how to tell Linux *not* to share IRQs ?
Quoted 21 lines · Mark H. Wood
On Wed, 8 Aug 2001, Andrew McNamara wrote:
[snippage]Quoted 8 lines
The problem is largely historical - each interrupt traditionally had a physically line associated with it, and lines on your backplane were a limited resource.
If you were to do it again these days, you might have some sort of shared serial bus, so devices could give detailed data to the cpu (not only to uniquely identify the interrupting device, but also identify sub-devices - say a USB peripheral).
See for example "vectored interrupts" on the PDP10. The device driver tells the device where the driver's ISR is, and when the device interrupts, it puts that address on the bus. The interrupt logic jumps directly to the ISR, which "knows" it is the only driver that would be interested in this interrupt. (You could set up a jump table if you wanted to, so that each device of the same type could identify itself uniquely, but that typically wasn't a big problem in '10 installations where multiples were most likely in a PDP11 on the other side of a DTE20, or Massbus devices on a single RH20.)
Quoted 1 line
Apparently this idea is now so old that it is new. :-)
Yeah - I believe the same was possible on the Z80 - though I'd have to go read the manual to be certain.
regards,
Per Jessen, Zurich
Windows 2001: "I'm sorry Dave ... I'm afraid I can't do that."
regards,
Per Jessen, Zurich
http://www.enidan.com - home of the J1 serial console.
Windows 2001: "I'm sorry Dave ... I'm afraid I can't do that."
Quoted 2 lines · Per Jessen
Yeah - I believe the same was possible on the Z80 - though I'd have to go read the manual to be certain.
It is. When I was reading the description in the previous e-mail, I was thinking, "that sounds a lot like the Z80".
I missed the beginning of this thead, what did the so old its new mean? Is there a new chip supporting vectored interupts? Or are they just being simulated in software?
-Chris
Signature · 3 lines
Two penguins were walking on an iceberg. The first penguin said to the second, "you look like you are wearing a tuxedo." The second penguin said, "I might be..." --David Lynch, Twin Peaks
Quoted 23 lines · Per Jessen
Quoted 21 lines · Mark H. Wood
On Wed, 8 Aug 2001, Andrew McNamara wrote:
[snippage]Quoted 8 lines
The problem is largely historical - each interrupt traditionally had a physically line associated with it, and lines on your backplane were a limited resource.
If you were to do it again these days, you might have some sort of shared serial bus, so devices could give detailed data to the cpu (not only to uniquely identify the interrupting device, but also identify sub-devices - say a USB peripheral).
See for example "vectored interrupts" on the PDP10. The device driver tells the device where the driver's ISR is, and when the device interrupts, it puts that address on the bus. The interrupt logic jumps directly to the ISR, which "knows" it is the only driver that would be interested in this interrupt. (You could set up a jump table if you wanted to, so that each device of the same type could identify itself uniquely, but that typically wasn't a big problem in '10 installations where multiples were most likely in a PDP11 on the other side of a DTE20, or Massbus devices on a single RH20.)
Quoted 4 lines
Quoted 1 line
Apparently this idea is now so old that it is new. :-)
Yeah - I believe the same was possible on the Z80 - though I'd have to go read the manual to be certain.
And with the m68k, too. In fact, that's what it always tries to do. It runs an interrupt acknowledge cycle when an interrupt is raised so that the interrupting device can respond with an 8 bit vector number which is used to load the address of the handler from the vector table.
When the /AVEC (autovector) input is true it generates a vector number automatically depending on the interrupt level. So it supports both modes of operation.
So it's not like the PDP-10 was the only to do that. If only a better design would have become standard instead of the ugly IBM PC architecture...
Signature · 1 line
Andreas E. Bombe <[email protected]> DSA key 0x04880A44
Followup to: <[email protected]> By author: Chris Meadors <[email protected]> In newsgroup: linux.dev.kernel
Quoted 6 lines · Per Jessen
Quoted 3 lines
Yeah - I believe the same was possible on the Z80 - though I'd have to go read the manual to be certain.
Quoted 7 lines
It is. When I was reading the description in the previous e-mail, I was thinking, "that sounds a lot like the Z80".
I missed the beginning of this thead, what did the so old its new mean? Is there a new chip supporting vectored interupts? Or are they just being simulated in software?
The x86 supports vectored interrupts, too (at least in the same sense the Z80 does.) There are 256 interrupt vectors possible, 32 of which are reserved (although that didn't keep IBM from foolishly use them rather than the 224 that weren't...) However, the PIC architecture in the PC doesn't use this -- it lets the vectors be generated by the PIC; the standard dual PIC allows only for 15 vectors and the APIC something like 24, and it's all controlled by which input line gets pulled.
Sad, isn't it?
-hpa
Signature · 3 lines
<[email protected]> at work, <[email protected]> in private! "Unix gives you enough rope to shoot yourself in the foot." http://www.zytor.com/~hpa/puzzle.txt <[email protected]>