Re: [PATCH 2/2] Documentation: gpio: board: describe the con_id parameter

[Dirk Behme <dirk.behme@gmail.com>]

On 21.08.2015 05:01, Alexandre Courbot wrote:
> On Fri, Aug 21, 2015 at 2:28 AM, Dirk Behme <dirk.behme@gmail.com> wrote:
>> On 25.07.2015 09:11, Dirk Behme wrote:
>>>
>>> The con_id parameter has to match the GPIO description and is
>>> automatically
>>> extended by the GPIO suffix if not NULL. I had to look into the code to
>>> understand this and properly find the GPIO I've been looking for, so
>>> document
>>> this.
>>>
>>> Signed-off-by: Dirk Behme <dirk.behme@gmail.com>
>>> ---
>>>    Documentation/gpio/board.txt    | 36
>>> ++++++++++++++++++++++++++++++++++++
>>>    Documentation/gpio/consumer.txt |  3 +++
>>>    2 files changed, 39 insertions(+)
>>>
>>> diff --git a/Documentation/gpio/board.txt b/Documentation/gpio/board.txt
>>> index 7605773..eac78fc 100644
>>> --- a/Documentation/gpio/board.txt
>>> +++ b/Documentation/gpio/board.txt
>>> @@ -48,6 +48,42 @@ This property will make GPIOs 15, 16 and 17 available
>>> to the driver under the
>>>    The led GPIOs will be active-high, while the power GPIO will be
>>> active-low (i.e.
>>>    gpiod_is_active_low(power) will be true).
>>>
>>> +If in the consumer device's node the property doesn't have a <function>-
>>> prefix,
>>> +the second parameter (con_id) of the gpiod_get*() functions has to be
>>> NULL:
>>> +
>>> +E.g. if in the above device tree example the GPIO array is just called
>>> 'gpios'
>>> +
>>> +       foo_device {
>>> +               compatible = "acme,foo";
>>> +               ...
>>> +               gpios = <&gpio 15 GPIO_ACTIVE_HIGH>, /* red */
>>> +                       <&gpio 16 GPIO_ACTIVE_HIGH>, /* green */
>>> +                       <&gpio 17 GPIO_ACTIVE_HIGH>; /* blue */
>>> +       };
>>> +
>>> +the gpiod_get*() calls will be:
>>> +
>>> +       struct gpio_desc *red, *green, *blue, *power;
>>> +
>>> +       red = gpiod_get_index(dev, NULL, 0, GPIOD_OUT_HIGH);
>>> +       green = gpiod_get_index(dev, NULL, 1, GPIOD_OUT_HIGH);
>>> +       blue = gpiod_get_index(dev, NULL, 2, GPIOD_OUT_HIGH);
>>> +
>>> +To summarize:
>>> +
>>> +The con_id string parameter has to be either NULL or the
>>> <function>-prefix
>>> +of the GPIO suffixes ("gpios" or "gpio", automatically looked up by the
>>> +gpiod functions internally):
>>> +
>>> +* If the GPIO description is just named with one of the GPIO suffixes
>>> +  ("gpios" or "gpio"), use NULL.
>>> +* If the GPIO description is prefixed with anything, e.g. "led-gpios",
>>> use the
>>> +  prefix without the "-" as con_id parameter (in this example "led").
>>> +
>>> +In case con_id is not NULL, the GPIO subsystem prefixes the GPIO suffix
>>> +("gpios" or "gpio") with the string passed in con_id to get the resulting
>>> string
>>> +(snprintf(... "%s-%s", con_id, gpio_suffixes[]).
>>> +
>>>    ACPI
>>>    ----
>>>    ACPI also supports function names for GPIOs in a similar fashion to DT.
>>> diff --git a/Documentation/gpio/consumer.txt
>>> b/Documentation/gpio/consumer.txt
>>> index 75542b9..47ce4fd 100644
>>> --- a/Documentation/gpio/consumer.txt
>>> +++ b/Documentation/gpio/consumer.txt
>>> @@ -39,6 +39,9 @@ device that displays digits), an additional index
>>> argument can be specified:
>>>                                            const char *con_id, unsigned int
>>> idx,
>>>                                            enum gpiod_flags flags)
>>>
>>> +For a more detailed description of the con_id parameter in the DeviceTree
>>> case
>>> +see Documentation/gpio/board.txt
>>> +
>>>    The flags parameter is used to optionally specify a direction and
>>> initial value
>>>    for the GPIO. Values can be:
>>
>>
>>
>> Any further comments on this? Or could this be applied?
>
> Using the "gpios" property is considered deprecated, so I am not sure
> that is a good idea to mention this too extensively in the boards file
> (people might be tempted to use that option while we clearly don't
> want them to).
>
> A precise description of the lookup method is clearly missing though.
> I don't know what the right place would be for it. consumer.txt?
> board.txt in a more concise (and with a big "DEPRECATED") way?

Ok, thanks!

In a previous comment there was the proposal to put it into board.txt, 
as this applies only to the device tree case and board.txt has a 
device tree section.

I'll drop the description of the deprecated part and send an update.

Best regards

Dirk