Hello,
we should have 1 interrupt every time a sensor data has been acquired. So in theory this is not something that should happen. But real world is always a different story.
Do you have a case where you saw that happen?
The best way if this happens would be to create the timestamp based on the sampling rate since we know it (last timestamp + sampling interval). That would be very similar to the real value since the only difference is the clock drift between the chip and the system.
Best regards,
Jean-Baptiste Maneyrol
From: linux-iio-owner@xxxxxxxxxxxxxxx <linux-iio-owner@xxxxxxxxxxxxxxx> on behalf of Jonathan Cameron <jic23@xxxxxxxxxx>
Sent: Saturday, March 24, 2018 1:35:19 PM
To: Martin Kelly
Cc: linux-iio@xxxxxxxxxxxxxxx; Jean-Baptiste Maneyrol
Subject: How to handle missing timestamps? (was Re: [PATCH] iio: imu: inv_mpu6050: improve missing timestamp handling)
On Fri, 23 Mar 2018 17:02:40 -0700
Martin Kelly <mkelly@xxxxxxxx> wrote:
> When interrupts are generated at a slower rate than the FIFO queue fills
> up, we will have fewer timestamps than samples. Currently, we fill in 0
> for any unmatched timestamps. However, this is very confusing for
> userspace, which does not expect discontinuities in timestamps and
> must somehow work around the issue.
>
> Improve the situation by using the most recent timestamp when a
> timestamp is missing. Although this guess is not perfectly accurate, it
> is still close to the correct timestamp and won't result in the
> confusion caused by using 0.
>
> Signed-off-by: Martin Kelly <mkelly@xxxxxxxx>
Hmm. I would like to see where other peoples opinions on this lie.
The decision to mark it as 0 was deliberately made. There are a number
of applications where you have to 'know' the timestamps are incorrect
- pretending simply doesn't work. Arguably it is fine for a system
to detect that it is seeing repeated values and hence 'fix them up.
This is a change in ABI however which is going to be unfortunate
if we have code out there which is doing the right thing - interpolating
timestamps only once we have another known point to build from.
So opinions anyone?
Jonathan
> ---
> drivers/iio/imu/inv_mpu6050/inv_mpu_ring.c | 8 ++++++--
> 1 file changed, 6 insertions(+), 2 deletions(-)
>
> diff --git a/drivers/iio/imu/inv_mpu6050/inv_mpu_ring.c b/drivers/iio/imu/inv_mpu6050/inv_mpu_ring.c
> index ff81c6aa009d..a982037d5dad 100644
> --- a/drivers/iio/imu/inv_mpu6050/inv_mpu_ring.c
> +++ b/drivers/iio/imu/inv_mpu6050/inv_mpu_ring.c
> @@ -126,6 +126,7 @@ irqreturn_t inv_mpu6050_read_fifo(int irq, void *p)
> int result;
> u8 data[INV_MPU6050_OUTPUT_DATA_SIZE];
> u16 fifo_count;
> + s64 last_timestamp;
> s64 timestamp;
>
> mutex_lock(&st->lock);
> @@ -159,6 +160,7 @@ irqreturn_t inv_mpu6050_read_fifo(int irq, void *p)
> if (kfifo_len(&st->timestamps) >
> fifo_count / bytes_per_datum + INV_MPU6050_TIME_STAMP_TOR)
> goto flush_fifo;
> + last_timestamp = 0;
> while (fifo_count >= bytes_per_datum) {
> result = regmap_bulk_read(st->map, st->reg->fifo_r_w,
> data, bytes_per_datum);
> @@ -166,9 +168,11 @@ irqreturn_t inv_mpu6050_read_fifo(int irq, void *p)
> goto flush_fifo;
>
> result = kfifo_out(&st->timestamps, ×tamp, 1);
> - /* when there is no timestamp, put timestamp as 0 */
> + /* when there is no timestamp, just use the last one we saw */
> if (result == 0)
> - timestamp = 0;
> + timestamp = last_timestamp;
> + else
> + last_timestamp = timestamp;
>
> result = iio_push_to_buffers_with_timestamp(indio_dev, data,
> timestamp);
--
To unsubscribe from this list: send the line "unsubscribe linux-iio" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
VGER.KERNEL.ORG - Majordomo info
vger.kernel.org
Very short Majordomo intro. Send request in email to address <majordomo@xxxxxxxxxxxxxxx> To subscribe a list (``linux-kernel'' is given as an example), use following ...
--
To unsubscribe from this list: send the line "unsubscribe linux-iio" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
