), or eax (32 bit).) Back in the old days, when you pressed a key on the keyboard, the keyboard controller would generate a hardware interrupt, which, unless masked, would immediately make the CPU jump to a registered interrupt handler, interrupting whatever else it was doing at the point. That interrupt handler would then usually save all registers on the stack, communicate with the keyboard controller to figure out what exactly happened, react to that, restore the old registers again and then jump back to where the CPU was before.
In modern times, USB keyboards are periodically actively polled instead.
I think there were a few early amd64 systems with genuine ps2, and I think you can still get one, but it wasn’t common, and honestly it’s probably usb->ps/2.
To be a pedantic asshole: mov eax, ecx? Unless you’re commenting on the insanity of interrupt driven i/O in the modern age of high performance, deep-pipelined superscalar OOO cores.
does that mean though that if I connect a PS/2 keyboard or mouse to my relatively modern computer (a “gamer” motherboard made ~6 years ago) 's PS/2 port, that it’ll still trigger such an interrupt?
The other commenter is on the right track but the chip controls both USB and PS/2 as well as others;
In the 90s and 2000s, for x86 machines, slower I/O was handled by a chip called the Southbridge which worked in conjunction with a chip called the Northbridge that handled faster I/O like IDE and PCI. Later these were integrated into a single chip and, as of recent processor generations, into the processor itself.
AFAIK ghosting and key rollover are issues when using PS/2 but it can offer some milliseconds off latency when used in high cpu games.
They are wholly independent from the protocol or interface. Ghosting is an electrical issue that is a result of keyboards being a bunch of switches arranged in a matrix. It makes the keyboard’s controller register an extra keypress in certain conditions. Nothing to do with how the thing communicates with the host computer.
Key rollover issues can be related to ghosting. The limit for it is once again the keyboard’s design at the circuit level, not its communication protocol.
Really they’re both related to how cheaply built the keyboard is. That’s the only thing.
Ghosting entirely depends on the wiring of the keyboard pcb. Key rollover can depend on the wiring of the keyboard pcb, but usually is limited by the usb HID protocol.
Generally speaking, usb can carry up to 6 keys of information in a single packet (I don’t remember off the top of my head if modifiers are included). It is possible to use extended packets and encode more info (and thus allow for more than 6 keys rollover) but it requires negotiation with the os so most vendors don’t bother as generally you don’t need to be able to press more than 6 keys at the same time for most applications.
I think there’s a USB device inside the mobo to handle dumb peripherals. So it would still trigger an interrupt, but it wouldn’t make it to the CPU. The USB keyboard controller would handle it and cache the strokes locally until polled by the CPU.
I would expect that any motherboard that went to the trouble of including a PS/2 port would handle it with a real hardware interrupt, because the whole point of still having those things is to avoid the latency overhead of USB.
This is x86 assembler. (Actually, looking at the register names, it’s probably x86_64. On old school x86, they were named something like al, ah (8 bit), ax (16 bit), or eax (32 bit).) Back in the old days, when you pressed a key on the keyboard, the keyboard controller would generate a hardware interrupt, which, unless masked, would immediately make the CPU jump to a registered interrupt handler, interrupting whatever else it was doing at the point. That interrupt handler would then usually save all registers on the stack, communicate with the keyboard controller to figure out what exactly happened, react to that, restore the old registers again and then jump back to where the CPU was before.
In modern times, USB keyboards are periodically actively polled instead.
Thanks for the explanation. Ironically this was the bit I didn’t know:
I was thinking the implication was that some computer had faulty interrupt handling that would smash the status register or something.
Honestly I think I’m just too old to understand memes.
The virgin USB: hey, uh, when you get a chance, uh, if it’s not too much trouble, could you, uh, put an ‘e’ there? Whenever you get the chance is fine
The chad PS/2: THE USER SAID E.
You’re right, but rax is amd64.
I think there were a few early amd64 systems with genuine ps2, and I think you can still get one, but it wasn’t common, and honestly it’s probably usb->ps/2.
To be a pedantic asshole: mov eax, ecx? Unless you’re commenting on the insanity of interrupt driven i/O in the modern age of high performance, deep-pipelined superscalar OOO cores.
does that mean though that if I connect a PS/2 keyboard or mouse to my relatively modern computer (a “gamer” motherboard made ~6 years ago) 's PS/2 port, that it’ll still trigger such an interrupt?
The other commenter is on the right track but the chip controls both USB and PS/2 as well as others;
In the 90s and 2000s, for x86 machines, slower I/O was handled by a chip called the Southbridge which worked in conjunction with a chip called the Northbridge that handled faster I/O like IDE and PCI. Later these were integrated into a single chip and, as of recent processor generations, into the processor itself.
AFAIK ghosting and key rollover are issues when using PS/2 but it can offer some milliseconds off latency when used in high cpu games.
I think it’s more of an issue for USB keyboards than PS/2 keyboards.
They are wholly independent from the protocol or interface. Ghosting is an electrical issue that is a result of keyboards being a bunch of switches arranged in a matrix. It makes the keyboard’s controller register an extra keypress in certain conditions. Nothing to do with how the thing communicates with the host computer.
Key rollover issues can be related to ghosting. The limit for it is once again the keyboard’s design at the circuit level, not its communication protocol.
Really they’re both related to how cheaply built the keyboard is. That’s the only thing.
Ghosting entirely depends on the wiring of the keyboard pcb. Key rollover can depend on the wiring of the keyboard pcb, but usually is limited by the usb HID protocol.
Generally speaking, usb can carry up to 6 keys of information in a single packet (I don’t remember off the top of my head if modifiers are included). It is possible to use extended packets and encode more info (and thus allow for more than 6 keys rollover) but it requires negotiation with the os so most vendors don’t bother as generally you don’t need to be able to press more than 6 keys at the same time for most applications.
I think there’s a USB device inside the mobo to handle dumb peripherals. So it would still trigger an interrupt, but it wouldn’t make it to the CPU. The USB keyboard controller would handle it and cache the strokes locally until polled by the CPU.
I would expect that any motherboard that went to the trouble of including a PS/2 port would handle it with a real hardware interrupt, because the whole point of still having those things is to avoid the latency overhead of USB.
Largely an urban legend. The internal electronics of the keyboard/mouse matter more than the protocol for end to end latency.
There are USB keyboards that beat a PS/2 one, at just 125 Hz polling. 1000 Hz polling pulls ahead even more.
https://www.youtube.com/watch?v=eEswl6kZq5k&t=650
I had to write a mini os and it handled keyboard interrupts. Certainly made it make a lot more since after writing it for my uni course