162587Sitojun/* $FreeBSD$ */ 295023Ssuz/* $KAME: if_stf.c,v 1.73 2001/12/03 11:08:30 keiichi Exp $ */ 362587Sitojun 4139823Simp/*- 562587Sitojun * Copyright (C) 2000 WIDE Project. 662587Sitojun * All rights reserved. 762587Sitojun * 862587Sitojun * Redistribution and use in source and binary forms, with or without 962587Sitojun * modification, are permitted provided that the following conditions 1062587Sitojun * are met: 1162587Sitojun * 1. Redistributions of source code must retain the above copyright 1262587Sitojun * notice, this list of conditions and the following disclaimer. 1362587Sitojun * 2. Redistributions in binary form must reproduce the above copyright 1462587Sitojun * notice, this list of conditions and the following disclaimer in the 1562587Sitojun * documentation and/or other materials provided with the distribution. 1662587Sitojun * 3. Neither the name of the project nor the names of its contributors 1762587Sitojun * may be used to endorse or promote products derived from this software 1862587Sitojun * without specific prior written permission. 1962587Sitojun * 2062587Sitojun * THIS SOFTWARE IS PROVIDED BY THE PROJECT AND CONTRIBUTORS ``AS IS'' AND 2162587Sitojun * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE 2262587Sitojun * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE 2362587Sitojun * ARE DISCLAIMED. IN NO EVENT SHALL THE PROJECT OR CONTRIBUTORS BE LIABLE 2462587Sitojun * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL 2562587Sitojun * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS 2662587Sitojun * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) 2762587Sitojun * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT 2862587Sitojun * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY 2962587Sitojun * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF 3062587Sitojun * SUCH DAMAGE. 3162587Sitojun */ 3262587Sitojun 3362587Sitojun/* 3478064Sume * 6to4 interface, based on RFC3056. 3562587Sitojun * 3662587Sitojun * 6to4 interface is NOT capable of link-layer (I mean, IPv4) multicasting. 3762587Sitojun * There is no address mapping defined from IPv6 multicast address to IPv4 3862587Sitojun * address. Therefore, we do not have IFF_MULTICAST on the interface. 3962587Sitojun * 4062587Sitojun * Due to the lack of address mapping for link-local addresses, we cannot 4162587Sitojun * throw packets toward link-local addresses (fe80::x). Also, we cannot throw 4262587Sitojun * packets to link-local multicast addresses (ff02::x). 4362587Sitojun * 4462587Sitojun * Here are interesting symptoms due to the lack of link-local address: 4562587Sitojun * 4662587Sitojun * Unicast routing exchange: 4762587Sitojun * - RIPng: Impossible. Uses link-local multicast packet toward ff02::9, 4862587Sitojun * and link-local addresses as nexthop. 4962587Sitojun * - OSPFv6: Impossible. OSPFv6 assumes that there's link-local address 5062587Sitojun * assigned to the link, and makes use of them. Also, HELLO packets use 5162587Sitojun * link-local multicast addresses (ff02::5 and ff02::6). 5262587Sitojun * - BGP4+: Maybe. You can only use global address as nexthop, and global 5362587Sitojun * address as TCP endpoint address. 5462587Sitojun * 5562587Sitojun * Multicast routing protocols: 5662587Sitojun * - PIM: Hello packet cannot be used to discover adjacent PIM routers. 5762587Sitojun * Adjacent PIM routers must be configured manually (is it really spec-wise 5862587Sitojun * correct thing to do?). 5962587Sitojun * 6062587Sitojun * ICMPv6: 6162587Sitojun * - Redirects cannot be used due to the lack of link-local address. 6262587Sitojun * 6378064Sume * stf interface does not have, and will not need, a link-local address. 6478064Sume * It seems to have no real benefit and does not help the above symptoms much. 6578064Sume * Even if we assign link-locals to interface, we cannot really 6678064Sume * use link-local unicast/multicast on top of 6to4 cloud (since there's no 6778064Sume * encapsulation defined for link-local address), and the above analysis does 6878064Sume * not change. RFC3056 does not mandate the assignment of link-local address 6978064Sume * either. 7062587Sitojun * 7162587Sitojun * 6to4 interface has security issues. Refer to 7262587Sitojun * http://playground.iijlab.net/i-d/draft-itojun-ipv6-transition-abuse-00.txt 7362587Sitojun * for details. The code tries to filter out some of malicious packets. 7462587Sitojun * Note that there is no way to be 100% secure. 7562587Sitojun */ 7662587Sitojun 7762587Sitojun#include "opt_inet.h" 7862587Sitojun#include "opt_inet6.h" 7962587Sitojun 8062587Sitojun#include <sys/param.h> 8162587Sitojun#include <sys/systm.h> 8262587Sitojun#include <sys/socket.h> 8362587Sitojun#include <sys/sockio.h> 8462587Sitojun#include <sys/mbuf.h> 8562587Sitojun#include <sys/errno.h> 8683655Sbrooks#include <sys/kernel.h> 87129880Sphk#include <sys/module.h> 8862587Sitojun#include <sys/protosw.h> 89178888Sjulian#include <sys/proc.h> 9083655Sbrooks#include <sys/queue.h> 91183351Sdwmalone#include <sys/sysctl.h> 9262587Sitojun#include <machine/cpu.h> 9362587Sitojun 9462587Sitojun#include <sys/malloc.h> 9562587Sitojun 9662587Sitojun#include <net/if.h> 97130933Sbrooks#include <net/if_clone.h> 9862587Sitojun#include <net/route.h> 9962587Sitojun#include <net/netisr.h> 10062587Sitojun#include <net/if_types.h> 10162587Sitojun#include <net/if_stf.h> 102196019Srwatson#include <net/vnet.h> 10362587Sitojun 10462587Sitojun#include <netinet/in.h> 10562587Sitojun#include <netinet/in_systm.h> 10662587Sitojun#include <netinet/ip.h> 10762587Sitojun#include <netinet/ip_var.h> 10862587Sitojun#include <netinet/in_var.h> 10962587Sitojun 11062587Sitojun#include <netinet/ip6.h> 11162587Sitojun#include <netinet6/ip6_var.h> 11262587Sitojun#include <netinet6/in6_var.h> 11362587Sitojun#include <netinet/ip_ecn.h> 11462587Sitojun 11562587Sitojun#include <netinet/ip_encap.h> 11662587Sitojun 11762587Sitojun#include <machine/stdarg.h> 11862587Sitojun 11962587Sitojun#include <net/bpf.h> 12062587Sitojun 121163606Srwatson#include <security/mac/mac_framework.h> 122163606Srwatson 123183351SdwmaloneSYSCTL_DECL(_net_link); 124248085Smariusstatic SYSCTL_NODE(_net_link, IFT_STF, stf, CTLFLAG_RW, 0, "6to4 Interface"); 125183351Sdwmalone 126183351Sdwmalonestatic int stf_route_cache = 1; 127183351SdwmaloneSYSCTL_INT(_net_link_stf, OID_AUTO, route_cache, CTLFLAG_RW, 128183351Sdwmalone &stf_route_cache, 0, "Caching of IPv4 routes for 6to4 Output"); 129183351Sdwmalone 130245258Saestatic int stf_permit_rfc1918 = 0; 131245258SaeTUNABLE_INT("net.link.stf.permit_rfc1918", &stf_permit_rfc1918); 132245258SaeSYSCTL_INT(_net_link_stf, OID_AUTO, permit_rfc1918, CTLFLAG_RW | CTLFLAG_TUN, 133245258Sae &stf_permit_rfc1918, 0, "Permit the use of private IPv4 addresses"); 134245258Sae 13583655Sbrooks#define STFNAME "stf" 136130933Sbrooks#define STFUNIT 0 13783655Sbrooks 13862587Sitojun#define IN6_IS_ADDR_6TO4(x) (ntohs((x)->s6_addr16[0]) == 0x2002) 13962587Sitojun 140109322Ssuz/* 141109322Ssuz * XXX: Return a pointer with 16-bit aligned. Don't cast it to 142109322Ssuz * struct in_addr *; use bcopy() instead. 143109322Ssuz */ 144109322Ssuz#define GET_V4(x) ((caddr_t)(&(x)->s6_addr16[1])) 145109322Ssuz 14662587Sitojunstruct stf_softc { 147147256Sbrooks struct ifnet *sc_ifp; 14862587Sitojun union { 14962587Sitojun struct route __sc_ro4; 15062587Sitojun struct route_in6 __sc_ro6; /* just for safety */ 15162587Sitojun } __sc_ro46; 15262587Sitojun#define sc_ro __sc_ro46.__sc_ro4 153183351Sdwmalone struct mtx sc_ro_mtx; 154178888Sjulian u_int sc_fibnum; 15562587Sitojun const struct encaptab *encap_cookie; 15662587Sitojun}; 157147256Sbrooks#define STF2IFP(sc) ((sc)->sc_ifp) 15862587Sitojun 159126783Srwatson/* 160183351Sdwmalone * Note that mutable fields in the softc are not currently locked. 161183351Sdwmalone * We do lock sc_ro in stf_output though. 162126783Srwatson */ 16383655Sbrooksstatic MALLOC_DEFINE(M_STF, STFNAME, "6to4 Tunnel Interface"); 164126709Srwatsonstatic const int ip_stf_ttl = 40; 16562587Sitojun 16679106Sbrooksextern struct domain inetdomain; 167152242Srustruct protosw in_stf_protosw = { 168152242Sru .pr_type = SOCK_RAW, 169152242Sru .pr_domain = &inetdomain, 170152242Sru .pr_protocol = IPPROTO_IPV6, 171152242Sru .pr_flags = PR_ATOMIC|PR_ADDR, 172152242Sru .pr_input = in_stf_input, 173152242Sru .pr_output = (pr_output_t *)rip_output, 174152242Sru .pr_ctloutput = rip_ctloutput, 175152242Sru .pr_usrreqs = &rip_usrreqs 17679106Sbrooks}; 17762587Sitojun 178130933Sbrooksstatic char *stfnames[] = {"stf0", "stf", "6to4", NULL}; 179130933Sbrooks 18092725Salfredstatic int stfmodevent(module_t, int, void *); 18192725Salfredstatic int stf_encapcheck(const struct mbuf *, int, int, void *); 18292725Salfredstatic struct in6_ifaddr *stf_getsrcifa6(struct ifnet *); 18392725Salfredstatic int stf_output(struct ifnet *, struct mbuf *, struct sockaddr *, 184191148Skmacy struct route *); 185103475Sumestatic int isrfc1918addr(struct in_addr *); 18692725Salfredstatic int stf_checkaddr4(struct stf_softc *, struct in_addr *, 18792725Salfred struct ifnet *); 18892725Salfredstatic int stf_checkaddr6(struct stf_softc *, struct in6_addr *, 18992725Salfred struct ifnet *); 19092725Salfredstatic void stf_rtrequest(int, struct rtentry *, struct rt_addrinfo *); 19192725Salfredstatic int stf_ioctl(struct ifnet *, u_long, caddr_t); 19262587Sitojun 193130933Sbrooksstatic int stf_clone_match(struct if_clone *, const char *); 194160195Ssamstatic int stf_clone_create(struct if_clone *, char *, size_t, caddr_t); 195130933Sbrooksstatic int stf_clone_destroy(struct if_clone *, struct ifnet *); 196130933Sbrooksstruct if_clone stf_cloner = IFC_CLONE_INITIALIZER(STFNAME, NULL, 0, 197130933Sbrooks NULL, stf_clone_match, stf_clone_create, stf_clone_destroy); 19883655Sbrooks 199130933Sbrooksstatic int 200130933Sbrooksstf_clone_match(struct if_clone *ifc, const char *name) 201130933Sbrooks{ 202130933Sbrooks int i; 20383655Sbrooks 204130933Sbrooks for(i = 0; stfnames[i] != NULL; i++) { 205130933Sbrooks if (strcmp(stfnames[i], name) == 0) 206130933Sbrooks return (1); 207130933Sbrooks } 208130933Sbrooks 209130933Sbrooks return (0); 210130933Sbrooks} 211130933Sbrooks 212128209Sbrooksstatic int 213160195Ssamstf_clone_create(struct if_clone *ifc, char *name, size_t len, caddr_t params) 21483655Sbrooks{ 215130933Sbrooks int err, unit; 21683655Sbrooks struct stf_softc *sc; 217128417Sbrooks struct ifnet *ifp; 21883655Sbrooks 219130933Sbrooks /* 220130933Sbrooks * We can only have one unit, but since unit allocation is 221130933Sbrooks * already locked, we use it to keep from allocating extra 222130933Sbrooks * interfaces. 223130933Sbrooks */ 224130933Sbrooks unit = STFUNIT; 225130933Sbrooks err = ifc_alloc_unit(ifc, &unit); 226130933Sbrooks if (err != 0) 227130933Sbrooks return (err); 228130933Sbrooks 229111119Simp sc = malloc(sizeof(struct stf_softc), M_STF, M_WAITOK | M_ZERO); 230147346Sbrooks ifp = STF2IFP(sc) = if_alloc(IFT_STF); 231147256Sbrooks if (ifp == NULL) { 232147256Sbrooks free(sc, M_STF); 233147256Sbrooks ifc_free_unit(ifc, unit); 234147256Sbrooks return (ENOSPC); 235147256Sbrooks } 236147346Sbrooks ifp->if_softc = sc; 237178888Sjulian sc->sc_fibnum = curthread->td_proc->p_fibnum; 238147346Sbrooks 239130933Sbrooks /* 240130933Sbrooks * Set the name manually rather then using if_initname because 241130933Sbrooks * we don't conform to the default naming convention for interfaces. 242130933Sbrooks */ 243130933Sbrooks strlcpy(ifp->if_xname, name, IFNAMSIZ); 244130933Sbrooks ifp->if_dname = ifc->ifc_name; 245130933Sbrooks ifp->if_dunit = IF_DUNIT_NONE; 24683655Sbrooks 247183351Sdwmalone mtx_init(&(sc)->sc_ro_mtx, "stf ro", NULL, MTX_DEF); 24883655Sbrooks sc->encap_cookie = encap_attach_func(AF_INET, IPPROTO_IPV6, 24983655Sbrooks stf_encapcheck, &in_stf_protosw, sc); 25083655Sbrooks if (sc->encap_cookie == NULL) { 251128417Sbrooks if_printf(ifp, "attach failed\n"); 25283655Sbrooks free(sc, M_STF); 253130933Sbrooks ifc_free_unit(ifc, unit); 25483655Sbrooks return (ENOMEM); 25583655Sbrooks } 25683655Sbrooks 257128417Sbrooks ifp->if_mtu = IPV6_MMTU; 258128417Sbrooks ifp->if_ioctl = stf_ioctl; 259128417Sbrooks ifp->if_output = stf_output; 260207554Ssobomax ifp->if_snd.ifq_maxlen = ifqmaxlen; 261128417Sbrooks if_attach(ifp); 262147611Sdwmalone bpfattach(ifp, DLT_NULL, sizeof(u_int32_t)); 26383655Sbrooks return (0); 26483655Sbrooks} 26583655Sbrooks 266130933Sbrooksstatic int 267130933Sbrooksstf_clone_destroy(struct if_clone *ifc, struct ifnet *ifp) 26883655Sbrooks{ 269147256Sbrooks struct stf_softc *sc = ifp->if_softc; 270151266Sthompsa int err; 27183655Sbrooks 272151266Sthompsa err = encap_detach(sc->encap_cookie); 273151266Sthompsa KASSERT(err == 0, ("Unexpected error detaching encap_cookie")); 274183351Sdwmalone mtx_destroy(&(sc)->sc_ro_mtx); 275151266Sthompsa bpfdetach(ifp); 276151266Sthompsa if_detach(ifp); 277151266Sthompsa if_free(ifp); 278151266Sthompsa 279151266Sthompsa free(sc, M_STF); 280130933Sbrooks ifc_free_unit(ifc, STFUNIT); 281130933Sbrooks 282130933Sbrooks return (0); 28383655Sbrooks} 28483655Sbrooks 28579106Sbrooksstatic int 28679106Sbrooksstfmodevent(mod, type, data) 28779106Sbrooks module_t mod; 28879106Sbrooks int type; 28979106Sbrooks void *data; 29062587Sitojun{ 29162587Sitojun 29279106Sbrooks switch (type) { 29379106Sbrooks case MOD_LOAD: 29483655Sbrooks if_clone_attach(&stf_cloner); 29579106Sbrooks break; 29679106Sbrooks case MOD_UNLOAD: 29783655Sbrooks if_clone_detach(&stf_cloner); 29879106Sbrooks break; 299132199Sphk default: 300132199Sphk return (EOPNOTSUPP); 30162587Sitojun } 30279106Sbrooks 30379106Sbrooks return (0); 30462587Sitojun} 30562587Sitojun 30679106Sbrooksstatic moduledata_t stf_mod = { 30779106Sbrooks "if_stf", 30879106Sbrooks stfmodevent, 30979106Sbrooks 0 31079106Sbrooks}; 31162587Sitojun 31279106SbrooksDECLARE_MODULE(if_stf, stf_mod, SI_SUB_PSEUDO, SI_ORDER_ANY); 31379106Sbrooks 31462587Sitojunstatic int 31562587Sitojunstf_encapcheck(m, off, proto, arg) 31662587Sitojun const struct mbuf *m; 31762587Sitojun int off; 31862587Sitojun int proto; 31962587Sitojun void *arg; 32062587Sitojun{ 32162587Sitojun struct ip ip; 32262587Sitojun struct in6_ifaddr *ia6; 32362587Sitojun struct stf_softc *sc; 324108710Sfenner struct in_addr a, b, mask; 32562587Sitojun 32662587Sitojun sc = (struct stf_softc *)arg; 32762587Sitojun if (sc == NULL) 32862587Sitojun return 0; 32962587Sitojun 330147256Sbrooks if ((STF2IFP(sc)->if_flags & IFF_UP) == 0) 33162587Sitojun return 0; 33262587Sitojun 33378064Sume /* IFF_LINK0 means "no decapsulation" */ 334147256Sbrooks if ((STF2IFP(sc)->if_flags & IFF_LINK0) != 0) 33578064Sume return 0; 33678064Sume 33762587Sitojun if (proto != IPPROTO_IPV6) 33862587Sitojun return 0; 33962587Sitojun 34062587Sitojun /* LINTED const cast */ 34191452Speter m_copydata((struct mbuf *)(uintptr_t)m, 0, sizeof(ip), (caddr_t)&ip); 34262587Sitojun 34362587Sitojun if (ip.ip_v != 4) 34462587Sitojun return 0; 34562587Sitojun 346147256Sbrooks ia6 = stf_getsrcifa6(STF2IFP(sc)); 34762587Sitojun if (ia6 == NULL) 34862587Sitojun return 0; 34962587Sitojun 35062587Sitojun /* 35162587Sitojun * check if IPv4 dst matches the IPv4 address derived from the 35262587Sitojun * local 6to4 address. 35362587Sitojun * success on: dst = 10.1.1.1, ia6->ia_addr = 2002:0a01:0101:... 35462587Sitojun */ 35562587Sitojun if (bcmp(GET_V4(&ia6->ia_addr.sin6_addr), &ip.ip_dst, 356194812Srwatson sizeof(ip.ip_dst)) != 0) { 357194812Srwatson ifa_free(&ia6->ia_ifa); 35862587Sitojun return 0; 359194812Srwatson } 36062587Sitojun 36162587Sitojun /* 36262587Sitojun * check if IPv4 src matches the IPv4 address derived from the 36362587Sitojun * local 6to4 address masked by prefixmask. 36462587Sitojun * success on: src = 10.1.1.1, ia6->ia_addr = 2002:0a00:.../24 36562587Sitojun * fail on: src = 10.1.1.1, ia6->ia_addr = 2002:0b00:.../24 36662587Sitojun */ 36762587Sitojun bzero(&a, sizeof(a)); 368108710Sfenner bcopy(GET_V4(&ia6->ia_addr.sin6_addr), &a, sizeof(a)); 369108710Sfenner bcopy(GET_V4(&ia6->ia_prefixmask.sin6_addr), &mask, sizeof(mask)); 370194812Srwatson ifa_free(&ia6->ia_ifa); 371108710Sfenner a.s_addr &= mask.s_addr; 37262587Sitojun b = ip.ip_src; 373108710Sfenner b.s_addr &= mask.s_addr; 37462587Sitojun if (a.s_addr != b.s_addr) 37562587Sitojun return 0; 37662587Sitojun 37762587Sitojun /* stf interface makes single side match only */ 37862587Sitojun return 32; 37962587Sitojun} 38062587Sitojun 38162587Sitojunstatic struct in6_ifaddr * 38262587Sitojunstf_getsrcifa6(ifp) 38362587Sitojun struct ifnet *ifp; 38462587Sitojun{ 38562587Sitojun struct ifaddr *ia; 38662587Sitojun struct in_ifaddr *ia4; 38762587Sitojun struct sockaddr_in6 *sin6; 38862587Sitojun struct in_addr in; 38962587Sitojun 390195022Srwatson if_addr_rlock(ifp); 391191339Srwatson TAILQ_FOREACH(ia, &ifp->if_addrhead, ifa_link) { 39262587Sitojun if (ia->ifa_addr->sa_family != AF_INET6) 39362587Sitojun continue; 39462587Sitojun sin6 = (struct sockaddr_in6 *)ia->ifa_addr; 39562587Sitojun if (!IN6_IS_ADDR_6TO4(&sin6->sin6_addr)) 39662587Sitojun continue; 39762587Sitojun 39862587Sitojun bcopy(GET_V4(&sin6->sin6_addr), &in, sizeof(in)); 39984104Sjlemon LIST_FOREACH(ia4, INADDR_HASH(in.s_addr), ia_hash) 40062587Sitojun if (ia4->ia_addr.sin_addr.s_addr == in.s_addr) 40162587Sitojun break; 40262587Sitojun if (ia4 == NULL) 40362587Sitojun continue; 40462587Sitojun 405194812Srwatson ifa_ref(ia); 406195022Srwatson if_addr_runlock(ifp); 40762587Sitojun return (struct in6_ifaddr *)ia; 40862587Sitojun } 409195022Srwatson if_addr_runlock(ifp); 41062587Sitojun 41162587Sitojun return NULL; 41262587Sitojun} 41362587Sitojun 41462587Sitojunstatic int 415191148Skmacystf_output(ifp, m, dst, ro) 41662587Sitojun struct ifnet *ifp; 41762587Sitojun struct mbuf *m; 41862587Sitojun struct sockaddr *dst; 419191148Skmacy struct route *ro; 42062587Sitojun{ 42162587Sitojun struct stf_softc *sc; 42262587Sitojun struct sockaddr_in6 *dst6; 423183351Sdwmalone struct route *cached_route; 424109322Ssuz struct in_addr in4; 425109322Ssuz caddr_t ptr; 42662587Sitojun struct sockaddr_in *dst4; 42762587Sitojun u_int8_t tos; 42862587Sitojun struct ip *ip; 42962587Sitojun struct ip6_hdr *ip6; 43062587Sitojun struct in6_ifaddr *ia6; 431147611Sdwmalone u_int32_t af; 432105580Srwatson int error; 43362587Sitojun 434183351Sdwmalone#ifdef MAC 435172930Srwatson error = mac_ifnet_check_transmit(ifp, m); 436105580Srwatson if (error) { 437105580Srwatson m_freem(m); 438105580Srwatson return (error); 439105580Srwatson } 440105580Srwatson#endif 441105580Srwatson 442147256Sbrooks sc = ifp->if_softc; 44362587Sitojun dst6 = (struct sockaddr_in6 *)dst; 44462587Sitojun 44562587Sitojun /* just in case */ 44662587Sitojun if ((ifp->if_flags & IFF_UP) == 0) { 44762587Sitojun m_freem(m); 448103487Sume ifp->if_oerrors++; 44962587Sitojun return ENETDOWN; 45062587Sitojun } 45162587Sitojun 45262587Sitojun /* 45362587Sitojun * If we don't have an ip4 address that match my inner ip6 address, 45462587Sitojun * we shouldn't generate output. Without this check, we'll end up 45562587Sitojun * using wrong IPv4 source. 45662587Sitojun */ 45762587Sitojun ia6 = stf_getsrcifa6(ifp); 45862587Sitojun if (ia6 == NULL) { 45962587Sitojun m_freem(m); 460103487Sume ifp->if_oerrors++; 46162587Sitojun return ENETDOWN; 46262587Sitojun } 46362587Sitojun 46462587Sitojun if (m->m_len < sizeof(*ip6)) { 46562587Sitojun m = m_pullup(m, sizeof(*ip6)); 466103487Sume if (!m) { 467194812Srwatson ifa_free(&ia6->ia_ifa); 468103487Sume ifp->if_oerrors++; 46962587Sitojun return ENOBUFS; 470103487Sume } 47162587Sitojun } 47262587Sitojun ip6 = mtod(m, struct ip6_hdr *); 47362587Sitojun tos = (ntohl(ip6->ip6_flow) >> 20) & 0xff; 47462587Sitojun 47578064Sume /* 476147611Sdwmalone * BPF writes need to be handled specially. 477147611Sdwmalone * This is a null operation, nothing here checks dst->sa_family. 478147611Sdwmalone */ 479147611Sdwmalone if (dst->sa_family == AF_UNSPEC) { 480147611Sdwmalone bcopy(dst->sa_data, &af, sizeof(af)); 481147611Sdwmalone dst->sa_family = af; 482147611Sdwmalone } 483147611Sdwmalone 484147611Sdwmalone /* 48578064Sume * Pickup the right outer dst addr from the list of candidates. 48678064Sume * ip6_dst has priority as it may be able to give us shorter IPv4 hops. 48778064Sume */ 488109322Ssuz ptr = NULL; 48978064Sume if (IN6_IS_ADDR_6TO4(&ip6->ip6_dst)) 490109322Ssuz ptr = GET_V4(&ip6->ip6_dst); 49178064Sume else if (IN6_IS_ADDR_6TO4(&dst6->sin6_addr)) 492109322Ssuz ptr = GET_V4(&dst6->sin6_addr); 49378064Sume else { 494194812Srwatson ifa_free(&ia6->ia_ifa); 49578064Sume m_freem(m); 496103487Sume ifp->if_oerrors++; 49778064Sume return ENETUNREACH; 49878064Sume } 499109322Ssuz bcopy(ptr, &in4, sizeof(in4)); 50078064Sume 501159180Scsjp if (bpf_peers_present(ifp->if_bpf)) { 50278701Sume /* 50378701Sume * We need to prepend the address family as 50478701Sume * a four byte field. Cons up a dummy header 50578701Sume * to pacify bpf. This is safe because bpf 50678701Sume * will only read from the mbuf (i.e., it won't 50778701Sume * try to free it or keep a pointer a to it). 50878701Sume */ 509147611Sdwmalone af = AF_INET6; 510123922Ssam bpf_mtap2(ifp->if_bpf, &af, sizeof(af), m); 51178701Sume } 51278701Sume 513111119Simp M_PREPEND(m, sizeof(struct ip), M_DONTWAIT); 51462587Sitojun if (m && m->m_len < sizeof(struct ip)) 51562587Sitojun m = m_pullup(m, sizeof(struct ip)); 516103487Sume if (m == NULL) { 517194812Srwatson ifa_free(&ia6->ia_ifa); 518103487Sume ifp->if_oerrors++; 51962587Sitojun return ENOBUFS; 520103487Sume } 52162587Sitojun ip = mtod(m, struct ip *); 52262587Sitojun 52362587Sitojun bzero(ip, sizeof(*ip)); 52462587Sitojun 52562587Sitojun bcopy(GET_V4(&((struct sockaddr_in6 *)&ia6->ia_addr)->sin6_addr), 52662587Sitojun &ip->ip_src, sizeof(ip->ip_src)); 527194812Srwatson ifa_free(&ia6->ia_ifa); 528109322Ssuz bcopy(&in4, &ip->ip_dst, sizeof(ip->ip_dst)); 52962587Sitojun ip->ip_p = IPPROTO_IPV6; 53079106Sbrooks ip->ip_ttl = ip_stf_ttl; 53162587Sitojun ip->ip_len = m->m_pkthdr.len; /*host order*/ 53262587Sitojun if (ifp->if_flags & IFF_LINK1) 53362587Sitojun ip_ecn_ingress(ECN_ALLOWED, &ip->ip_tos, &tos); 53478064Sume else 53578064Sume ip_ecn_ingress(ECN_NOCARE, &ip->ip_tos, &tos); 53662587Sitojun 537183351Sdwmalone if (!stf_route_cache) { 538183351Sdwmalone cached_route = NULL; 539183351Sdwmalone goto sendit; 540183351Sdwmalone } 541183351Sdwmalone 542126783Srwatson /* 543183351Sdwmalone * Do we have a cached route? 544126783Srwatson */ 545183351Sdwmalone mtx_lock(&(sc)->sc_ro_mtx); 54662587Sitojun dst4 = (struct sockaddr_in *)&sc->sc_ro.ro_dst; 54762587Sitojun if (dst4->sin_family != AF_INET || 54862587Sitojun bcmp(&dst4->sin_addr, &ip->ip_dst, sizeof(ip->ip_dst)) != 0) { 54962587Sitojun /* cache route doesn't match */ 55062587Sitojun dst4->sin_family = AF_INET; 55162587Sitojun dst4->sin_len = sizeof(struct sockaddr_in); 55262587Sitojun bcopy(&ip->ip_dst, &dst4->sin_addr, sizeof(dst4->sin_addr)); 55362587Sitojun if (sc->sc_ro.ro_rt) { 55462587Sitojun RTFREE(sc->sc_ro.ro_rt); 55562587Sitojun sc->sc_ro.ro_rt = NULL; 55662587Sitojun } 55762587Sitojun } 55862587Sitojun 55962587Sitojun if (sc->sc_ro.ro_rt == NULL) { 560178888Sjulian rtalloc_fib(&sc->sc_ro, sc->sc_fibnum); 56162587Sitojun if (sc->sc_ro.ro_rt == NULL) { 56262587Sitojun m_freem(m); 563183351Sdwmalone mtx_unlock(&(sc)->sc_ro_mtx); 564103487Sume ifp->if_oerrors++; 56562587Sitojun return ENETUNREACH; 56662587Sitojun } 56762587Sitojun } 568183351Sdwmalone cached_route = &sc->sc_ro; 56962587Sitojun 570183351Sdwmalonesendit: 571178888Sjulian M_SETFIB(m, sc->sc_fibnum); 572103487Sume ifp->if_opackets++; 573183351Sdwmalone error = ip_output(m, NULL, cached_route, 0, NULL, NULL); 574183351Sdwmalone 575183351Sdwmalone if (cached_route != NULL) 576183351Sdwmalone mtx_unlock(&(sc)->sc_ro_mtx); 577183351Sdwmalone return error; 57862587Sitojun} 57962587Sitojun 58062587Sitojunstatic int 581103475Sumeisrfc1918addr(in) 582109322Ssuz struct in_addr *in; 583103475Sume{ 584103475Sume /* 585103475Sume * returns 1 if private address range: 586103475Sume * 10.0.0.0/8 172.16.0.0/12 192.168.0.0/16 587103475Sume */ 588245258Sae if (stf_permit_rfc1918 == 0 && ( 589245258Sae (ntohl(in->s_addr) & 0xff000000) >> 24 == 10 || 590109322Ssuz (ntohl(in->s_addr) & 0xfff00000) >> 16 == 172 * 256 + 16 || 591245258Sae (ntohl(in->s_addr) & 0xffff0000) >> 16 == 192 * 256 + 168)) 592103475Sume return 1; 593103475Sume 594103475Sume return 0; 595103475Sume} 596103475Sume 597103475Sumestatic int 59878064Sumestf_checkaddr4(sc, in, inifp) 59978064Sume struct stf_softc *sc; 600109322Ssuz struct in_addr *in; 60178064Sume struct ifnet *inifp; /* incoming interface */ 60262587Sitojun{ 60362587Sitojun struct in_ifaddr *ia4; 60462587Sitojun 60562587Sitojun /* 60662587Sitojun * reject packets with the following address: 60762587Sitojun * 224.0.0.0/4 0.0.0.0/8 127.0.0.0/8 255.0.0.0/8 60862587Sitojun */ 609109322Ssuz if (IN_MULTICAST(ntohl(in->s_addr))) 61062587Sitojun return -1; 611109322Ssuz switch ((ntohl(in->s_addr) & 0xff000000) >> 24) { 612109322Ssuz case 0: case 127: case 255: 61362587Sitojun return -1; 614109322Ssuz } 61562587Sitojun 61662587Sitojun /* 617103475Sume * reject packets with private address range. 618103475Sume * (requirement from RFC3056 section 2 1st paragraph) 619103475Sume */ 620103475Sume if (isrfc1918addr(in)) 621103475Sume return -1; 622103475Sume 623103475Sume /* 62462587Sitojun * reject packets with broadcast 62562587Sitojun */ 626194951Srwatson IN_IFADDR_RLOCK(); 627181803Sbz for (ia4 = TAILQ_FIRST(&V_in_ifaddrhead); 62862587Sitojun ia4; 62962587Sitojun ia4 = TAILQ_NEXT(ia4, ia_link)) 63062587Sitojun { 63162587Sitojun if ((ia4->ia_ifa.ifa_ifp->if_flags & IFF_BROADCAST) == 0) 63262587Sitojun continue; 633194951Srwatson if (in->s_addr == ia4->ia_broadaddr.sin_addr.s_addr) { 634194951Srwatson IN_IFADDR_RUNLOCK(); 63562587Sitojun return -1; 636194951Srwatson } 63762587Sitojun } 638194951Srwatson IN_IFADDR_RUNLOCK(); 63962587Sitojun 64062587Sitojun /* 64162587Sitojun * perform ingress filter 64262587Sitojun */ 643147256Sbrooks if (sc && (STF2IFP(sc)->if_flags & IFF_LINK2) == 0 && inifp) { 64462587Sitojun struct sockaddr_in sin; 64562587Sitojun struct rtentry *rt; 64662587Sitojun 64762587Sitojun bzero(&sin, sizeof(sin)); 64862587Sitojun sin.sin_family = AF_INET; 64962587Sitojun sin.sin_len = sizeof(struct sockaddr_in); 650109322Ssuz sin.sin_addr = *in; 651178888Sjulian rt = rtalloc1_fib((struct sockaddr *)&sin, 0, 652178888Sjulian 0UL, sc->sc_fibnum); 65378064Sume if (!rt || rt->rt_ifp != inifp) { 65478064Sume#if 0 65578064Sume log(LOG_WARNING, "%s: packet from 0x%x dropped " 656147256Sbrooks "due to ingress filter\n", if_name(STF2IFP(sc)), 65778064Sume (u_int32_t)ntohl(sin.sin_addr.s_addr)); 65878064Sume#endif 65978064Sume if (rt) 660172307Scsjp RTFREE_LOCKED(rt); 66162587Sitojun return -1; 66262587Sitojun } 663172307Scsjp RTFREE_LOCKED(rt); 66462587Sitojun } 66562587Sitojun 66662587Sitojun return 0; 66762587Sitojun} 66862587Sitojun 66962587Sitojunstatic int 67078064Sumestf_checkaddr6(sc, in6, inifp) 67178064Sume struct stf_softc *sc; 67262587Sitojun struct in6_addr *in6; 67378064Sume struct ifnet *inifp; /* incoming interface */ 67462587Sitojun{ 67562587Sitojun /* 67662587Sitojun * check 6to4 addresses 67762587Sitojun */ 678109322Ssuz if (IN6_IS_ADDR_6TO4(in6)) { 679109322Ssuz struct in_addr in4; 680109322Ssuz bcopy(GET_V4(in6), &in4, sizeof(in4)); 681109322Ssuz return stf_checkaddr4(sc, &in4, inifp); 682109322Ssuz } 68362587Sitojun 68462587Sitojun /* 68562587Sitojun * reject anything that look suspicious. the test is implemented 68662587Sitojun * in ip6_input too, but we check here as well to 68762587Sitojun * (1) reject bad packets earlier, and 68862587Sitojun * (2) to be safe against future ip6_input change. 68962587Sitojun */ 69062587Sitojun if (IN6_IS_ADDR_V4COMPAT(in6) || IN6_IS_ADDR_V4MAPPED(in6)) 69162587Sitojun return -1; 69262587Sitojun 69362587Sitojun return 0; 69462587Sitojun} 69562587Sitojun 69662587Sitojunvoid 69783187Sjulianin_stf_input(m, off) 69878064Sume struct mbuf *m; 69983187Sjulian int off; 70062587Sitojun{ 70183187Sjulian int proto; 70262587Sitojun struct stf_softc *sc; 70362587Sitojun struct ip *ip; 70462587Sitojun struct ip6_hdr *ip6; 70562587Sitojun u_int8_t otos, itos; 70662587Sitojun struct ifnet *ifp; 70762587Sitojun 70882884Sjulian proto = mtod(m, struct ip *)->ip_p; 70962587Sitojun 71062587Sitojun if (proto != IPPROTO_IPV6) { 71162587Sitojun m_freem(m); 71262587Sitojun return; 71362587Sitojun } 71462587Sitojun 71562587Sitojun ip = mtod(m, struct ip *); 71662587Sitojun 71762587Sitojun sc = (struct stf_softc *)encap_getarg(m); 71862587Sitojun 719147256Sbrooks if (sc == NULL || (STF2IFP(sc)->if_flags & IFF_UP) == 0) { 72062587Sitojun m_freem(m); 72162587Sitojun return; 72262587Sitojun } 72362587Sitojun 724147256Sbrooks ifp = STF2IFP(sc); 72562587Sitojun 726105580Srwatson#ifdef MAC 727172930Srwatson mac_ifnet_create_mbuf(ifp, m); 728105580Srwatson#endif 729105580Srwatson 73062587Sitojun /* 73162587Sitojun * perform sanity check against outer src/dst. 73262587Sitojun * for source, perform ingress filter as well. 73362587Sitojun */ 73478064Sume if (stf_checkaddr4(sc, &ip->ip_dst, NULL) < 0 || 73578064Sume stf_checkaddr4(sc, &ip->ip_src, m->m_pkthdr.rcvif) < 0) { 73662587Sitojun m_freem(m); 73762587Sitojun return; 73862587Sitojun } 73962587Sitojun 74062587Sitojun otos = ip->ip_tos; 74162587Sitojun m_adj(m, off); 74262587Sitojun 74362587Sitojun if (m->m_len < sizeof(*ip6)) { 74462587Sitojun m = m_pullup(m, sizeof(*ip6)); 74562587Sitojun if (!m) 74662587Sitojun return; 74762587Sitojun } 74862587Sitojun ip6 = mtod(m, struct ip6_hdr *); 74962587Sitojun 75062587Sitojun /* 75162587Sitojun * perform sanity check against inner src/dst. 75262587Sitojun * for source, perform ingress filter as well. 75362587Sitojun */ 75478064Sume if (stf_checkaddr6(sc, &ip6->ip6_dst, NULL) < 0 || 75578064Sume stf_checkaddr6(sc, &ip6->ip6_src, m->m_pkthdr.rcvif) < 0) { 75662587Sitojun m_freem(m); 75762587Sitojun return; 75862587Sitojun } 75962587Sitojun 76062587Sitojun itos = (ntohl(ip6->ip6_flow) >> 20) & 0xff; 76162587Sitojun if ((ifp->if_flags & IFF_LINK1) != 0) 76262587Sitojun ip_ecn_egress(ECN_ALLOWED, &otos, &itos); 76378064Sume else 76478064Sume ip_ecn_egress(ECN_NOCARE, &otos, &itos); 76562587Sitojun ip6->ip6_flow &= ~htonl(0xff << 20); 76662587Sitojun ip6->ip6_flow |= htonl((u_int32_t)itos << 20); 76762587Sitojun 76862587Sitojun m->m_pkthdr.rcvif = ifp; 76962587Sitojun 770159180Scsjp if (bpf_peers_present(ifp->if_bpf)) { 77162587Sitojun /* 77262587Sitojun * We need to prepend the address family as 77362587Sitojun * a four byte field. Cons up a dummy header 77462587Sitojun * to pacify bpf. This is safe because bpf 77562587Sitojun * will only read from the mbuf (i.e., it won't 77662587Sitojun * try to free it or keep a pointer a to it). 77762587Sitojun */ 778123922Ssam u_int32_t af = AF_INET6; 779140042Sume bpf_mtap2(ifp->if_bpf, &af, sizeof(af), m); 78062587Sitojun } 78162587Sitojun 78262587Sitojun /* 78362587Sitojun * Put the packet to the network layer input queue according to the 78462587Sitojun * specified address family. 78562587Sitojun * See net/if_gif.c for possible issues with packet processing 78662587Sitojun * reorder due to extra queueing. 78762587Sitojun */ 78862587Sitojun ifp->if_ipackets++; 789111888Sjlemon ifp->if_ibytes += m->m_pkthdr.len; 790223741Sbz M_SETFIB(m, ifp->if_fib); 791111888Sjlemon netisr_dispatch(NETISR_IPV6, m); 79262587Sitojun} 79362587Sitojun 79462587Sitojun/* ARGSUSED */ 79562587Sitojunstatic void 79685074Srustf_rtrequest(cmd, rt, info) 79762587Sitojun int cmd; 79862587Sitojun struct rtentry *rt; 79985074Sru struct rt_addrinfo *info; 80062587Sitojun{ 801120727Ssam RT_LOCK_ASSERT(rt); 802248743Smelifaro rt->rt_rmx.rmx_mtu = rt->rt_ifp->if_mtu; 80362587Sitojun} 80462587Sitojun 80562587Sitojunstatic int 80662587Sitojunstf_ioctl(ifp, cmd, data) 80762587Sitojun struct ifnet *ifp; 80862587Sitojun u_long cmd; 80962587Sitojun caddr_t data; 81062587Sitojun{ 81162587Sitojun struct ifaddr *ifa; 81262587Sitojun struct ifreq *ifr; 81362587Sitojun struct sockaddr_in6 *sin6; 814109322Ssuz struct in_addr addr; 815248743Smelifaro int error, mtu; 81662587Sitojun 81762587Sitojun error = 0; 81862587Sitojun switch (cmd) { 81962587Sitojun case SIOCSIFADDR: 82062587Sitojun ifa = (struct ifaddr *)data; 82162587Sitojun if (ifa == NULL || ifa->ifa_addr->sa_family != AF_INET6) { 82262587Sitojun error = EAFNOSUPPORT; 82362587Sitojun break; 82462587Sitojun } 82562587Sitojun sin6 = (struct sockaddr_in6 *)ifa->ifa_addr; 826109322Ssuz if (!IN6_IS_ADDR_6TO4(&sin6->sin6_addr)) { 82762587Sitojun error = EINVAL; 828109322Ssuz break; 829109322Ssuz } 830109322Ssuz bcopy(GET_V4(&sin6->sin6_addr), &addr, sizeof(addr)); 831109322Ssuz if (isrfc1918addr(&addr)) { 832109322Ssuz error = EINVAL; 833109322Ssuz break; 834109322Ssuz } 835109322Ssuz 836109322Ssuz ifa->ifa_rtrequest = stf_rtrequest; 837109322Ssuz ifp->if_flags |= IFF_UP; 83862587Sitojun break; 83962587Sitojun 84062587Sitojun case SIOCADDMULTI: 84162587Sitojun case SIOCDELMULTI: 84262587Sitojun ifr = (struct ifreq *)data; 84362587Sitojun if (ifr && ifr->ifr_addr.sa_family == AF_INET6) 84462587Sitojun ; 84562587Sitojun else 84662587Sitojun error = EAFNOSUPPORT; 84762587Sitojun break; 84862587Sitojun 849248743Smelifaro case SIOCGIFMTU: 850248743Smelifaro break; 851248743Smelifaro 852248743Smelifaro case SIOCSIFMTU: 853248743Smelifaro ifr = (struct ifreq *)data; 854248743Smelifaro mtu = ifr->ifr_mtu; 855248743Smelifaro /* RFC 4213 3.2 ideal world MTU */ 856248743Smelifaro if (mtu < IPV6_MINMTU || mtu > IF_MAXMTU - 20) 857248743Smelifaro return (EINVAL); 858248743Smelifaro ifp->if_mtu = mtu; 859248743Smelifaro break; 860248743Smelifaro 86162587Sitojun default: 86262587Sitojun error = EINVAL; 86362587Sitojun break; 86462587Sitojun } 86562587Sitojun 86662587Sitojun return error; 86762587Sitojun} 868