1 /* $FreeBSD: src/sys/netinet6/ip6_input.c,v 1.11.2.15 2003/01/24 05:11:35 sam Exp $ */ 2 /* $KAME: ip6_input.c,v 1.259 2002/01/21 04:58:09 jinmei Exp $ */ 3 4 /* 5 * Copyright (C) 1995, 1996, 1997, and 1998 WIDE Project. 6 * All rights reserved. 7 * 8 * Redistribution and use in source and binary forms, with or without 9 * modification, are permitted provided that the following conditions 10 * are met: 11 * 1. Redistributions of source code must retain the above copyright 12 * notice, this list of conditions and the following disclaimer. 13 * 2. Redistributions in binary form must reproduce the above copyright 14 * notice, this list of conditions and the following disclaimer in the 15 * documentation and/or other materials provided with the distribution. 16 * 3. Neither the name of the project nor the names of its contributors 17 * may be used to endorse or promote products derived from this software 18 * without specific prior written permission. 19 * 20 * THIS SOFTWARE IS PROVIDED BY THE PROJECT AND CONTRIBUTORS ``AS IS'' AND 21 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE 22 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE 23 * ARE DISCLAIMED. IN NO EVENT SHALL THE PROJECT OR CONTRIBUTORS BE LIABLE 24 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL 25 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS 26 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) 27 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT 28 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY 29 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF 30 * SUCH DAMAGE. 31 */ 32 33 /* 34 * Copyright (c) 1982, 1986, 1988, 1993 35 * The Regents of the University of California. All rights reserved. 36 * 37 * Redistribution and use in source and binary forms, with or without 38 * modification, are permitted provided that the following conditions 39 * are met: 40 * 1. Redistributions of source code must retain the above copyright 41 * notice, this list of conditions and the following disclaimer. 42 * 2. Redistributions in binary form must reproduce the above copyright 43 * notice, this list of conditions and the following disclaimer in the 44 * documentation and/or other materials provided with the distribution. 45 * 3. Neither the name of the University nor the names of its contributors 46 * may be used to endorse or promote products derived from this software 47 * without specific prior written permission. 48 * 49 * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND 50 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE 51 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE 52 * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE 53 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL 54 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS 55 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) 56 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT 57 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY 58 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF 59 * SUCH DAMAGE. 60 * 61 * @(#)ip_input.c 8.2 (Berkeley) 1/4/94 62 */ 63 64 #include "opt_ip6fw.h" 65 #include "opt_inet.h" 66 #include "opt_inet6.h" 67 68 #include <sys/param.h> 69 #include <sys/systm.h> 70 #include <sys/malloc.h> 71 #include <sys/mbuf.h> 72 #include <sys/domain.h> 73 #include <sys/protosw.h> 74 #include <sys/socket.h> 75 #include <sys/socketvar.h> 76 #include <sys/errno.h> 77 #include <sys/time.h> 78 #include <sys/kernel.h> 79 #include <sys/syslog.h> 80 #include <sys/proc.h> 81 #include <sys/priv.h> 82 83 #include <net/if.h> 84 #include <net/if_types.h> 85 #include <net/if_dl.h> 86 #include <net/route.h> 87 #include <net/netisr.h> 88 #include <net/pfil.h> 89 90 #include <sys/msgport2.h> 91 #include <net/netmsg2.h> 92 #include <net/netisr2.h> 93 94 #include <netinet/in.h> 95 #include <netinet/in_systm.h> 96 #ifdef INET 97 #include <netinet/ip.h> 98 #include <netinet/ip_icmp.h> 99 #endif /* INET */ 100 #include <netinet/ip6.h> 101 #include <netinet6/in6_var.h> 102 #include <netinet6/ip6_var.h> 103 #include <netinet/in_pcb.h> 104 #include <netinet/icmp6.h> 105 #include <netinet6/scope6_var.h> 106 #include <netinet6/in6_ifattach.h> 107 #include <netinet6/nd6.h> 108 109 #include <net/ip6fw/ip6_fw.h> 110 111 #include <netinet6/ip6protosw.h> 112 113 #include <net/net_osdep.h> 114 115 extern struct domain inet6domain; 116 extern struct protosw inet6sw[]; 117 118 u_char ip6_protox[IPPROTO_MAX]; 119 struct in6_ifaddr *in6_ifaddr; 120 121 int ip6_forward_srcrt; /* XXX */ 122 int ip6_sourcecheck; /* XXX */ 123 int ip6_sourcecheck_interval; /* XXX */ 124 125 int ip6_ours_check_algorithm; 126 127 struct pfil_head inet6_pfil_hook; 128 129 /* firewall hooks */ 130 ip6_fw_chk_t *ip6_fw_chk_ptr; 131 ip6_fw_ctl_t *ip6_fw_ctl_ptr; 132 int ip6_fw_enable = 1; 133 134 struct ip6stat ip6stat; 135 136 static void ip6_init2 (void *); 137 static struct ip6aux *ip6_setdstifaddr (struct mbuf *, struct in6_ifaddr *); 138 static int ip6_hopopts_input (u_int32_t *, u_int32_t *, struct mbuf **, int *); 139 static void ip6_input(netmsg_t msg); 140 #ifdef PULLDOWN_TEST 141 static struct mbuf *ip6_pullexthdr (struct mbuf *, size_t, int); 142 #endif 143 static void transport6_processing_handler(netmsg_t netmsg); 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 protosw *pr; 153 int i; 154 struct timeval tv; 155 156 pr = pffindproto(PF_INET6, IPPROTO_RAW, SOCK_RAW); 157 if (pr == NULL) 158 panic("ip6_init"); 159 for (i = 0; i < IPPROTO_MAX; i++) 160 ip6_protox[i] = pr - inet6sw; 161 for (pr = inet6domain.dom_protosw; 162 pr < inet6domain.dom_protoswNPROTOSW; pr++) 163 if (pr->pr_domain->dom_family == PF_INET6 && 164 pr->pr_protocol && pr->pr_protocol != IPPROTO_RAW) 165 ip6_protox[pr->pr_protocol] = pr - inet6sw; 166 167 inet6_pfil_hook.ph_type = PFIL_TYPE_AF; 168 inet6_pfil_hook.ph_af = AF_INET6; 169 if ((i = pfil_head_register(&inet6_pfil_hook)) != 0) { 170 kprintf("%s: WARNING: unable to register pfil hook, " 171 "error %d\n", __func__, i); 172 } 173 174 netisr_register(NETISR_IPV6, ip6_input, NULL); /* XXX hashfn */ 175 scope6_init(); 176 addrsel_policy_init(); 177 nd6_init(); 178 frag6_init(); 179 /* 180 * in many cases, random() here does NOT return random number 181 * as initialization during bootstrap time occur in fixed order. 182 */ 183 microtime(&tv); 184 ip6_flow_seq = krandom() ^ tv.tv_usec; 185 microtime(&tv); 186 ip6_desync_factor = (krandom() ^ tv.tv_usec) % MAX_TEMP_DESYNC_FACTOR; 187 } 188 189 static void 190 ip6_init2(void *dummy) 191 { 192 nd6_timer_init(); 193 in6_tmpaddrtimer_init(); 194 } 195 196 /* cheat */ 197 /* This must be after route_init(), which is now SI_ORDER_THIRD */ 198 SYSINIT(netinet6init2, SI_SUB_PROTO_DOMAIN, SI_ORDER_MIDDLE, ip6_init2, NULL); 199 200 extern struct route_in6 ip6_forward_rt; 201 202 static void 203 ip6_input(netmsg_t msg) 204 { 205 struct mbuf *m = msg->packet.nm_packet; 206 struct ip6_hdr *ip6; 207 int off = sizeof(struct ip6_hdr), nest; 208 u_int32_t plen; 209 u_int32_t rtalert = ~0; 210 int nxt, ours = 0, rh_present = 0; 211 struct ifnet *deliverifp = NULL; 212 struct in6_addr odst; 213 int srcrt = 0; 214 215 /* 216 * make sure we don't have onion peering information into m_aux. 217 */ 218 ip6_delaux(m); 219 220 /* 221 * mbuf statistics 222 */ 223 if (m->m_flags & M_EXT) { 224 if (m->m_next) 225 ip6stat.ip6s_mext2m++; 226 else 227 ip6stat.ip6s_mext1++; 228 } else { 229 #define M2MMAX NELEM(ip6stat.ip6s_m2m) 230 if (m->m_next) { 231 if (m->m_flags & M_LOOP) { 232 ip6stat.ip6s_m2m[loif->if_index]++; /* XXX */ 233 } else if (m->m_pkthdr.rcvif->if_index < M2MMAX) 234 ip6stat.ip6s_m2m[m->m_pkthdr.rcvif->if_index]++; 235 else 236 ip6stat.ip6s_m2m[0]++; 237 } else 238 ip6stat.ip6s_m1++; 239 #undef M2MMAX 240 } 241 242 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_receive); 243 ip6stat.ip6s_total++; 244 245 #ifndef PULLDOWN_TEST 246 /* 247 * L2 bridge code and some other code can return mbuf chain 248 * that does not conform to KAME requirement. too bad. 249 * XXX: fails to join if interface MTU > MCLBYTES. jumbogram? 250 */ 251 if (m && m->m_next != NULL && m->m_pkthdr.len < MCLBYTES) { 252 struct mbuf *n; 253 254 n = m_getb(m->m_pkthdr.len, M_NOWAIT, MT_HEADER, M_PKTHDR); 255 if (n == NULL) 256 goto bad; 257 M_MOVE_PKTHDR(n, m); 258 259 m_copydata(m, 0, n->m_pkthdr.len, mtod(n, caddr_t)); 260 n->m_len = n->m_pkthdr.len; 261 m_freem(m); 262 m = n; 263 } 264 IP6_EXTHDR_CHECK_VOIDRET(m, 0, sizeof(struct ip6_hdr)); 265 #endif 266 267 if (m->m_len < sizeof(struct ip6_hdr)) { 268 struct ifnet *inifp; 269 inifp = m->m_pkthdr.rcvif; 270 if ((m = m_pullup(m, sizeof(struct ip6_hdr))) == NULL) { 271 ip6stat.ip6s_toosmall++; 272 in6_ifstat_inc(inifp, ifs6_in_hdrerr); 273 goto bad2; 274 } 275 } 276 277 ip6 = mtod(m, struct ip6_hdr *); 278 279 if ((ip6->ip6_vfc & IPV6_VERSION_MASK) != IPV6_VERSION) { 280 ip6stat.ip6s_badvers++; 281 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_hdrerr); 282 goto bad; 283 } 284 285 /* 286 * Run through list of hooks for input packets. 287 * 288 * NB: Beware of the destination address changing 289 * (e.g. by NAT rewriting). When this happens, 290 * tell ip6_forward to do the right thing. 291 */ 292 if (pfil_has_hooks(&inet6_pfil_hook)) { 293 odst = ip6->ip6_dst; 294 if (pfil_run_hooks(&inet6_pfil_hook, &m, 295 m->m_pkthdr.rcvif, PFIL_IN)) { 296 goto bad2; 297 } 298 if (m == NULL) /* consumed by filter */ 299 goto bad2; 300 ip6 = mtod(m, struct ip6_hdr *); 301 srcrt = !IN6_ARE_ADDR_EQUAL(&odst, &ip6->ip6_dst); 302 } 303 304 ip6stat.ip6s_nxthist[ip6->ip6_nxt]++; 305 306 #ifdef ALTQ 307 if (altq_input != NULL && (*altq_input)(m, AF_INET6) == 0) { 308 /* packet is dropped by traffic conditioner */ 309 return; 310 } 311 #endif 312 313 /* 314 * Check with the firewall... 315 */ 316 if (ip6_fw_enable && ip6_fw_chk_ptr) { 317 u_short port = 0; 318 /* If ipfw says divert, we have to just drop packet */ 319 /* use port as a dummy argument */ 320 if ((*ip6_fw_chk_ptr)(&ip6, NULL, &port, &m)) { 321 m_freem(m); 322 m = NULL; 323 } 324 if (!m) 325 goto bad2; 326 } 327 328 /* 329 * Check against address spoofing/corruption. 330 */ 331 if (IN6_IS_ADDR_MULTICAST(&ip6->ip6_src) || 332 IN6_IS_ADDR_UNSPECIFIED(&ip6->ip6_dst)) { 333 /* 334 * XXX: "badscope" is not very suitable for a multicast source. 335 */ 336 ip6stat.ip6s_badscope++; 337 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_addrerr); 338 goto bad; 339 } 340 if ((IN6_IS_ADDR_LOOPBACK(&ip6->ip6_src) || 341 IN6_IS_ADDR_LOOPBACK(&ip6->ip6_dst)) && 342 (m->m_pkthdr.rcvif->if_flags & IFF_LOOPBACK) == 0) { 343 ip6stat.ip6s_badscope++; 344 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_addrerr); 345 goto bad; 346 } 347 348 /* 349 * The following check is not documented in specs. A malicious 350 * party may be able to use IPv4 mapped addr to confuse tcp/udp stack 351 * and bypass security checks (act as if it was from 127.0.0.1 by using 352 * IPv6 src ::ffff:127.0.0.1). Be cautious. 353 * 354 * This check chokes if we are in an SIIT cloud. As none of BSDs 355 * support IPv4-less kernel compilation, we cannot support SIIT 356 * environment at all. So, it makes more sense for us to reject any 357 * malicious packets for non-SIIT environment, than try to do a 358 * partial support for SIIT environment. 359 */ 360 if (IN6_IS_ADDR_V4MAPPED(&ip6->ip6_src) || 361 IN6_IS_ADDR_V4MAPPED(&ip6->ip6_dst)) { 362 ip6stat.ip6s_badscope++; 363 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_addrerr); 364 goto bad; 365 } 366 #if 0 367 /* 368 * Reject packets with IPv4 compatible addresses (auto tunnel). 369 * 370 * The code forbids auto tunnel relay case in RFC1933 (the check is 371 * stronger than RFC1933). We may want to re-enable it if mech-xx 372 * is revised to forbid relaying case. 373 */ 374 if (IN6_IS_ADDR_V4COMPAT(&ip6->ip6_src) || 375 IN6_IS_ADDR_V4COMPAT(&ip6->ip6_dst)) { 376 ip6stat.ip6s_badscope++; 377 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_addrerr); 378 goto bad; 379 } 380 #endif 381 382 /* drop packets if interface ID portion is already filled */ 383 if ((m->m_pkthdr.rcvif->if_flags & IFF_LOOPBACK) == 0) { 384 if (IN6_IS_SCOPE_LINKLOCAL(&ip6->ip6_src) && 385 ip6->ip6_src.s6_addr16[1]) { 386 ip6stat.ip6s_badscope++; 387 goto bad; 388 } 389 if ((IN6_IS_ADDR_MC_INTFACELOCAL(&ip6->ip6_dst) || 390 IN6_IS_SCOPE_LINKLOCAL(&ip6->ip6_dst)) && 391 ip6->ip6_dst.s6_addr16[1]) { 392 ip6stat.ip6s_badscope++; 393 goto bad; 394 } 395 } 396 397 if (IN6_IS_SCOPE_LINKLOCAL(&ip6->ip6_src)) 398 ip6->ip6_src.s6_addr16[1] 399 = htons(m->m_pkthdr.rcvif->if_index); 400 if (IN6_IS_SCOPE_LINKLOCAL(&ip6->ip6_dst)) 401 ip6->ip6_dst.s6_addr16[1] 402 = htons(m->m_pkthdr.rcvif->if_index); 403 404 /* 405 * Multicast check 406 * 407 * WARNING! For general subnet proxying the interface hw will 408 * likely filter out the multicast solicitations, so 409 * the interface must be in promiscuous mode. 410 */ 411 if (IN6_IS_ADDR_MULTICAST(&ip6->ip6_dst)) { 412 ours = 1; 413 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_mcast); 414 deliverifp = m->m_pkthdr.rcvif; 415 goto hbhcheck; 416 } 417 418 /* 419 * Unicast check 420 */ 421 if (ip6_forward_rt.ro_rt != NULL && 422 (ip6_forward_rt.ro_rt->rt_flags & RTF_UP) && 423 IN6_ARE_ADDR_EQUAL(&ip6->ip6_dst, 424 &((struct sockaddr_in6 *)(&ip6_forward_rt.ro_dst))->sin6_addr)) { 425 ip6stat.ip6s_forward_cachehit++; 426 } else { 427 struct sockaddr_in6 *dst6; 428 429 if (ip6_forward_rt.ro_rt) { 430 /* route is down or destination is different */ 431 ip6stat.ip6s_forward_cachemiss++; 432 RTFREE(ip6_forward_rt.ro_rt); 433 ip6_forward_rt.ro_rt = 0; 434 } 435 436 bzero(&ip6_forward_rt.ro_dst, sizeof(struct sockaddr_in6)); 437 dst6 = &ip6_forward_rt.ro_dst; 438 dst6->sin6_len = sizeof(struct sockaddr_in6); 439 dst6->sin6_family = AF_INET6; 440 dst6->sin6_addr = ip6->ip6_dst; 441 442 rtalloc_ign((struct route *)&ip6_forward_rt, RTF_PRCLONING); 443 } 444 445 #define rt6_key(r) ((struct sockaddr_in6 *)((r)->rt_nodes->rn_key)) 446 447 /* 448 * Accept the packet if the forwarding interface to the destination 449 * according to the routing table is the loopback interface, 450 * unless the associated route has a gateway. 451 * Note that this approach causes to accept a packet if there is a 452 * route to the loopback interface for the destination of the packet. 453 * But we think it's even useful in some situations, e.g. when using 454 * a special daemon which wants to intercept the packet. 455 * 456 * XXX: some OSes automatically make a cloned route for the destination 457 * of an outgoing packet. If the outgoing interface of the packet 458 * is a loopback one, the kernel would consider the packet to be 459 * accepted, even if we have no such address assinged on the interface. 460 * We check the cloned flag of the route entry to reject such cases, 461 * assuming that route entries for our own addresses are not made by 462 * cloning (it should be true because in6_addloop explicitly installs 463 * the host route). However, we might have to do an explicit check 464 * while it would be less efficient. Or, should we rather install a 465 * reject route for such a case? 466 */ 467 if (ip6_forward_rt.ro_rt && 468 (ip6_forward_rt.ro_rt->rt_flags & 469 (RTF_HOST|RTF_GATEWAY)) == RTF_HOST && 470 #ifdef RTF_WASCLONED 471 !(ip6_forward_rt.ro_rt->rt_flags & RTF_WASCLONED) && 472 #endif 473 #ifdef RTF_CLONED 474 !(ip6_forward_rt.ro_rt->rt_flags & RTF_CLONED) && 475 #endif 476 #if 0 477 /* 478 * The check below is redundant since the comparison of 479 * the destination and the key of the rtentry has 480 * already done through looking up the routing table. 481 */ 482 IN6_ARE_ADDR_EQUAL(&ip6->ip6_dst, 483 &rt6_key(ip6_forward_rt.ro_rt)->sin6_addr) 484 #endif 485 ip6_forward_rt.ro_rt->rt_ifp->if_type == IFT_LOOP) { 486 struct in6_ifaddr *ia6 = 487 (struct in6_ifaddr *)ip6_forward_rt.ro_rt->rt_ifa; 488 489 /* 490 * record address information into m_aux. 491 */ 492 ip6_setdstifaddr(m, ia6); 493 494 /* 495 * packets to a tentative, duplicated, or somehow invalid 496 * address must not be accepted. 497 */ 498 if (!(ia6->ia6_flags & IN6_IFF_NOTREADY)) { 499 /* this address is ready */ 500 ours = 1; 501 deliverifp = ia6->ia_ifp; /* correct? */ 502 /* Count the packet in the ip address stats */ 503 IFA_STAT_INC(&ia6->ia_ifa, ipackets, 1); 504 IFA_STAT_INC(&ia6->ia_ifa, ibytes, m->m_pkthdr.len); 505 goto hbhcheck; 506 } else { 507 /* address is not ready, so discard the packet. */ 508 nd6log((LOG_INFO, 509 "ip6_input: packet to an unready address %s->%s\n", 510 ip6_sprintf(&ip6->ip6_src), 511 ip6_sprintf(&ip6->ip6_dst))); 512 513 goto bad; 514 } 515 } 516 517 /* 518 * Now there is no reason to process the packet if it's not our own 519 * and we're not a router. 520 */ 521 if (!ip6_forwarding) { 522 ip6stat.ip6s_cantforward++; 523 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_discard); 524 goto bad; 525 } 526 527 hbhcheck: 528 /* 529 * record address information into m_aux, if we don't have one yet. 530 * note that we are unable to record it, if the address is not listed 531 * as our interface address (e.g. multicast addresses and such). 532 */ 533 if (deliverifp && !ip6_getdstifaddr(m)) { 534 struct in6_ifaddr *ia6; 535 536 ia6 = in6_ifawithifp(deliverifp, &ip6->ip6_dst); 537 if (ia6) { 538 if (!ip6_setdstifaddr(m, ia6)) { 539 /* 540 * XXX maybe we should drop the packet here, 541 * as we could not provide enough information 542 * to the upper layers. 543 */ 544 } 545 } 546 } 547 548 /* 549 * Process Hop-by-Hop options header if it's contained. 550 * m may be modified in ip6_hopopts_input(), catch this via mp 551 * If a JumboPayload option is included, plen will also be modified. 552 */ 553 plen = (u_int32_t)ntohs(ip6->ip6_plen); 554 if (ip6->ip6_nxt == IPPROTO_HOPOPTS) { 555 struct ip6_hbh *hbh; 556 struct mbuf **mp = &m; 557 558 if (ip6_hopopts_input(&plen, &rtalert, mp, &off)) { 559 #if 0 /*touches NULL pointer*/ 560 in6_ifstat_inc((*mp)->m_pkthdr.rcvif, ifs6_in_discard); 561 #endif 562 goto bad2; /* m have already been freed */ 563 } 564 565 /* adjust pointer */ 566 m = *mp; 567 ip6 = mtod(m, struct ip6_hdr *); 568 569 /* 570 * if the payload length field is 0 and the next header field 571 * indicates Hop-by-Hop Options header, then a Jumbo Payload 572 * option MUST be included. 573 */ 574 if (ip6->ip6_plen == 0 && plen == 0) { 575 /* 576 * Note that if a valid jumbo payload option is 577 * contained, ip6_hoptops_input() must set a valid 578 * (non-zero) payload length to the variable plen. 579 */ 580 ip6stat.ip6s_badoptions++; 581 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_discard); 582 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_hdrerr); 583 icmp6_error(m, ICMP6_PARAM_PROB, 584 ICMP6_PARAMPROB_HEADER, 585 (caddr_t)&ip6->ip6_plen - (caddr_t)ip6); 586 goto bad2; 587 } 588 #ifndef PULLDOWN_TEST 589 /* ip6_hopopts_input() ensures that mbuf is contiguous */ 590 hbh = (struct ip6_hbh *)(ip6 + 1); 591 #else 592 IP6_EXTHDR_GET(hbh, struct ip6_hbh *, m, sizeof(struct ip6_hdr), 593 sizeof(struct ip6_hbh)); 594 if (hbh == NULL) { 595 ip6stat.ip6s_tooshort++; 596 goto bad2; 597 } 598 #endif 599 nxt = hbh->ip6h_nxt; 600 601 /* 602 * If we are acting as a router and the packet contains a 603 * router alert option, see if we know the option value. 604 * Currently, we only support the option value for MLD, in which 605 * case we should pass the packet to the multicast routing 606 * daemon. 607 */ 608 if (rtalert != ~0 && ip6_forwarding) { 609 switch (rtalert) { 610 case IP6OPT_RTALERT_MLD: 611 ours = 1; 612 break; 613 default: 614 /* 615 * RFC2711 requires unrecognized values must be 616 * silently ignored. 617 */ 618 break; 619 } 620 } 621 } else 622 nxt = ip6->ip6_nxt; 623 624 /* 625 * Check that the amount of data in the buffers 626 * is as at least much as the IPv6 header would have us expect. 627 * Trim mbufs if longer than we expect. 628 * Drop packet if shorter than we expect. 629 */ 630 if (m->m_pkthdr.len - sizeof(struct ip6_hdr) < plen) { 631 ip6stat.ip6s_tooshort++; 632 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_truncated); 633 goto bad; 634 } 635 if (m->m_pkthdr.len > sizeof(struct ip6_hdr) + plen) { 636 if (m->m_len == m->m_pkthdr.len) { 637 m->m_len = sizeof(struct ip6_hdr) + plen; 638 m->m_pkthdr.len = sizeof(struct ip6_hdr) + plen; 639 } else 640 m_adj(m, sizeof(struct ip6_hdr) + plen - m->m_pkthdr.len); 641 } 642 643 /* 644 * Forward if desirable. 645 */ 646 if (IN6_IS_ADDR_MULTICAST(&ip6->ip6_dst)) { 647 /* 648 * If we are acting as a multicast router, all 649 * incoming multicast packets are passed to the 650 * kernel-level multicast forwarding function. 651 * The packet is returned (relatively) intact; if 652 * ip6_mforward() returns a non-zero value, the packet 653 * must be discarded, else it may be accepted below. 654 */ 655 if (ip6_mrouter && ip6_mforward(ip6, m->m_pkthdr.rcvif, m)) { 656 ip6stat.ip6s_cantforward++; 657 goto bad; 658 } 659 if (!ours) 660 goto bad; 661 } else if (!ours) { 662 ip6_forward(m, srcrt); 663 goto bad2; 664 } 665 666 ip6 = mtod(m, struct ip6_hdr *); 667 668 /* 669 * Malicious party may be able to use IPv4 mapped addr to confuse 670 * tcp/udp stack and bypass security checks (act as if it was from 671 * 127.0.0.1 by using IPv6 src ::ffff:127.0.0.1). Be cautious. 672 * 673 * For SIIT end node behavior, you may want to disable the check. 674 * However, you will become vulnerable to attacks using IPv4 mapped 675 * source. 676 */ 677 if (IN6_IS_ADDR_V4MAPPED(&ip6->ip6_src) || 678 IN6_IS_ADDR_V4MAPPED(&ip6->ip6_dst)) { 679 ip6stat.ip6s_badscope++; 680 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_addrerr); 681 goto bad; 682 } 683 684 /* 685 * Tell launch routine the next header 686 */ 687 ip6stat.ip6s_delivered++; 688 in6_ifstat_inc(deliverifp, ifs6_in_deliver); 689 nest = 0; 690 691 rh_present = 0; 692 while (nxt != IPPROTO_DONE) { 693 struct protosw *sw6; 694 695 if (ip6_hdrnestlimit && (++nest > ip6_hdrnestlimit)) { 696 ip6stat.ip6s_toomanyhdr++; 697 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_hdrerr); 698 goto bad; 699 } 700 701 /* 702 * protection against faulty packet - there should be 703 * more sanity checks in header chain processing. 704 */ 705 if (m->m_pkthdr.len < off) { 706 ip6stat.ip6s_tooshort++; 707 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_truncated); 708 goto bad; 709 } 710 711 #if 0 712 /* 713 * do we need to do it for every header? yeah, other 714 * functions can play with it (like re-allocate and copy). 715 */ 716 mhist = ip6_addaux(m); 717 if (mhist && M_TRAILINGSPACE(mhist) >= sizeof(nxt)) { 718 hist = mtod(mhist, caddr_t) + mhist->m_len; 719 bcopy(&nxt, hist, sizeof(nxt)); 720 mhist->m_len += sizeof(nxt); 721 } else { 722 ip6stat.ip6s_toomanyhdr++; 723 goto bad; 724 } 725 #endif 726 727 if (nxt == IPPROTO_ROUTING) { 728 if (rh_present++) { 729 in6_ifstat_inc(m->m_pkthdr.rcvif, 730 ifs6_in_hdrerr); 731 ip6stat.ip6s_badoptions++; 732 goto bad; 733 } 734 } 735 736 sw6 = &inet6sw[ip6_protox[nxt]]; 737 /* 738 * If this is a terminal header forward to the port, otherwise 739 * process synchronously for more headers. 740 */ 741 if (sw6->pr_flags & PR_LASTHDR) { 742 struct netmsg_packet *pmsg; 743 lwkt_port_t port; 744 745 port = netisr_cpuport(0); /* XXX */ 746 if ((m->m_flags & M_HASH) == 0) 747 m_sethash(m, 0); /* XXX */ 748 749 KKASSERT(port != NULL); 750 pmsg = &m->m_hdr.mh_netmsg; 751 netmsg_init(&pmsg->base, NULL, 752 &netisr_apanic_rport, 753 0, transport6_processing_handler); 754 pmsg->nm_packet = m; 755 pmsg->nm_nxt = nxt; 756 pmsg->base.lmsg.u.ms_result = off; 757 lwkt_sendmsg(port, &pmsg->base.lmsg); 758 /* done with m */ 759 nxt = IPPROTO_DONE; 760 } else { 761 nxt = sw6->pr_input(&m, &off, nxt); 762 } 763 } 764 goto bad2; 765 bad: 766 m_freem(m); 767 bad2: 768 ; 769 /* msg was embedded in the mbuf, do not reply! */ 770 } 771 772 /* 773 * We have to call the pr_input() function from the correct protocol 774 * thread. The sw6->pr_soport() request at the end of ip6_input() 775 * returns the port and we forward a netmsg to the port to execute 776 * this function. 777 */ 778 static void 779 transport6_processing_handler(netmsg_t netmsg) 780 { 781 struct netmsg_packet *pmsg = (struct netmsg_packet *)netmsg; 782 struct protosw *sw6; 783 int hlen; 784 int nxt; 785 786 sw6 = &inet6sw[ip6_protox[pmsg->nm_nxt]]; 787 hlen = pmsg->base.lmsg.u.ms_result; 788 789 nxt = sw6->pr_input(&pmsg->nm_packet, &hlen, pmsg->nm_nxt); 790 KKASSERT(nxt == IPPROTO_DONE); 791 /* netmsg was embedded in the mbuf, do not reply! */ 792 } 793 794 /* 795 * set/grab in6_ifaddr correspond to IPv6 destination address. 796 * XXX backward compatibility wrapper 797 */ 798 static struct ip6aux * 799 ip6_setdstifaddr(struct mbuf *m, struct in6_ifaddr *ia6) 800 { 801 struct ip6aux *n; 802 803 n = ip6_addaux(m); 804 if (n) 805 n->ip6a_dstia6 = ia6; 806 return n; /* NULL if failed to set */ 807 } 808 809 struct in6_ifaddr * 810 ip6_getdstifaddr(struct mbuf *m) 811 { 812 struct ip6aux *n; 813 814 n = ip6_findaux(m); 815 if (n) 816 return n->ip6a_dstia6; 817 else 818 return NULL; 819 } 820 821 /* 822 * Hop-by-Hop options header processing. If a valid jumbo payload option is 823 * included, the real payload length will be stored in plenp. 824 */ 825 static int 826 ip6_hopopts_input(u_int32_t *plenp, 827 u_int32_t *rtalertp,/* XXX: should be stored more smart way */ 828 struct mbuf **mp, 829 int *offp) 830 { 831 struct mbuf *m = *mp; 832 int off = *offp, hbhlen; 833 struct ip6_hbh *hbh; 834 835 /* validation of the length of the header */ 836 #ifndef PULLDOWN_TEST 837 IP6_EXTHDR_CHECK(m, off, sizeof(*hbh), -1); 838 hbh = (struct ip6_hbh *)(mtod(m, caddr_t) + off); 839 hbhlen = (hbh->ip6h_len + 1) << 3; 840 841 IP6_EXTHDR_CHECK(m, off, hbhlen, -1); 842 hbh = (struct ip6_hbh *)(mtod(m, caddr_t) + off); 843 #else 844 IP6_EXTHDR_GET(hbh, struct ip6_hbh *, m, 845 sizeof(struct ip6_hdr), sizeof(struct ip6_hbh)); 846 if (hbh == NULL) { 847 ip6stat.ip6s_tooshort++; 848 return -1; 849 } 850 hbhlen = (hbh->ip6h_len + 1) << 3; 851 IP6_EXTHDR_GET(hbh, struct ip6_hbh *, m, sizeof(struct ip6_hdr), 852 hbhlen); 853 if (hbh == NULL) { 854 ip6stat.ip6s_tooshort++; 855 return -1; 856 } 857 #endif 858 off += hbhlen; 859 hbhlen -= sizeof(struct ip6_hbh); 860 861 if (ip6_process_hopopts(m, (u_int8_t *)hbh + sizeof(struct ip6_hbh), 862 hbhlen, rtalertp, plenp) < 0) 863 return (-1); 864 865 *offp = off; 866 *mp = m; 867 return (0); 868 } 869 870 /* 871 * Search header for all Hop-by-hop options and process each option. 872 * This function is separate from ip6_hopopts_input() in order to 873 * handle a case where the sending node itself process its hop-by-hop 874 * options header. In such a case, the function is called from ip6_output(). 875 * 876 * The function assumes that hbh header is located right after the IPv6 header 877 * (RFC2460 p7), opthead is pointer into data content in m, and opthead to 878 * opthead + hbhlen is located in continuous memory region. 879 */ 880 int 881 ip6_process_hopopts(struct mbuf *m, u_int8_t *opthead, int hbhlen, 882 u_int32_t *rtalertp, u_int32_t *plenp) 883 { 884 struct ip6_hdr *ip6; 885 int optlen = 0; 886 u_int8_t *opt = opthead; 887 u_int16_t rtalert_val; 888 u_int32_t jumboplen; 889 const int erroff = sizeof(struct ip6_hdr) + sizeof(struct ip6_hbh); 890 891 for (; hbhlen > 0; hbhlen -= optlen, opt += optlen) { 892 switch (*opt) { 893 case IP6OPT_PAD1: 894 optlen = 1; 895 break; 896 case IP6OPT_PADN: 897 if (hbhlen < IP6OPT_MINLEN) { 898 ip6stat.ip6s_toosmall++; 899 goto bad; 900 } 901 optlen = *(opt + 1) + 2; 902 break; 903 case IP6OPT_RTALERT: 904 /* XXX may need check for alignment */ 905 if (hbhlen < IP6OPT_RTALERT_LEN) { 906 ip6stat.ip6s_toosmall++; 907 goto bad; 908 } 909 if (*(opt + 1) != IP6OPT_RTALERT_LEN - 2) { 910 /* XXX stat */ 911 icmp6_error(m, ICMP6_PARAM_PROB, 912 ICMP6_PARAMPROB_HEADER, 913 erroff + opt + 1 - opthead); 914 return (-1); 915 } 916 optlen = IP6OPT_RTALERT_LEN; 917 bcopy((caddr_t)(opt + 2), (caddr_t)&rtalert_val, 2); 918 *rtalertp = ntohs(rtalert_val); 919 break; 920 case IP6OPT_JUMBO: 921 /* XXX may need check for alignment */ 922 if (hbhlen < IP6OPT_JUMBO_LEN) { 923 ip6stat.ip6s_toosmall++; 924 goto bad; 925 } 926 if (*(opt + 1) != IP6OPT_JUMBO_LEN - 2) { 927 /* XXX stat */ 928 icmp6_error(m, ICMP6_PARAM_PROB, 929 ICMP6_PARAMPROB_HEADER, 930 erroff + opt + 1 - opthead); 931 return (-1); 932 } 933 optlen = IP6OPT_JUMBO_LEN; 934 935 /* 936 * IPv6 packets that have non 0 payload length 937 * must not contain a jumbo payload option. 938 */ 939 ip6 = mtod(m, struct ip6_hdr *); 940 if (ip6->ip6_plen) { 941 ip6stat.ip6s_badoptions++; 942 icmp6_error(m, ICMP6_PARAM_PROB, 943 ICMP6_PARAMPROB_HEADER, 944 erroff + opt - opthead); 945 return (-1); 946 } 947 948 /* 949 * We may see jumbolen in unaligned location, so 950 * we'd need to perform bcopy(). 951 */ 952 bcopy(opt + 2, &jumboplen, sizeof(jumboplen)); 953 jumboplen = (u_int32_t)htonl(jumboplen); 954 955 #if 1 956 /* 957 * if there are multiple jumbo payload options, 958 * *plenp will be non-zero and the packet will be 959 * rejected. 960 * the behavior may need some debate in ipngwg - 961 * multiple options does not make sense, however, 962 * there's no explicit mention in specification. 963 */ 964 if (*plenp != 0) { 965 ip6stat.ip6s_badoptions++; 966 icmp6_error(m, ICMP6_PARAM_PROB, 967 ICMP6_PARAMPROB_HEADER, 968 erroff + opt + 2 - opthead); 969 return (-1); 970 } 971 #endif 972 973 /* 974 * jumbo payload length must be larger than 65535. 975 */ 976 if (jumboplen <= IPV6_MAXPACKET) { 977 ip6stat.ip6s_badoptions++; 978 icmp6_error(m, ICMP6_PARAM_PROB, 979 ICMP6_PARAMPROB_HEADER, 980 erroff + opt + 2 - opthead); 981 return (-1); 982 } 983 *plenp = jumboplen; 984 985 break; 986 default: /* unknown option */ 987 if (hbhlen < IP6OPT_MINLEN) { 988 ip6stat.ip6s_toosmall++; 989 goto bad; 990 } 991 optlen = ip6_unknown_opt(opt, m, 992 erroff + opt - opthead); 993 if (optlen == -1) 994 return (-1); 995 optlen += 2; 996 break; 997 } 998 } 999 1000 return (0); 1001 1002 bad: 1003 m_freem(m); 1004 return (-1); 1005 } 1006 1007 /* 1008 * Unknown option processing. 1009 * The third argument `off' is the offset from the IPv6 header to the option, 1010 * which is necessary if the IPv6 header the and option header and IPv6 header 1011 * is not continuous in order to return an ICMPv6 error. 1012 */ 1013 int 1014 ip6_unknown_opt(u_int8_t *optp, struct mbuf *m, int off) 1015 { 1016 struct ip6_hdr *ip6; 1017 1018 switch (IP6OPT_TYPE(*optp)) { 1019 case IP6OPT_TYPE_SKIP: /* ignore the option */ 1020 return ((int)*(optp + 1)); 1021 case IP6OPT_TYPE_DISCARD: /* silently discard */ 1022 m_freem(m); 1023 return (-1); 1024 case IP6OPT_TYPE_FORCEICMP: /* send ICMP even if multicasted */ 1025 ip6stat.ip6s_badoptions++; 1026 icmp6_error(m, ICMP6_PARAM_PROB, ICMP6_PARAMPROB_OPTION, off); 1027 return (-1); 1028 case IP6OPT_TYPE_ICMP: /* send ICMP if not multicasted */ 1029 ip6stat.ip6s_badoptions++; 1030 ip6 = mtod(m, struct ip6_hdr *); 1031 if (IN6_IS_ADDR_MULTICAST(&ip6->ip6_dst) || 1032 (m->m_flags & (M_BCAST|M_MCAST))) 1033 m_freem(m); 1034 else 1035 icmp6_error(m, ICMP6_PARAM_PROB, 1036 ICMP6_PARAMPROB_OPTION, off); 1037 return (-1); 1038 } 1039 1040 m_freem(m); /* XXX: NOTREACHED */ 1041 return (-1); 1042 } 1043 1044 /* 1045 * Create the "control" list for this pcb. 1046 * The function will not modify mbuf chain at all. 1047 * 1048 * with KAME mbuf chain restriction: 1049 * The routine will be called from upper layer handlers like tcp6_input(). 1050 * Thus the routine assumes that the caller (tcp6_input) have already 1051 * called IP6_EXTHDR_CHECK() and all the extension headers are located in the 1052 * very first mbuf on the mbuf chain. 1053 */ 1054 void 1055 ip6_savecontrol(struct inpcb *in6p, struct mbuf **mp, struct ip6_hdr *ip6, 1056 struct mbuf *m) 1057 { 1058 #define IS2292(x, y) ((in6p->in6p_flags & IN6P_RFC2292) ? (x) : (y)) 1059 struct thread *td = curthread; /* XXX */ 1060 int privileged = 0; 1061 1062 1063 if (priv_check(td, PRIV_ROOT) == 0) 1064 privileged++; 1065 1066 #ifdef SO_TIMESTAMP 1067 if (in6p->in6p_socket->so_options & SO_TIMESTAMP) { 1068 struct timeval tv; 1069 1070 microtime(&tv); 1071 *mp = sbcreatecontrol((caddr_t)&tv, sizeof(tv), 1072 SCM_TIMESTAMP, SOL_SOCKET); 1073 if (*mp) { 1074 mp = &(*mp)->m_next; 1075 } 1076 } 1077 #endif 1078 1079 /* RFC 2292 sec. 5 */ 1080 if (in6p->in6p_flags & IN6P_PKTINFO) { 1081 struct in6_pktinfo pi6; 1082 1083 bcopy(&ip6->ip6_dst, &pi6.ipi6_addr, sizeof(struct in6_addr)); 1084 in6_clearscope(&pi6.ipi6_addr); /* XXX */ 1085 pi6.ipi6_ifindex = (m && m->m_pkthdr.rcvif) ? 1086 m->m_pkthdr.rcvif->if_index : 0; 1087 *mp = sbcreatecontrol((caddr_t)&pi6, sizeof(struct in6_pktinfo), 1088 IS2292(IPV6_2292PKTINFO, IPV6_PKTINFO), IPPROTO_IPV6); 1089 if (*mp) 1090 mp = &(*mp)->m_next; 1091 } 1092 1093 if (in6p->in6p_flags & IN6P_HOPLIMIT) { 1094 int hlim = ip6->ip6_hlim & 0xff; 1095 1096 *mp = sbcreatecontrol((caddr_t)&hlim, sizeof(int), 1097 IS2292(IPV6_2292HOPLIMIT, IPV6_HOPLIMIT), IPPROTO_IPV6); 1098 if (*mp) 1099 mp = &(*mp)->m_next; 1100 } 1101 1102 if ((in6p->in6p_flags & IN6P_TCLASS) != 0) { 1103 u_int32_t flowinfo; 1104 int tclass; 1105 1106 flowinfo = (u_int32_t)ntohl(ip6->ip6_flow & IPV6_FLOWINFO_MASK); 1107 flowinfo >>= 20; 1108 1109 tclass = flowinfo & 0xff; 1110 *mp = sbcreatecontrol((caddr_t) &tclass, sizeof(tclass), 1111 IPV6_TCLASS, IPPROTO_IPV6); 1112 if (*mp) 1113 mp = &(*mp)->m_next; 1114 } 1115 1116 /* 1117 * IPV6_HOPOPTS socket option. We require super-user privilege 1118 * for the option, but it might be too strict, since there might 1119 * be some hop-by-hop options which can be returned to normal user. 1120 * See RFC 2292 section 6. 1121 */ 1122 if ((in6p->in6p_flags & IN6P_HOPOPTS) && privileged) { 1123 /* 1124 * Check if a hop-by-hop options header is contatined in the 1125 * received packet, and if so, store the options as ancillary 1126 * data. Note that a hop-by-hop options header must be 1127 * just after the IPv6 header, which is assured through the 1128 * IPv6 input processing. 1129 */ 1130 struct ip6_hdr *ip6 = mtod(m, struct ip6_hdr *); 1131 if (ip6->ip6_nxt == IPPROTO_HOPOPTS) { 1132 struct ip6_hbh *hbh; 1133 int hbhlen = 0; 1134 #ifdef PULLDOWN_TEST 1135 struct mbuf *ext; 1136 #endif 1137 1138 #ifndef PULLDOWN_TEST 1139 hbh = (struct ip6_hbh *)(ip6 + 1); 1140 hbhlen = (hbh->ip6h_len + 1) << 3; 1141 #else 1142 ext = ip6_pullexthdr(m, sizeof(struct ip6_hdr), 1143 ip6->ip6_nxt); 1144 if (ext == NULL) { 1145 ip6stat.ip6s_tooshort++; 1146 return; 1147 } 1148 hbh = mtod(ext, struct ip6_hbh *); 1149 hbhlen = (hbh->ip6h_len + 1) << 3; 1150 if (hbhlen != ext->m_len) { 1151 m_freem(ext); 1152 ip6stat.ip6s_tooshort++; 1153 return; 1154 } 1155 #endif 1156 1157 /* 1158 * XXX: We copy the whole header even if a 1159 * jumbo payload option is included, the option which 1160 * is to be removed before returning according to 1161 * RFC 2292. 1162 * Note: this constraint is removed in 2292bis. 1163 */ 1164 *mp = sbcreatecontrol((caddr_t)hbh, hbhlen, 1165 IS2292(IPV6_2292HOPOPTS, IPV6_HOPOPTS), 1166 IPPROTO_IPV6); 1167 if (*mp) 1168 mp = &(*mp)->m_next; 1169 #ifdef PULLDOWN_TEST 1170 m_freem(ext); 1171 #endif 1172 } 1173 } 1174 1175 /* IPV6_DSTOPTS and IPV6_RTHDR socket options */ 1176 if ((in6p->in6p_flags & (IN6P_DSTOPTS | IN6P_RTHDRDSTOPTS)) != 0) { 1177 int proto, off, nxt; 1178 1179 /* 1180 * go through the header chain to see if a routing header is 1181 * contained in the packet. We need this information to store 1182 * destination options headers (if any) properly. 1183 * XXX: performance issue. We should record this info when 1184 * processing extension headers in incoming routine. 1185 * (todo) use m_aux? 1186 */ 1187 proto = IPPROTO_IPV6; 1188 off = 0; 1189 nxt = -1; 1190 while (1) { 1191 int newoff; 1192 1193 newoff = ip6_nexthdr(m, off, proto, &nxt); 1194 if (newoff < 0) 1195 break; 1196 if (newoff < off) /* invalid, check for safety */ 1197 break; 1198 if ((proto = nxt) == IPPROTO_ROUTING) 1199 break; 1200 off = newoff; 1201 } 1202 } 1203 1204 if ((in6p->in6p_flags & 1205 (IN6P_RTHDR | IN6P_DSTOPTS | IN6P_RTHDRDSTOPTS)) != 0) { 1206 struct ip6_hdr *ip6 = mtod(m, struct ip6_hdr *); 1207 int nxt = ip6->ip6_nxt, off = sizeof(struct ip6_hdr); 1208 1209 /* 1210 * Search for destination options headers or routing 1211 * header(s) through the header chain, and stores each 1212 * header as ancillary data. 1213 * Note that the order of the headers remains in 1214 * the chain of ancillary data. 1215 */ 1216 while (1) { /* is explicit loop prevention necessary? */ 1217 struct ip6_ext *ip6e = NULL; 1218 int elen; 1219 #ifdef PULLDOWN_TEST 1220 struct mbuf *ext = NULL; 1221 #endif 1222 1223 /* 1224 * if it is not an extension header, don't try to 1225 * pull it from the chain. 1226 */ 1227 switch (nxt) { 1228 case IPPROTO_DSTOPTS: 1229 case IPPROTO_ROUTING: 1230 case IPPROTO_HOPOPTS: 1231 case IPPROTO_AH: /* is it possible? */ 1232 break; 1233 default: 1234 goto loopend; 1235 } 1236 1237 #ifndef PULLDOWN_TEST 1238 if (off + sizeof(*ip6e) > m->m_len) 1239 goto loopend; 1240 ip6e = (struct ip6_ext *)(mtod(m, caddr_t) + off); 1241 if (nxt == IPPROTO_AH) 1242 elen = (ip6e->ip6e_len + 2) << 2; 1243 else 1244 elen = (ip6e->ip6e_len + 1) << 3; 1245 if (off + elen > m->m_len) 1246 goto loopend; 1247 #else 1248 ext = ip6_pullexthdr(m, off, nxt); 1249 if (ext == NULL) { 1250 ip6stat.ip6s_tooshort++; 1251 return; 1252 } 1253 ip6e = mtod(ext, struct ip6_ext *); 1254 if (nxt == IPPROTO_AH) 1255 elen = (ip6e->ip6e_len + 2) << 2; 1256 else 1257 elen = (ip6e->ip6e_len + 1) << 3; 1258 if (elen != ext->m_len) { 1259 m_freem(ext); 1260 ip6stat.ip6s_tooshort++; 1261 return; 1262 } 1263 #endif 1264 1265 switch (nxt) { 1266 case IPPROTO_DSTOPTS: 1267 if ((in6p->in6p_flags & IN6P_DSTOPTS) == 0) 1268 break; 1269 1270 /* 1271 * We also require super-user privilege for 1272 * the option. See comments on IN6_HOPOPTS. 1273 */ 1274 if (!privileged) 1275 break; 1276 1277 *mp = sbcreatecontrol((caddr_t)ip6e, elen, 1278 IS2292(IPV6_2292DSTOPTS, IPV6_DSTOPTS), 1279 IPPROTO_IPV6); 1280 if (*mp) 1281 mp = &(*mp)->m_next; 1282 break; 1283 case IPPROTO_ROUTING: 1284 if (!(in6p->in6p_flags & IN6P_RTHDR)) 1285 break; 1286 1287 *mp = sbcreatecontrol((caddr_t)ip6e, elen, 1288 IS2292(IPV6_2292RTHDR, IPV6_RTHDR), 1289 IPPROTO_IPV6); 1290 if (*mp) 1291 mp = &(*mp)->m_next; 1292 break; 1293 case IPPROTO_HOPOPTS: 1294 case IPPROTO_AH: /* is it possible? */ 1295 break; 1296 1297 default: 1298 /* 1299 * other cases have been filtered in the above. 1300 * none will visit this case. here we supply 1301 * the code just in case (nxt overwritten or 1302 * other cases). 1303 */ 1304 #ifdef PULLDOWN_TEST 1305 m_freem(ext); 1306 #endif 1307 goto loopend; 1308 1309 } 1310 1311 /* proceed with the next header. */ 1312 off += elen; 1313 nxt = ip6e->ip6e_nxt; 1314 ip6e = NULL; 1315 #ifdef PULLDOWN_TEST 1316 m_freem(ext); 1317 ext = NULL; 1318 #endif 1319 } 1320 loopend: 1321 ; 1322 } 1323 #undef IS2292 1324 } 1325 1326 void 1327 ip6_notify_pmtu(struct inpcb *in6p, struct sockaddr_in6 *dst, u_int32_t *mtu) 1328 { 1329 struct socket *so; 1330 struct mbuf *m_mtu; 1331 struct ip6_mtuinfo mtuctl; 1332 1333 so = in6p->inp_socket; 1334 1335 if (mtu == NULL) 1336 return; 1337 1338 #ifdef DIAGNOSTIC 1339 if (so == NULL) /* I believe this is impossible */ 1340 panic("ip6_notify_pmtu: socket is NULL"); 1341 #endif 1342 1343 bzero(&mtuctl, sizeof(mtuctl)); /* zero-clear for safety */ 1344 mtuctl.ip6m_mtu = *mtu; 1345 mtuctl.ip6m_addr = *dst; 1346 1347 if ((m_mtu = sbcreatecontrol((caddr_t)&mtuctl, sizeof(mtuctl), 1348 IPV6_PATHMTU, IPPROTO_IPV6)) == NULL) 1349 return; 1350 1351 lwkt_gettoken(&so->so_rcv.ssb_token); 1352 if (sbappendaddr(&so->so_rcv.sb, (struct sockaddr *)dst, NULL, m_mtu) 1353 == 0) { 1354 m_freem(m_mtu); 1355 /* XXX: should count statistics */ 1356 } else { 1357 sorwakeup(so); 1358 } 1359 lwkt_reltoken(&so->so_rcv.ssb_token); 1360 } 1361 1362 #ifdef PULLDOWN_TEST 1363 /* 1364 * pull single extension header from mbuf chain. returns single mbuf that 1365 * contains the result, or NULL on error. 1366 */ 1367 static struct mbuf * 1368 ip6_pullexthdr(struct mbuf *m, size_t off, int nxt) 1369 { 1370 struct ip6_ext ip6e; 1371 size_t elen; 1372 struct mbuf *n; 1373 1374 #ifdef DIAGNOSTIC 1375 switch (nxt) { 1376 case IPPROTO_DSTOPTS: 1377 case IPPROTO_ROUTING: 1378 case IPPROTO_HOPOPTS: 1379 case IPPROTO_AH: /* is it possible? */ 1380 break; 1381 default: 1382 kprintf("ip6_pullexthdr: invalid nxt=%d\n", nxt); 1383 } 1384 #endif 1385 1386 m_copydata(m, off, sizeof(ip6e), (caddr_t)&ip6e); 1387 if (nxt == IPPROTO_AH) 1388 elen = (ip6e.ip6e_len + 2) << 2; 1389 else 1390 elen = (ip6e.ip6e_len + 1) << 3; 1391 1392 n = m_getb(elen, M_NOWAIT, MT_DATA, 0); 1393 if (n == NULL) 1394 return NULL; 1395 n->m_len = 0; 1396 1397 if (elen >= M_TRAILINGSPACE(n)) { 1398 m_free(n); 1399 return NULL; 1400 } 1401 1402 m_copydata(m, off, elen, mtod(n, caddr_t)); 1403 n->m_len = elen; 1404 return n; 1405 } 1406 #endif 1407 1408 /* 1409 * Get pointer to the previous header followed by the header 1410 * currently processed. 1411 * XXX: This function supposes that 1412 * M includes all headers, 1413 * the next header field and the header length field of each header 1414 * are valid, and 1415 * the sum of each header length equals to OFF. 1416 * Because of these assumptions, this function must be called very 1417 * carefully. Moreover, it will not be used in the near future when 1418 * we develop `neater' mechanism to process extension headers. 1419 */ 1420 char * 1421 ip6_get_prevhdr(struct mbuf *m, int off) 1422 { 1423 struct ip6_hdr *ip6 = mtod(m, struct ip6_hdr *); 1424 1425 if (off == sizeof(struct ip6_hdr)) 1426 return (&ip6->ip6_nxt); 1427 else { 1428 int len, nxt; 1429 struct ip6_ext *ip6e = NULL; 1430 1431 nxt = ip6->ip6_nxt; 1432 len = sizeof(struct ip6_hdr); 1433 while (len < off) { 1434 ip6e = (struct ip6_ext *)(mtod(m, caddr_t) + len); 1435 1436 switch (nxt) { 1437 case IPPROTO_FRAGMENT: 1438 len += sizeof(struct ip6_frag); 1439 break; 1440 case IPPROTO_AH: 1441 len += (ip6e->ip6e_len + 2) << 2; 1442 break; 1443 default: 1444 len += (ip6e->ip6e_len + 1) << 3; 1445 break; 1446 } 1447 nxt = ip6e->ip6e_nxt; 1448 } 1449 if (ip6e) 1450 return (&ip6e->ip6e_nxt); 1451 else 1452 return NULL; 1453 } 1454 } 1455 1456 /* 1457 * get next header offset. m will be retained. 1458 */ 1459 int 1460 ip6_nexthdr(struct mbuf *m, int off, int proto, int *nxtp) 1461 { 1462 struct ip6_hdr ip6; 1463 struct ip6_ext ip6e; 1464 struct ip6_frag fh; 1465 1466 /* just in case */ 1467 if (m == NULL) 1468 panic("ip6_nexthdr: m == NULL"); 1469 if ((m->m_flags & M_PKTHDR) == 0 || m->m_pkthdr.len < off) 1470 return -1; 1471 1472 switch (proto) { 1473 case IPPROTO_IPV6: 1474 if (m->m_pkthdr.len < off + sizeof(ip6)) 1475 return -1; 1476 m_copydata(m, off, sizeof(ip6), (caddr_t)&ip6); 1477 if (nxtp) 1478 *nxtp = ip6.ip6_nxt; 1479 off += sizeof(ip6); 1480 return off; 1481 1482 case IPPROTO_FRAGMENT: 1483 /* 1484 * terminate parsing if it is not the first fragment, 1485 * it does not make sense to parse through it. 1486 */ 1487 if (m->m_pkthdr.len < off + sizeof(fh)) 1488 return -1; 1489 m_copydata(m, off, sizeof(fh), (caddr_t)&fh); 1490 /* IP6F_OFF_MASK = 0xfff8(BigEndian), 0xf8ff(LittleEndian) */ 1491 if (fh.ip6f_offlg & IP6F_OFF_MASK) 1492 return -1; 1493 if (nxtp) 1494 *nxtp = fh.ip6f_nxt; 1495 off += sizeof(struct ip6_frag); 1496 return off; 1497 1498 case IPPROTO_AH: 1499 if (m->m_pkthdr.len < off + sizeof(ip6e)) 1500 return -1; 1501 m_copydata(m, off, sizeof(ip6e), (caddr_t)&ip6e); 1502 if (nxtp) 1503 *nxtp = ip6e.ip6e_nxt; 1504 off += (ip6e.ip6e_len + 2) << 2; 1505 return off; 1506 1507 case IPPROTO_HOPOPTS: 1508 case IPPROTO_ROUTING: 1509 case IPPROTO_DSTOPTS: 1510 if (m->m_pkthdr.len < off + sizeof(ip6e)) 1511 return -1; 1512 m_copydata(m, off, sizeof(ip6e), (caddr_t)&ip6e); 1513 if (nxtp) 1514 *nxtp = ip6e.ip6e_nxt; 1515 off += (ip6e.ip6e_len + 1) << 3; 1516 return off; 1517 1518 case IPPROTO_NONE: 1519 case IPPROTO_ESP: 1520 case IPPROTO_IPCOMP: 1521 /* give up */ 1522 return -1; 1523 1524 default: 1525 return -1; 1526 } 1527 1528 return -1; 1529 } 1530 1531 /* 1532 * get offset for the last header in the chain. m will be kept untainted. 1533 */ 1534 int 1535 ip6_lasthdr(struct mbuf *m, int off, int proto, int *nxtp) 1536 { 1537 int newoff; 1538 int nxt; 1539 1540 if (!nxtp) { 1541 nxt = -1; 1542 nxtp = &nxt; 1543 } 1544 while (1) { 1545 newoff = ip6_nexthdr(m, off, proto, nxtp); 1546 if (newoff < 0) 1547 return off; 1548 else if (newoff < off) 1549 return -1; /* invalid */ 1550 else if (newoff == off) 1551 return newoff; 1552 1553 off = newoff; 1554 proto = *nxtp; 1555 } 1556 } 1557 1558 struct ip6aux * 1559 ip6_addaux(struct mbuf *m) 1560 { 1561 struct m_tag *mtag; 1562 1563 mtag = m_tag_find(m, PACKET_TAG_IPV6_INPUT, NULL); 1564 if (!mtag) { 1565 mtag = m_tag_get(PACKET_TAG_IPV6_INPUT, sizeof(struct ip6aux), 1566 M_NOWAIT); 1567 if (mtag) 1568 m_tag_prepend(m, mtag); 1569 } 1570 if (mtag) 1571 bzero(m_tag_data(mtag), sizeof(struct ip6aux)); 1572 return mtag ? (struct ip6aux *)m_tag_data(mtag) : NULL; 1573 } 1574 1575 struct ip6aux * 1576 ip6_findaux(struct mbuf *m) 1577 { 1578 struct m_tag *mtag; 1579 1580 mtag = m_tag_find(m, PACKET_TAG_IPV6_INPUT, NULL); 1581 return mtag ? (struct ip6aux *)m_tag_data(mtag) : NULL; 1582 } 1583 1584 void 1585 ip6_delaux(struct mbuf *m) 1586 { 1587 struct m_tag *mtag; 1588 1589 mtag = m_tag_find(m, PACKET_TAG_IPV6_INPUT, NULL); 1590 if (mtag) 1591 m_tag_delete(m, mtag); 1592 } 1593 1594 /* 1595 * System control for IP6 1596 */ 1597 1598 u_char inet6ctlerrmap[PRC_NCMDS] = { 1599 0, 0, 0, 0, 1600 0, EMSGSIZE, EHOSTDOWN, EHOSTUNREACH, 1601 EHOSTUNREACH, EHOSTUNREACH, ECONNREFUSED, ECONNREFUSED, 1602 EMSGSIZE, EHOSTUNREACH, 0, 0, 1603 0, 0, 0, 0, 1604 ENOPROTOOPT 1605 }; 1606