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: icmp6.c,v 1.211 2001/04/04 05:56:20 itojun 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_icmp.c 8.2 (Berkeley) 1/4/94 61 */ 62 63 #include <sys/cdefs.h> 64 __FBSDID("$FreeBSD$"); 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/domain.h> 72 #include <sys/jail.h> 73 #include <sys/kernel.h> 74 #include <sys/lock.h> 75 #include <sys/malloc.h> 76 #include <sys/mbuf.h> 77 #include <sys/proc.h> 78 #include <sys/protosw.h> 79 #include <sys/signalvar.h> 80 #include <sys/socket.h> 81 #include <sys/socketvar.h> 82 #include <sys/sx.h> 83 #include <sys/syslog.h> 84 #include <sys/systm.h> 85 #include <sys/time.h> 86 87 #include <net/if.h> 88 #include <net/if_dl.h> 89 #include <net/if_llatbl.h> 90 #include <net/if_types.h> 91 #include <net/route.h> 92 #include <net/vnet.h> 93 94 #include <netinet/in.h> 95 #include <netinet/in_pcb.h> 96 #include <netinet/in_var.h> 97 #include <netinet/ip6.h> 98 #include <netinet/icmp6.h> 99 #include <netinet/tcp_var.h> 100 101 #include <netinet6/in6_ifattach.h> 102 #include <netinet6/in6_pcb.h> 103 #include <netinet6/ip6protosw.h> 104 #include <netinet6/ip6_var.h> 105 #include <netinet6/scope6_var.h> 106 #include <netinet6/mld6_var.h> 107 #include <netinet6/nd6.h> 108 #include <netinet6/send.h> 109 110 #ifdef IPSEC 111 #include <netipsec/ipsec.h> 112 #include <netipsec/key.h> 113 #endif 114 115 extern struct domain inet6domain; 116 117 VNET_DEFINE(struct icmp6stat, icmp6stat); 118 119 VNET_DECLARE(struct inpcbinfo, ripcbinfo); 120 VNET_DECLARE(struct inpcbhead, ripcb); 121 VNET_DECLARE(int, icmp6errppslim); 122 static VNET_DEFINE(int, icmp6errpps_count) = 0; 123 static VNET_DEFINE(struct timeval, icmp6errppslim_last); 124 VNET_DECLARE(int, icmp6_nodeinfo); 125 126 #define V_ripcbinfo VNET(ripcbinfo) 127 #define V_ripcb VNET(ripcb) 128 #define V_icmp6errppslim VNET(icmp6errppslim) 129 #define V_icmp6errpps_count VNET(icmp6errpps_count) 130 #define V_icmp6errppslim_last VNET(icmp6errppslim_last) 131 #define V_icmp6_nodeinfo VNET(icmp6_nodeinfo) 132 133 static void icmp6_errcount(struct icmp6errstat *, int, int); 134 static int icmp6_rip6_input(struct mbuf **, int); 135 static int icmp6_ratelimit(const struct in6_addr *, const int, const int); 136 static const char *icmp6_redirect_diag __P((struct in6_addr *, 137 struct in6_addr *, struct in6_addr *)); 138 static struct mbuf *ni6_input(struct mbuf *, int); 139 static struct mbuf *ni6_nametodns(const char *, int, int); 140 static int ni6_dnsmatch(const char *, int, const char *, int); 141 static int ni6_addrs __P((struct icmp6_nodeinfo *, struct mbuf *, 142 struct ifnet **, struct in6_addr *)); 143 static int ni6_store_addrs __P((struct icmp6_nodeinfo *, struct icmp6_nodeinfo *, 144 struct ifnet *, int)); 145 static int icmp6_notify_error(struct mbuf **, int, int, int); 146 147 /* 148 * Kernel module interface for updating icmp6stat. The argument is an index 149 * into icmp6stat treated as an array of u_quad_t. While this encodes the 150 * general layout of icmp6stat into the caller, it doesn't encode its 151 * location, so that future changes to add, for example, per-CPU stats 152 * support won't cause binary compatibility problems for kernel modules. 153 */ 154 void 155 kmod_icmp6stat_inc(int statnum) 156 { 157 158 (*((u_quad_t *)&V_icmp6stat + statnum))++; 159 } 160 161 static void 162 icmp6_errcount(struct icmp6errstat *stat, int type, int code) 163 { 164 switch (type) { 165 case ICMP6_DST_UNREACH: 166 switch (code) { 167 case ICMP6_DST_UNREACH_NOROUTE: 168 stat->icp6errs_dst_unreach_noroute++; 169 return; 170 case ICMP6_DST_UNREACH_ADMIN: 171 stat->icp6errs_dst_unreach_admin++; 172 return; 173 case ICMP6_DST_UNREACH_BEYONDSCOPE: 174 stat->icp6errs_dst_unreach_beyondscope++; 175 return; 176 case ICMP6_DST_UNREACH_ADDR: 177 stat->icp6errs_dst_unreach_addr++; 178 return; 179 case ICMP6_DST_UNREACH_NOPORT: 180 stat->icp6errs_dst_unreach_noport++; 181 return; 182 } 183 break; 184 case ICMP6_PACKET_TOO_BIG: 185 stat->icp6errs_packet_too_big++; 186 return; 187 case ICMP6_TIME_EXCEEDED: 188 switch (code) { 189 case ICMP6_TIME_EXCEED_TRANSIT: 190 stat->icp6errs_time_exceed_transit++; 191 return; 192 case ICMP6_TIME_EXCEED_REASSEMBLY: 193 stat->icp6errs_time_exceed_reassembly++; 194 return; 195 } 196 break; 197 case ICMP6_PARAM_PROB: 198 switch (code) { 199 case ICMP6_PARAMPROB_HEADER: 200 stat->icp6errs_paramprob_header++; 201 return; 202 case ICMP6_PARAMPROB_NEXTHEADER: 203 stat->icp6errs_paramprob_nextheader++; 204 return; 205 case ICMP6_PARAMPROB_OPTION: 206 stat->icp6errs_paramprob_option++; 207 return; 208 } 209 break; 210 case ND_REDIRECT: 211 stat->icp6errs_redirect++; 212 return; 213 } 214 stat->icp6errs_unknown++; 215 } 216 217 /* 218 * A wrapper function for icmp6_error() necessary when the erroneous packet 219 * may not contain enough scope zone information. 220 */ 221 void 222 icmp6_error2(struct mbuf *m, int type, int code, int param, 223 struct ifnet *ifp) 224 { 225 struct ip6_hdr *ip6; 226 227 if (ifp == NULL) 228 return; 229 230 #ifndef PULLDOWN_TEST 231 IP6_EXTHDR_CHECK(m, 0, sizeof(struct ip6_hdr), ); 232 #else 233 if (m->m_len < sizeof(struct ip6_hdr)) { 234 m = m_pullup(m, sizeof(struct ip6_hdr)); 235 if (m == NULL) 236 return; 237 } 238 #endif 239 240 ip6 = mtod(m, struct ip6_hdr *); 241 242 if (in6_setscope(&ip6->ip6_src, ifp, NULL) != 0) 243 return; 244 if (in6_setscope(&ip6->ip6_dst, ifp, NULL) != 0) 245 return; 246 247 icmp6_error(m, type, code, param); 248 } 249 250 /* 251 * Generate an error packet of type error in response to bad IP6 packet. 252 */ 253 void 254 icmp6_error(struct mbuf *m, int type, int code, int param) 255 { 256 struct ip6_hdr *oip6, *nip6; 257 struct icmp6_hdr *icmp6; 258 u_int preplen; 259 int off; 260 int nxt; 261 262 ICMP6STAT_INC(icp6s_error); 263 264 /* count per-type-code statistics */ 265 icmp6_errcount(&V_icmp6stat.icp6s_outerrhist, type, code); 266 267 #ifdef M_DECRYPTED /*not openbsd*/ 268 if (m->m_flags & M_DECRYPTED) { 269 ICMP6STAT_INC(icp6s_canterror); 270 goto freeit; 271 } 272 #endif 273 274 #ifndef PULLDOWN_TEST 275 IP6_EXTHDR_CHECK(m, 0, sizeof(struct ip6_hdr), ); 276 #else 277 if (m->m_len < sizeof(struct ip6_hdr)) { 278 m = m_pullup(m, sizeof(struct ip6_hdr)); 279 if (m == NULL) 280 return; 281 } 282 #endif 283 oip6 = mtod(m, struct ip6_hdr *); 284 285 /* 286 * If the destination address of the erroneous packet is a multicast 287 * address, or the packet was sent using link-layer multicast, 288 * we should basically suppress sending an error (RFC 2463, Section 289 * 2.4). 290 * We have two exceptions (the item e.2 in that section): 291 * - the Packet Too Big message can be sent for path MTU discovery. 292 * - the Parameter Problem Message that can be allowed an icmp6 error 293 * in the option type field. This check has been done in 294 * ip6_unknown_opt(), so we can just check the type and code. 295 */ 296 if ((m->m_flags & (M_BCAST|M_MCAST) || 297 IN6_IS_ADDR_MULTICAST(&oip6->ip6_dst)) && 298 (type != ICMP6_PACKET_TOO_BIG && 299 (type != ICMP6_PARAM_PROB || 300 code != ICMP6_PARAMPROB_OPTION))) 301 goto freeit; 302 303 /* 304 * RFC 2463, 2.4 (e.5): source address check. 305 * XXX: the case of anycast source? 306 */ 307 if (IN6_IS_ADDR_UNSPECIFIED(&oip6->ip6_src) || 308 IN6_IS_ADDR_MULTICAST(&oip6->ip6_src)) 309 goto freeit; 310 311 /* 312 * If we are about to send ICMPv6 against ICMPv6 error/redirect, 313 * don't do it. 314 */ 315 nxt = -1; 316 off = ip6_lasthdr(m, 0, IPPROTO_IPV6, &nxt); 317 if (off >= 0 && nxt == IPPROTO_ICMPV6) { 318 struct icmp6_hdr *icp; 319 320 #ifndef PULLDOWN_TEST 321 IP6_EXTHDR_CHECK(m, 0, off + sizeof(struct icmp6_hdr), ); 322 icp = (struct icmp6_hdr *)(mtod(m, caddr_t) + off); 323 #else 324 IP6_EXTHDR_GET(icp, struct icmp6_hdr *, m, off, 325 sizeof(*icp)); 326 if (icp == NULL) { 327 ICMP6STAT_INC(icp6s_tooshort); 328 return; 329 } 330 #endif 331 if (icp->icmp6_type < ICMP6_ECHO_REQUEST || 332 icp->icmp6_type == ND_REDIRECT) { 333 /* 334 * ICMPv6 error 335 * Special case: for redirect (which is 336 * informational) we must not send icmp6 error. 337 */ 338 ICMP6STAT_INC(icp6s_canterror); 339 goto freeit; 340 } else { 341 /* ICMPv6 informational - send the error */ 342 } 343 } else { 344 /* non-ICMPv6 - send the error */ 345 } 346 347 oip6 = mtod(m, struct ip6_hdr *); /* adjust pointer */ 348 349 /* Finally, do rate limitation check. */ 350 if (icmp6_ratelimit(&oip6->ip6_src, type, code)) { 351 ICMP6STAT_INC(icp6s_toofreq); 352 goto freeit; 353 } 354 355 /* 356 * OK, ICMP6 can be generated. 357 */ 358 359 if (m->m_pkthdr.len >= ICMPV6_PLD_MAXLEN) 360 m_adj(m, ICMPV6_PLD_MAXLEN - m->m_pkthdr.len); 361 362 preplen = sizeof(struct ip6_hdr) + sizeof(struct icmp6_hdr); 363 M_PREPEND(m, preplen, M_DONTWAIT); /* FIB is also copied over. */ 364 if (m && m->m_len < preplen) 365 m = m_pullup(m, preplen); 366 if (m == NULL) { 367 nd6log((LOG_DEBUG, "ENOBUFS in icmp6_error %d\n", __LINE__)); 368 return; 369 } 370 371 nip6 = mtod(m, struct ip6_hdr *); 372 nip6->ip6_src = oip6->ip6_src; 373 nip6->ip6_dst = oip6->ip6_dst; 374 375 in6_clearscope(&oip6->ip6_src); 376 in6_clearscope(&oip6->ip6_dst); 377 378 icmp6 = (struct icmp6_hdr *)(nip6 + 1); 379 icmp6->icmp6_type = type; 380 icmp6->icmp6_code = code; 381 icmp6->icmp6_pptr = htonl((u_int32_t)param); 382 383 /* 384 * icmp6_reflect() is designed to be in the input path. 385 * icmp6_error() can be called from both input and output path, 386 * and if we are in output path rcvif could contain bogus value. 387 * clear m->m_pkthdr.rcvif for safety, we should have enough scope 388 * information in ip header (nip6). 389 */ 390 m->m_pkthdr.rcvif = NULL; 391 392 ICMP6STAT_INC(icp6s_outhist[type]); 393 icmp6_reflect(m, sizeof(struct ip6_hdr)); /* header order: IPv6 - ICMPv6 */ 394 395 return; 396 397 freeit: 398 /* 399 * If we can't tell whether or not we can generate ICMP6, free it. 400 */ 401 m_freem(m); 402 } 403 404 /* 405 * Process a received ICMP6 message. 406 */ 407 int 408 icmp6_input(struct mbuf **mp, int *offp, int proto) 409 { 410 struct mbuf *m = *mp, *n; 411 struct ifnet *ifp; 412 struct ip6_hdr *ip6, *nip6; 413 struct icmp6_hdr *icmp6, *nicmp6; 414 int off = *offp; 415 int icmp6len = m->m_pkthdr.len - *offp; 416 int code, sum, noff; 417 char ip6bufs[INET6_ADDRSTRLEN], ip6bufd[INET6_ADDRSTRLEN]; 418 int ip6len, error; 419 420 ifp = m->m_pkthdr.rcvif; 421 422 #ifndef PULLDOWN_TEST 423 IP6_EXTHDR_CHECK(m, off, sizeof(struct icmp6_hdr), IPPROTO_DONE); 424 /* m might change if M_LOOP. So, call mtod after this */ 425 #endif 426 427 /* 428 * Locate icmp6 structure in mbuf, and check 429 * that not corrupted and of at least minimum length 430 */ 431 432 ip6 = mtod(m, struct ip6_hdr *); 433 ip6len = sizeof(struct ip6_hdr) + ntohs(ip6->ip6_plen); 434 if (icmp6len < sizeof(struct icmp6_hdr)) { 435 ICMP6STAT_INC(icp6s_tooshort); 436 goto freeit; 437 } 438 439 /* 440 * Check multicast group membership. 441 * Note: SSM filters are not applied for ICMPv6 traffic. 442 */ 443 if (IN6_IS_ADDR_MULTICAST(&ip6->ip6_dst)) { 444 struct in6_multi *inm; 445 446 inm = in6m_lookup(ifp, &ip6->ip6_dst); 447 if (inm == NULL) { 448 IP6STAT_INC(ip6s_notmember); 449 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_discard); 450 goto freeit; 451 } 452 } 453 454 /* 455 * calculate the checksum 456 */ 457 #ifndef PULLDOWN_TEST 458 icmp6 = (struct icmp6_hdr *)((caddr_t)ip6 + off); 459 #else 460 IP6_EXTHDR_GET(icmp6, struct icmp6_hdr *, m, off, sizeof(*icmp6)); 461 if (icmp6 == NULL) { 462 ICMP6STAT_INC(icp6s_tooshort); 463 return IPPROTO_DONE; 464 } 465 #endif 466 code = icmp6->icmp6_code; 467 468 if ((sum = in6_cksum(m, IPPROTO_ICMPV6, off, icmp6len)) != 0) { 469 nd6log((LOG_ERR, 470 "ICMP6 checksum error(%d|%x) %s\n", 471 icmp6->icmp6_type, sum, 472 ip6_sprintf(ip6bufs, &ip6->ip6_src))); 473 ICMP6STAT_INC(icp6s_checksum); 474 goto freeit; 475 } 476 477 if (faithprefix_p != NULL && (*faithprefix_p)(&ip6->ip6_dst)) { 478 /* 479 * Deliver very specific ICMP6 type only. 480 * This is important to deliver TOOBIG. Otherwise PMTUD 481 * will not work. 482 */ 483 switch (icmp6->icmp6_type) { 484 case ICMP6_DST_UNREACH: 485 case ICMP6_PACKET_TOO_BIG: 486 case ICMP6_TIME_EXCEEDED: 487 break; 488 default: 489 goto freeit; 490 } 491 } 492 493 ICMP6STAT_INC(icp6s_inhist[icmp6->icmp6_type]); 494 icmp6_ifstat_inc(ifp, ifs6_in_msg); 495 if (icmp6->icmp6_type < ICMP6_INFOMSG_MASK) 496 icmp6_ifstat_inc(ifp, ifs6_in_error); 497 498 switch (icmp6->icmp6_type) { 499 case ICMP6_DST_UNREACH: 500 icmp6_ifstat_inc(ifp, ifs6_in_dstunreach); 501 switch (code) { 502 case ICMP6_DST_UNREACH_NOROUTE: 503 code = PRC_UNREACH_NET; 504 break; 505 case ICMP6_DST_UNREACH_ADMIN: 506 icmp6_ifstat_inc(ifp, ifs6_in_adminprohib); 507 code = PRC_UNREACH_PROTOCOL; /* is this a good code? */ 508 break; 509 case ICMP6_DST_UNREACH_ADDR: 510 code = PRC_HOSTDEAD; 511 break; 512 case ICMP6_DST_UNREACH_BEYONDSCOPE: 513 /* I mean "source address was incorrect." */ 514 code = PRC_PARAMPROB; 515 break; 516 case ICMP6_DST_UNREACH_NOPORT: 517 code = PRC_UNREACH_PORT; 518 break; 519 default: 520 goto badcode; 521 } 522 goto deliver; 523 break; 524 525 case ICMP6_PACKET_TOO_BIG: 526 icmp6_ifstat_inc(ifp, ifs6_in_pkttoobig); 527 528 /* validation is made in icmp6_mtudisc_update */ 529 530 code = PRC_MSGSIZE; 531 532 /* 533 * Updating the path MTU will be done after examining 534 * intermediate extension headers. 535 */ 536 goto deliver; 537 break; 538 539 case ICMP6_TIME_EXCEEDED: 540 icmp6_ifstat_inc(ifp, ifs6_in_timeexceed); 541 switch (code) { 542 case ICMP6_TIME_EXCEED_TRANSIT: 543 code = PRC_TIMXCEED_INTRANS; 544 break; 545 case ICMP6_TIME_EXCEED_REASSEMBLY: 546 code = PRC_TIMXCEED_REASS; 547 break; 548 default: 549 goto badcode; 550 } 551 goto deliver; 552 break; 553 554 case ICMP6_PARAM_PROB: 555 icmp6_ifstat_inc(ifp, ifs6_in_paramprob); 556 switch (code) { 557 case ICMP6_PARAMPROB_NEXTHEADER: 558 code = PRC_UNREACH_PROTOCOL; 559 break; 560 case ICMP6_PARAMPROB_HEADER: 561 case ICMP6_PARAMPROB_OPTION: 562 code = PRC_PARAMPROB; 563 break; 564 default: 565 goto badcode; 566 } 567 goto deliver; 568 break; 569 570 case ICMP6_ECHO_REQUEST: 571 icmp6_ifstat_inc(ifp, ifs6_in_echo); 572 if (code != 0) 573 goto badcode; 574 if ((n = m_copy(m, 0, M_COPYALL)) == NULL) { 575 /* Give up remote */ 576 break; 577 } 578 if ((n->m_flags & M_EXT) != 0 579 || n->m_len < off + sizeof(struct icmp6_hdr)) { 580 struct mbuf *n0 = n; 581 const int maxlen = sizeof(*nip6) + sizeof(*nicmp6); 582 int n0len; 583 584 MGETHDR(n, M_DONTWAIT, n0->m_type); 585 n0len = n0->m_pkthdr.len; /* save for use below */ 586 if (n) 587 M_MOVE_PKTHDR(n, n0); /* FIB copied. */ 588 if (n && maxlen >= MHLEN) { 589 MCLGET(n, M_DONTWAIT); 590 if ((n->m_flags & M_EXT) == 0) { 591 m_free(n); 592 n = NULL; 593 } 594 } 595 if (n == NULL) { 596 /* Give up remote */ 597 m_freem(n0); 598 break; 599 } 600 /* 601 * Copy IPv6 and ICMPv6 only. 602 */ 603 nip6 = mtod(n, struct ip6_hdr *); 604 bcopy(ip6, nip6, sizeof(struct ip6_hdr)); 605 nicmp6 = (struct icmp6_hdr *)(nip6 + 1); 606 bcopy(icmp6, nicmp6, sizeof(struct icmp6_hdr)); 607 noff = sizeof(struct ip6_hdr); 608 /* new mbuf contains only ipv6+icmpv6 headers */ 609 n->m_len = noff + sizeof(struct icmp6_hdr); 610 /* 611 * Adjust mbuf. ip6_plen will be adjusted in 612 * ip6_output(). 613 */ 614 m_adj(n0, off + sizeof(struct icmp6_hdr)); 615 /* recalculate complete packet size */ 616 n->m_pkthdr.len = n0len + (noff - off); 617 n->m_next = n0; 618 } else { 619 nip6 = mtod(n, struct ip6_hdr *); 620 IP6_EXTHDR_GET(nicmp6, struct icmp6_hdr *, n, off, 621 sizeof(*nicmp6)); 622 noff = off; 623 } 624 nicmp6->icmp6_type = ICMP6_ECHO_REPLY; 625 nicmp6->icmp6_code = 0; 626 if (n) { 627 ICMP6STAT_INC(icp6s_reflect); 628 ICMP6STAT_INC(icp6s_outhist[ICMP6_ECHO_REPLY]); 629 icmp6_reflect(n, noff); 630 } 631 break; 632 633 case ICMP6_ECHO_REPLY: 634 icmp6_ifstat_inc(ifp, ifs6_in_echoreply); 635 if (code != 0) 636 goto badcode; 637 break; 638 639 case MLD_LISTENER_QUERY: 640 case MLD_LISTENER_REPORT: 641 case MLD_LISTENER_DONE: 642 case MLDV2_LISTENER_REPORT: 643 /* 644 * Drop MLD traffic which is not link-local, has a hop limit 645 * of greater than 1 hop, or which does not have the 646 * IPv6 HBH Router Alert option. 647 * As IPv6 HBH options are stripped in ip6_input() we must 648 * check an mbuf header flag. 649 * XXX Should we also sanity check that these messages 650 * were directed to a link-local multicast prefix? 651 */ 652 if ((ip6->ip6_hlim != 1) || (m->m_flags & M_RTALERT_MLD) == 0) 653 goto freeit; 654 if (mld_input(m, off, icmp6len) != 0) 655 return (IPPROTO_DONE); 656 /* m stays. */ 657 break; 658 659 case ICMP6_WRUREQUEST: /* ICMP6_FQDN_QUERY */ 660 { 661 enum { WRU, FQDN } mode; 662 663 if (!V_icmp6_nodeinfo) 664 break; 665 666 if (icmp6len == sizeof(struct icmp6_hdr) + 4) 667 mode = WRU; 668 else if (icmp6len >= sizeof(struct icmp6_nodeinfo)) 669 mode = FQDN; 670 else 671 goto badlen; 672 673 if (mode == FQDN) { 674 #ifndef PULLDOWN_TEST 675 IP6_EXTHDR_CHECK(m, off, sizeof(struct icmp6_nodeinfo), 676 IPPROTO_DONE); 677 #endif 678 n = m_copy(m, 0, M_COPYALL); 679 if (n) 680 n = ni6_input(n, off); 681 /* XXX meaningless if n == NULL */ 682 noff = sizeof(struct ip6_hdr); 683 } else { 684 struct prison *pr; 685 u_char *p; 686 int maxlen, maxhlen, hlen; 687 688 /* 689 * XXX: this combination of flags is pointless, 690 * but should we keep this for compatibility? 691 */ 692 if ((V_icmp6_nodeinfo & 5) != 5) 693 break; 694 695 if (code != 0) 696 goto badcode; 697 maxlen = sizeof(*nip6) + sizeof(*nicmp6) + 4; 698 if (maxlen >= MCLBYTES) { 699 /* Give up remote */ 700 break; 701 } 702 MGETHDR(n, M_DONTWAIT, m->m_type); 703 if (n && maxlen > MHLEN) { 704 MCLGET(n, M_DONTWAIT); 705 if ((n->m_flags & M_EXT) == 0) { 706 m_free(n); 707 n = NULL; 708 } 709 } 710 if (n && !m_dup_pkthdr(n, m, M_DONTWAIT)) { 711 /* 712 * Previous code did a blind M_COPY_PKTHDR 713 * and said "just for rcvif". If true, then 714 * we could tolerate the dup failing (due to 715 * the deep copy of the tag chain). For now 716 * be conservative and just fail. 717 */ 718 m_free(n); 719 n = NULL; 720 } 721 if (n == NULL) { 722 /* Give up remote */ 723 break; 724 } 725 n->m_pkthdr.rcvif = NULL; 726 n->m_len = 0; 727 maxhlen = M_TRAILINGSPACE(n) - maxlen; 728 pr = curthread->td_ucred->cr_prison; 729 mtx_lock(&pr->pr_mtx); 730 hlen = strlen(pr->pr_hostname); 731 if (maxhlen > hlen) 732 maxhlen = hlen; 733 /* 734 * Copy IPv6 and ICMPv6 only. 735 */ 736 nip6 = mtod(n, struct ip6_hdr *); 737 bcopy(ip6, nip6, sizeof(struct ip6_hdr)); 738 nicmp6 = (struct icmp6_hdr *)(nip6 + 1); 739 bcopy(icmp6, nicmp6, sizeof(struct icmp6_hdr)); 740 p = (u_char *)(nicmp6 + 1); 741 bzero(p, 4); 742 /* meaningless TTL */ 743 bcopy(pr->pr_hostname, p + 4, maxhlen); 744 mtx_unlock(&pr->pr_mtx); 745 noff = sizeof(struct ip6_hdr); 746 n->m_pkthdr.len = n->m_len = sizeof(struct ip6_hdr) + 747 sizeof(struct icmp6_hdr) + 4 + maxhlen; 748 nicmp6->icmp6_type = ICMP6_WRUREPLY; 749 nicmp6->icmp6_code = 0; 750 } 751 if (n) { 752 ICMP6STAT_INC(icp6s_reflect); 753 ICMP6STAT_INC(icp6s_outhist[ICMP6_WRUREPLY]); 754 icmp6_reflect(n, noff); 755 } 756 break; 757 } 758 759 case ICMP6_WRUREPLY: 760 if (code != 0) 761 goto badcode; 762 break; 763 764 case ND_ROUTER_SOLICIT: 765 icmp6_ifstat_inc(ifp, ifs6_in_routersolicit); 766 if (code != 0) 767 goto badcode; 768 if (icmp6len < sizeof(struct nd_router_solicit)) 769 goto badlen; 770 if ((n = m_copym(m, 0, M_COPYALL, M_DONTWAIT)) == NULL) { 771 /* give up local */ 772 773 /* Send incoming SeND packet to user space. */ 774 if (send_sendso_input_hook != NULL) { 775 IP6_EXTHDR_CHECK(m, off, 776 icmp6len, IPPROTO_DONE); 777 error = send_sendso_input_hook(m, ifp, 778 SND_IN, ip6len); 779 /* -1 == no app on SEND socket */ 780 if (error == 0) 781 return (IPPROTO_DONE); 782 nd6_rs_input(m, off, icmp6len); 783 } else 784 nd6_rs_input(m, off, icmp6len); 785 m = NULL; 786 goto freeit; 787 } 788 if (send_sendso_input_hook != NULL) { 789 IP6_EXTHDR_CHECK(n, off, 790 icmp6len, IPPROTO_DONE); 791 error = send_sendso_input_hook(n, ifp, 792 SND_IN, ip6len); 793 if (error == 0) 794 goto freeit; 795 /* -1 == no app on SEND socket */ 796 nd6_rs_input(n, off, icmp6len); 797 } else 798 nd6_rs_input(n, off, icmp6len); 799 /* m stays. */ 800 break; 801 802 case ND_ROUTER_ADVERT: 803 icmp6_ifstat_inc(ifp, ifs6_in_routeradvert); 804 if (code != 0) 805 goto badcode; 806 if (icmp6len < sizeof(struct nd_router_advert)) 807 goto badlen; 808 if ((n = m_copym(m, 0, M_COPYALL, M_DONTWAIT)) == NULL) { 809 810 /* Send incoming SeND-protected/ND packet to user space. */ 811 if (send_sendso_input_hook != NULL) { 812 error = send_sendso_input_hook(m, ifp, 813 SND_IN, ip6len); 814 if (error == 0) 815 return (IPPROTO_DONE); 816 nd6_ra_input(m, off, icmp6len); 817 } else 818 nd6_ra_input(m, off, icmp6len); 819 m = NULL; 820 goto freeit; 821 } 822 if (send_sendso_input_hook != NULL) { 823 error = send_sendso_input_hook(n, ifp, 824 SND_IN, ip6len); 825 if (error == 0) 826 goto freeit; 827 nd6_ra_input(n, off, icmp6len); 828 } else 829 nd6_ra_input(n, off, icmp6len); 830 /* m stays. */ 831 break; 832 833 case ND_NEIGHBOR_SOLICIT: 834 icmp6_ifstat_inc(ifp, ifs6_in_neighborsolicit); 835 if (code != 0) 836 goto badcode; 837 if (icmp6len < sizeof(struct nd_neighbor_solicit)) 838 goto badlen; 839 if ((n = m_copym(m, 0, M_COPYALL, M_DONTWAIT)) == NULL) { 840 if (send_sendso_input_hook != NULL) { 841 error = send_sendso_input_hook(m, ifp, 842 SND_IN, ip6len); 843 if (error == 0) 844 return (IPPROTO_DONE); 845 nd6_ns_input(m, off, icmp6len); 846 } else 847 nd6_ns_input(m, off, icmp6len); 848 m = NULL; 849 goto freeit; 850 } 851 if (send_sendso_input_hook != NULL) { 852 error = send_sendso_input_hook(n, ifp, 853 SND_IN, ip6len); 854 if (error == 0) 855 goto freeit; 856 nd6_ns_input(n, off, icmp6len); 857 } else 858 nd6_ns_input(n, off, icmp6len); 859 /* m stays. */ 860 break; 861 862 case ND_NEIGHBOR_ADVERT: 863 icmp6_ifstat_inc(ifp, ifs6_in_neighboradvert); 864 if (code != 0) 865 goto badcode; 866 if (icmp6len < sizeof(struct nd_neighbor_advert)) 867 goto badlen; 868 if ((n = m_copym(m, 0, M_COPYALL, M_DONTWAIT)) == NULL) { 869 870 /* Send incoming SeND-protected/ND packet to user space. */ 871 if (send_sendso_input_hook != NULL) { 872 error = send_sendso_input_hook(m, ifp, 873 SND_IN, ip6len); 874 if (error == 0) 875 return (IPPROTO_DONE); 876 nd6_na_input(m, off, icmp6len); 877 } else 878 nd6_na_input(m, off, icmp6len); 879 m = NULL; 880 goto freeit; 881 } 882 if (send_sendso_input_hook != NULL) { 883 error = send_sendso_input_hook(n, ifp, 884 SND_IN, ip6len); 885 if (error == 0) 886 goto freeit; 887 nd6_na_input(n, off, icmp6len); 888 } else 889 nd6_na_input(n, off, icmp6len); 890 /* m stays. */ 891 break; 892 893 case ND_REDIRECT: 894 icmp6_ifstat_inc(ifp, ifs6_in_redirect); 895 if (code != 0) 896 goto badcode; 897 if (icmp6len < sizeof(struct nd_redirect)) 898 goto badlen; 899 if ((n = m_copym(m, 0, M_COPYALL, M_DONTWAIT)) == NULL) { 900 if (send_sendso_input_hook != NULL) { 901 error = send_sendso_input_hook(m, ifp, 902 SND_IN, ip6len); 903 if (error == 0) 904 return (IPPROTO_DONE); 905 icmp6_redirect_input(m, off); 906 } else 907 icmp6_redirect_input(m, off); 908 m = NULL; 909 goto freeit; 910 } 911 if (send_sendso_input_hook != NULL) { 912 error = send_sendso_input_hook(n, ifp, 913 SND_IN, ip6len); 914 if (error == 0) 915 goto freeit; 916 icmp6_redirect_input(n, off); 917 } else 918 icmp6_redirect_input(n, off); 919 /* m stays. */ 920 break; 921 922 case ICMP6_ROUTER_RENUMBERING: 923 if (code != ICMP6_ROUTER_RENUMBERING_COMMAND && 924 code != ICMP6_ROUTER_RENUMBERING_RESULT) 925 goto badcode; 926 if (icmp6len < sizeof(struct icmp6_router_renum)) 927 goto badlen; 928 break; 929 930 default: 931 nd6log((LOG_DEBUG, 932 "icmp6_input: unknown type %d(src=%s, dst=%s, ifid=%d)\n", 933 icmp6->icmp6_type, ip6_sprintf(ip6bufs, &ip6->ip6_src), 934 ip6_sprintf(ip6bufd, &ip6->ip6_dst), 935 ifp ? ifp->if_index : 0)); 936 if (icmp6->icmp6_type < ICMP6_ECHO_REQUEST) { 937 /* ICMPv6 error: MUST deliver it by spec... */ 938 code = PRC_NCMDS; 939 /* deliver */ 940 } else { 941 /* ICMPv6 informational: MUST not deliver */ 942 break; 943 } 944 deliver: 945 if (icmp6_notify_error(&m, off, icmp6len, code) != 0) { 946 /* In this case, m should've been freed. */ 947 return (IPPROTO_DONE); 948 } 949 break; 950 951 badcode: 952 ICMP6STAT_INC(icp6s_badcode); 953 break; 954 955 badlen: 956 ICMP6STAT_INC(icp6s_badlen); 957 break; 958 } 959 960 /* deliver the packet to appropriate sockets */ 961 icmp6_rip6_input(&m, *offp); 962 963 return IPPROTO_DONE; 964 965 freeit: 966 m_freem(m); 967 return IPPROTO_DONE; 968 } 969 970 static int 971 icmp6_notify_error(struct mbuf **mp, int off, int icmp6len, int code) 972 { 973 struct mbuf *m = *mp; 974 struct icmp6_hdr *icmp6; 975 struct ip6_hdr *eip6; 976 u_int32_t notifymtu; 977 struct sockaddr_in6 icmp6src, icmp6dst; 978 979 if (icmp6len < sizeof(struct icmp6_hdr) + sizeof(struct ip6_hdr)) { 980 ICMP6STAT_INC(icp6s_tooshort); 981 goto freeit; 982 } 983 #ifndef PULLDOWN_TEST 984 IP6_EXTHDR_CHECK(m, off, 985 sizeof(struct icmp6_hdr) + sizeof(struct ip6_hdr), -1); 986 icmp6 = (struct icmp6_hdr *)(mtod(m, caddr_t) + off); 987 #else 988 IP6_EXTHDR_GET(icmp6, struct icmp6_hdr *, m, off, 989 sizeof(*icmp6) + sizeof(struct ip6_hdr)); 990 if (icmp6 == NULL) { 991 ICMP6STAT_INC(icp6s_tooshort); 992 return (-1); 993 } 994 #endif 995 eip6 = (struct ip6_hdr *)(icmp6 + 1); 996 997 /* Detect the upper level protocol */ 998 { 999 void (*ctlfunc)(int, struct sockaddr *, void *); 1000 u_int8_t nxt = eip6->ip6_nxt; 1001 int eoff = off + sizeof(struct icmp6_hdr) + 1002 sizeof(struct ip6_hdr); 1003 struct ip6ctlparam ip6cp; 1004 struct in6_addr *finaldst = NULL; 1005 int icmp6type = icmp6->icmp6_type; 1006 struct ip6_frag *fh; 1007 struct ip6_rthdr *rth; 1008 struct ip6_rthdr0 *rth0; 1009 int rthlen; 1010 1011 while (1) { /* XXX: should avoid infinite loop explicitly? */ 1012 struct ip6_ext *eh; 1013 1014 switch (nxt) { 1015 case IPPROTO_HOPOPTS: 1016 case IPPROTO_DSTOPTS: 1017 case IPPROTO_AH: 1018 #ifndef PULLDOWN_TEST 1019 IP6_EXTHDR_CHECK(m, 0, 1020 eoff + sizeof(struct ip6_ext), -1); 1021 eh = (struct ip6_ext *)(mtod(m, caddr_t) + eoff); 1022 #else 1023 IP6_EXTHDR_GET(eh, struct ip6_ext *, m, 1024 eoff, sizeof(*eh)); 1025 if (eh == NULL) { 1026 ICMP6STAT_INC(icp6s_tooshort); 1027 return (-1); 1028 } 1029 #endif 1030 1031 if (nxt == IPPROTO_AH) 1032 eoff += (eh->ip6e_len + 2) << 2; 1033 else 1034 eoff += (eh->ip6e_len + 1) << 3; 1035 nxt = eh->ip6e_nxt; 1036 break; 1037 case IPPROTO_ROUTING: 1038 /* 1039 * When the erroneous packet contains a 1040 * routing header, we should examine the 1041 * header to determine the final destination. 1042 * Otherwise, we can't properly update 1043 * information that depends on the final 1044 * destination (e.g. path MTU). 1045 */ 1046 #ifndef PULLDOWN_TEST 1047 IP6_EXTHDR_CHECK(m, 0, eoff + sizeof(*rth), -1); 1048 rth = (struct ip6_rthdr *) 1049 (mtod(m, caddr_t) + eoff); 1050 #else 1051 IP6_EXTHDR_GET(rth, struct ip6_rthdr *, m, 1052 eoff, sizeof(*rth)); 1053 if (rth == NULL) { 1054 ICMP6STAT_INC(icp6s_tooshort); 1055 return (-1); 1056 } 1057 #endif 1058 rthlen = (rth->ip6r_len + 1) << 3; 1059 /* 1060 * XXX: currently there is no 1061 * officially defined type other 1062 * than type-0. 1063 * Note that if the segment left field 1064 * is 0, all intermediate hops must 1065 * have been passed. 1066 */ 1067 if (rth->ip6r_segleft && 1068 rth->ip6r_type == IPV6_RTHDR_TYPE_0) { 1069 int hops; 1070 1071 #ifndef PULLDOWN_TEST 1072 IP6_EXTHDR_CHECK(m, 0, eoff + rthlen, -1); 1073 rth0 = (struct ip6_rthdr0 *) 1074 (mtod(m, caddr_t) + eoff); 1075 #else 1076 IP6_EXTHDR_GET(rth0, 1077 struct ip6_rthdr0 *, m, 1078 eoff, rthlen); 1079 if (rth0 == NULL) { 1080 ICMP6STAT_INC(icp6s_tooshort); 1081 return (-1); 1082 } 1083 #endif 1084 /* just ignore a bogus header */ 1085 if ((rth0->ip6r0_len % 2) == 0 && 1086 (hops = rth0->ip6r0_len/2)) 1087 finaldst = (struct in6_addr *)(rth0 + 1) + (hops - 1); 1088 } 1089 eoff += rthlen; 1090 nxt = rth->ip6r_nxt; 1091 break; 1092 case IPPROTO_FRAGMENT: 1093 #ifndef PULLDOWN_TEST 1094 IP6_EXTHDR_CHECK(m, 0, eoff + 1095 sizeof(struct ip6_frag), -1); 1096 fh = (struct ip6_frag *)(mtod(m, caddr_t) + 1097 eoff); 1098 #else 1099 IP6_EXTHDR_GET(fh, struct ip6_frag *, m, 1100 eoff, sizeof(*fh)); 1101 if (fh == NULL) { 1102 ICMP6STAT_INC(icp6s_tooshort); 1103 return (-1); 1104 } 1105 #endif 1106 /* 1107 * Data after a fragment header is meaningless 1108 * unless it is the first fragment, but 1109 * we'll go to the notify label for path MTU 1110 * discovery. 1111 */ 1112 if (fh->ip6f_offlg & IP6F_OFF_MASK) 1113 goto notify; 1114 1115 eoff += sizeof(struct ip6_frag); 1116 nxt = fh->ip6f_nxt; 1117 break; 1118 default: 1119 /* 1120 * This case includes ESP and the No Next 1121 * Header. In such cases going to the notify 1122 * label does not have any meaning 1123 * (i.e. ctlfunc will be NULL), but we go 1124 * anyway since we might have to update 1125 * path MTU information. 1126 */ 1127 goto notify; 1128 } 1129 } 1130 notify: 1131 #ifndef PULLDOWN_TEST 1132 icmp6 = (struct icmp6_hdr *)(mtod(m, caddr_t) + off); 1133 #else 1134 IP6_EXTHDR_GET(icmp6, struct icmp6_hdr *, m, off, 1135 sizeof(*icmp6) + sizeof(struct ip6_hdr)); 1136 if (icmp6 == NULL) { 1137 ICMP6STAT_INC(icp6s_tooshort); 1138 return (-1); 1139 } 1140 #endif 1141 1142 /* 1143 * retrieve parameters from the inner IPv6 header, and convert 1144 * them into sockaddr structures. 1145 * XXX: there is no guarantee that the source or destination 1146 * addresses of the inner packet are in the same scope as 1147 * the addresses of the icmp packet. But there is no other 1148 * way to determine the zone. 1149 */ 1150 eip6 = (struct ip6_hdr *)(icmp6 + 1); 1151 1152 bzero(&icmp6dst, sizeof(icmp6dst)); 1153 icmp6dst.sin6_len = sizeof(struct sockaddr_in6); 1154 icmp6dst.sin6_family = AF_INET6; 1155 if (finaldst == NULL) 1156 icmp6dst.sin6_addr = eip6->ip6_dst; 1157 else 1158 icmp6dst.sin6_addr = *finaldst; 1159 if (in6_setscope(&icmp6dst.sin6_addr, m->m_pkthdr.rcvif, NULL)) 1160 goto freeit; 1161 bzero(&icmp6src, sizeof(icmp6src)); 1162 icmp6src.sin6_len = sizeof(struct sockaddr_in6); 1163 icmp6src.sin6_family = AF_INET6; 1164 icmp6src.sin6_addr = eip6->ip6_src; 1165 if (in6_setscope(&icmp6src.sin6_addr, m->m_pkthdr.rcvif, NULL)) 1166 goto freeit; 1167 icmp6src.sin6_flowinfo = 1168 (eip6->ip6_flow & IPV6_FLOWLABEL_MASK); 1169 1170 if (finaldst == NULL) 1171 finaldst = &eip6->ip6_dst; 1172 ip6cp.ip6c_m = m; 1173 ip6cp.ip6c_icmp6 = icmp6; 1174 ip6cp.ip6c_ip6 = (struct ip6_hdr *)(icmp6 + 1); 1175 ip6cp.ip6c_off = eoff; 1176 ip6cp.ip6c_finaldst = finaldst; 1177 ip6cp.ip6c_src = &icmp6src; 1178 ip6cp.ip6c_nxt = nxt; 1179 1180 m_addr_changed(m); 1181 1182 if (icmp6type == ICMP6_PACKET_TOO_BIG) { 1183 notifymtu = ntohl(icmp6->icmp6_mtu); 1184 ip6cp.ip6c_cmdarg = (void *)¬ifymtu; 1185 icmp6_mtudisc_update(&ip6cp, 1); /*XXX*/ 1186 } 1187 1188 ctlfunc = (void (*)(int, struct sockaddr *, void *)) 1189 (inet6sw[ip6_protox[nxt]].pr_ctlinput); 1190 if (ctlfunc) { 1191 (void) (*ctlfunc)(code, (struct sockaddr *)&icmp6dst, 1192 &ip6cp); 1193 } 1194 } 1195 *mp = m; 1196 return (0); 1197 1198 freeit: 1199 m_freem(m); 1200 return (-1); 1201 } 1202 1203 void 1204 icmp6_mtudisc_update(struct ip6ctlparam *ip6cp, int validated) 1205 { 1206 struct in6_addr *dst = ip6cp->ip6c_finaldst; 1207 struct icmp6_hdr *icmp6 = ip6cp->ip6c_icmp6; 1208 struct mbuf *m = ip6cp->ip6c_m; /* will be necessary for scope issue */ 1209 u_int mtu = ntohl(icmp6->icmp6_mtu); 1210 struct in_conninfo inc; 1211 1212 #if 0 1213 /* 1214 * RFC2460 section 5, last paragraph. 1215 * even though minimum link MTU for IPv6 is IPV6_MMTU, 1216 * we may see ICMPv6 too big with mtu < IPV6_MMTU 1217 * due to packet translator in the middle. 1218 * see ip6_output() and ip6_getpmtu() "alwaysfrag" case for 1219 * special handling. 1220 */ 1221 if (mtu < IPV6_MMTU) 1222 return; 1223 #endif 1224 1225 /* 1226 * we reject ICMPv6 too big with abnormally small value. 1227 * XXX what is the good definition of "abnormally small"? 1228 */ 1229 if (mtu < sizeof(struct ip6_hdr) + sizeof(struct ip6_frag) + 8) 1230 return; 1231 1232 if (!validated) 1233 return; 1234 1235 /* 1236 * In case the suggested mtu is less than IPV6_MMTU, we 1237 * only need to remember that it was for above mentioned 1238 * "alwaysfrag" case. 1239 * Try to be as close to the spec as possible. 1240 */ 1241 if (mtu < IPV6_MMTU) 1242 mtu = IPV6_MMTU - 8; 1243 1244 bzero(&inc, sizeof(inc)); 1245 inc.inc_flags |= INC_ISIPV6; 1246 inc.inc6_faddr = *dst; 1247 if (in6_setscope(&inc.inc6_faddr, m->m_pkthdr.rcvif, NULL)) 1248 return; 1249 1250 if (mtu < tcp_maxmtu6(&inc, NULL)) { 1251 tcp_hc_updatemtu(&inc, mtu); 1252 ICMP6STAT_INC(icp6s_pmtuchg); 1253 } 1254 } 1255 1256 /* 1257 * Process a Node Information Query packet, based on 1258 * draft-ietf-ipngwg-icmp-name-lookups-07. 1259 * 1260 * Spec incompatibilities: 1261 * - IPv6 Subject address handling 1262 * - IPv4 Subject address handling support missing 1263 * - Proxy reply (answer even if it's not for me) 1264 * - joins NI group address at in6_ifattach() time only, does not cope 1265 * with hostname changes by sethostname(3) 1266 */ 1267 static struct mbuf * 1268 ni6_input(struct mbuf *m, int off) 1269 { 1270 struct icmp6_nodeinfo *ni6, *nni6; 1271 struct mbuf *n = NULL; 1272 struct prison *pr; 1273 u_int16_t qtype; 1274 int subjlen; 1275 int replylen = sizeof(struct ip6_hdr) + sizeof(struct icmp6_nodeinfo); 1276 struct ni_reply_fqdn *fqdn; 1277 int addrs; /* for NI_QTYPE_NODEADDR */ 1278 struct ifnet *ifp = NULL; /* for NI_QTYPE_NODEADDR */ 1279 struct in6_addr in6_subj; /* subject address */ 1280 struct ip6_hdr *ip6; 1281 int oldfqdn = 0; /* if 1, return pascal string (03 draft) */ 1282 char *subj = NULL; 1283 struct in6_ifaddr *ia6 = NULL; 1284 1285 ip6 = mtod(m, struct ip6_hdr *); 1286 #ifndef PULLDOWN_TEST 1287 ni6 = (struct icmp6_nodeinfo *)(mtod(m, caddr_t) + off); 1288 #else 1289 IP6_EXTHDR_GET(ni6, struct icmp6_nodeinfo *, m, off, sizeof(*ni6)); 1290 if (ni6 == NULL) { 1291 /* m is already reclaimed */ 1292 return (NULL); 1293 } 1294 #endif 1295 1296 /* 1297 * Validate IPv6 source address. 1298 * The default configuration MUST be to refuse answering queries from 1299 * global-scope addresses according to RFC4602. 1300 * Notes: 1301 * - it's not very clear what "refuse" means; this implementation 1302 * simply drops it. 1303 * - it's not very easy to identify global-scope (unicast) addresses 1304 * since there are many prefixes for them. It should be safer 1305 * and in practice sufficient to check "all" but loopback and 1306 * link-local (note that site-local unicast was deprecated and 1307 * ULA is defined as global scope-wise) 1308 */ 1309 if ((V_icmp6_nodeinfo & ICMP6_NODEINFO_GLOBALOK) == 0 && 1310 !IN6_IS_ADDR_LOOPBACK(&ip6->ip6_src) && 1311 !IN6_IS_ADDR_LINKLOCAL(&ip6->ip6_src)) 1312 goto bad; 1313 1314 /* 1315 * Validate IPv6 destination address. 1316 * 1317 * The Responder must discard the Query without further processing 1318 * unless it is one of the Responder's unicast or anycast addresses, or 1319 * a link-local scope multicast address which the Responder has joined. 1320 * [RFC4602, Section 5.] 1321 */ 1322 if (IN6_IS_ADDR_MULTICAST(&ip6->ip6_dst)) { 1323 if (!IN6_IS_ADDR_MC_LINKLOCAL(&ip6->ip6_dst)) 1324 goto bad; 1325 /* else it's a link-local multicast, fine */ 1326 } else { /* unicast or anycast */ 1327 if ((ia6 = ip6_getdstifaddr(m)) == NULL) 1328 goto bad; /* XXX impossible */ 1329 1330 if ((ia6->ia6_flags & IN6_IFF_TEMPORARY) && 1331 !(V_icmp6_nodeinfo & ICMP6_NODEINFO_TMPADDROK)) { 1332 ifa_free(&ia6->ia_ifa); 1333 nd6log((LOG_DEBUG, "ni6_input: ignore node info to " 1334 "a temporary address in %s:%d", 1335 __FILE__, __LINE__)); 1336 goto bad; 1337 } 1338 ifa_free(&ia6->ia_ifa); 1339 } 1340 1341 /* validate query Subject field. */ 1342 qtype = ntohs(ni6->ni_qtype); 1343 subjlen = m->m_pkthdr.len - off - sizeof(struct icmp6_nodeinfo); 1344 switch (qtype) { 1345 case NI_QTYPE_NOOP: 1346 case NI_QTYPE_SUPTYPES: 1347 /* 07 draft */ 1348 if (ni6->ni_code == ICMP6_NI_SUBJ_FQDN && subjlen == 0) 1349 break; 1350 /* FALLTHROUGH */ 1351 case NI_QTYPE_FQDN: 1352 case NI_QTYPE_NODEADDR: 1353 case NI_QTYPE_IPV4ADDR: 1354 switch (ni6->ni_code) { 1355 case ICMP6_NI_SUBJ_IPV6: 1356 #if ICMP6_NI_SUBJ_IPV6 != 0 1357 case 0: 1358 #endif 1359 /* 1360 * backward compatibility - try to accept 03 draft 1361 * format, where no Subject is present. 1362 */ 1363 if (qtype == NI_QTYPE_FQDN && ni6->ni_code == 0 && 1364 subjlen == 0) { 1365 oldfqdn++; 1366 break; 1367 } 1368 #if ICMP6_NI_SUBJ_IPV6 != 0 1369 if (ni6->ni_code != ICMP6_NI_SUBJ_IPV6) 1370 goto bad; 1371 #endif 1372 1373 if (subjlen != sizeof(struct in6_addr)) 1374 goto bad; 1375 1376 /* 1377 * Validate Subject address. 1378 * 1379 * Not sure what exactly "address belongs to the node" 1380 * means in the spec, is it just unicast, or what? 1381 * 1382 * At this moment we consider Subject address as 1383 * "belong to the node" if the Subject address equals 1384 * to the IPv6 destination address; validation for 1385 * IPv6 destination address should have done enough 1386 * check for us. 1387 * 1388 * We do not do proxy at this moment. 1389 */ 1390 /* m_pulldown instead of copy? */ 1391 m_copydata(m, off + sizeof(struct icmp6_nodeinfo), 1392 subjlen, (caddr_t)&in6_subj); 1393 if (in6_setscope(&in6_subj, m->m_pkthdr.rcvif, NULL)) 1394 goto bad; 1395 1396 subj = (char *)&in6_subj; 1397 if (IN6_ARE_ADDR_EQUAL(&ip6->ip6_dst, &in6_subj)) 1398 break; 1399 1400 /* 1401 * XXX if we are to allow other cases, we should really 1402 * be careful about scope here. 1403 * basically, we should disallow queries toward IPv6 1404 * destination X with subject Y, 1405 * if scope(X) > scope(Y). 1406 * if we allow scope(X) > scope(Y), it will result in 1407 * information leakage across scope boundary. 1408 */ 1409 goto bad; 1410 1411 case ICMP6_NI_SUBJ_FQDN: 1412 /* 1413 * Validate Subject name with gethostname(3). 1414 * 1415 * The behavior may need some debate, since: 1416 * - we are not sure if the node has FQDN as 1417 * hostname (returned by gethostname(3)). 1418 * - the code does wildcard match for truncated names. 1419 * however, we are not sure if we want to perform 1420 * wildcard match, if gethostname(3) side has 1421 * truncated hostname. 1422 */ 1423 pr = curthread->td_ucred->cr_prison; 1424 mtx_lock(&pr->pr_mtx); 1425 n = ni6_nametodns(pr->pr_hostname, 1426 strlen(pr->pr_hostname), 0); 1427 mtx_unlock(&pr->pr_mtx); 1428 if (!n || n->m_next || n->m_len == 0) 1429 goto bad; 1430 IP6_EXTHDR_GET(subj, char *, m, 1431 off + sizeof(struct icmp6_nodeinfo), subjlen); 1432 if (subj == NULL) 1433 goto bad; 1434 if (!ni6_dnsmatch(subj, subjlen, mtod(n, const char *), 1435 n->m_len)) { 1436 goto bad; 1437 } 1438 m_freem(n); 1439 n = NULL; 1440 break; 1441 1442 case ICMP6_NI_SUBJ_IPV4: /* XXX: to be implemented? */ 1443 default: 1444 goto bad; 1445 } 1446 break; 1447 } 1448 1449 /* refuse based on configuration. XXX ICMP6_NI_REFUSED? */ 1450 switch (qtype) { 1451 case NI_QTYPE_FQDN: 1452 if ((V_icmp6_nodeinfo & ICMP6_NODEINFO_FQDNOK) == 0) 1453 goto bad; 1454 break; 1455 case NI_QTYPE_NODEADDR: 1456 case NI_QTYPE_IPV4ADDR: 1457 if ((V_icmp6_nodeinfo & ICMP6_NODEINFO_NODEADDROK) == 0) 1458 goto bad; 1459 break; 1460 } 1461 1462 /* guess reply length */ 1463 switch (qtype) { 1464 case NI_QTYPE_NOOP: 1465 break; /* no reply data */ 1466 case NI_QTYPE_SUPTYPES: 1467 replylen += sizeof(u_int32_t); 1468 break; 1469 case NI_QTYPE_FQDN: 1470 /* XXX will append an mbuf */ 1471 replylen += offsetof(struct ni_reply_fqdn, ni_fqdn_namelen); 1472 break; 1473 case NI_QTYPE_NODEADDR: 1474 addrs = ni6_addrs(ni6, m, &ifp, (struct in6_addr *)subj); 1475 if ((replylen += addrs * (sizeof(struct in6_addr) + 1476 sizeof(u_int32_t))) > MCLBYTES) 1477 replylen = MCLBYTES; /* XXX: will truncate pkt later */ 1478 break; 1479 case NI_QTYPE_IPV4ADDR: 1480 /* unsupported - should respond with unknown Qtype? */ 1481 break; 1482 default: 1483 /* 1484 * XXX: We must return a reply with the ICMP6 code 1485 * `unknown Qtype' in this case. However we regard the case 1486 * as an FQDN query for backward compatibility. 1487 * Older versions set a random value to this field, 1488 * so it rarely varies in the defined qtypes. 1489 * But the mechanism is not reliable... 1490 * maybe we should obsolete older versions. 1491 */ 1492 qtype = NI_QTYPE_FQDN; 1493 /* XXX will append an mbuf */ 1494 replylen += offsetof(struct ni_reply_fqdn, ni_fqdn_namelen); 1495 oldfqdn++; 1496 break; 1497 } 1498 1499 /* allocate an mbuf to reply. */ 1500 MGETHDR(n, M_DONTWAIT, m->m_type); 1501 if (n == NULL) { 1502 m_freem(m); 1503 return (NULL); 1504 } 1505 M_MOVE_PKTHDR(n, m); /* just for recvif and FIB */ 1506 if (replylen > MHLEN) { 1507 if (replylen > MCLBYTES) { 1508 /* 1509 * XXX: should we try to allocate more? But MCLBYTES 1510 * is probably much larger than IPV6_MMTU... 1511 */ 1512 goto bad; 1513 } 1514 MCLGET(n, M_DONTWAIT); 1515 if ((n->m_flags & M_EXT) == 0) { 1516 goto bad; 1517 } 1518 } 1519 n->m_pkthdr.len = n->m_len = replylen; 1520 1521 /* copy mbuf header and IPv6 + Node Information base headers */ 1522 bcopy(mtod(m, caddr_t), mtod(n, caddr_t), sizeof(struct ip6_hdr)); 1523 nni6 = (struct icmp6_nodeinfo *)(mtod(n, struct ip6_hdr *) + 1); 1524 bcopy((caddr_t)ni6, (caddr_t)nni6, sizeof(struct icmp6_nodeinfo)); 1525 1526 /* qtype dependent procedure */ 1527 switch (qtype) { 1528 case NI_QTYPE_NOOP: 1529 nni6->ni_code = ICMP6_NI_SUCCESS; 1530 nni6->ni_flags = 0; 1531 break; 1532 case NI_QTYPE_SUPTYPES: 1533 { 1534 u_int32_t v; 1535 nni6->ni_code = ICMP6_NI_SUCCESS; 1536 nni6->ni_flags = htons(0x0000); /* raw bitmap */ 1537 /* supports NOOP, SUPTYPES, FQDN, and NODEADDR */ 1538 v = (u_int32_t)htonl(0x0000000f); 1539 bcopy(&v, nni6 + 1, sizeof(u_int32_t)); 1540 break; 1541 } 1542 case NI_QTYPE_FQDN: 1543 nni6->ni_code = ICMP6_NI_SUCCESS; 1544 fqdn = (struct ni_reply_fqdn *)(mtod(n, caddr_t) + 1545 sizeof(struct ip6_hdr) + sizeof(struct icmp6_nodeinfo)); 1546 nni6->ni_flags = 0; /* XXX: meaningless TTL */ 1547 fqdn->ni_fqdn_ttl = 0; /* ditto. */ 1548 /* 1549 * XXX do we really have FQDN in hostname? 1550 */ 1551 pr = curthread->td_ucred->cr_prison; 1552 mtx_lock(&pr->pr_mtx); 1553 n->m_next = ni6_nametodns(pr->pr_hostname, 1554 strlen(pr->pr_hostname), oldfqdn); 1555 mtx_unlock(&pr->pr_mtx); 1556 if (n->m_next == NULL) 1557 goto bad; 1558 /* XXX we assume that n->m_next is not a chain */ 1559 if (n->m_next->m_next != NULL) 1560 goto bad; 1561 n->m_pkthdr.len += n->m_next->m_len; 1562 break; 1563 case NI_QTYPE_NODEADDR: 1564 { 1565 int lenlim, copied; 1566 1567 nni6->ni_code = ICMP6_NI_SUCCESS; 1568 n->m_pkthdr.len = n->m_len = 1569 sizeof(struct ip6_hdr) + sizeof(struct icmp6_nodeinfo); 1570 lenlim = M_TRAILINGSPACE(n); 1571 copied = ni6_store_addrs(ni6, nni6, ifp, lenlim); 1572 /* XXX: reset mbuf length */ 1573 n->m_pkthdr.len = n->m_len = sizeof(struct ip6_hdr) + 1574 sizeof(struct icmp6_nodeinfo) + copied; 1575 break; 1576 } 1577 default: 1578 break; /* XXX impossible! */ 1579 } 1580 1581 nni6->ni_type = ICMP6_NI_REPLY; 1582 m_freem(m); 1583 return (n); 1584 1585 bad: 1586 m_freem(m); 1587 if (n) 1588 m_freem(n); 1589 return (NULL); 1590 } 1591 1592 /* 1593 * make a mbuf with DNS-encoded string. no compression support. 1594 * 1595 * XXX names with less than 2 dots (like "foo" or "foo.section") will be 1596 * treated as truncated name (two \0 at the end). this is a wild guess. 1597 * 1598 * old - return pascal string if non-zero 1599 */ 1600 static struct mbuf * 1601 ni6_nametodns(const char *name, int namelen, int old) 1602 { 1603 struct mbuf *m; 1604 char *cp, *ep; 1605 const char *p, *q; 1606 int i, len, nterm; 1607 1608 if (old) 1609 len = namelen + 1; 1610 else 1611 len = MCLBYTES; 1612 1613 /* because MAXHOSTNAMELEN is usually 256, we use cluster mbuf */ 1614 MGET(m, M_DONTWAIT, MT_DATA); 1615 if (m && len > MLEN) { 1616 MCLGET(m, M_DONTWAIT); 1617 if ((m->m_flags & M_EXT) == 0) 1618 goto fail; 1619 } 1620 if (!m) 1621 goto fail; 1622 m->m_next = NULL; 1623 1624 if (old) { 1625 m->m_len = len; 1626 *mtod(m, char *) = namelen; 1627 bcopy(name, mtod(m, char *) + 1, namelen); 1628 return m; 1629 } else { 1630 m->m_len = 0; 1631 cp = mtod(m, char *); 1632 ep = mtod(m, char *) + M_TRAILINGSPACE(m); 1633 1634 /* if not certain about my name, return empty buffer */ 1635 if (namelen == 0) 1636 return m; 1637 1638 /* 1639 * guess if it looks like shortened hostname, or FQDN. 1640 * shortened hostname needs two trailing "\0". 1641 */ 1642 i = 0; 1643 for (p = name; p < name + namelen; p++) { 1644 if (*p && *p == '.') 1645 i++; 1646 } 1647 if (i < 2) 1648 nterm = 2; 1649 else 1650 nterm = 1; 1651 1652 p = name; 1653 while (cp < ep && p < name + namelen) { 1654 i = 0; 1655 for (q = p; q < name + namelen && *q && *q != '.'; q++) 1656 i++; 1657 /* result does not fit into mbuf */ 1658 if (cp + i + 1 >= ep) 1659 goto fail; 1660 /* 1661 * DNS label length restriction, RFC1035 page 8. 1662 * "i == 0" case is included here to avoid returning 1663 * 0-length label on "foo..bar". 1664 */ 1665 if (i <= 0 || i >= 64) 1666 goto fail; 1667 *cp++ = i; 1668 bcopy(p, cp, i); 1669 cp += i; 1670 p = q; 1671 if (p < name + namelen && *p == '.') 1672 p++; 1673 } 1674 /* termination */ 1675 if (cp + nterm >= ep) 1676 goto fail; 1677 while (nterm-- > 0) 1678 *cp++ = '\0'; 1679 m->m_len = cp - mtod(m, char *); 1680 return m; 1681 } 1682 1683 panic("should not reach here"); 1684 /* NOTREACHED */ 1685 1686 fail: 1687 if (m) 1688 m_freem(m); 1689 return NULL; 1690 } 1691 1692 /* 1693 * check if two DNS-encoded string matches. takes care of truncated 1694 * form (with \0\0 at the end). no compression support. 1695 * XXX upper/lowercase match (see RFC2065) 1696 */ 1697 static int 1698 ni6_dnsmatch(const char *a, int alen, const char *b, int blen) 1699 { 1700 const char *a0, *b0; 1701 int l; 1702 1703 /* simplest case - need validation? */ 1704 if (alen == blen && bcmp(a, b, alen) == 0) 1705 return 1; 1706 1707 a0 = a; 1708 b0 = b; 1709 1710 /* termination is mandatory */ 1711 if (alen < 2 || blen < 2) 1712 return 0; 1713 if (a0[alen - 1] != '\0' || b0[blen - 1] != '\0') 1714 return 0; 1715 alen--; 1716 blen--; 1717 1718 while (a - a0 < alen && b - b0 < blen) { 1719 if (a - a0 + 1 > alen || b - b0 + 1 > blen) 1720 return 0; 1721 1722 if ((signed char)a[0] < 0 || (signed char)b[0] < 0) 1723 return 0; 1724 /* we don't support compression yet */ 1725 if (a[0] >= 64 || b[0] >= 64) 1726 return 0; 1727 1728 /* truncated case */ 1729 if (a[0] == 0 && a - a0 == alen - 1) 1730 return 1; 1731 if (b[0] == 0 && b - b0 == blen - 1) 1732 return 1; 1733 if (a[0] == 0 || b[0] == 0) 1734 return 0; 1735 1736 if (a[0] != b[0]) 1737 return 0; 1738 l = a[0]; 1739 if (a - a0 + 1 + l > alen || b - b0 + 1 + l > blen) 1740 return 0; 1741 if (bcmp(a + 1, b + 1, l) != 0) 1742 return 0; 1743 1744 a += 1 + l; 1745 b += 1 + l; 1746 } 1747 1748 if (a - a0 == alen && b - b0 == blen) 1749 return 1; 1750 else 1751 return 0; 1752 } 1753 1754 /* 1755 * calculate the number of addresses to be returned in the node info reply. 1756 */ 1757 static int 1758 ni6_addrs(struct icmp6_nodeinfo *ni6, struct mbuf *m, struct ifnet **ifpp, 1759 struct in6_addr *subj) 1760 { 1761 struct ifnet *ifp; 1762 struct in6_ifaddr *ifa6; 1763 struct ifaddr *ifa; 1764 int addrs = 0, addrsofif, iffound = 0; 1765 int niflags = ni6->ni_flags; 1766 1767 if ((niflags & NI_NODEADDR_FLAG_ALL) == 0) { 1768 switch (ni6->ni_code) { 1769 case ICMP6_NI_SUBJ_IPV6: 1770 if (subj == NULL) /* must be impossible... */ 1771 return (0); 1772 break; 1773 default: 1774 /* 1775 * XXX: we only support IPv6 subject address for 1776 * this Qtype. 1777 */ 1778 return (0); 1779 } 1780 } 1781 1782 IFNET_RLOCK_NOSLEEP(); 1783 TAILQ_FOREACH(ifp, &V_ifnet, if_list) { 1784 addrsofif = 0; 1785 IF_ADDR_RLOCK(ifp); 1786 TAILQ_FOREACH(ifa, &ifp->if_addrhead, ifa_link) { 1787 if (ifa->ifa_addr->sa_family != AF_INET6) 1788 continue; 1789 ifa6 = (struct in6_ifaddr *)ifa; 1790 1791 if ((niflags & NI_NODEADDR_FLAG_ALL) == 0 && 1792 IN6_ARE_ADDR_EQUAL(subj, &ifa6->ia_addr.sin6_addr)) 1793 iffound = 1; 1794 1795 /* 1796 * IPv4-mapped addresses can only be returned by a 1797 * Node Information proxy, since they represent 1798 * addresses of IPv4-only nodes, which perforce do 1799 * not implement this protocol. 1800 * [icmp-name-lookups-07, Section 5.4] 1801 * So we don't support NI_NODEADDR_FLAG_COMPAT in 1802 * this function at this moment. 1803 */ 1804 1805 /* What do we have to do about ::1? */ 1806 switch (in6_addrscope(&ifa6->ia_addr.sin6_addr)) { 1807 case IPV6_ADDR_SCOPE_LINKLOCAL: 1808 if ((niflags & NI_NODEADDR_FLAG_LINKLOCAL) == 0) 1809 continue; 1810 break; 1811 case IPV6_ADDR_SCOPE_SITELOCAL: 1812 if ((niflags & NI_NODEADDR_FLAG_SITELOCAL) == 0) 1813 continue; 1814 break; 1815 case IPV6_ADDR_SCOPE_GLOBAL: 1816 if ((niflags & NI_NODEADDR_FLAG_GLOBAL) == 0) 1817 continue; 1818 break; 1819 default: 1820 continue; 1821 } 1822 1823 /* 1824 * check if anycast is okay. 1825 * XXX: just experimental. not in the spec. 1826 */ 1827 if ((ifa6->ia6_flags & IN6_IFF_ANYCAST) != 0 && 1828 (niflags & NI_NODEADDR_FLAG_ANYCAST) == 0) 1829 continue; /* we need only unicast addresses */ 1830 if ((ifa6->ia6_flags & IN6_IFF_TEMPORARY) != 0 && 1831 (V_icmp6_nodeinfo & ICMP6_NODEINFO_TMPADDROK) == 0) { 1832 continue; 1833 } 1834 addrsofif++; /* count the address */ 1835 } 1836 IF_ADDR_RUNLOCK(ifp); 1837 if (iffound) { 1838 *ifpp = ifp; 1839 IFNET_RUNLOCK_NOSLEEP(); 1840 return (addrsofif); 1841 } 1842 1843 addrs += addrsofif; 1844 } 1845 IFNET_RUNLOCK_NOSLEEP(); 1846 1847 return (addrs); 1848 } 1849 1850 static int 1851 ni6_store_addrs(struct icmp6_nodeinfo *ni6, struct icmp6_nodeinfo *nni6, 1852 struct ifnet *ifp0, int resid) 1853 { 1854 struct ifnet *ifp; 1855 struct in6_ifaddr *ifa6; 1856 struct ifaddr *ifa; 1857 struct ifnet *ifp_dep = NULL; 1858 int copied = 0, allow_deprecated = 0; 1859 u_char *cp = (u_char *)(nni6 + 1); 1860 int niflags = ni6->ni_flags; 1861 u_int32_t ltime; 1862 1863 if (ifp0 == NULL && !(niflags & NI_NODEADDR_FLAG_ALL)) 1864 return (0); /* needless to copy */ 1865 1866 IFNET_RLOCK_NOSLEEP(); 1867 ifp = ifp0 ? ifp0 : TAILQ_FIRST(&V_ifnet); 1868 again: 1869 1870 for (; ifp; ifp = TAILQ_NEXT(ifp, if_list)) { 1871 IF_ADDR_RLOCK(ifp); 1872 TAILQ_FOREACH(ifa, &ifp->if_addrhead, ifa_link) { 1873 if (ifa->ifa_addr->sa_family != AF_INET6) 1874 continue; 1875 ifa6 = (struct in6_ifaddr *)ifa; 1876 1877 if ((ifa6->ia6_flags & IN6_IFF_DEPRECATED) != 0 && 1878 allow_deprecated == 0) { 1879 /* 1880 * prefererred address should be put before 1881 * deprecated addresses. 1882 */ 1883 1884 /* record the interface for later search */ 1885 if (ifp_dep == NULL) 1886 ifp_dep = ifp; 1887 1888 continue; 1889 } else if ((ifa6->ia6_flags & IN6_IFF_DEPRECATED) == 0 && 1890 allow_deprecated != 0) 1891 continue; /* we now collect deprecated addrs */ 1892 1893 /* What do we have to do about ::1? */ 1894 switch (in6_addrscope(&ifa6->ia_addr.sin6_addr)) { 1895 case IPV6_ADDR_SCOPE_LINKLOCAL: 1896 if ((niflags & NI_NODEADDR_FLAG_LINKLOCAL) == 0) 1897 continue; 1898 break; 1899 case IPV6_ADDR_SCOPE_SITELOCAL: 1900 if ((niflags & NI_NODEADDR_FLAG_SITELOCAL) == 0) 1901 continue; 1902 break; 1903 case IPV6_ADDR_SCOPE_GLOBAL: 1904 if ((niflags & NI_NODEADDR_FLAG_GLOBAL) == 0) 1905 continue; 1906 break; 1907 default: 1908 continue; 1909 } 1910 1911 /* 1912 * check if anycast is okay. 1913 * XXX: just experimental. not in the spec. 1914 */ 1915 if ((ifa6->ia6_flags & IN6_IFF_ANYCAST) != 0 && 1916 (niflags & NI_NODEADDR_FLAG_ANYCAST) == 0) 1917 continue; 1918 if ((ifa6->ia6_flags & IN6_IFF_TEMPORARY) != 0 && 1919 (V_icmp6_nodeinfo & ICMP6_NODEINFO_TMPADDROK) == 0) { 1920 continue; 1921 } 1922 1923 /* now we can copy the address */ 1924 if (resid < sizeof(struct in6_addr) + 1925 sizeof(u_int32_t)) { 1926 IF_ADDR_RUNLOCK(ifp); 1927 /* 1928 * We give up much more copy. 1929 * Set the truncate flag and return. 1930 */ 1931 nni6->ni_flags |= NI_NODEADDR_FLAG_TRUNCATE; 1932 IFNET_RUNLOCK_NOSLEEP(); 1933 return (copied); 1934 } 1935 1936 /* 1937 * Set the TTL of the address. 1938 * The TTL value should be one of the following 1939 * according to the specification: 1940 * 1941 * 1. The remaining lifetime of a DHCP lease on the 1942 * address, or 1943 * 2. The remaining Valid Lifetime of a prefix from 1944 * which the address was derived through Stateless 1945 * Autoconfiguration. 1946 * 1947 * Note that we currently do not support stateful 1948 * address configuration by DHCPv6, so the former 1949 * case can't happen. 1950 */ 1951 if (ifa6->ia6_lifetime.ia6t_expire == 0) 1952 ltime = ND6_INFINITE_LIFETIME; 1953 else { 1954 if (ifa6->ia6_lifetime.ia6t_expire > 1955 time_second) 1956 ltime = htonl(ifa6->ia6_lifetime.ia6t_expire - time_second); 1957 else 1958 ltime = 0; 1959 } 1960 1961 bcopy(<ime, cp, sizeof(u_int32_t)); 1962 cp += sizeof(u_int32_t); 1963 1964 /* copy the address itself */ 1965 bcopy(&ifa6->ia_addr.sin6_addr, cp, 1966 sizeof(struct in6_addr)); 1967 in6_clearscope((struct in6_addr *)cp); /* XXX */ 1968 cp += sizeof(struct in6_addr); 1969 1970 resid -= (sizeof(struct in6_addr) + sizeof(u_int32_t)); 1971 copied += (sizeof(struct in6_addr) + sizeof(u_int32_t)); 1972 } 1973 IF_ADDR_RUNLOCK(ifp); 1974 if (ifp0) /* we need search only on the specified IF */ 1975 break; 1976 } 1977 1978 if (allow_deprecated == 0 && ifp_dep != NULL) { 1979 ifp = ifp_dep; 1980 allow_deprecated = 1; 1981 1982 goto again; 1983 } 1984 1985 IFNET_RUNLOCK_NOSLEEP(); 1986 1987 return (copied); 1988 } 1989 1990 /* 1991 * XXX almost dup'ed code with rip6_input. 1992 */ 1993 static int 1994 icmp6_rip6_input(struct mbuf **mp, int off) 1995 { 1996 struct mbuf *m = *mp; 1997 struct ip6_hdr *ip6 = mtod(m, struct ip6_hdr *); 1998 struct inpcb *in6p; 1999 struct inpcb *last = NULL; 2000 struct sockaddr_in6 fromsa; 2001 struct icmp6_hdr *icmp6; 2002 struct mbuf *opts = NULL; 2003 2004 #ifndef PULLDOWN_TEST 2005 /* this is assumed to be safe. */ 2006 icmp6 = (struct icmp6_hdr *)((caddr_t)ip6 + off); 2007 #else 2008 IP6_EXTHDR_GET(icmp6, struct icmp6_hdr *, m, off, sizeof(*icmp6)); 2009 if (icmp6 == NULL) { 2010 /* m is already reclaimed */ 2011 return (IPPROTO_DONE); 2012 } 2013 #endif 2014 2015 /* 2016 * XXX: the address may have embedded scope zone ID, which should be 2017 * hidden from applications. 2018 */ 2019 bzero(&fromsa, sizeof(fromsa)); 2020 fromsa.sin6_family = AF_INET6; 2021 fromsa.sin6_len = sizeof(struct sockaddr_in6); 2022 fromsa.sin6_addr = ip6->ip6_src; 2023 if (sa6_recoverscope(&fromsa)) { 2024 m_freem(m); 2025 return (IPPROTO_DONE); 2026 } 2027 2028 INP_INFO_RLOCK(&V_ripcbinfo); 2029 LIST_FOREACH(in6p, &V_ripcb, inp_list) { 2030 if ((in6p->inp_vflag & INP_IPV6) == 0) 2031 continue; 2032 if (in6p->inp_ip_p != IPPROTO_ICMPV6) 2033 continue; 2034 if (!IN6_IS_ADDR_UNSPECIFIED(&in6p->in6p_laddr) && 2035 !IN6_ARE_ADDR_EQUAL(&in6p->in6p_laddr, &ip6->ip6_dst)) 2036 continue; 2037 if (!IN6_IS_ADDR_UNSPECIFIED(&in6p->in6p_faddr) && 2038 !IN6_ARE_ADDR_EQUAL(&in6p->in6p_faddr, &ip6->ip6_src)) 2039 continue; 2040 INP_RLOCK(in6p); 2041 if (ICMP6_FILTER_WILLBLOCK(icmp6->icmp6_type, 2042 in6p->in6p_icmp6filt)) { 2043 INP_RUNLOCK(in6p); 2044 continue; 2045 } 2046 if (last != NULL) { 2047 struct mbuf *n = NULL; 2048 2049 /* 2050 * Recent network drivers tend to allocate a single 2051 * mbuf cluster, rather than to make a couple of 2052 * mbufs without clusters. Also, since the IPv6 code 2053 * path tries to avoid m_pullup(), it is highly 2054 * probable that we still have an mbuf cluster here 2055 * even though the necessary length can be stored in an 2056 * mbuf's internal buffer. 2057 * Meanwhile, the default size of the receive socket 2058 * buffer for raw sockets is not so large. This means 2059 * the possibility of packet loss is relatively higher 2060 * than before. To avoid this scenario, we copy the 2061 * received data to a separate mbuf that does not use 2062 * a cluster, if possible. 2063 * XXX: it is better to copy the data after stripping 2064 * intermediate headers. 2065 */ 2066 if ((m->m_flags & M_EXT) && m->m_next == NULL && 2067 m->m_len <= MHLEN) { 2068 MGET(n, M_DONTWAIT, m->m_type); 2069 if (n != NULL) { 2070 if (m_dup_pkthdr(n, m, M_NOWAIT)) { 2071 bcopy(m->m_data, n->m_data, 2072 m->m_len); 2073 n->m_len = m->m_len; 2074 } else { 2075 m_free(n); 2076 n = NULL; 2077 } 2078 } 2079 } 2080 if (n != NULL || 2081 (n = m_copy(m, 0, (int)M_COPYALL)) != NULL) { 2082 if (last->inp_flags & INP_CONTROLOPTS) 2083 ip6_savecontrol(last, n, &opts); 2084 /* strip intermediate headers */ 2085 m_adj(n, off); 2086 SOCKBUF_LOCK(&last->inp_socket->so_rcv); 2087 if (sbappendaddr_locked( 2088 &last->inp_socket->so_rcv, 2089 (struct sockaddr *)&fromsa, n, opts) 2090 == 0) { 2091 /* should notify about lost packet */ 2092 m_freem(n); 2093 if (opts) { 2094 m_freem(opts); 2095 } 2096 SOCKBUF_UNLOCK( 2097 &last->inp_socket->so_rcv); 2098 } else 2099 sorwakeup_locked(last->inp_socket); 2100 opts = NULL; 2101 } 2102 INP_RUNLOCK(last); 2103 } 2104 last = in6p; 2105 } 2106 INP_INFO_RUNLOCK(&V_ripcbinfo); 2107 if (last != NULL) { 2108 if (last->inp_flags & INP_CONTROLOPTS) 2109 ip6_savecontrol(last, m, &opts); 2110 /* strip intermediate headers */ 2111 m_adj(m, off); 2112 2113 /* avoid using mbuf clusters if possible (see above) */ 2114 if ((m->m_flags & M_EXT) && m->m_next == NULL && 2115 m->m_len <= MHLEN) { 2116 struct mbuf *n; 2117 2118 MGET(n, M_DONTWAIT, m->m_type); 2119 if (n != NULL) { 2120 if (m_dup_pkthdr(n, m, M_NOWAIT)) { 2121 bcopy(m->m_data, n->m_data, m->m_len); 2122 n->m_len = m->m_len; 2123 2124 m_freem(m); 2125 m = n; 2126 } else { 2127 m_freem(n); 2128 n = NULL; 2129 } 2130 } 2131 } 2132 SOCKBUF_LOCK(&last->inp_socket->so_rcv); 2133 if (sbappendaddr_locked(&last->inp_socket->so_rcv, 2134 (struct sockaddr *)&fromsa, m, opts) == 0) { 2135 m_freem(m); 2136 if (opts) 2137 m_freem(opts); 2138 SOCKBUF_UNLOCK(&last->inp_socket->so_rcv); 2139 } else 2140 sorwakeup_locked(last->inp_socket); 2141 INP_RUNLOCK(last); 2142 } else { 2143 m_freem(m); 2144 IP6STAT_DEC(ip6s_delivered); 2145 } 2146 return IPPROTO_DONE; 2147 } 2148 2149 /* 2150 * Reflect the ip6 packet back to the source. 2151 * OFF points to the icmp6 header, counted from the top of the mbuf. 2152 */ 2153 void 2154 icmp6_reflect(struct mbuf *m, size_t off) 2155 { 2156 struct ip6_hdr *ip6; 2157 struct icmp6_hdr *icmp6; 2158 struct in6_ifaddr *ia = NULL; 2159 int plen; 2160 int type, code; 2161 struct ifnet *outif = NULL; 2162 struct in6_addr origdst, src, *srcp = NULL; 2163 2164 /* too short to reflect */ 2165 if (off < sizeof(struct ip6_hdr)) { 2166 nd6log((LOG_DEBUG, 2167 "sanity fail: off=%lx, sizeof(ip6)=%lx in %s:%d\n", 2168 (u_long)off, (u_long)sizeof(struct ip6_hdr), 2169 __FILE__, __LINE__)); 2170 goto bad; 2171 } 2172 2173 /* 2174 * If there are extra headers between IPv6 and ICMPv6, strip 2175 * off that header first. 2176 */ 2177 #ifdef DIAGNOSTIC 2178 if (sizeof(struct ip6_hdr) + sizeof(struct icmp6_hdr) > MHLEN) 2179 panic("assumption failed in icmp6_reflect"); 2180 #endif 2181 if (off > sizeof(struct ip6_hdr)) { 2182 size_t l; 2183 struct ip6_hdr nip6; 2184 2185 l = off - sizeof(struct ip6_hdr); 2186 m_copydata(m, 0, sizeof(nip6), (caddr_t)&nip6); 2187 m_adj(m, l); 2188 l = sizeof(struct ip6_hdr) + sizeof(struct icmp6_hdr); 2189 if (m->m_len < l) { 2190 if ((m = m_pullup(m, l)) == NULL) 2191 return; 2192 } 2193 bcopy((caddr_t)&nip6, mtod(m, caddr_t), sizeof(nip6)); 2194 } else /* off == sizeof(struct ip6_hdr) */ { 2195 size_t l; 2196 l = sizeof(struct ip6_hdr) + sizeof(struct icmp6_hdr); 2197 if (m->m_len < l) { 2198 if ((m = m_pullup(m, l)) == NULL) 2199 return; 2200 } 2201 } 2202 plen = m->m_pkthdr.len - sizeof(struct ip6_hdr); 2203 ip6 = mtod(m, struct ip6_hdr *); 2204 ip6->ip6_nxt = IPPROTO_ICMPV6; 2205 icmp6 = (struct icmp6_hdr *)(ip6 + 1); 2206 type = icmp6->icmp6_type; /* keep type for statistics */ 2207 code = icmp6->icmp6_code; /* ditto. */ 2208 2209 origdst = ip6->ip6_dst; 2210 /* 2211 * ip6_input() drops a packet if its src is multicast. 2212 * So, the src is never multicast. 2213 */ 2214 ip6->ip6_dst = ip6->ip6_src; 2215 2216 /* 2217 * If the incoming packet was addressed directly to us (i.e. unicast), 2218 * use dst as the src for the reply. 2219 * The IN6_IFF_NOTREADY case should be VERY rare, but is possible 2220 * (for example) when we encounter an error while forwarding procedure 2221 * destined to a duplicated address of ours. 2222 * Note that ip6_getdstifaddr() may fail if we are in an error handling 2223 * procedure of an outgoing packet of our own, in which case we need 2224 * to search in the ifaddr list. 2225 */ 2226 if (!IN6_IS_ADDR_MULTICAST(&origdst)) { 2227 if ((ia = ip6_getdstifaddr(m))) { 2228 if (!(ia->ia6_flags & 2229 (IN6_IFF_ANYCAST|IN6_IFF_NOTREADY))) 2230 srcp = &ia->ia_addr.sin6_addr; 2231 } else { 2232 struct sockaddr_in6 d; 2233 2234 bzero(&d, sizeof(d)); 2235 d.sin6_family = AF_INET6; 2236 d.sin6_len = sizeof(d); 2237 d.sin6_addr = origdst; 2238 ia = (struct in6_ifaddr *) 2239 ifa_ifwithaddr((struct sockaddr *)&d); 2240 if (ia && 2241 !(ia->ia6_flags & 2242 (IN6_IFF_ANYCAST|IN6_IFF_NOTREADY))) { 2243 srcp = &ia->ia_addr.sin6_addr; 2244 } 2245 } 2246 } 2247 2248 if (srcp == NULL) { 2249 int e; 2250 struct sockaddr_in6 sin6; 2251 struct route_in6 ro; 2252 2253 /* 2254 * This case matches to multicasts, our anycast, or unicasts 2255 * that we do not own. Select a source address based on the 2256 * source address of the erroneous packet. 2257 */ 2258 bzero(&sin6, sizeof(sin6)); 2259 sin6.sin6_family = AF_INET6; 2260 sin6.sin6_len = sizeof(sin6); 2261 sin6.sin6_addr = ip6->ip6_dst; /* zone ID should be embedded */ 2262 2263 bzero(&ro, sizeof(ro)); 2264 e = in6_selectsrc(&sin6, NULL, NULL, &ro, NULL, &outif, &src); 2265 if (ro.ro_rt) 2266 RTFREE(ro.ro_rt); /* XXX: we could use this */ 2267 if (e) { 2268 char ip6buf[INET6_ADDRSTRLEN]; 2269 nd6log((LOG_DEBUG, 2270 "icmp6_reflect: source can't be determined: " 2271 "dst=%s, error=%d\n", 2272 ip6_sprintf(ip6buf, &sin6.sin6_addr), e)); 2273 goto bad; 2274 } 2275 srcp = &src; 2276 } 2277 2278 ip6->ip6_src = *srcp; 2279 ip6->ip6_flow = 0; 2280 ip6->ip6_vfc &= ~IPV6_VERSION_MASK; 2281 ip6->ip6_vfc |= IPV6_VERSION; 2282 ip6->ip6_nxt = IPPROTO_ICMPV6; 2283 if (outif) 2284 ip6->ip6_hlim = ND_IFINFO(outif)->chlim; 2285 else if (m->m_pkthdr.rcvif) { 2286 /* XXX: This may not be the outgoing interface */ 2287 ip6->ip6_hlim = ND_IFINFO(m->m_pkthdr.rcvif)->chlim; 2288 } else 2289 ip6->ip6_hlim = V_ip6_defhlim; 2290 2291 icmp6->icmp6_cksum = 0; 2292 icmp6->icmp6_cksum = in6_cksum(m, IPPROTO_ICMPV6, 2293 sizeof(struct ip6_hdr), plen); 2294 2295 /* 2296 * XXX option handling 2297 */ 2298 2299 m->m_flags &= ~(M_BCAST|M_MCAST); 2300 2301 m_addr_changed(m); 2302 2303 ip6_output(m, NULL, NULL, 0, NULL, &outif, NULL); 2304 if (outif) 2305 icmp6_ifoutstat_inc(outif, type, code); 2306 2307 if (ia != NULL) 2308 ifa_free(&ia->ia_ifa); 2309 return; 2310 2311 bad: 2312 if (ia != NULL) 2313 ifa_free(&ia->ia_ifa); 2314 m_freem(m); 2315 return; 2316 } 2317 2318 void 2319 icmp6_fasttimo(void) 2320 { 2321 2322 mld_fasttimo(); 2323 } 2324 2325 void 2326 icmp6_slowtimo(void) 2327 { 2328 2329 mld_slowtimo(); 2330 } 2331 2332 static const char * 2333 icmp6_redirect_diag(struct in6_addr *src6, struct in6_addr *dst6, 2334 struct in6_addr *tgt6) 2335 { 2336 static char buf[1024]; 2337 char ip6bufs[INET6_ADDRSTRLEN]; 2338 char ip6bufd[INET6_ADDRSTRLEN]; 2339 char ip6buft[INET6_ADDRSTRLEN]; 2340 snprintf(buf, sizeof(buf), "(src=%s dst=%s tgt=%s)", 2341 ip6_sprintf(ip6bufs, src6), ip6_sprintf(ip6bufd, dst6), 2342 ip6_sprintf(ip6buft, tgt6)); 2343 return buf; 2344 } 2345 2346 void 2347 icmp6_redirect_input(struct mbuf *m, int off) 2348 { 2349 struct ifnet *ifp; 2350 struct ip6_hdr *ip6 = mtod(m, struct ip6_hdr *); 2351 struct nd_redirect *nd_rd; 2352 int icmp6len = ntohs(ip6->ip6_plen); 2353 char *lladdr = NULL; 2354 int lladdrlen = 0; 2355 struct rtentry *rt = NULL; 2356 int is_router; 2357 int is_onlink; 2358 struct in6_addr src6 = ip6->ip6_src; 2359 struct in6_addr redtgt6; 2360 struct in6_addr reddst6; 2361 union nd_opts ndopts; 2362 char ip6buf[INET6_ADDRSTRLEN]; 2363 2364 M_ASSERTPKTHDR(m); 2365 KASSERT(m->m_pkthdr.rcvif != NULL, ("%s: no rcvif", __func__)); 2366 2367 ifp = m->m_pkthdr.rcvif; 2368 2369 /* XXX if we are router, we don't update route by icmp6 redirect */ 2370 if (V_ip6_forwarding) 2371 goto freeit; 2372 if (!V_icmp6_rediraccept) 2373 goto freeit; 2374 2375 #ifndef PULLDOWN_TEST 2376 IP6_EXTHDR_CHECK(m, off, icmp6len,); 2377 nd_rd = (struct nd_redirect *)((caddr_t)ip6 + off); 2378 #else 2379 IP6_EXTHDR_GET(nd_rd, struct nd_redirect *, m, off, icmp6len); 2380 if (nd_rd == NULL) { 2381 ICMP6STAT_INC(icp6s_tooshort); 2382 return; 2383 } 2384 #endif 2385 redtgt6 = nd_rd->nd_rd_target; 2386 reddst6 = nd_rd->nd_rd_dst; 2387 2388 if (in6_setscope(&redtgt6, m->m_pkthdr.rcvif, NULL) || 2389 in6_setscope(&reddst6, m->m_pkthdr.rcvif, NULL)) { 2390 goto freeit; 2391 } 2392 2393 /* validation */ 2394 if (!IN6_IS_ADDR_LINKLOCAL(&src6)) { 2395 nd6log((LOG_ERR, 2396 "ICMP6 redirect sent from %s rejected; " 2397 "must be from linklocal\n", 2398 ip6_sprintf(ip6buf, &src6))); 2399 goto bad; 2400 } 2401 if (ip6->ip6_hlim != 255) { 2402 nd6log((LOG_ERR, 2403 "ICMP6 redirect sent from %s rejected; " 2404 "hlim=%d (must be 255)\n", 2405 ip6_sprintf(ip6buf, &src6), ip6->ip6_hlim)); 2406 goto bad; 2407 } 2408 { 2409 /* ip6->ip6_src must be equal to gw for icmp6->icmp6_reddst */ 2410 struct sockaddr_in6 sin6; 2411 struct in6_addr *gw6; 2412 2413 bzero(&sin6, sizeof(sin6)); 2414 sin6.sin6_family = AF_INET6; 2415 sin6.sin6_len = sizeof(struct sockaddr_in6); 2416 bcopy(&reddst6, &sin6.sin6_addr, sizeof(reddst6)); 2417 rt = in6_rtalloc1((struct sockaddr *)&sin6, 0, 0UL, RT_DEFAULT_FIB); 2418 if (rt) { 2419 if (rt->rt_gateway == NULL || 2420 rt->rt_gateway->sa_family != AF_INET6) { 2421 nd6log((LOG_ERR, 2422 "ICMP6 redirect rejected; no route " 2423 "with inet6 gateway found for redirect dst: %s\n", 2424 icmp6_redirect_diag(&src6, &reddst6, &redtgt6))); 2425 RTFREE_LOCKED(rt); 2426 goto bad; 2427 } 2428 2429 gw6 = &(((struct sockaddr_in6 *)rt->rt_gateway)->sin6_addr); 2430 if (bcmp(&src6, gw6, sizeof(struct in6_addr)) != 0) { 2431 nd6log((LOG_ERR, 2432 "ICMP6 redirect rejected; " 2433 "not equal to gw-for-src=%s (must be same): " 2434 "%s\n", 2435 ip6_sprintf(ip6buf, gw6), 2436 icmp6_redirect_diag(&src6, &reddst6, &redtgt6))); 2437 RTFREE_LOCKED(rt); 2438 goto bad; 2439 } 2440 } else { 2441 nd6log((LOG_ERR, 2442 "ICMP6 redirect rejected; " 2443 "no route found for redirect dst: %s\n", 2444 icmp6_redirect_diag(&src6, &reddst6, &redtgt6))); 2445 goto bad; 2446 } 2447 RTFREE_LOCKED(rt); 2448 rt = NULL; 2449 } 2450 if (IN6_IS_ADDR_MULTICAST(&reddst6)) { 2451 nd6log((LOG_ERR, 2452 "ICMP6 redirect rejected; " 2453 "redirect dst must be unicast: %s\n", 2454 icmp6_redirect_diag(&src6, &reddst6, &redtgt6))); 2455 goto bad; 2456 } 2457 2458 is_router = is_onlink = 0; 2459 if (IN6_IS_ADDR_LINKLOCAL(&redtgt6)) 2460 is_router = 1; /* router case */ 2461 if (bcmp(&redtgt6, &reddst6, sizeof(redtgt6)) == 0) 2462 is_onlink = 1; /* on-link destination case */ 2463 if (!is_router && !is_onlink) { 2464 nd6log((LOG_ERR, 2465 "ICMP6 redirect rejected; " 2466 "neither router case nor onlink case: %s\n", 2467 icmp6_redirect_diag(&src6, &reddst6, &redtgt6))); 2468 goto bad; 2469 } 2470 /* validation passed */ 2471 2472 icmp6len -= sizeof(*nd_rd); 2473 nd6_option_init(nd_rd + 1, icmp6len, &ndopts); 2474 if (nd6_options(&ndopts) < 0) { 2475 nd6log((LOG_INFO, "%s: invalid ND option, rejected: %s\n", 2476 __func__, icmp6_redirect_diag(&src6, &reddst6, &redtgt6))); 2477 /* nd6_options have incremented stats */ 2478 goto freeit; 2479 } 2480 2481 if (ndopts.nd_opts_tgt_lladdr) { 2482 lladdr = (char *)(ndopts.nd_opts_tgt_lladdr + 1); 2483 lladdrlen = ndopts.nd_opts_tgt_lladdr->nd_opt_len << 3; 2484 } 2485 2486 if (lladdr && ((ifp->if_addrlen + 2 + 7) & ~7) != lladdrlen) { 2487 nd6log((LOG_INFO, "%s: lladdrlen mismatch for %s " 2488 "(if %d, icmp6 packet %d): %s\n", 2489 __func__, ip6_sprintf(ip6buf, &redtgt6), 2490 ifp->if_addrlen, lladdrlen - 2, 2491 icmp6_redirect_diag(&src6, &reddst6, &redtgt6))); 2492 goto bad; 2493 } 2494 2495 /* RFC 2461 8.3 */ 2496 nd6_cache_lladdr(ifp, &redtgt6, lladdr, lladdrlen, ND_REDIRECT, 2497 is_onlink ? ND_REDIRECT_ONLINK : ND_REDIRECT_ROUTER); 2498 2499 if (!is_onlink) { /* better router case. perform rtredirect. */ 2500 /* perform rtredirect */ 2501 struct sockaddr_in6 sdst; 2502 struct sockaddr_in6 sgw; 2503 struct sockaddr_in6 ssrc; 2504 u_int fibnum; 2505 2506 bzero(&sdst, sizeof(sdst)); 2507 bzero(&sgw, sizeof(sgw)); 2508 bzero(&ssrc, sizeof(ssrc)); 2509 sdst.sin6_family = sgw.sin6_family = ssrc.sin6_family = AF_INET6; 2510 sdst.sin6_len = sgw.sin6_len = ssrc.sin6_len = 2511 sizeof(struct sockaddr_in6); 2512 bcopy(&redtgt6, &sgw.sin6_addr, sizeof(struct in6_addr)); 2513 bcopy(&reddst6, &sdst.sin6_addr, sizeof(struct in6_addr)); 2514 bcopy(&src6, &ssrc.sin6_addr, sizeof(struct in6_addr)); 2515 for (fibnum = 0; fibnum < rt_numfibs; fibnum++) 2516 in6_rtredirect((struct sockaddr *)&sdst, 2517 (struct sockaddr *)&sgw, (struct sockaddr *)NULL, 2518 RTF_GATEWAY | RTF_HOST, (struct sockaddr *)&ssrc, 2519 fibnum); 2520 } 2521 /* finally update cached route in each socket via pfctlinput */ 2522 { 2523 struct sockaddr_in6 sdst; 2524 2525 bzero(&sdst, sizeof(sdst)); 2526 sdst.sin6_family = AF_INET6; 2527 sdst.sin6_len = sizeof(struct sockaddr_in6); 2528 bcopy(&reddst6, &sdst.sin6_addr, sizeof(struct in6_addr)); 2529 pfctlinput(PRC_REDIRECT_HOST, (struct sockaddr *)&sdst); 2530 #ifdef IPSEC 2531 key_sa_routechange((struct sockaddr *)&sdst); 2532 #endif /* IPSEC */ 2533 } 2534 2535 freeit: 2536 m_freem(m); 2537 return; 2538 2539 bad: 2540 ICMP6STAT_INC(icp6s_badredirect); 2541 m_freem(m); 2542 } 2543 2544 void 2545 icmp6_redirect_output(struct mbuf *m0, struct rtentry *rt) 2546 { 2547 struct ifnet *ifp; /* my outgoing interface */ 2548 struct in6_addr *ifp_ll6; 2549 struct in6_addr *router_ll6; 2550 struct ip6_hdr *sip6; /* m0 as struct ip6_hdr */ 2551 struct mbuf *m = NULL; /* newly allocated one */ 2552 struct m_tag *mtag; 2553 struct ip6_hdr *ip6; /* m as struct ip6_hdr */ 2554 struct nd_redirect *nd_rd; 2555 struct llentry *ln = NULL; 2556 size_t maxlen; 2557 u_char *p; 2558 struct ifnet *outif = NULL; 2559 struct sockaddr_in6 src_sa; 2560 2561 icmp6_errcount(&V_icmp6stat.icp6s_outerrhist, ND_REDIRECT, 0); 2562 2563 /* if we are not router, we don't send icmp6 redirect */ 2564 if (!V_ip6_forwarding) 2565 goto fail; 2566 2567 /* sanity check */ 2568 if (!m0 || !rt || !(rt->rt_flags & RTF_UP) || !(ifp = rt->rt_ifp)) 2569 goto fail; 2570 2571 /* 2572 * Address check: 2573 * the source address must identify a neighbor, and 2574 * the destination address must not be a multicast address 2575 * [RFC 2461, sec 8.2] 2576 */ 2577 sip6 = mtod(m0, struct ip6_hdr *); 2578 bzero(&src_sa, sizeof(src_sa)); 2579 src_sa.sin6_family = AF_INET6; 2580 src_sa.sin6_len = sizeof(src_sa); 2581 src_sa.sin6_addr = sip6->ip6_src; 2582 if (nd6_is_addr_neighbor(&src_sa, ifp) == 0) 2583 goto fail; 2584 if (IN6_IS_ADDR_MULTICAST(&sip6->ip6_dst)) 2585 goto fail; /* what should we do here? */ 2586 2587 /* rate limit */ 2588 if (icmp6_ratelimit(&sip6->ip6_src, ND_REDIRECT, 0)) 2589 goto fail; 2590 2591 /* 2592 * Since we are going to append up to 1280 bytes (= IPV6_MMTU), 2593 * we almost always ask for an mbuf cluster for simplicity. 2594 * (MHLEN < IPV6_MMTU is almost always true) 2595 */ 2596 #if IPV6_MMTU >= MCLBYTES 2597 # error assumption failed about IPV6_MMTU and MCLBYTES 2598 #endif 2599 MGETHDR(m, M_DONTWAIT, MT_HEADER); 2600 if (m && IPV6_MMTU >= MHLEN) 2601 MCLGET(m, M_DONTWAIT); 2602 if (!m) 2603 goto fail; 2604 M_SETFIB(m, rt->rt_fibnum); 2605 m->m_pkthdr.rcvif = NULL; 2606 m->m_len = 0; 2607 maxlen = M_TRAILINGSPACE(m); 2608 maxlen = min(IPV6_MMTU, maxlen); 2609 /* just for safety */ 2610 if (maxlen < sizeof(struct ip6_hdr) + sizeof(struct icmp6_hdr) + 2611 ((sizeof(struct nd_opt_hdr) + ifp->if_addrlen + 7) & ~7)) { 2612 goto fail; 2613 } 2614 2615 { 2616 /* get ip6 linklocal address for ifp(my outgoing interface). */ 2617 struct in6_ifaddr *ia; 2618 if ((ia = in6ifa_ifpforlinklocal(ifp, 2619 IN6_IFF_NOTREADY| 2620 IN6_IFF_ANYCAST)) == NULL) 2621 goto fail; 2622 ifp_ll6 = &ia->ia_addr.sin6_addr; 2623 /* XXXRW: reference released prematurely. */ 2624 ifa_free(&ia->ia_ifa); 2625 } 2626 2627 /* get ip6 linklocal address for the router. */ 2628 if (rt->rt_gateway && (rt->rt_flags & RTF_GATEWAY)) { 2629 struct sockaddr_in6 *sin6; 2630 sin6 = (struct sockaddr_in6 *)rt->rt_gateway; 2631 router_ll6 = &sin6->sin6_addr; 2632 if (!IN6_IS_ADDR_LINKLOCAL(router_ll6)) 2633 router_ll6 = (struct in6_addr *)NULL; 2634 } else 2635 router_ll6 = (struct in6_addr *)NULL; 2636 2637 /* ip6 */ 2638 ip6 = mtod(m, struct ip6_hdr *); 2639 ip6->ip6_flow = 0; 2640 ip6->ip6_vfc &= ~IPV6_VERSION_MASK; 2641 ip6->ip6_vfc |= IPV6_VERSION; 2642 /* ip6->ip6_plen will be set later */ 2643 ip6->ip6_nxt = IPPROTO_ICMPV6; 2644 ip6->ip6_hlim = 255; 2645 /* ip6->ip6_src must be linklocal addr for my outgoing if. */ 2646 bcopy(ifp_ll6, &ip6->ip6_src, sizeof(struct in6_addr)); 2647 bcopy(&sip6->ip6_src, &ip6->ip6_dst, sizeof(struct in6_addr)); 2648 2649 /* ND Redirect */ 2650 nd_rd = (struct nd_redirect *)(ip6 + 1); 2651 nd_rd->nd_rd_type = ND_REDIRECT; 2652 nd_rd->nd_rd_code = 0; 2653 nd_rd->nd_rd_reserved = 0; 2654 if (rt->rt_flags & RTF_GATEWAY) { 2655 /* 2656 * nd_rd->nd_rd_target must be a link-local address in 2657 * better router cases. 2658 */ 2659 if (!router_ll6) 2660 goto fail; 2661 bcopy(router_ll6, &nd_rd->nd_rd_target, 2662 sizeof(nd_rd->nd_rd_target)); 2663 bcopy(&sip6->ip6_dst, &nd_rd->nd_rd_dst, 2664 sizeof(nd_rd->nd_rd_dst)); 2665 } else { 2666 /* make sure redtgt == reddst */ 2667 bcopy(&sip6->ip6_dst, &nd_rd->nd_rd_target, 2668 sizeof(nd_rd->nd_rd_target)); 2669 bcopy(&sip6->ip6_dst, &nd_rd->nd_rd_dst, 2670 sizeof(nd_rd->nd_rd_dst)); 2671 } 2672 2673 p = (u_char *)(nd_rd + 1); 2674 2675 if (!router_ll6) 2676 goto nolladdropt; 2677 2678 { 2679 /* target lladdr option */ 2680 int len; 2681 struct nd_opt_hdr *nd_opt; 2682 char *lladdr; 2683 2684 IF_AFDATA_LOCK(ifp); 2685 ln = nd6_lookup(router_ll6, 0, ifp); 2686 IF_AFDATA_UNLOCK(ifp); 2687 if (ln == NULL) 2688 goto nolladdropt; 2689 2690 len = sizeof(*nd_opt) + ifp->if_addrlen; 2691 len = (len + 7) & ~7; /* round by 8 */ 2692 /* safety check */ 2693 if (len + (p - (u_char *)ip6) > maxlen) 2694 goto nolladdropt; 2695 2696 if (ln->la_flags & LLE_VALID) { 2697 nd_opt = (struct nd_opt_hdr *)p; 2698 nd_opt->nd_opt_type = ND_OPT_TARGET_LINKADDR; 2699 nd_opt->nd_opt_len = len >> 3; 2700 lladdr = (char *)(nd_opt + 1); 2701 bcopy(&ln->ll_addr, lladdr, ifp->if_addrlen); 2702 p += len; 2703 } 2704 } 2705 nolladdropt: 2706 if (ln != NULL) 2707 LLE_RUNLOCK(ln); 2708 2709 m->m_pkthdr.len = m->m_len = p - (u_char *)ip6; 2710 2711 /* just to be safe */ 2712 #ifdef M_DECRYPTED /*not openbsd*/ 2713 if (m0->m_flags & M_DECRYPTED) 2714 goto noredhdropt; 2715 #endif 2716 if (p - (u_char *)ip6 > maxlen) 2717 goto noredhdropt; 2718 2719 { 2720 /* redirected header option */ 2721 int len; 2722 struct nd_opt_rd_hdr *nd_opt_rh; 2723 2724 /* 2725 * compute the maximum size for icmp6 redirect header option. 2726 * XXX room for auth header? 2727 */ 2728 len = maxlen - (p - (u_char *)ip6); 2729 len &= ~7; 2730 2731 /* This is just for simplicity. */ 2732 if (m0->m_pkthdr.len != m0->m_len) { 2733 if (m0->m_next) { 2734 m_freem(m0->m_next); 2735 m0->m_next = NULL; 2736 } 2737 m0->m_pkthdr.len = m0->m_len; 2738 } 2739 2740 /* 2741 * Redirected header option spec (RFC2461 4.6.3) talks nothing 2742 * about padding/truncate rule for the original IP packet. 2743 * From the discussion on IPv6imp in Feb 1999, 2744 * the consensus was: 2745 * - "attach as much as possible" is the goal 2746 * - pad if not aligned (original size can be guessed by 2747 * original ip6 header) 2748 * Following code adds the padding if it is simple enough, 2749 * and truncates if not. 2750 */ 2751 if (m0->m_next || m0->m_pkthdr.len != m0->m_len) 2752 panic("assumption failed in %s:%d", __FILE__, 2753 __LINE__); 2754 2755 if (len - sizeof(*nd_opt_rh) < m0->m_pkthdr.len) { 2756 /* not enough room, truncate */ 2757 m0->m_pkthdr.len = m0->m_len = len - 2758 sizeof(*nd_opt_rh); 2759 } else { 2760 /* enough room, pad or truncate */ 2761 size_t extra; 2762 2763 extra = m0->m_pkthdr.len % 8; 2764 if (extra) { 2765 /* pad if easy enough, truncate if not */ 2766 if (8 - extra <= M_TRAILINGSPACE(m0)) { 2767 /* pad */ 2768 m0->m_len += (8 - extra); 2769 m0->m_pkthdr.len += (8 - extra); 2770 } else { 2771 /* truncate */ 2772 m0->m_pkthdr.len -= extra; 2773 m0->m_len -= extra; 2774 } 2775 } 2776 len = m0->m_pkthdr.len + sizeof(*nd_opt_rh); 2777 m0->m_pkthdr.len = m0->m_len = len - 2778 sizeof(*nd_opt_rh); 2779 } 2780 2781 nd_opt_rh = (struct nd_opt_rd_hdr *)p; 2782 bzero(nd_opt_rh, sizeof(*nd_opt_rh)); 2783 nd_opt_rh->nd_opt_rh_type = ND_OPT_REDIRECTED_HEADER; 2784 nd_opt_rh->nd_opt_rh_len = len >> 3; 2785 p += sizeof(*nd_opt_rh); 2786 m->m_pkthdr.len = m->m_len = p - (u_char *)ip6; 2787 2788 /* connect m0 to m */ 2789 m_tag_delete_chain(m0, NULL); 2790 m0->m_flags &= ~M_PKTHDR; 2791 m->m_next = m0; 2792 m->m_pkthdr.len = m->m_len + m0->m_len; 2793 m0 = NULL; 2794 } 2795 noredhdropt:; 2796 if (m0) { 2797 m_freem(m0); 2798 m0 = NULL; 2799 } 2800 2801 /* XXX: clear embedded link IDs in the inner header */ 2802 in6_clearscope(&sip6->ip6_src); 2803 in6_clearscope(&sip6->ip6_dst); 2804 in6_clearscope(&nd_rd->nd_rd_target); 2805 in6_clearscope(&nd_rd->nd_rd_dst); 2806 2807 ip6->ip6_plen = htons(m->m_pkthdr.len - sizeof(struct ip6_hdr)); 2808 2809 nd_rd->nd_rd_cksum = 0; 2810 nd_rd->nd_rd_cksum = in6_cksum(m, IPPROTO_ICMPV6, 2811 sizeof(*ip6), ntohs(ip6->ip6_plen)); 2812 2813 if (send_sendso_input_hook != NULL) { 2814 mtag = m_tag_get(PACKET_TAG_ND_OUTGOING, sizeof(unsigned short), 2815 M_NOWAIT); 2816 if (mtag == NULL) 2817 goto fail; 2818 *(unsigned short *)(mtag + 1) = nd_rd->nd_rd_type; 2819 m_tag_prepend(m, mtag); 2820 } 2821 2822 /* send the packet to outside... */ 2823 ip6_output(m, NULL, NULL, 0, NULL, &outif, NULL); 2824 if (outif) { 2825 icmp6_ifstat_inc(outif, ifs6_out_msg); 2826 icmp6_ifstat_inc(outif, ifs6_out_redirect); 2827 } 2828 ICMP6STAT_INC(icp6s_outhist[ND_REDIRECT]); 2829 2830 return; 2831 2832 fail: 2833 if (m) 2834 m_freem(m); 2835 if (m0) 2836 m_freem(m0); 2837 } 2838 2839 /* 2840 * ICMPv6 socket option processing. 2841 */ 2842 int 2843 icmp6_ctloutput(struct socket *so, struct sockopt *sopt) 2844 { 2845 int error = 0; 2846 int optlen; 2847 struct inpcb *inp = sotoinpcb(so); 2848 int level, op, optname; 2849 2850 if (sopt) { 2851 level = sopt->sopt_level; 2852 op = sopt->sopt_dir; 2853 optname = sopt->sopt_name; 2854 optlen = sopt->sopt_valsize; 2855 } else 2856 level = op = optname = optlen = 0; 2857 2858 if (level != IPPROTO_ICMPV6) { 2859 return EINVAL; 2860 } 2861 2862 switch (op) { 2863 case PRCO_SETOPT: 2864 switch (optname) { 2865 case ICMP6_FILTER: 2866 { 2867 struct icmp6_filter ic6f; 2868 2869 if (optlen != sizeof(ic6f)) { 2870 error = EMSGSIZE; 2871 break; 2872 } 2873 error = sooptcopyin(sopt, &ic6f, optlen, optlen); 2874 if (error == 0) { 2875 INP_WLOCK(inp); 2876 *inp->in6p_icmp6filt = ic6f; 2877 INP_WUNLOCK(inp); 2878 } 2879 break; 2880 } 2881 2882 default: 2883 error = ENOPROTOOPT; 2884 break; 2885 } 2886 break; 2887 2888 case PRCO_GETOPT: 2889 switch (optname) { 2890 case ICMP6_FILTER: 2891 { 2892 struct icmp6_filter ic6f; 2893 2894 INP_RLOCK(inp); 2895 ic6f = *inp->in6p_icmp6filt; 2896 INP_RUNLOCK(inp); 2897 error = sooptcopyout(sopt, &ic6f, sizeof(ic6f)); 2898 break; 2899 } 2900 2901 default: 2902 error = ENOPROTOOPT; 2903 break; 2904 } 2905 break; 2906 } 2907 2908 return (error); 2909 } 2910 2911 /* 2912 * Perform rate limit check. 2913 * Returns 0 if it is okay to send the icmp6 packet. 2914 * Returns 1 if the router SHOULD NOT send this icmp6 packet due to rate 2915 * limitation. 2916 * 2917 * XXX per-destination/type check necessary? 2918 * 2919 * dst - not used at this moment 2920 * type - not used at this moment 2921 * code - not used at this moment 2922 */ 2923 static int 2924 icmp6_ratelimit(const struct in6_addr *dst, const int type, 2925 const int code) 2926 { 2927 int ret; 2928 2929 ret = 0; /* okay to send */ 2930 2931 /* PPS limit */ 2932 if (!ppsratecheck(&V_icmp6errppslim_last, &V_icmp6errpps_count, 2933 V_icmp6errppslim)) { 2934 /* The packet is subject to rate limit */ 2935 ret++; 2936 } 2937 2938 return ret; 2939 } 2940