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/route.h> 58 #include <net/vnet.h> 59 60 #include <netinet/in.h> 61 #include <netinet/in_systm.h> 62 #ifdef INET 63 #include <netinet/ip.h> 64 #include <netinet/ip_ecn.h> 65 #endif 66 #include <netinet/ip_encap.h> 67 #include <netinet/ip6.h> 68 #include <netinet6/ip6_var.h> 69 #include <netinet6/in6_var.h> 70 #include <netinet6/scope6_var.h> 71 #include <netinet6/ip6_ecn.h> 72 #include <netinet6/in6_fib.h> 73 74 #include <net/if_gif.h> 75 76 #define GIF_HLIM 30 77 VNET_DEFINE_STATIC(int, ip6_gif_hlim) = GIF_HLIM; 78 #define V_ip6_gif_hlim VNET(ip6_gif_hlim) 79 80 SYSCTL_DECL(_net_inet6_ip6); 81 SYSCTL_INT(_net_inet6_ip6, IPV6CTL_GIF_HLIM, gifhlim, 82 CTLFLAG_VNET | CTLFLAG_RW, &VNET_NAME(ip6_gif_hlim), 0, 83 "Default hop limit for encapsulated packets"); 84 85 /* 86 * We keep interfaces in a hash table using src+dst as key. 87 * Interfaces with GIF_IGNORE_SOURCE flag are linked into plain list. 88 */ 89 VNET_DEFINE_STATIC(struct gif_list *, ipv6_hashtbl) = NULL; 90 VNET_DEFINE_STATIC(struct gif_list *, ipv6_srchashtbl) = NULL; 91 VNET_DEFINE_STATIC(struct gif_list, ipv6_list) = CK_LIST_HEAD_INITIALIZER(); 92 #define V_ipv6_hashtbl VNET(ipv6_hashtbl) 93 #define V_ipv6_srchashtbl VNET(ipv6_srchashtbl) 94 #define V_ipv6_list VNET(ipv6_list) 95 96 #define GIF_HASH(src, dst) (V_ipv6_hashtbl[\ 97 in6_gif_hashval((src), (dst)) & (GIF_HASH_SIZE - 1)]) 98 #define GIF_SRCHASH(src) (V_ipv6_srchashtbl[\ 99 fnv_32_buf((src), sizeof(*src), FNV1_32_INIT) & (GIF_HASH_SIZE - 1)]) 100 #define GIF_HASH_SC(sc) GIF_HASH(&(sc)->gif_ip6hdr->ip6_src,\ 101 &(sc)->gif_ip6hdr->ip6_dst) 102 static uint32_t 103 in6_gif_hashval(const struct in6_addr *src, const struct in6_addr *dst) 104 { 105 uint32_t ret; 106 107 ret = fnv_32_buf(src, sizeof(*src), FNV1_32_INIT); 108 return (fnv_32_buf(dst, sizeof(*dst), ret)); 109 } 110 111 static int 112 in6_gif_checkdup(const struct gif_softc *sc, const struct in6_addr *src, 113 const struct in6_addr *dst) 114 { 115 struct gif_softc *tmp; 116 117 if (sc->gif_family == AF_INET6 && 118 IN6_ARE_ADDR_EQUAL(&sc->gif_ip6hdr->ip6_src, src) && 119 IN6_ARE_ADDR_EQUAL(&sc->gif_ip6hdr->ip6_dst, dst)) 120 return (EEXIST); 121 122 CK_LIST_FOREACH(tmp, &GIF_HASH(src, dst), chain) { 123 if (tmp == sc) 124 continue; 125 if (IN6_ARE_ADDR_EQUAL(&tmp->gif_ip6hdr->ip6_src, src) && 126 IN6_ARE_ADDR_EQUAL(&tmp->gif_ip6hdr->ip6_dst, dst)) 127 return (EADDRNOTAVAIL); 128 } 129 return (0); 130 } 131 132 /* 133 * Check that ingress address belongs to local host. 134 */ 135 static void 136 in6_gif_set_running(struct gif_softc *sc) 137 { 138 139 if (in6_localip(&sc->gif_ip6hdr->ip6_src)) 140 GIF2IFP(sc)->if_drv_flags |= IFF_DRV_RUNNING; 141 else 142 GIF2IFP(sc)->if_drv_flags &= ~IFF_DRV_RUNNING; 143 } 144 145 /* 146 * ifaddr_event handler. 147 * Clear IFF_DRV_RUNNING flag when ingress address disappears to prevent 148 * source address spoofing. 149 */ 150 static void 151 in6_gif_srcaddr(void *arg __unused, const struct sockaddr *sa, int event) 152 { 153 const struct sockaddr_in6 *sin; 154 struct gif_softc *sc; 155 156 if (V_ipv6_srchashtbl == NULL) 157 return; 158 159 MPASS(in_epoch(net_epoch_preempt)); 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 return (0); 198 } 199 200 int 201 in6_gif_ioctl(struct gif_softc *sc, u_long cmd, caddr_t data) 202 { 203 struct in6_ifreq *ifr = (struct in6_ifreq *)data; 204 struct sockaddr_in6 *dst, *src; 205 struct ip6_hdr *ip6; 206 int error; 207 208 /* NOTE: we are protected with gif_ioctl_sx lock */ 209 error = EINVAL; 210 switch (cmd) { 211 case SIOCSIFPHYADDR_IN6: 212 src = &((struct in6_aliasreq *)data)->ifra_addr; 213 dst = &((struct in6_aliasreq *)data)->ifra_dstaddr; 214 215 /* sanity checks */ 216 if (src->sin6_family != dst->sin6_family || 217 src->sin6_family != AF_INET6 || 218 src->sin6_len != dst->sin6_len || 219 src->sin6_len != sizeof(*src)) 220 break; 221 if (IN6_IS_ADDR_UNSPECIFIED(&src->sin6_addr) || 222 IN6_IS_ADDR_UNSPECIFIED(&dst->sin6_addr)) { 223 error = EADDRNOTAVAIL; 224 break; 225 } 226 /* 227 * Check validity of the scope zone ID of the 228 * addresses, and convert it into the kernel 229 * internal form if necessary. 230 */ 231 if ((error = sa6_embedscope(src, 0)) != 0 || 232 (error = sa6_embedscope(dst, 0)) != 0) 233 break; 234 235 if (V_ipv6_hashtbl == NULL) { 236 V_ipv6_hashtbl = gif_hashinit(); 237 V_ipv6_srchashtbl = gif_hashinit(); 238 } 239 error = in6_gif_checkdup(sc, &src->sin6_addr, 240 &dst->sin6_addr); 241 if (error == EADDRNOTAVAIL) 242 break; 243 if (error == EEXIST) { 244 /* Addresses are the same. Just return. */ 245 error = 0; 246 break; 247 } 248 ip6 = malloc(sizeof(*ip6), M_GIF, M_WAITOK | M_ZERO); 249 ip6->ip6_src = src->sin6_addr; 250 ip6->ip6_dst = dst->sin6_addr; 251 ip6->ip6_vfc = IPV6_VERSION; 252 if (sc->gif_family != 0) { 253 /* Detach existing tunnel first */ 254 CK_LIST_REMOVE(sc, srchash); 255 CK_LIST_REMOVE(sc, chain); 256 GIF_WAIT(); 257 free(sc->gif_hdr, M_GIF); 258 /* XXX: should we notify about link state change? */ 259 } 260 sc->gif_family = AF_INET6; 261 sc->gif_ip6hdr = ip6; 262 in6_gif_attach(sc); 263 in6_gif_set_running(sc); 264 break; 265 case SIOCGIFPSRCADDR_IN6: 266 case SIOCGIFPDSTADDR_IN6: 267 if (sc->gif_family != AF_INET6) { 268 error = EADDRNOTAVAIL; 269 break; 270 } 271 src = (struct sockaddr_in6 *)&ifr->ifr_addr; 272 memset(src, 0, sizeof(*src)); 273 src->sin6_family = AF_INET6; 274 src->sin6_len = sizeof(*src); 275 src->sin6_addr = (cmd == SIOCGIFPSRCADDR_IN6) ? 276 sc->gif_ip6hdr->ip6_src: sc->gif_ip6hdr->ip6_dst; 277 error = prison_if(curthread->td_ucred, (struct sockaddr *)src); 278 if (error == 0) 279 error = sa6_recoverscope(src); 280 if (error != 0) 281 memset(src, 0, sizeof(*src)); 282 break; 283 } 284 return (error); 285 } 286 287 int 288 in6_gif_output(struct ifnet *ifp, struct mbuf *m, int proto, uint8_t ecn) 289 { 290 struct gif_softc *sc = ifp->if_softc; 291 struct ip6_hdr *ip6; 292 int len; 293 294 /* prepend new IP header */ 295 MPASS(in_epoch(net_epoch_preempt)); 296 len = sizeof(struct ip6_hdr); 297 #ifndef __NO_STRICT_ALIGNMENT 298 if (proto == IPPROTO_ETHERIP) 299 len += ETHERIP_ALIGN; 300 #endif 301 M_PREPEND(m, len, M_NOWAIT); 302 if (m == NULL) 303 return (ENOBUFS); 304 #ifndef __NO_STRICT_ALIGNMENT 305 if (proto == IPPROTO_ETHERIP) { 306 len = mtod(m, vm_offset_t) & 3; 307 KASSERT(len == 0 || len == ETHERIP_ALIGN, 308 ("in6_gif_output: unexpected misalignment")); 309 m->m_data += len; 310 m->m_len -= ETHERIP_ALIGN; 311 } 312 #endif 313 314 ip6 = mtod(m, struct ip6_hdr *); 315 MPASS(sc->gif_family == AF_INET6); 316 bcopy(sc->gif_ip6hdr, ip6, sizeof(struct ip6_hdr)); 317 318 ip6->ip6_flow |= htonl((uint32_t)ecn << 20); 319 ip6->ip6_nxt = proto; 320 ip6->ip6_hlim = V_ip6_gif_hlim; 321 /* 322 * force fragmentation to minimum MTU, to avoid path MTU discovery. 323 * it is too painful to ask for resend of inner packet, to achieve 324 * path MTU discovery for encapsulated packets. 325 */ 326 return (ip6_output(m, 0, NULL, IPV6_MINMTU, 0, NULL, NULL)); 327 } 328 329 static int 330 in6_gif_input(struct mbuf *m, int off, int proto, void *arg) 331 { 332 struct gif_softc *sc = arg; 333 struct ifnet *gifp; 334 struct ip6_hdr *ip6; 335 uint8_t ecn; 336 337 MPASS(in_epoch(net_epoch_preempt)); 338 if (sc == NULL) { 339 m_freem(m); 340 IP6STAT_INC(ip6s_nogif); 341 return (IPPROTO_DONE); 342 } 343 gifp = GIF2IFP(sc); 344 if ((gifp->if_flags & IFF_UP) != 0) { 345 ip6 = mtod(m, struct ip6_hdr *); 346 ecn = (ntohl(ip6->ip6_flow) >> 20) & 0xff; 347 m_adj(m, off); 348 gif_input(m, gifp, proto, ecn); 349 } else { 350 m_freem(m); 351 IP6STAT_INC(ip6s_nogif); 352 } 353 return (IPPROTO_DONE); 354 } 355 356 static int 357 in6_gif_lookup(const struct mbuf *m, int off, int proto, void **arg) 358 { 359 const struct ip6_hdr *ip6; 360 struct gif_softc *sc; 361 int ret; 362 363 if (V_ipv6_hashtbl == NULL) 364 return (0); 365 366 MPASS(in_epoch(net_epoch_preempt)); 367 /* 368 * NOTE: it is safe to iterate without any locking here, because softc 369 * can be reclaimed only when we are not within net_epoch_preempt 370 * section, but ip_encap lookup+input are executed in epoch section. 371 */ 372 ip6 = mtod(m, const struct ip6_hdr *); 373 ret = 0; 374 CK_LIST_FOREACH(sc, &GIF_HASH(&ip6->ip6_dst, &ip6->ip6_src), chain) { 375 /* 376 * This is an inbound packet, its ip6_dst is source address 377 * in softc. 378 */ 379 if (IN6_ARE_ADDR_EQUAL(&sc->gif_ip6hdr->ip6_src, 380 &ip6->ip6_dst) && 381 IN6_ARE_ADDR_EQUAL(&sc->gif_ip6hdr->ip6_dst, 382 &ip6->ip6_src)) { 383 ret = ENCAP_DRV_LOOKUP; 384 goto done; 385 } 386 } 387 /* 388 * No exact match. 389 * Check the list of interfaces with GIF_IGNORE_SOURCE flag. 390 */ 391 CK_LIST_FOREACH(sc, &V_ipv6_list, chain) { 392 if (IN6_ARE_ADDR_EQUAL(&sc->gif_ip6hdr->ip6_src, 393 &ip6->ip6_dst)) { 394 ret = 128 + 8; /* src + proto */ 395 goto done; 396 } 397 } 398 return (0); 399 done: 400 if ((GIF2IFP(sc)->if_flags & IFF_UP) == 0) 401 return (0); 402 /* ingress filters on outer source */ 403 if ((GIF2IFP(sc)->if_flags & IFF_LINK2) == 0) { 404 struct nhop6_basic nh6; 405 406 if (fib6_lookup_nh_basic(sc->gif_fibnum, &ip6->ip6_src, 407 ntohs(in6_getscope(&ip6->ip6_src)), 0, 0, &nh6) != 0) 408 return (0); 409 410 if (nh6.nh_ifp != m->m_pkthdr.rcvif) 411 return (0); 412 } 413 *arg = sc; 414 return (ret); 415 } 416 417 static const struct srcaddrtab *ipv6_srcaddrtab; 418 static struct { 419 const struct encap_config encap; 420 const struct encaptab *cookie; 421 } ipv6_encap_cfg[] = { 422 #ifdef INET 423 { 424 .encap = { 425 .proto = IPPROTO_IPV4, 426 .min_length = sizeof(struct ip6_hdr) + 427 sizeof(struct ip), 428 .exact_match = ENCAP_DRV_LOOKUP, 429 .lookup = in6_gif_lookup, 430 .input = in6_gif_input 431 }, 432 }, 433 #endif 434 { 435 .encap = { 436 .proto = IPPROTO_IPV6, 437 .min_length = 2 * sizeof(struct ip6_hdr), 438 .exact_match = ENCAP_DRV_LOOKUP, 439 .lookup = in6_gif_lookup, 440 .input = in6_gif_input 441 }, 442 }, 443 { 444 .encap = { 445 .proto = IPPROTO_ETHERIP, 446 .min_length = sizeof(struct ip6_hdr) + 447 sizeof(struct etherip_header) + 448 sizeof(struct ether_header), 449 .exact_match = ENCAP_DRV_LOOKUP, 450 .lookup = in6_gif_lookup, 451 .input = in6_gif_input 452 }, 453 } 454 }; 455 456 void 457 in6_gif_init(void) 458 { 459 int i; 460 461 if (!IS_DEFAULT_VNET(curvnet)) 462 return; 463 464 ipv6_srcaddrtab = ip6_encap_register_srcaddr(in6_gif_srcaddr, 465 NULL, M_WAITOK); 466 for (i = 0; i < nitems(ipv6_encap_cfg); i++) 467 ipv6_encap_cfg[i].cookie = ip6_encap_attach( 468 &ipv6_encap_cfg[i].encap, NULL, M_WAITOK); 469 } 470 471 void 472 in6_gif_uninit(void) 473 { 474 int i; 475 476 if (IS_DEFAULT_VNET(curvnet)) { 477 for (i = 0; i < nitems(ipv6_encap_cfg); i++) 478 ip6_encap_detach(ipv6_encap_cfg[i].cookie); 479 ip6_encap_unregister_srcaddr(ipv6_srcaddrtab); 480 } 481 if (V_ipv6_hashtbl != NULL) { 482 gif_hashdestroy(V_ipv6_hashtbl); 483 gif_hashdestroy(V_ipv6_srchashtbl); 484 } 485 } 486