Re: IRQ-Routing for 5.3-BETA

Bruce M Simpson wrote:
> On Mon, Nov 08, 2004 at 09:25:21PM -0800, Nate Lawson wrote:
> 
>>I MFC'd the change to RELENG_5 as well so if you can test there or on 
>>6-current, you should fine that the card works as before.  The change 
>>was to allow the SCI (irq 9 on PIC systems) to always be used for 
>>devices, even if the ACPI _PRS setting for the device doesn't explicitly 
>>allow it.
> 
> 
> Whoah. Thanks for this. I remember discussing with Warner around the time
> of 4.5-STABLE that I had problems with interrupt routing on laptops.
> 
> So I hope this is not a stupid question...
> 
> The specific problem I had was that the OLDCARD code saw that the PCMCIA
> device I was trying to plug in did not allow irq 9 to be used as the
> function interrupt. The control interrupt of course was irq 9. I hacked
> OLDCARD somewhat to allow me to assign the function and control interrupts
> separately so I could get my PCMCIA smart card reader to work.
> 
> I wonder if it's still possible to do something like this in these days
> of ACPI and cardbus, or is it no longer necessary (can I lie about which
> IRQ is routed to the pccard/isa device?)

You can override which irqs are assigned a given PCI device (including 
cardbus) by setting tunables for ACPI and PCI (see the manpages for 
syntax).  However, with cardbus, the irq is assigned to the slot, not 
the device so once you route say irq 10 to a slot, all devices plugged 
into it get irq 10.  On most cardbus chips, a single controller has an 
irq so both slots by design have the same irq.

-Nate