Re: [PATCH v2] scsi: ufs: core: Fix data race in CPU latency PM QoS request handling

[Bart Van Assche <bvanassche@acm.org>]

On 9/2/25 12:48 AM, Zhongqiu Han wrote:
> -	return sysfs_emit(buf, "%d\n", hba->pm_qos_enabled);
> +	return sysfs_emit(buf, "%d\n", READ_ONCE(hba->pm_qos_enabled));

Using READ_ONCE() here is inconsistent since none of the modifications
of hba->pm_qos_enabled use WRITE_ONCE(). Protecting hba->pm_qos_enabled
modifications with a mutex is not sufficient since the above read of
hba->pm_qos_enabled is not protected by the same mutex.

Has it been considered to leave out the READ_ONCE() from the above code
and instead to add the following above the sysfs_emit() call?

guard(mutex)(&hba->pm_qos_mutex);

> diff --git a/drivers/ufs/core/ufshcd.c b/drivers/ufs/core/ufshcd.c
> index 926650412eaa..98b9ce583386 100644
> --- a/drivers/ufs/core/ufshcd.c
> +++ b/drivers/ufs/core/ufshcd.c
> @@ -1047,14 +1047,19 @@ EXPORT_SYMBOL_GPL(ufshcd_is_hba_active);
>    */
>   void ufshcd_pm_qos_init(struct ufs_hba *hba)
>   {
> +	mutex_lock(&hba->pm_qos_mutex);
>   
> -	if (hba->pm_qos_enabled)
> +	if (hba->pm_qos_enabled) {
> +		mutex_unlock(&hba->pm_qos_mutex);
>   		return;
> +	}
>   
>   	cpu_latency_qos_add_request(&hba->pm_qos_req, PM_QOS_DEFAULT_VALUE);
>   
>   	if (cpu_latency_qos_request_active(&hba->pm_qos_req))
>   		hba->pm_qos_enabled = true;
> +
> +	mutex_unlock(&hba->pm_qos_mutex);
>   }

Please make the above code easier to review by using
guard(mutex)(&hba->pm_qos_mutex) instead of explicit mutex_lock() and
mutex_unlock() calls.

> @@ -1063,11 +1068,16 @@ void ufshcd_pm_qos_init(struct ufs_hba *hba)
>    */
>   void ufshcd_pm_qos_exit(struct ufs_hba *hba)
>   {
> -	if (!hba->pm_qos_enabled)
> +	mutex_lock(&hba->pm_qos_mutex);
> +
> +	if (!hba->pm_qos_enabled) {
> +		mutex_unlock(&hba->pm_qos_mutex);
>   		return;
> +	}
>   
>   	cpu_latency_qos_remove_request(&hba->pm_qos_req);
>   	hba->pm_qos_enabled = false;
> +	mutex_unlock(&hba->pm_qos_mutex);
>   }

Same comment here: please make the above code easier to review by using
guard(mutex)(&hba->pm_qos_mutex) instead of explicit mutex_lock() and
mutex_unlock() calls.

> @@ -1077,10 +1087,15 @@ void ufshcd_pm_qos_exit(struct ufs_hba *hba)
>    */
>   static void ufshcd_pm_qos_update(struct ufs_hba *hba, bool on)
>   {
> -	if (!hba->pm_qos_enabled)
> +	mutex_lock(&hba->pm_qos_mutex);
> +
> +	if (!hba->pm_qos_enabled) {
> +		mutex_unlock(&hba->pm_qos_mutex);
>   		return;
> +	}
>   
>   	cpu_latency_qos_update_request(&hba->pm_qos_req, on ? 0 : PM_QOS_DEFAULT_VALUE);
> +	mutex_unlock(&hba->pm_qos_mutex);
>   }

Also in the above code, please use the guard()() macro instead of
explicit mutex_lock() and mutex_unlock() calls.

Thanks,

Bart.