Re: [PATCH v20 1/4] dt-bindings: i2c: Split AST2600 binding into a new YAML

[Krzysztof Kozlowski <krzk@kernel.org>]

On 24/10/2025 10:40, Jeremy Kerr wrote:
> Hi Krzysztof,
> 
>> On 24/10/2025 09:56, Jeremy Kerr wrote:
>>> Hi Krzysztof,
>>>
>>>> Although now I saw next patch, so clearly this commit is
>>>> incomplete.
>>>
>>> The split that Ryan has done here - by shifting to an identical
>>> separate
>>> binding, then making the changes explicit - allows us to review the
>>> actual changes without losing them in the move. Sounds like a
>>> benefit to
>>> me?
>>
>> Not related. I commented that rationale is incomplete. We do not move
>> parts of bindings because new device is someway different. There are
>> hundreds of bindings which cover different devices. We move them
>> because the binding is different.
> 
> OK, but in that case I think we're after guidance on the threshold for
> "difference" here.
> 
>> Not much different than every other soc. All of them are separate IPs.
>> Look at any Samsung, NXP or Qualcomm binding. Separate IPs.
> 
> So, something like this?
> 
>     allOf:
>       - $ref: /schemas/i2c/i2c-controller.yaml#
>       - if:
>           properties:
>             compatible:
>               contains:
>                 enum:
>                   - aspeed,ast2600-i2c-bus
>         then:
>           required:
>             - aspeed,global-regs


else:
  properties:
   ... : false

> 
> 
> I can't see how we could represent aspeed,transfer-mode though, as it's
> optional on aspeed,ast2600-i2c-bus, but prohibited on others. Any hints
> on that?


It's shown in the example-schema, if we go that way.

Best regards,
Krzysztof