Here's a refined series based on my previous RFC. I now have this
working through VFIO-based device assignment in Qemu. I've split
up the patches and tried to generalize the interface as one of
potentially a couple ways we might support level irqfds. I also
tried to document the requirements around re-asserting interrupts
and expectations for avoiding races.
The last patch in the series is just an FYI/RFC of something I
expect we're also going to need. We want to take advantage of KVM
irqchip, but not be dependent on it. That means we'll want to come
up with a way for Qemu to notify drivers of an EOI. We won't want
that interface to break when when enabling the in-kernel irqchip.
The needs are just different enough that I don't see how to combine
them together. Thanks,
Alex
---
Alex Williamson (4):
[RFC] kvm: eoi_eventfd
kvm: Extend irqfd to support level interrupts
kvm: Add missing KVM_IRQFD API documentation
kvm: Pass kvm_irqfd to functions
Documentation/virtual/kvm/api.txt | 46 ++++++++
arch/x86/kvm/x86.c | 2
include/linux/kvm.h | 18 +++
include/linux/kvm_host.h | 11 ++
virt/kvm/eventfd.c | 201 ++++++++++++++++++++++++++++++++++---
virt/kvm/kvm_main.c | 11 ++
6 files changed, 268 insertions(+), 21 deletions(-)
--
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
