diff mbox series

[RESEND,2/2] thermal: sprd: add missing of_node_put for loop iteration

Message ID 20210528115036.18222-2-krzysztof.kozlowski@canonical.com
State New
Headers show
Series None | expand

Commit Message

Krzysztof Kozlowski May 28, 2021, 11:50 a.m. UTC
Early exits from for_each_available_child_of_node() should decrement the
node reference counter.  Reported by Coccinelle:

  drivers/thermal/sprd_thermal.c:387:1-23: WARNING:
    Function "for_each_child_of_node" should have of_node_put() before goto around lines 391.

Signed-off-by: Krzysztof Kozlowski <krzysztof.kozlowski@canonical.com>
Acked-by: Chunyan Zhang <zhang.lyra@gmail.com>
---
 drivers/thermal/sprd_thermal.c | 4 ++++
 1 file changed, 4 insertions(+)

Comments

Daniel Lezcano June 14, 2021, 5:28 p.m. UTC | #1
On 28/05/2021 13:50, Krzysztof Kozlowski wrote:
> Early exits from for_each_available_child_of_node() should decrement the

> node reference counter.  Reported by Coccinelle:

> 

>   drivers/thermal/sprd_thermal.c:387:1-23: WARNING:

>     Function "for_each_child_of_node" should have of_node_put() before goto around lines 391.

> 

> Signed-off-by: Krzysztof Kozlowski <krzysztof.kozlowski@canonical.com>

> Acked-by: Chunyan Zhang <zhang.lyra@gmail.com>

> ---

>  drivers/thermal/sprd_thermal.c | 4 ++++

>  1 file changed, 4 insertions(+)

> 

> diff --git a/drivers/thermal/sprd_thermal.c b/drivers/thermal/sprd_thermal.c

> index 3682edb2f466..2778971aaf03 100644

> --- a/drivers/thermal/sprd_thermal.c

> +++ b/drivers/thermal/sprd_thermal.c

> @@ -388,6 +388,7 @@ static int sprd_thm_probe(struct platform_device *pdev)

>  		sen = devm_kzalloc(&pdev->dev, sizeof(*sen), GFP_KERNEL);

>  		if (!sen) {

>  			ret = -ENOMEM;

> +			of_node_put(sen_child);


Why not a new label for the rollback routine?

eg. goto out_of_node_put;

Or a new function devm_for_each_available_child_of_node() ?

>  			goto disable_clk;

>  		}

>  

> @@ -397,12 +398,14 @@ static int sprd_thm_probe(struct platform_device *pdev)

>  		ret = of_property_read_u32(sen_child, "reg", &sen->id);

>  		if (ret) {

>  			dev_err(&pdev->dev, "get sensor reg failed");

> +			of_node_put(sen_child);

>  			goto disable_clk;

>  		}

>  

>  		ret = sprd_thm_sensor_calibration(sen_child, thm, sen);

>  		if (ret) {

>  			dev_err(&pdev->dev, "efuse cal analysis failed");

> +			of_node_put(sen_child);

>  			goto disable_clk;

>  		}

>  

> @@ -416,6 +419,7 @@ static int sprd_thm_probe(struct platform_device *pdev)

>  			dev_err(&pdev->dev, "register thermal zone failed %d\n",

>  				sen->id);

>  			ret = PTR_ERR(sen->tzd);

> +			of_node_put(sen_child);

>  			goto disable_clk;

>  		}

>  

> 



-- 
<http://www.linaro.org/> Linaro.org │ Open source software for ARM SoCs

Follow Linaro:  <http://www.facebook.com/pages/Linaro> Facebook |
<http://twitter.com/#!/linaroorg> Twitter |
<http://www.linaro.org/linaro-blog/> Blog
Krzysztof Kozlowski June 14, 2021, 7:16 p.m. UTC | #2
On 14/06/2021 19:28, Daniel Lezcano wrote:
> On 28/05/2021 13:50, Krzysztof Kozlowski wrote:

>> Early exits from for_each_available_child_of_node() should decrement the

>> node reference counter.  Reported by Coccinelle:

>>

>>   drivers/thermal/sprd_thermal.c:387:1-23: WARNING:

>>     Function "for_each_child_of_node" should have of_node_put() before goto around lines 391.

>>

>> Signed-off-by: Krzysztof Kozlowski <krzysztof.kozlowski@canonical.com>

>> Acked-by: Chunyan Zhang <zhang.lyra@gmail.com>

>> ---

>>  drivers/thermal/sprd_thermal.c | 4 ++++

>>  1 file changed, 4 insertions(+)

>>

>> diff --git a/drivers/thermal/sprd_thermal.c b/drivers/thermal/sprd_thermal.c

>> index 3682edb2f466..2778971aaf03 100644

>> --- a/drivers/thermal/sprd_thermal.c

>> +++ b/drivers/thermal/sprd_thermal.c

>> @@ -388,6 +388,7 @@ static int sprd_thm_probe(struct platform_device *pdev)

>>  		sen = devm_kzalloc(&pdev->dev, sizeof(*sen), GFP_KERNEL);

>>  		if (!sen) {

>>  			ret = -ENOMEM;

>> +			of_node_put(sen_child);

> 

> Why not a new label for the rollback routine?

> 

> eg. goto out_of_node_put;


Indeed, this should work. When loops end, the sen_child is set to NULL
so of_node_putting it should be fine.

> 

> Or a new function devm_for_each_available_child_of_node() ?


Since there are multiple flavors of this for_each_loop, that might be
too much...


Best regards,
Krzysztof
diff mbox series

Patch

diff --git a/drivers/thermal/sprd_thermal.c b/drivers/thermal/sprd_thermal.c
index 3682edb2f466..2778971aaf03 100644
--- a/drivers/thermal/sprd_thermal.c
+++ b/drivers/thermal/sprd_thermal.c
@@ -388,6 +388,7 @@  static int sprd_thm_probe(struct platform_device *pdev)
 		sen = devm_kzalloc(&pdev->dev, sizeof(*sen), GFP_KERNEL);
 		if (!sen) {
 			ret = -ENOMEM;
+			of_node_put(sen_child);
 			goto disable_clk;
 		}
 
@@ -397,12 +398,14 @@  static int sprd_thm_probe(struct platform_device *pdev)
 		ret = of_property_read_u32(sen_child, "reg", &sen->id);
 		if (ret) {
 			dev_err(&pdev->dev, "get sensor reg failed");
+			of_node_put(sen_child);
 			goto disable_clk;
 		}
 
 		ret = sprd_thm_sensor_calibration(sen_child, thm, sen);
 		if (ret) {
 			dev_err(&pdev->dev, "efuse cal analysis failed");
+			of_node_put(sen_child);
 			goto disable_clk;
 		}
 
@@ -416,6 +419,7 @@  static int sprd_thm_probe(struct platform_device *pdev)
 			dev_err(&pdev->dev, "register thermal zone failed %d\n",
 				sen->id);
 			ret = PTR_ERR(sen->tzd);
+			of_node_put(sen_child);
 			goto disable_clk;
 		}