Re: [PATCH v2 06/10] iio: light: veml6030: use read_avail() for available attributes

[Javier Carrasco <javier.carrasco.cruz@gmail.com>]

On 28/09/2024 18:19, Jonathan Cameron wrote:
> On Mon, 23 Sep 2024 00:17:54 +0200
> Javier Carrasco <javier.carrasco.cruz@gmail.com> wrote:
> 
>> Drop custom attributes by using the standard read_avail() callback to
>> read scale and integration time. When at it, define these attributes as
>> available by all channels, as they affect the values of both the ALS and
>> the WHITE channel.
>>
>> Signed-off-by: Javier Carrasco <javier.carrasco.cruz@gmail.com>
> Hi Javier
> 
> Some comments inline
> 
> Thanks,
> 
> Jonathan
> 
>> ---
>>  drivers/iio/light/veml6030.c | 64 +++++++++++++++++++++++++++++++-------------
>>  1 file changed, 45 insertions(+), 19 deletions(-)
>>
>> diff --git a/drivers/iio/light/veml6030.c b/drivers/iio/light/veml6030.c
>> index 89c98bfc5191..a3190fab3add 100644
>> --- a/drivers/iio/light/veml6030.c
>> +++ b/drivers/iio/light/veml6030.c
>> @@ -58,25 +58,24 @@ struct veml6030_data {
>>  	int cur_integration_time;
>>  };
>>  
>> -/* Integration time available in seconds */
>> -static IIO_CONST_ATTR(in_illuminance_integration_time_available,
>> -				"0.025 0.05 0.1 0.2 0.4 0.8");
>> +static const int veml6030_it_times[][2] = {
>> +	{0, 25000},
> Really minor but I'm trying to get IIO standardized on formatting for this
> sort of array and I'd like not to introduce more instances of it
> done without the extra spaces as it will just give more to clean up
> at some point.
> 
> 	{ 0, 25000 },
> etc please.
>> +	{0, 50000},
>> +	{0, 100000},
>> +	{0, 200000},
>> +	{0, 400000},
>> +	{0, 800000},
>> +};
>>  
>>  /*
>>   * Scale is 1/gain. Value 0.125 is ALS gain x (1/8), 0.25 is
>>   * ALS gain x (1/4), 1.0 = ALS gain x 1 and 2.0 is ALS gain x 2.
>>   */
>> -static IIO_CONST_ATTR(in_illuminance_scale_available,
>> -				"0.125 0.25 1.0 2.0");
>> -
>> -static struct attribute *veml6030_attributes[] = {
>> -	&iio_const_attr_in_illuminance_integration_time_available.dev_attr.attr,
>> -	&iio_const_attr_in_illuminance_scale_available.dev_attr.attr,
>> -	NULL
>> -};
>> -
>> -static const struct attribute_group veml6030_attr_group = {
>> -	.attrs = veml6030_attributes,
>> +static const int veml6030_scale_vals[][2] = {
>> +	{0, 125000},
>> +	{0, 250000},
>> +	{1, 0},
>> +	{2, 0},
> 
> As above, add some spaces for minor readability improvement.
> 
>>  };
>>  
>>  /*
>> @@ -197,9 +196,11 @@ static const struct iio_chan_spec veml6030_channels[] = {
>>  		.type = IIO_LIGHT,
>>  		.channel = CH_ALS,
>>  		.info_mask_separate = BIT(IIO_CHAN_INFO_RAW) |
>> -				BIT(IIO_CHAN_INFO_PROCESSED) |
>> -				BIT(IIO_CHAN_INFO_INT_TIME) |
>> -				BIT(IIO_CHAN_INFO_SCALE),
>> +				BIT(IIO_CHAN_INFO_PROCESSED),
>> +		.info_mask_shared_by_all = BIT(IIO_CHAN_INFO_INT_TIME) |
>> +					       BIT(IIO_CHAN_INFO_SCALE),
> This bit is an ABI change and technically old code wasn't a bug, so
> we don't really have a good enough reason to change it.  So Please
> leave these as separate.
> 
>> +		.info_mask_shared_by_all_available = BIT(IIO_CHAN_INFO_INT_TIME) |
>> +						     BIT(IIO_CHAN_INFO_SCALE),
> That doesn't stop us sharing the available as that always was shared
> in the attribute naming above.
> 
>>  		.event_spec = veml6030_event_spec,
>>  		.num_event_specs = ARRAY_SIZE(veml6030_event_spec),
>>  	},
>> @@ -210,6 +211,10 @@ static const struct iio_chan_spec veml6030_channels[] = {
>>  		.channel2 = IIO_MOD_LIGHT_BOTH,
>>  		.info_mask_separate = BIT(IIO_CHAN_INFO_RAW) |
>>  				BIT(IIO_CHAN_INFO_PROCESSED),
>> +		.info_mask_shared_by_all = BIT(IIO_CHAN_INFO_INT_TIME) |
>> +					       BIT(IIO_CHAN_INFO_SCALE),
> This confuses me.  Is it fixing a bug by effectively adding attributes for this
> channel that were previously missing? If so we'll have to go with searpte
> even though they are shared to avoid breaking the ABI for other channel.
> 
>> +		.info_mask_shared_by_all_available = BIT(IIO_CHAN_INFO_INT_TIME) |
>> +						     BIT(IIO_CHAN_INFO_SCALE),
>>  	},
>>  };
>>  
> 
> 

This confused me as well, because even though the attributes where
defined as separate for the ALS channel, modifying their values affected
the values from the WHITE channel.

The integration time and the scale affect both channels, and therefore I
thought they should be shared attributes. But in that case, and to avoid
breaking the ABI for the other channel, I will make them separate even
though writing to one of them will change the value of the other as well.

Thanks and best regards,
Javier Carrasco