2022-08-22 17:12:49

by Paolo Bonzini

[permalink] [raw]
Subject: [PATCH v3 3/7] KVM: nVMX: Make an event request when pending an MTF nested VM-Exit

From: Sean Christopherson <[email protected]>

Set KVM_REQ_EVENT when MTF becomes pending to ensure that KVM will run
through inject_pending_event() and thus vmx_check_nested_events() prior
to re-entering the guest.

MTF currently works by virtue of KVM's hack that calls
kvm_check_nested_events() from kvm_vcpu_running(), but that hack will
be removed in the near future. Until that call is removed, the patch
introduces no functional change.

Fixes: 5ef8acbdd687 ("KVM: nVMX: Emulate MTF when performing instruction emulation")
Cc: [email protected]
Signed-off-by: Sean Christopherson <[email protected]>
Reviewed-by: Maxim Levitsky <[email protected]>
Signed-off-by: Paolo Bonzini <[email protected]>
---
arch/x86/kvm/vmx/vmx.c | 6 ++++--
1 file changed, 4 insertions(+), 2 deletions(-)

diff --git a/arch/x86/kvm/vmx/vmx.c b/arch/x86/kvm/vmx/vmx.c
index c9b49a09e6b5..d22e67ac1e69 100644
--- a/arch/x86/kvm/vmx/vmx.c
+++ b/arch/x86/kvm/vmx/vmx.c
@@ -1659,10 +1659,12 @@ static void vmx_update_emulated_instruction(struct kvm_vcpu *vcpu)
*/
if (nested_cpu_has_mtf(vmcs12) &&
(!vcpu->arch.exception.pending ||
- vcpu->arch.exception.nr == DB_VECTOR))
+ vcpu->arch.exception.nr == DB_VECTOR)) {
vmx->nested.mtf_pending = true;
- else
+ kvm_make_request(KVM_REQ_EVENT, vcpu);
+ } else {
vmx->nested.mtf_pending = false;
+ }
}

static int vmx_skip_emulated_instruction(struct kvm_vcpu *vcpu)
--
2.31.1



2022-08-22 18:07:22

by Jim Mattson

[permalink] [raw]
Subject: Re: [PATCH v3 3/7] KVM: nVMX: Make an event request when pending an MTF nested VM-Exit

On Mon, Aug 22, 2022 at 10:08 AM Paolo Bonzini <[email protected]> wrote:
>
> From: Sean Christopherson <[email protected]>
>
> Set KVM_REQ_EVENT when MTF becomes pending to ensure that KVM will run
> through inject_pending_event() and thus vmx_check_nested_events() prior
> to re-entering the guest.
>
> MTF currently works by virtue of KVM's hack that calls
> kvm_check_nested_events() from kvm_vcpu_running(), but that hack will
> be removed in the near future. Until that call is removed, the patch
> introduces no functional change.
>
> Fixes: 5ef8acbdd687 ("KVM: nVMX: Emulate MTF when performing instruction emulation")
> Cc: [email protected]
> Signed-off-by: Sean Christopherson <[email protected]>
> Reviewed-by: Maxim Levitsky <[email protected]>
> Signed-off-by: Paolo Bonzini <[email protected]>
> ---
What happens if live migration occurs before the KVM_REQ_EVENT is processed?

2022-08-22 19:58:59

by Paolo Bonzini

[permalink] [raw]
Subject: Re: [PATCH v3 3/7] KVM: nVMX: Make an event request when pending an MTF nested VM-Exit

On 8/22/22 19:52, Jim Mattson wrote:
> On Mon, Aug 22, 2022 at 10:08 AM Paolo Bonzini <[email protected]> wrote:
>>
>> From: Sean Christopherson <[email protected]>
>>
>> Set KVM_REQ_EVENT when MTF becomes pending to ensure that KVM will run
>> through inject_pending_event() and thus vmx_check_nested_events() prior
>> to re-entering the guest.
>>
>> MTF currently works by virtue of KVM's hack that calls
>> kvm_check_nested_events() from kvm_vcpu_running(), but that hack will
>> be removed in the near future. Until that call is removed, the patch
>> introduces no functional change.
>>
>> Fixes: 5ef8acbdd687 ("KVM: nVMX: Emulate MTF when performing instruction emulation")
>> Cc: [email protected]
>> Signed-off-by: Sean Christopherson <[email protected]>
>> Reviewed-by: Maxim Levitsky <[email protected]>
>> Signed-off-by: Paolo Bonzini <[email protected]>
>> ---
> What happens if live migration occurs before the KVM_REQ_EVENT is processed?

Good point, this must be squashed in:

diff --git a/arch/x86/kvm/vmx/nested.c b/arch/x86/kvm/vmx/nested.c
index ab135f9ef52f..a703b71d675d 100644
--- a/arch/x86/kvm/vmx/nested.c
+++ b/arch/x86/kvm/vmx/nested.c
@@ -6481,6 +6481,8 @@ static int vmx_set_nested_state(struct kvm_vcpu *vcpu,
if (ret)
goto error_guest_mode;

+ if (vmx->nested.mtf_pending)
+ kvm_make_request(KVM_REQ_EVENT, vcpu);
return 0;

error_guest_mode:

so that the other side restores the request.

Paolo