Re: [PATCH] dt-bindings:leds: Add function define for POE

[Krzysztof Kozlowski <krzk@kernel.org>]

On 23/02/2025 21:54, Chris Packham wrote:
> Hi Krzysztof,
> 
> On 17/02/2025 22:13, Krzysztof Kozlowski wrote:
>> On 16/02/2025 22:37, Tony O'Brien wrote:
>>> Hi Krzysztof -
>>>
>>>> Where did these two reviews happen?
>>> They were in-house reviews.  Please feel free to remove them from the
>>> patch.
>>>
>>>> Where is any user of this?
>>> We are adding Kernel control of PoE LEDs and thought this might be
>>> useful to others, maybe those working on netdev, or anyone implementing
>>> PoE on their devices.  Also, the Kernel >> Docs >> LEDs page states:
>>> "If required color or function is missing, please submit a patch to
>>> linux-leds@vger.kernel.org", which is included here.
>> You did not answer: where is the user? By "we are adding" you mean
>> downstream?
> 
> It's downstream. One of our PoE switches in development which uses a 


So for me that's a no. We don't care about downstream. Otherwise we
would need to accept whatever else people invented and never bothered
with upstreaming.

> pca9552 LED controller. I suggested to Tony that we get this upstream as 
> I know there is some work going on to support PoE PSEs in netdev. I did 
> wonder if we wanted to make this more specific i.e. have "pse" and "pd" 
> as different functions but something like "poe" seemed fine as you're 
> not going to have something that is both a PSE and a PD on the same port.

To me, this is not a catalog of all possible LED functions. Come with
any sort of user - either driver or DTS (and I still remember discussion
with Hans de Goede about taking such patches without DTS user, but that
was under condition there is driver user).

Best regards,
Krzysztof