tcp: export count for rehash attempts

Using IPv6 flow-label to swiftly route around avoid congested or
disconnected network path can greatly improve TCP reliability.

This patch adds SNMP counters and a OPT_STATS counter to track both
host-level and connection-level statistics. Network administrators
can use these counters to evaluate the impact of this new ability better.

Export count for rehash attempts to
1) two SNMP counters: TcpTimeoutRehash (rehash due to timeouts),
and TcpDuplicateDataRehash (rehash due to receiving duplicate
packets)
2) Timestamping API SOF_TIMESTAMPING_OPT_STATS.

Signed-off-by: Abdul Kabbani <[e-mail]>
Signed-off-by: Neal Cardwell <[e-mail]>
Signed-off-by: Yuchung Cheng <[e-mail]>
Signed-off-by: Kevin(Yudong) Yang <[e-mail]>
Signed-off-by: Eric Dumazet <[e-mail]>
Signed-off-by: David S. Miller <[e-mail]>

include/linux/tcp.h

···
 #define BPF_SOCK_OPS_TEST_FLAG(TP, ARG) 0
 #endif
 
+	u16 timeout_rehash;	/* Timeout-triggered rehash attempts */
+
 	u32 rcv_ooopack; /* Received out-of-order packets, for tcpinfo */
 
 /* Receiver side RTT estimation */

include/uapi/linux/snmp.h

···
 	LINUX_MIB_TCPRCVQDROP,			/* TCPRcvQDrop */
 	LINUX_MIB_TCPWQUEUETOOBIG,		/* TCPWqueueTooBig */
 	LINUX_MIB_TCPFASTOPENPASSIVEALTKEY,	/* TCPFastOpenPassiveAltKey */
+	LINUX_MIB_TCPTIMEOUTREHASH,		/* TCPTimeoutRehash */
+	LINUX_MIB_TCPDUPLICATEDATAREHASH,	/* TCPDuplicateDataRehash */
 	__LINUX_MIB_MAX
 };
 

include/uapi/linux/tcp.h

···
 	TCP_NLA_DSACK_DUPS,	/* DSACK blocks received */
 	TCP_NLA_REORD_SEEN,	/* reordering events seen */
 	TCP_NLA_SRTT,		/* smoothed RTT in usecs */
+	TCP_NLA_TIMEOUT_REHASH, /* Timeout-triggered rehash attempts */
 };
 
 /* for TCP_MD5SIG socket option */

net/ipv4/proc.c

···
 	SNMP_MIB_ITEM("TCPRcvQDrop", LINUX_MIB_TCPRCVQDROP),
 	SNMP_MIB_ITEM("TCPWqueueTooBig", LINUX_MIB_TCPWQUEUETOOBIG),
 	SNMP_MIB_ITEM("TCPFastOpenPassiveAltKey", LINUX_MIB_TCPFASTOPENPASSIVEALTKEY),
+	SNMP_MIB_ITEM("TcpTimeoutRehash", LINUX_MIB_TCPTIMEOUTREHASH),
+	SNMP_MIB_ITEM("TcpDuplicateDataRehash", LINUX_MIB_TCPDUPLICATEDATAREHASH),
 	SNMP_MIB_SENTINEL
 };
 

net/ipv4/tcp.c

···
 		nla_total_size(sizeof(u32)) + /* TCP_NLA_DSACK_DUPS */
 		nla_total_size(sizeof(u32)) + /* TCP_NLA_REORD_SEEN */
 		nla_total_size(sizeof(u32)) + /* TCP_NLA_SRTT */
+		nla_total_size(sizeof(u16)) + /* TCP_NLA_TIMEOUT_REHASH */
 		0;
 }
 
···
 	nla_put_u32(stats, TCP_NLA_DSACK_DUPS, tp->dsack_dups);
 	nla_put_u32(stats, TCP_NLA_REORD_SEEN, tp->reord_seen);
 	nla_put_u32(stats, TCP_NLA_SRTT, tp->srtt_us >> 3);
+	nla_put_u16(stats, TCP_NLA_TIMEOUT_REHASH, tp->timeout_rehash);
 
 	return stats;
 }

net/ipv4/tcp_input.c

···
 	 * The receiver remembers and reflects via DSACKs. Leverage the
 	 * DSACK state and change the txhash to re-route speculatively.
 	 */
-	if (TCP_SKB_CB(skb)->seq == tcp_sk(sk)->duplicate_sack[0].start_seq)
+	if (TCP_SKB_CB(skb)->seq == tcp_sk(sk)->duplicate_sack[0].start_seq) {
 		sk_rethink_txhash(sk);
+		NET_INC_STATS(sock_net(sk), LINUX_MIB_TCPDUPLICATEDATAREHASH);
+	}
 }
 
 static void tcp_send_dupack(struct sock *sk, const struct sk_buff *skb)

net/ipv4/tcp_timer.c

···
 			dst_negative_advice(sk);
 		} else {
 			sk_rethink_txhash(sk);
+			tp->timeout_rehash++;
+			__NET_INC_STATS(sock_net(sk),
+					LINUX_MIB_TCPTIMEOUTREHASH);
 		}
 		retry_until = icsk->icsk_syn_retries ? : net->ipv4.sysctl_tcp_syn_retries;
 		expired = icsk->icsk_retransmits >= retry_until;
···
 			dst_negative_advice(sk);
 		} else {
 			sk_rethink_txhash(sk);
+			tp->timeout_rehash++;
+			__NET_INC_STATS(sock_net(sk),
+					LINUX_MIB_TCPTIMEOUTREHASH);
 		}
 
 		retry_until = net->ipv4.sysctl_tcp_retries2;