Re: [PATCH 3/4] KVM: arm64: Enforce reserved bits for PSCI target affinities

[Marc Zyngier <maz@kernel.org>]

On Wed, 18 Aug 2021 09:50:46 +0100,
Oliver Upton <oupton@google.com> wrote:
> 
> Some calls in PSCI take a target affinity argument, defined to be
> bit-compatible with the affinity fields in MPIDR_EL1. All other bits in
> the parameter are reserved and must be 0.

For future reference, it may be worth quoting the spec (ARM DEN 0022D,
5.1.4 "CPU_ON").

> Return INVALID_PARAMETERS if the guest incorrectly sets a reserved
> bit.
> 
> Signed-off-by: Oliver Upton <oupton@google.com>
> ---
>  arch/arm64/kvm/psci.c | 20 +++++++++++++++++---
>  1 file changed, 17 insertions(+), 3 deletions(-)
> 
> diff --git a/arch/arm64/kvm/psci.c b/arch/arm64/kvm/psci.c
> index db4056ecccfd..bb76be01abd2 100644
> --- a/arch/arm64/kvm/psci.c
> +++ b/arch/arm64/kvm/psci.c
> @@ -59,6 +59,17 @@ static void kvm_psci_vcpu_off(struct kvm_vcpu *vcpu)
>  	kvm_vcpu_kick(vcpu);
>  }
>  
> +static inline bool kvm_psci_valid_affinity(struct kvm_vcpu *vcpu,
> +					   unsigned long affinity)
> +{
> +	unsigned long mask = MPIDR_HWID_BITMASK;
> +
> +	if (vcpu_mode_is_32bit(vcpu))
> +		mask &= ~((u32) 0);

I don't think we need this anymore since 5.7:

- fdc9999e20cd ("KVM: arm64: PSCI: Forbid 64bit functions for 32bit
  guests") guarantees that the guest can't trick KVM into using the
  SMC64 functions.

- with 2890ac993daa ("KVM: arm64: PSCI: Narrow input registers when
  using 32bit functions"), the registers are always narrowed down to
  32bit

Put the two together, and you can't have a 32bit guest issuing a PSCI
operation with crap in the upper 32bits.

> +
> +	return !(affinity & ~mask);

So the whole helper can now be rewritten as

	return !(affinity & ~MPIDR_HWID_BITMASK);

> +}
> +
>  static unsigned long kvm_psci_vcpu_on(struct kvm_vcpu *source_vcpu)
>  {
>  	struct vcpu_reset_state *reset_state;
> @@ -66,9 +77,9 @@ static unsigned long kvm_psci_vcpu_on(struct kvm_vcpu *source_vcpu)
>  	struct kvm_vcpu *vcpu = NULL;
>  	unsigned long cpu_id;
>  
> -	cpu_id = smccc_get_arg1(source_vcpu) & MPIDR_HWID_BITMASK;
> -	if (vcpu_mode_is_32bit(source_vcpu))
> -		cpu_id &= ~((u32) 0);
> +	cpu_id = smccc_get_arg1(source_vcpu);
> +	if (!kvm_psci_valid_affinity(source_vcpu, cpu_id))
> +		return PSCI_RET_INVALID_PARAMS;
>  
>  	vcpu = kvm_mpidr_to_vcpu(kvm, cpu_id);
>  
> @@ -126,6 +137,9 @@ static unsigned long kvm_psci_vcpu_affinity_info(struct kvm_vcpu *vcpu)
>  	target_affinity = smccc_get_arg1(vcpu);
>  	lowest_affinity_level = smccc_get_arg2(vcpu);
>  
> +	if (!kvm_psci_valid_affinity(vcpu, target_affinity))
> +		return PSCI_RET_INVALID_PARAMS;
> +
>  	/* Determine target affinity mask */
>  	target_affinity_mask = psci_affinity_mask(lowest_affinity_level);
>  	if (!target_affinity_mask)

Otherwise, looks good to me.

Thanks,

	M.

-- 
Without deviation from the norm, progress is not possible.