Re: [PATCH v4] iio: cros_ec_sensors: add cros_ec_activity driver

[Jonathan Cameron <jic23@kernel.org>]

> > > +     for_each_set_bit(i, &activities, BITS_PER_LONG) {
> > > +             /* List all available triggers */
> > > +             if (i == MOTIONSENSE_ACTIVITY_BODY_DETECTION) {
> > > +                     channel->type = IIO_PROXIMITY;
> > > +                     channel->info_mask_separate = BIT(IIO_CHAN_INFO_RAW);
> > > +                     channel->modified = 0;  
> >
> > Not modified is the 'obvious' default so can probably drop that one.
> >  
> > > +                     channel->event_spec = cros_ec_body_detect_events;
> > > +                     channel->num_event_specs =
> > > +                             ARRAY_SIZE(cros_ec_body_detect_events);
> > > +                     st->body_detection_channel_index = index;
> > > +             } else {
> > > +                     channel->type = IIO_ACTIVITY;  
> >
> > This is getting a little creative.  Do we know anything about what the
> > underlying detection actually is?  Activity sensor are often some form
> > of magnitude sensor on a specific type of data.  I guess if we have no
> > idea what this then activity detection makes some sense.  
> As written earlier, the event is triggered when a threshold is
> crossed. So contrary to other activity sensors like walking or
> swimming, there is no confidence level, it just happened that we are
> not still anymore.

The confidence level is a bit of a fiction for all sensors. I just wanted
the ABI to allow for algorithms getting more clever over time.  Right now
they all claim to be 0 or 100.

Anyhow, I don't think that matters anyway.