Re: [PATCH 3/5] gpio/omap: Add DT support to GPIO driver

[Stephen Warren <swarren@wwwdotorg.org>]

On 04/16/2013 05:14 PM, Jon Hunter wrote:
> 
> On 04/16/2013 05:11 PM, Stephen Warren wrote:
>> On 04/16/2013 01:27 PM, Jon Hunter wrote:
>>>
>>> On 04/16/2013 01:40 PM, Stephen Warren wrote:
>>>> On 04/15/2013 05:04 PM, Jon Hunter wrote:
>> ...
>>>>> If some driver is calling gpio_request() directly, then they will most
>>>>> likely just call gpio_to_irq() when requesting the interrupt and so the
>>>>> xlate function would not be called in this case (mmc drivers are a good
>>>>> example). So I don't see that as being a problem. In fact that's the
>>>>> benefit of this approach as AFAICT it solves this problem.
>>>>
>>>> Oh. That assumption seems very fragile. What about drivers that actually
>>>> do have platform data (or DT bindings) that provide both the IRQ and
>>>> GPIO IDs, and hence don't use gpio_to_irq()? That's entirely possible.
>>>
>>> Right. In the DT case though, if someone does provide the IRQ and GPIO
>>> IDs then at least they would use a different xlate function. Another
>>> option to consider would be defining the #interrupt-cells = <3> where we
>>> would have ...
>>>
>>> cell-#1 --> IRQ domain ID
>>> cell-#2 --> Trigger type
>>> cell-#3 --> GPIO ID
>>>
>>> Then we could have a generic xlate for 3 cells that would also request
>>> the GPIO. Again not sure if people are against a gpio being requested in
>>> the xlate but just an idea. Or given that irq_of_parse_and_map() calls
>>> the xlate, we could have this function call gpio_request() if the
>>> interrupt controller is a gpio and there are 3 cells.
>>
>> I rather dislike this approach since:
>>
>> a) It requires changes to the DT bindings, which are already defined.
>> Admittedly it's backwards-compatible, but still.
>>
>> b) There isn't really any need for the DT to represent this; the
>> GPIO+IRQ driver itself already knows which IRQ ID is which GPIO ID and
>> vice-versa (if the HW has such a concept), so there's no need for the DT
>> to contain this information. This seems like pushing Linux's internal
>> requirements into the design of the DT binding.
> 
> Yes, so the only alternative is to use irq_to_gpio to avoid this.
> 
>> c) I have the feeling that hooking the of_xlate function for this is a
>> bit of an abuse of the function.
> 
> I was wondering about that. So I was grep'ing through the various xlate
> implementations and found this [1]. Also you may recall that in the
> of_dma_simple_xlate() we call the dma_request_channel() to allocate the
> channel, which is very similar. However, I don't wish to get a
> reputation as abusing APIs so would be good to know if this really is
> abuse or not ;-)
> 
> Cheers
> Jon
> 
> [1] http://permalink.gmane.org/gmane.linux.ports.arm.kernel/195124
> 

Interesting.

This is really something that the core DT and GPIO and IRQ maintainers
should weigh in on. Hence, changing them from Cc: to To: in this message
and/or adding them.