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 __FBSDID("$FreeBSD$"); 37 38 #include "opt_inet.h" 39 #include "opt_inet6.h" 40 41 #include <sys/param.h> 42 #include <sys/systm.h> 43 #include <sys/jail.h> 44 #include <sys/socket.h> 45 #include <sys/sockio.h> 46 #include <sys/mbuf.h> 47 #include <sys/errno.h> 48 #include <sys/kernel.h> 49 #include <sys/syslog.h> 50 #include <sys/sysctl.h> 51 #include <sys/malloc.h> 52 #include <sys/proc.h> 53 54 #include <net/ethernet.h> 55 #include <net/if.h> 56 #include <net/if_var.h> 57 #include <net/if_private.h> 58 #include <net/route.h> 59 #include <net/vnet.h> 60 61 #include <netinet/in.h> 62 #include <netinet/in_systm.h> 63 #ifdef INET 64 #include <netinet/ip.h> 65 #include <netinet/ip_ecn.h> 66 #endif 67 #include <netinet/ip_encap.h> 68 #include <netinet/ip6.h> 69 #include <netinet6/ip6_var.h> 70 #include <netinet6/in6_var.h> 71 #include <netinet6/scope6_var.h> 72 #include <netinet6/ip6_ecn.h> 73 #include <netinet6/in6_fib.h> 74 75 #include <net/if_gif.h> 76 77 #define GIF_HLIM 30 78 VNET_DEFINE_STATIC(int, ip6_gif_hlim) = GIF_HLIM; 79 #define V_ip6_gif_hlim VNET(ip6_gif_hlim) 80 81 SYSCTL_DECL(_net_inet6_ip6); 82 SYSCTL_INT(_net_inet6_ip6, IPV6CTL_GIF_HLIM, gifhlim, 83 CTLFLAG_VNET | CTLFLAG_RW, &VNET_NAME(ip6_gif_hlim), 0, 84 "Default hop limit for encapsulated packets"); 85 86 /* 87 * We keep interfaces in a hash table using src+dst as key. 88 * Interfaces with GIF_IGNORE_SOURCE flag are linked into plain list. 89 */ 90 VNET_DEFINE_STATIC(struct gif_list *, ipv6_hashtbl) = NULL; 91 VNET_DEFINE_STATIC(struct gif_list *, ipv6_srchashtbl) = NULL; 92 VNET_DEFINE_STATIC(struct gif_list, ipv6_list) = CK_LIST_HEAD_INITIALIZER(); 93 #define V_ipv6_hashtbl VNET(ipv6_hashtbl) 94 #define V_ipv6_srchashtbl VNET(ipv6_srchashtbl) 95 #define V_ipv6_list VNET(ipv6_list) 96 97 #define GIF_HASH(src, dst) (V_ipv6_hashtbl[\ 98 in6_gif_hashval((src), (dst)) & (GIF_HASH_SIZE - 1)]) 99 #define GIF_SRCHASH(src) (V_ipv6_srchashtbl[\ 100 fnv_32_buf((src), sizeof(*src), FNV1_32_INIT) & (GIF_HASH_SIZE - 1)]) 101 #define GIF_HASH_SC(sc) GIF_HASH(&(sc)->gif_ip6hdr->ip6_src,\ 102 &(sc)->gif_ip6hdr->ip6_dst) 103 static uint32_t 104 in6_gif_hashval(const struct in6_addr *src, const struct in6_addr *dst) 105 { 106 uint32_t ret; 107 108 ret = fnv_32_buf(src, sizeof(*src), FNV1_32_INIT); 109 return (fnv_32_buf(dst, sizeof(*dst), ret)); 110 } 111 112 static int 113 in6_gif_checkdup(const struct gif_softc *sc, const struct in6_addr *src, 114 const struct in6_addr *dst) 115 { 116 struct gif_softc *tmp; 117 118 if (sc->gif_family == AF_INET6 && 119 IN6_ARE_ADDR_EQUAL(&sc->gif_ip6hdr->ip6_src, src) && 120 IN6_ARE_ADDR_EQUAL(&sc->gif_ip6hdr->ip6_dst, dst)) 121 return (EEXIST); 122 123 CK_LIST_FOREACH(tmp, &GIF_HASH(src, dst), chain) { 124 if (tmp == sc) 125 continue; 126 if (IN6_ARE_ADDR_EQUAL(&tmp->gif_ip6hdr->ip6_src, src) && 127 IN6_ARE_ADDR_EQUAL(&tmp->gif_ip6hdr->ip6_dst, dst)) 128 return (EADDRNOTAVAIL); 129 } 130 return (0); 131 } 132 133 /* 134 * Check that ingress address belongs to local host. 135 */ 136 static void 137 in6_gif_set_running(struct gif_softc *sc) 138 { 139 140 if (in6_localip(&sc->gif_ip6hdr->ip6_src)) 141 GIF2IFP(sc)->if_drv_flags |= IFF_DRV_RUNNING; 142 else 143 GIF2IFP(sc)->if_drv_flags &= ~IFF_DRV_RUNNING; 144 } 145 146 /* 147 * ifaddr_event handler. 148 * Clear IFF_DRV_RUNNING flag when ingress address disappears to prevent 149 * source address spoofing. 150 */ 151 static void 152 in6_gif_srcaddr(void *arg __unused, const struct sockaddr *sa, int event) 153 { 154 const struct sockaddr_in6 *sin; 155 struct gif_softc *sc; 156 157 /* Check that VNET is ready */ 158 if (V_ipv6_hashtbl == NULL) 159 return; 160 161 NET_EPOCH_ASSERT(); 162 sin = (const struct sockaddr_in6 *)sa; 163 CK_LIST_FOREACH(sc, &GIF_SRCHASH(&sin->sin6_addr), srchash) { 164 if (IN6_ARE_ADDR_EQUAL(&sc->gif_ip6hdr->ip6_src, 165 &sin->sin6_addr) == 0) 166 continue; 167 in6_gif_set_running(sc); 168 } 169 } 170 171 static void 172 in6_gif_attach(struct gif_softc *sc) 173 { 174 175 if (sc->gif_options & GIF_IGNORE_SOURCE) 176 CK_LIST_INSERT_HEAD(&V_ipv6_list, sc, chain); 177 else 178 CK_LIST_INSERT_HEAD(&GIF_HASH_SC(sc), sc, chain); 179 180 CK_LIST_INSERT_HEAD(&GIF_SRCHASH(&sc->gif_ip6hdr->ip6_src), 181 sc, srchash); 182 } 183 184 int 185 in6_gif_setopts(struct gif_softc *sc, u_int options) 186 { 187 188 /* NOTE: we are protected with gif_ioctl_sx lock */ 189 MPASS(sc->gif_family == AF_INET6); 190 MPASS(sc->gif_options != options); 191 192 if ((options & GIF_IGNORE_SOURCE) != 193 (sc->gif_options & GIF_IGNORE_SOURCE)) { 194 CK_LIST_REMOVE(sc, srchash); 195 CK_LIST_REMOVE(sc, chain); 196 sc->gif_options = options; 197 in6_gif_attach(sc); 198 } 199 return (0); 200 } 201 202 int 203 in6_gif_ioctl(struct gif_softc *sc, u_long cmd, caddr_t data) 204 { 205 struct in6_ifreq *ifr = (struct in6_ifreq *)data; 206 struct sockaddr_in6 *dst, *src; 207 struct ip6_hdr *ip6; 208 int error; 209 210 /* NOTE: we are protected with gif_ioctl_sx lock */ 211 error = EINVAL; 212 switch (cmd) { 213 case SIOCSIFPHYADDR_IN6: 214 src = &((struct in6_aliasreq *)data)->ifra_addr; 215 dst = &((struct in6_aliasreq *)data)->ifra_dstaddr; 216 217 /* sanity checks */ 218 if (src->sin6_family != dst->sin6_family || 219 src->sin6_family != AF_INET6 || 220 src->sin6_len != dst->sin6_len || 221 src->sin6_len != sizeof(*src)) 222 break; 223 if (IN6_IS_ADDR_UNSPECIFIED(&src->sin6_addr) || 224 IN6_IS_ADDR_UNSPECIFIED(&dst->sin6_addr)) { 225 error = EADDRNOTAVAIL; 226 break; 227 } 228 /* 229 * Check validity of the scope zone ID of the 230 * addresses, and convert it into the kernel 231 * internal form if necessary. 232 */ 233 if ((error = sa6_embedscope(src, 0)) != 0 || 234 (error = sa6_embedscope(dst, 0)) != 0) 235 break; 236 237 if (V_ipv6_hashtbl == NULL) { 238 V_ipv6_hashtbl = gif_hashinit(); 239 V_ipv6_srchashtbl = gif_hashinit(); 240 } 241 error = in6_gif_checkdup(sc, &src->sin6_addr, 242 &dst->sin6_addr); 243 if (error == EADDRNOTAVAIL) 244 break; 245 if (error == EEXIST) { 246 /* Addresses are the same. Just return. */ 247 error = 0; 248 break; 249 } 250 ip6 = malloc(sizeof(*ip6), M_GIF, M_WAITOK | M_ZERO); 251 ip6->ip6_src = src->sin6_addr; 252 ip6->ip6_dst = dst->sin6_addr; 253 ip6->ip6_vfc = IPV6_VERSION; 254 if (sc->gif_family != 0) { 255 /* Detach existing tunnel first */ 256 CK_LIST_REMOVE(sc, srchash); 257 CK_LIST_REMOVE(sc, chain); 258 GIF_WAIT(); 259 free(sc->gif_hdr, M_GIF); 260 /* XXX: should we notify about link state change? */ 261 } 262 sc->gif_family = AF_INET6; 263 sc->gif_ip6hdr = ip6; 264 in6_gif_attach(sc); 265 in6_gif_set_running(sc); 266 break; 267 case SIOCGIFPSRCADDR_IN6: 268 case SIOCGIFPDSTADDR_IN6: 269 if (sc->gif_family != AF_INET6) { 270 error = EADDRNOTAVAIL; 271 break; 272 } 273 src = (struct sockaddr_in6 *)&ifr->ifr_addr; 274 memset(src, 0, sizeof(*src)); 275 src->sin6_family = AF_INET6; 276 src->sin6_len = sizeof(*src); 277 src->sin6_addr = (cmd == SIOCGIFPSRCADDR_IN6) ? 278 sc->gif_ip6hdr->ip6_src: sc->gif_ip6hdr->ip6_dst; 279 error = prison_if(curthread->td_ucred, (struct sockaddr *)src); 280 if (error == 0) 281 error = sa6_recoverscope(src); 282 if (error != 0) 283 memset(src, 0, sizeof(*src)); 284 break; 285 } 286 return (error); 287 } 288 289 int 290 in6_gif_output(struct ifnet *ifp, struct mbuf *m, int proto, uint8_t ecn) 291 { 292 struct gif_softc *sc = ifp->if_softc; 293 struct ip6_hdr *ip6; 294 int len; 295 296 /* prepend new IP header */ 297 NET_EPOCH_ASSERT(); 298 len = sizeof(struct ip6_hdr); 299 #ifndef __NO_STRICT_ALIGNMENT 300 if (proto == IPPROTO_ETHERIP) 301 len += ETHERIP_ALIGN; 302 #endif 303 M_PREPEND(m, len, M_NOWAIT); 304 if (m == NULL) 305 return (ENOBUFS); 306 #ifndef __NO_STRICT_ALIGNMENT 307 if (proto == IPPROTO_ETHERIP) { 308 len = mtod(m, vm_offset_t) & 3; 309 KASSERT(len == 0 || len == ETHERIP_ALIGN, 310 ("in6_gif_output: unexpected misalignment")); 311 m->m_data += len; 312 m->m_len -= ETHERIP_ALIGN; 313 } 314 #endif 315 316 ip6 = mtod(m, struct ip6_hdr *); 317 MPASS(sc->gif_family == AF_INET6); 318 bcopy(sc->gif_ip6hdr, ip6, sizeof(struct ip6_hdr)); 319 320 ip6->ip6_flow |= htonl((uint32_t)ecn << 20); 321 ip6->ip6_nxt = proto; 322 ip6->ip6_hlim = V_ip6_gif_hlim; 323 /* 324 * force fragmentation to minimum MTU, to avoid path MTU discovery. 325 * it is too painful to ask for resend of inner packet, to achieve 326 * path MTU discovery for encapsulated packets. 327 */ 328 return (ip6_output(m, 0, NULL, IPV6_MINMTU, 0, NULL, NULL)); 329 } 330 331 static int 332 in6_gif_input(struct mbuf *m, int off, int proto, void *arg) 333 { 334 struct gif_softc *sc = arg; 335 struct ifnet *gifp; 336 struct ip6_hdr *ip6; 337 uint8_t ecn; 338 339 NET_EPOCH_ASSERT(); 340 if (sc == NULL) { 341 m_freem(m); 342 IP6STAT_INC(ip6s_nogif); 343 return (IPPROTO_DONE); 344 } 345 gifp = GIF2IFP(sc); 346 if ((gifp->if_flags & IFF_UP) != 0) { 347 ip6 = mtod(m, struct ip6_hdr *); 348 ecn = IPV6_TRAFFIC_CLASS(ip6); 349 m_adj(m, off); 350 gif_input(m, gifp, proto, ecn); 351 } else { 352 m_freem(m); 353 IP6STAT_INC(ip6s_nogif); 354 } 355 return (IPPROTO_DONE); 356 } 357 358 static int 359 in6_gif_lookup(const struct mbuf *m, int off, int proto, void **arg) 360 { 361 const struct ip6_hdr *ip6; 362 struct gif_softc *sc; 363 int ret; 364 365 if (V_ipv6_hashtbl == NULL) 366 return (0); 367 368 NET_EPOCH_ASSERT(); 369 /* 370 * NOTE: it is safe to iterate without any locking here, because softc 371 * can be reclaimed only when we are not within net_epoch_preempt 372 * section, but ip_encap lookup+input are executed in epoch section. 373 */ 374 ip6 = mtod(m, const struct ip6_hdr *); 375 ret = 0; 376 CK_LIST_FOREACH(sc, &GIF_HASH(&ip6->ip6_dst, &ip6->ip6_src), chain) { 377 /* 378 * This is an inbound packet, its ip6_dst is source address 379 * in softc. 380 */ 381 if (IN6_ARE_ADDR_EQUAL(&sc->gif_ip6hdr->ip6_src, 382 &ip6->ip6_dst) && 383 IN6_ARE_ADDR_EQUAL(&sc->gif_ip6hdr->ip6_dst, 384 &ip6->ip6_src)) { 385 ret = ENCAP_DRV_LOOKUP; 386 goto done; 387 } 388 } 389 /* 390 * No exact match. 391 * Check the list of interfaces with GIF_IGNORE_SOURCE flag. 392 */ 393 CK_LIST_FOREACH(sc, &V_ipv6_list, chain) { 394 if (IN6_ARE_ADDR_EQUAL(&sc->gif_ip6hdr->ip6_src, 395 &ip6->ip6_dst)) { 396 ret = 128 + 8; /* src + proto */ 397 goto done; 398 } 399 } 400 return (0); 401 done: 402 if ((GIF2IFP(sc)->if_flags & IFF_UP) == 0) 403 return (0); 404 /* ingress filters on outer source */ 405 if ((GIF2IFP(sc)->if_flags & IFF_LINK2) == 0) { 406 if (fib6_check_urpf(sc->gif_fibnum, &ip6->ip6_src, 407 ntohs(in6_getscope(&ip6->ip6_src)), NHR_NONE, 408 m->m_pkthdr.rcvif) == 0) 409 return (0); 410 } 411 *arg = sc; 412 return (ret); 413 } 414 415 static const struct srcaddrtab *ipv6_srcaddrtab; 416 static struct { 417 const struct encap_config encap; 418 const struct encaptab *cookie; 419 } ipv6_encap_cfg[] = { 420 #ifdef INET 421 { 422 .encap = { 423 .proto = IPPROTO_IPV4, 424 .min_length = sizeof(struct ip6_hdr) + 425 sizeof(struct ip), 426 .exact_match = ENCAP_DRV_LOOKUP, 427 .lookup = in6_gif_lookup, 428 .input = in6_gif_input 429 }, 430 }, 431 #endif 432 { 433 .encap = { 434 .proto = IPPROTO_IPV6, 435 .min_length = 2 * sizeof(struct ip6_hdr), 436 .exact_match = ENCAP_DRV_LOOKUP, 437 .lookup = in6_gif_lookup, 438 .input = in6_gif_input 439 }, 440 }, 441 { 442 .encap = { 443 .proto = IPPROTO_ETHERIP, 444 .min_length = sizeof(struct ip6_hdr) + 445 sizeof(struct etherip_header) + 446 sizeof(struct ether_header), 447 .exact_match = ENCAP_DRV_LOOKUP, 448 .lookup = in6_gif_lookup, 449 .input = in6_gif_input 450 }, 451 } 452 }; 453 454 void 455 in6_gif_init(void) 456 { 457 int i; 458 459 if (!IS_DEFAULT_VNET(curvnet)) 460 return; 461 462 ipv6_srcaddrtab = ip6_encap_register_srcaddr(in6_gif_srcaddr, 463 NULL, M_WAITOK); 464 for (i = 0; i < nitems(ipv6_encap_cfg); i++) 465 ipv6_encap_cfg[i].cookie = ip6_encap_attach( 466 &ipv6_encap_cfg[i].encap, NULL, M_WAITOK); 467 } 468 469 void 470 in6_gif_uninit(void) 471 { 472 int i; 473 474 if (IS_DEFAULT_VNET(curvnet)) { 475 for (i = 0; i < nitems(ipv6_encap_cfg); i++) 476 ip6_encap_detach(ipv6_encap_cfg[i].cookie); 477 ip6_encap_unregister_srcaddr(ipv6_srcaddrtab); 478 } 479 if (V_ipv6_hashtbl != NULL) { 480 gif_hashdestroy(V_ipv6_hashtbl); 481 V_ipv6_hashtbl = NULL; 482 GIF_WAIT(); 483 gif_hashdestroy(V_ipv6_srchashtbl); 484 } 485 } 486