1 /* $NetBSD: ip6_mroute.c,v 1.64 2005/05/29 21:43:09 christos Exp $ */ 2 /* $KAME: ip6_mroute.c,v 1.49 2001/07/25 09:21:18 jinmei 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) 1992, 1993 37 * The Regents of the University of California. All rights reserved. 38 * 39 * This code is derived from software contributed to Berkeley by 40 * Stephen Deering of Stanford University. 41 * 42 * Redistribution and use in source and binary forms, with or without 43 * modification, are permitted provided that the following conditions 44 * are met: 45 * 1. Redistributions of source code must retain the above copyright 46 * notice, this list of conditions and the following disclaimer. 47 * 2. Redistributions in binary form must reproduce the above copyright 48 * notice, this list of conditions and the following disclaimer in the 49 * documentation and/or other materials provided with the distribution. 50 * 3. Neither the name of the University nor the names of its contributors 51 * may be used to endorse or promote products derived from this software 52 * without specific prior written permission. 53 * 54 * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND 55 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE 56 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE 57 * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE 58 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL 59 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS 60 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) 61 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT 62 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY 63 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF 64 * SUCH DAMAGE. 65 * 66 * @(#)ip_mroute.c 8.2 (Berkeley) 11/15/93 67 */ 68 69 /* 70 * Copyright (c) 1989 Stephen Deering 71 * 72 * This code is derived from software contributed to Berkeley by 73 * Stephen Deering of Stanford University. 74 * 75 * Redistribution and use in source and binary forms, with or without 76 * modification, are permitted provided that the following conditions 77 * are met: 78 * 1. Redistributions of source code must retain the above copyright 79 * notice, this list of conditions and the following disclaimer. 80 * 2. Redistributions in binary form must reproduce the above copyright 81 * notice, this list of conditions and the following disclaimer in the 82 * documentation and/or other materials provided with the distribution. 83 * 3. All advertising materials mentioning features or use of this software 84 * must display the following acknowledgement: 85 * This product includes software developed by the University of 86 * California, Berkeley and its contributors. 87 * 4. Neither the name of the University nor the names of its contributors 88 * may be used to endorse or promote products derived from this software 89 * without specific prior written permission. 90 * 91 * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND 92 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE 93 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE 94 * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE 95 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL 96 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS 97 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) 98 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT 99 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY 100 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF 101 * SUCH DAMAGE. 102 * 103 * @(#)ip_mroute.c 8.2 (Berkeley) 11/15/93 104 */ 105 106 /* 107 * IP multicast forwarding procedures 108 * 109 * Written by David Waitzman, BBN Labs, August 1988. 110 * Modified by Steve Deering, Stanford, February 1989. 111 * Modified by Mark J. Steiglitz, Stanford, May, 1991 112 * Modified by Van Jacobson, LBL, January 1993 113 * Modified by Ajit Thyagarajan, PARC, August 1993 114 * Modified by Bill Fenner, PARC, April 1994 115 * 116 * MROUTING Revision: 3.5.1.2 + PIM-SMv2 (pimd) Support 117 */ 118 119 #include <sys/cdefs.h> 120 __KERNEL_RCSID(0, "$NetBSD: ip6_mroute.c,v 1.64 2005/05/29 21:43:09 christos Exp $"); 121 122 #include "opt_inet.h" 123 #include "opt_mrouting.h" 124 125 #include <sys/param.h> 126 #include <sys/systm.h> 127 #include <sys/callout.h> 128 #include <sys/mbuf.h> 129 #include <sys/socket.h> 130 #include <sys/socketvar.h> 131 #include <sys/sockio.h> 132 #include <sys/protosw.h> 133 #include <sys/errno.h> 134 #include <sys/time.h> 135 #include <sys/kernel.h> 136 #include <sys/ioctl.h> 137 #include <sys/syslog.h> 138 139 #include <net/if.h> 140 #include <net/route.h> 141 #include <net/raw_cb.h> 142 143 #include <netinet/in.h> 144 #include <netinet/in_var.h> 145 #ifdef MULTICAST_PMTUD 146 #include <netinet/icmp6.h> 147 #endif 148 149 #include <netinet/ip6.h> 150 #include <netinet6/ip6_var.h> 151 #include <netinet6/ip6_mroute.h> 152 #include <netinet6/pim6.h> 153 #include <netinet6/pim6_var.h> 154 #include <netinet6/nd6.h> 155 156 #include <net/net_osdep.h> 157 158 static int ip6_mdq __P((struct mbuf *, struct ifnet *, struct mf6c *)); 159 static void phyint_send __P((struct ip6_hdr *, struct mif6 *, struct mbuf *)); 160 161 static int set_pim6 __P((int *)); 162 static int get_pim6 __P((struct mbuf *)); 163 static int socket_send __P((struct socket *, struct mbuf *, 164 struct sockaddr_in6 *)); 165 static int register_send __P((struct ip6_hdr *, struct mif6 *, struct mbuf *)); 166 167 /* 168 * Globals. All but ip6_mrouter, ip6_mrtproto and mrt6stat could be static, 169 * except for netstat or debugging purposes. 170 */ 171 struct socket *ip6_mrouter = NULL; 172 int ip6_mrouter_ver = 0; 173 int ip6_mrtproto = IPPROTO_PIM; /* for netstat only */ 174 struct mrt6stat mrt6stat; 175 176 #define NO_RTE_FOUND 0x1 177 #define RTE_FOUND 0x2 178 179 struct mf6c *mf6ctable[MF6CTBLSIZ]; 180 u_char n6expire[MF6CTBLSIZ]; 181 static struct mif6 mif6table[MAXMIFS]; 182 #ifdef MRT6DEBUG 183 u_int mrt6debug = 0; /* debug level */ 184 #define DEBUG_MFC 0x02 185 #define DEBUG_FORWARD 0x04 186 #define DEBUG_EXPIRE 0x08 187 #define DEBUG_XMIT 0x10 188 #define DEBUG_REG 0x20 189 #define DEBUG_PIM 0x40 190 #endif 191 192 static void expire_upcalls __P((void *)); 193 #define EXPIRE_TIMEOUT (hz / 4) /* 4x / second */ 194 #define UPCALL_EXPIRE 6 /* number of timeouts */ 195 196 #ifdef INET 197 #ifdef MROUTING 198 extern struct socket *ip_mrouter; 199 #endif 200 #endif 201 202 /* 203 * 'Interfaces' associated with decapsulator (so we can tell 204 * packets that went through it from ones that get reflected 205 * by a broken gateway). These interfaces are never linked into 206 * the system ifnet list & no routes point to them. I.e., packets 207 * can't be sent this way. They only exist as a placeholder for 208 * multicast source verification. 209 */ 210 struct ifnet multicast_register_if; 211 212 #define ENCAP_HOPS 64 213 214 /* 215 * Private variables. 216 */ 217 static mifi_t nummifs = 0; 218 static mifi_t reg_mif_num = (mifi_t)-1; 219 220 struct pim6stat pim6stat; 221 static int pim6; 222 223 /* 224 * Hash function for a source, group entry 225 */ 226 #define MF6CHASH(a, g) MF6CHASHMOD((a).s6_addr32[0] ^ (a).s6_addr32[1] ^ \ 227 (a).s6_addr32[2] ^ (a).s6_addr32[3] ^ \ 228 (g).s6_addr32[0] ^ (g).s6_addr32[1] ^ \ 229 (g).s6_addr32[2] ^ (g).s6_addr32[3]) 230 231 /* 232 * Find a route for a given origin IPv6 address and Multicast group address. 233 * Quality of service parameter to be added in the future!!! 234 */ 235 236 #define MF6CFIND(o, g, rt) do { \ 237 struct mf6c *_rt = mf6ctable[MF6CHASH(o,g)]; \ 238 rt = NULL; \ 239 mrt6stat.mrt6s_mfc_lookups++; \ 240 while (_rt) { \ 241 if (IN6_ARE_ADDR_EQUAL(&_rt->mf6c_origin.sin6_addr, &(o)) && \ 242 IN6_ARE_ADDR_EQUAL(&_rt->mf6c_mcastgrp.sin6_addr, &(g)) && \ 243 (_rt->mf6c_stall == NULL)) { \ 244 rt = _rt; \ 245 break; \ 246 } \ 247 _rt = _rt->mf6c_next; \ 248 } \ 249 if (rt == NULL) { \ 250 mrt6stat.mrt6s_mfc_misses++; \ 251 } \ 252 } while (/*CONSTCOND*/ 0) 253 254 /* 255 * Macros to compute elapsed time efficiently 256 * Borrowed from Van Jacobson's scheduling code 257 */ 258 #define TV_DELTA(a, b, delta) do { \ 259 int xxs; \ 260 \ 261 delta = (a).tv_usec - (b).tv_usec; \ 262 if ((xxs = (a).tv_sec - (b).tv_sec)) { \ 263 switch (xxs) { \ 264 case 2: \ 265 delta += 1000000; \ 266 /* FALLTHROUGH */ \ 267 case 1: \ 268 delta += 1000000; \ 269 break; \ 270 default: \ 271 delta += (1000000 * xxs); \ 272 } \ 273 } \ 274 } while (/*CONSTCOND*/ 0) 275 276 #define TV_LT(a, b) (((a).tv_usec < (b).tv_usec && \ 277 (a).tv_sec <= (b).tv_sec) || (a).tv_sec < (b).tv_sec) 278 279 #ifdef UPCALL_TIMING 280 #define UPCALL_MAX 50 281 u_long upcall_data[UPCALL_MAX + 1]; 282 static void collate(); 283 #endif /* UPCALL_TIMING */ 284 285 static int get_sg_cnt __P((struct sioc_sg_req6 *)); 286 static int get_mif6_cnt __P((struct sioc_mif_req6 *)); 287 static int ip6_mrouter_init __P((struct socket *, int, int)); 288 static int add_m6if __P((struct mif6ctl *)); 289 static int del_m6if __P((mifi_t *)); 290 static int add_m6fc __P((struct mf6cctl *)); 291 static int del_m6fc __P((struct mf6cctl *)); 292 293 static struct callout expire_upcalls_ch = CALLOUT_INITIALIZER; 294 295 /* 296 * Handle MRT setsockopt commands to modify the multicast routing tables. 297 */ 298 int 299 ip6_mrouter_set(cmd, so, m) 300 int cmd; 301 struct socket *so; 302 struct mbuf *m; 303 { 304 if (cmd != MRT6_INIT && so != ip6_mrouter) 305 return (EACCES); 306 307 switch (cmd) { 308 #ifdef MRT6_OINIT 309 case MRT6_OINIT: 310 #endif 311 case MRT6_INIT: 312 if (m == NULL || m->m_len < sizeof(int)) 313 return (EINVAL); 314 return (ip6_mrouter_init(so, *mtod(m, int *), cmd)); 315 case MRT6_DONE: 316 return (ip6_mrouter_done()); 317 case MRT6_ADD_MIF: 318 if (m == NULL || m->m_len < sizeof(struct mif6ctl)) 319 return (EINVAL); 320 return (add_m6if(mtod(m, struct mif6ctl *))); 321 case MRT6_DEL_MIF: 322 if (m == NULL || m->m_len < sizeof(mifi_t)) 323 return (EINVAL); 324 return (del_m6if(mtod(m, mifi_t *))); 325 case MRT6_ADD_MFC: 326 if (m == NULL || m->m_len < sizeof(struct mf6cctl)) 327 return (EINVAL); 328 return (add_m6fc(mtod(m, struct mf6cctl *))); 329 case MRT6_DEL_MFC: 330 if (m == NULL || m->m_len < sizeof(struct mf6cctl)) 331 return (EINVAL); 332 return (del_m6fc(mtod(m, struct mf6cctl *))); 333 case MRT6_PIM: 334 if (m == NULL || m->m_len < sizeof(int)) 335 return (EINVAL); 336 return (set_pim6(mtod(m, int *))); 337 default: 338 return (EOPNOTSUPP); 339 } 340 } 341 342 /* 343 * Handle MRT getsockopt commands 344 */ 345 int 346 ip6_mrouter_get(cmd, so, m) 347 int cmd; 348 struct socket *so; 349 struct mbuf **m; 350 { 351 struct mbuf *mb; 352 353 if (so != ip6_mrouter) return EACCES; 354 355 *m = mb = m_get(M_WAIT, MT_SOOPTS); 356 357 switch (cmd) { 358 case MRT6_PIM: 359 return get_pim6(mb); 360 default: 361 m_free(mb); 362 return EOPNOTSUPP; 363 } 364 } 365 366 /* 367 * Handle ioctl commands to obtain information from the cache 368 */ 369 int 370 mrt6_ioctl(cmd, data) 371 int cmd; 372 caddr_t data; 373 { 374 375 switch (cmd) { 376 case SIOCGETSGCNT_IN6: 377 return (get_sg_cnt((struct sioc_sg_req6 *)data)); 378 case SIOCGETMIFCNT_IN6: 379 return (get_mif6_cnt((struct sioc_mif_req6 *)data)); 380 default: 381 return (EINVAL); 382 } 383 } 384 385 /* 386 * returns the packet, byte, rpf-failure count for the source group provided 387 */ 388 static int 389 get_sg_cnt(req) 390 struct sioc_sg_req6 *req; 391 { 392 struct mf6c *rt; 393 int s; 394 395 s = splsoftnet(); 396 MF6CFIND(req->src.sin6_addr, req->grp.sin6_addr, rt); 397 splx(s); 398 if (rt != NULL) { 399 req->pktcnt = rt->mf6c_pkt_cnt; 400 req->bytecnt = rt->mf6c_byte_cnt; 401 req->wrong_if = rt->mf6c_wrong_if; 402 } else 403 return (ESRCH); 404 #if 0 405 req->pktcnt = req->bytecnt = req->wrong_if = 0xffffffff; 406 #endif 407 408 return 0; 409 } 410 411 /* 412 * returns the input and output packet and byte counts on the mif provided 413 */ 414 static int 415 get_mif6_cnt(req) 416 struct sioc_mif_req6 *req; 417 { 418 mifi_t mifi = req->mifi; 419 420 if (mifi >= nummifs) 421 return EINVAL; 422 423 req->icount = mif6table[mifi].m6_pkt_in; 424 req->ocount = mif6table[mifi].m6_pkt_out; 425 req->ibytes = mif6table[mifi].m6_bytes_in; 426 req->obytes = mif6table[mifi].m6_bytes_out; 427 428 return 0; 429 } 430 431 /* 432 * Get PIM processiong global 433 */ 434 static int 435 get_pim6(m) 436 struct mbuf *m; 437 { 438 int *i; 439 440 i = mtod(m, int *); 441 442 *i = pim6; 443 444 return 0; 445 } 446 447 static int 448 set_pim6(i) 449 int *i; 450 { 451 if ((*i != 1) && (*i != 0)) 452 return EINVAL; 453 454 pim6 = *i; 455 456 return 0; 457 } 458 459 /* 460 * Enable multicast routing 461 */ 462 static int 463 ip6_mrouter_init(so, v, cmd) 464 struct socket *so; 465 int v; 466 int cmd; 467 { 468 #ifdef MRT6DEBUG 469 if (mrt6debug) 470 log(LOG_DEBUG, 471 "ip6_mrouter_init: so_type = %d, pr_protocol = %d\n", 472 so->so_type, so->so_proto->pr_protocol); 473 #endif 474 475 if (so->so_type != SOCK_RAW || 476 so->so_proto->pr_protocol != IPPROTO_ICMPV6) 477 return (EOPNOTSUPP); 478 479 if (v != 1) 480 return (ENOPROTOOPT); 481 482 if (ip6_mrouter != NULL) 483 return (EADDRINUSE); 484 485 ip6_mrouter = so; 486 ip6_mrouter_ver = cmd; 487 488 bzero((caddr_t)mf6ctable, sizeof(mf6ctable)); 489 bzero((caddr_t)n6expire, sizeof(n6expire)); 490 491 pim6 = 0;/* used for stubbing out/in pim stuff */ 492 493 callout_reset(&expire_upcalls_ch, EXPIRE_TIMEOUT, 494 expire_upcalls, NULL); 495 496 #ifdef MRT6DEBUG 497 if (mrt6debug) 498 log(LOG_DEBUG, "ip6_mrouter_init\n"); 499 #endif 500 501 return 0; 502 } 503 504 /* 505 * Disable multicast routing 506 */ 507 int 508 ip6_mrouter_done() 509 { 510 mifi_t mifi; 511 int i; 512 struct ifnet *ifp; 513 struct in6_ifreq ifr; 514 struct mf6c *rt; 515 struct rtdetq *rte; 516 int s; 517 518 s = splsoftnet(); 519 520 /* 521 * For each phyint in use, disable promiscuous reception of all IPv6 522 * multicasts. 523 */ 524 #ifdef INET 525 #ifdef MROUTING 526 /* 527 * If there is still IPv4 multicast routing daemon, 528 * we remain interfaces to receive all muliticasted packets. 529 * XXX: there may be an interface in which the IPv4 multicast 530 * daemon is not interested... 531 */ 532 if (!ip_mrouter) 533 #endif 534 #endif 535 { 536 for (mifi = 0; mifi < nummifs; mifi++) { 537 if (mif6table[mifi].m6_ifp && 538 !(mif6table[mifi].m6_flags & MIFF_REGISTER)) { 539 ifr.ifr_addr.sin6_family = AF_INET6; 540 ifr.ifr_addr.sin6_addr= in6addr_any; 541 ifp = mif6table[mifi].m6_ifp; 542 (*ifp->if_ioctl)(ifp, SIOCDELMULTI, 543 (caddr_t)&ifr); 544 } 545 } 546 } 547 #ifdef notyet 548 bzero((caddr_t)qtable, sizeof(qtable)); 549 bzero((caddr_t)tbftable, sizeof(tbftable)); 550 #endif 551 bzero((caddr_t)mif6table, sizeof(mif6table)); 552 nummifs = 0; 553 554 pim6 = 0; /* used to stub out/in pim specific code */ 555 556 callout_stop(&expire_upcalls_ch); 557 558 /* 559 * Free all multicast forwarding cache entries. 560 */ 561 for (i = 0; i < MF6CTBLSIZ; i++) { 562 rt = mf6ctable[i]; 563 while (rt) { 564 struct mf6c *frt; 565 566 for (rte = rt->mf6c_stall; rte != NULL; ) { 567 struct rtdetq *n = rte->next; 568 569 m_free(rte->m); 570 free(rte, M_MRTABLE); 571 rte = n; 572 } 573 frt = rt; 574 rt = rt->mf6c_next; 575 free(frt, M_MRTABLE); 576 } 577 } 578 579 bzero((caddr_t)mf6ctable, sizeof(mf6ctable)); 580 581 /* 582 * Reset de-encapsulation cache 583 */ 584 reg_mif_num = -1; 585 586 ip6_mrouter = NULL; 587 ip6_mrouter_ver = 0; 588 589 splx(s); 590 591 #ifdef MRT6DEBUG 592 if (mrt6debug) 593 log(LOG_DEBUG, "ip6_mrouter_done\n"); 594 #endif 595 596 return 0; 597 } 598 599 void 600 ip6_mrouter_detach(ifp) 601 struct ifnet *ifp; 602 { 603 struct rtdetq *rte; 604 struct mf6c *mfc; 605 mifi_t mifi; 606 int i; 607 608 /* 609 * Delete a mif which points to ifp. 610 */ 611 for (mifi = 0; mifi < nummifs; mifi++) 612 if (mif6table[mifi].m6_ifp == ifp) 613 del_m6if(&mifi); 614 615 /* 616 * Clear rte->ifp of cache entries received on ifp. 617 */ 618 for (i = 0; i < MF6CTBLSIZ; i++) { 619 if (n6expire[i] == 0) 620 continue; 621 622 for (mfc = mf6ctable[i]; mfc != NULL; mfc = mfc->mf6c_next) { 623 for (rte = mfc->mf6c_stall; rte != NULL; rte = rte->next) { 624 if (rte->ifp == ifp) 625 rte->ifp = NULL; 626 } 627 } 628 } 629 } 630 631 632 /* 633 * Add a mif to the mif table 634 */ 635 static int 636 add_m6if(mifcp) 637 struct mif6ctl *mifcp; 638 { 639 struct mif6 *mifp; 640 struct ifnet *ifp; 641 struct in6_ifreq ifr; 642 int error, s; 643 #ifdef notyet 644 struct tbf *m_tbf = tbftable + mifcp->mif6c_mifi; 645 #endif 646 647 if (mifcp->mif6c_mifi >= MAXMIFS) 648 return EINVAL; 649 mifp = mif6table + mifcp->mif6c_mifi; 650 if (mifp->m6_ifp) 651 return EADDRINUSE; /* XXX: is it appropriate? */ 652 if (mifcp->mif6c_pifi == 0 || mifcp->mif6c_pifi >= if_indexlim) 653 return ENXIO; 654 /* 655 * XXX: some OSes can remove ifp and clear ifindex2ifnet[id] 656 * even for id between 0 and if_index. 657 */ 658 if ((ifp = ifindex2ifnet[mifcp->mif6c_pifi]) == NULL) 659 return ENXIO; 660 661 if (mifcp->mif6c_flags & MIFF_REGISTER) { 662 if (reg_mif_num == (mifi_t)-1) { 663 strlcpy(multicast_register_if.if_xname, 664 "register_mif", 665 sizeof(multicast_register_if.if_xname)); 666 multicast_register_if.if_flags |= IFF_LOOPBACK; 667 multicast_register_if.if_index = mifcp->mif6c_mifi; 668 reg_mif_num = mifcp->mif6c_mifi; 669 } 670 671 ifp = &multicast_register_if; 672 673 } /* if REGISTER */ 674 else { 675 /* Make sure the interface supports multicast */ 676 if ((ifp->if_flags & IFF_MULTICAST) == 0) 677 return EOPNOTSUPP; 678 679 s = splsoftnet(); 680 /* 681 * Enable promiscuous reception of all IPv6 multicasts 682 * from the interface. 683 */ 684 ifr.ifr_addr.sin6_family = AF_INET6; 685 ifr.ifr_addr.sin6_addr = in6addr_any; 686 error = (*ifp->if_ioctl)(ifp, SIOCADDMULTI, (caddr_t)&ifr); 687 splx(s); 688 if (error) 689 return error; 690 } 691 692 s = splsoftnet(); 693 mifp->m6_flags = mifcp->mif6c_flags; 694 mifp->m6_ifp = ifp; 695 #ifdef notyet 696 /* scaling up here allows division by 1024 in critical code */ 697 mifp->m6_rate_limit = mifcp->mif6c_rate_limit * 1024 / 1000; 698 #endif 699 /* initialize per mif pkt counters */ 700 mifp->m6_pkt_in = 0; 701 mifp->m6_pkt_out = 0; 702 mifp->m6_bytes_in = 0; 703 mifp->m6_bytes_out = 0; 704 splx(s); 705 706 /* Adjust nummifs up if the mifi is higher than nummifs */ 707 if (nummifs <= mifcp->mif6c_mifi) 708 nummifs = mifcp->mif6c_mifi + 1; 709 710 #ifdef MRT6DEBUG 711 if (mrt6debug) 712 log(LOG_DEBUG, 713 "add_mif #%d, phyint %s%d\n", 714 mifcp->mif6c_mifi, 715 ifp->if_name, ifp->if_unit); 716 #endif 717 718 return 0; 719 } 720 721 /* 722 * Delete a mif from the mif table 723 */ 724 static int 725 del_m6if(mifip) 726 mifi_t *mifip; 727 { 728 struct mif6 *mifp = mif6table + *mifip; 729 mifi_t mifi; 730 struct ifnet *ifp; 731 struct in6_ifreq ifr; 732 int s; 733 734 if (*mifip >= nummifs) 735 return EINVAL; 736 if (mifp->m6_ifp == NULL) 737 return EINVAL; 738 739 s = splsoftnet(); 740 741 if (!(mifp->m6_flags & MIFF_REGISTER)) { 742 /* 743 * XXX: what if there is yet IPv4 multicast daemon 744 * using the interface? 745 */ 746 ifp = mifp->m6_ifp; 747 748 ifr.ifr_addr.sin6_family = AF_INET6; 749 ifr.ifr_addr.sin6_addr = in6addr_any; 750 (*ifp->if_ioctl)(ifp, SIOCDELMULTI, (caddr_t)&ifr); 751 } 752 753 #ifdef notyet 754 bzero((caddr_t)qtable[*mifip], sizeof(qtable[*mifip])); 755 bzero((caddr_t)mifp->m6_tbf, sizeof(*(mifp->m6_tbf))); 756 #endif 757 bzero((caddr_t)mifp, sizeof (*mifp)); 758 759 /* Adjust nummifs down */ 760 for (mifi = nummifs; mifi > 0; mifi--) 761 if (mif6table[mifi - 1].m6_ifp) 762 break; 763 nummifs = mifi; 764 765 splx(s); 766 767 #ifdef MRT6DEBUG 768 if (mrt6debug) 769 log(LOG_DEBUG, "del_m6if %d, nummifs %d\n", *mifip, nummifs); 770 #endif 771 772 return 0; 773 } 774 775 /* 776 * Add an mfc entry 777 */ 778 static int 779 add_m6fc(mfccp) 780 struct mf6cctl *mfccp; 781 { 782 struct mf6c *rt; 783 u_long hash; 784 struct rtdetq *rte; 785 u_short nstl; 786 int s; 787 788 MF6CFIND(mfccp->mf6cc_origin.sin6_addr, 789 mfccp->mf6cc_mcastgrp.sin6_addr, rt); 790 791 /* If an entry already exists, just update the fields */ 792 if (rt) { 793 #ifdef MRT6DEBUG 794 if (mrt6debug & DEBUG_MFC) 795 log(LOG_DEBUG,"add_m6fc update o %s g %s p %x\n", 796 ip6_sprintf(&mfccp->mf6cc_origin.sin6_addr), 797 ip6_sprintf(&mfccp->mf6cc_mcastgrp.sin6_addr), 798 mfccp->mf6cc_parent); 799 #endif 800 801 s = splsoftnet(); 802 rt->mf6c_parent = mfccp->mf6cc_parent; 803 rt->mf6c_ifset = mfccp->mf6cc_ifset; 804 splx(s); 805 return 0; 806 } 807 808 /* 809 * Find the entry for which the upcall was made and update 810 */ 811 s = splsoftnet(); 812 hash = MF6CHASH(mfccp->mf6cc_origin.sin6_addr, 813 mfccp->mf6cc_mcastgrp.sin6_addr); 814 for (rt = mf6ctable[hash], nstl = 0; rt; rt = rt->mf6c_next) { 815 if (IN6_ARE_ADDR_EQUAL(&rt->mf6c_origin.sin6_addr, 816 &mfccp->mf6cc_origin.sin6_addr) && 817 IN6_ARE_ADDR_EQUAL(&rt->mf6c_mcastgrp.sin6_addr, 818 &mfccp->mf6cc_mcastgrp.sin6_addr) && 819 (rt->mf6c_stall != NULL)) { 820 821 if (nstl++) 822 log(LOG_ERR, 823 "add_m6fc: %s o %s g %s p %x dbx %p\n", 824 "multiple kernel entries", 825 ip6_sprintf(&mfccp->mf6cc_origin.sin6_addr), 826 ip6_sprintf(&mfccp->mf6cc_mcastgrp.sin6_addr), 827 mfccp->mf6cc_parent, rt->mf6c_stall); 828 829 #ifdef MRT6DEBUG 830 if (mrt6debug & DEBUG_MFC) 831 log(LOG_DEBUG, 832 "add_m6fc o %s g %s p %x dbg %x\n", 833 ip6_sprintf(&mfccp->mf6cc_origin.sin6_addr), 834 ip6_sprintf(&mfccp->mf6cc_mcastgrp.sin6_addr), 835 mfccp->mf6cc_parent, rt->mf6c_stall); 836 #endif 837 838 rt->mf6c_origin = mfccp->mf6cc_origin; 839 rt->mf6c_mcastgrp = mfccp->mf6cc_mcastgrp; 840 rt->mf6c_parent = mfccp->mf6cc_parent; 841 rt->mf6c_ifset = mfccp->mf6cc_ifset; 842 /* initialize pkt counters per src-grp */ 843 rt->mf6c_pkt_cnt = 0; 844 rt->mf6c_byte_cnt = 0; 845 rt->mf6c_wrong_if = 0; 846 847 rt->mf6c_expire = 0; /* Don't clean this guy up */ 848 n6expire[hash]--; 849 850 /* free packets Qed at the end of this entry */ 851 for (rte = rt->mf6c_stall; rte != NULL; ) { 852 struct rtdetq *n = rte->next; 853 if (rte->ifp) { 854 ip6_mdq(rte->m, rte->ifp, rt); 855 } 856 m_freem(rte->m); 857 #ifdef UPCALL_TIMING 858 collate(&(rte->t)); 859 #endif /* UPCALL_TIMING */ 860 free(rte, M_MRTABLE); 861 rte = n; 862 } 863 rt->mf6c_stall = NULL; 864 } 865 } 866 867 /* 868 * It is possible that an entry is being inserted without an upcall 869 */ 870 if (nstl == 0) { 871 #ifdef MRT6DEBUG 872 if (mrt6debug & DEBUG_MFC) 873 log(LOG_DEBUG, 874 "add_mfc no upcall h %d o %s g %s p %x\n", 875 hash, 876 ip6_sprintf(&mfccp->mf6cc_origin.sin6_addr), 877 ip6_sprintf(&mfccp->mf6cc_mcastgrp.sin6_addr), 878 mfccp->mf6cc_parent); 879 #endif 880 881 for (rt = mf6ctable[hash]; rt; rt = rt->mf6c_next) { 882 883 if (IN6_ARE_ADDR_EQUAL(&rt->mf6c_origin.sin6_addr, 884 &mfccp->mf6cc_origin.sin6_addr)&& 885 IN6_ARE_ADDR_EQUAL(&rt->mf6c_mcastgrp.sin6_addr, 886 &mfccp->mf6cc_mcastgrp.sin6_addr)) { 887 888 rt->mf6c_origin = mfccp->mf6cc_origin; 889 rt->mf6c_mcastgrp = mfccp->mf6cc_mcastgrp; 890 rt->mf6c_parent = mfccp->mf6cc_parent; 891 rt->mf6c_ifset = mfccp->mf6cc_ifset; 892 /* initialize pkt counters per src-grp */ 893 rt->mf6c_pkt_cnt = 0; 894 rt->mf6c_byte_cnt = 0; 895 rt->mf6c_wrong_if = 0; 896 897 if (rt->mf6c_expire) 898 n6expire[hash]--; 899 rt->mf6c_expire = 0; 900 } 901 } 902 if (rt == NULL) { 903 /* no upcall, so make a new entry */ 904 rt = (struct mf6c *)malloc(sizeof(*rt), M_MRTABLE, 905 M_NOWAIT); 906 if (rt == NULL) { 907 splx(s); 908 return ENOBUFS; 909 } 910 911 /* insert new entry at head of hash chain */ 912 rt->mf6c_origin = mfccp->mf6cc_origin; 913 rt->mf6c_mcastgrp = mfccp->mf6cc_mcastgrp; 914 rt->mf6c_parent = mfccp->mf6cc_parent; 915 rt->mf6c_ifset = mfccp->mf6cc_ifset; 916 /* initialize pkt counters per src-grp */ 917 rt->mf6c_pkt_cnt = 0; 918 rt->mf6c_byte_cnt = 0; 919 rt->mf6c_wrong_if = 0; 920 rt->mf6c_expire = 0; 921 rt->mf6c_stall = NULL; 922 923 /* link into table */ 924 rt->mf6c_next = mf6ctable[hash]; 925 mf6ctable[hash] = rt; 926 } 927 } 928 splx(s); 929 return 0; 930 } 931 932 #ifdef UPCALL_TIMING 933 /* 934 * collect delay statistics on the upcalls 935 */ 936 static void 937 collate(t) 938 struct timeval *t; 939 { 940 u_long d; 941 struct timeval tp; 942 u_long delta; 943 944 GET_TIME(tp); 945 946 if (TV_LT(*t, tp)) 947 { 948 TV_DELTA(tp, *t, delta); 949 950 d = delta >> 10; 951 if (d > UPCALL_MAX) 952 d = UPCALL_MAX; 953 954 ++upcall_data[d]; 955 } 956 } 957 #endif /* UPCALL_TIMING */ 958 959 /* 960 * Delete an mfc entry 961 */ 962 static int 963 del_m6fc(mfccp) 964 struct mf6cctl *mfccp; 965 { 966 struct sockaddr_in6 origin; 967 struct sockaddr_in6 mcastgrp; 968 struct mf6c *rt; 969 struct mf6c **nptr; 970 u_long hash; 971 int s; 972 973 origin = mfccp->mf6cc_origin; 974 mcastgrp = mfccp->mf6cc_mcastgrp; 975 hash = MF6CHASH(origin.sin6_addr, mcastgrp.sin6_addr); 976 977 #ifdef MRT6DEBUG 978 if (mrt6debug & DEBUG_MFC) 979 log(LOG_DEBUG,"del_m6fc orig %s mcastgrp %s\n", 980 ip6_sprintf(&origin.sin6_addr), 981 ip6_sprintf(&mcastgrp.sin6_addr)); 982 #endif 983 984 s = splsoftnet(); 985 986 nptr = &mf6ctable[hash]; 987 while ((rt = *nptr) != NULL) { 988 if (IN6_ARE_ADDR_EQUAL(&origin.sin6_addr, 989 &rt->mf6c_origin.sin6_addr) && 990 IN6_ARE_ADDR_EQUAL(&mcastgrp.sin6_addr, 991 &rt->mf6c_mcastgrp.sin6_addr) && 992 rt->mf6c_stall == NULL) 993 break; 994 995 nptr = &rt->mf6c_next; 996 } 997 if (rt == NULL) { 998 splx(s); 999 return EADDRNOTAVAIL; 1000 } 1001 1002 *nptr = rt->mf6c_next; 1003 free(rt, M_MRTABLE); 1004 1005 splx(s); 1006 1007 return 0; 1008 } 1009 1010 static int 1011 socket_send(s, mm, src) 1012 struct socket *s; 1013 struct mbuf *mm; 1014 struct sockaddr_in6 *src; 1015 { 1016 if (s) { 1017 if (sbappendaddr(&s->so_rcv, 1018 (struct sockaddr *)src, 1019 mm, (struct mbuf *)0) != 0) { 1020 sorwakeup(s); 1021 return 0; 1022 } 1023 } 1024 m_freem(mm); 1025 return -1; 1026 } 1027 1028 /* 1029 * IPv6 multicast forwarding function. This function assumes that the packet 1030 * pointed to by "ip6" has arrived on (or is about to be sent to) the interface 1031 * pointed to by "ifp", and the packet is to be relayed to other networks 1032 * that have members of the packet's destination IPv6 multicast group. 1033 * 1034 * The packet is returned unscathed to the caller, unless it is 1035 * erroneous, in which case a non-zero return value tells the caller to 1036 * discard it. 1037 */ 1038 1039 int 1040 ip6_mforward(ip6, ifp, m) 1041 struct ip6_hdr *ip6; 1042 struct ifnet *ifp; 1043 struct mbuf *m; 1044 { 1045 struct mf6c *rt; 1046 struct mif6 *mifp; 1047 struct mbuf *mm; 1048 int s; 1049 mifi_t mifi; 1050 struct sockaddr_in6 sin6; 1051 1052 #ifdef MRT6DEBUG 1053 if (mrt6debug & DEBUG_FORWARD) 1054 log(LOG_DEBUG, "ip6_mforward: src %s, dst %s, ifindex %d\n", 1055 ip6_sprintf(&ip6->ip6_src), ip6_sprintf(&ip6->ip6_dst), 1056 ifp->if_index); 1057 #endif 1058 1059 /* 1060 * Don't forward a packet with Hop limit of zero or one, 1061 * or a packet destined to a local-only group. 1062 */ 1063 if (ip6->ip6_hlim <= 1 || IN6_IS_ADDR_MC_NODELOCAL(&ip6->ip6_dst) || 1064 IN6_IS_ADDR_MC_LINKLOCAL(&ip6->ip6_dst)) 1065 return 0; 1066 ip6->ip6_hlim--; 1067 1068 /* 1069 * Source address check: do not forward packets with unspecified 1070 * source. It was discussed in July 2000, on ipngwg mailing list. 1071 * This is rather more serious than unicast cases, because some 1072 * MLD packets can be sent with the unspecified source address 1073 * (although such packets must normally set 1 to the hop limit field). 1074 */ 1075 if (IN6_IS_ADDR_UNSPECIFIED(&ip6->ip6_src)) { 1076 ip6stat.ip6s_cantforward++; 1077 if (ip6_log_time + ip6_log_interval < time.tv_sec) { 1078 ip6_log_time = time.tv_sec; 1079 log(LOG_DEBUG, 1080 "cannot forward " 1081 "from %s to %s nxt %d received on %s\n", 1082 ip6_sprintf(&ip6->ip6_src), 1083 ip6_sprintf(&ip6->ip6_dst), 1084 ip6->ip6_nxt, 1085 m->m_pkthdr.rcvif ? 1086 if_name(m->m_pkthdr.rcvif) : "?"); 1087 } 1088 return 0; 1089 } 1090 1091 /* 1092 * Determine forwarding mifs from the forwarding cache table 1093 */ 1094 s = splsoftnet(); 1095 MF6CFIND(ip6->ip6_src, ip6->ip6_dst, rt); 1096 1097 /* Entry exists, so forward if necessary */ 1098 if (rt) { 1099 splx(s); 1100 return (ip6_mdq(m, ifp, rt)); 1101 } else { 1102 /* 1103 * If we don't have a route for packet's origin, 1104 * Make a copy of the packet & 1105 * send message to routing daemon 1106 */ 1107 1108 struct mbuf *mb0; 1109 struct rtdetq *rte; 1110 u_long hash; 1111 /* int i, npkts;*/ 1112 #ifdef UPCALL_TIMING 1113 struct timeval tp; 1114 1115 GET_TIME(tp); 1116 #endif /* UPCALL_TIMING */ 1117 1118 mrt6stat.mrt6s_no_route++; 1119 #ifdef MRT6DEBUG 1120 if (mrt6debug & (DEBUG_FORWARD | DEBUG_MFC)) 1121 log(LOG_DEBUG, "ip6_mforward: no rte s %s g %s\n", 1122 ip6_sprintf(&ip6->ip6_src), 1123 ip6_sprintf(&ip6->ip6_dst)); 1124 #endif 1125 1126 /* 1127 * Allocate mbufs early so that we don't do extra work if we 1128 * are just going to fail anyway. 1129 */ 1130 rte = (struct rtdetq *)malloc(sizeof(*rte), M_MRTABLE, 1131 M_NOWAIT); 1132 if (rte == NULL) { 1133 splx(s); 1134 return ENOBUFS; 1135 } 1136 mb0 = m_copy(m, 0, M_COPYALL); 1137 /* 1138 * Pullup packet header if needed before storing it, 1139 * as other references may modify it in the meantime. 1140 */ 1141 if (mb0 && 1142 (M_READONLY(mb0) || mb0->m_len < sizeof(struct ip6_hdr))) 1143 mb0 = m_pullup(mb0, sizeof(struct ip6_hdr)); 1144 if (mb0 == NULL) { 1145 free(rte, M_MRTABLE); 1146 splx(s); 1147 return ENOBUFS; 1148 } 1149 1150 /* is there an upcall waiting for this packet? */ 1151 hash = MF6CHASH(ip6->ip6_src, ip6->ip6_dst); 1152 for (rt = mf6ctable[hash]; rt; rt = rt->mf6c_next) { 1153 if (IN6_ARE_ADDR_EQUAL(&ip6->ip6_src, 1154 &rt->mf6c_origin.sin6_addr) && 1155 IN6_ARE_ADDR_EQUAL(&ip6->ip6_dst, 1156 &rt->mf6c_mcastgrp.sin6_addr) && 1157 (rt->mf6c_stall != NULL)) 1158 break; 1159 } 1160 1161 if (rt == NULL) { 1162 struct mrt6msg *im; 1163 struct omrt6msg *oim; 1164 1165 /* no upcall, so make a new entry */ 1166 rt = (struct mf6c *)malloc(sizeof(*rt), M_MRTABLE, 1167 M_NOWAIT); 1168 if (rt == NULL) { 1169 free(rte, M_MRTABLE); 1170 m_freem(mb0); 1171 splx(s); 1172 return ENOBUFS; 1173 } 1174 /* 1175 * Make a copy of the header to send to the user 1176 * level process 1177 */ 1178 mm = m_copy(mb0, 0, sizeof(struct ip6_hdr)); 1179 1180 if (mm == NULL) { 1181 free(rte, M_MRTABLE); 1182 m_freem(mb0); 1183 free(rt, M_MRTABLE); 1184 splx(s); 1185 return ENOBUFS; 1186 } 1187 1188 /* 1189 * Send message to routing daemon 1190 */ 1191 (void)memset(&sin6, 0, sizeof(sin6)); 1192 sin6.sin6_len = sizeof(sin6); 1193 sin6.sin6_family = AF_INET6; 1194 sin6.sin6_addr = ip6->ip6_src; 1195 1196 im = NULL; 1197 oim = NULL; 1198 switch (ip6_mrouter_ver) { 1199 case MRT6_OINIT: 1200 oim = mtod(mm, struct omrt6msg *); 1201 oim->im6_msgtype = MRT6MSG_NOCACHE; 1202 oim->im6_mbz = 0; 1203 break; 1204 case MRT6_INIT: 1205 im = mtod(mm, struct mrt6msg *); 1206 im->im6_msgtype = MRT6MSG_NOCACHE; 1207 im->im6_mbz = 0; 1208 break; 1209 default: 1210 free(rte, M_MRTABLE); 1211 m_freem(mb0); 1212 free(rt, M_MRTABLE); 1213 splx(s); 1214 return EINVAL; 1215 } 1216 1217 #ifdef MRT6DEBUG 1218 if (mrt6debug & DEBUG_FORWARD) 1219 log(LOG_DEBUG, 1220 "getting the iif info in the kernel\n"); 1221 #endif 1222 1223 for (mifp = mif6table, mifi = 0; 1224 mifi < nummifs && mifp->m6_ifp != ifp; 1225 mifp++, mifi++) 1226 ; 1227 1228 switch (ip6_mrouter_ver) { 1229 case MRT6_OINIT: 1230 oim->im6_mif = mifi; 1231 break; 1232 case MRT6_INIT: 1233 im->im6_mif = mifi; 1234 break; 1235 } 1236 1237 if (socket_send(ip6_mrouter, mm, &sin6) < 0) { 1238 log(LOG_WARNING, "ip6_mforward: ip6_mrouter " 1239 "socket queue full\n"); 1240 mrt6stat.mrt6s_upq_sockfull++; 1241 free(rte, M_MRTABLE); 1242 m_freem(mb0); 1243 free(rt, M_MRTABLE); 1244 splx(s); 1245 return ENOBUFS; 1246 } 1247 1248 mrt6stat.mrt6s_upcalls++; 1249 1250 /* insert new entry at head of hash chain */ 1251 bzero(rt, sizeof(*rt)); 1252 rt->mf6c_origin.sin6_family = AF_INET6; 1253 rt->mf6c_origin.sin6_len = sizeof(struct sockaddr_in6); 1254 rt->mf6c_origin.sin6_addr = ip6->ip6_src; 1255 rt->mf6c_mcastgrp.sin6_family = AF_INET6; 1256 rt->mf6c_mcastgrp.sin6_len = sizeof(struct sockaddr_in6); 1257 rt->mf6c_mcastgrp.sin6_addr = ip6->ip6_dst; 1258 rt->mf6c_expire = UPCALL_EXPIRE; 1259 n6expire[hash]++; 1260 rt->mf6c_parent = MF6C_INCOMPLETE_PARENT; 1261 1262 /* link into table */ 1263 rt->mf6c_next = mf6ctable[hash]; 1264 mf6ctable[hash] = rt; 1265 /* Add this entry to the end of the queue */ 1266 rt->mf6c_stall = rte; 1267 } else { 1268 /* determine if q has overflowed */ 1269 struct rtdetq **p; 1270 int npkts = 0; 1271 1272 for (p = &rt->mf6c_stall; *p != NULL; p = &(*p)->next) 1273 if (++npkts > MAX_UPQ6) { 1274 mrt6stat.mrt6s_upq_ovflw++; 1275 free(rte, M_MRTABLE); 1276 m_freem(mb0); 1277 splx(s); 1278 return 0; 1279 } 1280 1281 /* Add this entry to the end of the queue */ 1282 *p = rte; 1283 } 1284 1285 rte->next = NULL; 1286 rte->m = mb0; 1287 rte->ifp = ifp; 1288 #ifdef UPCALL_TIMING 1289 rte->t = tp; 1290 #endif /* UPCALL_TIMING */ 1291 1292 splx(s); 1293 1294 return 0; 1295 } 1296 } 1297 1298 /* 1299 * Clean up cache entries if upcalls are not serviced 1300 * Call from the Slow Timeout mechanism, every half second. 1301 */ 1302 static void 1303 expire_upcalls(unused) 1304 void *unused; 1305 { 1306 struct rtdetq *rte; 1307 struct mf6c *mfc, **nptr; 1308 int i; 1309 int s; 1310 1311 s = splsoftnet(); 1312 for (i = 0; i < MF6CTBLSIZ; i++) { 1313 if (n6expire[i] == 0) 1314 continue; 1315 nptr = &mf6ctable[i]; 1316 while ((mfc = *nptr) != NULL) { 1317 rte = mfc->mf6c_stall; 1318 /* 1319 * Skip real cache entries 1320 * Make sure it wasn't marked to not expire (shouldn't happen) 1321 * If it expires now 1322 */ 1323 if (rte != NULL && 1324 mfc->mf6c_expire != 0 && 1325 --mfc->mf6c_expire == 0) { 1326 #ifdef MRT6DEBUG 1327 if (mrt6debug & DEBUG_EXPIRE) 1328 log(LOG_DEBUG, "expire_upcalls: expiring (%s %s)\n", 1329 ip6_sprintf(&mfc->mf6c_origin.sin6_addr), 1330 ip6_sprintf(&mfc->mf6c_mcastgrp.sin6_addr)); 1331 #endif 1332 /* 1333 * drop all the packets 1334 * free the mbuf with the pkt, if, timing info 1335 */ 1336 do { 1337 struct rtdetq *n = rte->next; 1338 m_freem(rte->m); 1339 free(rte, M_MRTABLE); 1340 rte = n; 1341 } while (rte != NULL); 1342 mrt6stat.mrt6s_cache_cleanups++; 1343 n6expire[i]--; 1344 1345 *nptr = mfc->mf6c_next; 1346 free(mfc, M_MRTABLE); 1347 } else { 1348 nptr = &mfc->mf6c_next; 1349 } 1350 } 1351 } 1352 splx(s); 1353 callout_reset(&expire_upcalls_ch, EXPIRE_TIMEOUT, 1354 expire_upcalls, NULL); 1355 } 1356 1357 /* 1358 * Packet forwarding routine once entry in the cache is made 1359 */ 1360 static int 1361 ip6_mdq(m, ifp, rt) 1362 struct mbuf *m; 1363 struct ifnet *ifp; 1364 struct mf6c *rt; 1365 { 1366 struct ip6_hdr *ip6 = mtod(m, struct ip6_hdr *); 1367 mifi_t mifi, iif; 1368 struct mif6 *mifp; 1369 int plen = m->m_pkthdr.len; 1370 1371 /* 1372 * Macro to send packet on mif. Since RSVP packets don't get counted on 1373 * input, they shouldn't get counted on output, so statistics keeping is 1374 * separate. 1375 */ 1376 1377 #define MC6_SEND(ip6, mifp, m) do { \ 1378 if ((mifp)->m6_flags & MIFF_REGISTER) \ 1379 register_send((ip6), (mifp), (m)); \ 1380 else \ 1381 phyint_send((ip6), (mifp), (m)); \ 1382 } while (/*CONSTCOND*/ 0) 1383 1384 /* 1385 * Don't forward if it didn't arrive from the parent mif 1386 * for its origin. 1387 */ 1388 mifi = rt->mf6c_parent; 1389 if ((mifi >= nummifs) || (mif6table[mifi].m6_ifp != ifp)) { 1390 /* came in the wrong interface */ 1391 #ifdef MRT6DEBUG 1392 if (mrt6debug & DEBUG_FORWARD) 1393 log(LOG_DEBUG, 1394 "wrong if: ifid %d mifi %d mififid %x\n", 1395 ifp->if_index, mifi, 1396 mif6table[mifi].m6_ifp ? 1397 mif6table[mifi].m6_ifp->if_index : -1); 1398 #endif 1399 mrt6stat.mrt6s_wrong_if++; 1400 rt->mf6c_wrong_if++; 1401 /* 1402 * If we are doing PIM processing, and we are forwarding 1403 * packets on this interface, send a message to the 1404 * routing daemon. 1405 */ 1406 /* have to make sure this is a valid mif */ 1407 if (mifi < nummifs && mif6table[mifi].m6_ifp) 1408 if (pim6 && (m->m_flags & M_LOOP) == 0) { 1409 /* 1410 * Check the M_LOOP flag to avoid an 1411 * unnecessary PIM assert. 1412 * XXX: M_LOOP is an ad-hoc hack... 1413 */ 1414 struct sockaddr_in6 sin6; 1415 1416 struct mbuf *mm; 1417 struct mrt6msg *im; 1418 struct omrt6msg *oim; 1419 1420 mm = m_copy(m, 0, sizeof(struct ip6_hdr)); 1421 if (mm && 1422 (M_READONLY(mm) || 1423 mm->m_len < sizeof(struct ip6_hdr))) 1424 mm = m_pullup(mm, sizeof(struct ip6_hdr)); 1425 if (mm == NULL) 1426 return ENOBUFS; 1427 1428 oim = NULL; 1429 im = NULL; 1430 switch (ip6_mrouter_ver) { 1431 case MRT6_OINIT: 1432 oim = mtod(mm, struct omrt6msg *); 1433 oim->im6_msgtype = MRT6MSG_WRONGMIF; 1434 oim->im6_mbz = 0; 1435 break; 1436 case MRT6_INIT: 1437 im = mtod(mm, struct mrt6msg *); 1438 im->im6_msgtype = MRT6MSG_WRONGMIF; 1439 im->im6_mbz = 0; 1440 break; 1441 default: 1442 m_freem(mm); 1443 return EINVAL; 1444 } 1445 1446 for (mifp = mif6table, iif = 0; 1447 iif < nummifs && mifp && 1448 mifp->m6_ifp != ifp; 1449 mifp++, iif++) 1450 ; 1451 1452 (void)memset(&sin6, 0, sizeof(sin6)); 1453 sin6.sin6_len = sizeof(sin6); 1454 sin6.sin6_family = AF_INET6; 1455 switch (ip6_mrouter_ver) { 1456 case MRT6_OINIT: 1457 oim->im6_mif = iif; 1458 sin6.sin6_addr = oim->im6_src; 1459 break; 1460 case MRT6_INIT: 1461 im->im6_mif = iif; 1462 sin6.sin6_addr = im->im6_src; 1463 break; 1464 } 1465 1466 mrt6stat.mrt6s_upcalls++; 1467 1468 if (socket_send(ip6_mrouter, mm, &sin6) < 0) { 1469 #ifdef MRT6DEBUG 1470 if (mrt6debug) 1471 log(LOG_WARNING, "mdq, ip6_mrouter socket queue full\n"); 1472 #endif 1473 ++mrt6stat.mrt6s_upq_sockfull; 1474 return ENOBUFS; 1475 } /* if socket Q full */ 1476 } /* if PIM */ 1477 return 0; 1478 } /* if wrong iif */ 1479 1480 /* If I sourced this packet, it counts as output, else it was input. */ 1481 if (m->m_pkthdr.rcvif == NULL) { 1482 /* XXX: is rcvif really NULL when output?? */ 1483 mif6table[mifi].m6_pkt_out++; 1484 mif6table[mifi].m6_bytes_out += plen; 1485 } else { 1486 mif6table[mifi].m6_pkt_in++; 1487 mif6table[mifi].m6_bytes_in += plen; 1488 } 1489 rt->mf6c_pkt_cnt++; 1490 rt->mf6c_byte_cnt += plen; 1491 1492 /* 1493 * For each mif, forward a copy of the packet if there are group 1494 * members downstream on the interface. 1495 */ 1496 for (mifp = mif6table, mifi = 0; mifi < nummifs; mifp++, mifi++) 1497 if (IF_ISSET(mifi, &rt->mf6c_ifset)) { 1498 if (mif6table[mifi].m6_ifp == NULL) 1499 continue; 1500 #ifdef notyet 1501 /* 1502 * check if the outgoing packet is going to break 1503 * a scope boundary. 1504 * XXX For packets through PIM register tunnel 1505 * interface, we believe a routing daemon. 1506 */ 1507 if ((mif6table[rt->mf6c_parent].m6_flags & 1508 MIFF_REGISTER) == 0 && 1509 (mif6table[mifi].m6_flags & MIFF_REGISTER) == 0 && 1510 (in6_addr2scopeid(ifp, &ip6->ip6_dst) != 1511 in6_addr2scopeid(mif6table[mifi].m6_ifp, 1512 &ip6->ip6_dst) || 1513 in6_addr2scopeid(ifp, &ip6->ip6_src) != 1514 in6_addr2scopeid(mif6table[mifi].m6_ifp, 1515 &ip6->ip6_src))) { 1516 ip6stat.ip6s_badscope++; 1517 continue; 1518 } 1519 #endif 1520 1521 mifp->m6_pkt_out++; 1522 mifp->m6_bytes_out += plen; 1523 MC6_SEND(ip6, mifp, m); 1524 } 1525 return 0; 1526 } 1527 1528 static void 1529 phyint_send(ip6, mifp, m) 1530 struct ip6_hdr *ip6; 1531 struct mif6 *mifp; 1532 struct mbuf *m; 1533 { 1534 struct mbuf *mb_copy; 1535 struct ifnet *ifp = mifp->m6_ifp; 1536 int error = 0; 1537 int s = splsoftnet(); 1538 static struct route_in6 ro; 1539 struct in6_multi *in6m; 1540 struct sockaddr_in6 *dst6; 1541 u_long linkmtu; 1542 1543 /* 1544 * Make a new reference to the packet; make sure that 1545 * the IPv6 header is actually copied, not just referenced, 1546 * so that ip6_output() only scribbles on the copy. 1547 */ 1548 mb_copy = m_copy(m, 0, M_COPYALL); 1549 if (mb_copy && 1550 (M_READONLY(mb_copy) || mb_copy->m_len < sizeof(struct ip6_hdr))) 1551 mb_copy = m_pullup(mb_copy, sizeof(struct ip6_hdr)); 1552 if (mb_copy == NULL) { 1553 splx(s); 1554 return; 1555 } 1556 /* set MCAST flag to the outgoing packet */ 1557 mb_copy->m_flags |= M_MCAST; 1558 1559 /* 1560 * If we sourced the packet, call ip6_output since we may devide 1561 * the packet into fragments when the packet is too big for the 1562 * outgoing interface. 1563 * Otherwise, we can simply send the packet to the interface 1564 * sending queue. 1565 */ 1566 if (m->m_pkthdr.rcvif == NULL) { 1567 struct ip6_moptions im6o; 1568 1569 im6o.im6o_multicast_ifp = ifp; 1570 /* XXX: ip6_output will override ip6->ip6_hlim */ 1571 im6o.im6o_multicast_hlim = ip6->ip6_hlim; 1572 im6o.im6o_multicast_loop = 1; 1573 error = ip6_output(mb_copy, NULL, &ro, IPV6_FORWARDING, 1574 &im6o, (struct socket *)0, NULL); 1575 1576 #ifdef MRT6DEBUG 1577 if (mrt6debug & DEBUG_XMIT) 1578 log(LOG_DEBUG, "phyint_send on mif %d err %d\n", 1579 mifp - mif6table, error); 1580 #endif 1581 splx(s); 1582 return; 1583 } 1584 1585 /* 1586 * If we belong to the destination multicast group 1587 * on the outgoing interface, loop back a copy. 1588 */ 1589 dst6 = (struct sockaddr_in6 *)&ro.ro_dst; 1590 IN6_LOOKUP_MULTI(ip6->ip6_dst, ifp, in6m); 1591 if (in6m != NULL) { 1592 dst6->sin6_len = sizeof(struct sockaddr_in6); 1593 dst6->sin6_family = AF_INET6; 1594 dst6->sin6_addr = ip6->ip6_dst; 1595 ip6_mloopback(ifp, m, (struct sockaddr_in6 *)&ro.ro_dst); 1596 } 1597 /* 1598 * Put the packet into the sending queue of the outgoing interface 1599 * if it would fit in the MTU of the interface. 1600 */ 1601 linkmtu = IN6_LINKMTU(ifp); 1602 if (mb_copy->m_pkthdr.len <= linkmtu || linkmtu < IPV6_MMTU) { 1603 dst6->sin6_len = sizeof(struct sockaddr_in6); 1604 dst6->sin6_family = AF_INET6; 1605 dst6->sin6_addr = ip6->ip6_dst; 1606 /* 1607 * We just call if_output instead of nd6_output here, since 1608 * we need no ND for a multicast forwarded packet...right? 1609 */ 1610 error = (*ifp->if_output)(ifp, mb_copy, 1611 (struct sockaddr *)&ro.ro_dst, NULL); 1612 #ifdef MRT6DEBUG 1613 if (mrt6debug & DEBUG_XMIT) 1614 log(LOG_DEBUG, "phyint_send on mif %d err %d\n", 1615 mifp - mif6table, error); 1616 #endif 1617 } else { 1618 #ifdef MULTICAST_PMTUD 1619 icmp6_error(mb_copy, ICMP6_PACKET_TOO_BIG, 0, linkmtu); 1620 #else 1621 #ifdef MRT6DEBUG 1622 if (mrt6debug & DEBUG_XMIT) 1623 log(LOG_DEBUG, 1624 "phyint_send: packet too big on %s o %s g %s" 1625 " size %d(discarded)\n", 1626 if_name(ifp), 1627 ip6_sprintf(&ip6->ip6_src), 1628 ip6_sprintf(&ip6->ip6_dst), 1629 mb_copy->m_pkthdr.len); 1630 #endif /* MRT6DEBUG */ 1631 m_freem(mb_copy); /* simply discard the packet */ 1632 #endif 1633 } 1634 1635 splx(s); 1636 } 1637 1638 static int 1639 register_send(ip6, mif, m) 1640 struct ip6_hdr *ip6; 1641 struct mif6 *mif; 1642 struct mbuf *m; 1643 { 1644 struct mbuf *mm; 1645 int i, len = m->m_pkthdr.len; 1646 struct sockaddr_in6 sin6; 1647 struct mrt6msg *im6; 1648 1649 #ifdef MRT6DEBUG 1650 if (mrt6debug) 1651 log(LOG_DEBUG, "** IPv6 register_send **\n src %s dst %s\n", 1652 ip6_sprintf(&ip6->ip6_src), ip6_sprintf(&ip6->ip6_dst)); 1653 #endif 1654 ++pim6stat.pim6s_snd_registers; 1655 1656 /* Make a copy of the packet to send to the user level process */ 1657 MGETHDR(mm, M_DONTWAIT, MT_HEADER); 1658 if (mm == NULL) 1659 return ENOBUFS; 1660 mm->m_data += max_linkhdr; 1661 mm->m_len = sizeof(struct ip6_hdr); 1662 1663 if ((mm->m_next = m_copy(m, 0, M_COPYALL)) == NULL) { 1664 m_freem(mm); 1665 return ENOBUFS; 1666 } 1667 i = MHLEN - M_LEADINGSPACE(mm); 1668 if (i > len) 1669 i = len; 1670 mm = m_pullup(mm, i); 1671 if (mm == NULL) 1672 return ENOBUFS; 1673 /* TODO: check it! */ 1674 mm->m_pkthdr.len = len + sizeof(struct ip6_hdr); 1675 1676 /* 1677 * Send message to routing daemon 1678 */ 1679 (void)memset(&sin6, 0, sizeof(sin6)); 1680 sin6.sin6_len = sizeof(sin6); 1681 sin6.sin6_family = AF_INET6; 1682 sin6.sin6_addr = ip6->ip6_src; 1683 1684 im6 = mtod(mm, struct mrt6msg *); 1685 im6->im6_msgtype = MRT6MSG_WHOLEPKT; 1686 im6->im6_mbz = 0; 1687 1688 im6->im6_mif = mif - mif6table; 1689 1690 /* iif info is not given for reg. encap.n */ 1691 mrt6stat.mrt6s_upcalls++; 1692 1693 if (socket_send(ip6_mrouter, mm, &sin6) < 0) { 1694 #ifdef MRT6DEBUG 1695 if (mrt6debug) 1696 log(LOG_WARNING, 1697 "register_send: ip6_mrouter socket queue full\n"); 1698 #endif 1699 ++mrt6stat.mrt6s_upq_sockfull; 1700 return ENOBUFS; 1701 } 1702 return 0; 1703 } 1704 1705 /* 1706 * PIM sparse mode hook 1707 * Receives the pim control messages, and passes them up to the listening 1708 * socket, using rip6_input. 1709 * The only message processed is the REGISTER pim message; the pim header 1710 * is stripped off, and the inner packet is passed to register_mforward. 1711 */ 1712 int 1713 pim6_input(mp, offp, proto) 1714 struct mbuf **mp; 1715 int *offp, proto; 1716 { 1717 struct pim *pim; /* pointer to a pim struct */ 1718 struct ip6_hdr *ip6; 1719 int pimlen; 1720 struct mbuf *m = *mp; 1721 int minlen; 1722 int off = *offp; 1723 1724 ++pim6stat.pim6s_rcv_total; 1725 1726 ip6 = mtod(m, struct ip6_hdr *); 1727 pimlen = m->m_pkthdr.len - *offp; 1728 1729 /* 1730 * Validate lengths 1731 */ 1732 if (pimlen < PIM_MINLEN) { 1733 ++pim6stat.pim6s_rcv_tooshort; 1734 #ifdef MRT6DEBUG 1735 if (mrt6debug & DEBUG_PIM) 1736 log(LOG_DEBUG,"pim6_input: PIM packet too short\n"); 1737 #endif 1738 m_freem(m); 1739 return (IPPROTO_DONE); 1740 } 1741 1742 /* 1743 * if the packet is at least as big as a REGISTER, go ahead 1744 * and grab the PIM REGISTER header size, to avoid another 1745 * possible m_pullup() later. 1746 * 1747 * PIM_MINLEN == pimhdr + u_int32 == 8 1748 * PIM6_REG_MINLEN == pimhdr + reghdr + eip6hdr == 4 + 4 + 40 1749 */ 1750 minlen = (pimlen >= PIM6_REG_MINLEN) ? PIM6_REG_MINLEN : PIM_MINLEN; 1751 1752 /* 1753 * Make sure that the IP6 and PIM headers in contiguous memory, and 1754 * possibly the PIM REGISTER header 1755 */ 1756 IP6_EXTHDR_GET(pim, struct pim *, m, off, minlen); 1757 if (pim == NULL) { 1758 pim6stat.pim6s_rcv_tooshort++; 1759 return IPPROTO_DONE; 1760 } 1761 1762 /* PIM version check */ 1763 if (pim->pim_ver != PIM_VERSION) { 1764 ++pim6stat.pim6s_rcv_badversion; 1765 #ifdef MRT6DEBUG 1766 log(LOG_ERR, 1767 "pim6_input: incorrect version %d, expecting %d\n", 1768 pim->pim_ver, PIM_VERSION); 1769 #endif 1770 m_freem(m); 1771 return (IPPROTO_DONE); 1772 } 1773 1774 #define PIM6_CHECKSUM 1775 #ifdef PIM6_CHECKSUM 1776 { 1777 int cksumlen; 1778 1779 /* 1780 * Validate checksum. 1781 * If PIM REGISTER, exclude the data packet 1782 */ 1783 if (pim->pim_type == PIM_REGISTER) 1784 cksumlen = PIM_MINLEN; 1785 else 1786 cksumlen = pimlen; 1787 1788 if (in6_cksum(m, IPPROTO_PIM, off, cksumlen)) { 1789 ++pim6stat.pim6s_rcv_badsum; 1790 #ifdef MRT6DEBUG 1791 if (mrt6debug & DEBUG_PIM) 1792 log(LOG_DEBUG, 1793 "pim6_input: invalid checksum\n"); 1794 #endif 1795 m_freem(m); 1796 return (IPPROTO_DONE); 1797 } 1798 } 1799 #endif /* PIM_CHECKSUM */ 1800 1801 if (pim->pim_type == PIM_REGISTER) { 1802 /* 1803 * since this is a REGISTER, we'll make a copy of the register 1804 * headers ip6+pim+u_int32_t+encap_ip6, to be passed up to the 1805 * routing daemon. 1806 */ 1807 static struct sockaddr_in6 dst = { sizeof(dst), AF_INET6 }; 1808 1809 struct mbuf *mcp; 1810 struct ip6_hdr *eip6; 1811 u_int32_t *reghdr; 1812 1813 ++pim6stat.pim6s_rcv_registers; 1814 1815 if ((reg_mif_num >= nummifs) || (reg_mif_num == (mifi_t) -1)) { 1816 #ifdef MRT6DEBUG 1817 if (mrt6debug & DEBUG_PIM) 1818 log(LOG_DEBUG, 1819 "pim6_input: register mif not set: %d\n", 1820 reg_mif_num); 1821 #endif 1822 m_freem(m); 1823 return (IPPROTO_DONE); 1824 } 1825 1826 reghdr = (u_int32_t *)(pim + 1); 1827 1828 if ((ntohl(*reghdr) & PIM_NULL_REGISTER)) 1829 goto pim6_input_to_daemon; 1830 1831 /* 1832 * Validate length 1833 */ 1834 if (pimlen < PIM6_REG_MINLEN) { 1835 ++pim6stat.pim6s_rcv_tooshort; 1836 ++pim6stat.pim6s_rcv_badregisters; 1837 #ifdef MRT6DEBUG 1838 log(LOG_ERR, 1839 "pim6_input: register packet size too " 1840 "small %d from %s\n", 1841 pimlen, ip6_sprintf(&ip6->ip6_src)); 1842 #endif 1843 m_freem(m); 1844 return (IPPROTO_DONE); 1845 } 1846 1847 eip6 = (struct ip6_hdr *) (reghdr + 1); 1848 #ifdef MRT6DEBUG 1849 if (mrt6debug & DEBUG_PIM) 1850 log(LOG_DEBUG, 1851 "pim6_input[register], eip6: %s -> %s, " 1852 "eip6 plen %d\n", 1853 ip6_sprintf(&eip6->ip6_src), 1854 ip6_sprintf(&eip6->ip6_dst), 1855 ntohs(eip6->ip6_plen)); 1856 #endif 1857 1858 /* verify the version number of the inner packet */ 1859 if ((eip6->ip6_vfc & IPV6_VERSION_MASK) != IPV6_VERSION) { 1860 ++pim6stat.pim6s_rcv_badregisters; 1861 #ifdef MRT6DEBUG 1862 log(LOG_DEBUG, "pim6_input: invalid IP version (%d) " 1863 "of the inner packet\n", 1864 (eip6->ip6_vfc & IPV6_VERSION)); 1865 #endif 1866 m_freem(m); 1867 return (IPPROTO_NONE); 1868 } 1869 1870 /* verify the inner packet is destined to a mcast group */ 1871 if (!IN6_IS_ADDR_MULTICAST(&eip6->ip6_dst)) { 1872 ++pim6stat.pim6s_rcv_badregisters; 1873 #ifdef MRT6DEBUG 1874 if (mrt6debug & DEBUG_PIM) 1875 log(LOG_DEBUG, 1876 "pim6_input: inner packet of register " 1877 "is not multicast %s\n", 1878 ip6_sprintf(&eip6->ip6_dst)); 1879 #endif 1880 m_freem(m); 1881 return (IPPROTO_DONE); 1882 } 1883 1884 /* 1885 * make a copy of the whole header to pass to the daemon later. 1886 */ 1887 mcp = m_copy(m, 0, off + PIM6_REG_MINLEN); 1888 if (mcp == NULL) { 1889 #ifdef MRT6DEBUG 1890 log(LOG_ERR, 1891 "pim6_input: pim register: " 1892 "could not copy register head\n"); 1893 #endif 1894 m_freem(m); 1895 return (IPPROTO_DONE); 1896 } 1897 1898 /* 1899 * forward the inner ip6 packet; point m_data at the inner ip6. 1900 */ 1901 m_adj(m, off + PIM_MINLEN); 1902 #ifdef MRT6DEBUG 1903 if (mrt6debug & DEBUG_PIM) { 1904 log(LOG_DEBUG, 1905 "pim6_input: forwarding decapsulated register: " 1906 "src %s, dst %s, mif %d\n", 1907 ip6_sprintf(&eip6->ip6_src), 1908 ip6_sprintf(&eip6->ip6_dst), 1909 reg_mif_num); 1910 } 1911 #endif 1912 1913 looutput(mif6table[reg_mif_num].m6_ifp, m, 1914 (struct sockaddr *) &dst, 1915 (struct rtentry *) NULL); 1916 1917 /* prepare the register head to send to the mrouting daemon */ 1918 m = mcp; 1919 } 1920 1921 /* 1922 * Pass the PIM message up to the daemon; if it is a register message 1923 * pass the 'head' only up to the daemon. This includes the 1924 * encapsulator ip6 header, pim header, register header and the 1925 * encapsulated ip6 header. 1926 */ 1927 pim6_input_to_daemon: 1928 rip6_input(&m, offp, proto); 1929 return (IPPROTO_DONE); 1930 } 1931