On Wed 2016-09-28 12:02:41, Florian Vaussard wrote:

Hi Pavel,

Le 24. 09. 16 à 13:58, Pavel Machek a écrit :

quoted

Hi!

quoted

+Example
+=======
+
+led1: ncp5623@38 {
+	#address-cells = <1>;
+	#size-cells = <0>;
+	compatible = "onnn,ncp5623";
+	reg = <0x38>;
+	onnn,led-iref-microamp = <10>;
+
+	led1r@0 {
+		label = "ncp:power:red";
+		linux,default-trigger = "default-on";

...

quoted

+	led1b@1 {
+		label = "ncp:power:blue";
+		reg = <1>;

Actually... the three LEDs are packaged such as this is one colorful
light to the user, right? Some day we'll need to group them, so that
kernel can automatically tell this is one led, and probably add extra
attributes, such as values that produce white light.

Actually, it's up to the hardware designer to choose. On my board for instance,
this chip is driving an RGB LED, but it can really drive three independent LEDs
if you want.

Yup. And driving RGB LED is really a bit different from driving three
independent LEDs: you'd for example like to be able to set the RGB LED
to white, and you need to know relative intensities for that.

So it would be good to have hardware description that captures
difference between RGB LED and three LEDs.

(And then, we'll want pattern engine to drive that. One day :-) ).

I agree that the RGB case is quite common nowadays and currently not very well
managed by the LED subsystem. But I do not think that this is specific to this
driver.

No, it is not.

Best regards,
									Pavel
-- 
(english) http://www.livejournal.com/~pavelmachek
(cesky, pictures) http://atrey.karlin.mff.cuni.cz/~pavel/picture/horses/blog.html