Re: [PATCH/RFC v10 03/19] DT: leds: Add led-sources property

From: Rob Herring
Date: Fri Jan 16 2015 - 08:49:08 EST


On Fri, Jan 16, 2015 at 3:07 AM, Jacek Anaszewski
<j.anaszewski@xxxxxxxxxxx> wrote:
> On 01/15/2015 03:24 PM, Rob Herring wrote:
>>
>> On Tue, Jan 13, 2015 at 2:42 AM, Jacek Anaszewski
>> <j.anaszewski@xxxxxxxxxxx> wrote:
>>>
>>> On 01/12/2015 05:55 PM, Rob Herring wrote:
>>>>
>>>>
>>>> Adding Mark B and Liam...
>>>>
>>>> On Mon, Jan 12, 2015 at 10:10 AM, Jacek Anaszewski
>>>> <j.anaszewski@xxxxxxxxxxx> wrote:
>>>>>
>>>>>
>>>>> On 01/12/2015 02:52 PM, Rob Herring wrote:
>>>>>>
>>>>>>
>>>>>>
>>>>>> On Mon, Jan 12, 2015 at 2:32 AM, Jacek Anaszewski
>>>>>> <j.anaszewski@xxxxxxxxxxx> wrote:
>>>>>>>
>>>>>>>
>>>>>>> On 01/09/2015 07:33 PM, Rob Herring wrote:
>>>>>>>>
>>>>>>>>
>>>>>>>> On Fri, Jan 9, 2015 at 9:22 AM, Jacek Anaszewski
>>>>>>>> <j.anaszewski@xxxxxxxxxxx> wrote:
>>>>>>>>>
>>>>>>>>>
>>>>>>>>> Add a property for defining the device outputs the LED
>>>>>>>>> represented by the DT child node is connected to.
>>>>
>>>>
>>>>
>>>> [...]
>>>>
>>>>>>>>> b/Documentation/devicetree/bindings/leds/common.txt
>>>>>>>>> index a2c3f7a..29295bf 100644
>>>>>>>>> --- a/Documentation/devicetree/bindings/leds/common.txt
>>>>>>>>> +++ b/Documentation/devicetree/bindings/leds/common.txt
>>>>>>>>> @@ -1,6 +1,10 @@
>>>>>>>>> Common leds properties.
>>>>>>>>>
>>>>>>>>> Optional properties for child nodes:
>>>>>>>>> +- led-sources : Array of bits signifying the LED current regulator
>>>>>>>>> outputs the
>>>>>>>>> + LED represented by the child node is connected to
>>>>>>>>> (1
>>>>>>>>> -
>>>>>>>>> the LED
>>>>>>>>> + is connected to the output, 0 - the LED isn't
>>>>>>>>> connected
>>>>>>>>> to the
>>>>>>>>> + output).
>>>>>>>>
>>>>>>>>
>>>>>>>>
>>>>>>>>
>>>>>>>>
>>>>>>>> Sorry, I just don't understand this.
>>>>>>>
>>>>>>>
>>>>>>>
>>>>>>>
>>>>>>>
>>>>>>> In some Flash LED devices one LED can be connected to one or more
>>>>>>> electric current outputs, which allows for multiplying the maximum
>>>>>>> current allowed for the LED. Each sub-LED is represented by a child
>>>>>>> node in the DT binding of the Flash LED device and it needs to
>>>>>>> declare
>>>>>>> which outputs it is connected to. In the example below the
>>>>>>> led-sources
>>>>>>> property is a two element array, which means that the flash LED
>>>>>>> device
>>>>>>> has two current outputs, and the bits signify if the LED is connected
>>>>>>> to the output.
>>>>>>
>>>>>>
>>>>>>
>>>>>>
>>>>>> Sounds like a regulator for which we already have bindings for and we
>>>>>> have a driver for regulator based LEDs (but no binding for it).
>>>>>
>>>>>
>>>>>
>>>>>
>>>>> Do you think of drivers/leds/leds-regulator.c driver? This driver just
>>>>> allows for registering an arbitrary regulator device as a LED subsystem
>>>>> device.
>>>>>
>>>>> There are however devices that don't fall into this category, i.e. they
>>>>> have many outputs, that can be connected to a single LED or to many
>>>>> LEDs
>>>>> and the driver has to know what is the actual arrangement.
>>>>
>>>>
>>>>
>>>> We may need to extend the regulator binding slightly and allow for
>>>> multiple phandles on a supply property, but wouldn't something like
>>>> this work:
>>>>
>>>> led-supply = <&led-reg0>, <&led-reg1>, <&led-reg2>, <&led-reg3>;
>>>>
>>>> The shared source is already supported by the regulator binding.
>>>
>>>
>>>
>>> I think that we shouldn't split the LED devices into power supply
>>> providers and consumers as in case of generic regulators. From this
>>> point of view a LED device current output is a provider and a discrete
>>> LED element is a consumer. In this approach each discrete LED element
>>> should have a related driver which is not how LED devices are being
>>> handled in the LED subsystem, where there is a single binding for a LED
>>> device and there is a single driver for it which creates separate LED
>>> class devices for each LED connected to the LED device output. Each
>>> discrete LED is represented by a child node in the LED device binding.
>>>
>>> I am aware that it may be tempting to treat LED devices as common
>>> regulators, but they have their specific features which gave a
>>> reason for introducing LED class for them. Besides, there is already
>>> drivers/leds/leds-regulator.c driver for LED devices which support only
>>> turning on/off and setting brightness level.
>>>
>>> In your proposition a separate regulator provider binding would have
>>> to be created for each current output and a separate binding for
>>> each discrete LED connected to the LED device. It would create
>>> unnecessary noise in a dts file.
>>>
>>> Moreover, using regulator binding implies that we want to treat it
>>> as a sheer power supply for our device (which would be a discrete LED
>>> element in this case), whereas LED devices provide more features like
>>> blinking pattern and for flash LED devices - flash timeout, external
>>> strobe and flash faults.
>>
>>
>> Okay, fair enough. Please include some of this explanation in the
>> binding description.
>>
>> I do still have some concerns about led-sources and whether it can
>> support other scenarios. It is very much tied to the parent node. Are
>> there any cases where we don't want the LEDs to be sub nodes? Perhaps
>> the LEDs are on a separate daughterboard from the driver/supply and we
>> can have different drivers. It's a stretch maybe.
>
>
> I think it is. Such arrangements would introduce problems also to the
> other existing bindings. Probably not only LED subsystem related ones.
>
>> Or are there cases
>> where you need more information than just the connection?
>
>
> Currently I can't think of any.
>
> Modified rough proposal of the description:
>
>
> -Optional properties for child nodes:
> +LED and flash LED devices provide the same basic functionality as
> +current regulators, but extended with LED and flash LED specific +features
> like blinking patterns, flash timeout, flash faults and
> +external flash strobe mode.
> +
> +Many LED devices expose more than one current output that can be
> +connected to one or more discrete LED component. Since the arrangement
> +of connections can influence the way of the LED device initialization,
> +the LED components have to be tightly coupled with the LED device
> +binding. They are represented in the form of its child nodes.
> +
> +Optional properties for child nodes (if a LED device exposes only one
> +current output the properties can be placed directly in the LED device
> +node):

Why special case 1 output case? Just always require a child node.

> +- led-sources : Array of connection states between all LED current
> + sources exposed by the device and this LED (1 - this LED
> + is connected to the current output with index N, 0 -
> + this LED isn't connected to the current output with
> + index N); the mapping of N-th element of the array to
> + the physical device output should be defined in the LED
> + driver binding.

I think this should be a list of connected output numbers rather than
effectively a bitmask.

You may want to add something like led-output-cnt or led-driver-cnt in
the parent so you know the max list size.

Rob
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/