Skip to contentSkip to conversation

kvm: nVMX: add tracepoint for kvm:kvm_nested_vmrun

PATCH In review Mingwei Zhang 2 participants 3 messages Jun 26, 2022 – Jun 29, 2022

Subsystems (3)
lore
Share & tools
lore
#0 lore raw

PATCH 1/1 kvm: nVMX: add tracepoint for kvm:kvm_nested_vmrun

From: David Matlack <[email protected]>

This tracepoint is called by nested SVM during emulated VMRUN. Call also during emulated VMLAUNCH and VMRESUME in nested VMX.

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

[Move the code into the nested_vmx_enter_non_root_mode().]
---
 arch/x86/kvm/vmx/nested.c | 7 +++++++
 1 file changed, 7 insertions(+)
arch/x86/kvm/vmx/nested.c +7 -0
index f5cb18e00e78..29cc36cf2568 100644
@@ -3367,6 +3367,13 @@ enum nvmx_vmentry_status nested_vmx_enter_non_root_mode(struct kvm_vcpu *vcpu,
 	};
 	u32 failed_index;
 
+	trace_kvm_nested_vmrun(
+		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_service_local_tlb_flush_requests(vcpu);
 
 	evaluate_pending_interrupts = exec_controls_get(vmx) &

base-commit: 922d4578cfd017da67f545bfd07331bda86f795d

Signature · 1 line
2.37.0.rc0.161.g10f37bed90-goog
#1 lore raw
Quoted 31 lines · Mingwei Zhang

From: David Matlack <[email protected]>

This tracepoint is called by nested SVM during emulated VMRUN. Call also during emulated VMLAUNCH and VMRESUME in nested VMX.

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

[Move the code into the nested_vmx_enter_non_root_mode().]
---
 arch/x86/kvm/vmx/nested.c | 7 +++++++
 1 file changed, 7 insertions(+)
arch/x86/kvm/vmx/nested.c +0 -0
index f5cb18e00e78..29cc36cf2568 100644

arch/x86/kvm/vmx/nested.c enum nvmx_vmentry_status nested_vmx_enter_non_root_mode(struct kvm_vcpu *vcpu, · line 3372

Quoted 6 lines
arch/x86/kvm/vmx/nested.c +3 -0
@@ -3367,6 +3367,13 @@ enum nvmx_vmentry_status nested_vmx_enter_non_root_mode(struct kvm_vcpu *vcpu,
 	};
 	u32 failed_index;
 
+	trace_kvm_nested_vmrun(
+		kvm_rip_read(vcpu), vmx->nested.current_vmptr,
+		vmcs12->guest_rip,

Be consistent; either put each parameter on it's own line or wrap only when necessary.

arch/x86/kvm/vmx/nested.c enum nvmx_vmentry_status nested_vmx_enter_non_root_mode(struct kvm_vcpu *vcpu, · line 3376

Quoted 4 lines
arch/x86/kvm/vmx/nested.c +4 -0
+		vmcs12->guest_intr_status,
+		vmcs12->vm_entry_intr_info_field,
+		vmcs12->secondary_vm_exec_control & SECONDARY_EXEC_ENABLE_EPT);
+

Align the parameters to the opening '(', that "rule" trumps the 80 char soft limit.

	trace_kvm_nested_vmrun(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);

And if we're going to add nVMX, we should clean up the tracepoint output. E.g. pass in KVM_ISA_{SVM,VMX} to different VMCB vs. VMCS and npt vs. ept (and maybe print nNPT and nEPT to make it obvious it's the vmcs12 setting?). The "nrip" field is wrong even for SVM; the tracepoint prints the L2 rip, not the next_rip field in vmcs12. Maybe "L2 rip"?

#2 lore raw
Quoted 9 lines
Quoted 9 lines
arch/x86/kvm/vmx/nested.c +0 -0
index f5cb18e00e78..29cc36cf2568 100644
@@ -3367,6 +3367,13 @@ enum nvmx_vmentry_status nested_vmx_enter_non_root_mode(struct kvm_vcpu *vcpu,
      };
      u32 failed_index;
 
Quoted 6 lines

arch/x86/kvm/vmx/nested.c enum nvmx_vmentry_status nested_vmx_enter_non_root_mode(struct kvm_vcpu *vcpu, · line 3372

Quoted 3 lines
arch/x86/kvm/vmx/nested.c +3 -0
+     trace_kvm_nested_vmrun(
+             kvm_rip_read(vcpu), vmx->nested.current_vmptr,
+             vmcs12->guest_rip,

Be consistent; either put each parameter on it's own line or wrap only when necessary.

will do.

Quoted 7 lines

arch/x86/kvm/vmx/nested.c enum nvmx_vmentry_status nested_vmx_enter_non_root_mode(struct kvm_vcpu *vcpu, · line 3376

Quoted 5 lines
arch/x86/kvm/vmx/nested.c +4 -0
+             vmcs12->guest_intr_status,
+             vmcs12->vm_entry_intr_info_field,
+             vmcs12->secondary_vm_exec_control & SECONDARY_EXEC_ENABLE_EPT);
+

Align the parameters to the opening '(', that "rule" trumps the 80 char soft limit.

will do.

Quoted 8 lines
        trace_kvm_nested_vmrun(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);
Quoted 5 lines

And if we're going to add nVMX, we should clean up the tracepoint output. E.g. pass in KVM_ISA_{SVM,VMX} to different VMCB vs. VMCS and npt vs. ept (and maybe print nNPT and nEPT to make it obvious it's the vmcs12 setting?). The "nrip" field is wrong even for SVM; the tracepoint prints the L2 rip, not the next_rip field in vmcs12. Maybe "L2 rip"?

I can change the 'nrip' to 'nested rip', since it may not necessarily the 'L2 rip'.

For the others, will do.