[PATCH] powercap: DTPM: Fix the missing cpufreq_cpu_put() calls

[PATCH] powercap: DTPM: Fix the missing cpufreq_cpu_put() calls

[Lukasz Luba]

The policy returned by cpufreq_cpu_get() has to be released with
the help of cpufreq_cpu_put() to balance its kobject reference counter
properly.

Add the missing calls to cpufreq_cpu_put() in the code.

Fixes: 0aea2e4ec2a2 ("powercap/dtpm_cpu: Reset per_cpu variable in the release function")
Fixes: 0e8f68d7f048 ("powercap/drivers/dtpm: Add CPU energy model based support")
Cc: <stable@vger.kernel.org> # v5.10+
Signed-off-by: Lukasz Luba <lukasz.luba@arm.com>
---
 drivers/powercap/dtpm_cpu.c | 17 +++++++++++++----
 1 file changed, 13 insertions(+), 4 deletions(-)

diff --git a/drivers/powercap/dtpm_cpu.c b/drivers/powercap/dtpm_cpu.c
index 45bb7e2849d7..aac278e162d9 100644
--- a/drivers/powercap/dtpm_cpu.c
+++ b/drivers/powercap/dtpm_cpu.c
@@ -152,6 +152,8 @@ static void pd_release(struct dtpm *dtpm)
 	if (policy) {
 		for_each_cpu(dtpm_cpu->cpu, policy->related_cpus)
 			per_cpu(dtpm_per_cpu, dtpm_cpu->cpu) = NULL;
+
+		cpufreq_cpu_put(policy);
 	}
 	
 	kfree(dtpm_cpu);
@@ -204,12 +206,16 @@ static int __dtpm_cpu_setup(int cpu, struct dtpm *parent)
 		return 0;
 
 	pd = em_cpu_get(cpu);
-	if (!pd || em_is_artificial(pd))
-		return -EINVAL;
+	if (!pd || em_is_artificial(pd)) {
+		ret = -EINVAL;
+		goto release_policy;
+	}
 
 	dtpm_cpu = kzalloc(sizeof(*dtpm_cpu), GFP_KERNEL);
-	if (!dtpm_cpu)
-		return -ENOMEM;
+	if (!dtpm_cpu) {
+		ret = -ENOMEM;
+		goto release_policy;
+	}
 
 	dtpm_init(&dtpm_cpu->dtpm, &dtpm_ops);
 	dtpm_cpu->cpu = cpu;
@@ -231,6 +237,7 @@ static int __dtpm_cpu_setup(int cpu, struct dtpm *parent)
 	if (ret)
 		goto out_dtpm_unregister;
 
+	cpufreq_cpu_put(policy);
 	return 0;
 
 out_dtpm_unregister:
@@ -242,6 +249,8 @@ static int __dtpm_cpu_setup(int cpu, struct dtpm *parent)
 		per_cpu(dtpm_per_cpu, cpu) = NULL;
 	kfree(dtpm_cpu);
 
+release_policy:
+	cpufreq_cpu_put(policy);
 	return ret;
 }
 
-- 
2.25.1

Re: [PATCH] powercap: DTPM: Fix the missing cpufreq_cpu_put() calls

[Greg KH]

On Fri, Dec 01, 2023 at 12:32:05PM +0000, Lukasz Luba wrote:
> The policy returned by cpufreq_cpu_get() has to be released with
> the help of cpufreq_cpu_put() to balance its kobject reference counter
> properly.
> 
> Add the missing calls to cpufreq_cpu_put() in the code.
> 
> Fixes: 0aea2e4ec2a2 ("powercap/dtpm_cpu: Reset per_cpu variable in the release function")
> Fixes: 0e8f68d7f048 ("powercap/drivers/dtpm: Add CPU energy model based support")
> Cc: <stable@vger.kernel.org> # v5.10+

But the Fixes: tags are for commits that are only in 5.12 and newer, how
can this be relevant for 5.10?

thanks,

greg k-h

Re: [PATCH] powercap: DTPM: Fix the missing cpufreq_cpu_put() calls

[Lukasz Luba]

Hi Greg,

On 12/1/23 12:44, Greg KH wrote:
> On Fri, Dec 01, 2023 at 12:32:05PM +0000, Lukasz Luba wrote:
>> The policy returned by cpufreq_cpu_get() has to be released with
>> the help of cpufreq_cpu_put() to balance its kobject reference counter
>> properly.
>>
>> Add the missing calls to cpufreq_cpu_put() in the code.
>>
>> Fixes: 0aea2e4ec2a2 ("powercap/dtpm_cpu: Reset per_cpu variable in the release function")
>> Fixes: 0e8f68d7f048 ("powercap/drivers/dtpm: Add CPU energy model based support")
>> Cc: <stable@vger.kernel.org> # v5.10+
> 
> But the Fixes: tags are for commits that are only in 5.12 and newer, how
> can this be relevant for 5.10?

My apologies, you're right. Somehow I checked that this dtpm_cpu.c
was introduced in v5.10. It was in v5.12 indeed. I messed that up.

Also, the code in that v5.12 had different implementation and there was
a function cpuhp_dtpm_cpu_offline() which had the cpufreq_cpu_get().

I can craft for that v5.12 special extra patch fix addressing it and
send directly to stable list. Would that make sense?

So this patch would only be applicable for v5.16+ AFAICS.

Regards,
Lukasz

Re: [PATCH] powercap: DTPM: Fix the missing cpufreq_cpu_put() calls

[Rafael J. Wysocki]

On Fri, Dec 1, 2023 at 1:31 PM Lukasz Luba <lukasz.luba@arm.com> wrote:
>
> The policy returned by cpufreq_cpu_get() has to be released with
> the help of cpufreq_cpu_put() to balance its kobject reference counter
> properly.
>
> Add the missing calls to cpufreq_cpu_put() in the code.
>
> Fixes: 0aea2e4ec2a2 ("powercap/dtpm_cpu: Reset per_cpu variable in the release function")
> Fixes: 0e8f68d7f048 ("powercap/drivers/dtpm: Add CPU energy model based support")
> Cc: <stable@vger.kernel.org> # v5.10+
> Signed-off-by: Lukasz Luba <lukasz.luba@arm.com>
> ---
>  drivers/powercap/dtpm_cpu.c | 17 +++++++++++++----
>  1 file changed, 13 insertions(+), 4 deletions(-)
>
> diff --git a/drivers/powercap/dtpm_cpu.c b/drivers/powercap/dtpm_cpu.c
> index 45bb7e2849d7..aac278e162d9 100644
> --- a/drivers/powercap/dtpm_cpu.c
> +++ b/drivers/powercap/dtpm_cpu.c
> @@ -152,6 +152,8 @@ static void pd_release(struct dtpm *dtpm)
>         if (policy) {
>                 for_each_cpu(dtpm_cpu->cpu, policy->related_cpus)
>                         per_cpu(dtpm_per_cpu, dtpm_cpu->cpu) = NULL;
> +
> +               cpufreq_cpu_put(policy);
>         }
>
>         kfree(dtpm_cpu);
> @@ -204,12 +206,16 @@ static int __dtpm_cpu_setup(int cpu, struct dtpm *parent)
>                 return 0;
>
>         pd = em_cpu_get(cpu);
> -       if (!pd || em_is_artificial(pd))
> -               return -EINVAL;
> +       if (!pd || em_is_artificial(pd)) {
> +               ret = -EINVAL;
> +               goto release_policy;
> +       }
>
>         dtpm_cpu = kzalloc(sizeof(*dtpm_cpu), GFP_KERNEL);
> -       if (!dtpm_cpu)
> -               return -ENOMEM;
> +       if (!dtpm_cpu) {
> +               ret = -ENOMEM;
> +               goto release_policy;
> +       }
>
>         dtpm_init(&dtpm_cpu->dtpm, &dtpm_ops);
>         dtpm_cpu->cpu = cpu;
> @@ -231,6 +237,7 @@ static int __dtpm_cpu_setup(int cpu, struct dtpm *parent)
>         if (ret)
>                 goto out_dtpm_unregister;
>
> +       cpufreq_cpu_put(policy);
>         return 0;
>
>  out_dtpm_unregister:
> @@ -242,6 +249,8 @@ static int __dtpm_cpu_setup(int cpu, struct dtpm *parent)
>                 per_cpu(dtpm_per_cpu, cpu) = NULL;
>         kfree(dtpm_cpu);
>
> +release_policy:
> +       cpufreq_cpu_put(policy);
>         return ret;
>  }
>
> --

Applied as 6.7-rc material with the Cc: stable tag fixed.

Thanks!

Re: [PATCH] powercap: DTPM: Fix the missing cpufreq_cpu_put() calls

[Lukasz Luba]

On 12/5/23 19:54, Rafael J. Wysocki wrote:
> On Fri, Dec 1, 2023 at 1:31 PM Lukasz Luba <lukasz.luba@arm.com> wrote:
>>
>> The policy returned by cpufreq_cpu_get() has to be released with
>> the help of cpufreq_cpu_put() to balance its kobject reference counter
>> properly.
>>
>> Add the missing calls to cpufreq_cpu_put() in the code.
>>
>> Fixes: 0aea2e4ec2a2 ("powercap/dtpm_cpu: Reset per_cpu variable in the release function")
>> Fixes: 0e8f68d7f048 ("powercap/drivers/dtpm: Add CPU energy model based support")
>> Cc: <stable@vger.kernel.org> # v5.10+
>> Signed-off-by: Lukasz Luba <lukasz.luba@arm.com>
>> ---
>>   drivers/powercap/dtpm_cpu.c | 17 +++++++++++++----
>>   1 file changed, 13 insertions(+), 4 deletions(-)
>>
>> diff --git a/drivers/powercap/dtpm_cpu.c b/drivers/powercap/dtpm_cpu.c
>> index 45bb7e2849d7..aac278e162d9 100644
>> --- a/drivers/powercap/dtpm_cpu.c
>> +++ b/drivers/powercap/dtpm_cpu.c
>> @@ -152,6 +152,8 @@ static void pd_release(struct dtpm *dtpm)
>>          if (policy) {
>>                  for_each_cpu(dtpm_cpu->cpu, policy->related_cpus)
>>                          per_cpu(dtpm_per_cpu, dtpm_cpu->cpu) = NULL;
>> +
>> +               cpufreq_cpu_put(policy);
>>          }
>>
>>          kfree(dtpm_cpu);
>> @@ -204,12 +206,16 @@ static int __dtpm_cpu_setup(int cpu, struct dtpm *parent)
>>                  return 0;
>>
>>          pd = em_cpu_get(cpu);
>> -       if (!pd || em_is_artificial(pd))
>> -               return -EINVAL;
>> +       if (!pd || em_is_artificial(pd)) {
>> +               ret = -EINVAL;
>> +               goto release_policy;
>> +       }
>>
>>          dtpm_cpu = kzalloc(sizeof(*dtpm_cpu), GFP_KERNEL);
>> -       if (!dtpm_cpu)
>> -               return -ENOMEM;
>> +       if (!dtpm_cpu) {
>> +               ret = -ENOMEM;
>> +               goto release_policy;
>> +       }
>>
>>          dtpm_init(&dtpm_cpu->dtpm, &dtpm_ops);
>>          dtpm_cpu->cpu = cpu;
>> @@ -231,6 +237,7 @@ static int __dtpm_cpu_setup(int cpu, struct dtpm *parent)
>>          if (ret)
>>                  goto out_dtpm_unregister;
>>
>> +       cpufreq_cpu_put(policy);
>>          return 0;
>>
>>   out_dtpm_unregister:
>> @@ -242,6 +249,8 @@ static int __dtpm_cpu_setup(int cpu, struct dtpm *parent)
>>                  per_cpu(dtpm_per_cpu, cpu) = NULL;
>>          kfree(dtpm_cpu);
>>
>> +release_policy:
>> +       cpufreq_cpu_put(policy);
>>          return ret;
>>   }
>>
>> --
> 
> Applied as 6.7-rc material with the Cc: stable tag fixed.
> 
> Thanks!
> 

Thank you Rafael!