1 /* $NetBSD: ip_input.c,v 1.326 2016/01/08 03:55:39 knakahara 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 1.326 2016/01/08 03:55:39 knakahara Exp $"); 95 96 #ifdef _KERNEL_OPT 97 #include "opt_inet.h" 98 #include "opt_compat_netbsd.h" 99 #include "opt_gateway.h" 100 #include "opt_ipsec.h" 101 #include "opt_mrouting.h" 102 #include "opt_mbuftrace.h" 103 #include "opt_inet_csum.h" 104 #endif 105 106 #include "arp.h" 107 108 #include <sys/param.h> 109 #include <sys/systm.h> 110 #include <sys/cpu.h> 111 #include <sys/mbuf.h> 112 #include <sys/domain.h> 113 #include <sys/protosw.h> 114 #include <sys/socket.h> 115 #include <sys/socketvar.h> 116 #include <sys/errno.h> 117 #include <sys/time.h> 118 #include <sys/kernel.h> 119 #include <sys/pool.h> 120 #include <sys/sysctl.h> 121 #include <sys/kauth.h> 122 123 #include <net/if.h> 124 #include <net/if_dl.h> 125 #include <net/route.h> 126 #include <net/pktqueue.h> 127 #include <net/pfil.h> 128 129 #include <netinet/in.h> 130 #include <netinet/in_systm.h> 131 #include <netinet/ip.h> 132 #include <netinet/in_pcb.h> 133 #include <netinet/in_proto.h> 134 #include <netinet/in_var.h> 135 #include <netinet/ip_var.h> 136 #include <netinet/ip_private.h> 137 #include <netinet/ip_icmp.h> 138 /* just for gif_ttl */ 139 #include <netinet/in_gif.h> 140 #include "gif.h" 141 #include <net/if_gre.h> 142 #include "gre.h" 143 144 #ifdef MROUTING 145 #include <netinet/ip_mroute.h> 146 #endif 147 #include <netinet/portalgo.h> 148 149 #ifdef IPSEC 150 #include <netipsec/ipsec.h> 151 #endif 152 153 #ifndef IPFORWARDING 154 #ifdef GATEWAY 155 #define IPFORWARDING 1 /* forward IP packets not for us */ 156 #else /* GATEWAY */ 157 #define IPFORWARDING 0 /* don't forward IP packets not for us */ 158 #endif /* GATEWAY */ 159 #endif /* IPFORWARDING */ 160 #ifndef IPSENDREDIRECTS 161 #define IPSENDREDIRECTS 1 162 #endif 163 #ifndef IPFORWSRCRT 164 #define IPFORWSRCRT 1 /* forward source-routed packets */ 165 #endif 166 #ifndef IPALLOWSRCRT 167 #define IPALLOWSRCRT 1 /* allow source-routed packets */ 168 #endif 169 #ifndef IPMTUDISC 170 #define IPMTUDISC 1 171 #endif 172 #ifndef IPMTUDISCTIMEOUT 173 #define IPMTUDISCTIMEOUT (10 * 60) /* as per RFC 1191 */ 174 #endif 175 176 #ifdef COMPAT_50 177 #include <compat/sys/time.h> 178 #include <compat/sys/socket.h> 179 #endif 180 181 /* 182 * Note: DIRECTED_BROADCAST is handled this way so that previous 183 * configuration using this option will Just Work. 184 */ 185 #ifndef IPDIRECTEDBCAST 186 #ifdef DIRECTED_BROADCAST 187 #define IPDIRECTEDBCAST 1 188 #else 189 #define IPDIRECTEDBCAST 0 190 #endif /* DIRECTED_BROADCAST */ 191 #endif /* IPDIRECTEDBCAST */ 192 int ipforwarding = IPFORWARDING; 193 int ipsendredirects = IPSENDREDIRECTS; 194 int ip_defttl = IPDEFTTL; 195 int ip_forwsrcrt = IPFORWSRCRT; 196 int ip_directedbcast = IPDIRECTEDBCAST; 197 int ip_allowsrcrt = IPALLOWSRCRT; 198 int ip_mtudisc = IPMTUDISC; 199 int ip_mtudisc_timeout = IPMTUDISCTIMEOUT; 200 #ifdef DIAGNOSTIC 201 int ipprintfs = 0; 202 #endif 203 204 int ip_do_randomid = 0; 205 206 /* 207 * XXX - Setting ip_checkinterface mostly implements the receive side of 208 * the Strong ES model described in RFC 1122, but since the routing table 209 * and transmit implementation do not implement the Strong ES model, 210 * setting this to 1 results in an odd hybrid. 211 * 212 * XXX - ip_checkinterface currently must be disabled if you use ipnat 213 * to translate the destination address to another local interface. 214 * 215 * XXX - ip_checkinterface must be disabled if you add IP aliases 216 * to the loopback interface instead of the interface where the 217 * packets for those addresses are received. 218 */ 219 static int ip_checkinterface __read_mostly = 0; 220 221 struct rttimer_queue *ip_mtudisc_timeout_q = NULL; 222 223 pktqueue_t * ip_pktq __read_mostly; 224 pfil_head_t * inet_pfil_hook __read_mostly; 225 ipid_state_t * ip_ids __read_mostly; 226 percpu_t * ipstat_percpu __read_mostly; 227 228 static struct route ipforward_rt __cacheline_aligned; 229 230 uint16_t ip_id; 231 232 #ifdef INET_CSUM_COUNTERS 233 #include <sys/device.h> 234 235 struct evcnt ip_hwcsum_bad = EVCNT_INITIALIZER(EVCNT_TYPE_MISC, 236 NULL, "inet", "hwcsum bad"); 237 struct evcnt ip_hwcsum_ok = EVCNT_INITIALIZER(EVCNT_TYPE_MISC, 238 NULL, "inet", "hwcsum ok"); 239 struct evcnt ip_swcsum = EVCNT_INITIALIZER(EVCNT_TYPE_MISC, 240 NULL, "inet", "swcsum"); 241 242 #define INET_CSUM_COUNTER_INCR(ev) (ev)->ev_count++ 243 244 EVCNT_ATTACH_STATIC(ip_hwcsum_bad); 245 EVCNT_ATTACH_STATIC(ip_hwcsum_ok); 246 EVCNT_ATTACH_STATIC(ip_swcsum); 247 248 #else 249 250 #define INET_CSUM_COUNTER_INCR(ev) /* nothing */ 251 252 #endif /* INET_CSUM_COUNTERS */ 253 254 /* 255 * We need to save the IP options in case a protocol wants to respond 256 * to an incoming packet over the same route if the packet got here 257 * using IP source routing. This allows connection establishment and 258 * maintenance when the remote end is on a network that is not known 259 * to us. 260 */ 261 262 static int ip_nhops = 0; 263 264 static struct ip_srcrt { 265 struct in_addr dst; /* final destination */ 266 char nop; /* one NOP to align */ 267 char srcopt[IPOPT_OFFSET + 1]; /* OPTVAL, OLEN and OFFSET */ 268 struct in_addr route[MAX_IPOPTLEN/sizeof(struct in_addr)]; 269 } ip_srcrt; 270 271 static int ip_drainwanted; 272 273 struct sockaddr_in ipaddr = { 274 .sin_len = sizeof(ipaddr), 275 .sin_family = AF_INET, 276 }; 277 278 static void save_rte(u_char *, struct in_addr); 279 280 #ifdef MBUFTRACE 281 struct mowner ip_rx_mowner = MOWNER_INIT("internet", "rx"); 282 struct mowner ip_tx_mowner = MOWNER_INIT("internet", "tx"); 283 #endif 284 285 static void ipintr(void *); 286 static void ip_input(struct mbuf *); 287 static void ip_forward(struct mbuf *, int); 288 static bool ip_dooptions(struct mbuf *); 289 static struct in_ifaddr *ip_rtaddr(struct in_addr); 290 static void sysctl_net_inet_ip_setup(struct sysctllog **); 291 292 /* XXX: Not yet enabled. */ 293 #define SOFTNET_LOCK() KASSERT(mutex_owned(softnet_lock)) 294 #define SOFTNET_UNLOCK() KASSERT(mutex_owned(softnet_lock)) 295 296 /* 297 * IP initialization: fill in IP protocol switch table. 298 * All protocols not implemented in kernel go to raw IP protocol handler. 299 */ 300 void 301 ip_init(void) 302 { 303 const struct protosw *pr; 304 305 in_init(); 306 sysctl_net_inet_ip_setup(NULL); 307 308 pr = pffindproto(PF_INET, IPPROTO_RAW, SOCK_RAW); 309 KASSERT(pr != NULL); 310 311 ip_pktq = pktq_create(IFQ_MAXLEN, ipintr, NULL); 312 KASSERT(ip_pktq != NULL); 313 314 for (u_int i = 0; i < IPPROTO_MAX; i++) { 315 ip_protox[i] = pr - inetsw; 316 } 317 for (pr = inetdomain.dom_protosw; 318 pr < inetdomain.dom_protoswNPROTOSW; pr++) 319 if (pr->pr_domain->dom_family == PF_INET && 320 pr->pr_protocol && pr->pr_protocol != IPPROTO_RAW) 321 ip_protox[pr->pr_protocol] = pr - inetsw; 322 323 ip_reass_init(); 324 325 ip_ids = ip_id_init(); 326 ip_id = time_uptime & 0xfffff; 327 328 ip_mtudisc_timeout_q = rt_timer_queue_create(ip_mtudisc_timeout); 329 #ifdef GATEWAY 330 ipflow_init(); 331 #endif 332 333 /* Register our Packet Filter hook. */ 334 inet_pfil_hook = pfil_head_create(PFIL_TYPE_AF, (void *)AF_INET); 335 KASSERT(inet_pfil_hook != NULL); 336 337 #ifdef MBUFTRACE 338 MOWNER_ATTACH(&ip_tx_mowner); 339 MOWNER_ATTACH(&ip_rx_mowner); 340 #endif /* MBUFTRACE */ 341 342 ipstat_percpu = percpu_alloc(sizeof(uint64_t) * IP_NSTATS); 343 } 344 345 /* 346 * IP software interrupt routine. 347 */ 348 static void 349 ipintr(void *arg __unused) 350 { 351 struct mbuf *m; 352 353 KASSERT(cpu_softintr_p()); 354 355 mutex_enter(softnet_lock); 356 while ((m = pktq_dequeue(ip_pktq)) != NULL) { 357 ip_input(m); 358 } 359 mutex_exit(softnet_lock); 360 } 361 362 /* 363 * IP input routine. Checksum and byte swap header. If fragmented 364 * try to reassemble. Process options. Pass to next level. 365 */ 366 static void 367 ip_input(struct mbuf *m) 368 { 369 struct ip *ip = NULL; 370 struct in_ifaddr *ia; 371 struct ifaddr *ifa; 372 int hlen = 0, len; 373 int downmatch; 374 int checkif; 375 int srcrt = 0; 376 ifnet_t *ifp; 377 378 KASSERTMSG(cpu_softintr_p(), "ip_input: not in the software " 379 "interrupt handler; synchronization assumptions violated"); 380 381 MCLAIM(m, &ip_rx_mowner); 382 KASSERT((m->m_flags & M_PKTHDR) != 0); 383 ifp = m->m_pkthdr.rcvif; 384 385 /* 386 * If no IP addresses have been set yet but the interfaces 387 * are receiving, can't do anything with incoming packets yet. 388 * Note: we pre-check without locks held. 389 */ 390 if (!TAILQ_FIRST(&in_ifaddrhead)) { 391 goto bad; 392 } 393 IP_STATINC(IP_STAT_TOTAL); 394 395 /* 396 * If the IP header is not aligned, slurp it up into a new 397 * mbuf with space for link headers, in the event we forward 398 * it. Otherwise, if it is aligned, make sure the entire 399 * base IP header is in the first mbuf of the chain. 400 */ 401 if (IP_HDR_ALIGNED_P(mtod(m, void *)) == 0) { 402 if ((m = m_copyup(m, sizeof(struct ip), 403 (max_linkhdr + 3) & ~3)) == NULL) { 404 /* XXXJRT new stat, please */ 405 IP_STATINC(IP_STAT_TOOSMALL); 406 return; 407 } 408 } else if (__predict_false(m->m_len < sizeof (struct ip))) { 409 if ((m = m_pullup(m, sizeof (struct ip))) == NULL) { 410 IP_STATINC(IP_STAT_TOOSMALL); 411 return; 412 } 413 } 414 ip = mtod(m, struct ip *); 415 if (ip->ip_v != IPVERSION) { 416 IP_STATINC(IP_STAT_BADVERS); 417 goto bad; 418 } 419 hlen = ip->ip_hl << 2; 420 if (hlen < sizeof(struct ip)) { /* minimum header length */ 421 IP_STATINC(IP_STAT_BADHLEN); 422 goto bad; 423 } 424 if (hlen > m->m_len) { 425 if ((m = m_pullup(m, hlen)) == NULL) { 426 IP_STATINC(IP_STAT_BADHLEN); 427 return; 428 } 429 ip = mtod(m, struct ip *); 430 } 431 432 /* 433 * RFC1122: packets with a multicast source address are 434 * not allowed. 435 */ 436 if (IN_MULTICAST(ip->ip_src.s_addr)) { 437 IP_STATINC(IP_STAT_BADADDR); 438 goto bad; 439 } 440 441 /* 127/8 must not appear on wire - RFC1122 */ 442 if ((ntohl(ip->ip_dst.s_addr) >> IN_CLASSA_NSHIFT) == IN_LOOPBACKNET || 443 (ntohl(ip->ip_src.s_addr) >> IN_CLASSA_NSHIFT) == IN_LOOPBACKNET) { 444 if ((ifp->if_flags & IFF_LOOPBACK) == 0) { 445 IP_STATINC(IP_STAT_BADADDR); 446 goto bad; 447 } 448 } 449 450 switch (m->m_pkthdr.csum_flags & 451 ((ifp->if_csum_flags_rx & M_CSUM_IPv4) | 452 M_CSUM_IPv4_BAD)) { 453 case M_CSUM_IPv4|M_CSUM_IPv4_BAD: 454 INET_CSUM_COUNTER_INCR(&ip_hwcsum_bad); 455 goto badcsum; 456 457 case M_CSUM_IPv4: 458 /* Checksum was okay. */ 459 INET_CSUM_COUNTER_INCR(&ip_hwcsum_ok); 460 break; 461 462 default: 463 /* 464 * Must compute it ourselves. Maybe skip checksum on 465 * loopback interfaces. 466 */ 467 if (__predict_true(!(ifp->if_flags & IFF_LOOPBACK) || 468 ip_do_loopback_cksum)) { 469 INET_CSUM_COUNTER_INCR(&ip_swcsum); 470 if (in_cksum(m, hlen) != 0) 471 goto badcsum; 472 } 473 break; 474 } 475 476 /* Retrieve the packet length. */ 477 len = ntohs(ip->ip_len); 478 479 /* 480 * Check for additional length bogosity 481 */ 482 if (len < hlen) { 483 IP_STATINC(IP_STAT_BADLEN); 484 goto bad; 485 } 486 487 /* 488 * Check that the amount of data in the buffers 489 * is as at least much as the IP header would have us expect. 490 * Trim mbufs if longer than we expect. 491 * Drop packet if shorter than we expect. 492 */ 493 if (m->m_pkthdr.len < len) { 494 IP_STATINC(IP_STAT_TOOSHORT); 495 goto bad; 496 } 497 if (m->m_pkthdr.len > len) { 498 if (m->m_len == m->m_pkthdr.len) { 499 m->m_len = len; 500 m->m_pkthdr.len = len; 501 } else 502 m_adj(m, len - m->m_pkthdr.len); 503 } 504 505 /* 506 * Assume that we can create a fast-forward IP flow entry 507 * based on this packet. 508 */ 509 m->m_flags |= M_CANFASTFWD; 510 511 /* 512 * Run through list of hooks for input packets. If there are any 513 * filters which require that additional packets in the flow are 514 * not fast-forwarded, they must clear the M_CANFASTFWD flag. 515 * Note that filters must _never_ set this flag, as another filter 516 * in the list may have previously cleared it. 517 */ 518 #if defined(IPSEC) 519 if (!ipsec_used || !ipsec_indone(m)) 520 #else 521 if (1) 522 #endif 523 { 524 struct in_addr odst = ip->ip_dst; 525 bool freed; 526 527 SOFTNET_LOCK(); 528 freed = pfil_run_hooks(inet_pfil_hook, &m, ifp, PFIL_IN) != 0; 529 SOFTNET_UNLOCK(); 530 if (freed || m == NULL) { 531 return; 532 } 533 ip = mtod(m, struct ip *); 534 hlen = ip->ip_hl << 2; 535 536 /* 537 * XXX The setting of "srcrt" here is to prevent ip_forward() 538 * from generating ICMP redirects for packets that have 539 * been redirected by a hook back out on to the same LAN that 540 * they came from and is not an indication that the packet 541 * is being inffluenced by source routing options. This 542 * allows things like 543 * "rdr tlp0 0/0 port 80 -> 1.1.1.200 3128 tcp" 544 * where tlp0 is both on the 1.1.1.0/24 network and is the 545 * default route for hosts on 1.1.1.0/24. Of course this 546 * also requires a "map tlp0 ..." to complete the story. 547 * One might argue whether or not this kind of network config. 548 * should be supported in this manner... 549 */ 550 srcrt = (odst.s_addr != ip->ip_dst.s_addr); 551 } 552 553 #ifdef ALTQ 554 /* XXX Temporary until ALTQ is changed to use a pfil hook */ 555 if (altq_input) { 556 SOFTNET_LOCK(); 557 if ((*altq_input)(m, AF_INET) == 0) { 558 /* Packet dropped by traffic conditioner. */ 559 SOFTNET_UNLOCK(); 560 return; 561 } 562 SOFTNET_UNLOCK(); 563 } 564 #endif 565 566 /* 567 * Process options and, if not destined for us, 568 * ship it on. ip_dooptions returns 1 when an 569 * error was detected (causing an icmp message 570 * to be sent and the original packet to be freed). 571 */ 572 ip_nhops = 0; /* for source routed packets */ 573 if (hlen > sizeof (struct ip) && ip_dooptions(m)) 574 return; 575 576 /* 577 * Enable a consistency check between the destination address 578 * and the arrival interface for a unicast packet (the RFC 1122 579 * strong ES model) if IP forwarding is disabled and the packet 580 * is not locally generated. 581 * 582 * XXX - Checking also should be disabled if the destination 583 * address is ipnat'ed to a different interface. 584 * 585 * XXX - Checking is incompatible with IP aliases added 586 * to the loopback interface instead of the interface where 587 * the packets are received. 588 * 589 * XXX - We need to add a per ifaddr flag for this so that 590 * we get finer grain control. 591 */ 592 checkif = ip_checkinterface && (ipforwarding == 0) && 593 (ifp->if_flags & IFF_LOOPBACK) == 0; 594 595 /* 596 * Check our list of addresses, to see if the packet is for us. 597 * 598 * Traditional 4.4BSD did not consult IFF_UP at all. 599 * The behavior here is to treat addresses on !IFF_UP interface 600 * or IN_IFF_NOTREADY addresses as not mine. 601 */ 602 downmatch = 0; 603 LIST_FOREACH(ia, &IN_IFADDR_HASH(ip->ip_dst.s_addr), ia_hash) { 604 if (in_hosteq(ia->ia_addr.sin_addr, ip->ip_dst)) { 605 if (ia->ia4_flags & IN_IFF_NOTREADY) 606 continue; 607 if (checkif && ia->ia_ifp != ifp) 608 continue; 609 if ((ia->ia_ifp->if_flags & IFF_UP) != 0) 610 break; 611 else 612 downmatch++; 613 } 614 } 615 if (ia != NULL) 616 goto ours; 617 if (ifp->if_flags & IFF_BROADCAST) { 618 IFADDR_FOREACH(ifa, ifp) { 619 if (ifa->ifa_addr->sa_family != AF_INET) 620 continue; 621 ia = ifatoia(ifa); 622 if (ia->ia4_flags & IN_IFF_NOTREADY) 623 continue; 624 if (in_hosteq(ip->ip_dst, ia->ia_broadaddr.sin_addr) || 625 in_hosteq(ip->ip_dst, ia->ia_netbroadcast) || 626 /* 627 * Look for all-0's host part (old broadcast addr), 628 * either for subnet or net. 629 */ 630 ip->ip_dst.s_addr == ia->ia_subnet || 631 ip->ip_dst.s_addr == ia->ia_net) 632 goto ours; 633 /* 634 * An interface with IP address zero accepts 635 * all packets that arrive on that interface. 636 */ 637 if (in_nullhost(ia->ia_addr.sin_addr)) 638 goto ours; 639 } 640 } 641 if (IN_MULTICAST(ip->ip_dst.s_addr)) { 642 #ifdef MROUTING 643 extern struct socket *ip_mrouter; 644 645 if (ip_mrouter) { 646 /* 647 * If we are acting as a multicast router, all 648 * incoming multicast packets are passed to the 649 * kernel-level multicast forwarding function. 650 * The packet is returned (relatively) intact; if 651 * ip_mforward() returns a non-zero value, the packet 652 * must be discarded, else it may be accepted below. 653 * 654 * (The IP ident field is put in the same byte order 655 * as expected when ip_mforward() is called from 656 * ip_output().) 657 */ 658 SOFTNET_LOCK(); 659 if (ip_mforward(m, ifp) != 0) { 660 SOFTNET_UNLOCK(); 661 IP_STATINC(IP_STAT_CANTFORWARD); 662 m_freem(m); 663 return; 664 } 665 SOFTNET_UNLOCK(); 666 667 /* 668 * The process-level routing demon needs to receive 669 * all multicast IGMP packets, whether or not this 670 * host belongs to their destination groups. 671 */ 672 if (ip->ip_p == IPPROTO_IGMP) { 673 goto ours; 674 } 675 IP_STATINC(IP_STAT_CANTFORWARD); 676 } 677 #endif 678 /* 679 * See if we belong to the destination multicast group on the 680 * arrival interface. 681 */ 682 if (!in_multi_group(ip->ip_dst, ifp, 0)) { 683 IP_STATINC(IP_STAT_CANTFORWARD); 684 m_freem(m); 685 return; 686 } 687 goto ours; 688 } 689 if (ip->ip_dst.s_addr == INADDR_BROADCAST || 690 in_nullhost(ip->ip_dst)) 691 goto ours; 692 693 /* 694 * Not for us; forward if possible and desirable. 695 */ 696 if (ipforwarding == 0) { 697 IP_STATINC(IP_STAT_CANTFORWARD); 698 m_freem(m); 699 } else { 700 /* 701 * If ip_dst matched any of my address on !IFF_UP interface, 702 * and there's no IFF_UP interface that matches ip_dst, 703 * send icmp unreach. Forwarding it will result in in-kernel 704 * forwarding loop till TTL goes to 0. 705 */ 706 if (downmatch) { 707 icmp_error(m, ICMP_UNREACH, ICMP_UNREACH_HOST, 0, 0); 708 IP_STATINC(IP_STAT_CANTFORWARD); 709 return; 710 } 711 #ifdef IPSEC 712 /* Perform IPsec, if any. */ 713 if (ipsec_used) { 714 SOFTNET_LOCK(); 715 if (ipsec4_input(m, IP_FORWARDING | 716 (ip_directedbcast ? IP_ALLOWBROADCAST : 0)) != 0) { 717 SOFTNET_UNLOCK(); 718 goto bad; 719 } 720 SOFTNET_UNLOCK(); 721 } 722 #endif 723 ip_forward(m, srcrt); 724 } 725 return; 726 727 ours: 728 /* 729 * If offset or IP_MF are set, must reassemble. 730 */ 731 if (ip->ip_off & ~htons(IP_DF|IP_RF)) { 732 /* 733 * Pass to IP reassembly mechanism. 734 */ 735 if (ip_reass_packet(&m, ip) != 0) { 736 /* Failed; invalid fragment(s) or packet. */ 737 goto bad; 738 } 739 if (m == NULL) { 740 /* More fragments should come; silently return. */ 741 return; 742 } 743 /* 744 * Reassembly is done, we have the final packet. 745 * Updated cached data in local variable(s). 746 */ 747 ip = mtod(m, struct ip *); 748 hlen = ip->ip_hl << 2; 749 } 750 751 #ifdef IPSEC 752 /* 753 * Enforce IPsec policy checking if we are seeing last header. 754 * Note that we do not visit this with protocols with PCB layer 755 * code - like UDP/TCP/raw IP. 756 */ 757 if (ipsec_used && 758 (inetsw[ip_protox[ip->ip_p]].pr_flags & PR_LASTHDR) != 0) { 759 SOFTNET_LOCK(); 760 if (ipsec4_input(m, 0) != 0) { 761 SOFTNET_UNLOCK(); 762 goto bad; 763 } 764 SOFTNET_UNLOCK(); 765 } 766 #endif 767 768 /* 769 * Switch out to protocol's input routine. 770 */ 771 #if IFA_STATS 772 if (ia && ip) 773 ia->ia_ifa.ifa_data.ifad_inbytes += ntohs(ip->ip_len); 774 #endif 775 IP_STATINC(IP_STAT_DELIVERED); 776 777 const int off = hlen, nh = ip->ip_p; 778 779 SOFTNET_LOCK(); 780 (*inetsw[ip_protox[nh]].pr_input)(m, off, nh); 781 SOFTNET_UNLOCK(); 782 return; 783 bad: 784 m_freem(m); 785 return; 786 787 badcsum: 788 IP_STATINC(IP_STAT_BADSUM); 789 m_freem(m); 790 } 791 792 /* 793 * IP timer processing. 794 */ 795 void 796 ip_slowtimo(void) 797 { 798 799 mutex_enter(softnet_lock); 800 KERNEL_LOCK(1, NULL); 801 802 ip_reass_slowtimo(); 803 804 KERNEL_UNLOCK_ONE(NULL); 805 mutex_exit(softnet_lock); 806 } 807 808 /* 809 * IP drain processing. 810 */ 811 void 812 ip_drain(void) 813 { 814 815 KERNEL_LOCK(1, NULL); 816 ip_reass_drain(); 817 KERNEL_UNLOCK_ONE(NULL); 818 } 819 820 /* 821 * ip_dooptions: perform option processing on a datagram, possibly discarding 822 * it if bad options are encountered, or forwarding it if source-routed. 823 * 824 * => Returns true if packet has been forwarded/freed. 825 * => Returns false if the packet should be processed further. 826 */ 827 static bool 828 ip_dooptions(struct mbuf *m) 829 { 830 struct ip *ip = mtod(m, struct ip *); 831 u_char *cp, *cp0; 832 struct ip_timestamp *ipt; 833 struct in_ifaddr *ia; 834 int opt, optlen, cnt, off, code, type = ICMP_PARAMPROB, forward = 0; 835 struct in_addr dst; 836 n_time ntime; 837 838 dst = ip->ip_dst; 839 cp = (u_char *)(ip + 1); 840 cnt = (ip->ip_hl << 2) - sizeof (struct ip); 841 for (; cnt > 0; cnt -= optlen, cp += optlen) { 842 opt = cp[IPOPT_OPTVAL]; 843 if (opt == IPOPT_EOL) 844 break; 845 if (opt == IPOPT_NOP) 846 optlen = 1; 847 else { 848 if (cnt < IPOPT_OLEN + sizeof(*cp)) { 849 code = &cp[IPOPT_OLEN] - (u_char *)ip; 850 goto bad; 851 } 852 optlen = cp[IPOPT_OLEN]; 853 if (optlen < IPOPT_OLEN + sizeof(*cp) || optlen > cnt) { 854 code = &cp[IPOPT_OLEN] - (u_char *)ip; 855 goto bad; 856 } 857 } 858 switch (opt) { 859 860 default: 861 break; 862 863 /* 864 * Source routing with record. 865 * Find interface with current destination address. 866 * If none on this machine then drop if strictly routed, 867 * or do nothing if loosely routed. 868 * Record interface address and bring up next address 869 * component. If strictly routed make sure next 870 * address is on directly accessible net. 871 */ 872 case IPOPT_LSRR: 873 case IPOPT_SSRR: 874 if (ip_allowsrcrt == 0) { 875 type = ICMP_UNREACH; 876 code = ICMP_UNREACH_NET_PROHIB; 877 goto bad; 878 } 879 if (optlen < IPOPT_OFFSET + sizeof(*cp)) { 880 code = &cp[IPOPT_OLEN] - (u_char *)ip; 881 goto bad; 882 } 883 if ((off = cp[IPOPT_OFFSET]) < IPOPT_MINOFF) { 884 code = &cp[IPOPT_OFFSET] - (u_char *)ip; 885 goto bad; 886 } 887 ipaddr.sin_addr = ip->ip_dst; 888 ia = ifatoia(ifa_ifwithaddr(sintosa(&ipaddr))); 889 if (ia == 0) { 890 if (opt == IPOPT_SSRR) { 891 type = ICMP_UNREACH; 892 code = ICMP_UNREACH_SRCFAIL; 893 goto bad; 894 } 895 /* 896 * Loose routing, and not at next destination 897 * yet; nothing to do except forward. 898 */ 899 break; 900 } 901 off--; /* 0 origin */ 902 if ((off + sizeof(struct in_addr)) > optlen) { 903 /* 904 * End of source route. Should be for us. 905 */ 906 save_rte(cp, ip->ip_src); 907 break; 908 } 909 /* 910 * locate outgoing interface 911 */ 912 memcpy((void *)&ipaddr.sin_addr, (void *)(cp + off), 913 sizeof(ipaddr.sin_addr)); 914 if (opt == IPOPT_SSRR) 915 ia = ifatoia(ifa_ifwithladdr(sintosa(&ipaddr))); 916 else 917 ia = ip_rtaddr(ipaddr.sin_addr); 918 if (ia == 0) { 919 type = ICMP_UNREACH; 920 code = ICMP_UNREACH_SRCFAIL; 921 goto bad; 922 } 923 ip->ip_dst = ipaddr.sin_addr; 924 bcopy((void *)&ia->ia_addr.sin_addr, 925 (void *)(cp + off), sizeof(struct in_addr)); 926 cp[IPOPT_OFFSET] += sizeof(struct in_addr); 927 /* 928 * Let ip_intr's mcast routing check handle mcast pkts 929 */ 930 forward = !IN_MULTICAST(ip->ip_dst.s_addr); 931 break; 932 933 case IPOPT_RR: 934 if (optlen < IPOPT_OFFSET + sizeof(*cp)) { 935 code = &cp[IPOPT_OLEN] - (u_char *)ip; 936 goto bad; 937 } 938 if ((off = cp[IPOPT_OFFSET]) < IPOPT_MINOFF) { 939 code = &cp[IPOPT_OFFSET] - (u_char *)ip; 940 goto bad; 941 } 942 /* 943 * If no space remains, ignore. 944 */ 945 off--; /* 0 origin */ 946 if ((off + sizeof(struct in_addr)) > optlen) 947 break; 948 memcpy((void *)&ipaddr.sin_addr, (void *)(&ip->ip_dst), 949 sizeof(ipaddr.sin_addr)); 950 /* 951 * locate outgoing interface; if we're the destination, 952 * use the incoming interface (should be same). 953 */ 954 if ((ia = ifatoia(ifa_ifwithaddr(sintosa(&ipaddr)))) 955 == NULL && 956 (ia = ip_rtaddr(ipaddr.sin_addr)) == NULL) { 957 type = ICMP_UNREACH; 958 code = ICMP_UNREACH_HOST; 959 goto bad; 960 } 961 bcopy((void *)&ia->ia_addr.sin_addr, 962 (void *)(cp + off), sizeof(struct in_addr)); 963 cp[IPOPT_OFFSET] += sizeof(struct in_addr); 964 break; 965 966 case IPOPT_TS: 967 code = cp - (u_char *)ip; 968 ipt = (struct ip_timestamp *)cp; 969 if (ipt->ipt_len < 4 || ipt->ipt_len > 40) { 970 code = (u_char *)&ipt->ipt_len - (u_char *)ip; 971 goto bad; 972 } 973 if (ipt->ipt_ptr < 5) { 974 code = (u_char *)&ipt->ipt_ptr - (u_char *)ip; 975 goto bad; 976 } 977 if (ipt->ipt_ptr > ipt->ipt_len - sizeof (int32_t)) { 978 if (++ipt->ipt_oflw == 0) { 979 code = (u_char *)&ipt->ipt_ptr - 980 (u_char *)ip; 981 goto bad; 982 } 983 break; 984 } 985 cp0 = (cp + ipt->ipt_ptr - 1); 986 switch (ipt->ipt_flg) { 987 988 case IPOPT_TS_TSONLY: 989 break; 990 991 case IPOPT_TS_TSANDADDR: 992 if (ipt->ipt_ptr - 1 + sizeof(n_time) + 993 sizeof(struct in_addr) > ipt->ipt_len) { 994 code = (u_char *)&ipt->ipt_ptr - 995 (u_char *)ip; 996 goto bad; 997 } 998 ipaddr.sin_addr = dst; 999 ia = ifatoia(ifaof_ifpforaddr(sintosa(&ipaddr), 1000 m->m_pkthdr.rcvif)); 1001 if (ia == 0) 1002 continue; 1003 bcopy(&ia->ia_addr.sin_addr, 1004 cp0, sizeof(struct in_addr)); 1005 ipt->ipt_ptr += sizeof(struct in_addr); 1006 break; 1007 1008 case IPOPT_TS_PRESPEC: 1009 if (ipt->ipt_ptr - 1 + sizeof(n_time) + 1010 sizeof(struct in_addr) > ipt->ipt_len) { 1011 code = (u_char *)&ipt->ipt_ptr - 1012 (u_char *)ip; 1013 goto bad; 1014 } 1015 memcpy(&ipaddr.sin_addr, cp0, 1016 sizeof(struct in_addr)); 1017 if (ifatoia(ifa_ifwithaddr(sintosa(&ipaddr))) 1018 == NULL) 1019 continue; 1020 ipt->ipt_ptr += sizeof(struct in_addr); 1021 break; 1022 1023 default: 1024 /* XXX can't take &ipt->ipt_flg */ 1025 code = (u_char *)&ipt->ipt_ptr - 1026 (u_char *)ip + 1; 1027 goto bad; 1028 } 1029 ntime = iptime(); 1030 cp0 = (u_char *) &ntime; /* XXX grumble, GCC... */ 1031 memmove((char *)cp + ipt->ipt_ptr - 1, cp0, 1032 sizeof(n_time)); 1033 ipt->ipt_ptr += sizeof(n_time); 1034 } 1035 } 1036 if (forward) { 1037 if (ip_forwsrcrt == 0) { 1038 type = ICMP_UNREACH; 1039 code = ICMP_UNREACH_SRCFAIL; 1040 goto bad; 1041 } 1042 ip_forward(m, 1); 1043 return true; 1044 } 1045 return false; 1046 bad: 1047 icmp_error(m, type, code, 0, 0); 1048 IP_STATINC(IP_STAT_BADOPTIONS); 1049 return true; 1050 } 1051 1052 /* 1053 * ip_rtaddr: given address of next destination (final or next hop), 1054 * return internet address info of interface to be used to get there. 1055 */ 1056 static struct in_ifaddr * 1057 ip_rtaddr(struct in_addr dst) 1058 { 1059 struct rtentry *rt; 1060 union { 1061 struct sockaddr dst; 1062 struct sockaddr_in dst4; 1063 } u; 1064 1065 sockaddr_in_init(&u.dst4, &dst, 0); 1066 1067 SOFTNET_LOCK(); 1068 rt = rtcache_lookup(&ipforward_rt, &u.dst); 1069 SOFTNET_UNLOCK(); 1070 if (rt == NULL) 1071 return NULL; 1072 1073 return ifatoia(rt->rt_ifa); 1074 } 1075 1076 /* 1077 * save_rte: save incoming source route for use in replies, to be picked 1078 * up later by ip_srcroute if the receiver is interested. 1079 */ 1080 static void 1081 save_rte(u_char *option, struct in_addr dst) 1082 { 1083 unsigned olen; 1084 1085 olen = option[IPOPT_OLEN]; 1086 if (olen > sizeof(ip_srcrt) - (1 + sizeof(dst))) 1087 return; 1088 memcpy((void *)ip_srcrt.srcopt, (void *)option, olen); 1089 ip_nhops = (olen - IPOPT_OFFSET - 1) / sizeof(struct in_addr); 1090 ip_srcrt.dst = dst; 1091 } 1092 1093 /* 1094 * Retrieve incoming source route for use in replies, 1095 * in the same form used by setsockopt. 1096 * The first hop is placed before the options, will be removed later. 1097 */ 1098 struct mbuf * 1099 ip_srcroute(void) 1100 { 1101 struct in_addr *p, *q; 1102 struct mbuf *m; 1103 1104 if (ip_nhops == 0) 1105 return NULL; 1106 m = m_get(M_DONTWAIT, MT_SOOPTS); 1107 if (m == 0) 1108 return NULL; 1109 1110 MCLAIM(m, &inetdomain.dom_mowner); 1111 #define OPTSIZ (sizeof(ip_srcrt.nop) + sizeof(ip_srcrt.srcopt)) 1112 1113 /* length is (nhops+1)*sizeof(addr) + sizeof(nop + srcrt header) */ 1114 m->m_len = ip_nhops * sizeof(struct in_addr) + sizeof(struct in_addr) + 1115 OPTSIZ; 1116 1117 /* 1118 * First save first hop for return route 1119 */ 1120 p = &ip_srcrt.route[ip_nhops - 1]; 1121 *(mtod(m, struct in_addr *)) = *p--; 1122 1123 /* 1124 * Copy option fields and padding (nop) to mbuf. 1125 */ 1126 ip_srcrt.nop = IPOPT_NOP; 1127 ip_srcrt.srcopt[IPOPT_OFFSET] = IPOPT_MINOFF; 1128 memmove(mtod(m, char *) + sizeof(struct in_addr), &ip_srcrt.nop, 1129 OPTSIZ); 1130 q = (struct in_addr *)(mtod(m, char *) + 1131 sizeof(struct in_addr) + OPTSIZ); 1132 #undef OPTSIZ 1133 /* 1134 * Record return path as an IP source route, 1135 * reversing the path (pointers are now aligned). 1136 */ 1137 while (p >= ip_srcrt.route) { 1138 *q++ = *p--; 1139 } 1140 /* 1141 * Last hop goes to final destination. 1142 */ 1143 *q = ip_srcrt.dst; 1144 return (m); 1145 } 1146 1147 const int inetctlerrmap[PRC_NCMDS] = { 1148 [PRC_MSGSIZE] = EMSGSIZE, 1149 [PRC_HOSTDEAD] = EHOSTDOWN, 1150 [PRC_HOSTUNREACH] = EHOSTUNREACH, 1151 [PRC_UNREACH_NET] = EHOSTUNREACH, 1152 [PRC_UNREACH_HOST] = EHOSTUNREACH, 1153 [PRC_UNREACH_PROTOCOL] = ECONNREFUSED, 1154 [PRC_UNREACH_PORT] = ECONNREFUSED, 1155 [PRC_UNREACH_SRCFAIL] = EHOSTUNREACH, 1156 [PRC_PARAMPROB] = ENOPROTOOPT, 1157 }; 1158 1159 void 1160 ip_fasttimo(void) 1161 { 1162 if (ip_drainwanted) { 1163 ip_drain(); 1164 ip_drainwanted = 0; 1165 } 1166 } 1167 1168 void 1169 ip_drainstub(void) 1170 { 1171 ip_drainwanted = 1; 1172 } 1173 1174 /* 1175 * Forward a packet. If some error occurs return the sender 1176 * an icmp packet. Note we can't always generate a meaningful 1177 * icmp message because icmp doesn't have a large enough repertoire 1178 * of codes and types. 1179 * 1180 * If not forwarding, just drop the packet. This could be confusing 1181 * if ipforwarding was zero but some routing protocol was advancing 1182 * us as a gateway to somewhere. However, we must let the routing 1183 * protocol deal with that. 1184 * 1185 * The srcrt parameter indicates whether the packet is being forwarded 1186 * via a source route. 1187 */ 1188 static void 1189 ip_forward(struct mbuf *m, int srcrt) 1190 { 1191 struct ip *ip = mtod(m, struct ip *); 1192 struct rtentry *rt; 1193 int error, type = 0, code = 0, destmtu = 0; 1194 struct mbuf *mcopy; 1195 n_long dest; 1196 union { 1197 struct sockaddr dst; 1198 struct sockaddr_in dst4; 1199 } u; 1200 uint64_t *ips; 1201 1202 KASSERTMSG(cpu_softintr_p(), "ip_forward: not in the software " 1203 "interrupt handler; synchronization assumptions violated"); 1204 1205 /* 1206 * We are now in the output path. 1207 */ 1208 MCLAIM(m, &ip_tx_mowner); 1209 1210 /* 1211 * Clear any in-bound checksum flags for this packet. 1212 */ 1213 m->m_pkthdr.csum_flags = 0; 1214 1215 dest = 0; 1216 if (m->m_flags & (M_BCAST|M_MCAST) || in_canforward(ip->ip_dst) == 0) { 1217 IP_STATINC(IP_STAT_CANTFORWARD); 1218 m_freem(m); 1219 return; 1220 } 1221 1222 SOFTNET_LOCK(); 1223 1224 if (ip->ip_ttl <= IPTTLDEC) { 1225 icmp_error(m, ICMP_TIMXCEED, ICMP_TIMXCEED_INTRANS, dest, 0); 1226 SOFTNET_UNLOCK(); 1227 return; 1228 } 1229 1230 sockaddr_in_init(&u.dst4, &ip->ip_dst, 0); 1231 1232 if ((rt = rtcache_lookup(&ipforward_rt, &u.dst)) == NULL) { 1233 icmp_error(m, ICMP_UNREACH, ICMP_UNREACH_NET, dest, 0); 1234 SOFTNET_UNLOCK(); 1235 return; 1236 } 1237 1238 /* 1239 * Save at most 68 bytes of the packet in case 1240 * we need to generate an ICMP message to the src. 1241 * Pullup to avoid sharing mbuf cluster between m and mcopy. 1242 */ 1243 mcopy = m_copym(m, 0, imin(ntohs(ip->ip_len), 68), M_DONTWAIT); 1244 if (mcopy) 1245 mcopy = m_pullup(mcopy, ip->ip_hl << 2); 1246 1247 ip->ip_ttl -= IPTTLDEC; 1248 1249 /* 1250 * If forwarding packet using same interface that it came in on, 1251 * perhaps should send a redirect to sender to shortcut a hop. 1252 * Only send redirect if source is sending directly to us, 1253 * and if packet was not source routed (or has any options). 1254 * Also, don't send redirect if forwarding using a default route 1255 * or a route modified by a redirect. 1256 */ 1257 if (rt->rt_ifp == m->m_pkthdr.rcvif && 1258 (rt->rt_flags & (RTF_DYNAMIC|RTF_MODIFIED)) == 0 && 1259 !in_nullhost(satocsin(rt_getkey(rt))->sin_addr) && 1260 ipsendredirects && !srcrt) { 1261 if (rt->rt_ifa && 1262 (ip->ip_src.s_addr & ifatoia(rt->rt_ifa)->ia_subnetmask) == 1263 ifatoia(rt->rt_ifa)->ia_subnet) { 1264 if (rt->rt_flags & RTF_GATEWAY) 1265 dest = satosin(rt->rt_gateway)->sin_addr.s_addr; 1266 else 1267 dest = ip->ip_dst.s_addr; 1268 /* 1269 * Router requirements says to only send host 1270 * redirects. 1271 */ 1272 type = ICMP_REDIRECT; 1273 code = ICMP_REDIRECT_HOST; 1274 } 1275 } 1276 1277 error = ip_output(m, NULL, &ipforward_rt, 1278 (IP_FORWARDING | (ip_directedbcast ? IP_ALLOWBROADCAST : 0)), 1279 NULL, NULL); 1280 1281 if (error) { 1282 IP_STATINC(IP_STAT_CANTFORWARD); 1283 goto error; 1284 } 1285 1286 ips = IP_STAT_GETREF(); 1287 ips[IP_STAT_FORWARD]++; 1288 1289 if (type) { 1290 ips[IP_STAT_REDIRECTSENT]++; 1291 IP_STAT_PUTREF(); 1292 goto redirect; 1293 } 1294 1295 IP_STAT_PUTREF(); 1296 if (mcopy) { 1297 #ifdef GATEWAY 1298 if (mcopy->m_flags & M_CANFASTFWD) 1299 ipflow_create(&ipforward_rt, mcopy); 1300 #endif 1301 m_freem(mcopy); 1302 } 1303 1304 SOFTNET_UNLOCK(); 1305 return; 1306 1307 redirect: 1308 error: 1309 if (mcopy == NULL) { 1310 SOFTNET_UNLOCK(); 1311 return; 1312 } 1313 1314 switch (error) { 1315 1316 case 0: /* forwarded, but need redirect */ 1317 /* type, code set above */ 1318 break; 1319 1320 case ENETUNREACH: /* shouldn't happen, checked above */ 1321 case EHOSTUNREACH: 1322 case ENETDOWN: 1323 case EHOSTDOWN: 1324 default: 1325 type = ICMP_UNREACH; 1326 code = ICMP_UNREACH_HOST; 1327 break; 1328 1329 case EMSGSIZE: 1330 type = ICMP_UNREACH; 1331 code = ICMP_UNREACH_NEEDFRAG; 1332 1333 if ((rt = rtcache_validate(&ipforward_rt)) != NULL) 1334 destmtu = rt->rt_ifp->if_mtu; 1335 #ifdef IPSEC 1336 if (ipsec_used) 1337 (void)ipsec4_forward(mcopy, &destmtu); 1338 #endif 1339 IP_STATINC(IP_STAT_CANTFRAG); 1340 break; 1341 1342 case ENOBUFS: 1343 /* 1344 * Do not generate ICMP_SOURCEQUENCH as required in RFC 1812, 1345 * Requirements for IP Version 4 Routers. Source quench can 1346 * big problem under DoS attacks or if the underlying 1347 * interface is rate-limited. 1348 */ 1349 if (mcopy) 1350 m_freem(mcopy); 1351 SOFTNET_UNLOCK(); 1352 return; 1353 } 1354 icmp_error(mcopy, type, code, dest, destmtu); 1355 SOFTNET_UNLOCK(); 1356 } 1357 1358 void 1359 ip_savecontrol(struct inpcb *inp, struct mbuf **mp, struct ip *ip, 1360 struct mbuf *m) 1361 { 1362 struct socket *so = inp->inp_socket; 1363 ifnet_t *ifp = m->m_pkthdr.rcvif; 1364 int inpflags = inp->inp_flags; 1365 1366 if (so->so_options & SO_TIMESTAMP 1367 #ifdef SO_OTIMESTAMP 1368 || so->so_options & SO_OTIMESTAMP 1369 #endif 1370 ) { 1371 struct timeval tv; 1372 1373 microtime(&tv); 1374 #ifdef SO_OTIMESTAMP 1375 if (so->so_options & SO_OTIMESTAMP) { 1376 struct timeval50 tv50; 1377 timeval_to_timeval50(&tv, &tv50); 1378 *mp = sbcreatecontrol((void *) &tv50, sizeof(tv50), 1379 SCM_OTIMESTAMP, SOL_SOCKET); 1380 } else 1381 #endif 1382 *mp = sbcreatecontrol((void *) &tv, sizeof(tv), 1383 SCM_TIMESTAMP, SOL_SOCKET); 1384 if (*mp) 1385 mp = &(*mp)->m_next; 1386 } 1387 if (inpflags & INP_RECVDSTADDR) { 1388 *mp = sbcreatecontrol((void *) &ip->ip_dst, 1389 sizeof(struct in_addr), IP_RECVDSTADDR, IPPROTO_IP); 1390 if (*mp) 1391 mp = &(*mp)->m_next; 1392 } 1393 if (inpflags & INP_RECVPKTINFO) { 1394 struct in_pktinfo ipi; 1395 ipi.ipi_addr = ip->ip_src; 1396 ipi.ipi_ifindex = ifp->if_index; 1397 *mp = sbcreatecontrol((void *) &ipi, 1398 sizeof(ipi), IP_RECVPKTINFO, IPPROTO_IP); 1399 if (*mp) 1400 mp = &(*mp)->m_next; 1401 } 1402 if (inpflags & INP_PKTINFO) { 1403 struct in_pktinfo ipi; 1404 ipi.ipi_addr = ip->ip_dst; 1405 ipi.ipi_ifindex = ifp->if_index; 1406 *mp = sbcreatecontrol((void *) &ipi, 1407 sizeof(ipi), IP_PKTINFO, IPPROTO_IP); 1408 if (*mp) 1409 mp = &(*mp)->m_next; 1410 } 1411 if (inpflags & INP_RECVIF) { 1412 struct sockaddr_dl sdl; 1413 1414 sockaddr_dl_init(&sdl, sizeof(sdl), ifp ? 1415 ifp->if_index : 0, 0, NULL, 0, NULL, 0); 1416 *mp = sbcreatecontrol(&sdl, sdl.sdl_len, IP_RECVIF, IPPROTO_IP); 1417 if (*mp) 1418 mp = &(*mp)->m_next; 1419 } 1420 if (inpflags & INP_RECVTTL) { 1421 *mp = sbcreatecontrol((void *) &ip->ip_ttl, 1422 sizeof(uint8_t), IP_RECVTTL, IPPROTO_IP); 1423 if (*mp) 1424 mp = &(*mp)->m_next; 1425 } 1426 } 1427 1428 /* 1429 * sysctl helper routine for net.inet.ip.forwsrcrt. 1430 */ 1431 static int 1432 sysctl_net_inet_ip_forwsrcrt(SYSCTLFN_ARGS) 1433 { 1434 int error, tmp; 1435 struct sysctlnode node; 1436 1437 node = *rnode; 1438 tmp = ip_forwsrcrt; 1439 node.sysctl_data = &tmp; 1440 error = sysctl_lookup(SYSCTLFN_CALL(&node)); 1441 if (error || newp == NULL) 1442 return (error); 1443 1444 error = kauth_authorize_network(l->l_cred, KAUTH_NETWORK_FORWSRCRT, 1445 0, NULL, NULL, NULL); 1446 if (error) 1447 return (error); 1448 1449 ip_forwsrcrt = tmp; 1450 1451 return (0); 1452 } 1453 1454 /* 1455 * sysctl helper routine for net.inet.ip.mtudisctimeout. checks the 1456 * range of the new value and tweaks timers if it changes. 1457 */ 1458 static int 1459 sysctl_net_inet_ip_pmtudto(SYSCTLFN_ARGS) 1460 { 1461 int error, tmp; 1462 struct sysctlnode node; 1463 1464 node = *rnode; 1465 tmp = ip_mtudisc_timeout; 1466 node.sysctl_data = &tmp; 1467 error = sysctl_lookup(SYSCTLFN_CALL(&node)); 1468 if (error || newp == NULL) 1469 return (error); 1470 if (tmp < 0) 1471 return (EINVAL); 1472 1473 mutex_enter(softnet_lock); 1474 1475 ip_mtudisc_timeout = tmp; 1476 rt_timer_queue_change(ip_mtudisc_timeout_q, ip_mtudisc_timeout); 1477 1478 mutex_exit(softnet_lock); 1479 1480 return (0); 1481 } 1482 1483 static int 1484 sysctl_net_inet_ip_stats(SYSCTLFN_ARGS) 1485 { 1486 1487 return (NETSTAT_SYSCTL(ipstat_percpu, IP_NSTATS)); 1488 } 1489 1490 static void 1491 sysctl_net_inet_ip_setup(struct sysctllog **clog) 1492 { 1493 sysctl_createv(clog, 0, NULL, NULL, 1494 CTLFLAG_PERMANENT, 1495 CTLTYPE_NODE, "inet", 1496 SYSCTL_DESCR("PF_INET related settings"), 1497 NULL, 0, NULL, 0, 1498 CTL_NET, PF_INET, CTL_EOL); 1499 sysctl_createv(clog, 0, NULL, NULL, 1500 CTLFLAG_PERMANENT, 1501 CTLTYPE_NODE, "ip", 1502 SYSCTL_DESCR("IPv4 related settings"), 1503 NULL, 0, NULL, 0, 1504 CTL_NET, PF_INET, IPPROTO_IP, CTL_EOL); 1505 1506 sysctl_createv(clog, 0, NULL, NULL, 1507 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1508 CTLTYPE_INT, "forwarding", 1509 SYSCTL_DESCR("Enable forwarding of INET datagrams"), 1510 NULL, 0, &ipforwarding, 0, 1511 CTL_NET, PF_INET, IPPROTO_IP, 1512 IPCTL_FORWARDING, CTL_EOL); 1513 sysctl_createv(clog, 0, NULL, NULL, 1514 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1515 CTLTYPE_INT, "redirect", 1516 SYSCTL_DESCR("Enable sending of ICMP redirect messages"), 1517 NULL, 0, &ipsendredirects, 0, 1518 CTL_NET, PF_INET, IPPROTO_IP, 1519 IPCTL_SENDREDIRECTS, CTL_EOL); 1520 sysctl_createv(clog, 0, NULL, NULL, 1521 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1522 CTLTYPE_INT, "ttl", 1523 SYSCTL_DESCR("Default TTL for an INET datagram"), 1524 NULL, 0, &ip_defttl, 0, 1525 CTL_NET, PF_INET, IPPROTO_IP, 1526 IPCTL_DEFTTL, CTL_EOL); 1527 #ifdef IPCTL_DEFMTU 1528 sysctl_createv(clog, 0, NULL, NULL, 1529 CTLFLAG_PERMANENT /* |CTLFLAG_READWRITE? */, 1530 CTLTYPE_INT, "mtu", 1531 SYSCTL_DESCR("Default MTA for an INET route"), 1532 NULL, 0, &ip_mtu, 0, 1533 CTL_NET, PF_INET, IPPROTO_IP, 1534 IPCTL_DEFMTU, CTL_EOL); 1535 #endif /* IPCTL_DEFMTU */ 1536 sysctl_createv(clog, 0, NULL, NULL, 1537 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1538 CTLTYPE_INT, "forwsrcrt", 1539 SYSCTL_DESCR("Enable forwarding of source-routed " 1540 "datagrams"), 1541 sysctl_net_inet_ip_forwsrcrt, 0, &ip_forwsrcrt, 0, 1542 CTL_NET, PF_INET, IPPROTO_IP, 1543 IPCTL_FORWSRCRT, CTL_EOL); 1544 sysctl_createv(clog, 0, NULL, NULL, 1545 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1546 CTLTYPE_INT, "directed-broadcast", 1547 SYSCTL_DESCR("Enable forwarding of broadcast datagrams"), 1548 NULL, 0, &ip_directedbcast, 0, 1549 CTL_NET, PF_INET, IPPROTO_IP, 1550 IPCTL_DIRECTEDBCAST, CTL_EOL); 1551 sysctl_createv(clog, 0, NULL, NULL, 1552 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1553 CTLTYPE_INT, "allowsrcrt", 1554 SYSCTL_DESCR("Accept source-routed datagrams"), 1555 NULL, 0, &ip_allowsrcrt, 0, 1556 CTL_NET, PF_INET, IPPROTO_IP, 1557 IPCTL_ALLOWSRCRT, CTL_EOL); 1558 1559 sysctl_createv(clog, 0, NULL, NULL, 1560 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1561 CTLTYPE_INT, "mtudisc", 1562 SYSCTL_DESCR("Use RFC1191 Path MTU Discovery"), 1563 NULL, 0, &ip_mtudisc, 0, 1564 CTL_NET, PF_INET, IPPROTO_IP, 1565 IPCTL_MTUDISC, CTL_EOL); 1566 sysctl_createv(clog, 0, NULL, NULL, 1567 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1568 CTLTYPE_INT, "anonportmin", 1569 SYSCTL_DESCR("Lowest ephemeral port number to assign"), 1570 sysctl_net_inet_ip_ports, 0, &anonportmin, 0, 1571 CTL_NET, PF_INET, IPPROTO_IP, 1572 IPCTL_ANONPORTMIN, CTL_EOL); 1573 sysctl_createv(clog, 0, NULL, NULL, 1574 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1575 CTLTYPE_INT, "anonportmax", 1576 SYSCTL_DESCR("Highest ephemeral port number to assign"), 1577 sysctl_net_inet_ip_ports, 0, &anonportmax, 0, 1578 CTL_NET, PF_INET, IPPROTO_IP, 1579 IPCTL_ANONPORTMAX, CTL_EOL); 1580 sysctl_createv(clog, 0, NULL, NULL, 1581 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1582 CTLTYPE_INT, "mtudisctimeout", 1583 SYSCTL_DESCR("Lifetime of a Path MTU Discovered route"), 1584 sysctl_net_inet_ip_pmtudto, 0, (void *)&ip_mtudisc_timeout, 0, 1585 CTL_NET, PF_INET, IPPROTO_IP, 1586 IPCTL_MTUDISCTIMEOUT, CTL_EOL); 1587 #ifndef IPNOPRIVPORTS 1588 sysctl_createv(clog, 0, NULL, NULL, 1589 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1590 CTLTYPE_INT, "lowportmin", 1591 SYSCTL_DESCR("Lowest privileged ephemeral port number " 1592 "to assign"), 1593 sysctl_net_inet_ip_ports, 0, &lowportmin, 0, 1594 CTL_NET, PF_INET, IPPROTO_IP, 1595 IPCTL_LOWPORTMIN, CTL_EOL); 1596 sysctl_createv(clog, 0, NULL, NULL, 1597 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1598 CTLTYPE_INT, "lowportmax", 1599 SYSCTL_DESCR("Highest privileged ephemeral port number " 1600 "to assign"), 1601 sysctl_net_inet_ip_ports, 0, &lowportmax, 0, 1602 CTL_NET, PF_INET, IPPROTO_IP, 1603 IPCTL_LOWPORTMAX, CTL_EOL); 1604 #endif /* IPNOPRIVPORTS */ 1605 #if NGRE > 0 1606 sysctl_createv(clog, 0, NULL, NULL, 1607 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1608 CTLTYPE_INT, "grettl", 1609 SYSCTL_DESCR("Default TTL for a gre tunnel datagram"), 1610 NULL, 0, &ip_gre_ttl, 0, 1611 CTL_NET, PF_INET, IPPROTO_IP, 1612 IPCTL_GRE_TTL, CTL_EOL); 1613 #endif /* NGRE */ 1614 sysctl_createv(clog, 0, NULL, NULL, 1615 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1616 CTLTYPE_INT, "checkinterface", 1617 SYSCTL_DESCR("Enable receive side of Strong ES model " 1618 "from RFC1122"), 1619 NULL, 0, &ip_checkinterface, 0, 1620 CTL_NET, PF_INET, IPPROTO_IP, 1621 IPCTL_CHECKINTERFACE, CTL_EOL); 1622 sysctl_createv(clog, 0, NULL, NULL, 1623 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1624 CTLTYPE_INT, "random_id", 1625 SYSCTL_DESCR("Assign random ip_id values"), 1626 NULL, 0, &ip_do_randomid, 0, 1627 CTL_NET, PF_INET, IPPROTO_IP, 1628 IPCTL_RANDOMID, CTL_EOL); 1629 sysctl_createv(clog, 0, NULL, NULL, 1630 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1631 CTLTYPE_INT, "do_loopback_cksum", 1632 SYSCTL_DESCR("Perform IP checksum on loopback"), 1633 NULL, 0, &ip_do_loopback_cksum, 0, 1634 CTL_NET, PF_INET, IPPROTO_IP, 1635 IPCTL_LOOPBACKCKSUM, CTL_EOL); 1636 sysctl_createv(clog, 0, NULL, NULL, 1637 CTLFLAG_PERMANENT, 1638 CTLTYPE_STRUCT, "stats", 1639 SYSCTL_DESCR("IP statistics"), 1640 sysctl_net_inet_ip_stats, 0, NULL, 0, 1641 CTL_NET, PF_INET, IPPROTO_IP, IPCTL_STATS, 1642 CTL_EOL); 1643 #if NARP 1644 sysctl_createv(clog, 0, NULL, NULL, 1645 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1646 CTLTYPE_INT, "dad_count", 1647 SYSCTL_DESCR("Number of Duplicate Address Detection " 1648 "probes to send"), 1649 NULL, 0, &ip_dad_count, 0, 1650 CTL_NET, PF_INET, IPPROTO_IP, 1651 IPCTL_DAD_COUNT, CTL_EOL); 1652 #endif 1653 1654 /* anonportalgo RFC6056 subtree */ 1655 const struct sysctlnode *portalgo_node; 1656 sysctl_createv(clog, 0, NULL, &portalgo_node, 1657 CTLFLAG_PERMANENT, 1658 CTLTYPE_NODE, "anonportalgo", 1659 SYSCTL_DESCR("Anonymous Port Algorithm Selection (RFC 6056)"), 1660 NULL, 0, NULL, 0, 1661 CTL_NET, PF_INET, IPPROTO_IP, CTL_CREATE, CTL_EOL); 1662 sysctl_createv(clog, 0, &portalgo_node, NULL, 1663 CTLFLAG_PERMANENT, 1664 CTLTYPE_STRING, "available", 1665 SYSCTL_DESCR("available algorithms"), 1666 sysctl_portalgo_available, 0, NULL, PORTALGO_MAXLEN, 1667 CTL_CREATE, CTL_EOL); 1668 sysctl_createv(clog, 0, &portalgo_node, NULL, 1669 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1670 CTLTYPE_STRING, "selected", 1671 SYSCTL_DESCR("selected algorithm"), 1672 sysctl_portalgo_selected4, 0, NULL, PORTALGO_MAXLEN, 1673 CTL_CREATE, CTL_EOL); 1674 sysctl_createv(clog, 0, &portalgo_node, NULL, 1675 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1676 CTLTYPE_STRUCT, "reserve", 1677 SYSCTL_DESCR("bitmap of reserved ports"), 1678 sysctl_portalgo_reserve4, 0, NULL, 0, 1679 CTL_CREATE, CTL_EOL); 1680 } 1681 1682 void 1683 ip_statinc(u_int stat) 1684 { 1685 1686 KASSERT(stat < IP_NSTATS); 1687 IP_STATINC(stat); 1688 } 1689