[PATCH] USB: misc: uss720: unregister parport on probe failure

[PATCH] USB: misc: uss720: unregister parport on probe failure

[Myeonghun Pak]

uss720_probe() registers a parport before reading the 1284 register used
to detect unsupported Belkin F5U002 adapters. If get_1284_register()
fails, the error path drops the driver private data and the USB device
reference, but leaves the parport device registered.

Undo the pre-announce registration with parport_del_port(). Clear
priv->pp before unregistering the port, matching the disconnect path and
avoiding leaving a stale pointer in the private data.

This issue was identified during our ongoing static-analysis research while
reviewing kernel code.

Fixes: 3295f1b866bf ("usb: misc: uss720: check for incompatible versions of the Belkin F5U002")
Cc: stable@vger.kernel.org
Co-developed-by: Ijae Kim <ae878000@gmail.com>
Signed-off-by: Ijae Kim <ae878000@gmail.com>
Signed-off-by: Myeonghun Pak <mhun512@gmail.com>
---
 drivers/usb/misc/uss720.c | 4 ++++
 1 file changed, 4 insertions(+)

diff --git a/drivers/usb/misc/uss720.c b/drivers/usb/misc/uss720.c
index a8af7615b1..8a5ffb19a5 100644
--- a/drivers/usb/misc/uss720.c
+++ b/drivers/usb/misc/uss720.c
@@ -749,6 +749,10 @@ static int uss720_probe(struct usb_interface *intf,
 	return 0;
 
 probe_abort:
+	if (pp) {
+		priv->pp = NULL;
+		parport_del_port(pp);
+	}
 	kill_all_async_requests_priv(priv);
 	kref_put(&priv->ref_count, destroy_priv);
 	return -ENODEV;
-- 
2.47.1

Re: [PATCH] USB: misc: uss720: unregister parport on probe failure

[Alex Henrie]

On Wed, Jul 1, 2026 at 6:23 AM Myeonghun Pak <mhun512@gmail.com> wrote:
>
> uss720_probe() registers a parport before reading the 1284 register used
> to detect unsupported Belkin F5U002 adapters. If get_1284_register()
> fails, the error path drops the driver private data and the USB device
> reference, but leaves the parport device registered.

>  probe_abort:
> +       if (pp) {
> +               priv->pp = NULL;
> +               parport_del_port(pp);
> +       }
>         kill_all_async_requests_priv(priv);
>         kref_put(&priv->ref_count, destroy_priv);
>         return -ENODEV;

I think it would make more sense to set priv->pp to NULL and call
parport_del_port(pp) right before the second `goto probe_abort`
instead of calling parport_del_port in probe_abort. That would avoid
the need for checking whether pp is null before cleaning it up.

Is this just a memory leak, or is there some visible negative
consequence to not cleaning up the struct parport on failure?

-Alex