On 07/29/2010 09:51 AM, Gleb Natapov wrote:
On Tue, Jul 27, 2010 at 04:19:35PM +0300, Avi Kivity wrote:
Instead of blindly attempting to inject an event before each guest entry,
check for a possible event first in vcpu->requests. Sites that can trigger
event injection are modified to set KVM_REQ_EVENT:
- interrupt, nmi window opening
- ppr updates
- i8259 output changes
- local apic irr changes
- rflags updates
- gif flag set
- event set on exit
What about userspace irq chip? Does it work with this patch? I don't see
that you set KVM_REQ_EVENT on ioctl(KVM_INTERRUPT) for instance and
vcpu->run->request_interrupt_window should be probably checked out of
if (KVM_REQ_EVEN).
Right.
It looks like with this approach we scatter irq
injection logic all over the code instead of having it in one place.
One place is better, but it means we have to poll all event types on
every entry. We can go back to one place by having a mini-API for
events (extending the kvm_queue_exception family) that would take care
of the details.
--
error compiling committee.c: too many arguments to function
--
To unsubscribe from this list: send the line "unsubscribe kvm" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
