On 08/16/2012 07:17 PM, Jonathan Cameron wrote:
> On 08/16/2012 10:18 AM, Lars-Peter Clausen wrote:
>> This series is just a request for comments right now and at this point it is
>> barley tested.
>>
>> Currently IIO uses a decimal fixed point representations for real type numbers.
>> This patch introduces a new representation for rational type numbers. The number
>> will be expressed by specifying a numerator and denominator. For converting a
>> raw value to a processed value multiply it by the numerator and divide it by the
>> denominator.
>>
>> The reasoning for introducing this new type is that for a lot of devices the
>> scale can be represented easily by a fractional number, but it is not possible
>> to represent it as fixed point number without rounding. E.g. for a simple DAC
>> the scale is often the reference voltage divided by the number of possible
>> values (Usually 2**n_bits - 1). Each driver currently implements the conversion
>> of this fraction to a fixed point number on its own.
>>
>> Also when it comes to the in-kernel interface this allows to directly use the
>> fractional factors to convert a raw value to a processed value. This should on
>> one hand require less instructions and on the other hand increase the
>> precession.
> Hmm.. Is the range of this going to be large enough? Or do we need to have
> variants of the fractional value?
Well the range is from +-2147483648 to +-0.0000000005, but of course you can
have any value in between. E.g 2147483647.0000000005 would not be possible,
but it shouldn't be needed. Another advantage of this encoding is that the
smaller your factor is the more effective decimal places you get.
On the other hand though do we really need a scale factor as large as
2147483648 or would 2147483.648 or even 2147.483648 be enough?
>
> Otherwise this looks like a reasonable idea to me. Few comments inline.
>>
>> Signed-off-by: Lars-Peter Clausen <lars@xxxxxxxxxx>
>> ---
>> drivers/iio/industrialio-core.c | 9 +++++++++
>> include/linux/iio/types.h | 1 +
>> 2 files changed, 10 insertions(+)
>>
>> diff --git a/drivers/iio/industrialio-core.c b/drivers/iio/industrialio-core.c
>> index 2ec266e..69e60c5 100644
>> --- a/drivers/iio/industrialio-core.c
>> +++ b/drivers/iio/industrialio-core.c
>> @@ -365,6 +365,7 @@ static ssize_t iio_read_channel_info(struct device *dev,
>> {
>> struct iio_dev *indio_dev = dev_to_iio_dev(dev);
>> struct iio_dev_attr *this_attr = to_iio_dev_attr(attr);
>> + unsigned long long tmp;
>> int val, val2;
>> bool scale_db = false;
>> int ret = indio_dev->info->read_raw(indio_dev, this_attr->c,
>> @@ -390,6 +391,14 @@ static ssize_t iio_read_channel_info(struct device *dev,
>> return sprintf(buf, "-%d.%09u\n", val, -val2);
>> else
>> return sprintf(buf, "%d.%09u\n", val, val2);
>> + case IIO_VAL_FRACTIONAL:
>> + tmp = div_s64((s64)val * 1000000000LL, val2);
>> + val2 = do_div(tmp, 1000000000LL);
>> + val = tmp;
>> + return sprintf(buf, "%d.%09u\n", val, val2);
>> +
>> + /* Or maybe we could do something like this:
>> + return sprintf(buf, "%d/%d\n", val, val2); */
> Not sure on this one... Will wait for anyone elses comments.
> So far we have pretty printed eveything the same. Hence this will
> add a fair bit of complexity to userspace code (and will cause
> issues for any existing code!)
>
>> default:
>> return 0;
>> }
>> diff --git a/include/linux/iio/types.h b/include/linux/iio/types.h
>> index 44e3977..8ef5fde 100644
>> --- a/include/linux/iio/types.h
>> +++ b/include/linux/iio/types.h
>> @@ -57,5 +57,6 @@ enum iio_modifier {
>> #define IIO_VAL_INT_PLUS_MICRO 2
>> #define IIO_VAL_INT_PLUS_NANO 3
>> #define IIO_VAL_INT_PLUS_MICRO_DB 4
>> +#define IIO_VAL_FRACTIONAL 5
> I wonder if it is worth leaving a gap here for
> any more of the int plus form?
I guess it wouldn't hurt.
>>
>> #endif /* _IIO_TYPES_H_ */
>>
> --
> 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
--
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
