Re: [PATCH v5 3/5] clk: meson: a1: add support for Amlogic A1 PLL clock driver

[Jian Hu]

On 2019/12/28 1:04, Martin Blumenstingl wrote:
> Hi Jian,
>
> On Fri, Dec 27, 2019 at 10:46 AM Jian Hu <jian.hu@xxxxxxxxxxx> wrote:
> [...]
> > +               .parent_data = &(const struct clk_parent_data){
> > +                       .fw_name = "xtal_fixpll",
> > +               },
> in the Meson8b and G12A (I assume it's the same on GXBB, I didn't
> check it) we have a space between " clk_parent_data)" and "{"
> this applies to at least one more occurrence below
I have checked G12A and Meson8b, there is a space.The space is missing 
here, the same as other place. I will fix it in next version.
> [...]
> > +               /*
> > +                * This clock is used by APB bus which setted in Romcode
> nit-pick: I'm not sure about the grammar here: setted -> "is set"?
> and to make sure I understand this correctly: do you mean the "boot
> ROM" with "Romcode"?
You are right, it is a mistake here. 'is set' is right.
Yes, Romcode means boot ROM. I will change it to 'boot ROM code'
> [...]
> > +static int meson_a1_pll_probe(struct platform_device *pdev)
> > +{
> > +       const struct meson_eeclkc_data *data;
> what do you need this "data" variable for?
>
> > +       struct device *dev = &pdev->dev;
> > +       struct resource *res;
> > +       void __iomem *base;
> > +       struct regmap *map;
> > +       int ret, i;
> > +
> > +       data = of_device_get_match_data(dev);
> > +       if (!data)
> > +               return -EINVAL;
> > +
> > +       res = platform_get_resource(pdev, IORESOURCE_MEM, 0);
> > +
> > +       base = devm_ioremap_resource(dev, res);
> > +       if (IS_ERR(base))
> > +               return PTR_ERR(base);
> > +
> > +       map = devm_regmap_init_mmio(dev, base, &clkc_regmap_config);
> > +       if (IS_ERR(map))
> > +              return PTR_ERR(map);
> > +
> > +       /* Populate regmap for the regmap backed clocks */
> > +       for (i = 0; i < data->regmap_clk_num; i++)
> > +               data->regmap_clks[i]->map = map;
> why can't we use a1_pll_regmaps directly here?
OK, I will use it directly .
> > +
> > +       for (i = 0; i < data->hw_onecell_data->num; i++) {
> > +               /* array might be sparse */
> > +               if (!data->hw_onecell_data->hws[i])
> > +                       continue;
> > +
> > +               ret = devm_clk_hw_register(dev, data->hw_onecell_data->hws[i]);
> and why can't we use a1_pll_hw_onecell_data directly here?
OK, I will use it directly.
> [...]
> > +static const struct meson_eeclkc_data a1_pll_data = {
> > +               .regmap_clks = a1_pll_regmaps,
> > +               .regmap_clk_num = ARRAY_SIZE(a1_pll_regmaps),
> > +               .hw_onecell_data = &a1_pll_hw_onecell_data,
> > +};
> if _probe would access these directly then you can drop meson_eeclkc_data
> that is a good thing in my opinion because I was confused by the
> "eeclk" since the patch description says that there's no EE or AO
> domain on the A1 SoCs
OK, I will remove it and verify it.
> Martin
>
> .