Re: [PATCH] dt-bindings: remoteproc: Add missing (unevaluated|additional)Properties on child nodes

[Krzysztof Kozlowski <krzysztof.kozlowski@linaro.org>]

On 25/08/2022 16:13, Rob Herring wrote:
> On Thu, Aug 25, 2022 at 3:23 AM Krzysztof Kozlowski
> <krzysztof.kozlowski@linaro.org> wrote:
>>
>> On 23/08/2022 17:56, Rob Herring wrote:
>>> In order to ensure only documented properties are present, node schemas
>>> must have unevaluatedProperties or additionalProperties set to false
>>> (typically).
>>>
>>> Signed-off-by: Rob Herring <robh@kernel.org>
>>> ---
>>>  .../devicetree/bindings/remoteproc/qcom,sc7180-mss-pil.yaml      | 1 +
>>>  .../devicetree/bindings/remoteproc/qcom,sc7280-mss-pil.yaml      | 1 +
>>>  .../devicetree/bindings/remoteproc/qcom,sc7280-wpss-pil.yaml     | 1 +
>>>  3 files changed, 3 insertions(+)
>>>
>>> diff --git a/Documentation/devicetree/bindings/remoteproc/qcom,sc7180-mss-pil.yaml b/Documentation/devicetree/bindings/remoteproc/qcom,sc7180-mss-pil.yaml
>>> index e76c861165dd..e4a7da8020f4 100644
>>> --- a/Documentation/devicetree/bindings/remoteproc/qcom,sc7180-mss-pil.yaml
>>> +++ b/Documentation/devicetree/bindings/remoteproc/qcom,sc7180-mss-pil.yaml
>>> @@ -140,6 +140,7 @@ properties:
>>>
>>>    glink-edge:
>>>      $ref: qcom,glink-edge.yaml#
>>> +    unevaluatedProperties: false
>>
>> Is it actually needed? The qcom,glink-edge.yaml has
>> additionalProperties:false, so I expect it to complain if anything
>> appears here.
> 
> Perhaps not, but I'm trying to come up with a meta-schema to check
> these though I'm not sure I can get to no warnings which is how I
> found all these cases. The main remaining warnings are bus child node
> pattern schemas which can perhaps be handled with
> 'additionalProperties: true'. The rule I have says if properties or
> patternProperties is present then unevaluatedProperties or
> additionalProperties must be. To handle this case, I think we'd have
> to walk the $ref and check it.
> 
> Anyways, we can hold off on this one until when and if there's a
> meta-schema in place.

For me adding unevaluatedProp:false everywhere with $ref is okay and it
makes the code easier to read - no need to dive into referenced schema
to remember if it allows or does not allow additional properties. It is
also a safer choice if referenced schema forgot to set additionalProp:false.

However if referenced schema has additionalProp:false, then
unevaluatedProp:false here is redundant and question is whether the
redundancy is worth additional readability/obviousness.

To me, unevaluatedProp:false here would during review save time - no
need to jump into referenced schema to check what is there. If we make
it a rule / coding convention, then I am in.

Reviewed-by: Krzysztof Kozlowski <krzysztof.kozlowski@linaro.org>

Best regards,
Krzysztof