[PATCH RFC v2 2/2] pinctrl: add pinctrl gpio binding support

[Dong Aisheng]

On Tue, May 22, 2012 at 1:11 AM, Stephen Warren <swarren at wwwdotorg.org> wrote:
> On 05/21/2012 06:39 AM, Dong Aisheng wrote:
>> On Sat, May 19, 2012 at 04:05:46AM +0800, Stephen Warren wrote:
>>> On 05/18/2012 07:12 AM, Dong Aisheng wrote:
>>>> The gpio ranges standard dt binding format is
>>>> <&gpio $gpio_offset $pin_offset $npin>
>>>>
>>>> The core will parse and register the pinctrl gpio ranges
>>>> from device tree.
> ...
>>> Do you need to xxx_get(ranges[i].gc) to prevent it going away, and put()
>>> it when removing the ranges?
>>
>> How would you suggest to implement xxx_get(ranges[i].gc)?
>> Since the parameter is a struct gpiochip, my first sense is that it may be
>> provided by gpio subsystem, but i did not find such a function.
>> Looking at gpio subsystem, i also can't see it should provide such function.
>>
>> I wonder if we need to implement it, if gpiochip is gone way,
>> the error will be detected in the higher gpio layer and will not pass
>> down to pinctrl.
>
> Yes, it looks like we should add new APIs for this; we need to
> try_module_get() on the module containing the GPIO chip so it doesn't
> disappear, similar to what gpio_request() does.
yes, i checked the gpio request code.
It looks to me try_module_get is just ok like:
if (!try_module_get(ranges[i].gc.owner))
    goto done;
Will add it.
Thanks for the info.

Regards
Dong Aisheng