pc64: Setup TSC_AUX register with each cpu's id, when rdtscp is available.
[dragonfly.git] / sys / netinet6 / ip6_forward.c
blob4d54c25fa29bd1ed2e0bf9c999113464895cb1d0
1 /* $FreeBSD: src/sys/netinet6/ip6_forward.c,v 1.4.2.7 2003/01/24 05:11:35 sam Exp $ */
2 /* $DragonFly: src/sys/netinet6/ip6_forward.c,v 1.13 2006/12/22 23:57:53 swildner Exp $ */
3 /* $KAME: ip6_forward.c,v 1.69 2001/05/17 03:48:30 itojun Exp $ */
5 /*
6 * Copyright (C) 1995, 1996, 1997, and 1998 WIDE Project.
7 * All rights reserved.
9 * Redistribution and use in source and binary forms, with or without
10 * modification, are permitted provided that the following conditions
11 * are met:
12 * 1. Redistributions of source code must retain the above copyright
13 * notice, this list of conditions and the following disclaimer.
14 * 2. Redistributions in binary form must reproduce the above copyright
15 * notice, this list of conditions and the following disclaimer in the
16 * documentation and/or other materials provided with the distribution.
17 * 3. Neither the name of the project nor the names of its contributors
18 * may be used to endorse or promote products derived from this software
19 * without specific prior written permission.
21 * THIS SOFTWARE IS PROVIDED BY THE PROJECT AND CONTRIBUTORS ``AS IS'' AND
22 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
23 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
24 * ARE DISCLAIMED. IN NO EVENT SHALL THE PROJECT OR CONTRIBUTORS BE LIABLE
25 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
26 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
27 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
28 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
29 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
30 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
31 * SUCH DAMAGE.
34 #include "opt_ip6fw.h"
35 #include "opt_inet.h"
36 #include "opt_inet6.h"
37 #include "opt_ipsec.h"
39 #include <sys/param.h>
40 #include <sys/systm.h>
41 #include <sys/malloc.h>
42 #include <sys/mbuf.h>
43 #include <sys/domain.h>
44 #include <sys/protosw.h>
45 #include <sys/socket.h>
46 #include <sys/errno.h>
47 #include <sys/time.h>
48 #include <sys/kernel.h>
49 #include <sys/syslog.h>
51 #include <net/if.h>
52 #include <net/route.h>
53 #include <net/pfil.h>
55 #include <netinet/in.h>
56 #include <netinet/in_var.h>
57 #include <netinet/in_systm.h>
58 #include <netinet/ip.h>
59 #include <netinet/ip_var.h>
60 #include <netinet6/in6_var.h>
61 #include <netinet/ip6.h>
62 #include <netinet6/ip6_var.h>
63 #include <netinet/icmp6.h>
64 #include <netinet6/nd6.h>
66 #include <netinet/in_pcb.h>
68 #ifdef IPSEC
69 #include <netinet6/ipsec.h>
70 #ifdef INET6
71 #include <netinet6/ipsec6.h>
72 #endif
73 #include <netproto/key/key.h>
74 #endif /* IPSEC */
76 #ifdef FAST_IPSEC
77 #include <netproto/ipsec/ipsec.h>
78 #include <netproto/ipsec/ipsec6.h>
79 #include <netproto/ipsec/key.h>
80 #define IPSEC
81 #endif /* FAST_IPSEC */
83 #include <net/ip6fw/ip6_fw.h>
85 #include <net/net_osdep.h>
87 struct route_in6 ip6_forward_rt;
90 * Forward a packet. If some error occurs return the sender
91 * an icmp packet. Note we can't always generate a meaningful
92 * icmp message because icmp doesn't have a large enough repertoire
93 * of codes and types.
95 * If not forwarding, just drop the packet. This could be confusing
96 * if ipforwarding was zero but some routing protocol was advancing
97 * us as a gateway to somewhere. However, we must let the routing
98 * protocol deal with that.
102 void
103 ip6_forward(struct mbuf *m, int srcrt)
105 struct ip6_hdr *ip6 = mtod(m, struct ip6_hdr *);
106 struct sockaddr_in6 *dst;
107 struct rtentry *rt;
108 int error, type = 0, code = 0;
109 struct mbuf *mcopy = NULL;
110 struct ifnet *origifp; /* maybe unnecessary */
111 u_int32_t srczone, dstzone;
112 #ifdef IPSEC
113 struct secpolicy *sp = NULL;
114 #endif
116 #ifdef IPSEC
118 * Check AH/ESP integrity.
121 * Don't increment ip6s_cantforward because this is the check
122 * before forwarding packet actually.
124 if (ipsec6_in_reject(m, NULL)) {
125 #if !defined(FAST_IPSEC)
126 ipsec6stat.in_polvio++;
127 #endif
128 m_freem(m);
129 return;
131 #endif /* IPSEC */
134 * Do not forward packets to multicast destination (should be handled
135 * by ip6_mforward().
136 * Do not forward packets with unspecified source. It was discussed
137 * in July 2000, on the ipngwg mailing list.
139 if ((m->m_flags & (M_BCAST | M_MCAST)) != 0 ||
140 IN6_IS_ADDR_MULTICAST(&ip6->ip6_dst) ||
141 IN6_IS_ADDR_UNSPECIFIED(&ip6->ip6_src)) {
142 ip6stat.ip6s_cantforward++;
143 /* XXX in6_ifstat_inc(rt->rt_ifp, ifs6_in_discard) */
144 if (ip6_log_time + ip6_log_interval < time_uptime) {
145 ip6_log_time = time_uptime;
146 log(LOG_DEBUG,
147 "cannot forward "
148 "from %s to %s nxt %d received on %s\n",
149 ip6_sprintf(&ip6->ip6_src),
150 ip6_sprintf(&ip6->ip6_dst),
151 ip6->ip6_nxt,
152 if_name(m->m_pkthdr.rcvif));
154 m_freem(m);
155 return;
158 if (ip6->ip6_hlim <= IPV6_HLIMDEC) {
159 /* XXX in6_ifstat_inc(rt->rt_ifp, ifs6_in_discard) */
160 icmp6_error(m, ICMP6_TIME_EXCEEDED,
161 ICMP6_TIME_EXCEED_TRANSIT, 0);
162 return;
164 ip6->ip6_hlim -= IPV6_HLIMDEC;
167 * Save at most ICMPV6_PLD_MAXLEN (= the min IPv6 MTU -
168 * size of IPv6 + ICMPv6 headers) bytes of the packet in case
169 * we need to generate an ICMP6 message to the src.
170 * Thanks to M_EXT, in most cases copy will not occur.
172 * It is important to save it before IPsec processing as IPsec
173 * processing may modify the mbuf.
175 mcopy = m_copy(m, 0, imin(m->m_pkthdr.len, ICMPV6_PLD_MAXLEN));
177 #ifdef IPSEC
178 /* get a security policy for this packet */
179 sp = ipsec6_getpolicybyaddr(m, IPSEC_DIR_OUTBOUND, IP_FORWARDING,
180 &error);
181 if (sp == NULL) {
182 ipsec6stat.out_inval++;
183 ip6stat.ip6s_cantforward++;
184 if (mcopy) {
185 #if 0
186 /* XXX: what icmp ? */
187 #else
188 m_freem(mcopy);
189 #endif
191 m_freem(m);
192 return;
195 error = 0;
197 /* check policy */
198 switch (sp->policy) {
199 case IPSEC_POLICY_DISCARD:
201 * This packet is just discarded.
203 ipsec6stat.out_polvio++;
204 ip6stat.ip6s_cantforward++;
205 key_freesp(sp);
206 if (mcopy) {
207 #if 0
208 /* XXX: what icmp ? */
209 #else
210 m_freem(mcopy);
211 #endif
213 m_freem(m);
214 return;
216 case IPSEC_POLICY_BYPASS:
217 case IPSEC_POLICY_NONE:
218 /* no need to do IPsec. */
219 key_freesp(sp);
220 goto skip_ipsec;
222 case IPSEC_POLICY_IPSEC:
223 if (sp->req == NULL) {
224 /* XXX should be panic ? */
225 kprintf("ip6_forward: No IPsec request specified.\n");
226 ip6stat.ip6s_cantforward++;
227 key_freesp(sp);
228 if (mcopy) {
229 #if 0
230 /* XXX: what icmp ? */
231 #else
232 m_freem(mcopy);
233 #endif
235 m_freem(m);
236 return;
238 /* do IPsec */
239 break;
241 case IPSEC_POLICY_ENTRUST:
242 default:
243 /* should be panic ?? */
244 kprintf("ip6_forward: Invalid policy found. %d\n", sp->policy);
245 key_freesp(sp);
246 goto skip_ipsec;
250 struct ipsec_output_state state;
253 * All the extension headers will become inaccessible
254 * (since they can be encrypted).
255 * Don't panic, we need no more updates to extension headers
256 * on inner IPv6 packet (since they are now encapsulated).
258 * IPv6 [ESP|AH] IPv6 [extension headers] payload
260 bzero(&state, sizeof(state));
261 state.m = m;
262 state.ro = NULL; /* update at ipsec6_output_tunnel() */
263 state.dst = NULL; /* update at ipsec6_output_tunnel() */
265 error = ipsec6_output_tunnel(&state, sp, 0);
267 m = state.m;
268 key_freesp(sp);
270 if (error) {
271 /* mbuf is already reclaimed in ipsec6_output_tunnel. */
272 switch (error) {
273 case EHOSTUNREACH:
274 case ENETUNREACH:
275 case EMSGSIZE:
276 case ENOBUFS:
277 case ENOMEM:
278 break;
279 default:
280 kprintf("ip6_output (ipsec): error code %d\n", error);
281 /* FALLTHROUGH */
282 case ENOENT:
283 /* don't show these error codes to the user */
284 break;
286 ip6stat.ip6s_cantforward++;
287 if (mcopy) {
288 #if 0
289 /* XXX: what icmp ? */
290 #else
291 m_freem(mcopy);
292 #endif
294 m_freem(m);
295 return;
298 skip_ipsec:
299 #endif /* IPSEC */
301 dst = (struct sockaddr_in6 *)&ip6_forward_rt.ro_dst;
302 if (!srcrt) {
303 /* ip6_forward_rt.ro_dst.sin6_addr is equal to ip6->ip6_dst */
304 if (ip6_forward_rt.ro_rt == NULL ||
305 !(ip6_forward_rt.ro_rt->rt_flags & RTF_UP)) {
306 if (ip6_forward_rt.ro_rt) {
307 RTFREE(ip6_forward_rt.ro_rt);
308 ip6_forward_rt.ro_rt = 0;
311 /* this probably fails but give it a try again */
312 rtalloc_ign((struct route *)&ip6_forward_rt,
313 RTF_PRCLONING);
316 if (ip6_forward_rt.ro_rt == NULL) {
317 ip6stat.ip6s_noroute++;
318 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_noroute);
319 if (mcopy) {
320 icmp6_error(mcopy, ICMP6_DST_UNREACH,
321 ICMP6_DST_UNREACH_NOROUTE, 0);
323 m_freem(m);
324 return;
326 } else if ((rt = ip6_forward_rt.ro_rt) == NULL ||
327 !IN6_ARE_ADDR_EQUAL(&ip6->ip6_dst, &dst->sin6_addr)) {
328 if (ip6_forward_rt.ro_rt) {
329 RTFREE(ip6_forward_rt.ro_rt);
330 ip6_forward_rt.ro_rt = 0;
332 bzero(dst, sizeof(*dst));
333 dst->sin6_len = sizeof(struct sockaddr_in6);
334 dst->sin6_family = AF_INET6;
335 dst->sin6_addr = ip6->ip6_dst;
337 rtalloc_ign((struct route *)&ip6_forward_rt, RTF_PRCLONING);
338 if (ip6_forward_rt.ro_rt == NULL) {
339 ip6stat.ip6s_noroute++;
340 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_noroute);
341 if (mcopy) {
342 icmp6_error(mcopy, ICMP6_DST_UNREACH,
343 ICMP6_DST_UNREACH_NOROUTE, 0);
345 m_freem(m);
346 return;
349 rt = ip6_forward_rt.ro_rt;
352 * Scope check: if a packet can't be delivered to its destination
353 * for the reason that the destination is beyond the scope of the
354 * source address, discard the packet and return an icmp6 destination
355 * unreachable error with Code 2 (beyond scope of source address).
356 * [draft-ietf-ipngwg-icmp-v3-02.txt, Section 3.1]
358 if (in6_addr2zoneid(m->m_pkthdr.rcvif, &ip6->ip6_src, &srczone) ||
359 in6_addr2zoneid(rt->rt_ifp, &ip6->ip6_src, &dstzone) ||
360 srczone != dstzone) {
361 ip6stat.ip6s_cantforward++;
362 ip6stat.ip6s_badscope++;
363 in6_ifstat_inc(rt->rt_ifp, ifs6_in_discard);
365 if (ip6_log_time + ip6_log_interval < time_uptime) {
366 ip6_log_time = time_uptime;
367 log(LOG_DEBUG,
368 "cannot forward "
369 "src %s, dst %s, nxt %d, rcvif %s, outif %s\n",
370 ip6_sprintf(&ip6->ip6_src),
371 ip6_sprintf(&ip6->ip6_dst),
372 ip6->ip6_nxt,
373 if_name(m->m_pkthdr.rcvif), if_name(rt->rt_ifp));
375 if (mcopy)
376 icmp6_error(mcopy, ICMP6_DST_UNREACH,
377 ICMP6_DST_UNREACH_BEYONDSCOPE, 0);
378 m_freem(m);
379 return;
382 if (m->m_pkthdr.len > IN6_LINKMTU(rt->rt_ifp)) {
383 in6_ifstat_inc(rt->rt_ifp, ifs6_in_toobig);
384 if (mcopy) {
385 u_long mtu;
386 #ifdef IPSEC
387 struct secpolicy *sp;
388 int ipsecerror;
389 size_t ipsechdrsiz;
390 #endif
392 mtu = IN6_LINKMTU(rt->rt_ifp);
393 #ifdef IPSEC
395 * When we do IPsec tunnel ingress, we need to play
396 * with if_mtu value (decrement IPsec header size
397 * from mtu value). The code is much simpler than v4
398 * case, as we have the outgoing interface for
399 * encapsulated packet as "rt->rt_ifp".
401 sp = ipsec6_getpolicybyaddr(mcopy, IPSEC_DIR_OUTBOUND,
402 IP_FORWARDING, &ipsecerror);
403 if (sp) {
404 ipsechdrsiz = ipsec6_hdrsiz(mcopy,
405 IPSEC_DIR_OUTBOUND, NULL);
406 if (ipsechdrsiz < mtu)
407 mtu -= ipsechdrsiz;
411 * if mtu becomes less than minimum MTU,
412 * tell minimum MTU (and I'll need to fragment it).
414 if (mtu < IPV6_MMTU)
415 mtu = IPV6_MMTU;
416 #endif
417 icmp6_error(mcopy, ICMP6_PACKET_TOO_BIG, 0, mtu);
419 m_freem(m);
420 return;
423 if (rt->rt_flags & RTF_GATEWAY)
424 dst = (struct sockaddr_in6 *)rt->rt_gateway;
427 * If we are to forward the packet using the same interface
428 * as one we got the packet from, perhaps we should send a redirect
429 * to sender to shortcut a hop.
430 * Only send redirect if source is sending directly to us,
431 * and if packet was not source routed (or has any options).
432 * Also, don't send redirect if forwarding using a route
433 * modified by a redirect.
435 if (rt->rt_ifp == m->m_pkthdr.rcvif && !srcrt &&
436 (rt->rt_flags & (RTF_DYNAMIC | RTF_MODIFIED)) == 0) {
437 if (rt->rt_ifp->if_flags & IFF_POINTOPOINT) {
439 * If the incoming interface is equal to the outgoing
440 * one, and the link attached to the interface is
441 * point-to-point, then it will be highly probable
442 * that a routing loop occurs. Thus, we immediately
443 * drop the packet and send an ICMPv6 error message.
445 * type/code is based on suggestion by Rich Draves.
446 * not sure if it is the best pick.
448 icmp6_error(mcopy, ICMP6_DST_UNREACH,
449 ICMP6_DST_UNREACH_ADDR, 0);
450 m_freem(m);
451 return;
453 type = ND_REDIRECT;
457 * Check with the firewall...
459 if (ip6_fw_enable && ip6_fw_chk_ptr) {
460 u_short port = 0;
461 /* If ipfw says divert, we have to just drop packet */
462 if ((*ip6_fw_chk_ptr)(&ip6, rt->rt_ifp, &port, &m)) {
463 m_freem(m);
464 goto freecopy;
466 if (!m)
467 goto freecopy;
471 * Fake scoped addresses. Note that even link-local source or
472 * destinaion can appear, if the originating node just sends the
473 * packet to us (without address resolution for the destination).
474 * Since both icmp6_error and icmp6_redirect_output fill the embedded
475 * link identifiers, we can do this stuff after making a copy for
476 * returning an error.
478 if (rt->rt_ifp->if_flags & IFF_LOOPBACK) {
480 * See corresponding comments in ip6_output.
481 * XXX: but is it possible that ip6_forward() sends a packet
482 * to a loopback interface? I don't think so, and thus
483 * I bark here. (jinmei@kame.net)
484 * XXX: it is common to route invalid packets to loopback.
485 * also, the codepath will be visited on use of ::1 in
486 * rthdr. (itojun)
488 #if 1
489 if (0)
490 #else
491 if ((rt->rt_flags & (RTF_BLACKHOLE | RTF_REJECT)) == 0)
492 #endif
494 kprintf("ip6_forward: outgoing interface is loopback. "
495 "src %s, dst %s, nxt %d, rcvif %s, outif %s\n",
496 ip6_sprintf(&ip6->ip6_src),
497 ip6_sprintf(&ip6->ip6_dst),
498 ip6->ip6_nxt, if_name(m->m_pkthdr.rcvif),
499 if_name(rt->rt_ifp));
502 /* we can just use rcvif in forwarding. */
503 origifp = m->m_pkthdr.rcvif;
505 else
506 origifp = rt->rt_ifp;
508 * clear embedded scope identifiers if necessary.
509 * in6_clearscope will touch the addresses only when necessary.
511 in6_clearscope(&ip6->ip6_src);
512 in6_clearscope(&ip6->ip6_dst);
515 * Run through list of hooks for output packets.
517 if (pfil_has_hooks(&inet6_pfil_hook)) {
518 error = pfil_run_hooks(&inet6_pfil_hook, &m, rt->rt_ifp,
519 PFIL_OUT);
520 if (error != 0)
521 goto senderr;
522 if (m == NULL)
523 goto freecopy;
524 ip6 = mtod(m, struct ip6_hdr *);
527 error = nd6_output(rt->rt_ifp, origifp, m, dst, rt);
528 if (error) {
529 in6_ifstat_inc(rt->rt_ifp, ifs6_out_discard);
530 ip6stat.ip6s_cantforward++;
531 } else {
532 ip6stat.ip6s_forward++;
533 in6_ifstat_inc(rt->rt_ifp, ifs6_out_forward);
534 if (type)
535 ip6stat.ip6s_redirectsent++;
536 else {
537 if (mcopy)
538 goto freecopy;
541 senderr:
542 if (mcopy == NULL)
543 return;
544 switch (error) {
545 case 0:
546 if (type == ND_REDIRECT) {
547 icmp6_redirect_output(mcopy, rt);
548 return;
550 goto freecopy;
552 case EMSGSIZE:
553 /* xxx MTU is constant in PPP? */
554 goto freecopy;
556 case ENOBUFS:
557 /* Tell source to slow down like source quench in IP? */
558 goto freecopy;
560 case ENETUNREACH: /* shouldn't happen, checked above */
561 case EHOSTUNREACH:
562 case ENETDOWN:
563 case EHOSTDOWN:
564 default:
565 type = ICMP6_DST_UNREACH;
566 code = ICMP6_DST_UNREACH_ADDR;
567 break;
569 icmp6_error(mcopy, type, code, 0);
570 return;
572 freecopy:
573 m_freem(mcopy);
574 return;