Re: [PATCH-v3 2/3] mfd: 88pm800: Allow configuration of interrupt clear method

[Vaibhav Hiremath <vaibhav.hiremath@linaro.org>]

On Thursday 25 June 2015 11:02 AM, Krzysztof Kozlowski wrote:
> On 25.06.2015 14:26, Vaibhav Hiremath wrote:
>>
>>
>> On Thursday 25 June 2015 05:33 AM, Krzysztof Kozlowski wrote:
>>> 2015-06-24 18:21 GMT+09:00 Vaibhav Hiremath
>>> <vaibhav.hiremath@linaro.org>:
>>>> As per the spec, bit 1 (INT_CLEAR_MODE) of reg addr 0xe
>>>> (page 0) controls the method of clearing interrupt
>>>> status of 88pm800 family of devices;
>>>>
>>>>     0: clear on read
>>>>     1: clear on write
>>>>
>>>> This patch allows to configure this field, through DT.
>>>>
>>>> Also, as suggested by "Lee Jones" renaming DT property and variable
>>>> field to appropriate name.
>>>>
>>>> Signed-off-by: Zhao Ye <zhaoy@marvell.com>
>>>> Signed-off-by: Vaibhav Hiremath <vaibhav.hiremath@linaro.org>
>>>
>>> It does not look like a property of the board. Instead it looks like a
>>> runtime configuration so it should not be part of DT bindings.
>>>
>>
>> Why do you say that?
>>
>> It is very well feature of 88PM860 device, where you can control irq
>> clear operation (either read/write).
>>
>>
>> Thanks,
>> Vaibhav
>>
>>> I understand that previously this was configured by platform data and
>>> now you want to move everything to DT. But this does not belong to
>>> DT...
>>>
>>
>> Thats not completely true.
>> I think DT is the right place for this configuration.
>
> DT and its bindings describe the specific board or device. Let me quote:
> <<The "Open Firmware Device Tree", or simply Device Tree (DT), is a data
> structure and language for describing hardware.  More specifically, it
> is a description of hardware that is readable by an operating system...>>
>
> Whether you clear interrupts by writing or reading is configured during
> runtime and it is completely independent to wiring. Each board with
> 88pm800 would allow both methods. So this is not a property of hardware
> in the terms of open firmware. This is a runtime configuration.
>

Yes,
Fair enough...

I see very little value in runtime configuration, why not just do it
only way (either read or write)?
I would prefer to just set it by default (during init), to clear irq on
write.


Thanks,
Vaibhav