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 1.20 2004/09/17 00:07:27 dillon 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 88 #include <sys/thread2.h> 89 #include <sys/msgport2.h> 90 91 #include <net/if.h> 92 #include <net/if_types.h> 93 #include <net/if_dl.h> 94 #include <net/route.h> 95 #include <net/netisr.h> 96 #include <net/pfil.h> 97 98 #include <netinet/in.h> 99 #include <netinet/in_systm.h> 100 #ifdef INET 101 #include <netinet/ip.h> 102 #include <netinet/ip_icmp.h> 103 #endif /* INET */ 104 #include <netinet/ip6.h> 105 #include <netinet6/in6_var.h> 106 #include <netinet6/ip6_var.h> 107 #include <netinet/in_pcb.h> 108 #include <netinet/icmp6.h> 109 #include <netinet6/in6_ifattach.h> 110 #include <netinet6/nd6.h> 111 #include <netinet6/in6_prefix.h> 112 113 #ifdef IPSEC 114 #include <netinet6/ipsec.h> 115 #ifdef INET6 116 #include <netinet6/ipsec6.h> 117 #endif 118 #endif 119 120 #ifdef FAST_IPSEC 121 #include "ipsec.h" 122 #include "ipsec6.h" 123 #define IPSEC 124 #endif /* FAST_IPSEC */ 125 126 #include <net/ip6fw/ip6_fw.h> 127 128 #include <netinet6/ip6protosw.h> 129 130 /* we need it for NLOOP. */ 131 #include "use_loop.h" 132 133 #include <net/net_osdep.h> 134 135 extern struct domain inet6domain; 136 extern struct ip6protosw inet6sw[]; 137 138 u_char ip6_protox[IPPROTO_MAX]; 139 struct in6_ifaddr *in6_ifaddr; 140 141 extern struct callout in6_tmpaddrtimer_ch; 142 143 int ip6_forward_srcrt; /* XXX */ 144 int ip6_sourcecheck; /* XXX */ 145 int ip6_sourcecheck_interval; /* XXX */ 146 147 int ip6_ours_check_algorithm; 148 149 struct pfil_head inet6_pfil_hook; 150 151 /* firewall hooks */ 152 ip6_fw_chk_t *ip6_fw_chk_ptr; 153 ip6_fw_ctl_t *ip6_fw_ctl_ptr; 154 int ip6_fw_enable = 1; 155 156 struct ip6stat ip6stat; 157 158 static void ip6_init2 (void *); 159 static struct ip6aux *ip6_setdstifaddr (struct mbuf *, struct in6_ifaddr *); 160 static int ip6_hopopts_input (u_int32_t *, u_int32_t *, struct mbuf **, int *); 161 static int ip6_input(struct netmsg *msg); 162 #ifdef PULLDOWN_TEST 163 static struct mbuf *ip6_pullexthdr (struct mbuf *, size_t, int); 164 #endif 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 ip6protosw *pr; 174 int i; 175 struct timeval tv; 176 177 #ifdef DIAGNOSTIC 178 if (sizeof(struct protosw) != sizeof(struct ip6protosw)) 179 panic("sizeof(protosw) != sizeof(ip6protosw)"); 180 #endif 181 pr = (struct ip6protosw *)pffindproto(PF_INET6, IPPROTO_RAW, SOCK_RAW); 182 if (pr == 0) 183 panic("ip6_init"); 184 for (i = 0; i < IPPROTO_MAX; i++) 185 ip6_protox[i] = pr - inet6sw; 186 for (pr = (struct ip6protosw *)inet6domain.dom_protosw; 187 pr < (struct ip6protosw *)inet6domain.dom_protoswNPROTOSW; pr++) 188 if (pr->pr_domain->dom_family == PF_INET6 && 189 pr->pr_protocol && pr->pr_protocol != IPPROTO_RAW) 190 ip6_protox[pr->pr_protocol] = pr - inet6sw; 191 192 inet6_pfil_hook.ph_type = PFIL_TYPE_AF; 193 inet6_pfil_hook.ph_af = AF_INET6; 194 if ((i = pfil_head_register(&inet6_pfil_hook)) != 0) { 195 printf("%s: WARNING: unable to register pfil hook, " 196 "error %d\n", __func__, i); 197 } 198 199 netisr_register(NETISR_IPV6, cpu0_portfn, ip6_input); 200 nd6_init(); 201 frag6_init(); 202 /* 203 * in many cases, random() here does NOT return random number 204 * as initialization during bootstrap time occur in fixed order. 205 */ 206 microtime(&tv); 207 ip6_flow_seq = random() ^ tv.tv_usec; 208 microtime(&tv); 209 ip6_desync_factor = (random() ^ tv.tv_usec) % MAX_TEMP_DESYNC_FACTOR; 210 } 211 212 static void 213 ip6_init2(void *dummy) 214 { 215 216 /* 217 * to route local address of p2p link to loopback, 218 * assign loopback address first. 219 */ 220 in6_ifattach(&loif[0], NULL); 221 222 /* nd6_timer_init */ 223 callout_init(&nd6_timer_ch); 224 callout_reset(&nd6_timer_ch, hz, nd6_timer, NULL); 225 226 /* router renumbering prefix list maintenance */ 227 callout_init(&in6_rr_timer_ch); 228 callout_reset(&in6_rr_timer_ch, hz, in6_rr_timer, NULL); 229 230 /* timer for regeneranation of temporary addresses randomize ID */ 231 callout_init(&in6_tmpaddrtimer_ch); 232 callout_reset(&in6_tmpaddrtimer_ch, 233 (ip6_temp_preferred_lifetime - ip6_desync_factor - 234 ip6_temp_regen_advance) * hz, 235 in6_tmpaddrtimer, NULL); 236 } 237 238 /* cheat */ 239 /* This must be after route_init(), which is now SI_ORDER_THIRD */ 240 SYSINIT(netinet6init2, SI_SUB_PROTO_DOMAIN, SI_ORDER_MIDDLE, ip6_init2, NULL); 241 242 extern struct route_in6 ip6_forward_rt; 243 244 static 245 int 246 ip6_input(struct netmsg *msg) 247 { 248 struct mbuf *m = ((struct netmsg_packet *)msg)->nm_packet; 249 struct ip6_hdr *ip6; 250 int off = sizeof(struct ip6_hdr), nest; 251 u_int32_t plen; 252 u_int32_t rtalert = ~0; 253 int nxt, ours = 0; 254 struct ifnet *deliverifp = NULL; 255 struct in6_addr odst; 256 int srcrt = 0; 257 258 #ifdef IPSEC 259 /* 260 * should the inner packet be considered authentic? 261 * see comment in ah4_input(). 262 */ 263 if (m) { 264 m->m_flags &= ~M_AUTHIPHDR; 265 m->m_flags &= ~M_AUTHIPDGM; 266 } 267 #endif 268 269 /* 270 * make sure we don't have onion peering information into m_aux. 271 */ 272 ip6_delaux(m); 273 274 /* 275 * mbuf statistics 276 */ 277 if (m->m_flags & M_EXT) { 278 if (m->m_next) 279 ip6stat.ip6s_mext2m++; 280 else 281 ip6stat.ip6s_mext1++; 282 } else { 283 #define M2MMAX (sizeof(ip6stat.ip6s_m2m)/sizeof(ip6stat.ip6s_m2m[0])) 284 if (m->m_next) { 285 if (m->m_flags & M_LOOP) { 286 ip6stat.ip6s_m2m[loif[0].if_index]++; /* XXX */ 287 } else if (m->m_pkthdr.rcvif->if_index < M2MMAX) 288 ip6stat.ip6s_m2m[m->m_pkthdr.rcvif->if_index]++; 289 else 290 ip6stat.ip6s_m2m[0]++; 291 } else 292 ip6stat.ip6s_m1++; 293 #undef M2MMAX 294 } 295 296 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_receive); 297 ip6stat.ip6s_total++; 298 299 #ifndef PULLDOWN_TEST 300 /* 301 * L2 bridge code and some other code can return mbuf chain 302 * that does not conform to KAME requirement. too bad. 303 * XXX: fails to join if interface MTU > MCLBYTES. jumbogram? 304 */ 305 if (m && m->m_next != NULL && m->m_pkthdr.len < MCLBYTES) { 306 struct mbuf *n; 307 308 MGETHDR(n, MB_DONTWAIT, MT_HEADER); 309 if (n) 310 M_MOVE_PKTHDR(n, m); 311 if (n && n->m_pkthdr.len > MHLEN) { 312 MCLGET(n, MB_DONTWAIT); 313 if ((n->m_flags & M_EXT) == 0) { 314 m_freem(n); 315 n = NULL; 316 } 317 } 318 if (n == NULL) 319 goto bad; 320 321 m_copydata(m, 0, n->m_pkthdr.len, mtod(n, caddr_t)); 322 n->m_len = n->m_pkthdr.len; 323 m_freem(m); 324 m = n; 325 } 326 IP6_EXTHDR_CHECK(m, 0, sizeof(struct ip6_hdr), EASYNC); 327 #endif 328 329 if (m->m_len < sizeof(struct ip6_hdr)) { 330 struct ifnet *inifp; 331 inifp = m->m_pkthdr.rcvif; 332 if ((m = m_pullup(m, sizeof(struct ip6_hdr))) == 0) { 333 ip6stat.ip6s_toosmall++; 334 in6_ifstat_inc(inifp, ifs6_in_hdrerr); 335 goto bad2; 336 } 337 } 338 339 ip6 = mtod(m, struct ip6_hdr *); 340 341 if ((ip6->ip6_vfc & IPV6_VERSION_MASK) != IPV6_VERSION) { 342 ip6stat.ip6s_badvers++; 343 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_hdrerr); 344 goto bad; 345 } 346 347 /* 348 * Run through list of hooks for input packets. 349 * 350 * NB: Beware of the destination address changing 351 * (e.g. by NAT rewriting). When this happens, 352 * tell ip6_forward to do the right thing. 353 */ 354 if (pfil_has_hooks(&inet6_pfil_hook)) { 355 odst = ip6->ip6_dst; 356 if (pfil_run_hooks(&inet6_pfil_hook, &m, 357 m->m_pkthdr.rcvif, PFIL_IN)) { 358 goto bad2; 359 } 360 if (m == NULL) /* consumed by filter */ 361 goto bad2; 362 ip6 = mtod(m, struct ip6_hdr *); 363 srcrt = !IN6_ARE_ADDR_EQUAL(&odst, &ip6->ip6_dst); 364 } 365 366 ip6stat.ip6s_nxthist[ip6->ip6_nxt]++; 367 368 /* 369 * Check with the firewall... 370 */ 371 if (ip6_fw_enable && ip6_fw_chk_ptr) { 372 u_short port = 0; 373 /* If ipfw says divert, we have to just drop packet */ 374 /* use port as a dummy argument */ 375 if ((*ip6_fw_chk_ptr)(&ip6, NULL, &port, &m)) { 376 m_freem(m); 377 m = NULL; 378 } 379 if (!m) 380 goto bad2; 381 } 382 383 /* 384 * Check against address spoofing/corruption. 385 */ 386 if (IN6_IS_ADDR_MULTICAST(&ip6->ip6_src) || 387 IN6_IS_ADDR_UNSPECIFIED(&ip6->ip6_dst)) { 388 /* 389 * XXX: "badscope" is not very suitable for a multicast source. 390 */ 391 ip6stat.ip6s_badscope++; 392 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_addrerr); 393 goto bad; 394 } 395 if ((IN6_IS_ADDR_LOOPBACK(&ip6->ip6_src) || 396 IN6_IS_ADDR_LOOPBACK(&ip6->ip6_dst)) && 397 (m->m_pkthdr.rcvif->if_flags & IFF_LOOPBACK) == 0) { 398 ip6stat.ip6s_badscope++; 399 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_addrerr); 400 goto bad; 401 } 402 403 /* 404 * The following check is not documented in specs. A malicious 405 * party may be able to use IPv4 mapped addr to confuse tcp/udp stack 406 * and bypass security checks (act as if it was from 127.0.0.1 by using 407 * IPv6 src ::ffff:127.0.0.1). Be cautious. 408 * 409 * This check chokes if we are in an SIIT cloud. As none of BSDs 410 * support IPv4-less kernel compilation, we cannot support SIIT 411 * environment at all. So, it makes more sense for us to reject any 412 * malicious packets for non-SIIT environment, than try to do a 413 * partical support for SIIT environment. 414 */ 415 if (IN6_IS_ADDR_V4MAPPED(&ip6->ip6_src) || 416 IN6_IS_ADDR_V4MAPPED(&ip6->ip6_dst)) { 417 ip6stat.ip6s_badscope++; 418 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_addrerr); 419 goto bad; 420 } 421 #if 0 422 /* 423 * Reject packets with IPv4 compatible addresses (auto tunnel). 424 * 425 * The code forbids auto tunnel relay case in RFC1933 (the check is 426 * stronger than RFC1933). We may want to re-enable it if mech-xx 427 * is revised to forbid relaying case. 428 */ 429 if (IN6_IS_ADDR_V4COMPAT(&ip6->ip6_src) || 430 IN6_IS_ADDR_V4COMPAT(&ip6->ip6_dst)) { 431 ip6stat.ip6s_badscope++; 432 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_addrerr); 433 goto bad; 434 } 435 #endif 436 437 /* drop packets if interface ID portion is already filled */ 438 if ((m->m_pkthdr.rcvif->if_flags & IFF_LOOPBACK) == 0) { 439 if (IN6_IS_SCOPE_LINKLOCAL(&ip6->ip6_src) && 440 ip6->ip6_src.s6_addr16[1]) { 441 ip6stat.ip6s_badscope++; 442 goto bad; 443 } 444 if (IN6_IS_SCOPE_LINKLOCAL(&ip6->ip6_dst) && 445 ip6->ip6_dst.s6_addr16[1]) { 446 ip6stat.ip6s_badscope++; 447 goto bad; 448 } 449 } 450 451 if (IN6_IS_SCOPE_LINKLOCAL(&ip6->ip6_src)) 452 ip6->ip6_src.s6_addr16[1] 453 = htons(m->m_pkthdr.rcvif->if_index); 454 if (IN6_IS_SCOPE_LINKLOCAL(&ip6->ip6_dst)) 455 ip6->ip6_dst.s6_addr16[1] 456 = htons(m->m_pkthdr.rcvif->if_index); 457 458 #if 0 /* this case seems to be unnecessary. (jinmei, 20010401) */ 459 /* 460 * We use rt->rt_ifp to determine if the address is ours or not. 461 * If rt_ifp is lo0, the address is ours. 462 * The problem here is, rt->rt_ifp for fe80::%lo0/64 is set to lo0, 463 * so any address under fe80::%lo0/64 will be mistakenly considered 464 * local. The special case is supplied to handle the case properly 465 * by actually looking at interface addresses 466 * (using in6ifa_ifpwithaddr). 467 */ 468 if ((m->m_pkthdr.rcvif->if_flags & IFF_LOOPBACK) != 0 && 469 IN6_IS_ADDR_LINKLOCAL(&ip6->ip6_dst)) { 470 if (!in6ifa_ifpwithaddr(m->m_pkthdr.rcvif, &ip6->ip6_dst)) { 471 icmp6_error(m, ICMP6_DST_UNREACH, 472 ICMP6_DST_UNREACH_ADDR, 0); 473 /* m is already freed */ 474 goto bad2; 475 } 476 477 ours = 1; 478 deliverifp = m->m_pkthdr.rcvif; 479 goto hbhcheck; 480 } 481 #endif 482 483 /* 484 * Multicast check 485 */ 486 if (IN6_IS_ADDR_MULTICAST(&ip6->ip6_dst)) { 487 struct in6_multi *in6m = 0; 488 489 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_mcast); 490 /* 491 * See if we belong to the destination multicast group on the 492 * arrival interface. 493 */ 494 IN6_LOOKUP_MULTI(ip6->ip6_dst, m->m_pkthdr.rcvif, in6m); 495 if (in6m) 496 ours = 1; 497 else if (!ip6_mrouter) { 498 ip6stat.ip6s_notmember++; 499 ip6stat.ip6s_cantforward++; 500 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_discard); 501 goto bad; 502 } 503 deliverifp = m->m_pkthdr.rcvif; 504 goto hbhcheck; 505 } 506 507 /* 508 * Unicast check 509 */ 510 switch (ip6_ours_check_algorithm) { 511 default: 512 /* 513 * XXX: I intentionally broke our indentation rule here, 514 * since this switch-case is just for measurement and 515 * therefore should soon be removed. 516 */ 517 if (ip6_forward_rt.ro_rt != NULL && 518 (ip6_forward_rt.ro_rt->rt_flags & RTF_UP) != 0 && 519 IN6_ARE_ADDR_EQUAL(&ip6->ip6_dst, 520 &((struct sockaddr_in6 *)(&ip6_forward_rt.ro_dst))->sin6_addr)) 521 ip6stat.ip6s_forward_cachehit++; 522 else { 523 struct sockaddr_in6 *dst6; 524 525 if (ip6_forward_rt.ro_rt) { 526 /* route is down or destination is different */ 527 ip6stat.ip6s_forward_cachemiss++; 528 RTFREE(ip6_forward_rt.ro_rt); 529 ip6_forward_rt.ro_rt = 0; 530 } 531 532 bzero(&ip6_forward_rt.ro_dst, sizeof(struct sockaddr_in6)); 533 dst6 = (struct sockaddr_in6 *)&ip6_forward_rt.ro_dst; 534 dst6->sin6_len = sizeof(struct sockaddr_in6); 535 dst6->sin6_family = AF_INET6; 536 dst6->sin6_addr = ip6->ip6_dst; 537 #ifdef SCOPEDROUTING 538 ip6_forward_rt.ro_dst.sin6_scope_id = 539 in6_addr2scopeid(m->m_pkthdr.rcvif, &ip6->ip6_dst); 540 #endif 541 542 rtalloc_ign((struct route *)&ip6_forward_rt, RTF_PRCLONING); 543 } 544 545 #define rt6_key(r) ((struct sockaddr_in6 *)((r)->rt_nodes->rn_key)) 546 547 /* 548 * Accept the packet if the forwarding interface to the destination 549 * according to the routing table is the loopback interface, 550 * unless the associated route has a gateway. 551 * Note that this approach causes to accept a packet if there is a 552 * route to the loopback interface for the destination of the packet. 553 * But we think it's even useful in some situations, e.g. when using 554 * a special daemon which wants to intercept the packet. 555 * 556 * XXX: some OSes automatically make a cloned route for the destination 557 * of an outgoing packet. If the outgoing interface of the packet 558 * is a loopback one, the kernel would consider the packet to be 559 * accepted, even if we have no such address assinged on the interface. 560 * We check the cloned flag of the route entry to reject such cases, 561 * assuming that route entries for our own addresses are not made by 562 * cloning (it should be true because in6_addloop explicitly installs 563 * the host route). However, we might have to do an explicit check 564 * while it would be less efficient. Or, should we rather install a 565 * reject route for such a case? 566 */ 567 if (ip6_forward_rt.ro_rt && 568 (ip6_forward_rt.ro_rt->rt_flags & 569 (RTF_HOST|RTF_GATEWAY)) == RTF_HOST && 570 #ifdef RTF_WASCLONED 571 !(ip6_forward_rt.ro_rt->rt_flags & RTF_WASCLONED) && 572 #endif 573 #ifdef RTF_CLONED 574 !(ip6_forward_rt.ro_rt->rt_flags & RTF_CLONED) && 575 #endif 576 #if 0 577 /* 578 * The check below is redundant since the comparison of 579 * the destination and the key of the rtentry has 580 * already done through looking up the routing table. 581 */ 582 IN6_ARE_ADDR_EQUAL(&ip6->ip6_dst, 583 &rt6_key(ip6_forward_rt.ro_rt)->sin6_addr) 584 #endif 585 ip6_forward_rt.ro_rt->rt_ifp->if_type == IFT_LOOP) { 586 struct in6_ifaddr *ia6 = 587 (struct in6_ifaddr *)ip6_forward_rt.ro_rt->rt_ifa; 588 589 /* 590 * record address information into m_aux. 591 */ 592 (void)ip6_setdstifaddr(m, ia6); 593 594 /* 595 * packets to a tentative, duplicated, or somehow invalid 596 * address must not be accepted. 597 */ 598 if (!(ia6->ia6_flags & IN6_IFF_NOTREADY)) { 599 /* this address is ready */ 600 ours = 1; 601 deliverifp = ia6->ia_ifp; /* correct? */ 602 /* Count the packet in the ip address stats */ 603 ia6->ia_ifa.if_ipackets++; 604 ia6->ia_ifa.if_ibytes += m->m_pkthdr.len; 605 goto hbhcheck; 606 } else { 607 /* address is not ready, so discard the packet. */ 608 nd6log((LOG_INFO, 609 "ip6_input: packet to an unready address %s->%s\n", 610 ip6_sprintf(&ip6->ip6_src), 611 ip6_sprintf(&ip6->ip6_dst))); 612 613 goto bad; 614 } 615 } 616 } /* XXX indentation (see above) */ 617 618 /* 619 * FAITH(Firewall Aided Internet Translator) 620 */ 621 if (ip6_keepfaith) { 622 if (ip6_forward_rt.ro_rt && ip6_forward_rt.ro_rt->rt_ifp 623 && ip6_forward_rt.ro_rt->rt_ifp->if_type == IFT_FAITH) { 624 /* XXX do we need more sanity checks? */ 625 ours = 1; 626 deliverifp = ip6_forward_rt.ro_rt->rt_ifp; /* faith */ 627 goto hbhcheck; 628 } 629 } 630 631 /* 632 * Now there is no reason to process the packet if it's not our own 633 * and we're not a router. 634 */ 635 if (!ip6_forwarding) { 636 ip6stat.ip6s_cantforward++; 637 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_discard); 638 goto bad; 639 } 640 641 hbhcheck: 642 /* 643 * record address information into m_aux, if we don't have one yet. 644 * note that we are unable to record it, if the address is not listed 645 * as our interface address (e.g. multicast addresses, addresses 646 * within FAITH prefixes and such). 647 */ 648 if (deliverifp && !ip6_getdstifaddr(m)) { 649 struct in6_ifaddr *ia6; 650 651 ia6 = in6_ifawithifp(deliverifp, &ip6->ip6_dst); 652 if (ia6) { 653 if (!ip6_setdstifaddr(m, ia6)) { 654 /* 655 * XXX maybe we should drop the packet here, 656 * as we could not provide enough information 657 * to the upper layers. 658 */ 659 } 660 } 661 } 662 663 /* 664 * Process Hop-by-Hop options header if it's contained. 665 * m may be modified in ip6_hopopts_input(). 666 * If a JumboPayload option is included, plen will also be modified. 667 */ 668 plen = (u_int32_t)ntohs(ip6->ip6_plen); 669 if (ip6->ip6_nxt == IPPROTO_HOPOPTS) { 670 struct ip6_hbh *hbh; 671 672 if (ip6_hopopts_input(&plen, &rtalert, &m, &off)) { 673 #if 0 /*touches NULL pointer*/ 674 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_discard); 675 #endif 676 goto bad2; /* m have already been freed */ 677 } 678 679 /* adjust pointer */ 680 ip6 = mtod(m, struct ip6_hdr *); 681 682 /* 683 * if the payload length field is 0 and the next header field 684 * indicates Hop-by-Hop Options header, then a Jumbo Payload 685 * option MUST be included. 686 */ 687 if (ip6->ip6_plen == 0 && plen == 0) { 688 /* 689 * Note that if a valid jumbo payload option is 690 * contained, ip6_hoptops_input() must set a valid 691 * (non-zero) payload length to the variable plen. 692 */ 693 ip6stat.ip6s_badoptions++; 694 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_discard); 695 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_hdrerr); 696 icmp6_error(m, ICMP6_PARAM_PROB, 697 ICMP6_PARAMPROB_HEADER, 698 (caddr_t)&ip6->ip6_plen - (caddr_t)ip6); 699 goto bad2; 700 } 701 #ifndef PULLDOWN_TEST 702 /* ip6_hopopts_input() ensures that mbuf is contiguous */ 703 hbh = (struct ip6_hbh *)(ip6 + 1); 704 #else 705 IP6_EXTHDR_GET(hbh, struct ip6_hbh *, m, sizeof(struct ip6_hdr), 706 sizeof(struct ip6_hbh)); 707 if (hbh == NULL) { 708 ip6stat.ip6s_tooshort++; 709 goto bad2; 710 } 711 #endif 712 nxt = hbh->ip6h_nxt; 713 714 /* 715 * accept the packet if a router alert option is included 716 * and we act as an IPv6 router. 717 */ 718 if (rtalert != ~0 && ip6_forwarding) 719 ours = 1; 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 while (nxt != IPPROTO_DONE) { 791 if (ip6_hdrnestlimit && (++nest > ip6_hdrnestlimit)) { 792 ip6stat.ip6s_toomanyhdr++; 793 goto bad; 794 } 795 796 /* 797 * protection against faulty packet - there should be 798 * more sanity checks in header chain processing. 799 */ 800 if (m->m_pkthdr.len < off) { 801 ip6stat.ip6s_tooshort++; 802 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_truncated); 803 goto bad; 804 } 805 806 #if 0 807 /* 808 * do we need to do it for every header? yeah, other 809 * functions can play with it (like re-allocate and copy). 810 */ 811 mhist = ip6_addaux(m); 812 if (mhist && M_TRAILINGSPACE(mhist) >= sizeof(nxt)) { 813 hist = mtod(mhist, caddr_t) + mhist->m_len; 814 bcopy(&nxt, hist, sizeof(nxt)); 815 mhist->m_len += sizeof(nxt); 816 } else { 817 ip6stat.ip6s_toomanyhdr++; 818 goto bad; 819 } 820 #endif 821 822 #ifdef IPSEC 823 /* 824 * enforce IPsec policy checking if we are seeing last header. 825 * note that we do not visit this with protocols with pcb layer 826 * code - like udp/tcp/raw ip. 827 */ 828 if ((inet6sw[ip6_protox[nxt]].pr_flags & PR_LASTHDR) != 0 && 829 ipsec6_in_reject(m, NULL)) { 830 ipsec6stat.in_polvio++; 831 goto bad; 832 } 833 #endif 834 835 nxt = (*inet6sw[ip6_protox[nxt]].pr_input)(&m, &off, nxt); 836 } 837 goto bad2; 838 bad: 839 m_freem(m); 840 bad2: 841 lwkt_replymsg(&msg->nm_lmsg, 0); 842 return(EASYNC); 843 } 844 845 /* 846 * set/grab in6_ifaddr correspond to IPv6 destination address. 847 * XXX backward compatibility wrapper 848 */ 849 static struct ip6aux * 850 ip6_setdstifaddr(struct mbuf *m, struct in6_ifaddr *ia6) 851 { 852 struct ip6aux *n; 853 854 n = ip6_addaux(m); 855 if (n) 856 n->ip6a_dstia6 = ia6; 857 return n; /* NULL if failed to set */ 858 } 859 860 struct in6_ifaddr * 861 ip6_getdstifaddr(struct mbuf *m) 862 { 863 struct ip6aux *n; 864 865 n = ip6_findaux(m); 866 if (n) 867 return n->ip6a_dstia6; 868 else 869 return NULL; 870 } 871 872 /* 873 * Hop-by-Hop options header processing. If a valid jumbo payload option is 874 * included, the real payload length will be stored in plenp. 875 */ 876 static int 877 ip6_hopopts_input(u_int32_t *plenp, 878 u_int32_t *rtalertp,/* XXX: should be stored more smart way */ 879 struct mbuf **mp, 880 int *offp) 881 { 882 struct mbuf *m = *mp; 883 int off = *offp, hbhlen; 884 struct ip6_hbh *hbh; 885 u_int8_t *opt; 886 887 /* validation of the length of the header */ 888 #ifndef PULLDOWN_TEST 889 IP6_EXTHDR_CHECK(m, off, sizeof(*hbh), -1); 890 hbh = (struct ip6_hbh *)(mtod(m, caddr_t) + off); 891 hbhlen = (hbh->ip6h_len + 1) << 3; 892 893 IP6_EXTHDR_CHECK(m, off, hbhlen, -1); 894 hbh = (struct ip6_hbh *)(mtod(m, caddr_t) + off); 895 #else 896 IP6_EXTHDR_GET(hbh, struct ip6_hbh *, m, 897 sizeof(struct ip6_hdr), sizeof(struct ip6_hbh)); 898 if (hbh == NULL) { 899 ip6stat.ip6s_tooshort++; 900 return -1; 901 } 902 hbhlen = (hbh->ip6h_len + 1) << 3; 903 IP6_EXTHDR_GET(hbh, struct ip6_hbh *, m, sizeof(struct ip6_hdr), 904 hbhlen); 905 if (hbh == NULL) { 906 ip6stat.ip6s_tooshort++; 907 return -1; 908 } 909 #endif 910 off += hbhlen; 911 hbhlen -= sizeof(struct ip6_hbh); 912 opt = (u_int8_t *)hbh + sizeof(struct ip6_hbh); 913 914 if (ip6_process_hopopts(m, (u_int8_t *)hbh + sizeof(struct ip6_hbh), 915 hbhlen, rtalertp, plenp) < 0) 916 return(-1); 917 918 *offp = off; 919 *mp = m; 920 return(0); 921 } 922 923 /* 924 * Search header for all Hop-by-hop options and process each option. 925 * This function is separate from ip6_hopopts_input() in order to 926 * handle a case where the sending node itself process its hop-by-hop 927 * options header. In such a case, the function is called from ip6_output(). 928 * 929 * The function assumes that hbh header is located right after the IPv6 header 930 * (RFC2460 p7), opthead is pointer into data content in m, and opthead to 931 * opthead + hbhlen is located in continuous memory region. 932 */ 933 int 934 ip6_process_hopopts(struct mbuf *m, u_int8_t *opthead, int hbhlen, 935 u_int32_t *rtalertp, u_int32_t *plenp) 936 { 937 struct ip6_hdr *ip6; 938 int optlen = 0; 939 u_int8_t *opt = opthead; 940 u_int16_t rtalert_val; 941 u_int32_t jumboplen; 942 const int erroff = sizeof(struct ip6_hdr) + sizeof(struct ip6_hbh); 943 944 for (; hbhlen > 0; hbhlen -= optlen, opt += optlen) { 945 switch (*opt) { 946 case IP6OPT_PAD1: 947 optlen = 1; 948 break; 949 case IP6OPT_PADN: 950 if (hbhlen < IP6OPT_MINLEN) { 951 ip6stat.ip6s_toosmall++; 952 goto bad; 953 } 954 optlen = *(opt + 1) + 2; 955 break; 956 case IP6OPT_RTALERT: 957 /* XXX may need check for alignment */ 958 if (hbhlen < IP6OPT_RTALERT_LEN) { 959 ip6stat.ip6s_toosmall++; 960 goto bad; 961 } 962 if (*(opt + 1) != IP6OPT_RTALERT_LEN - 2) { 963 /* XXX stat */ 964 icmp6_error(m, ICMP6_PARAM_PROB, 965 ICMP6_PARAMPROB_HEADER, 966 erroff + opt + 1 - opthead); 967 return(-1); 968 } 969 optlen = IP6OPT_RTALERT_LEN; 970 bcopy((caddr_t)(opt + 2), (caddr_t)&rtalert_val, 2); 971 *rtalertp = ntohs(rtalert_val); 972 break; 973 case IP6OPT_JUMBO: 974 /* XXX may need check for alignment */ 975 if (hbhlen < IP6OPT_JUMBO_LEN) { 976 ip6stat.ip6s_toosmall++; 977 goto bad; 978 } 979 if (*(opt + 1) != IP6OPT_JUMBO_LEN - 2) { 980 /* XXX stat */ 981 icmp6_error(m, ICMP6_PARAM_PROB, 982 ICMP6_PARAMPROB_HEADER, 983 erroff + opt + 1 - opthead); 984 return(-1); 985 } 986 optlen = IP6OPT_JUMBO_LEN; 987 988 /* 989 * IPv6 packets that have non 0 payload length 990 * must not contain a jumbo payload option. 991 */ 992 ip6 = mtod(m, struct ip6_hdr *); 993 if (ip6->ip6_plen) { 994 ip6stat.ip6s_badoptions++; 995 icmp6_error(m, ICMP6_PARAM_PROB, 996 ICMP6_PARAMPROB_HEADER, 997 erroff + opt - opthead); 998 return(-1); 999 } 1000 1001 /* 1002 * We may see jumbolen in unaligned location, so 1003 * we'd need to perform bcopy(). 1004 */ 1005 bcopy(opt + 2, &jumboplen, sizeof(jumboplen)); 1006 jumboplen = (u_int32_t)htonl(jumboplen); 1007 1008 #if 1 1009 /* 1010 * if there are multiple jumbo payload options, 1011 * *plenp will be non-zero and the packet will be 1012 * rejected. 1013 * the behavior may need some debate in ipngwg - 1014 * multiple options does not make sense, however, 1015 * there's no explicit mention in specification. 1016 */ 1017 if (*plenp != 0) { 1018 ip6stat.ip6s_badoptions++; 1019 icmp6_error(m, ICMP6_PARAM_PROB, 1020 ICMP6_PARAMPROB_HEADER, 1021 erroff + opt + 2 - opthead); 1022 return(-1); 1023 } 1024 #endif 1025 1026 /* 1027 * jumbo payload length must be larger than 65535. 1028 */ 1029 if (jumboplen <= IPV6_MAXPACKET) { 1030 ip6stat.ip6s_badoptions++; 1031 icmp6_error(m, ICMP6_PARAM_PROB, 1032 ICMP6_PARAMPROB_HEADER, 1033 erroff + opt + 2 - opthead); 1034 return(-1); 1035 } 1036 *plenp = jumboplen; 1037 1038 break; 1039 default: /* unknown option */ 1040 if (hbhlen < IP6OPT_MINLEN) { 1041 ip6stat.ip6s_toosmall++; 1042 goto bad; 1043 } 1044 optlen = ip6_unknown_opt(opt, m, 1045 erroff + opt - opthead); 1046 if (optlen == -1) 1047 return(-1); 1048 optlen += 2; 1049 break; 1050 } 1051 } 1052 1053 return(0); 1054 1055 bad: 1056 m_freem(m); 1057 return(-1); 1058 } 1059 1060 /* 1061 * Unknown option processing. 1062 * The third argument `off' is the offset from the IPv6 header to the option, 1063 * which is necessary if the IPv6 header the and option header and IPv6 header 1064 * is not continuous in order to return an ICMPv6 error. 1065 */ 1066 int 1067 ip6_unknown_opt(u_int8_t *optp, struct mbuf *m, int off) 1068 { 1069 struct ip6_hdr *ip6; 1070 1071 switch (IP6OPT_TYPE(*optp)) { 1072 case IP6OPT_TYPE_SKIP: /* ignore the option */ 1073 return((int)*(optp + 1)); 1074 case IP6OPT_TYPE_DISCARD: /* silently discard */ 1075 m_freem(m); 1076 return(-1); 1077 case IP6OPT_TYPE_FORCEICMP: /* send ICMP even if multicasted */ 1078 ip6stat.ip6s_badoptions++; 1079 icmp6_error(m, ICMP6_PARAM_PROB, ICMP6_PARAMPROB_OPTION, off); 1080 return(-1); 1081 case IP6OPT_TYPE_ICMP: /* send ICMP if not multicasted */ 1082 ip6stat.ip6s_badoptions++; 1083 ip6 = mtod(m, struct ip6_hdr *); 1084 if (IN6_IS_ADDR_MULTICAST(&ip6->ip6_dst) || 1085 (m->m_flags & (M_BCAST|M_MCAST))) 1086 m_freem(m); 1087 else 1088 icmp6_error(m, ICMP6_PARAM_PROB, 1089 ICMP6_PARAMPROB_OPTION, off); 1090 return(-1); 1091 } 1092 1093 m_freem(m); /* XXX: NOTREACHED */ 1094 return(-1); 1095 } 1096 1097 /* 1098 * Create the "control" list for this pcb. 1099 * The function will not modify mbuf chain at all. 1100 * 1101 * with KAME mbuf chain restriction: 1102 * The routine will be called from upper layer handlers like tcp6_input(). 1103 * Thus the routine assumes that the caller (tcp6_input) have already 1104 * called IP6_EXTHDR_CHECK() and all the extension headers are located in the 1105 * very first mbuf on the mbuf chain. 1106 */ 1107 void 1108 ip6_savecontrol(struct inpcb *in6p, struct mbuf **mp, struct ip6_hdr *ip6, 1109 struct mbuf *m) 1110 { 1111 struct thread *td = curthread; /* XXX */ 1112 int privileged = 0; 1113 int rthdr_exist = 0; 1114 1115 1116 if (suser(td) == 0) 1117 privileged++; 1118 1119 #ifdef SO_TIMESTAMP 1120 if ((in6p->in6p_socket->so_options & SO_TIMESTAMP) != 0) { 1121 struct timeval tv; 1122 1123 microtime(&tv); 1124 *mp = sbcreatecontrol((caddr_t) &tv, sizeof(tv), 1125 SCM_TIMESTAMP, SOL_SOCKET); 1126 if (*mp) { 1127 mp = &(*mp)->m_next; 1128 } 1129 } 1130 #endif 1131 1132 /* RFC 2292 sec. 5 */ 1133 if ((in6p->in6p_flags & IN6P_PKTINFO) != 0) { 1134 struct in6_pktinfo pi6; 1135 bcopy(&ip6->ip6_dst, &pi6.ipi6_addr, sizeof(struct in6_addr)); 1136 if (IN6_IS_SCOPE_LINKLOCAL(&pi6.ipi6_addr)) 1137 pi6.ipi6_addr.s6_addr16[1] = 0; 1138 pi6.ipi6_ifindex = (m && m->m_pkthdr.rcvif) 1139 ? m->m_pkthdr.rcvif->if_index 1140 : 0; 1141 *mp = sbcreatecontrol((caddr_t) &pi6, 1142 sizeof(struct in6_pktinfo), IPV6_PKTINFO, 1143 IPPROTO_IPV6); 1144 if (*mp) 1145 mp = &(*mp)->m_next; 1146 } 1147 1148 if ((in6p->in6p_flags & IN6P_HOPLIMIT) != 0) { 1149 int hlim = ip6->ip6_hlim & 0xff; 1150 *mp = sbcreatecontrol((caddr_t) &hlim, 1151 sizeof(int), IPV6_HOPLIMIT, IPPROTO_IPV6); 1152 if (*mp) 1153 mp = &(*mp)->m_next; 1154 } 1155 1156 /* 1157 * IPV6_HOPOPTS socket option. We require super-user privilege 1158 * for the option, but it might be too strict, since there might 1159 * be some hop-by-hop options which can be returned to normal user. 1160 * See RFC 2292 section 6. 1161 */ 1162 if ((in6p->in6p_flags & IN6P_HOPOPTS) != 0 && privileged) { 1163 /* 1164 * Check if a hop-by-hop options header is contatined in the 1165 * received packet, and if so, store the options as ancillary 1166 * data. Note that a hop-by-hop options header must be 1167 * just after the IPv6 header, which fact is assured through 1168 * the IPv6 input processing. 1169 */ 1170 struct ip6_hdr *ip6 = mtod(m, struct ip6_hdr *); 1171 if (ip6->ip6_nxt == IPPROTO_HOPOPTS) { 1172 struct ip6_hbh *hbh; 1173 int hbhlen = 0; 1174 #ifdef PULLDOWN_TEST 1175 struct mbuf *ext; 1176 #endif 1177 1178 #ifndef PULLDOWN_TEST 1179 hbh = (struct ip6_hbh *)(ip6 + 1); 1180 hbhlen = (hbh->ip6h_len + 1) << 3; 1181 #else 1182 ext = ip6_pullexthdr(m, sizeof(struct ip6_hdr), 1183 ip6->ip6_nxt); 1184 if (ext == NULL) { 1185 ip6stat.ip6s_tooshort++; 1186 return; 1187 } 1188 hbh = mtod(ext, struct ip6_hbh *); 1189 hbhlen = (hbh->ip6h_len + 1) << 3; 1190 if (hbhlen != ext->m_len) { 1191 m_freem(ext); 1192 ip6stat.ip6s_tooshort++; 1193 return; 1194 } 1195 #endif 1196 1197 /* 1198 * XXX: We copy whole the header even if a jumbo 1199 * payload option is included, which option is to 1200 * be removed before returning in the RFC 2292. 1201 * Note: this constraint is removed in 2292bis. 1202 */ 1203 *mp = sbcreatecontrol((caddr_t)hbh, hbhlen, 1204 IPV6_HOPOPTS, IPPROTO_IPV6); 1205 if (*mp) 1206 mp = &(*mp)->m_next; 1207 #ifdef PULLDOWN_TEST 1208 m_freem(ext); 1209 #endif 1210 } 1211 } 1212 1213 /* IPV6_DSTOPTS and IPV6_RTHDR socket options */ 1214 if ((in6p->in6p_flags & (IN6P_DSTOPTS | IN6P_RTHDRDSTOPTS)) != 0) { 1215 int proto, off, nxt; 1216 1217 /* 1218 * go through the header chain to see if a routing header is 1219 * contained in the packet. We need this information to store 1220 * destination options headers (if any) properly. 1221 * XXX: performance issue. We should record this info when 1222 * processing extension headers in incoming routine. 1223 * (todo) use m_aux? 1224 */ 1225 proto = IPPROTO_IPV6; 1226 off = 0; 1227 nxt = -1; 1228 while (1) { 1229 int newoff; 1230 1231 newoff = ip6_nexthdr(m, off, proto, &nxt); 1232 if (newoff < 0) 1233 break; 1234 if (newoff < off) /* invalid, check for safety */ 1235 break; 1236 if ((proto = nxt) == IPPROTO_ROUTING) { 1237 rthdr_exist = 1; 1238 break; 1239 } 1240 off = newoff; 1241 } 1242 } 1243 1244 if ((in6p->in6p_flags & 1245 (IN6P_RTHDR | IN6P_DSTOPTS | IN6P_RTHDRDSTOPTS)) != 0) { 1246 struct ip6_hdr *ip6 = mtod(m, struct ip6_hdr *); 1247 int nxt = ip6->ip6_nxt, off = sizeof(struct ip6_hdr); 1248 1249 /* 1250 * Search for destination options headers or routing 1251 * header(s) through the header chain, and stores each 1252 * header as ancillary data. 1253 * Note that the order of the headers remains in 1254 * the chain of ancillary data. 1255 */ 1256 while (1) { /* is explicit loop prevention necessary? */ 1257 struct ip6_ext *ip6e = NULL; 1258 int elen; 1259 #ifdef PULLDOWN_TEST 1260 struct mbuf *ext = NULL; 1261 #endif 1262 1263 /* 1264 * if it is not an extension header, don't try to 1265 * pull it from the chain. 1266 */ 1267 switch (nxt) { 1268 case IPPROTO_DSTOPTS: 1269 case IPPROTO_ROUTING: 1270 case IPPROTO_HOPOPTS: 1271 case IPPROTO_AH: /* is it possible? */ 1272 break; 1273 default: 1274 goto loopend; 1275 } 1276 1277 #ifndef PULLDOWN_TEST 1278 if (off + sizeof(*ip6e) > m->m_len) 1279 goto loopend; 1280 ip6e = (struct ip6_ext *)(mtod(m, caddr_t) + off); 1281 if (nxt == IPPROTO_AH) 1282 elen = (ip6e->ip6e_len + 2) << 2; 1283 else 1284 elen = (ip6e->ip6e_len + 1) << 3; 1285 if (off + elen > m->m_len) 1286 goto loopend; 1287 #else 1288 ext = ip6_pullexthdr(m, off, nxt); 1289 if (ext == NULL) { 1290 ip6stat.ip6s_tooshort++; 1291 return; 1292 } 1293 ip6e = mtod(ext, struct ip6_ext *); 1294 if (nxt == IPPROTO_AH) 1295 elen = (ip6e->ip6e_len + 2) << 2; 1296 else 1297 elen = (ip6e->ip6e_len + 1) << 3; 1298 if (elen != ext->m_len) { 1299 m_freem(ext); 1300 ip6stat.ip6s_tooshort++; 1301 return; 1302 } 1303 #endif 1304 1305 switch (nxt) { 1306 case IPPROTO_DSTOPTS: 1307 if ((in6p->in6p_flags & IN6P_DSTOPTS) == 0) 1308 break; 1309 1310 /* 1311 * We also require super-user privilege for 1312 * the option. 1313 * See the comments on IN6_HOPOPTS. 1314 */ 1315 if (!privileged) 1316 break; 1317 1318 *mp = sbcreatecontrol((caddr_t)ip6e, elen, 1319 IPV6_DSTOPTS, 1320 IPPROTO_IPV6); 1321 if (*mp) 1322 mp = &(*mp)->m_next; 1323 break; 1324 case IPPROTO_ROUTING: 1325 if (!in6p->in6p_flags & IN6P_RTHDR) 1326 break; 1327 1328 *mp = sbcreatecontrol((caddr_t)ip6e, elen, 1329 IPV6_RTHDR, 1330 IPPROTO_IPV6); 1331 if (*mp) 1332 mp = &(*mp)->m_next; 1333 break; 1334 case IPPROTO_HOPOPTS: 1335 case IPPROTO_AH: /* is it possible? */ 1336 break; 1337 1338 default: 1339 /* 1340 * other cases have been filtered in the above. 1341 * none will visit this case. here we supply 1342 * the code just in case (nxt overwritten or 1343 * other cases). 1344 */ 1345 #ifdef PULLDOWN_TEST 1346 m_freem(ext); 1347 #endif 1348 goto loopend; 1349 1350 } 1351 1352 /* proceed with the next header. */ 1353 off += elen; 1354 nxt = ip6e->ip6e_nxt; 1355 ip6e = NULL; 1356 #ifdef PULLDOWN_TEST 1357 m_freem(ext); 1358 ext = NULL; 1359 #endif 1360 } 1361 loopend: 1362 ; 1363 } 1364 1365 } 1366 1367 #ifdef PULLDOWN_TEST 1368 /* 1369 * pull single extension header from mbuf chain. returns single mbuf that 1370 * contains the result, or NULL on error. 1371 */ 1372 static struct mbuf * 1373 ip6_pullexthdr(struct mbuf *m, size_t off, int nxt) 1374 { 1375 struct ip6_ext ip6e; 1376 size_t elen; 1377 struct mbuf *n; 1378 1379 #ifdef DIAGNOSTIC 1380 switch (nxt) { 1381 case IPPROTO_DSTOPTS: 1382 case IPPROTO_ROUTING: 1383 case IPPROTO_HOPOPTS: 1384 case IPPROTO_AH: /* is it possible? */ 1385 break; 1386 default: 1387 printf("ip6_pullexthdr: invalid nxt=%d\n", nxt); 1388 } 1389 #endif 1390 1391 m_copydata(m, off, sizeof(ip6e), (caddr_t)&ip6e); 1392 if (nxt == IPPROTO_AH) 1393 elen = (ip6e.ip6e_len + 2) << 2; 1394 else 1395 elen = (ip6e.ip6e_len + 1) << 3; 1396 1397 MGET(n, MB_DONTWAIT, MT_DATA); 1398 if (n && elen >= MLEN) { 1399 MCLGET(n, MB_DONTWAIT); 1400 if ((n->m_flags & M_EXT) == 0) { 1401 m_free(n); 1402 n = NULL; 1403 } 1404 } 1405 if (!n) 1406 return NULL; 1407 1408 n->m_len = 0; 1409 if (elen >= M_TRAILINGSPACE(n)) { 1410 m_free(n); 1411 return NULL; 1412 } 1413 1414 m_copydata(m, off, elen, mtod(n, caddr_t)); 1415 n->m_len = elen; 1416 return n; 1417 } 1418 #endif 1419 1420 /* 1421 * Get pointer to the previous header followed by the header 1422 * currently processed. 1423 * XXX: This function supposes that 1424 * M includes all headers, 1425 * the next header field and the header length field of each header 1426 * are valid, and 1427 * the sum of each header length equals to OFF. 1428 * Because of these assumptions, this function must be called very 1429 * carefully. Moreover, it will not be used in the near future when 1430 * we develop `neater' mechanism to process extension headers. 1431 */ 1432 char * 1433 ip6_get_prevhdr(struct mbuf *m, int off) 1434 { 1435 struct ip6_hdr *ip6 = mtod(m, struct ip6_hdr *); 1436 1437 if (off == sizeof(struct ip6_hdr)) 1438 return(&ip6->ip6_nxt); 1439 else { 1440 int len, nxt; 1441 struct ip6_ext *ip6e = NULL; 1442 1443 nxt = ip6->ip6_nxt; 1444 len = sizeof(struct ip6_hdr); 1445 while (len < off) { 1446 ip6e = (struct ip6_ext *)(mtod(m, caddr_t) + len); 1447 1448 switch (nxt) { 1449 case IPPROTO_FRAGMENT: 1450 len += sizeof(struct ip6_frag); 1451 break; 1452 case IPPROTO_AH: 1453 len += (ip6e->ip6e_len + 2) << 2; 1454 break; 1455 default: 1456 len += (ip6e->ip6e_len + 1) << 3; 1457 break; 1458 } 1459 nxt = ip6e->ip6e_nxt; 1460 } 1461 if (ip6e) 1462 return(&ip6e->ip6e_nxt); 1463 else 1464 return NULL; 1465 } 1466 } 1467 1468 /* 1469 * get next header offset. m will be retained. 1470 */ 1471 int 1472 ip6_nexthdr(struct mbuf *m, int off, int proto, int *nxtp) 1473 { 1474 struct ip6_hdr ip6; 1475 struct ip6_ext ip6e; 1476 struct ip6_frag fh; 1477 1478 /* just in case */ 1479 if (m == NULL) 1480 panic("ip6_nexthdr: m == NULL"); 1481 if ((m->m_flags & M_PKTHDR) == 0 || m->m_pkthdr.len < off) 1482 return -1; 1483 1484 switch (proto) { 1485 case IPPROTO_IPV6: 1486 if (m->m_pkthdr.len < off + sizeof(ip6)) 1487 return -1; 1488 m_copydata(m, off, sizeof(ip6), (caddr_t)&ip6); 1489 if (nxtp) 1490 *nxtp = ip6.ip6_nxt; 1491 off += sizeof(ip6); 1492 return off; 1493 1494 case IPPROTO_FRAGMENT: 1495 /* 1496 * terminate parsing if it is not the first fragment, 1497 * it does not make sense to parse through it. 1498 */ 1499 if (m->m_pkthdr.len < off + sizeof(fh)) 1500 return -1; 1501 m_copydata(m, off, sizeof(fh), (caddr_t)&fh); 1502 if ((ntohs(fh.ip6f_offlg) & IP6F_OFF_MASK) != 0) 1503 return -1; 1504 if (nxtp) 1505 *nxtp = fh.ip6f_nxt; 1506 off += sizeof(struct ip6_frag); 1507 return off; 1508 1509 case IPPROTO_AH: 1510 if (m->m_pkthdr.len < off + sizeof(ip6e)) 1511 return -1; 1512 m_copydata(m, off, sizeof(ip6e), (caddr_t)&ip6e); 1513 if (nxtp) 1514 *nxtp = ip6e.ip6e_nxt; 1515 off += (ip6e.ip6e_len + 2) << 2; 1516 return off; 1517 1518 case IPPROTO_HOPOPTS: 1519 case IPPROTO_ROUTING: 1520 case IPPROTO_DSTOPTS: 1521 if (m->m_pkthdr.len < off + sizeof(ip6e)) 1522 return -1; 1523 m_copydata(m, off, sizeof(ip6e), (caddr_t)&ip6e); 1524 if (nxtp) 1525 *nxtp = ip6e.ip6e_nxt; 1526 off += (ip6e.ip6e_len + 1) << 3; 1527 return off; 1528 1529 case IPPROTO_NONE: 1530 case IPPROTO_ESP: 1531 case IPPROTO_IPCOMP: 1532 /* give up */ 1533 return -1; 1534 1535 default: 1536 return -1; 1537 } 1538 1539 return -1; 1540 } 1541 1542 /* 1543 * get offset for the last header in the chain. m will be kept untainted. 1544 */ 1545 int 1546 ip6_lasthdr(struct mbuf *m, int off, int proto, int *nxtp) 1547 { 1548 int newoff; 1549 int nxt; 1550 1551 if (!nxtp) { 1552 nxt = -1; 1553 nxtp = &nxt; 1554 } 1555 while (1) { 1556 newoff = ip6_nexthdr(m, off, proto, nxtp); 1557 if (newoff < 0) 1558 return off; 1559 else if (newoff < off) 1560 return -1; /* invalid */ 1561 else if (newoff == off) 1562 return newoff; 1563 1564 off = newoff; 1565 proto = *nxtp; 1566 } 1567 } 1568 1569 struct ip6aux * 1570 ip6_addaux(struct mbuf *m) 1571 { 1572 struct m_tag *tag = m_tag_find(m, PACKET_TAG_IPV6_INPUT, NULL); 1573 if (!tag) { 1574 tag = m_tag_get(PACKET_TAG_IPV6_INPUT, 1575 sizeof (struct ip6aux), 1576 MB_DONTWAIT); 1577 if (tag) 1578 m_tag_prepend(m, tag); 1579 } 1580 if (tag) 1581 bzero(tag+1, sizeof (struct ip6aux)); 1582 return tag ? (struct ip6aux*)(tag+1) : NULL; 1583 } 1584 1585 struct ip6aux * 1586 ip6_findaux(struct mbuf *m) 1587 { 1588 struct m_tag *tag = m_tag_find(m, PACKET_TAG_IPV6_INPUT, NULL); 1589 return tag ? (struct ip6aux*)(tag+1) : NULL; 1590 } 1591 1592 void 1593 ip6_delaux(struct mbuf *m) 1594 { 1595 struct m_tag *tag = m_tag_find(m, PACKET_TAG_IPV6_INPUT, NULL); 1596 if (tag) 1597 m_tag_delete(m, tag); 1598 } 1599 1600 /* 1601 * System control for IP6 1602 */ 1603 1604 u_char inet6ctlerrmap[PRC_NCMDS] = { 1605 0, 0, 0, 0, 1606 0, EMSGSIZE, EHOSTDOWN, EHOSTUNREACH, 1607 EHOSTUNREACH, EHOSTUNREACH, ECONNREFUSED, ECONNREFUSED, 1608 EMSGSIZE, EHOSTUNREACH, 0, 0, 1609 0, 0, 0, 0, 1610 ENOPROTOOPT 1611 }; 1612