Skip to contentSkip to conversation

imu: st_lsm6dsx: Add support for rotation sensor

PATCH Superseded Francesco Lavra 3 participants 19 messages Jan 9 – Jan 16

Subsystems (2)

v1 of 9 Superseded Latest is v9 →

lore
Share & tools
lore
#0 lore raw

cover imu: st_lsm6dsx: Add support for rotation sensor

This series adds support for the rotation sensor functionality present in some chips from the ST LSM6DSX IMU family.
The first commit is a fix to a previous commit of mine [1] which made it to 6.19-rcX; the fix technically changes userspace, but this should be OK as long as it goes in during this release cycle.
Tested on LSM6DSV16X.

[1] https://lore.kernel.org/linux-iio/[email protected]/

Francesco Lavra (3):
iio: imu: st_lsm6dsx: set buffer sampling frequency for accelerometer
only
iio: imu: st_lsm6dsx: set FIFO ODR for accelerometer and magnetometer
only
iio: imu: st_lsm6dsx: add support for rotation sensor

 drivers/iio/imu/st_lsm6dsx/Makefile           |   2 +-
 drivers/iio/imu/st_lsm6dsx/st_lsm6dsx.h       |  26 +-
 .../iio/imu/st_lsm6dsx/st_lsm6dsx_buffer.c    |  28 ++-
 drivers/iio/imu/st_lsm6dsx/st_lsm6dsx_core.c  |  58 +++++
 .../iio/imu/st_lsm6dsx/st_lsm6dsx_fusion.c    | 224 ++++++++++++++++++
 5 files changed, 331 insertions(+), 7 deletions(-)
 create mode 100644 drivers/iio/imu/st_lsm6dsx/st_lsm6dsx_fusion.c
Signature · 1 line
2.39.5
#1 lore raw

PATCH 1/3 iio: imu: st_lsm6dsx: set buffer sampling frequency for accelerometer only

The st_lsm6dsx_hwfifo_odr_store() function, which is called when userspace writes the buffer sampling frequency sysfs attribute, calls st_lsm6dsx_check_odr(), which accesses the odr_table array at index sensor->id; since this array is only 2 entries long, an access for any sensor type other than accelerometer or gyroscope is an out-of-bounds access.

To prevent userspace from triggering an out-of-bounds array access, and to support the only use case for which FIFO sampling frequency values different from the sensor sampling frequency may be needed (which is for keeping FIFO data rate low while sampling acceleration data at high rates for accurate event detection), do not create the buffer sampling frequency attribute for sensor types other than the accelerometer.

Fixes 6b648a36c200 ("iio: imu: st_lsm6dsx: Decouple sensor ODR from FIFO batch data rate")
---
 drivers/iio/imu/st_lsm6dsx/st_lsm6dsx_buffer.c | 11 ++++++++++-
 1 file changed, 10 insertions(+), 1 deletion(-)
drivers/iio/imu/st_lsm6dsx/st_lsm6dsx_buffer.c +10 -1
index 55d877745575..5ac45e6230b5 100644
@@ -858,12 +858,21 @@ int st_lsm6dsx_fifo_setup(struct st_lsm6dsx_hw *hw)
 	int i, ret;
 
 	for (i = 0; i < ST_LSM6DSX_ID_MAX; i++) {
+		const struct iio_dev_attr **attrs;
+
 		if (!hw->iio_devs[i])
 			continue;
 
+		/*
+		 * For the accelerometer, allow setting FIFO sampling frequency
+		 * values different from the sensor sampling frequency, which
+		 * may be needed to keep FIFO data rate low while sampling
+		 * acceleration data at high rates for accurate event detection.
+		 */
+		attrs = (i == ST_LSM6DSX_ID_ACC) ? st_lsm6dsx_buffer_attrs : NULL;
 		ret = devm_iio_kfifo_buffer_setup_ext(hw->dev, hw->iio_devs[i],
 						      &st_lsm6dsx_buffer_ops,
-						      st_lsm6dsx_buffer_attrs);
+						      attrs);
 		if (ret)
 			return ret;
 	}
Signature · 1 line
2.39.5
maintainer
#2 lore raw

Re: [PATCH 1/3] iio: imu: st_lsm6dsx: set buffer sampling frequency for accelerometer only

Quoted 7 lines · On Fri, 9 Jan 2026 19:15:26 +0100 Francesco Lavra <[email protected]> wrote:

The st_lsm6dsx_hwfifo_odr_store() function, which is called when userspace writes the buffer sampling frequency sysfs attribute, calls st_lsm6dsx_check_odr(), which accesses the odr_table array at index sensor->id; since this array is only 2 entries long, an access for any sensor type other than accelerometer or gyroscope is an out-of-bounds access.

Quoted 6 lines

To prevent userspace from triggering an out-of-bounds array access, and to support the only use case for which FIFO sampling frequency values different from the sensor sampling frequency may be needed (which is for keeping FIFO data rate low while sampling acceleration data at high rates for accurate event detection), do not create the buffer sampling frequency attribute for sensor types other than the accelerometer.

I'm not following why we need to drop this attribute for the gyroscope. Perhaps lay out what the combinations of controls are and the attributes we end up with.

As you note in the cover letter we can change this now with ABI issues as it is just in my tree, so I don't mind the change, just want to understand it a little better than I currently do!

Quoted 34 lines
Fixes 6b648a36c200 ("iio: imu: st_lsm6dsx: Decouple sensor ODR from FIFO batch data rate")
---
 drivers/iio/imu/st_lsm6dsx/st_lsm6dsx_buffer.c | 11 ++++++++++-
 1 file changed, 10 insertions(+), 1 deletion(-)
drivers/iio/imu/st_lsm6dsx/st_lsm6dsx_buffer.c +10 -1
index 55d877745575..5ac45e6230b5 100644
@@ -858,12 +858,21 @@ int st_lsm6dsx_fifo_setup(struct st_lsm6dsx_hw *hw)
 	int i, ret;
 
 	for (i = 0; i < ST_LSM6DSX_ID_MAX; i++) {
+		const struct iio_dev_attr **attrs;
+
 		if (!hw->iio_devs[i])
 			continue;
 
+		/*
+		 * For the accelerometer, allow setting FIFO sampling frequency
+		 * values different from the sensor sampling frequency, which
+		 * may be needed to keep FIFO data rate low while sampling
+		 * acceleration data at high rates for accurate event detection.
+		 */
+		attrs = (i == ST_LSM6DSX_ID_ACC) ? st_lsm6dsx_buffer_attrs : NULL;
 		ret = devm_iio_kfifo_buffer_setup_ext(hw->dev, hw->iio_devs[i],
 						      &st_lsm6dsx_buffer_ops,
-						      st_lsm6dsx_buffer_attrs);
+						      attrs);
 		if (ret)
 			return ret;
 	}
#3 lore raw

Re: [PATCH 1/3] iio: imu: st_lsm6dsx: set buffer sampling frequency for accelerometer only

Quoted 23 lines · Jonathan Cameron
Quoted 9 lines · On Fri,  9 Jan 2026 19:15:26 +0100 Francesco Lavra <[email protected]> wrote:

The st_lsm6dsx_hwfifo_odr_store() function, which is called when userspace
writes the buffer sampling frequency sysfs attribute, calls st_lsm6dsx_check_odr(), which accesses the odr_table array at index sensor->id; since this array is only 2 entries long, an access for any
sensor type other than accelerometer or gyroscope is an out-of-bounds access.

Quoted 10 lines

To prevent userspace from triggering an out-of-bounds array access, and to
support the only use case for which FIFO sampling frequency values different from the sensor sampling frequency may be needed (which is for
keeping FIFO data rate low while sampling acceleration data at high rates
for accurate event detection), do not create the buffer sampling frequency
attribute for sensor types other than the accelerometer.

Quoted 3 lines

I'm not following why we need to drop this attribute for the gyroscope. Perhaps lay out what the combinations of controls are and the attributes we end up with.

It's not like we need to drop this attribute, it's just that I don't see a need for it. The only reason I added this attribute was to be able to control (e.g. lower) the rate of data coming from the sensor while maintaining a high accuracy for event detection; and accurate event detection requires a high sampling rate for the accelerometer. So the gyroscope is not involved here, and the attribute is only needed for the accelerometer.

Before this change, we have:
- accel IIO device with separate samp_freq and buffer/samp_freq
- gyro IIO device with separate samp_freq and buffer/samp_freq
- (optionally) external sensor IIO devices with separate samp_freq and buffer/samp_freq (and trying to set buffer/samp_freq for these triggers an out-of-bounds array access)

After this change, we have the accel IIO device with separate samp_freq and buffer/samp_freq, while the other IIO devices have only a single samp_freq attribute.

Quoted 4 lines

As you note in the cover letter we can change this now with ABI issues as it is just in my tree, so I don't mind the change, just want to understand
it a little better than I currently do!

It's not just in your tree, it has been pulled into Linus's tree for 6.19.

maintainer
#4 lore raw

Re: [PATCH 1/3] iio: imu: st_lsm6dsx: set buffer sampling frequency for accelerometer only

Quoted 28 lines · On Mon, 12 Jan 2026 18:10:32 +0100 Francesco Lavra <[email protected]> wrote:
Quoted 23 lines · Jonathan Cameron
Quoted 9 lines · On Fri,  9 Jan 2026 19:15:26 +0100 Francesco Lavra <[email protected]> wrote:

The st_lsm6dsx_hwfifo_odr_store() function, which is called when userspace
writes the buffer sampling frequency sysfs attribute, calls st_lsm6dsx_check_odr(), which accesses the odr_table array at index sensor->id; since this array is only 2 entries long, an access for any
sensor type other than accelerometer or gyroscope is an out-of-bounds access.

Quoted 10 lines

To prevent userspace from triggering an out-of-bounds array access, and to
support the only use case for which FIFO sampling frequency values different from the sensor sampling frequency may be needed (which is for
keeping FIFO data rate low while sampling acceleration data at high rates
for accurate event detection), do not create the buffer sampling frequency
attribute for sensor types other than the accelerometer.

Quoted 3 lines

I'm not following why we need to drop this attribute for the gyroscope. Perhaps lay out what the combinations of controls are and the attributes we end up with.

Quoted 5 lines

It's not like we need to drop this attribute, it's just that I don't see a need for it. The only reason I added this attribute was to be able to control (e.g. lower) the rate of data coming from the sensor while maintaining a high accuracy for event detection; and accurate event detection requires a high sampling rate for the accelerometer.

Ok. So key here is for accelerations we are looking at impacts as a typical use case, whereas gyroscope tends to be slow orientation change stuff. That sounds a bit usecase specific. If someone is using these to detect shaft rotation issues they are going to care about sampling rates on the gyro as well, or is there something inherent in the gyroscope events (i.e. maybe there aren't any gyro events?) that makes this not relevant?

Quoted 15 lines

So the
gyroscope is not involved here, and the attribute is only needed for the accelerometer.

Before this change, we have:
- accel IIO device with separate samp_freq and buffer/samp_freq
- gyro IIO device with separate samp_freq and buffer/samp_freq
- (optionally) external sensor IIO devices with separate samp_freq and buffer/samp_freq (and trying to set buffer/samp_freq for these triggers an out-of-bounds array access)

After this change, we have the accel IIO device with separate samp_freq and buffer/samp_freq, while the other IIO devices have only a single samp_freq attribute.

Quoted 6 lines
Quoted 4 lines

As you note in the cover letter we can change this now with ABI issues as it is just in my tree, so I don't mind the change, just want to understand
it a little better than I currently do!

It's not just in your tree, it has been pulled into Linus's tree for 6.19.

Bit tight to get a fix in, though I gather we are going to rc8 this time so maybe.

Jonathan

Quoted 1 line
#5 lore raw

Re: [PATCH 1/3] iio: imu: st_lsm6dsx: set buffer sampling frequency for accelerometer only

Quoted 44 lines · Jonathan Cameron
Quoted 34 lines · On Mon, 12 Jan 2026 18:10:32 +0100 Francesco Lavra <[email protected]> wrote:
Quoted 27 lines · Jonathan Cameron

On Fri,  9 Jan 2026 19:15:26 +0100
Francesco Lavra <[email protected]> wrote:

Quoted 11 lines

The st_lsm6dsx_hwfifo_odr_store() function, which is called when userspace
writes the buffer sampling frequency sysfs attribute, calls st_lsm6dsx_check_odr(), which accesses the odr_table array at index sensor->id; since this array is only 2 entries long, an access for
any
sensor type other than accelerometer or gyroscope is an out-of- bounds
access.

Quoted 12 lines

To prevent userspace from triggering an out-of-bounds array access, and
to
support the only use case for which FIFO sampling frequency values different from the sensor sampling frequency may be needed (which is
for
keeping FIFO data rate low while sampling acceleration data at high rates
for accurate event detection), do not create the buffer sampling frequency
attribute for sensor types other than the accelerometer.

Quoted 5 lines

I'm not following why we need to drop this attribute for the gyroscope.
Perhaps lay out what the combinations of controls are and the attributes
we end up with.

Quoted 6 lines

It's not like we need to drop this attribute, it's just that I don't see a
need for it. The only reason I added this attribute was to be able to control (e.g. lower) the rate of data coming from the sensor while maintaining a high accuracy for event detection; and accurate event detection requires a high sampling rate for the accelerometer.

Quoted 8 lines

Ok. So key here is for accelerations we are looking at impacts as a typical
use case, whereas gyroscope tends to be slow orientation change stuff. That sounds a bit usecase specific. If someone is using these to detect shaft rotation
issues they are going to care about sampling rates on the gyro as well, or is there something inherent in the gyroscope events (i.e. maybe there aren't any gyro events?) that makes this not relevant?

All the events supported by this driver (motion detection and tap detection) use acceleration data only.
Some chip variants (e.g LSM6DSV) have more advanced features such as configurable finite state machines that can take inputs from both the accelerometer and the gyroscope and generate event interrupts; but I don't think these events would map cleanly to standard IIO event types.

maintainer
#6 lore raw

Re: [PATCH 1/3] iio: imu: st_lsm6dsx: set buffer sampling frequency for accelerometer only

Quoted 54 lines · On Fri, 16 Jan 2026 20:06:46 +0100 Francesco Lavra <[email protected]> wrote:
Quoted 44 lines · Jonathan Cameron
Quoted 34 lines · On Mon, 12 Jan 2026 18:10:32 +0100 Francesco Lavra <[email protected]> wrote:
Quoted 27 lines · Jonathan Cameron

On Fri,  9 Jan 2026 19:15:26 +0100
Francesco Lavra <[email protected]> wrote:

Quoted 11 lines

The st_lsm6dsx_hwfifo_odr_store() function, which is called when userspace
writes the buffer sampling frequency sysfs attribute, calls st_lsm6dsx_check_odr(), which accesses the odr_table array at index sensor->id; since this array is only 2 entries long, an access for
any
sensor type other than accelerometer or gyroscope is an out-of- bounds
access.

Quoted 12 lines

To prevent userspace from triggering an out-of-bounds array access, and
to
support the only use case for which FIFO sampling frequency values different from the sensor sampling frequency may be needed (which is
for
keeping FIFO data rate low while sampling acceleration data at high rates
for accurate event detection), do not create the buffer sampling frequency
attribute for sensor types other than the accelerometer.

Quoted 5 lines

I'm not following why we need to drop this attribute for the gyroscope.
Perhaps lay out what the combinations of controls are and the attributes
we end up with.

Quoted 6 lines

It's not like we need to drop this attribute, it's just that I don't see a
need for it. The only reason I added this attribute was to be able to control (e.g. lower) the rate of data coming from the sensor while maintaining a high accuracy for event detection; and accurate event detection requires a high sampling rate for the accelerometer.

Quoted 8 lines

Ok. So key here is for accelerations we are looking at impacts as a typical
use case, whereas gyroscope tends to be slow orientation change stuff. That sounds a bit usecase specific. If someone is using these to detect shaft rotation
issues they are going to care about sampling rates on the gyro as well, or is there something inherent in the gyroscope events (i.e. maybe there aren't any gyro events?) that makes this not relevant?

Quoted 7 lines

All the events supported by this driver (motion detection and tap detection) use acceleration data only.
Some chip variants (e.g LSM6DSV) have more advanced features such as configurable finite state machines that can take inputs from both the accelerometer and the gyroscope and generate event interrupts; but I don't think these events would map cleanly to standard IIO event types.

Doh! The fact we don't support gyro events was the detail I was completely missing ;(

Please add that to the patch description.

Thanks

Jonathan

#7 lore raw

PATCH 2/3 iio: imu: st_lsm6dsx: set FIFO ODR for accelerometer and magnetometer only

In reply to Francesco Lavra

The st_lsm6dsx_set_fifo_odr() function, which is called when enabling and disabling the hardware FIFO, checks the contents of the hw->settings->batch array at index sensor->id, and then sets the current ODR value in sensor registers that depend on whether the register address is set in the above array element. This logic is valid for internal sensors only, i.e. the accelerometer and magnetometer; however, since commit c91c1c844ebd ("iio:
imu: st_lsm6dsx: add i2c embedded controller support"), this function is called also when configuring the hardware FIFO for external sensors (i.e. sensors accessed through the sensor hub functionality), which can result in unrelated device registers being written.

Add a check to the beginning of st_lsm6dsx_set_fifo_odr() so that it does not touch any registers unless it is called for internal sensors.

Fixes c91c1c844ebd ("iio: imu: st_lsm6dsx: add i2c embedded controller support")
---
 drivers/iio/imu/st_lsm6dsx/st_lsm6dsx_buffer.c | 3 +++
 1 file changed, 3 insertions(+)
drivers/iio/imu/st_lsm6dsx/st_lsm6dsx_buffer.c +3 -0
index 5ac45e6230b5..9db48e835d4f 100644
@@ -225,6 +225,9 @@ static int st_lsm6dsx_set_fifo_odr(struct st_lsm6dsx_sensor *sensor,
 	const struct st_lsm6dsx_reg *batch_reg;
 	u8 data;
 
+	/* Only accel and gyro have batch registers. */
+	if (sensor->id >= ARRAY_SIZE(hw->settings->batch))
+		return 0;
 	batch_reg = &hw->settings->batch[sensor->id];
 	if (batch_reg->addr) {
 		int val;
Signature · 1 line
2.39.5
#8 lore raw

Re: [PATCH 2/3] iio: imu: st_lsm6dsx: set FIFO ODR for accelerometer and magnetometer only

Quoted 13 lines · Francesco Lavra

The st_lsm6dsx_set_fifo_odr() function, which is called when enabling and disabling the hardware FIFO, checks the contents of the hw->settings->batch array at index sensor->id, and then sets the current ODR value in sensor registers that depend on whether the register address is set in the above array element. This logic is valid for internal sensors only, i.e. the accelerometer and magnetometer; however, since commit c91c1c844ebd ("iio:
imu: st_lsm6dsx: add i2c embedded controller support"), this function is called also when configuring the hardware FIFO for external sensors (i.e. sensors accessed through the sensor hub functionality), which can result in unrelated device registers being written.

Add a check to the beginning of st_lsm6dsx_set_fifo_odr() so that it does not touch any registers unless it is called for internal sensors.

...

drivers/iio/imu/st_lsm6dsx/st_lsm6dsx_buffer.c static int st_lsm6dsx_set_fifo_odr(struct st_lsm6dsx_sensor *sensor, · line 230

Quoted 6 lines
drivers/iio/imu/st_lsm6dsx/st_lsm6dsx_buffer.c +3 -0
 	const struct st_lsm6dsx_reg *batch_reg;
 	u8 data;
 
+	/* Only accel and gyro have batch registers. */
+	if (sensor->id >= ARRAY_SIZE(hw->settings->batch))
+		return 0;

+ Blank line.

Quoted 1 line

batch_reg = &hw->settings->batch[sensor->id];

Signature · 2 lines
With Best Regards,
Andy Shevchenko
maintainer
#9 lore raw

Re: [PATCH 2/3] iio: imu: st_lsm6dsx: set FIFO ODR for accelerometer and magnetometer only

In reply to Francesco Lavra
Quoted 11 lines · On Fri, 9 Jan 2026 19:15:27 +0100 Francesco Lavra <[email protected]> wrote:

The st_lsm6dsx_set_fifo_odr() function, which is called when enabling and disabling the hardware FIFO, checks the contents of the hw->settings->batch array at index sensor->id, and then sets the current ODR value in sensor registers that depend on whether the register address is set in the above array element. This logic is valid for internal sensors only, i.e. the accelerometer and magnetometer; however, since commit c91c1c844ebd ("iio:
imu: st_lsm6dsx: add i2c embedded controller support"), this function is called also when configuring the hardware FIFO for external sensors (i.e. sensors accessed through the sensor hub functionality), which can result in unrelated device registers being written.

Quoted 5 lines

Add a check to the beginning of st_lsm6dsx_set_fifo_odr() so that it does not touch any registers unless it is called for internal sensors.

Fixes c91c1c844ebd ("iio: imu: st_lsm6dsx: add i2c embedded controller support")

This seems fine to me. Ideally it would have been first patch in the series as this is one we want to backport. I'll leave it on list little while though to see if Lorenzo or anyone else has time to take a look.

Thanks,

Jonathan

Quoted 18 lines
---
 drivers/iio/imu/st_lsm6dsx/st_lsm6dsx_buffer.c | 3 +++
 1 file changed, 3 insertions(+)
drivers/iio/imu/st_lsm6dsx/st_lsm6dsx_buffer.c +3 -0
index 5ac45e6230b5..9db48e835d4f 100644
@@ -225,6 +225,9 @@ static int st_lsm6dsx_set_fifo_odr(struct st_lsm6dsx_sensor *sensor,
 	const struct st_lsm6dsx_reg *batch_reg;
 	u8 data;
 
+	/* Only accel and gyro have batch registers. */
+	if (sensor->id >= ARRAY_SIZE(hw->settings->batch))
+		return 0;
 	batch_reg = &hw->settings->batch[sensor->id];
 	if (batch_reg->addr) {
 		int val;
maintainer
#10 lore raw

Re: [PATCH 2/3] iio: imu: st_lsm6dsx: set FIFO ODR for accelerometer and magnetometer only

Quoted 14 lines · On Sun, 11 Jan 2026 16:23:51 +0000 Jonathan Cameron <[email protected]> wrote:
Quoted 11 lines · On Fri, 9 Jan 2026 19:15:27 +0100 Francesco Lavra <[email protected]> wrote:

The st_lsm6dsx_set_fifo_odr() function, which is called when enabling and disabling the hardware FIFO, checks the contents of the hw->settings->batch array at index sensor->id, and then sets the current ODR value in sensor registers that depend on whether the register address is set in the above array element. This logic is valid for internal sensors only, i.e. the accelerometer and magnetometer; however, since commit c91c1c844ebd ("iio:
imu: st_lsm6dsx: add i2c embedded controller support"), this function is called also when configuring the hardware FIFO for external sensors (i.e. sensors accessed through the sensor hub functionality), which can result in unrelated device registers being written.

Quoted 9 lines
Quoted 5 lines

Add a check to the beginning of st_lsm6dsx_set_fifo_odr() so that it does not touch any registers unless it is called for internal sensors.

Fixes c91c1c844ebd ("iio: imu: st_lsm6dsx: add i2c embedded controller support")

This seems fine to me. Ideally it would have been first patch in the series as this is one we want to backport. I'll leave it on list little while though to see if Lorenzo or anyone else has time to take a look.

One thing...

Quoted 17 lines

Thanks,

Jonathan

Quoted 12 lines
---
 drivers/iio/imu/st_lsm6dsx/st_lsm6dsx_buffer.c | 3 +++
 1 file changed, 3 insertions(+)
drivers/iio/imu/st_lsm6dsx/st_lsm6dsx_buffer.c +0 -0
index 5ac45e6230b5..9db48e835d4f 100644
@@ -225,6 +225,9 @@ static int st_lsm6dsx_set_fifo_odr(struct st_lsm6dsx_sensor *sensor,
 	const struct st_lsm6dsx_reg *batch_reg;
 	u8 data;
 
Quoted 1 line

drivers/iio/imu/st_lsm6dsx/st_lsm6dsx_buffer.c static int st_lsm6dsx_set_fifo_odr(struct st_lsm6dsx_sensor *sensor, · line 228

Quoted 1 line
drivers/iio/imu/st_lsm6dsx/st_lsm6dsx_buffer.c +1 -0
+	/* Only accel and gyro have batch registers. */

It's true we can't use the batch register, but not clear from this comment that the else path below is inappropriate. Is it the absence of patch register or just that the set_fifo_odr is meaningless for other sensors that matters? I think this comment needs to provide more detail.

Quoted 6 lines
Quoted 5 lines
drivers/iio/imu/st_lsm6dsx/st_lsm6dsx_buffer.c +2 -0
+	if (sensor->id >= ARRAY_SIZE(hw->settings->batch))
+		return 0;
 	batch_reg = &hw->settings->batch[sensor->id];
 	if (batch_reg->addr) {
 		int val;
#11 lore raw

Re: [PATCH 2/3] iio: imu: st_lsm6dsx: set FIFO ODR for accelerometer and magnetometer only

Quoted 9 lines · Jonathan Cameron
Quoted 6 lines · On Sun, 11 Jan 2026 16:23:51 +0000 Jonathan Cameron <[email protected]> wrote:
Quoted 3 lines · On Fri,  9 Jan 2026 19:15:27 +0100 Francesco Lavra <[email protected]> wrote:

The st_lsm6dsx_set_fifo_odr() function, which is called when enabling and
disabling the hardware FIFO, checks the contents of the hw->settings-

Quoted 8 lines
Quoted 8 lines
Quoted 8 lines
Quoted 1 line

batch

array at index sensor->id, and then sets the current ODR value in sensor
registers that depend on whether the register address is set in the above
array element. This logic is valid for internal sensors only, i.e. the
accelerometer and magnetometer; however, since commit c91c1c844ebd

Oops: s/magnetometer/gyroscope/ (will fix this in the next iteration)
Quoted 44 lines
Quoted 13 lines
Quoted 13 lines

("iio:
imu: st_lsm6dsx: add i2c embedded controller support"), this function is
called also when configuring the hardware FIFO for external sensors (i.e.
sensors accessed through the sensor hub functionality), which can result in
unrelated device registers being written.

Add a check to the beginning of st_lsm6dsx_set_fifo_odr() so that it does
not touch any registers unless it is called for internal sensors.

Quoted 8 lines
Quoted 3 lines
Fixes c91c1c844ebd ("iio: imu: st_lsm6dsx: add i2c embedded

controller support")

This seems fine to me. Ideally it would have been first patch in the series
as this is one we want to backport.  I'll leave it on list little while though to see if Lorenzo or anyone else has time to take a look.

One thing...

Quoted 9 lines

Thanks,

Jonathan

Quoted 4 lines

---
 drivers/iio/imu/st_lsm6dsx/st_lsm6dsx_buffer.c | 3 +++

 1 file changed, 3 insertions(+)
Quoted 11 lines
Quoted 11 lines
drivers/iio/imu/st_lsm6dsx/st_lsm6dsx_buffer.c +0 -0
index 5ac45e6230b5..9db48e835d4f 100644
@@ -225,6 +225,9 @@ static int st_lsm6dsx_set_fifo_odr(struct
 st_lsm6dsx_sensor *sensor,
         const struct st_lsm6dsx_reg *batch_reg;
         u8 data;


+       /* Only accel and gyro have batch registers. */

Quoted 7 lines

It's true we can't use the batch register, but not clear from this comment
that the else path below is inappropriate.  Is it the absence of patch register
or just that the set_fifo_odr is meaningless for other sensors that matters?
I think this comment needs to provide more detail.

The set_fifo_odr is meaningless for external sensors, so the else path below is also inappropriate for external sensors; I will amend the comment to clarify that only internal sensors have a FIFO ODR configuration register.

Quoted 5 lines
Quoted 5 lines
Quoted 5 lines

+       if (sensor->id >= ARRAY_SIZE(hw->settings->batch)) +               return 0;
        batch_reg = &hw->settings->batch[sensor->id]; if (batch_reg->addr) {
                int val;

#12 lore raw

PATCH 3/3 iio: imu: st_lsm6dsx: add support for rotation sensor

In reply to Francesco Lavra

Some IMU chips in the LSM6DSX family have sensor fusion features that combine data from the accelerometer and gyroscope. One of these features generates rotation vector data and makes it available in the hardware FIFO as a quaternion (more specifically, the X, Y and Z components of the quaternion vector, expressed as 16-bit half-precision floating-point numbers).

Add support for a new sensor instance that allows receiving sensor fusion data, by defining a new struct st_lsm6dsx_sf_settings (which contains chip-specific details for the sensor fusion functionality), and adding this struct as a new field in struct st_lsm6dsx_settings. In st_lsm6dsx_core.c, populate this new struct for the LSM6DSV and LSM6DSV16X chips, and add the logic to initialize an additional IIO device if this struct is populated for the hardware type being probed.
Note: a new IIO device is being defined (as opposed to adding channels to an existing device) because each of the existing devices handles data coming from a single sensor, while sensor fusion data comes from multiple sensors.

Tested on LSMDSV16X.

---
 drivers/iio/imu/st_lsm6dsx/Makefile           |   2 +-
 drivers/iio/imu/st_lsm6dsx/st_lsm6dsx.h       |  26 +-
 .../iio/imu/st_lsm6dsx/st_lsm6dsx_buffer.c    |  16 +-
 drivers/iio/imu/st_lsm6dsx/st_lsm6dsx_core.c  |  58 +++++
 .../iio/imu/st_lsm6dsx/st_lsm6dsx_fusion.c    | 224 ++++++++++++++++++
 5 files changed, 319 insertions(+), 7 deletions(-)
 create mode 100644 drivers/iio/imu/st_lsm6dsx/st_lsm6dsx_fusion.c
drivers/iio/imu/st_lsm6dsx/Makefile +1 -1
index 57cbcd67d64f..19a488254de3 100644
@@ -1,6 +1,6 @@
 # SPDX-License-Identifier: GPL-2.0-only
 st_lsm6dsx-y := st_lsm6dsx_core.o st_lsm6dsx_buffer.o \
-		st_lsm6dsx_shub.o
+		st_lsm6dsx_shub.o st_lsm6dsx_fusion.o
 
 obj-$(CONFIG_IIO_ST_LSM6DSX) += st_lsm6dsx.o
 obj-$(CONFIG_IIO_ST_LSM6DSX_I2C) += st_lsm6dsx_i2c.o
drivers/iio/imu/st_lsm6dsx/st_lsm6dsx.h +21 -5
index 07b1773c87bd..4173f670f7af 100644
@@ -294,6 +294,7 @@ enum st_lsm6dsx_sensor_id {
 	ST_LSM6DSX_ID_EXT0,
 	ST_LSM6DSX_ID_EXT1,
 	ST_LSM6DSX_ID_EXT2,
+	ST_LSM6DSX_ID_SF,
 	ST_LSM6DSX_ID_MAX,
 };
 
@@ -301,6 +302,15 @@ enum st_lsm6dsx_ext_sensor_id {
 	ST_LSM6DSX_ID_MAGN,
 };
 
+struct st_lsm6dsx_sf_settings {
+	const struct iio_chan_spec *chan;
+	int chan_len;
+	struct st_lsm6dsx_odr_table_entry odr_table;
+	struct st_lsm6dsx_reg fifo_enable;
+	struct st_lsm6dsx_reg page_mux;
+	struct st_lsm6dsx_reg enable;
+};
+
 /**
  * struct st_lsm6dsx_ext_dev_settings - i2c controller slave settings
  * @i2c_addr: I2c slave address list.
@@ -388,6 +398,7 @@ struct st_lsm6dsx_settings {
 	struct st_lsm6dsx_hw_ts_settings ts_settings;
 	struct st_lsm6dsx_shub_settings shub_settings;
 	struct st_lsm6dsx_event_settings event_settings;
+	struct st_lsm6dsx_sf_settings sf_settings;
 };
 
 enum st_lsm6dsx_sensor_id {
@@ -510,6 +521,9 @@ int st_lsm6dsx_check_odr(struct st_lsm6dsx_sensor *sensor, u32 odr, u8 *val);
 int st_lsm6dsx_shub_probe(struct st_lsm6dsx_hw *hw, const char *name);
 int st_lsm6dsx_shub_set_enable(struct st_lsm6dsx_sensor *sensor, bool enable);
 int st_lsm6dsx_shub_read_output(struct st_lsm6dsx_hw *hw, u8 *data, int len);
+int st_lsm6dsx_sf_probe(struct st_lsm6dsx_hw *hw, const char *name);
+int st_lsm6dsx_sf_set_enable(struct st_lsm6dsx_sensor *sensor, bool enable);
+int st_lsm6dsx_sf_set_odr(struct st_lsm6dsx_sensor *sensor, bool enable);
 int st_lsm6dsx_set_page(struct st_lsm6dsx_hw *hw, bool enable);
 
 static inline int
@@ -564,12 +578,14 @@ st_lsm6dsx_get_mount_matrix(const struct iio_dev *iio_dev,
 static inline int
 st_lsm6dsx_device_set_enable(struct st_lsm6dsx_sensor *sensor, bool enable)
 {
-	if (sensor->id == ST_LSM6DSX_ID_EXT0 ||
-	    sensor->id == ST_LSM6DSX_ID_EXT1 ||
-	    sensor->id == ST_LSM6DSX_ID_EXT2)
+	switch (sensor->id) {
+	case ST_LSM6DSX_ID_EXT0 ... ST_LSM6DSX_ID_EXT2:
 		return st_lsm6dsx_shub_set_enable(sensor, enable);
-
-	return st_lsm6dsx_sensor_set_enable(sensor, enable);
+	case ST_LSM6DSX_ID_SF:
+		return st_lsm6dsx_sf_set_enable(sensor, enable);
+	default:
+		return st_lsm6dsx_sensor_set_enable(sensor, enable);
+	}
 }
 
 static const
drivers/iio/imu/st_lsm6dsx/st_lsm6dsx_buffer.c +15 -1
index 9db48e835d4f..a689c081c186 100644
@@ -88,6 +88,7 @@ enum st_lsm6dsx_fifo_tag {
 	ST_LSM6DSX_EXT0_TAG = 0x0f,
 	ST_LSM6DSX_EXT1_TAG = 0x10,
 	ST_LSM6DSX_EXT2_TAG = 0x11,
+	ST_LSM6DSX_ROT_TAG = 0x13,
 };
 
 static const
@@ -226,8 +227,11 @@ static int st_lsm6dsx_set_fifo_odr(struct st_lsm6dsx_sensor *sensor,
 	u8 data;
 
 	/* Only accel and gyro have batch registers. */
-	if (sensor->id >= ARRAY_SIZE(hw->settings->batch))
+	if (sensor->id >= ARRAY_SIZE(hw->settings->batch)) {
+		if (sensor->id == ST_LSM6DSX_ID_SF)
+			return st_lsm6dsx_sf_set_odr(sensor, enable);
 		return 0;
+	}
 	batch_reg = &hw->settings->batch[sensor->id];
 	if (batch_reg->addr) {
 		int val;
@@ -579,6 +583,16 @@ st_lsm6dsx_push_tagged_data(struct st_lsm6dsx_hw *hw, u8 tag,
 	case ST_LSM6DSX_EXT2_TAG:
 		iio_dev = hw->iio_devs[ST_LSM6DSX_ID_EXT2];
 		break;
+	case ST_LSM6DSX_ROT_TAG:
+		/**
+		 * The sensor reports only the {X, Y, Z} elements of the
+		 * quaternion vector; set the W value to 0 (it can be derived
+		 * from the {X, Y, Z} values due to the property that the vector
+		 * is normalized).
+		 */
+		*(u16 *)(data + ST_LSM6DSX_SAMPLE_SIZE) = 0;
+		iio_dev = hw->iio_devs[ST_LSM6DSX_ID_SF];
+		break;
 	default:
 		return -EINVAL;
 	}
drivers/iio/imu/st_lsm6dsx/st_lsm6dsx_core.c +58 -0

82 lines not on this page · raw message

drivers/iio/imu/st_lsm6dsx/st_lsm6dsx_fusion.c new file +224 -0

224 lines not on this page · raw message

Signature · 1 line
2.39.5
#13 lore raw

Re: [PATCH 3/3] iio: imu: st_lsm6dsx: add support for rotation sensor

Quoted 20 lines · Francesco Lavra

Some IMU chips in the LSM6DSX family have sensor fusion features that combine data from the accelerometer and gyroscope. One of these features generates rotation vector data and makes it available in the hardware FIFO as a quaternion (more specifically, the X, Y and Z components of the quaternion vector, expressed as 16-bit half-precision floating-point numbers).

Add support for a new sensor instance that allows receiving sensor fusion data, by defining a new struct st_lsm6dsx_sf_settings (which contains chip-specific details for the sensor fusion functionality), and adding this struct as a new field in struct st_lsm6dsx_settings. In st_lsm6dsx_core.c, populate this new struct for the LSM6DSV and LSM6DSV16X chips, and add the logic to initialize an additional IIO device if this struct is populated for the hardware type being probed.
Note: a new IIO device is being defined (as opposed to adding channels to an existing device) because each of the existing devices handles data coming from a single sensor, while sensor fusion data comes from multiple sensors.

Tested on LSMDSV16X.

...

drivers/iio/imu/st_lsm6dsx/st_lsm6dsx.h enum st_lsm6dsx_sensor_id { · line 298

Quoted 7 lines

enum st_lsm6dsx_sensor_id {

drivers/iio/imu/st_lsm6dsx/st_lsm6dsx.h +1 -0
 	ST_LSM6DSX_ID_EXT0,
 	ST_LSM6DSX_ID_EXT1,
 	ST_LSM6DSX_ID_EXT2,
+	ST_LSM6DSX_ID_SF,
 	ST_LSM6DSX_ID_MAX,

At some point please either get rid of _ID_MAX, or drop the trailing comma (maybe some other places also need the same treatment).

...

Quoted 4 lines
drivers/iio/imu/st_lsm6dsx/st_lsm6dsx_fusion.c new file +4 -0
+static int st_lsm6dsx_sf_set_page(struct st_lsm6dsx_hw *hw, bool enable)
+{
+	const struct st_lsm6dsx_reg *mux;
+	int err;

drivers/iio/imu/st_lsm6dsx/st_lsm6dsx_fusion.c line 59

Quoted 10 lines
drivers/iio/imu/st_lsm6dsx/st_lsm6dsx_fusion.c new file +10 -0
+
+	mux = &hw->settings->sf_settings.page_mux;
+	if (enable)
+		mutex_lock(&hw->page_lock);
+	err = regmap_assign_bits(hw->regmap, mux->addr, mux->mask, enable);
+	if (!enable || err < 0)
+		mutex_unlock(&hw->page_lock);
+
+	return err;
+}

Why not having properly made functions with the respective sparse annotations? And drop this "enable" parameter.

...

drivers/iio/imu/st_lsm6dsx/st_lsm6dsx_fusion.c line 73

Quoted 13 lines
drivers/iio/imu/st_lsm6dsx/st_lsm6dsx_fusion.c new file +13 -0
+int st_lsm6dsx_sf_set_enable(struct st_lsm6dsx_sensor *sensor, bool enable)
+{
+	struct st_lsm6dsx_hw *hw = sensor->hw;
+	const struct st_lsm6dsx_reg *enable_reg;
+	int err;
+
+	enable_reg = &hw->settings->sf_settings.enable;
+	err = st_lsm6dsx_sf_set_page(hw, true);
+	if (err < 0)
+		return err;
+
+	err = regmap_assign_bits(hw->regmap, enable_reg->addr, enable_reg->mask,
+				 enable);

One line? The variable name can be shortened as well.

Quoted 4 lines
drivers/iio/imu/st_lsm6dsx/st_lsm6dsx_fusion.c new file +4 -0
+	st_lsm6dsx_sf_set_page(hw, false);
+
+	return err;
+}

...

drivers/iio/imu/st_lsm6dsx/st_lsm6dsx_fusion.c line 121

Quoted 10 lines
drivers/iio/imu/st_lsm6dsx/st_lsm6dsx_fusion.c new file +10 -0
+static int st_lsm6dsx_sf_read_raw(struct iio_dev *iio_dev,
+				  struct iio_chan_spec const *ch,
+				  int *val, int *val2, long mask)
+{
+	struct st_lsm6dsx_sensor *sensor = iio_priv(iio_dev);
+
+	switch (mask) {
+	case IIO_CHAN_INFO_SAMP_FREQ:
+		*val = sensor->hwfifo_odr_mHz / MILLI;
+		*val2 = (sensor->hwfifo_odr_mHz % MILLI) * MILLI;

Strictly speaking the multiplier in the second one should be "(MICRO / MILLI)".

Quoted 5 lines
drivers/iio/imu/st_lsm6dsx/st_lsm6dsx_fusion.c new file +5 -0
+		return IIO_VAL_INT_PLUS_MICRO;
+	default:
+		return -EINVAL;
+	}
+}

...

Quoted 5 lines
drivers/iio/imu/st_lsm6dsx/st_lsm6dsx_fusion.c new file +5 -0
+static int st_lsm6dsx_sf_write_raw(struct iio_dev *iio_dev,
+				   struct iio_chan_spec const *chan,
+				   int val, int val2, long mask)
+{
+	struct st_lsm6dsx_sensor *sensor = iio_priv(iio_dev);

drivers/iio/imu/st_lsm6dsx/st_lsm6dsx_fusion.c line 142

Quoted 10 lines
drivers/iio/imu/st_lsm6dsx/st_lsm6dsx_fusion.c new file +10 -0
+	const struct st_lsm6dsx_sf_settings *settings;
+	int err;
+
+	settings = &sensor->hw->settings->sf_settings;
+	switch (mask) {
+	case IIO_CHAN_INFO_SAMP_FREQ: {
+		u32 odr_mHz;
+		u8 odr_val;
+
+		odr_mHz = val * MILLI + val2 / MILLI;

Ditto.

drivers/iio/imu/st_lsm6dsx/st_lsm6dsx_fusion.c line 154

Quoted 12 lines
drivers/iio/imu/st_lsm6dsx/st_lsm6dsx_fusion.c new file +11 -0
+		err = st_lsm6dsx_sf_get_odr_val(settings, odr_mHz, &odr_val);
+		if (err)
+			return err;
+
+		sensor->hwfifo_odr_mHz = odr_mHz;
+		break;
+	}
+	default:
+		return -EINVAL;
+	}
 
+	return 0;

Perhaps move it to the only case that needs it?

Quoted 1 line

+}

...

drivers/iio/imu/st_lsm6dsx/st_lsm6dsx_fusion.c line 166

Quoted 10 lines
drivers/iio/imu/st_lsm6dsx/st_lsm6dsx_fusion.c new file +10 -0
+static ssize_t st_lsm6dsx_sf_sampling_freq_avail(struct device *dev,
+						 struct device_attribute *attr,
+						 char *buf)
+{
+	struct st_lsm6dsx_sensor *sensor = iio_priv(dev_to_iio_dev(dev));
+	const struct st_lsm6dsx_sf_settings *settings;
+	int i, len = 0;
+
+	settings = &sensor->hw->settings->sf_settings;
+	for (i = 0; i < settings->odr_table.odr_len; i++) {
	for (unsigned int i ...) {

drivers/iio/imu/st_lsm6dsx/st_lsm6dsx_fusion.c line 170

Quoted 4 lines
drivers/iio/imu/st_lsm6dsx/st_lsm6dsx_fusion.c new file +4 -0
+		u32 val = settings->odr_table.odr_avl[i].milli_hz;
+
+		len += scnprintf(buf + len, PAGE_SIZE - len, "%lu.%03lu ",
+				 val / MILLI, val % MILLI);

Hmm... I think this has to be sysfs_emit_at().

Quoted 5 lines
drivers/iio/imu/st_lsm6dsx/st_lsm6dsx_fusion.c new file +5 -0
+	}
+	buf[len - 1] = '\n';
+
+	return len;
+}

...

drivers/iio/imu/st_lsm6dsx/st_lsm6dsx_fusion.c line 180

Quoted 4 lines
drivers/iio/imu/st_lsm6dsx/st_lsm6dsx_fusion.c new file +3 -0
+static struct attribute *st_lsm6dsx_sf_attributes[] = {
+	&iio_dev_attr_sampling_frequency_available.dev_attr.attr,
 
+	NULL,

No comma here.

Quoted 1 line

+};

...

Quoted 12 lines
drivers/iio/imu/st_lsm6dsx/st_lsm6dsx_fusion.c new file +12 -0
+int st_lsm6dsx_sf_probe(struct st_lsm6dsx_hw *hw, const char *name)
+{
+	const struct st_lsm6dsx_sf_settings *settings;
+	struct st_lsm6dsx_sensor *sensor;
+	struct iio_dev *iio_dev;
+
+	iio_dev = devm_iio_device_alloc(hw->dev, sizeof(*sensor));
+	if (!iio_dev)
+		return -ENOMEM;
+
+	settings = &hw->settings->sf_settings;
+	sensor = iio_priv(iio_dev);

drivers/iio/imu/st_lsm6dsx/st_lsm6dsx_fusion.c line 214

Quoted 10 lines
drivers/iio/imu/st_lsm6dsx/st_lsm6dsx_fusion.c new file +9 -0
+	sensor->id = ST_LSM6DSX_ID_SF;
+	sensor->hw = hw;
+	sensor->hwfifo_odr_mHz = settings->odr_table.odr_avl[0].milli_hz;
+	sensor->watermark = 1;
+	iio_dev->modes = INDIO_DIRECT_MODE;
+	iio_dev->info = &st_lsm6dsx_sf_info;
+	iio_dev->channels = settings->chan;
+	iio_dev->num_channels = settings->chan_len;
 
+	scnprintf(sensor->name, sizeof(sensor->name), "%s_sf", name);

What's the point of "c" here, please?

drivers/iio/imu/st_lsm6dsx/st_lsm6dsx_fusion.c line 217

Quoted 3 lines
drivers/iio/imu/st_lsm6dsx/st_lsm6dsx_fusion.c new file +3 -0
+	iio_dev->name = sensor->name;
+
+	/**

Huh?!

Quoted 7 lines
drivers/iio/imu/st_lsm6dsx/st_lsm6dsx_fusion.c new file +7 -0
+	 *  Put the IIO device pointer in the iio_devs array so that the caller
+	 *  can set up a buffer and register this IIO device.
+	 */
+	hw->iio_devs[ST_LSM6DSX_ID_SF] = iio_dev;
+
+	return 0;
+}
Signature · 2 lines
With Best Regards,
Andy Shevchenko
maintainer
#14 lore raw

Re: [PATCH 3/3] iio: imu: st_lsm6dsx: add support for rotation sensor

Quoted 26 lines · On Fri, 9 Jan 2026 21:22:09 +0200 Andy Shevchenko…
Quoted 20 lines · Francesco Lavra

Some IMU chips in the LSM6DSX family have sensor fusion features that combine data from the accelerometer and gyroscope. One of these features generates rotation vector data and makes it available in the hardware FIFO as a quaternion (more specifically, the X, Y and Z components of the quaternion vector, expressed as 16-bit half-precision floating-point numbers).

Add support for a new sensor instance that allows receiving sensor fusion data, by defining a new struct st_lsm6dsx_sf_settings (which contains chip-specific details for the sensor fusion functionality), and adding this struct as a new field in struct st_lsm6dsx_settings. In st_lsm6dsx_core.c, populate this new struct for the LSM6DSV and LSM6DSV16X chips, and add the logic to initialize an additional IIO device if this struct is populated for the hardware type being probed.
Note: a new IIO device is being defined (as opposed to adding channels to an existing device) because each of the existing devices handles data coming from a single sensor, while sensor fusion data comes from multiple sensors.

Tested on LSMDSV16X.

...

Quoted 1 line

enum st_lsm6dsx_sensor_id {

Quoted 8 lines

drivers/iio/imu/st_lsm6dsx/st_lsm6dsx.h enum st_lsm6dsx_sensor_id { · line 298

Quoted 5 lines
drivers/iio/imu/st_lsm6dsx/st_lsm6dsx.h +1 -0
 	ST_LSM6DSX_ID_EXT0,
 	ST_LSM6DSX_ID_EXT1,
 	ST_LSM6DSX_ID_EXT2,
+	ST_LSM6DSX_ID_SF,
 	ST_LSM6DSX_ID_MAX,  

At some point please either get rid of _ID_MAX, or drop the trailing comma (maybe some other places also need the same treatment).

It's already gone, this needs a rebase on the upstream tree.

Thanks,

Jonathan

maintainer
#15 lore raw

Re: [PATCH 3/3] iio: imu: st_lsm6dsx: add support for rotation sensor

In reply to Francesco Lavra
Quoted 7 lines · On Fri, 9 Jan 2026 19:15:28 +0100 Francesco Lavra <[email protected]> wrote:

Some IMU chips in the LSM6DSX family have sensor fusion features that combine data from the accelerometer and gyroscope. One of these features generates rotation vector data and makes it available in the hardware FIFO as a quaternion (more specifically, the X, Y and Z components of the quaternion vector, expressed as 16-bit half-precision floating-point numbers).

Quoted 11 lines

Add support for a new sensor instance that allows receiving sensor fusion data, by defining a new struct st_lsm6dsx_sf_settings (which contains chip-specific details for the sensor fusion functionality), and adding this struct as a new field in struct st_lsm6dsx_settings. In st_lsm6dsx_core.c, populate this new struct for the LSM6DSV and LSM6DSV16X chips, and add the logic to initialize an additional IIO device if this struct is populated for the hardware type being probed.
Note: a new IIO device is being defined (as opposed to adding channels to an existing device) because each of the existing devices handles data coming from a single sensor, while sensor fusion data comes from multiple sensors.

That doesn't really justify the extra IIO device. We used to do this because the IIO buffers aren't tagged with channel info (unlike say driver/input) and so if the data for different channels /sub parts of the sensor can come out at different rates, we have to split them up. So real question here is what data rate is this generated at? If it matches an existing sensor we should add the channel there, if not fine to have yet another IIO device. All this changes is the explanation in this patch description. Code is fine.

Note as an FYI, we now support multiple buffers per IIO device. I think there are probably still some corners that need ironing out, but in theory that was to handle this sort of sensor in a single IIO device.

Quoted 32 lines

Tested on LSMDSV16X.

drivers/iio/imu/st_lsm6dsx/st_lsm6dsx_fusion.c new file +19 -0
new file mode 100644
index 000000000000..7c78f14cbb91
@@ -0,0 +1,224 @@
+// SPDX-License-Identifier: GPL-2.0-only
+/*
+ * STMicroelectronics st_lsm6dsx IMU sensor fusion
+ */
+
+#include <linux/iio/iio.h>
+#include <linux/iio/sysfs.h>
+#include <linux/mutex.h>
+#include <linux/regmap.h>
+#include <linux/sprintf.h>
+#include <linux/types.h>
+#include <linux/units.h>
+
+#include "st_lsm6dsx.h"
 
+/**
+ * st_lsm6dsx_sf_set_page - Enable or disable access to sensor fusion
+ * configuration registers.
+ * @hw: Sensor hardware instance.
+ * @enable: True to enable access, false to disable access.

drivers/iio/imu/st_lsm6dsx/st_lsm6dsx_fusion.c line 47

Quoted 10 lines
drivers/iio/imu/st_lsm6dsx/st_lsm6dsx_fusion.c new file +10 -0
+ *
+ * The register page lock is acquired when enabling access, and released when
+ * disabling access. Therefore, a function call with @enable set to true must be
+ * followed by a call with @enable set to false (unless the first call returns
+ * an error value).
+ *
+ * Return: 0 on success, negative value on error.
+ */
+static int st_lsm6dsx_sf_set_page(struct st_lsm6dsx_hw *hw, bool enable)
+{

I'm curious what sparse thinks of this. Try a make C=1 build. I'd expect it to be fussy that it can't figure out the acquires and releases. Maybe it can see enough to figure it out. If not you might need to break it into two functions so you can add the markings __acquires() etc

Personally I think I'd just take the page lock out of here and use guard() in the callers.

drivers/iio/imu/st_lsm6dsx/st_lsm6dsx_fusion.c line 51

Quoted 4 lines
drivers/iio/imu/st_lsm6dsx/st_lsm6dsx_fusion.c new file +4 -0
+	const struct st_lsm6dsx_reg *mux;
+	int err;
+
+	mux = &hw->settings->sf_settings.page_mux;

Just to check, can you use regmaps support for paging instead of doing this by hand? That would avoid future problems with enabling caching or similar

Quoted 28 lines
drivers/iio/imu/st_lsm6dsx/st_lsm6dsx_fusion.c new file +27 -0
+	if (enable)
+		mutex_lock(&hw->page_lock);
+	err = regmap_assign_bits(hw->regmap, mux->addr, mux->mask, enable);
+	if (!enable || err < 0)
+		mutex_unlock(&hw->page_lock);
+
+	return err;
+}
+
 
+
+static int st_lsm6dsx_sf_write_raw(struct iio_dev *iio_dev,
+				   struct iio_chan_spec const *chan,
+				   int val, int val2, long mask)
+{
+	struct st_lsm6dsx_sensor *sensor = iio_priv(iio_dev);
+	const struct st_lsm6dsx_sf_settings *settings;
+	int err;
+
+	settings = &sensor->hw->settings->sf_settings;
+	switch (mask) {
+	case IIO_CHAN_INFO_SAMP_FREQ: {
+		u32 odr_mHz;
+		u8 odr_val;
+
+		odr_mHz = val * MILLI + val2 / MILLI;
+		err = st_lsm6dsx_sf_get_odr_val(settings, odr_mHz, &odr_val);
+		if (err)

drivers/iio/imu/st_lsm6dsx/st_lsm6dsx_fusion.c line 154

Quoted 10 lines
drivers/iio/imu/st_lsm6dsx/st_lsm6dsx_fusion.c new file +10 -0
+			return err;
+
+		sensor->hwfifo_odr_mHz = odr_mHz;
+		break;
+	}
+	default:
+		return -EINVAL;
+	}
+
+	return 0;

Might as well return instead of break above and save a couple of lines + avoid anyone having to go see what happens after the switch statement.

drivers/iio/imu/st_lsm6dsx/st_lsm6dsx_fusion.c line 180

Quoted 7 lines
drivers/iio/imu/st_lsm6dsx/st_lsm6dsx_fusion.c new file +6 -0
+}
 
+
+static IIO_DEV_ATTR_SAMP_FREQ_AVAIL(st_lsm6dsx_sf_sampling_freq_avail);
+static struct attribute *st_lsm6dsx_sf_attributes[] = {
+	&iio_dev_attr_sampling_frequency_available.dev_attr.attr,
+	NULL,

No comma as nothing should come after this. Ideally would be replaced with code using the read_avail callback and appropriate bit set in the info mask avail bitmaps.

Quoted 1 line

+};

Thanks

Jonathan

#16 lore raw

Re: [PATCH 3/3] iio: imu: st_lsm6dsx: add support for rotation sensor

Quoted 30 lines · Jonathan Cameron
Quoted 19 lines · On Fri,  9 Jan 2026 19:15:28 +0100 Francesco Lavra <[email protected]> wrote:

Some IMU chips in the LSM6DSX family have sensor fusion features that combine data from the accelerometer and gyroscope. One of these features
generates rotation vector data and makes it available in the hardware FIFO as a quaternion (more specifically, the X, Y and Z components of the
quaternion vector, expressed as 16-bit half-precision floating-point numbers).

Add support for a new sensor instance that allows receiving sensor fusion
data, by defining a new struct st_lsm6dsx_sf_settings (which contains chip-specific details for the sensor fusion functionality), and adding this
struct as a new field in struct st_lsm6dsx_settings. In st_lsm6dsx_core.c,
populate this new struct for the LSM6DSV and LSM6DSV16X chips, and add the
logic to initialize an additional IIO device if this struct is

Quoted 8 lines

populated
for the hardware type being probed.
Note: a new IIO device is being defined (as opposed to adding channels to
an existing device) because each of the existing devices handles data coming from a single sensor, while sensor fusion data comes from multiple
sensors.

Quoted 8 lines

That doesn't really justify the extra IIO device.  We used to do this because the IIO buffers aren't tagged with channel info (unlike say driver/input) and so if the data for different channels /sub parts of the sensor can come out at different rates, we have to split them up.  So real question here is what data rate is this generated at?  If it matches an existing sensor we should add the channel there, if not fine to have yet another IIO device. All this changes is the explanation in this patch description. Code is fine.

The data rate from the rotation sensor may not match the rate from any of the existing sensors. I changed the explanation in the patch description.

Quoted 59 lines

Note as an FYI, we now support multiple buffers per IIO device. I think there are probably still some corners that need ironing out, but in theory
that was to handle this sort of sensor in a single IIO device.

Quoted 39 lines

Tested on LSMDSV16X.

drivers/iio/imu/st_lsm6dsx/st_lsm6dsx_fusion.c new file +21 -0
new file mode 100644
index 000000000000..7c78f14cbb91
@@ -0,0 +1,224 @@
+// SPDX-License-Identifier: GPL-2.0-only
+/*
+ * STMicroelectronics st_lsm6dsx IMU sensor fusion
+ */
+
+#include <linux/iio/iio.h>
+#include <linux/iio/sysfs.h>
+#include <linux/mutex.h>
+#include <linux/regmap.h>
+#include <linux/sprintf.h>
+#include <linux/types.h>
+#include <linux/units.h>
+
+#include "st_lsm6dsx.h"
 
+/**
+ * st_lsm6dsx_sf_set_page - Enable or disable access to sensor fusion
+ * configuration registers.
+ * @hw: Sensor hardware instance.
+ * @enable: True to enable access, false to disable access.
+ *
+ * The register page lock is acquired when enabling access, and

released when
+ * disabling access. Therefore, a function call with @enable set to true must be
+ * followed by a call with @enable set to false (unless the first call

Quoted 8 lines

returns
+ * an error value).
+ *
+ * Return: 0 on success, negative value on error.
+ */
+static int st_lsm6dsx_sf_set_page(struct st_lsm6dsx_hw *hw, bool enable)
+{

I'm curious what sparse thinks of this.  Try a make C=1 build. I'd expect it to be fussy that it can't figure out the acquires and releases.
Maybe it can see enough to figure it out.  If not you might need to break it into two functions so you can add the markings __acquires() etc

Quoted 3 lines

Personally I think I'd just take the page lock out of here and use guard() in the
callers.

Building with make C=1 doesn't generate any sparse warnings (sparse is not invoked for this source file). On the other hand, if I break this function into two functions with the __acquires() and __releases() annotations, sparse generates a "context imbalance - wrong count at exit" warning for both functions (and their callers).
I ended up just taking the page lock out of here and using guard() in the callers.

Quoted 8 lines
Quoted 4 lines

+       const struct st_lsm6dsx_reg *mux; +       int err;
+
+       mux = &hw->settings->sf_settings.page_mux;

Just to check, can you use regmaps support for paging instead of doing this by hand?
That would avoid future problems with enabling caching or similar

No, the chip has multiple register sets with overlapping addresses, and struct regmap_range_cfg is not usable here.

Quoted 12 lines
Quoted 7 lines
drivers/iio/imu/st_lsm6dsx/st_lsm6dsx_fusion.c new file +2 -0
+
+static

IIO_DEV_ATTR_SAMP_FREQ_AVAIL(st_lsm6dsx_sf_sampling_freq_avail); +static struct attribute *st_lsm6dsx_sf_attributes[] = { +       &iio_dev_attr_sampling_frequency_available.dev_attr.attr, +       NULL,

No comma as nothing should come after this.  Ideally would be replaced with
code using the read_avail callback and appropriate bit set in the info mask avail bitmaps.

The read_avail callback would require available values to be in an int array, with one of the available IIO_VAL_* formats, but the driver uses a table (similarly to other parts of the existing code) where frequency values are expressed in mHz (which does not match any IIO_VAL_* format) and are interleaved with corresponding hardware register values. So we can't use the existing available frequency values in a read_avail callback.

maintainer
#17 lore raw

Re: [PATCH 3/3] iio: imu: st_lsm6dsx: add support for rotation sensor

Quoted 7 lines
Quoted 7 lines
Quoted 6 lines

+       const struct st_lsm6dsx_reg *mux; +       int err;
+
+       mux = &hw->settings->sf_settings.page_mux;

Quoted 6 lines
Quoted 3 lines

Just to check, can you use regmaps support for paging instead of doing this by hand?
That would avoid future problems with enabling caching or similar

No, the chip has multiple register sets with overlapping addresses, and struct regmap_range_cfg is not usable here.

Paged addressing always has overlapping addresses, so I'm not following. The trick this does is to map those higher pages to a fake set of addresses. An example is the ICM42600:
https://invensense.tdk.com/wp-content/uploads/2020/04/ds-000347_icm-42688-p-datasheet.pdf
for a suitable datasheet (some of the newer parts have a much more complex scheme)

I took a look at the datasheet and seems there are multiple types of paging going on and effectively ends up with two nested paging controls. So fair enough it doesn't fit here.

Quoted 14 lines
Quoted 13 lines
Quoted 8 lines
(quoted diff) +2 -0
+
+static

IIO_DEV_ATTR_SAMP_FREQ_AVAIL(st_lsm6dsx_sf_sampling_freq_avail); +static struct attribute *st_lsm6dsx_sf_attributes[] = { +       &iio_dev_attr_sampling_frequency_available.dev_attr.attr, +       NULL,

No comma as nothing should come after this.  Ideally would be replaced with
code using the read_avail callback and appropriate bit set in the info mask avail bitmaps.

Quoted 6 lines

The read_avail callback would require available values to be in an int array, with one of the available IIO_VAL_* formats, but the driver uses a table (similarly to other parts of the existing code) where frequency values are expressed in mHz (which does not match any IIO_VAL_* format) and are interleaved with corresponding hardware register values. So we can't use the existing available frequency values in a read_avail callback.

You would need to do some data mangling to create the relevant table in order to use that callback. We can always do that later I suppose if we find that it matters for some inkernel consumer or other reason.

Jonathan

Quoted 1 line
#18 lore raw

Re: [PATCH 3/3] iio: imu: st_lsm6dsx: add support for rotation sensor

Quoted 12 lines · Jonathan Cameron
Quoted 12 lines
Quoted 11 lines
Quoted 6 lines

+       const struct st_lsm6dsx_reg *mux; +       int err;
+
+       mux = &hw->settings->sf_settings.page_mux;

Just to check, can you use regmaps support for paging instead of doing
this by hand?
That would avoid future problems with enabling caching or similar

Quoted 6 lines
Quoted 2 lines

No, the chip has multiple register sets with overlapping addresses, and struct regmap_range_cfg is not usable here.

Paged addressing always has overlapping addresses, so I'm not following. The trick this does is to map those higher pages to a fake set of addresses.

I thought having a fake set of addresses in the driver would be frowned upon since it would create a mismatch between the code and the device datasheet. But I guess it wouldn't be so bad if appropriate offsets between register pages are chosen.
Anyway, as you noted in this case the different pages cannot be selected via a simple bank selector register.

Quoted 10 lines

An example is the ICM42600:
https://invensense.tdk.com/wp-content/uploads/2020/04/ds-000347_icm-42688-p-datasheet.pdf
for a suitable datasheet (some of the newer parts have a much more complex scheme)

I took a look at the datasheet and seems there are multiple types of paging
going on and effectively ends up with two nested paging controls. So fair enough it
doesn't fit here.