linux/net
Eric Dumazet 10d3be5692 tcp-tso: do not split TSO packets at retransmit time
Linux TCP stack painfully segments all TSO/GSO packets before retransmits.

This was fine back in the days when TSO/GSO were emerging, with their
bugs, but we believe the dark age is over.

Keeping big packets in write queues, but also in stack traversal
has a lot of benefits.
 - Less memory overhead, because write queues have less skbs
 - Less cpu overhead at ACK processing.
 - Better SACK processing, as lot of studies mentioned how
   awful linux was at this ;)
 - Less cpu overhead to send the rtx packets
   (IP stack traversal, netfilter traversal, drivers...)
 - Better latencies in presence of losses.
 - Smaller spikes in fq like packet schedulers, as retransmits
   are not constrained by TCP Small Queues.

1 % packet losses are common today, and at 100Gbit speeds, this
translates to ~80,000 losses per second.
Losses are often correlated, and we see many retransmit events
leading to 1-MSS train of packets, at the time hosts are already
under stress.

Signed-off-by: Eric Dumazet <edumazet@google.com>
Acked-by: Yuchung Cheng <ycheng@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2016-04-24 14:43:59 -04:00
..
6lowpan 6lowpan: iphc: fix handling of link-local compression 2016-04-08 19:28:13 +02:00
9p net/9p: convert to new CQ API 2016-03-10 20:54:09 -05:00
802
8021q vlan: propagate gso_max_segs 2016-03-17 21:05:01 -04:00
appletalk appletalk: fix erroneous return value 2016-02-18 14:59:34 -05:00
atm net: Generalise wq_has_sleeper helper 2015-11-30 14:47:33 -05:00
ax25 ax25: add link layer header validation function 2016-03-09 22:13:01 -05:00
batman-adv batman-adv: clarify CFG80211 dependency 2016-03-02 13:45:47 -05:00
bluetooth sock: tigthen lockdep checks for sock_owned_by_user 2016-04-13 22:37:20 -04:00
bridge Merge git://git.kernel.org/pub/scm/linux/kernel/git/davem/net 2016-04-23 18:51:33 -04:00
caif net: caif: fix misleading indentation 2016-03-14 13:09:50 -04:00
can sock: enable timestamping using control messages 2016-04-04 15:50:30 -04:00
ceph mm, fs: get rid of PAGE_CACHE_* and page_cache_{get,release} macros 2016-04-04 10:41:08 -07:00
core libnl: nla_put_msecs(): align on a 64-bit area 2016-04-23 20:13:24 -04:00
dcb net/dcb: make dcbnl.c explicitly non-modular 2015-10-09 07:52:27 -07:00
dccp net: introduce lockdep_is_held and update various places to use it 2016-04-07 16:44:14 -04:00
decnet decnet: Do not build routes to devices without decnet private data. 2016-04-10 23:01:30 -04:00
dns_resolver net: dns_resolver: convert time_t to time64_t 2015-11-18 16:27:46 -05:00
dsa net: dsa: remove tag_protocol from dsa_switch 2016-04-21 13:43:11 -04:00
ethernet eth: Pull header from first fragment via eth_get_headlen 2016-02-24 13:58:05 -05:00
hsr NLA_BINARY misuse bug in HSR 2016-04-21 13:59:08 -04:00
ieee802154 libnl: nla_put_le64(): align on a 64-bit area 2016-04-23 20:13:24 -04:00
ipv4 tcp-tso: do not split TSO packets at retransmit time 2016-04-24 14:43:59 -04:00
ipv6 Merge git://git.kernel.org/pub/scm/linux/kernel/git/pablo/nf-next 2016-04-24 00:12:08 -04:00
ipx
irda Merge 4.5-rc4 into tty-next 2016-02-14 14:36:04 -08:00
iucv af_iucv: Validate socket address length in iucv_sock_bind() 2016-01-19 14:21:08 -05:00
kcm kcm: Add receive message timeout 2016-03-09 16:36:15 -05:00
key af_key: fix two typos 2015-10-23 03:05:19 -07:00
l2tp libnl: nla_put_msecs(): align on a 64-bit area 2016-04-23 20:13:24 -04:00
l3mdev net: l3mdev: address selection should only consider devices in L3 domain 2016-02-26 14:22:26 -05:00
lapb
llc sock: tigthen lockdep checks for sock_owned_by_user 2016-04-13 22:37:20 -04:00
mac80211 cfg80211: remove enum ieee80211_band 2016-04-12 15:56:15 +02:00
mac802154 Merge git://git.kernel.org/pub/scm/linux/kernel/git/davem/net-next 2016-03-19 10:05:34 -07:00
mpls GSO: Add GSO type for fixed IPv4 ID 2016-04-14 16:23:40 -04:00
netfilter Merge git://git.kernel.org/pub/scm/linux/kernel/git/pablo/nf-next 2016-04-24 00:12:08 -04:00
netlabel netlabel: do not initialise statics to NULL 2016-03-07 11:08:26 -05:00
netlink Merge git://git.kernel.org/pub/scm/linux/kernel/git/davem/net 2016-04-23 18:51:33 -04:00
netrom
nfc Merge git://git.kernel.org/pub/scm/linux/kernel/git/davem/net-next 2016-03-19 10:05:34 -07:00
openvswitch Merge git://git.kernel.org/pub/scm/linux/kernel/git/pablo/nf-next 2016-04-24 00:12:08 -04:00
packet Merge git://git.kernel.org/pub/scm/linux/kernel/git/davem/net 2016-04-23 18:51:33 -04:00
phonet sock: struct proto hash function may error 2016-02-11 03:54:14 -05:00
rds RDS: Fix the atomicity for congestion map update 2016-04-16 19:01:05 -04:00
rfkill rfkill: Use switch to demux userspace operations 2016-04-05 10:48:53 +02:00
rose
rxrpc rxrpc: Create a null security type and get rid of conditional calls 2016-04-11 15:34:41 -04:00
sched Merge git://git.kernel.org/pub/scm/linux/kernel/git/davem/net 2016-04-23 18:51:33 -04:00
sctp Merge git://git.kernel.org/pub/scm/linux/kernel/git/davem/net 2016-04-23 18:51:33 -04:00
sunrpc Merge git://git.kernel.org/pub/scm/linux/kernel/git/davem/net 2016-04-23 18:51:33 -04:00
switchdev switchdev: fix typo in comments/doc 2016-03-24 14:51:24 -04:00
tipc tipc: fix stale links after re-enabling bearer 2016-04-24 14:35:07 -04:00
unix Merge git://git.kernel.org/pub/scm/linux/kernel/git/davem/net 2016-02-23 00:09:14 -05:00
vmw_vsock VSOCK: Only check error on skb_recv_datagram when skb is NULL 2016-04-19 20:42:01 -04:00
wimax
wireless Merge git://git.kernel.org/pub/scm/linux/kernel/git/davem/net 2016-04-23 18:51:33 -04:00
x25
xfrm xfrm: align nlattr properly when needed 2016-04-23 20:13:25 -04:00
compat.c
Kconfig Make DST_CACHE a silent config option 2016-03-21 22:56:38 -04:00
Makefile kcm: Kernel Connection Multiplexor module 2016-03-09 16:36:14 -05:00
socket.c Merge branch 'for-davem' of git://git.kernel.org/pub/scm/linux/kernel/git/viro/vfs 2016-04-14 00:39:15 -04:00
sysctl_net.c net: sysctl: fix a kmemleak warning 2015-10-23 06:22:08 -07:00