iio: imu: inv_mpu6050: stabilized timestamping in interrupt

Use IRQ ONESHOT flag to ensure the timestamp is not updated in the
hard handler during the thread handler. And use a fixed value of 1
sample that correspond to this first timestamp.

This way we can ensure the timestamp is always corresponding to the
value used by the timestamping mechanism. Otherwise, it is possible
that between FIFO count read and FIFO processing the timestamp is
overwritten in the hard handler.

Fixes: 111e1abd00 ("iio: imu: inv_mpu6050: use the common inv_sensors timestamp module")
Cc: stable@vger.kernel.org
Signed-off-by: Jean-Baptiste Maneyrol <jean-baptiste.maneyrol@tdk.com>
Link: https://lore.kernel.org/r/20240527150117.608792-1-inv.git-commit@tdk.com
Signed-off-by: Jonathan Cameron <Jonathan.Cameron@huawei.com>
This commit is contained in:
Jean-Baptiste Maneyrol 2024-05-27 15:01:17 +00:00 committed by Jonathan Cameron
parent 182bc496dc
commit 8844ed0a6e
2 changed files with 3 additions and 2 deletions

View file

@ -100,8 +100,8 @@ irqreturn_t inv_mpu6050_read_fifo(int irq, void *p)
goto end_session;
/* Each FIFO data contains all sensors, so same number for FIFO and sensor data */
fifo_period = NSEC_PER_SEC / INV_MPU6050_DIVIDER_TO_FIFO_RATE(st->chip_config.divider);
inv_sensors_timestamp_interrupt(&st->timestamp, nb, pf->timestamp);
inv_sensors_timestamp_apply_odr(&st->timestamp, fifo_period, nb, 0);
inv_sensors_timestamp_interrupt(&st->timestamp, 1, pf->timestamp);
inv_sensors_timestamp_apply_odr(&st->timestamp, fifo_period, 1, 0);
/* clear internal data buffer for avoiding kernel data leak */
memset(data, 0, sizeof(data));

View file

@ -300,6 +300,7 @@ int inv_mpu6050_probe_trigger(struct iio_dev *indio_dev, int irq_type)
if (!st->trig)
return -ENOMEM;
irq_type |= IRQF_ONESHOT;
ret = devm_request_threaded_irq(&indio_dev->dev, st->irq,
&inv_mpu6050_interrupt_timestamp,
&inv_mpu6050_interrupt_handle,