ip6_mroute.c revision 1.46
1 /* $NetBSD: ip6_mroute.c,v 1.46 2003/05/16 03:56:49 itojun 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) 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 * 3. All advertising materials mentioning features or use of this software 52 * must display the following acknowledgement: 53 * This product includes software developed by the University of 54 * California, Berkeley and its contributors. 55 * 4. Neither the name of the University nor the names of its contributors 56 * may be used to endorse or promote products derived from this software 57 * without specific prior written permission. 58 * 59 * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND 60 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE 61 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE 62 * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE 63 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL 64 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS 65 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) 66 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT 67 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY 68 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF 69 * SUCH DAMAGE. 70 * 71 * @(#)ip_mroute.c 8.2 (Berkeley) 11/15/93 72 */ 73 74 /* 75 * IP multicast forwarding procedures 76 * 77 * Written by David Waitzman, BBN Labs, August 1988. 78 * Modified by Steve Deering, Stanford, February 1989. 79 * Modified by Mark J. Steiglitz, Stanford, May, 1991 80 * Modified by Van Jacobson, LBL, January 1993 81 * Modified by Ajit Thyagarajan, PARC, August 1993 82 * Modified by Bill Fenner, PARC, April 1994 83 * 84 * MROUTING Revision: 3.5.1.2 + PIM-SMv2 (pimd) Support 85 */ 86 87 #include <sys/cdefs.h> 88 __KERNEL_RCSID(0, "$NetBSD: ip6_mroute.c,v 1.46 2003/05/16 03:56:49 itojun Exp $"); 89 90 #include "opt_inet.h" 91 92 #ifndef _KERNEL 93 # ifdef KERNEL 94 # define _KERNEL 95 # endif 96 #endif 97 98 #include <sys/param.h> 99 #include <sys/systm.h> 100 #include <sys/callout.h> 101 #include <sys/mbuf.h> 102 #include <sys/socket.h> 103 #include <sys/socketvar.h> 104 #include <sys/sockio.h> 105 #include <sys/protosw.h> 106 #include <sys/errno.h> 107 #include <sys/time.h> 108 #include <sys/kernel.h> 109 #include <sys/ioctl.h> 110 #include <sys/syslog.h> 111 112 #include <net/if.h> 113 #include <net/route.h> 114 #include <net/raw_cb.h> 115 116 #include <netinet/in.h> 117 #include <netinet/in_var.h> 118 #ifdef MULTICAST_PMTUD 119 #include <netinet/icmp6.h> 120 #endif 121 122 #include <netinet/ip6.h> 123 #include <netinet6/ip6_var.h> 124 #include <netinet6/ip6_mroute.h> 125 #include <netinet6/pim6.h> 126 #include <netinet6/pim6_var.h> 127 #include <netinet6/nd6.h> 128 129 #include <net/net_osdep.h> 130 131 static int ip6_mdq __P((struct mbuf *, struct ifnet *, struct mf6c *)); 132 static void phyint_send __P((struct ip6_hdr *, struct mif6 *, struct mbuf *)); 133 134 static int set_pim6 __P((int *)); 135 static int get_pim6 __P((struct mbuf *)); 136 static int socket_send __P((struct socket *, struct mbuf *, 137 struct sockaddr_in6 *)); 138 static int register_send __P((struct ip6_hdr *, struct mif6 *, 139 struct mbuf *)); 140 141 /* 142 * Globals. All but ip6_mrouter, ip6_mrtproto and mrt6stat could be static, 143 * except for netstat or debugging purposes. 144 */ 145 struct socket *ip6_mrouter = NULL; 146 int ip6_mrouter_ver = 0; 147 int ip6_mrtproto = IPPROTO_PIM; /* for netstat only */ 148 struct mrt6stat mrt6stat; 149 150 #define NO_RTE_FOUND 0x1 151 #define RTE_FOUND 0x2 152 153 struct mf6c *mf6ctable[MF6CTBLSIZ]; 154 u_char n6expire[MF6CTBLSIZ]; 155 static struct mif6 mif6table[MAXMIFS]; 156 #ifdef MRT6DEBUG 157 u_int mrt6debug = 0; /* debug level */ 158 #define DEBUG_MFC 0x02 159 #define DEBUG_FORWARD 0x04 160 #define DEBUG_EXPIRE 0x08 161 #define DEBUG_XMIT 0x10 162 #define DEBUG_REG 0x20 163 #define DEBUG_PIM 0x40 164 #endif 165 166 static void expire_upcalls __P((void *)); 167 #define EXPIRE_TIMEOUT (hz / 4) /* 4x / second */ 168 #define UPCALL_EXPIRE 6 /* number of timeouts */ 169 170 #ifdef INET 171 #ifdef MROUTING 172 extern struct socket *ip_mrouter; 173 #endif 174 #endif 175 176 /* 177 * 'Interfaces' associated with decapsulator (so we can tell 178 * packets that went through it from ones that get reflected 179 * by a broken gateway). These interfaces are never linked into 180 * the system ifnet list & no routes point to them. I.e., packets 181 * can't be sent this way. They only exist as a placeholder for 182 * multicast source verification. 183 */ 184 struct ifnet multicast_register_if; 185 186 #define ENCAP_HOPS 64 187 188 /* 189 * Private variables. 190 */ 191 static mifi_t nummifs = 0; 192 static mifi_t reg_mif_num = (mifi_t)-1; 193 194 static struct pim6stat pim6stat; 195 static int pim6; 196 197 /* 198 * Hash function for a source, group entry 199 */ 200 #define MF6CHASH(a, g) MF6CHASHMOD((a).s6_addr32[0] ^ (a).s6_addr32[1] ^ \ 201 (a).s6_addr32[2] ^ (a).s6_addr32[3] ^ \ 202 (g).s6_addr32[0] ^ (g).s6_addr32[1] ^ \ 203 (g).s6_addr32[2] ^ (g).s6_addr32[3]) 204 205 /* 206 * Find a route for a given origin IPv6 address and Multicast group address. 207 * Quality of service parameter to be added in the future!!! 208 */ 209 210 #define MF6CFIND(o, g, rt) do { \ 211 struct mf6c *_rt = mf6ctable[MF6CHASH(o,g)]; \ 212 rt = NULL; \ 213 mrt6stat.mrt6s_mfc_lookups++; \ 214 while (_rt) { \ 215 if (IN6_ARE_ADDR_EQUAL(&_rt->mf6c_origin.sin6_addr, &(o)) && \ 216 IN6_ARE_ADDR_EQUAL(&_rt->mf6c_mcastgrp.sin6_addr, &(g)) && \ 217 (_rt->mf6c_stall == NULL)) { \ 218 rt = _rt; \ 219 break; \ 220 } \ 221 _rt = _rt->mf6c_next; \ 222 } \ 223 if (rt == NULL) { \ 224 mrt6stat.mrt6s_mfc_misses++; \ 225 } \ 226 } while (/*CONSTCOND*/ 0) 227 228 /* 229 * Macros to compute elapsed time efficiently 230 * Borrowed from Van Jacobson's scheduling code 231 */ 232 #define TV_DELTA(a, b, delta) do { \ 233 int xxs; \ 234 \ 235 delta = (a).tv_usec - (b).tv_usec; \ 236 if ((xxs = (a).tv_sec - (b).tv_sec)) { \ 237 switch (xxs) { \ 238 case 2: \ 239 delta += 1000000; \ 240 /* FALLTHROUGH */ \ 241 case 1: \ 242 delta += 1000000; \ 243 break; \ 244 default: \ 245 delta += (1000000 * xxs); \ 246 } \ 247 } \ 248 } while (/*CONSTCOND*/ 0) 249 250 #define TV_LT(a, b) (((a).tv_usec < (b).tv_usec && \ 251 (a).tv_sec <= (b).tv_sec) || (a).tv_sec < (b).tv_sec) 252 253 #ifdef UPCALL_TIMING 254 #define UPCALL_MAX 50 255 u_long upcall_data[UPCALL_MAX + 1]; 256 static void collate(); 257 #endif /* UPCALL_TIMING */ 258 259 static int get_sg_cnt __P((struct sioc_sg_req6 *)); 260 static int get_mif6_cnt __P((struct sioc_mif_req6 *)); 261 static int ip6_mrouter_init __P((struct socket *, struct mbuf *, int)); 262 static int add_m6if __P((struct mif6ctl *)); 263 static int del_m6if __P((mifi_t *)); 264 static int add_m6fc __P((struct mf6cctl *)); 265 static int del_m6fc __P((struct mf6cctl *)); 266 267 static struct callout expire_upcalls_ch = CALLOUT_INITIALIZER; 268 269 /* 270 * Handle MRT setsockopt commands to modify the multicast routing tables. 271 */ 272 int 273 ip6_mrouter_set(cmd, so, m) 274 int cmd; 275 struct socket *so; 276 struct mbuf *m; 277 { 278 if (cmd != MRT6_INIT && so != ip6_mrouter) 279 return EACCES; 280 281 switch (cmd) { 282 case MRT6_OINIT: return ip6_mrouter_init(so, m, cmd); 283 case MRT6_INIT: return ip6_mrouter_init(so, m, cmd); 284 case MRT6_DONE: return ip6_mrouter_done(); 285 case MRT6_ADD_MIF: return add_m6if(mtod(m, struct mif6ctl *)); 286 case MRT6_DEL_MIF: return del_m6if(mtod(m, mifi_t *)); 287 case MRT6_ADD_MFC: return add_m6fc(mtod(m, struct mf6cctl *)); 288 case MRT6_DEL_MFC: return del_m6fc(mtod(m, struct mf6cctl *)); 289 case MRT6_PIM: return set_pim6(mtod(m, int *)); 290 default: return EOPNOTSUPP; 291 } 292 } 293 294 /* 295 * Handle MRT getsockopt commands 296 */ 297 int 298 ip6_mrouter_get(cmd, so, m) 299 int cmd; 300 struct socket *so; 301 struct mbuf **m; 302 { 303 struct mbuf *mb; 304 305 if (so != ip6_mrouter) return EACCES; 306 307 *m = mb = m_get(M_WAIT, MT_SOOPTS); 308 309 switch (cmd) { 310 case MRT6_PIM: 311 return get_pim6(mb); 312 default: 313 m_free(mb); 314 return EOPNOTSUPP; 315 } 316 } 317 318 /* 319 * Handle ioctl commands to obtain information from the cache 320 */ 321 int 322 mrt6_ioctl(cmd, data) 323 int cmd; 324 caddr_t data; 325 { 326 327 switch (cmd) { 328 case SIOCGETSGCNT_IN6: 329 return (get_sg_cnt((struct sioc_sg_req6 *)data)); 330 case SIOCGETMIFCNT_IN6: 331 return (get_mif6_cnt((struct sioc_mif_req6 *)data)); 332 default: 333 return (EINVAL); 334 } 335 } 336 337 /* 338 * returns the packet, byte, rpf-failure count for the source group provided 339 */ 340 static int 341 get_sg_cnt(req) 342 struct sioc_sg_req6 *req; 343 { 344 struct mf6c *rt; 345 int s; 346 347 s = splsoftnet(); 348 MF6CFIND(req->src.sin6_addr, req->grp.sin6_addr, rt); 349 splx(s); 350 if (rt != NULL) { 351 req->pktcnt = rt->mf6c_pkt_cnt; 352 req->bytecnt = rt->mf6c_byte_cnt; 353 req->wrong_if = rt->mf6c_wrong_if; 354 } else 355 return (ESRCH); 356 #if 0 357 req->pktcnt = req->bytecnt = req->wrong_if = 0xffffffff; 358 #endif 359 360 return 0; 361 } 362 363 /* 364 * returns the input and output packet and byte counts on the mif provided 365 */ 366 static int 367 get_mif6_cnt(req) 368 struct sioc_mif_req6 *req; 369 { 370 mifi_t mifi = req->mifi; 371 372 if (mifi >= nummifs) 373 return EINVAL; 374 375 req->icount = mif6table[mifi].m6_pkt_in; 376 req->ocount = mif6table[mifi].m6_pkt_out; 377 req->ibytes = mif6table[mifi].m6_bytes_in; 378 req->obytes = mif6table[mifi].m6_bytes_out; 379 380 return 0; 381 } 382 383 /* 384 * Get PIM processiong global 385 */ 386 static int 387 get_pim6(m) 388 struct mbuf *m; 389 { 390 int *i; 391 392 i = mtod(m, int *); 393 394 *i = pim6; 395 396 return 0; 397 } 398 399 static int 400 set_pim6(i) 401 int *i; 402 { 403 if ((*i != 1) && (*i != 0)) 404 return EINVAL; 405 406 pim6 = *i; 407 408 return 0; 409 } 410 411 /* 412 * Enable multicast routing 413 */ 414 static int 415 ip6_mrouter_init(so, m, cmd) 416 struct socket *so; 417 struct mbuf *m; 418 int cmd; 419 { 420 int *v; 421 422 #ifdef MRT6DEBUG 423 if (mrt6debug) 424 log(LOG_DEBUG, 425 "ip6_mrouter_init: so_type = %d, pr_protocol = %d\n", 426 so->so_type, so->so_proto->pr_protocol); 427 #endif 428 429 if (so->so_type != SOCK_RAW || 430 so->so_proto->pr_protocol != IPPROTO_ICMPV6) 431 return EOPNOTSUPP; 432 433 if (!m || (m->m_len != sizeof(int *))) 434 return ENOPROTOOPT; 435 436 v = mtod(m, int *); 437 if (*v != 1) 438 return ENOPROTOOPT; 439 440 if (ip6_mrouter != NULL) return EADDRINUSE; 441 442 ip6_mrouter = so; 443 ip6_mrouter_ver = cmd; 444 445 bzero((caddr_t)mf6ctable, sizeof(mf6ctable)); 446 bzero((caddr_t)n6expire, sizeof(n6expire)); 447 448 pim6 = 0;/* used for stubbing out/in pim stuff */ 449 450 callout_reset(&expire_upcalls_ch, EXPIRE_TIMEOUT, 451 expire_upcalls, NULL); 452 453 #ifdef MRT6DEBUG 454 if (mrt6debug) 455 log(LOG_DEBUG, "ip6_mrouter_init\n"); 456 #endif 457 458 return 0; 459 } 460 461 /* 462 * Disable multicast routing 463 */ 464 int 465 ip6_mrouter_done() 466 { 467 mifi_t mifi; 468 int i; 469 struct ifnet *ifp; 470 struct in6_ifreq ifr; 471 struct mf6c *rt; 472 struct rtdetq *rte; 473 int s; 474 475 s = splsoftnet(); 476 477 /* 478 * For each phyint in use, disable promiscuous reception of all IPv6 479 * multicasts. 480 */ 481 #ifdef INET 482 #ifdef MROUTING 483 /* 484 * If there is still IPv4 multicast routing daemon, 485 * we remain interfaces to receive all muliticasted packets. 486 * XXX: there may be an interface in which the IPv4 multicast 487 * daemon is not interested... 488 */ 489 if (!ip_mrouter) 490 #endif 491 #endif 492 { 493 for (mifi = 0; mifi < nummifs; mifi++) { 494 if (mif6table[mifi].m6_ifp && 495 !(mif6table[mifi].m6_flags & MIFF_REGISTER)) { 496 ifr.ifr_addr.sin6_family = AF_INET6; 497 ifr.ifr_addr.sin6_addr= in6addr_any; 498 ifp = mif6table[mifi].m6_ifp; 499 (*ifp->if_ioctl)(ifp, SIOCDELMULTI, 500 (caddr_t)&ifr); 501 } 502 } 503 } 504 #ifdef notyet 505 bzero((caddr_t)qtable, sizeof(qtable)); 506 bzero((caddr_t)tbftable, sizeof(tbftable)); 507 #endif 508 bzero((caddr_t)mif6table, sizeof(mif6table)); 509 nummifs = 0; 510 511 pim6 = 0; /* used to stub out/in pim specific code */ 512 513 callout_stop(&expire_upcalls_ch); 514 515 /* 516 * Free all multicast forwarding cache entries. 517 */ 518 for (i = 0; i < MF6CTBLSIZ; i++) { 519 rt = mf6ctable[i]; 520 while (rt) { 521 struct mf6c *frt; 522 523 for (rte = rt->mf6c_stall; rte != NULL; ) { 524 struct rtdetq *n = rte->next; 525 526 m_free(rte->m); 527 free(rte, M_MRTABLE); 528 rte = n; 529 } 530 frt = rt; 531 rt = rt->mf6c_next; 532 free(frt, M_MRTABLE); 533 } 534 } 535 536 bzero((caddr_t)mf6ctable, sizeof(mf6ctable)); 537 538 /* 539 * Reset de-encapsulation cache 540 */ 541 reg_mif_num = -1; 542 543 ip6_mrouter = NULL; 544 ip6_mrouter_ver = 0; 545 546 splx(s); 547 548 #ifdef MRT6DEBUG 549 if (mrt6debug) 550 log(LOG_DEBUG, "ip6_mrouter_done\n"); 551 #endif 552 553 return 0; 554 } 555 556 static struct sockaddr_in6 sin6 = { sizeof(sin6), AF_INET6 }; 557 558 /* 559 * Add a mif to the mif table 560 */ 561 static int 562 add_m6if(mifcp) 563 struct mif6ctl *mifcp; 564 { 565 struct mif6 *mifp; 566 struct ifnet *ifp; 567 struct in6_ifreq ifr; 568 int error, s; 569 #ifdef notyet 570 struct tbf *m_tbf = tbftable + mifcp->mif6c_mifi; 571 #endif 572 573 if (mifcp->mif6c_mifi >= MAXMIFS) 574 return EINVAL; 575 mifp = mif6table + mifcp->mif6c_mifi; 576 if (mifp->m6_ifp) 577 return EADDRINUSE; /* XXX: is it appropriate? */ 578 if (mifcp->mif6c_pifi == 0 || mifcp->mif6c_pifi > if_index) 579 return ENXIO; 580 /* 581 * XXX: some OSes can remove ifp and clear ifindex2ifnet[id] 582 * even for id between 0 and if_index. 583 */ 584 if ((ifp = ifindex2ifnet[mifcp->mif6c_pifi]) == NULL) 585 return ENXIO; 586 587 if (mifcp->mif6c_flags & MIFF_REGISTER) { 588 if (reg_mif_num == (mifi_t)-1) { 589 strlcpy(multicast_register_if.if_xname, 590 "register_mif", 591 sizeof(multicast_register_if.if_xname)); 592 multicast_register_if.if_flags |= IFF_LOOPBACK; 593 multicast_register_if.if_index = mifcp->mif6c_mifi; 594 reg_mif_num = mifcp->mif6c_mifi; 595 } 596 597 ifp = &multicast_register_if; 598 599 } /* if REGISTER */ 600 else { 601 /* Make sure the interface supports multicast */ 602 if ((ifp->if_flags & IFF_MULTICAST) == 0) 603 return EOPNOTSUPP; 604 605 s = splsoftnet(); 606 /* 607 * Enable promiscuous reception of all IPv6 multicasts 608 * from the interface. 609 */ 610 ifr.ifr_addr.sin6_family = AF_INET6; 611 ifr.ifr_addr.sin6_addr = in6addr_any; 612 error = (*ifp->if_ioctl)(ifp, SIOCADDMULTI, (caddr_t)&ifr); 613 splx(s); 614 if (error) 615 return error; 616 } 617 618 s = splsoftnet(); 619 mifp->m6_flags = mifcp->mif6c_flags; 620 mifp->m6_ifp = ifp; 621 #ifdef notyet 622 /* scaling up here allows division by 1024 in critical code */ 623 mifp->m6_rate_limit = mifcp->mif6c_rate_limit * 1024 / 1000; 624 #endif 625 /* initialize per mif pkt counters */ 626 mifp->m6_pkt_in = 0; 627 mifp->m6_pkt_out = 0; 628 mifp->m6_bytes_in = 0; 629 mifp->m6_bytes_out = 0; 630 splx(s); 631 632 /* Adjust nummifs up if the mifi is higher than nummifs */ 633 if (nummifs <= mifcp->mif6c_mifi) 634 nummifs = mifcp->mif6c_mifi + 1; 635 636 #ifdef MRT6DEBUG 637 if (mrt6debug) 638 log(LOG_DEBUG, 639 "add_mif #%d, phyint %s%d\n", 640 mifcp->mif6c_mifi, 641 ifp->if_name, ifp->if_unit); 642 #endif 643 644 return 0; 645 } 646 647 /* 648 * Delete a mif from the mif table 649 */ 650 static int 651 del_m6if(mifip) 652 mifi_t *mifip; 653 { 654 struct mif6 *mifp = mif6table + *mifip; 655 mifi_t mifi; 656 struct ifnet *ifp; 657 struct in6_ifreq ifr; 658 int s; 659 660 if (*mifip >= nummifs) 661 return EINVAL; 662 if (mifp->m6_ifp == NULL) 663 return EINVAL; 664 665 s = splsoftnet(); 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 ifr.ifr_addr.sin6_family = AF_INET6; 675 ifr.ifr_addr.sin6_addr = in6addr_any; 676 (*ifp->if_ioctl)(ifp, SIOCDELMULTI, (caddr_t)&ifr); 677 } 678 679 #ifdef notyet 680 bzero((caddr_t)qtable[*mifip], sizeof(qtable[*mifip])); 681 bzero((caddr_t)mifp->m6_tbf, sizeof(*(mifp->m6_tbf))); 682 #endif 683 bzero((caddr_t)mifp, sizeof (*mifp)); 684 685 /* Adjust nummifs down */ 686 for (mifi = nummifs; mifi > 0; mifi--) 687 if (mif6table[mifi - 1].m6_ifp) 688 break; 689 nummifs = mifi; 690 691 splx(s); 692 693 #ifdef MRT6DEBUG 694 if (mrt6debug) 695 log(LOG_DEBUG, "del_m6if %d, nummifs %d\n", *mifip, nummifs); 696 #endif 697 698 return 0; 699 } 700 701 /* 702 * Add an mfc entry 703 */ 704 static int 705 add_m6fc(mfccp) 706 struct mf6cctl *mfccp; 707 { 708 struct mf6c *rt; 709 u_long hash; 710 struct rtdetq *rte; 711 u_short nstl; 712 int s; 713 714 MF6CFIND(mfccp->mf6cc_origin.sin6_addr, 715 mfccp->mf6cc_mcastgrp.sin6_addr, rt); 716 717 /* If an entry already exists, just update the fields */ 718 if (rt) { 719 #ifdef MRT6DEBUG 720 if (mrt6debug & DEBUG_MFC) 721 log(LOG_DEBUG,"add_m6fc update o %s g %s p %x\n", 722 ip6_sprintf(&mfccp->mf6cc_origin.sin6_addr), 723 ip6_sprintf(&mfccp->mf6cc_mcastgrp.sin6_addr), 724 mfccp->mf6cc_parent); 725 #endif 726 727 s = splsoftnet(); 728 rt->mf6c_parent = mfccp->mf6cc_parent; 729 rt->mf6c_ifset = mfccp->mf6cc_ifset; 730 splx(s); 731 return 0; 732 } 733 734 /* 735 * Find the entry for which the upcall was made and update 736 */ 737 s = splsoftnet(); 738 hash = MF6CHASH(mfccp->mf6cc_origin.sin6_addr, 739 mfccp->mf6cc_mcastgrp.sin6_addr); 740 for (rt = mf6ctable[hash], nstl = 0; rt; rt = rt->mf6c_next) { 741 if (IN6_ARE_ADDR_EQUAL(&rt->mf6c_origin.sin6_addr, 742 &mfccp->mf6cc_origin.sin6_addr) && 743 IN6_ARE_ADDR_EQUAL(&rt->mf6c_mcastgrp.sin6_addr, 744 &mfccp->mf6cc_mcastgrp.sin6_addr) && 745 (rt->mf6c_stall != NULL)) { 746 747 if (nstl++) 748 log(LOG_ERR, 749 "add_m6fc: %s o %s g %s p %x dbx %p\n", 750 "multiple kernel entries", 751 ip6_sprintf(&mfccp->mf6cc_origin.sin6_addr), 752 ip6_sprintf(&mfccp->mf6cc_mcastgrp.sin6_addr), 753 mfccp->mf6cc_parent, rt->mf6c_stall); 754 755 #ifdef MRT6DEBUG 756 if (mrt6debug & DEBUG_MFC) 757 log(LOG_DEBUG, 758 "add_m6fc o %s g %s p %x dbg %x\n", 759 ip6_sprintf(&mfccp->mf6cc_origin.sin6_addr), 760 ip6_sprintf(&mfccp->mf6cc_mcastgrp.sin6_addr), 761 mfccp->mf6cc_parent, rt->mf6c_stall); 762 #endif 763 764 rt->mf6c_origin = mfccp->mf6cc_origin; 765 rt->mf6c_mcastgrp = mfccp->mf6cc_mcastgrp; 766 rt->mf6c_parent = mfccp->mf6cc_parent; 767 rt->mf6c_ifset = mfccp->mf6cc_ifset; 768 /* initialize pkt counters per src-grp */ 769 rt->mf6c_pkt_cnt = 0; 770 rt->mf6c_byte_cnt = 0; 771 rt->mf6c_wrong_if = 0; 772 773 rt->mf6c_expire = 0; /* Don't clean this guy up */ 774 n6expire[hash]--; 775 776 /* free packets Qed at the end of this entry */ 777 for (rte = rt->mf6c_stall; rte != NULL; ) { 778 struct rtdetq *n = rte->next; 779 ip6_mdq(rte->m, rte->ifp, rt); 780 m_freem(rte->m); 781 #ifdef UPCALL_TIMING 782 collate(&(rte->t)); 783 #endif /* UPCALL_TIMING */ 784 free(rte, M_MRTABLE); 785 rte = n; 786 } 787 rt->mf6c_stall = NULL; 788 } 789 } 790 791 /* 792 * It is possible that an entry is being inserted without an upcall 793 */ 794 if (nstl == 0) { 795 #ifdef MRT6DEBUG 796 if (mrt6debug & DEBUG_MFC) 797 log(LOG_DEBUG, 798 "add_mfc no upcall h %d o %s g %s p %x\n", 799 hash, 800 ip6_sprintf(&mfccp->mf6cc_origin.sin6_addr), 801 ip6_sprintf(&mfccp->mf6cc_mcastgrp.sin6_addr), 802 mfccp->mf6cc_parent); 803 #endif 804 805 for (rt = mf6ctable[hash]; rt; rt = rt->mf6c_next) { 806 807 if (IN6_ARE_ADDR_EQUAL(&rt->mf6c_origin.sin6_addr, 808 &mfccp->mf6cc_origin.sin6_addr)&& 809 IN6_ARE_ADDR_EQUAL(&rt->mf6c_mcastgrp.sin6_addr, 810 &mfccp->mf6cc_mcastgrp.sin6_addr)) { 811 812 rt->mf6c_origin = mfccp->mf6cc_origin; 813 rt->mf6c_mcastgrp = mfccp->mf6cc_mcastgrp; 814 rt->mf6c_parent = mfccp->mf6cc_parent; 815 rt->mf6c_ifset = mfccp->mf6cc_ifset; 816 /* initialize pkt counters per src-grp */ 817 rt->mf6c_pkt_cnt = 0; 818 rt->mf6c_byte_cnt = 0; 819 rt->mf6c_wrong_if = 0; 820 821 if (rt->mf6c_expire) 822 n6expire[hash]--; 823 rt->mf6c_expire = 0; 824 } 825 } 826 if (rt == NULL) { 827 /* no upcall, so make a new entry */ 828 rt = (struct mf6c *)malloc(sizeof(*rt), M_MRTABLE, 829 M_NOWAIT); 830 if (rt == NULL) { 831 splx(s); 832 return ENOBUFS; 833 } 834 835 /* insert new entry at head of hash chain */ 836 rt->mf6c_origin = mfccp->mf6cc_origin; 837 rt->mf6c_mcastgrp = mfccp->mf6cc_mcastgrp; 838 rt->mf6c_parent = mfccp->mf6cc_parent; 839 rt->mf6c_ifset = mfccp->mf6cc_ifset; 840 /* initialize pkt counters per src-grp */ 841 rt->mf6c_pkt_cnt = 0; 842 rt->mf6c_byte_cnt = 0; 843 rt->mf6c_wrong_if = 0; 844 rt->mf6c_expire = 0; 845 rt->mf6c_stall = NULL; 846 847 /* link into table */ 848 rt->mf6c_next = mf6ctable[hash]; 849 mf6ctable[hash] = rt; 850 } 851 } 852 splx(s); 853 return 0; 854 } 855 856 #ifdef UPCALL_TIMING 857 /* 858 * collect delay statistics on the upcalls 859 */ 860 static void 861 collate(t) 862 struct timeval *t; 863 { 864 u_long d; 865 struct timeval tp; 866 u_long delta; 867 868 GET_TIME(tp); 869 870 if (TV_LT(*t, tp)) 871 { 872 TV_DELTA(tp, *t, delta); 873 874 d = delta >> 10; 875 if (d > UPCALL_MAX) 876 d = UPCALL_MAX; 877 878 ++upcall_data[d]; 879 } 880 } 881 #endif /* UPCALL_TIMING */ 882 883 /* 884 * Delete an mfc entry 885 */ 886 static int 887 del_m6fc(mfccp) 888 struct mf6cctl *mfccp; 889 { 890 struct sockaddr_in6 origin; 891 struct sockaddr_in6 mcastgrp; 892 struct mf6c *rt; 893 struct mf6c **nptr; 894 u_long hash; 895 int s; 896 897 origin = mfccp->mf6cc_origin; 898 mcastgrp = mfccp->mf6cc_mcastgrp; 899 hash = MF6CHASH(origin.sin6_addr, mcastgrp.sin6_addr); 900 901 #ifdef MRT6DEBUG 902 if (mrt6debug & DEBUG_MFC) 903 log(LOG_DEBUG,"del_m6fc orig %s mcastgrp %s\n", 904 ip6_sprintf(&origin.sin6_addr), 905 ip6_sprintf(&mcastgrp.sin6_addr)); 906 #endif 907 908 s = splsoftnet(); 909 910 nptr = &mf6ctable[hash]; 911 while ((rt = *nptr) != NULL) { 912 if (IN6_ARE_ADDR_EQUAL(&origin.sin6_addr, 913 &rt->mf6c_origin.sin6_addr) && 914 IN6_ARE_ADDR_EQUAL(&mcastgrp.sin6_addr, 915 &rt->mf6c_mcastgrp.sin6_addr) && 916 rt->mf6c_stall == NULL) 917 break; 918 919 nptr = &rt->mf6c_next; 920 } 921 if (rt == NULL) { 922 splx(s); 923 return EADDRNOTAVAIL; 924 } 925 926 *nptr = rt->mf6c_next; 927 free(rt, M_MRTABLE); 928 929 splx(s); 930 931 return 0; 932 } 933 934 static int 935 socket_send(s, mm, src) 936 struct socket *s; 937 struct mbuf *mm; 938 struct sockaddr_in6 *src; 939 { 940 if (s) { 941 if (sbappendaddr(&s->so_rcv, 942 (struct sockaddr *)src, 943 mm, (struct mbuf *)0) != 0) { 944 sorwakeup(s); 945 return 0; 946 } 947 } 948 m_freem(mm); 949 return -1; 950 } 951 952 /* 953 * IPv6 multicast forwarding function. This function assumes that the packet 954 * pointed to by "ip6" has arrived on (or is about to be sent to) the interface 955 * pointed to by "ifp", and the packet is to be relayed to other networks 956 * that have members of the packet's destination IPv6 multicast group. 957 * 958 * The packet is returned unscathed to the caller, unless it is 959 * erroneous, in which case a non-zero return value tells the caller to 960 * discard it. 961 */ 962 963 int 964 ip6_mforward(ip6, ifp, m) 965 struct ip6_hdr *ip6; 966 struct ifnet *ifp; 967 struct mbuf *m; 968 { 969 struct mf6c *rt; 970 struct mif6 *mifp; 971 struct mbuf *mm; 972 int s; 973 mifi_t mifi; 974 long time_second = time.tv_sec; 975 976 #ifdef MRT6DEBUG 977 if (mrt6debug & DEBUG_FORWARD) 978 log(LOG_DEBUG, "ip6_mforward: src %s, dst %s, ifindex %d\n", 979 ip6_sprintf(&ip6->ip6_src), ip6_sprintf(&ip6->ip6_dst), 980 ifp->if_index); 981 #endif 982 983 /* 984 * Don't forward a packet with Hop limit of zero or one, 985 * or a packet destined to a local-only group. 986 */ 987 if (ip6->ip6_hlim <= 1 || IN6_IS_ADDR_MC_NODELOCAL(&ip6->ip6_dst) || 988 IN6_IS_ADDR_MC_LINKLOCAL(&ip6->ip6_dst)) 989 return 0; 990 ip6->ip6_hlim--; 991 992 /* 993 * Source address check: do not forward packets with unspecified 994 * source. It was discussed in July 2000, on ipngwg mailing list. 995 * This is rather more serious than unicast cases, because some 996 * MLD packets can be sent with the unspecified source address 997 * (although such packets must normally set 1 to the hop limit field). 998 */ 999 if (IN6_IS_ADDR_UNSPECIFIED(&ip6->ip6_src)) { 1000 ip6stat.ip6s_cantforward++; 1001 if (ip6_log_time + ip6_log_interval < time_second) { 1002 ip6_log_time = time_second; 1003 log(LOG_DEBUG, 1004 "cannot forward " 1005 "from %s to %s nxt %d received on %s\n", 1006 ip6_sprintf(&ip6->ip6_src), 1007 ip6_sprintf(&ip6->ip6_dst), 1008 ip6->ip6_nxt, 1009 m->m_pkthdr.rcvif ? 1010 if_name(m->m_pkthdr.rcvif) : "?"); 1011 } 1012 return 0; 1013 } 1014 1015 /* 1016 * Determine forwarding mifs from the forwarding cache table 1017 */ 1018 s = splsoftnet(); 1019 MF6CFIND(ip6->ip6_src, ip6->ip6_dst, rt); 1020 1021 /* Entry exists, so forward if necessary */ 1022 if (rt) { 1023 splx(s); 1024 return (ip6_mdq(m, ifp, rt)); 1025 } else { 1026 /* 1027 * If we don't have a route for packet's origin, 1028 * Make a copy of the packet & 1029 * send message to routing daemon 1030 */ 1031 1032 struct mbuf *mb0; 1033 struct rtdetq *rte; 1034 u_long hash; 1035 /* int i, npkts;*/ 1036 #ifdef UPCALL_TIMING 1037 struct timeval tp; 1038 1039 GET_TIME(tp); 1040 #endif /* UPCALL_TIMING */ 1041 1042 mrt6stat.mrt6s_no_route++; 1043 #ifdef MRT6DEBUG 1044 if (mrt6debug & (DEBUG_FORWARD | DEBUG_MFC)) 1045 log(LOG_DEBUG, "ip6_mforward: no rte s %s g %s\n", 1046 ip6_sprintf(&ip6->ip6_src), 1047 ip6_sprintf(&ip6->ip6_dst)); 1048 #endif 1049 1050 /* 1051 * Allocate mbufs early so that we don't do extra work if we 1052 * are just going to fail anyway. 1053 */ 1054 rte = (struct rtdetq *)malloc(sizeof(*rte), M_MRTABLE, 1055 M_NOWAIT); 1056 if (rte == NULL) { 1057 splx(s); 1058 return ENOBUFS; 1059 } 1060 mb0 = m_copy(m, 0, M_COPYALL); 1061 /* 1062 * Pullup packet header if needed before storing it, 1063 * as other references may modify it in the meantime. 1064 */ 1065 if (mb0 && 1066 (M_READONLY(mb0) || mb0->m_len < sizeof(struct ip6_hdr))) 1067 mb0 = m_pullup(mb0, sizeof(struct ip6_hdr)); 1068 if (mb0 == NULL) { 1069 free(rte, M_MRTABLE); 1070 splx(s); 1071 return ENOBUFS; 1072 } 1073 1074 /* is there an upcall waiting for this packet? */ 1075 hash = MF6CHASH(ip6->ip6_src, ip6->ip6_dst); 1076 for (rt = mf6ctable[hash]; rt; rt = rt->mf6c_next) { 1077 if (IN6_ARE_ADDR_EQUAL(&ip6->ip6_src, 1078 &rt->mf6c_origin.sin6_addr) && 1079 IN6_ARE_ADDR_EQUAL(&ip6->ip6_dst, 1080 &rt->mf6c_mcastgrp.sin6_addr) && 1081 (rt->mf6c_stall != NULL)) 1082 break; 1083 } 1084 1085 if (rt == NULL) { 1086 struct mrt6msg *im; 1087 struct omrt6msg *oim; 1088 1089 /* no upcall, so make a new entry */ 1090 rt = (struct mf6c *)malloc(sizeof(*rt), M_MRTABLE, 1091 M_NOWAIT); 1092 if (rt == NULL) { 1093 free(rte, M_MRTABLE); 1094 m_freem(mb0); 1095 splx(s); 1096 return ENOBUFS; 1097 } 1098 /* 1099 * Make a copy of the header to send to the user 1100 * level process 1101 */ 1102 mm = m_copy(mb0, 0, sizeof(struct ip6_hdr)); 1103 1104 if (mm == NULL) { 1105 free(rte, M_MRTABLE); 1106 m_freem(mb0); 1107 free(rt, M_MRTABLE); 1108 splx(s); 1109 return ENOBUFS; 1110 } 1111 1112 /* 1113 * Send message to routing daemon 1114 */ 1115 sin6.sin6_addr = ip6->ip6_src; 1116 1117 im = NULL; 1118 oim = NULL; 1119 switch (ip6_mrouter_ver) { 1120 case MRT6_OINIT: 1121 oim = mtod(mm, struct omrt6msg *); 1122 oim->im6_msgtype = MRT6MSG_NOCACHE; 1123 oim->im6_mbz = 0; 1124 break; 1125 case MRT6_INIT: 1126 im = mtod(mm, struct mrt6msg *); 1127 im->im6_msgtype = MRT6MSG_NOCACHE; 1128 im->im6_mbz = 0; 1129 break; 1130 default: 1131 free(rte, M_MRTABLE); 1132 m_freem(mb0); 1133 free(rt, M_MRTABLE); 1134 splx(s); 1135 return EINVAL; 1136 } 1137 1138 #ifdef MRT6DEBUG 1139 if (mrt6debug & DEBUG_FORWARD) 1140 log(LOG_DEBUG, 1141 "getting the iif info in the kernel\n"); 1142 #endif 1143 1144 for (mifp = mif6table, mifi = 0; 1145 mifi < nummifs && mifp->m6_ifp != ifp; 1146 mifp++, mifi++) 1147 ; 1148 1149 switch (ip6_mrouter_ver) { 1150 case MRT6_OINIT: 1151 oim->im6_mif = mifi; 1152 break; 1153 case MRT6_INIT: 1154 im->im6_mif = mifi; 1155 break; 1156 } 1157 1158 if (socket_send(ip6_mrouter, mm, &sin6) < 0) { 1159 log(LOG_WARNING, "ip6_mforward: ip6_mrouter " 1160 "socket queue full\n"); 1161 mrt6stat.mrt6s_upq_sockfull++; 1162 free(rte, M_MRTABLE); 1163 m_freem(mb0); 1164 free(rt, M_MRTABLE); 1165 splx(s); 1166 return ENOBUFS; 1167 } 1168 1169 mrt6stat.mrt6s_upcalls++; 1170 1171 /* insert new entry at head of hash chain */ 1172 bzero(rt, sizeof(*rt)); 1173 rt->mf6c_origin.sin6_family = AF_INET6; 1174 rt->mf6c_origin.sin6_len = sizeof(struct sockaddr_in6); 1175 rt->mf6c_origin.sin6_addr = ip6->ip6_src; 1176 rt->mf6c_mcastgrp.sin6_family = AF_INET6; 1177 rt->mf6c_mcastgrp.sin6_len = sizeof(struct sockaddr_in6); 1178 rt->mf6c_mcastgrp.sin6_addr = ip6->ip6_dst; 1179 rt->mf6c_expire = UPCALL_EXPIRE; 1180 n6expire[hash]++; 1181 rt->mf6c_parent = MF6C_INCOMPLETE_PARENT; 1182 1183 /* link into table */ 1184 rt->mf6c_next = mf6ctable[hash]; 1185 mf6ctable[hash] = rt; 1186 /* Add this entry to the end of the queue */ 1187 rt->mf6c_stall = rte; 1188 } else { 1189 /* determine if q has overflowed */ 1190 struct rtdetq **p; 1191 int npkts = 0; 1192 1193 for (p = &rt->mf6c_stall; *p != NULL; p = &(*p)->next) 1194 if (++npkts > MAX_UPQ6) { 1195 mrt6stat.mrt6s_upq_ovflw++; 1196 free(rte, M_MRTABLE); 1197 m_freem(mb0); 1198 splx(s); 1199 return 0; 1200 } 1201 1202 /* Add this entry to the end of the queue */ 1203 *p = rte; 1204 } 1205 1206 rte->next = NULL; 1207 rte->m = mb0; 1208 rte->ifp = ifp; 1209 #ifdef UPCALL_TIMING 1210 rte->t = tp; 1211 #endif /* UPCALL_TIMING */ 1212 1213 splx(s); 1214 1215 return 0; 1216 } 1217 } 1218 1219 /* 1220 * Clean up cache entries if upcalls are not serviced 1221 * Call from the Slow Timeout mechanism, every half second. 1222 */ 1223 static void 1224 expire_upcalls(unused) 1225 void *unused; 1226 { 1227 struct rtdetq *rte; 1228 struct mf6c *mfc, **nptr; 1229 int i; 1230 int s; 1231 1232 s = splsoftnet(); 1233 for (i = 0; i < MF6CTBLSIZ; i++) { 1234 if (n6expire[i] == 0) 1235 continue; 1236 nptr = &mf6ctable[i]; 1237 while ((mfc = *nptr) != NULL) { 1238 rte = mfc->mf6c_stall; 1239 /* 1240 * Skip real cache entries 1241 * Make sure it wasn't marked to not expire (shouldn't happen) 1242 * If it expires now 1243 */ 1244 if (rte != NULL && 1245 mfc->mf6c_expire != 0 && 1246 --mfc->mf6c_expire == 0) { 1247 #ifdef MRT6DEBUG 1248 if (mrt6debug & DEBUG_EXPIRE) 1249 log(LOG_DEBUG, "expire_upcalls: expiring (%s %s)\n", 1250 ip6_sprintf(&mfc->mf6c_origin.sin6_addr), 1251 ip6_sprintf(&mfc->mf6c_mcastgrp.sin6_addr)); 1252 #endif 1253 /* 1254 * drop all the packets 1255 * free the mbuf with the pkt, if, timing info 1256 */ 1257 do { 1258 struct rtdetq *n = rte->next; 1259 m_freem(rte->m); 1260 free(rte, M_MRTABLE); 1261 rte = n; 1262 } while (rte != NULL); 1263 mrt6stat.mrt6s_cache_cleanups++; 1264 n6expire[i]--; 1265 1266 *nptr = mfc->mf6c_next; 1267 free(mfc, M_MRTABLE); 1268 } else { 1269 nptr = &mfc->mf6c_next; 1270 } 1271 } 1272 } 1273 splx(s); 1274 callout_reset(&expire_upcalls_ch, EXPIRE_TIMEOUT, 1275 expire_upcalls, NULL); 1276 } 1277 1278 /* 1279 * Packet forwarding routine once entry in the cache is made 1280 */ 1281 static int 1282 ip6_mdq(m, ifp, rt) 1283 struct mbuf *m; 1284 struct ifnet *ifp; 1285 struct mf6c *rt; 1286 { 1287 struct ip6_hdr *ip6 = mtod(m, struct ip6_hdr *); 1288 mifi_t mifi, iif; 1289 struct mif6 *mifp; 1290 int plen = m->m_pkthdr.len; 1291 1292 /* 1293 * Macro to send packet on mif. Since RSVP packets don't get counted on 1294 * input, they shouldn't get counted on output, so statistics keeping is 1295 * separate. 1296 */ 1297 1298 #define MC6_SEND(ip6, mifp, m) do { \ 1299 if ((mifp)->m6_flags & MIFF_REGISTER) \ 1300 register_send((ip6), (mifp), (m)); \ 1301 else \ 1302 phyint_send((ip6), (mifp), (m)); \ 1303 } while (/*CONSTCOND*/ 0) 1304 1305 /* 1306 * Don't forward if it didn't arrive from the parent mif 1307 * for its origin. 1308 */ 1309 mifi = rt->mf6c_parent; 1310 if ((mifi >= nummifs) || (mif6table[mifi].m6_ifp != ifp)) { 1311 /* came in the wrong interface */ 1312 #ifdef MRT6DEBUG 1313 if (mrt6debug & DEBUG_FORWARD) 1314 log(LOG_DEBUG, 1315 "wrong if: ifid %d mifi %d mififid %x\n", 1316 ifp->if_index, mifi, 1317 mif6table[mifi].m6_ifp->if_index); 1318 #endif 1319 mrt6stat.mrt6s_wrong_if++; 1320 rt->mf6c_wrong_if++; 1321 /* 1322 * If we are doing PIM processing, and we are forwarding 1323 * packets on this interface, send a message to the 1324 * routing daemon. 1325 */ 1326 /* have to make sure this is a valid mif */ 1327 if (mifi < nummifs && mif6table[mifi].m6_ifp) 1328 if (pim6 && (m->m_flags & M_LOOP) == 0) { 1329 /* 1330 * Check the M_LOOP flag to avoid an 1331 * unnecessary PIM assert. 1332 * XXX: M_LOOP is an ad-hoc hack... 1333 */ 1334 static struct sockaddr_in6 sin6 = 1335 { sizeof(sin6), AF_INET6 }; 1336 1337 struct mbuf *mm; 1338 struct mrt6msg *im; 1339 struct omrt6msg *oim; 1340 1341 mm = m_copy(m, 0, sizeof(struct ip6_hdr)); 1342 if (mm && 1343 (M_READONLY(mm) || 1344 mm->m_len < sizeof(struct ip6_hdr))) 1345 mm = m_pullup(mm, sizeof(struct ip6_hdr)); 1346 if (mm == NULL) 1347 return ENOBUFS; 1348 1349 oim = NULL; 1350 im = NULL; 1351 switch (ip6_mrouter_ver) { 1352 case MRT6_OINIT: 1353 oim = mtod(mm, struct omrt6msg *); 1354 oim->im6_msgtype = MRT6MSG_WRONGMIF; 1355 oim->im6_mbz = 0; 1356 break; 1357 case MRT6_INIT: 1358 im = mtod(mm, struct mrt6msg *); 1359 im->im6_msgtype = MRT6MSG_WRONGMIF; 1360 im->im6_mbz = 0; 1361 break; 1362 default: 1363 m_freem(mm); 1364 return EINVAL; 1365 } 1366 1367 for (mifp = mif6table, iif = 0; 1368 iif < nummifs && mifp && 1369 mifp->m6_ifp != ifp; 1370 mifp++, iif++) 1371 ; 1372 1373 switch (ip6_mrouter_ver) { 1374 case MRT6_OINIT: 1375 oim->im6_mif = iif; 1376 sin6.sin6_addr = oim->im6_src; 1377 break; 1378 case MRT6_INIT: 1379 im->im6_mif = iif; 1380 sin6.sin6_addr = im->im6_src; 1381 break; 1382 } 1383 1384 mrt6stat.mrt6s_upcalls++; 1385 1386 if (socket_send(ip6_mrouter, mm, &sin6) < 0) { 1387 #ifdef MRT6DEBUG 1388 if (mrt6debug) 1389 log(LOG_WARNING, "mdq, ip6_mrouter socket queue full\n"); 1390 #endif 1391 ++mrt6stat.mrt6s_upq_sockfull; 1392 return ENOBUFS; 1393 } /* if socket Q full */ 1394 } /* if PIM */ 1395 return 0; 1396 } /* if wrong iif */ 1397 1398 /* If I sourced this packet, it counts as output, else it was input. */ 1399 if (m->m_pkthdr.rcvif == NULL) { 1400 /* XXX: is rcvif really NULL when output?? */ 1401 mif6table[mifi].m6_pkt_out++; 1402 mif6table[mifi].m6_bytes_out += plen; 1403 } else { 1404 mif6table[mifi].m6_pkt_in++; 1405 mif6table[mifi].m6_bytes_in += plen; 1406 } 1407 rt->mf6c_pkt_cnt++; 1408 rt->mf6c_byte_cnt += plen; 1409 1410 /* 1411 * For each mif, forward a copy of the packet if there are group 1412 * members downstream on the interface. 1413 */ 1414 for (mifp = mif6table, mifi = 0; mifi < nummifs; mifp++, mifi++) 1415 if (IF_ISSET(mifi, &rt->mf6c_ifset)) { 1416 #ifdef notyet 1417 /* 1418 * check if the outgoing packet is going to break 1419 * a scope boundary. 1420 * XXX For packets through PIM register tunnel 1421 * interface, we believe a routing daemon. 1422 */ 1423 if ((mif6table[rt->mf6c_parent].m6_flags & 1424 MIFF_REGISTER) == 0 && 1425 (mif6table[mifi].m6_flags & MIFF_REGISTER) == 0 && 1426 (in6_addr2scopeid(ifp, &ip6->ip6_dst) != 1427 in6_addr2scopeid(mif6table[mifi].m6_ifp, 1428 &ip6->ip6_dst) || 1429 in6_addr2scopeid(ifp, &ip6->ip6_src) != 1430 in6_addr2scopeid(mif6table[mifi].m6_ifp, 1431 &ip6->ip6_src))) { 1432 ip6stat.ip6s_badscope++; 1433 continue; 1434 } 1435 #endif 1436 1437 mifp->m6_pkt_out++; 1438 mifp->m6_bytes_out += plen; 1439 MC6_SEND(ip6, mifp, m); 1440 } 1441 return 0; 1442 } 1443 1444 static void 1445 phyint_send(ip6, mifp, m) 1446 struct ip6_hdr *ip6; 1447 struct mif6 *mifp; 1448 struct mbuf *m; 1449 { 1450 struct mbuf *mb_copy; 1451 struct ifnet *ifp = mifp->m6_ifp; 1452 int error = 0; 1453 int s = splsoftnet(); 1454 static struct route_in6 ro; 1455 struct in6_multi *in6m; 1456 struct sockaddr_in6 *dst6; 1457 u_long linkmtu; 1458 1459 /* 1460 * Make a new reference to the packet; make sure that 1461 * the IPv6 header is actually copied, not just referenced, 1462 * so that ip6_output() only scribbles on the copy. 1463 */ 1464 mb_copy = m_copy(m, 0, M_COPYALL); 1465 if (mb_copy && 1466 (M_READONLY(mb_copy) || mb_copy->m_len < sizeof(struct ip6_hdr))) 1467 mb_copy = m_pullup(mb_copy, sizeof(struct ip6_hdr)); 1468 if (mb_copy == NULL) { 1469 splx(s); 1470 return; 1471 } 1472 /* set MCAST flag to the outgoing packet */ 1473 mb_copy->m_flags |= M_MCAST; 1474 1475 /* 1476 * If we sourced the packet, call ip6_output since we may devide 1477 * the packet into fragments when the packet is too big for the 1478 * outgoing interface. 1479 * Otherwise, we can simply send the packet to the interface 1480 * sending queue. 1481 */ 1482 if (m->m_pkthdr.rcvif == NULL) { 1483 struct ip6_moptions im6o; 1484 1485 im6o.im6o_multicast_ifp = ifp; 1486 /* XXX: ip6_output will override ip6->ip6_hlim */ 1487 im6o.im6o_multicast_hlim = ip6->ip6_hlim; 1488 im6o.im6o_multicast_loop = 1; 1489 error = ip6_output(mb_copy, NULL, &ro, 1490 IPV6_FORWARDING, &im6o, NULL); 1491 1492 #ifdef MRT6DEBUG 1493 if (mrt6debug & DEBUG_XMIT) 1494 log(LOG_DEBUG, "phyint_send on mif %d err %d\n", 1495 mifp - mif6table, error); 1496 #endif 1497 splx(s); 1498 return; 1499 } 1500 1501 /* 1502 * If we belong to the destination multicast group 1503 * on the outgoing interface, loop back a copy. 1504 */ 1505 dst6 = (struct sockaddr_in6 *)&ro.ro_dst; 1506 IN6_LOOKUP_MULTI(ip6->ip6_dst, ifp, in6m); 1507 if (in6m != NULL) { 1508 dst6->sin6_len = sizeof(struct sockaddr_in6); 1509 dst6->sin6_family = AF_INET6; 1510 dst6->sin6_addr = ip6->ip6_dst; 1511 ip6_mloopback(ifp, m, (struct sockaddr_in6 *)&ro.ro_dst); 1512 } 1513 /* 1514 * Put the packet into the sending queue of the outgoing interface 1515 * if it would fit in the MTU of the interface. 1516 */ 1517 linkmtu = IN6_LINKMTU(ifp); 1518 if (mb_copy->m_pkthdr.len <= linkmtu || linkmtu < IPV6_MMTU) { 1519 dst6->sin6_len = sizeof(struct sockaddr_in6); 1520 dst6->sin6_family = AF_INET6; 1521 dst6->sin6_addr = ip6->ip6_dst; 1522 /* 1523 * We just call if_output instead of nd6_output here, since 1524 * we need no ND for a multicast forwarded packet...right? 1525 */ 1526 error = (*ifp->if_output)(ifp, mb_copy, 1527 (struct sockaddr *)&ro.ro_dst, NULL); 1528 #ifdef MRT6DEBUG 1529 if (mrt6debug & DEBUG_XMIT) 1530 log(LOG_DEBUG, "phyint_send on mif %d err %d\n", 1531 mifp - mif6table, error); 1532 #endif 1533 } else { 1534 #ifdef MULTICAST_PMTUD 1535 icmp6_error(mb_copy, ICMP6_PACKET_TOO_BIG, 0, linkmtu); 1536 #else 1537 #ifdef MRT6DEBUG 1538 if (mrt6debug & DEBUG_XMIT) 1539 log(LOG_DEBUG, 1540 "phyint_send: packet too big on %s o %s g %s" 1541 " size %d(discarded)\n", 1542 if_name(ifp), 1543 ip6_sprintf(&ip6->ip6_src), 1544 ip6_sprintf(&ip6->ip6_dst), 1545 mb_copy->m_pkthdr.len); 1546 #endif /* MRT6DEBUG */ 1547 m_freem(mb_copy); /* simply discard the packet */ 1548 #endif 1549 } 1550 1551 splx(s); 1552 } 1553 1554 static int 1555 register_send(ip6, mif, m) 1556 struct ip6_hdr *ip6; 1557 struct mif6 *mif; 1558 struct mbuf *m; 1559 { 1560 struct mbuf *mm; 1561 int i, len = m->m_pkthdr.len; 1562 static struct sockaddr_in6 sin6 = { sizeof(sin6), AF_INET6 }; 1563 struct mrt6msg *im6; 1564 1565 #ifdef MRT6DEBUG 1566 if (mrt6debug) 1567 log(LOG_DEBUG, "** IPv6 register_send **\n src %s dst %s\n", 1568 ip6_sprintf(&ip6->ip6_src), ip6_sprintf(&ip6->ip6_dst)); 1569 #endif 1570 ++pim6stat.pim6s_snd_registers; 1571 1572 /* Make a copy of the packet to send to the user level process */ 1573 MGETHDR(mm, M_DONTWAIT, MT_HEADER); 1574 if (mm == NULL) 1575 return ENOBUFS; 1576 mm->m_data += max_linkhdr; 1577 mm->m_len = sizeof(struct ip6_hdr); 1578 1579 if ((mm->m_next = m_copy(m, 0, M_COPYALL)) == NULL) { 1580 m_freem(mm); 1581 return ENOBUFS; 1582 } 1583 i = MHLEN - M_LEADINGSPACE(mm); 1584 if (i > len) 1585 i = len; 1586 mm = m_pullup(mm, i); 1587 if (mm == NULL) 1588 return ENOBUFS; 1589 /* TODO: check it! */ 1590 mm->m_pkthdr.len = len + sizeof(struct ip6_hdr); 1591 1592 /* 1593 * Send message to routing daemon 1594 */ 1595 sin6.sin6_addr = ip6->ip6_src; 1596 1597 im6 = mtod(mm, struct mrt6msg *); 1598 im6->im6_msgtype = MRT6MSG_WHOLEPKT; 1599 im6->im6_mbz = 0; 1600 1601 im6->im6_mif = mif - mif6table; 1602 1603 /* iif info is not given for reg. encap.n */ 1604 mrt6stat.mrt6s_upcalls++; 1605 1606 if (socket_send(ip6_mrouter, mm, &sin6) < 0) { 1607 #ifdef MRT6DEBUG 1608 if (mrt6debug) 1609 log(LOG_WARNING, 1610 "register_send: ip6_mrouter socket queue full\n"); 1611 #endif 1612 ++mrt6stat.mrt6s_upq_sockfull; 1613 return ENOBUFS; 1614 } 1615 return 0; 1616 } 1617 1618 /* 1619 * PIM sparse mode hook 1620 * Receives the pim control messages, and passes them up to the listening 1621 * socket, using rip6_input. 1622 * The only message processed is the REGISTER pim message; the pim header 1623 * is stripped off, and the inner packet is passed to register_mforward. 1624 */ 1625 int 1626 pim6_input(mp, offp, proto) 1627 struct mbuf **mp; 1628 int *offp, proto; 1629 { 1630 struct pim *pim; /* pointer to a pim struct */ 1631 struct ip6_hdr *ip6; 1632 int pimlen; 1633 struct mbuf *m = *mp; 1634 int minlen; 1635 int off = *offp; 1636 1637 ++pim6stat.pim6s_rcv_total; 1638 1639 ip6 = mtod(m, struct ip6_hdr *); 1640 pimlen = m->m_pkthdr.len - *offp; 1641 1642 /* 1643 * Validate lengths 1644 */ 1645 if (pimlen < PIM_MINLEN) { 1646 ++pim6stat.pim6s_rcv_tooshort; 1647 #ifdef MRT6DEBUG 1648 if (mrt6debug & DEBUG_PIM) 1649 log(LOG_DEBUG,"pim6_input: PIM packet too short\n"); 1650 #endif 1651 m_freem(m); 1652 return (IPPROTO_DONE); 1653 } 1654 1655 /* 1656 * if the packet is at least as big as a REGISTER, go ahead 1657 * and grab the PIM REGISTER header size, to avoid another 1658 * possible m_pullup() later. 1659 * 1660 * PIM_MINLEN == pimhdr + u_int32 == 8 1661 * PIM6_REG_MINLEN == pimhdr + reghdr + eip6hdr == 4 + 4 + 40 1662 */ 1663 minlen = (pimlen >= PIM6_REG_MINLEN) ? PIM6_REG_MINLEN : PIM_MINLEN; 1664 1665 /* 1666 * Make sure that the IP6 and PIM headers in contiguous memory, and 1667 * possibly the PIM REGISTER header 1668 */ 1669 IP6_EXTHDR_GET(pim, struct pim *, m, off, minlen); 1670 if (pim == NULL) { 1671 pim6stat.pim6s_rcv_tooshort++; 1672 return IPPROTO_DONE; 1673 } 1674 1675 /* PIM version check */ 1676 if (pim->pim_ver != PIM_VERSION) { 1677 ++pim6stat.pim6s_rcv_badversion; 1678 #ifdef MRT6DEBUG 1679 log(LOG_ERR, 1680 "pim6_input: incorrect version %d, expecting %d\n", 1681 pim->pim_ver, PIM_VERSION); 1682 #endif 1683 m_freem(m); 1684 return (IPPROTO_DONE); 1685 } 1686 1687 #define PIM6_CHECKSUM 1688 #ifdef PIM6_CHECKSUM 1689 { 1690 int cksumlen; 1691 1692 /* 1693 * Validate checksum. 1694 * If PIM REGISTER, exclude the data packet 1695 */ 1696 if (pim->pim_type == PIM_REGISTER) 1697 cksumlen = PIM_MINLEN; 1698 else 1699 cksumlen = pimlen; 1700 1701 if (in6_cksum(m, IPPROTO_PIM, off, cksumlen)) { 1702 ++pim6stat.pim6s_rcv_badsum; 1703 #ifdef MRT6DEBUG 1704 if (mrt6debug & DEBUG_PIM) 1705 log(LOG_DEBUG, 1706 "pim6_input: invalid checksum\n"); 1707 #endif 1708 m_freem(m); 1709 return (IPPROTO_DONE); 1710 } 1711 } 1712 #endif /* PIM_CHECKSUM */ 1713 1714 if (pim->pim_type == PIM_REGISTER) { 1715 /* 1716 * since this is a REGISTER, we'll make a copy of the register 1717 * headers ip6+pim+u_int32_t+encap_ip6, to be passed up to the 1718 * routing daemon. 1719 */ 1720 static struct sockaddr_in6 dst = { sizeof(dst), AF_INET6 }; 1721 1722 struct mbuf *mcp; 1723 struct ip6_hdr *eip6; 1724 u_int32_t *reghdr; 1725 int rc; 1726 1727 ++pim6stat.pim6s_rcv_registers; 1728 1729 if ((reg_mif_num >= nummifs) || (reg_mif_num == (mifi_t) -1)) { 1730 #ifdef MRT6DEBUG 1731 if (mrt6debug & DEBUG_PIM) 1732 log(LOG_DEBUG, 1733 "pim6_input: register mif not set: %d\n", 1734 reg_mif_num); 1735 #endif 1736 m_freem(m); 1737 return (IPPROTO_DONE); 1738 } 1739 1740 reghdr = (u_int32_t *)(pim + 1); 1741 1742 if ((ntohl(*reghdr) & PIM_NULL_REGISTER)) 1743 goto pim6_input_to_daemon; 1744 1745 /* 1746 * Validate length 1747 */ 1748 if (pimlen < PIM6_REG_MINLEN) { 1749 ++pim6stat.pim6s_rcv_tooshort; 1750 ++pim6stat.pim6s_rcv_badregisters; 1751 #ifdef MRT6DEBUG 1752 log(LOG_ERR, 1753 "pim6_input: register packet size too " 1754 "small %d from %s\n", 1755 pimlen, ip6_sprintf(&ip6->ip6_src)); 1756 #endif 1757 m_freem(m); 1758 return (IPPROTO_DONE); 1759 } 1760 1761 eip6 = (struct ip6_hdr *) (reghdr + 1); 1762 #ifdef MRT6DEBUG 1763 if (mrt6debug & DEBUG_PIM) 1764 log(LOG_DEBUG, 1765 "pim6_input[register], eip6: %s -> %s, " 1766 "eip6 plen %d\n", 1767 ip6_sprintf(&eip6->ip6_src), 1768 ip6_sprintf(&eip6->ip6_dst), 1769 ntohs(eip6->ip6_plen)); 1770 #endif 1771 1772 /* verify the version number of the inner packet */ 1773 if ((eip6->ip6_vfc & IPV6_VERSION_MASK) != IPV6_VERSION) { 1774 ++pim6stat.pim6s_rcv_badregisters; 1775 #ifdef MRT6DEBUG 1776 log(LOG_DEBUG, "pim6_input: invalid IP version (%d) " 1777 "of the inner packet\n", 1778 (eip6->ip6_vfc & IPV6_VERSION)); 1779 #endif 1780 m_freem(m); 1781 return (IPPROTO_NONE); 1782 } 1783 1784 /* verify the inner packet is destined to a mcast group */ 1785 if (!IN6_IS_ADDR_MULTICAST(&eip6->ip6_dst)) { 1786 ++pim6stat.pim6s_rcv_badregisters; 1787 #ifdef MRT6DEBUG 1788 if (mrt6debug & DEBUG_PIM) 1789 log(LOG_DEBUG, 1790 "pim6_input: inner packet of register " 1791 "is not multicast %s\n", 1792 ip6_sprintf(&eip6->ip6_dst)); 1793 #endif 1794 m_freem(m); 1795 return (IPPROTO_DONE); 1796 } 1797 1798 /* 1799 * make a copy of the whole header to pass to the daemon later. 1800 */ 1801 mcp = m_copy(m, 0, off + PIM6_REG_MINLEN); 1802 if (mcp == NULL) { 1803 #ifdef MRT6DEBUG 1804 log(LOG_ERR, 1805 "pim6_input: pim register: " 1806 "could not copy register head\n"); 1807 #endif 1808 m_freem(m); 1809 return (IPPROTO_DONE); 1810 } 1811 1812 /* 1813 * forward the inner ip6 packet; point m_data at the inner ip6. 1814 */ 1815 m_adj(m, off + PIM_MINLEN); 1816 #ifdef MRT6DEBUG 1817 if (mrt6debug & DEBUG_PIM) { 1818 log(LOG_DEBUG, 1819 "pim6_input: forwarding decapsulated register: " 1820 "src %s, dst %s, mif %d\n", 1821 ip6_sprintf(&eip6->ip6_src), 1822 ip6_sprintf(&eip6->ip6_dst), 1823 reg_mif_num); 1824 } 1825 #endif 1826 1827 rc = looutput(mif6table[reg_mif_num].m6_ifp, m, 1828 (struct sockaddr *) &dst, 1829 (struct rtentry *) NULL); 1830 1831 /* prepare the register head to send to the mrouting daemon */ 1832 m = mcp; 1833 } 1834 1835 /* 1836 * Pass the PIM message up to the daemon; if it is a register message 1837 * pass the 'head' only up to the daemon. This includes the 1838 * encapsulator ip6 header, pim header, register header and the 1839 * encapsulated ip6 header. 1840 */ 1841 pim6_input_to_daemon: 1842 rip6_input(&m, offp, proto); 1843 return (IPPROTO_DONE); 1844 } 1845
Indexes created Wed Sep 24 05:09:52 GMT 2025