Subject: Re: PCMCIA "ne0: device timeout"
To: Jason Thorpe <thorpej@nas.nasa.gov>
From: Manuel Bouyer <bouyer@antioche.lip6.fr>
List: port-i386
Date: 08/01/1999 16:25:45
On Sat, Jul 31, 1999 at 10:02:11AM -0700, Jason Thorpe wrote:
> In fact, on the Inspiron 3500, you want to set it to 0x8000.  As far as
> I can tell, that's the only interrupt that PCMCIA devices can use on
> this laptop[*].
> 
> [*] I actually think that IRQ 14 might be usable, too.  But that happens to
> be the "compatibility" interrupt on the PCI IDE controller.  I would like
> to know if PCI IDE "compatibility" interrupts are really edge triggered
> (like our code claims) or if they're actually level triggered (like other
> PCI interrupts).  They would have to be level triggered in order for
> PCMCIA devices to use them.

I thinks they are as ISA interrupts. In fact, I believe on PCs it is routed
internally to the PCI chipset, and is not useable by anything else (isa or
PCI). This is true at last for the INTEL PIIX* and VIA chipsets. I tried to
share IRQ 14 with a PCI network card (after hacking pciide.c a bit, to test
IRQ sharing with the PCI IDE driver) and I never got any IRQ from the network
board.
I recently got the same with irq 4: when used by the PIIX4 built-in serial
controller it can't be used by an ISA board, irqs from the isa bus are not
routed (and unfortunably I can change the I/O ports of the add-on ISA board
I needed to use but not IRQs :(

--
Manuel Bouyer <bouyer@antioche.eu.org>
--