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 1.82.6.1 retrieving revision 1.149.2.1.6.2 diff -u -p -r1.82.6.1 -r1.149.2.1.6.2 --- src/sys/netinet6/ip6_input.c 2006/03/13 09:07:39 1.82.6.1 +++ src/sys/netinet6/ip6_input.c 2018/02/25 23:17:37 1.149.2.1.6.2 @@ -1,4 +1,4 @@ -/* $NetBSD: ip6_input.c,v 1.82.6.1 2006/03/13 09:07:39 yamt Exp $ */ +/* $NetBSD: ip6_input.c,v 1.149.2.1.6.2 2018/02/25 23:17:37 snj Exp $ */ /* $KAME: ip6_input.c,v 1.188 2001/03/29 05:34:31 itojun Exp $ */ /* @@ -62,11 +62,13 @@ */ #include -__KERNEL_RCSID(0, "$NetBSD: ip6_input.c,v 1.82.6.1 2006/03/13 09:07:39 yamt Exp $"); +__KERNEL_RCSID(0, "$NetBSD: ip6_input.c,v 1.149.2.1.6.2 2018/02/25 23:17:37 snj Exp $"); +#include "opt_gateway.h" #include "opt_inet.h" +#include "opt_inet6.h" #include "opt_ipsec.h" -#include "opt_pfil_hooks.h" +#include "opt_compat_netbsd.h" #include #include @@ -82,25 +84,27 @@ __KERNEL_RCSID(0, "$NetBSD: ip6_input.c, #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 @@ -108,7 +112,14 @@ __KERNEL_RCSID(0, "$NetBSD: ip6_input.c, #include #ifdef IPSEC -#include +#include +#include +#include +#endif /* IPSEC */ + +#ifdef COMPAT_50 +#include +#include #endif #include @@ -125,38 +136,39 @@ __KERNEL_RCSID(0, "$NetBSD: ip6_input.c, 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 callout in6_tmpaddrtimer_ch; +extern callout_t in6_tmpaddrtimer_ch; int ip6_forward_srcrt; /* XXX */ int ip6_sourcecheck; /* XXX */ int ip6_sourcecheck_interval; /* XXX */ -#ifdef PFIL_HOOKS -struct pfil_head inet6_pfil_hook; -#endif - -struct ip6stat ip6stat; +pfil_head_t *inet6_pfil_hook; -static void ip6_init2 __P((void *)); -static struct m_tag *ip6_setdstifaddr __P((struct mbuf *, struct in6_ifaddr *)); +percpu_t *ip6stat_percpu; -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 **); /* * 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; + sysctl_net_inet6_ip6_setup(NULL); pr = (const struct ip6protosw *)pffindproto(PF_INET6, IPPROTO_RAW, SOCK_RAW); if (pr == 0) panic("ip6_init"); @@ -167,37 +179,37 @@ 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_init2((void *)0); + ip6_desync_factor = cprng_fast32() % MAX_TEMP_DESYNC_FACTOR; -#ifdef PFIL_HOOKS + ip6_init2(NULL); +#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); } static void -ip6_init2(dummy) - void *dummy; +ip6_init2(void *dummy) { /* nd6_timer_init */ - callout_init(&nd6_timer_ch); + callout_init(&nd6_timer_ch, CALLOUT_MPSAFE); callout_reset(&nd6_timer_ch, hz, nd6_timer, NULL); /* timer for regeneranation of temporary addresses randomize ID */ - callout_init(&in6_tmpaddrtimer_ch); + callout_init(&in6_tmpaddrtimer_ch, CALLOUT_MPSAFE); callout_reset(&in6_tmpaddrtimer_ch, (ip6_temp_preferred_lifetime - ip6_desync_factor - ip6_temp_regen_advance) * hz, @@ -207,49 +219,44 @@ ip6_init2(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)) { + mutex_enter(softnet_lock); + while ((m = pktq_dequeue(ip6_pktq)) != NULL) { + const ifnet_t *ifp = m->m_pkthdr.rcvif; + + /* + * Drop the packet if IPv6 is disabled on the interface. + */ + if ((ND_IFINFO(ifp)->flags & ND6_IFF_IFDISABLED)) { m_freem(m); - return; + continue; } ip6_input(m); } + mutex_exit(softnet_lock); } -extern struct route_in6 ip6_forward_rt; +extern struct route ip6_forward_rt; void -ip6_input(m) - struct mbuf *m; +ip6_input(struct mbuf *m) { struct ip6_hdr *ip6; - int off = sizeof(struct ip6_hdr), nest; + int hit, off = sizeof(struct ip6_hdr), nest; u_int32_t plen; u_int32_t rtalert = ~0; - int nxt, ours = 0; + int nxt, ours = 0, rh_present = 0, frg_present; struct ifnet *deliverifp = NULL; int srcrt = 0; - -#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 + const struct rtentry *rt; + union { + struct sockaddr dst; + struct sockaddr_in6 dst6; + } u; /* * make sure we don't have onion peering information into m_tag. @@ -261,25 +268,26 @@ ip6_input(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 (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[lo0ifp->if_index]++; /* XXX */ - } else if (m->m_pkthdr.rcvif->if_index < M2MMAX) - ip6stat.ip6s_m2m[m->m_pkthdr.rcvif->if_index]++; - else - ip6stat.ip6s_m2m[0]++; + /*XXX*/ IP6_STATINC(IP6_STAT_M2M + lo0ifp->if_index); + } else if (m->m_pkthdr.rcvif->if_index < M2MMAX) { + IP6_STATINC(IP6_STAT_M2M + + m->m_pkthdr.rcvif->if_index); + } else + 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++; + IP6_STATINC(IP6_STAT_TOTAL); /* * If the IPv6 header is not aligned, slurp it up into a new @@ -287,19 +295,19 @@ ip6_input(m) * it. Otherwise, if it is aligned, make sure the entire base * IPv6 header is in the first mbuf of the chain. */ - if (IP6_HDR_ALIGNED_P(mtod(m, caddr_t)) == 0) { + if (IP6_HDR_ALIGNED_P(mtod(m, void *)) == 0) { struct ifnet *inifp = m->m_pkthdr.rcvif; if ((m = m_copyup(m, sizeof(struct ip6_hdr), (max_linkhdr + 3) & ~3)) == NULL) { /* XXXJRT new stat, please */ - ip6stat.ip6s_toosmall++; + IP6_STATINC(IP6_STAT_TOOSMALL); in6_ifstat_inc(inifp, ifs6_in_hdrerr); return; } } else if (__predict_false(m->m_len < sizeof(struct ip6_hdr))) { struct ifnet *inifp = m->m_pkthdr.rcvif; if ((m = m_pullup(m, sizeof(struct ip6_hdr))) == NULL) { - ip6stat.ip6s_toosmall++; + IP6_STATINC(IP6_STAT_TOOSMALL); in6_ifstat_inc(inifp, ifs6_in_hdrerr); return; } @@ -308,12 +316,17 @@ ip6_input(m) ip6 = mtod(m, struct ip6_hdr *); if ((ip6->ip6_vfc & IPV6_VERSION_MASK) != IPV6_VERSION) { - ip6stat.ip6s_badvers++; + IP6_STATINC(IP6_STAT_BADVERS); in6_ifstat_inc(m->m_pkthdr.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 @@ -325,8 +338,8 @@ 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 @@ -334,7 +347,7 @@ ip6_input(m) struct in6_addr odst; odst = ip6->ip6_dst; - if (pfil_run_hooks(&inet6_pfil_hook, &m, m->m_pkthdr.rcvif, + if (pfil_run_hooks(inet6_pfil_hook, &m, m->m_pkthdr.rcvif, PFIL_IN) != 0) return; if (m == NULL) @@ -342,9 +355,8 @@ ip6_input(m) 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) { @@ -361,7 +373,7 @@ ip6_input(m) /* * XXX: "badscope" is not very suitable for a multicast source. */ - ip6stat.ip6s_badscope++; + IP6_STATINC(IP6_STAT_BADSCOPE); in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_addrerr); goto bad; } @@ -379,7 +391,7 @@ ip6_input(m) */ if (IN6_IS_ADDR_V4MAPPED(&ip6->ip6_src) || IN6_IS_ADDR_V4MAPPED(&ip6->ip6_dst)) { - ip6stat.ip6s_badscope++; + IP6_STATINC(IP6_STAT_BADSCOPE); in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_addrerr); goto bad; } @@ -393,7 +405,7 @@ ip6_input(m) */ if (IN6_IS_ADDR_V4COMPAT(&ip6->ip6_src) || IN6_IS_ADDR_V4COMPAT(&ip6->ip6_dst)) { - ip6stat.ip6s_badscope++; + IP6_STATINC(IP6_STAT_BADSCOPE); in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_addrerr); goto bad; } @@ -409,13 +421,17 @@ ip6_input(m) * dst are the loopback address and the receiving interface * is not loopback. */ + if (__predict_false( + m_makewritable(&m, 0, sizeof(struct ip6_hdr), M_DONTWAIT))) + goto bad; + ip6 = mtod(m, struct ip6_hdr *); if (in6_clearscope(&ip6->ip6_src) || in6_clearscope(&ip6->ip6_dst)) { - ip6stat.ip6s_badscope++; /* XXX */ + 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++; + IP6_STATINC(IP6_STAT_BADSCOPE); goto bad; } @@ -434,8 +450,10 @@ ip6_input(m) if (in6m) ours = 1; else if (!ip6_mrouter) { - ip6stat.ip6s_notmember++; - ip6stat.ip6s_cantforward++; + uint64_t *ip6s = IP6_STAT_GETREF(); + ip6s[IP6_STAT_NOTMEMBER]++; + ip6s[IP6_STAT_CANTFORWARD]++; + IP6_STAT_PUTREF(); in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_discard); goto bad; } @@ -443,34 +461,18 @@ ip6_input(m) goto hbhcheck; } + sockaddr_in6_init(&u.dst6, &ip6->ip6_dst, 0, 0, 0); + /* * Unicast check */ - if (ip6_forward_rt.ro_rt != NULL && - (ip6_forward_rt.ro_rt->rt_flags & RTF_UP) != 0 && - IN6_ARE_ADDR_EQUAL(&ip6->ip6_dst, - &((struct sockaddr_in6 *)(&ip6_forward_rt.ro_dst))->sin6_addr)) - 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(&ip6_forward_rt, &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 @@ -481,22 +483,19 @@ ip6_input(m) * But we think it's even useful in some situations, e.g. when using * a special daemon which wants to intercept the packet. */ - if (ip6_forward_rt.ro_rt && - (ip6_forward_rt.ro_rt->rt_flags & - (RTF_HOST|RTF_GATEWAY)) == RTF_HOST && - !(ip6_forward_rt.ro_rt->rt_flags & RTF_CLONED) && + if (rt != NULL && + (rt->rt_flags & (RTF_HOST|RTF_GATEWAY)) == RTF_HOST && + !(rt->rt_flags & RTF_CLONED) && #if 0 /* * The check below is redundant since the comparison of * the destination and the key of the rtentry has * already done through looking up the routing table. */ - IN6_ARE_ADDR_EQUAL(&ip6->ip6_dst, - &rt6_key(ip6_forward_rt.ro_rt)->sin6_addr) && + IN6_ARE_ADDR_EQUAL(&ip6->ip6_dst, &rt6_getkey(rt)->sin6_addr) && #endif - ip6_forward_rt.ro_rt->rt_ifp->if_type == IFT_LOOP) { - struct in6_ifaddr *ia6 = - (struct in6_ifaddr *)ip6_forward_rt.ro_rt->rt_ifa; + rt->rt_ifp->if_type == IFT_LOOP) { + struct in6_ifaddr *ia6 = (struct in6_ifaddr *)rt->rt_ifa; if (ia6->ia6_flags & IN6_IFF_ANYCAST) m->m_flags |= M_ANYCAST6; /* @@ -524,11 +523,11 @@ 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; } } @@ -542,9 +541,7 @@ 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) { + IFADDR_FOREACH(ifa, m->m_pkthdr.rcvif) { if (ifa->ifa_addr == NULL) continue; /* just for safety */ if (ifa->ifa_addr->sa_family != AF_INET6) @@ -563,7 +560,7 @@ ip6_input(m) * and we're not a router. */ if (!ip6_forwarding) { - ip6stat.ip6s_cantforward++; + IP6_STATINC(IP6_STAT_CANTFORWARD); in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_discard); goto bad; } @@ -575,18 +572,16 @@ ip6_input(m) * as our interface address (e.g. multicast addresses, addresses * within FAITH prefixes and such). */ - if (deliverifp && !ip6_getdstifaddr(m)) { + if (deliverifp && ip6_getdstifaddr(m) == NULL) { struct in6_ifaddr *ia6; ia6 = in6_ifawithifp(deliverifp, &ip6->ip6_dst); - if (ia6) { - if (!ip6_setdstifaddr(m, ia6)) { - /* - * XXX maybe we should drop the packet here, - * as we could not provide enough information - * to the upper layers. - */ - } + 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. + */ } } @@ -620,18 +615,18 @@ ip6_input(m) * contained, ip6_hopopts_input() must set a valid * (non-zero) payload length to the variable plen. */ - ip6stat.ip6s_badoptions++; + IP6_STATINC(IP6_STAT_BADOPTIONS); in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_discard); in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_hdrerr); icmp6_error(m, ICMP6_PARAM_PROB, ICMP6_PARAMPROB_HEADER, - (caddr_t)&ip6->ip6_plen - (caddr_t)ip6); + (char *)&ip6->ip6_plen - (char *)ip6); return; } 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; } KASSERT(IP6_HDR_ALIGNED_P(hbh)); @@ -653,7 +648,7 @@ ip6_input(m) * Drop packet if shorter than we expect. */ if (m->m_pkthdr.len - sizeof(struct ip6_hdr) < plen) { - ip6stat.ip6s_tooshort++; + IP6_STATINC(IP6_STAT_TOOSHORT); in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_truncated); goto bad; } @@ -678,7 +673,7 @@ ip6_input(m) * must be discarded, else it may be accepted below. */ if (ip6_mrouter && ip6_mforward(ip6, m->m_pkthdr.rcvif, m)) { - ip6stat.ip6s_cantforward++; + IP6_STATINC(IP6_STAT_CANTFORWARD); m_freem(m); return; } @@ -704,7 +699,7 @@ ip6_input(m) */ if (IN6_IS_ADDR_V4MAPPED(&ip6->ip6_src) || IN6_IS_ADDR_V4MAPPED(&ip6->ip6_dst)) { - ip6stat.ip6s_badscope++; + IP6_STATINC(IP6_STAT_BADSCOPE); in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_addrerr); goto bad; } @@ -720,13 +715,16 @@ ip6_input(m) ia6->ia_ifa.ifa_data.ifad_inbytes += m->m_pkthdr.len; } #endif - ip6stat.ip6s_delivered++; + IP6_STATINC(IP6_STAT_DELIVERED); in6_ifstat_inc(deliverifp, ifs6_in_deliver); nest = 0; + rh_present = 0; + frg_present = 0; while (nxt != IPPROTO_DONE) { if (ip6_hdrnestlimit && (++nest > ip6_hdrnestlimit)) { - ip6stat.ip6s_toomanyhdr++; + IP6_STATINC(IP6_STAT_TOOMANYHDR); + in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_hdrerr); goto bad; } @@ -735,23 +733,79 @@ ip6_input(m) * more sanity checks in header chain processing. */ if (m->m_pkthdr.len < off) { - ip6stat.ip6s_tooshort++; + IP6_STATINC(IP6_STAT_TOOSHORT); in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_truncated); goto bad; } + if (nxt == IPPROTO_ROUTING) { + if (rh_present++) { + in6_ifstat_inc(m->m_pkthdr.rcvif, + ifs6_in_hdrerr); + IP6_STATINC(IP6_STAT_BADOPTIONS); + goto bad; + } + } else if (nxt == IPPROTO_FRAGMENT) { + if (frg_present++) { + in6_ifstat_inc(m->m_pkthdr.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) { + struct m_tag *mtag; + struct tdb_ident *tdbi; + struct secpolicy *sp; + int s, error; + + /* + * enforce IPsec policy checking if we are seeing last + * header. note that we do not visit this with + * protocols with pcb layer code - like udp/tcp/raw ip. + */ + if ((inet6sw[ip_protox[nxt]].pr_flags + & PR_LASTHDR) != 0) { + /* + * Check if the packet has already had IPsec + * processing done. If so, then just pass it + * along. This tag gets set during AH, ESP, + * etc. input handling, before the packet is + * returned to the ip input queue for delivery. + */ + mtag = m_tag_find(m, PACKET_TAG_IPSEC_IN_DONE, + NULL); + s = splsoftnet(); + if (mtag != NULL) { + tdbi = (struct tdb_ident *)(mtag + 1); + sp = ipsec_getpolicy(tdbi, + IPSEC_DIR_INBOUND); + } else { + sp = ipsec_getpolicybyaddr(m, + IPSEC_DIR_INBOUND, IP_FORWARDING, + &error); + } + if (sp != NULL) { + /* + * Check security policy against packet + * attributes. + */ + error = ipsec_in_reject(sp, m); + KEY_FREESP(&sp); + } else { + /* XXX error stat??? */ + error = EINVAL; + DPRINTF(("ip6_input: no SP, packet" + " discarded\n"));/*XXX*/ + } + splx(s); + if (error) + goto bad; + } } -#endif +#endif /* IPSEC */ nxt = (*inet6sw[ip6_protox[nxt]].pr_input)(&m, &off, nxt); } @@ -764,27 +818,34 @@ ip6_input(m) * set/grab in6_ifaddr correspond to IPv6 destination address. */ static struct m_tag * -ip6_setdstifaddr(m, ia6) - struct mbuf *m; - struct in6_ifaddr *ia6; +ip6_setdstifaddr(struct mbuf *m, const struct in6_ifaddr *ia) { struct m_tag *mtag; + struct ip6aux *ip6a; mtag = ip6_addaux(m); - if (mtag) - ((struct ip6aux *)(mtag + 1))->ip6a_dstia6 = ia6; - return mtag; /* NULL if failed to set */ + 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; } -struct in6_ifaddr * -ip6_getdstifaddr(m) - struct mbuf *m; +const struct ip6aux * +ip6_getdstifaddr(struct mbuf *m) { struct m_tag *mtag; mtag = ip6_findaux(m); - if (mtag) - return ((struct ip6aux *)(mtag + 1))->ip6a_dstia6; + if (mtag != NULL) + return (struct ip6aux *)(mtag + 1); else return NULL; } @@ -792,13 +853,12 @@ ip6_getdstifaddr(m) /* * 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; @@ -808,14 +868,14 @@ 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)); @@ -841,13 +901,9 @@ 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; @@ -863,7 +919,7 @@ 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; @@ -871,7 +927,7 @@ 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) { @@ -882,13 +938,13 @@ 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) { @@ -906,7 +962,7 @@ 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); @@ -917,7 +973,7 @@ 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 @@ -930,7 +986,7 @@ 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); @@ -942,7 +998,7 @@ 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); @@ -953,7 +1009,7 @@ 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, @@ -979,10 +1035,7 @@ 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; @@ -993,11 +1046,11 @@ 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))) @@ -1024,68 +1077,86 @@ 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)); + + memcpy(&pi6.ipi6_addr, &ip6->ip6_dst, sizeof(struct in6_addr)); in6_clearscope(&pi6.ipi6_addr); /* XXX */ - 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); + pi6.ipi6_ifindex = m->m_pkthdr.rcvif ? + m->m_pkthdr.rcvif->if_index : 0; + *mp = sbcreatecontrol((void *) &pi6, + sizeof(struct in6_pktinfo), + IS2292(IPV6_2292PKTINFO, IPV6_PKTINFO), IPPROTO_IPV6); if (*mp) mp = &(*mp)->m_next; } + if (in6p->in6p_flags & IN6P_HOPLIMIT) { int hlim = ip6->ip6_hlim & 0xff; - *mp = sbcreatecontrol((caddr_t) &hlim, sizeof(int), - IPV6_HOPLIMIT, IPPROTO_IPV6); + + *mp = sbcreatecontrol((void *) &hlim, sizeof(int), + IS2292(IPV6_2292HOPLIMIT, IPV6_HOPLIMIT), IPPROTO_IPV6); + if (*mp) + mp = &(*mp)->m_next; + } + + if ((in6p->in6p_flags & IN6P_TCLASS) != 0) { + u_int32_t flowinfo; + int tclass; + + flowinfo = (u_int32_t)ntohl(ip6->ip6_flow & IPV6_FLOWINFO_MASK); + flowinfo >>= 20; + + tclass = flowinfo & 0xff; + *mp = sbcreatecontrol((void *)&tclass, sizeof(tclass), + IPV6_TCLASS, IPPROTO_IPV6); + if (*mp) mp = &(*mp)->m_next; } - /* IN6P_NEXTHOP - for outgoing packet only */ /* * IPV6_HOPOPTS socket option. 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) { /* @@ -1104,14 +1175,14 @@ ip6_savecontrol(in6p, mp, ip6, m) 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; } @@ -1119,10 +1190,11 @@ 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); @@ -1141,7 +1213,7 @@ 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; @@ -1162,7 +1234,7 @@ 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 *); @@ -1172,28 +1244,30 @@ 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; @@ -1225,16 +1299,53 @@ 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; @@ -1252,7 +1363,7 @@ 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 @@ -1275,58 +1386,50 @@ 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; } /* - * 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 * -ip6_get_prevhdr(m, off) - struct mbuf *m; - int off; +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, caddr_t) + 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); } } @@ -1334,11 +1437,7 @@ 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; @@ -1357,7 +1456,7 @@ ip6_nexthdr(m, off, proto, nxtp) 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); @@ -1370,7 +1469,7 @@ 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) @@ -1381,7 +1480,7 @@ 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; @@ -1394,7 +1493,7 @@ 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; @@ -1417,11 +1516,7 @@ 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; @@ -1430,7 +1525,7 @@ 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; @@ -1445,8 +1540,7 @@ ip6_lasthdr(m, off, proto, nxtp) } struct m_tag * -ip6_addaux(m) - struct mbuf *m; +ip6_addaux(struct mbuf *m) { struct m_tag *mtag; @@ -1456,15 +1550,14 @@ ip6_addaux(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; } struct m_tag * -ip6_findaux(m) - struct mbuf *m; +ip6_findaux(struct mbuf *m) { struct m_tag *mtag; @@ -1473,8 +1566,7 @@ ip6_findaux(m) } void -ip6_delaux(m) - struct mbuf *m; +ip6_delaux(struct mbuf *m) { struct m_tag *mtag; @@ -1483,11 +1575,71 @@ ip6_delaux(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 error; + + error = sysctl_lookup(SYSCTLFN_CALL(rnode)); + if (error || newp == NULL) + return (error); + + mutex_enter(softnet_lock); + KERNEL_LOCK(1, NULL); + + ip6flow_reap(0); + + KERNEL_UNLOCK_ONE(NULL); + mutex_exit(softnet_lock); + + 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() + */ + mutex_enter(softnet_lock); + KERNEL_LOCK(1, NULL); + + error = ip6flow_invalidate_all(tmp); + + KERNEL_UNLOCK_ONE(NULL); + mutex_exit(softnet_lock); + } else { + /* + * EINVAL if not a power of 2 + */ + error = EINVAL; + } + + return error; +} +#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, @@ -1496,14 +1648,22 @@ u_char inet6ctlerrmap[PRC_NCMDS] = { ENOPROTOOPT }; -SYSCTL_SETUP(sysctl_net_inet6_ip6_setup, "sysctl net.inet6.ip6 subtree setup") +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", @@ -1597,6 +1757,20 @@ 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, @@ -1682,6 +1856,15 @@ SYSCTL_SETUP(sysctl_net_inet6_ip6_setup, IPV6CTL_V6ONLY, CTL_EOL); 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_READWRITE, CTLTYPE_INT, "anonportmin", SYSCTL_DESCR("Lowest ephemeral port number to assign"), sysctl_net_inet_ip_ports, 0, &ip6_anonportmin, 0, @@ -1721,6 +1904,14 @@ 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, @@ -1744,7 +1935,7 @@ 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, @@ -1761,4 +1952,86 @@ 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 + 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 + /* 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); }