Re: [PATCH] iio: inv_sensors: prevent timestamps going back in time in transitions

[Jonathan Cameron <jic23@kernel.org>]

On Tue, 20 Oct 2026 17:26:00 +0200
Jean-Baptiste Maneyrol via B4 Relay <devnull+jean-baptiste.maneyrol.tdk.com@kernel.org> wrote:

> From: Jean-Baptiste Maneyrol <jean-baptiste.maneyrol@tdk.com>
> 
> Ensure when computing first timestamp and when applying odr change,
> that the next timestamp will not be lower than the previous one.
> 
> This way there is no risk to have a timestamp going back in time, meaning
> being lower than the previous one. It already can never happen without
> a transition because we are generating timestamps by adding the period
> to the last timestamp.
> 
> Fixes: 0ecc363ccea7 ("iio: make invensense timestamp module generic")
> Cc: stable@vger.kernel.org
> Signed-off-by: Jean-Baptiste Maneyrol <jean-baptiste.maneyrol@tdk.com>
Maybe it is just me, but it does feel like there have been a lot of
bugs in this code.  It's complex fiddly stuff so perhaps not surprising.

Anyhow, applied to the fixes-togreg branch of iio.git

Thanks,

Jonathan

> ---
>  drivers/iio/common/inv_sensors/inv_sensors_timestamp.c | 16 ++++++++++++----
>  include/linux/iio/common/inv_sensors_timestamp.h       |  3 ++-
>  2 files changed, 14 insertions(+), 5 deletions(-)
> 
> diff --git a/drivers/iio/common/inv_sensors/inv_sensors_timestamp.c b/drivers/iio/common/inv_sensors/inv_sensors_timestamp.c
> index 8c9e81a77e68..9040da80f0c7 100644
> --- a/drivers/iio/common/inv_sensors/inv_sensors_timestamp.c
> +++ b/drivers/iio/common/inv_sensors/inv_sensors_timestamp.c
> @@ -133,10 +133,15 @@ void inv_sensors_timestamp_interrupt(struct inv_sensors_timestamp *ts,
>  		return;
>  
>  	/* no previous data, compute theoretical value from interrupt */
> -	if (ts->timestamp == 0) {
> +	if (!ts->init) {
>  		/* elapsed time: sensor period * sensor samples number */
>  		interval = (s64)ts->period * (s64)sample_nb;
> -		ts->timestamp = timestamp - interval;
> +		delta = timestamp - interval;
> +		/* prevent next timestamp to be lower than the last one */
> +		if ((delta + ts->period) > ts->timestamp)
> +			ts->timestamp = delta;
> +		ts->init = true;
> +
>  	}
>  
>  	/* update delta timestamps and estimated period */
> @@ -184,7 +189,7 @@ void inv_sensors_timestamp_apply_odr(struct inv_sensors_timestamp *ts,
>  				     u32 fifo_period, size_t fifo_nb,
>  				     unsigned int fifo_no)
>  {
> -	s64 interval;
> +	s64 interval, timestamp;
>  	u32 fifo_mult;
>  
>  	if (ts->new_mult == 0)
> @@ -207,7 +212,10 @@ void inv_sensors_timestamp_apply_odr(struct inv_sensors_timestamp *ts,
>  		fifo_period = fifo_mult * ts->chip_period.val;
>  		/* computes time interval between interrupt and this sample */
>  		interval = (s64)(fifo_nb - fifo_no) * (s64)fifo_period;
> -		ts->timestamp = ts->it.up - interval;
> +		timestamp = ts->it.up - interval;
> +		/* prevent next timestamp to be lower than current one */
> +		if ((timestamp + ts->period) > ts->timestamp)
> +			ts->timestamp = timestamp;
>  	}
>  }
>  EXPORT_SYMBOL_NS_GPL(inv_sensors_timestamp_apply_odr, "IIO_INV_SENSORS_TIMESTAMP");
> diff --git a/include/linux/iio/common/inv_sensors_timestamp.h b/include/linux/iio/common/inv_sensors_timestamp.h
> index 4f08204ede3b..91690778ae88 100644
> --- a/include/linux/iio/common/inv_sensors_timestamp.h
> +++ b/include/linux/iio/common/inv_sensors_timestamp.h
> @@ -66,6 +66,7 @@ struct inv_sensors_timestamp {
>  	u32 new_mult;
>  	u32 period;
>  	struct inv_sensors_timestamp_acc chip_period;
> +	bool init;
>  };
>  
>  void inv_sensors_timestamp_init(struct inv_sensors_timestamp *ts,
> @@ -94,7 +95,7 @@ static inline void inv_sensors_timestamp_reset(struct inv_sensors_timestamp *ts)
>  	ts->it = interval_init;
>  	ts->delta = interval_init;
>  	ts->delta_counter = 0;
> -	ts->timestamp = 0;
> +	ts->init = false;
>  }
>  
>  #endif
> 
> ---
> base-commit: 43ffda3c7f8456bbe641ae8249a077a361b97880
> change-id: 20261020-iio-common-inv-sensors-fixes-for-preventing-timestamp-in-past-8da19f683084
> 
> Best regards,
> --  
> Jean-Baptiste Maneyrol <jean-baptiste.maneyrol@tdk.com>
> 
>