2 * IPv6 virtual tunneling interface
4 * Copyright (C) 2013 secunet Security Networks AG
7 * Steffen Klassert <steffen.klassert@secunet.com>
10 * net/ipv6/ip6_tunnel.c
12 * This program is free software; you can redistribute it and/or
13 * modify it under the terms of the GNU General Public License
14 * as published by the Free Software Foundation; either version
15 * 2 of the License, or (at your option) any later version.
18 #include <linux/module.h>
19 #include <linux/capability.h>
20 #include <linux/errno.h>
21 #include <linux/types.h>
22 #include <linux/sockios.h>
23 #include <linux/icmp.h>
27 #include <linux/net.h>
28 #include <linux/in6.h>
29 #include <linux/netdevice.h>
30 #include <linux/if_arp.h>
31 #include <linux/icmpv6.h>
32 #include <linux/init.h>
33 #include <linux/route.h>
34 #include <linux/rtnetlink.h>
35 #include <linux/netfilter_ipv6.h>
36 #include <linux/slab.h>
37 #include <linux/hash.h>
39 #include <linux/uaccess.h>
40 #include <linux/atomic.h>
44 #include <net/ip_tunnels.h>
46 #include <net/ip6_route.h>
47 #include <net/addrconf.h>
48 #include <net/ip6_tunnel.h>
50 #include <net/net_namespace.h>
51 #include <net/netns/generic.h>
52 #include <linux/etherdevice.h>
54 #define IP6_VTI_HASH_SIZE_SHIFT 5
55 #define IP6_VTI_HASH_SIZE (1 << IP6_VTI_HASH_SIZE_SHIFT)
57 static u32 HASH(const struct in6_addr *addr1, const struct in6_addr *addr2)
59 u32 hash = ipv6_addr_hash(addr1) ^ ipv6_addr_hash(addr2);
61 return hash_32(hash, IP6_VTI_HASH_SIZE_SHIFT);
64 static int vti6_dev_init(struct net_device *dev);
65 static void vti6_dev_setup(struct net_device *dev);
66 static struct rtnl_link_ops vti6_link_ops __read_mostly;
68 static unsigned int vti6_net_id __read_mostly;
70 /* the vti6 tunnel fallback device */
71 struct net_device *fb_tnl_dev;
72 /* lists for storing tunnels in use */
73 struct ip6_tnl __rcu *tnls_r_l[IP6_VTI_HASH_SIZE];
74 struct ip6_tnl __rcu *tnls_wc[1];
75 struct ip6_tnl __rcu **tnls[2];
78 #define for_each_vti6_tunnel_rcu(start) \
79 for (t = rcu_dereference(start); t; t = rcu_dereference(t->next))
82 * vti6_tnl_lookup - fetch tunnel matching the end-point addresses
83 * @net: network namespace
84 * @remote: the address of the tunnel exit-point
85 * @local: the address of the tunnel entry-point
88 * tunnel matching given end-points if found,
89 * else fallback tunnel if its device is up,
92 static struct ip6_tnl *
93 vti6_tnl_lookup(struct net *net, const struct in6_addr *remote,
94 const struct in6_addr *local)
96 unsigned int hash = HASH(remote, local);
98 struct vti6_net *ip6n = net_generic(net, vti6_net_id);
101 for_each_vti6_tunnel_rcu(ip6n->tnls_r_l[hash]) {
102 if (ipv6_addr_equal(local, &t->parms.laddr) &&
103 ipv6_addr_equal(remote, &t->parms.raddr) &&
104 (t->dev->flags & IFF_UP))
108 memset(&any, 0, sizeof(any));
109 hash = HASH(&any, local);
110 for_each_vti6_tunnel_rcu(ip6n->tnls_r_l[hash]) {
111 if (ipv6_addr_equal(local, &t->parms.laddr) &&
112 (t->dev->flags & IFF_UP))
116 hash = HASH(remote, &any);
117 for_each_vti6_tunnel_rcu(ip6n->tnls_r_l[hash]) {
118 if (ipv6_addr_equal(remote, &t->parms.raddr) &&
119 (t->dev->flags & IFF_UP))
123 t = rcu_dereference(ip6n->tnls_wc[0]);
124 if (t && (t->dev->flags & IFF_UP))
131 * vti6_tnl_bucket - get head of list matching given tunnel parameters
132 * @p: parameters containing tunnel end-points
135 * vti6_tnl_bucket() returns the head of the list matching the
136 * &struct in6_addr entries laddr and raddr in @p.
138 * Return: head of IPv6 tunnel list
140 static struct ip6_tnl __rcu **
141 vti6_tnl_bucket(struct vti6_net *ip6n, const struct __ip6_tnl_parm *p)
143 const struct in6_addr *remote = &p->raddr;
144 const struct in6_addr *local = &p->laddr;
148 if (!ipv6_addr_any(remote) || !ipv6_addr_any(local)) {
150 h = HASH(remote, local);
152 return &ip6n->tnls[prio][h];
156 vti6_tnl_link(struct vti6_net *ip6n, struct ip6_tnl *t)
158 struct ip6_tnl __rcu **tp = vti6_tnl_bucket(ip6n, &t->parms);
160 rcu_assign_pointer(t->next , rtnl_dereference(*tp));
161 rcu_assign_pointer(*tp, t);
165 vti6_tnl_unlink(struct vti6_net *ip6n, struct ip6_tnl *t)
167 struct ip6_tnl __rcu **tp;
168 struct ip6_tnl *iter;
170 for (tp = vti6_tnl_bucket(ip6n, &t->parms);
171 (iter = rtnl_dereference(*tp)) != NULL;
174 rcu_assign_pointer(*tp, t->next);
180 static void vti6_dev_free(struct net_device *dev)
182 free_percpu(dev->tstats);
186 static int vti6_tnl_create2(struct net_device *dev)
188 struct ip6_tnl *t = netdev_priv(dev);
189 struct net *net = dev_net(dev);
190 struct vti6_net *ip6n = net_generic(net, vti6_net_id);
193 dev->rtnl_link_ops = &vti6_link_ops;
194 err = register_netdevice(dev);
198 strcpy(t->parms.name, dev->name);
201 vti6_tnl_link(ip6n, t);
209 static struct ip6_tnl *vti6_tnl_create(struct net *net, struct __ip6_tnl_parm *p)
211 struct net_device *dev;
217 strlcpy(name, p->name, IFNAMSIZ);
219 sprintf(name, "ip6_vti%%d");
221 dev = alloc_netdev(sizeof(*t), name, NET_NAME_UNKNOWN, vti6_dev_setup);
225 dev_net_set(dev, net);
227 t = netdev_priv(dev);
229 t->net = dev_net(dev);
231 err = vti6_tnl_create2(dev);
244 * vti6_locate - find or create tunnel matching given parameters
245 * @net: network namespace
246 * @p: tunnel parameters
247 * @create: != 0 if allowed to create new tunnel if no match found
250 * vti6_locate() first tries to locate an existing tunnel
251 * based on @parms. If this is unsuccessful, but @create is set a new
252 * tunnel device is created and registered for use.
255 * matching tunnel or NULL
257 static struct ip6_tnl *vti6_locate(struct net *net, struct __ip6_tnl_parm *p,
260 const struct in6_addr *remote = &p->raddr;
261 const struct in6_addr *local = &p->laddr;
262 struct ip6_tnl __rcu **tp;
264 struct vti6_net *ip6n = net_generic(net, vti6_net_id);
266 for (tp = vti6_tnl_bucket(ip6n, p);
267 (t = rtnl_dereference(*tp)) != NULL;
269 if (ipv6_addr_equal(local, &t->parms.laddr) &&
270 ipv6_addr_equal(remote, &t->parms.raddr)) {
279 return vti6_tnl_create(net, p);
283 * vti6_dev_uninit - tunnel device uninitializer
284 * @dev: the device to be destroyed
287 * vti6_dev_uninit() removes tunnel from its list
289 static void vti6_dev_uninit(struct net_device *dev)
291 struct ip6_tnl *t = netdev_priv(dev);
292 struct vti6_net *ip6n = net_generic(t->net, vti6_net_id);
294 if (dev == ip6n->fb_tnl_dev)
295 RCU_INIT_POINTER(ip6n->tnls_wc[0], NULL);
297 vti6_tnl_unlink(ip6n, t);
301 static int vti6_rcv(struct sk_buff *skb)
304 const struct ipv6hdr *ipv6h = ipv6_hdr(skb);
307 t = vti6_tnl_lookup(dev_net(skb->dev), &ipv6h->saddr, &ipv6h->daddr);
309 if (t->parms.proto != IPPROTO_IPV6 && t->parms.proto != 0) {
314 if (!xfrm6_policy_check(NULL, XFRM_POLICY_IN, skb)) {
319 if (!ip6_tnl_rcv_ctl(t, &ipv6h->daddr, &ipv6h->saddr)) {
320 t->dev->stats.rx_dropped++;
327 return xfrm6_rcv_tnl(skb, t);
336 static int vti6_rcv_cb(struct sk_buff *skb, int err)
338 unsigned short family;
339 struct net_device *dev;
340 struct pcpu_sw_netstats *tstats;
341 struct xfrm_state *x;
342 struct xfrm_mode *inner_mode;
343 struct ip6_tnl *t = XFRM_TUNNEL_SKB_CB(skb)->tunnel.ip6;
344 u32 orig_mark = skb->mark;
353 dev->stats.rx_errors++;
354 dev->stats.rx_dropped++;
359 x = xfrm_input_state(skb);
361 inner_mode = x->inner_mode;
363 if (x->sel.family == AF_UNSPEC) {
364 inner_mode = xfrm_ip2inner_mode(x, XFRM_MODE_SKB_CB(skb)->protocol);
365 if (inner_mode == NULL) {
366 XFRM_INC_STATS(dev_net(skb->dev),
367 LINUX_MIB_XFRMINSTATEMODEERROR);
372 family = inner_mode->afinfo->family;
374 skb->mark = be32_to_cpu(t->parms.i_key);
375 ret = xfrm_policy_check(NULL, XFRM_POLICY_IN, skb, family);
376 skb->mark = orig_mark;
381 skb_scrub_packet(skb, !net_eq(t->net, dev_net(skb->dev)));
384 tstats = this_cpu_ptr(dev->tstats);
385 u64_stats_update_begin(&tstats->syncp);
386 tstats->rx_packets++;
387 tstats->rx_bytes += skb->len;
388 u64_stats_update_end(&tstats->syncp);
394 * vti6_addr_conflict - compare packet addresses to tunnel's own
395 * @t: the outgoing tunnel device
396 * @hdr: IPv6 header from the incoming packet
399 * Avoid trivial tunneling loop by checking that tunnel exit-point
400 * doesn't match source of incoming packet.
407 vti6_addr_conflict(const struct ip6_tnl *t, const struct ipv6hdr *hdr)
409 return ipv6_addr_equal(&t->parms.raddr, &hdr->saddr);
412 static bool vti6_state_check(const struct xfrm_state *x,
413 const struct in6_addr *dst,
414 const struct in6_addr *src)
416 xfrm_address_t *daddr = (xfrm_address_t *)dst;
417 xfrm_address_t *saddr = (xfrm_address_t *)src;
419 /* if there is no transform then this tunnel is not functional.
420 * Or if the xfrm is not mode tunnel.
422 if (!x || x->props.mode != XFRM_MODE_TUNNEL ||
423 x->props.family != AF_INET6)
426 if (ipv6_addr_any(dst))
427 return xfrm_addr_equal(saddr, &x->props.saddr, AF_INET6);
429 if (!xfrm_state_addr_check(x, daddr, saddr, AF_INET6))
436 * vti6_xmit - send a packet
437 * @skb: the outgoing socket buffer
438 * @dev: the outgoing tunnel device
439 * @fl: the flow informations for the xfrm_lookup
442 vti6_xmit(struct sk_buff *skb, struct net_device *dev, struct flowi *fl)
444 struct ip6_tnl *t = netdev_priv(dev);
445 struct net_device_stats *stats = &t->dev->stats;
446 struct dst_entry *dst = skb_dst(skb);
447 struct net_device *tdev;
448 struct xfrm_state *x;
453 goto tx_err_link_failure;
456 dst = xfrm_lookup(t->net, dst, fl, NULL, 0);
460 goto tx_err_link_failure;
464 if (!vti6_state_check(x, &t->parms.raddr, &t->parms.laddr))
465 goto tx_err_link_failure;
467 if (!ip6_tnl_xmit_ctl(t, (const struct in6_addr *)&x->props.saddr,
468 (const struct in6_addr *)&x->id.daddr))
469 goto tx_err_link_failure;
475 net_warn_ratelimited("%s: Local routing loop detected!\n",
477 goto tx_err_dst_release;
480 skb_scrub_packet(skb, !net_eq(t->net, dev_net(dev)));
481 skb_dst_set(skb, dst);
482 skb->dev = skb_dst(skb)->dev;
485 if (!skb->ignore_df && skb->len > mtu) {
486 skb_dst(skb)->ops->update_pmtu(dst, NULL, skb, mtu);
488 if (skb->protocol == htons(ETH_P_IPV6))
489 icmpv6_send(skb, ICMPV6_PKT_TOOBIG, 0, mtu);
491 icmp_send(skb, ICMP_DEST_UNREACH, ICMP_FRAG_NEEDED,
497 err = dst_output(t->net, skb->sk, skb);
498 if (net_xmit_eval(err) == 0) {
499 struct pcpu_sw_netstats *tstats = this_cpu_ptr(dev->tstats);
501 u64_stats_update_begin(&tstats->syncp);
502 tstats->tx_bytes += skb->len;
503 tstats->tx_packets++;
504 u64_stats_update_end(&tstats->syncp);
507 stats->tx_aborted_errors++;
512 stats->tx_carrier_errors++;
513 dst_link_failure(skb);
520 vti6_tnl_xmit(struct sk_buff *skb, struct net_device *dev)
522 struct ip6_tnl *t = netdev_priv(dev);
523 struct net_device_stats *stats = &t->dev->stats;
524 struct ipv6hdr *ipv6h;
528 memset(&fl, 0, sizeof(fl));
530 switch (skb->protocol) {
531 case htons(ETH_P_IPV6):
532 ipv6h = ipv6_hdr(skb);
534 if ((t->parms.proto != IPPROTO_IPV6 && t->parms.proto != 0) ||
535 vti6_addr_conflict(t, ipv6h))
538 xfrm_decode_session(skb, &fl, AF_INET6);
539 memset(IP6CB(skb), 0, sizeof(*IP6CB(skb)));
541 case htons(ETH_P_IP):
542 xfrm_decode_session(skb, &fl, AF_INET);
543 memset(IPCB(skb), 0, sizeof(*IPCB(skb)));
549 /* override mark with tunnel output key */
550 fl.flowi_mark = be32_to_cpu(t->parms.o_key);
552 ret = vti6_xmit(skb, dev, &fl);
565 static int vti6_err(struct sk_buff *skb, struct inet6_skb_parm *opt,
566 u8 type, u8 code, int offset, __be32 info)
570 struct xfrm_state *x;
572 struct ip_esp_hdr *esph;
573 struct ip_auth_hdr *ah;
574 struct ip_comp_hdr *ipch;
575 struct net *net = dev_net(skb->dev);
576 const struct ipv6hdr *iph = (const struct ipv6hdr *)skb->data;
577 int protocol = iph->nexthdr;
579 t = vti6_tnl_lookup(dev_net(skb->dev), &iph->daddr, &iph->saddr);
583 mark = be32_to_cpu(t->parms.o_key);
587 esph = (struct ip_esp_hdr *)(skb->data + offset);
591 ah = (struct ip_auth_hdr *)(skb->data + offset);
595 ipch = (struct ip_comp_hdr *)(skb->data + offset);
596 spi = htonl(ntohs(ipch->cpi));
602 if (type != ICMPV6_PKT_TOOBIG &&
603 type != NDISC_REDIRECT)
606 x = xfrm_state_lookup(net, mark, (const xfrm_address_t *)&iph->daddr,
607 spi, protocol, AF_INET6);
611 if (type == NDISC_REDIRECT)
612 ip6_redirect(skb, net, skb->dev->ifindex, 0,
613 sock_net_uid(net, NULL));
615 ip6_update_pmtu(skb, net, info, 0, 0, sock_net_uid(net, NULL));
621 static void vti6_link_config(struct ip6_tnl *t)
623 struct net_device *dev = t->dev;
624 struct __ip6_tnl_parm *p = &t->parms;
626 memcpy(dev->dev_addr, &p->laddr, sizeof(struct in6_addr));
627 memcpy(dev->broadcast, &p->raddr, sizeof(struct in6_addr));
629 p->flags &= ~(IP6_TNL_F_CAP_XMIT | IP6_TNL_F_CAP_RCV |
630 IP6_TNL_F_CAP_PER_PACKET);
631 p->flags |= ip6_tnl_get_cap(t, &p->laddr, &p->raddr);
633 if (p->flags & IP6_TNL_F_CAP_XMIT && p->flags & IP6_TNL_F_CAP_RCV)
634 dev->flags |= IFF_POINTOPOINT;
636 dev->flags &= ~IFF_POINTOPOINT;
640 * vti6_tnl_change - update the tunnel parameters
641 * @t: tunnel to be changed
642 * @p: tunnel configuration parameters
645 * vti6_tnl_change() updates the tunnel parameters
648 vti6_tnl_change(struct ip6_tnl *t, const struct __ip6_tnl_parm *p)
650 t->parms.laddr = p->laddr;
651 t->parms.raddr = p->raddr;
652 t->parms.link = p->link;
653 t->parms.i_key = p->i_key;
654 t->parms.o_key = p->o_key;
655 t->parms.proto = p->proto;
656 dst_cache_reset(&t->dst_cache);
661 static int vti6_update(struct ip6_tnl *t, struct __ip6_tnl_parm *p)
663 struct net *net = dev_net(t->dev);
664 struct vti6_net *ip6n = net_generic(net, vti6_net_id);
667 vti6_tnl_unlink(ip6n, t);
669 err = vti6_tnl_change(t, p);
670 vti6_tnl_link(ip6n, t);
671 netdev_state_change(t->dev);
676 vti6_parm_from_user(struct __ip6_tnl_parm *p, const struct ip6_tnl_parm2 *u)
685 memcpy(p->name, u->name, sizeof(u->name));
689 vti6_parm_to_user(struct ip6_tnl_parm2 *u, const struct __ip6_tnl_parm *p)
697 u->i_flags |= GRE_KEY;
699 u->o_flags |= GRE_KEY;
702 memcpy(u->name, p->name, sizeof(u->name));
706 * vti6_tnl_ioctl - configure vti6 tunnels from userspace
707 * @dev: virtual device associated with tunnel
708 * @ifr: parameters passed from userspace
709 * @cmd: command to be performed
712 * vti6_ioctl() is used for managing vti6 tunnels
715 * The possible commands are the following:
716 * %SIOCGETTUNNEL: get tunnel parameters for device
717 * %SIOCADDTUNNEL: add tunnel matching given tunnel parameters
718 * %SIOCCHGTUNNEL: change tunnel parameters to those given
719 * %SIOCDELTUNNEL: delete tunnel
721 * The fallback device "ip6_vti0", created during module
722 * initialization, can be used for creating other tunnel devices.
726 * %-EFAULT if unable to copy data to or from userspace,
727 * %-EPERM if current process hasn't %CAP_NET_ADMIN set
728 * %-EINVAL if passed tunnel parameters are invalid,
729 * %-EEXIST if changing a tunnel's parameters would cause a conflict
730 * %-ENODEV if attempting to change or delete a nonexisting device
733 vti6_ioctl(struct net_device *dev, struct ifreq *ifr, int cmd)
736 struct ip6_tnl_parm2 p;
737 struct __ip6_tnl_parm p1;
738 struct ip6_tnl *t = NULL;
739 struct net *net = dev_net(dev);
740 struct vti6_net *ip6n = net_generic(net, vti6_net_id);
744 if (dev == ip6n->fb_tnl_dev) {
745 if (copy_from_user(&p, ifr->ifr_ifru.ifru_data, sizeof(p))) {
749 vti6_parm_from_user(&p1, &p);
750 t = vti6_locate(net, &p1, 0);
752 memset(&p, 0, sizeof(p));
755 t = netdev_priv(dev);
756 vti6_parm_to_user(&p, &t->parms);
757 if (copy_to_user(ifr->ifr_ifru.ifru_data, &p, sizeof(p)))
763 if (!ns_capable(net->user_ns, CAP_NET_ADMIN))
766 if (copy_from_user(&p, ifr->ifr_ifru.ifru_data, sizeof(p)))
769 if (p.proto != IPPROTO_IPV6 && p.proto != 0)
771 vti6_parm_from_user(&p1, &p);
772 t = vti6_locate(net, &p1, cmd == SIOCADDTUNNEL);
773 if (dev != ip6n->fb_tnl_dev && cmd == SIOCCHGTUNNEL) {
780 t = netdev_priv(dev);
782 err = vti6_update(t, &p1);
786 vti6_parm_to_user(&p, &t->parms);
787 if (copy_to_user(ifr->ifr_ifru.ifru_data, &p, sizeof(p)))
791 err = (cmd == SIOCADDTUNNEL ? -ENOBUFS : -ENOENT);
795 if (!ns_capable(net->user_ns, CAP_NET_ADMIN))
798 if (dev == ip6n->fb_tnl_dev) {
800 if (copy_from_user(&p, ifr->ifr_ifru.ifru_data, sizeof(p)))
803 vti6_parm_from_user(&p1, &p);
804 t = vti6_locate(net, &p1, 0);
808 if (t->dev == ip6n->fb_tnl_dev)
813 unregister_netdevice(dev);
821 static const struct net_device_ops vti6_netdev_ops = {
822 .ndo_init = vti6_dev_init,
823 .ndo_uninit = vti6_dev_uninit,
824 .ndo_start_xmit = vti6_tnl_xmit,
825 .ndo_do_ioctl = vti6_ioctl,
826 .ndo_get_stats64 = ip_tunnel_get_stats64,
827 .ndo_get_iflink = ip6_tnl_get_iflink,
831 * vti6_dev_setup - setup virtual tunnel device
832 * @dev: virtual device associated with tunnel
835 * Initialize function pointers and device parameters
837 static void vti6_dev_setup(struct net_device *dev)
839 dev->netdev_ops = &vti6_netdev_ops;
840 dev->destructor = vti6_dev_free;
842 dev->type = ARPHRD_TUNNEL6;
843 dev->hard_header_len = LL_MAX_HEADER + sizeof(struct ipv6hdr);
844 dev->mtu = ETH_DATA_LEN;
845 dev->min_mtu = IPV6_MIN_MTU;
846 dev->max_mtu = IP_MAX_MTU;
847 dev->flags |= IFF_NOARP;
848 dev->addr_len = sizeof(struct in6_addr);
850 /* This perm addr will be used as interface identifier by IPv6 */
851 dev->addr_assign_type = NET_ADDR_RANDOM;
852 eth_random_addr(dev->perm_addr);
856 * vti6_dev_init_gen - general initializer for all tunnel devices
857 * @dev: virtual device associated with tunnel
859 static inline int vti6_dev_init_gen(struct net_device *dev)
861 struct ip6_tnl *t = netdev_priv(dev);
864 t->net = dev_net(dev);
865 dev->tstats = netdev_alloc_pcpu_stats(struct pcpu_sw_netstats);
872 * vti6_dev_init - initializer for all non fallback tunnel devices
873 * @dev: virtual device associated with tunnel
875 static int vti6_dev_init(struct net_device *dev)
877 struct ip6_tnl *t = netdev_priv(dev);
878 int err = vti6_dev_init_gen(dev);
887 * vti6_fb_tnl_dev_init - initializer for fallback tunnel device
888 * @dev: fallback device
892 static int __net_init vti6_fb_tnl_dev_init(struct net_device *dev)
894 struct ip6_tnl *t = netdev_priv(dev);
895 struct net *net = dev_net(dev);
896 struct vti6_net *ip6n = net_generic(net, vti6_net_id);
898 t->parms.proto = IPPROTO_IPV6;
901 rcu_assign_pointer(ip6n->tnls_wc[0], t);
905 static int vti6_validate(struct nlattr *tb[], struct nlattr *data[])
910 static void vti6_netlink_parms(struct nlattr *data[],
911 struct __ip6_tnl_parm *parms)
913 memset(parms, 0, sizeof(*parms));
918 if (data[IFLA_VTI_LINK])
919 parms->link = nla_get_u32(data[IFLA_VTI_LINK]);
921 if (data[IFLA_VTI_LOCAL])
922 parms->laddr = nla_get_in6_addr(data[IFLA_VTI_LOCAL]);
924 if (data[IFLA_VTI_REMOTE])
925 parms->raddr = nla_get_in6_addr(data[IFLA_VTI_REMOTE]);
927 if (data[IFLA_VTI_IKEY])
928 parms->i_key = nla_get_be32(data[IFLA_VTI_IKEY]);
930 if (data[IFLA_VTI_OKEY])
931 parms->o_key = nla_get_be32(data[IFLA_VTI_OKEY]);
934 static int vti6_newlink(struct net *src_net, struct net_device *dev,
935 struct nlattr *tb[], struct nlattr *data[])
937 struct net *net = dev_net(dev);
940 nt = netdev_priv(dev);
941 vti6_netlink_parms(data, &nt->parms);
943 nt->parms.proto = IPPROTO_IPV6;
945 if (vti6_locate(net, &nt->parms, 0))
948 return vti6_tnl_create2(dev);
951 static void vti6_dellink(struct net_device *dev, struct list_head *head)
953 struct net *net = dev_net(dev);
954 struct vti6_net *ip6n = net_generic(net, vti6_net_id);
956 if (dev != ip6n->fb_tnl_dev)
957 unregister_netdevice_queue(dev, head);
960 static int vti6_changelink(struct net_device *dev, struct nlattr *tb[],
961 struct nlattr *data[])
964 struct __ip6_tnl_parm p;
965 struct net *net = dev_net(dev);
966 struct vti6_net *ip6n = net_generic(net, vti6_net_id);
968 if (dev == ip6n->fb_tnl_dev)
971 vti6_netlink_parms(data, &p);
973 t = vti6_locate(net, &p, 0);
979 t = netdev_priv(dev);
981 return vti6_update(t, &p);
984 static size_t vti6_get_size(const struct net_device *dev)
990 nla_total_size(sizeof(struct in6_addr)) +
991 /* IFLA_VTI_REMOTE */
992 nla_total_size(sizeof(struct in6_addr)) +
1000 static int vti6_fill_info(struct sk_buff *skb, const struct net_device *dev)
1002 struct ip6_tnl *tunnel = netdev_priv(dev);
1003 struct __ip6_tnl_parm *parm = &tunnel->parms;
1005 if (nla_put_u32(skb, IFLA_VTI_LINK, parm->link) ||
1006 nla_put_in6_addr(skb, IFLA_VTI_LOCAL, &parm->laddr) ||
1007 nla_put_in6_addr(skb, IFLA_VTI_REMOTE, &parm->raddr) ||
1008 nla_put_be32(skb, IFLA_VTI_IKEY, parm->i_key) ||
1009 nla_put_be32(skb, IFLA_VTI_OKEY, parm->o_key))
1010 goto nla_put_failure;
1017 static const struct nla_policy vti6_policy[IFLA_VTI_MAX + 1] = {
1018 [IFLA_VTI_LINK] = { .type = NLA_U32 },
1019 [IFLA_VTI_LOCAL] = { .len = sizeof(struct in6_addr) },
1020 [IFLA_VTI_REMOTE] = { .len = sizeof(struct in6_addr) },
1021 [IFLA_VTI_IKEY] = { .type = NLA_U32 },
1022 [IFLA_VTI_OKEY] = { .type = NLA_U32 },
1025 static struct rtnl_link_ops vti6_link_ops __read_mostly = {
1027 .maxtype = IFLA_VTI_MAX,
1028 .policy = vti6_policy,
1029 .priv_size = sizeof(struct ip6_tnl),
1030 .setup = vti6_dev_setup,
1031 .validate = vti6_validate,
1032 .newlink = vti6_newlink,
1033 .dellink = vti6_dellink,
1034 .changelink = vti6_changelink,
1035 .get_size = vti6_get_size,
1036 .fill_info = vti6_fill_info,
1037 .get_link_net = ip6_tnl_get_link_net,
1040 static void __net_exit vti6_destroy_tunnels(struct vti6_net *ip6n)
1046 for (h = 0; h < IP6_VTI_HASH_SIZE; h++) {
1047 t = rtnl_dereference(ip6n->tnls_r_l[h]);
1049 unregister_netdevice_queue(t->dev, &list);
1050 t = rtnl_dereference(t->next);
1054 t = rtnl_dereference(ip6n->tnls_wc[0]);
1055 unregister_netdevice_queue(t->dev, &list);
1056 unregister_netdevice_many(&list);
1059 static int __net_init vti6_init_net(struct net *net)
1061 struct vti6_net *ip6n = net_generic(net, vti6_net_id);
1062 struct ip6_tnl *t = NULL;
1065 ip6n->tnls[0] = ip6n->tnls_wc;
1066 ip6n->tnls[1] = ip6n->tnls_r_l;
1069 ip6n->fb_tnl_dev = alloc_netdev(sizeof(struct ip6_tnl), "ip6_vti0",
1070 NET_NAME_UNKNOWN, vti6_dev_setup);
1072 if (!ip6n->fb_tnl_dev)
1074 dev_net_set(ip6n->fb_tnl_dev, net);
1075 ip6n->fb_tnl_dev->rtnl_link_ops = &vti6_link_ops;
1077 err = vti6_fb_tnl_dev_init(ip6n->fb_tnl_dev);
1081 err = register_netdev(ip6n->fb_tnl_dev);
1085 t = netdev_priv(ip6n->fb_tnl_dev);
1087 strcpy(t->parms.name, ip6n->fb_tnl_dev->name);
1091 vti6_dev_free(ip6n->fb_tnl_dev);
1096 static void __net_exit vti6_exit_net(struct net *net)
1098 struct vti6_net *ip6n = net_generic(net, vti6_net_id);
1101 vti6_destroy_tunnels(ip6n);
1105 static struct pernet_operations vti6_net_ops = {
1106 .init = vti6_init_net,
1107 .exit = vti6_exit_net,
1109 .size = sizeof(struct vti6_net),
1112 static struct xfrm6_protocol vti_esp6_protocol __read_mostly = {
1113 .handler = vti6_rcv,
1114 .cb_handler = vti6_rcv_cb,
1115 .err_handler = vti6_err,
1119 static struct xfrm6_protocol vti_ah6_protocol __read_mostly = {
1120 .handler = vti6_rcv,
1121 .cb_handler = vti6_rcv_cb,
1122 .err_handler = vti6_err,
1126 static struct xfrm6_protocol vti_ipcomp6_protocol __read_mostly = {
1127 .handler = vti6_rcv,
1128 .cb_handler = vti6_rcv_cb,
1129 .err_handler = vti6_err,
1133 static bool is_vti6_tunnel(const struct net_device *dev)
1135 return dev->netdev_ops == &vti6_netdev_ops;
1138 static int vti6_device_event(struct notifier_block *unused,
1139 unsigned long event, void *ptr)
1141 struct net_device *dev = netdev_notifier_info_to_dev(ptr);
1142 struct ip6_tnl *t = netdev_priv(dev);
1144 if (!is_vti6_tunnel(dev))
1149 if (!net_eq(t->net, dev_net(dev)))
1150 xfrm_garbage_collect(t->net);
1156 static struct notifier_block vti6_notifier_block __read_mostly = {
1157 .notifier_call = vti6_device_event,
1161 * vti6_tunnel_init - register protocol and reserve needed resources
1163 * Return: 0 on success
1165 static int __init vti6_tunnel_init(void)
1170 register_netdevice_notifier(&vti6_notifier_block);
1172 msg = "tunnel device";
1173 err = register_pernet_device(&vti6_net_ops);
1175 goto pernet_dev_failed;
1177 msg = "tunnel protocols";
1178 err = xfrm6_protocol_register(&vti_esp6_protocol, IPPROTO_ESP);
1180 goto xfrm_proto_esp_failed;
1181 err = xfrm6_protocol_register(&vti_ah6_protocol, IPPROTO_AH);
1183 goto xfrm_proto_ah_failed;
1184 err = xfrm6_protocol_register(&vti_ipcomp6_protocol, IPPROTO_COMP);
1186 goto xfrm_proto_comp_failed;
1188 msg = "netlink interface";
1189 err = rtnl_link_register(&vti6_link_ops);
1191 goto rtnl_link_failed;
1196 xfrm6_protocol_deregister(&vti_ipcomp6_protocol, IPPROTO_COMP);
1197 xfrm_proto_comp_failed:
1198 xfrm6_protocol_deregister(&vti_ah6_protocol, IPPROTO_AH);
1199 xfrm_proto_ah_failed:
1200 xfrm6_protocol_deregister(&vti_esp6_protocol, IPPROTO_ESP);
1201 xfrm_proto_esp_failed:
1202 unregister_pernet_device(&vti6_net_ops);
1204 unregister_netdevice_notifier(&vti6_notifier_block);
1205 pr_err("vti6 init: failed to register %s\n", msg);
1210 * vti6_tunnel_cleanup - free resources and unregister protocol
1212 static void __exit vti6_tunnel_cleanup(void)
1214 rtnl_link_unregister(&vti6_link_ops);
1215 xfrm6_protocol_deregister(&vti_ipcomp6_protocol, IPPROTO_COMP);
1216 xfrm6_protocol_deregister(&vti_ah6_protocol, IPPROTO_AH);
1217 xfrm6_protocol_deregister(&vti_esp6_protocol, IPPROTO_ESP);
1218 unregister_pernet_device(&vti6_net_ops);
1219 unregister_netdevice_notifier(&vti6_notifier_block);
1222 module_init(vti6_tunnel_init);
1223 module_exit(vti6_tunnel_cleanup);
1224 MODULE_LICENSE("GPL");
1225 MODULE_ALIAS_RTNL_LINK("vti6");
1226 MODULE_ALIAS_NETDEV("ip6_vti0");
1227 MODULE_AUTHOR("Steffen Klassert");
1228 MODULE_DESCRIPTION("IPv6 virtual tunnel interface");