1 /* $FreeBSD: src/sys/netinet6/ip6_input.c,v 1.11.2.15 2003/01/24 05:11:35 sam Exp $ */ 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. Neither the name of the University nor the names of its contributors 46 * may be used to endorse or promote products derived from this software 47 * without specific prior written permission. 48 * 49 * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND 50 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE 51 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE 52 * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE 53 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL 54 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS 55 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) 56 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT 57 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY 58 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF 59 * SUCH DAMAGE. 60 * 61 * @(#)ip_input.c 8.2 (Berkeley) 1/4/94 62 */ 63 64 #include "opt_ip6fw.h" 65 #include "opt_inet.h" 66 #include "opt_inet6.h" 67 #include "opt_ipsec.h" 68 69 #include <sys/param.h> 70 #include <sys/systm.h> 71 #include <sys/malloc.h> 72 #include <sys/mbuf.h> 73 #include <sys/domain.h> 74 #include <sys/protosw.h> 75 #include <sys/socket.h> 76 #include <sys/socketvar.h> 77 #include <sys/errno.h> 78 #include <sys/time.h> 79 #include <sys/kernel.h> 80 #include <sys/syslog.h> 81 #include <sys/proc.h> 82 #include <sys/priv.h> 83 84 #include <net/if.h> 85 #include <net/if_types.h> 86 #include <net/if_dl.h> 87 #include <net/route.h> 88 #include <net/netisr.h> 89 #include <net/pfil.h> 90 91 #include <sys/thread2.h> 92 #include <sys/msgport2.h> 93 #include <net/netmsg2.h> 94 #include <net/netisr2.h> 95 96 #include <netinet/in.h> 97 #include <netinet/in_systm.h> 98 #ifdef INET 99 #include <netinet/ip.h> 100 #include <netinet/ip_icmp.h> 101 #endif /* INET */ 102 #include <netinet/ip6.h> 103 #include <netinet6/in6_var.h> 104 #include <netinet6/ip6_var.h> 105 #include <netinet/in_pcb.h> 106 #include <netinet/icmp6.h> 107 #include <netinet6/scope6_var.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 <netproto/ipsec/ipsec.h> 121 #include <netproto/ipsec/ipsec6.h> 122 #define IPSEC 123 #endif /* FAST_IPSEC */ 124 125 #include <net/ip6fw/ip6_fw.h> 126 127 #include <netinet6/ip6protosw.h> 128 129 #include <net/net_osdep.h> 130 131 extern struct domain inet6domain; 132 extern struct protosw inet6sw[]; 133 134 u_char ip6_protox[IPPROTO_MAX]; 135 struct in6_ifaddr *in6_ifaddr; 136 137 extern struct callout in6_tmpaddrtimer_ch; 138 139 int ip6_forward_srcrt; /* XXX */ 140 int ip6_sourcecheck; /* XXX */ 141 int ip6_sourcecheck_interval; /* XXX */ 142 143 int ip6_ours_check_algorithm; 144 145 struct pfil_head inet6_pfil_hook; 146 147 /* firewall hooks */ 148 ip6_fw_chk_t *ip6_fw_chk_ptr; 149 ip6_fw_ctl_t *ip6_fw_ctl_ptr; 150 int ip6_fw_enable = 1; 151 152 struct ip6stat ip6stat; 153 154 static void ip6_init2 (void *); 155 static struct ip6aux *ip6_setdstifaddr (struct mbuf *, struct in6_ifaddr *); 156 static int ip6_hopopts_input (u_int32_t *, u_int32_t *, struct mbuf **, int *); 157 static void ip6_input(netmsg_t msg); 158 #ifdef PULLDOWN_TEST 159 static struct mbuf *ip6_pullexthdr (struct mbuf *, size_t, int); 160 #endif 161 static void transport6_processing_handler(netmsg_t netmsg); 162 163 /* 164 * IP6 initialization: fill in IP6 protocol switch table. 165 * All protocols not implemented in kernel go to raw IP6 protocol handler. 166 */ 167 void 168 ip6_init(void) 169 { 170 struct protosw *pr; 171 int i; 172 struct timeval tv; 173 174 pr = pffindproto(PF_INET6, IPPROTO_RAW, SOCK_RAW); 175 if (pr == NULL) 176 panic("ip6_init"); 177 for (i = 0; i < IPPROTO_MAX; i++) 178 ip6_protox[i] = pr - inet6sw; 179 for (pr = inet6domain.dom_protosw; 180 pr < inet6domain.dom_protoswNPROTOSW; pr++) 181 if (pr->pr_domain->dom_family == PF_INET6 && 182 pr->pr_protocol && pr->pr_protocol != IPPROTO_RAW) 183 ip6_protox[pr->pr_protocol] = pr - inet6sw; 184 185 inet6_pfil_hook.ph_type = PFIL_TYPE_AF; 186 inet6_pfil_hook.ph_af = AF_INET6; 187 if ((i = pfil_head_register(&inet6_pfil_hook)) != 0) { 188 kprintf("%s: WARNING: unable to register pfil hook, " 189 "error %d\n", __func__, i); 190 } 191 192 netisr_register(NETISR_IPV6, ip6_input, NULL); /* XXX hashfn */ 193 scope6_init(); 194 addrsel_policy_init(); 195 nd6_init(); 196 frag6_init(); 197 /* 198 * in many cases, random() here does NOT return random number 199 * as initialization during bootstrap time occur in fixed order. 200 */ 201 microtime(&tv); 202 ip6_flow_seq = krandom() ^ tv.tv_usec; 203 microtime(&tv); 204 ip6_desync_factor = (krandom() ^ tv.tv_usec) % MAX_TEMP_DESYNC_FACTOR; 205 } 206 207 static void 208 ip6_init2(void *dummy) 209 { 210 /* nd6_timer_init */ 211 callout_init(&nd6_timer_ch); 212 callout_reset(&nd6_timer_ch, hz, nd6_timer, NULL); 213 214 /* router renumbering prefix list maintenance */ 215 callout_init(&in6_rr_timer_ch); 216 callout_reset(&in6_rr_timer_ch, hz, in6_rr_timer, NULL); 217 218 /* timer for regeneranation of temporary addresses randomize ID */ 219 callout_init(&in6_tmpaddrtimer_ch); 220 callout_reset(&in6_tmpaddrtimer_ch, 221 (ip6_temp_preferred_lifetime - ip6_desync_factor - 222 ip6_temp_regen_advance) * hz, 223 in6_tmpaddrtimer, NULL); 224 } 225 226 /* cheat */ 227 /* This must be after route_init(), which is now SI_ORDER_THIRD */ 228 SYSINIT(netinet6init2, SI_SUB_PROTO_DOMAIN, SI_ORDER_MIDDLE, ip6_init2, NULL); 229 230 extern struct route_in6 ip6_forward_rt; 231 232 static 233 void 234 ip6_input(netmsg_t msg) 235 { 236 struct mbuf *m = msg->packet.nm_packet; 237 struct ip6_hdr *ip6; 238 int off = sizeof(struct ip6_hdr), nest; 239 u_int32_t plen; 240 u_int32_t rtalert = ~0; 241 int nxt, ours = 0, rh_present = 0; 242 struct ifnet *deliverifp = NULL; 243 struct in6_addr odst; 244 int srcrt = 0; 245 246 #ifdef IPSEC 247 /* 248 * should the inner packet be considered authentic? 249 * see comment in ah4_input(). 250 */ 251 if (m) { 252 m->m_flags &= ~M_AUTHIPHDR; 253 m->m_flags &= ~M_AUTHIPDGM; 254 } 255 #endif 256 257 /* 258 * make sure we don't have onion peering information into m_aux. 259 */ 260 ip6_delaux(m); 261 262 /* 263 * mbuf statistics 264 */ 265 if (m->m_flags & M_EXT) { 266 if (m->m_next) 267 ip6stat.ip6s_mext2m++; 268 else 269 ip6stat.ip6s_mext1++; 270 } else { 271 #define M2MMAX NELEM(ip6stat.ip6s_m2m) 272 if (m->m_next) { 273 if (m->m_flags & M_LOOP) { 274 ip6stat.ip6s_m2m[loif[0].if_index]++; /* XXX */ 275 } else if (m->m_pkthdr.rcvif->if_index < M2MMAX) 276 ip6stat.ip6s_m2m[m->m_pkthdr.rcvif->if_index]++; 277 else 278 ip6stat.ip6s_m2m[0]++; 279 } else 280 ip6stat.ip6s_m1++; 281 #undef M2MMAX 282 } 283 284 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_receive); 285 ip6stat.ip6s_total++; 286 287 #ifndef PULLDOWN_TEST 288 /* 289 * L2 bridge code and some other code can return mbuf chain 290 * that does not conform to KAME requirement. too bad. 291 * XXX: fails to join if interface MTU > MCLBYTES. jumbogram? 292 */ 293 if (m && m->m_next != NULL && m->m_pkthdr.len < MCLBYTES) { 294 struct mbuf *n; 295 296 n = m_getb(m->m_pkthdr.len, MB_DONTWAIT, MT_HEADER, M_PKTHDR); 297 if (n == NULL) 298 goto bad; 299 M_MOVE_PKTHDR(n, m); 300 301 m_copydata(m, 0, n->m_pkthdr.len, mtod(n, caddr_t)); 302 n->m_len = n->m_pkthdr.len; 303 m_freem(m); 304 m = n; 305 } 306 IP6_EXTHDR_CHECK_VOIDRET(m, 0, sizeof(struct ip6_hdr)); 307 #endif 308 309 if (m->m_len < sizeof(struct ip6_hdr)) { 310 struct ifnet *inifp; 311 inifp = m->m_pkthdr.rcvif; 312 if ((m = m_pullup(m, sizeof(struct ip6_hdr))) == NULL) { 313 ip6stat.ip6s_toosmall++; 314 in6_ifstat_inc(inifp, ifs6_in_hdrerr); 315 goto bad2; 316 } 317 } 318 319 ip6 = mtod(m, struct ip6_hdr *); 320 321 if ((ip6->ip6_vfc & IPV6_VERSION_MASK) != IPV6_VERSION) { 322 ip6stat.ip6s_badvers++; 323 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_hdrerr); 324 goto bad; 325 } 326 327 /* 328 * Run through list of hooks for input packets. 329 * 330 * NB: Beware of the destination address changing 331 * (e.g. by NAT rewriting). When this happens, 332 * tell ip6_forward to do the right thing. 333 */ 334 if (pfil_has_hooks(&inet6_pfil_hook)) { 335 odst = ip6->ip6_dst; 336 if (pfil_run_hooks(&inet6_pfil_hook, &m, 337 m->m_pkthdr.rcvif, PFIL_IN)) { 338 goto bad2; 339 } 340 if (m == NULL) /* consumed by filter */ 341 goto bad2; 342 ip6 = mtod(m, struct ip6_hdr *); 343 srcrt = !IN6_ARE_ADDR_EQUAL(&odst, &ip6->ip6_dst); 344 } 345 346 ip6stat.ip6s_nxthist[ip6->ip6_nxt]++; 347 348 #ifdef ALTQ 349 if (altq_input != NULL && (*altq_input)(m, AF_INET6) == 0) { 350 /* packet is dropped by traffic conditioner */ 351 return; 352 } 353 #endif 354 355 /* 356 * Check with the firewall... 357 */ 358 if (ip6_fw_enable && ip6_fw_chk_ptr) { 359 u_short port = 0; 360 /* If ipfw says divert, we have to just drop packet */ 361 /* use port as a dummy argument */ 362 if ((*ip6_fw_chk_ptr)(&ip6, NULL, &port, &m)) { 363 m_freem(m); 364 m = NULL; 365 } 366 if (!m) 367 goto bad2; 368 } 369 370 /* 371 * Check against address spoofing/corruption. 372 */ 373 if (IN6_IS_ADDR_MULTICAST(&ip6->ip6_src) || 374 IN6_IS_ADDR_UNSPECIFIED(&ip6->ip6_dst)) { 375 /* 376 * XXX: "badscope" is not very suitable for a multicast source. 377 */ 378 ip6stat.ip6s_badscope++; 379 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_addrerr); 380 goto bad; 381 } 382 if ((IN6_IS_ADDR_LOOPBACK(&ip6->ip6_src) || 383 IN6_IS_ADDR_LOOPBACK(&ip6->ip6_dst)) && 384 (m->m_pkthdr.rcvif->if_flags & IFF_LOOPBACK) == 0) { 385 ip6stat.ip6s_badscope++; 386 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_addrerr); 387 goto bad; 388 } 389 390 /* 391 * The following check is not documented in specs. A malicious 392 * party may be able to use IPv4 mapped addr to confuse tcp/udp stack 393 * and bypass security checks (act as if it was from 127.0.0.1 by using 394 * IPv6 src ::ffff:127.0.0.1). Be cautious. 395 * 396 * This check chokes if we are in an SIIT cloud. As none of BSDs 397 * support IPv4-less kernel compilation, we cannot support SIIT 398 * environment at all. So, it makes more sense for us to reject any 399 * malicious packets for non-SIIT environment, than try to do a 400 * partical support for SIIT environment. 401 */ 402 if (IN6_IS_ADDR_V4MAPPED(&ip6->ip6_src) || 403 IN6_IS_ADDR_V4MAPPED(&ip6->ip6_dst)) { 404 ip6stat.ip6s_badscope++; 405 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_addrerr); 406 goto bad; 407 } 408 #if 0 409 /* 410 * Reject packets with IPv4 compatible addresses (auto tunnel). 411 * 412 * The code forbids auto tunnel relay case in RFC1933 (the check is 413 * stronger than RFC1933). We may want to re-enable it if mech-xx 414 * is revised to forbid relaying case. 415 */ 416 if (IN6_IS_ADDR_V4COMPAT(&ip6->ip6_src) || 417 IN6_IS_ADDR_V4COMPAT(&ip6->ip6_dst)) { 418 ip6stat.ip6s_badscope++; 419 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_addrerr); 420 goto bad; 421 } 422 #endif 423 424 /* drop packets if interface ID portion is already filled */ 425 if ((m->m_pkthdr.rcvif->if_flags & IFF_LOOPBACK) == 0) { 426 if (IN6_IS_SCOPE_LINKLOCAL(&ip6->ip6_src) && 427 ip6->ip6_src.s6_addr16[1]) { 428 ip6stat.ip6s_badscope++; 429 goto bad; 430 } 431 if (IN6_IS_SCOPE_LINKLOCAL(&ip6->ip6_dst) && 432 ip6->ip6_dst.s6_addr16[1]) { 433 ip6stat.ip6s_badscope++; 434 goto bad; 435 } 436 } 437 438 if (IN6_IS_SCOPE_LINKLOCAL(&ip6->ip6_src)) 439 ip6->ip6_src.s6_addr16[1] 440 = htons(m->m_pkthdr.rcvif->if_index); 441 if (IN6_IS_SCOPE_LINKLOCAL(&ip6->ip6_dst)) 442 ip6->ip6_dst.s6_addr16[1] 443 = htons(m->m_pkthdr.rcvif->if_index); 444 445 #if 0 /* this case seems to be unnecessary. (jinmei, 20010401) */ 446 /* 447 * We use rt->rt_ifp to determine if the address is ours or not. 448 * If rt_ifp is lo0, the address is ours. 449 * The problem here is, rt->rt_ifp for fe80::%lo0/64 is set to lo0, 450 * so any address under fe80::%lo0/64 will be mistakenly considered 451 * local. The special case is supplied to handle the case properly 452 * by actually looking at interface addresses 453 * (using in6ifa_ifpwithaddr). 454 */ 455 if ((m->m_pkthdr.rcvif->if_flags & IFF_LOOPBACK) && 456 IN6_IS_ADDR_LINKLOCAL(&ip6->ip6_dst)) { 457 if (!in6ifa_ifpwithaddr(m->m_pkthdr.rcvif, &ip6->ip6_dst)) { 458 icmp6_error(m, ICMP6_DST_UNREACH, 459 ICMP6_DST_UNREACH_ADDR, 0); 460 /* m is already freed */ 461 goto bad2; 462 } 463 464 ours = 1; 465 deliverifp = m->m_pkthdr.rcvif; 466 goto hbhcheck; 467 } 468 #endif 469 470 /* 471 * Multicast check 472 * 473 * WARNING! For general subnet proxying the interface hw will 474 * likely filter out the multicast solicitations, so 475 * the interface must be in promiscuous mode. 476 */ 477 if (IN6_IS_ADDR_MULTICAST(&ip6->ip6_dst)) { 478 ours = 1; 479 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_mcast); 480 deliverifp = m->m_pkthdr.rcvif; 481 goto hbhcheck; 482 } 483 484 /* 485 * Unicast check 486 */ 487 switch (ip6_ours_check_algorithm) { 488 default: 489 /* 490 * XXX: I intentionally broke our indentation rule here, 491 * since this switch-case is just for measurement and 492 * therefore should soon be removed. 493 */ 494 if (ip6_forward_rt.ro_rt != NULL && 495 (ip6_forward_rt.ro_rt->rt_flags & RTF_UP) && 496 IN6_ARE_ADDR_EQUAL(&ip6->ip6_dst, 497 &((struct sockaddr_in6 *)(&ip6_forward_rt.ro_dst))->sin6_addr)) 498 ip6stat.ip6s_forward_cachehit++; 499 else { 500 struct sockaddr_in6 *dst6; 501 502 if (ip6_forward_rt.ro_rt) { 503 /* route is down or destination is different */ 504 ip6stat.ip6s_forward_cachemiss++; 505 RTFREE(ip6_forward_rt.ro_rt); 506 ip6_forward_rt.ro_rt = 0; 507 } 508 509 bzero(&ip6_forward_rt.ro_dst, sizeof(struct sockaddr_in6)); 510 dst6 = &ip6_forward_rt.ro_dst; 511 dst6->sin6_len = sizeof(struct sockaddr_in6); 512 dst6->sin6_family = AF_INET6; 513 dst6->sin6_addr = ip6->ip6_dst; 514 515 rtalloc_ign((struct route *)&ip6_forward_rt, RTF_PRCLONING); 516 } 517 518 #define rt6_key(r) ((struct sockaddr_in6 *)((r)->rt_nodes->rn_key)) 519 520 /* 521 * Accept the packet if the forwarding interface to the destination 522 * according to the routing table is the loopback interface, 523 * unless the associated route has a gateway. 524 * Note that this approach causes to accept a packet if there is a 525 * route to the loopback interface for the destination of the packet. 526 * But we think it's even useful in some situations, e.g. when using 527 * a special daemon which wants to intercept the packet. 528 * 529 * XXX: some OSes automatically make a cloned route for the destination 530 * of an outgoing packet. If the outgoing interface of the packet 531 * is a loopback one, the kernel would consider the packet to be 532 * accepted, even if we have no such address assinged on the interface. 533 * We check the cloned flag of the route entry to reject such cases, 534 * assuming that route entries for our own addresses are not made by 535 * cloning (it should be true because in6_addloop explicitly installs 536 * the host route). However, we might have to do an explicit check 537 * while it would be less efficient. Or, should we rather install a 538 * reject route for such a case? 539 */ 540 if (ip6_forward_rt.ro_rt && 541 (ip6_forward_rt.ro_rt->rt_flags & 542 (RTF_HOST|RTF_GATEWAY)) == RTF_HOST && 543 #ifdef RTF_WASCLONED 544 !(ip6_forward_rt.ro_rt->rt_flags & RTF_WASCLONED) && 545 #endif 546 #ifdef RTF_CLONED 547 !(ip6_forward_rt.ro_rt->rt_flags & RTF_CLONED) && 548 #endif 549 #if 0 550 /* 551 * The check below is redundant since the comparison of 552 * the destination and the key of the rtentry has 553 * already done through looking up the routing table. 554 */ 555 IN6_ARE_ADDR_EQUAL(&ip6->ip6_dst, 556 &rt6_key(ip6_forward_rt.ro_rt)->sin6_addr) 557 #endif 558 ip6_forward_rt.ro_rt->rt_ifp->if_type == IFT_LOOP) { 559 struct in6_ifaddr *ia6 = 560 (struct in6_ifaddr *)ip6_forward_rt.ro_rt->rt_ifa; 561 562 /* 563 * record address information into m_aux. 564 */ 565 ip6_setdstifaddr(m, ia6); 566 567 /* 568 * packets to a tentative, duplicated, or somehow invalid 569 * address must not be accepted. 570 */ 571 if (!(ia6->ia6_flags & IN6_IFF_NOTREADY)) { 572 /* this address is ready */ 573 ours = 1; 574 deliverifp = ia6->ia_ifp; /* correct? */ 575 /* Count the packet in the ip address stats */ 576 IFA_STAT_INC(&ia6->ia_ifa, ipackets, 1); 577 IFA_STAT_INC(&ia6->ia_ifa, ibytes, m->m_pkthdr.len); 578 goto hbhcheck; 579 } else { 580 /* address is not ready, so discard the packet. */ 581 nd6log((LOG_INFO, 582 "ip6_input: packet to an unready address %s->%s\n", 583 ip6_sprintf(&ip6->ip6_src), 584 ip6_sprintf(&ip6->ip6_dst))); 585 586 goto bad; 587 } 588 } 589 } /* XXX indentation (see above) */ 590 591 /* 592 * FAITH(Firewall Aided Internet Translator) 593 */ 594 if (ip6_keepfaith) { 595 if (ip6_forward_rt.ro_rt && ip6_forward_rt.ro_rt->rt_ifp 596 && ip6_forward_rt.ro_rt->rt_ifp->if_type == IFT_FAITH) { 597 /* XXX do we need more sanity checks? */ 598 ours = 1; 599 deliverifp = ip6_forward_rt.ro_rt->rt_ifp; /* faith */ 600 goto hbhcheck; 601 } 602 } 603 604 /* 605 * Now there is no reason to process the packet if it's not our own 606 * and we're not a router. 607 */ 608 if (!ip6_forwarding) { 609 ip6stat.ip6s_cantforward++; 610 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_discard); 611 goto bad; 612 } 613 614 hbhcheck: 615 /* 616 * record address information into m_aux, if we don't have one yet. 617 * note that we are unable to record it, if the address is not listed 618 * as our interface address (e.g. multicast addresses, addresses 619 * within FAITH prefixes and such). 620 */ 621 if (deliverifp && !ip6_getdstifaddr(m)) { 622 struct in6_ifaddr *ia6; 623 624 ia6 = in6_ifawithifp(deliverifp, &ip6->ip6_dst); 625 if (ia6) { 626 if (!ip6_setdstifaddr(m, ia6)) { 627 /* 628 * XXX maybe we should drop the packet here, 629 * as we could not provide enough information 630 * to the upper layers. 631 */ 632 } 633 } 634 } 635 636 /* 637 * Process Hop-by-Hop options header if it's contained. 638 * m may be modified in ip6_hopopts_input(). 639 * If a JumboPayload option is included, plen will also be modified. 640 */ 641 plen = (u_int32_t)ntohs(ip6->ip6_plen); 642 if (ip6->ip6_nxt == IPPROTO_HOPOPTS) { 643 struct ip6_hbh *hbh; 644 645 if (ip6_hopopts_input(&plen, &rtalert, &m, &off)) { 646 #if 0 /*touches NULL pointer*/ 647 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_discard); 648 #endif 649 goto bad2; /* m have already been freed */ 650 } 651 652 /* adjust pointer */ 653 ip6 = mtod(m, struct ip6_hdr *); 654 655 /* 656 * if the payload length field is 0 and the next header field 657 * indicates Hop-by-Hop Options header, then a Jumbo Payload 658 * option MUST be included. 659 */ 660 if (ip6->ip6_plen == 0 && plen == 0) { 661 /* 662 * Note that if a valid jumbo payload option is 663 * contained, ip6_hoptops_input() must set a valid 664 * (non-zero) payload length to the variable plen. 665 */ 666 ip6stat.ip6s_badoptions++; 667 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_discard); 668 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_hdrerr); 669 icmp6_error(m, ICMP6_PARAM_PROB, 670 ICMP6_PARAMPROB_HEADER, 671 (caddr_t)&ip6->ip6_plen - (caddr_t)ip6); 672 goto bad2; 673 } 674 #ifndef PULLDOWN_TEST 675 /* ip6_hopopts_input() ensures that mbuf is contiguous */ 676 hbh = (struct ip6_hbh *)(ip6 + 1); 677 #else 678 IP6_EXTHDR_GET(hbh, struct ip6_hbh *, m, sizeof(struct ip6_hdr), 679 sizeof(struct ip6_hbh)); 680 if (hbh == NULL) { 681 ip6stat.ip6s_tooshort++; 682 goto bad2; 683 } 684 #endif 685 nxt = hbh->ip6h_nxt; 686 687 /* 688 * If we are acting as a router and the packet contains a 689 * router alert option, see if we know the option value. 690 * Currently, we only support the option value for MLD, in which 691 * case we should pass the packet to the multicast routing 692 * daemon. 693 */ 694 if (rtalert != ~0 && ip6_forwarding) { 695 switch (rtalert) { 696 case IP6OPT_RTALERT_MLD: 697 ours = 1; 698 break; 699 default: 700 /* 701 * RFC2711 requires unrecognized values must be 702 * silently ignored. 703 */ 704 break; 705 } 706 } 707 } else 708 nxt = ip6->ip6_nxt; 709 710 /* 711 * Check that the amount of data in the buffers 712 * is as at least much as the IPv6 header would have us expect. 713 * Trim mbufs if longer than we expect. 714 * Drop packet if shorter than we expect. 715 */ 716 if (m->m_pkthdr.len - sizeof(struct ip6_hdr) < plen) { 717 ip6stat.ip6s_tooshort++; 718 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_truncated); 719 goto bad; 720 } 721 if (m->m_pkthdr.len > sizeof(struct ip6_hdr) + plen) { 722 if (m->m_len == m->m_pkthdr.len) { 723 m->m_len = sizeof(struct ip6_hdr) + plen; 724 m->m_pkthdr.len = sizeof(struct ip6_hdr) + plen; 725 } else 726 m_adj(m, sizeof(struct ip6_hdr) + plen - m->m_pkthdr.len); 727 } 728 729 /* 730 * Forward if desirable. 731 */ 732 if (IN6_IS_ADDR_MULTICAST(&ip6->ip6_dst)) { 733 /* 734 * If we are acting as a multicast router, all 735 * incoming multicast packets are passed to the 736 * kernel-level multicast forwarding function. 737 * The packet is returned (relatively) intact; if 738 * ip6_mforward() returns a non-zero value, the packet 739 * must be discarded, else it may be accepted below. 740 */ 741 if (ip6_mrouter && ip6_mforward(ip6, m->m_pkthdr.rcvif, m)) { 742 ip6stat.ip6s_cantforward++; 743 goto bad; 744 } 745 if (!ours) 746 goto bad; 747 } else if (!ours) { 748 ip6_forward(m, srcrt); 749 goto bad2; 750 } 751 752 ip6 = mtod(m, struct ip6_hdr *); 753 754 /* 755 * Malicious party may be able to use IPv4 mapped addr to confuse 756 * tcp/udp stack and bypass security checks (act as if it was from 757 * 127.0.0.1 by using IPv6 src ::ffff:127.0.0.1). Be cautious. 758 * 759 * For SIIT end node behavior, you may want to disable the check. 760 * However, you will become vulnerable to attacks using IPv4 mapped 761 * source. 762 */ 763 if (IN6_IS_ADDR_V4MAPPED(&ip6->ip6_src) || 764 IN6_IS_ADDR_V4MAPPED(&ip6->ip6_dst)) { 765 ip6stat.ip6s_badscope++; 766 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_addrerr); 767 goto bad; 768 } 769 770 /* 771 * Tell launch routine the next header 772 */ 773 ip6stat.ip6s_delivered++; 774 in6_ifstat_inc(deliverifp, ifs6_in_deliver); 775 nest = 0; 776 777 rh_present = 0; 778 while (nxt != IPPROTO_DONE) { 779 struct protosw *sw6; 780 781 if (ip6_hdrnestlimit && (++nest > ip6_hdrnestlimit)) { 782 ip6stat.ip6s_toomanyhdr++; 783 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_hdrerr); 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 if (nxt == IPPROTO_ROUTING) { 814 if (rh_present++) { 815 in6_ifstat_inc(m->m_pkthdr.rcvif, 816 ifs6_in_hdrerr); 817 ip6stat.ip6s_badoptions++; 818 goto bad; 819 } 820 } 821 822 sw6 = &inet6sw[ip6_protox[nxt]]; 823 #ifdef IPSEC 824 /* 825 * enforce IPsec policy checking if we are seeing last header. 826 * note that we do not visit this with protocols with pcb layer 827 * code - like udp/tcp/raw ip. 828 */ 829 if ((sw6->pr_flags & PR_LASTHDR) && ipsec6_in_reject(m, NULL)) { 830 ipsec6stat.in_polvio++; 831 goto bad; 832 } 833 #endif 834 /* 835 * If this is a terminal header forward to the port, otherwise 836 * process synchronously for more headers. 837 */ 838 if (sw6->pr_flags & PR_LASTHDR) { 839 struct netmsg_packet *pmsg; 840 lwkt_port_t port; 841 842 port = netisr_cpuport(0); /* XXX */ 843 KKASSERT(port != NULL); 844 pmsg = &m->m_hdr.mh_netmsg; 845 netmsg_init(&pmsg->base, NULL, 846 &netisr_apanic_rport, 847 0, transport6_processing_handler); 848 pmsg->nm_packet = m; 849 pmsg->nm_nxt = nxt; 850 pmsg->base.lmsg.u.ms_result = off; 851 lwkt_sendmsg(port, &pmsg->base.lmsg); 852 /* done with m */ 853 nxt = IPPROTO_DONE; 854 } else { 855 nxt = sw6->pr_input(&m, &off, nxt); 856 } 857 } 858 goto bad2; 859 bad: 860 m_freem(m); 861 bad2: 862 ; 863 /* msg was embedded in the mbuf, do not reply! */ 864 } 865 866 /* 867 * We have to call the pr_input() function from the correct protocol 868 * thread. The sw6->pr_soport() request at the end of ip6_input() 869 * returns the port and we forward a netmsg to the port to execute 870 * this function. 871 */ 872 static void 873 transport6_processing_handler(netmsg_t netmsg) 874 { 875 struct netmsg_packet *pmsg = (struct netmsg_packet *)netmsg; 876 struct protosw *sw6; 877 int hlen; 878 int nxt; 879 880 sw6 = &inet6sw[ip6_protox[pmsg->nm_nxt]]; 881 hlen = pmsg->base.lmsg.u.ms_result; 882 883 nxt = sw6->pr_input(&pmsg->nm_packet, &hlen, pmsg->nm_nxt); 884 KKASSERT(nxt == IPPROTO_DONE); 885 /* netmsg was embedded in the mbuf, do not reply! */ 886 } 887 888 /* 889 * set/grab in6_ifaddr correspond to IPv6 destination address. 890 * XXX backward compatibility wrapper 891 */ 892 static struct ip6aux * 893 ip6_setdstifaddr(struct mbuf *m, struct in6_ifaddr *ia6) 894 { 895 struct ip6aux *n; 896 897 n = ip6_addaux(m); 898 if (n) 899 n->ip6a_dstia6 = ia6; 900 return n; /* NULL if failed to set */ 901 } 902 903 struct in6_ifaddr * 904 ip6_getdstifaddr(struct mbuf *m) 905 { 906 struct ip6aux *n; 907 908 n = ip6_findaux(m); 909 if (n) 910 return n->ip6a_dstia6; 911 else 912 return NULL; 913 } 914 915 /* 916 * Hop-by-Hop options header processing. If a valid jumbo payload option is 917 * included, the real payload length will be stored in plenp. 918 */ 919 static int 920 ip6_hopopts_input(u_int32_t *plenp, 921 u_int32_t *rtalertp,/* XXX: should be stored more smart way */ 922 struct mbuf **mp, 923 int *offp) 924 { 925 struct mbuf *m = *mp; 926 int off = *offp, hbhlen; 927 struct ip6_hbh *hbh; 928 929 /* validation of the length of the header */ 930 #ifndef PULLDOWN_TEST 931 IP6_EXTHDR_CHECK(m, off, sizeof(*hbh), -1); 932 hbh = (struct ip6_hbh *)(mtod(m, caddr_t) + off); 933 hbhlen = (hbh->ip6h_len + 1) << 3; 934 935 IP6_EXTHDR_CHECK(m, off, hbhlen, -1); 936 hbh = (struct ip6_hbh *)(mtod(m, caddr_t) + off); 937 #else 938 IP6_EXTHDR_GET(hbh, struct ip6_hbh *, m, 939 sizeof(struct ip6_hdr), sizeof(struct ip6_hbh)); 940 if (hbh == NULL) { 941 ip6stat.ip6s_tooshort++; 942 return -1; 943 } 944 hbhlen = (hbh->ip6h_len + 1) << 3; 945 IP6_EXTHDR_GET(hbh, struct ip6_hbh *, m, sizeof(struct ip6_hdr), 946 hbhlen); 947 if (hbh == NULL) { 948 ip6stat.ip6s_tooshort++; 949 return -1; 950 } 951 #endif 952 off += hbhlen; 953 hbhlen -= sizeof(struct ip6_hbh); 954 955 if (ip6_process_hopopts(m, (u_int8_t *)hbh + sizeof(struct ip6_hbh), 956 hbhlen, rtalertp, plenp) < 0) 957 return (-1); 958 959 *offp = off; 960 *mp = m; 961 return (0); 962 } 963 964 /* 965 * Search header for all Hop-by-hop options and process each option. 966 * This function is separate from ip6_hopopts_input() in order to 967 * handle a case where the sending node itself process its hop-by-hop 968 * options header. In such a case, the function is called from ip6_output(). 969 * 970 * The function assumes that hbh header is located right after the IPv6 header 971 * (RFC2460 p7), opthead is pointer into data content in m, and opthead to 972 * opthead + hbhlen is located in continuous memory region. 973 */ 974 int 975 ip6_process_hopopts(struct mbuf *m, u_int8_t *opthead, int hbhlen, 976 u_int32_t *rtalertp, u_int32_t *plenp) 977 { 978 struct ip6_hdr *ip6; 979 int optlen = 0; 980 u_int8_t *opt = opthead; 981 u_int16_t rtalert_val; 982 u_int32_t jumboplen; 983 const int erroff = sizeof(struct ip6_hdr) + sizeof(struct ip6_hbh); 984 985 for (; hbhlen > 0; hbhlen -= optlen, opt += optlen) { 986 switch (*opt) { 987 case IP6OPT_PAD1: 988 optlen = 1; 989 break; 990 case IP6OPT_PADN: 991 if (hbhlen < IP6OPT_MINLEN) { 992 ip6stat.ip6s_toosmall++; 993 goto bad; 994 } 995 optlen = *(opt + 1) + 2; 996 break; 997 case IP6OPT_RTALERT: 998 /* XXX may need check for alignment */ 999 if (hbhlen < IP6OPT_RTALERT_LEN) { 1000 ip6stat.ip6s_toosmall++; 1001 goto bad; 1002 } 1003 if (*(opt + 1) != IP6OPT_RTALERT_LEN - 2) { 1004 /* XXX stat */ 1005 icmp6_error(m, ICMP6_PARAM_PROB, 1006 ICMP6_PARAMPROB_HEADER, 1007 erroff + opt + 1 - opthead); 1008 return (-1); 1009 } 1010 optlen = IP6OPT_RTALERT_LEN; 1011 bcopy((caddr_t)(opt + 2), (caddr_t)&rtalert_val, 2); 1012 *rtalertp = ntohs(rtalert_val); 1013 break; 1014 case IP6OPT_JUMBO: 1015 /* XXX may need check for alignment */ 1016 if (hbhlen < IP6OPT_JUMBO_LEN) { 1017 ip6stat.ip6s_toosmall++; 1018 goto bad; 1019 } 1020 if (*(opt + 1) != IP6OPT_JUMBO_LEN - 2) { 1021 /* XXX stat */ 1022 icmp6_error(m, ICMP6_PARAM_PROB, 1023 ICMP6_PARAMPROB_HEADER, 1024 erroff + opt + 1 - opthead); 1025 return (-1); 1026 } 1027 optlen = IP6OPT_JUMBO_LEN; 1028 1029 /* 1030 * IPv6 packets that have non 0 payload length 1031 * must not contain a jumbo payload option. 1032 */ 1033 ip6 = mtod(m, struct ip6_hdr *); 1034 if (ip6->ip6_plen) { 1035 ip6stat.ip6s_badoptions++; 1036 icmp6_error(m, ICMP6_PARAM_PROB, 1037 ICMP6_PARAMPROB_HEADER, 1038 erroff + opt - opthead); 1039 return (-1); 1040 } 1041 1042 /* 1043 * We may see jumbolen in unaligned location, so 1044 * we'd need to perform bcopy(). 1045 */ 1046 bcopy(opt + 2, &jumboplen, sizeof(jumboplen)); 1047 jumboplen = (u_int32_t)htonl(jumboplen); 1048 1049 #if 1 1050 /* 1051 * if there are multiple jumbo payload options, 1052 * *plenp will be non-zero and the packet will be 1053 * rejected. 1054 * the behavior may need some debate in ipngwg - 1055 * multiple options does not make sense, however, 1056 * there's no explicit mention in specification. 1057 */ 1058 if (*plenp != 0) { 1059 ip6stat.ip6s_badoptions++; 1060 icmp6_error(m, ICMP6_PARAM_PROB, 1061 ICMP6_PARAMPROB_HEADER, 1062 erroff + opt + 2 - opthead); 1063 return (-1); 1064 } 1065 #endif 1066 1067 /* 1068 * jumbo payload length must be larger than 65535. 1069 */ 1070 if (jumboplen <= IPV6_MAXPACKET) { 1071 ip6stat.ip6s_badoptions++; 1072 icmp6_error(m, ICMP6_PARAM_PROB, 1073 ICMP6_PARAMPROB_HEADER, 1074 erroff + opt + 2 - opthead); 1075 return (-1); 1076 } 1077 *plenp = jumboplen; 1078 1079 break; 1080 default: /* unknown option */ 1081 if (hbhlen < IP6OPT_MINLEN) { 1082 ip6stat.ip6s_toosmall++; 1083 goto bad; 1084 } 1085 optlen = ip6_unknown_opt(opt, m, 1086 erroff + opt - opthead); 1087 if (optlen == -1) 1088 return (-1); 1089 optlen += 2; 1090 break; 1091 } 1092 } 1093 1094 return (0); 1095 1096 bad: 1097 m_freem(m); 1098 return (-1); 1099 } 1100 1101 /* 1102 * Unknown option processing. 1103 * The third argument `off' is the offset from the IPv6 header to the option, 1104 * which is necessary if the IPv6 header the and option header and IPv6 header 1105 * is not continuous in order to return an ICMPv6 error. 1106 */ 1107 int 1108 ip6_unknown_opt(u_int8_t *optp, struct mbuf *m, int off) 1109 { 1110 struct ip6_hdr *ip6; 1111 1112 switch (IP6OPT_TYPE(*optp)) { 1113 case IP6OPT_TYPE_SKIP: /* ignore the option */ 1114 return ((int)*(optp + 1)); 1115 case IP6OPT_TYPE_DISCARD: /* silently discard */ 1116 m_freem(m); 1117 return (-1); 1118 case IP6OPT_TYPE_FORCEICMP: /* send ICMP even if multicasted */ 1119 ip6stat.ip6s_badoptions++; 1120 icmp6_error(m, ICMP6_PARAM_PROB, ICMP6_PARAMPROB_OPTION, off); 1121 return (-1); 1122 case IP6OPT_TYPE_ICMP: /* send ICMP if not multicasted */ 1123 ip6stat.ip6s_badoptions++; 1124 ip6 = mtod(m, struct ip6_hdr *); 1125 if (IN6_IS_ADDR_MULTICAST(&ip6->ip6_dst) || 1126 (m->m_flags & (M_BCAST|M_MCAST))) 1127 m_freem(m); 1128 else 1129 icmp6_error(m, ICMP6_PARAM_PROB, 1130 ICMP6_PARAMPROB_OPTION, off); 1131 return (-1); 1132 } 1133 1134 m_freem(m); /* XXX: NOTREACHED */ 1135 return (-1); 1136 } 1137 1138 /* 1139 * Create the "control" list for this pcb. 1140 * The function will not modify mbuf chain at all. 1141 * 1142 * with KAME mbuf chain restriction: 1143 * The routine will be called from upper layer handlers like tcp6_input(). 1144 * Thus the routine assumes that the caller (tcp6_input) have already 1145 * called IP6_EXTHDR_CHECK() and all the extension headers are located in the 1146 * very first mbuf on the mbuf chain. 1147 */ 1148 void 1149 ip6_savecontrol(struct inpcb *in6p, struct mbuf **mp, struct ip6_hdr *ip6, 1150 struct mbuf *m) 1151 { 1152 #define IS2292(x, y) ((in6p->in6p_flags & IN6P_RFC2292) ? (x) : (y)) 1153 struct thread *td = curthread; /* XXX */ 1154 int privileged = 0; 1155 1156 1157 if (priv_check(td, PRIV_ROOT) == 0) 1158 privileged++; 1159 1160 #ifdef SO_TIMESTAMP 1161 if (in6p->in6p_socket->so_options & SO_TIMESTAMP) { 1162 struct timeval tv; 1163 1164 microtime(&tv); 1165 *mp = sbcreatecontrol((caddr_t) &tv, sizeof(tv), 1166 SCM_TIMESTAMP, SOL_SOCKET); 1167 if (*mp) { 1168 mp = &(*mp)->m_next; 1169 } 1170 } 1171 #endif 1172 1173 /* RFC 2292 sec. 5 */ 1174 if (in6p->in6p_flags & IN6P_PKTINFO) { 1175 struct in6_pktinfo pi6; 1176 bcopy(&ip6->ip6_dst, &pi6.ipi6_addr, sizeof(struct in6_addr)); 1177 if (IN6_IS_SCOPE_LINKLOCAL(&pi6.ipi6_addr)) 1178 pi6.ipi6_addr.s6_addr16[1] = 0; 1179 pi6.ipi6_ifindex = (m && m->m_pkthdr.rcvif) 1180 ? m->m_pkthdr.rcvif->if_index 1181 : 0; 1182 *mp = sbcreatecontrol((caddr_t) &pi6, 1183 sizeof(struct in6_pktinfo), IS2292(IPV6_2292PKTINFO, IPV6_PKTINFO), 1184 IPPROTO_IPV6); 1185 if (*mp) 1186 mp = &(*mp)->m_next; 1187 } 1188 1189 if (in6p->in6p_flags & IN6P_HOPLIMIT) { 1190 int hlim = ip6->ip6_hlim & 0xff; 1191 *mp = sbcreatecontrol((caddr_t) &hlim, 1192 sizeof(int), IS2292(IPV6_2292HOPLIMIT, IPV6_HOPLIMIT), IPPROTO_IPV6); 1193 if (*mp) 1194 mp = &(*mp)->m_next; 1195 } 1196 1197 if ((in6p->in6p_flags & IN6P_TCLASS) != 0) { 1198 u_int32_t flowinfo; 1199 int tclass; 1200 1201 flowinfo = (u_int32_t)ntohl(ip6->ip6_flow & IPV6_FLOWINFO_MASK); 1202 flowinfo >>= 20; 1203 1204 tclass = flowinfo & 0xff; 1205 *mp = sbcreatecontrol((caddr_t) &tclass, sizeof(tclass), 1206 IPV6_TCLASS, IPPROTO_IPV6); 1207 if (*mp) 1208 mp = &(*mp)->m_next; 1209 } 1210 1211 /* 1212 * IPV6_HOPOPTS socket option. We require super-user privilege 1213 * for the option, but it might be too strict, since there might 1214 * be some hop-by-hop options which can be returned to normal user. 1215 * See RFC 2292 section 6. 1216 */ 1217 if ((in6p->in6p_flags & IN6P_HOPOPTS) && privileged) { 1218 /* 1219 * Check if a hop-by-hop options header is contatined in the 1220 * received packet, and if so, store the options as ancillary 1221 * data. Note that a hop-by-hop options header must be 1222 * just after the IPv6 header, which fact is assured through 1223 * the IPv6 input processing. 1224 */ 1225 struct ip6_hdr *ip6 = mtod(m, struct ip6_hdr *); 1226 if (ip6->ip6_nxt == IPPROTO_HOPOPTS) { 1227 struct ip6_hbh *hbh; 1228 int hbhlen = 0; 1229 #ifdef PULLDOWN_TEST 1230 struct mbuf *ext; 1231 #endif 1232 1233 #ifndef PULLDOWN_TEST 1234 hbh = (struct ip6_hbh *)(ip6 + 1); 1235 hbhlen = (hbh->ip6h_len + 1) << 3; 1236 #else 1237 ext = ip6_pullexthdr(m, sizeof(struct ip6_hdr), 1238 ip6->ip6_nxt); 1239 if (ext == NULL) { 1240 ip6stat.ip6s_tooshort++; 1241 return; 1242 } 1243 hbh = mtod(ext, struct ip6_hbh *); 1244 hbhlen = (hbh->ip6h_len + 1) << 3; 1245 if (hbhlen != ext->m_len) { 1246 m_freem(ext); 1247 ip6stat.ip6s_tooshort++; 1248 return; 1249 } 1250 #endif 1251 1252 /* 1253 * XXX: We copy whole the header even if a jumbo 1254 * payload option is included, which option is to 1255 * be removed before returning in the RFC 2292. 1256 * Note: this constraint is removed in 2292bis. 1257 */ 1258 *mp = sbcreatecontrol((caddr_t)hbh, hbhlen, 1259 IS2292(IPV6_2292HOPOPTS, IPV6_HOPOPTS), IPPROTO_IPV6); 1260 if (*mp) 1261 mp = &(*mp)->m_next; 1262 #ifdef PULLDOWN_TEST 1263 m_freem(ext); 1264 #endif 1265 } 1266 } 1267 1268 /* IPV6_DSTOPTS and IPV6_RTHDR socket options */ 1269 if ((in6p->in6p_flags & (IN6P_DSTOPTS | IN6P_RTHDRDSTOPTS)) != 0) { 1270 int proto, off, nxt; 1271 1272 /* 1273 * go through the header chain to see if a routing header is 1274 * contained in the packet. We need this information to store 1275 * destination options headers (if any) properly. 1276 * XXX: performance issue. We should record this info when 1277 * processing extension headers in incoming routine. 1278 * (todo) use m_aux? 1279 */ 1280 proto = IPPROTO_IPV6; 1281 off = 0; 1282 nxt = -1; 1283 while (1) { 1284 int newoff; 1285 1286 newoff = ip6_nexthdr(m, off, proto, &nxt); 1287 if (newoff < 0) 1288 break; 1289 if (newoff < off) /* invalid, check for safety */ 1290 break; 1291 if ((proto = nxt) == IPPROTO_ROUTING) 1292 break; 1293 off = newoff; 1294 } 1295 } 1296 1297 if ((in6p->in6p_flags & 1298 (IN6P_RTHDR | IN6P_DSTOPTS | IN6P_RTHDRDSTOPTS)) != 0) { 1299 struct ip6_hdr *ip6 = mtod(m, struct ip6_hdr *); 1300 int nxt = ip6->ip6_nxt, off = sizeof(struct ip6_hdr); 1301 1302 /* 1303 * Search for destination options headers or routing 1304 * header(s) through the header chain, and stores each 1305 * header as ancillary data. 1306 * Note that the order of the headers remains in 1307 * the chain of ancillary data. 1308 */ 1309 while (1) { /* is explicit loop prevention necessary? */ 1310 struct ip6_ext *ip6e = NULL; 1311 int elen; 1312 #ifdef PULLDOWN_TEST 1313 struct mbuf *ext = NULL; 1314 #endif 1315 1316 /* 1317 * if it is not an extension header, don't try to 1318 * pull it from the chain. 1319 */ 1320 switch (nxt) { 1321 case IPPROTO_DSTOPTS: 1322 case IPPROTO_ROUTING: 1323 case IPPROTO_HOPOPTS: 1324 case IPPROTO_AH: /* is it possible? */ 1325 break; 1326 default: 1327 goto loopend; 1328 } 1329 1330 #ifndef PULLDOWN_TEST 1331 if (off + sizeof(*ip6e) > m->m_len) 1332 goto loopend; 1333 ip6e = (struct ip6_ext *)(mtod(m, caddr_t) + off); 1334 if (nxt == IPPROTO_AH) 1335 elen = (ip6e->ip6e_len + 2) << 2; 1336 else 1337 elen = (ip6e->ip6e_len + 1) << 3; 1338 if (off + elen > m->m_len) 1339 goto loopend; 1340 #else 1341 ext = ip6_pullexthdr(m, off, nxt); 1342 if (ext == NULL) { 1343 ip6stat.ip6s_tooshort++; 1344 return; 1345 } 1346 ip6e = mtod(ext, struct ip6_ext *); 1347 if (nxt == IPPROTO_AH) 1348 elen = (ip6e->ip6e_len + 2) << 2; 1349 else 1350 elen = (ip6e->ip6e_len + 1) << 3; 1351 if (elen != ext->m_len) { 1352 m_freem(ext); 1353 ip6stat.ip6s_tooshort++; 1354 return; 1355 } 1356 #endif 1357 1358 switch (nxt) { 1359 case IPPROTO_DSTOPTS: 1360 if ((in6p->in6p_flags & IN6P_DSTOPTS) == 0) 1361 break; 1362 1363 /* 1364 * We also require super-user privilege for 1365 * the option. 1366 * See the comments on IN6_HOPOPTS. 1367 */ 1368 if (!privileged) 1369 break; 1370 1371 *mp = sbcreatecontrol((caddr_t)ip6e, elen, 1372 IS2292(IPV6_2292DSTOPTS, IPV6_DSTOPTS), 1373 IPPROTO_IPV6); 1374 if (*mp) 1375 mp = &(*mp)->m_next; 1376 break; 1377 case IPPROTO_ROUTING: 1378 if (!(in6p->in6p_flags & IN6P_RTHDR)) 1379 break; 1380 1381 *mp = sbcreatecontrol((caddr_t)ip6e, elen, 1382 IS2292(IPV6_2292RTHDR, IPV6_RTHDR), 1383 IPPROTO_IPV6); 1384 if (*mp) 1385 mp = &(*mp)->m_next; 1386 break; 1387 case IPPROTO_HOPOPTS: 1388 case IPPROTO_AH: /* is it possible? */ 1389 break; 1390 1391 default: 1392 /* 1393 * other cases have been filtered in the above. 1394 * none will visit this case. here we supply 1395 * the code just in case (nxt overwritten or 1396 * other cases). 1397 */ 1398 #ifdef PULLDOWN_TEST 1399 m_freem(ext); 1400 #endif 1401 goto loopend; 1402 1403 } 1404 1405 /* proceed with the next header. */ 1406 off += elen; 1407 nxt = ip6e->ip6e_nxt; 1408 ip6e = NULL; 1409 #ifdef PULLDOWN_TEST 1410 m_freem(ext); 1411 ext = NULL; 1412 #endif 1413 } 1414 loopend: 1415 ; 1416 } 1417 #undef IS2292 1418 } 1419 1420 void 1421 ip6_notify_pmtu(struct inpcb *in6p, struct sockaddr_in6 *dst, u_int32_t *mtu) 1422 { 1423 struct socket *so; 1424 struct mbuf *m_mtu; 1425 struct ip6_mtuinfo mtuctl; 1426 1427 so = in6p->inp_socket; 1428 1429 if (mtu == NULL) 1430 return; 1431 1432 #ifdef DIAGNOSTIC 1433 if (so == NULL) /* I believe this is impossible */ 1434 panic("ip6_notify_pmtu: socket is NULL"); 1435 #endif 1436 1437 bzero(&mtuctl, sizeof(mtuctl)); /* zero-clear for safety */ 1438 mtuctl.ip6m_mtu = *mtu; 1439 mtuctl.ip6m_addr = *dst; 1440 1441 if ((m_mtu = sbcreatecontrol((caddr_t)&mtuctl, sizeof(mtuctl), 1442 IPV6_PATHMTU, IPPROTO_IPV6)) == NULL) 1443 return; 1444 1445 lwkt_gettoken(&so->so_rcv.ssb_token); 1446 if (sbappendaddr(&so->so_rcv.sb, (struct sockaddr *)dst, NULL, m_mtu) 1447 == 0) { 1448 m_freem(m_mtu); 1449 /* XXX: should count statistics */ 1450 } else { 1451 sorwakeup(so); 1452 } 1453 lwkt_reltoken(&so->so_rcv.ssb_token); 1454 } 1455 1456 #ifdef PULLDOWN_TEST 1457 /* 1458 * pull single extension header from mbuf chain. returns single mbuf that 1459 * contains the result, or NULL on error. 1460 */ 1461 static struct mbuf * 1462 ip6_pullexthdr(struct mbuf *m, size_t off, int nxt) 1463 { 1464 struct ip6_ext ip6e; 1465 size_t elen; 1466 struct mbuf *n; 1467 1468 #ifdef DIAGNOSTIC 1469 switch (nxt) { 1470 case IPPROTO_DSTOPTS: 1471 case IPPROTO_ROUTING: 1472 case IPPROTO_HOPOPTS: 1473 case IPPROTO_AH: /* is it possible? */ 1474 break; 1475 default: 1476 kprintf("ip6_pullexthdr: invalid nxt=%d\n", nxt); 1477 } 1478 #endif 1479 1480 m_copydata(m, off, sizeof(ip6e), (caddr_t)&ip6e); 1481 if (nxt == IPPROTO_AH) 1482 elen = (ip6e.ip6e_len + 2) << 2; 1483 else 1484 elen = (ip6e.ip6e_len + 1) << 3; 1485 1486 n = m_getb(elen, MB_DONTWAIT, MT_DATA, 0); 1487 if (n == NULL) 1488 return NULL; 1489 n->m_len = 0; 1490 1491 if (elen >= M_TRAILINGSPACE(n)) { 1492 m_free(n); 1493 return NULL; 1494 } 1495 1496 m_copydata(m, off, elen, mtod(n, caddr_t)); 1497 n->m_len = elen; 1498 return n; 1499 } 1500 #endif 1501 1502 /* 1503 * Get pointer to the previous header followed by the header 1504 * currently processed. 1505 * XXX: This function supposes that 1506 * M includes all headers, 1507 * the next header field and the header length field of each header 1508 * are valid, and 1509 * the sum of each header length equals to OFF. 1510 * Because of these assumptions, this function must be called very 1511 * carefully. Moreover, it will not be used in the near future when 1512 * we develop `neater' mechanism to process extension headers. 1513 */ 1514 char * 1515 ip6_get_prevhdr(struct mbuf *m, int off) 1516 { 1517 struct ip6_hdr *ip6 = mtod(m, struct ip6_hdr *); 1518 1519 if (off == sizeof(struct ip6_hdr)) 1520 return (&ip6->ip6_nxt); 1521 else { 1522 int len, nxt; 1523 struct ip6_ext *ip6e = NULL; 1524 1525 nxt = ip6->ip6_nxt; 1526 len = sizeof(struct ip6_hdr); 1527 while (len < off) { 1528 ip6e = (struct ip6_ext *)(mtod(m, caddr_t) + len); 1529 1530 switch (nxt) { 1531 case IPPROTO_FRAGMENT: 1532 len += sizeof(struct ip6_frag); 1533 break; 1534 case IPPROTO_AH: 1535 len += (ip6e->ip6e_len + 2) << 2; 1536 break; 1537 default: 1538 len += (ip6e->ip6e_len + 1) << 3; 1539 break; 1540 } 1541 nxt = ip6e->ip6e_nxt; 1542 } 1543 if (ip6e) 1544 return (&ip6e->ip6e_nxt); 1545 else 1546 return NULL; 1547 } 1548 } 1549 1550 /* 1551 * get next header offset. m will be retained. 1552 */ 1553 int 1554 ip6_nexthdr(struct mbuf *m, int off, int proto, int *nxtp) 1555 { 1556 struct ip6_hdr ip6; 1557 struct ip6_ext ip6e; 1558 struct ip6_frag fh; 1559 1560 /* just in case */ 1561 if (m == NULL) 1562 panic("ip6_nexthdr: m == NULL"); 1563 if ((m->m_flags & M_PKTHDR) == 0 || m->m_pkthdr.len < off) 1564 return -1; 1565 1566 switch (proto) { 1567 case IPPROTO_IPV6: 1568 if (m->m_pkthdr.len < off + sizeof(ip6)) 1569 return -1; 1570 m_copydata(m, off, sizeof(ip6), (caddr_t)&ip6); 1571 if (nxtp) 1572 *nxtp = ip6.ip6_nxt; 1573 off += sizeof(ip6); 1574 return off; 1575 1576 case IPPROTO_FRAGMENT: 1577 /* 1578 * terminate parsing if it is not the first fragment, 1579 * it does not make sense to parse through it. 1580 */ 1581 if (m->m_pkthdr.len < off + sizeof(fh)) 1582 return -1; 1583 m_copydata(m, off, sizeof(fh), (caddr_t)&fh); 1584 /* IP6F_OFF_MASK = 0xfff8(BigEndian), 0xf8ff(LittleEndian) */ 1585 if (fh.ip6f_offlg & IP6F_OFF_MASK) 1586 return -1; 1587 if (nxtp) 1588 *nxtp = fh.ip6f_nxt; 1589 off += sizeof(struct ip6_frag); 1590 return off; 1591 1592 case IPPROTO_AH: 1593 if (m->m_pkthdr.len < off + sizeof(ip6e)) 1594 return -1; 1595 m_copydata(m, off, sizeof(ip6e), (caddr_t)&ip6e); 1596 if (nxtp) 1597 *nxtp = ip6e.ip6e_nxt; 1598 off += (ip6e.ip6e_len + 2) << 2; 1599 return off; 1600 1601 case IPPROTO_HOPOPTS: 1602 case IPPROTO_ROUTING: 1603 case IPPROTO_DSTOPTS: 1604 if (m->m_pkthdr.len < off + sizeof(ip6e)) 1605 return -1; 1606 m_copydata(m, off, sizeof(ip6e), (caddr_t)&ip6e); 1607 if (nxtp) 1608 *nxtp = ip6e.ip6e_nxt; 1609 off += (ip6e.ip6e_len + 1) << 3; 1610 return off; 1611 1612 case IPPROTO_NONE: 1613 case IPPROTO_ESP: 1614 case IPPROTO_IPCOMP: 1615 /* give up */ 1616 return -1; 1617 1618 default: 1619 return -1; 1620 } 1621 1622 return -1; 1623 } 1624 1625 /* 1626 * get offset for the last header in the chain. m will be kept untainted. 1627 */ 1628 int 1629 ip6_lasthdr(struct mbuf *m, int off, int proto, int *nxtp) 1630 { 1631 int newoff; 1632 int nxt; 1633 1634 if (!nxtp) { 1635 nxt = -1; 1636 nxtp = &nxt; 1637 } 1638 while (1) { 1639 newoff = ip6_nexthdr(m, off, proto, nxtp); 1640 if (newoff < 0) 1641 return off; 1642 else if (newoff < off) 1643 return -1; /* invalid */ 1644 else if (newoff == off) 1645 return newoff; 1646 1647 off = newoff; 1648 proto = *nxtp; 1649 } 1650 } 1651 1652 struct ip6aux * 1653 ip6_addaux(struct mbuf *m) 1654 { 1655 struct m_tag *tag = m_tag_find(m, PACKET_TAG_IPV6_INPUT, NULL); 1656 if (!tag) { 1657 tag = m_tag_get(PACKET_TAG_IPV6_INPUT, 1658 sizeof (struct ip6aux), 1659 MB_DONTWAIT); 1660 if (tag) 1661 m_tag_prepend(m, tag); 1662 } 1663 if (tag) 1664 bzero(m_tag_data(tag), sizeof (struct ip6aux)); 1665 return tag ? (struct ip6aux *)m_tag_data(tag) : NULL; 1666 } 1667 1668 struct ip6aux * 1669 ip6_findaux(struct mbuf *m) 1670 { 1671 struct m_tag *tag = m_tag_find(m, PACKET_TAG_IPV6_INPUT, NULL); 1672 return tag ? (struct ip6aux*)(tag+1) : NULL; 1673 } 1674 1675 void 1676 ip6_delaux(struct mbuf *m) 1677 { 1678 struct m_tag *tag = m_tag_find(m, PACKET_TAG_IPV6_INPUT, NULL); 1679 if (tag) 1680 m_tag_delete(m, tag); 1681 } 1682 1683 /* 1684 * System control for IP6 1685 */ 1686 1687 u_char inet6ctlerrmap[PRC_NCMDS] = { 1688 0, 0, 0, 0, 1689 0, EMSGSIZE, EHOSTDOWN, EHOSTUNREACH, 1690 EHOSTUNREACH, EHOSTUNREACH, ECONNREFUSED, ECONNREFUSED, 1691 EMSGSIZE, EHOSTUNREACH, 0, 0, 1692 0, 0, 0, 0, 1693 ENOPROTOOPT 1694 }; 1695