Skip to contentSkip to conversation

Add light channel for LTR390

PATCH In review Abhash Jha 2 participants 8 messages Aug 14, 2024 – Aug 18, 2024

Subsystems (2)
lore
Share & tools
lore

Abhash Jha

author
#0 lore raw

v7 cover Add light channel for LTR390

Hello,

The first patch adds a new channel for the ALS feature of the sensor. The same configuration of gain and resolution has to be provided for this channel as well. As there are two IIO channels now, we would need to switch the sensor's mode of operation depending on which sensor is being accessed. Hence, mode switching is also provided.

Then the second patch adds support for calculating counts_per_uvi based on the current gain and resolution value.

Changes in v7:
- Changed the ltr390_set_mode function to do better error handling.
- Link to v6: https://lore.kernel.org/linux-iio/[email protected]/T/#t

Changes in v6:
- Changed IIO_CHAN_INFO_PROCESSED to IIO_CHAN_INFO_RAW
- Changed the scaling code
- Link to v5: https://lore.kernel.org/linux-iio/CAG=0Rq+q0WJzMroYwQy-4Ng0aSkTvaw-FEMx68i3MqAZwfteCg@mail.gmail.com/T/#t

Changes in v5:
- Replaced the IIO_INTENSITY channel with IIO_LIGHT channel
- We calculate the lux value directly using als_data / (gain * int_time)
- Provided a scale channel where the scale is 0.6 * WINDOW_FACTOR
- Link to v4: https://lore.kernel.org/linux-iio/[email protected]/T/#m

Changes in v4:
- Added "bitfield.h" include to fix -Wimplicit-function-declaration.
- Link to v3: https://lore.kernel.org/linux-iio/[email protected]/

Changes in v3:
- Added cover letter to the patch series.
- Fixed indentation in the patch description.
- Patch specific changes are listed below.

[PATCH v3 1/3]
- Cleaned up the spurious changes made in v2.
- ltr390_set_int_time and ltr390_set_gain now return -EINVAL to indicate no match.

[PATCH v3 2/3]
- Used enum ltr390_mode inside the ltr390_data struct.
- Refactored ltr390_set_mode function according to the comments in v2.

[PATCH v3 3/3]
- Simplified the formula for counts_per_uvi calculation.
- Removed spurious whitespace changes introduced in v2.

- Link to v2: https://lore.kernel.org/linux-iio/[email protected]/

Changes in v2:
- Split the single patch into 3 patches.
- Used FIELD_PREP to perform bit shifting.
- Used enum for mode selection instead of defines.
- Fixed indentation and whitespace issues pointed out in the comments
- Replaced mutex_lock(&data->lock) with guard(mutex)(&data->lock)
- Provided available values for gain and resolution via read_avail instead of sysfs attributes.
- Refactored ltr390_set_gain and ltr390_set_int_time.
- Used early returns instead of single exit points.

- Link to v1: https://lore.kernel.org/linux-iio/[email protected]/

Regards,
Abhash

Abhash Jha (2):
iio: light: ltr390: Add ALS channel and support for gain and
resolution
iio: light: ltr390: Calculate 'counts_per_uvi' dynamically

 drivers/iio/light/ltr390.c | 115 ++++++++++++++++++++++++++++++++-----
 1 file changed, 100 insertions(+), 15 deletions(-)
Signature · 1 line
2.43.0
author
#1 lore raw

PATCH v7 1/2 iio: light: ltr390: Add ALS channel and support for gain and resolution

Add new ALS channel and allow reading lux and scale values. Also provide gain and resolution configuration for ALS channel. Add automatic mode switching between the UVS and ALS channel based on which channel is being accessed.
The default mode in which the sensor start is ALS mode.

---
 drivers/iio/light/ltr390.c | 105 +++++++++++++++++++++++++++++++------
 1 file changed, 90 insertions(+), 15 deletions(-)
drivers/iio/light/ltr390.c +90 -15
index ee3d30075..16e06f2ab 100644
@@ -63,11 +63,17 @@
  */
 #define LTR390_WINDOW_FACTOR 1
 
+enum ltr390_mode {
+	LTR390_SET_ALS_MODE,
+	LTR390_SET_UVS_MODE,
+};
+
 struct ltr390_data {
 	struct regmap *regmap;
 	struct i2c_client *client;
 	/* Protects device from simulataneous reads */
 	struct mutex lock;
+	enum ltr390_mode mode;
 	int gain;
 	int int_time_us;
 };
@@ -95,6 +101,30 @@ static int ltr390_register_read(struct ltr390_data *data, u8 register_address)
 	return get_unaligned_le24(recieve_buffer);
 }
 
+static int ltr390_set_mode(struct ltr390_data *data, enum ltr390_mode mode)
+{
+	int ret;
+
+	if (data->mode == mode)
+		return 0;
+
+	switch (mode) {
+	case LTR390_SET_ALS_MODE:
+		ret = regmap_clear_bits(data->regmap, LTR390_MAIN_CTRL, LTR390_UVS_MODE);
+		break;
+
+	case LTR390_SET_UVS_MODE:
+		ret = regmap_set_bits(data->regmap, LTR390_MAIN_CTRL, LTR390_UVS_MODE);
+		break;
+	}
+
+	if (ret)
+		return ret;
+
+	data->mode = mode;
+	return 0;
+}
+
 static int ltr390_read_raw(struct iio_dev *iio_device,
 			   struct iio_chan_spec const *chan, int *val,
 			   int *val2, long mask)
@@ -105,15 +135,47 @@ static int ltr390_read_raw(struct iio_dev *iio_device,
 	guard(mutex)(&data->lock);
 	switch (mask) {
 	case IIO_CHAN_INFO_RAW:
-		ret = ltr390_register_read(data, LTR390_UVS_DATA);
-		if (ret < 0)
-			return ret;
+		switch (chan->type) {
+		case IIO_UVINDEX:
+			ret = ltr390_set_mode(data, LTR390_SET_UVS_MODE);
+			if (ret < 0)
+				return ret;
+
+			ret = ltr390_register_read(data, LTR390_UVS_DATA);
+			if (ret < 0)
+				return ret;
+			break;
+
+		case IIO_LIGHT:
+			ret = ltr390_set_mode(data, LTR390_SET_ALS_MODE);
+			if (ret < 0)
+				return ret;
+
+			ret = ltr390_register_read(data, LTR390_ALS_DATA);
+			if (ret < 0)
+				return ret;
+			break;
+
+		default:
+			return -EINVAL;
+		}
 		*val = ret;
 		return IIO_VAL_INT;
 	case IIO_CHAN_INFO_SCALE:
-		*val = LTR390_WINDOW_FACTOR;
-		*val2 = LTR390_COUNTS_PER_UVI;
-		return IIO_VAL_FRACTIONAL;
+		switch (chan->type) {
+		case IIO_UVINDEX:
+			*val = LTR390_WINDOW_FACTOR;
+			*val2 = LTR390_COUNTS_PER_UVI;
+			return IIO_VAL_FRACTIONAL;
+
+		case IIO_LIGHT:
+			*val = LTR390_WINDOW_FACTOR * 6 * 100;
+			*val2 = data->gain * data->int_time_us;
+			return IIO_VAL_FRACTIONAL;
+
+		default:
+			return -EINVAL;
+		}
 
 	case IIO_CHAN_INFO_INT_TIME:
 		*val = data->int_time_us;
@@ -128,11 +190,23 @@ static int ltr390_read_raw(struct iio_dev *iio_device,
 static const int ltr390_int_time_map_us[] = { 400000, 200000, 100000, 50000, 25000, 12500 };
 static const int ltr390_gain_map[] = { 1, 3, 6, 9, 18 };
 
-static const struct iio_chan_spec ltr390_channel = {
-	.type = IIO_UVINDEX,
-	.info_mask_separate = BIT(IIO_CHAN_INFO_RAW) | BIT(IIO_CHAN_INFO_SCALE),
-	.info_mask_shared_by_all = BIT(IIO_CHAN_INFO_INT_TIME),
-	.info_mask_shared_by_all_available = BIT(IIO_CHAN_INFO_INT_TIME) | BIT(IIO_CHAN_INFO_SCALE)
+static const struct iio_chan_spec ltr390_channels[] = {
+	/* UV sensor */
+	{
+		.type = IIO_UVINDEX,
+		.scan_index = 0,
+		.info_mask_separate = BIT(IIO_CHAN_INFO_RAW) | BIT(IIO_CHAN_INFO_SCALE),
+		.info_mask_shared_by_all = BIT(IIO_CHAN_INFO_INT_TIME),
+		.info_mask_shared_by_all_available = BIT(IIO_CHAN_INFO_INT_TIME) | BIT(IIO_CHAN_INFO_SCALE)
+	},
+	/* ALS sensor */
+	{
+		.type = IIO_LIGHT,
+		.scan_index = 1,
+		.info_mask_separate = BIT(IIO_CHAN_INFO_RAW) | BIT(IIO_CHAN_INFO_SCALE),
+		.info_mask_shared_by_all = BIT(IIO_CHAN_INFO_INT_TIME),
+		.info_mask_shared_by_all_available = BIT(IIO_CHAN_INFO_INT_TIME) | BIT(IIO_CHAN_INFO_SCALE)
+	},
 };
 
 static int ltr390_set_gain(struct ltr390_data *data, int val)
@@ -252,12 +326,14 @@ static int ltr390_probe(struct i2c_client *client)
 	data->int_time_us = 100000;
 	/* default value of gain from pg: 16 of the datasheet */
 	data->gain = 3;
+	/* default mode for ltr390 is ALS mode */
+	data->mode = LTR390_SET_ALS_MODE;
 
 	mutex_init(&data->lock);
 
 	indio_dev->info = &ltr390_info;
-	indio_dev->channels = &ltr390_channel;
-	indio_dev->num_channels = 1;
+	indio_dev->channels = ltr390_channels;
+	indio_dev->num_channels = ARRAY_SIZE(ltr390_channels);
 	indio_dev->name = "ltr390";
 
 	ret = regmap_read(data->regmap, LTR390_PART_ID, &part_number);
@@ -275,8 +351,7 @@ static int ltr390_probe(struct i2c_client *client)
 	/* Wait for the registers to reset before proceeding */
 	usleep_range(1000, 2000);
 
-	ret = regmap_set_bits(data->regmap, LTR390_MAIN_CTRL,
-			      LTR390_SENSOR_ENABLE | LTR390_UVS_MODE);
+	ret = regmap_set_bits(data->regmap, LTR390_MAIN_CTRL, LTR390_SENSOR_ENABLE);
 	if (ret)
 		return dev_err_probe(dev, ret, "failed to enable the sensor\n");
 
Signature · 1 line
2.43.0
author
#2 lore raw

PATCH v7 2/2 iio: light: ltr390: Calculate 'counts_per_uvi' dynamically

In reply to Abhash Jha

counts_per_uvi depends on the current value of gain and resolution. Hence, we cannot use the hardcoded value 96.
The counts_per_uvi function gives the count based on the current gain and resolution (integration time).

---
 drivers/iio/light/ltr390.c | 14 ++++++++++++--
 1 file changed, 12 insertions(+), 2 deletions(-)
drivers/iio/light/ltr390.c +12 -2
index 16e06f2ab..7e58b50f3 100644
@@ -46,6 +46,8 @@
 #define LTR390_UVS_MODE	      BIT(3)
 #define LTR390_SENSOR_ENABLE  BIT(1)
 
+#define LTR390_FRACTIONAL_PRECISION 100
+
 /*
  * At 20-bit resolution (integration time: 400ms) and 18x gain, 2300 counts of
  * the sensor are equal to 1 UV Index [Datasheet Page#8].
@@ -125,6 +127,14 @@ static int ltr390_set_mode(struct ltr390_data *data, enum ltr390_mode mode)
 	return 0;
 }
 
+static int ltr390_counts_per_uvi(struct ltr390_data *data)
+{
+	const int orig_gain = 18;
+	const int orig_int_time = 400;
+
+	return DIV_ROUND_CLOSEST(23 * data->gain * data->int_time_us, 10 * orig_gain * orig_int_time);
+}
+
 static int ltr390_read_raw(struct iio_dev *iio_device,
 			   struct iio_chan_spec const *chan, int *val,
 			   int *val2, long mask)
@@ -164,8 +174,8 @@ static int ltr390_read_raw(struct iio_dev *iio_device,
 	case IIO_CHAN_INFO_SCALE:
 		switch (chan->type) {
 		case IIO_UVINDEX:
-			*val = LTR390_WINDOW_FACTOR;
-			*val2 = LTR390_COUNTS_PER_UVI;
+			*val = LTR390_WINDOW_FACTOR * LTR390_FRACTIONAL_PRECISION;
+			*val2 = ltr390_counts_per_uvi(data);
 			return IIO_VAL_FRACTIONAL;
 
 		case IIO_LIGHT:
Signature · 1 line
2.43.0
maintainer
#3 lore raw
In reply to Abhash Jha
Quoted 10 lines · On Wed, 14 Aug 2024 17:01:32 +0530 Abhash Jha <[email protected]>…

Hello,

The first patch adds a new channel for the ALS feature of the sensor. The same configuration of gain and resolution has to be provided for this channel as well. As there are two IIO channels now, we would need to switch the sensor's mode of operation depending on which sensor is being accessed. Hence, mode switching is also provided.

Then the second patch adds support for calculating counts_per_uvi based on the current gain and resolution value.

This is v7 mark 2. I'm confused, but I think I picked up this version (Seems I'd queued an earlier one and not mentioned it on the list though so I've dropped that in favour of this).

Quoted 64 lines

Changes in v7:
- Changed the ltr390_set_mode function to do better error handling.
- Link to v6: https://lore.kernel.org/linux-iio/[email protected]/T/#t

Changes in v6:
- Changed IIO_CHAN_INFO_PROCESSED to IIO_CHAN_INFO_RAW
- Changed the scaling code
- Link to v5: https://lore.kernel.org/linux-iio/CAG=0Rq+q0WJzMroYwQy-4Ng0aSkTvaw-FEMx68i3MqAZwfteCg@mail.gmail.com/T/#t

Changes in v5:
- Replaced the IIO_INTENSITY channel with IIO_LIGHT channel
- We calculate the lux value directly using als_data / (gain * int_time)
- Provided a scale channel where the scale is 0.6 * WINDOW_FACTOR
- Link to v4: https://lore.kernel.org/linux-iio/[email protected]/T/#m

Changes in v4:
- Added "bitfield.h" include to fix -Wimplicit-function-declaration.
- Link to v3: https://lore.kernel.org/linux-iio/[email protected]/

Changes in v3:
- Added cover letter to the patch series.
- Fixed indentation in the patch description.
- Patch specific changes are listed below.

[PATCH v3 1/3]
- Cleaned up the spurious changes made in v2.
- ltr390_set_int_time and ltr390_set_gain now return -EINVAL to indicate no match.

[PATCH v3 2/3]
- Used enum ltr390_mode inside the ltr390_data struct.
- Refactored ltr390_set_mode function according to the comments in v2.

[PATCH v3 3/3]
- Simplified the formula for counts_per_uvi calculation.
- Removed spurious whitespace changes introduced in v2.

- Link to v2: https://lore.kernel.org/linux-iio/[email protected]/

Changes in v2:
- Split the single patch into 3 patches.
- Used FIELD_PREP to perform bit shifting.
- Used enum for mode selection instead of defines.
- Fixed indentation and whitespace issues pointed out in the comments
- Replaced mutex_lock(&data->lock) with guard(mutex)(&data->lock)
- Provided available values for gain and resolution via read_avail instead of sysfs attributes.
- Refactored ltr390_set_gain and ltr390_set_int_time.
- Used early returns instead of single exit points.

- Link to v1: https://lore.kernel.org/linux-iio/[email protected]/

Regards,
Abhash

Abhash Jha (2):
iio: light: ltr390: Add ALS channel and support for gain and
resolution
iio: light: ltr390: Calculate 'counts_per_uvi' dynamically

 drivers/iio/light/ltr390.c | 115 ++++++++++++++++++++++++++++++++-----
 1 file changed, 100 insertions(+), 15 deletions(-)
author
#4 lore raw
Quoted 4 lines

This is v7 mark 2. I'm confused, but I think I picked up this version (Seems I'd queued an earlier one and not mentioned it on the list though so I've dropped that in favour of this).

Does that mean that you have picked up the patch 1/2 and 2/2 of the v7 series as well ?

Thanks,
Abhash

maintainer
#5 lore raw
Quoted 6 lines · On Sat, 17 Aug 2024 21:20:10 +0530 Abhash jha <[email protected]>…
Quoted 4 lines

This is v7 mark 2. I'm confused, but I think I picked up this version (Seems I'd queued an earlier one and not mentioned it on the list though so I've dropped that in favour of this).

Does that mean that you have picked up the patch 1/2 and 2/2 of the v7 series as well ?

I think I have. But with two versions of v7 I'm not 100% sure which one got picked up. I've pushed out now as testing, so take a look.

https://git.kernel.org/pub/scm/linux/kernel/git/jic23/iio.git/log/?h=testing

Quoted 3 lines

Thanks,
Abhash

maintainer
#7 lore raw
Quoted 9 lines · On Sat, 17 Aug 2024 23:07:37 +0530 Abhash jha <[email protected]>…
Quoted 5 lines

I think I have. But with two versions of v7 I'm not 100% sure which one got picked up. I've pushed out now as testing, so take a look.

https://git.kernel.org/pub/scm/linux/kernel/git/jic23/iio.git/log/?h=testing

Quoted 1 line

The two versions v7 patches are the same. I had sent the same thing again because
I thought it might have gotten lost in your mail.
My apologies for getting you confused.

Ah. Never bother doing that. Just send a 'ping' to the original thread.

Most maintainers now use a lot of automation so tend not to drop messages any more (it used to happen occasionally).

Also convention is to wait at least 2 weeks before pinging.

Jonathan

Quoted 3 lines

Thanks,
Abhash