Re: [PATCH RESEND RFC 2/4] drivers: pinctrl: qcom: add wakeup gpio map for sdm845

[Marc Zyngier <marc.zyngier@arm.com>]

On 13/08/18 20:41, Lina Iyer wrote:
> On Wed, Aug 01 2018 at 14:04 -0600, Lina Iyer wrote:
>> On Wed, Aug 01 2018 at 02:42 -0600, Marc Zyngier wrote:
>>> On Wed, 01 Aug 2018 03:00:19 +0100,
>>> Lina Iyer <ilina@codeaurora.org> wrote:
>>>>
>>>> Add GPIO to PDC pin map for the SDM845 SoC.
>>>>
>>>> Signed-off-by: Lina Iyer <ilina@codeaurora.org>
>>>> ---
>>>> drivers/pinctrl/qcom/pinctrl-sdm845.c | 76 +++++++++++++++++++++++++++
>>>> 1 file changed, 76 insertions(+)
>>>>
>>>> diff --git a/drivers/pinctrl/qcom/pinctrl-sdm845.c b/drivers/pinctrl/qcom/pinctrl-sdm845.c
>>>> index 2ab7a8885757..e93660922dc2 100644
>>>> --- a/drivers/pinctrl/qcom/pinctrl-sdm845.c
>>>> +++ b/drivers/pinctrl/qcom/pinctrl-sdm845.c
>>>> @@ -1277,6 +1277,80 @@ static const struct msm_pingroup sdm845_groups[] = {
>>>> 	UFS_RESET(ufs_reset, 0x99f000),
>>>> };
>>>>
>>>> +static struct msm_pinctrl_pdc_map sdm845_wakeup_gpios[] = {
>>>
>>> [huge array]
>>>
>>>> +};
>>>
>>> Why isn't that array part of the DT? I'd expect other SoCs to
>>> eventually use a similar mechanism, no?
>>>
>> I agree and it should be.
>>
>> One place I am thinking is to add it to the DT definition of PDC
>> controller as a data argument -
>>
>> 	tlmm: pinctrl@000000{
>> 	[...]
>> 		interrupts-extended = <&pdc 30 IRQ_TYPE_LEVEL_HIGH 1>,
>>                     <&pdc 31 IRQ_TYPE_LEVEL_HIGH 3>,
>> 		      <&pdc 32 IRQ_TYPE_LEVEL_HIGH 5>,
>> 						   ^
>> 						   |--- Provide the GPIO
>> 						   for the PDC pin here.
>> 	};
>>
>> 	pdc: interrupt-controller@b220000 {
>> 		compatible = "qcom,sdm845-pdc";
>> 		reg = <0xb220000 0x30000>;
>> 		qcom,pdc-ranges = <0 512 94>, <94 641 15>, <115 662 7>;
>> 		#interrupt-cells = <3>; <-------- Increase this from 2 ?
>> 		interrupt-parent = <&intc>;
>> 		interrupt-controller;
>> 	};
>>
>> Would that be acceptable?
>>
> Any ideas on how to do this better?

I don't think adding an extra argument to the PDC interrupt specifier is
that great. I'd rather see some associated array in the PDC binding
mapping an interrupt to a pin on which special treatment must be applied.

Thanks,

	M.
-- 
Jazz is not dead. It just smells funny...