kvm: nVMX: move nested events check to kvm_vcpu_running
authorPaolo Bonzini <pbonzini@redhat.com>
Mon, 19 Dec 2016 14:23:54 +0000 (15:23 +0100)
committerPaolo Bonzini <pbonzini@redhat.com>
Wed, 15 Feb 2017 13:54:33 +0000 (14:54 +0100)
commit0ad3bed6c5ec6dbb093a26802c85088a85fb9757
tree1889f7af427e86b290c5b7b849d6e98b87eaa321
parent967235d320329e4a7a2bd1a36b04293063e985ae
kvm: nVMX: move nested events check to kvm_vcpu_running

vcpu_run calls kvm_vcpu_running, not kvm_arch_vcpu_runnable,
and the former does not call check_nested_events.

Once KVM_REQ_EVENT is removed from the APICv interrupt injection
path, however, this would leave no place to trigger a vmexit
from L2 to L1, causing a missed interrupt delivery while in guest
mode.  This is caught by the "ack interrupt on exit" test in
vmx.flat.

[This does not change the calls to check_nested_events in
 inject_pending_event.  That is material for a separate cleanup.]

Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
arch/x86/kvm/x86.c