On Mon, 19 Feb 2024 13:48:27 +0200
Andy Shevchenko <andriy.shevchenko@xxxxxxxxxxxxxxx> wrote:
> On Sat, Feb 17, 2024 at 04:42:49PM +0000, Jonathan Cameron wrote:
> > From: Jonathan Cameron <Jonathan.Cameron@xxxxxxxxxx>
> >
> > Switching to the _scoped() version removes the need for manual
> > calling of fwnode_handle_put() in the paths where the code
> > exits the loop early. In this case that's all in error paths.
>
> ...
>
> > span = ltc2688_span_lookup(st, (int)tmp[0] / 1000,
> > tmp[1] / 1000);
> > - if (span < 0) {
> > - fwnode_handle_put(child);
> > + if (span < 0)
> > return dev_err_probe(dev, -EINVAL,
> > "output range not valid:[%d %d]\n",
> > tmp[0], tmp[1]);
>
> Last minute observation, should not we return span instead of -EINVAL?
> (Haven't checked the semantics of the former though.)
It returns 0 or -EINVAL, so yes we should use span here.
>
> ...
>
> > + if (ret)
> > return dev_err_probe(dev, -EINVAL,
> > "failed to set chan settings\n");
>
> Ditto.
Definitely on that one.
I'll aim to fold those two in whilst picking this up with a note in the patch description.
(or I'll incorporate them if I do a v5 for other reasons!)
Thanks,
Jonathan
>
>
