diff options
| author | Florian Westphal <fw@strlen.de> | 2017-08-30 19:24:58 +0200 |
|---|---|---|
| committer | David S. Miller <davem@davemloft.net> | 2017-08-30 11:20:09 -0700 |
| commit | 31770e34e43d6c8dee129bfee77e56c34e61f0e5 (patch) | |
| tree | a4635a632732b39ef560a8f72eb3a81d04f01605 /net/ipv4/proc.c | |
| parent | c1d2b4c3e204e602c97680335d082b8d012d08cd (diff) | |
| download | linux-31770e34e43d6c8dee129bfee77e56c34e61f0e5.tar.bz2 | |
tcp: Revert "tcp: remove header prediction"
This reverts commit 45f119bf936b1f9f546a0b139c5b56f9bb2bdc78.
Eric Dumazet says:
We found at Google a significant regression caused by
45f119bf936b1f9f546a0b139c5b56f9bb2bdc78 tcp: remove header prediction
In typical RPC (TCP_RR), when a TCP socket receives data, we now call
tcp_ack() while we used to not call it.
This touches enough cache lines to cause a slowdown.
so problem does not seem to be HP removal itself but the tcp_ack()
call. Therefore, it might be possible to remove HP after all, provided
one finds a way to elide tcp_ack for most cases.
Reported-by: Eric Dumazet <edumazet@google.com>
Signed-off-by: Florian Westphal <fw@strlen.de>
Signed-off-by: David S. Miller <davem@davemloft.net>
Diffstat (limited to 'net/ipv4/proc.c')
| -rw-r--r-- | net/ipv4/proc.c | 2 |
1 files changed, 2 insertions, 0 deletions
diff --git a/net/ipv4/proc.c b/net/ipv4/proc.c index b6d3fe03feb3..127153f1ed8a 100644 --- a/net/ipv4/proc.c +++ b/net/ipv4/proc.c @@ -206,7 +206,9 @@ static const struct snmp_mib snmp4_net_list[] = { SNMP_MIB_ITEM("DelayedACKLost", LINUX_MIB_DELAYEDACKLOST), SNMP_MIB_ITEM("ListenOverflows", LINUX_MIB_LISTENOVERFLOWS), SNMP_MIB_ITEM("ListenDrops", LINUX_MIB_LISTENDROPS), + SNMP_MIB_ITEM("TCPHPHits", LINUX_MIB_TCPHPHITS), SNMP_MIB_ITEM("TCPPureAcks", LINUX_MIB_TCPPUREACKS), + SNMP_MIB_ITEM("TCPHPAcks", LINUX_MIB_TCPHPACKS), SNMP_MIB_ITEM("TCPRenoRecovery", LINUX_MIB_TCPRENORECOVERY), SNMP_MIB_ITEM("TCPSackRecovery", LINUX_MIB_TCPSACKRECOVERY), SNMP_MIB_ITEM("TCPSACKReneging", LINUX_MIB_TCPSACKRENEGING), |