[<prev] [next>] [day] [month] [year] [list]
Message-ID: <alpine.DEB.2.02.1407162056020.30101@tomh.mtv.corp.google.com>
Date: Wed, 16 Jul 2014 21:15:59 -0700 (PDT)
From: Tom Herbert <therbert@...gle.com>
To: davem@...emloft.net, netdev@...r.kernel.org
Subject: [PATCH net-next 8/9] ipip: RX path for IPIP/UDP Generic UDP
Encapsulation
Tested with TCP_RR netperf and 200 streams:
No GUE
70.86% CPU utilization
179/253/359 90/95/99% latencies
1.05203e+06 tps
With GUE
73.62% CPU utilization
183/265/389 90/95/99% latencies
1.02159e+06 tps
With GUE and UDP RSS
94.62% CPU utilization
155/247/458 90/95/99% latencies
1.16926e+06 tps
Signed-off-by: Tom Herbert <therbert@...gle.com>
---
net/ipv4/ipip.c | 10 ++++++++++
1 file changed, 10 insertions(+)
diff --git a/net/ipv4/ipip.c b/net/ipv4/ipip.c
index 91a1c89..75578fa 100644
--- a/net/ipv4/ipip.c
+++ b/net/ipv4/ipip.c
@@ -116,11 +116,14 @@
#include <net/xfrm.h>
#include <net/net_namespace.h>
#include <net/netns/generic.h>
+#include <net/gue.h>
static bool log_ecn_error = true;
module_param(log_ecn_error, bool, 0644);
MODULE_PARM_DESC(log_ecn_error, "Log packets received with corrupted ECN");
+GUE_TUNNEL_PARMS("IPIP", 0)
+
static int ipip_net_id __read_mostly;
static int ipip_tunnel_init(struct net_device *dev);
@@ -470,9 +473,15 @@ static int __init ipip_init(void)
if (err < 0)
goto rtnl_link_failed;
+ err = open_gue_port(IPPROTO_IPIP);
+ if (err < 0)
+ goto gue_udp_port_failed;
+
out:
return err;
+gue_udp_port_failed:
+ rtnl_link_unregister(&ipip_link_ops);
rtnl_link_failed:
xfrm4_tunnel_deregister(&ipip_handler, AF_INET);
xfrm_tunnel_failed:
@@ -482,6 +491,7 @@ xfrm_tunnel_failed:
static void __exit ipip_fini(void)
{
+ close_gue_port();
rtnl_link_unregister(&ipip_link_ops);
if (xfrm4_tunnel_deregister(&ipip_handler, AF_INET))
pr_info("%s: can't deregister tunnel\n", __func__);
--
2.0.0.526.g5318336
--
To unsubscribe from this list: send the line "unsubscribe netdev" in
the body of a message to majordomo@...r.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
Powered by blists - more mailing lists