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 1.73.2.1 retrieving revision 1.109.4.3 diff -u -p -r1.73.2.1 -r1.109.4.3 --- src/sys/netinet6/ip6_input.c 2004/05/28 07:23:48 1.73.2.1 +++ src/sys/netinet6/ip6_input.c 2007/10/31 23:14:11 1.109.4.3 @@ -1,4 +1,4 @@ -/* $NetBSD: ip6_input.c,v 1.73.2.1 2004/05/28 07:23:48 tron Exp $ */ +/* $NetBSD: ip6_input.c,v 1.109.4.3 2007/10/31 23:14:11 joerg Exp $ */ /* $KAME: ip6_input.c,v 1.188 2001/03/29 05:34:31 itojun Exp $ */ /* @@ -62,9 +62,10 @@ */ #include -__KERNEL_RCSID(0, "$NetBSD: ip6_input.c,v 1.73.2.1 2004/05/28 07:23:48 tron Exp $"); +__KERNEL_RCSID(0, "$NetBSD: ip6_input.c,v 1.109.4.3 2007/10/31 23:14:11 joerg Exp $"); #include "opt_inet.h" +#include "opt_inet6.h" #include "opt_ipsec.h" #include "opt_pfil_hooks.h" @@ -103,6 +104,7 @@ __KERNEL_RCSID(0, "$NetBSD: ip6_input.c, #include #include #include +#include #include #include @@ -110,13 +112,16 @@ __KERNEL_RCSID(0, "$NetBSD: ip6_input.c, #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 @@ -131,7 +136,8 @@ static int ip6qmaxlen = IFQ_MAXLEN; struct in6_ifaddr *in6_ifaddr; struct ifqueue ip6intrq; -extern struct ifnet loif[NLOOP]; +extern callout_t in6_tmpaddrtimer_ch; + int ip6_forward_srcrt; /* XXX */ int ip6_sourcecheck; /* XXX */ int ip6_sourcecheck_interval; /* XXX */ @@ -142,10 +148,11 @@ struct pfil_head inet6_pfil_hook; struct ip6stat ip6stat; -static void ip6_init2 __P((void *)); +static void ip6_init2(void *); +static struct m_tag *ip6_setdstifaddr(struct mbuf *, const 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)); +static int ip6_hopopts_input(u_int32_t *, u_int32_t *, struct mbuf **, int *); +static struct mbuf *ip6_pullexthdr(struct mbuf *, size_t, int); /* * IP6 initialization: fill in IP6 protocol switch table. @@ -154,24 +161,30 @@ static struct mbuf *ip6_pullexthdr __P(( void ip6_init() { - struct ip6protosw *pr; + 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; + scope6_init(); + addrsel_policy_init(); nd6_init(); frag6_init(); + 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. */ @@ -185,13 +198,19 @@ ip6_init() } static void -ip6_init2(dummy) - void *dummy; +ip6_init2(void *dummy) { /* nd6_timer_init */ - callout_init(&nd6_timer_ch); + callout_init(&nd6_timer_ch, 0); callout_reset(&nd6_timer_ch, hz, nd6_timer, NULL); + + /* timer for regeneranation of temporary addresses randomize ID */ + callout_init(&in6_tmpaddrtimer_ch, 0); + callout_reset(&in6_tmpaddrtimer_ch, + (ip6_temp_preferred_lifetime - ip6_desync_factor - + ip6_temp_regen_advance) * hz, + in6_tmpaddrtimer, NULL); } /* @@ -209,23 +228,38 @@ 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); } } -extern struct route_in6 ip6_forward_rt; +extern struct route 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; + int hit, off = sizeof(struct ip6_hdr), nest; u_int32_t plen; u_int32_t rtalert = ~0; - int nxt, ours = 0; + int nxt, ours = 0, rh_present = 0; struct ifnet *deliverifp = NULL; int srcrt = 0; + const struct rtentry *rt; + union { + struct sockaddr dst; + struct sockaddr_in6 dst6; + } u; +#ifdef FAST_IPSEC + struct m_tag *mtag; + struct tdb_ident *tdbi; + struct secpolicy *sp; + int s, error; +#endif #ifdef IPSEC /* @@ -237,6 +271,11 @@ ip6_input(m) #endif /* + * make sure we don't have onion peering information into m_tag. + */ + ip6_delaux(m); + + /* * mbuf statistics */ if (m->m_flags & M_EXT) { @@ -245,10 +284,10 @@ 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 @@ -264,10 +303,10 @@ ip6_input(m) /* * 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 + * 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) { + 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) { @@ -293,6 +332,17 @@ 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 @@ -307,6 +357,8 @@ ip6_input(m) */ #ifdef IPSEC if (!ipsec_getnhist(m)) +#elif defined(FAST_IPSEC) + if (!ipsec_indone(m)) #else if (1) #endif @@ -379,61 +431,28 @@ 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; } /* @@ -460,34 +479,18 @@ ip6_input(m) goto hbhcheck; } + sockaddr_in6_init(&u.dst6, &ip6->ip6_dst, 0, 0, 0); + /* * 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)) + rt = rtcache_lookup2(&ip6_forward_rt, &u.dst, 1, &hit); + if (hit) 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)); - 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; - - rtalloc((struct route *)&ip6_forward_rt); - } + else + ip6stat.ip6s_forward_cachemiss++; -#define rt6_key(r) ((struct sockaddr_in6 *)((r)->rt_nodes->rn_key)) +#define rt6_getkey(__rt) satocsin6(rt_getkey(__rt)) /* * Accept the packet if the forwarding interface to the destination @@ -498,22 +501,19 @@ 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 && - (ip6_forward_rt.ro_rt->rt_flags & - (RTF_HOST|RTF_GATEWAY)) == RTF_HOST && - !(ip6_forward_rt.ro_rt->rt_flags & RTF_CLONED) && + if (rt != NULL && + (rt->rt_flags & (RTF_HOST|RTF_GATEWAY)) == RTF_HOST && + !(rt->rt_flags & RTF_CLONED) && #if 0 /* * The check below is redundant since the comparison of * the destination and the key of the rtentry has * already done through looking up the routing table. */ - IN6_ARE_ADDR_EQUAL(&ip6->ip6_dst, - &rt6_key(ip6_forward_rt.ro_rt)->sin6_addr) && + IN6_ARE_ADDR_EQUAL(&ip6->ip6_dst, &rt6_getkey(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; + rt->rt_ifp->if_type == IFT_LOOP) { + struct in6_ifaddr *ia6 = (struct in6_ifaddr *)rt->rt_ifa; if (ia6->ia6_flags & IN6_IFF_ANYCAST) m->m_flags |= M_ANYCAST6; /* @@ -541,11 +541,11 @@ ip6_input(m) */ #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 (rt != NULL && rt->rt_ifp != NULL && + 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 = rt->rt_ifp; /* faith */ goto hbhcheck; } } @@ -559,9 +559,7 @@ 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) @@ -587,6 +585,25 @@ 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) == NULL) { + struct in6_ifaddr *ia6; + + ia6 = in6_ifawithifp(deliverifp, &ip6->ip6_dst); + if (ia6 != NULL && ip6_setdstifaddr(m, ia6) == NULL) { + /* + * 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. @@ -613,7 +630,7 @@ ip6_input(m) if (ip6->ip6_plen == 0 && plen == 0) { /* * Note that if a valid jumbo payload option is - * contained, ip6_hoptops_input() must set a valid + * contained, ip6_hopopts_input() must set a valid * (non-zero) payload length to the variable plen. */ ip6stat.ip6s_badoptions++; @@ -621,7 +638,7 @@ ip6_input(m) 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; } IP6_EXTHDR_GET(hbh, struct ip6_hbh *, m, sizeof(struct ip6_hdr), @@ -720,9 +737,11 @@ ip6_input(m) in6_ifstat_inc(deliverifp, ifs6_in_deliver); nest = 0; + rh_present = 0; while (nxt != IPPROTO_DONE) { if (ip6_hdrnestlimit && (++nest > ip6_hdrnestlimit)) { ip6stat.ip6s_toomanyhdr++; + in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_hdrerr); goto bad; } @@ -736,6 +755,15 @@ ip6_input(m) goto bad; } + if (nxt == IPPROTO_ROUTING) { + if (rh_present++) { + in6_ifstat_inc(m->m_pkthdr.rcvif, + ifs6_in_hdrerr); + ip6stat.ip6s_badoptions++; + goto bad; + } + } + #ifdef IPSEC /* * enforce IPsec policy checking if we are seeing last header. @@ -748,6 +776,45 @@ 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*/ + } + splx(s); + if (error) + goto bad; + } +#endif /* FAST_IPSEC */ + nxt = (*inet6sw[ip6_protox[nxt]].pr_input)(&m, &off, nxt); } @@ -757,15 +824,46 @@ ip6_input(m) } /* + * set/grab in6_ifaddr correspond to IPv6 destination address. + */ +static struct m_tag * +ip6_setdstifaddr(struct mbuf *m, const struct in6_ifaddr *ia) +{ + struct m_tag *mtag; + + mtag = ip6_addaux(m); + if (mtag != NULL) { + struct ip6aux *ip6a; + + ip6a = (struct ip6aux *)(mtag + 1); + in6_setscope(&ip6a->ip6a_src, ia->ia_ifp, &ip6a->ip6a_scope_id); + ip6a->ip6a_src = ia->ia_addr.sin6_addr; + ip6a->ip6a_flags = ia->ia6_flags; + } + return mtag; /* NULL if failed to set */ +} + +const struct ip6aux * +ip6_getdstifaddr(struct mbuf *m) +{ + struct m_tag *mtag; + + mtag = ip6_findaux(m); + if (mtag != NULL) + return (struct ip6aux *)(mtag + 1); + 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. + * + * rtalertp - XXX: should be stored more smart way */ static int -ip6_hopopts_input(plenp, rtalertp, mp, offp) - u_int32_t *plenp; - u_int32_t *rtalertp; /* XXX: should be stored more smart way */ - struct mbuf **mp; - int *offp; +ip6_hopopts_input(u_int32_t *plenp, u_int32_t *rtalertp, + struct mbuf **mp, int *offp) { struct mbuf *m = *mp; int off = *offp, hbhlen; @@ -809,12 +907,8 @@ ip6_hopopts_input(plenp, rtalertp, mp, o * opthead + hbhlen is located in continuous memory region. */ int -ip6_process_hopopts(m, opthead, hbhlen, rtalertp, plenp) - struct mbuf *m; - u_int8_t *opthead; - int hbhlen; - u_int32_t *rtalertp; - u_int32_t *plenp; +ip6_process_hopopts(struct mbuf *m, u_int8_t *opthead, int hbhlen, + u_int32_t *rtalertp, u_int32_t *plenp) { struct ip6_hdr *ip6; int optlen = 0; @@ -849,7 +943,7 @@ ip6_process_hopopts(m, opthead, hbhlen, 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: @@ -946,10 +1040,7 @@ ip6_process_hopopts(m, opthead, hbhlen, * is not continuous in order to return an ICMPv6 error. */ int -ip6_unknown_opt(optp, m, off) - u_int8_t *optp; - struct mbuf *m; - int off; +ip6_unknown_opt(u_int8_t *optp, struct mbuf *m, int off) { struct ip6_hdr *ip6; @@ -991,76 +1082,78 @@ ip6_unknown_opt(optp, m, off) * you are using IP6_EXTHDR_CHECK() not m_pulldown()) */ void -ip6_savecontrol(in6p, mp, ip6, m) - struct in6pcb *in6p; - struct mbuf **mp; - struct ip6_hdr *ip6; - struct mbuf *m; +ip6_savecontrol(struct in6pcb *in6p, struct mbuf **mp, + 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), + *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 @@ -1068,14 +1161,14 @@ 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; ext = ip6_pullexthdr(m, sizeof(struct ip6_hdr), - ip6->ip6_nxt); + xip6->ip6_nxt); if (ext == NULL) { ip6stat.ip6s_tooshort++; return; @@ -1092,10 +1185,11 @@ ip6_savecontrol(in6p, mp, ip6, m) * 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); @@ -1104,8 +1198,8 @@ ip6_savecontrol(in6p, mp, ip6, m) /* 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 @@ -1114,7 +1208,7 @@ 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? */ + for (;;) { /* is explicit loop prevention necessary? */ struct ip6_ext *ip6e = NULL; int elen; struct mbuf *ext = NULL; @@ -1155,16 +1249,9 @@ ip6_savecontrol(in6p, mp, ip6, m) 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; @@ -1173,8 +1260,9 @@ 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; @@ -1206,16 +1294,53 @@ ip6_savecontrol(in6p, mp, ip6, m) ; } } +#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; +ip6_pullexthdr(struct mbuf *m, size_t off, int nxt) { struct ip6_ext ip6e; size_t elen; @@ -1233,7 +1358,7 @@ ip6_pullexthdr(m, off, nxt) } #endif - m_copydata(m, off, sizeof(ip6e), (caddr_t)&ip6e); + m_copydata(m, off, sizeof(ip6e), (void *)&ip6e); if (nxt == IPPROTO_AH) elen = (ip6e.ip6e_len + 2) << 2; else @@ -1256,7 +1381,7 @@ ip6_pullexthdr(m, off, nxt) return NULL; } - m_copydata(m, off, elen, mtod(n, caddr_t)); + m_copydata(m, off, elen, mtod(n, void *)); n->m_len = elen; return n; } @@ -1274,9 +1399,7 @@ ip6_pullexthdr(m, off, nxt) * we develop `neater' mechanism to process extension headers. */ u_int8_t * -ip6_get_prevhdr(m, off) - struct mbuf *m; - int off; +ip6_get_prevhdr(struct mbuf *m, int off) { struct ip6_hdr *ip6 = mtod(m, struct ip6_hdr *); @@ -1289,7 +1412,7 @@ 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: @@ -1315,11 +1438,7 @@ ip6_get_prevhdr(m, off) * get next header offset. m will be retained. */ int -ip6_nexthdr(m, off, proto, nxtp) - struct mbuf *m; - int off; - int proto; - int *nxtp; +ip6_nexthdr(struct mbuf *m, int off, int proto, int *nxtp) { struct ip6_hdr ip6; struct ip6_ext ip6e; @@ -1333,9 +1452,12 @@ 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); @@ -1348,7 +1470,7 @@ 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); + m_copydata(m, off, sizeof(fh), (void *)&fh); if ((fh.ip6f_offlg & IP6F_OFF_MASK) != 0) return -1; if (nxtp) @@ -1359,7 +1481,7 @@ 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; @@ -1372,7 +1494,7 @@ 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; @@ -1395,11 +1517,7 @@ ip6_nexthdr(m, off, proto, nxtp) * 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; +ip6_lasthdr(struct mbuf *m, int off, int proto, int *nxtp) { int newoff; int nxt; @@ -1408,7 +1526,7 @@ 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; @@ -1422,6 +1540,93 @@ ip6_lasthdr(m, off, proto, nxtp) } } +struct m_tag * +ip6_addaux(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(struct mbuf *m) +{ + struct m_tag *mtag; + + mtag = m_tag_find(m, PACKET_TAG_INET6, NULL); + return mtag; +} + +void +ip6_delaux(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 */ @@ -1437,6 +1642,11 @@ u_char inet6ctlerrmap[PRC_NCMDS] = { SYSCTL_SETUP(sysctl_net_inet6_ip6_setup, "sysctl net.inet6.ip6 subtree setup") { +#ifdef RFC2292 +#define IS2292(x, y) ((in6p->in6p_flags & IN6P_RFC2292) ? (x) : (y)) +#else +#define IS2292(x, y) (y) +#endif sysctl_createv(clog, 0, NULL, NULL, CTLFLAG_PERMANENT, @@ -1494,12 +1704,6 @@ SYSCTL_SETUP(sysctl_net_inet6_ip6_setup, 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, @@ -1597,7 +1801,7 @@ SYSCTL_SETUP(sysctl_net_inet6_ip6_setup, CTLFLAG_PERMANENT, CTLTYPE_STRING, "kame_version", SYSCTL_DESCR("KAME Version"), - NULL, 0, __KAME_VERSION, 0, + NULL, 0, __UNCONST(__KAME_VERSION), 0, CTL_NET, PF_INET6, IPPROTO_IPV6, IPV6CTL_KAME_VERSION, CTL_EOL); sysctl_createv(clog, 0, NULL, NULL, @@ -1659,9 +1863,67 @@ SYSCTL_SETUP(sysctl_net_inet6_ip6_setup, #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 }