Re: [PATCH 02/17] KVM: arm64: Advertise the capabilities that protected VMs support

[Vincent Donnefort]

On Mon, Aug 31, 2026 at 05:34:06PM +0100, Fuad Tabba wrote:
> kvm_pkvm_ext_allowed() denies every capability it does not name, so a
> protected VM reports 0 for interfaces it implements: KVM_CHECK_EXTENSION
> denies KVM_CAP_ONE_REG while KVM_{GET,SET}_ONE_REG stage the guest's
> boot state.
> 
> Allow the capabilities that work for a protected guest, the vGIC and
> the I/O bus being host-managed, subject to the restrictions the rest of
> this series adds.
> 
> Signed-off-by: Fuad Tabba <fuad.tabba@xxxxxxxxx>
> ---
>  arch/arm64/include/asm/kvm_pkvm.h | 9 +++++++++
>  1 file changed, 9 insertions(+)
> 
> diff --git a/arch/arm64/include/asm/kvm_pkvm.h b/arch/arm64/include/asm/kvm_pkvm.h
> index beea00e693a0a..c4c834d55e503 100644
> --- a/arch/arm64/include/asm/kvm_pkvm.h
> +++ b/arch/arm64/include/asm/kvm_pkvm.h
> @@ -42,6 +42,15 @@ static inline bool kvm_pkvm_ext_allowed(struct kvm *kvm, long ext)
>  	case KVM_CAP_ARM_VM_IPA_SIZE:
>  	case KVM_CAP_ARM_PTRAUTH_ADDRESS:
>  	case KVM_CAP_ARM_PTRAUTH_GENERIC:
> +	case KVM_CAP_ONE_REG:

Shouldn't we bring the VCPU_PKVM_FINALIZED restriction in kvm_arch_vcpu_ioctl()
first before we allow SET_ONE_REG/GET_ONE_REG?

> +	case KVM_CAP_MP_STATE:
> +	case KVM_CAP_VCPU_EVENTS:
> +	case KVM_CAP_VCPU_ATTRIBUTES:
> +	case KVM_CAP_IMMEDIATE_EXIT:
> +	case KVM_CAP_IOEVENTFD:
> +	case KVM_CAP_IRQFD_RESAMPLE:
> +	case KVM_CAP_ARM_IRQ_LINE_LAYOUT_2:
> +	case KVM_CAP_ARM_INJECT_SERROR_ESR:
>  		return true;
>  	case KVM_CAP_ARM_MTE:
>  		return false;
> -- 
> 2.39.5
>