Re: [PATCH v2 2/3] dt-bindings: clock: add QCOM SM6125 display clock bindings

[Krzysztof Kozlowski]

On 02/03/2022 15:48, Bjorn Andersson wrote:
> On Wed 02 Mar 05:51 PST 2022, Krzysztof Kozlowski wrote:
> 
>> On 02/03/2022 13:54, Marijn Suijten wrote:
>>> On 2022-02-28 10:23:19, Krzysztof Kozlowski wrote:
>>>> On 27/02/2022 22:43, Dmitry Baryshkov wrote:
>>>>> On 27/02/2022 13:03, Krzysztof Kozlowski wrote:
>>>>>> On 26/02/2022 21:09, Marijn Suijten wrote:
>>>>>>> From: Martin Botka <martin.botka@xxxxxxxxxxxxxx>
>>>>>>>
>>>>>>> Add device tree bindings for display clock controller for
>>>>>>> Qualcomm Technology Inc's SM6125 SoC.
>>>>>>>
>>>>>>> Signed-off-by: Martin Botka <martin.botka@xxxxxxxxxxxxxx>
>>>>>>> ---
>>>>>>>   .../bindings/clock/qcom,dispcc-sm6125.yaml    | 87 +++++++++++++++++++
>>>>>>>   .../dt-bindings/clock/qcom,dispcc-sm6125.h    | 41 +++++++++
>>>>>>>   2 files changed, 128 insertions(+)
>>>>>>>   create mode 100644 Documentation/devicetree/bindings/clock/qcom,dispcc-sm6125.yaml
>>>>>>>   create mode 100644 include/dt-bindings/clock/qcom,dispcc-sm6125.h
>>>>>>>
>>>>>>> diff --git a/Documentation/devicetree/bindings/clock/qcom,dispcc-sm6125.yaml b/Documentation/devicetree/bindings/clock/qcom,dispcc-sm6125.yaml
>>>>>>> new file mode 100644
>>>>>>> index 000000000000..3465042d0d9f
>>>>>>> --- /dev/null
>>>>>>> +++ b/Documentation/devicetree/bindings/clock/qcom,dispcc-sm6125.yaml
>>>>>>> @@ -0,0 +1,87 @@
>>>>>>> +# SPDX-License-Identifier: (GPL-2.0-only OR BSD-2-Clause)
>>>>>>> +%YAML 1.2
>>>>>>> +---
>>>>>>> +$id: http://devicetree.org/schemas/clock/qcom,dispcc-sm6125.yaml#
>>>>>>> +$schema: http://devicetree.org/meta-schemas/core.yaml#
>>>>>>> +
>>>>>>> +title: Qualcomm Display Clock Controller Binding for SM6125
>>>>>>> +
>>>>>>> +maintainers:
>>>>>>> +  - Martin Botka <martin.botka@xxxxxxxxxxxxxx>
>>>>>>> +
>>>>>>> +description: |
>>>>>>> +  Qualcomm display clock control module which supports the clocks and
>>>>>>> +  power domains on SM6125.
>>>>>>> +
>>>>>>> +  See also:
>>>>>>> +    dt-bindings/clock/qcom,dispcc-sm6125.h
>>>>>>> +
>>>>>>> +properties:
>>>>>>> +  compatible:
>>>>>>> +    enum:
>>>>>>> +      - qcom,sm6125-dispcc
>>>>>>> +
>>>>>>> +  clocks:
>>>>>>> +    items:
>>>>>>> +      - description: Board XO source
>>>>>>> +      - description: Byte clock from DSI PHY0
>>>>>>> +      - description: Pixel clock from DSI PHY0
>>>>>>> +      - description: Pixel clock from DSI PHY1
>>>>>>> +      - description: Link clock from DP PHY
>>>>>>> +      - description: VCO DIV clock from DP PHY
>>>>>>> +      - description: AHB config clock from GCC
>>>>>>> +
>>>>>>> +  clock-names:
>>>>>>> +    items:
>>>>>>> +      - const: bi_tcxo
>>>>>>> +      - const: dsi0_phy_pll_out_byteclk
>>>>>>> +      - const: dsi0_phy_pll_out_dsiclk
>>>>>>> +      - const: dsi1_phy_pll_out_dsiclk
>>>>>>> +      - const: dp_phy_pll_link_clk
>>>>>>> +      - const: dp_phy_pll_vco_div_clk
>>>>>>> +      - const: cfg_ahb_clk
>>>>>>> +
>>>>>>> +  '#clock-cells':
>>>>>>> +    const: 1
>>>>>>> +
>>>>>>> +  '#power-domain-cells':
>>>>>>> +    const: 1
>>>>>>> +
>>>>>>> +  reg:
>>>>>>> +    maxItems: 1
>>>>>>> +
>>>>>>> +required:
>>>>>>> +  - compatible
>>>>>>> +  - reg
>>>>>>> +  - clocks
>>>>>>> +  - clock-names
>>>>>>> +  - '#clock-cells'
>>>>>>> +  - '#power-domain-cells'
>>>>>>> +
>>>>>>> +additionalProperties: false
>>>>>>> +
>>>>>>> +examples:
>>>>>>> +  - |
>>>>>>> +    #include <dt-bindings/clock/qcom,rpmcc.h>
>>>>>>> +    #include <dt-bindings/clock/qcom,gcc-sm6125.h>
>>>>>>> +    clock-controller@5f00000 {
>>>>>>> +      compatible = "qcom,sm6125-dispcc";
>>>>>>> +      reg = <0x5f00000 0x20000>;
>>>>>>> +      clocks = <&rpmcc RPM_SMD_XO_CLK_SRC>,
>>>>>>> +               <&dsi0_phy 0>,
>>>>>>> +               <&dsi0_phy 1>,
>>>>>>> +               <0>,
>>>>>>
>>>>>> This does not look like a valid phandle. This clock is required, isn't it?
>>>
>>> I remember it being used like this before, though upon closer inspection
>>> only qcom,gcc-msm8998.yaml uses it as example.
>>>
>>> The clock should be optional, in that case it is perhaps desired to omit
>>> it from clock-names instead, or pretend there's a `dsi1_phy 1`?
>>
>> I propose to omit it.
>>
> 
> The wire is there, it's only optional because we don't have the other
> side represented in DT yet.
> 
> I believe we started filling out 0s like this because omitting elements
> that are not yet possible to fill out means that the order will change
> as we add more functions, something Rob has objected to. Further more as
> we add more functions the existing dts will fail validation, even though
> the hardware hasn't changed.
> 
> 
> That said, even though we don't have the other piece on this particular
> platform we do know where this signal comes from. So we should be able
> to have a valid (or at least strongly plausible) example in the binding
> - and then fill out the dts with 0s to keep validation happy until the
> other pieces are filled out.

So based on this, this clock is not actually optional and the bindings
should stay like this. The example should be more-or-less complete, so
there is not much sense to have there clock "0". DTS is of course different.

BR,
Krzysztof