On 22/10/2022 05:17, Dmitry Baryshkov wrote: > On 22/10/2022 03:59, Krzysztof Kozlowski wrote: >> On 21/10/2022 12:55, Dmitry Baryshkov wrote: >>> Add missing compatibles used for Adreno SMMU on sc7280 and sm8450 >>> platforms and for the Qualcomm v2 SMMU used on SDM630 platform. >>> >>> Signed-off-by: Dmitry Baryshkov <dmitry.baryshkov@xxxxxxxxxx> >>> --- >>> Documentation/devicetree/bindings/iommu/arm,smmu.yaml | 11 +++++++++++ >>> 1 file changed, 11 insertions(+) >>> >>> diff --git a/Documentation/devicetree/bindings/iommu/arm,smmu.yaml b/Documentation/devicetree/bindings/iommu/arm,smmu.yaml >>> index 9066e6df1ba1..34ee33a62ba5 100644 >>> --- a/Documentation/devicetree/bindings/iommu/arm,smmu.yaml >>> +++ b/Documentation/devicetree/bindings/iommu/arm,smmu.yaml >>> @@ -28,6 +28,7 @@ properties: >>> - enum: >>> - qcom,msm8996-smmu-v2 >>> - qcom,msm8998-smmu-v2 >>> + - qcom,sdm630-smmu-v2 >> >> So qcom,adreno-smmu is not compatible with Adreno? See below. >> >>> - const: qcom,smmu-v2 >>> >>> - description: Qcom SoCs implementing "arm,mmu-500" >>> @@ -48,10 +49,20 @@ properties: >>> - qcom,sm8350-smmu-500 >>> - qcom,sm8450-smmu-500 >>> - const: arm,mmu-500 >>> + >>> + - description: Qcom Adreno GPUs implementing "arm,smmu-500" >>> + items: >>> + - enum: >>> + - qcom,sc7280-smmu-500 >>> + - qcom,sm8250-smmu-500 >>> + - const: qcom,adreno-smmu >>> + - const: arm,mmu-500 >>> - description: Qcom Adreno GPUs implementing "arm,smmu-v2" >>> items: >>> - enum: >>> + - qcom,msm8996-smmu-v2 >>> - qcom,sc7180-smmu-v2 >>> + - qcom,sdm630-smmu-v2 >> >> This does not look correct. The same compatible should not be present in >> two different setups. >> >> If qcom,msm8996-smmu-v2 is compatible with qcom,adreno-smmu, then your >> first hunk is not correct. > > Currently the qcom,adreno-smmu compat string is used as a flag, telling > the kernel that this SMMU instance needs some special setup to work with > Adreno GPU driver Indeed, I see the usage in DTS, > > For example, we have the following compat lists in the existing DT files: > - "qcom,msm8996-smmu-v2", "qcom,adreno-smmu", "qcom,smmu-v2" > - "qcom,msm8996-smmu-v2", "qcom,smmu-v2" // not handled by arm-qcom-smmu > > - "qcom,sdm630-smmu-v2", "qcom,adreno-smmu", "qcom,smmu-v2" > - "qcom,sdm630-smmu-v2", "qcom,smmu-v2" > > - "qcom,sdm845-smmu-v2", "qcom,adreno-smmu", "qcom,smmu-v2" > - "qcom,sdm845-smmu-500", "arm,mmu-500" > - "qcom,sdm845-smmu-v2", "qcom,smmu-v2" // special setup used on Cheza > > - "qcom,sm8250-smmu-500", "qcom,adreno-smmu", "arm,mmu-500" > - "qcom,sm8250-smmu-500", "arm,mmu-500" > > > As we are trying to refactor the IOMMU bindings, what would be your > recommendation? > > To introduce minimal changes, I wanted to have the following lists: > - "qcom,SOC-smmu-500", "qcom,adreno-smmu", "qcom,smmu-500", "arm,mmu-500" > > - "qcom,SOC-smmu-500", "qcom,smmu-500", "arm,mmu-500" > > However maybe you would prefer the following model: > > - "qcom,SOC-adreno-smmu-500", "qcom,adreno-smmu-500", "arm,mmu-500" > - "qcom,SOC-smmu-500", "qcom,smmu-500", "arm,mmu-500" If we started from scratch, I would prefer this one, however as DTSes are already using your previous method, It's fine. It's a bit confusing to have most specific compatible followed by different fallbacks, but we already have few cases for this (e.g. Renesas boards), so I guess it is fine here as well. At the end entire compatible list uniquely describes the hardware. > > > Or: > - "qcom,SOC-smmu-500", "qcom,smmu-500", "arm,mmu-500" + > 'qcom,adreno-smmu' flag/property? > > >> >>> - qcom,sdm845-smmu-v2 >>> - const: qcom,adreno-smmu >>> - const: qcom,smmu-v2 >> >> Best regards, >> Krzysztof >> > Best regards, Krzysztof