Re: [PATCH v3 3/5] iio: common: ssp_sensors: Add sensorhub iio commons

[Jonathan Cameron <jic23@kernel.org>]

On 13/01/15 16:03, Karol Wrona wrote:
> [...]
>>> +EXPORT_SYMBOL(ssp_common_setup_buffer);
>>> diff --git a/drivers/iio/common/ssp_sensors/ssp_iio_sensor.h b/drivers/iio/common/ssp_sensors/ssp_iio_sensor.h
>>> new file mode 100644
>>> index 0000000..4b79be0
>>> --- /dev/null
>>> +++ b/drivers/iio/common/ssp_sensors/ssp_iio_sensor.h
>>> @@ -0,0 +1,56 @@
>>> +#ifndef __SSP_IIO_SENSOR_H__
>>> +#define __SSP_IIO_SENSOR_H__
>>> +
>>> +#define SSP_CHANNEL_AG(_type, _mod, _index) \
>>> +{ \
>>> +		.type = _type,\
>>> +		.modified = 1,\
>>> +		.channel2 = _mod,\
>>> +		.info_mask_shared_by_type = BIT(IIO_CHAN_INFO_SAMP_FREQ),\
>>> +		.scan_index = _index,\
>>> +		.scan_type = {\
>>> +			.sign = 's',\
>>> +			.realbits = 16,\
>> Seems a little unlikely unless the device is padding appropriately. Even
>> then it is generally a bad idea to claim more accuracy than is real. Very
>> few accelerometers etc are 16 bit.  Please confirm.
>>> +			.storagebits = 16,\
> Physically it is mpu6500:
> http://www.invensense.com/mems/gyro/documents/PS-MPU-6500A-01.pdf.
> It has 16-bit ADC.  Other thing linux device does not get these data directly
> but after some processing by sensorhub and in this case it always will be 16-bit
> sample.
Fair enough, it'll be somewhat noisy but there might be some value in those low
bits ;)
> 
> 
> 
>>> +			.shift = 0,\
>>> +			.endianness = IIO_LE,\
>>> +		},\
>>> +}
>>> +
>>> +#define SSP_DIVISOR	1000000ULL
>>> +#define SSP_MS_PER_S	1000
> [...]
>