Re: [PATCH -next 12/25] spi: dw-mmio: Use helper function devm_clk_get_*()

[Serge Semin <fancer.lancer@xxxxxxxxx>]

Hi Li

On Tue, Aug 22, 2023 at 09:12:24PM +0800, Li Zetao wrote:
> Since commit 7ef9651e9792 ("clk: Provide new devm_clk helpers for prepared
> and enabled clocks"), devm_clk_get() and clk_prepare_enable() can now be
> replaced by devm_clk_get_enabled() when driver enables (and possibly
> prepares) the clocks for the whole lifetime of the device. Moreover, it is
> no longer necessary to unprepare and disable the clocks explicitly. Also,
> devm_clk_get_optional() and clk_prepare_enable() can now be replaced by
> devm_clk_get_optional_enabled(). Moreover, the lable "out_clk" no longer
> makes sense, rename it to "out_reset".
> 
> Signed-off-by: Li Zetao <lizetao1@xxxxxxxxxx>

Several nitpicks below please since you are touching the code anyway.

> ---
>  drivers/spi/spi-dw-mmio.c | 20 +++++---------------
>  1 file changed, 5 insertions(+), 15 deletions(-)
> 
> diff --git a/drivers/spi/spi-dw-mmio.c b/drivers/spi/spi-dw-mmio.c
> index 805264c9c65c..6d9f64ecbfd6 100644
> --- a/drivers/spi/spi-dw-mmio.c
> +++ b/drivers/spi/spi-dw-mmio.c
> @@ -340,28 +340,22 @@ static int dw_spi_mmio_probe(struct platform_device *pdev)
>  	if (dws->irq < 0)
>  		return dws->irq; /* -ENXIO */
>  
> -	dwsmmio->clk = devm_clk_get(&pdev->dev, NULL);
> +	dwsmmio->clk = devm_clk_get_enabled(&pdev->dev, NULL);
>  	if (IS_ERR(dwsmmio->clk))
>  		return PTR_ERR(dwsmmio->clk);
> -	ret = clk_prepare_enable(dwsmmio->clk);
> -	if (ret)
> -		return ret;
>  
>  	/* Optional clock needed to access the registers */
> -	dwsmmio->pclk = devm_clk_get_optional(&pdev->dev, "pclk");
> +	dwsmmio->pclk = devm_clk_get_optional_enabled(&pdev->dev, "pclk");

>  	if (IS_ERR(dwsmmio->pclk)) {
>  		ret = PTR_ERR(dwsmmio->pclk);
> -		goto out_clk;
> +		goto out_reset;
>  	}

Just return from here please. There is no point in calling
reset_control_deassert() on error because the reset control handler
hasn't been requested yet. So the argument will be NULL and considered
as optional reset-control.

> -	ret = clk_prepare_enable(dwsmmio->pclk);
> -	if (ret)
> -		goto out_clk;
>  
>  	/* find an optional reset controller */
>  	dwsmmio->rstc = devm_reset_control_get_optional_exclusive(&pdev->dev, "spi");

>  	if (IS_ERR(dwsmmio->rstc)) {
>  		ret = PTR_ERR(dwsmmio->rstc);
> -		goto out_clk;
> +		goto out_reset;
>  	}

ditto


Please use the "out_reset" label in the next part of the code:
	init_func = device_get_match_data(&pdev->dev);
	if (init_func) {
		ret = init_func(pdev, dwsmmio);
		if (ret)
-			goto out;
+			goto out_reset;
	}

I know this changes the cleanup-on-error path logic. But since you are
touching exactly that code anyway it would be good to have it properly
fixed.

>  	reset_control_deassert(dwsmmio->rstc);
>  
> @@ -397,9 +391,7 @@ static int dw_spi_mmio_probe(struct platform_device *pdev)
>  
>  out:
>  	pm_runtime_disable(&pdev->dev);
> -	clk_disable_unprepare(dwsmmio->pclk);
> -out_clk:
> -	clk_disable_unprepare(dwsmmio->clk);
> +out_reset:
>  	reset_control_assert(dwsmmio->rstc);
>  
>  	return ret;
> @@ -411,8 +403,6 @@ static void dw_spi_mmio_remove(struct platform_device *pdev)
>  

>  	dw_spi_remove_host(&dwsmmio->dws);
>  	pm_runtime_disable(&pdev->dev);
> -	clk_disable_unprepare(dwsmmio->pclk);
> -	clk_disable_unprepare(dwsmmio->clk);
>  	reset_control_assert(dwsmmio->rstc);

The order here isn't correct too. But I guess we can live with it.
It isn't that important.

-Serge(y)

>  }
>  
> -- 
> 2.34.1
>