1 /* $FreeBSD$ */ 2 /* $KAME: ip6_mroute.c,v 1.58 2001/12/18 02:36:31 itojun Exp $ */ 3 4 /*- 5 * Copyright (C) 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 /* BSDI ip_mroute.c,v 2.10 1996/11/14 00:29:52 jch Exp */ 34 35 /*- 36 * Copyright (c) 1989 Stephen Deering 37 * Copyright (c) 1992, 1993 38 * The Regents of the University of California. All rights reserved. 39 * 40 * This code is derived from software contributed to Berkeley by 41 * Stephen Deering of Stanford University. 42 * 43 * Redistribution and use in source and binary forms, with or without 44 * modification, are permitted provided that the following conditions 45 * are met: 46 * 1. Redistributions of source code must retain the above copyright 47 * notice, this list of conditions and the following disclaimer. 48 * 2. Redistributions in binary form must reproduce the above copyright 49 * notice, this list of conditions and the following disclaimer in the 50 * documentation and/or other materials provided with the distribution. 51 * 4. Neither the name of the University nor the names of its contributors 52 * may be used to endorse or promote products derived from this software 53 * without specific prior written permission. 54 * 55 * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND 56 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE 57 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE 58 * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE 59 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL 60 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS 61 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) 62 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT 63 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY 64 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF 65 * SUCH DAMAGE. 66 * 67 * @(#)ip_mroute.c 8.2 (Berkeley) 11/15/93 68 */ 69 70 /* 71 * IP multicast forwarding procedures 72 * 73 * Written by David Waitzman, BBN Labs, August 1988. 74 * Modified by Steve Deering, Stanford, February 1989. 75 * Modified by Mark J. Steiglitz, Stanford, May, 1991 76 * Modified by Van Jacobson, LBL, January 1993 77 * Modified by Ajit Thyagarajan, PARC, August 1993 78 * Modified by Bill Fenner, PARC, April 1994 79 * 80 * MROUTING Revision: 3.5.1.2 + PIM-SMv2 (pimd) Support 81 */ 82 83 #include "opt_inet.h" 84 #include "opt_inet6.h" 85 86 #include <sys/param.h> 87 #include <sys/callout.h> 88 #include <sys/errno.h> 89 #include <sys/kernel.h> 90 #include <sys/lock.h> 91 #include <sys/malloc.h> 92 #include <sys/mbuf.h> 93 #include <sys/protosw.h> 94 #include <sys/signalvar.h> 95 #include <sys/socket.h> 96 #include <sys/socketvar.h> 97 #include <sys/sockio.h> 98 #include <sys/sx.h> 99 #include <sys/syslog.h> 100 #include <sys/systm.h> 101 #include <sys/time.h> 102 103 #include <net/if.h> 104 #include <net/if_types.h> 105 #include <net/raw_cb.h> 106 #include <net/route.h> 107 108 #include <netinet/in.h> 109 #include <netinet/in_var.h> 110 #include <netinet/icmp6.h> 111 112 #include <netinet/ip6.h> 113 #include <netinet6/ip6_var.h> 114 #include <netinet6/scope6_var.h> 115 #include <netinet6/nd6.h> 116 #include <netinet6/ip6_mroute.h> 117 #include <netinet6/pim6.h> 118 #include <netinet6/pim6_var.h> 119 120 static MALLOC_DEFINE(M_MRTABLE6, "mf6c", "multicast forwarding cache entry"); 121 122 #define M_HASCL(m) ((m)->m_flags & M_EXT) 123 124 static int ip6_mdq __P((struct mbuf *, struct ifnet *, struct mf6c *)); 125 static void phyint_send __P((struct ip6_hdr *, struct mif6 *, struct mbuf *)); 126 127 static int set_pim6 __P((int *)); 128 static int socket_send __P((struct socket *, struct mbuf *, 129 struct sockaddr_in6 *)); 130 static int register_send __P((struct ip6_hdr *, struct mif6 *, 131 struct mbuf *)); 132 133 /* 134 * Globals. All but ip6_mrouter, ip6_mrtproto and mrt6stat could be static, 135 * except for netstat or debugging purposes. 136 */ 137 struct socket *ip6_mrouter = NULL; 138 int ip6_mrouter_ver = 0; 139 int ip6_mrtproto = IPPROTO_PIM; /* for netstat only */ 140 struct mrt6stat mrt6stat; 141 142 #define NO_RTE_FOUND 0x1 143 #define RTE_FOUND 0x2 144 145 struct mf6c *mf6ctable[MF6CTBLSIZ]; 146 u_char n6expire[MF6CTBLSIZ]; 147 static struct mif6 mif6table[MAXMIFS]; 148 #ifdef MRT6DEBUG 149 u_int mrt6debug = 0; /* debug level */ 150 #define DEBUG_MFC 0x02 151 #define DEBUG_FORWARD 0x04 152 #define DEBUG_EXPIRE 0x08 153 #define DEBUG_XMIT 0x10 154 #define DEBUG_REG 0x20 155 #define DEBUG_PIM 0x40 156 #endif 157 158 static void expire_upcalls __P((void *)); 159 #define EXPIRE_TIMEOUT (hz / 4) /* 4x / second */ 160 #define UPCALL_EXPIRE 6 /* number of timeouts */ 161 162 #ifdef INET 163 #ifdef MROUTING 164 extern struct socket *ip_mrouter; 165 #endif 166 #endif 167 168 /* 169 * 'Interfaces' associated with decapsulator (so we can tell 170 * packets that went through it from ones that get reflected 171 * by a broken gateway). Different from IPv4 register_if, 172 * these interfaces are linked into the system ifnet list, 173 * because per-interface IPv6 statistics are maintained in 174 * ifp->if_afdata. But it does not have any routes point 175 * to them. I.e., packets can't be sent this way. They 176 * only exist as a placeholder for multicast source 177 * verification. 178 */ 179 static struct ifnet *multicast_register_if6; 180 181 #define ENCAP_HOPS 64 182 183 /* 184 * Private variables. 185 */ 186 static mifi_t nummifs = 0; 187 static mifi_t reg_mif_num = (mifi_t)-1; 188 189 static struct pim6stat pim6stat; 190 static int pim6; 191 192 /* 193 * Hash function for a source, group entry 194 */ 195 #define MF6CHASH(a, g) MF6CHASHMOD((a).s6_addr32[0] ^ (a).s6_addr32[1] ^ \ 196 (a).s6_addr32[2] ^ (a).s6_addr32[3] ^ \ 197 (g).s6_addr32[0] ^ (g).s6_addr32[1] ^ \ 198 (g).s6_addr32[2] ^ (g).s6_addr32[3]) 199 200 /* 201 * Find a route for a given origin IPv6 address and Multicast group address. 202 * Quality of service parameter to be added in the future!!! 203 */ 204 205 #define MF6CFIND(o, g, rt) do { \ 206 struct mf6c *_rt = mf6ctable[MF6CHASH(o,g)]; \ 207 rt = NULL; \ 208 mrt6stat.mrt6s_mfc_lookups++; \ 209 while (_rt) { \ 210 if (IN6_ARE_ADDR_EQUAL(&_rt->mf6c_origin.sin6_addr, &(o)) && \ 211 IN6_ARE_ADDR_EQUAL(&_rt->mf6c_mcastgrp.sin6_addr, &(g)) && \ 212 (_rt->mf6c_stall == NULL)) { \ 213 rt = _rt; \ 214 break; \ 215 } \ 216 _rt = _rt->mf6c_next; \ 217 } \ 218 if (rt == NULL) { \ 219 mrt6stat.mrt6s_mfc_misses++; \ 220 } \ 221 } while (/*CONSTCOND*/ 0) 222 223 /* 224 * Macros to compute elapsed time efficiently 225 * Borrowed from Van Jacobson's scheduling code 226 */ 227 #define TV_DELTA(a, b, delta) do { \ 228 int xxs; \ 229 \ 230 delta = (a).tv_usec - (b).tv_usec; \ 231 if ((xxs = (a).tv_sec - (b).tv_sec)) { \ 232 switch (xxs) { \ 233 case 2: \ 234 delta += 1000000; \ 235 /* FALLTHROUGH */ \ 236 case 1: \ 237 delta += 1000000; \ 238 break; \ 239 default: \ 240 delta += (1000000 * xxs); \ 241 } \ 242 } \ 243 } while (/*CONSTCOND*/ 0) 244 245 #define TV_LT(a, b) (((a).tv_usec < (b).tv_usec && \ 246 (a).tv_sec <= (b).tv_sec) || (a).tv_sec < (b).tv_sec) 247 248 #ifdef UPCALL_TIMING 249 #define UPCALL_MAX 50 250 u_long upcall_data[UPCALL_MAX + 1]; 251 static void collate(); 252 #endif /* UPCALL_TIMING */ 253 254 static int get_sg_cnt __P((struct sioc_sg_req6 *)); 255 static int get_mif6_cnt __P((struct sioc_mif_req6 *)); 256 static int ip6_mrouter_init __P((struct socket *, int, int)); 257 static int add_m6if __P((struct mif6ctl *)); 258 static int del_m6if __P((mifi_t *)); 259 static int add_m6fc __P((struct mf6cctl *)); 260 static int del_m6fc __P((struct mf6cctl *)); 261 262 static struct callout expire_upcalls_ch; 263 264 /* 265 * Handle MRT setsockopt commands to modify the multicast routing tables. 266 */ 267 int 268 ip6_mrouter_set(so, sopt) 269 struct socket *so; 270 struct sockopt *sopt; 271 { 272 int error = 0; 273 int optval; 274 struct mif6ctl mifc; 275 struct mf6cctl mfcc; 276 mifi_t mifi; 277 278 if (so != ip6_mrouter && sopt->sopt_name != MRT6_INIT) 279 return (EACCES); 280 281 switch (sopt->sopt_name) { 282 case MRT6_INIT: 283 #ifdef MRT6_OINIT 284 case MRT6_OINIT: 285 #endif 286 error = sooptcopyin(sopt, &optval, sizeof(optval), 287 sizeof(optval)); 288 if (error) 289 break; 290 error = ip6_mrouter_init(so, optval, sopt->sopt_name); 291 break; 292 case MRT6_DONE: 293 error = ip6_mrouter_done(); 294 break; 295 case MRT6_ADD_MIF: 296 error = sooptcopyin(sopt, &mifc, sizeof(mifc), sizeof(mifc)); 297 if (error) 298 break; 299 error = add_m6if(&mifc); 300 break; 301 case MRT6_ADD_MFC: 302 error = sooptcopyin(sopt, &mfcc, sizeof(mfcc), sizeof(mfcc)); 303 if (error) 304 break; 305 error = add_m6fc(&mfcc); 306 break; 307 case MRT6_DEL_MFC: 308 error = sooptcopyin(sopt, &mfcc, sizeof(mfcc), sizeof(mfcc)); 309 if (error) 310 break; 311 error = del_m6fc(&mfcc); 312 break; 313 case MRT6_DEL_MIF: 314 error = sooptcopyin(sopt, &mifi, sizeof(mifi), sizeof(mifi)); 315 if (error) 316 break; 317 error = del_m6if(&mifi); 318 break; 319 case MRT6_PIM: 320 error = sooptcopyin(sopt, &optval, sizeof(optval), 321 sizeof(optval)); 322 if (error) 323 break; 324 error = set_pim6(&optval); 325 break; 326 default: 327 error = EOPNOTSUPP; 328 break; 329 } 330 331 return (error); 332 } 333 334 /* 335 * Handle MRT getsockopt commands 336 */ 337 int 338 ip6_mrouter_get(so, sopt) 339 struct socket *so; 340 struct sockopt *sopt; 341 { 342 int error = 0; 343 344 if (so != ip6_mrouter) 345 return (EACCES); 346 347 switch (sopt->sopt_name) { 348 case MRT6_PIM: 349 error = sooptcopyout(sopt, &pim6, sizeof(pim6)); 350 break; 351 } 352 return (error); 353 } 354 355 /* 356 * Handle ioctl commands to obtain information from the cache 357 */ 358 int 359 mrt6_ioctl(cmd, data) 360 int cmd; 361 caddr_t data; 362 { 363 switch (cmd) { 364 case SIOCGETSGCNT_IN6: 365 return (get_sg_cnt((struct sioc_sg_req6 *)data)); 366 case SIOCGETMIFCNT_IN6: 367 return (get_mif6_cnt((struct sioc_mif_req6 *)data)); 368 default: 369 return (EINVAL); 370 } 371 } 372 373 /* 374 * returns the packet, byte, rpf-failure count for the source group provided 375 */ 376 static int 377 get_sg_cnt(req) 378 struct sioc_sg_req6 *req; 379 { 380 struct mf6c *rt; 381 int s; 382 383 s = splnet(); 384 MF6CFIND(req->src.sin6_addr, req->grp.sin6_addr, rt); 385 splx(s); 386 if (rt != NULL) { 387 req->pktcnt = rt->mf6c_pkt_cnt; 388 req->bytecnt = rt->mf6c_byte_cnt; 389 req->wrong_if = rt->mf6c_wrong_if; 390 } else 391 return (ESRCH); 392 #if 0 393 req->pktcnt = req->bytecnt = req->wrong_if = 0xffffffff; 394 #endif 395 396 return (0); 397 } 398 399 /* 400 * returns the input and output packet and byte counts on the mif provided 401 */ 402 static int 403 get_mif6_cnt(req) 404 struct sioc_mif_req6 *req; 405 { 406 mifi_t mifi = req->mifi; 407 408 if (mifi >= nummifs) 409 return (EINVAL); 410 411 req->icount = mif6table[mifi].m6_pkt_in; 412 req->ocount = mif6table[mifi].m6_pkt_out; 413 req->ibytes = mif6table[mifi].m6_bytes_in; 414 req->obytes = mif6table[mifi].m6_bytes_out; 415 416 return (0); 417 } 418 419 static int 420 set_pim6(i) 421 int *i; 422 { 423 if ((*i != 1) && (*i != 0)) 424 return (EINVAL); 425 426 pim6 = *i; 427 428 return (0); 429 } 430 431 /* 432 * Enable multicast routing 433 */ 434 static int 435 ip6_mrouter_init(so, v, cmd) 436 struct socket *so; 437 int v; 438 int cmd; 439 { 440 #ifdef MRT6DEBUG 441 if (mrt6debug) 442 log(LOG_DEBUG, 443 "ip6_mrouter_init: so_type = %d, pr_protocol = %d\n", 444 so->so_type, so->so_proto->pr_protocol); 445 #endif 446 447 if (so->so_type != SOCK_RAW || 448 so->so_proto->pr_protocol != IPPROTO_ICMPV6) 449 return (EOPNOTSUPP); 450 451 if (v != 1) 452 return (ENOPROTOOPT); 453 454 if (ip6_mrouter != NULL) 455 return (EADDRINUSE); 456 457 ip6_mrouter = so; 458 ip6_mrouter_ver = cmd; 459 460 bzero((caddr_t)mf6ctable, sizeof(mf6ctable)); 461 bzero((caddr_t)n6expire, sizeof(n6expire)); 462 463 pim6 = 0;/* used for stubbing out/in pim stuff */ 464 465 callout_init(&expire_upcalls_ch, 0); 466 callout_reset(&expire_upcalls_ch, EXPIRE_TIMEOUT, 467 expire_upcalls, NULL); 468 469 #ifdef MRT6DEBUG 470 if (mrt6debug) 471 log(LOG_DEBUG, "ip6_mrouter_init\n"); 472 #endif 473 474 return (0); 475 } 476 477 /* 478 * Disable multicast routing 479 */ 480 int 481 ip6_mrouter_done() 482 { 483 mifi_t mifi; 484 int i; 485 struct mf6c *rt; 486 struct rtdetq *rte; 487 int s; 488 489 s = splnet(); 490 491 /* 492 * For each phyint in use, disable promiscuous reception of all IPv6 493 * multicasts. 494 */ 495 #ifdef INET 496 #ifdef MROUTING 497 /* 498 * If there is still IPv4 multicast routing daemon, 499 * we remain interfaces to receive all muliticasted packets. 500 * XXX: there may be an interface in which the IPv4 multicast 501 * daemon is not interested... 502 */ 503 if (!ip_mrouter) 504 #endif 505 #endif 506 { 507 for (mifi = 0; mifi < nummifs; mifi++) { 508 if (mif6table[mifi].m6_ifp && 509 !(mif6table[mifi].m6_flags & MIFF_REGISTER)) { 510 if_allmulti(mif6table[mifi].m6_ifp, 0); 511 } 512 } 513 } 514 #ifdef notyet 515 bzero((caddr_t)qtable, sizeof(qtable)); 516 bzero((caddr_t)tbftable, sizeof(tbftable)); 517 #endif 518 bzero((caddr_t)mif6table, sizeof(mif6table)); 519 nummifs = 0; 520 521 pim6 = 0; /* used to stub out/in pim specific code */ 522 523 callout_stop(&expire_upcalls_ch); 524 525 /* 526 * Free all multicast forwarding cache entries. 527 */ 528 for (i = 0; i < MF6CTBLSIZ; i++) { 529 rt = mf6ctable[i]; 530 while (rt) { 531 struct mf6c *frt; 532 533 for (rte = rt->mf6c_stall; rte != NULL; ) { 534 struct rtdetq *n = rte->next; 535 536 m_free(rte->m); 537 free(rte, M_MRTABLE6); 538 rte = n; 539 } 540 frt = rt; 541 rt = rt->mf6c_next; 542 free(frt, M_MRTABLE6); 543 } 544 } 545 546 bzero((caddr_t)mf6ctable, sizeof(mf6ctable)); 547 548 /* 549 * Reset register interface 550 */ 551 if (reg_mif_num != (mifi_t)-1 && multicast_register_if6 != NULL) { 552 if_detach(multicast_register_if6); 553 if_free(multicast_register_if6); 554 reg_mif_num = (mifi_t)-1; 555 multicast_register_if6 = NULL; 556 } 557 558 ip6_mrouter = NULL; 559 ip6_mrouter_ver = 0; 560 561 splx(s); 562 563 #ifdef MRT6DEBUG 564 if (mrt6debug) 565 log(LOG_DEBUG, "ip6_mrouter_done\n"); 566 #endif 567 568 return (0); 569 } 570 571 static struct sockaddr_in6 sin6 = { sizeof(sin6), AF_INET6 }; 572 573 /* 574 * Add a mif to the mif table 575 */ 576 static int 577 add_m6if(mifcp) 578 struct mif6ctl *mifcp; 579 { 580 struct mif6 *mifp; 581 struct ifnet *ifp; 582 int error, s; 583 #ifdef notyet 584 struct tbf *m_tbf = tbftable + mifcp->mif6c_mifi; 585 #endif 586 587 if (mifcp->mif6c_mifi >= MAXMIFS) 588 return (EINVAL); 589 mifp = mif6table + mifcp->mif6c_mifi; 590 if (mifp->m6_ifp) 591 return (EADDRINUSE); /* XXX: is it appropriate? */ 592 if (mifcp->mif6c_pifi == 0 || mifcp->mif6c_pifi > if_index) 593 return (ENXIO); 594 ifp = ifnet_byindex(mifcp->mif6c_pifi); 595 596 if (mifcp->mif6c_flags & MIFF_REGISTER) { 597 if (reg_mif_num == (mifi_t)-1) { 598 ifp = if_alloc(IFT_OTHER); 599 600 if_initname(ifp, "register_mif", 0); 601 ifp->if_flags |= IFF_LOOPBACK; 602 if_attach(ifp); 603 multicast_register_if6 = ifp; 604 reg_mif_num = mifcp->mif6c_mifi; 605 /* 606 * it is impossible to guess the ifindex of the 607 * register interface. So mif6c_pifi is automatically 608 * calculated. 609 */ 610 mifcp->mif6c_pifi = ifp->if_index; 611 } else { 612 ifp = multicast_register_if6; 613 } 614 615 } /* if REGISTER */ 616 else { 617 /* Make sure the interface supports multicast */ 618 if ((ifp->if_flags & IFF_MULTICAST) == 0) 619 return (EOPNOTSUPP); 620 621 s = splnet(); 622 error = if_allmulti(ifp, 1); 623 splx(s); 624 if (error) 625 return (error); 626 } 627 628 s = splnet(); 629 mifp->m6_flags = mifcp->mif6c_flags; 630 mifp->m6_ifp = ifp; 631 #ifdef notyet 632 /* scaling up here allows division by 1024 in critical code */ 633 mifp->m6_rate_limit = mifcp->mif6c_rate_limit * 1024 / 1000; 634 #endif 635 /* initialize per mif pkt counters */ 636 mifp->m6_pkt_in = 0; 637 mifp->m6_pkt_out = 0; 638 mifp->m6_bytes_in = 0; 639 mifp->m6_bytes_out = 0; 640 splx(s); 641 642 /* Adjust nummifs up if the mifi is higher than nummifs */ 643 if (nummifs <= mifcp->mif6c_mifi) 644 nummifs = mifcp->mif6c_mifi + 1; 645 646 #ifdef MRT6DEBUG 647 if (mrt6debug) 648 log(LOG_DEBUG, 649 "add_mif #%d, phyint %s\n", 650 mifcp->mif6c_mifi, 651 ifp->if_xname); 652 #endif 653 654 return (0); 655 } 656 657 /* 658 * Delete a mif from the mif table 659 */ 660 static int 661 del_m6if(mifip) 662 mifi_t *mifip; 663 { 664 struct mif6 *mifp = mif6table + *mifip; 665 mifi_t mifi; 666 struct ifnet *ifp; 667 int s; 668 669 if (*mifip >= nummifs) 670 return (EINVAL); 671 if (mifp->m6_ifp == NULL) 672 return (EINVAL); 673 674 s = splnet(); 675 676 if (!(mifp->m6_flags & MIFF_REGISTER)) { 677 /* 678 * XXX: what if there is yet IPv4 multicast daemon 679 * using the interface? 680 */ 681 ifp = mifp->m6_ifp; 682 683 if_allmulti(ifp, 0); 684 } else { 685 if (reg_mif_num != (mifi_t)-1 && 686 multicast_register_if6 != NULL) { 687 if_detach(multicast_register_if6); 688 if_free(multicast_register_if6); 689 reg_mif_num = (mifi_t)-1; 690 multicast_register_if6 = NULL; 691 } 692 } 693 694 #ifdef notyet 695 bzero((caddr_t)qtable[*mifip], sizeof(qtable[*mifip])); 696 bzero((caddr_t)mifp->m6_tbf, sizeof(*(mifp->m6_tbf))); 697 #endif 698 bzero((caddr_t)mifp, sizeof(*mifp)); 699 700 /* Adjust nummifs down */ 701 for (mifi = nummifs; mifi > 0; mifi--) 702 if (mif6table[mifi - 1].m6_ifp) 703 break; 704 nummifs = mifi; 705 706 splx(s); 707 708 #ifdef MRT6DEBUG 709 if (mrt6debug) 710 log(LOG_DEBUG, "del_m6if %d, nummifs %d\n", *mifip, nummifs); 711 #endif 712 713 return (0); 714 } 715 716 /* 717 * Add an mfc entry 718 */ 719 static int 720 add_m6fc(mfccp) 721 struct mf6cctl *mfccp; 722 { 723 struct mf6c *rt; 724 u_long hash; 725 struct rtdetq *rte; 726 u_short nstl; 727 int s; 728 char ip6bufo[INET6_ADDRSTRLEN], ip6bufg[INET6_ADDRSTRLEN]; 729 730 MF6CFIND(mfccp->mf6cc_origin.sin6_addr, 731 mfccp->mf6cc_mcastgrp.sin6_addr, rt); 732 733 /* If an entry already exists, just update the fields */ 734 if (rt) { 735 #ifdef MRT6DEBUG 736 if (mrt6debug & DEBUG_MFC) { 737 log(LOG_DEBUG, 738 "add_m6fc no upcall h %d o %s g %s p %x\n", 739 ip6_sprintf(ip6bufo, &mfccp->mf6cc_origin.sin6_addr), 740 ip6_sprintf(ip6bufg, &mfccp->mf6cc_mcastgrp.sin6_addr), 741 mfccp->mf6cc_parent); 742 } 743 #endif 744 745 s = splnet(); 746 rt->mf6c_parent = mfccp->mf6cc_parent; 747 rt->mf6c_ifset = mfccp->mf6cc_ifset; 748 splx(s); 749 return (0); 750 } 751 752 /* 753 * Find the entry for which the upcall was made and update 754 */ 755 s = splnet(); 756 hash = MF6CHASH(mfccp->mf6cc_origin.sin6_addr, 757 mfccp->mf6cc_mcastgrp.sin6_addr); 758 for (rt = mf6ctable[hash], nstl = 0; rt; rt = rt->mf6c_next) { 759 if (IN6_ARE_ADDR_EQUAL(&rt->mf6c_origin.sin6_addr, 760 &mfccp->mf6cc_origin.sin6_addr) && 761 IN6_ARE_ADDR_EQUAL(&rt->mf6c_mcastgrp.sin6_addr, 762 &mfccp->mf6cc_mcastgrp.sin6_addr) && 763 (rt->mf6c_stall != NULL)) { 764 765 if (nstl++) 766 log(LOG_ERR, 767 "add_m6fc: %s o %s g %s p %x dbx %p\n", 768 "multiple kernel entries", 769 ip6_sprintf(ip6bufo, 770 &mfccp->mf6cc_origin.sin6_addr), 771 ip6_sprintf(ip6bufg, 772 &mfccp->mf6cc_mcastgrp.sin6_addr), 773 mfccp->mf6cc_parent, rt->mf6c_stall); 774 775 #ifdef MRT6DEBUG 776 if (mrt6debug & DEBUG_MFC) 777 log(LOG_DEBUG, 778 "add_m6fc o %s g %s p %x dbg %x\n", 779 ip6_sprintf(ip6bufo, 780 &mfccp->mf6cc_origin.sin6_addr), 781 ip6_sprintf(ip6bufg, 782 &mfccp->mf6cc_mcastgrp.sin6_addr), 783 mfccp->mf6cc_parent, rt->mf6c_stall); 784 #endif 785 786 rt->mf6c_origin = mfccp->mf6cc_origin; 787 rt->mf6c_mcastgrp = mfccp->mf6cc_mcastgrp; 788 rt->mf6c_parent = mfccp->mf6cc_parent; 789 rt->mf6c_ifset = mfccp->mf6cc_ifset; 790 /* initialize pkt counters per src-grp */ 791 rt->mf6c_pkt_cnt = 0; 792 rt->mf6c_byte_cnt = 0; 793 rt->mf6c_wrong_if = 0; 794 795 rt->mf6c_expire = 0; /* Don't clean this guy up */ 796 n6expire[hash]--; 797 798 /* free packets Qed at the end of this entry */ 799 for (rte = rt->mf6c_stall; rte != NULL; ) { 800 struct rtdetq *n = rte->next; 801 ip6_mdq(rte->m, rte->ifp, rt); 802 m_freem(rte->m); 803 #ifdef UPCALL_TIMING 804 collate(&(rte->t)); 805 #endif /* UPCALL_TIMING */ 806 free(rte, M_MRTABLE6); 807 rte = n; 808 } 809 rt->mf6c_stall = NULL; 810 } 811 } 812 813 /* 814 * It is possible that an entry is being inserted without an upcall 815 */ 816 if (nstl == 0) { 817 #ifdef MRT6DEBUG 818 if (mrt6debug & DEBUG_MFC) 819 log(LOG_DEBUG, 820 "add_mfc no upcall h %d o %s g %s p %x\n", 821 hash, 822 ip6_sprintf(ip6bufo, &mfccp->mf6cc_origin.sin6_addr), 823 ip6_sprintf(ip6bufg, &mfccp->mf6cc_mcastgrp.sin6_addr), 824 mfccp->mf6cc_parent); 825 #endif 826 827 for (rt = mf6ctable[hash]; rt; rt = rt->mf6c_next) { 828 829 if (IN6_ARE_ADDR_EQUAL(&rt->mf6c_origin.sin6_addr, 830 &mfccp->mf6cc_origin.sin6_addr)&& 831 IN6_ARE_ADDR_EQUAL(&rt->mf6c_mcastgrp.sin6_addr, 832 &mfccp->mf6cc_mcastgrp.sin6_addr)) { 833 834 rt->mf6c_origin = mfccp->mf6cc_origin; 835 rt->mf6c_mcastgrp = mfccp->mf6cc_mcastgrp; 836 rt->mf6c_parent = mfccp->mf6cc_parent; 837 rt->mf6c_ifset = mfccp->mf6cc_ifset; 838 /* initialize pkt counters per src-grp */ 839 rt->mf6c_pkt_cnt = 0; 840 rt->mf6c_byte_cnt = 0; 841 rt->mf6c_wrong_if = 0; 842 843 if (rt->mf6c_expire) 844 n6expire[hash]--; 845 rt->mf6c_expire = 0; 846 } 847 } 848 if (rt == NULL) { 849 /* no upcall, so make a new entry */ 850 rt = (struct mf6c *)malloc(sizeof(*rt), M_MRTABLE6, 851 M_NOWAIT); 852 if (rt == NULL) { 853 splx(s); 854 return (ENOBUFS); 855 } 856 857 /* insert new entry at head of hash chain */ 858 rt->mf6c_origin = mfccp->mf6cc_origin; 859 rt->mf6c_mcastgrp = mfccp->mf6cc_mcastgrp; 860 rt->mf6c_parent = mfccp->mf6cc_parent; 861 rt->mf6c_ifset = mfccp->mf6cc_ifset; 862 /* initialize pkt counters per src-grp */ 863 rt->mf6c_pkt_cnt = 0; 864 rt->mf6c_byte_cnt = 0; 865 rt->mf6c_wrong_if = 0; 866 rt->mf6c_expire = 0; 867 rt->mf6c_stall = NULL; 868 869 /* link into table */ 870 rt->mf6c_next = mf6ctable[hash]; 871 mf6ctable[hash] = rt; 872 } 873 } 874 splx(s); 875 return (0); 876 } 877 878 #ifdef UPCALL_TIMING 879 /* 880 * collect delay statistics on the upcalls 881 */ 882 static void 883 collate(t) 884 struct timeval *t; 885 { 886 u_long d; 887 struct timeval tp; 888 u_long delta; 889 890 GET_TIME(tp); 891 892 if (TV_LT(*t, tp)) 893 { 894 TV_DELTA(tp, *t, delta); 895 896 d = delta >> 10; 897 if (d > UPCALL_MAX) 898 d = UPCALL_MAX; 899 900 ++upcall_data[d]; 901 } 902 } 903 #endif /* UPCALL_TIMING */ 904 905 /* 906 * Delete an mfc entry 907 */ 908 static int 909 del_m6fc(mfccp) 910 struct mf6cctl *mfccp; 911 { 912 struct sockaddr_in6 origin; 913 struct sockaddr_in6 mcastgrp; 914 struct mf6c *rt; 915 struct mf6c **nptr; 916 u_long hash; 917 int s; 918 919 origin = mfccp->mf6cc_origin; 920 mcastgrp = mfccp->mf6cc_mcastgrp; 921 hash = MF6CHASH(origin.sin6_addr, mcastgrp.sin6_addr); 922 923 #ifdef MRT6DEBUG 924 if (mrt6debug & DEBUG_MFC) { 925 char ip6bufo[INET6_ADDRSTRLEN], ip6bufg[INET6_ADDRSTRLEN]; 926 log(LOG_DEBUG,"del_m6fc orig %s mcastgrp %s\n", 927 ip6_sprintf(ip6bufo, &origin.sin6_addr), 928 ip6_sprintf(ip6bufg, &mcastgrp.sin6_addr)); 929 } 930 #endif 931 932 s = splnet(); 933 934 nptr = &mf6ctable[hash]; 935 while ((rt = *nptr) != NULL) { 936 if (IN6_ARE_ADDR_EQUAL(&origin.sin6_addr, 937 &rt->mf6c_origin.sin6_addr) && 938 IN6_ARE_ADDR_EQUAL(&mcastgrp.sin6_addr, 939 &rt->mf6c_mcastgrp.sin6_addr) && 940 rt->mf6c_stall == NULL) 941 break; 942 943 nptr = &rt->mf6c_next; 944 } 945 if (rt == NULL) { 946 splx(s); 947 return (EADDRNOTAVAIL); 948 } 949 950 *nptr = rt->mf6c_next; 951 free(rt, M_MRTABLE6); 952 953 splx(s); 954 955 return (0); 956 } 957 958 static int 959 socket_send(s, mm, src) 960 struct socket *s; 961 struct mbuf *mm; 962 struct sockaddr_in6 *src; 963 { 964 if (s) { 965 if (sbappendaddr(&s->so_rcv, 966 (struct sockaddr *)src, 967 mm, (struct mbuf *)0) != 0) { 968 sorwakeup(s); 969 return (0); 970 } 971 } 972 m_freem(mm); 973 return (-1); 974 } 975 976 /* 977 * IPv6 multicast forwarding function. This function assumes that the packet 978 * pointed to by "ip6" has arrived on (or is about to be sent to) the interface 979 * pointed to by "ifp", and the packet is to be relayed to other networks 980 * that have members of the packet's destination IPv6 multicast group. 981 * 982 * The packet is returned unscathed to the caller, unless it is 983 * erroneous, in which case a non-zero return value tells the caller to 984 * discard it. 985 * 986 * NOTE: this implementation assumes that m->m_pkthdr.rcvif is NULL iff 987 * this function is called in the originating context (i.e., not when 988 * forwarding a packet from other node). ip6_output(), which is currently the 989 * only function that calls this function is called in the originating context, 990 * explicitly ensures this condition. It is caller's responsibility to ensure 991 * that if this function is called from somewhere else in the originating 992 * context in the future. 993 */ 994 995 int 996 ip6_mforward(ip6, ifp, m) 997 struct ip6_hdr *ip6; 998 struct ifnet *ifp; 999 struct mbuf *m; 1000 { 1001 struct mf6c *rt; 1002 struct mif6 *mifp; 1003 struct mbuf *mm; 1004 int s; 1005 mifi_t mifi; 1006 char ip6bufs[INET6_ADDRSTRLEN], ip6bufd[INET6_ADDRSTRLEN]; 1007 1008 #ifdef MRT6DEBUG 1009 if (mrt6debug & DEBUG_FORWARD) 1010 log(LOG_DEBUG, "ip6_mforward: src %s, dst %s, ifindex %d\n", 1011 ip6_sprintf(ip6bufs, &ip6->ip6_src), 1012 ip6_sprintf(ip6bufd, &ip6->ip6_dst), 1013 ifp->if_index); 1014 #endif 1015 1016 /* 1017 * Don't forward a packet with Hop limit of zero or one, 1018 * or a packet destined to a local-only group. 1019 */ 1020 if (ip6->ip6_hlim <= 1 || IN6_IS_ADDR_MC_INTFACELOCAL(&ip6->ip6_dst) || 1021 IN6_IS_ADDR_MC_LINKLOCAL(&ip6->ip6_dst)) 1022 return (0); 1023 ip6->ip6_hlim--; 1024 1025 /* 1026 * Source address check: do not forward packets with unspecified 1027 * source. It was discussed in July 2000, on ipngwg mailing list. 1028 * This is rather more serious than unicast cases, because some 1029 * MLD packets can be sent with the unspecified source address 1030 * (although such packets must normally set 1 to the hop limit field). 1031 */ 1032 if (IN6_IS_ADDR_UNSPECIFIED(&ip6->ip6_src)) { 1033 ip6stat.ip6s_cantforward++; 1034 if (ip6_log_time + ip6_log_interval < time_second) { 1035 ip6_log_time = time_second; 1036 log(LOG_DEBUG, 1037 "cannot forward " 1038 "from %s to %s nxt %d received on %s\n", 1039 ip6_sprintf(ip6bufs, &ip6->ip6_src), 1040 ip6_sprintf(ip6bufd, &ip6->ip6_dst), 1041 ip6->ip6_nxt, 1042 if_name(m->m_pkthdr.rcvif)); 1043 } 1044 return (0); 1045 } 1046 1047 /* 1048 * Determine forwarding mifs from the forwarding cache table 1049 */ 1050 s = splnet(); 1051 MF6CFIND(ip6->ip6_src, ip6->ip6_dst, rt); 1052 1053 /* Entry exists, so forward if necessary */ 1054 if (rt) { 1055 splx(s); 1056 return (ip6_mdq(m, ifp, rt)); 1057 } else { 1058 /* 1059 * If we don't have a route for packet's origin, 1060 * Make a copy of the packet & 1061 * send message to routing daemon 1062 */ 1063 1064 struct mbuf *mb0; 1065 struct rtdetq *rte; 1066 u_long hash; 1067 /* int i, npkts;*/ 1068 #ifdef UPCALL_TIMING 1069 struct timeval tp; 1070 1071 GET_TIME(tp); 1072 #endif /* UPCALL_TIMING */ 1073 1074 mrt6stat.mrt6s_no_route++; 1075 #ifdef MRT6DEBUG 1076 if (mrt6debug & (DEBUG_FORWARD | DEBUG_MFC)) 1077 log(LOG_DEBUG, "ip6_mforward: no rte s %s g %s\n", 1078 ip6_sprintf(ip6bufs, &ip6->ip6_src), 1079 ip6_sprintf(ip6bufd, &ip6->ip6_dst)); 1080 #endif 1081 1082 /* 1083 * Allocate mbufs early so that we don't do extra work if we 1084 * are just going to fail anyway. 1085 */ 1086 rte = (struct rtdetq *)malloc(sizeof(*rte), M_MRTABLE6, 1087 M_NOWAIT); 1088 if (rte == NULL) { 1089 splx(s); 1090 return (ENOBUFS); 1091 } 1092 mb0 = m_copy(m, 0, M_COPYALL); 1093 /* 1094 * Pullup packet header if needed before storing it, 1095 * as other references may modify it in the meantime. 1096 */ 1097 if (mb0 && 1098 (M_HASCL(mb0) || mb0->m_len < sizeof(struct ip6_hdr))) 1099 mb0 = m_pullup(mb0, sizeof(struct ip6_hdr)); 1100 if (mb0 == NULL) { 1101 free(rte, M_MRTABLE6); 1102 splx(s); 1103 return (ENOBUFS); 1104 } 1105 1106 /* is there an upcall waiting for this packet? */ 1107 hash = MF6CHASH(ip6->ip6_src, ip6->ip6_dst); 1108 for (rt = mf6ctable[hash]; rt; rt = rt->mf6c_next) { 1109 if (IN6_ARE_ADDR_EQUAL(&ip6->ip6_src, 1110 &rt->mf6c_origin.sin6_addr) && 1111 IN6_ARE_ADDR_EQUAL(&ip6->ip6_dst, 1112 &rt->mf6c_mcastgrp.sin6_addr) && 1113 (rt->mf6c_stall != NULL)) 1114 break; 1115 } 1116 1117 if (rt == NULL) { 1118 struct mrt6msg *im; 1119 #ifdef MRT6_OINIT 1120 struct omrt6msg *oim; 1121 #endif 1122 1123 /* no upcall, so make a new entry */ 1124 rt = (struct mf6c *)malloc(sizeof(*rt), M_MRTABLE6, 1125 M_NOWAIT); 1126 if (rt == NULL) { 1127 free(rte, M_MRTABLE6); 1128 m_freem(mb0); 1129 splx(s); 1130 return (ENOBUFS); 1131 } 1132 /* 1133 * Make a copy of the header to send to the user 1134 * level process 1135 */ 1136 mm = m_copy(mb0, 0, sizeof(struct ip6_hdr)); 1137 1138 if (mm == NULL) { 1139 free(rte, M_MRTABLE6); 1140 m_freem(mb0); 1141 free(rt, M_MRTABLE6); 1142 splx(s); 1143 return (ENOBUFS); 1144 } 1145 1146 /* 1147 * Send message to routing daemon 1148 */ 1149 sin6.sin6_addr = ip6->ip6_src; 1150 1151 im = NULL; 1152 #ifdef MRT6_OINIT 1153 oim = NULL; 1154 #endif 1155 switch (ip6_mrouter_ver) { 1156 #ifdef MRT6_OINIT 1157 case MRT6_OINIT: 1158 oim = mtod(mm, struct omrt6msg *); 1159 oim->im6_msgtype = MRT6MSG_NOCACHE; 1160 oim->im6_mbz = 0; 1161 break; 1162 #endif 1163 case MRT6_INIT: 1164 im = mtod(mm, struct mrt6msg *); 1165 im->im6_msgtype = MRT6MSG_NOCACHE; 1166 im->im6_mbz = 0; 1167 break; 1168 default: 1169 free(rte, M_MRTABLE6); 1170 m_freem(mb0); 1171 free(rt, M_MRTABLE6); 1172 splx(s); 1173 return (EINVAL); 1174 } 1175 1176 #ifdef MRT6DEBUG 1177 if (mrt6debug & DEBUG_FORWARD) 1178 log(LOG_DEBUG, 1179 "getting the iif info in the kernel\n"); 1180 #endif 1181 1182 for (mifp = mif6table, mifi = 0; 1183 mifi < nummifs && mifp->m6_ifp != ifp; 1184 mifp++, mifi++) 1185 ; 1186 1187 switch (ip6_mrouter_ver) { 1188 #ifdef MRT6_OINIT 1189 case MRT6_OINIT: 1190 oim->im6_mif = mifi; 1191 break; 1192 #endif 1193 case MRT6_INIT: 1194 im->im6_mif = mifi; 1195 break; 1196 } 1197 1198 if (socket_send(ip6_mrouter, mm, &sin6) < 0) { 1199 log(LOG_WARNING, "ip6_mforward: ip6_mrouter " 1200 "socket queue full\n"); 1201 mrt6stat.mrt6s_upq_sockfull++; 1202 free(rte, M_MRTABLE6); 1203 m_freem(mb0); 1204 free(rt, M_MRTABLE6); 1205 splx(s); 1206 return (ENOBUFS); 1207 } 1208 1209 mrt6stat.mrt6s_upcalls++; 1210 1211 /* insert new entry at head of hash chain */ 1212 bzero(rt, sizeof(*rt)); 1213 rt->mf6c_origin.sin6_family = AF_INET6; 1214 rt->mf6c_origin.sin6_len = sizeof(struct sockaddr_in6); 1215 rt->mf6c_origin.sin6_addr = ip6->ip6_src; 1216 rt->mf6c_mcastgrp.sin6_family = AF_INET6; 1217 rt->mf6c_mcastgrp.sin6_len = sizeof(struct sockaddr_in6); 1218 rt->mf6c_mcastgrp.sin6_addr = ip6->ip6_dst; 1219 rt->mf6c_expire = UPCALL_EXPIRE; 1220 n6expire[hash]++; 1221 rt->mf6c_parent = MF6C_INCOMPLETE_PARENT; 1222 1223 /* link into table */ 1224 rt->mf6c_next = mf6ctable[hash]; 1225 mf6ctable[hash] = rt; 1226 /* Add this entry to the end of the queue */ 1227 rt->mf6c_stall = rte; 1228 } else { 1229 /* determine if q has overflowed */ 1230 struct rtdetq **p; 1231 int npkts = 0; 1232 1233 for (p = &rt->mf6c_stall; *p != NULL; p = &(*p)->next) 1234 if (++npkts > MAX_UPQ6) { 1235 mrt6stat.mrt6s_upq_ovflw++; 1236 free(rte, M_MRTABLE6); 1237 m_freem(mb0); 1238 splx(s); 1239 return (0); 1240 } 1241 1242 /* Add this entry to the end of the queue */ 1243 *p = rte; 1244 } 1245 1246 rte->next = NULL; 1247 rte->m = mb0; 1248 rte->ifp = ifp; 1249 #ifdef UPCALL_TIMING 1250 rte->t = tp; 1251 #endif /* UPCALL_TIMING */ 1252 1253 splx(s); 1254 1255 return (0); 1256 } 1257 } 1258 1259 /* 1260 * Clean up cache entries if upcalls are not serviced 1261 * Call from the Slow Timeout mechanism, every half second. 1262 */ 1263 static void 1264 expire_upcalls(unused) 1265 void *unused; 1266 { 1267 struct rtdetq *rte; 1268 struct mf6c *mfc, **nptr; 1269 int i; 1270 int s; 1271 1272 s = splnet(); 1273 for (i = 0; i < MF6CTBLSIZ; i++) { 1274 if (n6expire[i] == 0) 1275 continue; 1276 nptr = &mf6ctable[i]; 1277 while ((mfc = *nptr) != NULL) { 1278 rte = mfc->mf6c_stall; 1279 /* 1280 * Skip real cache entries 1281 * Make sure it wasn't marked to not expire (shouldn't happen) 1282 * If it expires now 1283 */ 1284 if (rte != NULL && 1285 mfc->mf6c_expire != 0 && 1286 --mfc->mf6c_expire == 0) { 1287 #ifdef MRT6DEBUG 1288 if (mrt6debug & DEBUG_EXPIRE) { 1289 char ip6bufo[INET6_ADDRSTRLEN]; 1290 char ip6bufg[INET6_ADDRSTRLEN]; 1291 log(LOG_DEBUG, "expire_upcalls: expiring (%s %s)\n", 1292 ip6_sprintf(ip6bufo, &mfc->mf6c_origin.sin6_addr), 1293 ip6_sprintf(ip6bufg, &mfc->mf6c_mcastgrp.sin6_addr)); 1294 } 1295 #endif 1296 /* 1297 * drop all the packets 1298 * free the mbuf with the pkt, if, timing info 1299 */ 1300 do { 1301 struct rtdetq *n = rte->next; 1302 m_freem(rte->m); 1303 free(rte, M_MRTABLE6); 1304 rte = n; 1305 } while (rte != NULL); 1306 mrt6stat.mrt6s_cache_cleanups++; 1307 n6expire[i]--; 1308 1309 *nptr = mfc->mf6c_next; 1310 free(mfc, M_MRTABLE6); 1311 } else { 1312 nptr = &mfc->mf6c_next; 1313 } 1314 } 1315 } 1316 splx(s); 1317 callout_reset(&expire_upcalls_ch, EXPIRE_TIMEOUT, 1318 expire_upcalls, NULL); 1319 } 1320 1321 /* 1322 * Packet forwarding routine once entry in the cache is made 1323 */ 1324 static int 1325 ip6_mdq(m, ifp, rt) 1326 struct mbuf *m; 1327 struct ifnet *ifp; 1328 struct mf6c *rt; 1329 { 1330 struct ip6_hdr *ip6 = mtod(m, struct ip6_hdr *); 1331 mifi_t mifi, iif; 1332 struct mif6 *mifp; 1333 int plen = m->m_pkthdr.len; 1334 struct in6_addr src0, dst0; /* copies for local work */ 1335 u_int32_t iszone, idzone, oszone, odzone; 1336 int error = 0; 1337 1338 /* 1339 * Macro to send packet on mif. Since RSVP packets don't get counted on 1340 * input, they shouldn't get counted on output, so statistics keeping is 1341 * separate. 1342 */ 1343 1344 #define MC6_SEND(ip6, mifp, m) do { \ 1345 if ((mifp)->m6_flags & MIFF_REGISTER) \ 1346 register_send((ip6), (mifp), (m)); \ 1347 else \ 1348 phyint_send((ip6), (mifp), (m)); \ 1349 } while (/*CONSTCOND*/ 0) 1350 1351 /* 1352 * Don't forward if it didn't arrive from the parent mif 1353 * for its origin. 1354 */ 1355 mifi = rt->mf6c_parent; 1356 if ((mifi >= nummifs) || (mif6table[mifi].m6_ifp != ifp)) { 1357 /* came in the wrong interface */ 1358 #ifdef MRT6DEBUG 1359 if (mrt6debug & DEBUG_FORWARD) 1360 log(LOG_DEBUG, 1361 "wrong if: ifid %d mifi %d mififid %x\n", 1362 ifp->if_index, mifi, 1363 mif6table[mifi].m6_ifp->if_index); 1364 #endif 1365 mrt6stat.mrt6s_wrong_if++; 1366 rt->mf6c_wrong_if++; 1367 /* 1368 * If we are doing PIM processing, and we are forwarding 1369 * packets on this interface, send a message to the 1370 * routing daemon. 1371 */ 1372 /* have to make sure this is a valid mif */ 1373 if (mifi < nummifs && mif6table[mifi].m6_ifp) 1374 if (pim6 && (m->m_flags & M_LOOP) == 0) { 1375 /* 1376 * Check the M_LOOP flag to avoid an 1377 * unnecessary PIM assert. 1378 * XXX: M_LOOP is an ad-hoc hack... 1379 */ 1380 static struct sockaddr_in6 sin6 = 1381 { sizeof(sin6), AF_INET6 }; 1382 1383 struct mbuf *mm; 1384 struct mrt6msg *im; 1385 #ifdef MRT6_OINIT 1386 struct omrt6msg *oim; 1387 #endif 1388 1389 mm = m_copy(m, 0, sizeof(struct ip6_hdr)); 1390 if (mm && 1391 (M_HASCL(mm) || 1392 mm->m_len < sizeof(struct ip6_hdr))) 1393 mm = m_pullup(mm, sizeof(struct ip6_hdr)); 1394 if (mm == NULL) 1395 return (ENOBUFS); 1396 1397 #ifdef MRT6_OINIT 1398 oim = NULL; 1399 #endif 1400 im = NULL; 1401 switch (ip6_mrouter_ver) { 1402 #ifdef MRT6_OINIT 1403 case MRT6_OINIT: 1404 oim = mtod(mm, struct omrt6msg *); 1405 oim->im6_msgtype = MRT6MSG_WRONGMIF; 1406 oim->im6_mbz = 0; 1407 break; 1408 #endif 1409 case MRT6_INIT: 1410 im = mtod(mm, struct mrt6msg *); 1411 im->im6_msgtype = MRT6MSG_WRONGMIF; 1412 im->im6_mbz = 0; 1413 break; 1414 default: 1415 m_freem(mm); 1416 return (EINVAL); 1417 } 1418 1419 for (mifp = mif6table, iif = 0; 1420 iif < nummifs && mifp && 1421 mifp->m6_ifp != ifp; 1422 mifp++, iif++) 1423 ; 1424 1425 switch (ip6_mrouter_ver) { 1426 #ifdef MRT6_OINIT 1427 case MRT6_OINIT: 1428 oim->im6_mif = iif; 1429 sin6.sin6_addr = oim->im6_src; 1430 break; 1431 #endif 1432 case MRT6_INIT: 1433 im->im6_mif = iif; 1434 sin6.sin6_addr = im->im6_src; 1435 break; 1436 } 1437 1438 mrt6stat.mrt6s_upcalls++; 1439 1440 if (socket_send(ip6_mrouter, mm, &sin6) < 0) { 1441 #ifdef MRT6DEBUG 1442 if (mrt6debug) 1443 log(LOG_WARNING, "mdq, ip6_mrouter socket queue full\n"); 1444 #endif 1445 ++mrt6stat.mrt6s_upq_sockfull; 1446 return (ENOBUFS); 1447 } /* if socket Q full */ 1448 } /* if PIM */ 1449 return (0); 1450 } /* if wrong iif */ 1451 1452 /* If I sourced this packet, it counts as output, else it was input. */ 1453 if (m->m_pkthdr.rcvif == NULL) { 1454 /* XXX: is rcvif really NULL when output?? */ 1455 mif6table[mifi].m6_pkt_out++; 1456 mif6table[mifi].m6_bytes_out += plen; 1457 } else { 1458 mif6table[mifi].m6_pkt_in++; 1459 mif6table[mifi].m6_bytes_in += plen; 1460 } 1461 rt->mf6c_pkt_cnt++; 1462 rt->mf6c_byte_cnt += plen; 1463 1464 /* 1465 * For each mif, forward a copy of the packet if there are group 1466 * members downstream on the interface. 1467 */ 1468 src0 = ip6->ip6_src; 1469 dst0 = ip6->ip6_dst; 1470 if ((error = in6_setscope(&src0, ifp, &iszone)) != 0 || 1471 (error = in6_setscope(&dst0, ifp, &idzone)) != 0) { 1472 ip6stat.ip6s_badscope++; 1473 return (error); 1474 } 1475 for (mifp = mif6table, mifi = 0; mifi < nummifs; mifp++, mifi++) { 1476 if (IF_ISSET(mifi, &rt->mf6c_ifset)) { 1477 /* 1478 * check if the outgoing packet is going to break 1479 * a scope boundary. 1480 * XXX For packets through PIM register tunnel 1481 * interface, we believe a routing daemon. 1482 */ 1483 if (!(mif6table[rt->mf6c_parent].m6_flags & 1484 MIFF_REGISTER) && 1485 !(mif6table[mifi].m6_flags & MIFF_REGISTER)) { 1486 if (in6_setscope(&src0, mif6table[mifi].m6_ifp, 1487 &oszone) || 1488 in6_setscope(&dst0, mif6table[mifi].m6_ifp, 1489 &odzone) || 1490 iszone != oszone || 1491 idzone != odzone) { 1492 ip6stat.ip6s_badscope++; 1493 continue; 1494 } 1495 } 1496 1497 mifp->m6_pkt_out++; 1498 mifp->m6_bytes_out += plen; 1499 MC6_SEND(ip6, mifp, m); 1500 } 1501 } 1502 return (0); 1503 } 1504 1505 static void 1506 phyint_send(ip6, mifp, m) 1507 struct ip6_hdr *ip6; 1508 struct mif6 *mifp; 1509 struct mbuf *m; 1510 { 1511 struct mbuf *mb_copy; 1512 struct ifnet *ifp = mifp->m6_ifp; 1513 int error = 0; 1514 int s = splnet(); /* needs to protect static "ro" below. */ 1515 static struct route_in6 ro; 1516 struct in6_multi *in6m; 1517 struct sockaddr_in6 *dst6; 1518 u_long linkmtu; 1519 1520 /* 1521 * Make a new reference to the packet; make sure that 1522 * the IPv6 header is actually copied, not just referenced, 1523 * so that ip6_output() only scribbles on the copy. 1524 */ 1525 mb_copy = m_copy(m, 0, M_COPYALL); 1526 if (mb_copy && 1527 (M_HASCL(mb_copy) || mb_copy->m_len < sizeof(struct ip6_hdr))) 1528 mb_copy = m_pullup(mb_copy, sizeof(struct ip6_hdr)); 1529 if (mb_copy == NULL) { 1530 splx(s); 1531 return; 1532 } 1533 /* set MCAST flag to the outgoing packet */ 1534 mb_copy->m_flags |= M_MCAST; 1535 1536 /* 1537 * If we sourced the packet, call ip6_output since we may devide 1538 * the packet into fragments when the packet is too big for the 1539 * outgoing interface. 1540 * Otherwise, we can simply send the packet to the interface 1541 * sending queue. 1542 */ 1543 if (m->m_pkthdr.rcvif == NULL) { 1544 struct ip6_moptions im6o; 1545 1546 im6o.im6o_multicast_ifp = ifp; 1547 /* XXX: ip6_output will override ip6->ip6_hlim */ 1548 im6o.im6o_multicast_hlim = ip6->ip6_hlim; 1549 im6o.im6o_multicast_loop = 1; 1550 error = ip6_output(mb_copy, NULL, &ro, 1551 IPV6_FORWARDING, &im6o, NULL, NULL); 1552 1553 #ifdef MRT6DEBUG 1554 if (mrt6debug & DEBUG_XMIT) 1555 log(LOG_DEBUG, "phyint_send on mif %d err %d\n", 1556 mifp - mif6table, error); 1557 #endif 1558 splx(s); 1559 return; 1560 } 1561 1562 /* 1563 * If we belong to the destination multicast group 1564 * on the outgoing interface, loop back a copy. 1565 */ 1566 dst6 = (struct sockaddr_in6 *)&ro.ro_dst; 1567 IN6_LOOKUP_MULTI(ip6->ip6_dst, ifp, in6m); 1568 if (in6m != NULL) { 1569 dst6->sin6_len = sizeof(struct sockaddr_in6); 1570 dst6->sin6_family = AF_INET6; 1571 dst6->sin6_addr = ip6->ip6_dst; 1572 ip6_mloopback(ifp, m, (struct sockaddr_in6 *)&ro.ro_dst); 1573 } 1574 /* 1575 * Put the packet into the sending queue of the outgoing interface 1576 * if it would fit in the MTU of the interface. 1577 */ 1578 linkmtu = IN6_LINKMTU(ifp); 1579 if (mb_copy->m_pkthdr.len <= linkmtu || linkmtu < IPV6_MMTU) { 1580 dst6->sin6_len = sizeof(struct sockaddr_in6); 1581 dst6->sin6_family = AF_INET6; 1582 dst6->sin6_addr = ip6->ip6_dst; 1583 /* 1584 * We just call if_output instead of nd6_output here, since 1585 * we need no ND for a multicast forwarded packet...right? 1586 */ 1587 error = (*ifp->if_output)(ifp, mb_copy, 1588 (struct sockaddr *)&ro.ro_dst, NULL); 1589 #ifdef MRT6DEBUG 1590 if (mrt6debug & DEBUG_XMIT) 1591 log(LOG_DEBUG, "phyint_send on mif %d err %d\n", 1592 mifp - mif6table, error); 1593 #endif 1594 } else { 1595 /* 1596 * pMTU discovery is intentionally disabled by default, since 1597 * various router may notify pMTU in multicast, which can be 1598 * a DDoS to a router 1599 */ 1600 if (ip6_mcast_pmtu) 1601 icmp6_error(mb_copy, ICMP6_PACKET_TOO_BIG, 0, linkmtu); 1602 else { 1603 #ifdef MRT6DEBUG 1604 if (mrt6debug & DEBUG_XMIT) { 1605 char ip6bufs[INET6_ADDRSTRLEN]; 1606 char ip6bufd[INET6_ADDRSTRLEN]; 1607 log(LOG_DEBUG, 1608 "phyint_send: packet too big on %s o %s " 1609 "g %s size %d(discarded)\n", 1610 if_name(ifp), 1611 ip6_sprintf(ip6bufs, &ip6->ip6_src), 1612 ip6_sprintf(ip6bufd, &ip6->ip6_dst), 1613 mb_copy->m_pkthdr.len); 1614 } 1615 #endif /* MRT6DEBUG */ 1616 m_freem(mb_copy); /* simply discard the packet */ 1617 } 1618 } 1619 1620 splx(s); 1621 } 1622 1623 static int 1624 register_send(ip6, mif, m) 1625 struct ip6_hdr *ip6; 1626 struct mif6 *mif; 1627 struct mbuf *m; 1628 { 1629 struct mbuf *mm; 1630 int i, len = m->m_pkthdr.len; 1631 static struct sockaddr_in6 sin6 = { sizeof(sin6), AF_INET6 }; 1632 struct mrt6msg *im6; 1633 1634 #ifdef MRT6DEBUG 1635 if (mrt6debug) { 1636 char ip6bufs[INET6_ADDRSTRLEN], ip6bufd[INET6_ADDRSTRLEN]; 1637 log(LOG_DEBUG, "** IPv6 register_send **\n src %s dst %s\n", 1638 ip6_sprintf(ip6bufs, &ip6->ip6_src), 1639 ip6_sprintf(ip6bufd, &ip6->ip6_dst)); 1640 } 1641 #endif 1642 ++pim6stat.pim6s_snd_registers; 1643 1644 /* Make a copy of the packet to send to the user level process */ 1645 MGETHDR(mm, M_DONTWAIT, MT_HEADER); 1646 if (mm == NULL) 1647 return (ENOBUFS); 1648 mm->m_pkthdr.rcvif = NULL; 1649 mm->m_data += max_linkhdr; 1650 mm->m_len = sizeof(struct ip6_hdr); 1651 1652 if ((mm->m_next = m_copy(m, 0, M_COPYALL)) == NULL) { 1653 m_freem(mm); 1654 return (ENOBUFS); 1655 } 1656 i = MHLEN - M_LEADINGSPACE(mm); 1657 if (i > len) 1658 i = len; 1659 mm = m_pullup(mm, i); 1660 if (mm == NULL) 1661 return (ENOBUFS); 1662 /* TODO: check it! */ 1663 mm->m_pkthdr.len = len + sizeof(struct ip6_hdr); 1664 1665 /* 1666 * Send message to routing daemon 1667 */ 1668 sin6.sin6_addr = ip6->ip6_src; 1669 1670 im6 = mtod(mm, struct mrt6msg *); 1671 im6->im6_msgtype = MRT6MSG_WHOLEPKT; 1672 im6->im6_mbz = 0; 1673 1674 im6->im6_mif = mif - mif6table; 1675 1676 /* iif info is not given for reg. encap.n */ 1677 mrt6stat.mrt6s_upcalls++; 1678 1679 if (socket_send(ip6_mrouter, mm, &sin6) < 0) { 1680 #ifdef MRT6DEBUG 1681 if (mrt6debug) 1682 log(LOG_WARNING, 1683 "register_send: ip6_mrouter socket queue full\n"); 1684 #endif 1685 ++mrt6stat.mrt6s_upq_sockfull; 1686 return (ENOBUFS); 1687 } 1688 return (0); 1689 } 1690 1691 /* 1692 * PIM sparse mode hook 1693 * Receives the pim control messages, and passes them up to the listening 1694 * socket, using rip6_input. 1695 * The only message processed is the REGISTER pim message; the pim header 1696 * is stripped off, and the inner packet is passed to register_mforward. 1697 */ 1698 int 1699 pim6_input(mp, offp, proto) 1700 struct mbuf **mp; 1701 int *offp, proto; 1702 { 1703 struct pim *pim; /* pointer to a pim struct */ 1704 struct ip6_hdr *ip6; 1705 int pimlen; 1706 struct mbuf *m = *mp; 1707 int minlen; 1708 int off = *offp; 1709 1710 ++pim6stat.pim6s_rcv_total; 1711 1712 ip6 = mtod(m, struct ip6_hdr *); 1713 pimlen = m->m_pkthdr.len - *offp; 1714 1715 /* 1716 * Validate lengths 1717 */ 1718 if (pimlen < PIM_MINLEN) { 1719 ++pim6stat.pim6s_rcv_tooshort; 1720 #ifdef MRT6DEBUG 1721 if (mrt6debug & DEBUG_PIM) 1722 log(LOG_DEBUG,"pim6_input: PIM packet too short\n"); 1723 #endif 1724 m_freem(m); 1725 return (IPPROTO_DONE); 1726 } 1727 1728 /* 1729 * if the packet is at least as big as a REGISTER, go ahead 1730 * and grab the PIM REGISTER header size, to avoid another 1731 * possible m_pullup() later. 1732 * 1733 * PIM_MINLEN == pimhdr + u_int32 == 8 1734 * PIM6_REG_MINLEN == pimhdr + reghdr + eip6hdr == 4 + 4 + 40 1735 */ 1736 minlen = (pimlen >= PIM6_REG_MINLEN) ? PIM6_REG_MINLEN : PIM_MINLEN; 1737 1738 /* 1739 * Make sure that the IP6 and PIM headers in contiguous memory, and 1740 * possibly the PIM REGISTER header 1741 */ 1742 #ifndef PULLDOWN_TEST 1743 IP6_EXTHDR_CHECK(m, off, minlen, IPPROTO_DONE); 1744 /* adjust pointer */ 1745 ip6 = mtod(m, struct ip6_hdr *); 1746 1747 /* adjust mbuf to point to the PIM header */ 1748 pim = (struct pim *)((caddr_t)ip6 + off); 1749 #else 1750 IP6_EXTHDR_GET(pim, struct pim *, m, off, minlen); 1751 if (pim == NULL) { 1752 pim6stat.pim6s_rcv_tooshort++; 1753 return (IPPROTO_DONE); 1754 } 1755 #endif 1756 1757 #define PIM6_CHECKSUM 1758 #ifdef PIM6_CHECKSUM 1759 { 1760 int cksumlen; 1761 1762 /* 1763 * Validate checksum. 1764 * If PIM REGISTER, exclude the data packet 1765 */ 1766 if (pim->pim_type == PIM_REGISTER) 1767 cksumlen = PIM_MINLEN; 1768 else 1769 cksumlen = pimlen; 1770 1771 if (in6_cksum(m, IPPROTO_PIM, off, cksumlen)) { 1772 ++pim6stat.pim6s_rcv_badsum; 1773 #ifdef MRT6DEBUG 1774 if (mrt6debug & DEBUG_PIM) 1775 log(LOG_DEBUG, 1776 "pim6_input: invalid checksum\n"); 1777 #endif 1778 m_freem(m); 1779 return (IPPROTO_DONE); 1780 } 1781 } 1782 #endif /* PIM_CHECKSUM */ 1783 1784 /* PIM version check */ 1785 if (pim->pim_ver != PIM_VERSION) { 1786 ++pim6stat.pim6s_rcv_badversion; 1787 #ifdef MRT6DEBUG 1788 log(LOG_ERR, 1789 "pim6_input: incorrect version %d, expecting %d\n", 1790 pim->pim_ver, PIM_VERSION); 1791 #endif 1792 m_freem(m); 1793 return (IPPROTO_DONE); 1794 } 1795 1796 if (pim->pim_type == PIM_REGISTER) { 1797 /* 1798 * since this is a REGISTER, we'll make a copy of the register 1799 * headers ip6+pim+u_int32_t+encap_ip6, to be passed up to the 1800 * routing daemon. 1801 */ 1802 static struct sockaddr_in6 dst = { sizeof(dst), AF_INET6 }; 1803 1804 struct mbuf *mcp; 1805 struct ip6_hdr *eip6; 1806 u_int32_t *reghdr; 1807 int rc; 1808 #ifdef MRT6DEBUG 1809 char ip6bufs[INET6_ADDRSTRLEN], ip6bufd[INET6_ADDRSTRLEN]; 1810 #endif 1811 1812 ++pim6stat.pim6s_rcv_registers; 1813 1814 if ((reg_mif_num >= nummifs) || (reg_mif_num == (mifi_t) -1)) { 1815 #ifdef MRT6DEBUG 1816 if (mrt6debug & DEBUG_PIM) 1817 log(LOG_DEBUG, 1818 "pim6_input: register mif not set: %d\n", 1819 reg_mif_num); 1820 #endif 1821 m_freem(m); 1822 return (IPPROTO_DONE); 1823 } 1824 1825 reghdr = (u_int32_t *)(pim + 1); 1826 1827 if ((ntohl(*reghdr) & PIM_NULL_REGISTER)) 1828 goto pim6_input_to_daemon; 1829 1830 /* 1831 * Validate length 1832 */ 1833 if (pimlen < PIM6_REG_MINLEN) { 1834 ++pim6stat.pim6s_rcv_tooshort; 1835 ++pim6stat.pim6s_rcv_badregisters; 1836 #ifdef MRT6DEBUG 1837 log(LOG_ERR, 1838 "pim6_input: register packet size too " 1839 "small %d from %s\n", 1840 pimlen, ip6_sprintf(ip6bufs, &ip6->ip6_src)); 1841 #endif 1842 m_freem(m); 1843 return (IPPROTO_DONE); 1844 } 1845 1846 eip6 = (struct ip6_hdr *) (reghdr + 1); 1847 #ifdef MRT6DEBUG 1848 if (mrt6debug & DEBUG_PIM) 1849 log(LOG_DEBUG, 1850 "pim6_input[register], eip6: %s -> %s, " 1851 "eip6 plen %d\n", 1852 ip6_sprintf(ip6bufs, &eip6->ip6_src), 1853 ip6_sprintf(ip6bufd, &eip6->ip6_dst), 1854 ntohs(eip6->ip6_plen)); 1855 #endif 1856 1857 /* verify the version number of the inner packet */ 1858 if ((eip6->ip6_vfc & IPV6_VERSION_MASK) != IPV6_VERSION) { 1859 ++pim6stat.pim6s_rcv_badregisters; 1860 #ifdef MRT6DEBUG 1861 log(LOG_DEBUG, "pim6_input: invalid IP version (%d) " 1862 "of the inner packet\n", 1863 (eip6->ip6_vfc & IPV6_VERSION)); 1864 #endif 1865 m_freem(m); 1866 return (IPPROTO_NONE); 1867 } 1868 1869 /* verify the inner packet is destined to a mcast group */ 1870 if (!IN6_IS_ADDR_MULTICAST(&eip6->ip6_dst)) { 1871 ++pim6stat.pim6s_rcv_badregisters; 1872 #ifdef MRT6DEBUG 1873 if (mrt6debug & DEBUG_PIM) 1874 log(LOG_DEBUG, 1875 "pim6_input: inner packet of register " 1876 "is not multicast %s\n", 1877 ip6_sprintf(ip6bufd, &eip6->ip6_dst)); 1878 #endif 1879 m_freem(m); 1880 return (IPPROTO_DONE); 1881 } 1882 1883 /* 1884 * make a copy of the whole header to pass to the daemon later. 1885 */ 1886 mcp = m_copy(m, 0, off + PIM6_REG_MINLEN); 1887 if (mcp == NULL) { 1888 #ifdef MRT6DEBUG 1889 log(LOG_ERR, 1890 "pim6_input: pim register: " 1891 "could not copy register head\n"); 1892 #endif 1893 m_freem(m); 1894 return (IPPROTO_DONE); 1895 } 1896 1897 /* 1898 * forward the inner ip6 packet; point m_data at the inner ip6. 1899 */ 1900 m_adj(m, off + PIM_MINLEN); 1901 #ifdef MRT6DEBUG 1902 if (mrt6debug & DEBUG_PIM) { 1903 log(LOG_DEBUG, 1904 "pim6_input: forwarding decapsulated register: " 1905 "src %s, dst %s, mif %d\n", 1906 ip6_sprintf(ip6bufs, &eip6->ip6_src), 1907 ip6_sprintf(ip6bufd, &eip6->ip6_dst), 1908 reg_mif_num); 1909 } 1910 #endif 1911 1912 rc = if_simloop(mif6table[reg_mif_num].m6_ifp, m, 1913 dst.sin6_family, 0); 1914 1915 /* prepare the register head to send to the mrouting daemon */ 1916 m = mcp; 1917 } 1918 1919 /* 1920 * Pass the PIM message up to the daemon; if it is a register message 1921 * pass the 'head' only up to the daemon. This includes the 1922 * encapsulator ip6 header, pim header, register header and the 1923 * encapsulated ip6 header. 1924 */ 1925 pim6_input_to_daemon: 1926 rip6_input(&m, offp, proto); 1927 return (IPPROTO_DONE); 1928 } 1929