Re: [PATCH v1 1/2] clk: x86: lpss-atom: Use predefined constants from units.h

From: Stephen Boyd
Date: Wed Aug 28 2024 - 14:39:16 EST


Quoting Andy Shevchenko (2024-08-28 06:11:55)
> On Tue, Aug 27, 2024 at 05:21:00PM -0700, Stephen Boyd wrote:
> > Quoting Andy Shevchenko (2024-08-22 09:14:07)
>
> ...
>
> > > static int lpss_atom_clk_probe(struct platform_device *pdev)
> > > {
> > > struct lpss_clk_data *drvdata;
> > > struct clk *clk;
> > > + u32 rate;
> >
> > Do we need a local variable?
>
> Hmm... The idea was to allow retrieving this via device properties, that's why
> a separate variable, but that patch wasn't included here.
>
> Nevertheless, despite above the separate variable makes code a bit better to
> read as we can see what is this value about.
>
> > > drvdata = devm_kzalloc(&pdev->dev, sizeof(*drvdata), GFP_KERNEL);
> > > if (!drvdata)
> > > return -ENOMEM;
> > >
> > > + /* Default frequency is 100MHz */
> > > + rate = 100 * HZ_PER_MHZ;
> > > +
> > > /* LPSS free running clock */
> > > drvdata->name = "lpss_clk";
> > > - clk = clk_register_fixed_rate(&pdev->dev, drvdata->name, NULL,
> > > - 0, 100000000);
> > > + clk = clk_register_fixed_rate(&pdev->dev, drvdata->name, NULL, 0, rate);
> >
> > This should be a one line patch.
>
> I don't get this. You mean the entire thingy?

Yes.

>
> It's possible, but as I mentioned above there is a rationale for making it with
> a temporary variable.

The rationale looks like future code will want a local variable. When
that happens the local variable would make sense.