| 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 | #include <net/sockaddr_utils.h> |
| 150 | |
| 151 | #define GET_V4(x) ((const struct in_addr *)(const void *)(&(x)->s6_addr16[1])) |
| 152 | |
| 153 | static LCK_GRP_DECLARE(stf_mtx_grp, "stf" ); |
| 154 | |
| 155 | struct stf_softc { |
| 156 | ifnet_t sc_if; /* common area */ |
| 157 | u_int32_t sc_protocol_family; /* dlil protocol attached */ |
| 158 | union { |
| 159 | struct route __sc_ro4; |
| 160 | struct route_in6 __sc_ro6; /* just for safety */ |
| 161 | } __sc_ro46; |
| 162 | #define sc_ro __sc_ro46.__sc_ro4 |
| 163 | decl_lck_mtx_data(, sc_ro_mtx); |
| 164 | const struct encaptab *encap_cookie; |
| 165 | bpf_tap_mode tap_mode; |
| 166 | bpf_packet_func tap_callback; |
| 167 | }; |
| 168 | |
| 169 | void stfattach(void); |
| 170 | |
| 171 | static int ip_stf_ttl = 40; |
| 172 | |
| 173 | static void in_stf_input(struct mbuf *, int); |
| 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 | /* |
| 199 | * gif_input is the input handler for IP and IPv6 attached to gif |
| 200 | */ |
| 201 | static errno_t |
| 202 | stf_media_input( |
| 203 | __unused ifnet_t ifp, |
| 204 | protocol_family_t protocol_family, |
| 205 | mbuf_t m, |
| 206 | __unused char *) |
| 207 | { |
| 208 | if (proto_input(protocol: protocol_family, packet: m) != 0) { |
| 209 | m_freem(m); |
| 210 | } |
| 211 | |
| 212 | return 0; |
| 213 | } |
| 214 | |
| 215 | |
| 216 | |
| 217 | static errno_t |
| 218 | stf_add_proto( |
| 219 | ifnet_t ifp, |
| 220 | protocol_family_t protocol_family, |
| 221 | __unused const struct ifnet_demux_desc *demux_array, |
| 222 | __unused u_int32_t demux_count) |
| 223 | { |
| 224 | /* Only one protocol may be attached at a time */ |
| 225 | struct stf_softc* stf = ifnet_softc(interface: ifp); |
| 226 | if (stf->sc_protocol_family == 0) { |
| 227 | stf->sc_protocol_family = protocol_family; |
| 228 | } else { |
| 229 | printf("stf_add_proto: stf already has a proto\n" ); |
| 230 | return EBUSY; |
| 231 | } |
| 232 | |
| 233 | return 0; |
| 234 | } |
| 235 | |
| 236 | static errno_t |
| 237 | stf_del_proto( |
| 238 | ifnet_t ifp, |
| 239 | protocol_family_t protocol_family) |
| 240 | { |
| 241 | if (((struct stf_softc*)ifnet_softc(interface: ifp))->sc_protocol_family == protocol_family) { |
| 242 | ((struct stf_softc*)ifnet_softc(interface: ifp))->sc_protocol_family = 0; |
| 243 | } |
| 244 | |
| 245 | return 0; |
| 246 | } |
| 247 | |
| 248 | static errno_t |
| 249 | stf_attach_inet6( |
| 250 | ifnet_t ifp, |
| 251 | protocol_family_t protocol_family) |
| 252 | { |
| 253 | struct ifnet_attach_proto_param reg; |
| 254 | errno_t stat; |
| 255 | |
| 256 | if (protocol_family != PF_INET6) { |
| 257 | return EPROTONOSUPPORT; |
| 258 | } |
| 259 | |
| 260 | bzero(s: ®, n: sizeof(reg)); |
| 261 | reg.input = stf_media_input; |
| 262 | reg.pre_output = stf_pre_output; |
| 263 | |
| 264 | stat = ifnet_attach_protocol(interface: ifp, protocol_family, proto_details: ®); |
| 265 | if (stat && stat != EEXIST) { |
| 266 | printf("stf_attach_proto_family can't attach interface fam=%d\n" , |
| 267 | protocol_family); |
| 268 | } |
| 269 | |
| 270 | return stat; |
| 271 | } |
| 272 | |
| 273 | static errno_t |
| 274 | stf_demux( |
| 275 | ifnet_t ifp, |
| 276 | __unused mbuf_t m, |
| 277 | __unused char *frame_ptr, |
| 278 | protocol_family_t *protocol_family) |
| 279 | { |
| 280 | struct stf_softc* stf = ifnet_softc(interface: ifp); |
| 281 | *protocol_family = stf->sc_protocol_family; |
| 282 | return 0; |
| 283 | } |
| 284 | |
| 285 | static errno_t |
| 286 | stf_set_bpf_tap( |
| 287 | ifnet_t ifp, |
| 288 | bpf_tap_mode mode, |
| 289 | bpf_packet_func callback) |
| 290 | { |
| 291 | struct stf_softc *sc = ifnet_softc(interface: ifp); |
| 292 | |
| 293 | sc->tap_mode = mode; |
| 294 | sc->tap_callback = callback; |
| 295 | |
| 296 | return 0; |
| 297 | } |
| 298 | |
| 299 | void |
| 300 | stfattach(void) |
| 301 | { |
| 302 | struct stf_softc *sc; |
| 303 | int error; |
| 304 | const struct encaptab *p; |
| 305 | struct ifnet_init_eparams stf_init; |
| 306 | |
| 307 | error = proto_register_plumber(PF_INET6, APPLE_IF_FAM_STF, |
| 308 | plumb: stf_attach_inet6, NULL); |
| 309 | if (error != 0) { |
| 310 | printf("proto_register_plumber failed for AF_INET6 error=%d\n" , error); |
| 311 | } |
| 312 | |
| 313 | sc = kalloc_type(struct stf_softc, Z_WAITOK_ZERO_NOFAIL); |
| 314 | lck_mtx_init(lck: &sc->sc_ro_mtx, grp: &stf_mtx_grp, LCK_ATTR_NULL); |
| 315 | |
| 316 | p = encap_attach_func(AF_INET, IPPROTO_IPV6, stf_encapcheck, |
| 317 | &in_stf_protosw, sc); |
| 318 | if (p == NULL) { |
| 319 | printf("sftattach encap_attach_func failed\n" ); |
| 320 | goto free_sc; |
| 321 | } |
| 322 | sc->encap_cookie = p; |
| 323 | |
| 324 | bzero(s: &stf_init, n: sizeof(stf_init)); |
| 325 | stf_init.ver = IFNET_INIT_CURRENT_VERSION; |
| 326 | stf_init.len = sizeof(stf_init); |
| 327 | stf_init.flags = IFNET_INIT_LEGACY; |
| 328 | stf_init.name = "stf" ; |
| 329 | stf_init.unit = 0; |
| 330 | stf_init.type = IFT_STF; |
| 331 | stf_init.family = IFNET_FAMILY_STF; |
| 332 | stf_init.output = stf_output; |
| 333 | stf_init.demux = stf_demux; |
| 334 | stf_init.add_proto = stf_add_proto; |
| 335 | stf_init.del_proto = stf_del_proto; |
| 336 | stf_init.softc = sc; |
| 337 | stf_init.ioctl = stf_ioctl; |
| 338 | stf_init.set_bpf_tap = stf_set_bpf_tap; |
| 339 | |
| 340 | error = ifnet_allocate_extended(init: &stf_init, interface: &sc->sc_if); |
| 341 | if (error != 0) { |
| 342 | printf("stfattach, ifnet_allocate failed - %d\n" , error); |
| 343 | encap_detach(sc->encap_cookie); |
| 344 | goto free_sc; |
| 345 | } |
| 346 | ifnet_set_mtu(interface: sc->sc_if, IPV6_MMTU); |
| 347 | ifnet_set_flags(interface: sc->sc_if, new_flags: 0, mask: 0xffff); /* clear all flags */ |
| 348 | #if 0 |
| 349 | /* turn off ingress filter */ |
| 350 | ifnet_set_flags(sc->sc_if, IFF_LINK2, IFF_LINK2); |
| 351 | #endif |
| 352 | |
| 353 | error = ifnet_attach(interface: sc->sc_if, NULL); |
| 354 | if (error != 0) { |
| 355 | printf("stfattach: ifnet_attach returned error=%d\n" , error); |
| 356 | encap_detach(sc->encap_cookie); |
| 357 | ifnet_release(interface: sc->sc_if); |
| 358 | goto free_sc; |
| 359 | } |
| 360 | |
| 361 | bpfattach(interface: sc->sc_if, DLT_NULL, header_length: sizeof(u_int)); |
| 362 | |
| 363 | return; |
| 364 | |
| 365 | free_sc: |
| 366 | lck_mtx_destroy(lck: &sc->sc_ro_mtx, grp: &stf_mtx_grp); |
| 367 | kfree_type(struct stf_softc, sc); |
| 368 | } |
| 369 | |
| 370 | static int |
| 371 | stf_encapcheck( |
| 372 | const struct mbuf *m, |
| 373 | __unused int off, |
| 374 | int proto, |
| 375 | void *arg) |
| 376 | { |
| 377 | struct ip ip; |
| 378 | struct in6_ifaddr *ia6; |
| 379 | struct stf_softc *sc; |
| 380 | struct in_addr a, b; |
| 381 | |
| 382 | sc = (struct stf_softc *)arg; |
| 383 | if (sc == NULL) { |
| 384 | return 0; |
| 385 | } |
| 386 | |
| 387 | if ((ifnet_flags(interface: sc->sc_if) & IFF_UP) == 0) { |
| 388 | return 0; |
| 389 | } |
| 390 | |
| 391 | /* IFF_LINK0 means "no decapsulation" */ |
| 392 | if ((ifnet_flags(interface: sc->sc_if) & IFF_LINK0) != 0) { |
| 393 | return 0; |
| 394 | } |
| 395 | |
| 396 | if (proto != IPPROTO_IPV6) { |
| 397 | return 0; |
| 398 | } |
| 399 | |
| 400 | mbuf_copydata(mbuf: (struct mbuf *)(size_t)m, offset: 0, length: sizeof(ip), out_data: &ip); |
| 401 | |
| 402 | if (ip.ip_v != 4) { |
| 403 | return 0; |
| 404 | } |
| 405 | |
| 406 | ia6 = stf_getsrcifa6(sc->sc_if); |
| 407 | if (ia6 == NULL) { |
| 408 | return 0; |
| 409 | } |
| 410 | |
| 411 | /* |
| 412 | * check if IPv4 dst matches the IPv4 address derived from the |
| 413 | * local 6to4 address. |
| 414 | * success on: dst = 10.1.1.1, ia6->ia_addr = 2002:0a01:0101:... |
| 415 | */ |
| 416 | IFA_LOCK(&ia6->ia_ifa); |
| 417 | if (bcmp(GET_V4(&ia6->ia_addr.sin6_addr), s2: &ip.ip_dst, |
| 418 | n: sizeof(ip.ip_dst)) != 0) { |
| 419 | IFA_UNLOCK(&ia6->ia_ifa); |
| 420 | ifa_remref(ifa: &ia6->ia_ifa); |
| 421 | return 0; |
| 422 | } |
| 423 | /* |
| 424 | * check if IPv4 src matches the IPv4 address derived from the |
| 425 | * local 6to4 address masked by prefixmask. |
| 426 | * success on: src = 10.1.1.1, ia6->ia_addr = 2002:0a00:.../24 |
| 427 | * fail on: src = 10.1.1.1, ia6->ia_addr = 2002:0b00:.../24 |
| 428 | */ |
| 429 | bzero(s: &a, n: sizeof(a)); |
| 430 | a.s_addr = GET_V4(&ia6->ia_addr.sin6_addr)->s_addr; |
| 431 | a.s_addr &= GET_V4(&ia6->ia_prefixmask.sin6_addr)->s_addr; |
| 432 | b = ip.ip_src; |
| 433 | b.s_addr &= GET_V4(&ia6->ia_prefixmask.sin6_addr)->s_addr; |
| 434 | if (a.s_addr != b.s_addr) { |
| 435 | IFA_UNLOCK(&ia6->ia_ifa); |
| 436 | ifa_remref(ifa: &ia6->ia_ifa); |
| 437 | return 0; |
| 438 | } |
| 439 | /* stf interface makes single side match only */ |
| 440 | IFA_UNLOCK(&ia6->ia_ifa); |
| 441 | ifa_remref(ifa: &ia6->ia_ifa); |
| 442 | return 32; |
| 443 | } |
| 444 | |
| 445 | static struct in6_ifaddr * |
| 446 | stf_getsrcifa6(struct ifnet *ifp) |
| 447 | { |
| 448 | struct ifaddr *ia; |
| 449 | struct in_ifaddr *ia4; |
| 450 | struct sockaddr_in6 *sin6; |
| 451 | struct in_addr in; |
| 452 | |
| 453 | ifnet_lock_shared(ifp); |
| 454 | for (ia = ifp->if_addrlist.tqh_first; ia; ia = ia->ifa_list.tqe_next) { |
| 455 | IFA_LOCK(ia); |
| 456 | if (ia->ifa_addr == NULL) { |
| 457 | IFA_UNLOCK(ia); |
| 458 | continue; |
| 459 | } |
| 460 | if (ia->ifa_addr->sa_family != AF_INET6) { |
| 461 | IFA_UNLOCK(ia); |
| 462 | continue; |
| 463 | } |
| 464 | sin6 = SIN6(ia->ifa_addr); |
| 465 | if (!IN6_IS_ADDR_6TO4(&sin6->sin6_addr)) { |
| 466 | IFA_UNLOCK(ia); |
| 467 | continue; |
| 468 | } |
| 469 | bcopy(GET_V4(&sin6->sin6_addr), dst: &in, n: sizeof(in)); |
| 470 | IFA_UNLOCK(ia); |
| 471 | lck_rw_lock_shared(lck: &in_ifaddr_rwlock); |
| 472 | for (ia4 = TAILQ_FIRST(&in_ifaddrhead); |
| 473 | ia4; |
| 474 | ia4 = TAILQ_NEXT(ia4, ia_link)) { |
| 475 | IFA_LOCK(&ia4->ia_ifa); |
| 476 | if (ia4->ia_addr.sin_addr.s_addr == in.s_addr) { |
| 477 | IFA_UNLOCK(&ia4->ia_ifa); |
| 478 | break; |
| 479 | } |
| 480 | IFA_UNLOCK(&ia4->ia_ifa); |
| 481 | } |
| 482 | lck_rw_done(lck: &in_ifaddr_rwlock); |
| 483 | if (ia4 == NULL) { |
| 484 | continue; |
| 485 | } |
| 486 | |
| 487 | ifa_addref(ifa: ia); /* for caller */ |
| 488 | ifnet_lock_done(ifp); |
| 489 | return (struct in6_ifaddr *)ia; |
| 490 | } |
| 491 | ifnet_lock_done(ifp); |
| 492 | |
| 493 | return NULL; |
| 494 | } |
| 495 | |
| 496 | int |
| 497 | stf_pre_output( |
| 498 | struct ifnet *ifp, |
| 499 | __unused protocol_family_t protocol_family, |
| 500 | struct mbuf **m0, |
| 501 | const struct sockaddr *dst, |
| 502 | __unused void *route, |
| 503 | __unused char *desk_linkaddr, |
| 504 | __unused char *frame_type) |
| 505 | { |
| 506 | struct mbuf *m = *m0; |
| 507 | struct stf_softc *sc; |
| 508 | const struct sockaddr_in6 *dst6; |
| 509 | const struct in_addr *in4; |
| 510 | u_int8_t tos; |
| 511 | struct ip *ip; |
| 512 | struct ip6_hdr *ip6; |
| 513 | struct in6_ifaddr *ia6; |
| 514 | struct sockaddr_in *dst4; |
| 515 | struct ip_out_args ipoa; |
| 516 | errno_t result = 0; |
| 517 | |
| 518 | bzero(s: &ipoa, n: sizeof(ipoa)); |
| 519 | ipoa.ipoa_boundif = IFSCOPE_NONE; |
| 520 | ipoa.ipoa_flags = IPOAF_SELECT_SRCIF; |
| 521 | ipoa.ipoa_sotc = SO_TC_UNSPEC; |
| 522 | ipoa.ipoa_netsvctype = _NET_SERVICE_TYPE_UNSPEC; |
| 523 | |
| 524 | sc = ifnet_softc(interface: ifp); |
| 525 | dst6 = SIN6(dst); |
| 526 | |
| 527 | /* just in case */ |
| 528 | if ((ifnet_flags(interface: ifp) & IFF_UP) == 0) { |
| 529 | printf("stf: IFF_DOWN\n" ); |
| 530 | return ENETDOWN; |
| 531 | } |
| 532 | |
| 533 | /* |
| 534 | * If we don't have an ip4 address that match my inner ip6 address, |
| 535 | * we shouldn't generate output. Without this check, we'll end up |
| 536 | * using wrong IPv4 source. |
| 537 | */ |
| 538 | ia6 = stf_getsrcifa6(ifp); |
| 539 | if (ia6 == NULL) { |
| 540 | return ENETDOWN; |
| 541 | } |
| 542 | |
| 543 | if (mbuf_len(mbuf: m) < sizeof(*ip6)) { |
| 544 | m = m_pullup(m, sizeof(*ip6)); |
| 545 | if (!m) { |
| 546 | *m0 = NULL; /* makes sure this won't be double freed */ |
| 547 | ifa_remref(ifa: &ia6->ia_ifa); |
| 548 | return ENOBUFS; |
| 549 | } |
| 550 | *m0 = m; |
| 551 | } |
| 552 | ip6 = mtod(m, struct ip6_hdr *); |
| 553 | tos = (ntohl(ip6->ip6_flow) >> 20) & 0xff; |
| 554 | |
| 555 | /* |
| 556 | * Pickup the right outer dst addr from the list of candidates. |
| 557 | * ip6_dst has priority as it may be able to give us shorter IPv4 hops. |
| 558 | */ |
| 559 | if (IN6_IS_ADDR_6TO4(&ip6->ip6_dst)) { |
| 560 | in4 = GET_V4(&ip6->ip6_dst); |
| 561 | } else if (IN6_IS_ADDR_6TO4(&dst6->sin6_addr)) { |
| 562 | in4 = GET_V4(&dst6->sin6_addr); |
| 563 | } else { |
| 564 | ifa_remref(ifa: &ia6->ia_ifa); |
| 565 | return ENETUNREACH; |
| 566 | } |
| 567 | |
| 568 | if (ifp->if_bpf) { |
| 569 | /* We need to prepend the address family as a four byte field. */ |
| 570 | u_int32_t af = AF_INET6; |
| 571 | |
| 572 | bpf_tap_out(interface: ifp, dlt: 0, packet: m, header: &af, header_len: sizeof(af)); |
| 573 | } |
| 574 | |
| 575 | M_PREPEND(m, sizeof(struct ip), M_DONTWAIT, 1); |
| 576 | if (m && mbuf_len(mbuf: m) < sizeof(struct ip)) { |
| 577 | m = m_pullup(m, sizeof(struct ip)); |
| 578 | } |
| 579 | if (m == NULL) { |
| 580 | *m0 = NULL; |
| 581 | ifa_remref(ifa: &ia6->ia_ifa); |
| 582 | return ENOBUFS; |
| 583 | } |
| 584 | |
| 585 | *m0 = m; |
| 586 | ip = mtod(m, struct ip *); |
| 587 | |
| 588 | bzero(s: ip, n: sizeof(*ip)); |
| 589 | |
| 590 | IFA_LOCK_SPIN(&ia6->ia_ifa); |
| 591 | bcopy(GET_V4(&(SIN6(&ia6->ia_addr))->sin6_addr), |
| 592 | dst: &ip->ip_src, n: sizeof(ip->ip_src)); |
| 593 | IFA_UNLOCK(&ia6->ia_ifa); |
| 594 | bcopy(src: in4, dst: &ip->ip_dst, n: sizeof(ip->ip_dst)); |
| 595 | ip->ip_p = IPPROTO_IPV6; |
| 596 | ip->ip_ttl = ip_stf_ttl; |
| 597 | ip->ip_len = m->m_pkthdr.len; /*host order*/ |
| 598 | if (ifp->if_flags & IFF_LINK1) { |
| 599 | ip_ecn_ingress(ECN_NORMAL, &ip->ip_tos, &tos); |
| 600 | } else { |
| 601 | ip_ecn_ingress(ECN_NOCARE, &ip->ip_tos, &tos); |
| 602 | } |
| 603 | |
| 604 | lck_mtx_lock(lck: &sc->sc_ro_mtx); |
| 605 | dst4 = SIN(&sc->sc_ro.ro_dst); |
| 606 | if (ROUTE_UNUSABLE(&sc->sc_ro) || dst4->sin_family != AF_INET || |
| 607 | bcmp(s1: &dst4->sin_addr, s2: &ip->ip_dst, n: sizeof(ip->ip_dst)) != 0) { |
| 608 | ROUTE_RELEASE(&sc->sc_ro); |
| 609 | /* cache route doesn't match: always the case during the first use */ |
| 610 | dst4->sin_family = AF_INET; |
| 611 | dst4->sin_len = sizeof(struct sockaddr_in); |
| 612 | bcopy(src: &ip->ip_dst, dst: &dst4->sin_addr, n: sizeof(dst4->sin_addr)); |
| 613 | } |
| 614 | |
| 615 | result = ip_output(m, NULL, &sc->sc_ro, IP_OUTARGS, NULL, &ipoa); |
| 616 | lck_mtx_unlock(lck: &sc->sc_ro_mtx); |
| 617 | |
| 618 | /* Assumption: ip_output will free mbuf on errors */ |
| 619 | /* All the output processing is done here, don't let stf_output be called */ |
| 620 | if (result == 0) { |
| 621 | result = EJUSTRETURN; |
| 622 | } |
| 623 | *m0 = NULL; |
| 624 | ifa_remref(ifa: &ia6->ia_ifa); |
| 625 | return result; |
| 626 | } |
| 627 | static errno_t |
| 628 | stf_output( |
| 629 | __unused ifnet_t ifp, |
| 630 | __unused mbuf_t m) |
| 631 | { |
| 632 | /* All processing is done in stf_pre_output |
| 633 | * this shouldn't be called as the pre_output returns "EJUSTRETURN" |
| 634 | */ |
| 635 | return 0; |
| 636 | } |
| 637 | |
| 638 | static int |
| 639 | stf_checkaddr4( |
| 640 | struct stf_softc *sc, |
| 641 | const struct in_addr *in, |
| 642 | struct ifnet *inifp) /* incoming interface */ |
| 643 | { |
| 644 | struct in_ifaddr *ia4; |
| 645 | |
| 646 | /* |
| 647 | * reject packets with the following address: |
| 648 | * 224.0.0.0/4 0.0.0.0/8 127.0.0.0/8 255.0.0.0/8 |
| 649 | */ |
| 650 | if (IN_MULTICAST(ntohl(in->s_addr))) { |
| 651 | return -1; |
| 652 | } |
| 653 | switch ((ntohl(in->s_addr) & 0xff000000) >> 24) { |
| 654 | case 0: case 127: case 255: |
| 655 | return -1; |
| 656 | } |
| 657 | |
| 658 | /* |
| 659 | * reject packets with broadcast |
| 660 | */ |
| 661 | lck_rw_lock_shared(lck: &in_ifaddr_rwlock); |
| 662 | for (ia4 = TAILQ_FIRST(&in_ifaddrhead); |
| 663 | ia4; |
| 664 | ia4 = TAILQ_NEXT(ia4, ia_link)) { |
| 665 | IFA_LOCK(&ia4->ia_ifa); |
| 666 | if ((ia4->ia_ifa.ifa_ifp->if_flags & IFF_BROADCAST) == 0) { |
| 667 | IFA_UNLOCK(&ia4->ia_ifa); |
| 668 | continue; |
| 669 | } |
| 670 | if (in->s_addr == ia4->ia_broadaddr.sin_addr.s_addr) { |
| 671 | IFA_UNLOCK(&ia4->ia_ifa); |
| 672 | lck_rw_done(lck: &in_ifaddr_rwlock); |
| 673 | return -1; |
| 674 | } |
| 675 | IFA_UNLOCK(&ia4->ia_ifa); |
| 676 | } |
| 677 | lck_rw_done(lck: &in_ifaddr_rwlock); |
| 678 | |
| 679 | /* |
| 680 | * perform ingress filter |
| 681 | */ |
| 682 | if (sc && (ifnet_flags(interface: sc->sc_if) & IFF_LINK2) == 0 && inifp) { |
| 683 | struct sockaddr_in sin; |
| 684 | struct rtentry *rt; |
| 685 | |
| 686 | SOCKADDR_ZERO(&sin, sizeof(sin)); |
| 687 | sin.sin_family = AF_INET; |
| 688 | sin.sin_len = sizeof(struct sockaddr_in); |
| 689 | sin.sin_addr = *in; |
| 690 | rt = rtalloc1(SA(&sin), 0, 0); |
| 691 | if (rt != NULL) { |
| 692 | RT_LOCK(rt); |
| 693 | } |
| 694 | if (rt == NULL || rt->rt_ifp != inifp) { |
| 695 | #if 1 |
| 696 | log(LOG_WARNING, "%s: packet from 0x%x dropped " |
| 697 | "due to ingress filter\n" , if_name(sc->sc_if), |
| 698 | (u_int32_t)ntohl(sin.sin_addr.s_addr)); |
| 699 | #endif |
| 700 | if (rt != NULL) { |
| 701 | RT_UNLOCK(rt); |
| 702 | rtfree(rt); |
| 703 | } |
| 704 | return -1; |
| 705 | } |
| 706 | RT_UNLOCK(rt); |
| 707 | rtfree(rt); |
| 708 | } |
| 709 | |
| 710 | return 0; |
| 711 | } |
| 712 | |
| 713 | static int |
| 714 | stf_checkaddr6( |
| 715 | struct stf_softc *sc, |
| 716 | struct in6_addr *in6, |
| 717 | struct ifnet *inifp) /* incoming interface */ |
| 718 | { |
| 719 | /* |
| 720 | * check 6to4 addresses |
| 721 | */ |
| 722 | if (IN6_IS_ADDR_6TO4(in6)) { |
| 723 | return stf_checkaddr4(sc, GET_V4(in6), inifp); |
| 724 | } |
| 725 | |
| 726 | /* |
| 727 | * reject anything that look suspicious. the test is implemented |
| 728 | * in ip6_input too, but we check here as well to |
| 729 | * (1) reject bad packets earlier, and |
| 730 | * (2) to be safe against future ip6_input change. |
| 731 | */ |
| 732 | if (IN6_IS_ADDR_V4COMPAT(in6) || IN6_IS_ADDR_V4MAPPED(in6)) { |
| 733 | return -1; |
| 734 | } |
| 735 | |
| 736 | return 0; |
| 737 | } |
| 738 | |
| 739 | static void |
| 740 | in_stf_input( |
| 741 | struct mbuf *m, |
| 742 | int off) |
| 743 | { |
| 744 | struct stf_softc *sc; |
| 745 | struct ip *ip; |
| 746 | struct ip6_hdr ip6; |
| 747 | u_int8_t otos, itos; |
| 748 | int proto; |
| 749 | struct ifnet *ifp; |
| 750 | struct ifnet_stat_increment_param stats; |
| 751 | |
| 752 | ip = mtod(m, struct ip *); |
| 753 | proto = ip->ip_p; |
| 754 | |
| 755 | if (proto != IPPROTO_IPV6) { |
| 756 | m_freem(m); |
| 757 | return; |
| 758 | } |
| 759 | |
| 760 | ip = mtod(m, struct ip *); |
| 761 | |
| 762 | sc = (struct stf_softc *)encap_getarg(m); |
| 763 | |
| 764 | if (sc == NULL || (ifnet_flags(interface: sc->sc_if) & IFF_UP) == 0) { |
| 765 | m_freem(m); |
| 766 | return; |
| 767 | } |
| 768 | |
| 769 | ifp = sc->sc_if; |
| 770 | |
| 771 | /* |
| 772 | * perform sanity check against outer src/dst. |
| 773 | * for source, perform ingress filter as well. |
| 774 | */ |
| 775 | if (stf_checkaddr4(sc, in: &ip->ip_dst, NULL) < 0 || |
| 776 | stf_checkaddr4(sc, in: &ip->ip_src, inifp: m->m_pkthdr.rcvif) < 0) { |
| 777 | m_freem(m); |
| 778 | return; |
| 779 | } |
| 780 | |
| 781 | otos = ip->ip_tos; |
| 782 | if (mbuf_copydata(mbuf: m, offset: off, length: sizeof(ip6), out_data: &ip6)) { |
| 783 | m_freem(m); |
| 784 | return; |
| 785 | } |
| 786 | |
| 787 | /* |
| 788 | * perform sanity check against inner src/dst. |
| 789 | * for source, perform ingress filter as well. |
| 790 | */ |
| 791 | if (stf_checkaddr6(sc, in6: &ip6.ip6_dst, NULL) < 0 || |
| 792 | stf_checkaddr6(sc, in6: &ip6.ip6_src, inifp: m->m_pkthdr.rcvif) < 0) { |
| 793 | m_freem(m); |
| 794 | return; |
| 795 | } |
| 796 | |
| 797 | itos = (ntohl(ip6.ip6_flow) >> 20) & 0xff; |
| 798 | if ((ifnet_flags(interface: ifp) & IFF_LINK1) != 0) { |
| 799 | ip_ecn_egress(ECN_NORMAL, &otos, &itos); |
| 800 | } else { |
| 801 | ip_ecn_egress(ECN_NOCARE, &otos, &itos); |
| 802 | } |
| 803 | ip6.ip6_flow &= ~htonl(0xff << 20); |
| 804 | ip6.ip6_flow |= htonl((u_int32_t)itos << 20); |
| 805 | |
| 806 | m->m_pkthdr.rcvif = ifp; |
| 807 | mbuf_pkthdr_setheader(mbuf: m, header: mbuf_data(mbuf: m)); |
| 808 | mbuf_adj(mbuf: m, len: off); |
| 809 | |
| 810 | if (ifp->if_bpf) { |
| 811 | /* We need to prepend the address family as a four byte field. */ |
| 812 | u_int32_t af = AF_INET6; |
| 813 | bpf_tap_in(interface: ifp, dlt: 0, packet: m, header: &af, header_len: sizeof(af)); |
| 814 | } |
| 815 | |
| 816 | /* |
| 817 | * Put the packet to the network layer input queue according to the |
| 818 | * specified address family. |
| 819 | * See net/if_gif.c for possible issues with packet processing |
| 820 | * reorder due to extra queueing. |
| 821 | */ |
| 822 | bzero(s: &stats, n: sizeof(stats)); |
| 823 | stats.packets_in = 1; |
| 824 | stats.bytes_in = mbuf_pkthdr_len(mbuf: m); |
| 825 | mbuf_pkthdr_setrcvif(mbuf: m, ifp); |
| 826 | ifnet_input(interface: ifp, first_packet: m, stats: &stats); |
| 827 | |
| 828 | return; |
| 829 | } |
| 830 | |
| 831 | static void |
| 832 | stf_rtrequest( |
| 833 | __unused int cmd, |
| 834 | struct rtentry *rt, |
| 835 | __unused struct sockaddr *sa) |
| 836 | { |
| 837 | if (rt != NULL) { |
| 838 | RT_LOCK_ASSERT_HELD(rt); |
| 839 | rt->rt_rmx.rmx_mtu = IPV6_MMTU; |
| 840 | } |
| 841 | } |
| 842 | |
| 843 | static errno_t |
| 844 | stf_ioctl( |
| 845 | ifnet_t ifp, |
| 846 | u_long cmd, |
| 847 | void *data) |
| 848 | { |
| 849 | struct ifaddr *ifa; |
| 850 | struct ifreq *ifr; |
| 851 | struct sockaddr_in6 *sin6; |
| 852 | int error; |
| 853 | |
| 854 | error = 0; |
| 855 | switch (cmd) { |
| 856 | case SIOCSIFADDR: |
| 857 | ifa = (struct ifaddr *)data; |
| 858 | if (ifa == NULL) { |
| 859 | error = EAFNOSUPPORT; |
| 860 | break; |
| 861 | } |
| 862 | IFA_LOCK(ifa); |
| 863 | if (ifa->ifa_addr->sa_family != AF_INET6) { |
| 864 | IFA_UNLOCK(ifa); |
| 865 | error = EAFNOSUPPORT; |
| 866 | break; |
| 867 | } |
| 868 | sin6 = SIN6(ifa->ifa_addr); |
| 869 | if (IN6_IS_ADDR_6TO4(&sin6->sin6_addr)) { |
| 870 | if (!(ifnet_flags( interface: ifp ) & IFF_UP)) { |
| 871 | /* do this only if the interface is not already up */ |
| 872 | ifa->ifa_rtrequest = stf_rtrequest; |
| 873 | IFA_UNLOCK(ifa); |
| 874 | ifnet_set_flags(interface: ifp, IFF_UP, IFF_UP); |
| 875 | } else { |
| 876 | IFA_UNLOCK(ifa); |
| 877 | } |
| 878 | } else { |
| 879 | IFA_UNLOCK(ifa); |
| 880 | error = EINVAL; |
| 881 | } |
| 882 | IFA_LOCK_ASSERT_NOTHELD(ifa); |
| 883 | break; |
| 884 | |
| 885 | case SIOCADDMULTI: |
| 886 | case SIOCDELMULTI: |
| 887 | ifr = (struct ifreq *)data; |
| 888 | if (ifr && ifr->ifr_addr.sa_family == AF_INET6) { |
| 889 | ; |
| 890 | } else { |
| 891 | error = EAFNOSUPPORT; |
| 892 | } |
| 893 | break; |
| 894 | |
| 895 | default: |
| 896 | error = EOPNOTSUPP; |
| 897 | break; |
| 898 | } |
| 899 | |
| 900 | return error; |
| 901 | } |
| 902 | |