On 11/03/2023 19:35, Jonathan Cameron wrote:
> On Sat, 11 Mar 2023 13:28:17 +0100
> Krzysztof Kozlowski <krzysztof.kozlowski@xxxxxxxxxx> wrote:
>
>> On 11/03/2023 13:26, Jonathan Cameron wrote:
>>> On Sat, 11 Mar 2023 12:14:56 +0100
>>> Krzysztof Kozlowski <krzysztof.kozlowski@xxxxxxxxxx> wrote:
>>>
>>>> The driver currently matches only via i2c_device_id, but also has
>>>> of_device_id table:
>>>>
>>>> drivers/iio/light/max44009.c:545:34: error: ‘max44009_of_match’ defined but not used [-Werror=unused-const-variable=]
>>>>
>>>> Fixes: 6aef699a7d7e ("iio: light: add driver for MAX44009")
>>>> Signed-off-by: Krzysztof Kozlowski <krzysztof.kozlowski@xxxxxxxxxx>
>>>
>>> Don't use of_match_ptr() unless you are absolutely sure no other firmware
>>> route will make use of the of_match_table.
>>>
>>> In this particular case ACPI using PRP0001 is broken by that macro.
>>
>> It's not broken because there was no matching via PRP0001 due to missing
>> table.
>>
>>>
>>> So good to set the of_match_table, but make sure to always set it
>>> and hence you don't need the __maybe_unused.
>>
>> So you want to add PRP0001? We can, the fix is for different issue, though.
>
> There is nothing to add. You need to do less than you have done in this patch.
> Drop the of_match_ptr() and the __maybe_unused and PRP0001 based matching will just
> work. The PRP0001 path just uses the of_device_id table and needs no
Sure, but that's *adding a feature*. You said that "ACPI using PRP0001
is broken", but it was never here in the first place. PRP0001 *was*
already broken here, not *is*. The patch does not decrease the
functionality.
> specific support in a driver - it doesn't need an ACPI id table or anything like
> that.
>
> It's a long story, but hindsight says that of_match_ptr() should never have
> existed as it only serves to stop things working that otherwise work for free.
Sure, I can go with ID table always present.
Best regards,
Krzysztof
