KVM: x86: svm: make sure NMI is injected after nmi_singlestep

PATCH Applied Vitaly Kuznetsov 2 participants 2 messages Apr 3, 2019 – Apr 10, 2019

Subsystems (3)

x86 KVM KVM x86

lore

Share & tools

lore

Vitaly Kuznetsov

author

2019-04-03 14:06 UTC #0 lore raw

PATCH 1/1 KVM: x86: svm: make sure NMI is injected after nmi_singlestep

I noticed that apic test from kvm-unit-tests always hangs on my EPYC 7401P, the hanging test nmi-after-sti is trying to deliver 30000 NMIs and tracing shows that we're sometimes able to deliver a few but never all.

When we're trying to inject an NMI we may fail to do so immediately for various reasons, however, we still need to inject it so enable_nmi_window() arms nmi_singlestep mode. #DB occurs as expected, but we're not checking for pending NMIs before entering the guest and unless there's a different event to process, the NMI will never get delivered.

Make KVM_REQ_EVENT request on the vCPU from db_interception() to make sure pending NMIs are checked and possibly injected.

Signed-off-by Vitaly Kuznetsov <[email protected]>

---
- I have to admit my RFC didn't gain much popularity, re-sending as non-RFC.

---
 arch/x86/kvm/svm.c | 3 +++
 1 file changed, 3 insertions(+)

arch/x86/kvm/svm.c +3 -0

index 426039285fd1..69a0edb23c28 100644
@@ -2692,6 +2692,7 @@ static int npf_interception(struct vcpu_svm *svm)
 static int db_interception(struct vcpu_svm *svm)
 {
 	struct kvm_run *kvm_run = svm->vcpu.run;
+	struct kvm_vcpu *vcpu = &svm->vcpu;
 
 	if (!(svm->vcpu.guest_debug &
 	      (KVM_GUESTDBG_SINGLESTEP | KVM_GUESTDBG_USE_HW_BP)) &&

@@ -2702,6 +2703,8 @@ static int db_interception(struct vcpu_svm *svm)
 
 	if (svm->nmi_singlestep) {
 		disable_nmi_singlestep(svm);
+		/* Make sure we check for pending NMIs upon entry */
+		kvm_make_request(KVM_REQ_EVENT, vcpu);
 	}
 
 	if (svm->vcpu.guest_debug &

Signature · 1 line

2.20.1

Paolo Bonzini

maintainer Applied

2019-04-10 09:08 UTC #1 lore raw

Quoted 35 lines · Vitaly Kuznetsov

I noticed that apic test from kvm-unit-tests always hangs on my EPYC 7401P, the hanging test nmi-after-sti is trying to deliver 30000 NMIs and tracing shows that we're sometimes able to deliver a few but never all.
When we're trying to inject an NMI we may fail to do so immediately for various reasons, however, we still need to inject it so enable_nmi_window() arms nmi_singlestep mode. #DB occurs as expected, but we're not checking for pending NMIs before entering the guest and unless there's a different event to process, the NMI will never get delivered.
Make KVM_REQ_EVENT request on the vCPU from db_interception() to make sure pending NMIs are checked and possibly injected.
Signed-off-by Vitaly Kuznetsov <[email protected]>
---
- I have to admit my RFC didn't gain much popularity, re-sending as non-RFC.
---
 arch/x86/kvm/svm.c | 3 +++
 1 file changed, 3 insertions(+)
arch/x86/kvm/svm.c +1 -0
index 426039285fd1..69a0edb23c28 100644
@@ -2692,6 +2692,7 @@ static int npf_interception(struct vcpu_svm *svm)
 static int db_interception(struct vcpu_svm *svm)
 {
 	struct kvm_run *kvm_run = svm->vcpu.run;
+	struct kvm_vcpu *vcpu = &svm->vcpu;
 
 	if (!(svm->vcpu.guest_debug &
 	      (KVM_GUESTDBG_SINGLESTEP | KVM_GUESTDBG_USE_HW_BP)) &&

arch/x86/kvm/svm.c static int db_interception(struct vcpu_svm *svm) · line 2710

Quoted 8 lines

arch/x86/kvm/svm.c +2 -0

@@ -2702,6 +2703,8 @@ static int db_interception(struct vcpu_svm *svm)
 
 	if (svm->nmi_singlestep) {
 		disable_nmi_singlestep(svm);
+		/* Make sure we check for pending NMIs upon entry */
+		kvm_make_request(KVM_REQ_EVENT, vcpu);
 	}
 
 	if (svm->vcpu.guest_debug &

Queued, thanks.

Paolo