Re: [PATCH RFC v2 1/3] arm64: dts: qcom: sm8750: Add Iris VPU v3.5

[Dmitry Baryshkov]

On Tue, Aug 12, 2025 at 09:01:36PM +0530, Vikash Garodia wrote:
> 
> On 8/12/2025 8:09 PM, Dmitry Baryshkov wrote:
> > On Tue, Aug 12, 2025 at 04:21:12PM +0200, Konrad Dybcio wrote:
> >> On 8/6/25 2:38 PM, Krzysztof Kozlowski wrote:
> >>> Add Iris video codec to SM8750 SoC, which comes with significantly
> >>> different powering up sequence than previous SM8650, thus different
> >>> clocks and resets.  For consistency keep existing clock and clock-names
> >>> naming, so the list shares common part.
> >>>
> >>> Signed-off-by: Krzysztof Kozlowski <krzysztof.kozlowski@xxxxxxxxxx>
> >>>
> >>> ---
> >>
> >> [...]
> >>
> >>> +			iris_opp_table: opp-table {
> >>> +				compatible = "operating-points-v2";
> >>> +
> >>> +				opp-240000000 {
> >>> +					opp-hz = /bits/ 64 <240000000>;
> >>> +					required-opps = <&rpmhpd_opp_low_svs_d1>,
> >>> +							<&rpmhpd_opp_low_svs_d1>;
> >>> +				};
> >>> +
> >>> +				opp-338000000 {
> >>> +					opp-hz = /bits/ 64 <338000000>;
> >>> +					required-opps = <&rpmhpd_opp_low_svs>,
> >>> +							<&rpmhpd_opp_low_svs>;
> >>> +				};
> >>> +
> >>> +				opp-420000000 {
> >>> +					opp-hz = /bits/ 64 <420000000>;
> >>> +					required-opps = <&rpmhpd_opp_svs>,
> >>> +							<&rpmhpd_opp_svs>;
> >>> +				};
> >>> +
> >>> +				opp-444000000 {
> >>> +					opp-hz = /bits/ 64 <444000000>;
> >>> +					required-opps = <&rpmhpd_opp_svs_l1>,
> >>> +							<&rpmhpd_opp_svs_l1>;
> >>> +				};
> >>> +
> >>> +				opp-533333334 {
> >>> +					opp-hz = /bits/ 64 <533333334>;
> >>> +					required-opps = <&rpmhpd_opp_nom>,
> >>> +							<&rpmhpd_opp_nom>;
> >>> +				};
> >>
> >> There's an additional OPP: 570 MHz @ NOM_L1
> >>
> >> +Dmitry, Vikash, please make sure you're OK with the iommu entries
> > 
> > We still don't have a way to describe it other way at this point.
> 
> I could validate the extended "iommu-map-masks" proposal. Given that we have a

Was it posted? If not, let's get it ASAP.

> new binding for SM8750 [1] , does it make sense to add iommus min/max as [1,5] ?

Why [1, 5]? It should be [1, 2] or just [1, 1] + your proposal.

> such that later if new property is introduced "iommu-map-mask", it does not
> break ABI.
> 
> iommus = <&apps_smmu 0x1940 0>;
> iommu-map-masks = <0 &apps_smmu 0x1947 1 0>;
> 
> [1] https://lore.kernel.org/all/20250804-sm8750-iris-v2-1-6d78407f8078@xxxxxxxxxx/
> 
> Regards,
> Vikash
> > 
> >>
> >> the other properties look OK
> >>
> >> Konrad
> > 

-- 
With best wishes
Dmitry