diff --git a/sys/netinet/ip_carp.c b/sys/netinet/ip_carp.c index d3d7957cf087..4f553b9aac5e 100644 --- a/sys/netinet/ip_carp.c +++ b/sys/netinet/ip_carp.c @@ -1,3110 +1,3123 @@ /*- * SPDX-License-Identifier: BSD-2-Clause * * Copyright (c) 2002 Michael Shalayeff. * Copyright (c) 2003 Ryan McBride. * Copyright (c) 2011 Gleb Smirnoff * All rights reserved. * * Redistribution and use in source and binary forms, with or without * modification, are permitted provided that the following conditions * are met: * 1. Redistributions of source code must retain the above copyright * notice, this list of conditions and the following disclaimer. * 2. Redistributions in binary form must reproduce the above copyright * notice, this list of conditions and the following disclaimer in the * documentation and/or other materials provided with the distribution. * * THIS SOFTWARE IS PROVIDED BY THE AUTHOR ``AS IS'' AND ANY EXPRESS OR * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES * OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. * IN NO EVENT SHALL THE AUTHOR OR HIS RELATIVES BE LIABLE FOR ANY DIRECT, * INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES * (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR * SERVICES; LOSS OF MIND, USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, * STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING * IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF * THE POSSIBILITY OF SUCH DAMAGE. */ #include #include "opt_bpf.h" #include "opt_inet.h" #include "opt_inet6.h" #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #if defined(INET) || defined(INET6) #include #include #include #include #include #include #endif #ifdef INET #include #include #endif #ifdef INET6 #include #include #include #include #include #include #endif #include #include #include #include #include static MALLOC_DEFINE(M_CARP, "CARP", "CARP addresses"); struct carp_softc { struct ifnet *sc_carpdev; /* Pointer to parent ifnet. */ struct ifaddr **sc_ifas; /* Our ifaddrs. */ carp_version_t sc_version; /* carp or VRRPv3 */ uint8_t sc_addr[ETHER_ADDR_LEN]; /* Our link level address. */ struct callout sc_ad_tmo; /* Advertising timeout. */ #ifdef INET struct callout sc_md_tmo; /* Master down timeout. */ #endif #ifdef INET6 struct callout sc_md6_tmo; /* XXX: Master down timeout. */ #endif struct mtx sc_mtx; int sc_vhid; union { struct { /* sc_version == CARP_VERSION_CARP */ int sc_advskew; int sc_advbase; struct in_addr sc_carpaddr; struct in6_addr sc_carpaddr6; uint64_t sc_counter; bool sc_init_counter; #define CARP_HMAC_PAD 64 unsigned char sc_key[CARP_KEY_LEN]; unsigned char sc_pad[CARP_HMAC_PAD]; SHA1_CTX sc_sha1; }; struct { /* sc_version == CARP_VERSION_VRRPv3 */ uint8_t sc_vrrp_prio; uint16_t sc_vrrp_adv_inter; uint16_t sc_vrrp_master_inter; }; }; int sc_naddrs; int sc_naddrs6; int sc_ifasiz; enum { INIT = 0, BACKUP, MASTER } sc_state; int sc_suppress; int sc_sendad_errors; #define CARP_SENDAD_MAX_ERRORS 3 int sc_sendad_success; #define CARP_SENDAD_MIN_SUCCESS 3 TAILQ_ENTRY(carp_softc) sc_list; /* On the carp_if list. */ LIST_ENTRY(carp_softc) sc_next; /* On the global list. */ }; struct carp_if { #ifdef INET int cif_naddrs; #endif #ifdef INET6 int cif_naddrs6; #endif TAILQ_HEAD(, carp_softc) cif_vrs; #ifdef INET struct ip_moptions cif_imo; #endif #ifdef INET6 struct ip6_moptions cif_im6o; #endif struct ifnet *cif_ifp; struct mtx cif_mtx; uint32_t cif_flags; #define CIF_PROMISC 0x00000001 }; /* Kernel equivalent of struct carpreq, but with more fields for new features. * */ struct carpkreq { int carpr_count; int carpr_vhid; int carpr_state; int carpr_advskew; int carpr_advbase; unsigned char carpr_key[CARP_KEY_LEN]; /* Everything above this is identical to carpreq */ struct in_addr carpr_addr; struct in6_addr carpr_addr6; carp_version_t carpr_version; uint8_t carpr_vrrp_priority; uint16_t carpr_vrrp_adv_inter; }; /* * Brief design of carp(4). * * Any carp-capable ifnet may have a list of carp softcs hanging off * its ifp->if_carp pointer. Each softc represents one unique virtual * host id, or vhid. The softc has a back pointer to the ifnet. All * softcs are joined in a global list, which has quite limited use. * * Any interface address that takes part in CARP negotiation has a * pointer to the softc of its vhid, ifa->ifa_carp. That could be either * AF_INET or AF_INET6 address. * * Although, one can get the softc's backpointer to ifnet and traverse * through its ifp->if_addrhead queue to find all interface addresses * involved in CARP, we keep a growable array of ifaddr pointers. This * allows us to avoid grabbing the IF_ADDR_LOCK() in many traversals that * do calls into the network stack, thus avoiding LORs. * * Locking: * * Each softc has a lock sc_mtx. It is used to synchronise carp_input_c(), * callout-driven events and ioctl()s. * * To traverse the list of softcs on an ifnet we use CIF_LOCK() or carp_sx. * To traverse the global list we use the mutex carp_mtx. * * Known issues with locking: * * - On module unload we may race (?) with packet processing thread * dereferencing our function pointers. */ /* Accept incoming CARP packets. */ VNET_DEFINE_STATIC(int, carp_allow) = 1; #define V_carp_allow VNET(carp_allow) /* Set DSCP in outgoing CARP packets. */ VNET_DEFINE_STATIC(int, carp_dscp) = 56; #define V_carp_dscp VNET(carp_dscp) /* Preempt slower nodes. */ VNET_DEFINE_STATIC(int, carp_preempt) = 0; #define V_carp_preempt VNET(carp_preempt) /* Log level. */ VNET_DEFINE_STATIC(int, carp_log) = 1; #define V_carp_log VNET(carp_log) /* Global advskew demotion. */ VNET_DEFINE_STATIC(int, carp_demotion) = 0; #define V_carp_demotion VNET(carp_demotion) /* Send error demotion factor. */ VNET_DEFINE_STATIC(int, carp_senderr_adj) = CARP_MAXSKEW; #define V_carp_senderr_adj VNET(carp_senderr_adj) /* Iface down demotion factor. */ VNET_DEFINE_STATIC(int, carp_ifdown_adj) = CARP_MAXSKEW; #define V_carp_ifdown_adj VNET(carp_ifdown_adj) static int carp_allow_sysctl(SYSCTL_HANDLER_ARGS); static int carp_dscp_sysctl(SYSCTL_HANDLER_ARGS); static int carp_demote_adj_sysctl(SYSCTL_HANDLER_ARGS); SYSCTL_NODE(_net_inet, IPPROTO_CARP, carp, CTLFLAG_RW | CTLFLAG_MPSAFE, 0, "CARP"); SYSCTL_PROC(_net_inet_carp, OID_AUTO, allow, CTLFLAG_VNET | CTLTYPE_INT | CTLFLAG_RWTUN | CTLFLAG_NOFETCH | CTLFLAG_MPSAFE, &VNET_NAME(carp_allow), 0, carp_allow_sysctl, "I", "Accept incoming CARP packets"); SYSCTL_PROC(_net_inet_carp, OID_AUTO, dscp, CTLFLAG_VNET | CTLTYPE_INT | CTLFLAG_RW | CTLFLAG_MPSAFE, 0, 0, carp_dscp_sysctl, "I", "DSCP value for carp packets"); SYSCTL_INT(_net_inet_carp, OID_AUTO, preempt, CTLFLAG_VNET | CTLFLAG_RW, &VNET_NAME(carp_preempt), 0, "High-priority backup preemption mode"); SYSCTL_INT(_net_inet_carp, OID_AUTO, log, CTLFLAG_VNET | CTLFLAG_RW, &VNET_NAME(carp_log), 0, "CARP log level"); SYSCTL_PROC(_net_inet_carp, OID_AUTO, demotion, CTLFLAG_VNET | CTLTYPE_INT | CTLFLAG_RW | CTLFLAG_MPSAFE, 0, 0, carp_demote_adj_sysctl, "I", "Adjust demotion factor (skew of advskew)"); SYSCTL_INT(_net_inet_carp, OID_AUTO, senderr_demotion_factor, CTLFLAG_VNET | CTLFLAG_RW, &VNET_NAME(carp_senderr_adj), 0, "Send error demotion factor adjustment"); SYSCTL_INT(_net_inet_carp, OID_AUTO, ifdown_demotion_factor, CTLFLAG_VNET | CTLFLAG_RW, &VNET_NAME(carp_ifdown_adj), 0, "Interface down demotion factor adjustment"); VNET_PCPUSTAT_DEFINE(struct carpstats, carpstats); VNET_PCPUSTAT_SYSINIT(carpstats); VNET_PCPUSTAT_SYSUNINIT(carpstats); #define CARPSTATS_ADD(name, val) \ counter_u64_add(VNET(carpstats)[offsetof(struct carpstats, name) / \ sizeof(uint64_t)], (val)) #define CARPSTATS_INC(name) CARPSTATS_ADD(name, 1) SYSCTL_VNET_PCPUSTAT(_net_inet_carp, OID_AUTO, stats, struct carpstats, carpstats, "CARP statistics (struct carpstats, netinet/ip_carp.h)"); #define CARP_LOCK_INIT(sc) mtx_init(&(sc)->sc_mtx, "carp_softc", \ NULL, MTX_DEF) #define CARP_LOCK_DESTROY(sc) mtx_destroy(&(sc)->sc_mtx) #define CARP_LOCK_ASSERT(sc) mtx_assert(&(sc)->sc_mtx, MA_OWNED) #define CARP_LOCK(sc) mtx_lock(&(sc)->sc_mtx) #define CARP_UNLOCK(sc) mtx_unlock(&(sc)->sc_mtx) #define CIF_LOCK_INIT(cif) mtx_init(&(cif)->cif_mtx, "carp_if", \ NULL, MTX_DEF) #define CIF_LOCK_DESTROY(cif) mtx_destroy(&(cif)->cif_mtx) #define CIF_LOCK_ASSERT(cif) mtx_assert(&(cif)->cif_mtx, MA_OWNED) #define CIF_LOCK(cif) mtx_lock(&(cif)->cif_mtx) #define CIF_UNLOCK(cif) mtx_unlock(&(cif)->cif_mtx) #define CIF_FREE(cif) do { \ CIF_LOCK(cif); \ if (TAILQ_EMPTY(&(cif)->cif_vrs)) \ carp_free_if(cif); \ else \ CIF_UNLOCK(cif); \ } while (0) #define CARP_LOG(...) do { \ if (V_carp_log > 0) \ log(LOG_INFO, "carp: " __VA_ARGS__); \ } while (0) #define CARP_DEBUG(...) do { \ if (V_carp_log > 1) \ log(LOG_DEBUG, __VA_ARGS__); \ } while (0) #define IFNET_FOREACH_IFA(ifp, ifa) \ CK_STAILQ_FOREACH((ifa), &(ifp)->if_addrhead, ifa_link) \ if ((ifa)->ifa_carp != NULL) #define CARP_FOREACH_IFA(sc, ifa) \ CARP_LOCK_ASSERT(sc); \ for (int _i = 0; \ _i < (sc)->sc_naddrs + (sc)->sc_naddrs6 && \ ((ifa) = sc->sc_ifas[_i]) != NULL; \ ++_i) #define IFNET_FOREACH_CARP(ifp, sc) \ KASSERT(mtx_owned(&ifp->if_carp->cif_mtx) || \ sx_xlocked(&carp_sx), ("cif_vrs not locked")); \ TAILQ_FOREACH((sc), &(ifp)->if_carp->cif_vrs, sc_list) #define DEMOTE_ADVSKEW(sc) \ (((sc)->sc_advskew + V_carp_demotion > CARP_MAXSKEW) ? \ CARP_MAXSKEW : \ (((sc)->sc_advskew + V_carp_demotion < 0) ? \ 0 : ((sc)->sc_advskew + V_carp_demotion))) static void carp_input_c(struct mbuf *, struct carp_header *, sa_family_t, int); static void vrrp_input_c(struct mbuf *, int, sa_family_t, int, int, uint16_t); static struct carp_softc *carp_alloc(struct ifnet *, carp_version_t, int); static void carp_destroy(struct carp_softc *); static struct carp_if *carp_alloc_if(struct ifnet *); static void carp_free_if(struct carp_if *); static void carp_set_state(struct carp_softc *, int, const char* reason); static void carp_sc_state(struct carp_softc *); static void carp_setrun(struct carp_softc *, sa_family_t); static void carp_master_down(void *); static void carp_master_down_locked(struct carp_softc *, const char* reason); static void carp_send_ad_locked(struct carp_softc *); static void vrrp_send_ad_locked(struct carp_softc *); static void carp_addroute(struct carp_softc *); static void carp_ifa_addroute(struct ifaddr *); static void carp_delroute(struct carp_softc *); static void carp_ifa_delroute(struct ifaddr *); static void carp_send_ad_all(void *, int); static void carp_demote_adj(int, char *); static LIST_HEAD(, carp_softc) carp_list = LIST_HEAD_INITIALIZER(carp_list); static struct mtx carp_mtx; static struct sx carp_sx; static struct task carp_sendall_task = TASK_INITIALIZER(0, carp_send_ad_all, NULL); static int carp_is_supported_if(if_t ifp) { if (ifp == NULL) return (ENXIO); switch (ifp->if_type) { case IFT_ETHER: case IFT_L2VLAN: case IFT_BRIDGE: break; default: return (EOPNOTSUPP); } return (0); } static void carp_hmac_prepare(struct carp_softc *sc) { uint8_t version = CARP_VERSION_CARP, type = CARP_ADVERTISEMENT; uint8_t vhid = sc->sc_vhid & 0xff; struct ifaddr *ifa; int i, found; #ifdef INET struct in_addr last, cur, in; #endif #ifdef INET6 struct in6_addr last6, cur6, in6; #endif CARP_LOCK_ASSERT(sc); MPASS(sc->sc_version == CARP_VERSION_CARP); /* Compute ipad from key. */ bzero(sc->sc_pad, sizeof(sc->sc_pad)); bcopy(sc->sc_key, sc->sc_pad, sizeof(sc->sc_key)); for (i = 0; i < sizeof(sc->sc_pad); i++) sc->sc_pad[i] ^= 0x36; /* Precompute first part of inner hash. */ SHA1Init(&sc->sc_sha1); SHA1Update(&sc->sc_sha1, sc->sc_pad, sizeof(sc->sc_pad)); SHA1Update(&sc->sc_sha1, (void *)&version, sizeof(version)); SHA1Update(&sc->sc_sha1, (void *)&type, sizeof(type)); SHA1Update(&sc->sc_sha1, (void *)&vhid, sizeof(vhid)); #ifdef INET cur.s_addr = 0; do { found = 0; last = cur; cur.s_addr = 0xffffffff; CARP_FOREACH_IFA(sc, ifa) { in.s_addr = ifatoia(ifa)->ia_addr.sin_addr.s_addr; if (ifa->ifa_addr->sa_family == AF_INET && ntohl(in.s_addr) > ntohl(last.s_addr) && ntohl(in.s_addr) < ntohl(cur.s_addr)) { cur.s_addr = in.s_addr; found++; } } if (found) SHA1Update(&sc->sc_sha1, (void *)&cur, sizeof(cur)); } while (found); #endif /* INET */ #ifdef INET6 memset(&cur6, 0, sizeof(cur6)); do { found = 0; last6 = cur6; memset(&cur6, 0xff, sizeof(cur6)); CARP_FOREACH_IFA(sc, ifa) { in6 = ifatoia6(ifa)->ia_addr.sin6_addr; if (IN6_IS_SCOPE_EMBED(&in6)) in6.s6_addr16[1] = 0; if (ifa->ifa_addr->sa_family == AF_INET6 && memcmp(&in6, &last6, sizeof(in6)) > 0 && memcmp(&in6, &cur6, sizeof(in6)) < 0) { cur6 = in6; found++; } } if (found) SHA1Update(&sc->sc_sha1, (void *)&cur6, sizeof(cur6)); } while (found); #endif /* INET6 */ /* convert ipad to opad */ for (i = 0; i < sizeof(sc->sc_pad); i++) sc->sc_pad[i] ^= 0x36 ^ 0x5c; } static void carp_hmac_generate(struct carp_softc *sc, uint32_t counter[2], unsigned char md[20]) { SHA1_CTX sha1ctx; CARP_LOCK_ASSERT(sc); /* fetch first half of inner hash */ bcopy(&sc->sc_sha1, &sha1ctx, sizeof(sha1ctx)); SHA1Update(&sha1ctx, (void *)counter, sizeof(sc->sc_counter)); SHA1Final(md, &sha1ctx); /* outer hash */ SHA1Init(&sha1ctx); SHA1Update(&sha1ctx, sc->sc_pad, sizeof(sc->sc_pad)); SHA1Update(&sha1ctx, md, 20); SHA1Final(md, &sha1ctx); } static int carp_hmac_verify(struct carp_softc *sc, uint32_t counter[2], unsigned char md[20]) { unsigned char md2[20]; CARP_LOCK_ASSERT(sc); carp_hmac_generate(sc, counter, md2); return (bcmp(md, md2, sizeof(md2))); } static int vrrp_checksum_verify(struct mbuf *m, int off, int len, uint16_t phdrcksum) { uint16_t cksum; /* * Note that VRRPv3 checksums are different from CARP checksums. * Carp just calculates the checksum over the packet. * VRRPv3 includes the pseudo-header checksum as well. */ cksum = in_cksum_skip(m, off + len, off); cksum -= phdrcksum; return (cksum); } /* * process input packet. * we have rearranged checks order compared to the rfc, * but it seems more efficient this way or not possible otherwise. */ #ifdef INET static int carp_input(struct mbuf **mp, int *offp, int proto) { struct mbuf *m = *mp; struct ip *ip; struct vrrpv3_header *vh; int iplen; int minlen; int totlen; iplen = *offp; *mp = NULL; CARPSTATS_INC(carps_ipackets); if (!V_carp_allow) { m_freem(m); return (IPPROTO_DONE); } /* Ensure we have enough header to figure out the version. */ if (m->m_pkthdr.len < iplen + sizeof(*vh)) { CARPSTATS_INC(carps_badlen); CARP_DEBUG("%s: received len %zd < sizeof(struct vrrpv3_header) " "on %s\n", __func__, m->m_len - sizeof(struct ip), if_name(m->m_pkthdr.rcvif)); m_freem(m); return (IPPROTO_DONE); } if (m->m_len < iplen + sizeof(*vh)) { if ((m = m_pullup(m, iplen + sizeof(*vh))) == NULL) { CARPSTATS_INC(carps_hdrops); CARP_DEBUG("%s():%d: pullup failed\n", __func__, __LINE__); return (IPPROTO_DONE); } } ip = mtod(m, struct ip *); totlen = ntohs(ip->ip_len); vh = (struct vrrpv3_header *)((char *)ip + iplen); switch (vh->vrrp_version) { case CARP_VERSION_CARP: minlen = sizeof(struct carp_header); break; case CARP_VERSION_VRRPv3: minlen = sizeof(struct vrrpv3_header); break; default: CARPSTATS_INC(carps_badver); CARP_DEBUG("%s: unsupported version %d on %s\n", __func__, vh->vrrp_version, if_name(m->m_pkthdr.rcvif)); m_freem(m); return (IPPROTO_DONE); } /* And now check the length again but with the real minimal length. */ if (m->m_pkthdr.len < iplen + minlen) { CARPSTATS_INC(carps_badlen); CARP_DEBUG("%s: received len %zd < %d " "on %s\n", __func__, m->m_len - sizeof(struct ip), iplen + minlen, if_name(m->m_pkthdr.rcvif)); m_freem(m); return (IPPROTO_DONE); } if (m->m_len < iplen + minlen) { if ((m = m_pullup(m, iplen + minlen)) == NULL) { CARPSTATS_INC(carps_hdrops); CARP_DEBUG("%s():%d: pullup failed\n", __func__, __LINE__); return (IPPROTO_DONE); } ip = mtod(m, struct ip *); vh = (struct vrrpv3_header *)((char *)ip + iplen); } switch (vh->vrrp_version) { case CARP_VERSION_CARP: { struct carp_header *ch; /* verify the CARP checksum */ if (in_cksum_skip(m, totlen, iplen)) { CARPSTATS_INC(carps_badsum); CARP_DEBUG("%s: checksum failed on %s\n", __func__, if_name(m->m_pkthdr.rcvif)); m_freem(m); break; } ch = (struct carp_header *)((char *)ip + iplen); carp_input_c(m, ch, AF_INET, ip->ip_ttl); break; } case CARP_VERSION_VRRPv3: { uint16_t phdrcksum; phdrcksum = in_pseudo(ip->ip_src.s_addr, ip->ip_dst.s_addr, htonl((u_short)(totlen - iplen) + ip->ip_p)); vrrp_input_c(m, iplen, AF_INET, ip->ip_ttl, totlen - iplen, phdrcksum); break; } default: KASSERT(false, ("Unsupported version %d", vh->vrrp_version)); } return (IPPROTO_DONE); } #endif #ifdef INET6 static int carp6_input(struct mbuf **mp, int *offp, int proto) { struct mbuf *m = *mp; struct ip6_hdr *ip6 = mtod(m, struct ip6_hdr *); struct vrrpv3_header *vh; u_int len, minlen; CARPSTATS_INC(carps_ipackets6); if (!V_carp_allow) { m_freem(m); return (IPPROTO_DONE); } /* check if received on a valid carp interface */ if (m->m_pkthdr.rcvif->if_carp == NULL) { CARPSTATS_INC(carps_badif); CARP_DEBUG("%s: packet received on non-carp interface: %s\n", __func__, if_name(m->m_pkthdr.rcvif)); m_freem(m); return (IPPROTO_DONE); } if (m->m_len < *offp + sizeof(*vh)) { len = m->m_len; m = m_pullup(m, *offp + sizeof(*vh)); if (m == NULL) { CARPSTATS_INC(carps_badlen); CARP_DEBUG("%s: packet size %u too small\n", __func__, len); return (IPPROTO_DONE); } ip6 = mtod(m, struct ip6_hdr *); } vh = (struct vrrpv3_header *)(mtod(m, char *) + *offp); switch (vh->vrrp_version) { case CARP_VERSION_CARP: minlen = sizeof(struct carp_header); break; case CARP_VERSION_VRRPv3: minlen = sizeof(struct vrrpv3_header); break; default: CARPSTATS_INC(carps_badver); CARP_DEBUG("%s: unsupported version %d on %s\n", __func__, vh->vrrp_version, if_name(m->m_pkthdr.rcvif)); m_freem(m); return (IPPROTO_DONE); } /* And now check the length again but with the real minimal length. */ if (m->m_pkthdr.len < sizeof(*ip6) + minlen) { CARPSTATS_INC(carps_badlen); CARP_DEBUG("%s: received len %zd < %zd " "on %s\n", __func__, m->m_len - sizeof(struct ip), sizeof(*ip6) + minlen, if_name(m->m_pkthdr.rcvif)); m_freem(m); return (IPPROTO_DONE); } if (m->m_len < sizeof(*ip6) + minlen) { if ((m = m_pullup(m, sizeof(*ip6) + minlen)) == NULL) { CARPSTATS_INC(carps_hdrops); CARP_DEBUG("%s():%d: pullup failed\n", __func__, __LINE__); return (IPPROTO_DONE); } ip6 = mtod(m, struct ip6_hdr *); vh = (struct vrrpv3_header *)mtodo(m, sizeof(*ip6)); } switch (vh->vrrp_version) { case CARP_VERSION_CARP: { struct carp_header *ch; /* verify the CARP checksum */ if (in_cksum_skip(m, *offp + sizeof(struct carp_header), *offp)) { CARPSTATS_INC(carps_badsum); CARP_DEBUG("%s: checksum failed, on %s\n", __func__, if_name(m->m_pkthdr.rcvif)); m_freem(m); break; } ch = (struct carp_header *)((char *)ip6 + sizeof(*ip6)); carp_input_c(m, ch, AF_INET6, ip6->ip6_hlim); break; } case CARP_VERSION_VRRPv3: { uint16_t phdrcksum; phdrcksum = in6_cksum_pseudo(ip6, ntohs(ip6->ip6_plen), ip6->ip6_nxt, 0); vrrp_input_c(m, sizeof(*ip6), AF_INET6, ip6->ip6_hlim, ntohs(ip6->ip6_plen), phdrcksum); break; } default: KASSERT(false, ("Unsupported version %d", vh->vrrp_version)); } return (IPPROTO_DONE); } #endif /* INET6 */ /* * This routine should not be necessary at all, but some switches * (VMWare ESX vswitches) can echo our own packets back at us, * and we must ignore them or they will cause us to drop out of * MASTER mode. * * We cannot catch all cases of network loops. Instead, what we * do here is catch any packet that arrives with a carp header * with a VHID of 0, that comes from an address that is our own. * These packets are by definition "from us" (even if they are from * a misconfigured host that is pretending to be us). * * The VHID test is outside this mini-function. */ static int carp_source_is_self(const struct mbuf *m, struct ifaddr *ifa, sa_family_t af) { #ifdef INET struct ip *ip4; struct in_addr in4; #endif #ifdef INET6 struct ip6_hdr *ip6; struct in6_addr in6; #endif switch (af) { #ifdef INET case AF_INET: ip4 = mtod(m, struct ip *); in4 = ifatoia(ifa)->ia_addr.sin_addr; return (in4.s_addr == ip4->ip_src.s_addr); #endif #ifdef INET6 case AF_INET6: ip6 = mtod(m, struct ip6_hdr *); in6 = ifatoia6(ifa)->ia_addr.sin6_addr; return (memcmp(&in6, &ip6->ip6_src, sizeof(in6)) == 0); #endif default: break; } return (0); } static struct ifaddr * carp_find_ifa(const struct mbuf *m, sa_family_t af, uint8_t vhid) { struct ifnet *ifp = m->m_pkthdr.rcvif; struct ifaddr *ifa, *match; int error; NET_EPOCH_ASSERT(); /* * Verify that the VHID is valid on the receiving interface. * * There should be just one match. If there are none * the VHID is not valid and we drop the packet. If * there are multiple VHID matches, take just the first * one, for compatibility with previous code. While we're * scanning, check for obvious loops in the network topology * (these should never happen, and as noted above, we may * miss real loops; this is just a double-check). */ error = 0; match = NULL; IFNET_FOREACH_IFA(ifp, ifa) { if (match == NULL && ifa->ifa_carp != NULL && ifa->ifa_addr->sa_family == af && ifa->ifa_carp->sc_vhid == vhid) match = ifa; if (vhid == 0 && carp_source_is_self(m, ifa, af)) error = ELOOP; } ifa = error ? NULL : match; if (ifa != NULL) ifa_ref(ifa); if (ifa == NULL) { if (error == ELOOP) { CARP_DEBUG("dropping looped packet on interface %s\n", if_name(ifp)); CARPSTATS_INC(carps_badif); /* ??? */ } else { CARPSTATS_INC(carps_badvhid); } } return (ifa); } static void carp_input_c(struct mbuf *m, struct carp_header *ch, sa_family_t af, int ttl) { struct ifnet *ifp = m->m_pkthdr.rcvif; struct ifaddr *ifa; struct carp_softc *sc; uint64_t tmp_counter; struct timeval sc_tv, ch_tv; bool multicast = false; NET_EPOCH_ASSERT(); MPASS(ch->carp_version == CARP_VERSION_CARP); ifa = carp_find_ifa(m, af, ch->carp_vhid); if (ifa == NULL) { m_freem(m); return; } sc = ifa->ifa_carp; CARP_LOCK(sc); /* verify the CARP version. */ if (sc->sc_version != CARP_VERSION_CARP) { CARP_UNLOCK(sc); CARPSTATS_INC(carps_badver); CARP_DEBUG("%s: invalid version %d\n", if_name(ifp), ch->carp_version); ifa_free(ifa); m_freem(m); return; } if (ifa->ifa_addr->sa_family == AF_INET) { multicast = IN_MULTICAST(ntohl(sc->sc_carpaddr.s_addr)); } else { multicast = IN6_IS_ADDR_MULTICAST(&sc->sc_carpaddr6); } ifa_free(ifa); /* verify that the IP TTL is 255, but only if we're not in unicast mode. */ if (multicast && ttl != CARP_DFLTTL) { CARPSTATS_INC(carps_badttl); CARP_DEBUG("%s: received ttl %d != 255 on %s\n", __func__, ttl, if_name(m->m_pkthdr.rcvif)); goto out; } if (carp_hmac_verify(sc, ch->carp_counter, ch->carp_md)) { CARPSTATS_INC(carps_badauth); CARP_DEBUG("%s: incorrect hash for VHID %u@%s\n", __func__, sc->sc_vhid, if_name(ifp)); goto out; } tmp_counter = ntohl(ch->carp_counter[0]); tmp_counter = tmp_counter<<32; tmp_counter += ntohl(ch->carp_counter[1]); /* XXX Replay protection goes here */ sc->sc_init_counter = false; sc->sc_counter = tmp_counter; sc_tv.tv_sec = sc->sc_advbase; sc_tv.tv_usec = DEMOTE_ADVSKEW(sc) * 1000000 / 256; ch_tv.tv_sec = ch->carp_advbase; ch_tv.tv_usec = ch->carp_advskew * 1000000 / 256; switch (sc->sc_state) { case INIT: break; case MASTER: /* * If we receive an advertisement from a master who's going to * be more frequent than us, go into BACKUP state. */ if (timevalcmp(&sc_tv, &ch_tv, >) || timevalcmp(&sc_tv, &ch_tv, ==)) { callout_stop(&sc->sc_ad_tmo); carp_set_state(sc, BACKUP, "more frequent advertisement received"); carp_setrun(sc, 0); carp_delroute(sc); } break; case BACKUP: /* * If we're pre-empting masters who advertise slower than us, * and this one claims to be slower, treat him as down. */ if (V_carp_preempt && timevalcmp(&sc_tv, &ch_tv, <)) { carp_master_down_locked(sc, "preempting a slower master"); break; } /* * If the master is going to advertise at such a low frequency * that he's guaranteed to time out, we'd might as well just * treat him as timed out now. */ sc_tv.tv_sec = sc->sc_advbase * 3; if (timevalcmp(&sc_tv, &ch_tv, <)) { carp_master_down_locked(sc, "master will time out"); break; } /* * Otherwise, we reset the counter and wait for the next * advertisement. */ carp_setrun(sc, af); break; } out: CARP_UNLOCK(sc); m_freem(m); } static void vrrp_input_c(struct mbuf *m, int off, sa_family_t af, int ttl, int len, uint16_t phdrcksum) { struct vrrpv3_header *vh = mtodo(m, off); struct ifnet *ifp = m->m_pkthdr.rcvif; struct ifaddr *ifa; struct carp_softc *sc; NET_EPOCH_ASSERT(); MPASS(vh->vrrp_version == CARP_VERSION_VRRPv3); ifa = carp_find_ifa(m, af, vh->vrrp_vrtid); if (ifa == NULL) { m_freem(m); return; } sc = ifa->ifa_carp; CARP_LOCK(sc); ifa_free(ifa); /* verify the CARP version. */ if (sc->sc_version != CARP_VERSION_VRRPv3) { CARP_UNLOCK(sc); CARPSTATS_INC(carps_badver); CARP_DEBUG("%s: invalid version %d\n", if_name(ifp), vh->vrrp_version); m_freem(m); return; } /* verify that the IP TTL is 255. */ if (ttl != CARP_DFLTTL) { CARPSTATS_INC(carps_badttl); CARP_DEBUG("%s: received ttl %d != 255 on %s\n", __func__, ttl, if_name(m->m_pkthdr.rcvif)); goto out; } if (vrrp_checksum_verify(m, off, len, phdrcksum)) { CARPSTATS_INC(carps_badsum); CARP_DEBUG("%s: incorrect checksum for VRID %u@%s\n", __func__, sc->sc_vhid, if_name(ifp)); goto out; } /* RFC9568, 7.1 Receiving VRRP packets. */ if (sc->sc_vrrp_prio == 255) { CARP_DEBUG("%s: our priority is 255. Ignore peer announcement.\n", __func__); goto out; } /* XXX TODO Check IP address payload. */ sc->sc_vrrp_master_inter = ntohs(vh->vrrp_max_adver_int); switch (sc->sc_state) { case INIT: break; case MASTER: /* * If we receive an advertisement from a master who's going to * be more frequent than us, go into BACKUP state. * Same if the peer has a higher priority than us. */ if (ntohs(vh->vrrp_max_adver_int) < sc->sc_vrrp_adv_inter || vh->vrrp_priority > sc->sc_vrrp_prio) { callout_stop(&sc->sc_ad_tmo); carp_set_state(sc, BACKUP, "more frequent advertisement received"); carp_setrun(sc, 0); carp_delroute(sc); } break; case BACKUP: /* * If we're pre-empting masters who advertise slower than us, * and this one claims to be slower, treat him as down. */ if (V_carp_preempt && (ntohs(vh->vrrp_max_adver_int) > sc->sc_vrrp_adv_inter || vh->vrrp_priority < sc->sc_vrrp_prio)) { carp_master_down_locked(sc, "preempting a slower master"); break; } /* * Otherwise, we reset the counter and wait for the next * advertisement. */ carp_setrun(sc, af); break; } out: CARP_UNLOCK(sc); m_freem(m); } static int carp_tag(struct carp_softc *sc, struct mbuf *m) { struct m_tag *mtag; /* Tag packet for carp_output */ if ((mtag = m_tag_get(PACKET_TAG_CARP, sizeof(sc->sc_vhid), M_NOWAIT)) == NULL) { m_freem(m); CARPSTATS_INC(carps_onomem); return (ENOMEM); } bcopy(&sc->sc_vhid, mtag + 1, sizeof(sc->sc_vhid)); m_tag_prepend(m, mtag); return (0); } static void carp_prepare_ad(struct mbuf *m, struct carp_softc *sc, struct carp_header *ch) { MPASS(sc->sc_version == CARP_VERSION_CARP); if (sc->sc_init_counter) { /* this could also be seconds since unix epoch */ sc->sc_counter = arc4random(); sc->sc_counter = sc->sc_counter << 32; sc->sc_counter += arc4random(); } else sc->sc_counter++; ch->carp_counter[0] = htonl((sc->sc_counter>>32)&0xffffffff); ch->carp_counter[1] = htonl(sc->sc_counter&0xffffffff); carp_hmac_generate(sc, ch->carp_counter, ch->carp_md); } static inline void send_ad_locked(struct carp_softc *sc) { switch (sc->sc_version) { case CARP_VERSION_CARP: carp_send_ad_locked(sc); break; case CARP_VERSION_VRRPv3: vrrp_send_ad_locked(sc); break; } } /* * To avoid LORs and possible recursions this function shouldn't * be called directly, but scheduled via taskqueue. */ static void carp_send_ad_all(void *ctx __unused, int pending __unused) { struct carp_softc *sc; struct epoch_tracker et; NET_EPOCH_ENTER(et); mtx_lock(&carp_mtx); LIST_FOREACH(sc, &carp_list, sc_next) if (sc->sc_state == MASTER) { CARP_LOCK(sc); CURVNET_SET(sc->sc_carpdev->if_vnet); send_ad_locked(sc); CURVNET_RESTORE(); CARP_UNLOCK(sc); } mtx_unlock(&carp_mtx); NET_EPOCH_EXIT(et); } /* Send a periodic advertisement, executed in callout context. */ static void carp_callout(void *v) { struct carp_softc *sc = v; struct epoch_tracker et; NET_EPOCH_ENTER(et); CARP_LOCK_ASSERT(sc); CURVNET_SET(sc->sc_carpdev->if_vnet); send_ad_locked(sc); CURVNET_RESTORE(); CARP_UNLOCK(sc); NET_EPOCH_EXIT(et); } static void carp_send_ad_error(struct carp_softc *sc, int error) { /* * We track errors and successful sends with this logic: * - Any error resets success counter to 0. * - MAX_ERRORS triggers demotion. * - MIN_SUCCESS successes resets error counter to 0. * - MIN_SUCCESS reverts demotion, if it was triggered before. */ if (error) { if (sc->sc_sendad_errors < INT_MAX) sc->sc_sendad_errors++; if (sc->sc_sendad_errors == CARP_SENDAD_MAX_ERRORS) { static const char fmt[] = "send error %d on %s"; char msg[sizeof(fmt) + IFNAMSIZ]; sprintf(msg, fmt, error, if_name(sc->sc_carpdev)); carp_demote_adj(V_carp_senderr_adj, msg); } sc->sc_sendad_success = 0; } else if (sc->sc_sendad_errors > 0) { if (++sc->sc_sendad_success >= CARP_SENDAD_MIN_SUCCESS) { if (sc->sc_sendad_errors >= CARP_SENDAD_MAX_ERRORS) { static const char fmt[] = "send ok on %s"; char msg[sizeof(fmt) + IFNAMSIZ]; sprintf(msg, fmt, if_name(sc->sc_carpdev)); carp_demote_adj(-V_carp_senderr_adj, msg); } sc->sc_sendad_errors = 0; } } } /* * Pick the best ifaddr on the given ifp for sending CARP * advertisements. * * "Best" here is defined by ifa_preferred(). This function is much * much like ifaof_ifpforaddr() except that we just use ifa_preferred(). * * (This could be simplified to return the actual address, except that * it has a different format in AF_INET and AF_INET6.) */ static struct ifaddr * carp_best_ifa(int af, struct ifnet *ifp) { struct ifaddr *ifa, *best; NET_EPOCH_ASSERT(); if (af >= AF_MAX) return (NULL); best = NULL; CK_STAILQ_FOREACH(ifa, &ifp->if_addrhead, ifa_link) { if (ifa->ifa_addr->sa_family == af && (best == NULL || ifa_preferred(best, ifa))) best = ifa; } if (best != NULL) ifa_ref(best); return (best); } static void carp_send_ad_locked(struct carp_softc *sc) { struct carp_header ch; struct timeval tv; struct ifaddr *ifa; struct carp_header *ch_ptr; struct mbuf *m; int len, advskew; NET_EPOCH_ASSERT(); CARP_LOCK_ASSERT(sc); MPASS(sc->sc_version == CARP_VERSION_CARP); advskew = DEMOTE_ADVSKEW(sc); tv.tv_sec = sc->sc_advbase; tv.tv_usec = advskew * 1000000 / 256; ch.carp_version = CARP_VERSION_CARP; ch.carp_type = CARP_ADVERTISEMENT; ch.carp_vhid = sc->sc_vhid; ch.carp_advbase = sc->sc_advbase; ch.carp_advskew = advskew; ch.carp_authlen = 7; /* XXX DEFINE */ ch.carp_pad1 = 0; /* must be zero */ ch.carp_cksum = 0; /* XXXGL: OpenBSD picks first ifaddr with needed family. */ #ifdef INET if (sc->sc_naddrs) { struct ip *ip; m = m_gethdr(M_NOWAIT, MT_DATA); if (m == NULL) { CARPSTATS_INC(carps_onomem); goto resched; } len = sizeof(*ip) + sizeof(ch); m->m_pkthdr.len = len; m->m_pkthdr.rcvif = NULL; m->m_len = len; M_ALIGN(m, m->m_len); if (IN_MULTICAST(ntohl(sc->sc_carpaddr.s_addr))) m->m_flags |= M_MCAST; ip = mtod(m, struct ip *); ip->ip_v = IPVERSION; ip->ip_hl = sizeof(*ip) >> 2; ip->ip_tos = V_carp_dscp << IPTOS_DSCP_OFFSET; ip->ip_len = htons(len); ip->ip_off = htons(IP_DF); ip->ip_ttl = CARP_DFLTTL; ip->ip_p = IPPROTO_CARP; ip->ip_sum = 0; ip_fillid(ip, V_ip_random_id); ifa = carp_best_ifa(AF_INET, sc->sc_carpdev); if (ifa != NULL) { ip->ip_src.s_addr = ifatoia(ifa)->ia_addr.sin_addr.s_addr; ifa_free(ifa); } else ip->ip_src.s_addr = 0; ip->ip_dst = sc->sc_carpaddr; ch_ptr = (struct carp_header *)(&ip[1]); bcopy(&ch, ch_ptr, sizeof(ch)); carp_prepare_ad(m, sc, ch_ptr); if (IN_MULTICAST(ntohl(sc->sc_carpaddr.s_addr)) && carp_tag(sc, m) != 0) goto resched; m->m_data += sizeof(*ip); ch_ptr->carp_cksum = in_cksum(m, len - sizeof(*ip)); m->m_data -= sizeof(*ip); CARPSTATS_INC(carps_opackets); carp_send_ad_error(sc, ip_output(m, NULL, NULL, IP_RAWOUTPUT, &sc->sc_carpdev->if_carp->cif_imo, NULL)); } #endif /* INET */ #ifdef INET6 if (sc->sc_naddrs6) { struct ip6_hdr *ip6; m = m_gethdr(M_NOWAIT, MT_DATA); if (m == NULL) { CARPSTATS_INC(carps_onomem); goto resched; } len = sizeof(*ip6) + sizeof(ch); m->m_pkthdr.len = len; m->m_pkthdr.rcvif = NULL; m->m_len = len; M_ALIGN(m, m->m_len); ip6 = mtod(m, struct ip6_hdr *); bzero(ip6, sizeof(*ip6)); ip6->ip6_vfc |= IPV6_VERSION; /* Traffic class isn't defined in ip6 struct instead * it gets offset into flowid field */ ip6->ip6_flow |= htonl(V_carp_dscp << (IPV6_FLOWLABEL_LEN + IPTOS_DSCP_OFFSET)); ip6->ip6_hlim = CARP_DFLTTL; ip6->ip6_nxt = IPPROTO_CARP; /* set the source address */ ifa = carp_best_ifa(AF_INET6, sc->sc_carpdev); if (ifa != NULL) { bcopy(IFA_IN6(ifa), &ip6->ip6_src, sizeof(struct in6_addr)); ifa_free(ifa); } else /* This should never happen with IPv6. */ bzero(&ip6->ip6_src, sizeof(struct in6_addr)); /* Set the multicast destination. */ memcpy(&ip6->ip6_dst, &sc->sc_carpaddr6, sizeof(ip6->ip6_dst)); if (IN6_IS_ADDR_MULTICAST(&ip6->ip6_dst) || IN6_IS_ADDR_LINKLOCAL(&ip6->ip6_dst)) { if (in6_setscope(&ip6->ip6_dst, sc->sc_carpdev, NULL) != 0) { m_freem(m); CARP_DEBUG("%s: in6_setscope failed\n", __func__); goto resched; } } ch_ptr = (struct carp_header *)(&ip6[1]); bcopy(&ch, ch_ptr, sizeof(ch)); carp_prepare_ad(m, sc, ch_ptr); if (IN6_IS_ADDR_MULTICAST(&sc->sc_carpaddr6) && carp_tag(sc, m) != 0) goto resched; m->m_data += sizeof(*ip6); ch_ptr->carp_cksum = in_cksum(m, len - sizeof(*ip6)); m->m_data -= sizeof(*ip6); CARPSTATS_INC(carps_opackets6); carp_send_ad_error(sc, ip6_output(m, NULL, NULL, 0, &sc->sc_carpdev->if_carp->cif_im6o, NULL, NULL)); } #endif /* INET6 */ resched: callout_reset(&sc->sc_ad_tmo, tvtohz(&tv), carp_callout, sc); } static void vrrp_send_ad_locked(struct carp_softc *sc) { struct vrrpv3_header *vh_ptr; struct ifaddr *ifa; struct mbuf *m; int len; struct vrrpv3_header vh = { .vrrp_version = CARP_VERSION_VRRPv3, .vrrp_type = VRRP_TYPE_ADVERTISEMENT, .vrrp_vrtid = sc->sc_vhid, .vrrp_priority = sc->sc_vrrp_prio, .vrrp_count_addr = 0, .vrrp_max_adver_int = htons(sc->sc_vrrp_adv_inter), .vrrp_checksum = 0, }; NET_EPOCH_ASSERT(); CARP_LOCK_ASSERT(sc); MPASS(sc->sc_version == CARP_VERSION_VRRPv3); #ifdef INET if (sc->sc_naddrs) { struct ip *ip; m = m_gethdr(M_NOWAIT, MT_DATA); if (m == NULL) { CARPSTATS_INC(carps_onomem); goto resched; } len = sizeof(*ip) + sizeof(vh); m->m_pkthdr.len = len; m->m_pkthdr.rcvif = NULL; m->m_len = len; M_ALIGN(m, m->m_len); m->m_flags |= M_MCAST; ip = mtod(m, struct ip *); ip->ip_v = IPVERSION; ip->ip_hl = sizeof(*ip) >> 2; ip->ip_tos = V_carp_dscp << IPTOS_DSCP_OFFSET; ip->ip_off = htons(IP_DF); ip->ip_ttl = CARP_DFLTTL; ip->ip_p = IPPROTO_CARP; ip->ip_sum = 0; ip_fillid(ip, V_ip_random_id); ifa = carp_best_ifa(AF_INET, sc->sc_carpdev); if (ifa != NULL) { ip->ip_src.s_addr = ifatoia(ifa)->ia_addr.sin_addr.s_addr; ifa_free(ifa); } else ip->ip_src.s_addr = 0; ip->ip_dst.s_addr = htonl(INADDR_CARP_GROUP); /* Include the IP addresses in the announcement. */ for (int i = 0; i < (sc->sc_naddrs + sc->sc_naddrs6); i++) { struct sockaddr_in *in; MPASS(sc->sc_ifas[i] != NULL); if (sc->sc_ifas[i]->ifa_addr->sa_family != AF_INET) continue; in = (struct sockaddr_in *)sc->sc_ifas[i]->ifa_addr; if (m_append(m, sizeof(in->sin_addr), (caddr_t)&in->sin_addr) != 1) { m_freem(m); goto resched; } vh.vrrp_count_addr++; len += sizeof(in->sin_addr); } ip->ip_len = htons(len); vh_ptr = (struct vrrpv3_header *)mtodo(m, sizeof(*ip)); bcopy(&vh, vh_ptr, sizeof(vh)); vh_ptr->vrrp_checksum = in_pseudo(ip->ip_src.s_addr, ip->ip_dst.s_addr, htonl((uint16_t)(len - sizeof(*ip)) + ip->ip_p)); vh_ptr->vrrp_checksum = in_cksum_skip(m, len, sizeof(*ip)); if (carp_tag(sc, m)) goto resched; CARPSTATS_INC(carps_opackets); carp_send_ad_error(sc, ip_output(m, NULL, NULL, IP_RAWOUTPUT, &sc->sc_carpdev->if_carp->cif_imo, NULL)); } #endif #ifdef INET6 if (sc->sc_naddrs6) { struct ip6_hdr *ip6; m = m_gethdr(M_NOWAIT, MT_DATA); if (m == NULL) { CARPSTATS_INC(carps_onomem); goto resched; } len = sizeof(*ip6) + sizeof(vh); m->m_pkthdr.len = len; m->m_pkthdr.rcvif = NULL; m->m_len = len; M_ALIGN(m, m->m_len); m->m_flags |= M_MCAST; ip6 = mtod(m, struct ip6_hdr *); bzero(ip6, sizeof(*ip6)); ip6->ip6_vfc |= IPV6_VERSION; /* Traffic class isn't defined in ip6 struct instead * it gets offset into flowid field */ ip6->ip6_flow |= htonl(V_carp_dscp << (IPV6_FLOWLABEL_LEN + IPTOS_DSCP_OFFSET)); ip6->ip6_hlim = CARP_DFLTTL; ip6->ip6_nxt = IPPROTO_CARP; /* set the source address */ ifa = carp_best_ifa(AF_INET6, sc->sc_carpdev); if (ifa != NULL) { bcopy(IFA_IN6(ifa), &ip6->ip6_src, sizeof(struct in6_addr)); ifa_free(ifa); } else /* This should never happen with IPv6. */ bzero(&ip6->ip6_src, sizeof(struct in6_addr)); /* Set the multicast destination. */ bzero(&ip6->ip6_dst, sizeof(ip6->ip6_dst)); ip6->ip6_dst.s6_addr16[0] = IPV6_ADDR_INT16_MLL; ip6->ip6_dst.s6_addr8[15] = 0x12; /* Include the IP addresses in the announcement. */ len = sizeof(vh); for (int i = 0; i < (sc->sc_naddrs + sc->sc_naddrs6); i++) { struct sockaddr_in6 *in6; MPASS(sc->sc_ifas[i] != NULL); if (sc->sc_ifas[i]->ifa_addr->sa_family != AF_INET6) continue; in6 = (struct sockaddr_in6 *)sc->sc_ifas[i]->ifa_addr; if (m_append(m, sizeof(in6->sin6_addr), (char *)&in6->sin6_addr) != 1) { m_freem(m); goto resched; } vh.vrrp_count_addr++; len += sizeof(in6->sin6_addr); } ip6->ip6_plen = htonl(len); vh_ptr = (struct vrrpv3_header *)mtodo(m, sizeof(*ip6)); bcopy(&vh, vh_ptr, sizeof(vh)); vh_ptr->vrrp_checksum = in6_cksum_pseudo(ip6, len, ip6->ip6_nxt, 0); vh_ptr->vrrp_checksum = in_cksum_skip(m, len + sizeof(*ip6), sizeof(*ip6)); if (in6_setscope(&ip6->ip6_dst, sc->sc_carpdev, NULL) != 0) { m_freem(m); CARP_DEBUG("%s: in6_setscope failed\n", __func__); goto resched; } if (carp_tag(sc, m)) goto resched; CARPSTATS_INC(carps_opackets6); carp_send_ad_error(sc, ip6_output(m, NULL, NULL, 0, &sc->sc_carpdev->if_carp->cif_im6o, NULL, NULL)); } #endif resched: callout_reset(&sc->sc_ad_tmo, sc->sc_vrrp_adv_inter * hz / 100, carp_callout, sc); } static void carp_addroute(struct carp_softc *sc) { struct ifaddr *ifa; CARP_FOREACH_IFA(sc, ifa) carp_ifa_addroute(ifa); } static void carp_ifa_addroute(struct ifaddr *ifa) { switch (ifa->ifa_addr->sa_family) { #ifdef INET case AF_INET: in_addprefix(ifatoia(ifa)); ifa_add_loopback_route(ifa, (struct sockaddr *)&ifatoia(ifa)->ia_addr); break; #endif #ifdef INET6 case AF_INET6: ifa_add_loopback_route(ifa, (struct sockaddr *)&ifatoia6(ifa)->ia_addr); nd6_add_ifa_lle(ifatoia6(ifa)); break; #endif } } static void carp_delroute(struct carp_softc *sc) { struct ifaddr *ifa; CARP_FOREACH_IFA(sc, ifa) carp_ifa_delroute(ifa); } static void carp_ifa_delroute(struct ifaddr *ifa) { switch (ifa->ifa_addr->sa_family) { #ifdef INET case AF_INET: ifa_del_loopback_route(ifa, (struct sockaddr *)&ifatoia(ifa)->ia_addr); in_scrubprefix(ifatoia(ifa), LLE_STATIC); break; #endif #ifdef INET6 case AF_INET6: ifa_del_loopback_route(ifa, (struct sockaddr *)&ifatoia6(ifa)->ia_addr); nd6_rem_ifa_lle(ifatoia6(ifa), 1); break; #endif } } int carp_master(struct ifaddr *ifa) { struct carp_softc *sc = ifa->ifa_carp; return (sc->sc_state == MASTER); } #ifdef INET /* * Broadcast a gratuitous ARP request containing * the virtual router MAC address for each IP address * associated with the virtual router. */ static void carp_send_arp(struct carp_softc *sc) { struct ifaddr *ifa; struct in_addr addr; NET_EPOCH_ASSERT(); CARP_FOREACH_IFA(sc, ifa) { if (ifa->ifa_addr->sa_family != AF_INET) continue; addr = ((struct sockaddr_in *)ifa->ifa_addr)->sin_addr; arp_announce_ifaddr(sc->sc_carpdev, addr, sc->sc_addr); } } int carp_iamatch(struct ifaddr *ifa, uint8_t **enaddr) { struct carp_softc *sc = ifa->ifa_carp; if (sc->sc_state == MASTER) { *enaddr = sc->sc_addr; return (1); } return (0); } #endif #ifdef INET6 static void carp_send_na(struct carp_softc *sc) { - static struct in6_addr mcast = IN6ADDR_LINKLOCAL_ALLNODES_INIT; struct ifaddr *ifa; - struct in6_addr *in6; + int flags; + /* + * Sending Unsolicited Neighbor Advertisements + * + * If the node is a router, we MUST set the Router flag to one. + * We set Override flag to one and send link-layer address option, + * thus neighboring nodes will install the new link-layer address. + */ + flags = ND_NA_FLAG_OVERRIDE; + if (V_ip6_forwarding) + flags |= ND_NA_FLAG_ROUTER; CARP_FOREACH_IFA(sc, ifa) { if (ifa->ifa_addr->sa_family != AF_INET6) continue; - - in6 = IFA_IN6(ifa); - nd6_na_output(sc->sc_carpdev, &mcast, in6, - ND_NA_FLAG_OVERRIDE, 1, NULL); - DELAY(1000); /* XXX */ + /* + * We use unspecified address as destination here to avoid + * scope initialization for each call. + * nd6_na_output() will use all nodes multicast address if + * destinaion address is unspecified. + */ + nd6_na_output(sc->sc_carpdev, &in6addr_any, IFA_IN6(ifa), + flags, ND6_NA_OPT_LLA | ND6_NA_CARP_MASTER, NULL); + DELAY(1000); /* RetransTimer */ } } /* * Returns ifa in case it's a carp address and it is MASTER, or if the address * matches and is not a carp address. Returns NULL otherwise. */ struct ifaddr * carp_iamatch6(struct ifnet *ifp, struct in6_addr *taddr) { struct ifaddr *ifa; NET_EPOCH_ASSERT(); ifa = NULL; CK_STAILQ_FOREACH(ifa, &ifp->if_addrhead, ifa_link) { if (ifa->ifa_addr->sa_family != AF_INET6) continue; if (!IN6_ARE_ADDR_EQUAL(taddr, IFA_IN6(ifa))) continue; if (ifa->ifa_carp && ifa->ifa_carp->sc_state != MASTER) ifa = NULL; else ifa_ref(ifa); break; } return (ifa); } char * carp_macmatch6(struct ifnet *ifp, struct mbuf *m, const struct in6_addr *taddr) { struct ifaddr *ifa; char *mac = NULL; NET_EPOCH_ASSERT(); IFNET_FOREACH_IFA(ifp, ifa) if (ifa->ifa_addr->sa_family == AF_INET6 && IN6_ARE_ADDR_EQUAL(taddr, IFA_IN6(ifa))) { struct carp_softc *sc = ifa->ifa_carp; struct m_tag *mtag; mtag = m_tag_get(PACKET_TAG_CARP, sizeof(sc->sc_vhid) + sizeof(sc->sc_addr), M_NOWAIT); if (mtag == NULL) { CARPSTATS_INC(carps_onomem); break; } /* carp_output expects sc_vhid first. */ bcopy(&sc->sc_vhid, mtag + 1, sizeof(sc->sc_vhid)); /* * Save sc_addr into mtag data after sc_vhid to avoid * possible access to destroyed softc. */ mac = (char *)(mtag + 1) + sizeof(sc->sc_vhid); bcopy(sc->sc_addr, mac, sizeof(sc->sc_addr)); m_tag_prepend(m, mtag); break; } return (mac); } #endif /* INET6 */ int carp_forus(struct ifnet *ifp, u_char *dhost) { struct carp_softc *sc; uint8_t *ena = dhost; if (ena[0] || ena[1] || ena[2] != 0x5e || ena[3] || ena[4] != 1) return (0); CIF_LOCK(ifp->if_carp); IFNET_FOREACH_CARP(ifp, sc) { /* * CARP_LOCK() is not here, since would protect nothing, but * cause deadlock with if_bridge, calling this under its lock. */ if (sc->sc_state == MASTER && !bcmp(dhost, sc->sc_addr, ETHER_ADDR_LEN)) { CIF_UNLOCK(ifp->if_carp); return (1); } } CIF_UNLOCK(ifp->if_carp); return (0); } /* Master down timeout event, executed in callout context. */ static void carp_master_down(void *v) { struct carp_softc *sc = v; struct epoch_tracker et; NET_EPOCH_ENTER(et); CARP_LOCK_ASSERT(sc); CURVNET_SET(sc->sc_carpdev->if_vnet); if (sc->sc_state == BACKUP) { carp_master_down_locked(sc, "master timed out"); } CURVNET_RESTORE(); CARP_UNLOCK(sc); NET_EPOCH_EXIT(et); } static void carp_master_down_locked(struct carp_softc *sc, const char *reason) { NET_EPOCH_ASSERT(); CARP_LOCK_ASSERT(sc); switch (sc->sc_state) { case BACKUP: carp_set_state(sc, MASTER, reason); send_ad_locked(sc); #ifdef INET carp_send_arp(sc); #endif #ifdef INET6 carp_send_na(sc); #endif carp_setrun(sc, 0); carp_addroute(sc); break; case INIT: case MASTER: #ifdef INVARIANTS panic("carp: VHID %u@%s: master_down event in %s state\n", sc->sc_vhid, if_name(sc->sc_carpdev), sc->sc_state ? "MASTER" : "INIT"); #endif break; } } /* * When in backup state, af indicates whether to reset the master down timer * for v4 or v6. If it's set to zero, reset the ones which are already pending. */ static void carp_setrun(struct carp_softc *sc, sa_family_t af) { struct timeval tv; int timeout; CARP_LOCK_ASSERT(sc); if ((sc->sc_carpdev->if_flags & IFF_UP) == 0 || sc->sc_carpdev->if_link_state != LINK_STATE_UP || (sc->sc_naddrs == 0 && sc->sc_naddrs6 == 0) || !V_carp_allow) return; switch (sc->sc_state) { case INIT: carp_set_state(sc, BACKUP, "initialization complete"); carp_setrun(sc, 0); break; case BACKUP: callout_stop(&sc->sc_ad_tmo); switch (sc->sc_version) { case CARP_VERSION_CARP: tv.tv_sec = 3 * sc->sc_advbase; tv.tv_usec = sc->sc_advskew * 1000000 / 256; timeout = tvtohz(&tv); break; case CARP_VERSION_VRRPv3: /* skew time */ timeout = (256 - sc->sc_vrrp_prio) * sc->sc_vrrp_master_inter / 256; timeout += (3 * sc->sc_vrrp_master_inter); timeout *= hz; timeout /= 100; /* master interval is in centiseconds */ break; } switch (af) { #ifdef INET case AF_INET: callout_reset(&sc->sc_md_tmo, timeout, carp_master_down, sc); break; #endif #ifdef INET6 case AF_INET6: callout_reset(&sc->sc_md6_tmo, timeout, carp_master_down, sc); break; #endif default: #ifdef INET if (sc->sc_naddrs) callout_reset(&sc->sc_md_tmo, timeout, carp_master_down, sc); #endif #ifdef INET6 if (sc->sc_naddrs6) callout_reset(&sc->sc_md6_tmo, timeout, carp_master_down, sc); #endif break; } break; case MASTER: switch (sc->sc_version) { case CARP_VERSION_CARP: tv.tv_sec = sc->sc_advbase; tv.tv_usec = sc->sc_advskew * 1000000 / 256; callout_reset(&sc->sc_ad_tmo, tvtohz(&tv), carp_callout, sc); break; case CARP_VERSION_VRRPv3: callout_reset(&sc->sc_ad_tmo, sc->sc_vrrp_adv_inter * hz / 100, carp_callout, sc); break; } break; } } /* * Setup multicast structures. */ static int carp_multicast_setup(struct carp_if *cif, sa_family_t sa) { struct ifnet *ifp = cif->cif_ifp; int error = 0; switch (sa) { #ifdef INET case AF_INET: { struct ip_moptions *imo = &cif->cif_imo; struct in_mfilter *imf; struct in_addr addr; if (ip_mfilter_first(&imo->imo_head) != NULL) return (0); imf = ip_mfilter_alloc(M_WAITOK, 0, 0); ip_mfilter_init(&imo->imo_head); imo->imo_multicast_vif = -1; addr.s_addr = htonl(INADDR_CARP_GROUP); if ((error = in_joingroup(ifp, &addr, NULL, &imf->imf_inm)) != 0) { ip_mfilter_free(imf); break; } ip_mfilter_insert(&imo->imo_head, imf); imo->imo_multicast_ifp = ifp; imo->imo_multicast_ttl = CARP_DFLTTL; imo->imo_multicast_loop = 0; break; } #endif #ifdef INET6 case AF_INET6: { struct ip6_moptions *im6o = &cif->cif_im6o; struct in6_mfilter *im6f[2]; struct in6_addr in6; if (ip6_mfilter_first(&im6o->im6o_head)) return (0); im6f[0] = ip6_mfilter_alloc(M_WAITOK, 0, 0); im6f[1] = ip6_mfilter_alloc(M_WAITOK, 0, 0); ip6_mfilter_init(&im6o->im6o_head); im6o->im6o_multicast_hlim = CARP_DFLTTL; im6o->im6o_multicast_ifp = ifp; /* Join IPv6 CARP multicast group. */ bzero(&in6, sizeof(in6)); in6.s6_addr16[0] = htons(0xff02); in6.s6_addr8[15] = 0x12; if ((error = in6_setscope(&in6, ifp, NULL)) != 0) { ip6_mfilter_free(im6f[0]); ip6_mfilter_free(im6f[1]); break; } if ((error = in6_joingroup(ifp, &in6, NULL, &im6f[0]->im6f_in6m, 0)) != 0) { ip6_mfilter_free(im6f[0]); ip6_mfilter_free(im6f[1]); break; } /* Join solicited multicast address. */ bzero(&in6, sizeof(in6)); in6.s6_addr16[0] = htons(0xff02); in6.s6_addr32[1] = 0; in6.s6_addr32[2] = htonl(1); in6.s6_addr32[3] = 0; in6.s6_addr8[12] = 0xff; if ((error = in6_setscope(&in6, ifp, NULL)) != 0) { ip6_mfilter_free(im6f[0]); ip6_mfilter_free(im6f[1]); break; } if ((error = in6_joingroup(ifp, &in6, NULL, &im6f[1]->im6f_in6m, 0)) != 0) { in6_leavegroup(im6f[0]->im6f_in6m, NULL); ip6_mfilter_free(im6f[0]); ip6_mfilter_free(im6f[1]); break; } ip6_mfilter_insert(&im6o->im6o_head, im6f[0]); ip6_mfilter_insert(&im6o->im6o_head, im6f[1]); break; } #endif } return (error); } /* * Free multicast structures. */ static void carp_multicast_cleanup(struct carp_if *cif, sa_family_t sa) { #ifdef INET struct ip_moptions *imo = &cif->cif_imo; struct in_mfilter *imf; #endif #ifdef INET6 struct ip6_moptions *im6o = &cif->cif_im6o; struct in6_mfilter *im6f; #endif sx_assert(&carp_sx, SA_XLOCKED); switch (sa) { #ifdef INET case AF_INET: if (cif->cif_naddrs != 0) break; while ((imf = ip_mfilter_first(&imo->imo_head)) != NULL) { ip_mfilter_remove(&imo->imo_head, imf); in_leavegroup(imf->imf_inm, NULL); ip_mfilter_free(imf); } break; #endif #ifdef INET6 case AF_INET6: if (cif->cif_naddrs6 != 0) break; while ((im6f = ip6_mfilter_first(&im6o->im6o_head)) != NULL) { ip6_mfilter_remove(&im6o->im6o_head, im6f); in6_leavegroup(im6f->im6f_in6m, NULL); ip6_mfilter_free(im6f); } break; #endif } } int carp_output(struct ifnet *ifp, struct mbuf *m, const struct sockaddr *sa) { struct m_tag *mtag; int vhid; if (!sa) return (0); switch (sa->sa_family) { #ifdef INET case AF_INET: break; #endif #ifdef INET6 case AF_INET6: break; #endif default: return (0); } mtag = m_tag_find(m, PACKET_TAG_CARP, NULL); if (mtag == NULL) return (0); bcopy(mtag + 1, &vhid, sizeof(vhid)); /* Set the source MAC address to the Virtual Router MAC Address. */ switch (ifp->if_type) { case IFT_ETHER: case IFT_BRIDGE: case IFT_L2VLAN: { struct ether_header *eh; eh = mtod(m, struct ether_header *); eh->ether_shost[0] = 0; eh->ether_shost[1] = 0; eh->ether_shost[2] = 0x5e; eh->ether_shost[3] = 0; eh->ether_shost[4] = 1; eh->ether_shost[5] = vhid; } break; default: printf("%s: carp is not supported for the %d interface type\n", if_name(ifp), ifp->if_type); return (EOPNOTSUPP); } return (0); } static struct carp_softc* carp_alloc(struct ifnet *ifp, carp_version_t version, int vhid) { struct carp_softc *sc; struct carp_if *cif; sx_assert(&carp_sx, SA_XLOCKED); if ((cif = ifp->if_carp) == NULL) cif = carp_alloc_if(ifp); sc = malloc(sizeof(*sc), M_CARP, M_WAITOK); *sc = (struct carp_softc ){ .sc_vhid = vhid, .sc_version = version, .sc_state = INIT, .sc_carpdev = ifp, .sc_ifasiz = sizeof(struct ifaddr *), .sc_addr = { 0, 0, 0x5e, 0, 1, vhid }, }; sc->sc_ifas = malloc(sc->sc_ifasiz, M_CARP, M_WAITOK|M_ZERO); switch (version) { case CARP_VERSION_CARP: sc->sc_advbase = CARP_DFLTINTV; sc->sc_init_counter = true; sc->sc_carpaddr.s_addr = htonl(INADDR_CARP_GROUP); sc->sc_carpaddr6.s6_addr16[0] = IPV6_ADDR_INT16_MLL; sc->sc_carpaddr6.s6_addr8[15] = 0x12; break; case CARP_VERSION_VRRPv3: sc->sc_vrrp_adv_inter = 100; sc->sc_vrrp_master_inter = sc->sc_vrrp_adv_inter; sc->sc_vrrp_prio = 100; break; } CARP_LOCK_INIT(sc); #ifdef INET callout_init_mtx(&sc->sc_md_tmo, &sc->sc_mtx, CALLOUT_RETURNUNLOCKED); #endif #ifdef INET6 callout_init_mtx(&sc->sc_md6_tmo, &sc->sc_mtx, CALLOUT_RETURNUNLOCKED); #endif callout_init_mtx(&sc->sc_ad_tmo, &sc->sc_mtx, CALLOUT_RETURNUNLOCKED); CIF_LOCK(cif); TAILQ_INSERT_TAIL(&cif->cif_vrs, sc, sc_list); CIF_UNLOCK(cif); mtx_lock(&carp_mtx); LIST_INSERT_HEAD(&carp_list, sc, sc_next); mtx_unlock(&carp_mtx); return (sc); } static void carp_grow_ifas(struct carp_softc *sc) { struct ifaddr **new; new = malloc(sc->sc_ifasiz * 2, M_CARP, M_WAITOK | M_ZERO); CARP_LOCK(sc); bcopy(sc->sc_ifas, new, sc->sc_ifasiz); free(sc->sc_ifas, M_CARP); sc->sc_ifas = new; sc->sc_ifasiz *= 2; CARP_UNLOCK(sc); } static void carp_destroy(struct carp_softc *sc) { struct ifnet *ifp = sc->sc_carpdev; struct carp_if *cif = ifp->if_carp; sx_assert(&carp_sx, SA_XLOCKED); if (sc->sc_suppress) carp_demote_adj(-V_carp_ifdown_adj, "vhid removed"); CARP_UNLOCK(sc); CIF_LOCK(cif); TAILQ_REMOVE(&cif->cif_vrs, sc, sc_list); CIF_UNLOCK(cif); mtx_lock(&carp_mtx); LIST_REMOVE(sc, sc_next); mtx_unlock(&carp_mtx); callout_drain(&sc->sc_ad_tmo); #ifdef INET callout_drain(&sc->sc_md_tmo); #endif #ifdef INET6 callout_drain(&sc->sc_md6_tmo); #endif CARP_LOCK_DESTROY(sc); free(sc->sc_ifas, M_CARP); free(sc, M_CARP); } static struct carp_if* carp_alloc_if(struct ifnet *ifp) { struct carp_if *cif; int error; cif = malloc(sizeof(*cif), M_CARP, M_WAITOK|M_ZERO); if ((error = ifpromisc(ifp, 1)) != 0) printf("%s: ifpromisc(%s) failed: %d\n", __func__, if_name(ifp), error); else cif->cif_flags |= CIF_PROMISC; CIF_LOCK_INIT(cif); cif->cif_ifp = ifp; TAILQ_INIT(&cif->cif_vrs); IF_ADDR_WLOCK(ifp); ifp->if_carp = cif; if_ref(ifp); IF_ADDR_WUNLOCK(ifp); return (cif); } static void carp_free_if(struct carp_if *cif) { struct ifnet *ifp = cif->cif_ifp; CIF_LOCK_ASSERT(cif); KASSERT(TAILQ_EMPTY(&cif->cif_vrs), ("%s: softc list not empty", __func__)); IF_ADDR_WLOCK(ifp); ifp->if_carp = NULL; IF_ADDR_WUNLOCK(ifp); CIF_LOCK_DESTROY(cif); if (cif->cif_flags & CIF_PROMISC) ifpromisc(ifp, 0); if_rele(ifp); free(cif, M_CARP); } static bool carp_carprcp(void *arg, struct carp_softc *sc, int priv) { struct carpreq *carpr = arg; CARP_LOCK(sc); carpr->carpr_state = sc->sc_state; carpr->carpr_vhid = sc->sc_vhid; switch (sc->sc_version) { case CARP_VERSION_CARP: carpr->carpr_advbase = sc->sc_advbase; carpr->carpr_advskew = sc->sc_advskew; if (priv) bcopy(sc->sc_key, carpr->carpr_key, sizeof(carpr->carpr_key)); else bzero(carpr->carpr_key, sizeof(carpr->carpr_key)); break; case CARP_VERSION_VRRPv3: break; } CARP_UNLOCK(sc); return (true); } static int carp_ioctl_set(if_t ifp, struct carpkreq *carpr) { struct epoch_tracker et; struct carp_softc *sc = NULL; int error = 0; if (carpr->carpr_vhid <= 0 || carpr->carpr_vhid > CARP_MAXVHID) return (EINVAL); switch (carpr->carpr_version) { case CARP_VERSION_CARP: if (carpr->carpr_advbase != 0 && (carpr->carpr_advbase > 255 || carpr->carpr_advbase < CARP_DFLTINTV)) return (EINVAL); if (carpr->carpr_advskew < 0 || carpr->carpr_advskew >= 255) return (EINVAL); break; case CARP_VERSION_VRRPv3: /* XXXGL: shouldn't we check anything? */ break; default: return (EINVAL); } if (ifp->if_carp) { IFNET_FOREACH_CARP(ifp, sc) if (sc->sc_vhid == carpr->carpr_vhid) break; } if (sc == NULL) sc = carp_alloc(ifp, carpr->carpr_version, carpr->carpr_vhid); else if (sc->sc_version != carpr->carpr_version) return (EINVAL); CARP_LOCK(sc); switch (sc->sc_version) { case CARP_VERSION_CARP: if (carpr->carpr_advbase != 0) sc->sc_advbase = carpr->carpr_advbase; sc->sc_advskew = carpr->carpr_advskew; if (carpr->carpr_addr.s_addr != INADDR_ANY) sc->sc_carpaddr = carpr->carpr_addr; if (!IN6_IS_ADDR_UNSPECIFIED(&carpr->carpr_addr6)) { memcpy(&sc->sc_carpaddr6, &carpr->carpr_addr6, sizeof(sc->sc_carpaddr6)); } if (carpr->carpr_key[0] != '\0') { bcopy(carpr->carpr_key, sc->sc_key, sizeof(sc->sc_key)); carp_hmac_prepare(sc); } break; case CARP_VERSION_VRRPv3: if (carpr->carpr_vrrp_priority != 0) sc->sc_vrrp_prio = carpr->carpr_vrrp_priority; if (carpr->carpr_vrrp_adv_inter) sc->sc_vrrp_adv_inter = carpr->carpr_vrrp_adv_inter; break; } if (sc->sc_state != INIT && carpr->carpr_state != sc->sc_state) { switch (carpr->carpr_state) { case BACKUP: callout_stop(&sc->sc_ad_tmo); carp_set_state(sc, BACKUP, "user requested via ifconfig"); carp_setrun(sc, 0); carp_delroute(sc); break; case MASTER: NET_EPOCH_ENTER(et); carp_master_down_locked(sc, "user requested via ifconfig"); NET_EPOCH_EXIT(et); break; default: break; } } CARP_UNLOCK(sc); return (error); } static int carp_ioctl_get(if_t ifp, struct ucred *cred, struct carpreq *carpr, bool (*outfn)(void *, struct carp_softc *, int), void *arg) { int priveleged; struct carp_softc *sc; if (carpr->carpr_vhid < 0 || carpr->carpr_vhid > CARP_MAXVHID) return (EINVAL); if (carpr->carpr_count < 1) return (EMSGSIZE); if (ifp->if_carp == NULL) return (ENOENT); priveleged = (priv_check_cred(cred, PRIV_NETINET_CARP) == 0); if (carpr->carpr_vhid != 0) { IFNET_FOREACH_CARP(ifp, sc) if (sc->sc_vhid == carpr->carpr_vhid) break; if (sc == NULL) return (ENOENT); if (! outfn(arg, sc, priveleged)) return (ENOMEM); carpr->carpr_count = 1; } else { int count; count = 0; IFNET_FOREACH_CARP(ifp, sc) count++; if (count > carpr->carpr_count) return (EMSGSIZE); IFNET_FOREACH_CARP(ifp, sc) { if (! outfn(arg, sc, priveleged)) return (ENOMEM); carpr->carpr_count = count; } } return (0); } int carp_ioctl(struct ifreq *ifr, u_long cmd, struct thread *td) { struct carpreq carpr; struct carpkreq carprk = { .carpr_version = CARP_VERSION_CARP, }; struct ifnet *ifp; int error = 0; if ((error = copyin(ifr_data_get_ptr(ifr), &carpr, sizeof carpr))) return (error); ifp = ifunit_ref(ifr->ifr_name); if ((error = carp_is_supported_if(ifp)) != 0) goto out; if ((ifp->if_flags & IFF_MULTICAST) == 0) { error = EADDRNOTAVAIL; goto out; } sx_xlock(&carp_sx); switch (cmd) { case SIOCSVH: if ((error = priv_check(td, PRIV_NETINET_CARP))) break; memcpy(&carprk, &carpr, sizeof(carpr)); error = carp_ioctl_set(ifp, &carprk); break; case SIOCGVH: error = carp_ioctl_get(ifp, td->td_ucred, &carpr, carp_carprcp, &carpr); if (error == 0) { error = copyout(&carpr, (char *)ifr_data_get_ptr(ifr), carpr.carpr_count * sizeof(carpr)); } break; default: error = EINVAL; } sx_xunlock(&carp_sx); out: if (ifp != NULL) if_rele(ifp); return (error); } static int carp_get_vhid(struct ifaddr *ifa) { if (ifa == NULL || ifa->ifa_carp == NULL) return (0); return (ifa->ifa_carp->sc_vhid); } int carp_attach(struct ifaddr *ifa, int vhid) { struct ifnet *ifp = ifa->ifa_ifp; struct carp_if *cif = ifp->if_carp; struct carp_softc *sc; int index, error; KASSERT(ifa->ifa_carp == NULL, ("%s: ifa %p attached", __func__, ifa)); switch (ifa->ifa_addr->sa_family) { #ifdef INET case AF_INET: #endif #ifdef INET6 case AF_INET6: #endif break; default: return (EPROTOTYPE); } sx_xlock(&carp_sx); if (ifp->if_carp == NULL) { sx_xunlock(&carp_sx); return (ENOPROTOOPT); } IFNET_FOREACH_CARP(ifp, sc) if (sc->sc_vhid == vhid) break; if (sc == NULL) { sx_xunlock(&carp_sx); return (ENOENT); } error = carp_multicast_setup(cif, ifa->ifa_addr->sa_family); if (error) { CIF_FREE(cif); sx_xunlock(&carp_sx); return (error); } index = sc->sc_naddrs + sc->sc_naddrs6 + 1; if (index > sc->sc_ifasiz / sizeof(struct ifaddr *)) carp_grow_ifas(sc); switch (ifa->ifa_addr->sa_family) { #ifdef INET case AF_INET: cif->cif_naddrs++; sc->sc_naddrs++; break; #endif #ifdef INET6 case AF_INET6: cif->cif_naddrs6++; sc->sc_naddrs6++; break; #endif } ifa_ref(ifa); CARP_LOCK(sc); sc->sc_ifas[index - 1] = ifa; ifa->ifa_carp = sc; if (sc->sc_version == CARP_VERSION_CARP) carp_hmac_prepare(sc); carp_sc_state(sc); CARP_UNLOCK(sc); sx_xunlock(&carp_sx); return (0); } void carp_detach(struct ifaddr *ifa, bool keep_cif) { struct ifnet *ifp = ifa->ifa_ifp; struct carp_if *cif = ifp->if_carp; struct carp_softc *sc = ifa->ifa_carp; int i, index; KASSERT(sc != NULL, ("%s: %p not attached", __func__, ifa)); sx_xlock(&carp_sx); CARP_LOCK(sc); /* Shift array. */ index = sc->sc_naddrs + sc->sc_naddrs6; for (i = 0; i < index; i++) if (sc->sc_ifas[i] == ifa) break; KASSERT(i < index, ("%s: %p no backref", __func__, ifa)); for (; i < index - 1; i++) sc->sc_ifas[i] = sc->sc_ifas[i+1]; sc->sc_ifas[index - 1] = NULL; switch (ifa->ifa_addr->sa_family) { #ifdef INET case AF_INET: cif->cif_naddrs--; sc->sc_naddrs--; break; #endif #ifdef INET6 case AF_INET6: cif->cif_naddrs6--; sc->sc_naddrs6--; break; #endif } carp_ifa_delroute(ifa); carp_multicast_cleanup(cif, ifa->ifa_addr->sa_family); ifa->ifa_carp = NULL; ifa_free(ifa); if (sc->sc_version == CARP_VERSION_CARP) carp_hmac_prepare(sc); carp_sc_state(sc); if (!keep_cif && sc->sc_naddrs == 0 && sc->sc_naddrs6 == 0) carp_destroy(sc); else CARP_UNLOCK(sc); if (!keep_cif) CIF_FREE(cif); sx_xunlock(&carp_sx); } static void carp_set_state(struct carp_softc *sc, int state, const char *reason) { CARP_LOCK_ASSERT(sc); if (sc->sc_state != state) { const char *carp_states[] = { CARP_STATES }; char subsys[IFNAMSIZ+5]; snprintf(subsys, IFNAMSIZ+5, "%u@%s", sc->sc_vhid, if_name(sc->sc_carpdev)); CARP_LOG("%s: %s -> %s (%s)\n", subsys, carp_states[sc->sc_state], carp_states[state], reason); sc->sc_state = state; devctl_notify("CARP", subsys, carp_states[state], NULL); } } static void carp_linkstate(struct ifnet *ifp) { struct carp_softc *sc; CIF_LOCK(ifp->if_carp); IFNET_FOREACH_CARP(ifp, sc) { CARP_LOCK(sc); carp_sc_state(sc); CARP_UNLOCK(sc); } CIF_UNLOCK(ifp->if_carp); } static void carp_sc_state(struct carp_softc *sc) { CARP_LOCK_ASSERT(sc); if (sc->sc_carpdev->if_link_state != LINK_STATE_UP || !(sc->sc_carpdev->if_flags & IFF_UP) || !V_carp_allow) { callout_stop(&sc->sc_ad_tmo); #ifdef INET callout_stop(&sc->sc_md_tmo); #endif #ifdef INET6 callout_stop(&sc->sc_md6_tmo); #endif carp_set_state(sc, INIT, "hardware interface down"); carp_setrun(sc, 0); carp_delroute(sc); if (!sc->sc_suppress) carp_demote_adj(V_carp_ifdown_adj, "interface down"); sc->sc_suppress = 1; } else { carp_set_state(sc, INIT, "hardware interface up"); carp_setrun(sc, 0); if (sc->sc_suppress) carp_demote_adj(-V_carp_ifdown_adj, "interface up"); sc->sc_suppress = 0; } } static void carp_demote_adj(int adj, char *reason) { atomic_add_int(&V_carp_demotion, adj); CARP_LOG("demoted by %d to %d (%s)\n", adj, V_carp_demotion, reason); taskqueue_enqueue(taskqueue_swi, &carp_sendall_task); } static int carp_allow_sysctl(SYSCTL_HANDLER_ARGS) { int new, error; struct carp_softc *sc; new = V_carp_allow; error = sysctl_handle_int(oidp, &new, 0, req); if (error || !req->newptr) return (error); if (V_carp_allow != new) { V_carp_allow = new; mtx_lock(&carp_mtx); LIST_FOREACH(sc, &carp_list, sc_next) { CARP_LOCK(sc); if (curvnet == sc->sc_carpdev->if_vnet) carp_sc_state(sc); CARP_UNLOCK(sc); } mtx_unlock(&carp_mtx); } return (0); } static int carp_dscp_sysctl(SYSCTL_HANDLER_ARGS) { int new, error; new = V_carp_dscp; error = sysctl_handle_int(oidp, &new, 0, req); if (error || !req->newptr) return (error); if (new < 0 || new > 63) return (EINVAL); V_carp_dscp = new; return (0); } static int carp_demote_adj_sysctl(SYSCTL_HANDLER_ARGS) { int new, error; new = V_carp_demotion; error = sysctl_handle_int(oidp, &new, 0, req); if (error || !req->newptr) return (error); carp_demote_adj(new, "sysctl"); return (0); } static int nlattr_get_carp_key(struct nlattr *nla, struct nl_pstate *npt, const void *arg, void *target) { if (__predict_false(NLA_DATA_LEN(nla) > CARP_KEY_LEN)) return (EINVAL); memcpy(target, NLA_DATA_CONST(nla), NLA_DATA_LEN(nla)); return (0); } struct carp_nl_send_args { struct nlmsghdr *hdr; struct nl_pstate *npt; }; static bool carp_nl_send(void *arg, struct carp_softc *sc, int priv) { struct carp_nl_send_args *nlsa = arg; struct nlmsghdr *hdr = nlsa->hdr; struct nl_pstate *npt = nlsa->npt; struct nl_writer *nw = npt->nw; struct genlmsghdr *ghdr_new; if (!nlmsg_reply(nw, hdr, sizeof(struct genlmsghdr))) { nlmsg_abort(nw); return (false); } ghdr_new = nlmsg_reserve_object(nw, struct genlmsghdr); if (ghdr_new == NULL) { nlmsg_abort(nw); return (false); } ghdr_new->cmd = CARP_NL_CMD_GET; ghdr_new->version = 0; ghdr_new->reserved = 0; CARP_LOCK(sc); nlattr_add_u32(nw, CARP_NL_VHID, sc->sc_vhid); nlattr_add_u32(nw, CARP_NL_STATE, sc->sc_state); nlattr_add_u8(nw, CARP_NL_VERSION, sc->sc_version); switch (sc->sc_version) { case CARP_VERSION_CARP: nlattr_add_s32(nw, CARP_NL_ADVBASE, sc->sc_advbase); nlattr_add_s32(nw, CARP_NL_ADVSKEW, sc->sc_advskew); nlattr_add_in_addr(nw, CARP_NL_ADDR, &sc->sc_carpaddr); nlattr_add_in6_addr(nw, CARP_NL_ADDR6, &sc->sc_carpaddr6); if (priv) nlattr_add(nw, CARP_NL_KEY, sizeof(sc->sc_key), sc->sc_key); break; case CARP_VERSION_VRRPv3: nlattr_add_u8(nw, CARP_NL_VRRP_PRIORITY, sc->sc_vrrp_prio); nlattr_add_u16(nw, CARP_NL_VRRP_ADV_INTER, sc->sc_vrrp_adv_inter); break; } CARP_UNLOCK(sc); if (! nlmsg_end(nw)) { nlmsg_abort(nw); return (false); } return (true); } struct nl_carp_parsed { unsigned int ifindex; char *ifname; uint32_t state; uint32_t vhid; int32_t advbase; int32_t advskew; char key[CARP_KEY_LEN]; struct in_addr addr; struct in6_addr addr6; carp_version_t version; uint8_t vrrp_prio; uint16_t vrrp_adv_inter; }; #define _OUT(_field) offsetof(struct nl_carp_parsed, _field) static const struct nlattr_parser nla_p_set[] = { { .type = CARP_NL_VHID, .off = _OUT(vhid), .cb = nlattr_get_uint32 }, { .type = CARP_NL_STATE, .off = _OUT(state), .cb = nlattr_get_uint32 }, { .type = CARP_NL_ADVBASE, .off = _OUT(advbase), .cb = nlattr_get_uint32 }, { .type = CARP_NL_ADVSKEW, .off = _OUT(advskew), .cb = nlattr_get_uint32 }, { .type = CARP_NL_KEY, .off = _OUT(key), .cb = nlattr_get_carp_key }, { .type = CARP_NL_IFINDEX, .off = _OUT(ifindex), .cb = nlattr_get_uint32 }, { .type = CARP_NL_ADDR, .off = _OUT(addr), .cb = nlattr_get_in_addr }, { .type = CARP_NL_ADDR6, .off = _OUT(addr6), .cb = nlattr_get_in6_addr }, { .type = CARP_NL_IFNAME, .off = _OUT(ifname), .cb = nlattr_get_string }, { .type = CARP_NL_VERSION, .off = _OUT(version), .cb = nlattr_get_uint8 }, { .type = CARP_NL_VRRP_PRIORITY, .off = _OUT(vrrp_prio), .cb = nlattr_get_uint8 }, { .type = CARP_NL_VRRP_ADV_INTER, .off = _OUT(vrrp_adv_inter), .cb = nlattr_get_uint16 }, }; NL_DECLARE_PARSER(carp_parser, struct genlmsghdr, nlf_p_empty, nla_p_set); #undef _OUT static int carp_nl_get(struct nlmsghdr *hdr, struct nl_pstate *npt) { struct nl_carp_parsed attrs = { }; struct carp_nl_send_args args; struct carpreq carpr = { }; struct epoch_tracker et; if_t ifp = NULL; int error; error = nl_parse_nlmsg(hdr, &carp_parser, npt, &attrs); if (error != 0) return (error); NET_EPOCH_ENTER(et); if (attrs.ifname != NULL) ifp = ifunit_ref(attrs.ifname); else if (attrs.ifindex != 0) ifp = ifnet_byindex_ref(attrs.ifindex); NET_EPOCH_EXIT(et); if ((error = carp_is_supported_if(ifp)) != 0) goto out; hdr->nlmsg_flags |= NLM_F_MULTI; args.hdr = hdr; args.npt = npt; carpr.carpr_vhid = attrs.vhid; carpr.carpr_count = CARP_MAXVHID; sx_xlock(&carp_sx); error = carp_ioctl_get(ifp, nlp_get_cred(npt->nlp), &carpr, carp_nl_send, &args); sx_xunlock(&carp_sx); if (! nlmsg_end_dump(npt->nw, error, hdr)) error = ENOMEM; out: if (ifp != NULL) if_rele(ifp); return (error); } static int carp_nl_set(struct nlmsghdr *hdr, struct nl_pstate *npt) { struct nl_carp_parsed attrs = { }; struct carpkreq carpr; struct epoch_tracker et; if_t ifp = NULL; int error; error = nl_parse_nlmsg(hdr, &carp_parser, npt, &attrs); if (error != 0) return (error); if (attrs.vhid <= 0 || attrs.vhid > CARP_MAXVHID) return (EINVAL); if (attrs.state > CARP_MAXSTATE) return (EINVAL); if (attrs.version == 0) /* compat with pre-VRRPv3 */ attrs.version = CARP_VERSION_CARP; switch (attrs.version) { case CARP_VERSION_CARP: if (attrs.advbase < 0 || attrs.advskew < 0) return (EINVAL); if (attrs.advbase > 255) return (EINVAL); if (attrs.advskew >= 255) return (EINVAL); break; case CARP_VERSION_VRRPv3: if (attrs.vrrp_adv_inter > VRRP_MAX_INTERVAL) return (EINVAL); break; default: return (EINVAL); } NET_EPOCH_ENTER(et); if (attrs.ifname != NULL) ifp = ifunit_ref(attrs.ifname); else if (attrs.ifindex != 0) ifp = ifnet_byindex_ref(attrs.ifindex); NET_EPOCH_EXIT(et); if ((error = carp_is_supported_if(ifp)) != 0) goto out; if ((ifp->if_flags & IFF_MULTICAST) == 0) { error = EADDRNOTAVAIL; goto out; } carpr.carpr_count = 1; carpr.carpr_vhid = attrs.vhid; carpr.carpr_state = attrs.state; carpr.carpr_version = attrs.version; switch (attrs.version) { case CARP_VERSION_CARP: carpr.carpr_advbase = attrs.advbase; carpr.carpr_advskew = attrs.advskew; carpr.carpr_addr = attrs.addr; carpr.carpr_addr6 = attrs.addr6; memcpy(&carpr.carpr_key, &attrs.key, sizeof(attrs.key)); break; case CARP_VERSION_VRRPv3: carpr.carpr_vrrp_priority = attrs.vrrp_prio; carpr.carpr_vrrp_adv_inter = attrs.vrrp_adv_inter; break; } sx_xlock(&carp_sx); error = carp_ioctl_set(ifp, &carpr); sx_xunlock(&carp_sx); out: if (ifp != NULL) if_rele(ifp); return (error); } static const struct nlhdr_parser *all_parsers[] = { &carp_parser }; static const struct genl_cmd carp_cmds[] = { { .cmd_num = CARP_NL_CMD_GET, .cmd_name = "SIOCGVH", .cmd_cb = carp_nl_get, .cmd_flags = GENL_CMD_CAP_DO | GENL_CMD_CAP_DUMP | GENL_CMD_CAP_HASPOL, }, { .cmd_num = CARP_NL_CMD_SET, .cmd_name = "SIOCSVH", .cmd_cb = carp_nl_set, .cmd_flags = GENL_CMD_CAP_DO | GENL_CMD_CAP_HASPOL, .cmd_priv = PRIV_NETINET_CARP, }, }; static uint16_t carp_family_id; static void carp_nl_register(void) { bool ret __diagused; NL_VERIFY_PARSERS(all_parsers); carp_family_id = genl_register_family(CARP_NL_FAMILY_NAME, 0, 2, CARP_NL_CMD_MAX); MPASS(carp_family_id != 0); ret = genl_register_cmds(carp_family_id, carp_cmds, nitems(carp_cmds)); MPASS(ret); } static void carp_nl_unregister(void) { genl_unregister_family(carp_family_id); } static void carp_mod_cleanup(void) { carp_nl_unregister(); #ifdef INET (void)ipproto_unregister(IPPROTO_CARP); carp_iamatch_p = NULL; #endif #ifdef INET6 (void)ip6proto_unregister(IPPROTO_CARP); carp_iamatch6_p = NULL; carp_macmatch6_p = NULL; #endif carp_ioctl_p = NULL; carp_attach_p = NULL; carp_detach_p = NULL; carp_get_vhid_p = NULL; carp_linkstate_p = NULL; carp_forus_p = NULL; carp_output_p = NULL; carp_demote_adj_p = NULL; carp_master_p = NULL; mtx_unlock(&carp_mtx); taskqueue_drain(taskqueue_swi, &carp_sendall_task); mtx_destroy(&carp_mtx); sx_destroy(&carp_sx); } static void ipcarp_sysinit(void) { /* Load allow as tunable so to postpone carp start after module load */ TUNABLE_INT_FETCH("net.inet.carp.allow", &V_carp_allow); } VNET_SYSINIT(ip_carp, SI_SUB_PROTO_DOMAIN, SI_ORDER_ANY, ipcarp_sysinit, NULL); static int carp_mod_load(void) { int err; mtx_init(&carp_mtx, "carp_mtx", NULL, MTX_DEF); sx_init(&carp_sx, "carp_sx"); carp_get_vhid_p = carp_get_vhid; carp_forus_p = carp_forus; carp_output_p = carp_output; carp_linkstate_p = carp_linkstate; carp_ioctl_p = carp_ioctl; carp_attach_p = carp_attach; carp_detach_p = carp_detach; carp_demote_adj_p = carp_demote_adj; carp_master_p = carp_master; #ifdef INET6 carp_iamatch6_p = carp_iamatch6; carp_macmatch6_p = carp_macmatch6; err = ip6proto_register(IPPROTO_CARP, carp6_input, NULL); if (err) { printf("carp: error %d registering with INET6\n", err); carp_mod_cleanup(); return (err); } #endif #ifdef INET carp_iamatch_p = carp_iamatch; err = ipproto_register(IPPROTO_CARP, carp_input, NULL); if (err) { printf("carp: error %d registering with INET\n", err); carp_mod_cleanup(); return (err); } #endif carp_nl_register(); return (0); } static int carp_modevent(module_t mod, int type, void *data) { switch (type) { case MOD_LOAD: return carp_mod_load(); /* NOTREACHED */ case MOD_UNLOAD: mtx_lock(&carp_mtx); if (LIST_EMPTY(&carp_list)) carp_mod_cleanup(); else { mtx_unlock(&carp_mtx); return (EBUSY); } break; default: return (EINVAL); } return (0); } static moduledata_t carp_mod = { "carp", carp_modevent, 0 }; DECLARE_MODULE(carp, carp_mod, SI_SUB_PROTO_DOMAIN, SI_ORDER_ANY); diff --git a/sys/netinet6/in6_src.c b/sys/netinet6/in6_src.c index dd6864482b3c..3e55c6e5fc05 100644 --- a/sys/netinet6/in6_src.c +++ b/sys/netinet6/in6_src.c @@ -1,1115 +1,1139 @@ /*- * SPDX-License-Identifier: BSD-3-Clause * * Copyright (C) 1995, 1996, 1997, and 1998 WIDE Project. * All rights reserved. * * Redistribution and use in source and binary forms, with or without * modification, are permitted provided that the following conditions * are met: * 1. Redistributions of source code must retain the above copyright * notice, this list of conditions and the following disclaimer. * 2. Redistributions in binary form must reproduce the above copyright * notice, this list of conditions and the following disclaimer in the * documentation and/or other materials provided with the distribution. * 3. Neither the name of the project nor the names of its contributors * may be used to endorse or promote products derived from this software * without specific prior written permission. * * THIS SOFTWARE IS PROVIDED BY THE PROJECT AND CONTRIBUTORS ``AS IS'' AND * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE * ARE DISCLAIMED. IN NO EVENT SHALL THE PROJECT OR CONTRIBUTORS BE LIABLE * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF * SUCH DAMAGE. * * $KAME: in6_src.c,v 1.132 2003/08/26 04:42:27 keiichi Exp $ */ /*- * Copyright (c) 1982, 1986, 1991, 1993 * The Regents of the University of California. All rights reserved. * * Redistribution and use in source and binary forms, with or without * modification, are permitted provided that the following conditions * are met: * 1. Redistributions of source code must retain the above copyright * notice, this list of conditions and the following disclaimer. * 2. Redistributions in binary form must reproduce the above copyright * notice, this list of conditions and the following disclaimer in the * documentation and/or other materials provided with the distribution. * 3. Neither the name of the University nor the names of its contributors * may be used to endorse or promote products derived from this software * without specific prior written permission. * * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF * SUCH DAMAGE. */ #include #include "opt_inet.h" #include "opt_inet6.h" #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include static struct mtx addrsel_lock; #define ADDRSEL_LOCK_INIT() mtx_init(&addrsel_lock, "addrsel_lock", NULL, MTX_DEF) #define ADDRSEL_LOCK() mtx_lock(&addrsel_lock) #define ADDRSEL_UNLOCK() mtx_unlock(&addrsel_lock) #define ADDRSEL_LOCK_ASSERT() mtx_assert(&addrsel_lock, MA_OWNED) static struct sx addrsel_sxlock; #define ADDRSEL_SXLOCK_INIT() sx_init(&addrsel_sxlock, "addrsel_sxlock") #define ADDRSEL_SLOCK() sx_slock(&addrsel_sxlock) #define ADDRSEL_SUNLOCK() sx_sunlock(&addrsel_sxlock) #define ADDRSEL_XLOCK() sx_xlock(&addrsel_sxlock) #define ADDRSEL_XUNLOCK() sx_xunlock(&addrsel_sxlock) #define ADDR_LABEL_NOTAPP (-1) VNET_DEFINE_STATIC(struct in6_addrpolicy, defaultaddrpolicy); #define V_defaultaddrpolicy VNET(defaultaddrpolicy) VNET_DEFINE(int, ip6_prefer_tempaddr) = 0; static int selectroute(struct sockaddr_in6 *, struct ip6_pktopts *, struct ip6_moptions *, struct route_in6 *, struct ifnet **, struct nhop_object **, int, u_int, uint32_t); static int in6_selectif(struct sockaddr_in6 *, struct ip6_pktopts *, struct ip6_moptions *, struct ifnet **, struct ifnet *, u_int); static int in6_selectsrc(uint32_t, struct sockaddr_in6 *, - struct ip6_pktopts *, struct inpcb *, struct ucred *, - struct ifnet **, struct in6_addr *); + struct ip6_pktopts *, struct ip6_moptions *, struct inpcb *, + struct ucred *, struct ifnet **, struct in6_addr *); static struct in6_addrpolicy *lookup_addrsel_policy(struct sockaddr_in6 *); static void init_policy_queue(void); static int add_addrsel_policyent(struct in6_addrpolicy *); static int delete_addrsel_policyent(struct in6_addrpolicy *); static int walk_addrsel_policy(int (*)(struct in6_addrpolicy *, void *), void *); static int dump_addrsel_policyent(struct in6_addrpolicy *, void *); static struct in6_addrpolicy *match_addrsel_policy(struct sockaddr_in6 *); /* * Return an IPv6 address, which is the most appropriate for a given * destination and user specified options. * If necessary, this function lookups the routing table and returns * an entry to the caller for later use. */ #define REPLACE(r) do {\ IP6STAT_INC2(ip6s_sources_rule, (r)); \ /* { \ char ip6buf[INET6_ADDRSTRLEN], ip6b[INET6_ADDRSTRLEN]; \ printf("in6_selectsrc: replace %s with %s by %d\n", ia_best ? ip6_sprintf(ip6buf, &ia_best->ia_addr.sin6_addr) : "none", ip6_sprintf(ip6b, &ia->ia_addr.sin6_addr), (r)); \ } */ \ goto replace; \ } while(0) #define NEXT(r) do {\ /* { \ char ip6buf[INET6_ADDRSTRLEN], ip6b[INET6_ADDRSTRLEN]; \ printf("in6_selectsrc: keep %s against %s by %d\n", ia_best ? ip6_sprintf(ip6buf, &ia_best->ia_addr.sin6_addr) : "none", ip6_sprintf(ip6b, &ia->ia_addr.sin6_addr), (r)); \ } */ \ goto next; /* XXX: we can't use 'continue' here */ \ } while(0) #define BREAK(r) do { \ IP6STAT_INC2(ip6s_sources_rule, (r)); \ goto out; /* XXX: we can't use 'break' here */ \ } while(0) static int in6_selectsrc(uint32_t fibnum, struct sockaddr_in6 *dstsock, - struct ip6_pktopts *opts, struct inpcb *inp, struct ucred *cred, - struct ifnet **ifpp, struct in6_addr *srcp) + struct ip6_pktopts *opts, struct ip6_moptions *mopts, struct inpcb *inp, + struct ucred *cred, struct ifnet **ifpp, struct in6_addr *srcp) { struct rm_priotracker in6_ifa_tracker; struct in6_addr dst, tmp; struct ifnet *ifp = NULL, *oifp = NULL; struct in6_ifaddr *ia = NULL, *ia_best = NULL; struct in6_pktinfo *pi = NULL; int dst_scope = -1, best_scope = -1, best_matchlen = -1; struct in6_addrpolicy *dst_policy = NULL, *best_policy = NULL; u_int32_t odstzone; int prefer_tempaddr; int error; - struct ip6_moptions *mopts; NET_EPOCH_ASSERT(); KASSERT(srcp != NULL, ("%s: srcp is NULL", __func__)); dst = dstsock->sin6_addr; /* make a copy for local operation */ if (ifpp) { /* * Save a possibly passed in ifp for in6_selectsrc. Only * neighbor discovery code should use this feature, where * we may know the interface but not the FIB number holding * the connected subnet in case someone deleted it from the * default FIB and we need to check the interface. */ if (*ifpp != NULL) oifp = *ifpp; *ifpp = NULL; } - if (inp != NULL) { - INP_LOCK_ASSERT(inp); - mopts = inp->in6p_moptions; - } else { - mopts = NULL; - } - /* * If the source address is explicitly specified by the caller, * check if the requested source address is indeed a unicast address * assigned to the node, and can be used as the packet's source * address. If everything is okay, use the address as source. */ if (opts && (pi = opts->ip6po_pktinfo) && !IN6_IS_ADDR_UNSPECIFIED(&pi->ipi6_addr)) { /* get the outgoing interface */ if ((error = in6_selectif(dstsock, opts, mopts, &ifp, oifp, fibnum)) != 0) return (error); /* * determine the appropriate zone id of the source based on * the zone of the destination and the outgoing interface. * If the specified address is ambiguous wrt the scope zone, * the interface must be specified; otherwise, ifa_ifwithaddr() * will fail matching the address. */ tmp = pi->ipi6_addr; if (ifp) { error = in6_setscope(&tmp, ifp, &odstzone); if (error) return (error); } if (cred != NULL && (error = prison_local_ip6(cred, &tmp, (inp->inp_flags & IN6P_IPV6_V6ONLY) != 0)) != 0) return (error); /* * If IPV6_BINDANY socket option is set, we allow to specify * non local addresses as source address in IPV6_PKTINFO * ancillary data. */ if ((inp->inp_flags & INP_BINDANY) == 0) { ia = in6ifa_ifwithaddr(&tmp, 0 /* XXX */, false); if (ia == NULL || (ia->ia6_flags & (IN6_IFF_ANYCAST | IN6_IFF_NOTREADY))) return (EADDRNOTAVAIL); bcopy(&ia->ia_addr.sin6_addr, srcp, sizeof(*srcp)); } else bcopy(&tmp, srcp, sizeof(*srcp)); pi->ipi6_addr = tmp; /* XXX: this overrides pi */ if (ifpp) *ifpp = ifp; return (0); } /* * Otherwise, if the socket has already bound the source, just use it. */ if (inp != NULL && !IN6_IS_ADDR_UNSPECIFIED(&inp->in6p_laddr)) { if (cred != NULL && (error = prison_local_ip6(cred, &inp->in6p_laddr, ((inp->inp_flags & IN6P_IPV6_V6ONLY) != 0))) != 0) return (error); bcopy(&inp->in6p_laddr, srcp, sizeof(*srcp)); return (0); } /* * Bypass source address selection and use the primary jail IP * if requested. */ if (cred != NULL && !prison_saddrsel_ip6(cred, srcp)) return (0); /* * If the address is not specified, choose the best one based on * the outgoing interface and the destination address. */ /* get the outgoing interface */ if ((error = in6_selectif(dstsock, opts, mopts, &ifp, oifp, (inp != NULL) ? inp->inp_inc.inc_fibnum : fibnum)) != 0) return (error); #ifdef DIAGNOSTIC if (ifp == NULL) /* this should not happen */ panic("in6_selectsrc: NULL ifp"); #endif error = in6_setscope(&dst, ifp, &odstzone); if (error) return (error); IN6_IFADDR_RLOCK(&in6_ifa_tracker); CK_STAILQ_FOREACH(ia, &V_in6_ifaddrhead, ia_link) { int new_scope = -1, new_matchlen = -1; struct in6_addrpolicy *new_policy = NULL; u_int32_t srczone, osrczone, dstzone; struct in6_addr src; struct ifnet *ifp1 = ia->ia_ifp; /* * We'll never take an address that breaks the scope zone * of the destination. We also skip an address if its zone * does not contain the outgoing interface. * XXX: we should probably use sin6_scope_id here. */ if (in6_setscope(&dst, ifp1, &dstzone) || odstzone != dstzone) { continue; } src = ia->ia_addr.sin6_addr; if (in6_setscope(&src, ifp, &osrczone) || in6_setscope(&src, ifp1, &srczone) || osrczone != srczone) { continue; } /* avoid unusable addresses */ if ((ia->ia6_flags & (IN6_IFF_NOTREADY | IN6_IFF_ANYCAST | IN6_IFF_DETACHED))) { continue; } if (!V_ip6_use_deprecated && IFA6_IS_DEPRECATED(ia)) continue; /* If jailed only take addresses of the jail into account. */ if (cred != NULL && prison_check_ip6(cred, &ia->ia_addr.sin6_addr) != 0) continue; /* Rule 1: Prefer same address */ if (IN6_ARE_ADDR_EQUAL(&dst, &ia->ia_addr.sin6_addr)) { ia_best = ia; BREAK(1); /* there should be no better candidate */ } if (ia_best == NULL) REPLACE(0); /* Rule 2: Prefer appropriate scope */ if (dst_scope < 0) dst_scope = in6_addrscope(&dst); new_scope = in6_addrscope(&ia->ia_addr.sin6_addr); if (IN6_ARE_SCOPE_CMP(best_scope, new_scope) < 0) { if (IN6_ARE_SCOPE_CMP(best_scope, dst_scope) < 0) REPLACE(2); NEXT(2); } else if (IN6_ARE_SCOPE_CMP(new_scope, best_scope) < 0) { if (IN6_ARE_SCOPE_CMP(new_scope, dst_scope) < 0) NEXT(2); REPLACE(2); } /* * Rule 3: Avoid deprecated addresses. Note that the case of * !ip6_use_deprecated is already rejected above. */ if (!IFA6_IS_DEPRECATED(ia_best) && IFA6_IS_DEPRECATED(ia)) NEXT(3); if (IFA6_IS_DEPRECATED(ia_best) && !IFA6_IS_DEPRECATED(ia)) REPLACE(3); /* Rule 4: Prefer home addresses */ /* * XXX: This is a TODO. We should probably merge the MIP6 * case above. */ /* Rule 5: Prefer outgoing interface */ if (!(ND_IFINFO(ifp)->flags & ND6_IFF_NO_PREFER_IFACE)) { if (ia_best->ia_ifp == ifp && ia->ia_ifp != ifp) NEXT(5); if (ia_best->ia_ifp != ifp && ia->ia_ifp == ifp) REPLACE(5); } /* * Rule 6: Prefer matching label * Note that best_policy should be non-NULL here. */ if (dst_policy == NULL) dst_policy = lookup_addrsel_policy(dstsock); if (dst_policy->label != ADDR_LABEL_NOTAPP) { new_policy = lookup_addrsel_policy(&ia->ia_addr); if (dst_policy->label == best_policy->label && dst_policy->label != new_policy->label) NEXT(6); if (dst_policy->label != best_policy->label && dst_policy->label == new_policy->label) REPLACE(6); } /* * Rule 7: Prefer public addresses. * We allow users to reverse the logic by configuring * a sysctl variable, so that privacy conscious users can * always prefer temporary addresses. */ if (opts == NULL || opts->ip6po_prefer_tempaddr == IP6PO_TEMPADDR_SYSTEM) { prefer_tempaddr = V_ip6_prefer_tempaddr; } else if (opts->ip6po_prefer_tempaddr == IP6PO_TEMPADDR_NOTPREFER) { prefer_tempaddr = 0; } else prefer_tempaddr = 1; if (!(ia_best->ia6_flags & IN6_IFF_TEMPORARY) && (ia->ia6_flags & IN6_IFF_TEMPORARY)) { if (prefer_tempaddr) REPLACE(7); else NEXT(7); } if ((ia_best->ia6_flags & IN6_IFF_TEMPORARY) && !(ia->ia6_flags & IN6_IFF_TEMPORARY)) { if (prefer_tempaddr) NEXT(7); else REPLACE(7); } /* * Rule 8: prefer addresses on alive interfaces. * This is a KAME specific rule. */ if ((ia_best->ia_ifp->if_flags & IFF_UP) && !(ia->ia_ifp->if_flags & IFF_UP)) NEXT(8); if (!(ia_best->ia_ifp->if_flags & IFF_UP) && (ia->ia_ifp->if_flags & IFF_UP)) REPLACE(8); /* * Rule 9: prefer address with better virtual status. */ if (ifa_preferred(&ia_best->ia_ifa, &ia->ia_ifa)) REPLACE(9); if (ifa_preferred(&ia->ia_ifa, &ia_best->ia_ifa)) NEXT(9); /* * Rule 10: prefer address with `prefer_source' flag. */ if ((ia_best->ia6_flags & IN6_IFF_PREFER_SOURCE) == 0 && (ia->ia6_flags & IN6_IFF_PREFER_SOURCE) != 0) REPLACE(10); if ((ia_best->ia6_flags & IN6_IFF_PREFER_SOURCE) != 0 && (ia->ia6_flags & IN6_IFF_PREFER_SOURCE) == 0) NEXT(10); /* * Rule 14: Use longest matching prefix. * Note: in the address selection draft, this rule is * documented as "Rule 8". However, since it is also * documented that this rule can be overridden, we assign * a large number so that it is easy to assign smaller numbers * to more preferred rules. */ new_matchlen = in6_matchlen(&ia->ia_addr.sin6_addr, &dst); if (best_matchlen < new_matchlen) REPLACE(14); if (new_matchlen < best_matchlen) NEXT(14); /* Rule 15 is reserved. */ /* * Last resort: just keep the current candidate. * Or, do we need more rules? */ continue; replace: ia_best = ia; best_scope = (new_scope >= 0 ? new_scope : in6_addrscope(&ia_best->ia_addr.sin6_addr)); best_policy = (new_policy ? new_policy : lookup_addrsel_policy(&ia_best->ia_addr)); best_matchlen = (new_matchlen >= 0 ? new_matchlen : in6_matchlen(&ia_best->ia_addr.sin6_addr, &dst)); next: continue; out: break; } if ((ia = ia_best) == NULL) { IN6_IFADDR_RUNLOCK(&in6_ifa_tracker); IP6STAT_INC(ip6s_sources_none); return (EADDRNOTAVAIL); } /* * At this point at least one of the addresses belonged to the jail * but it could still be, that we want to further restrict it, e.g. * theoratically IN6_IS_ADDR_LOOPBACK. * It must not be IN6_IS_ADDR_UNSPECIFIED anymore. * prison_local_ip6() will fix an IN6_IS_ADDR_LOOPBACK but should * let all others previously selected pass. * Use tmp to not change ::1 on lo0 to the primary jail address. */ tmp = ia->ia_addr.sin6_addr; if (cred != NULL && prison_local_ip6(cred, &tmp, (inp != NULL && (inp->inp_flags & IN6P_IPV6_V6ONLY) != 0)) != 0) { IN6_IFADDR_RUNLOCK(&in6_ifa_tracker); IP6STAT_INC(ip6s_sources_none); return (EADDRNOTAVAIL); } if (ifpp) *ifpp = ifp; bcopy(&tmp, srcp, sizeof(*srcp)); if (ia->ia_ifp == ifp) IP6STAT_INC2(ip6s_sources_sameif, best_scope); else IP6STAT_INC2(ip6s_sources_otherif, best_scope); if (dst_scope == best_scope) IP6STAT_INC2(ip6s_sources_samescope, best_scope); else IP6STAT_INC2(ip6s_sources_otherscope, best_scope); if (IFA6_IS_DEPRECATED(ia)) IP6STAT_INC2(ip6s_sources_deprecated, best_scope); IN6_IFADDR_RUNLOCK(&in6_ifa_tracker); return (0); } /* * Select source address based on @inp, @dstsock and @opts. * Stores selected address to @srcp. If @scope_ambiguous is set, * embed scope from selected outgoing interface. If @hlim pointer * is provided, stores calculated hop limit there. * Returns 0 on success. */ int in6_selectsrc_socket(struct sockaddr_in6 *dstsock, struct ip6_pktopts *opts, struct inpcb *inp, struct ucred *cred, int scope_ambiguous, struct in6_addr *srcp, int *hlim) { struct ifnet *retifp; uint32_t fibnum; int error; + INP_LOCK_ASSERT(inp); + fibnum = inp->inp_inc.inc_fibnum; retifp = NULL; - error = in6_selectsrc(fibnum, dstsock, opts, inp, cred, &retifp, srcp); + error = in6_selectsrc(fibnum, dstsock, opts, inp->in6p_moptions, + inp, cred, &retifp, srcp); if (error != 0) return (error); if (hlim != NULL) *hlim = in6_selecthlim(inp, retifp); if (retifp == NULL || scope_ambiguous == 0) return (0); /* * Application should provide a proper zone ID or the use of * default zone IDs should be enabled. Unfortunately, some * applications do not behave as it should, so we need a * workaround. Even if an appropriate ID is not determined * (when it's required), if we can determine the outgoing * interface. determine the zone ID based on the interface. */ error = in6_setscope(&dstsock->sin6_addr, retifp, NULL); return (error); } /* * Select source address based on @fibnum, @dst and @scopeid. * Stores selected address to @srcp. * Returns 0 on success. * - * Used by non-socket based consumers (ND code mostly) + * Used by non-socket based consumers */ int in6_selectsrc_addr(uint32_t fibnum, const struct in6_addr *dst, uint32_t scopeid, struct ifnet *ifp, struct in6_addr *srcp, int *hlim) { struct ifnet *retifp; struct sockaddr_in6 dst_sa; int error; retifp = ifp; bzero(&dst_sa, sizeof(dst_sa)); dst_sa.sin6_family = AF_INET6; dst_sa.sin6_len = sizeof(dst_sa); dst_sa.sin6_addr = *dst; dst_sa.sin6_scope_id = scopeid; sa6_embedscope(&dst_sa, 0); - error = in6_selectsrc(fibnum, &dst_sa, NULL, NULL, NULL, &retifp, srcp); + error = in6_selectsrc(fibnum, &dst_sa, NULL, NULL, + NULL, NULL, &retifp, srcp); if (hlim != NULL) *hlim = in6_selecthlim(NULL, retifp); return (error); } +/* + * Select source address based on @fibnum, @dst and @mopts. + * Stores selected address to @srcp. + * Returns 0 on success. + * + * Used by non-socket based consumers (ND code mostly) + */ +int +in6_selectsrc_nbr(uint32_t fibnum, const struct in6_addr *dst, + struct ip6_moptions *mopts, struct ifnet *ifp, struct in6_addr *srcp) +{ + struct sockaddr_in6 dst_sa; + struct ifnet *retifp; + int error; + + retifp = ifp; + bzero(&dst_sa, sizeof(dst_sa)); + dst_sa.sin6_family = AF_INET6; + dst_sa.sin6_len = sizeof(dst_sa); + dst_sa.sin6_addr = *dst; + dst_sa.sin6_scope_id = ntohs(in6_getscope(dst)); + sa6_embedscope(&dst_sa, 0); + + error = in6_selectsrc(fibnum, &dst_sa, NULL, mopts, + NULL, NULL, &retifp, srcp); + return (error); +} + static struct nhop_object * cache_route(uint32_t fibnum, const struct sockaddr_in6 *dst, struct route_in6 *ro, uint32_t flowid) { /* * Use a cached route if it exists and is valid, else try to allocate * a new one. Note that we should check the address family of the * cached destination, in case of sharing the cache with IPv4. * Assumes that 'struct route_in6' is exclusively locked. */ if (ro->ro_nh != NULL && ( !NH_IS_VALID(ro->ro_nh) || ro->ro_dst.sin6_family != AF_INET6 || !IN6_ARE_ADDR_EQUAL(&ro->ro_dst.sin6_addr, &dst->sin6_addr))) RO_NHFREE(ro); if (ro->ro_nh == NULL) { ro->ro_dst = *dst; const struct in6_addr *paddr; struct in6_addr unscoped_addr; uint32_t scopeid = 0; if (IN6_IS_SCOPE_LINKLOCAL(&dst->sin6_addr)) { in6_splitscope(&dst->sin6_addr, &unscoped_addr, &scopeid); paddr = &unscoped_addr; } else paddr = &dst->sin6_addr; ro->ro_nh = fib6_lookup(fibnum, paddr, scopeid, NHR_REF, flowid); } return (ro->ro_nh); } static struct nhop_object * lookup_route(uint32_t fibnum, struct sockaddr_in6 *dst, struct route_in6 *ro, struct ip6_pktopts *opts, uint32_t flowid) { struct nhop_object *nh = NULL; /* * If the next hop address for the packet is specified by the caller, * use it as the gateway. */ if (opts && opts->ip6po_nexthop) { struct route_in6 *ron = &opts->ip6po_nextroute; struct sockaddr_in6 *sin6_next = satosin6(opts->ip6po_nexthop); nh = cache_route(fibnum, sin6_next, ron, flowid); /* * The node identified by that address must be a * neighbor of the sending host. */ if (nh != NULL && (nh->nh_flags & NHF_GATEWAY) != 0) nh = NULL; } else if (ro != NULL) { nh = cache_route(fibnum, dst, ro, flowid); if (nh == NULL) return (NULL); /* * Check if the outgoing interface conflicts with * the interface specified by ipi6_ifindex (if specified). */ struct in6_pktinfo *pi; if (opts && (pi = opts->ip6po_pktinfo) != NULL && pi->ipi6_ifindex) { if (nh->nh_aifp->if_index != pi->ipi6_ifindex) nh = NULL; } } return (nh); } /* * Finds outgoing nexthop or the outgoing interface for the * @dstsock. * Return 0 on success and stores the lookup result in @retnh and @retifp */ static int selectroute(struct sockaddr_in6 *dstsock, struct ip6_pktopts *opts, struct ip6_moptions *mopts, struct route_in6 *ro, struct ifnet **retifp, struct nhop_object **retnh, int norouteok, u_int fibnum, uint32_t flowid) { int error = 0; struct ifnet *ifp = NULL; struct in6_pktinfo *pi = NULL; struct in6_addr *dst = &dstsock->sin6_addr; /* If the caller specify the outgoing interface explicitly, use it. */ if (opts && (pi = opts->ip6po_pktinfo) != NULL && pi->ipi6_ifindex) { /* XXX boundary check is assumed to be already done. */ ifp = ifnet_byindex(pi->ipi6_ifindex); if (ifp != NULL && (norouteok || IN6_IS_ADDR_MULTICAST(dst))) { /* * we do not have to check or get the route for * multicast. */ goto done; } else goto getroute; } /* * If the destination address is a multicast address and the outgoing * interface for the address is specified by the caller, use it. */ if (IN6_IS_ADDR_MULTICAST(dst) && mopts != NULL && (ifp = mopts->im6o_multicast_ifp) != NULL) { goto done; /* we do not need a route for multicast. */ } /* * If destination address is LLA or link- or node-local multicast, * use it's embedded scope zone id to determine outgoing interface. */ if (IN6_IS_ADDR_MC_LINKLOCAL(dst) || IN6_IS_ADDR_MC_NODELOCAL(dst)) { uint32_t zoneid = ntohs(in6_getscope(dst)); if (zoneid > 0) { ifp = in6_getlinkifnet(zoneid); goto done; } } getroute:; struct nhop_object *nh = lookup_route(fibnum, dstsock, ro, opts, flowid); if (nh != NULL) { *retifp = nh->nh_aifp; error = 0; } else { *retifp = NULL; IP6STAT_INC(ip6s_noroute); error = EHOSTUNREACH; } *retnh = nh; return (error); done: if (ifp == NULL) { /* * This can happen if the caller did not pass a cached route * nor any other hints. We treat this case an error. */ error = EHOSTUNREACH; } if (error == EHOSTUNREACH) IP6STAT_INC(ip6s_noroute); *retifp = ifp; *retnh = NULL; return (error); } static int in6_selectif(struct sockaddr_in6 *dstsock, struct ip6_pktopts *opts, struct ip6_moptions *mopts, struct ifnet **retifp, struct ifnet *oifp, u_int fibnum) { int error; struct route_in6 sro; struct nhop_object *nh = NULL; uint16_t nh_flags; KASSERT(retifp != NULL, ("%s: retifp is NULL", __func__)); bzero(&sro, sizeof(sro)); nh_flags = 0; error = selectroute(dstsock, opts, mopts, &sro, retifp, &nh, 1, fibnum, 0); if (nh != NULL) nh_flags = nh->nh_flags; if (nh != NULL && nh == sro.ro_nh) NH_FREE(nh); if (error != 0) { /* Help ND. See oifp comment in in6_selectsrc(). */ if (oifp != NULL && fibnum == RT_DEFAULT_FIB) { *retifp = oifp; error = 0; } return (error); } /* * do not use a rejected or black hole route. * XXX: this check should be done in the L2 output routine. * However, if we skipped this check here, we'd see the following * scenario: * - install a rejected route for a scoped address prefix * (like fe80::/10) * - send a packet to a destination that matches the scoped prefix, * with ambiguity about the scope zone. * - pick the outgoing interface from the route, and disambiguate the * scope zone with the interface. * - ip6_output() would try to get another route with the "new" * destination, which may be valid. * - we'd see no error on output. * Although this may not be very harmful, it should still be confusing. * We thus reject the case here. */ if (nh_flags & (NHF_REJECT | NHF_BLACKHOLE)) { error = (nh_flags & NHF_HOST ? EHOSTUNREACH : ENETUNREACH); return (error); } return (0); } /* Public wrapper function to selectroute(). */ int in6_selectroute(struct sockaddr_in6 *dstsock, struct ip6_pktopts *opts, struct ip6_moptions *mopts, struct route_in6 *ro, struct ifnet **retifp, struct nhop_object **retnh, u_int fibnum, uint32_t flowid) { MPASS(retifp != NULL); MPASS(retnh != NULL); return (selectroute(dstsock, opts, mopts, ro, retifp, retnh, 0, fibnum, flowid)); } /* * Default hop limit selection. The precedence is as follows: * 1. Hoplimit value specified via ioctl. * 2. (If the outgoing interface is detected) the current * hop limit of the interface specified by router advertisement. * 3. The system default hoplimit. */ int in6_selecthlim(struct inpcb *inp, struct ifnet *ifp) { if (inp && inp->in6p_hops >= 0) return (inp->in6p_hops); else if (ifp) return (ND_IFINFO(ifp)->chlim); else if (inp && !IN6_IS_ADDR_UNSPECIFIED(&inp->in6p_faddr)) { struct nhop_object *nh; struct in6_addr dst; uint32_t fibnum, scopeid; int hlim; fibnum = inp->inp_inc.inc_fibnum; in6_splitscope(&inp->in6p_faddr, &dst, &scopeid); nh = fib6_lookup(fibnum, &dst, scopeid, 0, 0); if (nh != NULL) { hlim = ND_IFINFO(nh->nh_ifp)->chlim; return (hlim); } } return (V_ip6_defhlim); } void addrsel_policy_init(void) { init_policy_queue(); /* initialize the "last resort" policy */ bzero(&V_defaultaddrpolicy, sizeof(V_defaultaddrpolicy)); V_defaultaddrpolicy.label = ADDR_LABEL_NOTAPP; if (!IS_DEFAULT_VNET(curvnet)) return; ADDRSEL_LOCK_INIT(); ADDRSEL_SXLOCK_INIT(); } static struct in6_addrpolicy * lookup_addrsel_policy(struct sockaddr_in6 *key) { struct in6_addrpolicy *match = NULL; ADDRSEL_LOCK(); match = match_addrsel_policy(key); if (match == NULL) match = &V_defaultaddrpolicy; else match->use++; ADDRSEL_UNLOCK(); return (match); } /* * Subroutines to manage the address selection policy table via sysctl. */ struct walkarg { struct sysctl_req *w_req; }; static int in6_src_sysctl(SYSCTL_HANDLER_ARGS); SYSCTL_DECL(_net_inet6_ip6); static SYSCTL_NODE(_net_inet6_ip6, IPV6CTL_ADDRCTLPOLICY, addrctlpolicy, CTLFLAG_RD | CTLFLAG_MPSAFE, in6_src_sysctl, ""); static int in6_src_sysctl(SYSCTL_HANDLER_ARGS) { struct walkarg w; if (req->newptr) return EPERM; bzero(&w, sizeof(w)); w.w_req = req; return (walk_addrsel_policy(dump_addrsel_policyent, &w)); } int in6_src_ioctl(u_long cmd, caddr_t data) { struct in6_addrpolicy ent0; if (cmd != SIOCAADDRCTL_POLICY && cmd != SIOCDADDRCTL_POLICY) return (EOPNOTSUPP); /* check for safety */ ent0 = *(struct in6_addrpolicy *)data; if (ent0.label == ADDR_LABEL_NOTAPP) return (EINVAL); /* check if the prefix mask is consecutive. */ if (in6_mask2len(&ent0.addrmask.sin6_addr, NULL) < 0) return (EINVAL); /* clear trailing garbages (if any) of the prefix address. */ IN6_MASK_ADDR(&ent0.addr.sin6_addr, &ent0.addrmask.sin6_addr); ent0.use = 0; switch (cmd) { case SIOCAADDRCTL_POLICY: return (add_addrsel_policyent(&ent0)); case SIOCDADDRCTL_POLICY: return (delete_addrsel_policyent(&ent0)); } return (0); /* XXX: compromise compilers */ } /* * The followings are implementation of the policy table using a * simple tail queue. * XXX such details should be hidden. * XXX implementation using binary tree should be more efficient. */ struct addrsel_policyent { TAILQ_ENTRY(addrsel_policyent) ape_entry; struct in6_addrpolicy ape_policy; }; TAILQ_HEAD(addrsel_policyhead, addrsel_policyent); VNET_DEFINE_STATIC(struct addrsel_policyhead, addrsel_policytab); #define V_addrsel_policytab VNET(addrsel_policytab) static void init_policy_queue(void) { TAILQ_INIT(&V_addrsel_policytab); } static int add_addrsel_policyent(struct in6_addrpolicy *newpolicy) { struct addrsel_policyent *new, *pol; new = malloc(sizeof(*new), M_IFADDR, M_WAITOK); ADDRSEL_XLOCK(); ADDRSEL_LOCK(); /* duplication check */ TAILQ_FOREACH(pol, &V_addrsel_policytab, ape_entry) { if (IN6_ARE_ADDR_EQUAL(&newpolicy->addr.sin6_addr, &pol->ape_policy.addr.sin6_addr) && IN6_ARE_ADDR_EQUAL(&newpolicy->addrmask.sin6_addr, &pol->ape_policy.addrmask.sin6_addr)) { ADDRSEL_UNLOCK(); ADDRSEL_XUNLOCK(); free(new, M_IFADDR); return (EEXIST); /* or override it? */ } } bzero(new, sizeof(*new)); /* XXX: should validate entry */ new->ape_policy = *newpolicy; TAILQ_INSERT_TAIL(&V_addrsel_policytab, new, ape_entry); ADDRSEL_UNLOCK(); ADDRSEL_XUNLOCK(); return (0); } static int delete_addrsel_policyent(struct in6_addrpolicy *key) { struct addrsel_policyent *pol; ADDRSEL_XLOCK(); ADDRSEL_LOCK(); /* search for the entry in the table */ TAILQ_FOREACH(pol, &V_addrsel_policytab, ape_entry) { if (IN6_ARE_ADDR_EQUAL(&key->addr.sin6_addr, &pol->ape_policy.addr.sin6_addr) && IN6_ARE_ADDR_EQUAL(&key->addrmask.sin6_addr, &pol->ape_policy.addrmask.sin6_addr)) { break; } } if (pol == NULL) { ADDRSEL_UNLOCK(); ADDRSEL_XUNLOCK(); return (ESRCH); } TAILQ_REMOVE(&V_addrsel_policytab, pol, ape_entry); ADDRSEL_UNLOCK(); ADDRSEL_XUNLOCK(); free(pol, M_IFADDR); return (0); } static int walk_addrsel_policy(int (*callback)(struct in6_addrpolicy *, void *), void *w) { struct addrsel_policyent *pol; int error = 0; ADDRSEL_SLOCK(); TAILQ_FOREACH(pol, &V_addrsel_policytab, ape_entry) { if ((error = (*callback)(&pol->ape_policy, w)) != 0) { ADDRSEL_SUNLOCK(); return (error); } } ADDRSEL_SUNLOCK(); return (error); } static int dump_addrsel_policyent(struct in6_addrpolicy *pol, void *arg) { int error = 0; struct walkarg *w = arg; error = SYSCTL_OUT(w->w_req, pol, sizeof(*pol)); return (error); } static struct in6_addrpolicy * match_addrsel_policy(struct sockaddr_in6 *key) { struct addrsel_policyent *pent; struct in6_addrpolicy *bestpol = NULL, *pol; int matchlen, bestmatchlen = -1; u_char *mp, *ep, *k, *p, m; TAILQ_FOREACH(pent, &V_addrsel_policytab, ape_entry) { matchlen = 0; pol = &pent->ape_policy; mp = (u_char *)&pol->addrmask.sin6_addr; ep = mp + 16; /* XXX: scope field? */ k = (u_char *)&key->sin6_addr; p = (u_char *)&pol->addr.sin6_addr; for (; mp < ep && *mp; mp++, k++, p++) { m = *mp; if ((*k & m) != *p) goto next; /* not match */ if (m == 0xff) /* short cut for a typical case */ matchlen += 8; else { while (m >= 0x80) { matchlen++; m <<= 1; } } } /* matched. check if this is better than the current best. */ if (bestpol == NULL || matchlen > bestmatchlen) { bestpol = pol; bestmatchlen = matchlen; } next: continue; } return (bestpol); } diff --git a/sys/netinet6/ip6_var.h b/sys/netinet6/ip6_var.h index e1a4e8678ebb..c28bfa5a9d08 100644 --- a/sys/netinet6/ip6_var.h +++ b/sys/netinet6/ip6_var.h @@ -1,496 +1,498 @@ /*- * SPDX-License-Identifier: BSD-3-Clause * * Copyright (C) 1995, 1996, 1997, and 1998 WIDE Project. * All rights reserved. * * Redistribution and use in source and binary forms, with or without * modification, are permitted provided that the following conditions * are met: * 1. Redistributions of source code must retain the above copyright * notice, this list of conditions and the following disclaimer. * 2. Redistributions in binary form must reproduce the above copyright * notice, this list of conditions and the following disclaimer in the * documentation and/or other materials provided with the distribution. * 3. Neither the name of the project nor the names of its contributors * may be used to endorse or promote products derived from this software * without specific prior written permission. * * THIS SOFTWARE IS PROVIDED BY THE PROJECT AND CONTRIBUTORS ``AS IS'' AND * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE * ARE DISCLAIMED. IN NO EVENT SHALL THE PROJECT OR CONTRIBUTORS BE LIABLE * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF * SUCH DAMAGE. * * $KAME: ip6_var.h,v 1.62 2001/05/03 14:51:48 itojun Exp $ */ /*- * Copyright (c) 1982, 1986, 1993 * The Regents of the University of California. All rights reserved. * * Redistribution and use in source and binary forms, with or without * modification, are permitted provided that the following conditions * are met: * 1. Redistributions of source code must retain the above copyright * notice, this list of conditions and the following disclaimer. * 2. Redistributions in binary form must reproduce the above copyright * notice, this list of conditions and the following disclaimer in the * documentation and/or other materials provided with the distribution. * 3. Neither the name of the University nor the names of its contributors * may be used to endorse or promote products derived from this software * without specific prior written permission. * * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF * SUCH DAMAGE. */ #ifndef _NETINET6_IP6_VAR_H_ #define _NETINET6_IP6_VAR_H_ #include #ifdef _KERNEL struct ip6asfrag; /* frag6.c */ TAILQ_HEAD(ip6fraghead, ip6asfrag); /* * IP6 reassembly queue structure. Each fragment * being reassembled is attached to one of these structures. */ struct ip6q { struct ip6fraghead ip6q_frags; u_int32_t ip6q_ident; u_int8_t ip6q_nxt; u_int8_t ip6q_ecn; u_int16_t ip6q_ttl; struct in6_addr ip6q_src, ip6q_dst; TAILQ_ENTRY(ip6q) ip6q_tq; int ip6q_unfrglen; /* len of unfragmentable part */ int ip6q_nfrag; /* # of fragments */ struct label *ip6q_label; }; #endif /* _KERNEL */ /* * IP6 reinjecting structure. */ struct ip6_direct_ctx { uint32_t ip6dc_nxt; /* next header to process */ uint32_t ip6dc_off; /* offset to next header */ }; #if defined(_NETINET6_IN6_VAR_H_) && defined(_KERNEL) /* * Structure attached to inpcb.in6p_moptions and * passed to ip6_output when IPv6 multicast options are in use. * This structure is lazy-allocated. */ struct ip6_moptions { struct ifnet *im6o_multicast_ifp; /* ifp for outgoing multicasts */ u_char im6o_multicast_hlim; /* hoplimit for outgoing multicasts */ u_char im6o_multicast_loop; /* 1 >= hear sends if a member */ struct ip6_mfilter_head im6o_head; /* group membership list */ }; #else struct ip6_moptions; #endif /* * Control options for outgoing packets */ /* Routing header related info */ struct ip6po_rhinfo { struct ip6_rthdr *ip6po_rhi_rthdr; /* Routing header */ struct route_in6 ip6po_rhi_route; /* Route to the 1st hop */ }; #define ip6po_rthdr ip6po_rhinfo.ip6po_rhi_rthdr #define ip6po_route ip6po_rhinfo.ip6po_rhi_route /* Nexthop related info */ struct ip6po_nhinfo { struct sockaddr *ip6po_nhi_nexthop; struct route_in6 ip6po_nhi_route; /* Route to the nexthop */ }; #define ip6po_nexthop ip6po_nhinfo.ip6po_nhi_nexthop #define ip6po_nextroute ip6po_nhinfo.ip6po_nhi_route /* * Note that fields with valid data must be flagged in ip6po_valid. * This is done to reduce cache misses in ip6_output(). Before * ip6po_valid, ip6_output needed to check all the individual fields * of ip6_pktopts needed to be checked themselves, and they are spread * across 4 cachelines. ip6_output() is currently the only consumer of * these flags, as it is in the critical path of every packet sent. */ struct ip6_pktopts { uint32_t ip6po_valid; #define IP6PO_VALID_HLIM 0x0001 #define IP6PO_VALID_PKTINFO 0x0002 #define IP6PO_VALID_NHINFO 0x0004 #define IP6PO_VALID_HBH 0x0008 #define IP6PO_VALID_DEST1 0x0010 #define IP6PO_VALID_RHINFO 0x0020 #define IP6PO_VALID_DEST2 0x0040 #define IP6PO_VALID_TC 0x0080 int ip6po_hlim; /* Hoplimit for outgoing packets */ int ip6po_tclass; /* traffic class */ int ip6po_minmtu; /* fragment vs PMTU discovery policy */ #define IP6PO_MINMTU_MCASTONLY -1 /* default; send at min MTU for multicast*/ #define IP6PO_MINMTU_DISABLE 0 /* always perform pmtu disc */ #define IP6PO_MINMTU_ALL 1 /* always send at min MTU */ int ip6po_prefer_tempaddr; /* whether temporary addresses are preferred as source address */ #define IP6PO_TEMPADDR_SYSTEM -1 /* follow the system default */ #define IP6PO_TEMPADDR_NOTPREFER 0 /* not prefer temporary address */ #define IP6PO_TEMPADDR_PREFER 1 /* prefer temporary address */ int ip6po_flags; #if 0 /* parameters in this block is obsolete. do not reuse the values. */ #define IP6PO_REACHCONF 0x01 /* upper-layer reachability confirmation. */ #define IP6PO_MINMTU 0x02 /* use minimum MTU (IPV6_USE_MIN_MTU) */ #endif #define IP6PO_DONTFRAG 0x04 /* disable fragmentation (IPV6_DONTFRAG) */ #define IP6PO_USECOA 0x08 /* use care of address */ struct mbuf *ip6po_m; /* Pointer to mbuf storing the data */ /* Outgoing IF/address information */ struct in6_pktinfo *ip6po_pktinfo; /* Next-hop address information */ struct ip6po_nhinfo ip6po_nhinfo; struct ip6_hbh *ip6po_hbh; /* Hop-by-Hop options header */ /* Destination options header (before a routing header) */ struct ip6_dest *ip6po_dest1; /* Routing header related info. */ struct ip6po_rhinfo ip6po_rhinfo; /* Destination options header (after a routing header) */ struct ip6_dest *ip6po_dest2; }; /* * Control options for incoming packets */ struct ip6stat { uint64_t ip6s_total; /* total packets received */ uint64_t ip6s_tooshort; /* packet too short */ uint64_t ip6s_toosmall; /* not enough data */ uint64_t ip6s_fragments; /* fragments received */ uint64_t ip6s_fragdropped; /* frags dropped(dups, out of space) */ uint64_t ip6s_fragtimeout; /* fragments timed out */ uint64_t ip6s_fragoverflow; /* fragments that exceeded limit */ uint64_t ip6s_forward; /* packets forwarded */ uint64_t ip6s_cantforward; /* packets rcvd for unreachable dest */ uint64_t ip6s_redirectsent; /* packets forwarded on same net */ uint64_t ip6s_delivered; /* datagrams delivered to upper level*/ uint64_t ip6s_localout; /* total ip packets generated here */ uint64_t ip6s_odropped; /* lost packets due to nobufs, etc. */ uint64_t ip6s_reassembled; /* total packets reassembled ok */ uint64_t ip6s_atomicfrags; /* atomic fragments */ uint64_t ip6s_fragmented; /* datagrams successfully fragmented */ uint64_t ip6s_ofragments; /* output fragments created */ uint64_t ip6s_cantfrag; /* don't fragment flag was set, etc. */ uint64_t ip6s_badoptions; /* error in option processing */ uint64_t ip6s_noroute; /* packets discarded due to no route */ uint64_t ip6s_badvers; /* ip6 version != 6 */ uint64_t ip6s_rawout; /* total raw ip packets generated */ uint64_t ip6s_badscope; /* scope error */ uint64_t ip6s_notmember; /* don't join this multicast group */ #define IP6S_HDRCNT 256 /* headers count */ uint64_t ip6s_nxthist[IP6S_HDRCNT]; /* next header history */ uint64_t ip6s_m1; /* one mbuf */ #define IP6S_M2MMAX 32 uint64_t ip6s_m2m[IP6S_M2MMAX]; /* two or more mbuf */ uint64_t ip6s_mext1; /* one ext mbuf */ uint64_t ip6s_mext2m; /* two or more ext mbuf */ uint64_t ip6s_exthdrtoolong; /* ext hdr are not contiguous */ uint64_t ip6s_nogif; /* no match gif found */ uint64_t ip6s_toomanyhdr; /* discarded due to too many headers */ /* * statistics for improvement of the source address selection * algorithm: * XXX: hardcoded 16 = # of ip6 multicast scope types + 1 */ #define IP6S_RULESMAX 16 #define IP6S_SCOPECNT 16 /* number of times that address selection fails */ uint64_t ip6s_sources_none; /* number of times that an address on the outgoing I/F is chosen */ uint64_t ip6s_sources_sameif[IP6S_SCOPECNT]; /* number of times that an address on a non-outgoing I/F is chosen */ uint64_t ip6s_sources_otherif[IP6S_SCOPECNT]; /* * number of times that an address that has the same scope * from the destination is chosen. */ uint64_t ip6s_sources_samescope[IP6S_SCOPECNT]; /* * number of times that an address that has a different scope * from the destination is chosen. */ uint64_t ip6s_sources_otherscope[IP6S_SCOPECNT]; /* number of times that a deprecated address is chosen */ uint64_t ip6s_sources_deprecated[IP6S_SCOPECNT]; /* number of times that each rule of source selection is applied. */ uint64_t ip6s_sources_rule[IP6S_RULESMAX]; }; #ifdef _KERNEL #include #include VNET_PCPUSTAT_DECLARE(struct ip6stat, ip6stat); #define IP6STAT_ADD(name, val) \ do { \ MIB_SDT_PROBE1(ip6, count, name, (val)); \ VNET_PCPUSTAT_ADD(struct ip6stat, ip6stat, name, (val)); \ } while (0) #define IP6STAT_SUB(name, val) IP6STAT_ADD(name, -(val)) #define IP6STAT_INC(name) IP6STAT_ADD(name, 1) #define IP6STAT_INC2(name, type) \ do { \ MIB_SDT_PROBE2(ip6, count, name, 1, type); \ VNET_PCPUSTAT_ADD(struct ip6stat, ip6stat, name, 1); \ } while (0) #define IP6STAT_DEC(name) IP6STAT_SUB(name, 1) #endif #ifdef _KERNEL /* flags passed to ip6_output as last parameter */ #define IPV6_UNSPECSRC 0x01 /* allow :: as the source address */ #define IPV6_FORWARDING 0x02 /* most of IPv6 header exists */ #define IPV6_MINMTU 0x04 /* use minimum MTU (IPV6_USE_MIN_MTU) */ #ifdef __NO_STRICT_ALIGNMENT #define IP6_HDR_ALIGNED_P(ip) 1 #else #define IP6_HDR_ALIGNED_P(ip) ((((intptr_t) (ip)) & 3) == 0) #endif VNET_DECLARE(int, ip6_defhlim); /* default hop limit */ VNET_DECLARE(int, ip6_defmcasthlim); /* default multicast hop limit */ VNET_DECLARE(int, ip6_forwarding); /* act as router? */ VNET_DECLARE(int, ip6_use_deprecated); /* allow deprecated addr as source */ VNET_DECLARE(int, ip6_rr_prune); /* router renumbering prefix * walk list every 5 sec. */ VNET_DECLARE(int, ip6_mcast_pmtu); /* enable pMTU discovery for multicast? */ VNET_DECLARE(int, ip6_v6only); #define V_ip6_defhlim VNET(ip6_defhlim) #define V_ip6_defmcasthlim VNET(ip6_defmcasthlim) #define V_ip6_forwarding VNET(ip6_forwarding) #define V_ip6_use_deprecated VNET(ip6_use_deprecated) #define V_ip6_rr_prune VNET(ip6_rr_prune) #define V_ip6_mcast_pmtu VNET(ip6_mcast_pmtu) #define V_ip6_v6only VNET(ip6_v6only) VNET_DECLARE(struct socket *, ip6_mrouter); /* multicast routing daemon */ VNET_DECLARE(int, ip6_sendredirects); /* send IP redirects when forwarding? */ VNET_DECLARE(int, ip6_accept_rtadv); /* Acts as a host not a router */ VNET_DECLARE(int, ip6_no_radr); /* No defroute from RA */ VNET_DECLARE(int, ip6_norbit_raif); /* Disable R-bit in NA on RA * receiving IF. */ VNET_DECLARE(int, ip6_rfc6204w3); /* Accept defroute from RA even when forwarding enabled */ VNET_DECLARE(int, ip6_hdrnestlimit); /* upper limit of # of extension * headers */ VNET_DECLARE(int, ip6_dad_count); /* DupAddrDetectionTransmits */ #define V_ip6_mrouter VNET(ip6_mrouter) #define V_ip6_sendredirects VNET(ip6_sendredirects) #define V_ip6_accept_rtadv VNET(ip6_accept_rtadv) #define V_ip6_no_radr VNET(ip6_no_radr) #define V_ip6_norbit_raif VNET(ip6_norbit_raif) #define V_ip6_rfc6204w3 VNET(ip6_rfc6204w3) #define V_ip6_hdrnestlimit VNET(ip6_hdrnestlimit) #define V_ip6_dad_count VNET(ip6_dad_count) VNET_DECLARE(int, ip6_auto_flowlabel); VNET_DECLARE(int, ip6_auto_linklocal); #define V_ip6_auto_flowlabel VNET(ip6_auto_flowlabel) #define V_ip6_auto_linklocal VNET(ip6_auto_linklocal) VNET_DECLARE(int, ip6_use_tempaddr); /* Whether to use temporary addresses */ VNET_DECLARE(int, ip6_prefer_tempaddr); /* Whether to prefer temporary * addresses in the source address * selection */ VNET_DECLARE(bool, ip6_use_stableaddr); /* Whether to use stable address generation (RFC 7217) */ #define V_ip6_use_tempaddr VNET(ip6_use_tempaddr) #define V_ip6_prefer_tempaddr VNET(ip6_prefer_tempaddr) #define V_ip6_use_stableaddr VNET(ip6_use_stableaddr) #define IP6_IDGEN_RETRIES 3 /* RFC 7217 section 7 default max retries */ VNET_DECLARE(int, ip6_stableaddr_maxretries); #define V_ip6_stableaddr_maxretries VNET(ip6_stableaddr_maxretries) #define IP6_STABLEADDR_NETIFSRC_NAME 0 #define IP6_STABLEADDR_NETIFSRC_ID 1 #define IP6_STABLEADDR_NETIFSRC_MAC 2 VNET_DECLARE(int, ip6_stableaddr_netifsource); #define V_ip6_stableaddr_netifsource VNET(ip6_stableaddr_netifsource) VNET_DECLARE(int, ip6_use_defzone); /* Whether to use the default scope * zone when unspecified */ #define V_ip6_use_defzone VNET(ip6_use_defzone) VNET_DECLARE(struct pfil_head *, inet6_pfil_head); #define V_inet6_pfil_head VNET(inet6_pfil_head) #define PFIL_INET6_NAME "inet6" VNET_DECLARE(struct pfil_head *, inet6_local_pfil_head); #define V_inet6_local_pfil_head VNET(inet6_local_pfil_head) #define PFIL_INET6_LOCAL_NAME "inet6-local" #ifdef IPSTEALTH VNET_DECLARE(int, ip6stealth); #define V_ip6stealth VNET(ip6stealth) #endif VNET_DECLARE(bool, ip6_log_cannot_forward); #define V_ip6_log_cannot_forward VNET(ip6_log_cannot_forward) extern struct pr_usrreqs rip6_usrreqs; struct sockopt; struct inpcb; struct ucred; int icmp6_ctloutput(struct socket *, struct sockopt *sopt); void ip6_input(struct mbuf *); void ip6_direct_input(struct mbuf *); void ip6_freepcbopts(struct ip6_pktopts *); int ip6_unknown_opt(u_int8_t *, struct mbuf *, int); int ip6_get_prevhdr(const struct mbuf *, int); int ip6_nexthdr(const struct mbuf *, int, int, int *); int ip6_lasthdr(const struct mbuf *, int, int, int *); extern int (*ip6_mforward)(struct ip6_hdr *, struct ifnet *, struct mbuf *); int ip6_process_hopopts(struct mbuf *, u_int8_t *, int, u_int32_t *, u_int32_t *); struct mbuf **ip6_savecontrol_v4(struct inpcb *, struct mbuf *, struct mbuf **, int *); void ip6_savecontrol(struct inpcb *, struct mbuf *, struct mbuf **); void ip6_notify_pmtu(struct inpcb *, struct sockaddr_in6 *, u_int32_t); int ip6_sysctl(int *, u_int, void *, size_t *, void *, size_t); void ip6_forward(struct mbuf *, int); void ip6_mloopback(struct ifnet *, struct mbuf *); int ip6_output(struct mbuf *, struct ip6_pktopts *, struct route_in6 *, int, struct ip6_moptions *, struct ifnet **, struct inpcb *); int ip6_ctloutput(struct socket *, struct sockopt *); int ip6_raw_ctloutput(struct socket *, struct sockopt *); void ip6_initpktopts(struct ip6_pktopts *); int ip6_setpktopts(struct mbuf *, struct ip6_pktopts *, struct ip6_pktopts *, struct ucred *, int); void ip6_clearpktopts(struct ip6_pktopts *, int); struct ip6_pktopts *ip6_copypktopts(struct ip6_pktopts *, int); int ip6_optlen(struct inpcb *); int ip6_deletefraghdr(struct mbuf *, int, int); int ip6_fragment(struct ifnet *, struct mbuf *, int, u_char, int, uint32_t); int route6_input(struct mbuf **, int *, int); void frag6_init(void); void frag6_destroy(void); int frag6_input(struct mbuf **, int *, int); void frag6_drain(void *, int); void rip6_init(void); int rip6_ctloutput(struct socket *, struct sockopt *); int rip6_usrreq(struct socket *, int, struct mbuf *, struct mbuf *, struct mbuf *, struct thread *); int dest6_input(struct mbuf **, int *, int); int none_input(struct mbuf **, int *, int); int in6_selectsrc_socket(struct sockaddr_in6 *, struct ip6_pktopts *, struct inpcb *, struct ucred *, int, struct in6_addr *, int *); int in6_selectsrc_addr(uint32_t, const struct in6_addr *, uint32_t, struct ifnet *, struct in6_addr *, int *); +int in6_selectsrc_nbr(uint32_t, const struct in6_addr *, + struct ip6_moptions *, struct ifnet *, struct in6_addr *); int in6_selectroute(struct sockaddr_in6 *, struct ip6_pktopts *, struct ip6_moptions *, struct route_in6 *, struct ifnet **, struct nhop_object **, u_int, uint32_t); u_int32_t ip6_randomid(void); u_int32_t ip6_randomflowlabel(void); void in6_delayed_cksum(struct mbuf *m, uint32_t plen, u_short offset); int ip6_log_ratelimit(void); /* * Argument type for the last arg of ip6proto_ctlinput_t(). * * IPv6 ICMP IPv6 [exthdrs] finalhdr payload * ^ ^ ^ ^ * | | ip6c_ip6 ip6c_off * | ip6c_icmp6 * ip6c_m * * ip6c_finaldst's sin6_addr usually points to ip6c_ip6->ip6_dst. If the * original * (internal) packet carries a routing header, it may point the * final * destination address in the routing header. * * ip6c_src: ip6c_ip6->ip6_src + scope info + flowlabel in ip6c_ip6 * (beware of flowlabel, if you try to compare it against others) * ip6c_dst: ip6c_finaldst + scope info */ struct ip6ctlparam { struct mbuf *ip6c_m; /* start of mbuf chain */ struct icmp6_hdr *ip6c_icmp6; /* icmp6 header of target packet */ struct ip6_hdr *ip6c_ip6; /* ip6 header of target packet */ int ip6c_off; /* offset of the target proto header */ struct sockaddr_in6 *ip6c_src; /* srcaddr w/ additional info */ struct sockaddr_in6 *ip6c_dst; /* (final) dstaddr w/ additional info */ struct sockaddr_in6 *ip6c_finaldst; /* final destination address */ void *ip6c_cmdarg; /* control command dependent data */ u_int8_t ip6c_nxt; /* final next header field */ }; typedef int ip6proto_input_t(struct mbuf **, int *, int); typedef void ip6proto_ctlinput_t(struct ip6ctlparam *); int ip6proto_register(uint8_t, ip6proto_input_t, ip6proto_ctlinput_t); int ip6proto_unregister(uint8_t); #define IP6PROTO_REGISTER(prot, input, ctl) do { \ int error __diagused; \ error = ip6proto_register(prot, input, ctl); \ MPASS(error == 0); \ } while (0) ip6proto_input_t rip6_input; ip6proto_ctlinput_t rip6_ctlinput; #endif /* _KERNEL */ #endif /* !_NETINET6_IP6_VAR_H_ */ diff --git a/sys/netinet6/nd6.h b/sys/netinet6/nd6.h index 5fe027ac5e7c..e484c709e29a 100644 --- a/sys/netinet6/nd6.h +++ b/sys/netinet6/nd6.h @@ -1,424 +1,428 @@ /*- * SPDX-License-Identifier: BSD-3-Clause * * Copyright (C) 1995, 1996, 1997, and 1998 WIDE Project. * All rights reserved. * * Redistribution and use in source and binary forms, with or without * modification, are permitted provided that the following conditions * are met: * 1. Redistributions of source code must retain the above copyright * notice, this list of conditions and the following disclaimer. * 2. Redistributions in binary form must reproduce the above copyright * notice, this list of conditions and the following disclaimer in the * documentation and/or other materials provided with the distribution. * 3. Neither the name of the project nor the names of its contributors * may be used to endorse or promote products derived from this software * without specific prior written permission. * * THIS SOFTWARE IS PROVIDED BY THE PROJECT AND CONTRIBUTORS ``AS IS'' AND * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE * ARE DISCLAIMED. IN NO EVENT SHALL THE PROJECT OR CONTRIBUTORS BE LIABLE * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF * SUCH DAMAGE. * * $KAME: nd6.h,v 1.76 2001/12/18 02:10:31 itojun Exp $ */ #ifndef _NETINET6_ND6_H_ #define _NETINET6_ND6_H_ /* see net/route.h, or net/if_inarp.h */ #ifndef RTF_ANNOUNCE #define RTF_ANNOUNCE RTF_PROTO2 #endif #include #include struct llentry; #define ND6_LLINFO_NOSTATE -2 /* * We don't need the WAITDELETE state any more, but we keep the definition * in a comment line instead of removing it. This is necessary to avoid * unintentionally reusing the value for another purpose, which might * affect backward compatibility with old applications. * (20000711 jinmei@kame.net) */ /* #define ND6_LLINFO_WAITDELETE -1 */ #define ND6_LLINFO_INCOMPLETE 0 #define ND6_LLINFO_REACHABLE 1 #define ND6_LLINFO_STALE 2 #define ND6_LLINFO_DELAY 3 #define ND6_LLINFO_PROBE 4 #define ND6_IS_LLINFO_PROBREACH(n) ((n)->ln_state > ND6_LLINFO_INCOMPLETE) #define ND6_LLINFO_PERMANENT(n) (((n)->la_expire == 0) && ((n)->ln_state > ND6_LLINFO_INCOMPLETE)) struct nd_ifinfo { u_int32_t linkmtu; /* LinkMTU */ u_int32_t maxmtu; /* Upper bound of LinkMTU */ u_int32_t basereachable; /* BaseReachableTime */ u_int32_t reachable; /* Reachable Time */ u_int32_t retrans; /* Retrans Timer */ u_int32_t flags; /* Flags */ int recalctm; /* BaseReacable re-calculation timer */ u_int8_t chlim; /* CurHopLimit */ u_int8_t initialized; /* Flag to see the entry is initialized */ /* the following 3 members are for privacy extension for addrconf */ u_int8_t randomseed0[8]; /* upper 64 bits of MD5 digest */ u_int8_t randomseed1[8]; /* lower 64 bits (usually the EUI64 IFID) */ u_int8_t randomid[8]; /* current random ID */ }; #define ND6_IFF_PERFORMNUD 0x1 #define ND6_IFF_ACCEPT_RTADV 0x2 #define ND6_IFF_PREFER_SOURCE 0x4 /* Not used in FreeBSD. */ #define ND6_IFF_IFDISABLED 0x8 /* IPv6 operation is disabled due to * DAD failure. (XXX: not ND-specific) */ #define ND6_IFF_DONT_SET_IFROUTE 0x10 #define ND6_IFF_AUTO_LINKLOCAL 0x20 #define ND6_IFF_NO_RADR 0x40 #define ND6_IFF_NO_PREFER_IFACE 0x80 /* XXX: not related to ND. */ #define ND6_IFF_NO_DAD 0x100 #define ND6_IFF_STABLEADDR 0x800 #ifdef EXPERIMENTAL /* XXX: not related to ND. */ #define ND6_IFF_IPV6_ONLY 0x200 /* draft-ietf-6man-ipv6only-flag */ #define ND6_IFF_IPV6_ONLY_MANUAL 0x400 #define ND6_IFF_IPV6_ONLY_MASK (ND6_IFF_IPV6_ONLY|ND6_IFF_IPV6_ONLY_MANUAL) #endif #ifdef _KERNEL #define ND_IFINFO(ifp) \ (((struct in6_ifextra *)if_getafdata(ifp, AF_INET6))->nd_ifinfo) #define IN6_LINKMTU(ifp) \ ((ND_IFINFO(ifp)->linkmtu && ND_IFINFO(ifp)->linkmtu < (ifp)->if_mtu) \ ? ND_IFINFO(ifp)->linkmtu \ : ((ND_IFINFO(ifp)->maxmtu && ND_IFINFO(ifp)->maxmtu < (ifp)->if_mtu) \ ? ND_IFINFO(ifp)->maxmtu : (ifp)->if_mtu)) #endif struct in6_nbrinfo { char ifname[IFNAMSIZ]; /* if name, e.g. "en0" */ struct in6_addr addr; /* IPv6 address of the neighbor */ long asked; /* number of queries already sent for this addr */ int isrouter; /* if it acts as a router */ int state; /* reachability state */ int expire; /* lifetime for NDP state transition */ }; /* Sysctls, shared with user space. */ struct in6_defrouter { struct sockaddr_in6 rtaddr; u_char flags; u_short rtlifetime; u_long expire; u_short if_index; }; struct in6_prefix { struct sockaddr_in6 prefix; struct prf_ra raflags; u_char prefixlen; u_char origin; u_int32_t vltime; u_int32_t pltime; time_t expire; u_int32_t flags; int refcnt; u_short if_index; u_short advrtrs; /* number of advertisement routers */ /* struct sockaddr_in6 advrtr[] */ }; #ifdef _KERNEL struct in6_ondireq { char ifname[IFNAMSIZ]; struct { u_int32_t linkmtu; /* LinkMTU */ u_int32_t maxmtu; /* Upper bound of LinkMTU */ u_int32_t basereachable; /* BaseReachableTime */ u_int32_t reachable; /* Reachable Time */ u_int32_t retrans; /* Retrans Timer */ u_int32_t flags; /* Flags */ int recalctm; /* BaseReacable re-calculation timer */ u_int8_t chlim; /* CurHopLimit */ u_int8_t receivedra; } ndi; }; #endif struct in6_ndireq { char ifname[IFNAMSIZ]; struct nd_ifinfo ndi; }; struct in6_ndifreq { char ifname[IFNAMSIZ]; u_long ifindex; }; /* Prefix status */ #define NDPRF_ONLINK 0x1 #define NDPRF_DETACHED 0x2 +/* ND6 NA output flags */ +#define ND6_NA_OPT_LLA 0x01 +#define ND6_NA_CARP_MASTER 0x02 + /* protocol constants */ #define MAX_RTR_SOLICITATION_DELAY 1 /* 1sec */ #define RTR_SOLICITATION_INTERVAL 4 /* 4sec */ #define MAX_RTR_SOLICITATIONS 3 #define ND6_INFINITE_LIFETIME 0xffffffff #ifdef _KERNEL /* node constants */ #define MAX_REACHABLE_TIME 3600000 /* msec */ #define REACHABLE_TIME 30000 /* msec */ #define RETRANS_TIMER 1000 /* msec */ #define MIN_RANDOM_FACTOR 512 /* 1024 * 0.5 */ #define MAX_RANDOM_FACTOR 1536 /* 1024 * 1.5 */ #define DEF_TEMP_VALID_LIFETIME 172800 /* 2 days */ #define DEF_TEMP_PREFERRED_LIFETIME 86400 /* 1 day */ #define TEMPADDR_REGEN_ADVANCE 5 /* sec */ #define TEMP_MAX_DESYNC_FACTOR_BASE 300 /* 5 min */ #define ND_COMPUTE_RTIME(x) \ (((MIN_RANDOM_FACTOR * (x >> 10)) + (arc4random() & \ ((MAX_RANDOM_FACTOR - MIN_RANDOM_FACTOR) * (x >> 10)))) /1000) struct nd_defrouter { TAILQ_ENTRY(nd_defrouter) dr_entry; struct in6_addr rtaddr; u_char raflags; /* flags on RA message */ u_short rtlifetime; u_long expire; struct ifnet *ifp; int installed; /* is installed into kernel routing table */ u_int refcnt; }; struct nd_prefixctl { struct ifnet *ndpr_ifp; /* prefix */ struct sockaddr_in6 ndpr_prefix; u_char ndpr_plen; u_int32_t ndpr_vltime; /* advertised valid lifetime */ u_int32_t ndpr_pltime; /* advertised preferred lifetime */ struct prf_ra ndpr_flags; }; LIST_HEAD(nd_prhead, nd_prefix); struct nd_prefix { struct ifnet *ndpr_ifp; LIST_ENTRY(nd_prefix) ndpr_entry; struct sockaddr_in6 ndpr_prefix; /* prefix */ struct in6_addr ndpr_mask; /* netmask derived from the prefix */ u_int32_t ndpr_vltime; /* advertised valid lifetime */ u_int32_t ndpr_pltime; /* advertised preferred lifetime */ time_t ndpr_expire; /* expiration time of the prefix */ time_t ndpr_preferred; /* preferred time of the prefix */ time_t ndpr_lastupdate; /* reception time of last advertisement */ struct prf_ra ndpr_flags; u_int32_t ndpr_stateflags; /* actual state flags */ /* list of routers that advertise the prefix: */ LIST_HEAD(pr_rtrhead, nd_pfxrouter) ndpr_advrtrs; u_char ndpr_plen; int ndpr_addrcnt; /* count of derived addresses */ volatile u_int ndpr_refcnt; }; #define ndpr_raf ndpr_flags #define ndpr_raf_onlink ndpr_flags.onlink #define ndpr_raf_auto ndpr_flags.autonomous #define ndpr_raf_router ndpr_flags.router struct nd_pfxrouter { LIST_ENTRY(nd_pfxrouter) pfr_entry; struct nd_defrouter *router; }; #ifdef MALLOC_DECLARE MALLOC_DECLARE(M_IP6NDP); #endif /* nd6.c */ VNET_DECLARE(int, nd6_mmaxtries); VNET_DECLARE(struct nd_prhead, nd_prefix); VNET_DECLARE(int, nd6_debug); #define V_nd6_mmaxtries VNET(nd6_mmaxtries) #define V_nd_prefix VNET(nd_prefix) #define V_nd6_debug VNET(nd6_debug) /* Lock for the prefix and default router lists. */ VNET_DECLARE(struct rwlock, nd6_lock); VNET_DECLARE(uint64_t, nd6_list_genid); #define V_nd6_lock VNET(nd6_lock) #define V_nd6_list_genid VNET(nd6_list_genid) #define ND6_RLOCK() rw_rlock(&V_nd6_lock) #define ND6_RUNLOCK() rw_runlock(&V_nd6_lock) #define ND6_WLOCK() rw_wlock(&V_nd6_lock) #define ND6_WUNLOCK() rw_wunlock(&V_nd6_lock) #define ND6_TRY_UPGRADE() rw_try_upgrade(&V_nd6_lock) #define ND6_WLOCK_ASSERT() rw_assert(&V_nd6_lock, RA_WLOCKED) #define ND6_RLOCK_ASSERT() rw_assert(&V_nd6_lock, RA_RLOCKED) #define ND6_LOCK_ASSERT() rw_assert(&V_nd6_lock, RA_LOCKED) #define ND6_UNLOCK_ASSERT() rw_assert(&V_nd6_lock, RA_UNLOCKED) /* Mutex for prefix onlink/offlink transitions. */ VNET_DECLARE(struct mtx, nd6_onlink_mtx); #define V_nd6_onlink_mtx VNET(nd6_onlink_mtx) #define ND6_ONLINK_LOCK() mtx_lock(&V_nd6_onlink_mtx) #define ND6_ONLINK_TRYLOCK() mtx_trylock(&V_nd6_onlink_mtx) #define ND6_ONLINK_UNLOCK() mtx_unlock(&V_nd6_onlink_mtx) #define ND6_ONLINK_LOCK_ASSERT() mtx_assert(&V_nd6_onlink_mtx, MA_OWNED) #define ND6_ONLINK_UNLOCK_ASSERT() mtx_assert(&V_nd6_onlink_mtx, MA_NOTOWNED) #define nd6log(x) do { if (V_nd6_debug) log x; } while (/*CONSTCOND*/ 0) /* nd6_rtr.c */ VNET_DECLARE(int, nd6_defifindex); VNET_DECLARE(int, ip6_desync_factor); /* seconds */ VNET_DECLARE(uint32_t, ip6_temp_max_desync_factor); /* seconds */ VNET_DECLARE(u_int32_t, ip6_temp_preferred_lifetime); /* seconds */ VNET_DECLARE(u_int32_t, ip6_temp_valid_lifetime); /* seconds */ VNET_DECLARE(int, ip6_temp_regen_advance); /* seconds */ #define V_nd6_defifindex VNET(nd6_defifindex) #define V_ip6_desync_factor VNET(ip6_desync_factor) #define V_ip6_temp_max_desync_factor VNET(ip6_temp_max_desync_factor) #define V_ip6_temp_preferred_lifetime VNET(ip6_temp_preferred_lifetime) #define V_ip6_temp_valid_lifetime VNET(ip6_temp_valid_lifetime) #define V_ip6_temp_regen_advance VNET(ip6_temp_regen_advance) union nd_opts { struct nd_opt_hdr *nd_opt_array[16]; /* max = ND_OPT_NONCE */ struct { struct nd_opt_hdr *zero; struct nd_opt_hdr *src_lladdr; struct nd_opt_hdr *tgt_lladdr; struct nd_opt_prefix_info *pi_beg; /* multiple opts, start */ struct nd_opt_rd_hdr *rh; struct nd_opt_mtu *mtu; struct nd_opt_hdr *__res6; struct nd_opt_hdr *__res7; struct nd_opt_hdr *__res8; struct nd_opt_hdr *__res9; struct nd_opt_hdr *__res10; struct nd_opt_hdr *__res11; struct nd_opt_hdr *__res12; struct nd_opt_hdr *__res13; struct nd_opt_nonce *nonce; struct nd_opt_hdr *__res15; struct nd_opt_hdr *search; /* multiple opts */ struct nd_opt_hdr *last; /* multiple opts */ int done; struct nd_opt_prefix_info *pi_end;/* multiple opts, end */ } nd_opt_each; }; #define nd_opts_src_lladdr nd_opt_each.src_lladdr #define nd_opts_tgt_lladdr nd_opt_each.tgt_lladdr #define nd_opts_pi nd_opt_each.pi_beg #define nd_opts_pi_end nd_opt_each.pi_end #define nd_opts_rh nd_opt_each.rh #define nd_opts_mtu nd_opt_each.mtu #define nd_opts_nonce nd_opt_each.nonce #define nd_opts_search nd_opt_each.search #define nd_opts_last nd_opt_each.last #define nd_opts_done nd_opt_each.done /* XXX: need nd6_var.h?? */ /* nd6.c */ void nd6_init(void); #ifdef VIMAGE void nd6_destroy(void); #endif struct nd_ifinfo *nd6_ifattach(struct ifnet *); void nd6_ifdetach(struct ifnet *, struct nd_ifinfo *); int nd6_is_addr_neighbor(const struct sockaddr_in6 *, struct ifnet *); void nd6_option_init(void *, int, union nd_opts *); struct nd_opt_hdr *nd6_option(union nd_opts *); int nd6_options(union nd_opts *); struct llentry *nd6_lookup(const struct in6_addr *, int, struct ifnet *); void nd6_setmtu(struct ifnet *); void nd6_llinfo_setstate(struct llentry *lle, int newstate); void nd6_timer(void *); void nd6_purge(struct ifnet *); int nd6_resolve_addr(struct ifnet *ifp, int flags, const struct sockaddr *dst, char *desten, uint32_t *pflags); int nd6_resolve(struct ifnet *, int, struct mbuf *, const struct sockaddr *, u_char *, uint32_t *, struct llentry **); int nd6_ioctl(u_long, caddr_t, struct ifnet *); void nd6_cache_lladdr(struct ifnet *, struct in6_addr *, char *, int, int, int); bool nd6_try_set_entry_addr(struct ifnet *ifp, struct llentry *lle, char *lladdr); struct mbuf *nd6_grab_holdchain(struct llentry *); int nd6_flush_holdchain(struct ifnet *, struct llentry *, struct mbuf *); void nd6_flush_children_holdchain(struct ifnet *, struct llentry *); int nd6_add_ifa_lle(struct in6_ifaddr *); void nd6_rem_ifa_lle(struct in6_ifaddr *, int); int nd6_output_ifp(struct ifnet *, struct ifnet *, struct mbuf *, struct sockaddr_in6 *, struct route *); struct rib_head; struct rib_cmd_info; void nd6_subscription_cb(struct rib_head *rnh, struct rib_cmd_info *rc, void *arg); /* nd6_nbr.c */ void nd6_na_input(struct mbuf *, int, int); void nd6_na_output(struct ifnet *, const struct in6_addr *, const struct in6_addr *, u_long, int, struct sockaddr *); void nd6_ns_input(struct mbuf *, int, int); void nd6_ns_output(struct ifnet *, const struct in6_addr *, const struct in6_addr *, const struct in6_addr *, uint8_t *); caddr_t nd6_ifptomac(struct ifnet *); void nd6_dad_init(void); void nd6_dad_start(struct ifaddr *, int); void nd6_dad_stop(struct ifaddr *); /* nd6_rtr.c */ void nd6_rs_input(struct mbuf *, int, int); void nd6_ra_input(struct mbuf *, int, int); void nd6_ifnet_link_event(void *, struct ifnet *, int); struct nd_defrouter *defrouter_lookup(const struct in6_addr *, struct ifnet *); struct nd_defrouter *defrouter_lookup_locked(const struct in6_addr *, struct ifnet *); void defrouter_reset(void); void defrouter_select_fib(int fibnum); void defrouter_rele(struct nd_defrouter *); bool defrouter_remove(struct in6_addr *, struct ifnet *); bool nd6_defrouter_list_empty(void); void nd6_defrouter_flush_all(void); void nd6_defrouter_purge(struct ifnet *); void nd6_defrouter_timer(void); void nd6_defrouter_init(void); int nd6_prelist_add(struct nd_prefixctl *, struct nd_defrouter *, struct nd_prefix **); void nd6_prefix_unlink(struct nd_prefix *, struct nd_prhead *); void nd6_prefix_del(struct nd_prefix *); void nd6_prefix_ref(struct nd_prefix *); void nd6_prefix_rele(struct nd_prefix *); int nd6_prefix_offlink(struct nd_prefix *); void pfxlist_onlink_check(void); struct nd_prefix *nd6_prefix_lookup(struct nd_prefixctl *); void rt6_flush(struct in6_addr *, struct ifnet *); int nd6_setdefaultiface(int); int in6_tmpifadd(const struct in6_ifaddr *, int, int); #endif /* _KERNEL */ #endif /* _NETINET6_ND6_H_ */ diff --git a/sys/netinet6/nd6_nbr.c b/sys/netinet6/nd6_nbr.c index cc17b4e1a402..e2db192bca15 100644 --- a/sys/netinet6/nd6_nbr.c +++ b/sys/netinet6/nd6_nbr.c @@ -1,1644 +1,1631 @@ /*- * SPDX-License-Identifier: BSD-3-Clause * * Copyright (C) 1995, 1996, 1997, and 1998 WIDE Project. * All rights reserved. * * Redistribution and use in source and binary forms, with or without * modification, are permitted provided that the following conditions * are met: * 1. Redistributions of source code must retain the above copyright * notice, this list of conditions and the following disclaimer. * 2. Redistributions in binary form must reproduce the above copyright * notice, this list of conditions and the following disclaimer in the * documentation and/or other materials provided with the distribution. * 3. Neither the name of the project nor the names of its contributors * may be used to endorse or promote products derived from this software * without specific prior written permission. * * THIS SOFTWARE IS PROVIDED BY THE PROJECT AND CONTRIBUTORS ``AS IS'' AND * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE * ARE DISCLAIMED. IN NO EVENT SHALL THE PROJECT OR CONTRIBUTORS BE LIABLE * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF * SUCH DAMAGE. * * $KAME: nd6_nbr.c,v 1.86 2002/01/21 02:33:04 jinmei Exp $ */ #include #include "opt_inet.h" #include "opt_inet6.h" #include "opt_ipsec.h" #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #define SDL(s) ((struct sockaddr_dl *)s) struct dadq; static struct dadq *nd6_dad_find(struct ifaddr *, struct nd_opt_nonce *); static void nd6_dad_add(struct dadq *dp); static void nd6_dad_del(struct dadq *dp); static void nd6_dad_rele(struct dadq *); static void nd6_dad_starttimer(struct dadq *, int); static void nd6_dad_stoptimer(struct dadq *); static void nd6_dad_timer(void *); static void nd6_dad_duplicated(struct ifaddr *, struct dadq *); static void nd6_dad_ns_output(struct dadq *); static void nd6_dad_ns_input(struct ifaddr *, struct nd_opt_nonce *); static void nd6_dad_na_input(struct ifaddr *); static void nd6_na_output_fib(struct ifnet *, const struct in6_addr *, const struct in6_addr *, u_long, int, struct sockaddr *, u_int); static void nd6_ns_output_fib(struct ifnet *, const struct in6_addr *, const struct in6_addr *, const struct in6_addr *, uint8_t *, u_int); static struct ifaddr *nd6_proxy_fill_sdl(struct ifnet *, const struct in6_addr *, struct sockaddr_dl *); VNET_DEFINE_STATIC(int, dad_enhanced) = 1; #define V_dad_enhanced VNET(dad_enhanced) SYSCTL_DECL(_net_inet6_ip6); SYSCTL_INT(_net_inet6_ip6, OID_AUTO, dad_enhanced, CTLFLAG_VNET | CTLFLAG_RW, &VNET_NAME(dad_enhanced), 0, "Enable Enhanced DAD, which adds a random nonce to NS messages for DAD."); VNET_DEFINE_STATIC(int, dad_maxtry) = 15; /* max # of *tries* to transmit DAD packet */ #define V_dad_maxtry VNET(dad_maxtry) VNET_DEFINE_STATIC(int, nd6_onlink_ns_rfc4861) = 0; #define V_nd6_onlink_ns_rfc4861 VNET(nd6_onlink_ns_rfc4861) SYSCTL_INT(_net_inet6_icmp6, ICMPV6CTL_ND6_ONLINKNSRFC4861, nd6_onlink_ns_rfc4861, CTLFLAG_VNET | CTLFLAG_RW, &VNET_NAME(nd6_onlink_ns_rfc4861), 0, "Accept 'on-link' ICMPv6 NS messages in compliance with RFC 4861"); /* * 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; struct ip6_hdr *ip6; struct nd_neighbor_solicit *nd_ns; struct in6_addr daddr6, myaddr6, saddr6, taddr6; struct ifaddr *ifa; struct sockaddr_dl proxydl; union nd_opts ndopts; char ip6bufs[INET6_ADDRSTRLEN], ip6bufd[INET6_ADDRSTRLEN]; char *lladdr; int anycast, lladdrlen, proxy, rflag, tentative, tlladdr; ifa = NULL; /* RFC 6980: Nodes MUST silently ignore fragments */ if(m->m_flags & M_FRAGMENTED) goto freeit; ifp = m->m_pkthdr.rcvif; ip6 = mtod(m, struct ip6_hdr *); if (__predict_false(ip6->ip6_hlim != 255)) { ICMP6STAT_INC(icp6s_invlhlim); 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 bads; } if (m->m_len < off + icmp6len) { m = m_pullup(m, off + icmp6len); if (m == NULL) { IP6STAT_INC(ip6s_exthdrtoolong); return; } } ip6 = mtod(m, struct ip6_hdr *); nd_ns = (struct nd_neighbor_solicit *)((caddr_t)ip6 + off); saddr6 = ip6->ip6_src; daddr6 = ip6->ip6_dst; taddr6 = nd_ns->nd_ns_target; if (in6_setscope(&taddr6, ifp, NULL) != 0) goto bad; rflag = (V_ip6_forwarding) ? ND_NA_FLAG_ROUTER : 0; if (ND_IFINFO(ifp)->flags & ND6_IFF_ACCEPT_RTADV && V_ip6_norbit_raif) rflag = 0; 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; } lladdr = NULL; lladdrlen = 0; 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; + tlladdr = 0; + if (IN6_IS_ADDR_MULTICAST(&daddr6)) + tlladdr |= ND6_NA_OPT_LLA; /* * 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) + if (ifp->if_carp) { ifa = (*carp_iamatch6_p)(ifp, &taddr6); - else + if (ifa != NULL) + tlladdr |= ND6_NA_CARP_MASTER; + } else ifa = (struct ifaddr *)in6ifa_ifpwithaddr(ifp, &taddr6); /* (2) check. */ proxy = 0; if (ifa == NULL) { if ((ifa = nd6_proxy_fill_sdl(ifp, &taddr6, &proxydl)) != NULL) proxy = 1; } if (ifa == NULL) { /* * We've got an NS packet, and we don't have that address * 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, ndopts.nd_opts_nonce); goto freeit; } + /* + * If the Target Address is either an anycast address or a unicast + * address for which the node is providing proxy service, or the Target + * Link-Layer Address option is not included, the Override flag SHOULD + * be set to zero. Otherwise, the Override flag SHOULD be set to one. + */ + if (anycast == 0 && proxy == 0 && (tlladdr & ND6_NA_OPT_LLA) != 0) + rflag |= ND_NA_FLAG_OVERRIDE; /* * 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. + * It looks that sender is performing DAD. nd6_na_output() will + * send 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; + if (!IN6_IS_ADDR_UNSPECIFIED(&saddr6)) { + nd6_cache_lladdr(ifp, &saddr6, lladdr, lladdrlen, + ND_NEIGHBOR_SOLICIT, 0); + rflag |= ND_NA_FLAG_SOLICITED; } - 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, + nd6_na_output_fib(ifp, &saddr6, &taddr6, rflag, 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))); bads: ICMP6STAT_INC(icp6s_badns); if (ifa != NULL) ifa_free(ifa); m_freem(m); } static struct ifaddr * nd6_proxy_fill_sdl(struct ifnet *ifp, const struct in6_addr *taddr6, struct sockaddr_dl *sdl) { struct ifaddr *ifa; struct llentry *ln; ifa = NULL; ln = nd6_lookup(taddr6, LLE_SF(AF_INET6, 0), ifp); if (ln == NULL) return (ifa); if ((ln->la_flags & (LLE_PUB | LLE_VALID)) == (LLE_PUB | LLE_VALID)) { link_init_sdl(ifp, (struct sockaddr *)sdl, ifp->if_type); sdl->sdl_alen = ifp->if_addrlen; bcopy(ln->ll_addr, &sdl->sdl_data, ifp->if_addrlen); LLE_RUNLOCK(ln); ifa = (struct ifaddr *)in6ifa_ifpforlinklocal(ifp, IN6_IFF_NOTREADY|IN6_IFF_ANYCAST); } else LLE_RUNLOCK(ln); return (ifa); } /* * Output a Neighbor Solicitation Message. Caller specifies: * - ICMP6 header source IP6 address * - ND6 header target IP6 address * - ND6 header source datalink address * * Based on RFC 2461 * Based on RFC 2462 (duplicate address detection) * * ln - for source address determination * nonce - If non-NULL, NS is used for duplicate address detection and * the value (length is ND_OPT_NONCE_LEN) is used as a random nonce. */ static void nd6_ns_output_fib(struct ifnet *ifp, const struct in6_addr *saddr6, const struct in6_addr *daddr6, const struct in6_addr *taddr6, uint8_t *nonce, u_int fibnum) { struct mbuf *m; struct m_tag *mtag; struct ip6_hdr *ip6; struct nd_neighbor_solicit *nd_ns; struct ip6_moptions im6o; int icmp6len; int maxlen; NET_EPOCH_ASSERT(); if (IN6_IS_ADDR_MULTICAST(taddr6)) return; /* estimate the size of message */ maxlen = sizeof(*ip6) + sizeof(*nd_ns); maxlen += (sizeof(struct nd_opt_hdr) + ifp->if_addrlen + 7) & ~7; KASSERT(max_linkhdr + maxlen <= MCLBYTES, ( "%s: max_linkhdr + maxlen > MCLBYTES (%d + %d > %d)", __func__, max_linkhdr, maxlen, MCLBYTES)); if (max_linkhdr + maxlen > MHLEN) m = m_getcl(M_NOWAIT, MT_DATA, M_PKTHDR); else m = m_gethdr(M_NOWAIT, MT_DATA); if (m == NULL) return; M_SETFIB(m, fibnum); - if (daddr6 == NULL || IN6_IS_ADDR_MULTICAST(daddr6)) { - m->m_flags |= M_MCAST; - im6o.im6o_multicast_ifp = ifp; - im6o.im6o_multicast_hlim = 255; - im6o.im6o_multicast_loop = 0; - } - icmp6len = sizeof(*nd_ns); m->m_pkthdr.len = m->m_len = sizeof(*ip6) + icmp6len; m->m_data += max_linkhdr; /* or M_ALIGN() equivalent? */ /* fill neighbor solicitation packet */ ip6 = mtod(m, struct ip6_hdr *); ip6->ip6_flow = 0; ip6->ip6_vfc &= ~IPV6_VERSION_MASK; ip6->ip6_vfc |= IPV6_VERSION; /* ip6->ip6_plen will be set later */ ip6->ip6_nxt = IPPROTO_ICMPV6; ip6->ip6_hlim = 255; if (daddr6) ip6->ip6_dst = *daddr6; else { ip6->ip6_dst.s6_addr16[0] = IPV6_ADDR_INT16_MLL; ip6->ip6_dst.s6_addr16[1] = 0; ip6->ip6_dst.s6_addr32[1] = 0; ip6->ip6_dst.s6_addr32[2] = IPV6_ADDR_INT32_ONE; ip6->ip6_dst.s6_addr32[3] = taddr6->s6_addr32[3]; ip6->ip6_dst.s6_addr8[12] = 0xff; if (in6_setscope(&ip6->ip6_dst, ifp, NULL) != 0) goto bad; } + if (IN6_IS_ADDR_MULTICAST(&ip6->ip6_dst)) { + m->m_flags |= M_MCAST; + im6o.im6o_multicast_ifp = ifp; + im6o.im6o_multicast_hlim = 255; + im6o.im6o_multicast_loop = 0; + } if (nonce == NULL) { char ip6buf[INET6_ADDRSTRLEN]; struct ifaddr *ifa = NULL; /* * RFC2461 7.2.2: * "If the source address of the packet prompting the * solicitation is the same as one of the addresses assigned * to the outgoing interface, that address SHOULD be placed * in the IP Source Address of the outgoing solicitation. * Otherwise, any one of the addresses assigned to the * interface should be used." * * We use the source address for the prompting packet * (saddr6), if saddr6 belongs to the outgoing interface. * Otherwise, we perform the source address selection as usual. */ if (saddr6 != NULL) ifa = (struct ifaddr *)in6ifa_ifpwithaddr(ifp, saddr6); if (ifa == NULL) { int error; - struct in6_addr dst6, src6; - uint32_t scopeid; - in6_splitscope(&ip6->ip6_dst, &dst6, &scopeid); - error = in6_selectsrc_addr(fibnum, &dst6, - scopeid, ifp, &src6, NULL); + error = in6_selectsrc_nbr(fibnum, &ip6->ip6_dst, &im6o, + ifp, &ip6->ip6_src); if (error) { nd6log((LOG_DEBUG, "%s: source can't be " "determined: dst=%s, error=%d\n", __func__, - ip6_sprintf(ip6buf, &dst6), + ip6_sprintf(ip6buf, &ip6->ip6_dst), error)); goto bad; } - ip6->ip6_src = src6; } else ip6->ip6_src = *saddr6; if (ifp->if_carp != NULL) { /* * Check that selected source address belongs to * CARP addresses. */ if (ifa == NULL) ifa = (struct ifaddr *)in6ifa_ifpwithaddr(ifp, &ip6->ip6_src); /* * Do not send NS for CARP address if we are not * the CARP master. */ if (ifa != NULL && ifa->ifa_carp != NULL && !(*carp_master_p)(ifa)) { nd6log((LOG_DEBUG, "nd6_ns_output: NS from BACKUP CARP address %s\n", ip6_sprintf(ip6buf, &ip6->ip6_src))); ifa_free(ifa); goto bad; } } if (ifa != NULL) ifa_free(ifa); } else { /* * Source address for DAD packet must always be IPv6 * unspecified address. (0::0) * We actually don't have to 0-clear the address (we did it * above), but we do so here explicitly to make the intention * clearer. */ bzero(&ip6->ip6_src, sizeof(ip6->ip6_src)); } nd_ns = (struct nd_neighbor_solicit *)(ip6 + 1); nd_ns->nd_ns_type = ND_NEIGHBOR_SOLICIT; nd_ns->nd_ns_code = 0; nd_ns->nd_ns_reserved = 0; nd_ns->nd_ns_target = *taddr6; in6_clearscope(&nd_ns->nd_ns_target); /* XXX */ /* * Add source link-layer address option. * * spec implementation * --- --- * DAD packet MUST NOT do not add the option * there's no link layer address: * impossible do not add the option * there's link layer address: * Multicast NS MUST add one add the option * Unicast NS SHOULD add one add the option */ if (nonce == NULL) { struct nd_opt_hdr *nd_opt; char *mac; int optlen; mac = NULL; if (ifp->if_carp) mac = (*carp_macmatch6_p)(ifp, m, &ip6->ip6_src); if (mac == NULL) mac = nd6_ifptomac(ifp); if (mac != NULL) { nd_opt = (struct nd_opt_hdr *)(nd_ns + 1); optlen = sizeof(struct nd_opt_hdr) + ifp->if_addrlen; /* 8 byte alignments... */ optlen = (optlen + 7) & ~7; m->m_pkthdr.len += optlen; m->m_len += optlen; icmp6len += optlen; bzero(nd_opt, optlen); nd_opt->nd_opt_type = ND_OPT_SOURCE_LINKADDR; nd_opt->nd_opt_len = optlen >> 3; bcopy(mac, nd_opt + 1, ifp->if_addrlen); } } /* * Add a Nonce option (RFC 3971) to detect looped back NS messages. * This behavior is documented as Enhanced Duplicate Address * Detection in RFC 7527. * net.inet6.ip6.dad_enhanced=0 disables this. */ if (V_dad_enhanced != 0 && nonce != NULL) { int optlen = sizeof(struct nd_opt_hdr) + ND_OPT_NONCE_LEN; struct nd_opt_hdr *nd_opt = (struct nd_opt_hdr *)(nd_ns + 1); /* 8-byte alignment is required. */ optlen = (optlen + 7) & ~7; m->m_pkthdr.len += optlen; m->m_len += optlen; icmp6len += optlen; bzero((caddr_t)nd_opt, optlen); nd_opt->nd_opt_type = ND_OPT_NONCE; nd_opt->nd_opt_len = optlen >> 3; bcopy(nonce, (caddr_t)(nd_opt + 1), ND_OPT_NONCE_LEN); } ip6->ip6_plen = htons((u_short)icmp6len); nd_ns->nd_ns_cksum = 0; nd_ns->nd_ns_cksum = in6_cksum(m, IPPROTO_ICMPV6, sizeof(*ip6), icmp6len); if (send_sendso_input_hook != NULL) { mtag = m_tag_get(PACKET_TAG_ND_OUTGOING, sizeof(unsigned short), M_NOWAIT); if (mtag == NULL) goto bad; *(unsigned short *)(mtag + 1) = nd_ns->nd_ns_type; m_tag_prepend(m, mtag); } ip6_output(m, NULL, NULL, (nonce != NULL) ? IPV6_UNSPECSRC : 0, &im6o, NULL, NULL); icmp6_ifstat_inc(ifp, ifs6_out_msg); icmp6_ifstat_inc(ifp, ifs6_out_neighborsolicit); ICMP6STAT_INC2(icp6s_outhist, ND_NEIGHBOR_SOLICIT); return; bad: m_freem(m); } #ifndef BURN_BRIDGES void nd6_ns_output(struct ifnet *ifp, const struct in6_addr *saddr6, const struct in6_addr *daddr6, const struct in6_addr *taddr6,uint8_t *nonce) { nd6_ns_output_fib(ifp, saddr6, daddr6, taddr6, nonce, RT_DEFAULT_FIB); } #endif /* * Neighbor advertisement input handling. * * Based on RFC 2461 * Based on RFC 2462 (duplicate address detection) * * the following items are not implemented yet: * - proxy advertisement delay rule (RFC2461 7.2.8, last paragraph, SHOULD) * - anycast advertisement delay rule (RFC2461 7.2.7, SHOULD) */ void nd6_na_input(struct mbuf *m, int off, int icmp6len) { struct ifnet *ifp; struct ip6_hdr *ip6; struct ifaddr *ifa; struct llentry *ln; struct mbuf *chain; struct nd_neighbor_advert *nd_na; struct in6_addr daddr6, taddr6; union nd_opts ndopts; u_char linkhdr[LLE_MAX_LINKHDR]; char ip6bufs[INET6_ADDRSTRLEN], ip6bufd[INET6_ADDRSTRLEN]; char *lladdr; size_t linkhdrsize; int flags, is_override, is_router, is_solicited; int lladdr_off, lladdrlen, checklink; bool flush_holdchain = false; NET_EPOCH_ASSERT(); chain = NULL; ln = NULL; checklink = 0; /* RFC 6980: Nodes MUST silently ignore fragments */ if(m->m_flags & M_FRAGMENTED) goto freeit; ifp = m->m_pkthdr.rcvif; ip6 = mtod(m, struct ip6_hdr *); if (__predict_false(ip6->ip6_hlim != 255)) { ICMP6STAT_INC(icp6s_invlhlim); nd6log((LOG_ERR, "nd6_na_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 (m->m_len < off + icmp6len) { m = m_pullup(m, off + icmp6len); if (m == NULL) { IP6STAT_INC(ip6s_exthdrtoolong); return; } } ip6 = mtod(m, struct ip6_hdr *); nd_na = (struct nd_neighbor_advert *)((caddr_t)ip6 + off); flags = nd_na->nd_na_flags_reserved; is_router = ((flags & ND_NA_FLAG_ROUTER) != 0); is_solicited = ((flags & ND_NA_FLAG_SOLICITED) != 0); is_override = ((flags & ND_NA_FLAG_OVERRIDE) != 0); taddr6 = nd_na->nd_na_target; if (in6_setscope(&taddr6, ifp, NULL)) goto bad; /* XXX: impossible */ if (IN6_IS_ADDR_MULTICAST(&taddr6)) { nd6log((LOG_ERR, "nd6_na_input: invalid target address %s\n", ip6_sprintf(ip6bufs, &taddr6))); goto bad; } daddr6 = ip6->ip6_dst; if (IN6_IS_ADDR_MULTICAST(&daddr6)) if (is_solicited) { nd6log((LOG_ERR, "nd6_na_input: a solicited adv is multicasted\n")); goto bad; } icmp6len -= sizeof(*nd_na); nd6_option_init(nd_na + 1, icmp6len, &ndopts); if (nd6_options(&ndopts) < 0) { nd6log((LOG_INFO, "nd6_na_input: invalid ND option, ignored\n")); /* nd6_options have incremented stats */ goto freeit; } lladdr = NULL; lladdrlen = 0; if (ndopts.nd_opts_tgt_lladdr) { lladdr = (char *)(ndopts.nd_opts_tgt_lladdr + 1); lladdrlen = ndopts.nd_opts_tgt_lladdr->nd_opt_len << 3; } ifa = (struct ifaddr *)in6ifa_ifpwithaddr(ifp, &taddr6); if (ifa != NULL && ifa->ifa_carp != NULL) { /* * Silently ignore NAs for CARP addresses if we are not * the CARP master. */ if (!(*carp_master_p)(ifa)) { nd6log((LOG_DEBUG, "nd6_na_input: NA for BACKUP CARP address %s\n", ip6_sprintf(ip6bufs, &taddr6))); ifa_free(ifa); goto freeit; } } /* * Target address matches one of my interface address. * * If my address is tentative, this means that there's somebody * already using the same address as mine. This indicates DAD failure. * This is defined in RFC 2462. * * Otherwise, process as defined in RFC 2461. */ if (ifa && (((struct in6_ifaddr *)ifa)->ia6_flags & IN6_IFF_TENTATIVE)) { nd6_dad_na_input(ifa); ifa_free(ifa); goto freeit; } /* Just for safety, maybe unnecessary. */ if (ifa) { ifa_free(ifa); log(LOG_ERR, "nd6_na_input: duplicate IP6 address %s\n", ip6_sprintf(ip6bufs, &taddr6)); goto freeit; } if (lladdr && ((ifp->if_addrlen + 2 + 7) & ~7) != lladdrlen) { nd6log((LOG_INFO, "nd6_na_input: lladdrlen mismatch for %s " "(if %d, NA packet %d)\n", ip6_sprintf(ip6bufs, &taddr6), ifp->if_addrlen, lladdrlen - 2)); goto bad; } /* * If no neighbor cache entry is found, NA SHOULD silently be * discarded. */ ln = nd6_lookup(&taddr6, LLE_SF(AF_INET6, LLE_EXCLUSIVE), ifp); if (ln == NULL) { goto freeit; } /* * Do not try to override static entry. */ if (ln->la_flags & LLE_STATIC) goto freeit; if (ln->ln_state == ND6_LLINFO_INCOMPLETE) { /* * If the link-layer has address, and no lladdr option came, * discard the packet. */ if (ifp->if_addrlen && lladdr == NULL) { goto freeit; } /* * Record link-layer address, and update the state. */ if (!nd6_try_set_entry_addr(ifp, ln, lladdr)) goto freeit; flush_holdchain = true; if (is_solicited) nd6_llinfo_setstate(ln, ND6_LLINFO_REACHABLE); else nd6_llinfo_setstate(ln, ND6_LLINFO_STALE); EVENTHANDLER_INVOKE(lle_event, ln, LLENTRY_RESOLVED); if ((ln->ln_router = is_router) != 0) { /* * This means a router's state has changed from * non-reachable to probably reachable, and might * affect the status of associated prefixes.. */ checklink = 1; } } else { int llchange; /* * Check if the link-layer address has changed or not. */ if (lladdr == NULL) llchange = 0; else { if (ln->la_flags & LLE_VALID) { if (bcmp(lladdr, ln->ll_addr, ifp->if_addrlen)) llchange = 1; else llchange = 0; } else llchange = 1; } /* * This is VERY complex. Look at it with care. * * override solicit lladdr llchange action * (L: record lladdr) * * 0 0 n -- (2c) * 0 0 y n (2b) L * 0 0 y y (1) REACHABLE->STALE * 0 1 n -- (2c) *->REACHABLE * 0 1 y n (2b) L *->REACHABLE * 0 1 y y (1) REACHABLE->STALE * 1 0 n -- (2a) * 1 0 y n (2a) L * 1 0 y y (2a) L *->STALE * 1 1 n -- (2a) *->REACHABLE * 1 1 y n (2a) L *->REACHABLE * 1 1 y y (2a) L *->REACHABLE */ if (!is_override && (lladdr != NULL && llchange)) { /* (1) */ /* * If state is REACHABLE, make it STALE. * no other updates should be done. */ if (ln->ln_state == ND6_LLINFO_REACHABLE) nd6_llinfo_setstate(ln, ND6_LLINFO_STALE); goto freeit; } else if (is_override /* (2a) */ || (!is_override && (lladdr != NULL && !llchange)) /* (2b) */ || lladdr == NULL) { /* (2c) */ /* * Update link-local address, if any. */ if (lladdr != NULL) { linkhdrsize = sizeof(linkhdr); if (lltable_calc_llheader(ifp, AF_INET6, lladdr, linkhdr, &linkhdrsize, &lladdr_off) != 0) goto freeit; if (lltable_try_set_entry_addr(ifp, ln, linkhdr, linkhdrsize, lladdr_off) == 0) goto freeit; EVENTHANDLER_INVOKE(lle_event, ln, LLENTRY_RESOLVED); } /* * If solicited, make the state REACHABLE. * If not solicited and the link-layer address was * changed, make it STALE. */ if (is_solicited) nd6_llinfo_setstate(ln, ND6_LLINFO_REACHABLE); else { if (lladdr != NULL && llchange) nd6_llinfo_setstate(ln, ND6_LLINFO_STALE); } } if (ln->ln_router && !is_router) { /* * The peer dropped the router flag. * Remove the sender from the Default Router List and * update the Destination Cache entries. */ struct ifnet *nd6_ifp; nd6_ifp = lltable_get_ifp(ln->lle_tbl); if (!defrouter_remove(&ln->r_l3addr.addr6, nd6_ifp) && (ND_IFINFO(nd6_ifp)->flags & ND6_IFF_ACCEPT_RTADV) != 0) /* * Even if the neighbor is not in the default * router list, the neighbor may be used as a * next hop for some destinations (e.g. redirect * case). So we must call rt6_flush explicitly. */ rt6_flush(&ip6->ip6_src, ifp); } ln->ln_router = is_router; } /* XXX - QL * Does this matter? * rt->rt_flags &= ~RTF_REJECT; */ ln->la_asked = 0; if (ln->la_hold != NULL) chain = nd6_grab_holdchain(ln); freeit: if (ln != NULL) LLE_WUNLOCK(ln); if (chain != NULL) nd6_flush_holdchain(ifp, ln, chain); if (flush_holdchain) nd6_flush_children_holdchain(ifp, ln); if (checklink) pfxlist_onlink_check(); m_freem(m); return; bad: if (ln != NULL) LLE_WUNLOCK(ln); ICMP6STAT_INC(icp6s_badna); m_freem(m); } /* * Neighbor advertisement output handling. * * Based on RFC 2461 * * the following items are not implemented yet: * - proxy advertisement delay rule (RFC2461 7.2.8, last paragraph, SHOULD) * - anycast advertisement delay rule (RFC2461 7.2.7, SHOULD) * - * tlladdr - 1 if include target link-layer address + * tlladdr: + * - 0x01 if include target link-layer address + * - 0x02 if target address is CARP MASTER * sdl0 - sockaddr_dl (= proxy NA) or NULL */ static void nd6_na_output_fib(struct ifnet *ifp, const struct in6_addr *daddr6_0, const struct in6_addr *taddr6, u_long flags, int tlladdr, struct sockaddr *sdl0, u_int fibnum) { struct mbuf *m; struct m_tag *mtag; struct ip6_hdr *ip6; struct nd_neighbor_advert *nd_na; struct ip6_moptions im6o; - struct in6_addr daddr6, dst6, src6; - uint32_t scopeid; + struct in6_addr daddr6; NET_EPOCH_ASSERT(); int icmp6len, maxlen, error; caddr_t mac = NULL; daddr6 = *daddr6_0; /* make a local copy for modification */ /* estimate the size of message */ maxlen = sizeof(*ip6) + sizeof(*nd_na); maxlen += (sizeof(struct nd_opt_hdr) + ifp->if_addrlen + 7) & ~7; KASSERT(max_linkhdr + maxlen <= MCLBYTES, ( "%s: max_linkhdr + maxlen > MCLBYTES (%d + %d > %d)", __func__, max_linkhdr, maxlen, MCLBYTES)); if (max_linkhdr + maxlen > MHLEN) m = m_getcl(M_NOWAIT, MT_DATA, M_PKTHDR); else m = m_gethdr(M_NOWAIT, MT_DATA); if (m == NULL) return; M_SETFIB(m, fibnum); - if (IN6_IS_ADDR_MULTICAST(&daddr6)) { - m->m_flags |= M_MCAST; - im6o.im6o_multicast_ifp = ifp; - im6o.im6o_multicast_hlim = 255; - im6o.im6o_multicast_loop = 0; - } - icmp6len = sizeof(*nd_na); m->m_pkthdr.len = m->m_len = sizeof(struct ip6_hdr) + icmp6len; m->m_data += max_linkhdr; /* or M_ALIGN() equivalent? */ /* fill neighbor advertisement packet */ ip6 = mtod(m, struct ip6_hdr *); ip6->ip6_flow = 0; ip6->ip6_vfc &= ~IPV6_VERSION_MASK; ip6->ip6_vfc |= IPV6_VERSION; ip6->ip6_nxt = IPPROTO_ICMPV6; ip6->ip6_hlim = 255; + if (IN6_IS_ADDR_UNSPECIFIED(&daddr6)) { /* reply to DAD */ - daddr6.s6_addr16[0] = IPV6_ADDR_INT16_MLL; - daddr6.s6_addr16[1] = 0; - daddr6.s6_addr32[1] = 0; - daddr6.s6_addr32[2] = 0; - daddr6.s6_addr32[3] = IPV6_ADDR_INT32_ONE; + daddr6 = in6addr_linklocal_allnodes; if (in6_setscope(&daddr6, ifp, NULL)) goto bad; flags &= ~ND_NA_FLAG_SOLICITED; } - ip6->ip6_dst = daddr6; + if (IN6_IS_ADDR_MULTICAST(&daddr6)) { + m->m_flags |= M_MCAST; + im6o.im6o_multicast_ifp = ifp; + im6o.im6o_multicast_hlim = 255; + im6o.im6o_multicast_loop = 0; + } - /* - * Select a source whose scope is the same as that of the dest. - */ - in6_splitscope(&daddr6, &dst6, &scopeid); - error = in6_selectsrc_addr(fibnum, &dst6, - scopeid, ifp, &src6, NULL); + ip6->ip6_dst = daddr6; + error = in6_selectsrc_nbr(fibnum, &daddr6, &im6o, ifp, &ip6->ip6_src); if (error) { char ip6buf[INET6_ADDRSTRLEN]; nd6log((LOG_DEBUG, "nd6_na_output: source can't be " "determined: dst=%s, error=%d\n", ip6_sprintf(ip6buf, &daddr6), error)); goto bad; } - ip6->ip6_src = src6; nd_na = (struct nd_neighbor_advert *)(ip6 + 1); nd_na->nd_na_type = ND_NEIGHBOR_ADVERT; nd_na->nd_na_code = 0; nd_na->nd_na_target = *taddr6; in6_clearscope(&nd_na->nd_na_target); /* XXX */ + /* + * If we respond from CARP address, we need to prepare mac address + * for carp_output(). + */ + if (ifp->if_carp && (tlladdr & ND6_NA_CARP_MASTER)) + mac = (*carp_macmatch6_p)(ifp, m, taddr6); /* * "tlladdr" indicates NS's condition for adding tlladdr or not. * see nd6_ns_input() for details. * Basically, if NS packet is sent to unicast/anycast addr, * target lladdr option SHOULD NOT be included. */ - if (tlladdr) { + if (tlladdr & ND6_NA_OPT_LLA) { /* * sdl0 != NULL indicates proxy NA. If we do proxy, use * lladdr in sdl0. If we are not proxying (sending NA for * my address) use lladdr configured for the interface. */ if (sdl0 == NULL) { - if (ifp->if_carp) - mac = (*carp_macmatch6_p)(ifp, m, taddr6); if (mac == NULL) mac = nd6_ifptomac(ifp); } else if (sdl0->sa_family == AF_LINK) { struct sockaddr_dl *sdl; sdl = (struct sockaddr_dl *)sdl0; if (sdl->sdl_alen == ifp->if_addrlen) mac = LLADDR(sdl); } } - if (tlladdr && mac) { + if ((tlladdr & ND6_NA_OPT_LLA) && mac != NULL) { int optlen = sizeof(struct nd_opt_hdr) + ifp->if_addrlen; struct nd_opt_hdr *nd_opt = (struct nd_opt_hdr *)(nd_na + 1); /* roundup to 8 bytes alignment! */ optlen = (optlen + 7) & ~7; m->m_pkthdr.len += optlen; m->m_len += optlen; icmp6len += optlen; bzero((caddr_t)nd_opt, optlen); nd_opt->nd_opt_type = ND_OPT_TARGET_LINKADDR; nd_opt->nd_opt_len = optlen >> 3; bcopy(mac, (caddr_t)(nd_opt + 1), ifp->if_addrlen); } else flags &= ~ND_NA_FLAG_OVERRIDE; ip6->ip6_plen = htons((u_short)icmp6len); nd_na->nd_na_flags_reserved = flags; nd_na->nd_na_cksum = 0; nd_na->nd_na_cksum = in6_cksum(m, IPPROTO_ICMPV6, sizeof(struct ip6_hdr), icmp6len); if (send_sendso_input_hook != NULL) { mtag = m_tag_get(PACKET_TAG_ND_OUTGOING, sizeof(unsigned short), M_NOWAIT); if (mtag == NULL) goto bad; *(unsigned short *)(mtag + 1) = nd_na->nd_na_type; m_tag_prepend(m, mtag); } ip6_output(m, NULL, NULL, 0, &im6o, NULL, NULL); icmp6_ifstat_inc(ifp, ifs6_out_msg); icmp6_ifstat_inc(ifp, ifs6_out_neighboradvert); ICMP6STAT_INC2(icp6s_outhist, ND_NEIGHBOR_ADVERT); return; bad: m_freem(m); } #ifndef BURN_BRIDGES void nd6_na_output(struct ifnet *ifp, const struct in6_addr *daddr6_0, const struct in6_addr *taddr6, u_long flags, int tlladdr, struct sockaddr *sdl0) { nd6_na_output_fib(ifp, daddr6_0, taddr6, flags, tlladdr, sdl0, RT_DEFAULT_FIB); } #endif caddr_t nd6_ifptomac(struct ifnet *ifp) { switch (ifp->if_type) { case IFT_ETHER: case IFT_IEEE1394: case IFT_L2VLAN: case IFT_INFINIBAND: case IFT_BRIDGE: return IF_LLADDR(ifp); default: return NULL; } } struct dadq { TAILQ_ENTRY(dadq) dad_list; struct ifaddr *dad_ifa; int dad_count; /* max NS to send */ int dad_ns_tcount; /* # of trials to send NS */ int dad_ns_ocount; /* NS sent so far */ int dad_ns_icount; int dad_na_icount; int dad_ns_lcount; /* looped back NS */ int dad_loopbackprobe; /* probing state for loopback detection */ struct callout dad_timer_ch; struct vnet *dad_vnet; u_int dad_refcnt; #define ND_OPT_NONCE_LEN32 \ ((ND_OPT_NONCE_LEN + sizeof(uint32_t) - 1)/sizeof(uint32_t)) uint32_t dad_nonce[ND_OPT_NONCE_LEN32]; bool dad_ondadq; /* on dadq? Protected by DADQ_WLOCK. */ }; VNET_DEFINE_STATIC(TAILQ_HEAD(, dadq), dadq); VNET_DEFINE_STATIC(struct rwlock, dad_rwlock); #define V_dadq VNET(dadq) #define V_dad_rwlock VNET(dad_rwlock) #define DADQ_LOCKPTR() (&V_dad_rwlock) #define DADQ_LOCK_INIT() rw_init(DADQ_LOCKPTR(), "nd6 DAD queue") #define DADQ_RLOCK() rw_rlock(DADQ_LOCKPTR()) #define DADQ_RUNLOCK() rw_runlock(DADQ_LOCKPTR()) #define DADQ_WLOCK() rw_wlock(DADQ_LOCKPTR()) #define DADQ_WUNLOCK() rw_wunlock(DADQ_LOCKPTR()) #define DADQ_LOCK_ASSERT() rw_assert(DADQ_LOCKPTR(), RA_LOCKED); #define DADQ_RLOCK_ASSERT() rw_assert(DADQ_LOCKPTR(), RA_RLOCKED); #define DADQ_WLOCK_ASSERT() rw_assert(DADQ_LOCKPTR(), RA_WLOCKED); static void nd6_dad_add(struct dadq *dp) { DADQ_WLOCK_ASSERT(); TAILQ_INSERT_TAIL(&V_dadq, dp, dad_list); dp->dad_ondadq = true; } static void nd6_dad_del(struct dadq *dp) { DADQ_WLOCK_ASSERT(); if (dp->dad_ondadq) { /* * Remove dp from the dadq and release the dadq's * reference. */ TAILQ_REMOVE(&V_dadq, dp, dad_list); dp->dad_ondadq = false; nd6_dad_rele(dp); } } static struct dadq * nd6_dad_find(struct ifaddr *ifa, struct nd_opt_nonce *n) { struct dadq *dp; DADQ_LOCK_ASSERT(); TAILQ_FOREACH(dp, &V_dadq, dad_list) { if (dp->dad_ifa != ifa) continue; /* * Skip if the nonce matches the received one. * +2 in the length is required because of type and * length fields are included in a header. */ if (n != NULL && n->nd_opt_nonce_len == (ND_OPT_NONCE_LEN + 2) / 8 && memcmp(&n->nd_opt_nonce[0], &dp->dad_nonce[0], ND_OPT_NONCE_LEN) == 0) { dp->dad_ns_lcount++; continue; } break; } return (dp); } static void nd6_dad_starttimer(struct dadq *dp, int ticks) { DADQ_WLOCK_ASSERT(); callout_reset(&dp->dad_timer_ch, ticks, nd6_dad_timer, dp); } static void nd6_dad_stoptimer(struct dadq *dp) { callout_drain(&dp->dad_timer_ch); } static void nd6_dad_rele(struct dadq *dp) { if (refcount_release(&dp->dad_refcnt)) { KASSERT(!dp->dad_ondadq, ("dp %p still on DAD queue", dp)); ifa_free(dp->dad_ifa); free(dp, M_IP6NDP); } } void nd6_dad_init(void) { DADQ_LOCK_INIT(); TAILQ_INIT(&V_dadq); } /* * Start Duplicate Address Detection (DAD) for specified interface address. */ void nd6_dad_start(struct ifaddr *ifa, int delay) { struct in6_ifaddr *ia = (struct in6_ifaddr *)ifa; struct dadq *dp; char ip6buf[INET6_ADDRSTRLEN]; KASSERT((ia->ia6_flags & IN6_IFF_TENTATIVE) != 0, ("starting DAD on non-tentative address %p", ifa)); /* * If we don't need DAD, don't do it. * There are several cases: * - DAD is disabled globally or on the interface * - the interface address is anycast */ if ((ia->ia6_flags & IN6_IFF_ANYCAST) != 0 || V_ip6_dad_count == 0 || (ND_IFINFO(ifa->ifa_ifp)->flags & ND6_IFF_NO_DAD) != 0) { ia->ia6_flags &= ~IN6_IFF_TENTATIVE; return; } if ((ifa->ifa_ifp->if_flags & IFF_UP) == 0 || (ifa->ifa_ifp->if_drv_flags & IFF_DRV_RUNNING) == 0 || (ND_IFINFO(ifa->ifa_ifp)->flags & ND6_IFF_IFDISABLED) != 0) return; DADQ_WLOCK(); if ((dp = nd6_dad_find(ifa, NULL)) != NULL) { /* * DAD is already in progress. Let the existing entry * finish it. */ DADQ_WUNLOCK(); return; } dp = malloc(sizeof(*dp), M_IP6NDP, M_NOWAIT | M_ZERO); if (dp == NULL) { log(LOG_ERR, "nd6_dad_start: memory allocation failed for " "%s(%s)\n", ip6_sprintf(ip6buf, &ia->ia_addr.sin6_addr), ifa->ifa_ifp ? if_name(ifa->ifa_ifp) : "???"); return; } callout_init_rw(&dp->dad_timer_ch, DADQ_LOCKPTR(), CALLOUT_RETURNUNLOCKED); #ifdef VIMAGE dp->dad_vnet = curvnet; #endif nd6log((LOG_DEBUG, "%s: starting DAD for %s\n", if_name(ifa->ifa_ifp), ip6_sprintf(ip6buf, &ia->ia_addr.sin6_addr))); /* * Send NS packet for DAD, ip6_dad_count times. * Note that we must delay the first transmission, if this is the * first packet to be sent from the interface after interface * (re)initialization. */ dp->dad_ifa = ifa; ifa_ref(dp->dad_ifa); dp->dad_count = V_ip6_dad_count; dp->dad_ns_icount = dp->dad_na_icount = 0; dp->dad_ns_ocount = dp->dad_ns_tcount = 0; dp->dad_ns_lcount = dp->dad_loopbackprobe = 0; /* Add this to the dadq and add a reference for the dadq. */ refcount_init(&dp->dad_refcnt, 1); nd6_dad_add(dp); nd6_dad_starttimer(dp, delay); DADQ_WUNLOCK(); } /* * terminate DAD unconditionally. used for address removals. */ void nd6_dad_stop(struct ifaddr *ifa) { struct dadq *dp; DADQ_WLOCK(); dp = nd6_dad_find(ifa, NULL); if (dp == NULL) { DADQ_WUNLOCK(); /* DAD wasn't started yet */ return; } /* * Acquire a temporary reference so that we can safely stop the callout. */ (void)refcount_acquire(&dp->dad_refcnt); nd6_dad_del(dp); DADQ_WUNLOCK(); nd6_dad_stoptimer(dp); nd6_dad_rele(dp); } static void nd6_dad_timer(void *arg) { struct dadq *dp = arg; struct ifaddr *ifa = dp->dad_ifa; struct ifnet *ifp = dp->dad_ifa->ifa_ifp; struct in6_ifaddr *ia = (struct in6_ifaddr *)ifa; char ip6buf[INET6_ADDRSTRLEN]; struct epoch_tracker et; CURVNET_SET(dp->dad_vnet); KASSERT(ia != NULL, ("DAD entry %p with no address", dp)); NET_EPOCH_ENTER(et); if (ND_IFINFO(ifp)->flags & ND6_IFF_IFDISABLED) { /* Do not need DAD for ifdisabled interface. */ log(LOG_ERR, "nd6_dad_timer: cancel DAD on %s because of " "ND6_IFF_IFDISABLED.\n", ifp->if_xname); goto err; } if (ia->ia6_flags & IN6_IFF_DUPLICATED) { log(LOG_ERR, "nd6_dad_timer: called with duplicated address " "%s(%s)\n", ip6_sprintf(ip6buf, &ia->ia_addr.sin6_addr), ifa->ifa_ifp ? if_name(ifa->ifa_ifp) : "???"); goto err; } if ((ia->ia6_flags & IN6_IFF_TENTATIVE) == 0) { log(LOG_ERR, "nd6_dad_timer: called with non-tentative address " "%s(%s)\n", ip6_sprintf(ip6buf, &ia->ia_addr.sin6_addr), ifa->ifa_ifp ? if_name(ifa->ifa_ifp) : "???"); goto err; } /* Stop DAD if the interface is down even after dad_maxtry attempts. */ if ((dp->dad_ns_tcount > V_dad_maxtry) && (((ifp->if_flags & IFF_UP) == 0) || ((ifp->if_drv_flags & IFF_DRV_RUNNING) == 0))) { nd6log((LOG_INFO, "%s: could not run DAD " "because the interface was down or not running.\n", if_name(ifa->ifa_ifp))); goto err; } /* Need more checks? */ if (dp->dad_ns_ocount < dp->dad_count) { /* * We have more NS to go. Send NS packet for DAD. */ nd6_dad_starttimer(dp, (long)ND_IFINFO(ifa->ifa_ifp)->retrans * hz / 1000); nd6_dad_ns_output(dp); goto done; } else { /* * We have transmitted sufficient number of DAD packets. * See what we've got. */ if (dp->dad_ns_icount > 0 || dp->dad_na_icount > 0) { /* We've seen NS or NA, means DAD has failed. */ nd6_dad_duplicated(ifa, dp); } else if (V_dad_enhanced != 0 && dp->dad_ns_lcount > 0 && dp->dad_ns_lcount > dp->dad_loopbackprobe) { /* * Sec. 4.1 in RFC 7527 requires transmission of * additional probes until the loopback condition * becomes clear when a looped back probe is detected. */ log(LOG_ERR, "%s: a looped back NS message is " "detected during DAD for %s. " "Another DAD probes are being sent.\n", if_name(ifa->ifa_ifp), ip6_sprintf(ip6buf, IFA_IN6(ifa))); dp->dad_loopbackprobe = dp->dad_ns_lcount; /* * Send an NS immediately and increase dad_count by * V_nd6_mmaxtries - 1. */ dp->dad_count = dp->dad_ns_ocount + V_nd6_mmaxtries - 1; nd6_dad_starttimer(dp, (long)ND_IFINFO(ifa->ifa_ifp)->retrans * hz / 1000); nd6_dad_ns_output(dp); goto done; } else { /* * We are done with DAD. No NA came, no NS came. * No duplicate address found. Check IFDISABLED flag * again in case that it is changed between the * beginning of this function and here. * * Reset DAD failures counter if using stable addresses. */ if ((ND_IFINFO(ifp)->flags & ND6_IFF_IFDISABLED) == 0) { ia->ia6_flags &= ~IN6_IFF_TENTATIVE; if ((ND_IFINFO(ifp)->flags & ND6_IFF_STABLEADDR) && !(ia->ia6_flags & IN6_IFF_TEMPORARY)) counter_u64_zero(DAD_FAILURES(ifp)); } nd6log((LOG_DEBUG, "%s: DAD complete for %s - no duplicates found\n", if_name(ifa->ifa_ifp), ip6_sprintf(ip6buf, &ia->ia_addr.sin6_addr))); if (dp->dad_ns_lcount > 0) log(LOG_ERR, "%s: DAD completed while " "a looped back NS message is detected " "during DAD for %s.\n", if_name(ifa->ifa_ifp), ip6_sprintf(ip6buf, IFA_IN6(ifa))); } } err: nd6_dad_del(dp); DADQ_WUNLOCK(); done: NET_EPOCH_EXIT(et); CURVNET_RESTORE(); } static void nd6_dad_duplicated(struct ifaddr *ifa, struct dadq *dp) { struct in6_ifaddr *ia = (struct in6_ifaddr *)ifa; struct ifnet *ifp; char ip6buf[INET6_ADDRSTRLEN]; ifp = ifa->ifa_ifp; log(LOG_ERR, "%s: DAD detected duplicate IPv6 address %s: " "NS in/out/loopback=%d/%d/%d, NA in=%d\n", if_name(ifp), ip6_sprintf(ip6buf, &ia->ia_addr.sin6_addr), dp->dad_ns_icount, dp->dad_ns_ocount, dp->dad_ns_lcount, dp->dad_na_icount); ia->ia6_flags &= ~IN6_IFF_TENTATIVE; ia->ia6_flags |= IN6_IFF_DUPLICATED; log(LOG_ERR, "%s: DAD complete for %s - duplicate found\n", if_name(ifp), ip6_sprintf(ip6buf, &ia->ia_addr.sin6_addr)); /* * For RFC 7217 stable addresses, increment failure counter here if we still have retries. * More addresses will be generated as long as retries are not exhausted. */ if ((ND_IFINFO(ifp)->flags & ND6_IFF_STABLEADDR) && !(ia->ia6_flags & IN6_IFF_TEMPORARY)) { uint64_t dad_failures = counter_u64_fetch(DAD_FAILURES(ifp)); if (dad_failures <= V_ip6_stableaddr_maxretries) { counter_u64_add(DAD_FAILURES(ifp), 1); /* if retries exhausted, output an informative error message */ if (dad_failures == V_ip6_stableaddr_maxretries) log(LOG_ERR, "%s: manual intervention required, consider disabling \"stableaddr\" on the interface" " or checking hostuuid for uniqueness\n", if_name(ifp)); } } else { log(LOG_ERR, "%s: manual intervention required\n", if_name(ifp)); } /* * If the address is a link-local address formed from an interface * identifier based on the hardware address which is supposed to be * uniquely assigned (e.g., EUI-64 for an Ethernet interface), IP * operation on the interface SHOULD be disabled. * [RFC 4862, Section 5.4.5] */ if (IN6_IS_ADDR_LINKLOCAL(&ia->ia_addr.sin6_addr)) { struct in6_addr in6; /* * To avoid over-reaction, we only apply this logic when we are * very sure that hardware addresses are supposed to be unique. */ switch (ifp->if_type) { case IFT_ETHER: case IFT_ATM: case IFT_IEEE1394: case IFT_INFINIBAND: in6 = ia->ia_addr.sin6_addr; if (in6_get_hw_ifid(ifp, &in6) == 0 && IN6_ARE_ADDR_EQUAL(&ia->ia_addr.sin6_addr, &in6)) { ND_IFINFO(ifp)->flags |= ND6_IFF_IFDISABLED; log(LOG_ERR, "%s: possible hardware address " "duplication detected, disable IPv6\n", if_name(ifp)); } break; } } } /* * Transmit a neighbour solicitation for the purpose of DAD. Returns with the * DAD queue unlocked. */ static void nd6_dad_ns_output(struct dadq *dp) { struct in6_ifaddr *ia = (struct in6_ifaddr *)dp->dad_ifa; struct ifnet *ifp = dp->dad_ifa->ifa_ifp; int i; DADQ_WLOCK_ASSERT(); dp->dad_ns_tcount++; if ((ifp->if_flags & IFF_UP) == 0) { DADQ_WUNLOCK(); return; } if ((ifp->if_drv_flags & IFF_DRV_RUNNING) == 0) { DADQ_WUNLOCK(); return; } dp->dad_ns_ocount++; if (V_dad_enhanced != 0) { for (i = 0; i < ND_OPT_NONCE_LEN32; i++) dp->dad_nonce[i] = arc4random(); /* * XXXHRS: Note that in the case that * DupAddrDetectTransmits > 1, multiple NS messages with * different nonces can be looped back in an unexpected * order. The current implementation recognizes only * the latest nonce on the sender side. Practically it * should work well in almost all cases. */ } DADQ_WUNLOCK(); nd6_ns_output(ifp, NULL, NULL, &ia->ia_addr.sin6_addr, (uint8_t *)&dp->dad_nonce[0]); } static void nd6_dad_ns_input(struct ifaddr *ifa, struct nd_opt_nonce *ndopt_nonce) { struct dadq *dp; if (ifa == NULL) panic("ifa == NULL in nd6_dad_ns_input"); /* Ignore Nonce option when Enhanced DAD is disabled. */ if (V_dad_enhanced == 0) ndopt_nonce = NULL; DADQ_RLOCK(); dp = nd6_dad_find(ifa, ndopt_nonce); if (dp != NULL) dp->dad_ns_icount++; DADQ_RUNLOCK(); } static void nd6_dad_na_input(struct ifaddr *ifa) { struct dadq *dp; if (ifa == NULL) panic("ifa == NULL in nd6_dad_na_input"); DADQ_RLOCK(); dp = nd6_dad_find(ifa, NULL); if (dp != NULL) dp->dad_na_icount++; DADQ_RUNLOCK(); }