On 13/03/2022 11:47, Michael Walle wrote:
> Hi Krzysztof,
>
> Am 2022-03-13 10:47, schrieb Krzysztof Kozlowski:
>> On 13/03/2022 01:25, Michael Walle wrote:
>>> The MDIO controller has support to release the internal PHYs from
>>> reset
>>> by specifying a second memory resource. This is different between the
>>> currently supported SparX-5 and the LAN966x. Add a new compatible to
>>> distiguish between these two.
Typo here, BTW.
>>>
>>> Signed-off-by: Michael Walle <michael@xxxxxxxx>
>>> ---
>>> Documentation/devicetree/bindings/net/mscc-miim.txt | 2 +-
>>> 1 file changed, 1 insertion(+), 1 deletion(-)
>>>
>>> diff --git a/Documentation/devicetree/bindings/net/mscc-miim.txt
>>> b/Documentation/devicetree/bindings/net/mscc-miim.txt
>>> index 7104679cf59d..a9efff252ca6 100644
>>> --- a/Documentation/devicetree/bindings/net/mscc-miim.txt
>>> +++ b/Documentation/devicetree/bindings/net/mscc-miim.txt
>>> @@ -2,7 +2,7 @@ Microsemi MII Management Controller (MIIM) / MDIO
>>> =================================================
>>>
>>> Properties:
>>> -- compatible: must be "mscc,ocelot-miim"
>>> +- compatible: must be "mscc,ocelot-miim" or "mscc,lan966x-miim"
>>
>> No wildcards, use one, specific compatible.
>
> I'm in a kind of dilemma here, have a look yourself:
> grep -r "lan966[28x]-" Documentation
>
> Should I deviate from the common "name" now? To make things
> worse, there was a similar request by Arnd [1]. But the
> solution feels like cheating ("lan966x" -> "lan966") ;)
The previous 966x cases were added by one person from Microchip, so he
actually might know something. But do you know whether lan966x will
cover all current and future designs from Microchip? E.g. lan9669 (if
ever made) will be the same? Avoiding wildcard is the easiest, just
choose one implementation, e.g. "lan9662".
Different topic is that all current lan966[28] are from Microchip and
you still add Microsemi, even though it was acquired by Microchip.
That's an inconsistency which should be rather fixed.
>
> On a side note, I understand that there should be no wildcards,
> because the compatible should target one specific implementation,
> right? But then the codename "ocelot" represents a whole series of
> chips. Therefore, names for whole families shouldn't be used neither,
> right?
You're not adding "ocelot" now, so it is separate topic. However a
compatible like "mscc,ocelot" feels wrong, unless it is used as a
fallback (see: git grep 'apple,').
Best regards,
Krzysztof
