ip6_input.c revision 96972
1/* $FreeBSD: head/sys/netinet6/ip6_input.c 96972 2002-05-20 05:41:09Z tanimura $ */ 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 74#include <sys/param.h> 75#include <sys/systm.h> 76#include <sys/malloc.h> 77#include <sys/mbuf.h> 78#include <sys/proc.h> 79#include <sys/domain.h> 80#include <sys/protosw.h> 81#include <sys/socket.h> 82#include <sys/socketvar.h> 83#include <sys/errno.h> 84#include <sys/time.h> 85#include <sys/kernel.h> 86#include <sys/syslog.h> 87 88#include <net/if.h> 89#include <net/if_types.h> 90#include <net/if_dl.h> 91#include <net/route.h> 92#include <net/netisr.h> 93#include <net/intrq.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/in6_ifattach.h> 110#include <netinet6/nd6.h> 111#include <netinet6/in6_prefix.h> 112 113#ifdef IPSEC 114#include <netinet6/ipsec.h> 115#ifdef INET6 116#include <netinet6/ipsec6.h> 117#endif 118#endif 119 120#include <netinet6/ip6_fw.h> 121 122#include <netinet6/ip6protosw.h> 123 124#include <net/net_osdep.h> 125 126extern struct domain inet6domain; 127 128u_char ip6_protox[IPPROTO_MAX]; 129static int ip6qmaxlen = IFQ_MAXLEN; 130struct in6_ifaddr *in6_ifaddr; 131 132extern struct callout in6_tmpaddrtimer_ch; 133 134int ip6_forward_srcrt; /* XXX */ 135int ip6_sourcecheck; /* XXX */ 136int ip6_sourcecheck_interval; /* XXX */ 137 138int ip6_ours_check_algorithm; 139 140 141/* firewall hooks */ 142ip6_fw_chk_t *ip6_fw_chk_ptr; 143ip6_fw_ctl_t *ip6_fw_ctl_ptr; 144int ip6_fw_enable = 1; 145 146struct ip6stat ip6stat; 147 148static void ip6_init2 __P((void *)); 149static struct mbuf *ip6_setdstifaddr __P((struct mbuf *, struct in6_ifaddr *)); 150static int ip6_hopopts_input __P((u_int32_t *, u_int32_t *, struct mbuf **, int *)); 151#ifdef PULLDOWN_TEST 152static struct mbuf *ip6_pullexthdr __P((struct mbuf *, size_t, int)); 153#endif 154 155 156/* 157 * IP6 initialization: fill in IP6 protocol switch table. 158 * All protocols not implemented in kernel go to raw IP6 protocol handler. 159 */ 160void 161ip6_init() 162{ 163 struct ip6protosw *pr; 164 int i; 165 struct timeval tv; 166 167#ifdef DIAGNOSTIC 168 if (sizeof(struct protosw) != sizeof(struct ip6protosw)) 169 panic("sizeof(protosw) != sizeof(ip6protosw)"); 170#endif 171 pr = (struct ip6protosw *)pffindproto(PF_INET6, IPPROTO_RAW, SOCK_RAW); 172 if (pr == 0) 173 panic("ip6_init"); 174 for (i = 0; i < IPPROTO_MAX; i++) 175 ip6_protox[i] = pr - inet6sw; 176 for (pr = (struct ip6protosw *)inet6domain.dom_protosw; 177 pr < (struct ip6protosw *)inet6domain.dom_protoswNPROTOSW; pr++) 178 if (pr->pr_domain->dom_family == PF_INET6 && 179 pr->pr_protocol && pr->pr_protocol != IPPROTO_RAW) 180 ip6_protox[pr->pr_protocol] = pr - inet6sw; 181 ip6intrq.ifq_maxlen = ip6qmaxlen; 182 mtx_init(&ip6intrq.ifq_mtx, "ip6_inq", NULL, MTX_DEF); 183 ip6intrq_present = 1; 184 register_netisr(NETISR_IPV6, ip6intr); 185 nd6_init(); 186 frag6_init(); 187 /* 188 * in many cases, random() here does NOT return random number 189 * as initialization during bootstrap time occur in fixed order. 190 */ 191 microtime(&tv); 192 ip6_flow_seq = random() ^ tv.tv_usec; 193 microtime(&tv); 194 ip6_desync_factor = (random() ^ tv.tv_usec) % MAX_TEMP_DESYNC_FACTOR; 195} 196 197static void 198ip6_init2(dummy) 199 void *dummy; 200{ 201 202 /* 203 * to route local address of p2p link to loopback, 204 * assign loopback address first. 205 */ 206 in6_ifattach(&loif[0], NULL); 207 208 /* nd6_timer_init */ 209 callout_init(&nd6_timer_ch, 0); 210 callout_reset(&nd6_timer_ch, hz, nd6_timer, NULL); 211 212 /* router renumbering prefix list maintenance */ 213 callout_init(&in6_rr_timer_ch, 0); 214 callout_reset(&in6_rr_timer_ch, hz, in6_rr_timer, NULL); 215 216 /* timer for regeneranation of temporary addresses randomize ID */ 217 callout_reset(&in6_tmpaddrtimer_ch, 218 (ip6_temp_preferred_lifetime - ip6_desync_factor - 219 ip6_temp_regen_advance) * hz, 220 in6_tmpaddrtimer, NULL); 221} 222 223/* cheat */ 224/* This must be after route_init(), which is now SI_ORDER_THIRD */ 225SYSINIT(netinet6init2, SI_SUB_PROTO_DOMAIN, SI_ORDER_MIDDLE, ip6_init2, NULL); 226 227/* 228 * IP6 input interrupt handling. Just pass the packet to ip6_input. 229 */ 230void 231ip6intr() 232{ 233 int s; 234 struct mbuf *m; 235 236 for (;;) { 237 s = splimp(); 238 IF_DEQUEUE(&ip6intrq, m); 239 splx(s); 240 if (m == 0) 241 return; 242 ip6_input(m); 243 } 244} 245 246extern struct route_in6 ip6_forward_rt; 247 248void 249ip6_input(m) 250 struct mbuf *m; 251{ 252 struct ip6_hdr *ip6; 253 int off = sizeof(struct ip6_hdr), nest; 254 u_int32_t plen; 255 u_int32_t rtalert = ~0; 256 int nxt, ours = 0; 257 struct ifnet *deliverifp = NULL; 258#ifdef PFIL_HOOKS 259 struct packet_filter_hook *pfh; 260 struct mbuf *m0; 261 int rv; 262#endif /* PFIL_HOOKS */ 263 264#ifdef IPSEC 265 /* 266 * should the inner packet be considered authentic? 267 * see comment in ah4_input(). 268 */ 269 if (m) { 270 m->m_flags &= ~M_AUTHIPHDR; 271 m->m_flags &= ~M_AUTHIPDGM; 272 } 273#endif 274 275 /* 276 * make sure we don't have onion peering information into m_aux. 277 */ 278 ip6_delaux(m); 279 280 /* 281 * mbuf statistics 282 */ 283 if (m->m_flags & M_EXT) { 284 if (m->m_next) 285 ip6stat.ip6s_mext2m++; 286 else 287 ip6stat.ip6s_mext1++; 288 } else { 289#define M2MMAX (sizeof(ip6stat.ip6s_m2m)/sizeof(ip6stat.ip6s_m2m[0])) 290 if (m->m_next) { 291 if (m->m_flags & M_LOOP) { 292 ip6stat.ip6s_m2m[loif[0].if_index]++; /* XXX */ 293 } else if (m->m_pkthdr.rcvif->if_index < M2MMAX) 294 ip6stat.ip6s_m2m[m->m_pkthdr.rcvif->if_index]++; 295 else 296 ip6stat.ip6s_m2m[0]++; 297 } else 298 ip6stat.ip6s_m1++; 299#undef M2MMAX 300 } 301 302 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_receive); 303 ip6stat.ip6s_total++; 304 305#ifndef PULLDOWN_TEST 306 /* 307 * L2 bridge code and some other code can return mbuf chain 308 * that does not conform to KAME requirement. too bad. 309 * XXX: fails to join if interface MTU > MCLBYTES. jumbogram? 310 */ 311 if (m && m->m_next != NULL && m->m_pkthdr.len < MCLBYTES) { 312 struct mbuf *n; 313 314 MGETHDR(n, M_DONTWAIT, MT_HEADER); 315 if (n) 316 M_COPY_PKTHDR(n, m); 317 if (n && m->m_pkthdr.len > MHLEN) { 318 MCLGET(n, M_DONTWAIT); 319 if ((n->m_flags & M_EXT) == 0) { 320 m_freem(n); 321 n = NULL; 322 } 323 } 324 if (n == NULL) { 325 m_freem(m); 326 return; /*ENOBUFS*/ 327 } 328 329 m_copydata(m, 0, m->m_pkthdr.len, mtod(n, caddr_t)); 330 n->m_len = m->m_pkthdr.len; 331 m_freem(m); 332 m = n; 333 } 334 IP6_EXTHDR_CHECK(m, 0, sizeof(struct ip6_hdr), /*nothing*/); 335#endif 336 337 if (m->m_len < sizeof(struct ip6_hdr)) { 338 struct ifnet *inifp; 339 inifp = m->m_pkthdr.rcvif; 340 if ((m = m_pullup(m, sizeof(struct ip6_hdr))) == 0) { 341 ip6stat.ip6s_toosmall++; 342 in6_ifstat_inc(inifp, ifs6_in_hdrerr); 343 return; 344 } 345 } 346 347 ip6 = mtod(m, struct ip6_hdr *); 348 349 if ((ip6->ip6_vfc & IPV6_VERSION_MASK) != IPV6_VERSION) { 350 ip6stat.ip6s_badvers++; 351 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_hdrerr); 352 goto bad; 353 } 354 355#ifdef PFIL_HOOKS 356 /* 357 * Run through list of hooks for input packets. If there are any 358 * filters which require that additional packets in the flow are 359 * not fast-forwarded, they must clear the M_CANFASTFWD flag. 360 * Note that filters must _never_ set this flag, as another filter 361 * in the list may have previously cleared it. 362 */ 363 m0 = m; 364 pfh = pfil_hook_get(PFIL_IN, &inet6sw[ip6_protox[IPPROTO_IPV6]].pr_pfh); 365 for (; pfh; pfh = pfh->pfil_link.tqe_next) 366 if (pfh->pfil_func) { 367 rv = pfh->pfil_func(ip6, sizeof(*ip6), 368 m->m_pkthdr.rcvif, 0, &m0); 369 if (rv) 370 return; 371 m = m0; 372 if (m == NULL) 373 return; 374 ip6 = mtod(m, struct ip6_hdr *); 375 } 376#endif /* PFIL_HOOKS */ 377 378 ip6stat.ip6s_nxthist[ip6->ip6_nxt]++; 379 380 /* 381 * Check with the firewall... 382 */ 383 if (ip6_fw_enable && ip6_fw_chk_ptr) { 384 u_short port = 0; 385 /* If ipfw says divert, we have to just drop packet */ 386 /* use port as a dummy argument */ 387 if ((*ip6_fw_chk_ptr)(&ip6, NULL, &port, &m)) { 388 m_freem(m); 389 m = NULL; 390 } 391 if (!m) 392 return; 393 } 394 395 /* 396 * Check against address spoofing/corruption. 397 */ 398 if (IN6_IS_ADDR_MULTICAST(&ip6->ip6_src) || 399 IN6_IS_ADDR_UNSPECIFIED(&ip6->ip6_dst)) { 400 /* 401 * XXX: "badscope" is not very suitable for a multicast source. 402 */ 403 ip6stat.ip6s_badscope++; 404 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_addrerr); 405 goto bad; 406 } 407 if ((IN6_IS_ADDR_LOOPBACK(&ip6->ip6_src) || 408 IN6_IS_ADDR_LOOPBACK(&ip6->ip6_dst)) && 409 (m->m_pkthdr.rcvif->if_flags & IFF_LOOPBACK) == 0) { 410 ip6stat.ip6s_badscope++; 411 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_addrerr); 412 goto bad; 413 } 414 415 /* 416 * The following check is not documented in specs. A malicious 417 * party may be able to use IPv4 mapped addr to confuse tcp/udp stack 418 * and bypass security checks (act as if it was from 127.0.0.1 by using 419 * IPv6 src ::ffff:127.0.0.1). Be cautious. 420 * 421 * This check chokes if we are in an SIIT cloud. As none of BSDs 422 * support IPv4-less kernel compilation, we cannot support SIIT 423 * environment at all. So, it makes more sense for us to reject any 424 * malicious packets for non-SIIT environment, than try to do a 425 * partical support for SIIT environment. 426 */ 427 if (IN6_IS_ADDR_V4MAPPED(&ip6->ip6_src) || 428 IN6_IS_ADDR_V4MAPPED(&ip6->ip6_dst)) { 429 ip6stat.ip6s_badscope++; 430 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_addrerr); 431 goto bad; 432 } 433#if 0 434 /* 435 * Reject packets with IPv4 compatible addresses (auto tunnel). 436 * 437 * The code forbids auto tunnel relay case in RFC1933 (the check is 438 * stronger than RFC1933). We may want to re-enable it if mech-xx 439 * is revised to forbid relaying case. 440 */ 441 if (IN6_IS_ADDR_V4COMPAT(&ip6->ip6_src) || 442 IN6_IS_ADDR_V4COMPAT(&ip6->ip6_dst)) { 443 ip6stat.ip6s_badscope++; 444 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_addrerr); 445 goto bad; 446 } 447#endif 448 449 /* drop packets if interface ID portion is already filled */ 450 if ((m->m_pkthdr.rcvif->if_flags & IFF_LOOPBACK) == 0) { 451 if (IN6_IS_SCOPE_LINKLOCAL(&ip6->ip6_src) && 452 ip6->ip6_src.s6_addr16[1]) { 453 ip6stat.ip6s_badscope++; 454 goto bad; 455 } 456 if (IN6_IS_SCOPE_LINKLOCAL(&ip6->ip6_dst) && 457 ip6->ip6_dst.s6_addr16[1]) { 458 ip6stat.ip6s_badscope++; 459 goto bad; 460 } 461 } 462 463 if (IN6_IS_SCOPE_LINKLOCAL(&ip6->ip6_src)) 464 ip6->ip6_src.s6_addr16[1] 465 = htons(m->m_pkthdr.rcvif->if_index); 466 if (IN6_IS_SCOPE_LINKLOCAL(&ip6->ip6_dst)) 467 ip6->ip6_dst.s6_addr16[1] 468 = htons(m->m_pkthdr.rcvif->if_index); 469 470#if 0 /* this case seems to be unnecessary. (jinmei, 20010401) */ 471 /* 472 * We use rt->rt_ifp to determine if the address is ours or not. 473 * If rt_ifp is lo0, the address is ours. 474 * The problem here is, rt->rt_ifp for fe80::%lo0/64 is set to lo0, 475 * so any address under fe80::%lo0/64 will be mistakenly considered 476 * local. The special case is supplied to handle the case properly 477 * by actually looking at interface addresses 478 * (using in6ifa_ifpwithaddr). 479 */ 480 if ((m->m_pkthdr.rcvif->if_flags & IFF_LOOPBACK) != 0 && 481 IN6_IS_ADDR_LINKLOCAL(&ip6->ip6_dst)) { 482 if (!in6ifa_ifpwithaddr(m->m_pkthdr.rcvif, &ip6->ip6_dst)) { 483 icmp6_error(m, ICMP6_DST_UNREACH, 484 ICMP6_DST_UNREACH_ADDR, 0); 485 /* m is already freed */ 486 return; 487 } 488 489 ours = 1; 490 deliverifp = m->m_pkthdr.rcvif; 491 goto hbhcheck; 492 } 493#endif 494 495 /* 496 * Multicast check 497 */ 498 if (IN6_IS_ADDR_MULTICAST(&ip6->ip6_dst)) { 499 struct in6_multi *in6m = 0; 500 501 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_mcast); 502 /* 503 * See if we belong to the destination multicast group on the 504 * arrival interface. 505 */ 506 IN6_LOOKUP_MULTI(ip6->ip6_dst, m->m_pkthdr.rcvif, in6m); 507 if (in6m) 508 ours = 1; 509 else if (!ip6_mrouter) { 510 ip6stat.ip6s_notmember++; 511 ip6stat.ip6s_cantforward++; 512 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_discard); 513 goto bad; 514 } 515 deliverifp = m->m_pkthdr.rcvif; 516 goto hbhcheck; 517 } 518 519 /* 520 * Unicast check 521 */ 522 switch (ip6_ours_check_algorithm) { 523 default: 524 /* 525 * XXX: I intentionally broke our indentation rule here, 526 * since this switch-case is just for measurement and 527 * therefore should soon be removed. 528 */ 529 if (ip6_forward_rt.ro_rt != NULL && 530 (ip6_forward_rt.ro_rt->rt_flags & RTF_UP) != 0 && 531 IN6_ARE_ADDR_EQUAL(&ip6->ip6_dst, 532 &((struct sockaddr_in6 *)(&ip6_forward_rt.ro_dst))->sin6_addr)) 533 ip6stat.ip6s_forward_cachehit++; 534 else { 535 struct sockaddr_in6 *dst6; 536 537 if (ip6_forward_rt.ro_rt) { 538 /* route is down or destination is different */ 539 ip6stat.ip6s_forward_cachemiss++; 540 RTFREE(ip6_forward_rt.ro_rt); 541 ip6_forward_rt.ro_rt = 0; 542 } 543 544 bzero(&ip6_forward_rt.ro_dst, sizeof(struct sockaddr_in6)); 545 dst6 = (struct sockaddr_in6 *)&ip6_forward_rt.ro_dst; 546 dst6->sin6_len = sizeof(struct sockaddr_in6); 547 dst6->sin6_family = AF_INET6; 548 dst6->sin6_addr = ip6->ip6_dst; 549#ifdef SCOPEDROUTING 550 ip6_forward_rt.ro_dst.sin6_scope_id = 551 in6_addr2scopeid(m->m_pkthdr.rcvif, &ip6->ip6_dst); 552#endif 553 554 rtalloc_ign((struct route *)&ip6_forward_rt, RTF_PRCLONING); 555 } 556 557#define rt6_key(r) ((struct sockaddr_in6 *)((r)->rt_nodes->rn_key)) 558 559 /* 560 * Accept the packet if the forwarding interface to the destination 561 * according to the routing table is the loopback interface, 562 * unless the associated route has a gateway. 563 * Note that this approach causes to accept a packet if there is a 564 * route to the loopback interface for the destination of the packet. 565 * But we think it's even useful in some situations, e.g. when using 566 * a special daemon which wants to intercept the packet. 567 * 568 * XXX: some OSes automatically make a cloned route for the destination 569 * of an outgoing packet. If the outgoing interface of the packet 570 * is a loopback one, the kernel would consider the packet to be 571 * accepted, even if we have no such address assinged on the interface. 572 * We check the cloned flag of the route entry to reject such cases, 573 * assuming that route entries for our own addresses are not made by 574 * cloning (it should be true because in6_addloop explicitly installs 575 * the host route). However, we might have to do an explicit check 576 * while it would be less efficient. Or, should we rather install a 577 * reject route for such a case? 578 */ 579 if (ip6_forward_rt.ro_rt && 580 (ip6_forward_rt.ro_rt->rt_flags & 581 (RTF_HOST|RTF_GATEWAY)) == RTF_HOST && 582#ifdef RTF_WASCLONED 583 !(ip6_forward_rt.ro_rt->rt_flags & RTF_WASCLONED) && 584#endif 585#ifdef RTF_CLONED 586 !(ip6_forward_rt.ro_rt->rt_flags & RTF_CLONED) && 587#endif 588#if 0 589 /* 590 * The check below is redundant since the comparison of 591 * the destination and the key of the rtentry has 592 * already done through looking up the routing table. 593 */ 594 IN6_ARE_ADDR_EQUAL(&ip6->ip6_dst, 595 &rt6_key(ip6_forward_rt.ro_rt)->sin6_addr) 596#endif 597 ip6_forward_rt.ro_rt->rt_ifp->if_type == IFT_LOOP) { 598 struct in6_ifaddr *ia6 = 599 (struct in6_ifaddr *)ip6_forward_rt.ro_rt->rt_ifa; 600 601 /* 602 * record address information into m_aux. 603 */ 604 (void)ip6_setdstifaddr(m, ia6); 605 606 /* 607 * packets to a tentative, duplicated, or somehow invalid 608 * address must not be accepted. 609 */ 610 if (!(ia6->ia6_flags & IN6_IFF_NOTREADY)) { 611 /* this address is ready */ 612 ours = 1; 613 deliverifp = ia6->ia_ifp; /* correct? */ 614 /* Count the packet in the ip address stats */ 615 ia6->ia_ifa.if_ipackets++; 616 ia6->ia_ifa.if_ibytes += m->m_pkthdr.len; 617 goto hbhcheck; 618 } else { 619 /* address is not ready, so discard the packet. */ 620 nd6log((LOG_INFO, 621 "ip6_input: packet to an unready address %s->%s\n", 622 ip6_sprintf(&ip6->ip6_src), 623 ip6_sprintf(&ip6->ip6_dst))); 624 625 goto bad; 626 } 627 } 628 } /* XXX indentation (see above) */ 629 630 /* 631 * FAITH(Firewall Aided Internet Translator) 632 */ 633 if (ip6_keepfaith) { 634 if (ip6_forward_rt.ro_rt && ip6_forward_rt.ro_rt->rt_ifp 635 && ip6_forward_rt.ro_rt->rt_ifp->if_type == IFT_FAITH) { 636 /* XXX do we need more sanity checks? */ 637 ours = 1; 638 deliverifp = ip6_forward_rt.ro_rt->rt_ifp; /* faith */ 639 goto hbhcheck; 640 } 641 } 642 643 /* 644 * Now there is no reason to process the packet if it's not our own 645 * and we're not a router. 646 */ 647 if (!ip6_forwarding) { 648 ip6stat.ip6s_cantforward++; 649 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_discard); 650 goto bad; 651 } 652 653 hbhcheck: 654 /* 655 * record address information into m_aux, if we don't have one yet. 656 * note that we are unable to record it, if the address is not listed 657 * as our interface address (e.g. multicast addresses, addresses 658 * within FAITH prefixes and such). 659 */ 660 if (deliverifp && !ip6_getdstifaddr(m)) { 661 struct in6_ifaddr *ia6; 662 663 ia6 = in6_ifawithifp(deliverifp, &ip6->ip6_dst); 664 if (ia6) { 665 if (!ip6_setdstifaddr(m, ia6)) { 666 /* 667 * XXX maybe we should drop the packet here, 668 * as we could not provide enough information 669 * to the upper layers. 670 */ 671 } 672 } 673 } 674 675 /* 676 * Process Hop-by-Hop options header if it's contained. 677 * m may be modified in ip6_hopopts_input(). 678 * If a JumboPayload option is included, plen will also be modified. 679 */ 680 plen = (u_int32_t)ntohs(ip6->ip6_plen); 681 if (ip6->ip6_nxt == IPPROTO_HOPOPTS) { 682 struct ip6_hbh *hbh; 683 684 if (ip6_hopopts_input(&plen, &rtalert, &m, &off)) { 685#if 0 /*touches NULL pointer*/ 686 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_discard); 687#endif 688 return; /* m have already been freed */ 689 } 690 691 /* adjust pointer */ 692 ip6 = mtod(m, struct ip6_hdr *); 693 694 /* 695 * if the payload length field is 0 and the next header field 696 * indicates Hop-by-Hop Options header, then a Jumbo Payload 697 * option MUST be included. 698 */ 699 if (ip6->ip6_plen == 0 && plen == 0) { 700 /* 701 * Note that if a valid jumbo payload option is 702 * contained, ip6_hoptops_input() must set a valid 703 * (non-zero) payload length to the variable plen. 704 */ 705 ip6stat.ip6s_badoptions++; 706 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_discard); 707 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_hdrerr); 708 icmp6_error(m, ICMP6_PARAM_PROB, 709 ICMP6_PARAMPROB_HEADER, 710 (caddr_t)&ip6->ip6_plen - (caddr_t)ip6); 711 return; 712 } 713#ifndef PULLDOWN_TEST 714 /* ip6_hopopts_input() ensures that mbuf is contiguous */ 715 hbh = (struct ip6_hbh *)(ip6 + 1); 716#else 717 IP6_EXTHDR_GET(hbh, struct ip6_hbh *, m, sizeof(struct ip6_hdr), 718 sizeof(struct ip6_hbh)); 719 if (hbh == NULL) { 720 ip6stat.ip6s_tooshort++; 721 return; 722 } 723#endif 724 nxt = hbh->ip6h_nxt; 725 726 /* 727 * accept the packet if a router alert option is included 728 * and we act as an IPv6 router. 729 */ 730 if (rtalert != ~0 && ip6_forwarding) 731 ours = 1; 732 } else 733 nxt = ip6->ip6_nxt; 734 735 /* 736 * Check that the amount of data in the buffers 737 * is as at least much as the IPv6 header would have us expect. 738 * Trim mbufs if longer than we expect. 739 * Drop packet if shorter than we expect. 740 */ 741 if (m->m_pkthdr.len - sizeof(struct ip6_hdr) < plen) { 742 ip6stat.ip6s_tooshort++; 743 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_truncated); 744 goto bad; 745 } 746 if (m->m_pkthdr.len > sizeof(struct ip6_hdr) + plen) { 747 if (m->m_len == m->m_pkthdr.len) { 748 m->m_len = sizeof(struct ip6_hdr) + plen; 749 m->m_pkthdr.len = sizeof(struct ip6_hdr) + plen; 750 } else 751 m_adj(m, sizeof(struct ip6_hdr) + plen - m->m_pkthdr.len); 752 } 753 754 /* 755 * Forward if desirable. 756 */ 757 if (IN6_IS_ADDR_MULTICAST(&ip6->ip6_dst)) { 758 /* 759 * If we are acting as a multicast router, all 760 * incoming multicast packets are passed to the 761 * kernel-level multicast forwarding function. 762 * The packet is returned (relatively) intact; if 763 * ip6_mforward() returns a non-zero value, the packet 764 * must be discarded, else it may be accepted below. 765 */ 766 if (ip6_mrouter && ip6_mforward(ip6, m->m_pkthdr.rcvif, m)) { 767 ip6stat.ip6s_cantforward++; 768 m_freem(m); 769 return; 770 } 771 if (!ours) { 772 m_freem(m); 773 return; 774 } 775 } else if (!ours) { 776 ip6_forward(m, 0); 777 return; 778 } 779 780 ip6 = mtod(m, struct ip6_hdr *); 781 782 /* 783 * Malicious party may be able to use IPv4 mapped addr to confuse 784 * tcp/udp stack and bypass security checks (act as if it was from 785 * 127.0.0.1 by using IPv6 src ::ffff:127.0.0.1). Be cautious. 786 * 787 * For SIIT end node behavior, you may want to disable the check. 788 * However, you will become vulnerable to attacks using IPv4 mapped 789 * source. 790 */ 791 if (IN6_IS_ADDR_V4MAPPED(&ip6->ip6_src) || 792 IN6_IS_ADDR_V4MAPPED(&ip6->ip6_dst)) { 793 ip6stat.ip6s_badscope++; 794 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_addrerr); 795 goto bad; 796 } 797 798 /* 799 * Tell launch routine the next header 800 */ 801 ip6stat.ip6s_delivered++; 802 in6_ifstat_inc(deliverifp, ifs6_in_deliver); 803 nest = 0; 804 805 while (nxt != IPPROTO_DONE) { 806 if (ip6_hdrnestlimit && (++nest > ip6_hdrnestlimit)) { 807 ip6stat.ip6s_toomanyhdr++; 808 goto bad; 809 } 810 811 /* 812 * protection against faulty packet - there should be 813 * more sanity checks in header chain processing. 814 */ 815 if (m->m_pkthdr.len < off) { 816 ip6stat.ip6s_tooshort++; 817 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_truncated); 818 goto bad; 819 } 820 821#if 0 822 /* 823 * do we need to do it for every header? yeah, other 824 * functions can play with it (like re-allocate and copy). 825 */ 826 mhist = ip6_addaux(m); 827 if (mhist && M_TRAILINGSPACE(mhist) >= sizeof(nxt)) { 828 hist = mtod(mhist, caddr_t) + mhist->m_len; 829 bcopy(&nxt, hist, sizeof(nxt)); 830 mhist->m_len += sizeof(nxt); 831 } else { 832 ip6stat.ip6s_toomanyhdr++; 833 goto bad; 834 } 835#endif 836 837#ifdef IPSEC 838 /* 839 * enforce IPsec policy checking if we are seeing last header. 840 * note that we do not visit this with protocols with pcb layer 841 * code - like udp/tcp/raw ip. 842 */ 843 if ((inet6sw[ip6_protox[nxt]].pr_flags & PR_LASTHDR) != 0 && 844 ipsec6_in_reject(m, NULL)) { 845 ipsec6stat.in_polvio++; 846 goto bad; 847 } 848#endif 849 850 nxt = (*inet6sw[ip6_protox[nxt]].pr_input)(&m, &off, nxt); 851 } 852 return; 853 bad: 854 m_freem(m); 855} 856 857/* 858 * set/grab in6_ifaddr correspond to IPv6 destination address. 859 * XXX backward compatibility wrapper 860 */ 861static struct mbuf * 862ip6_setdstifaddr(m, ia6) 863 struct mbuf *m; 864 struct in6_ifaddr *ia6; 865{ 866 struct mbuf *n; 867 868 n = ip6_addaux(m); 869 if (n) 870 mtod(n, struct ip6aux *)->ip6a_dstia6 = ia6; 871 return n; /* NULL if failed to set */ 872} 873 874struct in6_ifaddr * 875ip6_getdstifaddr(m) 876 struct mbuf *m; 877{ 878 struct mbuf *n; 879 880 n = ip6_findaux(m); 881 if (n) 882 return mtod(n, struct ip6aux *)->ip6a_dstia6; 883 else 884 return NULL; 885} 886 887/* 888 * Hop-by-Hop options header processing. If a valid jumbo payload option is 889 * included, the real payload length will be stored in plenp. 890 */ 891static int 892ip6_hopopts_input(plenp, rtalertp, mp, offp) 893 u_int32_t *plenp; 894 u_int32_t *rtalertp; /* XXX: should be stored more smart way */ 895 struct mbuf **mp; 896 int *offp; 897{ 898 struct mbuf *m = *mp; 899 int off = *offp, hbhlen; 900 struct ip6_hbh *hbh; 901 u_int8_t *opt; 902 903 /* validation of the length of the header */ 904#ifndef PULLDOWN_TEST 905 IP6_EXTHDR_CHECK(m, off, sizeof(*hbh), -1); 906 hbh = (struct ip6_hbh *)(mtod(m, caddr_t) + off); 907 hbhlen = (hbh->ip6h_len + 1) << 3; 908 909 IP6_EXTHDR_CHECK(m, off, hbhlen, -1); 910 hbh = (struct ip6_hbh *)(mtod(m, caddr_t) + off); 911#else 912 IP6_EXTHDR_GET(hbh, struct ip6_hbh *, m, 913 sizeof(struct ip6_hdr), sizeof(struct ip6_hbh)); 914 if (hbh == NULL) { 915 ip6stat.ip6s_tooshort++; 916 return -1; 917 } 918 hbhlen = (hbh->ip6h_len + 1) << 3; 919 IP6_EXTHDR_GET(hbh, struct ip6_hbh *, m, sizeof(struct ip6_hdr), 920 hbhlen); 921 if (hbh == NULL) { 922 ip6stat.ip6s_tooshort++; 923 return -1; 924 } 925#endif 926 off += hbhlen; 927 hbhlen -= sizeof(struct ip6_hbh); 928 opt = (u_int8_t *)hbh + sizeof(struct ip6_hbh); 929 930 if (ip6_process_hopopts(m, (u_int8_t *)hbh + sizeof(struct ip6_hbh), 931 hbhlen, rtalertp, plenp) < 0) 932 return(-1); 933 934 *offp = off; 935 *mp = m; 936 return(0); 937} 938 939/* 940 * Search header for all Hop-by-hop options and process each option. 941 * This function is separate from ip6_hopopts_input() in order to 942 * handle a case where the sending node itself process its hop-by-hop 943 * options header. In such a case, the function is called from ip6_output(). 944 * 945 * The function assumes that hbh header is located right after the IPv6 header 946 * (RFC2460 p7), opthead is pointer into data content in m, and opthead to 947 * opthead + hbhlen is located in continuous memory region. 948 */ 949int 950ip6_process_hopopts(m, opthead, hbhlen, rtalertp, plenp) 951 struct mbuf *m; 952 u_int8_t *opthead; 953 int hbhlen; 954 u_int32_t *rtalertp; 955 u_int32_t *plenp; 956{ 957 struct ip6_hdr *ip6; 958 int optlen = 0; 959 u_int8_t *opt = opthead; 960 u_int16_t rtalert_val; 961 u_int32_t jumboplen; 962 const int erroff = sizeof(struct ip6_hdr) + sizeof(struct ip6_hbh); 963 964 for (; hbhlen > 0; hbhlen -= optlen, opt += optlen) { 965 switch (*opt) { 966 case IP6OPT_PAD1: 967 optlen = 1; 968 break; 969 case IP6OPT_PADN: 970 if (hbhlen < IP6OPT_MINLEN) { 971 ip6stat.ip6s_toosmall++; 972 goto bad; 973 } 974 optlen = *(opt + 1) + 2; 975 break; 976 case IP6OPT_RTALERT: 977 /* XXX may need check for alignment */ 978 if (hbhlen < IP6OPT_RTALERT_LEN) { 979 ip6stat.ip6s_toosmall++; 980 goto bad; 981 } 982 if (*(opt + 1) != IP6OPT_RTALERT_LEN - 2) { 983 /* XXX stat */ 984 icmp6_error(m, ICMP6_PARAM_PROB, 985 ICMP6_PARAMPROB_HEADER, 986 erroff + opt + 1 - opthead); 987 return(-1); 988 } 989 optlen = IP6OPT_RTALERT_LEN; 990 bcopy((caddr_t)(opt + 2), (caddr_t)&rtalert_val, 2); 991 *rtalertp = ntohs(rtalert_val); 992 break; 993 case IP6OPT_JUMBO: 994 /* XXX may need check for alignment */ 995 if (hbhlen < IP6OPT_JUMBO_LEN) { 996 ip6stat.ip6s_toosmall++; 997 goto bad; 998 } 999 if (*(opt + 1) != IP6OPT_JUMBO_LEN - 2) { 1000 /* XXX stat */ 1001 icmp6_error(m, ICMP6_PARAM_PROB, 1002 ICMP6_PARAMPROB_HEADER, 1003 erroff + opt + 1 - opthead); 1004 return(-1); 1005 } 1006 optlen = IP6OPT_JUMBO_LEN; 1007 1008 /* 1009 * IPv6 packets that have non 0 payload length 1010 * must not contain a jumbo payload option. 1011 */ 1012 ip6 = mtod(m, struct ip6_hdr *); 1013 if (ip6->ip6_plen) { 1014 ip6stat.ip6s_badoptions++; 1015 icmp6_error(m, ICMP6_PARAM_PROB, 1016 ICMP6_PARAMPROB_HEADER, 1017 erroff + opt - opthead); 1018 return(-1); 1019 } 1020 1021 /* 1022 * We may see jumbolen in unaligned location, so 1023 * we'd need to perform bcopy(). 1024 */ 1025 bcopy(opt + 2, &jumboplen, sizeof(jumboplen)); 1026 jumboplen = (u_int32_t)htonl(jumboplen); 1027 1028#if 1 1029 /* 1030 * if there are multiple jumbo payload options, 1031 * *plenp will be non-zero and the packet will be 1032 * rejected. 1033 * the behavior may need some debate in ipngwg - 1034 * multiple options does not make sense, however, 1035 * there's no explicit mention in specification. 1036 */ 1037 if (*plenp != 0) { 1038 ip6stat.ip6s_badoptions++; 1039 icmp6_error(m, ICMP6_PARAM_PROB, 1040 ICMP6_PARAMPROB_HEADER, 1041 erroff + opt + 2 - opthead); 1042 return(-1); 1043 } 1044#endif 1045 1046 /* 1047 * jumbo payload length must be larger than 65535. 1048 */ 1049 if (jumboplen <= IPV6_MAXPACKET) { 1050 ip6stat.ip6s_badoptions++; 1051 icmp6_error(m, ICMP6_PARAM_PROB, 1052 ICMP6_PARAMPROB_HEADER, 1053 erroff + opt + 2 - opthead); 1054 return(-1); 1055 } 1056 *plenp = jumboplen; 1057 1058 break; 1059 default: /* unknown option */ 1060 if (hbhlen < IP6OPT_MINLEN) { 1061 ip6stat.ip6s_toosmall++; 1062 goto bad; 1063 } 1064 optlen = ip6_unknown_opt(opt, m, 1065 erroff + opt - opthead); 1066 if (optlen == -1) 1067 return(-1); 1068 optlen += 2; 1069 break; 1070 } 1071 } 1072 1073 return(0); 1074 1075 bad: 1076 m_freem(m); 1077 return(-1); 1078} 1079 1080/* 1081 * Unknown option processing. 1082 * The third argument `off' is the offset from the IPv6 header to the option, 1083 * which is necessary if the IPv6 header the and option header and IPv6 header 1084 * is not continuous in order to return an ICMPv6 error. 1085 */ 1086int 1087ip6_unknown_opt(optp, m, off) 1088 u_int8_t *optp; 1089 struct mbuf *m; 1090 int off; 1091{ 1092 struct ip6_hdr *ip6; 1093 1094 switch (IP6OPT_TYPE(*optp)) { 1095 case IP6OPT_TYPE_SKIP: /* ignore the option */ 1096 return((int)*(optp + 1)); 1097 case IP6OPT_TYPE_DISCARD: /* silently discard */ 1098 m_freem(m); 1099 return(-1); 1100 case IP6OPT_TYPE_FORCEICMP: /* send ICMP even if multicasted */ 1101 ip6stat.ip6s_badoptions++; 1102 icmp6_error(m, ICMP6_PARAM_PROB, ICMP6_PARAMPROB_OPTION, off); 1103 return(-1); 1104 case IP6OPT_TYPE_ICMP: /* send ICMP if not multicasted */ 1105 ip6stat.ip6s_badoptions++; 1106 ip6 = mtod(m, struct ip6_hdr *); 1107 if (IN6_IS_ADDR_MULTICAST(&ip6->ip6_dst) || 1108 (m->m_flags & (M_BCAST|M_MCAST))) 1109 m_freem(m); 1110 else 1111 icmp6_error(m, ICMP6_PARAM_PROB, 1112 ICMP6_PARAMPROB_OPTION, off); 1113 return(-1); 1114 } 1115 1116 m_freem(m); /* XXX: NOTREACHED */ 1117 return(-1); 1118} 1119 1120/* 1121 * Create the "control" list for this pcb. 1122 * The function will not modify mbuf chain at all. 1123 * 1124 * with KAME mbuf chain restriction: 1125 * The routine will be called from upper layer handlers like tcp6_input(). 1126 * Thus the routine assumes that the caller (tcp6_input) have already 1127 * called IP6_EXTHDR_CHECK() and all the extension headers are located in the 1128 * very first mbuf on the mbuf chain. 1129 */ 1130void 1131ip6_savecontrol(in6p, mp, ip6, m) 1132 struct inpcb *in6p; 1133 struct mbuf **mp; 1134 struct ip6_hdr *ip6; 1135 struct mbuf *m; 1136{ 1137#if __FreeBSD__ >= 5 1138 struct thread *td = curthread; /* XXX */ 1139#else 1140 struct proc *td = curproc; /* XXX */ 1141#endif 1142 int privileged = 0; 1143 int rthdr_exist = 0; 1144 1145 1146 if (td && !suser(td)) 1147 privileged++; 1148 1149#ifdef SO_TIMESTAMP 1150 SOCK_LOCK(in6p->in6p_socket); 1151 if ((in6p->in6p_socket->so_options & SO_TIMESTAMP) != 0) { 1152 struct timeval tv; 1153 1154 SOCK_UNLOCK(in6p->in6p_socket); 1155 microtime(&tv); 1156 *mp = sbcreatecontrol((caddr_t) &tv, sizeof(tv), 1157 SCM_TIMESTAMP, SOL_SOCKET); 1158 if (*mp) { 1159 mp = &(*mp)->m_next; 1160 } 1161 } else 1162 SOCK_UNLOCK(in6p->in6p_socket); 1163#endif 1164 1165 /* RFC 2292 sec. 5 */ 1166 if ((in6p->in6p_flags & IN6P_PKTINFO) != 0) { 1167 struct in6_pktinfo pi6; 1168 bcopy(&ip6->ip6_dst, &pi6.ipi6_addr, sizeof(struct in6_addr)); 1169 if (IN6_IS_SCOPE_LINKLOCAL(&pi6.ipi6_addr)) 1170 pi6.ipi6_addr.s6_addr16[1] = 0; 1171 pi6.ipi6_ifindex = (m && m->m_pkthdr.rcvif) 1172 ? m->m_pkthdr.rcvif->if_index 1173 : 0; 1174 *mp = sbcreatecontrol((caddr_t) &pi6, 1175 sizeof(struct in6_pktinfo), IPV6_PKTINFO, 1176 IPPROTO_IPV6); 1177 if (*mp) 1178 mp = &(*mp)->m_next; 1179 } 1180 1181 if ((in6p->in6p_flags & IN6P_HOPLIMIT) != 0) { 1182 int hlim = ip6->ip6_hlim & 0xff; 1183 *mp = sbcreatecontrol((caddr_t) &hlim, 1184 sizeof(int), IPV6_HOPLIMIT, IPPROTO_IPV6); 1185 if (*mp) 1186 mp = &(*mp)->m_next; 1187 } 1188 1189 /* 1190 * IPV6_HOPOPTS socket option. We require super-user privilege 1191 * for the option, but it might be too strict, since there might 1192 * be some hop-by-hop options which can be returned to normal user. 1193 * See RFC 2292 section 6. 1194 */ 1195 if ((in6p->in6p_flags & IN6P_HOPOPTS) != 0 && privileged) { 1196 /* 1197 * Check if a hop-by-hop options header is contatined in the 1198 * received packet, and if so, store the options as ancillary 1199 * data. Note that a hop-by-hop options header must be 1200 * just after the IPv6 header, which fact is assured through 1201 * the IPv6 input processing. 1202 */ 1203 struct ip6_hdr *ip6 = mtod(m, struct ip6_hdr *); 1204 if (ip6->ip6_nxt == IPPROTO_HOPOPTS) { 1205 struct ip6_hbh *hbh; 1206 int hbhlen = 0; 1207#ifdef PULLDOWN_TEST 1208 struct mbuf *ext; 1209#endif 1210 1211#ifndef PULLDOWN_TEST 1212 hbh = (struct ip6_hbh *)(ip6 + 1); 1213 hbhlen = (hbh->ip6h_len + 1) << 3; 1214#else 1215 ext = ip6_pullexthdr(m, sizeof(struct ip6_hdr), 1216 ip6->ip6_nxt); 1217 if (ext == NULL) { 1218 ip6stat.ip6s_tooshort++; 1219 return; 1220 } 1221 hbh = mtod(ext, struct ip6_hbh *); 1222 hbhlen = (hbh->ip6h_len + 1) << 3; 1223 if (hbhlen != ext->m_len) { 1224 m_freem(ext); 1225 ip6stat.ip6s_tooshort++; 1226 return; 1227 } 1228#endif 1229 1230 /* 1231 * XXX: We copy whole the header even if a jumbo 1232 * payload option is included, which option is to 1233 * be removed before returning in the RFC 2292. 1234 * Note: this constraint is removed in 2292bis. 1235 */ 1236 *mp = sbcreatecontrol((caddr_t)hbh, hbhlen, 1237 IPV6_HOPOPTS, IPPROTO_IPV6); 1238 if (*mp) 1239 mp = &(*mp)->m_next; 1240#ifdef PULLDOWN_TEST 1241 m_freem(ext); 1242#endif 1243 } 1244 } 1245 1246 /* IPV6_DSTOPTS and IPV6_RTHDR socket options */ 1247 if ((in6p->in6p_flags & (IN6P_DSTOPTS | IN6P_RTHDRDSTOPTS)) != 0) { 1248 int proto, off, nxt; 1249 1250 /* 1251 * go through the header chain to see if a routing header is 1252 * contained in the packet. We need this information to store 1253 * destination options headers (if any) properly. 1254 * XXX: performance issue. We should record this info when 1255 * processing extension headers in incoming routine. 1256 * (todo) use m_aux? 1257 */ 1258 proto = IPPROTO_IPV6; 1259 off = 0; 1260 nxt = -1; 1261 while (1) { 1262 int newoff; 1263 1264 newoff = ip6_nexthdr(m, off, proto, &nxt); 1265 if (newoff < 0) 1266 break; 1267 if (newoff < off) /* invalid, check for safety */ 1268 break; 1269 if ((proto = nxt) == IPPROTO_ROUTING) { 1270 rthdr_exist = 1; 1271 break; 1272 } 1273 off = newoff; 1274 } 1275 } 1276 1277 if ((in6p->in6p_flags & 1278 (IN6P_RTHDR | IN6P_DSTOPTS | IN6P_RTHDRDSTOPTS)) != 0) { 1279 struct ip6_hdr *ip6 = mtod(m, struct ip6_hdr *); 1280 int nxt = ip6->ip6_nxt, off = sizeof(struct ip6_hdr); 1281 1282 /* 1283 * Search for destination options headers or routing 1284 * header(s) through the header chain, and stores each 1285 * header as ancillary data. 1286 * Note that the order of the headers remains in 1287 * the chain of ancillary data. 1288 */ 1289 while (1) { /* is explicit loop prevention necessary? */ 1290 struct ip6_ext *ip6e = NULL; 1291 int elen; 1292#ifdef PULLDOWN_TEST 1293 struct mbuf *ext = NULL; 1294#endif 1295 1296 /* 1297 * if it is not an extension header, don't try to 1298 * pull it from the chain. 1299 */ 1300 switch (nxt) { 1301 case IPPROTO_DSTOPTS: 1302 case IPPROTO_ROUTING: 1303 case IPPROTO_HOPOPTS: 1304 case IPPROTO_AH: /* is it possible? */ 1305 break; 1306 default: 1307 goto loopend; 1308 } 1309 1310#ifndef PULLDOWN_TEST 1311 if (off + sizeof(*ip6e) > m->m_len) 1312 goto loopend; 1313 ip6e = (struct ip6_ext *)(mtod(m, caddr_t) + off); 1314 if (nxt == IPPROTO_AH) 1315 elen = (ip6e->ip6e_len + 2) << 2; 1316 else 1317 elen = (ip6e->ip6e_len + 1) << 3; 1318 if (off + elen > m->m_len) 1319 goto loopend; 1320#else 1321 ext = ip6_pullexthdr(m, off, nxt); 1322 if (ext == NULL) { 1323 ip6stat.ip6s_tooshort++; 1324 return; 1325 } 1326 ip6e = mtod(ext, struct ip6_ext *); 1327 if (nxt == IPPROTO_AH) 1328 elen = (ip6e->ip6e_len + 2) << 2; 1329 else 1330 elen = (ip6e->ip6e_len + 1) << 3; 1331 if (elen != ext->m_len) { 1332 m_freem(ext); 1333 ip6stat.ip6s_tooshort++; 1334 return; 1335 } 1336#endif 1337 1338 switch (nxt) { 1339 case IPPROTO_DSTOPTS: 1340 if ((in6p->in6p_flags & IN6P_DSTOPTS) == 0) 1341 break; 1342 1343 /* 1344 * We also require super-user privilege for 1345 * the option. 1346 * See the comments on IN6_HOPOPTS. 1347 */ 1348 if (!privileged) 1349 break; 1350 1351 *mp = sbcreatecontrol((caddr_t)ip6e, elen, 1352 IPV6_DSTOPTS, 1353 IPPROTO_IPV6); 1354 if (*mp) 1355 mp = &(*mp)->m_next; 1356 break; 1357 case IPPROTO_ROUTING: 1358 if (!in6p->in6p_flags & IN6P_RTHDR) 1359 break; 1360 1361 *mp = sbcreatecontrol((caddr_t)ip6e, elen, 1362 IPV6_RTHDR, 1363 IPPROTO_IPV6); 1364 if (*mp) 1365 mp = &(*mp)->m_next; 1366 break; 1367 case IPPROTO_HOPOPTS: 1368 case IPPROTO_AH: /* is it possible? */ 1369 break; 1370 1371 default: 1372 /* 1373 * other cases have been filtered in the above. 1374 * none will visit this case. here we supply 1375 * the code just in case (nxt overwritten or 1376 * other cases). 1377 */ 1378#ifdef PULLDOWN_TEST 1379 m_freem(ext); 1380#endif 1381 goto loopend; 1382 1383 } 1384 1385 /* proceed with the next header. */ 1386 off += elen; 1387 nxt = ip6e->ip6e_nxt; 1388 ip6e = NULL; 1389#ifdef PULLDOWN_TEST 1390 m_freem(ext); 1391 ext = NULL; 1392#endif 1393 } 1394 loopend: 1395 ; 1396 } 1397 1398} 1399 1400#ifdef PULLDOWN_TEST 1401/* 1402 * pull single extension header from mbuf chain. returns single mbuf that 1403 * contains the result, or NULL on error. 1404 */ 1405static struct mbuf * 1406ip6_pullexthdr(m, off, nxt) 1407 struct mbuf *m; 1408 size_t off; 1409 int nxt; 1410{ 1411 struct ip6_ext ip6e; 1412 size_t elen; 1413 struct mbuf *n; 1414 1415#ifdef DIAGNOSTIC 1416 switch (nxt) { 1417 case IPPROTO_DSTOPTS: 1418 case IPPROTO_ROUTING: 1419 case IPPROTO_HOPOPTS: 1420 case IPPROTO_AH: /* is it possible? */ 1421 break; 1422 default: 1423 printf("ip6_pullexthdr: invalid nxt=%d\n", nxt); 1424 } 1425#endif 1426 1427 m_copydata(m, off, sizeof(ip6e), (caddr_t)&ip6e); 1428 if (nxt == IPPROTO_AH) 1429 elen = (ip6e.ip6e_len + 2) << 2; 1430 else 1431 elen = (ip6e.ip6e_len + 1) << 3; 1432 1433 MGET(n, M_DONTWAIT, MT_DATA); 1434 if (n && elen >= MLEN) { 1435 MCLGET(n, M_DONTWAIT); 1436 if ((n->m_flags & M_EXT) == 0) { 1437 m_free(n); 1438 n = NULL; 1439 } 1440 } 1441 if (!n) 1442 return NULL; 1443 1444 n->m_len = 0; 1445 if (elen >= M_TRAILINGSPACE(n)) { 1446 m_free(n); 1447 return NULL; 1448 } 1449 1450 m_copydata(m, off, elen, mtod(n, caddr_t)); 1451 n->m_len = elen; 1452 return n; 1453} 1454#endif 1455 1456/* 1457 * Get pointer to the previous header followed by the header 1458 * currently processed. 1459 * XXX: This function supposes that 1460 * M includes all headers, 1461 * the next header field and the header length field of each header 1462 * are valid, and 1463 * the sum of each header length equals to OFF. 1464 * Because of these assumptions, this function must be called very 1465 * carefully. Moreover, it will not be used in the near future when 1466 * we develop `neater' mechanism to process extension headers. 1467 */ 1468char * 1469ip6_get_prevhdr(m, off) 1470 struct mbuf *m; 1471 int off; 1472{ 1473 struct ip6_hdr *ip6 = mtod(m, struct ip6_hdr *); 1474 1475 if (off == sizeof(struct ip6_hdr)) 1476 return(&ip6->ip6_nxt); 1477 else { 1478 int len, nxt; 1479 struct ip6_ext *ip6e = NULL; 1480 1481 nxt = ip6->ip6_nxt; 1482 len = sizeof(struct ip6_hdr); 1483 while (len < off) { 1484 ip6e = (struct ip6_ext *)(mtod(m, caddr_t) + len); 1485 1486 switch (nxt) { 1487 case IPPROTO_FRAGMENT: 1488 len += sizeof(struct ip6_frag); 1489 break; 1490 case IPPROTO_AH: 1491 len += (ip6e->ip6e_len + 2) << 2; 1492 break; 1493 default: 1494 len += (ip6e->ip6e_len + 1) << 3; 1495 break; 1496 } 1497 nxt = ip6e->ip6e_nxt; 1498 } 1499 if (ip6e) 1500 return(&ip6e->ip6e_nxt); 1501 else 1502 return NULL; 1503 } 1504} 1505 1506/* 1507 * get next header offset. m will be retained. 1508 */ 1509int 1510ip6_nexthdr(m, off, proto, nxtp) 1511 struct mbuf *m; 1512 int off; 1513 int proto; 1514 int *nxtp; 1515{ 1516 struct ip6_hdr ip6; 1517 struct ip6_ext ip6e; 1518 struct ip6_frag fh; 1519 1520 /* just in case */ 1521 if (m == NULL) 1522 panic("ip6_nexthdr: m == NULL"); 1523 if ((m->m_flags & M_PKTHDR) == 0 || m->m_pkthdr.len < off) 1524 return -1; 1525 1526 switch (proto) { 1527 case IPPROTO_IPV6: 1528 if (m->m_pkthdr.len < off + sizeof(ip6)) 1529 return -1; 1530 m_copydata(m, off, sizeof(ip6), (caddr_t)&ip6); 1531 if (nxtp) 1532 *nxtp = ip6.ip6_nxt; 1533 off += sizeof(ip6); 1534 return off; 1535 1536 case IPPROTO_FRAGMENT: 1537 /* 1538 * terminate parsing if it is not the first fragment, 1539 * it does not make sense to parse through it. 1540 */ 1541 if (m->m_pkthdr.len < off + sizeof(fh)) 1542 return -1; 1543 m_copydata(m, off, sizeof(fh), (caddr_t)&fh); 1544 if ((ntohs(fh.ip6f_offlg) & IP6F_OFF_MASK) != 0) 1545 return -1; 1546 if (nxtp) 1547 *nxtp = fh.ip6f_nxt; 1548 off += sizeof(struct ip6_frag); 1549 return off; 1550 1551 case IPPROTO_AH: 1552 if (m->m_pkthdr.len < off + sizeof(ip6e)) 1553 return -1; 1554 m_copydata(m, off, sizeof(ip6e), (caddr_t)&ip6e); 1555 if (nxtp) 1556 *nxtp = ip6e.ip6e_nxt; 1557 off += (ip6e.ip6e_len + 2) << 2; 1558 return off; 1559 1560 case IPPROTO_HOPOPTS: 1561 case IPPROTO_ROUTING: 1562 case IPPROTO_DSTOPTS: 1563 if (m->m_pkthdr.len < off + sizeof(ip6e)) 1564 return -1; 1565 m_copydata(m, off, sizeof(ip6e), (caddr_t)&ip6e); 1566 if (nxtp) 1567 *nxtp = ip6e.ip6e_nxt; 1568 off += (ip6e.ip6e_len + 1) << 3; 1569 return off; 1570 1571 case IPPROTO_NONE: 1572 case IPPROTO_ESP: 1573 case IPPROTO_IPCOMP: 1574 /* give up */ 1575 return -1; 1576 1577 default: 1578 return -1; 1579 } 1580 1581 return -1; 1582} 1583 1584/* 1585 * get offset for the last header in the chain. m will be kept untainted. 1586 */ 1587int 1588ip6_lasthdr(m, off, proto, nxtp) 1589 struct mbuf *m; 1590 int off; 1591 int proto; 1592 int *nxtp; 1593{ 1594 int newoff; 1595 int nxt; 1596 1597 if (!nxtp) { 1598 nxt = -1; 1599 nxtp = &nxt; 1600 } 1601 while (1) { 1602 newoff = ip6_nexthdr(m, off, proto, nxtp); 1603 if (newoff < 0) 1604 return off; 1605 else if (newoff < off) 1606 return -1; /* invalid */ 1607 else if (newoff == off) 1608 return newoff; 1609 1610 off = newoff; 1611 proto = *nxtp; 1612 } 1613} 1614 1615struct mbuf * 1616ip6_addaux(m) 1617 struct mbuf *m; 1618{ 1619 struct mbuf *n; 1620 1621#ifdef DIAGNOSTIC 1622 if (sizeof(struct ip6aux) > MHLEN) 1623 panic("assumption failed on sizeof(ip6aux)"); 1624#endif 1625 n = m_aux_find(m, AF_INET6, -1); 1626 if (n) { 1627 if (n->m_len < sizeof(struct ip6aux)) { 1628 printf("conflicting use of ip6aux"); 1629 return NULL; 1630 } 1631 } else { 1632 n = m_aux_add(m, AF_INET6, -1); 1633 n->m_len = sizeof(struct ip6aux); 1634 bzero(mtod(n, caddr_t), n->m_len); 1635 } 1636 return n; 1637} 1638 1639struct mbuf * 1640ip6_findaux(m) 1641 struct mbuf *m; 1642{ 1643 struct mbuf *n; 1644 1645 n = m_aux_find(m, AF_INET6, -1); 1646 if (n && n->m_len < sizeof(struct ip6aux)) { 1647 printf("conflicting use of ip6aux"); 1648 n = NULL; 1649 } 1650 return n; 1651} 1652 1653void 1654ip6_delaux(m) 1655 struct mbuf *m; 1656{ 1657 struct mbuf *n; 1658 1659 n = m_aux_find(m, AF_INET6, -1); 1660 if (n) 1661 m_aux_delete(m, n); 1662} 1663 1664/* 1665 * System control for IP6 1666 */ 1667 1668u_char inet6ctlerrmap[PRC_NCMDS] = { 1669 0, 0, 0, 0, 1670 0, EMSGSIZE, EHOSTDOWN, EHOSTUNREACH, 1671 EHOSTUNREACH, EHOSTUNREACH, ECONNREFUSED, ECONNREFUSED, 1672 EMSGSIZE, EHOSTUNREACH, 0, 0, 1673 0, 0, 0, 0, 1674 ENOPROTOOPT 1675}; 1676