[PATCH RESEND] cpufreq/longhaul: handle NULL policy in longhaul_exit

[PATCH RESEND] cpufreq/longhaul: handle NULL policy in longhaul_exit

[Dennis Beier]

Resending this patch to include cpufreq maintainers.

---

longhaul_exit() was calling cpufreq_cpu_get(0) without checking
for a NULL policy pointer. On some systems, this could lead to a
NULL dereference and a kernel warning or panic.

This patch adds a check using unlikely() and prints a warning
if the policy is NULL, then returns early. Also, the loop variable
is now declared inside the for-loop to match modern kernel style.

Bugzilla: #219962

Signed-off-by: Dennis Beier <nanovim@gmail.com>

---
 drivers/cpufreq/longhaul.c | 9 +++++++--
 1 file changed, 7 insertions(+), 2 deletions(-)

diff --git a/drivers/cpufreq/longhaul.c b/drivers/cpufreq/longhaul.c
index ba0e08c8486a..9698d56bfe6c 100644
--- a/drivers/cpufreq/longhaul.c
+++ b/drivers/cpufreq/longhaul.c
@@ -951,9 +951,14 @@ static int __init longhaul_init(void)
 static void __exit longhaul_exit(void)
 {
 	struct cpufreq_policy *policy = cpufreq_cpu_get(0);
-	int i;
+	if (unlikely(!policy)) {
+		pr_warn_once("longhaul_exit: policy is NULL\n");
+
+		return;
+	}
+
 
-	for (i = 0; i < numscales; i++) {
+	for (int i = 0; i < numscales; i++) {
 		if (mults[i] == maxmult) {
 			struct cpufreq_freqs freqs;
 
-- 
2.50.1

Re: [PATCH RESEND] cpufreq/longhaul: handle NULL policy in longhaul_exit

[Viresh Kumar]

On 28-08-25, 23:29, Dennis Beier wrote:
> 
> Resending this patch to include cpufreq maintainers.

This should have been added ...

> ---
> 
> longhaul_exit() was calling cpufreq_cpu_get(0) without checking
> for a NULL policy pointer. On some systems, this could lead to a
> NULL dereference and a kernel warning or panic.
> 
> This patch adds a check using unlikely() and prints a warning
> if the policy is NULL, then returns early. Also, the loop variable
> is now declared inside the for-loop to match modern kernel style.
> 
> Bugzilla: #219962
> 
> Signed-off-by: Dennis Beier <nanovim@gmail.com>
> 
> ---

... here. Right now if we apply the patch, it will only contain the
top line.

This is what I get now if I apply your patch with `git am`

Author: Dennis Beier <nanovim@gmail.com>
Date:   Thu Aug 28 23:29:36 2025 +0200

    cpufreq/longhaul: handle NULL policy in longhaul_exit

    Resending this patch to include cpufreq maintainers.

    Signed-off-by: Viresh Kumar <viresh.kumar@linaro.org>
---
 drivers/cpufreq/longhaul.c | 9 +++++++--
 1 file changed, 7 insertions(+), 2 deletions(-)


>  drivers/cpufreq/longhaul.c | 9 +++++++--
>  1 file changed, 7 insertions(+), 2 deletions(-)
> 
> diff --git a/drivers/cpufreq/longhaul.c b/drivers/cpufreq/longhaul.c
> index ba0e08c8486a..9698d56bfe6c 100644
> --- a/drivers/cpufreq/longhaul.c
> +++ b/drivers/cpufreq/longhaul.c
> @@ -951,9 +951,14 @@ static int __init longhaul_init(void)
>  static void __exit longhaul_exit(void)
>  {
>  	struct cpufreq_policy *policy = cpufreq_cpu_get(0);
> -	int i;

Please don't make another unnecessary change along with a bug fix. If
you really want that, you should do that in a separate patch. Also, it
is just not required.

Add a blank line here please.

> +	if (unlikely(!policy)) {
> +		pr_warn_once("longhaul_exit: policy is NULL\n");

Not sure if the warning is going to be of any use for the user. I
would just return silently, the module is going away anyway.

> +
> +		return;
> +	}
> +
>  
> -	for (i = 0; i < numscales; i++) {
> +	for (int i = 0; i < numscales; i++) {
>  		if (mults[i] == maxmult) {
>  			struct cpufreq_freqs freqs;
>  
> -- 
> 2.50.1

-- 
viresh