leandrof/linux
1
0
Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity
9aee400061
linux/net/ipv4
History
Lawrence Brakmo 9aee400061 tcp: ack immediately when a cwr packet arrives 
We observed high 99 and 99.9% latencies when doing RPCs with DCTCP. The
problem is triggered when the last packet of a request arrives CE
marked. The reply will carry the ECE mark causing TCP to shrink its cwnd
to 1 (because there are no packets in flight). When the 1st packet of
the next request arrives, the ACK was sometimes delayed even though it
is CWR marked, adding up to 40ms to the RPC latency.

This patch insures that CWR marked data packets arriving will be acked
immediately.

Packetdrill script to reproduce the problem:

0.000 socket(..., SOCK_STREAM, IPPROTO_TCP) = 3
0.000 setsockopt(3, SOL_SOCKET, SO_REUSEADDR, [1], 4) = 0
0.000 setsockopt(3, SOL_TCP, TCP_CONGESTION, "dctcp", 5) = 0
0.000 bind(3, ..., ...) = 0
0.000 listen(3, 1) = 0

0.100 < [ect0] SEW 0:0(0) win 32792 <mss 1000,sackOK,nop,nop,nop,wscale 7>
0.100 > SE. 0:0(0) ack 1 <mss 1460,nop,nop,sackOK,nop,wscale 8>
0.110 < [ect0] . 1:1(0) ack 1 win 257
0.200 accept(3, ..., ...) = 4

0.200 < [ect0] . 1:1001(1000) ack 1 win 257
0.200 > [ect01] . 1:1(0) ack 1001

0.200 write(4, ..., 1) = 1
0.200 > [ect01] P. 1:2(1) ack 1001

0.200 < [ect0] . 1001:2001(1000) ack 2 win 257
0.200 write(4, ..., 1) = 1
0.200 > [ect01] P. 2:3(1) ack 2001

0.200 < [ect0] . 2001:3001(1000) ack 3 win 257
0.200 < [ect0] . 3001:4001(1000) ack 3 win 257
0.200 > [ect01] . 3:3(0) ack 4001

0.210 < [ce] P. 4001:4501(500) ack 3 win 257

+0.001 read(4, ..., 4500) = 4500
+0 write(4, ..., 1) = 1
+0 > [ect01] PE. 3:4(1) ack 4501

+0.010 < [ect0] W. 4501:5501(1000) ack 4 win 257
// Previously the ACK sequence below would be 4501, causing a long RTO
+0.040~+0.045 > [ect01] . 4:4(0) ack 5501   // delayed ack

+0.311 < [ect0] . 5501:6501(1000) ack 4 win 257  // More data
+0 > [ect01] . 4:4(0) ack 6501     // now acks everything

+0.500 < F. 9501:9501(0) ack 4 win 257

Modified based on comments by Neal Cardwell <ncardwell@google.com>

Signed-off-by: Lawrence Brakmo <brakmo@fb.com>
Acked-by: Neal Cardwell <ncardwell@google.com>
Acked-by: Yuchung Cheng <ycheng@google.com>
Signed-off-by: Eric Dumazet <edumazet@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2018-07-25 16:20:35 -07:00
..
bpfilter net: bpfilter: make function bpfilter_mbox_request() static 2018-05-29 09:51:44 -04:00
netfilter netfilter: nf_tproxy: fix possible non-linear access to transport header 2018-07-06 14:32:44 +02:00
af_inet.c Revert changes to convert to ->poll_mask() and aio IOCB_CMD_POLL 2018-06-28 10:40:47 -07:00
ah4.c net: use -ENOSPC for transient busy indication 2017-11-03 22:11:17 +08:00
arp.c proc: introduce proc_create_net{,_data} 2018-05-16 07:24:30 +02:00
cipso_ipv4.c tcp/dccp: fix ireq->opt races 2017-10-21 01:33:19 +01:00
datagram.c
devinet.c net/ipv4: Add support for specifying metric of connected routes 2018-05-29 10:12:45 -04:00
esp4_offload.c esp: check the NETIF_F_HW_ESP_TX_CSUM bit before segmenting 2018-02-27 10:46:01 +01:00
esp4.c esp4: remove redundant initialization of pointer esph 2018-02-13 13:59:03 +01:00
fib_frontend.c net/ipv4: Set oif in fib_compute_spec_dst 2018-07-08 10:54:58 +09:00
fib_lookup.h License cleanup: add SPDX GPL-2.0 license identifier to files with no license 2017-11-02 11:10:55 +01:00
fib_notifier.c License cleanup: add SPDX GPL-2.0 license identifier to files with no license 2017-11-02 11:10:55 +01:00
fib_rules.c net: fib_rules: add extack support 2018-04-23 10:21:24 -04:00
fib_semantics.c net: metrics: add proper netlink validation 2018-06-05 12:29:43 -04:00
fib_trie.c Merge git://git.kernel.org/pub/scm/linux/kernel/git/davem/net-next 2018-06-06 18:39:49 -07:00
fou.c net: fix use-after-free in GRO with ESP 2018-07-02 20:34:04 +09:00
gre_demux.c
gre_offload.c net: fix use-after-free in GRO with ESP 2018-07-02 20:34:04 +09:00
icmp.c net: Drop pernet_operations::async 2018-03-27 13:18:09 -04:00
igmp.c multicast: do not restore deleted record source filter mode to new one 2018-07-21 22:58:17 -07:00
inet_connection_sock.c net: ipv4: remove define INET_CSK_DEBUG and unnecessary EXPORT_SYMBOL 2018-05-10 17:43:55 -04:00
inet_diag.c sock_diag: request _diag module only when the family or proto has been registered 2018-03-12 11:03:42 -04:00
inet_fragment.c ipfrag: really prevent allocation on netns exit 2018-07-08 13:05:33 +09:00
inet_hashtables.c net/tcp: Fix socket lookups with SO_BINDTODEVICE 2018-06-20 08:03:06 +09:00
inet_timewait_sock.c soreuseport: initialise timewait reuseport field 2018-04-07 22:32:32 -04:00
inetpeer.c inetpeer: fix uninit-value in inet_getpeer 2018-04-09 10:57:35 -04:00
ip_forward.c net: rename skb_gso_validate_mtu -> skb_gso_validate_network_len 2018-03-04 17:49:17 -05:00
ip_fragment.c inet: frags: fix ip6frag_low_thresh boundary 2018-04-04 12:04:59 -04:00
ip_gre.c Merge git://git.kernel.org/pub/scm/linux/kernel/git/davem/net 2018-05-21 16:01:54 -04:00
ip_input.c net: Make ip_ra_chain per struct net 2018-03-22 15:12:56 -04:00
ip_options.c License cleanup: add SPDX GPL-2.0 license identifier to files with no license 2017-11-02 11:10:55 +01:00
ip_output.c ip: hash fragments consistently 2018-07-23 11:39:30 -07:00
ip_sockglue.c ip: in cmsg IP(V6)_ORIGDSTADDR call pskb_may_pull 2018-07-24 16:35:58 -07:00
ip_tunnel_core.c net/ipv4: Update ip_tunnel_metadata_cnt static key to modern api 2018-05-10 15:13:33 -04:00
ip_tunnel.c ip_tunnel: Fix name string concatenate in __ip_tunnel_create() 2018-06-07 16:27:16 -04:00
ip_vti.c net: Drop pernet_operations::async 2018-03-27 13:18:09 -04:00
ipcomp.c
ipconfig.c Merge git://git.kernel.org/pub/scm/linux/kernel/git/davem/net-next 2018-06-06 18:39:49 -07:00
ipip.c net: Drop pernet_operations::async 2018-03-27 13:18:09 -04:00
ipmr_base.c ipmr: fix error path when ipmr_new_table fails 2018-06-05 12:26:41 -04:00
ipmr.c Merge git://git.kernel.org/pub/scm/linux/kernel/git/davem/net-next 2018-06-06 18:39:49 -07:00
Kconfig ipmr,ipmr6: Define a uniform vif_device 2018-03-01 13:13:23 -05:00
Makefile ipv4: support sport, dport and ip_proto in RTM_GETROUTE 2018-05-23 15:14:12 -04:00
metrics.c net: metrics: add proper netlink validation 2018-06-05 12:29:43 -04:00
netfilter.c netfilter: remove struct nf_afinfo and its helper functions 2018-01-08 18:11:02 +01:00
netlink.c ipv4: support sport, dport and ip_proto in RTM_GETROUTE 2018-05-23 15:14:12 -04:00
ping.c proc: introduce proc_create_net{,_data} 2018-05-16 07:24:30 +02:00
proc.c Merge git://git.kernel.org/pub/scm/linux/kernel/git/davem/net-next 2018-06-06 18:39:49 -07:00
protocol.c
raw_diag.c
raw.c proc: introduce proc_create_net{,_data} 2018-05-16 07:24:30 +02:00
route.c treewide: kzalloc() -> kcalloc() 2018-06-12 16:19:22 -07:00
syncookies.c net/ipv4: disable SMC TCP option with SYN Cookies 2018-03-25 20:53:54 -04:00
sysctl_net_ipv4.c ipv4: Return EINVAL when ping_group_range sysctl doesn't map to user ns 2018-07-06 11:51:18 +09:00
tcp_bbr.c tcp_bbr: fix to zero idle_restart only upon S/ACKed data 2018-05-02 11:12:32 -04:00
tcp_bic.c
tcp_cdg.c
tcp_cong.c tcp: Namespace-ify sysctl_tcp_default_congestion_control 2017-11-15 14:09:52 +09:00
tcp_cubic.c
tcp_dctcp.c tcp: do not delay ACK in DCTCP upon CE status change 2018-07-20 14:32:23 -07:00
tcp_diag.c net: sock: replace sk_state_load with inet_sk_state_load and remove sk_state_store 2017-12-20 14:00:25 -05:00
tcp_fastopen.c tcp: pause Fast Open globally after third consecutive timeout 2017-12-13 15:51:12 -05:00
tcp_highspeed.c
tcp_htcp.c
tcp_hybla.c
tcp_illinois.c net/tcp/illinois: replace broken algorithm reference link 2018-02-28 12:03:47 -05:00
tcp_input.c tcp: ack immediately when a cwr packet arrives 2018-07-25 16:20:35 -07:00
tcp_ipv4.c tcp: fix sequence numbers for repaired sockets re-using TIME-WAIT sockets 2018-07-12 14:33:45 -07:00
tcp_lp.c
tcp_metrics.c net: Drop pernet_operations::async 2018-03-27 13:18:09 -04:00
tcp_minisocks.c net-tcp: remove useless tw_timeout field 2018-06-05 10:45:24 -04:00
tcp_nv.c tcp_nv: fix potential integer overflow in tcpnv_acked 2018-01-31 10:26:30 -05:00
tcp_offload.c tcp: Do not reload skb pointer after skb_gro_receive(). 2018-06-11 20:00:56 -07:00
tcp_output.c tcp: do not cancel delay-AcK on DCTCP special ACK 2018-07-20 14:32:23 -07:00
tcp_rate.c tcp: invalidate rate samples during SACK reneging 2017-12-08 10:07:02 -05:00
tcp_recovery.c tcp: tcp_rack_reo_wnd() can be static 2018-05-18 13:28:40 -04:00
tcp_scalable.c
tcp_timer.c tcp: add SACK compression 2018-05-18 11:40:27 -04:00
tcp_ulp.c net: add a UID to use for ULP socket assignment 2018-02-06 11:39:31 +01:00
tcp_vegas.c
tcp_vegas.h License cleanup: add SPDX GPL-2.0 license identifier to files with no license 2017-11-02 11:10:55 +01:00
tcp_veno.c
tcp_westwood.c
tcp_yeah.c
tcp.c tcp: identify cryptic messages as TCP seq # bugs 2018-07-18 15:26:33 -07:00
tunnel4.c inet: whitespace cleanup 2018-02-28 11:43:28 -05:00
udp_diag.c udp: fix rx queue len reported by diag and proc interface 2018-06-08 19:55:15 -04:00
udp_impl.h License cleanup: add SPDX GPL-2.0 license identifier to files with no license 2017-11-02 11:10:55 +01:00
udp_offload.c net: fix use-after-free in GRO with ESP 2018-07-02 20:34:04 +09:00
udp_tunnel.c
udp.c Revert changes to convert to ->poll_mask() and aio IOCB_CMD_POLL 2018-06-28 10:40:47 -07:00
udplite.c proc: introduce proc_create_net{,_data} 2018-05-16 07:24:30 +02:00
xfrm4_input.c xfrm: Reinject transport-mode packets through tasklet 2017-12-19 08:23:21 +01:00
xfrm4_mode_beet.c
xfrm4_mode_transport.c
xfrm4_mode_tunnel.c xfrm: Verify MAC header exists before overwriting eth_hdr(skb)->h_proto 2018-03-07 10:54:29 +01:00
xfrm4_output.c net: xfrm: use skb_gso_validate_network_len() to check gso sizes 2018-03-04 17:49:17 -05:00
xfrm4_policy.c net: Drop pernet_operations::async 2018-03-27 13:18:09 -04:00
xfrm4_protocol.c
xfrm4_state.c License cleanup: add SPDX GPL-2.0 license identifier to files with no license 2017-11-02 11:10:55 +01:00
xfrm4_tunnel.c