Re: [PATCH v3 3/4] mmc: sdhci-xenon: use clk only with DT

[Ulf Hansson]

On Wed, 2 Dec 2020 at 19:51, Marcin Wojtas <mw@xxxxxxxxxxxx> wrote:
>
> As a preparation for supporting ACPI, modify the driver
> to use the clk framework only when booting with DT -
> otherwise rely on the configuration done by firmware.
> For that purpose introduce also a custom SDHCI get_max_clock
> callback.
>
> Signed-off-by: Marcin Wojtas <mw@xxxxxxxxxxxx>
> Acked-by: Adrian Hunter <adrian.hunter@xxxxxxxxx>
> ---
>  drivers/mmc/host/sdhci-xenon.c | 61 ++++++++++++--------
>  1 file changed, 38 insertions(+), 23 deletions(-)
>
> diff --git a/drivers/mmc/host/sdhci-xenon.c b/drivers/mmc/host/sdhci-xenon.c

[...]

> @@ -637,10 +650,12 @@ static int xenon_runtime_resume(struct device *dev)
>         struct xenon_priv *priv = sdhci_pltfm_priv(pltfm_host);
>         int ret;
>
> -       ret = clk_prepare_enable(pltfm_host->clk);
> -       if (ret) {
> -               dev_err(dev, "can't enable mainck\n");
> -               return ret;
> +       if (dev->of_node) {

I didn't notice this in the earlier version, my apologies, but there
is no need for this check.

clk_prepare_enable() should cope fine with a NULL argument - and you
only reach this path, if the clock was successfully fetched during the
probe or that it was left to stay NULL for non-DT case.

> +               ret = clk_prepare_enable(pltfm_host->clk);
> +               if (ret) {
> +                       dev_err(dev, "can't enable mainck\n");
> +                       return ret;
> +               }
>         }
>
>         if (priv->restore_needed) {
> --
> 2.29.0
>

Kind regards
Uffe