[PATCH v7 02/12] KVM: arm64: PMU: Set the default PMU for the guest before vCPU reset

[Raghavendra Rao Ananta]

On Tue, Oct 10, 2023 at 3:25 PM Oliver Upton <oliver.upton at linux.dev> wrote:
>
> Hi Raghu,
>
> On Mon, Oct 09, 2023 at 11:08:48PM +0000, Raghavendra Rao Ananta wrote:
> > From: Reiji Watanabe <reijiw at google.com>
> >
> > The following patches will use the number of counters information
> > from the arm_pmu and use this to set the PMCR.N for the guest
> > during vCPU reset. However, since the guest is not associated
> > with any arm_pmu until userspace configures the vPMU device
> > attributes, and a reset can happen before this event, assign a
> > default PMU to the guest just before doing the reset.
> >
> > Signed-off-by: Reiji Watanabe <reijiw at google.com>
> > Signed-off-by: Raghavendra Rao Ananta <rananta at google.com>
> > ---
> >  arch/arm64/kvm/arm.c      | 20 ++++++++++++++++++++
> >  arch/arm64/kvm/pmu-emul.c | 12 ++----------
> >  include/kvm/arm_pmu.h     |  6 ++++++
> >  3 files changed, 28 insertions(+), 10 deletions(-)
> >
> > diff --git a/arch/arm64/kvm/arm.c b/arch/arm64/kvm/arm.c
> > index 78b0970eb8e6..708a53b70a7b 100644
> > --- a/arch/arm64/kvm/arm.c
> > +++ b/arch/arm64/kvm/arm.c
> > @@ -1313,6 +1313,23 @@ static bool kvm_vcpu_init_changed(struct kvm_vcpu *vcpu,
> >                            KVM_VCPU_MAX_FEATURES);
> >  }
> >
> > +static int kvm_vcpu_set_pmu(struct kvm_vcpu *vcpu)
> > +{
> > +     struct kvm *kvm = vcpu->kvm;
> > +
> > +     if (!kvm_arm_support_pmu_v3())
> > +             return -EINVAL;
>
> This check is pointless; the vCPU feature flags have been sanitised at
> this point, and a requirement of having PMUv3 is that this predicate is
> true.
>
Oh yes. I'll avoid this in v8.

> > +     /*
> > +      * When the vCPU has a PMU, but no PMU is set for the guest
> > +      * yet, set the default one.
> > +      */
> > +     if (unlikely(!kvm->arch.arm_pmu))
> > +             return kvm_arm_set_default_pmu(kvm);
> > +
> > +     return 0;
> > +}
> > +
>
> Apologies, I believe I was unclear last time around as to what I was
> wanting here. Let's call this thing kvm_setup_vcpu() such that we can
> add other one-time setup activities to it in the future.
>
> Something like:
>
> diff --git a/arch/arm64/kvm/arm.c b/arch/arm64/kvm/arm.c
> index 96641e442039..4896a44108e0 100644
> --- a/arch/arm64/kvm/arm.c
> +++ b/arch/arm64/kvm/arm.c
> @@ -1265,19 +1265,17 @@ static bool kvm_vcpu_init_changed(struct kvm_vcpu *vcpu,
>                              KVM_VCPU_MAX_FEATURES);
>  }
>
> -static int kvm_vcpu_set_pmu(struct kvm_vcpu *vcpu)
> +static int kvm_setup_vcpu(struct kvm_vcpu *vcpu)
>  {
>         struct kvm *kvm = vcpu->kvm;
>
> -       if (!kvm_arm_support_pmu_v3())
> -               return -EINVAL;
> -
>         /*
>          * When the vCPU has a PMU, but no PMU is set for the guest
>          * yet, set the default one.
>          */
> -       if (unlikely(!kvm->arch.arm_pmu))
> -               return kvm_arm_set_default_pmu(kvm);
> +       if (kvm_vcpu_has_pmu(vcpu) && !kvm->arch.arm_pmu &&
> +           kvm_arm_set_default_pmu(kvm))
> +               return -EINVAL;
>
>         return 0;
>  }
> @@ -1297,7 +1295,8 @@ static int __kvm_vcpu_set_target(struct kvm_vcpu *vcpu,
>
>         bitmap_copy(kvm->arch.vcpu_features, &features, KVM_VCPU_MAX_FEATURES);
>
> -       if (kvm_vcpu_has_pmu(vcpu) && kvm_vcpu_set_pmu(vcpu))
> +       ret = kvm_setup_vcpu(vcpu);
> +       if (ret)
>                 goto out_unlock;
>
>         /* Now we know what it is, we can reset it. */
>
Introducing kvm_setup_vcpu() seems better than directly calling
kvm_vcpu_set_pmu(), which feels like it's crashing a party.

Thank you.
Raghavendra
> --
> Thanks,
> Oliver