Commit f8636849 authored by Avi Kivity's avatar Avi Kivity Committed by Marcelo Tosatti

KVM: Fix race between nmi injection and enabling nmi window

The interrupt injection logic looks something like

  if an nmi is pending, and nmi injection allowed
    inject nmi
  if an nmi is pending
    request exit on nmi window

the problem is that "nmi is pending" can be set asynchronously by
the PIT; if it happens to fire between the two if statements, we
will request an nmi window even though nmi injection is allowed.  On
SVM, this has disasterous results, since it causes eflags.TF to be
set in random guest code.

The fix is simple; make nmi_pending synchronous using the standard
vcpu->requests mechanism; this ensures the code above is completely
synchronous wrt nmi_pending.
Signed-off-by: default avatarAvi Kivity <avi@redhat.com>
parent 217ece61
...@@ -360,8 +360,8 @@ void kvm_propagate_fault(struct kvm_vcpu *vcpu, struct x86_exception *fault) ...@@ -360,8 +360,8 @@ void kvm_propagate_fault(struct kvm_vcpu *vcpu, struct x86_exception *fault)
void kvm_inject_nmi(struct kvm_vcpu *vcpu) void kvm_inject_nmi(struct kvm_vcpu *vcpu)
{ {
kvm_make_request(KVM_REQ_NMI, vcpu);
kvm_make_request(KVM_REQ_EVENT, vcpu); kvm_make_request(KVM_REQ_EVENT, vcpu);
vcpu->arch.nmi_pending = 1;
} }
EXPORT_SYMBOL_GPL(kvm_inject_nmi); EXPORT_SYMBOL_GPL(kvm_inject_nmi);
...@@ -5180,6 +5180,8 @@ static int vcpu_enter_guest(struct kvm_vcpu *vcpu) ...@@ -5180,6 +5180,8 @@ static int vcpu_enter_guest(struct kvm_vcpu *vcpu)
r = 1; r = 1;
goto out; goto out;
} }
if (kvm_check_request(KVM_REQ_NMI, vcpu))
vcpu->arch.nmi_pending = true;
} }
r = kvm_mmu_reload(vcpu); r = kvm_mmu_reload(vcpu);
......
...@@ -43,6 +43,7 @@ ...@@ -43,6 +43,7 @@
#define KVM_REQ_DEACTIVATE_FPU 10 #define KVM_REQ_DEACTIVATE_FPU 10
#define KVM_REQ_EVENT 11 #define KVM_REQ_EVENT 11
#define KVM_REQ_APF_HALT 12 #define KVM_REQ_APF_HALT 12
#define KVM_REQ_NMI 13
#define KVM_USERSPACE_IRQ_SOURCE_ID 0 #define KVM_USERSPACE_IRQ_SOURCE_ID 0
......
Markdown is supported
0%
or
You are about to add 0 people to the discussion. Proceed with caution.
Finish editing this message first!
Please register or to comment