[PATCH 0/2] iio: light: adux1020: fix proximity measurement issues

[PATCH 0/2] iio: light: adux1020: fix proximity measurement issues

[Raffaele Tranfaglia via B4 Relay]

While testing the guard() conversion of this driver with i2c-stub, I
found two issues in adux1020_measure().

Patch 1 fixes a real bug: the interrupt clear value is computed as
~u16, which integer promotion turns into 0xffffff0f. Plain I2C
adapters silently truncate it to 0xff0f, but SMBus-only adapters reject
it with -EINVAL, so every proximity read fails on them.

Patch 2 extracts the FIFO byte count with FIELD_GET(). The current code
compares the unshifted field against the packet size. Per the
datasheet, the FIFO only stores complete packets, so this does not
misbehave on a working device, hence no Fixes tag.

Tested on arm64 QEMU with i2c-stub (an SMBus-only adapter), lockdep and
UBSAN enabled, planting FIFO byte counts in INT_STATUS by hand:
- before: reads fail with -EINVAL
- patch 1: count 2 returns the planted sample; count 1 also returns it
  (issue fixed by patch 2); count 0 times out with -EIO
- patch 1+2: count 2 returns the sample; counts 1 and 0 time out with
  -EIO
No lockdep or UBSAN reports.

I prepared these patches with the help of an AI coding assistant, and
reviewed and tested them myself.

---
Raffaele Tranfaglia (2):
      iio: light: adux1020: fix interrupt clear value on SMBus-only adapters
      iio: light: adux1020: extract the FIFO byte count with FIELD_GET()

 drivers/iio/light/adux1020.c | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)
---
base-commit: b70566610b48ba49dec636019da4d696db64b085
change-id: 20261109-adux1020-measure-fixes-9f3bb844f758

Best regards,
--  
Raffaele Tranfaglia <raffaele.tranfaglia@gmail.com>

[PATCH 1/2] iio: light: adux1020: fix interrupt clear value on SMBus-only adapters

[Raffaele Tranfaglia via B4 Relay]

From: Raffaele Tranfaglia <raffaele.tranfaglia@gmail.com>

adux1020_measure() clears the mode interrupt by writing the inverted
interrupt enable mask to INT_STATUS. The mask is a u16, but integer
promotion makes ~0x00f0 evaluate to 0xffffff0f, which does not fit in
the 16-bit register.

On adapters supporting plain I2C, regmap formats the value with
put_unaligned_be16() and silently truncates it to 0xff0f. On adapters
that only support SMBus word transfers, regmap_smbus_word_write_swapped()
rejects values above 0xffff with -EINVAL, so every proximity read fails.

Mask the value to 16 bits so that SMBus-only adapters send the same
0xff0f that has always been written on plain I2C adapters.

Fixes: 356543545576 ("iio: light: Add support for ADUX1020 sensor")
Signed-off-by: Raffaele Tranfaglia <raffaele.tranfaglia@gmail.com>
---
 drivers/iio/light/adux1020.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/drivers/iio/light/adux1020.c b/drivers/iio/light/adux1020.c
index d4281b9754c3..cccd509dd18a 100644
--- a/drivers/iio/light/adux1020.c
+++ b/drivers/iio/light/adux1020.c
@@ -346,7 +346,7 @@ static int adux1020_measure(struct adux1020_data *data,
 
 	/* Clear mode interrupt */
 	ret = regmap_write(data->regmap, ADUX1020_REG_INT_STATUS,
-			   (~adux1020_modes[mode].int_en));
+			   ~adux1020_modes[mode].int_en & 0xffff);
 	if (ret < 0)
 		return ret;
 

-- 
2.56.0

[PATCH 2/2] iio: light: adux1020: extract the FIFO byte count with FIELD_GET()

[Raffaele Tranfaglia via B4 Relay]

From: Raffaele Tranfaglia <raffaele.tranfaglia@gmail.com>

adux1020_measure() masks INT_STATUS with ADUX1020_FIFO_STATUS_MASK but
does not shift the result down, so it compares 256 times the FIFO byte
count against the packet size. A count of 1 would thus be treated as a
complete 2-byte proximity packet.

This does not happen on a correctly working device, because the FIFO
only stores complete packets and the driver always reads complete
packets, so the count is always a multiple of the packet size. Still,
use FIELD_GET() so the comparison matches the datasheet's polling
procedure and does not depend on that.

Signed-off-by: Raffaele Tranfaglia <raffaele.tranfaglia@gmail.com>
---
 drivers/iio/light/adux1020.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/drivers/iio/light/adux1020.c b/drivers/iio/light/adux1020.c
index cccd509dd18a..dd5230c6dceb 100644
--- a/drivers/iio/light/adux1020.c
+++ b/drivers/iio/light/adux1020.c
@@ -331,7 +331,7 @@ static int adux1020_measure(struct adux1020_data *data,
 		if (ret < 0)
 			return ret;
 
-		status &= ADUX1020_FIFO_STATUS_MASK;
+		status = FIELD_GET(ADUX1020_FIFO_STATUS_MASK, status);
 		if (status >= adux1020_modes[mode].bytes)
 			break;
 		msleep(20);

-- 
2.56.0

Re: [PATCH 1/2] iio: light: adux1020: fix interrupt clear value on SMBus-only adapters

[Andy Shevchenko]

On Mon, Nov 09, 2026 at 11:39:02AM +0000, Raffaele Tranfaglia via B4 Relay wrote:
> 
> adux1020_measure() clears the mode interrupt by writing the inverted
> interrupt enable mask to INT_STATUS. The mask is a u16, but integer
> promotion makes ~0x00f0 evaluate to 0xffffff0f, which does not fit in
> the 16-bit register.

Huh?! Where did you get it from? the mentioned promotion may only happen for
a signed char which may not hold 0x00f0 in the first place. The above is some
nonsense.

> On adapters supporting plain I2C, regmap formats the value with
> put_unaligned_be16() and silently truncates it to 0xff0f. On adapters
> that only support SMBus word transfers, regmap_smbus_word_write_swapped()
> rejects values above 0xffff with -EINVAL, so every proximity read fails.
> 
> Mask the value to 16 bits so that SMBus-only adapters send the same
> 0xff0f that has always been written on plain I2C adapters.

...

>  	ret = regmap_write(data->regmap, ADUX1020_REG_INT_STATUS,
> -			   (~adux1020_modes[mode].int_en));
> +			   ~adux1020_modes[mode].int_en & 0xffff);

On top of that regmap should limit the value to 16-bit anyway. If there is a
problem in regmap to cut the upper bits, it should be addressed rather there.

>  	if (ret < 0)
>  		return ret;

-- 
With Best Regards,
Andy Shevchenko