bsd/net/if_stf.c

   1 /*
   2  * Copyright (c) 2000-2018 Apple Inc. All rights reserved.
   3  *
   4  * @APPLE_OSREFERENCE_LICENSE_HEADER_START@
   5  *
   6  * This file contains Original Code and/or Modifications of Original Code
   7  * as defined in and that are subject to the Apple Public Source License
   8  * Version 2.0 (the 'License'). You may not use this file except in
   9  * compliance with the License. The rights granted to you under the License
  10  * may not be used to create, or enable the creation or redistribution of,
  11  * unlawful or unlicensed copies of an Apple operating system, or to
  12  * circumvent, violate, or enable the circumvention or violation of, any
  13  * terms of an Apple operating system software license agreement.
  14  *
  15  * Please obtain a copy of the License at
  16  * http://www.opensource.apple.com/apsl/ and read it before using this file.
  17  *
  18  * The Original Code and all software distributed under the License are
  19  * distributed on an 'AS IS' basis, WITHOUT WARRANTY OF ANY KIND, EITHER
  20  * EXPRESS OR IMPLIED, AND APPLE HEREBY DISCLAIMS ALL SUCH WARRANTIES,
  21  * INCLUDING WITHOUT LIMITATION, ANY WARRANTIES OF MERCHANTABILITY,
  22  * FITNESS FOR A PARTICULAR PURPOSE, QUIET ENJOYMENT OR NON-INFRINGEMENT.
  23  * Please see the License for the specific language governing rights and
  24  * limitations under the License.
  25  *
  26  * @APPLE_OSREFERENCE_LICENSE_HEADER_END@
  27  */
  28
  29 /*      $FreeBSD: src/sys/net/if_stf.c,v 1.1.2.6 2001/07/24 19:10:18 brooks Exp $       */
  30 /*      $KAME: if_stf.c,v 1.62 2001/06/07 22:32:16 itojun Exp $ */
  31
  32 /*
  33  * Copyright (C) 2000 WIDE Project.
  34  * All rights reserved.
  35  *
  36  * Redistribution and use in source and binary forms, with or without
  37  * modification, are permitted provided that the following conditions
  38  * are met:
  39  * 1. Redistributions of source code must retain the above copyright
  40  *    notice, this list of conditions and the following disclaimer.
  41  * 2. Redistributions in binary form must reproduce the above copyright
  42  *    notice, this list of conditions and the following disclaimer in the
  43  *    documentation and/or other materials provided with the distribution.
  44  * 3. Neither the name of the project nor the names of its contributors
  45  *    may be used to endorse or promote products derived from this software
  46  *    without specific prior written permission.
  47  *
  48  * THIS SOFTWARE IS PROVIDED BY THE PROJECT AND CONTRIBUTORS ``AS IS'' AND
  49  * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
  50  * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
  51  * ARE DISCLAIMED.  IN NO EVENT SHALL THE PROJECT OR CONTRIBUTORS BE LIABLE
  52  * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
  53  * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
  54  * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
  55  * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
  56  * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
  57  * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
  58  * SUCH DAMAGE.
  59  */
  60 /*
  61  * NOTICE: This file was modified by SPARTA, Inc. in 2006 to introduce
  62  * support for mandatory and extensible security protections.  This notice
  63  * is included in support of clause 2.2 (b) of the Apple Public License,
  64  * Version 2.0.
  65  */
  66
  67 /*
  68  * 6to4 interface, based on RFC3056.
  69  *
  70  * 6to4 interface is NOT capable of link-layer (I mean, IPv4) multicasting.
  71  * There is no address mapping defined from IPv6 multicast address to IPv4
  72  * address.  Therefore, we do not have IFF_MULTICAST on the interface.
  73  *
  74  * Due to the lack of address mapping for link-local addresses, we cannot
  75  * throw packets toward link-local addresses (fe80::x).  Also, we cannot throw
  76  * packets to link-local multicast addresses (ff02::x).
  77  *
  78  * Here are interesting symptoms due to the lack of link-local address:
  79  *
  80  * Unicast routing exchange:
  81  * - RIPng: Impossible.  Uses link-local multicast packet toward ff02::9,
  82  *   and link-local addresses as nexthop.
  83  * - OSPFv6: Impossible.  OSPFv6 assumes that there's link-local address
  84  *   assigned to the link, and makes use of them.  Also, HELLO packets use
  85  *   link-local multicast addresses (ff02::5 and ff02::6).
  86  * - BGP4+: Maybe.  You can only use global address as nexthop, and global
  87  *   address as TCP endpoint address.
  88  *
  89  * Multicast routing protocols:
  90  * - PIM: Hello packet cannot be used to discover adjacent PIM routers.
  91  *   Adjacent PIM routers must be configured manually (is it really spec-wise
  92  *   correct thing to do?).
  93  *
  94  * ICMPv6:
  95  * - Redirects cannot be used due to the lack of link-local address.
  96  *
  97  * stf interface does not have, and will not need, a link-local address.
  98  * It seems to have no real benefit and does not help the above symptoms much.
  99  * Even if we assign link-locals to interface, we cannot really
 100  * use link-local unicast/multicast on top of 6to4 cloud (since there's no
 101  * encapsulation defined for link-local address), and the above analysis does
 102  * not change.  RFC3056 does not mandate the assignment of link-local address
 103  * either.
 104  *
 105  * 6to4 interface has security issues.  Refer to
 106  * http://playground.iijlab.net/i-d/draft-itojun-ipv6-transition-abuse-00.txt
 107  * for details.  The code tries to filter out some of malicious packets.
 108  * Note that there is no way to be 100% secure.
 109  */
 110
 111 #include <sys/param.h>
 112 #include <sys/systm.h>
 113 #include <sys/socket.h>
 114 #include <sys/sockio.h>
 115 #include <sys/mbuf.h>
 116 #include <sys/errno.h>
 117 #include <sys/protosw.h>
 118 #include <sys/kernel.h>
 119 #include <sys/syslog.h>
 120
 121 #include <sys/malloc.h>
 122
 123 #include <kern/locks.h>
 124
 125 #include <net/if.h>
 126 #include <net/route.h>
 127 #include <net/if_types.h>
 128
 129 #include <netinet/in.h>
 130 #include <netinet/in_systm.h>
 131 #include <netinet/ip.h>
 132 #include <netinet/ip_var.h>
 133 #include <netinet/in_var.h>
 134
 135 #include <netinet/ip6.h>
 136 #include <netinet6/ip6_var.h>
 137 #include <netinet6/in6_var.h>
 138 #include <netinet/ip_ecn.h>
 139
 140 #include <netinet/ip_encap.h>
 141 #include <net/kpi_interface.h>
 142 #include <net/kpi_protocol.h>
 143
 144
 145 #include <net/net_osdep.h>
 146
 147 #include <net/bpf.h>
 148
 149 #if CONFIG_MACF_NET
 150 #include <security/mac_framework.h>
 151 #endif
 152
 153 #define GET_V4(x) ((const struct in_addr *)(const void *)(&(x)->s6_addr16[1]))
 154
 155 static lck_grp_t *stf_mtx_grp;
 156
 157 struct stf_softc {
 158         ifnet_t                         sc_if;     /* common area */
 159         u_int32_t                               sc_protocol_family; /* dlil protocol attached */
 160         union {
 161                 struct route  __sc_ro4;
 162                 struct route_in6 __sc_ro6; /* just for safety */
 163         } __sc_ro46;
 164 #define sc_ro   __sc_ro46.__sc_ro4
 165         decl_lck_mtx_data(, sc_ro_mtx);
 166         const struct encaptab *encap_cookie;
 167         bpf_tap_mode            tap_mode;
 168         bpf_packet_func         tap_callback;
 169 };
 170
 171 void stfattach(void);
 172
 173 static int ip_stf_ttl = 40;
 174 static int stf_init_done;
 175
 176 static void in_stf_input(struct mbuf *, int);
 177 static void stfinit(void);
 178
 179 static struct protosw in_stf_protosw =
 180 {
 181         .pr_type =              SOCK_RAW,
 182         .pr_protocol =          IPPROTO_IPV6,
 183         .pr_flags =             PR_ATOMIC | PR_ADDR,
 184         .pr_input =             in_stf_input,
 185         .pr_ctloutput =         rip_ctloutput,
 186         .pr_usrreqs =           &rip_usrreqs,
 187         .pr_unlock =            rip_unlock,
 188 };
 189
 190 static int stf_encapcheck(const struct mbuf *, int, int, void *);
 191 static struct in6_ifaddr *stf_getsrcifa6(struct ifnet *);
 192 int stf_pre_output(struct ifnet *, protocol_family_t, struct mbuf **,
 193     const struct sockaddr *, void *, char *, char *);
 194 static int stf_checkaddr4(struct stf_softc *, const struct in_addr *,
 195     struct ifnet *);
 196 static int stf_checkaddr6(struct stf_softc *, struct in6_addr *,
 197     struct ifnet *);
 198 static void stf_rtrequest(int, struct rtentry *, struct sockaddr *);
 199 static errno_t stf_ioctl(ifnet_t ifp, u_long cmd, void *data);
 200 static errno_t stf_output(ifnet_t ifp, mbuf_t m);
 201
 202 static void
 203 stfinit(void)
 204 {
 205         if (!stf_init_done) {
 206                 stf_mtx_grp = lck_grp_alloc_init("stf", LCK_GRP_ATTR_NULL);
 207                 stf_init_done = 1;
 208         }
 209 }
 210
 211 /*
 212  * gif_input is the input handler for IP and IPv6 attached to gif
 213  */
 214 static errno_t
 215 stf_media_input(
 216         __unused ifnet_t        ifp,
 217         protocol_family_t       protocol_family,
 218         mbuf_t                          m,
 219         __unused char           *frame_header)
 220 {
 221         if (proto_input(protocol_family, m) != 0) {
 222                 m_freem(m);
 223         }
 224
 225         return 0;
 226 }
 227
 228
 229
 230 static errno_t
 231 stf_add_proto(
 232         ifnet_t                                                                 ifp,
 233         protocol_family_t                                               protocol_family,
 234         __unused const struct ifnet_demux_desc  *demux_array,
 235         __unused u_int32_t                                              demux_count)
 236 {
 237         /* Only one protocol may be attached at a time */
 238         struct stf_softc* stf = ifnet_softc(ifp);
 239         if (stf->sc_protocol_family == 0) {
 240                 stf->sc_protocol_family = protocol_family;
 241         } else {
 242                 printf("stf_add_proto: stf already has a proto\n");
 243                 return EBUSY;
 244         }
 245
 246         return 0;
 247 }
 248
 249 static errno_t
 250 stf_del_proto(
 251         ifnet_t                         ifp,
 252         protocol_family_t       protocol_family)
 253 {
 254         if (((struct stf_softc*)ifnet_softc(ifp))->sc_protocol_family == protocol_family) {
 255                 ((struct stf_softc*)ifnet_softc(ifp))->sc_protocol_family = 0;
 256         }
 257
 258         return 0;
 259 }
 260
 261 static errno_t
 262 stf_attach_inet6(
 263         ifnet_t                         ifp,
 264         protocol_family_t       protocol_family)
 265 {
 266         struct ifnet_attach_proto_param     reg;
 267         errno_t                                                     stat;
 268
 269         if (protocol_family != PF_INET6) {
 270                 return EPROTONOSUPPORT;
 271         }
 272
 273         bzero(&reg, sizeof(reg));
 274         reg.input = stf_media_input;
 275         reg.pre_output = stf_pre_output;
 276
 277         stat = ifnet_attach_protocol(ifp, protocol_family, &reg);
 278         if (stat && stat != EEXIST) {
 279                 printf("stf_attach_proto_family can't attach interface fam=%d\n",
 280                     protocol_family);
 281         }
 282
 283         return stat;
 284 }
 285
 286 static errno_t
 287 stf_demux(
 288         ifnet_t                                 ifp,
 289         __unused mbuf_t                 m,
 290         __unused char                   *frame_ptr,
 291         protocol_family_t               *protocol_family)
 292 {
 293         struct stf_softc* stf = ifnet_softc(ifp);
 294         *protocol_family = stf->sc_protocol_family;
 295         return 0;
 296 }
 297
 298 static errno_t
 299 stf_set_bpf_tap(
 300         ifnet_t                 ifp,
 301         bpf_tap_mode    mode,
 302         bpf_packet_func callback)
 303 {
 304         struct stf_softc        *sc = ifnet_softc(ifp);
 305
 306         sc->tap_mode = mode;
 307         sc->tap_callback = callback;
 308
 309         return 0;
 310 }
 311
 312 void
 313 stfattach(void)
 314 {
 315         struct stf_softc *sc;
 316         int error;
 317         const struct encaptab *p;
 318         struct ifnet_init_eparams       stf_init;
 319
 320         stfinit();
 321
 322         error = proto_register_plumber(PF_INET6, APPLE_IF_FAM_STF,
 323             stf_attach_inet6, NULL);
 324         if (error != 0) {
 325                 printf("proto_register_plumber failed for AF_INET6 error=%d\n", error);
 326         }
 327
 328         sc = _MALLOC(sizeof(struct stf_softc), M_DEVBUF, M_WAITOK | M_ZERO);
 329         if (sc == 0) {
 330                 printf("stf softc attach failed\n" );
 331                 return;
 332         }
 333
 334         p = encap_attach_func(AF_INET, IPPROTO_IPV6, stf_encapcheck,
 335             &in_stf_protosw, sc);
 336         if (p == NULL) {
 337                 printf("sftattach encap_attach_func failed\n");
 338                 FREE(sc, M_DEVBUF);
 339                 return;
 340         }
 341         sc->encap_cookie = p;
 342         lck_mtx_init(&sc->sc_ro_mtx, stf_mtx_grp, LCK_ATTR_NULL);
 343
 344         bzero(&stf_init, sizeof(stf_init));
 345         stf_init.ver = IFNET_INIT_CURRENT_VERSION;
 346         stf_init.len = sizeof(stf_init);
 347         stf_init.flags = IFNET_INIT_LEGACY;
 348         stf_init.name = "stf";
 349         stf_init.unit = 0;
 350         stf_init.type = IFT_STF;
 351         stf_init.family = IFNET_FAMILY_STF;
 352         stf_init.output = stf_output;
 353         stf_init.demux = stf_demux;
 354         stf_init.add_proto = stf_add_proto;
 355         stf_init.del_proto = stf_del_proto;
 356         stf_init.softc = sc;
 357         stf_init.ioctl = stf_ioctl;
 358         stf_init.set_bpf_tap = stf_set_bpf_tap;
 359
 360         error = ifnet_allocate_extended(&stf_init, &sc->sc_if);
 361         if (error != 0) {
 362                 printf("stfattach, ifnet_allocate failed - %d\n", error);
 363                 encap_detach(sc->encap_cookie);
 364                 lck_mtx_destroy(&sc->sc_ro_mtx, stf_mtx_grp);
 365                 FREE(sc, M_DEVBUF);
 366                 return;
 367         }
 368         ifnet_set_mtu(sc->sc_if, IPV6_MMTU);
 369         ifnet_set_flags(sc->sc_if, 0, 0xffff); /* clear all flags */
 370 #if 0
 371         /* turn off ingress filter */
 372         ifnet_set_flags(sc->sc_if, IFF_LINK2, IFF_LINK2);
 373 #endif
 374
 375 #if CONFIG_MACF_NET
 376         mac_ifnet_label_init(&sc->sc_if);
 377 #endif
 378
 379         error = ifnet_attach(sc->sc_if, NULL);
 380         if (error != 0) {
 381                 printf("stfattach: ifnet_attach returned error=%d\n", error);
 382                 encap_detach(sc->encap_cookie);
 383                 ifnet_release(sc->sc_if);
 384                 lck_mtx_destroy(&sc->sc_ro_mtx, stf_mtx_grp);
 385                 FREE(sc, M_DEVBUF);
 386                 return;
 387         }
 388
 389         bpfattach(sc->sc_if, DLT_NULL, sizeof(u_int));
 390
 391         return;
 392 }
 393
 394 static int
 395 stf_encapcheck(
 396         const struct mbuf *m,
 397         __unused int off,
 398         int proto,
 399         void *arg)
 400 {
 401         struct ip ip;
 402         struct in6_ifaddr *ia6;
 403         struct stf_softc *sc;
 404         struct in_addr a, b;
 405
 406         sc = (struct stf_softc *)arg;
 407         if (sc == NULL) {
 408                 return 0;
 409         }
 410
 411         if ((ifnet_flags(sc->sc_if) & IFF_UP) == 0) {
 412                 return 0;
 413         }
 414
 415         /* IFF_LINK0 means "no decapsulation" */
 416         if ((ifnet_flags(sc->sc_if) & IFF_LINK0) != 0) {
 417                 return 0;
 418         }
 419
 420         if (proto != IPPROTO_IPV6) {
 421                 return 0;
 422         }
 423
 424         mbuf_copydata((struct mbuf *)(size_t)m, 0, sizeof(ip), &ip);
 425
 426         if (ip.ip_v != 4) {
 427                 return 0;
 428         }
 429
 430         ia6 = stf_getsrcifa6(sc->sc_if);
 431         if (ia6 == NULL) {
 432                 return 0;
 433         }
 434
 435         /*
 436          * check if IPv4 dst matches the IPv4 address derived from the
 437          * local 6to4 address.
 438          * success on: dst = 10.1.1.1, ia6->ia_addr = 2002:0a01:0101:...
 439          */
 440         IFA_LOCK(&ia6->ia_ifa);
 441         if (bcmp(GET_V4(&ia6->ia_addr.sin6_addr), &ip.ip_dst,
 442             sizeof(ip.ip_dst)) != 0) {
 443                 IFA_UNLOCK(&ia6->ia_ifa);
 444                 IFA_REMREF(&ia6->ia_ifa);
 445                 return 0;
 446         }
 447         /*
 448          * check if IPv4 src matches the IPv4 address derived from the
 449          * local 6to4 address masked by prefixmask.
 450          * success on: src = 10.1.1.1, ia6->ia_addr = 2002:0a00:.../24
 451          * fail on: src = 10.1.1.1, ia6->ia_addr = 2002:0b00:.../24
 452          */
 453         bzero(&a, sizeof(a));
 454         a.s_addr = GET_V4(&ia6->ia_addr.sin6_addr)->s_addr;
 455         a.s_addr &= GET_V4(&ia6->ia_prefixmask.sin6_addr)->s_addr;
 456         b = ip.ip_src;
 457         b.s_addr &= GET_V4(&ia6->ia_prefixmask.sin6_addr)->s_addr;
 458         if (a.s_addr != b.s_addr) {
 459                 IFA_UNLOCK(&ia6->ia_ifa);
 460                 IFA_REMREF(&ia6->ia_ifa);
 461                 return 0;
 462         }
 463         /* stf interface makes single side match only */
 464         IFA_UNLOCK(&ia6->ia_ifa);
 465         IFA_REMREF(&ia6->ia_ifa);
 466         return 32;
 467 }
 468
 469 static struct in6_ifaddr *
 470 stf_getsrcifa6(struct ifnet *ifp)
 471 {
 472         struct ifaddr *ia;
 473         struct in_ifaddr *ia4;
 474         struct sockaddr_in6 *sin6;
 475         struct in_addr in;
 476
 477         ifnet_lock_shared(ifp);
 478         for (ia = ifp->if_addrlist.tqh_first; ia; ia = ia->ifa_list.tqe_next) {
 479                 IFA_LOCK(ia);
 480                 if (ia->ifa_addr == NULL) {
 481                         IFA_UNLOCK(ia);
 482                         continue;
 483                 }
 484                 if (ia->ifa_addr->sa_family != AF_INET6) {
 485                         IFA_UNLOCK(ia);
 486                         continue;
 487                 }
 488                 sin6 = (struct sockaddr_in6 *)(void *)ia->ifa_addr;
 489                 if (!IN6_IS_ADDR_6TO4(&sin6->sin6_addr)) {
 490                         IFA_UNLOCK(ia);
 491                         continue;
 492                 }
 493                 bcopy(GET_V4(&sin6->sin6_addr), &in, sizeof(in));
 494                 IFA_UNLOCK(ia);
 495                 lck_rw_lock_shared(in_ifaddr_rwlock);
 496                 for (ia4 = TAILQ_FIRST(&in_ifaddrhead);
 497                     ia4;
 498                     ia4 = TAILQ_NEXT(ia4, ia_link)) {
 499                         IFA_LOCK(&ia4->ia_ifa);
 500                         if (ia4->ia_addr.sin_addr.s_addr == in.s_addr) {
 501                                 IFA_UNLOCK(&ia4->ia_ifa);
 502                                 break;
 503                         }
 504                         IFA_UNLOCK(&ia4->ia_ifa);
 505                 }
 506                 lck_rw_done(in_ifaddr_rwlock);
 507                 if (ia4 == NULL) {
 508                         continue;
 509                 }
 510
 511                 IFA_ADDREF(ia);         /* for caller */
 512                 ifnet_lock_done(ifp);
 513                 return (struct in6_ifaddr *)ia;
 514         }
 515         ifnet_lock_done(ifp);
 516
 517         return NULL;
 518 }
 519
 520 int
 521 stf_pre_output(
 522         struct ifnet    *ifp,
 523         __unused protocol_family_t  protocol_family,
 524         struct mbuf     **m0,
 525         const struct sockaddr   *dst,
 526         __unused void *route,
 527         __unused char *desk_linkaddr,
 528         __unused char *frame_type)
 529 {
 530         struct mbuf *m = *m0;
 531         struct stf_softc *sc;
 532         const struct sockaddr_in6 *dst6;
 533         const struct in_addr *in4;
 534         u_int8_t tos;
 535         struct ip *ip;
 536         struct ip6_hdr *ip6;
 537         struct in6_ifaddr *ia6;
 538         struct sockaddr_in      *dst4;
 539         struct ip_out_args ipoa;
 540         errno_t result = 0;
 541
 542         bzero(&ipoa, sizeof(ipoa));
 543         ipoa.ipoa_boundif = IFSCOPE_NONE;
 544         ipoa.ipoa_flags = IPOAF_SELECT_SRCIF;
 545         ipoa.ipoa_sotc = SO_TC_UNSPEC;
 546         ipoa.ipoa_netsvctype = _NET_SERVICE_TYPE_UNSPEC;
 547
 548         sc = ifnet_softc(ifp);
 549         dst6 = (const struct sockaddr_in6 *)(const void *)dst;
 550
 551         /* just in case */
 552         if ((ifnet_flags(ifp) & IFF_UP) == 0) {
 553                 printf("stf: IFF_DOWN\n");
 554                 return ENETDOWN;
 555         }
 556
 557         /*
 558          * If we don't have an ip4 address that match my inner ip6 address,
 559          * we shouldn't generate output.  Without this check, we'll end up
 560          * using wrong IPv4 source.
 561          */
 562         ia6 = stf_getsrcifa6(ifp);
 563         if (ia6 == NULL) {
 564                 return ENETDOWN;
 565         }
 566
 567         if (mbuf_len(m) < sizeof(*ip6)) {
 568                 m = m_pullup(m, sizeof(*ip6));
 569                 if (!m) {
 570                         *m0 = NULL; /* makes sure this won't be double freed */
 571                         IFA_REMREF(&ia6->ia_ifa);
 572                         return ENOBUFS;
 573                 }
 574         }
 575         ip6 = mtod(m, struct ip6_hdr *);
 576         tos = (ntohl(ip6->ip6_flow) >> 20) & 0xff;
 577
 578         /*
 579          * Pickup the right outer dst addr from the list of candidates.
 580          * ip6_dst has priority as it may be able to give us shorter IPv4 hops.
 581          */
 582         if (IN6_IS_ADDR_6TO4(&ip6->ip6_dst)) {
 583                 in4 = GET_V4(&ip6->ip6_dst);
 584         } else if (IN6_IS_ADDR_6TO4(&dst6->sin6_addr)) {
 585                 in4 = GET_V4(&dst6->sin6_addr);
 586         } else {
 587                 IFA_REMREF(&ia6->ia_ifa);
 588                 return ENETUNREACH;
 589         }
 590
 591         if (ifp->if_bpf) {
 592                 /* We need to prepend the address family as a four byte field. */
 593                 u_int32_t af = AF_INET6;
 594
 595                 bpf_tap_out(ifp, 0, m, &af, sizeof(af));
 596         }
 597
 598         M_PREPEND(m, sizeof(struct ip), M_DONTWAIT, 1);
 599         if (m && mbuf_len(m) < sizeof(struct ip)) {
 600                 m = m_pullup(m, sizeof(struct ip));
 601         }
 602         if (m == NULL) {
 603                 *m0 = NULL;
 604                 IFA_REMREF(&ia6->ia_ifa);
 605                 return ENOBUFS;
 606         }
 607         ip = mtod(m, struct ip *);
 608
 609         bzero(ip, sizeof(*ip));
 610
 611         IFA_LOCK_SPIN(&ia6->ia_ifa);
 612         bcopy(GET_V4(&((struct sockaddr_in6 *)&ia6->ia_addr)->sin6_addr),
 613             &ip->ip_src, sizeof(ip->ip_src));
 614         IFA_UNLOCK(&ia6->ia_ifa);
 615         bcopy(in4, &ip->ip_dst, sizeof(ip->ip_dst));
 616         ip->ip_p = IPPROTO_IPV6;
 617         ip->ip_ttl = ip_stf_ttl;
 618         ip->ip_len = m->m_pkthdr.len;   /*host order*/
 619         if (ifp->if_flags & IFF_LINK1) {
 620                 ip_ecn_ingress(ECN_NORMAL, &ip->ip_tos, &tos);
 621         } else {
 622                 ip_ecn_ingress(ECN_NOCARE, &ip->ip_tos, &tos);
 623         }
 624
 625         lck_mtx_lock(&sc->sc_ro_mtx);
 626         dst4 = (struct sockaddr_in *)(void *)&sc->sc_ro.ro_dst;
 627         if (ROUTE_UNUSABLE(&sc->sc_ro) || dst4->sin_family != AF_INET ||
 628             bcmp(&dst4->sin_addr, &ip->ip_dst, sizeof(ip->ip_dst)) != 0) {
 629                 ROUTE_RELEASE(&sc->sc_ro);
 630                 /* cache route doesn't match: always the case during the first use */
 631                 dst4->sin_family = AF_INET;
 632                 dst4->sin_len = sizeof(struct sockaddr_in);
 633                 bcopy(&ip->ip_dst, &dst4->sin_addr, sizeof(dst4->sin_addr));
 634         }
 635
 636         result = ip_output(m, NULL, &sc->sc_ro, IP_OUTARGS, NULL, &ipoa);
 637         lck_mtx_unlock(&sc->sc_ro_mtx);
 638
 639         /* Assumption: ip_output will free mbuf on errors */
 640         /* All the output processing is done here, don't let stf_output be called */
 641         if (result == 0) {
 642                 result = EJUSTRETURN;
 643         }
 644         *m0 = NULL;
 645         IFA_REMREF(&ia6->ia_ifa);
 646         return result;
 647 }
 648 static errno_t
 649 stf_output(
 650         __unused ifnet_t        ifp,
 651         __unused mbuf_t m)
 652 {
 653         /* All processing is done in stf_pre_output
 654          * this shouldn't be called as the pre_output returns "EJUSTRETURN"
 655          */
 656         return 0;
 657 }
 658
 659 static int
 660 stf_checkaddr4(
 661         struct stf_softc *sc,
 662         const struct in_addr *in,
 663         struct ifnet *inifp)    /* incoming interface */
 664 {
 665         struct in_ifaddr *ia4;
 666
 667         /*
 668          * reject packets with the following address:
 669          * 224.0.0.0/4 0.0.0.0/8 127.0.0.0/8 255.0.0.0/8
 670          */
 671         if (IN_MULTICAST(ntohl(in->s_addr))) {
 672                 return -1;
 673         }
 674         switch ((ntohl(in->s_addr) & 0xff000000) >> 24) {
 675         case 0: case 127: case 255:
 676                 return -1;
 677         }
 678
 679         /*
 680          * reject packets with broadcast
 681          */
 682         lck_rw_lock_shared(in_ifaddr_rwlock);
 683         for (ia4 = TAILQ_FIRST(&in_ifaddrhead);
 684             ia4;
 685             ia4 = TAILQ_NEXT(ia4, ia_link)) {
 686                 IFA_LOCK(&ia4->ia_ifa);
 687                 if ((ia4->ia_ifa.ifa_ifp->if_flags & IFF_BROADCAST) == 0) {
 688                         IFA_UNLOCK(&ia4->ia_ifa);
 689                         continue;
 690                 }
 691                 if (in->s_addr == ia4->ia_broadaddr.sin_addr.s_addr) {
 692                         IFA_UNLOCK(&ia4->ia_ifa);
 693                         lck_rw_done(in_ifaddr_rwlock);
 694                         return -1;
 695                 }
 696                 IFA_UNLOCK(&ia4->ia_ifa);
 697         }
 698         lck_rw_done(in_ifaddr_rwlock);
 699
 700         /*
 701          * perform ingress filter
 702          */
 703         if (sc && (ifnet_flags(sc->sc_if) & IFF_LINK2) == 0 && inifp) {
 704                 struct sockaddr_in sin;
 705                 struct rtentry *rt;
 706
 707                 bzero(&sin, sizeof(sin));
 708                 sin.sin_family = AF_INET;
 709                 sin.sin_len = sizeof(struct sockaddr_in);
 710                 sin.sin_addr = *in;
 711                 rt = rtalloc1((struct sockaddr *)&sin, 0, 0);
 712                 if (rt != NULL) {
 713                         RT_LOCK(rt);
 714                 }
 715                 if (rt == NULL || rt->rt_ifp != inifp) {
 716 #if 1
 717                         log(LOG_WARNING, "%s: packet from 0x%x dropped "
 718                             "due to ingress filter\n", if_name(sc->sc_if),
 719                             (u_int32_t)ntohl(sin.sin_addr.s_addr));
 720 #endif
 721                         if (rt != NULL) {
 722                                 RT_UNLOCK(rt);
 723                                 rtfree(rt);
 724                         }
 725                         return -1;
 726                 }
 727                 RT_UNLOCK(rt);
 728                 rtfree(rt);
 729         }
 730
 731         return 0;
 732 }
 733
 734 static int
 735 stf_checkaddr6(
 736         struct stf_softc *sc,
 737         struct in6_addr *in6,
 738         struct ifnet *inifp)    /* incoming interface */
 739 {
 740         /*
 741          * check 6to4 addresses
 742          */
 743         if (IN6_IS_ADDR_6TO4(in6)) {
 744                 return stf_checkaddr4(sc, GET_V4(in6), inifp);
 745         }
 746
 747         /*
 748          * reject anything that look suspicious.  the test is implemented
 749          * in ip6_input too, but we check here as well to
 750          * (1) reject bad packets earlier, and
 751          * (2) to be safe against future ip6_input change.
 752          */
 753         if (IN6_IS_ADDR_V4COMPAT(in6) || IN6_IS_ADDR_V4MAPPED(in6)) {
 754                 return -1;
 755         }
 756
 757         return 0;
 758 }
 759
 760 static void
 761 in_stf_input(
 762         struct mbuf *m,
 763         int off)
 764 {
 765         struct stf_softc *sc;
 766         struct ip *ip;
 767         struct ip6_hdr ip6;
 768         u_int8_t otos, itos;
 769         int proto;
 770         struct ifnet *ifp;
 771         struct ifnet_stat_increment_param       stats;
 772
 773         ip = mtod(m, struct ip *);
 774         proto = ip->ip_p;
 775
 776         if (proto != IPPROTO_IPV6) {
 777                 m_freem(m);
 778                 return;
 779         }
 780
 781         ip = mtod(m, struct ip *);
 782
 783         sc = (struct stf_softc *)encap_getarg(m);
 784
 785         if (sc == NULL || (ifnet_flags(sc->sc_if) & IFF_UP) == 0) {
 786                 m_freem(m);
 787                 return;
 788         }
 789
 790         ifp = sc->sc_if;
 791
 792 #if MAC_LABEL
 793         mac_mbuf_label_associate_ifnet(ifp, m);
 794 #endif
 795
 796         /*
 797          * perform sanity check against outer src/dst.
 798          * for source, perform ingress filter as well.
 799          */
 800         if (stf_checkaddr4(sc, &ip->ip_dst, NULL) < 0 ||
 801             stf_checkaddr4(sc, &ip->ip_src, m->m_pkthdr.rcvif) < 0) {
 802                 m_freem(m);
 803                 return;
 804         }
 805
 806         otos = ip->ip_tos;
 807         mbuf_copydata(m, off, sizeof(ip6), &ip6);
 808
 809         /*
 810          * perform sanity check against inner src/dst.
 811          * for source, perform ingress filter as well.
 812          */
 813         if (stf_checkaddr6(sc, &ip6.ip6_dst, NULL) < 0 ||
 814             stf_checkaddr6(sc, &ip6.ip6_src, m->m_pkthdr.rcvif) < 0) {
 815                 m_freem(m);
 816                 return;
 817         }
 818
 819         itos = (ntohl(ip6.ip6_flow) >> 20) & 0xff;
 820         if ((ifnet_flags(ifp) & IFF_LINK1) != 0) {
 821                 ip_ecn_egress(ECN_NORMAL, &otos, &itos);
 822         } else {
 823                 ip_ecn_egress(ECN_NOCARE, &otos, &itos);
 824         }
 825         ip6.ip6_flow &= ~htonl(0xff << 20);
 826         ip6.ip6_flow |= htonl((u_int32_t)itos << 20);
 827
 828         m->m_pkthdr.rcvif = ifp;
 829         mbuf_pkthdr_setheader(m, mbuf_data(m));
 830         mbuf_adj(m, off);
 831
 832         if (ifp->if_bpf) {
 833                 /* We need to prepend the address family as a four byte field. */
 834                 u_int32_t af = AF_INET6;
 835                 bpf_tap_in(ifp, 0, m, &af, sizeof(af));
 836         }
 837
 838         /*
 839          * Put the packet to the network layer input queue according to the
 840          * specified address family.
 841          * See net/if_gif.c for possible issues with packet processing
 842          * reorder due to extra queueing.
 843          */
 844         bzero(&stats, sizeof(stats));
 845         stats.packets_in = 1;
 846         stats.bytes_in = mbuf_pkthdr_len(m);
 847         mbuf_pkthdr_setrcvif(m, ifp);
 848         ifnet_input(ifp, m, &stats);
 849
 850         return;
 851 }
 852
 853 static void
 854 stf_rtrequest(
 855         __unused int cmd,
 856         struct rtentry *rt,
 857         __unused struct sockaddr *sa)
 858 {
 859         if (rt != NULL) {
 860                 RT_LOCK_ASSERT_HELD(rt);
 861                 rt->rt_rmx.rmx_mtu = IPV6_MMTU;
 862         }
 863 }
 864
 865 static errno_t
 866 stf_ioctl(
 867         ifnet_t         ifp,
 868         u_long          cmd,
 869         void            *data)
 870 {
 871         struct ifaddr *ifa;
 872         struct ifreq *ifr;
 873         struct sockaddr_in6 *sin6;
 874         int error;
 875
 876         error = 0;
 877         switch (cmd) {
 878         case SIOCSIFADDR:
 879                 ifa = (struct ifaddr *)data;
 880                 if (ifa == NULL) {
 881                         error = EAFNOSUPPORT;
 882                         break;
 883                 }
 884                 IFA_LOCK(ifa);
 885                 if (ifa->ifa_addr->sa_family != AF_INET6) {
 886                         IFA_UNLOCK(ifa);
 887                         error = EAFNOSUPPORT;
 888                         break;
 889                 }
 890                 sin6 = (struct sockaddr_in6 *)(void *)ifa->ifa_addr;
 891                 if (IN6_IS_ADDR_6TO4(&sin6->sin6_addr)) {
 892                         if (!(ifnet_flags( ifp ) & IFF_UP)) {
 893                                 /* do this only if the interface is not already up */
 894                                 ifa->ifa_rtrequest = stf_rtrequest;
 895                                 IFA_UNLOCK(ifa);
 896                                 ifnet_set_flags(ifp, IFF_UP, IFF_UP);
 897                         } else {
 898                                 IFA_UNLOCK(ifa);
 899                         }
 900                 } else {
 901                         IFA_UNLOCK(ifa);
 902                         error = EINVAL;
 903                 }
 904                 IFA_LOCK_ASSERT_NOTHELD(ifa);
 905                 break;
 906
 907         case SIOCADDMULTI:
 908         case SIOCDELMULTI:
 909                 ifr = (struct ifreq *)data;
 910                 if (ifr && ifr->ifr_addr.sa_family == AF_INET6) {
 911                         ;
 912                 } else {
 913                         error = EAFNOSUPPORT;
 914                 }
 915                 break;
 916
 917         default:
 918                 error = EOPNOTSUPP;
 919                 break;
 920         }
 921
 922         return error;
 923 }