tcp: attach SYNACK messages to request sockets instead of listener

If a listen backlog is very big (to avoid syncookies), then
the listener sk->sk_wmem_alloc is the main source of false
sharing, as we need to touch it twice per SYNACK re-transmit
and TX completion.

(One SYN packet takes listener lock once, but up to 6 SYNACK
are generated)

By attaching the skb to the request socket, we remove this
source of contention.

Tested:

 listen(fd, 10485760); // single listener (no SO_REUSEPORT)
 16 RX/TX queue NIC
 Sustain a SYNFLOOD attack of ~320,000 SYN per second,
 Sending ~1,400,000 SYNACK per second.
 Perf profiles now show listener spinlock being next bottleneck.

    20.29%  [kernel]  [k] queued_spin_lock_slowpath
    10.06%  [kernel]  [k] __inet_lookup_established
     5.12%  [kernel]  [k] reqsk_timer_handler
     3.22%  [kernel]  [k] get_next_timer_interrupt
     3.00%  [kernel]  [k] tcp_make_synack
     2.77%  [kernel]  [k] ipt_do_table
     2.70%  [kernel]  [k] run_timer_softirq
     2.50%  [kernel]  [k] ip_finish_output
     2.04%  [kernel]  [k] cascade

Signed-off-by: Eric Dumazet <edumazet@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

1 files changed, 15 insertions, 7 deletions

diff --git a/net/ipv4/tcp_output.c b/net/ipv4/tcp_output.c
index 09bb082ca1a7..55ed3266b05f 100644
--- a/net/ipv4/tcp_output.c
+++ b/net/ipv4/tcp_output.c
@@ -2947,7 +2947,8 @@ int tcp_send_synack(struct sock *sk)
  */
 struct sk_buff *tcp_make_synack(const struct sock *sk, struct dst_entry *dst,
 				struct request_sock *req,
-				struct tcp_fastopen_cookie *foc)
+				struct tcp_fastopen_cookie *foc,
+				bool attach_req)
 {
 	struct inet_request_sock *ireq = inet_rsk(req);
 	const struct tcp_sock *tp = tcp_sk(sk);
@@ -2959,11 +2960,7 @@ struct sk_buff *tcp_make_synack(const struct sock *sk, struct dst_entry *dst,
 	u16 user_mss;
 	int mss;
 
-	/* sk is a const pointer, because we want to express multiple cpus
-	 * might call us concurrently.
-	 * sock_wmalloc() will change sk->sk_wmem_alloc in an atomic way.
-	 */
-	skb = sock_wmalloc((struct sock *)sk, MAX_TCP_HEADER, 1, GFP_ATOMIC);
+	skb = alloc_skb(MAX_TCP_HEADER, GFP_ATOMIC);
 	if (unlikely(!skb)) {
 		dst_release(dst);
 		return NULL;
@@ -2971,6 +2968,17 @@ struct sk_buff *tcp_make_synack(const struct sock *sk, struct dst_entry *dst,
 	/* Reserve space for headers. */
 	skb_reserve(skb, MAX_TCP_HEADER);
 
+	if (attach_req) {
+		skb->destructor = sock_edemux;
+		sock_hold(req_to_sk(req));
+		skb->sk = req_to_sk(req);
+	} else {
+		/* sk is a const pointer, because we want to express multiple
+		 * cpu might call us concurrently.
+		 * sk->sk_wmem_alloc in an atomic, we can promote to rw.
+		 */
+		skb_set_owner_w(skb, (struct sock *)sk);
+	}
 	skb_dst_set(skb, dst);
 
 	mss = dst_metric_advmss(dst);
@@ -3510,7 +3518,7 @@ int tcp_rtx_synack(const struct sock *sk, struct request_sock *req)
 	int res;
 
 	tcp_rsk(req)->txhash = net_tx_rndhash();
-	res = af_ops->send_synack(sk, NULL, &fl, req, 0, NULL);
+	res = af_ops->send_synack(sk, NULL, &fl, req, 0, NULL, true);
 	if (!res) {
 		TCP_INC_STATS_BH(sock_net(sk), TCP_MIB_RETRANSSEGS);
 		NET_INC_STATS_BH(sock_net(sk), LINUX_MIB_TCPSYNRETRANS);