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 1.9 retrieving revision 1.78 diff -u -p -r1.9 -r1.78 --- src/sys/netinet6/ip6_input.c 1999/12/13 15:17:22 1.9 +++ src/sys/netinet6/ip6_input.c 2005/05/29 21:43:51 1.78 @@ -1,9 +1,10 @@ -/* $NetBSD: ip6_input.c,v 1.9 1999/12/13 15:17:22 itojun Exp $ */ +/* $NetBSD: ip6_input.c,v 1.78 2005/05/29 21:43:51 christos Exp $ */ +/* $KAME: ip6_input.c,v 1.188 2001/03/29 05:34:31 itojun Exp $ */ /* * Copyright (C) 1995, 1996, 1997, and 1998 WIDE Project. * All rights reserved. - * + * * Redistribution and use in source and binary forms, with or without * modification, are permitted provided that the following conditions * are met: @@ -15,7 +16,7 @@ * 3. Neither the name of the project nor the names of its contributors * may be used to endorse or promote products derived from this software * without specific prior written permission. - * + * * THIS SOFTWARE IS PROVIDED BY THE PROJECT AND CONTRIBUTORS ``AS IS'' AND * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE @@ -41,11 +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. * @@ -64,15 +61,12 @@ * @(#)ip_input.c 8.2 (Berkeley) 1/4/94 */ -#ifdef __FreeBSD__ -#include "opt_ip6fw.h" -#endif -#if (defined(__FreeBSD__) && __FreeBSD__ >= 3) || defined(__NetBSD__) +#include +__KERNEL_RCSID(0, "$NetBSD: ip6_input.c,v 1.78 2005/05/29 21:43:51 christos Exp $"); + #include "opt_inet.h" -#ifdef __NetBSD__ /*XXX*/ #include "opt_ipsec.h" -#endif -#endif +#include "opt_pfil_hooks.h" #include #include @@ -86,83 +80,60 @@ #include #include #include -#if !defined(__bsdi__) && !(defined(__FreeBSD__) && __FreeBSD__ < 3) #include -#endif +#include #include #include #include #include #include +#ifdef PFIL_HOOKS +#include +#endif #include #include #ifdef INET #include #include -#endif /*INET*/ -#if (defined(__FreeBSD__) && __FreeBSD__ >= 3) || defined(__OpenBSD__) || (defined(__bsdi__) && _BSDI_VERSION >= 199802) -#include -#endif +#endif /* INET */ +#include #include -#include -#if !((defined(__FreeBSD__) && __FreeBSD__ >= 3) || defined(__OpenBSD__) || (defined(__bsdi__) && _BSDI_VERSION >= 199802)) -#include -#endif #include -#include +#include +#include #include #include -#include -#ifdef IPV6FIREWALL -#include +#ifdef IPSEC +#include #endif #include -/* we need it for NLOOP. */ -#ifndef __bsdi__ -#include "loop.h" -#endif #include "faith.h" - #include "gif.h" -#include "bpfilter.h" - -#include -#ifdef __OpenBSD__ /*KAME IPSEC*/ -#undef IPSEC +#if NGIF > 0 +#include #endif +#include + extern struct domain inet6domain; -extern struct ip6protosw inet6sw[]; -#ifdef __bsdi__ -#if _BSDI_VERSION < 199802 -extern struct ifnet loif; -#else -extern struct ifnet *loifp; -#endif -#endif u_char ip6_protox[IPPROTO_MAX]; static int ip6qmaxlen = IFQ_MAXLEN; struct in6_ifaddr *in6_ifaddr; struct ifqueue ip6intrq; -#if defined(__NetBSD__) || defined(__OpenBSD__) -extern struct ifnet loif[NLOOP]; -#endif int ip6_forward_srcrt; /* XXX */ int ip6_sourcecheck; /* XXX */ int ip6_sourcecheck_interval; /* XXX */ -#ifdef IPV6FIREWALL -/* firewall hooks */ -ip6_fw_chk_t *ip6_fw_chk_ptr; -ip6_fw_ctl_t *ip6_fw_ctl_ptr; +#ifdef PFIL_HOOKS +struct pfil_head inet6_pfil_hook; #endif struct ip6stat ip6stat; @@ -170,13 +141,7 @@ struct ip6stat ip6stat; static void ip6_init2 __P((void *)); static int ip6_hopopts_input __P((u_int32_t *, u_int32_t *, struct mbuf **, int *)); - -#if defined(PTR) -extern int ip6_protocol_tr; - -int ptr_in6 __P((struct mbuf *, struct mbuf **)); -extern void ip_forward __P((struct mbuf *, int)); -#endif +static struct mbuf *ip6_pullexthdr __P((struct mbuf *, size_t, int)); /* * IP6 initialization: fill in IP6 protocol switch table. @@ -185,71 +150,46 @@ extern void ip_forward __P((struct mbuf void ip6_init() { - register struct ip6protosw *pr; - register int i; - struct timeval tv; + const struct ip6protosw *pr; + int i; - 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; nd6_init(); frag6_init(); -#ifdef IPV6FIREWALL - ip6_fw_init(); -#endif - /* - * 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; -#ifndef __FreeBSD__ ip6_init2((void *)0); -#endif + +#ifdef PFIL_HOOKS + /* Register our Packet Filter hook. */ + inet6_pfil_hook.ph_type = PFIL_TYPE_AF; + inet6_pfil_hook.ph_af = AF_INET6; + i = pfil_head_register(&inet6_pfil_hook); + if (i != 0) + printf("ip6_init: WARNING: unable to register pfil hook, " + "error %d\n", i); +#endif /* PFIL_HOOKS */ } static void ip6_init2(dummy) void *dummy; { - int ret; -#if defined(__bsdi__) && _BSDI_VERSION < 199802 - struct ifnet *loifp = &loif; -#endif - - /* get EUI64 from somewhere */ - ret = in6_ifattach_getifid(NULL); - - /* - * to route local address of p2p link to loopback, - * assign loopback address first. - */ -#ifdef __bsdi__ - in6_ifattach(loifp, IN6_IFT_LOOP, NULL, 0); -#else - in6_ifattach(&loif[0], IN6_IFT_LOOP, NULL, 0); -#endif /* nd6_timer_init */ - timeout(nd6_timer, (caddr_t)0, hz); - /* router renumbering prefix list maintenance */ - timeout(in6_rr_timer, (caddr_t)0, hz); + callout_init(&nd6_timer_ch); + callout_reset(&nd6_timer_ch, hz, nd6_timer, NULL); } -#ifdef __FreeBSD__ -/* cheat */ -SYSINIT(netinet6init2, SI_SUB_PROTO_DOMAIN, SI_ORDER_THIRD, ip6_init2, NULL); -#endif - /* * IP6 input interrupt handling. Just pass the packet to ip6_input. */ @@ -260,7 +200,7 @@ ip6intr() struct mbuf *m; for (;;) { - s = splimp(); + s = splnet(); IF_DEQUEUE(&ip6intrq, m); splx(s); if (m == 0) @@ -269,10 +209,6 @@ ip6intr() } } -#ifdef __FreeBSD__ -NETISR_SET(NETISR_IPV6, ip6intr); -#endif - extern struct route_in6 ip6_forward_rt; void @@ -285,23 +221,19 @@ ip6_input(m) u_int32_t rtalert = ~0; int nxt, ours = 0; struct ifnet *deliverifp = NULL; -#if defined(__bsdi__) && _BSDI_VERSION < 199802 - struct ifnet *loifp = &loif; -#endif + int srcrt = 0; #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 + * mbuf statistics */ if (m->m_flags & M_EXT) { if (m->m_next) @@ -309,34 +241,40 @@ ip6_input(m) else ip6stat.ip6s_mext1++; } else { +#define M2MMAX (sizeof(ip6stat.ip6s_m2m)/sizeof(ip6stat.ip6s_m2m[0])) if (m->m_next) { if (m->m_flags & M_LOOP) { -#ifdef __bsdi__ - ip6stat.ip6s_m2m[loifp->if_index]++; /*XXX*/ -#else - ip6stat.ip6s_m2m[loif[0].if_index]++; /*XXX*/ -#endif - } - else if (m->m_pkthdr.rcvif->if_index <= 31) + 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 ip6stat.ip6s_m2m[0]++; } else ip6stat.ip6s_m1++; +#undef M2MMAX } 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, caddr_t)) == 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; @@ -351,34 +289,92 @@ ip6_input(m) goto bad; } - ip6stat.ip6s_nxthist[ip6->ip6_nxt]++; - -#ifdef IPV6FIREWALL +#ifdef PFIL_HOOKS /* - * Check with the firewall... + * Run through list of hooks for input packets. If there are any + * filters which require that additional packets in the flow are + * not fast-forwarded, they must clear the M_CANFASTFWD flag. + * Note that filters must _never_ set this flag, as another filter + * in the list may have previously cleared it. */ - if (ip6_fw_chk_ptr) { - u_short port = 0; - /* If ipfw says divert, we have to just drop packet */ - /* use port as a dummy argument */ - if ((*ip6_fw_chk_ptr)(&ip6, NULL, &port, &m)) { - m_freem(m); - m = NULL; - } - if (!m) + /* + * let ipfilter look at packet on the wire, + * not the decapsulated packet. + */ +#ifdef IPSEC + if (!ipsec_getnhist(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 + if (altq_input != NULL && (*altq_input)(m, AF_INET6) == 0) { + /* packet is dropped by traffic conditioner */ + return; } #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 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 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)) { ip6stat.ip6s_badscope++; in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_addrerr); goto bad; } +#if 0 + /* + * Reject packets with IPv4 compatible addresses (auto tunnel). + * + * The code forbids auto tunnel relay case in RFC1933 (the check is + * stronger than RFC1933). We may want to re-enable it if mech-xx + * is revised to forbid relaying case. + */ + if (IN6_IS_ADDR_V4COMPAT(&ip6->ip6_src) || + IN6_IS_ADDR_V4COMPAT(&ip6->ip6_dst)) { + ip6stat.ip6s_badscope++; + in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_addrerr); + goto bad; + } +#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) { @@ -392,47 +388,50 @@ ip6_input(m) } } - if (m->m_pkthdr.rcvif->if_flags & IFF_LOOPBACK) { - if (IN6_IS_ADDR_LINKLOCAL(&ip6->ip6_dst)) { - ours = 1; - deliverifp = m->m_pkthdr.rcvif; - goto hbhcheck; + /* 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; } - } else { - 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); } -#if defined(PTR) + 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 (ip6_protocol_tr) - { - struct mbuf *m1 = NULL; - - switch (ptr_in6(m, &m1)) - { - case IPPROTO_IP: goto mcastcheck; - case IPPROTO_IPV4: ip_forward(m1, 0); break; - case IPPROTO_IPV6: ip6_forward(m1, 0); break; - case IPPROTO_MAX: /* discard this packet */ - default: - } - - if (m != m1) - m_freem(m); + 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; + } - return; + ours = 1; + deliverifp = m->m_pkthdr.rcvif; + goto hbhcheck; } - mcastcheck: -#endif - /* * Multicast check */ @@ -460,23 +459,28 @@ ip6_input(m) /* * Unicast check */ - if (ip6_forward_rt.ro_rt == 0 || - !IN6_ARE_ADDR_EQUAL(&ip6->ip6_dst, - &ip6_forward_rt.ro_dst.sin6_addr)) { + 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)) + ip6stat.ip6s_forward_cachehit++; + 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; } + bzero(&ip6_forward_rt.ro_dst, sizeof(struct sockaddr_in6)); - ip6_forward_rt.ro_dst.sin6_len = sizeof(struct sockaddr_in6); - ip6_forward_rt.ro_dst.sin6_family = AF_INET6; - ip6_forward_rt.ro_dst.sin6_addr = ip6->ip6_dst; + dst6 = (struct sockaddr_in6 *)&ip6_forward_rt.ro_dst; + dst6->sin6_len = sizeof(struct sockaddr_in6); + dst6->sin6_family = AF_INET6; + dst6->sin6_addr = ip6->ip6_dst; -#ifdef __FreeBSD__ - rtalloc_ign((struct route *)&ip6_forward_rt, RTF_PRCLONING); -#else rtalloc((struct route *)&ip6_forward_rt); -#endif } #define rt6_key(r) ((struct sockaddr_in6 *)((r)->rt_nodes->rn_key)) @@ -493,6 +497,7 @@ ip6_input(m) if (ip6_forward_rt.ro_rt && (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 @@ -500,34 +505,43 @@ 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 = (struct in6_ifaddr *)ip6_forward_rt.ro_rt->rt_ifa; - /* packet to tentative address must not be received */ if (ia6->ia6_flags & IN6_IFF_ANYCAST) m->m_flags |= M_ANYCAST6; + /* + * packets to a tentative, duplicated, or somehow invalid + * address must not be accepted. + */ if (!(ia6->ia6_flags & IN6_IFF_NOTREADY)) { - /* this interface is ready */ + /* this address is ready */ ours = 1; deliverifp = ia6->ia_ifp; /* correct? */ goto hbhcheck; } else { - /* this interface is not ready, fall through */ + /* address is not ready, so discard the packet. */ + nd6log((LOG_INFO, + "ip6_input: packet to an unready address %s->%s\n", + ip6_sprintf(&ip6->ip6_src), + ip6_sprintf(&ip6->ip6_dst))); + + goto bad; } } /* - * 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 && ip6_forward_rt.ro_rt->rt_ifp && + 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; } } @@ -583,19 +597,36 @@ ip6_input(m) #endif return; /* m have already been freed */ } + /* adjust pointer */ ip6 = mtod(m, struct ip6_hdr *); -#ifndef PULLDOWN_TEST - /* ip6_hopopts_input() ensures that mbuf is contiguous */ - hbh = (struct ip6_hbh *)(ip6 + 1); -#else + + /* + * 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. + */ + 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); + return; + } 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; /* @@ -647,19 +678,35 @@ ip6_input(m) m_freem(m); return; } - } - else if (!ours) { - ip6_forward(m, 0); + } else if (!ours) { + ip6_forward(m, srcrt); return; - } + } + + ip6 = mtod(m, struct ip6_hdr *); + + /* + * 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. + * + * For SIIT end node behavior, you may want to disable the check. + * However, you will become vulnerable to attacks using IPv4 mapped + * source. + */ + if (IN6_IS_ADDR_V4MAPPED(&ip6->ip6_src) || + IN6_IS_ADDR_V4MAPPED(&ip6->ip6_dst)) { + ip6stat.ip6s_badscope++; + in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_addrerr); + goto bad; + } /* * Tell launch routine the next header */ -#if defined(__NetBSD__) && defined(IFA_STATS) - if (IFA_STATS && deliverifp != NULL) { +#ifdef IFA_STATS + if (deliverifp != NULL) { struct in6_ifaddr *ia6; - ip6 = mtod(m, struct ip6_hdr *); ia6 = in6_ifawithifp(deliverifp, &ip6->ip6_dst); if (ia6) ia6->ia_ifa.ifa_data.ifad_inbytes += m->m_pkthdr.len; @@ -668,6 +715,7 @@ ip6_input(m) ip6stat.ip6s_delivered++; in6_ifstat_inc(deliverifp, ifs6_in_deliver); nest = 0; + while (nxt != IPPROTO_DONE) { if (ip6_hdrnestlimit && (++nest > ip6_hdrnestlimit)) { ip6stat.ip6s_toomanyhdr++; @@ -684,6 +732,19 @@ ip6_input(m) goto bad; } +#ifdef 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[ip6_protox[nxt]].pr_flags & PR_LASTHDR) != 0 && + ipsec6_in_reject(m, NULL)) { + ipsec6stat.in_polvio++; + goto bad; + } +#endif + nxt = (*inet6sw[ip6_protox[nxt]].pr_input)(&m, &off, nxt); } return; @@ -702,20 +763,11 @@ ip6_hopopts_input(plenp, rtalertp, mp, o struct mbuf **mp; int *offp; { - register struct mbuf *m = *mp; + 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) { @@ -729,18 +781,17 @@ 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); } /* @@ -748,6 +799,10 @@ 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) @@ -761,105 +816,123 @@ ip6_process_hopopts(m, opthead, hbhlen, int optlen = 0; 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) { - case IP6OPT_PAD1: - optlen = 1; - break; - case IP6OPT_PADN: - if (hbhlen < IP6OPT_MINLEN) { - ip6stat.ip6s_toosmall++; - goto bad; - } - optlen = *(opt + 1) + 2; - break; - case IP6OPT_RTALERT: - /* XXX may need check for alignment */ - if (hbhlen < IP6OPT_RTALERT_LEN) { - ip6stat.ip6s_toosmall++; - 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)); - optlen = IP6OPT_RTALERT_LEN; - bcopy((caddr_t)(opt + 2), (caddr_t)&rtalert_val, 2); - *rtalertp = ntohs(rtalert_val); - break; - case IP6OPT_JUMBO: - /* XXX may need check for alignment */ - if (hbhlen < IP6OPT_JUMBO_LEN) { - ip6stat.ip6s_toosmall++; - 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)", - *(opt + 1)); - optlen = IP6OPT_JUMBO_LEN; - - /* - * We can simply cast because of the alignment - * requirement of the jumbo payload option. - */ -#if 0 - *plenp = ntohl(*(u_int32_t *)(opt + 2)); -#else - bcopy(opt + 2, plenp, sizeof(*plenp)); - *plenp = htonl(*plenp); + switch (*opt) { + case IP6OPT_PAD1: + optlen = 1; + break; + case IP6OPT_PADN: + if (hbhlen < IP6OPT_MINLEN) { + ip6stat.ip6s_toosmall++; + goto bad; + } + optlen = *(opt + 1) + 2; + break; + case IP6OPT_RTALERT: + /* XXX may need check for alignment */ + if (hbhlen < IP6OPT_RTALERT_LEN) { + ip6stat.ip6s_toosmall++; + goto bad; + } + if (*(opt + 1) != IP6OPT_RTALERT_LEN - 2) { + /* 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); + *rtalertp = ntohs(rtalert_val); + break; + case IP6OPT_JUMBO: + /* XXX may need check for alignment */ + if (hbhlen < IP6OPT_JUMBO_LEN) { + ip6stat.ip6s_toosmall++; + goto bad; + } + if (*(opt + 1) != IP6OPT_JUMBO_LEN - 2) { + /* XXX stat */ + icmp6_error(m, ICMP6_PARAM_PROB, + ICMP6_PARAMPROB_HEADER, + erroff + opt + 1 - opthead); + return (-1); + } + optlen = IP6OPT_JUMBO_LEN; + + /* + * IPv6 packets that have non 0 payload length + * must not contain a jumbo payload option. + */ + ip6 = mtod(m, struct ip6_hdr *); + if (ip6->ip6_plen) { + ip6stat.ip6s_badoptions++; + icmp6_error(m, ICMP6_PARAM_PROB, + ICMP6_PARAMPROB_HEADER, + erroff + opt - opthead); + return (-1); + } + + /* + * We may see jumbolen in unaligned location, so + * we'd need to perform bcopy(). + */ + bcopy(opt + 2, &jumboplen, sizeof(jumboplen)); + jumboplen = (u_int32_t)htonl(jumboplen); + +#if 1 + /* + * if there are multiple jumbo payload options, + * *plenp will be non-zero and the packet will be + * rejected. + * the behavior may need some debate in ipngwg - + * multiple options does not make sense, however, + * there's no explicit mention in specification. + */ + if (*plenp != 0) { + ip6stat.ip6s_badoptions++; + icmp6_error(m, ICMP6_PARAM_PROB, + ICMP6_PARAMPROB_HEADER, + erroff + opt + 2 - opthead); + return (-1); + } #endif - if (*plenp <= IPV6_MAXPACKET) { - /* - * jumbo payload length must be larger - * than 65535 - */ - 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); - } - - ip6 = mtod(m, struct ip6_hdr *); - if (ip6->ip6_plen) { - /* - * IPv6 packets that have non 0 payload length - * must not contain a jumbo paylod option. - */ - ip6stat.ip6s_badoptions++; - icmp6_error(m, ICMP6_PARAM_PROB, - ICMP6_PARAMPROB_HEADER, - sizeof(struct ip6_hdr) + - sizeof(struct ip6_hbh) + - opt - opthead); - return(-1); - } - break; - default: /* unknown option */ - if (hbhlen < IP6OPT_MINLEN) { - 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 += 2; - break; + + /* + * jumbo payload length must be larger than 65535. + */ + if (jumboplen <= IPV6_MAXPACKET) { + ip6stat.ip6s_badoptions++; + icmp6_error(m, ICMP6_PARAM_PROB, + ICMP6_PARAMPROB_HEADER, + erroff + opt + 2 - opthead); + return (-1); + } + *plenp = jumboplen; + + break; + default: /* unknown option */ + if (hbhlen < IP6OPT_MINLEN) { + ip6stat.ip6s_toosmall++; + goto bad; + } + 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); } /* @@ -876,30 +949,30 @@ ip6_unknown_opt(optp, m, off) { struct ip6_hdr *ip6; - switch(IP6OPT_TYPE(*optp)) { - case IP6OPT_TYPE_SKIP: /* ignore the option */ - return((int)*(optp + 1)); - case IP6OPT_TYPE_DISCARD: /* silently discard */ - m_freem(m); - 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); - case IP6OPT_TYPE_ICMP: /* send ICMP if not multicasted */ - ip6stat.ip6s_badoptions++; - ip6 = mtod(m, struct ip6_hdr *); - if (IN6_IS_ADDR_MULTICAST(&ip6->ip6_dst) || - (m->m_flags & (M_BCAST|M_MCAST))) - m_freem(m); - else - icmp6_error(m, ICMP6_PARAM_PROB, - ICMP6_PARAMPROB_OPTION, off); - return(-1); + switch (IP6OPT_TYPE(*optp)) { + case IP6OPT_TYPE_SKIP: /* ignore the option */ + return ((int)*(optp + 1)); + case IP6OPT_TYPE_DISCARD: /* silently discard */ + m_freem(m); + 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); + case IP6OPT_TYPE_ICMP: /* send ICMP if not multicasted */ + ip6stat.ip6s_badoptions++; + ip6 = mtod(m, struct ip6_hdr *); + if (IN6_IS_ADDR_MULTICAST(&ip6->ip6_dst) || + (m->m_flags & (M_BCAST|M_MCAST))) + m_freem(m); + else + icmp6_error(m, ICMP6_PARAM_PROB, + ICMP6_PARAMPROB_OPTION, off); + return (-1); } m_freem(m); /* XXX: NOTREACHED */ - return(-1); + return (-1); } /* @@ -915,39 +988,11 @@ ip6_unknown_opt(optp, m, off) */ void ip6_savecontrol(in6p, mp, ip6, m) -#if (defined(__FreeBSD__) && __FreeBSD__ >= 3) || defined(HAVE_NRL_INPCB) - register struct inpcb *in6p; -#else - register struct in6pcb *in6p; -#endif - register struct mbuf **mp; - register struct ip6_hdr *ip6; - register struct mbuf *m; + struct in6pcb *in6p; + struct mbuf **mp; + struct ip6_hdr *ip6; + struct mbuf *m; { -#ifdef HAVE_NRL_INPCB -# define in6p_flags inp_flags -#endif -#if defined(__NetBSD__) || (defined(__FreeBSD__) && __FreeBSD__ >= 3) - struct proc *p = curproc; /* XXX */ -#endif -#ifdef __bsdi__ -# define sbcreatecontrol so_cmsg -#endif - int privileged; - - privileged = 0; -#if defined(__NetBSD__) || (defined(__FreeBSD__) && __FreeBSD__ >= 3) - if (p && !suser(p->p_ucred, &p->p_acflag)) - privileged++; -#else -#ifdef HAVE_NRL_INPCB - if ((in6p->inp_socket->so_state & SS_PRIV) != 0) - privileged++; -#else - if ((in6p->in6p_socket->so_state & SS_PRIV) != 0) - privileged++; -#endif -#endif #ifdef SO_TIMESTAMP if (in6p->in6p_socket->so_options & SO_TIMESTAMP) { @@ -955,15 +1000,14 @@ ip6_savecontrol(in6p, mp, ip6, m) microtime(&tv); *mp = sbcreatecontrol((caddr_t) &tv, sizeof(tv), - SCM_TIMESTAMP, SOL_SOCKET); + 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); + sizeof(struct in6_addr), IPV6_RECVDSTADDR, IPPROTO_IPV6); if (*mp) mp = &(*mp)->m_next; } @@ -978,7 +1022,7 @@ ip6_savecontrol(in6p, mp, ip6, m) #endif /* RFC 2292 sec. 5 */ - if (in6p->in6p_flags & IN6P_PKTINFO) { + 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)) @@ -987,27 +1031,27 @@ ip6_savecontrol(in6p, mp, ip6, m) ? m->m_pkthdr.rcvif->if_index : 0; *mp = sbcreatecontrol((caddr_t) &pi6, - sizeof(struct in6_pktinfo), IPV6_PKTINFO, - IPPROTO_IPV6); + sizeof(struct in6_pktinfo), 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((caddr_t) &hlim, sizeof(int), + IPV6_HOPLIMIT, 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 RFC 2292 section 6. */ - if ((in6p->in6p_flags & IN6P_HOPOPTS) && 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 @@ -1015,29 +1059,25 @@ 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 @@ -1046,16 +1086,17 @@ ip6_savecontrol(in6p, mp, ip6, m) * But it's too painful operation... */ *mp = sbcreatecontrol((caddr_t)hbh, hbhlen, - IPV6_HOPOPTS, IPPROTO_IPV6); + 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 @@ -1064,100 +1105,143 @@ 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; + while (1) { /* 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: + if (!in6p->in6p_flags & IN6P_DSTOPTS) + break; + + *mp = sbcreatecontrol((caddr_t)ip6e, elen, + IPV6_DSTOPTS, IPPROTO_IPV6); + if (*mp) + mp = &(*mp)->m_next; + break; + + case IPPROTO_ROUTING: + if (!in6p->in6p_flags & IN6P_RTHDR) + break; + + *mp = sbcreatecontrol((caddr_t)ip6e, elen, + IPV6_RTHDR, IPPROTO_IPV6); + if (*mp) + mp = &(*mp)->m_next; + break; + + case IPPROTO_HOPOPTS: + case IPPROTO_AH: /* is it possible? */ + break; + + default: + /* + * 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; - switch(nxt) { - 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); - if (*mp) - mp = &(*mp)->m_next; - break; - - case IPPROTO_ROUTING: - if (!in6p->in6p_flags & IN6P_RTHDR) - break; - - *mp = sbcreatecontrol((caddr_t)ip6e, elen, - IPV6_RTHDR, - IPPROTO_IPV6); - if (*mp) - mp = &(*mp)->m_next; - break; - - case IPPROTO_UDP: - case IPPROTO_TCP: - case IPPROTO_ICMPV6: - default: - /* - * stop search if we encounter an upper - * layer protocol headers. - */ - 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 */ +} + +/* + * 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), (caddr_t)&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; + } } - /* IN6P_RTHDR - to be done */ + if (!n) + return NULL; -#ifdef __bsdi__ -# undef sbcreatecontrol -#endif -#ifdef __OpenBSD__ -# undef in6p_flags -#endif + n->m_len = 0; + if (elen >= M_TRAILINGSPACE(n)) { + m_free(n); + return NULL; + } + + m_copydata(m, off, elen, mtod(n, caddr_t)); + n->m_len = elen; + return n; } /* @@ -1172,7 +1256,7 @@ 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; @@ -1180,7 +1264,7 @@ 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; @@ -1190,7 +1274,7 @@ ip6_get_prevhdr(m, off) while (len < off) { ip6e = (struct ip6_ext *)(mtod(m, caddr_t) + len); - switch(nxt) { + switch (nxt) { case IPPROTO_FRAGMENT: len += sizeof(struct ip6_frag); break; @@ -1204,13 +1288,124 @@ ip6_get_prevhdr(m, off) nxt = ip6e->ip6e_nxt; } if (ip6e) - return(&ip6e->ip6e_nxt); + return (&ip6e->ip6e_nxt); else return NULL; } } /* + * get next header offset. m will be retained. + */ +int +ip6_nexthdr(m, off, proto, nxtp) + struct mbuf *m; + int off; + int proto; + int *nxtp; +{ + struct ip6_hdr ip6; + struct ip6_ext ip6e; + struct ip6_frag fh; + + /* just in case */ + if (m == NULL) + panic("ip6_nexthdr: m == NULL"); + if ((m->m_flags & M_PKTHDR) == 0 || m->m_pkthdr.len < off) + return -1; + + switch (proto) { + case IPPROTO_IPV6: + if (m->m_pkthdr.len < off + sizeof(ip6)) + return -1; + m_copydata(m, off, sizeof(ip6), (caddr_t)&ip6); + if (nxtp) + *nxtp = ip6.ip6_nxt; + off += sizeof(ip6); + return off; + + case IPPROTO_FRAGMENT: + /* + * terminate parsing if it is not the first fragment, + * it does not make sense to parse through it. + */ + if (m->m_pkthdr.len < off + sizeof(fh)) + return -1; + m_copydata(m, off, sizeof(fh), (caddr_t)&fh); + if ((fh.ip6f_offlg & IP6F_OFF_MASK) != 0) + return -1; + if (nxtp) + *nxtp = fh.ip6f_nxt; + off += sizeof(struct ip6_frag); + return off; + + case IPPROTO_AH: + if (m->m_pkthdr.len < off + sizeof(ip6e)) + return -1; + m_copydata(m, off, sizeof(ip6e), (caddr_t)&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: + case IPPROTO_ROUTING: + case IPPROTO_DSTOPTS: + if (m->m_pkthdr.len < off + sizeof(ip6e)) + return -1; + m_copydata(m, off, sizeof(ip6e), (caddr_t)&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: + case IPPROTO_ESP: + case IPPROTO_IPCOMP: + /* give up */ + return -1; + + default: + return -1; + } +} + +/* + * get offset for the last header in the chain. m will be kept untainted. + */ +int +ip6_lasthdr(m, off, proto, nxtp) + struct mbuf *m; + int off; + int proto; + int *nxtp; +{ + int newoff; + int nxt; + + if (!nxtp) { + nxt = -1; + nxtp = &nxt; + } + while (1) { + newoff = ip6_nexthdr(m, off, proto, nxtp); + if (newoff < 0) + return off; + else if (newoff < off) + return -1; /* invalid */ + else if (newoff == off) + return newoff; + + off = newoff; + proto = *nxtp; + } +} + +/* * System control for IP6 */ @@ -1223,94 +1418,233 @@ u_char inet6ctlerrmap[PRC_NCMDS] = { ENOPROTOOPT }; -#if defined(__NetBSD__) || defined(__OpenBSD__) -#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") { - /* 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); - default: - return EOPNOTSUPP; - } - /* NOTREACHED */ -} -#endif /* __NetBSD__ || __OpenBSD__ */ - -#ifdef __bsdi__ -int *ip6_sysvars[] = IPV6CTL_VARS; - -int -ip6_sysctl(name, namelen, oldp, oldlenp, newp, newlen) - int *name; - u_int namelen; - void *oldp; - size_t *oldlenp; - void *newp; - size_t newlen; -{ - if (name[0] >= IPV6CTL_MAXID) - return (EOPNOTSUPP); - switch (name[0]) { - case IPV6CTL_STATS: - return sysctl_rdtrunc(oldp, oldlenp, newp, &ip6stat, - sizeof(ip6stat)); - case IPV6CTL_KAME_VERSION: - return sysctl_rdstring(oldp, oldlenp, newp, __KAME_VERSION); - default: - return (sysctl_int_arr(ip6_sysvars, name, namelen, - oldp, oldlenp, newp, newlen)); - } + 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, "stats", NULL, + NULL, 0, &?, sizeof(?), + CTL_NET, PF_INET6, IPPROTO_IPV6, + IPV6CTL_STATS, 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 + |CTLFLAG_READWRITE, +#endif + 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 + 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, "maxfrags", + SYSCTL_DESCR("Maximum fragments in reassembly queue"), + NULL, 0, &ip6_maxfrags, 0, + CTL_NET, PF_INET6, IPPROTO_IPV6, + IPV6CTL_MAXFRAGS, CTL_EOL); } -#endif /* __bsdi__ */