Quoting Casey Connolly (2025-05-28 04:23:01)
> On 5/24/25 00:58, Stephen Boyd wrote:
> > @@ -561,6 +549,16 @@ static int msm_serial_probe(struct udevice *dev)
> > {
> > struct msm_serial_data *priv = dev_get_priv(dev);
> > int ret;
> > + struct clk *clk;
> > +
> > + clk = devm_clk_get(dev, NULL);
> > + if (IS_ERR(clk))
> > + return PTR_ERR(clk);
> > + priv->se = clk;
> > +
> > + ret = clk_enable(clk);
> > + if (ret)
> > + return ret;
>
> There's a subtle change in behaviour here, since we'll now fail probe if
> the clock isn't found. I think it would be best to treat this clock
> being missing as non-fatal since it can make bringup a little easier (no
> need to stub the clock).
I tried to return NULL from the sc7180 clk driver as a test and the
serial driver still probed. I think clk_enable() doesn't do anything in
this case, which is how it should be for other platforms? TL;DR is I
don't think this breaks anything.
