Old timers - What was the SysRq key used for?

[Bozz]

...?

 

[Lord Evermore]

It's used in Linux as a key combo to allow things like killing all processes or setting log levels (known as the magic sysrq key).

Just a label on a keyboard now really. The Print Screen function is what it's primarily used for now, but that's an operating system convention. If the OS was programmed for it, it could act as the letter A.

The original use, from the little information I can glean and the assumption that it means "System Request", is that it interrupts the system to allow debugging or to force keyboard input to be recognized, like when the OS is locking up. It was added as the shift function of the PrtScrn key but of course Windows doesn't care about the shift. It also appears to have been used to alert a mainframe that a terminal was still actively in use while a program was running on the mainframe, or to allow the user to issue other commands on the mainframe or pause/cancel a program running.

It apparently by default just generates a BIOS interrupt which is ignored. Software or the OS can be configured to interpret it as something, which is how Linux uses it (but you have to configure it specifically for the magic sysrq key) but for the most part it's useless.

 

[Bozz]

I did a google search prior to creating the thread and found the exact same details as you've posted

However - Linux was released around '91 if I'm not mistaken. This key has been on keyboards I've owned since 1985. Hence it may have some Microsoft DOS application use.... Apparently.

 

[AndyHui]

I remember reading that SysRq was supposed to be used in DOS...but MS never got around to actually using it.

 

[emjem]

The SysRq key is basically a system interupt technique. That is, when used in combination with other keys is causes a running application to halt and look for instructions. It is still used in Windows:

Debugger

However, here's what the old master himself had to say about it:

DAVE BRADLEY: That's not my perception of what happened.

But what, really, I think the problem was is we did a half-good job on what became the ISA bus. You had to fiddle around with DIP switches in order to plug anything new in there. And we decided the time had come to do something different.

And we wanted to solve a lot of problems with the MicroChannel. So we sort of all rolled them up into one thing and put it out there.

And a separate event happened at the same time that I think affected everyone's perception. That is, IBM changed the way they licensed patents at that time. Before that time, you essentially could get them for a song. About 1986, 1987, we changed the practice to still a very -- what I believe to be a very reasonable fee, but one that was more than we had charged before. So everybody said, okay, they're closing it up, the MicroChannel is going to cost you an arm and a leg to do it.

I don't think anybody paid more money to build a MicroChannel machine than they did to build an AT bus machine. But that's just what the perception was.

And so -- But remember, I'm an engineer. I wasn't making grand strategy. From an engineering viewpoint, we were just trying to solve a bunch of the problems that we had built into the ISA bus.

ANDY GROVE: Well, I can shed possibly a thin beam of light on that. Because the PS/2 machines were slated to be built around an internally developed CMOS 286 developed inside IBM, manufactured inside IBM, under license by Intel, and we were thanked for our services and bid an affectionate goodbye.

BILL GATES: Same here.

(Laughter.)

BILL GATES: That was the machine that had this special hypervisor in it that subsumed DOS. And that's where the SysRq key on the AT machine is, because they had bonded out the 286 to use the special debug instructions. So it was an attempt to regain the high ground.

ANDY GROVE: That's my perception.

BILL GATES: Yeah.

BRENT SCHLENDER: How dare they.

BILL GATES: It's fair game. It didn't work. But....

You can read more old stuff here:


Olden Days