Now available: The Design and Implementation of the FreeBSD Operating System (Second Edition)
[ source navigation ] [ diff markup ] [ identifier search ] [ freetext search ] [ file search ] [ list types ] [ track identifier ]
FreeBSD/Linux Kernel Cross Reference
sys/netinet6/ip6_input.c
Version:
- FREEBSD - FREEBSD-13-STABLE - FREEBSD-13-0 - FREEBSD-12-STABLE - FREEBSD-12-0 - FREEBSD-11-STABLE - FREEBSD-11-0 - FREEBSD-10-STABLE - FREEBSD-10-0 - FREEBSD-9-STABLE - FREEBSD-9-0 - FREEBSD-8-STABLE - FREEBSD-8-0 - FREEBSD-7-STABLE - FREEBSD-7-0 - FREEBSD-6-STABLE - FREEBSD-6-0 - FREEBSD-5-STABLE - FREEBSD-5-0 - FREEBSD-4-STABLE - FREEBSD-3-STABLE - FREEBSD22 - l41 - OPENBSD - linux-2.6 - MK84 - PLAN9 - xnu-8792
SearchContext: - none - 3 - 10
SearchContext: - none - 3 - 10
1 /*- 2 * Copyright (C) 1995, 1996, 1997, and 1998 WIDE Project. 3 * All rights reserved. 4 * 5 * Redistribution and use in source and binary forms, with or without 6 * modification, are permitted provided that the following conditions 7 * are met: 8 * 1. Redistributions of source code must retain the above copyright 9 * notice, this list of conditions and the following disclaimer. 10 * 2. Redistributions in binary form must reproduce the above copyright 11 * notice, this list of conditions and the following disclaimer in the 12 * documentation and/or other materials provided with the distribution. 13 * 3. Neither the name of the project nor the names of its contributors 14 * may be used to endorse or promote products derived from this software 15 * without specific prior written permission. 16 * 17 * THIS SOFTWARE IS PROVIDED BY THE PROJECT AND CONTRIBUTORS ``AS IS'' AND 18 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE 19 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE 20 * ARE DISCLAIMED. IN NO EVENT SHALL THE PROJECT OR CONTRIBUTORS BE LIABLE 21 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL 22 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS 23 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) 24 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT 25 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY 26 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF 27 * SUCH DAMAGE. 28 * 29 * $KAME: ip6_input.c,v 1.259 2002/01/21 04:58:09 jinmei Exp $ 30 */ 31 32 /*- 33 * Copyright (c) 1982, 1986, 1988, 1993 34 * The Regents of the University of California. All rights reserved. 35 * 36 * Redistribution and use in source and binary forms, with or without 37 * modification, are permitted provided that the following conditions 38 * are met: 39 * 1. Redistributions of source code must retain the above copyright 40 * notice, this list of conditions and the following disclaimer. 41 * 2. Redistributions in binary form must reproduce the above copyright 42 * notice, this list of conditions and the following disclaimer in the 43 * documentation and/or other materials provided with the distribution. 44 * 4. Neither the name of the University nor the names of its contributors 45 * may be used to endorse or promote products derived from this software 46 * without specific prior written permission. 47 * 48 * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND 49 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE 50 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE 51 * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE 52 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL 53 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS 54 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) 55 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT 56 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY 57 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF 58 * SUCH DAMAGE. 59 * 60 * @(#)ip_input.c 8.2 (Berkeley) 1/4/94 61 */ 62 63 #include <sys/cdefs.h> 64 __FBSDID("$FreeBSD: releng/8.3/sys/netinet6/ip6_input.c 236953 2012-06-12 12:10:10Z bz $"); 65 66 #include "opt_inet.h" 67 #include "opt_inet6.h" 68 #include "opt_ipsec.h" 69 70 #include <sys/param.h> 71 #include <sys/systm.h> 72 #include <sys/malloc.h> 73 #include <sys/mbuf.h> 74 #include <sys/proc.h> 75 #include <sys/domain.h> 76 #include <sys/protosw.h> 77 #include <sys/socket.h> 78 #include <sys/socketvar.h> 79 #include <sys/errno.h> 80 #include <sys/time.h> 81 #include <sys/kernel.h> 82 #include <sys/syslog.h> 83 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 #include <net/vnet.h> 91 92 #include <netinet/in.h> 93 #include <netinet/in_systm.h> 94 #include <net/if_llatbl.h> 95 #ifdef INET 96 #include <netinet/ip.h> 97 #include <netinet/ip_icmp.h> 98 #endif /* INET */ 99 #include <netinet/ip6.h> 100 #include <netinet6/in6_var.h> 101 #include <netinet6/ip6_var.h> 102 #include <netinet/in_pcb.h> 103 #include <netinet/icmp6.h> 104 #include <netinet6/scope6_var.h> 105 #include <netinet6/in6_ifattach.h> 106 #include <netinet6/nd6.h> 107 108 #ifdef IPSEC 109 #include <netipsec/ipsec.h> 110 #include <netinet6/ip6_ipsec.h> 111 #include <netipsec/ipsec6.h> 112 #endif /* IPSEC */ 113 114 #include <netinet6/ip6protosw.h> 115 116 extern struct domain inet6domain; 117 118 u_char ip6_protox[IPPROTO_MAX]; 119 VNET_DEFINE(struct in6_ifaddrhead, in6_ifaddrhead); 120 121 static struct netisr_handler ip6_nh = { 122 .nh_name = "ip6", 123 .nh_handler = ip6_input, 124 .nh_proto = NETISR_IPV6, 125 .nh_policy = NETISR_POLICY_FLOW, 126 }; 127 128 VNET_DECLARE(struct callout, in6_tmpaddrtimer_ch); 129 #define V_in6_tmpaddrtimer_ch VNET(in6_tmpaddrtimer_ch) 130 131 VNET_DEFINE(struct pfil_head, inet6_pfil_hook); 132 133 VNET_DEFINE(struct ip6stat, ip6stat); 134 135 struct rwlock in6_ifaddr_lock; 136 RW_SYSINIT(in6_ifaddr_lock, &in6_ifaddr_lock, "in6_ifaddr_lock"); 137 138 static void ip6_init2(void *); 139 static struct ip6aux *ip6_setdstifaddr(struct mbuf *, struct in6_ifaddr *); 140 static int ip6_hopopts_input(u_int32_t *, u_int32_t *, struct mbuf **, int *); 141 #ifdef PULLDOWN_TEST 142 static struct mbuf *ip6_pullexthdr(struct mbuf *, size_t, int); 143 #endif 144 145 /* 146 * IP6 initialization: fill in IP6 protocol switch table. 147 * All protocols not implemented in kernel go to raw IP6 protocol handler. 148 */ 149 void 150 ip6_init(void) 151 { 152 struct ip6protosw *pr; 153 int i; 154 155 TUNABLE_INT_FETCH("net.inet6.ip6.auto_linklocal", 156 &V_ip6_auto_linklocal); 157 158 TAILQ_INIT(&V_in6_ifaddrhead); 159 160 /* Initialize packet filter hooks. */ 161 V_inet6_pfil_hook.ph_type = PFIL_TYPE_AF; 162 V_inet6_pfil_hook.ph_af = AF_INET6; 163 if ((i = pfil_head_register(&V_inet6_pfil_hook)) != 0) 164 printf("%s: WARNING: unable to register pfil hook, " 165 "error %d\n", __func__, i); 166 167 scope6_init(); 168 addrsel_policy_init(); 169 nd6_init(); 170 frag6_init(); 171 172 V_ip6_desync_factor = arc4random() % MAX_TEMP_DESYNC_FACTOR; 173 174 /* Skip global initialization stuff for non-default instances. */ 175 if (!IS_DEFAULT_VNET(curvnet)) 176 return; 177 178 #ifdef DIAGNOSTIC 179 if (sizeof(struct protosw) != sizeof(struct ip6protosw)) 180 panic("sizeof(protosw) != sizeof(ip6protosw)"); 181 #endif 182 pr = (struct ip6protosw *)pffindproto(PF_INET6, IPPROTO_RAW, SOCK_RAW); 183 if (pr == NULL) 184 panic("ip6_init"); 185 186 /* Initialize the entire ip6_protox[] array to IPPROTO_RAW. */ 187 for (i = 0; i < IPPROTO_MAX; i++) 188 ip6_protox[i] = pr - inet6sw; 189 /* 190 * Cycle through IP protocols and put them into the appropriate place 191 * in ip6_protox[]. 192 */ 193 for (pr = (struct ip6protosw *)inet6domain.dom_protosw; 194 pr < (struct ip6protosw *)inet6domain.dom_protoswNPROTOSW; pr++) 195 if (pr->pr_domain->dom_family == PF_INET6 && 196 pr->pr_protocol && pr->pr_protocol != IPPROTO_RAW) { 197 /* Be careful to only index valid IP protocols. */ 198 if (pr->pr_protocol < IPPROTO_MAX) 199 ip6_protox[pr->pr_protocol] = pr - inet6sw; 200 } 201 202 netisr_register(&ip6_nh); 203 } 204 205 /* 206 * The protocol to be inserted into ip6_protox[] must be already registered 207 * in inet6sw[], either statically or through pf_proto_register(). 208 */ 209 int 210 ip6proto_register(short ip6proto) 211 { 212 struct ip6protosw *pr; 213 214 /* Sanity checks. */ 215 if (ip6proto <= 0 || ip6proto >= IPPROTO_MAX) 216 return (EPROTONOSUPPORT); 217 218 /* 219 * The protocol slot must not be occupied by another protocol 220 * already. An index pointing to IPPROTO_RAW is unused. 221 */ 222 pr = (struct ip6protosw *)pffindproto(PF_INET6, IPPROTO_RAW, SOCK_RAW); 223 if (pr == NULL) 224 return (EPFNOSUPPORT); 225 if (ip6_protox[ip6proto] != pr - inet6sw) /* IPPROTO_RAW */ 226 return (EEXIST); 227 228 /* 229 * Find the protocol position in inet6sw[] and set the index. 230 */ 231 for (pr = (struct ip6protosw *)inet6domain.dom_protosw; 232 pr < (struct ip6protosw *)inet6domain.dom_protoswNPROTOSW; pr++) { 233 if (pr->pr_domain->dom_family == PF_INET6 && 234 pr->pr_protocol && pr->pr_protocol == ip6proto) { 235 ip6_protox[pr->pr_protocol] = pr - inet6sw; 236 return (0); 237 } 238 } 239 return (EPROTONOSUPPORT); 240 } 241 242 int 243 ip6proto_unregister(short ip6proto) 244 { 245 struct ip6protosw *pr; 246 247 /* Sanity checks. */ 248 if (ip6proto <= 0 || ip6proto >= IPPROTO_MAX) 249 return (EPROTONOSUPPORT); 250 251 /* Check if the protocol was indeed registered. */ 252 pr = (struct ip6protosw *)pffindproto(PF_INET6, IPPROTO_RAW, SOCK_RAW); 253 if (pr == NULL) 254 return (EPFNOSUPPORT); 255 if (ip6_protox[ip6proto] == pr - inet6sw) /* IPPROTO_RAW */ 256 return (ENOENT); 257 258 /* Reset the protocol slot to IPPROTO_RAW. */ 259 ip6_protox[ip6proto] = pr - inet6sw; 260 return (0); 261 } 262 263 #ifdef VIMAGE 264 void 265 ip6_destroy() 266 { 267 268 nd6_destroy(); 269 callout_drain(&V_in6_tmpaddrtimer_ch); 270 } 271 #endif 272 273 static int 274 ip6_init2_vnet(const void *unused __unused) 275 { 276 277 /* nd6_timer_init */ 278 callout_init(&V_nd6_timer_ch, 0); 279 callout_reset(&V_nd6_timer_ch, hz, nd6_timer, curvnet); 280 281 /* timer for regeneranation of temporary addresses randomize ID */ 282 callout_init(&V_in6_tmpaddrtimer_ch, 0); 283 callout_reset(&V_in6_tmpaddrtimer_ch, 284 (V_ip6_temp_preferred_lifetime - V_ip6_desync_factor - 285 V_ip6_temp_regen_advance) * hz, 286 in6_tmpaddrtimer, curvnet); 287 288 return (0); 289 } 290 291 static void 292 ip6_init2(void *dummy) 293 { 294 295 ip6_init2_vnet(NULL); 296 } 297 298 /* cheat */ 299 /* This must be after route_init(), which is now SI_ORDER_THIRD */ 300 SYSINIT(netinet6init2, SI_SUB_PROTO_DOMAIN, SI_ORDER_MIDDLE, ip6_init2, NULL); 301 302 void 303 ip6_input(struct mbuf *m) 304 { 305 struct ip6_hdr *ip6; 306 int off = sizeof(struct ip6_hdr), nest; 307 u_int32_t plen; 308 u_int32_t rtalert = ~0; 309 int nxt, ours = 0; 310 struct ifnet *deliverifp = NULL, *ifp = NULL; 311 struct in6_addr odst; 312 struct route_in6 rin6; 313 int srcrt = 0; 314 struct llentry *lle = NULL; 315 struct sockaddr_in6 dst6, *dst; 316 317 bzero(&rin6, sizeof(struct route_in6)); 318 #ifdef IPSEC 319 /* 320 * should the inner packet be considered authentic? 321 * see comment in ah4_input(). 322 * NB: m cannot be NULL when passed to the input routine 323 */ 324 325 m->m_flags &= ~M_AUTHIPHDR; 326 m->m_flags &= ~M_AUTHIPDGM; 327 328 #endif /* IPSEC */ 329 330 /* 331 * make sure we don't have onion peering information into m_tag. 332 */ 333 ip6_delaux(m); 334 335 /* 336 * mbuf statistics 337 */ 338 if (m->m_flags & M_EXT) { 339 if (m->m_next) 340 V_ip6stat.ip6s_mext2m++; 341 else 342 V_ip6stat.ip6s_mext1++; 343 } else { 344 #define M2MMAX (sizeof(V_ip6stat.ip6s_m2m)/sizeof(V_ip6stat.ip6s_m2m[0])) 345 if (m->m_next) { 346 if (m->m_flags & M_LOOP) { 347 V_ip6stat.ip6s_m2m[V_loif->if_index]++; 348 } else if (m->m_pkthdr.rcvif->if_index < M2MMAX) 349 V_ip6stat.ip6s_m2m[m->m_pkthdr.rcvif->if_index]++; 350 else 351 V_ip6stat.ip6s_m2m[0]++; 352 } else 353 V_ip6stat.ip6s_m1++; 354 #undef M2MMAX 355 } 356 357 /* drop the packet if IPv6 operation is disabled on the IF */ 358 if ((ND_IFINFO(m->m_pkthdr.rcvif)->flags & ND6_IFF_IFDISABLED)) { 359 m_freem(m); 360 return; 361 } 362 363 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_receive); 364 V_ip6stat.ip6s_total++; 365 366 #ifndef PULLDOWN_TEST 367 /* 368 * L2 bridge code and some other code can return mbuf chain 369 * that does not conform to KAME requirement. too bad. 370 * XXX: fails to join if interface MTU > MCLBYTES. jumbogram? 371 */ 372 if (m && m->m_next != NULL && m->m_pkthdr.len < MCLBYTES) { 373 struct mbuf *n; 374 375 MGETHDR(n, M_DONTWAIT, MT_HEADER); 376 if (n) 377 M_MOVE_PKTHDR(n, m); 378 if (n && n->m_pkthdr.len > MHLEN) { 379 MCLGET(n, M_DONTWAIT); 380 if ((n->m_flags & M_EXT) == 0) { 381 m_freem(n); 382 n = NULL; 383 } 384 } 385 if (n == NULL) { 386 m_freem(m); 387 return; /* ENOBUFS */ 388 } 389 390 m_copydata(m, 0, n->m_pkthdr.len, mtod(n, caddr_t)); 391 n->m_len = n->m_pkthdr.len; 392 m_freem(m); 393 m = n; 394 } 395 IP6_EXTHDR_CHECK(m, 0, sizeof(struct ip6_hdr), /* nothing */); 396 #endif 397 398 if (m->m_len < sizeof(struct ip6_hdr)) { 399 struct ifnet *inifp; 400 inifp = m->m_pkthdr.rcvif; 401 if ((m = m_pullup(m, sizeof(struct ip6_hdr))) == NULL) { 402 V_ip6stat.ip6s_toosmall++; 403 in6_ifstat_inc(inifp, ifs6_in_hdrerr); 404 return; 405 } 406 } 407 408 ip6 = mtod(m, struct ip6_hdr *); 409 410 if ((ip6->ip6_vfc & IPV6_VERSION_MASK) != IPV6_VERSION) { 411 V_ip6stat.ip6s_badvers++; 412 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_hdrerr); 413 goto bad; 414 } 415 416 V_ip6stat.ip6s_nxthist[ip6->ip6_nxt]++; 417 418 /* 419 * Check against address spoofing/corruption. 420 */ 421 if (IN6_IS_ADDR_MULTICAST(&ip6->ip6_src) || 422 IN6_IS_ADDR_UNSPECIFIED(&ip6->ip6_dst)) { 423 /* 424 * XXX: "badscope" is not very suitable for a multicast source. 425 */ 426 V_ip6stat.ip6s_badscope++; 427 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_addrerr); 428 goto bad; 429 } 430 if (IN6_IS_ADDR_MC_INTFACELOCAL(&ip6->ip6_dst) && 431 !(m->m_flags & M_LOOP)) { 432 /* 433 * In this case, the packet should come from the loopback 434 * interface. However, we cannot just check the if_flags, 435 * because ip6_mloopback() passes the "actual" interface 436 * as the outgoing/incoming interface. 437 */ 438 V_ip6stat.ip6s_badscope++; 439 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_addrerr); 440 goto bad; 441 } 442 443 #ifdef ALTQ 444 if (altq_input != NULL && (*altq_input)(m, AF_INET6) == 0) { 445 /* packet is dropped by traffic conditioner */ 446 return; 447 } 448 #endif 449 /* 450 * The following check is not documented in specs. A malicious 451 * party may be able to use IPv4 mapped addr to confuse tcp/udp stack 452 * and bypass security checks (act as if it was from 127.0.0.1 by using 453 * IPv6 src ::ffff:127.0.0.1). Be cautious. 454 * 455 * This check chokes if we are in an SIIT cloud. As none of BSDs 456 * support IPv4-less kernel compilation, we cannot support SIIT 457 * environment at all. So, it makes more sense for us to reject any 458 * malicious packets for non-SIIT environment, than try to do a 459 * partial support for SIIT environment. 460 */ 461 if (IN6_IS_ADDR_V4MAPPED(&ip6->ip6_src) || 462 IN6_IS_ADDR_V4MAPPED(&ip6->ip6_dst)) { 463 V_ip6stat.ip6s_badscope++; 464 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_addrerr); 465 goto bad; 466 } 467 #if 0 468 /* 469 * Reject packets with IPv4 compatible addresses (auto tunnel). 470 * 471 * The code forbids auto tunnel relay case in RFC1933 (the check is 472 * stronger than RFC1933). We may want to re-enable it if mech-xx 473 * is revised to forbid relaying case. 474 */ 475 if (IN6_IS_ADDR_V4COMPAT(&ip6->ip6_src) || 476 IN6_IS_ADDR_V4COMPAT(&ip6->ip6_dst)) { 477 V_ip6stat.ip6s_badscope++; 478 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_addrerr); 479 goto bad; 480 } 481 #endif 482 483 /* 484 * Run through list of hooks for input packets. 485 * 486 * NB: Beware of the destination address changing 487 * (e.g. by NAT rewriting). When this happens, 488 * tell ip6_forward to do the right thing. 489 */ 490 odst = ip6->ip6_dst; 491 492 /* Jump over all PFIL processing if hooks are not active. */ 493 if (!PFIL_HOOKED(&V_inet6_pfil_hook)) 494 goto passin; 495 496 if (pfil_run_hooks(&V_inet6_pfil_hook, &m, 497 m->m_pkthdr.rcvif, PFIL_IN, NULL)) 498 return; 499 if (m == NULL) /* consumed by filter */ 500 return; 501 ip6 = mtod(m, struct ip6_hdr *); 502 srcrt = !IN6_ARE_ADDR_EQUAL(&odst, &ip6->ip6_dst); 503 504 passin: 505 /* 506 * Disambiguate address scope zones (if there is ambiguity). 507 * We first make sure that the original source or destination address 508 * is not in our internal form for scoped addresses. Such addresses 509 * are not necessarily invalid spec-wise, but we cannot accept them due 510 * to the usage conflict. 511 * in6_setscope() then also checks and rejects the cases where src or 512 * dst are the loopback address and the receiving interface 513 * is not loopback. 514 */ 515 if (in6_clearscope(&ip6->ip6_src) || in6_clearscope(&ip6->ip6_dst)) { 516 V_ip6stat.ip6s_badscope++; /* XXX */ 517 goto bad; 518 } 519 if (in6_setscope(&ip6->ip6_src, m->m_pkthdr.rcvif, NULL) || 520 in6_setscope(&ip6->ip6_dst, m->m_pkthdr.rcvif, NULL)) { 521 V_ip6stat.ip6s_badscope++; 522 goto bad; 523 } 524 525 /* 526 * Multicast check. Assume packet is for us to avoid 527 * prematurely taking locks. 528 */ 529 if (IN6_IS_ADDR_MULTICAST(&ip6->ip6_dst)) { 530 ours = 1; 531 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_mcast); 532 deliverifp = m->m_pkthdr.rcvif; 533 goto hbhcheck; 534 } 535 536 /* 537 * Unicast check 538 */ 539 540 bzero(&dst6, sizeof(dst6)); 541 dst6.sin6_family = AF_INET6; 542 dst6.sin6_len = sizeof(struct sockaddr_in6); 543 dst6.sin6_addr = ip6->ip6_dst; 544 ifp = m->m_pkthdr.rcvif; 545 IF_AFDATA_LOCK(ifp); 546 lle = lla_lookup(LLTABLE6(ifp), 0, 547 (struct sockaddr *)&dst6); 548 IF_AFDATA_UNLOCK(ifp); 549 if ((lle != NULL) && (lle->la_flags & LLE_IFADDR)) { 550 struct ifaddr *ifa; 551 struct in6_ifaddr *ia6; 552 int bad; 553 554 bad = 1; 555 #define sa_equal(a1, a2) \ 556 (bcmp((a1), (a2), ((a1))->sin6_len) == 0) 557 IF_ADDR_LOCK(ifp); 558 TAILQ_FOREACH(ifa, &ifp->if_addrhead, ifa_link) { 559 if (ifa->ifa_addr->sa_family != dst6.sin6_family) 560 continue; 561 if (sa_equal(&dst6, ifa->ifa_addr)) 562 break; 563 } 564 KASSERT(ifa != NULL, ("%s: ifa not found for lle %p", 565 __func__, lle)); 566 #undef sa_equal 567 568 ia6 = (struct in6_ifaddr *)ifa; 569 if (!(ia6->ia6_flags & IN6_IFF_NOTREADY)) { 570 /* Count the packet in the ip address stats */ 571 ia6->ia_ifa.if_ipackets++; 572 ia6->ia_ifa.if_ibytes += m->m_pkthdr.len; 573 574 /* 575 * record address information into m_tag. 576 */ 577 (void)ip6_setdstifaddr(m, ia6); 578 579 bad = 0; 580 } else { 581 char ip6bufs[INET6_ADDRSTRLEN]; 582 char ip6bufd[INET6_ADDRSTRLEN]; 583 /* address is not ready, so discard the packet. */ 584 nd6log((LOG_INFO, 585 "ip6_input: packet to an unready address %s->%s\n", 586 ip6_sprintf(ip6bufs, &ip6->ip6_src), 587 ip6_sprintf(ip6bufd, &ip6->ip6_dst))); 588 } 589 IF_ADDR_UNLOCK(ifp); 590 LLE_RUNLOCK(lle); 591 if (bad) 592 goto bad; 593 else { 594 ours = 1; 595 deliverifp = ifp; 596 goto hbhcheck; 597 } 598 } 599 if (lle != NULL) 600 LLE_RUNLOCK(lle); 601 602 dst = &rin6.ro_dst; 603 dst->sin6_len = sizeof(struct sockaddr_in6); 604 dst->sin6_family = AF_INET6; 605 dst->sin6_addr = ip6->ip6_dst; 606 rin6.ro_rt = rtalloc1((struct sockaddr *)dst, 0, 0); 607 if (rin6.ro_rt) 608 RT_UNLOCK(rin6.ro_rt); 609 610 #define rt6_key(r) ((struct sockaddr_in6 *)((r)->rt_nodes->rn_key)) 611 612 /* 613 * Accept the packet if the forwarding interface to the destination 614 * according to the routing table is the loopback interface, 615 * unless the associated route has a gateway. 616 * Note that this approach causes to accept a packet if there is a 617 * route to the loopback interface for the destination of the packet. 618 * But we think it's even useful in some situations, e.g. when using 619 * a special daemon which wants to intercept the packet. 620 * 621 * XXX: some OSes automatically make a cloned route for the destination 622 * of an outgoing packet. If the outgoing interface of the packet 623 * is a loopback one, the kernel would consider the packet to be 624 * accepted, even if we have no such address assinged on the interface. 625 * We check the cloned flag of the route entry to reject such cases, 626 * assuming that route entries for our own addresses are not made by 627 * cloning (it should be true because in6_addloop explicitly installs 628 * the host route). However, we might have to do an explicit check 629 * while it would be less efficient. Or, should we rather install a 630 * reject route for such a case? 631 */ 632 if (rin6.ro_rt && 633 (rin6.ro_rt->rt_flags & 634 (RTF_HOST|RTF_GATEWAY)) == RTF_HOST && 635 #ifdef RTF_WASCLONED 636 !(rin6.ro_rt->rt_flags & RTF_WASCLONED) && 637 #endif 638 #ifdef RTF_CLONED 639 !(rin6.ro_rt->rt_flags & RTF_CLONED) && 640 #endif 641 #if 0 642 /* 643 * The check below is redundant since the comparison of 644 * the destination and the key of the rtentry has 645 * already done through looking up the routing table. 646 */ 647 IN6_ARE_ADDR_EQUAL(&ip6->ip6_dst, 648 &rt6_key(rin6.ro_rt)->sin6_addr) 649 #endif 650 rin6.ro_rt->rt_ifp->if_type == IFT_LOOP) { 651 int free_ia6 = 0; 652 struct in6_ifaddr *ia6; 653 654 /* 655 * found the loopback route to the interface address 656 */ 657 if (rin6.ro_rt->rt_gateway->sa_family == AF_LINK) { 658 struct sockaddr_in6 dest6; 659 660 bzero(&dest6, sizeof(dest6)); 661 dest6.sin6_family = AF_INET6; 662 dest6.sin6_len = sizeof(dest6); 663 dest6.sin6_addr = ip6->ip6_dst; 664 ia6 = (struct in6_ifaddr *) 665 ifa_ifwithaddr((struct sockaddr *)&dest6); 666 if (ia6 == NULL) 667 goto bad; 668 free_ia6 = 1; 669 } 670 else 671 ia6 = (struct in6_ifaddr *)rin6.ro_rt->rt_ifa; 672 673 /* 674 * record address information into m_tag. 675 */ 676 (void)ip6_setdstifaddr(m, ia6); 677 678 /* 679 * packets to a tentative, duplicated, or somehow invalid 680 * address must not be accepted. 681 */ 682 if (!(ia6->ia6_flags & IN6_IFF_NOTREADY)) { 683 /* this address is ready */ 684 ours = 1; 685 deliverifp = ia6->ia_ifp; /* correct? */ 686 /* Count the packet in the ip address stats */ 687 ia6->ia_ifa.if_ipackets++; 688 ia6->ia_ifa.if_ibytes += m->m_pkthdr.len; 689 if (ia6 != NULL && free_ia6 != 0) 690 ifa_free(&ia6->ia_ifa); 691 goto hbhcheck; 692 } else { 693 char ip6bufs[INET6_ADDRSTRLEN]; 694 char ip6bufd[INET6_ADDRSTRLEN]; 695 /* address is not ready, so discard the packet. */ 696 nd6log((LOG_INFO, 697 "ip6_input: packet to an unready address %s->%s\n", 698 ip6_sprintf(ip6bufs, &ip6->ip6_src), 699 ip6_sprintf(ip6bufd, &ip6->ip6_dst))); 700 701 if (ia6 != NULL && free_ia6 != 0) 702 ifa_free(&ia6->ia_ifa); 703 goto bad; 704 } 705 } 706 707 /* 708 * FAITH (Firewall Aided Internet Translator) 709 */ 710 if (V_ip6_keepfaith) { 711 if (rin6.ro_rt && rin6.ro_rt->rt_ifp && 712 rin6.ro_rt->rt_ifp->if_type == IFT_FAITH) { 713 /* XXX do we need more sanity checks? */ 714 ours = 1; 715 deliverifp = rin6.ro_rt->rt_ifp; /* faith */ 716 goto hbhcheck; 717 } 718 } 719 720 /* 721 * Now there is no reason to process the packet if it's not our own 722 * and we're not a router. 723 */ 724 if (!V_ip6_forwarding) { 725 V_ip6stat.ip6s_cantforward++; 726 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_discard); 727 goto bad; 728 } 729 730 hbhcheck: 731 /* 732 * record address information into m_tag, if we don't have one yet. 733 * note that we are unable to record it, if the address is not listed 734 * as our interface address (e.g. multicast addresses, addresses 735 * within FAITH prefixes and such). 736 */ 737 if (deliverifp) { 738 struct in6_ifaddr *ia6; 739 740 if ((ia6 = ip6_getdstifaddr(m)) != NULL) { 741 ifa_free(&ia6->ia_ifa); 742 } else { 743 ia6 = in6_ifawithifp(deliverifp, &ip6->ip6_dst); 744 if (ia6) { 745 if (!ip6_setdstifaddr(m, ia6)) { 746 /* 747 * XXX maybe we should drop the packet here, 748 * as we could not provide enough information 749 * to the upper layers. 750 */ 751 } 752 ifa_free(&ia6->ia_ifa); 753 } 754 } 755 } 756 757 /* 758 * Process Hop-by-Hop options header if it's contained. 759 * m may be modified in ip6_hopopts_input(). 760 * If a JumboPayload option is included, plen will also be modified. 761 */ 762 plen = (u_int32_t)ntohs(ip6->ip6_plen); 763 if (ip6->ip6_nxt == IPPROTO_HOPOPTS) { 764 struct ip6_hbh *hbh; 765 766 if (ip6_hopopts_input(&plen, &rtalert, &m, &off)) { 767 #if 0 /*touches NULL pointer*/ 768 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_discard); 769 #endif 770 goto out; /* m have already been freed */ 771 } 772 773 /* adjust pointer */ 774 ip6 = mtod(m, struct ip6_hdr *); 775 776 /* 777 * if the payload length field is 0 and the next header field 778 * indicates Hop-by-Hop Options header, then a Jumbo Payload 779 * option MUST be included. 780 */ 781 if (ip6->ip6_plen == 0 && plen == 0) { 782 /* 783 * Note that if a valid jumbo payload option is 784 * contained, ip6_hopopts_input() must set a valid 785 * (non-zero) payload length to the variable plen. 786 */ 787 V_ip6stat.ip6s_badoptions++; 788 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_discard); 789 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_hdrerr); 790 icmp6_error(m, ICMP6_PARAM_PROB, 791 ICMP6_PARAMPROB_HEADER, 792 (caddr_t)&ip6->ip6_plen - (caddr_t)ip6); 793 goto out; 794 } 795 #ifndef PULLDOWN_TEST 796 /* ip6_hopopts_input() ensures that mbuf is contiguous */ 797 hbh = (struct ip6_hbh *)(ip6 + 1); 798 #else 799 IP6_EXTHDR_GET(hbh, struct ip6_hbh *, m, sizeof(struct ip6_hdr), 800 sizeof(struct ip6_hbh)); 801 if (hbh == NULL) { 802 V_ip6stat.ip6s_tooshort++; 803 goto out; 804 } 805 #endif 806 nxt = hbh->ip6h_nxt; 807 808 /* 809 * If we are acting as a router and the packet contains a 810 * router alert option, see if we know the option value. 811 * Currently, we only support the option value for MLD, in which 812 * case we should pass the packet to the multicast routing 813 * daemon. 814 */ 815 if (rtalert != ~0) { 816 switch (rtalert) { 817 case IP6OPT_RTALERT_MLD: 818 if (V_ip6_forwarding) 819 ours = 1; 820 break; 821 default: 822 /* 823 * RFC2711 requires unrecognized values must be 824 * silently ignored. 825 */ 826 break; 827 } 828 } 829 } else 830 nxt = ip6->ip6_nxt; 831 832 /* 833 * Check that the amount of data in the buffers 834 * is as at least much as the IPv6 header would have us expect. 835 * Trim mbufs if longer than we expect. 836 * Drop packet if shorter than we expect. 837 */ 838 if (m->m_pkthdr.len - sizeof(struct ip6_hdr) < plen) { 839 V_ip6stat.ip6s_tooshort++; 840 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_truncated); 841 goto bad; 842 } 843 if (m->m_pkthdr.len > sizeof(struct ip6_hdr) + plen) { 844 if (m->m_len == m->m_pkthdr.len) { 845 m->m_len = sizeof(struct ip6_hdr) + plen; 846 m->m_pkthdr.len = sizeof(struct ip6_hdr) + plen; 847 } else 848 m_adj(m, sizeof(struct ip6_hdr) + plen - m->m_pkthdr.len); 849 } 850 851 /* 852 * Forward if desirable. 853 */ 854 if (V_ip6_mrouter && 855 IN6_IS_ADDR_MULTICAST(&ip6->ip6_dst)) { 856 /* 857 * If we are acting as a multicast router, all 858 * incoming multicast packets are passed to the 859 * kernel-level multicast forwarding function. 860 * The packet is returned (relatively) intact; if 861 * ip6_mforward() returns a non-zero value, the packet 862 * must be discarded, else it may be accepted below. 863 * 864 * XXX TODO: Check hlim and multicast scope here to avoid 865 * unnecessarily calling into ip6_mforward(). 866 */ 867 if (ip6_mforward && 868 ip6_mforward(ip6, m->m_pkthdr.rcvif, m)) { 869 IP6STAT_INC(ip6s_cantforward); 870 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_discard); 871 goto bad; 872 } 873 } else if (!ours) { 874 ip6_forward(m, srcrt); 875 goto out; 876 } 877 878 ip6 = mtod(m, struct ip6_hdr *); 879 880 /* 881 * Malicious party may be able to use IPv4 mapped addr to confuse 882 * tcp/udp stack and bypass security checks (act as if it was from 883 * 127.0.0.1 by using IPv6 src ::ffff:127.0.0.1). Be cautious. 884 * 885 * For SIIT end node behavior, you may want to disable the check. 886 * However, you will become vulnerable to attacks using IPv4 mapped 887 * source. 888 */ 889 if (IN6_IS_ADDR_V4MAPPED(&ip6->ip6_src) || 890 IN6_IS_ADDR_V4MAPPED(&ip6->ip6_dst)) { 891 V_ip6stat.ip6s_badscope++; 892 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_addrerr); 893 goto bad; 894 } 895 896 /* 897 * Tell launch routine the next header 898 */ 899 V_ip6stat.ip6s_delivered++; 900 in6_ifstat_inc(deliverifp, ifs6_in_deliver); 901 nest = 0; 902 903 while (nxt != IPPROTO_DONE) { 904 if (V_ip6_hdrnestlimit && (++nest > V_ip6_hdrnestlimit)) { 905 V_ip6stat.ip6s_toomanyhdr++; 906 goto bad; 907 } 908 909 /* 910 * protection against faulty packet - there should be 911 * more sanity checks in header chain processing. 912 */ 913 if (m->m_pkthdr.len < off) { 914 V_ip6stat.ip6s_tooshort++; 915 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_truncated); 916 goto bad; 917 } 918 919 #ifdef IPSEC 920 /* 921 * enforce IPsec policy checking if we are seeing last header. 922 * note that we do not visit this with protocols with pcb layer 923 * code - like udp/tcp/raw ip. 924 */ 925 if (ip6_ipsec_input(m, nxt)) 926 goto bad; 927 #endif /* IPSEC */ 928 929 /* 930 * Use mbuf flags to propagate Router Alert option to 931 * ICMPv6 layer, as hop-by-hop options have been stripped. 932 */ 933 if (nxt == IPPROTO_ICMPV6 && rtalert != ~0) 934 m->m_flags |= M_RTALERT_MLD; 935 936 nxt = (*inet6sw[ip6_protox[nxt]].pr_input)(&m, &off, nxt); 937 } 938 goto out; 939 bad: 940 m_freem(m); 941 out: 942 if (rin6.ro_rt) 943 RTFREE(rin6.ro_rt); 944 } 945 946 /* 947 * set/grab in6_ifaddr correspond to IPv6 destination address. 948 * XXX backward compatibility wrapper 949 * 950 * XXXRW: We should bump the refcount on ia6 before sticking it in the m_tag, 951 * and then bump it when the tag is copied, and release it when the tag is 952 * freed. Unfortunately, m_tags don't support deep copies (yet), so instead 953 * we just bump the ia refcount when we receive it. This should be fixed. 954 */ 955 static struct ip6aux * 956 ip6_setdstifaddr(struct mbuf *m, struct in6_ifaddr *ia6) 957 { 958 struct ip6aux *ip6a; 959 960 ip6a = ip6_addaux(m); 961 if (ip6a) 962 ip6a->ip6a_dstia6 = ia6; 963 return ip6a; /* NULL if failed to set */ 964 } 965 966 struct in6_ifaddr * 967 ip6_getdstifaddr(struct mbuf *m) 968 { 969 struct ip6aux *ip6a; 970 struct in6_ifaddr *ia; 971 972 ip6a = ip6_findaux(m); 973 if (ip6a) { 974 ia = ip6a->ip6a_dstia6; 975 ifa_ref(&ia->ia_ifa); 976 return ia; 977 } else 978 return NULL; 979 } 980 981 /* 982 * Hop-by-Hop options header processing. If a valid jumbo payload option is 983 * included, the real payload length will be stored in plenp. 984 * 985 * rtalertp - XXX: should be stored more smart way 986 */ 987 static int 988 ip6_hopopts_input(u_int32_t *plenp, u_int32_t *rtalertp, 989 struct mbuf **mp, int *offp) 990 { 991 struct mbuf *m = *mp; 992 int off = *offp, hbhlen; 993 struct ip6_hbh *hbh; 994 u_int8_t *opt; 995 996 /* validation of the length of the header */ 997 #ifndef PULLDOWN_TEST 998 IP6_EXTHDR_CHECK(m, off, sizeof(*hbh), -1); 999 hbh = (struct ip6_hbh *)(mtod(m, caddr_t) + off); 1000 hbhlen = (hbh->ip6h_len + 1) << 3; 1001 1002 IP6_EXTHDR_CHECK(m, off, hbhlen, -1); 1003 hbh = (struct ip6_hbh *)(mtod(m, caddr_t) + off); 1004 #else 1005 IP6_EXTHDR_GET(hbh, struct ip6_hbh *, m, 1006 sizeof(struct ip6_hdr), sizeof(struct ip6_hbh)); 1007 if (hbh == NULL) { 1008 V_ip6stat.ip6s_tooshort++; 1009 return -1; 1010 } 1011 hbhlen = (hbh->ip6h_len + 1) << 3; 1012 IP6_EXTHDR_GET(hbh, struct ip6_hbh *, m, sizeof(struct ip6_hdr), 1013 hbhlen); 1014 if (hbh == NULL) { 1015 V_ip6stat.ip6s_tooshort++; 1016 return -1; 1017 } 1018 #endif 1019 off += hbhlen; 1020 hbhlen -= sizeof(struct ip6_hbh); 1021 opt = (u_int8_t *)hbh + sizeof(struct ip6_hbh); 1022 1023 if (ip6_process_hopopts(m, (u_int8_t *)hbh + sizeof(struct ip6_hbh), 1024 hbhlen, rtalertp, plenp) < 0) 1025 return (-1); 1026 1027 *offp = off; 1028 *mp = m; 1029 return (0); 1030 } 1031 1032 /* 1033 * Search header for all Hop-by-hop options and process each option. 1034 * This function is separate from ip6_hopopts_input() in order to 1035 * handle a case where the sending node itself process its hop-by-hop 1036 * options header. In such a case, the function is called from ip6_output(). 1037 * 1038 * The function assumes that hbh header is located right after the IPv6 header 1039 * (RFC2460 p7), opthead is pointer into data content in m, and opthead to 1040 * opthead + hbhlen is located in continuous memory region. 1041 */ 1042 int 1043 ip6_process_hopopts(struct mbuf *m, u_int8_t *opthead, int hbhlen, 1044 u_int32_t *rtalertp, u_int32_t *plenp) 1045 { 1046 struct ip6_hdr *ip6; 1047 int optlen = 0; 1048 u_int8_t *opt = opthead; 1049 u_int16_t rtalert_val; 1050 u_int32_t jumboplen; 1051 const int erroff = sizeof(struct ip6_hdr) + sizeof(struct ip6_hbh); 1052 1053 for (; hbhlen > 0; hbhlen -= optlen, opt += optlen) { 1054 switch (*opt) { 1055 case IP6OPT_PAD1: 1056 optlen = 1; 1057 break; 1058 case IP6OPT_PADN: 1059 if (hbhlen < IP6OPT_MINLEN) { 1060 V_ip6stat.ip6s_toosmall++; 1061 goto bad; 1062 } 1063 optlen = *(opt + 1) + 2; 1064 break; 1065 case IP6OPT_ROUTER_ALERT: 1066 /* XXX may need check for alignment */ 1067 if (hbhlen < IP6OPT_RTALERT_LEN) { 1068 V_ip6stat.ip6s_toosmall++; 1069 goto bad; 1070 } 1071 if (*(opt + 1) != IP6OPT_RTALERT_LEN - 2) { 1072 /* XXX stat */ 1073 icmp6_error(m, ICMP6_PARAM_PROB, 1074 ICMP6_PARAMPROB_HEADER, 1075 erroff + opt + 1 - opthead); 1076 return (-1); 1077 } 1078 optlen = IP6OPT_RTALERT_LEN; 1079 bcopy((caddr_t)(opt + 2), (caddr_t)&rtalert_val, 2); 1080 *rtalertp = ntohs(rtalert_val); 1081 break; 1082 case IP6OPT_JUMBO: 1083 /* XXX may need check for alignment */ 1084 if (hbhlen < IP6OPT_JUMBO_LEN) { 1085 V_ip6stat.ip6s_toosmall++; 1086 goto bad; 1087 } 1088 if (*(opt + 1) != IP6OPT_JUMBO_LEN - 2) { 1089 /* XXX stat */ 1090 icmp6_error(m, ICMP6_PARAM_PROB, 1091 ICMP6_PARAMPROB_HEADER, 1092 erroff + opt + 1 - opthead); 1093 return (-1); 1094 } 1095 optlen = IP6OPT_JUMBO_LEN; 1096 1097 /* 1098 * IPv6 packets that have non 0 payload length 1099 * must not contain a jumbo payload option. 1100 */ 1101 ip6 = mtod(m, struct ip6_hdr *); 1102 if (ip6->ip6_plen) { 1103 V_ip6stat.ip6s_badoptions++; 1104 icmp6_error(m, ICMP6_PARAM_PROB, 1105 ICMP6_PARAMPROB_HEADER, 1106 erroff + opt - opthead); 1107 return (-1); 1108 } 1109 1110 /* 1111 * We may see jumbolen in unaligned location, so 1112 * we'd need to perform bcopy(). 1113 */ 1114 bcopy(opt + 2, &jumboplen, sizeof(jumboplen)); 1115 jumboplen = (u_int32_t)htonl(jumboplen); 1116 1117 #if 1 1118 /* 1119 * if there are multiple jumbo payload options, 1120 * *plenp will be non-zero and the packet will be 1121 * rejected. 1122 * the behavior may need some debate in ipngwg - 1123 * multiple options does not make sense, however, 1124 * there's no explicit mention in specification. 1125 */ 1126 if (*plenp != 0) { 1127 V_ip6stat.ip6s_badoptions++; 1128 icmp6_error(m, ICMP6_PARAM_PROB, 1129 ICMP6_PARAMPROB_HEADER, 1130 erroff + opt + 2 - opthead); 1131 return (-1); 1132 } 1133 #endif 1134 1135 /* 1136 * jumbo payload length must be larger than 65535. 1137 */ 1138 if (jumboplen <= IPV6_MAXPACKET) { 1139 V_ip6stat.ip6s_badoptions++; 1140 icmp6_error(m, ICMP6_PARAM_PROB, 1141 ICMP6_PARAMPROB_HEADER, 1142 erroff + opt + 2 - opthead); 1143 return (-1); 1144 } 1145 *plenp = jumboplen; 1146 1147 break; 1148 default: /* unknown option */ 1149 if (hbhlen < IP6OPT_MINLEN) { 1150 V_ip6stat.ip6s_toosmall++; 1151 goto bad; 1152 } 1153 optlen = ip6_unknown_opt(opt, m, 1154 erroff + opt - opthead); 1155 if (optlen == -1) 1156 return (-1); 1157 optlen += 2; 1158 break; 1159 } 1160 } 1161 1162 return (0); 1163 1164 bad: 1165 m_freem(m); 1166 return (-1); 1167 } 1168 1169 /* 1170 * Unknown option processing. 1171 * The third argument `off' is the offset from the IPv6 header to the option, 1172 * which is necessary if the IPv6 header the and option header and IPv6 header 1173 * is not continuous in order to return an ICMPv6 error. 1174 */ 1175 int 1176 ip6_unknown_opt(u_int8_t *optp, struct mbuf *m, int off) 1177 { 1178 struct ip6_hdr *ip6; 1179 1180 switch (IP6OPT_TYPE(*optp)) { 1181 case IP6OPT_TYPE_SKIP: /* ignore the option */ 1182 return ((int)*(optp + 1)); 1183 case IP6OPT_TYPE_DISCARD: /* silently discard */ 1184 m_freem(m); 1185 return (-1); 1186 case IP6OPT_TYPE_FORCEICMP: /* send ICMP even if multicasted */ 1187 V_ip6stat.ip6s_badoptions++; 1188 icmp6_error(m, ICMP6_PARAM_PROB, ICMP6_PARAMPROB_OPTION, off); 1189 return (-1); 1190 case IP6OPT_TYPE_ICMP: /* send ICMP if not multicasted */ 1191 V_ip6stat.ip6s_badoptions++; 1192 ip6 = mtod(m, struct ip6_hdr *); 1193 if (IN6_IS_ADDR_MULTICAST(&ip6->ip6_dst) || 1194 (m->m_flags & (M_BCAST|M_MCAST))) 1195 m_freem(m); 1196 else 1197 icmp6_error(m, ICMP6_PARAM_PROB, 1198 ICMP6_PARAMPROB_OPTION, off); 1199 return (-1); 1200 } 1201 1202 m_freem(m); /* XXX: NOTREACHED */ 1203 return (-1); 1204 } 1205 1206 /* 1207 * Create the "control" list for this pcb. 1208 * These functions will not modify mbuf chain at all. 1209 * 1210 * With KAME mbuf chain restriction: 1211 * The routine will be called from upper layer handlers like tcp6_input(). 1212 * Thus the routine assumes that the caller (tcp6_input) have already 1213 * called IP6_EXTHDR_CHECK() and all the extension headers are located in the 1214 * very first mbuf on the mbuf chain. 1215 * 1216 * ip6_savecontrol_v4 will handle those options that are possible to be 1217 * set on a v4-mapped socket. 1218 * ip6_savecontrol will directly call ip6_savecontrol_v4 to handle those 1219 * options and handle the v6-only ones itself. 1220 */ 1221 struct mbuf ** 1222 ip6_savecontrol_v4(struct inpcb *inp, struct mbuf *m, struct mbuf **mp, 1223 int *v4only) 1224 { 1225 struct ip6_hdr *ip6 = mtod(m, struct ip6_hdr *); 1226 1227 #ifdef SO_TIMESTAMP 1228 if ((inp->inp_socket->so_options & SO_TIMESTAMP) != 0) { 1229 struct timeval tv; 1230 1231 microtime(&tv); 1232 *mp = sbcreatecontrol((caddr_t) &tv, sizeof(tv), 1233 SCM_TIMESTAMP, SOL_SOCKET); 1234 if (*mp) 1235 mp = &(*mp)->m_next; 1236 } 1237 #endif 1238 1239 if ((ip6->ip6_vfc & IPV6_VERSION_MASK) != IPV6_VERSION) { 1240 if (v4only != NULL) 1241 *v4only = 1; 1242 return (mp); 1243 } 1244 1245 #define IS2292(inp, x, y) (((inp)->inp_flags & IN6P_RFC2292) ? (x) : (y)) 1246 /* RFC 2292 sec. 5 */ 1247 if ((inp->inp_flags & IN6P_PKTINFO) != 0) { 1248 struct in6_pktinfo pi6; 1249 1250 bcopy(&ip6->ip6_dst, &pi6.ipi6_addr, sizeof(struct in6_addr)); 1251 in6_clearscope(&pi6.ipi6_addr); /* XXX */ 1252 pi6.ipi6_ifindex = 1253 (m && m->m_pkthdr.rcvif) ? m->m_pkthdr.rcvif->if_index : 0; 1254 1255 *mp = sbcreatecontrol((caddr_t) &pi6, 1256 sizeof(struct in6_pktinfo), 1257 IS2292(inp, IPV6_2292PKTINFO, IPV6_PKTINFO), IPPROTO_IPV6); 1258 if (*mp) 1259 mp = &(*mp)->m_next; 1260 } 1261 1262 if ((inp->inp_flags & IN6P_HOPLIMIT) != 0) { 1263 int hlim = ip6->ip6_hlim & 0xff; 1264 1265 *mp = sbcreatecontrol((caddr_t) &hlim, sizeof(int), 1266 IS2292(inp, IPV6_2292HOPLIMIT, IPV6_HOPLIMIT), 1267 IPPROTO_IPV6); 1268 if (*mp) 1269 mp = &(*mp)->m_next; 1270 } 1271 1272 if (v4only != NULL) 1273 *v4only = 0; 1274 return (mp); 1275 } 1276 1277 void 1278 ip6_savecontrol(struct inpcb *in6p, struct mbuf *m, struct mbuf **mp) 1279 { 1280 struct ip6_hdr *ip6 = mtod(m, struct ip6_hdr *); 1281 int v4only = 0; 1282 1283 mp = ip6_savecontrol_v4(in6p, m, mp, &v4only); 1284 if (v4only) 1285 return; 1286 1287 if ((in6p->inp_flags & IN6P_TCLASS) != 0) { 1288 u_int32_t flowinfo; 1289 int tclass; 1290 1291 flowinfo = (u_int32_t)ntohl(ip6->ip6_flow & IPV6_FLOWINFO_MASK); 1292 flowinfo >>= 20; 1293 1294 tclass = flowinfo & 0xff; 1295 *mp = sbcreatecontrol((caddr_t) &tclass, sizeof(tclass), 1296 IPV6_TCLASS, IPPROTO_IPV6); 1297 if (*mp) 1298 mp = &(*mp)->m_next; 1299 } 1300 1301 /* 1302 * IPV6_HOPOPTS socket option. Recall that we required super-user 1303 * privilege for the option (see ip6_ctloutput), but it might be too 1304 * strict, since there might be some hop-by-hop options which can be 1305 * returned to normal user. 1306 * See also RFC 2292 section 6 (or RFC 3542 section 8). 1307 */ 1308 if ((in6p->inp_flags & IN6P_HOPOPTS) != 0) { 1309 /* 1310 * Check if a hop-by-hop options header is contatined in the 1311 * received packet, and if so, store the options as ancillary 1312 * data. Note that a hop-by-hop options header must be 1313 * just after the IPv6 header, which is assured through the 1314 * IPv6 input processing. 1315 */ 1316 if (ip6->ip6_nxt == IPPROTO_HOPOPTS) { 1317 struct ip6_hbh *hbh; 1318 int hbhlen = 0; 1319 #ifdef PULLDOWN_TEST 1320 struct mbuf *ext; 1321 #endif 1322 1323 #ifndef PULLDOWN_TEST 1324 hbh = (struct ip6_hbh *)(ip6 + 1); 1325 hbhlen = (hbh->ip6h_len + 1) << 3; 1326 #else 1327 ext = ip6_pullexthdr(m, sizeof(struct ip6_hdr), 1328 ip6->ip6_nxt); 1329 if (ext == NULL) { 1330 V_ip6stat.ip6s_tooshort++; 1331 return; 1332 } 1333 hbh = mtod(ext, struct ip6_hbh *); 1334 hbhlen = (hbh->ip6h_len + 1) << 3; 1335 if (hbhlen != ext->m_len) { 1336 m_freem(ext); 1337 V_ip6stat.ip6s_tooshort++; 1338 return; 1339 } 1340 #endif 1341 1342 /* 1343 * XXX: We copy the whole header even if a 1344 * jumbo payload option is included, the option which 1345 * is to be removed before returning according to 1346 * RFC2292. 1347 * Note: this constraint is removed in RFC3542 1348 */ 1349 *mp = sbcreatecontrol((caddr_t)hbh, hbhlen, 1350 IS2292(in6p, IPV6_2292HOPOPTS, IPV6_HOPOPTS), 1351 IPPROTO_IPV6); 1352 if (*mp) 1353 mp = &(*mp)->m_next; 1354 #ifdef PULLDOWN_TEST 1355 m_freem(ext); 1356 #endif 1357 } 1358 } 1359 1360 if ((in6p->inp_flags & (IN6P_RTHDR | IN6P_DSTOPTS)) != 0) { 1361 int nxt = ip6->ip6_nxt, off = sizeof(struct ip6_hdr); 1362 1363 /* 1364 * Search for destination options headers or routing 1365 * header(s) through the header chain, and stores each 1366 * header as ancillary data. 1367 * Note that the order of the headers remains in 1368 * the chain of ancillary data. 1369 */ 1370 while (1) { /* is explicit loop prevention necessary? */ 1371 struct ip6_ext *ip6e = NULL; 1372 int elen; 1373 #ifdef PULLDOWN_TEST 1374 struct mbuf *ext = NULL; 1375 #endif 1376 1377 /* 1378 * if it is not an extension header, don't try to 1379 * pull it from the chain. 1380 */ 1381 switch (nxt) { 1382 case IPPROTO_DSTOPTS: 1383 case IPPROTO_ROUTING: 1384 case IPPROTO_HOPOPTS: 1385 case IPPROTO_AH: /* is it possible? */ 1386 break; 1387 default: 1388 goto loopend; 1389 } 1390 1391 #ifndef PULLDOWN_TEST 1392 if (off + sizeof(*ip6e) > m->m_len) 1393 goto loopend; 1394 ip6e = (struct ip6_ext *)(mtod(m, caddr_t) + off); 1395 if (nxt == IPPROTO_AH) 1396 elen = (ip6e->ip6e_len + 2) << 2; 1397 else 1398 elen = (ip6e->ip6e_len + 1) << 3; 1399 if (off + elen > m->m_len) 1400 goto loopend; 1401 #else 1402 ext = ip6_pullexthdr(m, off, nxt); 1403 if (ext == NULL) { 1404 V_ip6stat.ip6s_tooshort++; 1405 return; 1406 } 1407 ip6e = mtod(ext, struct ip6_ext *); 1408 if (nxt == IPPROTO_AH) 1409 elen = (ip6e->ip6e_len + 2) << 2; 1410 else 1411 elen = (ip6e->ip6e_len + 1) << 3; 1412 if (elen != ext->m_len) { 1413 m_freem(ext); 1414 V_ip6stat.ip6s_tooshort++; 1415 return; 1416 } 1417 #endif 1418 1419 switch (nxt) { 1420 case IPPROTO_DSTOPTS: 1421 if (!(in6p->inp_flags & IN6P_DSTOPTS)) 1422 break; 1423 1424 *mp = sbcreatecontrol((caddr_t)ip6e, elen, 1425 IS2292(in6p, 1426 IPV6_2292DSTOPTS, IPV6_DSTOPTS), 1427 IPPROTO_IPV6); 1428 if (*mp) 1429 mp = &(*mp)->m_next; 1430 break; 1431 case IPPROTO_ROUTING: 1432 if (!in6p->inp_flags & IN6P_RTHDR) 1433 break; 1434 1435 *mp = sbcreatecontrol((caddr_t)ip6e, elen, 1436 IS2292(in6p, IPV6_2292RTHDR, IPV6_RTHDR), 1437 IPPROTO_IPV6); 1438 if (*mp) 1439 mp = &(*mp)->m_next; 1440 break; 1441 case IPPROTO_HOPOPTS: 1442 case IPPROTO_AH: /* is it possible? */ 1443 break; 1444 1445 default: 1446 /* 1447 * other cases have been filtered in the above. 1448 * none will visit this case. here we supply 1449 * the code just in case (nxt overwritten or 1450 * other cases). 1451 */ 1452 #ifdef PULLDOWN_TEST 1453 m_freem(ext); 1454 #endif 1455 goto loopend; 1456 1457 } 1458 1459 /* proceed with the next header. */ 1460 off += elen; 1461 nxt = ip6e->ip6e_nxt; 1462 ip6e = NULL; 1463 #ifdef PULLDOWN_TEST 1464 m_freem(ext); 1465 ext = NULL; 1466 #endif 1467 } 1468 loopend: 1469 ; 1470 } 1471 } 1472 #undef IS2292 1473 1474 void 1475 ip6_notify_pmtu(struct inpcb *in6p, struct sockaddr_in6 *dst, u_int32_t *mtu) 1476 { 1477 struct socket *so; 1478 struct mbuf *m_mtu; 1479 struct ip6_mtuinfo mtuctl; 1480 1481 so = in6p->inp_socket; 1482 1483 if (mtu == NULL) 1484 return; 1485 1486 #ifdef DIAGNOSTIC 1487 if (so == NULL) /* I believe this is impossible */ 1488 panic("ip6_notify_pmtu: socket is NULL"); 1489 #endif 1490 1491 bzero(&mtuctl, sizeof(mtuctl)); /* zero-clear for safety */ 1492 mtuctl.ip6m_mtu = *mtu; 1493 mtuctl.ip6m_addr = *dst; 1494 if (sa6_recoverscope(&mtuctl.ip6m_addr)) 1495 return; 1496 1497 if ((m_mtu = sbcreatecontrol((caddr_t)&mtuctl, sizeof(mtuctl), 1498 IPV6_PATHMTU, IPPROTO_IPV6)) == NULL) 1499 return; 1500 1501 if (sbappendaddr(&so->so_rcv, (struct sockaddr *)dst, NULL, m_mtu) 1502 == 0) { 1503 m_freem(m_mtu); 1504 /* XXX: should count statistics */ 1505 } else 1506 sorwakeup(so); 1507 1508 return; 1509 } 1510 1511 #ifdef PULLDOWN_TEST 1512 /* 1513 * pull single extension header from mbuf chain. returns single mbuf that 1514 * contains the result, or NULL on error. 1515 */ 1516 static struct mbuf * 1517 ip6_pullexthdr(struct mbuf *m, size_t off, int nxt) 1518 { 1519 struct ip6_ext ip6e; 1520 size_t elen; 1521 struct mbuf *n; 1522 1523 #ifdef DIAGNOSTIC 1524 switch (nxt) { 1525 case IPPROTO_DSTOPTS: 1526 case IPPROTO_ROUTING: 1527 case IPPROTO_HOPOPTS: 1528 case IPPROTO_AH: /* is it possible? */ 1529 break; 1530 default: 1531 printf("ip6_pullexthdr: invalid nxt=%d\n", nxt); 1532 } 1533 #endif 1534 1535 m_copydata(m, off, sizeof(ip6e), (caddr_t)&ip6e); 1536 if (nxt == IPPROTO_AH) 1537 elen = (ip6e.ip6e_len + 2) << 2; 1538 else 1539 elen = (ip6e.ip6e_len + 1) << 3; 1540 1541 MGET(n, M_DONTWAIT, MT_DATA); 1542 if (n && elen >= MLEN) { 1543 MCLGET(n, M_DONTWAIT); 1544 if ((n->m_flags & M_EXT) == 0) { 1545 m_free(n); 1546 n = NULL; 1547 } 1548 } 1549 if (!n) 1550 return NULL; 1551 1552 n->m_len = 0; 1553 if (elen >= M_TRAILINGSPACE(n)) { 1554 m_free(n); 1555 return NULL; 1556 } 1557 1558 m_copydata(m, off, elen, mtod(n, caddr_t)); 1559 n->m_len = elen; 1560 return n; 1561 } 1562 #endif 1563 1564 /* 1565 * Get pointer to the previous header followed by the header 1566 * currently processed. 1567 * XXX: This function supposes that 1568 * M includes all headers, 1569 * the next header field and the header length field of each header 1570 * are valid, and 1571 * the sum of each header length equals to OFF. 1572 * Because of these assumptions, this function must be called very 1573 * carefully. Moreover, it will not be used in the near future when 1574 * we develop `neater' mechanism to process extension headers. 1575 */ 1576 char * 1577 ip6_get_prevhdr(struct mbuf *m, int off) 1578 { 1579 struct ip6_hdr *ip6 = mtod(m, struct ip6_hdr *); 1580 1581 if (off == sizeof(struct ip6_hdr)) 1582 return (&ip6->ip6_nxt); 1583 else { 1584 int len, nxt; 1585 struct ip6_ext *ip6e = NULL; 1586 1587 nxt = ip6->ip6_nxt; 1588 len = sizeof(struct ip6_hdr); 1589 while (len < off) { 1590 ip6e = (struct ip6_ext *)(mtod(m, caddr_t) + len); 1591 1592 switch (nxt) { 1593 case IPPROTO_FRAGMENT: 1594 len += sizeof(struct ip6_frag); 1595 break; 1596 case IPPROTO_AH: 1597 len += (ip6e->ip6e_len + 2) << 2; 1598 break; 1599 default: 1600 len += (ip6e->ip6e_len + 1) << 3; 1601 break; 1602 } 1603 nxt = ip6e->ip6e_nxt; 1604 } 1605 if (ip6e) 1606 return (&ip6e->ip6e_nxt); 1607 else 1608 return NULL; 1609 } 1610 } 1611 1612 /* 1613 * get next header offset. m will be retained. 1614 */ 1615 int 1616 ip6_nexthdr(struct mbuf *m, int off, int proto, int *nxtp) 1617 { 1618 struct ip6_hdr ip6; 1619 struct ip6_ext ip6e; 1620 struct ip6_frag fh; 1621 1622 /* just in case */ 1623 if (m == NULL) 1624 panic("ip6_nexthdr: m == NULL"); 1625 if ((m->m_flags & M_PKTHDR) == 0 || m->m_pkthdr.len < off) 1626 return -1; 1627 1628 switch (proto) { 1629 case IPPROTO_IPV6: 1630 if (m->m_pkthdr.len < off + sizeof(ip6)) 1631 return -1; 1632 m_copydata(m, off, sizeof(ip6), (caddr_t)&ip6); 1633 if (nxtp) 1634 *nxtp = ip6.ip6_nxt; 1635 off += sizeof(ip6); 1636 return off; 1637 1638 case IPPROTO_FRAGMENT: 1639 /* 1640 * terminate parsing if it is not the first fragment, 1641 * it does not make sense to parse through it. 1642 */ 1643 if (m->m_pkthdr.len < off + sizeof(fh)) 1644 return -1; 1645 m_copydata(m, off, sizeof(fh), (caddr_t)&fh); 1646 /* IP6F_OFF_MASK = 0xfff8(BigEndian), 0xf8ff(LittleEndian) */ 1647 if (fh.ip6f_offlg & IP6F_OFF_MASK) 1648 return -1; 1649 if (nxtp) 1650 *nxtp = fh.ip6f_nxt; 1651 off += sizeof(struct ip6_frag); 1652 return off; 1653 1654 case IPPROTO_AH: 1655 if (m->m_pkthdr.len < off + sizeof(ip6e)) 1656 return -1; 1657 m_copydata(m, off, sizeof(ip6e), (caddr_t)&ip6e); 1658 if (nxtp) 1659 *nxtp = ip6e.ip6e_nxt; 1660 off += (ip6e.ip6e_len + 2) << 2; 1661 return off; 1662 1663 case IPPROTO_HOPOPTS: 1664 case IPPROTO_ROUTING: 1665 case IPPROTO_DSTOPTS: 1666 if (m->m_pkthdr.len < off + sizeof(ip6e)) 1667 return -1; 1668 m_copydata(m, off, sizeof(ip6e), (caddr_t)&ip6e); 1669 if (nxtp) 1670 *nxtp = ip6e.ip6e_nxt; 1671 off += (ip6e.ip6e_len + 1) << 3; 1672 return off; 1673 1674 case IPPROTO_NONE: 1675 case IPPROTO_ESP: 1676 case IPPROTO_IPCOMP: 1677 /* give up */ 1678 return -1; 1679 1680 default: 1681 return -1; 1682 } 1683 1684 return -1; 1685 } 1686 1687 /* 1688 * get offset for the last header in the chain. m will be kept untainted. 1689 */ 1690 int 1691 ip6_lasthdr(struct mbuf *m, int off, int proto, int *nxtp) 1692 { 1693 int newoff; 1694 int nxt; 1695 1696 if (!nxtp) { 1697 nxt = -1; 1698 nxtp = &nxt; 1699 } 1700 while (1) { 1701 newoff = ip6_nexthdr(m, off, proto, nxtp); 1702 if (newoff < 0) 1703 return off; 1704 else if (newoff < off) 1705 return -1; /* invalid */ 1706 else if (newoff == off) 1707 return newoff; 1708 1709 off = newoff; 1710 proto = *nxtp; 1711 } 1712 } 1713 1714 struct ip6aux * 1715 ip6_addaux(struct mbuf *m) 1716 { 1717 struct m_tag *mtag; 1718 1719 mtag = m_tag_find(m, PACKET_TAG_IPV6_INPUT, NULL); 1720 if (!mtag) { 1721 mtag = m_tag_get(PACKET_TAG_IPV6_INPUT, sizeof(struct ip6aux), 1722 M_NOWAIT); 1723 if (mtag) { 1724 m_tag_prepend(m, mtag); 1725 bzero(mtag + 1, sizeof(struct ip6aux)); 1726 } 1727 } 1728 return mtag ? (struct ip6aux *)(mtag + 1) : NULL; 1729 } 1730 1731 struct ip6aux * 1732 ip6_findaux(struct mbuf *m) 1733 { 1734 struct m_tag *mtag; 1735 1736 mtag = m_tag_find(m, PACKET_TAG_IPV6_INPUT, NULL); 1737 return mtag ? (struct ip6aux *)(mtag + 1) : NULL; 1738 } 1739 1740 void 1741 ip6_delaux(struct mbuf *m) 1742 { 1743 struct m_tag *mtag; 1744 1745 mtag = m_tag_find(m, PACKET_TAG_IPV6_INPUT, NULL); 1746 if (mtag) 1747 m_tag_delete(m, mtag); 1748 } 1749 1750 /* 1751 * System control for IP6 1752 */ 1753 1754 u_char inet6ctlerrmap[PRC_NCMDS] = { 1755 0, 0, 0, 0, 1756 0, EMSGSIZE, EHOSTDOWN, EHOSTUNREACH, 1757 EHOSTUNREACH, EHOSTUNREACH, ECONNREFUSED, ECONNREFUSED, 1758 EMSGSIZE, EHOSTUNREACH, 0, 0, 1759 0, 0, 0, 0, 1760 ENOPROTOOPT 1761 };
Cache object: dbd8a28e3c16e3ce18b78ce6951b0e88
[ source navigation ] [ diff markup ] [ identifier search ] [ freetext search ] [ file search ] [ list types ] [ track identifier ]