Introduce tcp_hostcache and remove the tcp specific metrics from

the routing table.  Move all usage and references in the tcp stack
from the routing table metrics to the tcp hostcache.

It caches measured parameters of past tcp sessions to provide better
initial start values for following connections from or to the same
source or destination.  Depending on the network parameters to/from
the remote host this can lead to significant speedups for new tcp
connections after the first one because they inherit and shortcut
the learning curve.

tcp_hostcache is designed for multiple concurrent access in SMP
environments with high contention and is hash indexed by remote
ip address.

It removes significant locking requirements from the tcp stack with
regard to the routing table.

Reviewed by:	sam (mentor), bms
Reviewed by:	-net, -current, core@kame.net (IPv6 parts)
Approved by:	re (scottl)

1 files changed, 20 insertions, 7 deletions

diff --git a/sys/netinet6/ip6_output.c b/sys/netinet6/ip6_output.c
index b95b197..3072851 100644
--- a/sys/netinet6/ip6_output.c
+++ b/sys/netinet6/ip6_output.c
@@ -96,6 +96,7 @@
 #include <netinet/icmp6.h>
 #include <netinet6/ip6_var.h>
 #include <netinet/in_pcb.h>
+#include <netinet/tcp_var.h>
 #include <netinet6/nd6.h>
 
 #ifdef IPSEC
@@ -661,7 +662,7 @@ skip_ipsec2:;
 		/* XXX rt not locked */
 		ia = ifatoia6(ro->ro_rt->rt_ifa);
 		ifp = ro->ro_rt->rt_ifp;
-		ro->ro_rt->rt_use++;
+		ro->ro_rt->rt_rmx.rmx_pksent++;
 		if (ro->ro_rt->rt_flags & RTF_GATEWAY)
 			dst = (struct sockaddr_in6 *)ro->ro_rt->rt_gateway;
 		m->m_flags &= ~(M_BCAST | M_MCAST);	/* just in case */
@@ -757,7 +758,7 @@ skip_ipsec2:;
 			}
 			ia = ifatoia6(ro->ro_rt->rt_ifa);
 			ifp = ro->ro_rt->rt_ifp;
-			ro->ro_rt->rt_use++;
+			ro->ro_rt->rt_rmx.rmx_pksent++;
 			RT_UNLOCK(ro->ro_rt);
 		}
 
@@ -1387,11 +1388,20 @@ ip6_getpmtu(ro_pmtu, ro, ifp, dst, mtup, alwaysfragp)
 	}
 	if (ro_pmtu->ro_rt) {
 		u_int32_t ifmtu;
+		struct in_conninfo inc;
+
+		bzero(&inc, sizeof(inc));
+		inc.inc_flags = 1; /* IPv6 */
+		inc.inc6_faddr = *dst;
 
 		if (ifp == NULL)
 			ifp = ro_pmtu->ro_rt->rt_ifp;
 		ifmtu = IN6_LINKMTU(ifp);
-		mtu = ro_pmtu->ro_rt->rt_rmx.rmx_mtu;
+		mtu = tcp_hc_getmtu(&inc);
+		if (mtu)
+			mtu = min(mtu, ro_pmtu->ro_rt->rt_rmx.rmx_mtu);
+		else
+			mtu = ro_pmtu->ro_rt->rt_rmx.rmx_mtu;
 		if (mtu == 0)
 			mtu = ifmtu;
 		else if (mtu < IPV6_MMTU) {
@@ -1415,8 +1425,7 @@ ip6_getpmtu(ro_pmtu, ro, ifp, dst, mtup, alwaysfragp)
 			 * field isn't locked).
 			 */
 			mtu = ifmtu;
-			if (!(ro_pmtu->ro_rt->rt_rmx.rmx_locks & RTV_MTU))
-				ro_pmtu->ro_rt->rt_rmx.rmx_mtu = mtu;
+			ro_pmtu->ro_rt->rt_rmx.rmx_mtu = mtu;
 		}
 	} else if (ifp) {
 		mtu = IN6_LINKMTU(ifp);
@@ -1993,7 +2002,9 @@ do { \
 			{
 				u_long pmtu = 0;
 				struct ip6_mtuinfo mtuinfo;
-				struct route_in6 *ro = (struct route_in6 *)&in6p->in6p_route;
+				struct route_in6 sro;
+
+				bzero(&sro, sizeof(sro));
 
 				if (!(so->so_state & SS_ISCONNECTED))
 					return (ENOTCONN);
@@ -2002,8 +2013,10 @@ do { \
 				 * routing, or optional information to specify
 				 * the outgoing interface.
 				 */
-				error = ip6_getpmtu(ro, NULL, NULL,
+				error = ip6_getpmtu(&sro, NULL, NULL,
 				    &in6p->in6p_faddr, &pmtu, NULL);
+				if (sro.ro_rt)
+					RTFREE(sro.ro_rt);
 				if (error)
 					break;
 				if (pmtu > IPV6_MAXPACKET)