[PATCH] pinctrl: sunxi: always look for apb block

[PATCH] pinctrl: sunxi: always look for apb block

[Andre Przywara]

The Allwinner pinctrl device tree binding suggests that a clock named
"apb" would drive the pin controller IP. However (for legacy reasons) we
rely on this clock actually being the first clock defined.
Since named clocks can be in any order, let's explicitly check for a
clock called "apb" first, then revert to using the first (and only!)
clock to keep compatibility with older DTs.

Signed-off-by: Andre Przywara <andre.przywara@arm.com>
---
 drivers/pinctrl/sunxi/pinctrl-sunxi.c | 17 ++++++++++++++---
 1 file changed, 14 insertions(+), 3 deletions(-)

diff --git a/drivers/pinctrl/sunxi/pinctrl-sunxi.c b/drivers/pinctrl/sunxi/pinctrl-sunxi.c
index 4b6cb25bc796..897ba13e6a03 100644
--- a/drivers/pinctrl/sunxi/pinctrl-sunxi.c
+++ b/drivers/pinctrl/sunxi/pinctrl-sunxi.c
@@ -1360,10 +1360,21 @@ int sunxi_pinctrl_init_with_variant(struct platform_device *pdev,
 			goto gpiochip_error;
 	}
 
-	clk = devm_clk_get(&pdev->dev, NULL);
+	clk = devm_clk_get(&pdev->dev, "apb");
 	if (IS_ERR(clk)) {
-		ret = PTR_ERR(clk);
-		goto gpiochip_error;
+		/*
+		 * If no "apb" clock can be found, try the first one, which
+		 * must be the only one defined.
+		 */
+		if (PTR_ERR(clk) == -ENOENT &&
+		    of_count_phandle_with_args(node, "clocks",
+					       "#clock-cells") == 1)
+			clk = devm_clk_get(&pdev->dev, NULL);
+
+		if (IS_ERR(clk)) {
+			ret = PTR_ERR(clk);
+			goto gpiochip_error;
+		}
 	}
 
 	ret = clk_prepare_enable(clk);
-- 
2.14.1

[PATCH] pinctrl: sunxi: always look for apb block

[Maxime Ripard]

Hi,

On Thu, Mar 01, 2018 at 11:37:01AM +0000, Andre Przywara wrote:
> The Allwinner pinctrl device tree binding suggests that a clock named
> "apb" would drive the pin controller IP. However (for legacy reasons) we
> rely on this clock actually being the first clock defined.
> Since named clocks can be in any order, let's explicitly check for a
> clock called "apb" first, then revert to using the first (and only!)
> clock to keep compatibility with older DTs.
> 
> Signed-off-by: Andre Przywara <andre.przywara@arm.com>
> ---
>  drivers/pinctrl/sunxi/pinctrl-sunxi.c | 17 ++++++++++++++---
>  1 file changed, 14 insertions(+), 3 deletions(-)
> 
> diff --git a/drivers/pinctrl/sunxi/pinctrl-sunxi.c b/drivers/pinctrl/sunxi/pinctrl-sunxi.c
> index 4b6cb25bc796..897ba13e6a03 100644
> --- a/drivers/pinctrl/sunxi/pinctrl-sunxi.c
> +++ b/drivers/pinctrl/sunxi/pinctrl-sunxi.c
> @@ -1360,10 +1360,21 @@ int sunxi_pinctrl_init_with_variant(struct platform_device *pdev,
>  			goto gpiochip_error;
>  	}
>  
> -	clk = devm_clk_get(&pdev->dev, NULL);
> +	clk = devm_clk_get(&pdev->dev, "apb");
>  	if (IS_ERR(clk)) {
> -		ret = PTR_ERR(clk);
> -		goto gpiochip_error;
> +		/*
> +		 * If no "apb" clock can be found, try the first one, which
> +		 * must be the only one defined.
> +		 */
> +		if (PTR_ERR(clk) == -ENOENT &&
> +		    of_count_phandle_with_args(node, "clocks",
> +					       "#clock-cells") == 1)
> +			clk = devm_clk_get(&pdev->dev, NULL);
> +
> +		if (IS_ERR(clk)) {
> +			ret = PTR_ERR(clk);
> +			goto gpiochip_error;
> +		}

I guess this can be made slightly less convoluted using something like

ret = of_count_phandle_with_args(node, "clocks", "#clock-cells");
clk = devm_clk_get(&pdev->dev, ret == 1 ? NULL : "apb");

Maxime

-- 
Maxime Ripard, Bootlin (formerly Free Electrons)
Embedded Linux and Kernel engineering
https://bootlin.com
-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 833 bytes
Desc: not available
URL: <http://lists.infradead.org/pipermail/linux-arm-kernel/attachments/20180301/bd3a1390/attachment.sig>

[PATCH] pinctrl: sunxi: always look for apb block

[Andre Przywara]

Hi Maxime,

On 01/03/18 11:48, Maxime Ripard wrote:
> Hi,
> 
> On Thu, Mar 01, 2018 at 11:37:01AM +0000, Andre Przywara wrote:
>> The Allwinner pinctrl device tree binding suggests that a clock named
>> "apb" would drive the pin controller IP. However (for legacy reasons) we
>> rely on this clock actually being the first clock defined.
>> Since named clocks can be in any order, let's explicitly check for a
>> clock called "apb" first, then revert to using the first (and only!)
>> clock to keep compatibility with older DTs.
>>
>> Signed-off-by: Andre Przywara <andre.przywara@arm.com>
>> ---
>>  drivers/pinctrl/sunxi/pinctrl-sunxi.c | 17 ++++++++++++++---
>>  1 file changed, 14 insertions(+), 3 deletions(-)
>>
>> diff --git a/drivers/pinctrl/sunxi/pinctrl-sunxi.c b/drivers/pinctrl/sunxi/pinctrl-sunxi.c
>> index 4b6cb25bc796..897ba13e6a03 100644
>> --- a/drivers/pinctrl/sunxi/pinctrl-sunxi.c
>> +++ b/drivers/pinctrl/sunxi/pinctrl-sunxi.c
>> @@ -1360,10 +1360,21 @@ int sunxi_pinctrl_init_with_variant(struct platform_device *pdev,
>>  			goto gpiochip_error;
>>  	}
>>  
>> -	clk = devm_clk_get(&pdev->dev, NULL);
>> +	clk = devm_clk_get(&pdev->dev, "apb");
>>  	if (IS_ERR(clk)) {
>> -		ret = PTR_ERR(clk);
>> -		goto gpiochip_error;
>> +		/*
>> +		 * If no "apb" clock can be found, try the first one, which
>> +		 * must be the only one defined.
>> +		 */
>> +		if (PTR_ERR(clk) == -ENOENT &&
>> +		    of_count_phandle_with_args(node, "clocks",
>> +					       "#clock-cells") == 1)
>> +			clk = devm_clk_get(&pdev->dev, NULL);
>> +
>> +		if (IS_ERR(clk)) {
>> +			ret = PTR_ERR(clk);
>> +			goto gpiochip_error;
>> +		}
> 
> I guess this can be made slightly less convoluted using something like

                           ^^^^^^^^ that's a slight understatement

> 
> ret = of_count_phandle_with_args(node, "clocks", "#clock-cells");
> clk = devm_clk_get(&pdev->dev, ret == 1 ? NULL : "apb");

Pah, that's just embarrassing.
I guess I will need to back your Kickstarter now to buy me out of this.

Will send v2 after testing tonight.

Thanks a lot for having a look!

Cheers,
Andre.