static int in_ifadownkill(struct rtentry *rt, void *xap) { struct in_ifadown_arg *ap = xap; RT_LOCK(rt); if (rt->rt_ifa == ap->ifa && (ap->del || !(rt->rt_flags & RTF_STATIC))) { /* * Aquire a reference so that it can later be freed * as the refcount would be 0 here in case of at least * ap->del. */ RT_ADDREF(rt); /* * Disconnect it from the tree and permit protocols * to cleanup. */ rt_expunge(ap->rnh, rt); /* * At this point it is an rttrash node, and in case * the above is the only reference we must free it. * If we do not noone will have a pointer and the * rtentry will be leaked forever. * In case someone else holds a reference, we are * fine as we only decrement the refcount. In that * case if the other entity calls RT_REMREF, we * will still be leaking but at least we tried. */ RTFREE_LOCKED(rt); return (0); } RT_UNLOCK(rt); return 0; }
void rtalloc_mpath_fib(struct route *ro, uint32_t hash, u_int fibnum) { struct radix_node *rn0, *rn; u_int32_t n; struct rtentry *rt; int64_t weight; /* * XXX we don't attempt to lookup cached route again; what should * be done for sendto(3) case? */ if (ro->ro_rt && ro->ro_rt->rt_ifp && (ro->ro_rt->rt_flags & RTF_UP) && RT_LINK_IS_UP(ro->ro_rt->rt_ifp)) return; ro->ro_rt = rtalloc1_fib(&ro->ro_dst, 1, 0, fibnum); /* if the route does not exist or it is not multipath, don't care */ if (ro->ro_rt == NULL) return; if (rn_mpath_next((struct radix_node *)ro->ro_rt) == NULL) { RT_UNLOCK(ro->ro_rt); return; } /* beyond here, we use rn as the master copy */ rn0 = rn = (struct radix_node *)ro->ro_rt; n = rn_mpath_count(rn0); /* gw selection by Modulo-N Hash (RFC2991) XXX need improvement? */ hash += hashjitter; hash %= n; for (weight = abs((int32_t)hash), rt = ro->ro_rt; weight >= rt->rt_weight && rn; weight -= rt->rt_weight) { /* stay within the multipath routes */ if (rn->rn_dupedkey && rn->rn_mask != rn->rn_dupedkey->rn_mask) break; rn = rn->rn_dupedkey; rt = (struct rtentry *)rn; } /* XXX try filling rt_gwroute and avoid unreachable gw */ /* gw selection has failed - there must be only zero weight routes */ if (!rn) { RT_UNLOCK(ro->ro_rt); ro->ro_rt = NULL; return; } if (ro->ro_rt != rt) { RTFREE_LOCKED(ro->ro_rt); ro->ro_rt = (struct rtentry *)rn; RT_LOCK(ro->ro_rt); RT_ADDREF(ro->ro_rt); } RT_UNLOCK(ro->ro_rt); }
void rtalloc_mpath_fib(struct route *ro, uint32_t hash, u_int fibnum) { struct rtentry *rt; /* * XXX we don't attempt to lookup cached route again; what should * be done for sendto(3) case? */ if (ro->ro_rt && ro->ro_rt->rt_ifp && (ro->ro_rt->rt_flags & RTF_UP) && RT_LINK_IS_UP(ro->ro_rt->rt_ifp)) return; ro->ro_rt = rtalloc1_fib(&ro->ro_dst, 1, 0, fibnum); /* if the route does not exist or it is not multipath, don't care */ if (ro->ro_rt == NULL) return; if (rn_mpath_next((struct radix_node *)ro->ro_rt) == NULL) { RT_UNLOCK(ro->ro_rt); return; } rt = rt_mpath_selectrte(ro->ro_rt, hash); /* XXX try filling rt_gwroute and avoid unreachable gw */ /* gw selection has failed - there must be only zero weight routes */ if (!rt) { RT_UNLOCK(ro->ro_rt); ro->ro_rt = NULL; return; } if (ro->ro_rt != rt) { RTFREE_LOCKED(ro->ro_rt); ro->ro_rt = rt; RT_LOCK(ro->ro_rt); RT_ADDREF(ro->ro_rt); } RT_UNLOCK(ro->ro_rt); }
/* * Input a Neighbor Solicitation Message. * * Based on RFC 2461 * Based on RFC 2462 (duplicate address detection) */ void nd6_ns_input(struct mbuf *m, int off, int icmp6len) { struct ifnet *ifp = m->m_pkthdr.rcvif; struct ip6_hdr *ip6 = mtod(m, struct ip6_hdr *); struct nd_neighbor_solicit *nd_ns; struct in6_addr saddr6 = ip6->ip6_src; struct in6_addr daddr6 = ip6->ip6_dst; struct in6_addr taddr6; struct in6_addr myaddr6; char *lladdr = NULL; struct ifaddr *ifa = NULL; int lladdrlen = 0; int anycast = 0, proxy = 0, tentative = 0; int tlladdr; int rflag; union nd_opts ndopts; struct sockaddr_dl proxydl; char ip6bufs[INET6_ADDRSTRLEN], ip6bufd[INET6_ADDRSTRLEN]; rflag = (V_ip6_forwarding) ? ND_NA_FLAG_ROUTER : 0; if (ND_IFINFO(ifp)->flags & ND6_IFF_ACCEPT_RTADV && V_ip6_norbit_raif) rflag = 0; #ifndef PULLDOWN_TEST IP6_EXTHDR_CHECK(m, off, icmp6len,); nd_ns = (struct nd_neighbor_solicit *)((caddr_t)ip6 + off); #else IP6_EXTHDR_GET(nd_ns, struct nd_neighbor_solicit *, m, off, icmp6len); if (nd_ns == NULL) { ICMP6STAT_INC(icp6s_tooshort); return; } #endif ip6 = mtod(m, struct ip6_hdr *); /* adjust pointer for safety */ taddr6 = nd_ns->nd_ns_target; if (in6_setscope(&taddr6, ifp, NULL) != 0) goto bad; if (ip6->ip6_hlim != 255) { nd6log((LOG_ERR, "nd6_ns_input: invalid hlim (%d) from %s to %s on %s\n", ip6->ip6_hlim, ip6_sprintf(ip6bufs, &ip6->ip6_src), ip6_sprintf(ip6bufd, &ip6->ip6_dst), if_name(ifp))); goto bad; } if (IN6_IS_ADDR_UNSPECIFIED(&saddr6)) { /* dst has to be a solicited node multicast address. */ if (daddr6.s6_addr16[0] == IPV6_ADDR_INT16_MLL && /* don't check ifindex portion */ daddr6.s6_addr32[1] == 0 && daddr6.s6_addr32[2] == IPV6_ADDR_INT32_ONE && daddr6.s6_addr8[12] == 0xff) { ; /* good */ } else { nd6log((LOG_INFO, "nd6_ns_input: bad DAD packet " "(wrong ip6 dst)\n")); goto bad; } } else if (!V_nd6_onlink_ns_rfc4861) { struct sockaddr_in6 src_sa6; /* * According to recent IETF discussions, it is not a good idea * to accept a NS from an address which would not be deemed * to be a neighbor otherwise. This point is expected to be * clarified in future revisions of the specification. */ bzero(&src_sa6, sizeof(src_sa6)); src_sa6.sin6_family = AF_INET6; src_sa6.sin6_len = sizeof(src_sa6); src_sa6.sin6_addr = saddr6; if (nd6_is_addr_neighbor(&src_sa6, ifp) == 0) { nd6log((LOG_INFO, "nd6_ns_input: " "NS packet from non-neighbor\n")); goto bad; } } if (IN6_IS_ADDR_MULTICAST(&taddr6)) { nd6log((LOG_INFO, "nd6_ns_input: bad NS target (multicast)\n")); goto bad; } icmp6len -= sizeof(*nd_ns); nd6_option_init(nd_ns + 1, icmp6len, &ndopts); if (nd6_options(&ndopts) < 0) { nd6log((LOG_INFO, "nd6_ns_input: invalid ND option, ignored\n")); /* nd6_options have incremented stats */ goto freeit; } if (ndopts.nd_opts_src_lladdr) { lladdr = (char *)(ndopts.nd_opts_src_lladdr + 1); lladdrlen = ndopts.nd_opts_src_lladdr->nd_opt_len << 3; } if (IN6_IS_ADDR_UNSPECIFIED(&ip6->ip6_src) && lladdr) { nd6log((LOG_INFO, "nd6_ns_input: bad DAD packet " "(link-layer address option)\n")); goto bad; } /* * Attaching target link-layer address to the NA? * (RFC 2461 7.2.4) * * NS IP dst is unicast/anycast MUST NOT add * NS IP dst is solicited-node multicast MUST add * * In implementation, we add target link-layer address by default. * We do not add one in MUST NOT cases. */ if (!IN6_IS_ADDR_MULTICAST(&daddr6)) tlladdr = 0; else tlladdr = 1; /* * Target address (taddr6) must be either: * (1) Valid unicast/anycast address for my receiving interface, * (2) Unicast address for which I'm offering proxy service, or * (3) "tentative" address on which DAD is being performed. */ /* (1) and (3) check. */ if (ifp->if_carp) ifa = (*carp_iamatch6_p)(ifp, &taddr6); else ifa = (struct ifaddr *)in6ifa_ifpwithaddr(ifp, &taddr6); /* (2) check. */ if (ifa == NULL) { struct rtentry *rt; struct sockaddr_in6 tsin6; int need_proxy; #ifdef RADIX_MPATH struct route_in6 ro; #endif bzero(&tsin6, sizeof tsin6); tsin6.sin6_len = sizeof(struct sockaddr_in6); tsin6.sin6_family = AF_INET6; tsin6.sin6_addr = taddr6; /* Always use the default FIB. */ #ifdef RADIX_MPATH bzero(&ro, sizeof(ro)); ro.ro_dst = tsin6; rtalloc_mpath_fib((struct route *)&ro, RTF_ANNOUNCE, RT_DEFAULT_FIB); rt = ro.ro_rt; #else rt = in6_rtalloc1((struct sockaddr *)&tsin6, 0, 0, RT_DEFAULT_FIB); #endif need_proxy = (rt && (rt->rt_flags & RTF_ANNOUNCE) != 0 && rt->rt_gateway->sa_family == AF_LINK); if (rt != NULL) { /* * Make a copy while we can be sure that rt_gateway * is still stable before unlocking to avoid lock * order problems. proxydl will only be used if * proxy will be set in the next block. */ if (need_proxy) proxydl = *SDL(rt->rt_gateway); RTFREE_LOCKED(rt); } if (need_proxy) { /* * proxy NDP for single entry */ ifa = (struct ifaddr *)in6ifa_ifpforlinklocal(ifp, IN6_IFF_NOTREADY|IN6_IFF_ANYCAST); if (ifa) proxy = 1; } } if (ifa == NULL) { /* * We've got an NS packet, and we don't have that adddress * assigned for us. We MUST silently ignore it. * See RFC2461 7.2.3. */ goto freeit; } myaddr6 = *IFA_IN6(ifa); anycast = ((struct in6_ifaddr *)ifa)->ia6_flags & IN6_IFF_ANYCAST; tentative = ((struct in6_ifaddr *)ifa)->ia6_flags & IN6_IFF_TENTATIVE; if (((struct in6_ifaddr *)ifa)->ia6_flags & IN6_IFF_DUPLICATED) goto freeit; if (lladdr && ((ifp->if_addrlen + 2 + 7) & ~7) != lladdrlen) { nd6log((LOG_INFO, "nd6_ns_input: lladdrlen mismatch for %s " "(if %d, NS packet %d)\n", ip6_sprintf(ip6bufs, &taddr6), ifp->if_addrlen, lladdrlen - 2)); goto bad; } if (IN6_ARE_ADDR_EQUAL(&myaddr6, &saddr6)) { nd6log((LOG_INFO, "nd6_ns_input: duplicate IP6 address %s\n", ip6_sprintf(ip6bufs, &saddr6))); goto freeit; } /* * We have neighbor solicitation packet, with target address equals to * one of my tentative address. * * src addr how to process? * --- --- * multicast of course, invalid (rejected in ip6_input) * unicast somebody is doing address resolution -> ignore * unspec dup address detection * * The processing is defined in RFC 2462. */ if (tentative) { /* * If source address is unspecified address, it is for * duplicate address detection. * * If not, the packet is for addess resolution; * silently ignore it. */ if (IN6_IS_ADDR_UNSPECIFIED(&saddr6)) nd6_dad_ns_input(ifa); goto freeit; } /* * If the source address is unspecified address, entries must not * be created or updated. * It looks that sender is performing DAD. Output NA toward * all-node multicast address, to tell the sender that I'm using * the address. * S bit ("solicited") must be zero. */ if (IN6_IS_ADDR_UNSPECIFIED(&saddr6)) { struct in6_addr in6_all; in6_all = in6addr_linklocal_allnodes; if (in6_setscope(&in6_all, ifp, NULL) != 0) goto bad; nd6_na_output_fib(ifp, &in6_all, &taddr6, ((anycast || proxy || !tlladdr) ? 0 : ND_NA_FLAG_OVERRIDE) | rflag, tlladdr, proxy ? (struct sockaddr *)&proxydl : NULL, M_GETFIB(m)); goto freeit; } nd6_cache_lladdr(ifp, &saddr6, lladdr, lladdrlen, ND_NEIGHBOR_SOLICIT, 0); nd6_na_output_fib(ifp, &saddr6, &taddr6, ((anycast || proxy || !tlladdr) ? 0 : ND_NA_FLAG_OVERRIDE) | rflag | ND_NA_FLAG_SOLICITED, tlladdr, proxy ? (struct sockaddr *)&proxydl : NULL, M_GETFIB(m)); freeit: if (ifa != NULL) ifa_free(ifa); m_freem(m); return; bad: nd6log((LOG_ERR, "nd6_ns_input: src=%s\n", ip6_sprintf(ip6bufs, &saddr6))); nd6log((LOG_ERR, "nd6_ns_input: dst=%s\n", ip6_sprintf(ip6bufs, &daddr6))); nd6log((LOG_ERR, "nd6_ns_input: tgt=%s\n", ip6_sprintf(ip6bufs, &taddr6))); ICMP6STAT_INC(icp6s_badns); if (ifa != NULL) ifa_free(ifa); m_freem(m); }
/* * Do what we need to do when inserting a route. */ static struct radix_node * in6_addroute(void *v_arg, void *n_arg, struct radix_node_head *head, struct radix_node *treenodes) { struct rtentry *rt = (struct rtentry *)treenodes; struct sockaddr_in6 *sin6 = (struct sockaddr_in6 *)rt_key(rt); struct radix_node *ret; if (IN6_IS_ADDR_MULTICAST(&sin6->sin6_addr)) rt->rt_flags |= RTF_MULTICAST; /* * A little bit of help for both IPv6 output and input: * For local addresses, we make sure that RTF_LOCAL is set, * with the thought that this might one day be used to speed up * ip_input(). * * We also mark routes to multicast addresses as such, because * it's easy to do and might be useful (but this is much more * dubious since it's so easy to inspect the address). (This * is done above.) * * XXX * should elaborate the code. */ if (rt->rt_flags & RTF_HOST) { if (IN6_ARE_ADDR_EQUAL(&satosin6(rt->rt_ifa->ifa_addr) ->sin6_addr, &sin6->sin6_addr)) { rt->rt_flags |= RTF_LOCAL; } } if (!rt->rt_rmx.rmx_mtu && rt->rt_ifp) rt->rt_rmx.rmx_mtu = IN6_LINKMTU(rt->rt_ifp); ret = rn_addroute(v_arg, n_arg, head, treenodes); if (ret == NULL && rt->rt_flags & RTF_HOST) { struct rtentry *rt2; /* * We are trying to add a host route, but can't. * Find out if it is because of an * ARP entry and delete it if so. */ rt2 = rtalloc1((struct sockaddr *)sin6, 0, RTF_CLONING); if (rt2) { if (rt2->rt_flags & RTF_LLINFO && rt2->rt_flags & RTF_HOST && rt2->rt_gateway && rt2->rt_gateway->sa_family == AF_LINK) { rtexpunge(rt2); RTFREE_LOCKED(rt2); ret = rn_addroute(v_arg, n_arg, head, treenodes); } else RTFREE_LOCKED(rt2); } } else if (ret == NULL && rt->rt_flags & RTF_CLONING) { struct rtentry *rt2; /* * We are trying to add a net route, but can't. * The following case should be allowed, so we'll make a * special check for this: * Two IPv6 addresses with the same prefix is assigned * to a single interrface. * # ifconfig if0 inet6 3ffe:0501::1 prefix 64 alias (*1) * # ifconfig if0 inet6 3ffe:0501::2 prefix 64 alias (*2) * In this case, (*1) and (*2) want to add the same * net route entry, 3ffe:0501:: -> if0. * This case should not raise an error. */ rt2 = rtalloc1((struct sockaddr *)sin6, 0, RTF_CLONING); if (rt2) { if ((rt2->rt_flags & (RTF_CLONING|RTF_HOST|RTF_GATEWAY)) == RTF_CLONING && rt2->rt_gateway && rt2->rt_gateway->sa_family == AF_LINK && rt2->rt_ifp == rt->rt_ifp) { ret = rt2->rt_nodes; } RTFREE_LOCKED(rt2); } } return ret; }
/* * Do what we need to do when inserting a route. */ static struct radix_node * in_addroute(void *v_arg, void *n_arg, struct radix_node_head *head, struct radix_node *treenodes) { struct rtentry *rt = (struct rtentry *)treenodes; struct sockaddr_in *sin = (struct sockaddr_in *)rt_key(rt); struct radix_node *ret; /* * A little bit of help for both IP output and input: * For host routes, we make sure that RTF_BROADCAST * is set for anything that looks like a broadcast address. * This way, we can avoid an expensive call to in_broadcast() * in ip_output() most of the time (because the route passed * to ip_output() is almost always a host route). * * We also do the same for local addresses, with the thought * that this might one day be used to speed up ip_input(). * * We also mark routes to multicast addresses as such, because * it's easy to do and might be useful (but this is much more * dubious since it's so easy to inspect the address). */ if (rt->rt_flags & RTF_HOST) { if (in_broadcast(sin->sin_addr, rt->rt_ifp)) { rt->rt_flags |= RTF_BROADCAST; } else if (satosin(rt->rt_ifa->ifa_addr)->sin_addr.s_addr == sin->sin_addr.s_addr) { rt->rt_flags |= RTF_LOCAL; } } if (IN_MULTICAST(ntohl(sin->sin_addr.s_addr))) rt->rt_flags |= RTF_MULTICAST; if (!rt->rt_rmx.rmx_mtu && rt->rt_ifp) rt->rt_rmx.rmx_mtu = rt->rt_ifp->if_mtu; ret = rn_addroute(v_arg, n_arg, head, treenodes); if (ret == NULL && rt->rt_flags & RTF_HOST) { struct rtentry *rt2; /* * We are trying to add a host route, but can't. * Find out if it is because of an * ARP entry and delete it if so. */ rt2 = rtalloc1((struct sockaddr *)sin, 0, RTF_CLONING); if (rt2) { if (rt2->rt_flags & RTF_LLINFO && rt2->rt_flags & RTF_HOST && rt2->rt_gateway && rt2->rt_gateway->sa_family == AF_LINK) { rtexpunge(rt2); RTFREE_LOCKED(rt2); ret = rn_addroute(v_arg, n_arg, head, treenodes); } else RTFREE_LOCKED(rt2); } } return ret; }
/* * Accept by socket(4) in AF_ROUTE received service requests. * * o RTAX_DST holds key x for fec = < x, nh > where nh -> ifp * * o RTAX_GATEWAY holds by sockaddr_ftn{} for MPLS label * binding necessary * * < seg_i , seg_j > * * tuple where seg_i denotes key for by nhlfe generated ilm, * Furthermore, seg_j denotes typically upstream label for * transmission downstream by interface ifp in link-layer. * * o In rt_addrinfo{} Service Primitive (spi) contained flags * encodes with MPLS label binding linked operation. * * RTF_{POP|PUSH|SWAP} - self expanatory. * * RTF_MPE, denotes initial label of Label Switch Path. * * RTF_STK, denotes label stacking, but not yet fully * implemented. */ int mpls_rt_output_fib(struct rt_msghdr *rtm, struct rt_addrinfo *rti, struct rtentry **rt, u_int fibnum) { struct rtentry *fec = NULL; struct ifnet *ifp = NULL; struct mpls_aliasreq ifra; int error = 0, cmd = 0; #ifdef MPLS_DEBUG (void)printf("%s\n", __func__); #endif /* MPLS_DEBUG */ if (rti_dst(rti)->sa_len > sizeof(ifra.ifra_x)) { log(LOG_INFO, "%s: destination x in fec invalid\n", __func__); error = EMSGSIZE; goto out; } if (rti_gateway(rti)->sa_family != AF_MPLS) { log(LOG_INFO, "%s: segment invalid\n", __func__); error = EINVAL; goto out; } if (rti_gateway(rti)->sa_len > sizeof(ifra.ifra_seg)) { log(LOG_INFO, "%s: segment invalid\n", __func__); error = EMSGSIZE; goto out; } /* * Fetch interface by Forward Equivalence Class (fec). */ fec = rtalloc1_fib(rti_dst(rti), 0, 0UL, fibnum); if ((fec == NULL) || (fec->rt_gateway == NULL) || ((ifp = fec->rt_ifp) == NULL) || (fec->rt_ifa == NULL) || ((fec->rt_flags & RTF_UP) == 0)) { error = ESRCH; goto out; } bzero(&ifra, sizeof(ifra)); bcopy(rti_dst(rti), &ifra.ifra_x, rti_dst(rti)->sa_len); switch ((int)rtm->rtm_type) { case RTM_ADD: /* * Apply MPLS label binding on Forward Equivalence Class (fec). */ cmd = SIOCAIFADDR; /* FALLTHROUGH */ case RTM_DELETE: /* * Delete MPLS label binding on fec. */ cmd = (cmd == 0) ? SIOCDIFADDR : cmd; /* * Perform MPLS control operations on interface-layer. */ bcopy(rti_gateway(rti), &ifra.ifra_seg, rti_gateway(rti)->sa_len); ifra.ifra_flags = rti_flags(rti); RT_UNLOCK(fec); error = mpls_control(NULL, cmd, (caddr_t)&ifra, ifp, NULL); RT_LOCK(fec); break; case RTM_GET: /* * XXX: looks ugly... I'll delegate this operation * XXX: back to rt_output, but I'm not yet sure, if * XXX: I'll should do that... */ ifra.ifra_seg.sftn_len = SMPLS_LEN; ifra.ifra_seg.sftn_family = AF_MPLS; ((struct sockaddr_mpls *)&ifra.ifra_seg)->smpls_label = satosmpls_label(rti_gateway(rti)); /* * Fetch Incoming Label Map (ilm) by MPLS label binding on fec. */ *rt = ((ifra.ifra_flags & RTF_MPE) == 0) ? rtalloc1_fib((struct sockaddr *)&ifra.ifra_seg, 0, 0UL, ifp->if_fib) : NULL; if (*rt != NULL) { /* * Update by socket(2) on route(4) used Service Data Unit (sdu). */ bcopy(rt_key(*rt), rti_dst(rti), rt_key(*rt)->sa_len); bcopy((*rt)->rt_gateway, rti_gateway(rti), (*rt)->rt_gateway->sa_len); } else error = EADDRNOTAVAIL; break; default: log(LOG_INFO, "%s: command invalid\n", __func__); error = EOPNOTSUPP; break; } out: if (fec != NULL) RTFREE_LOCKED(fec); return (error); }