2024-05-22 01:42:36

by Sean Christopherson

Subject: [PATCH v2 5/6] KVM: x86: Unconditionally set l1tf_flush_l1d during vCPU load

Always set l1tf_flush_l1d during kvm_arch_vcpu_load() instead of setting
it only when the vCPU is being scheduled back in. The flag is processed
only when VM-Enter is imminent, and KVM obviously needs to load the vCPU
before VM-Enter, so attempting to precisely set l1tf_flush_l1d provides no
meaningful value. I.e. the flag _will_ be set either way, it's simply a
matter of when.

Signed-off-by: Sean Christopherson <[email protected]>
---
arch/x86/kvm/x86.c | 11 +++++------
1 file changed, 5 insertions(+), 6 deletions(-)

diff --git a/arch/x86/kvm/x86.c b/arch/x86/kvm/x86.c
index 59aa772af755..60fea297f91f 100644
--- a/arch/x86/kvm/x86.c
+++ b/arch/x86/kvm/x86.c
@@ -5006,12 +5006,11 @@ void kvm_arch_vcpu_load(struct kvm_vcpu *vcpu, int cpu)
{
struct kvm_pmu *pmu = vcpu_to_pmu(vcpu);

- if (vcpu->scheduled_out) {
- vcpu->arch.l1tf_flush_l1d = true;
- if (pmu->version && unlikely(pmu->event_count)) {
- pmu->need_cleanup = true;
- kvm_make_request(KVM_REQ_PMU, vcpu);
- }
+ vcpu->arch.l1tf_flush_l1d = true;
+
+ if (vcpu->scheduled_out && pmu->version && pmu->event_count) {
+ pmu->need_cleanup = true;
+ kvm_make_request(KVM_REQ_PMU, vcpu);
}

/* Address WBINVD may be executed by guest */
--
2.45.0.215.g3402c0e53f-goog

2024-05-23 22:43:23

by Kai Huang

Subject: Re: [PATCH v2 5/6] KVM: x86: Unconditionally set l1tf_flush_l1d during vCPU load

On 22/05/2024 1:40 pm, Sean Christopherson wrote:
> Always set l1tf_flush_l1d during kvm_arch_vcpu_load() instead of setting
> it only when the vCPU is being scheduled back in. The flag is processed
> only when VM-Enter is imminent, and KVM obviously needs to load the vCPU
> before VM-Enter, so attempting to precisely set l1tf_flush_l1d provides no
> meaningful value. I.e. the flag _will_ be set either way, it's simply a
> matter of when.

Seems reasonable.

>
> Signed-off-by: Sean Christopherson <[email protected]>
> ---

Acked-by: Kai Huang <[email protected]>

> arch/x86/kvm/x86.c | 11 +++++------
> 1 file changed, 5 insertions(+), 6 deletions(-)
>
> diff --git a/arch/x86/kvm/x86.c b/arch/x86/kvm/x86.c
> index 59aa772af755..60fea297f91f 100644
> --- a/arch/x86/kvm/x86.c
> +++ b/arch/x86/kvm/x86.c
> @@ -5006,12 +5006,11 @@ void kvm_arch_vcpu_load(struct kvm_vcpu *vcpu, int cpu)
> {
> struct kvm_pmu *pmu = vcpu_to_pmu(vcpu);
>
> - if (vcpu->scheduled_out) {
> - vcpu->arch.l1tf_flush_l1d = true;
> - if (pmu->version && unlikely(pmu->event_count)) {
> - pmu->need_cleanup = true;
> - kvm_make_request(KVM_REQ_PMU, vcpu);
> - }
> + vcpu->arch.l1tf_flush_l1d = true;
> +
> + if (vcpu->scheduled_out && pmu->version && pmu->event_count) {
> + pmu->need_cleanup = true;
> + kvm_make_request(KVM_REQ_PMU, vcpu);
> }

Nit, the unlikely() is lost, but I guess it is OK?