Update: Programming PCI-PCI bridge

Brian Topping (topping@mauswerks.com)
Mon, 8 Apr 1996 18:55:14 -0700

For those wondering what ever happened to this problem, here is the scoop.
I am not much of a board hacker on x86, so feel free to elucidate here...

The Znyx multiport boards were designed a couple of years ago when there
was little support for multifunction boards (and something to do with INTA)
in the standard BIOSs. Znyx decided to wire all of the IRQ's to the same
interrupt, so what I was seeing was something that was expected and was a

I guess newbies don't recognize nonstandard, so they bump their heads on it
repeatedly until someone points out the doorframe is a little lower than
they would expect. (Yes, my head hurts now ;-)

I have a couple of printk's to clean up and I should be able to coordinate
the release of the changes with Donald.

Thanks for all of your help in getting me this far!


>Hi all,
>Does anyone know how I can get the kernel to program a PCI-PCI bridge chip
>to force all of the PCI devices on the far end of the bridge to share an
>interrupt? I have a Znyx 4 port Ether board with a DEC 21050 PCI-PCI
>bridge, but I can't find much in the kernel source in this regard.
>The problem I am up against is the driver probe routine is called with the
>results of the BIOS mapping of the IRQs, with has a different IRQ for each
>of the NIC's. In reality though, the IRQ that the NIC's raise is all the
>same one.