Subject: Re: ACPI Interrupt Problems with 3.99.22 (and 3.99.21)
To: Martin Husemann <martin@duskware.de>
From: Thor Lancelot Simon <tls@rek.tjls.com>
List: port-amd64
Date: 08/06/2006 21:14:28
On Sun, Aug 06, 2006 at 01:37:57PM +0200, Martin Husemann wrote:
> On Sun, Aug 06, 2006 at 05:30:36AM +0000, Mathieu Boespflug wrote:
> > I'm experiencing the exact same problem as Ed on a D600 Dell laptop. Has any
> > workaround or fix to this issue been found yet? I can't find any bug reports
> > related to this.
> 
> Maybe not the same, but could be related:
> 
>   http://www.netbsd.org/cgi-bin/query-pr-single.pl?number=33922
> 
> I think there is some basic problem with shared interrupts.

I have seen a completely disastrous problem with shared interrupts in
amd64 ACPI kernels, where devices with the same legacy IRQ but *different*
ioapic interrupts -- in fact, interrupts on *different ioapics* -- get
each others' interrupts.

I cannot figure out how this happens, but it happens only when the devices
have the same ISA IRQ, in an ACPI/ioapic kernel that shouldn't use the ISA
IRQ to route interrupts at all.  Sigh.

Thor