1 /*-
2 * Copyright (C) 1995, 1996, 1997, and 1998 WIDE Project.
3 * All rights reserved.
4 *
5 * Redistribution and use in source and binary forms, with or without
6 * modification, are permitted provided that the following conditions
7 * are met:
8 * 1. Redistributions of source code must retain the above copyright
9 * notice, this list of conditions and the following disclaimer.
10 * 2. Redistributions in binary form must reproduce the above copyright
11 * notice, this list of conditions and the following disclaimer in the
12 * documentation and/or other materials provided with the distribution.
13 * 3. Neither the name of the project nor the names of its contributors
14 * may be used to endorse or promote products derived from this software
15 * without specific prior written permission.
16 *
17 * THIS SOFTWARE IS PROVIDED BY THE PROJECT AND CONTRIBUTORS ``AS IS'' AND
18 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
19 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
20 * ARE DISCLAIMED. IN NO EVENT SHALL THE PROJECT OR CONTRIBUTORS BE LIABLE
21 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
22 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
23 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
24 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
25 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
26 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
27 * SUCH DAMAGE.
28 *
29 * $KAME: ip6_input.c,v 1.259 2002/01/21 04:58:09 jinmei Exp $
30 */
31
32 /*-
33 * Copyright (c) 1982, 1986, 1988, 1993
34 * The Regents of the University of California. All rights reserved.
35 *
36 * Redistribution and use in source and binary forms, with or without
37 * modification, are permitted provided that the following conditions
38 * are met:
39 * 1. Redistributions of source code must retain the above copyright
40 * notice, this list of conditions and the following disclaimer.
41 * 2. Redistributions in binary form must reproduce the above copyright
42 * notice, this list of conditions and the following disclaimer in the
43 * documentation and/or other materials provided with the distribution.
44 * 4. Neither the name of the University nor the names of its contributors
45 * may be used to endorse or promote products derived from this software
46 * without specific prior written permission.
47 *
48 * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
49 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
50 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
51 * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
52 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
53 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
54 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
55 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
56 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
57 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
58 * SUCH DAMAGE.
59 *
60 * @(#)ip_input.c 8.2 (Berkeley) 1/4/94
61 */
62
63 #include <sys/cdefs.h>
64 __FBSDID("$FreeBSD: releng/8.1/sys/netinet6/ip6_input.c 236953 2012-06-12 12:10:10Z bz $");
65
66 #include "opt_inet.h"
67 #include "opt_inet6.h"
68 #include "opt_ipsec.h"
69
70 #include <sys/param.h>
71 #include <sys/systm.h>
72 #include <sys/malloc.h>
73 #include <sys/mbuf.h>
74 #include <sys/proc.h>
75 #include <sys/domain.h>
76 #include <sys/protosw.h>
77 #include <sys/socket.h>
78 #include <sys/socketvar.h>
79 #include <sys/errno.h>
80 #include <sys/time.h>
81 #include <sys/kernel.h>
82 #include <sys/syslog.h>
83
84 #include <net/if.h>
85 #include <net/if_types.h>
86 #include <net/if_dl.h>
87 #include <net/route.h>
88 #include <net/netisr.h>
89 #include <net/pfil.h>
90 #include <net/vnet.h>
91
92 #include <netinet/in.h>
93 #include <netinet/in_systm.h>
94 #include <net/if_llatbl.h>
95 #ifdef INET
96 #include <netinet/ip.h>
97 #include <netinet/ip_icmp.h>
98 #endif /* INET */
99 #include <netinet/ip6.h>
100 #include <netinet6/in6_var.h>
101 #include <netinet6/ip6_var.h>
102 #include <netinet/in_pcb.h>
103 #include <netinet/icmp6.h>
104 #include <netinet6/scope6_var.h>
105 #include <netinet6/in6_ifattach.h>
106 #include <netinet6/nd6.h>
107
108 #ifdef IPSEC
109 #include <netipsec/ipsec.h>
110 #include <netinet6/ip6_ipsec.h>
111 #include <netipsec/ipsec6.h>
112 #endif /* IPSEC */
113
114 #include <netinet6/ip6protosw.h>
115
116 extern struct domain inet6domain;
117
118 u_char ip6_protox[IPPROTO_MAX];
119 VNET_DEFINE(struct in6_ifaddrhead, in6_ifaddrhead);
120
121 static struct netisr_handler ip6_nh = {
122 .nh_name = "ip6",
123 .nh_handler = ip6_input,
124 .nh_proto = NETISR_IPV6,
125 .nh_policy = NETISR_POLICY_FLOW,
126 };
127
128 VNET_DECLARE(struct callout, in6_tmpaddrtimer_ch);
129 #define V_in6_tmpaddrtimer_ch VNET(in6_tmpaddrtimer_ch)
130
131 VNET_DEFINE(struct pfil_head, inet6_pfil_hook);
132
133 VNET_DEFINE(struct ip6stat, ip6stat);
134
135 struct rwlock in6_ifaddr_lock;
136 RW_SYSINIT(in6_ifaddr_lock, &in6_ifaddr_lock, "in6_ifaddr_lock");
137
138 static void ip6_init2(void *);
139 static struct ip6aux *ip6_setdstifaddr(struct mbuf *, struct in6_ifaddr *);
140 static int ip6_hopopts_input(u_int32_t *, u_int32_t *, struct mbuf **, int *);
141 #ifdef PULLDOWN_TEST
142 static struct mbuf *ip6_pullexthdr(struct mbuf *, size_t, int);
143 #endif
144
145 /*
146 * IP6 initialization: fill in IP6 protocol switch table.
147 * All protocols not implemented in kernel go to raw IP6 protocol handler.
148 */
149 void
150 ip6_init(void)
151 {
152 struct ip6protosw *pr;
153 int i;
154
155 TUNABLE_INT_FETCH("net.inet6.ip6.auto_linklocal",
156 &V_ip6_auto_linklocal);
157
158 TAILQ_INIT(&V_in6_ifaddrhead);
159
160 /* Initialize packet filter hooks. */
161 V_inet6_pfil_hook.ph_type = PFIL_TYPE_AF;
162 V_inet6_pfil_hook.ph_af = AF_INET6;
163 if ((i = pfil_head_register(&V_inet6_pfil_hook)) != 0)
164 printf("%s: WARNING: unable to register pfil hook, "
165 "error %d\n", __func__, i);
166
167 scope6_init();
168 addrsel_policy_init();
169 nd6_init();
170 frag6_init();
171
172 V_ip6_desync_factor = arc4random() % MAX_TEMP_DESYNC_FACTOR;
173
174 /* Skip global initialization stuff for non-default instances. */
175 if (!IS_DEFAULT_VNET(curvnet))
176 return;
177
178 #ifdef DIAGNOSTIC
179 if (sizeof(struct protosw) != sizeof(struct ip6protosw))
180 panic("sizeof(protosw) != sizeof(ip6protosw)");
181 #endif
182 pr = (struct ip6protosw *)pffindproto(PF_INET6, IPPROTO_RAW, SOCK_RAW);
183 if (pr == NULL)
184 panic("ip6_init");
185
186 /* Initialize the entire ip6_protox[] array to IPPROTO_RAW. */
187 for (i = 0; i < IPPROTO_MAX; i++)
188 ip6_protox[i] = pr - inet6sw;
189 /*
190 * Cycle through IP protocols and put them into the appropriate place
191 * in ip6_protox[].
192 */
193 for (pr = (struct ip6protosw *)inet6domain.dom_protosw;
194 pr < (struct ip6protosw *)inet6domain.dom_protoswNPROTOSW; pr++)
195 if (pr->pr_domain->dom_family == PF_INET6 &&
196 pr->pr_protocol && pr->pr_protocol != IPPROTO_RAW) {
197 /* Be careful to only index valid IP protocols. */
198 if (pr->pr_protocol < IPPROTO_MAX)
199 ip6_protox[pr->pr_protocol] = pr - inet6sw;
200 }
201
202 netisr_register(&ip6_nh);
203 }
204
205 #ifdef VIMAGE
206 void
207 ip6_destroy()
208 {
209
210 nd6_destroy();
211 callout_drain(&V_in6_tmpaddrtimer_ch);
212 }
213 #endif
214
215 static int
216 ip6_init2_vnet(const void *unused __unused)
217 {
218
219 /* nd6_timer_init */
220 callout_init(&V_nd6_timer_ch, 0);
221 callout_reset(&V_nd6_timer_ch, hz, nd6_timer, curvnet);
222
223 /* timer for regeneranation of temporary addresses randomize ID */
224 callout_init(&V_in6_tmpaddrtimer_ch, 0);
225 callout_reset(&V_in6_tmpaddrtimer_ch,
226 (V_ip6_temp_preferred_lifetime - V_ip6_desync_factor -
227 V_ip6_temp_regen_advance) * hz,
228 in6_tmpaddrtimer, curvnet);
229
230 return (0);
231 }
232
233 static void
234 ip6_init2(void *dummy)
235 {
236
237 ip6_init2_vnet(NULL);
238 }
239
240 /* cheat */
241 /* This must be after route_init(), which is now SI_ORDER_THIRD */
242 SYSINIT(netinet6init2, SI_SUB_PROTO_DOMAIN, SI_ORDER_MIDDLE, ip6_init2, NULL);
243
244 void
245 ip6_input(struct mbuf *m)
246 {
247 struct ip6_hdr *ip6;
248 int off = sizeof(struct ip6_hdr), nest;
249 u_int32_t plen;
250 u_int32_t rtalert = ~0;
251 int nxt, ours = 0;
252 struct ifnet *deliverifp = NULL, *ifp = NULL;
253 struct in6_addr odst;
254 struct route_in6 rin6;
255 int srcrt = 0;
256 struct llentry *lle = NULL;
257 struct sockaddr_in6 dst6, *dst;
258
259 bzero(&rin6, sizeof(struct route_in6));
260 #ifdef IPSEC
261 /*
262 * should the inner packet be considered authentic?
263 * see comment in ah4_input().
264 * NB: m cannot be NULL when passed to the input routine
265 */
266
267 m->m_flags &= ~M_AUTHIPHDR;
268 m->m_flags &= ~M_AUTHIPDGM;
269
270 #endif /* IPSEC */
271
272 /*
273 * make sure we don't have onion peering information into m_tag.
274 */
275 ip6_delaux(m);
276
277 /*
278 * mbuf statistics
279 */
280 if (m->m_flags & M_EXT) {
281 if (m->m_next)
282 V_ip6stat.ip6s_mext2m++;
283 else
284 V_ip6stat.ip6s_mext1++;
285 } else {
286 #define M2MMAX (sizeof(V_ip6stat.ip6s_m2m)/sizeof(V_ip6stat.ip6s_m2m[0]))
287 if (m->m_next) {
288 if (m->m_flags & M_LOOP) {
289 V_ip6stat.ip6s_m2m[V_loif->if_index]++;
290 } else if (m->m_pkthdr.rcvif->if_index < M2MMAX)
291 V_ip6stat.ip6s_m2m[m->m_pkthdr.rcvif->if_index]++;
292 else
293 V_ip6stat.ip6s_m2m[0]++;
294 } else
295 V_ip6stat.ip6s_m1++;
296 #undef M2MMAX
297 }
298
299 /* drop the packet if IPv6 operation is disabled on the IF */
300 if ((ND_IFINFO(m->m_pkthdr.rcvif)->flags & ND6_IFF_IFDISABLED)) {
301 m_freem(m);
302 return;
303 }
304
305 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_receive);
306 V_ip6stat.ip6s_total++;
307
308 #ifndef PULLDOWN_TEST
309 /*
310 * L2 bridge code and some other code can return mbuf chain
311 * that does not conform to KAME requirement. too bad.
312 * XXX: fails to join if interface MTU > MCLBYTES. jumbogram?
313 */
314 if (m && m->m_next != NULL && m->m_pkthdr.len < MCLBYTES) {
315 struct mbuf *n;
316
317 MGETHDR(n, M_DONTWAIT, MT_HEADER);
318 if (n)
319 M_MOVE_PKTHDR(n, m);
320 if (n && n->m_pkthdr.len > MHLEN) {
321 MCLGET(n, M_DONTWAIT);
322 if ((n->m_flags & M_EXT) == 0) {
323 m_freem(n);
324 n = NULL;
325 }
326 }
327 if (n == NULL) {
328 m_freem(m);
329 return; /* ENOBUFS */
330 }
331
332 m_copydata(m, 0, n->m_pkthdr.len, mtod(n, caddr_t));
333 n->m_len = n->m_pkthdr.len;
334 m_freem(m);
335 m = n;
336 }
337 IP6_EXTHDR_CHECK(m, 0, sizeof(struct ip6_hdr), /* nothing */);
338 #endif
339
340 if (m->m_len < sizeof(struct ip6_hdr)) {
341 struct ifnet *inifp;
342 inifp = m->m_pkthdr.rcvif;
343 if ((m = m_pullup(m, sizeof(struct ip6_hdr))) == NULL) {
344 V_ip6stat.ip6s_toosmall++;
345 in6_ifstat_inc(inifp, ifs6_in_hdrerr);
346 return;
347 }
348 }
349
350 ip6 = mtod(m, struct ip6_hdr *);
351
352 if ((ip6->ip6_vfc & IPV6_VERSION_MASK) != IPV6_VERSION) {
353 V_ip6stat.ip6s_badvers++;
354 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_hdrerr);
355 goto bad;
356 }
357
358 V_ip6stat.ip6s_nxthist[ip6->ip6_nxt]++;
359
360 /*
361 * Check against address spoofing/corruption.
362 */
363 if (IN6_IS_ADDR_MULTICAST(&ip6->ip6_src) ||
364 IN6_IS_ADDR_UNSPECIFIED(&ip6->ip6_dst)) {
365 /*
366 * XXX: "badscope" is not very suitable for a multicast source.
367 */
368 V_ip6stat.ip6s_badscope++;
369 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_addrerr);
370 goto bad;
371 }
372 if (IN6_IS_ADDR_MC_INTFACELOCAL(&ip6->ip6_dst) &&
373 !(m->m_flags & M_LOOP)) {
374 /*
375 * In this case, the packet should come from the loopback
376 * interface. However, we cannot just check the if_flags,
377 * because ip6_mloopback() passes the "actual" interface
378 * as the outgoing/incoming interface.
379 */
380 V_ip6stat.ip6s_badscope++;
381 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_addrerr);
382 goto bad;
383 }
384
385 #ifdef ALTQ
386 if (altq_input != NULL && (*altq_input)(m, AF_INET6) == 0) {
387 /* packet is dropped by traffic conditioner */
388 return;
389 }
390 #endif
391 /*
392 * The following check is not documented in specs. A malicious
393 * party may be able to use IPv4 mapped addr to confuse tcp/udp stack
394 * and bypass security checks (act as if it was from 127.0.0.1 by using
395 * IPv6 src ::ffff:127.0.0.1). Be cautious.
396 *
397 * This check chokes if we are in an SIIT cloud. As none of BSDs
398 * support IPv4-less kernel compilation, we cannot support SIIT
399 * environment at all. So, it makes more sense for us to reject any
400 * malicious packets for non-SIIT environment, than try to do a
401 * partial support for SIIT environment.
402 */
403 if (IN6_IS_ADDR_V4MAPPED(&ip6->ip6_src) ||
404 IN6_IS_ADDR_V4MAPPED(&ip6->ip6_dst)) {
405 V_ip6stat.ip6s_badscope++;
406 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_addrerr);
407 goto bad;
408 }
409 #if 0
410 /*
411 * Reject packets with IPv4 compatible addresses (auto tunnel).
412 *
413 * The code forbids auto tunnel relay case in RFC1933 (the check is
414 * stronger than RFC1933). We may want to re-enable it if mech-xx
415 * is revised to forbid relaying case.
416 */
417 if (IN6_IS_ADDR_V4COMPAT(&ip6->ip6_src) ||
418 IN6_IS_ADDR_V4COMPAT(&ip6->ip6_dst)) {
419 V_ip6stat.ip6s_badscope++;
420 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_addrerr);
421 goto bad;
422 }
423 #endif
424
425 /*
426 * Run through list of hooks for input packets.
427 *
428 * NB: Beware of the destination address changing
429 * (e.g. by NAT rewriting). When this happens,
430 * tell ip6_forward to do the right thing.
431 */
432 odst = ip6->ip6_dst;
433
434 /* Jump over all PFIL processing if hooks are not active. */
435 if (!PFIL_HOOKED(&V_inet6_pfil_hook))
436 goto passin;
437
438 if (pfil_run_hooks(&V_inet6_pfil_hook, &m,
439 m->m_pkthdr.rcvif, PFIL_IN, NULL))
440 return;
441 if (m == NULL) /* consumed by filter */
442 return;
443 ip6 = mtod(m, struct ip6_hdr *);
444 srcrt = !IN6_ARE_ADDR_EQUAL(&odst, &ip6->ip6_dst);
445
446 passin:
447 /*
448 * Disambiguate address scope zones (if there is ambiguity).
449 * We first make sure that the original source or destination address
450 * is not in our internal form for scoped addresses. Such addresses
451 * are not necessarily invalid spec-wise, but we cannot accept them due
452 * to the usage conflict.
453 * in6_setscope() then also checks and rejects the cases where src or
454 * dst are the loopback address and the receiving interface
455 * is not loopback.
456 */
457 if (in6_clearscope(&ip6->ip6_src) || in6_clearscope(&ip6->ip6_dst)) {
458 V_ip6stat.ip6s_badscope++; /* XXX */
459 goto bad;
460 }
461 if (in6_setscope(&ip6->ip6_src, m->m_pkthdr.rcvif, NULL) ||
462 in6_setscope(&ip6->ip6_dst, m->m_pkthdr.rcvif, NULL)) {
463 V_ip6stat.ip6s_badscope++;
464 goto bad;
465 }
466
467 /*
468 * Multicast check. Assume packet is for us to avoid
469 * prematurely taking locks.
470 */
471 if (IN6_IS_ADDR_MULTICAST(&ip6->ip6_dst)) {
472 ours = 1;
473 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_mcast);
474 deliverifp = m->m_pkthdr.rcvif;
475 goto hbhcheck;
476 }
477
478 /*
479 * Unicast check
480 */
481
482 bzero(&dst6, sizeof(dst6));
483 dst6.sin6_family = AF_INET6;
484 dst6.sin6_len = sizeof(struct sockaddr_in6);
485 dst6.sin6_addr = ip6->ip6_dst;
486 ifp = m->m_pkthdr.rcvif;
487 IF_AFDATA_LOCK(ifp);
488 lle = lla_lookup(LLTABLE6(ifp), 0,
489 (struct sockaddr *)&dst6);
490 IF_AFDATA_UNLOCK(ifp);
491 if ((lle != NULL) && (lle->la_flags & LLE_IFADDR)) {
492 ours = 1;
493 deliverifp = ifp;
494 LLE_RUNLOCK(lle);
495 goto hbhcheck;
496 }
497 if (lle != NULL)
498 LLE_RUNLOCK(lle);
499
500 dst = &rin6.ro_dst;
501 dst->sin6_len = sizeof(struct sockaddr_in6);
502 dst->sin6_family = AF_INET6;
503 dst->sin6_addr = ip6->ip6_dst;
504 rin6.ro_rt = rtalloc1((struct sockaddr *)dst, 0, 0);
505 if (rin6.ro_rt)
506 RT_UNLOCK(rin6.ro_rt);
507
508 #define rt6_key(r) ((struct sockaddr_in6 *)((r)->rt_nodes->rn_key))
509
510 /*
511 * Accept the packet if the forwarding interface to the destination
512 * according to the routing table is the loopback interface,
513 * unless the associated route has a gateway.
514 * Note that this approach causes to accept a packet if there is a
515 * route to the loopback interface for the destination of the packet.
516 * But we think it's even useful in some situations, e.g. when using
517 * a special daemon which wants to intercept the packet.
518 *
519 * XXX: some OSes automatically make a cloned route for the destination
520 * of an outgoing packet. If the outgoing interface of the packet
521 * is a loopback one, the kernel would consider the packet to be
522 * accepted, even if we have no such address assinged on the interface.
523 * We check the cloned flag of the route entry to reject such cases,
524 * assuming that route entries for our own addresses are not made by
525 * cloning (it should be true because in6_addloop explicitly installs
526 * the host route). However, we might have to do an explicit check
527 * while it would be less efficient. Or, should we rather install a
528 * reject route for such a case?
529 */
530 if (rin6.ro_rt &&
531 (rin6.ro_rt->rt_flags &
532 (RTF_HOST|RTF_GATEWAY)) == RTF_HOST &&
533 #ifdef RTF_WASCLONED
534 !(rin6.ro_rt->rt_flags & RTF_WASCLONED) &&
535 #endif
536 #ifdef RTF_CLONED
537 !(rin6.ro_rt->rt_flags & RTF_CLONED) &&
538 #endif
539 #if 0
540 /*
541 * The check below is redundant since the comparison of
542 * the destination and the key of the rtentry has
543 * already done through looking up the routing table.
544 */
545 IN6_ARE_ADDR_EQUAL(&ip6->ip6_dst,
546 &rt6_key(rin6.ro_rt)->sin6_addr)
547 #endif
548 rin6.ro_rt->rt_ifp->if_type == IFT_LOOP) {
549 int free_ia6 = 0;
550 struct in6_ifaddr *ia6;
551
552 /*
553 * found the loopback route to the interface address
554 */
555 if (rin6.ro_rt->rt_gateway->sa_family == AF_LINK) {
556 struct sockaddr_in6 dest6;
557
558 bzero(&dest6, sizeof(dest6));
559 dest6.sin6_family = AF_INET6;
560 dest6.sin6_len = sizeof(dest6);
561 dest6.sin6_addr = ip6->ip6_dst;
562 ia6 = (struct in6_ifaddr *)
563 ifa_ifwithaddr((struct sockaddr *)&dest6);
564 if (ia6 == NULL)
565 goto bad;
566 free_ia6 = 1;
567 }
568 else
569 ia6 = (struct in6_ifaddr *)rin6.ro_rt->rt_ifa;
570
571 /*
572 * record address information into m_tag.
573 */
574 (void)ip6_setdstifaddr(m, ia6);
575
576 /*
577 * packets to a tentative, duplicated, or somehow invalid
578 * address must not be accepted.
579 */
580 if (!(ia6->ia6_flags & IN6_IFF_NOTREADY)) {
581 /* this address is ready */
582 ours = 1;
583 deliverifp = ia6->ia_ifp; /* correct? */
584 /* Count the packet in the ip address stats */
585 ia6->ia_ifa.if_ipackets++;
586 ia6->ia_ifa.if_ibytes += m->m_pkthdr.len;
587 if (ia6 != NULL && free_ia6 != 0)
588 ifa_free(&ia6->ia_ifa);
589 goto hbhcheck;
590 } else {
591 char ip6bufs[INET6_ADDRSTRLEN];
592 char ip6bufd[INET6_ADDRSTRLEN];
593 /* address is not ready, so discard the packet. */
594 nd6log((LOG_INFO,
595 "ip6_input: packet to an unready address %s->%s\n",
596 ip6_sprintf(ip6bufs, &ip6->ip6_src),
597 ip6_sprintf(ip6bufd, &ip6->ip6_dst)));
598
599 if (ia6 != NULL && free_ia6 != 0)
600 ifa_free(&ia6->ia_ifa);
601 goto bad;
602 }
603 }
604
605 /*
606 * FAITH (Firewall Aided Internet Translator)
607 */
608 if (V_ip6_keepfaith) {
609 if (rin6.ro_rt && rin6.ro_rt->rt_ifp &&
610 rin6.ro_rt->rt_ifp->if_type == IFT_FAITH) {
611 /* XXX do we need more sanity checks? */
612 ours = 1;
613 deliverifp = rin6.ro_rt->rt_ifp; /* faith */
614 goto hbhcheck;
615 }
616 }
617
618 /*
619 * Now there is no reason to process the packet if it's not our own
620 * and we're not a router.
621 */
622 if (!V_ip6_forwarding) {
623 V_ip6stat.ip6s_cantforward++;
624 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_discard);
625 goto bad;
626 }
627
628 hbhcheck:
629 /*
630 * record address information into m_tag, if we don't have one yet.
631 * note that we are unable to record it, if the address is not listed
632 * as our interface address (e.g. multicast addresses, addresses
633 * within FAITH prefixes and such).
634 */
635 if (deliverifp) {
636 struct in6_ifaddr *ia6;
637
638 if ((ia6 = ip6_getdstifaddr(m)) != NULL) {
639 ifa_free(&ia6->ia_ifa);
640 } else {
641 ia6 = in6_ifawithifp(deliverifp, &ip6->ip6_dst);
642 if (ia6) {
643 if (!ip6_setdstifaddr(m, ia6)) {
644 /*
645 * XXX maybe we should drop the packet here,
646 * as we could not provide enough information
647 * to the upper layers.
648 */
649 }
650 ifa_free(&ia6->ia_ifa);
651 }
652 }
653 }
654
655 /*
656 * Process Hop-by-Hop options header if it's contained.
657 * m may be modified in ip6_hopopts_input().
658 * If a JumboPayload option is included, plen will also be modified.
659 */
660 plen = (u_int32_t)ntohs(ip6->ip6_plen);
661 if (ip6->ip6_nxt == IPPROTO_HOPOPTS) {
662 struct ip6_hbh *hbh;
663
664 if (ip6_hopopts_input(&plen, &rtalert, &m, &off)) {
665 #if 0 /*touches NULL pointer*/
666 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_discard);
667 #endif
668 goto out; /* m have already been freed */
669 }
670
671 /* adjust pointer */
672 ip6 = mtod(m, struct ip6_hdr *);
673
674 /*
675 * if the payload length field is 0 and the next header field
676 * indicates Hop-by-Hop Options header, then a Jumbo Payload
677 * option MUST be included.
678 */
679 if (ip6->ip6_plen == 0 && plen == 0) {
680 /*
681 * Note that if a valid jumbo payload option is
682 * contained, ip6_hopopts_input() must set a valid
683 * (non-zero) payload length to the variable plen.
684 */
685 V_ip6stat.ip6s_badoptions++;
686 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_discard);
687 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_hdrerr);
688 icmp6_error(m, ICMP6_PARAM_PROB,
689 ICMP6_PARAMPROB_HEADER,
690 (caddr_t)&ip6->ip6_plen - (caddr_t)ip6);
691 goto out;
692 }
693 #ifndef PULLDOWN_TEST
694 /* ip6_hopopts_input() ensures that mbuf is contiguous */
695 hbh = (struct ip6_hbh *)(ip6 + 1);
696 #else
697 IP6_EXTHDR_GET(hbh, struct ip6_hbh *, m, sizeof(struct ip6_hdr),
698 sizeof(struct ip6_hbh));
699 if (hbh == NULL) {
700 V_ip6stat.ip6s_tooshort++;
701 goto out;
702 }
703 #endif
704 nxt = hbh->ip6h_nxt;
705
706 /*
707 * If we are acting as a router and the packet contains a
708 * router alert option, see if we know the option value.
709 * Currently, we only support the option value for MLD, in which
710 * case we should pass the packet to the multicast routing
711 * daemon.
712 */
713 if (rtalert != ~0) {
714 switch (rtalert) {
715 case IP6OPT_RTALERT_MLD:
716 if (V_ip6_forwarding)
717 ours = 1;
718 break;
719 default:
720 /*
721 * RFC2711 requires unrecognized values must be
722 * silently ignored.
723 */
724 break;
725 }
726 }
727 } else
728 nxt = ip6->ip6_nxt;
729
730 /*
731 * Check that the amount of data in the buffers
732 * is as at least much as the IPv6 header would have us expect.
733 * Trim mbufs if longer than we expect.
734 * Drop packet if shorter than we expect.
735 */
736 if (m->m_pkthdr.len - sizeof(struct ip6_hdr) < plen) {
737 V_ip6stat.ip6s_tooshort++;
738 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_truncated);
739 goto bad;
740 }
741 if (m->m_pkthdr.len > sizeof(struct ip6_hdr) + plen) {
742 if (m->m_len == m->m_pkthdr.len) {
743 m->m_len = sizeof(struct ip6_hdr) + plen;
744 m->m_pkthdr.len = sizeof(struct ip6_hdr) + plen;
745 } else
746 m_adj(m, sizeof(struct ip6_hdr) + plen - m->m_pkthdr.len);
747 }
748
749 /*
750 * Forward if desirable.
751 */
752 if (V_ip6_mrouter &&
753 IN6_IS_ADDR_MULTICAST(&ip6->ip6_dst)) {
754 /*
755 * If we are acting as a multicast router, all
756 * incoming multicast packets are passed to the
757 * kernel-level multicast forwarding function.
758 * The packet is returned (relatively) intact; if
759 * ip6_mforward() returns a non-zero value, the packet
760 * must be discarded, else it may be accepted below.
761 *
762 * XXX TODO: Check hlim and multicast scope here to avoid
763 * unnecessarily calling into ip6_mforward().
764 */
765 if (ip6_mforward &&
766 ip6_mforward(ip6, m->m_pkthdr.rcvif, m)) {
767 IP6STAT_INC(ip6s_cantforward);
768 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_discard);
769 goto bad;
770 }
771 } else if (!ours) {
772 ip6_forward(m, srcrt);
773 goto out;
774 }
775
776 ip6 = mtod(m, struct ip6_hdr *);
777
778 /*
779 * Malicious party may be able to use IPv4 mapped addr to confuse
780 * tcp/udp stack and bypass security checks (act as if it was from
781 * 127.0.0.1 by using IPv6 src ::ffff:127.0.0.1). Be cautious.
782 *
783 * For SIIT end node behavior, you may want to disable the check.
784 * However, you will become vulnerable to attacks using IPv4 mapped
785 * source.
786 */
787 if (IN6_IS_ADDR_V4MAPPED(&ip6->ip6_src) ||
788 IN6_IS_ADDR_V4MAPPED(&ip6->ip6_dst)) {
789 V_ip6stat.ip6s_badscope++;
790 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_addrerr);
791 goto bad;
792 }
793
794 /*
795 * Tell launch routine the next header
796 */
797 V_ip6stat.ip6s_delivered++;
798 in6_ifstat_inc(deliverifp, ifs6_in_deliver);
799 nest = 0;
800
801 while (nxt != IPPROTO_DONE) {
802 if (V_ip6_hdrnestlimit && (++nest > V_ip6_hdrnestlimit)) {
803 V_ip6stat.ip6s_toomanyhdr++;
804 goto bad;
805 }
806
807 /*
808 * protection against faulty packet - there should be
809 * more sanity checks in header chain processing.
810 */
811 if (m->m_pkthdr.len < off) {
812 V_ip6stat.ip6s_tooshort++;
813 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_truncated);
814 goto bad;
815 }
816
817 #ifdef IPSEC
818 /*
819 * enforce IPsec policy checking if we are seeing last header.
820 * note that we do not visit this with protocols with pcb layer
821 * code - like udp/tcp/raw ip.
822 */
823 if (ip6_ipsec_input(m, nxt))
824 goto bad;
825 #endif /* IPSEC */
826
827 /*
828 * Use mbuf flags to propagate Router Alert option to
829 * ICMPv6 layer, as hop-by-hop options have been stripped.
830 */
831 if (nxt == IPPROTO_ICMPV6 && rtalert != ~0)
832 m->m_flags |= M_RTALERT_MLD;
833
834 nxt = (*inet6sw[ip6_protox[nxt]].pr_input)(&m, &off, nxt);
835 }
836 goto out;
837 bad:
838 m_freem(m);
839 out:
840 if (rin6.ro_rt)
841 RTFREE(rin6.ro_rt);
842 }
843
844 /*
845 * set/grab in6_ifaddr correspond to IPv6 destination address.
846 * XXX backward compatibility wrapper
847 *
848 * XXXRW: We should bump the refcount on ia6 before sticking it in the m_tag,
849 * and then bump it when the tag is copied, and release it when the tag is
850 * freed. Unfortunately, m_tags don't support deep copies (yet), so instead
851 * we just bump the ia refcount when we receive it. This should be fixed.
852 */
853 static struct ip6aux *
854 ip6_setdstifaddr(struct mbuf *m, struct in6_ifaddr *ia6)
855 {
856 struct ip6aux *ip6a;
857
858 ip6a = ip6_addaux(m);
859 if (ip6a)
860 ip6a->ip6a_dstia6 = ia6;
861 return ip6a; /* NULL if failed to set */
862 }
863
864 struct in6_ifaddr *
865 ip6_getdstifaddr(struct mbuf *m)
866 {
867 struct ip6aux *ip6a;
868 struct in6_ifaddr *ia;
869
870 ip6a = ip6_findaux(m);
871 if (ip6a) {
872 ia = ip6a->ip6a_dstia6;
873 ifa_ref(&ia->ia_ifa);
874 return ia;
875 } else
876 return NULL;
877 }
878
879 /*
880 * Hop-by-Hop options header processing. If a valid jumbo payload option is
881 * included, the real payload length will be stored in plenp.
882 *
883 * rtalertp - XXX: should be stored more smart way
884 */
885 static int
886 ip6_hopopts_input(u_int32_t *plenp, u_int32_t *rtalertp,
887 struct mbuf **mp, int *offp)
888 {
889 struct mbuf *m = *mp;
890 int off = *offp, hbhlen;
891 struct ip6_hbh *hbh;
892 u_int8_t *opt;
893
894 /* validation of the length of the header */
895 #ifndef PULLDOWN_TEST
896 IP6_EXTHDR_CHECK(m, off, sizeof(*hbh), -1);
897 hbh = (struct ip6_hbh *)(mtod(m, caddr_t) + off);
898 hbhlen = (hbh->ip6h_len + 1) << 3;
899
900 IP6_EXTHDR_CHECK(m, off, hbhlen, -1);
901 hbh = (struct ip6_hbh *)(mtod(m, caddr_t) + off);
902 #else
903 IP6_EXTHDR_GET(hbh, struct ip6_hbh *, m,
904 sizeof(struct ip6_hdr), sizeof(struct ip6_hbh));
905 if (hbh == NULL) {
906 V_ip6stat.ip6s_tooshort++;
907 return -1;
908 }
909 hbhlen = (hbh->ip6h_len + 1) << 3;
910 IP6_EXTHDR_GET(hbh, struct ip6_hbh *, m, sizeof(struct ip6_hdr),
911 hbhlen);
912 if (hbh == NULL) {
913 V_ip6stat.ip6s_tooshort++;
914 return -1;
915 }
916 #endif
917 off += hbhlen;
918 hbhlen -= sizeof(struct ip6_hbh);
919 opt = (u_int8_t *)hbh + sizeof(struct ip6_hbh);
920
921 if (ip6_process_hopopts(m, (u_int8_t *)hbh + sizeof(struct ip6_hbh),
922 hbhlen, rtalertp, plenp) < 0)
923 return (-1);
924
925 *offp = off;
926 *mp = m;
927 return (0);
928 }
929
930 /*
931 * Search header for all Hop-by-hop options and process each option.
932 * This function is separate from ip6_hopopts_input() in order to
933 * handle a case where the sending node itself process its hop-by-hop
934 * options header. In such a case, the function is called from ip6_output().
935 *
936 * The function assumes that hbh header is located right after the IPv6 header
937 * (RFC2460 p7), opthead is pointer into data content in m, and opthead to
938 * opthead + hbhlen is located in continuous memory region.
939 */
940 int
941 ip6_process_hopopts(struct mbuf *m, u_int8_t *opthead, int hbhlen,
942 u_int32_t *rtalertp, u_int32_t *plenp)
943 {
944 struct ip6_hdr *ip6;
945 int optlen = 0;
946 u_int8_t *opt = opthead;
947 u_int16_t rtalert_val;
948 u_int32_t jumboplen;
949 const int erroff = sizeof(struct ip6_hdr) + sizeof(struct ip6_hbh);
950
951 for (; hbhlen > 0; hbhlen -= optlen, opt += optlen) {
952 switch (*opt) {
953 case IP6OPT_PAD1:
954 optlen = 1;
955 break;
956 case IP6OPT_PADN:
957 if (hbhlen < IP6OPT_MINLEN) {
958 V_ip6stat.ip6s_toosmall++;
959 goto bad;
960 }
961 optlen = *(opt + 1) + 2;
962 break;
963 case IP6OPT_ROUTER_ALERT:
964 /* XXX may need check for alignment */
965 if (hbhlen < IP6OPT_RTALERT_LEN) {
966 V_ip6stat.ip6s_toosmall++;
967 goto bad;
968 }
969 if (*(opt + 1) != IP6OPT_RTALERT_LEN - 2) {
970 /* XXX stat */
971 icmp6_error(m, ICMP6_PARAM_PROB,
972 ICMP6_PARAMPROB_HEADER,
973 erroff + opt + 1 - opthead);
974 return (-1);
975 }
976 optlen = IP6OPT_RTALERT_LEN;
977 bcopy((caddr_t)(opt + 2), (caddr_t)&rtalert_val, 2);
978 *rtalertp = ntohs(rtalert_val);
979 break;
980 case IP6OPT_JUMBO:
981 /* XXX may need check for alignment */
982 if (hbhlen < IP6OPT_JUMBO_LEN) {
983 V_ip6stat.ip6s_toosmall++;
984 goto bad;
985 }
986 if (*(opt + 1) != IP6OPT_JUMBO_LEN - 2) {
987 /* XXX stat */
988 icmp6_error(m, ICMP6_PARAM_PROB,
989 ICMP6_PARAMPROB_HEADER,
990 erroff + opt + 1 - opthead);
991 return (-1);
992 }
993 optlen = IP6OPT_JUMBO_LEN;
994
995 /*
996 * IPv6 packets that have non 0 payload length
997 * must not contain a jumbo payload option.
998 */
999 ip6 = mtod(m, struct ip6_hdr *);
1000 if (ip6->ip6_plen) {
1001 V_ip6stat.ip6s_badoptions++;
1002 icmp6_error(m, ICMP6_PARAM_PROB,
1003 ICMP6_PARAMPROB_HEADER,
1004 erroff + opt - opthead);
1005 return (-1);
1006 }
1007
1008 /*
1009 * We may see jumbolen in unaligned location, so
1010 * we'd need to perform bcopy().
1011 */
1012 bcopy(opt + 2, &jumboplen, sizeof(jumboplen));
1013 jumboplen = (u_int32_t)htonl(jumboplen);
1014
1015 #if 1
1016 /*
1017 * if there are multiple jumbo payload options,
1018 * *plenp will be non-zero and the packet will be
1019 * rejected.
1020 * the behavior may need some debate in ipngwg -
1021 * multiple options does not make sense, however,
1022 * there's no explicit mention in specification.
1023 */
1024 if (*plenp != 0) {
1025 V_ip6stat.ip6s_badoptions++;
1026 icmp6_error(m, ICMP6_PARAM_PROB,
1027 ICMP6_PARAMPROB_HEADER,
1028 erroff + opt + 2 - opthead);
1029 return (-1);
1030 }
1031 #endif
1032
1033 /*
1034 * jumbo payload length must be larger than 65535.
1035 */
1036 if (jumboplen <= IPV6_MAXPACKET) {
1037 V_ip6stat.ip6s_badoptions++;
1038 icmp6_error(m, ICMP6_PARAM_PROB,
1039 ICMP6_PARAMPROB_HEADER,
1040 erroff + opt + 2 - opthead);
1041 return (-1);
1042 }
1043 *plenp = jumboplen;
1044
1045 break;
1046 default: /* unknown option */
1047 if (hbhlen < IP6OPT_MINLEN) {
1048 V_ip6stat.ip6s_toosmall++;
1049 goto bad;
1050 }
1051 optlen = ip6_unknown_opt(opt, m,
1052 erroff + opt - opthead);
1053 if (optlen == -1)
1054 return (-1);
1055 optlen += 2;
1056 break;
1057 }
1058 }
1059
1060 return (0);
1061
1062 bad:
1063 m_freem(m);
1064 return (-1);
1065 }
1066
1067 /*
1068 * Unknown option processing.
1069 * The third argument `off' is the offset from the IPv6 header to the option,
1070 * which is necessary if the IPv6 header the and option header and IPv6 header
1071 * is not continuous in order to return an ICMPv6 error.
1072 */
1073 int
1074 ip6_unknown_opt(u_int8_t *optp, struct mbuf *m, int off)
1075 {
1076 struct ip6_hdr *ip6;
1077
1078 switch (IP6OPT_TYPE(*optp)) {
1079 case IP6OPT_TYPE_SKIP: /* ignore the option */
1080 return ((int)*(optp + 1));
1081 case IP6OPT_TYPE_DISCARD: /* silently discard */
1082 m_freem(m);
1083 return (-1);
1084 case IP6OPT_TYPE_FORCEICMP: /* send ICMP even if multicasted */
1085 V_ip6stat.ip6s_badoptions++;
1086 icmp6_error(m, ICMP6_PARAM_PROB, ICMP6_PARAMPROB_OPTION, off);
1087 return (-1);
1088 case IP6OPT_TYPE_ICMP: /* send ICMP if not multicasted */
1089 V_ip6stat.ip6s_badoptions++;
1090 ip6 = mtod(m, struct ip6_hdr *);
1091 if (IN6_IS_ADDR_MULTICAST(&ip6->ip6_dst) ||
1092 (m->m_flags & (M_BCAST|M_MCAST)))
1093 m_freem(m);
1094 else
1095 icmp6_error(m, ICMP6_PARAM_PROB,
1096 ICMP6_PARAMPROB_OPTION, off);
1097 return (-1);
1098 }
1099
1100 m_freem(m); /* XXX: NOTREACHED */
1101 return (-1);
1102 }
1103
1104 /*
1105 * Create the "control" list for this pcb.
1106 * These functions will not modify mbuf chain at all.
1107 *
1108 * With KAME mbuf chain restriction:
1109 * The routine will be called from upper layer handlers like tcp6_input().
1110 * Thus the routine assumes that the caller (tcp6_input) have already
1111 * called IP6_EXTHDR_CHECK() and all the extension headers are located in the
1112 * very first mbuf on the mbuf chain.
1113 *
1114 * ip6_savecontrol_v4 will handle those options that are possible to be
1115 * set on a v4-mapped socket.
1116 * ip6_savecontrol will directly call ip6_savecontrol_v4 to handle those
1117 * options and handle the v6-only ones itself.
1118 */
1119 struct mbuf **
1120 ip6_savecontrol_v4(struct inpcb *inp, struct mbuf *m, struct mbuf **mp,
1121 int *v4only)
1122 {
1123 struct ip6_hdr *ip6 = mtod(m, struct ip6_hdr *);
1124
1125 #ifdef SO_TIMESTAMP
1126 if ((inp->inp_socket->so_options & SO_TIMESTAMP) != 0) {
1127 struct timeval tv;
1128
1129 microtime(&tv);
1130 *mp = sbcreatecontrol((caddr_t) &tv, sizeof(tv),
1131 SCM_TIMESTAMP, SOL_SOCKET);
1132 if (*mp)
1133 mp = &(*mp)->m_next;
1134 }
1135 #endif
1136
1137 if ((ip6->ip6_vfc & IPV6_VERSION_MASK) != IPV6_VERSION) {
1138 if (v4only != NULL)
1139 *v4only = 1;
1140 return (mp);
1141 }
1142
1143 #define IS2292(inp, x, y) (((inp)->inp_flags & IN6P_RFC2292) ? (x) : (y))
1144 /* RFC 2292 sec. 5 */
1145 if ((inp->inp_flags & IN6P_PKTINFO) != 0) {
1146 struct in6_pktinfo pi6;
1147
1148 bcopy(&ip6->ip6_dst, &pi6.ipi6_addr, sizeof(struct in6_addr));
1149 in6_clearscope(&pi6.ipi6_addr); /* XXX */
1150 pi6.ipi6_ifindex =
1151 (m && m->m_pkthdr.rcvif) ? m->m_pkthdr.rcvif->if_index : 0;
1152
1153 *mp = sbcreatecontrol((caddr_t) &pi6,
1154 sizeof(struct in6_pktinfo),
1155 IS2292(inp, IPV6_2292PKTINFO, IPV6_PKTINFO), IPPROTO_IPV6);
1156 if (*mp)
1157 mp = &(*mp)->m_next;
1158 }
1159
1160 if ((inp->inp_flags & IN6P_HOPLIMIT) != 0) {
1161 int hlim = ip6->ip6_hlim & 0xff;
1162
1163 *mp = sbcreatecontrol((caddr_t) &hlim, sizeof(int),
1164 IS2292(inp, IPV6_2292HOPLIMIT, IPV6_HOPLIMIT),
1165 IPPROTO_IPV6);
1166 if (*mp)
1167 mp = &(*mp)->m_next;
1168 }
1169
1170 if (v4only != NULL)
1171 *v4only = 0;
1172 return (mp);
1173 }
1174
1175 void
1176 ip6_savecontrol(struct inpcb *in6p, struct mbuf *m, struct mbuf **mp)
1177 {
1178 struct ip6_hdr *ip6 = mtod(m, struct ip6_hdr *);
1179 int v4only = 0;
1180
1181 mp = ip6_savecontrol_v4(in6p, m, mp, &v4only);
1182 if (v4only)
1183 return;
1184
1185 if ((in6p->inp_flags & IN6P_TCLASS) != 0) {
1186 u_int32_t flowinfo;
1187 int tclass;
1188
1189 flowinfo = (u_int32_t)ntohl(ip6->ip6_flow & IPV6_FLOWINFO_MASK);
1190 flowinfo >>= 20;
1191
1192 tclass = flowinfo & 0xff;
1193 *mp = sbcreatecontrol((caddr_t) &tclass, sizeof(tclass),
1194 IPV6_TCLASS, IPPROTO_IPV6);
1195 if (*mp)
1196 mp = &(*mp)->m_next;
1197 }
1198
1199 /*
1200 * IPV6_HOPOPTS socket option. Recall that we required super-user
1201 * privilege for the option (see ip6_ctloutput), but it might be too
1202 * strict, since there might be some hop-by-hop options which can be
1203 * returned to normal user.
1204 * See also RFC 2292 section 6 (or RFC 3542 section 8).
1205 */
1206 if ((in6p->inp_flags & IN6P_HOPOPTS) != 0) {
1207 /*
1208 * Check if a hop-by-hop options header is contatined in the
1209 * received packet, and if so, store the options as ancillary
1210 * data. Note that a hop-by-hop options header must be
1211 * just after the IPv6 header, which is assured through the
1212 * IPv6 input processing.
1213 */
1214 if (ip6->ip6_nxt == IPPROTO_HOPOPTS) {
1215 struct ip6_hbh *hbh;
1216 int hbhlen = 0;
1217 #ifdef PULLDOWN_TEST
1218 struct mbuf *ext;
1219 #endif
1220
1221 #ifndef PULLDOWN_TEST
1222 hbh = (struct ip6_hbh *)(ip6 + 1);
1223 hbhlen = (hbh->ip6h_len + 1) << 3;
1224 #else
1225 ext = ip6_pullexthdr(m, sizeof(struct ip6_hdr),
1226 ip6->ip6_nxt);
1227 if (ext == NULL) {
1228 V_ip6stat.ip6s_tooshort++;
1229 return;
1230 }
1231 hbh = mtod(ext, struct ip6_hbh *);
1232 hbhlen = (hbh->ip6h_len + 1) << 3;
1233 if (hbhlen != ext->m_len) {
1234 m_freem(ext);
1235 V_ip6stat.ip6s_tooshort++;
1236 return;
1237 }
1238 #endif
1239
1240 /*
1241 * XXX: We copy the whole header even if a
1242 * jumbo payload option is included, the option which
1243 * is to be removed before returning according to
1244 * RFC2292.
1245 * Note: this constraint is removed in RFC3542
1246 */
1247 *mp = sbcreatecontrol((caddr_t)hbh, hbhlen,
1248 IS2292(in6p, IPV6_2292HOPOPTS, IPV6_HOPOPTS),
1249 IPPROTO_IPV6);
1250 if (*mp)
1251 mp = &(*mp)->m_next;
1252 #ifdef PULLDOWN_TEST
1253 m_freem(ext);
1254 #endif
1255 }
1256 }
1257
1258 if ((in6p->inp_flags & (IN6P_RTHDR | IN6P_DSTOPTS)) != 0) {
1259 int nxt = ip6->ip6_nxt, off = sizeof(struct ip6_hdr);
1260
1261 /*
1262 * Search for destination options headers or routing
1263 * header(s) through the header chain, and stores each
1264 * header as ancillary data.
1265 * Note that the order of the headers remains in
1266 * the chain of ancillary data.
1267 */
1268 while (1) { /* is explicit loop prevention necessary? */
1269 struct ip6_ext *ip6e = NULL;
1270 int elen;
1271 #ifdef PULLDOWN_TEST
1272 struct mbuf *ext = NULL;
1273 #endif
1274
1275 /*
1276 * if it is not an extension header, don't try to
1277 * pull it from the chain.
1278 */
1279 switch (nxt) {
1280 case IPPROTO_DSTOPTS:
1281 case IPPROTO_ROUTING:
1282 case IPPROTO_HOPOPTS:
1283 case IPPROTO_AH: /* is it possible? */
1284 break;
1285 default:
1286 goto loopend;
1287 }
1288
1289 #ifndef PULLDOWN_TEST
1290 if (off + sizeof(*ip6e) > m->m_len)
1291 goto loopend;
1292 ip6e = (struct ip6_ext *)(mtod(m, caddr_t) + off);
1293 if (nxt == IPPROTO_AH)
1294 elen = (ip6e->ip6e_len + 2) << 2;
1295 else
1296 elen = (ip6e->ip6e_len + 1) << 3;
1297 if (off + elen > m->m_len)
1298 goto loopend;
1299 #else
1300 ext = ip6_pullexthdr(m, off, nxt);
1301 if (ext == NULL) {
1302 V_ip6stat.ip6s_tooshort++;
1303 return;
1304 }
1305 ip6e = mtod(ext, struct ip6_ext *);
1306 if (nxt == IPPROTO_AH)
1307 elen = (ip6e->ip6e_len + 2) << 2;
1308 else
1309 elen = (ip6e->ip6e_len + 1) << 3;
1310 if (elen != ext->m_len) {
1311 m_freem(ext);
1312 V_ip6stat.ip6s_tooshort++;
1313 return;
1314 }
1315 #endif
1316
1317 switch (nxt) {
1318 case IPPROTO_DSTOPTS:
1319 if (!(in6p->inp_flags & IN6P_DSTOPTS))
1320 break;
1321
1322 *mp = sbcreatecontrol((caddr_t)ip6e, elen,
1323 IS2292(in6p,
1324 IPV6_2292DSTOPTS, IPV6_DSTOPTS),
1325 IPPROTO_IPV6);
1326 if (*mp)
1327 mp = &(*mp)->m_next;
1328 break;
1329 case IPPROTO_ROUTING:
1330 if (!in6p->inp_flags & IN6P_RTHDR)
1331 break;
1332
1333 *mp = sbcreatecontrol((caddr_t)ip6e, elen,
1334 IS2292(in6p, IPV6_2292RTHDR, IPV6_RTHDR),
1335 IPPROTO_IPV6);
1336 if (*mp)
1337 mp = &(*mp)->m_next;
1338 break;
1339 case IPPROTO_HOPOPTS:
1340 case IPPROTO_AH: /* is it possible? */
1341 break;
1342
1343 default:
1344 /*
1345 * other cases have been filtered in the above.
1346 * none will visit this case. here we supply
1347 * the code just in case (nxt overwritten or
1348 * other cases).
1349 */
1350 #ifdef PULLDOWN_TEST
1351 m_freem(ext);
1352 #endif
1353 goto loopend;
1354
1355 }
1356
1357 /* proceed with the next header. */
1358 off += elen;
1359 nxt = ip6e->ip6e_nxt;
1360 ip6e = NULL;
1361 #ifdef PULLDOWN_TEST
1362 m_freem(ext);
1363 ext = NULL;
1364 #endif
1365 }
1366 loopend:
1367 ;
1368 }
1369 }
1370 #undef IS2292
1371
1372 void
1373 ip6_notify_pmtu(struct inpcb *in6p, struct sockaddr_in6 *dst, u_int32_t *mtu)
1374 {
1375 struct socket *so;
1376 struct mbuf *m_mtu;
1377 struct ip6_mtuinfo mtuctl;
1378
1379 so = in6p->inp_socket;
1380
1381 if (mtu == NULL)
1382 return;
1383
1384 #ifdef DIAGNOSTIC
1385 if (so == NULL) /* I believe this is impossible */
1386 panic("ip6_notify_pmtu: socket is NULL");
1387 #endif
1388
1389 bzero(&mtuctl, sizeof(mtuctl)); /* zero-clear for safety */
1390 mtuctl.ip6m_mtu = *mtu;
1391 mtuctl.ip6m_addr = *dst;
1392 if (sa6_recoverscope(&mtuctl.ip6m_addr))
1393 return;
1394
1395 if ((m_mtu = sbcreatecontrol((caddr_t)&mtuctl, sizeof(mtuctl),
1396 IPV6_PATHMTU, IPPROTO_IPV6)) == NULL)
1397 return;
1398
1399 if (sbappendaddr(&so->so_rcv, (struct sockaddr *)dst, NULL, m_mtu)
1400 == 0) {
1401 m_freem(m_mtu);
1402 /* XXX: should count statistics */
1403 } else
1404 sorwakeup(so);
1405
1406 return;
1407 }
1408
1409 #ifdef PULLDOWN_TEST
1410 /*
1411 * pull single extension header from mbuf chain. returns single mbuf that
1412 * contains the result, or NULL on error.
1413 */
1414 static struct mbuf *
1415 ip6_pullexthdr(struct mbuf *m, size_t off, int nxt)
1416 {
1417 struct ip6_ext ip6e;
1418 size_t elen;
1419 struct mbuf *n;
1420
1421 #ifdef DIAGNOSTIC
1422 switch (nxt) {
1423 case IPPROTO_DSTOPTS:
1424 case IPPROTO_ROUTING:
1425 case IPPROTO_HOPOPTS:
1426 case IPPROTO_AH: /* is it possible? */
1427 break;
1428 default:
1429 printf("ip6_pullexthdr: invalid nxt=%d\n", nxt);
1430 }
1431 #endif
1432
1433 m_copydata(m, off, sizeof(ip6e), (caddr_t)&ip6e);
1434 if (nxt == IPPROTO_AH)
1435 elen = (ip6e.ip6e_len + 2) << 2;
1436 else
1437 elen = (ip6e.ip6e_len + 1) << 3;
1438
1439 MGET(n, M_DONTWAIT, MT_DATA);
1440 if (n && elen >= MLEN) {
1441 MCLGET(n, M_DONTWAIT);
1442 if ((n->m_flags & M_EXT) == 0) {
1443 m_free(n);
1444 n = NULL;
1445 }
1446 }
1447 if (!n)
1448 return NULL;
1449
1450 n->m_len = 0;
1451 if (elen >= M_TRAILINGSPACE(n)) {
1452 m_free(n);
1453 return NULL;
1454 }
1455
1456 m_copydata(m, off, elen, mtod(n, caddr_t));
1457 n->m_len = elen;
1458 return n;
1459 }
1460 #endif
1461
1462 /*
1463 * Get pointer to the previous header followed by the header
1464 * currently processed.
1465 * XXX: This function supposes that
1466 * M includes all headers,
1467 * the next header field and the header length field of each header
1468 * are valid, and
1469 * the sum of each header length equals to OFF.
1470 * Because of these assumptions, this function must be called very
1471 * carefully. Moreover, it will not be used in the near future when
1472 * we develop `neater' mechanism to process extension headers.
1473 */
1474 char *
1475 ip6_get_prevhdr(struct mbuf *m, int off)
1476 {
1477 struct ip6_hdr *ip6 = mtod(m, struct ip6_hdr *);
1478
1479 if (off == sizeof(struct ip6_hdr))
1480 return (&ip6->ip6_nxt);
1481 else {
1482 int len, nxt;
1483 struct ip6_ext *ip6e = NULL;
1484
1485 nxt = ip6->ip6_nxt;
1486 len = sizeof(struct ip6_hdr);
1487 while (len < off) {
1488 ip6e = (struct ip6_ext *)(mtod(m, caddr_t) + len);
1489
1490 switch (nxt) {
1491 case IPPROTO_FRAGMENT:
1492 len += sizeof(struct ip6_frag);
1493 break;
1494 case IPPROTO_AH:
1495 len += (ip6e->ip6e_len + 2) << 2;
1496 break;
1497 default:
1498 len += (ip6e->ip6e_len + 1) << 3;
1499 break;
1500 }
1501 nxt = ip6e->ip6e_nxt;
1502 }
1503 if (ip6e)
1504 return (&ip6e->ip6e_nxt);
1505 else
1506 return NULL;
1507 }
1508 }
1509
1510 /*
1511 * get next header offset. m will be retained.
1512 */
1513 int
1514 ip6_nexthdr(struct mbuf *m, int off, int proto, int *nxtp)
1515 {
1516 struct ip6_hdr ip6;
1517 struct ip6_ext ip6e;
1518 struct ip6_frag fh;
1519
1520 /* just in case */
1521 if (m == NULL)
1522 panic("ip6_nexthdr: m == NULL");
1523 if ((m->m_flags & M_PKTHDR) == 0 || m->m_pkthdr.len < off)
1524 return -1;
1525
1526 switch (proto) {
1527 case IPPROTO_IPV6:
1528 if (m->m_pkthdr.len < off + sizeof(ip6))
1529 return -1;
1530 m_copydata(m, off, sizeof(ip6), (caddr_t)&ip6);
1531 if (nxtp)
1532 *nxtp = ip6.ip6_nxt;
1533 off += sizeof(ip6);
1534 return off;
1535
1536 case IPPROTO_FRAGMENT:
1537 /*
1538 * terminate parsing if it is not the first fragment,
1539 * it does not make sense to parse through it.
1540 */
1541 if (m->m_pkthdr.len < off + sizeof(fh))
1542 return -1;
1543 m_copydata(m, off, sizeof(fh), (caddr_t)&fh);
1544 /* IP6F_OFF_MASK = 0xfff8(BigEndian), 0xf8ff(LittleEndian) */
1545 if (fh.ip6f_offlg & IP6F_OFF_MASK)
1546 return -1;
1547 if (nxtp)
1548 *nxtp = fh.ip6f_nxt;
1549 off += sizeof(struct ip6_frag);
1550 return off;
1551
1552 case IPPROTO_AH:
1553 if (m->m_pkthdr.len < off + sizeof(ip6e))
1554 return -1;
1555 m_copydata(m, off, sizeof(ip6e), (caddr_t)&ip6e);
1556 if (nxtp)
1557 *nxtp = ip6e.ip6e_nxt;
1558 off += (ip6e.ip6e_len + 2) << 2;
1559 return off;
1560
1561 case IPPROTO_HOPOPTS:
1562 case IPPROTO_ROUTING:
1563 case IPPROTO_DSTOPTS:
1564 if (m->m_pkthdr.len < off + sizeof(ip6e))
1565 return -1;
1566 m_copydata(m, off, sizeof(ip6e), (caddr_t)&ip6e);
1567 if (nxtp)
1568 *nxtp = ip6e.ip6e_nxt;
1569 off += (ip6e.ip6e_len + 1) << 3;
1570 return off;
1571
1572 case IPPROTO_NONE:
1573 case IPPROTO_ESP:
1574 case IPPROTO_IPCOMP:
1575 /* give up */
1576 return -1;
1577
1578 default:
1579 return -1;
1580 }
1581
1582 return -1;
1583 }
1584
1585 /*
1586 * get offset for the last header in the chain. m will be kept untainted.
1587 */
1588 int
1589 ip6_lasthdr(struct mbuf *m, int off, int proto, int *nxtp)
1590 {
1591 int newoff;
1592 int nxt;
1593
1594 if (!nxtp) {
1595 nxt = -1;
1596 nxtp = &nxt;
1597 }
1598 while (1) {
1599 newoff = ip6_nexthdr(m, off, proto, nxtp);
1600 if (newoff < 0)
1601 return off;
1602 else if (newoff < off)
1603 return -1; /* invalid */
1604 else if (newoff == off)
1605 return newoff;
1606
1607 off = newoff;
1608 proto = *nxtp;
1609 }
1610 }
1611
1612 struct ip6aux *
1613 ip6_addaux(struct mbuf *m)
1614 {
1615 struct m_tag *mtag;
1616
1617 mtag = m_tag_find(m, PACKET_TAG_IPV6_INPUT, NULL);
1618 if (!mtag) {
1619 mtag = m_tag_get(PACKET_TAG_IPV6_INPUT, sizeof(struct ip6aux),
1620 M_NOWAIT);
1621 if (mtag) {
1622 m_tag_prepend(m, mtag);
1623 bzero(mtag + 1, sizeof(struct ip6aux));
1624 }
1625 }
1626 return mtag ? (struct ip6aux *)(mtag + 1) : NULL;
1627 }
1628
1629 struct ip6aux *
1630 ip6_findaux(struct mbuf *m)
1631 {
1632 struct m_tag *mtag;
1633
1634 mtag = m_tag_find(m, PACKET_TAG_IPV6_INPUT, NULL);
1635 return mtag ? (struct ip6aux *)(mtag + 1) : NULL;
1636 }
1637
1638 void
1639 ip6_delaux(struct mbuf *m)
1640 {
1641 struct m_tag *mtag;
1642
1643 mtag = m_tag_find(m, PACKET_TAG_IPV6_INPUT, NULL);
1644 if (mtag)
1645 m_tag_delete(m, mtag);
1646 }
1647
1648 /*
1649 * System control for IP6
1650 */
1651
1652 u_char inet6ctlerrmap[PRC_NCMDS] = {
1653 0, 0, 0, 0,
1654 0, EMSGSIZE, EHOSTDOWN, EHOSTUNREACH,
1655 EHOSTUNREACH, EHOSTUNREACH, ECONNREFUSED, ECONNREFUSED,
1656 EMSGSIZE, EHOSTUNREACH, 0, 0,
1657 0, 0, 0, 0,
1658 ENOPROTOOPT
1659 };
Cache object: 1244bb4eff2a908198406ce610166759
|