KVM: nSVM: remove useless kvm_clear_*_queue
authorPaolo Bonzini <pbonzini@redhat.com>
Mon, 2 Aug 2021 12:52:50 +0000 (08:52 -0400)
committerPaolo Bonzini <pbonzini@redhat.com>
Mon, 2 Aug 2021 15:02:00 +0000 (11:02 -0400)
commitdb105fab8d141fc0d9179600c51eba0d168dad34
tree9a7391dcd6f3c4653b0a3100d4e23f0e37390dc8
parent4c72ab5aa6e0ac2a5c11f9180e1fff89d7f2d38b
KVM: nSVM: remove useless kvm_clear_*_queue

For an event to be in injected state when nested_svm_vmrun executes,
it must have come from exitintinfo when svm_complete_interrupts ran:

  vcpu_enter_guest
   static_call(kvm_x86_run) -> svm_vcpu_run
    svm_complete_interrupts
     // now the event went from "exitintinfo" to "injected"
   static_call(kvm_x86_handle_exit) -> handle_exit
    svm_invoke_exit_handler
      vmrun_interception
       nested_svm_vmrun

However, no event could have been in exitintinfo before a VMRUN
vmexit.  The code in svm.c is a bit more permissive than the one
in vmx.c:

        if (is_external_interrupt(svm->vmcb->control.exit_int_info) &&
            exit_code != SVM_EXIT_EXCP_BASE + PF_VECTOR &&
            exit_code != SVM_EXIT_NPF && exit_code != SVM_EXIT_TASK_SWITCH &&
            exit_code != SVM_EXIT_INTR && exit_code != SVM_EXIT_NMI)

but in any case, a VMRUN instruction would not even start to execute
during an attempted event delivery.

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