Re: [PATCH] can: j1939: j1939_sk_bind(): fix j1939_ecu leak when re-bind failed

Hello, sashiko.

On 2026/08/21 1:40, [email protected] wrote:
> Thank you for your contribution! Sashiko AI review found 1 potential issue(s) to consider:
> 
> Pre-existing issues:
> - [High] Netdevice refcount leak due to a race condition where the unregister notifier misses a socket during bind or re-bind list manipulations.
> --

Thanks for the comment. However, I consider this is a false positive and won't cause a permanent
hang. If j1939_netdev_stop() is temporarily skipped because a socket is missing from the list
during a re-bind window, priv->rx_kref remains non-zero. This prevents __j1939_rx_release() from
clearing the netdev mapping (j1939_priv_set(priv->ndev, NULL)). Since the netdev core repeatedly
retries NETDEV_UNREGISTER notifications as long as references exist, the next retry window will
successfully capture the returned socket on the list and execute j1939_netdev_stop() properly.