Re: [PATCH v1 2/6] dt-bindings: i2c: convert ocores binding to yaml

[Conor.Dooley]

On 06/06/2022 09:02, Krzysztof Kozlowski wrote:
> EXTERNAL EMAIL: Do not click links or open attachments unless you know the content is safe
> 
> On 05/06/2022 15:32, mail@xxxxxxxxxxx wrote:
>> From: Conor Dooley <conor.dooley@xxxxxxxxxxxxx>
>>
>> Convert the open cores i2c controller binding from text to yaml.
>>
>> Signed-off-by: Conor Dooley <conor.dooley@xxxxxxxxxxxxx>
>> ---
>>  .../devicetree/bindings/i2c/i2c-ocores.txt    |  78 -----------
>>  .../devicetree/bindings/i2c/i2c-ocores.yaml   | 132 ++++++++++++++++++
>>  2 files changed, 132 insertions(+), 78 deletions(-)
>>  delete mode 100644 Documentation/devicetree/bindings/i2c/i2c-ocores.txt
>>  create mode 100644 Documentation/devicetree/bindings/i2c/i2c-ocores.yaml
>>
---8<---
>> +  reg-shift:
>> +    $ref: /schemas/types.yaml#/definitions/uint32
> 
> No need for ref, it is coming from the dtschema.
> 
>> +    description: |
>> +      device register offsets are shifted by this value
> 
> I guess 0 is default?
> 

I had a look around:
sifive and the users of opencores,i2c-ocores use 2
aeroflexgaisler,i2cmst does not exist in a devicetree

From the driver, it looks like if neither this nor regstep
is present it just carries on without setting a value.
So yeah, I guess it is zero.

> 
>> +
>> +  regstep:
>> +    description: |
>> +      deprecated, use reg-shift above
>> +    deprecated: true
>> +
>> +  opencores,ip-clock-frequency:
>> +    $ref: /schemas/types.yaml#/definitions/uint32
>> +    description: |
>> +      Frequency of the controller clock in Hz. Mutually exclusive with clocks.
>> +      See the note above.
>> +
>> +required:
>> +  - compatible
>> +  - reg
>> +  - "#address-cells"
>> +  - "#size-cells"
>> +
>> +oneOf:
>> +  - required:
>> +      - opencores,ip-clock-frequency
>> +  - required:
>> +      - clocks
> 
> This is correct if your intention was to require one of these
> properties, which seems to match the old bindings.


Cool, thanks! I did test adding both and got a wall-of-text error
from dt_binding_check, so it seemed about right.

Thanks,
Conor.