From: Jan Dakinevich <[email protected]>

Previously if no PMU counters were exposed to guest, KVM skipped whole
remaining PMU-related initialisation, including filling of LBR-related
information. It is wrong, since LBR support doesn't depend on
existence of PMU counters.

https://jira.sw.ru/browse/PSBM-94429
Signed-off-by: Jan Dakinevich <[email protected]>

https://jira.sw.ru/browse/PSBM-127846
(cherry-picked from vz7 commit 39b4be609936 ("KVM: x86/vPMU: handle
access to LBR stack regardless of PMU counters presence"))
Fixes: 1fb36a7c3b57 ("KVM: x86/vPMU: Ignore access to LBR-related MSRs")
Signed-off-by: Pavel Tikhomirov <[email protected]>
---
 arch/x86/kvm/vmx/pmu_intel.c | 10 +++++-----
 1 file changed, 5 insertions(+), 5 deletions(-)

diff --git a/arch/x86/kvm/vmx/pmu_intel.c b/arch/x86/kvm/vmx/pmu_intel.c
index 0df709f5b856..ccfe7ce4b78f 100644
--- a/arch/x86/kvm/vmx/pmu_intel.c
+++ b/arch/x86/kvm/vmx/pmu_intel.c
@@ -333,6 +333,11 @@ static void intel_pmu_refresh(struct kvm_vcpu *vcpu)
        pmu->version = 0;
        pmu->reserved_bits = 0xffffffff00200000ull;
 
+       entry = kvm_find_cpuid_entry(vcpu, 1, 0);
+       if (entry)
+               intel_pmu_lbr_fill(&pmu->lbr,
+                       x86_family(entry->eax), x86_model(entry->eax));
+
        entry = kvm_find_cpuid_entry(vcpu, 0xa, 0);
        if (!entry)
                return;
@@ -383,11 +388,6 @@ static void intel_pmu_refresh(struct kvm_vcpu *vcpu)
                INTEL_PMC_MAX_GENERIC, pmu->nr_arch_fixed_counters);
 
        nested_vmx_pmu_entry_exit_ctls_update(vcpu);
-
-       entry = kvm_find_cpuid_entry(vcpu, 1, 0);
-       if (entry)
-               intel_pmu_lbr_fill(&pmu->lbr,
-                       x86_family(entry->eax), x86_model(entry->eax));
 }
 
 static void intel_pmu_init(struct kvm_vcpu *vcpu)
-- 
2.31.1

_______________________________________________
Devel mailing list
[email protected]
https://lists.openvz.org/mailman/listinfo/devel

Reply via email to