1 /* $FreeBSD: src/sys/netinet6/ip6_input.c,v 1.11.2.15 2003/01/24 05:11:35 sam Exp $ */ 2 /* $DragonFly: src/sys/netinet6/ip6_input.c,v 1.38 2008/09/24 14:26:39 sephe Exp $ */ 3 /* $KAME: ip6_input.c,v 1.259 2002/01/21 04:58:09 jinmei Exp $ */ 4 5 /* 6 * Copyright (C) 1995, 1996, 1997, and 1998 WIDE Project. 7 * All rights reserved. 8 * 9 * Redistribution and use in source and binary forms, with or without 10 * modification, are permitted provided that the following conditions 11 * are met: 12 * 1. Redistributions of source code must retain the above copyright 13 * notice, this list of conditions and the following disclaimer. 14 * 2. Redistributions in binary form must reproduce the above copyright 15 * notice, this list of conditions and the following disclaimer in the 16 * documentation and/or other materials provided with the distribution. 17 * 3. Neither the name of the project nor the names of its contributors 18 * may be used to endorse or promote products derived from this software 19 * without specific prior written permission. 20 * 21 * THIS SOFTWARE IS PROVIDED BY THE PROJECT AND CONTRIBUTORS ``AS IS'' AND 22 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE 23 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE 24 * ARE DISCLAIMED. IN NO EVENT SHALL THE PROJECT OR CONTRIBUTORS BE LIABLE 25 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL 26 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS 27 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) 28 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT 29 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY 30 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF 31 * SUCH DAMAGE. 32 */ 33 34 /* 35 * Copyright (c) 1982, 1986, 1988, 1993 36 * The Regents of the University of California. All rights reserved. 37 * 38 * Redistribution and use in source and binary forms, with or without 39 * modification, are permitted provided that the following conditions 40 * are met: 41 * 1. Redistributions of source code must retain the above copyright 42 * notice, this list of conditions and the following disclaimer. 43 * 2. Redistributions in binary form must reproduce the above copyright 44 * notice, this list of conditions and the following disclaimer in the 45 * documentation and/or other materials provided with the distribution. 46 * 3. All advertising materials mentioning features or use of this software 47 * must display the following acknowledgement: 48 * This product includes software developed by the University of 49 * California, Berkeley and its contributors. 50 * 4. Neither the name of the University nor the names of its contributors 51 * may be used to endorse or promote products derived from this software 52 * without specific prior written permission. 53 * 54 * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND 55 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE 56 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE 57 * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE 58 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL 59 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS 60 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) 61 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT 62 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY 63 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF 64 * SUCH DAMAGE. 65 * 66 * @(#)ip_input.c 8.2 (Berkeley) 1/4/94 67 */ 68 69 #include "opt_ip6fw.h" 70 #include "opt_inet.h" 71 #include "opt_inet6.h" 72 #include "opt_ipsec.h" 73 74 #include <sys/param.h> 75 #include <sys/systm.h> 76 #include <sys/malloc.h> 77 #include <sys/mbuf.h> 78 #include <sys/domain.h> 79 #include <sys/protosw.h> 80 #include <sys/socket.h> 81 #include <sys/socketvar.h> 82 #include <sys/errno.h> 83 #include <sys/time.h> 84 #include <sys/kernel.h> 85 #include <sys/syslog.h> 86 #include <sys/proc.h> 87 #include <sys/priv.h> 88 89 #include <sys/thread2.h> 90 #include <sys/msgport2.h> 91 92 #include <net/if.h> 93 #include <net/if_types.h> 94 #include <net/if_dl.h> 95 #include <net/route.h> 96 #include <net/netisr.h> 97 #include <net/pfil.h> 98 99 #include <netinet/in.h> 100 #include <netinet/in_systm.h> 101 #ifdef INET 102 #include <netinet/ip.h> 103 #include <netinet/ip_icmp.h> 104 #endif /* INET */ 105 #include <netinet/ip6.h> 106 #include <netinet6/in6_var.h> 107 #include <netinet6/ip6_var.h> 108 #include <netinet/in_pcb.h> 109 #include <netinet/icmp6.h> 110 #include <netinet6/scope6_var.h> 111 #include <netinet6/in6_ifattach.h> 112 #include <netinet6/nd6.h> 113 #include <netinet6/in6_prefix.h> 114 115 #ifdef IPSEC 116 #include <netinet6/ipsec.h> 117 #ifdef INET6 118 #include <netinet6/ipsec6.h> 119 #endif 120 #endif 121 122 #ifdef FAST_IPSEC 123 #include <netproto/ipsec/ipsec.h> 124 #include <netproto/ipsec/ipsec6.h> 125 #define IPSEC 126 #endif /* FAST_IPSEC */ 127 128 #include <net/ip6fw/ip6_fw.h> 129 130 #include <netinet6/ip6protosw.h> 131 132 /* we need it for NLOOP. */ 133 #include "use_loop.h" 134 135 #include <net/net_osdep.h> 136 137 extern struct domain inet6domain; 138 extern struct ip6protosw inet6sw[]; 139 140 u_char ip6_protox[IPPROTO_MAX]; 141 struct in6_ifaddr *in6_ifaddr; 142 143 extern struct callout in6_tmpaddrtimer_ch; 144 145 int ip6_forward_srcrt; /* XXX */ 146 int ip6_sourcecheck; /* XXX */ 147 int ip6_sourcecheck_interval; /* XXX */ 148 149 int ip6_ours_check_algorithm; 150 151 struct pfil_head inet6_pfil_hook; 152 153 /* firewall hooks */ 154 ip6_fw_chk_t *ip6_fw_chk_ptr; 155 ip6_fw_ctl_t *ip6_fw_ctl_ptr; 156 int ip6_fw_enable = 1; 157 158 struct ip6stat ip6stat; 159 160 static void ip6_init2 (void *); 161 static struct ip6aux *ip6_setdstifaddr (struct mbuf *, struct in6_ifaddr *); 162 static int ip6_hopopts_input (u_int32_t *, u_int32_t *, struct mbuf **, int *); 163 static void ip6_input(struct netmsg *msg); 164 #ifdef PULLDOWN_TEST 165 static struct mbuf *ip6_pullexthdr (struct mbuf *, size_t, int); 166 #endif 167 168 /* 169 * IP6 initialization: fill in IP6 protocol switch table. 170 * All protocols not implemented in kernel go to raw IP6 protocol handler. 171 */ 172 void 173 ip6_init(void) 174 { 175 struct ip6protosw *pr; 176 int i; 177 struct timeval tv; 178 179 #ifdef DIAGNOSTIC 180 if (sizeof(struct protosw) != sizeof(struct ip6protosw)) 181 panic("sizeof(protosw) != sizeof(ip6protosw)"); 182 #endif 183 pr = (struct ip6protosw *)pffindproto(PF_INET6, IPPROTO_RAW, SOCK_RAW); 184 if (pr == 0) 185 panic("ip6_init"); 186 for (i = 0; i < IPPROTO_MAX; i++) 187 ip6_protox[i] = pr - inet6sw; 188 for (pr = (struct ip6protosw *)inet6domain.dom_protosw; 189 pr < (struct ip6protosw *)inet6domain.dom_protoswNPROTOSW; pr++) 190 if (pr->pr_domain->dom_family == PF_INET6 && 191 pr->pr_protocol && pr->pr_protocol != IPPROTO_RAW) 192 ip6_protox[pr->pr_protocol] = pr - inet6sw; 193 194 inet6_pfil_hook.ph_type = PFIL_TYPE_AF; 195 inet6_pfil_hook.ph_af = AF_INET6; 196 if ((i = pfil_head_register(&inet6_pfil_hook)) != 0) { 197 kprintf("%s: WARNING: unable to register pfil hook, " 198 "error %d\n", __func__, i); 199 } 200 201 netisr_register(NETISR_IPV6, cpu0_portfn, pktinfo_portfn_cpu0, 202 ip6_input, NETISR_FLAG_NOTMPSAFE); 203 scope6_init(); 204 addrsel_policy_init(); 205 nd6_init(); 206 frag6_init(); 207 /* 208 * in many cases, random() here does NOT return random number 209 * as initialization during bootstrap time occur in fixed order. 210 */ 211 microtime(&tv); 212 ip6_flow_seq = krandom() ^ tv.tv_usec; 213 microtime(&tv); 214 ip6_desync_factor = (krandom() ^ tv.tv_usec) % MAX_TEMP_DESYNC_FACTOR; 215 } 216 217 static void 218 ip6_init2(void *dummy) 219 { 220 /* nd6_timer_init */ 221 callout_init(&nd6_timer_ch); 222 callout_reset(&nd6_timer_ch, hz, nd6_timer, NULL); 223 224 /* router renumbering prefix list maintenance */ 225 callout_init(&in6_rr_timer_ch); 226 callout_reset(&in6_rr_timer_ch, hz, in6_rr_timer, NULL); 227 228 /* timer for regeneranation of temporary addresses randomize ID */ 229 callout_init(&in6_tmpaddrtimer_ch); 230 callout_reset(&in6_tmpaddrtimer_ch, 231 (ip6_temp_preferred_lifetime - ip6_desync_factor - 232 ip6_temp_regen_advance) * hz, 233 in6_tmpaddrtimer, NULL); 234 } 235 236 /* cheat */ 237 /* This must be after route_init(), which is now SI_ORDER_THIRD */ 238 SYSINIT(netinet6init2, SI_SUB_PROTO_DOMAIN, SI_ORDER_MIDDLE, ip6_init2, NULL); 239 240 extern struct route_in6 ip6_forward_rt; 241 242 static 243 void 244 ip6_input(struct netmsg *msg) 245 { 246 struct mbuf *m = ((struct netmsg_packet *)msg)->nm_packet; 247 struct ip6_hdr *ip6; 248 int off = sizeof(struct ip6_hdr), nest; 249 u_int32_t plen; 250 u_int32_t rtalert = ~0; 251 int nxt, ours = 0, rh_present = 0; 252 struct ifnet *deliverifp = NULL; 253 struct in6_addr odst; 254 int srcrt = 0; 255 256 #ifdef IPSEC 257 /* 258 * should the inner packet be considered authentic? 259 * see comment in ah4_input(). 260 */ 261 if (m) { 262 m->m_flags &= ~M_AUTHIPHDR; 263 m->m_flags &= ~M_AUTHIPDGM; 264 } 265 #endif 266 267 /* 268 * make sure we don't have onion peering information into m_aux. 269 */ 270 ip6_delaux(m); 271 272 /* 273 * mbuf statistics 274 */ 275 if (m->m_flags & M_EXT) { 276 if (m->m_next) 277 ip6stat.ip6s_mext2m++; 278 else 279 ip6stat.ip6s_mext1++; 280 } else { 281 #define M2MMAX (sizeof(ip6stat.ip6s_m2m)/sizeof(ip6stat.ip6s_m2m[0])) 282 if (m->m_next) { 283 if (m->m_flags & M_LOOP) { 284 ip6stat.ip6s_m2m[loif[0].if_index]++; /* XXX */ 285 } else if (m->m_pkthdr.rcvif->if_index < M2MMAX) 286 ip6stat.ip6s_m2m[m->m_pkthdr.rcvif->if_index]++; 287 else 288 ip6stat.ip6s_m2m[0]++; 289 } else 290 ip6stat.ip6s_m1++; 291 #undef M2MMAX 292 } 293 294 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_receive); 295 ip6stat.ip6s_total++; 296 297 #ifndef PULLDOWN_TEST 298 /* 299 * L2 bridge code and some other code can return mbuf chain 300 * that does not conform to KAME requirement. too bad. 301 * XXX: fails to join if interface MTU > MCLBYTES. jumbogram? 302 */ 303 if (m && m->m_next != NULL && m->m_pkthdr.len < MCLBYTES) { 304 struct mbuf *n; 305 306 n = m_getb(m->m_pkthdr.len, MB_DONTWAIT, MT_HEADER, M_PKTHDR); 307 if (n == NULL) 308 goto bad; 309 M_MOVE_PKTHDR(n, m); 310 311 m_copydata(m, 0, n->m_pkthdr.len, mtod(n, caddr_t)); 312 n->m_len = n->m_pkthdr.len; 313 m_freem(m); 314 m = n; 315 } 316 IP6_EXTHDR_CHECK_VOIDRET(m, 0, sizeof(struct ip6_hdr)); 317 #endif 318 319 if (m->m_len < sizeof(struct ip6_hdr)) { 320 struct ifnet *inifp; 321 inifp = m->m_pkthdr.rcvif; 322 if ((m = m_pullup(m, sizeof(struct ip6_hdr))) == 0) { 323 ip6stat.ip6s_toosmall++; 324 in6_ifstat_inc(inifp, ifs6_in_hdrerr); 325 goto bad2; 326 } 327 } 328 329 ip6 = mtod(m, struct ip6_hdr *); 330 331 if ((ip6->ip6_vfc & IPV6_VERSION_MASK) != IPV6_VERSION) { 332 ip6stat.ip6s_badvers++; 333 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_hdrerr); 334 goto bad; 335 } 336 337 /* 338 * Run through list of hooks for input packets. 339 * 340 * NB: Beware of the destination address changing 341 * (e.g. by NAT rewriting). When this happens, 342 * tell ip6_forward to do the right thing. 343 */ 344 if (pfil_has_hooks(&inet6_pfil_hook)) { 345 odst = ip6->ip6_dst; 346 if (pfil_run_hooks(&inet6_pfil_hook, &m, 347 m->m_pkthdr.rcvif, PFIL_IN)) { 348 goto bad2; 349 } 350 if (m == NULL) /* consumed by filter */ 351 goto bad2; 352 ip6 = mtod(m, struct ip6_hdr *); 353 srcrt = !IN6_ARE_ADDR_EQUAL(&odst, &ip6->ip6_dst); 354 } 355 356 ip6stat.ip6s_nxthist[ip6->ip6_nxt]++; 357 358 #ifdef ALTQ 359 if (altq_input != NULL && (*altq_input)(m, AF_INET6) == 0) { 360 /* packet is dropped by traffic conditioner */ 361 return; 362 } 363 #endif 364 365 /* 366 * Check with the firewall... 367 */ 368 if (ip6_fw_enable && ip6_fw_chk_ptr) { 369 u_short port = 0; 370 /* If ipfw says divert, we have to just drop packet */ 371 /* use port as a dummy argument */ 372 if ((*ip6_fw_chk_ptr)(&ip6, NULL, &port, &m)) { 373 m_freem(m); 374 m = NULL; 375 } 376 if (!m) 377 goto bad2; 378 } 379 380 /* 381 * Check against address spoofing/corruption. 382 */ 383 if (IN6_IS_ADDR_MULTICAST(&ip6->ip6_src) || 384 IN6_IS_ADDR_UNSPECIFIED(&ip6->ip6_dst)) { 385 /* 386 * XXX: "badscope" is not very suitable for a multicast source. 387 */ 388 ip6stat.ip6s_badscope++; 389 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_addrerr); 390 goto bad; 391 } 392 if ((IN6_IS_ADDR_LOOPBACK(&ip6->ip6_src) || 393 IN6_IS_ADDR_LOOPBACK(&ip6->ip6_dst)) && 394 (m->m_pkthdr.rcvif->if_flags & IFF_LOOPBACK) == 0) { 395 ip6stat.ip6s_badscope++; 396 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_addrerr); 397 goto bad; 398 } 399 400 /* 401 * The following check is not documented in specs. A malicious 402 * party may be able to use IPv4 mapped addr to confuse tcp/udp stack 403 * and bypass security checks (act as if it was from 127.0.0.1 by using 404 * IPv6 src ::ffff:127.0.0.1). Be cautious. 405 * 406 * This check chokes if we are in an SIIT cloud. As none of BSDs 407 * support IPv4-less kernel compilation, we cannot support SIIT 408 * environment at all. So, it makes more sense for us to reject any 409 * malicious packets for non-SIIT environment, than try to do a 410 * partical support for SIIT environment. 411 */ 412 if (IN6_IS_ADDR_V4MAPPED(&ip6->ip6_src) || 413 IN6_IS_ADDR_V4MAPPED(&ip6->ip6_dst)) { 414 ip6stat.ip6s_badscope++; 415 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_addrerr); 416 goto bad; 417 } 418 #if 0 419 /* 420 * Reject packets with IPv4 compatible addresses (auto tunnel). 421 * 422 * The code forbids auto tunnel relay case in RFC1933 (the check is 423 * stronger than RFC1933). We may want to re-enable it if mech-xx 424 * is revised to forbid relaying case. 425 */ 426 if (IN6_IS_ADDR_V4COMPAT(&ip6->ip6_src) || 427 IN6_IS_ADDR_V4COMPAT(&ip6->ip6_dst)) { 428 ip6stat.ip6s_badscope++; 429 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_addrerr); 430 goto bad; 431 } 432 #endif 433 434 /* drop packets if interface ID portion is already filled */ 435 if ((m->m_pkthdr.rcvif->if_flags & IFF_LOOPBACK) == 0) { 436 if (IN6_IS_SCOPE_LINKLOCAL(&ip6->ip6_src) && 437 ip6->ip6_src.s6_addr16[1]) { 438 ip6stat.ip6s_badscope++; 439 goto bad; 440 } 441 if (IN6_IS_SCOPE_LINKLOCAL(&ip6->ip6_dst) && 442 ip6->ip6_dst.s6_addr16[1]) { 443 ip6stat.ip6s_badscope++; 444 goto bad; 445 } 446 } 447 448 if (IN6_IS_SCOPE_LINKLOCAL(&ip6->ip6_src)) 449 ip6->ip6_src.s6_addr16[1] 450 = htons(m->m_pkthdr.rcvif->if_index); 451 if (IN6_IS_SCOPE_LINKLOCAL(&ip6->ip6_dst)) 452 ip6->ip6_dst.s6_addr16[1] 453 = htons(m->m_pkthdr.rcvif->if_index); 454 455 #if 0 /* this case seems to be unnecessary. (jinmei, 20010401) */ 456 /* 457 * We use rt->rt_ifp to determine if the address is ours or not. 458 * If rt_ifp is lo0, the address is ours. 459 * The problem here is, rt->rt_ifp for fe80::%lo0/64 is set to lo0, 460 * so any address under fe80::%lo0/64 will be mistakenly considered 461 * local. The special case is supplied to handle the case properly 462 * by actually looking at interface addresses 463 * (using in6ifa_ifpwithaddr). 464 */ 465 if ((m->m_pkthdr.rcvif->if_flags & IFF_LOOPBACK) && 466 IN6_IS_ADDR_LINKLOCAL(&ip6->ip6_dst)) { 467 if (!in6ifa_ifpwithaddr(m->m_pkthdr.rcvif, &ip6->ip6_dst)) { 468 icmp6_error(m, ICMP6_DST_UNREACH, 469 ICMP6_DST_UNREACH_ADDR, 0); 470 /* m is already freed */ 471 goto bad2; 472 } 473 474 ours = 1; 475 deliverifp = m->m_pkthdr.rcvif; 476 goto hbhcheck; 477 } 478 #endif 479 480 /* 481 * Multicast check 482 */ 483 if (IN6_IS_ADDR_MULTICAST(&ip6->ip6_dst)) { 484 struct in6_multi *in6m = 0; 485 486 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_mcast); 487 /* 488 * See if we belong to the destination multicast group on the 489 * arrival interface. 490 */ 491 IN6_LOOKUP_MULTI(ip6->ip6_dst, m->m_pkthdr.rcvif, in6m); 492 if (in6m) 493 ours = 1; 494 else if (!ip6_mrouter) { 495 ip6stat.ip6s_notmember++; 496 ip6stat.ip6s_cantforward++; 497 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_discard); 498 goto bad; 499 } 500 deliverifp = m->m_pkthdr.rcvif; 501 goto hbhcheck; 502 } 503 504 /* 505 * Unicast check 506 */ 507 switch (ip6_ours_check_algorithm) { 508 default: 509 /* 510 * XXX: I intentionally broke our indentation rule here, 511 * since this switch-case is just for measurement and 512 * therefore should soon be removed. 513 */ 514 if (ip6_forward_rt.ro_rt != NULL && 515 (ip6_forward_rt.ro_rt->rt_flags & RTF_UP) && 516 IN6_ARE_ADDR_EQUAL(&ip6->ip6_dst, 517 &((struct sockaddr_in6 *)(&ip6_forward_rt.ro_dst))->sin6_addr)) 518 ip6stat.ip6s_forward_cachehit++; 519 else { 520 struct sockaddr_in6 *dst6; 521 522 if (ip6_forward_rt.ro_rt) { 523 /* route is down or destination is different */ 524 ip6stat.ip6s_forward_cachemiss++; 525 RTFREE(ip6_forward_rt.ro_rt); 526 ip6_forward_rt.ro_rt = 0; 527 } 528 529 bzero(&ip6_forward_rt.ro_dst, sizeof(struct sockaddr_in6)); 530 dst6 = &ip6_forward_rt.ro_dst; 531 dst6->sin6_len = sizeof(struct sockaddr_in6); 532 dst6->sin6_family = AF_INET6; 533 dst6->sin6_addr = ip6->ip6_dst; 534 535 rtalloc_ign((struct route *)&ip6_forward_rt, RTF_PRCLONING); 536 } 537 538 #define rt6_key(r) ((struct sockaddr_in6 *)((r)->rt_nodes->rn_key)) 539 540 /* 541 * Accept the packet if the forwarding interface to the destination 542 * according to the routing table is the loopback interface, 543 * unless the associated route has a gateway. 544 * Note that this approach causes to accept a packet if there is a 545 * route to the loopback interface for the destination of the packet. 546 * But we think it's even useful in some situations, e.g. when using 547 * a special daemon which wants to intercept the packet. 548 * 549 * XXX: some OSes automatically make a cloned route for the destination 550 * of an outgoing packet. If the outgoing interface of the packet 551 * is a loopback one, the kernel would consider the packet to be 552 * accepted, even if we have no such address assinged on the interface. 553 * We check the cloned flag of the route entry to reject such cases, 554 * assuming that route entries for our own addresses are not made by 555 * cloning (it should be true because in6_addloop explicitly installs 556 * the host route). However, we might have to do an explicit check 557 * while it would be less efficient. Or, should we rather install a 558 * reject route for such a case? 559 */ 560 if (ip6_forward_rt.ro_rt && 561 (ip6_forward_rt.ro_rt->rt_flags & 562 (RTF_HOST|RTF_GATEWAY)) == RTF_HOST && 563 #ifdef RTF_WASCLONED 564 !(ip6_forward_rt.ro_rt->rt_flags & RTF_WASCLONED) && 565 #endif 566 #ifdef RTF_CLONED 567 !(ip6_forward_rt.ro_rt->rt_flags & RTF_CLONED) && 568 #endif 569 #if 0 570 /* 571 * The check below is redundant since the comparison of 572 * the destination and the key of the rtentry has 573 * already done through looking up the routing table. 574 */ 575 IN6_ARE_ADDR_EQUAL(&ip6->ip6_dst, 576 &rt6_key(ip6_forward_rt.ro_rt)->sin6_addr) 577 #endif 578 ip6_forward_rt.ro_rt->rt_ifp->if_type == IFT_LOOP) { 579 struct in6_ifaddr *ia6 = 580 (struct in6_ifaddr *)ip6_forward_rt.ro_rt->rt_ifa; 581 582 /* 583 * record address information into m_aux. 584 */ 585 ip6_setdstifaddr(m, ia6); 586 587 /* 588 * packets to a tentative, duplicated, or somehow invalid 589 * address must not be accepted. 590 */ 591 if (!(ia6->ia6_flags & IN6_IFF_NOTREADY)) { 592 /* this address is ready */ 593 ours = 1; 594 deliverifp = ia6->ia_ifp; /* correct? */ 595 /* Count the packet in the ip address stats */ 596 ia6->ia_ifa.if_ipackets++; 597 ia6->ia_ifa.if_ibytes += m->m_pkthdr.len; 598 goto hbhcheck; 599 } else { 600 /* address is not ready, so discard the packet. */ 601 nd6log((LOG_INFO, 602 "ip6_input: packet to an unready address %s->%s\n", 603 ip6_sprintf(&ip6->ip6_src), 604 ip6_sprintf(&ip6->ip6_dst))); 605 606 goto bad; 607 } 608 } 609 } /* XXX indentation (see above) */ 610 611 /* 612 * FAITH(Firewall Aided Internet Translator) 613 */ 614 if (ip6_keepfaith) { 615 if (ip6_forward_rt.ro_rt && ip6_forward_rt.ro_rt->rt_ifp 616 && ip6_forward_rt.ro_rt->rt_ifp->if_type == IFT_FAITH) { 617 /* XXX do we need more sanity checks? */ 618 ours = 1; 619 deliverifp = ip6_forward_rt.ro_rt->rt_ifp; /* faith */ 620 goto hbhcheck; 621 } 622 } 623 624 /* 625 * Now there is no reason to process the packet if it's not our own 626 * and we're not a router. 627 */ 628 if (!ip6_forwarding) { 629 ip6stat.ip6s_cantforward++; 630 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_discard); 631 goto bad; 632 } 633 634 hbhcheck: 635 /* 636 * record address information into m_aux, if we don't have one yet. 637 * note that we are unable to record it, if the address is not listed 638 * as our interface address (e.g. multicast addresses, addresses 639 * within FAITH prefixes and such). 640 */ 641 if (deliverifp && !ip6_getdstifaddr(m)) { 642 struct in6_ifaddr *ia6; 643 644 ia6 = in6_ifawithifp(deliverifp, &ip6->ip6_dst); 645 if (ia6) { 646 if (!ip6_setdstifaddr(m, ia6)) { 647 /* 648 * XXX maybe we should drop the packet here, 649 * as we could not provide enough information 650 * to the upper layers. 651 */ 652 } 653 } 654 } 655 656 /* 657 * Process Hop-by-Hop options header if it's contained. 658 * m may be modified in ip6_hopopts_input(). 659 * If a JumboPayload option is included, plen will also be modified. 660 */ 661 plen = (u_int32_t)ntohs(ip6->ip6_plen); 662 if (ip6->ip6_nxt == IPPROTO_HOPOPTS) { 663 struct ip6_hbh *hbh; 664 665 if (ip6_hopopts_input(&plen, &rtalert, &m, &off)) { 666 #if 0 /*touches NULL pointer*/ 667 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_discard); 668 #endif 669 goto bad2; /* m have already been freed */ 670 } 671 672 /* adjust pointer */ 673 ip6 = mtod(m, struct ip6_hdr *); 674 675 /* 676 * if the payload length field is 0 and the next header field 677 * indicates Hop-by-Hop Options header, then a Jumbo Payload 678 * option MUST be included. 679 */ 680 if (ip6->ip6_plen == 0 && plen == 0) { 681 /* 682 * Note that if a valid jumbo payload option is 683 * contained, ip6_hoptops_input() must set a valid 684 * (non-zero) payload length to the variable plen. 685 */ 686 ip6stat.ip6s_badoptions++; 687 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_discard); 688 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_hdrerr); 689 icmp6_error(m, ICMP6_PARAM_PROB, 690 ICMP6_PARAMPROB_HEADER, 691 (caddr_t)&ip6->ip6_plen - (caddr_t)ip6); 692 goto bad2; 693 } 694 #ifndef PULLDOWN_TEST 695 /* ip6_hopopts_input() ensures that mbuf is contiguous */ 696 hbh = (struct ip6_hbh *)(ip6 + 1); 697 #else 698 IP6_EXTHDR_GET(hbh, struct ip6_hbh *, m, sizeof(struct ip6_hdr), 699 sizeof(struct ip6_hbh)); 700 if (hbh == NULL) { 701 ip6stat.ip6s_tooshort++; 702 goto bad2; 703 } 704 #endif 705 nxt = hbh->ip6h_nxt; 706 707 /* 708 * If we are acting as a router and the packet contains a 709 * router alert option, see if we know the option value. 710 * Currently, we only support the option value for MLD, in which 711 * case we should pass the packet to the multicast routing 712 * daemon. 713 */ 714 if (rtalert != ~0 && ip6_forwarding) { 715 switch (rtalert) { 716 case IP6OPT_RTALERT_MLD: 717 ours = 1; 718 break; 719 default: 720 /* 721 * RFC2711 requires unrecognized values must be 722 * silently ignored. 723 */ 724 break; 725 } 726 } 727 } else 728 nxt = ip6->ip6_nxt; 729 730 /* 731 * Check that the amount of data in the buffers 732 * is as at least much as the IPv6 header would have us expect. 733 * Trim mbufs if longer than we expect. 734 * Drop packet if shorter than we expect. 735 */ 736 if (m->m_pkthdr.len - sizeof(struct ip6_hdr) < plen) { 737 ip6stat.ip6s_tooshort++; 738 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_truncated); 739 goto bad; 740 } 741 if (m->m_pkthdr.len > sizeof(struct ip6_hdr) + plen) { 742 if (m->m_len == m->m_pkthdr.len) { 743 m->m_len = sizeof(struct ip6_hdr) + plen; 744 m->m_pkthdr.len = sizeof(struct ip6_hdr) + plen; 745 } else 746 m_adj(m, sizeof(struct ip6_hdr) + plen - m->m_pkthdr.len); 747 } 748 749 /* 750 * Forward if desirable. 751 */ 752 if (IN6_IS_ADDR_MULTICAST(&ip6->ip6_dst)) { 753 /* 754 * If we are acting as a multicast router, all 755 * incoming multicast packets are passed to the 756 * kernel-level multicast forwarding function. 757 * The packet is returned (relatively) intact; if 758 * ip6_mforward() returns a non-zero value, the packet 759 * must be discarded, else it may be accepted below. 760 */ 761 if (ip6_mrouter && ip6_mforward(ip6, m->m_pkthdr.rcvif, m)) { 762 ip6stat.ip6s_cantforward++; 763 goto bad; 764 } 765 if (!ours) 766 goto bad; 767 } else if (!ours) { 768 ip6_forward(m, srcrt); 769 goto bad2; 770 } 771 772 ip6 = mtod(m, struct ip6_hdr *); 773 774 /* 775 * Malicious party may be able to use IPv4 mapped addr to confuse 776 * tcp/udp stack and bypass security checks (act as if it was from 777 * 127.0.0.1 by using IPv6 src ::ffff:127.0.0.1). Be cautious. 778 * 779 * For SIIT end node behavior, you may want to disable the check. 780 * However, you will become vulnerable to attacks using IPv4 mapped 781 * source. 782 */ 783 if (IN6_IS_ADDR_V4MAPPED(&ip6->ip6_src) || 784 IN6_IS_ADDR_V4MAPPED(&ip6->ip6_dst)) { 785 ip6stat.ip6s_badscope++; 786 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_addrerr); 787 goto bad; 788 } 789 790 /* 791 * Tell launch routine the next header 792 */ 793 ip6stat.ip6s_delivered++; 794 in6_ifstat_inc(deliverifp, ifs6_in_deliver); 795 nest = 0; 796 797 rh_present = 0; 798 while (nxt != IPPROTO_DONE) { 799 if (ip6_hdrnestlimit && (++nest > ip6_hdrnestlimit)) { 800 ip6stat.ip6s_toomanyhdr++; 801 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_hdrerr); 802 goto bad; 803 } 804 805 /* 806 * protection against faulty packet - there should be 807 * more sanity checks in header chain processing. 808 */ 809 if (m->m_pkthdr.len < off) { 810 ip6stat.ip6s_tooshort++; 811 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_truncated); 812 goto bad; 813 } 814 815 #if 0 816 /* 817 * do we need to do it for every header? yeah, other 818 * functions can play with it (like re-allocate and copy). 819 */ 820 mhist = ip6_addaux(m); 821 if (mhist && M_TRAILINGSPACE(mhist) >= sizeof(nxt)) { 822 hist = mtod(mhist, caddr_t) + mhist->m_len; 823 bcopy(&nxt, hist, sizeof(nxt)); 824 mhist->m_len += sizeof(nxt); 825 } else { 826 ip6stat.ip6s_toomanyhdr++; 827 goto bad; 828 } 829 #endif 830 831 if (nxt == IPPROTO_ROUTING) { 832 if (rh_present++) { 833 in6_ifstat_inc(m->m_pkthdr.rcvif, 834 ifs6_in_hdrerr); 835 ip6stat.ip6s_badoptions++; 836 goto bad; 837 } 838 } 839 840 #ifdef IPSEC 841 /* 842 * enforce IPsec policy checking if we are seeing last header. 843 * note that we do not visit this with protocols with pcb layer 844 * code - like udp/tcp/raw ip. 845 */ 846 if ((inet6sw[ip6_protox[nxt]].pr_flags & PR_LASTHDR) && 847 ipsec6_in_reject(m, NULL)) { 848 ipsec6stat.in_polvio++; 849 goto bad; 850 } 851 #endif 852 853 nxt = (*inet6sw[ip6_protox[nxt]].pr_input)(&m, &off, nxt); 854 } 855 goto bad2; 856 bad: 857 m_freem(m); 858 bad2: 859 ; 860 /* msg was embedded in the mbuf, do not reply! */ 861 } 862 863 /* 864 * set/grab in6_ifaddr correspond to IPv6 destination address. 865 * XXX backward compatibility wrapper 866 */ 867 static struct ip6aux * 868 ip6_setdstifaddr(struct mbuf *m, struct in6_ifaddr *ia6) 869 { 870 struct ip6aux *n; 871 872 n = ip6_addaux(m); 873 if (n) 874 n->ip6a_dstia6 = ia6; 875 return n; /* NULL if failed to set */ 876 } 877 878 struct in6_ifaddr * 879 ip6_getdstifaddr(struct mbuf *m) 880 { 881 struct ip6aux *n; 882 883 n = ip6_findaux(m); 884 if (n) 885 return n->ip6a_dstia6; 886 else 887 return NULL; 888 } 889 890 /* 891 * Hop-by-Hop options header processing. If a valid jumbo payload option is 892 * included, the real payload length will be stored in plenp. 893 */ 894 static int 895 ip6_hopopts_input(u_int32_t *plenp, 896 u_int32_t *rtalertp,/* XXX: should be stored more smart way */ 897 struct mbuf **mp, 898 int *offp) 899 { 900 struct mbuf *m = *mp; 901 int off = *offp, hbhlen; 902 struct ip6_hbh *hbh; 903 u_int8_t *opt; 904 905 /* validation of the length of the header */ 906 #ifndef PULLDOWN_TEST 907 IP6_EXTHDR_CHECK(m, off, sizeof(*hbh), -1); 908 hbh = (struct ip6_hbh *)(mtod(m, caddr_t) + off); 909 hbhlen = (hbh->ip6h_len + 1) << 3; 910 911 IP6_EXTHDR_CHECK(m, off, hbhlen, -1); 912 hbh = (struct ip6_hbh *)(mtod(m, caddr_t) + off); 913 #else 914 IP6_EXTHDR_GET(hbh, struct ip6_hbh *, m, 915 sizeof(struct ip6_hdr), sizeof(struct ip6_hbh)); 916 if (hbh == NULL) { 917 ip6stat.ip6s_tooshort++; 918 return -1; 919 } 920 hbhlen = (hbh->ip6h_len + 1) << 3; 921 IP6_EXTHDR_GET(hbh, struct ip6_hbh *, m, sizeof(struct ip6_hdr), 922 hbhlen); 923 if (hbh == NULL) { 924 ip6stat.ip6s_tooshort++; 925 return -1; 926 } 927 #endif 928 off += hbhlen; 929 hbhlen -= sizeof(struct ip6_hbh); 930 opt = (u_int8_t *)hbh + sizeof(struct ip6_hbh); 931 932 if (ip6_process_hopopts(m, (u_int8_t *)hbh + sizeof(struct ip6_hbh), 933 hbhlen, rtalertp, plenp) < 0) 934 return (-1); 935 936 *offp = off; 937 *mp = m; 938 return (0); 939 } 940 941 /* 942 * Search header for all Hop-by-hop options and process each option. 943 * This function is separate from ip6_hopopts_input() in order to 944 * handle a case where the sending node itself process its hop-by-hop 945 * options header. In such a case, the function is called from ip6_output(). 946 * 947 * The function assumes that hbh header is located right after the IPv6 header 948 * (RFC2460 p7), opthead is pointer into data content in m, and opthead to 949 * opthead + hbhlen is located in continuous memory region. 950 */ 951 int 952 ip6_process_hopopts(struct mbuf *m, u_int8_t *opthead, int hbhlen, 953 u_int32_t *rtalertp, u_int32_t *plenp) 954 { 955 struct ip6_hdr *ip6; 956 int optlen = 0; 957 u_int8_t *opt = opthead; 958 u_int16_t rtalert_val; 959 u_int32_t jumboplen; 960 const int erroff = sizeof(struct ip6_hdr) + sizeof(struct ip6_hbh); 961 962 for (; hbhlen > 0; hbhlen -= optlen, opt += optlen) { 963 switch (*opt) { 964 case IP6OPT_PAD1: 965 optlen = 1; 966 break; 967 case IP6OPT_PADN: 968 if (hbhlen < IP6OPT_MINLEN) { 969 ip6stat.ip6s_toosmall++; 970 goto bad; 971 } 972 optlen = *(opt + 1) + 2; 973 break; 974 case IP6OPT_RTALERT: 975 /* XXX may need check for alignment */ 976 if (hbhlen < IP6OPT_RTALERT_LEN) { 977 ip6stat.ip6s_toosmall++; 978 goto bad; 979 } 980 if (*(opt + 1) != IP6OPT_RTALERT_LEN - 2) { 981 /* XXX stat */ 982 icmp6_error(m, ICMP6_PARAM_PROB, 983 ICMP6_PARAMPROB_HEADER, 984 erroff + opt + 1 - opthead); 985 return (-1); 986 } 987 optlen = IP6OPT_RTALERT_LEN; 988 bcopy((caddr_t)(opt + 2), (caddr_t)&rtalert_val, 2); 989 *rtalertp = ntohs(rtalert_val); 990 break; 991 case IP6OPT_JUMBO: 992 /* XXX may need check for alignment */ 993 if (hbhlen < IP6OPT_JUMBO_LEN) { 994 ip6stat.ip6s_toosmall++; 995 goto bad; 996 } 997 if (*(opt + 1) != IP6OPT_JUMBO_LEN - 2) { 998 /* XXX stat */ 999 icmp6_error(m, ICMP6_PARAM_PROB, 1000 ICMP6_PARAMPROB_HEADER, 1001 erroff + opt + 1 - opthead); 1002 return (-1); 1003 } 1004 optlen = IP6OPT_JUMBO_LEN; 1005 1006 /* 1007 * IPv6 packets that have non 0 payload length 1008 * must not contain a jumbo payload option. 1009 */ 1010 ip6 = mtod(m, struct ip6_hdr *); 1011 if (ip6->ip6_plen) { 1012 ip6stat.ip6s_badoptions++; 1013 icmp6_error(m, ICMP6_PARAM_PROB, 1014 ICMP6_PARAMPROB_HEADER, 1015 erroff + opt - opthead); 1016 return (-1); 1017 } 1018 1019 /* 1020 * We may see jumbolen in unaligned location, so 1021 * we'd need to perform bcopy(). 1022 */ 1023 bcopy(opt + 2, &jumboplen, sizeof(jumboplen)); 1024 jumboplen = (u_int32_t)htonl(jumboplen); 1025 1026 #if 1 1027 /* 1028 * if there are multiple jumbo payload options, 1029 * *plenp will be non-zero and the packet will be 1030 * rejected. 1031 * the behavior may need some debate in ipngwg - 1032 * multiple options does not make sense, however, 1033 * there's no explicit mention in specification. 1034 */ 1035 if (*plenp != 0) { 1036 ip6stat.ip6s_badoptions++; 1037 icmp6_error(m, ICMP6_PARAM_PROB, 1038 ICMP6_PARAMPROB_HEADER, 1039 erroff + opt + 2 - opthead); 1040 return (-1); 1041 } 1042 #endif 1043 1044 /* 1045 * jumbo payload length must be larger than 65535. 1046 */ 1047 if (jumboplen <= IPV6_MAXPACKET) { 1048 ip6stat.ip6s_badoptions++; 1049 icmp6_error(m, ICMP6_PARAM_PROB, 1050 ICMP6_PARAMPROB_HEADER, 1051 erroff + opt + 2 - opthead); 1052 return (-1); 1053 } 1054 *plenp = jumboplen; 1055 1056 break; 1057 default: /* unknown option */ 1058 if (hbhlen < IP6OPT_MINLEN) { 1059 ip6stat.ip6s_toosmall++; 1060 goto bad; 1061 } 1062 optlen = ip6_unknown_opt(opt, m, 1063 erroff + opt - opthead); 1064 if (optlen == -1) 1065 return (-1); 1066 optlen += 2; 1067 break; 1068 } 1069 } 1070 1071 return (0); 1072 1073 bad: 1074 m_freem(m); 1075 return (-1); 1076 } 1077 1078 /* 1079 * Unknown option processing. 1080 * The third argument `off' is the offset from the IPv6 header to the option, 1081 * which is necessary if the IPv6 header the and option header and IPv6 header 1082 * is not continuous in order to return an ICMPv6 error. 1083 */ 1084 int 1085 ip6_unknown_opt(u_int8_t *optp, struct mbuf *m, int off) 1086 { 1087 struct ip6_hdr *ip6; 1088 1089 switch (IP6OPT_TYPE(*optp)) { 1090 case IP6OPT_TYPE_SKIP: /* ignore the option */ 1091 return ((int)*(optp + 1)); 1092 case IP6OPT_TYPE_DISCARD: /* silently discard */ 1093 m_freem(m); 1094 return (-1); 1095 case IP6OPT_TYPE_FORCEICMP: /* send ICMP even if multicasted */ 1096 ip6stat.ip6s_badoptions++; 1097 icmp6_error(m, ICMP6_PARAM_PROB, ICMP6_PARAMPROB_OPTION, off); 1098 return (-1); 1099 case IP6OPT_TYPE_ICMP: /* send ICMP if not multicasted */ 1100 ip6stat.ip6s_badoptions++; 1101 ip6 = mtod(m, struct ip6_hdr *); 1102 if (IN6_IS_ADDR_MULTICAST(&ip6->ip6_dst) || 1103 (m->m_flags & (M_BCAST|M_MCAST))) 1104 m_freem(m); 1105 else 1106 icmp6_error(m, ICMP6_PARAM_PROB, 1107 ICMP6_PARAMPROB_OPTION, off); 1108 return (-1); 1109 } 1110 1111 m_freem(m); /* XXX: NOTREACHED */ 1112 return (-1); 1113 } 1114 1115 /* 1116 * Create the "control" list for this pcb. 1117 * The function will not modify mbuf chain at all. 1118 * 1119 * with KAME mbuf chain restriction: 1120 * The routine will be called from upper layer handlers like tcp6_input(). 1121 * Thus the routine assumes that the caller (tcp6_input) have already 1122 * called IP6_EXTHDR_CHECK() and all the extension headers are located in the 1123 * very first mbuf on the mbuf chain. 1124 */ 1125 void 1126 ip6_savecontrol(struct inpcb *in6p, struct mbuf **mp, struct ip6_hdr *ip6, 1127 struct mbuf *m) 1128 { 1129 #define IS2292(x, y) ((in6p->in6p_flags & IN6P_RFC2292) ? (x) : (y)) 1130 struct thread *td = curthread; /* XXX */ 1131 int privileged = 0; 1132 int rthdr_exist = 0; 1133 1134 1135 if (priv_check(td, PRIV_ROOT) == 0) 1136 privileged++; 1137 1138 #ifdef SO_TIMESTAMP 1139 if (in6p->in6p_socket->so_options & SO_TIMESTAMP) { 1140 struct timeval tv; 1141 1142 microtime(&tv); 1143 *mp = sbcreatecontrol((caddr_t) &tv, sizeof(tv), 1144 SCM_TIMESTAMP, SOL_SOCKET); 1145 if (*mp) { 1146 mp = &(*mp)->m_next; 1147 } 1148 } 1149 #endif 1150 1151 /* RFC 2292 sec. 5 */ 1152 if (in6p->in6p_flags & IN6P_PKTINFO) { 1153 struct in6_pktinfo pi6; 1154 bcopy(&ip6->ip6_dst, &pi6.ipi6_addr, sizeof(struct in6_addr)); 1155 if (IN6_IS_SCOPE_LINKLOCAL(&pi6.ipi6_addr)) 1156 pi6.ipi6_addr.s6_addr16[1] = 0; 1157 pi6.ipi6_ifindex = (m && m->m_pkthdr.rcvif) 1158 ? m->m_pkthdr.rcvif->if_index 1159 : 0; 1160 *mp = sbcreatecontrol((caddr_t) &pi6, 1161 sizeof(struct in6_pktinfo), IS2292(IPV6_2292PKTINFO, IPV6_PKTINFO), 1162 IPPROTO_IPV6); 1163 if (*mp) 1164 mp = &(*mp)->m_next; 1165 } 1166 1167 if (in6p->in6p_flags & IN6P_HOPLIMIT) { 1168 int hlim = ip6->ip6_hlim & 0xff; 1169 *mp = sbcreatecontrol((caddr_t) &hlim, 1170 sizeof(int), IS2292(IPV6_2292HOPLIMIT, IPV6_HOPLIMIT), IPPROTO_IPV6); 1171 if (*mp) 1172 mp = &(*mp)->m_next; 1173 } 1174 1175 if ((in6p->in6p_flags & IN6P_TCLASS) != 0) { 1176 u_int32_t flowinfo; 1177 int tclass; 1178 1179 flowinfo = (u_int32_t)ntohl(ip6->ip6_flow & IPV6_FLOWINFO_MASK); 1180 flowinfo >>= 20; 1181 1182 tclass = flowinfo & 0xff; 1183 *mp = sbcreatecontrol((caddr_t) &tclass, sizeof(tclass), 1184 IPV6_TCLASS, IPPROTO_IPV6); 1185 if (*mp) 1186 mp = &(*mp)->m_next; 1187 } 1188 1189 /* 1190 * IPV6_HOPOPTS socket option. We require super-user privilege 1191 * for the option, but it might be too strict, since there might 1192 * be some hop-by-hop options which can be returned to normal user. 1193 * See RFC 2292 section 6. 1194 */ 1195 if ((in6p->in6p_flags & IN6P_HOPOPTS) && privileged) { 1196 /* 1197 * Check if a hop-by-hop options header is contatined in the 1198 * received packet, and if so, store the options as ancillary 1199 * data. Note that a hop-by-hop options header must be 1200 * just after the IPv6 header, which fact is assured through 1201 * the IPv6 input processing. 1202 */ 1203 struct ip6_hdr *ip6 = mtod(m, struct ip6_hdr *); 1204 if (ip6->ip6_nxt == IPPROTO_HOPOPTS) { 1205 struct ip6_hbh *hbh; 1206 int hbhlen = 0; 1207 #ifdef PULLDOWN_TEST 1208 struct mbuf *ext; 1209 #endif 1210 1211 #ifndef PULLDOWN_TEST 1212 hbh = (struct ip6_hbh *)(ip6 + 1); 1213 hbhlen = (hbh->ip6h_len + 1) << 3; 1214 #else 1215 ext = ip6_pullexthdr(m, sizeof(struct ip6_hdr), 1216 ip6->ip6_nxt); 1217 if (ext == NULL) { 1218 ip6stat.ip6s_tooshort++; 1219 return; 1220 } 1221 hbh = mtod(ext, struct ip6_hbh *); 1222 hbhlen = (hbh->ip6h_len + 1) << 3; 1223 if (hbhlen != ext->m_len) { 1224 m_freem(ext); 1225 ip6stat.ip6s_tooshort++; 1226 return; 1227 } 1228 #endif 1229 1230 /* 1231 * XXX: We copy whole the header even if a jumbo 1232 * payload option is included, which option is to 1233 * be removed before returning in the RFC 2292. 1234 * Note: this constraint is removed in 2292bis. 1235 */ 1236 *mp = sbcreatecontrol((caddr_t)hbh, hbhlen, 1237 IS2292(IPV6_2292HOPOPTS, IPV6_HOPOPTS), IPPROTO_IPV6); 1238 if (*mp) 1239 mp = &(*mp)->m_next; 1240 #ifdef PULLDOWN_TEST 1241 m_freem(ext); 1242 #endif 1243 } 1244 } 1245 1246 /* IPV6_DSTOPTS and IPV6_RTHDR socket options */ 1247 if ((in6p->in6p_flags & (IN6P_DSTOPTS | IN6P_RTHDRDSTOPTS)) != 0) { 1248 int proto, off, nxt; 1249 1250 /* 1251 * go through the header chain to see if a routing header is 1252 * contained in the packet. We need this information to store 1253 * destination options headers (if any) properly. 1254 * XXX: performance issue. We should record this info when 1255 * processing extension headers in incoming routine. 1256 * (todo) use m_aux? 1257 */ 1258 proto = IPPROTO_IPV6; 1259 off = 0; 1260 nxt = -1; 1261 while (1) { 1262 int newoff; 1263 1264 newoff = ip6_nexthdr(m, off, proto, &nxt); 1265 if (newoff < 0) 1266 break; 1267 if (newoff < off) /* invalid, check for safety */ 1268 break; 1269 if ((proto = nxt) == IPPROTO_ROUTING) { 1270 rthdr_exist = 1; 1271 break; 1272 } 1273 off = newoff; 1274 } 1275 } 1276 1277 if ((in6p->in6p_flags & 1278 (IN6P_RTHDR | IN6P_DSTOPTS | IN6P_RTHDRDSTOPTS)) != 0) { 1279 struct ip6_hdr *ip6 = mtod(m, struct ip6_hdr *); 1280 int nxt = ip6->ip6_nxt, off = sizeof(struct ip6_hdr); 1281 1282 /* 1283 * Search for destination options headers or routing 1284 * header(s) through the header chain, and stores each 1285 * header as ancillary data. 1286 * Note that the order of the headers remains in 1287 * the chain of ancillary data. 1288 */ 1289 while (1) { /* is explicit loop prevention necessary? */ 1290 struct ip6_ext *ip6e = NULL; 1291 int elen; 1292 #ifdef PULLDOWN_TEST 1293 struct mbuf *ext = NULL; 1294 #endif 1295 1296 /* 1297 * if it is not an extension header, don't try to 1298 * pull it from the chain. 1299 */ 1300 switch (nxt) { 1301 case IPPROTO_DSTOPTS: 1302 case IPPROTO_ROUTING: 1303 case IPPROTO_HOPOPTS: 1304 case IPPROTO_AH: /* is it possible? */ 1305 break; 1306 default: 1307 goto loopend; 1308 } 1309 1310 #ifndef PULLDOWN_TEST 1311 if (off + sizeof(*ip6e) > m->m_len) 1312 goto loopend; 1313 ip6e = (struct ip6_ext *)(mtod(m, caddr_t) + off); 1314 if (nxt == IPPROTO_AH) 1315 elen = (ip6e->ip6e_len + 2) << 2; 1316 else 1317 elen = (ip6e->ip6e_len + 1) << 3; 1318 if (off + elen > m->m_len) 1319 goto loopend; 1320 #else 1321 ext = ip6_pullexthdr(m, off, nxt); 1322 if (ext == NULL) { 1323 ip6stat.ip6s_tooshort++; 1324 return; 1325 } 1326 ip6e = mtod(ext, struct ip6_ext *); 1327 if (nxt == IPPROTO_AH) 1328 elen = (ip6e->ip6e_len + 2) << 2; 1329 else 1330 elen = (ip6e->ip6e_len + 1) << 3; 1331 if (elen != ext->m_len) { 1332 m_freem(ext); 1333 ip6stat.ip6s_tooshort++; 1334 return; 1335 } 1336 #endif 1337 1338 switch (nxt) { 1339 case IPPROTO_DSTOPTS: 1340 if ((in6p->in6p_flags & IN6P_DSTOPTS) == 0) 1341 break; 1342 1343 /* 1344 * We also require super-user privilege for 1345 * the option. 1346 * See the comments on IN6_HOPOPTS. 1347 */ 1348 if (!privileged) 1349 break; 1350 1351 *mp = sbcreatecontrol((caddr_t)ip6e, elen, 1352 IS2292(IPV6_2292DSTOPTS, IPV6_DSTOPTS), 1353 IPPROTO_IPV6); 1354 if (*mp) 1355 mp = &(*mp)->m_next; 1356 break; 1357 case IPPROTO_ROUTING: 1358 if (!in6p->in6p_flags & IN6P_RTHDR) 1359 break; 1360 1361 *mp = sbcreatecontrol((caddr_t)ip6e, elen, 1362 IS2292(IPV6_2292RTHDR, IPV6_RTHDR), 1363 IPPROTO_IPV6); 1364 if (*mp) 1365 mp = &(*mp)->m_next; 1366 break; 1367 case IPPROTO_HOPOPTS: 1368 case IPPROTO_AH: /* is it possible? */ 1369 break; 1370 1371 default: 1372 /* 1373 * other cases have been filtered in the above. 1374 * none will visit this case. here we supply 1375 * the code just in case (nxt overwritten or 1376 * other cases). 1377 */ 1378 #ifdef PULLDOWN_TEST 1379 m_freem(ext); 1380 #endif 1381 goto loopend; 1382 1383 } 1384 1385 /* proceed with the next header. */ 1386 off += elen; 1387 nxt = ip6e->ip6e_nxt; 1388 ip6e = NULL; 1389 #ifdef PULLDOWN_TEST 1390 m_freem(ext); 1391 ext = NULL; 1392 #endif 1393 } 1394 loopend: 1395 ; 1396 } 1397 #undef IS2292 1398 } 1399 1400 void 1401 ip6_notify_pmtu(struct inpcb *in6p, struct sockaddr_in6 *dst, u_int32_t *mtu) 1402 { 1403 struct socket *so; 1404 struct mbuf *m_mtu; 1405 struct ip6_mtuinfo mtuctl; 1406 1407 so = in6p->inp_socket; 1408 1409 if (mtu == NULL) 1410 return; 1411 1412 #ifdef DIAGNOSTIC 1413 if (so == NULL) /* I believe this is impossible */ 1414 panic("ip6_notify_pmtu: socket is NULL"); 1415 #endif 1416 1417 bzero(&mtuctl, sizeof(mtuctl)); /* zero-clear for safety */ 1418 mtuctl.ip6m_mtu = *mtu; 1419 mtuctl.ip6m_addr = *dst; 1420 1421 if ((m_mtu = sbcreatecontrol((caddr_t)&mtuctl, sizeof(mtuctl), 1422 IPV6_PATHMTU, IPPROTO_IPV6)) == NULL) 1423 return; 1424 1425 if (sbappendaddr(&so->so_rcv, (struct sockaddr *)dst, NULL, m_mtu) 1426 == 0) { 1427 m_freem(m_mtu); 1428 /* XXX: should count statistics */ 1429 } else 1430 sorwakeup(so); 1431 1432 return; 1433 } 1434 1435 #ifdef PULLDOWN_TEST 1436 /* 1437 * pull single extension header from mbuf chain. returns single mbuf that 1438 * contains the result, or NULL on error. 1439 */ 1440 static struct mbuf * 1441 ip6_pullexthdr(struct mbuf *m, size_t off, int nxt) 1442 { 1443 struct ip6_ext ip6e; 1444 size_t elen; 1445 struct mbuf *n; 1446 1447 #ifdef DIAGNOSTIC 1448 switch (nxt) { 1449 case IPPROTO_DSTOPTS: 1450 case IPPROTO_ROUTING: 1451 case IPPROTO_HOPOPTS: 1452 case IPPROTO_AH: /* is it possible? */ 1453 break; 1454 default: 1455 kprintf("ip6_pullexthdr: invalid nxt=%d\n", nxt); 1456 } 1457 #endif 1458 1459 m_copydata(m, off, sizeof(ip6e), (caddr_t)&ip6e); 1460 if (nxt == IPPROTO_AH) 1461 elen = (ip6e.ip6e_len + 2) << 2; 1462 else 1463 elen = (ip6e.ip6e_len + 1) << 3; 1464 1465 n = m_getb(elen, MB_DONTWAIT, MT_DATA, 0); 1466 if (n == NULL) 1467 return NULL; 1468 n->m_len = 0; 1469 1470 if (elen >= M_TRAILINGSPACE(n)) { 1471 m_free(n); 1472 return NULL; 1473 } 1474 1475 m_copydata(m, off, elen, mtod(n, caddr_t)); 1476 n->m_len = elen; 1477 return n; 1478 } 1479 #endif 1480 1481 /* 1482 * Get pointer to the previous header followed by the header 1483 * currently processed. 1484 * XXX: This function supposes that 1485 * M includes all headers, 1486 * the next header field and the header length field of each header 1487 * are valid, and 1488 * the sum of each header length equals to OFF. 1489 * Because of these assumptions, this function must be called very 1490 * carefully. Moreover, it will not be used in the near future when 1491 * we develop `neater' mechanism to process extension headers. 1492 */ 1493 char * 1494 ip6_get_prevhdr(struct mbuf *m, int off) 1495 { 1496 struct ip6_hdr *ip6 = mtod(m, struct ip6_hdr *); 1497 1498 if (off == sizeof(struct ip6_hdr)) 1499 return (&ip6->ip6_nxt); 1500 else { 1501 int len, nxt; 1502 struct ip6_ext *ip6e = NULL; 1503 1504 nxt = ip6->ip6_nxt; 1505 len = sizeof(struct ip6_hdr); 1506 while (len < off) { 1507 ip6e = (struct ip6_ext *)(mtod(m, caddr_t) + len); 1508 1509 switch (nxt) { 1510 case IPPROTO_FRAGMENT: 1511 len += sizeof(struct ip6_frag); 1512 break; 1513 case IPPROTO_AH: 1514 len += (ip6e->ip6e_len + 2) << 2; 1515 break; 1516 default: 1517 len += (ip6e->ip6e_len + 1) << 3; 1518 break; 1519 } 1520 nxt = ip6e->ip6e_nxt; 1521 } 1522 if (ip6e) 1523 return (&ip6e->ip6e_nxt); 1524 else 1525 return NULL; 1526 } 1527 } 1528 1529 /* 1530 * get next header offset. m will be retained. 1531 */ 1532 int 1533 ip6_nexthdr(struct mbuf *m, int off, int proto, int *nxtp) 1534 { 1535 struct ip6_hdr ip6; 1536 struct ip6_ext ip6e; 1537 struct ip6_frag fh; 1538 1539 /* just in case */ 1540 if (m == NULL) 1541 panic("ip6_nexthdr: m == NULL"); 1542 if ((m->m_flags & M_PKTHDR) == 0 || m->m_pkthdr.len < off) 1543 return -1; 1544 1545 switch (proto) { 1546 case IPPROTO_IPV6: 1547 if (m->m_pkthdr.len < off + sizeof(ip6)) 1548 return -1; 1549 m_copydata(m, off, sizeof(ip6), (caddr_t)&ip6); 1550 if (nxtp) 1551 *nxtp = ip6.ip6_nxt; 1552 off += sizeof(ip6); 1553 return off; 1554 1555 case IPPROTO_FRAGMENT: 1556 /* 1557 * terminate parsing if it is not the first fragment, 1558 * it does not make sense to parse through it. 1559 */ 1560 if (m->m_pkthdr.len < off + sizeof(fh)) 1561 return -1; 1562 m_copydata(m, off, sizeof(fh), (caddr_t)&fh); 1563 /* IP6F_OFF_MASK = 0xfff8(BigEndian), 0xf8ff(LittleEndian) */ 1564 if (fh.ip6f_offlg & IP6F_OFF_MASK) 1565 return -1; 1566 if (nxtp) 1567 *nxtp = fh.ip6f_nxt; 1568 off += sizeof(struct ip6_frag); 1569 return off; 1570 1571 case IPPROTO_AH: 1572 if (m->m_pkthdr.len < off + sizeof(ip6e)) 1573 return -1; 1574 m_copydata(m, off, sizeof(ip6e), (caddr_t)&ip6e); 1575 if (nxtp) 1576 *nxtp = ip6e.ip6e_nxt; 1577 off += (ip6e.ip6e_len + 2) << 2; 1578 return off; 1579 1580 case IPPROTO_HOPOPTS: 1581 case IPPROTO_ROUTING: 1582 case IPPROTO_DSTOPTS: 1583 if (m->m_pkthdr.len < off + sizeof(ip6e)) 1584 return -1; 1585 m_copydata(m, off, sizeof(ip6e), (caddr_t)&ip6e); 1586 if (nxtp) 1587 *nxtp = ip6e.ip6e_nxt; 1588 off += (ip6e.ip6e_len + 1) << 3; 1589 return off; 1590 1591 case IPPROTO_NONE: 1592 case IPPROTO_ESP: 1593 case IPPROTO_IPCOMP: 1594 /* give up */ 1595 return -1; 1596 1597 default: 1598 return -1; 1599 } 1600 1601 return -1; 1602 } 1603 1604 /* 1605 * get offset for the last header in the chain. m will be kept untainted. 1606 */ 1607 int 1608 ip6_lasthdr(struct mbuf *m, int off, int proto, int *nxtp) 1609 { 1610 int newoff; 1611 int nxt; 1612 1613 if (!nxtp) { 1614 nxt = -1; 1615 nxtp = &nxt; 1616 } 1617 while (1) { 1618 newoff = ip6_nexthdr(m, off, proto, nxtp); 1619 if (newoff < 0) 1620 return off; 1621 else if (newoff < off) 1622 return -1; /* invalid */ 1623 else if (newoff == off) 1624 return newoff; 1625 1626 off = newoff; 1627 proto = *nxtp; 1628 } 1629 } 1630 1631 struct ip6aux * 1632 ip6_addaux(struct mbuf *m) 1633 { 1634 struct m_tag *tag = m_tag_find(m, PACKET_TAG_IPV6_INPUT, NULL); 1635 if (!tag) { 1636 tag = m_tag_get(PACKET_TAG_IPV6_INPUT, 1637 sizeof (struct ip6aux), 1638 MB_DONTWAIT); 1639 if (tag) 1640 m_tag_prepend(m, tag); 1641 } 1642 if (tag) 1643 bzero(m_tag_data(tag), sizeof (struct ip6aux)); 1644 return tag ? (struct ip6aux *)m_tag_data(tag) : NULL; 1645 } 1646 1647 struct ip6aux * 1648 ip6_findaux(struct mbuf *m) 1649 { 1650 struct m_tag *tag = m_tag_find(m, PACKET_TAG_IPV6_INPUT, NULL); 1651 return tag ? (struct ip6aux*)(tag+1) : NULL; 1652 } 1653 1654 void 1655 ip6_delaux(struct mbuf *m) 1656 { 1657 struct m_tag *tag = m_tag_find(m, PACKET_TAG_IPV6_INPUT, NULL); 1658 if (tag) 1659 m_tag_delete(m, tag); 1660 } 1661 1662 /* 1663 * System control for IP6 1664 */ 1665 1666 u_char inet6ctlerrmap[PRC_NCMDS] = { 1667 0, 0, 0, 0, 1668 0, EMSGSIZE, EHOSTDOWN, EHOSTUNREACH, 1669 EHOSTUNREACH, EHOSTUNREACH, ECONNREFUSED, ECONNREFUSED, 1670 EMSGSIZE, EHOSTUNREACH, 0, 0, 1671 0, 0, 0, 0, 1672 ENOPROTOOPT 1673 }; 1674