The mceusb_disconnect() function calls rc_unregister_device() before
usb_kill_urb(), creating a race condition similar to the one fixed in
commit 'f656cfbc7a29'. If rc_unregister_device() executes while the URB
callback mceusb_dev_recv() is still running, mceusb_process_ir_data()
may call ir_raw_event_store_with_filter() or ir_raw_event_handle() with
a freed ir->rc->raw pointer, leading to a NULL pointer dereference.

Call usb_kill_urb() before rc_unregister_device() to ensure all URB
callbacks complete before freeing device resources.

Fixes: d8b4b5822f51 ("[media] ir-core: make struct rc_dev the primary interface")
Signed-off-by: Haotian Zhang <vulab@iscas.ac.cn>
---
 drivers/media/rc/mceusb.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/drivers/media/rc/mceusb.c b/drivers/media/rc/mceusb.c
index 044767eb3a38..fb5c393dcb94 100644
--- a/drivers/media/rc/mceusb.c
+++ b/drivers/media/rc/mceusb.c
@@ -1845,8 +1845,8 @@ static void mceusb_dev_disconnect(struct usb_interface *intf)
 
 	ir->usbdev = NULL;
 	cancel_work_sync(&ir->kevent);
-	rc_unregister_device(ir->rc);
 	usb_kill_urb(ir->urb_in);
+	rc_unregister_device(ir->rc);
 	usb_free_urb(ir->urb_in);
 	usb_free_coherent(dev, ir->len_in, ir->buf_in, ir->dma_in);
 	usb_put_dev(dev);
-- 
2.50.1.windows.1

On Fri, Nov 14, 2025 at 06:14:31PM +0800, Haotian Zhang wrote:
> The mceusb_disconnect() function calls rc_unregister_device() before
> usb_kill_urb(), creating a race condition similar to the one fixed in
> commit 'f656cfbc7a29'. If rc_unregister_device() executes while the URB
> callback mceusb_dev_recv() is still running, mceusb_process_ir_data()
> may call ir_raw_event_store_with_filter() or ir_raw_event_handle() with
> a freed ir->rc->raw pointer, leading to a NULL pointer dereference.
> 
> Call usb_kill_urb() before rc_unregister_device() to ensure all URB
> callbacks complete before freeing device resources.

First of all, I must apologise for not responding earlier. Haotian, you
have found a real issue here, thank you for finding and reporting it.

I've been thinking about this for some time now. I think I've found an
alternative solution, which I'l shortly be posting to the list.

Thank you 

Sean

> 
> Fixes: d8b4b5822f51 ("[media] ir-core: make struct rc_dev the primary interface")
> Signed-off-by: Haotian Zhang <vulab@iscas.ac.cn>
> ---
>  drivers/media/rc/mceusb.c | 2 +-
>  1 file changed, 1 insertion(+), 1 deletion(-)
> 
> diff --git a/drivers/media/rc/mceusb.c b/drivers/media/rc/mceusb.c
> index 044767eb3a38..fb5c393dcb94 100644
> --- a/drivers/media/rc/mceusb.c
> +++ b/drivers/media/rc/mceusb.c
> @@ -1845,8 +1845,8 @@ static void mceusb_dev_disconnect(struct usb_interface *intf)
>  
>  	ir->usbdev = NULL;
>  	cancel_work_sync(&ir->kevent);
> -	rc_unregister_device(ir->rc);
>  	usb_kill_urb(ir->urb_in);
> +	rc_unregister_device(ir->rc);
>  	usb_free_urb(ir->urb_in);
>  	usb_free_coherent(dev, ir->len_in, ir->buf_in, ir->dma_in);
>  	usb_put_dev(dev);
> -- 
> 2.50.1.windows.1