Re: [PATCH] iio: gyro: adxrs290: propagate chip ID read errors
From: Joshua Crofts
Date: Thu Nov 19 2026 - 05:16:00 EST
On 19-Nov-26 4:07, Siqi Chen wrote:
> spi_w8r8() returns either an unsigned 8-bit value or a negative error.
> The probe function stores its return value in a u8, truncating errors and
> treating them as unexpected chip IDs. Consequently, an SPI transfer failure
> is reported as -ENODEV instead of the actual error.
>
> Store each result in an int and return negative errors before validating
> the chip IDs.
>
> OpenCode was used to identify the issue and draft the code and changelog.
>
> Compile-tested with CONFIG_ADXRS290=m using:
>
> make O=/tmp/opencode/linux-build-adxrs290 W=1 \
> drivers/iio/gyro/adxrs290.o
>
> No hardware or SPI fault-injection testing was performed.
Remove the testing info from the commit message and put it under the ---.
Unless you tested the patch on specific hardware then no need to immortalize
it in the git log.
Side note: one would expect that patches sent to the mailing list are at
least compile tested :)
>
> Fixes: 2c8920fff145 ("iio: gyro: Add driver support for ADXRS290")
> Signed-off-by: Siqi Chen <chensiqi@xxxxxxxxxxxxx>
> ---
> drivers/iio/gyro/adxrs290.c | 26 ++++++++++++++++----------
> 1 file changed, 16 insertions(+), 10 deletions(-)
>
> diff --git a/drivers/iio/gyro/adxrs290.c b/drivers/iio/gyro/adxrs290.c
> index b61cc81122a6..eb3306215217 100644
> --- a/drivers/iio/gyro/adxrs290.c
> +++ b/drivers/iio/gyro/adxrs290.c
> @@ -604,7 +604,7 @@ static int adxrs290_probe(struct spi_device *spi)
> struct iio_dev *indio_dev;
> struct adxrs290_state *st;
> u8 val, val2;
> - int ret;
> + int id, ret;
Why a new variable? ret should be sufficient in this case.
>
> indio_dev = devm_iio_device_alloc(&spi->dev, sizeof(*st));
> if (!indio_dev)
> @@ -622,21 +622,27 @@ static int adxrs290_probe(struct spi_device *spi)
>
> mutex_init(&st->lock);
>
> - val = spi_w8r8(spi, ADXRS290_READ_REG(ADXRS290_REG_ADI_ID));
> - if (val != ADXRS290_ADI_ID) {
> - dev_err(&spi->dev, "Wrong ADI ID 0x%02x\n", val);
> + id = spi_w8r8(spi, ADXRS290_READ_REG(ADXRS290_REG_ADI_ID));
> + if (id < 0)
> + return dev_err_probe(&spi->dev, id, "failed to read ADI ID\n");
> + if (id != ADXRS290_ADI_ID) {
> + dev_err(&spi->dev, "Wrong ADI ID 0x%02x\n", id);
> return -ENODEV;
Not sure how this device works, but better to just call dev_info() to
ensure that any potential compatible fallback can be matched.
Also, why not just use dev_err_probe() if the previous if check uses
it? Style unification is a good idea.
--
Kind regards,
Joshua Crofts