On nested VMRUN, KVM ensures AVIC is inhibited by requesting KVM_REQ_APICV_UPDATE, triggering a check of inhibit reasons, finding APICV_INHIBIT_REASON_NESTED, and disabling AVIC. However, when KVM_SET_NESTED_STATE is performed on a vCPU not in guest mode with AVIC enabled, KVM_REQ_APICV_UPDATE is not requested, and AVIC is not inhibited. Request KVM_REQ_APICV_UPDATE in the KVM_SET_NESTED_STATE path if AVIC is active, similar to the nested VMRUN path. Fixes: f44509f849fe ("KVM: x86: SVM: allow AVIC to co-exist with a nested guest running") Cc: stable@vger.kernel.org Signed-off-by: Yosry Ahmed --- After the dust settles from all the inflight changes touching this area, we really need to reuse enter_svm_guest_mode() in svm_set_nested_state() imo. There are a few differences that will need to be addressed: 1. nested_vmcb02_prepare_save() needs to be skipped. 2. nested_svm_copy_common_state() may need to be skipped (or honestly, just folded into nested_vmcb02_prepare_save()?). 2. The CR3 value passed to nested_svm_load_cr3() needs to change. 3. Need to check if calling nested_svm_hv_update_vm_vp_ids() is okay/needed (might be another bug actually). I think mostly likely we'll pass a boolean arg to enter_svm_guest_mode(), something like use_cached_save, and key 1 to 3 off of it. I don't like it (smells like from_vmrun), but it's the simplest way to avoid bugs like this in the future. --- arch/x86/kvm/svm/nested.c | 3 +++ 1 file changed, 3 insertions(+) diff --git a/arch/x86/kvm/svm/nested.c b/arch/x86/kvm/svm/nested.c index de90b104a0dd5..82e512631e514 100644 --- a/arch/x86/kvm/svm/nested.c +++ b/arch/x86/kvm/svm/nested.c @@ -1922,6 +1922,9 @@ static int svm_set_nested_state(struct kvm_vcpu *vcpu, svm->nested.force_msr_bitmap_recalc = true; + if (kvm_vcpu_apicv_active(vcpu)) + kvm_make_request(KVM_REQ_APICV_UPDATE, vcpu); + kvm_make_request(KVM_REQ_GET_NESTED_STATE_PAGES, vcpu); ret = 0; out_free: -- 2.53.0.414.gf7e9f6c205-goog