1 /* $FreeBSD: src/sys/netinet6/ip6_input.c,v 1.11.2.15 2003/01/24 05:11:35 sam Exp $ */ 2 /* $KAME: ip6_input.c,v 1.259 2002/01/21 04:58:09 jinmei Exp $ */ 3 4 /* 5 * Copyright (C) 1995, 1996, 1997, and 1998 WIDE Project. 6 * All rights reserved. 7 * 8 * Redistribution and use in source and binary forms, with or without 9 * modification, are permitted provided that the following conditions 10 * are met: 11 * 1. Redistributions of source code must retain the above copyright 12 * notice, this list of conditions and the following disclaimer. 13 * 2. Redistributions in binary form must reproduce the above copyright 14 * notice, this list of conditions and the following disclaimer in the 15 * documentation and/or other materials provided with the distribution. 16 * 3. Neither the name of the project nor the names of its contributors 17 * may be used to endorse or promote products derived from this software 18 * without specific prior written permission. 19 * 20 * THIS SOFTWARE IS PROVIDED BY THE PROJECT AND CONTRIBUTORS ``AS IS'' AND 21 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE 22 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE 23 * ARE DISCLAIMED. IN NO EVENT SHALL THE PROJECT OR CONTRIBUTORS BE LIABLE 24 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL 25 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS 26 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) 27 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT 28 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY 29 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF 30 * SUCH DAMAGE. 31 */ 32 33 /* 34 * Copyright (c) 1982, 1986, 1988, 1993 35 * The Regents of the University of California. All rights reserved. 36 * 37 * Redistribution and use in source and binary forms, with or without 38 * modification, are permitted provided that the following conditions 39 * are met: 40 * 1. Redistributions of source code must retain the above copyright 41 * notice, this list of conditions and the following disclaimer. 42 * 2. Redistributions in binary form must reproduce the above copyright 43 * notice, this list of conditions and the following disclaimer in the 44 * documentation and/or other materials provided with the distribution. 45 * 3. Neither the name of the University nor the names of its contributors 46 * may be used to endorse or promote products derived from this software 47 * without specific prior written permission. 48 * 49 * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND 50 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE 51 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE 52 * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE 53 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL 54 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS 55 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) 56 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT 57 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY 58 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF 59 * SUCH DAMAGE. 60 * 61 * @(#)ip_input.c 8.2 (Berkeley) 1/4/94 62 */ 63 64 #include "opt_ip6fw.h" 65 #include "opt_inet.h" 66 #include "opt_inet6.h" 67 #include "opt_ipsec.h" 68 69 #include <sys/param.h> 70 #include <sys/systm.h> 71 #include <sys/malloc.h> 72 #include <sys/mbuf.h> 73 #include <sys/domain.h> 74 #include <sys/protosw.h> 75 #include <sys/socket.h> 76 #include <sys/socketvar.h> 77 #include <sys/errno.h> 78 #include <sys/time.h> 79 #include <sys/kernel.h> 80 #include <sys/syslog.h> 81 #include <sys/proc.h> 82 #include <sys/priv.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 <sys/thread2.h> 92 #include <sys/msgport2.h> 93 #include <net/netmsg2.h> 94 #include <net/netisr2.h> 95 96 #include <netinet/in.h> 97 #include <netinet/in_systm.h> 98 #ifdef INET 99 #include <netinet/ip.h> 100 #include <netinet/ip_icmp.h> 101 #endif /* INET */ 102 #include <netinet/ip6.h> 103 #include <netinet6/in6_var.h> 104 #include <netinet6/ip6_var.h> 105 #include <netinet/in_pcb.h> 106 #include <netinet/icmp6.h> 107 #include <netinet6/scope6_var.h> 108 #include <netinet6/in6_ifattach.h> 109 #include <netinet6/nd6.h> 110 111 #ifdef IPSEC 112 #include <netinet6/ipsec.h> 113 #ifdef INET6 114 #include <netinet6/ipsec6.h> 115 #endif 116 #endif 117 118 #ifdef FAST_IPSEC 119 #include <netproto/ipsec/ipsec.h> 120 #include <netproto/ipsec/ipsec6.h> 121 #define IPSEC 122 #endif /* FAST_IPSEC */ 123 124 #include <net/ip6fw/ip6_fw.h> 125 126 #include <netinet6/ip6protosw.h> 127 128 #include <net/net_osdep.h> 129 130 extern struct domain inet6domain; 131 extern struct protosw inet6sw[]; 132 133 u_char ip6_protox[IPPROTO_MAX]; 134 struct in6_ifaddr *in6_ifaddr; 135 136 int ip6_forward_srcrt; /* XXX */ 137 int ip6_sourcecheck; /* XXX */ 138 int ip6_sourcecheck_interval; /* XXX */ 139 140 int ip6_ours_check_algorithm; 141 142 struct pfil_head inet6_pfil_hook; 143 144 /* firewall hooks */ 145 ip6_fw_chk_t *ip6_fw_chk_ptr; 146 ip6_fw_ctl_t *ip6_fw_ctl_ptr; 147 int ip6_fw_enable = 1; 148 149 struct ip6stat ip6stat; 150 151 static void ip6_init2 (void *); 152 static struct ip6aux *ip6_setdstifaddr (struct mbuf *, struct in6_ifaddr *); 153 static int ip6_hopopts_input (u_int32_t *, u_int32_t *, struct mbuf **, int *); 154 static void ip6_input(netmsg_t msg); 155 #ifdef PULLDOWN_TEST 156 static struct mbuf *ip6_pullexthdr (struct mbuf *, size_t, int); 157 #endif 158 static void transport6_processing_handler(netmsg_t netmsg); 159 160 /* 161 * IP6 initialization: fill in IP6 protocol switch table. 162 * All protocols not implemented in kernel go to raw IP6 protocol handler. 163 */ 164 void 165 ip6_init(void) 166 { 167 struct protosw *pr; 168 int i; 169 struct timeval tv; 170 171 pr = pffindproto(PF_INET6, IPPROTO_RAW, SOCK_RAW); 172 if (pr == NULL) 173 panic("ip6_init"); 174 for (i = 0; i < IPPROTO_MAX; i++) 175 ip6_protox[i] = pr - inet6sw; 176 for (pr = inet6domain.dom_protosw; 177 pr < inet6domain.dom_protoswNPROTOSW; pr++) 178 if (pr->pr_domain->dom_family == PF_INET6 && 179 pr->pr_protocol && pr->pr_protocol != IPPROTO_RAW) 180 ip6_protox[pr->pr_protocol] = pr - inet6sw; 181 182 inet6_pfil_hook.ph_type = PFIL_TYPE_AF; 183 inet6_pfil_hook.ph_af = AF_INET6; 184 if ((i = pfil_head_register(&inet6_pfil_hook)) != 0) { 185 kprintf("%s: WARNING: unable to register pfil hook, " 186 "error %d\n", __func__, i); 187 } 188 189 netisr_register(NETISR_IPV6, ip6_input, NULL); /* XXX hashfn */ 190 scope6_init(); 191 addrsel_policy_init(); 192 nd6_init(); 193 frag6_init(); 194 /* 195 * in many cases, random() here does NOT return random number 196 * as initialization during bootstrap time occur in fixed order. 197 */ 198 microtime(&tv); 199 ip6_flow_seq = krandom() ^ tv.tv_usec; 200 microtime(&tv); 201 ip6_desync_factor = (krandom() ^ tv.tv_usec) % MAX_TEMP_DESYNC_FACTOR; 202 } 203 204 static void 205 ip6_init2(void *dummy) 206 { 207 nd6_timer_init(); 208 in6_tmpaddrtimer_init(); 209 } 210 211 /* cheat */ 212 /* This must be after route_init(), which is now SI_ORDER_THIRD */ 213 SYSINIT(netinet6init2, SI_SUB_PROTO_DOMAIN, SI_ORDER_MIDDLE, ip6_init2, NULL); 214 215 extern struct route_in6 ip6_forward_rt; 216 217 static 218 void 219 ip6_input(netmsg_t msg) 220 { 221 struct mbuf *m = msg->packet.nm_packet; 222 struct ip6_hdr *ip6; 223 int off = sizeof(struct ip6_hdr), nest; 224 u_int32_t plen; 225 u_int32_t rtalert = ~0; 226 int nxt, ours = 0, rh_present = 0; 227 struct ifnet *deliverifp = NULL; 228 struct in6_addr odst; 229 int srcrt = 0; 230 231 #ifdef IPSEC 232 /* 233 * should the inner packet be considered authentic? 234 * see comment in ah4_input(). 235 */ 236 if (m) { 237 m->m_flags &= ~M_AUTHIPHDR; 238 m->m_flags &= ~M_AUTHIPDGM; 239 } 240 #endif 241 242 /* 243 * make sure we don't have onion peering information into m_aux. 244 */ 245 ip6_delaux(m); 246 247 /* 248 * mbuf statistics 249 */ 250 if (m->m_flags & M_EXT) { 251 if (m->m_next) 252 ip6stat.ip6s_mext2m++; 253 else 254 ip6stat.ip6s_mext1++; 255 } else { 256 #define M2MMAX NELEM(ip6stat.ip6s_m2m) 257 if (m->m_next) { 258 if (m->m_flags & M_LOOP) { 259 ip6stat.ip6s_m2m[loif[0].if_index]++; /* XXX */ 260 } else if (m->m_pkthdr.rcvif->if_index < M2MMAX) 261 ip6stat.ip6s_m2m[m->m_pkthdr.rcvif->if_index]++; 262 else 263 ip6stat.ip6s_m2m[0]++; 264 } else 265 ip6stat.ip6s_m1++; 266 #undef M2MMAX 267 } 268 269 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_receive); 270 ip6stat.ip6s_total++; 271 272 #ifndef PULLDOWN_TEST 273 /* 274 * L2 bridge code and some other code can return mbuf chain 275 * that does not conform to KAME requirement. too bad. 276 * XXX: fails to join if interface MTU > MCLBYTES. jumbogram? 277 */ 278 if (m && m->m_next != NULL && m->m_pkthdr.len < MCLBYTES) { 279 struct mbuf *n; 280 281 n = m_getb(m->m_pkthdr.len, MB_DONTWAIT, MT_HEADER, M_PKTHDR); 282 if (n == NULL) 283 goto bad; 284 M_MOVE_PKTHDR(n, m); 285 286 m_copydata(m, 0, n->m_pkthdr.len, mtod(n, caddr_t)); 287 n->m_len = n->m_pkthdr.len; 288 m_freem(m); 289 m = n; 290 } 291 IP6_EXTHDR_CHECK_VOIDRET(m, 0, sizeof(struct ip6_hdr)); 292 #endif 293 294 if (m->m_len < sizeof(struct ip6_hdr)) { 295 struct ifnet *inifp; 296 inifp = m->m_pkthdr.rcvif; 297 if ((m = m_pullup(m, sizeof(struct ip6_hdr))) == NULL) { 298 ip6stat.ip6s_toosmall++; 299 in6_ifstat_inc(inifp, ifs6_in_hdrerr); 300 goto bad2; 301 } 302 } 303 304 ip6 = mtod(m, struct ip6_hdr *); 305 306 if ((ip6->ip6_vfc & IPV6_VERSION_MASK) != IPV6_VERSION) { 307 ip6stat.ip6s_badvers++; 308 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_hdrerr); 309 goto bad; 310 } 311 312 /* 313 * Run through list of hooks for input packets. 314 * 315 * NB: Beware of the destination address changing 316 * (e.g. by NAT rewriting). When this happens, 317 * tell ip6_forward to do the right thing. 318 */ 319 if (pfil_has_hooks(&inet6_pfil_hook)) { 320 odst = ip6->ip6_dst; 321 if (pfil_run_hooks(&inet6_pfil_hook, &m, 322 m->m_pkthdr.rcvif, PFIL_IN)) { 323 goto bad2; 324 } 325 if (m == NULL) /* consumed by filter */ 326 goto bad2; 327 ip6 = mtod(m, struct ip6_hdr *); 328 srcrt = !IN6_ARE_ADDR_EQUAL(&odst, &ip6->ip6_dst); 329 } 330 331 ip6stat.ip6s_nxthist[ip6->ip6_nxt]++; 332 333 #ifdef ALTQ 334 if (altq_input != NULL && (*altq_input)(m, AF_INET6) == 0) { 335 /* packet is dropped by traffic conditioner */ 336 return; 337 } 338 #endif 339 340 /* 341 * Check with the firewall... 342 */ 343 if (ip6_fw_enable && ip6_fw_chk_ptr) { 344 u_short port = 0; 345 /* If ipfw says divert, we have to just drop packet */ 346 /* use port as a dummy argument */ 347 if ((*ip6_fw_chk_ptr)(&ip6, NULL, &port, &m)) { 348 m_freem(m); 349 m = NULL; 350 } 351 if (!m) 352 goto bad2; 353 } 354 355 /* 356 * Check against address spoofing/corruption. 357 */ 358 if (IN6_IS_ADDR_MULTICAST(&ip6->ip6_src) || 359 IN6_IS_ADDR_UNSPECIFIED(&ip6->ip6_dst)) { 360 /* 361 * XXX: "badscope" is not very suitable for a multicast source. 362 */ 363 ip6stat.ip6s_badscope++; 364 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_addrerr); 365 goto bad; 366 } 367 if ((IN6_IS_ADDR_LOOPBACK(&ip6->ip6_src) || 368 IN6_IS_ADDR_LOOPBACK(&ip6->ip6_dst)) && 369 (m->m_pkthdr.rcvif->if_flags & IFF_LOOPBACK) == 0) { 370 ip6stat.ip6s_badscope++; 371 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_addrerr); 372 goto bad; 373 } 374 375 /* 376 * The following check is not documented in specs. A malicious 377 * party may be able to use IPv4 mapped addr to confuse tcp/udp stack 378 * and bypass security checks (act as if it was from 127.0.0.1 by using 379 * IPv6 src ::ffff:127.0.0.1). Be cautious. 380 * 381 * This check chokes if we are in an SIIT cloud. As none of BSDs 382 * support IPv4-less kernel compilation, we cannot support SIIT 383 * environment at all. So, it makes more sense for us to reject any 384 * malicious packets for non-SIIT environment, than try to do a 385 * partical support for SIIT environment. 386 */ 387 if (IN6_IS_ADDR_V4MAPPED(&ip6->ip6_src) || 388 IN6_IS_ADDR_V4MAPPED(&ip6->ip6_dst)) { 389 ip6stat.ip6s_badscope++; 390 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_addrerr); 391 goto bad; 392 } 393 #if 0 394 /* 395 * Reject packets with IPv4 compatible addresses (auto tunnel). 396 * 397 * The code forbids auto tunnel relay case in RFC1933 (the check is 398 * stronger than RFC1933). We may want to re-enable it if mech-xx 399 * is revised to forbid relaying case. 400 */ 401 if (IN6_IS_ADDR_V4COMPAT(&ip6->ip6_src) || 402 IN6_IS_ADDR_V4COMPAT(&ip6->ip6_dst)) { 403 ip6stat.ip6s_badscope++; 404 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_addrerr); 405 goto bad; 406 } 407 #endif 408 409 /* drop packets if interface ID portion is already filled */ 410 if ((m->m_pkthdr.rcvif->if_flags & IFF_LOOPBACK) == 0) { 411 if (IN6_IS_SCOPE_LINKLOCAL(&ip6->ip6_src) && 412 ip6->ip6_src.s6_addr16[1]) { 413 ip6stat.ip6s_badscope++; 414 goto bad; 415 } 416 if (IN6_IS_SCOPE_LINKLOCAL(&ip6->ip6_dst) && 417 ip6->ip6_dst.s6_addr16[1]) { 418 ip6stat.ip6s_badscope++; 419 goto bad; 420 } 421 } 422 423 if (IN6_IS_SCOPE_LINKLOCAL(&ip6->ip6_src)) 424 ip6->ip6_src.s6_addr16[1] 425 = htons(m->m_pkthdr.rcvif->if_index); 426 if (IN6_IS_SCOPE_LINKLOCAL(&ip6->ip6_dst)) 427 ip6->ip6_dst.s6_addr16[1] 428 = htons(m->m_pkthdr.rcvif->if_index); 429 430 #if 0 /* this case seems to be unnecessary. (jinmei, 20010401) */ 431 /* 432 * We use rt->rt_ifp to determine if the address is ours or not. 433 * If rt_ifp is lo0, the address is ours. 434 * The problem here is, rt->rt_ifp for fe80::%lo0/64 is set to lo0, 435 * so any address under fe80::%lo0/64 will be mistakenly considered 436 * local. The special case is supplied to handle the case properly 437 * by actually looking at interface addresses 438 * (using in6ifa_ifpwithaddr). 439 */ 440 if ((m->m_pkthdr.rcvif->if_flags & IFF_LOOPBACK) && 441 IN6_IS_ADDR_LINKLOCAL(&ip6->ip6_dst)) { 442 if (!in6ifa_ifpwithaddr(m->m_pkthdr.rcvif, &ip6->ip6_dst)) { 443 icmp6_error(m, ICMP6_DST_UNREACH, 444 ICMP6_DST_UNREACH_ADDR, 0); 445 /* m is already freed */ 446 goto bad2; 447 } 448 449 ours = 1; 450 deliverifp = m->m_pkthdr.rcvif; 451 goto hbhcheck; 452 } 453 #endif 454 455 /* 456 * Multicast check 457 * 458 * WARNING! For general subnet proxying the interface hw will 459 * likely filter out the multicast solicitations, so 460 * the interface must be in promiscuous mode. 461 */ 462 if (IN6_IS_ADDR_MULTICAST(&ip6->ip6_dst)) { 463 ours = 1; 464 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_mcast); 465 deliverifp = m->m_pkthdr.rcvif; 466 goto hbhcheck; 467 } 468 469 /* 470 * Unicast check 471 */ 472 switch (ip6_ours_check_algorithm) { 473 default: 474 /* 475 * XXX: I intentionally broke our indentation rule here, 476 * since this switch-case is just for measurement and 477 * therefore should soon be removed. 478 */ 479 if (ip6_forward_rt.ro_rt != NULL && 480 (ip6_forward_rt.ro_rt->rt_flags & RTF_UP) && 481 IN6_ARE_ADDR_EQUAL(&ip6->ip6_dst, 482 &((struct sockaddr_in6 *)(&ip6_forward_rt.ro_dst))->sin6_addr)) 483 ip6stat.ip6s_forward_cachehit++; 484 else { 485 struct sockaddr_in6 *dst6; 486 487 if (ip6_forward_rt.ro_rt) { 488 /* route is down or destination is different */ 489 ip6stat.ip6s_forward_cachemiss++; 490 RTFREE(ip6_forward_rt.ro_rt); 491 ip6_forward_rt.ro_rt = 0; 492 } 493 494 bzero(&ip6_forward_rt.ro_dst, sizeof(struct sockaddr_in6)); 495 dst6 = &ip6_forward_rt.ro_dst; 496 dst6->sin6_len = sizeof(struct sockaddr_in6); 497 dst6->sin6_family = AF_INET6; 498 dst6->sin6_addr = ip6->ip6_dst; 499 500 rtalloc_ign((struct route *)&ip6_forward_rt, RTF_PRCLONING); 501 } 502 503 #define rt6_key(r) ((struct sockaddr_in6 *)((r)->rt_nodes->rn_key)) 504 505 /* 506 * Accept the packet if the forwarding interface to the destination 507 * according to the routing table is the loopback interface, 508 * unless the associated route has a gateway. 509 * Note that this approach causes to accept a packet if there is a 510 * route to the loopback interface for the destination of the packet. 511 * But we think it's even useful in some situations, e.g. when using 512 * a special daemon which wants to intercept the packet. 513 * 514 * XXX: some OSes automatically make a cloned route for the destination 515 * of an outgoing packet. If the outgoing interface of the packet 516 * is a loopback one, the kernel would consider the packet to be 517 * accepted, even if we have no such address assinged on the interface. 518 * We check the cloned flag of the route entry to reject such cases, 519 * assuming that route entries for our own addresses are not made by 520 * cloning (it should be true because in6_addloop explicitly installs 521 * the host route). However, we might have to do an explicit check 522 * while it would be less efficient. Or, should we rather install a 523 * reject route for such a case? 524 */ 525 if (ip6_forward_rt.ro_rt && 526 (ip6_forward_rt.ro_rt->rt_flags & 527 (RTF_HOST|RTF_GATEWAY)) == RTF_HOST && 528 #ifdef RTF_WASCLONED 529 !(ip6_forward_rt.ro_rt->rt_flags & RTF_WASCLONED) && 530 #endif 531 #ifdef RTF_CLONED 532 !(ip6_forward_rt.ro_rt->rt_flags & RTF_CLONED) && 533 #endif 534 #if 0 535 /* 536 * The check below is redundant since the comparison of 537 * the destination and the key of the rtentry has 538 * already done through looking up the routing table. 539 */ 540 IN6_ARE_ADDR_EQUAL(&ip6->ip6_dst, 541 &rt6_key(ip6_forward_rt.ro_rt)->sin6_addr) 542 #endif 543 ip6_forward_rt.ro_rt->rt_ifp->if_type == IFT_LOOP) { 544 struct in6_ifaddr *ia6 = 545 (struct in6_ifaddr *)ip6_forward_rt.ro_rt->rt_ifa; 546 547 /* 548 * record address information into m_aux. 549 */ 550 ip6_setdstifaddr(m, ia6); 551 552 /* 553 * packets to a tentative, duplicated, or somehow invalid 554 * address must not be accepted. 555 */ 556 if (!(ia6->ia6_flags & IN6_IFF_NOTREADY)) { 557 /* this address is ready */ 558 ours = 1; 559 deliverifp = ia6->ia_ifp; /* correct? */ 560 /* Count the packet in the ip address stats */ 561 IFA_STAT_INC(&ia6->ia_ifa, ipackets, 1); 562 IFA_STAT_INC(&ia6->ia_ifa, ibytes, m->m_pkthdr.len); 563 goto hbhcheck; 564 } else { 565 /* address is not ready, so discard the packet. */ 566 nd6log((LOG_INFO, 567 "ip6_input: packet to an unready address %s->%s\n", 568 ip6_sprintf(&ip6->ip6_src), 569 ip6_sprintf(&ip6->ip6_dst))); 570 571 goto bad; 572 } 573 } 574 } /* XXX indentation (see above) */ 575 576 /* 577 * FAITH(Firewall Aided Internet Translator) 578 */ 579 if (ip6_keepfaith) { 580 if (ip6_forward_rt.ro_rt && ip6_forward_rt.ro_rt->rt_ifp 581 && ip6_forward_rt.ro_rt->rt_ifp->if_type == IFT_FAITH) { 582 /* XXX do we need more sanity checks? */ 583 ours = 1; 584 deliverifp = ip6_forward_rt.ro_rt->rt_ifp; /* faith */ 585 goto hbhcheck; 586 } 587 } 588 589 /* 590 * Now there is no reason to process the packet if it's not our own 591 * and we're not a router. 592 */ 593 if (!ip6_forwarding) { 594 ip6stat.ip6s_cantforward++; 595 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_discard); 596 goto bad; 597 } 598 599 hbhcheck: 600 /* 601 * record address information into m_aux, if we don't have one yet. 602 * note that we are unable to record it, if the address is not listed 603 * as our interface address (e.g. multicast addresses, addresses 604 * within FAITH prefixes and such). 605 */ 606 if (deliverifp && !ip6_getdstifaddr(m)) { 607 struct in6_ifaddr *ia6; 608 609 ia6 = in6_ifawithifp(deliverifp, &ip6->ip6_dst); 610 if (ia6) { 611 if (!ip6_setdstifaddr(m, ia6)) { 612 /* 613 * XXX maybe we should drop the packet here, 614 * as we could not provide enough information 615 * to the upper layers. 616 */ 617 } 618 } 619 } 620 621 /* 622 * Process Hop-by-Hop options header if it's contained. 623 * m may be modified in ip6_hopopts_input(). 624 * If a JumboPayload option is included, plen will also be modified. 625 */ 626 plen = (u_int32_t)ntohs(ip6->ip6_plen); 627 if (ip6->ip6_nxt == IPPROTO_HOPOPTS) { 628 struct ip6_hbh *hbh; 629 630 if (ip6_hopopts_input(&plen, &rtalert, &m, &off)) { 631 #if 0 /*touches NULL pointer*/ 632 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_discard); 633 #endif 634 goto bad2; /* m have already been freed */ 635 } 636 637 /* adjust pointer */ 638 ip6 = mtod(m, struct ip6_hdr *); 639 640 /* 641 * if the payload length field is 0 and the next header field 642 * indicates Hop-by-Hop Options header, then a Jumbo Payload 643 * option MUST be included. 644 */ 645 if (ip6->ip6_plen == 0 && plen == 0) { 646 /* 647 * Note that if a valid jumbo payload option is 648 * contained, ip6_hoptops_input() must set a valid 649 * (non-zero) payload length to the variable plen. 650 */ 651 ip6stat.ip6s_badoptions++; 652 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_discard); 653 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_hdrerr); 654 icmp6_error(m, ICMP6_PARAM_PROB, 655 ICMP6_PARAMPROB_HEADER, 656 (caddr_t)&ip6->ip6_plen - (caddr_t)ip6); 657 goto bad2; 658 } 659 #ifndef PULLDOWN_TEST 660 /* ip6_hopopts_input() ensures that mbuf is contiguous */ 661 hbh = (struct ip6_hbh *)(ip6 + 1); 662 #else 663 IP6_EXTHDR_GET(hbh, struct ip6_hbh *, m, sizeof(struct ip6_hdr), 664 sizeof(struct ip6_hbh)); 665 if (hbh == NULL) { 666 ip6stat.ip6s_tooshort++; 667 goto bad2; 668 } 669 #endif 670 nxt = hbh->ip6h_nxt; 671 672 /* 673 * If we are acting as a router and the packet contains a 674 * router alert option, see if we know the option value. 675 * Currently, we only support the option value for MLD, in which 676 * case we should pass the packet to the multicast routing 677 * daemon. 678 */ 679 if (rtalert != ~0 && ip6_forwarding) { 680 switch (rtalert) { 681 case IP6OPT_RTALERT_MLD: 682 ours = 1; 683 break; 684 default: 685 /* 686 * RFC2711 requires unrecognized values must be 687 * silently ignored. 688 */ 689 break; 690 } 691 } 692 } else 693 nxt = ip6->ip6_nxt; 694 695 /* 696 * Check that the amount of data in the buffers 697 * is as at least much as the IPv6 header would have us expect. 698 * Trim mbufs if longer than we expect. 699 * Drop packet if shorter than we expect. 700 */ 701 if (m->m_pkthdr.len - sizeof(struct ip6_hdr) < plen) { 702 ip6stat.ip6s_tooshort++; 703 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_truncated); 704 goto bad; 705 } 706 if (m->m_pkthdr.len > sizeof(struct ip6_hdr) + plen) { 707 if (m->m_len == m->m_pkthdr.len) { 708 m->m_len = sizeof(struct ip6_hdr) + plen; 709 m->m_pkthdr.len = sizeof(struct ip6_hdr) + plen; 710 } else 711 m_adj(m, sizeof(struct ip6_hdr) + plen - m->m_pkthdr.len); 712 } 713 714 /* 715 * Forward if desirable. 716 */ 717 if (IN6_IS_ADDR_MULTICAST(&ip6->ip6_dst)) { 718 /* 719 * If we are acting as a multicast router, all 720 * incoming multicast packets are passed to the 721 * kernel-level multicast forwarding function. 722 * The packet is returned (relatively) intact; if 723 * ip6_mforward() returns a non-zero value, the packet 724 * must be discarded, else it may be accepted below. 725 */ 726 if (ip6_mrouter && ip6_mforward(ip6, m->m_pkthdr.rcvif, m)) { 727 ip6stat.ip6s_cantforward++; 728 goto bad; 729 } 730 if (!ours) 731 goto bad; 732 } else if (!ours) { 733 ip6_forward(m, srcrt); 734 goto bad2; 735 } 736 737 ip6 = mtod(m, struct ip6_hdr *); 738 739 /* 740 * Malicious party may be able to use IPv4 mapped addr to confuse 741 * tcp/udp stack and bypass security checks (act as if it was from 742 * 127.0.0.1 by using IPv6 src ::ffff:127.0.0.1). Be cautious. 743 * 744 * For SIIT end node behavior, you may want to disable the check. 745 * However, you will become vulnerable to attacks using IPv4 mapped 746 * source. 747 */ 748 if (IN6_IS_ADDR_V4MAPPED(&ip6->ip6_src) || 749 IN6_IS_ADDR_V4MAPPED(&ip6->ip6_dst)) { 750 ip6stat.ip6s_badscope++; 751 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_addrerr); 752 goto bad; 753 } 754 755 /* 756 * Tell launch routine the next header 757 */ 758 ip6stat.ip6s_delivered++; 759 in6_ifstat_inc(deliverifp, ifs6_in_deliver); 760 nest = 0; 761 762 rh_present = 0; 763 while (nxt != IPPROTO_DONE) { 764 struct protosw *sw6; 765 766 if (ip6_hdrnestlimit && (++nest > ip6_hdrnestlimit)) { 767 ip6stat.ip6s_toomanyhdr++; 768 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_hdrerr); 769 goto bad; 770 } 771 772 /* 773 * protection against faulty packet - there should be 774 * more sanity checks in header chain processing. 775 */ 776 if (m->m_pkthdr.len < off) { 777 ip6stat.ip6s_tooshort++; 778 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_truncated); 779 goto bad; 780 } 781 782 #if 0 783 /* 784 * do we need to do it for every header? yeah, other 785 * functions can play with it (like re-allocate and copy). 786 */ 787 mhist = ip6_addaux(m); 788 if (mhist && M_TRAILINGSPACE(mhist) >= sizeof(nxt)) { 789 hist = mtod(mhist, caddr_t) + mhist->m_len; 790 bcopy(&nxt, hist, sizeof(nxt)); 791 mhist->m_len += sizeof(nxt); 792 } else { 793 ip6stat.ip6s_toomanyhdr++; 794 goto bad; 795 } 796 #endif 797 798 if (nxt == IPPROTO_ROUTING) { 799 if (rh_present++) { 800 in6_ifstat_inc(m->m_pkthdr.rcvif, 801 ifs6_in_hdrerr); 802 ip6stat.ip6s_badoptions++; 803 goto bad; 804 } 805 } 806 807 sw6 = &inet6sw[ip6_protox[nxt]]; 808 #ifdef IPSEC 809 /* 810 * enforce IPsec policy checking if we are seeing last header. 811 * note that we do not visit this with protocols with pcb layer 812 * code - like udp/tcp/raw ip. 813 */ 814 if ((sw6->pr_flags & PR_LASTHDR) && ipsec6_in_reject(m, NULL)) { 815 ipsec6stat.in_polvio++; 816 goto bad; 817 } 818 #endif 819 /* 820 * If this is a terminal header forward to the port, otherwise 821 * process synchronously for more headers. 822 */ 823 if (sw6->pr_flags & PR_LASTHDR) { 824 struct netmsg_packet *pmsg; 825 lwkt_port_t port; 826 827 port = netisr_cpuport(0); /* XXX */ 828 KKASSERT(port != NULL); 829 pmsg = &m->m_hdr.mh_netmsg; 830 netmsg_init(&pmsg->base, NULL, 831 &netisr_apanic_rport, 832 0, transport6_processing_handler); 833 pmsg->nm_packet = m; 834 pmsg->nm_nxt = nxt; 835 pmsg->base.lmsg.u.ms_result = off; 836 lwkt_sendmsg(port, &pmsg->base.lmsg); 837 /* done with m */ 838 nxt = IPPROTO_DONE; 839 } else { 840 nxt = sw6->pr_input(&m, &off, nxt); 841 } 842 } 843 goto bad2; 844 bad: 845 m_freem(m); 846 bad2: 847 ; 848 /* msg was embedded in the mbuf, do not reply! */ 849 } 850 851 /* 852 * We have to call the pr_input() function from the correct protocol 853 * thread. The sw6->pr_soport() request at the end of ip6_input() 854 * returns the port and we forward a netmsg to the port to execute 855 * this function. 856 */ 857 static void 858 transport6_processing_handler(netmsg_t netmsg) 859 { 860 struct netmsg_packet *pmsg = (struct netmsg_packet *)netmsg; 861 struct protosw *sw6; 862 int hlen; 863 int nxt; 864 865 sw6 = &inet6sw[ip6_protox[pmsg->nm_nxt]]; 866 hlen = pmsg->base.lmsg.u.ms_result; 867 868 nxt = sw6->pr_input(&pmsg->nm_packet, &hlen, pmsg->nm_nxt); 869 KKASSERT(nxt == IPPROTO_DONE); 870 /* netmsg was embedded in the mbuf, do not reply! */ 871 } 872 873 /* 874 * set/grab in6_ifaddr correspond to IPv6 destination address. 875 * XXX backward compatibility wrapper 876 */ 877 static struct ip6aux * 878 ip6_setdstifaddr(struct mbuf *m, struct in6_ifaddr *ia6) 879 { 880 struct ip6aux *n; 881 882 n = ip6_addaux(m); 883 if (n) 884 n->ip6a_dstia6 = ia6; 885 return n; /* NULL if failed to set */ 886 } 887 888 struct in6_ifaddr * 889 ip6_getdstifaddr(struct mbuf *m) 890 { 891 struct ip6aux *n; 892 893 n = ip6_findaux(m); 894 if (n) 895 return n->ip6a_dstia6; 896 else 897 return NULL; 898 } 899 900 /* 901 * Hop-by-Hop options header processing. If a valid jumbo payload option is 902 * included, the real payload length will be stored in plenp. 903 */ 904 static int 905 ip6_hopopts_input(u_int32_t *plenp, 906 u_int32_t *rtalertp,/* XXX: should be stored more smart way */ 907 struct mbuf **mp, 908 int *offp) 909 { 910 struct mbuf *m = *mp; 911 int off = *offp, hbhlen; 912 struct ip6_hbh *hbh; 913 914 /* validation of the length of the header */ 915 #ifndef PULLDOWN_TEST 916 IP6_EXTHDR_CHECK(m, off, sizeof(*hbh), -1); 917 hbh = (struct ip6_hbh *)(mtod(m, caddr_t) + off); 918 hbhlen = (hbh->ip6h_len + 1) << 3; 919 920 IP6_EXTHDR_CHECK(m, off, hbhlen, -1); 921 hbh = (struct ip6_hbh *)(mtod(m, caddr_t) + off); 922 #else 923 IP6_EXTHDR_GET(hbh, struct ip6_hbh *, m, 924 sizeof(struct ip6_hdr), sizeof(struct ip6_hbh)); 925 if (hbh == NULL) { 926 ip6stat.ip6s_tooshort++; 927 return -1; 928 } 929 hbhlen = (hbh->ip6h_len + 1) << 3; 930 IP6_EXTHDR_GET(hbh, struct ip6_hbh *, m, sizeof(struct ip6_hdr), 931 hbhlen); 932 if (hbh == NULL) { 933 ip6stat.ip6s_tooshort++; 934 return -1; 935 } 936 #endif 937 off += hbhlen; 938 hbhlen -= sizeof(struct ip6_hbh); 939 940 if (ip6_process_hopopts(m, (u_int8_t *)hbh + sizeof(struct ip6_hbh), 941 hbhlen, rtalertp, plenp) < 0) 942 return (-1); 943 944 *offp = off; 945 *mp = m; 946 return (0); 947 } 948 949 /* 950 * Search header for all Hop-by-hop options and process each option. 951 * This function is separate from ip6_hopopts_input() in order to 952 * handle a case where the sending node itself process its hop-by-hop 953 * options header. In such a case, the function is called from ip6_output(). 954 * 955 * The function assumes that hbh header is located right after the IPv6 header 956 * (RFC2460 p7), opthead is pointer into data content in m, and opthead to 957 * opthead + hbhlen is located in continuous memory region. 958 */ 959 int 960 ip6_process_hopopts(struct mbuf *m, u_int8_t *opthead, int hbhlen, 961 u_int32_t *rtalertp, u_int32_t *plenp) 962 { 963 struct ip6_hdr *ip6; 964 int optlen = 0; 965 u_int8_t *opt = opthead; 966 u_int16_t rtalert_val; 967 u_int32_t jumboplen; 968 const int erroff = sizeof(struct ip6_hdr) + sizeof(struct ip6_hbh); 969 970 for (; hbhlen > 0; hbhlen -= optlen, opt += optlen) { 971 switch (*opt) { 972 case IP6OPT_PAD1: 973 optlen = 1; 974 break; 975 case IP6OPT_PADN: 976 if (hbhlen < IP6OPT_MINLEN) { 977 ip6stat.ip6s_toosmall++; 978 goto bad; 979 } 980 optlen = *(opt + 1) + 2; 981 break; 982 case IP6OPT_RTALERT: 983 /* XXX may need check for alignment */ 984 if (hbhlen < IP6OPT_RTALERT_LEN) { 985 ip6stat.ip6s_toosmall++; 986 goto bad; 987 } 988 if (*(opt + 1) != IP6OPT_RTALERT_LEN - 2) { 989 /* XXX stat */ 990 icmp6_error(m, ICMP6_PARAM_PROB, 991 ICMP6_PARAMPROB_HEADER, 992 erroff + opt + 1 - opthead); 993 return (-1); 994 } 995 optlen = IP6OPT_RTALERT_LEN; 996 bcopy((caddr_t)(opt + 2), (caddr_t)&rtalert_val, 2); 997 *rtalertp = ntohs(rtalert_val); 998 break; 999 case IP6OPT_JUMBO: 1000 /* XXX may need check for alignment */ 1001 if (hbhlen < IP6OPT_JUMBO_LEN) { 1002 ip6stat.ip6s_toosmall++; 1003 goto bad; 1004 } 1005 if (*(opt + 1) != IP6OPT_JUMBO_LEN - 2) { 1006 /* XXX stat */ 1007 icmp6_error(m, ICMP6_PARAM_PROB, 1008 ICMP6_PARAMPROB_HEADER, 1009 erroff + opt + 1 - opthead); 1010 return (-1); 1011 } 1012 optlen = IP6OPT_JUMBO_LEN; 1013 1014 /* 1015 * IPv6 packets that have non 0 payload length 1016 * must not contain a jumbo payload option. 1017 */ 1018 ip6 = mtod(m, struct ip6_hdr *); 1019 if (ip6->ip6_plen) { 1020 ip6stat.ip6s_badoptions++; 1021 icmp6_error(m, ICMP6_PARAM_PROB, 1022 ICMP6_PARAMPROB_HEADER, 1023 erroff + opt - opthead); 1024 return (-1); 1025 } 1026 1027 /* 1028 * We may see jumbolen in unaligned location, so 1029 * we'd need to perform bcopy(). 1030 */ 1031 bcopy(opt + 2, &jumboplen, sizeof(jumboplen)); 1032 jumboplen = (u_int32_t)htonl(jumboplen); 1033 1034 #if 1 1035 /* 1036 * if there are multiple jumbo payload options, 1037 * *plenp will be non-zero and the packet will be 1038 * rejected. 1039 * the behavior may need some debate in ipngwg - 1040 * multiple options does not make sense, however, 1041 * there's no explicit mention in specification. 1042 */ 1043 if (*plenp != 0) { 1044 ip6stat.ip6s_badoptions++; 1045 icmp6_error(m, ICMP6_PARAM_PROB, 1046 ICMP6_PARAMPROB_HEADER, 1047 erroff + opt + 2 - opthead); 1048 return (-1); 1049 } 1050 #endif 1051 1052 /* 1053 * jumbo payload length must be larger than 65535. 1054 */ 1055 if (jumboplen <= IPV6_MAXPACKET) { 1056 ip6stat.ip6s_badoptions++; 1057 icmp6_error(m, ICMP6_PARAM_PROB, 1058 ICMP6_PARAMPROB_HEADER, 1059 erroff + opt + 2 - opthead); 1060 return (-1); 1061 } 1062 *plenp = jumboplen; 1063 1064 break; 1065 default: /* unknown option */ 1066 if (hbhlen < IP6OPT_MINLEN) { 1067 ip6stat.ip6s_toosmall++; 1068 goto bad; 1069 } 1070 optlen = ip6_unknown_opt(opt, m, 1071 erroff + opt - opthead); 1072 if (optlen == -1) 1073 return (-1); 1074 optlen += 2; 1075 break; 1076 } 1077 } 1078 1079 return (0); 1080 1081 bad: 1082 m_freem(m); 1083 return (-1); 1084 } 1085 1086 /* 1087 * Unknown option processing. 1088 * The third argument `off' is the offset from the IPv6 header to the option, 1089 * which is necessary if the IPv6 header the and option header and IPv6 header 1090 * is not continuous in order to return an ICMPv6 error. 1091 */ 1092 int 1093 ip6_unknown_opt(u_int8_t *optp, struct mbuf *m, int off) 1094 { 1095 struct ip6_hdr *ip6; 1096 1097 switch (IP6OPT_TYPE(*optp)) { 1098 case IP6OPT_TYPE_SKIP: /* ignore the option */ 1099 return ((int)*(optp + 1)); 1100 case IP6OPT_TYPE_DISCARD: /* silently discard */ 1101 m_freem(m); 1102 return (-1); 1103 case IP6OPT_TYPE_FORCEICMP: /* send ICMP even if multicasted */ 1104 ip6stat.ip6s_badoptions++; 1105 icmp6_error(m, ICMP6_PARAM_PROB, ICMP6_PARAMPROB_OPTION, off); 1106 return (-1); 1107 case IP6OPT_TYPE_ICMP: /* send ICMP if not multicasted */ 1108 ip6stat.ip6s_badoptions++; 1109 ip6 = mtod(m, struct ip6_hdr *); 1110 if (IN6_IS_ADDR_MULTICAST(&ip6->ip6_dst) || 1111 (m->m_flags & (M_BCAST|M_MCAST))) 1112 m_freem(m); 1113 else 1114 icmp6_error(m, ICMP6_PARAM_PROB, 1115 ICMP6_PARAMPROB_OPTION, off); 1116 return (-1); 1117 } 1118 1119 m_freem(m); /* XXX: NOTREACHED */ 1120 return (-1); 1121 } 1122 1123 /* 1124 * Create the "control" list for this pcb. 1125 * The function will not modify mbuf chain at all. 1126 * 1127 * with KAME mbuf chain restriction: 1128 * The routine will be called from upper layer handlers like tcp6_input(). 1129 * Thus the routine assumes that the caller (tcp6_input) have already 1130 * called IP6_EXTHDR_CHECK() and all the extension headers are located in the 1131 * very first mbuf on the mbuf chain. 1132 */ 1133 void 1134 ip6_savecontrol(struct inpcb *in6p, struct mbuf **mp, struct ip6_hdr *ip6, 1135 struct mbuf *m) 1136 { 1137 #define IS2292(x, y) ((in6p->in6p_flags & IN6P_RFC2292) ? (x) : (y)) 1138 struct thread *td = curthread; /* XXX */ 1139 int privileged = 0; 1140 1141 1142 if (priv_check(td, PRIV_ROOT) == 0) 1143 privileged++; 1144 1145 #ifdef SO_TIMESTAMP 1146 if (in6p->in6p_socket->so_options & SO_TIMESTAMP) { 1147 struct timeval tv; 1148 1149 microtime(&tv); 1150 *mp = sbcreatecontrol((caddr_t) &tv, sizeof(tv), 1151 SCM_TIMESTAMP, SOL_SOCKET); 1152 if (*mp) { 1153 mp = &(*mp)->m_next; 1154 } 1155 } 1156 #endif 1157 1158 /* RFC 2292 sec. 5 */ 1159 if (in6p->in6p_flags & IN6P_PKTINFO) { 1160 struct in6_pktinfo pi6; 1161 bcopy(&ip6->ip6_dst, &pi6.ipi6_addr, sizeof(struct in6_addr)); 1162 if (IN6_IS_SCOPE_LINKLOCAL(&pi6.ipi6_addr)) 1163 pi6.ipi6_addr.s6_addr16[1] = 0; 1164 pi6.ipi6_ifindex = (m && m->m_pkthdr.rcvif) 1165 ? m->m_pkthdr.rcvif->if_index 1166 : 0; 1167 *mp = sbcreatecontrol((caddr_t) &pi6, 1168 sizeof(struct in6_pktinfo), IS2292(IPV6_2292PKTINFO, IPV6_PKTINFO), 1169 IPPROTO_IPV6); 1170 if (*mp) 1171 mp = &(*mp)->m_next; 1172 } 1173 1174 if (in6p->in6p_flags & IN6P_HOPLIMIT) { 1175 int hlim = ip6->ip6_hlim & 0xff; 1176 *mp = sbcreatecontrol((caddr_t) &hlim, 1177 sizeof(int), IS2292(IPV6_2292HOPLIMIT, IPV6_HOPLIMIT), IPPROTO_IPV6); 1178 if (*mp) 1179 mp = &(*mp)->m_next; 1180 } 1181 1182 if ((in6p->in6p_flags & IN6P_TCLASS) != 0) { 1183 u_int32_t flowinfo; 1184 int tclass; 1185 1186 flowinfo = (u_int32_t)ntohl(ip6->ip6_flow & IPV6_FLOWINFO_MASK); 1187 flowinfo >>= 20; 1188 1189 tclass = flowinfo & 0xff; 1190 *mp = sbcreatecontrol((caddr_t) &tclass, sizeof(tclass), 1191 IPV6_TCLASS, IPPROTO_IPV6); 1192 if (*mp) 1193 mp = &(*mp)->m_next; 1194 } 1195 1196 /* 1197 * IPV6_HOPOPTS socket option. We require super-user privilege 1198 * for the option, but it might be too strict, since there might 1199 * be some hop-by-hop options which can be returned to normal user. 1200 * See RFC 2292 section 6. 1201 */ 1202 if ((in6p->in6p_flags & IN6P_HOPOPTS) && privileged) { 1203 /* 1204 * Check if a hop-by-hop options header is contatined in the 1205 * received packet, and if so, store the options as ancillary 1206 * data. Note that a hop-by-hop options header must be 1207 * just after the IPv6 header, which fact is assured through 1208 * the IPv6 input processing. 1209 */ 1210 struct ip6_hdr *ip6 = mtod(m, struct ip6_hdr *); 1211 if (ip6->ip6_nxt == IPPROTO_HOPOPTS) { 1212 struct ip6_hbh *hbh; 1213 int hbhlen = 0; 1214 #ifdef PULLDOWN_TEST 1215 struct mbuf *ext; 1216 #endif 1217 1218 #ifndef PULLDOWN_TEST 1219 hbh = (struct ip6_hbh *)(ip6 + 1); 1220 hbhlen = (hbh->ip6h_len + 1) << 3; 1221 #else 1222 ext = ip6_pullexthdr(m, sizeof(struct ip6_hdr), 1223 ip6->ip6_nxt); 1224 if (ext == NULL) { 1225 ip6stat.ip6s_tooshort++; 1226 return; 1227 } 1228 hbh = mtod(ext, struct ip6_hbh *); 1229 hbhlen = (hbh->ip6h_len + 1) << 3; 1230 if (hbhlen != ext->m_len) { 1231 m_freem(ext); 1232 ip6stat.ip6s_tooshort++; 1233 return; 1234 } 1235 #endif 1236 1237 /* 1238 * XXX: We copy whole the header even if a jumbo 1239 * payload option is included, which option is to 1240 * be removed before returning in the RFC 2292. 1241 * Note: this constraint is removed in 2292bis. 1242 */ 1243 *mp = sbcreatecontrol((caddr_t)hbh, hbhlen, 1244 IS2292(IPV6_2292HOPOPTS, IPV6_HOPOPTS), IPPROTO_IPV6); 1245 if (*mp) 1246 mp = &(*mp)->m_next; 1247 #ifdef PULLDOWN_TEST 1248 m_freem(ext); 1249 #endif 1250 } 1251 } 1252 1253 /* IPV6_DSTOPTS and IPV6_RTHDR socket options */ 1254 if ((in6p->in6p_flags & (IN6P_DSTOPTS | IN6P_RTHDRDSTOPTS)) != 0) { 1255 int proto, off, nxt; 1256 1257 /* 1258 * go through the header chain to see if a routing header is 1259 * contained in the packet. We need this information to store 1260 * destination options headers (if any) properly. 1261 * XXX: performance issue. We should record this info when 1262 * processing extension headers in incoming routine. 1263 * (todo) use m_aux? 1264 */ 1265 proto = IPPROTO_IPV6; 1266 off = 0; 1267 nxt = -1; 1268 while (1) { 1269 int newoff; 1270 1271 newoff = ip6_nexthdr(m, off, proto, &nxt); 1272 if (newoff < 0) 1273 break; 1274 if (newoff < off) /* invalid, check for safety */ 1275 break; 1276 if ((proto = nxt) == IPPROTO_ROUTING) 1277 break; 1278 off = newoff; 1279 } 1280 } 1281 1282 if ((in6p->in6p_flags & 1283 (IN6P_RTHDR | IN6P_DSTOPTS | IN6P_RTHDRDSTOPTS)) != 0) { 1284 struct ip6_hdr *ip6 = mtod(m, struct ip6_hdr *); 1285 int nxt = ip6->ip6_nxt, off = sizeof(struct ip6_hdr); 1286 1287 /* 1288 * Search for destination options headers or routing 1289 * header(s) through the header chain, and stores each 1290 * header as ancillary data. 1291 * Note that the order of the headers remains in 1292 * the chain of ancillary data. 1293 */ 1294 while (1) { /* is explicit loop prevention necessary? */ 1295 struct ip6_ext *ip6e = NULL; 1296 int elen; 1297 #ifdef PULLDOWN_TEST 1298 struct mbuf *ext = NULL; 1299 #endif 1300 1301 /* 1302 * if it is not an extension header, don't try to 1303 * pull it from the chain. 1304 */ 1305 switch (nxt) { 1306 case IPPROTO_DSTOPTS: 1307 case IPPROTO_ROUTING: 1308 case IPPROTO_HOPOPTS: 1309 case IPPROTO_AH: /* is it possible? */ 1310 break; 1311 default: 1312 goto loopend; 1313 } 1314 1315 #ifndef PULLDOWN_TEST 1316 if (off + sizeof(*ip6e) > m->m_len) 1317 goto loopend; 1318 ip6e = (struct ip6_ext *)(mtod(m, caddr_t) + off); 1319 if (nxt == IPPROTO_AH) 1320 elen = (ip6e->ip6e_len + 2) << 2; 1321 else 1322 elen = (ip6e->ip6e_len + 1) << 3; 1323 if (off + elen > m->m_len) 1324 goto loopend; 1325 #else 1326 ext = ip6_pullexthdr(m, off, nxt); 1327 if (ext == NULL) { 1328 ip6stat.ip6s_tooshort++; 1329 return; 1330 } 1331 ip6e = mtod(ext, struct ip6_ext *); 1332 if (nxt == IPPROTO_AH) 1333 elen = (ip6e->ip6e_len + 2) << 2; 1334 else 1335 elen = (ip6e->ip6e_len + 1) << 3; 1336 if (elen != ext->m_len) { 1337 m_freem(ext); 1338 ip6stat.ip6s_tooshort++; 1339 return; 1340 } 1341 #endif 1342 1343 switch (nxt) { 1344 case IPPROTO_DSTOPTS: 1345 if ((in6p->in6p_flags & IN6P_DSTOPTS) == 0) 1346 break; 1347 1348 /* 1349 * We also require super-user privilege for 1350 * the option. 1351 * See the comments on IN6_HOPOPTS. 1352 */ 1353 if (!privileged) 1354 break; 1355 1356 *mp = sbcreatecontrol((caddr_t)ip6e, elen, 1357 IS2292(IPV6_2292DSTOPTS, IPV6_DSTOPTS), 1358 IPPROTO_IPV6); 1359 if (*mp) 1360 mp = &(*mp)->m_next; 1361 break; 1362 case IPPROTO_ROUTING: 1363 if (!(in6p->in6p_flags & IN6P_RTHDR)) 1364 break; 1365 1366 *mp = sbcreatecontrol((caddr_t)ip6e, elen, 1367 IS2292(IPV6_2292RTHDR, IPV6_RTHDR), 1368 IPPROTO_IPV6); 1369 if (*mp) 1370 mp = &(*mp)->m_next; 1371 break; 1372 case IPPROTO_HOPOPTS: 1373 case IPPROTO_AH: /* is it possible? */ 1374 break; 1375 1376 default: 1377 /* 1378 * other cases have been filtered in the above. 1379 * none will visit this case. here we supply 1380 * the code just in case (nxt overwritten or 1381 * other cases). 1382 */ 1383 #ifdef PULLDOWN_TEST 1384 m_freem(ext); 1385 #endif 1386 goto loopend; 1387 1388 } 1389 1390 /* proceed with the next header. */ 1391 off += elen; 1392 nxt = ip6e->ip6e_nxt; 1393 ip6e = NULL; 1394 #ifdef PULLDOWN_TEST 1395 m_freem(ext); 1396 ext = NULL; 1397 #endif 1398 } 1399 loopend: 1400 ; 1401 } 1402 #undef IS2292 1403 } 1404 1405 void 1406 ip6_notify_pmtu(struct inpcb *in6p, struct sockaddr_in6 *dst, u_int32_t *mtu) 1407 { 1408 struct socket *so; 1409 struct mbuf *m_mtu; 1410 struct ip6_mtuinfo mtuctl; 1411 1412 so = in6p->inp_socket; 1413 1414 if (mtu == NULL) 1415 return; 1416 1417 #ifdef DIAGNOSTIC 1418 if (so == NULL) /* I believe this is impossible */ 1419 panic("ip6_notify_pmtu: socket is NULL"); 1420 #endif 1421 1422 bzero(&mtuctl, sizeof(mtuctl)); /* zero-clear for safety */ 1423 mtuctl.ip6m_mtu = *mtu; 1424 mtuctl.ip6m_addr = *dst; 1425 1426 if ((m_mtu = sbcreatecontrol((caddr_t)&mtuctl, sizeof(mtuctl), 1427 IPV6_PATHMTU, IPPROTO_IPV6)) == NULL) 1428 return; 1429 1430 lwkt_gettoken(&so->so_rcv.ssb_token); 1431 if (sbappendaddr(&so->so_rcv.sb, (struct sockaddr *)dst, NULL, m_mtu) 1432 == 0) { 1433 m_freem(m_mtu); 1434 /* XXX: should count statistics */ 1435 } else { 1436 sorwakeup(so); 1437 } 1438 lwkt_reltoken(&so->so_rcv.ssb_token); 1439 } 1440 1441 #ifdef PULLDOWN_TEST 1442 /* 1443 * pull single extension header from mbuf chain. returns single mbuf that 1444 * contains the result, or NULL on error. 1445 */ 1446 static struct mbuf * 1447 ip6_pullexthdr(struct mbuf *m, size_t off, int nxt) 1448 { 1449 struct ip6_ext ip6e; 1450 size_t elen; 1451 struct mbuf *n; 1452 1453 #ifdef DIAGNOSTIC 1454 switch (nxt) { 1455 case IPPROTO_DSTOPTS: 1456 case IPPROTO_ROUTING: 1457 case IPPROTO_HOPOPTS: 1458 case IPPROTO_AH: /* is it possible? */ 1459 break; 1460 default: 1461 kprintf("ip6_pullexthdr: invalid nxt=%d\n", nxt); 1462 } 1463 #endif 1464 1465 m_copydata(m, off, sizeof(ip6e), (caddr_t)&ip6e); 1466 if (nxt == IPPROTO_AH) 1467 elen = (ip6e.ip6e_len + 2) << 2; 1468 else 1469 elen = (ip6e.ip6e_len + 1) << 3; 1470 1471 n = m_getb(elen, MB_DONTWAIT, MT_DATA, 0); 1472 if (n == NULL) 1473 return NULL; 1474 n->m_len = 0; 1475 1476 if (elen >= M_TRAILINGSPACE(n)) { 1477 m_free(n); 1478 return NULL; 1479 } 1480 1481 m_copydata(m, off, elen, mtod(n, caddr_t)); 1482 n->m_len = elen; 1483 return n; 1484 } 1485 #endif 1486 1487 /* 1488 * Get pointer to the previous header followed by the header 1489 * currently processed. 1490 * XXX: This function supposes that 1491 * M includes all headers, 1492 * the next header field and the header length field of each header 1493 * are valid, and 1494 * the sum of each header length equals to OFF. 1495 * Because of these assumptions, this function must be called very 1496 * carefully. Moreover, it will not be used in the near future when 1497 * we develop `neater' mechanism to process extension headers. 1498 */ 1499 char * 1500 ip6_get_prevhdr(struct mbuf *m, int off) 1501 { 1502 struct ip6_hdr *ip6 = mtod(m, struct ip6_hdr *); 1503 1504 if (off == sizeof(struct ip6_hdr)) 1505 return (&ip6->ip6_nxt); 1506 else { 1507 int len, nxt; 1508 struct ip6_ext *ip6e = NULL; 1509 1510 nxt = ip6->ip6_nxt; 1511 len = sizeof(struct ip6_hdr); 1512 while (len < off) { 1513 ip6e = (struct ip6_ext *)(mtod(m, caddr_t) + len); 1514 1515 switch (nxt) { 1516 case IPPROTO_FRAGMENT: 1517 len += sizeof(struct ip6_frag); 1518 break; 1519 case IPPROTO_AH: 1520 len += (ip6e->ip6e_len + 2) << 2; 1521 break; 1522 default: 1523 len += (ip6e->ip6e_len + 1) << 3; 1524 break; 1525 } 1526 nxt = ip6e->ip6e_nxt; 1527 } 1528 if (ip6e) 1529 return (&ip6e->ip6e_nxt); 1530 else 1531 return NULL; 1532 } 1533 } 1534 1535 /* 1536 * get next header offset. m will be retained. 1537 */ 1538 int 1539 ip6_nexthdr(struct mbuf *m, int off, int proto, int *nxtp) 1540 { 1541 struct ip6_hdr ip6; 1542 struct ip6_ext ip6e; 1543 struct ip6_frag fh; 1544 1545 /* just in case */ 1546 if (m == NULL) 1547 panic("ip6_nexthdr: m == NULL"); 1548 if ((m->m_flags & M_PKTHDR) == 0 || m->m_pkthdr.len < off) 1549 return -1; 1550 1551 switch (proto) { 1552 case IPPROTO_IPV6: 1553 if (m->m_pkthdr.len < off + sizeof(ip6)) 1554 return -1; 1555 m_copydata(m, off, sizeof(ip6), (caddr_t)&ip6); 1556 if (nxtp) 1557 *nxtp = ip6.ip6_nxt; 1558 off += sizeof(ip6); 1559 return off; 1560 1561 case IPPROTO_FRAGMENT: 1562 /* 1563 * terminate parsing if it is not the first fragment, 1564 * it does not make sense to parse through it. 1565 */ 1566 if (m->m_pkthdr.len < off + sizeof(fh)) 1567 return -1; 1568 m_copydata(m, off, sizeof(fh), (caddr_t)&fh); 1569 /* IP6F_OFF_MASK = 0xfff8(BigEndian), 0xf8ff(LittleEndian) */ 1570 if (fh.ip6f_offlg & IP6F_OFF_MASK) 1571 return -1; 1572 if (nxtp) 1573 *nxtp = fh.ip6f_nxt; 1574 off += sizeof(struct ip6_frag); 1575 return off; 1576 1577 case IPPROTO_AH: 1578 if (m->m_pkthdr.len < off + sizeof(ip6e)) 1579 return -1; 1580 m_copydata(m, off, sizeof(ip6e), (caddr_t)&ip6e); 1581 if (nxtp) 1582 *nxtp = ip6e.ip6e_nxt; 1583 off += (ip6e.ip6e_len + 2) << 2; 1584 return off; 1585 1586 case IPPROTO_HOPOPTS: 1587 case IPPROTO_ROUTING: 1588 case IPPROTO_DSTOPTS: 1589 if (m->m_pkthdr.len < off + sizeof(ip6e)) 1590 return -1; 1591 m_copydata(m, off, sizeof(ip6e), (caddr_t)&ip6e); 1592 if (nxtp) 1593 *nxtp = ip6e.ip6e_nxt; 1594 off += (ip6e.ip6e_len + 1) << 3; 1595 return off; 1596 1597 case IPPROTO_NONE: 1598 case IPPROTO_ESP: 1599 case IPPROTO_IPCOMP: 1600 /* give up */ 1601 return -1; 1602 1603 default: 1604 return -1; 1605 } 1606 1607 return -1; 1608 } 1609 1610 /* 1611 * get offset for the last header in the chain. m will be kept untainted. 1612 */ 1613 int 1614 ip6_lasthdr(struct mbuf *m, int off, int proto, int *nxtp) 1615 { 1616 int newoff; 1617 int nxt; 1618 1619 if (!nxtp) { 1620 nxt = -1; 1621 nxtp = &nxt; 1622 } 1623 while (1) { 1624 newoff = ip6_nexthdr(m, off, proto, nxtp); 1625 if (newoff < 0) 1626 return off; 1627 else if (newoff < off) 1628 return -1; /* invalid */ 1629 else if (newoff == off) 1630 return newoff; 1631 1632 off = newoff; 1633 proto = *nxtp; 1634 } 1635 } 1636 1637 struct ip6aux * 1638 ip6_addaux(struct mbuf *m) 1639 { 1640 struct m_tag *tag = m_tag_find(m, PACKET_TAG_IPV6_INPUT, NULL); 1641 if (!tag) { 1642 tag = m_tag_get(PACKET_TAG_IPV6_INPUT, 1643 sizeof (struct ip6aux), 1644 MB_DONTWAIT); 1645 if (tag) 1646 m_tag_prepend(m, tag); 1647 } 1648 if (tag) 1649 bzero(m_tag_data(tag), sizeof (struct ip6aux)); 1650 return tag ? (struct ip6aux *)m_tag_data(tag) : NULL; 1651 } 1652 1653 struct ip6aux * 1654 ip6_findaux(struct mbuf *m) 1655 { 1656 struct m_tag *tag = m_tag_find(m, PACKET_TAG_IPV6_INPUT, NULL); 1657 return tag ? (struct ip6aux*)(tag+1) : NULL; 1658 } 1659 1660 void 1661 ip6_delaux(struct mbuf *m) 1662 { 1663 struct m_tag *tag = m_tag_find(m, PACKET_TAG_IPV6_INPUT, NULL); 1664 if (tag) 1665 m_tag_delete(m, tag); 1666 } 1667 1668 /* 1669 * System control for IP6 1670 */ 1671 1672 u_char inet6ctlerrmap[PRC_NCMDS] = { 1673 0, 0, 0, 0, 1674 0, EMSGSIZE, EHOSTDOWN, EHOSTUNREACH, 1675 EHOSTUNREACH, EHOSTUNREACH, ECONNREFUSED, ECONNREFUSED, 1676 EMSGSIZE, EHOSTUNREACH, 0, 0, 1677 0, 0, 0, 0, 1678 ENOPROTOOPT 1679 }; 1680