KVM: nVMX: Expose RDPMC-exiting only when guest supports PMU
authorLiran Alon <liran.alon@oracle.com>
Mon, 25 Mar 2019 19:09:17 +0000 (21:09 +0200)
committerPaolo Bonzini <pbonzini@redhat.com>
Tue, 16 Apr 2019 13:37:34 +0000 (15:37 +0200)
Issue was discovered when running kvm-unit-tests on KVM running as L1 on
top of Hyper-V.

When vmx_instruction_intercept unit-test attempts to run RDPMC to test
RDPMC-exiting, it is intercepted by L1 KVM which it's EXIT_REASON_RDPMC
handler raise #GP because vCPU exposed by Hyper-V doesn't support PMU.
Instead of unit-test expectation to be reflected with EXIT_REASON_RDPMC.

The reason vmx_instruction_intercept unit-test attempts to run RDPMC
even though Hyper-V doesn't support PMU is because L1 expose to L2
support for RDPMC-exiting. Which is reasonable to assume that is
supported only in case CPU supports PMU to being with.

Above issue can easily be simulated by modifying
vmx_instruction_intercept config in x86/unittests.cfg to run QEMU with
"-cpu host,+vmx,-pmu" and run unit-test.

To handle issue, change KVM to expose RDPMC-exiting only when guest
supports PMU.

Reported-by: Saar Amar <saaramar@microsoft.com>
Reviewed-by: Mihai Carabas <mihai.carabas@oracle.com>
Reviewed-by: Jim Mattson <jmattson@google.com>
Signed-off-by: Liran Alon <liran.alon@oracle.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
arch/x86/kvm/vmx/vmx.c

index 88060a6..5866e9e 100644 (file)
@@ -6856,6 +6856,30 @@ static void nested_vmx_entry_exit_ctls_update(struct kvm_vcpu *vcpu)
        }
 }
 
+static bool guest_cpuid_has_pmu(struct kvm_vcpu *vcpu)
+{
+       struct kvm_cpuid_entry2 *entry;
+       union cpuid10_eax eax;
+
+       entry = kvm_find_cpuid_entry(vcpu, 0xa, 0);
+       if (!entry)
+               return false;
+
+       eax.full = entry->eax;
+       return (eax.split.version_id > 0);
+}
+
+static void nested_vmx_procbased_ctls_update(struct kvm_vcpu *vcpu)
+{
+       struct vcpu_vmx *vmx = to_vmx(vcpu);
+       bool pmu_enabled = guest_cpuid_has_pmu(vcpu);
+
+       if (pmu_enabled)
+               vmx->nested.msrs.procbased_ctls_high |= CPU_BASED_RDPMC_EXITING;
+       else
+               vmx->nested.msrs.procbased_ctls_high &= ~CPU_BASED_RDPMC_EXITING;
+}
+
 static void update_intel_pt_cfg(struct kvm_vcpu *vcpu)
 {
        struct vcpu_vmx *vmx = to_vmx(vcpu);
@@ -6944,6 +6968,7 @@ static void vmx_cpuid_update(struct kvm_vcpu *vcpu)
        if (nested_vmx_allowed(vcpu)) {
                nested_vmx_cr_fixed1_bits_update(vcpu);
                nested_vmx_entry_exit_ctls_update(vcpu);
+               nested_vmx_procbased_ctls_update(vcpu);
        }
 
        if (boot_cpu_has(X86_FEATURE_INTEL_PT) &&