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 1.64.2.3 retrieving revision 1.176 diff -u -p -r1.64.2.3 -r1.176 --- src/sys/netinet6/ip6_input.c 2004/09/21 13:37:34 1.64.2.3 +++ src/sys/netinet6/ip6_input.c 2017/03/01 08:54:12 1.176 @@ -1,4 +1,4 @@ -/* $NetBSD: ip6_input.c,v 1.64.2.3 2004/09/21 13:37:34 skrll Exp $ */ +/* $NetBSD: ip6_input.c,v 1.176 2017/03/01 08:54:12 ozaki-r Exp $ */ /* $KAME: ip6_input.c,v 1.188 2001/03/29 05:34:31 itojun Exp $ */ /* @@ -62,15 +62,19 @@ */ #include -__KERNEL_RCSID(0, "$NetBSD: ip6_input.c,v 1.64.2.3 2004/09/21 13:37:34 skrll Exp $"); +__KERNEL_RCSID(0, "$NetBSD: ip6_input.c,v 1.176 2017/03/01 08:54:12 ozaki-r Exp $"); +#ifdef _KERNEL_OPT +#include "opt_gateway.h" #include "opt_inet.h" +#include "opt_inet6.h" #include "opt_ipsec.h" -#include "opt_pfil_hooks.h" +#include "opt_compat_netbsd.h" +#include "opt_net_mpsafe.h" +#endif #include #include -#include #include #include #include @@ -82,205 +86,243 @@ __KERNEL_RCSID(0, "$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 #include #include #ifdef IPSEC -#include +#include +#include +#include +#endif /* IPSEC */ + +#ifdef COMPAT_50 +#include +#include #endif #include -/* we need it for NLOOP. */ -#include "loop.h" #include "faith.h" -#include "gif.h" -#include "bpfilter.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; +pktqueue_t *ip6_pktq __read_mostly; -extern struct ifnet loif[NLOOP]; int ip6_forward_srcrt; /* XXX */ int ip6_sourcecheck; /* XXX */ int ip6_sourcecheck_interval; /* XXX */ -#ifdef PFIL_HOOKS -struct pfil_head inet6_pfil_hook; -#endif +pfil_head_t *inet6_pfil_hook; -struct ip6stat ip6stat; +percpu_t *ip6stat_percpu; -static void ip6_init2 __P((void *)); +percpu_t *ip6_forward_rt_percpu __cacheline_aligned; -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 void ip6_init2(void); +static void ip6intr(void *); +static struct m_tag *ip6_setdstifaddr(struct mbuf *, const struct in6_ifaddr *); + +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) { - struct ip6protosw *pr; + const struct ip6protosw *pr; int i; - pr = (struct ip6protosw *)pffindproto(PF_INET6, IPPROTO_RAW, SOCK_RAW); + 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"); 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; + + ip6_pktq = pktq_create(IFQ_MAXLEN, ip6intr, NULL); + KASSERT(ip6_pktq != NULL); + + scope6_init(); + addrsel_policy_init(); nd6_init(); frag6_init(); + ip6_desync_factor = cprng_fast32() % MAX_TEMP_DESYNC_FACTOR; - ip6_init2((void *)0); - -#ifdef PFIL_HOOKS + ip6_init2(); +#ifdef GATEWAY + ip6flow_init(ip6_hashsize); +#endif /* 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)); + if (ip6_forward_rt_percpu == NULL) + panic("failed to alllocate ip6_forward_rt_percpu"); } static void -ip6_init2(dummy) - void *dummy; +ip6_init2(void) { - /* nd6_timer_init */ - callout_init(&nd6_timer_ch); - callout_reset(&nd6_timer_ch, hz, nd6_timer, NULL); + /* timer for regeneranation 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, + in6_tmpaddrtimer, NULL); } /* * 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; - ip6_input(m); +#ifndef NET_MPSAFE + mutex_enter(softnet_lock); +#endif + 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); + 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, rcvif); + m_put_rcvif_psref(rcvif, &psref); } +#ifndef NET_MPSAFE + mutex_exit(softnet_lock); +#endif } -extern struct route_in6 ip6_forward_rt; - void -ip6_input(m) - struct mbuf *m; +ip6_input(struct mbuf *m, struct ifnet *rcvif) { 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; + struct rtentry *rt = NULL; + union { + struct sockaddr dst; + struct sockaddr_in6 dst6; + } u; + struct route *ro; -#ifdef IPSEC /* - * should the inner packet be considered authentic? - * see comment in ah4_input(). + * make sure we don't have onion peering information into m_tag. */ - m->m_flags &= ~M_AUTHIPHDR; - m->m_flags &= ~M_AUTHIPDGM; -#endif + ip6_delaux(m); /* * mbuf statistics */ 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 (sizeof(ip6stat.ip6s_m2m)/sizeof(ip6stat.ip6s_m2m[0])) +#define M2MMAX 32 if (m->m_next) { - if (m->m_flags & M_LOOP) { - ip6stat.ip6s_m2m[loif[0].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 * 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) { - struct ifnet *inifp = m->m_pkthdr.rcvif; + if (IP6_HDR_ALIGNED_P(mtod(m, void *)) == 0) { 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); + 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; } } @@ -288,12 +330,17 @@ ip6_input(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; } -#ifdef PFIL_HOOKS + /* + * 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 @@ -305,8 +352,8 @@ ip6_input(m) * let ipfilter look at packet on the wire, * not the decapsulated packet. */ -#ifdef IPSEC - if (!ipsec_getnhist(m)) +#if defined(IPSEC) + if (!ipsec_used || !ipsec_indone(m)) #else if (1) #endif @@ -314,22 +361,25 @@ ip6_input(m) struct in6_addr odst; odst = ip6->ip6_dst; - if (pfil_run_hooks(&inet6_pfil_hook, &m, m->m_pkthdr.rcvif, - PFIL_IN) != 0) + 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); } -#endif /* PFIL_HOOKS */ - ip6stat.ip6s_nxthist[ip6->ip6_nxt]++; + IP6_STATINC(IP6_STAT_NXTHIST + ip6->ip6_nxt); #ifdef ALTQ - if (altq_input != NULL && (*altq_input)(m, AF_INET6) == 0) { - /* packet is dropped by traffic conditioner */ - return; + 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 @@ -341,8 +391,8 @@ ip6_input(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; } /* @@ -359,8 +409,8 @@ ip6_input(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 @@ -373,121 +423,75 @@ ip6_input(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 - 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)) { + IP6_STATINC(IP6_STAT_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, 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; + 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 */ - 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)); - 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); - } + rt = rtcache_lookup2(ro, &u.dst, 1, &hit); + if (hit) + IP6_STATINC(IP6_STAT_FORWARD_CACHEHIT); + else + IP6_STATINC(IP6_STAT_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 @@ -498,41 +502,38 @@ 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 && #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; /* * 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 | IN6_IFF_DETACHED))) { /* 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; } } @@ -541,11 +542,11 @@ 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; } } @@ -559,11 +560,7 @@ 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) { - if (ifa->ifa_addr == NULL) - continue; /* just for safety */ + IFADDR_READER_FOREACH(ifa, rcvif) { if (ifa->ifa_addr->sa_family != AF_INET6) continue; if (IN6_ARE_ADDR_EQUAL(IFA_IN6(ifa), &ip6->ip6_dst)) { @@ -580,13 +577,35 @@ ip6_input(m) * 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: /* + * 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, + * as we could not provide enough information + * to the upper layers. + */ + } + pserialize_read_exit(s); + } + + /* * 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. @@ -597,8 +616,10 @@ ip6_input(m) if (ip6_hopopts_input(&plen, &rtalert, &m, &off)) { #if 0 /*touches NULL pointer*/ - in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_discard); + in6_ifstat_inc(rcvif, ifs6_in_discard); #endif + rtcache_unref(rt, ro); + percpu_putref(ip6_forward_rt_percpu); return; /* m have already been freed */ } @@ -613,21 +634,25 @@ 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++; - 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, - (caddr_t)&ip6->ip6_plen - (caddr_t)ip6); + (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)); @@ -649,9 +674,9 @@ ip6_input(m) * 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) { @@ -673,16 +698,25 @@ ip6_input(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; } @@ -700,9 +734,9 @@ ip6_input(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; } /* @@ -711,18 +745,28 @@ ip6_input(m) #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; while (nxt != IPPROTO_DONE) { if (ip6_hdrnestlimit && (++nest > ip6_hdrnestlimit)) { - ip6stat.ip6s_toomanyhdr++; + IP6_STATINC(IP6_STAT_TOOMANYHDR); + in6_ifstat_inc(rcvif, ifs6_in_hdrerr); goto bad; } @@ -731,41 +775,98 @@ ip6_input(m) * 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(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; + if (ipsec_used) { + /* + * 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) { + int error; + + SOFTNET_LOCK(); + error = ipsec6_input(m); + SOFTNET_UNLOCK(); + if (error) + goto bad; + } } -#endif +#endif /* IPSEC */ + SOFTNET_LOCK(); nxt = (*inet6sw[ip6_protox[nxt]].pr_input)(&m, &off, nxt); + SOFTNET_UNLOCK(); } return; + + bad_unref: + rtcache_unref(rt, ro); + percpu_putref(ip6_forward_rt_percpu); bad: m_freem(m); + return; +} + +/* + * 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; + struct ip6aux *ip6a; + + mtag = ip6_addaux(m); + if (mtag == NULL) + return NULL; + + 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; + } + + ip6a->ip6a_src = ia->ia_addr.sin6_addr; + ip6a->ip6a_flags = ia->ia6_flags; + return mtag; +} + +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; +int +ip6_hopopts_input(u_int32_t *plenp, u_int32_t *rtalertp, + struct mbuf **mp, int *offp) { struct mbuf *m = *mp; int off = *offp, hbhlen; @@ -775,14 +876,14 @@ ip6_hopopts_input(plenp, rtalertp, mp, o 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); return -1; } hbhlen = (hbh->ip6h_len + 1) << 3; IP6_EXTHDR_GET(hbh, struct ip6_hbh *, m, sizeof(struct ip6_hdr), hbhlen); if (hbh == NULL) { - ip6stat.ip6s_tooshort++; + IP6_STATINC(IP6_STAT_TOOSHORT); return -1; } KASSERT(IP6_HDR_ALIGNED_P(hbh)); @@ -808,13 +909,9 @@ ip6_hopopts_input(plenp, rtalertp, mp, o * (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) - struct mbuf *m; - u_int8_t *opthead; - int hbhlen; - u_int32_t *rtalertp; - u_int32_t *plenp; +static int +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; @@ -830,7 +927,7 @@ ip6_process_hopopts(m, opthead, hbhlen, break; case IP6OPT_PADN: if (hbhlen < IP6OPT_MINLEN) { - ip6stat.ip6s_toosmall++; + IP6_STATINC(IP6_STAT_TOOSMALL); goto bad; } optlen = *(opt + 1) + 2; @@ -838,7 +935,7 @@ ip6_process_hopopts(m, opthead, hbhlen, 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) { @@ -849,13 +946,13 @@ ip6_process_hopopts(m, opthead, hbhlen, return (-1); } optlen = IP6OPT_RTALERT_LEN; - bcopy((caddr_t)(opt + 2), (caddr_t)&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) { @@ -873,7 +970,7 @@ ip6_process_hopopts(m, opthead, hbhlen, */ 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); @@ -884,7 +981,7 @@ ip6_process_hopopts(m, opthead, hbhlen, * We may see jumbolen in unaligned location, so * we'd need to perform bcopy(). */ - bcopy(opt + 2, &jumboplen, sizeof(jumboplen)); + memcpy(&jumboplen, opt + 2, sizeof(jumboplen)); jumboplen = (u_int32_t)htonl(jumboplen); #if 1 @@ -897,7 +994,7 @@ ip6_process_hopopts(m, opthead, hbhlen, * 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); @@ -909,7 +1006,7 @@ ip6_process_hopopts(m, opthead, hbhlen, * 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); @@ -920,7 +1017,7 @@ ip6_process_hopopts(m, opthead, hbhlen, break; default: /* unknown option */ if (hbhlen < IP6OPT_MINLEN) { - ip6stat.ip6s_toosmall++; + IP6_STATINC(IP6_STAT_TOOSMALL); goto bad; } optlen = ip6_unknown_opt(opt, m, @@ -946,10 +1043,7 @@ 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; @@ -960,11 +1054,11 @@ ip6_unknown_opt(optp, m, off) 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))) @@ -991,69 +1085,85 @@ 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) { +#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) { + if (in6p->in6p_socket->so_options & SO_TIMESTAMP +#ifdef SO_OTIMESTAMP + || in6p->in6p_socket->so_options & SO_OTIMESTAMP +#endif + ) { struct timeval tv; microtime(&tv); - *mp = sbcreatecontrol((caddr_t) &tv, sizeof(tv), - SCM_TIMESTAMP, SOL_SOCKET); - if (*mp) - mp = &(*mp)->m_next; - } +#ifdef SO_OTIMESTAMP + if (in6p->in6p_socket->so_options & SO_OTIMESTAMP) { + struct timeval50 tv50; + timeval_to_timeval50(&tv, &tv50); + *mp = sbcreatecontrol((void *) &tv50, sizeof(tv50), + SCM_OTIMESTAMP, SOL_SOCKET); + } else #endif - if (in6p->in6p_flags & IN6P_RECVDSTADDR) { - *mp = sbcreatecontrol((caddr_t) &ip6->ip6_dst, - sizeof(struct in6_addr), IPV6_RECVDSTADDR, IPPROTO_IPV6); + *mp = sbcreatecontrol((void *) &tv, sizeof(tv), + SCM_TIMESTAMP, SOL_SOCKET); 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); + + memcpy(&pi6.ipi6_addr, &ip6->ip6_dst, sizeof(struct in6_addr)); + in6_clearscope(&pi6.ipi6_addr); /* XXX */ + pi6.ipi6_ifindex = m->m_pkthdr.rcvif_index; + *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. 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. + * See also RFC3542 section 8 (or RFC2292 section 6). */ if ((in6p->in6p_flags & IN6P_HOPOPTS) != 0) { /* @@ -1063,23 +1173,23 @@ 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++; + 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; } @@ -1087,10 +1197,11 @@ 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); @@ -1099,8 +1210,8 @@ 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 @@ -1109,7 +1220,7 @@ 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; @@ -1130,7 +1241,7 @@ ip6_savecontrol(in6p, mp, ip6, m) ext = ip6_pullexthdr(m, off, nxt); if (ext == NULL) { - ip6stat.ip6s_tooshort++; + IP6_STATINC(IP6_STAT_TOOSHORT); return; } ip6e = mtod(ext, struct ip6_ext *); @@ -1140,28 +1251,30 @@ ip6_savecontrol(in6p, mp, ip6, m) 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((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; case IPPROTO_ROUTING: - if (!in6p->in6p_flags & IN6P_RTHDR) + 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; @@ -1193,16 +1306,53 @@ 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; @@ -1220,7 +1370,7 @@ 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 @@ -1243,7 +1393,7 @@ 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; } @@ -1261,9 +1411,7 @@ 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 *); @@ -1276,7 +1424,7 @@ 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: @@ -1302,11 +1450,7 @@ 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; @@ -1320,9 +1464,12 @@ 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); @@ -1335,7 +1482,7 @@ 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) @@ -1346,7 +1493,7 @@ 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; @@ -1359,7 +1506,7 @@ 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; @@ -1382,11 +1529,7 @@ 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; @@ -1395,7 +1538,7 @@ 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; @@ -1409,11 +1552,47 @@ 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); + memset(mtag + 1, 0, 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); +} + /* * 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, @@ -1422,14 +1601,24 @@ 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) { - sysctl_createv(clog, 0, NULL, NULL, - CTLFLAG_PERMANENT, - CTLTYPE_NODE, "net", NULL, - NULL, 0, NULL, 0, - CTL_NET, CTL_EOL); + 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, "inet6", @@ -1481,12 +1670,6 @@ 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, @@ -1529,6 +1712,20 @@ SYSCTL_SETUP(sysctl_net_inet6_ip6_setup, IPV6CTL_ACCEPT_RTADV, CTL_EOL); sysctl_createv(clog, 0, NULL, NULL, CTLFLAG_PERMANENT|CTLFLAG_READWRITE, + 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_RTADV_MAXROUTES, CTL_EOL); + sysctl_createv(clog, 0, NULL, NULL, + 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_RTADV_NUMROUTES, 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, @@ -1571,20 +1768,11 @@ 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", 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, @@ -1646,9 +1834,142 @@ 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, + CTL_NET, PF_INET6, IPPROTO_IPV6, + 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, + 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"), + sysctl_net_inet6_ip6_stats, 0, NULL, 0, + 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); + /* 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, "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); }