Re: [PATCH v6 3/6] clk: qcom: smd: Add XO clock for MSM8998

[Stephen Boyd <sboyd@kernel.org>]

Quoting Jeffrey Hugo (2019-10-01 12:57:08)
> diff --git a/drivers/clk/qcom/gcc-msm8998.c b/drivers/clk/qcom/gcc-msm8998.c
> index 091acd59c1d6..1651a2f47ab8 100644
> --- a/drivers/clk/qcom/gcc-msm8998.c
> +++ b/drivers/clk/qcom/gcc-msm8998.c
> @@ -2971,14 +2957,23 @@ static const struct qcom_cc_desc gcc_msm8998_desc = {
>         .num_resets = ARRAY_SIZE(gcc_msm8998_resets),
>         .gdscs = gcc_msm8998_gdscs,
>         .num_gdscs = ARRAY_SIZE(gcc_msm8998_gdscs),
> -       .clk_hws = gcc_msm8998_hws,
> -       .num_clk_hws = ARRAY_SIZE(gcc_msm8998_hws),
>  };
>  
>  static int gcc_msm8998_probe(struct platform_device *pdev)
>  {
>         struct regmap *regmap;
>         int ret;
> +       struct clk *xo;
> +
> +       /*
> +        * We must have a valid XO to continue, otherwise having a missing
> +        * parent on a system critical clock like the uart core clock can
> +        * result in strange bugs.  We know XO will be provided by rpmcc,
> +        * but it might not be specified in DT like it should.
> +        */
> +       xo = __clk_lookup("xo");

I very much dislike __clk_lookup(). I think we can not have this patch?

> +       if (!xo)
> +               return -EPROBE_DEFER;
>