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/raw_cb.h> 105 #include <net/route.h> 106 107 #include <netinet/in.h> 108 #include <netinet/in_var.h> 109 #include <netinet/icmp6.h> 110 111 #include <netinet/ip6.h> 112 #include <netinet6/ip6_var.h> 113 #include <netinet6/scope6_var.h> 114 #include <netinet6/nd6.h> 115 #include <netinet6/ip6_mroute.h> 116 #include <netinet6/pim6.h> 117 #include <netinet6/pim6_var.h> 118 119 #include <net/net_osdep.h> 120 121 static MALLOC_DEFINE(M_MRTABLE6, "mf6c", "multicast forwarding cache entry"); 122 123 #define M_HASCL(m) ((m)->m_flags & M_EXT) 124 125 static int ip6_mdq __P((struct mbuf *, struct ifnet *, struct mf6c *)); 126 static void phyint_send __P((struct ip6_hdr *, struct mif6 *, struct mbuf *)); 127 128 static int set_pim6 __P((int *)); 129 static int socket_send __P((struct socket *, struct mbuf *, 130 struct sockaddr_in6 *)); 131 static int register_send __P((struct ip6_hdr *, struct mif6 *, 132 struct mbuf *)); 133 134 /* 135 * Globals. All but ip6_mrouter, ip6_mrtproto and mrt6stat could be static, 136 * except for netstat or debugging purposes. 137 */ 138 struct socket *ip6_mrouter = NULL; 139 int ip6_mrouter_ver = 0; 140 int ip6_mrtproto = IPPROTO_PIM; /* for netstat only */ 141 struct mrt6stat mrt6stat; 142 143 #define NO_RTE_FOUND 0x1 144 #define RTE_FOUND 0x2 145 146 struct mf6c *mf6ctable[MF6CTBLSIZ]; 147 u_char n6expire[MF6CTBLSIZ]; 148 static struct mif6 mif6table[MAXMIFS]; 149 #ifdef MRT6DEBUG 150 u_int mrt6debug = 0; /* debug level */ 151 #define DEBUG_MFC 0x02 152 #define DEBUG_FORWARD 0x04 153 #define DEBUG_EXPIRE 0x08 154 #define DEBUG_XMIT 0x10 155 #define DEBUG_REG 0x20 156 #define DEBUG_PIM 0x40 157 #endif 158 159 static void expire_upcalls __P((void *)); 160 #define EXPIRE_TIMEOUT (hz / 4) /* 4x / second */ 161 #define UPCALL_EXPIRE 6 /* number of timeouts */ 162 163 #ifdef INET 164 #ifdef MROUTING 165 extern struct socket *ip_mrouter; 166 #endif 167 #endif 168 169 /* 170 * 'Interfaces' associated with decapsulator (so we can tell 171 * packets that went through it from ones that get reflected 172 * by a broken gateway). These interfaces are never linked into 173 * the system ifnet list & no routes point to them. I.e., packets 174 * can't be sent this way. They only exist as a placeholder for 175 * multicast source verification. 176 */ 177 struct ifnet multicast_register_if; 178 179 #define ENCAP_HOPS 64 180 181 /* 182 * Private variables. 183 */ 184 static mifi_t nummifs = 0; 185 static mifi_t reg_mif_num = (mifi_t)-1; 186 187 static struct pim6stat pim6stat; 188 static int pim6; 189 190 /* 191 * Hash function for a source, group entry 192 */ 193 #define MF6CHASH(a, g) MF6CHASHMOD((a).s6_addr32[0] ^ (a).s6_addr32[1] ^ \ 194 (a).s6_addr32[2] ^ (a).s6_addr32[3] ^ \ 195 (g).s6_addr32[0] ^ (g).s6_addr32[1] ^ \ 196 (g).s6_addr32[2] ^ (g).s6_addr32[3]) 197 198 /* 199 * Find a route for a given origin IPv6 address and Multicast group address. 200 * Quality of service parameter to be added in the future!!! 201 */ 202 203 #define MF6CFIND(o, g, rt) do { \ 204 struct mf6c *_rt = mf6ctable[MF6CHASH(o,g)]; \ 205 rt = NULL; \ 206 mrt6stat.mrt6s_mfc_lookups++; \ 207 while (_rt) { \ 208 if (IN6_ARE_ADDR_EQUAL(&_rt->mf6c_origin.sin6_addr, &(o)) && \ 209 IN6_ARE_ADDR_EQUAL(&_rt->mf6c_mcastgrp.sin6_addr, &(g)) && \ 210 (_rt->mf6c_stall == NULL)) { \ 211 rt = _rt; \ 212 break; \ 213 } \ 214 _rt = _rt->mf6c_next; \ 215 } \ 216 if (rt == NULL) { \ 217 mrt6stat.mrt6s_mfc_misses++; \ 218 } \ 219 } while (/*CONSTCOND*/ 0) 220 221 /* 222 * Macros to compute elapsed time efficiently 223 * Borrowed from Van Jacobson's scheduling code 224 */ 225 #define TV_DELTA(a, b, delta) do { \ 226 int xxs; \ 227 \ 228 delta = (a).tv_usec - (b).tv_usec; \ 229 if ((xxs = (a).tv_sec - (b).tv_sec)) { \ 230 switch (xxs) { \ 231 case 2: \ 232 delta += 1000000; \ 233 /* FALLTHROUGH */ \ 234 case 1: \ 235 delta += 1000000; \ 236 break; \ 237 default: \ 238 delta += (1000000 * xxs); \ 239 } \ 240 } \ 241 } while (/*CONSTCOND*/ 0) 242 243 #define TV_LT(a, b) (((a).tv_usec < (b).tv_usec && \ 244 (a).tv_sec <= (b).tv_sec) || (a).tv_sec < (b).tv_sec) 245 246 #ifdef UPCALL_TIMING 247 #define UPCALL_MAX 50 248 u_long upcall_data[UPCALL_MAX + 1]; 249 static void collate(); 250 #endif /* UPCALL_TIMING */ 251 252 static int get_sg_cnt __P((struct sioc_sg_req6 *)); 253 static int get_mif6_cnt __P((struct sioc_mif_req6 *)); 254 static int ip6_mrouter_init __P((struct socket *, int, int)); 255 static int add_m6if __P((struct mif6ctl *)); 256 static int del_m6if __P((mifi_t *)); 257 static int add_m6fc __P((struct mf6cctl *)); 258 static int del_m6fc __P((struct mf6cctl *)); 259 260 static struct callout expire_upcalls_ch; 261 262 /* 263 * Handle MRT setsockopt commands to modify the multicast routing tables. 264 */ 265 int 266 ip6_mrouter_set(so, sopt) 267 struct socket *so; 268 struct sockopt *sopt; 269 { 270 int error = 0; 271 int optval; 272 struct mif6ctl mifc; 273 struct mf6cctl mfcc; 274 mifi_t mifi; 275 276 if (so != ip6_mrouter && sopt->sopt_name != MRT6_INIT) 277 return (EACCES); 278 279 switch (sopt->sopt_name) { 280 case MRT6_INIT: 281 #ifdef MRT6_OINIT 282 case MRT6_OINIT: 283 #endif 284 error = sooptcopyin(sopt, &optval, sizeof(optval), 285 sizeof(optval)); 286 if (error) 287 break; 288 error = ip6_mrouter_init(so, optval, sopt->sopt_name); 289 break; 290 case MRT6_DONE: 291 error = ip6_mrouter_done(); 292 break; 293 case MRT6_ADD_MIF: 294 error = sooptcopyin(sopt, &mifc, sizeof(mifc), sizeof(mifc)); 295 if (error) 296 break; 297 error = add_m6if(&mifc); 298 break; 299 case MRT6_ADD_MFC: 300 error = sooptcopyin(sopt, &mfcc, sizeof(mfcc), sizeof(mfcc)); 301 if (error) 302 break; 303 error = add_m6fc(&mfcc); 304 break; 305 case MRT6_DEL_MFC: 306 error = sooptcopyin(sopt, &mfcc, sizeof(mfcc), sizeof(mfcc)); 307 if (error) 308 break; 309 error = del_m6fc(&mfcc); 310 break; 311 case MRT6_DEL_MIF: 312 error = sooptcopyin(sopt, &mifi, sizeof(mifi), sizeof(mifi)); 313 if (error) 314 break; 315 error = del_m6if(&mifi); 316 break; 317 case MRT6_PIM: 318 error = sooptcopyin(sopt, &optval, sizeof(optval), 319 sizeof(optval)); 320 if (error) 321 break; 322 error = set_pim6(&optval); 323 break; 324 default: 325 error = EOPNOTSUPP; 326 break; 327 } 328 329 return (error); 330 } 331 332 /* 333 * Handle MRT getsockopt commands 334 */ 335 int 336 ip6_mrouter_get(so, sopt) 337 struct socket *so; 338 struct sockopt *sopt; 339 { 340 int error = 0; 341 342 if (so != ip6_mrouter) 343 return (EACCES); 344 345 switch (sopt->sopt_name) { 346 case MRT6_PIM: 347 error = sooptcopyout(sopt, &pim6, sizeof(pim6)); 348 break; 349 } 350 return (error); 351 } 352 353 /* 354 * Handle ioctl commands to obtain information from the cache 355 */ 356 int 357 mrt6_ioctl(cmd, data) 358 int cmd; 359 caddr_t data; 360 { 361 switch (cmd) { 362 case SIOCGETSGCNT_IN6: 363 return (get_sg_cnt((struct sioc_sg_req6 *)data)); 364 case SIOCGETMIFCNT_IN6: 365 return (get_mif6_cnt((struct sioc_mif_req6 *)data)); 366 default: 367 return (EINVAL); 368 } 369 } 370 371 /* 372 * returns the packet, byte, rpf-failure count for the source group provided 373 */ 374 static int 375 get_sg_cnt(req) 376 struct sioc_sg_req6 *req; 377 { 378 struct mf6c *rt; 379 int s; 380 381 s = splnet(); 382 MF6CFIND(req->src.sin6_addr, req->grp.sin6_addr, rt); 383 splx(s); 384 if (rt != NULL) { 385 req->pktcnt = rt->mf6c_pkt_cnt; 386 req->bytecnt = rt->mf6c_byte_cnt; 387 req->wrong_if = rt->mf6c_wrong_if; 388 } else 389 return (ESRCH); 390 #if 0 391 req->pktcnt = req->bytecnt = req->wrong_if = 0xffffffff; 392 #endif 393 394 return (0); 395 } 396 397 /* 398 * returns the input and output packet and byte counts on the mif provided 399 */ 400 static int 401 get_mif6_cnt(req) 402 struct sioc_mif_req6 *req; 403 { 404 mifi_t mifi = req->mifi; 405 406 if (mifi >= nummifs) 407 return (EINVAL); 408 409 req->icount = mif6table[mifi].m6_pkt_in; 410 req->ocount = mif6table[mifi].m6_pkt_out; 411 req->ibytes = mif6table[mifi].m6_bytes_in; 412 req->obytes = mif6table[mifi].m6_bytes_out; 413 414 return (0); 415 } 416 417 static int 418 set_pim6(i) 419 int *i; 420 { 421 if ((*i != 1) && (*i != 0)) 422 return (EINVAL); 423 424 pim6 = *i; 425 426 return (0); 427 } 428 429 /* 430 * Enable multicast routing 431 */ 432 static int 433 ip6_mrouter_init(so, v, cmd) 434 struct socket *so; 435 int v; 436 int cmd; 437 { 438 #ifdef MRT6DEBUG 439 if (mrt6debug) 440 log(LOG_DEBUG, 441 "ip6_mrouter_init: so_type = %d, pr_protocol = %d\n", 442 so->so_type, so->so_proto->pr_protocol); 443 #endif 444 445 if (so->so_type != SOCK_RAW || 446 so->so_proto->pr_protocol != IPPROTO_ICMPV6) 447 return (EOPNOTSUPP); 448 449 if (v != 1) 450 return (ENOPROTOOPT); 451 452 if (ip6_mrouter != NULL) 453 return (EADDRINUSE); 454 455 ip6_mrouter = so; 456 ip6_mrouter_ver = cmd; 457 458 bzero((caddr_t)mf6ctable, sizeof(mf6ctable)); 459 bzero((caddr_t)n6expire, sizeof(n6expire)); 460 461 pim6 = 0;/* used for stubbing out/in pim stuff */ 462 463 callout_init(&expire_upcalls_ch, 0); 464 callout_reset(&expire_upcalls_ch, EXPIRE_TIMEOUT, 465 expire_upcalls, NULL); 466 467 #ifdef MRT6DEBUG 468 if (mrt6debug) 469 log(LOG_DEBUG, "ip6_mrouter_init\n"); 470 #endif 471 472 return (0); 473 } 474 475 /* 476 * Disable multicast routing 477 */ 478 int 479 ip6_mrouter_done() 480 { 481 mifi_t mifi; 482 int i; 483 struct ifnet *ifp; 484 struct in6_ifreq ifr; 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 ifr.ifr_addr.sin6_family = AF_INET6; 511 ifr.ifr_addr.sin6_addr = in6addr_any; 512 ifp = mif6table[mifi].m6_ifp; 513 (*ifp->if_ioctl)(ifp, SIOCDELMULTI, 514 (caddr_t)&ifr); 515 } 516 } 517 } 518 #ifdef notyet 519 bzero((caddr_t)qtable, sizeof(qtable)); 520 bzero((caddr_t)tbftable, sizeof(tbftable)); 521 #endif 522 bzero((caddr_t)mif6table, sizeof(mif6table)); 523 nummifs = 0; 524 525 pim6 = 0; /* used to stub out/in pim specific code */ 526 527 callout_stop(&expire_upcalls_ch); 528 529 /* 530 * Free all multicast forwarding cache entries. 531 */ 532 for (i = 0; i < MF6CTBLSIZ; i++) { 533 rt = mf6ctable[i]; 534 while (rt) { 535 struct mf6c *frt; 536 537 for (rte = rt->mf6c_stall; rte != NULL; ) { 538 struct rtdetq *n = rte->next; 539 540 m_free(rte->m); 541 free(rte, M_MRTABLE6); 542 rte = n; 543 } 544 frt = rt; 545 rt = rt->mf6c_next; 546 free(frt, M_MRTABLE6); 547 } 548 } 549 550 bzero((caddr_t)mf6ctable, sizeof(mf6ctable)); 551 552 /* 553 * Reset de-encapsulation cache 554 */ 555 reg_mif_num = -1; 556 557 ip6_mrouter = NULL; 558 ip6_mrouter_ver = 0; 559 560 splx(s); 561 562 #ifdef MRT6DEBUG 563 if (mrt6debug) 564 log(LOG_DEBUG, "ip6_mrouter_done\n"); 565 #endif 566 567 return (0); 568 } 569 570 static struct sockaddr_in6 sin6 = { sizeof(sin6), AF_INET6 }; 571 572 /* 573 * Add a mif to the mif table 574 */ 575 static int 576 add_m6if(mifcp) 577 struct mif6ctl *mifcp; 578 { 579 struct mif6 *mifp; 580 struct ifnet *ifp; 581 int error, s; 582 #ifdef notyet 583 struct tbf *m_tbf = tbftable + mifcp->mif6c_mifi; 584 #endif 585 586 if (mifcp->mif6c_mifi >= MAXMIFS) 587 return (EINVAL); 588 mifp = mif6table + mifcp->mif6c_mifi; 589 if (mifp->m6_ifp) 590 return (EADDRINUSE); /* XXX: is it appropriate? */ 591 if (mifcp->mif6c_pifi == 0 || mifcp->mif6c_pifi > if_index) 592 return (ENXIO); 593 ifp = ifnet_byindex(mifcp->mif6c_pifi); 594 595 if (mifcp->mif6c_flags & MIFF_REGISTER) { 596 if (reg_mif_num == (mifi_t)-1) { 597 strlcpy(multicast_register_if.if_xname, "register_mif", 598 IFNAMSIZ); 599 multicast_register_if.if_flags |= IFF_LOOPBACK; 600 multicast_register_if.if_index = mifcp->mif6c_mifi; 601 reg_mif_num = mifcp->mif6c_mifi; 602 } 603 604 ifp = &multicast_register_if; 605 606 } /* if REGISTER */ 607 else { 608 /* Make sure the interface supports multicast */ 609 if ((ifp->if_flags & IFF_MULTICAST) == 0) 610 return (EOPNOTSUPP); 611 612 s = splnet(); 613 error = if_allmulti(ifp, 1); 614 splx(s); 615 if (error) 616 return (error); 617 } 618 619 s = splnet(); 620 mifp->m6_flags = mifcp->mif6c_flags; 621 mifp->m6_ifp = ifp; 622 #ifdef notyet 623 /* scaling up here allows division by 1024 in critical code */ 624 mifp->m6_rate_limit = mifcp->mif6c_rate_limit * 1024 / 1000; 625 #endif 626 /* initialize per mif pkt counters */ 627 mifp->m6_pkt_in = 0; 628 mifp->m6_pkt_out = 0; 629 mifp->m6_bytes_in = 0; 630 mifp->m6_bytes_out = 0; 631 splx(s); 632 633 /* Adjust nummifs up if the mifi is higher than nummifs */ 634 if (nummifs <= mifcp->mif6c_mifi) 635 nummifs = mifcp->mif6c_mifi + 1; 636 637 #ifdef MRT6DEBUG 638 if (mrt6debug) 639 log(LOG_DEBUG, 640 "add_mif #%d, phyint %s\n", 641 mifcp->mif6c_mifi, 642 ifp->if_xname); 643 #endif 644 645 return (0); 646 } 647 648 /* 649 * Delete a mif from the mif table 650 */ 651 static int 652 del_m6if(mifip) 653 mifi_t *mifip; 654 { 655 struct mif6 *mifp = mif6table + *mifip; 656 mifi_t mifi; 657 struct ifnet *ifp; 658 int s; 659 660 if (*mifip >= nummifs) 661 return (EINVAL); 662 if (mifp->m6_ifp == NULL) 663 return (EINVAL); 664 665 s = splnet(); 666 667 if (!(mifp->m6_flags & MIFF_REGISTER)) { 668 /* 669 * XXX: what if there is yet IPv4 multicast daemon 670 * using the interface? 671 */ 672 ifp = mifp->m6_ifp; 673 674 if_allmulti(ifp, 0); 675 } 676 677 #ifdef notyet 678 bzero((caddr_t)qtable[*mifip], sizeof(qtable[*mifip])); 679 bzero((caddr_t)mifp->m6_tbf, sizeof(*(mifp->m6_tbf))); 680 #endif 681 bzero((caddr_t)mifp, sizeof(*mifp)); 682 683 /* Adjust nummifs down */ 684 for (mifi = nummifs; mifi > 0; mifi--) 685 if (mif6table[mifi - 1].m6_ifp) 686 break; 687 nummifs = mifi; 688 689 splx(s); 690 691 #ifdef MRT6DEBUG 692 if (mrt6debug) 693 log(LOG_DEBUG, "del_m6if %d, nummifs %d\n", *mifip, nummifs); 694 #endif 695 696 return (0); 697 } 698 699 /* 700 * Add an mfc entry 701 */ 702 static int 703 add_m6fc(mfccp) 704 struct mf6cctl *mfccp; 705 { 706 struct mf6c *rt; 707 u_long hash; 708 struct rtdetq *rte; 709 u_short nstl; 710 int s; 711 712 MF6CFIND(mfccp->mf6cc_origin.sin6_addr, 713 mfccp->mf6cc_mcastgrp.sin6_addr, rt); 714 715 /* If an entry already exists, just update the fields */ 716 if (rt) { 717 #ifdef MRT6DEBUG 718 if (mrt6debug & DEBUG_MFC) 719 log(LOG_DEBUG, 720 "add_m6fc no upcall h %d o %s g %s p %x\n", 721 ip6_sprintf(&mfccp->mf6cc_origin.sin6_addr), 722 ip6_sprintf(&mfccp->mf6cc_mcastgrp.sin6_addr), 723 mfccp->mf6cc_parent); 724 #endif 725 726 s = splnet(); 727 rt->mf6c_parent = mfccp->mf6cc_parent; 728 rt->mf6c_ifset = mfccp->mf6cc_ifset; 729 splx(s); 730 return (0); 731 } 732 733 /* 734 * Find the entry for which the upcall was made and update 735 */ 736 s = splnet(); 737 hash = MF6CHASH(mfccp->mf6cc_origin.sin6_addr, 738 mfccp->mf6cc_mcastgrp.sin6_addr); 739 for (rt = mf6ctable[hash], nstl = 0; rt; rt = rt->mf6c_next) { 740 if (IN6_ARE_ADDR_EQUAL(&rt->mf6c_origin.sin6_addr, 741 &mfccp->mf6cc_origin.sin6_addr) && 742 IN6_ARE_ADDR_EQUAL(&rt->mf6c_mcastgrp.sin6_addr, 743 &mfccp->mf6cc_mcastgrp.sin6_addr) && 744 (rt->mf6c_stall != NULL)) { 745 746 if (nstl++) 747 log(LOG_ERR, 748 "add_m6fc: %s o %s g %s p %x dbx %p\n", 749 "multiple kernel entries", 750 ip6_sprintf(&mfccp->mf6cc_origin.sin6_addr), 751 ip6_sprintf(&mfccp->mf6cc_mcastgrp.sin6_addr), 752 mfccp->mf6cc_parent, rt->mf6c_stall); 753 754 #ifdef MRT6DEBUG 755 if (mrt6debug & DEBUG_MFC) 756 log(LOG_DEBUG, 757 "add_m6fc o %s g %s p %x dbg %x\n", 758 ip6_sprintf(&mfccp->mf6cc_origin.sin6_addr), 759 ip6_sprintf(&mfccp->mf6cc_mcastgrp.sin6_addr), 760 mfccp->mf6cc_parent, rt->mf6c_stall); 761 #endif 762 763 rt->mf6c_origin = mfccp->mf6cc_origin; 764 rt->mf6c_mcastgrp = mfccp->mf6cc_mcastgrp; 765 rt->mf6c_parent = mfccp->mf6cc_parent; 766 rt->mf6c_ifset = mfccp->mf6cc_ifset; 767 /* initialize pkt counters per src-grp */ 768 rt->mf6c_pkt_cnt = 0; 769 rt->mf6c_byte_cnt = 0; 770 rt->mf6c_wrong_if = 0; 771 772 rt->mf6c_expire = 0; /* Don't clean this guy up */ 773 n6expire[hash]--; 774 775 /* free packets Qed at the end of this entry */ 776 for (rte = rt->mf6c_stall; rte != NULL; ) { 777 struct rtdetq *n = rte->next; 778 ip6_mdq(rte->m, rte->ifp, rt); 779 m_freem(rte->m); 780 #ifdef UPCALL_TIMING 781 collate(&(rte->t)); 782 #endif /* UPCALL_TIMING */ 783 free(rte, M_MRTABLE6); 784 rte = n; 785 } 786 rt->mf6c_stall = NULL; 787 } 788 } 789 790 /* 791 * It is possible that an entry is being inserted without an upcall 792 */ 793 if (nstl == 0) { 794 #ifdef MRT6DEBUG 795 if (mrt6debug & DEBUG_MFC) 796 log(LOG_DEBUG, 797 "add_mfc no upcall h %d o %s g %s p %x\n", 798 hash, 799 ip6_sprintf(&mfccp->mf6cc_origin.sin6_addr), 800 ip6_sprintf(&mfccp->mf6cc_mcastgrp.sin6_addr), 801 mfccp->mf6cc_parent); 802 #endif 803 804 for (rt = mf6ctable[hash]; rt; rt = rt->mf6c_next) { 805 806 if (IN6_ARE_ADDR_EQUAL(&rt->mf6c_origin.sin6_addr, 807 &mfccp->mf6cc_origin.sin6_addr)&& 808 IN6_ARE_ADDR_EQUAL(&rt->mf6c_mcastgrp.sin6_addr, 809 &mfccp->mf6cc_mcastgrp.sin6_addr)) { 810 811 rt->mf6c_origin = mfccp->mf6cc_origin; 812 rt->mf6c_mcastgrp = mfccp->mf6cc_mcastgrp; 813 rt->mf6c_parent = mfccp->mf6cc_parent; 814 rt->mf6c_ifset = mfccp->mf6cc_ifset; 815 /* initialize pkt counters per src-grp */ 816 rt->mf6c_pkt_cnt = 0; 817 rt->mf6c_byte_cnt = 0; 818 rt->mf6c_wrong_if = 0; 819 820 if (rt->mf6c_expire) 821 n6expire[hash]--; 822 rt->mf6c_expire = 0; 823 } 824 } 825 if (rt == NULL) { 826 /* no upcall, so make a new entry */ 827 rt = (struct mf6c *)malloc(sizeof(*rt), M_MRTABLE6, 828 M_NOWAIT); 829 if (rt == NULL) { 830 splx(s); 831 return (ENOBUFS); 832 } 833 834 /* insert new entry at head of hash chain */ 835 rt->mf6c_origin = mfccp->mf6cc_origin; 836 rt->mf6c_mcastgrp = mfccp->mf6cc_mcastgrp; 837 rt->mf6c_parent = mfccp->mf6cc_parent; 838 rt->mf6c_ifset = mfccp->mf6cc_ifset; 839 /* initialize pkt counters per src-grp */ 840 rt->mf6c_pkt_cnt = 0; 841 rt->mf6c_byte_cnt = 0; 842 rt->mf6c_wrong_if = 0; 843 rt->mf6c_expire = 0; 844 rt->mf6c_stall = NULL; 845 846 /* link into table */ 847 rt->mf6c_next = mf6ctable[hash]; 848 mf6ctable[hash] = rt; 849 } 850 } 851 splx(s); 852 return (0); 853 } 854 855 #ifdef UPCALL_TIMING 856 /* 857 * collect delay statistics on the upcalls 858 */ 859 static void 860 collate(t) 861 struct timeval *t; 862 { 863 u_long d; 864 struct timeval tp; 865 u_long delta; 866 867 GET_TIME(tp); 868 869 if (TV_LT(*t, tp)) 870 { 871 TV_DELTA(tp, *t, delta); 872 873 d = delta >> 10; 874 if (d > UPCALL_MAX) 875 d = UPCALL_MAX; 876 877 ++upcall_data[d]; 878 } 879 } 880 #endif /* UPCALL_TIMING */ 881 882 /* 883 * Delete an mfc entry 884 */ 885 static int 886 del_m6fc(mfccp) 887 struct mf6cctl *mfccp; 888 { 889 struct sockaddr_in6 origin; 890 struct sockaddr_in6 mcastgrp; 891 struct mf6c *rt; 892 struct mf6c **nptr; 893 u_long hash; 894 int s; 895 896 origin = mfccp->mf6cc_origin; 897 mcastgrp = mfccp->mf6cc_mcastgrp; 898 hash = MF6CHASH(origin.sin6_addr, mcastgrp.sin6_addr); 899 900 #ifdef MRT6DEBUG 901 if (mrt6debug & DEBUG_MFC) 902 log(LOG_DEBUG,"del_m6fc orig %s mcastgrp %s\n", 903 ip6_sprintf(&origin.sin6_addr), 904 ip6_sprintf(&mcastgrp.sin6_addr)); 905 #endif 906 907 s = splnet(); 908 909 nptr = &mf6ctable[hash]; 910 while ((rt = *nptr) != NULL) { 911 if (IN6_ARE_ADDR_EQUAL(&origin.sin6_addr, 912 &rt->mf6c_origin.sin6_addr) && 913 IN6_ARE_ADDR_EQUAL(&mcastgrp.sin6_addr, 914 &rt->mf6c_mcastgrp.sin6_addr) && 915 rt->mf6c_stall == NULL) 916 break; 917 918 nptr = &rt->mf6c_next; 919 } 920 if (rt == NULL) { 921 splx(s); 922 return (EADDRNOTAVAIL); 923 } 924 925 *nptr = rt->mf6c_next; 926 free(rt, M_MRTABLE6); 927 928 splx(s); 929 930 return (0); 931 } 932 933 static int 934 socket_send(s, mm, src) 935 struct socket *s; 936 struct mbuf *mm; 937 struct sockaddr_in6 *src; 938 { 939 if (s) { 940 if (sbappendaddr(&s->so_rcv, 941 (struct sockaddr *)src, 942 mm, (struct mbuf *)0) != 0) { 943 sorwakeup(s); 944 return (0); 945 } 946 } 947 m_freem(mm); 948 return (-1); 949 } 950 951 /* 952 * IPv6 multicast forwarding function. This function assumes that the packet 953 * pointed to by "ip6" has arrived on (or is about to be sent to) the interface 954 * pointed to by "ifp", and the packet is to be relayed to other networks 955 * that have members of the packet's destination IPv6 multicast group. 956 * 957 * The packet is returned unscathed to the caller, unless it is 958 * erroneous, in which case a non-zero return value tells the caller to 959 * discard it. 960 */ 961 962 int 963 ip6_mforward(ip6, ifp, m) 964 struct ip6_hdr *ip6; 965 struct ifnet *ifp; 966 struct mbuf *m; 967 { 968 struct mf6c *rt; 969 struct mif6 *mifp; 970 struct mbuf *mm; 971 int s; 972 mifi_t mifi; 973 974 #ifdef MRT6DEBUG 975 if (mrt6debug & DEBUG_FORWARD) 976 log(LOG_DEBUG, "ip6_mforward: src %s, dst %s, ifindex %d\n", 977 ip6_sprintf(&ip6->ip6_src), ip6_sprintf(&ip6->ip6_dst), 978 ifp->if_index); 979 #endif 980 981 /* 982 * Don't forward a packet with Hop limit of zero or one, 983 * or a packet destined to a local-only group. 984 */ 985 if (ip6->ip6_hlim <= 1 || IN6_IS_ADDR_MC_INTFACELOCAL(&ip6->ip6_dst) || 986 IN6_IS_ADDR_MC_LINKLOCAL(&ip6->ip6_dst)) 987 return (0); 988 ip6->ip6_hlim--; 989 990 /* 991 * Source address check: do not forward packets with unspecified 992 * source. It was discussed in July 2000, on ipngwg mailing list. 993 * This is rather more serious than unicast cases, because some 994 * MLD packets can be sent with the unspecified source address 995 * (although such packets must normally set 1 to the hop limit field). 996 */ 997 if (IN6_IS_ADDR_UNSPECIFIED(&ip6->ip6_src)) { 998 ip6stat.ip6s_cantforward++; 999 if (ip6_log_time + ip6_log_interval < time_second) { 1000 ip6_log_time = time_second; 1001 log(LOG_DEBUG, 1002 "cannot forward " 1003 "from %s to %s nxt %d received on %s\n", 1004 ip6_sprintf(&ip6->ip6_src), 1005 ip6_sprintf(&ip6->ip6_dst), 1006 ip6->ip6_nxt, 1007 if_name(m->m_pkthdr.rcvif)); 1008 } 1009 return (0); 1010 } 1011 1012 /* 1013 * Determine forwarding mifs from the forwarding cache table 1014 */ 1015 s = splnet(); 1016 MF6CFIND(ip6->ip6_src, ip6->ip6_dst, rt); 1017 1018 /* Entry exists, so forward if necessary */ 1019 if (rt) { 1020 splx(s); 1021 return (ip6_mdq(m, ifp, rt)); 1022 } else { 1023 /* 1024 * If we don't have a route for packet's origin, 1025 * Make a copy of the packet & 1026 * send message to routing daemon 1027 */ 1028 1029 struct mbuf *mb0; 1030 struct rtdetq *rte; 1031 u_long hash; 1032 /* int i, npkts;*/ 1033 #ifdef UPCALL_TIMING 1034 struct timeval tp; 1035 1036 GET_TIME(tp); 1037 #endif /* UPCALL_TIMING */ 1038 1039 mrt6stat.mrt6s_no_route++; 1040 #ifdef MRT6DEBUG 1041 if (mrt6debug & (DEBUG_FORWARD | DEBUG_MFC)) 1042 log(LOG_DEBUG, "ip6_mforward: no rte s %s g %s\n", 1043 ip6_sprintf(&ip6->ip6_src), 1044 ip6_sprintf(&ip6->ip6_dst)); 1045 #endif 1046 1047 /* 1048 * Allocate mbufs early so that we don't do extra work if we 1049 * are just going to fail anyway. 1050 */ 1051 rte = (struct rtdetq *)malloc(sizeof(*rte), M_MRTABLE6, 1052 M_NOWAIT); 1053 if (rte == NULL) { 1054 splx(s); 1055 return (ENOBUFS); 1056 } 1057 mb0 = m_copy(m, 0, M_COPYALL); 1058 /* 1059 * Pullup packet header if needed before storing it, 1060 * as other references may modify it in the meantime. 1061 */ 1062 if (mb0 && 1063 (M_HASCL(mb0) || mb0->m_len < sizeof(struct ip6_hdr))) 1064 mb0 = m_pullup(mb0, sizeof(struct ip6_hdr)); 1065 if (mb0 == NULL) { 1066 free(rte, M_MRTABLE6); 1067 splx(s); 1068 return (ENOBUFS); 1069 } 1070 1071 /* is there an upcall waiting for this packet? */ 1072 hash = MF6CHASH(ip6->ip6_src, ip6->ip6_dst); 1073 for (rt = mf6ctable[hash]; rt; rt = rt->mf6c_next) { 1074 if (IN6_ARE_ADDR_EQUAL(&ip6->ip6_src, 1075 &rt->mf6c_origin.sin6_addr) && 1076 IN6_ARE_ADDR_EQUAL(&ip6->ip6_dst, 1077 &rt->mf6c_mcastgrp.sin6_addr) && 1078 (rt->mf6c_stall != NULL)) 1079 break; 1080 } 1081 1082 if (rt == NULL) { 1083 struct mrt6msg *im; 1084 #ifdef MRT6_OINIT 1085 struct omrt6msg *oim; 1086 #endif 1087 1088 /* no upcall, so make a new entry */ 1089 rt = (struct mf6c *)malloc(sizeof(*rt), M_MRTABLE6, 1090 M_NOWAIT); 1091 if (rt == NULL) { 1092 free(rte, M_MRTABLE6); 1093 m_freem(mb0); 1094 splx(s); 1095 return (ENOBUFS); 1096 } 1097 /* 1098 * Make a copy of the header to send to the user 1099 * level process 1100 */ 1101 mm = m_copy(mb0, 0, sizeof(struct ip6_hdr)); 1102 1103 if (mm == NULL) { 1104 free(rte, M_MRTABLE6); 1105 m_freem(mb0); 1106 free(rt, M_MRTABLE6); 1107 splx(s); 1108 return (ENOBUFS); 1109 } 1110 1111 /* 1112 * Send message to routing daemon 1113 */ 1114 sin6.sin6_addr = ip6->ip6_src; 1115 1116 im = NULL; 1117 #ifdef MRT6_OINIT 1118 oim = NULL; 1119 #endif 1120 switch (ip6_mrouter_ver) { 1121 #ifdef MRT6_OINIT 1122 case MRT6_OINIT: 1123 oim = mtod(mm, struct omrt6msg *); 1124 oim->im6_msgtype = MRT6MSG_NOCACHE; 1125 oim->im6_mbz = 0; 1126 break; 1127 #endif 1128 case MRT6_INIT: 1129 im = mtod(mm, struct mrt6msg *); 1130 im->im6_msgtype = MRT6MSG_NOCACHE; 1131 im->im6_mbz = 0; 1132 break; 1133 default: 1134 free(rte, M_MRTABLE6); 1135 m_freem(mb0); 1136 free(rt, M_MRTABLE6); 1137 splx(s); 1138 return (EINVAL); 1139 } 1140 1141 #ifdef MRT6DEBUG 1142 if (mrt6debug & DEBUG_FORWARD) 1143 log(LOG_DEBUG, 1144 "getting the iif info in the kernel\n"); 1145 #endif 1146 1147 for (mifp = mif6table, mifi = 0; 1148 mifi < nummifs && mifp->m6_ifp != ifp; 1149 mifp++, mifi++) 1150 ; 1151 1152 switch (ip6_mrouter_ver) { 1153 #ifdef MRT6_OINIT 1154 case MRT6_OINIT: 1155 oim->im6_mif = mifi; 1156 break; 1157 #endif 1158 case MRT6_INIT: 1159 im->im6_mif = mifi; 1160 break; 1161 } 1162 1163 if (socket_send(ip6_mrouter, mm, &sin6) < 0) { 1164 log(LOG_WARNING, "ip6_mforward: ip6_mrouter " 1165 "socket queue full\n"); 1166 mrt6stat.mrt6s_upq_sockfull++; 1167 free(rte, M_MRTABLE6); 1168 m_freem(mb0); 1169 free(rt, M_MRTABLE6); 1170 splx(s); 1171 return (ENOBUFS); 1172 } 1173 1174 mrt6stat.mrt6s_upcalls++; 1175 1176 /* insert new entry at head of hash chain */ 1177 bzero(rt, sizeof(*rt)); 1178 rt->mf6c_origin.sin6_family = AF_INET6; 1179 rt->mf6c_origin.sin6_len = sizeof(struct sockaddr_in6); 1180 rt->mf6c_origin.sin6_addr = ip6->ip6_src; 1181 rt->mf6c_mcastgrp.sin6_family = AF_INET6; 1182 rt->mf6c_mcastgrp.sin6_len = sizeof(struct sockaddr_in6); 1183 rt->mf6c_mcastgrp.sin6_addr = ip6->ip6_dst; 1184 rt->mf6c_expire = UPCALL_EXPIRE; 1185 n6expire[hash]++; 1186 rt->mf6c_parent = MF6C_INCOMPLETE_PARENT; 1187 1188 /* link into table */ 1189 rt->mf6c_next = mf6ctable[hash]; 1190 mf6ctable[hash] = rt; 1191 /* Add this entry to the end of the queue */ 1192 rt->mf6c_stall = rte; 1193 } else { 1194 /* determine if q has overflowed */ 1195 struct rtdetq **p; 1196 int npkts = 0; 1197 1198 for (p = &rt->mf6c_stall; *p != NULL; p = &(*p)->next) 1199 if (++npkts > MAX_UPQ6) { 1200 mrt6stat.mrt6s_upq_ovflw++; 1201 free(rte, M_MRTABLE6); 1202 m_freem(mb0); 1203 splx(s); 1204 return (0); 1205 } 1206 1207 /* Add this entry to the end of the queue */ 1208 *p = rte; 1209 } 1210 1211 rte->next = NULL; 1212 rte->m = mb0; 1213 rte->ifp = ifp; 1214 #ifdef UPCALL_TIMING 1215 rte->t = tp; 1216 #endif /* UPCALL_TIMING */ 1217 1218 splx(s); 1219 1220 return (0); 1221 } 1222 } 1223 1224 /* 1225 * Clean up cache entries if upcalls are not serviced 1226 * Call from the Slow Timeout mechanism, every half second. 1227 */ 1228 static void 1229 expire_upcalls(unused) 1230 void *unused; 1231 { 1232 struct rtdetq *rte; 1233 struct mf6c *mfc, **nptr; 1234 int i; 1235 int s; 1236 1237 s = splnet(); 1238 for (i = 0; i < MF6CTBLSIZ; i++) { 1239 if (n6expire[i] == 0) 1240 continue; 1241 nptr = &mf6ctable[i]; 1242 while ((mfc = *nptr) != NULL) { 1243 rte = mfc->mf6c_stall; 1244 /* 1245 * Skip real cache entries 1246 * Make sure it wasn't marked to not expire (shouldn't happen) 1247 * If it expires now 1248 */ 1249 if (rte != NULL && 1250 mfc->mf6c_expire != 0 && 1251 --mfc->mf6c_expire == 0) { 1252 #ifdef MRT6DEBUG 1253 if (mrt6debug & DEBUG_EXPIRE) 1254 log(LOG_DEBUG, "expire_upcalls: expiring (%s %s)\n", 1255 ip6_sprintf(&mfc->mf6c_origin.sin6_addr), 1256 ip6_sprintf(&mfc->mf6c_mcastgrp.sin6_addr)); 1257 #endif 1258 /* 1259 * drop all the packets 1260 * free the mbuf with the pkt, if, timing info 1261 */ 1262 do { 1263 struct rtdetq *n = rte->next; 1264 m_freem(rte->m); 1265 free(rte, M_MRTABLE6); 1266 rte = n; 1267 } while (rte != NULL); 1268 mrt6stat.mrt6s_cache_cleanups++; 1269 n6expire[i]--; 1270 1271 *nptr = mfc->mf6c_next; 1272 free(mfc, M_MRTABLE6); 1273 } else { 1274 nptr = &mfc->mf6c_next; 1275 } 1276 } 1277 } 1278 splx(s); 1279 callout_reset(&expire_upcalls_ch, EXPIRE_TIMEOUT, 1280 expire_upcalls, NULL); 1281 } 1282 1283 /* 1284 * Packet forwarding routine once entry in the cache is made 1285 */ 1286 static int 1287 ip6_mdq(m, ifp, rt) 1288 struct mbuf *m; 1289 struct ifnet *ifp; 1290 struct mf6c *rt; 1291 { 1292 struct ip6_hdr *ip6 = mtod(m, struct ip6_hdr *); 1293 mifi_t mifi, iif; 1294 struct mif6 *mifp; 1295 int plen = m->m_pkthdr.len; 1296 struct in6_addr src0, dst0; /* copies for local work */ 1297 u_int32_t iszone, idzone, oszone, odzone; 1298 int error = 0; 1299 1300 /* 1301 * Macro to send packet on mif. Since RSVP packets don't get counted on 1302 * input, they shouldn't get counted on output, so statistics keeping is 1303 * separate. 1304 */ 1305 1306 #define MC6_SEND(ip6, mifp, m) do { \ 1307 if ((mifp)->m6_flags & MIFF_REGISTER) \ 1308 register_send((ip6), (mifp), (m)); \ 1309 else \ 1310 phyint_send((ip6), (mifp), (m)); \ 1311 } while (/*CONSTCOND*/ 0) 1312 1313 /* 1314 * Don't forward if it didn't arrive from the parent mif 1315 * for its origin. 1316 */ 1317 mifi = rt->mf6c_parent; 1318 if ((mifi >= nummifs) || (mif6table[mifi].m6_ifp != ifp)) { 1319 /* came in the wrong interface */ 1320 #ifdef MRT6DEBUG 1321 if (mrt6debug & DEBUG_FORWARD) 1322 log(LOG_DEBUG, 1323 "wrong if: ifid %d mifi %d mififid %x\n", 1324 ifp->if_index, mifi, 1325 mif6table[mifi].m6_ifp->if_index); 1326 #endif 1327 mrt6stat.mrt6s_wrong_if++; 1328 rt->mf6c_wrong_if++; 1329 /* 1330 * If we are doing PIM processing, and we are forwarding 1331 * packets on this interface, send a message to the 1332 * routing daemon. 1333 */ 1334 /* have to make sure this is a valid mif */ 1335 if (mifi < nummifs && mif6table[mifi].m6_ifp) 1336 if (pim6 && (m->m_flags & M_LOOP) == 0) { 1337 /* 1338 * Check the M_LOOP flag to avoid an 1339 * unnecessary PIM assert. 1340 * XXX: M_LOOP is an ad-hoc hack... 1341 */ 1342 static struct sockaddr_in6 sin6 = 1343 { sizeof(sin6), AF_INET6 }; 1344 1345 struct mbuf *mm; 1346 struct mrt6msg *im; 1347 #ifdef MRT6_OINIT 1348 struct omrt6msg *oim; 1349 #endif 1350 1351 mm = m_copy(m, 0, sizeof(struct ip6_hdr)); 1352 if (mm && 1353 (M_HASCL(mm) || 1354 mm->m_len < sizeof(struct ip6_hdr))) 1355 mm = m_pullup(mm, sizeof(struct ip6_hdr)); 1356 if (mm == NULL) 1357 return (ENOBUFS); 1358 1359 #ifdef MRT6_OINIT 1360 oim = NULL; 1361 #endif 1362 im = NULL; 1363 switch (ip6_mrouter_ver) { 1364 #ifdef MRT6_OINIT 1365 case MRT6_OINIT: 1366 oim = mtod(mm, struct omrt6msg *); 1367 oim->im6_msgtype = MRT6MSG_WRONGMIF; 1368 oim->im6_mbz = 0; 1369 break; 1370 #endif 1371 case MRT6_INIT: 1372 im = mtod(mm, struct mrt6msg *); 1373 im->im6_msgtype = MRT6MSG_WRONGMIF; 1374 im->im6_mbz = 0; 1375 break; 1376 default: 1377 m_freem(mm); 1378 return (EINVAL); 1379 } 1380 1381 for (mifp = mif6table, iif = 0; 1382 iif < nummifs && mifp && 1383 mifp->m6_ifp != ifp; 1384 mifp++, iif++) 1385 ; 1386 1387 switch (ip6_mrouter_ver) { 1388 #ifdef MRT6_OINIT 1389 case MRT6_OINIT: 1390 oim->im6_mif = iif; 1391 sin6.sin6_addr = oim->im6_src; 1392 break; 1393 #endif 1394 case MRT6_INIT: 1395 im->im6_mif = iif; 1396 sin6.sin6_addr = im->im6_src; 1397 break; 1398 } 1399 1400 mrt6stat.mrt6s_upcalls++; 1401 1402 if (socket_send(ip6_mrouter, mm, &sin6) < 0) { 1403 #ifdef MRT6DEBUG 1404 if (mrt6debug) 1405 log(LOG_WARNING, "mdq, ip6_mrouter socket queue full\n"); 1406 #endif 1407 ++mrt6stat.mrt6s_upq_sockfull; 1408 return (ENOBUFS); 1409 } /* if socket Q full */ 1410 } /* if PIM */ 1411 return (0); 1412 } /* if wrong iif */ 1413 1414 /* If I sourced this packet, it counts as output, else it was input. */ 1415 if (m->m_pkthdr.rcvif == NULL) { 1416 /* XXX: is rcvif really NULL when output?? */ 1417 mif6table[mifi].m6_pkt_out++; 1418 mif6table[mifi].m6_bytes_out += plen; 1419 } else { 1420 mif6table[mifi].m6_pkt_in++; 1421 mif6table[mifi].m6_bytes_in += plen; 1422 } 1423 rt->mf6c_pkt_cnt++; 1424 rt->mf6c_byte_cnt += plen; 1425 1426 /* 1427 * For each mif, forward a copy of the packet if there are group 1428 * members downstream on the interface. 1429 */ 1430 src0 = ip6->ip6_src; 1431 dst0 = ip6->ip6_dst; 1432 if ((error = in6_setscope(&src0, ifp, &iszone)) != 0 || 1433 (error = in6_setscope(&dst0, ifp, &idzone)) != 0) { 1434 ip6stat.ip6s_badscope++; 1435 return (error); 1436 } 1437 for (mifp = mif6table, mifi = 0; mifi < nummifs; mifp++, mifi++) { 1438 if (IF_ISSET(mifi, &rt->mf6c_ifset)) { 1439 /* 1440 * check if the outgoing packet is going to break 1441 * a scope boundary. 1442 * XXX For packets through PIM register tunnel 1443 * interface, we believe a routing daemon. 1444 */ 1445 if (!(mif6table[rt->mf6c_parent].m6_flags & 1446 MIFF_REGISTER) && 1447 !(mif6table[mifi].m6_flags & MIFF_REGISTER)) { 1448 if (in6_setscope(&src0, mif6table[mifi].m6_ifp, 1449 &oszone) || 1450 in6_setscope(&dst0, mif6table[mifi].m6_ifp, 1451 &odzone) || 1452 iszone != oszone || 1453 idzone != odzone) { 1454 ip6stat.ip6s_badscope++; 1455 continue; 1456 } 1457 } 1458 1459 mifp->m6_pkt_out++; 1460 mifp->m6_bytes_out += plen; 1461 MC6_SEND(ip6, mifp, m); 1462 } 1463 } 1464 return (0); 1465 } 1466 1467 static void 1468 phyint_send(ip6, mifp, m) 1469 struct ip6_hdr *ip6; 1470 struct mif6 *mifp; 1471 struct mbuf *m; 1472 { 1473 struct mbuf *mb_copy; 1474 struct ifnet *ifp = mifp->m6_ifp; 1475 int error = 0; 1476 int s = splnet(); /* needs to protect static "ro" below. */ 1477 static struct route_in6 ro; 1478 struct in6_multi *in6m; 1479 struct sockaddr_in6 *dst6; 1480 u_long linkmtu; 1481 1482 /* 1483 * Make a new reference to the packet; make sure that 1484 * the IPv6 header is actually copied, not just referenced, 1485 * so that ip6_output() only scribbles on the copy. 1486 */ 1487 mb_copy = m_copy(m, 0, M_COPYALL); 1488 if (mb_copy && 1489 (M_HASCL(mb_copy) || mb_copy->m_len < sizeof(struct ip6_hdr))) 1490 mb_copy = m_pullup(mb_copy, sizeof(struct ip6_hdr)); 1491 if (mb_copy == NULL) { 1492 splx(s); 1493 return; 1494 } 1495 /* set MCAST flag to the outgoing packet */ 1496 mb_copy->m_flags |= M_MCAST; 1497 1498 /* 1499 * If we sourced the packet, call ip6_output since we may devide 1500 * the packet into fragments when the packet is too big for the 1501 * outgoing interface. 1502 * Otherwise, we can simply send the packet to the interface 1503 * sending queue. 1504 */ 1505 if (m->m_pkthdr.rcvif == NULL) { 1506 struct ip6_moptions im6o; 1507 1508 im6o.im6o_multicast_ifp = ifp; 1509 /* XXX: ip6_output will override ip6->ip6_hlim */ 1510 im6o.im6o_multicast_hlim = ip6->ip6_hlim; 1511 im6o.im6o_multicast_loop = 1; 1512 error = ip6_output(mb_copy, NULL, &ro, 1513 IPV6_FORWARDING, &im6o, NULL, NULL); 1514 1515 #ifdef MRT6DEBUG 1516 if (mrt6debug & DEBUG_XMIT) 1517 log(LOG_DEBUG, "phyint_send on mif %d err %d\n", 1518 mifp - mif6table, error); 1519 #endif 1520 splx(s); 1521 return; 1522 } 1523 1524 /* 1525 * If we belong to the destination multicast group 1526 * on the outgoing interface, loop back a copy. 1527 */ 1528 dst6 = (struct sockaddr_in6 *)&ro.ro_dst; 1529 IN6_LOOKUP_MULTI(ip6->ip6_dst, ifp, in6m); 1530 if (in6m != NULL) { 1531 dst6->sin6_len = sizeof(struct sockaddr_in6); 1532 dst6->sin6_family = AF_INET6; 1533 dst6->sin6_addr = ip6->ip6_dst; 1534 ip6_mloopback(ifp, m, (struct sockaddr_in6 *)&ro.ro_dst); 1535 } 1536 /* 1537 * Put the packet into the sending queue of the outgoing interface 1538 * if it would fit in the MTU of the interface. 1539 */ 1540 linkmtu = IN6_LINKMTU(ifp); 1541 if (mb_copy->m_pkthdr.len <= linkmtu || linkmtu < IPV6_MMTU) { 1542 dst6->sin6_len = sizeof(struct sockaddr_in6); 1543 dst6->sin6_family = AF_INET6; 1544 dst6->sin6_addr = ip6->ip6_dst; 1545 /* 1546 * We just call if_output instead of nd6_output here, since 1547 * we need no ND for a multicast forwarded packet...right? 1548 */ 1549 error = (*ifp->if_output)(ifp, mb_copy, 1550 (struct sockaddr *)&ro.ro_dst, NULL); 1551 #ifdef MRT6DEBUG 1552 if (mrt6debug & DEBUG_XMIT) 1553 log(LOG_DEBUG, "phyint_send on mif %d err %d\n", 1554 mifp - mif6table, error); 1555 #endif 1556 } else { 1557 /* 1558 * pMTU discovery is intentionally disabled by default, since 1559 * various router may notify pMTU in multicast, which can be 1560 * a DDoS to a router 1561 */ 1562 if (ip6_mcast_pmtu) 1563 icmp6_error(mb_copy, ICMP6_PACKET_TOO_BIG, 0, linkmtu); 1564 else { 1565 #ifdef MRT6DEBUG 1566 if (mrt6debug & DEBUG_XMIT) 1567 log(LOG_DEBUG, 1568 "phyint_send: packet too big on %s o %s " 1569 "g %s size %d(discarded)\n", 1570 if_name(ifp), 1571 ip6_sprintf(&ip6->ip6_src), 1572 ip6_sprintf(&ip6->ip6_dst), 1573 mb_copy->m_pkthdr.len); 1574 #endif /* MRT6DEBUG */ 1575 m_freem(mb_copy); /* simply discard the packet */ 1576 } 1577 } 1578 1579 splx(s); 1580 } 1581 1582 static int 1583 register_send(ip6, mif, m) 1584 struct ip6_hdr *ip6; 1585 struct mif6 *mif; 1586 struct mbuf *m; 1587 { 1588 struct mbuf *mm; 1589 int i, len = m->m_pkthdr.len; 1590 static struct sockaddr_in6 sin6 = { sizeof(sin6), AF_INET6 }; 1591 struct mrt6msg *im6; 1592 1593 #ifdef MRT6DEBUG 1594 if (mrt6debug) 1595 log(LOG_DEBUG, "** IPv6 register_send **\n src %s dst %s\n", 1596 ip6_sprintf(&ip6->ip6_src), ip6_sprintf(&ip6->ip6_dst)); 1597 #endif 1598 ++pim6stat.pim6s_snd_registers; 1599 1600 /* Make a copy of the packet to send to the user level process */ 1601 MGETHDR(mm, M_DONTWAIT, MT_HEADER); 1602 if (mm == NULL) 1603 return (ENOBUFS); 1604 mm->m_pkthdr.rcvif = NULL; 1605 mm->m_data += max_linkhdr; 1606 mm->m_len = sizeof(struct ip6_hdr); 1607 1608 if ((mm->m_next = m_copy(m, 0, M_COPYALL)) == NULL) { 1609 m_freem(mm); 1610 return (ENOBUFS); 1611 } 1612 i = MHLEN - M_LEADINGSPACE(mm); 1613 if (i > len) 1614 i = len; 1615 mm = m_pullup(mm, i); 1616 if (mm == NULL) 1617 return (ENOBUFS); 1618 /* TODO: check it! */ 1619 mm->m_pkthdr.len = len + sizeof(struct ip6_hdr); 1620 1621 /* 1622 * Send message to routing daemon 1623 */ 1624 sin6.sin6_addr = ip6->ip6_src; 1625 1626 im6 = mtod(mm, struct mrt6msg *); 1627 im6->im6_msgtype = MRT6MSG_WHOLEPKT; 1628 im6->im6_mbz = 0; 1629 1630 im6->im6_mif = mif - mif6table; 1631 1632 /* iif info is not given for reg. encap.n */ 1633 mrt6stat.mrt6s_upcalls++; 1634 1635 if (socket_send(ip6_mrouter, mm, &sin6) < 0) { 1636 #ifdef MRT6DEBUG 1637 if (mrt6debug) 1638 log(LOG_WARNING, 1639 "register_send: ip6_mrouter socket queue full\n"); 1640 #endif 1641 ++mrt6stat.mrt6s_upq_sockfull; 1642 return (ENOBUFS); 1643 } 1644 return (0); 1645 } 1646 1647 /* 1648 * PIM sparse mode hook 1649 * Receives the pim control messages, and passes them up to the listening 1650 * socket, using rip6_input. 1651 * The only message processed is the REGISTER pim message; the pim header 1652 * is stripped off, and the inner packet is passed to register_mforward. 1653 */ 1654 int 1655 pim6_input(mp, offp, proto) 1656 struct mbuf **mp; 1657 int *offp, proto; 1658 { 1659 struct pim *pim; /* pointer to a pim struct */ 1660 struct ip6_hdr *ip6; 1661 int pimlen; 1662 struct mbuf *m = *mp; 1663 int minlen; 1664 int off = *offp; 1665 1666 ++pim6stat.pim6s_rcv_total; 1667 1668 ip6 = mtod(m, struct ip6_hdr *); 1669 pimlen = m->m_pkthdr.len - *offp; 1670 1671 /* 1672 * Validate lengths 1673 */ 1674 if (pimlen < PIM_MINLEN) { 1675 ++pim6stat.pim6s_rcv_tooshort; 1676 #ifdef MRT6DEBUG 1677 if (mrt6debug & DEBUG_PIM) 1678 log(LOG_DEBUG,"pim6_input: PIM packet too short\n"); 1679 #endif 1680 m_freem(m); 1681 return (IPPROTO_DONE); 1682 } 1683 1684 /* 1685 * if the packet is at least as big as a REGISTER, go ahead 1686 * and grab the PIM REGISTER header size, to avoid another 1687 * possible m_pullup() later. 1688 * 1689 * PIM_MINLEN == pimhdr + u_int32 == 8 1690 * PIM6_REG_MINLEN == pimhdr + reghdr + eip6hdr == 4 + 4 + 40 1691 */ 1692 minlen = (pimlen >= PIM6_REG_MINLEN) ? PIM6_REG_MINLEN : PIM_MINLEN; 1693 1694 /* 1695 * Make sure that the IP6 and PIM headers in contiguous memory, and 1696 * possibly the PIM REGISTER header 1697 */ 1698 #ifndef PULLDOWN_TEST 1699 IP6_EXTHDR_CHECK(m, off, minlen, IPPROTO_DONE); 1700 /* adjust pointer */ 1701 ip6 = mtod(m, struct ip6_hdr *); 1702 1703 /* adjust mbuf to point to the PIM header */ 1704 pim = (struct pim *)((caddr_t)ip6 + off); 1705 #else 1706 IP6_EXTHDR_GET(pim, struct pim *, m, off, minlen); 1707 if (pim == NULL) { 1708 pim6stat.pim6s_rcv_tooshort++; 1709 return (IPPROTO_DONE); 1710 } 1711 #endif 1712 1713 #define PIM6_CHECKSUM 1714 #ifdef PIM6_CHECKSUM 1715 { 1716 int cksumlen; 1717 1718 /* 1719 * Validate checksum. 1720 * If PIM REGISTER, exclude the data packet 1721 */ 1722 if (pim->pim_type == PIM_REGISTER) 1723 cksumlen = PIM_MINLEN; 1724 else 1725 cksumlen = pimlen; 1726 1727 if (in6_cksum(m, IPPROTO_PIM, off, cksumlen)) { 1728 ++pim6stat.pim6s_rcv_badsum; 1729 #ifdef MRT6DEBUG 1730 if (mrt6debug & DEBUG_PIM) 1731 log(LOG_DEBUG, 1732 "pim6_input: invalid checksum\n"); 1733 #endif 1734 m_freem(m); 1735 return (IPPROTO_DONE); 1736 } 1737 } 1738 #endif /* PIM_CHECKSUM */ 1739 1740 /* PIM version check */ 1741 if (pim->pim_ver != PIM_VERSION) { 1742 ++pim6stat.pim6s_rcv_badversion; 1743 #ifdef MRT6DEBUG 1744 log(LOG_ERR, 1745 "pim6_input: incorrect version %d, expecting %d\n", 1746 pim->pim_ver, PIM_VERSION); 1747 #endif 1748 m_freem(m); 1749 return (IPPROTO_DONE); 1750 } 1751 1752 if (pim->pim_type == PIM_REGISTER) { 1753 /* 1754 * since this is a REGISTER, we'll make a copy of the register 1755 * headers ip6+pim+u_int32_t+encap_ip6, to be passed up to the 1756 * routing daemon. 1757 */ 1758 static struct sockaddr_in6 dst = { sizeof(dst), AF_INET6 }; 1759 1760 struct mbuf *mcp; 1761 struct ip6_hdr *eip6; 1762 u_int32_t *reghdr; 1763 int rc; 1764 1765 ++pim6stat.pim6s_rcv_registers; 1766 1767 if ((reg_mif_num >= nummifs) || (reg_mif_num == (mifi_t) -1)) { 1768 #ifdef MRT6DEBUG 1769 if (mrt6debug & DEBUG_PIM) 1770 log(LOG_DEBUG, 1771 "pim6_input: register mif not set: %d\n", 1772 reg_mif_num); 1773 #endif 1774 m_freem(m); 1775 return (IPPROTO_DONE); 1776 } 1777 1778 reghdr = (u_int32_t *)(pim + 1); 1779 1780 if ((ntohl(*reghdr) & PIM_NULL_REGISTER)) 1781 goto pim6_input_to_daemon; 1782 1783 /* 1784 * Validate length 1785 */ 1786 if (pimlen < PIM6_REG_MINLEN) { 1787 ++pim6stat.pim6s_rcv_tooshort; 1788 ++pim6stat.pim6s_rcv_badregisters; 1789 #ifdef MRT6DEBUG 1790 log(LOG_ERR, 1791 "pim6_input: register packet size too " 1792 "small %d from %s\n", 1793 pimlen, ip6_sprintf(&ip6->ip6_src)); 1794 #endif 1795 m_freem(m); 1796 return (IPPROTO_DONE); 1797 } 1798 1799 eip6 = (struct ip6_hdr *) (reghdr + 1); 1800 #ifdef MRT6DEBUG 1801 if (mrt6debug & DEBUG_PIM) 1802 log(LOG_DEBUG, 1803 "pim6_input[register], eip6: %s -> %s, " 1804 "eip6 plen %d\n", 1805 ip6_sprintf(&eip6->ip6_src), 1806 ip6_sprintf(&eip6->ip6_dst), 1807 ntohs(eip6->ip6_plen)); 1808 #endif 1809 1810 /* verify the version number of the inner packet */ 1811 if ((eip6->ip6_vfc & IPV6_VERSION_MASK) != IPV6_VERSION) { 1812 ++pim6stat.pim6s_rcv_badregisters; 1813 #ifdef MRT6DEBUG 1814 log(LOG_DEBUG, "pim6_input: invalid IP version (%d) " 1815 "of the inner packet\n", 1816 (eip6->ip6_vfc & IPV6_VERSION)); 1817 #endif 1818 m_freem(m); 1819 return (IPPROTO_NONE); 1820 } 1821 1822 /* verify the inner packet is destined to a mcast group */ 1823 if (!IN6_IS_ADDR_MULTICAST(&eip6->ip6_dst)) { 1824 ++pim6stat.pim6s_rcv_badregisters; 1825 #ifdef MRT6DEBUG 1826 if (mrt6debug & DEBUG_PIM) 1827 log(LOG_DEBUG, 1828 "pim6_input: inner packet of register " 1829 "is not multicast %s\n", 1830 ip6_sprintf(&eip6->ip6_dst)); 1831 #endif 1832 m_freem(m); 1833 return (IPPROTO_DONE); 1834 } 1835 1836 /* 1837 * make a copy of the whole header to pass to the daemon later. 1838 */ 1839 mcp = m_copy(m, 0, off + PIM6_REG_MINLEN); 1840 if (mcp == NULL) { 1841 #ifdef MRT6DEBUG 1842 log(LOG_ERR, 1843 "pim6_input: pim register: " 1844 "could not copy register head\n"); 1845 #endif 1846 m_freem(m); 1847 return (IPPROTO_DONE); 1848 } 1849 1850 /* 1851 * forward the inner ip6 packet; point m_data at the inner ip6. 1852 */ 1853 m_adj(m, off + PIM_MINLEN); 1854 #ifdef MRT6DEBUG 1855 if (mrt6debug & DEBUG_PIM) { 1856 log(LOG_DEBUG, 1857 "pim6_input: forwarding decapsulated register: " 1858 "src %s, dst %s, mif %d\n", 1859 ip6_sprintf(&eip6->ip6_src), 1860 ip6_sprintf(&eip6->ip6_dst), 1861 reg_mif_num); 1862 } 1863 #endif 1864 1865 rc = if_simloop(mif6table[reg_mif_num].m6_ifp, m, 1866 dst.sin6_family, 0); 1867 1868 /* prepare the register head to send to the mrouting daemon */ 1869 m = mcp; 1870 } 1871 1872 /* 1873 * Pass the PIM message up to the daemon; if it is a register message 1874 * pass the 'head' only up to the daemon. This includes the 1875 * encapsulator ip6 header, pim header, register header and the 1876 * encapsulated ip6 header. 1877 */ 1878 pim6_input_to_daemon: 1879 rip6_input(&m, offp, proto); 1880 return (IPPROTO_DONE); 1881 } 1882