On Mon, Aug 26, 2024 at 9:45 AM Guoqing Jiang
<guoqing.jiang@xxxxxxxxxxxxx> wrote:
>
> Add mtk_rng_remove function which calles pm_runtime relevant funcs
> and unregister hwrng to paired with mtk_rng_probe.
>
> And without remove function, pm_runtime complains below when reload
> the driver.
>
> mtk_rng 1020f000.rng: Unbalanced pm_runtime_enable!
>
> Signed-off-by: Guoqing Jiang <guoqing.jiang@xxxxxxxxxxxxx>
> ---
> drivers/char/hw_random/mtk-rng.c | 10 ++++++++++
> 1 file changed, 10 insertions(+)
>
> diff --git a/drivers/char/hw_random/mtk-rng.c b/drivers/char/hw_random/mtk-rng.c
> index 302e201b51c2..1b6aa9406b11 100644
> --- a/drivers/char/hw_random/mtk-rng.c
> +++ b/drivers/char/hw_random/mtk-rng.c
> @@ -149,6 +149,15 @@ static int mtk_rng_probe(struct platform_device *pdev)
> return 0;
> }
>
> +static void mtk_rng_remove(struct platform_device *pdev)
> +{
> + struct mtk_rng *priv = platform_get_drvdata(pdev);
> +
> + pm_runtime_disable(&pdev->dev);
Instead maybe just replace pm_runtime_enable() with devm_pm_runtime_enable()
in the probe function?
> + pm_runtime_set_suspended(&pdev->dev);
Not sure if this is needed? I'm not super familiar with runtime PM.
> + devm_hwrng_unregister(&pdev->dev, &priv->rng);
The fact that it is already devm_* means that you shouldn't need to
call it.
ChenYu
> +}
> +
> #ifdef CONFIG_PM
> static int mtk_rng_runtime_suspend(struct device *dev)
> {
> @@ -186,6 +195,7 @@ MODULE_DEVICE_TABLE(of, mtk_rng_match);
>
> static struct platform_driver mtk_rng_driver = {
> .probe = mtk_rng_probe,
> + .remove_new = mtk_rng_remove,
> .driver = {
> .name = MTK_RNG_DEV,
> .pm = MTK_RNG_PM_OPS,
> --
> 2.34.1
>
>
