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 1.113.8.1 retrieving revision 1.193.2.3 diff -u -p -r1.113.8.1 -r1.193.2.3 --- src/sys/netinet6/ip6_input.c 2008/02/22 02:53:33 1.113.8.1 +++ src/sys/netinet6/ip6_input.c 2018/04/16 02:00:09 1.193.2.3 @@ -1,4 +1,4 @@ -/* $NetBSD: ip6_input.c,v 1.113.8.1 2008/02/22 02:53:33 keiichi Exp $ */ +/* $NetBSD: ip6_input.c,v 1.193.2.3 2018/04/16 02:00:09 pgoyette Exp $ */ /* $KAME: ip6_input.c,v 1.188 2001/03/29 05:34:31 itojun Exp $ */ /* @@ -62,16 +62,18 @@ */ #include -__KERNEL_RCSID(0, "$NetBSD: ip6_input.c,v 1.113.8.1 2008/02/22 02:53:33 keiichi Exp $"); +__KERNEL_RCSID(0, "$NetBSD: ip6_input.c,v 1.193.2.3 2018/04/16 02:00:09 pgoyette Exp $"); +#ifdef _KERNEL_OPT +#include "opt_gateway.h" #include "opt_inet.h" #include "opt_inet6.h" #include "opt_ipsec.h" -#include "opt_pfil_hooks.h" +#include "opt_net_mpsafe.h" +#endif #include #include -#include #include #include #include @@ -83,25 +85,28 @@ __KERNEL_RCSID(0, "$NetBSD: ip6_input.c, #include #include #include +#include +#include #include #include #include #include -#include -#ifdef PFIL_HOOKS +#include #include -#endif #include #include #ifdef INET #include +#include #include #endif /* INET */ #include +#include #include #include +#include #include #include #include @@ -109,61 +114,58 @@ __KERNEL_RCSID(0, "$NetBSD: ip6_input.c, #include #ifdef IPSEC -#include -#endif - -#ifdef FAST_IPSEC #include #include #include -#endif /* FAST_IPSEC */ +#endif /* IPSEC */ #include #include "faith.h" -#include "gif.h" - -#if NGIF > 0 -#include -#endif #include extern struct domain inet6domain; u_char ip6_protox[IPPROTO_MAX]; -static int ip6qmaxlen = IFQ_MAXLEN; -struct in6_ifaddr *in6_ifaddr; -struct ifqueue ip6intrq; - -extern callout_t in6_tmpaddrtimer_ch; - -int ip6_forward_srcrt; /* XXX */ -int ip6_sourcecheck; /* XXX */ -int ip6_sourcecheck_interval; /* XXX */ +pktqueue_t *ip6_pktq __read_mostly; -#ifdef PFIL_HOOKS -struct pfil_head inet6_pfil_hook; -#endif +pfil_head_t *inet6_pfil_hook; + +percpu_t *ip6stat_percpu; -struct ip6stat ip6stat; +percpu_t *ip6_forward_rt_percpu __cacheline_aligned; -static void ip6_init2(void *); +static void ip6_init2(void); +static void ip6intr(void *); static struct m_tag *ip6_setdstifaddr(struct mbuf *, const struct in6_ifaddr *); -static int ip6_hopopts_input(u_int32_t *, u_int32_t *, struct mbuf **, int *); +static int ip6_process_hopopts(struct mbuf *, u_int8_t *, int, u_int32_t *, + u_int32_t *); static struct mbuf *ip6_pullexthdr(struct mbuf *, size_t, int); +static void sysctl_net_inet6_ip6_setup(struct sysctllog **); + +#ifdef NET_MPSAFE +#define SOFTNET_LOCK() mutex_enter(softnet_lock) +#define SOFTNET_UNLOCK() mutex_exit(softnet_lock) +#else +#define SOFTNET_LOCK() KASSERT(mutex_owned(softnet_lock)) +#define SOFTNET_UNLOCK() KASSERT(mutex_owned(softnet_lock)) +#endif /* * IP6 initialization: fill in IP6 protocol switch table. * All protocols not implemented in kernel go to raw IP6 protocol handler. */ void -ip6_init() +ip6_init(void) { const struct ip6protosw *pr; int i; + in6_init(); + + sysctl_net_inet6_ip6_setup(NULL); pr = (const struct ip6protosw *)pffindproto(PF_INET6, IPPROTO_RAW, SOCK_RAW); if (pr == 0) panic("ip6_init"); @@ -174,39 +176,34 @@ ip6_init() 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; + + ip6_pktq = pktq_create(IFQ_MAXLEN, ip6intr, NULL); + KASSERT(ip6_pktq != NULL); + scope6_init(); addrsel_policy_init(); nd6_init(); frag6_init(); - ip6_desync_factor = arc4random() % MAX_TEMP_DESYNC_FACTOR; + ip6_desync_factor = cprng_fast32() % MAX_TEMP_DESYNC_FACTOR; - ip6_init2((void *)0); + ip6_init2(); #ifdef GATEWAY ip6flow_init(ip6_hashsize); #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 */ + inet6_pfil_hook = pfil_head_create(PFIL_TYPE_AF, (void *)AF_INET6); + KASSERT(inet6_pfil_hook != NULL); + + ip6stat_percpu = percpu_alloc(sizeof(uint64_t) * IP6_NSTATS); + ip6_forward_rt_percpu = percpu_alloc(sizeof(struct route)); } static void -ip6_init2(void *dummy) +ip6_init2(void) { - /* nd6_timer_init */ - 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); + /* timer for regeneration of temporary addresses randomize ID */ + callout_init(&in6_tmpaddrtimer_ch, CALLOUT_MPSAFE); callout_reset(&in6_tmpaddrtimer_ch, (ip6_temp_preferred_lifetime - ip6_desync_factor - ip6_temp_regen_advance) * hz, @@ -216,59 +213,52 @@ ip6_init2(void *dummy) /* * IP6 input interrupt handling. Just pass the packet to ip6_input. */ -void -ip6intr() +static void +ip6intr(void *arg __unused) { - int s; struct mbuf *m; - for (;;) { - s = splnet(); - IF_DEQUEUE(&ip6intrq, m); - 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)) { + SOFTNET_KERNEL_LOCK_UNLESS_NET_MPSAFE(); + while ((m = pktq_dequeue(ip6_pktq)) != NULL) { + struct psref psref; + struct ifnet *rcvif = m_get_rcvif_psref(m, &psref); + + if (rcvif == NULL) { m_freem(m); - return; + continue; + } + /* + * Drop the packet if IPv6 is disabled on the interface. + */ + if ((ND_IFINFO(rcvif)->flags & ND6_IFF_IFDISABLED)) { + m_put_rcvif_psref(rcvif, &psref); + m_freem(m); + continue; } - ip6_input(m); + ip6_input(m, rcvif); + m_put_rcvif_psref(rcvif, &psref); } + SOFTNET_KERNEL_UNLOCK_UNLESS_NET_MPSAFE(); } -extern struct route ip6_forward_rt; - void -ip6_input(struct mbuf *m) +ip6_input(struct mbuf *m, struct ifnet *rcvif) { struct ip6_hdr *ip6; int hit, off = sizeof(struct ip6_hdr), nest; u_int32_t plen; u_int32_t rtalert = ~0; - int nxt, ours = 0, rh_present = 0; + int nxt, ours = 0, rh_present = 0, frg_present; struct ifnet *deliverifp = NULL; int srcrt = 0; - const struct rtentry *rt; + struct rtentry *rt = NULL; 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 + struct route *ro; -#ifdef IPSEC - /* - * should the inner packet be considered authentic? - * see comment in ah4_input(). - */ - m->m_flags &= ~M_AUTHIPHDR; - m->m_flags &= ~M_AUTHIPDGM; -#endif + KASSERT(rcvif != NULL); /* * make sure we don't have onion peering information into m_tag. @@ -280,25 +270,25 @@ ip6_input(struct mbuf *m) */ if (m->m_flags & M_EXT) { if (m->m_next) - ip6stat.ip6s_mext2m++; + IP6_STATINC(IP6_STAT_MEXT2M); else - ip6stat.ip6s_mext1++; + IP6_STATINC(IP6_STAT_MEXT1); } else { -#define M2MMAX __arraycount(ip6stat.ip6s_m2m) +#define M2MMAX 32 if (m->m_next) { - if (m->m_flags & M_LOOP) { - 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]++; + if (m->m_flags & M_LOOP) + /*XXX*/ IP6_STATINC(IP6_STAT_M2M + lo0ifp->if_index); + else if (rcvif->if_index < M2MMAX) + IP6_STATINC(IP6_STAT_M2M + rcvif->if_index); else - ip6stat.ip6s_m2m[0]++; + IP6_STATINC(IP6_STAT_M2M); } else - ip6stat.ip6s_m1++; + IP6_STATINC(IP6_STAT_M1); #undef M2MMAX } - in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_receive); - ip6stat.ip6s_total++; + in6_ifstat_inc(rcvif, ifs6_in_receive); + IP6_STATINC(IP6_STAT_TOTAL); /* * If the IPv6 header is not aligned, slurp it up into a new @@ -307,19 +297,17 @@ ip6_input(struct mbuf *m) * 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) { + (max_linkhdr + 3) & ~3)) == NULL) { /* XXXJRT new stat, please */ - ip6stat.ip6s_toosmall++; - in6_ifstat_inc(inifp, ifs6_in_hdrerr); + IP6_STATINC(IP6_STAT_TOOSMALL); + in6_ifstat_inc(rcvif, 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); + IP6_STATINC(IP6_STAT_TOOSMALL); + in6_ifstat_inc(rcvif, ifs6_in_hdrerr); return; } } @@ -327,64 +315,11 @@ ip6_input(struct mbuf *m) ip6 = mtod(m, struct ip6_hdr *); if ((ip6->ip6_vfc & IPV6_VERSION_MASK) != IPV6_VERSION) { - ip6stat.ip6s_badvers++; - in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_hdrerr); + IP6_STATINC(IP6_STAT_BADVERS); + in6_ifstat_inc(rcvif, ifs6_in_hdrerr); 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 - * 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. - */ - /* - * let ipfilter look at packet on the wire, - * not the decapsulated packet. - */ -#ifdef IPSEC - 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 - if (altq_input != NULL && (*altq_input)(m, AF_INET6) == 0) { - /* packet is dropped by traffic conditioner */ - return; - } -#endif - /* * Check against address spoofing/corruption. */ @@ -393,10 +328,11 @@ ip6_input(struct mbuf *m) /* * XXX: "badscope" is not very suitable for a multicast source. */ - ip6stat.ip6s_badscope++; - in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_addrerr); + IP6_STATINC(IP6_STAT_BADSCOPE); + in6_ifstat_inc(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 @@ -411,10 +347,11 @@ ip6_input(struct mbuf *m) */ 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); + IP6_STATINC(IP6_STAT_BADSCOPE); + in6_ifstat_inc(rcvif, ifs6_in_addrerr); goto bad; } + #if 0 /* * Reject packets with IPv4 compatible addresses (auto tunnel). @@ -425,13 +362,60 @@ ip6_input(struct mbuf *m) */ 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); + IP6_STATINC(IP6_STAT_BADSCOPE); + in6_ifstat_inc(rcvif, ifs6_in_addrerr); goto bad; } #endif /* + * Assume that we can create a fast-forward IP flow entry + * based on this packet. + */ + m->m_flags |= M_CANFASTFWD; + + /* + * 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. + * + * Don't call hooks if the packet has already been processed by + * IPsec (encapsulated, tunnel mode). + */ +#if defined(IPSEC) + if (!ipsec_used || !ipsec_indone(m)) +#else + if (1) +#endif + { + struct in6_addr odst; + + odst = ip6->ip6_dst; + if (pfil_run_hooks(inet6_pfil_hook, &m, rcvif, PFIL_IN) != 0) + return; + if (m == NULL) + return; + ip6 = mtod(m, struct ip6_hdr *); + srcrt = !IN6_ARE_ADDR_EQUAL(&odst, &ip6->ip6_dst); + } + + IP6_STATINC(IP6_STAT_NXTHIST + ip6->ip6_nxt); + +#ifdef ALTQ + if (altq_input != NULL) { + SOFTNET_LOCK(); + if ((*altq_input)(m, AF_INET6) == 0) { + SOFTNET_UNLOCK(); + /* packet is dropped by traffic conditioner */ + return; + } + SOFTNET_UNLOCK(); + } +#endif + + /* * 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 @@ -439,63 +423,70 @@ ip6_input(struct mbuf *m) * 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. + * 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 */ + IP6_STATINC(IP6_STAT_BADSCOPE); /* XXX */ goto bad; } - if (in6_setscope(&ip6->ip6_src, m->m_pkthdr.rcvif, NULL) || - in6_setscope(&ip6->ip6_dst, m->m_pkthdr.rcvif, NULL)) { - ip6stat.ip6s_badscope++; + if (in6_setscope(&ip6->ip6_src, rcvif, NULL) || + in6_setscope(&ip6->ip6_dst, rcvif, NULL)) { + IP6_STATINC(IP6_STAT_BADSCOPE); goto bad; } + ro = percpu_getref(ip6_forward_rt_percpu); + /* * Multicast check */ if (IN6_IS_ADDR_MULTICAST(&ip6->ip6_dst)) { - struct in6_multi *in6m = 0; + bool ingroup; - in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_mcast); + in6_ifstat_inc(rcvif, ifs6_in_mcast); /* * See if we belong to the destination multicast group on the * arrival interface. */ - IN6_LOOKUP_MULTI(ip6->ip6_dst, m->m_pkthdr.rcvif, in6m); - if (in6m) + ingroup = in6_multi_group(&ip6->ip6_dst, rcvif); + if (ingroup) { ours = 1; - else if (!ip6_mrouter) { - ip6stat.ip6s_notmember++; - ip6stat.ip6s_cantforward++; - in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_discard); - goto bad; + } else if (!ip6_mrouter) { + uint64_t *ip6s = IP6_STAT_GETREF(); + ip6s[IP6_STAT_NOTMEMBER]++; + ip6s[IP6_STAT_CANTFORWARD]++; + IP6_STAT_PUTREF(); + in6_ifstat_inc(rcvif, ifs6_in_discard); + goto bad_unref; } - deliverifp = m->m_pkthdr.rcvif; + deliverifp = rcvif; goto hbhcheck; } sockaddr_in6_init(&u.dst6, &ip6->ip6_dst, 0, 0, 0); /* - * Unicast check + * Unicast check */ - rt = rtcache_lookup2(&ip6_forward_rt, &u.dst, 1, &hit); + rt = rtcache_lookup2(ro, &u.dst, 1, &hit); if (hit) - ip6stat.ip6s_forward_cachehit++; + IP6_STATINC(IP6_STAT_FORWARD_CACHEHIT); else - ip6stat.ip6s_forward_cachemiss++; - -#define rt6_getkey(__rt) satocsin6(rt_getkey(__rt)) + IP6_STATINC(IP6_STAT_FORWARD_CACHEMISS); /* * Accept the packet if the forwarding interface to the destination - * according to the routing table is the loopback interface, + * (according to the routing table) is the loopback interface, * unless the associated route has a gateway. + * + * We don't explicitly match ip6_dst against an interface here. It + * is already done in rtcache_lookup2: rt->rt_ifp->if_type will be + * IFT_LOOP if the packet is for us. + * * Note that this approach causes to accept a packet if there is a * route to the loopback interface for the destination of the packet. * But we think it's even useful in some situations, e.g. when using @@ -503,71 +494,51 @@ ip6_input(struct mbuf *m) */ 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_getkey(rt)->sin6_addr) && -#endif -#ifdef MOBILE_IPV6 - ((rt->rt_flags & RTF_ANNOUNCE) || - rt->rt_ifp->if_type == IFT_LOOP) -#else - rt->rt_ifp->if_type == IFT_LOOP -#endif /* MOBILE_IPV6 */ - ) { + rt->rt_ifp->if_type == IFT_LOOP) { struct in6_ifaddr *ia6 = (struct in6_ifaddr *)rt->rt_ifa; + int addrok; + if (ia6->ia6_flags & IN6_IFF_ANYCAST) m->m_flags |= M_ANYCAST6; -#ifdef MOBILE_IPV6 - /* check unicast NS */ - if ((rtcache_validate(&ip6_forward_rt)->rt_flags & RTF_ANNOUNCE) != 0) { - /* This route shows proxy nd. thus the packet was - * captured. this packet should be tunneled to - * actual coa with tunneling unless this is NS. - */ - int nh, loff; - struct icmp6_hdr *icp; - loff = ip6_lasthdr(m, 0, IPPROTO_IPV6, &nh); - if (loff < 0 || nh != IPPROTO_ICMPV6) - goto mip6_forwarding; - IP6_EXTHDR_GET(icp, struct icmp6_hdr *, m, loff, - sizeof(*icp)); - if (icp == NULL) { - icmp6stat.icp6s_tooshort++; - return; - } - if (icp->icmp6_type != ND_NEIGHBOR_SOLICIT) - goto mip6_forwarding; - } -#endif /* MOBILE_IPV6 */ /* * packets to a tentative, duplicated, or somehow invalid * address must not be accepted. */ - if (!(ia6->ia6_flags & IN6_IFF_NOTREADY)) { + if (ia6->ia6_flags & IN6_IFF_NOTREADY) + addrok = 0; + else if (ia6->ia6_flags & IN6_IFF_DETACHED && + !IN6_IS_ADDR_UNSPECIFIED(&ip6->ip6_src)) + { + /* Allow internal traffic to DETACHED addresses */ + struct sockaddr_in6 sin6; + int s; + + memset(&sin6, 0, sizeof(sin6)); + sin6.sin6_family = AF_INET6; + sin6.sin6_len = sizeof(sin6); + sin6.sin6_addr = ip6->ip6_src; + s = pserialize_read_enter(); + addrok = (ifa_ifwithaddr(sin6tosa(&sin6)) != NULL); + pserialize_read_exit(s); + } else + addrok = 1; + if (addrok) { /* this address is ready */ ours = 1; deliverifp = ia6->ia_ifp; /* correct? */ goto hbhcheck; } else { /* 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))); + char ip6bufs[INET6_ADDRSTRLEN]; + char ip6bufd[INET6_ADDRSTRLEN]; + nd6log(LOG_INFO, "packet to an unready address %s->%s\n", + IN6_PRINT(ip6bufs, &ip6->ip6_src), + IN6_PRINT(ip6bufd, &ip6->ip6_dst)); - goto bad; + goto bad_unref; } } -#ifdef MOBILE_IPV6 - mip6_forwarding: -#endif /* MOBILE_IPV6 */ - /* * FAITH (Firewall Aided Internet Translator) */ @@ -583,49 +554,29 @@ ip6_input(struct mbuf *m) } #endif -#if 0 - { - /* - * Last resort: check in6_ifaddr for incoming interface. - * The code is here until I update the "goto ours hack" code above - * working right. - */ - struct ifaddr *ifa; - IFADDR_FOREACH(ifa, m->m_pkthdr.rcvif) { - if (ifa->ifa_addr == NULL) - continue; /* just for safety */ - if (ifa->ifa_addr->sa_family != AF_INET6) - continue; - if (IN6_ARE_ADDR_EQUAL(IFA_IN6(ifa), &ip6->ip6_dst)) { - ours = 1; - deliverifp = ifa->ifa_ifp; - goto hbhcheck; - } - } - } -#endif - /* * Now there is no reason to process the packet if it's not our own * and we're not a router. */ if (!ip6_forwarding) { - ip6stat.ip6s_cantforward++; - in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_discard); - goto bad; + IP6_STATINC(IP6_STAT_CANTFORWARD); + in6_ifstat_inc(rcvif, ifs6_in_discard); + goto bad_unref; } - hbhcheck: +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 + * 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; + int s = pserialize_read_enter(); ia6 = in6_ifawithifp(deliverifp, &ip6->ip6_dst); + /* Depends on ip6_setdstifaddr never sleep */ if (ia6 != NULL && ip6_setdstifaddr(m, ia6) == NULL) { /* * XXX maybe we should drop the packet here, @@ -633,6 +584,7 @@ ip6_input(struct mbuf *m) * to the upper layers. */ } + pserialize_read_exit(s); } /* @@ -645,10 +597,11 @@ ip6_input(struct mbuf *m) struct ip6_hbh *hbh; if (ip6_hopopts_input(&plen, &rtalert, &m, &off)) { -#if 0 /*touches NULL pointer*/ - in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_discard); -#endif - return; /* m have already been freed */ + /* m already freed */ + in6_ifstat_inc(rcvif, ifs6_in_discard); + rtcache_unref(rt, ro); + percpu_putref(ip6_forward_rt_percpu); + return; } /* adjust pointer */ @@ -665,18 +618,22 @@ ip6_input(struct mbuf *m) * 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); + IP6_STATINC(IP6_STAT_BADOPTIONS); + in6_ifstat_inc(rcvif, ifs6_in_discard); + in6_ifstat_inc(rcvif, ifs6_in_hdrerr); icmp6_error(m, ICMP6_PARAM_PROB, ICMP6_PARAMPROB_HEADER, (char *)&ip6->ip6_plen - (char *)ip6); + rtcache_unref(rt, ro); + percpu_putref(ip6_forward_rt_percpu); return; } IP6_EXTHDR_GET(hbh, struct ip6_hbh *, m, sizeof(struct ip6_hdr), sizeof(struct ip6_hbh)); if (hbh == NULL) { - ip6stat.ip6s_tooshort++; + IP6_STATINC(IP6_STAT_TOOSHORT); + rtcache_unref(rt, ro); + percpu_putref(ip6_forward_rt_percpu); return; } KASSERT(IP6_HDR_ALIGNED_P(hbh)); @@ -692,15 +649,14 @@ ip6_input(struct mbuf *m) nxt = ip6->ip6_nxt; /* - * Check that the amount of data in the buffers - * is as at least much as the IPv6 header would have us expect. - * Trim mbufs if longer than we expect. - * Drop packet if shorter than we expect. + * Check that the amount of data in the buffers is at least much as + * the IPv6 header would have us expect. Trim mbufs if longer than we + * expect. Drop packet if shorter than we expect. */ if (m->m_pkthdr.len - sizeof(struct ip6_hdr) < plen) { - ip6stat.ip6s_tooshort++; - in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_truncated); - goto bad; + IP6_STATINC(IP6_STAT_TOOSHORT); + in6_ifstat_inc(rcvif, ifs6_in_truncated); + goto bad_unref; } if (m->m_pkthdr.len > sizeof(struct ip6_hdr) + plen) { if (m->m_len == m->m_pkthdr.len) { @@ -722,16 +678,25 @@ ip6_input(struct mbuf *m) * ip6_mforward() returns a non-zero value, the packet * must be discarded, else it may be accepted below. */ - if (ip6_mrouter && ip6_mforward(ip6, m->m_pkthdr.rcvif, m)) { - ip6stat.ip6s_cantforward++; - m_freem(m); - return; - } - if (!ours) { - m_freem(m); - return; + if (ip6_mrouter != NULL) { + int error; + + SOFTNET_LOCK(); + error = ip6_mforward(ip6, rcvif, m); + SOFTNET_UNLOCK(); + + if (error != 0) { + rtcache_unref(rt, ro); + percpu_putref(ip6_forward_rt_percpu); + IP6_STATINC(IP6_STAT_CANTFORWARD); + goto bad; + } } + if (!ours) + goto bad_unref; } else if (!ours) { + rtcache_unref(rt, ro); + percpu_putref(ip6_forward_rt_percpu); ip6_forward(m, srcrt); return; } @@ -749,112 +714,94 @@ ip6_input(struct mbuf *m) */ 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; + IP6_STATINC(IP6_STAT_BADSCOPE); + in6_ifstat_inc(rcvif, ifs6_in_addrerr); + goto bad_unref; } - /* - * Tell launch routine the next header - */ #ifdef IFA_STATS if (deliverifp != NULL) { struct in6_ifaddr *ia6; + int s = pserialize_read_enter(); ia6 = in6_ifawithifp(deliverifp, &ip6->ip6_dst); if (ia6) ia6->ia_ifa.ifa_data.ifad_inbytes += m->m_pkthdr.len; + pserialize_read_exit(s); } #endif - ip6stat.ip6s_delivered++; + IP6_STATINC(IP6_STAT_DELIVERED); in6_ifstat_inc(deliverifp, ifs6_in_deliver); nest = 0; + if (rt != NULL) { + rtcache_unref(rt, ro); + rt = NULL; + } + percpu_putref(ip6_forward_rt_percpu); + rh_present = 0; + frg_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); + IP6_STATINC(IP6_STAT_TOOMANYHDR); + in6_ifstat_inc(rcvif, ifs6_in_hdrerr); goto bad; } + M_VERIFY_PACKET(m); + /* * protection against faulty packet - there should be * more sanity checks in header chain processing. */ if (m->m_pkthdr.len < off) { - ip6stat.ip6s_tooshort++; - in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_truncated); + IP6_STATINC(IP6_STAT_TOOSHORT); + in6_ifstat_inc(rcvif, ifs6_in_truncated); goto bad; } if (nxt == IPPROTO_ROUTING) { if (rh_present++) { - in6_ifstat_inc(m->m_pkthdr.rcvif, - ifs6_in_hdrerr); - ip6stat.ip6s_badoptions++; + in6_ifstat_inc(rcvif, ifs6_in_hdrerr); + IP6_STATINC(IP6_STAT_BADOPTIONS); + goto bad; + } + } else if (nxt == IPPROTO_FRAGMENT) { + if (frg_present++) { + in6_ifstat_inc(rcvif, ifs6_in_hdrerr); + IP6_STATINC(IP6_STAT_BADOPTIONS); 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 -#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) { + if (ipsec_used) { /* - * Check security policy against packet attributes. + * 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. */ - error = ipsec_in_reject(sp, m); - KEY_FREESP(&sp); - } else { - /* XXX error stat??? */ - error = EINVAL; - DPRINTF(("ip6_input: no SP, packet discarded\n"));/*XXX*/ + if ((inet6sw[ip_protox[nxt]].pr_flags + & PR_LASTHDR) != 0) { + int error; + + error = ipsec6_input(m); + if (error) + goto bad; + } } - splx(s); - if (error) - goto bad; - } -#endif /* FAST_IPSEC */ - - if (dest6_mip6_hao(m, off, nxt) < 0) - goto bad; +#endif nxt = (*inet6sw[ip6_protox[nxt]].pr_input)(&m, &off, nxt); } return; - bad: + +bad_unref: + rtcache_unref(rt, ro); + percpu_putref(ip6_forward_rt_percpu); +bad: m_freem(m); + return; } /* @@ -864,17 +811,21 @@ static struct m_tag * ip6_setdstifaddr(struct mbuf *m, const struct in6_ifaddr *ia) { struct m_tag *mtag; + struct ip6aux *ip6a; mtag = ip6_addaux(m); - if (mtag != NULL) { - struct ip6aux *ip6a; + if (mtag == NULL) + return NULL; - ip6a = (struct ip6aux *)(mtag + 1); - in6_setscope(&ip6a->ip6a_src, ia->ia_ifp, &ip6a->ip6a_src_scope_id); - ip6a->ip6a_src = ia->ia_addr.sin6_addr; - ip6a->ip6a_src_flags = ia->ia6_flags; + ip6a = (struct ip6aux *)(mtag + 1); + if (in6_setscope(&ip6a->ip6a_src, ia->ia_ifp, &ip6a->ip6a_scope_id)) { + IP6_STATINC(IP6_STAT_BADSCOPE); + return NULL; } - return mtag; /* NULL if failed to set */ + + ip6a->ip6a_src = ia->ia_addr.sin6_addr; + ip6a->ip6a_flags = ia->ia6_flags; + return mtag; } const struct ip6aux * @@ -895,7 +846,7 @@ ip6_getdstifaddr(struct mbuf *m) * * rtalertp - XXX: should be stored more smart way */ -static int +int ip6_hopopts_input(u_int32_t *plenp, u_int32_t *rtalertp, struct mbuf **mp, int *offp) { @@ -905,16 +856,16 @@ ip6_hopopts_input(u_int32_t *plenp, u_in /* validation of the length of the header */ IP6_EXTHDR_GET(hbh, struct ip6_hbh *, m, - sizeof(struct ip6_hdr), sizeof(struct ip6_hbh)); + sizeof(struct ip6_hdr), sizeof(struct ip6_hbh)); if (hbh == NULL) { - ip6stat.ip6s_tooshort++; + IP6_STATINC(IP6_STAT_TOOSHORT); return -1; } hbhlen = (hbh->ip6h_len + 1) << 3; IP6_EXTHDR_GET(hbh, struct ip6_hbh *, m, sizeof(struct ip6_hdr), - hbhlen); + hbhlen); if (hbh == NULL) { - ip6stat.ip6s_tooshort++; + IP6_STATINC(IP6_STAT_TOOSHORT); return -1; } KASSERT(IP6_HDR_ALIGNED_P(hbh)); @@ -922,12 +873,12 @@ ip6_hopopts_input(u_int32_t *plenp, u_in hbhlen -= sizeof(struct ip6_hbh); if (ip6_process_hopopts(m, (u_int8_t *)hbh + sizeof(struct ip6_hbh), - hbhlen, rtalertp, plenp) < 0) - return (-1); + hbhlen, rtalertp, plenp) < 0) + return -1; *offp = off; *mp = m; - return (0); + return 0; } /* @@ -940,7 +891,7 @@ ip6_hopopts_input(u_int32_t *plenp, u_in * (RFC2460 p7), opthead is pointer into data content in m, and opthead to * opthead + hbhlen is located in continuous memory region. */ -int +static int ip6_process_hopopts(struct mbuf *m, u_int8_t *opthead, int hbhlen, u_int32_t *rtalertp, u_int32_t *plenp) { @@ -958,7 +909,7 @@ ip6_process_hopopts(struct mbuf *m, u_in break; case IP6OPT_PADN: if (hbhlen < IP6OPT_MINLEN) { - ip6stat.ip6s_toosmall++; + IP6_STATINC(IP6_STAT_TOOSMALL); goto bad; } optlen = *(opt + 1) + 2; @@ -966,7 +917,7 @@ ip6_process_hopopts(struct mbuf *m, u_in case IP6OPT_RTALERT: /* XXX may need check for alignment */ if (hbhlen < IP6OPT_RTALERT_LEN) { - ip6stat.ip6s_toosmall++; + IP6_STATINC(IP6_STAT_TOOSMALL); goto bad; } if (*(opt + 1) != IP6OPT_RTALERT_LEN - 2) { @@ -977,13 +928,13 @@ ip6_process_hopopts(struct mbuf *m, u_in return (-1); } optlen = IP6OPT_RTALERT_LEN; - bcopy((void *)(opt + 2), (void *)&rtalert_val, 2); + memcpy((void *)&rtalert_val, (void *)(opt + 2), 2); *rtalertp = ntohs(rtalert_val); break; case IP6OPT_JUMBO: /* XXX may need check for alignment */ if (hbhlen < IP6OPT_JUMBO_LEN) { - ip6stat.ip6s_toosmall++; + IP6_STATINC(IP6_STAT_TOOSMALL); goto bad; } if (*(opt + 1) != IP6OPT_JUMBO_LEN - 2) { @@ -1001,7 +952,7 @@ ip6_process_hopopts(struct mbuf *m, u_in */ ip6 = mtod(m, struct ip6_hdr *); if (ip6->ip6_plen) { - ip6stat.ip6s_badoptions++; + IP6_STATINC(IP6_STAT_BADOPTIONS); icmp6_error(m, ICMP6_PARAM_PROB, ICMP6_PARAMPROB_HEADER, erroff + opt - opthead); @@ -1010,9 +961,9 @@ ip6_process_hopopts(struct mbuf *m, u_in /* * We may see jumbolen in unaligned location, so - * we'd need to perform bcopy(). + * we'd need to perform memcpy(). */ - bcopy(opt + 2, &jumboplen, sizeof(jumboplen)); + memcpy(&jumboplen, opt + 2, sizeof(jumboplen)); jumboplen = (u_int32_t)htonl(jumboplen); #if 1 @@ -1025,7 +976,7 @@ ip6_process_hopopts(struct mbuf *m, u_in * there's no explicit mention in specification. */ if (*plenp != 0) { - ip6stat.ip6s_badoptions++; + IP6_STATINC(IP6_STAT_BADOPTIONS); icmp6_error(m, ICMP6_PARAM_PROB, ICMP6_PARAMPROB_HEADER, erroff + opt + 2 - opthead); @@ -1037,7 +988,7 @@ ip6_process_hopopts(struct mbuf *m, u_in * jumbo payload length must be larger than 65535. */ if (jumboplen <= IPV6_MAXPACKET) { - ip6stat.ip6s_badoptions++; + IP6_STATINC(IP6_STAT_BADOPTIONS); icmp6_error(m, ICMP6_PARAM_PROB, ICMP6_PARAMPROB_HEADER, erroff + opt + 2 - opthead); @@ -1048,7 +999,7 @@ ip6_process_hopopts(struct mbuf *m, u_in break; default: /* unknown option */ if (hbhlen < IP6OPT_MINLEN) { - ip6stat.ip6s_toosmall++; + IP6_STATINC(IP6_STAT_TOOSMALL); goto bad; } optlen = ip6_unknown_opt(opt, m, @@ -1085,11 +1036,11 @@ ip6_unknown_opt(u_int8_t *optp, struct m m_freem(m); return (-1); case IP6OPT_TYPE_FORCEICMP: /* send ICMP even if multicasted */ - ip6stat.ip6s_badoptions++; + IP6_STATINC(IP6_STAT_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_STATINC(IP6_STAT_BADOPTIONS); ip6 = mtod(m, struct ip6_hdr *); if (IN6_IS_ADDR_MULTICAST(&ip6->ip6_dst) || (m->m_flags & (M_BCAST|M_MCAST))) @@ -1119,23 +1070,15 @@ void ip6_savecontrol(struct in6pcb *in6p, struct mbuf **mp, struct ip6_hdr *ip6, struct mbuf *m) { + struct socket *so = in6p->in6p_socket; #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((void *) &tv, sizeof(tv), - SCM_TIMESTAMP, SOL_SOCKET); - if (*mp) - mp = &(*mp)->m_next; - } -#endif + if (SOOPT_TIMESTAMP(so->so_options)) + mp = sbsavetimestamp(so->so_options, m, mp); /* some OSes call this logic with IPv4 packet, for SO_TIMESTAMP */ if ((ip6->ip6_vfc & IPV6_VERSION_MASK) != IPV6_VERSION) @@ -1145,12 +1088,10 @@ ip6_savecontrol(struct in6pcb *in6p, str if ((in6p->in6p_flags & IN6P_PKTINFO) != 0) { struct in6_pktinfo pi6; - bcopy(&ip6->ip6_dst, &pi6.ipi6_addr, sizeof(struct in6_addr)); + memcpy(&pi6.ipi6_addr, &ip6->ip6_dst, sizeof(struct in6_addr)); 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), + pi6.ipi6_ifindex = m->m_pkthdr.rcvif_index; + *mp = sbcreatecontrol(&pi6, sizeof(pi6), IS2292(IPV6_2292PKTINFO, IPV6_PKTINFO), IPPROTO_IPV6); if (*mp) mp = &(*mp)->m_next; @@ -1159,7 +1100,7 @@ ip6_savecontrol(struct in6pcb *in6p, str if (in6p->in6p_flags & IN6P_HOPLIMIT) { int hlim = ip6->ip6_hlim & 0xff; - *mp = sbcreatecontrol((void *) &hlim, sizeof(int), + *mp = sbcreatecontrol(&hlim, sizeof(hlim), IS2292(IPV6_2292HOPLIMIT, IPV6_HOPLIMIT), IPPROTO_IPV6); if (*mp) mp = &(*mp)->m_next; @@ -1173,7 +1114,7 @@ ip6_savecontrol(struct in6pcb *in6p, str flowinfo >>= 20; tclass = flowinfo & 0xff; - *mp = sbcreatecontrol((void *)&tclass, sizeof(tclass), + *mp = sbcreatecontrol(&tclass, sizeof(tclass), IPV6_TCLASS, IPPROTO_IPV6); if (*mp) @@ -1204,14 +1145,14 @@ ip6_savecontrol(struct in6pcb *in6p, str ext = ip6_pullexthdr(m, sizeof(struct ip6_hdr), xip6->ip6_nxt); if (ext == NULL) { - ip6stat.ip6s_tooshort++; + IP6_STATINC(IP6_STAT_TOOSHORT); return; } hbh = mtod(ext, struct ip6_hbh *); hbhlen = (hbh->ip6h_len + 1) << 3; if (hbhlen != ext->m_len) { m_freem(ext); - ip6stat.ip6s_tooshort++; + IP6_STATINC(IP6_STAT_TOOSHORT); return; } @@ -1221,7 +1162,7 @@ ip6_savecontrol(struct in6pcb *in6p, str * be removed before returning in the RFC 2292. * Note: this constraint is removed in RFC3542. */ - *mp = sbcreatecontrol((void *)hbh, hbhlen, + *mp = sbcreatecontrol(hbh, hbhlen, IS2292(IPV6_2292HOPOPTS, IPV6_HOPOPTS), IPPROTO_IPV6); if (*mp) @@ -1263,7 +1204,7 @@ ip6_savecontrol(struct in6pcb *in6p, str ext = ip6_pullexthdr(m, off, nxt); if (ext == NULL) { - ip6stat.ip6s_tooshort++; + IP6_STATINC(IP6_STAT_TOOSHORT); return; } ip6e = mtod(ext, struct ip6_ext *); @@ -1273,17 +1214,17 @@ ip6_savecontrol(struct in6pcb *in6p, str elen = (ip6e->ip6e_len + 1) << 3; if (elen != ext->m_len) { m_freem(ext); - ip6stat.ip6s_tooshort++; + IP6_STATINC(IP6_STAT_TOOSHORT); return; } KASSERT(IP6_HDR_ALIGNED_P(ip6e)); switch (nxt) { case IPPROTO_DSTOPTS: - if (!in6p->in6p_flags & IN6P_DSTOPTS) + if (!(in6p->in6p_flags & IN6P_DSTOPTS)) break; - *mp = sbcreatecontrol((void *)ip6e, elen, + *mp = sbcreatecontrol(ip6e, elen, IS2292(IPV6_2292DSTOPTS, IPV6_DSTOPTS), IPPROTO_IPV6); if (*mp) @@ -1291,10 +1232,10 @@ ip6_savecontrol(struct in6pcb *in6p, str break; case IPPROTO_ROUTING: - if (!in6p->in6p_flags & IN6P_RTHDR) + if (!(in6p->in6p_flags & IN6P_RTHDR)) break; - *mp = sbcreatecontrol((void *)ip6e, elen, + *mp = sbcreatecontrol(ip6e, elen, IS2292(IPV6_2292RTHDR, IPV6_RTHDR), IPPROTO_IPV6); if (*mp) @@ -1344,10 +1285,7 @@ ip6_notify_pmtu(struct in6pcb *in6p, con if (mtu == NULL) return; -#ifdef DIAGNOSTIC - if (so == NULL) /* I believe this is impossible */ - panic("ip6_notify_pmtu: socket is NULL"); -#endif + KASSERT(so != NULL); memset(&mtuctl, 0, sizeof(mtuctl)); /* zero-clear for safety */ mtuctl.ip6m_mtu = *mtu; @@ -1355,14 +1293,14 @@ ip6_notify_pmtu(struct in6pcb *in6p, con if (sa6_recoverscope(&mtuctl.ip6m_addr)) return; - if ((m_mtu = sbcreatecontrol((void *)&mtuctl, sizeof(mtuctl), + if ((m_mtu = sbcreatecontrol(&mtuctl, sizeof(mtuctl), IPV6_PATHMTU, IPPROTO_IPV6)) == NULL) return; if (sbappendaddr(&so->so_rcv, (const struct sockaddr *)dst, NULL, m_mtu) == 0) { + soroverflow(so); m_freem(m_mtu); - /* XXX: should count statistics */ } else sorwakeup(so); @@ -1380,18 +1318,6 @@ ip6_pullexthdr(struct mbuf *m, size_t of 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); - } -#endif - m_copydata(m, off, sizeof(ip6e), (void *)&ip6e); if (nxt == IPPROTO_AH) elen = (ip6e.ip6e_len + 2) << 2; @@ -1421,50 +1347,44 @@ ip6_pullexthdr(struct mbuf *m, size_t of } /* - * Get pointer to the previous header followed by the header + * Get offset to the previous header followed by the header * currently processed. - * XXX: This function supposes that - * M includes all headers, - * the next header field and the header length field of each header - * are valid, and - * the sum of each header length equals to OFF. - * Because of these assumptions, this function must be called very - * carefully. Moreover, it will not be used in the near future when - * we develop `neater' mechanism to process extension headers. */ -u_int8_t * +int ip6_get_prevhdr(struct mbuf *m, int off) { struct ip6_hdr *ip6 = mtod(m, struct ip6_hdr *); - if (off == sizeof(struct ip6_hdr)) - return (&ip6->ip6_nxt); - else { - int len, nxt; - struct ip6_ext *ip6e = NULL; + if (off == sizeof(struct ip6_hdr)) { + return offsetof(struct ip6_hdr, ip6_nxt); + } else if (off < sizeof(struct ip6_hdr)) { + panic("%s: off < sizeof(struct ip6_hdr)", __func__); + } else { + int len, nlen, nxt; + struct ip6_ext ip6e; nxt = ip6->ip6_nxt; len = sizeof(struct ip6_hdr); + nlen = 0; while (len < off) { - ip6e = (struct ip6_ext *)(mtod(m, char *) + len); + m_copydata(m, len, sizeof(ip6e), &ip6e); switch (nxt) { case IPPROTO_FRAGMENT: - len += sizeof(struct ip6_frag); + nlen = sizeof(struct ip6_frag); break; case IPPROTO_AH: - len += (ip6e->ip6e_len + 2) << 2; + nlen = (ip6e.ip6e_len + 2) << 2; break; default: - len += (ip6e->ip6e_len + 1) << 3; + nlen = (ip6e.ip6e_len + 1) << 3; break; } - nxt = ip6e->ip6e_nxt; + len += nlen; + nxt = ip6e.ip6e_nxt; } - if (ip6e) - return (&ip6e->ip6e_nxt); - else - return NULL; + + return (len - nlen); } } @@ -1480,7 +1400,7 @@ ip6_nexthdr(struct mbuf *m, int off, int /* just in case */ if (m == NULL) - panic("ip6_nexthdr: m == NULL"); + panic("%s: m == NULL", __func__); if ((m->m_flags & M_PKTHDR) == 0 || m->m_pkthdr.len < off) return -1; @@ -1585,7 +1505,7 @@ ip6_addaux(struct mbuf *m) M_NOWAIT); if (mtag) { m_tag_prepend(m, mtag); - bzero(mtag + 1, sizeof(struct ip6aux)); + memset(mtag + 1, 0, sizeof(struct ip6aux)); } } return mtag; @@ -1610,62 +1530,11 @@ ip6_delaux(struct mbuf *m) 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 */ -u_char inet6ctlerrmap[PRC_NCMDS] = { +const u_char inet6ctlerrmap[PRC_NCMDS] = { 0, 0, 0, 0, 0, EMSGSIZE, EHOSTDOWN, EHOSTUNREACH, EHOSTUNREACH, EHOSTUNREACH, ECONNREFUSED, ECONNREFUSED, @@ -1674,19 +1543,19 @@ u_char inet6ctlerrmap[PRC_NCMDS] = { ENOPROTOOPT }; -SYSCTL_SETUP(sysctl_net_inet6_ip6_setup, "sysctl net.inet6.ip6 subtree setup") +extern int sysctl_net_inet6_addrctlpolicy(SYSCTLFN_ARGS); + +static int +sysctl_net_inet6_ip6_stats(SYSCTLFN_ARGS) +{ + + return (NETSTAT_SYSCTL(ip6stat_percpu, IP6_NSTATS)); +} + +static void +sysctl_net_inet6_ip6_setup(struct sysctllog **clog) { -#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, - CTLTYPE_NODE, "net", NULL, - NULL, 0, NULL, 0, - CTL_NET, CTL_EOL); sysctl_createv(clog, 0, NULL, NULL, CTLFLAG_PERMANENT, CTLTYPE_NODE, "inet6", @@ -1721,34 +1590,6 @@ SYSCTL_SETUP(sysctl_net_inet6_ip6_setup, 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", @@ -1757,27 +1598,27 @@ SYSCTL_SETUP(sysctl_net_inet6_ip6_setup, 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, + CTLTYPE_INT, "accept_rtadv", + SYSCTL_DESCR("Accept router advertisements"), + NULL, 0, &ip6_accept_rtadv, 0, CTL_NET, PF_INET6, IPPROTO_IPV6, - IPV6CTL_SOURCECHECK, CTL_EOL); + IPV6CTL_ACCEPT_RTADV, CTL_EOL); sysctl_createv(clog, 0, NULL, NULL, CTLFLAG_PERMANENT|CTLFLAG_READWRITE, - CTLTYPE_INT, "sourcecheck_logint", NULL, - NULL, 0, &?, 0, + CTLTYPE_INT, "rtadv_maxroutes", + SYSCTL_DESCR("Maximum number of routes accepted via router advertisements"), + NULL, 0, &ip6_rtadv_maxroutes, 0, CTL_NET, PF_INET6, IPPROTO_IPV6, - IPV6CTL_SOURCECHECK_LOGINT, CTL_EOL); -#endif + IPV6CTL_RTADV_MAXROUTES, CTL_EOL); 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, + CTLFLAG_PERMANENT, + CTLTYPE_INT, "rtadv_numroutes", + SYSCTL_DESCR("Current number of routes accepted via router advertisements"), + NULL, 0, &nd6_numroutes, 0, CTL_NET, PF_INET6, IPPROTO_IPV6, - IPV6CTL_ACCEPT_RTADV, CTL_EOL); + IPV6CTL_RTADV_NUMROUTES, CTL_EOL); sysctl_createv(clog, 0, NULL, NULL, CTLFLAG_PERMANENT|CTLFLAG_READWRITE, CTLTYPE_INT, "keepfaith", @@ -1788,7 +1629,7 @@ SYSCTL_SETUP(sysctl_net_inet6_ip6_setup, sysctl_createv(clog, 0, NULL, NULL, CTLFLAG_PERMANENT|CTLFLAG_READWRITE, CTLTYPE_INT, "log_interval", - SYSCTL_DESCR("Minumum interval between logging " + SYSCTL_DESCR("Minimum interval between logging " "unroutable packets"), NULL, 0, &ip6_log_interval, 0, CTL_NET, PF_INET6, IPPROTO_IPV6, @@ -1822,15 +1663,6 @@ SYSCTL_SETUP(sysctl_net_inet6_ip6_setup, 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", @@ -1897,6 +1729,23 @@ SYSCTL_SETUP(sysctl_net_inet6_ip6_setup, #endif /* IPNOPRIVPORTS */ sysctl_createv(clog, 0, NULL, NULL, CTLFLAG_PERMANENT|CTLFLAG_READWRITE, + CTLTYPE_INT, "auto_linklocal", + SYSCTL_DESCR("Default value of per-interface flag for " + "adding an IPv6 link-local address to " + "interfaces when attached"), + NULL, 0, &ip6_auto_linklocal, 0, + CTL_NET, PF_INET6, IPPROTO_IPV6, + IPV6CTL_AUTO_LINKLOCAL, CTL_EOL); + sysctl_createv(clog, 0, NULL, NULL, + CTLFLAG_PERMANENT|CTLFLAG_READONLY, + CTLTYPE_STRUCT, "addctlpolicy", + SYSCTL_DESCR("Return the current address control" + " policy"), + sysctl_net_inet6_addrctlpolicy, 0, NULL, 0, + CTL_NET, PF_INET6, IPPROTO_IPV6, + IPV6CTL_ADDRCTLPOLICY, CTL_EOL); + 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, @@ -1904,6 +1753,14 @@ SYSCTL_SETUP(sysctl_net_inet6_ip6_setup, CTL_CREATE, CTL_EOL); sysctl_createv(clog, 0, NULL, NULL, CTLFLAG_PERMANENT|CTLFLAG_READWRITE, + CTLTYPE_INT, "prefer_tempaddr", + SYSCTL_DESCR("Prefer temporary address as source " + "address"), + NULL, 0, &ip6_prefer_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, @@ -1927,7 +1784,7 @@ SYSCTL_SETUP(sysctl_net_inet6_ip6_setup, CTLFLAG_PERMANENT, CTLTYPE_STRUCT, "stats", SYSCTL_DESCR("IPv6 statistics"), - NULL, 0, &ip6stat, sizeof(ip6stat), + sysctl_net_inet6_ip6_stats, 0, NULL, 0, CTL_NET, PF_INET6, IPPROTO_IPV6, IPV6CTL_STATS, CTL_EOL); sysctl_createv(clog, 0, NULL, NULL, @@ -1944,20 +1801,70 @@ SYSCTL_SETUP(sysctl_net_inet6_ip6_setup, NULL, 0, &ip6_mcast_pmtu, 0, CTL_NET, PF_INET6, IPPROTO_IPV6, CTL_CREATE, CTL_EOL); -#ifdef GATEWAY + /* anonportalgo RFC6056 subtree */ + const struct sysctlnode *portalgo_node; + sysctl_createv(clog, 0, NULL, &portalgo_node, + CTLFLAG_PERMANENT, + CTLTYPE_NODE, "anonportalgo", + SYSCTL_DESCR("Anonymous port algorithm selection (RFC 6056)"), + NULL, 0, NULL, 0, + CTL_NET, PF_INET6, IPPROTO_IPV6, CTL_CREATE, CTL_EOL); + sysctl_createv(clog, 0, &portalgo_node, NULL, + CTLFLAG_PERMANENT, + CTLTYPE_STRING, "available", + SYSCTL_DESCR("available algorithms"), + sysctl_portalgo_available, 0, NULL, PORTALGO_MAXLEN, + CTL_CREATE, CTL_EOL); + sysctl_createv(clog, 0, &portalgo_node, NULL, + CTLFLAG_PERMANENT|CTLFLAG_READWRITE, + CTLTYPE_STRING, "selected", + SYSCTL_DESCR("selected algorithm"), + sysctl_portalgo_selected6, 0, NULL, PORTALGO_MAXLEN, + CTL_CREATE, CTL_EOL); + sysctl_createv(clog, 0, &portalgo_node, NULL, + CTLFLAG_PERMANENT|CTLFLAG_READWRITE, + CTLTYPE_STRUCT, "reserve", + SYSCTL_DESCR("bitmap of reserved ports"), + sysctl_portalgo_reserve6, 0, NULL, 0, + CTL_CREATE, CTL_EOL); 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 + CTLFLAG_PERMANENT|CTLFLAG_READWRITE, + CTLTYPE_INT, "neighborgcthresh", + SYSCTL_DESCR("Maximum number of entries in neighbor" + " cache"), + NULL, 1, &ip6_neighborgcthresh, 0, + CTL_NET, PF_INET6, IPPROTO_IPV6, + CTL_CREATE, CTL_EOL); + sysctl_createv(clog, 0, NULL, NULL, + CTLFLAG_PERMANENT|CTLFLAG_READWRITE, + CTLTYPE_INT, "maxifprefixes", + SYSCTL_DESCR("Maximum number of prefixes created by" + " route advertisement per interface"), + NULL, 1, &ip6_maxifprefixes, 0, + CTL_NET, PF_INET6, IPPROTO_IPV6, + CTL_CREATE, CTL_EOL); + sysctl_createv(clog, 0, NULL, NULL, + CTLFLAG_PERMANENT|CTLFLAG_READWRITE, + CTLTYPE_INT, "maxifdefrouters", + SYSCTL_DESCR("Maximum number of default routers created" + " by route advertisement per interface"), + NULL, 1, &ip6_maxifdefrouters, 0, + CTL_NET, PF_INET6, IPPROTO_IPV6, + CTL_CREATE, CTL_EOL); + sysctl_createv(clog, 0, NULL, NULL, + CTLFLAG_PERMANENT|CTLFLAG_READWRITE, + CTLTYPE_INT, "maxdynroutes", + SYSCTL_DESCR("Maximum number of routes created via" + " redirect"), + NULL, 1, &ip6_maxdynroutes, 0, + CTL_NET, PF_INET6, IPPROTO_IPV6, + CTL_CREATE, CTL_EOL); +} + +void +ip6_statinc(u_int stat) +{ + + KASSERT(stat < IP6_NSTATS); + IP6_STATINC(stat); }