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 #include <netinet6/in6_prefix.h> 111 112 #ifdef IPSEC 113 #include <netinet6/ipsec.h> 114 #ifdef INET6 115 #include <netinet6/ipsec6.h> 116 #endif 117 #endif 118 119 #ifdef FAST_IPSEC 120 #include <netproto/ipsec/ipsec.h> 121 #include <netproto/ipsec/ipsec6.h> 122 #define IPSEC 123 #endif /* FAST_IPSEC */ 124 125 #include <net/ip6fw/ip6_fw.h> 126 127 #include <netinet6/ip6protosw.h> 128 129 #include <net/net_osdep.h> 130 131 extern struct domain inet6domain; 132 extern struct protosw inet6sw[]; 133 134 u_char ip6_protox[IPPROTO_MAX]; 135 struct in6_ifaddr *in6_ifaddr; 136 137 extern struct callout in6_tmpaddrtimer_ch; 138 139 int ip6_forward_srcrt; /* XXX */ 140 int ip6_sourcecheck; /* XXX */ 141 int ip6_sourcecheck_interval; /* XXX */ 142 143 int ip6_ours_check_algorithm; 144 145 struct pfil_head inet6_pfil_hook; 146 147 /* firewall hooks */ 148 ip6_fw_chk_t *ip6_fw_chk_ptr; 149 ip6_fw_ctl_t *ip6_fw_ctl_ptr; 150 int ip6_fw_enable = 1; 151 152 struct ip6stat ip6stat; 153 154 static void ip6_init2 (void *); 155 static struct ip6aux *ip6_setdstifaddr (struct mbuf *, struct in6_ifaddr *); 156 static int ip6_hopopts_input (u_int32_t *, u_int32_t *, struct mbuf **, int *); 157 static void ip6_input(netmsg_t msg); 158 #ifdef PULLDOWN_TEST 159 static struct mbuf *ip6_pullexthdr (struct mbuf *, size_t, int); 160 #endif 161 static void transport6_processing_handler(netmsg_t netmsg); 162 163 /* 164 * IP6 initialization: fill in IP6 protocol switch table. 165 * All protocols not implemented in kernel go to raw IP6 protocol handler. 166 */ 167 void 168 ip6_init(void) 169 { 170 struct protosw *pr; 171 int i; 172 struct timeval tv; 173 174 pr = pffindproto(PF_INET6, IPPROTO_RAW, SOCK_RAW); 175 if (pr == NULL) 176 panic("ip6_init"); 177 for (i = 0; i < IPPROTO_MAX; i++) 178 ip6_protox[i] = pr - inet6sw; 179 for (pr = inet6domain.dom_protosw; 180 pr < inet6domain.dom_protoswNPROTOSW; pr++) 181 if (pr->pr_domain->dom_family == PF_INET6 && 182 pr->pr_protocol && pr->pr_protocol != IPPROTO_RAW) 183 ip6_protox[pr->pr_protocol] = pr - inet6sw; 184 185 inet6_pfil_hook.ph_type = PFIL_TYPE_AF; 186 inet6_pfil_hook.ph_af = AF_INET6; 187 if ((i = pfil_head_register(&inet6_pfil_hook)) != 0) { 188 kprintf("%s: WARNING: unable to register pfil hook, " 189 "error %d\n", __func__, i); 190 } 191 192 netisr_register(NETISR_IPV6, ip6_input, NULL); /* XXX hashfn */ 193 scope6_init(); 194 addrsel_policy_init(); 195 nd6_init(); 196 frag6_init(); 197 /* 198 * in many cases, random() here does NOT return random number 199 * as initialization during bootstrap time occur in fixed order. 200 */ 201 microtime(&tv); 202 ip6_flow_seq = krandom() ^ tv.tv_usec; 203 microtime(&tv); 204 ip6_desync_factor = (krandom() ^ tv.tv_usec) % MAX_TEMP_DESYNC_FACTOR; 205 } 206 207 static void 208 ip6_init2(void *dummy) 209 { 210 /* nd6_timer_init */ 211 callout_init(&nd6_timer_ch); 212 callout_reset(&nd6_timer_ch, hz, nd6_timer, NULL); 213 214 /* router renumbering prefix list maintenance */ 215 callout_init(&in6_rr_timer_ch); 216 callout_reset(&in6_rr_timer_ch, hz, in6_rr_timer, NULL); 217 218 /* timer for regeneranation of temporary addresses randomize ID */ 219 callout_init(&in6_tmpaddrtimer_ch); 220 callout_reset(&in6_tmpaddrtimer_ch, 221 (ip6_temp_preferred_lifetime - ip6_desync_factor - 222 ip6_temp_regen_advance) * hz, 223 in6_tmpaddrtimer, NULL); 224 } 225 226 /* cheat */ 227 /* This must be after route_init(), which is now SI_ORDER_THIRD */ 228 SYSINIT(netinet6init2, SI_SUB_PROTO_DOMAIN, SI_ORDER_MIDDLE, ip6_init2, NULL); 229 230 extern struct route_in6 ip6_forward_rt; 231 232 static 233 void 234 ip6_input(netmsg_t msg) 235 { 236 struct mbuf *m = msg->packet.nm_packet; 237 struct ip6_hdr *ip6; 238 int off = sizeof(struct ip6_hdr), nest; 239 u_int32_t plen; 240 u_int32_t rtalert = ~0; 241 int nxt, ours = 0, rh_present = 0; 242 struct ifnet *deliverifp = NULL; 243 struct in6_addr odst; 244 int srcrt = 0; 245 246 #ifdef IPSEC 247 /* 248 * should the inner packet be considered authentic? 249 * see comment in ah4_input(). 250 */ 251 if (m) { 252 m->m_flags &= ~M_AUTHIPHDR; 253 m->m_flags &= ~M_AUTHIPDGM; 254 } 255 #endif 256 257 /* 258 * make sure we don't have onion peering information into m_aux. 259 */ 260 ip6_delaux(m); 261 262 /* 263 * mbuf statistics 264 */ 265 if (m->m_flags & M_EXT) { 266 if (m->m_next) 267 ip6stat.ip6s_mext2m++; 268 else 269 ip6stat.ip6s_mext1++; 270 } else { 271 #define M2MMAX NELEM(ip6stat.ip6s_m2m) 272 if (m->m_next) { 273 if (m->m_flags & M_LOOP) { 274 ip6stat.ip6s_m2m[loif[0].if_index]++; /* XXX */ 275 } else if (m->m_pkthdr.rcvif->if_index < M2MMAX) 276 ip6stat.ip6s_m2m[m->m_pkthdr.rcvif->if_index]++; 277 else 278 ip6stat.ip6s_m2m[0]++; 279 } else 280 ip6stat.ip6s_m1++; 281 #undef M2MMAX 282 } 283 284 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_receive); 285 ip6stat.ip6s_total++; 286 287 #ifndef PULLDOWN_TEST 288 /* 289 * L2 bridge code and some other code can return mbuf chain 290 * that does not conform to KAME requirement. too bad. 291 * XXX: fails to join if interface MTU > MCLBYTES. jumbogram? 292 */ 293 if (m && m->m_next != NULL && m->m_pkthdr.len < MCLBYTES) { 294 struct mbuf *n; 295 296 n = m_getb(m->m_pkthdr.len, MB_DONTWAIT, MT_HEADER, M_PKTHDR); 297 if (n == NULL) 298 goto bad; 299 M_MOVE_PKTHDR(n, m); 300 301 m_copydata(m, 0, n->m_pkthdr.len, mtod(n, caddr_t)); 302 n->m_len = n->m_pkthdr.len; 303 m_freem(m); 304 m = n; 305 } 306 IP6_EXTHDR_CHECK_VOIDRET(m, 0, sizeof(struct ip6_hdr)); 307 #endif 308 309 if (m->m_len < sizeof(struct ip6_hdr)) { 310 struct ifnet *inifp; 311 inifp = m->m_pkthdr.rcvif; 312 if ((m = m_pullup(m, sizeof(struct ip6_hdr))) == NULL) { 313 ip6stat.ip6s_toosmall++; 314 in6_ifstat_inc(inifp, ifs6_in_hdrerr); 315 goto bad2; 316 } 317 } 318 319 ip6 = mtod(m, struct ip6_hdr *); 320 321 if ((ip6->ip6_vfc & IPV6_VERSION_MASK) != IPV6_VERSION) { 322 ip6stat.ip6s_badvers++; 323 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_hdrerr); 324 goto bad; 325 } 326 327 /* 328 * Run through list of hooks for input packets. 329 * 330 * NB: Beware of the destination address changing 331 * (e.g. by NAT rewriting). When this happens, 332 * tell ip6_forward to do the right thing. 333 */ 334 if (pfil_has_hooks(&inet6_pfil_hook)) { 335 odst = ip6->ip6_dst; 336 if (pfil_run_hooks(&inet6_pfil_hook, &m, 337 m->m_pkthdr.rcvif, PFIL_IN)) { 338 goto bad2; 339 } 340 if (m == NULL) /* consumed by filter */ 341 goto bad2; 342 ip6 = mtod(m, struct ip6_hdr *); 343 srcrt = !IN6_ARE_ADDR_EQUAL(&odst, &ip6->ip6_dst); 344 } 345 346 ip6stat.ip6s_nxthist[ip6->ip6_nxt]++; 347 348 #ifdef ALTQ 349 if (altq_input != NULL && (*altq_input)(m, AF_INET6) == 0) { 350 /* packet is dropped by traffic conditioner */ 351 return; 352 } 353 #endif 354 355 /* 356 * Check with the firewall... 357 */ 358 if (ip6_fw_enable && ip6_fw_chk_ptr) { 359 u_short port = 0; 360 /* If ipfw says divert, we have to just drop packet */ 361 /* use port as a dummy argument */ 362 if ((*ip6_fw_chk_ptr)(&ip6, NULL, &port, &m)) { 363 m_freem(m); 364 m = NULL; 365 } 366 if (!m) 367 goto bad2; 368 } 369 370 /* 371 * Check against address spoofing/corruption. 372 */ 373 if (IN6_IS_ADDR_MULTICAST(&ip6->ip6_src) || 374 IN6_IS_ADDR_UNSPECIFIED(&ip6->ip6_dst)) { 375 /* 376 * XXX: "badscope" is not very suitable for a multicast source. 377 */ 378 ip6stat.ip6s_badscope++; 379 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_addrerr); 380 goto bad; 381 } 382 if ((IN6_IS_ADDR_LOOPBACK(&ip6->ip6_src) || 383 IN6_IS_ADDR_LOOPBACK(&ip6->ip6_dst)) && 384 (m->m_pkthdr.rcvif->if_flags & IFF_LOOPBACK) == 0) { 385 ip6stat.ip6s_badscope++; 386 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_addrerr); 387 goto bad; 388 } 389 390 /* 391 * The following check is not documented in specs. A malicious 392 * party may be able to use IPv4 mapped addr to confuse tcp/udp stack 393 * and bypass security checks (act as if it was from 127.0.0.1 by using 394 * IPv6 src ::ffff:127.0.0.1). Be cautious. 395 * 396 * This check chokes if we are in an SIIT cloud. As none of BSDs 397 * support IPv4-less kernel compilation, we cannot support SIIT 398 * environment at all. So, it makes more sense for us to reject any 399 * malicious packets for non-SIIT environment, than try to do a 400 * partical support for SIIT environment. 401 */ 402 if (IN6_IS_ADDR_V4MAPPED(&ip6->ip6_src) || 403 IN6_IS_ADDR_V4MAPPED(&ip6->ip6_dst)) { 404 ip6stat.ip6s_badscope++; 405 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_addrerr); 406 goto bad; 407 } 408 #if 0 409 /* 410 * Reject packets with IPv4 compatible addresses (auto tunnel). 411 * 412 * The code forbids auto tunnel relay case in RFC1933 (the check is 413 * stronger than RFC1933). We may want to re-enable it if mech-xx 414 * is revised to forbid relaying case. 415 */ 416 if (IN6_IS_ADDR_V4COMPAT(&ip6->ip6_src) || 417 IN6_IS_ADDR_V4COMPAT(&ip6->ip6_dst)) { 418 ip6stat.ip6s_badscope++; 419 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_addrerr); 420 goto bad; 421 } 422 #endif 423 424 /* drop packets if interface ID portion is already filled */ 425 if ((m->m_pkthdr.rcvif->if_flags & IFF_LOOPBACK) == 0) { 426 if (IN6_IS_SCOPE_LINKLOCAL(&ip6->ip6_src) && 427 ip6->ip6_src.s6_addr16[1]) { 428 ip6stat.ip6s_badscope++; 429 goto bad; 430 } 431 if (IN6_IS_SCOPE_LINKLOCAL(&ip6->ip6_dst) && 432 ip6->ip6_dst.s6_addr16[1]) { 433 ip6stat.ip6s_badscope++; 434 goto bad; 435 } 436 } 437 438 if (IN6_IS_SCOPE_LINKLOCAL(&ip6->ip6_src)) 439 ip6->ip6_src.s6_addr16[1] 440 = htons(m->m_pkthdr.rcvif->if_index); 441 if (IN6_IS_SCOPE_LINKLOCAL(&ip6->ip6_dst)) 442 ip6->ip6_dst.s6_addr16[1] 443 = htons(m->m_pkthdr.rcvif->if_index); 444 445 #if 0 /* this case seems to be unnecessary. (jinmei, 20010401) */ 446 /* 447 * We use rt->rt_ifp to determine if the address is ours or not. 448 * If rt_ifp is lo0, the address is ours. 449 * The problem here is, rt->rt_ifp for fe80::%lo0/64 is set to lo0, 450 * so any address under fe80::%lo0/64 will be mistakenly considered 451 * local. The special case is supplied to handle the case properly 452 * by actually looking at interface addresses 453 * (using in6ifa_ifpwithaddr). 454 */ 455 if ((m->m_pkthdr.rcvif->if_flags & IFF_LOOPBACK) && 456 IN6_IS_ADDR_LINKLOCAL(&ip6->ip6_dst)) { 457 if (!in6ifa_ifpwithaddr(m->m_pkthdr.rcvif, &ip6->ip6_dst)) { 458 icmp6_error(m, ICMP6_DST_UNREACH, 459 ICMP6_DST_UNREACH_ADDR, 0); 460 /* m is already freed */ 461 goto bad2; 462 } 463 464 ours = 1; 465 deliverifp = m->m_pkthdr.rcvif; 466 goto hbhcheck; 467 } 468 #endif 469 470 /* 471 * Multicast check 472 */ 473 if (IN6_IS_ADDR_MULTICAST(&ip6->ip6_dst)) { 474 struct in6_multi *in6m = NULL; 475 476 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_mcast); 477 /* 478 * See if we belong to the destination multicast group on the 479 * arrival interface. 480 */ 481 IN6_LOOKUP_MULTI(ip6->ip6_dst, m->m_pkthdr.rcvif, in6m); 482 if (in6m) 483 ours = 1; 484 else if (!ip6_mrouter) { 485 ip6stat.ip6s_notmember++; 486 ip6stat.ip6s_cantforward++; 487 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_discard); 488 goto bad; 489 } 490 deliverifp = m->m_pkthdr.rcvif; 491 goto hbhcheck; 492 } 493 494 /* 495 * Unicast check 496 */ 497 switch (ip6_ours_check_algorithm) { 498 default: 499 /* 500 * XXX: I intentionally broke our indentation rule here, 501 * since this switch-case is just for measurement and 502 * therefore should soon be removed. 503 */ 504 if (ip6_forward_rt.ro_rt != NULL && 505 (ip6_forward_rt.ro_rt->rt_flags & RTF_UP) && 506 IN6_ARE_ADDR_EQUAL(&ip6->ip6_dst, 507 &((struct sockaddr_in6 *)(&ip6_forward_rt.ro_dst))->sin6_addr)) 508 ip6stat.ip6s_forward_cachehit++; 509 else { 510 struct sockaddr_in6 *dst6; 511 512 if (ip6_forward_rt.ro_rt) { 513 /* route is down or destination is different */ 514 ip6stat.ip6s_forward_cachemiss++; 515 RTFREE(ip6_forward_rt.ro_rt); 516 ip6_forward_rt.ro_rt = 0; 517 } 518 519 bzero(&ip6_forward_rt.ro_dst, sizeof(struct sockaddr_in6)); 520 dst6 = &ip6_forward_rt.ro_dst; 521 dst6->sin6_len = sizeof(struct sockaddr_in6); 522 dst6->sin6_family = AF_INET6; 523 dst6->sin6_addr = ip6->ip6_dst; 524 525 rtalloc_ign((struct route *)&ip6_forward_rt, RTF_PRCLONING); 526 } 527 528 #define rt6_key(r) ((struct sockaddr_in6 *)((r)->rt_nodes->rn_key)) 529 530 /* 531 * Accept the packet if the forwarding interface to the destination 532 * according to the routing table is the loopback interface, 533 * unless the associated route has a gateway. 534 * Note that this approach causes to accept a packet if there is a 535 * route to the loopback interface for the destination of the packet. 536 * But we think it's even useful in some situations, e.g. when using 537 * a special daemon which wants to intercept the packet. 538 * 539 * XXX: some OSes automatically make a cloned route for the destination 540 * of an outgoing packet. If the outgoing interface of the packet 541 * is a loopback one, the kernel would consider the packet to be 542 * accepted, even if we have no such address assinged on the interface. 543 * We check the cloned flag of the route entry to reject such cases, 544 * assuming that route entries for our own addresses are not made by 545 * cloning (it should be true because in6_addloop explicitly installs 546 * the host route). However, we might have to do an explicit check 547 * while it would be less efficient. Or, should we rather install a 548 * reject route for such a case? 549 */ 550 if (ip6_forward_rt.ro_rt && 551 (ip6_forward_rt.ro_rt->rt_flags & 552 (RTF_HOST|RTF_GATEWAY)) == RTF_HOST && 553 #ifdef RTF_WASCLONED 554 !(ip6_forward_rt.ro_rt->rt_flags & RTF_WASCLONED) && 555 #endif 556 #ifdef RTF_CLONED 557 !(ip6_forward_rt.ro_rt->rt_flags & RTF_CLONED) && 558 #endif 559 #if 0 560 /* 561 * The check below is redundant since the comparison of 562 * the destination and the key of the rtentry has 563 * already done through looking up the routing table. 564 */ 565 IN6_ARE_ADDR_EQUAL(&ip6->ip6_dst, 566 &rt6_key(ip6_forward_rt.ro_rt)->sin6_addr) 567 #endif 568 ip6_forward_rt.ro_rt->rt_ifp->if_type == IFT_LOOP) { 569 struct in6_ifaddr *ia6 = 570 (struct in6_ifaddr *)ip6_forward_rt.ro_rt->rt_ifa; 571 572 /* 573 * record address information into m_aux. 574 */ 575 ip6_setdstifaddr(m, ia6); 576 577 /* 578 * packets to a tentative, duplicated, or somehow invalid 579 * address must not be accepted. 580 */ 581 if (!(ia6->ia6_flags & IN6_IFF_NOTREADY)) { 582 /* this address is ready */ 583 ours = 1; 584 deliverifp = ia6->ia_ifp; /* correct? */ 585 /* Count the packet in the ip address stats */ 586 IFA_STAT_INC(&ia6->ia_ifa, ipackets, 1); 587 IFA_STAT_INC(&ia6->ia_ifa, ibytes, m->m_pkthdr.len); 588 goto hbhcheck; 589 } else { 590 /* address is not ready, so discard the packet. */ 591 nd6log((LOG_INFO, 592 "ip6_input: packet to an unready address %s->%s\n", 593 ip6_sprintf(&ip6->ip6_src), 594 ip6_sprintf(&ip6->ip6_dst))); 595 596 goto bad; 597 } 598 } 599 } /* XXX indentation (see above) */ 600 601 /* 602 * FAITH(Firewall Aided Internet Translator) 603 */ 604 if (ip6_keepfaith) { 605 if (ip6_forward_rt.ro_rt && ip6_forward_rt.ro_rt->rt_ifp 606 && ip6_forward_rt.ro_rt->rt_ifp->if_type == IFT_FAITH) { 607 /* XXX do we need more sanity checks? */ 608 ours = 1; 609 deliverifp = ip6_forward_rt.ro_rt->rt_ifp; /* faith */ 610 goto hbhcheck; 611 } 612 } 613 614 /* 615 * Now there is no reason to process the packet if it's not our own 616 * and we're not a router. 617 */ 618 if (!ip6_forwarding) { 619 ip6stat.ip6s_cantforward++; 620 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_discard); 621 goto bad; 622 } 623 624 hbhcheck: 625 /* 626 * record address information into m_aux, if we don't have one yet. 627 * note that we are unable to record it, if the address is not listed 628 * as our interface address (e.g. multicast addresses, addresses 629 * within FAITH prefixes and such). 630 */ 631 if (deliverifp && !ip6_getdstifaddr(m)) { 632 struct in6_ifaddr *ia6; 633 634 ia6 = in6_ifawithifp(deliverifp, &ip6->ip6_dst); 635 if (ia6) { 636 if (!ip6_setdstifaddr(m, ia6)) { 637 /* 638 * XXX maybe we should drop the packet here, 639 * as we could not provide enough information 640 * to the upper layers. 641 */ 642 } 643 } 644 } 645 646 /* 647 * Process Hop-by-Hop options header if it's contained. 648 * m may be modified in ip6_hopopts_input(). 649 * If a JumboPayload option is included, plen will also be modified. 650 */ 651 plen = (u_int32_t)ntohs(ip6->ip6_plen); 652 if (ip6->ip6_nxt == IPPROTO_HOPOPTS) { 653 struct ip6_hbh *hbh; 654 655 if (ip6_hopopts_input(&plen, &rtalert, &m, &off)) { 656 #if 0 /*touches NULL pointer*/ 657 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_discard); 658 #endif 659 goto bad2; /* m have already been freed */ 660 } 661 662 /* adjust pointer */ 663 ip6 = mtod(m, struct ip6_hdr *); 664 665 /* 666 * if the payload length field is 0 and the next header field 667 * indicates Hop-by-Hop Options header, then a Jumbo Payload 668 * option MUST be included. 669 */ 670 if (ip6->ip6_plen == 0 && plen == 0) { 671 /* 672 * Note that if a valid jumbo payload option is 673 * contained, ip6_hoptops_input() must set a valid 674 * (non-zero) payload length to the variable plen. 675 */ 676 ip6stat.ip6s_badoptions++; 677 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_discard); 678 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_hdrerr); 679 icmp6_error(m, ICMP6_PARAM_PROB, 680 ICMP6_PARAMPROB_HEADER, 681 (caddr_t)&ip6->ip6_plen - (caddr_t)ip6); 682 goto bad2; 683 } 684 #ifndef PULLDOWN_TEST 685 /* ip6_hopopts_input() ensures that mbuf is contiguous */ 686 hbh = (struct ip6_hbh *)(ip6 + 1); 687 #else 688 IP6_EXTHDR_GET(hbh, struct ip6_hbh *, m, sizeof(struct ip6_hdr), 689 sizeof(struct ip6_hbh)); 690 if (hbh == NULL) { 691 ip6stat.ip6s_tooshort++; 692 goto bad2; 693 } 694 #endif 695 nxt = hbh->ip6h_nxt; 696 697 /* 698 * If we are acting as a router and the packet contains a 699 * router alert option, see if we know the option value. 700 * Currently, we only support the option value for MLD, in which 701 * case we should pass the packet to the multicast routing 702 * daemon. 703 */ 704 if (rtalert != ~0 && ip6_forwarding) { 705 switch (rtalert) { 706 case IP6OPT_RTALERT_MLD: 707 ours = 1; 708 break; 709 default: 710 /* 711 * RFC2711 requires unrecognized values must be 712 * silently ignored. 713 */ 714 break; 715 } 716 } 717 } else 718 nxt = ip6->ip6_nxt; 719 720 /* 721 * Check that the amount of data in the buffers 722 * is as at least much as the IPv6 header would have us expect. 723 * Trim mbufs if longer than we expect. 724 * Drop packet if shorter than we expect. 725 */ 726 if (m->m_pkthdr.len - sizeof(struct ip6_hdr) < plen) { 727 ip6stat.ip6s_tooshort++; 728 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_truncated); 729 goto bad; 730 } 731 if (m->m_pkthdr.len > sizeof(struct ip6_hdr) + plen) { 732 if (m->m_len == m->m_pkthdr.len) { 733 m->m_len = sizeof(struct ip6_hdr) + plen; 734 m->m_pkthdr.len = sizeof(struct ip6_hdr) + plen; 735 } else 736 m_adj(m, sizeof(struct ip6_hdr) + plen - m->m_pkthdr.len); 737 } 738 739 /* 740 * Forward if desirable. 741 */ 742 if (IN6_IS_ADDR_MULTICAST(&ip6->ip6_dst)) { 743 /* 744 * If we are acting as a multicast router, all 745 * incoming multicast packets are passed to the 746 * kernel-level multicast forwarding function. 747 * The packet is returned (relatively) intact; if 748 * ip6_mforward() returns a non-zero value, the packet 749 * must be discarded, else it may be accepted below. 750 */ 751 if (ip6_mrouter && ip6_mforward(ip6, m->m_pkthdr.rcvif, m)) { 752 ip6stat.ip6s_cantforward++; 753 goto bad; 754 } 755 if (!ours) 756 goto bad; 757 } else if (!ours) { 758 ip6_forward(m, srcrt); 759 goto bad2; 760 } 761 762 ip6 = mtod(m, struct ip6_hdr *); 763 764 /* 765 * Malicious party may be able to use IPv4 mapped addr to confuse 766 * tcp/udp stack and bypass security checks (act as if it was from 767 * 127.0.0.1 by using IPv6 src ::ffff:127.0.0.1). Be cautious. 768 * 769 * For SIIT end node behavior, you may want to disable the check. 770 * However, you will become vulnerable to attacks using IPv4 mapped 771 * source. 772 */ 773 if (IN6_IS_ADDR_V4MAPPED(&ip6->ip6_src) || 774 IN6_IS_ADDR_V4MAPPED(&ip6->ip6_dst)) { 775 ip6stat.ip6s_badscope++; 776 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_addrerr); 777 goto bad; 778 } 779 780 /* 781 * Tell launch routine the next header 782 */ 783 ip6stat.ip6s_delivered++; 784 in6_ifstat_inc(deliverifp, ifs6_in_deliver); 785 nest = 0; 786 787 rh_present = 0; 788 while (nxt != IPPROTO_DONE) { 789 struct protosw *sw6; 790 791 if (ip6_hdrnestlimit && (++nest > ip6_hdrnestlimit)) { 792 ip6stat.ip6s_toomanyhdr++; 793 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_hdrerr); 794 goto bad; 795 } 796 797 /* 798 * protection against faulty packet - there should be 799 * more sanity checks in header chain processing. 800 */ 801 if (m->m_pkthdr.len < off) { 802 ip6stat.ip6s_tooshort++; 803 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_truncated); 804 goto bad; 805 } 806 807 #if 0 808 /* 809 * do we need to do it for every header? yeah, other 810 * functions can play with it (like re-allocate and copy). 811 */ 812 mhist = ip6_addaux(m); 813 if (mhist && M_TRAILINGSPACE(mhist) >= sizeof(nxt)) { 814 hist = mtod(mhist, caddr_t) + mhist->m_len; 815 bcopy(&nxt, hist, sizeof(nxt)); 816 mhist->m_len += sizeof(nxt); 817 } else { 818 ip6stat.ip6s_toomanyhdr++; 819 goto bad; 820 } 821 #endif 822 823 if (nxt == IPPROTO_ROUTING) { 824 if (rh_present++) { 825 in6_ifstat_inc(m->m_pkthdr.rcvif, 826 ifs6_in_hdrerr); 827 ip6stat.ip6s_badoptions++; 828 goto bad; 829 } 830 } 831 832 sw6 = &inet6sw[ip6_protox[nxt]]; 833 #ifdef IPSEC 834 /* 835 * enforce IPsec policy checking if we are seeing last header. 836 * note that we do not visit this with protocols with pcb layer 837 * code - like udp/tcp/raw ip. 838 */ 839 if ((sw6->pr_flags & PR_LASTHDR) && ipsec6_in_reject(m, NULL)) { 840 ipsec6stat.in_polvio++; 841 goto bad; 842 } 843 #endif 844 /* 845 * If this is a terminal header forward to the port, otherwise 846 * process synchronously for more headers. 847 */ 848 if (sw6->pr_flags & PR_LASTHDR) { 849 struct netmsg_packet *pmsg; 850 lwkt_port_t port; 851 852 port = netisr_cpuport(0); /* XXX */ 853 KKASSERT(port != NULL); 854 pmsg = &m->m_hdr.mh_netmsg; 855 netmsg_init(&pmsg->base, NULL, 856 &netisr_apanic_rport, 857 0, transport6_processing_handler); 858 pmsg->nm_packet = m; 859 pmsg->nm_nxt = nxt; 860 pmsg->base.lmsg.u.ms_result = off; 861 lwkt_sendmsg(port, &pmsg->base.lmsg); 862 /* done with m */ 863 nxt = IPPROTO_DONE; 864 } else { 865 nxt = sw6->pr_input(&m, &off, nxt); 866 } 867 } 868 goto bad2; 869 bad: 870 m_freem(m); 871 bad2: 872 ; 873 /* msg was embedded in the mbuf, do not reply! */ 874 } 875 876 /* 877 * We have to call the pr_input() function from the correct protocol 878 * thread. The sw6->pr_soport() request at the end of ip6_input() 879 * returns the port and we forward a netmsg to the port to execute 880 * this function. 881 */ 882 static void 883 transport6_processing_handler(netmsg_t netmsg) 884 { 885 struct netmsg_packet *pmsg = (struct netmsg_packet *)netmsg; 886 struct protosw *sw6; 887 int hlen; 888 int nxt; 889 890 sw6 = &inet6sw[ip6_protox[pmsg->nm_nxt]]; 891 hlen = pmsg->base.lmsg.u.ms_result; 892 893 nxt = sw6->pr_input(&pmsg->nm_packet, &hlen, pmsg->nm_nxt); 894 KKASSERT(nxt == IPPROTO_DONE); 895 /* netmsg was embedded in the mbuf, do not reply! */ 896 } 897 898 /* 899 * set/grab in6_ifaddr correspond to IPv6 destination address. 900 * XXX backward compatibility wrapper 901 */ 902 static struct ip6aux * 903 ip6_setdstifaddr(struct mbuf *m, struct in6_ifaddr *ia6) 904 { 905 struct ip6aux *n; 906 907 n = ip6_addaux(m); 908 if (n) 909 n->ip6a_dstia6 = ia6; 910 return n; /* NULL if failed to set */ 911 } 912 913 struct in6_ifaddr * 914 ip6_getdstifaddr(struct mbuf *m) 915 { 916 struct ip6aux *n; 917 918 n = ip6_findaux(m); 919 if (n) 920 return n->ip6a_dstia6; 921 else 922 return NULL; 923 } 924 925 /* 926 * Hop-by-Hop options header processing. If a valid jumbo payload option is 927 * included, the real payload length will be stored in plenp. 928 */ 929 static int 930 ip6_hopopts_input(u_int32_t *plenp, 931 u_int32_t *rtalertp,/* XXX: should be stored more smart way */ 932 struct mbuf **mp, 933 int *offp) 934 { 935 struct mbuf *m = *mp; 936 int off = *offp, hbhlen; 937 struct ip6_hbh *hbh; 938 939 /* validation of the length of the header */ 940 #ifndef PULLDOWN_TEST 941 IP6_EXTHDR_CHECK(m, off, sizeof(*hbh), -1); 942 hbh = (struct ip6_hbh *)(mtod(m, caddr_t) + off); 943 hbhlen = (hbh->ip6h_len + 1) << 3; 944 945 IP6_EXTHDR_CHECK(m, off, hbhlen, -1); 946 hbh = (struct ip6_hbh *)(mtod(m, caddr_t) + off); 947 #else 948 IP6_EXTHDR_GET(hbh, struct ip6_hbh *, m, 949 sizeof(struct ip6_hdr), sizeof(struct ip6_hbh)); 950 if (hbh == NULL) { 951 ip6stat.ip6s_tooshort++; 952 return -1; 953 } 954 hbhlen = (hbh->ip6h_len + 1) << 3; 955 IP6_EXTHDR_GET(hbh, struct ip6_hbh *, m, sizeof(struct ip6_hdr), 956 hbhlen); 957 if (hbh == NULL) { 958 ip6stat.ip6s_tooshort++; 959 return -1; 960 } 961 #endif 962 off += hbhlen; 963 hbhlen -= sizeof(struct ip6_hbh); 964 965 if (ip6_process_hopopts(m, (u_int8_t *)hbh + sizeof(struct ip6_hbh), 966 hbhlen, rtalertp, plenp) < 0) 967 return (-1); 968 969 *offp = off; 970 *mp = m; 971 return (0); 972 } 973 974 /* 975 * Search header for all Hop-by-hop options and process each option. 976 * This function is separate from ip6_hopopts_input() in order to 977 * handle a case where the sending node itself process its hop-by-hop 978 * options header. In such a case, the function is called from ip6_output(). 979 * 980 * The function assumes that hbh header is located right after the IPv6 header 981 * (RFC2460 p7), opthead is pointer into data content in m, and opthead to 982 * opthead + hbhlen is located in continuous memory region. 983 */ 984 int 985 ip6_process_hopopts(struct mbuf *m, u_int8_t *opthead, int hbhlen, 986 u_int32_t *rtalertp, u_int32_t *plenp) 987 { 988 struct ip6_hdr *ip6; 989 int optlen = 0; 990 u_int8_t *opt = opthead; 991 u_int16_t rtalert_val; 992 u_int32_t jumboplen; 993 const int erroff = sizeof(struct ip6_hdr) + sizeof(struct ip6_hbh); 994 995 for (; hbhlen > 0; hbhlen -= optlen, opt += optlen) { 996 switch (*opt) { 997 case IP6OPT_PAD1: 998 optlen = 1; 999 break; 1000 case IP6OPT_PADN: 1001 if (hbhlen < IP6OPT_MINLEN) { 1002 ip6stat.ip6s_toosmall++; 1003 goto bad; 1004 } 1005 optlen = *(opt + 1) + 2; 1006 break; 1007 case IP6OPT_RTALERT: 1008 /* XXX may need check for alignment */ 1009 if (hbhlen < IP6OPT_RTALERT_LEN) { 1010 ip6stat.ip6s_toosmall++; 1011 goto bad; 1012 } 1013 if (*(opt + 1) != IP6OPT_RTALERT_LEN - 2) { 1014 /* XXX stat */ 1015 icmp6_error(m, ICMP6_PARAM_PROB, 1016 ICMP6_PARAMPROB_HEADER, 1017 erroff + opt + 1 - opthead); 1018 return (-1); 1019 } 1020 optlen = IP6OPT_RTALERT_LEN; 1021 bcopy((caddr_t)(opt + 2), (caddr_t)&rtalert_val, 2); 1022 *rtalertp = ntohs(rtalert_val); 1023 break; 1024 case IP6OPT_JUMBO: 1025 /* XXX may need check for alignment */ 1026 if (hbhlen < IP6OPT_JUMBO_LEN) { 1027 ip6stat.ip6s_toosmall++; 1028 goto bad; 1029 } 1030 if (*(opt + 1) != IP6OPT_JUMBO_LEN - 2) { 1031 /* XXX stat */ 1032 icmp6_error(m, ICMP6_PARAM_PROB, 1033 ICMP6_PARAMPROB_HEADER, 1034 erroff + opt + 1 - opthead); 1035 return (-1); 1036 } 1037 optlen = IP6OPT_JUMBO_LEN; 1038 1039 /* 1040 * IPv6 packets that have non 0 payload length 1041 * must not contain a jumbo payload option. 1042 */ 1043 ip6 = mtod(m, struct ip6_hdr *); 1044 if (ip6->ip6_plen) { 1045 ip6stat.ip6s_badoptions++; 1046 icmp6_error(m, ICMP6_PARAM_PROB, 1047 ICMP6_PARAMPROB_HEADER, 1048 erroff + opt - opthead); 1049 return (-1); 1050 } 1051 1052 /* 1053 * We may see jumbolen in unaligned location, so 1054 * we'd need to perform bcopy(). 1055 */ 1056 bcopy(opt + 2, &jumboplen, sizeof(jumboplen)); 1057 jumboplen = (u_int32_t)htonl(jumboplen); 1058 1059 #if 1 1060 /* 1061 * if there are multiple jumbo payload options, 1062 * *plenp will be non-zero and the packet will be 1063 * rejected. 1064 * the behavior may need some debate in ipngwg - 1065 * multiple options does not make sense, however, 1066 * there's no explicit mention in specification. 1067 */ 1068 if (*plenp != 0) { 1069 ip6stat.ip6s_badoptions++; 1070 icmp6_error(m, ICMP6_PARAM_PROB, 1071 ICMP6_PARAMPROB_HEADER, 1072 erroff + opt + 2 - opthead); 1073 return (-1); 1074 } 1075 #endif 1076 1077 /* 1078 * jumbo payload length must be larger than 65535. 1079 */ 1080 if (jumboplen <= IPV6_MAXPACKET) { 1081 ip6stat.ip6s_badoptions++; 1082 icmp6_error(m, ICMP6_PARAM_PROB, 1083 ICMP6_PARAMPROB_HEADER, 1084 erroff + opt + 2 - opthead); 1085 return (-1); 1086 } 1087 *plenp = jumboplen; 1088 1089 break; 1090 default: /* unknown option */ 1091 if (hbhlen < IP6OPT_MINLEN) { 1092 ip6stat.ip6s_toosmall++; 1093 goto bad; 1094 } 1095 optlen = ip6_unknown_opt(opt, m, 1096 erroff + opt - opthead); 1097 if (optlen == -1) 1098 return (-1); 1099 optlen += 2; 1100 break; 1101 } 1102 } 1103 1104 return (0); 1105 1106 bad: 1107 m_freem(m); 1108 return (-1); 1109 } 1110 1111 /* 1112 * Unknown option processing. 1113 * The third argument `off' is the offset from the IPv6 header to the option, 1114 * which is necessary if the IPv6 header the and option header and IPv6 header 1115 * is not continuous in order to return an ICMPv6 error. 1116 */ 1117 int 1118 ip6_unknown_opt(u_int8_t *optp, struct mbuf *m, int off) 1119 { 1120 struct ip6_hdr *ip6; 1121 1122 switch (IP6OPT_TYPE(*optp)) { 1123 case IP6OPT_TYPE_SKIP: /* ignore the option */ 1124 return ((int)*(optp + 1)); 1125 case IP6OPT_TYPE_DISCARD: /* silently discard */ 1126 m_freem(m); 1127 return (-1); 1128 case IP6OPT_TYPE_FORCEICMP: /* send ICMP even if multicasted */ 1129 ip6stat.ip6s_badoptions++; 1130 icmp6_error(m, ICMP6_PARAM_PROB, ICMP6_PARAMPROB_OPTION, off); 1131 return (-1); 1132 case IP6OPT_TYPE_ICMP: /* send ICMP if not multicasted */ 1133 ip6stat.ip6s_badoptions++; 1134 ip6 = mtod(m, struct ip6_hdr *); 1135 if (IN6_IS_ADDR_MULTICAST(&ip6->ip6_dst) || 1136 (m->m_flags & (M_BCAST|M_MCAST))) 1137 m_freem(m); 1138 else 1139 icmp6_error(m, ICMP6_PARAM_PROB, 1140 ICMP6_PARAMPROB_OPTION, off); 1141 return (-1); 1142 } 1143 1144 m_freem(m); /* XXX: NOTREACHED */ 1145 return (-1); 1146 } 1147 1148 /* 1149 * Create the "control" list for this pcb. 1150 * The function will not modify mbuf chain at all. 1151 * 1152 * with KAME mbuf chain restriction: 1153 * The routine will be called from upper layer handlers like tcp6_input(). 1154 * Thus the routine assumes that the caller (tcp6_input) have already 1155 * called IP6_EXTHDR_CHECK() and all the extension headers are located in the 1156 * very first mbuf on the mbuf chain. 1157 */ 1158 void 1159 ip6_savecontrol(struct inpcb *in6p, struct mbuf **mp, struct ip6_hdr *ip6, 1160 struct mbuf *m) 1161 { 1162 #define IS2292(x, y) ((in6p->in6p_flags & IN6P_RFC2292) ? (x) : (y)) 1163 struct thread *td = curthread; /* XXX */ 1164 int privileged = 0; 1165 1166 1167 if (priv_check(td, PRIV_ROOT) == 0) 1168 privileged++; 1169 1170 #ifdef SO_TIMESTAMP 1171 if (in6p->in6p_socket->so_options & SO_TIMESTAMP) { 1172 struct timeval tv; 1173 1174 microtime(&tv); 1175 *mp = sbcreatecontrol((caddr_t) &tv, sizeof(tv), 1176 SCM_TIMESTAMP, SOL_SOCKET); 1177 if (*mp) { 1178 mp = &(*mp)->m_next; 1179 } 1180 } 1181 #endif 1182 1183 /* RFC 2292 sec. 5 */ 1184 if (in6p->in6p_flags & IN6P_PKTINFO) { 1185 struct in6_pktinfo pi6; 1186 bcopy(&ip6->ip6_dst, &pi6.ipi6_addr, sizeof(struct in6_addr)); 1187 if (IN6_IS_SCOPE_LINKLOCAL(&pi6.ipi6_addr)) 1188 pi6.ipi6_addr.s6_addr16[1] = 0; 1189 pi6.ipi6_ifindex = (m && m->m_pkthdr.rcvif) 1190 ? m->m_pkthdr.rcvif->if_index 1191 : 0; 1192 *mp = sbcreatecontrol((caddr_t) &pi6, 1193 sizeof(struct in6_pktinfo), IS2292(IPV6_2292PKTINFO, IPV6_PKTINFO), 1194 IPPROTO_IPV6); 1195 if (*mp) 1196 mp = &(*mp)->m_next; 1197 } 1198 1199 if (in6p->in6p_flags & IN6P_HOPLIMIT) { 1200 int hlim = ip6->ip6_hlim & 0xff; 1201 *mp = sbcreatecontrol((caddr_t) &hlim, 1202 sizeof(int), IS2292(IPV6_2292HOPLIMIT, IPV6_HOPLIMIT), IPPROTO_IPV6); 1203 if (*mp) 1204 mp = &(*mp)->m_next; 1205 } 1206 1207 if ((in6p->in6p_flags & IN6P_TCLASS) != 0) { 1208 u_int32_t flowinfo; 1209 int tclass; 1210 1211 flowinfo = (u_int32_t)ntohl(ip6->ip6_flow & IPV6_FLOWINFO_MASK); 1212 flowinfo >>= 20; 1213 1214 tclass = flowinfo & 0xff; 1215 *mp = sbcreatecontrol((caddr_t) &tclass, sizeof(tclass), 1216 IPV6_TCLASS, IPPROTO_IPV6); 1217 if (*mp) 1218 mp = &(*mp)->m_next; 1219 } 1220 1221 /* 1222 * IPV6_HOPOPTS socket option. We require super-user privilege 1223 * for the option, but it might be too strict, since there might 1224 * be some hop-by-hop options which can be returned to normal user. 1225 * See RFC 2292 section 6. 1226 */ 1227 if ((in6p->in6p_flags & IN6P_HOPOPTS) && privileged) { 1228 /* 1229 * Check if a hop-by-hop options header is contatined in the 1230 * received packet, and if so, store the options as ancillary 1231 * data. Note that a hop-by-hop options header must be 1232 * just after the IPv6 header, which fact is assured through 1233 * the IPv6 input processing. 1234 */ 1235 struct ip6_hdr *ip6 = mtod(m, struct ip6_hdr *); 1236 if (ip6->ip6_nxt == IPPROTO_HOPOPTS) { 1237 struct ip6_hbh *hbh; 1238 int hbhlen = 0; 1239 #ifdef PULLDOWN_TEST 1240 struct mbuf *ext; 1241 #endif 1242 1243 #ifndef PULLDOWN_TEST 1244 hbh = (struct ip6_hbh *)(ip6 + 1); 1245 hbhlen = (hbh->ip6h_len + 1) << 3; 1246 #else 1247 ext = ip6_pullexthdr(m, sizeof(struct ip6_hdr), 1248 ip6->ip6_nxt); 1249 if (ext == NULL) { 1250 ip6stat.ip6s_tooshort++; 1251 return; 1252 } 1253 hbh = mtod(ext, struct ip6_hbh *); 1254 hbhlen = (hbh->ip6h_len + 1) << 3; 1255 if (hbhlen != ext->m_len) { 1256 m_freem(ext); 1257 ip6stat.ip6s_tooshort++; 1258 return; 1259 } 1260 #endif 1261 1262 /* 1263 * XXX: We copy whole the header even if a jumbo 1264 * payload option is included, which option is to 1265 * be removed before returning in the RFC 2292. 1266 * Note: this constraint is removed in 2292bis. 1267 */ 1268 *mp = sbcreatecontrol((caddr_t)hbh, hbhlen, 1269 IS2292(IPV6_2292HOPOPTS, IPV6_HOPOPTS), IPPROTO_IPV6); 1270 if (*mp) 1271 mp = &(*mp)->m_next; 1272 #ifdef PULLDOWN_TEST 1273 m_freem(ext); 1274 #endif 1275 } 1276 } 1277 1278 /* IPV6_DSTOPTS and IPV6_RTHDR socket options */ 1279 if ((in6p->in6p_flags & (IN6P_DSTOPTS | IN6P_RTHDRDSTOPTS)) != 0) { 1280 int proto, off, nxt; 1281 1282 /* 1283 * go through the header chain to see if a routing header is 1284 * contained in the packet. We need this information to store 1285 * destination options headers (if any) properly. 1286 * XXX: performance issue. We should record this info when 1287 * processing extension headers in incoming routine. 1288 * (todo) use m_aux? 1289 */ 1290 proto = IPPROTO_IPV6; 1291 off = 0; 1292 nxt = -1; 1293 while (1) { 1294 int newoff; 1295 1296 newoff = ip6_nexthdr(m, off, proto, &nxt); 1297 if (newoff < 0) 1298 break; 1299 if (newoff < off) /* invalid, check for safety */ 1300 break; 1301 if ((proto = nxt) == IPPROTO_ROUTING) 1302 break; 1303 off = newoff; 1304 } 1305 } 1306 1307 if ((in6p->in6p_flags & 1308 (IN6P_RTHDR | IN6P_DSTOPTS | IN6P_RTHDRDSTOPTS)) != 0) { 1309 struct ip6_hdr *ip6 = mtod(m, struct ip6_hdr *); 1310 int nxt = ip6->ip6_nxt, off = sizeof(struct ip6_hdr); 1311 1312 /* 1313 * Search for destination options headers or routing 1314 * header(s) through the header chain, and stores each 1315 * header as ancillary data. 1316 * Note that the order of the headers remains in 1317 * the chain of ancillary data. 1318 */ 1319 while (1) { /* is explicit loop prevention necessary? */ 1320 struct ip6_ext *ip6e = NULL; 1321 int elen; 1322 #ifdef PULLDOWN_TEST 1323 struct mbuf *ext = NULL; 1324 #endif 1325 1326 /* 1327 * if it is not an extension header, don't try to 1328 * pull it from the chain. 1329 */ 1330 switch (nxt) { 1331 case IPPROTO_DSTOPTS: 1332 case IPPROTO_ROUTING: 1333 case IPPROTO_HOPOPTS: 1334 case IPPROTO_AH: /* is it possible? */ 1335 break; 1336 default: 1337 goto loopend; 1338 } 1339 1340 #ifndef PULLDOWN_TEST 1341 if (off + sizeof(*ip6e) > m->m_len) 1342 goto loopend; 1343 ip6e = (struct ip6_ext *)(mtod(m, caddr_t) + off); 1344 if (nxt == IPPROTO_AH) 1345 elen = (ip6e->ip6e_len + 2) << 2; 1346 else 1347 elen = (ip6e->ip6e_len + 1) << 3; 1348 if (off + elen > m->m_len) 1349 goto loopend; 1350 #else 1351 ext = ip6_pullexthdr(m, off, nxt); 1352 if (ext == NULL) { 1353 ip6stat.ip6s_tooshort++; 1354 return; 1355 } 1356 ip6e = mtod(ext, struct ip6_ext *); 1357 if (nxt == IPPROTO_AH) 1358 elen = (ip6e->ip6e_len + 2) << 2; 1359 else 1360 elen = (ip6e->ip6e_len + 1) << 3; 1361 if (elen != ext->m_len) { 1362 m_freem(ext); 1363 ip6stat.ip6s_tooshort++; 1364 return; 1365 } 1366 #endif 1367 1368 switch (nxt) { 1369 case IPPROTO_DSTOPTS: 1370 if ((in6p->in6p_flags & IN6P_DSTOPTS) == 0) 1371 break; 1372 1373 /* 1374 * We also require super-user privilege for 1375 * the option. 1376 * See the comments on IN6_HOPOPTS. 1377 */ 1378 if (!privileged) 1379 break; 1380 1381 *mp = sbcreatecontrol((caddr_t)ip6e, elen, 1382 IS2292(IPV6_2292DSTOPTS, IPV6_DSTOPTS), 1383 IPPROTO_IPV6); 1384 if (*mp) 1385 mp = &(*mp)->m_next; 1386 break; 1387 case IPPROTO_ROUTING: 1388 if (!(in6p->in6p_flags & IN6P_RTHDR)) 1389 break; 1390 1391 *mp = sbcreatecontrol((caddr_t)ip6e, elen, 1392 IS2292(IPV6_2292RTHDR, IPV6_RTHDR), 1393 IPPROTO_IPV6); 1394 if (*mp) 1395 mp = &(*mp)->m_next; 1396 break; 1397 case IPPROTO_HOPOPTS: 1398 case IPPROTO_AH: /* is it possible? */ 1399 break; 1400 1401 default: 1402 /* 1403 * other cases have been filtered in the above. 1404 * none will visit this case. here we supply 1405 * the code just in case (nxt overwritten or 1406 * other cases). 1407 */ 1408 #ifdef PULLDOWN_TEST 1409 m_freem(ext); 1410 #endif 1411 goto loopend; 1412 1413 } 1414 1415 /* proceed with the next header. */ 1416 off += elen; 1417 nxt = ip6e->ip6e_nxt; 1418 ip6e = NULL; 1419 #ifdef PULLDOWN_TEST 1420 m_freem(ext); 1421 ext = NULL; 1422 #endif 1423 } 1424 loopend: 1425 ; 1426 } 1427 #undef IS2292 1428 } 1429 1430 void 1431 ip6_notify_pmtu(struct inpcb *in6p, struct sockaddr_in6 *dst, u_int32_t *mtu) 1432 { 1433 struct socket *so; 1434 struct mbuf *m_mtu; 1435 struct ip6_mtuinfo mtuctl; 1436 1437 so = in6p->inp_socket; 1438 1439 if (mtu == NULL) 1440 return; 1441 1442 #ifdef DIAGNOSTIC 1443 if (so == NULL) /* I believe this is impossible */ 1444 panic("ip6_notify_pmtu: socket is NULL"); 1445 #endif 1446 1447 bzero(&mtuctl, sizeof(mtuctl)); /* zero-clear for safety */ 1448 mtuctl.ip6m_mtu = *mtu; 1449 mtuctl.ip6m_addr = *dst; 1450 1451 if ((m_mtu = sbcreatecontrol((caddr_t)&mtuctl, sizeof(mtuctl), 1452 IPV6_PATHMTU, IPPROTO_IPV6)) == NULL) 1453 return; 1454 1455 lwkt_gettoken(&so->so_rcv.ssb_token); 1456 if (sbappendaddr(&so->so_rcv.sb, (struct sockaddr *)dst, NULL, m_mtu) 1457 == 0) { 1458 m_freem(m_mtu); 1459 /* XXX: should count statistics */ 1460 } else { 1461 sorwakeup(so); 1462 } 1463 lwkt_reltoken(&so->so_rcv.ssb_token); 1464 } 1465 1466 #ifdef PULLDOWN_TEST 1467 /* 1468 * pull single extension header from mbuf chain. returns single mbuf that 1469 * contains the result, or NULL on error. 1470 */ 1471 static struct mbuf * 1472 ip6_pullexthdr(struct mbuf *m, size_t off, int nxt) 1473 { 1474 struct ip6_ext ip6e; 1475 size_t elen; 1476 struct mbuf *n; 1477 1478 #ifdef DIAGNOSTIC 1479 switch (nxt) { 1480 case IPPROTO_DSTOPTS: 1481 case IPPROTO_ROUTING: 1482 case IPPROTO_HOPOPTS: 1483 case IPPROTO_AH: /* is it possible? */ 1484 break; 1485 default: 1486 kprintf("ip6_pullexthdr: invalid nxt=%d\n", nxt); 1487 } 1488 #endif 1489 1490 m_copydata(m, off, sizeof(ip6e), (caddr_t)&ip6e); 1491 if (nxt == IPPROTO_AH) 1492 elen = (ip6e.ip6e_len + 2) << 2; 1493 else 1494 elen = (ip6e.ip6e_len + 1) << 3; 1495 1496 n = m_getb(elen, MB_DONTWAIT, MT_DATA, 0); 1497 if (n == NULL) 1498 return NULL; 1499 n->m_len = 0; 1500 1501 if (elen >= M_TRAILINGSPACE(n)) { 1502 m_free(n); 1503 return NULL; 1504 } 1505 1506 m_copydata(m, off, elen, mtod(n, caddr_t)); 1507 n->m_len = elen; 1508 return n; 1509 } 1510 #endif 1511 1512 /* 1513 * Get pointer to the previous header followed by the header 1514 * currently processed. 1515 * XXX: This function supposes that 1516 * M includes all headers, 1517 * the next header field and the header length field of each header 1518 * are valid, and 1519 * the sum of each header length equals to OFF. 1520 * Because of these assumptions, this function must be called very 1521 * carefully. Moreover, it will not be used in the near future when 1522 * we develop `neater' mechanism to process extension headers. 1523 */ 1524 char * 1525 ip6_get_prevhdr(struct mbuf *m, int off) 1526 { 1527 struct ip6_hdr *ip6 = mtod(m, struct ip6_hdr *); 1528 1529 if (off == sizeof(struct ip6_hdr)) 1530 return (&ip6->ip6_nxt); 1531 else { 1532 int len, nxt; 1533 struct ip6_ext *ip6e = NULL; 1534 1535 nxt = ip6->ip6_nxt; 1536 len = sizeof(struct ip6_hdr); 1537 while (len < off) { 1538 ip6e = (struct ip6_ext *)(mtod(m, caddr_t) + len); 1539 1540 switch (nxt) { 1541 case IPPROTO_FRAGMENT: 1542 len += sizeof(struct ip6_frag); 1543 break; 1544 case IPPROTO_AH: 1545 len += (ip6e->ip6e_len + 2) << 2; 1546 break; 1547 default: 1548 len += (ip6e->ip6e_len + 1) << 3; 1549 break; 1550 } 1551 nxt = ip6e->ip6e_nxt; 1552 } 1553 if (ip6e) 1554 return (&ip6e->ip6e_nxt); 1555 else 1556 return NULL; 1557 } 1558 } 1559 1560 /* 1561 * get next header offset. m will be retained. 1562 */ 1563 int 1564 ip6_nexthdr(struct mbuf *m, int off, int proto, int *nxtp) 1565 { 1566 struct ip6_hdr ip6; 1567 struct ip6_ext ip6e; 1568 struct ip6_frag fh; 1569 1570 /* just in case */ 1571 if (m == NULL) 1572 panic("ip6_nexthdr: m == NULL"); 1573 if ((m->m_flags & M_PKTHDR) == 0 || m->m_pkthdr.len < off) 1574 return -1; 1575 1576 switch (proto) { 1577 case IPPROTO_IPV6: 1578 if (m->m_pkthdr.len < off + sizeof(ip6)) 1579 return -1; 1580 m_copydata(m, off, sizeof(ip6), (caddr_t)&ip6); 1581 if (nxtp) 1582 *nxtp = ip6.ip6_nxt; 1583 off += sizeof(ip6); 1584 return off; 1585 1586 case IPPROTO_FRAGMENT: 1587 /* 1588 * terminate parsing if it is not the first fragment, 1589 * it does not make sense to parse through it. 1590 */ 1591 if (m->m_pkthdr.len < off + sizeof(fh)) 1592 return -1; 1593 m_copydata(m, off, sizeof(fh), (caddr_t)&fh); 1594 /* IP6F_OFF_MASK = 0xfff8(BigEndian), 0xf8ff(LittleEndian) */ 1595 if (fh.ip6f_offlg & IP6F_OFF_MASK) 1596 return -1; 1597 if (nxtp) 1598 *nxtp = fh.ip6f_nxt; 1599 off += sizeof(struct ip6_frag); 1600 return off; 1601 1602 case IPPROTO_AH: 1603 if (m->m_pkthdr.len < off + sizeof(ip6e)) 1604 return -1; 1605 m_copydata(m, off, sizeof(ip6e), (caddr_t)&ip6e); 1606 if (nxtp) 1607 *nxtp = ip6e.ip6e_nxt; 1608 off += (ip6e.ip6e_len + 2) << 2; 1609 return off; 1610 1611 case IPPROTO_HOPOPTS: 1612 case IPPROTO_ROUTING: 1613 case IPPROTO_DSTOPTS: 1614 if (m->m_pkthdr.len < off + sizeof(ip6e)) 1615 return -1; 1616 m_copydata(m, off, sizeof(ip6e), (caddr_t)&ip6e); 1617 if (nxtp) 1618 *nxtp = ip6e.ip6e_nxt; 1619 off += (ip6e.ip6e_len + 1) << 3; 1620 return off; 1621 1622 case IPPROTO_NONE: 1623 case IPPROTO_ESP: 1624 case IPPROTO_IPCOMP: 1625 /* give up */ 1626 return -1; 1627 1628 default: 1629 return -1; 1630 } 1631 1632 return -1; 1633 } 1634 1635 /* 1636 * get offset for the last header in the chain. m will be kept untainted. 1637 */ 1638 int 1639 ip6_lasthdr(struct mbuf *m, int off, int proto, int *nxtp) 1640 { 1641 int newoff; 1642 int nxt; 1643 1644 if (!nxtp) { 1645 nxt = -1; 1646 nxtp = &nxt; 1647 } 1648 while (1) { 1649 newoff = ip6_nexthdr(m, off, proto, nxtp); 1650 if (newoff < 0) 1651 return off; 1652 else if (newoff < off) 1653 return -1; /* invalid */ 1654 else if (newoff == off) 1655 return newoff; 1656 1657 off = newoff; 1658 proto = *nxtp; 1659 } 1660 } 1661 1662 struct ip6aux * 1663 ip6_addaux(struct mbuf *m) 1664 { 1665 struct m_tag *tag = m_tag_find(m, PACKET_TAG_IPV6_INPUT, NULL); 1666 if (!tag) { 1667 tag = m_tag_get(PACKET_TAG_IPV6_INPUT, 1668 sizeof (struct ip6aux), 1669 MB_DONTWAIT); 1670 if (tag) 1671 m_tag_prepend(m, tag); 1672 } 1673 if (tag) 1674 bzero(m_tag_data(tag), sizeof (struct ip6aux)); 1675 return tag ? (struct ip6aux *)m_tag_data(tag) : NULL; 1676 } 1677 1678 struct ip6aux * 1679 ip6_findaux(struct mbuf *m) 1680 { 1681 struct m_tag *tag = m_tag_find(m, PACKET_TAG_IPV6_INPUT, NULL); 1682 return tag ? (struct ip6aux*)(tag+1) : NULL; 1683 } 1684 1685 void 1686 ip6_delaux(struct mbuf *m) 1687 { 1688 struct m_tag *tag = m_tag_find(m, PACKET_TAG_IPV6_INPUT, NULL); 1689 if (tag) 1690 m_tag_delete(m, tag); 1691 } 1692 1693 /* 1694 * System control for IP6 1695 */ 1696 1697 u_char inet6ctlerrmap[PRC_NCMDS] = { 1698 0, 0, 0, 0, 1699 0, EMSGSIZE, EHOSTDOWN, EHOSTUNREACH, 1700 EHOSTUNREACH, EHOSTUNREACH, ECONNREFUSED, ECONNREFUSED, 1701 EMSGSIZE, EHOSTUNREACH, 0, 0, 1702 0, 0, 0, 0, 1703 ENOPROTOOPT 1704 }; 1705