On 12/03/2023 10:42, Laurent Pinchart wrote:
> Hi Krzysztof,
>
> Thank you for the patch.
>
> On Fri, Mar 10, 2023 at 10:45:46PM +0100, Krzysztof Kozlowski wrote:
>> The driver can be compile tested with !CONFIG_OF making certain data
>> unused:
>>
>> drivers/regulator/max20086-regulator.c:289:34: error: ‘max20086_dt_ids’ defined but not used [-Werror=unused-const-variable=]
>>
>> Signed-off-by: Krzysztof Kozlowski <krzysztof.kozlowski@xxxxxxxxxx>
>> ---
>> drivers/regulator/max20086-regulator.c | 2 +-
>> 1 file changed, 1 insertion(+), 1 deletion(-)
>>
>> diff --git a/drivers/regulator/max20086-regulator.c b/drivers/regulator/max20086-regulator.c
>> index b8bf76c170fe..c98a72f43935 100644
>> --- a/drivers/regulator/max20086-regulator.c
>> +++ b/drivers/regulator/max20086-regulator.c
>> @@ -286,7 +286,7 @@ static const struct i2c_device_id max20086_i2c_id[] = {
>>
>> MODULE_DEVICE_TABLE(i2c, max20086_i2c_id);
>>
>> -static const struct of_device_id max20086_dt_ids[] = {
>> +static const struct of_device_id max20086_dt_ids[] __maybe_unused = {
>
> The following change would also work, as the of_match_table field of
> struct device_driver isn't conditioned by CONFIG_OF:
>
> diff --git a/drivers/regulator/max20086-regulator.c b/drivers/regulator/max20086-regulator.c
> index b8bf76c170fe..ad92f84b4abb 100644
> --- a/drivers/regulator/max20086-regulator.c
> +++ b/drivers/regulator/max20086-regulator.c
> @@ -320,7 +320,7 @@ MODULE_DEVICE_TABLE(of, max20086_dt_ids);
> static struct i2c_driver max20086_regulator_driver = {
> .driver = {
> .name = "max20086",
> - .of_match_table = of_match_ptr(max20086_dt_ids),
> + .of_match_table = max20086_dt_ids,
> },
> .probe_new = max20086_i2c_probe,
> .id_table = max20086_i2c_id,
>
> Your patch should reduce the module size without any real drawback as
> far as I can see, so that's probably best. I'm fine with either
> approach, so
I know it would work. If you check all my patches you see both patterns
used depending on the needs:
https://lore.kernel.org/all/?q=f%3Akrzysztof+of_device_id
(~100 patches so far)
The point is that here the device can actually match via ID table, so OF
table could stay optional. I don't think PRP0001 is relevant here, thus
I proposed to keep OF table optional. Different folks have different
opinion on that, so if general consensus is that availability of OF ID
table (for PRP0001) is preferred, I can rework the patch towards it.
>
> Reviewed-by: Laurent Pinchart <laurent.pinchart@xxxxxxxxxxxxxxxx>
Thanks
Best regards,
Krzysztof
