On Mon, 23 Oct 2023 11:29:30 +0800
Su Hui <suhui@xxxxxxxxxxxx> wrote:
> On 2023/10/23 09:33, Su Hui wrote:
> > On 2023/10/20 23:55, Jonathan Cameron wrote:
> >> I'm not sure why inv_mpu6050_sensor_show() doesn't return
> >> the actual error code from the regmap_bulk_read() and instead
> >> replaces it
> >> with -EINVAL. Given you are tidying up this related issues perhaps
> >> change
> >> that as well?
> >>
> >> static int inv_mpu6050_sensor_show(struct inv_mpu6050_state *st, int
> >> reg,
> >> int axis, int *val)
> >> {
> >> int ind, result;
> >> __be16 d;
> >>
> >> ind = (axis - IIO_MOD_X) * 2;
> >> result = regmap_bulk_read(st->map, reg + ind, &d, sizeof(d));
> >> if (result)
> >> return -EINVAL;
> >> //Make this return result;
> >
> > Sure, I will tidy up this, Thanks for your suggestion!
>
> I'm not sure whether the caller could handler this when return
> 'result' rather than '-EINVAL'.
>
> This is not a big problem, maybe we shouldn't modify this code.
If the ultimate caller (userspace in this case) doesn't handle other
error codes it won't work with lots of other devices and is broken.
So I doubt any code is that fragile. If we get a report of a regression
we can look into whether it's sensible to resolve it.
So fine to change this, but could be a separate patch.
J
>
> Su Hui
>
