Skip to contentSkip to conversation

grundler

parisc-linux-cvs Philipp Rumpf 4 participants 9 messages Sep 2, 1999 – Sep 4, 1999 1 message not in the archive

lore
Share & tools
lore
#2 lore raw
Quoted 4 lines · Philipp Rumpf
Quoted 2 lines

Modified Files:
pci.c

- removed support for more than one bus

Not quite - removed support for more than one type of PCI bus adapter. PA platforms only support one type of PCI bus adapter at a time. We could link in more than on PCI bus driver but only one will ever get used. (At least that's my goal).

Quoted 1 line

we don't want that, do we ?

Rather we don't need it.
If you want it you can have it. :^)

Quoted 1 line

That's it for today, I promise.

ditto.

thanks,
grant

Grant Grundler
Communications Infrastructure Computer Operations
+1.408.447.7253

#3 lore raw
Quoted 2 lines

Not quite - removed support for more than one type of PCI bus adapter. PA platforms only support one type of PCI bus adapter at a time.

Sorry ? Alex told me the exact opposite when I asked him and I relied on that.

So: Are there parisc boxen with more than one type of interface to PCI-like busses ? For example, can I take a Dino-on-a-card card and put it into a box which does not use Dino as native-to-PCI bridge ?

Philipp Rumpf

#4 lore raw
Quoted 4 lines · Philipp Rumpf
Quoted 2 lines

Not quite - removed support for more than one type of PCI bus adapter. PA platforms only support one type of PCI bus adapter at a time.

Sorry ? Alex told me the exact opposite when I asked him and I relied on that.

I don't *think* I ever said that, but if I did, I was wrong.

Quoted 3 lines

So: Are there parisc boxen with more than one type of interface to PCI-like busses ? For example, can I take a Dino-on-a-card card and put it into a box which does not use Dino as native-to-PCI bridge ?

And does the dino-on-a-card have it's own IODC on the card?

- Alex

Signature · 3 lines
Alex deVries
Vice President of Engineering
The Puffin Group
#9 lore raw

Re: PCI-like busses

In reply to Grant Grundler
Quoted 5 lines
Quoted 2 lines

So: Are there parisc boxen with more than one type of interface to PCI-like busses?

B/C/D class boxen haben PCI u. EISA (u. GSC) "slots". Since learning EISA/ISA busses are "PCI-like", the answer is "yes".

Of course, strictly speaking we could separate between pci busses (where we have configuration space accesses) and isa-like busses (where we don't). I don't see any advantage. For port accesses, we don't have to separate.

Quoted 3 lines

And if this answer is right, then I'll put back the "multi-bus type" support in pci.h. I don't get how EISA would use these services (perhaps just the IRQ stuff?) but I'll trust you on that.

Remember I still want to remove the IRQ handling out of struct pci_bus_ops.

Quoted 2 lines

Yes - but I'm only aware of K/T class "form factor" (shape/size). These won't normally fit in the A/B/C/D class boxes which also have

The Dino-on-a-card I have in the A180C is quite long, yes.

Quoted 2 lines

(Next week I'll see if I can wedge the K-class one into a workstation anyway)

Don't destroy any hardware ;)

Philipp Rumpf

#8 lore raw

[parisc-linux] PCI-like busses

In reply to Philipp Rumpf
Quoted 5 lines · Philipp Rumpf
Quoted 2 lines

Not quite - removed support for more than one type of PCI bus adapter. PA platforms only support one type of PCI bus adapter at a time.

Sorry ? Alex told me the exact opposite when I asked him and I relied on that
.

Ok.

Quoted 2 lines

So: Are there parisc boxen with more than one type of interface to PCI-like busses?

B/C/D class boxen haben PCI u. EISA (u. GSC) "slots". Since learning EISA/ISA busses are "PCI-like", the answer is "yes". And if this answer is right, then I'll put back the "multi-bus type" support in pci.h. I don't get how EISA would use these services (perhaps just the IRQ stuff?) but I'll trust you on that.

Quoted 2 lines

For example, can I take a Dino-on-a-card card and put it into a box which does not use Dino as native-to-PCI bridge ?

Yes - but I'm only aware of K/T class "form factor" (shape/size). These won't normally fit in the A/B/C/D class boxes which also have GSC slots. Any other HP person know if any card-mode Dino's were made for workstations?

(Next week I'll see if I can wedge the K-class one into a workstation anyway)

grant

Grant Grundler
Communications Infrastructure Computer Operations
+1.408.447.7253

#5 lore raw

[parisc-linux] Card-mode Dino

In reply to Alex deVries
Quoted 1 line · Alex deVries

And does the dino-on-a-card have it's own IODC on the card?

No. Dino implements the PA I/O architected registers but no IODC. I suspect the problem is it's really a bus converter and that needs PDC support - IODC entry points don't apply to this. Can anyone confirm this?

grant

#6 lore raw

Re: [parisc-linux] Card-mode Dino

Quoted 5 lines
Quoted 1 line

And does the dino-on-a-card have it's own IODC on the card?

No. Dino implements the PA I/O architected registers but no IODC. I suspect the problem is it's really a bus converter and that needs PDC support - IODC entry points don't apply to this. Can anyone confirm this?

Can't confirm or deny, but it is possible (we've done it at least once) to build a bus converter that does not require PDC help, just architecture knowledge.

lamont

#7 lore raw

Re: [parisc-linux] Card-mode Dino

Quoted 9 lines · LaMont Jones
Quoted 5 lines
Quoted 1 line

And does the dino-on-a-card have it's own IODC on the card?

No. Dino implements the PA I/O architected registers but no IODC. I suspect the problem is it's really a bus converter and that needs PDC support - IODC entry points don't apply to this. Can anyone confirm this?

Can't confirm or deny, but it is possible (we've done it at least once) to build a bus converter that does not require PDC help, just architecture knowledge.

Let me add some depth to my suspicion.
Platform IO space (which Dino "translates" 1:1 to PCI MMIO space) is assigned exclusively by PDC on platforms with GSC busses. The OS can't assign I/O space addresses to devices unless (a) knows where I/O space is (could make assumptions about this), (b) exactly which parts have been used by other bus converters or devices, and (c) which parts are reserved for "special" devices (eg graphics)?

I suspect IODC can't provide this sort of functionality. Let me rephrase my question: does anyone know if architected PDC interfaces provide enough information to satisfy a/b/c above?

thanks,
grant

Grant Grundler
Communications Infrastructure Computer Operations
+1.408.447.7253