On Mi, 2023-03-15 at 18:44 +0800, Minda Chen wrote:
> There is a Cadence USB3 core for JH7110 SoCs, the cdns
> core is the child of this USB wrapper module device.
>
> Signed-off-by: Minda Chen <minda.chen@xxxxxxxxxxxxxxxx>
> ---
[...]
> diff --git a/drivers/usb/cdns3/cdns3-starfive.c b/drivers/usb/cdns3/cdns3-starfive.c
> new file mode 100644
> index 000000000000..a99f98f85235
> --- /dev/null
> +++ b/drivers/usb/cdns3/cdns3-starfive.c
> @@ -0,0 +1,305 @@
[...]
> +static int cdns_clk_rst_init(struct cdns_starfive *data)
> +{
> + int ret;
> +
> + data->num_clks = devm_clk_bulk_get_all(data->dev, &data->clks);
> + if (data->num_clks < 0)
> + return dev_err_probe(data->dev, -ENODEV,
> + "Failed to get clocks\n");
> +
> + ret = clk_bulk_prepare_enable(data->num_clks, data->clks);
> + if (ret)
> + return dev_err_probe(data->dev, ret,
> + "failed to enable clocks\n");
In general, it's better to acquire all resources first and only then
start interacting with them, and to order all devm_ calls before non-
devm calls to make sure cleanup is done in reverse order.
In this case you can switch clk_bulk_prepare_enable() with
devm_reset_control_array_get_exclusive() and simplify the error path.
> + data->resets = devm_reset_control_array_get_exclusive(data->dev);
> + if (IS_ERR(data->resets)) {
> + ret = dev_err_probe(data->dev, PTR_ERR(data->resets),
> + "Failed to get resets");
> + goto err_clk_init;
> + }
> +
> + ret = reset_control_deassert(data->resets);
> + if (ret) {
> + ret = dev_err_probe(data->dev, ret,
> + "failed to reset clocks\n");
> + goto err_clk_init;
> + }
> +
> + return ret;
> +
> +err_clk_init:
> + clk_bulk_disable_unprepare(data->num_clks, data->clks);
> + return ret;
> +}
regards
Philipp
