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. All advertising materials mentioning features or use of this software 46 * must display the following acknowledgement: 47 * This product includes software developed by the University of 48 * California, Berkeley and its contributors. 49 * 4. Neither the name of the University nor the names of its contributors 50 * may be used to endorse or promote products derived from this software 51 * without specific prior written permission. 52 * 53 * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND 54 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE 55 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE 56 * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE 57 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL 58 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS 59 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) 60 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT 61 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY 62 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF 63 * SUCH DAMAGE. 64 * 65 * @(#)ip_input.c 8.2 (Berkeley) 1/4/94 66 */ 67 68 #include "opt_ip6fw.h" 69 #include "opt_inet.h" 70 #include "opt_inet6.h" 71 #include "opt_ipsec.h" 72 73 #include <sys/param.h> 74 #include <sys/systm.h> 75 #include <sys/malloc.h> 76 #include <sys/mbuf.h> 77 #include <sys/domain.h> 78 #include <sys/protosw.h> 79 #include <sys/socket.h> 80 #include <sys/socketvar.h> 81 #include <sys/errno.h> 82 #include <sys/time.h> 83 #include <sys/kernel.h> 84 #include <sys/syslog.h> 85 #include <sys/proc.h> 86 #include <sys/priv.h> 87 88 #include <net/if.h> 89 #include <net/if_types.h> 90 #include <net/if_dl.h> 91 #include <net/route.h> 92 #include <net/netisr.h> 93 #include <net/pfil.h> 94 95 #include <sys/thread2.h> 96 #include <sys/msgport2.h> 97 #include <net/netmsg2.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 #include <net/net_osdep.h> 133 134 extern struct domain inet6domain; 135 extern struct protosw inet6sw[]; 136 137 u_char ip6_protox[IPPROTO_MAX]; 138 struct in6_ifaddr *in6_ifaddr; 139 140 extern struct callout in6_tmpaddrtimer_ch; 141 142 int ip6_forward_srcrt; /* XXX */ 143 int ip6_sourcecheck; /* XXX */ 144 int ip6_sourcecheck_interval; /* XXX */ 145 146 int ip6_ours_check_algorithm; 147 148 struct pfil_head inet6_pfil_hook; 149 150 /* firewall hooks */ 151 ip6_fw_chk_t *ip6_fw_chk_ptr; 152 ip6_fw_ctl_t *ip6_fw_ctl_ptr; 153 int ip6_fw_enable = 1; 154 155 struct ip6stat ip6stat; 156 157 static void ip6_init2 (void *); 158 static struct ip6aux *ip6_setdstifaddr (struct mbuf *, struct in6_ifaddr *); 159 static int ip6_hopopts_input (u_int32_t *, u_int32_t *, struct mbuf **, int *); 160 static void ip6_input(netmsg_t msg); 161 #ifdef PULLDOWN_TEST 162 static struct mbuf *ip6_pullexthdr (struct mbuf *, size_t, int); 163 #endif 164 static void transport6_processing_handler(netmsg_t netmsg); 165 166 /* 167 * IP6 initialization: fill in IP6 protocol switch table. 168 * All protocols not implemented in kernel go to raw IP6 protocol handler. 169 */ 170 void 171 ip6_init(void) 172 { 173 struct protosw *pr; 174 int i; 175 struct timeval tv; 176 177 pr = pffindproto(PF_INET6, IPPROTO_RAW, SOCK_RAW); 178 if (pr == NULL) 179 panic("ip6_init"); 180 for (i = 0; i < IPPROTO_MAX; i++) 181 ip6_protox[i] = pr - inet6sw; 182 for (pr = inet6domain.dom_protosw; 183 pr < inet6domain.dom_protoswNPROTOSW; pr++) 184 if (pr->pr_domain->dom_family == PF_INET6 && 185 pr->pr_protocol && pr->pr_protocol != IPPROTO_RAW) 186 ip6_protox[pr->pr_protocol] = pr - inet6sw; 187 188 inet6_pfil_hook.ph_type = PFIL_TYPE_AF; 189 inet6_pfil_hook.ph_af = AF_INET6; 190 if ((i = pfil_head_register(&inet6_pfil_hook)) != 0) { 191 kprintf("%s: WARNING: unable to register pfil hook, " 192 "error %d\n", __func__, i); 193 } 194 195 netisr_register(NETISR_IPV6, ip6_input, NULL); /* XXX cpufn */ 196 scope6_init(); 197 addrsel_policy_init(); 198 nd6_init(); 199 frag6_init(); 200 /* 201 * in many cases, random() here does NOT return random number 202 * as initialization during bootstrap time occur in fixed order. 203 */ 204 microtime(&tv); 205 ip6_flow_seq = krandom() ^ tv.tv_usec; 206 microtime(&tv); 207 ip6_desync_factor = (krandom() ^ tv.tv_usec) % MAX_TEMP_DESYNC_FACTOR; 208 } 209 210 static void 211 ip6_init2(void *dummy) 212 { 213 /* nd6_timer_init */ 214 callout_init(&nd6_timer_ch); 215 callout_reset(&nd6_timer_ch, hz, nd6_timer, NULL); 216 217 /* router renumbering prefix list maintenance */ 218 callout_init(&in6_rr_timer_ch); 219 callout_reset(&in6_rr_timer_ch, hz, in6_rr_timer, NULL); 220 221 /* timer for regeneranation of temporary addresses randomize ID */ 222 callout_init(&in6_tmpaddrtimer_ch); 223 callout_reset(&in6_tmpaddrtimer_ch, 224 (ip6_temp_preferred_lifetime - ip6_desync_factor - 225 ip6_temp_regen_advance) * hz, 226 in6_tmpaddrtimer, NULL); 227 } 228 229 /* cheat */ 230 /* This must be after route_init(), which is now SI_ORDER_THIRD */ 231 SYSINIT(netinet6init2, SI_SUB_PROTO_DOMAIN, SI_ORDER_MIDDLE, ip6_init2, NULL); 232 233 extern struct route_in6 ip6_forward_rt; 234 235 static 236 void 237 ip6_input(netmsg_t msg) 238 { 239 struct mbuf *m = msg->packet.nm_packet; 240 struct ip6_hdr *ip6; 241 int off = sizeof(struct ip6_hdr), nest; 242 u_int32_t plen; 243 u_int32_t rtalert = ~0; 244 int nxt, ours = 0, rh_present = 0; 245 struct ifnet *deliverifp = NULL; 246 struct in6_addr odst; 247 int srcrt = 0; 248 249 #ifdef IPSEC 250 /* 251 * should the inner packet be considered authentic? 252 * see comment in ah4_input(). 253 */ 254 if (m) { 255 m->m_flags &= ~M_AUTHIPHDR; 256 m->m_flags &= ~M_AUTHIPDGM; 257 } 258 #endif 259 260 /* 261 * make sure we don't have onion peering information into m_aux. 262 */ 263 ip6_delaux(m); 264 265 /* 266 * mbuf statistics 267 */ 268 if (m->m_flags & M_EXT) { 269 if (m->m_next) 270 ip6stat.ip6s_mext2m++; 271 else 272 ip6stat.ip6s_mext1++; 273 } else { 274 #define M2MMAX NELEM(ip6stat.ip6s_m2m) 275 if (m->m_next) { 276 if (m->m_flags & M_LOOP) { 277 ip6stat.ip6s_m2m[loif[0].if_index]++; /* XXX */ 278 } else if (m->m_pkthdr.rcvif->if_index < M2MMAX) 279 ip6stat.ip6s_m2m[m->m_pkthdr.rcvif->if_index]++; 280 else 281 ip6stat.ip6s_m2m[0]++; 282 } else 283 ip6stat.ip6s_m1++; 284 #undef M2MMAX 285 } 286 287 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_receive); 288 ip6stat.ip6s_total++; 289 290 #ifndef PULLDOWN_TEST 291 /* 292 * L2 bridge code and some other code can return mbuf chain 293 * that does not conform to KAME requirement. too bad. 294 * XXX: fails to join if interface MTU > MCLBYTES. jumbogram? 295 */ 296 if (m && m->m_next != NULL && m->m_pkthdr.len < MCLBYTES) { 297 struct mbuf *n; 298 299 n = m_getb(m->m_pkthdr.len, MB_DONTWAIT, MT_HEADER, M_PKTHDR); 300 if (n == NULL) 301 goto bad; 302 M_MOVE_PKTHDR(n, m); 303 304 m_copydata(m, 0, n->m_pkthdr.len, mtod(n, caddr_t)); 305 n->m_len = n->m_pkthdr.len; 306 m_freem(m); 307 m = n; 308 } 309 IP6_EXTHDR_CHECK_VOIDRET(m, 0, sizeof(struct ip6_hdr)); 310 #endif 311 312 if (m->m_len < sizeof(struct ip6_hdr)) { 313 struct ifnet *inifp; 314 inifp = m->m_pkthdr.rcvif; 315 if ((m = m_pullup(m, sizeof(struct ip6_hdr))) == NULL) { 316 ip6stat.ip6s_toosmall++; 317 in6_ifstat_inc(inifp, ifs6_in_hdrerr); 318 goto bad2; 319 } 320 } 321 322 ip6 = mtod(m, struct ip6_hdr *); 323 324 if ((ip6->ip6_vfc & IPV6_VERSION_MASK) != IPV6_VERSION) { 325 ip6stat.ip6s_badvers++; 326 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_hdrerr); 327 goto bad; 328 } 329 330 /* 331 * Run through list of hooks for input packets. 332 * 333 * NB: Beware of the destination address changing 334 * (e.g. by NAT rewriting). When this happens, 335 * tell ip6_forward to do the right thing. 336 */ 337 if (pfil_has_hooks(&inet6_pfil_hook)) { 338 odst = ip6->ip6_dst; 339 if (pfil_run_hooks(&inet6_pfil_hook, &m, 340 m->m_pkthdr.rcvif, PFIL_IN)) { 341 goto bad2; 342 } 343 if (m == NULL) /* consumed by filter */ 344 goto bad2; 345 ip6 = mtod(m, struct ip6_hdr *); 346 srcrt = !IN6_ARE_ADDR_EQUAL(&odst, &ip6->ip6_dst); 347 } 348 349 ip6stat.ip6s_nxthist[ip6->ip6_nxt]++; 350 351 #ifdef ALTQ 352 if (altq_input != NULL && (*altq_input)(m, AF_INET6) == 0) { 353 /* packet is dropped by traffic conditioner */ 354 return; 355 } 356 #endif 357 358 /* 359 * Check with the firewall... 360 */ 361 if (ip6_fw_enable && ip6_fw_chk_ptr) { 362 u_short port = 0; 363 /* If ipfw says divert, we have to just drop packet */ 364 /* use port as a dummy argument */ 365 if ((*ip6_fw_chk_ptr)(&ip6, NULL, &port, &m)) { 366 m_freem(m); 367 m = NULL; 368 } 369 if (!m) 370 goto bad2; 371 } 372 373 /* 374 * Check against address spoofing/corruption. 375 */ 376 if (IN6_IS_ADDR_MULTICAST(&ip6->ip6_src) || 377 IN6_IS_ADDR_UNSPECIFIED(&ip6->ip6_dst)) { 378 /* 379 * XXX: "badscope" is not very suitable for a multicast source. 380 */ 381 ip6stat.ip6s_badscope++; 382 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_addrerr); 383 goto bad; 384 } 385 if ((IN6_IS_ADDR_LOOPBACK(&ip6->ip6_src) || 386 IN6_IS_ADDR_LOOPBACK(&ip6->ip6_dst)) && 387 (m->m_pkthdr.rcvif->if_flags & IFF_LOOPBACK) == 0) { 388 ip6stat.ip6s_badscope++; 389 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_addrerr); 390 goto bad; 391 } 392 393 /* 394 * The following check is not documented in specs. A malicious 395 * party may be able to use IPv4 mapped addr to confuse tcp/udp stack 396 * and bypass security checks (act as if it was from 127.0.0.1 by using 397 * IPv6 src ::ffff:127.0.0.1). Be cautious. 398 * 399 * This check chokes if we are in an SIIT cloud. As none of BSDs 400 * support IPv4-less kernel compilation, we cannot support SIIT 401 * environment at all. So, it makes more sense for us to reject any 402 * malicious packets for non-SIIT environment, than try to do a 403 * partical support for SIIT environment. 404 */ 405 if (IN6_IS_ADDR_V4MAPPED(&ip6->ip6_src) || 406 IN6_IS_ADDR_V4MAPPED(&ip6->ip6_dst)) { 407 ip6stat.ip6s_badscope++; 408 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_addrerr); 409 goto bad; 410 } 411 #if 0 412 /* 413 * Reject packets with IPv4 compatible addresses (auto tunnel). 414 * 415 * The code forbids auto tunnel relay case in RFC1933 (the check is 416 * stronger than RFC1933). We may want to re-enable it if mech-xx 417 * is revised to forbid relaying case. 418 */ 419 if (IN6_IS_ADDR_V4COMPAT(&ip6->ip6_src) || 420 IN6_IS_ADDR_V4COMPAT(&ip6->ip6_dst)) { 421 ip6stat.ip6s_badscope++; 422 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_addrerr); 423 goto bad; 424 } 425 #endif 426 427 /* drop packets if interface ID portion is already filled */ 428 if ((m->m_pkthdr.rcvif->if_flags & IFF_LOOPBACK) == 0) { 429 if (IN6_IS_SCOPE_LINKLOCAL(&ip6->ip6_src) && 430 ip6->ip6_src.s6_addr16[1]) { 431 ip6stat.ip6s_badscope++; 432 goto bad; 433 } 434 if (IN6_IS_SCOPE_LINKLOCAL(&ip6->ip6_dst) && 435 ip6->ip6_dst.s6_addr16[1]) { 436 ip6stat.ip6s_badscope++; 437 goto bad; 438 } 439 } 440 441 if (IN6_IS_SCOPE_LINKLOCAL(&ip6->ip6_src)) 442 ip6->ip6_src.s6_addr16[1] 443 = htons(m->m_pkthdr.rcvif->if_index); 444 if (IN6_IS_SCOPE_LINKLOCAL(&ip6->ip6_dst)) 445 ip6->ip6_dst.s6_addr16[1] 446 = htons(m->m_pkthdr.rcvif->if_index); 447 448 #if 0 /* this case seems to be unnecessary. (jinmei, 20010401) */ 449 /* 450 * We use rt->rt_ifp to determine if the address is ours or not. 451 * If rt_ifp is lo0, the address is ours. 452 * The problem here is, rt->rt_ifp for fe80::%lo0/64 is set to lo0, 453 * so any address under fe80::%lo0/64 will be mistakenly considered 454 * local. The special case is supplied to handle the case properly 455 * by actually looking at interface addresses 456 * (using in6ifa_ifpwithaddr). 457 */ 458 if ((m->m_pkthdr.rcvif->if_flags & IFF_LOOPBACK) && 459 IN6_IS_ADDR_LINKLOCAL(&ip6->ip6_dst)) { 460 if (!in6ifa_ifpwithaddr(m->m_pkthdr.rcvif, &ip6->ip6_dst)) { 461 icmp6_error(m, ICMP6_DST_UNREACH, 462 ICMP6_DST_UNREACH_ADDR, 0); 463 /* m is already freed */ 464 goto bad2; 465 } 466 467 ours = 1; 468 deliverifp = m->m_pkthdr.rcvif; 469 goto hbhcheck; 470 } 471 #endif 472 473 /* 474 * Multicast check 475 */ 476 if (IN6_IS_ADDR_MULTICAST(&ip6->ip6_dst)) { 477 struct in6_multi *in6m = NULL; 478 479 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_mcast); 480 /* 481 * See if we belong to the destination multicast group on the 482 * arrival interface. 483 */ 484 IN6_LOOKUP_MULTI(ip6->ip6_dst, m->m_pkthdr.rcvif, in6m); 485 if (in6m) 486 ours = 1; 487 else if (!ip6_mrouter) { 488 ip6stat.ip6s_notmember++; 489 ip6stat.ip6s_cantforward++; 490 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_discard); 491 goto bad; 492 } 493 deliverifp = m->m_pkthdr.rcvif; 494 goto hbhcheck; 495 } 496 497 /* 498 * Unicast check 499 */ 500 switch (ip6_ours_check_algorithm) { 501 default: 502 /* 503 * XXX: I intentionally broke our indentation rule here, 504 * since this switch-case is just for measurement and 505 * therefore should soon be removed. 506 */ 507 if (ip6_forward_rt.ro_rt != NULL && 508 (ip6_forward_rt.ro_rt->rt_flags & RTF_UP) && 509 IN6_ARE_ADDR_EQUAL(&ip6->ip6_dst, 510 &((struct sockaddr_in6 *)(&ip6_forward_rt.ro_dst))->sin6_addr)) 511 ip6stat.ip6s_forward_cachehit++; 512 else { 513 struct sockaddr_in6 *dst6; 514 515 if (ip6_forward_rt.ro_rt) { 516 /* route is down or destination is different */ 517 ip6stat.ip6s_forward_cachemiss++; 518 RTFREE(ip6_forward_rt.ro_rt); 519 ip6_forward_rt.ro_rt = 0; 520 } 521 522 bzero(&ip6_forward_rt.ro_dst, sizeof(struct sockaddr_in6)); 523 dst6 = &ip6_forward_rt.ro_dst; 524 dst6->sin6_len = sizeof(struct sockaddr_in6); 525 dst6->sin6_family = AF_INET6; 526 dst6->sin6_addr = ip6->ip6_dst; 527 528 rtalloc_ign((struct route *)&ip6_forward_rt, RTF_PRCLONING); 529 } 530 531 #define rt6_key(r) ((struct sockaddr_in6 *)((r)->rt_nodes->rn_key)) 532 533 /* 534 * Accept the packet if the forwarding interface to the destination 535 * according to the routing table is the loopback interface, 536 * unless the associated route has a gateway. 537 * Note that this approach causes to accept a packet if there is a 538 * route to the loopback interface for the destination of the packet. 539 * But we think it's even useful in some situations, e.g. when using 540 * a special daemon which wants to intercept the packet. 541 * 542 * XXX: some OSes automatically make a cloned route for the destination 543 * of an outgoing packet. If the outgoing interface of the packet 544 * is a loopback one, the kernel would consider the packet to be 545 * accepted, even if we have no such address assinged on the interface. 546 * We check the cloned flag of the route entry to reject such cases, 547 * assuming that route entries for our own addresses are not made by 548 * cloning (it should be true because in6_addloop explicitly installs 549 * the host route). However, we might have to do an explicit check 550 * while it would be less efficient. Or, should we rather install a 551 * reject route for such a case? 552 */ 553 if (ip6_forward_rt.ro_rt && 554 (ip6_forward_rt.ro_rt->rt_flags & 555 (RTF_HOST|RTF_GATEWAY)) == RTF_HOST && 556 #ifdef RTF_WASCLONED 557 !(ip6_forward_rt.ro_rt->rt_flags & RTF_WASCLONED) && 558 #endif 559 #ifdef RTF_CLONED 560 !(ip6_forward_rt.ro_rt->rt_flags & RTF_CLONED) && 561 #endif 562 #if 0 563 /* 564 * The check below is redundant since the comparison of 565 * the destination and the key of the rtentry has 566 * already done through looking up the routing table. 567 */ 568 IN6_ARE_ADDR_EQUAL(&ip6->ip6_dst, 569 &rt6_key(ip6_forward_rt.ro_rt)->sin6_addr) 570 #endif 571 ip6_forward_rt.ro_rt->rt_ifp->if_type == IFT_LOOP) { 572 struct in6_ifaddr *ia6 = 573 (struct in6_ifaddr *)ip6_forward_rt.ro_rt->rt_ifa; 574 575 /* 576 * record address information into m_aux. 577 */ 578 ip6_setdstifaddr(m, ia6); 579 580 /* 581 * packets to a tentative, duplicated, or somehow invalid 582 * address must not be accepted. 583 */ 584 if (!(ia6->ia6_flags & IN6_IFF_NOTREADY)) { 585 /* this address is ready */ 586 ours = 1; 587 deliverifp = ia6->ia_ifp; /* correct? */ 588 /* Count the packet in the ip address stats */ 589 IFA_STAT_INC(&ia6->ia_ifa, ipackets, 1); 590 IFA_STAT_INC(&ia6->ia_ifa, ibytes, m->m_pkthdr.len); 591 goto hbhcheck; 592 } else { 593 /* address is not ready, so discard the packet. */ 594 nd6log((LOG_INFO, 595 "ip6_input: packet to an unready address %s->%s\n", 596 ip6_sprintf(&ip6->ip6_src), 597 ip6_sprintf(&ip6->ip6_dst))); 598 599 goto bad; 600 } 601 } 602 } /* XXX indentation (see above) */ 603 604 /* 605 * FAITH(Firewall Aided Internet Translator) 606 */ 607 if (ip6_keepfaith) { 608 if (ip6_forward_rt.ro_rt && ip6_forward_rt.ro_rt->rt_ifp 609 && ip6_forward_rt.ro_rt->rt_ifp->if_type == IFT_FAITH) { 610 /* XXX do we need more sanity checks? */ 611 ours = 1; 612 deliverifp = ip6_forward_rt.ro_rt->rt_ifp; /* faith */ 613 goto hbhcheck; 614 } 615 } 616 617 /* 618 * Now there is no reason to process the packet if it's not our own 619 * and we're not a router. 620 */ 621 if (!ip6_forwarding) { 622 ip6stat.ip6s_cantforward++; 623 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_discard); 624 goto bad; 625 } 626 627 hbhcheck: 628 /* 629 * record address information into m_aux, if we don't have one yet. 630 * note that we are unable to record it, if the address is not listed 631 * as our interface address (e.g. multicast addresses, addresses 632 * within FAITH prefixes and such). 633 */ 634 if (deliverifp && !ip6_getdstifaddr(m)) { 635 struct in6_ifaddr *ia6; 636 637 ia6 = in6_ifawithifp(deliverifp, &ip6->ip6_dst); 638 if (ia6) { 639 if (!ip6_setdstifaddr(m, ia6)) { 640 /* 641 * XXX maybe we should drop the packet here, 642 * as we could not provide enough information 643 * to the upper layers. 644 */ 645 } 646 } 647 } 648 649 /* 650 * Process Hop-by-Hop options header if it's contained. 651 * m may be modified in ip6_hopopts_input(). 652 * If a JumboPayload option is included, plen will also be modified. 653 */ 654 plen = (u_int32_t)ntohs(ip6->ip6_plen); 655 if (ip6->ip6_nxt == IPPROTO_HOPOPTS) { 656 struct ip6_hbh *hbh; 657 658 if (ip6_hopopts_input(&plen, &rtalert, &m, &off)) { 659 #if 0 /*touches NULL pointer*/ 660 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_discard); 661 #endif 662 goto bad2; /* m have already been freed */ 663 } 664 665 /* adjust pointer */ 666 ip6 = mtod(m, struct ip6_hdr *); 667 668 /* 669 * if the payload length field is 0 and the next header field 670 * indicates Hop-by-Hop Options header, then a Jumbo Payload 671 * option MUST be included. 672 */ 673 if (ip6->ip6_plen == 0 && plen == 0) { 674 /* 675 * Note that if a valid jumbo payload option is 676 * contained, ip6_hoptops_input() must set a valid 677 * (non-zero) payload length to the variable plen. 678 */ 679 ip6stat.ip6s_badoptions++; 680 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_discard); 681 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_hdrerr); 682 icmp6_error(m, ICMP6_PARAM_PROB, 683 ICMP6_PARAMPROB_HEADER, 684 (caddr_t)&ip6->ip6_plen - (caddr_t)ip6); 685 goto bad2; 686 } 687 #ifndef PULLDOWN_TEST 688 /* ip6_hopopts_input() ensures that mbuf is contiguous */ 689 hbh = (struct ip6_hbh *)(ip6 + 1); 690 #else 691 IP6_EXTHDR_GET(hbh, struct ip6_hbh *, m, sizeof(struct ip6_hdr), 692 sizeof(struct ip6_hbh)); 693 if (hbh == NULL) { 694 ip6stat.ip6s_tooshort++; 695 goto bad2; 696 } 697 #endif 698 nxt = hbh->ip6h_nxt; 699 700 /* 701 * If we are acting as a router and the packet contains a 702 * router alert option, see if we know the option value. 703 * Currently, we only support the option value for MLD, in which 704 * case we should pass the packet to the multicast routing 705 * daemon. 706 */ 707 if (rtalert != ~0 && ip6_forwarding) { 708 switch (rtalert) { 709 case IP6OPT_RTALERT_MLD: 710 ours = 1; 711 break; 712 default: 713 /* 714 * RFC2711 requires unrecognized values must be 715 * silently ignored. 716 */ 717 break; 718 } 719 } 720 } else 721 nxt = ip6->ip6_nxt; 722 723 /* 724 * Check that the amount of data in the buffers 725 * is as at least much as the IPv6 header would have us expect. 726 * Trim mbufs if longer than we expect. 727 * Drop packet if shorter than we expect. 728 */ 729 if (m->m_pkthdr.len - sizeof(struct ip6_hdr) < plen) { 730 ip6stat.ip6s_tooshort++; 731 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_truncated); 732 goto bad; 733 } 734 if (m->m_pkthdr.len > sizeof(struct ip6_hdr) + plen) { 735 if (m->m_len == m->m_pkthdr.len) { 736 m->m_len = sizeof(struct ip6_hdr) + plen; 737 m->m_pkthdr.len = sizeof(struct ip6_hdr) + plen; 738 } else 739 m_adj(m, sizeof(struct ip6_hdr) + plen - m->m_pkthdr.len); 740 } 741 742 /* 743 * Forward if desirable. 744 */ 745 if (IN6_IS_ADDR_MULTICAST(&ip6->ip6_dst)) { 746 /* 747 * If we are acting as a multicast router, all 748 * incoming multicast packets are passed to the 749 * kernel-level multicast forwarding function. 750 * The packet is returned (relatively) intact; if 751 * ip6_mforward() returns a non-zero value, the packet 752 * must be discarded, else it may be accepted below. 753 */ 754 if (ip6_mrouter && ip6_mforward(ip6, m->m_pkthdr.rcvif, m)) { 755 ip6stat.ip6s_cantforward++; 756 goto bad; 757 } 758 if (!ours) 759 goto bad; 760 } else if (!ours) { 761 ip6_forward(m, srcrt); 762 goto bad2; 763 } 764 765 ip6 = mtod(m, struct ip6_hdr *); 766 767 /* 768 * Malicious party may be able to use IPv4 mapped addr to confuse 769 * tcp/udp stack and bypass security checks (act as if it was from 770 * 127.0.0.1 by using IPv6 src ::ffff:127.0.0.1). Be cautious. 771 * 772 * For SIIT end node behavior, you may want to disable the check. 773 * However, you will become vulnerable to attacks using IPv4 mapped 774 * source. 775 */ 776 if (IN6_IS_ADDR_V4MAPPED(&ip6->ip6_src) || 777 IN6_IS_ADDR_V4MAPPED(&ip6->ip6_dst)) { 778 ip6stat.ip6s_badscope++; 779 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_addrerr); 780 goto bad; 781 } 782 783 /* 784 * Tell launch routine the next header 785 */ 786 ip6stat.ip6s_delivered++; 787 in6_ifstat_inc(deliverifp, ifs6_in_deliver); 788 nest = 0; 789 790 rh_present = 0; 791 while (nxt != IPPROTO_DONE) { 792 struct protosw *sw6; 793 794 if (ip6_hdrnestlimit && (++nest > ip6_hdrnestlimit)) { 795 ip6stat.ip6s_toomanyhdr++; 796 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_hdrerr); 797 goto bad; 798 } 799 800 /* 801 * protection against faulty packet - there should be 802 * more sanity checks in header chain processing. 803 */ 804 if (m->m_pkthdr.len < off) { 805 ip6stat.ip6s_tooshort++; 806 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_truncated); 807 goto bad; 808 } 809 810 #if 0 811 /* 812 * do we need to do it for every header? yeah, other 813 * functions can play with it (like re-allocate and copy). 814 */ 815 mhist = ip6_addaux(m); 816 if (mhist && M_TRAILINGSPACE(mhist) >= sizeof(nxt)) { 817 hist = mtod(mhist, caddr_t) + mhist->m_len; 818 bcopy(&nxt, hist, sizeof(nxt)); 819 mhist->m_len += sizeof(nxt); 820 } else { 821 ip6stat.ip6s_toomanyhdr++; 822 goto bad; 823 } 824 #endif 825 826 if (nxt == IPPROTO_ROUTING) { 827 if (rh_present++) { 828 in6_ifstat_inc(m->m_pkthdr.rcvif, 829 ifs6_in_hdrerr); 830 ip6stat.ip6s_badoptions++; 831 goto bad; 832 } 833 } 834 835 sw6 = &inet6sw[ip6_protox[nxt]]; 836 #ifdef IPSEC 837 /* 838 * enforce IPsec policy checking if we are seeing last header. 839 * note that we do not visit this with protocols with pcb layer 840 * code - like udp/tcp/raw ip. 841 */ 842 if ((sw6->pr_flags & PR_LASTHDR) && ipsec6_in_reject(m, NULL)) { 843 ipsec6stat.in_polvio++; 844 goto bad; 845 } 846 #endif 847 /* 848 * If this is a terminal header forward to the port, otherwise 849 * process synchronously for more headers. 850 */ 851 if (sw6->pr_flags & PR_LASTHDR) { 852 struct netmsg_packet *pmsg; 853 lwkt_port_t port; 854 855 port = netisr_portfn(0); /* XXX */ 856 KKASSERT(port != NULL); 857 pmsg = &m->m_hdr.mh_netmsg; 858 netmsg_init(&pmsg->base, NULL, 859 &netisr_apanic_rport, 860 0, transport6_processing_handler); 861 pmsg->nm_packet = m; 862 pmsg->nm_nxt = nxt; 863 pmsg->base.lmsg.u.ms_result = off; 864 lwkt_sendmsg(port, &pmsg->base.lmsg); 865 /* done with m */ 866 nxt = IPPROTO_DONE; 867 } else { 868 nxt = sw6->pr_input(&m, &off, nxt); 869 } 870 } 871 goto bad2; 872 bad: 873 m_freem(m); 874 bad2: 875 ; 876 /* msg was embedded in the mbuf, do not reply! */ 877 } 878 879 /* 880 * We have to call the pr_input() function from the correct protocol 881 * thread. The sw6->pr_soport() request at the end of ip6_input() 882 * returns the port and we forward a netmsg to the port to execute 883 * this function. 884 */ 885 static void 886 transport6_processing_handler(netmsg_t netmsg) 887 { 888 struct netmsg_packet *pmsg = (struct netmsg_packet *)netmsg; 889 struct protosw *sw6; 890 int hlen; 891 int nxt; 892 893 sw6 = &inet6sw[ip6_protox[pmsg->nm_nxt]]; 894 hlen = pmsg->base.lmsg.u.ms_result; 895 896 nxt = sw6->pr_input(&pmsg->nm_packet, &hlen, pmsg->nm_nxt); 897 KKASSERT(nxt == IPPROTO_DONE); 898 /* netmsg was embedded in the mbuf, do not reply! */ 899 } 900 901 /* 902 * set/grab in6_ifaddr correspond to IPv6 destination address. 903 * XXX backward compatibility wrapper 904 */ 905 static struct ip6aux * 906 ip6_setdstifaddr(struct mbuf *m, struct in6_ifaddr *ia6) 907 { 908 struct ip6aux *n; 909 910 n = ip6_addaux(m); 911 if (n) 912 n->ip6a_dstia6 = ia6; 913 return n; /* NULL if failed to set */ 914 } 915 916 struct in6_ifaddr * 917 ip6_getdstifaddr(struct mbuf *m) 918 { 919 struct ip6aux *n; 920 921 n = ip6_findaux(m); 922 if (n) 923 return n->ip6a_dstia6; 924 else 925 return NULL; 926 } 927 928 /* 929 * Hop-by-Hop options header processing. If a valid jumbo payload option is 930 * included, the real payload length will be stored in plenp. 931 */ 932 static int 933 ip6_hopopts_input(u_int32_t *plenp, 934 u_int32_t *rtalertp,/* XXX: should be stored more smart way */ 935 struct mbuf **mp, 936 int *offp) 937 { 938 struct mbuf *m = *mp; 939 int off = *offp, hbhlen; 940 struct ip6_hbh *hbh; 941 942 /* validation of the length of the header */ 943 #ifndef PULLDOWN_TEST 944 IP6_EXTHDR_CHECK(m, off, sizeof(*hbh), -1); 945 hbh = (struct ip6_hbh *)(mtod(m, caddr_t) + off); 946 hbhlen = (hbh->ip6h_len + 1) << 3; 947 948 IP6_EXTHDR_CHECK(m, off, hbhlen, -1); 949 hbh = (struct ip6_hbh *)(mtod(m, caddr_t) + off); 950 #else 951 IP6_EXTHDR_GET(hbh, struct ip6_hbh *, m, 952 sizeof(struct ip6_hdr), sizeof(struct ip6_hbh)); 953 if (hbh == NULL) { 954 ip6stat.ip6s_tooshort++; 955 return -1; 956 } 957 hbhlen = (hbh->ip6h_len + 1) << 3; 958 IP6_EXTHDR_GET(hbh, struct ip6_hbh *, m, sizeof(struct ip6_hdr), 959 hbhlen); 960 if (hbh == NULL) { 961 ip6stat.ip6s_tooshort++; 962 return -1; 963 } 964 #endif 965 off += hbhlen; 966 hbhlen -= sizeof(struct ip6_hbh); 967 968 if (ip6_process_hopopts(m, (u_int8_t *)hbh + sizeof(struct ip6_hbh), 969 hbhlen, rtalertp, plenp) < 0) 970 return (-1); 971 972 *offp = off; 973 *mp = m; 974 return (0); 975 } 976 977 /* 978 * Search header for all Hop-by-hop options and process each option. 979 * This function is separate from ip6_hopopts_input() in order to 980 * handle a case where the sending node itself process its hop-by-hop 981 * options header. In such a case, the function is called from ip6_output(). 982 * 983 * The function assumes that hbh header is located right after the IPv6 header 984 * (RFC2460 p7), opthead is pointer into data content in m, and opthead to 985 * opthead + hbhlen is located in continuous memory region. 986 */ 987 int 988 ip6_process_hopopts(struct mbuf *m, u_int8_t *opthead, int hbhlen, 989 u_int32_t *rtalertp, u_int32_t *plenp) 990 { 991 struct ip6_hdr *ip6; 992 int optlen = 0; 993 u_int8_t *opt = opthead; 994 u_int16_t rtalert_val; 995 u_int32_t jumboplen; 996 const int erroff = sizeof(struct ip6_hdr) + sizeof(struct ip6_hbh); 997 998 for (; hbhlen > 0; hbhlen -= optlen, opt += optlen) { 999 switch (*opt) { 1000 case IP6OPT_PAD1: 1001 optlen = 1; 1002 break; 1003 case IP6OPT_PADN: 1004 if (hbhlen < IP6OPT_MINLEN) { 1005 ip6stat.ip6s_toosmall++; 1006 goto bad; 1007 } 1008 optlen = *(opt + 1) + 2; 1009 break; 1010 case IP6OPT_RTALERT: 1011 /* XXX may need check for alignment */ 1012 if (hbhlen < IP6OPT_RTALERT_LEN) { 1013 ip6stat.ip6s_toosmall++; 1014 goto bad; 1015 } 1016 if (*(opt + 1) != IP6OPT_RTALERT_LEN - 2) { 1017 /* XXX stat */ 1018 icmp6_error(m, ICMP6_PARAM_PROB, 1019 ICMP6_PARAMPROB_HEADER, 1020 erroff + opt + 1 - opthead); 1021 return (-1); 1022 } 1023 optlen = IP6OPT_RTALERT_LEN; 1024 bcopy((caddr_t)(opt + 2), (caddr_t)&rtalert_val, 2); 1025 *rtalertp = ntohs(rtalert_val); 1026 break; 1027 case IP6OPT_JUMBO: 1028 /* XXX may need check for alignment */ 1029 if (hbhlen < IP6OPT_JUMBO_LEN) { 1030 ip6stat.ip6s_toosmall++; 1031 goto bad; 1032 } 1033 if (*(opt + 1) != IP6OPT_JUMBO_LEN - 2) { 1034 /* XXX stat */ 1035 icmp6_error(m, ICMP6_PARAM_PROB, 1036 ICMP6_PARAMPROB_HEADER, 1037 erroff + opt + 1 - opthead); 1038 return (-1); 1039 } 1040 optlen = IP6OPT_JUMBO_LEN; 1041 1042 /* 1043 * IPv6 packets that have non 0 payload length 1044 * must not contain a jumbo payload option. 1045 */ 1046 ip6 = mtod(m, struct ip6_hdr *); 1047 if (ip6->ip6_plen) { 1048 ip6stat.ip6s_badoptions++; 1049 icmp6_error(m, ICMP6_PARAM_PROB, 1050 ICMP6_PARAMPROB_HEADER, 1051 erroff + opt - opthead); 1052 return (-1); 1053 } 1054 1055 /* 1056 * We may see jumbolen in unaligned location, so 1057 * we'd need to perform bcopy(). 1058 */ 1059 bcopy(opt + 2, &jumboplen, sizeof(jumboplen)); 1060 jumboplen = (u_int32_t)htonl(jumboplen); 1061 1062 #if 1 1063 /* 1064 * if there are multiple jumbo payload options, 1065 * *plenp will be non-zero and the packet will be 1066 * rejected. 1067 * the behavior may need some debate in ipngwg - 1068 * multiple options does not make sense, however, 1069 * there's no explicit mention in specification. 1070 */ 1071 if (*plenp != 0) { 1072 ip6stat.ip6s_badoptions++; 1073 icmp6_error(m, ICMP6_PARAM_PROB, 1074 ICMP6_PARAMPROB_HEADER, 1075 erroff + opt + 2 - opthead); 1076 return (-1); 1077 } 1078 #endif 1079 1080 /* 1081 * jumbo payload length must be larger than 65535. 1082 */ 1083 if (jumboplen <= IPV6_MAXPACKET) { 1084 ip6stat.ip6s_badoptions++; 1085 icmp6_error(m, ICMP6_PARAM_PROB, 1086 ICMP6_PARAMPROB_HEADER, 1087 erroff + opt + 2 - opthead); 1088 return (-1); 1089 } 1090 *plenp = jumboplen; 1091 1092 break; 1093 default: /* unknown option */ 1094 if (hbhlen < IP6OPT_MINLEN) { 1095 ip6stat.ip6s_toosmall++; 1096 goto bad; 1097 } 1098 optlen = ip6_unknown_opt(opt, m, 1099 erroff + opt - opthead); 1100 if (optlen == -1) 1101 return (-1); 1102 optlen += 2; 1103 break; 1104 } 1105 } 1106 1107 return (0); 1108 1109 bad: 1110 m_freem(m); 1111 return (-1); 1112 } 1113 1114 /* 1115 * Unknown option processing. 1116 * The third argument `off' is the offset from the IPv6 header to the option, 1117 * which is necessary if the IPv6 header the and option header and IPv6 header 1118 * is not continuous in order to return an ICMPv6 error. 1119 */ 1120 int 1121 ip6_unknown_opt(u_int8_t *optp, struct mbuf *m, int off) 1122 { 1123 struct ip6_hdr *ip6; 1124 1125 switch (IP6OPT_TYPE(*optp)) { 1126 case IP6OPT_TYPE_SKIP: /* ignore the option */ 1127 return ((int)*(optp + 1)); 1128 case IP6OPT_TYPE_DISCARD: /* silently discard */ 1129 m_freem(m); 1130 return (-1); 1131 case IP6OPT_TYPE_FORCEICMP: /* send ICMP even if multicasted */ 1132 ip6stat.ip6s_badoptions++; 1133 icmp6_error(m, ICMP6_PARAM_PROB, ICMP6_PARAMPROB_OPTION, off); 1134 return (-1); 1135 case IP6OPT_TYPE_ICMP: /* send ICMP if not multicasted */ 1136 ip6stat.ip6s_badoptions++; 1137 ip6 = mtod(m, struct ip6_hdr *); 1138 if (IN6_IS_ADDR_MULTICAST(&ip6->ip6_dst) || 1139 (m->m_flags & (M_BCAST|M_MCAST))) 1140 m_freem(m); 1141 else 1142 icmp6_error(m, ICMP6_PARAM_PROB, 1143 ICMP6_PARAMPROB_OPTION, off); 1144 return (-1); 1145 } 1146 1147 m_freem(m); /* XXX: NOTREACHED */ 1148 return (-1); 1149 } 1150 1151 /* 1152 * Create the "control" list for this pcb. 1153 * The function will not modify mbuf chain at all. 1154 * 1155 * with KAME mbuf chain restriction: 1156 * The routine will be called from upper layer handlers like tcp6_input(). 1157 * Thus the routine assumes that the caller (tcp6_input) have already 1158 * called IP6_EXTHDR_CHECK() and all the extension headers are located in the 1159 * very first mbuf on the mbuf chain. 1160 */ 1161 void 1162 ip6_savecontrol(struct inpcb *in6p, struct mbuf **mp, struct ip6_hdr *ip6, 1163 struct mbuf *m) 1164 { 1165 #define IS2292(x, y) ((in6p->in6p_flags & IN6P_RFC2292) ? (x) : (y)) 1166 struct thread *td = curthread; /* XXX */ 1167 int privileged = 0; 1168 1169 1170 if (priv_check(td, PRIV_ROOT) == 0) 1171 privileged++; 1172 1173 #ifdef SO_TIMESTAMP 1174 if (in6p->in6p_socket->so_options & SO_TIMESTAMP) { 1175 struct timeval tv; 1176 1177 microtime(&tv); 1178 *mp = sbcreatecontrol((caddr_t) &tv, sizeof(tv), 1179 SCM_TIMESTAMP, SOL_SOCKET); 1180 if (*mp) { 1181 mp = &(*mp)->m_next; 1182 } 1183 } 1184 #endif 1185 1186 /* RFC 2292 sec. 5 */ 1187 if (in6p->in6p_flags & IN6P_PKTINFO) { 1188 struct in6_pktinfo pi6; 1189 bcopy(&ip6->ip6_dst, &pi6.ipi6_addr, sizeof(struct in6_addr)); 1190 if (IN6_IS_SCOPE_LINKLOCAL(&pi6.ipi6_addr)) 1191 pi6.ipi6_addr.s6_addr16[1] = 0; 1192 pi6.ipi6_ifindex = (m && m->m_pkthdr.rcvif) 1193 ? m->m_pkthdr.rcvif->if_index 1194 : 0; 1195 *mp = sbcreatecontrol((caddr_t) &pi6, 1196 sizeof(struct in6_pktinfo), IS2292(IPV6_2292PKTINFO, IPV6_PKTINFO), 1197 IPPROTO_IPV6); 1198 if (*mp) 1199 mp = &(*mp)->m_next; 1200 } 1201 1202 if (in6p->in6p_flags & IN6P_HOPLIMIT) { 1203 int hlim = ip6->ip6_hlim & 0xff; 1204 *mp = sbcreatecontrol((caddr_t) &hlim, 1205 sizeof(int), IS2292(IPV6_2292HOPLIMIT, IPV6_HOPLIMIT), IPPROTO_IPV6); 1206 if (*mp) 1207 mp = &(*mp)->m_next; 1208 } 1209 1210 if ((in6p->in6p_flags & IN6P_TCLASS) != 0) { 1211 u_int32_t flowinfo; 1212 int tclass; 1213 1214 flowinfo = (u_int32_t)ntohl(ip6->ip6_flow & IPV6_FLOWINFO_MASK); 1215 flowinfo >>= 20; 1216 1217 tclass = flowinfo & 0xff; 1218 *mp = sbcreatecontrol((caddr_t) &tclass, sizeof(tclass), 1219 IPV6_TCLASS, IPPROTO_IPV6); 1220 if (*mp) 1221 mp = &(*mp)->m_next; 1222 } 1223 1224 /* 1225 * IPV6_HOPOPTS socket option. We require super-user privilege 1226 * for the option, but it might be too strict, since there might 1227 * be some hop-by-hop options which can be returned to normal user. 1228 * See RFC 2292 section 6. 1229 */ 1230 if ((in6p->in6p_flags & IN6P_HOPOPTS) && privileged) { 1231 /* 1232 * Check if a hop-by-hop options header is contatined in the 1233 * received packet, and if so, store the options as ancillary 1234 * data. Note that a hop-by-hop options header must be 1235 * just after the IPv6 header, which fact is assured through 1236 * the IPv6 input processing. 1237 */ 1238 struct ip6_hdr *ip6 = mtod(m, struct ip6_hdr *); 1239 if (ip6->ip6_nxt == IPPROTO_HOPOPTS) { 1240 struct ip6_hbh *hbh; 1241 int hbhlen = 0; 1242 #ifdef PULLDOWN_TEST 1243 struct mbuf *ext; 1244 #endif 1245 1246 #ifndef PULLDOWN_TEST 1247 hbh = (struct ip6_hbh *)(ip6 + 1); 1248 hbhlen = (hbh->ip6h_len + 1) << 3; 1249 #else 1250 ext = ip6_pullexthdr(m, sizeof(struct ip6_hdr), 1251 ip6->ip6_nxt); 1252 if (ext == NULL) { 1253 ip6stat.ip6s_tooshort++; 1254 return; 1255 } 1256 hbh = mtod(ext, struct ip6_hbh *); 1257 hbhlen = (hbh->ip6h_len + 1) << 3; 1258 if (hbhlen != ext->m_len) { 1259 m_freem(ext); 1260 ip6stat.ip6s_tooshort++; 1261 return; 1262 } 1263 #endif 1264 1265 /* 1266 * XXX: We copy whole the header even if a jumbo 1267 * payload option is included, which option is to 1268 * be removed before returning in the RFC 2292. 1269 * Note: this constraint is removed in 2292bis. 1270 */ 1271 *mp = sbcreatecontrol((caddr_t)hbh, hbhlen, 1272 IS2292(IPV6_2292HOPOPTS, IPV6_HOPOPTS), IPPROTO_IPV6); 1273 if (*mp) 1274 mp = &(*mp)->m_next; 1275 #ifdef PULLDOWN_TEST 1276 m_freem(ext); 1277 #endif 1278 } 1279 } 1280 1281 /* IPV6_DSTOPTS and IPV6_RTHDR socket options */ 1282 if ((in6p->in6p_flags & (IN6P_DSTOPTS | IN6P_RTHDRDSTOPTS)) != 0) { 1283 int proto, off, nxt; 1284 1285 /* 1286 * go through the header chain to see if a routing header is 1287 * contained in the packet. We need this information to store 1288 * destination options headers (if any) properly. 1289 * XXX: performance issue. We should record this info when 1290 * processing extension headers in incoming routine. 1291 * (todo) use m_aux? 1292 */ 1293 proto = IPPROTO_IPV6; 1294 off = 0; 1295 nxt = -1; 1296 while (1) { 1297 int newoff; 1298 1299 newoff = ip6_nexthdr(m, off, proto, &nxt); 1300 if (newoff < 0) 1301 break; 1302 if (newoff < off) /* invalid, check for safety */ 1303 break; 1304 if ((proto = nxt) == IPPROTO_ROUTING) 1305 break; 1306 off = newoff; 1307 } 1308 } 1309 1310 if ((in6p->in6p_flags & 1311 (IN6P_RTHDR | IN6P_DSTOPTS | IN6P_RTHDRDSTOPTS)) != 0) { 1312 struct ip6_hdr *ip6 = mtod(m, struct ip6_hdr *); 1313 int nxt = ip6->ip6_nxt, off = sizeof(struct ip6_hdr); 1314 1315 /* 1316 * Search for destination options headers or routing 1317 * header(s) through the header chain, and stores each 1318 * header as ancillary data. 1319 * Note that the order of the headers remains in 1320 * the chain of ancillary data. 1321 */ 1322 while (1) { /* is explicit loop prevention necessary? */ 1323 struct ip6_ext *ip6e = NULL; 1324 int elen; 1325 #ifdef PULLDOWN_TEST 1326 struct mbuf *ext = NULL; 1327 #endif 1328 1329 /* 1330 * if it is not an extension header, don't try to 1331 * pull it from the chain. 1332 */ 1333 switch (nxt) { 1334 case IPPROTO_DSTOPTS: 1335 case IPPROTO_ROUTING: 1336 case IPPROTO_HOPOPTS: 1337 case IPPROTO_AH: /* is it possible? */ 1338 break; 1339 default: 1340 goto loopend; 1341 } 1342 1343 #ifndef PULLDOWN_TEST 1344 if (off + sizeof(*ip6e) > m->m_len) 1345 goto loopend; 1346 ip6e = (struct ip6_ext *)(mtod(m, caddr_t) + off); 1347 if (nxt == IPPROTO_AH) 1348 elen = (ip6e->ip6e_len + 2) << 2; 1349 else 1350 elen = (ip6e->ip6e_len + 1) << 3; 1351 if (off + elen > m->m_len) 1352 goto loopend; 1353 #else 1354 ext = ip6_pullexthdr(m, off, nxt); 1355 if (ext == NULL) { 1356 ip6stat.ip6s_tooshort++; 1357 return; 1358 } 1359 ip6e = mtod(ext, struct ip6_ext *); 1360 if (nxt == IPPROTO_AH) 1361 elen = (ip6e->ip6e_len + 2) << 2; 1362 else 1363 elen = (ip6e->ip6e_len + 1) << 3; 1364 if (elen != ext->m_len) { 1365 m_freem(ext); 1366 ip6stat.ip6s_tooshort++; 1367 return; 1368 } 1369 #endif 1370 1371 switch (nxt) { 1372 case IPPROTO_DSTOPTS: 1373 if ((in6p->in6p_flags & IN6P_DSTOPTS) == 0) 1374 break; 1375 1376 /* 1377 * We also require super-user privilege for 1378 * the option. 1379 * See the comments on IN6_HOPOPTS. 1380 */ 1381 if (!privileged) 1382 break; 1383 1384 *mp = sbcreatecontrol((caddr_t)ip6e, elen, 1385 IS2292(IPV6_2292DSTOPTS, IPV6_DSTOPTS), 1386 IPPROTO_IPV6); 1387 if (*mp) 1388 mp = &(*mp)->m_next; 1389 break; 1390 case IPPROTO_ROUTING: 1391 if (!(in6p->in6p_flags & IN6P_RTHDR)) 1392 break; 1393 1394 *mp = sbcreatecontrol((caddr_t)ip6e, elen, 1395 IS2292(IPV6_2292RTHDR, IPV6_RTHDR), 1396 IPPROTO_IPV6); 1397 if (*mp) 1398 mp = &(*mp)->m_next; 1399 break; 1400 case IPPROTO_HOPOPTS: 1401 case IPPROTO_AH: /* is it possible? */ 1402 break; 1403 1404 default: 1405 /* 1406 * other cases have been filtered in the above. 1407 * none will visit this case. here we supply 1408 * the code just in case (nxt overwritten or 1409 * other cases). 1410 */ 1411 #ifdef PULLDOWN_TEST 1412 m_freem(ext); 1413 #endif 1414 goto loopend; 1415 1416 } 1417 1418 /* proceed with the next header. */ 1419 off += elen; 1420 nxt = ip6e->ip6e_nxt; 1421 ip6e = NULL; 1422 #ifdef PULLDOWN_TEST 1423 m_freem(ext); 1424 ext = NULL; 1425 #endif 1426 } 1427 loopend: 1428 ; 1429 } 1430 #undef IS2292 1431 } 1432 1433 void 1434 ip6_notify_pmtu(struct inpcb *in6p, struct sockaddr_in6 *dst, u_int32_t *mtu) 1435 { 1436 struct socket *so; 1437 struct mbuf *m_mtu; 1438 struct ip6_mtuinfo mtuctl; 1439 1440 so = in6p->inp_socket; 1441 1442 if (mtu == NULL) 1443 return; 1444 1445 #ifdef DIAGNOSTIC 1446 if (so == NULL) /* I believe this is impossible */ 1447 panic("ip6_notify_pmtu: socket is NULL"); 1448 #endif 1449 1450 bzero(&mtuctl, sizeof(mtuctl)); /* zero-clear for safety */ 1451 mtuctl.ip6m_mtu = *mtu; 1452 mtuctl.ip6m_addr = *dst; 1453 1454 if ((m_mtu = sbcreatecontrol((caddr_t)&mtuctl, sizeof(mtuctl), 1455 IPV6_PATHMTU, IPPROTO_IPV6)) == NULL) 1456 return; 1457 1458 lwkt_gettoken(&so->so_rcv.ssb_token); 1459 if (sbappendaddr(&so->so_rcv.sb, (struct sockaddr *)dst, NULL, m_mtu) 1460 == 0) { 1461 m_freem(m_mtu); 1462 /* XXX: should count statistics */ 1463 } else { 1464 sorwakeup(so); 1465 } 1466 lwkt_reltoken(&so->so_rcv.ssb_token); 1467 } 1468 1469 #ifdef PULLDOWN_TEST 1470 /* 1471 * pull single extension header from mbuf chain. returns single mbuf that 1472 * contains the result, or NULL on error. 1473 */ 1474 static struct mbuf * 1475 ip6_pullexthdr(struct mbuf *m, size_t off, int nxt) 1476 { 1477 struct ip6_ext ip6e; 1478 size_t elen; 1479 struct mbuf *n; 1480 1481 #ifdef DIAGNOSTIC 1482 switch (nxt) { 1483 case IPPROTO_DSTOPTS: 1484 case IPPROTO_ROUTING: 1485 case IPPROTO_HOPOPTS: 1486 case IPPROTO_AH: /* is it possible? */ 1487 break; 1488 default: 1489 kprintf("ip6_pullexthdr: invalid nxt=%d\n", nxt); 1490 } 1491 #endif 1492 1493 m_copydata(m, off, sizeof(ip6e), (caddr_t)&ip6e); 1494 if (nxt == IPPROTO_AH) 1495 elen = (ip6e.ip6e_len + 2) << 2; 1496 else 1497 elen = (ip6e.ip6e_len + 1) << 3; 1498 1499 n = m_getb(elen, MB_DONTWAIT, MT_DATA, 0); 1500 if (n == NULL) 1501 return NULL; 1502 n->m_len = 0; 1503 1504 if (elen >= M_TRAILINGSPACE(n)) { 1505 m_free(n); 1506 return NULL; 1507 } 1508 1509 m_copydata(m, off, elen, mtod(n, caddr_t)); 1510 n->m_len = elen; 1511 return n; 1512 } 1513 #endif 1514 1515 /* 1516 * Get pointer to the previous header followed by the header 1517 * currently processed. 1518 * XXX: This function supposes that 1519 * M includes all headers, 1520 * the next header field and the header length field of each header 1521 * are valid, and 1522 * the sum of each header length equals to OFF. 1523 * Because of these assumptions, this function must be called very 1524 * carefully. Moreover, it will not be used in the near future when 1525 * we develop `neater' mechanism to process extension headers. 1526 */ 1527 char * 1528 ip6_get_prevhdr(struct mbuf *m, int off) 1529 { 1530 struct ip6_hdr *ip6 = mtod(m, struct ip6_hdr *); 1531 1532 if (off == sizeof(struct ip6_hdr)) 1533 return (&ip6->ip6_nxt); 1534 else { 1535 int len, nxt; 1536 struct ip6_ext *ip6e = NULL; 1537 1538 nxt = ip6->ip6_nxt; 1539 len = sizeof(struct ip6_hdr); 1540 while (len < off) { 1541 ip6e = (struct ip6_ext *)(mtod(m, caddr_t) + len); 1542 1543 switch (nxt) { 1544 case IPPROTO_FRAGMENT: 1545 len += sizeof(struct ip6_frag); 1546 break; 1547 case IPPROTO_AH: 1548 len += (ip6e->ip6e_len + 2) << 2; 1549 break; 1550 default: 1551 len += (ip6e->ip6e_len + 1) << 3; 1552 break; 1553 } 1554 nxt = ip6e->ip6e_nxt; 1555 } 1556 if (ip6e) 1557 return (&ip6e->ip6e_nxt); 1558 else 1559 return NULL; 1560 } 1561 } 1562 1563 /* 1564 * get next header offset. m will be retained. 1565 */ 1566 int 1567 ip6_nexthdr(struct mbuf *m, int off, int proto, int *nxtp) 1568 { 1569 struct ip6_hdr ip6; 1570 struct ip6_ext ip6e; 1571 struct ip6_frag fh; 1572 1573 /* just in case */ 1574 if (m == NULL) 1575 panic("ip6_nexthdr: m == NULL"); 1576 if ((m->m_flags & M_PKTHDR) == 0 || m->m_pkthdr.len < off) 1577 return -1; 1578 1579 switch (proto) { 1580 case IPPROTO_IPV6: 1581 if (m->m_pkthdr.len < off + sizeof(ip6)) 1582 return -1; 1583 m_copydata(m, off, sizeof(ip6), (caddr_t)&ip6); 1584 if (nxtp) 1585 *nxtp = ip6.ip6_nxt; 1586 off += sizeof(ip6); 1587 return off; 1588 1589 case IPPROTO_FRAGMENT: 1590 /* 1591 * terminate parsing if it is not the first fragment, 1592 * it does not make sense to parse through it. 1593 */ 1594 if (m->m_pkthdr.len < off + sizeof(fh)) 1595 return -1; 1596 m_copydata(m, off, sizeof(fh), (caddr_t)&fh); 1597 /* IP6F_OFF_MASK = 0xfff8(BigEndian), 0xf8ff(LittleEndian) */ 1598 if (fh.ip6f_offlg & IP6F_OFF_MASK) 1599 return -1; 1600 if (nxtp) 1601 *nxtp = fh.ip6f_nxt; 1602 off += sizeof(struct ip6_frag); 1603 return off; 1604 1605 case IPPROTO_AH: 1606 if (m->m_pkthdr.len < off + sizeof(ip6e)) 1607 return -1; 1608 m_copydata(m, off, sizeof(ip6e), (caddr_t)&ip6e); 1609 if (nxtp) 1610 *nxtp = ip6e.ip6e_nxt; 1611 off += (ip6e.ip6e_len + 2) << 2; 1612 return off; 1613 1614 case IPPROTO_HOPOPTS: 1615 case IPPROTO_ROUTING: 1616 case IPPROTO_DSTOPTS: 1617 if (m->m_pkthdr.len < off + sizeof(ip6e)) 1618 return -1; 1619 m_copydata(m, off, sizeof(ip6e), (caddr_t)&ip6e); 1620 if (nxtp) 1621 *nxtp = ip6e.ip6e_nxt; 1622 off += (ip6e.ip6e_len + 1) << 3; 1623 return off; 1624 1625 case IPPROTO_NONE: 1626 case IPPROTO_ESP: 1627 case IPPROTO_IPCOMP: 1628 /* give up */ 1629 return -1; 1630 1631 default: 1632 return -1; 1633 } 1634 1635 return -1; 1636 } 1637 1638 /* 1639 * get offset for the last header in the chain. m will be kept untainted. 1640 */ 1641 int 1642 ip6_lasthdr(struct mbuf *m, int off, int proto, int *nxtp) 1643 { 1644 int newoff; 1645 int nxt; 1646 1647 if (!nxtp) { 1648 nxt = -1; 1649 nxtp = &nxt; 1650 } 1651 while (1) { 1652 newoff = ip6_nexthdr(m, off, proto, nxtp); 1653 if (newoff < 0) 1654 return off; 1655 else if (newoff < off) 1656 return -1; /* invalid */ 1657 else if (newoff == off) 1658 return newoff; 1659 1660 off = newoff; 1661 proto = *nxtp; 1662 } 1663 } 1664 1665 struct ip6aux * 1666 ip6_addaux(struct mbuf *m) 1667 { 1668 struct m_tag *tag = m_tag_find(m, PACKET_TAG_IPV6_INPUT, NULL); 1669 if (!tag) { 1670 tag = m_tag_get(PACKET_TAG_IPV6_INPUT, 1671 sizeof (struct ip6aux), 1672 MB_DONTWAIT); 1673 if (tag) 1674 m_tag_prepend(m, tag); 1675 } 1676 if (tag) 1677 bzero(m_tag_data(tag), sizeof (struct ip6aux)); 1678 return tag ? (struct ip6aux *)m_tag_data(tag) : NULL; 1679 } 1680 1681 struct ip6aux * 1682 ip6_findaux(struct mbuf *m) 1683 { 1684 struct m_tag *tag = m_tag_find(m, PACKET_TAG_IPV6_INPUT, NULL); 1685 return tag ? (struct ip6aux*)(tag+1) : NULL; 1686 } 1687 1688 void 1689 ip6_delaux(struct mbuf *m) 1690 { 1691 struct m_tag *tag = m_tag_find(m, PACKET_TAG_IPV6_INPUT, NULL); 1692 if (tag) 1693 m_tag_delete(m, tag); 1694 } 1695 1696 /* 1697 * System control for IP6 1698 */ 1699 1700 u_char inet6ctlerrmap[PRC_NCMDS] = { 1701 0, 0, 0, 0, 1702 0, EMSGSIZE, EHOSTDOWN, EHOSTUNREACH, 1703 EHOSTUNREACH, EHOSTUNREACH, ECONNREFUSED, ECONNREFUSED, 1704 EMSGSIZE, EHOSTUNREACH, 0, 0, 1705 0, 0, 0, 0, 1706 ENOPROTOOPT 1707 }; 1708