ip6_input.c revision 120648
1/* $FreeBSD: head/sys/netinet6/ip6_input.c 120648 2003-10-01 21:10:02Z 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 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#ifdef PFIL_HOOKS 94#include <net/pfil.h> 95#endif 96 97#include <netinet/in.h> 98#include <netinet/in_systm.h> 99#ifdef INET 100#include <netinet/ip.h> 101#include <netinet/ip_icmp.h> 102#endif /* INET */ 103#include <netinet/ip6.h> 104#include <netinet6/in6_var.h> 105#include <netinet6/ip6_var.h> 106#include <netinet/in_pcb.h> 107#include <netinet/icmp6.h> 108#include <netinet6/in6_ifattach.h> 109#include <netinet6/nd6.h> 110#include <netinet6/in6_prefix.h> 111 112#ifdef IPSEC 113#include <netinet6/ipsec.h> 114#ifdef INET6 115#include <netinet6/ipsec6.h> 116#endif 117#endif 118 119#ifdef FAST_IPSEC 120#include <netipsec/ipsec.h> 121#include <netipsec/ipsec6.h> 122#define IPSEC 123#endif /* FAST_IPSEC */ 124 125#include <netinet6/ip6_fw.h> 126 127#include <netinet6/ip6protosw.h> 128 129#include <net/net_osdep.h> 130 131extern struct domain inet6domain; 132 133u_char ip6_protox[IPPROTO_MAX]; 134static struct ifqueue ip6intrq; 135static int ip6qmaxlen = IFQ_MAXLEN; 136struct in6_ifaddr *in6_ifaddr; 137 138extern struct callout in6_tmpaddrtimer_ch; 139 140int ip6_forward_srcrt; /* XXX */ 141int ip6_sourcecheck; /* XXX */ 142int ip6_sourcecheck_interval; /* XXX */ 143 144int ip6_ours_check_algorithm; 145 146#ifdef PFIL_HOOKS 147struct pfil_head inet6_pfil_hook; 148#endif 149 150/* firewall hooks */ 151ip6_fw_chk_t *ip6_fw_chk_ptr; 152ip6_fw_ctl_t *ip6_fw_ctl_ptr; 153int ip6_fw_enable = 1; 154 155struct ip6stat ip6stat; 156 157static void ip6_init2 __P((void *)); 158static struct ip6aux *ip6_setdstifaddr __P((struct mbuf *, struct in6_ifaddr *)); 159static int ip6_hopopts_input __P((u_int32_t *, u_int32_t *, struct mbuf **, int *)); 160#ifdef PULLDOWN_TEST 161static struct mbuf *ip6_pullexthdr __P((struct mbuf *, size_t, int)); 162#endif 163 164 165/* 166 * IP6 initialization: fill in IP6 protocol switch table. 167 * All protocols not implemented in kernel go to raw IP6 protocol handler. 168 */ 169void 170ip6_init() 171{ 172 struct ip6protosw *pr; 173 int i; 174 175#ifdef DIAGNOSTIC 176 if (sizeof(struct protosw) != sizeof(struct ip6protosw)) 177 panic("sizeof(protosw) != sizeof(ip6protosw)"); 178#endif 179 pr = (struct ip6protosw *)pffindproto(PF_INET6, IPPROTO_RAW, SOCK_RAW); 180 if (pr == 0) 181 panic("ip6_init"); 182 for (i = 0; i < IPPROTO_MAX; i++) 183 ip6_protox[i] = pr - inet6sw; 184 for (pr = (struct ip6protosw *)inet6domain.dom_protosw; 185 pr < (struct ip6protosw *)inet6domain.dom_protoswNPROTOSW; pr++) 186 if (pr->pr_domain->dom_family == PF_INET6 && 187 pr->pr_protocol && pr->pr_protocol != IPPROTO_RAW) 188 ip6_protox[pr->pr_protocol] = pr - inet6sw; 189#ifdef PFIL_HOOKS 190 inet6_pfil_hook.ph_type = PFIL_TYPE_AF; 191 inet6_pfil_hook.ph_af = AF_INET6; 192 if ((i = pfil_head_register(&inet6_pfil_hook)) != 0) 193 printf("%s: WARNING: unable to register pfil hook, " 194 "error %d\n", __func__, i); 195#endif /* PFIL_HOOKS */ 196 ip6intrq.ifq_maxlen = ip6qmaxlen; 197 mtx_init(&ip6intrq.ifq_mtx, "ip6_inq", NULL, MTX_DEF); 198 netisr_register(NETISR_IPV6, ip6_input, &ip6intrq); 199 nd6_init(); 200 frag6_init(); 201 /* 202 * in many cases, random() here does NOT return random number 203 * as initialization during bootstrap time occur in fixed order. 204 */ 205 ip6_flow_seq = arc4random(); 206 ip6_desync_factor = arc4random() % MAX_TEMP_DESYNC_FACTOR; 207} 208 209static void 210ip6_init2(dummy) 211 void *dummy; 212{ 213 214 /* 215 * to route local address of p2p link to loopback, 216 * assign loopback address first. 217 */ 218 in6_ifattach(&loif[0], NULL); 219 220 /* nd6_timer_init */ 221 callout_init(&nd6_timer_ch, 0); 222 callout_reset(&nd6_timer_ch, hz, nd6_timer, NULL); 223 224 /* router renumbering prefix list maintenance */ 225 callout_init(&in6_rr_timer_ch, 0); 226 callout_reset(&in6_rr_timer_ch, hz, in6_rr_timer, NULL); 227 228 /* timer for regeneranation of temporary addresses randomize ID */ 229 callout_reset(&in6_tmpaddrtimer_ch, 230 (ip6_temp_preferred_lifetime - ip6_desync_factor - 231 ip6_temp_regen_advance) * hz, 232 in6_tmpaddrtimer, NULL); 233} 234 235/* cheat */ 236/* This must be after route_init(), which is now SI_ORDER_THIRD */ 237SYSINIT(netinet6init2, SI_SUB_PROTO_DOMAIN, SI_ORDER_MIDDLE, ip6_init2, NULL); 238 239extern struct route_in6 ip6_forward_rt; 240 241void 242ip6_input(m) 243 struct mbuf *m; 244{ 245 struct ip6_hdr *ip6; 246 int off = sizeof(struct ip6_hdr), nest; 247 u_int32_t plen; 248 u_int32_t rtalert = ~0; 249 int nxt, ours = 0; 250 struct ifnet *deliverifp = NULL; 251 252#ifdef IPSEC 253 /* 254 * should the inner packet be considered authentic? 255 * see comment in ah4_input(). 256 */ 257 if (m) { 258 m->m_flags &= ~M_AUTHIPHDR; 259 m->m_flags &= ~M_AUTHIPDGM; 260 } 261#endif 262 263 /* 264 * make sure we don't have onion peering information into m_aux. 265 */ 266 ip6_delaux(m); 267 268 /* 269 * mbuf statistics 270 */ 271 if (m->m_flags & M_EXT) { 272 if (m->m_next) 273 ip6stat.ip6s_mext2m++; 274 else 275 ip6stat.ip6s_mext1++; 276 } else { 277#define M2MMAX (sizeof(ip6stat.ip6s_m2m)/sizeof(ip6stat.ip6s_m2m[0])) 278 if (m->m_next) { 279 if (m->m_flags & M_LOOP) { 280 ip6stat.ip6s_m2m[loif[0].if_index]++; /* XXX */ 281 } else if (m->m_pkthdr.rcvif->if_index < M2MMAX) 282 ip6stat.ip6s_m2m[m->m_pkthdr.rcvif->if_index]++; 283 else 284 ip6stat.ip6s_m2m[0]++; 285 } else 286 ip6stat.ip6s_m1++; 287#undef M2MMAX 288 } 289 290 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_receive); 291 ip6stat.ip6s_total++; 292 293#ifndef PULLDOWN_TEST 294 /* 295 * L2 bridge code and some other code can return mbuf chain 296 * that does not conform to KAME requirement. too bad. 297 * XXX: fails to join if interface MTU > MCLBYTES. jumbogram? 298 */ 299 if (m && m->m_next != NULL && m->m_pkthdr.len < MCLBYTES) { 300 struct mbuf *n; 301 302 MGETHDR(n, M_DONTWAIT, MT_HEADER); 303 if (n) 304 M_MOVE_PKTHDR(n, m); 305 if (n && n->m_pkthdr.len > MHLEN) { 306 MCLGET(n, M_DONTWAIT); 307 if ((n->m_flags & M_EXT) == 0) { 308 m_freem(n); 309 n = NULL; 310 } 311 } 312 if (n == NULL) { 313 m_freem(m); 314 return; /*ENOBUFS*/ 315 } 316 317 m_copydata(m, 0, n->m_pkthdr.len, mtod(n, caddr_t)); 318 n->m_len = n->m_pkthdr.len; 319 m_freem(m); 320 m = n; 321 } 322 IP6_EXTHDR_CHECK(m, 0, sizeof(struct ip6_hdr), /*nothing*/); 323#endif 324 325 if (m->m_len < sizeof(struct ip6_hdr)) { 326 struct ifnet *inifp; 327 inifp = m->m_pkthdr.rcvif; 328 if ((m = m_pullup(m, sizeof(struct ip6_hdr))) == 0) { 329 ip6stat.ip6s_toosmall++; 330 in6_ifstat_inc(inifp, ifs6_in_hdrerr); 331 return; 332 } 333 } 334 335 ip6 = mtod(m, struct ip6_hdr *); 336 337 if ((ip6->ip6_vfc & IPV6_VERSION_MASK) != IPV6_VERSION) { 338 ip6stat.ip6s_badvers++; 339 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_hdrerr); 340 goto bad; 341 } 342 343#ifdef PFIL_HOOKS 344 /* 345 * Run through list of hooks for input packets. 346 */ 347 if (pfil_run_hooks(&inet6_pfil_hook, &m, m->m_pkthdr.rcvif, PFIL_IN)) 348 return; 349 if (m == NULL) /* consumed by filter */ 350 return; 351 ip6 = mtod(m, struct ip6_hdr *); 352#endif /* PFIL_HOOKS */ 353 354 ip6stat.ip6s_nxthist[ip6->ip6_nxt]++; 355 356 /* 357 * Check with the firewall... 358 */ 359 if (ip6_fw_enable && ip6_fw_chk_ptr) { 360 u_short port = 0; 361 /* If ipfw says divert, we have to just drop packet */ 362 /* use port as a dummy argument */ 363 if ((*ip6_fw_chk_ptr)(&ip6, NULL, &port, &m)) { 364 m_freem(m); 365 m = NULL; 366 } 367 if (!m) 368 return; 369 } 370 371 /* 372 * Check against address spoofing/corruption. 373 */ 374 if (IN6_IS_ADDR_MULTICAST(&ip6->ip6_src) || 375 IN6_IS_ADDR_UNSPECIFIED(&ip6->ip6_dst)) { 376 /* 377 * XXX: "badscope" is not very suitable for a multicast source. 378 */ 379 ip6stat.ip6s_badscope++; 380 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_addrerr); 381 goto bad; 382 } 383 if ((IN6_IS_ADDR_LOOPBACK(&ip6->ip6_src) || 384 IN6_IS_ADDR_LOOPBACK(&ip6->ip6_dst)) && 385 (m->m_pkthdr.rcvif->if_flags & IFF_LOOPBACK) == 0) { 386 ip6stat.ip6s_badscope++; 387 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_addrerr); 388 goto bad; 389 } 390 391 /* 392 * The following check is not documented in specs. A malicious 393 * party may be able to use IPv4 mapped addr to confuse tcp/udp stack 394 * and bypass security checks (act as if it was from 127.0.0.1 by using 395 * IPv6 src ::ffff:127.0.0.1). Be cautious. 396 * 397 * This check chokes if we are in an SIIT cloud. As none of BSDs 398 * support IPv4-less kernel compilation, we cannot support SIIT 399 * environment at all. So, it makes more sense for us to reject any 400 * malicious packets for non-SIIT environment, than try to do a 401 * partical support for SIIT environment. 402 */ 403 if (IN6_IS_ADDR_V4MAPPED(&ip6->ip6_src) || 404 IN6_IS_ADDR_V4MAPPED(&ip6->ip6_dst)) { 405 ip6stat.ip6s_badscope++; 406 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_addrerr); 407 goto bad; 408 } 409#if 0 410 /* 411 * Reject packets with IPv4 compatible addresses (auto tunnel). 412 * 413 * The code forbids auto tunnel relay case in RFC1933 (the check is 414 * stronger than RFC1933). We may want to re-enable it if mech-xx 415 * is revised to forbid relaying case. 416 */ 417 if (IN6_IS_ADDR_V4COMPAT(&ip6->ip6_src) || 418 IN6_IS_ADDR_V4COMPAT(&ip6->ip6_dst)) { 419 ip6stat.ip6s_badscope++; 420 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_addrerr); 421 goto bad; 422 } 423#endif 424 425 /* drop packets if interface ID portion is already filled */ 426 if ((m->m_pkthdr.rcvif->if_flags & IFF_LOOPBACK) == 0) { 427 if (IN6_IS_SCOPE_LINKLOCAL(&ip6->ip6_src) && 428 ip6->ip6_src.s6_addr16[1]) { 429 ip6stat.ip6s_badscope++; 430 goto bad; 431 } 432 if (IN6_IS_SCOPE_LINKLOCAL(&ip6->ip6_dst) && 433 ip6->ip6_dst.s6_addr16[1]) { 434 ip6stat.ip6s_badscope++; 435 goto bad; 436 } 437 } 438 439 if (IN6_IS_SCOPE_LINKLOCAL(&ip6->ip6_src)) 440 ip6->ip6_src.s6_addr16[1] 441 = htons(m->m_pkthdr.rcvif->if_index); 442 if (IN6_IS_SCOPE_LINKLOCAL(&ip6->ip6_dst)) 443 ip6->ip6_dst.s6_addr16[1] 444 = htons(m->m_pkthdr.rcvif->if_index); 445 446#if 0 /* this case seems to be unnecessary. (jinmei, 20010401) */ 447 /* 448 * We use rt->rt_ifp to determine if the address is ours or not. 449 * If rt_ifp is lo0, the address is ours. 450 * The problem here is, rt->rt_ifp for fe80::%lo0/64 is set to lo0, 451 * so any address under fe80::%lo0/64 will be mistakenly considered 452 * local. The special case is supplied to handle the case properly 453 * by actually looking at interface addresses 454 * (using in6ifa_ifpwithaddr). 455 */ 456 if ((m->m_pkthdr.rcvif->if_flags & IFF_LOOPBACK) != 0 && 457 IN6_IS_ADDR_LINKLOCAL(&ip6->ip6_dst)) { 458 if (!in6ifa_ifpwithaddr(m->m_pkthdr.rcvif, &ip6->ip6_dst)) { 459 icmp6_error(m, ICMP6_DST_UNREACH, 460 ICMP6_DST_UNREACH_ADDR, 0); 461 /* m is already freed */ 462 return; 463 } 464 465 ours = 1; 466 deliverifp = m->m_pkthdr.rcvif; 467 goto hbhcheck; 468 } 469#endif 470 471 /* 472 * Multicast check 473 */ 474 if (IN6_IS_ADDR_MULTICAST(&ip6->ip6_dst)) { 475 struct in6_multi *in6m = 0; 476 477 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_mcast); 478 /* 479 * See if we belong to the destination multicast group on the 480 * arrival interface. 481 */ 482 IN6_LOOKUP_MULTI(ip6->ip6_dst, m->m_pkthdr.rcvif, in6m); 483 if (in6m) 484 ours = 1; 485 else if (!ip6_mrouter) { 486 ip6stat.ip6s_notmember++; 487 ip6stat.ip6s_cantforward++; 488 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_discard); 489 goto bad; 490 } 491 deliverifp = m->m_pkthdr.rcvif; 492 goto hbhcheck; 493 } 494 495 /* 496 * Unicast check 497 */ 498 switch (ip6_ours_check_algorithm) { 499 default: 500 /* 501 * XXX: I intentionally broke our indentation rule here, 502 * since this switch-case is just for measurement and 503 * therefore should soon be removed. 504 */ 505 if (ip6_forward_rt.ro_rt != NULL && 506 (ip6_forward_rt.ro_rt->rt_flags & RTF_UP) != 0 && 507 IN6_ARE_ADDR_EQUAL(&ip6->ip6_dst, 508 &((struct sockaddr_in6 *)(&ip6_forward_rt.ro_dst))->sin6_addr)) 509 ip6stat.ip6s_forward_cachehit++; 510 else { 511 struct sockaddr_in6 *dst6; 512 513 if (ip6_forward_rt.ro_rt) { 514 /* route is down or destination is different */ 515 ip6stat.ip6s_forward_cachemiss++; 516 RTFREE(ip6_forward_rt.ro_rt); 517 ip6_forward_rt.ro_rt = 0; 518 } 519 520 bzero(&ip6_forward_rt.ro_dst, sizeof(struct sockaddr_in6)); 521 dst6 = (struct sockaddr_in6 *)&ip6_forward_rt.ro_dst; 522 dst6->sin6_len = sizeof(struct sockaddr_in6); 523 dst6->sin6_family = AF_INET6; 524 dst6->sin6_addr = ip6->ip6_dst; 525#ifdef SCOPEDROUTING 526 ip6_forward_rt.ro_dst.sin6_scope_id = 527 in6_addr2scopeid(m->m_pkthdr.rcvif, &ip6->ip6_dst); 528#endif 529 530 rtalloc_ign((struct route *)&ip6_forward_rt, RTF_PRCLONING); 531 } 532 533#define rt6_key(r) ((struct sockaddr_in6 *)((r)->rt_nodes->rn_key)) 534 535 /* 536 * Accept the packet if the forwarding interface to the destination 537 * according to the routing table is the loopback interface, 538 * unless the associated route has a gateway. 539 * Note that this approach causes to accept a packet if there is a 540 * route to the loopback interface for the destination of the packet. 541 * But we think it's even useful in some situations, e.g. when using 542 * a special daemon which wants to intercept the packet. 543 * 544 * XXX: some OSes automatically make a cloned route for the destination 545 * of an outgoing packet. If the outgoing interface of the packet 546 * is a loopback one, the kernel would consider the packet to be 547 * accepted, even if we have no such address assinged on the interface. 548 * We check the cloned flag of the route entry to reject such cases, 549 * assuming that route entries for our own addresses are not made by 550 * cloning (it should be true because in6_addloop explicitly installs 551 * the host route). However, we might have to do an explicit check 552 * while it would be less efficient. Or, should we rather install a 553 * reject route for such a case? 554 */ 555 if (ip6_forward_rt.ro_rt && 556 (ip6_forward_rt.ro_rt->rt_flags & 557 (RTF_HOST|RTF_GATEWAY)) == RTF_HOST && 558#ifdef RTF_WASCLONED 559 !(ip6_forward_rt.ro_rt->rt_flags & RTF_WASCLONED) && 560#endif 561#ifdef RTF_CLONED 562 !(ip6_forward_rt.ro_rt->rt_flags & RTF_CLONED) && 563#endif 564#if 0 565 /* 566 * The check below is redundant since the comparison of 567 * the destination and the key of the rtentry has 568 * already done through looking up the routing table. 569 */ 570 IN6_ARE_ADDR_EQUAL(&ip6->ip6_dst, 571 &rt6_key(ip6_forward_rt.ro_rt)->sin6_addr) 572#endif 573 ip6_forward_rt.ro_rt->rt_ifp->if_type == IFT_LOOP) { 574 struct in6_ifaddr *ia6 = 575 (struct in6_ifaddr *)ip6_forward_rt.ro_rt->rt_ifa; 576 577 /* 578 * record address information into m_aux. 579 */ 580 (void)ip6_setdstifaddr(m, ia6); 581 582 /* 583 * packets to a tentative, duplicated, or somehow invalid 584 * address must not be accepted. 585 */ 586 if (!(ia6->ia6_flags & IN6_IFF_NOTREADY)) { 587 /* this address is ready */ 588 ours = 1; 589 deliverifp = ia6->ia_ifp; /* correct? */ 590 /* Count the packet in the ip address stats */ 591 ia6->ia_ifa.if_ipackets++; 592 ia6->ia_ifa.if_ibytes += m->m_pkthdr.len; 593 goto hbhcheck; 594 } else { 595 /* address is not ready, so discard the packet. */ 596 nd6log((LOG_INFO, 597 "ip6_input: packet to an unready address %s->%s\n", 598 ip6_sprintf(&ip6->ip6_src), 599 ip6_sprintf(&ip6->ip6_dst))); 600 601 goto bad; 602 } 603 } 604 } /* XXX indentation (see above) */ 605 606 /* 607 * FAITH(Firewall Aided Internet Translator) 608 */ 609 if (ip6_keepfaith) { 610 if (ip6_forward_rt.ro_rt && ip6_forward_rt.ro_rt->rt_ifp 611 && ip6_forward_rt.ro_rt->rt_ifp->if_type == IFT_FAITH) { 612 /* XXX do we need more sanity checks? */ 613 ours = 1; 614 deliverifp = ip6_forward_rt.ro_rt->rt_ifp; /* faith */ 615 goto hbhcheck; 616 } 617 } 618 619 /* 620 * Now there is no reason to process the packet if it's not our own 621 * and we're not a router. 622 */ 623 if (!ip6_forwarding) { 624 ip6stat.ip6s_cantforward++; 625 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_discard); 626 goto bad; 627 } 628 629 hbhcheck: 630 /* 631 * record address information into m_aux, if we don't have one yet. 632 * note that we are unable to record it, if the address is not listed 633 * as our interface address (e.g. multicast addresses, addresses 634 * within FAITH prefixes and such). 635 */ 636 if (deliverifp && !ip6_getdstifaddr(m)) { 637 struct in6_ifaddr *ia6; 638 639 ia6 = in6_ifawithifp(deliverifp, &ip6->ip6_dst); 640 if (ia6) { 641 if (!ip6_setdstifaddr(m, ia6)) { 642 /* 643 * XXX maybe we should drop the packet here, 644 * as we could not provide enough information 645 * to the upper layers. 646 */ 647 } 648 } 649 } 650 651 /* 652 * Process Hop-by-Hop options header if it's contained. 653 * m may be modified in ip6_hopopts_input(). 654 * If a JumboPayload option is included, plen will also be modified. 655 */ 656 plen = (u_int32_t)ntohs(ip6->ip6_plen); 657 if (ip6->ip6_nxt == IPPROTO_HOPOPTS) { 658 struct ip6_hbh *hbh; 659 660 if (ip6_hopopts_input(&plen, &rtalert, &m, &off)) { 661#if 0 /*touches NULL pointer*/ 662 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_discard); 663#endif 664 return; /* m have already been freed */ 665 } 666 667 /* adjust pointer */ 668 ip6 = mtod(m, struct ip6_hdr *); 669 670 /* 671 * if the payload length field is 0 and the next header field 672 * indicates Hop-by-Hop Options header, then a Jumbo Payload 673 * option MUST be included. 674 */ 675 if (ip6->ip6_plen == 0 && plen == 0) { 676 /* 677 * Note that if a valid jumbo payload option is 678 * contained, ip6_hoptops_input() must set a valid 679 * (non-zero) payload length to the variable plen. 680 */ 681 ip6stat.ip6s_badoptions++; 682 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_discard); 683 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_hdrerr); 684 icmp6_error(m, ICMP6_PARAM_PROB, 685 ICMP6_PARAMPROB_HEADER, 686 (caddr_t)&ip6->ip6_plen - (caddr_t)ip6); 687 return; 688 } 689#ifndef PULLDOWN_TEST 690 /* ip6_hopopts_input() ensures that mbuf is contiguous */ 691 hbh = (struct ip6_hbh *)(ip6 + 1); 692#else 693 IP6_EXTHDR_GET(hbh, struct ip6_hbh *, m, sizeof(struct ip6_hdr), 694 sizeof(struct ip6_hbh)); 695 if (hbh == NULL) { 696 ip6stat.ip6s_tooshort++; 697 return; 698 } 699#endif 700 nxt = hbh->ip6h_nxt; 701 702 /* 703 * accept the packet if a router alert option is included 704 * and we act as an IPv6 router. 705 */ 706 if (rtalert != ~0 && ip6_forwarding) 707 ours = 1; 708 } else 709 nxt = ip6->ip6_nxt; 710 711 /* 712 * Check that the amount of data in the buffers 713 * is as at least much as the IPv6 header would have us expect. 714 * Trim mbufs if longer than we expect. 715 * Drop packet if shorter than we expect. 716 */ 717 if (m->m_pkthdr.len - sizeof(struct ip6_hdr) < plen) { 718 ip6stat.ip6s_tooshort++; 719 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_truncated); 720 goto bad; 721 } 722 if (m->m_pkthdr.len > sizeof(struct ip6_hdr) + plen) { 723 if (m->m_len == m->m_pkthdr.len) { 724 m->m_len = sizeof(struct ip6_hdr) + plen; 725 m->m_pkthdr.len = sizeof(struct ip6_hdr) + plen; 726 } else 727 m_adj(m, sizeof(struct ip6_hdr) + plen - m->m_pkthdr.len); 728 } 729 730 /* 731 * Forward if desirable. 732 */ 733 if (IN6_IS_ADDR_MULTICAST(&ip6->ip6_dst)) { 734 /* 735 * If we are acting as a multicast router, all 736 * incoming multicast packets are passed to the 737 * kernel-level multicast forwarding function. 738 * The packet is returned (relatively) intact; if 739 * ip6_mforward() returns a non-zero value, the packet 740 * must be discarded, else it may be accepted below. 741 */ 742 if (ip6_mrouter && ip6_mforward(ip6, m->m_pkthdr.rcvif, m)) { 743 ip6stat.ip6s_cantforward++; 744 m_freem(m); 745 return; 746 } 747 if (!ours) { 748 m_freem(m); 749 return; 750 } 751 } else if (!ours) { 752 ip6_forward(m, 0); 753 return; 754 } 755 756 ip6 = mtod(m, struct ip6_hdr *); 757 758 /* 759 * Malicious party may be able to use IPv4 mapped addr to confuse 760 * tcp/udp stack and bypass security checks (act as if it was from 761 * 127.0.0.1 by using IPv6 src ::ffff:127.0.0.1). Be cautious. 762 * 763 * For SIIT end node behavior, you may want to disable the check. 764 * However, you will become vulnerable to attacks using IPv4 mapped 765 * source. 766 */ 767 if (IN6_IS_ADDR_V4MAPPED(&ip6->ip6_src) || 768 IN6_IS_ADDR_V4MAPPED(&ip6->ip6_dst)) { 769 ip6stat.ip6s_badscope++; 770 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_addrerr); 771 goto bad; 772 } 773 774 /* 775 * Tell launch routine the next header 776 */ 777 ip6stat.ip6s_delivered++; 778 in6_ifstat_inc(deliverifp, ifs6_in_deliver); 779 nest = 0; 780 781 while (nxt != IPPROTO_DONE) { 782 if (ip6_hdrnestlimit && (++nest > ip6_hdrnestlimit)) { 783 ip6stat.ip6s_toomanyhdr++; 784 goto bad; 785 } 786 787 /* 788 * protection against faulty packet - there should be 789 * more sanity checks in header chain processing. 790 */ 791 if (m->m_pkthdr.len < off) { 792 ip6stat.ip6s_tooshort++; 793 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_truncated); 794 goto bad; 795 } 796 797#if 0 798 /* 799 * do we need to do it for every header? yeah, other 800 * functions can play with it (like re-allocate and copy). 801 */ 802 mhist = ip6_addaux(m); 803 if (mhist && M_TRAILINGSPACE(mhist) >= sizeof(nxt)) { 804 hist = mtod(mhist, caddr_t) + mhist->m_len; 805 bcopy(&nxt, hist, sizeof(nxt)); 806 mhist->m_len += sizeof(nxt); 807 } else { 808 ip6stat.ip6s_toomanyhdr++; 809 goto bad; 810 } 811#endif 812 813#ifdef IPSEC 814 /* 815 * enforce IPsec policy checking if we are seeing last header. 816 * note that we do not visit this with protocols with pcb layer 817 * code - like udp/tcp/raw ip. 818 */ 819 if ((inet6sw[ip6_protox[nxt]].pr_flags & PR_LASTHDR) != 0 && 820 ipsec6_in_reject(m, NULL)) { 821 ipsec6stat.in_polvio++; 822 goto bad; 823 } 824#endif 825 826 nxt = (*inet6sw[ip6_protox[nxt]].pr_input)(&m, &off, nxt); 827 } 828 return; 829 bad: 830 m_freem(m); 831} 832 833/* 834 * set/grab in6_ifaddr correspond to IPv6 destination address. 835 * XXX backward compatibility wrapper 836 */ 837static struct ip6aux * 838ip6_setdstifaddr(m, ia6) 839 struct mbuf *m; 840 struct in6_ifaddr *ia6; 841{ 842 struct ip6aux *n; 843 844 n = ip6_addaux(m); 845 if (n) 846 n->ip6a_dstia6 = ia6; 847 return n; /* NULL if failed to set */ 848} 849 850struct in6_ifaddr * 851ip6_getdstifaddr(m) 852 struct mbuf *m; 853{ 854 struct ip6aux *n; 855 856 n = ip6_findaux(m); 857 if (n) 858 return n->ip6a_dstia6; 859 else 860 return NULL; 861} 862 863/* 864 * Hop-by-Hop options header processing. If a valid jumbo payload option is 865 * included, the real payload length will be stored in plenp. 866 */ 867static int 868ip6_hopopts_input(plenp, rtalertp, mp, offp) 869 u_int32_t *plenp; 870 u_int32_t *rtalertp; /* XXX: should be stored more smart way */ 871 struct mbuf **mp; 872 int *offp; 873{ 874 struct mbuf *m = *mp; 875 int off = *offp, hbhlen; 876 struct ip6_hbh *hbh; 877 u_int8_t *opt; 878 879 /* validation of the length of the header */ 880#ifndef PULLDOWN_TEST 881 IP6_EXTHDR_CHECK(m, off, sizeof(*hbh), -1); 882 hbh = (struct ip6_hbh *)(mtod(m, caddr_t) + off); 883 hbhlen = (hbh->ip6h_len + 1) << 3; 884 885 IP6_EXTHDR_CHECK(m, off, hbhlen, -1); 886 hbh = (struct ip6_hbh *)(mtod(m, caddr_t) + off); 887#else 888 IP6_EXTHDR_GET(hbh, struct ip6_hbh *, m, 889 sizeof(struct ip6_hdr), sizeof(struct ip6_hbh)); 890 if (hbh == NULL) { 891 ip6stat.ip6s_tooshort++; 892 return -1; 893 } 894 hbhlen = (hbh->ip6h_len + 1) << 3; 895 IP6_EXTHDR_GET(hbh, struct ip6_hbh *, m, sizeof(struct ip6_hdr), 896 hbhlen); 897 if (hbh == NULL) { 898 ip6stat.ip6s_tooshort++; 899 return -1; 900 } 901#endif 902 off += hbhlen; 903 hbhlen -= sizeof(struct ip6_hbh); 904 opt = (u_int8_t *)hbh + sizeof(struct ip6_hbh); 905 906 if (ip6_process_hopopts(m, (u_int8_t *)hbh + sizeof(struct ip6_hbh), 907 hbhlen, rtalertp, plenp) < 0) 908 return(-1); 909 910 *offp = off; 911 *mp = m; 912 return(0); 913} 914 915/* 916 * Search header for all Hop-by-hop options and process each option. 917 * This function is separate from ip6_hopopts_input() in order to 918 * handle a case where the sending node itself process its hop-by-hop 919 * options header. In such a case, the function is called from ip6_output(). 920 * 921 * The function assumes that hbh header is located right after the IPv6 header 922 * (RFC2460 p7), opthead is pointer into data content in m, and opthead to 923 * opthead + hbhlen is located in continuous memory region. 924 */ 925int 926ip6_process_hopopts(m, opthead, hbhlen, rtalertp, plenp) 927 struct mbuf *m; 928 u_int8_t *opthead; 929 int hbhlen; 930 u_int32_t *rtalertp; 931 u_int32_t *plenp; 932{ 933 struct ip6_hdr *ip6; 934 int optlen = 0; 935 u_int8_t *opt = opthead; 936 u_int16_t rtalert_val; 937 u_int32_t jumboplen; 938 const int erroff = sizeof(struct ip6_hdr) + sizeof(struct ip6_hbh); 939 940 for (; hbhlen > 0; hbhlen -= optlen, opt += optlen) { 941 switch (*opt) { 942 case IP6OPT_PAD1: 943 optlen = 1; 944 break; 945 case IP6OPT_PADN: 946 if (hbhlen < IP6OPT_MINLEN) { 947 ip6stat.ip6s_toosmall++; 948 goto bad; 949 } 950 optlen = *(opt + 1) + 2; 951 break; 952 case IP6OPT_RTALERT: 953 /* XXX may need check for alignment */ 954 if (hbhlen < IP6OPT_RTALERT_LEN) { 955 ip6stat.ip6s_toosmall++; 956 goto bad; 957 } 958 if (*(opt + 1) != IP6OPT_RTALERT_LEN - 2) { 959 /* XXX stat */ 960 icmp6_error(m, ICMP6_PARAM_PROB, 961 ICMP6_PARAMPROB_HEADER, 962 erroff + opt + 1 - opthead); 963 return(-1); 964 } 965 optlen = IP6OPT_RTALERT_LEN; 966 bcopy((caddr_t)(opt + 2), (caddr_t)&rtalert_val, 2); 967 *rtalertp = ntohs(rtalert_val); 968 break; 969 case IP6OPT_JUMBO: 970 /* XXX may need check for alignment */ 971 if (hbhlen < IP6OPT_JUMBO_LEN) { 972 ip6stat.ip6s_toosmall++; 973 goto bad; 974 } 975 if (*(opt + 1) != IP6OPT_JUMBO_LEN - 2) { 976 /* XXX stat */ 977 icmp6_error(m, ICMP6_PARAM_PROB, 978 ICMP6_PARAMPROB_HEADER, 979 erroff + opt + 1 - opthead); 980 return(-1); 981 } 982 optlen = IP6OPT_JUMBO_LEN; 983 984 /* 985 * IPv6 packets that have non 0 payload length 986 * must not contain a jumbo payload option. 987 */ 988 ip6 = mtod(m, struct ip6_hdr *); 989 if (ip6->ip6_plen) { 990 ip6stat.ip6s_badoptions++; 991 icmp6_error(m, ICMP6_PARAM_PROB, 992 ICMP6_PARAMPROB_HEADER, 993 erroff + opt - opthead); 994 return(-1); 995 } 996 997 /* 998 * We may see jumbolen in unaligned location, so 999 * we'd need to perform bcopy(). 1000 */ 1001 bcopy(opt + 2, &jumboplen, sizeof(jumboplen)); 1002 jumboplen = (u_int32_t)htonl(jumboplen); 1003 1004#if 1 1005 /* 1006 * if there are multiple jumbo payload options, 1007 * *plenp will be non-zero and the packet will be 1008 * rejected. 1009 * the behavior may need some debate in ipngwg - 1010 * multiple options does not make sense, however, 1011 * there's no explicit mention in specification. 1012 */ 1013 if (*plenp != 0) { 1014 ip6stat.ip6s_badoptions++; 1015 icmp6_error(m, ICMP6_PARAM_PROB, 1016 ICMP6_PARAMPROB_HEADER, 1017 erroff + opt + 2 - opthead); 1018 return(-1); 1019 } 1020#endif 1021 1022 /* 1023 * jumbo payload length must be larger than 65535. 1024 */ 1025 if (jumboplen <= IPV6_MAXPACKET) { 1026 ip6stat.ip6s_badoptions++; 1027 icmp6_error(m, ICMP6_PARAM_PROB, 1028 ICMP6_PARAMPROB_HEADER, 1029 erroff + opt + 2 - opthead); 1030 return(-1); 1031 } 1032 *plenp = jumboplen; 1033 1034 break; 1035 default: /* unknown option */ 1036 if (hbhlen < IP6OPT_MINLEN) { 1037 ip6stat.ip6s_toosmall++; 1038 goto bad; 1039 } 1040 optlen = ip6_unknown_opt(opt, m, 1041 erroff + opt - opthead); 1042 if (optlen == -1) 1043 return(-1); 1044 optlen += 2; 1045 break; 1046 } 1047 } 1048 1049 return(0); 1050 1051 bad: 1052 m_freem(m); 1053 return(-1); 1054} 1055 1056/* 1057 * Unknown option processing. 1058 * The third argument `off' is the offset from the IPv6 header to the option, 1059 * which is necessary if the IPv6 header the and option header and IPv6 header 1060 * is not continuous in order to return an ICMPv6 error. 1061 */ 1062int 1063ip6_unknown_opt(optp, m, off) 1064 u_int8_t *optp; 1065 struct mbuf *m; 1066 int off; 1067{ 1068 struct ip6_hdr *ip6; 1069 1070 switch (IP6OPT_TYPE(*optp)) { 1071 case IP6OPT_TYPE_SKIP: /* ignore the option */ 1072 return((int)*(optp + 1)); 1073 case IP6OPT_TYPE_DISCARD: /* silently discard */ 1074 m_freem(m); 1075 return(-1); 1076 case IP6OPT_TYPE_FORCEICMP: /* send ICMP even if multicasted */ 1077 ip6stat.ip6s_badoptions++; 1078 icmp6_error(m, ICMP6_PARAM_PROB, ICMP6_PARAMPROB_OPTION, off); 1079 return(-1); 1080 case IP6OPT_TYPE_ICMP: /* send ICMP if not multicasted */ 1081 ip6stat.ip6s_badoptions++; 1082 ip6 = mtod(m, struct ip6_hdr *); 1083 if (IN6_IS_ADDR_MULTICAST(&ip6->ip6_dst) || 1084 (m->m_flags & (M_BCAST|M_MCAST))) 1085 m_freem(m); 1086 else 1087 icmp6_error(m, ICMP6_PARAM_PROB, 1088 ICMP6_PARAMPROB_OPTION, off); 1089 return(-1); 1090 } 1091 1092 m_freem(m); /* XXX: NOTREACHED */ 1093 return(-1); 1094} 1095 1096/* 1097 * Create the "control" list for this pcb. 1098 * The function will not modify mbuf chain at all. 1099 * 1100 * with KAME mbuf chain restriction: 1101 * The routine will be called from upper layer handlers like tcp6_input(). 1102 * Thus the routine assumes that the caller (tcp6_input) have already 1103 * called IP6_EXTHDR_CHECK() and all the extension headers are located in the 1104 * very first mbuf on the mbuf chain. 1105 */ 1106void 1107ip6_savecontrol(in6p, mp, ip6, m) 1108 struct inpcb *in6p; 1109 struct mbuf **mp; 1110 struct ip6_hdr *ip6; 1111 struct mbuf *m; 1112{ 1113#if __FreeBSD_version >= 500000 1114 struct thread *td = curthread; /* XXX */ 1115#else 1116 struct proc *td = curproc; /* XXX */ 1117#endif 1118 int privileged = 0; 1119 int rthdr_exist = 0; 1120 1121 1122 if (td && !suser(td)) 1123 privileged++; 1124 1125#ifdef SO_TIMESTAMP 1126 if ((in6p->in6p_socket->so_options & SO_TIMESTAMP) != 0) { 1127 struct timeval tv; 1128 1129 microtime(&tv); 1130 *mp = sbcreatecontrol((caddr_t) &tv, sizeof(tv), 1131 SCM_TIMESTAMP, SOL_SOCKET); 1132 if (*mp) { 1133 mp = &(*mp)->m_next; 1134 } 1135 } 1136#endif 1137 1138 /* RFC 2292 sec. 5 */ 1139 if ((in6p->in6p_flags & IN6P_PKTINFO) != 0) { 1140 struct in6_pktinfo pi6; 1141 bcopy(&ip6->ip6_dst, &pi6.ipi6_addr, sizeof(struct in6_addr)); 1142 if (IN6_IS_SCOPE_LINKLOCAL(&pi6.ipi6_addr)) 1143 pi6.ipi6_addr.s6_addr16[1] = 0; 1144 pi6.ipi6_ifindex = (m && m->m_pkthdr.rcvif) 1145 ? m->m_pkthdr.rcvif->if_index 1146 : 0; 1147 *mp = sbcreatecontrol((caddr_t) &pi6, 1148 sizeof(struct in6_pktinfo), IPV6_PKTINFO, 1149 IPPROTO_IPV6); 1150 if (*mp) 1151 mp = &(*mp)->m_next; 1152 } 1153 1154 if ((in6p->in6p_flags & IN6P_HOPLIMIT) != 0) { 1155 int hlim = ip6->ip6_hlim & 0xff; 1156 *mp = sbcreatecontrol((caddr_t) &hlim, 1157 sizeof(int), IPV6_HOPLIMIT, IPPROTO_IPV6); 1158 if (*mp) 1159 mp = &(*mp)->m_next; 1160 } 1161 1162 /* 1163 * IPV6_HOPOPTS socket option. We require super-user privilege 1164 * for the option, but it might be too strict, since there might 1165 * be some hop-by-hop options which can be returned to normal user. 1166 * See RFC 2292 section 6. 1167 */ 1168 if ((in6p->in6p_flags & IN6P_HOPOPTS) != 0 && privileged) { 1169 /* 1170 * Check if a hop-by-hop options header is contatined in the 1171 * received packet, and if so, store the options as ancillary 1172 * data. Note that a hop-by-hop options header must be 1173 * just after the IPv6 header, which fact is assured through 1174 * the IPv6 input processing. 1175 */ 1176 struct ip6_hdr *ip6 = mtod(m, struct ip6_hdr *); 1177 if (ip6->ip6_nxt == IPPROTO_HOPOPTS) { 1178 struct ip6_hbh *hbh; 1179 int hbhlen = 0; 1180#ifdef PULLDOWN_TEST 1181 struct mbuf *ext; 1182#endif 1183 1184#ifndef PULLDOWN_TEST 1185 hbh = (struct ip6_hbh *)(ip6 + 1); 1186 hbhlen = (hbh->ip6h_len + 1) << 3; 1187#else 1188 ext = ip6_pullexthdr(m, sizeof(struct ip6_hdr), 1189 ip6->ip6_nxt); 1190 if (ext == NULL) { 1191 ip6stat.ip6s_tooshort++; 1192 return; 1193 } 1194 hbh = mtod(ext, struct ip6_hbh *); 1195 hbhlen = (hbh->ip6h_len + 1) << 3; 1196 if (hbhlen != ext->m_len) { 1197 m_freem(ext); 1198 ip6stat.ip6s_tooshort++; 1199 return; 1200 } 1201#endif 1202 1203 /* 1204 * XXX: We copy whole the header even if a jumbo 1205 * payload option is included, which option is to 1206 * be removed before returning in the RFC 2292. 1207 * Note: this constraint is removed in 2292bis. 1208 */ 1209 *mp = sbcreatecontrol((caddr_t)hbh, hbhlen, 1210 IPV6_HOPOPTS, IPPROTO_IPV6); 1211 if (*mp) 1212 mp = &(*mp)->m_next; 1213#ifdef PULLDOWN_TEST 1214 m_freem(ext); 1215#endif 1216 } 1217 } 1218 1219 /* IPV6_DSTOPTS and IPV6_RTHDR socket options */ 1220 if ((in6p->in6p_flags & (IN6P_DSTOPTS | IN6P_RTHDRDSTOPTS)) != 0) { 1221 int proto, off, nxt; 1222 1223 /* 1224 * go through the header chain to see if a routing header is 1225 * contained in the packet. We need this information to store 1226 * destination options headers (if any) properly. 1227 * XXX: performance issue. We should record this info when 1228 * processing extension headers in incoming routine. 1229 * (todo) use m_aux? 1230 */ 1231 proto = IPPROTO_IPV6; 1232 off = 0; 1233 nxt = -1; 1234 while (1) { 1235 int newoff; 1236 1237 newoff = ip6_nexthdr(m, off, proto, &nxt); 1238 if (newoff < 0) 1239 break; 1240 if (newoff < off) /* invalid, check for safety */ 1241 break; 1242 if ((proto = nxt) == IPPROTO_ROUTING) { 1243 rthdr_exist = 1; 1244 break; 1245 } 1246 off = newoff; 1247 } 1248 } 1249 1250 if ((in6p->in6p_flags & 1251 (IN6P_RTHDR | IN6P_DSTOPTS | IN6P_RTHDRDSTOPTS)) != 0) { 1252 struct ip6_hdr *ip6 = mtod(m, struct ip6_hdr *); 1253 int nxt = ip6->ip6_nxt, off = sizeof(struct ip6_hdr); 1254 1255 /* 1256 * Search for destination options headers or routing 1257 * header(s) through the header chain, and stores each 1258 * header as ancillary data. 1259 * Note that the order of the headers remains in 1260 * the chain of ancillary data. 1261 */ 1262 while (1) { /* is explicit loop prevention necessary? */ 1263 struct ip6_ext *ip6e = NULL; 1264 int elen; 1265#ifdef PULLDOWN_TEST 1266 struct mbuf *ext = NULL; 1267#endif 1268 1269 /* 1270 * if it is not an extension header, don't try to 1271 * pull it from the chain. 1272 */ 1273 switch (nxt) { 1274 case IPPROTO_DSTOPTS: 1275 case IPPROTO_ROUTING: 1276 case IPPROTO_HOPOPTS: 1277 case IPPROTO_AH: /* is it possible? */ 1278 break; 1279 default: 1280 goto loopend; 1281 } 1282 1283#ifndef PULLDOWN_TEST 1284 if (off + sizeof(*ip6e) > m->m_len) 1285 goto loopend; 1286 ip6e = (struct ip6_ext *)(mtod(m, caddr_t) + off); 1287 if (nxt == IPPROTO_AH) 1288 elen = (ip6e->ip6e_len + 2) << 2; 1289 else 1290 elen = (ip6e->ip6e_len + 1) << 3; 1291 if (off + elen > m->m_len) 1292 goto loopend; 1293#else 1294 ext = ip6_pullexthdr(m, off, nxt); 1295 if (ext == NULL) { 1296 ip6stat.ip6s_tooshort++; 1297 return; 1298 } 1299 ip6e = mtod(ext, struct ip6_ext *); 1300 if (nxt == IPPROTO_AH) 1301 elen = (ip6e->ip6e_len + 2) << 2; 1302 else 1303 elen = (ip6e->ip6e_len + 1) << 3; 1304 if (elen != ext->m_len) { 1305 m_freem(ext); 1306 ip6stat.ip6s_tooshort++; 1307 return; 1308 } 1309#endif 1310 1311 switch (nxt) { 1312 case IPPROTO_DSTOPTS: 1313 if ((in6p->in6p_flags & IN6P_DSTOPTS) == 0) 1314 break; 1315 1316 /* 1317 * We also require super-user privilege for 1318 * the option. 1319 * See the comments on IN6_HOPOPTS. 1320 */ 1321 if (!privileged) 1322 break; 1323 1324 *mp = sbcreatecontrol((caddr_t)ip6e, elen, 1325 IPV6_DSTOPTS, 1326 IPPROTO_IPV6); 1327 if (*mp) 1328 mp = &(*mp)->m_next; 1329 break; 1330 case IPPROTO_ROUTING: 1331 if (!in6p->in6p_flags & IN6P_RTHDR) 1332 break; 1333 1334 *mp = sbcreatecontrol((caddr_t)ip6e, elen, 1335 IPV6_RTHDR, 1336 IPPROTO_IPV6); 1337 if (*mp) 1338 mp = &(*mp)->m_next; 1339 break; 1340 case IPPROTO_HOPOPTS: 1341 case IPPROTO_AH: /* is it possible? */ 1342 break; 1343 1344 default: 1345 /* 1346 * other cases have been filtered in the above. 1347 * none will visit this case. here we supply 1348 * the code just in case (nxt overwritten or 1349 * other cases). 1350 */ 1351#ifdef PULLDOWN_TEST 1352 m_freem(ext); 1353#endif 1354 goto loopend; 1355 1356 } 1357 1358 /* proceed with the next header. */ 1359 off += elen; 1360 nxt = ip6e->ip6e_nxt; 1361 ip6e = NULL; 1362#ifdef PULLDOWN_TEST 1363 m_freem(ext); 1364 ext = NULL; 1365#endif 1366 } 1367 loopend: 1368 ; 1369 } 1370 1371} 1372 1373#ifdef PULLDOWN_TEST 1374/* 1375 * pull single extension header from mbuf chain. returns single mbuf that 1376 * contains the result, or NULL on error. 1377 */ 1378static struct mbuf * 1379ip6_pullexthdr(m, off, nxt) 1380 struct mbuf *m; 1381 size_t off; 1382 int nxt; 1383{ 1384 struct ip6_ext ip6e; 1385 size_t elen; 1386 struct mbuf *n; 1387 1388#ifdef DIAGNOSTIC 1389 switch (nxt) { 1390 case IPPROTO_DSTOPTS: 1391 case IPPROTO_ROUTING: 1392 case IPPROTO_HOPOPTS: 1393 case IPPROTO_AH: /* is it possible? */ 1394 break; 1395 default: 1396 printf("ip6_pullexthdr: invalid nxt=%d\n", nxt); 1397 } 1398#endif 1399 1400 m_copydata(m, off, sizeof(ip6e), (caddr_t)&ip6e); 1401 if (nxt == IPPROTO_AH) 1402 elen = (ip6e.ip6e_len + 2) << 2; 1403 else 1404 elen = (ip6e.ip6e_len + 1) << 3; 1405 1406 MGET(n, M_DONTWAIT, MT_DATA); 1407 if (n && elen >= MLEN) { 1408 MCLGET(n, M_DONTWAIT); 1409 if ((n->m_flags & M_EXT) == 0) { 1410 m_free(n); 1411 n = NULL; 1412 } 1413 } 1414 if (!n) 1415 return NULL; 1416 1417 n->m_len = 0; 1418 if (elen >= M_TRAILINGSPACE(n)) { 1419 m_free(n); 1420 return NULL; 1421 } 1422 1423 m_copydata(m, off, elen, mtod(n, caddr_t)); 1424 n->m_len = elen; 1425 return n; 1426} 1427#endif 1428 1429/* 1430 * Get pointer to the previous header followed by the header 1431 * currently processed. 1432 * XXX: This function supposes that 1433 * M includes all headers, 1434 * the next header field and the header length field of each header 1435 * are valid, and 1436 * the sum of each header length equals to OFF. 1437 * Because of these assumptions, this function must be called very 1438 * carefully. Moreover, it will not be used in the near future when 1439 * we develop `neater' mechanism to process extension headers. 1440 */ 1441char * 1442ip6_get_prevhdr(m, off) 1443 struct mbuf *m; 1444 int off; 1445{ 1446 struct ip6_hdr *ip6 = mtod(m, struct ip6_hdr *); 1447 1448 if (off == sizeof(struct ip6_hdr)) 1449 return(&ip6->ip6_nxt); 1450 else { 1451 int len, nxt; 1452 struct ip6_ext *ip6e = NULL; 1453 1454 nxt = ip6->ip6_nxt; 1455 len = sizeof(struct ip6_hdr); 1456 while (len < off) { 1457 ip6e = (struct ip6_ext *)(mtod(m, caddr_t) + len); 1458 1459 switch (nxt) { 1460 case IPPROTO_FRAGMENT: 1461 len += sizeof(struct ip6_frag); 1462 break; 1463 case IPPROTO_AH: 1464 len += (ip6e->ip6e_len + 2) << 2; 1465 break; 1466 default: 1467 len += (ip6e->ip6e_len + 1) << 3; 1468 break; 1469 } 1470 nxt = ip6e->ip6e_nxt; 1471 } 1472 if (ip6e) 1473 return(&ip6e->ip6e_nxt); 1474 else 1475 return NULL; 1476 } 1477} 1478 1479/* 1480 * get next header offset. m will be retained. 1481 */ 1482int 1483ip6_nexthdr(m, off, proto, nxtp) 1484 struct mbuf *m; 1485 int off; 1486 int proto; 1487 int *nxtp; 1488{ 1489 struct ip6_hdr ip6; 1490 struct ip6_ext ip6e; 1491 struct ip6_frag fh; 1492 1493 /* just in case */ 1494 if (m == NULL) 1495 panic("ip6_nexthdr: m == NULL"); 1496 if ((m->m_flags & M_PKTHDR) == 0 || m->m_pkthdr.len < off) 1497 return -1; 1498 1499 switch (proto) { 1500 case IPPROTO_IPV6: 1501 if (m->m_pkthdr.len < off + sizeof(ip6)) 1502 return -1; 1503 m_copydata(m, off, sizeof(ip6), (caddr_t)&ip6); 1504 if (nxtp) 1505 *nxtp = ip6.ip6_nxt; 1506 off += sizeof(ip6); 1507 return off; 1508 1509 case IPPROTO_FRAGMENT: 1510 /* 1511 * terminate parsing if it is not the first fragment, 1512 * it does not make sense to parse through it. 1513 */ 1514 if (m->m_pkthdr.len < off + sizeof(fh)) 1515 return -1; 1516 m_copydata(m, off, sizeof(fh), (caddr_t)&fh); 1517 if ((ntohs(fh.ip6f_offlg) & IP6F_OFF_MASK) != 0) 1518 return -1; 1519 if (nxtp) 1520 *nxtp = fh.ip6f_nxt; 1521 off += sizeof(struct ip6_frag); 1522 return off; 1523 1524 case IPPROTO_AH: 1525 if (m->m_pkthdr.len < off + sizeof(ip6e)) 1526 return -1; 1527 m_copydata(m, off, sizeof(ip6e), (caddr_t)&ip6e); 1528 if (nxtp) 1529 *nxtp = ip6e.ip6e_nxt; 1530 off += (ip6e.ip6e_len + 2) << 2; 1531 return off; 1532 1533 case IPPROTO_HOPOPTS: 1534 case IPPROTO_ROUTING: 1535 case IPPROTO_DSTOPTS: 1536 if (m->m_pkthdr.len < off + sizeof(ip6e)) 1537 return -1; 1538 m_copydata(m, off, sizeof(ip6e), (caddr_t)&ip6e); 1539 if (nxtp) 1540 *nxtp = ip6e.ip6e_nxt; 1541 off += (ip6e.ip6e_len + 1) << 3; 1542 return off; 1543 1544 case IPPROTO_NONE: 1545 case IPPROTO_ESP: 1546 case IPPROTO_IPCOMP: 1547 /* give up */ 1548 return -1; 1549 1550 default: 1551 return -1; 1552 } 1553 1554 return -1; 1555} 1556 1557/* 1558 * get offset for the last header in the chain. m will be kept untainted. 1559 */ 1560int 1561ip6_lasthdr(m, off, proto, nxtp) 1562 struct mbuf *m; 1563 int off; 1564 int proto; 1565 int *nxtp; 1566{ 1567 int newoff; 1568 int nxt; 1569 1570 if (!nxtp) { 1571 nxt = -1; 1572 nxtp = &nxt; 1573 } 1574 while (1) { 1575 newoff = ip6_nexthdr(m, off, proto, nxtp); 1576 if (newoff < 0) 1577 return off; 1578 else if (newoff < off) 1579 return -1; /* invalid */ 1580 else if (newoff == off) 1581 return newoff; 1582 1583 off = newoff; 1584 proto = *nxtp; 1585 } 1586} 1587 1588struct ip6aux * 1589ip6_addaux(m) 1590 struct mbuf *m; 1591{ 1592 struct m_tag *tag = m_tag_find(m, PACKET_TAG_IPV6_INPUT, NULL); 1593 if (!tag) { 1594 tag = m_tag_get(PACKET_TAG_IPV6_INPUT, 1595 sizeof (struct ip6aux), 1596 M_NOWAIT); 1597 if (tag) 1598 m_tag_prepend(m, tag); 1599 } 1600 if (tag) 1601 bzero(tag+1, sizeof (struct ip6aux)); 1602 return tag ? (struct ip6aux*)(tag+1) : NULL; 1603} 1604 1605struct ip6aux * 1606ip6_findaux(m) 1607 struct mbuf *m; 1608{ 1609 struct m_tag *tag = m_tag_find(m, PACKET_TAG_IPV6_INPUT, NULL); 1610 return tag ? (struct ip6aux*)(tag+1) : NULL; 1611} 1612 1613void 1614ip6_delaux(m) 1615 struct mbuf *m; 1616{ 1617 struct m_tag *tag = m_tag_find(m, PACKET_TAG_IPV6_INPUT, NULL); 1618 if (tag) 1619 m_tag_delete(m, tag); 1620} 1621 1622/* 1623 * System control for IP6 1624 */ 1625 1626u_char inet6ctlerrmap[PRC_NCMDS] = { 1627 0, 0, 0, 0, 1628 0, EMSGSIZE, EHOSTDOWN, EHOSTUNREACH, 1629 EHOSTUNREACH, EHOSTUNREACH, ECONNREFUSED, ECONNREFUSED, 1630 EMSGSIZE, EHOSTUNREACH, 0, 0, 1631 0, 0, 0, 0, 1632 ENOPROTOOPT 1633}; 1634