clk: at91: pmc: Use common error handling code in pmc_register_ops()

[PATCH] clk: at91: pmc: Use common error handling code in pmc_register_ops()

Posted by Markus Elfring 1 year, 4 months ago

From: Markus Elfring <elfring@users.sourceforge.net>
Date: Tue, 17 Sep 2024 14:28:22 +0200

Add a jump target so that a bit of exception handling can be better reused
at the end of this function implementation.

This issue was detected by using the Coccinelle software.

Signed-off-by: Markus Elfring <elfring@users.sourceforge.net>
---
 drivers/clk/at91/pmc.c | 18 ++++++++++--------
 1 file changed, 10 insertions(+), 8 deletions(-)

diff --git a/drivers/clk/at91/pmc.c b/drivers/clk/at91/pmc.c
index 5aa9c1f1c886..040b70e1ffbc 100644
--- a/drivers/clk/at91/pmc.c
+++ b/drivers/clk/at91/pmc.c
@@ -162,20 +162,18 @@ static int __init pmc_register_ops(void)
 	if (!np)
 		return -ENODEV;

-	if (!of_device_is_available(np)) {
-		of_node_put(np);
-		return -ENODEV;
-	}
+	if (!of_device_is_available(np))
+		goto put_node;
+
 	of_node_put(np);

 	np = of_find_compatible_node(NULL, NULL, "atmel,sama5d2-securam");
 	if (!np)
 		return -ENODEV;

-	if (!of_device_is_available(np)) {
-		of_node_put(np);
-		return -ENODEV;
-	}
+	if (!of_device_is_available(np))
+		goto put_node;
+
 	of_node_put(np);

 	at91_pmc_backup_suspend = of_iomap(np, 0);
@@ -187,6 +185,10 @@ static int __init pmc_register_ops(void)
 	register_syscore_ops(&pmc_syscore_ops);

 	return 0;
+
+put_node:
+	of_node_put(np);
+	return -ENODEV;
 }
 /* This has to happen before arch_initcall because of the tcb_clksrc driver */
 postcore_initcall(pmc_register_ops);
--
2.46.0

Re: [PATCH] clk: at91: pmc: Use common error handling code in pmc_register_ops()

Posted by Nicolas Ferre 1 year, 4 months ago

On 17/09/2024 at 14:34, Markus Elfring wrote:
> From: Markus Elfring <elfring@users.sourceforge.net>
> Date: Tue, 17 Sep 2024 14:28:22 +0200
> 
> Add a jump target so that a bit of exception handling can be better reused
> at the end of this function implementation.
> 
> This issue was detected by using the Coccinelle software.
> 
> Signed-off-by: Markus Elfring <elfring@users.sourceforge.net>

Bad track record and no real benefit from the patch:
NACK, sorry.

Regards,
   Nicolas

> ---
>   drivers/clk/at91/pmc.c | 18 ++++++++++--------
>   1 file changed, 10 insertions(+), 8 deletions(-)
> 
> diff --git a/drivers/clk/at91/pmc.c b/drivers/clk/at91/pmc.c
> index 5aa9c1f1c886..040b70e1ffbc 100644
> --- a/drivers/clk/at91/pmc.c
> +++ b/drivers/clk/at91/pmc.c
> @@ -162,20 +162,18 @@ static int __init pmc_register_ops(void)
>          if (!np)
>                  return -ENODEV;
> 
> -       if (!of_device_is_available(np)) {
> -               of_node_put(np);
> -               return -ENODEV;
> -       }
> +       if (!of_device_is_available(np))
> +               goto put_node;
> +
>          of_node_put(np);
> 
>          np = of_find_compatible_node(NULL, NULL, "atmel,sama5d2-securam");
>          if (!np)
>                  return -ENODEV;
> 
> -       if (!of_device_is_available(np)) {
> -               of_node_put(np);
> -               return -ENODEV;
> -       }
> +       if (!of_device_is_available(np))
> +               goto put_node;
> +
>          of_node_put(np);
> 
>          at91_pmc_backup_suspend = of_iomap(np, 0);
> @@ -187,6 +185,10 @@ static int __init pmc_register_ops(void)
>          register_syscore_ops(&pmc_syscore_ops);
> 
>          return 0;
> +
> +put_node:
> +       of_node_put(np);
> +       return -ENODEV;
>   }
>   /* This has to happen before arch_initcall because of the tcb_clksrc driver */
>   postcore_initcall(pmc_register_ops);
> --
> 2.46.0
>

Re: [PATCH] clk: at91: pmc: Use common error handling code in pmc_register_ops()

Posted by Markus Elfring 1 year, 4 months ago

>> Add a jump target so that a bit of exception handling can be better reused
>> at the end of this function implementation.
…
> Bad track record and no real benefit from the patch:
> NACK, sorry.
…
>> +++ b/drivers/clk/at91/pmc.c
>> @@ -162,20 +162,18 @@ static int __init pmc_register_ops(void)
>>          if (!np)
>>                  return -ENODEV;
>>
>> -       if (!of_device_is_available(np)) {
>> -               of_node_put(np);
>> -               return -ENODEV;
>> -       }
>> +       if (!of_device_is_available(np))
>> +               goto put_node;
>> +
>>          of_node_put(np);
>>
>>          np = of_find_compatible_node(NULL, NULL, "atmel,sama5d2-securam");
>>          if (!np)
>>                  return -ENODEV;
>>
>> -       if (!of_device_is_available(np)) {
>> -               of_node_put(np);
>> -               return -ENODEV;
>> -       }
>> +       if (!of_device_is_available(np))
>> +               goto put_node;
>> +
>>          of_node_put(np);
>>
>>          at91_pmc_backup_suspend = of_iomap(np, 0);
>> @@ -187,6 +185,10 @@ static int __init pmc_register_ops(void)
>>          register_syscore_ops(&pmc_syscore_ops);
>>
>>          return 0;
>> +
>> +put_node:
>> +       of_node_put(np);
>> +       return -ENODEV;
>>   }
…

* Do you really disagree to advice from related information sources?
  https://wiki.sei.cmu.edu/confluence/display/c/MEM12-C.+Consider+using+a+goto+chain+when+leaving+a+function+on+error+when+using+and+releasing+resources?

* How do you think about to increase the application of scope-based resource management?


Regards,
Markus