[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: MVME5500 status, and on to PMC interrupts




On Feb 28, 2005, at 1:04 PM, Kate Feng wrote:

it seems that the IRQ of the  2nd PMC730  returns as a VME
interrupt 3.  My understanding is that the PMCspan  card is
a VME module featuring the DEC21150 PCI-to-PCI interface
bridge.  You do not need  to write a  newdriver for the bridge to
search for the PMC module.  However,  I think you need to program
the PMCspan VME module corrctly to access the returned address
space.


It isn't really a VME module though it takes up a VME slot. It hooks up to the PCI expansion connector on the MVME5500. You can also add a "PMCSPAN-010" for two more PMC slots. I think it is only picking up power from the VME and using the P2 connectors for user I/O.


If you look at table 2-1 of the MVME5500 Programming Manual you'll see that the PMC expansion returns "PCI AD Line AD20" and INTA# through INTD# of 12, 13, 14, 15, that second set matches the VME INTA# through INTD# assignments, but the VME uses "PCI AD LINE AD16". Maybe you understand this, I don't yet. I'm assuming there is some offset to these interrupt numbers based on how they feed into the GT64260 that get you from 0x08 to 0x48.

Also, if you look at the log I sent you and at the mvme5500 pci.c, I don't think the code is setting up that DEC21150 bus properly. I don't see any iteration over that bus once it is discovered, similar to the way the code iterates over new busses in the "shared/pci" directory, so I don't think things are properly enabled for the boards on the other side of the bridge. By brute force looking at higher PCI bus numbers I am seeing it after startup but I don't think it is setup and thus the memory problem when I try to look at registers. Note that access to the first PMC730 and the PMC341 are logged during startup, but the second PMC730 never is, the first you see it is after my code is running.


It will help if one has a programmer's manual of the PMCspan module, which I am still working on.


I found a copy on a British Motorola site, but now I don't see it. I'll send a copy to you off-line.


Peter

Peter Dufault
HD Associates, Inc.