Re: [PATCH] iommu/io-pgtable-arm: Add misisng concatenated PGD cases

[Will Deacon <will@kernel.org>]

On Wed, Dec 10, 2025 at 12:06:10PM +0000, Robin Murphy wrote:
> On 2025-12-10 12:42 am, Will Deacon wrote:
> > On Tue, Dec 09, 2025 at 01:33:36PM +0000, Robin Murphy wrote:
> > > On 2025-12-09 12:37 pm, Mostafa Saleh wrote:
> > > > On Tue, Dec 09, 2025 at 11:34:34AM +0000, Robin Murphy wrote:
> > > > > On 2025-11-30 7:45 pm, Mostafa Saleh wrote:
> > > > > > arm_lpae_concat_mandatory() assumes that OAS >= IAS which is not
> > > > > > correct for SMMUs supporting AArch32, and have OAS = 32/36 bits,
> > > > > > as IAS would be 40 bits.
> > > > > 
> > > > > But that is only when *using* AArch32 format. The bit in chapter 3.4 of the
> > > > > SMMU architecture is talking about the maximum IAS that an SMMU
> > > > > implementation needs to be able to accommodate based on its configuration,
> > > > > but it does then attempt to clarify that the actual IPA size in use by any
> > > > > given context should depend on the VMSA format in use:
> > > > > 
> > > > > "VMSAv8-32 LPAE always supports an IPA size of 40 bits, whereas VMSAv8-64
> > > > > and VMSAv9-128 limits the maximum IPA size to the maximum PA size."
> > > > > 
> > > > > Rule R_SRKBC in the Arm ARM lays out the exact T0SZ constraints with this
> > > > > AArch32/AArch64 detail.
> > > > 
> > > > I see, thanks a lot for the explanation, I got confused by the this
> > > > statement:
> > > > Note: If AArch32 is implemented, IAS == MAX(40, OAS), otherwise IAS == OAS.
> > > 
> > > Indeed, that appears confusingly contradictory; I've filed a bug.
> > 
> > I think the spec has always been worded like this. My reading is that, in
> > isolation:
> > 
> >    - VMSAv8-32 LPAE always uses a 40-bit IAS
> >    - VMSAv8-64 has IAS == OAS and this can be smaller than 40 bits
> > 
> > so if AArch32 is implemented, we know that the hardware supports at
> > least a 40-bit IAS and in that case the VMSAv8-64 IAS can be bigger
> > than the OAS.
> 
> No, the VMSAv8-64 IAS can never be bigger than OAS, as that would violate
> VMSA (see rules R_DTLMN and R_SRKBC). The SMMU spec seems mostly fixated on
> the notional maximum IAS that an SMMU must accommodate in general across all
> supported formats; the limitations of using any particular format must still
> apply though (similarly, the fact that AArch32 has a fixed 40-bit output
> doesn't mean that OAS and S2PS don't still matter for AArch64 format.)

A VMSAv8-32 stage-1 could still input a 40-bit IAS into a VMSAv8-64
stage-2 when the OAS < 40-bit though, right? If that's the only case
where this happens, then I agree we should ditch the complexity from the
driver on the grounds that we don't ever use the 32-bit formats.

Will