X-Git-Url: https://git.saurik.com/apple/xnu.git/blobdiff_plain/cf7d32b81c573a0536dc4da4157f9c26f8d0bed3..d190cdc3f5544636abb56dc1874be391d3e1b148:/bsd/netinet6/icmp6.c diff --git a/bsd/netinet6/icmp6.c b/bsd/netinet6/icmp6.c index 11b390413..23a9178d3 100644 --- a/bsd/netinet6/icmp6.c +++ b/bsd/netinet6/icmp6.c @@ -1,3 +1,31 @@ +/* + * Copyright (c) 2000-2016 Apple Inc. All rights reserved. + * + * @APPLE_OSREFERENCE_LICENSE_HEADER_START@ + * + * This file contains Original Code and/or Modifications of Original Code + * as defined in and that are subject to the Apple Public Source License + * Version 2.0 (the 'License'). You may not use this file except in + * compliance with the License. The rights granted to you under the License + * may not be used to create, or enable the creation or redistribution of, + * unlawful or unlicensed copies of an Apple operating system, or to + * circumvent, violate, or enable the circumvention or violation of, any + * terms of an Apple operating system software license agreement. + * + * Please obtain a copy of the License at + * http://www.opensource.apple.com/apsl/ and read it before using this file. + * + * The Original Code and all software distributed under the License are + * distributed on an 'AS IS' basis, WITHOUT WARRANTY OF ANY KIND, EITHER + * EXPRESS OR IMPLIED, AND APPLE HEREBY DISCLAIMS ALL SUCH WARRANTIES, + * INCLUDING WITHOUT LIMITATION, ANY WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE, QUIET ENJOYMENT OR NON-INFRINGEMENT. + * Please see the License for the specific language governing rights and + * limitations under the License. + * + * @APPLE_OSREFERENCE_LICENSE_HEADER_END@ + */ + /* $FreeBSD: src/sys/netinet6/icmp6.c,v 1.6.2.6 2001/07/10 09:44:16 ume Exp $ */ /* $KAME: icmp6.c,v 1.211 2001/04/04 05:56:20 itojun Exp $ */ @@ -70,6 +98,7 @@ #include #include #include +#include #include #include #include @@ -78,6 +107,7 @@ #include #include #include +#include #include #include @@ -92,69 +122,75 @@ #include #include #include +#include #include #include #include +#include #if IPSEC #include #include - -extern int ipsec_bypass; #endif #include -extern struct domain inet6domain; -extern struct ip6protosw inet6sw[]; +#if NECP +#include +#endif + extern struct ip6protosw *ip6_protox[]; -extern u_long rip_sendspace; -extern u_long rip_recvspace; +extern uint32_t rip_sendspace; +extern uint32_t rip_recvspace; struct icmp6stat icmp6stat; extern struct inpcbhead ripcb; extern int icmp6errppslim; +extern int icmp6rappslim; static int icmp6errpps_count = 0; +static int icmp6rapps_count = 0; static struct timeval icmp6errppslim_last; +static struct timeval icmp6rappslim_last; extern int icmp6_nodeinfo; extern struct inpcbinfo ripcbinfo; -extern lck_mtx_t *ip6_mutex; -extern lck_mtx_t *nd6_mutex; static void icmp6_errcount(struct icmp6errstat *, int, int); static int icmp6_rip6_input(struct mbuf **, int); static int icmp6_ratelimit(const struct in6_addr *, const int, const int); static const char *icmp6_redirect_diag(struct in6_addr *, struct in6_addr *, struct in6_addr *); -#ifndef HAVE_PPSRATECHECK -static int ppsratecheck(struct timeval *, int *, int); -#endif static struct mbuf *ni6_input(struct mbuf *, int); static struct mbuf *ni6_nametodns(const char *, int, int); static int ni6_dnsmatch(const char *, int, const char *, int); -static int ni6_addrs(struct icmp6_nodeinfo *, +static int ni6_addrs(struct icmp6_nodeinfo *, struct ifnet **, char *); static int ni6_store_addrs(struct icmp6_nodeinfo *, struct icmp6_nodeinfo *, struct ifnet *, int); static int icmp6_notify_error(struct mbuf *, int, int, int); -#ifdef COMPAT_RFC1885 -static struct route_in6 icmp6_reflect_rt; -#endif void -icmp6_init() +icmp6_init(struct ip6protosw *pp, struct domain *dp) { - mld6_init(); +#pragma unused(dp) + static int icmp6_initialized = 0; + + /* Also called from ip6_init() without pp */ + VERIFY(pp == NULL || + (pp->pr_flags & (PR_INITIALIZED|PR_ATTACHED)) == PR_ATTACHED); + + /* This gets called by more than one protocols, so initialize once */ + if (!icmp6_initialized) { + icmp6_initialized = 1; + mld_init(); + } } static void -icmp6_errcount(stat, type, code) - struct icmp6errstat *stat; - int type, code; +icmp6_errcount(struct icmp6errstat *stat, int type, int code) { switch (type) { case ICMP6_DST_UNREACH: @@ -209,13 +245,48 @@ icmp6_errcount(stat, type, code) stat->icp6errs_unknown++; } +/* + * A wrapper function for icmp6_error() necessary when the erroneous packet + * may not contain enough scope zone information. + */ +void +icmp6_error2(struct mbuf *m, int type, int code, int param, + struct ifnet *ifp) +{ + struct ip6_hdr *ip6; + + if (ifp == NULL) + return; + +#ifndef PULLDOWN_TEST + IP6_EXTHDR_CHECK(m, 0, sizeof(struct ip6_hdr),return ); +#else + if (m->m_len < sizeof(struct ip6_hdr)) { + m = m_pullup(m, sizeof(struct ip6_hdr)); + if (m == NULL) + return; + } +#endif + + ip6 = mtod(m, struct ip6_hdr *); + + if (in6_setscope(&ip6->ip6_src, ifp, NULL) != 0) + return; + if (in6_setscope(&ip6->ip6_dst, ifp, NULL) != 0) + return; + + icmp6_error(m, type, code, param); +} + /* * Generate an error packet of type error in response to bad IP6 packet. */ void -icmp6_error(m, type, code, param) - struct mbuf *m; - int type, code, param; +icmp6_error(struct mbuf *m, int type, int code, int param) { + icmp6_error_flag(m, type, code, param, ICMP6_ERROR_RST_MRCVIF); +} + +void icmp6_error_flag (struct mbuf *m, int type, int code, int param, int flags) { struct ip6_hdr *oip6, *nip6; struct icmp6_hdr *icmp6; @@ -225,7 +296,6 @@ icmp6_error(m, type, code, param) icmp6stat.icp6s_error++; - lck_mtx_assert(ip6_mutex, LCK_MTX_ASSERT_NOTOWNED); /* count per-type-code statistics */ icmp6_errcount(&icmp6stat.icp6s_outerrhist, type, code); @@ -248,9 +318,15 @@ icmp6_error(m, type, code, param) oip6 = mtod(m, struct ip6_hdr *); /* - * Multicast destination check. For unrecognized option errors, - * this check has already done in ip6_unknown_opt(), so we can - * check only for other errors. + * If the destination address of the erroneous packet is a multicast + * address, or the packet was sent using link-layer multicast, + * we should basically suppress sending an error (RFC 2463, Section + * 2.4). + * We have two exceptions (the item e.2 in that section): + * - the Pakcet Too Big message can be sent for path MTU discovery. + * - the Parameter Problem Message that can be allowed an icmp6 error + * in the option type field. This check has been done in + * ip6_unknown_opt(), so we can just check the type and code. */ if ((m->m_flags & (M_BCAST|M_MCAST) || IN6_IS_ADDR_MULTICAST(&oip6->ip6_dst)) && @@ -259,7 +335,10 @@ icmp6_error(m, type, code, param) code != ICMP6_PARAMPROB_OPTION))) goto freeit; - /* Source address check. XXX: the case of anycast source? */ + /* + * RFC 2463, 2.4 (e.5): source address check. + * XXX: the case of anycast source? + */ if (IN6_IS_ADDR_UNSPECIFIED(&oip6->ip6_src) || IN6_IS_ADDR_MULTICAST(&oip6->ip6_src)) goto freeit; @@ -316,7 +395,7 @@ icmp6_error(m, type, code, param) m_adj(m, ICMPV6_PLD_MAXLEN - m->m_pkthdr.len); preplen = sizeof(struct ip6_hdr) + sizeof(struct icmp6_hdr); - M_PREPEND(m, preplen, M_DONTWAIT); + M_PREPEND(m, preplen, M_DONTWAIT, 1); if (m && m->m_len < preplen) m = m_pullup(m, preplen); if (m == NULL) { @@ -328,10 +407,8 @@ icmp6_error(m, type, code, param) nip6->ip6_src = oip6->ip6_src; nip6->ip6_dst = oip6->ip6_dst; - if (IN6_IS_SCOPE_LINKLOCAL(&oip6->ip6_src)) - oip6->ip6_src.s6_addr16[1] = 0; - if (IN6_IS_SCOPE_LINKLOCAL(&oip6->ip6_dst)) - oip6->ip6_dst.s6_addr16[1] = 0; + in6_clearscope(&oip6->ip6_src); + in6_clearscope(&oip6->ip6_dst); icmp6 = (struct icmp6_hdr *)(nip6 + 1); icmp6->icmp6_type = type; @@ -340,12 +417,14 @@ icmp6_error(m, type, code, param) /* * icmp6_reflect() is designed to be in the input path. - * icmp6_error() can be called from both input and outut path, + * icmp6_error() can be called from both input and output path, * and if we are in output path rcvif could contain bogus value. * clear m->m_pkthdr.rcvif for safety, we should have enough scope * information in ip header (nip6). */ - m->m_pkthdr.rcvif = NULL; + if (flags & ICMP6_ERROR_RST_MRCVIF) { + m->m_pkthdr.rcvif = NULL; + } icmp6stat.icp6s_outhist[type]++; icmp6_reflect(m, sizeof(struct ip6_hdr)); /* header order: IPv6 - ICMPv6 */ @@ -354,7 +433,7 @@ icmp6_error(m, type, code, param) freeit: /* - * If we can't tell wheter or not we can generate ICMP6, free it. + * If we can't tell whether or not we can generate ICMP6, free it. */ m_freem(m); } @@ -363,36 +442,37 @@ icmp6_error(m, type, code, param) * Process a received ICMP6 message. */ int -icmp6_input(mp, offp) - struct mbuf **mp; - int *offp; +icmp6_input(struct mbuf **mp, int *offp, int proto) { +#pragma unused(proto) struct mbuf *m = *mp, *n; + struct ifnet *ifp; struct ip6_hdr *ip6, *nip6; struct icmp6_hdr *icmp6, *nicmp6; int off = *offp; int icmp6len = m->m_pkthdr.len - *offp; - int code, sum, noff; + int code, sum, noff, proxy = 0; + + ifp = m->m_pkthdr.rcvif; #ifndef PULLDOWN_TEST IP6_EXTHDR_CHECK(m, off, sizeof(struct icmp6_hdr), return IPPROTO_DONE); /* m might change if M_LOOP. So, call mtod after this */ #endif + /* Expect 32-bit aligned data pointer on strict-align platforms */ + MBUF_STRICT_DATA_ALIGNMENT_CHECK_32(m); + /* * Locate icmp6 structure in mbuf, and check * that not corrupted and of at least minimum length */ - ip6 = mtod(m, struct ip6_hdr *); if (icmp6len < sizeof(struct icmp6_hdr)) { icmp6stat.icp6s_tooshort++; goto freeit; } - /* - * calculate the checksum - */ #ifndef PULLDOWN_TEST icmp6 = (struct icmp6_hdr *)((caddr_t)ip6 + off); #else @@ -404,6 +484,61 @@ icmp6_input(mp, offp) #endif code = icmp6->icmp6_code; + /* + * Early check for RFC 6980 + * Drop certain NDP packets if they came in fragmented + */ + switch (icmp6->icmp6_type) { + case ND_ROUTER_SOLICIT: + case ND_ROUTER_ADVERT: + case ND_NEIGHBOR_SOLICIT: + case ND_NEIGHBOR_ADVERT: + case ND_REDIRECT: + if (m->m_pkthdr.pkt_flags & PKTF_REASSEMBLED) { + icmp6stat.icp6s_rfc6980_drop++; + goto freeit; + } + break; + default: + break; + } + + /* Apply rate limit before checksum validation. */ + if (icmp6_ratelimit(&ip6->ip6_dst, icmp6->icmp6_type, code)) { + icmp6stat.icp6s_toofreq++; + goto freeit; + } + + /* + * Check multicast group membership. + * Note: SSM filters are not applied for ICMPv6 traffic. + */ + if (IN6_IS_ADDR_MULTICAST(&ip6->ip6_dst)) { + struct in6_multi *inm; + + in6_multihead_lock_shared(); + IN6_LOOKUP_MULTI(&ip6->ip6_dst, ifp, inm); + in6_multihead_lock_done(); + + if (inm == NULL) { + /* + * Don't discard if this is a Neighbor Solicitation + * that needs to be proxied (see check down below.) + */ + if (!(m->m_pkthdr.pkt_flags & PKTF_PROXY_DST)) { + ip6stat.ip6s_notmember++; + in6_ifstat_inc(m->m_pkthdr.rcvif, + ifs6_in_discard); + goto freeit; + } + } else { + IN6M_REMREF(inm); + } + } + + /* + * calculate the checksum + */ if ((sum = in6_cksum(m, IPPROTO_ICMPV6, off, icmp6len)) != 0) { nd6log((LOG_ERR, "ICMP6 checksum error(%d|%x) %s\n", @@ -412,23 +547,22 @@ icmp6_input(mp, offp) goto freeit; } -#if defined(NFAITH) && 0 < NFAITH - if (faithprefix(&ip6->ip6_dst)) { + if (m->m_pkthdr.pkt_flags & PKTF_PROXY_DST) { /* - * Deliver very specific ICMP6 type only. - * This is important to deilver TOOBIG. Otherwise PMTUD - * will not work. + * This is the special case of proxying NS (dst is either + * solicited-node multicast or unicast); process it locally + * but don't deliver it to sockets. It practically lets us + * steer the packet to nd6_prproxy_ns_input, where more + * specific tests and actions will be taken. */ switch (icmp6->icmp6_type) { - case ICMP6_DST_UNREACH: - case ICMP6_PACKET_TOO_BIG: - case ICMP6_TIME_EXCEEDED: + case ND_NEIGHBOR_SOLICIT: + proxy = 1; break; default: goto freeit; } } -#endif icmp6stat.icp6s_inhist[icmp6->icmp6_type]++; icmp6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_msg); @@ -449,16 +583,10 @@ icmp6_input(mp, offp) case ICMP6_DST_UNREACH_ADDR: code = PRC_HOSTDEAD; break; -#ifdef COMPAT_RFC1885 - case ICMP6_DST_UNREACH_NOTNEIGHBOR: - code = PRC_UNREACH_SRCFAIL; - break; -#else case ICMP6_DST_UNREACH_BEYONDSCOPE: /* I mean "source address was incorrect." */ code = PRC_PARAMPROB; break; -#endif case ICMP6_DST_UNREACH_NOPORT: code = PRC_UNREACH_PORT; break; @@ -466,12 +594,9 @@ icmp6_input(mp, offp) goto badcode; } goto deliver; - break; case ICMP6_PACKET_TOO_BIG: icmp6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_pkttoobig); - if (code != 0) - goto badcode; code = PRC_MSGSIZE; @@ -480,20 +605,20 @@ icmp6_input(mp, offp) * intermediate extension headers. */ goto deliver; - break; case ICMP6_TIME_EXCEEDED: icmp6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_timeexceed); switch (code) { case ICMP6_TIME_EXCEED_TRANSIT: + code = PRC_TIMXCEED_INTRANS; + break; case ICMP6_TIME_EXCEED_REASSEMBLY: - code += PRC_TIMXCEED_INTRANS; + code = PRC_TIMXCEED_REASS; break; default: goto badcode; } goto deliver; - break; case ICMP6_PARAM_PROB: icmp6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_paramprob); @@ -509,15 +634,15 @@ icmp6_input(mp, offp) goto badcode; } goto deliver; - break; case ICMP6_ECHO_REQUEST: icmp6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_echo); if (code != 0) goto badcode; + if ((n = m_copy(m, 0, M_COPYALL)) == NULL) { /* Give up remote */ - break; + goto rate_limit_checked; } if ((n->m_flags & M_EXT) != 0 || n->m_len < off + sizeof(struct icmp6_hdr)) { @@ -531,7 +656,7 @@ icmp6_input(mp, offp) if (maxlen >= MCLBYTES) { /* Give up remote */ m_freem(n0); - break; + goto rate_limit_checked; } MGETHDR(n, M_DONTWAIT, n0->m_type); /* MAC-OK */ if (n && maxlen >= MHLEN) { @@ -544,7 +669,7 @@ icmp6_input(mp, offp) if (n == NULL) { /* Give up remote */ m_freem(n0); - break; + goto rate_limit_checked; } M_COPY_PKTHDR(n, n0); /* @@ -571,14 +696,18 @@ icmp6_input(mp, offp) sizeof(*nicmp6)); noff = off; } - nicmp6->icmp6_type = ICMP6_ECHO_REPLY; - nicmp6->icmp6_code = 0; + if(nicmp6 == NULL) + panic("nicmp6 is NULL in %s, which isn't good!\n", __FUNCTION__); + else { + nicmp6->icmp6_type = ICMP6_ECHO_REPLY; + nicmp6->icmp6_code = 0; + } if (n) { icmp6stat.icp6s_reflect++; icmp6stat.icp6s_outhist[ICMP6_ECHO_REPLY]++; icmp6_reflect(n, noff); } - break; + goto rate_limit_checked; case ICMP6_ECHO_REPLY: icmp6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_echoreply); @@ -586,41 +715,44 @@ icmp6_input(mp, offp) goto badcode; break; - case MLD6_LISTENER_QUERY: - case MLD6_LISTENER_REPORT: - if (icmp6len < sizeof(struct mld6_hdr)) + case MLD_LISTENER_QUERY: + case MLD_LISTENER_REPORT: + + if (icmp6len < sizeof(struct mld_hdr)) goto badlen; - if (icmp6->icmp6_type == MLD6_LISTENER_QUERY) /* XXX: ugly... */ + if (icmp6->icmp6_type == MLD_LISTENER_QUERY) /* XXX: ugly... */ icmp6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_mldquery); else icmp6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_mldreport); + if ((n = m_copym(m, 0, M_COPYALL, M_DONTWAIT)) == NULL) { /* give up local */ - mld6_input(m, off); - m = NULL; + if (mld_input(m, off, icmp6len) == IPPROTO_DONE) + m = NULL; goto freeit; } - mld6_input(n, off); + if (mld_input(n, off, icmp6len) != IPPROTO_DONE) + m_freem(n); /* m stays. */ - break; + goto rate_limit_checked; - case MLD6_LISTENER_DONE: + case MLD_LISTENER_DONE: icmp6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_mlddone); - if (icmp6len < sizeof(struct mld6_hdr)) /* necessary? */ + if (icmp6len < sizeof(struct mld_hdr)) /* necessary? */ goto badlen; break; /* nothing to be done in kernel */ - case MLD6_MTRACE_RESP: - case MLD6_MTRACE: - /* XXX: these two are experimental. not officially defind. */ + case MLD_MTRACE_RESP: + case MLD_MTRACE: + /* XXX: these two are experimental. not officially defined. */ /* XXX: per-interface statistics? */ break; /* just pass it to applications */ case ICMP6_NI_QUERY: if (!icmp6_nodeinfo) break; - - /* By RFC 4620 refuse to answer queries from global scope addresses */ +//### LD 10/20 Check fbsd differences here. Not sure we're more advanced or not. + /* By RFC 4620 refuse to answer queries from global scope addresses */ if ((icmp6_nodeinfo & 8) != 8 && in6_addrscope(&ip6->ip6_src) == IPV6_ADDR_SCOPE_GLOBAL) break; @@ -631,6 +763,7 @@ icmp6_input(mp, offp) IP6_EXTHDR_CHECK(m, off, sizeof(struct icmp6_nodeinfo), return IPPROTO_DONE); #endif + n = m_copy(m, 0, M_COPYALL); if (n) n = ni6_input(n, off); @@ -640,7 +773,7 @@ icmp6_input(mp, offp) icmp6stat.icp6s_outhist[ICMP6_WRUREPLY]++; icmp6_reflect(n, noff); } - break; + goto rate_limit_checked; case ICMP6_WRUREPLY: if (code != 0) @@ -653,6 +786,7 @@ icmp6_input(mp, offp) goto badcode; if (icmp6len < sizeof(struct nd_router_solicit)) goto badlen; + if ((n = m_copym(m, 0, M_COPYALL, M_DONTWAIT)) == NULL) { /* give up local */ nd6_rs_input(m, off, icmp6len); @@ -661,7 +795,7 @@ icmp6_input(mp, offp) } nd6_rs_input(n, off, icmp6len); /* m stays. */ - break; + goto rate_limit_checked; case ND_ROUTER_ADVERT: icmp6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_routeradvert); @@ -669,6 +803,7 @@ icmp6_input(mp, offp) goto badcode; if (icmp6len < sizeof(struct nd_router_advert)) goto badlen; + if ((n = m_copym(m, 0, M_COPYALL, M_DONTWAIT)) == NULL) { /* give up local */ nd6_ra_input(m, off, icmp6len); @@ -677,7 +812,7 @@ icmp6_input(mp, offp) } nd6_ra_input(n, off, icmp6len); /* m stays. */ - break; + goto rate_limit_checked; case ND_NEIGHBOR_SOLICIT: icmp6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_neighborsolicit); @@ -685,7 +820,9 @@ icmp6_input(mp, offp) goto badcode; if (icmp6len < sizeof(struct nd_neighbor_solicit)) goto badlen; - if ((n = m_copym(m, 0, M_COPYALL, M_DONTWAIT)) == NULL) { + + if (proxy || + ((n = m_copym(m, 0, M_COPYALL, M_DONTWAIT)) == NULL)) { /* give up local */ nd6_ns_input(m, off, icmp6len); m = NULL; @@ -693,7 +830,7 @@ icmp6_input(mp, offp) } nd6_ns_input(n, off, icmp6len); /* m stays. */ - break; + goto rate_limit_checked; case ND_NEIGHBOR_ADVERT: icmp6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_neighboradvert); @@ -701,6 +838,7 @@ icmp6_input(mp, offp) goto badcode; if (icmp6len < sizeof(struct nd_neighbor_advert)) goto badlen; + if ((n = m_copym(m, 0, M_COPYALL, M_DONTWAIT)) == NULL) { /* give up local */ nd6_na_input(m, off, icmp6len); @@ -709,7 +847,7 @@ icmp6_input(mp, offp) } nd6_na_input(n, off, icmp6len); /* m stays. */ - break; + goto rate_limit_checked; case ND_REDIRECT: icmp6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_redirect); @@ -717,6 +855,7 @@ icmp6_input(mp, offp) goto badcode; if (icmp6len < sizeof(struct nd_redirect)) goto badlen; + if ((n = m_copym(m, 0, M_COPYALL, M_DONTWAIT)) == NULL) { /* give up local */ icmp6_redirect_input(m, off); @@ -725,7 +864,7 @@ icmp6_input(mp, offp) } icmp6_redirect_input(n, off); /* m stays. */ - break; + goto rate_limit_checked; case ICMP6_ROUTER_RENUMBERING: if (code != ICMP6_ROUTER_RENUMBERING_COMMAND && @@ -747,7 +886,7 @@ icmp6_input(mp, offp) /* deliver */ } else { /* ICMPv6 informational: MUST not deliver */ - break; + goto rate_limit_checked; } deliver: if (icmp6_notify_error(m, off, icmp6len, code)) { @@ -765,20 +904,17 @@ icmp6_input(mp, offp) break; } - /* deliver the packet to appropriate sockets */ +rate_limit_checked: icmp6_rip6_input(&m, *offp); - return IPPROTO_DONE; - freeit: +freeit: m_freem(m); return IPPROTO_DONE; } static int -icmp6_notify_error(m, off, icmp6len, code) - struct mbuf *m; - int off, icmp6len, code; +icmp6_notify_error(struct mbuf *m, int off, int icmp6len, int code) { struct icmp6_hdr *icmp6; struct ip6_hdr *eip6; @@ -839,7 +975,7 @@ icmp6_notify_error(m, off, icmp6len, code) return(-1); } #endif - + if (nxt == IPPROTO_AH) eoff += (eh->ip6e_len + 2) << 2; else @@ -897,7 +1033,7 @@ icmp6_notify_error(m, off, icmp6len, code) /* just ignore a bogus header */ if ((rth0->ip6r0_len % 2) == 0 && (hops = rth0->ip6r0_len/2)) - finaldst = (struct in6_addr *)(rth0 + 1) + (hops - 1); + finaldst = (struct in6_addr *)(void *)(rth0 + 1) + (hops - 1); } eoff += rthlen; nxt = rth->ip6r_nxt; @@ -914,7 +1050,7 @@ icmp6_notify_error(m, off, icmp6len, code) eoff, sizeof(*fh)); if (fh == NULL) { icmp6stat.icp6s_tooshort++; - return(-1); + return (-1); } #endif /* @@ -946,14 +1082,23 @@ icmp6_notify_error(m, off, icmp6len, code) icmp6 = (struct icmp6_hdr *)(mtod(m, caddr_t) + off); #else IP6_EXTHDR_GET(icmp6, struct icmp6_hdr *, m, off, - sizeof(*icmp6) + sizeof(struct ip6_hdr)); + sizeof(*icmp6) + sizeof(struct ip6_hdr)); if (icmp6 == NULL) { icmp6stat.icp6s_tooshort++; - return(-1); + return (-1); } #endif + /* + * retrieve parameters from the inner IPv6 header, and convert + * them into sockaddr structures. + * XXX: there is no guarantee that the source or destination + * addresses of the inner packet are in the same scope as + * the addresses of the icmp packet. But there is no other + * way to determine the zone. + */ eip6 = (struct ip6_hdr *)(icmp6 + 1); + bzero(&icmp6dst, sizeof(icmp6dst)); icmp6dst.sin6_len = sizeof(struct sockaddr_in6); icmp6dst.sin6_family = AF_INET6; @@ -961,39 +1106,16 @@ icmp6_notify_error(m, off, icmp6len, code) icmp6dst.sin6_addr = eip6->ip6_dst; else icmp6dst.sin6_addr = *finaldst; - icmp6dst.sin6_scope_id = in6_addr2scopeid(m->m_pkthdr.rcvif, - &icmp6dst.sin6_addr); -#ifndef SCOPEDROUTING - if (in6_embedscope(&icmp6dst.sin6_addr, &icmp6dst, - NULL, NULL)) { - /* should be impossbile */ - nd6log((LOG_DEBUG, - "icmp6_notify_error: in6_embedscope failed\n")); + if (in6_setscope(&icmp6dst.sin6_addr, m->m_pkthdr.rcvif, NULL)) goto freeit; - } -#endif - - /* - * retrieve parameters from the inner IPv6 header, and convert - * them into sockaddr structures. - */ bzero(&icmp6src, sizeof(icmp6src)); icmp6src.sin6_len = sizeof(struct sockaddr_in6); icmp6src.sin6_family = AF_INET6; icmp6src.sin6_addr = eip6->ip6_src; - icmp6src.sin6_scope_id = in6_addr2scopeid(m->m_pkthdr.rcvif, - &icmp6src.sin6_addr); -#ifndef SCOPEDROUTING - if (in6_embedscope(&icmp6src.sin6_addr, &icmp6src, - NULL, NULL)) { - /* should be impossbile */ - nd6log((LOG_DEBUG, - "icmp6_notify_error: in6_embedscope failed\n")); + if (in6_setscope(&icmp6src.sin6_addr, m->m_pkthdr.rcvif, NULL)) goto freeit; - } -#endif icmp6src.sin6_flowinfo = - (eip6->ip6_flow & IPV6_FLOWLABEL_MASK); + (eip6->ip6_flow & IPV6_FLOWLABEL_MASK); if (finaldst == NULL) finaldst = &eip6->ip6_dst; @@ -1020,15 +1142,13 @@ icmp6_notify_error(m, off, icmp6len, code) } return(0); - freeit: +freeit: m_freem(m); return(-1); } void -icmp6_mtudisc_update(ip6cp, validated) - struct ip6ctlparam *ip6cp; - int validated; +icmp6_mtudisc_update(struct ip6ctlparam *ip6cp, int validated) { struct in6_addr *dst = ip6cp->ip6c_finaldst; struct icmp6_hdr *icmp6 = ip6cp->ip6c_icmp6; @@ -1036,10 +1156,25 @@ icmp6_mtudisc_update(ip6cp, validated) u_int mtu = ntohl(icmp6->icmp6_mtu); struct rtentry *rt = NULL; struct sockaddr_in6 sin6; + /* + * we reject ICMPv6 too big with abnormally small value. + * XXX what is the good definition of "abnormally small"? + */ + if (mtu < sizeof(struct ip6_hdr) + sizeof(struct ip6_frag) + 8) + return; if (!validated) return; + /* + * In case the suggested mtu is less than IPV6_MMTU, we + * only need to remember that it was for above mentioned + * "alwaysfrag" case. + * Try to be as close to the spec as possible. + */ + if (mtu < IPV6_MMTU) + mtu = IPV6_MMTU - 8; + bzero(&sin6, sizeof(sin6)); sin6.sin6_family = PF_INET6; sin6.sin6_len = sizeof(struct sockaddr_in6); @@ -1050,27 +1185,33 @@ icmp6_mtudisc_update(ip6cp, validated) htons(m->m_pkthdr.rcvif->if_index); } /* sin6.sin6_scope_id = XXX: should be set if DST is a scoped addr */ - rt = rtalloc1((struct sockaddr *)&sin6, 0, RTF_CLONING | RTF_PRCLONING); - - if (rt && (rt->rt_flags & RTF_HOST) - && !(rt->rt_rmx.rmx_locks & RTV_MTU)) { - if (mtu < IPV6_MMTU) { - /* xxx */ - rt->rt_rmx.rmx_locks |= RTV_MTU; - } else if (mtu < rt->rt_ifp->if_mtu && - rt->rt_rmx.rmx_mtu > mtu) { + /* + * XXX On a side note, for asymmetric data-path + * the lookup on receive interace is probably not + * what we want to do. + * That requires looking at the cached route for the + * protocol control block. + */ + rt = rtalloc1_scoped((struct sockaddr *)&sin6, 0, + RTF_CLONING | RTF_PRCLONING, m->m_pkthdr.rcvif->if_index); + if (rt != NULL) { + RT_LOCK(rt); + if ((rt->rt_flags & RTF_HOST) && + !(rt->rt_rmx.rmx_locks & RTV_MTU) && + mtu < IN6_LINKMTU(rt->rt_ifp) && + rt->rt_rmx.rmx_mtu > mtu) { icmp6stat.icp6s_pmtuchg++; rt->rt_rmx.rmx_mtu = mtu; } - } - if (rt) + RT_UNLOCK(rt); rtfree(rt); + } } /* * Process a Node Information Query packet, based on * draft-ietf-ipngwg-icmp-name-lookups-07. - * + * * Spec incompatibilities: * - IPv6 Subject address handling * - IPv4 Subject address handling support missing @@ -1080,9 +1221,7 @@ icmp6_mtudisc_update(ip6cp, validated) */ #define hostnamelen strlen(hostname) static struct mbuf * -ni6_input(m, off) - struct mbuf *m; - int off; +ni6_input(struct mbuf *m, int off) { struct icmp6_nodeinfo *ni6, *nni6; struct mbuf *n = NULL; @@ -1097,7 +1236,6 @@ ni6_input(m, off) struct ip6_hdr *ip6; int oldfqdn = 0; /* if 1, return pascal string (03 draft) */ char *subj = NULL; - struct in6_ifaddr *ia6 = NULL; ip6 = mtod(m, struct ip6_hdr *); #ifndef PULLDOWN_TEST @@ -1106,40 +1244,54 @@ ni6_input(m, off) IP6_EXTHDR_GET(ni6, struct icmp6_nodeinfo *, m, off, sizeof(*ni6)); if (ni6 == NULL) { /* m is already reclaimed */ - return NULL; + return (NULL); } #endif + /* + * Validate IPv6 source address. + * The default configuration MUST be to refuse answering queries from + * global-scope addresses according to RFC4602. + * Notes: + * - it's not very clear what "refuse" means; this implementation + * simply drops it. + * - it's not very easy to identify global-scope (unicast) addresses + * since there are many prefixes for them. It should be safer + * and in practice sufficient to check "all" but loopback and + * link-local (note that site-local unicast was deprecated and + * ULA is defined as global scope-wise) + */ + if ((icmp6_nodeinfo & ICMP6_NODEINFO_GLOBALOK) == 0 && + !IN6_IS_ADDR_LOOPBACK(&ip6->ip6_src) && + !IN6_IS_ADDR_LINKLOCAL(&ip6->ip6_src)) + goto bad; + /* * Validate IPv6 destination address. * * The Responder must discard the Query without further processing * unless it is one of the Responder's unicast or anycast addresses, or * a link-local scope multicast address which the Responder has joined. - * [icmp-name-lookups-07, Section 4.] + * [RFC4602, Section 5.] */ - bzero(&sin6, sizeof(sin6)); - sin6.sin6_family = AF_INET6; - sin6.sin6_len = sizeof(struct sockaddr_in6); - bcopy(&ip6->ip6_dst, &sin6.sin6_addr, sizeof(sin6.sin6_addr)); - /* XXX scopeid */ - if ((ia6 = (struct in6_ifaddr *)ifa_ifwithaddr((struct sockaddr *)&sin6)) != NULL) { - /* unicast/anycast, fine */ - if ((ia6->ia6_flags & IN6_IFF_TEMPORARY) != 0 && - (icmp6_nodeinfo & 4) == 0) { - ifafree(&ia6->ia_ifa); - ia6 = NULL; + if (IN6_IS_ADDR_MULTICAST(&ip6->ip6_dst)) { + if (!IN6_IS_ADDR_MC_LINKLOCAL(&ip6->ip6_dst)) + goto bad; + /* else it's a link-local multicast, fine */ + } else { /* unicast or anycast */ + uint32_t ia6_flags; + + if (ip6_getdstifaddr_info(m, NULL, &ia6_flags) != 0) + goto bad; /* XXX impossible */ + + if ((ia6_flags & IN6_IFF_TEMPORARY) && + !(icmp6_nodeinfo & ICMP6_NODEINFO_TMPADDROK)) { nd6log((LOG_DEBUG, "ni6_input: ignore node info to " "a temporary address in %s:%d", __FILE__, __LINE__)); goto bad; } - ifafree(&ia6->ia_ifa); - ia6 = NULL; - } else if (IN6_IS_ADDR_MC_LINKLOCAL(&sin6.sin6_addr)) - ; /* link-local multicast, fine */ - else - goto bad; + } /* validate query Subject field. */ qtype = ntohs(ni6->ni_qtype); @@ -1153,6 +1305,7 @@ ni6_input(m, off) /* FALLTHROUGH */ case NI_QTYPE_FQDN: case NI_QTYPE_NODEADDR: + case NI_QTYPE_IPV4ADDR: switch (ni6->ni_code) { case ICMP6_NI_SUBJ_IPV6: #if ICMP6_NI_SUBJ_IPV6 != 0 @@ -1172,7 +1325,7 @@ ni6_input(m, off) goto bad; #endif - if (subjlen != sizeof(sin6.sin6_addr)) + if (subjlen != sizeof(struct in6_addr)) goto bad; /* @@ -1194,18 +1347,16 @@ ni6_input(m, off) subjlen, (caddr_t)&sin6.sin6_addr); sin6.sin6_scope_id = in6_addr2scopeid(m->m_pkthdr.rcvif, &sin6.sin6_addr); -#ifndef SCOPEDROUTING - in6_embedscope(&sin6.sin6_addr, &sin6, NULL, NULL); -#endif + in6_embedscope(&sin6.sin6_addr, &sin6, NULL, NULL, + NULL); bzero(&sin6_d, sizeof(sin6_d)); sin6_d.sin6_family = AF_INET6; /* not used, actually */ sin6_d.sin6_len = sizeof(sin6_d); /* ditto */ sin6_d.sin6_addr = ip6->ip6_dst; sin6_d.sin6_scope_id = in6_addr2scopeid(m->m_pkthdr.rcvif, &ip6->ip6_dst); -#ifndef SCOPEDROUTING - in6_embedscope(&sin6_d.sin6_addr, &sin6_d, NULL, NULL); -#endif + in6_embedscope(&sin6_d.sin6_addr, &sin6_d, NULL, NULL, + NULL); subj = (char *)&sin6; if (SA6_ARE_ADDR_EQUAL(&sin6, &sin6_d)) break; @@ -1214,7 +1365,8 @@ ni6_input(m, off) * XXX if we are to allow other cases, we should really * be careful about scope here. * basically, we should disallow queries toward IPv6 - * destination X with subject Y, if scope(X) > scope(Y). + * destination X with subject Y, + * if scope(X) > scope(Y). * if we allow scope(X) > scope(Y), it will result in * information leakage across scope boundary. */ @@ -1257,11 +1409,12 @@ ni6_input(m, off) /* refuse based on configuration. XXX ICMP6_NI_REFUSED? */ switch (qtype) { case NI_QTYPE_FQDN: - if ((icmp6_nodeinfo & 1) == 0) + if ((icmp6_nodeinfo & ICMP6_NODEINFO_FQDNOK) == 0) goto bad; break; case NI_QTYPE_NODEADDR: - if ((icmp6_nodeinfo & 2) == 0) + case NI_QTYPE_IPV4ADDR: + if ((icmp6_nodeinfo & ICMP6_NODEINFO_NODEADDROK) == 0) goto bad; break; } @@ -1280,13 +1433,16 @@ ni6_input(m, off) case NI_QTYPE_NODEADDR: addrs = ni6_addrs(ni6, &ifp, subj); if ((replylen += addrs * (sizeof(struct in6_addr) + - sizeof(u_int32_t))) > MCLBYTES) + sizeof(u_int32_t))) > MCLBYTES) replylen = MCLBYTES; /* XXX: will truncate pkt later */ break; + case NI_QTYPE_IPV4ADDR: + /* unsupported - should respond with unknown Qtype? */ + break; default: /* * XXX: We must return a reply with the ICMP6 code - * `unknown Qtype' in this case. However we regard the case + * `unknown Qtype' in this case. However we regard the case * as an FQDN query for backward compatibility. * Older versions set a random value to this field, * so it rarely varies in the defined qtypes. @@ -1304,7 +1460,9 @@ ni6_input(m, off) MGETHDR(n, M_DONTWAIT, m->m_type); /* MAC-OK */ if (n == NULL) { m_freem(m); - return(NULL); + if (ifp != NULL) + ifnet_release(ifp); + return (NULL); } M_COPY_PKTHDR(n, m); /* just for recvif */ if (replylen > MHLEN) { @@ -1381,13 +1539,17 @@ ni6_input(m, off) nni6->ni_type = ICMP6_NI_REPLY; m_freem(m); - return(n); + if (ifp != NULL) + ifnet_release(ifp); + return (n); - bad: +bad: m_freem(m); if (n) m_freem(n); - return(NULL); + if (ifp != NULL) + ifnet_release(ifp); + return (NULL); } #undef hostnamelen @@ -1398,10 +1560,10 @@ ni6_input(m, off) * treated as truncated name (two \0 at the end). this is a wild guess. */ static struct mbuf * -ni6_nametodns(name, namelen, old) - const char *name; - int namelen; - int old; /* return pascal string if non-zero */ +ni6_nametodns( + const char *name, + int namelen, + int old) /* return pascal string if non-zero */ { struct mbuf *m; char *cp, *ep; @@ -1498,11 +1660,7 @@ ni6_nametodns(name, namelen, old) * XXX upper/lowercase match (see RFC2065) */ static int -ni6_dnsmatch(a, alen, b, blen) - const char *a; - int alen; - const char *b; - int blen; +ni6_dnsmatch(const char *a, int alen, const char *b, int blen) { const char *a0, *b0; int l; @@ -1562,10 +1720,7 @@ ni6_dnsmatch(a, alen, b, blen) * calculate the number of addresses to be returned in the node info reply. */ static int -ni6_addrs(ni6, ifpp, subj) - struct icmp6_nodeinfo *ni6; - struct ifnet **ifpp; - char *subj; +ni6_addrs(struct icmp6_nodeinfo *ni6, struct ifnet **ifpp, char *subj) { struct ifnet *ifp; struct in6_ifaddr *ifa6; @@ -1574,19 +1729,22 @@ ni6_addrs(ni6, ifpp, subj) int addrs = 0, addrsofif, iffound = 0; int niflags = ni6->ni_flags; + if (ifpp != NULL) + *ifpp = NULL; + if ((niflags & NI_NODEADDR_FLAG_ALL) == 0) { switch (ni6->ni_code) { case ICMP6_NI_SUBJ_IPV6: if (subj == NULL) /* must be impossible... */ return(0); - subj_ip6 = (struct sockaddr_in6 *)subj; + subj_ip6 = (struct sockaddr_in6 *)(void *)subj; break; default: /* * XXX: we only support IPv6 subject address for * this Qtype. */ - return(0); + return (0); } } @@ -1596,8 +1754,11 @@ ni6_addrs(ni6, ifpp, subj) ifnet_lock_shared(ifp); TAILQ_FOREACH(ifa, &ifp->if_addrlist, ifa_list) { - if (ifa->ifa_addr->sa_family != AF_INET6) + IFA_LOCK(ifa); + if (ifa->ifa_addr->sa_family != AF_INET6) { + IFA_UNLOCK(ifa); continue; + } ifa6 = (struct in6_ifaddr *)ifa; if ((niflags & NI_NODEADDR_FLAG_ALL) == 0 && @@ -1618,18 +1779,25 @@ ni6_addrs(ni6, ifpp, subj) /* What do we have to do about ::1? */ switch (in6_addrscope(&ifa6->ia_addr.sin6_addr)) { case IPV6_ADDR_SCOPE_LINKLOCAL: - if ((niflags & NI_NODEADDR_FLAG_LINKLOCAL) == 0) + if ((niflags & NI_NODEADDR_FLAG_LINKLOCAL) == 0) { + IFA_UNLOCK(ifa); continue; + } break; case IPV6_ADDR_SCOPE_SITELOCAL: - if ((niflags & NI_NODEADDR_FLAG_SITELOCAL) == 0) + if ((niflags & NI_NODEADDR_FLAG_SITELOCAL) == 0) { + IFA_UNLOCK(ifa); continue; + } break; case IPV6_ADDR_SCOPE_GLOBAL: - if ((niflags & NI_NODEADDR_FLAG_GLOBAL) == 0) + if ((niflags & NI_NODEADDR_FLAG_GLOBAL) == 0) { + IFA_UNLOCK(ifa); continue; + } break; default: + IFA_UNLOCK(ifa); continue; } @@ -1638,17 +1806,24 @@ ni6_addrs(ni6, ifpp, subj) * XXX: just experimental. not in the spec. */ if ((ifa6->ia6_flags & IN6_IFF_ANYCAST) != 0 && - (niflags & NI_NODEADDR_FLAG_ANYCAST) == 0) + (niflags & NI_NODEADDR_FLAG_ANYCAST) == 0) { + IFA_UNLOCK(ifa); continue; /* we need only unicast addresses */ + } if ((ifa6->ia6_flags & IN6_IFF_TEMPORARY) != 0 && - (icmp6_nodeinfo & 4) == 0) { + (icmp6_nodeinfo & ICMP6_NODEINFO_TMPADDROK) == 0) { + IFA_UNLOCK(ifa); continue; } addrsofif++; /* count the address */ + IFA_UNLOCK(ifa); } ifnet_lock_done(ifp); if (iffound) { - *ifpp = ifp; + if (ifpp != NULL) { + *ifpp = ifp; + ifnet_reference(ifp); + } ifnet_head_done(); return(addrsofif); } @@ -1657,14 +1832,12 @@ ni6_addrs(ni6, ifpp, subj) } ifnet_head_done(); - return(addrs); + return (addrs); } static int -ni6_store_addrs(ni6, nni6, ifp0, resid) - struct icmp6_nodeinfo *ni6, *nni6; - struct ifnet *ifp0; - int resid; +ni6_store_addrs(struct icmp6_nodeinfo *ni6, struct icmp6_nodeinfo *nni6, + struct ifnet *ifp0, int resid) { struct ifnet *ifp = ifp0; struct in6_ifaddr *ifa6; @@ -1674,25 +1847,28 @@ ni6_store_addrs(ni6, nni6, ifp0, resid) u_char *cp = (u_char *)(nni6 + 1); int niflags = ni6->ni_flags; u_int32_t ltime; - struct timeval timenow; - - getmicrotime(&timenow); + uint64_t now = net_uptime(); if (ifp0 == NULL && !(niflags & NI_NODEADDR_FLAG_ALL)) - return(0); /* needless to copy */ + return (0); /* needless to copy */ again: ifnet_head_lock_shared(); - if (ifp == NULL) ifp = TAILQ_FIRST(&ifnet_head); - + if (ifp == NULL) + ifp = TAILQ_FIRST(&ifnet_head); + for (; ifp; ifp = TAILQ_NEXT(ifp, if_list)) { ifnet_lock_shared(ifp); for (ifa = ifp->if_addrlist.tqh_first; ifa; - ifa = ifa->ifa_list.tqe_next) - { - if (ifa->ifa_addr->sa_family != AF_INET6) + ifa = ifa->ifa_list.tqe_next) { + struct in6_addrlifetime_i *lt; + + IFA_LOCK(ifa); + if (ifa->ifa_addr->sa_family != AF_INET6) { + IFA_UNLOCK(ifa); continue; + } ifa6 = (struct in6_ifaddr *)ifa; if ((ifa6->ia6_flags & IN6_IFF_DEPRECATED) != 0 && @@ -1706,45 +1882,57 @@ ni6_store_addrs(ni6, nni6, ifp0, resid) if (ifp_dep == NULL) ifp_dep = ifp; + IFA_UNLOCK(ifa); continue; - } - else if ((ifa6->ia6_flags & IN6_IFF_DEPRECATED) == 0 && - allow_deprecated != 0) + } else if ((ifa6->ia6_flags & IN6_IFF_DEPRECATED) == 0 && + allow_deprecated != 0) { + IFA_UNLOCK(ifa); continue; /* we now collect deprecated addrs */ - + } /* What do we have to do about ::1? */ switch (in6_addrscope(&ifa6->ia_addr.sin6_addr)) { case IPV6_ADDR_SCOPE_LINKLOCAL: - if ((niflags & NI_NODEADDR_FLAG_LINKLOCAL) == 0) + if ((niflags & NI_NODEADDR_FLAG_LINKLOCAL) == 0) { + IFA_UNLOCK(ifa); continue; + } break; case IPV6_ADDR_SCOPE_SITELOCAL: - if ((niflags & NI_NODEADDR_FLAG_SITELOCAL) == 0) + if ((niflags & NI_NODEADDR_FLAG_SITELOCAL) == 0) { + IFA_UNLOCK(ifa); continue; + } break; case IPV6_ADDR_SCOPE_GLOBAL: - if ((niflags & NI_NODEADDR_FLAG_GLOBAL) == 0) + if ((niflags & NI_NODEADDR_FLAG_GLOBAL) == 0) { + IFA_UNLOCK(ifa); continue; + } break; default: + IFA_UNLOCK(ifa); continue; } /* * check if anycast is okay. - * XXX: just experimental. not in the spec. + * XXX: just experimental. not in the spec. */ if ((ifa6->ia6_flags & IN6_IFF_ANYCAST) != 0 && - (niflags & NI_NODEADDR_FLAG_ANYCAST) == 0) + (niflags & NI_NODEADDR_FLAG_ANYCAST) == 0) { + IFA_UNLOCK(ifa); continue; + } if ((ifa6->ia6_flags & IN6_IFF_TEMPORARY) != 0 && - (icmp6_nodeinfo & 4) == 0) { + (icmp6_nodeinfo & ICMP6_NODEINFO_TMPADDROK) == 0) { + IFA_UNLOCK(ifa); continue; } /* now we can copy the address */ if (resid < sizeof(struct in6_addr) + sizeof(u_int32_t)) { + IFA_UNLOCK(ifa); /* * We give up much more copy. * Set the truncate flag and return. @@ -1771,16 +1959,16 @@ ni6_store_addrs(ni6, nni6, ifp0, resid) * address configuration by DHCPv6, so the former * case can't happen. */ - if (ifa6->ia6_lifetime.ia6t_expire == 0) + lt = &ifa6->ia6_lifetime; + if (lt->ia6ti_expire == 0) { ltime = ND6_INFINITE_LIFETIME; - else { - if (ifa6->ia6_lifetime.ia6t_expire > - timenow.tv_sec) - ltime = htonl(ifa6->ia6_lifetime.ia6t_expire - timenow.tv_sec); + } else { + if (lt->ia6ti_expire > now) + ltime = htonl(lt->ia6ti_expire - now); else ltime = 0; } - + bcopy(<ime, cp, sizeof(u_int32_t)); cp += sizeof(u_int32_t); @@ -1789,12 +1977,13 @@ ni6_store_addrs(ni6, nni6, ifp0, resid) sizeof(struct in6_addr)); /* XXX: KAME link-local hack; remove ifindex */ if (IN6_IS_ADDR_LINKLOCAL(&ifa6->ia_addr.sin6_addr)) - ((struct in6_addr *)cp)->s6_addr16[1] = 0; + ((struct in6_addr *)(void *)cp)->s6_addr16[1] = 0; cp += sizeof(struct in6_addr); - + resid -= (sizeof(struct in6_addr) + sizeof(u_int32_t)); copied += (sizeof(struct in6_addr) + sizeof(u_int32_t)); + IFA_UNLOCK(ifa); } ifnet_lock_done(ifp); if (ifp0) /* we need search only on the specified IF */ @@ -1816,9 +2005,7 @@ ni6_store_addrs(ni6, nni6, ifp0, resid) * XXX almost dup'ed code with rip6_input. */ static int -icmp6_rip6_input(mp, off) - struct mbuf **mp; - int off; +icmp6_rip6_input(struct mbuf **mp, int off) { struct mbuf *m = *mp; struct ip6_hdr *ip6 = mtod(m, struct ip6_hdr *); @@ -1827,6 +2014,8 @@ icmp6_rip6_input(mp, off) struct sockaddr_in6 rip6src; struct icmp6_hdr *icmp6; struct mbuf *opts = NULL; + int ret = 0; + struct ifnet *ifp = m->m_pkthdr.rcvif; #ifndef PULLDOWN_TEST /* this is assumed to be safe. */ @@ -1839,21 +2028,22 @@ icmp6_rip6_input(mp, off) } #endif + /* + * XXX: the address may have embedded scope zone ID, which should be + * hidden from applications. + */ bzero(&rip6src, sizeof(rip6src)); - rip6src.sin6_len = sizeof(struct sockaddr_in6); rip6src.sin6_family = AF_INET6; - /* KAME hack: recover scopeid */ - (void)in6_recoverscope(&rip6src, &ip6->ip6_src, m->m_pkthdr.rcvif); + rip6src.sin6_len = sizeof(struct sockaddr_in6); + rip6src.sin6_addr = ip6->ip6_src; + if (sa6_recoverscope(&rip6src, TRUE)) + return (IPPROTO_DONE); - lck_rw_lock_shared(ripcbinfo.mtx); + lck_rw_lock_shared(ripcbinfo.ipi_lock); LIST_FOREACH(in6p, &ripcb, inp_list) { if ((in6p->inp_vflag & INP_IPV6) == 0) continue; -#if HAVE_NRL_INPCB - if (!(in6p->in6p_flags & INP_IPV6)) - continue; -#endif if (in6p->in6p_ip6_nxt != IPPROTO_ICMPV6) continue; if (!IN6_IS_ADDR_UNSPECIFIED(&in6p->in6p_laddr) && @@ -1866,13 +2056,27 @@ icmp6_rip6_input(mp, off) && ICMP6_FILTER_WILLBLOCK(icmp6->icmp6_type, in6p->in6p_icmp6filt)) continue; + + if (inp_restricted_recv(in6p, ifp)) + continue; + if (last) { struct mbuf *n; if ((n = m_copy(m, 0, (int)M_COPYALL)) != NULL) { - if (last->in6p_flags & IN6P_CONTROLOPTS) - ip6_savecontrol(last, &opts, ip6, n); + if ((last->in6p_flags & INP_CONTROLOPTS) != 0 || + (last->in6p_socket->so_options & SO_TIMESTAMP) != 0 || + (last->in6p_socket->so_options & SO_TIMESTAMP_MONOTONIC) != 0) { + ret = ip6_savecontrol(last, n, &opts); + if (ret != 0) { + m_freem(n); + m_freem(opts); + last = in6p; + continue; + } + } /* strip intermediate headers */ m_adj(n, off); + so_recv_data_stat(last->in6p_socket, m, 0); if (sbappendaddr(&last->in6p_socket->so_rcv, (struct sockaddr *)&rip6src, n, opts, NULL) != 0) { @@ -1883,20 +2087,33 @@ icmp6_rip6_input(mp, off) } last = in6p; } - lck_rw_done(ripcbinfo.mtx); if (last) { - if (last->in6p_flags & IN6P_CONTROLOPTS) - ip6_savecontrol(last, &opts, ip6, m); + if ((last->in6p_flags & INP_CONTROLOPTS) != 0 || + (last->in6p_socket->so_options & SO_TIMESTAMP) != 0 || + (last->in6p_socket->so_options & SO_TIMESTAMP_MONOTONIC) != 0) { + ret = ip6_savecontrol(last, m, &opts); + if (ret != 0) { + goto error; + } + } /* strip intermediate headers */ m_adj(m, off); + so_recv_data_stat(last->in6p_socket, m, 0); if (sbappendaddr(&last->in6p_socket->so_rcv, (struct sockaddr *)&rip6src, m, opts, NULL) != 0) { sorwakeup(last->in6p_socket); } } else { - m_freem(m); - ip6stat.ip6s_delivered--; + goto error; } + lck_rw_done(ripcbinfo.ipi_lock); + return IPPROTO_DONE; + +error: + lck_rw_done(ripcbinfo.ipi_lock); + m_freem(m); + m_freem(opts); + ip6stat.ip6s_delivered--; return IPPROTO_DONE; } @@ -1905,10 +2122,9 @@ icmp6_rip6_input(mp, off) * OFF points to the icmp6 header, counted from the top of the mbuf. */ void -icmp6_reflect(m, off) - struct mbuf *m; - size_t off; +icmp6_reflect(struct mbuf *m, size_t off) { + struct mbuf *m_ip6hdr = m; struct ip6_hdr *ip6; struct icmp6_hdr *icmp6; struct in6_ifaddr *ia; @@ -1917,16 +2133,23 @@ icmp6_reflect(m, off) int type, code; struct ifnet *outif = NULL; struct sockaddr_in6 sa6_src, sa6_dst; -#ifdef COMPAT_RFC1885 - int mtu = IPV6_MMTU; - struct sockaddr_in6 *sin6 = &icmp6_reflect_rt.ro_dst; -#endif + struct nd_ifinfo *ndi = NULL; + u_int32_t oflow; + struct ip6_out_args ip6oa = { IFSCOPE_NONE, { 0 }, + IP6OAF_SELECT_SRCIF | IP6OAF_BOUND_SRCADDR | + IP6OAF_INTCOPROC_ALLOWED | IP6OAF_AWDL_UNRESTRICTED, 0, + SO_TC_UNSPEC, _NET_SERVICE_TYPE_UNSPEC }; + + if (!(m->m_pkthdr.pkt_flags & PKTF_LOOP) && m->m_pkthdr.rcvif != NULL) { + ip6oa.ip6oa_boundif = m->m_pkthdr.rcvif->if_index; + ip6oa.ip6oa_flags |= IP6OAF_BOUND_IF; + } /* too short to reflect */ if (off < sizeof(struct ip6_hdr)) { nd6log((LOG_DEBUG, "sanity fail: off=%lx, sizeof(ip6)=%lx in %s:%d\n", - (u_long)off, (u_long)sizeof(struct ip6_hdr), + (u_int32_t)off, (u_int32_t)sizeof(struct ip6_hdr), __FILE__, __LINE__)); goto bad; } @@ -1935,10 +2158,6 @@ icmp6_reflect(m, off) * If there are extra headers between IPv6 and ICMPv6, strip * off that header first. */ -#if DIAGNOSTIC - if (sizeof(struct ip6_hdr) + sizeof(struct icmp6_hdr) > MHLEN) - panic("assumption failed in icmp6_reflect"); -#endif if (off > sizeof(struct ip6_hdr)) { size_t l; struct ip6_hdr nip6; @@ -1948,7 +2167,7 @@ icmp6_reflect(m, off) m_adj(m, l); l = sizeof(struct ip6_hdr) + sizeof(struct icmp6_hdr); if (m->m_len < l) { - if ((m = m_pullup(m, l)) == NULL) + if ((m_ip6hdr = m_pulldown(m, 0, l, NULL)) == NULL) return; } bcopy((caddr_t)&nip6, mtod(m, caddr_t), sizeof(nip6)); @@ -1956,12 +2175,12 @@ icmp6_reflect(m, off) size_t l; l = sizeof(struct ip6_hdr) + sizeof(struct icmp6_hdr); if (m->m_len < l) { - if ((m = m_pullup(m, l)) == NULL) + if ((m_ip6hdr = m_pulldown(m, 0, l, NULL)) == NULL) return; } } plen = m->m_pkthdr.len - sizeof(struct ip6_hdr); - ip6 = mtod(m, struct ip6_hdr *); + ip6 = mtod(m_ip6hdr, struct ip6_hdr *); ip6->ip6_nxt = IPPROTO_ICMPV6; icmp6 = (struct icmp6_hdr *)(ip6 + 1); type = icmp6->icmp6_type; /* keep type for statistics */ @@ -1978,80 +2197,57 @@ icmp6_reflect(m, off) * XXX: make sure to embed scope zone information, using * already embedded IDs or the received interface (if any). * Note that rcvif may be NULL. - * TODO: scoped routing case (XXX). */ bzero(&sa6_src, sizeof(sa6_src)); sa6_src.sin6_family = AF_INET6; sa6_src.sin6_len = sizeof(sa6_src); sa6_src.sin6_addr = ip6->ip6_dst; in6_recoverscope(&sa6_src, &ip6->ip6_dst, m->m_pkthdr.rcvif); - in6_embedscope(&ip6->ip6_dst, &sa6_src, NULL, NULL); + in6_embedscope(&ip6->ip6_dst, &sa6_src, NULL, NULL, NULL); bzero(&sa6_dst, sizeof(sa6_dst)); sa6_dst.sin6_family = AF_INET6; sa6_dst.sin6_len = sizeof(sa6_dst); sa6_dst.sin6_addr = t; in6_recoverscope(&sa6_dst, &t, m->m_pkthdr.rcvif); - in6_embedscope(&t, &sa6_dst, NULL, NULL); + in6_embedscope(&t, &sa6_dst, NULL, NULL, NULL); -#ifdef COMPAT_RFC1885 - /* - * xxx guess MTU - * RFC 1885 requires that echo reply should be truncated if it - * does not fit in with (return) path MTU, but the description was - * removed in the new spec. - */ - if (icmp6_reflect_rt.ro_rt == 0 || - ! (IN6_ARE_ADDR_EQUAL(&sin6->sin6_addr, &ip6->ip6_dst))) { - if (icmp6_reflect_rt.ro_rt) { - rtfree(icmp6_reflect_rt.ro_rt); - icmp6_reflect_rt.ro_rt = 0; - } - bzero(sin6, sizeof(*sin6)); - sin6->sin6_family = PF_INET6; - sin6->sin6_len = sizeof(struct sockaddr_in6); - sin6->sin6_addr = ip6->ip6_dst; - - rtalloc_ign((struct route *)&icmp6_reflect_rt.ro_rt, - RTF_PRCLONING); - } - - if (icmp6_reflect_rt.ro_rt == 0) - goto bad; - - if ((icmp6_reflect_rt.ro_rt->rt_flags & RTF_HOST) - && mtu < icmp6_reflect_rt.ro_rt->rt_ifp->if_mtu) - mtu = icmp6_reflect_rt.ro_rt->rt_rmx.rmx_mtu; - - if (mtu < m->m_pkthdr.len) { - plen -= (m->m_pkthdr.len - mtu); - m_adj(m, mtu - m->m_pkthdr.len); - } -#endif /* * If the incoming packet was addressed directly to us(i.e. unicast), * use dst as the src for the reply. - * The IN6_IFF_NOTREADY case would be VERY rare, but is possible + * The IN6_IFF_NOTREADY case should be VERY rare, but is possible * (for example) when we encounter an error while forwarding procedure * destined to a duplicated address of ours. + * Note that ip6_getdstifaddr() may fail if we are in an error handling + * procedure of an outgoing packet of our own, in which case we need + * to search in the ifaddr list. */ - lck_mtx_lock(nd6_mutex); - for (ia = in6_ifaddrs; ia; ia = ia->ia_next) + lck_rw_lock_shared(&in6_ifaddr_rwlock); + for (ia = in6_ifaddrs; ia; ia = ia->ia_next) { + IFA_LOCK(&ia->ia_ifa); if (IN6_ARE_ADDR_EQUAL(&t, &ia->ia_addr.sin6_addr) && (ia->ia6_flags & (IN6_IFF_ANYCAST|IN6_IFF_NOTREADY)) == 0) { + IFA_UNLOCK(&ia->ia_ifa); src = &t; break; } - lck_mtx_unlock(nd6_mutex); - if (ia == NULL && IN6_IS_ADDR_LINKLOCAL(&t) && (m->m_flags & M_LOOP)) { + IFA_UNLOCK(&ia->ia_ifa); + } + lck_rw_done(&in6_ifaddr_rwlock); + if (ia == NULL && IN6_IS_ADDR_LINKLOCAL(&t) && + ((m->m_flags & M_LOOP) || (m->m_pkthdr.pkt_flags & PKTF_LOOP))) { /* * This is the case if the dst is our link-local address - * and the sender is also ourselves. + * and the sender is also ourselves. Here we test for both + * M_LOOP and PKTF_LOOP, since the former may have been set + * in ip6_output() and that we get here as part of callling + * ip6_process_hopopts(). See comments in */ src = &t; } - if (src == 0) { + if (src == NULL) { int e; + struct sockaddr_in6 sin6; struct route_in6 ro; /* @@ -2059,10 +2255,20 @@ icmp6_reflect(m, off) * that we do not own. Select a source address based on the * source address of the erroneous packet. */ + bzero(&sin6, sizeof(sin6)); + sin6.sin6_family = AF_INET6; + sin6.sin6_len = sizeof(sin6); + sin6.sin6_addr = ip6->ip6_dst; /* zone ID should be embedded */ + bzero(&ro, sizeof(ro)); - src = in6_selectsrc(&sa6_src, NULL, NULL, &ro, NULL, &src_storage, &e); - if (ro.ro_rt) - rtfree(ro.ro_rt); /* XXX: we could use this */ + /* + * in6_selectsrc() might return outif with its reference held + * even in the error case, so we always need to release it + * if non-NULL. + */ + src = in6_selectsrc(&sin6, NULL, NULL, &ro, &outif, + &src_storage, ip6oa.ip6oa_boundif, &e); + ROUTE_RELEASE(&ro); if (src == NULL) { nd6log((LOG_DEBUG, "icmp6_reflect: source can't be determined: " @@ -2071,61 +2277,66 @@ icmp6_reflect(m, off) goto bad; } } - + oflow = ip6->ip6_flow; /* Save for later */ ip6->ip6_src = *src; - ip6->ip6_flow = 0; ip6->ip6_vfc &= ~IPV6_VERSION_MASK; ip6->ip6_vfc |= IPV6_VERSION; + if (icmp6->icmp6_type == ICMP6_ECHO_REPLY && icmp6->icmp6_code == 0) { + ip6->ip6_flow |= (oflow & htonl(0x0ff00000)); + } ip6->ip6_nxt = IPPROTO_ICMPV6; - if (m->m_pkthdr.rcvif) { + if (outif != NULL && (ndi = ND_IFINFO(outif)) != NULL && + ndi->initialized) { + lck_mtx_lock(&ndi->lock); + ip6->ip6_hlim = ndi->chlim; + lck_mtx_unlock(&ndi->lock); + } + if (m->m_pkthdr.rcvif != NULL && + (ndi = ND_IFINFO(m->m_pkthdr.rcvif)) != NULL && + ndi->initialized) { /* XXX: This may not be the outgoing interface */ - ip6->ip6_hlim = nd_ifinfo[m->m_pkthdr.rcvif->if_index].chlim; - } else + lck_mtx_lock(&ndi->lock); + ip6->ip6_hlim = ndi->chlim; + lck_mtx_unlock(&ndi->lock); + } else { ip6->ip6_hlim = ip6_defhlim; - + } + /* Use the same traffic class as in the request to match IPv4 */ icmp6->icmp6_cksum = 0; icmp6->icmp6_cksum = in6_cksum(m, IPPROTO_ICMPV6, - sizeof(struct ip6_hdr), plen); + sizeof(struct ip6_hdr), plen); /* * XXX option handling */ - m->m_flags &= ~(M_BCAST|M_MCAST); -#if IPSEC - /* Don't lookup socket */ - if (ipsec_bypass == 0) - (void)ipsec_setsocket(m, NULL); -#endif /*IPSEC*/ -#ifdef COMPAT_RFC1885 - ip6_output(m, NULL, &icmp6_reflect_rt, 0, NULL, &outif, 0); -#else - ip6_output(m, NULL, NULL, 0, NULL, &outif, 0); -#endif - if (outif) - icmp6_ifoutstat_inc(outif, type, code); + if (outif != NULL) { + ifnet_release(outif); + outif = NULL; + } + m->m_pkthdr.csum_data = 0; + m->m_pkthdr.csum_flags = 0; + ip6_output(m, NULL, NULL, IPV6_OUTARGS, NULL, &outif, &ip6oa); + if (outif != NULL) { + icmp6_ifoutstat_inc(outif, type, code); + ifnet_release(outif); + } return; - bad: +bad: m_freem(m); + if (outif != NULL) + ifnet_release(outif); return; } -void -icmp6_fasttimo() -{ - - mld6_fasttimeo(); -} - static const char * -icmp6_redirect_diag(src6, dst6, tgt6) - struct in6_addr *src6; - struct in6_addr *dst6; - struct in6_addr *tgt6; +icmp6_redirect_diag(struct in6_addr *src6, + struct in6_addr *dst6, + struct in6_addr *tgt6) { static char buf[1024]; snprintf(buf, sizeof(buf), "(src=%s dst=%s tgt=%s)", @@ -2134,9 +2345,7 @@ icmp6_redirect_diag(src6, dst6, tgt6) } void -icmp6_redirect_input(m, off) - struct mbuf *m; - int off; +icmp6_redirect_input(struct mbuf *m, int off) { struct ifnet *ifp = m->m_pkthdr.rcvif; struct ip6_hdr *ip6 = mtod(m, struct ip6_hdr *); @@ -2157,8 +2366,11 @@ icmp6_redirect_input(m, off) if (!m || !ifp) return; - /* XXX if we are router, we don't update route by icmp6 redirect */ - if (ip6_forwarding) + /* + * If we are an advertising router on this interface, + * don't update route by icmp6 redirect. + */ + if (ifp->if_eflags & IFEF_IPV6_ROUTER) goto freeit; if (!icmp6_rediraccept) goto freeit; @@ -2176,10 +2388,10 @@ icmp6_redirect_input(m, off) redtgt6 = nd_rd->nd_rd_target; reddst6 = nd_rd->nd_rd_dst; - if (IN6_IS_ADDR_LINKLOCAL(&redtgt6)) - redtgt6.s6_addr16[1] = htons(ifp->if_index); - if (IN6_IS_ADDR_LINKLOCAL(&reddst6)) - reddst6.s6_addr16[1] = htons(ifp->if_index); + if (in6_setscope(&redtgt6, m->m_pkthdr.rcvif, NULL) || + in6_setscope(&reddst6, m->m_pkthdr.rcvif, NULL)) { + goto freeit; + } /* validation */ if (!IN6_IS_ADDR_LINKLOCAL(&src6)) { @@ -2204,19 +2416,22 @@ icmp6_redirect_input(m, off) sin6.sin6_family = AF_INET6; sin6.sin6_len = sizeof(struct sockaddr_in6); bcopy(&reddst6, &sin6.sin6_addr, sizeof(reddst6)); - rt = rtalloc1((struct sockaddr *)&sin6, 0, 0UL); + rt = rtalloc1_scoped((struct sockaddr *)&sin6, 0, 0, ifp->if_index); if (rt) { + RT_LOCK(rt); if (rt->rt_gateway == NULL || rt->rt_gateway->sa_family != AF_INET6) { nd6log((LOG_ERR, "ICMP6 redirect rejected; no route " "with inet6 gateway found for redirect dst: %s\n", icmp6_redirect_diag(&src6, &reddst6, &redtgt6))); + RT_UNLOCK(rt); rtfree(rt); goto bad; } - gw6 = &(((struct sockaddr_in6 *)rt->rt_gateway)->sin6_addr); + gw6 = &(((struct sockaddr_in6 *)(void *) + rt->rt_gateway)->sin6_addr); if (bcmp(&src6, gw6, sizeof(struct in6_addr)) != 0) { nd6log((LOG_ERR, "ICMP6 redirect rejected; " @@ -2224,6 +2439,7 @@ icmp6_redirect_input(m, off) "%s\n", ip6_sprintf(gw6), icmp6_redirect_diag(&src6, &reddst6, &redtgt6))); + RT_UNLOCK(rt); rtfree(rt); goto bad; } @@ -2234,6 +2450,7 @@ icmp6_redirect_input(m, off) icmp6_redirect_diag(&src6, &reddst6, &redtgt6))); goto bad; } + RT_UNLOCK(rt); rtfree(rt); rt = NULL; } @@ -2307,10 +2524,9 @@ icmp6_redirect_input(m, off) bcopy(&redtgt6, &sgw.sin6_addr, sizeof(struct in6_addr)); bcopy(&reddst6, &sdst.sin6_addr, sizeof(struct in6_addr)); bcopy(&src6, &ssrc.sin6_addr, sizeof(struct in6_addr)); - rtredirect((struct sockaddr *)&sdst, (struct sockaddr *)&sgw, - (struct sockaddr *)NULL, RTF_GATEWAY | RTF_HOST, - (struct sockaddr *)&ssrc, - (struct rtentry **)NULL); + rtredirect(ifp, (struct sockaddr *)&sdst, + (struct sockaddr *)&sgw, NULL, RTF_GATEWAY | RTF_HOST, + (struct sockaddr *)&ssrc, NULL); } /* finally update cached route in each socket via pfctlinput */ { @@ -2320,6 +2536,7 @@ icmp6_redirect_input(m, off) sdst.sin6_family = AF_INET6; sdst.sin6_len = sizeof(struct sockaddr_in6); bcopy(&reddst6, &sdst.sin6_addr, sizeof(struct in6_addr)); + pfctlinput(PRC_REDIRECT_HOST, (struct sockaddr *)&sdst); #if IPSEC key_sa_routechange((struct sockaddr *)&sdst); @@ -2336,9 +2553,7 @@ icmp6_redirect_input(m, off) } void -icmp6_redirect_output(m0, rt) - struct mbuf *m0; - struct rtentry *rt; +icmp6_redirect_output(struct mbuf *m0, struct rtentry *rt) { struct ifnet *ifp; /* my outgoing interface */ struct in6_addr ifp_ll6; @@ -2351,15 +2566,24 @@ icmp6_redirect_output(m0, rt) u_char *p; struct ifnet *outif = NULL; struct sockaddr_in6 src_sa; + struct ip6_out_args ip6oa = { IFSCOPE_NONE, { 0 }, + IP6OAF_SELECT_SRCIF | IP6OAF_BOUND_SRCADDR, 0, + SO_TC_UNSPEC, _NET_SERVICE_TYPE_UNSPEC }; icmp6_errcount(&icmp6stat.icp6s_outerrhist, ND_REDIRECT, 0); + if (rt != NULL) + RT_LOCK(rt); + /* sanity check */ if (!m0 || !rt || !(rt->rt_flags & RTF_UP) || !(ifp = rt->rt_ifp)) goto fail; - /* if we are not router, we don't send icmp6 redirect */ - if (!ip6_forwarding || ip6_accept_rtadv || (ifp->if_eflags & IFEF_ACCEPT_RTADVD)) + /* + * If we are not a router to begin with, or not an advertising + * router on this interface, don't send icmp6 redirect. + */ + if (!ip6_forwarding || !(ifp->if_eflags & IFEF_IPV6_ROUTER)) goto fail; /* @@ -2375,8 +2599,13 @@ icmp6_redirect_output(m0, rt) src_sa.sin6_addr = sip6->ip6_src; /* we don't currently use sin6_scope_id, but eventually use it */ src_sa.sin6_scope_id = in6_addr2scopeid(ifp, &sip6->ip6_src); - if (nd6_is_addr_neighbor(&src_sa, ifp, 0) == 0) + RT_UNLOCK(rt); + if (nd6_is_addr_neighbor(&src_sa, ifp, 0) == 0) { + /* already unlocked */ + rt = NULL; goto fail; + } + RT_LOCK(rt); if (IN6_IS_ADDR_MULTICAST(&sip6->ip6_dst)) goto fail; /* what should we do here? */ @@ -2414,14 +2643,16 @@ icmp6_redirect_output(m0, rt) IN6_IFF_NOTREADY| IN6_IFF_ANYCAST)) == NULL) goto fail; + IFA_LOCK(&ia->ia_ifa); ifp_ll6 = ia->ia_addr.sin6_addr; - ifafree(&ia->ia_ifa); + IFA_UNLOCK(&ia->ia_ifa); + IFA_REMREF(&ia->ia_ifa); } /* get ip6 linklocal address for the router. */ if (rt->rt_gateway && (rt->rt_flags & RTF_GATEWAY)) { struct sockaddr_in6 *sin6; - sin6 = (struct sockaddr_in6 *)rt->rt_gateway; + sin6 = (struct sockaddr_in6 *)(void *)rt->rt_gateway; router_ll6 = &sin6->sin6_addr; if (!IN6_IS_ADDR_LINKLOCAL(router_ll6)) router_ll6 = (struct in6_addr *)NULL; @@ -2463,41 +2694,52 @@ icmp6_redirect_output(m0, rt) bcopy(&sip6->ip6_dst, &nd_rd->nd_rd_dst, sizeof(nd_rd->nd_rd_dst)); } + RT_UNLOCK(rt); + rt = NULL; p = (u_char *)(nd_rd + 1); if (!router_ll6) goto nolladdropt; - { - /* target lladdr option */ - struct rtentry *rt_router = NULL; - int len; - struct sockaddr_dl *sdl; - struct nd_opt_hdr *nd_opt; - char *lladdr; + { + /* target lladdr option */ + struct rtentry *rt_router = NULL; + int len; + struct sockaddr_dl *sdl; + struct nd_opt_hdr *nd_opt; + char *lladdr; + + /* Callee returns a locked route upon success */ + rt_router = nd6_lookup(router_ll6, 0, ifp, 0); + if (!rt_router) + goto nolladdropt; + RT_LOCK_ASSERT_HELD(rt_router); + len = sizeof(*nd_opt) + ifp->if_addrlen; + len = (len + 7) & ~7; /* round by 8 */ + /* safety check */ + if (len + (p - (u_char *)ip6) > maxlen) { + RT_REMREF_LOCKED(rt_router); + RT_UNLOCK(rt_router); + goto nolladdropt; + } - rt_router = nd6_lookup(router_ll6, 0, ifp, 0); - if (!rt_router) - goto nolladdropt; - len = sizeof(*nd_opt) + ifp->if_addrlen; - len = (len + 7) & ~7; /* round by 8 */ - /* safety check */ - if (len + (p - (u_char *)ip6) > maxlen) - goto nolladdropt; - if (!(rt_router->rt_flags & RTF_GATEWAY) && - (rt_router->rt_flags & RTF_LLINFO) && - (rt_router->rt_gateway->sa_family == AF_LINK) && - (sdl = (struct sockaddr_dl *)rt_router->rt_gateway) && - sdl->sdl_alen) { - nd_opt = (struct nd_opt_hdr *)p; - nd_opt->nd_opt_type = ND_OPT_TARGET_LINKADDR; - nd_opt->nd_opt_len = len >> 3; - lladdr = (char *)(nd_opt + 1); - bcopy(LLADDR(sdl), lladdr, ifp->if_addrlen); - p += len; + if (!(rt_router->rt_flags & RTF_GATEWAY) && + (rt_router->rt_flags & RTF_LLINFO) && + (rt_router->rt_gateway->sa_family == AF_LINK) && + (sdl = (struct sockaddr_dl *)(void *) + rt_router->rt_gateway) && sdl->sdl_alen) { + nd_opt = (struct nd_opt_hdr *)p; + nd_opt->nd_opt_type = ND_OPT_TARGET_LINKADDR; + nd_opt->nd_opt_len = len >> 3; + lladdr = (char *)(nd_opt + 1); + bcopy(LLADDR(sdl), lladdr, ifp->if_addrlen); + p += len; + } + RT_REMREF_LOCKED(rt_router); + RT_UNLOCK(rt_router); } - } + nolladdropt:; m->m_pkthdr.len = m->m_len = p - (u_char *)ip6; @@ -2581,20 +2823,11 @@ nolladdropt:; } noredhdropt:; - if (IN6_IS_ADDR_LINKLOCAL(&sip6->ip6_src)) - sip6->ip6_src.s6_addr16[1] = 0; - if (IN6_IS_ADDR_LINKLOCAL(&sip6->ip6_dst)) - sip6->ip6_dst.s6_addr16[1] = 0; -#if 0 - if (IN6_IS_ADDR_LINKLOCAL(&ip6->ip6_src)) - ip6->ip6_src.s6_addr16[1] = 0; - if (IN6_IS_ADDR_LINKLOCAL(&ip6->ip6_dst)) - ip6->ip6_dst.s6_addr16[1] = 0; -#endif - if (IN6_IS_ADDR_LINKLOCAL(&nd_rd->nd_rd_target)) - nd_rd->nd_rd_target.s6_addr16[1] = 0; - if (IN6_IS_ADDR_LINKLOCAL(&nd_rd->nd_rd_dst)) - nd_rd->nd_rd_dst.s6_addr16[1] = 0; + /* XXX: clear embedded link IDs in the inner header */ + in6_clearscope(&sip6->ip6_src); + in6_clearscope(&sip6->ip6_dst); + in6_clearscope(&nd_rd->nd_rd_target); + in6_clearscope(&nd_rd->nd_rd_dst); ip6->ip6_plen = htons(m->m_pkthdr.len - sizeof(struct ip6_hdr)); @@ -2603,39 +2836,33 @@ noredhdropt:; = in6_cksum(m, IPPROTO_ICMPV6, sizeof(*ip6), ntohs(ip6->ip6_plen)); /* send the packet to outside... */ -#if IPSEC - /* Don't lookup socket */ - if (ipsec_bypass == 0) - (void)ipsec_setsocket(m, NULL); -#endif /*IPSEC*/ - ip6_output(m, NULL, NULL, 0, NULL, &outif, 0); + ip6oa.ip6oa_boundif = ifp->if_index; + ip6oa.ip6oa_flags |= IP6OAF_BOUND_IF; + + ip6_output(m, NULL, NULL, IPV6_OUTARGS, NULL, &outif, &ip6oa); if (outif) { icmp6_ifstat_inc(outif, ifs6_out_msg); icmp6_ifstat_inc(outif, ifs6_out_redirect); + ifnet_release(outif); } icmp6stat.icp6s_outhist[ND_REDIRECT]++; return; fail: + if (rt != NULL) + RT_UNLOCK(rt); if (m) m_freem(m); if (m0) m_freem(m0); } -#if HAVE_NRL_INPCB -#define sotoin6pcb sotoinpcb -#define in6pcb inpcb -#define in6p_icmp6filt inp_icmp6filt -#endif /* * ICMPv6 socket option processing. */ int -icmp6_ctloutput(so, sopt) - struct socket *so; - struct sockopt *sopt; +icmp6_ctloutput(struct socket *so, struct sockopt *sopt) { int error = 0; int optlen; @@ -2661,7 +2888,7 @@ icmp6_ctloutput(so, sopt) { struct icmp6_filter *p; - if (optlen != sizeof(*p)) { + if (optlen != 0 && optlen != sizeof(*p)) { error = EMSGSIZE; break; } @@ -2669,8 +2896,17 @@ icmp6_ctloutput(so, sopt) error = EINVAL; break; } - error = sooptcopyin(sopt, inp->in6p_icmp6filt, optlen, - optlen); + + if (optlen == 0) { + /* According to RFC 3542, an installed filter can be + * cleared by issuing a setsockopt for ICMP6_FILTER + * with a zero length. + */ + ICMP6_FILTER_SETPASSALL(inp->in6p_icmp6filt); + } else { + error = sooptcopyin(sopt, inp->in6p_icmp6filt, optlen, + optlen); + } break; } @@ -2689,7 +2925,7 @@ icmp6_ctloutput(so, sopt) break; } error = sooptcopyout(sopt, inp->in6p_icmp6filt, - sizeof(struct icmp6_filter)); + min(sizeof(struct icmp6_filter), optlen)); break; } @@ -2702,11 +2938,6 @@ icmp6_ctloutput(so, sopt) return(error); } -#if HAVE_NRL_INPCB -#undef sotoin6pcb -#undef in6pcb -#undef in6p_icmp6filt -#endif /* * ICMPv6 socket datagram option processing. @@ -2714,7 +2945,7 @@ icmp6_ctloutput(so, sopt) int icmp6_dgram_ctloutput(struct socket *so, struct sockopt *sopt) { - if (so->so_uid == 0) + if (kauth_cred_issuser(so->so_cred)) return icmp6_ctloutput(so, sopt); if (sopt->sopt_level == IPPROTO_ICMPV6) { @@ -2725,21 +2956,23 @@ icmp6_dgram_ctloutput(struct socket *so, struct sockopt *sopt) return EPERM; } } - + if (sopt->sopt_level != IPPROTO_IPV6) return EINVAL; - + switch (sopt->sopt_name) { - case IPV6_PKTOPTIONS: case IPV6_UNICAST_HOPS: case IPV6_CHECKSUM: - case IPV6_FAITH: case IPV6_V6ONLY: + case IPV6_USE_MIN_MTU: + case IPV6_RECVRTHDR: + case IPV6_RECVPKTINFO: + case IPV6_RECVHOPLIMIT: + case IPV6_PATHMTU: case IPV6_PKTINFO: case IPV6_HOPLIMIT: case IPV6_HOPOPTS: case IPV6_DSTOPTS: - case IPV6_RTHDR: case IPV6_MULTICAST_IF: case IPV6_MULTICAST_HOPS: case IPV6_MULTICAST_LOOP: @@ -2747,33 +2980,52 @@ icmp6_dgram_ctloutput(struct socket *so, struct sockopt *sopt) case IPV6_LEAVE_GROUP: case IPV6_PORTRANGE: case IPV6_IPSEC_POLICY: + case IPV6_RECVTCLASS: + case IPV6_TCLASS: + case IPV6_2292PKTOPTIONS: + case IPV6_2292PKTINFO: + case IPV6_2292HOPLIMIT: + case IPV6_2292HOPOPTS: + case IPV6_2292DSTOPTS: + case IPV6_2292RTHDR: + case IPV6_BOUND_IF: + case IPV6_NO_IFT_CELLULAR: + return ip6_ctloutput(so, sopt); - + default: return EPERM; - - } } __private_extern__ int -icmp6_dgram_send(struct socket *so, __unused int flags, struct mbuf *m, struct sockaddr *nam, - struct mbuf *control, __unused struct proc *p) +icmp6_dgram_send(struct socket *so, int flags, struct mbuf *m, + struct sockaddr *nam, struct mbuf *control, struct proc *p) { +#pragma unused(flags, p) int error = 0; struct inpcb *inp = sotoinpcb(so); struct sockaddr_in6 tmp; - struct sockaddr_in6 *dst; + struct sockaddr_in6 *dst = (struct sockaddr_in6 *)(void *)nam; struct icmp6_hdr *icmp6; - if (so->so_uid == 0) - return rip6_output(m, so, (struct sockaddr_in6 *) nam, control); + if (inp == NULL +#if NECP + || (necp_socket_should_use_flow_divert(inp)) +#endif /* NECP */ + ) { + error = (inp == NULL ? EINVAL : EPROTOTYPE); + goto bad; + } + + if (kauth_cred_issuser(so->so_cred)) + return (rip6_output(m, so, SIN6(nam), control, 0)); /* always copy sockaddr to avoid overwrites */ if (so->so_state & SS_ISCONNECTED) { - if (nam) { - m_freem(m); - return EISCONN; + if (nam != NULL) { + error = EISCONN; + goto bad; } /* XXX */ bzero(&tmp, sizeof(tmp)); @@ -2784,29 +3036,36 @@ icmp6_dgram_send(struct socket *so, __unused int flags, struct mbuf *m, struct s dst = &tmp; } else { if (nam == NULL) { - m_freem(m); - return ENOTCONN; + error = ENOTCONN; + goto bad; } - tmp = *(struct sockaddr_in6 *)nam; + tmp = *(struct sockaddr_in6 *)(void *)nam; dst = &tmp; } /* * For an ICMPv6 packet, we should know its type and code */ - if (so->so_proto->pr_protocol == IPPROTO_ICMPV6) { + if (SOCK_PROTO(so) == IPPROTO_ICMPV6) { if (m->m_len < sizeof(struct icmp6_hdr) && (m = m_pullup(m, sizeof(struct icmp6_hdr))) == NULL) { error = ENOBUFS; goto bad; } icmp6 = mtod(m, struct icmp6_hdr *); - + /* - * Allow only to send echo request type 128 with code 0 + * Allow only to send echo request and node information request * See RFC 2463 for Echo Request Message format */ - if (icmp6->icmp6_type != 128 || icmp6->icmp6_code != 0) { + if ((icmp6->icmp6_type == ICMP6_ECHO_REQUEST && + icmp6->icmp6_code == 0) || + (icmp6->icmp6_type == ICMP6_NI_QUERY && + (icmp6->icmp6_code == ICMP6_NI_SUBJ_IPV6 || + icmp6->icmp6_code == ICMP6_NI_SUBJ_FQDN))) { + /* Good */ + ; + } else { error = EPERM; goto bad; } @@ -2814,14 +3073,20 @@ icmp6_dgram_send(struct socket *so, __unused int flags, struct mbuf *m, struct s #if ENABLE_DEFAULT_SCOPE if (dst->sin6_scope_id == 0) { /* not change if specified */ - dst->sin6_scope_id = scope6_addr2default(&dst->sin6_addr); + dst->sin6_scope_id = scope6_addr2default(&dst->sin6_addr); } #endif - return rip6_output(m, so, (struct sockaddr_in6 *) nam, control); + return (rip6_output(m, so, dst, control, 0)); bad: - m_freem(m); - return error; + VERIFY(error != 0); + + if (m != NULL) + m_freem(m); + if (control != NULL) + m_freem(control); + + return (error); } /* Like rip6_attach but without root privilege enforcement */ @@ -2834,10 +3099,10 @@ icmp6_dgram_attach(struct socket *so, int proto, struct proc *p) inp = sotoinpcb(so); if (inp) panic("icmp6_dgram_attach"); - + if (proto != IPPROTO_ICMPV6) return EINVAL; - + error = soreserve(so, rip_sendspace, rip_recvspace); if (error) return error; @@ -2858,95 +3123,30 @@ icmp6_dgram_attach(struct socket *so, int proto, struct proc *p) } - -#ifndef HAVE_PPSRATECHECK -#ifndef timersub -#define timersub(tvp, uvp, vvp) \ - do { \ - (vvp)->tv_sec = (tvp)->tv_sec - (uvp)->tv_sec; \ - (vvp)->tv_usec = (tvp)->tv_usec - (uvp)->tv_usec; \ - if ((vvp)->tv_usec < 0) { \ - (vvp)->tv_sec--; \ - (vvp)->tv_usec += 1000000; \ - } \ - } while (0) -#endif - -/* - * ppsratecheck(): packets (or events) per second limitation. - */ -static int -ppsratecheck(lasttime, curpps, maxpps) - struct timeval *lasttime; - int *curpps; - int maxpps; /* maximum pps allowed */ -{ - struct timeval tv, delta; - int rv; - - microtime(&tv); - - timersub(&tv, lasttime, &delta); - - /* - * Check for 0,0 so that the message will be seen at least once. - * If more than one second has passed since the last update of - * lasttime, reset the counter. - * - * we do increment *curpps even in *curpps < maxpps case, as some may - * try to use *curpps for stat purposes as well. - */ - if ((lasttime->tv_sec == 0 && lasttime->tv_usec == 0) || - delta.tv_sec >= 1) { - *lasttime = tv; - *curpps = 0; - rv = 1; - } else if (maxpps < 0) - rv = 1; - else if (*curpps < maxpps) - rv = 1; - else - rv = 0; - -#if 1 /* DIAGNOSTIC? */ - /* be careful about wrap-around */ - if (*curpps + 1 > 0) - *curpps = *curpps + 1; -#else - /* - * assume that there's not too many calls to this function. - * not sure if the assumption holds, as it depends on *caller's* - * behavior, not the behavior of this function. - * IMHO it is wrong to make assumption on the caller's behavior, - * so the above #if is #if 1, not #ifdef DIAGNOSTIC. - */ - *curpps = *curpps + 1; -#endif - - return (rv); -} -#endif - /* * Perform rate limit check. * Returns 0 if it is okay to send the icmp6 packet. * Returns 1 if the router SHOULD NOT send this icmp6 packet due to rate * limitation. * - * XXX per-destination/type check necessary? + * XXX per-destination check necessary? */ static int icmp6_ratelimit( __unused const struct in6_addr *dst, /* not used at this moment */ - __unused const int type, /* not used at this moment */ - __unused const int code) /* not used at this moment */ + const int type, + __unused const int code) { int ret; ret = 0; /* okay to send */ /* PPS limit */ - if (!ppsratecheck(&icmp6errppslim_last, &icmp6errpps_count, + if (type == ND_ROUTER_ADVERT) { + if (!ppsratecheck(&icmp6rappslim_last, &icmp6rapps_count, + icmp6rappslim)) + ret++; + } else if (!ppsratecheck(&icmp6errppslim_last, &icmp6errpps_count, icmp6errppslim)) { /* The packet is subject to rate limit */ ret++; @@ -2954,4 +3154,3 @@ icmp6_ratelimit( return ret; } -