Re: [PATCH v2] drm/panthor: Fix UAF race between device unplug and FW event processing

[Steven Price]

On 22/10/2025 11:30, Ketil Johnsen wrote:
> The function panthor_fw_unplug() will free the FW memory sections.
> The problem is that there could still be pending FW events which are yet
> not handled at this point. process_fw_events_work() can in this case try
> to access said freed memory.
> 
> This fix introduces a destroyed state for the panthor_scheduler object,
> and we check for this before processing FW events.
> 
> Signed-off-by: Ketil Johnsen <ketil.johnsen@xxxxxxx>
> Fixes: de85488138247 ("drm/panthor: Add the scheduler logical block")

Reviewed-by: Steven Price <steven.price@xxxxxxx>

> ---
> v2:
> - Followed Boris's advice and handle the race purely within the
>   scheduler block (by adding a destroyed state)
> ---
>  drivers/gpu/drm/panthor/panthor_sched.c | 15 ++++++++++++---
>  1 file changed, 12 insertions(+), 3 deletions(-)
> 
> diff --git a/drivers/gpu/drm/panthor/panthor_sched.c b/drivers/gpu/drm/panthor/panthor_sched.c
> index 0cc9055f4ee52..4996f987b8183 100644
> --- a/drivers/gpu/drm/panthor/panthor_sched.c
> +++ b/drivers/gpu/drm/panthor/panthor_sched.c
> @@ -315,6 +315,13 @@ struct panthor_scheduler {
>  		 */
>  		struct list_head stopped_groups;
>  	} reset;
> +
> +	/**
> +	 * @destroyed: Scheduler object is (being) destroyed
> +	 *
> +	 * Normal scheduler operations should no longer take place.
> +	 */
> +	bool destroyed;
>  };
>  
>  /**
> @@ -1765,7 +1772,10 @@ static void process_fw_events_work(struct work_struct *work)
>  	u32 events = atomic_xchg(&sched->fw_events, 0);
>  	struct panthor_device *ptdev = sched->ptdev;
>  
> -	mutex_lock(&sched->lock);
> +	guard(mutex)(&sched->lock);
> +
> +	if (sched->destroyed)
> +		return;
>  
>  	if (events & JOB_INT_GLOBAL_IF) {
>  		sched_process_global_irq_locked(ptdev);
> @@ -1778,8 +1788,6 @@ static void process_fw_events_work(struct work_struct *work)
>  		sched_process_csg_irq_locked(ptdev, csg_id);
>  		events &= ~BIT(csg_id);
>  	}
> -
> -	mutex_unlock(&sched->lock);
>  }
>  
>  /**
> @@ -3882,6 +3890,7 @@ void panthor_sched_unplug(struct panthor_device *ptdev)
>  	cancel_delayed_work_sync(&sched->tick_work);
>  
>  	mutex_lock(&sched->lock);
> +	sched->destroyed = true;
>  	if (sched->pm.has_ref) {
>  		pm_runtime_put(ptdev->base.dev);
>  		sched->pm.has_ref = false;