On Mon, 2024-11-11 at 12:53 +0100, Uwe Kleine-König wrote:
> Hello Nuno,
>
> On Mon, Nov 11, 2024 at 11:37:46AM +0100, Nuno Sá wrote:
> > On Fri, 2024-11-08 at 19:18 +0100, Uwe Kleine-König wrote:
> > > @@ -821,6 +822,11 @@ static int ad7124_parse_channel_config(struct iio_dev
> > > *indio_dev,
> > > if (!st->num_channels)
> > > return dev_err_probe(dev, -ENODEV, "no channel
> > > children\n");
> > >
> > > + if (st->num_channels > AD7124_MAX_CHANNELS) {
> > > + dev_warn(dev, "Limit number of channels to "
> > > __stringify(AD7124_MAX_CHANNELS) "\n");
> > > + st->num_channels = AD7124_MAX_CHANNELS;
> > > + }
> >
> > Hmmm, I would treat it as an error...
>
> Well, it probably results in an error further below when the first child
> is hit that uses a too high reg property. I considered erroring out
Assuming no one points two different nodes to the same reg :)
> here, but thought this might not be justified if some children are not
> logical channels. I'm not sure either way, but can rework accordingly if
> all other concerns are resolved.
>
Anyways, I'm not totally sure about the logical channels you and David are
discussing (did not went through it), but FWIW, I agree that we should stop
probe if something is not supported rather than proceed just to get non obvious,
subtle bugs.
- Nuno Sá
