[IPV4]: Provide full proto/ports in flowi route lookups.

include/net/route.h

@@ -139,12 +139,16 @@ static inline char rt_tos2priority(u8 tos)
 	return ip_tos2prio[IPTOS_TOS(tos)>>1];
 }
 
-static inline int ip_route_connect(struct rtable **rp, u32 dst, u32 src, u32 tos, int oif)
+static inline int ip_route_connect(struct rtable **rp, u32 dst, u32 src, u32 tos, int oif, u8 protocol, u16 sport, u16 dport)
 {
-	struct flowi fl = { .nl_u = { .ip4_u = { .daddr = dst,
+	struct flowi fl = { .oif = oif,
+			    .nl_u = { .ip4_u = { .daddr = dst,
 						 .saddr = src,
 						 .tos   = tos } },
-			    .oif = oif };
+			    .proto = protocol,
+			    .uli_u = { .ports =
+				       { .sport = sport,
+					 .dport = dport } } };
 
 	int err;
 	err = ip_route_output_key(rp, &fl);
@@ -157,6 +161,21 @@ static inline int ip_route_connect(struct rtable **rp, u32 dst, u32 src, u32 tos
 	return ip_route_output_key(rp, &fl);
 }
 
+static inline int ip_route_newports(struct rtable **rp, u16 sport, u16 dport)
+{
+	if (sport != (*rp)->fl.uli_u.ports.sport ||
+	    dport != (*rp)->fl.uli_u.ports.dport) {
+		struct flowi fl;
+
+		memcpy(&fl, &(*rp)->fl, sizeof(fl));
+		fl.uli_u.ports.sport = sport;
+		fl.uli_u.ports.dport = dport;
+		*rp = NULL;
+		return ip_route_output_key(rp, &fl);
+	}
+	return 0;
+}
+
 extern void rt_bind_peer(struct rtable *rt, int create);
 
 static inline struct inet_peer *rt_get_peer(struct rtable *rt)

net/ipv4/icmp.c

@@ -428,7 +428,8 @@ static void icmp_reply(struct icmp_bxm *icmp_param, struct sk_buff *skb)
 		struct flowi fl = { .nl_u = { .ip4_u =
 					      { .daddr = daddr,
 						.saddr = rt->rt_spec_dst,
-						.tos = RT_TOS(skb->nh.iph->tos) } } };
+						.tos = RT_TOS(skb->nh.iph->tos) } },
+				    .proto = IPPROTO_ICMP };
 		if (ip_route_output_key(&rt, &fl))
 			goto out_unlock;
 	}
@@ -550,7 +551,8 @@ void icmp_send(struct sk_buff *skb_in, int type, int code, u32 info)
 	{
 		struct flowi fl = { .nl_u = { .ip4_u = { .daddr = iph->saddr,
 							 .saddr = saddr,
-							 .tos = RT_TOS(tos) } } };
+							 .tos = RT_TOS(tos) } },
+				    .proto = IPPROTO_ICMP };
 		if (ip_route_output_key(&rt, &fl))
 		    goto out_unlock;
 	}
@@ -577,7 +579,8 @@ void icmp_send(struct sk_buff *skb_in, int type, int code, u32 info)
 		struct flowi fl = { .nl_u = { .ip4_u =
 					      { .daddr = icmp_param.replyopts.faddr,
 						.saddr = saddr,
-						.tos = RT_TOS(tos) } } };
+						.tos = RT_TOS(tos) } },
+				    .proto = IPPROTO_ICMP };
 		ip_rt_put(rt);
 		if (ip_route_output_key(&rt, &fl))
 			goto out_unlock;

net/ipv4/igmp.c

@@ -208,8 +208,9 @@ static int igmp_send_report(struct net_device *dev, u32 group, int type)
 		dst = IGMP_ALL_ROUTER;
 
 	{
-		struct flowi fl = { .nl_u = { .ip4_u = { .daddr = dst } },
-				    .oif = dev->ifindex };
+		struct flowi fl = { .oif = dev->ifindex,
+				    .nl_u = { .ip4_u = { .daddr = dst } },
+				    .proto = IPPROTO_IGMP };
 		if (ip_route_output_key(&rt, &fl))
 			return -1;
 	}

net/ipv4/ip_gre.c

@@ -492,6 +492,7 @@ void ipgre_err(struct sk_buff *skb, u32 info)
 	memset(&fl, 0, sizeof(fl));
 	fl.fl4_dst = eiph->saddr;
 	fl.fl4_tos = RT_TOS(eiph->tos);
+	fl.proto = IPPROTO_GRE;
 	if (ip_route_output_key(&rt, &fl)) {
 		kfree_skb(skb2);
 		return;
@@ -757,11 +758,12 @@ static int ipgre_tunnel_xmit(struct sk_buff *skb, struct net_device *dev)
 	}
 
 	{
-		struct flowi fl = { .nl_u = { .ip4_u =
+		struct flowi fl = { .oif = tunnel->parms.link,
+				    .nl_u = { .ip4_u =
 					      { .daddr = dst,
 						.saddr = tiph->saddr,
 						.tos = RT_TOS(tos) } },
-				    .oif = tunnel->parms.link };
+				    .proto = IPPROTO_GRE };
 		if (ip_route_output_key(&rt, &fl)) {
 			tunnel->stat.tx_carrier_errors++;
 			goto tx_error;
@@ -1118,11 +1120,12 @@ static int ipgre_open(struct net_device *dev)
 
 	MOD_INC_USE_COUNT;
 	if (MULTICAST(t->parms.iph.daddr)) {
-		struct flowi fl = { .nl_u = { .ip4_u =
+		struct flowi fl = { .oif = t->parms.link,
+				    .nl_u = { .ip4_u =
 					      { .daddr = t->parms.iph.daddr,
 						.saddr = t->parms.iph.saddr,
 						.tos = RT_TOS(t->parms.iph.tos) } },
-				    .oif = t->parms.link };
+				    .proto = IPPROTO_GRE };
 		struct rtable *rt;
 		if (ip_route_output_key(&rt, &fl)) {
 			MOD_DEC_USE_COUNT;
@@ -1194,11 +1197,12 @@ static int ipgre_tunnel_init(struct net_device *dev)
 	/* Guess output device to choose reasonable mtu and hard_header_len */
 
 	if (iph->daddr) {
-		struct flowi fl = { .nl_u = { .ip4_u =
+		struct flowi fl = { .oif = tunnel->parms.link,
+				    .nl_u = { .ip4_u =
 					      { .daddr = iph->daddr,
 						.saddr = iph->saddr,
 						.tos = RT_TOS(iph->tos) } },
-				    .oif = tunnel->parms.link };
+				    .proto = IPPROTO_GRE };
 		struct rtable *rt;
 		if (!ip_route_output_key(&rt, &fl)) {
 			tdev = rt->u.dst.dev;

net/ipv4/ip_nat_dumb.c

@@ -118,7 +118,9 @@ ip_do_nat(struct sk_buff *skb)
 				if (ciph->daddr != osaddr) {
 					struct   fib_result res;
 					unsigned flags = 0;
-					struct flowi fl = { .nl_u =
+					struct flowi fl = {
+						.iif = skb->dev->ifindex,
+						.nl_u =
 						{ .ip4_u =
 						  { .daddr = ciph->saddr,
 						    .saddr = ciph->daddr,
@@ -126,7 +128,7 @@ ip_do_nat(struct sk_buff *skb)
 						    .tos = RT_TOS(ciph->tos)
 #endif
 						  } },
-							    .iif = skb->dev->ifindex };
+						.proto = ciph->protocol };
 
 					/* Use fib_lookup() until we get our own
 					 * hash table of NATed hosts -- Rani

net/ipv4/ip_output.c

@@ -306,15 +306,19 @@ int ip_queue_xmit(struct sk_buff *skb)
 			daddr = opt->faddr;
 
 		{
-			struct flowi fl = { .nl_u = { .ip4_u =
+			struct flowi fl = { .oif = sk->bound_dev_if,
+					    .nl_u = { .ip4_u =
 						      { .daddr = daddr,
 							.saddr = inet->saddr,
 							.tos = RT_CONN_FLAGS(sk) } },
-					    .oif = sk->bound_dev_if };
+					    .proto = sk->protocol,
+					    .uli_u = { .ports =
+						       { .sport = inet->sport,
+							 .dport = inet->dport } } };
 
 			/* If this fails, retransmit mechanism of transport layer will
-			 * keep trying until route appears or the connection times itself
-			 * out.
+			 * keep trying until route appears or the connection times
+			 * itself out.
 			 */
 			if (ip_route_output_key(&rt, &fl))
 				goto no_route;
@@ -1206,7 +1210,8 @@ void ip_send_reply(struct sock *sk, struct sk_buff *skb, struct ip_reply_arg *ar
 		struct flowi fl = { .nl_u = { .ip4_u =
 					      { .daddr = daddr,
 						.saddr = rt->rt_spec_dst,
-						.tos = RT_TOS(skb->nh.iph->tos) } } };
+						.tos = RT_TOS(skb->nh.iph->tos) } },
+				    .proto = sk->protocol };
 		if (ip_route_output_key(&rt, &fl))
 			return;
 	}

net/ipv4/ipip.c

@@ -421,6 +421,7 @@ void ipip_err(struct sk_buff *skb, u32 info)
 	memset(&fl, 0, sizeof(fl));
 	fl.fl4_daddr = eiph->saddr;
 	fl.fl4_tos = RT_TOS(eiph->tos);
+	fl.proto = IPPROTO_IPIP;
 	if (ip_route_output_key(&rt, &key)) {
 		kfree_skb(skb2);
 		return;
@@ -568,11 +569,12 @@ static int ipip_tunnel_xmit(struct sk_buff *skb, struct net_device *dev)
 	}
 
 	{
-		struct flowi fl = { .nl_u = { .ip4_u =
+		struct flowi fl = { .oif = tunnel->parms.link,
+				    .nl_u = { .ip4_u =
 					      { .daddr = dst,
 						.saddr = tiph->saddr,
 						.tos = RT_TOS(tos) } },
-				    .oif = tunnel->parms.link };
+				    .proto = IPPROTO_IPIP };
 		if (ip_route_output_key(&rt, &fl)) {
 			tunnel->stat.tx_carrier_errors++;
 			goto tx_error_icmp;
@@ -836,11 +838,12 @@ static int ipip_tunnel_init(struct net_device *dev)
 	ipip_tunnel_init_gen(dev);
 
 	if (iph->daddr) {
-		struct flowi fl = { .nl_u = { .ip4_u =
+		struct flowi fl = { .oif = tunnel->parms.link,
+				    .nl_u = { .ip4_u =
 					      { .daddr = iph->daddr,
 						.saddr = iph->saddr,
 						.tos = RT_TOS(iph->tos) } },
-				    .oif = tunnel->parms.link };
+				    .proto = IPPROTO_IPIP };
 		struct rtable *rt;
 		if (!ip_route_output_key(&rt, &fl)) {
 			tdev = rt->u.dst.dev;

net/ipv4/ipmr.c

@@ -1146,19 +1146,21 @@ static void ipmr_queue_xmit(struct sk_buff *skb, struct mfc_cache *c,
 #endif
 
 	if (vif->flags&VIFF_TUNNEL) {
-		struct flowi fl = { .nl_u = { .ip4_u =
+		struct flowi fl = { .oif = vif->link,
+				    .nl_u = { .ip4_u =
 					      { .daddr = vif->remote,
 						.saddr = vif->local,
 						.tos = RT_TOS(iph->tos) } },
-				    .oif = vif->link };
+				    .proto = IPPROTO_IPIP };
 		if (ip_route_output_key(&rt, &fl))
 			return;
 		encap = sizeof(struct iphdr);
 	} else {
-		struct flowi fl = { .nl_u = { .ip4_u =
+		struct flowi fl = { .oif = vif->link,
+				    .nl_u = { .ip4_u =
 					      { .daddr = iph->daddr,
 						.tos = RT_TOS(iph->tos) } },
-				    .oif = vif->link };
+				    .proto = IPPROTO_IPIP };
 		if (ip_route_output_key(&rt, &fl))
 			return;
 	}

net/ipv4/raw.c

@@ -420,11 +420,12 @@ static int raw_sendmsg(struct kiocb *iocb, struct sock *sk, struct msghdr *msg,
 	}
 
 	{
-		struct flowi fl = { .nl_u = { .ip4_u =
+		struct flowi fl = { .oif = ipc.oif,
+				    .nl_u = { .ip4_u =
 					      { .daddr = daddr,
 						.saddr = saddr,
 						.tos = tos } },
-				    .oif = ipc.oif };
+				    .proto = IPPROTO_RAW };
 		err = ip_route_output_key(&rt, &fl);
 	}
 	if (err)

net/ipv4/syncookies.c

@@ -177,7 +177,11 @@ struct sock *cookie_v4_check(struct sock *sk, struct sk_buff *skb,
 							  opt->faddr :
 							  req->af.v4_req.rmt_addr),
 						.saddr = req->af.v4_req.loc_addr,
-						.tos = RT_CONN_FLAGS(sk) } } };
+						.tos = RT_CONN_FLAGS(sk) } },
+				    .proto = IPPROTO_TCP,
+				    .uli_u = { .ports =
+					       { .sport = skb->h.th->dest,
+						 .dport = skb->h.th->source } } };
 		if (ip_route_output_key(&rt, &fl)) {
 			tcp_openreq_free(req);
 			goto out; 

net/ipv4/tcp_ipv4.c

@@ -770,7 +770,9 @@ int tcp_v4_connect(struct sock *sk, struct sockaddr *uaddr, int addr_len)
 	}
 
 	tmp = ip_route_connect(&rt, nexthop, inet->saddr,
-			       RT_CONN_FLAGS(sk), sk->bound_dev_if);
+			       RT_CONN_FLAGS(sk), sk->bound_dev_if,
+			       IPPROTO_TCP,
+			       inet->sport, usin->sin_port);
 	if (tmp < 0)
 		return tmp;
 
@@ -779,10 +781,6 @@ int tcp_v4_connect(struct sock *sk, struct sockaddr *uaddr, int addr_len)
 		return -ENETUNREACH;
 	}
 
-	__sk_dst_set(sk, &rt->u.dst);
-	tcp_v4_setup_caps(sk, &rt->u.dst);
-	tp->ext_header_len += rt->u.dst.header_len;
-
 	if (!inet->opt || !inet->opt->srr)
 		daddr = rt->rt_dst;
 
@@ -831,6 +829,19 @@ int tcp_v4_connect(struct sock *sk, struct sockaddr *uaddr, int addr_len)
 	if (err)
 		goto failure;
 
+	err = ip_route_newports(&rt, inet->sport, inet->dport);
+	if (err)
+		goto failure;
+
+	/* OK, now commit destination to socket.  */
+	__sk_dst_set(sk, &rt->u.dst);
+	tcp_v4_setup_caps(sk, &rt->u.dst);
+
+	/* DAVEM REDPEN: This used to sit above forced ext_header_len = 0
+	 *               above, it was real bug.  Is this one correct?
+	 */
+	tp->ext_header_len += rt->u.dst.header_len;
+
 	if (!tp->write_seq)
 		tp->write_seq = secure_tcp_sequence_number(inet->saddr,
 							   inet->daddr,
@@ -846,8 +857,9 @@ int tcp_v4_connect(struct sock *sk, struct sockaddr *uaddr, int addr_len)
 	return 0;
 
 failure:
+	/* This unhashes the socket and releases the local port, if necessary. */
 	tcp_set_state(sk, TCP_CLOSE);
-	__sk_dst_reset(sk);
+	ip_rt_put(rt);
 	sk->route_caps = 0;
 	inet->dport = 0;
 	return err;
@@ -1265,13 +1277,17 @@ static struct dst_entry* tcp_v4_route_req(struct sock *sk,
 {
 	struct rtable *rt;
 	struct ip_options *opt = req->af.v4_req.opt;
-	struct flowi fl = { .nl_u = { .ip4_u =
+	struct flowi fl = { .oif = sk->bound_dev_if,
+			    .nl_u = { .ip4_u =
 				      { .daddr = ((opt && opt->srr) ?
 						  opt->faddr :
 						  req->af.v4_req.rmt_addr),
 					.saddr = req->af.v4_req.loc_addr,
 					.tos = RT_CONN_FLAGS(sk) } },
-			    .oif = sk->bound_dev_if };
+			    .proto = IPPROTO_TCP,
+			    .uli_u = { .ports =
+				       { .sport = inet_sk(sk)->sport,
+					 .dport = req->rmt_port } } };
 
 	if (ip_route_output_key(&rt, &fl)) {
 		IP_INC_STATS_BH(IpOutNoRoutes);
@@ -1864,7 +1880,9 @@ static int tcp_v4_reselect_saddr(struct sock *sk)
 	/* Query new route. */
 	err = ip_route_connect(&rt, daddr, 0,
 			       RT_TOS(inet->tos) | sk->localroute,
-			       sk->bound_dev_if);
+			       sk->bound_dev_if,
+			       IPPROTO_TCP,
+			       inet->sport, inet->dport);
 	if (err)
 		return err;
 
@@ -1914,11 +1932,15 @@ int tcp_v4_rebuild_header(struct sock *sk)
 		daddr = inet->opt->faddr;
 
 	{
-		struct flowi fl = { .nl_u = { .ip4_u =
+		struct flowi fl = { .oif = sk->bound_dev_if,
+				    .nl_u = { .ip4_u =
 					      { .daddr = daddr,
 						.saddr = inet->saddr,
 						.tos = RT_CONN_FLAGS(sk) } },
-				    .oif = sk->bound_dev_if };
+				    .proto = IPPROTO_TCP,
+				    .uli_u = { .ports =
+					       { .sport = inet->sport,
+						 .dport = inet->dport } } };
 						
 		err = ip_route_output_key(&rt, &fl);
 	}

net/ipv4/udp.c

@@ -586,11 +586,15 @@ int udp_sendmsg(struct kiocb *iocb, struct sock *sk, struct msghdr *msg,
 		rt = (struct rtable*)sk_dst_check(sk, 0);
 
 	if (rt == NULL) {
-		struct flowi fl = { .nl_u = { .ip4_u =
+		struct flowi fl = { .oif = ipc.oif,
+				    .nl_u = { .ip4_u =
 					      { .daddr = faddr,
 						.saddr = saddr,
 						.tos = tos } },
-				    .oif = ipc.oif };
+				    .proto = IPPROTO_UDP,
+				    .uli_u = { .ports =
+					       { .sport = inet->sport,
+						 .dport = dport } } };
 		err = ip_route_output_key(&rt, &fl);
 		if (err)
 			goto out;
@@ -872,7 +876,9 @@ int udp_connect(struct sock *sk, struct sockaddr *uaddr, int addr_len)
 			saddr = inet->mc_addr;
 	}
 	err = ip_route_connect(&rt, usin->sin_addr.s_addr, saddr,
-			       RT_CONN_FLAGS(sk), oif);
+			       RT_CONN_FLAGS(sk), oif,
+			       IPPROTO_UDP,
+			       inet->sport, usin->sin_port);
 	if (err)
 		return err;
 	if ((rt->rt_flags&RTCF_BROADCAST) && !sk->broadcast) {