From: "Matthias Scheler" <tron%zhadum.de@localhost>
On Sat, Mar 19, 2005 at 10:17:53AM -0000, Christian Limpach wrote:It handles FPU errors, the context switching part of FPU exception handlingis in npxdna_*. Both types of exceptions get always passed up to the domain causing the exception. I would try removing the panic and alsocommenting out the bus_space_write which is just following it. The rest ofthe code looks like it could just work...Who will clear the interrupt latch? The Hypervisor?
Clearing the interrupt latch is only necessary when you're running on a 386 or older where you (choose to) get FPU errors on interrupt 13. On newer chips, you (choose to) get FPU errors on trap #16 and there's no interrupt latch to clear. I wonder what the bus_space_write actually writes to on NetBSD/i386...
Xen always acknowledges hardware interrupts, the guest only needs to acknoweledge the virtual interrupt it gets on the event channel which is used to propagate the hardware interrupt to the guest (or guests in case of interrupt line sharing with drivers in different domains -- really not a sane setup though...).
christian