Merge branch 'ip_tunnel-collect_md'

Alexei Starovoitov says:

====================
ip_tunnel: add collect_md mode to IPv4/IPv6 tunnels

Similar to geneve, vxlan, gre tunnels implement 'collect metadata' mode
in ipip, ipip6, ip6ip6 tunnels.
====================
Signed-off-by: David S. Miller <davem@davemloft.net>

include/net/ip6_tunnel.h

@@ -23,6 +23,7 @@ struct __ip6_tnl_parm {
 	__u8 proto;		/* tunnel protocol */
 	__u8 encap_limit;	/* encapsulation limit for tunnel */
 	__u8 hop_limit;		/* hop limit for tunnel */
+	bool collect_md;
 	__be32 flowinfo;	/* traffic class and flowlabel for tunnel */
 	__u32 flags;		/* tunnel flags */
 	struct in6_addr laddr;	/* local tunnel end-point address */

include/net/ip_tunnels.h

@@ -255,6 +255,8 @@ void ip_tunnel_delete_net(struct ip_tunnel_net *itn, struct rtnl_link_ops *ops);
 
 void ip_tunnel_xmit(struct sk_buff *skb, struct net_device *dev,
 		    const struct iphdr *tnl_params, const u8 protocol);
+void ip_md_tunnel_xmit(struct sk_buff *skb, struct net_device *dev,
+		       const u8 proto);
 int ip_tunnel_ioctl(struct net_device *dev, struct ip_tunnel_parm *p, int cmd);
 int __ip_tunnel_change_mtu(struct net_device *dev, int new_mtu, bool strict);
 int ip_tunnel_change_mtu(struct net_device *dev, int new_mtu);

include/uapi/linux/if_tunnel.h

@@ -73,6 +73,7 @@ enum {
 	IFLA_IPTUN_ENCAP_FLAGS,
 	IFLA_IPTUN_ENCAP_SPORT,
 	IFLA_IPTUN_ENCAP_DPORT,
+	IFLA_IPTUN_COLLECT_METADATA,
 	__IFLA_IPTUN_MAX,
 };
 #define IFLA_IPTUN_MAX	(__IFLA_IPTUN_MAX - 1)

net/ipv4/ip_tunnel.c

@@ -55,6 +55,7 @@
 #include <net/netns/generic.h>
 #include <net/rtnetlink.h>
 #include <net/udp.h>
+#include <net/dst_metadata.h>
 
 #if IS_ENABLED(CONFIG_IPV6)
 #include <net/ipv6.h>
@@ -546,6 +547,81 @@ static int tnl_update_pmtu(struct net_device *dev, struct sk_buff *skb,
 	return 0;
 }
 
+void ip_md_tunnel_xmit(struct sk_buff *skb, struct net_device *dev, u8 proto)
+{
+	struct ip_tunnel *tunnel = netdev_priv(dev);
+	u32 headroom = sizeof(struct iphdr);
+	struct ip_tunnel_info *tun_info;
+	const struct ip_tunnel_key *key;
+	const struct iphdr *inner_iph;
+	struct rtable *rt;
+	struct flowi4 fl4;
+	__be16 df = 0;
+	u8 tos, ttl;
+
+	tun_info = skb_tunnel_info(skb);
+	if (unlikely(!tun_info || !(tun_info->mode & IP_TUNNEL_INFO_TX) ||
+		     ip_tunnel_info_af(tun_info) != AF_INET))
+		goto tx_error;
+	key = &tun_info->key;
+	memset(&(IPCB(skb)->opt), 0, sizeof(IPCB(skb)->opt));
+	inner_iph = (const struct iphdr *)skb_inner_network_header(skb);
+	tos = key->tos;
+	if (tos == 1) {
+		if (skb->protocol == htons(ETH_P_IP))
+			tos = inner_iph->tos;
+		else if (skb->protocol == htons(ETH_P_IPV6))
+			tos = ipv6_get_dsfield((const struct ipv6hdr *)inner_iph);
+	}
+	init_tunnel_flow(&fl4, proto, key->u.ipv4.dst, key->u.ipv4.src, 0,
+			 RT_TOS(tos), tunnel->parms.link);
+	if (tunnel->encap.type != TUNNEL_ENCAP_NONE)
+		goto tx_error;
+	rt = ip_route_output_key(tunnel->net, &fl4);
+	if (IS_ERR(rt)) {
+		dev->stats.tx_carrier_errors++;
+		goto tx_error;
+	}
+	if (rt->dst.dev == dev) {
+		ip_rt_put(rt);
+		dev->stats.collisions++;
+		goto tx_error;
+	}
+	tos = ip_tunnel_ecn_encap(tos, inner_iph, skb);
+	ttl = key->ttl;
+	if (ttl == 0) {
+		if (skb->protocol == htons(ETH_P_IP))
+			ttl = inner_iph->ttl;
+		else if (skb->protocol == htons(ETH_P_IPV6))
+			ttl = ((const struct ipv6hdr *)inner_iph)->hop_limit;
+		else
+			ttl = ip4_dst_hoplimit(&rt->dst);
+	}
+	if (key->tun_flags & TUNNEL_DONT_FRAGMENT)
+		df = htons(IP_DF);
+	else if (skb->protocol == htons(ETH_P_IP))
+		df = inner_iph->frag_off & htons(IP_DF);
+	headroom += LL_RESERVED_SPACE(rt->dst.dev) + rt->dst.header_len;
+	if (headroom > dev->needed_headroom)
+		dev->needed_headroom = headroom;
+
+	if (skb_cow_head(skb, dev->needed_headroom)) {
+		ip_rt_put(rt);
+		goto tx_dropped;
+	}
+	iptunnel_xmit(NULL, rt, skb, fl4.saddr, fl4.daddr, proto, key->tos,
+		      key->ttl, df, !net_eq(tunnel->net, dev_net(dev)));
+	return;
+tx_error:
+	dev->stats.tx_errors++;
+	goto kfree;
+tx_dropped:
+	dev->stats.tx_dropped++;
+kfree:
+	kfree_skb(skb);
+}
+EXPORT_SYMBOL_GPL(ip_md_tunnel_xmit);
+
 void ip_tunnel_xmit(struct sk_buff *skb, struct net_device *dev,
 		    const struct iphdr *tnl_params, u8 protocol)
 {

net/ipv4/ipip.c

@@ -115,6 +115,7 @@
 #include <net/xfrm.h>
 #include <net/net_namespace.h>
 #include <net/netns/generic.h>
+#include <net/dst_metadata.h>
 
 static bool log_ecn_error = true;
 module_param(log_ecn_error, bool, 0644);
@@ -193,6 +194,7 @@ static int ipip_tunnel_rcv(struct sk_buff *skb, u8 ipproto)
 {
 	struct net *net = dev_net(skb->dev);
 	struct ip_tunnel_net *itn = net_generic(net, ipip_net_id);
+	struct metadata_dst *tun_dst = NULL;
 	struct ip_tunnel *tunnel;
 	const struct iphdr *iph;
 
@@ -216,7 +218,12 @@ static int ipip_tunnel_rcv(struct sk_buff *skb, u8 ipproto)
 			tpi = &ipip_tpi;
 		if (iptunnel_pull_header(skb, 0, tpi->proto, false))
 			goto drop;
-		return ip_tunnel_rcv(tunnel, skb, tpi, NULL, log_ecn_error);
+		if (tunnel->collect_md) {
+			tun_dst = ip_tun_rx_dst(skb, 0, 0, 0);
+			if (!tun_dst)
+				return 0;
+		}
+		return ip_tunnel_rcv(tunnel, skb, tpi, tun_dst, log_ecn_error);
 	}
 
 	return -1;
@@ -270,6 +277,9 @@ static netdev_tx_t ipip_tunnel_xmit(struct sk_buff *skb,
 
 	skb_set_inner_ipproto(skb, ipproto);
 
+	if (tunnel->collect_md)
+		ip_md_tunnel_xmit(skb, dev, ipproto);
+	else
 		ip_tunnel_xmit(skb, dev, tiph, ipproto);
 	return NETDEV_TX_OK;
 
@@ -380,13 +390,14 @@ static int ipip_tunnel_validate(struct nlattr *tb[], struct nlattr *data[])
 }
 
 static void ipip_netlink_parms(struct nlattr *data[],
-			       struct ip_tunnel_parm *parms)
+			       struct ip_tunnel_parm *parms, bool *collect_md)
 {
 	memset(parms, 0, sizeof(*parms));
 
 	parms->iph.version = 4;
 	parms->iph.protocol = IPPROTO_IPIP;
 	parms->iph.ihl = 5;
+	*collect_md = false;
 
 	if (!data)
 		return;
@@ -414,6 +425,9 @@ static void ipip_netlink_parms(struct nlattr *data[],
 
 	if (!data[IFLA_IPTUN_PMTUDISC] || nla_get_u8(data[IFLA_IPTUN_PMTUDISC]))
 		parms->iph.frag_off = htons(IP_DF);
+
+	if (data[IFLA_IPTUN_COLLECT_METADATA])
+		*collect_md = true;
 }
 
 /* This function returns true when ENCAP attributes are present in the nl msg */
@@ -453,18 +467,18 @@ static bool ipip_netlink_encap_parms(struct nlattr *data[],
 static int ipip_newlink(struct net *src_net, struct net_device *dev,
 			struct nlattr *tb[], struct nlattr *data[])
 {
+	struct ip_tunnel *t = netdev_priv(dev);
 	struct ip_tunnel_parm p;
 	struct ip_tunnel_encap ipencap;
 
 	if (ipip_netlink_encap_parms(data, &ipencap)) {
-		struct ip_tunnel *t = netdev_priv(dev);
 		int err = ip_tunnel_encap_setup(t, &ipencap);
 
 		if (err < 0)
 			return err;
 	}
 
-	ipip_netlink_parms(data, &p);
+	ipip_netlink_parms(data, &p, &t->collect_md);
 	return ip_tunnel_newlink(dev, tb, &p);
 }
 
@@ -473,6 +487,7 @@ static int ipip_changelink(struct net_device *dev, struct nlattr *tb[],
 {
 	struct ip_tunnel_parm p;
 	struct ip_tunnel_encap ipencap;
+	bool collect_md;
 
 	if (ipip_netlink_encap_parms(data, &ipencap)) {
 		struct ip_tunnel *t = netdev_priv(dev);
@@ -482,7 +497,9 @@ static int ipip_changelink(struct net_device *dev, struct nlattr *tb[],
 			return err;
 	}
 
-	ipip_netlink_parms(data, &p);
+	ipip_netlink_parms(data, &p, &collect_md);
+	if (collect_md)
+		return -EINVAL;
 
 	if (((dev->flags & IFF_POINTOPOINT) && !p.iph.daddr) ||
 	    (!(dev->flags & IFF_POINTOPOINT) && p.iph.daddr))
@@ -516,6 +533,8 @@ static size_t ipip_get_size(const struct net_device *dev)
 		nla_total_size(2) +
 		/* IFLA_IPTUN_ENCAP_DPORT */
 		nla_total_size(2) +
+		/* IFLA_IPTUN_COLLECT_METADATA */
+		nla_total_size(0) +
 		0;
 }
 
@@ -544,6 +563,9 @@ static int ipip_fill_info(struct sk_buff *skb, const struct net_device *dev)
 			tunnel->encap.flags))
 		goto nla_put_failure;
 
+	if (tunnel->collect_md)
+		if (nla_put_flag(skb, IFLA_IPTUN_COLLECT_METADATA))
+			goto nla_put_failure;
 	return 0;
 
 nla_put_failure:
@@ -562,6 +584,7 @@ static const struct nla_policy ipip_policy[IFLA_IPTUN_MAX + 1] = {
 	[IFLA_IPTUN_ENCAP_FLAGS]	= { .type = NLA_U16 },
 	[IFLA_IPTUN_ENCAP_SPORT]	= { .type = NLA_U16 },
 	[IFLA_IPTUN_ENCAP_DPORT]	= { .type = NLA_U16 },
+	[IFLA_IPTUN_COLLECT_METADATA]	= { .type = NLA_FLAG },
 };
 
 static struct rtnl_link_ops ipip_link_ops __read_mostly = {

net/ipv6/ip6_tunnel.c

@@ -57,6 +57,7 @@
 #include <net/inet_ecn.h>
 #include <net/net_namespace.h>
 #include <net/netns/generic.h>
+#include <net/dst_metadata.h>
 
 MODULE_AUTHOR("Ville Nuorvala");
 MODULE_DESCRIPTION("IPv6 tunneling device");
@@ -90,6 +91,7 @@ struct ip6_tnl_net {
 	struct ip6_tnl __rcu *tnls_r_l[IP6_TUNNEL_HASH_SIZE];
 	struct ip6_tnl __rcu *tnls_wc[1];
 	struct ip6_tnl __rcu **tnls[2];
+	struct ip6_tnl __rcu *collect_md_tun;
 };
 
 static struct net_device_stats *ip6_get_stats(struct net_device *dev)
@@ -166,6 +168,10 @@ ip6_tnl_lookup(struct net *net, const struct in6_addr *remote, const struct in6_
 			return t;
 	}
 
+	t = rcu_dereference(ip6n->collect_md_tun);
+	if (t)
+		return t;
+
 	t = rcu_dereference(ip6n->tnls_wc[0]);
 	if (t && (t->dev->flags & IFF_UP))
 		return t;
@@ -209,6 +215,8 @@ ip6_tnl_link(struct ip6_tnl_net *ip6n, struct ip6_tnl *t)
 {
 	struct ip6_tnl __rcu **tp = ip6_tnl_bucket(ip6n, &t->parms);
 
+	if (t->parms.collect_md)
+		rcu_assign_pointer(ip6n->collect_md_tun, t);
 	rcu_assign_pointer(t->next , rtnl_dereference(*tp));
 	rcu_assign_pointer(*tp, t);
 }
@@ -224,6 +232,9 @@ ip6_tnl_unlink(struct ip6_tnl_net *ip6n, struct ip6_tnl *t)
 	struct ip6_tnl __rcu **tp;
 	struct ip6_tnl *iter;
 
+	if (t->parms.collect_md)
+		rcu_assign_pointer(ip6n->collect_md_tun, NULL);
+
 	for (tp = ip6_tnl_bucket(ip6n, &t->parms);
 	     (iter = rtnl_dereference(*tp)) != NULL;
 	     tp = &iter->next) {
@@ -829,6 +840,9 @@ static int __ip6_tnl_rcv(struct ip6_tnl *tunnel, struct sk_buff *skb,
 
 	skb_scrub_packet(skb, !net_eq(tunnel->net, dev_net(tunnel->dev)));
 
+	if (tun_dst)
+		skb_dst_set(skb, (struct dst_entry *)tun_dst);
+
 	gro_cells_receive(&tunnel->gro_cells, skb);
 	return 0;
 
@@ -865,6 +879,7 @@ static int ipxip6_rcv(struct sk_buff *skb, u8 ipproto,
 {
 	struct ip6_tnl *t;
 	const struct ipv6hdr *ipv6h = ipv6_hdr(skb);
+	struct metadata_dst *tun_dst = NULL;
 	int ret = -1;
 
 	rcu_read_lock();
@@ -881,7 +896,12 @@ static int ipxip6_rcv(struct sk_buff *skb, u8 ipproto,
 			goto drop;
 		if (iptunnel_pull_header(skb, 0, tpi->proto, false))
 			goto drop;
-		ret = __ip6_tnl_rcv(t, skb, tpi, NULL, dscp_ecn_decapsulate,
+		if (t->parms.collect_md) {
+			tun_dst = ipv6_tun_rx_dst(skb, 0, 0, 0);
+			if (!tun_dst)
+				return 0;
+		}
+		ret = __ip6_tnl_rcv(t, skb, tpi, tun_dst, dscp_ecn_decapsulate,
 				    log_ecn_error);
 	}
 
@@ -1012,8 +1032,16 @@ int ip6_tnl_xmit(struct sk_buff *skb, struct net_device *dev, __u8 dsfield,
 	int mtu;
 	unsigned int psh_hlen = sizeof(struct ipv6hdr) + t->encap_hlen;
 	unsigned int max_headroom = psh_hlen;
+	u8 hop_limit;
 	int err = -1;
 
+	if (t->parms.collect_md) {
+		hop_limit = skb_tunnel_info(skb)->key.ttl;
+		goto route_lookup;
+	} else {
+		hop_limit = t->parms.hop_limit;
+	}
+
 	/* NBMA tunnel */
 	if (ipv6_addr_any(&t->parms.raddr)) {
 		struct in6_addr *addr6;
@@ -1043,6 +1071,7 @@ int ip6_tnl_xmit(struct sk_buff *skb, struct net_device *dev, __u8 dsfield,
 		goto tx_err_link_failure;
 
 	if (!dst) {
+route_lookup:
 		dst = ip6_route_output(net, NULL, fl6);
 
 		if (dst->error)
@@ -1053,6 +1082,10 @@ int ip6_tnl_xmit(struct sk_buff *skb, struct net_device *dev, __u8 dsfield,
 			dst = NULL;
 			goto tx_err_link_failure;
 		}
+		if (t->parms.collect_md &&
+		    ipv6_dev_get_saddr(net, ip6_dst_idev(dst)->dev,
+				       &fl6->daddr, 0, &fl6->saddr))
+			goto tx_err_link_failure;
 		ndst = dst;
 	}
 
@@ -1071,7 +1104,7 @@ int ip6_tnl_xmit(struct sk_buff *skb, struct net_device *dev, __u8 dsfield,
 	}
 	if (mtu < IPV6_MIN_MTU)
 		mtu = IPV6_MIN_MTU;
-	if (skb_dst(skb))
+	if (skb_dst(skb) && !t->parms.collect_md)
 		skb_dst(skb)->ops->update_pmtu(skb_dst(skb), NULL, skb, mtu);
 	if (skb->len > mtu && !skb_is_gso(skb)) {
 		*pmtu = mtu;
@@ -1111,8 +1144,13 @@ int ip6_tnl_xmit(struct sk_buff *skb, struct net_device *dev, __u8 dsfield,
 		skb = new_skb;
 	}
 
+	if (t->parms.collect_md) {
+		if (t->encap.type != TUNNEL_ENCAP_NONE)
+			goto tx_err_dst_release;
+	} else {
 		if (!fl6->flowi6_mark && ndst)
 			dst_cache_set_ip6(&t->dst_cache, ndst, &fl6->saddr);
+	}
 	skb_dst_set(skb, dst);
 
 	if (encap_limit >= 0) {
@@ -1137,7 +1175,7 @@ int ip6_tnl_xmit(struct sk_buff *skb, struct net_device *dev, __u8 dsfield,
 	ipv6h = ipv6_hdr(skb);
 	ip6_flow_hdr(ipv6h, INET_ECN_encapsulate(0, dsfield),
 		     ip6_make_flowlabel(net, skb, fl6->flowlabel, true, fl6));
-	ipv6h->hop_limit = t->parms.hop_limit;
+	ipv6h->hop_limit = hop_limit;
 	ipv6h->nexthdr = proto;
 	ipv6h->saddr = fl6->saddr;
 	ipv6h->daddr = fl6->daddr;
@@ -1170,19 +1208,34 @@ ip4ip6_tnl_xmit(struct sk_buff *skb, struct net_device *dev)
 	if (tproto != IPPROTO_IPIP && tproto != 0)
 		return -1;
 
+	dsfield = ipv4_get_dsfield(iph);
+
+	if (t->parms.collect_md) {
+		struct ip_tunnel_info *tun_info;
+		const struct ip_tunnel_key *key;
+
+		tun_info = skb_tunnel_info(skb);
+		if (unlikely(!tun_info || !(tun_info->mode & IP_TUNNEL_INFO_TX) ||
+			     ip_tunnel_info_af(tun_info) != AF_INET6))
+			return -1;
+		key = &tun_info->key;
+		memset(&fl6, 0, sizeof(fl6));
+		fl6.flowi6_proto = IPPROTO_IPIP;
+		fl6.daddr = key->u.ipv6.dst;
+		fl6.flowlabel = key->label;
+	} else {
 		if (!(t->parms.flags & IP6_TNL_F_IGN_ENCAP_LIMIT))
 			encap_limit = t->parms.encap_limit;
 
 		memcpy(&fl6, &t->fl.u.ip6, sizeof(fl6));
 		fl6.flowi6_proto = IPPROTO_IPIP;
 
-	dsfield = ipv4_get_dsfield(iph);
-
 		if (t->parms.flags & IP6_TNL_F_USE_ORIG_TCLASS)
 			fl6.flowlabel |= htonl((__u32)iph->tos << IPV6_TCLASS_SHIFT)
 					 & IPV6_TCLASS_MASK;
 		if (t->parms.flags & IP6_TNL_F_USE_ORIG_FWMARK)
 			fl6.flowi6_mark = skb->mark;
+	}
 
 	if (iptunnel_handle_offloads(skb, SKB_GSO_IPXIP6))
 		return -1;
@@ -1220,29 +1273,47 @@ ip6ip6_tnl_xmit(struct sk_buff *skb, struct net_device *dev)
 	    ip6_tnl_addr_conflict(t, ipv6h))
 		return -1;
 
+	dsfield = ipv6_get_dsfield(ipv6h);
+
+	if (t->parms.collect_md) {
+		struct ip_tunnel_info *tun_info;
+		const struct ip_tunnel_key *key;
+
+		tun_info = skb_tunnel_info(skb);
+		if (unlikely(!tun_info || !(tun_info->mode & IP_TUNNEL_INFO_TX) ||
+			     ip_tunnel_info_af(tun_info) != AF_INET6))
+			return -1;
+		key = &tun_info->key;
+		memset(&fl6, 0, sizeof(fl6));
+		fl6.flowi6_proto = IPPROTO_IPV6;
+		fl6.daddr = key->u.ipv6.dst;
+		fl6.flowlabel = key->label;
+	} else {
 		offset = ip6_tnl_parse_tlv_enc_lim(skb, skb_network_header(skb));
 		if (offset > 0) {
 			struct ipv6_tlv_tnl_enc_lim *tel;
-		tel = (struct ipv6_tlv_tnl_enc_lim *)&skb_network_header(skb)[offset];
+
+			tel = (void *)&skb_network_header(skb)[offset];
 			if (tel->encap_limit == 0) {
 				icmpv6_send(skb, ICMPV6_PARAMPROB,
 					    ICMPV6_HDR_FIELD, offset + 2);
 				return -1;
 			}
 			encap_limit = tel->encap_limit - 1;
-	} else if (!(t->parms.flags & IP6_TNL_F_IGN_ENCAP_LIMIT))
+		} else if (!(t->parms.flags & IP6_TNL_F_IGN_ENCAP_LIMIT)) {
 			encap_limit = t->parms.encap_limit;
+		}
 
 		memcpy(&fl6, &t->fl.u.ip6, sizeof(fl6));
 		fl6.flowi6_proto = IPPROTO_IPV6;
 
-	dsfield = ipv6_get_dsfield(ipv6h);
 		if (t->parms.flags & IP6_TNL_F_USE_ORIG_TCLASS)
-		fl6.flowlabel |= (*(__be32 *) ipv6h & IPV6_TCLASS_MASK);
+			fl6.flowlabel |= (*(__be32 *)ipv6h & IPV6_TCLASS_MASK);
 		if (t->parms.flags & IP6_TNL_F_USE_ORIG_FLOWLABEL)
 			fl6.flowlabel |= ip6_flowlabel(ipv6h);
 		if (t->parms.flags & IP6_TNL_F_USE_ORIG_FWMARK)
 			fl6.flowi6_mark = skb->mark;
+	}
 
 	if (iptunnel_handle_offloads(skb, SKB_GSO_IPXIP6))
 		return -1;
@@ -1741,6 +1812,10 @@ static int ip6_tnl_dev_init(struct net_device *dev)
 	if (err)
 		return err;
 	ip6_tnl_link_config(t);
+	if (t->parms.collect_md) {
+		dev->features |= NETIF_F_NETNS_LOCAL;
+		netif_keep_dst(dev);
+	}
 	return 0;
 }
 
@@ -1811,6 +1886,9 @@ static void ip6_tnl_netlink_parms(struct nlattr *data[],
 
 	if (data[IFLA_IPTUN_PROTO])
 		parms->proto = nla_get_u8(data[IFLA_IPTUN_PROTO]);
+
+	if (data[IFLA_IPTUN_COLLECT_METADATA])
+		parms->collect_md = true;
 }
 
 static bool ip6_tnl_netlink_encap_parms(struct nlattr *data[],
@@ -1850,6 +1928,7 @@ static int ip6_tnl_newlink(struct net *src_net, struct net_device *dev,
 			   struct nlattr *tb[], struct nlattr *data[])
 {
 	struct net *net = dev_net(dev);
+	struct ip6_tnl_net *ip6n = net_generic(net, ip6_tnl_net_id);
 	struct ip6_tnl *nt, *t;
 	struct ip_tunnel_encap ipencap;
 
@@ -1864,9 +1943,14 @@ static int ip6_tnl_newlink(struct net *src_net, struct net_device *dev,
 
 	ip6_tnl_netlink_parms(data, &nt->parms);
 
+	if (nt->parms.collect_md) {
+		if (rtnl_dereference(ip6n->collect_md_tun))
+			return -EEXIST;
+	} else {
 		t = ip6_tnl_locate(net, &nt->parms, 0);
 		if (!IS_ERR(t))
 			return -EEXIST;
+	}
 
 	return ip6_tnl_create2(dev);
 }
@@ -1890,6 +1974,8 @@ static int ip6_tnl_changelink(struct net_device *dev, struct nlattr *tb[],
 			return err;
 	}
 	ip6_tnl_netlink_parms(data, &p);
+	if (p.collect_md)
+		return -EINVAL;
 
 	t = ip6_tnl_locate(net, &p, 0);
 	if (!IS_ERR(t)) {
@@ -1937,6 +2023,8 @@ static size_t ip6_tnl_get_size(const struct net_device *dev)
 		nla_total_size(2) +
 		/* IFLA_IPTUN_ENCAP_DPORT */
 		nla_total_size(2) +
+		/* IFLA_IPTUN_COLLECT_METADATA */
+		nla_total_size(0) +
 		0;
 }
 
@@ -1955,16 +2043,15 @@ static int ip6_tnl_fill_info(struct sk_buff *skb, const struct net_device *dev)
 	    nla_put_u8(skb, IFLA_IPTUN_PROTO, parm->proto))
 		goto nla_put_failure;
 
-	if (nla_put_u16(skb, IFLA_IPTUN_ENCAP_TYPE,
-			tunnel->encap.type) ||
-	nla_put_be16(skb, IFLA_IPTUN_ENCAP_SPORT,
-		     tunnel->encap.sport) ||
-	nla_put_be16(skb, IFLA_IPTUN_ENCAP_DPORT,
-		     tunnel->encap.dport) ||
-	nla_put_u16(skb, IFLA_IPTUN_ENCAP_FLAGS,
-		    tunnel->encap.flags))
+	if (nla_put_u16(skb, IFLA_IPTUN_ENCAP_TYPE, tunnel->encap.type) ||
+	    nla_put_be16(skb, IFLA_IPTUN_ENCAP_SPORT, tunnel->encap.sport) ||
+	    nla_put_be16(skb, IFLA_IPTUN_ENCAP_DPORT, tunnel->encap.dport) ||
+	    nla_put_u16(skb, IFLA_IPTUN_ENCAP_FLAGS, tunnel->encap.flags))
 		goto nla_put_failure;
 
+	if (parm->collect_md)
+		if (nla_put_flag(skb, IFLA_IPTUN_COLLECT_METADATA))
+			goto nla_put_failure;
 	return 0;
 
 nla_put_failure:
@@ -1992,6 +2079,7 @@ static const struct nla_policy ip6_tnl_policy[IFLA_IPTUN_MAX + 1] = {
 	[IFLA_IPTUN_ENCAP_FLAGS]	= { .type = NLA_U16 },
 	[IFLA_IPTUN_ENCAP_SPORT]	= { .type = NLA_U16 },
 	[IFLA_IPTUN_ENCAP_DPORT]	= { .type = NLA_U16 },
+	[IFLA_IPTUN_COLLECT_METADATA]	= { .type = NLA_FLAG },
 };
 
 static struct rtnl_link_ops ip6_link_ops __read_mostly = {

samples/bpf/tcbpf2_kern.c

 /* Copyright (c) 2016 VMware
+ * Copyright (c) 2016 Facebook
  *
  * This program is free software; you can redistribute it and/or
  * modify it under the terms of version 2 of the GNU General Public
@@ -8,12 +9,15 @@
 #include <uapi/linux/if_ether.h>
 #include <uapi/linux/if_packet.h>
 #include <uapi/linux/ip.h>
+#include <uapi/linux/ipv6.h>
 #include <uapi/linux/in.h>
 #include <uapi/linux/tcp.h>
 #include <uapi/linux/filter.h>
 #include <uapi/linux/pkt_cls.h>
+#include <net/ipv6.h>
 #include "bpf_helpers.h"
 
+#define _htonl __builtin_bswap32
 #define ERROR(ret) do {\
 		char fmt[] = "ERROR line:%d ret:%d\n";\
 		bpf_trace_printk(fmt, sizeof(fmt), __LINE__, ret); \
@@ -188,4 +192,190 @@ int _geneve_get_tunnel(struct __sk_buff *skb)
 	return TC_ACT_OK;
 }
 
+SEC("ipip_set_tunnel")
+int _ipip_set_tunnel(struct __sk_buff *skb)
+{
+	struct bpf_tunnel_key key = {};
+	void *data = (void *)(long)skb->data;
+	struct iphdr *iph = data;
+	struct tcphdr *tcp = data + sizeof(*iph);
+	void *data_end = (void *)(long)skb->data_end;
+	int ret;
+
+	/* single length check */
+	if (data + sizeof(*iph) + sizeof(*tcp) > data_end) {
+		ERROR(1);
+		return TC_ACT_SHOT;
+	}
+
+	key.tunnel_ttl = 64;
+	if (iph->protocol == IPPROTO_ICMP) {
+		key.remote_ipv4 = 0xac100164; /* 172.16.1.100 */
+	} else {
+		if (iph->protocol != IPPROTO_TCP || iph->ihl != 5)
+			return TC_ACT_SHOT;
+
+		if (tcp->dest == htons(5200))
+			key.remote_ipv4 = 0xac100164; /* 172.16.1.100 */
+		else if (tcp->dest == htons(5201))
+			key.remote_ipv4 = 0xac100165; /* 172.16.1.101 */
+		else
+			return TC_ACT_SHOT;
+	}
+
+	ret = bpf_skb_set_tunnel_key(skb, &key, sizeof(key), 0);
+	if (ret < 0) {
+		ERROR(ret);
+		return TC_ACT_SHOT;
+	}
+
+	return TC_ACT_OK;
+}
+
+SEC("ipip_get_tunnel")
+int _ipip_get_tunnel(struct __sk_buff *skb)
+{
+	int ret;
+	struct bpf_tunnel_key key;
+	char fmt[] = "remote ip 0x%x\n";
+
+	ret = bpf_skb_get_tunnel_key(skb, &key, sizeof(key), 0);
+	if (ret < 0) {
+		ERROR(ret);
+		return TC_ACT_SHOT;
+	}
+
+	bpf_trace_printk(fmt, sizeof(fmt), key.remote_ipv4);
+	return TC_ACT_OK;
+}
+
+SEC("ipip6_set_tunnel")
+int _ipip6_set_tunnel(struct __sk_buff *skb)
+{
+	struct bpf_tunnel_key key = {};
+	void *data = (void *)(long)skb->data;
+	struct iphdr *iph = data;
+	struct tcphdr *tcp = data + sizeof(*iph);
+	void *data_end = (void *)(long)skb->data_end;
+	int ret;
+
+	/* single length check */
+	if (data + sizeof(*iph) + sizeof(*tcp) > data_end) {
+		ERROR(1);
+		return TC_ACT_SHOT;
+	}
+
+	key.remote_ipv6[0] = _htonl(0x2401db00);
+	key.tunnel_ttl = 64;
+
+	if (iph->protocol == IPPROTO_ICMP) {
+		key.remote_ipv6[3] = _htonl(1);
+	} else {
+		if (iph->protocol != IPPROTO_TCP || iph->ihl != 5) {
+			ERROR(iph->protocol);
+			return TC_ACT_SHOT;
+		}
+
+		if (tcp->dest == htons(5200)) {
+			key.remote_ipv6[3] = _htonl(1);
+		} else if (tcp->dest == htons(5201)) {
+			key.remote_ipv6[3] = _htonl(2);
+		} else {
+			ERROR(tcp->dest);
+			return TC_ACT_SHOT;
+		}
+	}
+
+	ret = bpf_skb_set_tunnel_key(skb, &key, sizeof(key), BPF_F_TUNINFO_IPV6);
+	if (ret < 0) {
+		ERROR(ret);
+		return TC_ACT_SHOT;
+	}
+
+	return TC_ACT_OK;
+}
+
+SEC("ipip6_get_tunnel")
+int _ipip6_get_tunnel(struct __sk_buff *skb)
+{
+	int ret;
+	struct bpf_tunnel_key key;
+	char fmt[] = "remote ip6 %x::%x\n";
+
+	ret = bpf_skb_get_tunnel_key(skb, &key, sizeof(key), BPF_F_TUNINFO_IPV6);
+	if (ret < 0) {
+		ERROR(ret);
+		return TC_ACT_SHOT;
+	}
+
+	bpf_trace_printk(fmt, sizeof(fmt), _htonl(key.remote_ipv6[0]),
+			 _htonl(key.remote_ipv6[3]));
+	return TC_ACT_OK;
+}
+
+SEC("ip6ip6_set_tunnel")
+int _ip6ip6_set_tunnel(struct __sk_buff *skb)
+{
+	struct bpf_tunnel_key key = {};
+	void *data = (void *)(long)skb->data;
+	struct ipv6hdr *iph = data;
+	struct tcphdr *tcp = data + sizeof(*iph);
+	void *data_end = (void *)(long)skb->data_end;
+	int ret;
+
+	/* single length check */
+	if (data + sizeof(*iph) + sizeof(*tcp) > data_end) {
+		ERROR(1);
+		return TC_ACT_SHOT;
+	}
+
+	key.remote_ipv6[0] = _htonl(0x2401db00);
+	key.tunnel_ttl = 64;
+
+	if (iph->nexthdr == NEXTHDR_ICMP) {
+		key.remote_ipv6[3] = _htonl(1);
+	} else {
+		if (iph->nexthdr != NEXTHDR_TCP) {
+			ERROR(iph->nexthdr);
+			return TC_ACT_SHOT;
+		}
+
+		if (tcp->dest == htons(5200)) {
+			key.remote_ipv6[3] = _htonl(1);
+		} else if (tcp->dest == htons(5201)) {
+			key.remote_ipv6[3] = _htonl(2);
+		} else {
+			ERROR(tcp->dest);
+			return TC_ACT_SHOT;
+		}
+	}
+
+	ret = bpf_skb_set_tunnel_key(skb, &key, sizeof(key), BPF_F_TUNINFO_IPV6);
+	if (ret < 0) {
+		ERROR(ret);
+		return TC_ACT_SHOT;
+	}
+
+	return TC_ACT_OK;
+}
+
+SEC("ip6ip6_get_tunnel")
+int _ip6ip6_get_tunnel(struct __sk_buff *skb)
+{
+	int ret;
+	struct bpf_tunnel_key key;
+	char fmt[] = "remote ip6 %x::%x\n";
+
+	ret = bpf_skb_get_tunnel_key(skb, &key, sizeof(key), BPF_F_TUNINFO_IPV6);
+	if (ret < 0) {
+		ERROR(ret);
+		return TC_ACT_SHOT;
+	}
+
+	bpf_trace_printk(fmt, sizeof(fmt), _htonl(key.remote_ipv6[0]),
+			 _htonl(key.remote_ipv6[3]));
+	return TC_ACT_OK;
+}
+
+
 char _license[] SEC("license") = "GPL";

samples/bpf/test_ipip.sh

+#!/bin/bash
+
+function config_device {
+	ip netns add at_ns0
+	ip netns add at_ns1
+	ip netns add at_ns2
+	ip link add veth0 type veth peer name veth0b
+	ip link add veth1 type veth peer name veth1b
+	ip link add veth2 type veth peer name veth2b
+	ip link set veth0b up
+	ip link set veth1b up
+	ip link set veth2b up
+	ip link set dev veth0b mtu 1500
+	ip link set dev veth1b mtu 1500
+	ip link set dev veth2b mtu 1500
+	ip link set veth0 netns at_ns0
+	ip link set veth1 netns at_ns1
+	ip link set veth2 netns at_ns2
+	ip netns exec at_ns0 ip addr add 172.16.1.100/24 dev veth0
+	ip netns exec at_ns0 ip addr add 2401:db00::1/64 dev veth0 nodad
+	ip netns exec at_ns0 ip link set dev veth0 up
+	ip netns exec at_ns1 ip addr add 172.16.1.101/24 dev veth1
+	ip netns exec at_ns1 ip addr add 2401:db00::2/64 dev veth1 nodad
+	ip netns exec at_ns1 ip link set dev veth1 up
+	ip netns exec at_ns2 ip addr add 172.16.1.200/24 dev veth2
+	ip netns exec at_ns2 ip addr add 2401:db00::3/64 dev veth2 nodad
+	ip netns exec at_ns2 ip link set dev veth2 up
+	ip link add br0 type bridge
+	ip link set br0 up
+	ip link set dev br0 mtu 1500
+	ip link set veth0b master br0
+	ip link set veth1b master br0
+	ip link set veth2b master br0
+}
+
+function add_ipip_tunnel {
+	ip netns exec at_ns0 \
+		ip link add dev $DEV_NS type ipip local 172.16.1.100 remote 172.16.1.200
+	ip netns exec at_ns0 ip link set dev $DEV_NS up
+	ip netns exec at_ns0 ip addr add dev $DEV_NS 10.1.1.100/24
+	ip netns exec at_ns1 \
+		ip link add dev $DEV_NS type ipip local 172.16.1.101 remote 172.16.1.200
+	ip netns exec at_ns1 ip link set dev $DEV_NS up
+	# same inner IP address in at_ns0 and at_ns1
+	ip netns exec at_ns1 ip addr add dev $DEV_NS 10.1.1.100/24
+
+	ip netns exec at_ns2 ip link add dev $DEV type ipip external
+	ip netns exec at_ns2 ip link set dev $DEV up
+	ip netns exec at_ns2 ip addr add dev $DEV 10.1.1.200/24
+}
+
+function add_ipip6_tunnel {
+	ip netns exec at_ns0 \
+		ip link add dev $DEV_NS type ip6tnl mode ipip6 local 2401:db00::1/64 remote 2401:db00::3/64
+	ip netns exec at_ns0 ip link set dev $DEV_NS up
+	ip netns exec at_ns0 ip addr add dev $DEV_NS 10.1.1.100/24
+	ip netns exec at_ns1 \
+		ip link add dev $DEV_NS type ip6tnl mode ipip6 local 2401:db00::2/64 remote 2401:db00::3/64
+	ip netns exec at_ns1 ip link set dev $DEV_NS up
+	# same inner IP address in at_ns0 and at_ns1
+	ip netns exec at_ns1 ip addr add dev $DEV_NS 10.1.1.100/24
+
+	ip netns exec at_ns2 ip link add dev $DEV type ip6tnl mode ipip6 external
+	ip netns exec at_ns2 ip link set dev $DEV up
+	ip netns exec at_ns2 ip addr add dev $DEV 10.1.1.200/24
+}
+
+function add_ip6ip6_tunnel {
+	ip netns exec at_ns0 \
+		ip link add dev $DEV_NS type ip6tnl mode ip6ip6 local 2401:db00::1/64 remote 2401:db00::3/64
+	ip netns exec at_ns0 ip link set dev $DEV_NS up
+	ip netns exec at_ns0 ip addr add dev $DEV_NS 2601:646::1/64
+	ip netns exec at_ns1 \
+		ip link add dev $DEV_NS type ip6tnl mode ip6ip6 local 2401:db00::2/64 remote 2401:db00::3/64
+	ip netns exec at_ns1 ip link set dev $DEV_NS up
+	# same inner IP address in at_ns0 and at_ns1
+	ip netns exec at_ns1 ip addr add dev $DEV_NS 2601:646::1/64
+
+	ip netns exec at_ns2 ip link add dev $DEV type ip6tnl mode ip6ip6 external
+	ip netns exec at_ns2 ip link set dev $DEV up
+	ip netns exec at_ns2 ip addr add dev $DEV 2601:646::2/64
+}
+
+function attach_bpf {
+	DEV=$1
+	SET_TUNNEL=$2
+	GET_TUNNEL=$3
+	ip netns exec at_ns2 tc qdisc add dev $DEV clsact
+	ip netns exec at_ns2 tc filter add dev $DEV egress bpf da obj tcbpf2_kern.o sec $SET_TUNNEL
+	ip netns exec at_ns2 tc filter add dev $DEV ingress bpf da obj tcbpf2_kern.o sec $GET_TUNNEL
+}
+
+function test_ipip {
+	DEV_NS=ipip_std
+	DEV=ipip_bpf
+	config_device
+#	tcpdump -nei br0 &
+	cat /sys/kernel/debug/tracing/trace_pipe &
+
+	add_ipip_tunnel
+	attach_bpf $DEV ipip_set_tunnel ipip_get_tunnel
+
+	ip netns exec at_ns0 ping -c 1 10.1.1.200
+	ip netns exec at_ns2 ping -c 1 10.1.1.100
+	ip netns exec at_ns0 iperf -sD -p 5200 > /dev/null
+	ip netns exec at_ns1 iperf -sD -p 5201 > /dev/null
+	sleep 0.2
+	# tcp check _same_ IP over different tunnels
+	ip netns exec at_ns2 iperf -c 10.1.1.100 -n 5k -p 5200
+	ip netns exec at_ns2 iperf -c 10.1.1.100 -n 5k -p 5201
+	cleanup
+}
+
+# IPv4 over IPv6 tunnel
+function test_ipip6 {
+	DEV_NS=ipip_std
+	DEV=ipip_bpf
+	config_device
+#	tcpdump -nei br0 &
+	cat /sys/kernel/debug/tracing/trace_pipe &
+
+	add_ipip6_tunnel
+	attach_bpf $DEV ipip6_set_tunnel ipip6_get_tunnel
+
+	ip netns exec at_ns0 ping -c 1 10.1.1.200
+	ip netns exec at_ns2 ping -c 1 10.1.1.100
+	ip netns exec at_ns0 iperf -sD -p 5200 > /dev/null
+	ip netns exec at_ns1 iperf -sD -p 5201 > /dev/null
+	sleep 0.2
+	# tcp check _same_ IP over different tunnels
+	ip netns exec at_ns2 iperf -c 10.1.1.100 -n 5k -p 5200
+	ip netns exec at_ns2 iperf -c 10.1.1.100 -n 5k -p 5201
+	cleanup
+}
+
+# IPv6 over IPv6 tunnel
+function test_ip6ip6 {
+	DEV_NS=ipip_std
+	DEV=ipip_bpf
+	config_device
+#	tcpdump -nei br0 &
+	cat /sys/kernel/debug/tracing/trace_pipe &
+
+	add_ip6ip6_tunnel
+	attach_bpf $DEV ip6ip6_set_tunnel ip6ip6_get_tunnel
+
+	ip netns exec at_ns0 ping -6 -c 1 2601:646::2
+	ip netns exec at_ns2 ping -6 -c 1 2601:646::1
+	ip netns exec at_ns0 iperf -6sD -p 5200 > /dev/null
+	ip netns exec at_ns1 iperf -6sD -p 5201 > /dev/null
+	sleep 0.2
+	# tcp check _same_ IP over different tunnels
+	ip netns exec at_ns2 iperf -6c 2601:646::1 -n 5k -p 5200
+	ip netns exec at_ns2 iperf -6c 2601:646::1 -n 5k -p 5201
+	cleanup
+}
+
+function cleanup {
+	set +ex
+	pkill iperf
+	ip netns delete at_ns0
+	ip netns delete at_ns1
+	ip netns delete at_ns2
+	ip link del veth0
+	ip link del veth1
+	ip link del veth2
+	ip link del br0
+	pkill tcpdump
+	pkill cat
+	set -ex
+}
+
+cleanup
+echo "Testing IP tunnels..."
+test_ipip
+test_ipip6
+test_ip6ip6
+echo "*** PASS ***"

samples/bpf/test_tunnel_bpf.sh

@@ -9,15 +9,13 @@
 # local 172.16.1.200 remote 172.16.1.100
 # veth1 IP: 172.16.1.200, tunnel dev <type>11
 
-set -e
-
 function config_device {
 	ip netns add at_ns0
 	ip link add veth0 type veth peer name veth1
 	ip link set veth0 netns at_ns0
 	ip netns exec at_ns0 ip addr add 172.16.1.100/24 dev veth0
 	ip netns exec at_ns0 ip link set dev veth0 up
-	ip link set dev veth1 up
+	ip link set dev veth1 up mtu 1500
 	ip addr add dev veth1 172.16.1.200/24
 }
 
@@ -67,6 +65,19 @@ function add_geneve_tunnel {
 	ip addr add dev $DEV 10.1.1.200/24
 }
 
+function add_ipip_tunnel {
+	# in namespace
+	ip netns exec at_ns0 \
+		ip link add dev $DEV_NS type $TYPE local 172.16.1.100 remote 172.16.1.200
+	ip netns exec at_ns0 ip link set dev $DEV_NS up
+	ip netns exec at_ns0 ip addr add dev $DEV_NS 10.1.1.100/24
+
+	# out of namespace
+	ip link add dev $DEV type $TYPE external
+	ip link set dev $DEV up
+	ip addr add dev $DEV 10.1.1.200/24
+}
+
 function attach_bpf {
 	DEV=$1
 	SET_TUNNEL=$2
@@ -85,6 +96,7 @@ function test_gre {
 	attach_bpf $DEV gre_set_tunnel gre_get_tunnel
 	ping -c 1 10.1.1.100
 	ip netns exec at_ns0 ping -c 1 10.1.1.200
+	cleanup
 }
 
 function test_vxlan {
@@ -96,6 +108,7 @@ function test_vxlan {
 	attach_bpf $DEV vxlan_set_tunnel vxlan_get_tunnel
 	ping -c 1 10.1.1.100
 	ip netns exec at_ns0 ping -c 1 10.1.1.200
+	cleanup
 }
 
 function test_geneve {
@@ -107,21 +120,48 @@ function test_geneve {
 	attach_bpf $DEV geneve_set_tunnel geneve_get_tunnel
 	ping -c 1 10.1.1.100
 	ip netns exec at_ns0 ping -c 1 10.1.1.200
+	cleanup
+}
+
+function test_ipip {
+	TYPE=ipip
+	DEV_NS=ipip00
+	DEV=ipip11
+	config_device
+	tcpdump -nei veth1 &
+	cat /sys/kernel/debug/tracing/trace_pipe &
+	add_ipip_tunnel
+	ethtool -K veth1 gso off gro off rx off tx off
+	ip link set dev veth1 mtu 1500
+	attach_bpf $DEV ipip_set_tunnel ipip_get_tunnel
+	ping -c 1 10.1.1.100
+	ip netns exec at_ns0 ping -c 1 10.1.1.200
+	ip netns exec at_ns0 iperf -sD -p 5200 > /dev/null
+	sleep 0.2
+	iperf -c 10.1.1.100 -n 5k -p 5200
+	cleanup
 }
 
 function cleanup {
+	set +ex
+	pkill iperf
 	ip netns delete at_ns0
 	ip link del veth1
-	ip link del $DEV
+	ip link del ipip11
+	ip link del gretap11
+	ip link del geneve11
+	pkill tcpdump
+	pkill cat
+	set -ex
 }
 
+cleanup
 echo "Testing GRE tunnel..."
 test_gre
-cleanup
 echo "Testing VXLAN tunnel..."
 test_vxlan
-cleanup
 echo "Testing GENEVE tunnel..."
 test_geneve
-cleanup
-echo "Success"
+echo "Testing IPIP tunnel..."
+test_ipip
+echo "*** PASS ***"