diff options
author | Linus Torvalds <torvalds@ppc970.osdl.org> | 2005-04-16 18:20:36 -0400 |
---|---|---|
committer | Linus Torvalds <torvalds@ppc970.osdl.org> | 2005-04-16 18:20:36 -0400 |
commit | 1da177e4c3f41524e886b7f1b8a0c1fc7321cac2 (patch) | |
tree | 0bba044c4ce775e45a88a51686b5d9f90697ea9d /net/ipv6/sit.c |
Linux-2.6.12-rc2v2.6.12-rc2
Initial git repository build. I'm not bothering with the full history,
even though we have it. We can create a separate "historical" git
archive of that later if we want to, and in the meantime it's about
3.2GB when imported into git - space that would just make the early
git days unnecessarily complicated, when we don't have a lot of good
infrastructure for it.
Let it rip!
Diffstat (limited to 'net/ipv6/sit.c')
-rw-r--r-- | net/ipv6/sit.c | 833 |
1 files changed, 833 insertions, 0 deletions
diff --git a/net/ipv6/sit.c b/net/ipv6/sit.c new file mode 100644 index 000000000000..b788f55e139b --- /dev/null +++ b/net/ipv6/sit.c | |||
@@ -0,0 +1,833 @@ | |||
1 | /* | ||
2 | * IPv6 over IPv4 tunnel device - Simple Internet Transition (SIT) | ||
3 | * Linux INET6 implementation | ||
4 | * | ||
5 | * Authors: | ||
6 | * Pedro Roque <roque@di.fc.ul.pt> | ||
7 | * Alexey Kuznetsov <kuznet@ms2.inr.ac.ru> | ||
8 | * | ||
9 | * $Id: sit.c,v 1.53 2001/09/25 05:09:53 davem Exp $ | ||
10 | * | ||
11 | * This program is free software; you can redistribute it and/or | ||
12 | * modify it under the terms of the GNU General Public License | ||
13 | * as published by the Free Software Foundation; either version | ||
14 | * 2 of the License, or (at your option) any later version. | ||
15 | * | ||
16 | * Changes: | ||
17 | * Roger Venning <r.venning@telstra.com>: 6to4 support | ||
18 | * Nate Thompson <nate@thebog.net>: 6to4 support | ||
19 | */ | ||
20 | |||
21 | #include <linux/config.h> | ||
22 | #include <linux/module.h> | ||
23 | #include <linux/errno.h> | ||
24 | #include <linux/types.h> | ||
25 | #include <linux/socket.h> | ||
26 | #include <linux/sockios.h> | ||
27 | #include <linux/sched.h> | ||
28 | #include <linux/net.h> | ||
29 | #include <linux/in6.h> | ||
30 | #include <linux/netdevice.h> | ||
31 | #include <linux/if_arp.h> | ||
32 | #include <linux/icmp.h> | ||
33 | #include <asm/uaccess.h> | ||
34 | #include <linux/init.h> | ||
35 | #include <linux/netfilter_ipv4.h> | ||
36 | |||
37 | #include <net/sock.h> | ||
38 | #include <net/snmp.h> | ||
39 | |||
40 | #include <net/ipv6.h> | ||
41 | #include <net/protocol.h> | ||
42 | #include <net/transp_v6.h> | ||
43 | #include <net/ip6_fib.h> | ||
44 | #include <net/ip6_route.h> | ||
45 | #include <net/ndisc.h> | ||
46 | #include <net/addrconf.h> | ||
47 | #include <net/ip.h> | ||
48 | #include <net/udp.h> | ||
49 | #include <net/icmp.h> | ||
50 | #include <net/ipip.h> | ||
51 | #include <net/inet_ecn.h> | ||
52 | #include <net/xfrm.h> | ||
53 | #include <net/dsfield.h> | ||
54 | |||
55 | /* | ||
56 | This version of net/ipv6/sit.c is cloned of net/ipv4/ip_gre.c | ||
57 | |||
58 | For comments look at net/ipv4/ip_gre.c --ANK | ||
59 | */ | ||
60 | |||
61 | #define HASH_SIZE 16 | ||
62 | #define HASH(addr) ((addr^(addr>>4))&0xF) | ||
63 | |||
64 | static int ipip6_fb_tunnel_init(struct net_device *dev); | ||
65 | static int ipip6_tunnel_init(struct net_device *dev); | ||
66 | static void ipip6_tunnel_setup(struct net_device *dev); | ||
67 | |||
68 | static struct net_device *ipip6_fb_tunnel_dev; | ||
69 | |||
70 | static struct ip_tunnel *tunnels_r_l[HASH_SIZE]; | ||
71 | static struct ip_tunnel *tunnels_r[HASH_SIZE]; | ||
72 | static struct ip_tunnel *tunnels_l[HASH_SIZE]; | ||
73 | static struct ip_tunnel *tunnels_wc[1]; | ||
74 | static struct ip_tunnel **tunnels[4] = { tunnels_wc, tunnels_l, tunnels_r, tunnels_r_l }; | ||
75 | |||
76 | static DEFINE_RWLOCK(ipip6_lock); | ||
77 | |||
78 | static struct ip_tunnel * ipip6_tunnel_lookup(u32 remote, u32 local) | ||
79 | { | ||
80 | unsigned h0 = HASH(remote); | ||
81 | unsigned h1 = HASH(local); | ||
82 | struct ip_tunnel *t; | ||
83 | |||
84 | for (t = tunnels_r_l[h0^h1]; t; t = t->next) { | ||
85 | if (local == t->parms.iph.saddr && | ||
86 | remote == t->parms.iph.daddr && (t->dev->flags&IFF_UP)) | ||
87 | return t; | ||
88 | } | ||
89 | for (t = tunnels_r[h0]; t; t = t->next) { | ||
90 | if (remote == t->parms.iph.daddr && (t->dev->flags&IFF_UP)) | ||
91 | return t; | ||
92 | } | ||
93 | for (t = tunnels_l[h1]; t; t = t->next) { | ||
94 | if (local == t->parms.iph.saddr && (t->dev->flags&IFF_UP)) | ||
95 | return t; | ||
96 | } | ||
97 | if ((t = tunnels_wc[0]) != NULL && (t->dev->flags&IFF_UP)) | ||
98 | return t; | ||
99 | return NULL; | ||
100 | } | ||
101 | |||
102 | static struct ip_tunnel ** ipip6_bucket(struct ip_tunnel *t) | ||
103 | { | ||
104 | u32 remote = t->parms.iph.daddr; | ||
105 | u32 local = t->parms.iph.saddr; | ||
106 | unsigned h = 0; | ||
107 | int prio = 0; | ||
108 | |||
109 | if (remote) { | ||
110 | prio |= 2; | ||
111 | h ^= HASH(remote); | ||
112 | } | ||
113 | if (local) { | ||
114 | prio |= 1; | ||
115 | h ^= HASH(local); | ||
116 | } | ||
117 | return &tunnels[prio][h]; | ||
118 | } | ||
119 | |||
120 | static void ipip6_tunnel_unlink(struct ip_tunnel *t) | ||
121 | { | ||
122 | struct ip_tunnel **tp; | ||
123 | |||
124 | for (tp = ipip6_bucket(t); *tp; tp = &(*tp)->next) { | ||
125 | if (t == *tp) { | ||
126 | write_lock_bh(&ipip6_lock); | ||
127 | *tp = t->next; | ||
128 | write_unlock_bh(&ipip6_lock); | ||
129 | break; | ||
130 | } | ||
131 | } | ||
132 | } | ||
133 | |||
134 | static void ipip6_tunnel_link(struct ip_tunnel *t) | ||
135 | { | ||
136 | struct ip_tunnel **tp = ipip6_bucket(t); | ||
137 | |||
138 | t->next = *tp; | ||
139 | write_lock_bh(&ipip6_lock); | ||
140 | *tp = t; | ||
141 | write_unlock_bh(&ipip6_lock); | ||
142 | } | ||
143 | |||
144 | static struct ip_tunnel * ipip6_tunnel_locate(struct ip_tunnel_parm *parms, int create) | ||
145 | { | ||
146 | u32 remote = parms->iph.daddr; | ||
147 | u32 local = parms->iph.saddr; | ||
148 | struct ip_tunnel *t, **tp, *nt; | ||
149 | struct net_device *dev; | ||
150 | unsigned h = 0; | ||
151 | int prio = 0; | ||
152 | char name[IFNAMSIZ]; | ||
153 | |||
154 | if (remote) { | ||
155 | prio |= 2; | ||
156 | h ^= HASH(remote); | ||
157 | } | ||
158 | if (local) { | ||
159 | prio |= 1; | ||
160 | h ^= HASH(local); | ||
161 | } | ||
162 | for (tp = &tunnels[prio][h]; (t = *tp) != NULL; tp = &t->next) { | ||
163 | if (local == t->parms.iph.saddr && remote == t->parms.iph.daddr) | ||
164 | return t; | ||
165 | } | ||
166 | if (!create) | ||
167 | goto failed; | ||
168 | |||
169 | if (parms->name[0]) | ||
170 | strlcpy(name, parms->name, IFNAMSIZ); | ||
171 | else { | ||
172 | int i; | ||
173 | for (i=1; i<100; i++) { | ||
174 | sprintf(name, "sit%d", i); | ||
175 | if (__dev_get_by_name(name) == NULL) | ||
176 | break; | ||
177 | } | ||
178 | if (i==100) | ||
179 | goto failed; | ||
180 | } | ||
181 | |||
182 | dev = alloc_netdev(sizeof(*t), name, ipip6_tunnel_setup); | ||
183 | if (dev == NULL) | ||
184 | return NULL; | ||
185 | |||
186 | nt = dev->priv; | ||
187 | dev->init = ipip6_tunnel_init; | ||
188 | nt->parms = *parms; | ||
189 | |||
190 | if (register_netdevice(dev) < 0) { | ||
191 | free_netdev(dev); | ||
192 | goto failed; | ||
193 | } | ||
194 | |||
195 | dev_hold(dev); | ||
196 | |||
197 | ipip6_tunnel_link(nt); | ||
198 | /* Do not decrement MOD_USE_COUNT here. */ | ||
199 | return nt; | ||
200 | |||
201 | failed: | ||
202 | return NULL; | ||
203 | } | ||
204 | |||
205 | static void ipip6_tunnel_uninit(struct net_device *dev) | ||
206 | { | ||
207 | if (dev == ipip6_fb_tunnel_dev) { | ||
208 | write_lock_bh(&ipip6_lock); | ||
209 | tunnels_wc[0] = NULL; | ||
210 | write_unlock_bh(&ipip6_lock); | ||
211 | dev_put(dev); | ||
212 | } else { | ||
213 | ipip6_tunnel_unlink((struct ip_tunnel*)dev->priv); | ||
214 | dev_put(dev); | ||
215 | } | ||
216 | } | ||
217 | |||
218 | |||
219 | static void ipip6_err(struct sk_buff *skb, u32 info) | ||
220 | { | ||
221 | #ifndef I_WISH_WORLD_WERE_PERFECT | ||
222 | |||
223 | /* It is not :-( All the routers (except for Linux) return only | ||
224 | 8 bytes of packet payload. It means, that precise relaying of | ||
225 | ICMP in the real Internet is absolutely infeasible. | ||
226 | */ | ||
227 | struct iphdr *iph = (struct iphdr*)skb->data; | ||
228 | int type = skb->h.icmph->type; | ||
229 | int code = skb->h.icmph->code; | ||
230 | struct ip_tunnel *t; | ||
231 | |||
232 | switch (type) { | ||
233 | default: | ||
234 | case ICMP_PARAMETERPROB: | ||
235 | return; | ||
236 | |||
237 | case ICMP_DEST_UNREACH: | ||
238 | switch (code) { | ||
239 | case ICMP_SR_FAILED: | ||
240 | case ICMP_PORT_UNREACH: | ||
241 | /* Impossible event. */ | ||
242 | return; | ||
243 | case ICMP_FRAG_NEEDED: | ||
244 | /* Soft state for pmtu is maintained by IP core. */ | ||
245 | return; | ||
246 | default: | ||
247 | /* All others are translated to HOST_UNREACH. | ||
248 | rfc2003 contains "deep thoughts" about NET_UNREACH, | ||
249 | I believe they are just ether pollution. --ANK | ||
250 | */ | ||
251 | break; | ||
252 | } | ||
253 | break; | ||
254 | case ICMP_TIME_EXCEEDED: | ||
255 | if (code != ICMP_EXC_TTL) | ||
256 | return; | ||
257 | break; | ||
258 | } | ||
259 | |||
260 | read_lock(&ipip6_lock); | ||
261 | t = ipip6_tunnel_lookup(iph->daddr, iph->saddr); | ||
262 | if (t == NULL || t->parms.iph.daddr == 0) | ||
263 | goto out; | ||
264 | if (t->parms.iph.ttl == 0 && type == ICMP_TIME_EXCEEDED) | ||
265 | goto out; | ||
266 | |||
267 | if (jiffies - t->err_time < IPTUNNEL_ERR_TIMEO) | ||
268 | t->err_count++; | ||
269 | else | ||
270 | t->err_count = 1; | ||
271 | t->err_time = jiffies; | ||
272 | out: | ||
273 | read_unlock(&ipip6_lock); | ||
274 | return; | ||
275 | #else | ||
276 | struct iphdr *iph = (struct iphdr*)dp; | ||
277 | int hlen = iph->ihl<<2; | ||
278 | struct ipv6hdr *iph6; | ||
279 | int type = skb->h.icmph->type; | ||
280 | int code = skb->h.icmph->code; | ||
281 | int rel_type = 0; | ||
282 | int rel_code = 0; | ||
283 | int rel_info = 0; | ||
284 | struct sk_buff *skb2; | ||
285 | struct rt6_info *rt6i; | ||
286 | |||
287 | if (len < hlen + sizeof(struct ipv6hdr)) | ||
288 | return; | ||
289 | iph6 = (struct ipv6hdr*)(dp + hlen); | ||
290 | |||
291 | switch (type) { | ||
292 | default: | ||
293 | return; | ||
294 | case ICMP_PARAMETERPROB: | ||
295 | if (skb->h.icmph->un.gateway < hlen) | ||
296 | return; | ||
297 | |||
298 | /* So... This guy found something strange INSIDE encapsulated | ||
299 | packet. Well, he is fool, but what can we do ? | ||
300 | */ | ||
301 | rel_type = ICMPV6_PARAMPROB; | ||
302 | rel_info = skb->h.icmph->un.gateway - hlen; | ||
303 | break; | ||
304 | |||
305 | case ICMP_DEST_UNREACH: | ||
306 | switch (code) { | ||
307 | case ICMP_SR_FAILED: | ||
308 | case ICMP_PORT_UNREACH: | ||
309 | /* Impossible event. */ | ||
310 | return; | ||
311 | case ICMP_FRAG_NEEDED: | ||
312 | /* Too complicated case ... */ | ||
313 | return; | ||
314 | default: | ||
315 | /* All others are translated to HOST_UNREACH. | ||
316 | rfc2003 contains "deep thoughts" about NET_UNREACH, | ||
317 | I believe, it is just ether pollution. --ANK | ||
318 | */ | ||
319 | rel_type = ICMPV6_DEST_UNREACH; | ||
320 | rel_code = ICMPV6_ADDR_UNREACH; | ||
321 | break; | ||
322 | } | ||
323 | break; | ||
324 | case ICMP_TIME_EXCEEDED: | ||
325 | if (code != ICMP_EXC_TTL) | ||
326 | return; | ||
327 | rel_type = ICMPV6_TIME_EXCEED; | ||
328 | rel_code = ICMPV6_EXC_HOPLIMIT; | ||
329 | break; | ||
330 | } | ||
331 | |||
332 | /* Prepare fake skb to feed it to icmpv6_send */ | ||
333 | skb2 = skb_clone(skb, GFP_ATOMIC); | ||
334 | if (skb2 == NULL) | ||
335 | return; | ||
336 | dst_release(skb2->dst); | ||
337 | skb2->dst = NULL; | ||
338 | skb_pull(skb2, skb->data - (u8*)iph6); | ||
339 | skb2->nh.raw = skb2->data; | ||
340 | |||
341 | /* Try to guess incoming interface */ | ||
342 | rt6i = rt6_lookup(&iph6->saddr, NULL, NULL, 0); | ||
343 | if (rt6i && rt6i->rt6i_dev) { | ||
344 | skb2->dev = rt6i->rt6i_dev; | ||
345 | |||
346 | rt6i = rt6_lookup(&iph6->daddr, &iph6->saddr, NULL, 0); | ||
347 | |||
348 | if (rt6i && rt6i->rt6i_dev && rt6i->rt6i_dev->type == ARPHRD_SIT) { | ||
349 | struct ip_tunnel * t = (struct ip_tunnel*)rt6i->rt6i_dev->priv; | ||
350 | if (rel_type == ICMPV6_TIME_EXCEED && t->parms.iph.ttl) { | ||
351 | rel_type = ICMPV6_DEST_UNREACH; | ||
352 | rel_code = ICMPV6_ADDR_UNREACH; | ||
353 | } | ||
354 | icmpv6_send(skb2, rel_type, rel_code, rel_info, skb2->dev); | ||
355 | } | ||
356 | } | ||
357 | kfree_skb(skb2); | ||
358 | return; | ||
359 | #endif | ||
360 | } | ||
361 | |||
362 | static inline void ipip6_ecn_decapsulate(struct iphdr *iph, struct sk_buff *skb) | ||
363 | { | ||
364 | if (INET_ECN_is_ce(iph->tos)) | ||
365 | IP6_ECN_set_ce(skb->nh.ipv6h); | ||
366 | } | ||
367 | |||
368 | static int ipip6_rcv(struct sk_buff *skb) | ||
369 | { | ||
370 | struct iphdr *iph; | ||
371 | struct ip_tunnel *tunnel; | ||
372 | |||
373 | if (!pskb_may_pull(skb, sizeof(struct ipv6hdr))) | ||
374 | goto out; | ||
375 | |||
376 | iph = skb->nh.iph; | ||
377 | |||
378 | read_lock(&ipip6_lock); | ||
379 | if ((tunnel = ipip6_tunnel_lookup(iph->saddr, iph->daddr)) != NULL) { | ||
380 | secpath_reset(skb); | ||
381 | skb->mac.raw = skb->nh.raw; | ||
382 | skb->nh.raw = skb->data; | ||
383 | memset(&(IPCB(skb)->opt), 0, sizeof(struct ip_options)); | ||
384 | skb->protocol = htons(ETH_P_IPV6); | ||
385 | skb->pkt_type = PACKET_HOST; | ||
386 | tunnel->stat.rx_packets++; | ||
387 | tunnel->stat.rx_bytes += skb->len; | ||
388 | skb->dev = tunnel->dev; | ||
389 | dst_release(skb->dst); | ||
390 | skb->dst = NULL; | ||
391 | nf_reset(skb); | ||
392 | ipip6_ecn_decapsulate(iph, skb); | ||
393 | netif_rx(skb); | ||
394 | read_unlock(&ipip6_lock); | ||
395 | return 0; | ||
396 | } | ||
397 | |||
398 | icmp_send(skb, ICMP_DEST_UNREACH, ICMP_PROT_UNREACH, 0); | ||
399 | kfree_skb(skb); | ||
400 | read_unlock(&ipip6_lock); | ||
401 | out: | ||
402 | return 0; | ||
403 | } | ||
404 | |||
405 | /* Returns the embedded IPv4 address if the IPv6 address | ||
406 | comes from 6to4 (RFC 3056) addr space */ | ||
407 | |||
408 | static inline u32 try_6to4(struct in6_addr *v6dst) | ||
409 | { | ||
410 | u32 dst = 0; | ||
411 | |||
412 | if (v6dst->s6_addr16[0] == htons(0x2002)) { | ||
413 | /* 6to4 v6 addr has 16 bits prefix, 32 v4addr, 16 SLA, ... */ | ||
414 | memcpy(&dst, &v6dst->s6_addr16[1], 4); | ||
415 | } | ||
416 | return dst; | ||
417 | } | ||
418 | |||
419 | /* | ||
420 | * This function assumes it is being called from dev_queue_xmit() | ||
421 | * and that skb is filled properly by that function. | ||
422 | */ | ||
423 | |||
424 | static int ipip6_tunnel_xmit(struct sk_buff *skb, struct net_device *dev) | ||
425 | { | ||
426 | struct ip_tunnel *tunnel = (struct ip_tunnel*)dev->priv; | ||
427 | struct net_device_stats *stats = &tunnel->stat; | ||
428 | struct iphdr *tiph = &tunnel->parms.iph; | ||
429 | struct ipv6hdr *iph6 = skb->nh.ipv6h; | ||
430 | u8 tos = tunnel->parms.iph.tos; | ||
431 | struct rtable *rt; /* Route to the other host */ | ||
432 | struct net_device *tdev; /* Device to other host */ | ||
433 | struct iphdr *iph; /* Our new IP header */ | ||
434 | int max_headroom; /* The extra header space needed */ | ||
435 | u32 dst = tiph->daddr; | ||
436 | int mtu; | ||
437 | struct in6_addr *addr6; | ||
438 | int addr_type; | ||
439 | |||
440 | if (tunnel->recursion++) { | ||
441 | tunnel->stat.collisions++; | ||
442 | goto tx_error; | ||
443 | } | ||
444 | |||
445 | if (skb->protocol != htons(ETH_P_IPV6)) | ||
446 | goto tx_error; | ||
447 | |||
448 | if (!dst) | ||
449 | dst = try_6to4(&iph6->daddr); | ||
450 | |||
451 | if (!dst) { | ||
452 | struct neighbour *neigh = NULL; | ||
453 | |||
454 | if (skb->dst) | ||
455 | neigh = skb->dst->neighbour; | ||
456 | |||
457 | if (neigh == NULL) { | ||
458 | if (net_ratelimit()) | ||
459 | printk(KERN_DEBUG "sit: nexthop == NULL\n"); | ||
460 | goto tx_error; | ||
461 | } | ||
462 | |||
463 | addr6 = (struct in6_addr*)&neigh->primary_key; | ||
464 | addr_type = ipv6_addr_type(addr6); | ||
465 | |||
466 | if (addr_type == IPV6_ADDR_ANY) { | ||
467 | addr6 = &skb->nh.ipv6h->daddr; | ||
468 | addr_type = ipv6_addr_type(addr6); | ||
469 | } | ||
470 | |||
471 | if ((addr_type & IPV6_ADDR_COMPATv4) == 0) | ||
472 | goto tx_error_icmp; | ||
473 | |||
474 | dst = addr6->s6_addr32[3]; | ||
475 | } | ||
476 | |||
477 | { | ||
478 | struct flowi fl = { .nl_u = { .ip4_u = | ||
479 | { .daddr = dst, | ||
480 | .saddr = tiph->saddr, | ||
481 | .tos = RT_TOS(tos) } }, | ||
482 | .oif = tunnel->parms.link, | ||
483 | .proto = IPPROTO_IPV6 }; | ||
484 | if (ip_route_output_key(&rt, &fl)) { | ||
485 | tunnel->stat.tx_carrier_errors++; | ||
486 | goto tx_error_icmp; | ||
487 | } | ||
488 | } | ||
489 | if (rt->rt_type != RTN_UNICAST) { | ||
490 | ip_rt_put(rt); | ||
491 | tunnel->stat.tx_carrier_errors++; | ||
492 | goto tx_error_icmp; | ||
493 | } | ||
494 | tdev = rt->u.dst.dev; | ||
495 | |||
496 | if (tdev == dev) { | ||
497 | ip_rt_put(rt); | ||
498 | tunnel->stat.collisions++; | ||
499 | goto tx_error; | ||
500 | } | ||
501 | |||
502 | if (tiph->frag_off) | ||
503 | mtu = dst_mtu(&rt->u.dst) - sizeof(struct iphdr); | ||
504 | else | ||
505 | mtu = skb->dst ? dst_mtu(skb->dst) : dev->mtu; | ||
506 | |||
507 | if (mtu < 68) { | ||
508 | tunnel->stat.collisions++; | ||
509 | ip_rt_put(rt); | ||
510 | goto tx_error; | ||
511 | } | ||
512 | if (mtu < IPV6_MIN_MTU) | ||
513 | mtu = IPV6_MIN_MTU; | ||
514 | if (tunnel->parms.iph.daddr && skb->dst) | ||
515 | skb->dst->ops->update_pmtu(skb->dst, mtu); | ||
516 | |||
517 | if (skb->len > mtu) { | ||
518 | icmpv6_send(skb, ICMPV6_PKT_TOOBIG, 0, mtu, dev); | ||
519 | ip_rt_put(rt); | ||
520 | goto tx_error; | ||
521 | } | ||
522 | |||
523 | if (tunnel->err_count > 0) { | ||
524 | if (jiffies - tunnel->err_time < IPTUNNEL_ERR_TIMEO) { | ||
525 | tunnel->err_count--; | ||
526 | dst_link_failure(skb); | ||
527 | } else | ||
528 | tunnel->err_count = 0; | ||
529 | } | ||
530 | |||
531 | /* | ||
532 | * Okay, now see if we can stuff it in the buffer as-is. | ||
533 | */ | ||
534 | max_headroom = LL_RESERVED_SPACE(tdev)+sizeof(struct iphdr); | ||
535 | |||
536 | if (skb_headroom(skb) < max_headroom || skb_cloned(skb) || skb_shared(skb)) { | ||
537 | struct sk_buff *new_skb = skb_realloc_headroom(skb, max_headroom); | ||
538 | if (!new_skb) { | ||
539 | ip_rt_put(rt); | ||
540 | stats->tx_dropped++; | ||
541 | dev_kfree_skb(skb); | ||
542 | tunnel->recursion--; | ||
543 | return 0; | ||
544 | } | ||
545 | if (skb->sk) | ||
546 | skb_set_owner_w(new_skb, skb->sk); | ||
547 | dev_kfree_skb(skb); | ||
548 | skb = new_skb; | ||
549 | iph6 = skb->nh.ipv6h; | ||
550 | } | ||
551 | |||
552 | skb->h.raw = skb->nh.raw; | ||
553 | skb->nh.raw = skb_push(skb, sizeof(struct iphdr)); | ||
554 | memset(&(IPCB(skb)->opt), 0, sizeof(IPCB(skb)->opt)); | ||
555 | dst_release(skb->dst); | ||
556 | skb->dst = &rt->u.dst; | ||
557 | |||
558 | /* | ||
559 | * Push down and install the IPIP header. | ||
560 | */ | ||
561 | |||
562 | iph = skb->nh.iph; | ||
563 | iph->version = 4; | ||
564 | iph->ihl = sizeof(struct iphdr)>>2; | ||
565 | if (mtu > IPV6_MIN_MTU) | ||
566 | iph->frag_off = htons(IP_DF); | ||
567 | else | ||
568 | iph->frag_off = 0; | ||
569 | |||
570 | iph->protocol = IPPROTO_IPV6; | ||
571 | iph->tos = INET_ECN_encapsulate(tos, ipv6_get_dsfield(iph6)); | ||
572 | iph->daddr = rt->rt_dst; | ||
573 | iph->saddr = rt->rt_src; | ||
574 | |||
575 | if ((iph->ttl = tiph->ttl) == 0) | ||
576 | iph->ttl = iph6->hop_limit; | ||
577 | |||
578 | nf_reset(skb); | ||
579 | |||
580 | IPTUNNEL_XMIT(); | ||
581 | tunnel->recursion--; | ||
582 | return 0; | ||
583 | |||
584 | tx_error_icmp: | ||
585 | dst_link_failure(skb); | ||
586 | tx_error: | ||
587 | stats->tx_errors++; | ||
588 | dev_kfree_skb(skb); | ||
589 | tunnel->recursion--; | ||
590 | return 0; | ||
591 | } | ||
592 | |||
593 | static int | ||
594 | ipip6_tunnel_ioctl (struct net_device *dev, struct ifreq *ifr, int cmd) | ||
595 | { | ||
596 | int err = 0; | ||
597 | struct ip_tunnel_parm p; | ||
598 | struct ip_tunnel *t; | ||
599 | |||
600 | switch (cmd) { | ||
601 | case SIOCGETTUNNEL: | ||
602 | t = NULL; | ||
603 | if (dev == ipip6_fb_tunnel_dev) { | ||
604 | if (copy_from_user(&p, ifr->ifr_ifru.ifru_data, sizeof(p))) { | ||
605 | err = -EFAULT; | ||
606 | break; | ||
607 | } | ||
608 | t = ipip6_tunnel_locate(&p, 0); | ||
609 | } | ||
610 | if (t == NULL) | ||
611 | t = (struct ip_tunnel*)dev->priv; | ||
612 | memcpy(&p, &t->parms, sizeof(p)); | ||
613 | if (copy_to_user(ifr->ifr_ifru.ifru_data, &p, sizeof(p))) | ||
614 | err = -EFAULT; | ||
615 | break; | ||
616 | |||
617 | case SIOCADDTUNNEL: | ||
618 | case SIOCCHGTUNNEL: | ||
619 | err = -EPERM; | ||
620 | if (!capable(CAP_NET_ADMIN)) | ||
621 | goto done; | ||
622 | |||
623 | err = -EFAULT; | ||
624 | if (copy_from_user(&p, ifr->ifr_ifru.ifru_data, sizeof(p))) | ||
625 | goto done; | ||
626 | |||
627 | err = -EINVAL; | ||
628 | if (p.iph.version != 4 || p.iph.protocol != IPPROTO_IPV6 || | ||
629 | p.iph.ihl != 5 || (p.iph.frag_off&htons(~IP_DF))) | ||
630 | goto done; | ||
631 | if (p.iph.ttl) | ||
632 | p.iph.frag_off |= htons(IP_DF); | ||
633 | |||
634 | t = ipip6_tunnel_locate(&p, cmd == SIOCADDTUNNEL); | ||
635 | |||
636 | if (dev != ipip6_fb_tunnel_dev && cmd == SIOCCHGTUNNEL) { | ||
637 | if (t != NULL) { | ||
638 | if (t->dev != dev) { | ||
639 | err = -EEXIST; | ||
640 | break; | ||
641 | } | ||
642 | } else { | ||
643 | if (((dev->flags&IFF_POINTOPOINT) && !p.iph.daddr) || | ||
644 | (!(dev->flags&IFF_POINTOPOINT) && p.iph.daddr)) { | ||
645 | err = -EINVAL; | ||
646 | break; | ||
647 | } | ||
648 | t = (struct ip_tunnel*)dev->priv; | ||
649 | ipip6_tunnel_unlink(t); | ||
650 | t->parms.iph.saddr = p.iph.saddr; | ||
651 | t->parms.iph.daddr = p.iph.daddr; | ||
652 | memcpy(dev->dev_addr, &p.iph.saddr, 4); | ||
653 | memcpy(dev->broadcast, &p.iph.daddr, 4); | ||
654 | ipip6_tunnel_link(t); | ||
655 | netdev_state_change(dev); | ||
656 | } | ||
657 | } | ||
658 | |||
659 | if (t) { | ||
660 | err = 0; | ||
661 | if (cmd == SIOCCHGTUNNEL) { | ||
662 | t->parms.iph.ttl = p.iph.ttl; | ||
663 | t->parms.iph.tos = p.iph.tos; | ||
664 | } | ||
665 | if (copy_to_user(ifr->ifr_ifru.ifru_data, &t->parms, sizeof(p))) | ||
666 | err = -EFAULT; | ||
667 | } else | ||
668 | err = (cmd == SIOCADDTUNNEL ? -ENOBUFS : -ENOENT); | ||
669 | break; | ||
670 | |||
671 | case SIOCDELTUNNEL: | ||
672 | err = -EPERM; | ||
673 | if (!capable(CAP_NET_ADMIN)) | ||
674 | goto done; | ||
675 | |||
676 | if (dev == ipip6_fb_tunnel_dev) { | ||
677 | err = -EFAULT; | ||
678 | if (copy_from_user(&p, ifr->ifr_ifru.ifru_data, sizeof(p))) | ||
679 | goto done; | ||
680 | err = -ENOENT; | ||
681 | if ((t = ipip6_tunnel_locate(&p, 0)) == NULL) | ||
682 | goto done; | ||
683 | err = -EPERM; | ||
684 | if (t == ipip6_fb_tunnel_dev->priv) | ||
685 | goto done; | ||
686 | dev = t->dev; | ||
687 | } | ||
688 | err = unregister_netdevice(dev); | ||
689 | break; | ||
690 | |||
691 | default: | ||
692 | err = -EINVAL; | ||
693 | } | ||
694 | |||
695 | done: | ||
696 | return err; | ||
697 | } | ||
698 | |||
699 | static struct net_device_stats *ipip6_tunnel_get_stats(struct net_device *dev) | ||
700 | { | ||
701 | return &(((struct ip_tunnel*)dev->priv)->stat); | ||
702 | } | ||
703 | |||
704 | static int ipip6_tunnel_change_mtu(struct net_device *dev, int new_mtu) | ||
705 | { | ||
706 | if (new_mtu < IPV6_MIN_MTU || new_mtu > 0xFFF8 - sizeof(struct iphdr)) | ||
707 | return -EINVAL; | ||
708 | dev->mtu = new_mtu; | ||
709 | return 0; | ||
710 | } | ||
711 | |||
712 | static void ipip6_tunnel_setup(struct net_device *dev) | ||
713 | { | ||
714 | SET_MODULE_OWNER(dev); | ||
715 | dev->uninit = ipip6_tunnel_uninit; | ||
716 | dev->destructor = free_netdev; | ||
717 | dev->hard_start_xmit = ipip6_tunnel_xmit; | ||
718 | dev->get_stats = ipip6_tunnel_get_stats; | ||
719 | dev->do_ioctl = ipip6_tunnel_ioctl; | ||
720 | dev->change_mtu = ipip6_tunnel_change_mtu; | ||
721 | |||
722 | dev->type = ARPHRD_SIT; | ||
723 | dev->hard_header_len = LL_MAX_HEADER + sizeof(struct iphdr); | ||
724 | dev->mtu = 1500 - sizeof(struct iphdr); | ||
725 | dev->flags = IFF_NOARP; | ||
726 | dev->iflink = 0; | ||
727 | dev->addr_len = 4; | ||
728 | } | ||
729 | |||
730 | static int ipip6_tunnel_init(struct net_device *dev) | ||
731 | { | ||
732 | struct net_device *tdev = NULL; | ||
733 | struct ip_tunnel *tunnel; | ||
734 | struct iphdr *iph; | ||
735 | |||
736 | tunnel = (struct ip_tunnel*)dev->priv; | ||
737 | iph = &tunnel->parms.iph; | ||
738 | |||
739 | tunnel->dev = dev; | ||
740 | strcpy(tunnel->parms.name, dev->name); | ||
741 | |||
742 | memcpy(dev->dev_addr, &tunnel->parms.iph.saddr, 4); | ||
743 | memcpy(dev->broadcast, &tunnel->parms.iph.daddr, 4); | ||
744 | |||
745 | if (iph->daddr) { | ||
746 | struct flowi fl = { .nl_u = { .ip4_u = | ||
747 | { .daddr = iph->daddr, | ||
748 | .saddr = iph->saddr, | ||
749 | .tos = RT_TOS(iph->tos) } }, | ||
750 | .oif = tunnel->parms.link, | ||
751 | .proto = IPPROTO_IPV6 }; | ||
752 | struct rtable *rt; | ||
753 | if (!ip_route_output_key(&rt, &fl)) { | ||
754 | tdev = rt->u.dst.dev; | ||
755 | ip_rt_put(rt); | ||
756 | } | ||
757 | dev->flags |= IFF_POINTOPOINT; | ||
758 | } | ||
759 | |||
760 | if (!tdev && tunnel->parms.link) | ||
761 | tdev = __dev_get_by_index(tunnel->parms.link); | ||
762 | |||
763 | if (tdev) { | ||
764 | dev->hard_header_len = tdev->hard_header_len + sizeof(struct iphdr); | ||
765 | dev->mtu = tdev->mtu - sizeof(struct iphdr); | ||
766 | if (dev->mtu < IPV6_MIN_MTU) | ||
767 | dev->mtu = IPV6_MIN_MTU; | ||
768 | } | ||
769 | dev->iflink = tunnel->parms.link; | ||
770 | |||
771 | return 0; | ||
772 | } | ||
773 | |||
774 | int __init ipip6_fb_tunnel_init(struct net_device *dev) | ||
775 | { | ||
776 | struct ip_tunnel *tunnel = dev->priv; | ||
777 | struct iphdr *iph = &tunnel->parms.iph; | ||
778 | |||
779 | tunnel->dev = dev; | ||
780 | strcpy(tunnel->parms.name, dev->name); | ||
781 | |||
782 | iph->version = 4; | ||
783 | iph->protocol = IPPROTO_IPV6; | ||
784 | iph->ihl = 5; | ||
785 | iph->ttl = 64; | ||
786 | |||
787 | dev_hold(dev); | ||
788 | tunnels_wc[0] = tunnel; | ||
789 | return 0; | ||
790 | } | ||
791 | |||
792 | static struct net_protocol sit_protocol = { | ||
793 | .handler = ipip6_rcv, | ||
794 | .err_handler = ipip6_err, | ||
795 | }; | ||
796 | |||
797 | void __exit sit_cleanup(void) | ||
798 | { | ||
799 | inet_del_protocol(&sit_protocol, IPPROTO_IPV6); | ||
800 | unregister_netdev(ipip6_fb_tunnel_dev); | ||
801 | } | ||
802 | |||
803 | int __init sit_init(void) | ||
804 | { | ||
805 | int err; | ||
806 | |||
807 | printk(KERN_INFO "IPv6 over IPv4 tunneling driver\n"); | ||
808 | |||
809 | if (inet_add_protocol(&sit_protocol, IPPROTO_IPV6) < 0) { | ||
810 | printk(KERN_INFO "sit init: Can't add protocol\n"); | ||
811 | return -EAGAIN; | ||
812 | } | ||
813 | |||
814 | ipip6_fb_tunnel_dev = alloc_netdev(sizeof(struct ip_tunnel), "sit0", | ||
815 | ipip6_tunnel_setup); | ||
816 | if (!ipip6_fb_tunnel_dev) { | ||
817 | err = -ENOMEM; | ||
818 | goto err1; | ||
819 | } | ||
820 | |||
821 | ipip6_fb_tunnel_dev->init = ipip6_fb_tunnel_init; | ||
822 | |||
823 | if ((err = register_netdev(ipip6_fb_tunnel_dev))) | ||
824 | goto err2; | ||
825 | |||
826 | out: | ||
827 | return err; | ||
828 | err2: | ||
829 | free_netdev(ipip6_fb_tunnel_dev); | ||
830 | err1: | ||
831 | inet_del_protocol(&sit_protocol, IPPROTO_IPV6); | ||
832 | goto out; | ||
833 | } | ||