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

TOMOYO Linux Cross Reference
Linux/net/ipv6/ip6_tunnel.c

Version: ~ [ linux-5.12-rc1 ] ~ [ linux-5.11.2 ] ~ [ linux-5.10.19 ] ~ [ linux-5.9.16 ] ~ [ linux-5.8.18 ] ~ [ linux-5.7.19 ] ~ [ linux-5.6.19 ] ~ [ linux-5.5.19 ] ~ [ linux-5.4.101 ] ~ [ linux-5.3.18 ] ~ [ linux-5.2.21 ] ~ [ linux-5.1.21 ] ~ [ linux-5.0.21 ] ~ [ linux-4.20.17 ] ~ [ linux-4.19.177 ] ~ [ linux-4.18.20 ] ~ [ linux-4.17.19 ] ~ [ linux-4.16.18 ] ~ [ linux-4.15.18 ] ~ [ linux-4.14.222 ] ~ [ linux-4.13.16 ] ~ [ linux-4.12.14 ] ~ [ linux-4.11.12 ] ~ [ linux-4.10.17 ] ~ [ linux-4.9.258 ] ~ [ linux-4.8.17 ] ~ [ linux-4.7.10 ] ~ [ linux-4.6.7 ] ~ [ linux-4.5.7 ] ~ [ linux-4.4.258 ] ~ [ 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 tunneling device
  3  *      Linux INET6 implementation
  4  *
  5  *      Authors:
  6  *      Ville Nuorvala          <vnuorval@tcs.hut.fi>
  7  *      Yasuyuki Kozakai        <kozakai@linux-ipv6.org>
  8  *
  9  *      Based on:
 10  *      linux/net/ipv6/sit.c and linux/net/ipv4/ipip.c
 11  *
 12  *      RFC 2473
 13  *
 14  *      This program is free software; you can redistribute it and/or
 15  *      modify it under the terms of the GNU General Public License
 16  *      as published by the Free Software Foundation; either version
 17  *      2 of the License, or (at your option) any later version.
 18  *
 19  */
 20 
 21 #include <linux/module.h>
 22 #include <linux/capability.h>
 23 #include <linux/errno.h>
 24 #include <linux/types.h>
 25 #include <linux/sockios.h>
 26 #include <linux/icmp.h>
 27 #include <linux/if.h>
 28 #include <linux/in.h>
 29 #include <linux/ip.h>
 30 #include <linux/if_tunnel.h>
 31 #include <linux/net.h>
 32 #include <linux/in6.h>
 33 #include <linux/netdevice.h>
 34 #include <linux/if_arp.h>
 35 #include <linux/icmpv6.h>
 36 #include <linux/init.h>
 37 #include <linux/route.h>
 38 #include <linux/rtnetlink.h>
 39 #include <linux/netfilter_ipv6.h>
 40 
 41 #include <asm/uaccess.h>
 42 #include <asm/atomic.h>
 43 
 44 #include <net/icmp.h>
 45 #include <net/ip.h>
 46 #include <net/ipv6.h>
 47 #include <net/ip6_route.h>
 48 #include <net/addrconf.h>
 49 #include <net/ip6_tunnel.h>
 50 #include <net/xfrm.h>
 51 #include <net/dsfield.h>
 52 #include <net/inet_ecn.h>
 53 #include <net/net_namespace.h>
 54 #include <net/netns/generic.h>
 55 
 56 MODULE_AUTHOR("Ville Nuorvala");
 57 MODULE_DESCRIPTION("IPv6 tunneling device");
 58 MODULE_LICENSE("GPL");
 59 MODULE_ALIAS_NETDEV("ip6tnl0");
 60 
 61 #define IPV6_TLV_TEL_DST_SIZE 8
 62 
 63 #ifdef IP6_TNL_DEBUG
 64 #define IP6_TNL_TRACE(x...) printk(KERN_DEBUG "%s:" x "\n", __func__)
 65 #else
 66 #define IP6_TNL_TRACE(x...) do {;} while(0)
 67 #endif
 68 
 69 #define IPV6_TCLASS_MASK (IPV6_FLOWINFO_MASK & ~IPV6_FLOWLABEL_MASK)
 70 #define IPV6_TCLASS_SHIFT 20
 71 
 72 #define HASH_SIZE  32
 73 
 74 #define HASH(addr) ((__force u32)((addr)->s6_addr32[0] ^ (addr)->s6_addr32[1] ^ \
 75                      (addr)->s6_addr32[2] ^ (addr)->s6_addr32[3]) & \
 76                     (HASH_SIZE - 1))
 77 
 78 static void ip6_fb_tnl_dev_init(struct net_device *dev);
 79 static void ip6_tnl_dev_init(struct net_device *dev);
 80 static void ip6_tnl_dev_setup(struct net_device *dev);
 81 
 82 static int ip6_tnl_net_id;
 83 struct ip6_tnl_net {
 84         /* the IPv6 tunnel fallback device */
 85         struct net_device *fb_tnl_dev;
 86         /* lists for storing tunnels in use */
 87         struct ip6_tnl *tnls_r_l[HASH_SIZE];
 88         struct ip6_tnl *tnls_wc[1];
 89         struct ip6_tnl **tnls[2];
 90 };
 91 
 92 /* lock for the tunnel lists */
 93 static DEFINE_RWLOCK(ip6_tnl_lock);
 94 
 95 static inline struct dst_entry *ip6_tnl_dst_check(struct ip6_tnl *t)
 96 {
 97         struct dst_entry *dst = t->dst_cache;
 98 
 99         if (dst && dst->obsolete &&
100             dst->ops->check(dst, t->dst_cookie) == NULL) {
101                 t->dst_cache = NULL;
102                 dst_release(dst);
103                 return NULL;
104         }
105 
106         return dst;
107 }
108 
109 static inline void ip6_tnl_dst_reset(struct ip6_tnl *t)
110 {
111         dst_release(t->dst_cache);
112         t->dst_cache = NULL;
113 }
114 
115 static inline void ip6_tnl_dst_store(struct ip6_tnl *t, struct dst_entry *dst)
116 {
117         struct rt6_info *rt = (struct rt6_info *) dst;
118         t->dst_cookie = rt->rt6i_node ? rt->rt6i_node->fn_sernum : 0;
119         dst_release(t->dst_cache);
120         t->dst_cache = dst;
121 }
122 
123 /**
124  * ip6_tnl_lookup - fetch tunnel matching the end-point addresses
125  *   @remote: the address of the tunnel exit-point
126  *   @local: the address of the tunnel entry-point
127  *
128  * Return:
129  *   tunnel matching given end-points if found,
130  *   else fallback tunnel if its device is up,
131  *   else %NULL
132  **/
133 
134 static struct ip6_tnl *
135 ip6_tnl_lookup(struct net *net, struct in6_addr *remote, struct in6_addr *local)
136 {
137         unsigned h0 = HASH(remote);
138         unsigned h1 = HASH(local);
139         struct ip6_tnl *t;
140         struct ip6_tnl_net *ip6n = net_generic(net, ip6_tnl_net_id);
141 
142         for (t = ip6n->tnls_r_l[h0 ^ h1]; t; t = t->next) {
143                 if (ipv6_addr_equal(local, &t->parms.laddr) &&
144                     ipv6_addr_equal(remote, &t->parms.raddr) &&
145                     (t->dev->flags & IFF_UP))
146                         return t;
147         }
148         if ((t = ip6n->tnls_wc[0]) != NULL && (t->dev->flags & IFF_UP))
149                 return t;
150 
151         return NULL;
152 }
153 
154 /**
155  * ip6_tnl_bucket - get head of list matching given tunnel parameters
156  *   @p: parameters containing tunnel end-points
157  *
158  * Description:
159  *   ip6_tnl_bucket() returns the head of the list matching the
160  *   &struct in6_addr entries laddr and raddr in @p.
161  *
162  * Return: head of IPv6 tunnel list
163  **/
164 
165 static struct ip6_tnl **
166 ip6_tnl_bucket(struct ip6_tnl_net *ip6n, struct ip6_tnl_parm *p)
167 {
168         struct in6_addr *remote = &p->raddr;
169         struct in6_addr *local = &p->laddr;
170         unsigned h = 0;
171         int prio = 0;
172 
173         if (!ipv6_addr_any(remote) || !ipv6_addr_any(local)) {
174                 prio = 1;
175                 h = HASH(remote) ^ HASH(local);
176         }
177         return &ip6n->tnls[prio][h];
178 }
179 
180 /**
181  * ip6_tnl_link - add tunnel to hash table
182  *   @t: tunnel to be added
183  **/
184 
185 static void
186 ip6_tnl_link(struct ip6_tnl_net *ip6n, struct ip6_tnl *t)
187 {
188         struct ip6_tnl **tp = ip6_tnl_bucket(ip6n, &t->parms);
189 
190         t->next = *tp;
191         write_lock_bh(&ip6_tnl_lock);
192         *tp = t;
193         write_unlock_bh(&ip6_tnl_lock);
194 }
195 
196 /**
197  * ip6_tnl_unlink - remove tunnel from hash table
198  *   @t: tunnel to be removed
199  **/
200 
201 static void
202 ip6_tnl_unlink(struct ip6_tnl_net *ip6n, struct ip6_tnl *t)
203 {
204         struct ip6_tnl **tp;
205 
206         for (tp = ip6_tnl_bucket(ip6n, &t->parms); *tp; tp = &(*tp)->next) {
207                 if (t == *tp) {
208                         write_lock_bh(&ip6_tnl_lock);
209                         *tp = t->next;
210                         write_unlock_bh(&ip6_tnl_lock);
211                         break;
212                 }
213         }
214 }
215 
216 /**
217  * ip6_tnl_create() - create a new tunnel
218  *   @p: tunnel parameters
219  *   @pt: pointer to new tunnel
220  *
221  * Description:
222  *   Create tunnel matching given parameters.
223  *
224  * Return:
225  *   created tunnel or NULL
226  **/
227 
228 static struct ip6_tnl *ip6_tnl_create(struct net *net, struct ip6_tnl_parm *p)
229 {
230         struct net_device *dev;
231         struct ip6_tnl *t;
232         char name[IFNAMSIZ];
233         int err;
234         struct ip6_tnl_net *ip6n = net_generic(net, ip6_tnl_net_id);
235 
236         if (p->name[0])
237                 strlcpy(name, p->name, IFNAMSIZ);
238         else
239                 sprintf(name, "ip6tnl%%d");
240 
241         dev = alloc_netdev(sizeof (*t), name, ip6_tnl_dev_setup);
242         if (dev == NULL)
243                 goto failed;
244 
245         dev_net_set(dev, net);
246 
247         if (strchr(name, '%')) {
248                 if (dev_alloc_name(dev, name) < 0)
249                         goto failed_free;
250         }
251 
252         t = netdev_priv(dev);
253         t->parms = *p;
254         ip6_tnl_dev_init(dev);
255 
256         if ((err = register_netdevice(dev)) < 0)
257                 goto failed_free;
258 
259         dev_hold(dev);
260         ip6_tnl_link(ip6n, t);
261         return t;
262 
263 failed_free:
264         free_netdev(dev);
265 failed:
266         return NULL;
267 }
268 
269 /**
270  * ip6_tnl_locate - find or create tunnel matching given parameters
271  *   @p: tunnel parameters
272  *   @create: != 0 if allowed to create new tunnel if no match found
273  *
274  * Description:
275  *   ip6_tnl_locate() first tries to locate an existing tunnel
276  *   based on @parms. If this is unsuccessful, but @create is set a new
277  *   tunnel device is created and registered for use.
278  *
279  * Return:
280  *   matching tunnel or NULL
281  **/
282 
283 static struct ip6_tnl *ip6_tnl_locate(struct net *net,
284                 struct ip6_tnl_parm *p, int create)
285 {
286         struct in6_addr *remote = &p->raddr;
287         struct in6_addr *local = &p->laddr;
288         struct ip6_tnl *t;
289         struct ip6_tnl_net *ip6n = net_generic(net, ip6_tnl_net_id);
290 
291         for (t = *ip6_tnl_bucket(ip6n, p); t; t = t->next) {
292                 if (ipv6_addr_equal(local, &t->parms.laddr) &&
293                     ipv6_addr_equal(remote, &t->parms.raddr))
294                         return t;
295         }
296         if (!create)
297                 return NULL;
298         return ip6_tnl_create(net, p);
299 }
300 
301 /**
302  * ip6_tnl_dev_uninit - tunnel device uninitializer
303  *   @dev: the device to be destroyed
304  *
305  * Description:
306  *   ip6_tnl_dev_uninit() removes tunnel from its list
307  **/
308 
309 static void
310 ip6_tnl_dev_uninit(struct net_device *dev)
311 {
312         struct ip6_tnl *t = netdev_priv(dev);
313         struct net *net = dev_net(dev);
314         struct ip6_tnl_net *ip6n = net_generic(net, ip6_tnl_net_id);
315 
316         if (dev == ip6n->fb_tnl_dev) {
317                 write_lock_bh(&ip6_tnl_lock);
318                 ip6n->tnls_wc[0] = NULL;
319                 write_unlock_bh(&ip6_tnl_lock);
320         } else {
321                 ip6_tnl_unlink(ip6n, t);
322         }
323         ip6_tnl_dst_reset(t);
324         dev_put(dev);
325 }
326 
327 /**
328  * parse_tvl_tnl_enc_lim - handle encapsulation limit option
329  *   @skb: received socket buffer
330  *
331  * Return:
332  *   0 if none was found,
333  *   else index to encapsulation limit
334  **/
335 
336 static __u16
337 parse_tlv_tnl_enc_lim(struct sk_buff *skb, __u8 * raw)
338 {
339         struct ipv6hdr *ipv6h = (struct ipv6hdr *) raw;
340         __u8 nexthdr = ipv6h->nexthdr;
341         __u16 off = sizeof (*ipv6h);
342 
343         while (ipv6_ext_hdr(nexthdr) && nexthdr != NEXTHDR_NONE) {
344                 __u16 optlen = 0;
345                 struct ipv6_opt_hdr *hdr;
346                 if (raw + off + sizeof (*hdr) > skb->data &&
347                     !pskb_may_pull(skb, raw - skb->data + off + sizeof (*hdr)))
348                         break;
349 
350                 hdr = (struct ipv6_opt_hdr *) (raw + off);
351                 if (nexthdr == NEXTHDR_FRAGMENT) {
352                         struct frag_hdr *frag_hdr = (struct frag_hdr *) hdr;
353                         if (frag_hdr->frag_off)
354                                 break;
355                         optlen = 8;
356                 } else if (nexthdr == NEXTHDR_AUTH) {
357                         optlen = (hdr->hdrlen + 2) << 2;
358                 } else {
359                         optlen = ipv6_optlen(hdr);
360                 }
361                 if (nexthdr == NEXTHDR_DEST) {
362                         __u16 i = off + 2;
363                         while (1) {
364                                 struct ipv6_tlv_tnl_enc_lim *tel;
365 
366                                 /* No more room for encapsulation limit */
367                                 if (i + sizeof (*tel) > off + optlen)
368                                         break;
369 
370                                 tel = (struct ipv6_tlv_tnl_enc_lim *) &raw[i];
371                                 /* return index of option if found and valid */
372                                 if (tel->type == IPV6_TLV_TNL_ENCAP_LIMIT &&
373                                     tel->length == 1)
374                                         return i;
375                                 /* else jump to next option */
376                                 if (tel->type)
377                                         i += tel->length + 2;
378                                 else
379                                         i++;
380                         }
381                 }
382                 nexthdr = hdr->nexthdr;
383                 off += optlen;
384         }
385         return 0;
386 }
387 
388 /**
389  * ip6_tnl_err - tunnel error handler
390  *
391  * Description:
392  *   ip6_tnl_err() should handle errors in the tunnel according
393  *   to the specifications in RFC 2473.
394  **/
395 
396 static int
397 ip6_tnl_err(struct sk_buff *skb, __u8 ipproto, struct inet6_skb_parm *opt,
398             u8 *type, u8 *code, int *msg, __u32 *info, int offset)
399 {
400         struct ipv6hdr *ipv6h = (struct ipv6hdr *) skb->data;
401         struct ip6_tnl *t;
402         int rel_msg = 0;
403         u8 rel_type = ICMPV6_DEST_UNREACH;
404         u8 rel_code = ICMPV6_ADDR_UNREACH;
405         __u32 rel_info = 0;
406         __u16 len;
407         int err = -ENOENT;
408 
409         /* If the packet doesn't contain the original IPv6 header we are
410            in trouble since we might need the source address for further
411            processing of the error. */
412 
413         read_lock(&ip6_tnl_lock);
414         if ((t = ip6_tnl_lookup(dev_net(skb->dev), &ipv6h->daddr,
415                                         &ipv6h->saddr)) == NULL)
416                 goto out;
417 
418         if (t->parms.proto != ipproto && t->parms.proto != 0)
419                 goto out;
420 
421         err = 0;
422 
423         switch (*type) {
424                 __u32 teli;
425                 struct ipv6_tlv_tnl_enc_lim *tel;
426                 __u32 mtu;
427         case ICMPV6_DEST_UNREACH:
428                 if (net_ratelimit())
429                         printk(KERN_WARNING
430                                "%s: Path to destination invalid "
431                                "or inactive!\n", t->parms.name);
432                 rel_msg = 1;
433                 break;
434         case ICMPV6_TIME_EXCEED:
435                 if ((*code) == ICMPV6_EXC_HOPLIMIT) {
436                         if (net_ratelimit())
437                                 printk(KERN_WARNING
438                                        "%s: Too small hop limit or "
439                                        "routing loop in tunnel!\n",
440                                        t->parms.name);
441                         rel_msg = 1;
442                 }
443                 break;
444         case ICMPV6_PARAMPROB:
445                 teli = 0;
446                 if ((*code) == ICMPV6_HDR_FIELD)
447                         teli = parse_tlv_tnl_enc_lim(skb, skb->data);
448 
449                 if (teli && teli == *info - 2) {
450                         tel = (struct ipv6_tlv_tnl_enc_lim *) &skb->data[teli];
451                         if (tel->encap_limit == 0) {
452                                 if (net_ratelimit())
453                                         printk(KERN_WARNING
454                                                "%s: Too small encapsulation "
455                                                "limit or routing loop in "
456                                                "tunnel!\n", t->parms.name);
457                                 rel_msg = 1;
458                         }
459                 } else if (net_ratelimit()) {
460                         printk(KERN_WARNING
461                                "%s: Recipient unable to parse tunneled "
462                                "packet!\n ", t->parms.name);
463                 }
464                 break;
465         case ICMPV6_PKT_TOOBIG:
466                 mtu = *info - offset;
467                 if (mtu < IPV6_MIN_MTU)
468                         mtu = IPV6_MIN_MTU;
469                 t->dev->mtu = mtu;
470 
471                 if ((len = sizeof (*ipv6h) + ntohs(ipv6h->payload_len)) > mtu) {
472                         rel_type = ICMPV6_PKT_TOOBIG;
473                         rel_code = 0;
474                         rel_info = mtu;
475                         rel_msg = 1;
476                 }
477                 break;
478         }
479 
480         *type = rel_type;
481         *code = rel_code;
482         *info = rel_info;
483         *msg = rel_msg;
484 
485 out:
486         read_unlock(&ip6_tnl_lock);
487         return err;
488 }
489 
490 static int
491 ip4ip6_err(struct sk_buff *skb, struct inet6_skb_parm *opt,
492            u8 type, u8 code, int offset, __be32 info)
493 {
494         int rel_msg = 0;
495         u8 rel_type = type;
496         u8 rel_code = code;
497         __u32 rel_info = ntohl(info);
498         int err;
499         struct sk_buff *skb2;
500         struct iphdr *eiph;
501         struct flowi fl;
502         struct rtable *rt;
503 
504         err = ip6_tnl_err(skb, IPPROTO_IPIP, opt, &rel_type, &rel_code,
505                           &rel_msg, &rel_info, offset);
506         if (err < 0)
507                 return err;
508 
509         if (rel_msg == 0)
510                 return 0;
511 
512         switch (rel_type) {
513         case ICMPV6_DEST_UNREACH:
514                 if (rel_code != ICMPV6_ADDR_UNREACH)
515                         return 0;
516                 rel_type = ICMP_DEST_UNREACH;
517                 rel_code = ICMP_HOST_UNREACH;
518                 break;
519         case ICMPV6_PKT_TOOBIG:
520                 if (rel_code != 0)
521                         return 0;
522                 rel_type = ICMP_DEST_UNREACH;
523                 rel_code = ICMP_FRAG_NEEDED;
524                 break;
525         default:
526                 return 0;
527         }
528 
529         if (!pskb_may_pull(skb, offset + sizeof(struct iphdr)))
530                 return 0;
531 
532         skb2 = skb_clone(skb, GFP_ATOMIC);
533         if (!skb2)
534                 return 0;
535 
536         skb_dst_drop(skb2);
537 
538         skb_pull(skb2, offset);
539         skb_reset_network_header(skb2);
540         eiph = ip_hdr(skb2);
541 
542         /* Try to guess incoming interface */
543         memset(&fl, 0, sizeof(fl));
544         fl.fl4_dst = eiph->saddr;
545         fl.fl4_tos = RT_TOS(eiph->tos);
546         fl.proto = IPPROTO_IPIP;
547         if (ip_route_output_key(dev_net(skb->dev), &rt, &fl))
548                 goto out;
549 
550         skb2->dev = rt->u.dst.dev;
551 
552         /* route "incoming" packet */
553         if (rt->rt_flags & RTCF_LOCAL) {
554                 ip_rt_put(rt);
555                 rt = NULL;
556                 fl.fl4_dst = eiph->daddr;
557                 fl.fl4_src = eiph->saddr;
558                 fl.fl4_tos = eiph->tos;
559                 if (ip_route_output_key(dev_net(skb->dev), &rt, &fl) ||
560                     rt->u.dst.dev->type != ARPHRD_TUNNEL) {
561                         ip_rt_put(rt);
562                         goto out;
563                 }
564                 skb_dst_set(skb2, (struct dst_entry *)rt);
565         } else {
566                 ip_rt_put(rt);
567                 if (ip_route_input(skb2, eiph->daddr, eiph->saddr, eiph->tos,
568                                    skb2->dev) ||
569                     skb_dst(skb2)->dev->type != ARPHRD_TUNNEL)
570                         goto out;
571         }
572 
573         /* change mtu on this route */
574         if (rel_type == ICMP_DEST_UNREACH && rel_code == ICMP_FRAG_NEEDED) {
575                 if (rel_info > dst_mtu(skb_dst(skb2)))
576                         goto out;
577 
578                 skb_dst(skb2)->ops->update_pmtu(skb_dst(skb2), rel_info);
579         }
580 
581         icmp_send(skb2, rel_type, rel_code, htonl(rel_info));
582 
583 out:
584         kfree_skb(skb2);
585         return 0;
586 }
587 
588 static int
589 ip6ip6_err(struct sk_buff *skb, struct inet6_skb_parm *opt,
590            u8 type, u8 code, int offset, __be32 info)
591 {
592         int rel_msg = 0;
593         u8 rel_type = type;
594         u8 rel_code = code;
595         __u32 rel_info = ntohl(info);
596         int err;
597 
598         err = ip6_tnl_err(skb, IPPROTO_IPV6, opt, &rel_type, &rel_code,
599                           &rel_msg, &rel_info, offset);
600         if (err < 0)
601                 return err;
602 
603         if (rel_msg && pskb_may_pull(skb, offset + sizeof(struct ipv6hdr))) {
604                 struct rt6_info *rt;
605                 struct sk_buff *skb2 = skb_clone(skb, GFP_ATOMIC);
606 
607                 if (!skb2)
608                         return 0;
609 
610                 skb_dst_drop(skb2);
611                 skb_pull(skb2, offset);
612                 skb_reset_network_header(skb2);
613 
614                 /* Try to guess incoming interface */
615                 rt = rt6_lookup(dev_net(skb->dev), &ipv6_hdr(skb2)->saddr,
616                                 NULL, 0, 0);
617 
618                 if (rt && rt->rt6i_dev)
619                         skb2->dev = rt->rt6i_dev;
620 
621                 icmpv6_send(skb2, rel_type, rel_code, rel_info, skb2->dev);
622 
623                 if (rt)
624                         dst_release(&rt->u.dst);
625 
626                 kfree_skb(skb2);
627         }
628 
629         return 0;
630 }
631 
632 static void ip4ip6_dscp_ecn_decapsulate(struct ip6_tnl *t,
633                                         struct ipv6hdr *ipv6h,
634                                         struct sk_buff *skb)
635 {
636         __u8 dsfield = ipv6_get_dsfield(ipv6h) & ~INET_ECN_MASK;
637 
638         if (t->parms.flags & IP6_TNL_F_RCV_DSCP_COPY)
639                 ipv4_change_dsfield(ip_hdr(skb), INET_ECN_MASK, dsfield);
640 
641         if (INET_ECN_is_ce(dsfield))
642                 IP_ECN_set_ce(ip_hdr(skb));
643 }
644 
645 static void ip6ip6_dscp_ecn_decapsulate(struct ip6_tnl *t,
646                                         struct ipv6hdr *ipv6h,
647                                         struct sk_buff *skb)
648 {
649         if (t->parms.flags & IP6_TNL_F_RCV_DSCP_COPY)
650                 ipv6_copy_dscp(ipv6_get_dsfield(ipv6h), ipv6_hdr(skb));
651 
652         if (INET_ECN_is_ce(ipv6_get_dsfield(ipv6h)))
653                 IP6_ECN_set_ce(ipv6_hdr(skb));
654 }
655 
656 static inline int ip6_tnl_rcv_ctl(struct ip6_tnl *t)
657 {
658         struct ip6_tnl_parm *p = &t->parms;
659         int ret = 0;
660         struct net *net = dev_net(t->dev);
661 
662         if (p->flags & IP6_TNL_F_CAP_RCV) {
663                 struct net_device *ldev = NULL;
664 
665                 if (p->link)
666                         ldev = dev_get_by_index(net, p->link);
667 
668                 if ((ipv6_addr_is_multicast(&p->laddr) ||
669                      likely(ipv6_chk_addr(net, &p->laddr, ldev, 0))) &&
670                     likely(!ipv6_chk_addr(net, &p->raddr, NULL, 0)))
671                         ret = 1;
672 
673                 if (ldev)
674                         dev_put(ldev);
675         }
676         return ret;
677 }
678 
679 /**
680  * ip6_tnl_rcv - decapsulate IPv6 packet and retransmit it locally
681  *   @skb: received socket buffer
682  *   @protocol: ethernet protocol ID
683  *   @dscp_ecn_decapsulate: the function to decapsulate DSCP code and ECN
684  *
685  * Return: 0
686  **/
687 
688 static int ip6_tnl_rcv(struct sk_buff *skb, __u16 protocol,
689                        __u8 ipproto,
690                        void (*dscp_ecn_decapsulate)(struct ip6_tnl *t,
691                                                     struct ipv6hdr *ipv6h,
692                                                     struct sk_buff *skb))
693 {
694         struct ip6_tnl *t;
695         struct ipv6hdr *ipv6h = ipv6_hdr(skb);
696 
697         read_lock(&ip6_tnl_lock);
698 
699         if ((t = ip6_tnl_lookup(dev_net(skb->dev), &ipv6h->saddr,
700                                         &ipv6h->daddr)) != NULL) {
701                 if (t->parms.proto != ipproto && t->parms.proto != 0) {
702                         read_unlock(&ip6_tnl_lock);
703                         goto discard;
704                 }
705 
706                 if (!xfrm6_policy_check(NULL, XFRM_POLICY_IN, skb)) {
707                         read_unlock(&ip6_tnl_lock);
708                         goto discard;
709                 }
710 
711                 if (!ip6_tnl_rcv_ctl(t)) {
712                         t->dev->stats.rx_dropped++;
713                         read_unlock(&ip6_tnl_lock);
714                         goto discard;
715                 }
716                 secpath_reset(skb);
717                 skb->mac_header = skb->network_header;
718                 skb_reset_network_header(skb);
719                 skb->protocol = htons(protocol);
720                 skb->pkt_type = PACKET_HOST;
721                 memset(skb->cb, 0, sizeof(struct inet6_skb_parm));
722                 skb->dev = t->dev;
723                 skb_dst_drop(skb);
724                 nf_reset(skb);
725 
726                 dscp_ecn_decapsulate(t, ipv6h, skb);
727 
728                 t->dev->stats.rx_packets++;
729                 t->dev->stats.rx_bytes += skb->len;
730                 netif_rx(skb);
731                 read_unlock(&ip6_tnl_lock);
732                 return 0;
733         }
734         read_unlock(&ip6_tnl_lock);
735         return 1;
736 
737 discard:
738         kfree_skb(skb);
739         return 0;
740 }
741 
742 static int ip4ip6_rcv(struct sk_buff *skb)
743 {
744         return ip6_tnl_rcv(skb, ETH_P_IP, IPPROTO_IPIP,
745                            ip4ip6_dscp_ecn_decapsulate);
746 }
747 
748 static int ip6ip6_rcv(struct sk_buff *skb)
749 {
750         return ip6_tnl_rcv(skb, ETH_P_IPV6, IPPROTO_IPV6,
751                            ip6ip6_dscp_ecn_decapsulate);
752 }
753 
754 struct ipv6_tel_txoption {
755         struct ipv6_txoptions ops;
756         __u8 dst_opt[8];
757 };
758 
759 static void init_tel_txopt(struct ipv6_tel_txoption *opt, __u8 encap_limit)
760 {
761         memset(opt, 0, sizeof(struct ipv6_tel_txoption));
762 
763         opt->dst_opt[2] = IPV6_TLV_TNL_ENCAP_LIMIT;
764         opt->dst_opt[3] = 1;
765         opt->dst_opt[4] = encap_limit;
766         opt->dst_opt[5] = IPV6_TLV_PADN;
767         opt->dst_opt[6] = 1;
768 
769         opt->ops.dst0opt = (struct ipv6_opt_hdr *) opt->dst_opt;
770         opt->ops.opt_nflen = 8;
771 }
772 
773 /**
774  * ip6_tnl_addr_conflict - compare packet addresses to tunnel's own
775  *   @t: the outgoing tunnel device
776  *   @hdr: IPv6 header from the incoming packet
777  *
778  * Description:
779  *   Avoid trivial tunneling loop by checking that tunnel exit-point
780  *   doesn't match source of incoming packet.
781  *
782  * Return:
783  *   1 if conflict,
784  *   0 else
785  **/
786 
787 static inline int
788 ip6_tnl_addr_conflict(struct ip6_tnl *t, struct ipv6hdr *hdr)
789 {
790         return ipv6_addr_equal(&t->parms.raddr, &hdr->saddr);
791 }
792 
793 static inline int ip6_tnl_xmit_ctl(struct ip6_tnl *t)
794 {
795         struct ip6_tnl_parm *p = &t->parms;
796         int ret = 0;
797         struct net *net = dev_net(t->dev);
798 
799         if (p->flags & IP6_TNL_F_CAP_XMIT) {
800                 struct net_device *ldev = NULL;
801 
802                 if (p->link)
803                         ldev = dev_get_by_index(net, p->link);
804 
805                 if (unlikely(!ipv6_chk_addr(net, &p->laddr, ldev, 0)))
806                         printk(KERN_WARNING
807                                "%s xmit: Local address not yet configured!\n",
808                                p->name);
809                 else if (!ipv6_addr_is_multicast(&p->raddr) &&
810                          unlikely(ipv6_chk_addr(net, &p->raddr, NULL, 0)))
811                         printk(KERN_WARNING
812                                "%s xmit: Routing loop! "
813                                "Remote address found on this node!\n",
814                                p->name);
815                 else
816                         ret = 1;
817                 if (ldev)
818                         dev_put(ldev);
819         }
820         return ret;
821 }
822 /**
823  * ip6_tnl_xmit2 - encapsulate packet and send
824  *   @skb: the outgoing socket buffer
825  *   @dev: the outgoing tunnel device
826  *   @dsfield: dscp code for outer header
827  *   @fl: flow of tunneled packet
828  *   @encap_limit: encapsulation limit
829  *   @pmtu: Path MTU is stored if packet is too big
830  *
831  * Description:
832  *   Build new header and do some sanity checks on the packet before sending
833  *   it.
834  *
835  * Return:
836  *   0 on success
837  *   -1 fail
838  *   %-EMSGSIZE message too big. return mtu in this case.
839  **/
840 
841 static int ip6_tnl_xmit2(struct sk_buff *skb,
842                          struct net_device *dev,
843                          __u8 dsfield,
844                          struct flowi *fl,
845                          int encap_limit,
846                          __u32 *pmtu)
847 {
848         struct net *net = dev_net(dev);
849         struct ip6_tnl *t = netdev_priv(dev);
850         struct net_device_stats *stats = &t->dev->stats;
851         struct ipv6hdr *ipv6h = ipv6_hdr(skb);
852         struct ipv6_tel_txoption opt;
853         struct dst_entry *dst;
854         struct net_device *tdev;
855         int mtu;
856         unsigned int max_headroom = sizeof(struct ipv6hdr);
857         u8 proto;
858         int err = -1;
859         int pkt_len;
860 
861         if ((dst = ip6_tnl_dst_check(t)) != NULL)
862                 dst_hold(dst);
863         else {
864                 dst = ip6_route_output(net, NULL, fl);
865 
866                 if (dst->error || xfrm_lookup(net, &dst, fl, NULL, 0) < 0)
867                         goto tx_err_link_failure;
868         }
869 
870         tdev = dst->dev;
871 
872         if (tdev == dev) {
873                 stats->collisions++;
874                 if (net_ratelimit())
875                         printk(KERN_WARNING
876                                "%s: Local routing loop detected!\n",
877                                t->parms.name);
878                 goto tx_err_dst_release;
879         }
880         mtu = dst_mtu(dst) - sizeof (*ipv6h);
881         if (encap_limit >= 0) {
882                 max_headroom += 8;
883                 mtu -= 8;
884         }
885         if (mtu < IPV6_MIN_MTU)
886                 mtu = IPV6_MIN_MTU;
887         if (skb_dst(skb))
888                 skb_dst(skb)->ops->update_pmtu(skb_dst(skb), mtu);
889         if (skb->len > mtu) {
890                 *pmtu = mtu;
891                 err = -EMSGSIZE;
892                 goto tx_err_dst_release;
893         }
894 
895         /*
896          * Okay, now see if we can stuff it in the buffer as-is.
897          */
898         max_headroom += LL_RESERVED_SPACE(tdev);
899 
900         if (skb_headroom(skb) < max_headroom || skb_shared(skb) ||
901             (skb_cloned(skb) && !skb_clone_writable(skb, 0))) {
902                 struct sk_buff *new_skb;
903 
904                 if (!(new_skb = skb_realloc_headroom(skb, max_headroom)))
905                         goto tx_err_dst_release;
906 
907                 if (skb->sk)
908                         skb_set_owner_w(new_skb, skb->sk);
909                 kfree_skb(skb);
910                 skb = new_skb;
911         }
912         skb_dst_drop(skb);
913         skb_dst_set(skb, dst_clone(dst));
914 
915         skb->transport_header = skb->network_header;
916 
917         proto = fl->proto;
918         if (encap_limit >= 0) {
919                 init_tel_txopt(&opt, encap_limit);
920                 ipv6_push_nfrag_opts(skb, &opt.ops, &proto, NULL);
921         }
922         skb_push(skb, sizeof(struct ipv6hdr));
923         skb_reset_network_header(skb);
924         ipv6h = ipv6_hdr(skb);
925         *(__be32*)ipv6h = fl->fl6_flowlabel | htonl(0x60000000);
926         dsfield = INET_ECN_encapsulate(0, dsfield);
927         ipv6_change_dsfield(ipv6h, ~INET_ECN_MASK, dsfield);
928         ipv6h->hop_limit = t->parms.hop_limit;
929         ipv6h->nexthdr = proto;
930         ipv6_addr_copy(&ipv6h->saddr, &fl->fl6_src);
931         ipv6_addr_copy(&ipv6h->daddr, &fl->fl6_dst);
932         nf_reset(skb);
933         pkt_len = skb->len;
934         err = ip6_local_out(skb);
935 
936         if (net_xmit_eval(err) == 0) {
937                 stats->tx_bytes += pkt_len;
938                 stats->tx_packets++;
939         } else {
940                 stats->tx_errors++;
941                 stats->tx_aborted_errors++;
942         }
943         ip6_tnl_dst_store(t, dst);
944         return 0;
945 tx_err_link_failure:
946         stats->tx_carrier_errors++;
947         dst_link_failure(skb);
948 tx_err_dst_release:
949         dst_release(dst);
950         return err;
951 }
952 
953 static inline int
954 ip4ip6_tnl_xmit(struct sk_buff *skb, struct net_device *dev)
955 {
956         struct ip6_tnl *t = netdev_priv(dev);
957         struct iphdr  *iph = ip_hdr(skb);
958         int encap_limit = -1;
959         struct flowi fl;
960         __u8 dsfield;
961         __u32 mtu;
962         int err;
963 
964         if ((t->parms.proto != IPPROTO_IPIP && t->parms.proto != 0) ||
965             !ip6_tnl_xmit_ctl(t))
966                 return -1;
967 
968         if (!(t->parms.flags & IP6_TNL_F_IGN_ENCAP_LIMIT))
969                 encap_limit = t->parms.encap_limit;
970 
971         memcpy(&fl, &t->fl, sizeof (fl));
972         fl.proto = IPPROTO_IPIP;
973 
974         dsfield = ipv4_get_dsfield(iph);
975 
976         if ((t->parms.flags & IP6_TNL_F_USE_ORIG_TCLASS))
977                 fl.fl6_flowlabel |= htonl((__u32)iph->tos << IPV6_TCLASS_SHIFT)
978                                           & IPV6_TCLASS_MASK;
979 
980         err = ip6_tnl_xmit2(skb, dev, dsfield, &fl, encap_limit, &mtu);
981         if (err != 0) {
982                 /* XXX: send ICMP error even if DF is not set. */
983                 if (err == -EMSGSIZE)
984                         icmp_send(skb, ICMP_DEST_UNREACH, ICMP_FRAG_NEEDED,
985                                   htonl(mtu));
986                 return -1;
987         }
988 
989         return 0;
990 }
991 
992 static inline int
993 ip6ip6_tnl_xmit(struct sk_buff *skb, struct net_device *dev)
994 {
995         struct ip6_tnl *t = netdev_priv(dev);
996         struct ipv6hdr *ipv6h = ipv6_hdr(skb);
997         int encap_limit = -1;
998         __u16 offset;
999         struct flowi fl;
1000         __u8 dsfield;
1001         __u32 mtu;
1002         int err;
1003 
1004         if ((t->parms.proto != IPPROTO_IPV6 && t->parms.proto != 0) ||
1005             !ip6_tnl_xmit_ctl(t) || ip6_tnl_addr_conflict(t, ipv6h))
1006                 return -1;
1007 
1008         offset = parse_tlv_tnl_enc_lim(skb, skb_network_header(skb));
1009         if (offset > 0) {
1010                 struct ipv6_tlv_tnl_enc_lim *tel;
1011                 tel = (struct ipv6_tlv_tnl_enc_lim *)&skb_network_header(skb)[offset];
1012                 if (tel->encap_limit == 0) {
1013                         icmpv6_send(skb, ICMPV6_PARAMPROB,
1014                                     ICMPV6_HDR_FIELD, offset + 2, skb->dev);
1015                         return -1;
1016                 }
1017                 encap_limit = tel->encap_limit - 1;
1018         } else if (!(t->parms.flags & IP6_TNL_F_IGN_ENCAP_LIMIT))
1019                 encap_limit = t->parms.encap_limit;
1020 
1021         memcpy(&fl, &t->fl, sizeof (fl));
1022         fl.proto = IPPROTO_IPV6;
1023 
1024         dsfield = ipv6_get_dsfield(ipv6h);
1025         if ((t->parms.flags & IP6_TNL_F_USE_ORIG_TCLASS))
1026                 fl.fl6_flowlabel |= (*(__be32 *) ipv6h & IPV6_TCLASS_MASK);
1027         if ((t->parms.flags & IP6_TNL_F_USE_ORIG_FLOWLABEL))
1028                 fl.fl6_flowlabel |= (*(__be32 *) ipv6h & IPV6_FLOWLABEL_MASK);
1029 
1030         err = ip6_tnl_xmit2(skb, dev, dsfield, &fl, encap_limit, &mtu);
1031         if (err != 0) {
1032                 if (err == -EMSGSIZE)
1033                         icmpv6_send(skb, ICMPV6_PKT_TOOBIG, 0, mtu, dev);
1034                 return -1;
1035         }
1036 
1037         return 0;
1038 }
1039 
1040 static netdev_tx_t
1041 ip6_tnl_xmit(struct sk_buff *skb, struct net_device *dev)
1042 {
1043         struct ip6_tnl *t = netdev_priv(dev);
1044         struct net_device_stats *stats = &t->dev->stats;
1045         int ret;
1046 
1047         switch (skb->protocol) {
1048         case htons(ETH_P_IP):
1049                 ret = ip4ip6_tnl_xmit(skb, dev);
1050                 break;
1051         case htons(ETH_P_IPV6):
1052                 ret = ip6ip6_tnl_xmit(skb, dev);
1053                 break;
1054         default:
1055                 goto tx_err;
1056         }
1057 
1058         if (ret < 0)
1059                 goto tx_err;
1060 
1061         return NETDEV_TX_OK;
1062 
1063 tx_err:
1064         stats->tx_errors++;
1065         stats->tx_dropped++;
1066         kfree_skb(skb);
1067         return NETDEV_TX_OK;
1068 }
1069 
1070 static void ip6_tnl_set_cap(struct ip6_tnl *t)
1071 {
1072         struct ip6_tnl_parm *p = &t->parms;
1073         int ltype = ipv6_addr_type(&p->laddr);
1074         int rtype = ipv6_addr_type(&p->raddr);
1075 
1076         p->flags &= ~(IP6_TNL_F_CAP_XMIT|IP6_TNL_F_CAP_RCV);
1077 
1078         if (ltype & (IPV6_ADDR_UNICAST|IPV6_ADDR_MULTICAST) &&
1079             rtype & (IPV6_ADDR_UNICAST|IPV6_ADDR_MULTICAST) &&
1080             !((ltype|rtype) & IPV6_ADDR_LOOPBACK) &&
1081             (!((ltype|rtype) & IPV6_ADDR_LINKLOCAL) || p->link)) {
1082                 if (ltype&IPV6_ADDR_UNICAST)
1083                         p->flags |= IP6_TNL_F_CAP_XMIT;
1084                 if (rtype&IPV6_ADDR_UNICAST)
1085                         p->flags |= IP6_TNL_F_CAP_RCV;
1086         }
1087 }
1088 
1089 static void ip6_tnl_link_config(struct ip6_tnl *t)
1090 {
1091         struct net_device *dev = t->dev;
1092         struct ip6_tnl_parm *p = &t->parms;
1093         struct flowi *fl = &t->fl;
1094 
1095         memcpy(dev->dev_addr, &p->laddr, sizeof(struct in6_addr));
1096         memcpy(dev->broadcast, &p->raddr, sizeof(struct in6_addr));
1097 
1098         /* Set up flowi template */
1099         ipv6_addr_copy(&fl->fl6_src, &p->laddr);
1100         ipv6_addr_copy(&fl->fl6_dst, &p->raddr);
1101         fl->oif = p->link;
1102         fl->fl6_flowlabel = 0;
1103 
1104         if (!(p->flags&IP6_TNL_F_USE_ORIG_TCLASS))
1105                 fl->fl6_flowlabel |= IPV6_TCLASS_MASK & p->flowinfo;
1106         if (!(p->flags&IP6_TNL_F_USE_ORIG_FLOWLABEL))
1107                 fl->fl6_flowlabel |= IPV6_FLOWLABEL_MASK & p->flowinfo;
1108 
1109         ip6_tnl_set_cap(t);
1110 
1111         if (p->flags&IP6_TNL_F_CAP_XMIT && p->flags&IP6_TNL_F_CAP_RCV)
1112                 dev->flags |= IFF_POINTOPOINT;
1113         else
1114                 dev->flags &= ~IFF_POINTOPOINT;
1115 
1116         dev->iflink = p->link;
1117 
1118         if (p->flags & IP6_TNL_F_CAP_XMIT) {
1119                 int strict = (ipv6_addr_type(&p->raddr) &
1120                               (IPV6_ADDR_MULTICAST|IPV6_ADDR_LINKLOCAL));
1121 
1122                 struct rt6_info *rt = rt6_lookup(dev_net(dev),
1123                                                  &p->raddr, &p->laddr,
1124                                                  p->link, strict);
1125 
1126                 if (rt == NULL)
1127                         return;
1128 
1129                 if (rt->rt6i_dev) {
1130                         dev->hard_header_len = rt->rt6i_dev->hard_header_len +
1131                                 sizeof (struct ipv6hdr);
1132 
1133                         dev->mtu = rt->rt6i_dev->mtu - sizeof (struct ipv6hdr);
1134 
1135                         if (dev->mtu < IPV6_MIN_MTU)
1136                                 dev->mtu = IPV6_MIN_MTU;
1137                 }
1138                 dst_release(&rt->u.dst);
1139         }
1140 }
1141 
1142 /**
1143  * ip6_tnl_change - update the tunnel parameters
1144  *   @t: tunnel to be changed
1145  *   @p: tunnel configuration parameters
1146  *
1147  * Description:
1148  *   ip6_tnl_change() updates the tunnel parameters
1149  **/
1150 
1151 static int
1152 ip6_tnl_change(struct ip6_tnl *t, struct ip6_tnl_parm *p)
1153 {
1154         ipv6_addr_copy(&t->parms.laddr, &p->laddr);
1155         ipv6_addr_copy(&t->parms.raddr, &p->raddr);
1156         t->parms.flags = p->flags;
1157         t->parms.hop_limit = p->hop_limit;
1158         t->parms.encap_limit = p->encap_limit;
1159         t->parms.flowinfo = p->flowinfo;
1160         t->parms.link = p->link;
1161         t->parms.proto = p->proto;
1162         ip6_tnl_dst_reset(t);
1163         ip6_tnl_link_config(t);
1164         return 0;
1165 }
1166 
1167 /**
1168  * ip6_tnl_ioctl - configure ipv6 tunnels from userspace
1169  *   @dev: virtual device associated with tunnel
1170  *   @ifr: parameters passed from userspace
1171  *   @cmd: command to be performed
1172  *
1173  * Description:
1174  *   ip6_tnl_ioctl() is used for managing IPv6 tunnels
1175  *   from userspace.
1176  *
1177  *   The possible commands are the following:
1178  *     %SIOCGETTUNNEL: get tunnel parameters for device
1179  *     %SIOCADDTUNNEL: add tunnel matching given tunnel parameters
1180  *     %SIOCCHGTUNNEL: change tunnel parameters to those given
1181  *     %SIOCDELTUNNEL: delete tunnel
1182  *
1183  *   The fallback device "ip6tnl0", created during module
1184  *   initialization, can be used for creating other tunnel devices.
1185  *
1186  * Return:
1187  *   0 on success,
1188  *   %-EFAULT if unable to copy data to or from userspace,
1189  *   %-EPERM if current process hasn't %CAP_NET_ADMIN set
1190  *   %-EINVAL if passed tunnel parameters are invalid,
1191  *   %-EEXIST if changing a tunnel's parameters would cause a conflict
1192  *   %-ENODEV if attempting to change or delete a nonexisting device
1193  **/
1194 
1195 static int
1196 ip6_tnl_ioctl(struct net_device *dev, struct ifreq *ifr, int cmd)
1197 {
1198         int err = 0;
1199         struct ip6_tnl_parm p;
1200         struct ip6_tnl *t = NULL;
1201         struct net *net = dev_net(dev);
1202         struct ip6_tnl_net *ip6n = net_generic(net, ip6_tnl_net_id);
1203 
1204         switch (cmd) {
1205         case SIOCGETTUNNEL:
1206                 if (dev == ip6n->fb_tnl_dev) {
1207                         if (copy_from_user(&p, ifr->ifr_ifru.ifru_data, sizeof (p))) {
1208                                 err = -EFAULT;
1209                                 break;
1210                         }
1211                         t = ip6_tnl_locate(net, &p, 0);
1212                 }
1213                 if (t == NULL)
1214                         t = netdev_priv(dev);
1215                 memcpy(&p, &t->parms, sizeof (p));
1216                 if (copy_to_user(ifr->ifr_ifru.ifru_data, &p, sizeof (p))) {
1217                         err = -EFAULT;
1218                 }
1219                 break;
1220         case SIOCADDTUNNEL:
1221         case SIOCCHGTUNNEL:
1222                 err = -EPERM;
1223                 if (!capable(CAP_NET_ADMIN))
1224                         break;
1225                 err = -EFAULT;
1226                 if (copy_from_user(&p, ifr->ifr_ifru.ifru_data, sizeof (p)))
1227                         break;
1228                 err = -EINVAL;
1229                 if (p.proto != IPPROTO_IPV6 && p.proto != IPPROTO_IPIP &&
1230                     p.proto != 0)
1231                         break;
1232                 t = ip6_tnl_locate(net, &p, cmd == SIOCADDTUNNEL);
1233                 if (dev != ip6n->fb_tnl_dev && cmd == SIOCCHGTUNNEL) {
1234                         if (t != NULL) {
1235                                 if (t->dev != dev) {
1236                                         err = -EEXIST;
1237                                         break;
1238                                 }
1239                         } else
1240                                 t = netdev_priv(dev);
1241 
1242                         ip6_tnl_unlink(ip6n, t);
1243                         err = ip6_tnl_change(t, &p);
1244                         ip6_tnl_link(ip6n, t);
1245                         netdev_state_change(dev);
1246                 }
1247                 if (t) {
1248                         err = 0;
1249                         if (copy_to_user(ifr->ifr_ifru.ifru_data, &t->parms, sizeof (p)))
1250                                 err = -EFAULT;
1251 
1252                 } else
1253                         err = (cmd == SIOCADDTUNNEL ? -ENOBUFS : -ENOENT);
1254                 break;
1255         case SIOCDELTUNNEL:
1256                 err = -EPERM;
1257                 if (!capable(CAP_NET_ADMIN))
1258                         break;
1259 
1260                 if (dev == ip6n->fb_tnl_dev) {
1261                         err = -EFAULT;
1262                         if (copy_from_user(&p, ifr->ifr_ifru.ifru_data, sizeof (p)))
1263                                 break;
1264                         err = -ENOENT;
1265                         if ((t = ip6_tnl_locate(net, &p, 0)) == NULL)
1266                                 break;
1267                         err = -EPERM;
1268                         if (t->dev == ip6n->fb_tnl_dev)
1269                                 break;
1270                         dev = t->dev;
1271                 }
1272                 err = 0;
1273                 unregister_netdevice(dev);
1274                 break;
1275         default:
1276                 err = -EINVAL;
1277         }
1278         return err;
1279 }
1280 
1281 /**
1282  * ip6_tnl_change_mtu - change mtu manually for tunnel device
1283  *   @dev: virtual device associated with tunnel
1284  *   @new_mtu: the new mtu
1285  *
1286  * Return:
1287  *   0 on success,
1288  *   %-EINVAL if mtu too small
1289  **/
1290 
1291 static int
1292 ip6_tnl_change_mtu(struct net_device *dev, int new_mtu)
1293 {
1294         if (new_mtu < IPV6_MIN_MTU) {
1295                 return -EINVAL;
1296         }
1297         dev->mtu = new_mtu;
1298         return 0;
1299 }
1300 
1301 
1302 static const struct net_device_ops ip6_tnl_netdev_ops = {
1303         .ndo_uninit = ip6_tnl_dev_uninit,
1304         .ndo_start_xmit = ip6_tnl_xmit,
1305         .ndo_do_ioctl = ip6_tnl_ioctl,
1306         .ndo_change_mtu = ip6_tnl_change_mtu,
1307 };
1308 
1309 /**
1310  * ip6_tnl_dev_setup - setup virtual tunnel device
1311  *   @dev: virtual device associated with tunnel
1312  *
1313  * Description:
1314  *   Initialize function pointers and device parameters
1315  **/
1316 
1317 static void ip6_tnl_dev_setup(struct net_device *dev)
1318 {
1319         dev->netdev_ops = &ip6_tnl_netdev_ops;
1320         dev->destructor = free_netdev;
1321 
1322         dev->type = ARPHRD_TUNNEL6;
1323         dev->hard_header_len = LL_MAX_HEADER + sizeof (struct ipv6hdr);
1324         dev->mtu = ETH_DATA_LEN - sizeof (struct ipv6hdr);
1325         dev->flags |= IFF_NOARP;
1326         dev->addr_len = sizeof(struct in6_addr);
1327         dev->features |= NETIF_F_NETNS_LOCAL;
1328 }
1329 
1330 
1331 /**
1332  * ip6_tnl_dev_init_gen - general initializer for all tunnel devices
1333  *   @dev: virtual device associated with tunnel
1334  **/
1335 
1336 static inline void
1337 ip6_tnl_dev_init_gen(struct net_device *dev)
1338 {
1339         struct ip6_tnl *t = netdev_priv(dev);
1340         t->dev = dev;
1341         strcpy(t->parms.name, dev->name);
1342 }
1343 
1344 /**
1345  * ip6_tnl_dev_init - initializer for all non fallback tunnel devices
1346  *   @dev: virtual device associated with tunnel
1347  **/
1348 
1349 static void ip6_tnl_dev_init(struct net_device *dev)
1350 {
1351         struct ip6_tnl *t = netdev_priv(dev);
1352         ip6_tnl_dev_init_gen(dev);
1353         ip6_tnl_link_config(t);
1354 }
1355 
1356 /**
1357  * ip6_fb_tnl_dev_init - initializer for fallback tunnel device
1358  *   @dev: fallback device
1359  *
1360  * Return: 0
1361  **/
1362 
1363 static void ip6_fb_tnl_dev_init(struct net_device *dev)
1364 {
1365         struct ip6_tnl *t = netdev_priv(dev);
1366         struct net *net = dev_net(dev);
1367         struct ip6_tnl_net *ip6n = net_generic(net, ip6_tnl_net_id);
1368 
1369         ip6_tnl_dev_init_gen(dev);
1370         t->parms.proto = IPPROTO_IPV6;
1371         dev_hold(dev);
1372         ip6n->tnls_wc[0] = t;
1373 }
1374 
1375 static struct xfrm6_tunnel ip4ip6_handler = {
1376         .handler        = ip4ip6_rcv,
1377         .err_handler    = ip4ip6_err,
1378         .priority       =       1,
1379 };
1380 
1381 static struct xfrm6_tunnel ip6ip6_handler = {
1382         .handler        = ip6ip6_rcv,
1383         .err_handler    = ip6ip6_err,
1384         .priority       =       1,
1385 };
1386 
1387 static void ip6_tnl_destroy_tunnels(struct ip6_tnl_net *ip6n)
1388 {
1389         int h;
1390         struct ip6_tnl *t;
1391 
1392         for (h = 0; h < HASH_SIZE; h++) {
1393                 while ((t = ip6n->tnls_r_l[h]) != NULL)
1394                         unregister_netdevice(t->dev);
1395         }
1396 
1397         t = ip6n->tnls_wc[0];
1398         unregister_netdevice(t->dev);
1399 }
1400 
1401 static int ip6_tnl_init_net(struct net *net)
1402 {
1403         int err;
1404         struct ip6_tnl_net *ip6n;
1405 
1406         err = -ENOMEM;
1407         ip6n = kzalloc(sizeof(struct ip6_tnl_net), GFP_KERNEL);
1408         if (ip6n == NULL)
1409                 goto err_alloc;
1410 
1411         err = net_assign_generic(net, ip6_tnl_net_id, ip6n);
1412         if (err < 0)
1413                 goto err_assign;
1414 
1415         ip6n->tnls[0] = ip6n->tnls_wc;
1416         ip6n->tnls[1] = ip6n->tnls_r_l;
1417 
1418         err = -ENOMEM;
1419         ip6n->fb_tnl_dev = alloc_netdev(sizeof(struct ip6_tnl), "ip6tnl0",
1420                                       ip6_tnl_dev_setup);
1421 
1422         if (!ip6n->fb_tnl_dev)
1423                 goto err_alloc_dev;
1424         dev_net_set(ip6n->fb_tnl_dev, net);
1425 
1426         ip6_fb_tnl_dev_init(ip6n->fb_tnl_dev);
1427 
1428         err = register_netdev(ip6n->fb_tnl_dev);
1429         if (err < 0)
1430                 goto err_register;
1431         return 0;
1432 
1433 err_register:
1434         free_netdev(ip6n->fb_tnl_dev);
1435 err_alloc_dev:
1436         /* nothing */
1437 err_assign:
1438         kfree(ip6n);
1439 err_alloc:
1440         return err;
1441 }
1442 
1443 static void ip6_tnl_exit_net(struct net *net)
1444 {
1445         struct ip6_tnl_net *ip6n;
1446 
1447         ip6n = net_generic(net, ip6_tnl_net_id);
1448         rtnl_lock();
1449         ip6_tnl_destroy_tunnels(ip6n);
1450         rtnl_unlock();
1451         kfree(ip6n);
1452 }
1453 
1454 static struct pernet_operations ip6_tnl_net_ops = {
1455         .init = ip6_tnl_init_net,
1456         .exit = ip6_tnl_exit_net,
1457 };
1458 
1459 /**
1460  * ip6_tunnel_init - register protocol and reserve needed resources
1461  *
1462  * Return: 0 on success
1463  **/
1464 
1465 static int __init ip6_tunnel_init(void)
1466 {
1467         int  err;
1468 
1469         err = register_pernet_gen_device(&ip6_tnl_net_id, &ip6_tnl_net_ops);
1470         if (err < 0)
1471                 goto out_pernet;
1472 
1473         err = xfrm6_tunnel_register(&ip4ip6_handler, AF_INET);
1474         if (err < 0) {
1475                 printk(KERN_ERR "ip6_tunnel init: can't register ip4ip6\n");
1476                 goto out_ip4ip6;
1477         }
1478 
1479         err = xfrm6_tunnel_register(&ip6ip6_handler, AF_INET6);
1480         if (err < 0) {
1481                 printk(KERN_ERR "ip6_tunnel init: can't register ip6ip6\n");
1482                 goto out_ip6ip6;
1483         }
1484 
1485         return 0;
1486 
1487 out_ip6ip6:
1488         xfrm6_tunnel_deregister(&ip4ip6_handler, AF_INET);
1489 out_ip4ip6:
1490         unregister_pernet_gen_device(ip6_tnl_net_id, &ip6_tnl_net_ops);
1491 out_pernet:
1492         return err;
1493 }
1494 
1495 /**
1496  * ip6_tunnel_cleanup - free resources and unregister protocol
1497  **/
1498 
1499 static void __exit ip6_tunnel_cleanup(void)
1500 {
1501         if (xfrm6_tunnel_deregister(&ip4ip6_handler, AF_INET))
1502                 printk(KERN_INFO "ip6_tunnel close: can't deregister ip4ip6\n");
1503 
1504         if (xfrm6_tunnel_deregister(&ip6ip6_handler, AF_INET6))
1505                 printk(KERN_INFO "ip6_tunnel close: can't deregister ip6ip6\n");
1506 
1507         unregister_pernet_gen_device(ip6_tnl_net_id, &ip6_tnl_net_ops);
1508 }
1509 
1510 module_init(ip6_tunnel_init);
1511 module_exit(ip6_tunnel_cleanup);
1512 

~ [ 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