tcp: fix req->saved_syn race

For the reasons explained in commit ce1050089c96 ("tcp/dccp: fix
ireq->pktopts race"), we need to make sure we do not access
req->saved_syn unless we own the request sock.

This fixes races for listeners using TCP_SAVE_SYN option.

Fixes: e994b2f0fb92 ("tcp: do not lock listener to process SYN packets")
Fixes: 079096f103fa ("tcp/dccp: install syn_recv requests into ehash table")
Signed-off-by: Eric Dumazet <edumazet@google.com>
Reported-by: Ying Cai <ycai@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
diff --git a/include/linux/tcp.h b/include/linux/tcp.h
index c906f45..b386361 100644
--- a/include/linux/tcp.h
+++ b/include/linux/tcp.h
@@ -397,6 +397,13 @@
 	queue->fastopenq.max_qlen = min_t(unsigned int, backlog, somaxconn);
 }
 
+static inline void tcp_move_syn(struct tcp_sock *tp,
+				struct request_sock *req)
+{
+	tp->saved_syn = req->saved_syn;
+	req->saved_syn = NULL;
+}
+
 static inline void tcp_saved_syn_free(struct tcp_sock *tp)
 {
 	kfree(tp->saved_syn);