On 2011-01-20 21:02, Blue Swirl wrote: > I think KVMState was designed to match KVM ioctl interface: all stuff > that is needed for talking to KVM or received from KVM are there. But > I think this shouldn't be a design driver. Agreed. The nice cleanup would probably be the complete assimilation of KVMState by something bigger of comparable scope. If a machine was brought up with KVM support, every device that refers to this machine (as it is supposed to become part of it) should be able to use KVM services in order to accelerate its model. > > If the only pieces of kvm_state that are needed by the devices are > irqchip_in_kernel, pit_in_kernel and many_ioeventfds, the problem of > passing kvm_state to devices becomes very different. Each of these are > just single bits, affecting only a few devices. Perhaps they could be > device properties which the board level sets when KVM is used? Forget about the static capabilities for now. The core of kvm_state are handles that enable you to use KVM services and maybe state fields that have machine scope (unless we find more local homes like a memory object). Those need to be accessible by the kvm layer when servicing requests of components that are related to that very same machine. Jan
Attachment:
signature.asc
Description: OpenPGP digital signature
