Re: [PATCH v2 1/5] dt-bindings: arm: bcm: NSP: add Meraki MX64/MX65

[Matthew Hagan]

Hi Vladimir,

Many thanks for taking the time to review the submission.

On 11/06/2021 20:46, Vladimir Oltean wrote:

> On Fri, Jun 11, 2021 at 12:27:13AM +0100, Matthew Hagan wrote:
>> Add bindings for the Meraki MX64/MX65 series. Note this patch should be
>> applied on top of "dt-bindings: arm: bcm: add NSP devices to SoCs".
>>
>> Signed-off-by: Matthew Hagan <mnhagan88@xxxxxxxxx>
>> ---
>>  Documentation/devicetree/bindings/arm/bcm/brcm,nsp.yaml | 6 ++++++
>>  1 file changed, 6 insertions(+)
>>
>> diff --git a/Documentation/devicetree/bindings/arm/bcm/brcm,nsp.yaml b/Documentation/devicetree/bindings/arm/bcm/brcm,nsp.yaml
>> index 78dfa315f3d0..7d184ba7d180 100644
>> --- a/Documentation/devicetree/bindings/arm/bcm/brcm,nsp.yaml
>> +++ b/Documentation/devicetree/bindings/arm/bcm/brcm,nsp.yaml
>> @@ -62,6 +62,12 @@ properties:
>>            - enum:
>>                - brcm,bcm958625hr
>>                - brcm,bcm958625k
>> +              - meraki,mx64
>> +              - meraki,mx64-a0
>> +              - meraki,mx64w
>> +              - meraki,mx64w-a0
>> +              - meraki,mx65
>> +              - meraki,mx65w
>>            - const: brcm,bcm58625
>>            - const: brcm,nsp
>>  
>> -- 
>> 2.26.3
>>
> I think these compatibles describe SoCs, whereas Meraki MX64/MX65 are
> boards, so this is a miscategorization. Can you not just describe the
> Northstar Plus SoC that you are using in your compatible string?

My understanding is that the bcm958625hr and bcm958625k are dev boards
using the BCM58625 SoC variant of NSP. For reference, a close example
can be found in brcm,bcm4708.yaml in the same directory:

      - description: BCM53012 based boards
        items:
          - enum:
              - brcm,bcm953012er
              - brcm,bcm953012hr
              - brcm,bcm953012k
              - meraki,mr32
          - const: brcm,brcm53012
          - const: brcm,brcm53016
          - const: brcm,bcm4708

For the compatible string we would definitely need to specify a unique
device name for identification by the OS, rather than just the SoC.

Of course I could be mistaken. Will await confirmation on this.

Matthew