From 58d607d3e52f2b15902f58a1161da9fb3b0f6d47 Mon Sep 17 00:00:00 2001 From: Eric Dumazet Date: Tue, 15 Sep 2015 15:24:20 -0700 Subject: tcp: provide skb->hash to synack packets In commit b73c3d0e4f0e ("net: Save TX flow hash in sock and set in skbuf on xmit"), Tom provided a l4 hash to most outgoing TCP packets. We'd like to provide one as well for SYNACK packets, so that all packets of a given flow share same txhash, to later enable bonding driver to also use skb->hash to perform slave selection. Note that a SYNACK retransmit shuffles the tx hash, as Tom did in commit 265f94ff54d62 ("net: Recompute sk_txhash on negative routing advice") for established sockets. This has nice effect making TCP flows resilient to some kind of black holes, even at connection establish phase. Signed-off-by: Eric Dumazet Cc: Tom Herbert Cc: Mahesh Bandewar Acked-by: Tom Herbert Signed-off-by: David S. Miller --- net/ipv4/tcp_input.c | 1 + 1 file changed, 1 insertion(+) (limited to 'net/ipv4/tcp_input.c') diff --git a/net/ipv4/tcp_input.c b/net/ipv4/tcp_input.c index a8f515bb19c4..a62e9c76d485 100644 --- a/net/ipv4/tcp_input.c +++ b/net/ipv4/tcp_input.c @@ -6228,6 +6228,7 @@ int tcp_conn_request(struct request_sock_ops *rsk_ops, } tcp_rsk(req)->snt_isn = isn; + tcp_rsk(req)->txhash = net_tx_rndhash(); tcp_openreq_init_rwin(req, sk, dst); fastopen = !want_cookie && tcp_try_fastopen(sk, skb, req, &foc, dst); -- cgit From 0f1c28ae74bb1a34d36fca2db5161611d58b3148 Mon Sep 17 00:00:00 2001 From: Yuchung Cheng Date: Fri, 18 Sep 2015 11:36:14 -0700 Subject: tcp: usec resolution SYN/ACK RTT Currently SYN/ACK RTT is measured in jiffies. For LAN the SYN/ACK RTT is often measured as 0ms or sometimes 1ms, which would affect RTT estimation and min RTT samping used by some congestion control. This patch improves SYN/ACK RTT to be usec resolution if platform supports it. While the timestamping of SYN/ACK is done in request sock, the RTT measurement is carefully arranged to avoid storing another u64 timestamp in tcp_sock. For regular handshake w/o SYNACK retransmission, the RTT is sampled right after the child socket is created and right before the request sock is released (tcp_check_req() in tcp_minisocks.c) For Fast Open the child socket is already created when SYN/ACK was sent, the RTT is sampled in tcp_rcv_state_process() after processing the final ACK an right before the request socket is released. If the SYN/ACK was retransmistted or SYN-cookie was used, we rely on TCP timestamps to measure the RTT. The sample is taken at the same place in tcp_rcv_state_process() after the timestamp values are validated in tcp_validate_incoming(). Note that we do not store TS echo value in request_sock for SYN-cookies, because the value is already stored in tp->rx_opt used by tcp_ack_update_rtt(). One side benefit is that the RTT measurement now happens before initializing congestion control (of the passive side). Therefore the congestion control can use the SYN/ACK RTT. Signed-off-by: Yuchung Cheng Signed-off-by: Neal Cardwell Signed-off-by: Eric Dumazet Signed-off-by: David S. Miller --- net/ipv4/tcp_input.c | 29 ++++++++++++++--------------- 1 file changed, 14 insertions(+), 15 deletions(-) (limited to 'net/ipv4/tcp_input.c') diff --git a/net/ipv4/tcp_input.c b/net/ipv4/tcp_input.c index a62e9c76d485..497adf58a6b8 100644 --- a/net/ipv4/tcp_input.c +++ b/net/ipv4/tcp_input.c @@ -2953,21 +2953,21 @@ static inline bool tcp_ack_update_rtt(struct sock *sk, const int flag, } /* Compute time elapsed between (last) SYNACK and the ACK completing 3WHS. */ -static void tcp_synack_rtt_meas(struct sock *sk, const u32 synack_stamp) +void tcp_synack_rtt_meas(struct sock *sk, struct request_sock *req) { - struct tcp_sock *tp = tcp_sk(sk); - long seq_rtt_us = -1L; + long rtt_us = -1L; - if (synack_stamp && !tp->total_retrans) - seq_rtt_us = jiffies_to_usecs(tcp_time_stamp - synack_stamp); + if (req && !req->num_retrans && tcp_rsk(req)->snt_synack.v64) { + struct skb_mstamp now; - /* If the ACK acks both the SYNACK and the (Fast Open'd) data packets - * sent in SYN_RECV, SYNACK RTT is the smooth RTT computed in tcp_ack() - */ - if (!tp->srtt_us) - tcp_ack_update_rtt(sk, FLAG_SYN_ACKED, seq_rtt_us, -1L); + skb_mstamp_get(&now); + rtt_us = skb_mstamp_us_delta(&now, &tcp_rsk(req)->snt_synack); + } + + tcp_ack_update_rtt(sk, FLAG_SYN_ACKED, rtt_us, -1L); } + static void tcp_cong_avoid(struct sock *sk, u32 ack, u32 acked) { const struct inet_connection_sock *icsk = inet_csk(sk); @@ -5706,7 +5706,6 @@ int tcp_rcv_state_process(struct sock *sk, struct sk_buff *skb, struct request_sock *req; int queued = 0; bool acceptable; - u32 synack_stamp; tp->rx_opt.saw_tstamp = 0; @@ -5785,15 +5784,16 @@ int tcp_rcv_state_process(struct sock *sk, struct sk_buff *skb, if (!acceptable) return 1; + if (!tp->srtt_us) + tcp_synack_rtt_meas(sk, req); + /* Once we leave TCP_SYN_RECV, we no longer need req * so release it. */ if (req) { - synack_stamp = tcp_rsk(req)->snt_synack; tp->total_retrans = req->num_retrans; reqsk_fastopen_remove(sk, req, false); } else { - synack_stamp = tp->lsndtime; /* Make sure socket is routed, for correct metrics. */ icsk->icsk_af_ops->rebuild_header(sk); tcp_init_congestion_control(sk); @@ -5816,7 +5816,6 @@ int tcp_rcv_state_process(struct sock *sk, struct sk_buff *skb, tp->snd_una = TCP_SKB_CB(skb)->ack_seq; tp->snd_wnd = ntohs(th->window) << tp->rx_opt.snd_wscale; tcp_init_wl(tp, TCP_SKB_CB(skb)->seq); - tcp_synack_rtt_meas(sk, synack_stamp); if (tp->rx_opt.tstamp_ok) tp->advmss -= TCPOLEN_TSTAMP_ALIGNED; @@ -6027,7 +6026,7 @@ static void tcp_openreq_init(struct request_sock *req, req->cookie_ts = 0; tcp_rsk(req)->rcv_isn = TCP_SKB_CB(skb)->seq; tcp_rsk(req)->rcv_nxt = TCP_SKB_CB(skb)->seq + 1; - tcp_rsk(req)->snt_synack = tcp_time_stamp; + skb_mstamp_get(&tcp_rsk(req)->snt_synack); tcp_rsk(req)->last_oow_ack_time = 0; req->mss = rx_opt->mss_clamp; req->ts_recent = rx_opt->saw_tstamp ? rx_opt->rcv_tsval : 0; -- cgit From 7c85af8810448d8ef59331be51e482413b5f503d Mon Sep 17 00:00:00 2001 From: Eric Dumazet Date: Thu, 24 Sep 2015 17:16:05 -0700 Subject: tcp: avoid reorders for TFO passive connections We found that a TCP Fast Open passive connection was vulnerable to reorders, as the exchange might look like [1] C -> S S [2] S -> C S. ack request [3] S -> C . packets [2] and [3] can be generated at almost the same time. If C receives the 3rd packet before the 2nd, it will drop it as the socket is in SYN_SENT state and expects a SYNACK. S will have to retransmit the answer. Current OOO avoidance in linux is defeated because SYNACK packets are attached to the LISTEN socket, while DATA packets are attached to the children. They might be sent by different cpus, and different TX queues might be selected. It turns out that for TFO, we created a child, which is a full blown socket in TCP_SYN_RECV state, and we simply can attach the SYNACK packet to this socket. This means that at the time tcp_sendmsg() pushes DATA packet, skb->ooo_okay will be set iff the SYNACK packet had been sent and TX completed. This removes the reorder source at the host level. We also removed the export of tcp_try_fastopen(), as it is no longer called from IPv6. Signed-off-by: Eric Dumazet Signed-off-by: Yuchung Cheng Signed-off-by: David S. Miller --- net/ipv4/tcp_input.c | 19 +++++++++++-------- 1 file changed, 11 insertions(+), 8 deletions(-) (limited to 'net/ipv4/tcp_input.c') diff --git a/net/ipv4/tcp_input.c b/net/ipv4/tcp_input.c index 497adf58a6b8..4964d53907e9 100644 --- a/net/ipv4/tcp_input.c +++ b/net/ipv4/tcp_input.c @@ -6111,14 +6111,15 @@ int tcp_conn_request(struct request_sock_ops *rsk_ops, const struct tcp_request_sock_ops *af_ops, struct sock *sk, struct sk_buff *skb) { + struct tcp_fastopen_cookie foc = { .len = -1 }; + __u32 isn = TCP_SKB_CB(skb)->tcp_tw_isn; struct tcp_options_received tmp_opt; - struct request_sock *req; struct tcp_sock *tp = tcp_sk(sk); + struct sock *fastopen_sk = NULL; struct dst_entry *dst = NULL; - __u32 isn = TCP_SKB_CB(skb)->tcp_tw_isn; - bool want_cookie = false, fastopen; + struct request_sock *req; + bool want_cookie = false; struct flowi fl; - struct tcp_fastopen_cookie foc = { .len = -1 }; int err; @@ -6229,11 +6230,13 @@ int tcp_conn_request(struct request_sock_ops *rsk_ops, tcp_rsk(req)->snt_isn = isn; tcp_rsk(req)->txhash = net_tx_rndhash(); tcp_openreq_init_rwin(req, sk, dst); - fastopen = !want_cookie && - tcp_try_fastopen(sk, skb, req, &foc, dst); - err = af_ops->send_synack(sk, dst, &fl, req, + if (!want_cookie) + fastopen_sk = tcp_try_fastopen(sk, skb, req, &foc, dst); + err = af_ops->send_synack(fastopen_sk ?: sk, dst, &fl, req, skb_get_queue_mapping(skb), &foc); - if (!fastopen) { + if (fastopen_sk) { + sock_put(fastopen_sk); + } else { if (err || want_cookie) goto drop_and_free; -- cgit From bda07a64c09c44ced789dbb815c71854f0c59839 Mon Sep 17 00:00:00 2001 From: Eric Dumazet Date: Tue, 29 Sep 2015 07:42:40 -0700 Subject: tcp: remove unused len argument from tcp_rcv_state_process() Once we realize tcp_rcv_synsent_state_process() does not use its 'len' argument and we get rid of it, then it becomes clear this argument is no longer used in tcp_rcv_state_process() Signed-off-by: Eric Dumazet Signed-off-by: David S. Miller --- net/ipv4/tcp_input.c | 6 +++--- 1 file changed, 3 insertions(+), 3 deletions(-) (limited to 'net/ipv4/tcp_input.c') diff --git a/net/ipv4/tcp_input.c b/net/ipv4/tcp_input.c index 4964d53907e9..dcbddf12f4b3 100644 --- a/net/ipv4/tcp_input.c +++ b/net/ipv4/tcp_input.c @@ -5472,7 +5472,7 @@ static bool tcp_rcv_fastopen_synack(struct sock *sk, struct sk_buff *synack, } static int tcp_rcv_synsent_state_process(struct sock *sk, struct sk_buff *skb, - const struct tcphdr *th, unsigned int len) + const struct tcphdr *th) { struct inet_connection_sock *icsk = inet_csk(sk); struct tcp_sock *tp = tcp_sk(sk); @@ -5699,7 +5699,7 @@ reset_and_undo: */ int tcp_rcv_state_process(struct sock *sk, struct sk_buff *skb, - const struct tcphdr *th, unsigned int len) + const struct tcphdr *th) { struct tcp_sock *tp = tcp_sk(sk); struct inet_connection_sock *icsk = inet_csk(sk); @@ -5749,7 +5749,7 @@ int tcp_rcv_state_process(struct sock *sk, struct sk_buff *skb, goto discard; case TCP_SYN_SENT: - queued = tcp_rcv_synsent_state_process(sk, skb, th, len); + queued = tcp_rcv_synsent_state_process(sk, skb, th); if (queued >= 0) return queued; -- cgit From 72ab4a86f7a260d4c2a320b49662da107ce77a81 Mon Sep 17 00:00:00 2001 From: Eric Dumazet Date: Tue, 29 Sep 2015 07:42:41 -0700 Subject: tcp: remove tcp_rcv_state_process() tcp_hdr argument Factorize code to get tcp header from skb. It makes no sense to duplicate code in callers. Signed-off-by: Eric Dumazet Signed-off-by: David S. Miller --- net/ipv4/tcp_input.c | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-) (limited to 'net/ipv4/tcp_input.c') diff --git a/net/ipv4/tcp_input.c b/net/ipv4/tcp_input.c index dcbddf12f4b3..67b27aee8d28 100644 --- a/net/ipv4/tcp_input.c +++ b/net/ipv4/tcp_input.c @@ -5698,11 +5698,11 @@ reset_and_undo: * address independent. */ -int tcp_rcv_state_process(struct sock *sk, struct sk_buff *skb, - const struct tcphdr *th) +int tcp_rcv_state_process(struct sock *sk, struct sk_buff *skb) { struct tcp_sock *tp = tcp_sk(sk); struct inet_connection_sock *icsk = inet_csk(sk); + const struct tcphdr *th = tcp_hdr(skb); struct request_sock *req; int queued = 0; bool acceptable; -- cgit From 2985aaac010ebd5e562ce1a22cc61acbb0e40cf2 Mon Sep 17 00:00:00 2001 From: Eric Dumazet Date: Tue, 29 Sep 2015 07:42:51 -0700 Subject: tcp: constify tcp_syn_flood_action() socket argument tcp_syn_flood_action() will soon be called with unlocked socket. In order to avoid SYN flood warning being emitted multiple times, use xchg(). Extend max_qlen_log and synflood_warned fields in struct listen_sock to u32 Signed-off-by: Eric Dumazet Signed-off-by: David S. Miller --- net/ipv4/tcp_input.c | 9 +++++---- 1 file changed, 5 insertions(+), 4 deletions(-) (limited to 'net/ipv4/tcp_input.c') diff --git a/net/ipv4/tcp_input.c b/net/ipv4/tcp_input.c index 67b27aee8d28..e58cbcd2f07e 100644 --- a/net/ipv4/tcp_input.c +++ b/net/ipv4/tcp_input.c @@ -6064,7 +6064,7 @@ EXPORT_SYMBOL(inet_reqsk_alloc); /* * Return true if a syncookie should be sent */ -static bool tcp_syn_flood_action(struct sock *sk, +static bool tcp_syn_flood_action(const struct sock *sk, const struct sk_buff *skb, const char *proto) { @@ -6082,11 +6082,12 @@ static bool tcp_syn_flood_action(struct sock *sk, NET_INC_STATS_BH(sock_net(sk), LINUX_MIB_TCPREQQFULLDROP); lopt = inet_csk(sk)->icsk_accept_queue.listen_opt; - if (!lopt->synflood_warned && sysctl_tcp_syncookies != 2) { - lopt->synflood_warned = 1; + if (!lopt->synflood_warned && + sysctl_tcp_syncookies != 2 && + xchg(&lopt->synflood_warned, 1) == 0) pr_info("%s: Possible SYN flooding on port %d. %s. Check SNMP counters.\n", proto, ntohs(tcp_hdr(skb)->dest), msg); - } + return want_cookie; } -- cgit From 8d2675f1e464aa5cedda63849adecffd8d33fead Mon Sep 17 00:00:00 2001 From: Eric Dumazet Date: Fri, 2 Oct 2015 11:43:25 -0700 Subject: tcp: move synflood_warned into struct request_sock_queue long term plan is to remove struct listen_sock when its hash table is no longer there. Signed-off-by: Eric Dumazet Signed-off-by: David S. Miller --- net/ipv4/tcp_input.c | 7 +++---- 1 file changed, 3 insertions(+), 4 deletions(-) (limited to 'net/ipv4/tcp_input.c') diff --git a/net/ipv4/tcp_input.c b/net/ipv4/tcp_input.c index e58cbcd2f07e..8b0ce73c2049 100644 --- a/net/ipv4/tcp_input.c +++ b/net/ipv4/tcp_input.c @@ -6068,9 +6068,9 @@ static bool tcp_syn_flood_action(const struct sock *sk, const struct sk_buff *skb, const char *proto) { + struct request_sock_queue *queue = &inet_csk(sk)->icsk_accept_queue; const char *msg = "Dropping request"; bool want_cookie = false; - struct listen_sock *lopt; #ifdef CONFIG_SYN_COOKIES if (sysctl_tcp_syncookies) { @@ -6081,10 +6081,9 @@ static bool tcp_syn_flood_action(const struct sock *sk, #endif NET_INC_STATS_BH(sock_net(sk), LINUX_MIB_TCPREQQFULLDROP); - lopt = inet_csk(sk)->icsk_accept_queue.listen_opt; - if (!lopt->synflood_warned && + if (!queue->synflood_warned && sysctl_tcp_syncookies != 2 && - xchg(&lopt->synflood_warned, 1) == 0) + xchg(&queue->synflood_warned, 1) == 0) pr_info("%s: Possible SYN flooding on port %d. %s. Check SNMP counters.\n", proto, ntohs(tcp_hdr(skb)->dest), msg); -- cgit From 079096f103faca2dd87342cca6f23d4b34da8871 Mon Sep 17 00:00:00 2001 From: Eric Dumazet Date: Fri, 2 Oct 2015 11:43:32 -0700 Subject: tcp/dccp: install syn_recv requests into ehash table In this patch, we insert request sockets into TCP/DCCP regular ehash table (where ESTABLISHED and TIMEWAIT sockets are) instead of using the per listener hash table. ACK packets find SYN_RECV pseudo sockets without having to find and lock the listener. In nominal conditions, this halves pressure on listener lock. Note that this will allow for SO_REUSEPORT refinements, so that we can select a listener using cpu/numa affinities instead of the prior 'consistent hash', since only SYN packets will apply this selection logic. We will shrink listen_sock in the following patch to ease code review. Signed-off-by: Eric Dumazet Cc: Ying Cai Cc: Willem de Bruijn Signed-off-by: David S. Miller --- net/ipv4/tcp_input.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) (limited to 'net/ipv4/tcp_input.c') diff --git a/net/ipv4/tcp_input.c b/net/ipv4/tcp_input.c index 8b0ce73c2049..a56912772354 100644 --- a/net/ipv4/tcp_input.c +++ b/net/ipv4/tcp_input.c @@ -6241,7 +6241,7 @@ int tcp_conn_request(struct request_sock_ops *rsk_ops, goto drop_and_free; tcp_rsk(req)->tfo_listener = false; - af_ops->queue_hash_add(sk, req, TCP_TIMEOUT_INIT); + inet_csk_reqsk_queue_hash_add(sk, req, TCP_TIMEOUT_INIT); } tcp_reqsk_record_syn(sk, req, skb); -- cgit From ca6fb06518836ef9b65dc0aac02ff97704d52a05 Mon Sep 17 00:00:00 2001 From: Eric Dumazet Date: Fri, 2 Oct 2015 11:43:35 -0700 Subject: 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 Signed-off-by: David S. Miller --- net/ipv4/tcp_input.c | 23 ++++++++++++----------- 1 file changed, 12 insertions(+), 11 deletions(-) (limited to 'net/ipv4/tcp_input.c') diff --git a/net/ipv4/tcp_input.c b/net/ipv4/tcp_input.c index a56912772354..27108757c310 100644 --- a/net/ipv4/tcp_input.c +++ b/net/ipv4/tcp_input.c @@ -6120,8 +6120,6 @@ int tcp_conn_request(struct request_sock_ops *rsk_ops, struct request_sock *req; bool want_cookie = false; struct flowi fl; - int err; - /* TW buckets are converted to open requests without * limitations, they conserve resources and peer is @@ -6230,21 +6228,24 @@ int tcp_conn_request(struct request_sock_ops *rsk_ops, tcp_rsk(req)->snt_isn = isn; tcp_rsk(req)->txhash = net_tx_rndhash(); tcp_openreq_init_rwin(req, sk, dst); - if (!want_cookie) + if (!want_cookie) { fastopen_sk = tcp_try_fastopen(sk, skb, req, &foc, dst); - err = af_ops->send_synack(fastopen_sk ?: sk, dst, &fl, req, - skb_get_queue_mapping(skb), &foc); + tcp_reqsk_record_syn(sk, req, skb); + } if (fastopen_sk) { + af_ops->send_synack(fastopen_sk, dst, &fl, req, + skb_get_queue_mapping(skb), &foc, false); sock_put(fastopen_sk); } else { - if (err || want_cookie) - goto drop_and_free; - tcp_rsk(req)->tfo_listener = false; - inet_csk_reqsk_queue_hash_add(sk, req, TCP_TIMEOUT_INIT); + if (!want_cookie) + inet_csk_reqsk_queue_hash_add(sk, req, TCP_TIMEOUT_INIT); + af_ops->send_synack(sk, dst, &fl, req, + skb_get_queue_mapping(skb), &foc, !want_cookie); + if (want_cookie) + goto drop_and_free; } - tcp_reqsk_record_syn(sk, req, skb); - + reqsk_put(req); return 0; drop_and_release: -- cgit From 7656d842de93fd2d2de7b403062cad757cadf1df Mon Sep 17 00:00:00 2001 From: Eric Dumazet Date: Sun, 4 Oct 2015 21:08:07 -0700 Subject: tcp: fix fastopen races vs lockless listener There are multiple races that need fixes : 1) skb_get() + queue skb + kfree_skb() is racy An accept() can be done on another cpu, data consumed immediately. tcp_recvmsg() uses __kfree_skb() as it is assumed all skb found in socket receive queue are private. Then the kfree_skb() in tcp_rcv_state_process() uses an already freed skb 2) tcp_reqsk_record_syn() needs to be done before tcp_try_fastopen() for the same reasons. 3) We want to send the SYNACK before queueing child into accept queue, otherwise we might reintroduce the ooo issue fixed in commit 7c85af881044 ("tcp: avoid reorders for TFO passive connections") Signed-off-by: Eric Dumazet Signed-off-by: David S. Miller --- net/ipv4/tcp_input.c | 6 +++++- 1 file changed, 5 insertions(+), 1 deletion(-) (limited to 'net/ipv4/tcp_input.c') diff --git a/net/ipv4/tcp_input.c b/net/ipv4/tcp_input.c index 27108757c310..a95c8eb04ff7 100644 --- a/net/ipv4/tcp_input.c +++ b/net/ipv4/tcp_input.c @@ -6229,12 +6229,16 @@ int tcp_conn_request(struct request_sock_ops *rsk_ops, tcp_rsk(req)->txhash = net_tx_rndhash(); tcp_openreq_init_rwin(req, sk, dst); if (!want_cookie) { - fastopen_sk = tcp_try_fastopen(sk, skb, req, &foc, dst); tcp_reqsk_record_syn(sk, req, skb); + fastopen_sk = tcp_try_fastopen(sk, skb, req, &foc, dst); } if (fastopen_sk) { af_ops->send_synack(fastopen_sk, dst, &fl, req, skb_get_queue_mapping(skb), &foc, false); + /* Add the child socket directly into the accept queue */ + inet_csk_reqsk_queue_add(sk, req, fastopen_sk); + sk->sk_data_ready(sk); + bh_unlock_sock(fastopen_sk); sock_put(fastopen_sk); } else { tcp_rsk(req)->tfo_listener = false; -- cgit From a1a5344ddbe8fd3e080013b317ac9a664490cfdf Mon Sep 17 00:00:00 2001 From: Eric Dumazet Date: Sun, 4 Oct 2015 21:08:11 -0700 Subject: tcp: avoid two atomic ops for syncookies inet_reqsk_alloc() is used to allocate a temporary request in order to generate a SYNACK with a cookie. Then later, syncookie validation also uses a temporary request. These paths already took a reference on listener refcount, we can avoid a couple of atomic operations. Signed-off-by: Eric Dumazet Signed-off-by: David S. Miller --- net/ipv4/tcp_input.c | 8 +++++--- 1 file changed, 5 insertions(+), 3 deletions(-) (limited to 'net/ipv4/tcp_input.c') diff --git a/net/ipv4/tcp_input.c b/net/ipv4/tcp_input.c index a95c8eb04ff7..ddadb318e850 100644 --- a/net/ipv4/tcp_input.c +++ b/net/ipv4/tcp_input.c @@ -6042,9 +6042,11 @@ static void tcp_openreq_init(struct request_sock *req, } struct request_sock *inet_reqsk_alloc(const struct request_sock_ops *ops, - struct sock *sk_listener) + struct sock *sk_listener, + bool attach_listener) { - struct request_sock *req = reqsk_alloc(ops, sk_listener); + struct request_sock *req = reqsk_alloc(ops, sk_listener, + attach_listener); if (req) { struct inet_request_sock *ireq = inet_rsk(req); @@ -6143,7 +6145,7 @@ int tcp_conn_request(struct request_sock_ops *rsk_ops, goto drop; } - req = inet_reqsk_alloc(rsk_ops, sk); + req = inet_reqsk_alloc(rsk_ops, sk, !want_cookie); if (!req) goto drop; -- cgit From ed53d0ab761f5c71d77c8dc05fd19c0a851200db Mon Sep 17 00:00:00 2001 From: Eric Dumazet Date: Thu, 8 Oct 2015 19:33:23 -0700 Subject: net: shrink struct sock and request_sock by 8 bytes One 32bit hole is following skc_refcnt, use it. skc_incoming_cpu can also be an union for request_sock rcv_wnd. Signed-off-by: Eric Dumazet Signed-off-by: David S. Miller --- net/ipv4/tcp_input.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) (limited to 'net/ipv4/tcp_input.c') diff --git a/net/ipv4/tcp_input.c b/net/ipv4/tcp_input.c index ddadb318e850..3b35c3f4d268 100644 --- a/net/ipv4/tcp_input.c +++ b/net/ipv4/tcp_input.c @@ -6022,7 +6022,7 @@ static void tcp_openreq_init(struct request_sock *req, { struct inet_request_sock *ireq = inet_rsk(req); - req->rcv_wnd = 0; /* So that tcp_send_synack() knows! */ + req->rsk_rcv_wnd = 0; /* So that tcp_send_synack() knows! */ req->cookie_ts = 0; tcp_rsk(req)->rcv_isn = TCP_SKB_CB(skb)->seq; tcp_rsk(req)->rcv_nxt = TCP_SKB_CB(skb)->seq + 1; -- cgit From dc6ef6be52154490c5c03f742e28bc781cc751b2 Mon Sep 17 00:00:00 2001 From: Eric Dumazet Date: Fri, 16 Oct 2015 13:00:01 -0700 Subject: tcp: do not set queue_mapping on SYNACK At the time of commit fff326990789 ("tcp: reflect SYN queue_mapping into SYNACK packets") we had little ways to cope with SYN floods. We no longer need to reflect incoming skb queue mappings, and instead can pick a TX queue based on cpu cooking the SYNACK, with normal XPS affinities. Note that all SYNACK retransmits were picking TX queue 0, this no longer is a win given that SYNACK rtx are now distributed on all cpus. Signed-off-by: Eric Dumazet Signed-off-by: David S. Miller --- net/ipv4/tcp_input.c | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-) (limited to 'net/ipv4/tcp_input.c') diff --git a/net/ipv4/tcp_input.c b/net/ipv4/tcp_input.c index 3b35c3f4d268..944eaca69115 100644 --- a/net/ipv4/tcp_input.c +++ b/net/ipv4/tcp_input.c @@ -6236,7 +6236,7 @@ int tcp_conn_request(struct request_sock_ops *rsk_ops, } if (fastopen_sk) { af_ops->send_synack(fastopen_sk, dst, &fl, req, - skb_get_queue_mapping(skb), &foc, false); + &foc, false); /* Add the child socket directly into the accept queue */ inet_csk_reqsk_queue_add(sk, req, fastopen_sk); sk->sk_data_ready(sk); @@ -6247,7 +6247,7 @@ int tcp_conn_request(struct request_sock_ops *rsk_ops, if (!want_cookie) inet_csk_reqsk_queue_hash_add(sk, req, TCP_TIMEOUT_INIT); af_ops->send_synack(sk, dst, &fl, req, - skb_get_queue_mapping(skb), &foc, !want_cookie); + &foc, !want_cookie); if (want_cookie) goto drop_and_free; } -- cgit From 9e45a3e36b363cc4c79c70f2b4f994e66543a219 Mon Sep 17 00:00:00 2001 From: Yuchung Cheng Date: Fri, 16 Oct 2015 21:57:41 -0700 Subject: tcp: apply Kern's check on RTTs used for congestion control Currently ca_seq_rtt_us does not use Kern's check. Fix that by checking if any packet acked is a retransmit, for both RTT used for RTT estimation and congestion control. Fixes: 5b08e47ca ("tcp: prefer packet timing to TS-ECR for RTT") Signed-off-by: Yuchung Cheng Signed-off-by: Neal Cardwell Signed-off-by: Eric Dumazet Signed-off-by: David S. Miller --- net/ipv4/tcp_input.c | 5 +---- 1 file changed, 1 insertion(+), 4 deletions(-) (limited to 'net/ipv4/tcp_input.c') diff --git a/net/ipv4/tcp_input.c b/net/ipv4/tcp_input.c index 944eaca69115..62ee71efd1ce 100644 --- a/net/ipv4/tcp_input.c +++ b/net/ipv4/tcp_input.c @@ -2925,9 +2925,6 @@ static inline bool tcp_ack_update_rtt(struct sock *sk, const int flag, * Karn's algorithm forbids taking RTT if some retransmitted data * is acked (RFC6298). */ - if (flag & FLAG_RETRANS_DATA_ACKED) - seq_rtt_us = -1L; - if (seq_rtt_us < 0) seq_rtt_us = sack_rtt_us; @@ -3169,7 +3166,7 @@ static int tcp_clean_rtx_queue(struct sock *sk, int prior_fackets, flag |= FLAG_SACK_RENEGING; skb_mstamp_get(&now); - if (likely(first_ackt.v64)) { + if (likely(first_ackt.v64) && !(flag & FLAG_RETRANS_DATA_ACKED)) { seq_rtt_us = skb_mstamp_us_delta(&now, &first_ackt); ca_rtt_us = skb_mstamp_us_delta(&now, &last_ackt); } -- cgit From f672258391b42a5c7cc2732c9c063e56a85c8dbe Mon Sep 17 00:00:00 2001 From: Yuchung Cheng Date: Fri, 16 Oct 2015 21:57:42 -0700 Subject: tcp: track min RTT using windowed min-filter Kathleen Nichols' algorithm for tracking the minimum RTT of a data stream over some measurement window. It uses constant space and constant time per update. Yet it almost always delivers the same minimum as an implementation that has to keep all the data in the window. The measurement window is tunable via sysctl.net.ipv4.tcp_min_rtt_wlen with a default value of 5 minutes. The algorithm keeps track of the best, 2nd best & 3rd best min values, maintaining an invariant that the measurement time of the n'th best >= n-1'th best. It also makes sure that the three values are widely separated in the time window since that bounds the worse case error when that data is monotonically increasing over the window. Upon getting a new min, we can forget everything earlier because it has no value - the new min is less than everything else in the window by definition and it's the most recent. So we restart fresh on every new min and overwrites the 2nd & 3rd choices. The same property holds for the 2nd & 3rd best. Therefore we have to maintain two invariants to maximize the information in the samples, one on values (1st.v <= 2nd.v <= 3rd.v) and the other on times (now-win <=1st.t <= 2nd.t <= 3rd.t <= now). These invariants determine the structure of the code The RTT input to the windowed filter is the minimum RTT measured from ACK or SACK, or as the last resort from TCP timestamps. The accessor tcp_min_rtt() returns the minimum RTT seen in the window. ~0U indicates it is not available. The minimum is 1usec even if the true RTT is below that. Signed-off-by: Yuchung Cheng Signed-off-by: Neal Cardwell Signed-off-by: Eric Dumazet Signed-off-by: David S. Miller --- net/ipv4/tcp_input.c | 78 ++++++++++++++++++++++++++++++++++++++++++++++++---- 1 file changed, 73 insertions(+), 5 deletions(-) (limited to 'net/ipv4/tcp_input.c') diff --git a/net/ipv4/tcp_input.c b/net/ipv4/tcp_input.c index 62ee71efd1ce..eedb25db3947 100644 --- a/net/ipv4/tcp_input.c +++ b/net/ipv4/tcp_input.c @@ -95,6 +95,7 @@ int sysctl_tcp_stdurg __read_mostly; int sysctl_tcp_rfc1337 __read_mostly; int sysctl_tcp_max_orphans __read_mostly = NR_FILE; int sysctl_tcp_frto __read_mostly = 2; +int sysctl_tcp_min_rtt_wlen __read_mostly = 300; int sysctl_tcp_thin_dupack __read_mostly; @@ -2915,8 +2916,69 @@ static void tcp_fastretrans_alert(struct sock *sk, const int acked, tcp_xmit_retransmit_queue(sk); } +/* Kathleen Nichols' algorithm for tracking the minimum value of + * a data stream over some fixed time interval. (E.g., the minimum + * RTT over the past five minutes.) It uses constant space and constant + * time per update yet almost always delivers the same minimum as an + * implementation that has to keep all the data in the window. + * + * The algorithm keeps track of the best, 2nd best & 3rd best min + * values, maintaining an invariant that the measurement time of the + * n'th best >= n-1'th best. It also makes sure that the three values + * are widely separated in the time window since that bounds the worse + * case error when that data is monotonically increasing over the window. + * + * Upon getting a new min, we can forget everything earlier because it + * has no value - the new min is <= everything else in the window by + * definition and it's the most recent. So we restart fresh on every new min + * and overwrites 2nd & 3rd choices. The same property holds for 2nd & 3rd + * best. + */ +static void tcp_update_rtt_min(struct sock *sk, u32 rtt_us) +{ + const u32 now = tcp_time_stamp, wlen = sysctl_tcp_min_rtt_wlen * HZ; + struct rtt_meas *m = tcp_sk(sk)->rtt_min; + struct rtt_meas rttm = { .rtt = (rtt_us ? : 1), .ts = now }; + u32 elapsed; + + /* Check if the new measurement updates the 1st, 2nd, or 3rd choices */ + if (unlikely(rttm.rtt <= m[0].rtt)) + m[0] = m[1] = m[2] = rttm; + else if (rttm.rtt <= m[1].rtt) + m[1] = m[2] = rttm; + else if (rttm.rtt <= m[2].rtt) + m[2] = rttm; + + elapsed = now - m[0].ts; + if (unlikely(elapsed > wlen)) { + /* Passed entire window without a new min so make 2nd choice + * the new min & 3rd choice the new 2nd. So forth and so on. + */ + m[0] = m[1]; + m[1] = m[2]; + m[2] = rttm; + if (now - m[0].ts > wlen) { + m[0] = m[1]; + m[1] = rttm; + if (now - m[0].ts > wlen) + m[0] = rttm; + } + } else if (m[1].ts == m[0].ts && elapsed > wlen / 4) { + /* Passed a quarter of the window without a new min so + * take 2nd choice from the 2nd quarter of the window. + */ + m[2] = m[1] = rttm; + } else if (m[2].ts == m[1].ts && elapsed > wlen / 2) { + /* Passed half the window without a new min so take the 3rd + * choice from the last half of the window. + */ + m[2] = rttm; + } +} + static inline bool tcp_ack_update_rtt(struct sock *sk, const int flag, - long seq_rtt_us, long sack_rtt_us) + long seq_rtt_us, long sack_rtt_us, + long ca_rtt_us) { const struct tcp_sock *tp = tcp_sk(sk); @@ -2936,11 +2998,16 @@ static inline bool tcp_ack_update_rtt(struct sock *sk, const int flag, */ if (seq_rtt_us < 0 && tp->rx_opt.saw_tstamp && tp->rx_opt.rcv_tsecr && flag & FLAG_ACKED) - seq_rtt_us = jiffies_to_usecs(tcp_time_stamp - tp->rx_opt.rcv_tsecr); - + seq_rtt_us = ca_rtt_us = jiffies_to_usecs(tcp_time_stamp - + tp->rx_opt.rcv_tsecr); if (seq_rtt_us < 0) return false; + /* ca_rtt_us >= 0 is counting on the invariant that ca_rtt_us is + * always taken together with ACK, SACK, or TS-opts. Any negative + * values will be skipped with the seq_rtt_us < 0 check above. + */ + tcp_update_rtt_min(sk, ca_rtt_us); tcp_rtt_estimator(sk, seq_rtt_us); tcp_set_rto(sk); @@ -2961,7 +3028,7 @@ void tcp_synack_rtt_meas(struct sock *sk, struct request_sock *req) rtt_us = skb_mstamp_us_delta(&now, &tcp_rsk(req)->snt_synack); } - tcp_ack_update_rtt(sk, FLAG_SYN_ACKED, rtt_us, -1L); + tcp_ack_update_rtt(sk, FLAG_SYN_ACKED, rtt_us, -1L, rtt_us); } @@ -3175,7 +3242,8 @@ static int tcp_clean_rtx_queue(struct sock *sk, int prior_fackets, ca_rtt_us = skb_mstamp_us_delta(&now, &sack->last_sackt); } - rtt_update = tcp_ack_update_rtt(sk, flag, seq_rtt_us, sack_rtt_us); + rtt_update = tcp_ack_update_rtt(sk, flag, seq_rtt_us, sack_rtt_us, + ca_rtt_us); if (flag & FLAG_ACKED) { tcp_rearm_rto(sk); -- cgit From af82f4e84866ecd360a53f770d6217637116e6c1 Mon Sep 17 00:00:00 2001 From: Yuchung Cheng Date: Fri, 16 Oct 2015 21:57:43 -0700 Subject: tcp: remove tcp_mark_lost_retrans() Remove the existing lost retransmit detection because RACK subsumes it completely. This also stops the overloading the ack_seq field of the skb control block. Signed-off-by: Yuchung Cheng Signed-off-by: Neal Cardwell Signed-off-by: Eric Dumazet Signed-off-by: David S. Miller --- net/ipv4/tcp_input.c | 65 ---------------------------------------------------- 1 file changed, 65 deletions(-) (limited to 'net/ipv4/tcp_input.c') diff --git a/net/ipv4/tcp_input.c b/net/ipv4/tcp_input.c index eedb25db3947..5a776897a8c7 100644 --- a/net/ipv4/tcp_input.c +++ b/net/ipv4/tcp_input.c @@ -1048,70 +1048,6 @@ static bool tcp_is_sackblock_valid(struct tcp_sock *tp, bool is_dsack, return !before(start_seq, end_seq - tp->max_window); } -/* Check for lost retransmit. This superb idea is borrowed from "ratehalving". - * Event "B". Later note: FACK people cheated me again 8), we have to account - * for reordering! Ugly, but should help. - * - * Search retransmitted skbs from write_queue that were sent when snd_nxt was - * less than what is now known to be received by the other end (derived from - * highest SACK block). Also calculate the lowest snd_nxt among the remaining - * retransmitted skbs to avoid some costly processing per ACKs. - */ -static void tcp_mark_lost_retrans(struct sock *sk, int *flag) -{ - const struct inet_connection_sock *icsk = inet_csk(sk); - struct tcp_sock *tp = tcp_sk(sk); - struct sk_buff *skb; - int cnt = 0; - u32 new_low_seq = tp->snd_nxt; - u32 received_upto = tcp_highest_sack_seq(tp); - - if (!tcp_is_fack(tp) || !tp->retrans_out || - !after(received_upto, tp->lost_retrans_low) || - icsk->icsk_ca_state != TCP_CA_Recovery) - return; - - tcp_for_write_queue(skb, sk) { - u32 ack_seq = TCP_SKB_CB(skb)->ack_seq; - - if (skb == tcp_send_head(sk)) - break; - if (cnt == tp->retrans_out) - break; - if (!after(TCP_SKB_CB(skb)->end_seq, tp->snd_una)) - continue; - - if (!(TCP_SKB_CB(skb)->sacked & TCPCB_SACKED_RETRANS)) - continue; - - /* TODO: We would like to get rid of tcp_is_fack(tp) only - * constraint here (see above) but figuring out that at - * least tp->reordering SACK blocks reside between ack_seq - * and received_upto is not easy task to do cheaply with - * the available datastructures. - * - * Whether FACK should check here for tp->reordering segs - * in-between one could argue for either way (it would be - * rather simple to implement as we could count fack_count - * during the walk and do tp->fackets_out - fack_count). - */ - if (after(received_upto, ack_seq)) { - TCP_SKB_CB(skb)->sacked &= ~TCPCB_SACKED_RETRANS; - tp->retrans_out -= tcp_skb_pcount(skb); - *flag |= FLAG_LOST_RETRANS; - tcp_skb_mark_lost_uncond_verify(tp, skb); - NET_INC_STATS_BH(sock_net(sk), LINUX_MIB_TCPLOSTRETRANSMIT); - } else { - if (before(ack_seq, new_low_seq)) - new_low_seq = ack_seq; - cnt += tcp_skb_pcount(skb); - } - } - - if (tp->retrans_out) - tp->lost_retrans_low = new_low_seq; -} - static bool tcp_check_dsack(struct sock *sk, const struct sk_buff *ack_skb, struct tcp_sack_block_wire *sp, int num_sacks, u32 prior_snd_una) @@ -1838,7 +1774,6 @@ advance_sp: ((inet_csk(sk)->icsk_ca_state != TCP_CA_Loss) || tp->undo_marker)) tcp_update_reordering(sk, tp->fackets_out - state->reord, 0); - tcp_mark_lost_retrans(sk, &state->flag); tcp_verify_left_out(tp); out: -- cgit From 77c631273dc1305a89698929decafa6e43bea645 Mon Sep 17 00:00:00 2001 From: Yuchung Cheng Date: Fri, 16 Oct 2015 21:57:44 -0700 Subject: tcp: add tcp_tsopt_ecr_before helper a helper to prepare the main RACK patch Signed-off-by: Yuchung Cheng Signed-off-by: Neal Cardwell Signed-off-by: Eric Dumazet Signed-off-by: David S. Miller --- net/ipv4/tcp_input.c | 9 +++++++-- 1 file changed, 7 insertions(+), 2 deletions(-) (limited to 'net/ipv4/tcp_input.c') diff --git a/net/ipv4/tcp_input.c b/net/ipv4/tcp_input.c index 5a776897a8c7..1e97e73e5ecf 100644 --- a/net/ipv4/tcp_input.c +++ b/net/ipv4/tcp_input.c @@ -2250,14 +2250,19 @@ static inline void tcp_moderate_cwnd(struct tcp_sock *tp) tp->snd_cwnd_stamp = tcp_time_stamp; } +static bool tcp_tsopt_ecr_before(const struct tcp_sock *tp, u32 when) +{ + return tp->rx_opt.saw_tstamp && tp->rx_opt.rcv_tsecr && + before(tp->rx_opt.rcv_tsecr, when); +} + /* Nothing was retransmitted or returned timestamp is less * than timestamp of the first retransmission. */ static inline bool tcp_packet_delayed(const struct tcp_sock *tp) { return !tp->retrans_stamp || - (tp->rx_opt.saw_tstamp && tp->rx_opt.rcv_tsecr && - before(tp->rx_opt.rcv_tsecr, tp->retrans_stamp)); + tcp_tsopt_ecr_before(tp, tp->retrans_stamp); } /* Undo procedures. */ -- cgit From 659a8ad56f490279f0efee43a62ffa1ac914a4e0 Mon Sep 17 00:00:00 2001 From: Yuchung Cheng Date: Fri, 16 Oct 2015 21:57:46 -0700 Subject: tcp: track the packet timings in RACK This patch is the first half of the RACK loss recovery. RACK loss recovery uses the notion of time instead of packet sequence (FACK) or counts (dupthresh). It's inspired by the previous FACK heuristic in tcp_mark_lost_retrans(): when a limited transmit (new data packet) is sacked, then current retransmitted sequence below the newly sacked sequence must been lost, since at least one round trip time has elapsed. But it has several limitations: 1) can't detect tail drops since it depends on limited transmit 2) is disabled upon reordering (assumes no reordering) 3) only enabled in fast recovery ut not timeout recovery RACK (Recently ACK) addresses these limitations with the notion of time instead: a packet P1 is lost if a later packet P2 is s/acked, as at least one round trip has passed. Since RACK cares about the time sequence instead of the data sequence of packets, it can detect tail drops when later retransmission is s/acked while FACK or dupthresh can't. For reordering RACK uses a dynamically adjusted reordering window ("reo_wnd") to reduce false positives on ever (small) degree of reordering. This patch implements tcp_advanced_rack() which tracks the most recent transmission time among the packets that have been delivered (ACKed or SACKed) in tp->rack.mstamp. This timestamp is the key to determine which packet has been lost. Consider an example that the sender sends six packets: T1: P1 (lost) T2: P2 T3: P3 T4: P4 T100: sack of P2. rack.mstamp = T2 T101: retransmit P1 T102: sack of P2,P3,P4. rack.mstamp = T4 T205: ACK of P4 since the hole is repaired. rack.mstamp = T101 We need to be careful about spurious retransmission because it may falsely advance tp->rack.mstamp by an RTT or an RTO, causing RACK to falsely mark all packets lost, just like a spurious timeout. We identify spurious retransmission by the ACK's TS echo value. If TS option is not applicable but the retransmission is acknowledged less than min-RTT ago, it is likely to be spurious. We refrain from using the transmission time of these spurious retransmissions. The second half is implemented in the next patch that marks packet lost using RACK timestamp. Signed-off-by: Yuchung Cheng Signed-off-by: Neal Cardwell Signed-off-by: Eric Dumazet Signed-off-by: David S. Miller --- net/ipv4/tcp_input.c | 14 ++++++++++++++ 1 file changed, 14 insertions(+) (limited to 'net/ipv4/tcp_input.c') diff --git a/net/ipv4/tcp_input.c b/net/ipv4/tcp_input.c index 1e97e73e5ecf..ce8370525832 100644 --- a/net/ipv4/tcp_input.c +++ b/net/ipv4/tcp_input.c @@ -1173,6 +1173,8 @@ static u8 tcp_sacktag_one(struct sock *sk, return sacked; if (!(sacked & TCPCB_SACKED_ACKED)) { + tcp_rack_advance(tp, xmit_time, sacked); + if (sacked & TCPCB_SACKED_RETRANS) { /* If the segment is not tagged as lost, * we do not clear RETRANS, believing @@ -2256,6 +2258,16 @@ static bool tcp_tsopt_ecr_before(const struct tcp_sock *tp, u32 when) before(tp->rx_opt.rcv_tsecr, when); } +/* skb is spurious retransmitted if the returned timestamp echo + * reply is prior to the skb transmission time + */ +static bool tcp_skb_spurious_retrans(const struct tcp_sock *tp, + const struct sk_buff *skb) +{ + return (TCP_SKB_CB(skb)->sacked & TCPCB_RETRANS) && + tcp_tsopt_ecr_before(tp, tcp_skb_timestamp(skb)); +} + /* Nothing was retransmitted or returned timestamp is less * than timestamp of the first retransmission. */ @@ -3135,6 +3147,8 @@ static int tcp_clean_rtx_queue(struct sock *sk, int prior_fackets, if (sacked & TCPCB_SACKED_ACKED) tp->sacked_out -= acked_pcount; + else if (tcp_is_sack(tp) && !tcp_skb_spurious_retrans(tp, skb)) + tcp_rack_advance(tp, &skb->skb_mstamp, sacked); if (sacked & TCPCB_LOST) tp->lost_out -= acked_pcount; -- cgit From 4f41b1c58a32537542f14c1150099131613a5e8a Mon Sep 17 00:00:00 2001 From: Yuchung Cheng Date: Fri, 16 Oct 2015 21:57:47 -0700 Subject: tcp: use RACK to detect losses This patch implements the second half of RACK that uses the the most recent transmit time among all delivered packets to detect losses. tcp_rack_mark_lost() is called upon receiving a dubious ACK. It then checks if an not-yet-sacked packet was sent at least "reo_wnd" prior to the sent time of the most recently delivered. If so the packet is deemed lost. The "reo_wnd" reordering window starts with 1msec for fast loss detection and changes to min-RTT/4 when reordering is observed. We found 1msec accommodates well on tiny degree of reordering (<3 pkts) on faster links. We use min-RTT instead of SRTT because reordering is more of a path property but SRTT can be inflated by self-inflicated congestion. The factor of 4 is borrowed from the delayed early retransmit and seems to work reasonably well. Since RACK is still experimental, it is now used as a supplemental loss detection on top of existing algorithms. It is only effective after the fast recovery starts or after the timeout occurs. The fast recovery is still triggered by FACK and/or dupack threshold instead of RACK. We introduce a new sysctl net.ipv4.tcp_recovery for future experiments of loss recoveries. For now RACK can be disabled by setting it to 0. Signed-off-by: Yuchung Cheng Signed-off-by: Neal Cardwell Signed-off-by: Eric Dumazet Signed-off-by: David S. Miller --- net/ipv4/tcp_input.c | 9 +++++++-- 1 file changed, 7 insertions(+), 2 deletions(-) (limited to 'net/ipv4/tcp_input.c') diff --git a/net/ipv4/tcp_input.c b/net/ipv4/tcp_input.c index ce8370525832..fdd88c3803a6 100644 --- a/net/ipv4/tcp_input.c +++ b/net/ipv4/tcp_input.c @@ -881,6 +881,7 @@ static void tcp_update_reordering(struct sock *sk, const int metric, if (metric > 0) tcp_disable_early_retrans(tp); + tp->rack.reord = 1; } /* This must be called before lost_out is incremented */ @@ -906,8 +907,7 @@ static void tcp_skb_mark_lost(struct tcp_sock *tp, struct sk_buff *skb) } } -static void tcp_skb_mark_lost_uncond_verify(struct tcp_sock *tp, - struct sk_buff *skb) +void tcp_skb_mark_lost_uncond_verify(struct tcp_sock *tp, struct sk_buff *skb) { tcp_verify_retransmit_hint(tp, skb); @@ -2806,6 +2806,11 @@ static void tcp_fastretrans_alert(struct sock *sk, const int acked, } } + /* Use RACK to detect loss */ + if (sysctl_tcp_recovery & TCP_RACK_LOST_RETRANS && + tcp_rack_mark_lost(sk)) + flag |= FLAG_LOST_RETRANS; + /* E. Process state. */ switch (icsk->icsk_ca_state) { case TCP_CA_Recovery: -- cgit