2 * IPv6 over IPv4 tunnel device - Simple Internet Transition (SIT)
3 * Linux INET6 implementation
6 * Pedro Roque <roque@di.fc.ul.pt>
7 * Alexey Kuznetsov <kuznet@ms2.inr.ac.ru>
9 * This program is free software; you can redistribute it and/or
10 * modify it under the terms of the GNU General Public License
11 * as published by the Free Software Foundation; either version
12 * 2 of the License, or (at your option) any later version.
15 * Roger Venning <r.venning@telstra.com>: 6to4 support
16 * Nate Thompson <nate@thebog.net>: 6to4 support
17 * Fred Templin <fred.l.templin@boeing.com>: isatap support
20 #define pr_fmt(fmt) KBUILD_MODNAME ": " fmt
22 #include <linux/module.h>
23 #include <linux/capability.h>
24 #include <linux/errno.h>
25 #include <linux/types.h>
26 #include <linux/socket.h>
27 #include <linux/sockios.h>
28 #include <linux/net.h>
29 #include <linux/in6.h>
30 #include <linux/netdevice.h>
31 #include <linux/if_arp.h>
32 #include <linux/icmp.h>
33 #include <linux/slab.h>
34 #include <asm/uaccess.h>
35 #include <linux/init.h>
36 #include <linux/netfilter_ipv4.h>
37 #include <linux/if_ether.h>
43 #include <net/protocol.h>
44 #include <net/transp_v6.h>
45 #include <net/ip6_fib.h>
46 #include <net/ip6_route.h>
47 #include <net/ndisc.h>
48 #include <net/addrconf.h>
52 #include <net/ip_tunnels.h>
53 #include <net/inet_ecn.h>
55 #include <net/dsfield.h>
56 #include <net/net_namespace.h>
57 #include <net/netns/generic.h>
60 This version of net/ipv6/sit.c is cloned of net/ipv4/ip_gre.c
62 For comments look at net/ipv4/ip_gre.c --ANK
66 #define HASH(addr) (((__force u32)addr^((__force u32)addr>>4))&0xF)
68 static bool log_ecn_error = true;
69 module_param(log_ecn_error, bool, 0644);
70 MODULE_PARM_DESC(log_ecn_error, "Log packets received with corrupted ECN");
72 static int ipip6_tunnel_init(struct net_device *dev);
73 static void ipip6_tunnel_setup(struct net_device *dev);
74 static void ipip6_dev_free(struct net_device *dev);
75 static bool check_6rd(struct ip_tunnel *tunnel, const struct in6_addr *v6dst,
77 static struct rtnl_link_ops sit_link_ops __read_mostly;
79 static int sit_net_id __read_mostly;
81 struct ip_tunnel __rcu *tunnels_r_l[HASH_SIZE];
82 struct ip_tunnel __rcu *tunnels_r[HASH_SIZE];
83 struct ip_tunnel __rcu *tunnels_l[HASH_SIZE];
84 struct ip_tunnel __rcu *tunnels_wc[1];
85 struct ip_tunnel __rcu **tunnels[4];
87 struct net_device *fb_tunnel_dev;
91 * Must be invoked with rcu_read_lock
93 static struct ip_tunnel *ipip6_tunnel_lookup(struct net *net,
94 struct net_device *dev, __be32 remote, __be32 local)
96 unsigned int h0 = HASH(remote);
97 unsigned int h1 = HASH(local);
99 struct sit_net *sitn = net_generic(net, sit_net_id);
101 for_each_ip_tunnel_rcu(t, sitn->tunnels_r_l[h0 ^ h1]) {
102 if (local == t->parms.iph.saddr &&
103 remote == t->parms.iph.daddr &&
104 (!dev || !t->parms.link || dev->ifindex == t->parms.link) &&
105 (t->dev->flags & IFF_UP))
108 for_each_ip_tunnel_rcu(t, sitn->tunnels_r[h0]) {
109 if (remote == t->parms.iph.daddr &&
110 (!dev || !t->parms.link || dev->ifindex == t->parms.link) &&
111 (t->dev->flags & IFF_UP))
114 for_each_ip_tunnel_rcu(t, sitn->tunnels_l[h1]) {
115 if (local == t->parms.iph.saddr &&
116 (!dev || !t->parms.link || dev->ifindex == t->parms.link) &&
117 (t->dev->flags & IFF_UP))
120 t = rcu_dereference(sitn->tunnels_wc[0]);
121 if (t && (t->dev->flags & IFF_UP))
126 static struct ip_tunnel __rcu **__ipip6_bucket(struct sit_net *sitn,
127 struct ip_tunnel_parm *parms)
129 __be32 remote = parms->iph.daddr;
130 __be32 local = parms->iph.saddr;
142 return &sitn->tunnels[prio][h];
145 static inline struct ip_tunnel __rcu **ipip6_bucket(struct sit_net *sitn,
148 return __ipip6_bucket(sitn, &t->parms);
151 static void ipip6_tunnel_unlink(struct sit_net *sitn, struct ip_tunnel *t)
153 struct ip_tunnel __rcu **tp;
154 struct ip_tunnel *iter;
156 for (tp = ipip6_bucket(sitn, t);
157 (iter = rtnl_dereference(*tp)) != NULL;
160 rcu_assign_pointer(*tp, t->next);
166 static void ipip6_tunnel_link(struct sit_net *sitn, struct ip_tunnel *t)
168 struct ip_tunnel __rcu **tp = ipip6_bucket(sitn, t);
170 rcu_assign_pointer(t->next, rtnl_dereference(*tp));
171 rcu_assign_pointer(*tp, t);
174 static void ipip6_tunnel_clone_6rd(struct net_device *dev, struct sit_net *sitn)
176 #ifdef CONFIG_IPV6_SIT_6RD
177 struct ip_tunnel *t = netdev_priv(dev);
179 if (t->dev == sitn->fb_tunnel_dev) {
180 ipv6_addr_set(&t->ip6rd.prefix, htonl(0x20020000), 0, 0, 0);
181 t->ip6rd.relay_prefix = 0;
182 t->ip6rd.prefixlen = 16;
183 t->ip6rd.relay_prefixlen = 0;
185 struct ip_tunnel *t0 = netdev_priv(sitn->fb_tunnel_dev);
186 memcpy(&t->ip6rd, &t0->ip6rd, sizeof(t->ip6rd));
191 static int ipip6_tunnel_create(struct net_device *dev)
193 struct ip_tunnel *t = netdev_priv(dev);
194 struct net *net = dev_net(dev);
195 struct sit_net *sitn = net_generic(net, sit_net_id);
198 memcpy(dev->dev_addr, &t->parms.iph.saddr, 4);
199 memcpy(dev->broadcast, &t->parms.iph.daddr, 4);
201 if ((__force u16)t->parms.i_flags & SIT_ISATAP)
202 dev->priv_flags |= IFF_ISATAP;
204 dev->rtnl_link_ops = &sit_link_ops;
206 err = register_netdevice(dev);
210 ipip6_tunnel_clone_6rd(dev, sitn);
214 ipip6_tunnel_link(sitn, t);
221 static struct ip_tunnel *ipip6_tunnel_locate(struct net *net,
222 struct ip_tunnel_parm *parms, int create)
224 __be32 remote = parms->iph.daddr;
225 __be32 local = parms->iph.saddr;
226 struct ip_tunnel *t, *nt;
227 struct ip_tunnel __rcu **tp;
228 struct net_device *dev;
230 struct sit_net *sitn = net_generic(net, sit_net_id);
232 for (tp = __ipip6_bucket(sitn, parms);
233 (t = rtnl_dereference(*tp)) != NULL;
235 if (local == t->parms.iph.saddr &&
236 remote == t->parms.iph.daddr &&
237 parms->link == t->parms.link) {
248 strlcpy(name, parms->name, IFNAMSIZ);
250 strcpy(name, "sit%d");
252 dev = alloc_netdev(sizeof(*t), name, NET_NAME_UNKNOWN,
257 dev_net_set(dev, net);
259 nt = netdev_priv(dev);
262 if (ipip6_tunnel_create(dev) < 0)
273 #define for_each_prl_rcu(start) \
274 for (prl = rcu_dereference(start); \
276 prl = rcu_dereference(prl->next))
278 static struct ip_tunnel_prl_entry *
279 __ipip6_tunnel_locate_prl(struct ip_tunnel *t, __be32 addr)
281 struct ip_tunnel_prl_entry *prl;
283 for_each_prl_rcu(t->prl)
284 if (prl->addr == addr)
290 static int ipip6_tunnel_get_prl(struct ip_tunnel *t,
291 struct ip_tunnel_prl __user *a)
293 struct ip_tunnel_prl kprl, *kp;
294 struct ip_tunnel_prl_entry *prl;
295 unsigned int cmax, c = 0, ca, len;
298 if (copy_from_user(&kprl, a, sizeof(kprl)))
300 cmax = kprl.datalen / sizeof(kprl);
301 if (cmax > 1 && kprl.addr != htonl(INADDR_ANY))
304 /* For simple GET or for root users,
305 * we try harder to allocate.
307 kp = (cmax <= 1 || capable(CAP_NET_ADMIN)) ?
308 kcalloc(cmax, sizeof(*kp), GFP_KERNEL) :
313 ca = t->prl_count < cmax ? t->prl_count : cmax;
316 /* We don't try hard to allocate much memory for
318 * For root users, retry allocating enough memory for
321 kp = kcalloc(ca, sizeof(*kp), GFP_ATOMIC);
329 for_each_prl_rcu(t->prl) {
332 if (kprl.addr != htonl(INADDR_ANY) && prl->addr != kprl.addr)
334 kp[c].addr = prl->addr;
335 kp[c].flags = prl->flags;
337 if (kprl.addr != htonl(INADDR_ANY))
343 len = sizeof(*kp) * c;
345 if ((len && copy_to_user(a + 1, kp, len)) || put_user(len, &a->datalen))
354 ipip6_tunnel_add_prl(struct ip_tunnel *t, struct ip_tunnel_prl *a, int chg)
356 struct ip_tunnel_prl_entry *p;
359 if (a->addr == htonl(INADDR_ANY))
364 for (p = rtnl_dereference(t->prl); p; p = rtnl_dereference(p->next)) {
365 if (p->addr == a->addr) {
380 p = kzalloc(sizeof(struct ip_tunnel_prl_entry), GFP_KERNEL);
390 rcu_assign_pointer(t->prl, p);
395 static void prl_list_destroy_rcu(struct rcu_head *head)
397 struct ip_tunnel_prl_entry *p, *n;
399 p = container_of(head, struct ip_tunnel_prl_entry, rcu_head);
401 n = rcu_dereference_protected(p->next, 1);
408 ipip6_tunnel_del_prl(struct ip_tunnel *t, struct ip_tunnel_prl *a)
410 struct ip_tunnel_prl_entry *x;
411 struct ip_tunnel_prl_entry __rcu **p;
416 if (a && a->addr != htonl(INADDR_ANY)) {
418 (x = rtnl_dereference(*p)) != NULL;
420 if (x->addr == a->addr) {
422 kfree_rcu(x, rcu_head);
429 x = rtnl_dereference(t->prl);
432 call_rcu(&x->rcu_head, prl_list_destroy_rcu);
441 isatap_chksrc(struct sk_buff *skb, const struct iphdr *iph, struct ip_tunnel *t)
443 struct ip_tunnel_prl_entry *p;
447 p = __ipip6_tunnel_locate_prl(t, iph->saddr);
449 if (p->flags & PRL_DEFAULT)
450 skb->ndisc_nodetype = NDISC_NODETYPE_DEFAULT;
452 skb->ndisc_nodetype = NDISC_NODETYPE_NODEFAULT;
454 const struct in6_addr *addr6 = &ipv6_hdr(skb)->saddr;
456 if (ipv6_addr_is_isatap(addr6) &&
457 (addr6->s6_addr32[3] == iph->saddr) &&
458 ipv6_chk_prefix(addr6, t->dev))
459 skb->ndisc_nodetype = NDISC_NODETYPE_HOST;
467 static void ipip6_tunnel_uninit(struct net_device *dev)
469 struct ip_tunnel *tunnel = netdev_priv(dev);
470 struct sit_net *sitn = net_generic(tunnel->net, sit_net_id);
472 if (dev == sitn->fb_tunnel_dev) {
473 RCU_INIT_POINTER(sitn->tunnels_wc[0], NULL);
475 ipip6_tunnel_unlink(sitn, tunnel);
476 ipip6_tunnel_del_prl(tunnel, NULL);
478 dst_cache_reset(&tunnel->dst_cache);
482 static int ipip6_err(struct sk_buff *skb, u32 info)
484 const struct iphdr *iph = (const struct iphdr *)skb->data;
485 const int type = icmp_hdr(skb)->type;
486 const int code = icmp_hdr(skb)->code;
487 unsigned int data_len = 0;
493 case ICMP_PARAMETERPROB:
496 case ICMP_DEST_UNREACH:
499 /* Impossible event. */
502 /* All others are translated to HOST_UNREACH.
503 rfc2003 contains "deep thoughts" about NET_UNREACH,
504 I believe they are just ether pollution. --ANK
509 case ICMP_TIME_EXCEEDED:
510 if (code != ICMP_EXC_TTL)
512 data_len = icmp_hdr(skb)->un.reserved[1] * 4; /* RFC 4884 4.1 */
520 t = ipip6_tunnel_lookup(dev_net(skb->dev),
527 if (type == ICMP_DEST_UNREACH && code == ICMP_FRAG_NEEDED) {
528 ipv4_update_pmtu(skb, dev_net(skb->dev), info,
529 t->parms.link, 0, IPPROTO_IPV6, 0);
533 if (type == ICMP_REDIRECT) {
534 ipv4_redirect(skb, dev_net(skb->dev), t->parms.link, 0,
541 if (!ip6_err_gen_icmpv6_unreach(skb, iph->ihl * 4, type, data_len))
544 if (t->parms.iph.daddr == 0)
547 if (t->parms.iph.ttl == 0 && type == ICMP_TIME_EXCEEDED)
550 if (time_before(jiffies, t->err_time + IPTUNNEL_ERR_TIMEO))
554 t->err_time = jiffies;
559 static inline bool is_spoofed_6rd(struct ip_tunnel *tunnel, const __be32 v4addr,
560 const struct in6_addr *v6addr)
563 if (check_6rd(tunnel, v6addr, &v4embed) && v4addr != v4embed)
568 /* Checks if an address matches an address on the tunnel interface.
569 * Used to detect the NAT of proto 41 packets and let them pass spoofing test.
571 * This function is called after we considered the packet as spoofed
573 * We may have a router that is doing NAT for proto 41 packets
574 * for an internal station. Destination a.a.a.a/PREFIX:bbbb:bbbb
575 * will be translated to n.n.n.n/PREFIX:bbbb:bbbb. And is_spoofed_6rd
576 * function will return true, dropping the packet.
577 * But, we can still check if is spoofed against the IP
578 * addresses associated with the interface.
580 static bool only_dnatted(const struct ip_tunnel *tunnel,
581 const struct in6_addr *v6dst)
585 #ifdef CONFIG_IPV6_SIT_6RD
586 prefix_len = tunnel->ip6rd.prefixlen + 32
587 - tunnel->ip6rd.relay_prefixlen;
591 return ipv6_chk_custom_prefix(v6dst, prefix_len, tunnel->dev);
594 /* Returns true if a packet is spoofed */
595 static bool packet_is_spoofed(struct sk_buff *skb,
596 const struct iphdr *iph,
597 struct ip_tunnel *tunnel)
599 const struct ipv6hdr *ipv6h;
601 if (tunnel->dev->priv_flags & IFF_ISATAP) {
602 if (!isatap_chksrc(skb, iph, tunnel))
608 if (tunnel->dev->flags & IFF_POINTOPOINT)
611 ipv6h = ipv6_hdr(skb);
613 if (unlikely(is_spoofed_6rd(tunnel, iph->saddr, &ipv6h->saddr))) {
614 net_warn_ratelimited("Src spoofed %pI4/%pI6c -> %pI4/%pI6c\n",
615 &iph->saddr, &ipv6h->saddr,
616 &iph->daddr, &ipv6h->daddr);
620 if (likely(!is_spoofed_6rd(tunnel, iph->daddr, &ipv6h->daddr)))
623 if (only_dnatted(tunnel, &ipv6h->daddr))
626 net_warn_ratelimited("Dst spoofed %pI4/%pI6c -> %pI4/%pI6c\n",
627 &iph->saddr, &ipv6h->saddr,
628 &iph->daddr, &ipv6h->daddr);
632 static int ipip6_rcv(struct sk_buff *skb)
634 const struct iphdr *iph = ip_hdr(skb);
635 struct ip_tunnel *tunnel;
638 tunnel = ipip6_tunnel_lookup(dev_net(skb->dev), skb->dev,
639 iph->saddr, iph->daddr);
641 struct pcpu_sw_netstats *tstats;
643 if (tunnel->parms.iph.protocol != IPPROTO_IPV6 &&
644 tunnel->parms.iph.protocol != 0)
647 skb->mac_header = skb->network_header;
648 skb_reset_network_header(skb);
649 IPCB(skb)->flags = 0;
650 skb->dev = tunnel->dev;
652 if (packet_is_spoofed(skb, iph, tunnel)) {
653 tunnel->dev->stats.rx_errors++;
657 if (iptunnel_pull_header(skb, 0, htons(ETH_P_IPV6),
658 !net_eq(tunnel->net, dev_net(tunnel->dev))))
661 err = IP_ECN_decapsulate(iph, skb);
664 net_info_ratelimited("non-ECT from %pI4 with TOS=%#x\n",
665 &iph->saddr, iph->tos);
667 ++tunnel->dev->stats.rx_frame_errors;
668 ++tunnel->dev->stats.rx_errors;
673 tstats = this_cpu_ptr(tunnel->dev->tstats);
674 u64_stats_update_begin(&tstats->syncp);
675 tstats->rx_packets++;
676 tstats->rx_bytes += skb->len;
677 u64_stats_update_end(&tstats->syncp);
684 /* no tunnel matched, let upstream know, ipsec may handle it */
691 static const struct tnl_ptk_info tpi = {
692 /* no tunnel info required for ipip. */
693 .proto = htons(ETH_P_IP),
696 static int ipip_rcv(struct sk_buff *skb)
698 const struct iphdr *iph;
699 struct ip_tunnel *tunnel;
702 tunnel = ipip6_tunnel_lookup(dev_net(skb->dev), skb->dev,
703 iph->saddr, iph->daddr);
705 if (tunnel->parms.iph.protocol != IPPROTO_IPIP &&
706 tunnel->parms.iph.protocol != 0)
709 if (!xfrm4_policy_check(NULL, XFRM_POLICY_IN, skb))
711 if (iptunnel_pull_header(skb, 0, tpi.proto, false))
713 return ip_tunnel_rcv(tunnel, skb, &tpi, NULL, log_ecn_error);
724 * If the IPv6 address comes from 6rd / 6to4 (RFC 3056) addr space this function
725 * stores the embedded IPv4 address in v4dst and returns true.
727 static bool check_6rd(struct ip_tunnel *tunnel, const struct in6_addr *v6dst,
730 #ifdef CONFIG_IPV6_SIT_6RD
731 if (ipv6_prefix_equal(v6dst, &tunnel->ip6rd.prefix,
732 tunnel->ip6rd.prefixlen)) {
733 unsigned int pbw0, pbi0;
737 pbw0 = tunnel->ip6rd.prefixlen >> 5;
738 pbi0 = tunnel->ip6rd.prefixlen & 0x1f;
740 d = (ntohl(v6dst->s6_addr32[pbw0]) << pbi0) >>
741 tunnel->ip6rd.relay_prefixlen;
743 pbi1 = pbi0 - tunnel->ip6rd.relay_prefixlen;
745 d |= ntohl(v6dst->s6_addr32[pbw0 + 1]) >>
748 *v4dst = tunnel->ip6rd.relay_prefix | htonl(d);
752 if (v6dst->s6_addr16[0] == htons(0x2002)) {
753 /* 6to4 v6 addr has 16 bits prefix, 32 v4addr, 16 SLA, ... */
754 memcpy(v4dst, &v6dst->s6_addr16[1], 4);
761 static inline __be32 try_6rd(struct ip_tunnel *tunnel,
762 const struct in6_addr *v6dst)
765 check_6rd(tunnel, v6dst, &dst);
770 * This function assumes it is being called from dev_queue_xmit()
771 * and that skb is filled properly by that function.
774 static netdev_tx_t ipip6_tunnel_xmit(struct sk_buff *skb,
775 struct net_device *dev)
777 struct ip_tunnel *tunnel = netdev_priv(dev);
778 const struct iphdr *tiph = &tunnel->parms.iph;
779 const struct ipv6hdr *iph6 = ipv6_hdr(skb);
780 u8 tos = tunnel->parms.iph.tos;
781 __be16 df = tiph->frag_off;
782 struct rtable *rt; /* Route to the other host */
783 struct net_device *tdev; /* Device to other host */
784 unsigned int max_headroom; /* The extra header space needed */
785 __be32 dst = tiph->daddr;
788 const struct in6_addr *addr6;
791 u8 protocol = IPPROTO_IPV6;
792 int t_hlen = tunnel->hlen + sizeof(struct iphdr);
795 tos = ipv6_get_dsfield(iph6);
797 /* ISATAP (RFC4214) - must come before 6to4 */
798 if (dev->priv_flags & IFF_ISATAP) {
799 struct neighbour *neigh = NULL;
800 bool do_tx_error = false;
803 neigh = dst_neigh_lookup(skb_dst(skb), &iph6->daddr);
806 net_dbg_ratelimited("nexthop == NULL\n");
810 addr6 = (const struct in6_addr *)&neigh->primary_key;
811 addr_type = ipv6_addr_type(addr6);
813 if ((addr_type & IPV6_ADDR_UNICAST) &&
814 ipv6_addr_is_isatap(addr6))
815 dst = addr6->s6_addr32[3];
819 neigh_release(neigh);
825 dst = try_6rd(tunnel, &iph6->daddr);
828 struct neighbour *neigh = NULL;
829 bool do_tx_error = false;
832 neigh = dst_neigh_lookup(skb_dst(skb), &iph6->daddr);
835 net_dbg_ratelimited("nexthop == NULL\n");
839 addr6 = (const struct in6_addr *)&neigh->primary_key;
840 addr_type = ipv6_addr_type(addr6);
842 if (addr_type == IPV6_ADDR_ANY) {
843 addr6 = &ipv6_hdr(skb)->daddr;
844 addr_type = ipv6_addr_type(addr6);
847 if ((addr_type & IPV6_ADDR_COMPATv4) != 0)
848 dst = addr6->s6_addr32[3];
852 neigh_release(neigh);
857 rt = ip_route_output_ports(tunnel->net, &fl4, NULL,
860 IPPROTO_IPV6, RT_TOS(tos),
863 dev->stats.tx_carrier_errors++;
866 if (rt->rt_type != RTN_UNICAST) {
868 dev->stats.tx_carrier_errors++;
875 dev->stats.collisions++;
879 if (iptunnel_handle_offloads(skb, SKB_GSO_IPXIP4)) {
885 mtu = dst_mtu(&rt->dst) - t_hlen;
888 dev->stats.collisions++;
893 if (mtu < IPV6_MIN_MTU) {
898 if (tunnel->parms.iph.daddr && skb_dst(skb))
899 skb_dst(skb)->ops->update_pmtu(skb_dst(skb), NULL, skb, mtu);
901 if (skb->len > mtu && !skb_is_gso(skb)) {
902 icmpv6_send(skb, ICMPV6_PKT_TOOBIG, 0, mtu);
908 if (tunnel->err_count > 0) {
909 if (time_before(jiffies,
910 tunnel->err_time + IPTUNNEL_ERR_TIMEO)) {
912 dst_link_failure(skb);
914 tunnel->err_count = 0;
918 * Okay, now see if we can stuff it in the buffer as-is.
920 max_headroom = LL_RESERVED_SPACE(tdev) + t_hlen;
922 if (skb_headroom(skb) < max_headroom || skb_shared(skb) ||
923 (skb_cloned(skb) && !skb_clone_writable(skb, 0))) {
924 struct sk_buff *new_skb = skb_realloc_headroom(skb, max_headroom);
927 dev->stats.tx_dropped++;
932 skb_set_owner_w(new_skb, skb->sk);
935 iph6 = ipv6_hdr(skb);
939 ttl = iph6->hop_limit;
940 tos = INET_ECN_encapsulate(tos, ipv6_get_dsfield(iph6));
942 if (ip_tunnel_encap(skb, tunnel, &protocol, &fl4) < 0) {
947 skb_set_inner_ipproto(skb, IPPROTO_IPV6);
949 iptunnel_xmit(NULL, rt, skb, fl4.saddr, fl4.daddr, protocol, tos, ttl,
950 df, !net_eq(tunnel->net, dev_net(dev)));
954 dst_link_failure(skb);
957 dev->stats.tx_errors++;
961 static netdev_tx_t ipip_tunnel_xmit(struct sk_buff *skb, struct net_device *dev)
963 struct ip_tunnel *tunnel = netdev_priv(dev);
964 const struct iphdr *tiph = &tunnel->parms.iph;
966 if (iptunnel_handle_offloads(skb, SKB_GSO_IPXIP4))
969 skb_set_inner_ipproto(skb, IPPROTO_IPIP);
971 ip_tunnel_xmit(skb, dev, tiph, IPPROTO_IPIP);
975 dev->stats.tx_errors++;
979 static netdev_tx_t sit_tunnel_xmit(struct sk_buff *skb,
980 struct net_device *dev)
982 switch (skb->protocol) {
983 case htons(ETH_P_IP):
984 ipip_tunnel_xmit(skb, dev);
986 case htons(ETH_P_IPV6):
987 ipip6_tunnel_xmit(skb, dev);
996 dev->stats.tx_errors++;
1002 static void ipip6_tunnel_bind_dev(struct net_device *dev)
1004 struct net_device *tdev = NULL;
1005 struct ip_tunnel *tunnel;
1006 const struct iphdr *iph;
1009 tunnel = netdev_priv(dev);
1010 iph = &tunnel->parms.iph;
1013 struct rtable *rt = ip_route_output_ports(tunnel->net, &fl4,
1015 iph->daddr, iph->saddr,
1019 tunnel->parms.link);
1025 dev->flags |= IFF_POINTOPOINT;
1028 if (!tdev && tunnel->parms.link)
1029 tdev = __dev_get_by_index(tunnel->net, tunnel->parms.link);
1032 int t_hlen = tunnel->hlen + sizeof(struct iphdr);
1034 dev->hard_header_len = tdev->hard_header_len + sizeof(struct iphdr);
1035 dev->mtu = tdev->mtu - t_hlen;
1036 if (dev->mtu < IPV6_MIN_MTU)
1037 dev->mtu = IPV6_MIN_MTU;
1041 static void ipip6_tunnel_update(struct ip_tunnel *t, struct ip_tunnel_parm *p)
1043 struct net *net = t->net;
1044 struct sit_net *sitn = net_generic(net, sit_net_id);
1046 ipip6_tunnel_unlink(sitn, t);
1048 t->parms.iph.saddr = p->iph.saddr;
1049 t->parms.iph.daddr = p->iph.daddr;
1050 memcpy(t->dev->dev_addr, &p->iph.saddr, 4);
1051 memcpy(t->dev->broadcast, &p->iph.daddr, 4);
1052 ipip6_tunnel_link(sitn, t);
1053 t->parms.iph.ttl = p->iph.ttl;
1054 t->parms.iph.tos = p->iph.tos;
1055 if (t->parms.link != p->link) {
1056 t->parms.link = p->link;
1057 ipip6_tunnel_bind_dev(t->dev);
1059 dst_cache_reset(&t->dst_cache);
1060 netdev_state_change(t->dev);
1063 #ifdef CONFIG_IPV6_SIT_6RD
1064 static int ipip6_tunnel_update_6rd(struct ip_tunnel *t,
1065 struct ip_tunnel_6rd *ip6rd)
1067 struct in6_addr prefix;
1068 __be32 relay_prefix;
1070 if (ip6rd->relay_prefixlen > 32 ||
1071 ip6rd->prefixlen + (32 - ip6rd->relay_prefixlen) > 64)
1074 ipv6_addr_prefix(&prefix, &ip6rd->prefix, ip6rd->prefixlen);
1075 if (!ipv6_addr_equal(&prefix, &ip6rd->prefix))
1077 if (ip6rd->relay_prefixlen)
1078 relay_prefix = ip6rd->relay_prefix &
1079 htonl(0xffffffffUL <<
1080 (32 - ip6rd->relay_prefixlen));
1083 if (relay_prefix != ip6rd->relay_prefix)
1086 t->ip6rd.prefix = prefix;
1087 t->ip6rd.relay_prefix = relay_prefix;
1088 t->ip6rd.prefixlen = ip6rd->prefixlen;
1089 t->ip6rd.relay_prefixlen = ip6rd->relay_prefixlen;
1090 dst_cache_reset(&t->dst_cache);
1091 netdev_state_change(t->dev);
1097 ipip6_tunnel_ioctl(struct net_device *dev, struct ifreq *ifr, int cmd)
1100 struct ip_tunnel_parm p;
1101 struct ip_tunnel_prl prl;
1102 struct ip_tunnel *t = netdev_priv(dev);
1103 struct net *net = t->net;
1104 struct sit_net *sitn = net_generic(net, sit_net_id);
1105 #ifdef CONFIG_IPV6_SIT_6RD
1106 struct ip_tunnel_6rd ip6rd;
1111 #ifdef CONFIG_IPV6_SIT_6RD
1114 if (dev == sitn->fb_tunnel_dev) {
1115 if (copy_from_user(&p, ifr->ifr_ifru.ifru_data, sizeof(p))) {
1119 t = ipip6_tunnel_locate(net, &p, 0);
1121 t = netdev_priv(dev);
1125 if (cmd == SIOCGETTUNNEL) {
1126 memcpy(&p, &t->parms, sizeof(p));
1127 if (copy_to_user(ifr->ifr_ifru.ifru_data, &p,
1130 #ifdef CONFIG_IPV6_SIT_6RD
1132 ip6rd.prefix = t->ip6rd.prefix;
1133 ip6rd.relay_prefix = t->ip6rd.relay_prefix;
1134 ip6rd.prefixlen = t->ip6rd.prefixlen;
1135 ip6rd.relay_prefixlen = t->ip6rd.relay_prefixlen;
1136 if (copy_to_user(ifr->ifr_ifru.ifru_data, &ip6rd,
1147 if (!ns_capable(net->user_ns, CAP_NET_ADMIN))
1151 if (copy_from_user(&p, ifr->ifr_ifru.ifru_data, sizeof(p)))
1155 if (p.iph.protocol != IPPROTO_IPV6 &&
1156 p.iph.protocol != IPPROTO_IPIP &&
1157 p.iph.protocol != 0)
1159 if (p.iph.version != 4 ||
1160 p.iph.ihl != 5 || (p.iph.frag_off&htons(~IP_DF)))
1163 p.iph.frag_off |= htons(IP_DF);
1165 t = ipip6_tunnel_locate(net, &p, cmd == SIOCADDTUNNEL);
1167 if (dev != sitn->fb_tunnel_dev && cmd == SIOCCHGTUNNEL) {
1169 if (t->dev != dev) {
1174 if (((dev->flags&IFF_POINTOPOINT) && !p.iph.daddr) ||
1175 (!(dev->flags&IFF_POINTOPOINT) && p.iph.daddr)) {
1179 t = netdev_priv(dev);
1182 ipip6_tunnel_update(t, &p);
1187 if (copy_to_user(ifr->ifr_ifru.ifru_data, &t->parms, sizeof(p)))
1190 err = (cmd == SIOCADDTUNNEL ? -ENOBUFS : -ENOENT);
1195 if (!ns_capable(net->user_ns, CAP_NET_ADMIN))
1198 if (dev == sitn->fb_tunnel_dev) {
1200 if (copy_from_user(&p, ifr->ifr_ifru.ifru_data, sizeof(p)))
1203 t = ipip6_tunnel_locate(net, &p, 0);
1207 if (t == netdev_priv(sitn->fb_tunnel_dev))
1211 unregister_netdevice(dev);
1217 if (dev == sitn->fb_tunnel_dev)
1219 err = ipip6_tunnel_get_prl(t, ifr->ifr_ifru.ifru_data);
1226 if (!ns_capable(net->user_ns, CAP_NET_ADMIN))
1229 if (dev == sitn->fb_tunnel_dev)
1232 if (copy_from_user(&prl, ifr->ifr_ifru.ifru_data, sizeof(prl)))
1237 err = ipip6_tunnel_del_prl(t, &prl);
1241 err = ipip6_tunnel_add_prl(t, &prl, cmd == SIOCCHGPRL);
1244 dst_cache_reset(&t->dst_cache);
1245 netdev_state_change(dev);
1248 #ifdef CONFIG_IPV6_SIT_6RD
1253 if (!ns_capable(net->user_ns, CAP_NET_ADMIN))
1257 if (copy_from_user(&ip6rd, ifr->ifr_ifru.ifru_data,
1261 if (cmd != SIOCDEL6RD) {
1262 err = ipip6_tunnel_update_6rd(t, &ip6rd);
1266 ipip6_tunnel_clone_6rd(dev, sitn);
1280 static int ipip6_tunnel_change_mtu(struct net_device *dev, int new_mtu)
1282 struct ip_tunnel *tunnel = netdev_priv(dev);
1283 int t_hlen = tunnel->hlen + sizeof(struct iphdr);
1285 if (new_mtu < IPV6_MIN_MTU || new_mtu > 0xFFF8 - t_hlen)
1291 static const struct net_device_ops ipip6_netdev_ops = {
1292 .ndo_init = ipip6_tunnel_init,
1293 .ndo_uninit = ipip6_tunnel_uninit,
1294 .ndo_start_xmit = sit_tunnel_xmit,
1295 .ndo_do_ioctl = ipip6_tunnel_ioctl,
1296 .ndo_change_mtu = ipip6_tunnel_change_mtu,
1297 .ndo_get_stats64 = ip_tunnel_get_stats64,
1298 .ndo_get_iflink = ip_tunnel_get_iflink,
1301 static void ipip6_dev_free(struct net_device *dev)
1303 struct ip_tunnel *tunnel = netdev_priv(dev);
1305 dst_cache_destroy(&tunnel->dst_cache);
1306 free_percpu(dev->tstats);
1310 #define SIT_FEATURES (NETIF_F_SG | \
1311 NETIF_F_FRAGLIST | \
1313 NETIF_F_GSO_SOFTWARE | \
1316 static void ipip6_tunnel_setup(struct net_device *dev)
1318 struct ip_tunnel *tunnel = netdev_priv(dev);
1319 int t_hlen = tunnel->hlen + sizeof(struct iphdr);
1321 dev->netdev_ops = &ipip6_netdev_ops;
1322 dev->destructor = ipip6_dev_free;
1324 dev->type = ARPHRD_SIT;
1325 dev->hard_header_len = LL_MAX_HEADER + t_hlen;
1326 dev->mtu = ETH_DATA_LEN - t_hlen;
1327 dev->flags = IFF_NOARP;
1328 netif_keep_dst(dev);
1330 dev->features |= NETIF_F_LLTX;
1331 dev->features |= SIT_FEATURES;
1332 dev->hw_features |= SIT_FEATURES;
1335 static int ipip6_tunnel_init(struct net_device *dev)
1337 struct ip_tunnel *tunnel = netdev_priv(dev);
1341 tunnel->net = dev_net(dev);
1342 strcpy(tunnel->parms.name, dev->name);
1344 ipip6_tunnel_bind_dev(dev);
1345 dev->tstats = netdev_alloc_pcpu_stats(struct pcpu_sw_netstats);
1349 err = dst_cache_init(&tunnel->dst_cache, GFP_KERNEL);
1351 free_percpu(dev->tstats);
1358 static void __net_init ipip6_fb_tunnel_init(struct net_device *dev)
1360 struct ip_tunnel *tunnel = netdev_priv(dev);
1361 struct iphdr *iph = &tunnel->parms.iph;
1362 struct net *net = dev_net(dev);
1363 struct sit_net *sitn = net_generic(net, sit_net_id);
1366 iph->protocol = IPPROTO_IPV6;
1371 rcu_assign_pointer(sitn->tunnels_wc[0], tunnel);
1374 static int ipip6_validate(struct nlattr *tb[], struct nlattr *data[])
1378 if (!data || !data[IFLA_IPTUN_PROTO])
1381 proto = nla_get_u8(data[IFLA_IPTUN_PROTO]);
1382 if (proto != IPPROTO_IPV6 &&
1383 proto != IPPROTO_IPIP &&
1390 static void ipip6_netlink_parms(struct nlattr *data[],
1391 struct ip_tunnel_parm *parms)
1393 memset(parms, 0, sizeof(*parms));
1395 parms->iph.version = 4;
1396 parms->iph.protocol = IPPROTO_IPV6;
1398 parms->iph.ttl = 64;
1403 if (data[IFLA_IPTUN_LINK])
1404 parms->link = nla_get_u32(data[IFLA_IPTUN_LINK]);
1406 if (data[IFLA_IPTUN_LOCAL])
1407 parms->iph.saddr = nla_get_be32(data[IFLA_IPTUN_LOCAL]);
1409 if (data[IFLA_IPTUN_REMOTE])
1410 parms->iph.daddr = nla_get_be32(data[IFLA_IPTUN_REMOTE]);
1412 if (data[IFLA_IPTUN_TTL]) {
1413 parms->iph.ttl = nla_get_u8(data[IFLA_IPTUN_TTL]);
1415 parms->iph.frag_off = htons(IP_DF);
1418 if (data[IFLA_IPTUN_TOS])
1419 parms->iph.tos = nla_get_u8(data[IFLA_IPTUN_TOS]);
1421 if (!data[IFLA_IPTUN_PMTUDISC] || nla_get_u8(data[IFLA_IPTUN_PMTUDISC]))
1422 parms->iph.frag_off = htons(IP_DF);
1424 if (data[IFLA_IPTUN_FLAGS])
1425 parms->i_flags = nla_get_be16(data[IFLA_IPTUN_FLAGS]);
1427 if (data[IFLA_IPTUN_PROTO])
1428 parms->iph.protocol = nla_get_u8(data[IFLA_IPTUN_PROTO]);
1432 /* This function returns true when ENCAP attributes are present in the nl msg */
1433 static bool ipip6_netlink_encap_parms(struct nlattr *data[],
1434 struct ip_tunnel_encap *ipencap)
1438 memset(ipencap, 0, sizeof(*ipencap));
1443 if (data[IFLA_IPTUN_ENCAP_TYPE]) {
1445 ipencap->type = nla_get_u16(data[IFLA_IPTUN_ENCAP_TYPE]);
1448 if (data[IFLA_IPTUN_ENCAP_FLAGS]) {
1450 ipencap->flags = nla_get_u16(data[IFLA_IPTUN_ENCAP_FLAGS]);
1453 if (data[IFLA_IPTUN_ENCAP_SPORT]) {
1455 ipencap->sport = nla_get_be16(data[IFLA_IPTUN_ENCAP_SPORT]);
1458 if (data[IFLA_IPTUN_ENCAP_DPORT]) {
1460 ipencap->dport = nla_get_be16(data[IFLA_IPTUN_ENCAP_DPORT]);
1466 #ifdef CONFIG_IPV6_SIT_6RD
1467 /* This function returns true when 6RD attributes are present in the nl msg */
1468 static bool ipip6_netlink_6rd_parms(struct nlattr *data[],
1469 struct ip_tunnel_6rd *ip6rd)
1472 memset(ip6rd, 0, sizeof(*ip6rd));
1477 if (data[IFLA_IPTUN_6RD_PREFIX]) {
1479 ip6rd->prefix = nla_get_in6_addr(data[IFLA_IPTUN_6RD_PREFIX]);
1482 if (data[IFLA_IPTUN_6RD_RELAY_PREFIX]) {
1484 ip6rd->relay_prefix =
1485 nla_get_be32(data[IFLA_IPTUN_6RD_RELAY_PREFIX]);
1488 if (data[IFLA_IPTUN_6RD_PREFIXLEN]) {
1490 ip6rd->prefixlen = nla_get_u16(data[IFLA_IPTUN_6RD_PREFIXLEN]);
1493 if (data[IFLA_IPTUN_6RD_RELAY_PREFIXLEN]) {
1495 ip6rd->relay_prefixlen =
1496 nla_get_u16(data[IFLA_IPTUN_6RD_RELAY_PREFIXLEN]);
1503 static int ipip6_newlink(struct net *src_net, struct net_device *dev,
1504 struct nlattr *tb[], struct nlattr *data[])
1506 struct net *net = dev_net(dev);
1507 struct ip_tunnel *nt;
1508 struct ip_tunnel_encap ipencap;
1509 #ifdef CONFIG_IPV6_SIT_6RD
1510 struct ip_tunnel_6rd ip6rd;
1514 nt = netdev_priv(dev);
1516 if (ipip6_netlink_encap_parms(data, &ipencap)) {
1517 err = ip_tunnel_encap_setup(nt, &ipencap);
1522 ipip6_netlink_parms(data, &nt->parms);
1524 if (ipip6_tunnel_locate(net, &nt->parms, 0))
1527 err = ipip6_tunnel_create(dev);
1531 #ifdef CONFIG_IPV6_SIT_6RD
1532 if (ipip6_netlink_6rd_parms(data, &ip6rd))
1533 err = ipip6_tunnel_update_6rd(nt, &ip6rd);
1539 static int ipip6_changelink(struct net_device *dev, struct nlattr *tb[],
1540 struct nlattr *data[])
1542 struct ip_tunnel *t = netdev_priv(dev);
1543 struct ip_tunnel_parm p;
1544 struct ip_tunnel_encap ipencap;
1545 struct net *net = t->net;
1546 struct sit_net *sitn = net_generic(net, sit_net_id);
1547 #ifdef CONFIG_IPV6_SIT_6RD
1548 struct ip_tunnel_6rd ip6rd;
1552 if (dev == sitn->fb_tunnel_dev)
1555 if (ipip6_netlink_encap_parms(data, &ipencap)) {
1556 err = ip_tunnel_encap_setup(t, &ipencap);
1561 ipip6_netlink_parms(data, &p);
1563 if (((dev->flags & IFF_POINTOPOINT) && !p.iph.daddr) ||
1564 (!(dev->flags & IFF_POINTOPOINT) && p.iph.daddr))
1567 t = ipip6_tunnel_locate(net, &p, 0);
1573 t = netdev_priv(dev);
1575 ipip6_tunnel_update(t, &p);
1577 #ifdef CONFIG_IPV6_SIT_6RD
1578 if (ipip6_netlink_6rd_parms(data, &ip6rd))
1579 return ipip6_tunnel_update_6rd(t, &ip6rd);
1585 static size_t ipip6_get_size(const struct net_device *dev)
1588 /* IFLA_IPTUN_LINK */
1590 /* IFLA_IPTUN_LOCAL */
1592 /* IFLA_IPTUN_REMOTE */
1594 /* IFLA_IPTUN_TTL */
1596 /* IFLA_IPTUN_TOS */
1598 /* IFLA_IPTUN_PMTUDISC */
1600 /* IFLA_IPTUN_FLAGS */
1602 /* IFLA_IPTUN_PROTO */
1604 #ifdef CONFIG_IPV6_SIT_6RD
1605 /* IFLA_IPTUN_6RD_PREFIX */
1606 nla_total_size(sizeof(struct in6_addr)) +
1607 /* IFLA_IPTUN_6RD_RELAY_PREFIX */
1609 /* IFLA_IPTUN_6RD_PREFIXLEN */
1611 /* IFLA_IPTUN_6RD_RELAY_PREFIXLEN */
1614 /* IFLA_IPTUN_ENCAP_TYPE */
1616 /* IFLA_IPTUN_ENCAP_FLAGS */
1618 /* IFLA_IPTUN_ENCAP_SPORT */
1620 /* IFLA_IPTUN_ENCAP_DPORT */
1625 static int ipip6_fill_info(struct sk_buff *skb, const struct net_device *dev)
1627 struct ip_tunnel *tunnel = netdev_priv(dev);
1628 struct ip_tunnel_parm *parm = &tunnel->parms;
1630 if (nla_put_u32(skb, IFLA_IPTUN_LINK, parm->link) ||
1631 nla_put_in_addr(skb, IFLA_IPTUN_LOCAL, parm->iph.saddr) ||
1632 nla_put_in_addr(skb, IFLA_IPTUN_REMOTE, parm->iph.daddr) ||
1633 nla_put_u8(skb, IFLA_IPTUN_TTL, parm->iph.ttl) ||
1634 nla_put_u8(skb, IFLA_IPTUN_TOS, parm->iph.tos) ||
1635 nla_put_u8(skb, IFLA_IPTUN_PMTUDISC,
1636 !!(parm->iph.frag_off & htons(IP_DF))) ||
1637 nla_put_u8(skb, IFLA_IPTUN_PROTO, parm->iph.protocol) ||
1638 nla_put_be16(skb, IFLA_IPTUN_FLAGS, parm->i_flags))
1639 goto nla_put_failure;
1641 #ifdef CONFIG_IPV6_SIT_6RD
1642 if (nla_put_in6_addr(skb, IFLA_IPTUN_6RD_PREFIX,
1643 &tunnel->ip6rd.prefix) ||
1644 nla_put_in_addr(skb, IFLA_IPTUN_6RD_RELAY_PREFIX,
1645 tunnel->ip6rd.relay_prefix) ||
1646 nla_put_u16(skb, IFLA_IPTUN_6RD_PREFIXLEN,
1647 tunnel->ip6rd.prefixlen) ||
1648 nla_put_u16(skb, IFLA_IPTUN_6RD_RELAY_PREFIXLEN,
1649 tunnel->ip6rd.relay_prefixlen))
1650 goto nla_put_failure;
1653 if (nla_put_u16(skb, IFLA_IPTUN_ENCAP_TYPE,
1654 tunnel->encap.type) ||
1655 nla_put_be16(skb, IFLA_IPTUN_ENCAP_SPORT,
1656 tunnel->encap.sport) ||
1657 nla_put_be16(skb, IFLA_IPTUN_ENCAP_DPORT,
1658 tunnel->encap.dport) ||
1659 nla_put_u16(skb, IFLA_IPTUN_ENCAP_FLAGS,
1660 tunnel->encap.flags))
1661 goto nla_put_failure;
1669 static const struct nla_policy ipip6_policy[IFLA_IPTUN_MAX + 1] = {
1670 [IFLA_IPTUN_LINK] = { .type = NLA_U32 },
1671 [IFLA_IPTUN_LOCAL] = { .type = NLA_U32 },
1672 [IFLA_IPTUN_REMOTE] = { .type = NLA_U32 },
1673 [IFLA_IPTUN_TTL] = { .type = NLA_U8 },
1674 [IFLA_IPTUN_TOS] = { .type = NLA_U8 },
1675 [IFLA_IPTUN_PMTUDISC] = { .type = NLA_U8 },
1676 [IFLA_IPTUN_FLAGS] = { .type = NLA_U16 },
1677 [IFLA_IPTUN_PROTO] = { .type = NLA_U8 },
1678 #ifdef CONFIG_IPV6_SIT_6RD
1679 [IFLA_IPTUN_6RD_PREFIX] = { .len = sizeof(struct in6_addr) },
1680 [IFLA_IPTUN_6RD_RELAY_PREFIX] = { .type = NLA_U32 },
1681 [IFLA_IPTUN_6RD_PREFIXLEN] = { .type = NLA_U16 },
1682 [IFLA_IPTUN_6RD_RELAY_PREFIXLEN] = { .type = NLA_U16 },
1684 [IFLA_IPTUN_ENCAP_TYPE] = { .type = NLA_U16 },
1685 [IFLA_IPTUN_ENCAP_FLAGS] = { .type = NLA_U16 },
1686 [IFLA_IPTUN_ENCAP_SPORT] = { .type = NLA_U16 },
1687 [IFLA_IPTUN_ENCAP_DPORT] = { .type = NLA_U16 },
1690 static void ipip6_dellink(struct net_device *dev, struct list_head *head)
1692 struct net *net = dev_net(dev);
1693 struct sit_net *sitn = net_generic(net, sit_net_id);
1695 if (dev != sitn->fb_tunnel_dev)
1696 unregister_netdevice_queue(dev, head);
1699 static struct rtnl_link_ops sit_link_ops __read_mostly = {
1701 .maxtype = IFLA_IPTUN_MAX,
1702 .policy = ipip6_policy,
1703 .priv_size = sizeof(struct ip_tunnel),
1704 .setup = ipip6_tunnel_setup,
1705 .validate = ipip6_validate,
1706 .newlink = ipip6_newlink,
1707 .changelink = ipip6_changelink,
1708 .get_size = ipip6_get_size,
1709 .fill_info = ipip6_fill_info,
1710 .dellink = ipip6_dellink,
1711 .get_link_net = ip_tunnel_get_link_net,
1714 static struct xfrm_tunnel sit_handler __read_mostly = {
1715 .handler = ipip6_rcv,
1716 .err_handler = ipip6_err,
1720 static struct xfrm_tunnel ipip_handler __read_mostly = {
1721 .handler = ipip_rcv,
1722 .err_handler = ipip6_err,
1726 static void __net_exit sit_destroy_tunnels(struct net *net,
1727 struct list_head *head)
1729 struct sit_net *sitn = net_generic(net, sit_net_id);
1730 struct net_device *dev, *aux;
1733 for_each_netdev_safe(net, dev, aux)
1734 if (dev->rtnl_link_ops == &sit_link_ops)
1735 unregister_netdevice_queue(dev, head);
1737 for (prio = 1; prio < 4; prio++) {
1739 for (h = 0; h < HASH_SIZE; h++) {
1740 struct ip_tunnel *t;
1742 t = rtnl_dereference(sitn->tunnels[prio][h]);
1744 /* If dev is in the same netns, it has already
1745 * been added to the list by the previous loop.
1747 if (!net_eq(dev_net(t->dev), net))
1748 unregister_netdevice_queue(t->dev,
1750 t = rtnl_dereference(t->next);
1756 static int __net_init sit_init_net(struct net *net)
1758 struct sit_net *sitn = net_generic(net, sit_net_id);
1759 struct ip_tunnel *t;
1762 sitn->tunnels[0] = sitn->tunnels_wc;
1763 sitn->tunnels[1] = sitn->tunnels_l;
1764 sitn->tunnels[2] = sitn->tunnels_r;
1765 sitn->tunnels[3] = sitn->tunnels_r_l;
1767 sitn->fb_tunnel_dev = alloc_netdev(sizeof(struct ip_tunnel), "sit0",
1769 ipip6_tunnel_setup);
1770 if (!sitn->fb_tunnel_dev) {
1774 dev_net_set(sitn->fb_tunnel_dev, net);
1775 sitn->fb_tunnel_dev->rtnl_link_ops = &sit_link_ops;
1776 /* FB netdevice is special: we have one, and only one per netns.
1777 * Allowing to move it to another netns is clearly unsafe.
1779 sitn->fb_tunnel_dev->features |= NETIF_F_NETNS_LOCAL;
1781 err = register_netdev(sitn->fb_tunnel_dev);
1785 ipip6_tunnel_clone_6rd(sitn->fb_tunnel_dev, sitn);
1786 ipip6_fb_tunnel_init(sitn->fb_tunnel_dev);
1788 t = netdev_priv(sitn->fb_tunnel_dev);
1790 strcpy(t->parms.name, sitn->fb_tunnel_dev->name);
1794 ipip6_dev_free(sitn->fb_tunnel_dev);
1799 static void __net_exit sit_exit_net(struct net *net)
1804 sit_destroy_tunnels(net, &list);
1805 unregister_netdevice_many(&list);
1809 static struct pernet_operations sit_net_ops = {
1810 .init = sit_init_net,
1811 .exit = sit_exit_net,
1813 .size = sizeof(struct sit_net),
1816 static void __exit sit_cleanup(void)
1818 rtnl_link_unregister(&sit_link_ops);
1819 xfrm4_tunnel_deregister(&sit_handler, AF_INET6);
1820 xfrm4_tunnel_deregister(&ipip_handler, AF_INET);
1822 unregister_pernet_device(&sit_net_ops);
1823 rcu_barrier(); /* Wait for completion of call_rcu()'s */
1826 static int __init sit_init(void)
1830 pr_info("IPv6 over IPv4 tunneling driver\n");
1832 err = register_pernet_device(&sit_net_ops);
1835 err = xfrm4_tunnel_register(&sit_handler, AF_INET6);
1837 pr_info("%s: can't register ip6ip4\n", __func__);
1838 goto xfrm_tunnel_failed;
1840 err = xfrm4_tunnel_register(&ipip_handler, AF_INET);
1842 pr_info("%s: can't register ip4ip4\n", __func__);
1843 goto xfrm_tunnel4_failed;
1845 err = rtnl_link_register(&sit_link_ops);
1847 goto rtnl_link_failed;
1853 xfrm4_tunnel_deregister(&ipip_handler, AF_INET);
1854 xfrm_tunnel4_failed:
1855 xfrm4_tunnel_deregister(&sit_handler, AF_INET6);
1857 unregister_pernet_device(&sit_net_ops);
1861 module_init(sit_init);
1862 module_exit(sit_cleanup);
1863 MODULE_LICENSE("GPL");
1864 MODULE_ALIAS_RTNL_LINK("sit");
1865 MODULE_ALIAS_NETDEV("sit0");