KVM: x86: DR7.GD should be cleared upon any #DB exception

Intel SDM 17.2.4 (Debug Control Register (DR7)) says: "The processor clears the
GD flag upon entering to the debug exception handler." This sentence may be
misunderstood as if it happens only on #DB due to debug-register protection,
but it happens regardless to the cause of the #DB.

Fix the behavior to match both real hardware and Bochs.

Signed-off-by: Nadav Amit <namit@cs.technion.ac.il>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>

diff --git a/arch/x86/kvm/vmx.c b/arch/x86/kvm/vmx.c
index 2627ffa..63e7cf1 100644 (file)
--- a/arch/x86/kvm/vmx.c
+++ b/arch/x86/kvm/vmx.c
@@ -5185,9 +5185,7 @@ static int handle_dr(struct kvm_vcpu *vcpu)
                        vcpu->run->exit_reason = KVM_EXIT_DEBUG;
                        return 0;
                } else {
-                       vcpu->arch.dr7 &= ~DR7_GD;
                        vcpu->arch.dr6 |= DR6_BD | DR6_RTM;
-                       vmcs_writel(GUEST_DR7, vcpu->arch.dr7);
                        kvm_queue_exception(vcpu, DB_VECTOR);
                        return 1;
                }

diff --git a/arch/x86/kvm/x86.c b/arch/x86/kvm/x86.c
index 0033df3..1ab7361 100644 (file)
--- a/arch/x86/kvm/x86.c
+++ b/arch/x86/kvm/x86.c
@@ -5965,6 +5965,12 @@ static int inject_pending_event(struct kvm_vcpu *vcpu, bool req_int_win)
                        __kvm_set_rflags(vcpu, kvm_get_rflags(vcpu) |
                                             X86_EFLAGS_RF);
 
+               if (vcpu->arch.exception.nr == DB_VECTOR &&
+                   (vcpu->arch.dr7 & DR7_GD)) {
+                       vcpu->arch.dr7 &= ~DR7_GD;
+                       kvm_update_dr7(vcpu);
+               }
+
                kvm_x86_ops->queue_exception(vcpu, vcpu->arch.exception.nr,
                                          vcpu->arch.exception.has_error_code,
                                          vcpu->arch.exception.error_code,