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("$FreeBSD$"); 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 91 #include <netinet/in.h> 92 #include <netinet/in_systm.h> 93 #ifdef INET 94 #include <netinet/ip.h> 95 #include <netinet/ip_icmp.h> 96 #endif /* INET */ 97 #include <netinet/ip6.h> 98 #include <netinet6/in6_var.h> 99 #include <netinet6/ip6_var.h> 100 #include <netinet/in_pcb.h> 101 #include <netinet/icmp6.h> 102 #include <netinet6/scope6_var.h> 103 #include <netinet6/in6_ifattach.h> 104 #include <netinet6/nd6.h> 105 106 #ifdef IPSEC 107 #include <netipsec/ipsec.h> 108 #include <netinet6/ip6_ipsec.h> 109 #include <netipsec/ipsec6.h> 110 #endif /* IPSEC */ 111 112 #include <netinet6/ip6protosw.h> 113 114 extern struct domain inet6domain; 115 116 u_char ip6_protox[IPPROTO_MAX]; 117 static struct ifqueue ip6intrq; 118 static int ip6qmaxlen = IFQ_MAXLEN; 119 struct in6_ifaddr *in6_ifaddr; 120 121 extern struct callout in6_tmpaddrtimer_ch; 122 123 int ip6_forward_srcrt; /* XXX */ 124 int ip6_sourcecheck; /* XXX */ 125 int ip6_sourcecheck_interval; /* XXX */ 126 127 int ip6_ours_check_algorithm; 128 129 struct pfil_head inet6_pfil_hook; 130 131 struct ip6stat ip6stat; 132 133 static void ip6_init2(void *); 134 static struct ip6aux *ip6_setdstifaddr(struct mbuf *, struct in6_ifaddr *); 135 static int ip6_hopopts_input(u_int32_t *, u_int32_t *, struct mbuf **, int *); 136 #ifdef PULLDOWN_TEST 137 static struct mbuf *ip6_pullexthdr(struct mbuf *, size_t, int); 138 #endif 139 140 /* 141 * IP6 initialization: fill in IP6 protocol switch table. 142 * All protocols not implemented in kernel go to raw IP6 protocol handler. 143 */ 144 void 145 ip6_init(void) 146 { 147 struct ip6protosw *pr; 148 int i; 149 150 #ifdef DIAGNOSTIC 151 if (sizeof(struct protosw) != sizeof(struct ip6protosw)) 152 panic("sizeof(protosw) != sizeof(ip6protosw)"); 153 #endif 154 pr = (struct ip6protosw *)pffindproto(PF_INET6, IPPROTO_RAW, SOCK_RAW); 155 if (pr == 0) 156 panic("ip6_init"); 157 158 /* Initialize the entire ip_protox[] array to IPPROTO_RAW. */ 159 for (i = 0; i < IPPROTO_MAX; i++) 160 ip6_protox[i] = pr - inet6sw; 161 /* 162 * Cycle through IP protocols and put them into the appropriate place 163 * in ip6_protox[]. 164 */ 165 for (pr = (struct ip6protosw *)inet6domain.dom_protosw; 166 pr < (struct ip6protosw *)inet6domain.dom_protoswNPROTOSW; pr++) 167 if (pr->pr_domain->dom_family == PF_INET6 && 168 pr->pr_protocol && pr->pr_protocol != IPPROTO_RAW) { 169 /* Be careful to only index valid IP protocols. */ 170 if (pr->pr_protocol < IPPROTO_MAX) 171 ip6_protox[pr->pr_protocol] = pr - inet6sw; 172 } 173 174 /* Initialize packet filter hooks. */ 175 inet6_pfil_hook.ph_type = PFIL_TYPE_AF; 176 inet6_pfil_hook.ph_af = AF_INET6; 177 if ((i = pfil_head_register(&inet6_pfil_hook)) != 0) 178 printf("%s: WARNING: unable to register pfil hook, " 179 "error %d\n", __func__, i); 180 181 ip6intrq.ifq_maxlen = ip6qmaxlen; 182 mtx_init(&ip6intrq.ifq_mtx, "ip6_inq", NULL, MTX_DEF); 183 netisr_register(NETISR_IPV6, ip6_input, &ip6intrq, 0); 184 scope6_init(); 185 addrsel_policy_init(); 186 nd6_init(); 187 frag6_init(); 188 ip6_desync_factor = arc4random() % MAX_TEMP_DESYNC_FACTOR; 189 } 190 191 static void 192 ip6_init2(void *dummy) 193 { 194 195 /* nd6_timer_init */ 196 callout_init(&nd6_timer_ch, 0); 197 callout_reset(&nd6_timer_ch, hz, nd6_timer, NULL); 198 199 /* timer for regeneranation of temporary addresses randomize ID */ 200 callout_init(&in6_tmpaddrtimer_ch, 0); 201 callout_reset(&in6_tmpaddrtimer_ch, 202 (ip6_temp_preferred_lifetime - ip6_desync_factor - 203 ip6_temp_regen_advance) * hz, 204 in6_tmpaddrtimer, NULL); 205 } 206 207 /* cheat */ 208 /* This must be after route_init(), which is now SI_ORDER_THIRD */ 209 SYSINIT(netinet6init2, SI_SUB_PROTO_DOMAIN, SI_ORDER_MIDDLE, ip6_init2, NULL); 210 211 extern struct route_in6 ip6_forward_rt; 212 213 void 214 ip6_input(struct mbuf *m) 215 { 216 struct ip6_hdr *ip6; 217 int off = sizeof(struct ip6_hdr), nest; 218 u_int32_t plen; 219 u_int32_t rtalert = ~0; 220 int nxt, ours = 0; 221 struct ifnet *deliverifp = NULL; 222 struct in6_addr odst; 223 int srcrt = 0; 224 225 GIANT_REQUIRED; /* XXX for now */ 226 227 #ifdef IPSEC 228 /* 229 * should the inner packet be considered authentic? 230 * see comment in ah4_input(). 231 * NB: m cannot be NULL when passed to the input routine 232 */ 233 234 m->m_flags &= ~M_AUTHIPHDR; 235 m->m_flags &= ~M_AUTHIPDGM; 236 237 #endif /* IPSEC */ 238 239 /* 240 * make sure we don't have onion peering information into m_tag. 241 */ 242 ip6_delaux(m); 243 244 /* 245 * mbuf statistics 246 */ 247 if (m->m_flags & M_EXT) { 248 if (m->m_next) 249 ip6stat.ip6s_mext2m++; 250 else 251 ip6stat.ip6s_mext1++; 252 } else { 253 #define M2MMAX (sizeof(ip6stat.ip6s_m2m)/sizeof(ip6stat.ip6s_m2m[0])) 254 if (m->m_next) { 255 if (m->m_flags & M_LOOP) { 256 ip6stat.ip6s_m2m[loif[0].if_index]++; /* XXX */ 257 } else if (m->m_pkthdr.rcvif->if_index < M2MMAX) 258 ip6stat.ip6s_m2m[m->m_pkthdr.rcvif->if_index]++; 259 else 260 ip6stat.ip6s_m2m[0]++; 261 } else 262 ip6stat.ip6s_m1++; 263 #undef M2MMAX 264 } 265 266 /* drop the packet if IPv6 operation is disabled on the IF */ 267 if ((ND_IFINFO(m->m_pkthdr.rcvif)->flags & ND6_IFF_IFDISABLED)) { 268 m_freem(m); 269 return; 270 } 271 272 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_receive); 273 ip6stat.ip6s_total++; 274 275 #ifndef PULLDOWN_TEST 276 /* 277 * L2 bridge code and some other code can return mbuf chain 278 * that does not conform to KAME requirement. too bad. 279 * XXX: fails to join if interface MTU > MCLBYTES. jumbogram? 280 */ 281 if (m && m->m_next != NULL && m->m_pkthdr.len < MCLBYTES) { 282 struct mbuf *n; 283 284 MGETHDR(n, M_DONTWAIT, MT_HEADER); 285 if (n) 286 M_MOVE_PKTHDR(n, m); 287 if (n && n->m_pkthdr.len > MHLEN) { 288 MCLGET(n, M_DONTWAIT); 289 if ((n->m_flags & M_EXT) == 0) { 290 m_freem(n); 291 n = NULL; 292 } 293 } 294 if (n == NULL) { 295 m_freem(m); 296 return; /* ENOBUFS */ 297 } 298 299 m_copydata(m, 0, n->m_pkthdr.len, mtod(n, caddr_t)); 300 n->m_len = n->m_pkthdr.len; 301 m_freem(m); 302 m = n; 303 } 304 IP6_EXTHDR_CHECK(m, 0, sizeof(struct ip6_hdr), /* nothing */); 305 #endif 306 307 if (m->m_len < sizeof(struct ip6_hdr)) { 308 struct ifnet *inifp; 309 inifp = m->m_pkthdr.rcvif; 310 if ((m = m_pullup(m, sizeof(struct ip6_hdr))) == NULL) { 311 ip6stat.ip6s_toosmall++; 312 in6_ifstat_inc(inifp, ifs6_in_hdrerr); 313 return; 314 } 315 } 316 317 ip6 = mtod(m, struct ip6_hdr *); 318 319 if ((ip6->ip6_vfc & IPV6_VERSION_MASK) != IPV6_VERSION) { 320 ip6stat.ip6s_badvers++; 321 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_hdrerr); 322 goto bad; 323 } 324 325 ip6stat.ip6s_nxthist[ip6->ip6_nxt]++; 326 327 /* 328 * Check against address spoofing/corruption. 329 */ 330 if (IN6_IS_ADDR_MULTICAST(&ip6->ip6_src) || 331 IN6_IS_ADDR_UNSPECIFIED(&ip6->ip6_dst)) { 332 /* 333 * XXX: "badscope" is not very suitable for a multicast source. 334 */ 335 ip6stat.ip6s_badscope++; 336 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_addrerr); 337 goto bad; 338 } 339 if (IN6_IS_ADDR_MC_INTFACELOCAL(&ip6->ip6_dst) && 340 !(m->m_flags & M_LOOP)) { 341 /* 342 * In this case, the packet should come from the loopback 343 * interface. However, we cannot just check the if_flags, 344 * because ip6_mloopback() passes the "actual" interface 345 * as the outgoing/incoming interface. 346 */ 347 ip6stat.ip6s_badscope++; 348 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_addrerr); 349 goto bad; 350 } 351 352 #ifdef ALTQ 353 if (altq_input != NULL && (*altq_input)(m, AF_INET6) == 0) { 354 /* packet is dropped by traffic conditioner */ 355 return; 356 } 357 #endif 358 /* 359 * The following check is not documented in specs. A malicious 360 * party may be able to use IPv4 mapped addr to confuse tcp/udp stack 361 * and bypass security checks (act as if it was from 127.0.0.1 by using 362 * IPv6 src ::ffff:127.0.0.1). Be cautious. 363 * 364 * This check chokes if we are in an SIIT cloud. As none of BSDs 365 * support IPv4-less kernel compilation, we cannot support SIIT 366 * environment at all. So, it makes more sense for us to reject any 367 * malicious packets for non-SIIT environment, than try to do a 368 * partial support for SIIT environment. 369 */ 370 if (IN6_IS_ADDR_V4MAPPED(&ip6->ip6_src) || 371 IN6_IS_ADDR_V4MAPPED(&ip6->ip6_dst)) { 372 ip6stat.ip6s_badscope++; 373 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_addrerr); 374 goto bad; 375 } 376 #if 0 377 /* 378 * Reject packets with IPv4 compatible addresses (auto tunnel). 379 * 380 * The code forbids auto tunnel relay case in RFC1933 (the check is 381 * stronger than RFC1933). We may want to re-enable it if mech-xx 382 * is revised to forbid relaying case. 383 */ 384 if (IN6_IS_ADDR_V4COMPAT(&ip6->ip6_src) || 385 IN6_IS_ADDR_V4COMPAT(&ip6->ip6_dst)) { 386 ip6stat.ip6s_badscope++; 387 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_addrerr); 388 goto bad; 389 } 390 #endif 391 392 /* 393 * Run through list of hooks for input packets. 394 * 395 * NB: Beware of the destination address changing 396 * (e.g. by NAT rewriting). When this happens, 397 * tell ip6_forward to do the right thing. 398 */ 399 odst = ip6->ip6_dst; 400 401 /* Jump over all PFIL processing if hooks are not active. */ 402 if (!PFIL_HOOKED(&inet6_pfil_hook)) 403 goto passin; 404 405 if (pfil_run_hooks(&inet6_pfil_hook, &m, m->m_pkthdr.rcvif, PFIL_IN, NULL)) 406 return; 407 if (m == NULL) /* consumed by filter */ 408 return; 409 ip6 = mtod(m, struct ip6_hdr *); 410 srcrt = !IN6_ARE_ADDR_EQUAL(&odst, &ip6->ip6_dst); 411 412 passin: 413 /* 414 * Disambiguate address scope zones (if there is ambiguity). 415 * We first make sure that the original source or destination address 416 * is not in our internal form for scoped addresses. Such addresses 417 * are not necessarily invalid spec-wise, but we cannot accept them due 418 * to the usage conflict. 419 * in6_setscope() then also checks and rejects the cases where src or 420 * dst are the loopback address and the receiving interface 421 * is not loopback. 422 */ 423 if (in6_clearscope(&ip6->ip6_src) || in6_clearscope(&ip6->ip6_dst)) { 424 ip6stat.ip6s_badscope++; /* XXX */ 425 goto bad; 426 } 427 if (in6_setscope(&ip6->ip6_src, m->m_pkthdr.rcvif, NULL) || 428 in6_setscope(&ip6->ip6_dst, m->m_pkthdr.rcvif, NULL)) { 429 ip6stat.ip6s_badscope++; 430 goto bad; 431 } 432 433 /* 434 * Multicast check 435 */ 436 if (IN6_IS_ADDR_MULTICAST(&ip6->ip6_dst)) { 437 struct in6_multi *in6m = 0; 438 439 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_mcast); 440 /* 441 * See if we belong to the destination multicast group on the 442 * arrival interface. 443 */ 444 IN6_LOOKUP_MULTI(ip6->ip6_dst, m->m_pkthdr.rcvif, in6m); 445 if (in6m) 446 ours = 1; 447 else if (!ip6_mrouter) { 448 ip6stat.ip6s_notmember++; 449 ip6stat.ip6s_cantforward++; 450 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_discard); 451 goto bad; 452 } 453 deliverifp = m->m_pkthdr.rcvif; 454 goto hbhcheck; 455 } 456 457 /* 458 * Unicast check 459 */ 460 if (ip6_forward_rt.ro_rt != NULL && 461 (ip6_forward_rt.ro_rt->rt_flags & RTF_UP) != 0 && 462 IN6_ARE_ADDR_EQUAL(&ip6->ip6_dst, 463 &((struct sockaddr_in6 *)(&ip6_forward_rt.ro_dst))->sin6_addr)) 464 ip6stat.ip6s_forward_cachehit++; 465 else { 466 struct sockaddr_in6 *dst6; 467 468 if (ip6_forward_rt.ro_rt) { 469 /* route is down or destination is different */ 470 ip6stat.ip6s_forward_cachemiss++; 471 RTFREE(ip6_forward_rt.ro_rt); 472 ip6_forward_rt.ro_rt = 0; 473 } 474 475 bzero(&ip6_forward_rt.ro_dst, sizeof(struct sockaddr_in6)); 476 dst6 = (struct sockaddr_in6 *)&ip6_forward_rt.ro_dst; 477 dst6->sin6_len = sizeof(struct sockaddr_in6); 478 dst6->sin6_family = AF_INET6; 479 dst6->sin6_addr = ip6->ip6_dst; 480 481 rtalloc((struct route *)&ip6_forward_rt); 482 } 483 484 #define rt6_key(r) ((struct sockaddr_in6 *)((r)->rt_nodes->rn_key)) 485 486 /* 487 * Accept the packet if the forwarding interface to the destination 488 * according to the routing table is the loopback interface, 489 * unless the associated route has a gateway. 490 * Note that this approach causes to accept a packet if there is a 491 * route to the loopback interface for the destination of the packet. 492 * But we think it's even useful in some situations, e.g. when using 493 * a special daemon which wants to intercept the packet. 494 * 495 * XXX: some OSes automatically make a cloned route for the destination 496 * of an outgoing packet. If the outgoing interface of the packet 497 * is a loopback one, the kernel would consider the packet to be 498 * accepted, even if we have no such address assinged on the interface. 499 * We check the cloned flag of the route entry to reject such cases, 500 * assuming that route entries for our own addresses are not made by 501 * cloning (it should be true because in6_addloop explicitly installs 502 * the host route). However, we might have to do an explicit check 503 * while it would be less efficient. Or, should we rather install a 504 * reject route for such a case? 505 */ 506 if (ip6_forward_rt.ro_rt && 507 (ip6_forward_rt.ro_rt->rt_flags & 508 (RTF_HOST|RTF_GATEWAY)) == RTF_HOST && 509 #ifdef RTF_WASCLONED 510 !(ip6_forward_rt.ro_rt->rt_flags & RTF_WASCLONED) && 511 #endif 512 #ifdef RTF_CLONED 513 !(ip6_forward_rt.ro_rt->rt_flags & RTF_CLONED) && 514 #endif 515 #if 0 516 /* 517 * The check below is redundant since the comparison of 518 * the destination and the key of the rtentry has 519 * already done through looking up the routing table. 520 */ 521 IN6_ARE_ADDR_EQUAL(&ip6->ip6_dst, 522 &rt6_key(ip6_forward_rt.ro_rt)->sin6_addr) 523 #endif 524 ip6_forward_rt.ro_rt->rt_ifp->if_type == IFT_LOOP) { 525 struct in6_ifaddr *ia6 = 526 (struct in6_ifaddr *)ip6_forward_rt.ro_rt->rt_ifa; 527 528 /* 529 * record address information into m_tag. 530 */ 531 (void)ip6_setdstifaddr(m, ia6); 532 533 /* 534 * packets to a tentative, duplicated, or somehow invalid 535 * address must not be accepted. 536 */ 537 if (!(ia6->ia6_flags & IN6_IFF_NOTREADY)) { 538 /* this address is ready */ 539 ours = 1; 540 deliverifp = ia6->ia_ifp; /* correct? */ 541 /* Count the packet in the ip address stats */ 542 ia6->ia_ifa.if_ipackets++; 543 ia6->ia_ifa.if_ibytes += m->m_pkthdr.len; 544 goto hbhcheck; 545 } else { 546 char ip6bufs[INET6_ADDRSTRLEN]; 547 char ip6bufd[INET6_ADDRSTRLEN]; 548 /* address is not ready, so discard the packet. */ 549 nd6log((LOG_INFO, 550 "ip6_input: packet to an unready address %s->%s\n", 551 ip6_sprintf(ip6bufs, &ip6->ip6_src), 552 ip6_sprintf(ip6bufd, &ip6->ip6_dst))); 553 554 goto bad; 555 } 556 } 557 558 /* 559 * FAITH (Firewall Aided Internet Translator) 560 */ 561 if (ip6_keepfaith) { 562 if (ip6_forward_rt.ro_rt && ip6_forward_rt.ro_rt->rt_ifp 563 && ip6_forward_rt.ro_rt->rt_ifp->if_type == IFT_FAITH) { 564 /* XXX do we need more sanity checks? */ 565 ours = 1; 566 deliverifp = ip6_forward_rt.ro_rt->rt_ifp; /* faith */ 567 goto hbhcheck; 568 } 569 } 570 571 /* 572 * Now there is no reason to process the packet if it's not our own 573 * and we're not a router. 574 */ 575 if (!ip6_forwarding) { 576 ip6stat.ip6s_cantforward++; 577 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_discard); 578 goto bad; 579 } 580 581 hbhcheck: 582 /* 583 * record address information into m_tag, if we don't have one yet. 584 * note that we are unable to record it, if the address is not listed 585 * as our interface address (e.g. multicast addresses, addresses 586 * within FAITH prefixes and such). 587 */ 588 if (deliverifp && !ip6_getdstifaddr(m)) { 589 struct in6_ifaddr *ia6; 590 591 ia6 = in6_ifawithifp(deliverifp, &ip6->ip6_dst); 592 if (ia6) { 593 if (!ip6_setdstifaddr(m, ia6)) { 594 /* 595 * XXX maybe we should drop the packet here, 596 * as we could not provide enough information 597 * to the upper layers. 598 */ 599 } 600 } 601 } 602 603 /* 604 * Process Hop-by-Hop options header if it's contained. 605 * m may be modified in ip6_hopopts_input(). 606 * If a JumboPayload option is included, plen will also be modified. 607 */ 608 plen = (u_int32_t)ntohs(ip6->ip6_plen); 609 if (ip6->ip6_nxt == IPPROTO_HOPOPTS) { 610 struct ip6_hbh *hbh; 611 612 if (ip6_hopopts_input(&plen, &rtalert, &m, &off)) { 613 #if 0 /*touches NULL pointer*/ 614 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_discard); 615 #endif 616 return; /* m have already been freed */ 617 } 618 619 /* adjust pointer */ 620 ip6 = mtod(m, struct ip6_hdr *); 621 622 /* 623 * if the payload length field is 0 and the next header field 624 * indicates Hop-by-Hop Options header, then a Jumbo Payload 625 * option MUST be included. 626 */ 627 if (ip6->ip6_plen == 0 && plen == 0) { 628 /* 629 * Note that if a valid jumbo payload option is 630 * contained, ip6_hopopts_input() must set a valid 631 * (non-zero) payload length to the variable plen. 632 */ 633 ip6stat.ip6s_badoptions++; 634 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_discard); 635 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_hdrerr); 636 icmp6_error(m, ICMP6_PARAM_PROB, 637 ICMP6_PARAMPROB_HEADER, 638 (caddr_t)&ip6->ip6_plen - (caddr_t)ip6); 639 return; 640 } 641 #ifndef PULLDOWN_TEST 642 /* ip6_hopopts_input() ensures that mbuf is contiguous */ 643 hbh = (struct ip6_hbh *)(ip6 + 1); 644 #else 645 IP6_EXTHDR_GET(hbh, struct ip6_hbh *, m, sizeof(struct ip6_hdr), 646 sizeof(struct ip6_hbh)); 647 if (hbh == NULL) { 648 ip6stat.ip6s_tooshort++; 649 return; 650 } 651 #endif 652 nxt = hbh->ip6h_nxt; 653 654 /* 655 * If we are acting as a router and the packet contains a 656 * router alert option, see if we know the option value. 657 * Currently, we only support the option value for MLD, in which 658 * case we should pass the packet to the multicast routing 659 * daemon. 660 */ 661 if (rtalert != ~0 && ip6_forwarding) { 662 switch (rtalert) { 663 case IP6OPT_RTALERT_MLD: 664 ours = 1; 665 break; 666 default: 667 /* 668 * RFC2711 requires unrecognized values must be 669 * silently ignored. 670 */ 671 break; 672 } 673 } 674 } else 675 nxt = ip6->ip6_nxt; 676 677 /* 678 * Check that the amount of data in the buffers 679 * is as at least much as the IPv6 header would have us expect. 680 * Trim mbufs if longer than we expect. 681 * Drop packet if shorter than we expect. 682 */ 683 if (m->m_pkthdr.len - sizeof(struct ip6_hdr) < plen) { 684 ip6stat.ip6s_tooshort++; 685 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_truncated); 686 goto bad; 687 } 688 if (m->m_pkthdr.len > sizeof(struct ip6_hdr) + plen) { 689 if (m->m_len == m->m_pkthdr.len) { 690 m->m_len = sizeof(struct ip6_hdr) + plen; 691 m->m_pkthdr.len = sizeof(struct ip6_hdr) + plen; 692 } else 693 m_adj(m, sizeof(struct ip6_hdr) + plen - m->m_pkthdr.len); 694 } 695 696 /* 697 * Forward if desirable. 698 */ 699 if (IN6_IS_ADDR_MULTICAST(&ip6->ip6_dst)) { 700 /* 701 * If we are acting as a multicast router, all 702 * incoming multicast packets are passed to the 703 * kernel-level multicast forwarding function. 704 * The packet is returned (relatively) intact; if 705 * ip6_mforward() returns a non-zero value, the packet 706 * must be discarded, else it may be accepted below. 707 */ 708 if (ip6_mrouter && ip6_mforward && 709 ip6_mforward(ip6, m->m_pkthdr.rcvif, m)) { 710 ip6stat.ip6s_cantforward++; 711 m_freem(m); 712 return; 713 } 714 if (!ours) { 715 m_freem(m); 716 return; 717 } 718 } else if (!ours) { 719 ip6_forward(m, srcrt); 720 return; 721 } 722 723 ip6 = mtod(m, struct ip6_hdr *); 724 725 /* 726 * Malicious party may be able to use IPv4 mapped addr to confuse 727 * tcp/udp stack and bypass security checks (act as if it was from 728 * 127.0.0.1 by using IPv6 src ::ffff:127.0.0.1). Be cautious. 729 * 730 * For SIIT end node behavior, you may want to disable the check. 731 * However, you will become vulnerable to attacks using IPv4 mapped 732 * source. 733 */ 734 if (IN6_IS_ADDR_V4MAPPED(&ip6->ip6_src) || 735 IN6_IS_ADDR_V4MAPPED(&ip6->ip6_dst)) { 736 ip6stat.ip6s_badscope++; 737 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_addrerr); 738 goto bad; 739 } 740 741 /* 742 * Tell launch routine the next header 743 */ 744 ip6stat.ip6s_delivered++; 745 in6_ifstat_inc(deliverifp, ifs6_in_deliver); 746 nest = 0; 747 748 while (nxt != IPPROTO_DONE) { 749 if (ip6_hdrnestlimit && (++nest > ip6_hdrnestlimit)) { 750 ip6stat.ip6s_toomanyhdr++; 751 goto bad; 752 } 753 754 /* 755 * protection against faulty packet - there should be 756 * more sanity checks in header chain processing. 757 */ 758 if (m->m_pkthdr.len < off) { 759 ip6stat.ip6s_tooshort++; 760 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_truncated); 761 goto bad; 762 } 763 764 #ifdef IPSEC 765 /* 766 * enforce IPsec policy checking if we are seeing last header. 767 * note that we do not visit this with protocols with pcb layer 768 * code - like udp/tcp/raw ip. 769 */ 770 if (ip6_ipsec_input(m, nxt)) 771 goto bad; 772 #endif /* IPSEC */ 773 nxt = (*inet6sw[ip6_protox[nxt]].pr_input)(&m, &off, nxt); 774 } 775 return; 776 bad: 777 m_freem(m); 778 } 779 780 /* 781 * set/grab in6_ifaddr correspond to IPv6 destination address. 782 * XXX backward compatibility wrapper 783 */ 784 static struct ip6aux * 785 ip6_setdstifaddr(struct mbuf *m, struct in6_ifaddr *ia6) 786 { 787 struct ip6aux *ip6a; 788 789 ip6a = ip6_addaux(m); 790 if (ip6a) 791 ip6a->ip6a_dstia6 = ia6; 792 return ip6a; /* NULL if failed to set */ 793 } 794 795 struct in6_ifaddr * 796 ip6_getdstifaddr(struct mbuf *m) 797 { 798 struct ip6aux *ip6a; 799 800 ip6a = ip6_findaux(m); 801 if (ip6a) 802 return ip6a->ip6a_dstia6; 803 else 804 return NULL; 805 } 806 807 /* 808 * Hop-by-Hop options header processing. If a valid jumbo payload option is 809 * included, the real payload length will be stored in plenp. 810 * 811 * rtalertp - XXX: should be stored more smart way 812 */ 813 static int 814 ip6_hopopts_input(u_int32_t *plenp, u_int32_t *rtalertp, 815 struct mbuf **mp, int *offp) 816 { 817 struct mbuf *m = *mp; 818 int off = *offp, hbhlen; 819 struct ip6_hbh *hbh; 820 u_int8_t *opt; 821 822 /* validation of the length of the header */ 823 #ifndef PULLDOWN_TEST 824 IP6_EXTHDR_CHECK(m, off, sizeof(*hbh), -1); 825 hbh = (struct ip6_hbh *)(mtod(m, caddr_t) + off); 826 hbhlen = (hbh->ip6h_len + 1) << 3; 827 828 IP6_EXTHDR_CHECK(m, off, hbhlen, -1); 829 hbh = (struct ip6_hbh *)(mtod(m, caddr_t) + off); 830 #else 831 IP6_EXTHDR_GET(hbh, struct ip6_hbh *, m, 832 sizeof(struct ip6_hdr), sizeof(struct ip6_hbh)); 833 if (hbh == NULL) { 834 ip6stat.ip6s_tooshort++; 835 return -1; 836 } 837 hbhlen = (hbh->ip6h_len + 1) << 3; 838 IP6_EXTHDR_GET(hbh, struct ip6_hbh *, m, sizeof(struct ip6_hdr), 839 hbhlen); 840 if (hbh == NULL) { 841 ip6stat.ip6s_tooshort++; 842 return -1; 843 } 844 #endif 845 off += hbhlen; 846 hbhlen -= sizeof(struct ip6_hbh); 847 opt = (u_int8_t *)hbh + sizeof(struct ip6_hbh); 848 849 if (ip6_process_hopopts(m, (u_int8_t *)hbh + sizeof(struct ip6_hbh), 850 hbhlen, rtalertp, plenp) < 0) 851 return (-1); 852 853 *offp = off; 854 *mp = m; 855 return (0); 856 } 857 858 /* 859 * Search header for all Hop-by-hop options and process each option. 860 * This function is separate from ip6_hopopts_input() in order to 861 * handle a case where the sending node itself process its hop-by-hop 862 * options header. In such a case, the function is called from ip6_output(). 863 * 864 * The function assumes that hbh header is located right after the IPv6 header 865 * (RFC2460 p7), opthead is pointer into data content in m, and opthead to 866 * opthead + hbhlen is located in continuous memory region. 867 */ 868 int 869 ip6_process_hopopts(struct mbuf *m, u_int8_t *opthead, int hbhlen, 870 u_int32_t *rtalertp, u_int32_t *plenp) 871 { 872 struct ip6_hdr *ip6; 873 int optlen = 0; 874 u_int8_t *opt = opthead; 875 u_int16_t rtalert_val; 876 u_int32_t jumboplen; 877 const int erroff = sizeof(struct ip6_hdr) + sizeof(struct ip6_hbh); 878 879 for (; hbhlen > 0; hbhlen -= optlen, opt += optlen) { 880 switch (*opt) { 881 case IP6OPT_PAD1: 882 optlen = 1; 883 break; 884 case IP6OPT_PADN: 885 if (hbhlen < IP6OPT_MINLEN) { 886 ip6stat.ip6s_toosmall++; 887 goto bad; 888 } 889 optlen = *(opt + 1) + 2; 890 break; 891 case IP6OPT_ROUTER_ALERT: 892 /* XXX may need check for alignment */ 893 if (hbhlen < IP6OPT_RTALERT_LEN) { 894 ip6stat.ip6s_toosmall++; 895 goto bad; 896 } 897 if (*(opt + 1) != IP6OPT_RTALERT_LEN - 2) { 898 /* XXX stat */ 899 icmp6_error(m, ICMP6_PARAM_PROB, 900 ICMP6_PARAMPROB_HEADER, 901 erroff + opt + 1 - opthead); 902 return (-1); 903 } 904 optlen = IP6OPT_RTALERT_LEN; 905 bcopy((caddr_t)(opt + 2), (caddr_t)&rtalert_val, 2); 906 *rtalertp = ntohs(rtalert_val); 907 break; 908 case IP6OPT_JUMBO: 909 /* XXX may need check for alignment */ 910 if (hbhlen < IP6OPT_JUMBO_LEN) { 911 ip6stat.ip6s_toosmall++; 912 goto bad; 913 } 914 if (*(opt + 1) != IP6OPT_JUMBO_LEN - 2) { 915 /* XXX stat */ 916 icmp6_error(m, ICMP6_PARAM_PROB, 917 ICMP6_PARAMPROB_HEADER, 918 erroff + opt + 1 - opthead); 919 return (-1); 920 } 921 optlen = IP6OPT_JUMBO_LEN; 922 923 /* 924 * IPv6 packets that have non 0 payload length 925 * must not contain a jumbo payload option. 926 */ 927 ip6 = mtod(m, struct ip6_hdr *); 928 if (ip6->ip6_plen) { 929 ip6stat.ip6s_badoptions++; 930 icmp6_error(m, ICMP6_PARAM_PROB, 931 ICMP6_PARAMPROB_HEADER, 932 erroff + opt - opthead); 933 return (-1); 934 } 935 936 /* 937 * We may see jumbolen in unaligned location, so 938 * we'd need to perform bcopy(). 939 */ 940 bcopy(opt + 2, &jumboplen, sizeof(jumboplen)); 941 jumboplen = (u_int32_t)htonl(jumboplen); 942 943 #if 1 944 /* 945 * if there are multiple jumbo payload options, 946 * *plenp will be non-zero and the packet will be 947 * rejected. 948 * the behavior may need some debate in ipngwg - 949 * multiple options does not make sense, however, 950 * there's no explicit mention in specification. 951 */ 952 if (*plenp != 0) { 953 ip6stat.ip6s_badoptions++; 954 icmp6_error(m, ICMP6_PARAM_PROB, 955 ICMP6_PARAMPROB_HEADER, 956 erroff + opt + 2 - opthead); 957 return (-1); 958 } 959 #endif 960 961 /* 962 * jumbo payload length must be larger than 65535. 963 */ 964 if (jumboplen <= IPV6_MAXPACKET) { 965 ip6stat.ip6s_badoptions++; 966 icmp6_error(m, ICMP6_PARAM_PROB, 967 ICMP6_PARAMPROB_HEADER, 968 erroff + opt + 2 - opthead); 969 return (-1); 970 } 971 *plenp = jumboplen; 972 973 break; 974 default: /* unknown option */ 975 if (hbhlen < IP6OPT_MINLEN) { 976 ip6stat.ip6s_toosmall++; 977 goto bad; 978 } 979 optlen = ip6_unknown_opt(opt, m, 980 erroff + opt - opthead); 981 if (optlen == -1) 982 return (-1); 983 optlen += 2; 984 break; 985 } 986 } 987 988 return (0); 989 990 bad: 991 m_freem(m); 992 return (-1); 993 } 994 995 /* 996 * Unknown option processing. 997 * The third argument `off' is the offset from the IPv6 header to the option, 998 * which is necessary if the IPv6 header the and option header and IPv6 header 999 * is not continuous in order to return an ICMPv6 error. 1000 */ 1001 int 1002 ip6_unknown_opt(u_int8_t *optp, struct mbuf *m, int off) 1003 { 1004 struct ip6_hdr *ip6; 1005 1006 switch (IP6OPT_TYPE(*optp)) { 1007 case IP6OPT_TYPE_SKIP: /* ignore the option */ 1008 return ((int)*(optp + 1)); 1009 case IP6OPT_TYPE_DISCARD: /* silently discard */ 1010 m_freem(m); 1011 return (-1); 1012 case IP6OPT_TYPE_FORCEICMP: /* send ICMP even if multicasted */ 1013 ip6stat.ip6s_badoptions++; 1014 icmp6_error(m, ICMP6_PARAM_PROB, ICMP6_PARAMPROB_OPTION, off); 1015 return (-1); 1016 case IP6OPT_TYPE_ICMP: /* send ICMP if not multicasted */ 1017 ip6stat.ip6s_badoptions++; 1018 ip6 = mtod(m, struct ip6_hdr *); 1019 if (IN6_IS_ADDR_MULTICAST(&ip6->ip6_dst) || 1020 (m->m_flags & (M_BCAST|M_MCAST))) 1021 m_freem(m); 1022 else 1023 icmp6_error(m, ICMP6_PARAM_PROB, 1024 ICMP6_PARAMPROB_OPTION, off); 1025 return (-1); 1026 } 1027 1028 m_freem(m); /* XXX: NOTREACHED */ 1029 return (-1); 1030 } 1031 1032 /* 1033 * Create the "control" list for this pcb. 1034 * These functions will not modify mbuf chain at all. 1035 * 1036 * With KAME mbuf chain restriction: 1037 * The routine will be called from upper layer handlers like tcp6_input(). 1038 * Thus the routine assumes that the caller (tcp6_input) have already 1039 * called IP6_EXTHDR_CHECK() and all the extension headers are located in the 1040 * very first mbuf on the mbuf chain. 1041 * 1042 * ip6_savecontrol_v4 will handle those options that are possible to be 1043 * set on a v4-mapped socket. 1044 * ip6_savecontrol will directly call ip6_savecontrol_v4 to handle those 1045 * options and handle the v6-only ones itself. 1046 */ 1047 int 1048 ip6_savecontrol_v4(struct inpcb *inp, struct mbuf *m, struct mbuf **mp) 1049 { 1050 struct ip6_hdr *ip6 = mtod(m, struct ip6_hdr *); 1051 1052 #ifdef SO_TIMESTAMP 1053 if ((inp->inp_socket->so_options & SO_TIMESTAMP) != 0) { 1054 struct timeval tv; 1055 1056 microtime(&tv); 1057 *mp = sbcreatecontrol((caddr_t) &tv, sizeof(tv), 1058 SCM_TIMESTAMP, SOL_SOCKET); 1059 if (*mp) 1060 mp = &(*mp)->m_next; 1061 } 1062 #endif 1063 1064 if ((ip6->ip6_vfc & IPV6_VERSION_MASK) != IPV6_VERSION) 1065 return (1); 1066 1067 #define IS2292(inp, x, y) (((inp)->inp_flags & IN6P_RFC2292) ? (x) : (y)) 1068 /* RFC 2292 sec. 5 */ 1069 if ((inp->inp_flags & IN6P_PKTINFO) != 0) { 1070 struct in6_pktinfo pi6; 1071 1072 bcopy(&ip6->ip6_dst, &pi6.ipi6_addr, sizeof(struct in6_addr)); 1073 in6_clearscope(&pi6.ipi6_addr); /* XXX */ 1074 pi6.ipi6_ifindex = 1075 (m && m->m_pkthdr.rcvif) ? m->m_pkthdr.rcvif->if_index : 0; 1076 1077 *mp = sbcreatecontrol((caddr_t) &pi6, 1078 sizeof(struct in6_pktinfo), 1079 IS2292(inp, IPV6_2292PKTINFO, IPV6_PKTINFO), IPPROTO_IPV6); 1080 if (*mp) 1081 mp = &(*mp)->m_next; 1082 } 1083 1084 if ((inp->inp_flags & IN6P_HOPLIMIT) != 0) { 1085 int hlim = ip6->ip6_hlim & 0xff; 1086 1087 *mp = sbcreatecontrol((caddr_t) &hlim, sizeof(int), 1088 IS2292(inp, IPV6_2292HOPLIMIT, IPV6_HOPLIMIT), 1089 IPPROTO_IPV6); 1090 if (*mp) 1091 mp = &(*mp)->m_next; 1092 } 1093 1094 return (0); 1095 } 1096 1097 void 1098 ip6_savecontrol(struct inpcb *in6p, struct mbuf *m, struct mbuf **mp) 1099 { 1100 struct ip6_hdr *ip6 = mtod(m, struct ip6_hdr *); 1101 1102 if (ip6_savecontrol_v4(in6p, m, mp) != 0) 1103 return; 1104 1105 if ((in6p->in6p_flags & IN6P_TCLASS) != 0) { 1106 u_int32_t flowinfo; 1107 int tclass; 1108 1109 flowinfo = (u_int32_t)ntohl(ip6->ip6_flow & IPV6_FLOWINFO_MASK); 1110 flowinfo >>= 20; 1111 1112 tclass = flowinfo & 0xff; 1113 *mp = sbcreatecontrol((caddr_t) &tclass, sizeof(tclass), 1114 IPV6_TCLASS, IPPROTO_IPV6); 1115 if (*mp) 1116 mp = &(*mp)->m_next; 1117 } 1118 1119 /* 1120 * IPV6_HOPOPTS socket option. Recall that we required super-user 1121 * privilege for the option (see ip6_ctloutput), but it might be too 1122 * strict, since there might be some hop-by-hop options which can be 1123 * returned to normal user. 1124 * See also RFC 2292 section 6 (or RFC 3542 section 8). 1125 */ 1126 if ((in6p->in6p_flags & IN6P_HOPOPTS) != 0) { 1127 /* 1128 * Check if a hop-by-hop options header is contatined in the 1129 * received packet, and if so, store the options as ancillary 1130 * data. Note that a hop-by-hop options header must be 1131 * just after the IPv6 header, which is assured through the 1132 * IPv6 input processing. 1133 */ 1134 if (ip6->ip6_nxt == IPPROTO_HOPOPTS) { 1135 struct ip6_hbh *hbh; 1136 int hbhlen = 0; 1137 #ifdef PULLDOWN_TEST 1138 struct mbuf *ext; 1139 #endif 1140 1141 #ifndef PULLDOWN_TEST 1142 hbh = (struct ip6_hbh *)(ip6 + 1); 1143 hbhlen = (hbh->ip6h_len + 1) << 3; 1144 #else 1145 ext = ip6_pullexthdr(m, sizeof(struct ip6_hdr), 1146 ip6->ip6_nxt); 1147 if (ext == NULL) { 1148 ip6stat.ip6s_tooshort++; 1149 return; 1150 } 1151 hbh = mtod(ext, struct ip6_hbh *); 1152 hbhlen = (hbh->ip6h_len + 1) << 3; 1153 if (hbhlen != ext->m_len) { 1154 m_freem(ext); 1155 ip6stat.ip6s_tooshort++; 1156 return; 1157 } 1158 #endif 1159 1160 /* 1161 * XXX: We copy the whole header even if a 1162 * jumbo payload option is included, the option which 1163 * is to be removed before returning according to 1164 * RFC2292. 1165 * Note: this constraint is removed in RFC3542 1166 */ 1167 *mp = sbcreatecontrol((caddr_t)hbh, hbhlen, 1168 IS2292(in6p, IPV6_2292HOPOPTS, IPV6_HOPOPTS), 1169 IPPROTO_IPV6); 1170 if (*mp) 1171 mp = &(*mp)->m_next; 1172 #ifdef PULLDOWN_TEST 1173 m_freem(ext); 1174 #endif 1175 } 1176 } 1177 1178 if ((in6p->in6p_flags & (IN6P_RTHDR | IN6P_DSTOPTS)) != 0) { 1179 int nxt = ip6->ip6_nxt, off = sizeof(struct ip6_hdr); 1180 1181 /* 1182 * Search for destination options headers or routing 1183 * header(s) through the header chain, and stores each 1184 * header as ancillary data. 1185 * Note that the order of the headers remains in 1186 * the chain of ancillary data. 1187 */ 1188 while (1) { /* is explicit loop prevention necessary? */ 1189 struct ip6_ext *ip6e = NULL; 1190 int elen; 1191 #ifdef PULLDOWN_TEST 1192 struct mbuf *ext = NULL; 1193 #endif 1194 1195 /* 1196 * if it is not an extension header, don't try to 1197 * pull it from the chain. 1198 */ 1199 switch (nxt) { 1200 case IPPROTO_DSTOPTS: 1201 case IPPROTO_ROUTING: 1202 case IPPROTO_HOPOPTS: 1203 case IPPROTO_AH: /* is it possible? */ 1204 break; 1205 default: 1206 goto loopend; 1207 } 1208 1209 #ifndef PULLDOWN_TEST 1210 if (off + sizeof(*ip6e) > m->m_len) 1211 goto loopend; 1212 ip6e = (struct ip6_ext *)(mtod(m, caddr_t) + off); 1213 if (nxt == IPPROTO_AH) 1214 elen = (ip6e->ip6e_len + 2) << 2; 1215 else 1216 elen = (ip6e->ip6e_len + 1) << 3; 1217 if (off + elen > m->m_len) 1218 goto loopend; 1219 #else 1220 ext = ip6_pullexthdr(m, off, nxt); 1221 if (ext == NULL) { 1222 ip6stat.ip6s_tooshort++; 1223 return; 1224 } 1225 ip6e = mtod(ext, struct ip6_ext *); 1226 if (nxt == IPPROTO_AH) 1227 elen = (ip6e->ip6e_len + 2) << 2; 1228 else 1229 elen = (ip6e->ip6e_len + 1) << 3; 1230 if (elen != ext->m_len) { 1231 m_freem(ext); 1232 ip6stat.ip6s_tooshort++; 1233 return; 1234 } 1235 #endif 1236 1237 switch (nxt) { 1238 case IPPROTO_DSTOPTS: 1239 if (!(in6p->in6p_flags & IN6P_DSTOPTS)) 1240 break; 1241 1242 *mp = sbcreatecontrol((caddr_t)ip6e, elen, 1243 IS2292(in6p, 1244 IPV6_2292DSTOPTS, IPV6_DSTOPTS), 1245 IPPROTO_IPV6); 1246 if (*mp) 1247 mp = &(*mp)->m_next; 1248 break; 1249 case IPPROTO_ROUTING: 1250 if (!in6p->in6p_flags & IN6P_RTHDR) 1251 break; 1252 1253 *mp = sbcreatecontrol((caddr_t)ip6e, elen, 1254 IS2292(in6p, IPV6_2292RTHDR, IPV6_RTHDR), 1255 IPPROTO_IPV6); 1256 if (*mp) 1257 mp = &(*mp)->m_next; 1258 break; 1259 case IPPROTO_HOPOPTS: 1260 case IPPROTO_AH: /* is it possible? */ 1261 break; 1262 1263 default: 1264 /* 1265 * other cases have been filtered in the above. 1266 * none will visit this case. here we supply 1267 * the code just in case (nxt overwritten or 1268 * other cases). 1269 */ 1270 #ifdef PULLDOWN_TEST 1271 m_freem(ext); 1272 #endif 1273 goto loopend; 1274 1275 } 1276 1277 /* proceed with the next header. */ 1278 off += elen; 1279 nxt = ip6e->ip6e_nxt; 1280 ip6e = NULL; 1281 #ifdef PULLDOWN_TEST 1282 m_freem(ext); 1283 ext = NULL; 1284 #endif 1285 } 1286 loopend: 1287 ; 1288 } 1289 } 1290 #undef IS2292 1291 1292 void 1293 ip6_notify_pmtu(struct inpcb *in6p, struct sockaddr_in6 *dst, u_int32_t *mtu) 1294 { 1295 struct socket *so; 1296 struct mbuf *m_mtu; 1297 struct ip6_mtuinfo mtuctl; 1298 1299 so = in6p->inp_socket; 1300 1301 if (mtu == NULL) 1302 return; 1303 1304 #ifdef DIAGNOSTIC 1305 if (so == NULL) /* I believe this is impossible */ 1306 panic("ip6_notify_pmtu: socket is NULL"); 1307 #endif 1308 1309 bzero(&mtuctl, sizeof(mtuctl)); /* zero-clear for safety */ 1310 mtuctl.ip6m_mtu = *mtu; 1311 mtuctl.ip6m_addr = *dst; 1312 if (sa6_recoverscope(&mtuctl.ip6m_addr)) 1313 return; 1314 1315 if ((m_mtu = sbcreatecontrol((caddr_t)&mtuctl, sizeof(mtuctl), 1316 IPV6_PATHMTU, IPPROTO_IPV6)) == NULL) 1317 return; 1318 1319 if (sbappendaddr(&so->so_rcv, (struct sockaddr *)dst, NULL, m_mtu) 1320 == 0) { 1321 m_freem(m_mtu); 1322 /* XXX: should count statistics */ 1323 } else 1324 sorwakeup(so); 1325 1326 return; 1327 } 1328 1329 #ifdef PULLDOWN_TEST 1330 /* 1331 * pull single extension header from mbuf chain. returns single mbuf that 1332 * contains the result, or NULL on error. 1333 */ 1334 static struct mbuf * 1335 ip6_pullexthdr(struct mbuf *m, size_t off, int nxt) 1336 { 1337 struct ip6_ext ip6e; 1338 size_t elen; 1339 struct mbuf *n; 1340 1341 #ifdef DIAGNOSTIC 1342 switch (nxt) { 1343 case IPPROTO_DSTOPTS: 1344 case IPPROTO_ROUTING: 1345 case IPPROTO_HOPOPTS: 1346 case IPPROTO_AH: /* is it possible? */ 1347 break; 1348 default: 1349 printf("ip6_pullexthdr: invalid nxt=%d\n", nxt); 1350 } 1351 #endif 1352 1353 m_copydata(m, off, sizeof(ip6e), (caddr_t)&ip6e); 1354 if (nxt == IPPROTO_AH) 1355 elen = (ip6e.ip6e_len + 2) << 2; 1356 else 1357 elen = (ip6e.ip6e_len + 1) << 3; 1358 1359 MGET(n, M_DONTWAIT, MT_DATA); 1360 if (n && elen >= MLEN) { 1361 MCLGET(n, M_DONTWAIT); 1362 if ((n->m_flags & M_EXT) == 0) { 1363 m_free(n); 1364 n = NULL; 1365 } 1366 } 1367 if (!n) 1368 return NULL; 1369 1370 n->m_len = 0; 1371 if (elen >= M_TRAILINGSPACE(n)) { 1372 m_free(n); 1373 return NULL; 1374 } 1375 1376 m_copydata(m, off, elen, mtod(n, caddr_t)); 1377 n->m_len = elen; 1378 return n; 1379 } 1380 #endif 1381 1382 /* 1383 * Get pointer to the previous header followed by the header 1384 * currently processed. 1385 * XXX: This function supposes that 1386 * M includes all headers, 1387 * the next header field and the header length field of each header 1388 * are valid, and 1389 * the sum of each header length equals to OFF. 1390 * Because of these assumptions, this function must be called very 1391 * carefully. Moreover, it will not be used in the near future when 1392 * we develop `neater' mechanism to process extension headers. 1393 */ 1394 char * 1395 ip6_get_prevhdr(struct mbuf *m, int off) 1396 { 1397 struct ip6_hdr *ip6 = mtod(m, struct ip6_hdr *); 1398 1399 if (off == sizeof(struct ip6_hdr)) 1400 return (&ip6->ip6_nxt); 1401 else { 1402 int len, nxt; 1403 struct ip6_ext *ip6e = NULL; 1404 1405 nxt = ip6->ip6_nxt; 1406 len = sizeof(struct ip6_hdr); 1407 while (len < off) { 1408 ip6e = (struct ip6_ext *)(mtod(m, caddr_t) + len); 1409 1410 switch (nxt) { 1411 case IPPROTO_FRAGMENT: 1412 len += sizeof(struct ip6_frag); 1413 break; 1414 case IPPROTO_AH: 1415 len += (ip6e->ip6e_len + 2) << 2; 1416 break; 1417 default: 1418 len += (ip6e->ip6e_len + 1) << 3; 1419 break; 1420 } 1421 nxt = ip6e->ip6e_nxt; 1422 } 1423 if (ip6e) 1424 return (&ip6e->ip6e_nxt); 1425 else 1426 return NULL; 1427 } 1428 } 1429 1430 /* 1431 * get next header offset. m will be retained. 1432 */ 1433 int 1434 ip6_nexthdr(struct mbuf *m, int off, int proto, int *nxtp) 1435 { 1436 struct ip6_hdr ip6; 1437 struct ip6_ext ip6e; 1438 struct ip6_frag fh; 1439 1440 /* just in case */ 1441 if (m == NULL) 1442 panic("ip6_nexthdr: m == NULL"); 1443 if ((m->m_flags & M_PKTHDR) == 0 || m->m_pkthdr.len < off) 1444 return -1; 1445 1446 switch (proto) { 1447 case IPPROTO_IPV6: 1448 if (m->m_pkthdr.len < off + sizeof(ip6)) 1449 return -1; 1450 m_copydata(m, off, sizeof(ip6), (caddr_t)&ip6); 1451 if (nxtp) 1452 *nxtp = ip6.ip6_nxt; 1453 off += sizeof(ip6); 1454 return off; 1455 1456 case IPPROTO_FRAGMENT: 1457 /* 1458 * terminate parsing if it is not the first fragment, 1459 * it does not make sense to parse through it. 1460 */ 1461 if (m->m_pkthdr.len < off + sizeof(fh)) 1462 return -1; 1463 m_copydata(m, off, sizeof(fh), (caddr_t)&fh); 1464 /* IP6F_OFF_MASK = 0xfff8(BigEndian), 0xf8ff(LittleEndian) */ 1465 if (fh.ip6f_offlg & IP6F_OFF_MASK) 1466 return -1; 1467 if (nxtp) 1468 *nxtp = fh.ip6f_nxt; 1469 off += sizeof(struct ip6_frag); 1470 return off; 1471 1472 case IPPROTO_AH: 1473 if (m->m_pkthdr.len < off + sizeof(ip6e)) 1474 return -1; 1475 m_copydata(m, off, sizeof(ip6e), (caddr_t)&ip6e); 1476 if (nxtp) 1477 *nxtp = ip6e.ip6e_nxt; 1478 off += (ip6e.ip6e_len + 2) << 2; 1479 return off; 1480 1481 case IPPROTO_HOPOPTS: 1482 case IPPROTO_ROUTING: 1483 case IPPROTO_DSTOPTS: 1484 if (m->m_pkthdr.len < off + sizeof(ip6e)) 1485 return -1; 1486 m_copydata(m, off, sizeof(ip6e), (caddr_t)&ip6e); 1487 if (nxtp) 1488 *nxtp = ip6e.ip6e_nxt; 1489 off += (ip6e.ip6e_len + 1) << 3; 1490 return off; 1491 1492 case IPPROTO_NONE: 1493 case IPPROTO_ESP: 1494 case IPPROTO_IPCOMP: 1495 /* give up */ 1496 return -1; 1497 1498 default: 1499 return -1; 1500 } 1501 1502 return -1; 1503 } 1504 1505 /* 1506 * get offset for the last header in the chain. m will be kept untainted. 1507 */ 1508 int 1509 ip6_lasthdr(struct mbuf *m, int off, int proto, int *nxtp) 1510 { 1511 int newoff; 1512 int nxt; 1513 1514 if (!nxtp) { 1515 nxt = -1; 1516 nxtp = &nxt; 1517 } 1518 while (1) { 1519 newoff = ip6_nexthdr(m, off, proto, nxtp); 1520 if (newoff < 0) 1521 return off; 1522 else if (newoff < off) 1523 return -1; /* invalid */ 1524 else if (newoff == off) 1525 return newoff; 1526 1527 off = newoff; 1528 proto = *nxtp; 1529 } 1530 } 1531 1532 struct ip6aux * 1533 ip6_addaux(struct mbuf *m) 1534 { 1535 struct m_tag *mtag; 1536 1537 mtag = m_tag_find(m, PACKET_TAG_IPV6_INPUT, NULL); 1538 if (!mtag) { 1539 mtag = m_tag_get(PACKET_TAG_IPV6_INPUT, sizeof(struct ip6aux), 1540 M_NOWAIT); 1541 if (mtag) { 1542 m_tag_prepend(m, mtag); 1543 bzero(mtag + 1, sizeof(struct ip6aux)); 1544 } 1545 } 1546 return mtag ? (struct ip6aux *)(mtag + 1) : NULL; 1547 } 1548 1549 struct ip6aux * 1550 ip6_findaux(struct mbuf *m) 1551 { 1552 struct m_tag *mtag; 1553 1554 mtag = m_tag_find(m, PACKET_TAG_IPV6_INPUT, NULL); 1555 return mtag ? (struct ip6aux *)(mtag + 1) : NULL; 1556 } 1557 1558 void 1559 ip6_delaux(struct mbuf *m) 1560 { 1561 struct m_tag *mtag; 1562 1563 mtag = m_tag_find(m, PACKET_TAG_IPV6_INPUT, NULL); 1564 if (mtag) 1565 m_tag_delete(m, mtag); 1566 } 1567 1568 /* 1569 * System control for IP6 1570 */ 1571 1572 u_char inet6ctlerrmap[PRC_NCMDS] = { 1573 0, 0, 0, 0, 1574 0, EMSGSIZE, EHOSTDOWN, EHOSTUNREACH, 1575 EHOSTUNREACH, EHOSTUNREACH, ECONNREFUSED, ECONNREFUSED, 1576 EMSGSIZE, EHOSTUNREACH, 0, 0, 1577 0, 0, 0, 0, 1578 ENOPROTOOPT 1579 }; 1580