ip6_mroute.c revision 148914
1/* $FreeBSD: head/sys/netinet6/ip6_mroute.c 148914 2005-08-10 05:28:11Z suz $ */ 2/* $KAME: ip6_mroute.c,v 1.58 2001/12/18 02:36:31 itojun Exp $ */ 3 4/*- 5 * Copyright (C) 1998 WIDE Project. 6 * All rights reserved. 7 * 8 * Redistribution and use in source and binary forms, with or without 9 * modification, are permitted provided that the following conditions 10 * are met: 11 * 1. Redistributions of source code must retain the above copyright 12 * notice, this list of conditions and the following disclaimer. 13 * 2. Redistributions in binary form must reproduce the above copyright 14 * notice, this list of conditions and the following disclaimer in the 15 * documentation and/or other materials provided with the distribution. 16 * 3. Neither the name of the project nor the names of its contributors 17 * may be used to endorse or promote products derived from this software 18 * without specific prior written permission. 19 * 20 * THIS SOFTWARE IS PROVIDED BY THE PROJECT AND CONTRIBUTORS ``AS IS'' AND 21 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE 22 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE 23 * ARE DISCLAIMED. IN NO EVENT SHALL THE PROJECT OR CONTRIBUTORS BE LIABLE 24 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL 25 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS 26 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) 27 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT 28 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY 29 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF 30 * SUCH DAMAGE. 31 */ 32 33/* BSDI ip_mroute.c,v 2.10 1996/11/14 00:29:52 jch Exp */ 34 35/*- 36 * Copyright (c) 1989 Stephen Deering 37 * Copyright (c) 1992, 1993 38 * The Regents of the University of California. All rights reserved. 39 * 40 * This code is derived from software contributed to Berkeley by 41 * Stephen Deering of Stanford University. 42 * 43 * Redistribution and use in source and binary forms, with or without 44 * modification, are permitted provided that the following conditions 45 * are met: 46 * 1. Redistributions of source code must retain the above copyright 47 * notice, this list of conditions and the following disclaimer. 48 * 2. Redistributions in binary form must reproduce the above copyright 49 * notice, this list of conditions and the following disclaimer in the 50 * documentation and/or other materials provided with the distribution. 51 * 4. Neither the name of the University nor the names of its contributors 52 * may be used to endorse or promote products derived from this software 53 * without specific prior written permission. 54 * 55 * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND 56 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE 57 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE 58 * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE 59 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL 60 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS 61 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) 62 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT 63 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY 64 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF 65 * SUCH DAMAGE. 66 * 67 * @(#)ip_mroute.c 8.2 (Berkeley) 11/15/93 68 */ 69 70/* 71 * IP multicast forwarding procedures 72 * 73 * Written by David Waitzman, BBN Labs, August 1988. 74 * Modified by Steve Deering, Stanford, February 1989. 75 * Modified by Mark J. Steiglitz, Stanford, May, 1991 76 * Modified by Van Jacobson, LBL, January 1993 77 * Modified by Ajit Thyagarajan, PARC, August 1993 78 * Modified by Bill Fenner, PARC, April 1994 79 * 80 * MROUTING Revision: 3.5.1.2 + PIM-SMv2 (pimd) Support 81 */ 82 83#include "opt_inet.h" 84#include "opt_inet6.h" 85 86#include <sys/param.h> 87#include <sys/callout.h> 88#include <sys/errno.h> 89#include <sys/kernel.h> 90#include <sys/lock.h> 91#include <sys/malloc.h> 92#include <sys/mbuf.h> 93#include <sys/protosw.h> 94#include <sys/signalvar.h> 95#include <sys/socket.h> 96#include <sys/socketvar.h> 97#include <sys/sockio.h> 98#include <sys/sx.h> 99#include <sys/syslog.h> 100#include <sys/systm.h> 101#include <sys/time.h> 102 103#include <net/if.h> 104#include <net/raw_cb.h> 105#include <net/route.h> 106 107#include <netinet/in.h> 108#include <netinet/in_var.h> 109 110#include <netinet/ip6.h> 111#include <netinet6/ip6_var.h> 112#include <netinet6/scope6_var.h> 113#include <netinet6/nd6.h> 114#include <netinet6/ip6_mroute.h> 115#include <netinet6/pim6.h> 116#include <netinet6/pim6_var.h> 117 118#include <net/net_osdep.h> 119 120static MALLOC_DEFINE(M_MRTABLE, "mf6c", "multicast forwarding cache entry"); 121 122#define M_HASCL(m) ((m)->m_flags & M_EXT) 123 124static int ip6_mdq __P((struct mbuf *, struct ifnet *, struct mf6c *)); 125static void phyint_send __P((struct ip6_hdr *, struct mif6 *, struct mbuf *)); 126 127static int set_pim6 __P((int *)); 128static int socket_send __P((struct socket *, struct mbuf *, 129 struct sockaddr_in6 *)); 130static int register_send __P((struct ip6_hdr *, struct mif6 *, 131 struct mbuf *)); 132 133/* 134 * Globals. All but ip6_mrouter, ip6_mrtproto and mrt6stat could be static, 135 * except for netstat or debugging purposes. 136 */ 137struct socket *ip6_mrouter = NULL; 138int ip6_mrouter_ver = 0; 139int ip6_mrtproto = IPPROTO_PIM; /* for netstat only */ 140struct mrt6stat mrt6stat; 141 142#define NO_RTE_FOUND 0x1 143#define RTE_FOUND 0x2 144 145struct mf6c *mf6ctable[MF6CTBLSIZ]; 146u_char n6expire[MF6CTBLSIZ]; 147static struct mif6 mif6table[MAXMIFS]; 148#ifdef MRT6DEBUG 149u_int mrt6debug = 0; /* debug level */ 150#define DEBUG_MFC 0x02 151#define DEBUG_FORWARD 0x04 152#define DEBUG_EXPIRE 0x08 153#define DEBUG_XMIT 0x10 154#define DEBUG_REG 0x20 155#define DEBUG_PIM 0x40 156#endif 157 158static void expire_upcalls __P((void *)); 159#define EXPIRE_TIMEOUT (hz / 4) /* 4x / second */ 160#define UPCALL_EXPIRE 6 /* number of timeouts */ 161 162#ifdef INET 163#ifdef MROUTING 164extern struct socket *ip_mrouter; 165#endif 166#endif 167 168/* 169 * 'Interfaces' associated with decapsulator (so we can tell 170 * packets that went through it from ones that get reflected 171 * by a broken gateway). These interfaces are never linked into 172 * the system ifnet list & no routes point to them. I.e., packets 173 * can't be sent this way. They only exist as a placeholder for 174 * multicast source verification. 175 */ 176struct ifnet multicast_register_if; 177 178#define ENCAP_HOPS 64 179 180/* 181 * Private variables. 182 */ 183static mifi_t nummifs = 0; 184static mifi_t reg_mif_num = (mifi_t)-1; 185 186static struct pim6stat pim6stat; 187static int pim6; 188 189/* 190 * Hash function for a source, group entry 191 */ 192#define MF6CHASH(a, g) MF6CHASHMOD((a).s6_addr32[0] ^ (a).s6_addr32[1] ^ \ 193 (a).s6_addr32[2] ^ (a).s6_addr32[3] ^ \ 194 (g).s6_addr32[0] ^ (g).s6_addr32[1] ^ \ 195 (g).s6_addr32[2] ^ (g).s6_addr32[3]) 196 197/* 198 * Find a route for a given origin IPv6 address and Multicast group address. 199 * Quality of service parameter to be added in the future!!! 200 */ 201 202#define MF6CFIND(o, g, rt) do { \ 203 struct mf6c *_rt = mf6ctable[MF6CHASH(o,g)]; \ 204 rt = NULL; \ 205 mrt6stat.mrt6s_mfc_lookups++; \ 206 while (_rt) { \ 207 if (IN6_ARE_ADDR_EQUAL(&_rt->mf6c_origin.sin6_addr, &(o)) && \ 208 IN6_ARE_ADDR_EQUAL(&_rt->mf6c_mcastgrp.sin6_addr, &(g)) && \ 209 (_rt->mf6c_stall == NULL)) { \ 210 rt = _rt; \ 211 break; \ 212 } \ 213 _rt = _rt->mf6c_next; \ 214 } \ 215 if (rt == NULL) { \ 216 mrt6stat.mrt6s_mfc_misses++; \ 217 } \ 218} while (/*CONSTCOND*/ 0) 219 220/* 221 * Macros to compute elapsed time efficiently 222 * Borrowed from Van Jacobson's scheduling code 223 */ 224#define TV_DELTA(a, b, delta) do { \ 225 int xxs; \ 226 \ 227 delta = (a).tv_usec - (b).tv_usec; \ 228 if ((xxs = (a).tv_sec - (b).tv_sec)) { \ 229 switch (xxs) { \ 230 case 2: \ 231 delta += 1000000; \ 232 /* FALLTHROUGH */ \ 233 case 1: \ 234 delta += 1000000; \ 235 break; \ 236 default: \ 237 delta += (1000000 * xxs); \ 238 } \ 239 } \ 240} while (/*CONSTCOND*/ 0) 241 242#define TV_LT(a, b) (((a).tv_usec < (b).tv_usec && \ 243 (a).tv_sec <= (b).tv_sec) || (a).tv_sec < (b).tv_sec) 244 245#ifdef UPCALL_TIMING 246#define UPCALL_MAX 50 247u_long upcall_data[UPCALL_MAX + 1]; 248static void collate(); 249#endif /* UPCALL_TIMING */ 250 251static int get_sg_cnt __P((struct sioc_sg_req6 *)); 252static int get_mif6_cnt __P((struct sioc_mif_req6 *)); 253static int ip6_mrouter_init __P((struct socket *, int, int)); 254static int add_m6if __P((struct mif6ctl *)); 255static int del_m6if __P((mifi_t *)); 256static int add_m6fc __P((struct mf6cctl *)); 257static int del_m6fc __P((struct mf6cctl *)); 258 259static struct callout expire_upcalls_ch; 260 261/* 262 * Handle MRT setsockopt commands to modify the multicast routing tables. 263 */ 264int 265ip6_mrouter_set(so, sopt) 266 struct socket *so; 267 struct sockopt *sopt; 268{ 269 int error = 0; 270 int optval; 271 struct mif6ctl mifc; 272 struct mf6cctl mfcc; 273 mifi_t mifi; 274 275 if (so != ip6_mrouter && sopt->sopt_name != MRT6_INIT) 276 return (EACCES); 277 278 switch (sopt->sopt_name) { 279 case MRT6_INIT: 280#ifdef MRT6_OINIT 281 case MRT6_OINIT: 282#endif 283 error = sooptcopyin(sopt, &optval, sizeof(optval), 284 sizeof(optval)); 285 if (error) 286 break; 287 error = ip6_mrouter_init(so, optval, sopt->sopt_name); 288 break; 289 case MRT6_DONE: 290 error = ip6_mrouter_done(); 291 break; 292 case MRT6_ADD_MIF: 293 error = sooptcopyin(sopt, &mifc, sizeof(mifc), sizeof(mifc)); 294 if (error) 295 break; 296 error = add_m6if(&mifc); 297 break; 298 case MRT6_ADD_MFC: 299 error = sooptcopyin(sopt, &mfcc, sizeof(mfcc), sizeof(mfcc)); 300 if (error) 301 break; 302 error = add_m6fc(&mfcc); 303 break; 304 case MRT6_DEL_MFC: 305 error = sooptcopyin(sopt, &mfcc, sizeof(mfcc), sizeof(mfcc)); 306 if (error) 307 break; 308 error = del_m6fc(&mfcc); 309 break; 310 case MRT6_DEL_MIF: 311 error = sooptcopyin(sopt, &mifi, sizeof(mifi), sizeof(mifi)); 312 if (error) 313 break; 314 error = del_m6if(&mifi); 315 break; 316 case MRT6_PIM: 317 error = sooptcopyin(sopt, &optval, sizeof(optval), 318 sizeof(optval)); 319 if (error) 320 break; 321 error = set_pim6(&optval); 322 break; 323 default: 324 error = EOPNOTSUPP; 325 break; 326 } 327 328 return (error); 329} 330 331/* 332 * Handle MRT getsockopt commands 333 */ 334int 335ip6_mrouter_get(so, sopt) 336 struct socket *so; 337 struct sockopt *sopt; 338{ 339 int error = 0; 340 341 if (so != ip6_mrouter) 342 return (EACCES); 343 344 switch (sopt->sopt_name) { 345 case MRT6_PIM: 346 error = sooptcopyout(sopt, &pim6, sizeof(pim6)); 347 break; 348 } 349 return (error); 350} 351 352/* 353 * Handle ioctl commands to obtain information from the cache 354 */ 355int 356mrt6_ioctl(cmd, data) 357 int cmd; 358 caddr_t data; 359{ 360 switch (cmd) { 361 case SIOCGETSGCNT_IN6: 362 return (get_sg_cnt((struct sioc_sg_req6 *)data)); 363 case SIOCGETMIFCNT_IN6: 364 return (get_mif6_cnt((struct sioc_mif_req6 *)data)); 365 default: 366 return (EINVAL); 367 } 368} 369 370/* 371 * returns the packet, byte, rpf-failure count for the source group provided 372 */ 373static int 374get_sg_cnt(req) 375 struct sioc_sg_req6 *req; 376{ 377 struct mf6c *rt; 378 int s; 379 380 s = splnet(); 381 MF6CFIND(req->src.sin6_addr, req->grp.sin6_addr, rt); 382 splx(s); 383 if (rt != NULL) { 384 req->pktcnt = rt->mf6c_pkt_cnt; 385 req->bytecnt = rt->mf6c_byte_cnt; 386 req->wrong_if = rt->mf6c_wrong_if; 387 } else 388 return (ESRCH); 389#if 0 390 req->pktcnt = req->bytecnt = req->wrong_if = 0xffffffff; 391#endif 392 393 return (0); 394} 395 396/* 397 * returns the input and output packet and byte counts on the mif provided 398 */ 399static int 400get_mif6_cnt(req) 401 struct sioc_mif_req6 *req; 402{ 403 mifi_t mifi = req->mifi; 404 405 if (mifi >= nummifs) 406 return (EINVAL); 407 408 req->icount = mif6table[mifi].m6_pkt_in; 409 req->ocount = mif6table[mifi].m6_pkt_out; 410 req->ibytes = mif6table[mifi].m6_bytes_in; 411 req->obytes = mif6table[mifi].m6_bytes_out; 412 413 return (0); 414} 415 416static int 417set_pim6(i) 418 int *i; 419{ 420 if ((*i != 1) && (*i != 0)) 421 return (EINVAL); 422 423 pim6 = *i; 424 425 return (0); 426} 427 428/* 429 * Enable multicast routing 430 */ 431static int 432ip6_mrouter_init(so, v, cmd) 433 struct socket *so; 434 int v; 435 int cmd; 436{ 437#ifdef MRT6DEBUG 438 if (mrt6debug) 439 log(LOG_DEBUG, 440 "ip6_mrouter_init: so_type = %d, pr_protocol = %d\n", 441 so->so_type, so->so_proto->pr_protocol); 442#endif 443 444 if (so->so_type != SOCK_RAW || 445 so->so_proto->pr_protocol != IPPROTO_ICMPV6) 446 return (EOPNOTSUPP); 447 448 if (v != 1) 449 return (ENOPROTOOPT); 450 451 if (ip6_mrouter != NULL) 452 return (EADDRINUSE); 453 454 ip6_mrouter = so; 455 ip6_mrouter_ver = cmd; 456 457 bzero((caddr_t)mf6ctable, sizeof(mf6ctable)); 458 bzero((caddr_t)n6expire, sizeof(n6expire)); 459 460 pim6 = 0;/* used for stubbing out/in pim stuff */ 461 462 callout_init(&expire_upcalls_ch, 0); 463 callout_reset(&expire_upcalls_ch, EXPIRE_TIMEOUT, 464 expire_upcalls, NULL); 465 466#ifdef MRT6DEBUG 467 if (mrt6debug) 468 log(LOG_DEBUG, "ip6_mrouter_init\n"); 469#endif 470 471 return (0); 472} 473 474/* 475 * Disable multicast routing 476 */ 477int 478ip6_mrouter_done() 479{ 480 mifi_t mifi; 481 int i; 482 struct ifnet *ifp; 483 struct in6_ifreq ifr; 484 struct mf6c *rt; 485 struct rtdetq *rte; 486 int s; 487 488 s = splnet(); 489 490 /* 491 * For each phyint in use, disable promiscuous reception of all IPv6 492 * multicasts. 493 */ 494#ifdef INET 495#ifdef MROUTING 496 /* 497 * If there is still IPv4 multicast routing daemon, 498 * we remain interfaces to receive all muliticasted packets. 499 * XXX: there may be an interface in which the IPv4 multicast 500 * daemon is not interested... 501 */ 502 if (!ip_mrouter) 503#endif 504#endif 505 { 506 for (mifi = 0; mifi < nummifs; mifi++) { 507 if (mif6table[mifi].m6_ifp && 508 !(mif6table[mifi].m6_flags & MIFF_REGISTER)) { 509 ifr.ifr_addr.sin6_family = AF_INET6; 510 ifr.ifr_addr.sin6_addr = in6addr_any; 511 ifp = mif6table[mifi].m6_ifp; 512 (*ifp->if_ioctl)(ifp, SIOCDELMULTI, 513 (caddr_t)&ifr); 514 } 515 } 516 } 517#ifdef notyet 518 bzero((caddr_t)qtable, sizeof(qtable)); 519 bzero((caddr_t)tbftable, sizeof(tbftable)); 520#endif 521 bzero((caddr_t)mif6table, sizeof(mif6table)); 522 nummifs = 0; 523 524 pim6 = 0; /* used to stub out/in pim specific code */ 525 526 callout_stop(&expire_upcalls_ch); 527 528 /* 529 * Free all multicast forwarding cache entries. 530 */ 531 for (i = 0; i < MF6CTBLSIZ; i++) { 532 rt = mf6ctable[i]; 533 while (rt) { 534 struct mf6c *frt; 535 536 for (rte = rt->mf6c_stall; rte != NULL; ) { 537 struct rtdetq *n = rte->next; 538 539 m_free(rte->m); 540 free(rte, M_MRTABLE); 541 rte = n; 542 } 543 frt = rt; 544 rt = rt->mf6c_next; 545 free(frt, M_MRTABLE); 546 } 547 } 548 549 bzero((caddr_t)mf6ctable, sizeof(mf6ctable)); 550 551 /* 552 * Reset de-encapsulation cache 553 */ 554 reg_mif_num = -1; 555 556 ip6_mrouter = NULL; 557 ip6_mrouter_ver = 0; 558 559 splx(s); 560 561#ifdef MRT6DEBUG 562 if (mrt6debug) 563 log(LOG_DEBUG, "ip6_mrouter_done\n"); 564#endif 565 566 return (0); 567} 568 569static struct sockaddr_in6 sin6 = { sizeof(sin6), AF_INET6 }; 570 571/* 572 * Add a mif to the mif table 573 */ 574static int 575add_m6if(mifcp) 576 struct mif6ctl *mifcp; 577{ 578 struct mif6 *mifp; 579 struct ifnet *ifp; 580 int error, s; 581#ifdef notyet 582 struct tbf *m_tbf = tbftable + mifcp->mif6c_mifi; 583#endif 584 585 if (mifcp->mif6c_mifi >= MAXMIFS) 586 return (EINVAL); 587 mifp = mif6table + mifcp->mif6c_mifi; 588 if (mifp->m6_ifp) 589 return (EADDRINUSE); /* XXX: is it appropriate? */ 590 if (mifcp->mif6c_pifi == 0 || mifcp->mif6c_pifi > if_index) 591 return (ENXIO); 592 ifp = ifnet_byindex(mifcp->mif6c_pifi); 593 594 if (mifcp->mif6c_flags & MIFF_REGISTER) { 595 if (reg_mif_num == (mifi_t)-1) { 596 strlcpy(multicast_register_if.if_xname, "register_mif", 597 IFNAMSIZ); 598 multicast_register_if.if_flags |= IFF_LOOPBACK; 599 multicast_register_if.if_index = mifcp->mif6c_mifi; 600 reg_mif_num = mifcp->mif6c_mifi; 601 } 602 603 ifp = &multicast_register_if; 604 605 } /* if REGISTER */ 606 else { 607 /* Make sure the interface supports multicast */ 608 if ((ifp->if_flags & IFF_MULTICAST) == 0) 609 return (EOPNOTSUPP); 610 611 s = splnet(); 612 error = if_allmulti(ifp, 1); 613 splx(s); 614 if (error) 615 return (error); 616 } 617 618 s = splnet(); 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\n", 640 mifcp->mif6c_mifi, 641 ifp->if_xname); 642#endif 643 644 return (0); 645} 646 647/* 648 * Delete a mif from the mif table 649 */ 650static int 651del_m6if(mifip) 652 mifi_t *mifip; 653{ 654 struct mif6 *mifp = mif6table + *mifip; 655 mifi_t mifi; 656 struct ifnet *ifp; 657 int s; 658 659 if (*mifip >= nummifs) 660 return (EINVAL); 661 if (mifp->m6_ifp == NULL) 662 return (EINVAL); 663 664 s = splnet(); 665 666 if (!(mifp->m6_flags & MIFF_REGISTER)) { 667 /* 668 * XXX: what if there is yet IPv4 multicast daemon 669 * using the interface? 670 */ 671 ifp = mifp->m6_ifp; 672 673 if_allmulti(ifp, 0); 674 } 675 676#ifdef notyet 677 bzero((caddr_t)qtable[*mifip], sizeof(qtable[*mifip])); 678 bzero((caddr_t)mifp->m6_tbf, sizeof(*(mifp->m6_tbf))); 679#endif 680 bzero((caddr_t)mifp, sizeof(*mifp)); 681 682 /* Adjust nummifs down */ 683 for (mifi = nummifs; mifi > 0; mifi--) 684 if (mif6table[mifi - 1].m6_ifp) 685 break; 686 nummifs = mifi; 687 688 splx(s); 689 690#ifdef MRT6DEBUG 691 if (mrt6debug) 692 log(LOG_DEBUG, "del_m6if %d, nummifs %d\n", *mifip, nummifs); 693#endif 694 695 return (0); 696} 697 698/* 699 * Add an mfc entry 700 */ 701static int 702add_m6fc(mfccp) 703 struct mf6cctl *mfccp; 704{ 705 struct mf6c *rt; 706 u_long hash; 707 struct rtdetq *rte; 708 u_short nstl; 709 int s; 710 711 MF6CFIND(mfccp->mf6cc_origin.sin6_addr, 712 mfccp->mf6cc_mcastgrp.sin6_addr, rt); 713 714 /* If an entry already exists, just update the fields */ 715 if (rt) { 716#ifdef MRT6DEBUG 717 if (mrt6debug & DEBUG_MFC) 718 log(LOG_DEBUG, 719 "add_m6fc no upcall h %d o %s g %s p %x\n", 720 ip6_sprintf(&mfccp->mf6cc_origin.sin6_addr), 721 ip6_sprintf(&mfccp->mf6cc_mcastgrp.sin6_addr), 722 mfccp->mf6cc_parent); 723#endif 724 725 s = splnet(); 726 rt->mf6c_parent = mfccp->mf6cc_parent; 727 rt->mf6c_ifset = mfccp->mf6cc_ifset; 728 splx(s); 729 return (0); 730 } 731 732 /* 733 * Find the entry for which the upcall was made and update 734 */ 735 s = splnet(); 736 hash = MF6CHASH(mfccp->mf6cc_origin.sin6_addr, 737 mfccp->mf6cc_mcastgrp.sin6_addr); 738 for (rt = mf6ctable[hash], nstl = 0; rt; rt = rt->mf6c_next) { 739 if (IN6_ARE_ADDR_EQUAL(&rt->mf6c_origin.sin6_addr, 740 &mfccp->mf6cc_origin.sin6_addr) && 741 IN6_ARE_ADDR_EQUAL(&rt->mf6c_mcastgrp.sin6_addr, 742 &mfccp->mf6cc_mcastgrp.sin6_addr) && 743 (rt->mf6c_stall != NULL)) { 744 745 if (nstl++) 746 log(LOG_ERR, 747 "add_m6fc: %s o %s g %s p %x dbx %p\n", 748 "multiple kernel entries", 749 ip6_sprintf(&mfccp->mf6cc_origin.sin6_addr), 750 ip6_sprintf(&mfccp->mf6cc_mcastgrp.sin6_addr), 751 mfccp->mf6cc_parent, rt->mf6c_stall); 752 753#ifdef MRT6DEBUG 754 if (mrt6debug & DEBUG_MFC) 755 log(LOG_DEBUG, 756 "add_m6fc o %s g %s p %x dbg %x\n", 757 ip6_sprintf(&mfccp->mf6cc_origin.sin6_addr), 758 ip6_sprintf(&mfccp->mf6cc_mcastgrp.sin6_addr), 759 mfccp->mf6cc_parent, rt->mf6c_stall); 760#endif 761 762 rt->mf6c_origin = mfccp->mf6cc_origin; 763 rt->mf6c_mcastgrp = mfccp->mf6cc_mcastgrp; 764 rt->mf6c_parent = mfccp->mf6cc_parent; 765 rt->mf6c_ifset = mfccp->mf6cc_ifset; 766 /* initialize pkt counters per src-grp */ 767 rt->mf6c_pkt_cnt = 0; 768 rt->mf6c_byte_cnt = 0; 769 rt->mf6c_wrong_if = 0; 770 771 rt->mf6c_expire = 0; /* Don't clean this guy up */ 772 n6expire[hash]--; 773 774 /* free packets Qed at the end of this entry */ 775 for (rte = rt->mf6c_stall; rte != NULL; ) { 776 struct rtdetq *n = rte->next; 777 ip6_mdq(rte->m, rte->ifp, rt); 778 m_freem(rte->m); 779#ifdef UPCALL_TIMING 780 collate(&(rte->t)); 781#endif /* UPCALL_TIMING */ 782 free(rte, M_MRTABLE); 783 rte = n; 784 } 785 rt->mf6c_stall = NULL; 786 } 787 } 788 789 /* 790 * It is possible that an entry is being inserted without an upcall 791 */ 792 if (nstl == 0) { 793#ifdef MRT6DEBUG 794 if (mrt6debug & DEBUG_MFC) 795 log(LOG_DEBUG, 796 "add_mfc no upcall h %d o %s g %s p %x\n", 797 hash, 798 ip6_sprintf(&mfccp->mf6cc_origin.sin6_addr), 799 ip6_sprintf(&mfccp->mf6cc_mcastgrp.sin6_addr), 800 mfccp->mf6cc_parent); 801#endif 802 803 for (rt = mf6ctable[hash]; rt; rt = rt->mf6c_next) { 804 805 if (IN6_ARE_ADDR_EQUAL(&rt->mf6c_origin.sin6_addr, 806 &mfccp->mf6cc_origin.sin6_addr)&& 807 IN6_ARE_ADDR_EQUAL(&rt->mf6c_mcastgrp.sin6_addr, 808 &mfccp->mf6cc_mcastgrp.sin6_addr)) { 809 810 rt->mf6c_origin = mfccp->mf6cc_origin; 811 rt->mf6c_mcastgrp = mfccp->mf6cc_mcastgrp; 812 rt->mf6c_parent = mfccp->mf6cc_parent; 813 rt->mf6c_ifset = mfccp->mf6cc_ifset; 814 /* initialize pkt counters per src-grp */ 815 rt->mf6c_pkt_cnt = 0; 816 rt->mf6c_byte_cnt = 0; 817 rt->mf6c_wrong_if = 0; 818 819 if (rt->mf6c_expire) 820 n6expire[hash]--; 821 rt->mf6c_expire = 0; 822 } 823 } 824 if (rt == NULL) { 825 /* no upcall, so make a new entry */ 826 rt = (struct mf6c *)malloc(sizeof(*rt), M_MRTABLE, 827 M_NOWAIT); 828 if (rt == NULL) { 829 splx(s); 830 return (ENOBUFS); 831 } 832 833 /* insert new entry at head of hash chain */ 834 rt->mf6c_origin = mfccp->mf6cc_origin; 835 rt->mf6c_mcastgrp = mfccp->mf6cc_mcastgrp; 836 rt->mf6c_parent = mfccp->mf6cc_parent; 837 rt->mf6c_ifset = mfccp->mf6cc_ifset; 838 /* initialize pkt counters per src-grp */ 839 rt->mf6c_pkt_cnt = 0; 840 rt->mf6c_byte_cnt = 0; 841 rt->mf6c_wrong_if = 0; 842 rt->mf6c_expire = 0; 843 rt->mf6c_stall = NULL; 844 845 /* link into table */ 846 rt->mf6c_next = mf6ctable[hash]; 847 mf6ctable[hash] = rt; 848 } 849 } 850 splx(s); 851 return (0); 852} 853 854#ifdef UPCALL_TIMING 855/* 856 * collect delay statistics on the upcalls 857 */ 858static void 859collate(t) 860 struct timeval *t; 861{ 862 u_long d; 863 struct timeval tp; 864 u_long delta; 865 866 GET_TIME(tp); 867 868 if (TV_LT(*t, tp)) 869 { 870 TV_DELTA(tp, *t, delta); 871 872 d = delta >> 10; 873 if (d > UPCALL_MAX) 874 d = UPCALL_MAX; 875 876 ++upcall_data[d]; 877 } 878} 879#endif /* UPCALL_TIMING */ 880 881/* 882 * Delete an mfc entry 883 */ 884static int 885del_m6fc(mfccp) 886 struct mf6cctl *mfccp; 887{ 888 struct sockaddr_in6 origin; 889 struct sockaddr_in6 mcastgrp; 890 struct mf6c *rt; 891 struct mf6c **nptr; 892 u_long hash; 893 int s; 894 895 origin = mfccp->mf6cc_origin; 896 mcastgrp = mfccp->mf6cc_mcastgrp; 897 hash = MF6CHASH(origin.sin6_addr, mcastgrp.sin6_addr); 898 899#ifdef MRT6DEBUG 900 if (mrt6debug & DEBUG_MFC) 901 log(LOG_DEBUG,"del_m6fc orig %s mcastgrp %s\n", 902 ip6_sprintf(&origin.sin6_addr), 903 ip6_sprintf(&mcastgrp.sin6_addr)); 904#endif 905 906 s = splnet(); 907 908 nptr = &mf6ctable[hash]; 909 while ((rt = *nptr) != NULL) { 910 if (IN6_ARE_ADDR_EQUAL(&origin.sin6_addr, 911 &rt->mf6c_origin.sin6_addr) && 912 IN6_ARE_ADDR_EQUAL(&mcastgrp.sin6_addr, 913 &rt->mf6c_mcastgrp.sin6_addr) && 914 rt->mf6c_stall == NULL) 915 break; 916 917 nptr = &rt->mf6c_next; 918 } 919 if (rt == NULL) { 920 splx(s); 921 return (EADDRNOTAVAIL); 922 } 923 924 *nptr = rt->mf6c_next; 925 free(rt, M_MRTABLE); 926 927 splx(s); 928 929 return (0); 930} 931 932static int 933socket_send(s, mm, src) 934 struct socket *s; 935 struct mbuf *mm; 936 struct sockaddr_in6 *src; 937{ 938 if (s) { 939 if (sbappendaddr(&s->so_rcv, 940 (struct sockaddr *)src, 941 mm, (struct mbuf *)0) != 0) { 942 sorwakeup(s); 943 return (0); 944 } 945 } 946 m_freem(mm); 947 return (-1); 948} 949 950/* 951 * IPv6 multicast forwarding function. This function assumes that the packet 952 * pointed to by "ip6" has arrived on (or is about to be sent to) the interface 953 * pointed to by "ifp", and the packet is to be relayed to other networks 954 * that have members of the packet's destination IPv6 multicast group. 955 * 956 * The packet is returned unscathed to the caller, unless it is 957 * erroneous, in which case a non-zero return value tells the caller to 958 * discard it. 959 */ 960 961int 962ip6_mforward(ip6, ifp, m) 963 struct ip6_hdr *ip6; 964 struct ifnet *ifp; 965 struct mbuf *m; 966{ 967 struct mf6c *rt; 968 struct mif6 *mifp; 969 struct mbuf *mm; 970 int s; 971 mifi_t mifi; 972 973#ifdef MRT6DEBUG 974 if (mrt6debug & DEBUG_FORWARD) 975 log(LOG_DEBUG, "ip6_mforward: src %s, dst %s, ifindex %d\n", 976 ip6_sprintf(&ip6->ip6_src), ip6_sprintf(&ip6->ip6_dst), 977 ifp->if_index); 978#endif 979 980 /* 981 * Don't forward a packet with Hop limit of zero or one, 982 * or a packet destined to a local-only group. 983 */ 984 if (ip6->ip6_hlim <= 1 || IN6_IS_ADDR_MC_INTFACELOCAL(&ip6->ip6_dst) || 985 IN6_IS_ADDR_MC_LINKLOCAL(&ip6->ip6_dst)) 986 return (0); 987 ip6->ip6_hlim--; 988 989 /* 990 * Source address check: do not forward packets with unspecified 991 * source. It was discussed in July 2000, on ipngwg mailing list. 992 * This is rather more serious than unicast cases, because some 993 * MLD packets can be sent with the unspecified source address 994 * (although such packets must normally set 1 to the hop limit field). 995 */ 996 if (IN6_IS_ADDR_UNSPECIFIED(&ip6->ip6_src)) { 997 ip6stat.ip6s_cantforward++; 998 if (ip6_log_time + ip6_log_interval < time_second) { 999 ip6_log_time = time_second; 1000 log(LOG_DEBUG, 1001 "cannot forward " 1002 "from %s to %s nxt %d received on %s\n", 1003 ip6_sprintf(&ip6->ip6_src), 1004 ip6_sprintf(&ip6->ip6_dst), 1005 ip6->ip6_nxt, 1006 if_name(m->m_pkthdr.rcvif)); 1007 } 1008 return (0); 1009 } 1010 1011 /* 1012 * Determine forwarding mifs from the forwarding cache table 1013 */ 1014 s = splnet(); 1015 MF6CFIND(ip6->ip6_src, ip6->ip6_dst, rt); 1016 1017 /* Entry exists, so forward if necessary */ 1018 if (rt) { 1019 splx(s); 1020 return (ip6_mdq(m, ifp, rt)); 1021 } else { 1022 /* 1023 * If we don't have a route for packet's origin, 1024 * Make a copy of the packet & 1025 * send message to routing daemon 1026 */ 1027 1028 struct mbuf *mb0; 1029 struct rtdetq *rte; 1030 u_long hash; 1031/* int i, npkts;*/ 1032#ifdef UPCALL_TIMING 1033 struct timeval tp; 1034 1035 GET_TIME(tp); 1036#endif /* UPCALL_TIMING */ 1037 1038 mrt6stat.mrt6s_no_route++; 1039#ifdef MRT6DEBUG 1040 if (mrt6debug & (DEBUG_FORWARD | DEBUG_MFC)) 1041 log(LOG_DEBUG, "ip6_mforward: no rte s %s g %s\n", 1042 ip6_sprintf(&ip6->ip6_src), 1043 ip6_sprintf(&ip6->ip6_dst)); 1044#endif 1045 1046 /* 1047 * Allocate mbufs early so that we don't do extra work if we 1048 * are just going to fail anyway. 1049 */ 1050 rte = (struct rtdetq *)malloc(sizeof(*rte), M_MRTABLE, 1051 M_NOWAIT); 1052 if (rte == NULL) { 1053 splx(s); 1054 return (ENOBUFS); 1055 } 1056 mb0 = m_copy(m, 0, M_COPYALL); 1057 /* 1058 * Pullup packet header if needed before storing it, 1059 * as other references may modify it in the meantime. 1060 */ 1061 if (mb0 && 1062 (M_HASCL(mb0) || mb0->m_len < sizeof(struct ip6_hdr))) 1063 mb0 = m_pullup(mb0, sizeof(struct ip6_hdr)); 1064 if (mb0 == NULL) { 1065 free(rte, M_MRTABLE); 1066 splx(s); 1067 return (ENOBUFS); 1068 } 1069 1070 /* is there an upcall waiting for this packet? */ 1071 hash = MF6CHASH(ip6->ip6_src, ip6->ip6_dst); 1072 for (rt = mf6ctable[hash]; rt; rt = rt->mf6c_next) { 1073 if (IN6_ARE_ADDR_EQUAL(&ip6->ip6_src, 1074 &rt->mf6c_origin.sin6_addr) && 1075 IN6_ARE_ADDR_EQUAL(&ip6->ip6_dst, 1076 &rt->mf6c_mcastgrp.sin6_addr) && 1077 (rt->mf6c_stall != NULL)) 1078 break; 1079 } 1080 1081 if (rt == NULL) { 1082 struct mrt6msg *im; 1083#ifdef MRT6_OINIT 1084 struct omrt6msg *oim; 1085#endif 1086 1087 /* no upcall, so make a new entry */ 1088 rt = (struct mf6c *)malloc(sizeof(*rt), M_MRTABLE, 1089 M_NOWAIT); 1090 if (rt == NULL) { 1091 free(rte, M_MRTABLE); 1092 m_freem(mb0); 1093 splx(s); 1094 return (ENOBUFS); 1095 } 1096 /* 1097 * Make a copy of the header to send to the user 1098 * level process 1099 */ 1100 mm = m_copy(mb0, 0, sizeof(struct ip6_hdr)); 1101 1102 if (mm == NULL) { 1103 free(rte, M_MRTABLE); 1104 m_freem(mb0); 1105 free(rt, M_MRTABLE); 1106 splx(s); 1107 return (ENOBUFS); 1108 } 1109 1110 /* 1111 * Send message to routing daemon 1112 */ 1113 sin6.sin6_addr = ip6->ip6_src; 1114 1115 im = NULL; 1116#ifdef MRT6_OINIT 1117 oim = NULL; 1118#endif 1119 switch (ip6_mrouter_ver) { 1120#ifdef MRT6_OINIT 1121 case MRT6_OINIT: 1122 oim = mtod(mm, struct omrt6msg *); 1123 oim->im6_msgtype = MRT6MSG_NOCACHE; 1124 oim->im6_mbz = 0; 1125 break; 1126#endif 1127 case MRT6_INIT: 1128 im = mtod(mm, struct mrt6msg *); 1129 im->im6_msgtype = MRT6MSG_NOCACHE; 1130 im->im6_mbz = 0; 1131 break; 1132 default: 1133 free(rte, M_MRTABLE); 1134 m_freem(mb0); 1135 free(rt, M_MRTABLE); 1136 splx(s); 1137 return (EINVAL); 1138 } 1139 1140#ifdef MRT6DEBUG 1141 if (mrt6debug & DEBUG_FORWARD) 1142 log(LOG_DEBUG, 1143 "getting the iif info in the kernel\n"); 1144#endif 1145 1146 for (mifp = mif6table, mifi = 0; 1147 mifi < nummifs && mifp->m6_ifp != ifp; 1148 mifp++, mifi++) 1149 ; 1150 1151 switch (ip6_mrouter_ver) { 1152#ifdef MRT6_OINIT 1153 case MRT6_OINIT: 1154 oim->im6_mif = mifi; 1155 break; 1156#endif 1157 case MRT6_INIT: 1158 im->im6_mif = mifi; 1159 break; 1160 } 1161 1162 if (socket_send(ip6_mrouter, mm, &sin6) < 0) { 1163 log(LOG_WARNING, "ip6_mforward: ip6_mrouter " 1164 "socket queue full\n"); 1165 mrt6stat.mrt6s_upq_sockfull++; 1166 free(rte, M_MRTABLE); 1167 m_freem(mb0); 1168 free(rt, M_MRTABLE); 1169 splx(s); 1170 return (ENOBUFS); 1171 } 1172 1173 mrt6stat.mrt6s_upcalls++; 1174 1175 /* insert new entry at head of hash chain */ 1176 bzero(rt, sizeof(*rt)); 1177 rt->mf6c_origin.sin6_family = AF_INET6; 1178 rt->mf6c_origin.sin6_len = sizeof(struct sockaddr_in6); 1179 rt->mf6c_origin.sin6_addr = ip6->ip6_src; 1180 rt->mf6c_mcastgrp.sin6_family = AF_INET6; 1181 rt->mf6c_mcastgrp.sin6_len = sizeof(struct sockaddr_in6); 1182 rt->mf6c_mcastgrp.sin6_addr = ip6->ip6_dst; 1183 rt->mf6c_expire = UPCALL_EXPIRE; 1184 n6expire[hash]++; 1185 rt->mf6c_parent = MF6C_INCOMPLETE_PARENT; 1186 1187 /* link into table */ 1188 rt->mf6c_next = mf6ctable[hash]; 1189 mf6ctable[hash] = rt; 1190 /* Add this entry to the end of the queue */ 1191 rt->mf6c_stall = rte; 1192 } else { 1193 /* determine if q has overflowed */ 1194 struct rtdetq **p; 1195 int npkts = 0; 1196 1197 for (p = &rt->mf6c_stall; *p != NULL; p = &(*p)->next) 1198 if (++npkts > MAX_UPQ6) { 1199 mrt6stat.mrt6s_upq_ovflw++; 1200 free(rte, M_MRTABLE); 1201 m_freem(mb0); 1202 splx(s); 1203 return (0); 1204 } 1205 1206 /* Add this entry to the end of the queue */ 1207 *p = rte; 1208 } 1209 1210 rte->next = NULL; 1211 rte->m = mb0; 1212 rte->ifp = ifp; 1213#ifdef UPCALL_TIMING 1214 rte->t = tp; 1215#endif /* UPCALL_TIMING */ 1216 1217 splx(s); 1218 1219 return (0); 1220 } 1221} 1222 1223/* 1224 * Clean up cache entries if upcalls are not serviced 1225 * Call from the Slow Timeout mechanism, every half second. 1226 */ 1227static void 1228expire_upcalls(unused) 1229 void *unused; 1230{ 1231 struct rtdetq *rte; 1232 struct mf6c *mfc, **nptr; 1233 int i; 1234 int s; 1235 1236 s = splnet(); 1237 for (i = 0; i < MF6CTBLSIZ; i++) { 1238 if (n6expire[i] == 0) 1239 continue; 1240 nptr = &mf6ctable[i]; 1241 while ((mfc = *nptr) != NULL) { 1242 rte = mfc->mf6c_stall; 1243 /* 1244 * Skip real cache entries 1245 * Make sure it wasn't marked to not expire (shouldn't happen) 1246 * If it expires now 1247 */ 1248 if (rte != NULL && 1249 mfc->mf6c_expire != 0 && 1250 --mfc->mf6c_expire == 0) { 1251#ifdef MRT6DEBUG 1252 if (mrt6debug & DEBUG_EXPIRE) 1253 log(LOG_DEBUG, "expire_upcalls: expiring (%s %s)\n", 1254 ip6_sprintf(&mfc->mf6c_origin.sin6_addr), 1255 ip6_sprintf(&mfc->mf6c_mcastgrp.sin6_addr)); 1256#endif 1257 /* 1258 * drop all the packets 1259 * free the mbuf with the pkt, if, timing info 1260 */ 1261 do { 1262 struct rtdetq *n = rte->next; 1263 m_freem(rte->m); 1264 free(rte, M_MRTABLE); 1265 rte = n; 1266 } while (rte != NULL); 1267 mrt6stat.mrt6s_cache_cleanups++; 1268 n6expire[i]--; 1269 1270 *nptr = mfc->mf6c_next; 1271 free(mfc, M_MRTABLE); 1272 } else { 1273 nptr = &mfc->mf6c_next; 1274 } 1275 } 1276 } 1277 splx(s); 1278 callout_reset(&expire_upcalls_ch, EXPIRE_TIMEOUT, 1279 expire_upcalls, NULL); 1280} 1281 1282/* 1283 * Packet forwarding routine once entry in the cache is made 1284 */ 1285static int 1286ip6_mdq(m, ifp, rt) 1287 struct mbuf *m; 1288 struct ifnet *ifp; 1289 struct mf6c *rt; 1290{ 1291 struct ip6_hdr *ip6 = mtod(m, struct ip6_hdr *); 1292 mifi_t mifi, iif; 1293 struct mif6 *mifp; 1294 int plen = m->m_pkthdr.len; 1295 struct in6_addr src0, dst0; /* copies for local work */ 1296 u_int32_t iszone, idzone, oszone, odzone; 1297 int error = 0; 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 src0 = ip6->ip6_src; 1430 dst0 = ip6->ip6_dst; 1431 if ((error = in6_setscope(&src0, ifp, &iszone)) != 0 || 1432 (error = in6_setscope(&dst0, ifp, &idzone)) != 0) { 1433 ip6stat.ip6s_badscope++; 1434 return (error); 1435 } 1436 for (mifp = mif6table, mifi = 0; mifi < nummifs; mifp++, mifi++) { 1437 if (IF_ISSET(mifi, &rt->mf6c_ifset)) { 1438 /* 1439 * check if the outgoing packet is going to break 1440 * a scope boundary. 1441 * XXX For packets through PIM register tunnel 1442 * interface, we believe a routing daemon. 1443 */ 1444 if (!(mif6table[rt->mf6c_parent].m6_flags & 1445 MIFF_REGISTER) && 1446 !(mif6table[mifi].m6_flags & MIFF_REGISTER)) { 1447 if (in6_setscope(&src0, mif6table[mifi].m6_ifp, 1448 &oszone) || 1449 in6_setscope(&dst0, mif6table[mifi].m6_ifp, 1450 &odzone) || 1451 iszone != oszone || 1452 idzone != odzone) { 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, 0); 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