Re: [PATCH] arm64: dts: rockchip: Remove non-existing pwm-delay-us DT property

From: Javier Martinez Canillas
Date: Thu Mar 30 2023 - 19:00:16 EST


Brian Norris <briannorris@xxxxxxxxxxxx> writes:

Hello Brian,

> On Thu, Mar 30, 2023 at 3:03 PM Javier Martinez Canillas
> <javierm@xxxxxxxxxx> wrote:
>>
>> There is no neither a driver that parses this nor a DT binding schema that

Ups, I noticed now that there's an unnecessary "no" and it should be instead:

"There is neither a driver..."

>> documents it so let's remove it from the DTS files that make use of this.
>>
>> The properties that exist are post-pwm-on-delay-ms and pwm-off-delay-ms,
>> defined in the pwm-backlight DT binding. So probably what these DTS want
>> is something like following:
>>
>> backlight: backlight {
>> compatible = "pwm-backlight";
>> enable-gpios = <&gpio4 21 GPIO_ACTIVE_HIGH>;
>> pinctrl-names = "default";
>> pinctrl-0 = <&bl_en>;
>> pwms = <&pwm1 0 1000000 0>;
>> post-pwm-on-delay-ms = <10>;
>> pwm-off-delay-ms = <10>;
>> };
>>
>> But that should be follow-up change if that is the case. Because otherwise
>> it would be change in behaviour, since currently pwm-delay-us is a no-op.
>>
>> Signed-off-by: Javier Martinez Canillas <javierm@xxxxxxxxxx>
>
> pwm-delay-us seems to have been a downstream-only ("CHROMIUM", if
> you're familiar with ChromiumOS kernel parlance) change that seems
> like a combination of the two now-upstream properties you point at. I

Yes, that's what I found too. So it seems that this was an oversight when
the DTS for these Chromebooks were upstreamed.

> looked through the first use of pwm-delay-us on RK3399 Gru systems,
> and I can't find a spec reference that said it was needed; perhaps it
> was needless copy/paste imitation?
>
> So, simple deletion is probably fine:
>
> Reviewed-by: Brian Norris <briannorris@xxxxxxxxxxxx>
>

Thanks for the confirmation and review!

--
Best regards,

Javier Martinez Canillas
Core Platforms
Red Hat