On Sun, Jun 5, 2011 at 5:14 PM, Jan Kiszka <jan.kiszka at web.de> wrote: > On 2011-06-05 17:10, Avi Kivity wrote: >> On 06/05/2011 05:58 PM, Jan Kiszka wrote: >>> > >>> > ?Note that with KVM_SET_GSI_ROUTING (bad name for ARM...) we can even >>> > ?choose if an irq line is connected to a kernel-emulated interrupt >>> > ?controller or to the core's irq input. >>> >>> Makes some sense: Add KVM_IRQ_ROUTING_CPU, and kvm_irq_routing_entry's >>> union would require some struct kvm_irq_routing_cpu containing the >>> target identifier. >> >> Right. ?Note it would be the default, so no need to implement >> KVM_SET_GSI_ROUTING just yet. >> >> An additional advantage is that this is a vm ioctl, not a vcpu ioctl, so >> no need to interrupt the vcpu thread in userspace in order to queue an >> interrupt. ?Of course, it still happens in the kernel, but it's easier >> for userspace to implement its device model this way. > > So supporting this over existing archs would have some benefits as well, > though a bit smaller if in-kernel irqchip is already implemented. > Could you elaborate what you mean here? I'm not really following. Are you suggesting a new arch-generic interface? (Pardon my ignorance).
