On Wed, 2019-10-30 at 13:55 +0200, Alexandru Ardelean wrote:
> The inline read functions in the ADIS library don't check the return
> value
> of the `adis_read_reg()` function and assign the value of `tmp`
> regardless.
>
> Fix this by checking if return value is zero and only then assigning the
> value of `tmp`.
>
@Jonathan:
Please disregard this.
I will re-spin this as a series.
It turned out that the whole thing/changeset is much bigger than I
initially anticipated.
I think I found the reason why we (at ADI) weren't catching the ADIS
warnings/errors, before upstreaming.
It turns out, that we were only testing with ARM & ARM64 compilers, while
the compilers that were hitting us with warnings are x86_64. I do remember
[from past experiences] that different GCC compilers report different
[often valid] types of errors. But, if I don't need to work with too many
ARCHs, I do forget about this aspect.
In any case, I've updated our CI to also include a build-run with x86_64
for all ADI drivers.
And we may also extend this to cover the entire kernel [later].
Thanks
Alex
> Fixes: 57a1228a06b7a ("iio:imu:adis: Add support for 32bit registers")
> Signed-off-by: Alexandru Ardelean <alexandru.ardelean@xxxxxxxxxx>
> ---
> include/linux/iio/imu/adis.h | 6 ++++--
> 1 file changed, 4 insertions(+), 2 deletions(-)
>
> diff --git a/include/linux/iio/imu/adis.h b/include/linux/iio/imu/adis.h
> index 4c53815bb729..92aae14593bf 100644
> --- a/include/linux/iio/imu/adis.h
> +++ b/include/linux/iio/imu/adis.h
> @@ -129,7 +129,8 @@ static inline int adis_read_reg_16(struct adis *adis,
> unsigned int reg,
> int ret;
>
> ret = adis_read_reg(adis, reg, &tmp, 2);
> - *val = tmp;
> + if (ret == 0)
> + *val = tmp;
>
> return ret;
> }
> @@ -147,7 +148,8 @@ static inline int adis_read_reg_32(struct adis *adis,
> unsigned int reg,
> int ret;
>
> ret = adis_read_reg(adis, reg, &tmp, 4);
> - *val = tmp;
> + if (ret == 0)
> + *val = tmp;
>
> return ret;
> }
