ip6_mroute.c revision 194739
193787Sdes/*- 294691Sdes * Copyright (C) 1998 WIDE Project. 393787Sdes * All rights reserved. 493787Sdes * 593787Sdes * Redistribution and use in source and binary forms, with or without 693787Sdes * modification, are permitted provided that the following conditions 793787Sdes * are met: 893787Sdes * 1. Redistributions of source code must retain the above copyright 993787Sdes * notice, this list of conditions and the following disclaimer. 1093787Sdes * 2. Redistributions in binary form must reproduce the above copyright 1193787Sdes * notice, this list of conditions and the following disclaimer in the 1293787Sdes * documentation and/or other materials provided with the distribution. 1393787Sdes * 3. Neither the name of the project nor the names of its contributors 1493787Sdes * may be used to endorse or promote products derived from this software 1593787Sdes * without specific prior written permission. 1693787Sdes * 1793787Sdes * THIS SOFTWARE IS PROVIDED BY THE PROJECT AND CONTRIBUTORS ``AS IS'' AND 1893787Sdes * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE 1993787Sdes * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE 2093787Sdes * ARE DISCLAIMED. IN NO EVENT SHALL THE PROJECT OR CONTRIBUTORS BE LIABLE 2193787Sdes * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL 2293787Sdes * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS 2393787Sdes * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) 2493787Sdes * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT 2593787Sdes * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY 2693787Sdes * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF 2793787Sdes * SUCH DAMAGE. 2893787Sdes * 2993787Sdes * $KAME: ip6_mroute.c,v 1.58 2001/12/18 02:36:31 itojun Exp $ 3093787Sdes */ 3194691Sdes 3294691Sdes/*- 3393787Sdes * Copyright (c) 1989 Stephen Deering 3493787Sdes * Copyright (c) 1992, 1993 3594691Sdes * The Regents of the University of California. All rights reserved. 3694691Sdes * 3794691Sdes * This code is derived from software contributed to Berkeley by 3893787Sdes * Stephen Deering of Stanford University. 3994691Sdes * 4093787Sdes * Redistribution and use in source and binary forms, with or without 4194691Sdes * modification, are permitted provided that the following conditions 4293787Sdes * are met: 4393787Sdes * 1. Redistributions of source code must retain the above copyright 4493787Sdes * notice, this list of conditions and the following disclaimer. 4593787Sdes * 2. Redistributions in binary form must reproduce the above copyright 4693787Sdes * notice, this list of conditions and the following disclaimer in the 4793787Sdes * documentation and/or other materials provided with the distribution. 4893787Sdes * 4. Neither the name of the University nor the names of its contributors 4993787Sdes * may be used to endorse or promote products derived from this software 5093787Sdes * without specific prior written permission. 5193787Sdes * 5293787Sdes * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND 5393787Sdes * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE 5493787Sdes * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE 5593787Sdes * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE 5693787Sdes * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL 5793787Sdes * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS 5893787Sdes * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) 5993787Sdes * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT 6093787Sdes * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY 6193787Sdes * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF 6293787Sdes * SUCH DAMAGE. 6393787Sdes * 6493787Sdes * @(#)ip_mroute.c 8.2 (Berkeley) 11/15/93 6593787Sdes * BSDI ip_mroute.c,v 2.10 1996/11/14 00:29:52 jch Exp 6693787Sdes */ 6793787Sdes 6893787Sdes/* 6993787Sdes * IP multicast forwarding procedures 7093787Sdes * 7193787Sdes * Written by David Waitzman, BBN Labs, August 1988. 7293787Sdes * Modified by Steve Deering, Stanford, February 1989. 7393787Sdes * Modified by Mark J. Steiglitz, Stanford, May, 1991 7493787Sdes * Modified by Van Jacobson, LBL, January 1993 7593787Sdes * Modified by Ajit Thyagarajan, PARC, August 1993 7693787Sdes * Modified by Bill Fenner, PARC, April 1994 7793787Sdes * 7893787Sdes * MROUTING Revision: 3.5.1.2 + PIM-SMv2 (pimd) Support 7993787Sdes */ 8093787Sdes 8193787Sdes#include <sys/cdefs.h> 8293787Sdes__FBSDID("$FreeBSD: head/sys/netinet6/ip6_mroute.c 194739 2009-06-23 17:03:45Z bz $"); 8393787Sdes 8493787Sdes#include "opt_inet.h" 8593787Sdes#include "opt_inet6.h" 8693787Sdes 8793787Sdes#include <sys/param.h> 8893787Sdes#include <sys/callout.h> 8993787Sdes#include <sys/errno.h> 9093787Sdes#include <sys/kernel.h> 9193787Sdes#include <sys/lock.h> 9293787Sdes#include <sys/malloc.h> 9393787Sdes#include <sys/mbuf.h> 9493787Sdes#include <sys/module.h> 9593787Sdes#include <sys/domain.h> 9693787Sdes#include <sys/protosw.h> 9793787Sdes#include <sys/signalvar.h> 9893787Sdes#include <sys/socket.h> 9993787Sdes#include <sys/socketvar.h> 10093787Sdes#include <sys/sockio.h> 10193787Sdes#include <sys/sx.h> 10293787Sdes#include <sys/sysctl.h> 10393787Sdes#include <sys/syslog.h> 10493787Sdes#include <sys/systm.h> 10593787Sdes#include <sys/time.h> 10693787Sdes#include <sys/vimage.h> 10793787Sdes 10893787Sdes#include <net/if.h> 10993787Sdes#include <net/if_types.h> 11093787Sdes#include <net/raw_cb.h> 11193787Sdes#include <net/vnet.h> 11293787Sdes 11393787Sdes#include <netinet/in.h> 11493787Sdes#include <netinet/in_var.h> 11593787Sdes#include <netinet/icmp6.h> 11693787Sdes#include <netinet/vinet.h> 11793787Sdes#include <netinet/ip_encap.h> 11893787Sdes 11993787Sdes#include <netinet/ip6.h> 12093787Sdes#include <netinet6/ip6_var.h> 12193787Sdes#include <netinet6/scope6_var.h> 12293787Sdes#include <netinet6/nd6.h> 12393787Sdes#include <netinet6/ip6_mroute.h> 12493787Sdes#include <netinet6/ip6protosw.h> 12593787Sdes#include <netinet6/pim6.h> 12693787Sdes#include <netinet6/pim6_var.h> 12793787Sdes#include <netinet6/vinet6.h> 12893787Sdes 12993787Sdesstatic MALLOC_DEFINE(M_MRTABLE6, "mf6c", "multicast forwarding cache entry"); 13093787Sdes 13193787Sdes/* XXX: this is a very common idiom; move to <sys/mbuf.h> ? */ 13294691Sdes#define M_HASCL(m) ((m)->m_flags & M_EXT) 13393787Sdes 13493787Sdesstatic int ip6_mdq(struct mbuf *, struct ifnet *, struct mf6c *); 13593787Sdesstatic void phyint_send(struct ip6_hdr *, struct mif6 *, struct mbuf *); 13693787Sdesstatic void pim6_init(void); 13793787Sdesstatic int register_send(struct ip6_hdr *, struct mif6 *, struct mbuf *); 13893787Sdesstatic int set_pim6(int *); 13993787Sdesstatic int socket_send(struct socket *, struct mbuf *, 14093787Sdes struct sockaddr_in6 *); 14193787Sdes 14293787Sdesextern int in6_mcast_loop; 14393787Sdesextern struct domain inet6domain; 14493787Sdes 14593787Sdesstatic const struct encaptab *pim6_encap_cookie; 14693787Sdesstatic const struct ip6protosw in6_pim_protosw = { 14793787Sdes .pr_type = SOCK_RAW, 14894691Sdes .pr_domain = &inet6domain, 14993787Sdes .pr_protocol = IPPROTO_PIM, 15093787Sdes .pr_flags = PR_ATOMIC|PR_ADDR|PR_LASTHDR, 15193787Sdes .pr_input = pim6_input, 15293787Sdes .pr_output = rip6_output, 15393787Sdes .pr_ctloutput = rip6_ctloutput, 15493787Sdes .pr_init = pim6_init, 15593787Sdes .pr_usrreqs = &rip6_usrreqs 15693787Sdes}; 15793787Sdesstatic int pim6_encapcheck(const struct mbuf *, int, int, void *); 15893787Sdes 15993787Sdes#ifdef VIMAGE_GLOBALS 16093787Sdesstatic int ip6_mrouter_ver; 16193787Sdes#endif 16293787Sdes 16393787SdesSYSCTL_DECL(_net_inet6); 16493787SdesSYSCTL_DECL(_net_inet6_ip6); 16593787SdesSYSCTL_NODE(_net_inet6, IPPROTO_PIM, pim, CTLFLAG_RW, 0, "PIM"); 16693787Sdes 16794691Sdesstatic struct mrt6stat mrt6stat; 16893787SdesSYSCTL_STRUCT(_net_inet6_ip6, OID_AUTO, mrt6stat, CTLFLAG_RW, 16993787Sdes &mrt6stat, mrt6stat, 17093787Sdes "Multicast Routing Statistics (struct mrt6stat, netinet6/ip6_mroute.h)"); 17193787Sdes 17293787Sdes#define NO_RTE_FOUND 0x1 17393787Sdes#define RTE_FOUND 0x2 17493787Sdes 17593787Sdesstatic struct mtx mrouter6_mtx; 17693787Sdes#define MROUTER6_LOCK() mtx_lock(&mrouter6_mtx) 17793787Sdes#define MROUTER6_UNLOCK() mtx_unlock(&mrouter6_mtx) 17894691Sdes#define MROUTER6_LOCK_ASSERT() do { \ 17993787Sdes mtx_assert(&mrouter6_mtx, MA_OWNED); \ 18094691Sdes NET_ASSERT_GIANT(); \ 18193787Sdes} while (0) 18293787Sdes#define MROUTER6_LOCK_INIT() \ 18393787Sdes mtx_init(&mrouter6_mtx, "IPv6 multicast forwarding", NULL, MTX_DEF) 18493787Sdes#define MROUTER6_LOCK_DESTROY() mtx_destroy(&mrouter6_mtx) 18593787Sdes 18693787Sdesstatic struct mf6c *mf6ctable[MF6CTBLSIZ]; 18793787SdesSYSCTL_OPAQUE(_net_inet6_ip6, OID_AUTO, mf6ctable, CTLFLAG_RD, 18893787Sdes &mf6ctable, sizeof(mf6ctable), "S,*mf6ctable[MF6CTBLSIZ]", 18993787Sdes "IPv6 Multicast Forwarding Table (struct *mf6ctable[MF6CTBLSIZ], " 19093787Sdes "netinet6/ip6_mroute.h)"); 19193787Sdes 19293787Sdesstatic struct mtx mfc6_mtx; 19393787Sdes#define MFC6_LOCK() mtx_lock(&mfc6_mtx) 19493787Sdes#define MFC6_UNLOCK() mtx_unlock(&mfc6_mtx) 19594691Sdes#define MFC6_LOCK_ASSERT() do { \ 19693787Sdes mtx_assert(&mfc6_mtx, MA_OWNED); \ 19793787Sdes NET_ASSERT_GIANT(); \ 19893787Sdes} while (0) 19993787Sdes#define MFC6_LOCK_INIT() \ 20093787Sdes mtx_init(&mfc6_mtx, "IPv6 multicast forwarding cache", NULL, MTX_DEF) 20193787Sdes#define MFC6_LOCK_DESTROY() mtx_destroy(&mfc6_mtx) 20293787Sdes 20393787Sdesstatic u_char n6expire[MF6CTBLSIZ]; 20493787Sdes 20593787Sdesstatic struct mif6 mif6table[MAXMIFS]; 20694691SdesSYSCTL_OPAQUE(_net_inet6_ip6, OID_AUTO, mif6table, CTLFLAG_RD, 20794691Sdes &mif6table, sizeof(mif6table), "S,mif6[MAXMIFS]", 20894691Sdes "IPv6 Multicast Interfaces (struct mif6[MAXMIFS], netinet6/ip6_mroute.h)"); 20994691Sdes 21093787Sdesstatic struct mtx mif6_mtx; 21193787Sdes#define MIF6_LOCK() mtx_lock(&mif6_mtx) 21294691Sdes#define MIF6_UNLOCK() mtx_unlock(&mif6_mtx) 21393787Sdes#define MIF6_LOCK_ASSERT() mtx_assert(&mif6_mtx, MA_OWNED) 21493787Sdes#define MIF6_LOCK_INIT() \ 21593787Sdes mtx_init(&mif6_mtx, "IPv6 multicast interfaces", NULL, MTX_DEF) 21693787Sdes#define MIF6_LOCK_DESTROY() mtx_destroy(&mif6_mtx) 21793787Sdes 21894691Sdes#ifdef MRT6DEBUG 21994691Sdes#ifdef VIMAGE_GLOBALS 22093787Sdesstatic u_int mrt6debug; /* debug level */ 22193787Sdes#endif 22293787Sdes#define DEBUG_MFC 0x02 22394691Sdes#define DEBUG_FORWARD 0x04 22494691Sdes#define DEBUG_EXPIRE 0x08 22593787Sdes#define DEBUG_XMIT 0x10 22693787Sdes#define DEBUG_REG 0x20 22793787Sdes#define DEBUG_PIM 0x40 22894691Sdes#endif 22994691Sdes 23093787Sdesstatic void expire_upcalls(void *); 23193787Sdes#define EXPIRE_TIMEOUT (hz / 4) /* 4x / second */ 23293787Sdes#define UPCALL_EXPIRE 6 /* number of timeouts */ 23393787Sdes 23493787Sdes/* 23593787Sdes * XXX TODO: maintain a count to if_allmulti() calls in struct ifnet. 23693787Sdes */ 23793787Sdes 23893787Sdes/* 23993787Sdes * 'Interfaces' associated with decapsulator (so we can tell 24093787Sdes * packets that went through it from ones that get reflected 24193787Sdes * by a broken gateway). Different from IPv4 register_if, 24294691Sdes * these interfaces are linked into the system ifnet list, 24394691Sdes * because per-interface IPv6 statistics are maintained in 24494691Sdes * ifp->if_afdata. But it does not have any routes point 24593787Sdes * to them. I.e., packets can't be sent this way. They 24693787Sdes * only exist as a placeholder for multicast source 24794691Sdes * verification. 24893787Sdes */ 24993787Sdesstatic struct ifnet *multicast_register_if6; 25093787Sdes 25193787Sdes#define ENCAP_HOPS 64 25293787Sdes 25393787Sdes/* 25493787Sdes * Private variables. 25593787Sdes */ 25693787Sdesstatic mifi_t nummifs = 0; 25793787Sdesstatic mifi_t reg_mif_num = (mifi_t)-1; 25893787Sdes 25993787Sdesstatic struct pim6stat pim6stat; 26093787SdesSYSCTL_STRUCT(_net_inet6_pim, PIM6CTL_STATS, stats, CTLFLAG_RD, 26193787Sdes &pim6stat, pim6stat, 26293787Sdes "PIM Statistics (struct pim6stat, netinet6/pim_var.h)"); 26393787Sdes 26493787Sdes#ifdef VIMAGE_GLOBALS 26593787Sdesstatic int pim6; 26694691Sdes#endif 26794691Sdes 26893787Sdes/* 26993787Sdes * Hash function for a source, group entry 27093787Sdes */ 27193787Sdes#define MF6CHASH(a, g) MF6CHASHMOD((a).s6_addr32[0] ^ (a).s6_addr32[1] ^ \ 27293787Sdes (a).s6_addr32[2] ^ (a).s6_addr32[3] ^ \ 27393787Sdes (g).s6_addr32[0] ^ (g).s6_addr32[1] ^ \ 27493787Sdes (g).s6_addr32[2] ^ (g).s6_addr32[3]) 27593787Sdes 27693787Sdes/* 27793787Sdes * Find a route for a given origin IPv6 address and Multicast group address. 27893787Sdes */ 27993787Sdes#define MF6CFIND(o, g, rt) do { \ 28093787Sdes struct mf6c *_rt = mf6ctable[MF6CHASH(o,g)]; \ 28193787Sdes rt = NULL; \ 28293787Sdes mrt6stat.mrt6s_mfc_lookups++; \ 28393787Sdes while (_rt) { \ 28493787Sdes if (IN6_ARE_ADDR_EQUAL(&_rt->mf6c_origin.sin6_addr, &(o)) && \ 28593787Sdes IN6_ARE_ADDR_EQUAL(&_rt->mf6c_mcastgrp.sin6_addr, &(g)) && \ 28693787Sdes (_rt->mf6c_stall == NULL)) { \ 28793787Sdes rt = _rt; \ 28893787Sdes break; \ 28993787Sdes } \ 29093787Sdes _rt = _rt->mf6c_next; \ 29193787Sdes } \ 29293787Sdes if (rt == NULL) { \ 29393787Sdes mrt6stat.mrt6s_mfc_misses++; \ 29493787Sdes } \ 29593787Sdes} while (/*CONSTCOND*/ 0) 29693787Sdes 29793787Sdes/* 29893787Sdes * Macros to compute elapsed time efficiently 29993787Sdes * Borrowed from Van Jacobson's scheduling code 30093787Sdes * XXX: replace with timersub() ? 30193787Sdes */ 30293787Sdes#define TV_DELTA(a, b, delta) do { \ 30393787Sdes int xxs; \ 30493787Sdes \ 30593787Sdes delta = (a).tv_usec - (b).tv_usec; \ 30693787Sdes if ((xxs = (a).tv_sec - (b).tv_sec)) { \ 30793787Sdes switch (xxs) { \ 30893787Sdes case 2: \ 30993787Sdes delta += 1000000; \ 31093787Sdes /* FALLTHROUGH */ \ 31193787Sdes case 1: \ 31293787Sdes delta += 1000000; \ 31393787Sdes break; \ 31493787Sdes default: \ 31594691Sdes delta += (1000000 * xxs); \ 31693787Sdes } \ 31793787Sdes } \ 31893787Sdes} while (/*CONSTCOND*/ 0) 31993787Sdes 32093787Sdes/* XXX: replace with timercmp(a, b, <) ? */ 32193787Sdes#define TV_LT(a, b) (((a).tv_usec < (b).tv_usec && \ 32293787Sdes (a).tv_sec <= (b).tv_sec) || (a).tv_sec < (b).tv_sec) 32393787Sdes 32493787Sdes#ifdef UPCALL_TIMING 32593787Sdes#define UPCALL_MAX 50 32693787Sdesstatic u_long upcall_data[UPCALL_MAX + 1]; 32793787Sdesstatic void collate(); 32893787Sdes#endif /* UPCALL_TIMING */ 32993787Sdes 33093787Sdesstatic int ip6_mrouter_init(struct socket *, int, int); 33193787Sdesstatic int add_m6fc(struct mf6cctl *); 33293787Sdesstatic int add_m6if(struct mif6ctl *); 33393787Sdesstatic int del_m6fc(struct mf6cctl *); 33493787Sdesstatic int del_m6if(mifi_t *); 33593787Sdesstatic int del_m6if_locked(mifi_t *); 33693787Sdesstatic int get_mif6_cnt(struct sioc_mif_req6 *); 33793787Sdesstatic int get_sg_cnt(struct sioc_sg_req6 *); 33893787Sdes 33993787Sdesstatic struct callout expire_upcalls_ch; 34093787Sdes 34193787Sdesint X_ip6_mforward(struct ip6_hdr *, struct ifnet *, struct mbuf *); 34293787Sdesint X_ip6_mrouter_done(void); 34393787Sdesint X_ip6_mrouter_set(struct socket *, struct sockopt *); 34493787Sdesint X_ip6_mrouter_get(struct socket *, struct sockopt *); 34593787Sdesint X_mrt6_ioctl(u_long, caddr_t); 34694691Sdes 34793787Sdesstatic void 34893787Sdespim6_init(void) 34993787Sdes{ 35093787Sdes INIT_VNET_INET6(curvnet); 35193787Sdes 35293787Sdes V_ip6_mrouter_ver = 0; 35393787Sdes#ifdef MRT6DEBUG 35493787Sdes V_mrt6debug = 0; /* debug level */ 35593787Sdes#endif 35693787Sdes} 35793787Sdes 35893787Sdes/* 35993787Sdes * Handle MRT setsockopt commands to modify the multicast routing tables. 36093787Sdes */ 36193787Sdesint 36293787SdesX_ip6_mrouter_set(struct socket *so, struct sockopt *sopt) 36393787Sdes{ 36493787Sdes INIT_VNET_INET6(curvnet); 36593787Sdes int error = 0; 36693787Sdes int optval; 36793787Sdes struct mif6ctl mifc; 36893787Sdes struct mf6cctl mfcc; 36993787Sdes mifi_t mifi; 37093787Sdes 37193787Sdes if (so != V_ip6_mrouter && sopt->sopt_name != MRT6_INIT) 37293787Sdes return (EACCES); 37393787Sdes 37493787Sdes switch (sopt->sopt_name) { 37593787Sdes case MRT6_INIT: 37693787Sdes#ifdef MRT6_OINIT 37793787Sdes case MRT6_OINIT: 37893787Sdes#endif 37993787Sdes error = sooptcopyin(sopt, &optval, sizeof(optval), 38093787Sdes sizeof(optval)); 38193787Sdes if (error) 38293787Sdes break; 38393787Sdes error = ip6_mrouter_init(so, optval, sopt->sopt_name); 38493787Sdes break; 38593787Sdes case MRT6_DONE: 38693787Sdes error = X_ip6_mrouter_done(); 38793787Sdes break; 38893787Sdes case MRT6_ADD_MIF: 38993787Sdes error = sooptcopyin(sopt, &mifc, sizeof(mifc), sizeof(mifc)); 39093787Sdes if (error) 39193787Sdes break; 39293787Sdes error = add_m6if(&mifc); 39393787Sdes break; 39493787Sdes case MRT6_ADD_MFC: 39593787Sdes error = sooptcopyin(sopt, &mfcc, sizeof(mfcc), sizeof(mfcc)); 39693787Sdes if (error) 39793787Sdes break; 39893787Sdes error = add_m6fc(&mfcc); 39993787Sdes break; 40093787Sdes case MRT6_DEL_MFC: 40193787Sdes error = sooptcopyin(sopt, &mfcc, sizeof(mfcc), sizeof(mfcc)); 40293787Sdes if (error) 40393787Sdes break; 40493787Sdes error = del_m6fc(&mfcc); 40593787Sdes break; 40693787Sdes case MRT6_DEL_MIF: 40793787Sdes error = sooptcopyin(sopt, &mifi, sizeof(mifi), sizeof(mifi)); 40893787Sdes if (error) 40993787Sdes break; 41093787Sdes error = del_m6if(&mifi); 41193787Sdes break; 41293787Sdes case MRT6_PIM: 41394889Sdes error = sooptcopyin(sopt, &optval, sizeof(optval), 41493787Sdes sizeof(optval)); 41593787Sdes if (error) 41693787Sdes break; 41793787Sdes error = set_pim6(&optval); 41894889Sdes break; 41993787Sdes default: 42093787Sdes error = EOPNOTSUPP; 42193787Sdes break; 42293787Sdes } 42393787Sdes 42493787Sdes return (error); 42593787Sdes} 42693787Sdes 42793787Sdes/* 42893787Sdes * Handle MRT getsockopt commands 42993787Sdes */ 43093787Sdesint 43193787SdesX_ip6_mrouter_get(struct socket *so, struct sockopt *sopt) 43293787Sdes{ 43393787Sdes INIT_VNET_INET6(curvnet); 43493787Sdes int error = 0; 43593787Sdes 43693787Sdes if (so != V_ip6_mrouter) 43793787Sdes return (EACCES); 43893787Sdes 43993787Sdes switch (sopt->sopt_name) { 44093787Sdes case MRT6_PIM: 44193787Sdes error = sooptcopyout(sopt, &V_pim6, sizeof(V_pim6)); 44293787Sdes break; 44393787Sdes } 44493787Sdes return (error); 44593787Sdes} 44693787Sdes 44793787Sdes/* 44893787Sdes * Handle ioctl commands to obtain information from the cache 44993787Sdes */ 45093787Sdesint 45193787SdesX_mrt6_ioctl(u_long cmd, caddr_t data) 45293787Sdes{ 45393787Sdes int ret; 45493787Sdes 45593787Sdes ret = EINVAL; 45693787Sdes 45793787Sdes switch (cmd) { 45893787Sdes case SIOCGETSGCNT_IN6: 45993787Sdes ret = get_sg_cnt((struct sioc_sg_req6 *)data); 46093787Sdes break; 46193787Sdes 46293787Sdes case SIOCGETMIFCNT_IN6: 46393787Sdes ret = get_mif6_cnt((struct sioc_mif_req6 *)data); 46493787Sdes break; 46593787Sdes 46693787Sdes default: 46793787Sdes break; 46893787Sdes } 46993787Sdes 47093787Sdes return (ret); 47193787Sdes} 47294889Sdes 47393787Sdes/* 47493787Sdes * returns the packet, byte, rpf-failure count for the source group provided 47593787Sdes */ 47693787Sdesstatic int 47794889Sdesget_sg_cnt(struct sioc_sg_req6 *req) 47893787Sdes{ 47993787Sdes struct mf6c *rt; 48093787Sdes int ret; 48193787Sdes 48293787Sdes ret = 0; 48393787Sdes 48493787Sdes MFC6_LOCK(); 48593787Sdes 48693787Sdes MF6CFIND(req->src.sin6_addr, req->grp.sin6_addr, rt); 48793787Sdes if (rt == NULL) { 48893787Sdes ret = ESRCH; 48993787Sdes } else { 49094889Sdes req->pktcnt = rt->mf6c_pkt_cnt; 49193787Sdes req->bytecnt = rt->mf6c_byte_cnt; 49293787Sdes req->wrong_if = rt->mf6c_wrong_if; 49393787Sdes } 49494889Sdes 49593787Sdes MFC6_UNLOCK(); 49693787Sdes 49793787Sdes return (ret); 49893787Sdes} 49993787Sdes 50093787Sdes/* 50193787Sdes * returns the input and output packet and byte counts on the mif provided 50293787Sdes */ 50393787Sdesstatic int 50493787Sdesget_mif6_cnt(struct sioc_mif_req6 *req) 50593787Sdes{ 50693787Sdes mifi_t mifi; 50793787Sdes int ret; 50894889Sdes 50993787Sdes ret = 0; 51093787Sdes mifi = req->mifi; 51193787Sdes 51293787Sdes MIF6_LOCK(); 51393787Sdes 51493787Sdes if (mifi >= nummifs) { 51593787Sdes ret = EINVAL; 51693787Sdes } else { 51793787Sdes req->icount = mif6table[mifi].m6_pkt_in; 51893787Sdes req->ocount = mif6table[mifi].m6_pkt_out; 51993787Sdes req->ibytes = mif6table[mifi].m6_bytes_in; 52093787Sdes req->obytes = mif6table[mifi].m6_bytes_out; 52193787Sdes } 52294889Sdes 52393787Sdes MIF6_UNLOCK(); 52493787Sdes 52593787Sdes return (ret); 52693787Sdes} 52793787Sdes 52894889Sdesstatic int 52993787Sdesset_pim6(int *i) 53093787Sdes{ 53193787Sdes INIT_VNET_INET6(curvnet); 53293787Sdes if ((*i != 1) && (*i != 0)) 53393787Sdes return (EINVAL); 53493787Sdes 53593787Sdes V_pim6 = *i; 53693787Sdes 53793787Sdes return (0); 53893787Sdes} 53993787Sdes 54093787Sdes/* 54193787Sdes * Enable multicast routing 54293787Sdes */ 54393787Sdesstatic int 54493787Sdesip6_mrouter_init(struct socket *so, int v, int cmd) 54593787Sdes{ 54693787Sdes INIT_VNET_INET6(curvnet); 54794691Sdes 54894691Sdes V_ip6_mrouter_ver = 0; 54994691Sdes 55093787Sdes#ifdef MRT6DEBUG 55193787Sdes V_mrt6debug = 0; 55293787Sdes 55393787Sdes if (V_mrt6debug) 55493787Sdes log(LOG_DEBUG, 55593787Sdes "ip6_mrouter_init: so_type = %d, pr_protocol = %d\n", 55693787Sdes so->so_type, so->so_proto->pr_protocol); 55793787Sdes#endif 55893787Sdes 55993787Sdes if (so->so_type != SOCK_RAW || 560 so->so_proto->pr_protocol != IPPROTO_ICMPV6) 561 return (EOPNOTSUPP); 562 563 if (v != 1) 564 return (ENOPROTOOPT); 565 566 MROUTER6_LOCK(); 567 568 if (V_ip6_mrouter != NULL) { 569 MROUTER6_UNLOCK(); 570 return (EADDRINUSE); 571 } 572 573 V_ip6_mrouter = so; 574 V_ip6_mrouter_ver = cmd; 575 576 bzero((caddr_t)mf6ctable, sizeof(mf6ctable)); 577 bzero((caddr_t)n6expire, sizeof(n6expire)); 578 579 V_pim6 = 0;/* used for stubbing out/in pim stuff */ 580 581 callout_init(&expire_upcalls_ch, 0); 582 callout_reset(&expire_upcalls_ch, EXPIRE_TIMEOUT, 583 expire_upcalls, NULL); 584 585 MROUTER6_UNLOCK(); 586 587#ifdef MRT6DEBUG 588 if (V_mrt6debug) 589 log(LOG_DEBUG, "ip6_mrouter_init\n"); 590#endif 591 592 return (0); 593} 594 595/* 596 * Disable IPv6 multicast forwarding. 597 */ 598int 599X_ip6_mrouter_done(void) 600{ 601 INIT_VNET_INET6(curvnet); 602 mifi_t mifi; 603 int i; 604 struct mf6c *rt; 605 struct rtdetq *rte; 606 607 MROUTER6_LOCK(); 608 609 if (V_ip6_mrouter == NULL) { 610 MROUTER6_UNLOCK(); 611 return (EINVAL); 612 } 613 614 /* 615 * For each phyint in use, disable promiscuous reception of all IPv6 616 * multicasts. 617 */ 618 for (mifi = 0; mifi < nummifs; mifi++) { 619 if (mif6table[mifi].m6_ifp && 620 !(mif6table[mifi].m6_flags & MIFF_REGISTER)) { 621 if_allmulti(mif6table[mifi].m6_ifp, 0); 622 } 623 } 624 bzero((caddr_t)mif6table, sizeof(mif6table)); 625 nummifs = 0; 626 627 V_pim6 = 0; /* used to stub out/in pim specific code */ 628 629 callout_stop(&expire_upcalls_ch); 630 631 /* 632 * Free all multicast forwarding cache entries. 633 */ 634 MFC6_LOCK(); 635 for (i = 0; i < MF6CTBLSIZ; i++) { 636 rt = mf6ctable[i]; 637 while (rt) { 638 struct mf6c *frt; 639 640 for (rte = rt->mf6c_stall; rte != NULL; ) { 641 struct rtdetq *n = rte->next; 642 643 m_free(rte->m); 644 free(rte, M_MRTABLE6); 645 rte = n; 646 } 647 frt = rt; 648 rt = rt->mf6c_next; 649 free(frt, M_MRTABLE6); 650 } 651 } 652 bzero((caddr_t)mf6ctable, sizeof(mf6ctable)); 653 MFC6_UNLOCK(); 654 655 /* 656 * Reset register interface 657 */ 658 if (reg_mif_num != (mifi_t)-1 && multicast_register_if6 != NULL) { 659 if_detach(multicast_register_if6); 660 if_free(multicast_register_if6); 661 reg_mif_num = (mifi_t)-1; 662 multicast_register_if6 = NULL; 663 } 664 665 V_ip6_mrouter = NULL; 666 V_ip6_mrouter_ver = 0; 667 668 MROUTER6_UNLOCK(); 669 670#ifdef MRT6DEBUG 671 if (V_mrt6debug) 672 log(LOG_DEBUG, "ip6_mrouter_done\n"); 673#endif 674 675 return (0); 676} 677 678static struct sockaddr_in6 sin6 = { sizeof(sin6), AF_INET6 }; 679 680/* 681 * Add a mif to the mif table 682 */ 683static int 684add_m6if(struct mif6ctl *mifcp) 685{ 686 INIT_VNET_NET(curvnet); 687 struct mif6 *mifp; 688 struct ifnet *ifp; 689 int error; 690 691 MIF6_LOCK(); 692 693 if (mifcp->mif6c_mifi >= MAXMIFS) { 694 MIF6_UNLOCK(); 695 return (EINVAL); 696 } 697 mifp = mif6table + mifcp->mif6c_mifi; 698 if (mifp->m6_ifp != NULL) { 699 MIF6_UNLOCK(); 700 return (EADDRINUSE); /* XXX: is it appropriate? */ 701 } 702 if (mifcp->mif6c_pifi == 0 || mifcp->mif6c_pifi > V_if_index) { 703 MIF6_UNLOCK(); 704 return (ENXIO); 705 } 706 707 ifp = ifnet_byindex(mifcp->mif6c_pifi); 708 709 if (mifcp->mif6c_flags & MIFF_REGISTER) { 710 if (reg_mif_num == (mifi_t)-1) { 711 ifp = if_alloc(IFT_OTHER); 712 713 if_initname(ifp, "register_mif", 0); 714 ifp->if_flags |= IFF_LOOPBACK; 715 if_attach(ifp); 716 multicast_register_if6 = ifp; 717 reg_mif_num = mifcp->mif6c_mifi; 718 /* 719 * it is impossible to guess the ifindex of the 720 * register interface. So mif6c_pifi is automatically 721 * calculated. 722 */ 723 mifcp->mif6c_pifi = ifp->if_index; 724 } else { 725 ifp = multicast_register_if6; 726 } 727 } else { 728 /* Make sure the interface supports multicast */ 729 if ((ifp->if_flags & IFF_MULTICAST) == 0) { 730 MIF6_UNLOCK(); 731 return (EOPNOTSUPP); 732 } 733 734 error = if_allmulti(ifp, 1); 735 if (error) { 736 MIF6_UNLOCK(); 737 return (error); 738 } 739 } 740 741 mifp->m6_flags = mifcp->mif6c_flags; 742 mifp->m6_ifp = ifp; 743 744 /* initialize per mif pkt counters */ 745 mifp->m6_pkt_in = 0; 746 mifp->m6_pkt_out = 0; 747 mifp->m6_bytes_in = 0; 748 mifp->m6_bytes_out = 0; 749 bzero(&mifp->m6_route, sizeof(mifp->m6_route)); 750 751 /* Adjust nummifs up if the mifi is higher than nummifs */ 752 if (nummifs <= mifcp->mif6c_mifi) 753 nummifs = mifcp->mif6c_mifi + 1; 754 755 MIF6_UNLOCK(); 756 757#ifdef MRT6DEBUG 758 if (V_mrt6debug) 759 log(LOG_DEBUG, 760 "add_mif #%d, phyint %s\n", 761 mifcp->mif6c_mifi, 762 ifp->if_xname); 763#endif 764 765 return (0); 766} 767 768/* 769 * Delete a mif from the mif table 770 */ 771static int 772del_m6if_locked(mifi_t *mifip) 773{ 774 struct mif6 *mifp = mif6table + *mifip; 775 mifi_t mifi; 776 struct ifnet *ifp; 777 778 MIF6_LOCK_ASSERT(); 779 780 if (*mifip >= nummifs) 781 return (EINVAL); 782 if (mifp->m6_ifp == NULL) 783 return (EINVAL); 784 785 if (!(mifp->m6_flags & MIFF_REGISTER)) { 786 /* XXX: TODO: Maintain an ALLMULTI refcount in struct ifnet. */ 787 ifp = mifp->m6_ifp; 788 if_allmulti(ifp, 0); 789 } else { 790 if (reg_mif_num != (mifi_t)-1 && 791 multicast_register_if6 != NULL) { 792 if_detach(multicast_register_if6); 793 if_free(multicast_register_if6); 794 reg_mif_num = (mifi_t)-1; 795 multicast_register_if6 = NULL; 796 } 797 } 798 799 bzero((caddr_t)mifp, sizeof(*mifp)); 800 801 /* Adjust nummifs down */ 802 for (mifi = nummifs; mifi > 0; mifi--) 803 if (mif6table[mifi - 1].m6_ifp) 804 break; 805 nummifs = mifi; 806 807#ifdef MRT6DEBUG 808 if (V_mrt6debug) 809 log(LOG_DEBUG, "del_m6if %d, nummifs %d\n", *mifip, nummifs); 810#endif 811 812 return (0); 813} 814 815static int 816del_m6if(mifi_t *mifip) 817{ 818 int cc; 819 820 MIF6_LOCK(); 821 cc = del_m6if_locked(mifip); 822 MIF6_UNLOCK(); 823 824 return (cc); 825} 826 827/* 828 * Add an mfc entry 829 */ 830static int 831add_m6fc(struct mf6cctl *mfccp) 832{ 833 struct mf6c *rt; 834 u_long hash; 835 struct rtdetq *rte; 836 u_short nstl; 837 char ip6bufo[INET6_ADDRSTRLEN], ip6bufg[INET6_ADDRSTRLEN]; 838 839 MFC6_LOCK(); 840 841 MF6CFIND(mfccp->mf6cc_origin.sin6_addr, 842 mfccp->mf6cc_mcastgrp.sin6_addr, rt); 843 844 /* If an entry already exists, just update the fields */ 845 if (rt) { 846#ifdef MRT6DEBUG 847 if (V_mrt6debug & DEBUG_MFC) { 848 log(LOG_DEBUG, 849 "add_m6fc no upcall h %d o %s g %s p %x\n", 850 ip6_sprintf(ip6bufo, &mfccp->mf6cc_origin.sin6_addr), 851 ip6_sprintf(ip6bufg, &mfccp->mf6cc_mcastgrp.sin6_addr), 852 mfccp->mf6cc_parent); 853 } 854#endif 855 856 rt->mf6c_parent = mfccp->mf6cc_parent; 857 rt->mf6c_ifset = mfccp->mf6cc_ifset; 858 859 MFC6_UNLOCK(); 860 return (0); 861 } 862 863 /* 864 * Find the entry for which the upcall was made and update 865 */ 866 hash = MF6CHASH(mfccp->mf6cc_origin.sin6_addr, 867 mfccp->mf6cc_mcastgrp.sin6_addr); 868 for (rt = mf6ctable[hash], nstl = 0; rt; rt = rt->mf6c_next) { 869 if (IN6_ARE_ADDR_EQUAL(&rt->mf6c_origin.sin6_addr, 870 &mfccp->mf6cc_origin.sin6_addr) && 871 IN6_ARE_ADDR_EQUAL(&rt->mf6c_mcastgrp.sin6_addr, 872 &mfccp->mf6cc_mcastgrp.sin6_addr) && 873 (rt->mf6c_stall != NULL)) { 874 875 if (nstl++) 876 log(LOG_ERR, 877 "add_m6fc: %s o %s g %s p %x dbx %p\n", 878 "multiple kernel entries", 879 ip6_sprintf(ip6bufo, 880 &mfccp->mf6cc_origin.sin6_addr), 881 ip6_sprintf(ip6bufg, 882 &mfccp->mf6cc_mcastgrp.sin6_addr), 883 mfccp->mf6cc_parent, rt->mf6c_stall); 884 885#ifdef MRT6DEBUG 886 if (V_mrt6debug & DEBUG_MFC) 887 log(LOG_DEBUG, 888 "add_m6fc o %s g %s p %x dbg %x\n", 889 ip6_sprintf(ip6bufo, 890 &mfccp->mf6cc_origin.sin6_addr), 891 ip6_sprintf(ip6bufg, 892 &mfccp->mf6cc_mcastgrp.sin6_addr), 893 mfccp->mf6cc_parent, rt->mf6c_stall); 894#endif 895 896 rt->mf6c_origin = mfccp->mf6cc_origin; 897 rt->mf6c_mcastgrp = mfccp->mf6cc_mcastgrp; 898 rt->mf6c_parent = mfccp->mf6cc_parent; 899 rt->mf6c_ifset = mfccp->mf6cc_ifset; 900 /* initialize pkt counters per src-grp */ 901 rt->mf6c_pkt_cnt = 0; 902 rt->mf6c_byte_cnt = 0; 903 rt->mf6c_wrong_if = 0; 904 905 rt->mf6c_expire = 0; /* Don't clean this guy up */ 906 n6expire[hash]--; 907 908 /* free packets Qed at the end of this entry */ 909 for (rte = rt->mf6c_stall; rte != NULL; ) { 910 struct rtdetq *n = rte->next; 911 ip6_mdq(rte->m, rte->ifp, rt); 912 m_freem(rte->m); 913#ifdef UPCALL_TIMING 914 collate(&(rte->t)); 915#endif /* UPCALL_TIMING */ 916 free(rte, M_MRTABLE6); 917 rte = n; 918 } 919 rt->mf6c_stall = NULL; 920 } 921 } 922 923 /* 924 * It is possible that an entry is being inserted without an upcall 925 */ 926 if (nstl == 0) { 927#ifdef MRT6DEBUG 928 if (V_mrt6debug & DEBUG_MFC) 929 log(LOG_DEBUG, 930 "add_mfc no upcall h %d o %s g %s p %x\n", 931 hash, 932 ip6_sprintf(ip6bufo, &mfccp->mf6cc_origin.sin6_addr), 933 ip6_sprintf(ip6bufg, &mfccp->mf6cc_mcastgrp.sin6_addr), 934 mfccp->mf6cc_parent); 935#endif 936 937 for (rt = mf6ctable[hash]; rt; rt = rt->mf6c_next) { 938 939 if (IN6_ARE_ADDR_EQUAL(&rt->mf6c_origin.sin6_addr, 940 &mfccp->mf6cc_origin.sin6_addr)&& 941 IN6_ARE_ADDR_EQUAL(&rt->mf6c_mcastgrp.sin6_addr, 942 &mfccp->mf6cc_mcastgrp.sin6_addr)) { 943 944 rt->mf6c_origin = mfccp->mf6cc_origin; 945 rt->mf6c_mcastgrp = mfccp->mf6cc_mcastgrp; 946 rt->mf6c_parent = mfccp->mf6cc_parent; 947 rt->mf6c_ifset = mfccp->mf6cc_ifset; 948 /* initialize pkt counters per src-grp */ 949 rt->mf6c_pkt_cnt = 0; 950 rt->mf6c_byte_cnt = 0; 951 rt->mf6c_wrong_if = 0; 952 953 if (rt->mf6c_expire) 954 n6expire[hash]--; 955 rt->mf6c_expire = 0; 956 } 957 } 958 if (rt == NULL) { 959 /* no upcall, so make a new entry */ 960 rt = (struct mf6c *)malloc(sizeof(*rt), M_MRTABLE6, 961 M_NOWAIT); 962 if (rt == NULL) { 963 MFC6_UNLOCK(); 964 return (ENOBUFS); 965 } 966 967 /* insert new entry at head of hash chain */ 968 rt->mf6c_origin = mfccp->mf6cc_origin; 969 rt->mf6c_mcastgrp = mfccp->mf6cc_mcastgrp; 970 rt->mf6c_parent = mfccp->mf6cc_parent; 971 rt->mf6c_ifset = mfccp->mf6cc_ifset; 972 /* initialize pkt counters per src-grp */ 973 rt->mf6c_pkt_cnt = 0; 974 rt->mf6c_byte_cnt = 0; 975 rt->mf6c_wrong_if = 0; 976 rt->mf6c_expire = 0; 977 rt->mf6c_stall = NULL; 978 979 /* link into table */ 980 rt->mf6c_next = mf6ctable[hash]; 981 mf6ctable[hash] = rt; 982 } 983 } 984 985 MFC6_UNLOCK(); 986 return (0); 987} 988 989#ifdef UPCALL_TIMING 990/* 991 * collect delay statistics on the upcalls 992 */ 993static void 994collate(struct timeval *t) 995{ 996 u_long d; 997 struct timeval tp; 998 u_long delta; 999 1000 GET_TIME(tp); 1001 1002 if (TV_LT(*t, tp)) 1003 { 1004 TV_DELTA(tp, *t, delta); 1005 1006 d = delta >> 10; 1007 if (d > UPCALL_MAX) 1008 d = UPCALL_MAX; 1009 1010 ++upcall_data[d]; 1011 } 1012} 1013#endif /* UPCALL_TIMING */ 1014 1015/* 1016 * Delete an mfc entry 1017 */ 1018static int 1019del_m6fc(struct mf6cctl *mfccp) 1020{ 1021 struct sockaddr_in6 origin; 1022 struct sockaddr_in6 mcastgrp; 1023 struct mf6c *rt; 1024 struct mf6c **nptr; 1025 u_long hash; 1026 1027 origin = mfccp->mf6cc_origin; 1028 mcastgrp = mfccp->mf6cc_mcastgrp; 1029 hash = MF6CHASH(origin.sin6_addr, mcastgrp.sin6_addr); 1030 1031#ifdef MRT6DEBUG 1032 if (V_mrt6debug & DEBUG_MFC) { 1033 char ip6bufo[INET6_ADDRSTRLEN], ip6bufg[INET6_ADDRSTRLEN]; 1034 log(LOG_DEBUG,"del_m6fc orig %s mcastgrp %s\n", 1035 ip6_sprintf(ip6bufo, &origin.sin6_addr), 1036 ip6_sprintf(ip6bufg, &mcastgrp.sin6_addr)); 1037 } 1038#endif 1039 1040 MFC6_LOCK(); 1041 1042 nptr = &mf6ctable[hash]; 1043 while ((rt = *nptr) != NULL) { 1044 if (IN6_ARE_ADDR_EQUAL(&origin.sin6_addr, 1045 &rt->mf6c_origin.sin6_addr) && 1046 IN6_ARE_ADDR_EQUAL(&mcastgrp.sin6_addr, 1047 &rt->mf6c_mcastgrp.sin6_addr) && 1048 rt->mf6c_stall == NULL) 1049 break; 1050 1051 nptr = &rt->mf6c_next; 1052 } 1053 if (rt == NULL) { 1054 MFC6_UNLOCK(); 1055 return (EADDRNOTAVAIL); 1056 } 1057 1058 *nptr = rt->mf6c_next; 1059 free(rt, M_MRTABLE6); 1060 1061 MFC6_UNLOCK(); 1062 1063 return (0); 1064} 1065 1066static int 1067socket_send(struct socket *s, struct mbuf *mm, struct sockaddr_in6 *src) 1068{ 1069 1070 if (s) { 1071 if (sbappendaddr(&s->so_rcv, 1072 (struct sockaddr *)src, 1073 mm, (struct mbuf *)0) != 0) { 1074 sorwakeup(s); 1075 return (0); 1076 } 1077 } 1078 m_freem(mm); 1079 return (-1); 1080} 1081 1082/* 1083 * IPv6 multicast forwarding function. This function assumes that the packet 1084 * pointed to by "ip6" has arrived on (or is about to be sent to) the interface 1085 * pointed to by "ifp", and the packet is to be relayed to other networks 1086 * that have members of the packet's destination IPv6 multicast group. 1087 * 1088 * The packet is returned unscathed to the caller, unless it is 1089 * erroneous, in which case a non-zero return value tells the caller to 1090 * discard it. 1091 * 1092 * NOTE: this implementation assumes that m->m_pkthdr.rcvif is NULL iff 1093 * this function is called in the originating context (i.e., not when 1094 * forwarding a packet from other node). ip6_output(), which is currently the 1095 * only function that calls this function is called in the originating context, 1096 * explicitly ensures this condition. It is caller's responsibility to ensure 1097 * that if this function is called from somewhere else in the originating 1098 * context in the future. 1099 */ 1100int 1101X_ip6_mforward(struct ip6_hdr *ip6, struct ifnet *ifp, struct mbuf *m) 1102{ 1103 INIT_VNET_INET6(curvnet); 1104 struct mf6c *rt; 1105 struct mif6 *mifp; 1106 struct mbuf *mm; 1107 mifi_t mifi; 1108 char ip6bufs[INET6_ADDRSTRLEN], ip6bufd[INET6_ADDRSTRLEN]; 1109 1110#ifdef MRT6DEBUG 1111 if (V_mrt6debug & DEBUG_FORWARD) 1112 log(LOG_DEBUG, "ip6_mforward: src %s, dst %s, ifindex %d\n", 1113 ip6_sprintf(ip6bufs, &ip6->ip6_src), 1114 ip6_sprintf(ip6bufd, &ip6->ip6_dst), 1115 ifp->if_index); 1116#endif 1117 1118 /* 1119 * Don't forward a packet with Hop limit of zero or one, 1120 * or a packet destined to a local-only group. 1121 */ 1122 if (ip6->ip6_hlim <= 1 || IN6_IS_ADDR_MC_INTFACELOCAL(&ip6->ip6_dst) || 1123 IN6_IS_ADDR_MC_LINKLOCAL(&ip6->ip6_dst)) 1124 return (0); 1125 ip6->ip6_hlim--; 1126 1127 /* 1128 * Source address check: do not forward packets with unspecified 1129 * source. It was discussed in July 2000, on ipngwg mailing list. 1130 * This is rather more serious than unicast cases, because some 1131 * MLD packets can be sent with the unspecified source address 1132 * (although such packets must normally set 1 to the hop limit field). 1133 */ 1134 if (IN6_IS_ADDR_UNSPECIFIED(&ip6->ip6_src)) { 1135 V_ip6stat.ip6s_cantforward++; 1136 if (V_ip6_log_time + V_ip6_log_interval < time_second) { 1137 V_ip6_log_time = time_second; 1138 log(LOG_DEBUG, 1139 "cannot forward " 1140 "from %s to %s nxt %d received on %s\n", 1141 ip6_sprintf(ip6bufs, &ip6->ip6_src), 1142 ip6_sprintf(ip6bufd, &ip6->ip6_dst), 1143 ip6->ip6_nxt, 1144 if_name(m->m_pkthdr.rcvif)); 1145 } 1146 return (0); 1147 } 1148 1149 MFC6_LOCK(); 1150 1151 /* 1152 * Determine forwarding mifs from the forwarding cache table 1153 */ 1154 MF6CFIND(ip6->ip6_src, ip6->ip6_dst, rt); 1155 1156 /* Entry exists, so forward if necessary */ 1157 if (rt) { 1158 MFC6_UNLOCK(); 1159 return (ip6_mdq(m, ifp, rt)); 1160 } else { 1161 /* 1162 * If we don't have a route for packet's origin, 1163 * Make a copy of the packet & 1164 * send message to routing daemon 1165 */ 1166 1167 struct mbuf *mb0; 1168 struct rtdetq *rte; 1169 u_long hash; 1170/* int i, npkts;*/ 1171#ifdef UPCALL_TIMING 1172 struct timeval tp; 1173 1174 GET_TIME(tp); 1175#endif /* UPCALL_TIMING */ 1176 1177 mrt6stat.mrt6s_no_route++; 1178#ifdef MRT6DEBUG 1179 if (V_mrt6debug & (DEBUG_FORWARD | DEBUG_MFC)) 1180 log(LOG_DEBUG, "ip6_mforward: no rte s %s g %s\n", 1181 ip6_sprintf(ip6bufs, &ip6->ip6_src), 1182 ip6_sprintf(ip6bufd, &ip6->ip6_dst)); 1183#endif 1184 1185 /* 1186 * Allocate mbufs early so that we don't do extra work if we 1187 * are just going to fail anyway. 1188 */ 1189 rte = (struct rtdetq *)malloc(sizeof(*rte), M_MRTABLE6, 1190 M_NOWAIT); 1191 if (rte == NULL) { 1192 MFC6_UNLOCK(); 1193 return (ENOBUFS); 1194 } 1195 mb0 = m_copy(m, 0, M_COPYALL); 1196 /* 1197 * Pullup packet header if needed before storing it, 1198 * as other references may modify it in the meantime. 1199 */ 1200 if (mb0 && 1201 (M_HASCL(mb0) || mb0->m_len < sizeof(struct ip6_hdr))) 1202 mb0 = m_pullup(mb0, sizeof(struct ip6_hdr)); 1203 if (mb0 == NULL) { 1204 free(rte, M_MRTABLE6); 1205 MFC6_UNLOCK(); 1206 return (ENOBUFS); 1207 } 1208 1209 /* is there an upcall waiting for this packet? */ 1210 hash = MF6CHASH(ip6->ip6_src, ip6->ip6_dst); 1211 for (rt = mf6ctable[hash]; rt; rt = rt->mf6c_next) { 1212 if (IN6_ARE_ADDR_EQUAL(&ip6->ip6_src, 1213 &rt->mf6c_origin.sin6_addr) && 1214 IN6_ARE_ADDR_EQUAL(&ip6->ip6_dst, 1215 &rt->mf6c_mcastgrp.sin6_addr) && 1216 (rt->mf6c_stall != NULL)) 1217 break; 1218 } 1219 1220 if (rt == NULL) { 1221 struct mrt6msg *im; 1222#ifdef MRT6_OINIT 1223 struct omrt6msg *oim; 1224#endif 1225 1226 /* no upcall, so make a new entry */ 1227 rt = (struct mf6c *)malloc(sizeof(*rt), M_MRTABLE6, 1228 M_NOWAIT); 1229 if (rt == NULL) { 1230 free(rte, M_MRTABLE6); 1231 m_freem(mb0); 1232 MFC6_UNLOCK(); 1233 return (ENOBUFS); 1234 } 1235 /* 1236 * Make a copy of the header to send to the user 1237 * level process 1238 */ 1239 mm = m_copy(mb0, 0, sizeof(struct ip6_hdr)); 1240 1241 if (mm == NULL) { 1242 free(rte, M_MRTABLE6); 1243 m_freem(mb0); 1244 free(rt, M_MRTABLE6); 1245 MFC6_UNLOCK(); 1246 return (ENOBUFS); 1247 } 1248 1249 /* 1250 * Send message to routing daemon 1251 */ 1252 sin6.sin6_addr = ip6->ip6_src; 1253 1254 im = NULL; 1255#ifdef MRT6_OINIT 1256 oim = NULL; 1257#endif 1258 switch (V_ip6_mrouter_ver) { 1259#ifdef MRT6_OINIT 1260 case MRT6_OINIT: 1261 oim = mtod(mm, struct omrt6msg *); 1262 oim->im6_msgtype = MRT6MSG_NOCACHE; 1263 oim->im6_mbz = 0; 1264 break; 1265#endif 1266 case MRT6_INIT: 1267 im = mtod(mm, struct mrt6msg *); 1268 im->im6_msgtype = MRT6MSG_NOCACHE; 1269 im->im6_mbz = 0; 1270 break; 1271 default: 1272 free(rte, M_MRTABLE6); 1273 m_freem(mb0); 1274 free(rt, M_MRTABLE6); 1275 MFC6_UNLOCK(); 1276 return (EINVAL); 1277 } 1278 1279#ifdef MRT6DEBUG 1280 if (V_mrt6debug & DEBUG_FORWARD) 1281 log(LOG_DEBUG, 1282 "getting the iif info in the kernel\n"); 1283#endif 1284 1285 for (mifp = mif6table, mifi = 0; 1286 mifi < nummifs && mifp->m6_ifp != ifp; 1287 mifp++, mifi++) 1288 ; 1289 1290 switch (V_ip6_mrouter_ver) { 1291#ifdef MRT6_OINIT 1292 case MRT6_OINIT: 1293 oim->im6_mif = mifi; 1294 break; 1295#endif 1296 case MRT6_INIT: 1297 im->im6_mif = mifi; 1298 break; 1299 } 1300 1301 if (socket_send(V_ip6_mrouter, mm, &sin6) < 0) { 1302 log(LOG_WARNING, "ip6_mforward: ip6_mrouter " 1303 "socket queue full\n"); 1304 mrt6stat.mrt6s_upq_sockfull++; 1305 free(rte, M_MRTABLE6); 1306 m_freem(mb0); 1307 free(rt, M_MRTABLE6); 1308 MFC6_UNLOCK(); 1309 return (ENOBUFS); 1310 } 1311 1312 mrt6stat.mrt6s_upcalls++; 1313 1314 /* insert new entry at head of hash chain */ 1315 bzero(rt, sizeof(*rt)); 1316 rt->mf6c_origin.sin6_family = AF_INET6; 1317 rt->mf6c_origin.sin6_len = sizeof(struct sockaddr_in6); 1318 rt->mf6c_origin.sin6_addr = ip6->ip6_src; 1319 rt->mf6c_mcastgrp.sin6_family = AF_INET6; 1320 rt->mf6c_mcastgrp.sin6_len = sizeof(struct sockaddr_in6); 1321 rt->mf6c_mcastgrp.sin6_addr = ip6->ip6_dst; 1322 rt->mf6c_expire = UPCALL_EXPIRE; 1323 n6expire[hash]++; 1324 rt->mf6c_parent = MF6C_INCOMPLETE_PARENT; 1325 1326 /* link into table */ 1327 rt->mf6c_next = mf6ctable[hash]; 1328 mf6ctable[hash] = rt; 1329 /* Add this entry to the end of the queue */ 1330 rt->mf6c_stall = rte; 1331 } else { 1332 /* determine if q has overflowed */ 1333 struct rtdetq **p; 1334 int npkts = 0; 1335 1336 for (p = &rt->mf6c_stall; *p != NULL; p = &(*p)->next) 1337 if (++npkts > MAX_UPQ6) { 1338 mrt6stat.mrt6s_upq_ovflw++; 1339 free(rte, M_MRTABLE6); 1340 m_freem(mb0); 1341 MFC6_UNLOCK(); 1342 return (0); 1343 } 1344 1345 /* Add this entry to the end of the queue */ 1346 *p = rte; 1347 } 1348 1349 rte->next = NULL; 1350 rte->m = mb0; 1351 rte->ifp = ifp; 1352#ifdef UPCALL_TIMING 1353 rte->t = tp; 1354#endif /* UPCALL_TIMING */ 1355 1356 MFC6_UNLOCK(); 1357 1358 return (0); 1359 } 1360} 1361 1362/* 1363 * Clean up cache entries if upcalls are not serviced 1364 * Call from the Slow Timeout mechanism, every half second. 1365 */ 1366static void 1367expire_upcalls(void *unused) 1368{ 1369 struct rtdetq *rte; 1370 struct mf6c *mfc, **nptr; 1371 int i; 1372 1373 MFC6_LOCK(); 1374 for (i = 0; i < MF6CTBLSIZ; i++) { 1375 if (n6expire[i] == 0) 1376 continue; 1377 nptr = &mf6ctable[i]; 1378 while ((mfc = *nptr) != NULL) { 1379 rte = mfc->mf6c_stall; 1380 /* 1381 * Skip real cache entries 1382 * Make sure it wasn't marked to not expire (shouldn't happen) 1383 * If it expires now 1384 */ 1385 if (rte != NULL && 1386 mfc->mf6c_expire != 0 && 1387 --mfc->mf6c_expire == 0) { 1388#ifdef MRT6DEBUG 1389 if (V_mrt6debug & DEBUG_EXPIRE) { 1390 char ip6bufo[INET6_ADDRSTRLEN]; 1391 char ip6bufg[INET6_ADDRSTRLEN]; 1392 log(LOG_DEBUG, "expire_upcalls: expiring (%s %s)\n", 1393 ip6_sprintf(ip6bufo, &mfc->mf6c_origin.sin6_addr), 1394 ip6_sprintf(ip6bufg, &mfc->mf6c_mcastgrp.sin6_addr)); 1395 } 1396#endif 1397 /* 1398 * drop all the packets 1399 * free the mbuf with the pkt, if, timing info 1400 */ 1401 do { 1402 struct rtdetq *n = rte->next; 1403 m_freem(rte->m); 1404 free(rte, M_MRTABLE6); 1405 rte = n; 1406 } while (rte != NULL); 1407 mrt6stat.mrt6s_cache_cleanups++; 1408 n6expire[i]--; 1409 1410 *nptr = mfc->mf6c_next; 1411 free(mfc, M_MRTABLE6); 1412 } else { 1413 nptr = &mfc->mf6c_next; 1414 } 1415 } 1416 } 1417 MFC6_UNLOCK(); 1418 callout_reset(&expire_upcalls_ch, EXPIRE_TIMEOUT, 1419 expire_upcalls, NULL); 1420} 1421 1422/* 1423 * Packet forwarding routine once entry in the cache is made 1424 */ 1425static int 1426ip6_mdq(struct mbuf *m, struct ifnet *ifp, struct mf6c *rt) 1427{ 1428 INIT_VNET_INET6(curvnet); 1429 struct ip6_hdr *ip6 = mtod(m, struct ip6_hdr *); 1430 mifi_t mifi, iif; 1431 struct mif6 *mifp; 1432 int plen = m->m_pkthdr.len; 1433 struct in6_addr src0, dst0; /* copies for local work */ 1434 u_int32_t iszone, idzone, oszone, odzone; 1435 int error = 0; 1436 1437/* 1438 * Macro to send packet on mif. Since RSVP packets don't get counted on 1439 * input, they shouldn't get counted on output, so statistics keeping is 1440 * separate. 1441 */ 1442 1443#define MC6_SEND(ip6, mifp, m) do { \ 1444 if ((mifp)->m6_flags & MIFF_REGISTER) \ 1445 register_send((ip6), (mifp), (m)); \ 1446 else \ 1447 phyint_send((ip6), (mifp), (m)); \ 1448} while (/*CONSTCOND*/ 0) 1449 1450 /* 1451 * Don't forward if it didn't arrive from the parent mif 1452 * for its origin. 1453 */ 1454 mifi = rt->mf6c_parent; 1455 if ((mifi >= nummifs) || (mif6table[mifi].m6_ifp != ifp)) { 1456 /* came in the wrong interface */ 1457#ifdef MRT6DEBUG 1458 if (V_mrt6debug & DEBUG_FORWARD) 1459 log(LOG_DEBUG, 1460 "wrong if: ifid %d mifi %d mififid %x\n", 1461 ifp->if_index, mifi, 1462 mif6table[mifi].m6_ifp->if_index); 1463#endif 1464 mrt6stat.mrt6s_wrong_if++; 1465 rt->mf6c_wrong_if++; 1466 /* 1467 * If we are doing PIM processing, and we are forwarding 1468 * packets on this interface, send a message to the 1469 * routing daemon. 1470 */ 1471 /* have to make sure this is a valid mif */ 1472 if (mifi < nummifs && mif6table[mifi].m6_ifp) 1473 if (V_pim6 && (m->m_flags & M_LOOP) == 0) { 1474 /* 1475 * Check the M_LOOP flag to avoid an 1476 * unnecessary PIM assert. 1477 * XXX: M_LOOP is an ad-hoc hack... 1478 */ 1479 static struct sockaddr_in6 sin6 = 1480 { sizeof(sin6), AF_INET6 }; 1481 1482 struct mbuf *mm; 1483 struct mrt6msg *im; 1484#ifdef MRT6_OINIT 1485 struct omrt6msg *oim; 1486#endif 1487 1488 mm = m_copy(m, 0, sizeof(struct ip6_hdr)); 1489 if (mm && 1490 (M_HASCL(mm) || 1491 mm->m_len < sizeof(struct ip6_hdr))) 1492 mm = m_pullup(mm, sizeof(struct ip6_hdr)); 1493 if (mm == NULL) 1494 return (ENOBUFS); 1495 1496#ifdef MRT6_OINIT 1497 oim = NULL; 1498#endif 1499 im = NULL; 1500 switch (V_ip6_mrouter_ver) { 1501#ifdef MRT6_OINIT 1502 case MRT6_OINIT: 1503 oim = mtod(mm, struct omrt6msg *); 1504 oim->im6_msgtype = MRT6MSG_WRONGMIF; 1505 oim->im6_mbz = 0; 1506 break; 1507#endif 1508 case MRT6_INIT: 1509 im = mtod(mm, struct mrt6msg *); 1510 im->im6_msgtype = MRT6MSG_WRONGMIF; 1511 im->im6_mbz = 0; 1512 break; 1513 default: 1514 m_freem(mm); 1515 return (EINVAL); 1516 } 1517 1518 for (mifp = mif6table, iif = 0; 1519 iif < nummifs && mifp && 1520 mifp->m6_ifp != ifp; 1521 mifp++, iif++) 1522 ; 1523 1524 switch (V_ip6_mrouter_ver) { 1525#ifdef MRT6_OINIT 1526 case MRT6_OINIT: 1527 oim->im6_mif = iif; 1528 sin6.sin6_addr = oim->im6_src; 1529 break; 1530#endif 1531 case MRT6_INIT: 1532 im->im6_mif = iif; 1533 sin6.sin6_addr = im->im6_src; 1534 break; 1535 } 1536 1537 mrt6stat.mrt6s_upcalls++; 1538 1539 if (socket_send(V_ip6_mrouter, mm, &sin6) < 0) { 1540#ifdef MRT6DEBUG 1541 if (V_mrt6debug) 1542 log(LOG_WARNING, "mdq, ip6_mrouter socket queue full\n"); 1543#endif 1544 ++mrt6stat.mrt6s_upq_sockfull; 1545 return (ENOBUFS); 1546 } /* if socket Q full */ 1547 } /* if PIM */ 1548 return (0); 1549 } /* if wrong iif */ 1550 1551 /* If I sourced this packet, it counts as output, else it was input. */ 1552 if (m->m_pkthdr.rcvif == NULL) { 1553 /* XXX: is rcvif really NULL when output?? */ 1554 mif6table[mifi].m6_pkt_out++; 1555 mif6table[mifi].m6_bytes_out += plen; 1556 } else { 1557 mif6table[mifi].m6_pkt_in++; 1558 mif6table[mifi].m6_bytes_in += plen; 1559 } 1560 rt->mf6c_pkt_cnt++; 1561 rt->mf6c_byte_cnt += plen; 1562 1563 /* 1564 * For each mif, forward a copy of the packet if there are group 1565 * members downstream on the interface. 1566 */ 1567 src0 = ip6->ip6_src; 1568 dst0 = ip6->ip6_dst; 1569 if ((error = in6_setscope(&src0, ifp, &iszone)) != 0 || 1570 (error = in6_setscope(&dst0, ifp, &idzone)) != 0) { 1571 V_ip6stat.ip6s_badscope++; 1572 return (error); 1573 } 1574 for (mifp = mif6table, mifi = 0; mifi < nummifs; mifp++, mifi++) { 1575 if (IF_ISSET(mifi, &rt->mf6c_ifset)) { 1576 /* 1577 * check if the outgoing packet is going to break 1578 * a scope boundary. 1579 * XXX For packets through PIM register tunnel 1580 * interface, we believe a routing daemon. 1581 */ 1582 if (!(mif6table[rt->mf6c_parent].m6_flags & 1583 MIFF_REGISTER) && 1584 !(mif6table[mifi].m6_flags & MIFF_REGISTER)) { 1585 if (in6_setscope(&src0, mif6table[mifi].m6_ifp, 1586 &oszone) || 1587 in6_setscope(&dst0, mif6table[mifi].m6_ifp, 1588 &odzone) || 1589 iszone != oszone || 1590 idzone != odzone) { 1591 V_ip6stat.ip6s_badscope++; 1592 continue; 1593 } 1594 } 1595 1596 mifp->m6_pkt_out++; 1597 mifp->m6_bytes_out += plen; 1598 MC6_SEND(ip6, mifp, m); 1599 } 1600 } 1601 return (0); 1602} 1603 1604static void 1605phyint_send(struct ip6_hdr *ip6, struct mif6 *mifp, struct mbuf *m) 1606{ 1607 INIT_VNET_INET6(curvnet); 1608 struct mbuf *mb_copy; 1609 struct ifnet *ifp = mifp->m6_ifp; 1610 int error = 0; 1611 struct sockaddr_in6 *dst6; 1612 u_long linkmtu; 1613 1614 dst6 = &mifp->m6_route.ro_dst; 1615 1616 /* 1617 * Make a new reference to the packet; make sure that 1618 * the IPv6 header is actually copied, not just referenced, 1619 * so that ip6_output() only scribbles on the copy. 1620 */ 1621 mb_copy = m_copy(m, 0, M_COPYALL); 1622 if (mb_copy && 1623 (M_HASCL(mb_copy) || mb_copy->m_len < sizeof(struct ip6_hdr))) 1624 mb_copy = m_pullup(mb_copy, sizeof(struct ip6_hdr)); 1625 if (mb_copy == NULL) { 1626 return; 1627 } 1628 /* set MCAST flag to the outgoing packet */ 1629 mb_copy->m_flags |= M_MCAST; 1630 1631 /* 1632 * If we sourced the packet, call ip6_output since we may devide 1633 * the packet into fragments when the packet is too big for the 1634 * outgoing interface. 1635 * Otherwise, we can simply send the packet to the interface 1636 * sending queue. 1637 */ 1638 if (m->m_pkthdr.rcvif == NULL) { 1639 struct ip6_moptions im6o; 1640 1641 im6o.im6o_multicast_ifp = ifp; 1642 /* XXX: ip6_output will override ip6->ip6_hlim */ 1643 im6o.im6o_multicast_hlim = ip6->ip6_hlim; 1644 im6o.im6o_multicast_loop = 1; 1645 error = ip6_output(mb_copy, NULL, &mifp->m6_route, 1646 IPV6_FORWARDING, &im6o, NULL, NULL); 1647 1648#ifdef MRT6DEBUG 1649 if (V_mrt6debug & DEBUG_XMIT) 1650 log(LOG_DEBUG, "phyint_send on mif %d err %d\n", 1651 mifp - mif6table, error); 1652#endif 1653 return; 1654 } 1655 1656 /* 1657 * If configured to loop back multicasts by default, 1658 * loop back a copy now. 1659 */ 1660 if (in6_mcast_loop) { 1661 dst6->sin6_len = sizeof(struct sockaddr_in6); 1662 dst6->sin6_family = AF_INET6; 1663 dst6->sin6_addr = ip6->ip6_dst; 1664 ip6_mloopback(ifp, m, &mifp->m6_route.ro_dst); 1665 } 1666 1667 /* 1668 * Put the packet into the sending queue of the outgoing interface 1669 * if it would fit in the MTU of the interface. 1670 */ 1671 linkmtu = IN6_LINKMTU(ifp); 1672 if (mb_copy->m_pkthdr.len <= linkmtu || linkmtu < IPV6_MMTU) { 1673 dst6->sin6_len = sizeof(struct sockaddr_in6); 1674 dst6->sin6_family = AF_INET6; 1675 dst6->sin6_addr = ip6->ip6_dst; 1676 /* 1677 * We just call if_output instead of nd6_output here, since 1678 * we need no ND for a multicast forwarded packet...right? 1679 */ 1680 error = (*ifp->if_output)(ifp, mb_copy, 1681 (struct sockaddr *)&mifp->m6_route.ro_dst, NULL); 1682#ifdef MRT6DEBUG 1683 if (V_mrt6debug & DEBUG_XMIT) 1684 log(LOG_DEBUG, "phyint_send on mif %d err %d\n", 1685 mifp - mif6table, error); 1686#endif 1687 } else { 1688 /* 1689 * pMTU discovery is intentionally disabled by default, since 1690 * various router may notify pMTU in multicast, which can be 1691 * a DDoS to a router 1692 */ 1693 if (V_ip6_mcast_pmtu) 1694 icmp6_error(mb_copy, ICMP6_PACKET_TOO_BIG, 0, linkmtu); 1695 else { 1696#ifdef MRT6DEBUG 1697 if (V_mrt6debug & DEBUG_XMIT) { 1698 char ip6bufs[INET6_ADDRSTRLEN]; 1699 char ip6bufd[INET6_ADDRSTRLEN]; 1700 log(LOG_DEBUG, 1701 "phyint_send: packet too big on %s o %s " 1702 "g %s size %d(discarded)\n", 1703 if_name(ifp), 1704 ip6_sprintf(ip6bufs, &ip6->ip6_src), 1705 ip6_sprintf(ip6bufd, &ip6->ip6_dst), 1706 mb_copy->m_pkthdr.len); 1707 } 1708#endif /* MRT6DEBUG */ 1709 m_freem(mb_copy); /* simply discard the packet */ 1710 } 1711 } 1712} 1713 1714static int 1715register_send(struct ip6_hdr *ip6, struct mif6 *mif, struct mbuf *m) 1716{ 1717 INIT_VNET_INET6(curvnet); 1718 struct mbuf *mm; 1719 int i, len = m->m_pkthdr.len; 1720 static struct sockaddr_in6 sin6 = { sizeof(sin6), AF_INET6 }; 1721 struct mrt6msg *im6; 1722 1723#ifdef MRT6DEBUG 1724 if (V_mrt6debug) { 1725 char ip6bufs[INET6_ADDRSTRLEN], ip6bufd[INET6_ADDRSTRLEN]; 1726 log(LOG_DEBUG, "** IPv6 register_send **\n src %s dst %s\n", 1727 ip6_sprintf(ip6bufs, &ip6->ip6_src), 1728 ip6_sprintf(ip6bufd, &ip6->ip6_dst)); 1729 } 1730#endif 1731 ++pim6stat.pim6s_snd_registers; 1732 1733 /* Make a copy of the packet to send to the user level process */ 1734 MGETHDR(mm, M_DONTWAIT, MT_HEADER); 1735 if (mm == NULL) 1736 return (ENOBUFS); 1737 mm->m_pkthdr.rcvif = NULL; 1738 mm->m_data += max_linkhdr; 1739 mm->m_len = sizeof(struct ip6_hdr); 1740 1741 if ((mm->m_next = m_copy(m, 0, M_COPYALL)) == NULL) { 1742 m_freem(mm); 1743 return (ENOBUFS); 1744 } 1745 i = MHLEN - M_LEADINGSPACE(mm); 1746 if (i > len) 1747 i = len; 1748 mm = m_pullup(mm, i); 1749 if (mm == NULL) 1750 return (ENOBUFS); 1751/* TODO: check it! */ 1752 mm->m_pkthdr.len = len + sizeof(struct ip6_hdr); 1753 1754 /* 1755 * Send message to routing daemon 1756 */ 1757 sin6.sin6_addr = ip6->ip6_src; 1758 1759 im6 = mtod(mm, struct mrt6msg *); 1760 im6->im6_msgtype = MRT6MSG_WHOLEPKT; 1761 im6->im6_mbz = 0; 1762 1763 im6->im6_mif = mif - mif6table; 1764 1765 /* iif info is not given for reg. encap.n */ 1766 mrt6stat.mrt6s_upcalls++; 1767 1768 if (socket_send(V_ip6_mrouter, mm, &sin6) < 0) { 1769#ifdef MRT6DEBUG 1770 if (V_mrt6debug) 1771 log(LOG_WARNING, 1772 "register_send: ip6_mrouter socket queue full\n"); 1773#endif 1774 ++mrt6stat.mrt6s_upq_sockfull; 1775 return (ENOBUFS); 1776 } 1777 return (0); 1778} 1779 1780/* 1781 * pim6_encapcheck() is called by the encap6_input() path at runtime to 1782 * determine if a packet is for PIM; allowing PIM to be dynamically loaded 1783 * into the kernel. 1784 */ 1785static int 1786pim6_encapcheck(const struct mbuf *m, int off, int proto, void *arg) 1787{ 1788 1789#ifdef DIAGNOSTIC 1790 KASSERT(proto == IPPROTO_PIM, ("not for IPPROTO_PIM")); 1791#endif 1792 if (proto != IPPROTO_PIM) 1793 return 0; /* not for us; reject the datagram. */ 1794 1795 return 64; /* claim the datagram. */ 1796} 1797 1798/* 1799 * PIM sparse mode hook 1800 * Receives the pim control messages, and passes them up to the listening 1801 * socket, using rip6_input. 1802 * The only message processed is the REGISTER pim message; the pim header 1803 * is stripped off, and the inner packet is passed to register_mforward. 1804 */ 1805int 1806pim6_input(struct mbuf **mp, int *offp, int proto) 1807{ 1808 INIT_VNET_INET6(curvnet); 1809 struct pim *pim; /* pointer to a pim struct */ 1810 struct ip6_hdr *ip6; 1811 int pimlen; 1812 struct mbuf *m = *mp; 1813 int minlen; 1814 int off = *offp; 1815 1816 ++pim6stat.pim6s_rcv_total; 1817 1818 ip6 = mtod(m, struct ip6_hdr *); 1819 pimlen = m->m_pkthdr.len - *offp; 1820 1821 /* 1822 * Validate lengths 1823 */ 1824 if (pimlen < PIM_MINLEN) { 1825 ++pim6stat.pim6s_rcv_tooshort; 1826#ifdef MRT6DEBUG 1827 if (V_mrt6debug & DEBUG_PIM) 1828 log(LOG_DEBUG,"pim6_input: PIM packet too short\n"); 1829#endif 1830 m_freem(m); 1831 return (IPPROTO_DONE); 1832 } 1833 1834 /* 1835 * if the packet is at least as big as a REGISTER, go ahead 1836 * and grab the PIM REGISTER header size, to avoid another 1837 * possible m_pullup() later. 1838 * 1839 * PIM_MINLEN == pimhdr + u_int32 == 8 1840 * PIM6_REG_MINLEN == pimhdr + reghdr + eip6hdr == 4 + 4 + 40 1841 */ 1842 minlen = (pimlen >= PIM6_REG_MINLEN) ? PIM6_REG_MINLEN : PIM_MINLEN; 1843 1844 /* 1845 * Make sure that the IP6 and PIM headers in contiguous memory, and 1846 * possibly the PIM REGISTER header 1847 */ 1848#ifndef PULLDOWN_TEST 1849 IP6_EXTHDR_CHECK(m, off, minlen, IPPROTO_DONE); 1850 /* adjust pointer */ 1851 ip6 = mtod(m, struct ip6_hdr *); 1852 1853 /* adjust mbuf to point to the PIM header */ 1854 pim = (struct pim *)((caddr_t)ip6 + off); 1855#else 1856 IP6_EXTHDR_GET(pim, struct pim *, m, off, minlen); 1857 if (pim == NULL) { 1858 pim6stat.pim6s_rcv_tooshort++; 1859 return (IPPROTO_DONE); 1860 } 1861#endif 1862 1863#define PIM6_CHECKSUM 1864#ifdef PIM6_CHECKSUM 1865 { 1866 int cksumlen; 1867 1868 /* 1869 * Validate checksum. 1870 * If PIM REGISTER, exclude the data packet 1871 */ 1872 if (pim->pim_type == PIM_REGISTER) 1873 cksumlen = PIM_MINLEN; 1874 else 1875 cksumlen = pimlen; 1876 1877 if (in6_cksum(m, IPPROTO_PIM, off, cksumlen)) { 1878 ++pim6stat.pim6s_rcv_badsum; 1879#ifdef MRT6DEBUG 1880 if (V_mrt6debug & DEBUG_PIM) 1881 log(LOG_DEBUG, 1882 "pim6_input: invalid checksum\n"); 1883#endif 1884 m_freem(m); 1885 return (IPPROTO_DONE); 1886 } 1887 } 1888#endif /* PIM_CHECKSUM */ 1889 1890 /* PIM version check */ 1891 if (pim->pim_ver != PIM_VERSION) { 1892 ++pim6stat.pim6s_rcv_badversion; 1893#ifdef MRT6DEBUG 1894 log(LOG_ERR, 1895 "pim6_input: incorrect version %d, expecting %d\n", 1896 pim->pim_ver, PIM_VERSION); 1897#endif 1898 m_freem(m); 1899 return (IPPROTO_DONE); 1900 } 1901 1902 if (pim->pim_type == PIM_REGISTER) { 1903 /* 1904 * since this is a REGISTER, we'll make a copy of the register 1905 * headers ip6+pim+u_int32_t+encap_ip6, to be passed up to the 1906 * routing daemon. 1907 */ 1908 static struct sockaddr_in6 dst = { sizeof(dst), AF_INET6 }; 1909 1910 struct mbuf *mcp; 1911 struct ip6_hdr *eip6; 1912 u_int32_t *reghdr; 1913 int rc; 1914#ifdef MRT6DEBUG 1915 char ip6bufs[INET6_ADDRSTRLEN], ip6bufd[INET6_ADDRSTRLEN]; 1916#endif 1917 1918 ++pim6stat.pim6s_rcv_registers; 1919 1920 if ((reg_mif_num >= nummifs) || (reg_mif_num == (mifi_t) -1)) { 1921#ifdef MRT6DEBUG 1922 if (V_mrt6debug & DEBUG_PIM) 1923 log(LOG_DEBUG, 1924 "pim6_input: register mif not set: %d\n", 1925 reg_mif_num); 1926#endif 1927 m_freem(m); 1928 return (IPPROTO_DONE); 1929 } 1930 1931 reghdr = (u_int32_t *)(pim + 1); 1932 1933 if ((ntohl(*reghdr) & PIM_NULL_REGISTER)) 1934 goto pim6_input_to_daemon; 1935 1936 /* 1937 * Validate length 1938 */ 1939 if (pimlen < PIM6_REG_MINLEN) { 1940 ++pim6stat.pim6s_rcv_tooshort; 1941 ++pim6stat.pim6s_rcv_badregisters; 1942#ifdef MRT6DEBUG 1943 log(LOG_ERR, 1944 "pim6_input: register packet size too " 1945 "small %d from %s\n", 1946 pimlen, ip6_sprintf(ip6bufs, &ip6->ip6_src)); 1947#endif 1948 m_freem(m); 1949 return (IPPROTO_DONE); 1950 } 1951 1952 eip6 = (struct ip6_hdr *) (reghdr + 1); 1953#ifdef MRT6DEBUG 1954 if (V_mrt6debug & DEBUG_PIM) 1955 log(LOG_DEBUG, 1956 "pim6_input[register], eip6: %s -> %s, " 1957 "eip6 plen %d\n", 1958 ip6_sprintf(ip6bufs, &eip6->ip6_src), 1959 ip6_sprintf(ip6bufd, &eip6->ip6_dst), 1960 ntohs(eip6->ip6_plen)); 1961#endif 1962 1963 /* verify the version number of the inner packet */ 1964 if ((eip6->ip6_vfc & IPV6_VERSION_MASK) != IPV6_VERSION) { 1965 ++pim6stat.pim6s_rcv_badregisters; 1966#ifdef MRT6DEBUG 1967 log(LOG_DEBUG, "pim6_input: invalid IP version (%d) " 1968 "of the inner packet\n", 1969 (eip6->ip6_vfc & IPV6_VERSION)); 1970#endif 1971 m_freem(m); 1972 return (IPPROTO_NONE); 1973 } 1974 1975 /* verify the inner packet is destined to a mcast group */ 1976 if (!IN6_IS_ADDR_MULTICAST(&eip6->ip6_dst)) { 1977 ++pim6stat.pim6s_rcv_badregisters; 1978#ifdef MRT6DEBUG 1979 if (V_mrt6debug & DEBUG_PIM) 1980 log(LOG_DEBUG, 1981 "pim6_input: inner packet of register " 1982 "is not multicast %s\n", 1983 ip6_sprintf(ip6bufd, &eip6->ip6_dst)); 1984#endif 1985 m_freem(m); 1986 return (IPPROTO_DONE); 1987 } 1988 1989 /* 1990 * make a copy of the whole header to pass to the daemon later. 1991 */ 1992 mcp = m_copy(m, 0, off + PIM6_REG_MINLEN); 1993 if (mcp == NULL) { 1994#ifdef MRT6DEBUG 1995 log(LOG_ERR, 1996 "pim6_input: pim register: " 1997 "could not copy register head\n"); 1998#endif 1999 m_freem(m); 2000 return (IPPROTO_DONE); 2001 } 2002 2003 /* 2004 * forward the inner ip6 packet; point m_data at the inner ip6. 2005 */ 2006 m_adj(m, off + PIM_MINLEN); 2007#ifdef MRT6DEBUG 2008 if (V_mrt6debug & DEBUG_PIM) { 2009 log(LOG_DEBUG, 2010 "pim6_input: forwarding decapsulated register: " 2011 "src %s, dst %s, mif %d\n", 2012 ip6_sprintf(ip6bufs, &eip6->ip6_src), 2013 ip6_sprintf(ip6bufd, &eip6->ip6_dst), 2014 reg_mif_num); 2015 } 2016#endif 2017 2018 rc = if_simloop(mif6table[reg_mif_num].m6_ifp, m, 2019 dst.sin6_family, 0); 2020 2021 /* prepare the register head to send to the mrouting daemon */ 2022 m = mcp; 2023 } 2024 2025 /* 2026 * Pass the PIM message up to the daemon; if it is a register message 2027 * pass the 'head' only up to the daemon. This includes the 2028 * encapsulator ip6 header, pim header, register header and the 2029 * encapsulated ip6 header. 2030 */ 2031 pim6_input_to_daemon: 2032 rip6_input(&m, offp, proto); 2033 return (IPPROTO_DONE); 2034} 2035 2036static int 2037ip6_mroute_modevent(module_t mod, int type, void *unused) 2038{ 2039 INIT_VNET_INET6(curvnet); 2040 2041 switch (type) { 2042 case MOD_LOAD: 2043 MROUTER6_LOCK_INIT(); 2044 MFC6_LOCK_INIT(); 2045 MIF6_LOCK_INIT(); 2046 2047 pim6_encap_cookie = encap_attach_func(AF_INET6, IPPROTO_PIM, 2048 pim6_encapcheck, 2049 (const struct protosw *)&in6_pim_protosw, NULL); 2050 if (pim6_encap_cookie == NULL) { 2051 printf("ip6_mroute: unable to attach pim6 encap\n"); 2052 MIF6_LOCK_DESTROY(); 2053 MFC6_LOCK_DESTROY(); 2054 MROUTER6_LOCK_DESTROY(); 2055 return (EINVAL); 2056 } 2057 2058 ip6_mforward = X_ip6_mforward; 2059 ip6_mrouter_done = X_ip6_mrouter_done; 2060 ip6_mrouter_get = X_ip6_mrouter_get; 2061 ip6_mrouter_set = X_ip6_mrouter_set; 2062 mrt6_ioctl = X_mrt6_ioctl; 2063 break; 2064 2065 case MOD_UNLOAD: 2066 if (V_ip6_mrouter != NULL) 2067 return EINVAL; 2068 2069 if (pim6_encap_cookie) { 2070 encap_detach(pim6_encap_cookie); 2071 pim6_encap_cookie = NULL; 2072 } 2073 X_ip6_mrouter_done(); 2074 ip6_mforward = NULL; 2075 ip6_mrouter_done = NULL; 2076 ip6_mrouter_get = NULL; 2077 ip6_mrouter_set = NULL; 2078 mrt6_ioctl = NULL; 2079 2080 MIF6_LOCK_DESTROY(); 2081 MFC6_LOCK_DESTROY(); 2082 MROUTER6_LOCK_DESTROY(); 2083 break; 2084 2085 default: 2086 return (EOPNOTSUPP); 2087 } 2088 2089 return (0); 2090} 2091 2092static moduledata_t ip6_mroutemod = { 2093 "ip6_mroute", 2094 ip6_mroute_modevent, 2095 0 2096}; 2097 2098DECLARE_MODULE(ip6_mroute, ip6_mroutemod, SI_SUB_PSEUDO, SI_ORDER_ANY); 2099