linux/net
Raghavendra K T a3a773726c net: Optimize snmp stat aggregation by walking all the percpu data at once
Docker container creation linearly increased from around 1.6 sec to 7.5 sec
(at 1000 containers) and perf data showed 50% ovehead in snmp_fold_field.

reason: currently __snmp6_fill_stats64 calls snmp_fold_field that walks
through per cpu data of an item (iteratively for around 36 items).

idea: This patch tries to aggregate the statistics by going through
all the items of each cpu sequentially which is reducing cache
misses.

Docker creation got faster by more than 2x after the patch.

Result:
                       Before           After
Docker creation time   6.836s           3.25s
cache miss             2.7%             1.41%

perf before:
    50.73%  docker           [kernel.kallsyms]       [k] snmp_fold_field
     9.07%  swapper          [kernel.kallsyms]       [k] snooze_loop
     3.49%  docker           [kernel.kallsyms]       [k] veth_stats_one
     2.85%  swapper          [kernel.kallsyms]       [k] _raw_spin_lock

perf after:
    10.57%  docker           docker                [.] scanblock
     8.37%  swapper          [kernel.kallsyms]     [k] snooze_loop
     6.91%  docker           [kernel.kallsyms]     [k] snmp_get_cpu_field
     6.67%  docker           [kernel.kallsyms]     [k] veth_stats_one

changes/ideas suggested:
Using buffer in stack (Eric), Usage of memset (David), Using memcpy in
place of unaligned_put (Joe).

Signed-off-by: Raghavendra K T <raghavendra.kt@linux.vnet.ibm.com>
Acked-by: Eric Dumazet <edumazet@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2015-08-30 21:48:58 -07:00
..
6lowpan 6lowpan: move module_init into core functionality 2015-08-11 22:05:36 +02:00
9p 9p: ensure err is initialized to 0 in p9_client_read/write 2015-08-22 21:35:02 -04:00
802 net: Kill dev_rebuild_header 2015-03-02 16:43:41 -05:00
8021q net: 8021q: convert to using IFF_NO_QUEUE 2015-08-18 11:55:06 -07:00
appletalk net: Pass kern from net_proto_family.create to sk_alloc 2015-05-11 10:50:17 -04:00
atm br2684: Remove unnecessary formatting macros b1 and bs 2015-07-31 15:25:52 -07:00
ax25 NET: AX.25: Stop heartbeat timer on disconnect. 2015-07-15 15:59:58 -07:00
batman-adv batman-adv: turn batadv_neigh_node_get() into local function 2015-08-27 20:15:34 +02:00
bluetooth Bluetooth: Fix SCO link type handling on connection complete 2015-08-28 21:03:00 +02:00
bridge Merge git://git.kernel.org/pub/scm/linux/kernel/git/pablo/nf-next 2015-08-28 16:29:59 -07:00
caif net: caif: convert to using IFF_NO_QUEUE 2015-08-18 11:55:07 -07:00
can can: replace timestamp as unique skb attribute 2015-07-12 21:13:22 +02:00
ceph libceph: treat sockaddr_storage with uninitialized family as blank 2015-07-09 20:30:34 +03:00
core ip_tunnels: record IP version in tunnel info 2015-08-29 13:07:54 -07:00
dcb net/dcb: Add IEEE QCN attribute 2015-03-06 21:50:02 -05:00
dccp tcp: fix recv with flags MSG_WAITALL | MSG_PEEK 2015-07-27 01:06:53 -07:00
decnet net: Pass kern from net_proto_family.create to sk_alloc 2015-05-11 10:50:17 -04:00
dns_resolver
dsa net: dsa: Allow multi hop routes to be expressed 2015-08-18 14:17:21 -07:00
ethernet net: ethernet: Fix double word "the the" in eth.c 2015-08-09 22:53:00 -07:00
hsr net: hsr: convert to using IFF_NO_QUEUE 2015-08-18 11:55:07 -07:00
ieee802154 Merge branch 'for-upstream' of git://git.kernel.org/pub/scm/linux/kernel/git/bluetooth/bluetooth-next 2015-08-29 13:15:03 -07:00
ipv4 net: Introduce helper functions to get the per cpu data 2015-08-30 21:48:58 -07:00
ipv6 net: Optimize snmp stat aggregation by walking all the percpu data at once 2015-08-30 21:48:58 -07:00
ipx net: Pass kern from net_proto_family.create to sk_alloc 2015-05-11 10:50:17 -04:00
irda irda: use msecs_to_jiffies for conversion to jiffies 2015-05-25 17:46:21 -04:00
iucv net: Pass kern from net_proto_family.create to sk_alloc 2015-05-11 10:50:17 -04:00
key net: Fix RCU splat in af_key 2015-08-24 14:48:10 -07:00
l2tp net: Modify sk_alloc to not reference count the netns of kernel sockets. 2015-05-11 10:50:18 -04:00
lapb
llc tcp: fix recv with flags MSG_WAITALL | MSG_PEEK 2015-07-27 01:06:53 -07:00
mac80211 Merge git://git.kernel.org/pub/scm/linux/kernel/git/davem/net 2015-08-21 11:44:04 -07:00
mac802154 ieee802154: add ack request default handling 2015-08-10 20:43:06 +02:00
mpls lwt: Add cfg argument to build_state 2015-08-24 10:34:40 -07:00
netfilter Merge git://git.kernel.org/pub/scm/linux/kernel/git/pablo/nf-next 2015-08-28 16:29:59 -07:00
netlabel netlink: implement nla_put_in_addr and nla_put_in6_addr 2015-03-31 13:58:35 -04:00
netlink netlink: mmap: fix lookup frame position 2015-08-28 22:25:42 -07:00
netrom netfilter: Remove spurios included of netfilter.h 2015-06-18 21:14:32 +02:00
nfc nfc: netlink: Add capability to reply to vendor_cmd with data 2015-08-20 22:00:11 +02:00
openvswitch openvswitch: Remove vport-net 2015-08-29 19:07:15 -07:00
packet packet: add extended BPF fanout mode 2015-08-17 14:22:48 -07:00
phonet net: Pass kern from net_proto_family.create to sk_alloc 2015-05-11 10:50:17 -04:00
rds RDS: remove superfluous from rds_ib_alloc_fmr() 2015-08-25 16:28:11 -07:00
rfkill Merge branch 'for-upstream' of git://git.kernel.org/pub/scm/linux/kernel/git/bluetooth/bluetooth-next 2015-08-17 15:41:21 -07:00
rose Merge git://git.kernel.org/pub/scm/linux/kernel/git/davem/net 2015-06-24 02:58:51 -07:00
rxrpc net: Pass kern from net_proto_family.create to sk_alloc 2015-05-11 10:50:17 -04:00
sched net: sched: don't break line in tc_classify loop notification 2015-08-28 14:01:15 -07:00
sctp Merge git://git.kernel.org/pub/scm/linux/kernel/git/davem/net 2015-08-30 21:45:01 -07:00
sunrpc NFS client bugfixes for Linux 4.2 2015-07-28 09:37:44 -07:00
switchdev net: switchdev: support static FDB addresses 2015-08-11 12:03:19 -07:00
tipc tipc: fix stale link problem during synchronization 2015-08-23 16:14:45 -07:00
unix net/unix: support SCM_SECURITY for stream sockets 2015-06-10 22:49:20 -07:00
vmw_vsock net: Pass kern from net_proto_family.create to sk_alloc 2015-05-11 10:50:17 -04:00
wimax net:wimax: Fix doucble word "the the" in networking.xml 2015-08-09 22:43:52 -07:00
wireless nl80211: Allow setting multicast rate on OCB interfaces 2015-08-14 17:49:48 +02:00
x25 net: Pass kern from net_proto_family.create to sk_alloc 2015-05-11 10:50:17 -04:00
xfrm xfrm: Add oif to dst lookups 2015-08-11 12:41:35 +02:00
compat.c net: switch importing msghdr from userland to {compat_,}import_iovec() 2015-04-09 00:02:26 -04:00
Kconfig lwtunnel: infrastructure for handling light weight tunnels like mpls 2015-07-21 10:39:03 -07:00
Makefile mpls: Refactor how the mpls module is built 2015-03-04 00:26:06 -05:00
socket.c net: Add a struct net parameter to sock_create_kern 2015-05-11 10:50:17 -04:00
sysctl_net.c