On Sat, 2007-05-26 at 15:35 +0200, Uwe Hermann wrote:
On Wed, May 23, 2007 at 02:46:53PM +0200, Philipp Degler wrote:
On Wednesday 23 May 2007 01:59:01 Jeremy Jackson wrote:
I'm working on a Tyan S2865 board, similar except the SuperIO chip. I'm having trouble with only one of four PCI slots generating interrupts. I wonder if you can try a PCI network card or something and post your status?
actually the asus board has 3 pci slots. I plugged in a 3com network card. The device was found and the module loaded but the assignment of an interrupt failed. I would assume this is an mptable issue.
Does the "irqpoll" kernel command line option help?
unfortunately not. I'm convinced the mptable is correct, as I have verified the physical wiring of the PCI slots with a multimeter. Based on the slot that has all 4 PIRQ[A-D] working, these should work.
If interrupts were being generated, but mptable was wrong, wouldn't the kernel complain about a spurious interrupt (and hence work with irqpoll)?
I think my next step will be to attach a counter/timer to the PIRQ lines and see what's going on.
As for the HOWTO, that's basically what I've done, and the IRQ steering can't be related because of the zero ohm connection between the PIRQ pins of all 4 slots.
What would be really helpful is to hear a report of someone with a CK804 that has more than 1 working PCI slot.
Also I'm looking for a PCI card that I can easily check interrupt pending status from a register, PC parallel printer port comes to mind, I'll have to look at the Linux kernel driver source.
Regards,
Jeremy
Jürgen Beisert has written a really nice HOWTO which may be useful, see http://linuxbios.org/Some_help_to_reverse_engineer_a_valid_interrupt_routing...