DragonFly users List (threaded) for 2005-02
Re: xl device doesn't work with ACPI enabled under -current and -stable
:On Thu, Feb 03, 2005 at 10:40:13AM -0800, Matthew Dillon wrote:
:> Well, ok, but there's got to be some fundamental issue here that we
:> are missing, because that patch was utterly trivial. Shouldn't
:> the code path be the same between the old PCIB_ROUTE_INTERRUPT and
:> the new PCI_ROUTE_INTERRUPT ? Or is it that it wasn't doing
:> anything at all before and now it is trying to actually change the
:> interrupt route from the default? That's what I don't get.
:It wasn't doing anything at all before.
Which means it wasn't routing cbb0 before. That's what is happening.
The default irq must be 11 but when it routed cbb0 it changed it to 3.
Since xl0 is on the same pin as cbb0, xl0's irq became 3 too.
I think this is a bug in our code somewhere and not a bug in the ACPI ASL.
When we route an interrupt pin we need to be aware that it may set the
IRQ for another device.
The question is why aren't there any bootversbose messages indicating
the routing of XL0's interrupt ? The XL device makes the call right
after it allocates its ports, but there aren't any bootverbose messages
indicating that the PCIB subsystem even saw the request.
pcib_alloc_resource() doesn't do any special handling of SYS_RES_IRQ,
it just passes it through to bus_generic_alloc_resource(). Who is
responsible for allocating the IRQ ?