ip6_input.c revision 194907
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: head/sys/netinet6/ip6_input.c 194907 2009-06-24 21:00:25Z rwatson $"); 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#include <sys/vimage.h> 84 85#include <net/if.h> 86#include <net/if_types.h> 87#include <net/if_dl.h> 88#include <net/route.h> 89#include <net/netisr.h> 90#include <net/pfil.h> 91#include <net/vnet.h> 92 93#include <netinet/in.h> 94#include <netinet/in_systm.h> 95#include <net/if_llatbl.h> 96#ifdef INET 97#include <netinet/ip.h> 98#include <netinet/ip_icmp.h> 99#include <netinet/vinet.h> 100#endif /* INET */ 101#include <netinet/ip6.h> 102#include <netinet6/in6_var.h> 103#include <netinet6/ip6_var.h> 104#include <netinet/in_pcb.h> 105#include <netinet/icmp6.h> 106#include <netinet6/scope6_var.h> 107#include <netinet6/in6_ifattach.h> 108#include <netinet6/nd6.h> 109#include <netinet6/vinet6.h> 110 111#ifdef IPSEC 112#include <netipsec/ipsec.h> 113#include <netinet6/ip6_ipsec.h> 114#include <netipsec/ipsec6.h> 115#endif /* IPSEC */ 116 117#include <netinet6/ip6protosw.h> 118 119extern struct domain inet6domain; 120 121u_char ip6_protox[IPPROTO_MAX]; 122 123static struct netisr_handler ip6_nh = { 124 .nh_name = "ip6", 125 .nh_handler = ip6_input, 126 .nh_proto = NETISR_IPV6, 127 .nh_policy = NETISR_POLICY_FLOW, 128}; 129 130#ifndef VIMAGE 131#ifndef VIMAGE_GLOBALS 132struct vnet_inet6 vnet_inet6_0; 133#endif 134#endif 135 136#ifdef VIMAGE_GLOBALS 137struct in6_ifaddrhead in6_ifaddrhead; 138struct ip6stat ip6stat; 139 140extern struct callout in6_tmpaddrtimer_ch; 141 142extern int dad_init; 143extern int pmtu_expire; 144extern int pmtu_probe; 145extern u_long rip6_sendspace; 146extern u_long rip6_recvspace; 147extern int icmp6errppslim; 148extern int icmp6_nodeinfo; 149extern int udp6_sendspace; 150extern int udp6_recvspace; 151#endif 152 153struct pfil_head inet6_pfil_hook; 154 155static void ip6_init2(void *); 156static struct ip6aux *ip6_setdstifaddr(struct mbuf *, struct in6_ifaddr *); 157static int ip6_hopopts_input(u_int32_t *, u_int32_t *, struct mbuf **, int *); 158#ifdef PULLDOWN_TEST 159static struct mbuf *ip6_pullexthdr(struct mbuf *, size_t, int); 160#endif 161 162#ifndef VIMAGE_GLOBALS 163static void vnet_inet6_register(void); 164 165static const vnet_modinfo_t vnet_inet6_modinfo = { 166 .vmi_id = VNET_MOD_INET6, 167 .vmi_name = "inet6", 168 .vmi_size = sizeof(struct vnet_inet6), 169 .vmi_dependson = VNET_MOD_INET /* XXX revisit - TCP/UDP needs this? */ 170}; 171 172static void 173vnet_inet6_register(void) 174{ 175 176 vnet_mod_register(&vnet_inet6_modinfo); 177} 178 179SYSINIT(inet6, SI_SUB_PROTO_BEGIN, SI_ORDER_FIRST, vnet_inet6_register, 0); 180#endif 181 182/* 183 * IP6 initialization: fill in IP6 protocol switch table. 184 * All protocols not implemented in kernel go to raw IP6 protocol handler. 185 */ 186void 187ip6_init(void) 188{ 189 INIT_VNET_INET6(curvnet); 190 struct ip6protosw *pr; 191 int i; 192 193 V_in6_maxmtu = 0; 194#ifdef IP6_AUTO_LINKLOCAL 195 V_ip6_auto_linklocal = IP6_AUTO_LINKLOCAL; 196#else 197 V_ip6_auto_linklocal = 1; /* enable by default */ 198#endif 199 TUNABLE_INT_FETCH("net.inet6.ip6.auto_linklocal", 200 &V_ip6_auto_linklocal); 201 202#ifndef IPV6FORWARDING 203#ifdef GATEWAY6 204#define IPV6FORWARDING 1 /* forward IP6 packets not for us */ 205#else 206#define IPV6FORWARDING 0 /* don't forward IP6 packets not for us */ 207#endif /* GATEWAY6 */ 208#endif /* !IPV6FORWARDING */ 209 210#ifndef IPV6_SENDREDIRECTS 211#define IPV6_SENDREDIRECTS 1 212#endif 213 214 V_ip6_forwarding = IPV6FORWARDING; /* act as router? */ 215 V_ip6_sendredirects = IPV6_SENDREDIRECTS; 216 V_ip6_defhlim = IPV6_DEFHLIM; 217 V_ip6_defmcasthlim = IPV6_DEFAULT_MULTICAST_HOPS; 218 V_ip6_accept_rtadv = 0; /* "IPV6FORWARDING ? 0 : 1" is dangerous */ 219 V_ip6_log_interval = 5; 220 V_ip6_hdrnestlimit = 15; /* How many header options will we process? */ 221 V_ip6_dad_count = 1; /* DupAddrDetectionTransmits */ 222 V_ip6_auto_flowlabel = 1; 223 V_ip6_use_deprecated = 1;/* allow deprecated addr (RFC2462 5.5.4) */ 224 V_ip6_rr_prune = 5; /* router renumbering prefix 225 * walk list every 5 sec. */ 226 V_ip6_mcast_pmtu = 0; /* enable pMTU discovery for multicast? */ 227 V_ip6_v6only = 1; 228 V_ip6_keepfaith = 0; 229 V_ip6_log_time = (time_t)0L; 230#ifdef IPSTEALTH 231 V_ip6stealth = 0; 232#endif 233 V_nd6_onlink_ns_rfc4861 = 0; /* allow 'on-link' nd6 NS (RFC 4861) */ 234 235 V_pmtu_expire = 60*10; 236 V_pmtu_probe = 60*2; 237 238 /* raw IP6 parameters */ 239 /* 240 * Nominal space allocated to a raw ip socket. 241 */ 242#define RIPV6SNDQ 8192 243#define RIPV6RCVQ 8192 244 V_rip6_sendspace = RIPV6SNDQ; 245 V_rip6_recvspace = RIPV6RCVQ; 246 247 /* ICMPV6 parameters */ 248 V_icmp6_rediraccept = 1; /* accept and process redirects */ 249 V_icmp6_redirtimeout = 10 * 60; /* 10 minutes */ 250 V_icmp6errppslim = 100; /* 100pps */ 251 /* control how to respond to NI queries */ 252 V_icmp6_nodeinfo = (ICMP6_NODEINFO_FQDNOK|ICMP6_NODEINFO_NODEADDROK); 253 254 /* UDP on IP6 parameters */ 255 V_udp6_sendspace = 9216; /* really max datagram size */ 256 V_udp6_recvspace = 40 * (1024 + sizeof(struct sockaddr_in6)); 257 /* 40 1K datagrams */ 258 V_dad_init = 0; 259 260 TAILQ_INIT(&V_in6_ifaddrhead); 261 262 scope6_init(); 263 addrsel_policy_init(); 264 nd6_init(); 265 frag6_init(); 266 267 V_ip6_desync_factor = arc4random() % MAX_TEMP_DESYNC_FACTOR; 268 269 /* Skip global initialization stuff for non-default instances. */ 270 if (!IS_DEFAULT_VNET(curvnet)) 271 return; 272 273#ifdef DIAGNOSTIC 274 if (sizeof(struct protosw) != sizeof(struct ip6protosw)) 275 panic("sizeof(protosw) != sizeof(ip6protosw)"); 276#endif 277 pr = (struct ip6protosw *)pffindproto(PF_INET6, IPPROTO_RAW, SOCK_RAW); 278 if (pr == NULL) 279 panic("ip6_init"); 280 281 /* Initialize the entire ip6_protox[] array to IPPROTO_RAW. */ 282 for (i = 0; i < IPPROTO_MAX; i++) 283 ip6_protox[i] = pr - inet6sw; 284 /* 285 * Cycle through IP protocols and put them into the appropriate place 286 * in ip6_protox[]. 287 */ 288 for (pr = (struct ip6protosw *)inet6domain.dom_protosw; 289 pr < (struct ip6protosw *)inet6domain.dom_protoswNPROTOSW; pr++) 290 if (pr->pr_domain->dom_family == PF_INET6 && 291 pr->pr_protocol && pr->pr_protocol != IPPROTO_RAW) { 292 /* Be careful to only index valid IP protocols. */ 293 if (pr->pr_protocol < IPPROTO_MAX) 294 ip6_protox[pr->pr_protocol] = pr - inet6sw; 295 } 296 297 /* Initialize packet filter hooks. */ 298 inet6_pfil_hook.ph_type = PFIL_TYPE_AF; 299 inet6_pfil_hook.ph_af = AF_INET6; 300 if ((i = pfil_head_register(&inet6_pfil_hook)) != 0) 301 printf("%s: WARNING: unable to register pfil hook, " 302 "error %d\n", __func__, i); 303 304 netisr_register(&ip6_nh); 305} 306 307#ifdef VIMAGE 308void 309ip6_destroy() 310{ 311 INIT_VNET_INET6(curvnet); 312 313 nd6_destroy(); 314 callout_drain(&V_in6_tmpaddrtimer_ch); 315} 316#endif 317 318static int 319ip6_init2_vnet(const void *unused __unused) 320{ 321 INIT_VNET_INET6(curvnet); 322 323 /* nd6_timer_init */ 324 callout_init(&V_nd6_timer_ch, 0); 325 callout_reset(&V_nd6_timer_ch, hz, nd6_timer, curvnet); 326 327 /* timer for regeneranation of temporary addresses randomize ID */ 328 callout_init(&V_in6_tmpaddrtimer_ch, 0); 329 callout_reset(&V_in6_tmpaddrtimer_ch, 330 (V_ip6_temp_preferred_lifetime - V_ip6_desync_factor - 331 V_ip6_temp_regen_advance) * hz, 332 in6_tmpaddrtimer, curvnet); 333 334 return (0); 335} 336 337static void 338ip6_init2(void *dummy) 339{ 340 341 ip6_init2_vnet(NULL); 342} 343 344/* cheat */ 345/* This must be after route_init(), which is now SI_ORDER_THIRD */ 346SYSINIT(netinet6init2, SI_SUB_PROTO_DOMAIN, SI_ORDER_MIDDLE, ip6_init2, NULL); 347 348void 349ip6_input(struct mbuf *m) 350{ 351 INIT_VNET_NET(curvnet); 352 INIT_VNET_INET6(curvnet); 353 struct ip6_hdr *ip6; 354 int off = sizeof(struct ip6_hdr), nest; 355 u_int32_t plen; 356 u_int32_t rtalert = ~0; 357 int nxt, ours = 0; 358 struct ifnet *deliverifp = NULL, *ifp = NULL; 359 struct in6_addr odst; 360 struct route_in6 rin6; 361 int srcrt = 0; 362 struct llentry *lle = NULL; 363 struct sockaddr_in6 dst6, *dst; 364 365 bzero(&rin6, sizeof(struct route_in6)); 366#ifdef IPSEC 367 /* 368 * should the inner packet be considered authentic? 369 * see comment in ah4_input(). 370 * NB: m cannot be NULL when passed to the input routine 371 */ 372 373 m->m_flags &= ~M_AUTHIPHDR; 374 m->m_flags &= ~M_AUTHIPDGM; 375 376#endif /* IPSEC */ 377 378 /* 379 * make sure we don't have onion peering information into m_tag. 380 */ 381 ip6_delaux(m); 382 383 /* 384 * mbuf statistics 385 */ 386 if (m->m_flags & M_EXT) { 387 if (m->m_next) 388 V_ip6stat.ip6s_mext2m++; 389 else 390 V_ip6stat.ip6s_mext1++; 391 } else { 392#define M2MMAX (sizeof(V_ip6stat.ip6s_m2m)/sizeof(V_ip6stat.ip6s_m2m[0])) 393 if (m->m_next) { 394 if (m->m_flags & M_LOOP) { 395 V_ip6stat.ip6s_m2m[V_loif->if_index]++; 396 } else if (m->m_pkthdr.rcvif->if_index < M2MMAX) 397 V_ip6stat.ip6s_m2m[m->m_pkthdr.rcvif->if_index]++; 398 else 399 V_ip6stat.ip6s_m2m[0]++; 400 } else 401 V_ip6stat.ip6s_m1++; 402#undef M2MMAX 403 } 404 405 /* drop the packet if IPv6 operation is disabled on the IF */ 406 if ((ND_IFINFO(m->m_pkthdr.rcvif)->flags & ND6_IFF_IFDISABLED)) { 407 m_freem(m); 408 return; 409 } 410 411 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_receive); 412 V_ip6stat.ip6s_total++; 413 414#ifndef PULLDOWN_TEST 415 /* 416 * L2 bridge code and some other code can return mbuf chain 417 * that does not conform to KAME requirement. too bad. 418 * XXX: fails to join if interface MTU > MCLBYTES. jumbogram? 419 */ 420 if (m && m->m_next != NULL && m->m_pkthdr.len < MCLBYTES) { 421 struct mbuf *n; 422 423 MGETHDR(n, M_DONTWAIT, MT_HEADER); 424 if (n) 425 M_MOVE_PKTHDR(n, m); 426 if (n && n->m_pkthdr.len > MHLEN) { 427 MCLGET(n, M_DONTWAIT); 428 if ((n->m_flags & M_EXT) == 0) { 429 m_freem(n); 430 n = NULL; 431 } 432 } 433 if (n == NULL) { 434 m_freem(m); 435 return; /* ENOBUFS */ 436 } 437 438 m_copydata(m, 0, n->m_pkthdr.len, mtod(n, caddr_t)); 439 n->m_len = n->m_pkthdr.len; 440 m_freem(m); 441 m = n; 442 } 443 IP6_EXTHDR_CHECK(m, 0, sizeof(struct ip6_hdr), /* nothing */); 444#endif 445 446 if (m->m_len < sizeof(struct ip6_hdr)) { 447 struct ifnet *inifp; 448 inifp = m->m_pkthdr.rcvif; 449 if ((m = m_pullup(m, sizeof(struct ip6_hdr))) == NULL) { 450 V_ip6stat.ip6s_toosmall++; 451 in6_ifstat_inc(inifp, ifs6_in_hdrerr); 452 return; 453 } 454 } 455 456 ip6 = mtod(m, struct ip6_hdr *); 457 458 if ((ip6->ip6_vfc & IPV6_VERSION_MASK) != IPV6_VERSION) { 459 V_ip6stat.ip6s_badvers++; 460 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_hdrerr); 461 goto bad; 462 } 463 464 V_ip6stat.ip6s_nxthist[ip6->ip6_nxt]++; 465 466 /* 467 * Check against address spoofing/corruption. 468 */ 469 if (IN6_IS_ADDR_MULTICAST(&ip6->ip6_src) || 470 IN6_IS_ADDR_UNSPECIFIED(&ip6->ip6_dst)) { 471 /* 472 * XXX: "badscope" is not very suitable for a multicast source. 473 */ 474 V_ip6stat.ip6s_badscope++; 475 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_addrerr); 476 goto bad; 477 } 478 if (IN6_IS_ADDR_MC_INTFACELOCAL(&ip6->ip6_dst) && 479 !(m->m_flags & M_LOOP)) { 480 /* 481 * In this case, the packet should come from the loopback 482 * interface. However, we cannot just check the if_flags, 483 * because ip6_mloopback() passes the "actual" interface 484 * as the outgoing/incoming interface. 485 */ 486 V_ip6stat.ip6s_badscope++; 487 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_addrerr); 488 goto bad; 489 } 490 491#ifdef ALTQ 492 if (altq_input != NULL && (*altq_input)(m, AF_INET6) == 0) { 493 /* packet is dropped by traffic conditioner */ 494 return; 495 } 496#endif 497 /* 498 * The following check is not documented in specs. A malicious 499 * party may be able to use IPv4 mapped addr to confuse tcp/udp stack 500 * and bypass security checks (act as if it was from 127.0.0.1 by using 501 * IPv6 src ::ffff:127.0.0.1). Be cautious. 502 * 503 * This check chokes if we are in an SIIT cloud. As none of BSDs 504 * support IPv4-less kernel compilation, we cannot support SIIT 505 * environment at all. So, it makes more sense for us to reject any 506 * malicious packets for non-SIIT environment, than try to do a 507 * partial support for SIIT environment. 508 */ 509 if (IN6_IS_ADDR_V4MAPPED(&ip6->ip6_src) || 510 IN6_IS_ADDR_V4MAPPED(&ip6->ip6_dst)) { 511 V_ip6stat.ip6s_badscope++; 512 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_addrerr); 513 goto bad; 514 } 515#if 0 516 /* 517 * Reject packets with IPv4 compatible addresses (auto tunnel). 518 * 519 * The code forbids auto tunnel relay case in RFC1933 (the check is 520 * stronger than RFC1933). We may want to re-enable it if mech-xx 521 * is revised to forbid relaying case. 522 */ 523 if (IN6_IS_ADDR_V4COMPAT(&ip6->ip6_src) || 524 IN6_IS_ADDR_V4COMPAT(&ip6->ip6_dst)) { 525 V_ip6stat.ip6s_badscope++; 526 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_addrerr); 527 goto bad; 528 } 529#endif 530 531 /* 532 * Run through list of hooks for input packets. 533 * 534 * NB: Beware of the destination address changing 535 * (e.g. by NAT rewriting). When this happens, 536 * tell ip6_forward to do the right thing. 537 */ 538 odst = ip6->ip6_dst; 539 540 /* Jump over all PFIL processing if hooks are not active. */ 541 if (!PFIL_HOOKED(&inet6_pfil_hook)) 542 goto passin; 543 544 if (pfil_run_hooks(&inet6_pfil_hook, &m, m->m_pkthdr.rcvif, PFIL_IN, NULL)) 545 return; 546 if (m == NULL) /* consumed by filter */ 547 return; 548 ip6 = mtod(m, struct ip6_hdr *); 549 srcrt = !IN6_ARE_ADDR_EQUAL(&odst, &ip6->ip6_dst); 550 551passin: 552 /* 553 * Disambiguate address scope zones (if there is ambiguity). 554 * We first make sure that the original source or destination address 555 * is not in our internal form for scoped addresses. Such addresses 556 * are not necessarily invalid spec-wise, but we cannot accept them due 557 * to the usage conflict. 558 * in6_setscope() then also checks and rejects the cases where src or 559 * dst are the loopback address and the receiving interface 560 * is not loopback. 561 */ 562 if (in6_clearscope(&ip6->ip6_src) || in6_clearscope(&ip6->ip6_dst)) { 563 V_ip6stat.ip6s_badscope++; /* XXX */ 564 goto bad; 565 } 566 if (in6_setscope(&ip6->ip6_src, m->m_pkthdr.rcvif, NULL) || 567 in6_setscope(&ip6->ip6_dst, m->m_pkthdr.rcvif, NULL)) { 568 V_ip6stat.ip6s_badscope++; 569 goto bad; 570 } 571 572 /* 573 * Multicast check. Assume packet is for us to avoid 574 * prematurely taking locks. 575 */ 576 if (IN6_IS_ADDR_MULTICAST(&ip6->ip6_dst)) { 577 ours = 1; 578 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_mcast); 579 deliverifp = m->m_pkthdr.rcvif; 580 goto hbhcheck; 581 } 582 583 /* 584 * Unicast check 585 */ 586 587 bzero(&dst6, sizeof(dst6)); 588 dst6.sin6_family = AF_INET6; 589 dst6.sin6_len = sizeof(struct sockaddr_in6); 590 dst6.sin6_addr = ip6->ip6_dst; 591 ifp = m->m_pkthdr.rcvif; 592 IF_AFDATA_LOCK(ifp); 593 lle = lla_lookup(LLTABLE6(ifp), 0, 594 (struct sockaddr *)&dst6); 595 IF_AFDATA_UNLOCK(ifp); 596 if ((lle != NULL) && (lle->la_flags & LLE_IFADDR)) { 597 ours = 1; 598 deliverifp = ifp; 599 LLE_RUNLOCK(lle); 600 goto hbhcheck; 601 } 602 if (lle != NULL) 603 LLE_RUNLOCK(lle); 604 605 dst = &rin6.ro_dst; 606 dst->sin6_len = sizeof(struct sockaddr_in6); 607 dst->sin6_family = AF_INET6; 608 dst->sin6_addr = ip6->ip6_dst; 609 rin6.ro_rt = rtalloc1((struct sockaddr *)dst, 0, 0); 610 if (rin6.ro_rt) 611 RT_UNLOCK(rin6.ro_rt); 612 613#define rt6_key(r) ((struct sockaddr_in6 *)((r)->rt_nodes->rn_key)) 614 615 /* 616 * Accept the packet if the forwarding interface to the destination 617 * according to the routing table is the loopback interface, 618 * unless the associated route has a gateway. 619 * Note that this approach causes to accept a packet if there is a 620 * route to the loopback interface for the destination of the packet. 621 * But we think it's even useful in some situations, e.g. when using 622 * a special daemon which wants to intercept the packet. 623 * 624 * XXX: some OSes automatically make a cloned route for the destination 625 * of an outgoing packet. If the outgoing interface of the packet 626 * is a loopback one, the kernel would consider the packet to be 627 * accepted, even if we have no such address assinged on the interface. 628 * We check the cloned flag of the route entry to reject such cases, 629 * assuming that route entries for our own addresses are not made by 630 * cloning (it should be true because in6_addloop explicitly installs 631 * the host route). However, we might have to do an explicit check 632 * while it would be less efficient. Or, should we rather install a 633 * reject route for such a case? 634 */ 635 if (rin6.ro_rt && 636 (rin6.ro_rt->rt_flags & 637 (RTF_HOST|RTF_GATEWAY)) == RTF_HOST && 638#ifdef RTF_WASCLONED 639 !(rin6.ro_rt->rt_flags & RTF_WASCLONED) && 640#endif 641#ifdef RTF_CLONED 642 !(rin6.ro_rt->rt_flags & RTF_CLONED) && 643#endif 644#if 0 645 /* 646 * The check below is redundant since the comparison of 647 * the destination and the key of the rtentry has 648 * already done through looking up the routing table. 649 */ 650 IN6_ARE_ADDR_EQUAL(&ip6->ip6_dst, 651 &rt6_key(rin6.ro_rt)->sin6_addr) 652#endif 653 rin6.ro_rt->rt_ifp->if_type == IFT_LOOP) { 654 struct in6_ifaddr *ia6 = 655 (struct in6_ifaddr *)rin6.ro_rt->rt_ifa; 656 657 /* 658 * record address information into m_tag. 659 */ 660 (void)ip6_setdstifaddr(m, ia6); 661 662 /* 663 * packets to a tentative, duplicated, or somehow invalid 664 * address must not be accepted. 665 */ 666 if (!(ia6->ia6_flags & IN6_IFF_NOTREADY)) { 667 /* this address is ready */ 668 ours = 1; 669 deliverifp = ia6->ia_ifp; /* correct? */ 670 /* Count the packet in the ip address stats */ 671 ia6->ia_ifa.if_ipackets++; 672 ia6->ia_ifa.if_ibytes += m->m_pkthdr.len; 673 goto hbhcheck; 674 } else { 675 char ip6bufs[INET6_ADDRSTRLEN]; 676 char ip6bufd[INET6_ADDRSTRLEN]; 677 /* address is not ready, so discard the packet. */ 678 nd6log((LOG_INFO, 679 "ip6_input: packet to an unready address %s->%s\n", 680 ip6_sprintf(ip6bufs, &ip6->ip6_src), 681 ip6_sprintf(ip6bufd, &ip6->ip6_dst))); 682 683 goto bad; 684 } 685 } 686 687 /* 688 * FAITH (Firewall Aided Internet Translator) 689 */ 690 if (V_ip6_keepfaith) { 691 if (rin6.ro_rt && rin6.ro_rt->rt_ifp && 692 rin6.ro_rt->rt_ifp->if_type == IFT_FAITH) { 693 /* XXX do we need more sanity checks? */ 694 ours = 1; 695 deliverifp = rin6.ro_rt->rt_ifp; /* faith */ 696 goto hbhcheck; 697 } 698 } 699 700 /* 701 * Now there is no reason to process the packet if it's not our own 702 * and we're not a router. 703 */ 704 if (!V_ip6_forwarding) { 705 V_ip6stat.ip6s_cantforward++; 706 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_discard); 707 goto bad; 708 } 709 710 hbhcheck: 711 /* 712 * record address information into m_tag, if we don't have one yet. 713 * note that we are unable to record it, if the address is not listed 714 * as our interface address (e.g. multicast addresses, addresses 715 * within FAITH prefixes and such). 716 */ 717 if (deliverifp && !ip6_getdstifaddr(m)) { 718 struct in6_ifaddr *ia6; 719 720 ia6 = in6_ifawithifp(deliverifp, &ip6->ip6_dst); 721 if (ia6) { 722 if (!ip6_setdstifaddr(m, ia6)) { 723 /* 724 * XXX maybe we should drop the packet here, 725 * as we could not provide enough information 726 * to the upper layers. 727 */ 728 } 729 ifa_free(&ia6->ia_ifa); 730 } 731 } 732 733 /* 734 * Process Hop-by-Hop options header if it's contained. 735 * m may be modified in ip6_hopopts_input(). 736 * If a JumboPayload option is included, plen will also be modified. 737 */ 738 plen = (u_int32_t)ntohs(ip6->ip6_plen); 739 if (ip6->ip6_nxt == IPPROTO_HOPOPTS) { 740 struct ip6_hbh *hbh; 741 742 if (ip6_hopopts_input(&plen, &rtalert, &m, &off)) { 743#if 0 /*touches NULL pointer*/ 744 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_discard); 745#endif 746 goto out; /* m have already been freed */ 747 } 748 749 /* adjust pointer */ 750 ip6 = mtod(m, struct ip6_hdr *); 751 752 /* 753 * if the payload length field is 0 and the next header field 754 * indicates Hop-by-Hop Options header, then a Jumbo Payload 755 * option MUST be included. 756 */ 757 if (ip6->ip6_plen == 0 && plen == 0) { 758 /* 759 * Note that if a valid jumbo payload option is 760 * contained, ip6_hopopts_input() must set a valid 761 * (non-zero) payload length to the variable plen. 762 */ 763 V_ip6stat.ip6s_badoptions++; 764 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_discard); 765 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_hdrerr); 766 icmp6_error(m, ICMP6_PARAM_PROB, 767 ICMP6_PARAMPROB_HEADER, 768 (caddr_t)&ip6->ip6_plen - (caddr_t)ip6); 769 goto out; 770 } 771#ifndef PULLDOWN_TEST 772 /* ip6_hopopts_input() ensures that mbuf is contiguous */ 773 hbh = (struct ip6_hbh *)(ip6 + 1); 774#else 775 IP6_EXTHDR_GET(hbh, struct ip6_hbh *, m, sizeof(struct ip6_hdr), 776 sizeof(struct ip6_hbh)); 777 if (hbh == NULL) { 778 V_ip6stat.ip6s_tooshort++; 779 goto out; 780 } 781#endif 782 nxt = hbh->ip6h_nxt; 783 784 /* 785 * If we are acting as a router and the packet contains a 786 * router alert option, see if we know the option value. 787 * Currently, we only support the option value for MLD, in which 788 * case we should pass the packet to the multicast routing 789 * daemon. 790 */ 791 if (rtalert != ~0) { 792 switch (rtalert) { 793 case IP6OPT_RTALERT_MLD: 794 if (V_ip6_forwarding) 795 ours = 1; 796 break; 797 default: 798 /* 799 * RFC2711 requires unrecognized values must be 800 * silently ignored. 801 */ 802 break; 803 } 804 } 805 } else 806 nxt = ip6->ip6_nxt; 807 808 /* 809 * Check that the amount of data in the buffers 810 * is as at least much as the IPv6 header would have us expect. 811 * Trim mbufs if longer than we expect. 812 * Drop packet if shorter than we expect. 813 */ 814 if (m->m_pkthdr.len - sizeof(struct ip6_hdr) < plen) { 815 V_ip6stat.ip6s_tooshort++; 816 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_truncated); 817 goto bad; 818 } 819 if (m->m_pkthdr.len > sizeof(struct ip6_hdr) + plen) { 820 if (m->m_len == m->m_pkthdr.len) { 821 m->m_len = sizeof(struct ip6_hdr) + plen; 822 m->m_pkthdr.len = sizeof(struct ip6_hdr) + plen; 823 } else 824 m_adj(m, sizeof(struct ip6_hdr) + plen - m->m_pkthdr.len); 825 } 826 827 /* 828 * Forward if desirable. 829 */ 830 if (V_ip6_mrouter && 831 IN6_IS_ADDR_MULTICAST(&ip6->ip6_dst)) { 832 /* 833 * If we are acting as a multicast router, all 834 * incoming multicast packets are passed to the 835 * kernel-level multicast forwarding function. 836 * The packet is returned (relatively) intact; if 837 * ip6_mforward() returns a non-zero value, the packet 838 * must be discarded, else it may be accepted below. 839 * 840 * XXX TODO: Check hlim and multicast scope here to avoid 841 * unnecessarily calling into ip6_mforward(). 842 */ 843 if (ip6_mforward && 844 ip6_mforward(ip6, m->m_pkthdr.rcvif, m)) { 845 IP6STAT_INC(ip6s_cantforward); 846 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_discard); 847 goto bad; 848 } 849 } else if (!ours) { 850 ip6_forward(m, srcrt); 851 goto out; 852 } 853 854 ip6 = mtod(m, struct ip6_hdr *); 855 856 /* 857 * Malicious party may be able to use IPv4 mapped addr to confuse 858 * tcp/udp stack and bypass security checks (act as if it was from 859 * 127.0.0.1 by using IPv6 src ::ffff:127.0.0.1). Be cautious. 860 * 861 * For SIIT end node behavior, you may want to disable the check. 862 * However, you will become vulnerable to attacks using IPv4 mapped 863 * source. 864 */ 865 if (IN6_IS_ADDR_V4MAPPED(&ip6->ip6_src) || 866 IN6_IS_ADDR_V4MAPPED(&ip6->ip6_dst)) { 867 V_ip6stat.ip6s_badscope++; 868 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_addrerr); 869 goto bad; 870 } 871 872 /* 873 * Tell launch routine the next header 874 */ 875 V_ip6stat.ip6s_delivered++; 876 in6_ifstat_inc(deliverifp, ifs6_in_deliver); 877 nest = 0; 878 879 while (nxt != IPPROTO_DONE) { 880 if (V_ip6_hdrnestlimit && (++nest > V_ip6_hdrnestlimit)) { 881 V_ip6stat.ip6s_toomanyhdr++; 882 goto bad; 883 } 884 885 /* 886 * protection against faulty packet - there should be 887 * more sanity checks in header chain processing. 888 */ 889 if (m->m_pkthdr.len < off) { 890 V_ip6stat.ip6s_tooshort++; 891 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_truncated); 892 goto bad; 893 } 894 895#ifdef IPSEC 896 /* 897 * enforce IPsec policy checking if we are seeing last header. 898 * note that we do not visit this with protocols with pcb layer 899 * code - like udp/tcp/raw ip. 900 */ 901 if (ip6_ipsec_input(m, nxt)) 902 goto bad; 903#endif /* IPSEC */ 904 905 /* 906 * Use mbuf flags to propagate Router Alert option to 907 * ICMPv6 layer, as hop-by-hop options have been stripped. 908 */ 909 if (nxt == IPPROTO_ICMPV6 && rtalert != ~0) 910 m->m_flags |= M_RTALERT_MLD; 911 912 nxt = (*inet6sw[ip6_protox[nxt]].pr_input)(&m, &off, nxt); 913 } 914 goto out; 915bad: 916 m_freem(m); 917out: 918 if (rin6.ro_rt) 919 RTFREE(rin6.ro_rt); 920} 921 922/* 923 * set/grab in6_ifaddr correspond to IPv6 destination address. 924 * XXX backward compatibility wrapper 925 * 926 * XXXRW: We should bump the refcount on ia6 before sticking it in the m_tag, 927 * and then bump it when the tag is copied, and release it when the tag is 928 * freed. Unfortunately, m_tags don't support deep copies (yet), so instead 929 * we just bump the ia refcount when we receive it. This should be fixed. 930 */ 931static struct ip6aux * 932ip6_setdstifaddr(struct mbuf *m, struct in6_ifaddr *ia6) 933{ 934 struct ip6aux *ip6a; 935 936 ip6a = ip6_addaux(m); 937 if (ip6a) 938 ip6a->ip6a_dstia6 = ia6; 939 return ip6a; /* NULL if failed to set */ 940} 941 942struct in6_ifaddr * 943ip6_getdstifaddr(struct mbuf *m) 944{ 945 struct ip6aux *ip6a; 946 struct in6_ifaddr *ia; 947 948 ip6a = ip6_findaux(m); 949 if (ip6a) { 950 ia = ip6a->ip6a_dstia6; 951 ifa_ref(&ia->ia_ifa); 952 return ia; 953 } else 954 return NULL; 955} 956 957/* 958 * Hop-by-Hop options header processing. If a valid jumbo payload option is 959 * included, the real payload length will be stored in plenp. 960 * 961 * rtalertp - XXX: should be stored more smart way 962 */ 963static int 964ip6_hopopts_input(u_int32_t *plenp, u_int32_t *rtalertp, 965 struct mbuf **mp, int *offp) 966{ 967 INIT_VNET_INET6(curvnet); 968 struct mbuf *m = *mp; 969 int off = *offp, hbhlen; 970 struct ip6_hbh *hbh; 971 u_int8_t *opt; 972 973 /* validation of the length of the header */ 974#ifndef PULLDOWN_TEST 975 IP6_EXTHDR_CHECK(m, off, sizeof(*hbh), -1); 976 hbh = (struct ip6_hbh *)(mtod(m, caddr_t) + off); 977 hbhlen = (hbh->ip6h_len + 1) << 3; 978 979 IP6_EXTHDR_CHECK(m, off, hbhlen, -1); 980 hbh = (struct ip6_hbh *)(mtod(m, caddr_t) + off); 981#else 982 IP6_EXTHDR_GET(hbh, struct ip6_hbh *, m, 983 sizeof(struct ip6_hdr), sizeof(struct ip6_hbh)); 984 if (hbh == NULL) { 985 V_ip6stat.ip6s_tooshort++; 986 return -1; 987 } 988 hbhlen = (hbh->ip6h_len + 1) << 3; 989 IP6_EXTHDR_GET(hbh, struct ip6_hbh *, m, sizeof(struct ip6_hdr), 990 hbhlen); 991 if (hbh == NULL) { 992 V_ip6stat.ip6s_tooshort++; 993 return -1; 994 } 995#endif 996 off += hbhlen; 997 hbhlen -= sizeof(struct ip6_hbh); 998 opt = (u_int8_t *)hbh + sizeof(struct ip6_hbh); 999 1000 if (ip6_process_hopopts(m, (u_int8_t *)hbh + sizeof(struct ip6_hbh), 1001 hbhlen, rtalertp, plenp) < 0) 1002 return (-1); 1003 1004 *offp = off; 1005 *mp = m; 1006 return (0); 1007} 1008 1009/* 1010 * Search header for all Hop-by-hop options and process each option. 1011 * This function is separate from ip6_hopopts_input() in order to 1012 * handle a case where the sending node itself process its hop-by-hop 1013 * options header. In such a case, the function is called from ip6_output(). 1014 * 1015 * The function assumes that hbh header is located right after the IPv6 header 1016 * (RFC2460 p7), opthead is pointer into data content in m, and opthead to 1017 * opthead + hbhlen is located in continuous memory region. 1018 */ 1019int 1020ip6_process_hopopts(struct mbuf *m, u_int8_t *opthead, int hbhlen, 1021 u_int32_t *rtalertp, u_int32_t *plenp) 1022{ 1023 INIT_VNET_INET6(curvnet); 1024 struct ip6_hdr *ip6; 1025 int optlen = 0; 1026 u_int8_t *opt = opthead; 1027 u_int16_t rtalert_val; 1028 u_int32_t jumboplen; 1029 const int erroff = sizeof(struct ip6_hdr) + sizeof(struct ip6_hbh); 1030 1031 for (; hbhlen > 0; hbhlen -= optlen, opt += optlen) { 1032 switch (*opt) { 1033 case IP6OPT_PAD1: 1034 optlen = 1; 1035 break; 1036 case IP6OPT_PADN: 1037 if (hbhlen < IP6OPT_MINLEN) { 1038 V_ip6stat.ip6s_toosmall++; 1039 goto bad; 1040 } 1041 optlen = *(opt + 1) + 2; 1042 break; 1043 case IP6OPT_ROUTER_ALERT: 1044 /* XXX may need check for alignment */ 1045 if (hbhlen < IP6OPT_RTALERT_LEN) { 1046 V_ip6stat.ip6s_toosmall++; 1047 goto bad; 1048 } 1049 if (*(opt + 1) != IP6OPT_RTALERT_LEN - 2) { 1050 /* XXX stat */ 1051 icmp6_error(m, ICMP6_PARAM_PROB, 1052 ICMP6_PARAMPROB_HEADER, 1053 erroff + opt + 1 - opthead); 1054 return (-1); 1055 } 1056 optlen = IP6OPT_RTALERT_LEN; 1057 bcopy((caddr_t)(opt + 2), (caddr_t)&rtalert_val, 2); 1058 *rtalertp = ntohs(rtalert_val); 1059 break; 1060 case IP6OPT_JUMBO: 1061 /* XXX may need check for alignment */ 1062 if (hbhlen < IP6OPT_JUMBO_LEN) { 1063 V_ip6stat.ip6s_toosmall++; 1064 goto bad; 1065 } 1066 if (*(opt + 1) != IP6OPT_JUMBO_LEN - 2) { 1067 /* XXX stat */ 1068 icmp6_error(m, ICMP6_PARAM_PROB, 1069 ICMP6_PARAMPROB_HEADER, 1070 erroff + opt + 1 - opthead); 1071 return (-1); 1072 } 1073 optlen = IP6OPT_JUMBO_LEN; 1074 1075 /* 1076 * IPv6 packets that have non 0 payload length 1077 * must not contain a jumbo payload option. 1078 */ 1079 ip6 = mtod(m, struct ip6_hdr *); 1080 if (ip6->ip6_plen) { 1081 V_ip6stat.ip6s_badoptions++; 1082 icmp6_error(m, ICMP6_PARAM_PROB, 1083 ICMP6_PARAMPROB_HEADER, 1084 erroff + opt - opthead); 1085 return (-1); 1086 } 1087 1088 /* 1089 * We may see jumbolen in unaligned location, so 1090 * we'd need to perform bcopy(). 1091 */ 1092 bcopy(opt + 2, &jumboplen, sizeof(jumboplen)); 1093 jumboplen = (u_int32_t)htonl(jumboplen); 1094 1095#if 1 1096 /* 1097 * if there are multiple jumbo payload options, 1098 * *plenp will be non-zero and the packet will be 1099 * rejected. 1100 * the behavior may need some debate in ipngwg - 1101 * multiple options does not make sense, however, 1102 * there's no explicit mention in specification. 1103 */ 1104 if (*plenp != 0) { 1105 V_ip6stat.ip6s_badoptions++; 1106 icmp6_error(m, ICMP6_PARAM_PROB, 1107 ICMP6_PARAMPROB_HEADER, 1108 erroff + opt + 2 - opthead); 1109 return (-1); 1110 } 1111#endif 1112 1113 /* 1114 * jumbo payload length must be larger than 65535. 1115 */ 1116 if (jumboplen <= IPV6_MAXPACKET) { 1117 V_ip6stat.ip6s_badoptions++; 1118 icmp6_error(m, ICMP6_PARAM_PROB, 1119 ICMP6_PARAMPROB_HEADER, 1120 erroff + opt + 2 - opthead); 1121 return (-1); 1122 } 1123 *plenp = jumboplen; 1124 1125 break; 1126 default: /* unknown option */ 1127 if (hbhlen < IP6OPT_MINLEN) { 1128 V_ip6stat.ip6s_toosmall++; 1129 goto bad; 1130 } 1131 optlen = ip6_unknown_opt(opt, m, 1132 erroff + opt - opthead); 1133 if (optlen == -1) 1134 return (-1); 1135 optlen += 2; 1136 break; 1137 } 1138 } 1139 1140 return (0); 1141 1142 bad: 1143 m_freem(m); 1144 return (-1); 1145} 1146 1147/* 1148 * Unknown option processing. 1149 * The third argument `off' is the offset from the IPv6 header to the option, 1150 * which is necessary if the IPv6 header the and option header and IPv6 header 1151 * is not continuous in order to return an ICMPv6 error. 1152 */ 1153int 1154ip6_unknown_opt(u_int8_t *optp, struct mbuf *m, int off) 1155{ 1156 INIT_VNET_INET6(curvnet); 1157 struct ip6_hdr *ip6; 1158 1159 switch (IP6OPT_TYPE(*optp)) { 1160 case IP6OPT_TYPE_SKIP: /* ignore the option */ 1161 return ((int)*(optp + 1)); 1162 case IP6OPT_TYPE_DISCARD: /* silently discard */ 1163 m_freem(m); 1164 return (-1); 1165 case IP6OPT_TYPE_FORCEICMP: /* send ICMP even if multicasted */ 1166 V_ip6stat.ip6s_badoptions++; 1167 icmp6_error(m, ICMP6_PARAM_PROB, ICMP6_PARAMPROB_OPTION, off); 1168 return (-1); 1169 case IP6OPT_TYPE_ICMP: /* send ICMP if not multicasted */ 1170 V_ip6stat.ip6s_badoptions++; 1171 ip6 = mtod(m, struct ip6_hdr *); 1172 if (IN6_IS_ADDR_MULTICAST(&ip6->ip6_dst) || 1173 (m->m_flags & (M_BCAST|M_MCAST))) 1174 m_freem(m); 1175 else 1176 icmp6_error(m, ICMP6_PARAM_PROB, 1177 ICMP6_PARAMPROB_OPTION, off); 1178 return (-1); 1179 } 1180 1181 m_freem(m); /* XXX: NOTREACHED */ 1182 return (-1); 1183} 1184 1185/* 1186 * Create the "control" list for this pcb. 1187 * These functions will not modify mbuf chain at all. 1188 * 1189 * With KAME mbuf chain restriction: 1190 * The routine will be called from upper layer handlers like tcp6_input(). 1191 * Thus the routine assumes that the caller (tcp6_input) have already 1192 * called IP6_EXTHDR_CHECK() and all the extension headers are located in the 1193 * very first mbuf on the mbuf chain. 1194 * 1195 * ip6_savecontrol_v4 will handle those options that are possible to be 1196 * set on a v4-mapped socket. 1197 * ip6_savecontrol will directly call ip6_savecontrol_v4 to handle those 1198 * options and handle the v6-only ones itself. 1199 */ 1200struct mbuf ** 1201ip6_savecontrol_v4(struct inpcb *inp, struct mbuf *m, struct mbuf **mp, 1202 int *v4only) 1203{ 1204 struct ip6_hdr *ip6 = mtod(m, struct ip6_hdr *); 1205 1206#ifdef SO_TIMESTAMP 1207 if ((inp->inp_socket->so_options & SO_TIMESTAMP) != 0) { 1208 struct timeval tv; 1209 1210 microtime(&tv); 1211 *mp = sbcreatecontrol((caddr_t) &tv, sizeof(tv), 1212 SCM_TIMESTAMP, SOL_SOCKET); 1213 if (*mp) 1214 mp = &(*mp)->m_next; 1215 } 1216#endif 1217 1218 if ((ip6->ip6_vfc & IPV6_VERSION_MASK) != IPV6_VERSION) { 1219 if (v4only != NULL) 1220 *v4only = 1; 1221 return (mp); 1222 } 1223 1224#define IS2292(inp, x, y) (((inp)->inp_flags & IN6P_RFC2292) ? (x) : (y)) 1225 /* RFC 2292 sec. 5 */ 1226 if ((inp->inp_flags & IN6P_PKTINFO) != 0) { 1227 struct in6_pktinfo pi6; 1228 1229 bcopy(&ip6->ip6_dst, &pi6.ipi6_addr, sizeof(struct in6_addr)); 1230 in6_clearscope(&pi6.ipi6_addr); /* XXX */ 1231 pi6.ipi6_ifindex = 1232 (m && m->m_pkthdr.rcvif) ? m->m_pkthdr.rcvif->if_index : 0; 1233 1234 *mp = sbcreatecontrol((caddr_t) &pi6, 1235 sizeof(struct in6_pktinfo), 1236 IS2292(inp, IPV6_2292PKTINFO, IPV6_PKTINFO), IPPROTO_IPV6); 1237 if (*mp) 1238 mp = &(*mp)->m_next; 1239 } 1240 1241 if ((inp->inp_flags & IN6P_HOPLIMIT) != 0) { 1242 int hlim = ip6->ip6_hlim & 0xff; 1243 1244 *mp = sbcreatecontrol((caddr_t) &hlim, sizeof(int), 1245 IS2292(inp, IPV6_2292HOPLIMIT, IPV6_HOPLIMIT), 1246 IPPROTO_IPV6); 1247 if (*mp) 1248 mp = &(*mp)->m_next; 1249 } 1250 1251 if (v4only != NULL) 1252 *v4only = 0; 1253 return (mp); 1254} 1255 1256void 1257ip6_savecontrol(struct inpcb *in6p, struct mbuf *m, struct mbuf **mp) 1258{ 1259 struct ip6_hdr *ip6 = mtod(m, struct ip6_hdr *); 1260 int v4only = 0; 1261 1262 mp = ip6_savecontrol_v4(in6p, m, mp, &v4only); 1263 if (v4only) 1264 return; 1265 1266 if ((in6p->inp_flags & IN6P_TCLASS) != 0) { 1267 u_int32_t flowinfo; 1268 int tclass; 1269 1270 flowinfo = (u_int32_t)ntohl(ip6->ip6_flow & IPV6_FLOWINFO_MASK); 1271 flowinfo >>= 20; 1272 1273 tclass = flowinfo & 0xff; 1274 *mp = sbcreatecontrol((caddr_t) &tclass, sizeof(tclass), 1275 IPV6_TCLASS, IPPROTO_IPV6); 1276 if (*mp) 1277 mp = &(*mp)->m_next; 1278 } 1279 1280 /* 1281 * IPV6_HOPOPTS socket option. Recall that we required super-user 1282 * privilege for the option (see ip6_ctloutput), but it might be too 1283 * strict, since there might be some hop-by-hop options which can be 1284 * returned to normal user. 1285 * See also RFC 2292 section 6 (or RFC 3542 section 8). 1286 */ 1287 if ((in6p->inp_flags & IN6P_HOPOPTS) != 0) { 1288 /* 1289 * Check if a hop-by-hop options header is contatined in the 1290 * received packet, and if so, store the options as ancillary 1291 * data. Note that a hop-by-hop options header must be 1292 * just after the IPv6 header, which is assured through the 1293 * IPv6 input processing. 1294 */ 1295 if (ip6->ip6_nxt == IPPROTO_HOPOPTS) { 1296 struct ip6_hbh *hbh; 1297 int hbhlen = 0; 1298#ifdef PULLDOWN_TEST 1299 struct mbuf *ext; 1300#endif 1301 1302#ifndef PULLDOWN_TEST 1303 hbh = (struct ip6_hbh *)(ip6 + 1); 1304 hbhlen = (hbh->ip6h_len + 1) << 3; 1305#else 1306 ext = ip6_pullexthdr(m, sizeof(struct ip6_hdr), 1307 ip6->ip6_nxt); 1308 if (ext == NULL) { 1309 V_ip6stat.ip6s_tooshort++; 1310 return; 1311 } 1312 hbh = mtod(ext, struct ip6_hbh *); 1313 hbhlen = (hbh->ip6h_len + 1) << 3; 1314 if (hbhlen != ext->m_len) { 1315 m_freem(ext); 1316 V_ip6stat.ip6s_tooshort++; 1317 return; 1318 } 1319#endif 1320 1321 /* 1322 * XXX: We copy the whole header even if a 1323 * jumbo payload option is included, the option which 1324 * is to be removed before returning according to 1325 * RFC2292. 1326 * Note: this constraint is removed in RFC3542 1327 */ 1328 *mp = sbcreatecontrol((caddr_t)hbh, hbhlen, 1329 IS2292(in6p, IPV6_2292HOPOPTS, IPV6_HOPOPTS), 1330 IPPROTO_IPV6); 1331 if (*mp) 1332 mp = &(*mp)->m_next; 1333#ifdef PULLDOWN_TEST 1334 m_freem(ext); 1335#endif 1336 } 1337 } 1338 1339 if ((in6p->inp_flags & (IN6P_RTHDR | IN6P_DSTOPTS)) != 0) { 1340 int nxt = ip6->ip6_nxt, off = sizeof(struct ip6_hdr); 1341 1342 /* 1343 * Search for destination options headers or routing 1344 * header(s) through the header chain, and stores each 1345 * header as ancillary data. 1346 * Note that the order of the headers remains in 1347 * the chain of ancillary data. 1348 */ 1349 while (1) { /* is explicit loop prevention necessary? */ 1350 struct ip6_ext *ip6e = NULL; 1351 int elen; 1352#ifdef PULLDOWN_TEST 1353 struct mbuf *ext = NULL; 1354#endif 1355 1356 /* 1357 * if it is not an extension header, don't try to 1358 * pull it from the chain. 1359 */ 1360 switch (nxt) { 1361 case IPPROTO_DSTOPTS: 1362 case IPPROTO_ROUTING: 1363 case IPPROTO_HOPOPTS: 1364 case IPPROTO_AH: /* is it possible? */ 1365 break; 1366 default: 1367 goto loopend; 1368 } 1369 1370#ifndef PULLDOWN_TEST 1371 if (off + sizeof(*ip6e) > m->m_len) 1372 goto loopend; 1373 ip6e = (struct ip6_ext *)(mtod(m, caddr_t) + off); 1374 if (nxt == IPPROTO_AH) 1375 elen = (ip6e->ip6e_len + 2) << 2; 1376 else 1377 elen = (ip6e->ip6e_len + 1) << 3; 1378 if (off + elen > m->m_len) 1379 goto loopend; 1380#else 1381 ext = ip6_pullexthdr(m, off, nxt); 1382 if (ext == NULL) { 1383 V_ip6stat.ip6s_tooshort++; 1384 return; 1385 } 1386 ip6e = mtod(ext, struct ip6_ext *); 1387 if (nxt == IPPROTO_AH) 1388 elen = (ip6e->ip6e_len + 2) << 2; 1389 else 1390 elen = (ip6e->ip6e_len + 1) << 3; 1391 if (elen != ext->m_len) { 1392 m_freem(ext); 1393 V_ip6stat.ip6s_tooshort++; 1394 return; 1395 } 1396#endif 1397 1398 switch (nxt) { 1399 case IPPROTO_DSTOPTS: 1400 if (!(in6p->inp_flags & IN6P_DSTOPTS)) 1401 break; 1402 1403 *mp = sbcreatecontrol((caddr_t)ip6e, elen, 1404 IS2292(in6p, 1405 IPV6_2292DSTOPTS, IPV6_DSTOPTS), 1406 IPPROTO_IPV6); 1407 if (*mp) 1408 mp = &(*mp)->m_next; 1409 break; 1410 case IPPROTO_ROUTING: 1411 if (!in6p->inp_flags & IN6P_RTHDR) 1412 break; 1413 1414 *mp = sbcreatecontrol((caddr_t)ip6e, elen, 1415 IS2292(in6p, IPV6_2292RTHDR, IPV6_RTHDR), 1416 IPPROTO_IPV6); 1417 if (*mp) 1418 mp = &(*mp)->m_next; 1419 break; 1420 case IPPROTO_HOPOPTS: 1421 case IPPROTO_AH: /* is it possible? */ 1422 break; 1423 1424 default: 1425 /* 1426 * other cases have been filtered in the above. 1427 * none will visit this case. here we supply 1428 * the code just in case (nxt overwritten or 1429 * other cases). 1430 */ 1431#ifdef PULLDOWN_TEST 1432 m_freem(ext); 1433#endif 1434 goto loopend; 1435 1436 } 1437 1438 /* proceed with the next header. */ 1439 off += elen; 1440 nxt = ip6e->ip6e_nxt; 1441 ip6e = NULL; 1442#ifdef PULLDOWN_TEST 1443 m_freem(ext); 1444 ext = NULL; 1445#endif 1446 } 1447 loopend: 1448 ; 1449 } 1450} 1451#undef IS2292 1452 1453void 1454ip6_notify_pmtu(struct inpcb *in6p, struct sockaddr_in6 *dst, u_int32_t *mtu) 1455{ 1456 struct socket *so; 1457 struct mbuf *m_mtu; 1458 struct ip6_mtuinfo mtuctl; 1459 1460 so = in6p->inp_socket; 1461 1462 if (mtu == NULL) 1463 return; 1464 1465#ifdef DIAGNOSTIC 1466 if (so == NULL) /* I believe this is impossible */ 1467 panic("ip6_notify_pmtu: socket is NULL"); 1468#endif 1469 1470 bzero(&mtuctl, sizeof(mtuctl)); /* zero-clear for safety */ 1471 mtuctl.ip6m_mtu = *mtu; 1472 mtuctl.ip6m_addr = *dst; 1473 if (sa6_recoverscope(&mtuctl.ip6m_addr)) 1474 return; 1475 1476 if ((m_mtu = sbcreatecontrol((caddr_t)&mtuctl, sizeof(mtuctl), 1477 IPV6_PATHMTU, IPPROTO_IPV6)) == NULL) 1478 return; 1479 1480 if (sbappendaddr(&so->so_rcv, (struct sockaddr *)dst, NULL, m_mtu) 1481 == 0) { 1482 m_freem(m_mtu); 1483 /* XXX: should count statistics */ 1484 } else 1485 sorwakeup(so); 1486 1487 return; 1488} 1489 1490#ifdef PULLDOWN_TEST 1491/* 1492 * pull single extension header from mbuf chain. returns single mbuf that 1493 * contains the result, or NULL on error. 1494 */ 1495static struct mbuf * 1496ip6_pullexthdr(struct mbuf *m, size_t off, int nxt) 1497{ 1498 struct ip6_ext ip6e; 1499 size_t elen; 1500 struct mbuf *n; 1501 1502#ifdef DIAGNOSTIC 1503 switch (nxt) { 1504 case IPPROTO_DSTOPTS: 1505 case IPPROTO_ROUTING: 1506 case IPPROTO_HOPOPTS: 1507 case IPPROTO_AH: /* is it possible? */ 1508 break; 1509 default: 1510 printf("ip6_pullexthdr: invalid nxt=%d\n", nxt); 1511 } 1512#endif 1513 1514 m_copydata(m, off, sizeof(ip6e), (caddr_t)&ip6e); 1515 if (nxt == IPPROTO_AH) 1516 elen = (ip6e.ip6e_len + 2) << 2; 1517 else 1518 elen = (ip6e.ip6e_len + 1) << 3; 1519 1520 MGET(n, M_DONTWAIT, MT_DATA); 1521 if (n && elen >= MLEN) { 1522 MCLGET(n, M_DONTWAIT); 1523 if ((n->m_flags & M_EXT) == 0) { 1524 m_free(n); 1525 n = NULL; 1526 } 1527 } 1528 if (!n) 1529 return NULL; 1530 1531 n->m_len = 0; 1532 if (elen >= M_TRAILINGSPACE(n)) { 1533 m_free(n); 1534 return NULL; 1535 } 1536 1537 m_copydata(m, off, elen, mtod(n, caddr_t)); 1538 n->m_len = elen; 1539 return n; 1540} 1541#endif 1542 1543/* 1544 * Get pointer to the previous header followed by the header 1545 * currently processed. 1546 * XXX: This function supposes that 1547 * M includes all headers, 1548 * the next header field and the header length field of each header 1549 * are valid, and 1550 * the sum of each header length equals to OFF. 1551 * Because of these assumptions, this function must be called very 1552 * carefully. Moreover, it will not be used in the near future when 1553 * we develop `neater' mechanism to process extension headers. 1554 */ 1555char * 1556ip6_get_prevhdr(struct mbuf *m, int off) 1557{ 1558 struct ip6_hdr *ip6 = mtod(m, struct ip6_hdr *); 1559 1560 if (off == sizeof(struct ip6_hdr)) 1561 return (&ip6->ip6_nxt); 1562 else { 1563 int len, nxt; 1564 struct ip6_ext *ip6e = NULL; 1565 1566 nxt = ip6->ip6_nxt; 1567 len = sizeof(struct ip6_hdr); 1568 while (len < off) { 1569 ip6e = (struct ip6_ext *)(mtod(m, caddr_t) + len); 1570 1571 switch (nxt) { 1572 case IPPROTO_FRAGMENT: 1573 len += sizeof(struct ip6_frag); 1574 break; 1575 case IPPROTO_AH: 1576 len += (ip6e->ip6e_len + 2) << 2; 1577 break; 1578 default: 1579 len += (ip6e->ip6e_len + 1) << 3; 1580 break; 1581 } 1582 nxt = ip6e->ip6e_nxt; 1583 } 1584 if (ip6e) 1585 return (&ip6e->ip6e_nxt); 1586 else 1587 return NULL; 1588 } 1589} 1590 1591/* 1592 * get next header offset. m will be retained. 1593 */ 1594int 1595ip6_nexthdr(struct mbuf *m, int off, int proto, int *nxtp) 1596{ 1597 struct ip6_hdr ip6; 1598 struct ip6_ext ip6e; 1599 struct ip6_frag fh; 1600 1601 /* just in case */ 1602 if (m == NULL) 1603 panic("ip6_nexthdr: m == NULL"); 1604 if ((m->m_flags & M_PKTHDR) == 0 || m->m_pkthdr.len < off) 1605 return -1; 1606 1607 switch (proto) { 1608 case IPPROTO_IPV6: 1609 if (m->m_pkthdr.len < off + sizeof(ip6)) 1610 return -1; 1611 m_copydata(m, off, sizeof(ip6), (caddr_t)&ip6); 1612 if (nxtp) 1613 *nxtp = ip6.ip6_nxt; 1614 off += sizeof(ip6); 1615 return off; 1616 1617 case IPPROTO_FRAGMENT: 1618 /* 1619 * terminate parsing if it is not the first fragment, 1620 * it does not make sense to parse through it. 1621 */ 1622 if (m->m_pkthdr.len < off + sizeof(fh)) 1623 return -1; 1624 m_copydata(m, off, sizeof(fh), (caddr_t)&fh); 1625 /* IP6F_OFF_MASK = 0xfff8(BigEndian), 0xf8ff(LittleEndian) */ 1626 if (fh.ip6f_offlg & IP6F_OFF_MASK) 1627 return -1; 1628 if (nxtp) 1629 *nxtp = fh.ip6f_nxt; 1630 off += sizeof(struct ip6_frag); 1631 return off; 1632 1633 case IPPROTO_AH: 1634 if (m->m_pkthdr.len < off + sizeof(ip6e)) 1635 return -1; 1636 m_copydata(m, off, sizeof(ip6e), (caddr_t)&ip6e); 1637 if (nxtp) 1638 *nxtp = ip6e.ip6e_nxt; 1639 off += (ip6e.ip6e_len + 2) << 2; 1640 return off; 1641 1642 case IPPROTO_HOPOPTS: 1643 case IPPROTO_ROUTING: 1644 case IPPROTO_DSTOPTS: 1645 if (m->m_pkthdr.len < off + sizeof(ip6e)) 1646 return -1; 1647 m_copydata(m, off, sizeof(ip6e), (caddr_t)&ip6e); 1648 if (nxtp) 1649 *nxtp = ip6e.ip6e_nxt; 1650 off += (ip6e.ip6e_len + 1) << 3; 1651 return off; 1652 1653 case IPPROTO_NONE: 1654 case IPPROTO_ESP: 1655 case IPPROTO_IPCOMP: 1656 /* give up */ 1657 return -1; 1658 1659 default: 1660 return -1; 1661 } 1662 1663 return -1; 1664} 1665 1666/* 1667 * get offset for the last header in the chain. m will be kept untainted. 1668 */ 1669int 1670ip6_lasthdr(struct mbuf *m, int off, int proto, int *nxtp) 1671{ 1672 int newoff; 1673 int nxt; 1674 1675 if (!nxtp) { 1676 nxt = -1; 1677 nxtp = &nxt; 1678 } 1679 while (1) { 1680 newoff = ip6_nexthdr(m, off, proto, nxtp); 1681 if (newoff < 0) 1682 return off; 1683 else if (newoff < off) 1684 return -1; /* invalid */ 1685 else if (newoff == off) 1686 return newoff; 1687 1688 off = newoff; 1689 proto = *nxtp; 1690 } 1691} 1692 1693struct ip6aux * 1694ip6_addaux(struct mbuf *m) 1695{ 1696 struct m_tag *mtag; 1697 1698 mtag = m_tag_find(m, PACKET_TAG_IPV6_INPUT, NULL); 1699 if (!mtag) { 1700 mtag = m_tag_get(PACKET_TAG_IPV6_INPUT, sizeof(struct ip6aux), 1701 M_NOWAIT); 1702 if (mtag) { 1703 m_tag_prepend(m, mtag); 1704 bzero(mtag + 1, sizeof(struct ip6aux)); 1705 } 1706 } 1707 return mtag ? (struct ip6aux *)(mtag + 1) : NULL; 1708} 1709 1710struct ip6aux * 1711ip6_findaux(struct mbuf *m) 1712{ 1713 struct m_tag *mtag; 1714 1715 mtag = m_tag_find(m, PACKET_TAG_IPV6_INPUT, NULL); 1716 return mtag ? (struct ip6aux *)(mtag + 1) : NULL; 1717} 1718 1719void 1720ip6_delaux(struct mbuf *m) 1721{ 1722 struct m_tag *mtag; 1723 1724 mtag = m_tag_find(m, PACKET_TAG_IPV6_INPUT, NULL); 1725 if (mtag) 1726 m_tag_delete(m, mtag); 1727} 1728 1729/* 1730 * System control for IP6 1731 */ 1732 1733u_char inet6ctlerrmap[PRC_NCMDS] = { 1734 0, 0, 0, 0, 1735 0, EMSGSIZE, EHOSTDOWN, EHOSTUNREACH, 1736 EHOSTUNREACH, EHOSTUNREACH, ECONNREFUSED, ECONNREFUSED, 1737 EMSGSIZE, EHOSTUNREACH, 0, 0, 1738 0, 0, 0, 0, 1739 ENOPROTOOPT 1740}; 1741