Eric Dumazet acb32ba3de ipv4: reduce percpu needs for icmpmsg mibs ... Reading /proc/net/snmp on a machine with a lot of cpus is very expensive (can be ~88000 us). This is because ICMPMSG MIB uses 4096 bytes per cpu, and folding values for all possible cpus can read 16 Mbytes of memory. ICMP messages are not considered as fast path on a typical server, and eventually few cpus handle them anyway. We can afford an atomic operation instead of using percpu data. This saves 4096 bytes per cpu and per network namespace. Signed-off-by: Eric Dumazet <eric.dumazet@gmail.com> Signed-off-by: David S. Miller <davem@davemloft.net>		2011-11-09 16:04:20 -05:00
..
conntrack.h	atomic: use <linux/atomic.h>	2011-07-26 16:49:47 -07:00
core.h	percpu: add __percpu sparse annotations to net	2010-02-16 23:05:38 -08:00
dccp.h	[NETNS][DCCPV6]: Move the dccp_v6_ctl_sk on the struct net.	2008-04-13 22:32:25 -07:00
generic.h	netns: let net_generic take pointer-to-const args	2010-11-21 10:05:10 -08:00
hash.h	netns: introduce the net_hash_mix "salt" for hashes	2008-06-16 17:14:11 -07:00
ipv4.h	net: ipv4: add IPPROTO_ICMP socket kind	2011-05-13 16:08:13 -04:00
ipv6.h	ipv6: ip6mr: support multiple tables	2010-05-11 14:40:55 +02:00
mib.h	ipv4: reduce percpu needs for icmpmsg mibs	2011-11-09 16:04:20 -05:00
packet.h	packet: convert socket list to RCU (v3)	2010-02-22 15:45:56 -08:00
unix.h
x_tables.h	net, netns_xt: shrink netns_xt members	2009-07-05 19:16:18 -07:00
xfrm.h	netns: reorder fields in struct net	2010-10-17 13:49:14 -07:00