projects / GitHub / mt8127 / android_kernel_alcatel_ttab.git / blame_incremental
? search:
summary | shortlog | log | commit | commitdiff | tree
blob | blame (non-incremental) | history | HEAD
drivers: power: report battery voltage in AOSP compatible format
[GitHub/mt8127/android_kernel_alcatel_ttab.git] / net / ipv6 / ip6_tunnel.c
... / ...
CommitLineData
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#define pr_fmt(fmt) KBUILD_MODNAME ": " fmt
22
23#include <linux/module.h>
24#include <linux/capability.h>
25#include <linux/errno.h>
26#include <linux/types.h>
27#include <linux/sockios.h>
28#include <linux/icmp.h>
29#include <linux/if.h>
30#include <linux/in.h>
31#include <linux/ip.h>
32#include <linux/if_tunnel.h>
33#include <linux/net.h>
34#include <linux/in6.h>
35#include <linux/netdevice.h>
36#include <linux/if_arp.h>
37#include <linux/icmpv6.h>
38#include <linux/init.h>
39#include <linux/route.h>
40#include <linux/rtnetlink.h>
41#include <linux/netfilter_ipv6.h>
42#include <linux/slab.h>
43#include <linux/hash.h>
44
45#include <asm/uaccess.h>
46#include <linux/atomic.h>
47
48#include <net/icmp.h>
49#include <net/ip.h>
50#include <net/ip_tunnels.h>
51#include <net/ipv6.h>
52#include <net/ip6_route.h>
53#include <net/addrconf.h>
54#include <net/ip6_tunnel.h>
55#include <net/xfrm.h>
56#include <net/dsfield.h>
57#include <net/inet_ecn.h>
58#include <net/net_namespace.h>
59#include <net/netns/generic.h>
60
61MODULE_AUTHOR("Ville Nuorvala");
62MODULE_DESCRIPTION("IPv6 tunneling device");
63MODULE_LICENSE("GPL");
64MODULE_ALIAS_RTNL_LINK("ip6tnl");
65MODULE_ALIAS_NETDEV("ip6tnl0");
66
67#ifdef IP6_TNL_DEBUG
68#define IP6_TNL_TRACE(x...) pr_debug("%s:" x "\n", __func__)
69#else
70#define IP6_TNL_TRACE(x...) do {;} while(0)
71#endif
72
73#define IPV6_TCLASS_MASK (IPV6_FLOWINFO_MASK & ~IPV6_FLOWLABEL_MASK)
74#define IPV6_TCLASS_SHIFT 20
75
76#define HASH_SIZE_SHIFT 5
77#define HASH_SIZE (1 << HASH_SIZE_SHIFT)
78
79static bool log_ecn_error = true;
80module_param(log_ecn_error, bool, 0644);
81MODULE_PARM_DESC(log_ecn_error, "Log packets received with corrupted ECN");
82
83static u32 HASH(const struct in6_addr *addr1, const struct in6_addr *addr2)
84{
85 u32 hash = ipv6_addr_hash(addr1) ^ ipv6_addr_hash(addr2);
86
87 return hash_32(hash, HASH_SIZE_SHIFT);
88}
89
90static int ip6_tnl_dev_init(struct net_device *dev);
91static void ip6_tnl_dev_setup(struct net_device *dev);
92static struct rtnl_link_ops ip6_link_ops __read_mostly;
93
94static int ip6_tnl_net_id __read_mostly;
95struct ip6_tnl_net {
96 /* the IPv6 tunnel fallback device */
97 struct net_device *fb_tnl_dev;
98 /* lists for storing tunnels in use */
99 struct ip6_tnl __rcu *tnls_r_l[HASH_SIZE];
100 struct ip6_tnl __rcu *tnls_wc[1];
101 struct ip6_tnl __rcu **tnls[2];
102};
103
104static struct net_device_stats *ip6_get_stats(struct net_device *dev)
105{
106 struct pcpu_tstats tmp, sum = { 0 };
107 int i;
108
109 for_each_possible_cpu(i) {
110 unsigned int start;
111 const struct pcpu_tstats *tstats = per_cpu_ptr(dev->tstats, i);
112
113 do {
114 start = u64_stats_fetch_begin_bh(&tstats->syncp);
115 tmp.rx_packets = tstats->rx_packets;
116 tmp.rx_bytes = tstats->rx_bytes;
117 tmp.tx_packets = tstats->tx_packets;
118 tmp.tx_bytes = tstats->tx_bytes;
119 } while (u64_stats_fetch_retry_bh(&tstats->syncp, start));
120
121 sum.rx_packets += tmp.rx_packets;
122 sum.rx_bytes += tmp.rx_bytes;
123 sum.tx_packets += tmp.tx_packets;
124 sum.tx_bytes += tmp.tx_bytes;
125 }
126 dev->stats.rx_packets = sum.rx_packets;
127 dev->stats.rx_bytes = sum.rx_bytes;
128 dev->stats.tx_packets = sum.tx_packets;
129 dev->stats.tx_bytes = sum.tx_bytes;
130 return &dev->stats;
131}
132
133/*
134 * Locking : hash tables are protected by RCU and RTNL
135 */
136
137struct dst_entry *ip6_tnl_dst_check(struct ip6_tnl *t)
138{
139 struct dst_entry *dst = t->dst_cache;
140
141 if (dst && dst->obsolete &&
142 dst->ops->check(dst, t->dst_cookie) == NULL) {
143 t->dst_cache = NULL;
144 dst_release(dst);
145 return NULL;
146 }
147
148 return dst;
149}
150EXPORT_SYMBOL_GPL(ip6_tnl_dst_check);
151
152void ip6_tnl_dst_reset(struct ip6_tnl *t)
153{
154 dst_release(t->dst_cache);
155 t->dst_cache = NULL;
156}
157EXPORT_SYMBOL_GPL(ip6_tnl_dst_reset);
158
159void ip6_tnl_dst_store(struct ip6_tnl *t, struct dst_entry *dst)
160{
161 struct rt6_info *rt = (struct rt6_info *) dst;
162 t->dst_cookie = rt->rt6i_node ? rt->rt6i_node->fn_sernum : 0;
163 dst_release(t->dst_cache);
164 t->dst_cache = dst;
165}
166EXPORT_SYMBOL_GPL(ip6_tnl_dst_store);
167
168/**
169 * ip6_tnl_lookup - fetch tunnel matching the end-point addresses
170 * @remote: the address of the tunnel exit-point
171 * @local: the address of the tunnel entry-point
172 *
173 * Return:
174 * tunnel matching given end-points if found,
175 * else fallback tunnel if its device is up,
176 * else %NULL
177 **/
178
179#define for_each_ip6_tunnel_rcu(start) \
180 for (t = rcu_dereference(start); t; t = rcu_dereference(t->next))
181
182static struct ip6_tnl *
183ip6_tnl_lookup(struct net *net, const struct in6_addr *remote, const struct in6_addr *local)
184{
185 unsigned int hash = HASH(remote, local);
186 struct ip6_tnl *t;
187 struct ip6_tnl_net *ip6n = net_generic(net, ip6_tnl_net_id);
188
189 for_each_ip6_tunnel_rcu(ip6n->tnls_r_l[hash]) {
190 if (ipv6_addr_equal(local, &t->parms.laddr) &&
191 ipv6_addr_equal(remote, &t->parms.raddr) &&
192 (t->dev->flags & IFF_UP))
193 return t;
194 }
195 t = rcu_dereference(ip6n->tnls_wc[0]);
196 if (t && (t->dev->flags & IFF_UP))
197 return t;
198
199 return NULL;
200}
201
202/**
203 * ip6_tnl_bucket - get head of list matching given tunnel parameters
204 * @p: parameters containing tunnel end-points
205 *
206 * Description:
207 * ip6_tnl_bucket() returns the head of the list matching the
208 * &struct in6_addr entries laddr and raddr in @p.
209 *
210 * Return: head of IPv6 tunnel list
211 **/
212
213static struct ip6_tnl __rcu **
214ip6_tnl_bucket(struct ip6_tnl_net *ip6n, const struct __ip6_tnl_parm *p)
215{
216 const struct in6_addr *remote = &p->raddr;
217 const struct in6_addr *local = &p->laddr;
218 unsigned int h = 0;
219 int prio = 0;
220
221 if (!ipv6_addr_any(remote) || !ipv6_addr_any(local)) {
222 prio = 1;
223 h = HASH(remote, local);
224 }
225 return &ip6n->tnls[prio][h];
226}
227
228/**
229 * ip6_tnl_link - add tunnel to hash table
230 * @t: tunnel to be added
231 **/
232
233static void
234ip6_tnl_link(struct ip6_tnl_net *ip6n, struct ip6_tnl *t)
235{
236 struct ip6_tnl __rcu **tp = ip6_tnl_bucket(ip6n, &t->parms);
237
238 rcu_assign_pointer(t->next , rtnl_dereference(*tp));
239 rcu_assign_pointer(*tp, t);
240}
241
242/**
243 * ip6_tnl_unlink - remove tunnel from hash table
244 * @t: tunnel to be removed
245 **/
246
247static void
248ip6_tnl_unlink(struct ip6_tnl_net *ip6n, struct ip6_tnl *t)
249{
250 struct ip6_tnl __rcu **tp;
251 struct ip6_tnl *iter;
252
253 for (tp = ip6_tnl_bucket(ip6n, &t->parms);
254 (iter = rtnl_dereference(*tp)) != NULL;
255 tp = &iter->next) {
256 if (t == iter) {
257 rcu_assign_pointer(*tp, t->next);
258 break;
259 }
260 }
261}
262
263static void ip6_dev_free(struct net_device *dev)
264{
265 free_percpu(dev->tstats);
266 free_netdev(dev);
267}
268
269static int ip6_tnl_create2(struct net_device *dev)
270{
271 struct ip6_tnl *t = netdev_priv(dev);
272 struct net *net = dev_net(dev);
273 struct ip6_tnl_net *ip6n = net_generic(net, ip6_tnl_net_id);
274 int err;
275
276 t = netdev_priv(dev);
277
278 dev->rtnl_link_ops = &ip6_link_ops;
279 err = register_netdevice(dev);
280 if (err < 0)
281 goto out;
282
283 strcpy(t->parms.name, dev->name);
284
285 dev_hold(dev);
286 ip6_tnl_link(ip6n, t);
287 return 0;
288
289out:
290 return err;
291}
292
293/**
294 * ip6_tnl_create - create a new tunnel
295 * @p: tunnel parameters
296 * @pt: pointer to new tunnel
297 *
298 * Description:
299 * Create tunnel matching given parameters.
300 *
301 * Return:
302 * created tunnel or NULL
303 **/
304
305static struct ip6_tnl *ip6_tnl_create(struct net *net, struct __ip6_tnl_parm *p)
306{
307 struct net_device *dev;
308 struct ip6_tnl *t;
309 char name[IFNAMSIZ];
310 int err;
311
312 if (p->name[0])
313 strlcpy(name, p->name, IFNAMSIZ);
314 else
315 sprintf(name, "ip6tnl%%d");
316
317 dev = alloc_netdev(sizeof (*t), name, ip6_tnl_dev_setup);
318 if (dev == NULL)
319 goto failed;
320
321 dev_net_set(dev, net);
322
323 t = netdev_priv(dev);
324 t->parms = *p;
325 err = ip6_tnl_create2(dev);
326 if (err < 0)
327 goto failed_free;
328
329 return t;
330
331failed_free:
332 ip6_dev_free(dev);
333failed:
334 return NULL;
335}
336
337/**
338 * ip6_tnl_locate - find or create tunnel matching given parameters
339 * @p: tunnel parameters
340 * @create: != 0 if allowed to create new tunnel if no match found
341 *
342 * Description:
343 * ip6_tnl_locate() first tries to locate an existing tunnel
344 * based on @parms. If this is unsuccessful, but @create is set a new
345 * tunnel device is created and registered for use.
346 *
347 * Return:
348 * matching tunnel or NULL
349 **/
350
351static struct ip6_tnl *ip6_tnl_locate(struct net *net,
352 struct __ip6_tnl_parm *p, int create)
353{
354 const struct in6_addr *remote = &p->raddr;
355 const struct in6_addr *local = &p->laddr;
356 struct ip6_tnl __rcu **tp;
357 struct ip6_tnl *t;
358 struct ip6_tnl_net *ip6n = net_generic(net, ip6_tnl_net_id);
359
360 for (tp = ip6_tnl_bucket(ip6n, p);
361 (t = rtnl_dereference(*tp)) != NULL;
362 tp = &t->next) {
363 if (ipv6_addr_equal(local, &t->parms.laddr) &&
364 ipv6_addr_equal(remote, &t->parms.raddr))
365 return t;
366 }
367 if (!create)
368 return NULL;
369 return ip6_tnl_create(net, p);
370}
371
372/**
373 * ip6_tnl_dev_uninit - tunnel device uninitializer
374 * @dev: the device to be destroyed
375 *
376 * Description:
377 * ip6_tnl_dev_uninit() removes tunnel from its list
378 **/
379
380static void
381ip6_tnl_dev_uninit(struct net_device *dev)
382{
383 struct ip6_tnl *t = netdev_priv(dev);
384 struct net *net = dev_net(dev);
385 struct ip6_tnl_net *ip6n = net_generic(net, ip6_tnl_net_id);
386
387 if (dev == ip6n->fb_tnl_dev)
388 RCU_INIT_POINTER(ip6n->tnls_wc[0], NULL);
389 else
390 ip6_tnl_unlink(ip6n, t);
391 ip6_tnl_dst_reset(t);
392 dev_put(dev);
393}
394
395/**
396 * parse_tvl_tnl_enc_lim - handle encapsulation limit option
397 * @skb: received socket buffer
398 *
399 * Return:
400 * 0 if none was found,
401 * else index to encapsulation limit
402 **/
403
404__u16 ip6_tnl_parse_tlv_enc_lim(struct sk_buff *skb, __u8 *raw)
405{
406 const struct ipv6hdr *ipv6h = (const struct ipv6hdr *)raw;
407 unsigned int nhoff = raw - skb->data;
408 unsigned int off = nhoff + sizeof(*ipv6h);
409 u8 next, nexthdr = ipv6h->nexthdr;
410
411 while (ipv6_ext_hdr(nexthdr) && nexthdr != NEXTHDR_NONE) {
412 struct ipv6_opt_hdr *hdr;
413 u16 optlen;
414
415 if (!pskb_may_pull(skb, off + sizeof(*hdr)))
416 break;
417
418 hdr = (struct ipv6_opt_hdr *)(skb->data + off);
419 if (nexthdr == NEXTHDR_FRAGMENT) {
420 struct frag_hdr *frag_hdr = (struct frag_hdr *) hdr;
421 if (frag_hdr->frag_off)
422 break;
423 optlen = 8;
424 } else if (nexthdr == NEXTHDR_AUTH) {
425 optlen = (hdr->hdrlen + 2) << 2;
426 } else {
427 optlen = ipv6_optlen(hdr);
428 }
429 /* cache hdr->nexthdr, since pskb_may_pull() might
430 * invalidate hdr
431 */
432 next = hdr->nexthdr;
433 if (nexthdr == NEXTHDR_DEST) {
434 u16 i = 2;
435
436 /* Remember : hdr is no longer valid at this point. */
437 if (!pskb_may_pull(skb, off + optlen))
438 break;
439
440 while (1) {
441 struct ipv6_tlv_tnl_enc_lim *tel;
442
443 /* No more room for encapsulation limit */
444 if (i + sizeof(*tel) > optlen)
445 break;
446
447 tel = (struct ipv6_tlv_tnl_enc_lim *)(skb->data + off + i);
448 /* return index of option if found and valid */
449 if (tel->type == IPV6_TLV_TNL_ENCAP_LIMIT &&
450 tel->length == 1)
451 return i + off - nhoff;
452 /* else jump to next option */
453 if (tel->type)
454 i += tel->length + 2;
455 else
456 i++;
457 }
458 }
459 nexthdr = next;
460 off += optlen;
461 }
462 return 0;
463}
464EXPORT_SYMBOL(ip6_tnl_parse_tlv_enc_lim);
465
466/**
467 * ip6_tnl_err - tunnel error handler
468 *
469 * Description:
470 * ip6_tnl_err() should handle errors in the tunnel according
471 * to the specifications in RFC 2473.
472 **/
473
474static int
475ip6_tnl_err(struct sk_buff *skb, __u8 ipproto, struct inet6_skb_parm *opt,
476 u8 *type, u8 *code, int *msg, __u32 *info, int offset)
477{
478 const struct ipv6hdr *ipv6h = (const struct ipv6hdr *) skb->data;
479 struct ip6_tnl *t;
480 int rel_msg = 0;
481 u8 rel_type = ICMPV6_DEST_UNREACH;
482 u8 rel_code = ICMPV6_ADDR_UNREACH;
483 __u32 rel_info = 0;
484 __u16 len;
485 int err = -ENOENT;
486
487 /* If the packet doesn't contain the original IPv6 header we are
488 in trouble since we might need the source address for further
489 processing of the error. */
490
491 rcu_read_lock();
492 if ((t = ip6_tnl_lookup(dev_net(skb->dev), &ipv6h->daddr,
493 &ipv6h->saddr)) == NULL)
494 goto out;
495
496 if (t->parms.proto != ipproto && t->parms.proto != 0)
497 goto out;
498
499 err = 0;
500
501 switch (*type) {
502 __u32 teli;
503 struct ipv6_tlv_tnl_enc_lim *tel;
504 __u32 mtu;
505 case ICMPV6_DEST_UNREACH:
506 net_warn_ratelimited("%s: Path to destination invalid or inactive!\n",
507 t->parms.name);
508 rel_msg = 1;
509 break;
510 case ICMPV6_TIME_EXCEED:
511 if ((*code) == ICMPV6_EXC_HOPLIMIT) {
512 net_warn_ratelimited("%s: Too small hop limit or routing loop in tunnel!\n",
513 t->parms.name);
514 rel_msg = 1;
515 }
516 break;
517 case ICMPV6_PARAMPROB:
518 teli = 0;
519 if ((*code) == ICMPV6_HDR_FIELD)
520 teli = ip6_tnl_parse_tlv_enc_lim(skb, skb->data);
521
522 if (teli && teli == *info - 2) {
523 tel = (struct ipv6_tlv_tnl_enc_lim *) &skb->data[teli];
524 if (tel->encap_limit == 0) {
525 net_warn_ratelimited("%s: Too small encapsulation limit or routing loop in tunnel!\n",
526 t->parms.name);
527 rel_msg = 1;
528 }
529 } else {
530 net_warn_ratelimited("%s: Recipient unable to parse tunneled packet!\n",
531 t->parms.name);
532 }
533 break;
534 case ICMPV6_PKT_TOOBIG:
535 mtu = *info - offset;
536 if (mtu < IPV6_MIN_MTU)
537 mtu = IPV6_MIN_MTU;
538 t->dev->mtu = mtu;
539
540 if ((len = sizeof (*ipv6h) + ntohs(ipv6h->payload_len)) > mtu) {
541 rel_type = ICMPV6_PKT_TOOBIG;
542 rel_code = 0;
543 rel_info = mtu;
544 rel_msg = 1;
545 }
546 break;
547 }
548
549 *type = rel_type;
550 *code = rel_code;
551 *info = rel_info;
552 *msg = rel_msg;
553
554out:
555 rcu_read_unlock();
556 return err;
557}
558
559static int
560ip4ip6_err(struct sk_buff *skb, struct inet6_skb_parm *opt,
561 u8 type, u8 code, int offset, __be32 info)
562{
563 int rel_msg = 0;
564 u8 rel_type = type;
565 u8 rel_code = code;
566 __u32 rel_info = ntohl(info);
567 int err;
568 struct sk_buff *skb2;
569 const struct iphdr *eiph;
570 struct rtable *rt;
571 struct flowi4 fl4;
572
573 err = ip6_tnl_err(skb, IPPROTO_IPIP, opt, &rel_type, &rel_code,
574 &rel_msg, &rel_info, offset);
575 if (err < 0)
576 return err;
577
578 if (rel_msg == 0)
579 return 0;
580
581 switch (rel_type) {
582 case ICMPV6_DEST_UNREACH:
583 if (rel_code != ICMPV6_ADDR_UNREACH)
584 return 0;
585 rel_type = ICMP_DEST_UNREACH;
586 rel_code = ICMP_HOST_UNREACH;
587 break;
588 case ICMPV6_PKT_TOOBIG:
589 if (rel_code != 0)
590 return 0;
591 rel_type = ICMP_DEST_UNREACH;
592 rel_code = ICMP_FRAG_NEEDED;
593 break;
594 case NDISC_REDIRECT:
595 rel_type = ICMP_REDIRECT;
596 rel_code = ICMP_REDIR_HOST;
597 default:
598 return 0;
599 }
600
601 if (!pskb_may_pull(skb, offset + sizeof(struct iphdr)))
602 return 0;
603
604 skb2 = skb_clone(skb, GFP_ATOMIC);
605 if (!skb2)
606 return 0;
607
608 skb_dst_drop(skb2);
609
610 skb_pull(skb2, offset);
611 skb_reset_network_header(skb2);
612 eiph = ip_hdr(skb2);
613
614 /* Try to guess incoming interface */
615 rt = ip_route_output_ports(dev_net(skb->dev), &fl4, NULL,
616 eiph->saddr, 0,
617 0, 0,
618 IPPROTO_IPIP, RT_TOS(eiph->tos), 0);
619 if (IS_ERR(rt))
620 goto out;
621
622 skb2->dev = rt->dst.dev;
623
624 /* route "incoming" packet */
625 if (rt->rt_flags & RTCF_LOCAL) {
626 ip_rt_put(rt);
627 rt = NULL;
628 rt = ip_route_output_ports(dev_net(skb->dev), &fl4, NULL,
629 eiph->daddr, eiph->saddr,
630 0, 0,
631 IPPROTO_IPIP,
632 RT_TOS(eiph->tos), 0);
633 if (IS_ERR(rt) ||
634 rt->dst.dev->type != ARPHRD_TUNNEL) {
635 if (!IS_ERR(rt))
636 ip_rt_put(rt);
637 goto out;
638 }
639 skb_dst_set(skb2, &rt->dst);
640 } else {
641 ip_rt_put(rt);
642 if (ip_route_input(skb2, eiph->daddr, eiph->saddr, eiph->tos,
643 skb2->dev) ||
644 skb_dst(skb2)->dev->type != ARPHRD_TUNNEL)
645 goto out;
646 }
647
648 /* change mtu on this route */
649 if (rel_type == ICMP_DEST_UNREACH && rel_code == ICMP_FRAG_NEEDED) {
650 if (rel_info > dst_mtu(skb_dst(skb2)))
651 goto out;
652
653 skb_dst(skb2)->ops->update_pmtu(skb_dst(skb2), NULL, skb2, rel_info);
654 }
655 if (rel_type == ICMP_REDIRECT)
656 skb_dst(skb2)->ops->redirect(skb_dst(skb2), NULL, skb2);
657
658 icmp_send(skb2, rel_type, rel_code, htonl(rel_info));
659
660out:
661 kfree_skb(skb2);
662 return 0;
663}
664
665static int
666ip6ip6_err(struct sk_buff *skb, struct inet6_skb_parm *opt,
667 u8 type, u8 code, int offset, __be32 info)
668{
669 int rel_msg = 0;
670 u8 rel_type = type;
671 u8 rel_code = code;
672 __u32 rel_info = ntohl(info);
673 int err;
674
675 err = ip6_tnl_err(skb, IPPROTO_IPV6, opt, &rel_type, &rel_code,
676 &rel_msg, &rel_info, offset);
677 if (err < 0)
678 return err;
679
680 if (rel_msg && pskb_may_pull(skb, offset + sizeof(struct ipv6hdr))) {
681 struct rt6_info *rt;
682 struct sk_buff *skb2 = skb_clone(skb, GFP_ATOMIC);
683
684 if (!skb2)
685 return 0;
686
687 skb_dst_drop(skb2);
688 skb_pull(skb2, offset);
689 skb_reset_network_header(skb2);
690
691 /* Try to guess incoming interface */
692 rt = rt6_lookup(dev_net(skb->dev), &ipv6_hdr(skb2)->saddr,
693 NULL, 0, 0);
694
695 if (rt && rt->dst.dev)
696 skb2->dev = rt->dst.dev;
697
698 icmpv6_send(skb2, rel_type, rel_code, rel_info);
699
700 ip6_rt_put(rt);
701
702 kfree_skb(skb2);
703 }
704
705 return 0;
706}
707
708static int ip4ip6_dscp_ecn_decapsulate(const struct ip6_tnl *t,
709 const struct ipv6hdr *ipv6h,
710 struct sk_buff *skb)
711{
712 __u8 dsfield = ipv6_get_dsfield(ipv6h) & ~INET_ECN_MASK;
713
714 if (t->parms.flags & IP6_TNL_F_RCV_DSCP_COPY)
715 ipv4_change_dsfield(ip_hdr(skb), INET_ECN_MASK, dsfield);
716
717 return IP6_ECN_decapsulate(ipv6h, skb);
718}
719
720static int ip6ip6_dscp_ecn_decapsulate(const struct ip6_tnl *t,
721 const struct ipv6hdr *ipv6h,
722 struct sk_buff *skb)
723{
724 if (t->parms.flags & IP6_TNL_F_RCV_DSCP_COPY)
725 ipv6_copy_dscp(ipv6_get_dsfield(ipv6h), ipv6_hdr(skb));
726
727 return IP6_ECN_decapsulate(ipv6h, skb);
728}
729
730__u32 ip6_tnl_get_cap(struct ip6_tnl *t,
731 const struct in6_addr *laddr,
732 const struct in6_addr *raddr)
733{
734 struct __ip6_tnl_parm *p = &t->parms;
735 int ltype = ipv6_addr_type(laddr);
736 int rtype = ipv6_addr_type(raddr);
737 __u32 flags = 0;
738
739 if (ltype == IPV6_ADDR_ANY || rtype == IPV6_ADDR_ANY) {
740 flags = IP6_TNL_F_CAP_PER_PACKET;
741 } else if (ltype & (IPV6_ADDR_UNICAST|IPV6_ADDR_MULTICAST) &&
742 rtype & (IPV6_ADDR_UNICAST|IPV6_ADDR_MULTICAST) &&
743 !((ltype|rtype) & IPV6_ADDR_LOOPBACK) &&
744 (!((ltype|rtype) & IPV6_ADDR_LINKLOCAL) || p->link)) {
745 if (ltype&IPV6_ADDR_UNICAST)
746 flags |= IP6_TNL_F_CAP_XMIT;
747 if (rtype&IPV6_ADDR_UNICAST)
748 flags |= IP6_TNL_F_CAP_RCV;
749 }
750 return flags;
751}
752EXPORT_SYMBOL(ip6_tnl_get_cap);
753
754/* called with rcu_read_lock() */
755int ip6_tnl_rcv_ctl(struct ip6_tnl *t,
756 const struct in6_addr *laddr,
757 const struct in6_addr *raddr)
758{
759 struct __ip6_tnl_parm *p = &t->parms;
760 int ret = 0;
761 struct net *net = dev_net(t->dev);
762
763 if ((p->flags & IP6_TNL_F_CAP_RCV) ||
764 ((p->flags & IP6_TNL_F_CAP_PER_PACKET) &&
765 (ip6_tnl_get_cap(t, laddr, raddr) & IP6_TNL_F_CAP_RCV))) {
766 struct net_device *ldev = NULL;
767
768 if (p->link)
769 ldev = dev_get_by_index_rcu(net, p->link);
770
771 if ((ipv6_addr_is_multicast(laddr) ||
772 likely(ipv6_chk_addr(net, laddr, ldev, 0))) &&
773 likely(!ipv6_chk_addr(net, raddr, NULL, 0)))
774 ret = 1;
775 }
776 return ret;
777}
778EXPORT_SYMBOL_GPL(ip6_tnl_rcv_ctl);
779
780/**
781 * ip6_tnl_rcv - decapsulate IPv6 packet and retransmit it locally
782 * @skb: received socket buffer
783 * @protocol: ethernet protocol ID
784 * @dscp_ecn_decapsulate: the function to decapsulate DSCP code and ECN
785 *
786 * Return: 0
787 **/
788
789static int ip6_tnl_rcv(struct sk_buff *skb, __u16 protocol,
790 __u8 ipproto,
791 int (*dscp_ecn_decapsulate)(const struct ip6_tnl *t,
792 const struct ipv6hdr *ipv6h,
793 struct sk_buff *skb))
794{
795 struct ip6_tnl *t;
796 const struct ipv6hdr *ipv6h = ipv6_hdr(skb);
797 int err;
798
799 rcu_read_lock();
800
801 if ((t = ip6_tnl_lookup(dev_net(skb->dev), &ipv6h->saddr,
802 &ipv6h->daddr)) != NULL) {
803 struct pcpu_tstats *tstats;
804
805 if (t->parms.proto != ipproto && t->parms.proto != 0) {
806 rcu_read_unlock();
807 goto discard;
808 }
809
810 if (!xfrm6_policy_check(NULL, XFRM_POLICY_IN, skb)) {
811 rcu_read_unlock();
812 goto discard;
813 }
814
815 if (!ip6_tnl_rcv_ctl(t, &ipv6h->daddr, &ipv6h->saddr)) {
816 t->dev->stats.rx_dropped++;
817 rcu_read_unlock();
818 goto discard;
819 }
820 secpath_reset(skb);
821 skb->mac_header = skb->network_header;
822 skb_reset_network_header(skb);
823 skb->protocol = htons(protocol);
824 skb->pkt_type = PACKET_HOST;
825 memset(skb->cb, 0, sizeof(struct inet6_skb_parm));
826
827 __skb_tunnel_rx(skb, t->dev);
828
829 err = dscp_ecn_decapsulate(t, ipv6h, skb);
830 if (unlikely(err)) {
831 if (log_ecn_error)
832 net_info_ratelimited("non-ECT from %pI6 with dsfield=%#x\n",
833 &ipv6h->saddr,
834 ipv6_get_dsfield(ipv6h));
835 if (err > 1) {
836 ++t->dev->stats.rx_frame_errors;
837 ++t->dev->stats.rx_errors;
838 rcu_read_unlock();
839 goto discard;
840 }
841 }
842
843 tstats = this_cpu_ptr(t->dev->tstats);
844 u64_stats_update_begin(&tstats->syncp);
845 tstats->rx_packets++;
846 tstats->rx_bytes += skb->len;
847 u64_stats_update_end(&tstats->syncp);
848
849 netif_rx(skb);
850
851 rcu_read_unlock();
852 return 0;
853 }
854 rcu_read_unlock();
855 return 1;
856
857discard:
858 kfree_skb(skb);
859 return 0;
860}
861
862static int ip4ip6_rcv(struct sk_buff *skb)
863{
864 return ip6_tnl_rcv(skb, ETH_P_IP, IPPROTO_IPIP,
865 ip4ip6_dscp_ecn_decapsulate);
866}
867
868static int ip6ip6_rcv(struct sk_buff *skb)
869{
870 return ip6_tnl_rcv(skb, ETH_P_IPV6, IPPROTO_IPV6,
871 ip6ip6_dscp_ecn_decapsulate);
872}
873
874struct ipv6_tel_txoption {
875 struct ipv6_txoptions ops;
876 __u8 dst_opt[8];
877};
878
879static void init_tel_txopt(struct ipv6_tel_txoption *opt, __u8 encap_limit)
880{
881 memset(opt, 0, sizeof(struct ipv6_tel_txoption));
882
883 opt->dst_opt[2] = IPV6_TLV_TNL_ENCAP_LIMIT;
884 opt->dst_opt[3] = 1;
885 opt->dst_opt[4] = encap_limit;
886 opt->dst_opt[5] = IPV6_TLV_PADN;
887 opt->dst_opt[6] = 1;
888
889 opt->ops.dst0opt = (struct ipv6_opt_hdr *) opt->dst_opt;
890 opt->ops.opt_nflen = 8;
891}
892
893/**
894 * ip6_tnl_addr_conflict - compare packet addresses to tunnel's own
895 * @t: the outgoing tunnel device
896 * @hdr: IPv6 header from the incoming packet
897 *
898 * Description:
899 * Avoid trivial tunneling loop by checking that tunnel exit-point
900 * doesn't match source of incoming packet.
901 *
902 * Return:
903 * 1 if conflict,
904 * 0 else
905 **/
906
907static inline bool
908ip6_tnl_addr_conflict(const struct ip6_tnl *t, const struct ipv6hdr *hdr)
909{
910 return ipv6_addr_equal(&t->parms.raddr, &hdr->saddr);
911}
912
913int ip6_tnl_xmit_ctl(struct ip6_tnl *t)
914{
915 struct __ip6_tnl_parm *p = &t->parms;
916 int ret = 0;
917 struct net *net = dev_net(t->dev);
918
919 if (p->flags & IP6_TNL_F_CAP_XMIT) {
920 struct net_device *ldev = NULL;
921
922 rcu_read_lock();
923 if (p->link)
924 ldev = dev_get_by_index_rcu(net, p->link);
925
926 if (unlikely(!ipv6_chk_addr(net, &p->laddr, ldev, 0)))
927 pr_warn("%s xmit: Local address not yet configured!\n",
928 p->name);
929 else if (!ipv6_addr_is_multicast(&p->raddr) &&
930 unlikely(ipv6_chk_addr(net, &p->raddr, NULL, 0)))
931 pr_warn("%s xmit: Routing loop! Remote address found on this node!\n",
932 p->name);
933 else
934 ret = 1;
935 rcu_read_unlock();
936 }
937 return ret;
938}
939EXPORT_SYMBOL_GPL(ip6_tnl_xmit_ctl);
940
941/**
942 * ip6_tnl_xmit2 - encapsulate packet and send
943 * @skb: the outgoing socket buffer
944 * @dev: the outgoing tunnel device
945 * @dsfield: dscp code for outer header
946 * @fl: flow of tunneled packet
947 * @encap_limit: encapsulation limit
948 * @pmtu: Path MTU is stored if packet is too big
949 *
950 * Description:
951 * Build new header and do some sanity checks on the packet before sending
952 * it.
953 *
954 * Return:
955 * 0 on success
956 * -1 fail
957 * %-EMSGSIZE message too big. return mtu in this case.
958 **/
959
960static int ip6_tnl_xmit2(struct sk_buff *skb,
961 struct net_device *dev,
962 __u8 dsfield,
963 struct flowi6 *fl6,
964 int encap_limit,
965 __u32 *pmtu)
966{
967 struct net *net = dev_net(dev);
968 struct ip6_tnl *t = netdev_priv(dev);
969 struct net_device_stats *stats = &t->dev->stats;
970 struct ipv6hdr *ipv6h = ipv6_hdr(skb);
971 struct ipv6_tel_txoption opt;
972 struct dst_entry *dst = NULL, *ndst = NULL;
973 struct net_device *tdev;
974 bool use_cache = false;
975 int mtu;
976 unsigned int max_headroom = sizeof(struct ipv6hdr);
977 u8 proto;
978 int err = -1;
979
980 if (!(t->parms.flags &
981 (IP6_TNL_F_USE_ORIG_TCLASS | IP6_TNL_F_USE_ORIG_FWMARK))) {
982 /* enable the cache only only if the routing decision does
983 * not depend on the current inner header value
984 */
985 use_cache = true;
986 }
987
988 if (use_cache)
989 dst = ip6_tnl_dst_check(t);
990 if (!dst) {
991 ndst = ip6_route_output(net, NULL, fl6);
992
993 if (ndst->error)
994 goto tx_err_link_failure;
995 ndst = xfrm_lookup(net, ndst, flowi6_to_flowi(fl6), NULL, 0);
996 if (IS_ERR(ndst)) {
997 err = PTR_ERR(ndst);
998 ndst = NULL;
999 goto tx_err_link_failure;
1000 }
1001 dst = ndst;
1002 }
1003
1004 tdev = dst->dev;
1005
1006 if (tdev == dev) {
1007 stats->collisions++;
1008 net_warn_ratelimited("%s: Local routing loop detected!\n",
1009 t->parms.name);
1010 goto tx_err_dst_release;
1011 }
1012 mtu = dst_mtu(dst) - sizeof (*ipv6h);
1013 if (encap_limit >= 0) {
1014 max_headroom += 8;
1015 mtu -= 8;
1016 }
1017 if (mtu < IPV6_MIN_MTU)
1018 mtu = IPV6_MIN_MTU;
1019 if (skb_dst(skb))
1020 skb_dst(skb)->ops->update_pmtu(skb_dst(skb), NULL, skb, mtu);
1021 if (skb->len > mtu) {
1022 *pmtu = mtu;
1023 err = -EMSGSIZE;
1024 goto tx_err_dst_release;
1025 }
1026
1027 /*
1028 * Okay, now see if we can stuff it in the buffer as-is.
1029 */
1030 max_headroom += LL_RESERVED_SPACE(tdev);
1031
1032 if (skb_headroom(skb) < max_headroom || skb_shared(skb) ||
1033 (skb_cloned(skb) && !skb_clone_writable(skb, 0))) {
1034 struct sk_buff *new_skb;
1035
1036 if (!(new_skb = skb_realloc_headroom(skb, max_headroom)))
1037 goto tx_err_dst_release;
1038
1039 if (skb->sk)
1040 skb_set_owner_w(new_skb, skb->sk);
1041 consume_skb(skb);
1042 skb = new_skb;
1043 }
1044 skb_dst_drop(skb);
1045 if (!use_cache) {
1046 skb_dst_set(skb, dst);
1047 ndst = NULL;
1048 } else {
1049 skb_dst_set_noref(skb, dst);
1050 }
1051 skb->transport_header = skb->network_header;
1052
1053 proto = fl6->flowi6_proto;
1054 if (encap_limit >= 0) {
1055 init_tel_txopt(&opt, encap_limit);
1056 ipv6_push_nfrag_opts(skb, &opt.ops, &proto, NULL);
1057 }
1058 skb_push(skb, sizeof(struct ipv6hdr));
1059 skb_reset_network_header(skb);
1060 ipv6h = ipv6_hdr(skb);
1061 ip6_flow_hdr(ipv6h, INET_ECN_encapsulate(0, dsfield), fl6->flowlabel);
1062 ipv6h->hop_limit = t->parms.hop_limit;
1063 ipv6h->nexthdr = proto;
1064 ipv6h->saddr = fl6->saddr;
1065 ipv6h->daddr = fl6->daddr;
1066 ip6tunnel_xmit(skb, dev);
1067 if (ndst)
1068 ip6_tnl_dst_store(t, ndst);
1069 return 0;
1070tx_err_link_failure:
1071 stats->tx_carrier_errors++;
1072 dst_link_failure(skb);
1073tx_err_dst_release:
1074 dst_release(ndst);
1075 return err;
1076}
1077
1078static inline int
1079ip4ip6_tnl_xmit(struct sk_buff *skb, struct net_device *dev)
1080{
1081 struct ip6_tnl *t = netdev_priv(dev);
1082 const struct iphdr *iph = ip_hdr(skb);
1083 int encap_limit = -1;
1084 struct flowi6 fl6;
1085 __u8 dsfield;
1086 __u32 mtu;
1087 int err;
1088
1089 if ((t->parms.proto != IPPROTO_IPIP && t->parms.proto != 0) ||
1090 !ip6_tnl_xmit_ctl(t))
1091 return -1;
1092
1093 if (!(t->parms.flags & IP6_TNL_F_IGN_ENCAP_LIMIT))
1094 encap_limit = t->parms.encap_limit;
1095
1096 memcpy(&fl6, &t->fl.u.ip6, sizeof (fl6));
1097 fl6.flowi6_proto = IPPROTO_IPIP;
1098
1099 dsfield = ipv4_get_dsfield(iph);
1100
1101 if (t->parms.flags & IP6_TNL_F_USE_ORIG_TCLASS)
1102 fl6.flowlabel |= htonl((__u32)iph->tos << IPV6_TCLASS_SHIFT)
1103 & IPV6_TCLASS_MASK;
1104 if (t->parms.flags & IP6_TNL_F_USE_ORIG_FWMARK)
1105 fl6.flowi6_mark = skb->mark;
1106
1107 err = ip6_tnl_xmit2(skb, dev, dsfield, &fl6, encap_limit, &mtu);
1108 if (err != 0) {
1109 /* XXX: send ICMP error even if DF is not set. */
1110 if (err == -EMSGSIZE)
1111 icmp_send(skb, ICMP_DEST_UNREACH, ICMP_FRAG_NEEDED,
1112 htonl(mtu));
1113 return -1;
1114 }
1115
1116 return 0;
1117}
1118
1119static inline int
1120ip6ip6_tnl_xmit(struct sk_buff *skb, struct net_device *dev)
1121{
1122 struct ip6_tnl *t = netdev_priv(dev);
1123 struct ipv6hdr *ipv6h = ipv6_hdr(skb);
1124 int encap_limit = -1;
1125 __u16 offset;
1126 struct flowi6 fl6;
1127 __u8 dsfield;
1128 __u32 mtu;
1129 int err;
1130
1131 if ((t->parms.proto != IPPROTO_IPV6 && t->parms.proto != 0) ||
1132 !ip6_tnl_xmit_ctl(t) || ip6_tnl_addr_conflict(t, ipv6h))
1133 return -1;
1134
1135 offset = ip6_tnl_parse_tlv_enc_lim(skb, skb_network_header(skb));
1136 if (offset > 0) {
1137 struct ipv6_tlv_tnl_enc_lim *tel;
1138 tel = (struct ipv6_tlv_tnl_enc_lim *)&skb_network_header(skb)[offset];
1139 if (tel->encap_limit == 0) {
1140 icmpv6_send(skb, ICMPV6_PARAMPROB,
1141 ICMPV6_HDR_FIELD, offset + 2);
1142 return -1;
1143 }
1144 encap_limit = tel->encap_limit - 1;
1145 } else if (!(t->parms.flags & IP6_TNL_F_IGN_ENCAP_LIMIT))
1146 encap_limit = t->parms.encap_limit;
1147
1148 memcpy(&fl6, &t->fl.u.ip6, sizeof (fl6));
1149 fl6.flowi6_proto = IPPROTO_IPV6;
1150
1151 dsfield = ipv6_get_dsfield(ipv6h);
1152 if (t->parms.flags & IP6_TNL_F_USE_ORIG_TCLASS)
1153 fl6.flowlabel |= (*(__be32 *) ipv6h & IPV6_TCLASS_MASK);
1154 if (t->parms.flags & IP6_TNL_F_USE_ORIG_FLOWLABEL)
1155 fl6.flowlabel |= (*(__be32 *) ipv6h & IPV6_FLOWLABEL_MASK);
1156 if (t->parms.flags & IP6_TNL_F_USE_ORIG_FWMARK)
1157 fl6.flowi6_mark = skb->mark;
1158
1159 err = ip6_tnl_xmit2(skb, dev, dsfield, &fl6, encap_limit, &mtu);
1160 if (err != 0) {
1161 if (err == -EMSGSIZE)
1162 icmpv6_send(skb, ICMPV6_PKT_TOOBIG, 0, mtu);
1163 return -1;
1164 }
1165
1166 return 0;
1167}
1168
1169static netdev_tx_t
1170ip6_tnl_xmit(struct sk_buff *skb, struct net_device *dev)
1171{
1172 struct ip6_tnl *t = netdev_priv(dev);
1173 struct net_device_stats *stats = &t->dev->stats;
1174 int ret;
1175
1176 switch (skb->protocol) {
1177 case htons(ETH_P_IP):
1178 ret = ip4ip6_tnl_xmit(skb, dev);
1179 break;
1180 case htons(ETH_P_IPV6):
1181 ret = ip6ip6_tnl_xmit(skb, dev);
1182 break;
1183 default:
1184 goto tx_err;
1185 }
1186
1187 if (ret < 0)
1188 goto tx_err;
1189
1190 return NETDEV_TX_OK;
1191
1192tx_err:
1193 stats->tx_errors++;
1194 stats->tx_dropped++;
1195 kfree_skb(skb);
1196 return NETDEV_TX_OK;
1197}
1198
1199static void ip6_tnl_link_config(struct ip6_tnl *t)
1200{
1201 struct net_device *dev = t->dev;
1202 struct __ip6_tnl_parm *p = &t->parms;
1203 struct flowi6 *fl6 = &t->fl.u.ip6;
1204
1205 memcpy(dev->dev_addr, &p->laddr, sizeof(struct in6_addr));
1206 memcpy(dev->broadcast, &p->raddr, sizeof(struct in6_addr));
1207
1208 /* Set up flowi template */
1209 fl6->saddr = p->laddr;
1210 fl6->daddr = p->raddr;
1211 fl6->flowi6_oif = p->link;
1212 fl6->flowlabel = 0;
1213
1214 if (!(p->flags&IP6_TNL_F_USE_ORIG_TCLASS))
1215 fl6->flowlabel |= IPV6_TCLASS_MASK & p->flowinfo;
1216 if (!(p->flags&IP6_TNL_F_USE_ORIG_FLOWLABEL))
1217 fl6->flowlabel |= IPV6_FLOWLABEL_MASK & p->flowinfo;
1218
1219 p->flags &= ~(IP6_TNL_F_CAP_XMIT|IP6_TNL_F_CAP_RCV|IP6_TNL_F_CAP_PER_PACKET);
1220 p->flags |= ip6_tnl_get_cap(t, &p->laddr, &p->raddr);
1221
1222 if (p->flags&IP6_TNL_F_CAP_XMIT && p->flags&IP6_TNL_F_CAP_RCV)
1223 dev->flags |= IFF_POINTOPOINT;
1224 else
1225 dev->flags &= ~IFF_POINTOPOINT;
1226
1227 dev->iflink = p->link;
1228
1229 if (p->flags & IP6_TNL_F_CAP_XMIT) {
1230 int strict = (ipv6_addr_type(&p->raddr) &
1231 (IPV6_ADDR_MULTICAST|IPV6_ADDR_LINKLOCAL));
1232
1233 struct rt6_info *rt = rt6_lookup(dev_net(dev),
1234 &p->raddr, &p->laddr,
1235 p->link, strict);
1236
1237 if (rt == NULL)
1238 return;
1239
1240 if (rt->dst.dev) {
1241 dev->hard_header_len = rt->dst.dev->hard_header_len +
1242 sizeof (struct ipv6hdr);
1243
1244 dev->mtu = rt->dst.dev->mtu - sizeof (struct ipv6hdr);
1245 if (!(t->parms.flags & IP6_TNL_F_IGN_ENCAP_LIMIT))
1246 dev->mtu-=8;
1247
1248 if (dev->mtu < IPV6_MIN_MTU)
1249 dev->mtu = IPV6_MIN_MTU;
1250 }
1251 ip6_rt_put(rt);
1252 }
1253}
1254
1255/**
1256 * ip6_tnl_change - update the tunnel parameters
1257 * @t: tunnel to be changed
1258 * @p: tunnel configuration parameters
1259 *
1260 * Description:
1261 * ip6_tnl_change() updates the tunnel parameters
1262 **/
1263
1264static int
1265ip6_tnl_change(struct ip6_tnl *t, const struct __ip6_tnl_parm *p)
1266{
1267 t->parms.laddr = p->laddr;
1268 t->parms.raddr = p->raddr;
1269 t->parms.flags = p->flags;
1270 t->parms.hop_limit = p->hop_limit;
1271 t->parms.encap_limit = p->encap_limit;
1272 t->parms.flowinfo = p->flowinfo;
1273 t->parms.link = p->link;
1274 t->parms.proto = p->proto;
1275 ip6_tnl_dst_reset(t);
1276 ip6_tnl_link_config(t);
1277 return 0;
1278}
1279
1280static int ip6_tnl_update(struct ip6_tnl *t, struct __ip6_tnl_parm *p)
1281{
1282 struct net *net = dev_net(t->dev);
1283 struct ip6_tnl_net *ip6n = net_generic(net, ip6_tnl_net_id);
1284 int err;
1285
1286 ip6_tnl_unlink(ip6n, t);
1287 synchronize_net();
1288 err = ip6_tnl_change(t, p);
1289 ip6_tnl_link(ip6n, t);
1290 netdev_state_change(t->dev);
1291 return err;
1292}
1293
1294static void
1295ip6_tnl_parm_from_user(struct __ip6_tnl_parm *p, const struct ip6_tnl_parm *u)
1296{
1297 p->laddr = u->laddr;
1298 p->raddr = u->raddr;
1299 p->flags = u->flags;
1300 p->hop_limit = u->hop_limit;
1301 p->encap_limit = u->encap_limit;
1302 p->flowinfo = u->flowinfo;
1303 p->link = u->link;
1304 p->proto = u->proto;
1305 memcpy(p->name, u->name, sizeof(u->name));
1306}
1307
1308static void
1309ip6_tnl_parm_to_user(struct ip6_tnl_parm *u, const struct __ip6_tnl_parm *p)
1310{
1311 u->laddr = p->laddr;
1312 u->raddr = p->raddr;
1313 u->flags = p->flags;
1314 u->hop_limit = p->hop_limit;
1315 u->encap_limit = p->encap_limit;
1316 u->flowinfo = p->flowinfo;
1317 u->link = p->link;
1318 u->proto = p->proto;
1319 memcpy(u->name, p->name, sizeof(u->name));
1320}
1321
1322/**
1323 * ip6_tnl_ioctl - configure ipv6 tunnels from userspace
1324 * @dev: virtual device associated with tunnel
1325 * @ifr: parameters passed from userspace
1326 * @cmd: command to be performed
1327 *
1328 * Description:
1329 * ip6_tnl_ioctl() is used for managing IPv6 tunnels
1330 * from userspace.
1331 *
1332 * The possible commands are the following:
1333 * %SIOCGETTUNNEL: get tunnel parameters for device
1334 * %SIOCADDTUNNEL: add tunnel matching given tunnel parameters
1335 * %SIOCCHGTUNNEL: change tunnel parameters to those given
1336 * %SIOCDELTUNNEL: delete tunnel
1337 *
1338 * The fallback device "ip6tnl0", created during module
1339 * initialization, can be used for creating other tunnel devices.
1340 *
1341 * Return:
1342 * 0 on success,
1343 * %-EFAULT if unable to copy data to or from userspace,
1344 * %-EPERM if current process hasn't %CAP_NET_ADMIN set
1345 * %-EINVAL if passed tunnel parameters are invalid,
1346 * %-EEXIST if changing a tunnel's parameters would cause a conflict
1347 * %-ENODEV if attempting to change or delete a nonexisting device
1348 **/
1349
1350static int
1351ip6_tnl_ioctl(struct net_device *dev, struct ifreq *ifr, int cmd)
1352{
1353 int err = 0;
1354 struct ip6_tnl_parm p;
1355 struct __ip6_tnl_parm p1;
1356 struct ip6_tnl *t = NULL;
1357 struct net *net = dev_net(dev);
1358 struct ip6_tnl_net *ip6n = net_generic(net, ip6_tnl_net_id);
1359
1360 switch (cmd) {
1361 case SIOCGETTUNNEL:
1362 if (dev == ip6n->fb_tnl_dev) {
1363 if (copy_from_user(&p, ifr->ifr_ifru.ifru_data, sizeof (p))) {
1364 err = -EFAULT;
1365 break;
1366 }
1367 ip6_tnl_parm_from_user(&p1, &p);
1368 t = ip6_tnl_locate(net, &p1, 0);
1369 } else {
1370 memset(&p, 0, sizeof(p));
1371 }
1372 if (t == NULL)
1373 t = netdev_priv(dev);
1374 ip6_tnl_parm_to_user(&p, &t->parms);
1375 if (copy_to_user(ifr->ifr_ifru.ifru_data, &p, sizeof (p))) {
1376 err = -EFAULT;
1377 }
1378 break;
1379 case SIOCADDTUNNEL:
1380 case SIOCCHGTUNNEL:
1381 err = -EPERM;
1382 if (!ns_capable(net->user_ns, CAP_NET_ADMIN))
1383 break;
1384 err = -EFAULT;
1385 if (copy_from_user(&p, ifr->ifr_ifru.ifru_data, sizeof (p)))
1386 break;
1387 err = -EINVAL;
1388 if (p.proto != IPPROTO_IPV6 && p.proto != IPPROTO_IPIP &&
1389 p.proto != 0)
1390 break;
1391 ip6_tnl_parm_from_user(&p1, &p);
1392 t = ip6_tnl_locate(net, &p1, cmd == SIOCADDTUNNEL);
1393 if (dev != ip6n->fb_tnl_dev && cmd == SIOCCHGTUNNEL) {
1394 if (t != NULL) {
1395 if (t->dev != dev) {
1396 err = -EEXIST;
1397 break;
1398 }
1399 } else
1400 t = netdev_priv(dev);
1401
1402 err = ip6_tnl_update(t, &p1);
1403 }
1404 if (t) {
1405 err = 0;
1406 ip6_tnl_parm_to_user(&p, &t->parms);
1407 if (copy_to_user(ifr->ifr_ifru.ifru_data, &p, sizeof(p)))
1408 err = -EFAULT;
1409
1410 } else
1411 err = (cmd == SIOCADDTUNNEL ? -ENOBUFS : -ENOENT);
1412 break;
1413 case SIOCDELTUNNEL:
1414 err = -EPERM;
1415 if (!ns_capable(net->user_ns, CAP_NET_ADMIN))
1416 break;
1417
1418 if (dev == ip6n->fb_tnl_dev) {
1419 err = -EFAULT;
1420 if (copy_from_user(&p, ifr->ifr_ifru.ifru_data, sizeof (p)))
1421 break;
1422 err = -ENOENT;
1423 ip6_tnl_parm_from_user(&p1, &p);
1424 t = ip6_tnl_locate(net, &p1, 0);
1425 if (t == NULL)
1426 break;
1427 err = -EPERM;
1428 if (t->dev == ip6n->fb_tnl_dev)
1429 break;
1430 dev = t->dev;
1431 }
1432 err = 0;
1433 unregister_netdevice(dev);
1434 break;
1435 default:
1436 err = -EINVAL;
1437 }
1438 return err;
1439}
1440
1441/**
1442 * ip6_tnl_change_mtu - change mtu manually for tunnel device
1443 * @dev: virtual device associated with tunnel
1444 * @new_mtu: the new mtu
1445 *
1446 * Return:
1447 * 0 on success,
1448 * %-EINVAL if mtu too small
1449 **/
1450
1451static int
1452ip6_tnl_change_mtu(struct net_device *dev, int new_mtu)
1453{
1454 if (new_mtu < IPV6_MIN_MTU) {
1455 return -EINVAL;
1456 }
1457 dev->mtu = new_mtu;
1458 return 0;
1459}
1460
1461
1462static const struct net_device_ops ip6_tnl_netdev_ops = {
1463 .ndo_init = ip6_tnl_dev_init,
1464 .ndo_uninit = ip6_tnl_dev_uninit,
1465 .ndo_start_xmit = ip6_tnl_xmit,
1466 .ndo_do_ioctl = ip6_tnl_ioctl,
1467 .ndo_change_mtu = ip6_tnl_change_mtu,
1468 .ndo_get_stats = ip6_get_stats,
1469};
1470
1471
1472/**
1473 * ip6_tnl_dev_setup - setup virtual tunnel device
1474 * @dev: virtual device associated with tunnel
1475 *
1476 * Description:
1477 * Initialize function pointers and device parameters
1478 **/
1479
1480static void ip6_tnl_dev_setup(struct net_device *dev)
1481{
1482 struct ip6_tnl *t;
1483
1484 dev->netdev_ops = &ip6_tnl_netdev_ops;
1485 dev->destructor = ip6_dev_free;
1486
1487 dev->type = ARPHRD_TUNNEL6;
1488 dev->hard_header_len = LL_MAX_HEADER + sizeof (struct ipv6hdr);
1489 dev->mtu = ETH_DATA_LEN - sizeof (struct ipv6hdr);
1490 t = netdev_priv(dev);
1491 if (!(t->parms.flags & IP6_TNL_F_IGN_ENCAP_LIMIT))
1492 dev->mtu-=8;
1493 dev->flags |= IFF_NOARP;
1494 dev->addr_len = sizeof(struct in6_addr);
1495 dev->features |= NETIF_F_NETNS_LOCAL;
1496 dev->priv_flags &= ~IFF_XMIT_DST_RELEASE;
1497}
1498
1499
1500/**
1501 * ip6_tnl_dev_init_gen - general initializer for all tunnel devices
1502 * @dev: virtual device associated with tunnel
1503 **/
1504
1505static inline int
1506ip6_tnl_dev_init_gen(struct net_device *dev)
1507{
1508 struct ip6_tnl *t = netdev_priv(dev);
1509
1510 t->dev = dev;
1511 dev->tstats = alloc_percpu(struct pcpu_tstats);
1512 if (!dev->tstats)
1513 return -ENOMEM;
1514 return 0;
1515}
1516
1517/**
1518 * ip6_tnl_dev_init - initializer for all non fallback tunnel devices
1519 * @dev: virtual device associated with tunnel
1520 **/
1521
1522static int ip6_tnl_dev_init(struct net_device *dev)
1523{
1524 struct ip6_tnl *t = netdev_priv(dev);
1525 int err = ip6_tnl_dev_init_gen(dev);
1526
1527 if (err)
1528 return err;
1529 ip6_tnl_link_config(t);
1530 return 0;
1531}
1532
1533/**
1534 * ip6_fb_tnl_dev_init - initializer for fallback tunnel device
1535 * @dev: fallback device
1536 *
1537 * Return: 0
1538 **/
1539
1540static int __net_init ip6_fb_tnl_dev_init(struct net_device *dev)
1541{
1542 struct ip6_tnl *t = netdev_priv(dev);
1543 struct net *net = dev_net(dev);
1544 struct ip6_tnl_net *ip6n = net_generic(net, ip6_tnl_net_id);
1545
1546 t->parms.proto = IPPROTO_IPV6;
1547 dev_hold(dev);
1548
1549 rcu_assign_pointer(ip6n->tnls_wc[0], t);
1550 return 0;
1551}
1552
1553static int ip6_tnl_validate(struct nlattr *tb[], struct nlattr *data[])
1554{
1555 u8 proto;
1556
1557 if (!data || !data[IFLA_IPTUN_PROTO])
1558 return 0;
1559
1560 proto = nla_get_u8(data[IFLA_IPTUN_PROTO]);
1561 if (proto != IPPROTO_IPV6 &&
1562 proto != IPPROTO_IPIP &&
1563 proto != 0)
1564 return -EINVAL;
1565
1566 return 0;
1567}
1568
1569static void ip6_tnl_netlink_parms(struct nlattr *data[],
1570 struct __ip6_tnl_parm *parms)
1571{
1572 memset(parms, 0, sizeof(*parms));
1573
1574 if (!data)
1575 return;
1576
1577 if (data[IFLA_IPTUN_LINK])
1578 parms->link = nla_get_u32(data[IFLA_IPTUN_LINK]);
1579
1580 if (data[IFLA_IPTUN_LOCAL])
1581 nla_memcpy(&parms->laddr, data[IFLA_IPTUN_LOCAL],
1582 sizeof(struct in6_addr));
1583
1584 if (data[IFLA_IPTUN_REMOTE])
1585 nla_memcpy(&parms->raddr, data[IFLA_IPTUN_REMOTE],
1586 sizeof(struct in6_addr));
1587
1588 if (data[IFLA_IPTUN_TTL])
1589 parms->hop_limit = nla_get_u8(data[IFLA_IPTUN_TTL]);
1590
1591 if (data[IFLA_IPTUN_ENCAP_LIMIT])
1592 parms->encap_limit = nla_get_u8(data[IFLA_IPTUN_ENCAP_LIMIT]);
1593
1594 if (data[IFLA_IPTUN_FLOWINFO])
1595 parms->flowinfo = nla_get_be32(data[IFLA_IPTUN_FLOWINFO]);
1596
1597 if (data[IFLA_IPTUN_FLAGS])
1598 parms->flags = nla_get_u32(data[IFLA_IPTUN_FLAGS]);
1599
1600 if (data[IFLA_IPTUN_PROTO])
1601 parms->proto = nla_get_u8(data[IFLA_IPTUN_PROTO]);
1602}
1603
1604static int ip6_tnl_newlink(struct net *src_net, struct net_device *dev,
1605 struct nlattr *tb[], struct nlattr *data[])
1606{
1607 struct net *net = dev_net(dev);
1608 struct ip6_tnl *nt;
1609
1610 nt = netdev_priv(dev);
1611 ip6_tnl_netlink_parms(data, &nt->parms);
1612
1613 if (ip6_tnl_locate(net, &nt->parms, 0))
1614 return -EEXIST;
1615
1616 return ip6_tnl_create2(dev);
1617}
1618
1619static int ip6_tnl_changelink(struct net_device *dev, struct nlattr *tb[],
1620 struct nlattr *data[])
1621{
1622 struct ip6_tnl *t;
1623 struct __ip6_tnl_parm p;
1624 struct net *net = dev_net(dev);
1625 struct ip6_tnl_net *ip6n = net_generic(net, ip6_tnl_net_id);
1626
1627 if (dev == ip6n->fb_tnl_dev)
1628 return -EINVAL;
1629
1630 ip6_tnl_netlink_parms(data, &p);
1631
1632 t = ip6_tnl_locate(net, &p, 0);
1633
1634 if (t) {
1635 if (t->dev != dev)
1636 return -EEXIST;
1637 } else
1638 t = netdev_priv(dev);
1639
1640 return ip6_tnl_update(t, &p);
1641}
1642
1643static void ip6_tnl_dellink(struct net_device *dev, struct list_head *head)
1644{
1645 struct net *net = dev_net(dev);
1646 struct ip6_tnl_net *ip6n = net_generic(net, ip6_tnl_net_id);
1647
1648 if (dev != ip6n->fb_tnl_dev)
1649 unregister_netdevice_queue(dev, head);
1650}
1651
1652static size_t ip6_tnl_get_size(const struct net_device *dev)
1653{
1654 return
1655 /* IFLA_IPTUN_LINK */
1656 nla_total_size(4) +
1657 /* IFLA_IPTUN_LOCAL */
1658 nla_total_size(sizeof(struct in6_addr)) +
1659 /* IFLA_IPTUN_REMOTE */
1660 nla_total_size(sizeof(struct in6_addr)) +
1661 /* IFLA_IPTUN_TTL */
1662 nla_total_size(1) +
1663 /* IFLA_IPTUN_ENCAP_LIMIT */
1664 nla_total_size(1) +
1665 /* IFLA_IPTUN_FLOWINFO */
1666 nla_total_size(4) +
1667 /* IFLA_IPTUN_FLAGS */
1668 nla_total_size(4) +
1669 /* IFLA_IPTUN_PROTO */
1670 nla_total_size(1) +
1671 0;
1672}
1673
1674static int ip6_tnl_fill_info(struct sk_buff *skb, const struct net_device *dev)
1675{
1676 struct ip6_tnl *tunnel = netdev_priv(dev);
1677 struct __ip6_tnl_parm *parm = &tunnel->parms;
1678
1679 if (nla_put_u32(skb, IFLA_IPTUN_LINK, parm->link) ||
1680 nla_put(skb, IFLA_IPTUN_LOCAL, sizeof(struct in6_addr),
1681 &parm->laddr) ||
1682 nla_put(skb, IFLA_IPTUN_REMOTE, sizeof(struct in6_addr),
1683 &parm->raddr) ||
1684 nla_put_u8(skb, IFLA_IPTUN_TTL, parm->hop_limit) ||
1685 nla_put_u8(skb, IFLA_IPTUN_ENCAP_LIMIT, parm->encap_limit) ||
1686 nla_put_be32(skb, IFLA_IPTUN_FLOWINFO, parm->flowinfo) ||
1687 nla_put_u32(skb, IFLA_IPTUN_FLAGS, parm->flags) ||
1688 nla_put_u8(skb, IFLA_IPTUN_PROTO, parm->proto))
1689 goto nla_put_failure;
1690 return 0;
1691
1692nla_put_failure:
1693 return -EMSGSIZE;
1694}
1695
1696static const struct nla_policy ip6_tnl_policy[IFLA_IPTUN_MAX + 1] = {
1697 [IFLA_IPTUN_LINK] = { .type = NLA_U32 },
1698 [IFLA_IPTUN_LOCAL] = { .len = sizeof(struct in6_addr) },
1699 [IFLA_IPTUN_REMOTE] = { .len = sizeof(struct in6_addr) },
1700 [IFLA_IPTUN_TTL] = { .type = NLA_U8 },
1701 [IFLA_IPTUN_ENCAP_LIMIT] = { .type = NLA_U8 },
1702 [IFLA_IPTUN_FLOWINFO] = { .type = NLA_U32 },
1703 [IFLA_IPTUN_FLAGS] = { .type = NLA_U32 },
1704 [IFLA_IPTUN_PROTO] = { .type = NLA_U8 },
1705};
1706
1707static struct rtnl_link_ops ip6_link_ops __read_mostly = {
1708 .kind = "ip6tnl",
1709 .maxtype = IFLA_IPTUN_MAX,
1710 .policy = ip6_tnl_policy,
1711 .priv_size = sizeof(struct ip6_tnl),
1712 .setup = ip6_tnl_dev_setup,
1713 .validate = ip6_tnl_validate,
1714 .newlink = ip6_tnl_newlink,
1715 .changelink = ip6_tnl_changelink,
1716 .dellink = ip6_tnl_dellink,
1717 .get_size = ip6_tnl_get_size,
1718 .fill_info = ip6_tnl_fill_info,
1719};
1720
1721static struct xfrm6_tunnel ip4ip6_handler __read_mostly = {
1722 .handler = ip4ip6_rcv,
1723 .err_handler = ip4ip6_err,
1724 .priority = 1,
1725};
1726
1727static struct xfrm6_tunnel ip6ip6_handler __read_mostly = {
1728 .handler = ip6ip6_rcv,
1729 .err_handler = ip6ip6_err,
1730 .priority = 1,
1731};
1732
1733static void __net_exit ip6_tnl_destroy_tunnels(struct ip6_tnl_net *ip6n)
1734{
1735 int h;
1736 struct ip6_tnl *t;
1737 LIST_HEAD(list);
1738
1739 for (h = 0; h < HASH_SIZE; h++) {
1740 t = rtnl_dereference(ip6n->tnls_r_l[h]);
1741 while (t != NULL) {
1742 unregister_netdevice_queue(t->dev, &list);
1743 t = rtnl_dereference(t->next);
1744 }
1745 }
1746
1747 t = rtnl_dereference(ip6n->tnls_wc[0]);
1748 unregister_netdevice_queue(t->dev, &list);
1749 unregister_netdevice_many(&list);
1750}
1751
1752static int __net_init ip6_tnl_init_net(struct net *net)
1753{
1754 struct ip6_tnl_net *ip6n = net_generic(net, ip6_tnl_net_id);
1755 struct ip6_tnl *t = NULL;
1756 int err;
1757
1758 ip6n->tnls[0] = ip6n->tnls_wc;
1759 ip6n->tnls[1] = ip6n->tnls_r_l;
1760
1761 err = -ENOMEM;
1762 ip6n->fb_tnl_dev = alloc_netdev(sizeof(struct ip6_tnl), "ip6tnl0",
1763 ip6_tnl_dev_setup);
1764
1765 if (!ip6n->fb_tnl_dev)
1766 goto err_alloc_dev;
1767 dev_net_set(ip6n->fb_tnl_dev, net);
1768 ip6n->fb_tnl_dev->rtnl_link_ops = &ip6_link_ops;
1769
1770 err = ip6_fb_tnl_dev_init(ip6n->fb_tnl_dev);
1771 if (err < 0)
1772 goto err_register;
1773
1774 err = register_netdev(ip6n->fb_tnl_dev);
1775 if (err < 0)
1776 goto err_register;
1777
1778 t = netdev_priv(ip6n->fb_tnl_dev);
1779
1780 strcpy(t->parms.name, ip6n->fb_tnl_dev->name);
1781 return 0;
1782
1783err_register:
1784 ip6_dev_free(ip6n->fb_tnl_dev);
1785err_alloc_dev:
1786 return err;
1787}
1788
1789static void __net_exit ip6_tnl_exit_net(struct net *net)
1790{
1791 struct ip6_tnl_net *ip6n = net_generic(net, ip6_tnl_net_id);
1792
1793 rtnl_lock();
1794 ip6_tnl_destroy_tunnels(ip6n);
1795 rtnl_unlock();
1796}
1797
1798static struct pernet_operations ip6_tnl_net_ops = {
1799 .init = ip6_tnl_init_net,
1800 .exit = ip6_tnl_exit_net,
1801 .id = &ip6_tnl_net_id,
1802 .size = sizeof(struct ip6_tnl_net),
1803};
1804
1805/**
1806 * ip6_tunnel_init - register protocol and reserve needed resources
1807 *
1808 * Return: 0 on success
1809 **/
1810
1811static int __init ip6_tunnel_init(void)
1812{
1813 int err;
1814
1815 err = register_pernet_device(&ip6_tnl_net_ops);
1816 if (err < 0)
1817 goto out_pernet;
1818
1819 err = xfrm6_tunnel_register(&ip4ip6_handler, AF_INET);
1820 if (err < 0) {
1821 pr_err("%s: can't register ip4ip6\n", __func__);
1822 goto out_ip4ip6;
1823 }
1824
1825 err = xfrm6_tunnel_register(&ip6ip6_handler, AF_INET6);
1826 if (err < 0) {
1827 pr_err("%s: can't register ip6ip6\n", __func__);
1828 goto out_ip6ip6;
1829 }
1830 err = rtnl_link_register(&ip6_link_ops);
1831 if (err < 0)
1832 goto rtnl_link_failed;
1833
1834 return 0;
1835
1836rtnl_link_failed:
1837 xfrm6_tunnel_deregister(&ip6ip6_handler, AF_INET6);
1838out_ip6ip6:
1839 xfrm6_tunnel_deregister(&ip4ip6_handler, AF_INET);
1840out_ip4ip6:
1841 unregister_pernet_device(&ip6_tnl_net_ops);
1842out_pernet:
1843 return err;
1844}
1845
1846/**
1847 * ip6_tunnel_cleanup - free resources and unregister protocol
1848 **/
1849
1850static void __exit ip6_tunnel_cleanup(void)
1851{
1852 rtnl_link_unregister(&ip6_link_ops);
1853 if (xfrm6_tunnel_deregister(&ip4ip6_handler, AF_INET))
1854 pr_info("%s: can't deregister ip4ip6\n", __func__);
1855
1856 if (xfrm6_tunnel_deregister(&ip6ip6_handler, AF_INET6))
1857 pr_info("%s: can't deregister ip6ip6\n", __func__);
1858
1859 unregister_pernet_device(&ip6_tnl_net_ops);
1860}
1861
1862module_init(ip6_tunnel_init);
1863module_exit(ip6_tunnel_cleanup);
kernel source for telekom puls (alcatel/mediatek)