Please note that diffs are not public domain; they are subject to the copyright notices on the relevant files. =================================================================== RCS file: /ftp/cvs/cvsroot/src/sys/netinet6/ip6_input.c,v rcsdiff: /ftp/cvs/cvsroot/src/sys/netinet6/ip6_input.c,v: warning: Unknown phrases like `commitid ...;' are present. retrieving revision 1.41 retrieving revision 1.101 diff -u -p -r1.41 -r1.101 --- src/sys/netinet6/ip6_input.c 2001/04/13 23:30:26 1.41 +++ src/sys/netinet6/ip6_input.c 2007/03/24 00:42:14 1.101 @@ -1,4 +1,4 @@ -/* $NetBSD: ip6_input.c,v 1.41 2001/04/13 23:30:26 thorpej Exp $ */ +/* $NetBSD: ip6_input.c,v 1.101 2007/03/24 00:42:14 liamjfoy Exp $ */ /* $KAME: ip6_input.c,v 1.188 2001/03/29 05:34:31 itojun Exp $ */ /* @@ -42,11 +42,7 @@ * 2. Redistributions in binary form must reproduce the above copyright * notice, this list of conditions and the following disclaimer in the * documentation and/or other materials provided with the distribution. - * 3. All advertising materials mentioning features or use of this software - * must display the following acknowledgement: - * This product includes software developed by the University of - * California, Berkeley and its contributors. - * 4. Neither the name of the University nor the names of its contributors + * 3. Neither the name of the University nor the names of its contributors * may be used to endorse or promote products derived from this software * without specific prior written permission. * @@ -65,7 +61,11 @@ * @(#)ip_input.c 8.2 (Berkeley) 1/4/94 */ +#include +__KERNEL_RCSID(0, "$NetBSD: ip6_input.c,v 1.101 2007/03/24 00:42:14 liamjfoy Exp $"); + #include "opt_inet.h" +#include "opt_inet6.h" #include "opt_ipsec.h" #include "opt_pfil_hooks.h" @@ -82,6 +82,7 @@ #include #include #include +#include #include #include @@ -97,28 +98,34 @@ #ifdef INET #include #include -#endif /*INET*/ +#endif /* INET */ #include #include #include #include #include +#include #include #include -#include #ifdef IPSEC #include #endif +#ifdef FAST_IPSEC +#include +#include +#include +#endif /* FAST_IPSEC */ + #include -/* we need it for NLOOP. */ -#include "loop.h" #include "faith.h" - #include "gif.h" -#include "bpfilter.h" + +#if NGIF > 0 +#include +#endif #include @@ -129,7 +136,8 @@ static int ip6qmaxlen = IFQ_MAXLEN; struct in6_ifaddr *in6_ifaddr; struct ifqueue ip6intrq; -extern struct ifnet loif[NLOOP]; +extern struct callout in6_tmpaddrtimer_ch; + int ip6_forward_srcrt; /* XXX */ int ip6_sourcecheck; /* XXX */ int ip6_sourcecheck_interval; /* XXX */ @@ -141,8 +149,10 @@ struct pfil_head inet6_pfil_hook; struct ip6stat ip6stat; static void ip6_init2 __P((void *)); +static struct m_tag *ip6_setdstifaddr __P((struct mbuf *, struct in6_ifaddr *)); static int ip6_hopopts_input __P((u_int32_t *, u_int32_t *, struct mbuf **, int *)); +static struct mbuf *ip6_pullexthdr __P((struct mbuf *, size_t, int)); /* * IP6 initialization: fill in IP6 protocol switch table. @@ -151,31 +161,30 @@ static int ip6_hopopts_input __P((u_int3 void ip6_init() { - struct ip6protosw *pr; + const struct ip6protosw *pr; int i; - struct timeval tv; - pr = (struct ip6protosw *)pffindproto(PF_INET6, IPPROTO_RAW, SOCK_RAW); + pr = (const struct ip6protosw *)pffindproto(PF_INET6, IPPROTO_RAW, SOCK_RAW); if (pr == 0) panic("ip6_init"); for (i = 0; i < IPPROTO_MAX; i++) ip6_protox[i] = pr - inet6sw; - for (pr = (struct ip6protosw *)inet6domain.dom_protosw; - pr < (struct ip6protosw *)inet6domain.dom_protoswNPROTOSW; pr++) + for (pr = (const struct ip6protosw *)inet6domain.dom_protosw; + pr < (const struct ip6protosw *)inet6domain.dom_protoswNPROTOSW; pr++) if (pr->pr_domain->dom_family == PF_INET6 && pr->pr_protocol && pr->pr_protocol != IPPROTO_RAW) ip6_protox[pr->pr_protocol] = pr - inet6sw; ip6intrq.ifq_maxlen = ip6qmaxlen; + scope6_init(); + addrsel_policy_init(); nd6_init(); frag6_init(); - /* - * in many cases, random() here does NOT return random number - * as initialization during bootstrap time occur in fixed order. - */ - microtime(&tv); - ip6_flow_seq = random() ^ tv.tv_usec; + ip6_desync_factor = arc4random() % MAX_TEMP_DESYNC_FACTOR; ip6_init2((void *)0); +#ifdef GATEWAY + ip6flow_init(ip6_hashsize); +#endif #ifdef PFIL_HOOKS /* Register our Packet Filter hook. */ @@ -189,21 +198,19 @@ ip6_init() } static void -ip6_init2(dummy) - void *dummy; +ip6_init2(void *dummy) { - /* - * to route local address of p2p link to loopback, - * assign loopback address first. - */ - in6_ifattach(&loif[0], NULL); /* nd6_timer_init */ callout_init(&nd6_timer_ch); callout_reset(&nd6_timer_ch, hz, nd6_timer, NULL); - /* router renumbering prefix list maintenance */ - callout_init(&in6_rr_timer_ch); - callout_reset(&in6_rr_timer_ch, hz, in6_rr_timer, NULL); + + /* timer for regeneranation of temporary addresses randomize ID */ + callout_init(&in6_tmpaddrtimer_ch); + callout_reset(&in6_tmpaddrtimer_ch, + (ip6_temp_preferred_lifetime - ip6_desync_factor - + ip6_temp_regen_advance) * hz, + in6_tmpaddrtimer, NULL); } /* @@ -221,6 +228,11 @@ ip6intr() splx(s); if (m == 0) return; + /* drop the packet if IPv6 operation is disabled on the IF */ + if ((ND_IFINFO(m->m_pkthdr.rcvif)->flags & ND6_IFF_IFDISABLED)) { + m_freem(m); + return; + } ip6_input(m); } } @@ -228,8 +240,7 @@ ip6intr() extern struct route_in6 ip6_forward_rt; void -ip6_input(m) - struct mbuf *m; +ip6_input(struct mbuf *m) { struct ip6_hdr *ip6; int off = sizeof(struct ip6_hdr), nest; @@ -237,20 +248,30 @@ ip6_input(m) u_int32_t rtalert = ~0; int nxt, ours = 0; struct ifnet *deliverifp = NULL; + int srcrt = 0; +#ifdef FAST_IPSEC + struct m_tag *mtag; + struct tdb_ident *tdbi; + struct secpolicy *sp; + int s, error; +#endif #ifdef IPSEC /* * should the inner packet be considered authentic? * see comment in ah4_input(). */ - if (m) { - m->m_flags &= ~M_AUTHIPHDR; - m->m_flags &= ~M_AUTHIPDGM; - } + m->m_flags &= ~M_AUTHIPHDR; + m->m_flags &= ~M_AUTHIPDGM; #endif /* - * mbuf statistics by kazu + * make sure we don't have onion peering information into m_tag. + */ + ip6_delaux(m); + + /* + * mbuf statistics */ if (m->m_flags & M_EXT) { if (m->m_next) @@ -258,10 +279,10 @@ ip6_input(m) else ip6stat.ip6s_mext1++; } else { -#define M2MMAX (sizeof(ip6stat.ip6s_m2m)/sizeof(ip6stat.ip6s_m2m[0])) +#define M2MMAX __arraycount(ip6stat.ip6s_m2m) if (m->m_next) { if (m->m_flags & M_LOOP) { - ip6stat.ip6s_m2m[loif[0].if_index]++; /*XXX*/ + ip6stat.ip6s_m2m[lo0ifp->if_index]++; /* XXX */ } else if (m->m_pkthdr.rcvif->if_index < M2MMAX) ip6stat.ip6s_m2m[m->m_pkthdr.rcvif->if_index]++; else @@ -274,15 +295,24 @@ ip6_input(m) in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_receive); ip6stat.ip6s_total++; -#ifndef PULLDOWN_TEST - /* XXX is the line really necessary? */ - IP6_EXTHDR_CHECK(m, 0, sizeof(struct ip6_hdr), /*nothing*/); -#endif - - if (m->m_len < sizeof(struct ip6_hdr)) { - struct ifnet *inifp; - inifp = m->m_pkthdr.rcvif; - if ((m = m_pullup(m, sizeof(struct ip6_hdr))) == 0) { + /* + * If the IPv6 header is not aligned, slurp it up into a new + * mbuf with space for link headers, in the event we forward + * it. Otherwise, if it is aligned, make sure the entire base + * IPv6 header is in the first mbuf of the chain. + */ + if (IP6_HDR_ALIGNED_P(mtod(m, void *)) == 0) { + struct ifnet *inifp = m->m_pkthdr.rcvif; + if ((m = m_copyup(m, sizeof(struct ip6_hdr), + (max_linkhdr + 3) & ~3)) == NULL) { + /* XXXJRT new stat, please */ + ip6stat.ip6s_toosmall++; + in6_ifstat_inc(inifp, ifs6_in_hdrerr); + return; + } + } else if (__predict_false(m->m_len < sizeof(struct ip6_hdr))) { + struct ifnet *inifp = m->m_pkthdr.rcvif; + if ((m = m_pullup(m, sizeof(struct ip6_hdr))) == NULL) { ip6stat.ip6s_toosmall++; in6_ifstat_inc(inifp, ifs6_in_hdrerr); return; @@ -297,6 +327,17 @@ ip6_input(m) goto bad; } +#if defined(IPSEC) + /* IPv6 fast forwarding is not compatible with IPsec. */ + m->m_flags &= ~M_CANFASTFWD; +#else + /* + * Assume that we can create a fast-forward IP flow entry + * based on this packet. + */ + m->m_flags |= M_CANFASTFWD; +#endif + #ifdef PFIL_HOOKS /* * Run through list of hooks for input packets. If there are any @@ -310,25 +351,29 @@ ip6_input(m) * not the decapsulated packet. */ #ifdef IPSEC - if (!ipsec_gethist(m, NULL)) + if (!ipsec_getnhist(m)) +#elif defined(FAST_IPSEC) + if (!ipsec_indone(m)) #else if (1) #endif { + struct in6_addr odst; + + odst = ip6->ip6_dst; if (pfil_run_hooks(&inet6_pfil_hook, &m, m->m_pkthdr.rcvif, PFIL_IN) != 0) return; if (m == NULL) return; ip6 = mtod(m, struct ip6_hdr *); + srcrt = !IN6_ARE_ADDR_EQUAL(&odst, &ip6->ip6_dst); } #endif /* PFIL_HOOKS */ - ip6stat.ip6s_nxthist[ip6->ip6_nxt]++; #ifdef ALTQ - /* XXX Temporary until ALTQ is changed to use a pfil hook */ if (altq_input != NULL && (*altq_input)(m, AF_INET6) == 0) { /* packet is dropped by traffic conditioner */ return; @@ -336,25 +381,28 @@ ip6_input(m) #endif /* - * Scope check + * Check against address spoofing/corruption. */ if (IN6_IS_ADDR_MULTICAST(&ip6->ip6_src) || IN6_IS_ADDR_UNSPECIFIED(&ip6->ip6_dst)) { + /* + * XXX: "badscope" is not very suitable for a multicast source. + */ ip6stat.ip6s_badscope++; in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_addrerr); goto bad; } /* - * The following check is not documented in the spec. Malicious party - * may be able to use IPv4 mapped addr to confuse tcp/udp stack and - * bypass security checks (act as if it was from 127.0.0.1 by using - * IPv6 src ::ffff:127.0.0.1). Be cautious. + * The following check is not documented in specs. A malicious + * party may be able to use IPv4 mapped addr to confuse tcp/udp stack + * and bypass security checks (act as if it was from 127.0.0.1 by using + * IPv6 src ::ffff:127.0.0.1). Be cautious. * - * This check chokes if we are in SIIT cloud. As none of BSDs support - * IPv4-less kernel compilation, we cannot support SIIT environment - * at all. So, it makes more sense for us to reject any malicious - * packets for non-SIIT environment, than try to do a partical support - * for SIIT environment. + * This check chokes if we are in an SIIT cloud. As none of BSDs + * support IPv4-less kernel compilation, we cannot support SIIT + * environment at all. So, it makes more sense for us to reject any + * malicious packets for non-SIIT environment, than try to do a + * partial support for SIIT environment. */ if (IN6_IS_ADDR_V4MAPPED(&ip6->ip6_src) || IN6_IS_ADDR_V4MAPPED(&ip6->ip6_dst)) { @@ -378,61 +426,28 @@ ip6_input(m) } #endif - if (IN6_IS_ADDR_LOOPBACK(&ip6->ip6_src) || - IN6_IS_ADDR_LOOPBACK(&ip6->ip6_dst)) { - if (m->m_pkthdr.rcvif->if_flags & IFF_LOOPBACK) { - ours = 1; - deliverifp = m->m_pkthdr.rcvif; - goto hbhcheck; - } else { - ip6stat.ip6s_badscope++; - in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_addrerr); - goto bad; - } - } - - /* drop packets if interface ID portion is already filled */ - if ((m->m_pkthdr.rcvif->if_flags & IFF_LOOPBACK) == 0) { - if (IN6_IS_SCOPE_LINKLOCAL(&ip6->ip6_src) && - ip6->ip6_src.s6_addr16[1]) { - ip6stat.ip6s_badscope++; - goto bad; - } - if (IN6_IS_SCOPE_LINKLOCAL(&ip6->ip6_dst) && - ip6->ip6_dst.s6_addr16[1]) { - ip6stat.ip6s_badscope++; - goto bad; - } + /* + * Disambiguate address scope zones (if there is ambiguity). + * We first make sure that the original source or destination address + * is not in our internal form for scoped addresses. Such addresses + * are not necessarily invalid spec-wise, but we cannot accept them due + * to the usage conflict. + * in6_setscope() then also checks and rejects the cases where src or + * dst are the loopback address and the receiving interface + * is not loopback. + */ + if (__predict_false( + m_makewritable(&m, 0, sizeof(struct ip6_hdr), M_DONTWAIT))) + goto bad; + ip6 = mtod(m, struct ip6_hdr *); + if (in6_clearscope(&ip6->ip6_src) || in6_clearscope(&ip6->ip6_dst)) { + ip6stat.ip6s_badscope++; /* XXX */ + goto bad; } - - if (IN6_IS_SCOPE_LINKLOCAL(&ip6->ip6_src)) - ip6->ip6_src.s6_addr16[1] - = htons(m->m_pkthdr.rcvif->if_index); - if (IN6_IS_SCOPE_LINKLOCAL(&ip6->ip6_dst)) - ip6->ip6_dst.s6_addr16[1] - = htons(m->m_pkthdr.rcvif->if_index); - - /* - * We use rt->rt_ifp to determine if the address is ours or not. - * If rt_ifp is lo0, the address is ours. - * The problem here is, rt->rt_ifp for fe80::%lo0/64 is set to lo0, - * so any address under fe80::%lo0/64 will be mistakenly considered - * local. The special case is supplied to handle the case properly - * by actually looking at interface addresses - * (using in6ifa_ifpwithaddr). - */ - if ((m->m_pkthdr.rcvif->if_flags & IFF_LOOPBACK) != 0 && - IN6_IS_ADDR_LINKLOCAL(&ip6->ip6_dst)) { - if (!in6ifa_ifpwithaddr(m->m_pkthdr.rcvif, &ip6->ip6_dst)) { - icmp6_error(m, ICMP6_DST_UNREACH, - ICMP6_DST_UNREACH_ADDR, 0); - /* m is already freed */ - return; - } - - ours = 1; - deliverifp = m->m_pkthdr.rcvif; - goto hbhcheck; + if (in6_setscope(&ip6->ip6_src, m->m_pkthdr.rcvif, NULL) || + in6_setscope(&ip6->ip6_dst, m->m_pkthdr.rcvif, NULL)) { + ip6stat.ip6s_badscope++; + goto bad; } /* @@ -462,28 +477,26 @@ ip6_input(m) /* * Unicast check */ - if (ip6_forward_rt.ro_rt != NULL && - (ip6_forward_rt.ro_rt->rt_flags & RTF_UP) != 0 && - IN6_ARE_ADDR_EQUAL(&ip6->ip6_dst, - &((struct sockaddr_in6 *)(&ip6_forward_rt.ro_dst))->sin6_addr)) + if (!IN6_ARE_ADDR_EQUAL(&ip6->ip6_dst, + &((const struct sockaddr_in6 *)rtcache_getdst((const struct route *)&ip6_forward_rt))->sin6_addr)) + rtcache_free((struct route *)&ip6_forward_rt); + else + rtcache_check((struct route *)&ip6_forward_rt); + if (ip6_forward_rt.ro_rt != NULL) { + /* XXX Revalidated route is accounted wrongly. */ ip6stat.ip6s_forward_cachehit++; - else { + } else { struct sockaddr_in6 *dst6; - if (ip6_forward_rt.ro_rt) { - /* route is down or destination is different */ - ip6stat.ip6s_forward_cachemiss++; - RTFREE(ip6_forward_rt.ro_rt); - ip6_forward_rt.ro_rt = 0; - } + ip6stat.ip6s_forward_cachemiss++; - bzero(&ip6_forward_rt.ro_dst, sizeof(struct sockaddr_in6)); - dst6 = (struct sockaddr_in6 *)&ip6_forward_rt.ro_dst; + dst6 = &ip6_forward_rt.ro_dst; + memset(dst6, 0, sizeof(*dst6)); dst6->sin6_len = sizeof(struct sockaddr_in6); dst6->sin6_family = AF_INET6; dst6->sin6_addr = ip6->ip6_dst; - rtalloc((struct route *)&ip6_forward_rt); + rtcache_init((struct route *)&ip6_forward_rt); } #define rt6_key(r) ((struct sockaddr_in6 *)((r)->rt_nodes->rn_key)) @@ -497,9 +510,10 @@ ip6_input(m) * But we think it's even useful in some situations, e.g. when using * a special daemon which wants to intercept the packet. */ - if (ip6_forward_rt.ro_rt && + if (ip6_forward_rt.ro_rt != NULL && (ip6_forward_rt.ro_rt->rt_flags & (RTF_HOST|RTF_GATEWAY)) == RTF_HOST && + !(ip6_forward_rt.ro_rt->rt_flags & RTF_CLONED) && #if 0 /* * The check below is redundant since the comparison of @@ -507,7 +521,7 @@ ip6_input(m) * already done through looking up the routing table. */ IN6_ARE_ADDR_EQUAL(&ip6->ip6_dst, - &rt6_key(ip6_forward_rt.ro_rt)->sin6_addr) && + &rt6_key(ip6_forward_rt.ro_rt)->sin6_addr) && #endif ip6_forward_rt.ro_rt->rt_ifp->if_type == IFT_LOOP) { struct in6_ifaddr *ia6 = @@ -535,15 +549,16 @@ ip6_input(m) } /* - * FAITH(Firewall Aided Internet Translator) + * FAITH (Firewall Aided Internet Translator) */ #if defined(NFAITH) && 0 < NFAITH if (ip6_keepfaith) { - if (ip6_forward_rt.ro_rt && ip6_forward_rt.ro_rt->rt_ifp - && ip6_forward_rt.ro_rt->rt_ifp->if_type == IFT_FAITH) { + if (ip6_forward_rt.ro_rt != NULL && + ip6_forward_rt.ro_rt->rt_ifp != NULL && + ip6_forward_rt.ro_rt->rt_ifp->if_type == IFT_FAITH) { /* XXX do we need more sanity checks? */ ours = 1; - deliverifp = ip6_forward_rt.ro_rt->rt_ifp; /*faith*/ + deliverifp = ip6_forward_rt.ro_rt->rt_ifp; /* faith */ goto hbhcheck; } } @@ -557,9 +572,7 @@ ip6_input(m) * working right. */ struct ifaddr *ifa; - for (ifa = m->m_pkthdr.rcvif->if_addrlist.tqh_first; - ifa; - ifa = ifa->ifa_list.tqe_next) { + TAILQ_FOREACH(ifa, &m->m_pkthdr.rcvif->if_addrlist, ifa_list) { if (ifa->ifa_addr == NULL) continue; /* just for safety */ if (ifa->ifa_addr->sa_family != AF_INET6) @@ -585,6 +598,27 @@ ip6_input(m) hbhcheck: /* + * record address information into m_tag, if we don't have one yet. + * note that we are unable to record it, if the address is not listed + * as our interface address (e.g. multicast addresses, addresses + * within FAITH prefixes and such). + */ + if (deliverifp && !ip6_getdstifaddr(m)) { + struct in6_ifaddr *ia6; + + ia6 = in6_ifawithifp(deliverifp, &ip6->ip6_dst); + if (ia6) { + if (!ip6_setdstifaddr(m, ia6)) { + /* + * XXX maybe we should drop the packet here, + * as we could not provide enough information + * to the upper layers. + */ + } + } + } + + /* * Process Hop-by-Hop options header if it's contained. * m may be modified in ip6_hopopts_input(). * If a JumboPayload option is included, plen will also be modified. @@ -604,35 +638,31 @@ ip6_input(m) ip6 = mtod(m, struct ip6_hdr *); /* - * if the payload length field is 0 and the next header field + * if the payload length field is 0 and the next header field * indicates Hop-by-Hop Options header, then a Jumbo Payload * option MUST be included. */ if (ip6->ip6_plen == 0 && plen == 0) { /* * Note that if a valid jumbo payload option is - * contained, ip6_hoptops_input() must set a valid - * (non-zero) payload length to the variable plen. + * contained, ip6_hopopts_input() must set a valid + * (non-zero) payload length to the variable plen. */ ip6stat.ip6s_badoptions++; in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_discard); in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_hdrerr); icmp6_error(m, ICMP6_PARAM_PROB, ICMP6_PARAMPROB_HEADER, - (caddr_t)&ip6->ip6_plen - (caddr_t)ip6); + (char *)&ip6->ip6_plen - (char *)ip6); return; } -#ifndef PULLDOWN_TEST - /* ip6_hopopts_input() ensures that mbuf is contiguous */ - hbh = (struct ip6_hbh *)(ip6 + 1); -#else IP6_EXTHDR_GET(hbh, struct ip6_hbh *, m, sizeof(struct ip6_hdr), sizeof(struct ip6_hbh)); if (hbh == NULL) { ip6stat.ip6s_tooshort++; return; } -#endif + KASSERT(IP6_HDR_ALIGNED_P(hbh)); nxt = hbh->ip6h_nxt; /* @@ -685,9 +715,9 @@ ip6_input(m) return; } } else if (!ours) { - ip6_forward(m, 0); + ip6_forward(m, srcrt); return; - } + } ip6 = mtod(m, struct ip6_hdr *); @@ -750,7 +780,47 @@ ip6_input(m) goto bad; } #endif - +#ifdef FAST_IPSEC + /* + * enforce IPsec policy checking if we are seeing last header. + * note that we do not visit this with protocols with pcb layer + * code - like udp/tcp/raw ip. + */ + if ((inet6sw[ip_protox[nxt]].pr_flags & PR_LASTHDR) != 0) { + /* + * Check if the packet has already had IPsec processing + * done. If so, then just pass it along. This tag gets + * set during AH, ESP, etc. input handling, before the + * packet is returned to the ip input queue for delivery. + */ + mtag = m_tag_find(m, PACKET_TAG_IPSEC_IN_DONE, NULL); + s = splsoftnet(); + if (mtag != NULL) { + tdbi = (struct tdb_ident *)(mtag + 1); + sp = ipsec_getpolicy(tdbi, IPSEC_DIR_INBOUND); + } else { + sp = ipsec_getpolicybyaddr(m, IPSEC_DIR_INBOUND, + IP_FORWARDING, &error); + } + if (sp != NULL) { + /* + * Check security policy against packet attributes. + */ + error = ipsec_in_reject(sp, m); + KEY_FREESP(&sp); + } else { + /* XXX error stat??? */ + error = EINVAL; + DPRINTF(("ip6_input: no SP, packet discarded\n"));/*XXX*/ + goto bad; + } + splx(s); + if (error) + goto bad; + } +#endif /* FAST_IPSEC */ + + nxt = (*inet6sw[ip6_protox[nxt]].pr_input)(&m, &off, nxt); } return; @@ -759,6 +829,35 @@ ip6_input(m) } /* + * set/grab in6_ifaddr correspond to IPv6 destination address. + */ +static struct m_tag * +ip6_setdstifaddr(m, ia6) + struct mbuf *m; + struct in6_ifaddr *ia6; +{ + struct m_tag *mtag; + + mtag = ip6_addaux(m); + if (mtag) + ((struct ip6aux *)(mtag + 1))->ip6a_dstia6 = ia6; + return mtag; /* NULL if failed to set */ +} + +struct in6_ifaddr * +ip6_getdstifaddr(m) + struct mbuf *m; +{ + struct m_tag *mtag; + + mtag = ip6_findaux(m); + if (mtag) + return ((struct ip6aux *)(mtag + 1))->ip6a_dstia6; + else + return NULL; +} + +/* * Hop-by-Hop options header processing. If a valid jumbo payload option is * included, the real payload length will be stored in plenp. */ @@ -772,17 +871,8 @@ ip6_hopopts_input(plenp, rtalertp, mp, o struct mbuf *m = *mp; int off = *offp, hbhlen; struct ip6_hbh *hbh; - u_int8_t *opt; /* validation of the length of the header */ -#ifndef PULLDOWN_TEST - IP6_EXTHDR_CHECK(m, off, sizeof(*hbh), -1); - hbh = (struct ip6_hbh *)(mtod(m, caddr_t) + off); - hbhlen = (hbh->ip6h_len + 1) << 3; - - IP6_EXTHDR_CHECK(m, off, hbhlen, -1); - hbh = (struct ip6_hbh *)(mtod(m, caddr_t) + off); -#else IP6_EXTHDR_GET(hbh, struct ip6_hbh *, m, sizeof(struct ip6_hdr), sizeof(struct ip6_hbh)); if (hbh == NULL) { @@ -796,18 +886,17 @@ ip6_hopopts_input(plenp, rtalertp, mp, o ip6stat.ip6s_tooshort++; return -1; } -#endif + KASSERT(IP6_HDR_ALIGNED_P(hbh)); off += hbhlen; hbhlen -= sizeof(struct ip6_hbh); - opt = (u_int8_t *)hbh + sizeof(struct ip6_hbh); if (ip6_process_hopopts(m, (u_int8_t *)hbh + sizeof(struct ip6_hbh), hbhlen, rtalertp, plenp) < 0) - return(-1); + return (-1); *offp = off; *mp = m; - return(0); + return (0); } /* @@ -815,6 +904,10 @@ ip6_hopopts_input(plenp, rtalertp, mp, o * This function is separate from ip6_hopopts_input() in order to * handle a case where the sending node itself process its hop-by-hop * options header. In such a case, the function is called from ip6_output(). + * + * The function assumes that hbh header is located right after the IPv6 header + * (RFC2460 p7), opthead is pointer into data content in m, and opthead to + * opthead + hbhlen is located in continuous memory region. */ int ip6_process_hopopts(m, opthead, hbhlen, rtalertp, plenp) @@ -829,6 +922,7 @@ ip6_process_hopopts(m, opthead, hbhlen, u_int8_t *opt = opthead; u_int16_t rtalert_val; u_int32_t jumboplen; + const int erroff = sizeof(struct ip6_hdr) + sizeof(struct ip6_hbh); for (; hbhlen > 0; hbhlen -= optlen, opt += optlen) { switch (*opt) { @@ -849,12 +943,14 @@ ip6_process_hopopts(m, opthead, hbhlen, goto bad; } if (*(opt + 1) != IP6OPT_RTALERT_LEN - 2) { - /* XXX: should we discard the packet? */ - log(LOG_ERR, "length of router alert opt is inconsitent(%d)", - *(opt + 1)); + /* XXX stat */ + icmp6_error(m, ICMP6_PARAM_PROB, + ICMP6_PARAMPROB_HEADER, + erroff + opt + 1 - opthead); + return (-1); } optlen = IP6OPT_RTALERT_LEN; - bcopy((caddr_t)(opt + 2), (caddr_t)&rtalert_val, 2); + bcopy((void *)(opt + 2), (void *)&rtalert_val, 2); *rtalertp = ntohs(rtalert_val); break; case IP6OPT_JUMBO: @@ -864,10 +960,11 @@ ip6_process_hopopts(m, opthead, hbhlen, goto bad; } if (*(opt + 1) != IP6OPT_JUMBO_LEN - 2) { - /* XXX: should we discard the packet? */ - log(LOG_ERR, "length of jumbopayload opt " - "is inconsistent(%d)\n", - *(opt + 1)); + /* XXX stat */ + icmp6_error(m, ICMP6_PARAM_PROB, + ICMP6_PARAMPROB_HEADER, + erroff + opt + 1 - opthead); + return (-1); } optlen = IP6OPT_JUMBO_LEN; @@ -879,11 +976,9 @@ ip6_process_hopopts(m, opthead, hbhlen, if (ip6->ip6_plen) { ip6stat.ip6s_badoptions++; icmp6_error(m, ICMP6_PARAM_PROB, - ICMP6_PARAMPROB_HEADER, - sizeof(struct ip6_hdr) + - sizeof(struct ip6_hbh) + - opt - opthead); - return(-1); + ICMP6_PARAMPROB_HEADER, + erroff + opt - opthead); + return (-1); } /* @@ -905,11 +1000,9 @@ ip6_process_hopopts(m, opthead, hbhlen, if (*plenp != 0) { ip6stat.ip6s_badoptions++; icmp6_error(m, ICMP6_PARAM_PROB, - ICMP6_PARAMPROB_HEADER, - sizeof(struct ip6_hdr) + - sizeof(struct ip6_hbh) + - opt + 2 - opthead); - return(-1); + ICMP6_PARAMPROB_HEADER, + erroff + opt + 2 - opthead); + return (-1); } #endif @@ -919,11 +1012,9 @@ ip6_process_hopopts(m, opthead, hbhlen, if (jumboplen <= IPV6_MAXPACKET) { ip6stat.ip6s_badoptions++; icmp6_error(m, ICMP6_PARAM_PROB, - ICMP6_PARAMPROB_HEADER, - sizeof(struct ip6_hdr) + - sizeof(struct ip6_hbh) + - opt + 2 - opthead); - return(-1); + ICMP6_PARAMPROB_HEADER, + erroff + opt + 2 - opthead); + return (-1); } *plenp = jumboplen; @@ -933,21 +1024,20 @@ ip6_process_hopopts(m, opthead, hbhlen, ip6stat.ip6s_toosmall++; goto bad; } - if ((optlen = ip6_unknown_opt(opt, m, - sizeof(struct ip6_hdr) + - sizeof(struct ip6_hbh) + - opt - opthead)) == -1) - return(-1); + optlen = ip6_unknown_opt(opt, m, + erroff + opt - opthead); + if (optlen == -1) + return (-1); optlen += 2; break; } } - return(0); + return (0); bad: m_freem(m); - return(-1); + return (-1); } /* @@ -966,14 +1056,14 @@ ip6_unknown_opt(optp, m, off) switch (IP6OPT_TYPE(*optp)) { case IP6OPT_TYPE_SKIP: /* ignore the option */ - return((int)*(optp + 1)); + return ((int)*(optp + 1)); case IP6OPT_TYPE_DISCARD: /* silently discard */ m_freem(m); - return(-1); + return (-1); case IP6OPT_TYPE_FORCEICMP: /* send ICMP even if multicasted */ ip6stat.ip6s_badoptions++; icmp6_error(m, ICMP6_PARAM_PROB, ICMP6_PARAMPROB_OPTION, off); - return(-1); + return (-1); case IP6OPT_TYPE_ICMP: /* send ICMP if not multicasted */ ip6stat.ip6s_badoptions++; ip6 = mtod(m, struct ip6_hdr *); @@ -983,11 +1073,11 @@ ip6_unknown_opt(optp, m, off) else icmp6_error(m, ICMP6_PARAM_PROB, ICMP6_PARAMPROB_OPTION, off); - return(-1); + return (-1); } m_freem(m); /* XXX: NOTREACHED */ - return(-1); + return (-1); } /* @@ -1008,72 +1098,75 @@ ip6_savecontrol(in6p, mp, ip6, m) struct ip6_hdr *ip6; struct mbuf *m; { - struct proc *p = curproc; /* XXX */ - int privileged; - - privileged = 0; - if (p && !suser(p->p_ucred, &p->p_acflag)) - privileged++; +#ifdef RFC2292 +#define IS2292(x, y) ((in6p->in6p_flags & IN6P_RFC2292) ? (x) : (y)) +#else +#define IS2292(x, y) (y) +#endif #ifdef SO_TIMESTAMP if (in6p->in6p_socket->so_options & SO_TIMESTAMP) { struct timeval tv; microtime(&tv); - *mp = sbcreatecontrol((caddr_t) &tv, sizeof(tv), - SCM_TIMESTAMP, SOL_SOCKET); + *mp = sbcreatecontrol((void *) &tv, sizeof(tv), + SCM_TIMESTAMP, SOL_SOCKET); if (*mp) mp = &(*mp)->m_next; } #endif - if (in6p->in6p_flags & IN6P_RECVDSTADDR) { - *mp = sbcreatecontrol((caddr_t) &ip6->ip6_dst, - sizeof(struct in6_addr), IPV6_RECVDSTADDR, - IPPROTO_IPV6); - if (*mp) - mp = &(*mp)->m_next; - } -#ifdef noyet - /* options were tossed above */ - if (in6p->in6p_flags & IN6P_RECVOPTS) - /* broken */ - /* ip6_srcroute doesn't do what we want here, need to fix */ - if (in6p->in6p_flags & IPV6P_RECVRETOPTS) - /* broken */ -#endif + /* some OSes call this logic with IPv4 packet, for SO_TIMESTAMP */ + if ((ip6->ip6_vfc & IPV6_VERSION_MASK) != IPV6_VERSION) + return; /* RFC 2292 sec. 5 */ if ((in6p->in6p_flags & IN6P_PKTINFO) != 0) { struct in6_pktinfo pi6; + bcopy(&ip6->ip6_dst, &pi6.ipi6_addr, sizeof(struct in6_addr)); - if (IN6_IS_SCOPE_LINKLOCAL(&pi6.ipi6_addr)) - pi6.ipi6_addr.s6_addr16[1] = 0; - pi6.ipi6_ifindex = (m && m->m_pkthdr.rcvif) - ? m->m_pkthdr.rcvif->if_index - : 0; - *mp = sbcreatecontrol((caddr_t) &pi6, - sizeof(struct in6_pktinfo), IPV6_PKTINFO, - IPPROTO_IPV6); + in6_clearscope(&pi6.ipi6_addr); /* XXX */ + pi6.ipi6_ifindex = m->m_pkthdr.rcvif ? + m->m_pkthdr.rcvif->if_index : 0; + *mp = sbcreatecontrol((void *) &pi6, + sizeof(struct in6_pktinfo), + IS2292(IPV6_2292PKTINFO, IPV6_PKTINFO), IPPROTO_IPV6); if (*mp) mp = &(*mp)->m_next; } + if (in6p->in6p_flags & IN6P_HOPLIMIT) { int hlim = ip6->ip6_hlim & 0xff; - *mp = sbcreatecontrol((caddr_t) &hlim, - sizeof(int), IPV6_HOPLIMIT, IPPROTO_IPV6); + + *mp = sbcreatecontrol((void *) &hlim, sizeof(int), + IS2292(IPV6_2292HOPLIMIT, IPV6_HOPLIMIT), IPPROTO_IPV6); + if (*mp) + mp = &(*mp)->m_next; + } + + if ((in6p->in6p_flags & IN6P_TCLASS) != 0) { + u_int32_t flowinfo; + int tclass; + + flowinfo = (u_int32_t)ntohl(ip6->ip6_flow & IPV6_FLOWINFO_MASK); + flowinfo >>= 20; + + tclass = flowinfo & 0xff; + *mp = sbcreatecontrol((void *)&tclass, sizeof(tclass), + IPV6_TCLASS, IPPROTO_IPV6); + if (*mp) mp = &(*mp)->m_next; } - /* IN6P_NEXTHOP - for outgoing packet only */ /* - * IPV6_HOPOPTS socket option. We require super-user privilege - * for the option, but it might be too strict, since there might - * be some hop-by-hop options which can be returned to normal user. - * See RFC 2292 section 6. + * IPV6_HOPOPTS socket option. Recall that we required super-user + * privilege for the option (see ip6_ctloutput), but it might be too + * strict, since there might be some hop-by-hop options which can be + * returned to normal user. + * See also RFC3542 section 8 (or RFC2292 section 6). */ - if ((in6p->in6p_flags & IN6P_HOPOPTS) != 0 && privileged) { + if ((in6p->in6p_flags & IN6P_HOPOPTS) != 0) { /* * Check if a hop-by-hop options header is contatined in the * received packet, and if so, store the options as ancillary @@ -1081,47 +1174,45 @@ ip6_savecontrol(in6p, mp, ip6, m) * just after the IPv6 header, which fact is assured through * the IPv6 input processing. */ - struct ip6_hdr *ip6 = mtod(m, struct ip6_hdr *); - if (ip6->ip6_nxt == IPPROTO_HOPOPTS) { + struct ip6_hdr *xip6 = mtod(m, struct ip6_hdr *); + if (xip6->ip6_nxt == IPPROTO_HOPOPTS) { struct ip6_hbh *hbh; int hbhlen; + struct mbuf *ext; -#ifndef PULLDOWN_TEST - hbh = (struct ip6_hbh *)(ip6 + 1); - hbhlen = (hbh->ip6h_len + 1) << 3; -#else - IP6_EXTHDR_GET(hbh, struct ip6_hbh *, m, - sizeof(struct ip6_hdr), sizeof(struct ip6_hbh)); - if (hbh == NULL) { + ext = ip6_pullexthdr(m, sizeof(struct ip6_hdr), + xip6->ip6_nxt); + if (ext == NULL) { ip6stat.ip6s_tooshort++; return; } + hbh = mtod(ext, struct ip6_hbh *); hbhlen = (hbh->ip6h_len + 1) << 3; - IP6_EXTHDR_GET(hbh, struct ip6_hbh *, m, - sizeof(struct ip6_hdr), hbhlen); - if (hbh == NULL) { + if (hbhlen != ext->m_len) { + m_freem(ext); ip6stat.ip6s_tooshort++; return; } -#endif /* * XXX: We copy whole the header even if a jumbo * payload option is included, which option is to * be removed before returning in the RFC 2292. - * But it's too painful operation... + * Note: this constraint is removed in RFC3542. */ - *mp = sbcreatecontrol((caddr_t)hbh, hbhlen, - IPV6_HOPOPTS, IPPROTO_IPV6); + *mp = sbcreatecontrol((void *)hbh, hbhlen, + IS2292(IPV6_2292HOPOPTS, IPV6_HOPOPTS), + IPPROTO_IPV6); if (*mp) mp = &(*mp)->m_next; + m_freem(ext); } } /* IPV6_DSTOPTS and IPV6_RTHDR socket options */ if (in6p->in6p_flags & (IN6P_DSTOPTS | IN6P_RTHDR)) { - struct ip6_hdr *ip6 = mtod(m, struct ip6_hdr *); - int nxt = ip6->ip6_nxt, off = sizeof(struct ip6_hdr);; + struct ip6_hdr *xip6 = mtod(m, struct ip6_hdr *); + int nxt = xip6->ip6_nxt, off = sizeof(struct ip6_hdr); /* * Search for destination options headers or routing @@ -1130,50 +1221,50 @@ ip6_savecontrol(in6p, mp, ip6, m) * Note that the order of the headers remains in * the chain of ancillary data. */ - while (1) { /* is explicit loop prevention necessary? */ - struct ip6_ext *ip6e; + for (;;) { /* is explicit loop prevention necessary? */ + struct ip6_ext *ip6e = NULL; int elen; + struct mbuf *ext = NULL; -#ifndef PULLDOWN_TEST - ip6e = (struct ip6_ext *)(mtod(m, caddr_t) + off); - if (nxt == IPPROTO_AH) - elen = (ip6e->ip6e_len + 2) << 2; - else - elen = (ip6e->ip6e_len + 1) << 3; -#else - IP6_EXTHDR_GET(ip6e, struct ip6_ext *, m, off, - sizeof(struct ip6_ext)); - if (ip6e == NULL) { + /* + * if it is not an extension header, don't try to + * pull it from the chain. + */ + switch (nxt) { + case IPPROTO_DSTOPTS: + case IPPROTO_ROUTING: + case IPPROTO_HOPOPTS: + case IPPROTO_AH: /* is it possible? */ + break; + default: + goto loopend; + } + + ext = ip6_pullexthdr(m, off, nxt); + if (ext == NULL) { ip6stat.ip6s_tooshort++; return; } + ip6e = mtod(ext, struct ip6_ext *); if (nxt == IPPROTO_AH) elen = (ip6e->ip6e_len + 2) << 2; else elen = (ip6e->ip6e_len + 1) << 3; - IP6_EXTHDR_GET(ip6e, struct ip6_ext *, m, off, elen); - if (ip6e == NULL) { + if (elen != ext->m_len) { + m_freem(ext); ip6stat.ip6s_tooshort++; return; } -#endif + KASSERT(IP6_HDR_ALIGNED_P(ip6e)); switch (nxt) { - case IPPROTO_DSTOPTS: + case IPPROTO_DSTOPTS: if (!in6p->in6p_flags & IN6P_DSTOPTS) break; - /* - * We also require super-user privilege for - * the option. - * See the comments on IN6_HOPOPTS. - */ - if (!privileged) - break; - - *mp = sbcreatecontrol((caddr_t)ip6e, elen, - IPV6_DSTOPTS, - IPPROTO_IPV6); + *mp = sbcreatecontrol((void *)ip6e, elen, + IS2292(IPV6_2292DSTOPTS, IPV6_DSTOPTS), + IPPROTO_IPV6); if (*mp) mp = &(*mp)->m_next; break; @@ -1182,43 +1273,133 @@ ip6_savecontrol(in6p, mp, ip6, m) if (!in6p->in6p_flags & IN6P_RTHDR) break; - *mp = sbcreatecontrol((caddr_t)ip6e, elen, - IPV6_RTHDR, - IPPROTO_IPV6); + *mp = sbcreatecontrol((void *)ip6e, elen, + IS2292(IPV6_2292RTHDR, IPV6_RTHDR), + IPPROTO_IPV6); if (*mp) mp = &(*mp)->m_next; break; - case IPPROTO_UDP: - case IPPROTO_TCP: - case IPPROTO_ICMPV6: + case IPPROTO_HOPOPTS: + case IPPROTO_AH: /* is it possible? */ + break; + default: /* - * stop search if we encounter an upper - * layer protocol headers. + * other cases have been filtered in the above. + * none will visit this case. here we supply + * the code just in case (nxt overwritten or + * other cases). */ + m_freem(ext); goto loopend; - case IPPROTO_HOPOPTS: - case IPPROTO_AH: /* is it possible? */ - break; } /* proceed with the next header. */ off += elen; nxt = ip6e->ip6e_nxt; + ip6e = NULL; + m_freem(ext); + ext = NULL; } loopend: - ; + ; } - if ((in6p->in6p_flags & IN6P_HOPOPTS) && privileged) { - /* to be done */ +} +#undef IS2292 + + +void +ip6_notify_pmtu(struct in6pcb *in6p, const struct sockaddr_in6 *dst, + uint32_t *mtu) +{ + struct socket *so; + struct mbuf *m_mtu; + struct ip6_mtuinfo mtuctl; + + so = in6p->in6p_socket; + + if (mtu == NULL) + return; + +#ifdef DIAGNOSTIC + if (so == NULL) /* I believe this is impossible */ + panic("ip6_notify_pmtu: socket is NULL"); +#endif + + memset(&mtuctl, 0, sizeof(mtuctl)); /* zero-clear for safety */ + mtuctl.ip6m_mtu = *mtu; + mtuctl.ip6m_addr = *dst; + if (sa6_recoverscope(&mtuctl.ip6m_addr)) + return; + + if ((m_mtu = sbcreatecontrol((void *)&mtuctl, sizeof(mtuctl), + IPV6_PATHMTU, IPPROTO_IPV6)) == NULL) + return; + + if (sbappendaddr(&so->so_rcv, (const struct sockaddr *)dst, NULL, m_mtu) + == 0) { + m_freem(m_mtu); + /* XXX: should count statistics */ + } else + sorwakeup(so); + + return; +} + +/* + * pull single extension header from mbuf chain. returns single mbuf that + * contains the result, or NULL on error. + */ +static struct mbuf * +ip6_pullexthdr(m, off, nxt) + struct mbuf *m; + size_t off; + int nxt; +{ + struct ip6_ext ip6e; + size_t elen; + struct mbuf *n; + +#ifdef DIAGNOSTIC + switch (nxt) { + case IPPROTO_DSTOPTS: + case IPPROTO_ROUTING: + case IPPROTO_HOPOPTS: + case IPPROTO_AH: /* is it possible? */ + break; + default: + printf("ip6_pullexthdr: invalid nxt=%d\n", nxt); } - if ((in6p->in6p_flags & IN6P_DSTOPTS) && privileged) { - /* to be done */ +#endif + + m_copydata(m, off, sizeof(ip6e), (void *)&ip6e); + if (nxt == IPPROTO_AH) + elen = (ip6e.ip6e_len + 2) << 2; + else + elen = (ip6e.ip6e_len + 1) << 3; + + MGET(n, M_DONTWAIT, MT_DATA); + if (n && elen >= MLEN) { + MCLGET(n, M_DONTWAIT); + if ((n->m_flags & M_EXT) == 0) { + m_free(n); + n = NULL; + } + } + if (!n) + return NULL; + + n->m_len = 0; + if (elen >= M_TRAILINGSPACE(n)) { + m_free(n); + return NULL; } - /* IN6P_RTHDR - to be done */ + m_copydata(m, off, elen, mtod(n, void *)); + n->m_len = elen; + return n; } /* @@ -1233,7 +1414,7 @@ ip6_savecontrol(in6p, mp, ip6, m) * carefully. Moreover, it will not be used in the near future when * we develop `neater' mechanism to process extension headers. */ -char * +u_int8_t * ip6_get_prevhdr(m, off) struct mbuf *m; int off; @@ -1241,7 +1422,7 @@ ip6_get_prevhdr(m, off) struct ip6_hdr *ip6 = mtod(m, struct ip6_hdr *); if (off == sizeof(struct ip6_hdr)) - return(&ip6->ip6_nxt); + return (&ip6->ip6_nxt); else { int len, nxt; struct ip6_ext *ip6e = NULL; @@ -1249,7 +1430,7 @@ ip6_get_prevhdr(m, off) nxt = ip6->ip6_nxt; len = sizeof(struct ip6_hdr); while (len < off) { - ip6e = (struct ip6_ext *)(mtod(m, caddr_t) + len); + ip6e = (struct ip6_ext *)(mtod(m, char *) + len); switch (nxt) { case IPPROTO_FRAGMENT: @@ -1265,7 +1446,7 @@ ip6_get_prevhdr(m, off) nxt = ip6e->ip6e_nxt; } if (ip6e) - return(&ip6e->ip6e_nxt); + return (&ip6e->ip6e_nxt); else return NULL; } @@ -1293,9 +1474,12 @@ ip6_nexthdr(m, off, proto, nxtp) switch (proto) { case IPPROTO_IPV6: + /* do not chase beyond intermediate IPv6 headers */ + if (off != 0) + return -1; if (m->m_pkthdr.len < off + sizeof(ip6)) return -1; - m_copydata(m, off, sizeof(ip6), (caddr_t)&ip6); + m_copydata(m, off, sizeof(ip6), (void *)&ip6); if (nxtp) *nxtp = ip6.ip6_nxt; off += sizeof(ip6); @@ -1308,8 +1492,8 @@ ip6_nexthdr(m, off, proto, nxtp) */ if (m->m_pkthdr.len < off + sizeof(fh)) return -1; - m_copydata(m, off, sizeof(fh), (caddr_t)&fh); - if ((ntohs(fh.ip6f_offlg) & IP6F_OFF_MASK) != 0) + m_copydata(m, off, sizeof(fh), (void *)&fh); + if ((fh.ip6f_offlg & IP6F_OFF_MASK) != 0) return -1; if (nxtp) *nxtp = fh.ip6f_nxt; @@ -1319,10 +1503,12 @@ ip6_nexthdr(m, off, proto, nxtp) case IPPROTO_AH: if (m->m_pkthdr.len < off + sizeof(ip6e)) return -1; - m_copydata(m, off, sizeof(ip6e), (caddr_t)&ip6e); + m_copydata(m, off, sizeof(ip6e), (void *)&ip6e); if (nxtp) *nxtp = ip6e.ip6e_nxt; off += (ip6e.ip6e_len + 2) << 2; + if (m->m_pkthdr.len < off) + return -1; return off; case IPPROTO_HOPOPTS: @@ -1330,10 +1516,12 @@ ip6_nexthdr(m, off, proto, nxtp) case IPPROTO_DSTOPTS: if (m->m_pkthdr.len < off + sizeof(ip6e)) return -1; - m_copydata(m, off, sizeof(ip6e), (caddr_t)&ip6e); + m_copydata(m, off, sizeof(ip6e), (void *)&ip6e); if (nxtp) *nxtp = ip6e.ip6e_nxt; off += (ip6e.ip6e_len + 1) << 3; + if (m->m_pkthdr.len < off) + return -1; return off; case IPPROTO_NONE: @@ -1345,8 +1533,6 @@ ip6_nexthdr(m, off, proto, nxtp) default: return -1; } - - return -1; } /* @@ -1366,7 +1552,7 @@ ip6_lasthdr(m, off, proto, nxtp) nxt = -1; nxtp = &nxt; } - while (1) { + for (;;) { newoff = ip6_nexthdr(m, off, proto, nxtp); if (newoff < 0) return off; @@ -1380,6 +1566,96 @@ ip6_lasthdr(m, off, proto, nxtp) } } +struct m_tag * +ip6_addaux(m) + struct mbuf *m; +{ + struct m_tag *mtag; + + mtag = m_tag_find(m, PACKET_TAG_INET6, NULL); + if (!mtag) { + mtag = m_tag_get(PACKET_TAG_INET6, sizeof(struct ip6aux), + M_NOWAIT); + if (mtag) { + m_tag_prepend(m, mtag); + bzero(mtag + 1, sizeof(struct ip6aux)); + } + } + return mtag; +} + +struct m_tag * +ip6_findaux(m) + struct mbuf *m; +{ + struct m_tag *mtag; + + mtag = m_tag_find(m, PACKET_TAG_INET6, NULL); + return mtag; +} + +void +ip6_delaux(m) + struct mbuf *m; +{ + struct m_tag *mtag; + + mtag = m_tag_find(m, PACKET_TAG_INET6, NULL); + if (mtag) + m_tag_delete(m, mtag); +} + +#ifdef GATEWAY +/* + * sysctl helper routine for net.inet.ip6.maxflows. Since + * we could reduce this value, call ip6flow_reap(); + */ +static int +sysctl_net_inet6_ip6_maxflows(SYSCTLFN_ARGS) +{ + int s; + + s = sysctl_lookup(SYSCTLFN_CALL(rnode)); + if (s || newp == NULL) + return (s); + + s = splsoftnet(); + ip6flow_reap(0); + splx(s); + + return (0); +} + +static int +sysctl_net_inet6_ip6_hashsize(SYSCTLFN_ARGS) +{ + int error, tmp; + struct sysctlnode node; + + node = *rnode; + tmp = ip6_hashsize; + node.sysctl_data = &tmp; + error = sysctl_lookup(SYSCTLFN_CALL(&node)); + if (error || newp == NULL) + return (error); + + if ((tmp & (tmp - 1)) == 0 && tmp != 0) { + /* + * Can only fail due to malloc() + */ + if (ip6flow_invalidate_all(tmp)) + return ENOMEM; + } else { + /* + * EINVAL if not a power of 2 + */ + return EINVAL; + } + + return (0); +} +#endif /* GATEWAY */ + /* * System control for IP6 */ @@ -1393,125 +1669,290 @@ u_char inet6ctlerrmap[PRC_NCMDS] = { ENOPROTOOPT }; -#include -#include - -int -ip6_sysctl(name, namelen, oldp, oldlenp, newp, newlen) - int *name; - u_int namelen; - void *oldp; - size_t *oldlenp; - void *newp; - size_t newlen; +SYSCTL_SETUP(sysctl_net_inet6_ip6_setup, "sysctl net.inet6.ip6 subtree setup") { - int old, error; +#ifdef RFC2292 +#define IS2292(x, y) ((in6p->in6p_flags & IN6P_RFC2292) ? (x) : (y)) +#else +#define IS2292(x, y) (y) +#endif - /* All sysctl names at this level are terminal. */ - if (namelen != 1) - return ENOTDIR; - - switch (name[0]) { - - case IPV6CTL_FORWARDING: - return sysctl_int(oldp, oldlenp, newp, newlen, - &ip6_forwarding); - case IPV6CTL_SENDREDIRECTS: - return sysctl_int(oldp, oldlenp, newp, newlen, - &ip6_sendredirects); - case IPV6CTL_DEFHLIM: - return sysctl_int(oldp, oldlenp, newp, newlen, &ip6_defhlim); - case IPV6CTL_MAXFRAGPACKETS: - return sysctl_int(oldp, oldlenp, newp, newlen, - &ip6_maxfragpackets); - case IPV6CTL_ACCEPT_RTADV: - return sysctl_int(oldp, oldlenp, newp, newlen, - &ip6_accept_rtadv); - case IPV6CTL_KEEPFAITH: - return sysctl_int(oldp, oldlenp, newp, newlen, &ip6_keepfaith); - case IPV6CTL_LOG_INTERVAL: - return sysctl_int(oldp, oldlenp, newp, newlen, - &ip6_log_interval); - case IPV6CTL_HDRNESTLIMIT: - return sysctl_int(oldp, oldlenp, newp, newlen, - &ip6_hdrnestlimit); - case IPV6CTL_DAD_COUNT: - return sysctl_int(oldp, oldlenp, newp, newlen, &ip6_dad_count); - case IPV6CTL_AUTO_FLOWLABEL: - return sysctl_int(oldp, oldlenp, newp, newlen, - &ip6_auto_flowlabel); - case IPV6CTL_DEFMCASTHLIM: - return sysctl_int(oldp, oldlenp, newp, newlen, - &ip6_defmcasthlim); - case IPV6CTL_GIF_HLIM: - return sysctl_int(oldp, oldlenp, newp, newlen, - &ip6_gif_hlim); - case IPV6CTL_KAME_VERSION: - return sysctl_rdstring(oldp, oldlenp, newp, __KAME_VERSION); - case IPV6CTL_USE_DEPRECATED: - return sysctl_int(oldp, oldlenp, newp, newlen, - &ip6_use_deprecated); - case IPV6CTL_RR_PRUNE: - return sysctl_int(oldp, oldlenp, newp, newlen, &ip6_rr_prune); + sysctl_createv(clog, 0, NULL, NULL, + CTLFLAG_PERMANENT, + CTLTYPE_NODE, "net", NULL, + NULL, 0, NULL, 0, + CTL_NET, CTL_EOL); + sysctl_createv(clog, 0, NULL, NULL, + CTLFLAG_PERMANENT, + CTLTYPE_NODE, "inet6", + SYSCTL_DESCR("PF_INET6 related settings"), + NULL, 0, NULL, 0, + CTL_NET, PF_INET6, CTL_EOL); + sysctl_createv(clog, 0, NULL, NULL, + CTLFLAG_PERMANENT, + CTLTYPE_NODE, "ip6", + SYSCTL_DESCR("IPv6 related settings"), + NULL, 0, NULL, 0, + CTL_NET, PF_INET6, IPPROTO_IPV6, CTL_EOL); + + sysctl_createv(clog, 0, NULL, NULL, + CTLFLAG_PERMANENT|CTLFLAG_READWRITE, + CTLTYPE_INT, "forwarding", + SYSCTL_DESCR("Enable forwarding of INET6 datagrams"), + NULL, 0, &ip6_forwarding, 0, + CTL_NET, PF_INET6, IPPROTO_IPV6, + IPV6CTL_FORWARDING, CTL_EOL); + sysctl_createv(clog, 0, NULL, NULL, + CTLFLAG_PERMANENT|CTLFLAG_READWRITE, + CTLTYPE_INT, "redirect", + SYSCTL_DESCR("Enable sending of ICMPv6 redirect messages"), + NULL, 0, &ip6_sendredirects, 0, + CTL_NET, PF_INET6, IPPROTO_IPV6, + IPV6CTL_SENDREDIRECTS, CTL_EOL); + sysctl_createv(clog, 0, NULL, NULL, + CTLFLAG_PERMANENT|CTLFLAG_READWRITE, + CTLTYPE_INT, "hlim", + SYSCTL_DESCR("Hop limit for an INET6 datagram"), + NULL, 0, &ip6_defhlim, 0, + CTL_NET, PF_INET6, IPPROTO_IPV6, + IPV6CTL_DEFHLIM, CTL_EOL); +#ifdef notyet + sysctl_createv(clog, 0, NULL, NULL, + CTLFLAG_PERMANENT|CTLFLAG_READWRITE, + CTLTYPE_INT, "mtu", NULL, + NULL, 0, &, 0, + CTL_NET, PF_INET6, IPPROTO_IPV6, + IPV6CTL_DEFMTU, CTL_EOL); +#endif +#ifdef __no_idea__ + sysctl_createv(clog, 0, NULL, NULL, + CTLFLAG_PERMANENT|CTLFLAG_READWRITE, + CTLTYPE_INT, "forwsrcrt", NULL, + NULL, 0, &?, 0, + CTL_NET, PF_INET6, IPPROTO_IPV6, + IPV6CTL_FORWSRCRT, CTL_EOL); + sysctl_createv(clog, 0, NULL, NULL, + CTLFLAG_PERMANENT|CTLFLAG_READWRITE, + CTLTYPE_STRUCT, "mrtstats", NULL, + NULL, 0, &?, sizeof(?), + CTL_NET, PF_INET6, IPPROTO_IPV6, + IPV6CTL_MRTSTATS, CTL_EOL); + sysctl_createv(clog, 0, NULL, NULL, + CTLFLAG_PERMANENT|CTLFLAG_READWRITE, + CTLTYPE_?, "mrtproto", NULL, + NULL, 0, &?, sizeof(?), + CTL_NET, PF_INET6, IPPROTO_IPV6, + IPV6CTL_MRTPROTO, CTL_EOL); +#endif + sysctl_createv(clog, 0, NULL, NULL, + CTLFLAG_PERMANENT|CTLFLAG_READWRITE, + CTLTYPE_INT, "maxfragpackets", + SYSCTL_DESCR("Maximum number of fragments to buffer " + "for reassembly"), + NULL, 0, &ip6_maxfragpackets, 0, + CTL_NET, PF_INET6, IPPROTO_IPV6, + IPV6CTL_MAXFRAGPACKETS, CTL_EOL); +#ifdef __no_idea__ + sysctl_createv(clog, 0, NULL, NULL, + CTLFLAG_PERMANENT|CTLFLAG_READWRITE, + CTLTYPE_INT, "sourcecheck", NULL, + NULL, 0, &?, 0, + CTL_NET, PF_INET6, IPPROTO_IPV6, + IPV6CTL_SOURCECHECK, CTL_EOL); + sysctl_createv(clog, 0, NULL, NULL, + CTLFLAG_PERMANENT|CTLFLAG_READWRITE, + CTLTYPE_INT, "sourcecheck_logint", NULL, + NULL, 0, &?, 0, + CTL_NET, PF_INET6, IPPROTO_IPV6, + IPV6CTL_SOURCECHECK_LOGINT, CTL_EOL); +#endif + sysctl_createv(clog, 0, NULL, NULL, + CTLFLAG_PERMANENT|CTLFLAG_READWRITE, + CTLTYPE_INT, "accept_rtadv", + SYSCTL_DESCR("Accept router advertisements"), + NULL, 0, &ip6_accept_rtadv, 0, + CTL_NET, PF_INET6, IPPROTO_IPV6, + IPV6CTL_ACCEPT_RTADV, CTL_EOL); + sysctl_createv(clog, 0, NULL, NULL, + CTLFLAG_PERMANENT|CTLFLAG_READWRITE, + CTLTYPE_INT, "keepfaith", + SYSCTL_DESCR("Activate faith interface"), + NULL, 0, &ip6_keepfaith, 0, + CTL_NET, PF_INET6, IPPROTO_IPV6, + IPV6CTL_KEEPFAITH, CTL_EOL); + sysctl_createv(clog, 0, NULL, NULL, + CTLFLAG_PERMANENT|CTLFLAG_READWRITE, + CTLTYPE_INT, "log_interval", + SYSCTL_DESCR("Minumum interval between logging " + "unroutable packets"), + NULL, 0, &ip6_log_interval, 0, + CTL_NET, PF_INET6, IPPROTO_IPV6, + IPV6CTL_LOG_INTERVAL, CTL_EOL); + sysctl_createv(clog, 0, NULL, NULL, + CTLFLAG_PERMANENT|CTLFLAG_READWRITE, + CTLTYPE_INT, "hdrnestlimit", + SYSCTL_DESCR("Maximum number of nested IPv6 headers"), + NULL, 0, &ip6_hdrnestlimit, 0, + CTL_NET, PF_INET6, IPPROTO_IPV6, + IPV6CTL_HDRNESTLIMIT, CTL_EOL); + sysctl_createv(clog, 0, NULL, NULL, + CTLFLAG_PERMANENT|CTLFLAG_READWRITE, + CTLTYPE_INT, "dad_count", + SYSCTL_DESCR("Number of Duplicate Address Detection " + "probes to send"), + NULL, 0, &ip6_dad_count, 0, + CTL_NET, PF_INET6, IPPROTO_IPV6, + IPV6CTL_DAD_COUNT, CTL_EOL); + sysctl_createv(clog, 0, NULL, NULL, + CTLFLAG_PERMANENT|CTLFLAG_READWRITE, + CTLTYPE_INT, "auto_flowlabel", + SYSCTL_DESCR("Assign random IPv6 flow labels"), + NULL, 0, &ip6_auto_flowlabel, 0, + CTL_NET, PF_INET6, IPPROTO_IPV6, + IPV6CTL_AUTO_FLOWLABEL, CTL_EOL); + sysctl_createv(clog, 0, NULL, NULL, + CTLFLAG_PERMANENT|CTLFLAG_READWRITE, + CTLTYPE_INT, "defmcasthlim", + SYSCTL_DESCR("Default multicast hop limit"), + NULL, 0, &ip6_defmcasthlim, 0, + CTL_NET, PF_INET6, IPPROTO_IPV6, + IPV6CTL_DEFMCASTHLIM, CTL_EOL); +#if NGIF > 0 + sysctl_createv(clog, 0, NULL, NULL, + CTLFLAG_PERMANENT|CTLFLAG_READWRITE, + CTLTYPE_INT, "gifhlim", + SYSCTL_DESCR("Default hop limit for a gif tunnel datagram"), + NULL, 0, &ip6_gif_hlim, 0, + CTL_NET, PF_INET6, IPPROTO_IPV6, + IPV6CTL_GIF_HLIM, CTL_EOL); +#endif /* NGIF */ + sysctl_createv(clog, 0, NULL, NULL, + CTLFLAG_PERMANENT, + CTLTYPE_STRING, "kame_version", + SYSCTL_DESCR("KAME Version"), + NULL, 0, __UNCONST(__KAME_VERSION), 0, + CTL_NET, PF_INET6, IPPROTO_IPV6, + IPV6CTL_KAME_VERSION, CTL_EOL); + sysctl_createv(clog, 0, NULL, NULL, + CTLFLAG_PERMANENT|CTLFLAG_READWRITE, + CTLTYPE_INT, "use_deprecated", + SYSCTL_DESCR("Allow use of deprecated addresses as " + "source addresses"), + NULL, 0, &ip6_use_deprecated, 0, + CTL_NET, PF_INET6, IPPROTO_IPV6, + IPV6CTL_USE_DEPRECATED, CTL_EOL); + sysctl_createv(clog, 0, NULL, NULL, + CTLFLAG_PERMANENT|CTLFLAG_READWRITE, + CTLTYPE_INT, "rr_prune", NULL, + NULL, 0, &ip6_rr_prune, 0, + CTL_NET, PF_INET6, IPPROTO_IPV6, + IPV6CTL_RR_PRUNE, CTL_EOL); + sysctl_createv(clog, 0, NULL, NULL, + CTLFLAG_PERMANENT #ifndef INET6_BINDV6ONLY - case IPV6CTL_BINDV6ONLY: - return sysctl_int(oldp, oldlenp, newp, newlen, - &ip6_bindv6only); -#endif - case IPV6CTL_ANONPORTMIN: - old = ip6_anonportmin; - error = sysctl_int(oldp, oldlenp, newp, newlen, - &ip6_anonportmin); - if (ip6_anonportmin >= ip6_anonportmax || ip6_anonportmin < 0 || - ip6_anonportmin > 65535 -#ifndef IPNOPRIVPORTS - || ip6_anonportmin < IPV6PORT_RESERVED -#endif - ) { - ip6_anonportmin = old; - return (EINVAL); - } - return (error); - case IPV6CTL_ANONPORTMAX: - old = ip6_anonportmax; - error = sysctl_int(oldp, oldlenp, newp, newlen, - &ip6_anonportmax); - if (ip6_anonportmin >= ip6_anonportmax || ip6_anonportmax < 0 || - ip6_anonportmax > 65535 -#ifndef IPNOPRIVPORTS - || ip6_anonportmax < IPV6PORT_RESERVED + |CTLFLAG_READWRITE, #endif - ) { - ip6_anonportmax = old; - return (EINVAL); - } - return (error); + CTLTYPE_INT, "v6only", + SYSCTL_DESCR("Disallow PF_INET6 sockets from connecting " + "to PF_INET sockets"), + NULL, 0, &ip6_v6only, 0, + CTL_NET, PF_INET6, IPPROTO_IPV6, + IPV6CTL_V6ONLY, CTL_EOL); + sysctl_createv(clog, 0, NULL, NULL, + CTLFLAG_PERMANENT|CTLFLAG_READWRITE, + CTLTYPE_INT, "anonportmin", + SYSCTL_DESCR("Lowest ephemeral port number to assign"), + sysctl_net_inet_ip_ports, 0, &ip6_anonportmin, 0, + CTL_NET, PF_INET6, IPPROTO_IPV6, + IPV6CTL_ANONPORTMIN, CTL_EOL); + sysctl_createv(clog, 0, NULL, NULL, + CTLFLAG_PERMANENT|CTLFLAG_READWRITE, + CTLTYPE_INT, "anonportmax", + SYSCTL_DESCR("Highest ephemeral port number to assign"), + sysctl_net_inet_ip_ports, 0, &ip6_anonportmax, 0, + CTL_NET, PF_INET6, IPPROTO_IPV6, + IPV6CTL_ANONPORTMAX, CTL_EOL); #ifndef IPNOPRIVPORTS - case IPV6CTL_LOWPORTMIN: - old = ip6_lowportmin; - error = sysctl_int(oldp, oldlenp, newp, newlen, - &ip6_lowportmin); - if (ip6_lowportmin >= ip6_lowportmax || - ip6_lowportmin > IPV6PORT_RESERVEDMAX || - ip6_lowportmin < IPV6PORT_RESERVEDMIN) { - ip6_lowportmin = old; - return (EINVAL); - } - return (error); - case IPV6CTL_LOWPORTMAX: - old = ip6_lowportmax; - error = sysctl_int(oldp, oldlenp, newp, newlen, - &ip6_lowportmax); - if (ip6_lowportmin >= ip6_lowportmax || - ip6_lowportmax > IPV6PORT_RESERVEDMAX || - ip6_lowportmax < IPV6PORT_RESERVEDMIN) { - ip6_lowportmax = old; - return (EINVAL); - } - return (error); + sysctl_createv(clog, 0, NULL, NULL, + CTLFLAG_PERMANENT|CTLFLAG_READWRITE, + CTLTYPE_INT, "lowportmin", + SYSCTL_DESCR("Lowest privileged ephemeral port number " + "to assign"), + sysctl_net_inet_ip_ports, 0, &ip6_lowportmin, 0, + CTL_NET, PF_INET6, IPPROTO_IPV6, + IPV6CTL_LOWPORTMIN, CTL_EOL); + sysctl_createv(clog, 0, NULL, NULL, + CTLFLAG_PERMANENT|CTLFLAG_READWRITE, + CTLTYPE_INT, "lowportmax", + SYSCTL_DESCR("Highest privileged ephemeral port number " + "to assign"), + sysctl_net_inet_ip_ports, 0, &ip6_lowportmax, 0, + CTL_NET, PF_INET6, IPPROTO_IPV6, + IPV6CTL_LOWPORTMAX, CTL_EOL); +#endif /* IPNOPRIVPORTS */ + sysctl_createv(clog, 0, NULL, NULL, + CTLFLAG_PERMANENT|CTLFLAG_READWRITE, + CTLTYPE_INT, "use_tempaddr", + SYSCTL_DESCR("Use temporary address"), + NULL, 0, &ip6_use_tempaddr, 0, + CTL_NET, PF_INET6, IPPROTO_IPV6, + CTL_CREATE, CTL_EOL); + sysctl_createv(clog, 0, NULL, NULL, + CTLFLAG_PERMANENT|CTLFLAG_READWRITE, + CTLTYPE_INT, "temppltime", + SYSCTL_DESCR("preferred lifetime of a temporary address"), + NULL, 0, &ip6_temp_preferred_lifetime, 0, + CTL_NET, PF_INET6, IPPROTO_IPV6, + CTL_CREATE, CTL_EOL); + sysctl_createv(clog, 0, NULL, NULL, + CTLFLAG_PERMANENT|CTLFLAG_READWRITE, + CTLTYPE_INT, "tempvltime", + SYSCTL_DESCR("valid lifetime of a temporary address"), + NULL, 0, &ip6_temp_valid_lifetime, 0, + CTL_NET, PF_INET6, IPPROTO_IPV6, + CTL_CREATE, CTL_EOL); + sysctl_createv(clog, 0, NULL, NULL, + CTLFLAG_PERMANENT|CTLFLAG_READWRITE, + CTLTYPE_INT, "maxfrags", + SYSCTL_DESCR("Maximum fragments in reassembly queue"), + NULL, 0, &ip6_maxfrags, 0, + CTL_NET, PF_INET6, IPPROTO_IPV6, + IPV6CTL_MAXFRAGS, CTL_EOL); + sysctl_createv(clog, 0, NULL, NULL, + CTLFLAG_PERMANENT, + CTLTYPE_STRUCT, "stats", + SYSCTL_DESCR("IPv6 statistics"), + NULL, 0, &ip6stat, sizeof(ip6stat), + CTL_NET, PF_INET6, IPPROTO_IPV6, + IPV6CTL_STATS, CTL_EOL); + sysctl_createv(clog, 0, NULL, NULL, + CTLFLAG_PERMANENT|CTLFLAG_READWRITE, + CTLTYPE_INT, "use_defaultzone", + SYSCTL_DESCR("Whether to use the default scope zones"), + NULL, 0, &ip6_use_defzone, 0, + CTL_NET, PF_INET6, IPPROTO_IPV6, + IPV6CTL_USE_DEFAULTZONE, CTL_EOL); + sysctl_createv(clog, 0, NULL, NULL, + CTLFLAG_PERMANENT|CTLFLAG_READWRITE, + CTLTYPE_INT, "mcast_pmtu", + SYSCTL_DESCR("Enable pMTU discovery for multicast packet"), + NULL, 0, &ip6_mcast_pmtu, 0, + CTL_NET, PF_INET6, IPPROTO_IPV6, + CTL_CREATE, CTL_EOL); +#ifdef GATEWAY + sysctl_createv(clog, 0, NULL, NULL, + CTLFLAG_PERMANENT|CTLFLAG_READWRITE, + CTLTYPE_INT, "maxflows", + SYSCTL_DESCR("Number of flows for fast forwarding (IPv6)"), + sysctl_net_inet6_ip6_maxflows, 0, &ip6_maxflows, 0, + CTL_NET, PF_INET6, IPPROTO_IPV6, + CTL_CREATE, CTL_EOL); + sysctl_createv(clog, 0, NULL, NULL, + CTLFLAG_PERMANENT|CTLFLAG_READWRITE, + CTLTYPE_INT, "hashsize", + SYSCTL_DESCR("Size of hash table for fast forwarding (IPv6)"), + sysctl_net_inet6_ip6_hashsize, 0, &ip6_hashsize, 0, + CTL_NET, PF_INET6, IPPROTO_IPV6, + CTL_CREATE, CTL_EOL); #endif - default: - return EOPNOTSUPP; - } - /* NOTREACHED */ }