drivers/iio/adc/ad7944.c | 2 ++ 1 file changed, 2 insertions(+)
Apply a mask to the raw value received over the SPI bus for unsigned
direct reads. As we found recently, SPI controllers may not set unused
bits to 0 when reading with bits_per_word != {8,16,32}. The ad7944 uses
bits_per_word of 14 and 18, so we need to mask the value to be sure we
returning the correct value to userspace during a direct read.
Fixes: d1efcf8871db ("iio: adc: ad7944: add driver for AD7944/AD7985/AD7986")
Signed-off-by: David Lechner <dlechner@baylibre.com>
---
The sign_extend32() already takes care of signed reads in case that
isn't obvious - it overwrites the unknown bits with the sign bit.
---
drivers/iio/adc/ad7944.c | 2 ++
1 file changed, 2 insertions(+)
diff --git a/drivers/iio/adc/ad7944.c b/drivers/iio/adc/ad7944.c
index 2f949fe5587318957f2e423029294ced0a6f803d..37a137bd83571b055e970a8cd483d8726972d637 100644
--- a/drivers/iio/adc/ad7944.c
+++ b/drivers/iio/adc/ad7944.c
@@ -377,6 +377,8 @@ static int ad7944_single_conversion(struct ad7944_adc *adc,
if (chan->scan_type.sign == 's')
*val = sign_extend32(*val, chan->scan_type.realbits - 1);
+ else
+ *val &= GENMASK(chan->scan_type.realbits - 1, 0);
return IIO_VAL_INT;
}
---
base-commit: b72f1157bfb9b92b0439e11469f7f94e47363460
change-id: 20250505-iio-adc-ad7944-max-high-bits-on-direct-read-b358289d0337
Best regards,
--
David Lechner <dlechner@baylibre.com>
Just realized the subject is supposed to say "mask", not "max".
On Thu, 8 May 2025 16:03:15 -0500 David Lechner <dlechner@baylibre.com> wrote: > Just realized the subject is supposed to say "mask", not "max". Applied, marked for stable and wording tweaked. Thanks, J
On Mon, 2025-05-05 at 13:28 -0500, David Lechner wrote:
> Apply a mask to the raw value received over the SPI bus for unsigned
> direct reads. As we found recently, SPI controllers may not set unused
> bits to 0 when reading with bits_per_word != {8,16,32}. The ad7944 uses
> bits_per_word of 14 and 18, so we need to mask the value to be sure we
> returning the correct value to userspace during a direct read.
>
> Fixes: d1efcf8871db ("iio: adc: ad7944: add driver for AD7944/AD7985/AD7986")
> Signed-off-by: David Lechner <dlechner@baylibre.com>
> ---
> The sign_extend32() already takes care of signed reads in case that
> isn't obvious - it overwrites the unknown bits with the sign bit.
> ---
Reviewed-by: Nuno Sá <nuno.sa@analog.com>
> drivers/iio/adc/ad7944.c | 2 ++
> 1 file changed, 2 insertions(+)
>
> diff --git a/drivers/iio/adc/ad7944.c b/drivers/iio/adc/ad7944.c
> index
> 2f949fe5587318957f2e423029294ced0a6f803d..37a137bd83571b055e970a8cd483d8726972d637
> 100644
> --- a/drivers/iio/adc/ad7944.c
> +++ b/drivers/iio/adc/ad7944.c
> @@ -377,6 +377,8 @@ static int ad7944_single_conversion(struct ad7944_adc *adc,
>
> if (chan->scan_type.sign == 's')
> *val = sign_extend32(*val, chan->scan_type.realbits - 1);
> + else
> + *val &= GENMASK(chan->scan_type.realbits - 1, 0);
>
> return IIO_VAL_INT;
> }
>
> ---
> base-commit: b72f1157bfb9b92b0439e11469f7f94e47363460
> change-id: 20250505-iio-adc-ad7944-max-high-bits-on-direct-read-b358289d0337
>
> Best regards,
© 2016 - 2025 Red Hat, Inc.