Re: [PATCH] dt-bindings: spi: nuvoton,ma35d1-qspi: Disallow GPIO chip selects

[email protected] 於 2026/8/3 上午 10:48 寫道:
> Thank you for your contribution! Sashiko AI review found 1 potential issue(s) to consider:
> - [Medium] The commit disallows the `cs-gpios` property in the DT binding based solely on a current Linux driver limitation, violating the core Device Tree principle that bindings describe hardware, not software capabilities.
> --
>
> --- Patch [1]: [PATCH] dt-bindings: spi: nuvoton,ma35d1-qspi: Disallow GPIO chip selects ---
> commit 3aecb6991adf3696410ad69c99107c7d732fd078
> Author: Chi-Wen Weng <[email protected]>
>
> dt-bindings: spi: nuvoton,ma35d1-qspi: Disallow GPIO chip selects
>
> This commit updates the nuvoton,ma35d1-qspi binding to explicitly disallow
> the cs-gpios property. The change is made because the current driver only
> supports native chip selects, preventing the use of GPIO-based chip selects.
>
>> diff --git a/Documentation/devicetree/bindings/spi/nuvoton,ma35d1-qspi.yaml b/Documentation/devicetree/bindings/spi/nuvoton,ma35d1-qspi.yaml
>> --- a/Documentation/devicetree/bindings/spi/nuvoton,ma35d1-qspi.yaml
>> +++ b/Documentation/devicetree/bindings/spi/nuvoton,ma35d1-qspi.yaml
>> @@ -33,6 +33,8 @@ properties:
>>       maximum: 2
>>       default: 2
>>   
>> +  cs-gpios: false
>> +
> [Severity: Medium]
> Does this restriction violate the device tree principle that bindings should
> describe hardware rather than software capabilities?
>
> By explicitly disallowing cs-gpios in the properties block based on a current
> Linux driver limitation, board designers who route a standard GPIO to a SPI
> peripheral's chip select pin will be incorrectly prevented from describing
> their actual physical board wiring. When running make dtbs_check on a board
> device tree that uses cs-gpios for this controller, the schema will reject it.
>
> Should the hardware schema allow standard hardware properties like cs-gpios
> even if the current software driver lacks support for them?
>
>>   required:
>>     - compatible
>>     - reg

Hi,

Thanks for pointing this out.

I agree that the binding should describe valid hardware and board wiring 
rather than the current driver limitation.

I will send a v2 that keeps cs-gpios supported by the binding and 
instead enables GPIO descriptor-based chip-select handling in the MA35D1 
QSPI driver. I will also keep active-high chip selects restricted to 
GPIO-based chip selects, since the controller's native chip selects 
currently support active-low operation only.

Best regards,
Chi-Wen