Re: [PATCH 2/3] iio: light: opt3001: add support for TI's opt3002 light sensor

[Emil Gedenryd <Emil.Gedenryd@axis.com>]

On Sat, 2024-09-07 at 18:35 +0100, Jonathan Cameron wrote:
> On Thu, 5 Sep 2024 12:20:46 +0200
> Emil Gedenryd <emil.gedenryd@axis.com> wrote:
> 
> > TI's opt3002 light sensor shares most properties with the opt3001
> > model, with the exception of supporting a wider spectrum range.
> > 
> > Add support for TI's opt3002 by extending the TI opt3001 driver.
> > 
> > See https://www.ti.com/product/OPT3002 for more information.
> Make that a Datasheet tag.
> > 
> Datasheet: https://www.ti.com/product/OPT3002
> > Signed-off-by: Emil Gedenryd <emil.gedenryd@axis.com>
> 
> Various comments inline.
> Thanks,
> 
> Jonathan

Thank you for having a look at the patch set.
I'll submit a new version that fixes the issues you highlighted
in the code and commit messages either later today or tomorrow.

Best regards,
Emil
> 
> > ---
> >  drivers/iio/light/Kconfig   |   2 +-
> >  drivers/iio/light/opt3001.c | 199 ++++++++++++++++++++++++++++++++++++--------
> >  2 files changed, 167 insertions(+), 34 deletions(-)
> > 
> > diff --git a/drivers/iio/light/Kconfig b/drivers/iio/light/Kconfig
> > index b68dcc1fbaca..c35bf962dae6 100644
> > --- a/drivers/iio/light/Kconfig
> > +++ b/drivers/iio/light/Kconfig
> > @@ -461,7 +461,7 @@ config OPT3001
> >  	depends on I2C
> >  	help
> >  	  If you say Y or M here, you get support for Texas Instruments
> > -	  OPT3001 Ambient Light Sensor.
> > +	  OPT3001 Ambient Light Sensor, OPT3002 Light-to-Digital Sensor.
> >  
> >  	  If built as a dynamically linked module, it will be called
> >  	  opt3001.
> > diff --git a/drivers/iio/light/opt3001.c b/drivers/iio/light/opt3001.c
> > index 176e54bb48c3..e6098f88dd04 100644
> > --- a/drivers/iio/light/opt3001.c
> > +++ b/drivers/iio/light/opt3001.c
> > @@ -70,6 +70,19 @@
> >  #define OPT3001_RESULT_READY_SHORT	150
> >  #define OPT3001_RESULT_READY_LONG	1000
> >  
> > +/* The opt3002 doesn't have a device id register, predefine value instead */
> > +#define OPT3002_DEVICE_ID_VALUE		3002
> 
> Why?  Just make the code not care about the value for this
> device.  Add a flag to the chip info structure to say it doesn't have
> one and check that before using it.

The device id is used to log the model. Should I not log the
model for the opt3002 then or should I have the callback just return
3002? I thought it would be cleaner to have the id value as a defined
constant instead of a "magic" number in the code. Is there a preferred
way of doing it?
> 
> 
> > +
> > +enum chip_model {
> > +	OPT3001,
> This should not be needed. See below.
> 
> > +	OPT3002,
> > +};
> > +
> > +struct opt300x_chip_info {
> > +	enum chip_model model;
> > +	enum iio_chan_type chan_type;
> > +};
> > +
> >  struct opt3001 {
> >  	struct i2c_client	*client;
> >  	struct device		*dev;
> > @@ -79,6 +92,7 @@ struct opt3001 {
> >  	bool			result_ready;
> >  	wait_queue_head_t	result_ready_queue;
> >  	u16			result;
> > +	const struct opt300x_chip_info *chip_info;
> >  
> >  	u32			int_time;
> >  	u32			mode;
> > @@ -97,6 +111,16 @@ struct opt3001_scale {
> >  	int	val2;
> >  };
> >  
> > +static const struct opt300x_chip_info opt3001_chip_info = {
> > +	.model = OPT3001,
> Having a model in a chip_info structure is almost always a sign
> of a design that won't scale well to lots of additional devices.
> 
> Get rid of that and instead add all the 'data' that you are looking
> up with that model number to this structure so it can be just
> referenced without caring which mode it is for.

Good point!
I'll move as much model-specific code as possible to the struct.

> 
> > +	.chan_type = IIO_LIGHT,
> > +};
> > +
> > +static const struct opt300x_chip_info opt3002_chip_info = {
> > +	.model = OPT3002,
> > +	.chan_type = IIO_INTENSITY,
> > +};
> 
> > +
> >  static int opt3001_find_scale(const struct opt3001 *opt, int val,
> >  		int val2, u8 *exponent)
> >  {
> >  	int i;
> > +	const struct opt3001_scale (*scale_arr)[12];
> >  
> > -	for (i = 0; i < ARRAY_SIZE(opt3001_scales); i++) {
> > -		const struct opt3001_scale *scale = &opt3001_scales[i];
> > +	switch (opt->chip_info->model) {
> > +	case OPT3001:
> > +		scale_arr = &opt3001_scales;
> Put them in chip_info directly, not look them up here.
> 
> > +		break;
> > +	case OPT3002:
> > +		scale_arr = &opt3002_scales;
> > +		break;
> > +	default:
> > +		dev_err(opt->dev, "scale not configured for chip model\n");
> > +		return -EINVAL;
> > +	}
> >  
> > +	for (i = 0; i < ARRAY_SIZE(*scale_arr); i++) {
> > +		const struct opt3001_scale *scale = &(*scale_arr)[i];
> >  		/*
> > -		 * Combine the integer and micro parts for comparison
> > -		 * purposes. Use milli lux precision to avoid 32-bit integer
> > -		 * overflows.
> > +		 * Compare the integer and micro parts to determine value scale.
> >  		 */
> > -		if ((val * 1000 + val2 / 1000) <=
> > -				(scale->val * 1000 + scale->val2 / 1000)) {
> > +		if (val < scale->val ||
> > +		    (val == scale->val && val2 <= scale->val2)) {
> >  			*exponent = i;
> >  			return 0;
> >  		}
> > @@ -174,11 +259,20 @@ static int opt3001_find_scale(const struct opt3001 *opt, int val,
> >  static void opt3001_to_iio_ret(struct opt3001 *opt, u8 exponent,
> >  		u16 mantissa, int *val, int *val2)
> >  {
> > -	int lux;
> > +	int ret;
> >  
> > -	lux = 10 * (mantissa << exponent);
> > -	*val = lux / 1000;
> > -	*val2 = (lux - (*val * 1000)) * 1000;
> > +	switch (opt->chip_info->model) {
> > +	case OPT3001:
> > +		ret = 10 * (mantissa << exponent);
> > +		*val = ret / 1000;
> > +		*val2 = (ret - (*val * 1000)) * 1000;
> > +		break;
> > +	case OPT3002:
> > +		ret = 12 * (mantissa << exponent);
> > +		*val = ret / 10;
> > +		*val2 = (ret - (*val * 10)) * 100000;
> 
> As below - constants in the chip_info structure so this becomes
> a simple case of using them without needing to know the chip type
> in the code.
> 
> > +		break;
> > +	}
> >  }
> 
> > @@ -497,7 +602,15 @@ static int opt3001_write_event_value(struct iio_dev *iio,
> >  		goto err;
> >  	}
> >  
> > -	mantissa = (((val * 1000) + (val2 / 1000)) / 10) >> exponent;
> > +	switch (opt->chip_info->model) {
> > +	case OPT3001:
> > +		mantissa = (((val * 1000) + (val2 / 1000)) / 10) >> exponent;
> 
> Encode the sections of this maths that is different as values in the chip
> info structure and use them directly here rather than having a switch statement.
> 
> > +		break;
> > +	case OPT3002:
> > +		mantissa = (((val * 10) + (val2 / 100000)) / 12) >> exponent;
> > +		break;
> > +	}
> > +
> >  	value = (exponent << 12) | mantissa;
> >  
> >  	switch (dir) {
> > @@ -607,15 +720,22 @@ static int opt3001_read_id(struct opt3001 *opt)
> >  	manufacturer[0] = ret >> 8;
> >  	manufacturer[1] = ret & 0xff;
> >  
> > -	ret = i2c_smbus_read_word_swapped(opt->client, OPT3001_DEVICE_ID);
> > -	if (ret < 0) {
> > -		dev_err(opt->dev, "failed to read register %02x\n",
> > +	switch (opt->chip_info->model) {
> 
> Add a callback for this to the chip_info structure. That will make it
> much cleaner to add future devices.
> 
> > +	case OPT3001:
> > +		ret = i2c_smbus_read_word_swapped(opt->client,
> > +						  OPT3001_DEVICE_ID);
> > +		if (ret == 0) {
> > +			dev_err(opt->dev, "failed to read register %02x\n",
> >  				OPT3001_DEVICE_ID);
> > -		return ret;
> > +			return ret;
> > +		}
> > +		device_id = ret;
> > +		break;
> > +	case OPT3002:
> > +		device_id = OPT3002_DEVICE_ID_VALUE;
> > +		break;
> 
> > @@ -755,6 +877,7 @@ static int opt3001_probe(struct i2c_client *client)
> >  	opt = iio_priv(iio);
> >  	opt->client = client;
> >  	opt->dev = dev;
> > +	opt->chip_info = device_get_match_data(&client->dev);
> >  
> >  	mutex_init(&opt->lock);
> >  	init_waitqueue_head(&opt->result_ready_queue);
> > @@ -769,10 +892,18 @@ static int opt3001_probe(struct i2c_client *client)
> >  		return ret;
> >  
> >  	iio->name = client->name;
> > -	iio->channels = opt3001_channels;
> > -	iio->num_channels = ARRAY_SIZE(opt3001_channels);
> >  	iio->modes = INDIO_DIRECT_MODE;
> >  	iio->info = &opt3001_info;
> > +	switch (opt->chip_info->model) {
> > +	case OPT3001:
> > +		iio->channels = opt3001_channels;
> > +		iio->num_channels = ARRAY_SIZE(opt3001_channels);
> Add this to the chip info structure so this can become a simple assignment
> rather than having to look up by model.
> 
> > +		break;
> > +	case OPT3002:
> > +		iio->channels = opt3002_channels;
> > +		iio->num_channels = ARRAY_SIZE(opt3002_channels);
> > +		break;
> > +	}
> >  
> >  	ret = devm_iio_device_register(dev, iio);
> >  	if (ret) {
> > @@ -826,13 +957,15 @@ static void opt3001_remove(struct i2c_client *client)
> >  }
> >  
> >  static const struct i2c_device_id opt3001_id[] = {
> > -	{ "opt3001" },
> > +	{ "opt3001", 0 },
> > +	{ "opt3002", 1 },
> >  	{ } /* Terminating Entry */
> >  };
> >  MODULE_DEVICE_TABLE(i2c, opt3001_id);
> >  
> >  static const struct of_device_id opt3001_of_match[] = {
> > -	{ .compatible = "ti,opt3001" },
> > +	{ .compatible = "ti,opt3001", .data = &opt3001_chip_info },
> > +	{ .compatible = "ti,opt3002", .data = &opt3002_chip_info },
> >  	{ }
> >  };
> >  MODULE_DEVICE_TABLE(of, opt3001_of_match);
> > 
>