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 1.82.2.3 retrieving revision 1.117 diff -u -p -r1.82.2.3 -r1.117 --- src/sys/netinet6/ip6_input.c 2006/09/09 02:58:55 1.82.2.3 +++ src/sys/netinet6/ip6_input.c 2008/04/23 06:09:05 1.117 @@ -1,4 +1,4 @@ -/* $NetBSD: ip6_input.c,v 1.82.2.3 2006/09/09 02:58:55 rpaulo Exp $ */ +/* $NetBSD: ip6_input.c,v 1.117 2008/04/23 06:09:05 thorpej Exp $ */ /* $KAME: ip6_input.c,v 1.188 2001/03/29 05:34:31 itojun Exp $ */ /* @@ -62,7 +62,7 @@ */ #include -__KERNEL_RCSID(0, "$NetBSD: ip6_input.c,v 1.82.2.3 2006/09/09 02:58:55 rpaulo Exp $"); +__KERNEL_RCSID(0, "$NetBSD: ip6_input.c,v 1.117 2008/04/23 06:09:05 thorpej Exp $"); #include "opt_inet.h" #include "opt_inet6.h" @@ -95,7 +95,6 @@ __KERNEL_RCSID(0, "$NetBSD: ip6_input.c, #include #include -#include #ifdef INET #include #include @@ -103,6 +102,8 @@ __KERNEL_RCSID(0, "$NetBSD: ip6_input.c, #include #include #include +#include +#include #include #include #include @@ -110,8 +111,15 @@ __KERNEL_RCSID(0, "$NetBSD: ip6_input.c, #ifdef IPSEC #include +#include #endif +#ifdef FAST_IPSEC +#include +#include +#include +#endif /* FAST_IPSEC */ + #include #include "faith.h" @@ -130,7 +138,7 @@ static int ip6qmaxlen = IFQ_MAXLEN; struct in6_ifaddr *in6_ifaddr; struct ifqueue ip6intrq; -extern struct callout in6_tmpaddrtimer_ch; +extern callout_t in6_tmpaddrtimer_ch; int ip6_forward_srcrt; /* XXX */ int ip6_sourcecheck; /* XXX */ @@ -140,20 +148,20 @@ int ip6_sourcecheck_interval; /* XXX */ struct pfil_head inet6_pfil_hook; #endif -struct ip6stat ip6stat; +percpu_t *ip6stat_percpu; -static void ip6_init2 __P((void *)); -static struct m_tag *ip6_setdstifaddr __P((struct mbuf *, struct in6_ifaddr *)); +static void ip6_init2(void *); +static struct m_tag *ip6_setdstifaddr(struct mbuf *, const struct in6_ifaddr *); -static int ip6_hopopts_input __P((u_int32_t *, u_int32_t *, struct mbuf **, int *)); -static struct mbuf *ip6_pullexthdr __P((struct mbuf *, size_t, int)); +static int ip6_hopopts_input(u_int32_t *, u_int32_t *, struct mbuf **, int *); +static struct mbuf *ip6_pullexthdr(struct mbuf *, size_t, int); /* * IP6 initialization: fill in IP6 protocol switch table. * All protocols not implemented in kernel go to raw IP6 protocol handler. */ void -ip6_init() +ip6_init(void) { const struct ip6protosw *pr; int i; @@ -176,6 +184,9 @@ ip6_init() ip6_desync_factor = arc4random() % MAX_TEMP_DESYNC_FACTOR; ip6_init2((void *)0); +#ifdef GATEWAY + ip6flow_init(ip6_hashsize); +#endif #ifdef PFIL_HOOKS /* Register our Packet Filter hook. */ @@ -186,19 +197,20 @@ ip6_init() printf("ip6_init: WARNING: unable to register pfil hook, " "error %d\n", i); #endif /* PFIL_HOOKS */ + + 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, 0); 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, 0); callout_reset(&in6_tmpaddrtimer_ch, (ip6_temp_preferred_lifetime - ip6_desync_factor - ip6_temp_regen_advance) * hz, @@ -209,7 +221,7 @@ ip6_init2(dummy) * IP6 input interrupt handling. Just pass the packet to ip6_input. */ void -ip6intr() +ip6intr(void) { int s; struct mbuf *m; @@ -229,19 +241,29 @@ ip6intr() } } -extern struct route_in6 ip6_forward_rt; +extern struct route ip6_forward_rt; void -ip6_input(m) - struct mbuf *m; +ip6_input(struct mbuf *m) { struct ip6_hdr *ip6; - int off = sizeof(struct ip6_hdr), nest; + int hit, off = sizeof(struct ip6_hdr), nest; u_int32_t plen; u_int32_t rtalert = ~0; - int nxt, ours = 0; + int nxt, ours = 0, rh_present = 0; struct ifnet *deliverifp = NULL; int srcrt = 0; + const struct rtentry *rt; + union { + struct sockaddr dst; + struct sockaddr_in6 dst6; + } u; +#ifdef FAST_IPSEC + struct m_tag *mtag; + struct tdb_ident *tdbi; + struct secpolicy *sp; + int s, error; +#endif #ifdef IPSEC /* @@ -262,25 +284,26 @@ 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 @@ -288,19 +311,19 @@ 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; } @@ -309,11 +332,22 @@ 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; } +#if defined(IPSEC) + /* IPv6 fast forwarding is not compatible with IPsec. */ + m->m_flags &= ~M_CANFASTFWD; +#else + /* + * Assume that we can create a fast-forward IP flow entry + * based on this packet. + */ + m->m_flags |= M_CANFASTFWD; +#endif + #ifdef PFIL_HOOKS /* * Run through list of hooks for input packets. If there are any @@ -328,6 +362,8 @@ ip6_input(m) */ #ifdef IPSEC if (!ipsec_getnhist(m)) +#elif defined(FAST_IPSEC) + if (!ipsec_indone(m)) #else if (1) #endif @@ -345,7 +381,7 @@ ip6_input(m) } #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) { @@ -362,7 +398,7 @@ 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; } @@ -380,7 +416,7 @@ 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; } @@ -394,7 +430,7 @@ 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; } @@ -415,12 +451,12 @@ ip6_input(m) 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; } @@ -439,8 +475,10 @@ 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; } @@ -448,34 +486,18 @@ 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 @@ -486,22 +508,19 @@ 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; /* @@ -529,11 +548,11 @@ 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; } } @@ -547,9 +566,7 @@ 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) @@ -568,7 +585,7 @@ 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; } @@ -580,18 +597,16 @@ 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. + */ } } @@ -625,18 +640,18 @@ 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)); @@ -658,7 +673,7 @@ 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; } @@ -683,7 +698,7 @@ 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; } @@ -709,7 +724,7 @@ 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; } @@ -725,13 +740,15 @@ 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; 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; } @@ -740,11 +757,20 @@ 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; + } + } + #ifdef IPSEC /* * enforce IPsec policy checking if we are seeing last header. @@ -753,10 +779,49 @@ ip6_input(m) */ if ((inet6sw[ip6_protox[nxt]].pr_flags & PR_LASTHDR) != 0 && ipsec6_in_reject(m, NULL)) { - ipsec6stat.in_polvio++; + IPSEC6_STATINC(IPSEC_STAT_IN_POLVIO); goto bad; } #endif +#ifdef FAST_IPSEC + /* + * enforce IPsec policy checking if we are seeing last header. + * note that we do not visit this with protocols with pcb layer + * code - like udp/tcp/raw ip. + */ + if ((inet6sw[ip_protox[nxt]].pr_flags & PR_LASTHDR) != 0) { + /* + * Check if the packet has already had IPsec processing + * done. If so, then just pass it along. This tag gets + * set during AH, ESP, etc. input handling, before the + * packet is returned to the ip input queue for delivery. + */ + mtag = m_tag_find(m, PACKET_TAG_IPSEC_IN_DONE, NULL); + s = splsoftnet(); + if (mtag != NULL) { + tdbi = (struct tdb_ident *)(mtag + 1); + sp = ipsec_getpolicy(tdbi, IPSEC_DIR_INBOUND); + } else { + sp = ipsec_getpolicybyaddr(m, IPSEC_DIR_INBOUND, + IP_FORWARDING, &error); + } + if (sp != NULL) { + /* + * Check security policy against packet attributes. + */ + error = ipsec_in_reject(sp, m); + KEY_FREESP(&sp); + } else { + /* XXX error stat??? */ + error = EINVAL; + DPRINTF(("ip6_input: no SP, packet discarded\n"));/*XXX*/ + } + splx(s); + if (error) + goto bad; + } +#endif /* FAST_IPSEC */ + nxt = (*inet6sw[ip6_protox[nxt]].pr_input)(&m, &off, nxt); } @@ -769,27 +834,30 @@ 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; mtag = ip6_addaux(m); - if (mtag) - ((struct ip6aux *)(mtag + 1))->ip6a_dstia6 = ia6; + if (mtag != NULL) { + struct ip6aux *ip6a; + + ip6a = (struct ip6aux *)(mtag + 1); + in6_setscope(&ip6a->ip6a_src, ia->ia_ifp, &ip6a->ip6a_scope_id); + ip6a->ip6a_src = ia->ia_addr.sin6_addr; + ip6a->ip6a_flags = ia->ia6_flags; + } return mtag; /* NULL if failed to set */ } -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; } @@ -797,13 +865,12 @@ 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; +ip6_hopopts_input(u_int32_t *plenp, u_int32_t *rtalertp, + struct mbuf **mp, int *offp) { struct mbuf *m = *mp; int off = *offp, hbhlen; @@ -813,14 +880,14 @@ 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)); @@ -847,12 +914,8 @@ ip6_hopopts_input(plenp, rtalertp, mp, o * opthead + hbhlen is located in continuous memory region. */ int -ip6_process_hopopts(m, opthead, hbhlen, rtalertp, plenp) - struct mbuf *m; - u_int8_t *opthead; - int hbhlen; - u_int32_t *rtalertp; - u_int32_t *plenp; +ip6_process_hopopts(struct mbuf *m, u_int8_t *opthead, int hbhlen, + u_int32_t *rtalertp, u_int32_t *plenp) { struct ip6_hdr *ip6; int optlen = 0; @@ -868,7 +931,7 @@ 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; @@ -876,7 +939,7 @@ 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) { @@ -887,13 +950,13 @@ ip6_process_hopopts(m, opthead, hbhlen, return (-1); } optlen = IP6OPT_RTALERT_LEN; - bcopy((caddr_t)(opt + 2), (caddr_t)&rtalert_val, 2); + bcopy((void *)(opt + 2), (void *)&rtalert_val, 2); *rtalertp = ntohs(rtalert_val); break; case IP6OPT_JUMBO: /* 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) { @@ -911,7 +974,7 @@ 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); @@ -935,7 +998,7 @@ 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); @@ -947,7 +1010,7 @@ 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); @@ -958,7 +1021,7 @@ 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, @@ -984,10 +1047,7 @@ 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; @@ -998,11 +1058,11 @@ 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))) @@ -1029,11 +1089,8 @@ ip6_unknown_opt(optp, m, off) * you are using IP6_EXTHDR_CHECK() not m_pulldown()) */ void -ip6_savecontrol(inp, mp, ip6, m) - struct inpcb *inp; - 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)) @@ -1042,11 +1099,11 @@ ip6_savecontrol(inp, mp, ip6, m) #endif #ifdef SO_TIMESTAMP - if (inp->inp_socket->so_options & SO_TIMESTAMP) { + if (in6p->in6p_socket->so_options & SO_TIMESTAMP) { struct timeval tv; microtime(&tv); - *mp = sbcreatecontrol((caddr_t) &tv, sizeof(tv), + *mp = sbcreatecontrol((void *) &tv, sizeof(tv), SCM_TIMESTAMP, SOL_SOCKET); if (*mp) mp = &(*mp)->m_next; @@ -1058,23 +1115,24 @@ ip6_savecontrol(inp, mp, ip6, m) return; /* RFC 2292 sec. 5 */ - if ((inp->inp_flags & IN6P_PKTINFO) != 0) { + if ((in6p->in6p_flags & IN6P_PKTINFO) != 0) { struct in6_pktinfo pi6; bcopy(&ip6->ip6_dst, &pi6.ipi6_addr, sizeof(struct in6_addr)); in6_clearscope(&pi6.ipi6_addr); /* XXX */ pi6.ipi6_ifindex = m->m_pkthdr.rcvif ? m->m_pkthdr.rcvif->if_index : 0; - *mp = sbcreatecontrol((caddr_t) &pi6, + *mp = sbcreatecontrol((void *) &pi6, sizeof(struct in6_pktinfo), IS2292(IPV6_2292PKTINFO, IPV6_PKTINFO), IPPROTO_IPV6); if (*mp) mp = &(*mp)->m_next; } - if (inp->inp_flags & IN6P_HOPLIMIT) { + + if (in6p->in6p_flags & IN6P_HOPLIMIT) { int hlim = ip6->ip6_hlim & 0xff; - *mp = sbcreatecontrol((caddr_t) &hlim, sizeof(int), + *mp = sbcreatecontrol((void *) &hlim, sizeof(int), IS2292(IPV6_2292HOPLIMIT, IPV6_HOPLIMIT), IPPROTO_IPV6); if (*mp) mp = &(*mp)->m_next; @@ -1088,7 +1146,7 @@ ip6_savecontrol(inp, mp, ip6, m) flowinfo >>= 20; tclass = flowinfo & 0xff; - *mp = sbcreatecontrol((caddr_t)&tclass, sizeof(tclass), + *mp = sbcreatecontrol((void *)&tclass, sizeof(tclass), IPV6_TCLASS, IPPROTO_IPV6); if (*mp) @@ -1102,7 +1160,7 @@ ip6_savecontrol(inp, mp, ip6, m) * returned to normal user. * See also RFC3542 section 8 (or RFC2292 section 6). */ - if ((inp->inp_flags & IN6P_HOPOPTS) != 0) { + if ((in6p->in6p_flags & IN6P_HOPOPTS) != 0) { /* * Check if a hop-by-hop options header is contatined in the * received packet, and if so, store the options as ancillary @@ -1119,14 +1177,14 @@ ip6_savecontrol(inp, 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; } @@ -1136,7 +1194,7 @@ ip6_savecontrol(inp, mp, ip6, m) * be removed before returning in the RFC 2292. * Note: this constraint is removed in RFC3542. */ - *mp = sbcreatecontrol((caddr_t)hbh, hbhlen, + *mp = sbcreatecontrol((void *)hbh, hbhlen, IS2292(IPV6_2292HOPOPTS, IPV6_HOPOPTS), IPPROTO_IPV6); if (*mp) @@ -1146,7 +1204,7 @@ ip6_savecontrol(inp, mp, ip6, m) } /* IPV6_DSTOPTS and IPV6_RTHDR socket options */ - if (inp->inp_flags & (IN6P_DSTOPTS | IN6P_RTHDR)) { + if (in6p->in6p_flags & (IN6P_DSTOPTS | IN6P_RTHDR)) { struct ip6_hdr *xip6 = mtod(m, struct ip6_hdr *); int nxt = xip6->ip6_nxt, off = sizeof(struct ip6_hdr); @@ -1178,7 +1236,7 @@ ip6_savecontrol(inp, 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 *); @@ -1188,17 +1246,17 @@ ip6_savecontrol(inp, 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 (!inp->inp_flags & IN6P_DSTOPTS) + if (!in6p->in6p_flags & IN6P_DSTOPTS) break; - *mp = sbcreatecontrol((caddr_t)ip6e, elen, + *mp = sbcreatecontrol((void *)ip6e, elen, IS2292(IPV6_2292DSTOPTS, IPV6_DSTOPTS), IPPROTO_IPV6); if (*mp) @@ -1206,10 +1264,10 @@ ip6_savecontrol(inp, mp, ip6, m) break; case IPPROTO_ROUTING: - if (!inp->inp_flags & IN6P_RTHDR) + if (!in6p->in6p_flags & IN6P_RTHDR) break; - *mp = sbcreatecontrol((caddr_t)ip6e, elen, + *mp = sbcreatecontrol((void *)ip6e, elen, IS2292(IPV6_2292RTHDR, IPV6_RTHDR), IPPROTO_IPV6); if (*mp) @@ -1247,7 +1305,8 @@ ip6_savecontrol(inp, mp, ip6, m) void -ip6_notify_pmtu(struct in6pcb *in6p, struct sockaddr_in6 *dst, uint32_t *mtu) +ip6_notify_pmtu(struct in6pcb *in6p, const struct sockaddr_in6 *dst, + uint32_t *mtu) { struct socket *so; struct mbuf *m_mtu; @@ -1269,11 +1328,11 @@ ip6_notify_pmtu(struct in6pcb *in6p, str if (sa6_recoverscope(&mtuctl.ip6m_addr)) return; - if ((m_mtu = sbcreatecontrol((caddr_t)&mtuctl, sizeof(mtuctl), + if ((m_mtu = sbcreatecontrol((void *)&mtuctl, sizeof(mtuctl), IPV6_PATHMTU, IPPROTO_IPV6)) == NULL) return; - if (sbappendaddr(&so->so_rcv, (struct sockaddr *)dst, NULL, m_mtu) + if (sbappendaddr(&so->so_rcv, (const struct sockaddr *)dst, NULL, m_mtu) == 0) { m_freem(m_mtu); /* XXX: should count statistics */ @@ -1288,10 +1347,7 @@ ip6_notify_pmtu(struct in6pcb *in6p, str * 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; @@ -1309,7 +1365,7 @@ 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 @@ -1332,7 +1388,7 @@ 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; } @@ -1350,9 +1406,7 @@ 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 *); @@ -1365,7 +1419,7 @@ 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: @@ -1391,11 +1445,7 @@ 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; @@ -1414,7 +1464,7 @@ 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); @@ -1427,7 +1477,7 @@ 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) @@ -1438,7 +1488,7 @@ 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; @@ -1451,7 +1501,7 @@ 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; @@ -1474,11 +1524,7 @@ 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; @@ -1502,8 +1548,7 @@ ip6_lasthdr(m, off, proto, nxtp) } struct m_tag * -ip6_addaux(m) - struct mbuf *m; +ip6_addaux(struct mbuf *m) { struct m_tag *mtag; @@ -1520,8 +1565,7 @@ ip6_addaux(m) } struct m_tag * -ip6_findaux(m) - struct mbuf *m; +ip6_findaux(struct mbuf *m) { struct m_tag *mtag; @@ -1530,8 +1574,7 @@ ip6_findaux(m) } void -ip6_delaux(m) - struct mbuf *m; +ip6_delaux(struct mbuf *m) { struct m_tag *mtag; @@ -1540,6 +1583,57 @@ 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 s; + + s = sysctl_lookup(SYSCTLFN_CALL(rnode)); + if (s || newp == NULL) + return (s); + + s = splsoftnet(); + ip6flow_reap(0); + splx(s); + + return (0); +} + +static int +sysctl_net_inet6_ip6_hashsize(SYSCTLFN_ARGS) +{ + int error, tmp; + struct sysctlnode node; + + node = *rnode; + tmp = ip6_hashsize; + node.sysctl_data = &tmp; + error = sysctl_lookup(SYSCTLFN_CALL(&node)); + if (error || newp == NULL) + return (error); + + if ((tmp & (tmp - 1)) == 0 && tmp != 0) { + /* + * Can only fail due to malloc() + */ + if (ip6flow_invalidate_all(tmp)) + return ENOMEM; + } else { + /* + * EINVAL if not a power of 2 + */ + return EINVAL; + } + + return (0); +} +#endif /* GATEWAY */ + /* * System control for IP6 */ @@ -1553,6 +1647,18 @@ u_char inet6ctlerrmap[PRC_NCMDS] = { ENOPROTOOPT }; +static int +sysctl_net_inet6_ip6_stats(SYSCTLFN_ARGS) +{ + netstat_sysctl_context ctx; + uint64_t ip6s[IP6_NSTATS]; + + ctx.ctx_stat = ip6stat_percpu; + ctx.ctx_counters = ip6s; + ctx.ctx_ncounters = IP6_NSTATS; + return (NETSTAT_SYSCTL(&ctx)); +} + SYSCTL_SETUP(sysctl_net_inet6_ip6_setup, "sysctl net.inet6.ip6 subtree setup") { #ifdef RFC2292 @@ -1806,7 +1912,7 @@ 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, @@ -1823,4 +1929,28 @@ 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 +} + +void +ip6_statinc(u_int stat) +{ + + KASSERT(stat < IP6_NSTATS); + IP6_STATINC(stat); }