linux/include/net
Eric Dumazet f54b311142 tcp: auto corking
With the introduction of TCP Small Queues, TSO auto sizing, and TCP
pacing, we can implement Automatic Corking in the kernel, to help
applications doing small write()/sendmsg() to TCP sockets.

Idea is to change tcp_push() to check if the current skb payload is
under skb optimal size (a multiple of MSS bytes)

If under 'size_goal', and at least one packet is still in Qdisc or
NIC TX queues, set the TCP Small Queue Throttled bit, so that the push
will be delayed up to TX completion time.

This delay might allow the application to coalesce more bytes
in the skb in following write()/sendmsg()/sendfile() system calls.

The exact duration of the delay is depending on the dynamics
of the system, and might be zero if no packet for this flow
is actually held in Qdisc or NIC TX ring.

Using FQ/pacing is a way to increase the probability of
autocorking being triggered.

Add a new sysctl (/proc/sys/net/ipv4/tcp_autocorking) to control
this feature and default it to 1 (enabled)

Add a new SNMP counter : nstat -a | grep TcpExtTCPAutoCorking
This counter is incremented every time we detected skb was under used
and its flush was deferred.

Tested:

Interesting effects when using line buffered commands under ssh.

Excellent performance results in term of cpu usage and total throughput.

lpq83:~# echo 1 >/proc/sys/net/ipv4/tcp_autocorking
lpq83:~# perf stat ./super_netperf 4 -t TCP_STREAM -H lpq84 -- -m 128
9410.39

 Performance counter stats for './super_netperf 4 -t TCP_STREAM -H lpq84 -- -m 128':

      35209.439626 task-clock                #    2.901 CPUs utilized
             2,294 context-switches          #    0.065 K/sec
               101 CPU-migrations            #    0.003 K/sec
             4,079 page-faults               #    0.116 K/sec
    97,923,241,298 cycles                    #    2.781 GHz                     [83.31%]
    51,832,908,236 stalled-cycles-frontend   #   52.93% frontend cycles idle    [83.30%]
    25,697,986,603 stalled-cycles-backend    #   26.24% backend  cycles idle    [66.70%]
   102,225,978,536 instructions              #    1.04  insns per cycle
                                             #    0.51  stalled cycles per insn [83.38%]
    18,657,696,819 branches                  #  529.906 M/sec                   [83.29%]
        91,679,646 branch-misses             #    0.49% of all branches         [83.40%]

      12.136204899 seconds time elapsed

lpq83:~# echo 0 >/proc/sys/net/ipv4/tcp_autocorking
lpq83:~# perf stat ./super_netperf 4 -t TCP_STREAM -H lpq84 -- -m 128
6624.89

 Performance counter stats for './super_netperf 4 -t TCP_STREAM -H lpq84 -- -m 128':
      40045.864494 task-clock                #    3.301 CPUs utilized
               171 context-switches          #    0.004 K/sec
                53 CPU-migrations            #    0.001 K/sec
             4,080 page-faults               #    0.102 K/sec
   111,340,458,645 cycles                    #    2.780 GHz                     [83.34%]
    61,778,039,277 stalled-cycles-frontend   #   55.49% frontend cycles idle    [83.31%]
    29,295,522,759 stalled-cycles-backend    #   26.31% backend  cycles idle    [66.67%]
   108,654,349,355 instructions              #    0.98  insns per cycle
                                             #    0.57  stalled cycles per insn [83.34%]
    19,552,170,748 branches                  #  488.244 M/sec                   [83.34%]
       157,875,417 branch-misses             #    0.81% of all branches         [83.34%]

      12.130267788 seconds time elapsed

Signed-off-by: Eric Dumazet <edumazet@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2013-12-06 12:51:41 -05:00
..
9p for-linus-3.12-merge minor 9p fixes and tweaks for 3.12 merge window 2013-09-11 12:34:13 -07:00
bluetooth Merge branch 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/jikos/trivial 2013-11-15 16:47:22 -08:00
caif caif_hsi.h: Remove extern from function prototypes 2013-09-23 16:29:41 -04:00
irda include/net/: Fix FSF address in file headers 2013-12-06 12:37:56 -05:00
iucv af_iucv: fix recvmsg by replacing skb_pull() function 2013-04-08 17:16:57 -04:00
netfilter netfilter: push reasm skb through instead of original frag skbs 2013-11-11 00:19:35 -05:00
netns tcp_memcontrol: Remove the per netns control. 2013-10-21 18:43:02 -04:00
nfc include/net/: Fix FSF address in file headers 2013-12-06 12:37:56 -05:00
phonet net: remove my future former mail address 2012-06-17 16:29:38 -07:00
sctp include/net/: Fix FSF address in file headers 2013-12-06 12:37:56 -05:00
tc_act include/net/: Fix FSF address in file headers 2013-12-06 12:37:56 -05:00
act_api.h net: Remove extern from include/net/ scheduling prototypes 2013-07-31 17:24:22 -07:00
addrconf.h IPv6 NAT: Do not drop DNATed 6to4/6rd packets 2013-09-28 15:56:15 -04:00
af_ieee802154.h
af_rxrpc.h af_rxrpc.h: Remove extern from function prototypes 2013-07-31 17:50:01 -07:00
af_unix.h af_unix: improve STREAM behavior with fragmented memory 2013-08-10 01:16:44 -07:00
af_vsock.h VSOCK: Move af_vsock.h and vsock_addr.h to include/net 2013-07-27 22:14:06 -07:00
ah.h
arp.h arp/neighbour.h: Remove extern from function prototypes 2013-07-31 17:50:02 -07:00
atmclip.h atm: clip: Use device neigh support on top of "arp_tbl". 2011-11-30 18:51:03 -05:00
ax25.h ax25.h: Remove extern from function prototypes 2013-07-31 17:50:02 -07:00
ax88796.h
busy_poll.h net: add cpu_relax to busy poll loop 2013-08-28 17:45:48 -04:00
cfg80211-wext.h
cfg80211.h cfg80211: export reg_initiator_name() 2013-10-28 15:05:27 +01:00
checksum.h net: checksum: fix warning in skb_checksum 2013-11-04 15:27:08 -05:00
cipso_ipv4.h include/net/: Fix FSF address in file headers 2013-12-06 12:37:56 -05:00
cls_cgroup.h Merge git://git.kernel.org/pub/scm/linux/kernel/git/davem/net-next 2013-09-05 14:54:29 -07:00
codel.h net: codel: Avoid undefined behavior from signed overflow 2013-11-04 20:01:29 -05:00
compat.h compat.h: Remove extern from function prototypes 2013-09-20 14:49:32 -04:00
datalink.h
dcbevent.h include/net/: Fix FSF address in file headers 2013-12-06 12:37:56 -05:00
dcbnl.h include/net/: Fix FSF address in file headers 2013-12-06 12:37:56 -05:00
dn.h decnet (dn*.h): Remove extern from function prototypes 2013-09-20 14:49:32 -04:00
dn_dev.h decnet (dn*.h): Remove extern from function prototypes 2013-09-20 14:49:32 -04:00
dn_fib.h decnet (dn*.h): Remove extern from function prototypes 2013-09-20 14:49:32 -04:00
dn_neigh.h decnet (dn*.h): Remove extern from function prototypes 2013-09-20 14:49:32 -04:00
dn_nsp.h decnet (dn*.h): Remove extern from function prototypes 2013-09-20 14:49:32 -04:00
dn_route.h decnet (dn*.h): Remove extern from function prototypes 2013-09-20 14:49:32 -04:00
dsa.h dsa: Include linux/if_ether.h to fix build error 2011-12-01 11:41:06 -05:00
dsfield.h ipv6: Optimize ipv6_change_dsfield(). 2013-01-09 23:59:53 -08:00
dst.h Merge git://git.kernel.org/pub/scm/linux/kernel/git/davem/net 2013-10-23 16:49:34 -04:00
dst_ops.h net: Fix warnings in dst_ops.h 2012-07-19 10:43:03 -07:00
esp.h net: move pskb_put() to core code 2013-11-07 19:28:58 -05:00
ethoc.h
fib_rules.h fib_rules.h: Remove extern from function prototypes 2013-09-20 14:49:33 -04:00
firewire.h firewire net, ipv4 arp: Extend hardware address and remove driver-level packet inspection. 2013-03-26 12:32:13 -04:00
flow.h flow.h/flow_keys.h: Remove extern from function prototypes 2013-09-20 14:49:33 -04:00
flow_keys.h flow_dissector: factor out the ports extraction in skb_flow_get_ports 2013-10-03 15:36:37 -04:00
garp.h garp.h: Remove extern from function prototypes 2013-09-20 14:49:33 -04:00
gen_stats.h gen_stats.h: Remove extern from function prototypes 2013-09-20 14:49:33 -04:00
genetlink.h genetlink: fix genl_set_err() group ID 2013-11-21 13:09:43 -05:00
gre.h ipv4: generalize gre_handle_offloads 2013-10-19 19:36:18 -04:00
gro_cells.h gro: Fix kcalloc argument order 2013-01-27 22:46:33 -05:00
icmp.h icmp.h: Remove extern from function prototypes 2013-09-20 14:49:33 -04:00
ieee80211_radiotap.h mac80211: add radiotap flag and handling for 5/10 MHz 2013-07-16 09:58:05 +03:00
ieee802154.h 6LoWPAN: add fragmentation support 2011-11-14 00:19:42 -05:00
ieee802154_netdev.h ieee802154/nl-mac.c: make some MLME operations optional 2013-04-08 12:00:16 -04:00
if_inet6.h ipv6: Remove privacy config option. 2013-10-28 20:07:50 -04:00
inet6_connection_sock.h inet*.h: Remove extern from function prototypes 2013-09-21 14:01:38 -04:00
inet6_hashtables.h ipv6: split inet6_ehashfn to hash functions per compilation unit 2013-10-19 19:45:34 -04:00
inet_common.h inet*.h: Remove extern from function prototypes 2013-09-21 14:01:38 -04:00
inet_connection_sock.h inet*.h: Remove extern from function prototypes 2013-09-21 14:01:38 -04:00
inet_ecn.h net: Correct comparisons and calculations using skb->tail and skb-transport_header 2013-05-28 23:49:07 -07:00
inet_frag.h inet: remove old fragmentation hash initializing 2013-10-23 17:01:41 -04:00
inet_hashtables.h tcp/dccp: remove twchain 2013-10-08 23:19:24 -04:00
inet_sock.h inet: convert inet_ehash_secret and ipv6_hash_secret to net_get_random_once 2013-10-19 19:45:35 -04:00
inet_timewait_sock.h netdev: inet_timewait_sock.h missing semi-colon when KMEMCHECK is enabled 2013-10-17 15:56:53 -04:00
inetpeer.h inet*.h: Remove extern from function prototypes 2013-09-21 14:01:38 -04:00
ip.h inet: fix addr_len/msg->msg_namelen assignment in recv_error and rxpmtu functions 2013-11-23 14:46:23 -08:00
ip6_checksum.h net: fix build errors if ipv6 is disabled 2013-10-09 13:04:03 -04:00
ip6_fib.h Merge git://git.kernel.org/pub/scm/linux/kernel/git/davem/net 2013-11-04 13:48:30 -05:00
ip6_route.h Merge git://git.kernel.org/pub/scm/linux/kernel/git/davem/net 2013-10-23 16:49:34 -04:00
ip6_tunnel.h tunnels: harmonize cleanup done on skb on xmit path 2013-09-04 00:27:25 -04:00
ip_fib.h ip*.h: Remove extern from function prototypes 2013-09-21 14:01:38 -04:00
ip_tunnels.h ipv4: generalize gre_handle_offloads 2013-10-19 19:36:18 -04:00
ip_vs.h netfilter: push reasm skb through instead of original frag skbs 2013-11-11 00:19:35 -05:00
ipcomp.h
ipconfig.h
ipv6.h inet: fix addr_len/msg->msg_namelen assignment in recv_error and rxpmtu functions 2013-11-23 14:46:23 -08:00
ipx.h ipx.h: Remove extern from function prototypes 2013-09-21 14:01:38 -04:00
iw_handler.h iw_handler.h: Remove extern from function prototypes 2013-09-21 14:01:39 -04:00
lapb.h lapb.h: Remove extern from function prototypes 2013-09-21 14:01:38 -04:00
lib80211.h hostap: Don't use create_proc_read_entry() 2013-04-29 15:41:56 -04:00
llc.h llc*.h: Remove extern from function prototypes 2013-09-21 14:01:38 -04:00
llc_c_ac.h llc*.h: Remove extern from function prototypes 2013-09-21 14:01:38 -04:00
llc_c_ev.h llc*.h: Remove extern from function prototypes 2013-09-21 14:01:38 -04:00
llc_c_st.h
llc_conn.h llc*.h: Remove extern from function prototypes 2013-09-21 14:01:38 -04:00
llc_if.h llc*.h: Remove extern from function prototypes 2013-09-21 14:01:38 -04:00
llc_pdu.h llc*.h: Remove extern from function prototypes 2013-09-21 14:01:38 -04:00
llc_s_ac.h llc*.h: Remove extern from function prototypes 2013-09-21 14:01:38 -04:00
llc_s_ev.h llc*.h: Remove extern from function prototypes 2013-09-21 14:01:38 -04:00
llc_s_st.h
llc_sap.h llc*.h: Remove extern from function prototypes 2013-09-21 14:01:38 -04:00
mac80211.h mac80211: add ieee80211_tx_prepare_skb() helper function 2013-10-28 15:05:27 +01:00
mac802154.h mac802154: correct a typo in ieee802154_alloc_device() prototype 2013-10-21 18:56:23 -04:00
mip6.h include/net/: Fix FSF address in file headers 2013-12-06 12:37:56 -05:00
mld.h net: ipv6: mld: get rid of MLDV2_MRC and simplify calculation 2013-09-04 14:53:20 -04:00
mrp.h Merge git://git.kernel.org/pub/scm/linux/kernel/git/davem/net 2013-10-01 17:06:14 -04:00
ndisc.h ndisc.h: Remove extern from function prototypes 2013-09-21 14:01:39 -04:00
neighbour.h arp/neighbour.h: Remove extern from function prototypes 2013-07-31 17:50:02 -07:00
net_namespace.h netfilter: nf_tables: complete net namespace support 2013-10-14 18:00:59 +02:00
net_ratelimit.h
netdma.h
netevent.h netevent/netlink.h: Remove extern from function prototypes 2013-09-21 14:01:39 -04:00
netlabel.h include/net/: Fix FSF address in file headers 2013-12-06 12:37:56 -05:00
netlink.h netevent/netlink.h: Remove extern from function prototypes 2013-09-21 14:01:39 -04:00
netprio_cgroup.h Merge git://git.kernel.org/pub/scm/linux/kernel/git/davem/net-next 2013-09-05 14:54:29 -07:00
netrom.h netrom.h: Remove extern from function prototypes 2013-09-21 14:01:39 -04:00
nexthop.h
nl802154.h
p8022.h p8022.h: Remove extern from function prototypes 2013-09-21 14:01:39 -04:00
ping.h inet: fix addr_len/msg->msg_namelen assignment in recv_error and rxpmtu functions 2013-11-23 14:46:23 -08:00
pkt_cls.h net: Remove extern from include/net/ scheduling prototypes 2013-07-31 17:24:22 -07:00
pkt_sched.h qdisc: allow setting default queuing discipline 2013-08-31 00:32:32 -04:00
protocol.h protocol.h: Remove extern from function prototypes 2013-09-23 01:51:08 -04:00
psnap.h psnap.h: Remove extern from function prototypes 2013-09-23 01:51:08 -04:00
raw.h raw/rawv6.h: Remove extern from function prototypes 2013-09-23 01:51:08 -04:00
rawv6.h raw/rawv6.h: Remove extern from function prototypes 2013-09-23 01:51:08 -04:00
red.h net_sched: red: Make minor corrections to comments 2012-04-16 23:53:11 -04:00
regulatory.h regulatory: use RCU to protect last_request 2013-01-03 13:01:30 +01:00
request_sock.h inet: includes a sock_common in request_sock 2013-10-10 00:08:07 -04:00
rose.h rose.h: Remove extern from function prototypes 2013-09-23 01:51:08 -04:00
route.h ipv4: introduce new IP_MTU_DISCOVER mode IP_PMTUDISC_INTERFACE 2013-11-05 21:52:27 -05:00
rtnetlink.h rtnetlink.h: Remove extern from function prototypes 2013-09-23 01:51:09 -04:00
sch_generic.h net_sched: add u64 rate to psched_ratecfg_precompute() 2013-09-20 14:41:02 -04:00
scm.h scm.h: Remove extern from function prototypes 2013-09-23 01:51:09 -04:00
secure_seq.h Merge git://git.kernel.org/pub/scm/linux/kernel/git/davem/net 2013-10-01 17:06:14 -04:00
slhc_vj.h
snmp.h net: avoid reloads in SNMP_UPD_PO_STATS 2012-08-06 13:40:47 -07:00
sock.h net: Dereference pointer-value of sk_prot->memory_pressure 2013-10-23 16:15:01 -04:00
stp.h stp.h: Remove extern from function prototypes 2013-09-23 01:51:09 -04:00
tcp.h tcp: auto corking 2013-12-06 12:51:41 -05:00
tcp_memcontrol.h tcp_memcontrol: Kill struct tcp_memcontrol 2013-10-21 18:43:02 -04:00
tcp_states.h
timewait_sock.h [PATCH] tcp: Cache inetpeer in timewait socket, and only when necessary. 2012-06-09 14:56:12 -07:00
transp_v6.h transp_v6.h: style neatening 2013-06-04 16:43:42 -07:00
udp.h udp: Remove unnecessary semicolon from do{}while (0) macro 2013-11-07 02:14:33 -05:00
udplite.h udplite.h: Remove extern from function prototypes 2013-09-23 16:29:40 -04:00
vsock_addr.h VSOCK: Move af_vsock.h and vsock_addr.h to include/net 2013-07-27 22:14:06 -07:00
vxlan.h vxlan: Have the NIC drivers do less work for offloads 2013-10-29 02:39:13 -07:00
wext.h wext.h: Remove extern from function prototypes 2013-09-23 16:29:40 -04:00
wimax.h wimax.h: Remove extern from function prototypes 2013-09-23 16:29:41 -04:00
wpan-phy.h mac802154: monitor device support 2012-05-16 15:17:08 -04:00
x25.h x25.h: Remove extern from function prototypes 2013-09-23 16:29:41 -04:00
x25device.h
xfrm.h ipv6: Add a receive path hook for vti6 in xfrm6_mode_tunnel. 2013-10-09 13:16:36 +02:00