Re: [PATCH] dt-bindings: leds: Allow differently named multicolor leds

["J. Neuschäfer" <j.ne@posteo.net>]

On Tue, Feb 11, 2025 at 08:10:28PM +0100, Krzysztof Kozlowski wrote:
> On 11/02/2025 19:50, J. Neuschäfer wrote:
> > On Tue, Feb 11, 2025 at 02:43:00PM +0000, Lee Jones wrote:
> >> On Sun, 09 Feb 2025, J. Neuschäfer via B4 Relay wrote:
> >>
> >>> From: "J. Neuschäfer" <j.ne@posteo.net>
> >>>
> >>> In some cases, a board may have multiple multi-leds, which can't be
> >>> distinguished by unit address. In such cases it should be possible to
> >>> name them differently, for example multi-led-a and multi-led-b.
> >>> This patch adds another node name pattern to leds-class-multicolor.yaml
> >>> to allow such names.
> >>
> >> Which H/W needs this?  Is it upstream?  Where is the doc / usage?
> > 
> > I encountered this situation while upstreaming the LANCOM NWAPP2 board,
> > which has multiple LED-group-based multicolor LEDs:
> > 
> >   https://lore.kernel.org/lkml/20250102-mpc83xx-v1-16-86f78ba2a7af@posteo.net/
> 
> Which LEDs are these?

These and a few more:

	led-power {
		label = "multicolor:power";
		compatible = "leds-group-multicolor";
		color = <LED_COLOR_ID_MULTI>;
		function = LED_FUNCTION_POWER;
		leds = <&led_power_red>, <&led_power_green>;
	};

	led-wlan-link {
		label = "multicolor:wlan-link";
		compatible = "leds-group-multicolor";
		color = <LED_COLOR_ID_MULTI>;
		function = LED_FUNCTION_WLAN;
		leds = <&led_wlan_link_red>, <&led_wlan_link_green>;
	};

According to the leds-class-multicolor.yaml binding, they should be
named "multi-led", optionally with a unit address. Unit addresses don't
make a lot of sense, as these nodes don't have (or need) a reg property.
They can't, however, have the same name, which brings me to the idea of
this patch: To allow different names that start with "multi-led-".

> I don't see multi-led there node name at all.

This was my mistake while writing the NWAPP2 devicetree.

> 
> This patch must come with user. It's fine to send the user separately,
> but please provide lore link to exact user.
> 
> Otherwise what certainty we have that this change is needed in the first
> place?

For ease of review, I'll include this patch in v2 of the NWAPP2 series,
and extend the commit message.

> 
> > 
> > Since they are based on leds-group-multicolor, they don't have a unit
> > address, but there is more than one on the same level (as direct
> > sub-nodes of the DT root node).
> 
> Which binding is this?

Documentation/devicetree/bindings/leds/leds-group-multicolor.yaml
  specifies compatible = "leds-group-multicolor", and includes:

Documentation/devicetree/bindings/leds/leds-class-multicolor.yaml
  defines node name pattern of "^multi-led(@[0-9a-f])?$"


Best regards,
J. Neuschäfer