CVE-2026-80849

In the Linux kernel, the following vulnerability has been resolved:

net/tcp-ao: fix use-after-free of current key on reconnect to another peer

tcp inbound ao hash() is called before bh lock sock nested() is taken, with only rcu read lock() held. On the fast path for established sockets, if the rnext keyid sent by the peer differs from current key->sndid, the key the peer asked for is looked up and stored in current key. The lookup is inside the RCU read side, but current key outlives it.

When the socket is disconnected and connect() is called again for another peer, tcp ao connect init() unlinks every key that does not match the new peer and frees it with call rcu(). If current key points at such a key, it is cleared to NULL.

The fast path reads sk state only once on entry, so a softirq that got into it while the socket was still established can update current key after that loop has already run. The update is inside the RCU read side, so it comes before the call rcu() callback, and once the callback frees the key, current key is left pointing at freed memory.

The next transmission picks that pointer up in tcp get current key(). tcp ao transmit skb() then reads the traffic key from the freed object, which is the use-after-free.

Wait for one grace period before unlinking, and only if a key is going to be removed. By the time tcp connect() runs the socket is already in TCP SYN SENT, and TCP AO ESTABLISHED does not contain TCPF SYN SENT, so a softirq entering after the wait cannot reach the fast path, and the ones already in it have finished. The existing NULL handling in the loop is then enough.