linux_dsm_epyc7002/include/net
Yuchung Cheng b9f64820fb tcp: track data delivery rate for a TCP connection
This patch generates data delivery rate (throughput) samples on a
per-ACK basis. These rate samples can be used by congestion control
modules, and specifically will be used by TCP BBR in later patches in
this series.

Key state:

tp->delivered: Tracks the total number of data packets (original or not)
	       delivered so far. This is an already-existing field.

tp->delivered_mstamp: the last time tp->delivered was updated.

Algorithm:

A rate sample is calculated as (d1 - d0)/(t1 - t0) on a per-ACK basis:

  d1: the current tp->delivered after processing the ACK
  t1: the current time after processing the ACK

  d0: the prior tp->delivered when the acked skb was transmitted
  t0: the prior tp->delivered_mstamp when the acked skb was transmitted

When an skb is transmitted, we snapshot d0 and t0 in its control
block in tcp_rate_skb_sent().

When an ACK arrives, it may SACK and ACK some skbs. For each SACKed
or ACKed skb, tcp_rate_skb_delivered() updates the rate_sample struct
to reflect the latest (d0, t0).

Finally, tcp_rate_gen() generates a rate sample by storing
(d1 - d0) in rs->delivered and (t1 - t0) in rs->interval_us.

One caveat: if an skb was sent with no packets in flight, then
tp->delivered_mstamp may be either invalid (if the connection is
starting) or outdated (if the connection was idle). In that case,
we'll re-stamp tp->delivered_mstamp.

At first glance it seems t0 should always be the time when an skb was
transmitted, but actually this could over-estimate the rate due to
phase mismatch between transmit and ACK events. To track the delivery
rate, we ensure that if packets are in flight then t0 and and t1 are
times at which packets were marked delivered.

If the initial and final RTTs are different then one may be corrupted
by some sort of noise. The noise we see most often is sending gaps
caused by delayed, compressed, or stretched acks. This either affects
both RTTs equally or artificially reduces the final RTT. We approach
this by recording the info we need to compute the initial RTT
(duration of the "send phase" of the window) when we recorded the
associated inflight. Then, for a filter to avoid bandwidth
overestimates, we generalize the per-sample bandwidth computation
from:

    bw = delivered / ack_phase_rtt

to the following:

    bw = delivered / max(send_phase_rtt, ack_phase_rtt)

In large-scale experiments, this filtering approach incorporating
send_phase_rtt is effective at avoiding bandwidth overestimates due to
ACK compression or stretched ACKs.

Signed-off-by: Van Jacobson <vanj@google.com>
Signed-off-by: Neal Cardwell <ncardwell@google.com>
Signed-off-by: Yuchung Cheng <ycheng@google.com>
Signed-off-by: Nandita Dukkipati <nanditad@google.com>
Signed-off-by: Eric Dumazet <edumazet@google.com>
Signed-off-by: Soheil Hassas Yeganeh <soheil@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2016-09-21 00:23:00 -04:00
..
9p 9p: switch p9_client_read() to passing struct iov_iter * 2015-04-11 22:28:27 -04:00
bluetooth Bluetooth: Add support for appearance in scan rsp 2016-09-19 20:19:34 +02:00
caif caif: fix a signedness bug in cfpkt_iterate() 2015-02-20 17:35:14 -05:00
irda irda: Convert function pointer arrays and uses to const 2014-12-10 15:33:16 -05:00
iucv s390/iucv: do not use arrays as argument 2015-09-21 16:03:04 -07:00
netfilter Merge git://git.kernel.org/pub/scm/linux/kernel/git/davem/net 2016-09-12 15:52:44 -07:00
netns Merge branch 'master' of git://git.kernel.org/pub/scm/linux/kernel/git/klassert/ipsec-next 2016-09-08 13:09:41 -07:00
nfc NFC: digital: Add support for NFC DEP Response Waiting Time 2016-07-11 02:01:14 +02:00
phonet sock: struct proto hash function may error 2016-02-11 03:54:14 -05:00
sctp sctp: make sctp_outq_flush/tail/uncork return void 2016-09-18 22:02:33 -04:00
tc_act net sched ife action: add 16 bit helpers 2016-09-19 21:55:28 -04:00
6lowpan.h 6lowpan: add 802.15.4 short addr slaac 2016-06-15 20:41:22 -07:00
act_api.h net_sched: convert tcf_exts from list to pointer array 2016-08-17 19:27:51 -04:00
addrconf.h ipv6: export several functions 2016-06-15 20:41:23 -07:00
af_ieee802154.h ieee802154: af_ieee802154: fix typo in comment. 2015-09-17 13:20:05 +02:00
af_rxrpc.h rxrpc: Rewrite the data and ack handling code 2016-09-08 11:10:12 +01:00
af_unix.h af_unix: split 'u->readlock' into two: 'iolock' and 'bindlock' 2016-09-04 13:29:29 -07:00
af_vsock.h VSOCK: Introduce virtio_vsock_common.ko 2016-08-02 02:57:29 +03:00
ah.h
arp.h neigh: Factor out ___neigh_lookup_noref 2015-03-04 00:23:23 -05:00
atmclip.h
ax25.h ax25: Stop using sock->sk_protinfo. 2015-06-28 16:55:44 -07:00
ax88796.h
bond_3ad.h bonding: 3ad: apply ad_actor settings changes immediately 2016-02-09 04:45:49 -05:00
bond_alb.h net: Move bonding headers under include/net 2014-11-10 13:27:49 -05:00
bond_options.h bonding: convert num_grat_arp to the new bonding option API 2015-07-27 01:05:24 -07:00
bonding.h bonding: prevent out of bound accesses 2016-07-01 06:06:09 -04:00
busy_poll.h net: un-inline sk_busy_loop() 2015-11-18 16:17:38 -05:00
calipso.h calipso: Add a label cache. 2016-06-27 15:06:17 -04:00
cfg80211-wext.h
cfg80211.h This time we have various things - all across the board: 2016-09-18 22:29:08 -04:00
cfg802154.h ieee802154: add netns support 2016-07-08 12:20:57 +02:00
checksum.h csum: Update csum_block_add to use rotate instead of byteswap 2016-03-13 15:01:00 -04:00
cipso_ipv4.h cipso: don't use IPCB() to locate the CIPSO IP option 2015-02-11 14:46:37 -05:00
cls_cgroup.h cls_cgroup: get sk_classid only from full sockets 2016-04-19 20:09:25 -04:00
codel_impl.h codel: split into multiple files 2016-04-25 16:44:27 -04:00
codel_qdisc.h net_sched: fq_codel: cache skb->truesize into skb->cb 2016-06-25 12:19:35 -04:00
codel.h codel: split into multiple files 2016-04-25 16:44:27 -04:00
compat.h packet: compat support for sock_fprog 2016-06-09 23:41:03 -07:00
datalink.h
dcbevent.h
dcbnl.h net/dcb: Add IEEE QCN attribute 2015-03-06 21:50:02 -05:00
devlink.h devlink: remove unused priv_size 2016-08-26 11:55:18 -07:00
dn_dev.h
dn_fib.h
dn_neigh.h netfilter: Pass net into okfn 2015-09-17 17:18:37 -07:00
dn_nsp.h
dn_route.h
dn.h
dsa.h net-next: dsa: add Qualcomm tag RX/TX handler 2016-09-16 04:31:51 -04:00
dsfield.h
dst_cache.h net: add dst_cache support 2016-02-16 20:21:48 -05:00
dst_metadata.h net/dst: Utility functions to build dst_metadata without supplying an skb 2016-09-10 20:53:55 -07:00
dst_ops.h ipv4, ipv6: Pass net into __ip_local_out and __ip6_local_out 2015-10-08 04:27:02 -07:00
dst.h route: move lwtunnel state to a single place 2016-04-25 16:20:09 -04:00
esp.h
ethoc.h net/ethoc: support big-endian register layout 2015-09-23 15:33:15 -07:00
fib_rules.h net: Add l3mdev rule 2016-06-08 11:36:02 -07:00
firewire.h
flow_dissector.h rps: flow_dissector: Add the const for the parameter of flow_keys_have_l4 2016-09-01 16:51:08 -07:00
flow.h net: flow: Remove FLOWI_FLAG_L3MDEV_SRC flag 2016-09-10 23:12:53 -07:00
flowcache.h
fou.h fou: Add encap ops for IPv6 tunnels 2016-05-20 18:03:16 -04:00
fq_impl.h fq: split out backlog update logic 2016-04-28 17:03:38 -04:00
fq.h fq: add fair queuing framework 2016-04-25 16:45:53 -04:00
garp.h
gen_stats.h net: sched: do not acquire qdisc spinlock in qdisc/class stats dump 2016-06-07 16:37:14 -07:00
genetlink.h Revert "genl: Add genlmsg_new_unicast() for unicast message allocation" 2016-02-18 11:42:19 -05:00
geneve.h net: Remove deprecated tunnel specific UDP offload functions 2016-06-17 20:23:32 -07:00
gre.h Merge git://git.kernel.org/pub/scm/linux/kernel/git/davem/net 2016-08-18 01:17:32 -04:00
gro_cells.h gro_cells: gro_cells_receive now return error code 2016-07-22 00:50:41 -04:00
gtp.h gtp: #define #define _GTP_H_ and not #define _GTP_H 2016-07-25 17:55:43 -07:00
gue.h gue: Protocol constants for remote checksum offload 2014-11-05 16:30:03 -05:00
hwbm.h net: add a hardware buffer management helper API 2016-03-14 12:19:46 -04:00
icmp.h net: snmp: kill STATS_BH macros 2016-04-27 22:48:25 -04:00
ieee80211_radiotap.h mac80211: add support for radiotap timestamp field 2016-09-12 11:45:45 +02:00
ieee802154_netdev.h mac802154: constify ieee802154_llsec_ops structure 2016-01-04 20:40:41 +01:00
if_inet6.h ipv6: do retries on stable privacy addresses 2015-03-23 22:12:09 -04:00
ila.h ila: Add generic ILA translation facility 2015-12-15 23:25:20 -05:00
inet6_connection_sock.h ipv6: remove unused in6_addr struct 2016-03-22 15:45:44 -04:00
inet6_hashtables.h tcp/dccp: do not touch listener sk_refcnt under synflood 2016-04-04 22:11:20 -04:00
inet_common.h ip4ip6: Support for GSO/GRO 2016-05-20 18:03:17 -04:00
inet_connection_sock.h tcp/dccp: fix another race at listener dismantle 2016-02-18 11:35:51 -05:00
inet_ecn.h ipv6: suppress sparse warnings in IP6_ECN_set_ce() 2016-08-13 15:08:00 -07:00
inet_frag.h ipv4: namespacify ip fragment max dist sysctl knob 2016-02-16 20:42:54 -05:00
inet_hashtables.h tcp/dccp: do not touch listener sk_refcnt under synflood 2016-04-04 22:11:20 -04:00
inet_sock.h ipv6: Allow request socks to contain IPv6 options. 2016-06-27 15:05:28 -04:00
inet_timewait_sock.h Merge git://git.kernel.org/pub/scm/linux/kernel/git/davem/net 2015-10-20 06:08:27 -07:00
inetpeer.h inet: tcp: fix inetpeer_set_addr_v4() 2015-12-16 00:14:12 -05:00
ip6_checksum.h ipv6: Pass proto to csum_ipv6_magic as __u8 instead of unsigned short 2016-03-13 23:55:13 -04:00
ip6_fib.h ipv6: Check rt->dst.from for the DST_NOCACHE route 2015-11-15 17:12:37 -05:00
ip6_route.h ipv6: Export p6_route_input_lookup symbol 2016-09-19 01:25:22 -04:00
ip6_tunnel.h ip6_tunnel: add collect_md mode to IPv6 tunnels 2016-09-17 10:13:07 -04:00
ip_fib.h net: Don't delete routes in different VRFs 2016-09-06 13:56:13 -07:00
ip_tunnels.h ip_tunnel: add collect_md mode to IPIP tunnel 2016-09-17 10:13:07 -04:00
ip_vs.h ipvs: update real-server binding of outgoing connections in SIP-pe 2016-06-06 09:47:25 +09:00
ip.h net/ipv4: Introduce IPSKB_FRAG_SEGS bit to inet_skb_parm.flags 2016-07-19 16:40:22 -07:00
ipcomp.h
ipconfig.h
ipv6.h Merge branch 'stable-4.8' of git://git.infradead.org/users/pcmoore/selinux into next 2016-07-07 10:15:34 +10:00
ipx.h switch ipxrtr_route_packet() from iovec to msghdr 2014-11-24 04:28:49 -05:00
iw_handler.h cfg80211/wext: fix message ordering 2016-01-29 17:13:43 +01:00
kcm.h kcm: Use stream parser 2016-08-17 19:36:23 -04:00
l3mdev.h net: l3mdev: Remove netif_index_is_l3_master 2016-09-17 10:05:05 -04:00
lapb.h
lib80211.h lib80211: remove unused print_ssid() 2014-10-14 02:18:27 +02:00
llc_c_ac.h
llc_c_ev.h
llc_c_st.h llc: Make llc_conn_ev_qfyr_t function pointer arrays const 2014-12-10 15:21:24 -05:00
llc_conn.h net: Pass kern from net_proto_family.create to sk_alloc 2015-05-11 10:50:17 -04:00
llc_if.h
llc_pdu.h
llc_s_ac.h
llc_s_ev.h
llc_s_st.h llc: Make llc_sap_action_t function pointer arrays const 2014-12-10 15:21:24 -05:00
llc_sap.h
llc.h
lwtunnel.h net: lwtunnel: Handle fragmentation 2016-08-30 22:27:18 -07:00
mac80211.h mac80211: allow driver to handle packet-loss mechanism 2016-09-15 16:46:20 +02:00
mac802154.h ieee802154: cleanup WARN_ON for fc fetch 2016-07-08 13:23:12 +02:00
mip6.h
mld.h
mpls_iptunnel.h mpls: multipath route support 2015-10-23 06:26:42 -07:00
mpls.h openvswitch: Add basic MPLS support to kernel 2014-11-05 23:52:33 -08:00
mrp.h
ncsi.h net/ncsi: Package and channel management 2016-07-19 20:49:17 -07:00
ndisc.h 6lowpan: introduce 6lowpan-nd 2016-06-15 20:41:23 -07:00
neighbour.h net: add explicit logging and stat for neighbour table overflow 2015-08-10 13:46:21 -07:00
net_namespace.h treewide: replace obsolete _refok by __ref 2016-08-02 17:31:41 -04:00
net_ratelimit.h
netevent.h neigh: Send a notification when DELAY_PROBE_TIME changes 2016-07-05 09:06:29 -07:00
netlabel.h netlabel: Implement CALIPSO config functions for SMACK. 2016-06-27 15:06:18 -04:00
netlink.h netlink: kill nla_put_u64() 2016-05-16 13:46:23 -04:00
netprio_cgroup.h net: wrap sock->sk_cgrp_prioidx and ->sk_classid inside a struct 2015-12-08 22:02:33 -05:00
netrom.h
nexthop.h
nl802154.h ieee802154: add netns support 2016-07-08 12:20:57 +02:00
p8022.h
ping.h net: ping: make ping_v6_sendmsg static 2016-03-23 22:09:58 -04:00
pkt_cls.h net_sched: properly handle failure case of tcf_exts_init() 2016-08-22 17:02:31 -07:00
pkt_sched.h net: sched: convert qdisc linked list to hashtable 2016-08-10 17:19:02 -07:00
pptp.h pptp: Refactor the struct and macros of PPTP codes 2016-08-15 10:55:53 -07:00
protocol.h udp: Remove udp_offloads 2016-04-07 16:53:30 -04:00
psnap.h
raw.h sock: struct proto hash function may error 2016-02-11 03:54:14 -05:00
rawv6.h
red.h
regulatory.h cfg80211: allow wiphy specific regdomain management 2014-12-17 11:49:55 +01:00
request_sock.h inet: reqsk_alloc() needs to take care of dead listeners 2016-04-04 22:11:19 -04:00
rose.h
route.h net: ipv4: Remove l3mdev_get_saddr 2016-09-10 23:12:53 -07:00
rtnetlink.h net: rtnetlink: add support for the IFLA_STATS_LINK_XSTATS_SLAVE attribute 2016-06-30 06:15:04 -04:00
sch_generic.h sched: add and use qdisc_skb_head helpers 2016-09-19 01:47:18 -04:00
scm.h unix: correctly track in-flight fds in sending process user_struct 2016-02-08 10:30:42 -05:00
secure_seq.h
slhc_vj.h
snmp.h net: snmp: fix 64bit stats on 32bit arches 2016-04-28 11:49:45 -04:00
sock_reuseport.h soreuseport: fix NULL ptr dereference SO_REUSEPORT after bind 2016-01-19 14:44:23 -05:00
sock.h net: remove clear_sk() method 2016-08-23 23:25:29 -07:00
Space.h
stp.h
strparser.h kcm: Remove TCP specific references from kcm and strparser 2016-08-28 23:32:41 -04:00
switchdev.h switchdev: Fix return value of switchdev_port_fdb_dump(). 2016-09-02 11:00:21 -07:00
tcp_states.h inet: add TCP_NEW_SYN_RECV state 2015-03-12 22:58:12 -04:00
tcp.h tcp: track data delivery rate for a TCP connection 2016-09-21 00:23:00 -04:00
timewait_sock.h inet: remove BUG_ON() in twsk_destructor() 2015-07-09 15:12:20 -07:00
transp_v6.h ipv6: add new struct ipcm6_cookie 2016-05-03 16:08:14 -04:00
tso.h net: tso: add support for IPv6 2015-10-26 22:24:22 -07:00
udp_tunnel.h vxlan: Add new UDP encapsulation offload type for VXLAN-GPE 2016-06-17 20:23:32 -07:00
udp.h net: diag: support SOCK_DESTROY for UDP sockets 2016-08-23 23:12:27 -07:00
udplite.h net: switch memcpy_fromiovec()/memcpy_fromiovecend() users to copy_from_iter() 2015-02-04 01:34:15 -05:00
vsock_addr.h
vxlan.h net/ip_tunnels: Introduce tunnel_id_to_key32() and key32_to_tunnel_id() 2016-09-10 20:53:55 -07:00
wext.h
wimax.h
x25.h
x25device.h
xfrm.h xfrm: constify xfrm_replay structures 2016-08-10 11:18:49 +02:00