[1/3] KVM: x86: make vendor code check for all nested events

Right now, the VMX preemption timer is special cased via the
hv_timer_pending, but the purpose of the callback can be easily
extended to observing any event that can occur only in non-root
mode.  Interrupts, NMIs etc. are already handled properly by
the *_interrupt_allowed callbacks, so what is missing is only
MTF.  Check it in the newly-renamed callback, so that
kvm_vcpu_running's call to kvm_check_nested_events
becomes redundant.

Cc: stable@vger.kernel.org
Reported-by: Maxim Levitsky <mlevitsk@redhat.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
---
 arch/x86/include/asm/kvm_host.h | 2 +-
 arch/x86/kvm/vmx/nested.c       | 7 ++++++-
 arch/x86/kvm/x86.c              | 8 ++++----
 3 files changed, 11 insertions(+), 6 deletions(-)

On Fri, Apr 29, 2022, Paolo Bonzini wrote:
> On 4/29/22 19:03, Sean Christopherson wrote:
> > This doesn't even compile...
> > 
> > arch/x86/kvm/vmx/nested.c: In function ‘vmx_has_nested_events’:
> > arch/x86/kvm/vmx/nested.c:3862:61: error: ‘vmx’ undeclared (first use in this function)
> >   3862 |         return nested_vmx_preemption_timer_pending(vcpu) || vmx->nested.mtf_pending;
> >        |                                                             ^~~
> > arch/x86/kvm/vmx/nested.c:3862:61: note: each undeclared identifier is reported only once for each function it appears in
> >    CC [M]  arch/x86/kvm/svm/svm_onhyperv.o
> > arch/x86/kvm/vmx/nested.c:3863:1: error: control reaches end of non-void function [-Werror=return-type]
> >   3863 | }
> >        | ^
> > cc1: all warnings being treated as errors
> >    LD [M]  arch/x86/kvm/kvm.o
> 
> Yeah, it doesn't.  Of course this will need a v2, also because there are
> failures in the vmx tests.

Heh, I suspected there would be failures, I was about to type up a response to
patch 3.  MTF is subtly relying on the call from kvm_vcpu_running() to inject
the event.

From: Sean Christopherson <seanjc@google.com>
Date: Fri, 29 Apr 2022 17:30:54 +0000
Subject: [PATCH] KVM: nVMX: Make an event request when pending an MTF nested
 VM-Exit

Set KVM_REQ_EVENT when MTF becomes pending to ensure that KVM will run
through inject_pending_event() and thus vmx_check_nested_events() prior
to re-entering the guest.  MTF currently works by virtue of KVM's hack
that calls kvm_check_nested_events() from kvm_vcpu_running(), but that
hack will be removed in the near future.

Fixes: 5ef8acbdd687 ("KVM: nVMX: Emulate MTF when performing instruction emulation")
Cc: stable@vger.kernel.org
Signed-off-by: Sean Christopherson <seanjc@google.com>
---
 arch/x86/kvm/vmx/vmx.c | 6 ++++--
 1 file changed, 4 insertions(+), 2 deletions(-)

diff --git a/arch/x86/kvm/vmx/vmx.c b/arch/x86/kvm/vmx/vmx.c
index d58b763df855..4c635bc08105 100644
--- a/arch/x86/kvm/vmx/vmx.c
+++ b/arch/x86/kvm/vmx/vmx.c
@@ -1577,10 +1577,12 @@ static void vmx_update_emulated_instruction(struct kvm_vcpu *vcpu)
 	 */
 	if (nested_cpu_has_mtf(vmcs12) &&
 	    (!vcpu->arch.exception.pending ||
-	     vcpu->arch.exception.nr == DB_VECTOR))
+	     vcpu->arch.exception.nr == DB_VECTOR)) {
 		vmx->nested.mtf_pending = true;
-	else
+		kvm_make_request(KVM_REQ_EVENT, vcpu);
+	} else {
 		vmx->nested.mtf_pending = false;
+	}
 }

 static int vmx_skip_emulated_instruction(struct kvm_vcpu *vcpu)

base-commit: 39aa5903e8c407e5128c15aeabb0717b275b007e
--

On 4/27/22 22:40, Maxim Levitsky wrote:
> 
> Wasn't able to test on my intel laptop, I am getting out of sudden in qemu:
> 
> 'cpuid_data is full, no space for cpuid(eax:0x8000001d,ecx:0x3e)'

Sending a patch soon, it's a QEMU bug that we have to work around.

Paolo