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         return (0);
 225 }
 226
 227
 228
 229 static errno_t
 230 stf_add_proto(
 231         ifnet_t                                                                 ifp,
 232         protocol_family_t                                               protocol_family,
 233         __unused const struct ifnet_demux_desc  *demux_array,
 234         __unused u_int32_t                                              demux_count)
 235 {
 236         /* Only one protocol may be attached at a time */
 237         struct stf_softc* stf = ifnet_softc(ifp);
 238         if (stf->sc_protocol_family == 0)
 239                 stf->sc_protocol_family = protocol_family;
 240         else {
 241                 printf("stf_add_proto: stf already has a proto\n");
 242                 return EBUSY;
 243         }
 244
 245         return 0;
 246 }
 247
 248 static errno_t
 249 stf_del_proto(
 250         ifnet_t                         ifp,
 251         protocol_family_t       protocol_family)
 252 {
 253         if (((struct stf_softc*)ifnet_softc(ifp))->sc_protocol_family == protocol_family)
 254                 ((struct stf_softc*)ifnet_softc(ifp))->sc_protocol_family = 0;
 255
 256         return 0;
 257 }
 258
 259 static errno_t
 260 stf_attach_inet6(
 261         ifnet_t                         ifp,
 262         protocol_family_t       protocol_family)
 263 {
 264     struct ifnet_attach_proto_param     reg;
 265     errno_t                                                     stat;
 266
 267     if (protocol_family != PF_INET6)
 268         return EPROTONOSUPPORT;
 269
 270         bzero(&reg, sizeof(reg));
 271     reg.input = stf_media_input;
 272     reg.pre_output = stf_pre_output;
 273
 274     stat = ifnet_attach_protocol(ifp, protocol_family, &reg);
 275     if (stat && stat != EEXIST) {
 276         printf("stf_attach_proto_family can't attach interface fam=%d\n",
 277                    protocol_family);
 278     }
 279
 280     return stat;
 281 }
 282
 283 static errno_t
 284 stf_demux(
 285         ifnet_t                                 ifp,
 286         __unused mbuf_t                 m,
 287         __unused char                   *frame_ptr,
 288         protocol_family_t               *protocol_family)
 289 {
 290         struct stf_softc* stf = ifnet_softc(ifp);
 291         *protocol_family = stf->sc_protocol_family;
 292         return 0;
 293 }
 294
 295 static errno_t
 296 stf_set_bpf_tap(
 297         ifnet_t                 ifp,
 298         bpf_tap_mode    mode,
 299         bpf_packet_func callback)
 300 {
 301         struct stf_softc        *sc = ifnet_softc(ifp);
 302
 303         sc->tap_mode = mode;
 304         sc->tap_callback = callback;
 305
 306         return 0;
 307 }
 308
 309 void
 310 stfattach(void)
 311 {
 312         struct stf_softc *sc;
 313         int error;
 314         const struct encaptab *p;
 315         struct ifnet_init_eparams       stf_init;
 316
 317         stfinit();
 318
 319         error = proto_register_plumber(PF_INET6, APPLE_IF_FAM_STF,
 320             stf_attach_inet6, NULL);
 321         if (error != 0)
 322                 printf("proto_register_plumber failed for AF_INET6 error=%d\n", error);
 323
 324         sc = _MALLOC(sizeof(struct stf_softc), M_DEVBUF, M_WAITOK | M_ZERO);
 325         if (sc == 0) {
 326                 printf("stf softc attach failed\n" );
 327                 return;
 328         }
 329
 330         p = encap_attach_func(AF_INET, IPPROTO_IPV6, stf_encapcheck,
 331             &in_stf_protosw, sc);
 332         if (p == NULL) {
 333                 printf("sftattach encap_attach_func failed\n");
 334                 FREE(sc, M_DEVBUF);
 335                 return;
 336         }
 337         sc->encap_cookie = p;
 338         lck_mtx_init(&sc->sc_ro_mtx, stf_mtx_grp, LCK_ATTR_NULL);
 339
 340         bzero(&stf_init, sizeof(stf_init));
 341         stf_init.ver = IFNET_INIT_CURRENT_VERSION;
 342         stf_init.len = sizeof (stf_init);
 343         stf_init.flags = IFNET_INIT_LEGACY;
 344         stf_init.name = "stf";
 345         stf_init.unit = 0;
 346         stf_init.type = IFT_STF;
 347         stf_init.family = IFNET_FAMILY_STF;
 348         stf_init.output = stf_output;
 349         stf_init.demux = stf_demux;
 350         stf_init.add_proto = stf_add_proto;
 351         stf_init.del_proto = stf_del_proto;
 352         stf_init.softc = sc;
 353         stf_init.ioctl = stf_ioctl;
 354         stf_init.set_bpf_tap = stf_set_bpf_tap;
 355
 356         error = ifnet_allocate_extended(&stf_init, &sc->sc_if);
 357         if (error != 0) {
 358                 printf("stfattach, ifnet_allocate failed - %d\n", error);
 359                 encap_detach(sc->encap_cookie);
 360                 lck_mtx_destroy(&sc->sc_ro_mtx, stf_mtx_grp);
 361                 FREE(sc, M_DEVBUF);
 362                 return;
 363         }
 364         ifnet_set_mtu(sc->sc_if, IPV6_MMTU);
 365         ifnet_set_flags(sc->sc_if, 0, 0xffff); /* clear all flags */
 366 #if 0
 367         /* turn off ingress filter */
 368         ifnet_set_flags(sc->sc_if, IFF_LINK2, IFF_LINK2);
 369 #endif
 370
 371 #if CONFIG_MACF_NET
 372         mac_ifnet_label_init(&sc->sc_if);
 373 #endif
 374
 375         error = ifnet_attach(sc->sc_if, NULL);
 376         if (error != 0) {
 377                 printf("stfattach: ifnet_attach returned error=%d\n", error);
 378                 encap_detach(sc->encap_cookie);
 379                 ifnet_release(sc->sc_if);
 380                 lck_mtx_destroy(&sc->sc_ro_mtx, stf_mtx_grp);
 381                 FREE(sc, M_DEVBUF);
 382                 return;
 383         }
 384
 385         bpfattach(sc->sc_if, DLT_NULL, sizeof(u_int));
 386
 387         return;
 388 }
 389
 390 static int
 391 stf_encapcheck(
 392         const struct mbuf *m,
 393         __unused int off,
 394         int proto,
 395         void *arg)
 396 {
 397         struct ip ip;
 398         struct in6_ifaddr *ia6;
 399         struct stf_softc *sc;
 400         struct in_addr a, b;
 401
 402         sc = (struct stf_softc *)arg;
 403         if (sc == NULL)
 404                 return 0;
 405
 406         if ((ifnet_flags(sc->sc_if) & IFF_UP) == 0)
 407                 return 0;
 408
 409         /* IFF_LINK0 means "no decapsulation" */
 410         if ((ifnet_flags(sc->sc_if) & IFF_LINK0) != 0)
 411                 return 0;
 412
 413         if (proto != IPPROTO_IPV6)
 414                 return 0;
 415
 416         mbuf_copydata((struct mbuf *)(size_t)m, 0, sizeof(ip), &ip);
 417
 418         if (ip.ip_v != 4)
 419                 return 0;
 420
 421         ia6 = stf_getsrcifa6(sc->sc_if);
 422         if (ia6 == NULL)
 423                 return 0;
 424
 425         /*
 426          * check if IPv4 dst matches the IPv4 address derived from the
 427          * local 6to4 address.
 428          * success on: dst = 10.1.1.1, ia6->ia_addr = 2002:0a01:0101:...
 429          */
 430         IFA_LOCK(&ia6->ia_ifa);
 431         if (bcmp(GET_V4(&ia6->ia_addr.sin6_addr), &ip.ip_dst,
 432             sizeof(ip.ip_dst)) != 0) {
 433                 IFA_UNLOCK(&ia6->ia_ifa);
 434                 IFA_REMREF(&ia6->ia_ifa);
 435                 return 0;
 436         }
 437         /*
 438          * check if IPv4 src matches the IPv4 address derived from the
 439          * local 6to4 address masked by prefixmask.
 440          * success on: src = 10.1.1.1, ia6->ia_addr = 2002:0a00:.../24
 441          * fail on: src = 10.1.1.1, ia6->ia_addr = 2002:0b00:.../24
 442          */
 443         bzero(&a, sizeof(a));
 444         a.s_addr = GET_V4(&ia6->ia_addr.sin6_addr)->s_addr;
 445         a.s_addr &= GET_V4(&ia6->ia_prefixmask.sin6_addr)->s_addr;
 446         b = ip.ip_src;
 447         b.s_addr &= GET_V4(&ia6->ia_prefixmask.sin6_addr)->s_addr;
 448         if (a.s_addr != b.s_addr) {
 449                 IFA_UNLOCK(&ia6->ia_ifa);
 450                 IFA_REMREF(&ia6->ia_ifa);
 451                 return 0;
 452         }
 453         /* stf interface makes single side match only */
 454         IFA_UNLOCK(&ia6->ia_ifa);
 455         IFA_REMREF(&ia6->ia_ifa);
 456         return 32;
 457 }
 458
 459 static struct in6_ifaddr *
 460 stf_getsrcifa6(struct ifnet *ifp)
 461 {
 462         struct ifaddr *ia;
 463         struct in_ifaddr *ia4;
 464         struct sockaddr_in6 *sin6;
 465         struct in_addr in;
 466
 467         ifnet_lock_shared(ifp);
 468         for (ia = ifp->if_addrlist.tqh_first; ia; ia = ia->ifa_list.tqe_next) {
 469                 IFA_LOCK(ia);
 470                 if (ia->ifa_addr == NULL) {
 471                         IFA_UNLOCK(ia);
 472                         continue;
 473                 }
 474                 if (ia->ifa_addr->sa_family != AF_INET6) {
 475                         IFA_UNLOCK(ia);
 476                         continue;
 477                 }
 478                 sin6 = (struct sockaddr_in6 *)(void *)ia->ifa_addr;
 479                 if (!IN6_IS_ADDR_6TO4(&sin6->sin6_addr)) {
 480                         IFA_UNLOCK(ia);
 481                         continue;
 482                 }
 483                 bcopy(GET_V4(&sin6->sin6_addr), &in, sizeof(in));
 484                 IFA_UNLOCK(ia);
 485                 lck_rw_lock_shared(in_ifaddr_rwlock);
 486                 for (ia4 = TAILQ_FIRST(&in_ifaddrhead);
 487                      ia4;
 488                      ia4 = TAILQ_NEXT(ia4, ia_link))
 489                 {
 490                         IFA_LOCK(&ia4->ia_ifa);
 491                         if (ia4->ia_addr.sin_addr.s_addr == in.s_addr) {
 492                                 IFA_UNLOCK(&ia4->ia_ifa);
 493                                 break;
 494                         }
 495                         IFA_UNLOCK(&ia4->ia_ifa);
 496                 }
 497                 lck_rw_done(in_ifaddr_rwlock);
 498                 if (ia4 == NULL)
 499                         continue;
 500
 501                 IFA_ADDREF(ia);         /* for caller */
 502                 ifnet_lock_done(ifp);
 503                 return ((struct in6_ifaddr *)ia);
 504         }
 505         ifnet_lock_done(ifp);
 506
 507         return (NULL);
 508 }
 509
 510 int
 511 stf_pre_output(
 512         struct ifnet    *ifp,
 513         __unused protocol_family_t  protocol_family,
 514         struct mbuf     **m0,
 515         const struct sockaddr   *dst,
 516         __unused void *route,
 517         __unused char *desk_linkaddr,
 518         __unused char *frame_type)
 519 {
 520         struct mbuf *m = *m0;
 521         struct stf_softc *sc;
 522         const struct sockaddr_in6 *dst6;
 523         const struct in_addr *in4;
 524         u_int8_t tos;
 525         struct ip *ip;
 526         struct ip6_hdr *ip6;
 527         struct in6_ifaddr *ia6;
 528         struct sockaddr_in      *dst4;
 529         struct ip_out_args ipoa;
 530         errno_t result = 0;
 531
 532         bzero(&ipoa, sizeof(ipoa));
 533         ipoa.ipoa_boundif = IFSCOPE_NONE;
 534         ipoa.ipoa_flags = IPOAF_SELECT_SRCIF;
 535         ipoa.ipoa_sotc = SO_TC_UNSPEC;
 536         ipoa.ipoa_netsvctype = _NET_SERVICE_TYPE_UNSPEC;
 537
 538         sc = ifnet_softc(ifp);
 539         dst6 = (const struct sockaddr_in6 *)(const void *)dst;
 540
 541         /* just in case */
 542         if ((ifnet_flags(ifp) & IFF_UP) == 0) {
 543                 printf("stf: IFF_DOWN\n");
 544                 return ENETDOWN;
 545         }
 546
 547         /*
 548          * If we don't have an ip4 address that match my inner ip6 address,
 549          * we shouldn't generate output.  Without this check, we'll end up
 550          * using wrong IPv4 source.
 551          */
 552         ia6 = stf_getsrcifa6(ifp);
 553         if (ia6 == NULL) {
 554                 return ENETDOWN;
 555         }
 556
 557         if (mbuf_len(m) < sizeof(*ip6)) {
 558                 m = m_pullup(m, sizeof(*ip6));
 559                 if (!m) {
 560                         *m0 = NULL; /* makes sure this won't be double freed */
 561                         IFA_REMREF(&ia6->ia_ifa);
 562                         return ENOBUFS;
 563                 }
 564         }
 565         ip6 = mtod(m, struct ip6_hdr *);
 566         tos = (ntohl(ip6->ip6_flow) >> 20) & 0xff;
 567
 568         /*
 569          * Pickup the right outer dst addr from the list of candidates.
 570          * ip6_dst has priority as it may be able to give us shorter IPv4 hops.
 571          */
 572         if (IN6_IS_ADDR_6TO4(&ip6->ip6_dst))
 573                 in4 = GET_V4(&ip6->ip6_dst);
 574         else if (IN6_IS_ADDR_6TO4(&dst6->sin6_addr))
 575                 in4 = GET_V4(&dst6->sin6_addr);
 576         else {
 577                 IFA_REMREF(&ia6->ia_ifa);
 578                 return ENETUNREACH;
 579         }
 580
 581         if (ifp->if_bpf) {
 582                 /* We need to prepend the address family as a four byte field. */
 583                 u_int32_t af = AF_INET6;
 584
 585                 bpf_tap_out(ifp, 0, m, &af, sizeof(af));
 586         }
 587
 588         M_PREPEND(m, sizeof(struct ip), M_DONTWAIT, 1);
 589         if (m && mbuf_len(m) < sizeof(struct ip))
 590                 m = m_pullup(m, sizeof(struct ip));
 591         if (m == NULL) {
 592                 *m0 = NULL;
 593                 IFA_REMREF(&ia6->ia_ifa);
 594                 return ENOBUFS;
 595         }
 596         ip = mtod(m, struct ip *);
 597
 598         bzero(ip, sizeof(*ip));
 599
 600         IFA_LOCK_SPIN(&ia6->ia_ifa);
 601         bcopy(GET_V4(&((struct sockaddr_in6 *)&ia6->ia_addr)->sin6_addr),
 602             &ip->ip_src, sizeof(ip->ip_src));
 603         IFA_UNLOCK(&ia6->ia_ifa);
 604         bcopy(in4, &ip->ip_dst, sizeof(ip->ip_dst));
 605         ip->ip_p = IPPROTO_IPV6;
 606         ip->ip_ttl = ip_stf_ttl;
 607         ip->ip_len = m->m_pkthdr.len;   /*host order*/
 608         if (ifp->if_flags & IFF_LINK1)
 609                 ip_ecn_ingress(ECN_NORMAL, &ip->ip_tos, &tos);
 610         else
 611                 ip_ecn_ingress(ECN_NOCARE, &ip->ip_tos, &tos);
 612
 613         lck_mtx_lock(&sc->sc_ro_mtx);
 614         dst4 = (struct sockaddr_in *)(void *)&sc->sc_ro.ro_dst;
 615         if (ROUTE_UNUSABLE(&sc->sc_ro) || dst4->sin_family != AF_INET ||
 616             bcmp(&dst4->sin_addr, &ip->ip_dst, sizeof(ip->ip_dst)) != 0) {
 617                 ROUTE_RELEASE(&sc->sc_ro);
 618                 /* cache route doesn't match: always the case during the first use */
 619                 dst4->sin_family = AF_INET;
 620                 dst4->sin_len = sizeof(struct sockaddr_in);
 621                 bcopy(&ip->ip_dst, &dst4->sin_addr, sizeof(dst4->sin_addr));
 622         }
 623
 624         result = ip_output(m, NULL, &sc->sc_ro, IP_OUTARGS, NULL, &ipoa);
 625         lck_mtx_unlock(&sc->sc_ro_mtx);
 626
 627         /* Assumption: ip_output will free mbuf on errors */
 628         /* All the output processing is done here, don't let stf_output be called */
 629         if (result == 0)
 630                 result = EJUSTRETURN;
 631         *m0 = NULL;
 632         IFA_REMREF(&ia6->ia_ifa);
 633         return result;
 634 }
 635 static errno_t
 636 stf_output(
 637         __unused ifnet_t        ifp,
 638         __unused mbuf_t m)
 639 {
 640         /* All processing is done in stf_pre_output
 641          * this shouldn't be called as the pre_output returns "EJUSTRETURN"
 642          */
 643         return 0;
 644 }
 645
 646 static int
 647 stf_checkaddr4(
 648         struct stf_softc *sc,
 649         const struct in_addr *in,
 650         struct ifnet *inifp)    /* incoming interface */
 651 {
 652         struct in_ifaddr *ia4;
 653
 654         /*
 655          * reject packets with the following address:
 656          * 224.0.0.0/4 0.0.0.0/8 127.0.0.0/8 255.0.0.0/8
 657          */
 658         if (IN_MULTICAST(ntohl(in->s_addr)))
 659                 return -1;
 660         switch ((ntohl(in->s_addr) & 0xff000000) >> 24) {
 661         case 0: case 127: case 255:
 662                 return -1;
 663         }
 664
 665         /*
 666          * reject packets with broadcast
 667          */
 668         lck_rw_lock_shared(in_ifaddr_rwlock);
 669         for (ia4 = TAILQ_FIRST(&in_ifaddrhead);
 670              ia4;
 671              ia4 = TAILQ_NEXT(ia4, ia_link))
 672         {
 673                 IFA_LOCK(&ia4->ia_ifa);
 674                 if ((ia4->ia_ifa.ifa_ifp->if_flags & IFF_BROADCAST) == 0) {
 675                         IFA_UNLOCK(&ia4->ia_ifa);
 676                         continue;
 677                 }
 678                 if (in->s_addr == ia4->ia_broadaddr.sin_addr.s_addr) {
 679                         IFA_UNLOCK(&ia4->ia_ifa);
 680                         lck_rw_done(in_ifaddr_rwlock);
 681                         return -1;
 682                 }
 683                 IFA_UNLOCK(&ia4->ia_ifa);
 684         }
 685         lck_rw_done(in_ifaddr_rwlock);
 686
 687         /*
 688          * perform ingress filter
 689          */
 690         if (sc && (ifnet_flags(sc->sc_if) & IFF_LINK2) == 0 && inifp) {
 691                 struct sockaddr_in sin;
 692                 struct rtentry *rt;
 693
 694                 bzero(&sin, sizeof(sin));
 695                 sin.sin_family = AF_INET;
 696                 sin.sin_len = sizeof(struct sockaddr_in);
 697                 sin.sin_addr = *in;
 698                 rt = rtalloc1((struct sockaddr *)&sin, 0, 0);
 699                 if (rt != NULL)
 700                         RT_LOCK(rt);
 701                 if (rt == NULL || rt->rt_ifp != inifp) {
 702 #if 1
 703                         log(LOG_WARNING, "%s: packet from 0x%x dropped "
 704                             "due to ingress filter\n", if_name(sc->sc_if),
 705                             (u_int32_t)ntohl(sin.sin_addr.s_addr));
 706 #endif
 707                         if (rt != NULL) {
 708                                 RT_UNLOCK(rt);
 709                                 rtfree(rt);
 710                         }
 711                         return -1;
 712                 }
 713                 RT_UNLOCK(rt);
 714                 rtfree(rt);
 715         }
 716
 717         return 0;
 718 }
 719
 720 static int
 721 stf_checkaddr6(
 722         struct stf_softc *sc,
 723         struct in6_addr *in6,
 724         struct ifnet *inifp)    /* incoming interface */
 725 {
 726         /*
 727          * check 6to4 addresses
 728          */
 729         if (IN6_IS_ADDR_6TO4(in6))
 730                 return stf_checkaddr4(sc, GET_V4(in6), inifp);
 731
 732         /*
 733          * reject anything that look suspicious.  the test is implemented
 734          * in ip6_input too, but we check here as well to
 735          * (1) reject bad packets earlier, and
 736          * (2) to be safe against future ip6_input change.
 737          */
 738         if (IN6_IS_ADDR_V4COMPAT(in6) || IN6_IS_ADDR_V4MAPPED(in6))
 739                 return -1;
 740
 741         return 0;
 742 }
 743
 744 static void
 745 in_stf_input(
 746         struct mbuf *m,
 747         int off)
 748 {
 749         struct stf_softc *sc;
 750         struct ip *ip;
 751         struct ip6_hdr ip6;
 752         u_int8_t otos, itos;
 753         int proto;
 754         struct ifnet *ifp;
 755         struct ifnet_stat_increment_param       stats;
 756
 757         ip = mtod(m, struct ip *);
 758         proto = ip->ip_p;
 759
 760         if (proto != IPPROTO_IPV6) {
 761                 m_freem(m);
 762                 return;
 763         }
 764
 765         ip = mtod(m, struct ip *);
 766
 767         sc = (struct stf_softc *)encap_getarg(m);
 768
 769         if (sc == NULL || (ifnet_flags(sc->sc_if) & IFF_UP) == 0) {
 770                 m_freem(m);
 771                 return;
 772         }
 773
 774         ifp = sc->sc_if;
 775
 776 #if MAC_LABEL
 777         mac_mbuf_label_associate_ifnet(ifp, m);
 778 #endif
 779
 780         /*
 781          * perform sanity check against outer src/dst.
 782          * for source, perform ingress filter as well.
 783          */
 784         if (stf_checkaddr4(sc, &ip->ip_dst, NULL) < 0 ||
 785             stf_checkaddr4(sc, &ip->ip_src, m->m_pkthdr.rcvif) < 0) {
 786                 m_freem(m);
 787                 return;
 788         }
 789
 790         otos = ip->ip_tos;
 791         mbuf_copydata(m, off, sizeof(ip6), &ip6);
 792
 793         /*
 794          * perform sanity check against inner src/dst.
 795          * for source, perform ingress filter as well.
 796          */
 797         if (stf_checkaddr6(sc, &ip6.ip6_dst, NULL) < 0 ||
 798             stf_checkaddr6(sc, &ip6.ip6_src, m->m_pkthdr.rcvif) < 0) {
 799                 m_freem(m);
 800                 return;
 801         }
 802
 803         itos = (ntohl(ip6.ip6_flow) >> 20) & 0xff;
 804         if ((ifnet_flags(ifp) & IFF_LINK1) != 0)
 805                 ip_ecn_egress(ECN_NORMAL, &otos, &itos);
 806         else
 807                 ip_ecn_egress(ECN_NOCARE, &otos, &itos);
 808         ip6.ip6_flow &= ~htonl(0xff << 20);
 809         ip6.ip6_flow |= htonl((u_int32_t)itos << 20);
 810
 811         m->m_pkthdr.rcvif = ifp;
 812         mbuf_pkthdr_setheader(m, mbuf_data(m));
 813         mbuf_adj(m, off);
 814
 815         if (ifp->if_bpf) {
 816                 /* We need to prepend the address family as a four byte field. */
 817                 u_int32_t af = AF_INET6;
 818                 bpf_tap_in(ifp, 0, m, &af, sizeof(af));
 819         }
 820
 821         /*
 822          * Put the packet to the network layer input queue according to the
 823          * specified address family.
 824          * See net/if_gif.c for possible issues with packet processing
 825          * reorder due to extra queueing.
 826          */
 827         bzero(&stats, sizeof(stats));
 828         stats.packets_in = 1;
 829         stats.bytes_in = mbuf_pkthdr_len(m);
 830         mbuf_pkthdr_setrcvif(m, ifp);
 831         ifnet_input(ifp, m, &stats);
 832
 833         return;
 834 }
 835
 836 static void
 837 stf_rtrequest(
 838         __unused int cmd,
 839         struct rtentry *rt,
 840         __unused struct sockaddr *sa)
 841 {
 842         if (rt != NULL) {
 843                 RT_LOCK_ASSERT_HELD(rt);
 844                 rt->rt_rmx.rmx_mtu = IPV6_MMTU;
 845         }
 846 }
 847
 848 static errno_t
 849 stf_ioctl(
 850         ifnet_t         ifp,
 851         u_long          cmd,
 852         void            *data)
 853 {
 854         struct ifaddr *ifa;
 855         struct ifreq *ifr;
 856         struct sockaddr_in6 *sin6;
 857         int error;
 858
 859         error = 0;
 860         switch (cmd) {
 861         case SIOCSIFADDR:
 862                 ifa = (struct ifaddr *)data;
 863                 if (ifa == NULL) {
 864                         error = EAFNOSUPPORT;
 865                         break;
 866                 }
 867                 IFA_LOCK(ifa);
 868                 if (ifa->ifa_addr->sa_family != AF_INET6) {
 869                         IFA_UNLOCK(ifa);
 870                         error = EAFNOSUPPORT;
 871                         break;
 872                 }
 873                 sin6 = (struct sockaddr_in6 *)(void *)ifa->ifa_addr;
 874                 if (IN6_IS_ADDR_6TO4(&sin6->sin6_addr)) {
 875                         if ( !(ifnet_flags( ifp ) & IFF_UP) ) {
 876                                 /* do this only if the interface is not already up */
 877                                 ifa->ifa_rtrequest = stf_rtrequest;
 878                                 IFA_UNLOCK(ifa);
 879                                 ifnet_set_flags(ifp, IFF_UP, IFF_UP);
 880                         } else {
 881                                 IFA_UNLOCK(ifa);
 882                         }
 883                 } else {
 884                         IFA_UNLOCK(ifa);
 885                         error = EINVAL;
 886                 }
 887                 IFA_LOCK_ASSERT_NOTHELD(ifa);
 888                 break;
 889
 890         case SIOCADDMULTI:
 891         case SIOCDELMULTI:
 892                 ifr = (struct ifreq *)data;
 893                 if (ifr && ifr->ifr_addr.sa_family == AF_INET6)
 894                         ;
 895                 else
 896                         error = EAFNOSUPPORT;
 897                 break;
 898
 899         default:
 900                 error = EOPNOTSUPP;
 901                 break;
 902         }
 903
 904         return error;
 905 }