Re: [PATCH v5 1/7] KVM: arm64: Enforce strict SBZ checks in the FF-A proxy

[Sebastian Ene]

On Thu, Jun 25, 2026 at 02:16:40PM +0100, Will Deacon wrote:
> Hi all,
> 
> On Tue, Jun 23, 2026 at 11:53:48AM +0000, Sebastian Ene wrote:
> > Introduce a helper method ffa_check_unused_args_sbz to enforce strict
> > arguments checking when the hypervisor acts as a relayer between the
> > host and Trustzone.
> > 
> > Signed-off-by: Sebastian Ene <sebastianene@xxxxxxxxxx>
> > Reviewed-by: Vincent Donnefort <vdonnefort@xxxxxxxxxx>
> > ---
> >  arch/arm64/kvm/hyp/nvhe/ffa.c | 54 +++++++++++++++++++++++++++++++++++
> >  1 file changed, 54 insertions(+)
> > 
> > diff --git a/arch/arm64/kvm/hyp/nvhe/ffa.c b/arch/arm64/kvm/hyp/nvhe/ffa.c
> > index 1af722771178..78bb043b33ee 100644
> > --- a/arch/arm64/kvm/hyp/nvhe/ffa.c
> > +++ b/arch/arm64/kvm/hyp/nvhe/ffa.c
> > @@ -71,6 +71,20 @@ static u32 hyp_ffa_version;
> >  static bool has_version_negotiated;
> >  static hyp_spinlock_t version_lock;
> >  
> > +static bool ffa_check_unused_args_sbz(struct kvm_cpu_context *ctxt, int first_reg)
> > +{
> > +	DECLARE_REG(u32, func_id, ctxt, 0);
> > +	int reg, end_reg;
> > +
> > +	end_reg = ARM_SMCCC_IS_64(func_id) ? 17 : 7;
> > +	for (reg = first_reg; reg <= end_reg; reg++) {
> > +		if (cpu_reg(ctxt, reg))
> > +			return true;
> > +	}
> > +
> > +	return false;
> > +}

Hello Will,

> 
> Seb and I tried taking this for a spin on some Android devices and, sadly,
> it leads to fireworks. The reason is that the FF-A spec quietly changed
> the list of unused parameter registers for 64-bit SMCs from v1.1 to v1.2
> of the spec so that pre-existing calls were affected.
> 
> For example, in v1.1 a 64-bit RXTX_MAP only has x4-x7 as MBZ, whereas in
> v1.2 the same call has x4-x17 as SBZ.
> 
> We can follow the spec by predicating the additional check on the FF-A
> version being >= 1.2, but I'm not hopeful that existing drivers are
> compliant. I also suggest moving this patch to the end of the series in
> case we need to revert it.

I spinned up a new series (v6) which moves the check at the end of the
series and I made it so that it takes the ff-a version into account.

https://lore.kernel.org/all/20260626074545.433234-1-sebastianene@xxxxxxxxxx/

> 
> Cheers,
> 
> Will

Thanks
Sebastian