1 /* $NetBSD: ip_input.c,v 1.317 2014/05/30 01:39:03 christos Exp $ */ 2 3 /* 4 * Copyright (C) 1995, 1996, 1997, and 1998 WIDE Project. 5 * All rights reserved. 6 * 7 * Redistribution and use in source and binary forms, with or without 8 * modification, are permitted provided that the following conditions 9 * are met: 10 * 1. Redistributions of source code must retain the above copyright 11 * notice, this list of conditions and the following disclaimer. 12 * 2. Redistributions in binary form must reproduce the above copyright 13 * notice, this list of conditions and the following disclaimer in the 14 * documentation and/or other materials provided with the distribution. 15 * 3. Neither the name of the project nor the names of its contributors 16 * may be used to endorse or promote products derived from this software 17 * without specific prior written permission. 18 * 19 * THIS SOFTWARE IS PROVIDED BY THE PROJECT AND CONTRIBUTORS ``AS IS'' AND 20 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE 21 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE 22 * ARE DISCLAIMED. IN NO EVENT SHALL THE PROJECT OR CONTRIBUTORS BE LIABLE 23 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL 24 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS 25 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) 26 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT 27 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY 28 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF 29 * SUCH DAMAGE. 30 */ 31 32 /*- 33 * Copyright (c) 1998 The NetBSD Foundation, Inc. 34 * All rights reserved. 35 * 36 * This code is derived from software contributed to The NetBSD Foundation 37 * by Public Access Networks Corporation ("Panix"). It was developed under 38 * contract to Panix by Eric Haszlakiewicz and Thor Lancelot Simon. 39 * 40 * Redistribution and use in source and binary forms, with or without 41 * modification, are permitted provided that the following conditions 42 * are met: 43 * 1. Redistributions of source code must retain the above copyright 44 * notice, this list of conditions and the following disclaimer. 45 * 2. Redistributions in binary form must reproduce the above copyright 46 * notice, this list of conditions and the following disclaimer in the 47 * documentation and/or other materials provided with the distribution. 48 * 49 * THIS SOFTWARE IS PROVIDED BY THE NETBSD FOUNDATION, INC. AND CONTRIBUTORS 50 * ``AS IS'' AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED 51 * TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR 52 * PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE FOUNDATION OR CONTRIBUTORS 53 * BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR 54 * CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF 55 * SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS 56 * INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN 57 * CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) 58 * ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE 59 * POSSIBILITY OF SUCH DAMAGE. 60 */ 61 62 /* 63 * Copyright (c) 1982, 1986, 1988, 1993 64 * The Regents of the University of California. All rights reserved. 65 * 66 * Redistribution and use in source and binary forms, with or without 67 * modification, are permitted provided that the following conditions 68 * are met: 69 * 1. Redistributions of source code must retain the above copyright 70 * notice, this list of conditions and the following disclaimer. 71 * 2. Redistributions in binary form must reproduce the above copyright 72 * notice, this list of conditions and the following disclaimer in the 73 * documentation and/or other materials provided with the distribution. 74 * 3. Neither the name of the University nor the names of its contributors 75 * may be used to endorse or promote products derived from this software 76 * without specific prior written permission. 77 * 78 * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND 79 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE 80 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE 81 * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE 82 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL 83 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS 84 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) 85 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT 86 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY 87 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF 88 * SUCH DAMAGE. 89 * 90 * @(#)ip_input.c 8.2 (Berkeley) 1/4/94 91 */ 92 93 #include <sys/cdefs.h> 94 __KERNEL_RCSID(0, "$NetBSD: ip_input.c,v 1.317 2014/05/30 01:39:03 christos Exp $"); 95 96 #include "opt_inet.h" 97 #include "opt_compat_netbsd.h" 98 #include "opt_gateway.h" 99 #include "opt_ipsec.h" 100 #include "opt_mrouting.h" 101 #include "opt_mbuftrace.h" 102 #include "opt_inet_csum.h" 103 104 #include <sys/param.h> 105 #include <sys/systm.h> 106 #include <sys/mbuf.h> 107 #include <sys/domain.h> 108 #include <sys/protosw.h> 109 #include <sys/socket.h> 110 #include <sys/socketvar.h> 111 #include <sys/errno.h> 112 #include <sys/time.h> 113 #include <sys/kernel.h> 114 #include <sys/pool.h> 115 #include <sys/sysctl.h> 116 #include <sys/kauth.h> 117 118 #include <net/if.h> 119 #include <net/if_dl.h> 120 #include <net/route.h> 121 #include <net/pfil.h> 122 123 #include <netinet/in.h> 124 #include <netinet/in_systm.h> 125 #include <netinet/ip.h> 126 #include <netinet/in_pcb.h> 127 #include <netinet/in_proto.h> 128 #include <netinet/in_var.h> 129 #include <netinet/ip_var.h> 130 #include <netinet/ip_private.h> 131 #include <netinet/ip_icmp.h> 132 /* just for gif_ttl */ 133 #include <netinet/in_gif.h> 134 #include "gif.h" 135 #include <net/if_gre.h> 136 #include "gre.h" 137 138 #ifdef MROUTING 139 #include <netinet/ip_mroute.h> 140 #endif 141 #include <netinet/portalgo.h> 142 143 #ifdef IPSEC 144 #include <netipsec/ipsec.h> 145 #endif 146 147 #ifndef IPFORWARDING 148 #ifdef GATEWAY 149 #define IPFORWARDING 1 /* forward IP packets not for us */ 150 #else /* GATEWAY */ 151 #define IPFORWARDING 0 /* don't forward IP packets not for us */ 152 #endif /* GATEWAY */ 153 #endif /* IPFORWARDING */ 154 #ifndef IPSENDREDIRECTS 155 #define IPSENDREDIRECTS 1 156 #endif 157 #ifndef IPFORWSRCRT 158 #define IPFORWSRCRT 1 /* forward source-routed packets */ 159 #endif 160 #ifndef IPALLOWSRCRT 161 #define IPALLOWSRCRT 1 /* allow source-routed packets */ 162 #endif 163 #ifndef IPMTUDISC 164 #define IPMTUDISC 1 165 #endif 166 #ifndef IPMTUDISCTIMEOUT 167 #define IPMTUDISCTIMEOUT (10 * 60) /* as per RFC 1191 */ 168 #endif 169 170 #ifdef COMPAT_50 171 #include <compat/sys/time.h> 172 #include <compat/sys/socket.h> 173 #endif 174 175 /* 176 * Note: DIRECTED_BROADCAST is handled this way so that previous 177 * configuration using this option will Just Work. 178 */ 179 #ifndef IPDIRECTEDBCAST 180 #ifdef DIRECTED_BROADCAST 181 #define IPDIRECTEDBCAST 1 182 #else 183 #define IPDIRECTEDBCAST 0 184 #endif /* DIRECTED_BROADCAST */ 185 #endif /* IPDIRECTEDBCAST */ 186 int ipforwarding = IPFORWARDING; 187 int ipsendredirects = IPSENDREDIRECTS; 188 int ip_defttl = IPDEFTTL; 189 int ip_forwsrcrt = IPFORWSRCRT; 190 int ip_directedbcast = IPDIRECTEDBCAST; 191 int ip_allowsrcrt = IPALLOWSRCRT; 192 int ip_mtudisc = IPMTUDISC; 193 int ip_mtudisc_timeout = IPMTUDISCTIMEOUT; 194 #ifdef DIAGNOSTIC 195 int ipprintfs = 0; 196 #endif 197 198 int ip_do_randomid = 0; 199 200 /* 201 * XXX - Setting ip_checkinterface mostly implements the receive side of 202 * the Strong ES model described in RFC 1122, but since the routing table 203 * and transmit implementation do not implement the Strong ES model, 204 * setting this to 1 results in an odd hybrid. 205 * 206 * XXX - ip_checkinterface currently must be disabled if you use ipnat 207 * to translate the destination address to another local interface. 208 * 209 * XXX - ip_checkinterface must be disabled if you add IP aliases 210 * to the loopback interface instead of the interface where the 211 * packets for those addresses are received. 212 */ 213 int ip_checkinterface = 0; 214 215 struct rttimer_queue *ip_mtudisc_timeout_q = NULL; 216 217 struct ifqueue ipintrq; 218 219 ipid_state_t * ip_ids; 220 uint16_t ip_id; 221 222 percpu_t *ipstat_percpu; 223 224 pfil_head_t *inet_pfil_hook; 225 226 #ifdef INET_CSUM_COUNTERS 227 #include <sys/device.h> 228 229 struct evcnt ip_hwcsum_bad = EVCNT_INITIALIZER(EVCNT_TYPE_MISC, 230 NULL, "inet", "hwcsum bad"); 231 struct evcnt ip_hwcsum_ok = EVCNT_INITIALIZER(EVCNT_TYPE_MISC, 232 NULL, "inet", "hwcsum ok"); 233 struct evcnt ip_swcsum = EVCNT_INITIALIZER(EVCNT_TYPE_MISC, 234 NULL, "inet", "swcsum"); 235 236 #define INET_CSUM_COUNTER_INCR(ev) (ev)->ev_count++ 237 238 EVCNT_ATTACH_STATIC(ip_hwcsum_bad); 239 EVCNT_ATTACH_STATIC(ip_hwcsum_ok); 240 EVCNT_ATTACH_STATIC(ip_swcsum); 241 242 #else 243 244 #define INET_CSUM_COUNTER_INCR(ev) /* nothing */ 245 246 #endif /* INET_CSUM_COUNTERS */ 247 248 /* 249 * We need to save the IP options in case a protocol wants to respond 250 * to an incoming packet over the same route if the packet got here 251 * using IP source routing. This allows connection establishment and 252 * maintenance when the remote end is on a network that is not known 253 * to us. 254 */ 255 256 static int ip_nhops = 0; 257 258 static struct ip_srcrt { 259 struct in_addr dst; /* final destination */ 260 char nop; /* one NOP to align */ 261 char srcopt[IPOPT_OFFSET + 1]; /* OPTVAL, OLEN and OFFSET */ 262 struct in_addr route[MAX_IPOPTLEN/sizeof(struct in_addr)]; 263 } ip_srcrt; 264 265 static int ip_drainwanted; 266 267 static void save_rte(u_char *, struct in_addr); 268 269 #ifdef MBUFTRACE 270 struct mowner ip_rx_mowner = MOWNER_INIT("internet", "rx"); 271 struct mowner ip_tx_mowner = MOWNER_INIT("internet", "tx"); 272 #endif 273 274 static void ip_input(struct mbuf *); 275 static void ip_forward(struct mbuf *, int); 276 static bool ip_dooptions(struct mbuf *); 277 static struct in_ifaddr *ip_rtaddr(struct in_addr); 278 static void sysctl_net_inet_ip_setup(struct sysctllog **); 279 280 /* 281 * IP initialization: fill in IP protocol switch table. 282 * All protocols not implemented in kernel go to raw IP protocol handler. 283 */ 284 void 285 ip_init(void) 286 { 287 const struct protosw *pr; 288 289 in_init(); 290 sysctl_net_inet_ip_setup(NULL); 291 292 pr = pffindproto(PF_INET, IPPROTO_RAW, SOCK_RAW); 293 KASSERT(pr != NULL); 294 295 for (u_int i = 0; i < IPPROTO_MAX; i++) { 296 ip_protox[i] = pr - inetsw; 297 } 298 for (pr = inetdomain.dom_protosw; 299 pr < inetdomain.dom_protoswNPROTOSW; pr++) 300 if (pr->pr_domain->dom_family == PF_INET && 301 pr->pr_protocol && pr->pr_protocol != IPPROTO_RAW) 302 ip_protox[pr->pr_protocol] = pr - inetsw; 303 304 ip_reass_init(); 305 306 ip_ids = ip_id_init(); 307 ip_id = time_second & 0xfffff; 308 309 ipintrq.ifq_maxlen = IFQ_MAXLEN; 310 311 ip_mtudisc_timeout_q = rt_timer_queue_create(ip_mtudisc_timeout); 312 #ifdef GATEWAY 313 ipflow_init(); 314 #endif 315 316 /* Register our Packet Filter hook. */ 317 inet_pfil_hook = pfil_head_create(PFIL_TYPE_AF, (void *)AF_INET); 318 KASSERT(inet_pfil_hook != NULL); 319 320 #ifdef MBUFTRACE 321 MOWNER_ATTACH(&ip_tx_mowner); 322 MOWNER_ATTACH(&ip_rx_mowner); 323 #endif /* MBUFTRACE */ 324 325 ipstat_percpu = percpu_alloc(sizeof(uint64_t) * IP_NSTATS); 326 } 327 328 struct sockaddr_in ipaddr = { 329 .sin_len = sizeof(ipaddr), 330 .sin_family = AF_INET, 331 }; 332 333 static struct route ipforward_rt; 334 335 /* 336 * IP software interrupt routine 337 */ 338 void 339 ipintr(void) 340 { 341 int s; 342 struct mbuf *m; 343 struct ifqueue lcl_intrq; 344 345 memset(&lcl_intrq, 0, sizeof(lcl_intrq)); 346 347 mutex_enter(softnet_lock); 348 KERNEL_LOCK(1, NULL); 349 if (!IF_IS_EMPTY(&ipintrq)) { 350 s = splnet(); 351 352 /* Take existing queue onto stack */ 353 lcl_intrq = ipintrq; 354 355 /* Zero out global queue, preserving maxlen and drops */ 356 ipintrq.ifq_head = NULL; 357 ipintrq.ifq_tail = NULL; 358 ipintrq.ifq_len = 0; 359 ipintrq.ifq_maxlen = lcl_intrq.ifq_maxlen; 360 ipintrq.ifq_drops = lcl_intrq.ifq_drops; 361 362 splx(s); 363 } 364 KERNEL_UNLOCK_ONE(NULL); 365 while (!IF_IS_EMPTY(&lcl_intrq)) { 366 IF_DEQUEUE(&lcl_intrq, m); 367 if (m == NULL) 368 break; 369 ip_input(m); 370 } 371 mutex_exit(softnet_lock); 372 } 373 374 /* 375 * Ip input routine. Checksum and byte swap header. If fragmented 376 * try to reassemble. Process options. Pass to next level. 377 */ 378 static void 379 ip_input(struct mbuf *m) 380 { 381 struct ip *ip = NULL; 382 struct in_ifaddr *ia; 383 struct ifaddr *ifa; 384 int hlen = 0, len; 385 int downmatch; 386 int checkif; 387 int srcrt = 0; 388 ifnet_t *ifp; 389 390 MCLAIM(m, &ip_rx_mowner); 391 KASSERT((m->m_flags & M_PKTHDR) != 0); 392 ifp = m->m_pkthdr.rcvif; 393 394 /* 395 * If no IP addresses have been set yet but the interfaces 396 * are receiving, can't do anything with incoming packets yet. 397 */ 398 if (TAILQ_FIRST(&in_ifaddrhead) == 0) 399 goto bad; 400 IP_STATINC(IP_STAT_TOTAL); 401 /* 402 * If the IP header is not aligned, slurp it up into a new 403 * mbuf with space for link headers, in the event we forward 404 * it. Otherwise, if it is aligned, make sure the entire 405 * base IP header is in the first mbuf of the chain. 406 */ 407 if (IP_HDR_ALIGNED_P(mtod(m, void *)) == 0) { 408 if ((m = m_copyup(m, sizeof(struct ip), 409 (max_linkhdr + 3) & ~3)) == NULL) { 410 /* XXXJRT new stat, please */ 411 IP_STATINC(IP_STAT_TOOSMALL); 412 return; 413 } 414 } else if (__predict_false(m->m_len < sizeof (struct ip))) { 415 if ((m = m_pullup(m, sizeof (struct ip))) == NULL) { 416 IP_STATINC(IP_STAT_TOOSMALL); 417 return; 418 } 419 } 420 ip = mtod(m, struct ip *); 421 if (ip->ip_v != IPVERSION) { 422 IP_STATINC(IP_STAT_BADVERS); 423 goto bad; 424 } 425 hlen = ip->ip_hl << 2; 426 if (hlen < sizeof(struct ip)) { /* minimum header length */ 427 IP_STATINC(IP_STAT_BADHLEN); 428 goto bad; 429 } 430 if (hlen > m->m_len) { 431 if ((m = m_pullup(m, hlen)) == NULL) { 432 IP_STATINC(IP_STAT_BADHLEN); 433 return; 434 } 435 ip = mtod(m, struct ip *); 436 } 437 438 /* 439 * RFC1122: packets with a multicast source address are 440 * not allowed. 441 */ 442 if (IN_MULTICAST(ip->ip_src.s_addr)) { 443 IP_STATINC(IP_STAT_BADADDR); 444 goto bad; 445 } 446 447 /* 127/8 must not appear on wire - RFC1122 */ 448 if ((ntohl(ip->ip_dst.s_addr) >> IN_CLASSA_NSHIFT) == IN_LOOPBACKNET || 449 (ntohl(ip->ip_src.s_addr) >> IN_CLASSA_NSHIFT) == IN_LOOPBACKNET) { 450 if ((ifp->if_flags & IFF_LOOPBACK) == 0) { 451 IP_STATINC(IP_STAT_BADADDR); 452 goto bad; 453 } 454 } 455 456 switch (m->m_pkthdr.csum_flags & 457 ((ifp->if_csum_flags_rx & M_CSUM_IPv4) | 458 M_CSUM_IPv4_BAD)) { 459 case M_CSUM_IPv4|M_CSUM_IPv4_BAD: 460 INET_CSUM_COUNTER_INCR(&ip_hwcsum_bad); 461 goto badcsum; 462 463 case M_CSUM_IPv4: 464 /* Checksum was okay. */ 465 INET_CSUM_COUNTER_INCR(&ip_hwcsum_ok); 466 break; 467 468 default: 469 /* 470 * Must compute it ourselves. Maybe skip checksum on 471 * loopback interfaces. 472 */ 473 if (__predict_true(!(ifp->if_flags & IFF_LOOPBACK) || 474 ip_do_loopback_cksum)) { 475 INET_CSUM_COUNTER_INCR(&ip_swcsum); 476 if (in_cksum(m, hlen) != 0) 477 goto badcsum; 478 } 479 break; 480 } 481 482 /* Retrieve the packet length. */ 483 len = ntohs(ip->ip_len); 484 485 /* 486 * Check for additional length bogosity 487 */ 488 if (len < hlen) { 489 IP_STATINC(IP_STAT_BADLEN); 490 goto bad; 491 } 492 493 /* 494 * Check that the amount of data in the buffers 495 * is as at least much as the IP header would have us expect. 496 * Trim mbufs if longer than we expect. 497 * Drop packet if shorter than we expect. 498 */ 499 if (m->m_pkthdr.len < len) { 500 IP_STATINC(IP_STAT_TOOSHORT); 501 goto bad; 502 } 503 if (m->m_pkthdr.len > len) { 504 if (m->m_len == m->m_pkthdr.len) { 505 m->m_len = len; 506 m->m_pkthdr.len = len; 507 } else 508 m_adj(m, len - m->m_pkthdr.len); 509 } 510 511 /* 512 * Assume that we can create a fast-forward IP flow entry 513 * based on this packet. 514 */ 515 m->m_flags |= M_CANFASTFWD; 516 517 /* 518 * Run through list of hooks for input packets. If there are any 519 * filters which require that additional packets in the flow are 520 * not fast-forwarded, they must clear the M_CANFASTFWD flag. 521 * Note that filters must _never_ set this flag, as another filter 522 * in the list may have previously cleared it. 523 */ 524 #if defined(IPSEC) 525 if (!ipsec_used || !ipsec_indone(m)) 526 #else 527 if (1) 528 #endif 529 { 530 struct in_addr odst = ip->ip_dst; 531 532 if (pfil_run_hooks(inet_pfil_hook, &m, ifp, PFIL_IN) != 0) 533 return; 534 if (m == NULL) 535 return; 536 ip = mtod(m, struct ip *); 537 hlen = ip->ip_hl << 2; 538 /* 539 * XXX The setting of "srcrt" here is to prevent ip_forward() 540 * from generating ICMP redirects for packets that have 541 * been redirected by a hook back out on to the same LAN that 542 * they came from and is not an indication that the packet 543 * is being inffluenced by source routing options. This 544 * allows things like 545 * "rdr tlp0 0/0 port 80 -> 1.1.1.200 3128 tcp" 546 * where tlp0 is both on the 1.1.1.0/24 network and is the 547 * default route for hosts on 1.1.1.0/24. Of course this 548 * also requires a "map tlp0 ..." to complete the story. 549 * One might argue whether or not this kind of network config. 550 * should be supported in this manner... 551 */ 552 srcrt = (odst.s_addr != ip->ip_dst.s_addr); 553 } 554 555 #ifdef ALTQ 556 /* XXX Temporary until ALTQ is changed to use a pfil hook */ 557 if (altq_input != NULL && (*altq_input)(m, AF_INET) == 0) { 558 /* packet dropped by traffic conditioner */ 559 return; 560 } 561 #endif 562 563 /* 564 * Process options and, if not destined for us, 565 * ship it on. ip_dooptions returns 1 when an 566 * error was detected (causing an icmp message 567 * to be sent and the original packet to be freed). 568 */ 569 ip_nhops = 0; /* for source routed packets */ 570 if (hlen > sizeof (struct ip) && ip_dooptions(m)) 571 return; 572 573 /* 574 * Enable a consistency check between the destination address 575 * and the arrival interface for a unicast packet (the RFC 1122 576 * strong ES model) if IP forwarding is disabled and the packet 577 * is not locally generated. 578 * 579 * XXX - Checking also should be disabled if the destination 580 * address is ipnat'ed to a different interface. 581 * 582 * XXX - Checking is incompatible with IP aliases added 583 * to the loopback interface instead of the interface where 584 * the packets are received. 585 * 586 * XXX - We need to add a per ifaddr flag for this so that 587 * we get finer grain control. 588 */ 589 checkif = ip_checkinterface && (ipforwarding == 0) && 590 (ifp->if_flags & IFF_LOOPBACK) == 0; 591 592 /* 593 * Check our list of addresses, to see if the packet is for us. 594 * 595 * Traditional 4.4BSD did not consult IFF_UP at all. 596 * The behavior here is to treat addresses on !IFF_UP interface 597 * as not mine. 598 */ 599 downmatch = 0; 600 LIST_FOREACH(ia, &IN_IFADDR_HASH(ip->ip_dst.s_addr), ia_hash) { 601 if (in_hosteq(ia->ia_addr.sin_addr, ip->ip_dst)) { 602 if (checkif && ia->ia_ifp != ifp) 603 continue; 604 if ((ia->ia_ifp->if_flags & IFF_UP) != 0) 605 break; 606 else 607 downmatch++; 608 } 609 } 610 if (ia != NULL) 611 goto ours; 612 if (ifp->if_flags & IFF_BROADCAST) { 613 IFADDR_FOREACH(ifa, ifp) { 614 if (ifa->ifa_addr->sa_family != AF_INET) 615 continue; 616 ia = ifatoia(ifa); 617 if (in_hosteq(ip->ip_dst, ia->ia_broadaddr.sin_addr) || 618 in_hosteq(ip->ip_dst, ia->ia_netbroadcast) || 619 /* 620 * Look for all-0's host part (old broadcast addr), 621 * either for subnet or net. 622 */ 623 ip->ip_dst.s_addr == ia->ia_subnet || 624 ip->ip_dst.s_addr == ia->ia_net) 625 goto ours; 626 /* 627 * An interface with IP address zero accepts 628 * all packets that arrive on that interface. 629 */ 630 if (in_nullhost(ia->ia_addr.sin_addr)) 631 goto ours; 632 } 633 } 634 if (IN_MULTICAST(ip->ip_dst.s_addr)) { 635 #ifdef MROUTING 636 extern struct socket *ip_mrouter; 637 638 if (ip_mrouter) { 639 /* 640 * If we are acting as a multicast router, all 641 * incoming multicast packets are passed to the 642 * kernel-level multicast forwarding function. 643 * The packet is returned (relatively) intact; if 644 * ip_mforward() returns a non-zero value, the packet 645 * must be discarded, else it may be accepted below. 646 * 647 * (The IP ident field is put in the same byte order 648 * as expected when ip_mforward() is called from 649 * ip_output().) 650 */ 651 if (ip_mforward(m, ifp) != 0) { 652 IP_STATINC(IP_STAT_CANTFORWARD); 653 m_freem(m); 654 return; 655 } 656 657 /* 658 * The process-level routing demon needs to receive 659 * all multicast IGMP packets, whether or not this 660 * host belongs to their destination groups. 661 */ 662 if (ip->ip_p == IPPROTO_IGMP) 663 goto ours; 664 IP_STATINC(IP_STAT_CANTFORWARD); 665 } 666 #endif 667 /* 668 * See if we belong to the destination multicast group on the 669 * arrival interface. 670 */ 671 if (!in_multi_group(ip->ip_dst, ifp, 0)) { 672 IP_STATINC(IP_STAT_CANTFORWARD); 673 m_freem(m); 674 return; 675 } 676 goto ours; 677 } 678 if (ip->ip_dst.s_addr == INADDR_BROADCAST || 679 in_nullhost(ip->ip_dst)) 680 goto ours; 681 682 /* 683 * Not for us; forward if possible and desirable. 684 */ 685 if (ipforwarding == 0) { 686 IP_STATINC(IP_STAT_CANTFORWARD); 687 m_freem(m); 688 } else { 689 /* 690 * If ip_dst matched any of my address on !IFF_UP interface, 691 * and there's no IFF_UP interface that matches ip_dst, 692 * send icmp unreach. Forwarding it will result in in-kernel 693 * forwarding loop till TTL goes to 0. 694 */ 695 if (downmatch) { 696 icmp_error(m, ICMP_UNREACH, ICMP_UNREACH_HOST, 0, 0); 697 IP_STATINC(IP_STAT_CANTFORWARD); 698 return; 699 } 700 #ifdef IPSEC 701 /* Perform IPsec, if any. */ 702 if (ipsec_used && 703 ipsec4_input(m, IP_FORWARDING | (ip_directedbcast ? 704 IP_ALLOWBROADCAST : 0)) != 0) { 705 goto bad; 706 } 707 #endif 708 ip_forward(m, srcrt); 709 } 710 return; 711 712 ours: 713 /* 714 * If offset or IP_MF are set, must reassemble. 715 */ 716 if (ip->ip_off & ~htons(IP_DF|IP_RF)) { 717 /* 718 * Pass to IP reassembly mechanism. 719 */ 720 if (ip_reass_packet(&m, ip) != 0) { 721 /* Failed; invalid fragment(s) or packet. */ 722 goto bad; 723 } 724 if (m == NULL) { 725 /* More fragments should come; silently return. */ 726 return; 727 } 728 /* 729 * Reassembly is done, we have the final packet. 730 * Updated cached data in local variable(s). 731 */ 732 ip = mtod(m, struct ip *); 733 hlen = ip->ip_hl << 2; 734 } 735 736 #ifdef IPSEC 737 /* 738 * Enforce IPsec policy checking if we are seeing last header. 739 * Note that we do not visit this with protocols with PCB layer 740 * code - like UDP/TCP/raw IP. 741 */ 742 if (ipsec_used && 743 (inetsw[ip_protox[ip->ip_p]].pr_flags & PR_LASTHDR) != 0) { 744 if (ipsec4_input(m, 0) != 0) { 745 goto bad; 746 } 747 } 748 #endif 749 750 /* 751 * Switch out to protocol's input routine. 752 */ 753 #if IFA_STATS 754 if (ia && ip) 755 ia->ia_ifa.ifa_data.ifad_inbytes += ntohs(ip->ip_len); 756 #endif 757 IP_STATINC(IP_STAT_DELIVERED); 758 { 759 int off = hlen, nh = ip->ip_p; 760 761 (*inetsw[ip_protox[nh]].pr_input)(m, off, nh); 762 return; 763 } 764 bad: 765 m_freem(m); 766 return; 767 768 badcsum: 769 IP_STATINC(IP_STAT_BADSUM); 770 m_freem(m); 771 } 772 773 /* 774 * IP timer processing. 775 */ 776 void 777 ip_slowtimo(void) 778 { 779 780 mutex_enter(softnet_lock); 781 KERNEL_LOCK(1, NULL); 782 783 ip_reass_slowtimo(); 784 785 KERNEL_UNLOCK_ONE(NULL); 786 mutex_exit(softnet_lock); 787 } 788 789 /* 790 * IP drain processing. 791 */ 792 void 793 ip_drain(void) 794 { 795 796 KERNEL_LOCK(1, NULL); 797 ip_reass_drain(); 798 KERNEL_UNLOCK_ONE(NULL); 799 } 800 801 /* 802 * ip_dooptions: perform option processing on a datagram, possibly discarding 803 * it if bad options are encountered, or forwarding it if source-routed. 804 * 805 * => Returns true if packet has been forwarded/freed. 806 * => Returns false if the packet should be processed further. 807 */ 808 static bool 809 ip_dooptions(struct mbuf *m) 810 { 811 struct ip *ip = mtod(m, struct ip *); 812 u_char *cp, *cp0; 813 struct ip_timestamp *ipt; 814 struct in_ifaddr *ia; 815 int opt, optlen, cnt, off, code, type = ICMP_PARAMPROB, forward = 0; 816 struct in_addr dst; 817 n_time ntime; 818 819 dst = ip->ip_dst; 820 cp = (u_char *)(ip + 1); 821 cnt = (ip->ip_hl << 2) - sizeof (struct ip); 822 for (; cnt > 0; cnt -= optlen, cp += optlen) { 823 opt = cp[IPOPT_OPTVAL]; 824 if (opt == IPOPT_EOL) 825 break; 826 if (opt == IPOPT_NOP) 827 optlen = 1; 828 else { 829 if (cnt < IPOPT_OLEN + sizeof(*cp)) { 830 code = &cp[IPOPT_OLEN] - (u_char *)ip; 831 goto bad; 832 } 833 optlen = cp[IPOPT_OLEN]; 834 if (optlen < IPOPT_OLEN + sizeof(*cp) || optlen > cnt) { 835 code = &cp[IPOPT_OLEN] - (u_char *)ip; 836 goto bad; 837 } 838 } 839 switch (opt) { 840 841 default: 842 break; 843 844 /* 845 * Source routing with record. 846 * Find interface with current destination address. 847 * If none on this machine then drop if strictly routed, 848 * or do nothing if loosely routed. 849 * Record interface address and bring up next address 850 * component. If strictly routed make sure next 851 * address is on directly accessible net. 852 */ 853 case IPOPT_LSRR: 854 case IPOPT_SSRR: 855 if (ip_allowsrcrt == 0) { 856 type = ICMP_UNREACH; 857 code = ICMP_UNREACH_NET_PROHIB; 858 goto bad; 859 } 860 if (optlen < IPOPT_OFFSET + sizeof(*cp)) { 861 code = &cp[IPOPT_OLEN] - (u_char *)ip; 862 goto bad; 863 } 864 if ((off = cp[IPOPT_OFFSET]) < IPOPT_MINOFF) { 865 code = &cp[IPOPT_OFFSET] - (u_char *)ip; 866 goto bad; 867 } 868 ipaddr.sin_addr = ip->ip_dst; 869 ia = ifatoia(ifa_ifwithaddr(sintosa(&ipaddr))); 870 if (ia == 0) { 871 if (opt == IPOPT_SSRR) { 872 type = ICMP_UNREACH; 873 code = ICMP_UNREACH_SRCFAIL; 874 goto bad; 875 } 876 /* 877 * Loose routing, and not at next destination 878 * yet; nothing to do except forward. 879 */ 880 break; 881 } 882 off--; /* 0 origin */ 883 if ((off + sizeof(struct in_addr)) > optlen) { 884 /* 885 * End of source route. Should be for us. 886 */ 887 save_rte(cp, ip->ip_src); 888 break; 889 } 890 /* 891 * locate outgoing interface 892 */ 893 memcpy((void *)&ipaddr.sin_addr, (void *)(cp + off), 894 sizeof(ipaddr.sin_addr)); 895 if (opt == IPOPT_SSRR) 896 ia = ifatoia(ifa_ifwithladdr(sintosa(&ipaddr))); 897 else 898 ia = ip_rtaddr(ipaddr.sin_addr); 899 if (ia == 0) { 900 type = ICMP_UNREACH; 901 code = ICMP_UNREACH_SRCFAIL; 902 goto bad; 903 } 904 ip->ip_dst = ipaddr.sin_addr; 905 bcopy((void *)&ia->ia_addr.sin_addr, 906 (void *)(cp + off), sizeof(struct in_addr)); 907 cp[IPOPT_OFFSET] += sizeof(struct in_addr); 908 /* 909 * Let ip_intr's mcast routing check handle mcast pkts 910 */ 911 forward = !IN_MULTICAST(ip->ip_dst.s_addr); 912 break; 913 914 case IPOPT_RR: 915 if (optlen < IPOPT_OFFSET + sizeof(*cp)) { 916 code = &cp[IPOPT_OLEN] - (u_char *)ip; 917 goto bad; 918 } 919 if ((off = cp[IPOPT_OFFSET]) < IPOPT_MINOFF) { 920 code = &cp[IPOPT_OFFSET] - (u_char *)ip; 921 goto bad; 922 } 923 /* 924 * If no space remains, ignore. 925 */ 926 off--; /* 0 origin */ 927 if ((off + sizeof(struct in_addr)) > optlen) 928 break; 929 memcpy((void *)&ipaddr.sin_addr, (void *)(&ip->ip_dst), 930 sizeof(ipaddr.sin_addr)); 931 /* 932 * locate outgoing interface; if we're the destination, 933 * use the incoming interface (should be same). 934 */ 935 if ((ia = ifatoia(ifa_ifwithaddr(sintosa(&ipaddr)))) 936 == NULL && 937 (ia = ip_rtaddr(ipaddr.sin_addr)) == NULL) { 938 type = ICMP_UNREACH; 939 code = ICMP_UNREACH_HOST; 940 goto bad; 941 } 942 bcopy((void *)&ia->ia_addr.sin_addr, 943 (void *)(cp + off), sizeof(struct in_addr)); 944 cp[IPOPT_OFFSET] += sizeof(struct in_addr); 945 break; 946 947 case IPOPT_TS: 948 code = cp - (u_char *)ip; 949 ipt = (struct ip_timestamp *)cp; 950 if (ipt->ipt_len < 4 || ipt->ipt_len > 40) { 951 code = (u_char *)&ipt->ipt_len - (u_char *)ip; 952 goto bad; 953 } 954 if (ipt->ipt_ptr < 5) { 955 code = (u_char *)&ipt->ipt_ptr - (u_char *)ip; 956 goto bad; 957 } 958 if (ipt->ipt_ptr > ipt->ipt_len - sizeof (int32_t)) { 959 if (++ipt->ipt_oflw == 0) { 960 code = (u_char *)&ipt->ipt_ptr - 961 (u_char *)ip; 962 goto bad; 963 } 964 break; 965 } 966 cp0 = (cp + ipt->ipt_ptr - 1); 967 switch (ipt->ipt_flg) { 968 969 case IPOPT_TS_TSONLY: 970 break; 971 972 case IPOPT_TS_TSANDADDR: 973 if (ipt->ipt_ptr - 1 + sizeof(n_time) + 974 sizeof(struct in_addr) > ipt->ipt_len) { 975 code = (u_char *)&ipt->ipt_ptr - 976 (u_char *)ip; 977 goto bad; 978 } 979 ipaddr.sin_addr = dst; 980 ia = ifatoia(ifaof_ifpforaddr(sintosa(&ipaddr), 981 m->m_pkthdr.rcvif)); 982 if (ia == 0) 983 continue; 984 bcopy(&ia->ia_addr.sin_addr, 985 cp0, sizeof(struct in_addr)); 986 ipt->ipt_ptr += sizeof(struct in_addr); 987 break; 988 989 case IPOPT_TS_PRESPEC: 990 if (ipt->ipt_ptr - 1 + sizeof(n_time) + 991 sizeof(struct in_addr) > ipt->ipt_len) { 992 code = (u_char *)&ipt->ipt_ptr - 993 (u_char *)ip; 994 goto bad; 995 } 996 memcpy(&ipaddr.sin_addr, cp0, 997 sizeof(struct in_addr)); 998 if (ifatoia(ifa_ifwithaddr(sintosa(&ipaddr))) 999 == NULL) 1000 continue; 1001 ipt->ipt_ptr += sizeof(struct in_addr); 1002 break; 1003 1004 default: 1005 /* XXX can't take &ipt->ipt_flg */ 1006 code = (u_char *)&ipt->ipt_ptr - 1007 (u_char *)ip + 1; 1008 goto bad; 1009 } 1010 ntime = iptime(); 1011 cp0 = (u_char *) &ntime; /* XXX grumble, GCC... */ 1012 memmove((char *)cp + ipt->ipt_ptr - 1, cp0, 1013 sizeof(n_time)); 1014 ipt->ipt_ptr += sizeof(n_time); 1015 } 1016 } 1017 if (forward) { 1018 if (ip_forwsrcrt == 0) { 1019 type = ICMP_UNREACH; 1020 code = ICMP_UNREACH_SRCFAIL; 1021 goto bad; 1022 } 1023 ip_forward(m, 1); 1024 return true; 1025 } 1026 return false; 1027 bad: 1028 icmp_error(m, type, code, 0, 0); 1029 IP_STATINC(IP_STAT_BADOPTIONS); 1030 return true; 1031 } 1032 1033 /* 1034 * ip_rtaddr: given address of next destination (final or next hop), 1035 * return internet address info of interface to be used to get there. 1036 */ 1037 static struct in_ifaddr * 1038 ip_rtaddr(struct in_addr dst) 1039 { 1040 struct rtentry *rt; 1041 union { 1042 struct sockaddr dst; 1043 struct sockaddr_in dst4; 1044 } u; 1045 1046 sockaddr_in_init(&u.dst4, &dst, 0); 1047 1048 if ((rt = rtcache_lookup(&ipforward_rt, &u.dst)) == NULL) 1049 return NULL; 1050 1051 return ifatoia(rt->rt_ifa); 1052 } 1053 1054 /* 1055 * save_rte: save incoming source route for use in replies, to be picked 1056 * up later by ip_srcroute if the receiver is interested. 1057 */ 1058 static void 1059 save_rte(u_char *option, struct in_addr dst) 1060 { 1061 unsigned olen; 1062 1063 olen = option[IPOPT_OLEN]; 1064 if (olen > sizeof(ip_srcrt) - (1 + sizeof(dst))) 1065 return; 1066 memcpy((void *)ip_srcrt.srcopt, (void *)option, olen); 1067 ip_nhops = (olen - IPOPT_OFFSET - 1) / sizeof(struct in_addr); 1068 ip_srcrt.dst = dst; 1069 } 1070 1071 /* 1072 * Retrieve incoming source route for use in replies, 1073 * in the same form used by setsockopt. 1074 * The first hop is placed before the options, will be removed later. 1075 */ 1076 struct mbuf * 1077 ip_srcroute(void) 1078 { 1079 struct in_addr *p, *q; 1080 struct mbuf *m; 1081 1082 if (ip_nhops == 0) 1083 return NULL; 1084 m = m_get(M_DONTWAIT, MT_SOOPTS); 1085 if (m == 0) 1086 return NULL; 1087 1088 MCLAIM(m, &inetdomain.dom_mowner); 1089 #define OPTSIZ (sizeof(ip_srcrt.nop) + sizeof(ip_srcrt.srcopt)) 1090 1091 /* length is (nhops+1)*sizeof(addr) + sizeof(nop + srcrt header) */ 1092 m->m_len = ip_nhops * sizeof(struct in_addr) + sizeof(struct in_addr) + 1093 OPTSIZ; 1094 1095 /* 1096 * First save first hop for return route 1097 */ 1098 p = &ip_srcrt.route[ip_nhops - 1]; 1099 *(mtod(m, struct in_addr *)) = *p--; 1100 1101 /* 1102 * Copy option fields and padding (nop) to mbuf. 1103 */ 1104 ip_srcrt.nop = IPOPT_NOP; 1105 ip_srcrt.srcopt[IPOPT_OFFSET] = IPOPT_MINOFF; 1106 memmove(mtod(m, char *) + sizeof(struct in_addr), &ip_srcrt.nop, 1107 OPTSIZ); 1108 q = (struct in_addr *)(mtod(m, char *) + 1109 sizeof(struct in_addr) + OPTSIZ); 1110 #undef OPTSIZ 1111 /* 1112 * Record return path as an IP source route, 1113 * reversing the path (pointers are now aligned). 1114 */ 1115 while (p >= ip_srcrt.route) { 1116 *q++ = *p--; 1117 } 1118 /* 1119 * Last hop goes to final destination. 1120 */ 1121 *q = ip_srcrt.dst; 1122 return (m); 1123 } 1124 1125 const int inetctlerrmap[PRC_NCMDS] = { 1126 [PRC_MSGSIZE] = EMSGSIZE, 1127 [PRC_HOSTDEAD] = EHOSTDOWN, 1128 [PRC_HOSTUNREACH] = EHOSTUNREACH, 1129 [PRC_UNREACH_NET] = EHOSTUNREACH, 1130 [PRC_UNREACH_HOST] = EHOSTUNREACH, 1131 [PRC_UNREACH_PROTOCOL] = ECONNREFUSED, 1132 [PRC_UNREACH_PORT] = ECONNREFUSED, 1133 [PRC_UNREACH_SRCFAIL] = EHOSTUNREACH, 1134 [PRC_PARAMPROB] = ENOPROTOOPT, 1135 }; 1136 1137 void 1138 ip_fasttimo(void) 1139 { 1140 if (ip_drainwanted) { 1141 ip_drain(); 1142 ip_drainwanted = 0; 1143 } 1144 } 1145 1146 void 1147 ip_drainstub(void) 1148 { 1149 ip_drainwanted = 1; 1150 } 1151 1152 /* 1153 * Forward a packet. If some error occurs return the sender 1154 * an icmp packet. Note we can't always generate a meaningful 1155 * icmp message because icmp doesn't have a large enough repertoire 1156 * of codes and types. 1157 * 1158 * If not forwarding, just drop the packet. This could be confusing 1159 * if ipforwarding was zero but some routing protocol was advancing 1160 * us as a gateway to somewhere. However, we must let the routing 1161 * protocol deal with that. 1162 * 1163 * The srcrt parameter indicates whether the packet is being forwarded 1164 * via a source route. 1165 */ 1166 static void 1167 ip_forward(struct mbuf *m, int srcrt) 1168 { 1169 struct ip *ip = mtod(m, struct ip *); 1170 struct rtentry *rt; 1171 int error, type = 0, code = 0, destmtu = 0; 1172 struct mbuf *mcopy; 1173 n_long dest; 1174 union { 1175 struct sockaddr dst; 1176 struct sockaddr_in dst4; 1177 } u; 1178 1179 /* 1180 * We are now in the output path. 1181 */ 1182 MCLAIM(m, &ip_tx_mowner); 1183 1184 /* 1185 * Clear any in-bound checksum flags for this packet. 1186 */ 1187 m->m_pkthdr.csum_flags = 0; 1188 1189 dest = 0; 1190 if (m->m_flags & (M_BCAST|M_MCAST) || in_canforward(ip->ip_dst) == 0) { 1191 IP_STATINC(IP_STAT_CANTFORWARD); 1192 m_freem(m); 1193 return; 1194 } 1195 if (ip->ip_ttl <= IPTTLDEC) { 1196 icmp_error(m, ICMP_TIMXCEED, ICMP_TIMXCEED_INTRANS, dest, 0); 1197 return; 1198 } 1199 1200 sockaddr_in_init(&u.dst4, &ip->ip_dst, 0); 1201 if ((rt = rtcache_lookup(&ipforward_rt, &u.dst)) == NULL) { 1202 icmp_error(m, ICMP_UNREACH, ICMP_UNREACH_NET, dest, 0); 1203 return; 1204 } 1205 1206 /* 1207 * Save at most 68 bytes of the packet in case 1208 * we need to generate an ICMP message to the src. 1209 * Pullup to avoid sharing mbuf cluster between m and mcopy. 1210 */ 1211 mcopy = m_copym(m, 0, imin(ntohs(ip->ip_len), 68), M_DONTWAIT); 1212 if (mcopy) 1213 mcopy = m_pullup(mcopy, ip->ip_hl << 2); 1214 1215 ip->ip_ttl -= IPTTLDEC; 1216 1217 /* 1218 * If forwarding packet using same interface that it came in on, 1219 * perhaps should send a redirect to sender to shortcut a hop. 1220 * Only send redirect if source is sending directly to us, 1221 * and if packet was not source routed (or has any options). 1222 * Also, don't send redirect if forwarding using a default route 1223 * or a route modified by a redirect. 1224 */ 1225 if (rt->rt_ifp == m->m_pkthdr.rcvif && 1226 (rt->rt_flags & (RTF_DYNAMIC|RTF_MODIFIED)) == 0 && 1227 !in_nullhost(satocsin(rt_getkey(rt))->sin_addr) && 1228 ipsendredirects && !srcrt) { 1229 if (rt->rt_ifa && 1230 (ip->ip_src.s_addr & ifatoia(rt->rt_ifa)->ia_subnetmask) == 1231 ifatoia(rt->rt_ifa)->ia_subnet) { 1232 if (rt->rt_flags & RTF_GATEWAY) 1233 dest = satosin(rt->rt_gateway)->sin_addr.s_addr; 1234 else 1235 dest = ip->ip_dst.s_addr; 1236 /* 1237 * Router requirements says to only send host 1238 * redirects. 1239 */ 1240 type = ICMP_REDIRECT; 1241 code = ICMP_REDIRECT_HOST; 1242 } 1243 } 1244 1245 error = ip_output(m, NULL, &ipforward_rt, 1246 (IP_FORWARDING | (ip_directedbcast ? IP_ALLOWBROADCAST : 0)), 1247 NULL, NULL); 1248 1249 if (error) 1250 IP_STATINC(IP_STAT_CANTFORWARD); 1251 else { 1252 uint64_t *ips = IP_STAT_GETREF(); 1253 ips[IP_STAT_FORWARD]++; 1254 if (type) { 1255 ips[IP_STAT_REDIRECTSENT]++; 1256 IP_STAT_PUTREF(); 1257 } else { 1258 IP_STAT_PUTREF(); 1259 if (mcopy) { 1260 #ifdef GATEWAY 1261 if (mcopy->m_flags & M_CANFASTFWD) 1262 ipflow_create(&ipforward_rt, mcopy); 1263 #endif 1264 m_freem(mcopy); 1265 } 1266 return; 1267 } 1268 } 1269 if (mcopy == NULL) 1270 return; 1271 1272 switch (error) { 1273 1274 case 0: /* forwarded, but need redirect */ 1275 /* type, code set above */ 1276 break; 1277 1278 case ENETUNREACH: /* shouldn't happen, checked above */ 1279 case EHOSTUNREACH: 1280 case ENETDOWN: 1281 case EHOSTDOWN: 1282 default: 1283 type = ICMP_UNREACH; 1284 code = ICMP_UNREACH_HOST; 1285 break; 1286 1287 case EMSGSIZE: 1288 type = ICMP_UNREACH; 1289 code = ICMP_UNREACH_NEEDFRAG; 1290 1291 if ((rt = rtcache_validate(&ipforward_rt)) != NULL) 1292 destmtu = rt->rt_ifp->if_mtu; 1293 #ifdef IPSEC 1294 if (ipsec_used) 1295 (void)ipsec4_forward(mcopy, &destmtu); 1296 #endif 1297 IP_STATINC(IP_STAT_CANTFRAG); 1298 break; 1299 1300 case ENOBUFS: 1301 /* 1302 * Do not generate ICMP_SOURCEQUENCH as required in RFC 1812, 1303 * Requirements for IP Version 4 Routers. Source quench can 1304 * big problem under DoS attacks or if the underlying 1305 * interface is rate-limited. 1306 */ 1307 if (mcopy) 1308 m_freem(mcopy); 1309 return; 1310 } 1311 icmp_error(mcopy, type, code, dest, destmtu); 1312 } 1313 1314 void 1315 ip_savecontrol(struct inpcb *inp, struct mbuf **mp, struct ip *ip, 1316 struct mbuf *m) 1317 { 1318 struct socket *so = inp->inp_socket; 1319 ifnet_t *ifp = m->m_pkthdr.rcvif; 1320 int inpflags = inp->inp_flags; 1321 1322 if (so->so_options & SO_TIMESTAMP 1323 #ifdef SO_OTIMESTAMP 1324 || so->so_options & SO_OTIMESTAMP 1325 #endif 1326 ) { 1327 struct timeval tv; 1328 1329 microtime(&tv); 1330 #ifdef SO_OTIMESTAMP 1331 if (so->so_options & SO_OTIMESTAMP) { 1332 struct timeval50 tv50; 1333 timeval_to_timeval50(&tv, &tv50); 1334 *mp = sbcreatecontrol((void *) &tv50, sizeof(tv50), 1335 SCM_OTIMESTAMP, SOL_SOCKET); 1336 } else 1337 #endif 1338 *mp = sbcreatecontrol((void *) &tv, sizeof(tv), 1339 SCM_TIMESTAMP, SOL_SOCKET); 1340 if (*mp) 1341 mp = &(*mp)->m_next; 1342 } 1343 if (inpflags & INP_RECVDSTADDR) { 1344 *mp = sbcreatecontrol((void *) &ip->ip_dst, 1345 sizeof(struct in_addr), IP_RECVDSTADDR, IPPROTO_IP); 1346 if (*mp) 1347 mp = &(*mp)->m_next; 1348 } 1349 if (inpflags & INP_RECVPKTINFO) { 1350 struct in_pktinfo ipi; 1351 ipi.ipi_addr = ip->ip_src; 1352 ipi.ipi_ifindex = ifp->if_index; 1353 *mp = sbcreatecontrol((void *) &ipi, 1354 sizeof(ipi), IP_RECVPKTINFO, IPPROTO_IP); 1355 if (*mp) 1356 mp = &(*mp)->m_next; 1357 } 1358 if (inpflags & INP_PKTINFO) { 1359 struct in_pktinfo ipi; 1360 ipi.ipi_addr = ip->ip_dst; 1361 ipi.ipi_ifindex = ifp->if_index; 1362 *mp = sbcreatecontrol((void *) &ipi, 1363 sizeof(ipi), IP_PKTINFO, IPPROTO_IP); 1364 if (*mp) 1365 mp = &(*mp)->m_next; 1366 } 1367 if (inpflags & INP_RECVIF) { 1368 struct sockaddr_dl sdl; 1369 1370 sockaddr_dl_init(&sdl, sizeof(sdl), ifp ? 1371 ifp->if_index : 0, 0, NULL, 0, NULL, 0); 1372 *mp = sbcreatecontrol(&sdl, sdl.sdl_len, IP_RECVIF, IPPROTO_IP); 1373 if (*mp) 1374 mp = &(*mp)->m_next; 1375 } 1376 if (inpflags & INP_RECVTTL) { 1377 *mp = sbcreatecontrol((void *) &ip->ip_ttl, 1378 sizeof(uint8_t), IP_RECVTTL, IPPROTO_IP); 1379 if (*mp) 1380 mp = &(*mp)->m_next; 1381 } 1382 } 1383 1384 /* 1385 * sysctl helper routine for net.inet.ip.forwsrcrt. 1386 */ 1387 static int 1388 sysctl_net_inet_ip_forwsrcrt(SYSCTLFN_ARGS) 1389 { 1390 int error, tmp; 1391 struct sysctlnode node; 1392 1393 node = *rnode; 1394 tmp = ip_forwsrcrt; 1395 node.sysctl_data = &tmp; 1396 error = sysctl_lookup(SYSCTLFN_CALL(&node)); 1397 if (error || newp == NULL) 1398 return (error); 1399 1400 error = kauth_authorize_network(l->l_cred, KAUTH_NETWORK_FORWSRCRT, 1401 0, NULL, NULL, NULL); 1402 if (error) 1403 return (error); 1404 1405 ip_forwsrcrt = tmp; 1406 1407 return (0); 1408 } 1409 1410 /* 1411 * sysctl helper routine for net.inet.ip.mtudisctimeout. checks the 1412 * range of the new value and tweaks timers if it changes. 1413 */ 1414 static int 1415 sysctl_net_inet_ip_pmtudto(SYSCTLFN_ARGS) 1416 { 1417 int error, tmp; 1418 struct sysctlnode node; 1419 1420 node = *rnode; 1421 tmp = ip_mtudisc_timeout; 1422 node.sysctl_data = &tmp; 1423 error = sysctl_lookup(SYSCTLFN_CALL(&node)); 1424 if (error || newp == NULL) 1425 return (error); 1426 if (tmp < 0) 1427 return (EINVAL); 1428 1429 mutex_enter(softnet_lock); 1430 1431 ip_mtudisc_timeout = tmp; 1432 rt_timer_queue_change(ip_mtudisc_timeout_q, ip_mtudisc_timeout); 1433 1434 mutex_exit(softnet_lock); 1435 1436 return (0); 1437 } 1438 1439 static int 1440 sysctl_net_inet_ip_stats(SYSCTLFN_ARGS) 1441 { 1442 1443 return (NETSTAT_SYSCTL(ipstat_percpu, IP_NSTATS)); 1444 } 1445 1446 static void 1447 sysctl_net_inet_ip_setup(struct sysctllog **clog) 1448 { 1449 sysctl_createv(clog, 0, NULL, NULL, 1450 CTLFLAG_PERMANENT, 1451 CTLTYPE_NODE, "inet", 1452 SYSCTL_DESCR("PF_INET related settings"), 1453 NULL, 0, NULL, 0, 1454 CTL_NET, PF_INET, CTL_EOL); 1455 sysctl_createv(clog, 0, NULL, NULL, 1456 CTLFLAG_PERMANENT, 1457 CTLTYPE_NODE, "ip", 1458 SYSCTL_DESCR("IPv4 related settings"), 1459 NULL, 0, NULL, 0, 1460 CTL_NET, PF_INET, IPPROTO_IP, CTL_EOL); 1461 1462 sysctl_createv(clog, 0, NULL, NULL, 1463 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1464 CTLTYPE_INT, "forwarding", 1465 SYSCTL_DESCR("Enable forwarding of INET datagrams"), 1466 NULL, 0, &ipforwarding, 0, 1467 CTL_NET, PF_INET, IPPROTO_IP, 1468 IPCTL_FORWARDING, CTL_EOL); 1469 sysctl_createv(clog, 0, NULL, NULL, 1470 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1471 CTLTYPE_INT, "redirect", 1472 SYSCTL_DESCR("Enable sending of ICMP redirect messages"), 1473 NULL, 0, &ipsendredirects, 0, 1474 CTL_NET, PF_INET, IPPROTO_IP, 1475 IPCTL_SENDREDIRECTS, CTL_EOL); 1476 sysctl_createv(clog, 0, NULL, NULL, 1477 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1478 CTLTYPE_INT, "ttl", 1479 SYSCTL_DESCR("Default TTL for an INET datagram"), 1480 NULL, 0, &ip_defttl, 0, 1481 CTL_NET, PF_INET, IPPROTO_IP, 1482 IPCTL_DEFTTL, CTL_EOL); 1483 #ifdef IPCTL_DEFMTU 1484 sysctl_createv(clog, 0, NULL, NULL, 1485 CTLFLAG_PERMANENT /* |CTLFLAG_READWRITE? */, 1486 CTLTYPE_INT, "mtu", 1487 SYSCTL_DESCR("Default MTA for an INET route"), 1488 NULL, 0, &ip_mtu, 0, 1489 CTL_NET, PF_INET, IPPROTO_IP, 1490 IPCTL_DEFMTU, CTL_EOL); 1491 #endif /* IPCTL_DEFMTU */ 1492 sysctl_createv(clog, 0, NULL, NULL, 1493 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1494 CTLTYPE_INT, "forwsrcrt", 1495 SYSCTL_DESCR("Enable forwarding of source-routed " 1496 "datagrams"), 1497 sysctl_net_inet_ip_forwsrcrt, 0, &ip_forwsrcrt, 0, 1498 CTL_NET, PF_INET, IPPROTO_IP, 1499 IPCTL_FORWSRCRT, CTL_EOL); 1500 sysctl_createv(clog, 0, NULL, NULL, 1501 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1502 CTLTYPE_INT, "directed-broadcast", 1503 SYSCTL_DESCR("Enable forwarding of broadcast datagrams"), 1504 NULL, 0, &ip_directedbcast, 0, 1505 CTL_NET, PF_INET, IPPROTO_IP, 1506 IPCTL_DIRECTEDBCAST, CTL_EOL); 1507 sysctl_createv(clog, 0, NULL, NULL, 1508 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1509 CTLTYPE_INT, "allowsrcrt", 1510 SYSCTL_DESCR("Accept source-routed datagrams"), 1511 NULL, 0, &ip_allowsrcrt, 0, 1512 CTL_NET, PF_INET, IPPROTO_IP, 1513 IPCTL_ALLOWSRCRT, CTL_EOL); 1514 1515 sysctl_createv(clog, 0, NULL, NULL, 1516 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1517 CTLTYPE_INT, "mtudisc", 1518 SYSCTL_DESCR("Use RFC1191 Path MTU Discovery"), 1519 NULL, 0, &ip_mtudisc, 0, 1520 CTL_NET, PF_INET, IPPROTO_IP, 1521 IPCTL_MTUDISC, CTL_EOL); 1522 sysctl_createv(clog, 0, NULL, NULL, 1523 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1524 CTLTYPE_INT, "anonportmin", 1525 SYSCTL_DESCR("Lowest ephemeral port number to assign"), 1526 sysctl_net_inet_ip_ports, 0, &anonportmin, 0, 1527 CTL_NET, PF_INET, IPPROTO_IP, 1528 IPCTL_ANONPORTMIN, CTL_EOL); 1529 sysctl_createv(clog, 0, NULL, NULL, 1530 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1531 CTLTYPE_INT, "anonportmax", 1532 SYSCTL_DESCR("Highest ephemeral port number to assign"), 1533 sysctl_net_inet_ip_ports, 0, &anonportmax, 0, 1534 CTL_NET, PF_INET, IPPROTO_IP, 1535 IPCTL_ANONPORTMAX, CTL_EOL); 1536 sysctl_createv(clog, 0, NULL, NULL, 1537 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1538 CTLTYPE_INT, "mtudisctimeout", 1539 SYSCTL_DESCR("Lifetime of a Path MTU Discovered route"), 1540 sysctl_net_inet_ip_pmtudto, 0, (void *)&ip_mtudisc_timeout, 0, 1541 CTL_NET, PF_INET, IPPROTO_IP, 1542 IPCTL_MTUDISCTIMEOUT, CTL_EOL); 1543 #if NGIF > 0 1544 sysctl_createv(clog, 0, NULL, NULL, 1545 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1546 CTLTYPE_INT, "gifttl", 1547 SYSCTL_DESCR("Default TTL for a gif tunnel datagram"), 1548 NULL, 0, &ip_gif_ttl, 0, 1549 CTL_NET, PF_INET, IPPROTO_IP, 1550 IPCTL_GIF_TTL, CTL_EOL); 1551 #endif /* NGIF */ 1552 #ifndef IPNOPRIVPORTS 1553 sysctl_createv(clog, 0, NULL, NULL, 1554 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1555 CTLTYPE_INT, "lowportmin", 1556 SYSCTL_DESCR("Lowest privileged ephemeral port number " 1557 "to assign"), 1558 sysctl_net_inet_ip_ports, 0, &lowportmin, 0, 1559 CTL_NET, PF_INET, IPPROTO_IP, 1560 IPCTL_LOWPORTMIN, CTL_EOL); 1561 sysctl_createv(clog, 0, NULL, NULL, 1562 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1563 CTLTYPE_INT, "lowportmax", 1564 SYSCTL_DESCR("Highest privileged ephemeral port number " 1565 "to assign"), 1566 sysctl_net_inet_ip_ports, 0, &lowportmax, 0, 1567 CTL_NET, PF_INET, IPPROTO_IP, 1568 IPCTL_LOWPORTMAX, CTL_EOL); 1569 #endif /* IPNOPRIVPORTS */ 1570 #if NGRE > 0 1571 sysctl_createv(clog, 0, NULL, NULL, 1572 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1573 CTLTYPE_INT, "grettl", 1574 SYSCTL_DESCR("Default TTL for a gre tunnel datagram"), 1575 NULL, 0, &ip_gre_ttl, 0, 1576 CTL_NET, PF_INET, IPPROTO_IP, 1577 IPCTL_GRE_TTL, CTL_EOL); 1578 #endif /* NGRE */ 1579 sysctl_createv(clog, 0, NULL, NULL, 1580 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1581 CTLTYPE_INT, "checkinterface", 1582 SYSCTL_DESCR("Enable receive side of Strong ES model " 1583 "from RFC1122"), 1584 NULL, 0, &ip_checkinterface, 0, 1585 CTL_NET, PF_INET, IPPROTO_IP, 1586 IPCTL_CHECKINTERFACE, CTL_EOL); 1587 sysctl_createv(clog, 0, NULL, NULL, 1588 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1589 CTLTYPE_INT, "random_id", 1590 SYSCTL_DESCR("Assign random ip_id values"), 1591 NULL, 0, &ip_do_randomid, 0, 1592 CTL_NET, PF_INET, IPPROTO_IP, 1593 IPCTL_RANDOMID, CTL_EOL); 1594 sysctl_createv(clog, 0, NULL, NULL, 1595 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1596 CTLTYPE_INT, "do_loopback_cksum", 1597 SYSCTL_DESCR("Perform IP checksum on loopback"), 1598 NULL, 0, &ip_do_loopback_cksum, 0, 1599 CTL_NET, PF_INET, IPPROTO_IP, 1600 IPCTL_LOOPBACKCKSUM, CTL_EOL); 1601 sysctl_createv(clog, 0, NULL, NULL, 1602 CTLFLAG_PERMANENT, 1603 CTLTYPE_STRUCT, "stats", 1604 SYSCTL_DESCR("IP statistics"), 1605 sysctl_net_inet_ip_stats, 0, NULL, 0, 1606 CTL_NET, PF_INET, IPPROTO_IP, IPCTL_STATS, 1607 CTL_EOL); 1608 1609 /* anonportalgo RFC6056 subtree */ 1610 const struct sysctlnode *portalgo_node; 1611 sysctl_createv(clog, 0, NULL, &portalgo_node, 1612 CTLFLAG_PERMANENT, 1613 CTLTYPE_NODE, "anonportalgo", 1614 SYSCTL_DESCR("Anonymous Port Algorithm Selection (RFC 6056)"), 1615 NULL, 0, NULL, 0, 1616 CTL_NET, PF_INET, IPPROTO_IP, CTL_CREATE, CTL_EOL); 1617 sysctl_createv(clog, 0, &portalgo_node, NULL, 1618 CTLFLAG_PERMANENT, 1619 CTLTYPE_STRING, "available", 1620 SYSCTL_DESCR("available algorithms"), 1621 sysctl_portalgo_available, 0, NULL, PORTALGO_MAXLEN, 1622 CTL_CREATE, CTL_EOL); 1623 sysctl_createv(clog, 0, &portalgo_node, NULL, 1624 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1625 CTLTYPE_STRING, "selected", 1626 SYSCTL_DESCR("selected algorithm"), 1627 sysctl_portalgo_selected4, 0, NULL, PORTALGO_MAXLEN, 1628 CTL_CREATE, CTL_EOL); 1629 sysctl_createv(clog, 0, &portalgo_node, NULL, 1630 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1631 CTLTYPE_STRUCT, "reserve", 1632 SYSCTL_DESCR("bitmap of reserved ports"), 1633 sysctl_portalgo_reserve4, 0, NULL, 0, 1634 CTL_CREATE, CTL_EOL); 1635 } 1636 1637 void 1638 ip_statinc(u_int stat) 1639 { 1640 1641 KASSERT(stat < IP_NSTATS); 1642 IP_STATINC(stat); 1643 } 1644