[PATCH] hwmon: (ntc_thermistor): Use library interpolation

[PATCH] hwmon: (ntc_thermistor): Use library interpolation

[Linus Walleij]

The kernel has a helper function for linear interpolation so
use it. It incidentally makes the code easier to read as well.

Cc: Peter Rosin <peda@axentia.se>
Cc: Chris Lesiak <chris.lesiak@licor.com>
Cc: linux-iio@vger.kernel.org
Signed-off-by: Linus Walleij <linus.walleij@linaro.org>
---
 drivers/hwmon/ntc_thermistor.c | 11 +++++++----
 1 file changed, 7 insertions(+), 4 deletions(-)

diff --git a/drivers/hwmon/ntc_thermistor.c b/drivers/hwmon/ntc_thermistor.c
index 8587189c7f15..61bd0e074ec9 100644
--- a/drivers/hwmon/ntc_thermistor.c
+++ b/drivers/hwmon/ntc_thermistor.c
@@ -14,6 +14,7 @@
 #include <linux/err.h>
 #include <linux/of.h>
 #include <linux/of_device.h>
+#include <linux/fixp-arith.h>
 
 #include <linux/platform_data/ntc_thermistor.h>
 
@@ -557,10 +558,12 @@ static int get_temp_mc(struct ntc_data *data, unsigned int ohm)
 		/* Unable to use linear approximation */
 		temp = data->comp[low].temp_c * 1000;
 	} else {
-		temp = data->comp[low].temp_c * 1000 +
-			((data->comp[high].temp_c - data->comp[low].temp_c) *
-			 1000 * ((int)ohm - (int)data->comp[low].ohm)) /
-			((int)data->comp[high].ohm - (int)data->comp[low].ohm);
+		temp = fixp_linear_interpolate(data->comp[low].ohm,
+					       data->comp[low].temp_c,
+					       data->comp[high].ohm,
+					       data->comp[high].temp_c,
+					       ohm);
+		temp *= 1000;
 	}
 	return temp;
 }
-- 
2.31.1

Re: [PATCH] hwmon: (ntc_thermistor): Use library interpolation

[Jonathan Cameron]

On Sat,  3 Jul 2021 20:01:31 +0200
Linus Walleij <linus.walleij@linaro.org> wrote:

> The kernel has a helper function for linear interpolation so
> use it. It incidentally makes the code easier to read as well.
> 
> Cc: Peter Rosin <peda@axentia.se>
> Cc: Chris Lesiak <chris.lesiak@licor.com>
> Cc: linux-iio@vger.kernel.org
> Signed-off-by: Linus Walleij <linus.walleij@linaro.org>

Potential precision problem because the multiplication by 1000 is now done
post division whereas before it was before it?

I'm only eyeballing the code so no idea if it makes a practical difference.

> ---
>  drivers/hwmon/ntc_thermistor.c | 11 +++++++----
>  1 file changed, 7 insertions(+), 4 deletions(-)
> 
> diff --git a/drivers/hwmon/ntc_thermistor.c b/drivers/hwmon/ntc_thermistor.c
> index 8587189c7f15..61bd0e074ec9 100644
> --- a/drivers/hwmon/ntc_thermistor.c
> +++ b/drivers/hwmon/ntc_thermistor.c
> @@ -14,6 +14,7 @@
>  #include <linux/err.h>
>  #include <linux/of.h>
>  #include <linux/of_device.h>
> +#include <linux/fixp-arith.h>
>  
>  #include <linux/platform_data/ntc_thermistor.h>
>  
> @@ -557,10 +558,12 @@ static int get_temp_mc(struct ntc_data *data, unsigned int ohm)
>  		/* Unable to use linear approximation */
>  		temp = data->comp[low].temp_c * 1000;
>  	} else {
> -		temp = data->comp[low].temp_c * 1000 +
> -			((data->comp[high].temp_c - data->comp[low].temp_c) *
> -			 1000 * ((int)ohm - (int)data->comp[low].ohm)) /
> -			((int)data->comp[high].ohm - (int)data->comp[low].ohm);
> +		temp = fixp_linear_interpolate(data->comp[low].ohm,
> +					       data->comp[low].temp_c,
> +					       data->comp[high].ohm,
> +					       data->comp[high].temp_c,
> +					       ohm);
> +		temp *= 1000;
>  	}
>  	return temp;
>  }

Re: [PATCH] hwmon: (ntc_thermistor): Use library interpolation

[Guenter Roeck]

On 7/4/21 9:14 AM, Jonathan Cameron wrote:
> On Sat,  3 Jul 2021 20:01:31 +0200
> Linus Walleij <linus.walleij@linaro.org> wrote:
> 
>> The kernel has a helper function for linear interpolation so
>> use it. It incidentally makes the code easier to read as well.
>>
>> Cc: Peter Rosin <peda@axentia.se>
>> Cc: Chris Lesiak <chris.lesiak@licor.com>
>> Cc: linux-iio@vger.kernel.org
>> Signed-off-by: Linus Walleij <linus.walleij@linaro.org>
> 
> Potential precision problem because the multiplication by 1000 is now done
> post division whereas before it was before it?
> 
> I'm only eyeballing the code so no idea if it makes a practical difference.
> 

I think so, since the code now only reports temperatures in full degrees C
which wasn't the case earlier. That should be easy to fix though by
passing data->comp[low].temp_c * 1000 and data->comp[high].temp_c * 1000
as parameters to fixp_linear_interpolate().

Also, the if/else in get_temp_mc() seems now unnecessary since
fixp_linear_interpolate() effectively does the same checks.

Guenter

>> ---
>>   drivers/hwmon/ntc_thermistor.c | 11 +++++++----
>>   1 file changed, 7 insertions(+), 4 deletions(-)
>>
>> diff --git a/drivers/hwmon/ntc_thermistor.c b/drivers/hwmon/ntc_thermistor.c
>> index 8587189c7f15..61bd0e074ec9 100644
>> --- a/drivers/hwmon/ntc_thermistor.c
>> +++ b/drivers/hwmon/ntc_thermistor.c
>> @@ -14,6 +14,7 @@
>>   #include <linux/err.h>
>>   #include <linux/of.h>
>>   #include <linux/of_device.h>
>> +#include <linux/fixp-arith.h>
>>   
>>   #include <linux/platform_data/ntc_thermistor.h>
>>   
>> @@ -557,10 +558,12 @@ static int get_temp_mc(struct ntc_data *data, unsigned int ohm)
>>   		/* Unable to use linear approximation */
>>   		temp = data->comp[low].temp_c * 1000;
>>   	} else {
>> -		temp = data->comp[low].temp_c * 1000 +
>> -			((data->comp[high].temp_c - data->comp[low].temp_c) *
>> -			 1000 * ((int)ohm - (int)data->comp[low].ohm)) /
>> -			((int)data->comp[high].ohm - (int)data->comp[low].ohm);
>> +		temp = fixp_linear_interpolate(data->comp[low].ohm,
>> +					       data->comp[low].temp_c,
>> +					       data->comp[high].ohm,
>> +					       data->comp[high].temp_c,
>> +					       ohm);
>> +		temp *= 1000;
>>   	}
>>   	return temp;
>>   }
>