~ [ source navigation ] ~ [ diff markup ] ~ [ identifier search ] ~

TOMOYO Linux Cross Reference
Linux/net/ipv6/ip6_vti.c

Version: ~ [ linux-5.15-rc1 ] ~ [ linux-5.14.5 ] ~ [ linux-5.13.18 ] ~ [ linux-5.12.19 ] ~ [ linux-5.11.22 ] ~ [ linux-5.10.66 ] ~ [ linux-5.9.16 ] ~ [ linux-5.8.18 ] ~ [ linux-5.7.19 ] ~ [ linux-5.6.19 ] ~ [ linux-5.5.19 ] ~ [ linux-5.4.147 ] ~ [ linux-5.3.18 ] ~ [ linux-5.2.21 ] ~ [ linux-5.1.21 ] ~ [ linux-5.0.21 ] ~ [ linux-4.20.17 ] ~ [ linux-4.19.206 ] ~ [ linux-4.18.20 ] ~ [ linux-4.17.19 ] ~ [ linux-4.16.18 ] ~ [ linux-4.15.18 ] ~ [ linux-4.14.246 ] ~ [ linux-4.13.16 ] ~ [ linux-4.12.14 ] ~ [ linux-4.11.12 ] ~ [ linux-4.10.17 ] ~ [ linux-4.9.282 ] ~ [ linux-4.8.17 ] ~ [ linux-4.7.10 ] ~ [ linux-4.6.7 ] ~ [ linux-4.5.7 ] ~ [ linux-4.4.283 ] ~ [ linux-4.3.6 ] ~ [ linux-4.2.8 ] ~ [ linux-4.1.52 ] ~ [ linux-4.0.9 ] ~ [ linux-3.18.140 ] ~ [ linux-3.16.85 ] ~ [ linux-3.14.79 ] ~ [ linux-3.12.74 ] ~ [ linux-3.10.108 ] ~ [ linux-2.6.32.71 ] ~ [ linux-2.6.0 ] ~ [ linux-2.4.37.11 ] ~ [ unix-v6-master ] ~ [ ccs-tools-1.8.5 ] ~ [ policy-sample ] ~
Architecture: ~ [ i386 ] ~ [ alpha ] ~ [ m68k ] ~ [ mips ] ~ [ ppc ] ~ [ sparc ] ~ [ sparc64 ] ~

  1 /*
  2  *      IPv6 virtual tunneling interface
  3  *
  4  *      Copyright (C) 2013 secunet Security Networks AG
  5  *
  6  *      Author:
  7  *      Steffen Klassert <steffen.klassert@secunet.com>
  8  *
  9  *      Based on:
 10  *      net/ipv6/ip6_tunnel.c
 11  *
 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.
 16  */
 17 
 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>
 24 #include <linux/if.h>
 25 #include <linux/in.h>
 26 #include <linux/ip.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>
 38 
 39 #include <linux/uaccess.h>
 40 #include <linux/atomic.h>
 41 
 42 #include <net/icmp.h>
 43 #include <net/ip.h>
 44 #include <net/ip_tunnels.h>
 45 #include <net/ipv6.h>
 46 #include <net/ip6_route.h>
 47 #include <net/addrconf.h>
 48 #include <net/ip6_tunnel.h>
 49 #include <net/xfrm.h>
 50 #include <net/net_namespace.h>
 51 #include <net/netns/generic.h>
 52 #include <linux/etherdevice.h>
 53 
 54 #define IP6_VTI_HASH_SIZE_SHIFT  5
 55 #define IP6_VTI_HASH_SIZE (1 << IP6_VTI_HASH_SIZE_SHIFT)
 56 
 57 static u32 HASH(const struct in6_addr *addr1, const struct in6_addr *addr2)
 58 {
 59         u32 hash = ipv6_addr_hash(addr1) ^ ipv6_addr_hash(addr2);
 60 
 61         return hash_32(hash, IP6_VTI_HASH_SIZE_SHIFT);
 62 }
 63 
 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;
 67 
 68 static unsigned int vti6_net_id __read_mostly;
 69 struct vti6_net {
 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];
 76 };
 77 
 78 #define for_each_vti6_tunnel_rcu(start) \
 79         for (t = rcu_dereference(start); t; t = rcu_dereference(t->next))
 80 
 81 /**
 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
 86  *
 87  * Return:
 88  *   tunnel matching given end-points if found,
 89  *   else fallback tunnel if its device is up,
 90  *   else %NULL
 91  **/
 92 static struct ip6_tnl *
 93 vti6_tnl_lookup(struct net *net, const struct in6_addr *remote,
 94                 const struct in6_addr *local)
 95 {
 96         unsigned int hash = HASH(remote, local);
 97         struct ip6_tnl *t;
 98         struct vti6_net *ip6n = net_generic(net, vti6_net_id);
 99         struct in6_addr any;
100 
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))
105                         return t;
106         }
107 
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))
113                         return t;
114         }
115 
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))
120                         return t;
121         }
122 
123         t = rcu_dereference(ip6n->tnls_wc[0]);
124         if (t && (t->dev->flags & IFF_UP))
125                 return t;
126 
127         return NULL;
128 }
129 
130 /**
131  * vti6_tnl_bucket - get head of list matching given tunnel parameters
132  *   @p: parameters containing tunnel end-points
133  *
134  * Description:
135  *   vti6_tnl_bucket() returns the head of the list matching the
136  *   &struct in6_addr entries laddr and raddr in @p.
137  *
138  * Return: head of IPv6 tunnel list
139  **/
140 static struct ip6_tnl __rcu **
141 vti6_tnl_bucket(struct vti6_net *ip6n, const struct __ip6_tnl_parm *p)
142 {
143         const struct in6_addr *remote = &p->raddr;
144         const struct in6_addr *local = &p->laddr;
145         unsigned int h = 0;
146         int prio = 0;
147 
148         if (!ipv6_addr_any(remote) || !ipv6_addr_any(local)) {
149                 prio = 1;
150                 h = HASH(remote, local);
151         }
152         return &ip6n->tnls[prio][h];
153 }
154 
155 static void
156 vti6_tnl_link(struct vti6_net *ip6n, struct ip6_tnl *t)
157 {
158         struct ip6_tnl __rcu **tp = vti6_tnl_bucket(ip6n, &t->parms);
159 
160         rcu_assign_pointer(t->next , rtnl_dereference(*tp));
161         rcu_assign_pointer(*tp, t);
162 }
163 
164 static void
165 vti6_tnl_unlink(struct vti6_net *ip6n, struct ip6_tnl *t)
166 {
167         struct ip6_tnl __rcu **tp;
168         struct ip6_tnl *iter;
169 
170         for (tp = vti6_tnl_bucket(ip6n, &t->parms);
171              (iter = rtnl_dereference(*tp)) != NULL;
172              tp = &iter->next) {
173                 if (t == iter) {
174                         rcu_assign_pointer(*tp, t->next);
175                         break;
176                 }
177         }
178 }
179 
180 static void vti6_dev_free(struct net_device *dev)
181 {
182         free_percpu(dev->tstats);
183 }
184 
185 static int vti6_tnl_create2(struct net_device *dev)
186 {
187         struct ip6_tnl *t = netdev_priv(dev);
188         struct net *net = dev_net(dev);
189         struct vti6_net *ip6n = net_generic(net, vti6_net_id);
190         int err;
191 
192         dev->rtnl_link_ops = &vti6_link_ops;
193         err = register_netdevice(dev);
194         if (err < 0)
195                 goto out;
196 
197         strcpy(t->parms.name, dev->name);
198 
199         dev_hold(dev);
200         vti6_tnl_link(ip6n, t);
201 
202         return 0;
203 
204 out:
205         return err;
206 }
207 
208 static struct ip6_tnl *vti6_tnl_create(struct net *net, struct __ip6_tnl_parm *p)
209 {
210         struct net_device *dev;
211         struct ip6_tnl *t;
212         char name[IFNAMSIZ];
213         int err;
214 
215         if (p->name[0])
216                 strlcpy(name, p->name, IFNAMSIZ);
217         else
218                 sprintf(name, "ip6_vti%%d");
219 
220         dev = alloc_netdev(sizeof(*t), name, NET_NAME_UNKNOWN, vti6_dev_setup);
221         if (!dev)
222                 goto failed;
223 
224         dev_net_set(dev, net);
225 
226         t = netdev_priv(dev);
227         t->parms = *p;
228         t->net = dev_net(dev);
229 
230         err = vti6_tnl_create2(dev);
231         if (err < 0)
232                 goto failed_free;
233 
234         return t;
235 
236 failed_free:
237         free_netdev(dev);
238 failed:
239         return NULL;
240 }
241 
242 /**
243  * vti6_locate - find or create tunnel matching given parameters
244  *   @net: network namespace
245  *   @p: tunnel parameters
246  *   @create: != 0 if allowed to create new tunnel if no match found
247  *
248  * Description:
249  *   vti6_locate() first tries to locate an existing tunnel
250  *   based on @parms. If this is unsuccessful, but @create is set a new
251  *   tunnel device is created and registered for use.
252  *
253  * Return:
254  *   matching tunnel or NULL
255  **/
256 static struct ip6_tnl *vti6_locate(struct net *net, struct __ip6_tnl_parm *p,
257                                    int create)
258 {
259         const struct in6_addr *remote = &p->raddr;
260         const struct in6_addr *local = &p->laddr;
261         struct ip6_tnl __rcu **tp;
262         struct ip6_tnl *t;
263         struct vti6_net *ip6n = net_generic(net, vti6_net_id);
264 
265         for (tp = vti6_tnl_bucket(ip6n, p);
266              (t = rtnl_dereference(*tp)) != NULL;
267              tp = &t->next) {
268                 if (ipv6_addr_equal(local, &t->parms.laddr) &&
269                     ipv6_addr_equal(remote, &t->parms.raddr)) {
270                         if (create)
271                                 return NULL;
272 
273                         return t;
274                 }
275         }
276         if (!create)
277                 return NULL;
278         return vti6_tnl_create(net, p);
279 }
280 
281 /**
282  * vti6_dev_uninit - tunnel device uninitializer
283  *   @dev: the device to be destroyed
284  *
285  * Description:
286  *   vti6_dev_uninit() removes tunnel from its list
287  **/
288 static void vti6_dev_uninit(struct net_device *dev)
289 {
290         struct ip6_tnl *t = netdev_priv(dev);
291         struct vti6_net *ip6n = net_generic(t->net, vti6_net_id);
292 
293         if (dev == ip6n->fb_tnl_dev)
294                 RCU_INIT_POINTER(ip6n->tnls_wc[0], NULL);
295         else
296                 vti6_tnl_unlink(ip6n, t);
297         dev_put(dev);
298 }
299 
300 static int vti6_rcv(struct sk_buff *skb)
301 {
302         struct ip6_tnl *t;
303         const struct ipv6hdr *ipv6h = ipv6_hdr(skb);
304 
305         rcu_read_lock();
306         t = vti6_tnl_lookup(dev_net(skb->dev), &ipv6h->saddr, &ipv6h->daddr);
307         if (t) {
308                 if (t->parms.proto != IPPROTO_IPV6 && t->parms.proto != 0) {
309                         rcu_read_unlock();
310                         goto discard;
311                 }
312 
313                 if (!xfrm6_policy_check(NULL, XFRM_POLICY_IN, skb)) {
314                         rcu_read_unlock();
315                         return 0;
316                 }
317 
318                 if (!ip6_tnl_rcv_ctl(t, &ipv6h->daddr, &ipv6h->saddr)) {
319                         t->dev->stats.rx_dropped++;
320                         rcu_read_unlock();
321                         goto discard;
322                 }
323 
324                 rcu_read_unlock();
325 
326                 return xfrm6_rcv_tnl(skb, t);
327         }
328         rcu_read_unlock();
329         return -EINVAL;
330 discard:
331         kfree_skb(skb);
332         return 0;
333 }
334 
335 static int vti6_rcv_cb(struct sk_buff *skb, int err)
336 {
337         unsigned short family;
338         struct net_device *dev;
339         struct pcpu_sw_netstats *tstats;
340         struct xfrm_state *x;
341         struct xfrm_mode *inner_mode;
342         struct ip6_tnl *t = XFRM_TUNNEL_SKB_CB(skb)->tunnel.ip6;
343         u32 orig_mark = skb->mark;
344         int ret;
345 
346         if (!t)
347                 return 1;
348 
349         dev = t->dev;
350 
351         if (err) {
352                 dev->stats.rx_errors++;
353                 dev->stats.rx_dropped++;
354 
355                 return 0;
356         }
357 
358         x = xfrm_input_state(skb);
359 
360         inner_mode = x->inner_mode;
361 
362         if (x->sel.family == AF_UNSPEC) {
363                 inner_mode = xfrm_ip2inner_mode(x, XFRM_MODE_SKB_CB(skb)->protocol);
364                 if (inner_mode == NULL) {
365                         XFRM_INC_STATS(dev_net(skb->dev),
366                                        LINUX_MIB_XFRMINSTATEMODEERROR);
367                         return -EINVAL;
368                 }
369         }
370 
371         family = inner_mode->afinfo->family;
372 
373         skb->mark = be32_to_cpu(t->parms.i_key);
374         ret = xfrm_policy_check(NULL, XFRM_POLICY_IN, skb, family);
375         skb->mark = orig_mark;
376 
377         if (!ret)
378                 return -EPERM;
379 
380         skb_scrub_packet(skb, !net_eq(t->net, dev_net(skb->dev)));
381         skb->dev = dev;
382 
383         tstats = this_cpu_ptr(dev->tstats);
384         u64_stats_update_begin(&tstats->syncp);
385         tstats->rx_packets++;
386         tstats->rx_bytes += skb->len;
387         u64_stats_update_end(&tstats->syncp);
388 
389         return 0;
390 }
391 
392 /**
393  * vti6_addr_conflict - compare packet addresses to tunnel's own
394  *   @t: the outgoing tunnel device
395  *   @hdr: IPv6 header from the incoming packet
396  *
397  * Description:
398  *   Avoid trivial tunneling loop by checking that tunnel exit-point
399  *   doesn't match source of incoming packet.
400  *
401  * Return:
402  *   1 if conflict,
403  *   0 else
404  **/
405 static inline bool
406 vti6_addr_conflict(const struct ip6_tnl *t, const struct ipv6hdr *hdr)
407 {
408         return ipv6_addr_equal(&t->parms.raddr, &hdr->saddr);
409 }
410 
411 static bool vti6_state_check(const struct xfrm_state *x,
412                              const struct in6_addr *dst,
413                              const struct in6_addr *src)
414 {
415         xfrm_address_t *daddr = (xfrm_address_t *)dst;
416         xfrm_address_t *saddr = (xfrm_address_t *)src;
417 
418         /* if there is no transform then this tunnel is not functional.
419          * Or if the xfrm is not mode tunnel.
420          */
421         if (!x || x->props.mode != XFRM_MODE_TUNNEL ||
422             x->props.family != AF_INET6)
423                 return false;
424 
425         if (ipv6_addr_any(dst))
426                 return xfrm_addr_equal(saddr, &x->props.saddr, AF_INET6);
427 
428         if (!xfrm_state_addr_check(x, daddr, saddr, AF_INET6))
429                 return false;
430 
431         return true;
432 }
433 
434 /**
435  * vti6_xmit - send a packet
436  *   @skb: the outgoing socket buffer
437  *   @dev: the outgoing tunnel device
438  *   @fl: the flow informations for the xfrm_lookup
439  **/
440 static int
441 vti6_xmit(struct sk_buff *skb, struct net_device *dev, struct flowi *fl)
442 {
443         struct ip6_tnl *t = netdev_priv(dev);
444         struct net_device_stats *stats = &t->dev->stats;
445         struct dst_entry *dst = skb_dst(skb);
446         struct net_device *tdev;
447         struct xfrm_state *x;
448         int pkt_len = skb->len;
449         int err = -1;
450         int mtu;
451 
452         if (!dst)
453                 goto tx_err_link_failure;
454 
455         dst_hold(dst);
456         dst = xfrm_lookup(t->net, dst, fl, NULL, 0);
457         if (IS_ERR(dst)) {
458                 err = PTR_ERR(dst);
459                 dst = NULL;
460                 goto tx_err_link_failure;
461         }
462 
463         x = dst->xfrm;
464         if (!vti6_state_check(x, &t->parms.raddr, &t->parms.laddr))
465                 goto tx_err_link_failure;
466 
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;
470 
471         tdev = dst->dev;
472 
473         if (tdev == dev) {
474                 stats->collisions++;
475                 net_warn_ratelimited("%s: Local routing loop detected!\n",
476                                      t->parms.name);
477                 goto tx_err_dst_release;
478         }
479 
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;
483 
484         mtu = dst_mtu(dst);
485         if (!skb->ignore_df && skb->len > mtu) {
486                 skb_dst(skb)->ops->update_pmtu(dst, NULL, skb, mtu);
487 
488                 if (skb->protocol == htons(ETH_P_IPV6)) {
489                         if (mtu < IPV6_MIN_MTU)
490                                 mtu = IPV6_MIN_MTU;
491 
492                         icmpv6_send(skb, ICMPV6_PKT_TOOBIG, 0, mtu);
493                 } else {
494                         icmp_send(skb, ICMP_DEST_UNREACH, ICMP_FRAG_NEEDED,
495                                   htonl(mtu));
496                 }
497 
498                 return -EMSGSIZE;
499         }
500 
501         err = dst_output(t->net, skb->sk, skb);
502         if (net_xmit_eval(err) == 0) {
503                 struct pcpu_sw_netstats *tstats = this_cpu_ptr(dev->tstats);
504 
505                 u64_stats_update_begin(&tstats->syncp);
506                 tstats->tx_bytes += pkt_len;
507                 tstats->tx_packets++;
508                 u64_stats_update_end(&tstats->syncp);
509         } else {
510                 stats->tx_errors++;
511                 stats->tx_aborted_errors++;
512         }
513 
514         return 0;
515 tx_err_link_failure:
516         stats->tx_carrier_errors++;
517         dst_link_failure(skb);
518 tx_err_dst_release:
519         dst_release(dst);
520         return err;
521 }
522 
523 static netdev_tx_t
524 vti6_tnl_xmit(struct sk_buff *skb, struct net_device *dev)
525 {
526         struct ip6_tnl *t = netdev_priv(dev);
527         struct net_device_stats *stats = &t->dev->stats;
528         struct ipv6hdr *ipv6h;
529         struct flowi fl;
530         int ret;
531 
532         memset(&fl, 0, sizeof(fl));
533 
534         switch (skb->protocol) {
535         case htons(ETH_P_IPV6):
536                 ipv6h = ipv6_hdr(skb);
537 
538                 if ((t->parms.proto != IPPROTO_IPV6 && t->parms.proto != 0) ||
539                     vti6_addr_conflict(t, ipv6h))
540                         goto tx_err;
541 
542                 xfrm_decode_session(skb, &fl, AF_INET6);
543                 memset(IP6CB(skb), 0, sizeof(*IP6CB(skb)));
544                 break;
545         case htons(ETH_P_IP):
546                 xfrm_decode_session(skb, &fl, AF_INET);
547                 memset(IPCB(skb), 0, sizeof(*IPCB(skb)));
548                 break;
549         default:
550                 goto tx_err;
551         }
552 
553         /* override mark with tunnel output key */
554         fl.flowi_mark = be32_to_cpu(t->parms.o_key);
555 
556         ret = vti6_xmit(skb, dev, &fl);
557         if (ret < 0)
558                 goto tx_err;
559 
560         return NETDEV_TX_OK;
561 
562 tx_err:
563         stats->tx_errors++;
564         stats->tx_dropped++;
565         kfree_skb(skb);
566         return NETDEV_TX_OK;
567 }
568 
569 static int vti6_err(struct sk_buff *skb, struct inet6_skb_parm *opt,
570                     u8 type, u8 code, int offset, __be32 info)
571 {
572         __be32 spi;
573         __u32 mark;
574         struct xfrm_state *x;
575         struct ip6_tnl *t;
576         struct ip_esp_hdr *esph;
577         struct ip_auth_hdr *ah;
578         struct ip_comp_hdr *ipch;
579         struct net *net = dev_net(skb->dev);
580         const struct ipv6hdr *iph = (const struct ipv6hdr *)skb->data;
581         int protocol = iph->nexthdr;
582 
583         t = vti6_tnl_lookup(dev_net(skb->dev), &iph->daddr, &iph->saddr);
584         if (!t)
585                 return -1;
586 
587         mark = be32_to_cpu(t->parms.o_key);
588 
589         switch (protocol) {
590         case IPPROTO_ESP:
591                 esph = (struct ip_esp_hdr *)(skb->data + offset);
592                 spi = esph->spi;
593                 break;
594         case IPPROTO_AH:
595                 ah = (struct ip_auth_hdr *)(skb->data + offset);
596                 spi = ah->spi;
597                 break;
598         case IPPROTO_COMP:
599                 ipch = (struct ip_comp_hdr *)(skb->data + offset);
600                 spi = htonl(ntohs(ipch->cpi));
601                 break;
602         default:
603                 return 0;
604         }
605 
606         if (type != ICMPV6_PKT_TOOBIG &&
607             type != NDISC_REDIRECT)
608                 return 0;
609 
610         x = xfrm_state_lookup(net, mark, (const xfrm_address_t *)&iph->daddr,
611                               spi, protocol, AF_INET6);
612         if (!x)
613                 return 0;
614 
615         if (type == NDISC_REDIRECT)
616                 ip6_redirect(skb, net, skb->dev->ifindex, 0,
617                              sock_net_uid(net, NULL));
618         else
619                 ip6_update_pmtu(skb, net, info, 0, 0, sock_net_uid(net, NULL));
620         xfrm_state_put(x);
621 
622         return 0;
623 }
624 
625 static void vti6_link_config(struct ip6_tnl *t)
626 {
627         struct net_device *dev = t->dev;
628         struct __ip6_tnl_parm *p = &t->parms;
629 
630         memcpy(dev->dev_addr, &p->laddr, sizeof(struct in6_addr));
631         memcpy(dev->broadcast, &p->raddr, sizeof(struct in6_addr));
632 
633         p->flags &= ~(IP6_TNL_F_CAP_XMIT | IP6_TNL_F_CAP_RCV |
634                       IP6_TNL_F_CAP_PER_PACKET);
635         p->flags |= ip6_tnl_get_cap(t, &p->laddr, &p->raddr);
636 
637         if (p->flags & IP6_TNL_F_CAP_XMIT && p->flags & IP6_TNL_F_CAP_RCV)
638                 dev->flags |= IFF_POINTOPOINT;
639         else
640                 dev->flags &= ~IFF_POINTOPOINT;
641 }
642 
643 /**
644  * vti6_tnl_change - update the tunnel parameters
645  *   @t: tunnel to be changed
646  *   @p: tunnel configuration parameters
647  *
648  * Description:
649  *   vti6_tnl_change() updates the tunnel parameters
650  **/
651 static int
652 vti6_tnl_change(struct ip6_tnl *t, const struct __ip6_tnl_parm *p)
653 {
654         t->parms.laddr = p->laddr;
655         t->parms.raddr = p->raddr;
656         t->parms.link = p->link;
657         t->parms.i_key = p->i_key;
658         t->parms.o_key = p->o_key;
659         t->parms.proto = p->proto;
660         t->parms.fwmark = p->fwmark;
661         dst_cache_reset(&t->dst_cache);
662         vti6_link_config(t);
663         return 0;
664 }
665 
666 static int vti6_update(struct ip6_tnl *t, struct __ip6_tnl_parm *p)
667 {
668         struct net *net = dev_net(t->dev);
669         struct vti6_net *ip6n = net_generic(net, vti6_net_id);
670         int err;
671 
672         vti6_tnl_unlink(ip6n, t);
673         synchronize_net();
674         err = vti6_tnl_change(t, p);
675         vti6_tnl_link(ip6n, t);
676         netdev_state_change(t->dev);
677         return err;
678 }
679 
680 static void
681 vti6_parm_from_user(struct __ip6_tnl_parm *p, const struct ip6_tnl_parm2 *u)
682 {
683         p->laddr = u->laddr;
684         p->raddr = u->raddr;
685         p->link = u->link;
686         p->i_key = u->i_key;
687         p->o_key = u->o_key;
688         p->proto = u->proto;
689 
690         memcpy(p->name, u->name, sizeof(u->name));
691 }
692 
693 static void
694 vti6_parm_to_user(struct ip6_tnl_parm2 *u, const struct __ip6_tnl_parm *p)
695 {
696         u->laddr = p->laddr;
697         u->raddr = p->raddr;
698         u->link = p->link;
699         u->i_key = p->i_key;
700         u->o_key = p->o_key;
701         if (u->i_key)
702                 u->i_flags |= GRE_KEY;
703         if (u->o_key)
704                 u->o_flags |= GRE_KEY;
705         u->proto = p->proto;
706 
707         memcpy(u->name, p->name, sizeof(u->name));
708 }
709 
710 /**
711  * vti6_tnl_ioctl - configure vti6 tunnels from userspace
712  *   @dev: virtual device associated with tunnel
713  *   @ifr: parameters passed from userspace
714  *   @cmd: command to be performed
715  *
716  * Description:
717  *   vti6_ioctl() is used for managing vti6 tunnels
718  *   from userspace.
719  *
720  *   The possible commands are the following:
721  *     %SIOCGETTUNNEL: get tunnel parameters for device
722  *     %SIOCADDTUNNEL: add tunnel matching given tunnel parameters
723  *     %SIOCCHGTUNNEL: change tunnel parameters to those given
724  *     %SIOCDELTUNNEL: delete tunnel
725  *
726  *   The fallback device "ip6_vti0", created during module
727  *   initialization, can be used for creating other tunnel devices.
728  *
729  * Return:
730  *   0 on success,
731  *   %-EFAULT if unable to copy data to or from userspace,
732  *   %-EPERM if current process hasn't %CAP_NET_ADMIN set
733  *   %-EINVAL if passed tunnel parameters are invalid,
734  *   %-EEXIST if changing a tunnel's parameters would cause a conflict
735  *   %-ENODEV if attempting to change or delete a nonexisting device
736  **/
737 static int
738 vti6_ioctl(struct net_device *dev, struct ifreq *ifr, int cmd)
739 {
740         int err = 0;
741         struct ip6_tnl_parm2 p;
742         struct __ip6_tnl_parm p1;
743         struct ip6_tnl *t = NULL;
744         struct net *net = dev_net(dev);
745         struct vti6_net *ip6n = net_generic(net, vti6_net_id);
746 
747         switch (cmd) {
748         case SIOCGETTUNNEL:
749                 if (dev == ip6n->fb_tnl_dev) {
750                         if (copy_from_user(&p, ifr->ifr_ifru.ifru_data, sizeof(p))) {
751                                 err = -EFAULT;
752                                 break;
753                         }
754                         vti6_parm_from_user(&p1, &p);
755                         t = vti6_locate(net, &p1, 0);
756                 } else {
757                         memset(&p, 0, sizeof(p));
758                 }
759                 if (!t)
760                         t = netdev_priv(dev);
761                 vti6_parm_to_user(&p, &t->parms);
762                 if (copy_to_user(ifr->ifr_ifru.ifru_data, &p, sizeof(p)))
763                         err = -EFAULT;
764                 break;
765         case SIOCADDTUNNEL:
766         case SIOCCHGTUNNEL:
767                 err = -EPERM;
768                 if (!ns_capable(net->user_ns, CAP_NET_ADMIN))
769                         break;
770                 err = -EFAULT;
771                 if (copy_from_user(&p, ifr->ifr_ifru.ifru_data, sizeof(p)))
772                         break;
773                 err = -EINVAL;
774                 if (p.proto != IPPROTO_IPV6  && p.proto != 0)
775                         break;
776                 vti6_parm_from_user(&p1, &p);
777                 t = vti6_locate(net, &p1, cmd == SIOCADDTUNNEL);
778                 if (dev != ip6n->fb_tnl_dev && cmd == SIOCCHGTUNNEL) {
779                         if (t) {
780                                 if (t->dev != dev) {
781                                         err = -EEXIST;
782                                         break;
783                                 }
784                         } else
785                                 t = netdev_priv(dev);
786 
787                         err = vti6_update(t, &p1);
788                 }
789                 if (t) {
790                         err = 0;
791                         vti6_parm_to_user(&p, &t->parms);
792                         if (copy_to_user(ifr->ifr_ifru.ifru_data, &p, sizeof(p)))
793                                 err = -EFAULT;
794 
795                 } else
796                         err = (cmd == SIOCADDTUNNEL ? -ENOBUFS : -ENOENT);
797                 break;
798         case SIOCDELTUNNEL:
799                 err = -EPERM;
800                 if (!ns_capable(net->user_ns, CAP_NET_ADMIN))
801                         break;
802 
803                 if (dev == ip6n->fb_tnl_dev) {
804                         err = -EFAULT;
805                         if (copy_from_user(&p, ifr->ifr_ifru.ifru_data, sizeof(p)))
806                                 break;
807                         err = -ENOENT;
808                         vti6_parm_from_user(&p1, &p);
809                         t = vti6_locate(net, &p1, 0);
810                         if (!t)
811                                 break;
812                         err = -EPERM;
813                         if (t->dev == ip6n->fb_tnl_dev)
814                                 break;
815                         dev = t->dev;
816                 }
817                 err = 0;
818                 unregister_netdevice(dev);
819                 break;
820         default:
821                 err = -EINVAL;
822         }
823         return err;
824 }
825 
826 static const struct net_device_ops vti6_netdev_ops = {
827         .ndo_init       = vti6_dev_init,
828         .ndo_uninit     = vti6_dev_uninit,
829         .ndo_start_xmit = vti6_tnl_xmit,
830         .ndo_do_ioctl   = vti6_ioctl,
831         .ndo_get_stats64 = ip_tunnel_get_stats64,
832         .ndo_get_iflink = ip6_tnl_get_iflink,
833 };
834 
835 /**
836  * vti6_dev_setup - setup virtual tunnel device
837  *   @dev: virtual device associated with tunnel
838  *
839  * Description:
840  *   Initialize function pointers and device parameters
841  **/
842 static void vti6_dev_setup(struct net_device *dev)
843 {
844         dev->netdev_ops = &vti6_netdev_ops;
845         dev->needs_free_netdev = true;
846         dev->priv_destructor = vti6_dev_free;
847 
848         dev->type = ARPHRD_TUNNEL6;
849         dev->hard_header_len = LL_MAX_HEADER + sizeof(struct ipv6hdr);
850         dev->mtu = ETH_DATA_LEN;
851         dev->min_mtu = IPV6_MIN_MTU;
852         dev->max_mtu = IP_MAX_MTU;
853         dev->flags |= IFF_NOARP;
854         dev->addr_len = sizeof(struct in6_addr);
855         netif_keep_dst(dev);
856         /* This perm addr will be used as interface identifier by IPv6 */
857         dev->addr_assign_type = NET_ADDR_RANDOM;
858         eth_random_addr(dev->perm_addr);
859 }
860 
861 /**
862  * vti6_dev_init_gen - general initializer for all tunnel devices
863  *   @dev: virtual device associated with tunnel
864  **/
865 static inline int vti6_dev_init_gen(struct net_device *dev)
866 {
867         struct ip6_tnl *t = netdev_priv(dev);
868 
869         t->dev = dev;
870         t->net = dev_net(dev);
871         dev->tstats = netdev_alloc_pcpu_stats(struct pcpu_sw_netstats);
872         if (!dev->tstats)
873                 return -ENOMEM;
874         return 0;
875 }
876 
877 /**
878  * vti6_dev_init - initializer for all non fallback tunnel devices
879  *   @dev: virtual device associated with tunnel
880  **/
881 static int vti6_dev_init(struct net_device *dev)
882 {
883         struct ip6_tnl *t = netdev_priv(dev);
884         int err = vti6_dev_init_gen(dev);
885 
886         if (err)
887                 return err;
888         vti6_link_config(t);
889         return 0;
890 }
891 
892 /**
893  * vti6_fb_tnl_dev_init - initializer for fallback tunnel device
894  *   @dev: fallback device
895  *
896  * Return: 0
897  **/
898 static int __net_init vti6_fb_tnl_dev_init(struct net_device *dev)
899 {
900         struct ip6_tnl *t = netdev_priv(dev);
901         struct net *net = dev_net(dev);
902         struct vti6_net *ip6n = net_generic(net, vti6_net_id);
903 
904         t->parms.proto = IPPROTO_IPV6;
905         dev_hold(dev);
906 
907         rcu_assign_pointer(ip6n->tnls_wc[0], t);
908         return 0;
909 }
910 
911 static int vti6_validate(struct nlattr *tb[], struct nlattr *data[],
912                          struct netlink_ext_ack *extack)
913 {
914         return 0;
915 }
916 
917 static void vti6_netlink_parms(struct nlattr *data[],
918                                struct __ip6_tnl_parm *parms)
919 {
920         memset(parms, 0, sizeof(*parms));
921 
922         if (!data)
923                 return;
924 
925         if (data[IFLA_VTI_LINK])
926                 parms->link = nla_get_u32(data[IFLA_VTI_LINK]);
927 
928         if (data[IFLA_VTI_LOCAL])
929                 parms->laddr = nla_get_in6_addr(data[IFLA_VTI_LOCAL]);
930 
931         if (data[IFLA_VTI_REMOTE])
932                 parms->raddr = nla_get_in6_addr(data[IFLA_VTI_REMOTE]);
933 
934         if (data[IFLA_VTI_IKEY])
935                 parms->i_key = nla_get_be32(data[IFLA_VTI_IKEY]);
936 
937         if (data[IFLA_VTI_OKEY])
938                 parms->o_key = nla_get_be32(data[IFLA_VTI_OKEY]);
939 
940         if (data[IFLA_VTI_FWMARK])
941                 parms->fwmark = nla_get_u32(data[IFLA_VTI_FWMARK]);
942 }
943 
944 static int vti6_newlink(struct net *src_net, struct net_device *dev,
945                         struct nlattr *tb[], struct nlattr *data[],
946                         struct netlink_ext_ack *extack)
947 {
948         struct net *net = dev_net(dev);
949         struct ip6_tnl *nt;
950 
951         nt = netdev_priv(dev);
952         vti6_netlink_parms(data, &nt->parms);
953 
954         nt->parms.proto = IPPROTO_IPV6;
955 
956         if (vti6_locate(net, &nt->parms, 0))
957                 return -EEXIST;
958 
959         return vti6_tnl_create2(dev);
960 }
961 
962 static void vti6_dellink(struct net_device *dev, struct list_head *head)
963 {
964         struct net *net = dev_net(dev);
965         struct vti6_net *ip6n = net_generic(net, vti6_net_id);
966 
967         if (dev != ip6n->fb_tnl_dev)
968                 unregister_netdevice_queue(dev, head);
969 }
970 
971 static int vti6_changelink(struct net_device *dev, struct nlattr *tb[],
972                            struct nlattr *data[],
973                            struct netlink_ext_ack *extack)
974 {
975         struct ip6_tnl *t;
976         struct __ip6_tnl_parm p;
977         struct net *net = dev_net(dev);
978         struct vti6_net *ip6n = net_generic(net, vti6_net_id);
979 
980         if (dev == ip6n->fb_tnl_dev)
981                 return -EINVAL;
982 
983         vti6_netlink_parms(data, &p);
984 
985         t = vti6_locate(net, &p, 0);
986 
987         if (t) {
988                 if (t->dev != dev)
989                         return -EEXIST;
990         } else
991                 t = netdev_priv(dev);
992 
993         return vti6_update(t, &p);
994 }
995 
996 static size_t vti6_get_size(const struct net_device *dev)
997 {
998         return
999                 /* IFLA_VTI_LINK */
1000                 nla_total_size(4) +
1001                 /* IFLA_VTI_LOCAL */
1002                 nla_total_size(sizeof(struct in6_addr)) +
1003                 /* IFLA_VTI_REMOTE */
1004                 nla_total_size(sizeof(struct in6_addr)) +
1005                 /* IFLA_VTI_IKEY */
1006                 nla_total_size(4) +
1007                 /* IFLA_VTI_OKEY */
1008                 nla_total_size(4) +
1009                 /* IFLA_VTI_FWMARK */
1010                 nla_total_size(4) +
1011                 0;
1012 }
1013 
1014 static int vti6_fill_info(struct sk_buff *skb, const struct net_device *dev)
1015 {
1016         struct ip6_tnl *tunnel = netdev_priv(dev);
1017         struct __ip6_tnl_parm *parm = &tunnel->parms;
1018 
1019         if (nla_put_u32(skb, IFLA_VTI_LINK, parm->link) ||
1020             nla_put_in6_addr(skb, IFLA_VTI_LOCAL, &parm->laddr) ||
1021             nla_put_in6_addr(skb, IFLA_VTI_REMOTE, &parm->raddr) ||
1022             nla_put_be32(skb, IFLA_VTI_IKEY, parm->i_key) ||
1023             nla_put_be32(skb, IFLA_VTI_OKEY, parm->o_key) ||
1024             nla_put_u32(skb, IFLA_VTI_FWMARK, parm->fwmark))
1025                 goto nla_put_failure;
1026         return 0;
1027 
1028 nla_put_failure:
1029         return -EMSGSIZE;
1030 }
1031 
1032 static const struct nla_policy vti6_policy[IFLA_VTI_MAX + 1] = {
1033         [IFLA_VTI_LINK]         = { .type = NLA_U32 },
1034         [IFLA_VTI_LOCAL]        = { .len = sizeof(struct in6_addr) },
1035         [IFLA_VTI_REMOTE]       = { .len = sizeof(struct in6_addr) },
1036         [IFLA_VTI_IKEY]         = { .type = NLA_U32 },
1037         [IFLA_VTI_OKEY]         = { .type = NLA_U32 },
1038         [IFLA_VTI_FWMARK]       = { .type = NLA_U32 },
1039 };
1040 
1041 static struct rtnl_link_ops vti6_link_ops __read_mostly = {
1042         .kind           = "vti6",
1043         .maxtype        = IFLA_VTI_MAX,
1044         .policy         = vti6_policy,
1045         .priv_size      = sizeof(struct ip6_tnl),
1046         .setup          = vti6_dev_setup,
1047         .validate       = vti6_validate,
1048         .newlink        = vti6_newlink,
1049         .dellink        = vti6_dellink,
1050         .changelink     = vti6_changelink,
1051         .get_size       = vti6_get_size,
1052         .fill_info      = vti6_fill_info,
1053         .get_link_net   = ip6_tnl_get_link_net,
1054 };
1055 
1056 static void __net_exit vti6_destroy_tunnels(struct vti6_net *ip6n)
1057 {
1058         int h;
1059         struct ip6_tnl *t;
1060         LIST_HEAD(list);
1061 
1062         for (h = 0; h < IP6_VTI_HASH_SIZE; h++) {
1063                 t = rtnl_dereference(ip6n->tnls_r_l[h]);
1064                 while (t) {
1065                         unregister_netdevice_queue(t->dev, &list);
1066                         t = rtnl_dereference(t->next);
1067                 }
1068         }
1069 
1070         t = rtnl_dereference(ip6n->tnls_wc[0]);
1071         unregister_netdevice_queue(t->dev, &list);
1072         unregister_netdevice_many(&list);
1073 }
1074 
1075 static int __net_init vti6_init_net(struct net *net)
1076 {
1077         struct vti6_net *ip6n = net_generic(net, vti6_net_id);
1078         struct ip6_tnl *t = NULL;
1079         int err;
1080 
1081         ip6n->tnls[0] = ip6n->tnls_wc;
1082         ip6n->tnls[1] = ip6n->tnls_r_l;
1083 
1084         err = -ENOMEM;
1085         ip6n->fb_tnl_dev = alloc_netdev(sizeof(struct ip6_tnl), "ip6_vti0",
1086                                         NET_NAME_UNKNOWN, vti6_dev_setup);
1087 
1088         if (!ip6n->fb_tnl_dev)
1089                 goto err_alloc_dev;
1090         dev_net_set(ip6n->fb_tnl_dev, net);
1091         ip6n->fb_tnl_dev->rtnl_link_ops = &vti6_link_ops;
1092 
1093         err = vti6_fb_tnl_dev_init(ip6n->fb_tnl_dev);
1094         if (err < 0)
1095                 goto err_register;
1096 
1097         err = register_netdev(ip6n->fb_tnl_dev);
1098         if (err < 0)
1099                 goto err_register;
1100 
1101         t = netdev_priv(ip6n->fb_tnl_dev);
1102 
1103         strcpy(t->parms.name, ip6n->fb_tnl_dev->name);
1104         return 0;
1105 
1106 err_register:
1107         free_netdev(ip6n->fb_tnl_dev);
1108 err_alloc_dev:
1109         return err;
1110 }
1111 
1112 static void __net_exit vti6_exit_net(struct net *net)
1113 {
1114         struct vti6_net *ip6n = net_generic(net, vti6_net_id);
1115 
1116         rtnl_lock();
1117         vti6_destroy_tunnels(ip6n);
1118         rtnl_unlock();
1119 }
1120 
1121 static struct pernet_operations vti6_net_ops = {
1122         .init = vti6_init_net,
1123         .exit = vti6_exit_net,
1124         .id   = &vti6_net_id,
1125         .size = sizeof(struct vti6_net),
1126 };
1127 
1128 static struct xfrm6_protocol vti_esp6_protocol __read_mostly = {
1129         .handler        =       vti6_rcv,
1130         .cb_handler     =       vti6_rcv_cb,
1131         .err_handler    =       vti6_err,
1132         .priority       =       100,
1133 };
1134 
1135 static struct xfrm6_protocol vti_ah6_protocol __read_mostly = {
1136         .handler        =       vti6_rcv,
1137         .cb_handler     =       vti6_rcv_cb,
1138         .err_handler    =       vti6_err,
1139         .priority       =       100,
1140 };
1141 
1142 static struct xfrm6_protocol vti_ipcomp6_protocol __read_mostly = {
1143         .handler        =       vti6_rcv,
1144         .cb_handler     =       vti6_rcv_cb,
1145         .err_handler    =       vti6_err,
1146         .priority       =       100,
1147 };
1148 
1149 static bool is_vti6_tunnel(const struct net_device *dev)
1150 {
1151         return dev->netdev_ops == &vti6_netdev_ops;
1152 }
1153 
1154 static int vti6_device_event(struct notifier_block *unused,
1155                              unsigned long event, void *ptr)
1156 {
1157         struct net_device *dev = netdev_notifier_info_to_dev(ptr);
1158         struct ip6_tnl *t = netdev_priv(dev);
1159 
1160         if (!is_vti6_tunnel(dev))
1161                 return NOTIFY_DONE;
1162 
1163         switch (event) {
1164         case NETDEV_DOWN:
1165                 if (!net_eq(t->net, dev_net(dev)))
1166                         xfrm_garbage_collect(t->net);
1167                 break;
1168         }
1169         return NOTIFY_DONE;
1170 }
1171 
1172 static struct notifier_block vti6_notifier_block __read_mostly = {
1173         .notifier_call = vti6_device_event,
1174 };
1175 
1176 /**
1177  * vti6_tunnel_init - register protocol and reserve needed resources
1178  *
1179  * Return: 0 on success
1180  **/
1181 static int __init vti6_tunnel_init(void)
1182 {
1183         const char *msg;
1184         int err;
1185 
1186         register_netdevice_notifier(&vti6_notifier_block);
1187 
1188         msg = "tunnel device";
1189         err = register_pernet_device(&vti6_net_ops);
1190         if (err < 0)
1191                 goto pernet_dev_failed;
1192 
1193         msg = "tunnel protocols";
1194         err = xfrm6_protocol_register(&vti_esp6_protocol, IPPROTO_ESP);
1195         if (err < 0)
1196                 goto xfrm_proto_esp_failed;
1197         err = xfrm6_protocol_register(&vti_ah6_protocol, IPPROTO_AH);
1198         if (err < 0)
1199                 goto xfrm_proto_ah_failed;
1200         err = xfrm6_protocol_register(&vti_ipcomp6_protocol, IPPROTO_COMP);
1201         if (err < 0)
1202                 goto xfrm_proto_comp_failed;
1203 
1204         msg = "netlink interface";
1205         err = rtnl_link_register(&vti6_link_ops);
1206         if (err < 0)
1207                 goto rtnl_link_failed;
1208 
1209         return 0;
1210 
1211 rtnl_link_failed:
1212         xfrm6_protocol_deregister(&vti_ipcomp6_protocol, IPPROTO_COMP);
1213 xfrm_proto_comp_failed:
1214         xfrm6_protocol_deregister(&vti_ah6_protocol, IPPROTO_AH);
1215 xfrm_proto_ah_failed:
1216         xfrm6_protocol_deregister(&vti_esp6_protocol, IPPROTO_ESP);
1217 xfrm_proto_esp_failed:
1218         unregister_pernet_device(&vti6_net_ops);
1219 pernet_dev_failed:
1220         unregister_netdevice_notifier(&vti6_notifier_block);
1221         pr_err("vti6 init: failed to register %s\n", msg);
1222         return err;
1223 }
1224 
1225 /**
1226  * vti6_tunnel_cleanup - free resources and unregister protocol
1227  **/
1228 static void __exit vti6_tunnel_cleanup(void)
1229 {
1230         rtnl_link_unregister(&vti6_link_ops);
1231         xfrm6_protocol_deregister(&vti_ipcomp6_protocol, IPPROTO_COMP);
1232         xfrm6_protocol_deregister(&vti_ah6_protocol, IPPROTO_AH);
1233         xfrm6_protocol_deregister(&vti_esp6_protocol, IPPROTO_ESP);
1234         unregister_pernet_device(&vti6_net_ops);
1235         unregister_netdevice_notifier(&vti6_notifier_block);
1236 }
1237 
1238 module_init(vti6_tunnel_init);
1239 module_exit(vti6_tunnel_cleanup);
1240 MODULE_LICENSE("GPL");
1241 MODULE_ALIAS_RTNL_LINK("vti6");
1242 MODULE_ALIAS_NETDEV("ip6_vti0");
1243 MODULE_AUTHOR("Steffen Klassert");
1244 MODULE_DESCRIPTION("IPv6 virtual tunnel interface");
1245 

~ [ source navigation ] ~ [ diff markup ] ~ [ identifier search ] ~

kernel.org | git.kernel.org | LWN.net | Project Home | Wiki (Japanese) | Wiki (English) | SVN repository | Mail admin

Linux® is a registered trademark of Linus Torvalds in the United States and other countries.
TOMOYO® is a registered trademark of NTT DATA CORPORATION.

osdn.jp