On Thu, Sep 23, 2021 at 08:47:52AM +0300, Dan Carpenter wrote:
> > > > if (ret != -ENODEV)
> > > > return ret;
> > > > /* unbuffered? */
> > > > ad7949_adc->vref = devm_regulator_get_optional(dev, "vref");
> > > > if (IS_ERR(ad7949_adc->vref)) {
> > > > ret = PTR_ERR(ad7949_adc->vref);
> > > > + ad7949_adc->vref = NULL;
> > >
> > > But this also won't work. Passing a NULL to regulator_enable() will
> > > cause an Oops. All the reference to ->vref need checks. :/
> >
> > I believe it still work since these conditions around
> > devm_regulator_get_optional() also set ad7949_adc->refsel.
> >
> > ad7949_adc->refsel is then checked before calling regulator_enable() and
> > regulator_get_voltage().
> >
> > Even without the patch, I don't think we can call regulor_enable()
> > without having it be defined. Am I missing something else?
Hi Dan,
> Actually, you're right. This warning is a 100% false positive. Smatch
> doesn't handle bit wise tests very well. I've been meaning to write
> that code but I haven't done it yet. When I do the false positive will
> go away.
>
> Sorry for the noise on this.
No worries, thanks for your support on this.
> regards,
> dan carpenter
Liam
