Re: [PATCH v5 1/1] iio/scmi: Add reading "raw" attribute.

[Andriy Tryshnivskyy]

On 17.10.21 14:51, Jonathan Cameron wrote:

> CAUTION: This email originated from outside of the organization.
> Do not click links or open attachments unless you recognize the sender and know the content is safe.
>
>
> On Fri,  8 Oct 2021 21:28:26 +0300
> Andriy Tryshnivskyy <andriy.tryshnivskyy@xxxxxxxxxxxxxxx> wrote:
>
> > Add scmi_iio_get_raw() to read "raw" attribute.
> >
> > Signed-off-by: Andriy Tryshnivskyy <andriy.tryshnivskyy@xxxxxxxxxxxxxxx>
> > ---
> For a single patch series, it is better to put a change log in the patch (here)
> Whilst I can see why you would use this approach rather than the read_raw callback
> there are significant disadvantages in doing so.  The channel can't be used
> by in kernel users such as iio-hwmon.   That may cause you more trouble than
> it is worth in the long run.
>
> Note that you could also define a new IIO_VAL type to still use the two
> (possibly) 32 bit values and return a 64 bit value. That way, with appropriate
> additions in the consumer drivers the channel could still be used.
>
> IIO_VAL_INT_64 perhaps with val as the lower 32 bits and val2 as the upper
> with appropriate care around the sign.

Hi Jonathan,

In the past I also was thinking about introducing a new type to return 64 bits value - IIO_VAL_INT_LONG.
However since it requires changes in

    include/linux/iio/types.h
    drivers/iio/industrialio-core.c

I thought it will be harder to get approval for this approach.
But now it seems there is no other way. So I will prepare a new patch version soon which introduces IIO_VAL_INT_LONG.

Thank you for your review!

> >   drivers/iio/common/scmi_sensors/scmi_iio.c | 61 ++++++++++++++++++++++
> >   1 file changed, 61 insertions(+)
> >
> > diff --git a/drivers/iio/common/scmi_sensors/scmi_iio.c b/drivers/iio/common/scmi_sensors/scmi_iio.c
> > index 7cf2bf282cef..691cbbd61e3a 100644
> > --- a/drivers/iio/common/scmi_sensors/scmi_iio.c
> > +++ b/drivers/iio/common/scmi_sensors/scmi_iio.c
> > @@ -311,6 +311,62 @@ static const struct iio_info scmi_iio_info = {
> >        .write_raw = scmi_iio_write_raw,
> >   };
> >
> > +static ssize_t scmi_iio_get_raw(struct iio_dev *iio_dev, uintptr_t private,
> > +                             const struct iio_chan_spec *chan, char *buf)
> > +{
> > +     struct scmi_iio_priv *sensor = iio_priv(iio_dev);
> > +     int err;
> > +     u32 sensor_config;
> > +     struct scmi_sensor_reading readings[SCMI_IIO_NUM_OF_AXIS];
> > +     int len = 0;
> > +
> > +     err = iio_device_claim_direct_mode(iio_dev);
> > +     if (err) {
> > +             dev_err(&iio_dev->dev,
> > +                     "Error in claiming direct mode for sensor %s err %d",
> > +                     sensor->sensor_info->name, err);
> It's not an error, it just means the device is busy, so at most dev_info()
> or just rely on userspace correctly interpreting EBUSY.

Agree. Will use dev_info().
Thank you!

> > +             goto err_release;
> > +     }
> > +
> > +     sensor_config = FIELD_PREP(SCMI_SENS_CFG_SENSOR_ENABLED_MASK,
> > +                                SCMI_SENS_CFG_SENSOR_ENABLE);
> > +     err = sensor->sensor_ops->config_set(
> > +             sensor->ph, sensor->sensor_info->id, sensor_config);
> > +     if (err) {
> > +             dev_err(&iio_dev->dev, "Error in enabling sensor %s err %d",
> > +                     sensor->sensor_info->name, err);
> > +             goto err_release;
> > +     }
> > +
> > +     err = sensor->sensor_ops->reading_get_timestamped(
> > +             sensor->ph, sensor->sensor_info->id,
> > +             sensor->sensor_info->num_axis, readings);
> > +     if (err) {
> > +             dev_err(&iio_dev->dev,
> > +                     "Error in reading raw attribute for sensor %s err %d",
> > +                     sensor->sensor_info->name, err);
> > +             goto err_release;
> > +     }
> > +
> > +     sensor_config = FIELD_PREP(SCMI_SENS_CFG_SENSOR_ENABLED_MASK,
> > +                                SCMI_SENS_CFG_SENSOR_DISABLE);
> > +     err = sensor->sensor_ops->config_set(
> > +             sensor->ph, sensor->sensor_info->id, sensor_config);
> > +     if (err) {
> > +             dev_err(&iio_dev->dev, "Error in disabling sensor %s err %d",
> > +                     sensor->sensor_info->name, err);
> > +             goto err_release;
> > +     }
> > +
> > +     len = scnprintf(buf, PAGE_SIZE, "%lld\n",
> > +                     readings[chan->scan_index].value);
> > +
> > +err_release:
> > +     iio_device_release_direct_mode(iio_dev);
> > +
> > +     return len;
> > +}
> > +
> >   static ssize_t scmi_iio_get_raw_available(struct iio_dev *iio_dev,
> >                                          uintptr_t private,
> >                                          const struct iio_chan_spec *chan,
> > @@ -355,6 +411,11 @@ static ssize_t scmi_iio_get_raw_available(struct iio_dev *iio_dev,
> >   }
> >
> >   static const struct iio_chan_spec_ext_info scmi_iio_ext_info[] = {
> > +     {
> > +             .name = "raw",
> > +             .read = scmi_iio_get_raw,
> > +             .shared = IIO_SEPARATE,
> > +     },
> >        {
> >                .name = "raw_available",
> >                .read = scmi_iio_get_raw_available,