BME-280 humidity over-/underflow

BME-280 humidity over-/underflow

[Hendrik Hoeth]

[-- Attachment #1: Type: text/plain, Size: 996 bytes --]

Hi,

I've discovered an over-/underflow problem for relative humidity
measurements in drivers/iio/pressure/bmp280-core.c. At very low
humidities the value measured by the sensor can fluctuate into the
negative humidity range. The kernel driver then reports this as
extremely large value (e.g. 4193085, corresponding to 4193% RH). The
same probably happens at high humidity values, but with our climate
chamber I was only able to look at the low end.

To avoid this, the reference code in the BME-280 datasheet catches
values < 0% RH and > 100% RH. These two lines are currently missing in
the kernel driver.

Attached you find a patch that fixed the problem for us. You can find
the sensor datasheet at
https://ae-bst.resource.bosch.com/media/_tech/media/datasheets/BST-BME280-DS002.pdf
The code I'm talking about is in section 4.2.3 on page 25/26.

Thanks a lot,

   Hendrik

-- 
Garrecht Avionik GmbH | Wieslocher Str. 38 | 69190 Walldorf
https://www.air-avionics.com/
Tel.: +49 6224 98 96 999

[-- Attachment #2: bmp280-core.c.patch --]
[-- Type: text/x-diff, Size: 622 bytes --]

Fix underflow / overflow for humidity measurements

diff --git a/drivers/iio/pressure/bmp280-core.c b/drivers/iio/pressure/bmp280-core.c
index 29c209cc1108..d1f36b044f05 100644
--- a/drivers/iio/pressure/bmp280-core.c
+++ b/drivers/iio/pressure/bmp280-core.c
@@ -270,6 +270,8 @@ static u32 bmp280_compensate_humidity(struct bmp280_data *data,
 		* (((var * (s32)calib->H3) >> 11) + (s32)32768)) >> 10)
 		+ (s32)2097152) * calib->H2 + 8192) >> 14);
 	var -= ((((var >> 15) * (var >> 15)) >> 7) * (s32)calib->H1) >> 4;
+	var = ((var < 0) ? 0 : var);
+	var = ((var > 419430400) ? 419430400 : var);
 
 	return var >> 12;
 };

Re: BME-280 humidity over-/underflow

[Tomasz Duszynski]

Hello Hendrik,

On Wed, Jun 10, 2020 at 03:45:07PM +0200, Hendrik Hoeth wrote:
> Hi,
>
> I've discovered an over-/underflow problem for relative humidity
> measurements in drivers/iio/pressure/bmp280-core.c. At very low
> humidities the value measured by the sensor can fluctuate into the
> negative humidity range. The kernel driver then reports this as
> extremely large value (e.g. 4193085, corresponding to 4193% RH). The
> same probably happens at high humidity values, but with our climate
> chamber I was only able to look at the low end.
>
> To avoid this, the reference code in the BME-280 datasheet catches
> values < 0% RH and > 100% RH. These two lines are currently missing in
> the kernel driver.
>
> Attached you find a patch that fixed the problem for us. You can find
> the sensor datasheet at
> https://ae-bst.resource.bosch.com/media/_tech/media/datasheets/BST-BME280-DS002.pdf
> The code I'm talking about is in section 4.2.3 on page 25/26.
>
> Thanks a lot,
>
>    Hendrik

Seems this particular issue has been already fixed here:
dee2dabc0e4115b80945fe2c91603e634f4b4686 iio: bmp280: fix compensation of humidity

>
> --
> Garrecht Avionik GmbH | Wieslocher Str. 38 | 69190 Walldorf
> https://www.air-avionics.com/
> Tel.: +49 6224 98 96 999

> Fix underflow / overflow for humidity measurements
>
> diff --git a/drivers/iio/pressure/bmp280-core.c b/drivers/iio/pressure/bmp280-core.c
> index 29c209cc1108..d1f36b044f05 100644
> --- a/drivers/iio/pressure/bmp280-core.c
> +++ b/drivers/iio/pressure/bmp280-core.c
> @@ -270,6 +270,8 @@ static u32 bmp280_compensate_humidity(struct bmp280_data *data,
>  		* (((var * (s32)calib->H3) >> 11) + (s32)32768)) >> 10)
>  		+ (s32)2097152) * calib->H2 + 8192) >> 14);
>  	var -= ((((var >> 15) * (var >> 15)) >> 7) * (s32)calib->H1) >> 4;
> +	var = ((var < 0) ? 0 : var);
> +	var = ((var > 419430400) ? 419430400 : var);
>
>  	return var >> 12;
>  };

Re: BME-280 humidity over-/underflow

[Hendrik Hoeth]

Hello Tomasz,

On Wed, 10 Jun 2020, Tomasz Duszynski <tomasz.duszynski@octakon.com> wrote:

> Seems this particular issue has been already fixed here:
> dee2dabc0e4115b80945fe2c91603e634f4b4686 iio: bmp280: fix compensation of humidity

Thanks, I hadn't seen that (I only checked stable).

Best regards, and sorry for the noise,

   Hendrik