]> git.proxmox.com Git - mirror_ubuntu-eoan-kernel.git/blob - net/ipv4/ip_vti.c
Merge tag 'selinux-pr-20190115' of git://git.kernel.org/pub/scm/linux/kernel/git...
[mirror_ubuntu-eoan-kernel.git] / net / ipv4 / ip_vti.c
1 /*
2 * Linux NET3: IP/IP protocol decoder modified to support
3 * virtual tunnel interface
4 *
5 * Authors:
6 * Saurabh Mohan (saurabh.mohan@vyatta.com) 05/07/2012
7 *
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.
12 *
13 */
14
15 /*
16 This version of net/ipv4/ip_vti.c is cloned of net/ipv4/ipip.c
17
18 For comments look at net/ipv4/ip_gre.c --ANK
19 */
20
21
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>
29 #include <linux/in.h>
30 #include <linux/tcp.h>
31 #include <linux/udp.h>
32 #include <linux/if_arp.h>
33 #include <linux/init.h>
34 #include <linux/netfilter_ipv4.h>
35 #include <linux/if_ether.h>
36 #include <linux/icmpv6.h>
37
38 #include <net/sock.h>
39 #include <net/ip.h>
40 #include <net/icmp.h>
41 #include <net/ip_tunnels.h>
42 #include <net/inet_ecn.h>
43 #include <net/xfrm.h>
44 #include <net/net_namespace.h>
45 #include <net/netns/generic.h>
46
47 static struct rtnl_link_ops vti_link_ops __read_mostly;
48
49 static unsigned int vti_net_id __read_mostly;
50 static int vti_tunnel_init(struct net_device *dev);
51
52 static int vti_input(struct sk_buff *skb, int nexthdr, __be32 spi,
53 int encap_type)
54 {
55 struct ip_tunnel *tunnel;
56 const struct iphdr *iph = ip_hdr(skb);
57 struct net *net = dev_net(skb->dev);
58 struct ip_tunnel_net *itn = net_generic(net, vti_net_id);
59
60 tunnel = ip_tunnel_lookup(itn, skb->dev->ifindex, TUNNEL_NO_KEY,
61 iph->saddr, iph->daddr, 0);
62 if (tunnel) {
63 if (!xfrm4_policy_check(NULL, XFRM_POLICY_IN, skb))
64 goto drop;
65
66 XFRM_TUNNEL_SKB_CB(skb)->tunnel.ip4 = tunnel;
67
68 return xfrm_input(skb, nexthdr, spi, encap_type);
69 }
70
71 return -EINVAL;
72 drop:
73 kfree_skb(skb);
74 return 0;
75 }
76
77 static int vti_rcv(struct sk_buff *skb)
78 {
79 XFRM_SPI_SKB_CB(skb)->family = AF_INET;
80 XFRM_SPI_SKB_CB(skb)->daddroff = offsetof(struct iphdr, daddr);
81
82 return vti_input(skb, ip_hdr(skb)->protocol, 0, 0);
83 }
84
85 static int vti_rcv_cb(struct sk_buff *skb, int err)
86 {
87 unsigned short family;
88 struct net_device *dev;
89 struct pcpu_sw_netstats *tstats;
90 struct xfrm_state *x;
91 struct xfrm_mode *inner_mode;
92 struct ip_tunnel *tunnel = XFRM_TUNNEL_SKB_CB(skb)->tunnel.ip4;
93 u32 orig_mark = skb->mark;
94 int ret;
95
96 if (!tunnel)
97 return 1;
98
99 dev = tunnel->dev;
100
101 if (err) {
102 dev->stats.rx_errors++;
103 dev->stats.rx_dropped++;
104
105 return 0;
106 }
107
108 x = xfrm_input_state(skb);
109
110 inner_mode = x->inner_mode;
111
112 if (x->sel.family == AF_UNSPEC) {
113 inner_mode = xfrm_ip2inner_mode(x, XFRM_MODE_SKB_CB(skb)->protocol);
114 if (inner_mode == NULL) {
115 XFRM_INC_STATS(dev_net(skb->dev),
116 LINUX_MIB_XFRMINSTATEMODEERROR);
117 return -EINVAL;
118 }
119 }
120
121 family = inner_mode->afinfo->family;
122
123 skb->mark = be32_to_cpu(tunnel->parms.i_key);
124 ret = xfrm_policy_check(NULL, XFRM_POLICY_IN, skb, family);
125 skb->mark = orig_mark;
126
127 if (!ret)
128 return -EPERM;
129
130 skb_scrub_packet(skb, !net_eq(tunnel->net, dev_net(skb->dev)));
131 skb->dev = dev;
132
133 tstats = this_cpu_ptr(dev->tstats);
134
135 u64_stats_update_begin(&tstats->syncp);
136 tstats->rx_packets++;
137 tstats->rx_bytes += skb->len;
138 u64_stats_update_end(&tstats->syncp);
139
140 return 0;
141 }
142
143 static bool vti_state_check(const struct xfrm_state *x, __be32 dst, __be32 src)
144 {
145 xfrm_address_t *daddr = (xfrm_address_t *)&dst;
146 xfrm_address_t *saddr = (xfrm_address_t *)&src;
147
148 /* if there is no transform then this tunnel is not functional.
149 * Or if the xfrm is not mode tunnel.
150 */
151 if (!x || x->props.mode != XFRM_MODE_TUNNEL ||
152 x->props.family != AF_INET)
153 return false;
154
155 if (!dst)
156 return xfrm_addr_equal(saddr, &x->props.saddr, AF_INET);
157
158 if (!xfrm_state_addr_check(x, daddr, saddr, AF_INET))
159 return false;
160
161 return true;
162 }
163
164 static netdev_tx_t vti_xmit(struct sk_buff *skb, struct net_device *dev,
165 struct flowi *fl)
166 {
167 struct ip_tunnel *tunnel = netdev_priv(dev);
168 struct ip_tunnel_parm *parms = &tunnel->parms;
169 struct dst_entry *dst = skb_dst(skb);
170 struct net_device *tdev; /* Device to other host */
171 int pkt_len = skb->len;
172 int err;
173 int mtu;
174
175 if (!dst) {
176 dev->stats.tx_carrier_errors++;
177 goto tx_error_icmp;
178 }
179
180 dst_hold(dst);
181 dst = xfrm_lookup(tunnel->net, dst, fl, NULL, 0);
182 if (IS_ERR(dst)) {
183 dev->stats.tx_carrier_errors++;
184 goto tx_error_icmp;
185 }
186
187 if (!vti_state_check(dst->xfrm, parms->iph.daddr, parms->iph.saddr)) {
188 dev->stats.tx_carrier_errors++;
189 dst_release(dst);
190 goto tx_error_icmp;
191 }
192
193 tdev = dst->dev;
194
195 if (tdev == dev) {
196 dst_release(dst);
197 dev->stats.collisions++;
198 goto tx_error;
199 }
200
201 mtu = dst_mtu(dst);
202 if (skb->len > mtu) {
203 skb_dst_update_pmtu(skb, mtu);
204 if (skb->protocol == htons(ETH_P_IP)) {
205 icmp_send(skb, ICMP_DEST_UNREACH, ICMP_FRAG_NEEDED,
206 htonl(mtu));
207 } else {
208 if (mtu < IPV6_MIN_MTU)
209 mtu = IPV6_MIN_MTU;
210
211 icmpv6_send(skb, ICMPV6_PKT_TOOBIG, 0, mtu);
212 }
213
214 dst_release(dst);
215 goto tx_error;
216 }
217
218 skb_scrub_packet(skb, !net_eq(tunnel->net, dev_net(dev)));
219 skb_dst_set(skb, dst);
220 skb->dev = skb_dst(skb)->dev;
221
222 err = dst_output(tunnel->net, skb->sk, skb);
223 if (net_xmit_eval(err) == 0)
224 err = pkt_len;
225 iptunnel_xmit_stats(dev, err);
226 return NETDEV_TX_OK;
227
228 tx_error_icmp:
229 dst_link_failure(skb);
230 tx_error:
231 dev->stats.tx_errors++;
232 kfree_skb(skb);
233 return NETDEV_TX_OK;
234 }
235
236 /* This function assumes it is being called from dev_queue_xmit()
237 * and that skb is filled properly by that function.
238 */
239 static netdev_tx_t vti_tunnel_xmit(struct sk_buff *skb, struct net_device *dev)
240 {
241 struct ip_tunnel *tunnel = netdev_priv(dev);
242 struct flowi fl;
243
244 if (!pskb_inet_may_pull(skb))
245 goto tx_err;
246
247 memset(&fl, 0, sizeof(fl));
248
249 switch (skb->protocol) {
250 case htons(ETH_P_IP):
251 xfrm_decode_session(skb, &fl, AF_INET);
252 memset(IPCB(skb), 0, sizeof(*IPCB(skb)));
253 break;
254 case htons(ETH_P_IPV6):
255 xfrm_decode_session(skb, &fl, AF_INET6);
256 memset(IP6CB(skb), 0, sizeof(*IP6CB(skb)));
257 break;
258 default:
259 goto tx_err;
260 }
261
262 /* override mark with tunnel output key */
263 fl.flowi_mark = be32_to_cpu(tunnel->parms.o_key);
264
265 return vti_xmit(skb, dev, &fl);
266
267 tx_err:
268 dev->stats.tx_errors++;
269 kfree_skb(skb);
270 return NETDEV_TX_OK;
271 }
272
273 static int vti4_err(struct sk_buff *skb, u32 info)
274 {
275 __be32 spi;
276 __u32 mark;
277 struct xfrm_state *x;
278 struct ip_tunnel *tunnel;
279 struct ip_esp_hdr *esph;
280 struct ip_auth_hdr *ah ;
281 struct ip_comp_hdr *ipch;
282 struct net *net = dev_net(skb->dev);
283 const struct iphdr *iph = (const struct iphdr *)skb->data;
284 int protocol = iph->protocol;
285 struct ip_tunnel_net *itn = net_generic(net, vti_net_id);
286
287 tunnel = ip_tunnel_lookup(itn, skb->dev->ifindex, TUNNEL_NO_KEY,
288 iph->daddr, iph->saddr, 0);
289 if (!tunnel)
290 return -1;
291
292 mark = be32_to_cpu(tunnel->parms.o_key);
293
294 switch (protocol) {
295 case IPPROTO_ESP:
296 esph = (struct ip_esp_hdr *)(skb->data+(iph->ihl<<2));
297 spi = esph->spi;
298 break;
299 case IPPROTO_AH:
300 ah = (struct ip_auth_hdr *)(skb->data+(iph->ihl<<2));
301 spi = ah->spi;
302 break;
303 case IPPROTO_COMP:
304 ipch = (struct ip_comp_hdr *)(skb->data+(iph->ihl<<2));
305 spi = htonl(ntohs(ipch->cpi));
306 break;
307 default:
308 return 0;
309 }
310
311 switch (icmp_hdr(skb)->type) {
312 case ICMP_DEST_UNREACH:
313 if (icmp_hdr(skb)->code != ICMP_FRAG_NEEDED)
314 return 0;
315 case ICMP_REDIRECT:
316 break;
317 default:
318 return 0;
319 }
320
321 x = xfrm_state_lookup(net, mark, (const xfrm_address_t *)&iph->daddr,
322 spi, protocol, AF_INET);
323 if (!x)
324 return 0;
325
326 if (icmp_hdr(skb)->type == ICMP_DEST_UNREACH)
327 ipv4_update_pmtu(skb, net, info, 0, protocol);
328 else
329 ipv4_redirect(skb, net, 0, protocol);
330 xfrm_state_put(x);
331
332 return 0;
333 }
334
335 static int
336 vti_tunnel_ioctl(struct net_device *dev, struct ifreq *ifr, int cmd)
337 {
338 int err = 0;
339 struct ip_tunnel_parm p;
340
341 if (copy_from_user(&p, ifr->ifr_ifru.ifru_data, sizeof(p)))
342 return -EFAULT;
343
344 if (cmd == SIOCADDTUNNEL || cmd == SIOCCHGTUNNEL) {
345 if (p.iph.version != 4 || p.iph.protocol != IPPROTO_IPIP ||
346 p.iph.ihl != 5)
347 return -EINVAL;
348 }
349
350 if (!(p.i_flags & GRE_KEY))
351 p.i_key = 0;
352 if (!(p.o_flags & GRE_KEY))
353 p.o_key = 0;
354
355 p.i_flags = VTI_ISVTI;
356
357 err = ip_tunnel_ioctl(dev, &p, cmd);
358 if (err)
359 return err;
360
361 if (cmd != SIOCDELTUNNEL) {
362 p.i_flags |= GRE_KEY;
363 p.o_flags |= GRE_KEY;
364 }
365
366 if (copy_to_user(ifr->ifr_ifru.ifru_data, &p, sizeof(p)))
367 return -EFAULT;
368 return 0;
369 }
370
371 static const struct net_device_ops vti_netdev_ops = {
372 .ndo_init = vti_tunnel_init,
373 .ndo_uninit = ip_tunnel_uninit,
374 .ndo_start_xmit = vti_tunnel_xmit,
375 .ndo_do_ioctl = vti_tunnel_ioctl,
376 .ndo_change_mtu = ip_tunnel_change_mtu,
377 .ndo_get_stats64 = ip_tunnel_get_stats64,
378 .ndo_get_iflink = ip_tunnel_get_iflink,
379 };
380
381 static void vti_tunnel_setup(struct net_device *dev)
382 {
383 dev->netdev_ops = &vti_netdev_ops;
384 dev->type = ARPHRD_TUNNEL;
385 ip_tunnel_setup(dev, vti_net_id);
386 }
387
388 static int vti_tunnel_init(struct net_device *dev)
389 {
390 struct ip_tunnel *tunnel = netdev_priv(dev);
391 struct iphdr *iph = &tunnel->parms.iph;
392
393 memcpy(dev->dev_addr, &iph->saddr, 4);
394 memcpy(dev->broadcast, &iph->daddr, 4);
395
396 dev->flags = IFF_NOARP;
397 dev->addr_len = 4;
398 dev->features |= NETIF_F_LLTX;
399 netif_keep_dst(dev);
400
401 return ip_tunnel_init(dev);
402 }
403
404 static void __net_init vti_fb_tunnel_init(struct net_device *dev)
405 {
406 struct ip_tunnel *tunnel = netdev_priv(dev);
407 struct iphdr *iph = &tunnel->parms.iph;
408
409 iph->version = 4;
410 iph->protocol = IPPROTO_IPIP;
411 iph->ihl = 5;
412 }
413
414 static struct xfrm4_protocol vti_esp4_protocol __read_mostly = {
415 .handler = vti_rcv,
416 .input_handler = vti_input,
417 .cb_handler = vti_rcv_cb,
418 .err_handler = vti4_err,
419 .priority = 100,
420 };
421
422 static struct xfrm4_protocol vti_ah4_protocol __read_mostly = {
423 .handler = vti_rcv,
424 .input_handler = vti_input,
425 .cb_handler = vti_rcv_cb,
426 .err_handler = vti4_err,
427 .priority = 100,
428 };
429
430 static struct xfrm4_protocol vti_ipcomp4_protocol __read_mostly = {
431 .handler = vti_rcv,
432 .input_handler = vti_input,
433 .cb_handler = vti_rcv_cb,
434 .err_handler = vti4_err,
435 .priority = 100,
436 };
437
438 static int __net_init vti_init_net(struct net *net)
439 {
440 int err;
441 struct ip_tunnel_net *itn;
442
443 err = ip_tunnel_init_net(net, vti_net_id, &vti_link_ops, "ip_vti0");
444 if (err)
445 return err;
446 itn = net_generic(net, vti_net_id);
447 if (itn->fb_tunnel_dev)
448 vti_fb_tunnel_init(itn->fb_tunnel_dev);
449 return 0;
450 }
451
452 static void __net_exit vti_exit_batch_net(struct list_head *list_net)
453 {
454 ip_tunnel_delete_nets(list_net, vti_net_id, &vti_link_ops);
455 }
456
457 static struct pernet_operations vti_net_ops = {
458 .init = vti_init_net,
459 .exit_batch = vti_exit_batch_net,
460 .id = &vti_net_id,
461 .size = sizeof(struct ip_tunnel_net),
462 };
463
464 static int vti_tunnel_validate(struct nlattr *tb[], struct nlattr *data[],
465 struct netlink_ext_ack *extack)
466 {
467 return 0;
468 }
469
470 static void vti_netlink_parms(struct nlattr *data[],
471 struct ip_tunnel_parm *parms,
472 __u32 *fwmark)
473 {
474 memset(parms, 0, sizeof(*parms));
475
476 parms->iph.protocol = IPPROTO_IPIP;
477
478 if (!data)
479 return;
480
481 parms->i_flags = VTI_ISVTI;
482
483 if (data[IFLA_VTI_LINK])
484 parms->link = nla_get_u32(data[IFLA_VTI_LINK]);
485
486 if (data[IFLA_VTI_IKEY])
487 parms->i_key = nla_get_be32(data[IFLA_VTI_IKEY]);
488
489 if (data[IFLA_VTI_OKEY])
490 parms->o_key = nla_get_be32(data[IFLA_VTI_OKEY]);
491
492 if (data[IFLA_VTI_LOCAL])
493 parms->iph.saddr = nla_get_in_addr(data[IFLA_VTI_LOCAL]);
494
495 if (data[IFLA_VTI_REMOTE])
496 parms->iph.daddr = nla_get_in_addr(data[IFLA_VTI_REMOTE]);
497
498 if (data[IFLA_VTI_FWMARK])
499 *fwmark = nla_get_u32(data[IFLA_VTI_FWMARK]);
500 }
501
502 static int vti_newlink(struct net *src_net, struct net_device *dev,
503 struct nlattr *tb[], struct nlattr *data[],
504 struct netlink_ext_ack *extack)
505 {
506 struct ip_tunnel_parm parms;
507 __u32 fwmark = 0;
508
509 vti_netlink_parms(data, &parms, &fwmark);
510 return ip_tunnel_newlink(dev, tb, &parms, fwmark);
511 }
512
513 static int vti_changelink(struct net_device *dev, struct nlattr *tb[],
514 struct nlattr *data[],
515 struct netlink_ext_ack *extack)
516 {
517 struct ip_tunnel *t = netdev_priv(dev);
518 __u32 fwmark = t->fwmark;
519 struct ip_tunnel_parm p;
520
521 vti_netlink_parms(data, &p, &fwmark);
522 return ip_tunnel_changelink(dev, tb, &p, fwmark);
523 }
524
525 static size_t vti_get_size(const struct net_device *dev)
526 {
527 return
528 /* IFLA_VTI_LINK */
529 nla_total_size(4) +
530 /* IFLA_VTI_IKEY */
531 nla_total_size(4) +
532 /* IFLA_VTI_OKEY */
533 nla_total_size(4) +
534 /* IFLA_VTI_LOCAL */
535 nla_total_size(4) +
536 /* IFLA_VTI_REMOTE */
537 nla_total_size(4) +
538 /* IFLA_VTI_FWMARK */
539 nla_total_size(4) +
540 0;
541 }
542
543 static int vti_fill_info(struct sk_buff *skb, const struct net_device *dev)
544 {
545 struct ip_tunnel *t = netdev_priv(dev);
546 struct ip_tunnel_parm *p = &t->parms;
547
548 if (nla_put_u32(skb, IFLA_VTI_LINK, p->link) ||
549 nla_put_be32(skb, IFLA_VTI_IKEY, p->i_key) ||
550 nla_put_be32(skb, IFLA_VTI_OKEY, p->o_key) ||
551 nla_put_in_addr(skb, IFLA_VTI_LOCAL, p->iph.saddr) ||
552 nla_put_in_addr(skb, IFLA_VTI_REMOTE, p->iph.daddr) ||
553 nla_put_u32(skb, IFLA_VTI_FWMARK, t->fwmark))
554 return -EMSGSIZE;
555
556 return 0;
557 }
558
559 static const struct nla_policy vti_policy[IFLA_VTI_MAX + 1] = {
560 [IFLA_VTI_LINK] = { .type = NLA_U32 },
561 [IFLA_VTI_IKEY] = { .type = NLA_U32 },
562 [IFLA_VTI_OKEY] = { .type = NLA_U32 },
563 [IFLA_VTI_LOCAL] = { .len = FIELD_SIZEOF(struct iphdr, saddr) },
564 [IFLA_VTI_REMOTE] = { .len = FIELD_SIZEOF(struct iphdr, daddr) },
565 [IFLA_VTI_FWMARK] = { .type = NLA_U32 },
566 };
567
568 static struct rtnl_link_ops vti_link_ops __read_mostly = {
569 .kind = "vti",
570 .maxtype = IFLA_VTI_MAX,
571 .policy = vti_policy,
572 .priv_size = sizeof(struct ip_tunnel),
573 .setup = vti_tunnel_setup,
574 .validate = vti_tunnel_validate,
575 .newlink = vti_newlink,
576 .changelink = vti_changelink,
577 .dellink = ip_tunnel_dellink,
578 .get_size = vti_get_size,
579 .fill_info = vti_fill_info,
580 .get_link_net = ip_tunnel_get_link_net,
581 };
582
583 static int __init vti_init(void)
584 {
585 const char *msg;
586 int err;
587
588 pr_info("IPv4 over IPsec tunneling driver\n");
589
590 msg = "tunnel device";
591 err = register_pernet_device(&vti_net_ops);
592 if (err < 0)
593 goto pernet_dev_failed;
594
595 msg = "tunnel protocols";
596 err = xfrm4_protocol_register(&vti_esp4_protocol, IPPROTO_ESP);
597 if (err < 0)
598 goto xfrm_proto_esp_failed;
599 err = xfrm4_protocol_register(&vti_ah4_protocol, IPPROTO_AH);
600 if (err < 0)
601 goto xfrm_proto_ah_failed;
602 err = xfrm4_protocol_register(&vti_ipcomp4_protocol, IPPROTO_COMP);
603 if (err < 0)
604 goto xfrm_proto_comp_failed;
605
606 msg = "netlink interface";
607 err = rtnl_link_register(&vti_link_ops);
608 if (err < 0)
609 goto rtnl_link_failed;
610
611 return err;
612
613 rtnl_link_failed:
614 xfrm4_protocol_deregister(&vti_ipcomp4_protocol, IPPROTO_COMP);
615 xfrm_proto_comp_failed:
616 xfrm4_protocol_deregister(&vti_ah4_protocol, IPPROTO_AH);
617 xfrm_proto_ah_failed:
618 xfrm4_protocol_deregister(&vti_esp4_protocol, IPPROTO_ESP);
619 xfrm_proto_esp_failed:
620 unregister_pernet_device(&vti_net_ops);
621 pernet_dev_failed:
622 pr_err("vti init: failed to register %s\n", msg);
623 return err;
624 }
625
626 static void __exit vti_fini(void)
627 {
628 rtnl_link_unregister(&vti_link_ops);
629 xfrm4_protocol_deregister(&vti_ipcomp4_protocol, IPPROTO_COMP);
630 xfrm4_protocol_deregister(&vti_ah4_protocol, IPPROTO_AH);
631 xfrm4_protocol_deregister(&vti_esp4_protocol, IPPROTO_ESP);
632 unregister_pernet_device(&vti_net_ops);
633 }
634
635 module_init(vti_init);
636 module_exit(vti_fini);
637 MODULE_LICENSE("GPL");
638 MODULE_ALIAS_RTNL_LINK("vti");
639 MODULE_ALIAS_NETDEV("ip_vti0");