bsd/net/if_stf.c

   1 /*
   2  * Copyright (c) 2000-2020 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 #define GET_V4(x) ((const struct in_addr *)(const void *)(&(x)->s6_addr16[1]))
 150
 151 static lck_grp_t *stf_mtx_grp;
 152
 153 struct stf_softc {
 154         ifnet_t                         sc_if;     /* common area */
 155         u_int32_t                               sc_protocol_family; /* dlil protocol attached */
 156         union {
 157                 struct route  __sc_ro4;
 158                 struct route_in6 __sc_ro6; /* just for safety */
 159         } __sc_ro46;
 160 #define sc_ro   __sc_ro46.__sc_ro4
 161         decl_lck_mtx_data(, sc_ro_mtx);
 162         const struct encaptab *encap_cookie;
 163         bpf_tap_mode            tap_mode;
 164         bpf_packet_func         tap_callback;
 165 };
 166
 167 void stfattach(void);
 168
 169 static int ip_stf_ttl = 40;
 170 static int stf_init_done;
 171
 172 static void in_stf_input(struct mbuf *, int);
 173 static void stfinit(void);
 174
 175 static struct protosw in_stf_protosw =
 176 {
 177         .pr_type =              SOCK_RAW,
 178         .pr_protocol =          IPPROTO_IPV6,
 179         .pr_flags =             PR_ATOMIC | PR_ADDR,
 180         .pr_input =             in_stf_input,
 181         .pr_ctloutput =         rip_ctloutput,
 182         .pr_usrreqs =           &rip_usrreqs,
 183         .pr_unlock =            rip_unlock,
 184 };
 185
 186 static int stf_encapcheck(const struct mbuf *, int, int, void *);
 187 static struct in6_ifaddr *stf_getsrcifa6(struct ifnet *);
 188 int stf_pre_output(struct ifnet *, protocol_family_t, struct mbuf **,
 189     const struct sockaddr *, void *, char *, char *);
 190 static int stf_checkaddr4(struct stf_softc *, const struct in_addr *,
 191     struct ifnet *);
 192 static int stf_checkaddr6(struct stf_softc *, struct in6_addr *,
 193     struct ifnet *);
 194 static void stf_rtrequest(int, struct rtentry *, struct sockaddr *);
 195 static errno_t stf_ioctl(ifnet_t ifp, u_long cmd, void *data);
 196 static errno_t stf_output(ifnet_t ifp, mbuf_t m);
 197
 198 static void
 199 stfinit(void)
 200 {
 201         if (!stf_init_done) {
 202                 stf_mtx_grp = lck_grp_alloc_init("stf", LCK_GRP_ATTR_NULL);
 203                 stf_init_done = 1;
 204         }
 205 }
 206
 207 /*
 208  * gif_input is the input handler for IP and IPv6 attached to gif
 209  */
 210 static errno_t
 211 stf_media_input(
 212         __unused ifnet_t        ifp,
 213         protocol_family_t       protocol_family,
 214         mbuf_t                          m,
 215         __unused char           *frame_header)
 216 {
 217         if (proto_input(protocol_family, m) != 0) {
 218                 m_freem(m);
 219         }
 220
 221         return 0;
 222 }
 223
 224
 225
 226 static errno_t
 227 stf_add_proto(
 228         ifnet_t                                                                 ifp,
 229         protocol_family_t                                               protocol_family,
 230         __unused const struct ifnet_demux_desc  *demux_array,
 231         __unused u_int32_t                                              demux_count)
 232 {
 233         /* Only one protocol may be attached at a time */
 234         struct stf_softc* stf = ifnet_softc(ifp);
 235         if (stf->sc_protocol_family == 0) {
 236                 stf->sc_protocol_family = protocol_family;
 237         } else {
 238                 printf("stf_add_proto: stf already has a proto\n");
 239                 return EBUSY;
 240         }
 241
 242         return 0;
 243 }
 244
 245 static errno_t
 246 stf_del_proto(
 247         ifnet_t                         ifp,
 248         protocol_family_t       protocol_family)
 249 {
 250         if (((struct stf_softc*)ifnet_softc(ifp))->sc_protocol_family == protocol_family) {
 251                 ((struct stf_softc*)ifnet_softc(ifp))->sc_protocol_family = 0;
 252         }
 253
 254         return 0;
 255 }
 256
 257 static errno_t
 258 stf_attach_inet6(
 259         ifnet_t                         ifp,
 260         protocol_family_t       protocol_family)
 261 {
 262         struct ifnet_attach_proto_param     reg;
 263         errno_t                                                     stat;
 264
 265         if (protocol_family != PF_INET6) {
 266                 return EPROTONOSUPPORT;
 267         }
 268
 269         bzero(&reg, sizeof(reg));
 270         reg.input = stf_media_input;
 271         reg.pre_output = stf_pre_output;
 272
 273         stat = ifnet_attach_protocol(ifp, protocol_family, &reg);
 274         if (stat && stat != EEXIST) {
 275                 printf("stf_attach_proto_family can't attach interface fam=%d\n",
 276                     protocol_family);
 277         }
 278
 279         return stat;
 280 }
 281
 282 static errno_t
 283 stf_demux(
 284         ifnet_t                                 ifp,
 285         __unused mbuf_t                 m,
 286         __unused char                   *frame_ptr,
 287         protocol_family_t               *protocol_family)
 288 {
 289         struct stf_softc* stf = ifnet_softc(ifp);
 290         *protocol_family = stf->sc_protocol_family;
 291         return 0;
 292 }
 293
 294 static errno_t
 295 stf_set_bpf_tap(
 296         ifnet_t                 ifp,
 297         bpf_tap_mode    mode,
 298         bpf_packet_func callback)
 299 {
 300         struct stf_softc        *sc = ifnet_softc(ifp);
 301
 302         sc->tap_mode = mode;
 303         sc->tap_callback = callback;
 304
 305         return 0;
 306 }
 307
 308 void
 309 stfattach(void)
 310 {
 311         struct stf_softc *sc;
 312         int error;
 313         const struct encaptab *p;
 314         struct ifnet_init_eparams       stf_init;
 315
 316         stfinit();
 317
 318         error = proto_register_plumber(PF_INET6, APPLE_IF_FAM_STF,
 319             stf_attach_inet6, NULL);
 320         if (error != 0) {
 321                 printf("proto_register_plumber failed for AF_INET6 error=%d\n", error);
 322         }
 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         error = ifnet_attach(sc->sc_if, NULL);
 372         if (error != 0) {
 373                 printf("stfattach: ifnet_attach returned error=%d\n", error);
 374                 encap_detach(sc->encap_cookie);
 375                 ifnet_release(sc->sc_if);
 376                 lck_mtx_destroy(&sc->sc_ro_mtx, stf_mtx_grp);
 377                 FREE(sc, M_DEVBUF);
 378                 return;
 379         }
 380
 381         bpfattach(sc->sc_if, DLT_NULL, sizeof(u_int));
 382
 383         return;
 384 }
 385
 386 static int
 387 stf_encapcheck(
 388         const struct mbuf *m,
 389         __unused int off,
 390         int proto,
 391         void *arg)
 392 {
 393         struct ip ip;
 394         struct in6_ifaddr *ia6;
 395         struct stf_softc *sc;
 396         struct in_addr a, b;
 397
 398         sc = (struct stf_softc *)arg;
 399         if (sc == NULL) {
 400                 return 0;
 401         }
 402
 403         if ((ifnet_flags(sc->sc_if) & IFF_UP) == 0) {
 404                 return 0;
 405         }
 406
 407         /* IFF_LINK0 means "no decapsulation" */
 408         if ((ifnet_flags(sc->sc_if) & IFF_LINK0) != 0) {
 409                 return 0;
 410         }
 411
 412         if (proto != IPPROTO_IPV6) {
 413                 return 0;
 414         }
 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
 422         ia6 = stf_getsrcifa6(sc->sc_if);
 423         if (ia6 == NULL) {
 424                 return 0;
 425         }
 426
 427         /*
 428          * check if IPv4 dst matches the IPv4 address derived from the
 429          * local 6to4 address.
 430          * success on: dst = 10.1.1.1, ia6->ia_addr = 2002:0a01:0101:...
 431          */
 432         IFA_LOCK(&ia6->ia_ifa);
 433         if (bcmp(GET_V4(&ia6->ia_addr.sin6_addr), &ip.ip_dst,
 434             sizeof(ip.ip_dst)) != 0) {
 435                 IFA_UNLOCK(&ia6->ia_ifa);
 436                 IFA_REMREF(&ia6->ia_ifa);
 437                 return 0;
 438         }
 439         /*
 440          * check if IPv4 src matches the IPv4 address derived from the
 441          * local 6to4 address masked by prefixmask.
 442          * success on: src = 10.1.1.1, ia6->ia_addr = 2002:0a00:.../24
 443          * fail on: src = 10.1.1.1, ia6->ia_addr = 2002:0b00:.../24
 444          */
 445         bzero(&a, sizeof(a));
 446         a.s_addr = GET_V4(&ia6->ia_addr.sin6_addr)->s_addr;
 447         a.s_addr &= GET_V4(&ia6->ia_prefixmask.sin6_addr)->s_addr;
 448         b = ip.ip_src;
 449         b.s_addr &= GET_V4(&ia6->ia_prefixmask.sin6_addr)->s_addr;
 450         if (a.s_addr != b.s_addr) {
 451                 IFA_UNLOCK(&ia6->ia_ifa);
 452                 IFA_REMREF(&ia6->ia_ifa);
 453                 return 0;
 454         }
 455         /* stf interface makes single side match only */
 456         IFA_UNLOCK(&ia6->ia_ifa);
 457         IFA_REMREF(&ia6->ia_ifa);
 458         return 32;
 459 }
 460
 461 static struct in6_ifaddr *
 462 stf_getsrcifa6(struct ifnet *ifp)
 463 {
 464         struct ifaddr *ia;
 465         struct in_ifaddr *ia4;
 466         struct sockaddr_in6 *sin6;
 467         struct in_addr in;
 468
 469         ifnet_lock_shared(ifp);
 470         for (ia = ifp->if_addrlist.tqh_first; ia; ia = ia->ifa_list.tqe_next) {
 471                 IFA_LOCK(ia);
 472                 if (ia->ifa_addr == NULL) {
 473                         IFA_UNLOCK(ia);
 474                         continue;
 475                 }
 476                 if (ia->ifa_addr->sa_family != AF_INET6) {
 477                         IFA_UNLOCK(ia);
 478                         continue;
 479                 }
 480                 sin6 = (struct sockaddr_in6 *)(void *)ia->ifa_addr;
 481                 if (!IN6_IS_ADDR_6TO4(&sin6->sin6_addr)) {
 482                         IFA_UNLOCK(ia);
 483                         continue;
 484                 }
 485                 bcopy(GET_V4(&sin6->sin6_addr), &in, sizeof(in));
 486                 IFA_UNLOCK(ia);
 487                 lck_rw_lock_shared(in_ifaddr_rwlock);
 488                 for (ia4 = TAILQ_FIRST(&in_ifaddrhead);
 489                     ia4;
 490                     ia4 = TAILQ_NEXT(ia4, ia_link)) {
 491                         IFA_LOCK(&ia4->ia_ifa);
 492                         if (ia4->ia_addr.sin_addr.s_addr == in.s_addr) {
 493                                 IFA_UNLOCK(&ia4->ia_ifa);
 494                                 break;
 495                         }
 496                         IFA_UNLOCK(&ia4->ia_ifa);
 497                 }
 498                 lck_rw_done(in_ifaddr_rwlock);
 499                 if (ia4 == NULL) {
 500                         continue;
 501                 }
 502
 503                 IFA_ADDREF(ia);         /* for caller */
 504                 ifnet_lock_done(ifp);
 505                 return (struct in6_ifaddr *)ia;
 506         }
 507         ifnet_lock_done(ifp);
 508
 509         return NULL;
 510 }
 511
 512 int
 513 stf_pre_output(
 514         struct ifnet    *ifp,
 515         __unused protocol_family_t  protocol_family,
 516         struct mbuf     **m0,
 517         const struct sockaddr   *dst,
 518         __unused void *route,
 519         __unused char *desk_linkaddr,
 520         __unused char *frame_type)
 521 {
 522         struct mbuf *m = *m0;
 523         struct stf_softc *sc;
 524         const struct sockaddr_in6 *dst6;
 525         const struct in_addr *in4;
 526         u_int8_t tos;
 527         struct ip *ip;
 528         struct ip6_hdr *ip6;
 529         struct in6_ifaddr *ia6;
 530         struct sockaddr_in      *dst4;
 531         struct ip_out_args ipoa;
 532         errno_t result = 0;
 533
 534         bzero(&ipoa, sizeof(ipoa));
 535         ipoa.ipoa_boundif = IFSCOPE_NONE;
 536         ipoa.ipoa_flags = IPOAF_SELECT_SRCIF;
 537         ipoa.ipoa_sotc = SO_TC_UNSPEC;
 538         ipoa.ipoa_netsvctype = _NET_SERVICE_TYPE_UNSPEC;
 539
 540         sc = ifnet_softc(ifp);
 541         dst6 = (const struct sockaddr_in6 *)(const void *)dst;
 542
 543         /* just in case */
 544         if ((ifnet_flags(ifp) & IFF_UP) == 0) {
 545                 printf("stf: IFF_DOWN\n");
 546                 return ENETDOWN;
 547         }
 548
 549         /*
 550          * If we don't have an ip4 address that match my inner ip6 address,
 551          * we shouldn't generate output.  Without this check, we'll end up
 552          * using wrong IPv4 source.
 553          */
 554         ia6 = stf_getsrcifa6(ifp);
 555         if (ia6 == NULL) {
 556                 return ENETDOWN;
 557         }
 558
 559         if (mbuf_len(m) < sizeof(*ip6)) {
 560                 m = m_pullup(m, sizeof(*ip6));
 561                 if (!m) {
 562                         *m0 = NULL; /* makes sure this won't be double freed */
 563                         IFA_REMREF(&ia6->ia_ifa);
 564                         return ENOBUFS;
 565                 }
 566                 *m0 = m;
 567         }
 568         ip6 = mtod(m, struct ip6_hdr *);
 569         tos = (ntohl(ip6->ip6_flow) >> 20) & 0xff;
 570
 571         /*
 572          * Pickup the right outer dst addr from the list of candidates.
 573          * ip6_dst has priority as it may be able to give us shorter IPv4 hops.
 574          */
 575         if (IN6_IS_ADDR_6TO4(&ip6->ip6_dst)) {
 576                 in4 = GET_V4(&ip6->ip6_dst);
 577         } else if (IN6_IS_ADDR_6TO4(&dst6->sin6_addr)) {
 578                 in4 = GET_V4(&dst6->sin6_addr);
 579         } else {
 580                 IFA_REMREF(&ia6->ia_ifa);
 581                 return ENETUNREACH;
 582         }
 583
 584         if (ifp->if_bpf) {
 585                 /* We need to prepend the address family as a four byte field. */
 586                 u_int32_t af = AF_INET6;
 587
 588                 bpf_tap_out(ifp, 0, m, &af, sizeof(af));
 589         }
 590
 591         M_PREPEND(m, sizeof(struct ip), M_DONTWAIT, 1);
 592         if (m && mbuf_len(m) < sizeof(struct ip)) {
 593                 m = m_pullup(m, sizeof(struct ip));
 594         }
 595         if (m == NULL) {
 596                 *m0 = NULL;
 597                 IFA_REMREF(&ia6->ia_ifa);
 598                 return ENOBUFS;
 599         }
 600
 601         *m0 = m;
 602         ip = mtod(m, struct ip *);
 603
 604         bzero(ip, sizeof(*ip));
 605
 606         IFA_LOCK_SPIN(&ia6->ia_ifa);
 607         bcopy(GET_V4(&((struct sockaddr_in6 *)&ia6->ia_addr)->sin6_addr),
 608             &ip->ip_src, sizeof(ip->ip_src));
 609         IFA_UNLOCK(&ia6->ia_ifa);
 610         bcopy(in4, &ip->ip_dst, sizeof(ip->ip_dst));
 611         ip->ip_p = IPPROTO_IPV6;
 612         ip->ip_ttl = ip_stf_ttl;
 613         ip->ip_len = m->m_pkthdr.len;   /*host order*/
 614         if (ifp->if_flags & IFF_LINK1) {
 615                 ip_ecn_ingress(ECN_NORMAL, &ip->ip_tos, &tos);
 616         } else {
 617                 ip_ecn_ingress(ECN_NOCARE, &ip->ip_tos, &tos);
 618         }
 619
 620         lck_mtx_lock(&sc->sc_ro_mtx);
 621         dst4 = (struct sockaddr_in *)(void *)&sc->sc_ro.ro_dst;
 622         if (ROUTE_UNUSABLE(&sc->sc_ro) || dst4->sin_family != AF_INET ||
 623             bcmp(&dst4->sin_addr, &ip->ip_dst, sizeof(ip->ip_dst)) != 0) {
 624                 ROUTE_RELEASE(&sc->sc_ro);
 625                 /* cache route doesn't match: always the case during the first use */
 626                 dst4->sin_family = AF_INET;
 627                 dst4->sin_len = sizeof(struct sockaddr_in);
 628                 bcopy(&ip->ip_dst, &dst4->sin_addr, sizeof(dst4->sin_addr));
 629         }
 630
 631         result = ip_output(m, NULL, &sc->sc_ro, IP_OUTARGS, NULL, &ipoa);
 632         lck_mtx_unlock(&sc->sc_ro_mtx);
 633
 634         /* Assumption: ip_output will free mbuf on errors */
 635         /* All the output processing is done here, don't let stf_output be called */
 636         if (result == 0) {
 637                 result = EJUSTRETURN;
 638         }
 639         *m0 = NULL;
 640         IFA_REMREF(&ia6->ia_ifa);
 641         return result;
 642 }
 643 static errno_t
 644 stf_output(
 645         __unused ifnet_t        ifp,
 646         __unused mbuf_t m)
 647 {
 648         /* All processing is done in stf_pre_output
 649          * this shouldn't be called as the pre_output returns "EJUSTRETURN"
 650          */
 651         return 0;
 652 }
 653
 654 static int
 655 stf_checkaddr4(
 656         struct stf_softc *sc,
 657         const struct in_addr *in,
 658         struct ifnet *inifp)    /* incoming interface */
 659 {
 660         struct in_ifaddr *ia4;
 661
 662         /*
 663          * reject packets with the following address:
 664          * 224.0.0.0/4 0.0.0.0/8 127.0.0.0/8 255.0.0.0/8
 665          */
 666         if (IN_MULTICAST(ntohl(in->s_addr))) {
 667                 return -1;
 668         }
 669         switch ((ntohl(in->s_addr) & 0xff000000) >> 24) {
 670         case 0: case 127: case 255:
 671                 return -1;
 672         }
 673
 674         /*
 675          * reject packets with broadcast
 676          */
 677         lck_rw_lock_shared(in_ifaddr_rwlock);
 678         for (ia4 = TAILQ_FIRST(&in_ifaddrhead);
 679             ia4;
 680             ia4 = TAILQ_NEXT(ia4, ia_link)) {
 681                 IFA_LOCK(&ia4->ia_ifa);
 682                 if ((ia4->ia_ifa.ifa_ifp->if_flags & IFF_BROADCAST) == 0) {
 683                         IFA_UNLOCK(&ia4->ia_ifa);
 684                         continue;
 685                 }
 686                 if (in->s_addr == ia4->ia_broadaddr.sin_addr.s_addr) {
 687                         IFA_UNLOCK(&ia4->ia_ifa);
 688                         lck_rw_done(in_ifaddr_rwlock);
 689                         return -1;
 690                 }
 691                 IFA_UNLOCK(&ia4->ia_ifa);
 692         }
 693         lck_rw_done(in_ifaddr_rwlock);
 694
 695         /*
 696          * perform ingress filter
 697          */
 698         if (sc && (ifnet_flags(sc->sc_if) & IFF_LINK2) == 0 && inifp) {
 699                 struct sockaddr_in sin;
 700                 struct rtentry *rt;
 701
 702                 bzero(&sin, sizeof(sin));
 703                 sin.sin_family = AF_INET;
 704                 sin.sin_len = sizeof(struct sockaddr_in);
 705                 sin.sin_addr = *in;
 706                 rt = rtalloc1((struct sockaddr *)&sin, 0, 0);
 707                 if (rt != NULL) {
 708                         RT_LOCK(rt);
 709                 }
 710                 if (rt == NULL || rt->rt_ifp != inifp) {
 711 #if 1
 712                         log(LOG_WARNING, "%s: packet from 0x%x dropped "
 713                             "due to ingress filter\n", if_name(sc->sc_if),
 714                             (u_int32_t)ntohl(sin.sin_addr.s_addr));
 715 #endif
 716                         if (rt != NULL) {
 717                                 RT_UNLOCK(rt);
 718                                 rtfree(rt);
 719                         }
 720                         return -1;
 721                 }
 722                 RT_UNLOCK(rt);
 723                 rtfree(rt);
 724         }
 725
 726         return 0;
 727 }
 728
 729 static int
 730 stf_checkaddr6(
 731         struct stf_softc *sc,
 732         struct in6_addr *in6,
 733         struct ifnet *inifp)    /* incoming interface */
 734 {
 735         /*
 736          * check 6to4 addresses
 737          */
 738         if (IN6_IS_ADDR_6TO4(in6)) {
 739                 return stf_checkaddr4(sc, GET_V4(in6), inifp);
 740         }
 741
 742         /*
 743          * reject anything that look suspicious.  the test is implemented
 744          * in ip6_input too, but we check here as well to
 745          * (1) reject bad packets earlier, and
 746          * (2) to be safe against future ip6_input change.
 747          */
 748         if (IN6_IS_ADDR_V4COMPAT(in6) || IN6_IS_ADDR_V4MAPPED(in6)) {
 749                 return -1;
 750         }
 751
 752         return 0;
 753 }
 754
 755 static void
 756 in_stf_input(
 757         struct mbuf *m,
 758         int off)
 759 {
 760         struct stf_softc *sc;
 761         struct ip *ip;
 762         struct ip6_hdr ip6;
 763         u_int8_t otos, itos;
 764         int proto;
 765         struct ifnet *ifp;
 766         struct ifnet_stat_increment_param       stats;
 767
 768         ip = mtod(m, struct ip *);
 769         proto = ip->ip_p;
 770
 771         if (proto != IPPROTO_IPV6) {
 772                 m_freem(m);
 773                 return;
 774         }
 775
 776         ip = mtod(m, struct ip *);
 777
 778         sc = (struct stf_softc *)encap_getarg(m);
 779
 780         if (sc == NULL || (ifnet_flags(sc->sc_if) & IFF_UP) == 0) {
 781                 m_freem(m);
 782                 return;
 783         }
 784
 785         ifp = sc->sc_if;
 786
 787         /*
 788          * perform sanity check against outer src/dst.
 789          * for source, perform ingress filter as well.
 790          */
 791         if (stf_checkaddr4(sc, &ip->ip_dst, NULL) < 0 ||
 792             stf_checkaddr4(sc, &ip->ip_src, m->m_pkthdr.rcvif) < 0) {
 793                 m_freem(m);
 794                 return;
 795         }
 796
 797         otos = ip->ip_tos;
 798         mbuf_copydata(m, off, sizeof(ip6), &ip6);
 799
 800         /*
 801          * perform sanity check against inner src/dst.
 802          * for source, perform ingress filter as well.
 803          */
 804         if (stf_checkaddr6(sc, &ip6.ip6_dst, NULL) < 0 ||
 805             stf_checkaddr6(sc, &ip6.ip6_src, m->m_pkthdr.rcvif) < 0) {
 806                 m_freem(m);
 807                 return;
 808         }
 809
 810         itos = (ntohl(ip6.ip6_flow) >> 20) & 0xff;
 811         if ((ifnet_flags(ifp) & IFF_LINK1) != 0) {
 812                 ip_ecn_egress(ECN_NORMAL, &otos, &itos);
 813         } else {
 814                 ip_ecn_egress(ECN_NOCARE, &otos, &itos);
 815         }
 816         ip6.ip6_flow &= ~htonl(0xff << 20);
 817         ip6.ip6_flow |= htonl((u_int32_t)itos << 20);
 818
 819         m->m_pkthdr.rcvif = ifp;
 820         mbuf_pkthdr_setheader(m, mbuf_data(m));
 821         mbuf_adj(m, off);
 822
 823         if (ifp->if_bpf) {
 824                 /* We need to prepend the address family as a four byte field. */
 825                 u_int32_t af = AF_INET6;
 826                 bpf_tap_in(ifp, 0, m, &af, sizeof(af));
 827         }
 828
 829         /*
 830          * Put the packet to the network layer input queue according to the
 831          * specified address family.
 832          * See net/if_gif.c for possible issues with packet processing
 833          * reorder due to extra queueing.
 834          */
 835         bzero(&stats, sizeof(stats));
 836         stats.packets_in = 1;
 837         stats.bytes_in = mbuf_pkthdr_len(m);
 838         mbuf_pkthdr_setrcvif(m, ifp);
 839         ifnet_input(ifp, m, &stats);
 840
 841         return;
 842 }
 843
 844 static void
 845 stf_rtrequest(
 846         __unused int cmd,
 847         struct rtentry *rt,
 848         __unused struct sockaddr *sa)
 849 {
 850         if (rt != NULL) {
 851                 RT_LOCK_ASSERT_HELD(rt);
 852                 rt->rt_rmx.rmx_mtu = IPV6_MMTU;
 853         }
 854 }
 855
 856 static errno_t
 857 stf_ioctl(
 858         ifnet_t         ifp,
 859         u_long          cmd,
 860         void            *data)
 861 {
 862         struct ifaddr *ifa;
 863         struct ifreq *ifr;
 864         struct sockaddr_in6 *sin6;
 865         int error;
 866
 867         error = 0;
 868         switch (cmd) {
 869         case SIOCSIFADDR:
 870                 ifa = (struct ifaddr *)data;
 871                 if (ifa == NULL) {
 872                         error = EAFNOSUPPORT;
 873                         break;
 874                 }
 875                 IFA_LOCK(ifa);
 876                 if (ifa->ifa_addr->sa_family != AF_INET6) {
 877                         IFA_UNLOCK(ifa);
 878                         error = EAFNOSUPPORT;
 879                         break;
 880                 }
 881                 sin6 = (struct sockaddr_in6 *)(void *)ifa->ifa_addr;
 882                 if (IN6_IS_ADDR_6TO4(&sin6->sin6_addr)) {
 883                         if (!(ifnet_flags( ifp ) & IFF_UP)) {
 884                                 /* do this only if the interface is not already up */
 885                                 ifa->ifa_rtrequest = stf_rtrequest;
 886                                 IFA_UNLOCK(ifa);
 887                                 ifnet_set_flags(ifp, IFF_UP, IFF_UP);
 888                         } else {
 889                                 IFA_UNLOCK(ifa);
 890                         }
 891                 } else {
 892                         IFA_UNLOCK(ifa);
 893                         error = EINVAL;
 894                 }
 895                 IFA_LOCK_ASSERT_NOTHELD(ifa);
 896                 break;
 897
 898         case SIOCADDMULTI:
 899         case SIOCDELMULTI:
 900                 ifr = (struct ifreq *)data;
 901                 if (ifr && ifr->ifr_addr.sa_family == AF_INET6) {
 902                         ;
 903                 } else {
 904                         error = EAFNOSUPPORT;
 905                 }
 906                 break;
 907
 908         default:
 909                 error = EOPNOTSUPP;
 910                 break;
 911         }
 912
 913         return error;
 914 }