[PATCH v1 1/1] cpufreq: loongson: Check for error code from devm_mutex_init() call

[PATCH v1 1/1] cpufreq: loongson: Check for error code from devm_mutex_init() call

[Andy Shevchenko]

Even if it's not critical, the avoidance of checking the error code
from devm_mutex_init() call today diminishes the point of using devm
variant of it. Tomorrow it may even leak something. Add the missed
check.

Fixes: ccf51454145b ("cpufreq: Add Loongson-3 CPUFreq driver support")
Signed-off-by: Andy Shevchenko <andriy.shevchenko@linux.intel.com>
---
 drivers/cpufreq/loongson3_cpufreq.c | 7 +++++--
 1 file changed, 5 insertions(+), 2 deletions(-)

diff --git a/drivers/cpufreq/loongson3_cpufreq.c b/drivers/cpufreq/loongson3_cpufreq.c
index 61ebebf69455..bd34bf0fafa5 100644
--- a/drivers/cpufreq/loongson3_cpufreq.c
+++ b/drivers/cpufreq/loongson3_cpufreq.c
@@ -346,8 +346,11 @@ static int loongson3_cpufreq_probe(struct platform_device *pdev)
 {
 	int i, ret;
 
-	for (i = 0; i < MAX_PACKAGES; i++)
-		devm_mutex_init(&pdev->dev, &cpufreq_mutex[i]);
+	for (i = 0; i < MAX_PACKAGES; i++) {
+		ret = devm_mutex_init(&pdev->dev, &cpufreq_mutex[i]);
+		if (ret)
+			return ret;
+	}
 
 	ret = do_service_request(0, 0, CMD_GET_VERSION, 0, 0);
 	if (ret <= 0)
-- 
2.43.0.rc1.1336.g36b5255a03ac

Re: [PATCH v1 1/1] cpufreq: loongson: Check for error code from devm_mutex_init() call

[Huacai Chen]

Hi, Andy,

On Thu, Oct 31, 2024 at 12:29 AM Andy Shevchenko
<andriy.shevchenko@linux.intel.com> wrote:
>
> Even if it's not critical, the avoidance of checking the error code
> from devm_mutex_init() call today diminishes the point of using devm
> variant of it. Tomorrow it may even leak something. Add the missed
> check.
>
> Fixes: ccf51454145b ("cpufreq: Add Loongson-3 CPUFreq driver support")
> Signed-off-by: Andy Shevchenko <andriy.shevchenko@linux.intel.com>
> ---
>  drivers/cpufreq/loongson3_cpufreq.c | 7 +++++--
>  1 file changed, 5 insertions(+), 2 deletions(-)
>
> diff --git a/drivers/cpufreq/loongson3_cpufreq.c b/drivers/cpufreq/loongson3_cpufreq.c
> index 61ebebf69455..bd34bf0fafa5 100644
> --- a/drivers/cpufreq/loongson3_cpufreq.c
> +++ b/drivers/cpufreq/loongson3_cpufreq.c
> @@ -346,8 +346,11 @@ static int loongson3_cpufreq_probe(struct platform_device *pdev)
>  {
>         int i, ret;
>
> -       for (i = 0; i < MAX_PACKAGES; i++)
> -               devm_mutex_init(&pdev->dev, &cpufreq_mutex[i]);
> +       for (i = 0; i < MAX_PACKAGES; i++) {
> +               ret = devm_mutex_init(&pdev->dev, &cpufreq_mutex[i]);
> +               if (ret)
Good catch, but I think "if (ret < 0)" is better? Sometimes a positive
return value is legal, even if not in this case.

And it is better to use loongson3 rather than loongson because there
is another loongson2 driver.

Huacai

> +                       return ret;
> +       }
>
>         ret = do_service_request(0, 0, CMD_GET_VERSION, 0, 0);
>         if (ret <= 0)
> --
> 2.43.0.rc1.1336.g36b5255a03ac
>
>

Re: [PATCH v1 1/1] cpufreq: loongson: Check for error code from devm_mutex_init() call

[Andy Shevchenko]

On Thu, Oct 31, 2024 at 09:29:52AM +0800, Huacai Chen wrote:
> On Thu, Oct 31, 2024 at 12:29 AM Andy Shevchenko
> <andriy.shevchenko@linux.intel.com> wrote:

...

> > -       for (i = 0; i < MAX_PACKAGES; i++)
> > -               devm_mutex_init(&pdev->dev, &cpufreq_mutex[i]);
> > +       for (i = 0; i < MAX_PACKAGES; i++) {
> > +               ret = devm_mutex_init(&pdev->dev, &cpufreq_mutex[i]);
> > +               if (ret)
> Good catch, but I think "if (ret < 0)" is better? Sometimes a positive
> return value is legal, even if not in this case.

I disagree on this.

During a tons of reviews I have done in the past this kind of check is
impediment and always rises the Q "why?" It means that the author hasn't
fully thought through the code and most likely done something is a cargo cult.
On top of that, if the callee is changed at some point to actually return
a positive code(s), the caller most likely has to be at least aware of that
change. The proposed modification makes this silently compile and hides
possible important details from the caller(s).

> And it is better to use loongson3 rather than loongson because there
>  is another loongson2 driver.

Thanks, I will change that in v2 (I believe you are talking about Subject?).

> > +                       return ret;
> > +       }

-- 
With Best Regards,
Andy Shevchenko