1 /* $NetBSD: ip6_input.c,v 1.75 2004/06/01 03:13:22 itojun Exp $ */ 2 /* $KAME: ip6_input.c,v 1.188 2001/03/29 05:34:31 itojun Exp $ */ 3 4 /* 5 * Copyright (C) 1995, 1996, 1997, and 1998 WIDE Project. 6 * All rights reserved. 7 * 8 * Redistribution and use in source and binary forms, with or without 9 * modification, are permitted provided that the following conditions 10 * are met: 11 * 1. Redistributions of source code must retain the above copyright 12 * notice, this list of conditions and the following disclaimer. 13 * 2. Redistributions in binary form must reproduce the above copyright 14 * notice, this list of conditions and the following disclaimer in the 15 * documentation and/or other materials provided with the distribution. 16 * 3. Neither the name of the project nor the names of its contributors 17 * may be used to endorse or promote products derived from this software 18 * without specific prior written permission. 19 * 20 * THIS SOFTWARE IS PROVIDED BY THE PROJECT AND CONTRIBUTORS ``AS IS'' AND 21 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE 22 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE 23 * ARE DISCLAIMED. IN NO EVENT SHALL THE PROJECT OR CONTRIBUTORS BE LIABLE 24 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL 25 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS 26 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) 27 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT 28 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY 29 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF 30 * SUCH DAMAGE. 31 */ 32 33 /* 34 * Copyright (c) 1982, 1986, 1988, 1993 35 * The Regents of the University of California. All rights reserved. 36 * 37 * Redistribution and use in source and binary forms, with or without 38 * modification, are permitted provided that the following conditions 39 * are met: 40 * 1. Redistributions of source code must retain the above copyright 41 * notice, this list of conditions and the following disclaimer. 42 * 2. Redistributions in binary form must reproduce the above copyright 43 * notice, this list of conditions and the following disclaimer in the 44 * documentation and/or other materials provided with the distribution. 45 * 3. Neither the name of the University nor the names of its contributors 46 * may be used to endorse or promote products derived from this software 47 * without specific prior written permission. 48 * 49 * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND 50 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE 51 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE 52 * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE 53 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL 54 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS 55 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) 56 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT 57 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY 58 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF 59 * SUCH DAMAGE. 60 * 61 * @(#)ip_input.c 8.2 (Berkeley) 1/4/94 62 */ 63 64 #include <sys/cdefs.h> 65 __KERNEL_RCSID(0, "$NetBSD: ip6_input.c,v 1.75 2004/06/01 03:13:22 itojun Exp $"); 66 67 #include "opt_inet.h" 68 #include "opt_ipsec.h" 69 #include "opt_pfil_hooks.h" 70 71 #include <sys/param.h> 72 #include <sys/systm.h> 73 #include <sys/malloc.h> 74 #include <sys/mbuf.h> 75 #include <sys/domain.h> 76 #include <sys/protosw.h> 77 #include <sys/socket.h> 78 #include <sys/socketvar.h> 79 #include <sys/errno.h> 80 #include <sys/time.h> 81 #include <sys/kernel.h> 82 #include <sys/syslog.h> 83 #include <sys/proc.h> 84 #include <sys/sysctl.h> 85 86 #include <net/if.h> 87 #include <net/if_types.h> 88 #include <net/if_dl.h> 89 #include <net/route.h> 90 #include <net/netisr.h> 91 #ifdef PFIL_HOOKS 92 #include <net/pfil.h> 93 #endif 94 95 #include <netinet/in.h> 96 #include <netinet/in_systm.h> 97 #ifdef INET 98 #include <netinet/ip.h> 99 #include <netinet/ip_icmp.h> 100 #endif /* INET */ 101 #include <netinet/ip6.h> 102 #include <netinet6/in6_var.h> 103 #include <netinet6/ip6_var.h> 104 #include <netinet6/in6_pcb.h> 105 #include <netinet/icmp6.h> 106 #include <netinet6/in6_ifattach.h> 107 #include <netinet6/nd6.h> 108 109 #ifdef IPSEC 110 #include <netinet6/ipsec.h> 111 #endif 112 113 #include <netinet6/ip6protosw.h> 114 115 /* we need it for NLOOP. */ 116 #include "loop.h" 117 #include "faith.h" 118 #include "gif.h" 119 #include "bpfilter.h" 120 121 #if NGIF > 0 122 #include <netinet6/in6_gif.h> 123 #endif 124 125 #include <net/net_osdep.h> 126 127 extern struct domain inet6domain; 128 129 u_char ip6_protox[IPPROTO_MAX]; 130 static int ip6qmaxlen = IFQ_MAXLEN; 131 struct in6_ifaddr *in6_ifaddr; 132 struct ifqueue ip6intrq; 133 134 extern struct ifnet loif[NLOOP]; 135 int ip6_forward_srcrt; /* XXX */ 136 int ip6_sourcecheck; /* XXX */ 137 int ip6_sourcecheck_interval; /* XXX */ 138 139 #ifdef PFIL_HOOKS 140 struct pfil_head inet6_pfil_hook; 141 #endif 142 143 struct ip6stat ip6stat; 144 145 static void ip6_init2 __P((void *)); 146 147 static int ip6_hopopts_input __P((u_int32_t *, u_int32_t *, struct mbuf **, int *)); 148 static struct mbuf *ip6_pullexthdr __P((struct mbuf *, size_t, int)); 149 150 /* 151 * IP6 initialization: fill in IP6 protocol switch table. 152 * All protocols not implemented in kernel go to raw IP6 protocol handler. 153 */ 154 void 155 ip6_init() 156 { 157 struct ip6protosw *pr; 158 int i; 159 160 pr = (struct ip6protosw *)pffindproto(PF_INET6, IPPROTO_RAW, SOCK_RAW); 161 if (pr == 0) 162 panic("ip6_init"); 163 for (i = 0; i < IPPROTO_MAX; i++) 164 ip6_protox[i] = pr - inet6sw; 165 for (pr = (struct ip6protosw *)inet6domain.dom_protosw; 166 pr < (struct ip6protosw *)inet6domain.dom_protoswNPROTOSW; pr++) 167 if (pr->pr_domain->dom_family == PF_INET6 && 168 pr->pr_protocol && pr->pr_protocol != IPPROTO_RAW) 169 ip6_protox[pr->pr_protocol] = pr - inet6sw; 170 ip6intrq.ifq_maxlen = ip6qmaxlen; 171 nd6_init(); 172 frag6_init(); 173 174 ip6_init2((void *)0); 175 176 #ifdef PFIL_HOOKS 177 /* Register our Packet Filter hook. */ 178 inet6_pfil_hook.ph_type = PFIL_TYPE_AF; 179 inet6_pfil_hook.ph_af = AF_INET6; 180 i = pfil_head_register(&inet6_pfil_hook); 181 if (i != 0) 182 printf("ip6_init: WARNING: unable to register pfil hook, " 183 "error %d\n", i); 184 #endif /* PFIL_HOOKS */ 185 } 186 187 static void 188 ip6_init2(dummy) 189 void *dummy; 190 { 191 192 /* nd6_timer_init */ 193 callout_init(&nd6_timer_ch); 194 callout_reset(&nd6_timer_ch, hz, nd6_timer, NULL); 195 } 196 197 /* 198 * IP6 input interrupt handling. Just pass the packet to ip6_input. 199 */ 200 void 201 ip6intr() 202 { 203 int s; 204 struct mbuf *m; 205 206 for (;;) { 207 s = splnet(); 208 IF_DEQUEUE(&ip6intrq, m); 209 splx(s); 210 if (m == 0) 211 return; 212 ip6_input(m); 213 } 214 } 215 216 extern struct route_in6 ip6_forward_rt; 217 218 void 219 ip6_input(m) 220 struct mbuf *m; 221 { 222 struct ip6_hdr *ip6; 223 int off = sizeof(struct ip6_hdr), nest; 224 u_int32_t plen; 225 u_int32_t rtalert = ~0; 226 int nxt, ours = 0; 227 struct ifnet *deliverifp = NULL; 228 int srcrt = 0; 229 230 #ifdef IPSEC 231 /* 232 * should the inner packet be considered authentic? 233 * see comment in ah4_input(). 234 */ 235 m->m_flags &= ~M_AUTHIPHDR; 236 m->m_flags &= ~M_AUTHIPDGM; 237 #endif 238 239 /* 240 * mbuf statistics 241 */ 242 if (m->m_flags & M_EXT) { 243 if (m->m_next) 244 ip6stat.ip6s_mext2m++; 245 else 246 ip6stat.ip6s_mext1++; 247 } else { 248 #define M2MMAX (sizeof(ip6stat.ip6s_m2m)/sizeof(ip6stat.ip6s_m2m[0])) 249 if (m->m_next) { 250 if (m->m_flags & M_LOOP) { 251 ip6stat.ip6s_m2m[loif[0].if_index]++; /* XXX */ 252 } else if (m->m_pkthdr.rcvif->if_index < M2MMAX) 253 ip6stat.ip6s_m2m[m->m_pkthdr.rcvif->if_index]++; 254 else 255 ip6stat.ip6s_m2m[0]++; 256 } else 257 ip6stat.ip6s_m1++; 258 #undef M2MMAX 259 } 260 261 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_receive); 262 ip6stat.ip6s_total++; 263 264 /* 265 * If the IPv6 header is not aligned, slurp it up into a new 266 * mbuf with space for link headers, in the event we forward 267 * it. OTherwise, if it is aligned, make sure the entire base 268 * IPv6 header is in the first mbuf of the chain. 269 */ 270 if (IP6_HDR_ALIGNED_P(mtod(m, caddr_t)) == 0) { 271 struct ifnet *inifp = m->m_pkthdr.rcvif; 272 if ((m = m_copyup(m, sizeof(struct ip6_hdr), 273 (max_linkhdr + 3) & ~3)) == NULL) { 274 /* XXXJRT new stat, please */ 275 ip6stat.ip6s_toosmall++; 276 in6_ifstat_inc(inifp, ifs6_in_hdrerr); 277 return; 278 } 279 } else if (__predict_false(m->m_len < sizeof(struct ip6_hdr))) { 280 struct ifnet *inifp = m->m_pkthdr.rcvif; 281 if ((m = m_pullup(m, sizeof(struct ip6_hdr))) == NULL) { 282 ip6stat.ip6s_toosmall++; 283 in6_ifstat_inc(inifp, ifs6_in_hdrerr); 284 return; 285 } 286 } 287 288 ip6 = mtod(m, struct ip6_hdr *); 289 290 if ((ip6->ip6_vfc & IPV6_VERSION_MASK) != IPV6_VERSION) { 291 ip6stat.ip6s_badvers++; 292 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_hdrerr); 293 goto bad; 294 } 295 296 #ifdef PFIL_HOOKS 297 /* 298 * Run through list of hooks for input packets. If there are any 299 * filters which require that additional packets in the flow are 300 * not fast-forwarded, they must clear the M_CANFASTFWD flag. 301 * Note that filters must _never_ set this flag, as another filter 302 * in the list may have previously cleared it. 303 */ 304 /* 305 * let ipfilter look at packet on the wire, 306 * not the decapsulated packet. 307 */ 308 #ifdef IPSEC 309 if (!ipsec_getnhist(m)) 310 #else 311 if (1) 312 #endif 313 { 314 struct in6_addr odst; 315 316 odst = ip6->ip6_dst; 317 if (pfil_run_hooks(&inet6_pfil_hook, &m, m->m_pkthdr.rcvif, 318 PFIL_IN) != 0) 319 return; 320 if (m == NULL) 321 return; 322 ip6 = mtod(m, struct ip6_hdr *); 323 srcrt = !IN6_ARE_ADDR_EQUAL(&odst, &ip6->ip6_dst); 324 } 325 #endif /* PFIL_HOOKS */ 326 327 ip6stat.ip6s_nxthist[ip6->ip6_nxt]++; 328 329 #ifdef ALTQ 330 if (altq_input != NULL && (*altq_input)(m, AF_INET6) == 0) { 331 /* packet is dropped by traffic conditioner */ 332 return; 333 } 334 #endif 335 336 /* 337 * Check against address spoofing/corruption. 338 */ 339 if (IN6_IS_ADDR_MULTICAST(&ip6->ip6_src) || 340 IN6_IS_ADDR_UNSPECIFIED(&ip6->ip6_dst)) { 341 /* 342 * XXX: "badscope" is not very suitable for a multicast source. 343 */ 344 ip6stat.ip6s_badscope++; 345 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_addrerr); 346 goto bad; 347 } 348 /* 349 * The following check is not documented in specs. A malicious 350 * party may be able to use IPv4 mapped addr to confuse tcp/udp stack 351 * and bypass security checks (act as if it was from 127.0.0.1 by using 352 * IPv6 src ::ffff:127.0.0.1). Be cautious. 353 * 354 * This check chokes if we are in an SIIT cloud. As none of BSDs 355 * support IPv4-less kernel compilation, we cannot support SIIT 356 * environment at all. So, it makes more sense for us to reject any 357 * malicious packets for non-SIIT environment, than try to do a 358 * partial support for SIIT environment. 359 */ 360 if (IN6_IS_ADDR_V4MAPPED(&ip6->ip6_src) || 361 IN6_IS_ADDR_V4MAPPED(&ip6->ip6_dst)) { 362 ip6stat.ip6s_badscope++; 363 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_addrerr); 364 goto bad; 365 } 366 #if 0 367 /* 368 * Reject packets with IPv4 compatible addresses (auto tunnel). 369 * 370 * The code forbids auto tunnel relay case in RFC1933 (the check is 371 * stronger than RFC1933). We may want to re-enable it if mech-xx 372 * is revised to forbid relaying case. 373 */ 374 if (IN6_IS_ADDR_V4COMPAT(&ip6->ip6_src) || 375 IN6_IS_ADDR_V4COMPAT(&ip6->ip6_dst)) { 376 ip6stat.ip6s_badscope++; 377 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_addrerr); 378 goto bad; 379 } 380 #endif 381 382 if (IN6_IS_ADDR_LOOPBACK(&ip6->ip6_src) || 383 IN6_IS_ADDR_LOOPBACK(&ip6->ip6_dst)) { 384 if (m->m_pkthdr.rcvif->if_flags & IFF_LOOPBACK) { 385 ours = 1; 386 deliverifp = m->m_pkthdr.rcvif; 387 goto hbhcheck; 388 } else { 389 ip6stat.ip6s_badscope++; 390 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_addrerr); 391 goto bad; 392 } 393 } 394 395 /* drop packets if interface ID portion is already filled */ 396 if ((m->m_pkthdr.rcvif->if_flags & IFF_LOOPBACK) == 0) { 397 if (IN6_IS_SCOPE_LINKLOCAL(&ip6->ip6_src) && 398 ip6->ip6_src.s6_addr16[1]) { 399 ip6stat.ip6s_badscope++; 400 goto bad; 401 } 402 if (IN6_IS_SCOPE_LINKLOCAL(&ip6->ip6_dst) && 403 ip6->ip6_dst.s6_addr16[1]) { 404 ip6stat.ip6s_badscope++; 405 goto bad; 406 } 407 } 408 409 if (IN6_IS_SCOPE_LINKLOCAL(&ip6->ip6_src)) 410 ip6->ip6_src.s6_addr16[1] 411 = htons(m->m_pkthdr.rcvif->if_index); 412 if (IN6_IS_SCOPE_LINKLOCAL(&ip6->ip6_dst)) 413 ip6->ip6_dst.s6_addr16[1] 414 = htons(m->m_pkthdr.rcvif->if_index); 415 416 /* 417 * We use rt->rt_ifp to determine if the address is ours or not. 418 * If rt_ifp is lo0, the address is ours. 419 * The problem here is, rt->rt_ifp for fe80::%lo0/64 is set to lo0, 420 * so any address under fe80::%lo0/64 will be mistakenly considered 421 * local. The special case is supplied to handle the case properly 422 * by actually looking at interface addresses 423 * (using in6ifa_ifpwithaddr). 424 */ 425 if ((m->m_pkthdr.rcvif->if_flags & IFF_LOOPBACK) != 0 && 426 IN6_IS_ADDR_LINKLOCAL(&ip6->ip6_dst)) { 427 if (!in6ifa_ifpwithaddr(m->m_pkthdr.rcvif, &ip6->ip6_dst)) { 428 icmp6_error(m, ICMP6_DST_UNREACH, 429 ICMP6_DST_UNREACH_ADDR, 0); 430 /* m is already freed */ 431 return; 432 } 433 434 ours = 1; 435 deliverifp = m->m_pkthdr.rcvif; 436 goto hbhcheck; 437 } 438 439 /* 440 * Multicast check 441 */ 442 if (IN6_IS_ADDR_MULTICAST(&ip6->ip6_dst)) { 443 struct in6_multi *in6m = 0; 444 445 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_mcast); 446 /* 447 * See if we belong to the destination multicast group on the 448 * arrival interface. 449 */ 450 IN6_LOOKUP_MULTI(ip6->ip6_dst, m->m_pkthdr.rcvif, in6m); 451 if (in6m) 452 ours = 1; 453 else if (!ip6_mrouter) { 454 ip6stat.ip6s_notmember++; 455 ip6stat.ip6s_cantforward++; 456 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_discard); 457 goto bad; 458 } 459 deliverifp = m->m_pkthdr.rcvif; 460 goto hbhcheck; 461 } 462 463 /* 464 * Unicast check 465 */ 466 if (ip6_forward_rt.ro_rt != NULL && 467 (ip6_forward_rt.ro_rt->rt_flags & RTF_UP) != 0 && 468 IN6_ARE_ADDR_EQUAL(&ip6->ip6_dst, 469 &((struct sockaddr_in6 *)(&ip6_forward_rt.ro_dst))->sin6_addr)) 470 ip6stat.ip6s_forward_cachehit++; 471 else { 472 struct sockaddr_in6 *dst6; 473 474 if (ip6_forward_rt.ro_rt) { 475 /* route is down or destination is different */ 476 ip6stat.ip6s_forward_cachemiss++; 477 RTFREE(ip6_forward_rt.ro_rt); 478 ip6_forward_rt.ro_rt = 0; 479 } 480 481 bzero(&ip6_forward_rt.ro_dst, sizeof(struct sockaddr_in6)); 482 dst6 = (struct sockaddr_in6 *)&ip6_forward_rt.ro_dst; 483 dst6->sin6_len = sizeof(struct sockaddr_in6); 484 dst6->sin6_family = AF_INET6; 485 dst6->sin6_addr = ip6->ip6_dst; 486 487 rtalloc((struct route *)&ip6_forward_rt); 488 } 489 490 #define rt6_key(r) ((struct sockaddr_in6 *)((r)->rt_nodes->rn_key)) 491 492 /* 493 * Accept the packet if the forwarding interface to the destination 494 * according to the routing table is the loopback interface, 495 * unless the associated route has a gateway. 496 * Note that this approach causes to accept a packet if there is a 497 * route to the loopback interface for the destination of the packet. 498 * But we think it's even useful in some situations, e.g. when using 499 * a special daemon which wants to intercept the packet. 500 */ 501 if (ip6_forward_rt.ro_rt && 502 (ip6_forward_rt.ro_rt->rt_flags & 503 (RTF_HOST|RTF_GATEWAY)) == RTF_HOST && 504 !(ip6_forward_rt.ro_rt->rt_flags & RTF_CLONED) && 505 #if 0 506 /* 507 * The check below is redundant since the comparison of 508 * the destination and the key of the rtentry has 509 * already done through looking up the routing table. 510 */ 511 IN6_ARE_ADDR_EQUAL(&ip6->ip6_dst, 512 &rt6_key(ip6_forward_rt.ro_rt)->sin6_addr) && 513 #endif 514 ip6_forward_rt.ro_rt->rt_ifp->if_type == IFT_LOOP) { 515 struct in6_ifaddr *ia6 = 516 (struct in6_ifaddr *)ip6_forward_rt.ro_rt->rt_ifa; 517 if (ia6->ia6_flags & IN6_IFF_ANYCAST) 518 m->m_flags |= M_ANYCAST6; 519 /* 520 * packets to a tentative, duplicated, or somehow invalid 521 * address must not be accepted. 522 */ 523 if (!(ia6->ia6_flags & IN6_IFF_NOTREADY)) { 524 /* this address is ready */ 525 ours = 1; 526 deliverifp = ia6->ia_ifp; /* correct? */ 527 goto hbhcheck; 528 } else { 529 /* address is not ready, so discard the packet. */ 530 nd6log((LOG_INFO, 531 "ip6_input: packet to an unready address %s->%s\n", 532 ip6_sprintf(&ip6->ip6_src), 533 ip6_sprintf(&ip6->ip6_dst))); 534 535 goto bad; 536 } 537 } 538 539 /* 540 * FAITH (Firewall Aided Internet Translator) 541 */ 542 #if defined(NFAITH) && 0 < NFAITH 543 if (ip6_keepfaith) { 544 if (ip6_forward_rt.ro_rt && ip6_forward_rt.ro_rt->rt_ifp && 545 ip6_forward_rt.ro_rt->rt_ifp->if_type == IFT_FAITH) { 546 /* XXX do we need more sanity checks? */ 547 ours = 1; 548 deliverifp = ip6_forward_rt.ro_rt->rt_ifp; /* faith */ 549 goto hbhcheck; 550 } 551 } 552 #endif 553 554 #if 0 555 { 556 /* 557 * Last resort: check in6_ifaddr for incoming interface. 558 * The code is here until I update the "goto ours hack" code above 559 * working right. 560 */ 561 struct ifaddr *ifa; 562 for (ifa = m->m_pkthdr.rcvif->if_addrlist.tqh_first; 563 ifa; 564 ifa = ifa->ifa_list.tqe_next) { 565 if (ifa->ifa_addr == NULL) 566 continue; /* just for safety */ 567 if (ifa->ifa_addr->sa_family != AF_INET6) 568 continue; 569 if (IN6_ARE_ADDR_EQUAL(IFA_IN6(ifa), &ip6->ip6_dst)) { 570 ours = 1; 571 deliverifp = ifa->ifa_ifp; 572 goto hbhcheck; 573 } 574 } 575 } 576 #endif 577 578 /* 579 * Now there is no reason to process the packet if it's not our own 580 * and we're not a router. 581 */ 582 if (!ip6_forwarding) { 583 ip6stat.ip6s_cantforward++; 584 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_discard); 585 goto bad; 586 } 587 588 hbhcheck: 589 /* 590 * Process Hop-by-Hop options header if it's contained. 591 * m may be modified in ip6_hopopts_input(). 592 * If a JumboPayload option is included, plen will also be modified. 593 */ 594 plen = (u_int32_t)ntohs(ip6->ip6_plen); 595 if (ip6->ip6_nxt == IPPROTO_HOPOPTS) { 596 struct ip6_hbh *hbh; 597 598 if (ip6_hopopts_input(&plen, &rtalert, &m, &off)) { 599 #if 0 /*touches NULL pointer*/ 600 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_discard); 601 #endif 602 return; /* m have already been freed */ 603 } 604 605 /* adjust pointer */ 606 ip6 = mtod(m, struct ip6_hdr *); 607 608 /* 609 * if the payload length field is 0 and the next header field 610 * indicates Hop-by-Hop Options header, then a Jumbo Payload 611 * option MUST be included. 612 */ 613 if (ip6->ip6_plen == 0 && plen == 0) { 614 /* 615 * Note that if a valid jumbo payload option is 616 * contained, ip6_hoptops_input() must set a valid 617 * (non-zero) payload length to the variable plen. 618 */ 619 ip6stat.ip6s_badoptions++; 620 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_discard); 621 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_hdrerr); 622 icmp6_error(m, ICMP6_PARAM_PROB, 623 ICMP6_PARAMPROB_HEADER, 624 (caddr_t)&ip6->ip6_plen - (caddr_t)ip6); 625 return; 626 } 627 IP6_EXTHDR_GET(hbh, struct ip6_hbh *, m, sizeof(struct ip6_hdr), 628 sizeof(struct ip6_hbh)); 629 if (hbh == NULL) { 630 ip6stat.ip6s_tooshort++; 631 return; 632 } 633 KASSERT(IP6_HDR_ALIGNED_P(hbh)); 634 nxt = hbh->ip6h_nxt; 635 636 /* 637 * accept the packet if a router alert option is included 638 * and we act as an IPv6 router. 639 */ 640 if (rtalert != ~0 && ip6_forwarding) 641 ours = 1; 642 } else 643 nxt = ip6->ip6_nxt; 644 645 /* 646 * Check that the amount of data in the buffers 647 * is as at least much as the IPv6 header would have us expect. 648 * Trim mbufs if longer than we expect. 649 * Drop packet if shorter than we expect. 650 */ 651 if (m->m_pkthdr.len - sizeof(struct ip6_hdr) < plen) { 652 ip6stat.ip6s_tooshort++; 653 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_truncated); 654 goto bad; 655 } 656 if (m->m_pkthdr.len > sizeof(struct ip6_hdr) + plen) { 657 if (m->m_len == m->m_pkthdr.len) { 658 m->m_len = sizeof(struct ip6_hdr) + plen; 659 m->m_pkthdr.len = sizeof(struct ip6_hdr) + plen; 660 } else 661 m_adj(m, sizeof(struct ip6_hdr) + plen - m->m_pkthdr.len); 662 } 663 664 /* 665 * Forward if desirable. 666 */ 667 if (IN6_IS_ADDR_MULTICAST(&ip6->ip6_dst)) { 668 /* 669 * If we are acting as a multicast router, all 670 * incoming multicast packets are passed to the 671 * kernel-level multicast forwarding function. 672 * The packet is returned (relatively) intact; if 673 * ip6_mforward() returns a non-zero value, the packet 674 * must be discarded, else it may be accepted below. 675 */ 676 if (ip6_mrouter && ip6_mforward(ip6, m->m_pkthdr.rcvif, m)) { 677 ip6stat.ip6s_cantforward++; 678 m_freem(m); 679 return; 680 } 681 if (!ours) { 682 m_freem(m); 683 return; 684 } 685 } else if (!ours) { 686 ip6_forward(m, srcrt); 687 return; 688 } 689 690 ip6 = mtod(m, struct ip6_hdr *); 691 692 /* 693 * Malicious party may be able to use IPv4 mapped addr to confuse 694 * tcp/udp stack and bypass security checks (act as if it was from 695 * 127.0.0.1 by using IPv6 src ::ffff:127.0.0.1). Be cautious. 696 * 697 * For SIIT end node behavior, you may want to disable the check. 698 * However, you will become vulnerable to attacks using IPv4 mapped 699 * source. 700 */ 701 if (IN6_IS_ADDR_V4MAPPED(&ip6->ip6_src) || 702 IN6_IS_ADDR_V4MAPPED(&ip6->ip6_dst)) { 703 ip6stat.ip6s_badscope++; 704 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_addrerr); 705 goto bad; 706 } 707 708 /* 709 * Tell launch routine the next header 710 */ 711 #ifdef IFA_STATS 712 if (deliverifp != NULL) { 713 struct in6_ifaddr *ia6; 714 ia6 = in6_ifawithifp(deliverifp, &ip6->ip6_dst); 715 if (ia6) 716 ia6->ia_ifa.ifa_data.ifad_inbytes += m->m_pkthdr.len; 717 } 718 #endif 719 ip6stat.ip6s_delivered++; 720 in6_ifstat_inc(deliverifp, ifs6_in_deliver); 721 nest = 0; 722 723 while (nxt != IPPROTO_DONE) { 724 if (ip6_hdrnestlimit && (++nest > ip6_hdrnestlimit)) { 725 ip6stat.ip6s_toomanyhdr++; 726 goto bad; 727 } 728 729 /* 730 * protection against faulty packet - there should be 731 * more sanity checks in header chain processing. 732 */ 733 if (m->m_pkthdr.len < off) { 734 ip6stat.ip6s_tooshort++; 735 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_truncated); 736 goto bad; 737 } 738 739 #ifdef IPSEC 740 /* 741 * enforce IPsec policy checking if we are seeing last header. 742 * note that we do not visit this with protocols with pcb layer 743 * code - like udp/tcp/raw ip. 744 */ 745 if ((inet6sw[ip6_protox[nxt]].pr_flags & PR_LASTHDR) != 0 && 746 ipsec6_in_reject(m, NULL)) { 747 ipsec6stat.in_polvio++; 748 goto bad; 749 } 750 #endif 751 752 nxt = (*inet6sw[ip6_protox[nxt]].pr_input)(&m, &off, nxt); 753 } 754 return; 755 bad: 756 m_freem(m); 757 } 758 759 /* 760 * Hop-by-Hop options header processing. If a valid jumbo payload option is 761 * included, the real payload length will be stored in plenp. 762 */ 763 static int 764 ip6_hopopts_input(plenp, rtalertp, mp, offp) 765 u_int32_t *plenp; 766 u_int32_t *rtalertp; /* XXX: should be stored more smart way */ 767 struct mbuf **mp; 768 int *offp; 769 { 770 struct mbuf *m = *mp; 771 int off = *offp, hbhlen; 772 struct ip6_hbh *hbh; 773 774 /* validation of the length of the header */ 775 IP6_EXTHDR_GET(hbh, struct ip6_hbh *, m, 776 sizeof(struct ip6_hdr), sizeof(struct ip6_hbh)); 777 if (hbh == NULL) { 778 ip6stat.ip6s_tooshort++; 779 return -1; 780 } 781 hbhlen = (hbh->ip6h_len + 1) << 3; 782 IP6_EXTHDR_GET(hbh, struct ip6_hbh *, m, sizeof(struct ip6_hdr), 783 hbhlen); 784 if (hbh == NULL) { 785 ip6stat.ip6s_tooshort++; 786 return -1; 787 } 788 KASSERT(IP6_HDR_ALIGNED_P(hbh)); 789 off += hbhlen; 790 hbhlen -= sizeof(struct ip6_hbh); 791 792 if (ip6_process_hopopts(m, (u_int8_t *)hbh + sizeof(struct ip6_hbh), 793 hbhlen, rtalertp, plenp) < 0) 794 return (-1); 795 796 *offp = off; 797 *mp = m; 798 return (0); 799 } 800 801 /* 802 * Search header for all Hop-by-hop options and process each option. 803 * This function is separate from ip6_hopopts_input() in order to 804 * handle a case where the sending node itself process its hop-by-hop 805 * options header. In such a case, the function is called from ip6_output(). 806 * 807 * The function assumes that hbh header is located right after the IPv6 header 808 * (RFC2460 p7), opthead is pointer into data content in m, and opthead to 809 * opthead + hbhlen is located in continuous memory region. 810 */ 811 int 812 ip6_process_hopopts(m, opthead, hbhlen, rtalertp, plenp) 813 struct mbuf *m; 814 u_int8_t *opthead; 815 int hbhlen; 816 u_int32_t *rtalertp; 817 u_int32_t *plenp; 818 { 819 struct ip6_hdr *ip6; 820 int optlen = 0; 821 u_int8_t *opt = opthead; 822 u_int16_t rtalert_val; 823 u_int32_t jumboplen; 824 const int erroff = sizeof(struct ip6_hdr) + sizeof(struct ip6_hbh); 825 826 for (; hbhlen > 0; hbhlen -= optlen, opt += optlen) { 827 switch (*opt) { 828 case IP6OPT_PAD1: 829 optlen = 1; 830 break; 831 case IP6OPT_PADN: 832 if (hbhlen < IP6OPT_MINLEN) { 833 ip6stat.ip6s_toosmall++; 834 goto bad; 835 } 836 optlen = *(opt + 1) + 2; 837 break; 838 case IP6OPT_RTALERT: 839 /* XXX may need check for alignment */ 840 if (hbhlen < IP6OPT_RTALERT_LEN) { 841 ip6stat.ip6s_toosmall++; 842 goto bad; 843 } 844 if (*(opt + 1) != IP6OPT_RTALERT_LEN - 2) { 845 /* XXX stat */ 846 icmp6_error(m, ICMP6_PARAM_PROB, 847 ICMP6_PARAMPROB_HEADER, 848 erroff + opt + 1 - opthead); 849 return (-1); 850 } 851 optlen = IP6OPT_RTALERT_LEN; 852 bcopy((caddr_t)(opt + 2), (caddr_t)&rtalert_val, 2); 853 *rtalertp = ntohs(rtalert_val); 854 break; 855 case IP6OPT_JUMBO: 856 /* XXX may need check for alignment */ 857 if (hbhlen < IP6OPT_JUMBO_LEN) { 858 ip6stat.ip6s_toosmall++; 859 goto bad; 860 } 861 if (*(opt + 1) != IP6OPT_JUMBO_LEN - 2) { 862 /* XXX stat */ 863 icmp6_error(m, ICMP6_PARAM_PROB, 864 ICMP6_PARAMPROB_HEADER, 865 erroff + opt + 1 - opthead); 866 return (-1); 867 } 868 optlen = IP6OPT_JUMBO_LEN; 869 870 /* 871 * IPv6 packets that have non 0 payload length 872 * must not contain a jumbo payload option. 873 */ 874 ip6 = mtod(m, struct ip6_hdr *); 875 if (ip6->ip6_plen) { 876 ip6stat.ip6s_badoptions++; 877 icmp6_error(m, ICMP6_PARAM_PROB, 878 ICMP6_PARAMPROB_HEADER, 879 erroff + opt - opthead); 880 return (-1); 881 } 882 883 /* 884 * We may see jumbolen in unaligned location, so 885 * we'd need to perform bcopy(). 886 */ 887 bcopy(opt + 2, &jumboplen, sizeof(jumboplen)); 888 jumboplen = (u_int32_t)htonl(jumboplen); 889 890 #if 1 891 /* 892 * if there are multiple jumbo payload options, 893 * *plenp will be non-zero and the packet will be 894 * rejected. 895 * the behavior may need some debate in ipngwg - 896 * multiple options does not make sense, however, 897 * there's no explicit mention in specification. 898 */ 899 if (*plenp != 0) { 900 ip6stat.ip6s_badoptions++; 901 icmp6_error(m, ICMP6_PARAM_PROB, 902 ICMP6_PARAMPROB_HEADER, 903 erroff + opt + 2 - opthead); 904 return (-1); 905 } 906 #endif 907 908 /* 909 * jumbo payload length must be larger than 65535. 910 */ 911 if (jumboplen <= IPV6_MAXPACKET) { 912 ip6stat.ip6s_badoptions++; 913 icmp6_error(m, ICMP6_PARAM_PROB, 914 ICMP6_PARAMPROB_HEADER, 915 erroff + opt + 2 - opthead); 916 return (-1); 917 } 918 *plenp = jumboplen; 919 920 break; 921 default: /* unknown option */ 922 if (hbhlen < IP6OPT_MINLEN) { 923 ip6stat.ip6s_toosmall++; 924 goto bad; 925 } 926 optlen = ip6_unknown_opt(opt, m, 927 erroff + opt - opthead); 928 if (optlen == -1) 929 return (-1); 930 optlen += 2; 931 break; 932 } 933 } 934 935 return (0); 936 937 bad: 938 m_freem(m); 939 return (-1); 940 } 941 942 /* 943 * Unknown option processing. 944 * The third argument `off' is the offset from the IPv6 header to the option, 945 * which is necessary if the IPv6 header the and option header and IPv6 header 946 * is not continuous in order to return an ICMPv6 error. 947 */ 948 int 949 ip6_unknown_opt(optp, m, off) 950 u_int8_t *optp; 951 struct mbuf *m; 952 int off; 953 { 954 struct ip6_hdr *ip6; 955 956 switch (IP6OPT_TYPE(*optp)) { 957 case IP6OPT_TYPE_SKIP: /* ignore the option */ 958 return ((int)*(optp + 1)); 959 case IP6OPT_TYPE_DISCARD: /* silently discard */ 960 m_freem(m); 961 return (-1); 962 case IP6OPT_TYPE_FORCEICMP: /* send ICMP even if multicasted */ 963 ip6stat.ip6s_badoptions++; 964 icmp6_error(m, ICMP6_PARAM_PROB, ICMP6_PARAMPROB_OPTION, off); 965 return (-1); 966 case IP6OPT_TYPE_ICMP: /* send ICMP if not multicasted */ 967 ip6stat.ip6s_badoptions++; 968 ip6 = mtod(m, struct ip6_hdr *); 969 if (IN6_IS_ADDR_MULTICAST(&ip6->ip6_dst) || 970 (m->m_flags & (M_BCAST|M_MCAST))) 971 m_freem(m); 972 else 973 icmp6_error(m, ICMP6_PARAM_PROB, 974 ICMP6_PARAMPROB_OPTION, off); 975 return (-1); 976 } 977 978 m_freem(m); /* XXX: NOTREACHED */ 979 return (-1); 980 } 981 982 /* 983 * Create the "control" list for this pcb. 984 * 985 * The routine will be called from upper layer handlers like tcp6_input(). 986 * Thus the routine assumes that the caller (tcp6_input) have already 987 * called IP6_EXTHDR_CHECK() and all the extension headers are located in the 988 * very first mbuf on the mbuf chain. 989 * We may want to add some infinite loop prevention or sanity checks for safety. 990 * (This applies only when you are using KAME mbuf chain restriction, i.e. 991 * you are using IP6_EXTHDR_CHECK() not m_pulldown()) 992 */ 993 void 994 ip6_savecontrol(in6p, mp, ip6, m) 995 struct in6pcb *in6p; 996 struct mbuf **mp; 997 struct ip6_hdr *ip6; 998 struct mbuf *m; 999 { 1000 1001 #ifdef SO_TIMESTAMP 1002 if (in6p->in6p_socket->so_options & SO_TIMESTAMP) { 1003 struct timeval tv; 1004 1005 microtime(&tv); 1006 *mp = sbcreatecontrol((caddr_t) &tv, sizeof(tv), 1007 SCM_TIMESTAMP, SOL_SOCKET); 1008 if (*mp) 1009 mp = &(*mp)->m_next; 1010 } 1011 #endif 1012 if (in6p->in6p_flags & IN6P_RECVDSTADDR) { 1013 *mp = sbcreatecontrol((caddr_t) &ip6->ip6_dst, 1014 sizeof(struct in6_addr), IPV6_RECVDSTADDR, IPPROTO_IPV6); 1015 if (*mp) 1016 mp = &(*mp)->m_next; 1017 } 1018 1019 #ifdef noyet 1020 /* options were tossed above */ 1021 if (in6p->in6p_flags & IN6P_RECVOPTS) 1022 /* broken */ 1023 /* ip6_srcroute doesn't do what we want here, need to fix */ 1024 if (in6p->in6p_flags & IPV6P_RECVRETOPTS) 1025 /* broken */ 1026 #endif 1027 1028 /* RFC 2292 sec. 5 */ 1029 if ((in6p->in6p_flags & IN6P_PKTINFO) != 0) { 1030 struct in6_pktinfo pi6; 1031 bcopy(&ip6->ip6_dst, &pi6.ipi6_addr, sizeof(struct in6_addr)); 1032 if (IN6_IS_SCOPE_LINKLOCAL(&pi6.ipi6_addr)) 1033 pi6.ipi6_addr.s6_addr16[1] = 0; 1034 pi6.ipi6_ifindex = (m && m->m_pkthdr.rcvif) 1035 ? m->m_pkthdr.rcvif->if_index 1036 : 0; 1037 *mp = sbcreatecontrol((caddr_t) &pi6, 1038 sizeof(struct in6_pktinfo), IPV6_PKTINFO, IPPROTO_IPV6); 1039 if (*mp) 1040 mp = &(*mp)->m_next; 1041 } 1042 if (in6p->in6p_flags & IN6P_HOPLIMIT) { 1043 int hlim = ip6->ip6_hlim & 0xff; 1044 *mp = sbcreatecontrol((caddr_t) &hlim, sizeof(int), 1045 IPV6_HOPLIMIT, IPPROTO_IPV6); 1046 if (*mp) 1047 mp = &(*mp)->m_next; 1048 } 1049 /* IN6P_NEXTHOP - for outgoing packet only */ 1050 1051 /* 1052 * IPV6_HOPOPTS socket option. Recall that we required super-user 1053 * privilege for the option (see ip6_ctloutput), but it might be too 1054 * strict, since there might be some hop-by-hop options which can be 1055 * returned to normal user. 1056 * See also RFC 2292 section 6. 1057 */ 1058 if ((in6p->in6p_flags & IN6P_HOPOPTS) != 0) { 1059 /* 1060 * Check if a hop-by-hop options header is contatined in the 1061 * received packet, and if so, store the options as ancillary 1062 * data. Note that a hop-by-hop options header must be 1063 * just after the IPv6 header, which fact is assured through 1064 * the IPv6 input processing. 1065 */ 1066 struct ip6_hdr *ip6 = mtod(m, struct ip6_hdr *); 1067 if (ip6->ip6_nxt == IPPROTO_HOPOPTS) { 1068 struct ip6_hbh *hbh; 1069 int hbhlen; 1070 struct mbuf *ext; 1071 1072 ext = ip6_pullexthdr(m, sizeof(struct ip6_hdr), 1073 ip6->ip6_nxt); 1074 if (ext == NULL) { 1075 ip6stat.ip6s_tooshort++; 1076 return; 1077 } 1078 hbh = mtod(ext, struct ip6_hbh *); 1079 hbhlen = (hbh->ip6h_len + 1) << 3; 1080 if (hbhlen != ext->m_len) { 1081 m_freem(ext); 1082 ip6stat.ip6s_tooshort++; 1083 return; 1084 } 1085 1086 /* 1087 * XXX: We copy whole the header even if a jumbo 1088 * payload option is included, which option is to 1089 * be removed before returning in the RFC 2292. 1090 * But it's too painful operation... 1091 */ 1092 *mp = sbcreatecontrol((caddr_t)hbh, hbhlen, 1093 IPV6_HOPOPTS, IPPROTO_IPV6); 1094 if (*mp) 1095 mp = &(*mp)->m_next; 1096 m_freem(ext); 1097 } 1098 } 1099 1100 /* IPV6_DSTOPTS and IPV6_RTHDR socket options */ 1101 if (in6p->in6p_flags & (IN6P_DSTOPTS | IN6P_RTHDR)) { 1102 struct ip6_hdr *ip6 = mtod(m, struct ip6_hdr *); 1103 int nxt = ip6->ip6_nxt, off = sizeof(struct ip6_hdr); 1104 1105 /* 1106 * Search for destination options headers or routing 1107 * header(s) through the header chain, and stores each 1108 * header as ancillary data. 1109 * Note that the order of the headers remains in 1110 * the chain of ancillary data. 1111 */ 1112 while (1) { /* is explicit loop prevention necessary? */ 1113 struct ip6_ext *ip6e = NULL; 1114 int elen; 1115 struct mbuf *ext = NULL; 1116 1117 /* 1118 * if it is not an extension header, don't try to 1119 * pull it from the chain. 1120 */ 1121 switch (nxt) { 1122 case IPPROTO_DSTOPTS: 1123 case IPPROTO_ROUTING: 1124 case IPPROTO_HOPOPTS: 1125 case IPPROTO_AH: /* is it possible? */ 1126 break; 1127 default: 1128 goto loopend; 1129 } 1130 1131 ext = ip6_pullexthdr(m, off, nxt); 1132 if (ext == NULL) { 1133 ip6stat.ip6s_tooshort++; 1134 return; 1135 } 1136 ip6e = mtod(ext, struct ip6_ext *); 1137 if (nxt == IPPROTO_AH) 1138 elen = (ip6e->ip6e_len + 2) << 2; 1139 else 1140 elen = (ip6e->ip6e_len + 1) << 3; 1141 if (elen != ext->m_len) { 1142 m_freem(ext); 1143 ip6stat.ip6s_tooshort++; 1144 return; 1145 } 1146 KASSERT(IP6_HDR_ALIGNED_P(ip6e)); 1147 1148 switch (nxt) { 1149 case IPPROTO_DSTOPTS: 1150 if (!in6p->in6p_flags & IN6P_DSTOPTS) 1151 break; 1152 1153 *mp = sbcreatecontrol((caddr_t)ip6e, elen, 1154 IPV6_DSTOPTS, IPPROTO_IPV6); 1155 if (*mp) 1156 mp = &(*mp)->m_next; 1157 break; 1158 1159 case IPPROTO_ROUTING: 1160 if (!in6p->in6p_flags & IN6P_RTHDR) 1161 break; 1162 1163 *mp = sbcreatecontrol((caddr_t)ip6e, elen, 1164 IPV6_RTHDR, IPPROTO_IPV6); 1165 if (*mp) 1166 mp = &(*mp)->m_next; 1167 break; 1168 1169 case IPPROTO_HOPOPTS: 1170 case IPPROTO_AH: /* is it possible? */ 1171 break; 1172 1173 default: 1174 /* 1175 * other cases have been filtered in the above. 1176 * none will visit this case. here we supply 1177 * the code just in case (nxt overwritten or 1178 * other cases). 1179 */ 1180 m_freem(ext); 1181 goto loopend; 1182 1183 } 1184 1185 /* proceed with the next header. */ 1186 off += elen; 1187 nxt = ip6e->ip6e_nxt; 1188 ip6e = NULL; 1189 m_freem(ext); 1190 ext = NULL; 1191 } 1192 loopend: 1193 ; 1194 } 1195 } 1196 1197 /* 1198 * pull single extension header from mbuf chain. returns single mbuf that 1199 * contains the result, or NULL on error. 1200 */ 1201 static struct mbuf * 1202 ip6_pullexthdr(m, off, nxt) 1203 struct mbuf *m; 1204 size_t off; 1205 int nxt; 1206 { 1207 struct ip6_ext ip6e; 1208 size_t elen; 1209 struct mbuf *n; 1210 1211 #ifdef DIAGNOSTIC 1212 switch (nxt) { 1213 case IPPROTO_DSTOPTS: 1214 case IPPROTO_ROUTING: 1215 case IPPROTO_HOPOPTS: 1216 case IPPROTO_AH: /* is it possible? */ 1217 break; 1218 default: 1219 printf("ip6_pullexthdr: invalid nxt=%d\n", nxt); 1220 } 1221 #endif 1222 1223 m_copydata(m, off, sizeof(ip6e), (caddr_t)&ip6e); 1224 if (nxt == IPPROTO_AH) 1225 elen = (ip6e.ip6e_len + 2) << 2; 1226 else 1227 elen = (ip6e.ip6e_len + 1) << 3; 1228 1229 MGET(n, M_DONTWAIT, MT_DATA); 1230 if (n && elen >= MLEN) { 1231 MCLGET(n, M_DONTWAIT); 1232 if ((n->m_flags & M_EXT) == 0) { 1233 m_free(n); 1234 n = NULL; 1235 } 1236 } 1237 if (!n) 1238 return NULL; 1239 1240 n->m_len = 0; 1241 if (elen >= M_TRAILINGSPACE(n)) { 1242 m_free(n); 1243 return NULL; 1244 } 1245 1246 m_copydata(m, off, elen, mtod(n, caddr_t)); 1247 n->m_len = elen; 1248 return n; 1249 } 1250 1251 /* 1252 * Get pointer to the previous header followed by the header 1253 * currently processed. 1254 * XXX: This function supposes that 1255 * M includes all headers, 1256 * the next header field and the header length field of each header 1257 * are valid, and 1258 * the sum of each header length equals to OFF. 1259 * Because of these assumptions, this function must be called very 1260 * carefully. Moreover, it will not be used in the near future when 1261 * we develop `neater' mechanism to process extension headers. 1262 */ 1263 u_int8_t * 1264 ip6_get_prevhdr(m, off) 1265 struct mbuf *m; 1266 int off; 1267 { 1268 struct ip6_hdr *ip6 = mtod(m, struct ip6_hdr *); 1269 1270 if (off == sizeof(struct ip6_hdr)) 1271 return (&ip6->ip6_nxt); 1272 else { 1273 int len, nxt; 1274 struct ip6_ext *ip6e = NULL; 1275 1276 nxt = ip6->ip6_nxt; 1277 len = sizeof(struct ip6_hdr); 1278 while (len < off) { 1279 ip6e = (struct ip6_ext *)(mtod(m, caddr_t) + len); 1280 1281 switch (nxt) { 1282 case IPPROTO_FRAGMENT: 1283 len += sizeof(struct ip6_frag); 1284 break; 1285 case IPPROTO_AH: 1286 len += (ip6e->ip6e_len + 2) << 2; 1287 break; 1288 default: 1289 len += (ip6e->ip6e_len + 1) << 3; 1290 break; 1291 } 1292 nxt = ip6e->ip6e_nxt; 1293 } 1294 if (ip6e) 1295 return (&ip6e->ip6e_nxt); 1296 else 1297 return NULL; 1298 } 1299 } 1300 1301 /* 1302 * get next header offset. m will be retained. 1303 */ 1304 int 1305 ip6_nexthdr(m, off, proto, nxtp) 1306 struct mbuf *m; 1307 int off; 1308 int proto; 1309 int *nxtp; 1310 { 1311 struct ip6_hdr ip6; 1312 struct ip6_ext ip6e; 1313 struct ip6_frag fh; 1314 1315 /* just in case */ 1316 if (m == NULL) 1317 panic("ip6_nexthdr: m == NULL"); 1318 if ((m->m_flags & M_PKTHDR) == 0 || m->m_pkthdr.len < off) 1319 return -1; 1320 1321 switch (proto) { 1322 case IPPROTO_IPV6: 1323 if (m->m_pkthdr.len < off + sizeof(ip6)) 1324 return -1; 1325 m_copydata(m, off, sizeof(ip6), (caddr_t)&ip6); 1326 if (nxtp) 1327 *nxtp = ip6.ip6_nxt; 1328 off += sizeof(ip6); 1329 return off; 1330 1331 case IPPROTO_FRAGMENT: 1332 /* 1333 * terminate parsing if it is not the first fragment, 1334 * it does not make sense to parse through it. 1335 */ 1336 if (m->m_pkthdr.len < off + sizeof(fh)) 1337 return -1; 1338 m_copydata(m, off, sizeof(fh), (caddr_t)&fh); 1339 if ((fh.ip6f_offlg & IP6F_OFF_MASK) != 0) 1340 return -1; 1341 if (nxtp) 1342 *nxtp = fh.ip6f_nxt; 1343 off += sizeof(struct ip6_frag); 1344 return off; 1345 1346 case IPPROTO_AH: 1347 if (m->m_pkthdr.len < off + sizeof(ip6e)) 1348 return -1; 1349 m_copydata(m, off, sizeof(ip6e), (caddr_t)&ip6e); 1350 if (nxtp) 1351 *nxtp = ip6e.ip6e_nxt; 1352 off += (ip6e.ip6e_len + 2) << 2; 1353 if (m->m_pkthdr.len < off) 1354 return -1; 1355 return off; 1356 1357 case IPPROTO_HOPOPTS: 1358 case IPPROTO_ROUTING: 1359 case IPPROTO_DSTOPTS: 1360 if (m->m_pkthdr.len < off + sizeof(ip6e)) 1361 return -1; 1362 m_copydata(m, off, sizeof(ip6e), (caddr_t)&ip6e); 1363 if (nxtp) 1364 *nxtp = ip6e.ip6e_nxt; 1365 off += (ip6e.ip6e_len + 1) << 3; 1366 if (m->m_pkthdr.len < off) 1367 return -1; 1368 return off; 1369 1370 case IPPROTO_NONE: 1371 case IPPROTO_ESP: 1372 case IPPROTO_IPCOMP: 1373 /* give up */ 1374 return -1; 1375 1376 default: 1377 return -1; 1378 } 1379 } 1380 1381 /* 1382 * get offset for the last header in the chain. m will be kept untainted. 1383 */ 1384 int 1385 ip6_lasthdr(m, off, proto, nxtp) 1386 struct mbuf *m; 1387 int off; 1388 int proto; 1389 int *nxtp; 1390 { 1391 int newoff; 1392 int nxt; 1393 1394 if (!nxtp) { 1395 nxt = -1; 1396 nxtp = &nxt; 1397 } 1398 while (1) { 1399 newoff = ip6_nexthdr(m, off, proto, nxtp); 1400 if (newoff < 0) 1401 return off; 1402 else if (newoff < off) 1403 return -1; /* invalid */ 1404 else if (newoff == off) 1405 return newoff; 1406 1407 off = newoff; 1408 proto = *nxtp; 1409 } 1410 } 1411 1412 /* 1413 * System control for IP6 1414 */ 1415 1416 u_char inet6ctlerrmap[PRC_NCMDS] = { 1417 0, 0, 0, 0, 1418 0, EMSGSIZE, EHOSTDOWN, EHOSTUNREACH, 1419 EHOSTUNREACH, EHOSTUNREACH, ECONNREFUSED, ECONNREFUSED, 1420 EMSGSIZE, EHOSTUNREACH, 0, 0, 1421 0, 0, 0, 0, 1422 ENOPROTOOPT 1423 }; 1424 1425 SYSCTL_SETUP(sysctl_net_inet6_ip6_setup, "sysctl net.inet6.ip6 subtree setup") 1426 { 1427 1428 sysctl_createv(clog, 0, NULL, NULL, 1429 CTLFLAG_PERMANENT, 1430 CTLTYPE_NODE, "net", NULL, 1431 NULL, 0, NULL, 0, 1432 CTL_NET, CTL_EOL); 1433 sysctl_createv(clog, 0, NULL, NULL, 1434 CTLFLAG_PERMANENT, 1435 CTLTYPE_NODE, "inet6", 1436 SYSCTL_DESCR("PF_INET6 related settings"), 1437 NULL, 0, NULL, 0, 1438 CTL_NET, PF_INET6, CTL_EOL); 1439 sysctl_createv(clog, 0, NULL, NULL, 1440 CTLFLAG_PERMANENT, 1441 CTLTYPE_NODE, "ip6", 1442 SYSCTL_DESCR("IPv6 related settings"), 1443 NULL, 0, NULL, 0, 1444 CTL_NET, PF_INET6, IPPROTO_IPV6, CTL_EOL); 1445 1446 sysctl_createv(clog, 0, NULL, NULL, 1447 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1448 CTLTYPE_INT, "forwarding", 1449 SYSCTL_DESCR("Enable forwarding of INET6 datagrams"), 1450 NULL, 0, &ip6_forwarding, 0, 1451 CTL_NET, PF_INET6, IPPROTO_IPV6, 1452 IPV6CTL_FORWARDING, CTL_EOL); 1453 sysctl_createv(clog, 0, NULL, NULL, 1454 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1455 CTLTYPE_INT, "redirect", 1456 SYSCTL_DESCR("Enable sending of ICMPv6 redirect messages"), 1457 NULL, 0, &ip6_sendredirects, 0, 1458 CTL_NET, PF_INET6, IPPROTO_IPV6, 1459 IPV6CTL_SENDREDIRECTS, CTL_EOL); 1460 sysctl_createv(clog, 0, NULL, NULL, 1461 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1462 CTLTYPE_INT, "hlim", 1463 SYSCTL_DESCR("Hop limit for an INET6 datagram"), 1464 NULL, 0, &ip6_defhlim, 0, 1465 CTL_NET, PF_INET6, IPPROTO_IPV6, 1466 IPV6CTL_DEFHLIM, CTL_EOL); 1467 #ifdef notyet 1468 sysctl_createv(clog, 0, NULL, NULL, 1469 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1470 CTLTYPE_INT, "mtu", NULL, 1471 NULL, 0, &, 0, 1472 CTL_NET, PF_INET6, IPPROTO_IPV6, 1473 IPV6CTL_DEFMTU, CTL_EOL); 1474 #endif 1475 #ifdef __no_idea__ 1476 sysctl_createv(clog, 0, NULL, NULL, 1477 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1478 CTLTYPE_INT, "forwsrcrt", NULL, 1479 NULL, 0, &?, 0, 1480 CTL_NET, PF_INET6, IPPROTO_IPV6, 1481 IPV6CTL_FORWSRCRT, CTL_EOL); 1482 sysctl_createv(clog, 0, NULL, NULL, 1483 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1484 CTLTYPE_STRUCT, "stats", NULL, 1485 NULL, 0, &?, sizeof(?), 1486 CTL_NET, PF_INET6, IPPROTO_IPV6, 1487 IPV6CTL_STATS, CTL_EOL); 1488 sysctl_createv(clog, 0, NULL, NULL, 1489 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1490 CTLTYPE_STRUCT, "mrtstats", NULL, 1491 NULL, 0, &?, sizeof(?), 1492 CTL_NET, PF_INET6, IPPROTO_IPV6, 1493 IPV6CTL_MRTSTATS, CTL_EOL); 1494 sysctl_createv(clog, 0, NULL, NULL, 1495 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1496 CTLTYPE_?, "mrtproto", NULL, 1497 NULL, 0, &?, sizeof(?), 1498 CTL_NET, PF_INET6, IPPROTO_IPV6, 1499 IPV6CTL_MRTPROTO, CTL_EOL); 1500 #endif 1501 sysctl_createv(clog, 0, NULL, NULL, 1502 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1503 CTLTYPE_INT, "maxfragpackets", 1504 SYSCTL_DESCR("Maximum number of fragments to buffer " 1505 "for reassembly"), 1506 NULL, 0, &ip6_maxfragpackets, 0, 1507 CTL_NET, PF_INET6, IPPROTO_IPV6, 1508 IPV6CTL_MAXFRAGPACKETS, CTL_EOL); 1509 #ifdef __no_idea__ 1510 sysctl_createv(clog, 0, NULL, NULL, 1511 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1512 CTLTYPE_INT, "sourcecheck", NULL, 1513 NULL, 0, &?, 0, 1514 CTL_NET, PF_INET6, IPPROTO_IPV6, 1515 IPV6CTL_SOURCECHECK, CTL_EOL); 1516 sysctl_createv(clog, 0, NULL, NULL, 1517 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1518 CTLTYPE_INT, "sourcecheck_logint", NULL, 1519 NULL, 0, &?, 0, 1520 CTL_NET, PF_INET6, IPPROTO_IPV6, 1521 IPV6CTL_SOURCECHECK_LOGINT, CTL_EOL); 1522 #endif 1523 sysctl_createv(clog, 0, NULL, NULL, 1524 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1525 CTLTYPE_INT, "accept_rtadv", 1526 SYSCTL_DESCR("Accept router advertisements"), 1527 NULL, 0, &ip6_accept_rtadv, 0, 1528 CTL_NET, PF_INET6, IPPROTO_IPV6, 1529 IPV6CTL_ACCEPT_RTADV, CTL_EOL); 1530 sysctl_createv(clog, 0, NULL, NULL, 1531 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1532 CTLTYPE_INT, "keepfaith", 1533 SYSCTL_DESCR("Activate faith interface"), 1534 NULL, 0, &ip6_keepfaith, 0, 1535 CTL_NET, PF_INET6, IPPROTO_IPV6, 1536 IPV6CTL_KEEPFAITH, CTL_EOL); 1537 sysctl_createv(clog, 0, NULL, NULL, 1538 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1539 CTLTYPE_INT, "log_interval", 1540 SYSCTL_DESCR("Minumum interval between logging " 1541 "unroutable packets"), 1542 NULL, 0, &ip6_log_interval, 0, 1543 CTL_NET, PF_INET6, IPPROTO_IPV6, 1544 IPV6CTL_LOG_INTERVAL, CTL_EOL); 1545 sysctl_createv(clog, 0, NULL, NULL, 1546 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1547 CTLTYPE_INT, "hdrnestlimit", 1548 SYSCTL_DESCR("Maximum number of nested IPv6 headers"), 1549 NULL, 0, &ip6_hdrnestlimit, 0, 1550 CTL_NET, PF_INET6, IPPROTO_IPV6, 1551 IPV6CTL_HDRNESTLIMIT, CTL_EOL); 1552 sysctl_createv(clog, 0, NULL, NULL, 1553 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1554 CTLTYPE_INT, "dad_count", 1555 SYSCTL_DESCR("Number of Duplicate Address Detection " 1556 "probes to send"), 1557 NULL, 0, &ip6_dad_count, 0, 1558 CTL_NET, PF_INET6, IPPROTO_IPV6, 1559 IPV6CTL_DAD_COUNT, CTL_EOL); 1560 sysctl_createv(clog, 0, NULL, NULL, 1561 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1562 CTLTYPE_INT, "auto_flowlabel", 1563 SYSCTL_DESCR("Assign random IPv6 flow labels"), 1564 NULL, 0, &ip6_auto_flowlabel, 0, 1565 CTL_NET, PF_INET6, IPPROTO_IPV6, 1566 IPV6CTL_AUTO_FLOWLABEL, CTL_EOL); 1567 sysctl_createv(clog, 0, NULL, NULL, 1568 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1569 CTLTYPE_INT, "defmcasthlim", 1570 SYSCTL_DESCR("Default multicast hop limit"), 1571 NULL, 0, &ip6_defmcasthlim, 0, 1572 CTL_NET, PF_INET6, IPPROTO_IPV6, 1573 IPV6CTL_DEFMCASTHLIM, CTL_EOL); 1574 #if NGIF > 0 1575 sysctl_createv(clog, 0, NULL, NULL, 1576 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1577 CTLTYPE_INT, "gifhlim", 1578 SYSCTL_DESCR("Default hop limit for a gif tunnel datagram"), 1579 NULL, 0, &ip6_gif_hlim, 0, 1580 CTL_NET, PF_INET6, IPPROTO_IPV6, 1581 IPV6CTL_GIF_HLIM, CTL_EOL); 1582 #endif /* NGIF */ 1583 sysctl_createv(clog, 0, NULL, NULL, 1584 CTLFLAG_PERMANENT, 1585 CTLTYPE_STRING, "kame_version", 1586 SYSCTL_DESCR("KAME Version"), 1587 NULL, 0, __KAME_VERSION, 0, 1588 CTL_NET, PF_INET6, IPPROTO_IPV6, 1589 IPV6CTL_KAME_VERSION, CTL_EOL); 1590 sysctl_createv(clog, 0, NULL, NULL, 1591 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1592 CTLTYPE_INT, "use_deprecated", 1593 SYSCTL_DESCR("Allow use of deprecated addresses as " 1594 "source addresses"), 1595 NULL, 0, &ip6_use_deprecated, 0, 1596 CTL_NET, PF_INET6, IPPROTO_IPV6, 1597 IPV6CTL_USE_DEPRECATED, CTL_EOL); 1598 sysctl_createv(clog, 0, NULL, NULL, 1599 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1600 CTLTYPE_INT, "rr_prune", NULL, 1601 NULL, 0, &ip6_rr_prune, 0, 1602 CTL_NET, PF_INET6, IPPROTO_IPV6, 1603 IPV6CTL_RR_PRUNE, CTL_EOL); 1604 sysctl_createv(clog, 0, NULL, NULL, 1605 CTLFLAG_PERMANENT 1606 #ifndef INET6_BINDV6ONLY 1607 |CTLFLAG_READWRITE, 1608 #endif 1609 CTLTYPE_INT, "v6only", 1610 SYSCTL_DESCR("Disallow PF_INET6 sockets from connecting " 1611 "to PF_INET sockets"), 1612 NULL, 0, &ip6_v6only, 0, 1613 CTL_NET, PF_INET6, IPPROTO_IPV6, 1614 IPV6CTL_V6ONLY, CTL_EOL); 1615 sysctl_createv(clog, 0, NULL, NULL, 1616 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1617 CTLTYPE_INT, "anonportmin", 1618 SYSCTL_DESCR("Lowest ephemeral port number to assign"), 1619 sysctl_net_inet_ip_ports, 0, &ip6_anonportmin, 0, 1620 CTL_NET, PF_INET6, IPPROTO_IPV6, 1621 IPV6CTL_ANONPORTMIN, CTL_EOL); 1622 sysctl_createv(clog, 0, NULL, NULL, 1623 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1624 CTLTYPE_INT, "anonportmax", 1625 SYSCTL_DESCR("Highest ephemeral port number to assign"), 1626 sysctl_net_inet_ip_ports, 0, &ip6_anonportmax, 0, 1627 CTL_NET, PF_INET6, IPPROTO_IPV6, 1628 IPV6CTL_ANONPORTMAX, CTL_EOL); 1629 #ifndef IPNOPRIVPORTS 1630 sysctl_createv(clog, 0, NULL, NULL, 1631 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1632 CTLTYPE_INT, "lowportmin", 1633 SYSCTL_DESCR("Lowest privileged ephemeral port number " 1634 "to assign"), 1635 sysctl_net_inet_ip_ports, 0, &ip6_lowportmin, 0, 1636 CTL_NET, PF_INET6, IPPROTO_IPV6, 1637 IPV6CTL_LOWPORTMIN, CTL_EOL); 1638 sysctl_createv(clog, 0, NULL, NULL, 1639 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1640 CTLTYPE_INT, "lowportmax", 1641 SYSCTL_DESCR("Highest privileged ephemeral port number " 1642 "to assign"), 1643 sysctl_net_inet_ip_ports, 0, &ip6_lowportmax, 0, 1644 CTL_NET, PF_INET6, IPPROTO_IPV6, 1645 IPV6CTL_LOWPORTMAX, CTL_EOL); 1646 #endif /* IPNOPRIVPORTS */ 1647 sysctl_createv(clog, 0, NULL, NULL, 1648 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1649 CTLTYPE_INT, "maxfrags", 1650 SYSCTL_DESCR("Maximum fragments in reassembly queue"), 1651 NULL, 0, &ip6_maxfrags, 0, 1652 CTL_NET, PF_INET6, IPPROTO_IPV6, 1653 IPV6CTL_MAXFRAGS, CTL_EOL); 1654 } 1655