leandrof/linux
1
0
Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity
c6e1a0d12c
linux/include/net
History
Tom Herbert c6e1a0d12c net: Allow no-cache copy from user on transmit 
This patch uses __copy_from_user_nocache on transmit to bypass data
cache for a performance improvement.  skb_add_data_nocache and
skb_copy_to_page_nocache can be called by sendmsg functions to use
this feature, initial support is in tcp_sendmsg.  This functionality is
configurable per device using ethtool.

Presumably, this feature would only be useful when the driver does
not touch the data.  The feature is turned on by default if a device
indicates that it does some form of checksum offload; it is off by
default for devices that do no checksum offload or indicate no checksum
is necessary.  For the former case copy-checksum is probably done
anyway, in the latter case the device is likely loopback in which case
the no cache copy is probably not beneficial.

This patch was tested using 200 instances of netperf TCP_RR with
1400 byte request and one byte reply.  Platform is 16 core AMD x86.

No-cache copy disabled:
   672703 tps, 97.13% utilization
   50/90/99% latency:244.31 484.205 1028.41

No-cache copy enabled:
   702113 tps, 96.16% utilization,
   50/90/99% latency 238.56 467.56 956.955

Using 14000 byte request and response sizes demonstrate the
effects more dramatically:

No-cache copy disabled:
   79571 tps, 34.34 %utlization
   50/90/95% latency 1584.46 2319.59 5001.76

No-cache copy enabled:
   83856 tps, 34.81% utilization
   50/90/95% latency 2508.42 2622.62 2735.88

Note especially the effect on latency tail (95th percentile).

This seems to provide a nice performance improvement and is
consistent in the tests I ran.  Presumably, this would provide
the greatest benfits in the presence of an application workload
stressing the cache and a lot of transmit data happening.

Signed-off-by: Tom Herbert <therbert@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2011-04-04 22:30:30 -07:00
..
9p net/9p: Implement syncfs 9P operation 2011-03-15 09:57:38 -05:00
bluetooth Bluetooth: Fix HCI_RESET command synchronization 2011-03-24 17:04:44 -03:00
caif include/net/caif/cfctrl.h: Remove unnecessary semicolons 2010-11-15 11:07:16 -08:00
irda tty: now phase out the ioctl file pointer for good 2011-02-17 11:59:56 -08:00
iucv include cleanup: Update gfp.h and slab.h includes to prepare for breaking implicit slab.h inclusion from percpu.h 2010-03-30 22:02:32 +09:00
netfilter Merge branch 'master' of master.kernel.org:/pub/scm/linux/kernel/git/davem/net-2.6 2011-02-19 19:17:35 -08:00
netns ipv4: Invalidate nexthop cache nh_saddr more correctly. 2011-03-24 17:42:21 -07:00
phonet Phonet: allocate sock from accept syscall rather than soft IRQ 2011-03-09 11:59:32 -08:00
sctp sctp: fix auth_hmacs field's length of struct sctp_cookie 2011-04-01 21:45:50 -07:00
tc_act net/sched: add ACT_CSUM action to update packets checksums 2010-08-20 01:42:59 -07:00
act_api.h pkt_sched: gen_kill_estimator() rcu fixes 2010-06-11 18:37:08 -07:00
addrconf.h net: kill unused macros from head file 2010-12-02 13:27:33 -08:00
af_ieee802154.h af_ieee802154: add support for WANT_ACK socket option 2009-08-12 21:54:50 -07:00
af_rxrpc.h
af_unix.h af_unix: limit recursion level 2010-11-29 09:45:15 -08:00
ah.h ipsec: update MAX_AH_AUTH_LEN to support sha512 2011-01-13 21:48:25 -08:00
arp.h arp: allow to invalidate specific ARP entries 2011-01-10 16:10:37 -08:00
atmclip.h
ax25.h include cleanup: Update gfp.h and slab.h includes to prepare for breaking implicit slab.h inclusion from percpu.h 2010-03-30 22:02:32 +09:00
ax88796.h ax88796: Add method to take MAC from platform data 2009-03-24 23:32:03 -07:00
cfg80211.h wireless: fix 80211 kernel-doc warnings 2011-03-21 15:19:48 -04:00
checksum.h
cipso_ipv4.h netlabel: Label incoming TCP connections correctly in SELinux 2009-03-28 15:01:36 +11:00
cls_cgroup.h Merge commit 'v2.6.36-rc7' into core/rcu 2010-10-07 09:43:45 +02:00
compat.h net: fix compat_sys_recvmmsg parameter type 2009-12-11 15:07:56 -08:00
datalink.h
dcbevent.h net_dcb: add application notifiers 2010-12-31 10:47:46 -08:00
dcbnl.h dcbnl: add support for retrieving peer configuration - cee 2011-03-02 21:58:55 -08:00
dn_dev.h decnet: RCU conversion and get rid of dev_base_lock 2010-11-08 13:50:08 -08:00
dn_fib.h decnet: Convert to use flowidn where applicable. 2011-03-12 15:08:55 -08:00
dn_neigh.h
dn_nsp.h net: use __packed annotation 2010-06-03 03:21:52 -07:00
dn_route.h decnet: Convert to use flowidn where applicable. 2011-03-12 15:08:55 -08:00
dn.h decnet: Convert to use flowidn where applicable. 2011-03-12 15:08:55 -08:00
dsa.h dsa: add switch chip cascading support 2009-03-21 19:06:54 -07:00
dsfield.h
dst_ops.h net: Implement read-only protection and COW'ing of metrics. 2011-01-26 20:51:05 -08:00
dst.h dst: Clone child entry in skb_dst_pop 2011-03-27 17:55:01 -07:00
esp.h
ethoc.h net: Add support for the OpenCores 10/100 Mbps Ethernet MAC. 2009-03-27 00:16:21 -07:00
fib_rules.h fib_rules: __rcu annotates ctarget 2010-10-27 11:37:32 -07:00
flow.h net: Order ports in same order as addresses in flow objects. 2011-03-31 18:03:35 -07:00
garp.h net/802: add __rcu annotations 2010-10-25 13:09:44 -07:00
gen_stats.h net: cleanup include/net 2009-11-04 05:06:25 -08:00
genetlink.h include/net/genetlink.h: Allow genlmsg_cancel to accept a NULL argument 2011-02-03 20:47:08 -08:00
gre.h PPTP: PPP over IPv4 (Point-to-Point Tunneling Protocol) 2010-08-21 23:05:39 -07:00
icmp.h inetpeer: Move ICMP rate limiting state into inet_peer entries. 2011-02-04 15:59:53 -08:00
ieee80211_radiotap.h mac80211: add MCS information to radiotap 2011-01-28 15:44:29 -05:00
ieee802154_netdev.h ieee802154: add an mlme_ops call to retrieve PHY object 2009-11-06 14:32:18 +03:00
ieee802154.h ieee802154: move headers out of extra directory 2009-07-23 17:08:51 +04:00
if_inet6.h net: gre: provide multicast mappings for ipv4 and ipv6 2011-03-30 00:10:47 -07:00
inet6_connection_sock.h ipv6: Create inet6_csk_route_req(). 2010-12-02 10:59:22 -08:00
inet6_hashtables.h tcp: Fix a connect() race with timewait sockets 2009-12-08 20:17:51 -08:00
inet_common.h inet, inet6: make tcp_sendmsg() and tcp_sendpage() through inet_sendmsg() and inet_sendpage() 2010-07-12 20:21:46 -07:00
inet_connection_sock.h net: kill unused macros 2010-12-19 21:59:35 -08:00
inet_ecn.h net: return operator cleanup 2010-09-23 14:33:39 -07:00
inet_frag.h fragment: add fast path for in-order fragments 2010-06-30 13:44:29 -07:00
inet_hashtables.h tproxy: fix hash locking issue when using port redirection in __inet_inherit_port() 2010-10-21 13:06:43 +02:00
inet_sock.h inet: Remove explicit write references to sk/inet in ip_append_data 2011-03-01 12:35:02 -08:00
inet_timewait_sock.h net: optimize INET input path further 2010-12-09 20:05:58 -08:00
inetpeer.h inetpeer: Add redirect and PMTU discovery cached info. 2011-02-10 13:29:30 -08:00
ip6_checksum.h
ip6_fib.h ipv6: Convert to use flowi6 where applicable. 2011-03-12 15:08:54 -08:00
ip6_route.h ipv6: ip6_route_output does not modify sk parameter, so make it const 2011-03-22 19:17:36 -07:00
ip6_tunnel.h tunnels: add _rcu annotations 2010-10-25 13:09:45 -07:00
ip_fib.h ipv4: Fix nexthop caching wrt. scoping. 2011-03-24 18:06:47 -07:00
ip_vs.h IPVS: Use global mutex in ip_vs_app.c 2011-03-21 20:39:24 -07:00
ip.h net: gre: provide multicast mappings for ipv4 and ipv6 2011-03-30 00:10:47 -07:00
ipcomp.h percpu: add __percpu sparse annotations to net 2010-02-16 23:05:38 -08:00
ipconfig.h
ipip.h tunnels: add __rcu annotations 2010-10-27 11:37:32 -07:00
ipv6.h ipv6: Convert to use flowi6 where applicable. 2011-03-12 15:08:54 -08:00
ipx.h net: use __packed annotation 2010-06-03 03:21:52 -07:00
iw_handler.h include/net/iw_handler.h: Use SIOCIWFIRST not SIOCSIWCOMMIT in comment 2010-03-31 14:49:12 -04:00
lapb.h
lib80211.h lib80211: remove unused host_build_iv option 2010-07-26 15:09:04 -04:00
llc_c_ac.h
llc_c_ev.h
llc_c_st.h
llc_conn.h llc: use a device based hash table to speed up multicast delivery 2009-12-26 20:43:57 -08:00
llc_if.h
llc_pdu.h
llc_s_ac.h
llc_s_ev.h
llc_s_st.h
llc_sap.h
llc.h llc: convert llc_sap_list to RCU 2009-12-26 20:46:28 -08:00
mac80211.h wireless: fix 80211 kernel-doc warnings 2011-03-21 15:19:48 -04:00
mip6.h net: use __packed annotation 2010-06-03 03:21:52 -07:00
mld.h ipv6 mcast: Introduce include/net/mld.h for MLD definitions. 2010-04-23 13:35:55 +09:00
ndisc.h net: kill unused macros from head file 2010-12-02 13:27:33 -08:00
neighbour.h Merge branch 'master' of master.kernel.org:/pub/scm/linux/kernel/git/davem/net-2.6 2010-11-19 13:13:47 -08:00
net_namespace.h ipvs: move struct netns_ipvs 2011-03-15 09:36:50 +09:00
netdma.h
netevent.h net: Kill NETEVENT_PMTU_UPDATE. 2011-02-08 16:17:55 -08:00
netlabel.h include cleanup: Update gfp.h and slab.h includes to prepare for breaking implicit slab.h inclusion from percpu.h 2010-03-30 22:02:32 +09:00
netlink.h netfilter: NFNL_SUBSYS_IPSET id and NLA_PUT_NET* macros 2011-02-01 15:20:14 +01:00
netrom.h include cleanup: Update gfp.h and slab.h includes to prepare for breaking implicit slab.h inclusion from percpu.h 2010-03-30 22:02:32 +09:00
nexthop.h
nl802154.h ieee802154: add support for channel pages from IEEE 802.15.4-2006 2009-08-19 23:08:22 +04:00
p8022.h
pkt_cls.h net: Fix range checks in tcf_valid_offset(). 2010-12-21 12:43:16 -08:00
pkt_sched.h net: Define accessors to manipulate QDISC_STATE_RUNNING 2010-06-02 03:23:51 -07:00
protocol.h net: change netdev->features to u32 2011-01-24 15:32:47 -08:00
psnap.h snap: use const for descriptor 2009-03-21 19:06:50 -07:00
raw.h include/net/raw.h: Convert raw_seq_private macro to inline 2010-09-08 13:42:22 -07:00
rawv6.h ipv6: Use correct data types for ICMPv6 type and code 2009-06-23 04:31:07 -07:00
red.h sched: remove unused backlog in RED stats 2011-01-12 19:00:39 -08:00
regulatory.h cfg80211: Fix regulatory bug with multiple cards and delays 2010-11-22 15:48:51 -05:00
request_sock.h tcp: account SYN-ACK timeouts & retransmissions 2010-01-17 19:09:39 -08:00
rose.h rose: Add length checks to CALL_REQUEST parsing 2011-03-27 17:59:04 -07:00
route.h ipv4: Use flowi4_init_output() in net/route.h 2011-03-31 04:52:59 -07:00
rtnetlink.h rtnl: make link af-specific updates atomic 2010-11-27 22:56:08 -08:00
sch_generic.h net_sched: fix THROTTLED/RUNNING race 2011-03-24 00:13:14 -07:00
scm.h scm: lower SCM_MAX_FD 2010-11-24 11:16:43 -08:00
slhc_vj.h
snmp.h snmp: SNMP_UPD_PO_STATS_BH() always called from softirq 2011-03-21 18:12:54 -07:00
sock.h net: Allow no-cache copy from user on transmit 2011-04-04 22:30:30 -07:00
stp.h
tcp_states.h
tcp.h tcp: Remove debug macro of TCP_CHECK_TIMER 2011-02-20 11:10:14 -08:00
timewait_sock.h timewait_sock: Create and use getpeer op. 2010-12-01 18:09:13 -08:00
transp_v6.h ipv6: Convert to use flowi6 where applicable. 2011-03-12 15:08:54 -08:00
udp.h udp: Switch to ip_finish_skb 2011-03-01 12:35:03 -08:00
udplite.h udp: Switch to ip_finish_skb 2011-03-01 12:35:03 -08:00
wext.h wext: refactor 2009-10-07 16:39:43 -04:00
wimax.h Merge branch 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/jikos/trivial 2009-12-09 19:43:33 -08:00
wpan-phy.h ieee802154: add support for creation/removal of logic interfaces 2009-11-06 14:32:24 +03:00
x25.h X25 remove bkl in subscription ioctls 2010-11-28 11:12:20 -08:00
x25device.h X25: Add if_x25.h and x25 to device identifiers 2010-04-22 16:12:36 -07:00
xfrm.h xfrm: Assign esn pointers when cloning a state 2011-03-28 23:34:52 -07:00