OSDN Git Service

KVM: nVMX: Add tracepoint for nested VM-Enter
authorDavid Matlack <dmatlack@google.com>
Thu, 25 Aug 2022 22:57:54 +0000 (22:57 +0000)
committerPaolo Bonzini <pbonzini@redhat.com>
Mon, 26 Sep 2022 16:02:32 +0000 (12:02 -0400)
Call trace_kvm_nested_vmenter() during nested VMLAUNCH/VMRESUME to bring
parity with nSVM's usage of the tracepoint during nested VMRUN.

Attempt to use analagous VMCS fields to the VMCB fields that are
reported in the SVM case:

"int_ctl": 32-bit field of the VMCB that the CPU uses to deliver virtual
interrupts. The analagous VMCS field is the 16-bit "guest interrupt
status".

"event_inj": 32-bit field of VMCB that is used to inject events
(exceptions and interrupts) into the guest. The analagous VMCS field
is the "VM-entry interruption-information field".

"npt_enabled": 1 when the VCPU has enabled nested paging. The analagous
VMCS field is the enable-EPT execution control.

"npt_addr": 64-bit field when the VCPU has enabled nested paging. The
analagous VMCS field is the ept_pointer.

Signed-off-by: David Matlack <dmatlack@google.com>
[move the code into the nested_vmx_enter_non_root_mode().]
Signed-off-by: Mingwei Zhang <mizhang@google.com>
Link: https://lore.kernel.org/r/20220825225755.907001-3-mizhang@google.com
Signed-off-by: Sean Christopherson <seanjc@google.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
arch/x86/kvm/vmx/nested.c

index ddd4367..f72fe94 100644 (file)
@@ -3364,6 +3364,14 @@ enum nvmx_vmentry_status nested_vmx_enter_non_root_mode(struct kvm_vcpu *vcpu,
        };
        u32 failed_index;
 
+       trace_kvm_nested_vmenter(kvm_rip_read(vcpu),
+                                vmx->nested.current_vmptr,
+                                vmcs12->guest_rip,
+                                vmcs12->guest_intr_status,
+                                vmcs12->vm_entry_intr_info_field,
+                                vmcs12->secondary_vm_exec_control & SECONDARY_EXEC_ENABLE_EPT,
+                                KVM_ISA_VMX);
+
        kvm_service_local_tlb_flush_requests(vcpu);
 
        evaluate_pending_interrupts = exec_controls_get(vmx) &