1 /*- 2 * SPDX-License-Identifier: BSD-3-Clause 3 * 4 * Copyright (C) 1995, 1996, 1997, and 1998 WIDE Project. 5 * Copyright (c) 2018 Andrey V. Elsukov <ae@FreeBSD.org> 6 * All rights reserved. 7 * 8 * Redistribution and use in source and binary forms, with or without 9 * modification, are permitted provided that the following conditions 10 * are met: 11 * 1. Redistributions of source code must retain the above copyright 12 * notice, this list of conditions and the following disclaimer. 13 * 2. Redistributions in binary form must reproduce the above copyright 14 * notice, this list of conditions and the following disclaimer in the 15 * documentation and/or other materials provided with the distribution. 16 * 3. Neither the name of the project nor the names of its contributors 17 * may be used to endorse or promote products derived from this software 18 * without specific prior written permission. 19 * 20 * THIS SOFTWARE IS PROVIDED BY THE PROJECT AND CONTRIBUTORS ``AS IS'' AND 21 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE 22 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE 23 * ARE DISCLAIMED. IN NO EVENT SHALL THE PROJECT OR CONTRIBUTORS BE LIABLE 24 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL 25 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS 26 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) 27 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT 28 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY 29 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF 30 * SUCH DAMAGE. 31 * 32 * $KAME: in6_gif.c,v 1.49 2001/05/14 14:02:17 itojun Exp $ 33 */ 34 35 #include <sys/cdefs.h> 36 #include "opt_inet.h" 37 #include "opt_inet6.h" 38 39 #include <sys/param.h> 40 #include <sys/systm.h> 41 #include <sys/jail.h> 42 #include <sys/socket.h> 43 #include <sys/sockio.h> 44 #include <sys/mbuf.h> 45 #include <sys/errno.h> 46 #include <sys/kernel.h> 47 #include <sys/syslog.h> 48 #include <sys/sysctl.h> 49 #include <sys/malloc.h> 50 #include <sys/proc.h> 51 52 #include <net/ethernet.h> 53 #include <net/if.h> 54 #include <net/if_var.h> 55 #include <net/if_private.h> 56 #include <net/route.h> 57 #include <net/vnet.h> 58 59 #include <netinet/in.h> 60 #include <netinet/in_systm.h> 61 #ifdef INET 62 #include <netinet/ip.h> 63 #include <netinet/ip_ecn.h> 64 #endif 65 #include <netinet/ip_encap.h> 66 #include <netinet/ip6.h> 67 #include <netinet6/ip6_var.h> 68 #include <netinet6/in6_var.h> 69 #include <netinet6/scope6_var.h> 70 #include <netinet6/ip6_ecn.h> 71 #include <netinet6/in6_fib.h> 72 73 #include <net/if_gif.h> 74 75 #define GIF_HLIM 30 76 VNET_DEFINE_STATIC(int, ip6_gif_hlim) = GIF_HLIM; 77 #define V_ip6_gif_hlim VNET(ip6_gif_hlim) 78 79 SYSCTL_DECL(_net_inet6_ip6); 80 SYSCTL_INT(_net_inet6_ip6, IPV6CTL_GIF_HLIM, gifhlim, 81 CTLFLAG_VNET | CTLFLAG_RW, &VNET_NAME(ip6_gif_hlim), 0, 82 "Default hop limit for encapsulated packets"); 83 84 /* 85 * We keep interfaces in a hash table using src+dst as key. 86 * Interfaces with GIF_IGNORE_SOURCE flag are linked into plain list. 87 */ 88 VNET_DEFINE_STATIC(struct gif_list *, ipv6_hashtbl) = NULL; 89 VNET_DEFINE_STATIC(struct gif_list *, ipv6_srchashtbl) = NULL; 90 VNET_DEFINE_STATIC(struct gif_list, ipv6_list) = CK_LIST_HEAD_INITIALIZER(); 91 #define V_ipv6_hashtbl VNET(ipv6_hashtbl) 92 #define V_ipv6_srchashtbl VNET(ipv6_srchashtbl) 93 #define V_ipv6_list VNET(ipv6_list) 94 95 #define GIF_HASH(src, dst) (V_ipv6_hashtbl[\ 96 in6_gif_hashval((src), (dst)) & (GIF_HASH_SIZE - 1)]) 97 #define GIF_SRCHASH(src) (V_ipv6_srchashtbl[\ 98 fnv_32_buf((src), sizeof(*src), FNV1_32_INIT) & (GIF_HASH_SIZE - 1)]) 99 #define GIF_HASH_SC(sc) GIF_HASH(&(sc)->gif_ip6hdr->ip6_src,\ 100 &(sc)->gif_ip6hdr->ip6_dst) 101 static uint32_t 102 in6_gif_hashval(const struct in6_addr *src, const struct in6_addr *dst) 103 { 104 uint32_t ret; 105 106 ret = fnv_32_buf(src, sizeof(*src), FNV1_32_INIT); 107 return (fnv_32_buf(dst, sizeof(*dst), ret)); 108 } 109 110 static int 111 in6_gif_checkdup(const struct gif_softc *sc, const struct in6_addr *src, 112 const struct in6_addr *dst) 113 { 114 struct gif_softc *tmp; 115 116 if (sc->gif_family == AF_INET6 && 117 IN6_ARE_ADDR_EQUAL(&sc->gif_ip6hdr->ip6_src, src) && 118 IN6_ARE_ADDR_EQUAL(&sc->gif_ip6hdr->ip6_dst, dst)) 119 return (EEXIST); 120 121 CK_LIST_FOREACH(tmp, &GIF_HASH(src, dst), chain) { 122 if (tmp == sc) 123 continue; 124 if (IN6_ARE_ADDR_EQUAL(&tmp->gif_ip6hdr->ip6_src, src) && 125 IN6_ARE_ADDR_EQUAL(&tmp->gif_ip6hdr->ip6_dst, dst)) 126 return (EADDRNOTAVAIL); 127 } 128 return (0); 129 } 130 131 /* 132 * Check that ingress address belongs to local host. 133 */ 134 static void 135 in6_gif_set_running(struct gif_softc *sc) 136 { 137 138 if (in6_localip(&sc->gif_ip6hdr->ip6_src)) 139 GIF2IFP(sc)->if_drv_flags |= IFF_DRV_RUNNING; 140 else 141 GIF2IFP(sc)->if_drv_flags &= ~IFF_DRV_RUNNING; 142 } 143 144 /* 145 * ifaddr_event handler. 146 * Clear IFF_DRV_RUNNING flag when ingress address disappears to prevent 147 * source address spoofing. 148 */ 149 static void 150 in6_gif_srcaddr(void *arg __unused, const struct sockaddr *sa, int event) 151 { 152 const struct sockaddr_in6 *sin; 153 struct gif_softc *sc; 154 155 /* Check that VNET is ready */ 156 if (V_ipv6_hashtbl == NULL) 157 return; 158 159 NET_EPOCH_ASSERT(); 160 sin = (const struct sockaddr_in6 *)sa; 161 CK_LIST_FOREACH(sc, &GIF_SRCHASH(&sin->sin6_addr), srchash) { 162 if (IN6_ARE_ADDR_EQUAL(&sc->gif_ip6hdr->ip6_src, 163 &sin->sin6_addr) == 0) 164 continue; 165 in6_gif_set_running(sc); 166 } 167 } 168 169 static void 170 in6_gif_attach(struct gif_softc *sc) 171 { 172 173 if (sc->gif_options & GIF_IGNORE_SOURCE) 174 CK_LIST_INSERT_HEAD(&V_ipv6_list, sc, chain); 175 else 176 CK_LIST_INSERT_HEAD(&GIF_HASH_SC(sc), sc, chain); 177 178 CK_LIST_INSERT_HEAD(&GIF_SRCHASH(&sc->gif_ip6hdr->ip6_src), 179 sc, srchash); 180 } 181 182 int 183 in6_gif_setopts(struct gif_softc *sc, u_int options) 184 { 185 186 /* NOTE: we are protected with gif_ioctl_sx lock */ 187 MPASS(sc->gif_family == AF_INET6); 188 MPASS(sc->gif_options != options); 189 190 if ((options & GIF_IGNORE_SOURCE) != 191 (sc->gif_options & GIF_IGNORE_SOURCE)) { 192 CK_LIST_REMOVE(sc, srchash); 193 CK_LIST_REMOVE(sc, chain); 194 sc->gif_options = options; 195 in6_gif_attach(sc); 196 } 197 198 if ((options & GIF_NOCLAMP) != 199 (sc->gif_options & GIF_NOCLAMP)) { 200 sc->gif_options = options; 201 } 202 return (0); 203 } 204 205 int 206 in6_gif_ioctl(struct gif_softc *sc, u_long cmd, caddr_t data) 207 { 208 struct in6_ifreq *ifr = (struct in6_ifreq *)data; 209 struct sockaddr_in6 *dst, *src; 210 struct ip6_hdr *ip6; 211 int error; 212 213 /* NOTE: we are protected with gif_ioctl_sx lock */ 214 error = EINVAL; 215 switch (cmd) { 216 case SIOCSIFPHYADDR_IN6: 217 src = &((struct in6_aliasreq *)data)->ifra_addr; 218 dst = &((struct in6_aliasreq *)data)->ifra_dstaddr; 219 220 /* sanity checks */ 221 if (src->sin6_family != dst->sin6_family || 222 src->sin6_family != AF_INET6 || 223 src->sin6_len != dst->sin6_len || 224 src->sin6_len != sizeof(*src)) 225 break; 226 if (IN6_IS_ADDR_UNSPECIFIED(&src->sin6_addr) || 227 IN6_IS_ADDR_UNSPECIFIED(&dst->sin6_addr)) { 228 error = EADDRNOTAVAIL; 229 break; 230 } 231 /* 232 * Check validity of the scope zone ID of the 233 * addresses, and convert it into the kernel 234 * internal form if necessary. 235 */ 236 if ((error = sa6_embedscope(src, 0)) != 0 || 237 (error = sa6_embedscope(dst, 0)) != 0) 238 break; 239 240 if (V_ipv6_hashtbl == NULL) { 241 V_ipv6_hashtbl = gif_hashinit(); 242 V_ipv6_srchashtbl = gif_hashinit(); 243 } 244 error = in6_gif_checkdup(sc, &src->sin6_addr, 245 &dst->sin6_addr); 246 if (error == EADDRNOTAVAIL) 247 break; 248 if (error == EEXIST) { 249 /* Addresses are the same. Just return. */ 250 error = 0; 251 break; 252 } 253 ip6 = malloc(sizeof(*ip6), M_GIF, M_WAITOK | M_ZERO); 254 ip6->ip6_src = src->sin6_addr; 255 ip6->ip6_dst = dst->sin6_addr; 256 ip6->ip6_vfc = IPV6_VERSION; 257 if (sc->gif_family != 0) { 258 /* Detach existing tunnel first */ 259 CK_LIST_REMOVE(sc, srchash); 260 CK_LIST_REMOVE(sc, chain); 261 GIF_WAIT(); 262 free(sc->gif_hdr, M_GIF); 263 /* XXX: should we notify about link state change? */ 264 } 265 sc->gif_family = AF_INET6; 266 sc->gif_ip6hdr = ip6; 267 in6_gif_attach(sc); 268 in6_gif_set_running(sc); 269 break; 270 case SIOCGIFPSRCADDR_IN6: 271 case SIOCGIFPDSTADDR_IN6: 272 if (sc->gif_family != AF_INET6) { 273 error = EADDRNOTAVAIL; 274 break; 275 } 276 src = (struct sockaddr_in6 *)&ifr->ifr_addr; 277 memset(src, 0, sizeof(*src)); 278 src->sin6_family = AF_INET6; 279 src->sin6_len = sizeof(*src); 280 src->sin6_addr = (cmd == SIOCGIFPSRCADDR_IN6) ? 281 sc->gif_ip6hdr->ip6_src: sc->gif_ip6hdr->ip6_dst; 282 error = prison_if(curthread->td_ucred, (struct sockaddr *)src); 283 if (error == 0) 284 error = sa6_recoverscope(src); 285 if (error != 0) 286 memset(src, 0, sizeof(*src)); 287 break; 288 } 289 return (error); 290 } 291 292 int 293 in6_gif_output(struct ifnet *ifp, struct mbuf *m, int proto, uint8_t ecn) 294 { 295 struct gif_softc *sc = ifp->if_softc; 296 struct ip6_hdr *ip6; 297 u_long mtu; 298 299 /* prepend new IP header */ 300 NET_EPOCH_ASSERT(); 301 M_PREPEND(m, sizeof(struct ip6_hdr), M_NOWAIT); 302 if (m == NULL) 303 return (ENOBUFS); 304 305 ip6 = mtod(m, struct ip6_hdr *); 306 MPASS(sc->gif_family == AF_INET6); 307 memcpy(ip6, sc->gif_ip6hdr, sizeof(struct ip6_hdr)); 308 309 ip6->ip6_flow |= htonl((uint32_t)ecn << 20); 310 ip6->ip6_nxt = proto; 311 ip6->ip6_hlim = V_ip6_gif_hlim; 312 /* 313 * Enforce fragmentation to minimum MTU, even if the interface MTU 314 * is larger, to avoid path MTU discovery when NOCLAMP is not 315 * set (default). IPv6 does not allow fragmentation on intermediate 316 * router nodes, so it is too painful to ask for resend of inner 317 * packet, to achieve path MTU discovery for encapsulated packets. 318 */ 319 mtu = ((sc->gif_options & GIF_NOCLAMP) == 0) ? IPV6_MINMTU : 0; 320 321 return (ip6_output(m, 0, NULL, mtu, 0, NULL, NULL)); 322 } 323 324 static int 325 in6_gif_input(struct mbuf *m, int off, int proto, void *arg) 326 { 327 struct gif_softc *sc = arg; 328 struct ifnet *gifp; 329 struct ip6_hdr *ip6; 330 uint8_t ecn; 331 332 NET_EPOCH_ASSERT(); 333 if (sc == NULL) { 334 m_freem(m); 335 IP6STAT_INC(ip6s_nogif); 336 return (IPPROTO_DONE); 337 } 338 gifp = GIF2IFP(sc); 339 if ((gifp->if_flags & IFF_UP) != 0) { 340 ip6 = mtod(m, struct ip6_hdr *); 341 ecn = IPV6_TRAFFIC_CLASS(ip6); 342 m_adj(m, off); 343 gif_input(m, gifp, proto, ecn); 344 } else { 345 m_freem(m); 346 IP6STAT_INC(ip6s_nogif); 347 } 348 return (IPPROTO_DONE); 349 } 350 351 static int 352 in6_gif_lookup(const struct mbuf *m, int off, int proto, void **arg) 353 { 354 const struct ip6_hdr *ip6; 355 struct gif_softc *sc; 356 int ret; 357 358 if (V_ipv6_hashtbl == NULL) 359 return (0); 360 361 NET_EPOCH_ASSERT(); 362 /* 363 * NOTE: it is safe to iterate without any locking here, because softc 364 * can be reclaimed only when we are not within net_epoch_preempt 365 * section, but ip_encap lookup+input are executed in epoch section. 366 */ 367 ip6 = mtod(m, const struct ip6_hdr *); 368 ret = 0; 369 CK_LIST_FOREACH(sc, &GIF_HASH(&ip6->ip6_dst, &ip6->ip6_src), chain) { 370 /* 371 * This is an inbound packet, its ip6_dst is source address 372 * in softc. 373 */ 374 if (IN6_ARE_ADDR_EQUAL(&sc->gif_ip6hdr->ip6_src, 375 &ip6->ip6_dst) && 376 IN6_ARE_ADDR_EQUAL(&sc->gif_ip6hdr->ip6_dst, 377 &ip6->ip6_src)) { 378 ret = ENCAP_DRV_LOOKUP; 379 goto done; 380 } 381 } 382 /* 383 * No exact match. 384 * Check the list of interfaces with GIF_IGNORE_SOURCE flag. 385 */ 386 CK_LIST_FOREACH(sc, &V_ipv6_list, chain) { 387 if (IN6_ARE_ADDR_EQUAL(&sc->gif_ip6hdr->ip6_src, 388 &ip6->ip6_dst)) { 389 ret = 128 + 8; /* src + proto */ 390 goto done; 391 } 392 } 393 return (0); 394 done: 395 if ((GIF2IFP(sc)->if_flags & IFF_UP) == 0) 396 return (0); 397 /* ingress filters on outer source */ 398 if ((GIF2IFP(sc)->if_flags & IFF_LINK2) == 0) { 399 if (fib6_check_urpf(sc->gif_fibnum, &ip6->ip6_src, 400 ntohs(in6_getscope(&ip6->ip6_src)), NHR_NONE, 401 m->m_pkthdr.rcvif) == 0) 402 return (0); 403 } 404 *arg = sc; 405 return (ret); 406 } 407 408 static const struct srcaddrtab *ipv6_srcaddrtab; 409 static struct { 410 const struct encap_config encap; 411 const struct encaptab *cookie; 412 } ipv6_encap_cfg[] = { 413 #ifdef INET 414 { 415 .encap = { 416 .proto = IPPROTO_IPV4, 417 .min_length = sizeof(struct ip6_hdr) + 418 sizeof(struct ip), 419 .exact_match = ENCAP_DRV_LOOKUP, 420 .lookup = in6_gif_lookup, 421 .input = in6_gif_input 422 }, 423 }, 424 #endif 425 { 426 .encap = { 427 .proto = IPPROTO_IPV6, 428 .min_length = 2 * sizeof(struct ip6_hdr), 429 .exact_match = ENCAP_DRV_LOOKUP, 430 .lookup = in6_gif_lookup, 431 .input = in6_gif_input 432 }, 433 }, 434 { 435 .encap = { 436 .proto = IPPROTO_ETHERIP, 437 .min_length = sizeof(struct ip6_hdr) + 438 sizeof(struct etherip_header) + 439 sizeof(struct ether_header), 440 .exact_match = ENCAP_DRV_LOOKUP, 441 .lookup = in6_gif_lookup, 442 .input = in6_gif_input 443 }, 444 } 445 }; 446 447 void 448 in6_gif_init(void) 449 { 450 int i; 451 452 if (!IS_DEFAULT_VNET(curvnet)) 453 return; 454 455 ipv6_srcaddrtab = ip6_encap_register_srcaddr(in6_gif_srcaddr, 456 NULL, M_WAITOK); 457 for (i = 0; i < nitems(ipv6_encap_cfg); i++) 458 ipv6_encap_cfg[i].cookie = ip6_encap_attach( 459 &ipv6_encap_cfg[i].encap, NULL, M_WAITOK); 460 } 461 462 void 463 in6_gif_uninit(void) 464 { 465 int i; 466 467 if (IS_DEFAULT_VNET(curvnet)) { 468 for (i = 0; i < nitems(ipv6_encap_cfg); i++) 469 ip6_encap_detach(ipv6_encap_cfg[i].cookie); 470 ip6_encap_unregister_srcaddr(ipv6_srcaddrtab); 471 } 472 if (V_ipv6_hashtbl != NULL) { 473 gif_hashdestroy(V_ipv6_hashtbl); 474 V_ipv6_hashtbl = NULL; 475 GIF_WAIT(); 476 gif_hashdestroy(V_ipv6_srchashtbl); 477 } 478 } 479