2014-06-30 08:35:56

by Nadav Amit

[permalink] [raw]
Subject: [PATCH] KVM: x86: Pending interrupt may be delivered after INIT

We encountered a scenario in which after an INIT is delivered, a pending
interrupt is delivered, although it was sent before the INIT. As the SDM
states in section 10.4.7.1, the ISR and the IRR should be cleared after INIT as
KVM does. This also means that pending interrupts should be cleared. This
patch clears upon reset (and INIT) the pending interrupts; and at the same
occassion clears the pending exceptions, since they may cause a similar issue.

Signed-off-by: Nadav Amit <[email protected]>
---
arch/x86/kvm/x86.c | 2 ++
1 file changed, 2 insertions(+)

diff --git a/arch/x86/kvm/x86.c b/arch/x86/kvm/x86.c
index f32a025..863ac07 100644
--- a/arch/x86/kvm/x86.c
+++ b/arch/x86/kvm/x86.c
@@ -6835,6 +6835,8 @@ void kvm_vcpu_reset(struct kvm_vcpu *vcpu)
atomic_set(&vcpu->arch.nmi_queued, 0);
vcpu->arch.nmi_pending = 0;
vcpu->arch.nmi_injected = false;
+ vcpu->arch.interrupt.pending = false;
+ vcpu->arch.exception.pending = false;

memset(vcpu->arch.db, 0, sizeof(vcpu->arch.db));
vcpu->arch.dr6 = DR6_FIXED_1;
--
1.9.1


2014-06-30 08:47:14

by Gleb Natapov

[permalink] [raw]
Subject: Re: [PATCH] KVM: x86: Pending interrupt may be delivered after INIT

On Mon, Jun 30, 2014 at 11:35:27AM +0300, Nadav Amit wrote:
> We encountered a scenario in which after an INIT is delivered, a pending
> interrupt is delivered, although it was sent before the INIT. As the SDM
> states in section 10.4.7.1, the ISR and the IRR should be cleared after INIT as
> KVM does. This also means that pending interrupts should be cleared. This
> patch clears upon reset (and INIT) the pending interrupts; and at the same
> occassion clears the pending exceptions, since they may cause a similar issue.
>
> Signed-off-by: Nadav Amit <[email protected]>
> ---
> arch/x86/kvm/x86.c | 2 ++
> 1 file changed, 2 insertions(+)
>
> diff --git a/arch/x86/kvm/x86.c b/arch/x86/kvm/x86.c
> index f32a025..863ac07 100644
> --- a/arch/x86/kvm/x86.c
> +++ b/arch/x86/kvm/x86.c
> @@ -6835,6 +6835,8 @@ void kvm_vcpu_reset(struct kvm_vcpu *vcpu)
> atomic_set(&vcpu->arch.nmi_queued, 0);
> vcpu->arch.nmi_pending = 0;
> vcpu->arch.nmi_injected = false;
> + vcpu->arch.interrupt.pending = false;
> + vcpu->arch.exception.pending = false;
kvm_clear_interrupt_queue(vcpu);
kvm_clear_exception_queue(vcpu);

>
> memset(vcpu->arch.db, 0, sizeof(vcpu->arch.db));
> vcpu->arch.dr6 = DR6_FIXED_1;
> --
> 1.9.1
>

--
Gleb.

2014-06-30 09:03:26

by Nadav Amit

[permalink] [raw]
Subject: [PATCH v2] KVM: x86: Pending interrupt may be delivered after INIT

We encountered a scenario in which after an INIT is delivered, a pending
interrupt is delivered, although it was sent before the INIT. As the SDM
states in section 10.4.7.1, the ISR and the IRR should be cleared after INIT as
KVM does. This also means that pending interrupts should be cleared. This
patch clears upon reset (and INIT) the pending interrupts; and at the same
occassion clears the pending exceptions, since they may cause a similar issue.

Signed-off-by: Nadav Amit <[email protected]>
---
arch/x86/kvm/x86.c | 2 ++
1 file changed, 2 insertions(+)

diff --git a/arch/x86/kvm/x86.c b/arch/x86/kvm/x86.c
index f32a025..6425a31 100644
--- a/arch/x86/kvm/x86.c
+++ b/arch/x86/kvm/x86.c
@@ -6835,6 +6835,8 @@ void kvm_vcpu_reset(struct kvm_vcpu *vcpu)
atomic_set(&vcpu->arch.nmi_queued, 0);
vcpu->arch.nmi_pending = 0;
vcpu->arch.nmi_injected = false;
+ kvm_clear_interrupt_queue(vcpu);
+ kvm_clear_exception_queue(vcpu);

memset(vcpu->arch.db, 0, sizeof(vcpu->arch.db));
vcpu->arch.dr6 = DR6_FIXED_1;
--
1.9.1

2014-07-09 16:16:13

by Paolo Bonzini

[permalink] [raw]
Subject: Re: [PATCH v2] KVM: x86: Pending interrupt may be delivered after INIT

Il 30/06/2014 11:03, Nadav Amit ha scritto:
> We encountered a scenario in which after an INIT is delivered, a pending
> interrupt is delivered, although it was sent before the INIT. As the SDM
> states in section 10.4.7.1, the ISR and the IRR should be cleared after INIT as
> KVM does. This also means that pending interrupts should be cleared. This
> patch clears upon reset (and INIT) the pending interrupts; and at the same
> occassion clears the pending exceptions, since they may cause a similar issue.
>
> Signed-off-by: Nadav Amit <[email protected]>
> ---
> arch/x86/kvm/x86.c | 2 ++
> 1 file changed, 2 insertions(+)
>
> diff --git a/arch/x86/kvm/x86.c b/arch/x86/kvm/x86.c
> index f32a025..6425a31 100644
> --- a/arch/x86/kvm/x86.c
> +++ b/arch/x86/kvm/x86.c
> @@ -6835,6 +6835,8 @@ void kvm_vcpu_reset(struct kvm_vcpu *vcpu)
> atomic_set(&vcpu->arch.nmi_queued, 0);
> vcpu->arch.nmi_pending = 0;
> vcpu->arch.nmi_injected = false;
> + kvm_clear_interrupt_queue(vcpu);
> + kvm_clear_exception_queue(vcpu);
>
> memset(vcpu->arch.db, 0, sizeof(vcpu->arch.db));
> vcpu->arch.dr6 = DR6_FIXED_1;
>

Applied to kvm/queue, thanks.

Paolo