Merge branch 'tcp-exponential-backoff-in-tcp_send_ack'

Eric Dumazet says:

====================
tcp: exponential backoff in tcp_send_ack()

We had outages caused by repeated skb allocation failures in tcp_send_ack()

It is time to add exponential backoff to reduce number of attempts.
Before doing so, first patch removes icsk_ack.blocked to make
room for a new field (icsk_ack.retry)
====================

Signed-off-by: David S. Miller <[email protected]>

6 files changed, 16 insertions, 17 deletions

diff --git a/include/net/inet_connection_sock.h b/include/net/inet_connection_sock.h
index dc763ca9413c..7338b3865a2a 100644
--- a/include/net/inet_connection_sock.h
+++ b/include/net/inet_connection_sock.h
@@ -110,7 +110,7 @@ struct inet_connection_sock {
 		__u8		  pending;	 /* ACK is pending			   */
 		__u8		  quick;	 /* Scheduled number of quick acks	   */
 		__u8		  pingpong;	 /* The session is interactive		   */
-		__u8		  blocked;	 /* Delayed ACK was blocked by socket lock */
+		__u8		  retry;	 /* Number of attempts			   */
 		__u32		  ato;		 /* Predicted tick of soft clock	   */
 		unsigned long	  timeout;	 /* Currently scheduled timeout		   */
 		__u32		  lrcvtime;	 /* timestamp of last received data packet */
@@ -198,7 +198,8 @@ static inline void inet_csk_clear_xmit_timer(struct sock *sk, const int what)
 		sk_stop_timer(sk, &icsk->icsk_retransmit_timer);
 #endif
 	} else if (what == ICSK_TIME_DACK) {
-		icsk->icsk_ack.blocked = icsk->icsk_ack.pending = 0;
+		icsk->icsk_ack.pending = 0;
+		icsk->icsk_ack.retry = 0;
 #ifdef INET_CSK_CLEAR_TIMERS
 		sk_stop_timer(sk, &icsk->icsk_delack_timer);
 #endif
diff --git a/net/dccp/timer.c b/net/dccp/timer.c
index 927c796d7682..a934d2932373 100644
--- a/net/dccp/timer.c
+++ b/net/dccp/timer.c
@@ -176,7 +176,6 @@ static void dccp_delack_timer(struct timer_list *t)
 	bh_lock_sock(sk);
 	if (sock_owned_by_user(sk)) {
 		/* Try again later. */
-		icsk->icsk_ack.blocked = 1;
 		__NET_INC_STATS(sock_net(sk), LINUX_MIB_DELAYEDACKLOCKED);
 		sk_reset_timer(sk, &icsk->icsk_delack_timer,
 			       jiffies + TCP_DELACK_MIN);
diff --git a/net/ipv4/inet_connection_sock.c b/net/ipv4/inet_connection_sock.c
index b457dd2d6c75..4148f5f78f31 100644
--- a/net/ipv4/inet_connection_sock.c
+++ b/net/ipv4/inet_connection_sock.c
@@ -564,7 +564,7 @@ void inet_csk_clear_xmit_timers(struct sock *sk)
 {
 	struct inet_connection_sock *icsk = inet_csk(sk);
 
-	icsk->icsk_pending = icsk->icsk_ack.pending = icsk->icsk_ack.blocked = 0;
+	icsk->icsk_pending = icsk->icsk_ack.pending = 0;
 
 	sk_stop_timer(sk, &icsk->icsk_retransmit_timer);
 	sk_stop_timer(sk, &icsk->icsk_delack_timer);
diff --git a/net/ipv4/tcp.c b/net/ipv4/tcp.c
index 2a8bfa89a515..ed2805564424 100644
--- a/net/ipv4/tcp.c
+++ b/net/ipv4/tcp.c
@@ -1538,10 +1538,8 @@ void tcp_cleanup_rbuf(struct sock *sk, int copied)
 
 	if (inet_csk_ack_scheduled(sk)) {
 		const struct inet_connection_sock *icsk = inet_csk(sk);
-		   /* Delayed ACKs frequently hit locked sockets during bulk
-		    * receive. */
-		if (icsk->icsk_ack.blocked ||
-		    /* Once-per-two-segments ACK was not sent by tcp_input.c */
+
+		if (/* Once-per-two-segments ACK was not sent by tcp_input.c */
 		    tp->rcv_nxt - tp->rcv_wup > icsk->icsk_ack.rcv_mss ||
 		    /*
 		     * If this read emptied read buffer, we send ACK, if
diff --git a/net/ipv4/tcp_output.c b/net/ipv4/tcp_output.c
index 386978dcd318..bf48cd73e967 100644
--- a/net/ipv4/tcp_output.c
+++ b/net/ipv4/tcp_output.c
@@ -3911,11 +3911,8 @@ void tcp_send_delayed_ack(struct sock *sk)
 
 	/* Use new timeout only if there wasn't a older one earlier. */
 	if (icsk->icsk_ack.pending & ICSK_ACK_TIMER) {
-		/* If delack timer was blocked or is about to expire,
-		 * send ACK now.
-		 */
-		if (icsk->icsk_ack.blocked ||
-		    time_before_eq(icsk->icsk_ack.timeout, jiffies + (ato >> 2))) {
+		/* If delack timer is about to expire, send ACK now. */
+		if (time_before_eq(icsk->icsk_ack.timeout, jiffies + (ato >> 2))) {
 			tcp_send_ack(sk);
 			return;
 		}
@@ -3944,10 +3941,15 @@ void __tcp_send_ack(struct sock *sk, u32 rcv_nxt)
 	buff = alloc_skb(MAX_TCP_HEADER,
 			 sk_gfp_mask(sk, GFP_ATOMIC | __GFP_NOWARN));
 	if (unlikely(!buff)) {
+		struct inet_connection_sock *icsk = inet_csk(sk);
+		unsigned long delay;
+
+		delay = TCP_DELACK_MAX << icsk->icsk_ack.retry;
+		if (delay < TCP_RTO_MAX)
+			icsk->icsk_ack.retry++;
 		inet_csk_schedule_ack(sk);
-		inet_csk(sk)->icsk_ack.ato = TCP_ATO_MIN;
-		inet_csk_reset_xmit_timer(sk, ICSK_TIME_DACK,
-					  TCP_DELACK_MAX, TCP_RTO_MAX);
+		icsk->icsk_ack.ato = TCP_ATO_MIN;
+		inet_csk_reset_xmit_timer(sk, ICSK_TIME_DACK, delay, TCP_RTO_MAX);
 		return;
 	}
 
diff --git a/net/ipv4/tcp_timer.c b/net/ipv4/tcp_timer.c
index 0c08c420fbc2..6c62b9ea1320 100644
--- a/net/ipv4/tcp_timer.c
+++ b/net/ipv4/tcp_timer.c
@@ -331,7 +331,6 @@ static void tcp_delack_timer(struct timer_list *t)
 	if (!sock_owned_by_user(sk)) {
 		tcp_delack_timer_handler(sk);
 	} else {
-		icsk->icsk_ack.blocked = 1;
 		__NET_INC_STATS(sock_net(sk), LINUX_MIB_DELAYEDACKLOCKED);
 		/* deleguate our work to tcp_release_cb() */
 		if (!test_and_set_bit(TCP_DELACK_TIMER_DEFERRED, &sk->sk_tsq_flags))