net: Do not clear the sock TX queue in sk_set_socket()

Clearing the sock TX queue in sk_set_socket() might cause unexpected
out-of-order transmit when called from sock_orphan(), as outstanding
packets can pick a different TX queue and bypass the ones already queued.

This is undesired in general. More specifically, it breaks the in-order
scheduling property guarantee for device-offloaded TLS sockets.

Remove the call to sk_tx_queue_clear() in sk_set_socket(), and add it
explicitly only where needed.

Fixes: e022f0b ("net: Introduce sk_tx_queue_mapping")
Signed-off-by: Tariq Toukan <[email protected]>
Reviewed-by: Boris Pismenny <[email protected]>
Signed-off-by: David S. Miller <[email protected]>

Author: Tariq Toukan

include/net/sock.h

@@ -1848,7 +1848,6 @@ static inline int sk_rx_queue_get(const struct sock *sk)
 
 static inline void sk_set_socket(struct sock *sk, struct socket *sock)
 {
-	sk_tx_queue_clear(sk);
 	sk->sk_socket = sock;
 }
 

net/core/sock.c

@@ -1767,6 +1767,7 @@ struct sock *sk_alloc(struct net *net, int family, gfp_t priority,
 		cgroup_sk_alloc(&sk->sk_cgrp_data);
 		sock_update_classid(&sk->sk_cgrp_data);
 		sock_update_netprioidx(&sk->sk_cgrp_data);
+		sk_tx_queue_clear(sk);
 	}
 
 	return sk;
@@ -1990,6 +1991,7 @@ struct sock *sk_clone_lock(const struct sock *sk, const gfp_t priority)
 		 */
 		sk_refcnt_debug_inc(newsk);
 		sk_set_socket(newsk, NULL);
+		sk_tx_queue_clear(newsk);
 		RCU_INIT_POINTER(newsk->sk_wq, NULL);
 
 		if (newsk->sk_prot->sockets_allocated)