tcp: reduce accepted window in NEW_SYN_RECV state

Jason commit made checks against ACK sequence less strict
and can be exploited by attackers to establish spoofed flows
with less probes.

Innocent users might use tcp_rmem[1] == 1,000,000,000,
or something more reasonable.

An attacker can use a regular TCP connection to learn the server
initial tp->rcv_wnd, and use it to optimize the attack.

If we make sure that only the announced window (smaller than 65535)
is used for ACK validation, we force an attacker to use
65537 packets to complete the 3WHS (assuming server ISN is unknown)

Fixes: 378979e9 ("tcp: remove 64 KByte limit for initial tp->rcv_wnd value")
Link: https://datatracker.ietf.org/meeting/119/materials/slides-119-tcpm-ghost-acks-00Signed-off-by: Eric Dumazet <edumazet@google.com>
Acked-by: Neal Cardwell <ncardwell@google.com>
Reviewed-by: Jason Xing <kerneljasonxing@gmail.com>
Link: https://lore.kernel.org/r/20240523130528.60376-1-edumazet@google.comSigned-off-by: Jakub Kicinski <kuba@kernel.org>

include/net/request_sock.h

@@ -285,4 +285,16 @@ static inline int reqsk_queue_len_young(const struct request_sock_queue *queue)
 	return atomic_read(&queue->young);
 }
 
+/* RFC 7323 2.3 Using the Window Scale Option
+ *  The window field (SEG.WND) of every outgoing segment, with the
+ *  exception of <SYN> segments, MUST be right-shifted by
+ *  Rcv.Wind.Shift bits.
+ *
+ * This means the SEG.WND carried in SYNACK can not exceed 65535.
+ * We use this property to harden TCP stack while in NEW_SYN_RECV state.
+ */
+static inline u32 tcp_synack_window(const struct request_sock *req)
+{
+	return min(req->rsk_rcv_wnd, 65535U);
+}
 #endif /* _REQUEST_SOCK_H */

net/ipv4/tcp_ipv4.c

@@ -1144,14 +1144,9 @@ static void tcp_v4_reqsk_send_ack(const struct sock *sk, struct sk_buff *skb,
 #endif
 	}
 
-	/* RFC 7323 2.3
-	 * The window field (SEG.WND) of every outgoing segment, with the
-	 * exception of <SYN> segments, MUST be right-shifted by
-	 * Rcv.Wind.Shift bits:
-	 */
 	tcp_v4_send_ack(sk, skb, seq,
 			tcp_rsk(req)->rcv_nxt,
-			req->rsk_rcv_wnd >> inet_rsk(req)->rcv_wscale,
+			tcp_synack_window(req) >> inet_rsk(req)->rcv_wscale,
 			tcp_rsk_tsval(tcp_rsk(req)),
 			READ_ONCE(req->ts_recent),
 			0, &key,

net/ipv4/tcp_minisocks.c

@@ -783,8 +783,11 @@ struct sock *tcp_check_req(struct sock *sk, struct sk_buff *skb,
 
 	/* RFC793: "first check sequence number". */
 
-	if (paws_reject || !tcp_in_window(TCP_SKB_CB(skb)->seq, TCP_SKB_CB(skb)->end_seq,
-					  tcp_rsk(req)->rcv_nxt, tcp_rsk(req)->rcv_nxt + req->rsk_rcv_wnd)) {
+	if (paws_reject || !tcp_in_window(TCP_SKB_CB(skb)->seq,
+					  TCP_SKB_CB(skb)->end_seq,
+					  tcp_rsk(req)->rcv_nxt,
+					  tcp_rsk(req)->rcv_nxt +
+					  tcp_synack_window(req))) {
 		/* Out of window: send ACK and drop. */
 		if (!(flg & TCP_FLAG_RST) &&
 		    !tcp_oow_rate_limited(sock_net(sk), skb,

net/ipv6/tcp_ipv6.c

@@ -1272,15 +1272,10 @@ static void tcp_v6_reqsk_send_ack(const struct sock *sk, struct sk_buff *skb,
 	/* sk->sk_state == TCP_LISTEN -> for regular TCP_SYN_RECV
 	 * sk->sk_state == TCP_SYN_RECV -> for Fast Open.
 	 */
-	/* RFC 7323 2.3
-	 * The window field (SEG.WND) of every outgoing segment, with the
-	 * exception of <SYN> segments, MUST be right-shifted by
-	 * Rcv.Wind.Shift bits:
-	 */
 	tcp_v6_send_ack(sk, skb, (sk->sk_state == TCP_LISTEN) ?
 			tcp_rsk(req)->snt_isn + 1 : tcp_sk(sk)->snd_nxt,
 			tcp_rsk(req)->rcv_nxt,
-			req->rsk_rcv_wnd >> inet_rsk(req)->rcv_wscale,
+			tcp_synack_window(req) >> inet_rsk(req)->rcv_wscale,
 			tcp_rsk_tsval(tcp_rsk(req)),
 			READ_ONCE(req->ts_recent), sk->sk_bound_dev_if,
 			&key, ipv6_get_dsfield(ipv6_hdr(skb)), 0,