Re: [PATCH 2/2] ASoC: dt-bindings: microchip,sama7g5-i2smcc: Add 'sound-name-prefix' property

On 09/09/2024 14:28, Andrei.Simion@microchip.com wrote:

quoted

quoted

     minItems: 1

+  sound-name-prefix:
+    pattern: "^I2SMCC[0-9]$"

This does not look correct. Name/prefix can be anything matching real
hardware, why are you restricting it? How can you predict all names?

Based on the datasheet, the SoC(s) have the following naming conventions:
  - sama7g5: I2SMCC0 and I2SMCC1
  - sam9x60/sam9x75: I2SMCC

To accommodate these variations, I propose using a more relaxed pattern: "^I2SMCC(0-9)?$".
This pattern allows for both the fixed prefix and an optional single digit at the end.
What are your thoughts on this approach?

I understand this does not differ per board, because it is component of
the SoC, yet still I do not see any value in enforcing name.

quoted

quoted

+    $ref: /schemas/types.yaml#/definitions/string
+    description:
+      Unique prefixes for the sink/source names of the component, ensuring
+      distinct identification among multiple instances.

You are duplicating property definitions. This is not needed at all.
Maybe your schema misses $ref to common schema.

I understand the concern about duplicating property definitions.
In the current file, I have referenced `dai-common` as shown here:
https://github.com/torvalds/linux/blob/master/Documentation/devicetree/bindings/sound/microchip%2Csama7g5-i2smcc.yaml#L74C1-L75C27

Could you please confirm if this reference is correctly implemented,

Yeah, the dai-common $ref is correct, so you should not need it. Do you
see any warning?

or suggest any adjustments needed to align with the common schema?

I claim nothing has to be done and entire patch is redundant or not much
helpful. Your commit msg did not explain *why* this is needed and what
problem you are fixing, so what I can say? I don't know why should be
aligned because I do not understand the problem being fixed.

Best regards,
Krzysztof