2 * Linux NET3: IP/IP protocol decoder modified to support
3 * virtual tunnel interface
6 * Saurabh Mohan (saurabh.mohan@vyatta.com) 05/07/2012
8 * This program is free software; you can redistribute it and/or
9 * modify it under the terms of the GNU General Public License
10 * as published by the Free Software Foundation; either version
11 * 2 of the License, or (at your option) any later version.
16 This version of net/ipv4/ip_vti.c is cloned of net/ipv4/ipip.c
18 For comments look at net/ipv4/ip_gre.c --ANK
22 #include <linux/capability.h>
23 #include <linux/module.h>
24 #include <linux/types.h>
25 #include <linux/kernel.h>
26 #include <linux/uaccess.h>
27 #include <linux/skbuff.h>
28 #include <linux/netdevice.h>
30 #include <linux/tcp.h>
31 #include <linux/udp.h>
32 #include <linux/if_arp.h>
33 #include <linux/mroute.h>
34 #include <linux/init.h>
35 #include <linux/netfilter_ipv4.h>
36 #include <linux/if_ether.h>
37 #include <linux/icmpv6.h>
42 #include <net/ip_tunnels.h>
43 #include <net/inet_ecn.h>
45 #include <net/net_namespace.h>
46 #include <net/netns/generic.h>
48 static struct rtnl_link_ops vti_link_ops __read_mostly;
50 static int vti_net_id __read_mostly;
51 static int vti_tunnel_init(struct net_device *dev);
53 static int vti_input(struct sk_buff *skb, int nexthdr, __be32 spi,
56 struct ip_tunnel *tunnel;
57 const struct iphdr *iph = ip_hdr(skb);
58 struct net *net = dev_net(skb->dev);
59 struct ip_tunnel_net *itn = net_generic(net, vti_net_id);
61 tunnel = ip_tunnel_lookup(itn, skb->dev->ifindex, TUNNEL_NO_KEY,
62 iph->saddr, iph->daddr, 0);
64 if (!xfrm4_policy_check(NULL, XFRM_POLICY_IN, skb))
67 XFRM_TUNNEL_SKB_CB(skb)->tunnel.ip4 = tunnel;
68 skb->mark = be32_to_cpu(tunnel->parms.i_key);
70 return xfrm_input(skb, nexthdr, spi, encap_type);
79 static int vti_rcv(struct sk_buff *skb)
81 XFRM_SPI_SKB_CB(skb)->family = AF_INET;
82 XFRM_SPI_SKB_CB(skb)->daddroff = offsetof(struct iphdr, daddr);
84 return vti_input(skb, ip_hdr(skb)->protocol, 0, 0);
87 static int vti_rcv_cb(struct sk_buff *skb, int err)
89 unsigned short family;
90 struct net_device *dev;
91 struct pcpu_sw_netstats *tstats;
93 struct ip_tunnel *tunnel = XFRM_TUNNEL_SKB_CB(skb)->tunnel.ip4;
101 dev->stats.rx_errors++;
102 dev->stats.rx_dropped++;
107 x = xfrm_input_state(skb);
108 family = x->inner_mode->afinfo->family;
110 if (!xfrm_policy_check(NULL, XFRM_POLICY_IN, skb, family))
113 skb_scrub_packet(skb, !net_eq(tunnel->net, dev_net(skb->dev)));
116 tstats = this_cpu_ptr(dev->tstats);
118 u64_stats_update_begin(&tstats->syncp);
119 tstats->rx_packets++;
120 tstats->rx_bytes += skb->len;
121 u64_stats_update_end(&tstats->syncp);
126 static bool vti_state_check(const struct xfrm_state *x, __be32 dst, __be32 src)
128 xfrm_address_t *daddr = (xfrm_address_t *)&dst;
129 xfrm_address_t *saddr = (xfrm_address_t *)&src;
131 /* if there is no transform then this tunnel is not functional.
132 * Or if the xfrm is not mode tunnel.
134 if (!x || x->props.mode != XFRM_MODE_TUNNEL ||
135 x->props.family != AF_INET)
139 return xfrm_addr_equal(saddr, &x->props.saddr, AF_INET);
141 if (!xfrm_state_addr_check(x, daddr, saddr, AF_INET))
147 static netdev_tx_t vti_xmit(struct sk_buff *skb, struct net_device *dev,
150 struct ip_tunnel *tunnel = netdev_priv(dev);
151 struct ip_tunnel_parm *parms = &tunnel->parms;
152 struct dst_entry *dst = skb_dst(skb);
153 struct net_device *tdev; /* Device to other host */
157 dev->stats.tx_carrier_errors++;
162 dst = xfrm_lookup(tunnel->net, dst, fl, NULL, 0);
164 dev->stats.tx_carrier_errors++;
168 if (!vti_state_check(dst->xfrm, parms->iph.daddr, parms->iph.saddr)) {
169 dev->stats.tx_carrier_errors++;
178 dev->stats.collisions++;
182 if (tunnel->err_count > 0) {
183 if (time_before(jiffies,
184 tunnel->err_time + IPTUNNEL_ERR_TIMEO)) {
186 dst_link_failure(skb);
188 tunnel->err_count = 0;
191 skb_scrub_packet(skb, !net_eq(tunnel->net, dev_net(dev)));
192 skb_dst_set(skb, dst);
193 skb->dev = skb_dst(skb)->dev;
195 err = dst_output(skb);
196 if (net_xmit_eval(err) == 0)
198 iptunnel_xmit_stats(err, &dev->stats, dev->tstats);
202 dst_link_failure(skb);
204 dev->stats.tx_errors++;
209 /* This function assumes it is being called from dev_queue_xmit()
210 * and that skb is filled properly by that function.
212 static netdev_tx_t vti_tunnel_xmit(struct sk_buff *skb, struct net_device *dev)
214 struct ip_tunnel *tunnel = netdev_priv(dev);
217 memset(&fl, 0, sizeof(fl));
219 skb->mark = be32_to_cpu(tunnel->parms.o_key);
221 switch (skb->protocol) {
222 case htons(ETH_P_IP):
223 xfrm_decode_session(skb, &fl, AF_INET);
224 memset(IPCB(skb), 0, sizeof(*IPCB(skb)));
226 case htons(ETH_P_IPV6):
227 xfrm_decode_session(skb, &fl, AF_INET6);
228 memset(IP6CB(skb), 0, sizeof(*IP6CB(skb)));
231 dev->stats.tx_errors++;
236 return vti_xmit(skb, dev, &fl);
239 static int vti4_err(struct sk_buff *skb, u32 info)
243 struct xfrm_state *x;
244 struct ip_tunnel *tunnel;
245 struct ip_esp_hdr *esph;
246 struct ip_auth_hdr *ah ;
247 struct ip_comp_hdr *ipch;
248 struct net *net = dev_net(skb->dev);
249 const struct iphdr *iph = (const struct iphdr *)skb->data;
250 int protocol = iph->protocol;
251 struct ip_tunnel_net *itn = net_generic(net, vti_net_id);
253 tunnel = ip_tunnel_lookup(itn, skb->dev->ifindex, TUNNEL_NO_KEY,
254 iph->daddr, iph->saddr, 0);
258 mark = be32_to_cpu(tunnel->parms.o_key);
262 esph = (struct ip_esp_hdr *)(skb->data+(iph->ihl<<2));
266 ah = (struct ip_auth_hdr *)(skb->data+(iph->ihl<<2));
270 ipch = (struct ip_comp_hdr *)(skb->data+(iph->ihl<<2));
271 spi = htonl(ntohs(ipch->cpi));
277 switch (icmp_hdr(skb)->type) {
278 case ICMP_DEST_UNREACH:
279 if (icmp_hdr(skb)->code != ICMP_FRAG_NEEDED)
287 x = xfrm_state_lookup(net, mark, (const xfrm_address_t *)&iph->daddr,
288 spi, protocol, AF_INET);
292 if (icmp_hdr(skb)->type == ICMP_DEST_UNREACH)
293 ipv4_update_pmtu(skb, net, info, 0, 0, protocol, 0);
295 ipv4_redirect(skb, net, 0, 0, protocol, 0);
302 vti_tunnel_ioctl(struct net_device *dev, struct ifreq *ifr, int cmd)
305 struct ip_tunnel_parm p;
307 if (copy_from_user(&p, ifr->ifr_ifru.ifru_data, sizeof(p)))
310 if (cmd == SIOCADDTUNNEL || cmd == SIOCCHGTUNNEL) {
311 if (p.iph.version != 4 || p.iph.protocol != IPPROTO_IPIP ||
316 p.i_flags |= VTI_ISVTI;
317 err = ip_tunnel_ioctl(dev, &p, cmd);
321 if (cmd != SIOCDELTUNNEL) {
322 p.i_flags |= GRE_KEY;
323 p.o_flags |= GRE_KEY;
326 if (copy_to_user(ifr->ifr_ifru.ifru_data, &p, sizeof(p)))
331 static const struct net_device_ops vti_netdev_ops = {
332 .ndo_init = vti_tunnel_init,
333 .ndo_uninit = ip_tunnel_uninit,
334 .ndo_start_xmit = vti_tunnel_xmit,
335 .ndo_do_ioctl = vti_tunnel_ioctl,
336 .ndo_change_mtu = ip_tunnel_change_mtu,
337 .ndo_get_stats64 = ip_tunnel_get_stats64,
340 static void vti_tunnel_setup(struct net_device *dev)
342 dev->netdev_ops = &vti_netdev_ops;
343 dev->type = ARPHRD_TUNNEL;
344 ip_tunnel_setup(dev, vti_net_id);
347 static int vti_tunnel_init(struct net_device *dev)
349 struct ip_tunnel *tunnel = netdev_priv(dev);
350 struct iphdr *iph = &tunnel->parms.iph;
352 memcpy(dev->dev_addr, &iph->saddr, 4);
353 memcpy(dev->broadcast, &iph->daddr, 4);
355 dev->hard_header_len = LL_MAX_HEADER + sizeof(struct iphdr);
356 dev->mtu = ETH_DATA_LEN;
357 dev->flags = IFF_NOARP;
360 dev->features |= NETIF_F_LLTX;
361 dev->priv_flags &= ~IFF_XMIT_DST_RELEASE;
363 return ip_tunnel_init(dev);
366 static void __net_init vti_fb_tunnel_init(struct net_device *dev)
368 struct ip_tunnel *tunnel = netdev_priv(dev);
369 struct iphdr *iph = &tunnel->parms.iph;
372 iph->protocol = IPPROTO_IPIP;
376 static struct xfrm4_protocol vti_esp4_protocol __read_mostly = {
378 .input_handler = vti_input,
379 .cb_handler = vti_rcv_cb,
380 .err_handler = vti4_err,
384 static struct xfrm4_protocol vti_ah4_protocol __read_mostly = {
386 .input_handler = vti_input,
387 .cb_handler = vti_rcv_cb,
388 .err_handler = vti4_err,
392 static struct xfrm4_protocol vti_ipcomp4_protocol __read_mostly = {
394 .input_handler = vti_input,
395 .cb_handler = vti_rcv_cb,
396 .err_handler = vti4_err,
400 static int __net_init vti_init_net(struct net *net)
403 struct ip_tunnel_net *itn;
405 err = ip_tunnel_init_net(net, vti_net_id, &vti_link_ops, "ip_vti0");
408 itn = net_generic(net, vti_net_id);
409 vti_fb_tunnel_init(itn->fb_tunnel_dev);
413 static void __net_exit vti_exit_net(struct net *net)
415 struct ip_tunnel_net *itn = net_generic(net, vti_net_id);
416 ip_tunnel_delete_net(itn, &vti_link_ops);
419 static struct pernet_operations vti_net_ops = {
420 .init = vti_init_net,
421 .exit = vti_exit_net,
423 .size = sizeof(struct ip_tunnel_net),
426 static int vti_tunnel_validate(struct nlattr *tb[], struct nlattr *data[])
431 static void vti_netlink_parms(struct nlattr *data[],
432 struct ip_tunnel_parm *parms)
434 memset(parms, 0, sizeof(*parms));
436 parms->iph.protocol = IPPROTO_IPIP;
441 parms->i_flags = VTI_ISVTI;
443 if (data[IFLA_VTI_LINK])
444 parms->link = nla_get_u32(data[IFLA_VTI_LINK]);
446 if (data[IFLA_VTI_IKEY])
447 parms->i_key = nla_get_be32(data[IFLA_VTI_IKEY]);
449 if (data[IFLA_VTI_OKEY])
450 parms->o_key = nla_get_be32(data[IFLA_VTI_OKEY]);
452 if (data[IFLA_VTI_LOCAL])
453 parms->iph.saddr = nla_get_be32(data[IFLA_VTI_LOCAL]);
455 if (data[IFLA_VTI_REMOTE])
456 parms->iph.daddr = nla_get_be32(data[IFLA_VTI_REMOTE]);
460 static int vti_newlink(struct net *src_net, struct net_device *dev,
461 struct nlattr *tb[], struct nlattr *data[])
463 struct ip_tunnel_parm parms;
465 vti_netlink_parms(data, &parms);
466 return ip_tunnel_newlink(dev, tb, &parms);
469 static int vti_changelink(struct net_device *dev, struct nlattr *tb[],
470 struct nlattr *data[])
472 struct ip_tunnel_parm p;
474 vti_netlink_parms(data, &p);
475 return ip_tunnel_changelink(dev, tb, &p);
478 static size_t vti_get_size(const struct net_device *dev)
489 /* IFLA_VTI_REMOTE */
494 static int vti_fill_info(struct sk_buff *skb, const struct net_device *dev)
496 struct ip_tunnel *t = netdev_priv(dev);
497 struct ip_tunnel_parm *p = &t->parms;
499 nla_put_u32(skb, IFLA_VTI_LINK, p->link);
500 nla_put_be32(skb, IFLA_VTI_IKEY, p->i_key);
501 nla_put_be32(skb, IFLA_VTI_OKEY, p->o_key);
502 nla_put_be32(skb, IFLA_VTI_LOCAL, p->iph.saddr);
503 nla_put_be32(skb, IFLA_VTI_REMOTE, p->iph.daddr);
508 static const struct nla_policy vti_policy[IFLA_VTI_MAX + 1] = {
509 [IFLA_VTI_LINK] = { .type = NLA_U32 },
510 [IFLA_VTI_IKEY] = { .type = NLA_U32 },
511 [IFLA_VTI_OKEY] = { .type = NLA_U32 },
512 [IFLA_VTI_LOCAL] = { .len = FIELD_SIZEOF(struct iphdr, saddr) },
513 [IFLA_VTI_REMOTE] = { .len = FIELD_SIZEOF(struct iphdr, daddr) },
516 static struct rtnl_link_ops vti_link_ops __read_mostly = {
518 .maxtype = IFLA_VTI_MAX,
519 .policy = vti_policy,
520 .priv_size = sizeof(struct ip_tunnel),
521 .setup = vti_tunnel_setup,
522 .validate = vti_tunnel_validate,
523 .newlink = vti_newlink,
524 .changelink = vti_changelink,
525 .get_size = vti_get_size,
526 .fill_info = vti_fill_info,
529 static int __init vti_init(void)
533 pr_info("IPv4 over IPSec tunneling driver\n");
535 err = register_pernet_device(&vti_net_ops);
538 err = xfrm4_protocol_register(&vti_esp4_protocol, IPPROTO_ESP);
540 unregister_pernet_device(&vti_net_ops);
541 pr_info("vti init: can't register tunnel\n");
546 err = xfrm4_protocol_register(&vti_ah4_protocol, IPPROTO_AH);
548 xfrm4_protocol_deregister(&vti_esp4_protocol, IPPROTO_ESP);
549 unregister_pernet_device(&vti_net_ops);
550 pr_info("vti init: can't register tunnel\n");
555 err = xfrm4_protocol_register(&vti_ipcomp4_protocol, IPPROTO_COMP);
557 xfrm4_protocol_deregister(&vti_ah4_protocol, IPPROTO_AH);
558 xfrm4_protocol_deregister(&vti_esp4_protocol, IPPROTO_ESP);
559 unregister_pernet_device(&vti_net_ops);
560 pr_info("vti init: can't register tunnel\n");
565 err = rtnl_link_register(&vti_link_ops);
567 goto rtnl_link_failed;
572 xfrm4_protocol_deregister(&vti_ipcomp4_protocol, IPPROTO_COMP);
573 xfrm4_protocol_deregister(&vti_ah4_protocol, IPPROTO_AH);
574 xfrm4_protocol_deregister(&vti_esp4_protocol, IPPROTO_ESP);
575 unregister_pernet_device(&vti_net_ops);
579 static void __exit vti_fini(void)
581 rtnl_link_unregister(&vti_link_ops);
582 if (xfrm4_protocol_deregister(&vti_ipcomp4_protocol, IPPROTO_COMP))
583 pr_info("vti close: can't deregister tunnel\n");
584 if (xfrm4_protocol_deregister(&vti_ah4_protocol, IPPROTO_AH))
585 pr_info("vti close: can't deregister tunnel\n");
586 if (xfrm4_protocol_deregister(&vti_esp4_protocol, IPPROTO_ESP))
587 pr_info("vti close: can't deregister tunnel\n");
590 unregister_pernet_device(&vti_net_ops);
593 module_init(vti_init);
594 module_exit(vti_fini);
595 MODULE_LICENSE("GPL");
596 MODULE_ALIAS_RTNL_LINK("vti");
597 MODULE_ALIAS_NETDEV("ip_vti0");