ip6_input.c revision 126444
1/* $FreeBSD: head/sys/netinet6/ip6_input.c 126444 2004-03-01 15:34:29Z ume $ */ 2/* $KAME: ip6_input.c,v 1.259 2002/01/21 04:58:09 jinmei Exp $ */ 3 4/* 5 * Copyright (C) 1995, 1996, 1997, and 1998 WIDE Project. 6 * All rights reserved. 7 * 8 * Redistribution and use in source and binary forms, with or without 9 * modification, are permitted provided that the following conditions 10 * are met: 11 * 1. Redistributions of source code must retain the above copyright 12 * notice, this list of conditions and the following disclaimer. 13 * 2. Redistributions in binary form must reproduce the above copyright 14 * notice, this list of conditions and the following disclaimer in the 15 * documentation and/or other materials provided with the distribution. 16 * 3. Neither the name of the project nor the names of its contributors 17 * may be used to endorse or promote products derived from this software 18 * without specific prior written permission. 19 * 20 * THIS SOFTWARE IS PROVIDED BY THE PROJECT AND CONTRIBUTORS ``AS IS'' AND 21 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE 22 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE 23 * ARE DISCLAIMED. IN NO EVENT SHALL THE PROJECT OR CONTRIBUTORS BE LIABLE 24 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL 25 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS 26 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) 27 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT 28 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY 29 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF 30 * SUCH DAMAGE. 31 */ 32 33/* 34 * Copyright (c) 1982, 1986, 1988, 1993 35 * The Regents of the University of California. All rights reserved. 36 * 37 * Redistribution and use in source and binary forms, with or without 38 * modification, are permitted provided that the following conditions 39 * are met: 40 * 1. Redistributions of source code must retain the above copyright 41 * notice, this list of conditions and the following disclaimer. 42 * 2. Redistributions in binary form must reproduce the above copyright 43 * notice, this list of conditions and the following disclaimer in the 44 * documentation and/or other materials provided with the distribution. 45 * 3. All advertising materials mentioning features or use of this software 46 * must display the following acknowledgement: 47 * This product includes software developed by the University of 48 * California, Berkeley and its contributors. 49 * 4. Neither the name of the University nor the names of its contributors 50 * may be used to endorse or promote products derived from this software 51 * without specific prior written permission. 52 * 53 * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND 54 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE 55 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE 56 * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE 57 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL 58 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS 59 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) 60 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT 61 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY 62 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF 63 * SUCH DAMAGE. 64 * 65 * @(#)ip_input.c 8.2 (Berkeley) 1/4/94 66 */ 67 68#include "opt_ip6fw.h" 69#include "opt_inet.h" 70#include "opt_inet6.h" 71#include "opt_ipsec.h" 72#include "opt_pfil_hooks.h" 73#include "opt_random_ip_id.h" 74 75#include <sys/param.h> 76#include <sys/systm.h> 77#include <sys/malloc.h> 78#include <sys/mbuf.h> 79#include <sys/proc.h> 80#include <sys/domain.h> 81#include <sys/protosw.h> 82#include <sys/socket.h> 83#include <sys/socketvar.h> 84#include <sys/errno.h> 85#include <sys/time.h> 86#include <sys/kernel.h> 87#include <sys/syslog.h> 88 89#include <net/if.h> 90#include <net/if_types.h> 91#include <net/if_dl.h> 92#include <net/route.h> 93#include <net/netisr.h> 94#ifdef PFIL_HOOKS 95#include <net/pfil.h> 96#endif 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/scope6_var.h> 110#include <netinet6/in6_ifattach.h> 111#include <netinet6/nd6.h> 112#include <netinet6/in6_prefix.h> 113 114#ifdef IPSEC 115#include <netinet6/ipsec.h> 116#ifdef INET6 117#include <netinet6/ipsec6.h> 118#endif 119#endif 120 121#ifdef FAST_IPSEC 122#include <netipsec/ipsec.h> 123#include <netipsec/ipsec6.h> 124#define IPSEC 125#endif /* FAST_IPSEC */ 126 127#include <netinet6/ip6_fw.h> 128 129#include <netinet6/ip6protosw.h> 130 131#include <net/net_osdep.h> 132 133extern struct domain inet6domain; 134 135u_char ip6_protox[IPPROTO_MAX]; 136static struct ifqueue ip6intrq; 137static int ip6qmaxlen = IFQ_MAXLEN; 138struct in6_ifaddr *in6_ifaddr; 139 140extern struct callout in6_tmpaddrtimer_ch; 141 142int ip6_forward_srcrt; /* XXX */ 143int ip6_sourcecheck; /* XXX */ 144int ip6_sourcecheck_interval; /* XXX */ 145 146int ip6_ours_check_algorithm; 147 148#ifdef PFIL_HOOKS 149struct pfil_head inet6_pfil_hook; 150#endif 151 152/* firewall hooks */ 153ip6_fw_chk_t *ip6_fw_chk_ptr; 154ip6_fw_ctl_t *ip6_fw_ctl_ptr; 155int ip6_fw_enable = 1; 156 157struct ip6stat ip6stat; 158 159static void ip6_init2 __P((void *)); 160static struct ip6aux *ip6_setdstifaddr __P((struct mbuf *, struct in6_ifaddr *)); 161static int ip6_hopopts_input __P((u_int32_t *, u_int32_t *, struct mbuf **, int *)); 162#ifdef PULLDOWN_TEST 163static struct mbuf *ip6_pullexthdr __P((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 */ 170void 171ip6_init() 172{ 173 struct ip6protosw *pr; 174 int i; 175 176#ifdef DIAGNOSTIC 177 if (sizeof(struct protosw) != sizeof(struct ip6protosw)) 178 panic("sizeof(protosw) != sizeof(ip6protosw)"); 179#endif 180 pr = (struct ip6protosw *)pffindproto(PF_INET6, IPPROTO_RAW, SOCK_RAW); 181 if (pr == 0) 182 panic("ip6_init"); 183 for (i = 0; i < IPPROTO_MAX; i++) 184 ip6_protox[i] = pr - inet6sw; 185 for (pr = (struct ip6protosw *)inet6domain.dom_protosw; 186 pr < (struct ip6protosw *)inet6domain.dom_protoswNPROTOSW; pr++) 187 if (pr->pr_domain->dom_family == PF_INET6 && 188 pr->pr_protocol && pr->pr_protocol != IPPROTO_RAW) 189 ip6_protox[pr->pr_protocol] = pr - inet6sw; 190#ifdef PFIL_HOOKS 191 inet6_pfil_hook.ph_type = PFIL_TYPE_AF; 192 inet6_pfil_hook.ph_af = AF_INET6; 193 if ((i = pfil_head_register(&inet6_pfil_hook)) != 0) 194 printf("%s: WARNING: unable to register pfil hook, " 195 "error %d\n", __func__, i); 196#endif /* PFIL_HOOKS */ 197 ip6intrq.ifq_maxlen = ip6qmaxlen; 198 mtx_init(&ip6intrq.ifq_mtx, "ip6_inq", NULL, MTX_DEF); 199 netisr_register(NETISR_IPV6, ip6_input, &ip6intrq, 0); 200 scope6_init(); 201 addrsel_policy_init(); 202 nd6_init(); 203 frag6_init(); 204#ifndef RANDOM_IP_ID 205 ip6_flow_seq = arc4random(); 206#endif 207 ip6_desync_factor = arc4random() % MAX_TEMP_DESYNC_FACTOR; 208} 209 210static void 211ip6_init2(dummy) 212 void *dummy; 213{ 214 215 /* nd6_timer_init */ 216 callout_init(&nd6_timer_ch, 0); 217 callout_reset(&nd6_timer_ch, hz, nd6_timer, NULL); 218 219 /* router renumbering prefix list maintenance */ 220 callout_init(&in6_rr_timer_ch, 0); 221 callout_reset(&in6_rr_timer_ch, hz, in6_rr_timer, NULL); 222 223 /* timer for regeneranation of temporary addresses randomize ID */ 224 callout_init(&in6_tmpaddrtimer_ch, 0); 225 callout_reset(&in6_tmpaddrtimer_ch, 226 (ip6_temp_preferred_lifetime - ip6_desync_factor - 227 ip6_temp_regen_advance) * hz, 228 in6_tmpaddrtimer, NULL); 229} 230 231/* cheat */ 232/* This must be after route_init(), which is now SI_ORDER_THIRD */ 233SYSINIT(netinet6init2, SI_SUB_PROTO_DOMAIN, SI_ORDER_MIDDLE, ip6_init2, NULL); 234 235extern struct route_in6 ip6_forward_rt; 236 237void 238ip6_input(m) 239 struct mbuf *m; 240{ 241 struct ip6_hdr *ip6; 242 int off = sizeof(struct ip6_hdr), nest; 243 u_int32_t plen; 244 u_int32_t rtalert = ~0; 245 int nxt, ours = 0; 246 struct ifnet *deliverifp = NULL; 247 struct sockaddr_in6 sa6; 248 u_int32_t srczone, dstzone; 249#ifdef PFIL_HOOKS 250 struct in6_addr odst; 251#endif 252 int srcrt = 0; 253 254 GIANT_REQUIRED; /* XXX for now */ 255#ifdef IPSEC 256 /* 257 * should the inner packet be considered authentic? 258 * see comment in ah4_input(). 259 */ 260 if (m) { 261 m->m_flags &= ~M_AUTHIPHDR; 262 m->m_flags &= ~M_AUTHIPDGM; 263 } 264#endif 265 266 /* 267 * make sure we don't have onion peering information into m_tag. 268 */ 269 ip6_delaux(m); 270 271 /* 272 * mbuf statistics 273 */ 274 if (m->m_flags & M_EXT) { 275 if (m->m_next) 276 ip6stat.ip6s_mext2m++; 277 else 278 ip6stat.ip6s_mext1++; 279 } else { 280#define M2MMAX (sizeof(ip6stat.ip6s_m2m)/sizeof(ip6stat.ip6s_m2m[0])) 281 if (m->m_next) { 282 if (m->m_flags & M_LOOP) { 283 ip6stat.ip6s_m2m[loif[0].if_index]++; /* XXX */ 284 } else if (m->m_pkthdr.rcvif->if_index < M2MMAX) 285 ip6stat.ip6s_m2m[m->m_pkthdr.rcvif->if_index]++; 286 else 287 ip6stat.ip6s_m2m[0]++; 288 } else 289 ip6stat.ip6s_m1++; 290#undef M2MMAX 291 } 292 293 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_receive); 294 ip6stat.ip6s_total++; 295 296#ifndef PULLDOWN_TEST 297 /* 298 * L2 bridge code and some other code can return mbuf chain 299 * that does not conform to KAME requirement. too bad. 300 * XXX: fails to join if interface MTU > MCLBYTES. jumbogram? 301 */ 302 if (m && m->m_next != NULL && m->m_pkthdr.len < MCLBYTES) { 303 struct mbuf *n; 304 305 MGETHDR(n, M_DONTWAIT, MT_HEADER); 306 if (n) 307 M_MOVE_PKTHDR(n, m); 308 if (n && n->m_pkthdr.len > MHLEN) { 309 MCLGET(n, M_DONTWAIT); 310 if ((n->m_flags & M_EXT) == 0) { 311 m_freem(n); 312 n = NULL; 313 } 314 } 315 if (n == NULL) { 316 m_freem(m); 317 return; /* ENOBUFS */ 318 } 319 320 m_copydata(m, 0, n->m_pkthdr.len, mtod(n, caddr_t)); 321 n->m_len = n->m_pkthdr.len; 322 m_freem(m); 323 m = n; 324 } 325 IP6_EXTHDR_CHECK(m, 0, sizeof(struct ip6_hdr), /* nothing */); 326#endif 327 328 if (m->m_len < sizeof(struct ip6_hdr)) { 329 struct ifnet *inifp; 330 inifp = m->m_pkthdr.rcvif; 331 if ((m = m_pullup(m, sizeof(struct ip6_hdr))) == NULL) { 332 ip6stat.ip6s_toosmall++; 333 in6_ifstat_inc(inifp, ifs6_in_hdrerr); 334 return; 335 } 336 } 337 338 ip6 = mtod(m, struct ip6_hdr *); 339 340 if ((ip6->ip6_vfc & IPV6_VERSION_MASK) != IPV6_VERSION) { 341 ip6stat.ip6s_badvers++; 342 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_hdrerr); 343 goto bad; 344 } 345 346#ifdef PFIL_HOOKS 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 odst = ip6->ip6_dst; 355 if (pfil_run_hooks(&inet6_pfil_hook, &m, m->m_pkthdr.rcvif, PFIL_IN)) 356 return; 357 if (m == NULL) /* consumed by filter */ 358 return; 359 ip6 = mtod(m, struct ip6_hdr *); 360 srcrt = !IN6_ARE_ADDR_EQUAL(&odst, &ip6->ip6_dst); 361#endif /* PFIL_HOOKS */ 362 363 ip6stat.ip6s_nxthist[ip6->ip6_nxt]++; 364 365 /* 366 * Check with the firewall... 367 */ 368 if (ip6_fw_enable && ip6_fw_chk_ptr) { 369 u_short port = 0; 370 /* If ipfw says divert, we have to just drop packet */ 371 /* use port as a dummy argument */ 372 if ((*ip6_fw_chk_ptr)(&ip6, NULL, &port, &m)) { 373 m_freem(m); 374 m = NULL; 375 } 376 if (!m) 377 return; 378 } 379 380 /* 381 * Check against address spoofing/corruption. 382 */ 383 if (IN6_IS_ADDR_MULTICAST(&ip6->ip6_src) || 384 IN6_IS_ADDR_UNSPECIFIED(&ip6->ip6_dst)) { 385 /* 386 * XXX: "badscope" is not very suitable for a multicast source. 387 */ 388 ip6stat.ip6s_badscope++; 389 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_addrerr); 390 goto bad; 391 } 392 if (IN6_IS_ADDR_MC_INTFACELOCAL(&ip6->ip6_dst) && 393 !(m->m_flags & M_LOOP)) { 394 /* 395 * In this case, the packet should come from the loopback 396 * interface. However, we cannot just check the if_flags, 397 * because ip6_mloopback() passes the "actual" interface 398 * as the outgoing/incoming interface. 399 */ 400 ip6stat.ip6s_badscope++; 401 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_addrerr); 402 goto bad; 403 } 404 405 /* 406 * The following check is not documented in specs. A malicious 407 * party may be able to use IPv4 mapped addr to confuse tcp/udp stack 408 * and bypass security checks (act as if it was from 127.0.0.1 by using 409 * IPv6 src ::ffff:127.0.0.1). Be cautious. 410 * 411 * This check chokes if we are in an SIIT cloud. As none of BSDs 412 * support IPv4-less kernel compilation, we cannot support SIIT 413 * environment at all. So, it makes more sense for us to reject any 414 * malicious packets for non-SIIT environment, than try to do a 415 * partial support for SIIT environment. 416 */ 417 if (IN6_IS_ADDR_V4MAPPED(&ip6->ip6_src) || 418 IN6_IS_ADDR_V4MAPPED(&ip6->ip6_dst)) { 419 ip6stat.ip6s_badscope++; 420 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_addrerr); 421 goto bad; 422 } 423#if 0 424 /* 425 * Reject packets with IPv4 compatible addresses (auto tunnel). 426 * 427 * The code forbids auto tunnel relay case in RFC1933 (the check is 428 * stronger than RFC1933). We may want to re-enable it if mech-xx 429 * is revised to forbid relaying case. 430 */ 431 if (IN6_IS_ADDR_V4COMPAT(&ip6->ip6_src) || 432 IN6_IS_ADDR_V4COMPAT(&ip6->ip6_dst)) { 433 ip6stat.ip6s_badscope++; 434 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_addrerr); 435 goto bad; 436 } 437#endif 438 439 /* 440 * Drop packets if the link ID portion is already filled. 441 * XXX: this is technically not a good behavior. But, we internally 442 * use the field to disambiguate link-local addresses, so we cannot 443 * be generous against those a bit strange addresses. 444 */ 445 if ((m->m_pkthdr.rcvif->if_flags & IFF_LOOPBACK) == 0) { 446 if (IN6_IS_SCOPE_LINKLOCAL(&ip6->ip6_src) && 447 ip6->ip6_src.s6_addr16[1]) { 448 ip6stat.ip6s_badscope++; 449 goto bad; 450 } 451 if ((IN6_IS_ADDR_MC_INTFACELOCAL(&ip6->ip6_dst) || 452 IN6_IS_SCOPE_LINKLOCAL(&ip6->ip6_dst)) && 453 ip6->ip6_dst.s6_addr16[1]) { 454 ip6stat.ip6s_badscope++; 455 goto bad; 456 } 457 } 458 459 /* 460 * construct source and destination address structures with 461 * disambiguating their scope zones (if there is ambiguity). 462 * XXX: sin6_family and sin6_len will NOT be referred to, but we fill 463 * in these fields just in case. 464 */ 465 if (in6_addr2zoneid(m->m_pkthdr.rcvif, &ip6->ip6_src, &srczone) || 466 in6_addr2zoneid(m->m_pkthdr.rcvif, &ip6->ip6_dst, &dstzone)) { 467 /* 468 * Note that these generic checks cover cases that src or 469 * dst are the loopback address and the receiving interface 470 * is not loopback. 471 */ 472 ip6stat.ip6s_badscope++; 473 goto bad; 474 } 475 476 bzero(&sa6, sizeof(sa6)); 477 sa6.sin6_family = AF_INET6; 478 sa6.sin6_len = sizeof(struct sockaddr_in6); 479 480 sa6.sin6_addr = ip6->ip6_src; 481 sa6.sin6_scope_id = srczone; 482 if (in6_embedscope(&ip6->ip6_src, &sa6, NULL, NULL)) { 483 /* XXX: should not happen */ 484 ip6stat.ip6s_badscope++; 485 goto bad; 486 } 487 488 sa6.sin6_addr = ip6->ip6_dst; 489 sa6.sin6_scope_id = dstzone; 490 if (in6_embedscope(&ip6->ip6_dst, &sa6, NULL, NULL)) { 491 /* XXX: should not happen */ 492 ip6stat.ip6s_badscope++; 493 goto bad; 494 } 495 496 /* 497 * Multicast check 498 */ 499 if (IN6_IS_ADDR_MULTICAST(&ip6->ip6_dst)) { 500 struct in6_multi *in6m = 0; 501 502 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_mcast); 503 /* 504 * See if we belong to the destination multicast group on the 505 * arrival interface. 506 */ 507 IN6_LOOKUP_MULTI(ip6->ip6_dst, m->m_pkthdr.rcvif, in6m); 508 if (in6m) 509 ours = 1; 510 else if (!ip6_mrouter) { 511 ip6stat.ip6s_notmember++; 512 ip6stat.ip6s_cantforward++; 513 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_discard); 514 goto bad; 515 } 516 deliverifp = m->m_pkthdr.rcvif; 517 goto hbhcheck; 518 } 519 520 /* 521 * Unicast check 522 */ 523 if (ip6_forward_rt.ro_rt != NULL && 524 (ip6_forward_rt.ro_rt->rt_flags & RTF_UP) != 0 && 525 IN6_ARE_ADDR_EQUAL(&ip6->ip6_dst, 526 &((struct sockaddr_in6 *)(&ip6_forward_rt.ro_dst))->sin6_addr)) 527 ip6stat.ip6s_forward_cachehit++; 528 else { 529 struct sockaddr_in6 *dst6; 530 531 if (ip6_forward_rt.ro_rt) { 532 /* route is down or destination is different */ 533 ip6stat.ip6s_forward_cachemiss++; 534 RTFREE(ip6_forward_rt.ro_rt); 535 ip6_forward_rt.ro_rt = 0; 536 } 537 538 bzero(&ip6_forward_rt.ro_dst, sizeof(struct sockaddr_in6)); 539 dst6 = (struct sockaddr_in6 *)&ip6_forward_rt.ro_dst; 540 dst6->sin6_len = sizeof(struct sockaddr_in6); 541 dst6->sin6_family = AF_INET6; 542 dst6->sin6_addr = ip6->ip6_dst; 543 544 rtalloc((struct route *)&ip6_forward_rt); 545 } 546 547#define rt6_key(r) ((struct sockaddr_in6 *)((r)->rt_nodes->rn_key)) 548 549 /* 550 * Accept the packet if the forwarding interface to the destination 551 * according to the routing table is the loopback interface, 552 * unless the associated route has a gateway. 553 * Note that this approach causes to accept a packet if there is a 554 * route to the loopback interface for the destination of the packet. 555 * But we think it's even useful in some situations, e.g. when using 556 * a special daemon which wants to intercept the packet. 557 * 558 * XXX: some OSes automatically make a cloned route for the destination 559 * of an outgoing packet. If the outgoing interface of the packet 560 * is a loopback one, the kernel would consider the packet to be 561 * accepted, even if we have no such address assinged on the interface. 562 * We check the cloned flag of the route entry to reject such cases, 563 * assuming that route entries for our own addresses are not made by 564 * cloning (it should be true because in6_addloop explicitly installs 565 * the host route). However, we might have to do an explicit check 566 * while it would be less efficient. Or, should we rather install a 567 * reject route for such a case? 568 */ 569 if (ip6_forward_rt.ro_rt && 570 (ip6_forward_rt.ro_rt->rt_flags & 571 (RTF_HOST|RTF_GATEWAY)) == RTF_HOST && 572#ifdef RTF_WASCLONED 573 !(ip6_forward_rt.ro_rt->rt_flags & RTF_WASCLONED) && 574#endif 575#ifdef RTF_CLONED 576 !(ip6_forward_rt.ro_rt->rt_flags & RTF_CLONED) && 577#endif 578#if 0 579 /* 580 * The check below is redundant since the comparison of 581 * the destination and the key of the rtentry has 582 * already done through looking up the routing table. 583 */ 584 IN6_ARE_ADDR_EQUAL(&ip6->ip6_dst, 585 &rt6_key(ip6_forward_rt.ro_rt)->sin6_addr) 586#endif 587 ip6_forward_rt.ro_rt->rt_ifp->if_type == IFT_LOOP) { 588 struct in6_ifaddr *ia6 = 589 (struct in6_ifaddr *)ip6_forward_rt.ro_rt->rt_ifa; 590 591 /* 592 * record address information into m_tag. 593 */ 594 (void)ip6_setdstifaddr(m, ia6); 595 596 /* 597 * packets to a tentative, duplicated, or somehow invalid 598 * address must not be accepted. 599 */ 600 if (!(ia6->ia6_flags & IN6_IFF_NOTREADY)) { 601 /* this address is ready */ 602 ours = 1; 603 deliverifp = ia6->ia_ifp; /* correct? */ 604 /* Count the packet in the ip address stats */ 605 ia6->ia_ifa.if_ipackets++; 606 ia6->ia_ifa.if_ibytes += m->m_pkthdr.len; 607 goto hbhcheck; 608 } else { 609 /* address is not ready, so discard the packet. */ 610 nd6log((LOG_INFO, 611 "ip6_input: packet to an unready address %s->%s\n", 612 ip6_sprintf(&ip6->ip6_src), 613 ip6_sprintf(&ip6->ip6_dst))); 614 615 goto bad; 616 } 617 } 618 619 /* 620 * FAITH (Firewall Aided Internet Translator) 621 */ 622 if (ip6_keepfaith) { 623 if (ip6_forward_rt.ro_rt && ip6_forward_rt.ro_rt->rt_ifp 624 && ip6_forward_rt.ro_rt->rt_ifp->if_type == IFT_FAITH) { 625 /* XXX do we need more sanity checks? */ 626 ours = 1; 627 deliverifp = ip6_forward_rt.ro_rt->rt_ifp; /* faith */ 628 goto hbhcheck; 629 } 630 } 631 632 /* 633 * Now there is no reason to process the packet if it's not our own 634 * and we're not a router. 635 */ 636 if (!ip6_forwarding) { 637 ip6stat.ip6s_cantforward++; 638 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_discard); 639 goto bad; 640 } 641 642 hbhcheck: 643 /* 644 * record address information into m_tag, if we don't have one yet. 645 * note that we are unable to record it, if the address is not listed 646 * as our interface address (e.g. multicast addresses, addresses 647 * within FAITH prefixes and such). 648 */ 649 if (deliverifp && !ip6_getdstifaddr(m)) { 650 struct in6_ifaddr *ia6; 651 652 ia6 = in6_ifawithifp(deliverifp, &ip6->ip6_dst); 653 if (ia6) { 654 if (!ip6_setdstifaddr(m, ia6)) { 655 /* 656 * XXX maybe we should drop the packet here, 657 * as we could not provide enough information 658 * to the upper layers. 659 */ 660 } 661 } 662 } 663 664 /* 665 * Process Hop-by-Hop options header if it's contained. 666 * m may be modified in ip6_hopopts_input(). 667 * If a JumboPayload option is included, plen will also be modified. 668 */ 669 plen = (u_int32_t)ntohs(ip6->ip6_plen); 670 if (ip6->ip6_nxt == IPPROTO_HOPOPTS) { 671 struct ip6_hbh *hbh; 672 673 if (ip6_hopopts_input(&plen, &rtalert, &m, &off)) { 674#if 0 /*touches NULL pointer*/ 675 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_discard); 676#endif 677 return; /* m have already been freed */ 678 } 679 680 /* adjust pointer */ 681 ip6 = mtod(m, struct ip6_hdr *); 682 683 /* 684 * if the payload length field is 0 and the next header field 685 * indicates Hop-by-Hop Options header, then a Jumbo Payload 686 * option MUST be included. 687 */ 688 if (ip6->ip6_plen == 0 && plen == 0) { 689 /* 690 * Note that if a valid jumbo payload option is 691 * contained, ip6_hopopts_input() must set a valid 692 * (non-zero) payload length to the variable plen. 693 */ 694 ip6stat.ip6s_badoptions++; 695 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_discard); 696 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_hdrerr); 697 icmp6_error(m, ICMP6_PARAM_PROB, 698 ICMP6_PARAMPROB_HEADER, 699 (caddr_t)&ip6->ip6_plen - (caddr_t)ip6); 700 return; 701 } 702#ifndef PULLDOWN_TEST 703 /* ip6_hopopts_input() ensures that mbuf is contiguous */ 704 hbh = (struct ip6_hbh *)(ip6 + 1); 705#else 706 IP6_EXTHDR_GET(hbh, struct ip6_hbh *, m, sizeof(struct ip6_hdr), 707 sizeof(struct ip6_hbh)); 708 if (hbh == NULL) { 709 ip6stat.ip6s_tooshort++; 710 return; 711 } 712#endif 713 nxt = hbh->ip6h_nxt; 714 715 /* 716 * accept the packet if a router alert option is included 717 * and we act as an IPv6 router. 718 */ 719 if (rtalert != ~0 && ip6_forwarding) 720 ours = 1; 721 } else 722 nxt = ip6->ip6_nxt; 723 724 /* 725 * Check that the amount of data in the buffers 726 * is as at least much as the IPv6 header would have us expect. 727 * Trim mbufs if longer than we expect. 728 * Drop packet if shorter than we expect. 729 */ 730 if (m->m_pkthdr.len - sizeof(struct ip6_hdr) < plen) { 731 ip6stat.ip6s_tooshort++; 732 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_truncated); 733 goto bad; 734 } 735 if (m->m_pkthdr.len > sizeof(struct ip6_hdr) + plen) { 736 if (m->m_len == m->m_pkthdr.len) { 737 m->m_len = sizeof(struct ip6_hdr) + plen; 738 m->m_pkthdr.len = sizeof(struct ip6_hdr) + plen; 739 } else 740 m_adj(m, sizeof(struct ip6_hdr) + plen - m->m_pkthdr.len); 741 } 742 743 /* 744 * Forward if desirable. 745 */ 746 if (IN6_IS_ADDR_MULTICAST(&ip6->ip6_dst)) { 747 /* 748 * If we are acting as a multicast router, all 749 * incoming multicast packets are passed to the 750 * kernel-level multicast forwarding function. 751 * The packet is returned (relatively) intact; if 752 * ip6_mforward() returns a non-zero value, the packet 753 * must be discarded, else it may be accepted below. 754 */ 755 if (ip6_mrouter && ip6_mforward(ip6, m->m_pkthdr.rcvif, m)) { 756 ip6stat.ip6s_cantforward++; 757 m_freem(m); 758 return; 759 } 760 if (!ours) { 761 m_freem(m); 762 return; 763 } 764 } else if (!ours) { 765 ip6_forward(m, srcrt); 766 return; 767 } 768 769 ip6 = mtod(m, struct ip6_hdr *); 770 771 /* 772 * Malicious party may be able to use IPv4 mapped addr to confuse 773 * tcp/udp stack and bypass security checks (act as if it was from 774 * 127.0.0.1 by using IPv6 src ::ffff:127.0.0.1). Be cautious. 775 * 776 * For SIIT end node behavior, you may want to disable the check. 777 * However, you will become vulnerable to attacks using IPv4 mapped 778 * source. 779 */ 780 if (IN6_IS_ADDR_V4MAPPED(&ip6->ip6_src) || 781 IN6_IS_ADDR_V4MAPPED(&ip6->ip6_dst)) { 782 ip6stat.ip6s_badscope++; 783 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_addrerr); 784 goto bad; 785 } 786 787 /* 788 * Tell launch routine the next header 789 */ 790 ip6stat.ip6s_delivered++; 791 in6_ifstat_inc(deliverifp, ifs6_in_deliver); 792 nest = 0; 793 794 while (nxt != IPPROTO_DONE) { 795 if (ip6_hdrnestlimit && (++nest > ip6_hdrnestlimit)) { 796 ip6stat.ip6s_toomanyhdr++; 797 goto bad; 798 } 799 800 /* 801 * protection against faulty packet - there should be 802 * more sanity checks in header chain processing. 803 */ 804 if (m->m_pkthdr.len < off) { 805 ip6stat.ip6s_tooshort++; 806 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_truncated); 807 goto bad; 808 } 809 810#ifdef IPSEC 811 /* 812 * enforce IPsec policy checking if we are seeing last header. 813 * note that we do not visit this with protocols with pcb layer 814 * code - like udp/tcp/raw ip. 815 */ 816 if ((inet6sw[ip6_protox[nxt]].pr_flags & PR_LASTHDR) != 0 && 817 ipsec6_in_reject(m, NULL)) { 818 ipsec6stat.in_polvio++; 819 goto bad; 820 } 821#endif 822 nxt = (*inet6sw[ip6_protox[nxt]].pr_input)(&m, &off, nxt); 823 } 824 return; 825 bad: 826 m_freem(m); 827} 828 829/* 830 * set/grab in6_ifaddr correspond to IPv6 destination address. 831 * XXX backward compatibility wrapper 832 */ 833static struct ip6aux * 834ip6_setdstifaddr(m, ia6) 835 struct mbuf *m; 836 struct in6_ifaddr *ia6; 837{ 838 struct ip6aux *ip6a; 839 840 ip6a = ip6_addaux(m); 841 if (ip6a) 842 ip6a->ip6a_dstia6 = ia6; 843 return ip6a; /* NULL if failed to set */ 844} 845 846struct in6_ifaddr * 847ip6_getdstifaddr(m) 848 struct mbuf *m; 849{ 850 struct ip6aux *ip6a; 851 852 ip6a = ip6_findaux(m); 853 if (ip6a) 854 return ip6a->ip6a_dstia6; 855 else 856 return NULL; 857} 858 859/* 860 * Hop-by-Hop options header processing. If a valid jumbo payload option is 861 * included, the real payload length will be stored in plenp. 862 */ 863static int 864ip6_hopopts_input(plenp, rtalertp, mp, offp) 865 u_int32_t *plenp; 866 u_int32_t *rtalertp; /* XXX: should be stored more smart way */ 867 struct mbuf **mp; 868 int *offp; 869{ 870 struct mbuf *m = *mp; 871 int off = *offp, hbhlen; 872 struct ip6_hbh *hbh; 873 u_int8_t *opt; 874 875 /* validation of the length of the header */ 876#ifndef PULLDOWN_TEST 877 IP6_EXTHDR_CHECK(m, off, sizeof(*hbh), -1); 878 hbh = (struct ip6_hbh *)(mtod(m, caddr_t) + off); 879 hbhlen = (hbh->ip6h_len + 1) << 3; 880 881 IP6_EXTHDR_CHECK(m, off, hbhlen, -1); 882 hbh = (struct ip6_hbh *)(mtod(m, caddr_t) + off); 883#else 884 IP6_EXTHDR_GET(hbh, struct ip6_hbh *, m, 885 sizeof(struct ip6_hdr), sizeof(struct ip6_hbh)); 886 if (hbh == NULL) { 887 ip6stat.ip6s_tooshort++; 888 return -1; 889 } 890 hbhlen = (hbh->ip6h_len + 1) << 3; 891 IP6_EXTHDR_GET(hbh, struct ip6_hbh *, m, sizeof(struct ip6_hdr), 892 hbhlen); 893 if (hbh == NULL) { 894 ip6stat.ip6s_tooshort++; 895 return -1; 896 } 897#endif 898 off += hbhlen; 899 hbhlen -= sizeof(struct ip6_hbh); 900 opt = (u_int8_t *)hbh + sizeof(struct ip6_hbh); 901 902 if (ip6_process_hopopts(m, (u_int8_t *)hbh + sizeof(struct ip6_hbh), 903 hbhlen, rtalertp, plenp) < 0) 904 return (-1); 905 906 *offp = off; 907 *mp = m; 908 return (0); 909} 910 911/* 912 * Search header for all Hop-by-hop options and process each option. 913 * This function is separate from ip6_hopopts_input() in order to 914 * handle a case where the sending node itself process its hop-by-hop 915 * options header. In such a case, the function is called from ip6_output(). 916 * 917 * The function assumes that hbh header is located right after the IPv6 header 918 * (RFC2460 p7), opthead is pointer into data content in m, and opthead to 919 * opthead + hbhlen is located in continuous memory region. 920 */ 921int 922ip6_process_hopopts(m, opthead, hbhlen, rtalertp, plenp) 923 struct mbuf *m; 924 u_int8_t *opthead; 925 int hbhlen; 926 u_int32_t *rtalertp; 927 u_int32_t *plenp; 928{ 929 struct ip6_hdr *ip6; 930 int optlen = 0; 931 u_int8_t *opt = opthead; 932 u_int16_t rtalert_val; 933 u_int32_t jumboplen; 934 const int erroff = sizeof(struct ip6_hdr) + sizeof(struct ip6_hbh); 935 936 for (; hbhlen > 0; hbhlen -= optlen, opt += optlen) { 937 switch (*opt) { 938 case IP6OPT_PAD1: 939 optlen = 1; 940 break; 941 case IP6OPT_PADN: 942 if (hbhlen < IP6OPT_MINLEN) { 943 ip6stat.ip6s_toosmall++; 944 goto bad; 945 } 946 optlen = *(opt + 1) + 2; 947 break; 948 case IP6OPT_ROUTER_ALERT: 949 /* XXX may need check for alignment */ 950 if (hbhlen < IP6OPT_RTALERT_LEN) { 951 ip6stat.ip6s_toosmall++; 952 goto bad; 953 } 954 if (*(opt + 1) != IP6OPT_RTALERT_LEN - 2) { 955 /* XXX stat */ 956 icmp6_error(m, ICMP6_PARAM_PROB, 957 ICMP6_PARAMPROB_HEADER, 958 erroff + opt + 1 - opthead); 959 return (-1); 960 } 961 optlen = IP6OPT_RTALERT_LEN; 962 bcopy((caddr_t)(opt + 2), (caddr_t)&rtalert_val, 2); 963 *rtalertp = ntohs(rtalert_val); 964 break; 965 case IP6OPT_JUMBO: 966 /* XXX may need check for alignment */ 967 if (hbhlen < IP6OPT_JUMBO_LEN) { 968 ip6stat.ip6s_toosmall++; 969 goto bad; 970 } 971 if (*(opt + 1) != IP6OPT_JUMBO_LEN - 2) { 972 /* XXX stat */ 973 icmp6_error(m, ICMP6_PARAM_PROB, 974 ICMP6_PARAMPROB_HEADER, 975 erroff + opt + 1 - opthead); 976 return (-1); 977 } 978 optlen = IP6OPT_JUMBO_LEN; 979 980 /* 981 * IPv6 packets that have non 0 payload length 982 * must not contain a jumbo payload option. 983 */ 984 ip6 = mtod(m, struct ip6_hdr *); 985 if (ip6->ip6_plen) { 986 ip6stat.ip6s_badoptions++; 987 icmp6_error(m, ICMP6_PARAM_PROB, 988 ICMP6_PARAMPROB_HEADER, 989 erroff + opt - opthead); 990 return (-1); 991 } 992 993 /* 994 * We may see jumbolen in unaligned location, so 995 * we'd need to perform bcopy(). 996 */ 997 bcopy(opt + 2, &jumboplen, sizeof(jumboplen)); 998 jumboplen = (u_int32_t)htonl(jumboplen); 999 1000#if 1 1001 /* 1002 * if there are multiple jumbo payload options, 1003 * *plenp will be non-zero and the packet will be 1004 * rejected. 1005 * the behavior may need some debate in ipngwg - 1006 * multiple options does not make sense, however, 1007 * there's no explicit mention in specification. 1008 */ 1009 if (*plenp != 0) { 1010 ip6stat.ip6s_badoptions++; 1011 icmp6_error(m, ICMP6_PARAM_PROB, 1012 ICMP6_PARAMPROB_HEADER, 1013 erroff + opt + 2 - opthead); 1014 return (-1); 1015 } 1016#endif 1017 1018 /* 1019 * jumbo payload length must be larger than 65535. 1020 */ 1021 if (jumboplen <= IPV6_MAXPACKET) { 1022 ip6stat.ip6s_badoptions++; 1023 icmp6_error(m, ICMP6_PARAM_PROB, 1024 ICMP6_PARAMPROB_HEADER, 1025 erroff + opt + 2 - opthead); 1026 return (-1); 1027 } 1028 *plenp = jumboplen; 1029 1030 break; 1031 default: /* unknown option */ 1032 if (hbhlen < IP6OPT_MINLEN) { 1033 ip6stat.ip6s_toosmall++; 1034 goto bad; 1035 } 1036 optlen = ip6_unknown_opt(opt, m, 1037 erroff + opt - opthead); 1038 if (optlen == -1) 1039 return (-1); 1040 optlen += 2; 1041 break; 1042 } 1043 } 1044 1045 return (0); 1046 1047 bad: 1048 m_freem(m); 1049 return (-1); 1050} 1051 1052/* 1053 * Unknown option processing. 1054 * The third argument `off' is the offset from the IPv6 header to the option, 1055 * which is necessary if the IPv6 header the and option header and IPv6 header 1056 * is not continuous in order to return an ICMPv6 error. 1057 */ 1058int 1059ip6_unknown_opt(optp, m, off) 1060 u_int8_t *optp; 1061 struct mbuf *m; 1062 int off; 1063{ 1064 struct ip6_hdr *ip6; 1065 1066 switch (IP6OPT_TYPE(*optp)) { 1067 case IP6OPT_TYPE_SKIP: /* ignore the option */ 1068 return ((int)*(optp + 1)); 1069 case IP6OPT_TYPE_DISCARD: /* silently discard */ 1070 m_freem(m); 1071 return (-1); 1072 case IP6OPT_TYPE_FORCEICMP: /* send ICMP even if multicasted */ 1073 ip6stat.ip6s_badoptions++; 1074 icmp6_error(m, ICMP6_PARAM_PROB, ICMP6_PARAMPROB_OPTION, off); 1075 return (-1); 1076 case IP6OPT_TYPE_ICMP: /* send ICMP if not multicasted */ 1077 ip6stat.ip6s_badoptions++; 1078 ip6 = mtod(m, struct ip6_hdr *); 1079 if (IN6_IS_ADDR_MULTICAST(&ip6->ip6_dst) || 1080 (m->m_flags & (M_BCAST|M_MCAST))) 1081 m_freem(m); 1082 else 1083 icmp6_error(m, ICMP6_PARAM_PROB, 1084 ICMP6_PARAMPROB_OPTION, off); 1085 return (-1); 1086 } 1087 1088 m_freem(m); /* XXX: NOTREACHED */ 1089 return (-1); 1090} 1091 1092/* 1093 * Create the "control" list for this pcb. 1094 * The function will not modify mbuf chain at all. 1095 * 1096 * with KAME mbuf chain restriction: 1097 * The routine will be called from upper layer handlers like tcp6_input(). 1098 * Thus the routine assumes that the caller (tcp6_input) have already 1099 * called IP6_EXTHDR_CHECK() and all the extension headers are located in the 1100 * very first mbuf on the mbuf chain. 1101 */ 1102void 1103ip6_savecontrol(in6p, m, mp) 1104 struct inpcb *in6p; 1105 struct mbuf *m, **mp; 1106{ 1107#define IS2292(x, y) ((in6p->in6p_flags & IN6P_RFC2292) ? (x) : (y)) 1108 struct thread *td = curthread; /* XXX */ 1109 int privileged = 0; 1110 struct ip6_hdr *ip6 = mtod(m, struct ip6_hdr *); 1111 1112 if (td && !suser(td)) 1113 privileged++; 1114 1115#ifdef SO_TIMESTAMP 1116 if ((in6p->in6p_socket->so_options & SO_TIMESTAMP) != 0) { 1117 struct timeval tv; 1118 1119 microtime(&tv); 1120 *mp = sbcreatecontrol((caddr_t) &tv, sizeof(tv), 1121 SCM_TIMESTAMP, SOL_SOCKET); 1122 if (*mp) 1123 mp = &(*mp)->m_next; 1124 } 1125#endif 1126 1127 if ((ip6->ip6_vfc & IPV6_VERSION_MASK) != IPV6_VERSION) 1128 return; 1129 1130 /* RFC 2292 sec. 5 */ 1131 if ((in6p->in6p_flags & IN6P_PKTINFO) != 0) { 1132 struct in6_pktinfo pi6; 1133 1134 bcopy(&ip6->ip6_dst, &pi6.ipi6_addr, sizeof(struct in6_addr)); 1135 in6_clearscope(&pi6.ipi6_addr); /* XXX */ 1136 pi6.ipi6_ifindex = 1137 (m && m->m_pkthdr.rcvif) ? m->m_pkthdr.rcvif->if_index : 0; 1138 1139 *mp = sbcreatecontrol((caddr_t) &pi6, 1140 sizeof(struct in6_pktinfo), 1141 IS2292(IPV6_2292PKTINFO, IPV6_PKTINFO), IPPROTO_IPV6); 1142 if (*mp) 1143 mp = &(*mp)->m_next; 1144 } 1145 1146 if ((in6p->in6p_flags & IN6P_HOPLIMIT) != 0) { 1147 int hlim = ip6->ip6_hlim & 0xff; 1148 1149 *mp = sbcreatecontrol((caddr_t) &hlim, sizeof(int), 1150 IS2292(IPV6_2292HOPLIMIT, IPV6_HOPLIMIT), IPPROTO_IPV6); 1151 if (*mp) 1152 mp = &(*mp)->m_next; 1153 } 1154 1155 if ((in6p->in6p_flags & IN6P_TCLASS) != 0) { 1156 u_int32_t flowinfo; 1157 int tclass; 1158 1159 flowinfo = (u_int32_t)ntohl(ip6->ip6_flow & IPV6_FLOWINFO_MASK); 1160 flowinfo >>= 20; 1161 1162 tclass = flowinfo & 0xff; 1163 *mp = sbcreatecontrol((caddr_t) &tclass, sizeof(tclass), 1164 IPV6_TCLASS, IPPROTO_IPV6); 1165 if (*mp) 1166 mp = &(*mp)->m_next; 1167 } 1168 1169 /* 1170 * IPV6_HOPOPTS socket option. We require super-user privilege 1171 * for the option, but it might be too strict, since there might 1172 * be some hop-by-hop options which can be returned to normal user. 1173 * See RFC 2292 section 6. 1174 */ 1175 if ((in6p->in6p_flags & IN6P_HOPOPTS) != 0) { 1176#ifdef DIAGNOSTIC 1177 if (!privileged) 1178 panic("IN6P_HOPOPTS is set for unprivileged socket"); 1179#endif 1180 /* 1181 * Check if a hop-by-hop options header is contatined in the 1182 * received packet, and if so, store the options as ancillary 1183 * data. Note that a hop-by-hop options header must be 1184 * just after the IPv6 header, which is assured through the 1185 * IPv6 input processing. 1186 */ 1187 if (ip6->ip6_nxt == IPPROTO_HOPOPTS) { 1188 struct ip6_hbh *hbh; 1189 int hbhlen = 0; 1190#ifdef PULLDOWN_TEST 1191 struct mbuf *ext; 1192#endif 1193 1194#ifndef PULLDOWN_TEST 1195 hbh = (struct ip6_hbh *)(ip6 + 1); 1196 hbhlen = (hbh->ip6h_len + 1) << 3; 1197#else 1198 ext = ip6_pullexthdr(m, sizeof(struct ip6_hdr), 1199 ip6->ip6_nxt); 1200 if (ext == NULL) { 1201 ip6stat.ip6s_tooshort++; 1202 return; 1203 } 1204 hbh = mtod(ext, struct ip6_hbh *); 1205 hbhlen = (hbh->ip6h_len + 1) << 3; 1206 if (hbhlen != ext->m_len) { 1207 m_freem(ext); 1208 ip6stat.ip6s_tooshort++; 1209 return; 1210 } 1211#endif 1212 1213 /* 1214 * XXX: We copy the whole header even if a 1215 * jumbo payload option is included, the option which 1216 * is to be removed before returning according to 1217 * RFC2292. 1218 * Note: this constraint is removed in 2292bis. 1219 */ 1220 *mp = sbcreatecontrol((caddr_t)hbh, hbhlen, 1221 IS2292(IPV6_2292HOPOPTS, IPV6_HOPOPTS), 1222 IPPROTO_IPV6); 1223 if (*mp) 1224 mp = &(*mp)->m_next; 1225#ifdef PULLDOWN_TEST 1226 m_freem(ext); 1227#endif 1228 } 1229 } 1230 1231 if ((in6p->in6p_flags & (IN6P_RTHDR | IN6P_DSTOPTS)) != 0) { 1232 int nxt = ip6->ip6_nxt, off = sizeof(struct ip6_hdr); 1233 1234 /* 1235 * Search for destination options headers or routing 1236 * header(s) through the header chain, and stores each 1237 * header as ancillary data. 1238 * Note that the order of the headers remains in 1239 * the chain of ancillary data. 1240 */ 1241 while (1) { /* is explicit loop prevention necessary? */ 1242 struct ip6_ext *ip6e = NULL; 1243 int elen; 1244#ifdef PULLDOWN_TEST 1245 struct mbuf *ext = NULL; 1246#endif 1247 1248 /* 1249 * if it is not an extension header, don't try to 1250 * pull it from the chain. 1251 */ 1252 switch (nxt) { 1253 case IPPROTO_DSTOPTS: 1254 case IPPROTO_ROUTING: 1255 case IPPROTO_HOPOPTS: 1256 case IPPROTO_AH: /* is it possible? */ 1257 break; 1258 default: 1259 goto loopend; 1260 } 1261 1262#ifndef PULLDOWN_TEST 1263 if (off + sizeof(*ip6e) > m->m_len) 1264 goto loopend; 1265 ip6e = (struct ip6_ext *)(mtod(m, caddr_t) + off); 1266 if (nxt == IPPROTO_AH) 1267 elen = (ip6e->ip6e_len + 2) << 2; 1268 else 1269 elen = (ip6e->ip6e_len + 1) << 3; 1270 if (off + elen > m->m_len) 1271 goto loopend; 1272#else 1273 ext = ip6_pullexthdr(m, off, nxt); 1274 if (ext == NULL) { 1275 ip6stat.ip6s_tooshort++; 1276 return; 1277 } 1278 ip6e = mtod(ext, struct ip6_ext *); 1279 if (nxt == IPPROTO_AH) 1280 elen = (ip6e->ip6e_len + 2) << 2; 1281 else 1282 elen = (ip6e->ip6e_len + 1) << 3; 1283 if (elen != ext->m_len) { 1284 m_freem(ext); 1285 ip6stat.ip6s_tooshort++; 1286 return; 1287 } 1288#endif 1289 1290 switch (nxt) { 1291 case IPPROTO_DSTOPTS: 1292 if (!(in6p->in6p_flags & IN6P_DSTOPTS)) 1293 break; 1294 1295 /* 1296 * We also require super-user privilege for 1297 * the option. See comments on IN6_HOPOPTS. 1298 */ 1299 if (!privileged) 1300 break; 1301 1302 *mp = sbcreatecontrol((caddr_t)ip6e, elen, 1303 IS2292(IPV6_2292DSTOPTS, IPV6_DSTOPTS), 1304 IPPROTO_IPV6); 1305 if (*mp) 1306 mp = &(*mp)->m_next; 1307 break; 1308 case IPPROTO_ROUTING: 1309 if (!in6p->in6p_flags & IN6P_RTHDR) 1310 break; 1311 1312 *mp = sbcreatecontrol((caddr_t)ip6e, elen, 1313 IS2292(IPV6_2292RTHDR, IPV6_RTHDR), 1314 IPPROTO_IPV6); 1315 if (*mp) 1316 mp = &(*mp)->m_next; 1317 break; 1318 case IPPROTO_HOPOPTS: 1319 case IPPROTO_AH: /* is it possible? */ 1320 break; 1321 1322 default: 1323 /* 1324 * other cases have been filtered in the above. 1325 * none will visit this case. here we supply 1326 * the code just in case (nxt overwritten or 1327 * other cases). 1328 */ 1329#ifdef PULLDOWN_TEST 1330 m_freem(ext); 1331#endif 1332 goto loopend; 1333 1334 } 1335 1336 /* proceed with the next header. */ 1337 off += elen; 1338 nxt = ip6e->ip6e_nxt; 1339 ip6e = NULL; 1340#ifdef PULLDOWN_TEST 1341 m_freem(ext); 1342 ext = NULL; 1343#endif 1344 } 1345 loopend: 1346 ; 1347 } 1348 1349#undef IS2292 1350} 1351 1352void 1353ip6_notify_pmtu(in6p, dst, mtu) 1354 struct inpcb *in6p; 1355 struct sockaddr_in6 *dst; 1356 u_int32_t *mtu; 1357{ 1358 struct socket *so; 1359 struct mbuf *m_mtu; 1360 struct ip6_mtuinfo mtuctl; 1361 1362 so = in6p->inp_socket; 1363 1364 if (mtu == NULL) 1365 return; 1366 1367#ifdef DIAGNOSTIC 1368 if (so == NULL) /* I believe this is impossible */ 1369 panic("ip6_notify_pmtu: socket is NULL"); 1370#endif 1371 1372 bzero(&mtuctl, sizeof(mtuctl)); /* zero-clear for safety */ 1373 mtuctl.ip6m_mtu = *mtu; 1374 mtuctl.ip6m_addr = *dst; 1375 in6_recoverscope(&mtuctl.ip6m_addr, &mtuctl.ip6m_addr.sin6_addr, NULL); 1376 1377 if ((m_mtu = sbcreatecontrol((caddr_t)&mtuctl, sizeof(mtuctl), 1378 IPV6_PATHMTU, IPPROTO_IPV6)) == NULL) 1379 return; 1380 1381 if (sbappendaddr(&so->so_rcv, (struct sockaddr *)dst, NULL, m_mtu) 1382 == 0) { 1383 m_freem(m_mtu); 1384 /* XXX: should count statistics */ 1385 } else 1386 sorwakeup(so); 1387 1388 return; 1389} 1390 1391#ifdef PULLDOWN_TEST 1392/* 1393 * pull single extension header from mbuf chain. returns single mbuf that 1394 * contains the result, or NULL on error. 1395 */ 1396static struct mbuf * 1397ip6_pullexthdr(m, off, nxt) 1398 struct mbuf *m; 1399 size_t off; 1400 int nxt; 1401{ 1402 struct ip6_ext ip6e; 1403 size_t elen; 1404 struct mbuf *n; 1405 1406#ifdef DIAGNOSTIC 1407 switch (nxt) { 1408 case IPPROTO_DSTOPTS: 1409 case IPPROTO_ROUTING: 1410 case IPPROTO_HOPOPTS: 1411 case IPPROTO_AH: /* is it possible? */ 1412 break; 1413 default: 1414 printf("ip6_pullexthdr: invalid nxt=%d\n", nxt); 1415 } 1416#endif 1417 1418 m_copydata(m, off, sizeof(ip6e), (caddr_t)&ip6e); 1419 if (nxt == IPPROTO_AH) 1420 elen = (ip6e.ip6e_len + 2) << 2; 1421 else 1422 elen = (ip6e.ip6e_len + 1) << 3; 1423 1424 MGET(n, M_DONTWAIT, MT_DATA); 1425 if (n && elen >= MLEN) { 1426 MCLGET(n, M_DONTWAIT); 1427 if ((n->m_flags & M_EXT) == 0) { 1428 m_free(n); 1429 n = NULL; 1430 } 1431 } 1432 if (!n) 1433 return NULL; 1434 1435 n->m_len = 0; 1436 if (elen >= M_TRAILINGSPACE(n)) { 1437 m_free(n); 1438 return NULL; 1439 } 1440 1441 m_copydata(m, off, elen, mtod(n, caddr_t)); 1442 n->m_len = elen; 1443 return n; 1444} 1445#endif 1446 1447/* 1448 * Get pointer to the previous header followed by the header 1449 * currently processed. 1450 * XXX: This function supposes that 1451 * M includes all headers, 1452 * the next header field and the header length field of each header 1453 * are valid, and 1454 * the sum of each header length equals to OFF. 1455 * Because of these assumptions, this function must be called very 1456 * carefully. Moreover, it will not be used in the near future when 1457 * we develop `neater' mechanism to process extension headers. 1458 */ 1459char * 1460ip6_get_prevhdr(m, off) 1461 struct mbuf *m; 1462 int off; 1463{ 1464 struct ip6_hdr *ip6 = mtod(m, struct ip6_hdr *); 1465 1466 if (off == sizeof(struct ip6_hdr)) 1467 return (&ip6->ip6_nxt); 1468 else { 1469 int len, nxt; 1470 struct ip6_ext *ip6e = NULL; 1471 1472 nxt = ip6->ip6_nxt; 1473 len = sizeof(struct ip6_hdr); 1474 while (len < off) { 1475 ip6e = (struct ip6_ext *)(mtod(m, caddr_t) + len); 1476 1477 switch (nxt) { 1478 case IPPROTO_FRAGMENT: 1479 len += sizeof(struct ip6_frag); 1480 break; 1481 case IPPROTO_AH: 1482 len += (ip6e->ip6e_len + 2) << 2; 1483 break; 1484 default: 1485 len += (ip6e->ip6e_len + 1) << 3; 1486 break; 1487 } 1488 nxt = ip6e->ip6e_nxt; 1489 } 1490 if (ip6e) 1491 return (&ip6e->ip6e_nxt); 1492 else 1493 return NULL; 1494 } 1495} 1496 1497/* 1498 * get next header offset. m will be retained. 1499 */ 1500int 1501ip6_nexthdr(m, off, proto, nxtp) 1502 struct mbuf *m; 1503 int off; 1504 int proto; 1505 int *nxtp; 1506{ 1507 struct ip6_hdr ip6; 1508 struct ip6_ext ip6e; 1509 struct ip6_frag fh; 1510 1511 /* just in case */ 1512 if (m == NULL) 1513 panic("ip6_nexthdr: m == NULL"); 1514 if ((m->m_flags & M_PKTHDR) == 0 || m->m_pkthdr.len < off) 1515 return -1; 1516 1517 switch (proto) { 1518 case IPPROTO_IPV6: 1519 if (m->m_pkthdr.len < off + sizeof(ip6)) 1520 return -1; 1521 m_copydata(m, off, sizeof(ip6), (caddr_t)&ip6); 1522 if (nxtp) 1523 *nxtp = ip6.ip6_nxt; 1524 off += sizeof(ip6); 1525 return off; 1526 1527 case IPPROTO_FRAGMENT: 1528 /* 1529 * terminate parsing if it is not the first fragment, 1530 * it does not make sense to parse through it. 1531 */ 1532 if (m->m_pkthdr.len < off + sizeof(fh)) 1533 return -1; 1534 m_copydata(m, off, sizeof(fh), (caddr_t)&fh); 1535 /* IP6F_OFF_MASK = 0xfff8(BigEndian), 0xf8ff(LittleEndian) */ 1536 if (fh.ip6f_offlg & IP6F_OFF_MASK) 1537 return -1; 1538 if (nxtp) 1539 *nxtp = fh.ip6f_nxt; 1540 off += sizeof(struct ip6_frag); 1541 return off; 1542 1543 case IPPROTO_AH: 1544 if (m->m_pkthdr.len < off + sizeof(ip6e)) 1545 return -1; 1546 m_copydata(m, off, sizeof(ip6e), (caddr_t)&ip6e); 1547 if (nxtp) 1548 *nxtp = ip6e.ip6e_nxt; 1549 off += (ip6e.ip6e_len + 2) << 2; 1550 return off; 1551 1552 case IPPROTO_HOPOPTS: 1553 case IPPROTO_ROUTING: 1554 case IPPROTO_DSTOPTS: 1555 if (m->m_pkthdr.len < off + sizeof(ip6e)) 1556 return -1; 1557 m_copydata(m, off, sizeof(ip6e), (caddr_t)&ip6e); 1558 if (nxtp) 1559 *nxtp = ip6e.ip6e_nxt; 1560 off += (ip6e.ip6e_len + 1) << 3; 1561 return off; 1562 1563 case IPPROTO_NONE: 1564 case IPPROTO_ESP: 1565 case IPPROTO_IPCOMP: 1566 /* give up */ 1567 return -1; 1568 1569 default: 1570 return -1; 1571 } 1572 1573 return -1; 1574} 1575 1576/* 1577 * get offset for the last header in the chain. m will be kept untainted. 1578 */ 1579int 1580ip6_lasthdr(m, off, proto, nxtp) 1581 struct mbuf *m; 1582 int off; 1583 int proto; 1584 int *nxtp; 1585{ 1586 int newoff; 1587 int nxt; 1588 1589 if (!nxtp) { 1590 nxt = -1; 1591 nxtp = &nxt; 1592 } 1593 while (1) { 1594 newoff = ip6_nexthdr(m, off, proto, nxtp); 1595 if (newoff < 0) 1596 return off; 1597 else if (newoff < off) 1598 return -1; /* invalid */ 1599 else if (newoff == off) 1600 return newoff; 1601 1602 off = newoff; 1603 proto = *nxtp; 1604 } 1605} 1606 1607struct ip6aux * 1608ip6_addaux(m) 1609 struct mbuf *m; 1610{ 1611 struct m_tag *mtag; 1612 1613 mtag = m_tag_find(m, PACKET_TAG_IPV6_INPUT, NULL); 1614 if (!mtag) { 1615 mtag = m_tag_get(PACKET_TAG_IPV6_INPUT, sizeof(struct ip6aux), 1616 M_NOWAIT); 1617 if (mtag) { 1618 m_tag_prepend(m, mtag); 1619 bzero(mtag + 1, sizeof(struct ip6aux)); 1620 } 1621 } 1622 return mtag ? (struct ip6aux *)(mtag + 1) : NULL; 1623} 1624 1625struct ip6aux * 1626ip6_findaux(m) 1627 struct mbuf *m; 1628{ 1629 struct m_tag *mtag; 1630 1631 mtag = m_tag_find(m, PACKET_TAG_IPV6_INPUT, NULL); 1632 return mtag ? (struct ip6aux *)(mtag + 1) : NULL; 1633} 1634 1635void 1636ip6_delaux(m) 1637 struct mbuf *m; 1638{ 1639 struct m_tag *mtag; 1640 1641 mtag = m_tag_find(m, PACKET_TAG_IPV6_INPUT, NULL); 1642 if (mtag) 1643 m_tag_delete(m, mtag); 1644} 1645 1646/* 1647 * System control for IP6 1648 */ 1649 1650u_char inet6ctlerrmap[PRC_NCMDS] = { 1651 0, 0, 0, 0, 1652 0, EMSGSIZE, EHOSTDOWN, EHOSTUNREACH, 1653 EHOSTUNREACH, EHOSTUNREACH, ECONNREFUSED, ECONNREFUSED, 1654 EMSGSIZE, EHOSTUNREACH, 0, 0, 1655 0, 0, 0, 0, 1656 ENOPROTOOPT 1657}; 1658