I really honestly don't understand what is going on other than there are an awful lot of threads involving IIsi and pseudo-slots and wanting more. And I was doing a bit of reading in the Guide to the Macintosh Family Hardware and Designing Cards and Drivers for the Macintosh Family that may or may not be relevant to the idea of a IIsi and pseudo-slots and this seemed like as good a place as any to talk about it, because there's still a lot I don't understand about what I was reading.
AFAIK, it should be reasonably (as these things go) straightforward to get 3 nubus slots out of an 030 PDS slot, since as you've pointed out, 3 pseudo-slots. The 030 PDS has 3 interrupt lines, one for each pseudo-slot, so all is well. There's no interrupt lines to go beyond 3, and you can't really share any of the existing interrupt lines because those get OR'd together to generate a SLOTIRQ signal and toggle bits on VIA2 which indicate which pseudo-slot generated the interrupt, which in turn is used to determine which interrupt handler should be run. So, assuming >3 pseudo-slots "work", there's no interrupt for them. Do you have an interruptless nubus card in mind to use with such a slot? Looking over the Nubus documentation, I'm not entirely clear how interrupts work with bus master cards. It looks like the /NMRQ is used as an interrupt for non-master cards, or at least cards currently operating in non-master mode. Is it possible to implement nubus (as implemented on Macintosh II computers, since Designing Cards and Drivers specifically mentions other ways of generating interrupts according to the nubus spec, but only dedicated /NMRQ lines are supported by the Macintosh II family) without any interrupt hardware at all?
Now if you wanted the nubus slots and a non-pseudo-slot 030 PDS card that used the 030 PDS reserved memory area and didn't use anything else, that seems feasible...
When the Slot Manager runs from ROM, it iterates over the nubus decl rom address locations looking for a valid decl rom, and parsing it, as best I can gather. I haven't seen anything specifically indicating the SE/30 or IIsi ROMs do or don't iterate over the addresses for slots C and D (E is taken by the video "card" in the SE/30 at least). In fact, it's likely they do iterate over those "blank" address ranges since the SE/30's video card is a pseudo-card complete with declrom that should be parsed by the Slot Manager like everything else.
So, I guess my questions are:
Does it make sense to have a nubus card with no way to interrupt the host processor? Or do you have a solution to the lack of interrupts problem?
In the quest for additional pseudo-slots, do you have any indication they aren't being probed?
I guess an 030 PDS to Nubus with the Nubus ID1-ID3 lines set to C or D and /NMRQ not connected would be a test?
Anyway, I hope that makes some amount of sense, I'm getting blurry eyed reading these books.
AFAIK, it should be reasonably (as these things go) straightforward to get 3 nubus slots out of an 030 PDS slot, since as you've pointed out, 3 pseudo-slots. The 030 PDS has 3 interrupt lines, one for each pseudo-slot, so all is well. There's no interrupt lines to go beyond 3, and you can't really share any of the existing interrupt lines because those get OR'd together to generate a SLOTIRQ signal and toggle bits on VIA2 which indicate which pseudo-slot generated the interrupt, which in turn is used to determine which interrupt handler should be run. So, assuming >3 pseudo-slots "work", there's no interrupt for them. Do you have an interruptless nubus card in mind to use with such a slot? Looking over the Nubus documentation, I'm not entirely clear how interrupts work with bus master cards. It looks like the /NMRQ is used as an interrupt for non-master cards, or at least cards currently operating in non-master mode. Is it possible to implement nubus (as implemented on Macintosh II computers, since Designing Cards and Drivers specifically mentions other ways of generating interrupts according to the nubus spec, but only dedicated /NMRQ lines are supported by the Macintosh II family) without any interrupt hardware at all?
Now if you wanted the nubus slots and a non-pseudo-slot 030 PDS card that used the 030 PDS reserved memory area and didn't use anything else, that seems feasible...
When the Slot Manager runs from ROM, it iterates over the nubus decl rom address locations looking for a valid decl rom, and parsing it, as best I can gather. I haven't seen anything specifically indicating the SE/30 or IIsi ROMs do or don't iterate over the addresses for slots C and D (E is taken by the video "card" in the SE/30 at least). In fact, it's likely they do iterate over those "blank" address ranges since the SE/30's video card is a pseudo-card complete with declrom that should be parsed by the Slot Manager like everything else.
So, I guess my questions are:
Does it make sense to have a nubus card with no way to interrupt the host processor? Or do you have a solution to the lack of interrupts problem?
In the quest for additional pseudo-slots, do you have any indication they aren't being probed?
I guess an 030 PDS to Nubus with the Nubus ID1-ID3 lines set to C or D and /NMRQ not connected would be a test?
Anyway, I hope that makes some amount of sense, I'm getting blurry eyed reading these books.