OK, this may sound stupid, but...
I've read mingo's IO-APIC.txt and I've attempted to search the linux-smp
archives (and partially failed since linuxhq seems to be off the air right
now), and I've even spent a while trying to grok the code, but I can't figure
this out.
Where (and how) do we decide to (re)map the PCI (device,pin) pairs to IRQs when
we are using the APIC? In particular, what is happening is that I have two PCI
cards which both use INTA, and seem to be in slots (1 and 5) where these two
INTA pins are connected to the same PIRQ (yes I know I shouldn't do this
according to mingo, but that is only a solution, not an answer). At some point,
the kernel remaps one of the PCI devices to use IRQ16 for INTA. Unfortunately,
it leaves the other one (the one with the lower PCI device number) still using
IRQ11 (which is what it gets when I use "noapic" and everything works), and
since these appear to be the same wire, the machine hangs when it is expecting
an IRQ11 from this board. I've tried various pirq= options but nothing seems to
work (and I'd be surprised if it did since the problem isn't mapping a PIRQ to
an IRQ, but keeping track of the fact that some of the (device,pin) pairs are
tied together).
Can anybody explain this to me?
Oh, in case it matters, the motherboard is a Tyan 1668D (440FX, 2 PPros). The
BIOS banner is something like
Award Modular Bios 4.51PG
TYN TITAN-PRO V3.03 10/31/96
David
-
Linux SMP list: FIRST see FAQ at http://www.irisa.fr/prive/mentre/smp-faq/
To Unsubscribe: send "unsubscribe linux-smp" to [EMAIL PROTECTED]
