]> git.baikalelectronics.ru Git - kernel.git/commit
netfilter: nf_ct_icmp: keep the ICMP ct entries longer
authorJan Kasprzak <kas@fi.muni.cz>
Mon, 8 Jun 2009 13:53:43 +0000 (15:53 +0200)
committerPatrick McHardy <kaber@trash.net>
Mon, 8 Jun 2009 13:53:43 +0000 (15:53 +0200)
commit3bd1a2b4c34a6a43a11bf6d27d8d7f26db22921a
tree0ec53ee8c373e6b4224b2fda40ed4fc49c1ed822
parente18f85c3934122138287f0486632a44ff1ec0404
netfilter: nf_ct_icmp: keep the ICMP ct entries longer

Current conntrack code kills the ICMP conntrack entry as soon as
the first reply is received. This is incorrect, as we then see only
the first ICMP echo reply out of several possible duplicates as
ESTABLISHED, while the rest will be INVALID. Also this unnecessarily
increases the conntrackd traffic on H-A firewalls.

Make all the ICMP conntrack entries (including the replied ones)
last for the default of nf_conntrack_icmp{,v6}_timeout seconds.

Signed-off-by: Jan "Yenya" Kasprzak <kas@fi.muni.cz>
Signed-off-by: Patrick McHardy <kaber@trash.net>
include/net/netfilter/ipv4/nf_conntrack_icmp.h [deleted file]
include/net/netfilter/ipv6/nf_conntrack_icmpv6.h
include/net/netfilter/nf_conntrack.h
net/ipv4/netfilter/nf_conntrack_proto_icmp.c
net/ipv6/netfilter/nf_conntrack_proto_icmpv6.c