ip6_mroute.c revision 123396
1/* $FreeBSD: head/sys/netinet6/ip6_mroute.c 123396 2003-12-10 18:17:05Z ume $ */ 2/* $KAME: ip6_mroute.c,v 1.58 2001/12/18 02:36:31 itojun Exp $ */ 3 4/* 5 * Copyright (C) 1998 WIDE Project. 6 * All rights reserved. 7 * 8 * Redistribution and use in source and binary forms, with or without 9 * modification, are permitted provided that the following conditions 10 * are met: 11 * 1. Redistributions of source code must retain the above copyright 12 * notice, this list of conditions and the following disclaimer. 13 * 2. Redistributions in binary form must reproduce the above copyright 14 * notice, this list of conditions and the following disclaimer in the 15 * documentation and/or other materials provided with the distribution. 16 * 3. Neither the name of the project nor the names of its contributors 17 * may be used to endorse or promote products derived from this software 18 * without specific prior written permission. 19 * 20 * THIS SOFTWARE IS PROVIDED BY THE PROJECT AND CONTRIBUTORS ``AS IS'' AND 21 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE 22 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE 23 * ARE DISCLAIMED. IN NO EVENT SHALL THE PROJECT OR CONTRIBUTORS BE LIABLE 24 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL 25 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS 26 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) 27 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT 28 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY 29 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF 30 * SUCH DAMAGE. 31 */ 32 33/* BSDI ip_mroute.c,v 2.10 1996/11/14 00:29:52 jch Exp */ 34 35/* 36 * Copyright (c) 1989 Stephen Deering 37 * Copyright (c) 1992, 1993 38 * The Regents of the University of California. All rights reserved. 39 * 40 * This code is derived from software contributed to Berkeley by 41 * Stephen Deering of Stanford University. 42 * 43 * Redistribution and use in source and binary forms, with or without 44 * modification, are permitted provided that the following conditions 45 * are met: 46 * 1. Redistributions of source code must retain the above copyright 47 * notice, this list of conditions and the following disclaimer. 48 * 2. Redistributions in binary form must reproduce the above copyright 49 * notice, this list of conditions and the following disclaimer in the 50 * documentation and/or other materials provided with the distribution. 51 * 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 "opt_inet.h" 88#include "opt_inet6.h" 89 90#include <sys/param.h> 91#include <sys/callout.h> 92#include <sys/errno.h> 93#include <sys/kernel.h> 94#include <sys/lock.h> 95#include <sys/malloc.h> 96#include <sys/mbuf.h> 97#include <sys/protosw.h> 98#include <sys/signalvar.h> 99#include <sys/socket.h> 100#include <sys/socketvar.h> 101#include <sys/sockio.h> 102#include <sys/sx.h> 103#include <sys/syslog.h> 104#include <sys/systm.h> 105#include <sys/time.h> 106 107#include <net/if.h> 108#include <net/raw_cb.h> 109#include <net/route.h> 110 111#include <netinet/in.h> 112#include <netinet/in_var.h> 113 114#include <netinet/ip6.h> 115#include <netinet6/ip6_var.h> 116#include <netinet6/nd6.h> 117#include <netinet6/ip6_mroute.h> 118#include <netinet6/pim6.h> 119#include <netinet6/pim6_var.h> 120 121#include <net/net_osdep.h> 122 123static MALLOC_DEFINE(M_MRTABLE, "mf6c", "multicast forwarding cache entry"); 124 125#define M_HASCL(m) ((m)->m_flags & M_EXT) 126 127static int ip6_mdq __P((struct mbuf *, struct ifnet *, struct mf6c *)); 128static void phyint_send __P((struct ip6_hdr *, struct mif6 *, struct mbuf *)); 129 130static int set_pim6 __P((int *)); 131static int socket_send __P((struct socket *, struct mbuf *, 132 struct sockaddr_in6 *)); 133static int register_send __P((struct ip6_hdr *, struct mif6 *, 134 struct mbuf *)); 135 136/* 137 * Globals. All but ip6_mrouter, ip6_mrtproto and mrt6stat could be static, 138 * except for netstat or debugging purposes. 139 */ 140struct socket *ip6_mrouter = NULL; 141int ip6_mrouter_ver = 0; 142int ip6_mrtproto = IPPROTO_PIM; /* for netstat only */ 143struct mrt6stat mrt6stat; 144 145#define NO_RTE_FOUND 0x1 146#define RTE_FOUND 0x2 147 148struct mf6c *mf6ctable[MF6CTBLSIZ]; 149u_char n6expire[MF6CTBLSIZ]; 150static struct mif6 mif6table[MAXMIFS]; 151#ifdef MRT6DEBUG 152u_int mrt6debug = 0; /* debug level */ 153#define DEBUG_MFC 0x02 154#define DEBUG_FORWARD 0x04 155#define DEBUG_EXPIRE 0x08 156#define DEBUG_XMIT 0x10 157#define DEBUG_REG 0x20 158#define DEBUG_PIM 0x40 159#endif 160 161static void expire_upcalls __P((void *)); 162#define EXPIRE_TIMEOUT (hz / 4) /* 4x / second */ 163#define UPCALL_EXPIRE 6 /* number of timeouts */ 164 165#ifdef INET 166#ifdef MROUTING 167extern struct socket *ip_mrouter; 168#endif 169#endif 170 171/* 172 * 'Interfaces' associated with decapsulator (so we can tell 173 * packets that went through it from ones that get reflected 174 * by a broken gateway). These interfaces are never linked into 175 * the system ifnet list & no routes point to them. I.e., packets 176 * can't be sent this way. They only exist as a placeholder for 177 * multicast source verification. 178 */ 179struct ifnet multicast_register_if; 180 181#define ENCAP_HOPS 64 182 183/* 184 * Private variables. 185 */ 186static mifi_t nummifs = 0; 187static mifi_t reg_mif_num = (mifi_t)-1; 188 189static struct pim6stat pim6stat; 190static int pim6; 191 192/* 193 * Hash function for a source, group entry 194 */ 195#define MF6CHASH(a, g) MF6CHASHMOD((a).s6_addr32[0] ^ (a).s6_addr32[1] ^ \ 196 (a).s6_addr32[2] ^ (a).s6_addr32[3] ^ \ 197 (g).s6_addr32[0] ^ (g).s6_addr32[1] ^ \ 198 (g).s6_addr32[2] ^ (g).s6_addr32[3]) 199 200/* 201 * Find a route for a given origin IPv6 address and Multicast group address. 202 * Quality of service parameter to be added in the future!!! 203 */ 204 205#define MF6CFIND(o, g, rt) do { \ 206 struct mf6c *_rt = mf6ctable[MF6CHASH(o,g)]; \ 207 rt = NULL; \ 208 mrt6stat.mrt6s_mfc_lookups++; \ 209 while (_rt) { \ 210 if (IN6_ARE_ADDR_EQUAL(&_rt->mf6c_origin.sin6_addr, &(o)) && \ 211 IN6_ARE_ADDR_EQUAL(&_rt->mf6c_mcastgrp.sin6_addr, &(g)) && \ 212 (_rt->mf6c_stall == NULL)) { \ 213 rt = _rt; \ 214 break; \ 215 } \ 216 _rt = _rt->mf6c_next; \ 217 } \ 218 if (rt == NULL) { \ 219 mrt6stat.mrt6s_mfc_misses++; \ 220 } \ 221} while (/*CONSTCOND*/ 0) 222 223/* 224 * Macros to compute elapsed time efficiently 225 * Borrowed from Van Jacobson's scheduling code 226 */ 227#define TV_DELTA(a, b, delta) do { \ 228 int xxs; \ 229 \ 230 delta = (a).tv_usec - (b).tv_usec; \ 231 if ((xxs = (a).tv_sec - (b).tv_sec)) { \ 232 switch (xxs) { \ 233 case 2: \ 234 delta += 1000000; \ 235 /* FALLTHROUGH */ \ 236 case 1: \ 237 delta += 1000000; \ 238 break; \ 239 default: \ 240 delta += (1000000 * xxs); \ 241 } \ 242 } \ 243} while (/*CONSTCOND*/ 0) 244 245#define TV_LT(a, b) (((a).tv_usec < (b).tv_usec && \ 246 (a).tv_sec <= (b).tv_sec) || (a).tv_sec < (b).tv_sec) 247 248#ifdef UPCALL_TIMING 249#define UPCALL_MAX 50 250u_long upcall_data[UPCALL_MAX + 1]; 251static void collate(); 252#endif /* UPCALL_TIMING */ 253 254static int get_sg_cnt __P((struct sioc_sg_req6 *)); 255static int get_mif6_cnt __P((struct sioc_mif_req6 *)); 256static int ip6_mrouter_init __P((struct socket *, int, int)); 257static int add_m6if __P((struct mif6ctl *)); 258static int del_m6if __P((mifi_t *)); 259static int add_m6fc __P((struct mf6cctl *)); 260static int del_m6fc __P((struct mf6cctl *)); 261 262static struct callout expire_upcalls_ch; 263 264/* 265 * Handle MRT setsockopt commands to modify the multicast routing tables. 266 */ 267int 268ip6_mrouter_set(so, sopt) 269 struct socket *so; 270 struct sockopt *sopt; 271{ 272 int error = 0; 273 int optval; 274 struct mif6ctl mifc; 275 struct mf6cctl mfcc; 276 mifi_t mifi; 277 278 if (so != ip6_mrouter && sopt->sopt_name != MRT6_INIT) 279 return (EACCES); 280 281 switch (sopt->sopt_name) { 282 case MRT6_INIT: 283#ifdef MRT6_OINIT 284 case MRT6_OINIT: 285#endif 286 error = sooptcopyin(sopt, &optval, sizeof(optval), 287 sizeof(optval)); 288 if (error) 289 break; 290 error = ip6_mrouter_init(so, optval, sopt->sopt_name); 291 break; 292 case MRT6_DONE: 293 error = ip6_mrouter_done(); 294 break; 295 case MRT6_ADD_MIF: 296 error = sooptcopyin(sopt, &mifc, sizeof(mifc), sizeof(mifc)); 297 if (error) 298 break; 299 error = add_m6if(&mifc); 300 break; 301 case MRT6_ADD_MFC: 302 error = sooptcopyin(sopt, &mfcc, sizeof(mfcc), sizeof(mfcc)); 303 if (error) 304 break; 305 error = add_m6fc(&mfcc); 306 break; 307 case MRT6_DEL_MFC: 308 error = sooptcopyin(sopt, &mfcc, sizeof(mfcc), sizeof(mfcc)); 309 if (error) 310 break; 311 error = del_m6fc(&mfcc); 312 break; 313 case MRT6_DEL_MIF: 314 error = sooptcopyin(sopt, &mifi, sizeof(mifi), sizeof(mifi)); 315 if (error) 316 break; 317 error = del_m6if(&mifi); 318 break; 319 case MRT6_PIM: 320 error = sooptcopyin(sopt, &optval, sizeof(optval), 321 sizeof(optval)); 322 if (error) 323 break; 324 error = set_pim6(&optval); 325 break; 326 default: 327 error = EOPNOTSUPP; 328 break; 329 } 330 331 return (error); 332} 333 334/* 335 * Handle MRT getsockopt commands 336 */ 337int 338ip6_mrouter_get(so, sopt) 339 struct socket *so; 340 struct sockopt *sopt; 341{ 342 int error = 0; 343 344 if (so != ip6_mrouter) 345 return (EACCES); 346 347 switch (sopt->sopt_name) { 348 case MRT6_PIM: 349 error = sooptcopyout(sopt, &pim6, sizeof(pim6)); 350 break; 351 } 352 return (error); 353} 354 355/* 356 * Handle ioctl commands to obtain information from the cache 357 */ 358int 359mrt6_ioctl(cmd, data) 360 int cmd; 361 caddr_t data; 362{ 363 switch (cmd) { 364 case SIOCGETSGCNT_IN6: 365 return (get_sg_cnt((struct sioc_sg_req6 *)data)); 366 case SIOCGETMIFCNT_IN6: 367 return (get_mif6_cnt((struct sioc_mif_req6 *)data)); 368 default: 369 return (EINVAL); 370 } 371} 372 373/* 374 * returns the packet, byte, rpf-failure count for the source group provided 375 */ 376static int 377get_sg_cnt(req) 378 struct sioc_sg_req6 *req; 379{ 380 struct mf6c *rt; 381 int s; 382 383 s = splnet(); 384 MF6CFIND(req->src.sin6_addr, req->grp.sin6_addr, rt); 385 splx(s); 386 if (rt != NULL) { 387 req->pktcnt = rt->mf6c_pkt_cnt; 388 req->bytecnt = rt->mf6c_byte_cnt; 389 req->wrong_if = rt->mf6c_wrong_if; 390 } else 391 return (ESRCH); 392#if 0 393 req->pktcnt = req->bytecnt = req->wrong_if = 0xffffffff; 394#endif 395 396 return (0); 397} 398 399/* 400 * returns the input and output packet and byte counts on the mif provided 401 */ 402static int 403get_mif6_cnt(req) 404 struct sioc_mif_req6 *req; 405{ 406 mifi_t mifi = req->mifi; 407 408 if (mifi >= nummifs) 409 return (EINVAL); 410 411 req->icount = mif6table[mifi].m6_pkt_in; 412 req->ocount = mif6table[mifi].m6_pkt_out; 413 req->ibytes = mif6table[mifi].m6_bytes_in; 414 req->obytes = mif6table[mifi].m6_bytes_out; 415 416 return (0); 417} 418 419static int 420set_pim6(i) 421 int *i; 422{ 423 if ((*i != 1) && (*i != 0)) 424 return (EINVAL); 425 426 pim6 = *i; 427 428 return (0); 429} 430 431/* 432 * Enable multicast routing 433 */ 434static int 435ip6_mrouter_init(so, v, cmd) 436 struct socket *so; 437 int v; 438 int cmd; 439{ 440#ifdef MRT6DEBUG 441 if (mrt6debug) 442 log(LOG_DEBUG, 443 "ip6_mrouter_init: so_type = %d, pr_protocol = %d\n", 444 so->so_type, so->so_proto->pr_protocol); 445#endif 446 447 if (so->so_type != SOCK_RAW || 448 so->so_proto->pr_protocol != IPPROTO_ICMPV6) 449 return (EOPNOTSUPP); 450 451 if (v != 1) 452 return (ENOPROTOOPT); 453 454 if (ip6_mrouter != NULL) 455 return (EADDRINUSE); 456 457 ip6_mrouter = so; 458 ip6_mrouter_ver = cmd; 459 460 bzero((caddr_t)mf6ctable, sizeof(mf6ctable)); 461 bzero((caddr_t)n6expire, sizeof(n6expire)); 462 463 pim6 = 0;/* used for stubbing out/in pim stuff */ 464 465 callout_reset(&expire_upcalls_ch, EXPIRE_TIMEOUT, 466 expire_upcalls, NULL); 467 468#ifdef MRT6DEBUG 469 if (mrt6debug) 470 log(LOG_DEBUG, "ip6_mrouter_init\n"); 471#endif 472 473 return (0); 474} 475 476/* 477 * Disable multicast routing 478 */ 479int 480ip6_mrouter_done() 481{ 482 mifi_t mifi; 483 int i; 484 struct ifnet *ifp; 485 struct in6_ifreq ifr; 486 struct mf6c *rt; 487 struct rtdetq *rte; 488 int s; 489 490 s = splnet(); 491 492 /* 493 * For each phyint in use, disable promiscuous reception of all IPv6 494 * multicasts. 495 */ 496#ifdef INET 497#ifdef MROUTING 498 /* 499 * If there is still IPv4 multicast routing daemon, 500 * we remain interfaces to receive all muliticasted packets. 501 * XXX: there may be an interface in which the IPv4 multicast 502 * daemon is not interested... 503 */ 504 if (!ip_mrouter) 505#endif 506#endif 507 { 508 for (mifi = 0; mifi < nummifs; mifi++) { 509 if (mif6table[mifi].m6_ifp && 510 !(mif6table[mifi].m6_flags & MIFF_REGISTER)) { 511 ifr.ifr_addr.sin6_family = AF_INET6; 512 ifr.ifr_addr.sin6_addr = in6addr_any; 513 ifp = mif6table[mifi].m6_ifp; 514 (*ifp->if_ioctl)(ifp, SIOCDELMULTI, 515 (caddr_t)&ifr); 516 } 517 } 518 } 519#ifdef notyet 520 bzero((caddr_t)qtable, sizeof(qtable)); 521 bzero((caddr_t)tbftable, sizeof(tbftable)); 522#endif 523 bzero((caddr_t)mif6table, sizeof(mif6table)); 524 nummifs = 0; 525 526 pim6 = 0; /* used to stub out/in pim specific code */ 527 528 callout_stop(&expire_upcalls_ch); 529 530 /* 531 * Free all multicast forwarding cache entries. 532 */ 533 for (i = 0; i < MF6CTBLSIZ; i++) { 534 rt = mf6ctable[i]; 535 while (rt) { 536 struct mf6c *frt; 537 538 for (rte = rt->mf6c_stall; rte != NULL; ) { 539 struct rtdetq *n = rte->next; 540 541 m_free(rte->m); 542 free(rte, M_MRTABLE); 543 rte = n; 544 } 545 frt = rt; 546 rt = rt->mf6c_next; 547 free(frt, M_MRTABLE); 548 } 549 } 550 551 bzero((caddr_t)mf6ctable, sizeof(mf6ctable)); 552 553 /* 554 * Reset de-encapsulation cache 555 */ 556 reg_mif_num = -1; 557 558 ip6_mrouter = NULL; 559 ip6_mrouter_ver = 0; 560 561 splx(s); 562 563#ifdef MRT6DEBUG 564 if (mrt6debug) 565 log(LOG_DEBUG, "ip6_mrouter_done\n"); 566#endif 567 568 return (0); 569} 570 571static struct sockaddr_in6 sin6 = { sizeof(sin6), AF_INET6 }; 572 573/* 574 * Add a mif to the mif table 575 */ 576static int 577add_m6if(mifcp) 578 struct mif6ctl *mifcp; 579{ 580 struct mif6 *mifp; 581 struct ifnet *ifp; 582 int error, s; 583#ifdef notyet 584 struct tbf *m_tbf = tbftable + mifcp->mif6c_mifi; 585#endif 586 587 if (mifcp->mif6c_mifi >= MAXMIFS) 588 return (EINVAL); 589 mifp = mif6table + mifcp->mif6c_mifi; 590 if (mifp->m6_ifp) 591 return (EADDRINUSE); /* XXX: is it appropriate? */ 592 if (mifcp->mif6c_pifi == 0 || mifcp->mif6c_pifi > if_index) 593 return (ENXIO); 594 ifp = ifnet_byindex(mifcp->mif6c_pifi); 595 596 if (mifcp->mif6c_flags & MIFF_REGISTER) { 597 if (reg_mif_num == (mifi_t)-1) { 598 strlcpy(multicast_register_if.if_xname, "register_mif", 599 IFNAMSIZ); 600 multicast_register_if.if_flags |= IFF_LOOPBACK; 601 multicast_register_if.if_index = mifcp->mif6c_mifi; 602 reg_mif_num = mifcp->mif6c_mifi; 603 } 604 605 ifp = &multicast_register_if; 606 607 } /* if REGISTER */ 608 else { 609 /* Make sure the interface supports multicast */ 610 if ((ifp->if_flags & IFF_MULTICAST) == 0) 611 return (EOPNOTSUPP); 612 613 s = splnet(); 614 error = if_allmulti(ifp, 1); 615 splx(s); 616 if (error) 617 return (error); 618 } 619 620 s = splnet(); 621 mifp->m6_flags = mifcp->mif6c_flags; 622 mifp->m6_ifp = ifp; 623#ifdef notyet 624 /* scaling up here allows division by 1024 in critical code */ 625 mifp->m6_rate_limit = mifcp->mif6c_rate_limit * 1024 / 1000; 626#endif 627 /* initialize per mif pkt counters */ 628 mifp->m6_pkt_in = 0; 629 mifp->m6_pkt_out = 0; 630 mifp->m6_bytes_in = 0; 631 mifp->m6_bytes_out = 0; 632 splx(s); 633 634 /* Adjust nummifs up if the mifi is higher than nummifs */ 635 if (nummifs <= mifcp->mif6c_mifi) 636 nummifs = mifcp->mif6c_mifi + 1; 637 638#ifdef MRT6DEBUG 639 if (mrt6debug) 640 log(LOG_DEBUG, 641 "add_mif #%d, phyint %s\n", 642 mifcp->mif6c_mifi, 643 ifp->if_xname); 644#endif 645 646 return (0); 647} 648 649/* 650 * Delete a mif from the mif table 651 */ 652static int 653del_m6if(mifip) 654 mifi_t *mifip; 655{ 656 struct mif6 *mifp = mif6table + *mifip; 657 mifi_t mifi; 658 struct ifnet *ifp; 659 int s; 660 661 if (*mifip >= nummifs) 662 return (EINVAL); 663 if (mifp->m6_ifp == NULL) 664 return (EINVAL); 665 666 s = splnet(); 667 668 if (!(mifp->m6_flags & MIFF_REGISTER)) { 669 /* 670 * XXX: what if there is yet IPv4 multicast daemon 671 * using the interface? 672 */ 673 ifp = mifp->m6_ifp; 674 675 if_allmulti(ifp, 0); 676 } 677 678#ifdef notyet 679 bzero((caddr_t)qtable[*mifip], sizeof(qtable[*mifip])); 680 bzero((caddr_t)mifp->m6_tbf, sizeof(*(mifp->m6_tbf))); 681#endif 682 bzero((caddr_t)mifp, sizeof(*mifp)); 683 684 /* Adjust nummifs down */ 685 for (mifi = nummifs; mifi > 0; mifi--) 686 if (mif6table[mifi - 1].m6_ifp) 687 break; 688 nummifs = mifi; 689 690 splx(s); 691 692#ifdef MRT6DEBUG 693 if (mrt6debug) 694 log(LOG_DEBUG, "del_m6if %d, nummifs %d\n", *mifip, nummifs); 695#endif 696 697 return (0); 698} 699 700/* 701 * Add an mfc entry 702 */ 703static int 704add_m6fc(mfccp) 705 struct mf6cctl *mfccp; 706{ 707 struct mf6c *rt; 708 u_long hash; 709 struct rtdetq *rte; 710 u_short nstl; 711 int s; 712 713 MF6CFIND(mfccp->mf6cc_origin.sin6_addr, 714 mfccp->mf6cc_mcastgrp.sin6_addr, rt); 715 716 /* If an entry already exists, just update the fields */ 717 if (rt) { 718#ifdef MRT6DEBUG 719 if (mrt6debug & DEBUG_MFC) 720 log(LOG_DEBUG, 721 "add_m6fc no upcall h %d 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 = splnet(); 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 = splnet(); 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 */ 860static void 861collate(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 */ 886static int 887del_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 = splnet(); 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 934static int 935socket_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 963int 964ip6_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; 974 975#ifdef MRT6DEBUG 976 if (mrt6debug & DEBUG_FORWARD) 977 log(LOG_DEBUG, "ip6_mforward: src %s, dst %s, ifindex %d\n", 978 ip6_sprintf(&ip6->ip6_src), ip6_sprintf(&ip6->ip6_dst), 979 ifp->if_index); 980#endif 981 982 /* 983 * Don't forward a packet with Hop limit of zero or one, 984 * or a packet destined to a local-only group. 985 */ 986 if (ip6->ip6_hlim <= 1 || IN6_IS_ADDR_MC_INTFACELOCAL(&ip6->ip6_dst) || 987 IN6_IS_ADDR_MC_LINKLOCAL(&ip6->ip6_dst)) 988 return (0); 989 ip6->ip6_hlim--; 990 991 /* 992 * Source address check: do not forward packets with unspecified 993 * source. It was discussed in July 2000, on ipngwg mailing list. 994 * This is rather more serious than unicast cases, because some 995 * MLD packets can be sent with the unspecified source address 996 * (although such packets must normally set 1 to the hop limit field). 997 */ 998 if (IN6_IS_ADDR_UNSPECIFIED(&ip6->ip6_src)) { 999 ip6stat.ip6s_cantforward++; 1000 if (ip6_log_time + ip6_log_interval < time_second) { 1001 ip6_log_time = time_second; 1002 log(LOG_DEBUG, 1003 "cannot forward " 1004 "from %s to %s nxt %d received on %s\n", 1005 ip6_sprintf(&ip6->ip6_src), 1006 ip6_sprintf(&ip6->ip6_dst), 1007 ip6->ip6_nxt, 1008 if_name(m->m_pkthdr.rcvif)); 1009 } 1010 return (0); 1011 } 1012 1013 /* 1014 * Determine forwarding mifs from the forwarding cache table 1015 */ 1016 s = splnet(); 1017 MF6CFIND(ip6->ip6_src, ip6->ip6_dst, rt); 1018 1019 /* Entry exists, so forward if necessary */ 1020 if (rt) { 1021 splx(s); 1022 return (ip6_mdq(m, ifp, rt)); 1023 } else { 1024 /* 1025 * If we don't have a route for packet's origin, 1026 * Make a copy of the packet & 1027 * send message to routing daemon 1028 */ 1029 1030 struct mbuf *mb0; 1031 struct rtdetq *rte; 1032 u_long hash; 1033/* int i, npkts;*/ 1034#ifdef UPCALL_TIMING 1035 struct timeval tp; 1036 1037 GET_TIME(tp); 1038#endif /* UPCALL_TIMING */ 1039 1040 mrt6stat.mrt6s_no_route++; 1041#ifdef MRT6DEBUG 1042 if (mrt6debug & (DEBUG_FORWARD | DEBUG_MFC)) 1043 log(LOG_DEBUG, "ip6_mforward: no rte s %s g %s\n", 1044 ip6_sprintf(&ip6->ip6_src), 1045 ip6_sprintf(&ip6->ip6_dst)); 1046#endif 1047 1048 /* 1049 * Allocate mbufs early so that we don't do extra work if we 1050 * are just going to fail anyway. 1051 */ 1052 rte = (struct rtdetq *)malloc(sizeof(*rte), M_MRTABLE, 1053 M_NOWAIT); 1054 if (rte == NULL) { 1055 splx(s); 1056 return (ENOBUFS); 1057 } 1058 mb0 = m_copy(m, 0, M_COPYALL); 1059 /* 1060 * Pullup packet header if needed before storing it, 1061 * as other references may modify it in the meantime. 1062 */ 1063 if (mb0 && 1064 (M_HASCL(mb0) || mb0->m_len < sizeof(struct ip6_hdr))) 1065 mb0 = m_pullup(mb0, sizeof(struct ip6_hdr)); 1066 if (mb0 == NULL) { 1067 free(rte, M_MRTABLE); 1068 splx(s); 1069 return (ENOBUFS); 1070 } 1071 1072 /* is there an upcall waiting for this packet? */ 1073 hash = MF6CHASH(ip6->ip6_src, ip6->ip6_dst); 1074 for (rt = mf6ctable[hash]; rt; rt = rt->mf6c_next) { 1075 if (IN6_ARE_ADDR_EQUAL(&ip6->ip6_src, 1076 &rt->mf6c_origin.sin6_addr) && 1077 IN6_ARE_ADDR_EQUAL(&ip6->ip6_dst, 1078 &rt->mf6c_mcastgrp.sin6_addr) && 1079 (rt->mf6c_stall != NULL)) 1080 break; 1081 } 1082 1083 if (rt == NULL) { 1084 struct mrt6msg *im; 1085#ifdef MRT6_OINIT 1086 struct omrt6msg *oim; 1087#endif 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#ifdef MRT6_OINIT 1119 oim = NULL; 1120#endif 1121 switch (ip6_mrouter_ver) { 1122#ifdef MRT6_OINIT 1123 case MRT6_OINIT: 1124 oim = mtod(mm, struct omrt6msg *); 1125 oim->im6_msgtype = MRT6MSG_NOCACHE; 1126 oim->im6_mbz = 0; 1127 break; 1128#endif 1129 case MRT6_INIT: 1130 im = mtod(mm, struct mrt6msg *); 1131 im->im6_msgtype = MRT6MSG_NOCACHE; 1132 im->im6_mbz = 0; 1133 break; 1134 default: 1135 free(rte, M_MRTABLE); 1136 m_freem(mb0); 1137 free(rt, M_MRTABLE); 1138 splx(s); 1139 return (EINVAL); 1140 } 1141 1142#ifdef MRT6DEBUG 1143 if (mrt6debug & DEBUG_FORWARD) 1144 log(LOG_DEBUG, 1145 "getting the iif info in the kernel\n"); 1146#endif 1147 1148 for (mifp = mif6table, mifi = 0; 1149 mifi < nummifs && mifp->m6_ifp != ifp; 1150 mifp++, mifi++) 1151 ; 1152 1153 switch (ip6_mrouter_ver) { 1154#ifdef MRT6_OINIT 1155 case MRT6_OINIT: 1156 oim->im6_mif = mifi; 1157 break; 1158#endif 1159 case MRT6_INIT: 1160 im->im6_mif = mifi; 1161 break; 1162 } 1163 1164 if (socket_send(ip6_mrouter, mm, &sin6) < 0) { 1165 log(LOG_WARNING, "ip6_mforward: ip6_mrouter " 1166 "socket queue full\n"); 1167 mrt6stat.mrt6s_upq_sockfull++; 1168 free(rte, M_MRTABLE); 1169 m_freem(mb0); 1170 free(rt, M_MRTABLE); 1171 splx(s); 1172 return (ENOBUFS); 1173 } 1174 1175 mrt6stat.mrt6s_upcalls++; 1176 1177 /* insert new entry at head of hash chain */ 1178 bzero(rt, sizeof(*rt)); 1179 rt->mf6c_origin.sin6_family = AF_INET6; 1180 rt->mf6c_origin.sin6_len = sizeof(struct sockaddr_in6); 1181 rt->mf6c_origin.sin6_addr = ip6->ip6_src; 1182 rt->mf6c_mcastgrp.sin6_family = AF_INET6; 1183 rt->mf6c_mcastgrp.sin6_len = sizeof(struct sockaddr_in6); 1184 rt->mf6c_mcastgrp.sin6_addr = ip6->ip6_dst; 1185 rt->mf6c_expire = UPCALL_EXPIRE; 1186 n6expire[hash]++; 1187 rt->mf6c_parent = MF6C_INCOMPLETE_PARENT; 1188 1189 /* link into table */ 1190 rt->mf6c_next = mf6ctable[hash]; 1191 mf6ctable[hash] = rt; 1192 /* Add this entry to the end of the queue */ 1193 rt->mf6c_stall = rte; 1194 } else { 1195 /* determine if q has overflowed */ 1196 struct rtdetq **p; 1197 int npkts = 0; 1198 1199 for (p = &rt->mf6c_stall; *p != NULL; p = &(*p)->next) 1200 if (++npkts > MAX_UPQ6) { 1201 mrt6stat.mrt6s_upq_ovflw++; 1202 free(rte, M_MRTABLE); 1203 m_freem(mb0); 1204 splx(s); 1205 return (0); 1206 } 1207 1208 /* Add this entry to the end of the queue */ 1209 *p = rte; 1210 } 1211 1212 rte->next = NULL; 1213 rte->m = mb0; 1214 rte->ifp = ifp; 1215#ifdef UPCALL_TIMING 1216 rte->t = tp; 1217#endif /* UPCALL_TIMING */ 1218 1219 splx(s); 1220 1221 return (0); 1222 } 1223} 1224 1225/* 1226 * Clean up cache entries if upcalls are not serviced 1227 * Call from the Slow Timeout mechanism, every half second. 1228 */ 1229static void 1230expire_upcalls(unused) 1231 void *unused; 1232{ 1233 struct rtdetq *rte; 1234 struct mf6c *mfc, **nptr; 1235 int i; 1236 int s; 1237 1238 s = splnet(); 1239 for (i = 0; i < MF6CTBLSIZ; i++) { 1240 if (n6expire[i] == 0) 1241 continue; 1242 nptr = &mf6ctable[i]; 1243 while ((mfc = *nptr) != NULL) { 1244 rte = mfc->mf6c_stall; 1245 /* 1246 * Skip real cache entries 1247 * Make sure it wasn't marked to not expire (shouldn't happen) 1248 * If it expires now 1249 */ 1250 if (rte != NULL && 1251 mfc->mf6c_expire != 0 && 1252 --mfc->mf6c_expire == 0) { 1253#ifdef MRT6DEBUG 1254 if (mrt6debug & DEBUG_EXPIRE) 1255 log(LOG_DEBUG, "expire_upcalls: expiring (%s %s)\n", 1256 ip6_sprintf(&mfc->mf6c_origin.sin6_addr), 1257 ip6_sprintf(&mfc->mf6c_mcastgrp.sin6_addr)); 1258#endif 1259 /* 1260 * drop all the packets 1261 * free the mbuf with the pkt, if, timing info 1262 */ 1263 do { 1264 struct rtdetq *n = rte->next; 1265 m_freem(rte->m); 1266 free(rte, M_MRTABLE); 1267 rte = n; 1268 } while (rte != NULL); 1269 mrt6stat.mrt6s_cache_cleanups++; 1270 n6expire[i]--; 1271 1272 *nptr = mfc->mf6c_next; 1273 free(mfc, M_MRTABLE); 1274 } else { 1275 nptr = &mfc->mf6c_next; 1276 } 1277 } 1278 } 1279 splx(s); 1280 callout_reset(&expire_upcalls_ch, EXPIRE_TIMEOUT, 1281 expire_upcalls, NULL); 1282} 1283 1284/* 1285 * Packet forwarding routine once entry in the cache is made 1286 */ 1287static int 1288ip6_mdq(m, ifp, rt) 1289 struct mbuf *m; 1290 struct ifnet *ifp; 1291 struct mf6c *rt; 1292{ 1293 struct ip6_hdr *ip6 = mtod(m, struct ip6_hdr *); 1294 mifi_t mifi, iif; 1295 struct mif6 *mifp; 1296 int plen = m->m_pkthdr.len; 1297 u_int32_t dscopein, sscopein; 1298 1299/* 1300 * Macro to send packet on mif. Since RSVP packets don't get counted on 1301 * input, they shouldn't get counted on output, so statistics keeping is 1302 * separate. 1303 */ 1304 1305#define MC6_SEND(ip6, mifp, m) do { \ 1306 if ((mifp)->m6_flags & MIFF_REGISTER) \ 1307 register_send((ip6), (mifp), (m)); \ 1308 else \ 1309 phyint_send((ip6), (mifp), (m)); \ 1310} while (/*CONSTCOND*/ 0) 1311 1312 /* 1313 * Don't forward if it didn't arrive from the parent mif 1314 * for its origin. 1315 */ 1316 mifi = rt->mf6c_parent; 1317 if ((mifi >= nummifs) || (mif6table[mifi].m6_ifp != ifp)) { 1318 /* came in the wrong interface */ 1319#ifdef MRT6DEBUG 1320 if (mrt6debug & DEBUG_FORWARD) 1321 log(LOG_DEBUG, 1322 "wrong if: ifid %d mifi %d mififid %x\n", 1323 ifp->if_index, mifi, 1324 mif6table[mifi].m6_ifp->if_index); 1325#endif 1326 mrt6stat.mrt6s_wrong_if++; 1327 rt->mf6c_wrong_if++; 1328 /* 1329 * If we are doing PIM processing, and we are forwarding 1330 * packets on this interface, send a message to the 1331 * routing daemon. 1332 */ 1333 /* have to make sure this is a valid mif */ 1334 if (mifi < nummifs && mif6table[mifi].m6_ifp) 1335 if (pim6 && (m->m_flags & M_LOOP) == 0) { 1336 /* 1337 * Check the M_LOOP flag to avoid an 1338 * unnecessary PIM assert. 1339 * XXX: M_LOOP is an ad-hoc hack... 1340 */ 1341 static struct sockaddr_in6 sin6 = 1342 { sizeof(sin6), AF_INET6 }; 1343 1344 struct mbuf *mm; 1345 struct mrt6msg *im; 1346#ifdef MRT6_OINIT 1347 struct omrt6msg *oim; 1348#endif 1349 1350 mm = m_copy(m, 0, sizeof(struct ip6_hdr)); 1351 if (mm && 1352 (M_HASCL(mm) || 1353 mm->m_len < sizeof(struct ip6_hdr))) 1354 mm = m_pullup(mm, sizeof(struct ip6_hdr)); 1355 if (mm == NULL) 1356 return (ENOBUFS); 1357 1358#ifdef MRT6_OINIT 1359 oim = NULL; 1360#endif 1361 im = NULL; 1362 switch (ip6_mrouter_ver) { 1363#ifdef MRT6_OINIT 1364 case MRT6_OINIT: 1365 oim = mtod(mm, struct omrt6msg *); 1366 oim->im6_msgtype = MRT6MSG_WRONGMIF; 1367 oim->im6_mbz = 0; 1368 break; 1369#endif 1370 case MRT6_INIT: 1371 im = mtod(mm, struct mrt6msg *); 1372 im->im6_msgtype = MRT6MSG_WRONGMIF; 1373 im->im6_mbz = 0; 1374 break; 1375 default: 1376 m_freem(mm); 1377 return (EINVAL); 1378 } 1379 1380 for (mifp = mif6table, iif = 0; 1381 iif < nummifs && mifp && 1382 mifp->m6_ifp != ifp; 1383 mifp++, iif++) 1384 ; 1385 1386 switch (ip6_mrouter_ver) { 1387#ifdef MRT6_OINIT 1388 case MRT6_OINIT: 1389 oim->im6_mif = iif; 1390 sin6.sin6_addr = oim->im6_src; 1391 break; 1392#endif 1393 case MRT6_INIT: 1394 im->im6_mif = iif; 1395 sin6.sin6_addr = im->im6_src; 1396 break; 1397 } 1398 1399 mrt6stat.mrt6s_upcalls++; 1400 1401 if (socket_send(ip6_mrouter, mm, &sin6) < 0) { 1402#ifdef MRT6DEBUG 1403 if (mrt6debug) 1404 log(LOG_WARNING, "mdq, ip6_mrouter socket queue full\n"); 1405#endif 1406 ++mrt6stat.mrt6s_upq_sockfull; 1407 return (ENOBUFS); 1408 } /* if socket Q full */ 1409 } /* if PIM */ 1410 return (0); 1411 } /* if wrong iif */ 1412 1413 /* If I sourced this packet, it counts as output, else it was input. */ 1414 if (m->m_pkthdr.rcvif == NULL) { 1415 /* XXX: is rcvif really NULL when output?? */ 1416 mif6table[mifi].m6_pkt_out++; 1417 mif6table[mifi].m6_bytes_out += plen; 1418 } else { 1419 mif6table[mifi].m6_pkt_in++; 1420 mif6table[mifi].m6_bytes_in += plen; 1421 } 1422 rt->mf6c_pkt_cnt++; 1423 rt->mf6c_byte_cnt += plen; 1424 1425 /* 1426 * For each mif, forward a copy of the packet if there are group 1427 * members downstream on the interface. 1428 */ 1429 if (in6_addr2zoneid(ifp, &ip6->ip6_dst, &dscopein) || 1430 in6_addr2zoneid(ifp, &ip6->ip6_src, &sscopein)) 1431 return (EINVAL); 1432 for (mifp = mif6table, mifi = 0; mifi < nummifs; mifp++, mifi++) { 1433 if (IF_ISSET(mifi, &rt->mf6c_ifset)) { 1434 u_int32_t dscopeout, sscopeout; 1435 1436 /* 1437 * check if the outgoing packet is going to break 1438 * a scope boundary. 1439 * XXX For packets through PIM register tunnel 1440 * interface, we believe a routing daemon. 1441 */ 1442 if (!(mif6table[rt->mf6c_parent].m6_flags & 1443 MIFF_REGISTER) && 1444 !(mif6table[mifi].m6_flags & MIFF_REGISTER)) { 1445 if (in6_addr2zoneid(mif6table[mifi].m6_ifp, 1446 &ip6->ip6_dst, 1447 &dscopeout) || 1448 in6_addr2zoneid(mif6table[mifi].m6_ifp, 1449 &ip6->ip6_src, 1450 &sscopeout) || 1451 dscopein != dscopeout || 1452 sscopein != sscopeout) { 1453 ip6stat.ip6s_badscope++; 1454 continue; 1455 } 1456 } 1457 1458 mifp->m6_pkt_out++; 1459 mifp->m6_bytes_out += plen; 1460 MC6_SEND(ip6, mifp, m); 1461 } 1462 } 1463 return (0); 1464} 1465 1466static void 1467phyint_send(ip6, mifp, m) 1468 struct ip6_hdr *ip6; 1469 struct mif6 *mifp; 1470 struct mbuf *m; 1471{ 1472 struct mbuf *mb_copy; 1473 struct ifnet *ifp = mifp->m6_ifp; 1474 int error = 0; 1475 int s = splnet(); /* needs to protect static "ro" below. */ 1476 static struct route_in6 ro; 1477 struct in6_multi *in6m; 1478 struct sockaddr_in6 *dst6; 1479 u_long linkmtu; 1480 1481 /* 1482 * Make a new reference to the packet; make sure that 1483 * the IPv6 header is actually copied, not just referenced, 1484 * so that ip6_output() only scribbles on the copy. 1485 */ 1486 mb_copy = m_copy(m, 0, M_COPYALL); 1487 if (mb_copy && 1488 (M_HASCL(mb_copy) || mb_copy->m_len < sizeof(struct ip6_hdr))) 1489 mb_copy = m_pullup(mb_copy, sizeof(struct ip6_hdr)); 1490 if (mb_copy == NULL) { 1491 splx(s); 1492 return; 1493 } 1494 /* set MCAST flag to the outgoing packet */ 1495 mb_copy->m_flags |= M_MCAST; 1496 1497 /* 1498 * If we sourced the packet, call ip6_output since we may devide 1499 * the packet into fragments when the packet is too big for the 1500 * outgoing interface. 1501 * Otherwise, we can simply send the packet to the interface 1502 * sending queue. 1503 */ 1504 if (m->m_pkthdr.rcvif == NULL) { 1505 struct ip6_moptions im6o; 1506 1507 im6o.im6o_multicast_ifp = ifp; 1508 /* XXX: ip6_output will override ip6->ip6_hlim */ 1509 im6o.im6o_multicast_hlim = ip6->ip6_hlim; 1510 im6o.im6o_multicast_loop = 1; 1511 error = ip6_output(mb_copy, NULL, &ro, 1512 IPV6_FORWARDING, &im6o, NULL, NULL); 1513 1514#ifdef MRT6DEBUG 1515 if (mrt6debug & DEBUG_XMIT) 1516 log(LOG_DEBUG, "phyint_send on mif %d err %d\n", 1517 mifp - mif6table, error); 1518#endif 1519 splx(s); 1520 return; 1521 } 1522 1523 /* 1524 * If we belong to the destination multicast group 1525 * on the outgoing interface, loop back a copy. 1526 */ 1527 dst6 = (struct sockaddr_in6 *)&ro.ro_dst; 1528 IN6_LOOKUP_MULTI(ip6->ip6_dst, ifp, in6m); 1529 if (in6m != NULL) { 1530 dst6->sin6_len = sizeof(struct sockaddr_in6); 1531 dst6->sin6_family = AF_INET6; 1532 dst6->sin6_addr = ip6->ip6_dst; 1533 ip6_mloopback(ifp, m, (struct sockaddr_in6 *)&ro.ro_dst); 1534 } 1535 /* 1536 * Put the packet into the sending queue of the outgoing interface 1537 * if it would fit in the MTU of the interface. 1538 */ 1539 linkmtu = IN6_LINKMTU(ifp); 1540 if (mb_copy->m_pkthdr.len <= linkmtu || linkmtu < IPV6_MMTU) { 1541 dst6->sin6_len = sizeof(struct sockaddr_in6); 1542 dst6->sin6_family = AF_INET6; 1543 dst6->sin6_addr = ip6->ip6_dst; 1544 /* 1545 * We just call if_output instead of nd6_output here, since 1546 * we need no ND for a multicast forwarded packet...right? 1547 */ 1548 error = (*ifp->if_output)(ifp, mb_copy, 1549 (struct sockaddr *)&ro.ro_dst, NULL); 1550#ifdef MRT6DEBUG 1551 if (mrt6debug & DEBUG_XMIT) 1552 log(LOG_DEBUG, "phyint_send on mif %d err %d\n", 1553 mifp - mif6table, error); 1554#endif 1555 } else { 1556#ifdef MULTICAST_PMTUD 1557 icmp6_error(mb_copy, ICMP6_PACKET_TOO_BIG, 0, linkmtu); 1558#else 1559#ifdef MRT6DEBUG 1560 if (mrt6debug & DEBUG_XMIT) 1561 log(LOG_DEBUG, 1562 "phyint_send: packet too big on %s o %s g %s" 1563 " size %d(discarded)\n", 1564 if_name(ifp), 1565 ip6_sprintf(&ip6->ip6_src), 1566 ip6_sprintf(&ip6->ip6_dst), 1567 mb_copy->m_pkthdr.len); 1568#endif /* MRT6DEBUG */ 1569 m_freem(mb_copy); /* simply discard the packet */ 1570#endif 1571 } 1572 1573 splx(s); 1574} 1575 1576static int 1577register_send(ip6, mif, m) 1578 struct ip6_hdr *ip6; 1579 struct mif6 *mif; 1580 struct mbuf *m; 1581{ 1582 struct mbuf *mm; 1583 int i, len = m->m_pkthdr.len; 1584 static struct sockaddr_in6 sin6 = { sizeof(sin6), AF_INET6 }; 1585 struct mrt6msg *im6; 1586 1587#ifdef MRT6DEBUG 1588 if (mrt6debug) 1589 log(LOG_DEBUG, "** IPv6 register_send **\n src %s dst %s\n", 1590 ip6_sprintf(&ip6->ip6_src), ip6_sprintf(&ip6->ip6_dst)); 1591#endif 1592 ++pim6stat.pim6s_snd_registers; 1593 1594 /* Make a copy of the packet to send to the user level process */ 1595 MGETHDR(mm, M_DONTWAIT, MT_HEADER); 1596 if (mm == NULL) 1597 return (ENOBUFS); 1598 mm->m_pkthdr.rcvif = NULL; 1599 mm->m_data += max_linkhdr; 1600 mm->m_len = sizeof(struct ip6_hdr); 1601 1602 if ((mm->m_next = m_copy(m, 0, M_COPYALL)) == NULL) { 1603 m_freem(mm); 1604 return (ENOBUFS); 1605 } 1606 i = MHLEN - M_LEADINGSPACE(mm); 1607 if (i > len) 1608 i = len; 1609 mm = m_pullup(mm, i); 1610 if (mm == NULL) 1611 return (ENOBUFS); 1612/* TODO: check it! */ 1613 mm->m_pkthdr.len = len + sizeof(struct ip6_hdr); 1614 1615 /* 1616 * Send message to routing daemon 1617 */ 1618 sin6.sin6_addr = ip6->ip6_src; 1619 1620 im6 = mtod(mm, struct mrt6msg *); 1621 im6->im6_msgtype = MRT6MSG_WHOLEPKT; 1622 im6->im6_mbz = 0; 1623 1624 im6->im6_mif = mif - mif6table; 1625 1626 /* iif info is not given for reg. encap.n */ 1627 mrt6stat.mrt6s_upcalls++; 1628 1629 if (socket_send(ip6_mrouter, mm, &sin6) < 0) { 1630#ifdef MRT6DEBUG 1631 if (mrt6debug) 1632 log(LOG_WARNING, 1633 "register_send: ip6_mrouter socket queue full\n"); 1634#endif 1635 ++mrt6stat.mrt6s_upq_sockfull; 1636 return (ENOBUFS); 1637 } 1638 return (0); 1639} 1640 1641/* 1642 * PIM sparse mode hook 1643 * Receives the pim control messages, and passes them up to the listening 1644 * socket, using rip6_input. 1645 * The only message processed is the REGISTER pim message; the pim header 1646 * is stripped off, and the inner packet is passed to register_mforward. 1647 */ 1648int 1649pim6_input(mp, offp, proto) 1650 struct mbuf **mp; 1651 int *offp, proto; 1652{ 1653 struct pim *pim; /* pointer to a pim struct */ 1654 struct ip6_hdr *ip6; 1655 int pimlen; 1656 struct mbuf *m = *mp; 1657 int minlen; 1658 int off = *offp; 1659 1660 ++pim6stat.pim6s_rcv_total; 1661 1662 ip6 = mtod(m, struct ip6_hdr *); 1663 pimlen = m->m_pkthdr.len - *offp; 1664 1665 /* 1666 * Validate lengths 1667 */ 1668 if (pimlen < PIM_MINLEN) { 1669 ++pim6stat.pim6s_rcv_tooshort; 1670#ifdef MRT6DEBUG 1671 if (mrt6debug & DEBUG_PIM) 1672 log(LOG_DEBUG,"pim6_input: PIM packet too short\n"); 1673#endif 1674 m_freem(m); 1675 return (IPPROTO_DONE); 1676 } 1677 1678 /* 1679 * if the packet is at least as big as a REGISTER, go ahead 1680 * and grab the PIM REGISTER header size, to avoid another 1681 * possible m_pullup() later. 1682 * 1683 * PIM_MINLEN == pimhdr + u_int32 == 8 1684 * PIM6_REG_MINLEN == pimhdr + reghdr + eip6hdr == 4 + 4 + 40 1685 */ 1686 minlen = (pimlen >= PIM6_REG_MINLEN) ? PIM6_REG_MINLEN : PIM_MINLEN; 1687 1688 /* 1689 * Make sure that the IP6 and PIM headers in contiguous memory, and 1690 * possibly the PIM REGISTER header 1691 */ 1692#ifndef PULLDOWN_TEST 1693 IP6_EXTHDR_CHECK(m, off, minlen, IPPROTO_DONE); 1694 /* adjust pointer */ 1695 ip6 = mtod(m, struct ip6_hdr *); 1696 1697 /* adjust mbuf to point to the PIM header */ 1698 pim = (struct pim *)((caddr_t)ip6 + off); 1699#else 1700 IP6_EXTHDR_GET(pim, struct pim *, m, off, minlen); 1701 if (pim == NULL) { 1702 pim6stat.pim6s_rcv_tooshort++; 1703 return (IPPROTO_DONE); 1704 } 1705#endif 1706 1707#define PIM6_CHECKSUM 1708#ifdef PIM6_CHECKSUM 1709 { 1710 int cksumlen; 1711 1712 /* 1713 * Validate checksum. 1714 * If PIM REGISTER, exclude the data packet 1715 */ 1716 if (pim->pim_type == PIM_REGISTER) 1717 cksumlen = PIM_MINLEN; 1718 else 1719 cksumlen = pimlen; 1720 1721 if (in6_cksum(m, IPPROTO_PIM, off, cksumlen)) { 1722 ++pim6stat.pim6s_rcv_badsum; 1723#ifdef MRT6DEBUG 1724 if (mrt6debug & DEBUG_PIM) 1725 log(LOG_DEBUG, 1726 "pim6_input: invalid checksum\n"); 1727#endif 1728 m_freem(m); 1729 return (IPPROTO_DONE); 1730 } 1731 } 1732#endif /* PIM_CHECKSUM */ 1733 1734 /* PIM version check */ 1735 if (pim->pim_ver != PIM_VERSION) { 1736 ++pim6stat.pim6s_rcv_badversion; 1737#ifdef MRT6DEBUG 1738 log(LOG_ERR, 1739 "pim6_input: incorrect version %d, expecting %d\n", 1740 pim->pim_ver, PIM_VERSION); 1741#endif 1742 m_freem(m); 1743 return (IPPROTO_DONE); 1744 } 1745 1746 if (pim->pim_type == PIM_REGISTER) { 1747 /* 1748 * since this is a REGISTER, we'll make a copy of the register 1749 * headers ip6+pim+u_int32_t+encap_ip6, to be passed up to the 1750 * routing daemon. 1751 */ 1752 static struct sockaddr_in6 dst = { sizeof(dst), AF_INET6 }; 1753 1754 struct mbuf *mcp; 1755 struct ip6_hdr *eip6; 1756 u_int32_t *reghdr; 1757 int rc; 1758 1759 ++pim6stat.pim6s_rcv_registers; 1760 1761 if ((reg_mif_num >= nummifs) || (reg_mif_num == (mifi_t) -1)) { 1762#ifdef MRT6DEBUG 1763 if (mrt6debug & DEBUG_PIM) 1764 log(LOG_DEBUG, 1765 "pim6_input: register mif not set: %d\n", 1766 reg_mif_num); 1767#endif 1768 m_freem(m); 1769 return (IPPROTO_DONE); 1770 } 1771 1772 reghdr = (u_int32_t *)(pim + 1); 1773 1774 if ((ntohl(*reghdr) & PIM_NULL_REGISTER)) 1775 goto pim6_input_to_daemon; 1776 1777 /* 1778 * Validate length 1779 */ 1780 if (pimlen < PIM6_REG_MINLEN) { 1781 ++pim6stat.pim6s_rcv_tooshort; 1782 ++pim6stat.pim6s_rcv_badregisters; 1783#ifdef MRT6DEBUG 1784 log(LOG_ERR, 1785 "pim6_input: register packet size too " 1786 "small %d from %s\n", 1787 pimlen, ip6_sprintf(&ip6->ip6_src)); 1788#endif 1789 m_freem(m); 1790 return (IPPROTO_DONE); 1791 } 1792 1793 eip6 = (struct ip6_hdr *) (reghdr + 1); 1794#ifdef MRT6DEBUG 1795 if (mrt6debug & DEBUG_PIM) 1796 log(LOG_DEBUG, 1797 "pim6_input[register], eip6: %s -> %s, " 1798 "eip6 plen %d\n", 1799 ip6_sprintf(&eip6->ip6_src), 1800 ip6_sprintf(&eip6->ip6_dst), 1801 ntohs(eip6->ip6_plen)); 1802#endif 1803 1804 /* verify the version number of the inner packet */ 1805 if ((eip6->ip6_vfc & IPV6_VERSION_MASK) != IPV6_VERSION) { 1806 ++pim6stat.pim6s_rcv_badregisters; 1807#ifdef MRT6DEBUG 1808 log(LOG_DEBUG, "pim6_input: invalid IP version (%d) " 1809 "of the inner packet\n", 1810 (eip6->ip6_vfc & IPV6_VERSION)); 1811#endif 1812 m_freem(m); 1813 return (IPPROTO_NONE); 1814 } 1815 1816 /* verify the inner packet is destined to a mcast group */ 1817 if (!IN6_IS_ADDR_MULTICAST(&eip6->ip6_dst)) { 1818 ++pim6stat.pim6s_rcv_badregisters; 1819#ifdef MRT6DEBUG 1820 if (mrt6debug & DEBUG_PIM) 1821 log(LOG_DEBUG, 1822 "pim6_input: inner packet of register " 1823 "is not multicast %s\n", 1824 ip6_sprintf(&eip6->ip6_dst)); 1825#endif 1826 m_freem(m); 1827 return (IPPROTO_DONE); 1828 } 1829 1830 /* 1831 * make a copy of the whole header to pass to the daemon later. 1832 */ 1833 mcp = m_copy(m, 0, off + PIM6_REG_MINLEN); 1834 if (mcp == NULL) { 1835#ifdef MRT6DEBUG 1836 log(LOG_ERR, 1837 "pim6_input: pim register: " 1838 "could not copy register head\n"); 1839#endif 1840 m_freem(m); 1841 return (IPPROTO_DONE); 1842 } 1843 1844 /* 1845 * forward the inner ip6 packet; point m_data at the inner ip6. 1846 */ 1847 m_adj(m, off + PIM_MINLEN); 1848#ifdef MRT6DEBUG 1849 if (mrt6debug & DEBUG_PIM) { 1850 log(LOG_DEBUG, 1851 "pim6_input: forwarding decapsulated register: " 1852 "src %s, dst %s, mif %d\n", 1853 ip6_sprintf(&eip6->ip6_src), 1854 ip6_sprintf(&eip6->ip6_dst), 1855 reg_mif_num); 1856 } 1857#endif 1858 1859 rc = if_simloop(mif6table[reg_mif_num].m6_ifp, m, 1860 dst.sin6_family, NULL); 1861 1862 /* prepare the register head to send to the mrouting daemon */ 1863 m = mcp; 1864 } 1865 1866 /* 1867 * Pass the PIM message up to the daemon; if it is a register message 1868 * pass the 'head' only up to the daemon. This includes the 1869 * encapsulator ip6 header, pim header, register header and the 1870 * encapsulated ip6 header. 1871 */ 1872 pim6_input_to_daemon: 1873 rip6_input(&m, offp, proto); 1874 return (IPPROTO_DONE); 1875} 1876