ip_input.c revision 1.337.2.2
1 /* $NetBSD: ip_input.c,v 1.337.2.2 2016/11/04 14:49:21 pgoyette 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 1.337.2.2 2016/11/04 14:49:21 pgoyette 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 #include "opt_net_mpsafe.h" 105 #endif 106 107 #include "arp.h" 108 109 #include <sys/param.h> 110 #include <sys/systm.h> 111 #include <sys/cpu.h> 112 #include <sys/mbuf.h> 113 #include <sys/domain.h> 114 #include <sys/protosw.h> 115 #include <sys/socket.h> 116 #include <sys/socketvar.h> 117 #include <sys/errno.h> 118 #include <sys/time.h> 119 #include <sys/kernel.h> 120 #include <sys/pool.h> 121 #include <sys/sysctl.h> 122 #include <sys/kauth.h> 123 124 #include <net/if.h> 125 #include <net/if_dl.h> 126 #include <net/route.h> 127 #include <net/pktqueue.h> 128 #include <net/pfil.h> 129 130 #include <netinet/in.h> 131 #include <netinet/in_systm.h> 132 #include <netinet/ip.h> 133 #include <netinet/in_pcb.h> 134 #include <netinet/in_proto.h> 135 #include <netinet/in_var.h> 136 #include <netinet/ip_var.h> 137 #include <netinet/ip_private.h> 138 #include <netinet/ip_icmp.h> 139 /* just for gif_ttl */ 140 #include <netinet/in_gif.h> 141 #include "gif.h" 142 #include <net/if_gre.h> 143 #include "gre.h" 144 145 #ifdef MROUTING 146 #include <netinet/ip_mroute.h> 147 #endif 148 #include <netinet/portalgo.h> 149 150 #ifdef IPSEC 151 #include <netipsec/ipsec.h> 152 #endif 153 154 #ifndef IPFORWARDING 155 #ifdef GATEWAY 156 #define IPFORWARDING 1 /* forward IP packets not for us */ 157 #else /* GATEWAY */ 158 #define IPFORWARDING 0 /* don't forward IP packets not for us */ 159 #endif /* GATEWAY */ 160 #endif /* IPFORWARDING */ 161 #ifndef IPSENDREDIRECTS 162 #define IPSENDREDIRECTS 1 163 #endif 164 #ifndef IPFORWSRCRT 165 #define IPFORWSRCRT 1 /* forward source-routed packets */ 166 #endif 167 #ifndef IPALLOWSRCRT 168 #define IPALLOWSRCRT 1 /* allow source-routed packets */ 169 #endif 170 #ifndef IPMTUDISC 171 #define IPMTUDISC 1 172 #endif 173 #ifndef IPMTUDISCTIMEOUT 174 #define IPMTUDISCTIMEOUT (10 * 60) /* as per RFC 1191 */ 175 #endif 176 177 #ifdef COMPAT_50 178 #include <compat/sys/time.h> 179 #include <compat/sys/socket.h> 180 #endif 181 182 /* 183 * Note: DIRECTED_BROADCAST is handled this way so that previous 184 * configuration using this option will Just Work. 185 */ 186 #ifndef IPDIRECTEDBCAST 187 #ifdef DIRECTED_BROADCAST 188 #define IPDIRECTEDBCAST 1 189 #else 190 #define IPDIRECTEDBCAST 0 191 #endif /* DIRECTED_BROADCAST */ 192 #endif /* IPDIRECTEDBCAST */ 193 int ipforwarding = IPFORWARDING; 194 int ipsendredirects = IPSENDREDIRECTS; 195 int ip_defttl = IPDEFTTL; 196 int ip_forwsrcrt = IPFORWSRCRT; 197 int ip_directedbcast = IPDIRECTEDBCAST; 198 int ip_allowsrcrt = IPALLOWSRCRT; 199 int ip_mtudisc = IPMTUDISC; 200 int ip_mtudisc_timeout = IPMTUDISCTIMEOUT; 201 #ifdef DIAGNOSTIC 202 int ipprintfs = 0; 203 #endif 204 205 int ip_do_randomid = 0; 206 207 /* 208 * XXX - Setting ip_checkinterface mostly implements the receive side of 209 * the Strong ES model described in RFC 1122, but since the routing table 210 * and transmit implementation do not implement the Strong ES model, 211 * setting this to 1 results in an odd hybrid. 212 * 213 * XXX - ip_checkinterface currently must be disabled if you use ipnat 214 * to translate the destination address to another local interface. 215 * 216 * XXX - ip_checkinterface must be disabled if you add IP aliases 217 * to the loopback interface instead of the interface where the 218 * packets for those addresses are received. 219 */ 220 static int ip_checkinterface __read_mostly = 0; 221 222 struct rttimer_queue *ip_mtudisc_timeout_q = NULL; 223 224 pktqueue_t * ip_pktq __read_mostly; 225 pfil_head_t * inet_pfil_hook __read_mostly; 226 ipid_state_t * ip_ids __read_mostly; 227 percpu_t * ipstat_percpu __read_mostly; 228 229 static struct route ipforward_rt __cacheline_aligned; 230 231 uint16_t ip_id; 232 233 #ifdef INET_CSUM_COUNTERS 234 #include <sys/device.h> 235 236 struct evcnt ip_hwcsum_bad = EVCNT_INITIALIZER(EVCNT_TYPE_MISC, 237 NULL, "inet", "hwcsum bad"); 238 struct evcnt ip_hwcsum_ok = EVCNT_INITIALIZER(EVCNT_TYPE_MISC, 239 NULL, "inet", "hwcsum ok"); 240 struct evcnt ip_swcsum = EVCNT_INITIALIZER(EVCNT_TYPE_MISC, 241 NULL, "inet", "swcsum"); 242 243 #define INET_CSUM_COUNTER_INCR(ev) (ev)->ev_count++ 244 245 EVCNT_ATTACH_STATIC(ip_hwcsum_bad); 246 EVCNT_ATTACH_STATIC(ip_hwcsum_ok); 247 EVCNT_ATTACH_STATIC(ip_swcsum); 248 249 #else 250 251 #define INET_CSUM_COUNTER_INCR(ev) /* nothing */ 252 253 #endif /* INET_CSUM_COUNTERS */ 254 255 /* 256 * We need to save the IP options in case a protocol wants to respond 257 * to an incoming packet over the same route if the packet got here 258 * using IP source routing. This allows connection establishment and 259 * maintenance when the remote end is on a network that is not known 260 * to us. 261 */ 262 263 static int ip_nhops = 0; 264 265 static struct ip_srcrt { 266 struct in_addr dst; /* final destination */ 267 char nop; /* one NOP to align */ 268 char srcopt[IPOPT_OFFSET + 1]; /* OPTVAL, OLEN and OFFSET */ 269 struct in_addr route[MAX_IPOPTLEN/sizeof(struct in_addr)]; 270 } ip_srcrt; 271 272 static int ip_drainwanted; 273 274 struct sockaddr_in ipaddr = { 275 .sin_len = sizeof(ipaddr), 276 .sin_family = AF_INET, 277 }; 278 279 static void save_rte(u_char *, struct in_addr); 280 281 #ifdef MBUFTRACE 282 struct mowner ip_rx_mowner = MOWNER_INIT("internet", "rx"); 283 struct mowner ip_tx_mowner = MOWNER_INIT("internet", "tx"); 284 #endif 285 286 static void ipintr(void *); 287 static void ip_input(struct mbuf *); 288 static void ip_forward(struct mbuf *, int, struct ifnet *); 289 static bool ip_dooptions(struct mbuf *); 290 static struct in_ifaddr *ip_rtaddr(struct in_addr); 291 static void sysctl_net_inet_ip_setup(struct sysctllog **); 292 293 static struct in_ifaddr *ip_match_our_address(struct ifnet *, struct ip *, 294 int *); 295 static struct in_ifaddr *ip_match_our_address_broadcast(struct ifnet *, 296 struct ip *); 297 298 #ifdef NET_MPSAFE 299 #define SOFTNET_LOCK() mutex_enter(softnet_lock) 300 #define SOFTNET_UNLOCK() mutex_exit(softnet_lock) 301 #else 302 #define SOFTNET_LOCK() KASSERT(mutex_owned(softnet_lock)) 303 #define SOFTNET_UNLOCK() KASSERT(mutex_owned(softnet_lock)) 304 #endif 305 306 /* 307 * IP initialization: fill in IP protocol switch table. 308 * All protocols not implemented in kernel go to raw IP protocol handler. 309 */ 310 void 311 ip_init(void) 312 { 313 const struct protosw *pr; 314 315 in_init(); 316 sysctl_net_inet_ip_setup(NULL); 317 318 pr = pffindproto(PF_INET, IPPROTO_RAW, SOCK_RAW); 319 KASSERT(pr != NULL); 320 321 ip_pktq = pktq_create(IFQ_MAXLEN, ipintr, NULL); 322 KASSERT(ip_pktq != NULL); 323 324 for (u_int i = 0; i < IPPROTO_MAX; i++) { 325 ip_protox[i] = pr - inetsw; 326 } 327 for (pr = inetdomain.dom_protosw; 328 pr < inetdomain.dom_protoswNPROTOSW; pr++) 329 if (pr->pr_domain->dom_family == PF_INET && 330 pr->pr_protocol && pr->pr_protocol != IPPROTO_RAW) 331 ip_protox[pr->pr_protocol] = pr - inetsw; 332 333 ip_reass_init(); 334 335 ip_ids = ip_id_init(); 336 ip_id = time_uptime & 0xfffff; 337 338 ip_mtudisc_timeout_q = rt_timer_queue_create(ip_mtudisc_timeout); 339 #ifdef GATEWAY 340 ipflow_init(); 341 #endif 342 343 /* Register our Packet Filter hook. */ 344 inet_pfil_hook = pfil_head_create(PFIL_TYPE_AF, (void *)AF_INET); 345 KASSERT(inet_pfil_hook != NULL); 346 347 #ifdef MBUFTRACE 348 MOWNER_ATTACH(&ip_tx_mowner); 349 MOWNER_ATTACH(&ip_rx_mowner); 350 #endif /* MBUFTRACE */ 351 352 ipstat_percpu = percpu_alloc(sizeof(uint64_t) * IP_NSTATS); 353 } 354 355 static struct in_ifaddr * 356 ip_match_our_address(struct ifnet *ifp, struct ip *ip, int *downmatch) 357 { 358 struct in_ifaddr *ia = NULL; 359 int checkif; 360 361 /* 362 * Enable a consistency check between the destination address 363 * and the arrival interface for a unicast packet (the RFC 1122 364 * strong ES model) if IP forwarding is disabled and the packet 365 * is not locally generated. 366 * 367 * XXX - Checking also should be disabled if the destination 368 * address is ipnat'ed to a different interface. 369 * 370 * XXX - Checking is incompatible with IP aliases added 371 * to the loopback interface instead of the interface where 372 * the packets are received. 373 * 374 * XXX - We need to add a per ifaddr flag for this so that 375 * we get finer grain control. 376 */ 377 checkif = ip_checkinterface && (ipforwarding == 0) && 378 (ifp->if_flags & IFF_LOOPBACK) == 0; 379 380 IN_ADDRHASH_READER_FOREACH(ia, ip->ip_dst.s_addr) { 381 if (in_hosteq(ia->ia_addr.sin_addr, ip->ip_dst)) { 382 if (ia->ia4_flags & IN_IFF_NOTREADY) 383 continue; 384 if (checkif && ia->ia_ifp != ifp) 385 continue; 386 if ((ia->ia_ifp->if_flags & IFF_UP) != 0) 387 break; 388 else 389 (*downmatch)++; 390 } 391 } 392 393 return ia; 394 } 395 396 static struct in_ifaddr * 397 ip_match_our_address_broadcast(struct ifnet *ifp, struct ip *ip) 398 { 399 struct in_ifaddr *ia = NULL; 400 struct ifaddr *ifa; 401 402 IFADDR_READER_FOREACH(ifa, ifp) { 403 if (ifa->ifa_addr->sa_family != AF_INET) 404 continue; 405 ia = ifatoia(ifa); 406 if (ia->ia4_flags & IN_IFF_NOTREADY) 407 continue; 408 if (in_hosteq(ip->ip_dst, ia->ia_broadaddr.sin_addr) || 409 in_hosteq(ip->ip_dst, ia->ia_netbroadcast) || 410 /* 411 * Look for all-0's host part (old broadcast addr), 412 * either for subnet or net. 413 */ 414 ip->ip_dst.s_addr == ia->ia_subnet || 415 ip->ip_dst.s_addr == ia->ia_net) 416 goto matched; 417 /* 418 * An interface with IP address zero accepts 419 * all packets that arrive on that interface. 420 */ 421 if (in_nullhost(ia->ia_addr.sin_addr)) 422 goto matched; 423 } 424 ia = NULL; 425 426 matched: 427 return ia; 428 } 429 430 /* 431 * IP software interrupt routine. 432 */ 433 static void 434 ipintr(void *arg __unused) 435 { 436 struct mbuf *m; 437 438 KASSERT(cpu_softintr_p()); 439 440 #ifndef NET_MPSAFE 441 mutex_enter(softnet_lock); 442 #endif 443 while ((m = pktq_dequeue(ip_pktq)) != NULL) { 444 ip_input(m); 445 } 446 #ifndef NET_MPSAFE 447 mutex_exit(softnet_lock); 448 #endif 449 } 450 451 /* 452 * IP input routine. Checksum and byte swap header. If fragmented 453 * try to reassemble. Process options. Pass to next level. 454 */ 455 static void 456 ip_input(struct mbuf *m) 457 { 458 struct ip *ip = NULL; 459 struct in_ifaddr *ia = NULL; 460 int hlen = 0, len; 461 int downmatch; 462 int srcrt = 0; 463 ifnet_t *ifp; 464 struct psref psref; 465 int s; 466 467 KASSERTMSG(cpu_softintr_p(), "ip_input: not in the software " 468 "interrupt handler; synchronization assumptions violated"); 469 470 MCLAIM(m, &ip_rx_mowner); 471 KASSERT((m->m_flags & M_PKTHDR) != 0); 472 473 ifp = m_get_rcvif_psref(m, &psref); 474 if (__predict_false(ifp == NULL)) 475 goto out; 476 477 /* 478 * If no IP addresses have been set yet but the interfaces 479 * are receiving, can't do anything with incoming packets yet. 480 * Note: we pre-check without locks held. 481 */ 482 if (IN_ADDRLIST_READER_EMPTY()) 483 goto out; 484 IP_STATINC(IP_STAT_TOTAL); 485 486 /* 487 * If the IP header is not aligned, slurp it up into a new 488 * mbuf with space for link headers, in the event we forward 489 * it. Otherwise, if it is aligned, make sure the entire 490 * base IP header is in the first mbuf of the chain. 491 */ 492 if (IP_HDR_ALIGNED_P(mtod(m, void *)) == 0) { 493 if ((m = m_copyup(m, sizeof(struct ip), 494 (max_linkhdr + 3) & ~3)) == NULL) { 495 /* XXXJRT new stat, please */ 496 IP_STATINC(IP_STAT_TOOSMALL); 497 goto out; 498 } 499 } else if (__predict_false(m->m_len < sizeof (struct ip))) { 500 if ((m = m_pullup(m, sizeof (struct ip))) == NULL) { 501 IP_STATINC(IP_STAT_TOOSMALL); 502 goto out; 503 } 504 } 505 ip = mtod(m, struct ip *); 506 if (ip->ip_v != IPVERSION) { 507 IP_STATINC(IP_STAT_BADVERS); 508 goto out; 509 } 510 hlen = ip->ip_hl << 2; 511 if (hlen < sizeof(struct ip)) { /* minimum header length */ 512 IP_STATINC(IP_STAT_BADHLEN); 513 goto out; 514 } 515 if (hlen > m->m_len) { 516 if ((m = m_pullup(m, hlen)) == NULL) { 517 IP_STATINC(IP_STAT_BADHLEN); 518 goto out; 519 } 520 ip = mtod(m, struct ip *); 521 } 522 523 /* 524 * RFC1122: packets with a multicast source address are 525 * not allowed. 526 */ 527 if (IN_MULTICAST(ip->ip_src.s_addr)) { 528 IP_STATINC(IP_STAT_BADADDR); 529 goto out; 530 } 531 532 /* 127/8 must not appear on wire - RFC1122 */ 533 if ((ntohl(ip->ip_dst.s_addr) >> IN_CLASSA_NSHIFT) == IN_LOOPBACKNET || 534 (ntohl(ip->ip_src.s_addr) >> IN_CLASSA_NSHIFT) == IN_LOOPBACKNET) { 535 if ((ifp->if_flags & IFF_LOOPBACK) == 0) { 536 IP_STATINC(IP_STAT_BADADDR); 537 goto out; 538 } 539 } 540 541 switch (m->m_pkthdr.csum_flags & 542 ((ifp->if_csum_flags_rx & M_CSUM_IPv4) | 543 M_CSUM_IPv4_BAD)) { 544 case M_CSUM_IPv4|M_CSUM_IPv4_BAD: 545 INET_CSUM_COUNTER_INCR(&ip_hwcsum_bad); 546 IP_STATINC(IP_STAT_BADSUM); 547 goto out; 548 549 case M_CSUM_IPv4: 550 /* Checksum was okay. */ 551 INET_CSUM_COUNTER_INCR(&ip_hwcsum_ok); 552 break; 553 554 default: 555 /* 556 * Must compute it ourselves. Maybe skip checksum on 557 * loopback interfaces. 558 */ 559 if (__predict_true(!(ifp->if_flags & IFF_LOOPBACK) || 560 ip_do_loopback_cksum)) { 561 INET_CSUM_COUNTER_INCR(&ip_swcsum); 562 if (in_cksum(m, hlen) != 0) { 563 IP_STATINC(IP_STAT_BADSUM); 564 goto out; 565 } 566 } 567 break; 568 } 569 570 /* Retrieve the packet length. */ 571 len = ntohs(ip->ip_len); 572 573 /* 574 * Check for additional length bogosity 575 */ 576 if (len < hlen) { 577 IP_STATINC(IP_STAT_BADLEN); 578 goto out; 579 } 580 581 /* 582 * Check that the amount of data in the buffers 583 * is as at least much as the IP header would have us expect. 584 * Trim mbufs if longer than we expect. 585 * Drop packet if shorter than we expect. 586 */ 587 if (m->m_pkthdr.len < len) { 588 IP_STATINC(IP_STAT_TOOSHORT); 589 goto out; 590 } 591 if (m->m_pkthdr.len > len) { 592 if (m->m_len == m->m_pkthdr.len) { 593 m->m_len = len; 594 m->m_pkthdr.len = len; 595 } else 596 m_adj(m, len - m->m_pkthdr.len); 597 } 598 599 /* 600 * Assume that we can create a fast-forward IP flow entry 601 * based on this packet. 602 */ 603 m->m_flags |= M_CANFASTFWD; 604 605 /* 606 * Run through list of hooks for input packets. If there are any 607 * filters which require that additional packets in the flow are 608 * not fast-forwarded, they must clear the M_CANFASTFWD flag. 609 * Note that filters must _never_ set this flag, as another filter 610 * in the list may have previously cleared it. 611 */ 612 #if defined(IPSEC) 613 if (!ipsec_used || !ipsec_indone(m)) 614 #else 615 if (1) 616 #endif 617 { 618 struct in_addr odst = ip->ip_dst; 619 bool freed; 620 621 freed = pfil_run_hooks(inet_pfil_hook, &m, ifp, PFIL_IN) != 0; 622 if (freed || m == NULL) { 623 m = NULL; 624 goto out; 625 } 626 ip = mtod(m, struct ip *); 627 hlen = ip->ip_hl << 2; 628 629 /* 630 * XXX The setting of "srcrt" here is to prevent ip_forward() 631 * from generating ICMP redirects for packets that have 632 * been redirected by a hook back out on to the same LAN that 633 * they came from and is not an indication that the packet 634 * is being inffluenced by source routing options. This 635 * allows things like 636 * "rdr tlp0 0/0 port 80 -> 1.1.1.200 3128 tcp" 637 * where tlp0 is both on the 1.1.1.0/24 network and is the 638 * default route for hosts on 1.1.1.0/24. Of course this 639 * also requires a "map tlp0 ..." to complete the story. 640 * One might argue whether or not this kind of network config. 641 * should be supported in this manner... 642 */ 643 srcrt = (odst.s_addr != ip->ip_dst.s_addr); 644 } 645 646 #ifdef ALTQ 647 /* XXX Temporary until ALTQ is changed to use a pfil hook */ 648 if (altq_input) { 649 SOFTNET_LOCK(); 650 if ((*altq_input)(m, AF_INET) == 0) { 651 /* Packet dropped by traffic conditioner. */ 652 SOFTNET_UNLOCK(); 653 m = NULL; 654 goto out; 655 } 656 SOFTNET_UNLOCK(); 657 } 658 #endif 659 660 /* 661 * Process options and, if not destined for us, 662 * ship it on. ip_dooptions returns 1 when an 663 * error was detected (causing an icmp message 664 * to be sent and the original packet to be freed). 665 */ 666 ip_nhops = 0; /* for source routed packets */ 667 if (hlen > sizeof (struct ip) && ip_dooptions(m)) { 668 m = NULL; 669 goto out; 670 } 671 672 /* 673 * Check our list of addresses, to see if the packet is for us. 674 * 675 * Traditional 4.4BSD did not consult IFF_UP at all. 676 * The behavior here is to treat addresses on !IFF_UP interface 677 * or IN_IFF_NOTREADY addresses as not mine. 678 */ 679 downmatch = 0; 680 s = pserialize_read_enter(); 681 ia = ip_match_our_address(ifp, ip, &downmatch); 682 if (ia != NULL) { 683 pserialize_read_exit(s); 684 goto ours; 685 } 686 687 if (ifp->if_flags & IFF_BROADCAST) { 688 ia = ip_match_our_address_broadcast(ifp, ip); 689 if (ia != NULL) { 690 pserialize_read_exit(s); 691 goto ours; 692 } 693 } 694 pserialize_read_exit(s); 695 696 if (IN_MULTICAST(ip->ip_dst.s_addr)) { 697 #ifdef MROUTING 698 extern struct socket *ip_mrouter; 699 700 if (ip_mrouter) { 701 /* 702 * If we are acting as a multicast router, all 703 * incoming multicast packets are passed to the 704 * kernel-level multicast forwarding function. 705 * The packet is returned (relatively) intact; if 706 * ip_mforward() returns a non-zero value, the packet 707 * must be discarded, else it may be accepted below. 708 * 709 * (The IP ident field is put in the same byte order 710 * as expected when ip_mforward() is called from 711 * ip_output().) 712 */ 713 SOFTNET_LOCK(); 714 if (ip_mforward(m, ifp) != 0) { 715 SOFTNET_UNLOCK(); 716 IP_STATINC(IP_STAT_CANTFORWARD); 717 goto out; 718 } 719 SOFTNET_UNLOCK(); 720 721 /* 722 * The process-level routing demon needs to receive 723 * all multicast IGMP packets, whether or not this 724 * host belongs to their destination groups. 725 */ 726 if (ip->ip_p == IPPROTO_IGMP) { 727 goto ours; 728 } 729 IP_STATINC(IP_STAT_CANTFORWARD); 730 } 731 #endif 732 /* 733 * See if we belong to the destination multicast group on the 734 * arrival interface. 735 */ 736 if (!in_multi_group(ip->ip_dst, ifp, 0)) { 737 IP_STATINC(IP_STAT_CANTFORWARD); 738 goto out; 739 } 740 goto ours; 741 } 742 if (ip->ip_dst.s_addr == INADDR_BROADCAST || 743 in_nullhost(ip->ip_dst)) 744 goto ours; 745 746 /* 747 * Not for us; forward if possible and desirable. 748 */ 749 if (ipforwarding == 0) { 750 m_put_rcvif_psref(ifp, &psref); 751 IP_STATINC(IP_STAT_CANTFORWARD); 752 m_freem(m); 753 } else { 754 /* 755 * If ip_dst matched any of my address on !IFF_UP interface, 756 * and there's no IFF_UP interface that matches ip_dst, 757 * send icmp unreach. Forwarding it will result in in-kernel 758 * forwarding loop till TTL goes to 0. 759 */ 760 if (downmatch) { 761 m_put_rcvif_psref(ifp, &psref); 762 icmp_error(m, ICMP_UNREACH, ICMP_UNREACH_HOST, 0, 0); 763 IP_STATINC(IP_STAT_CANTFORWARD); 764 return; 765 } 766 #ifdef IPSEC 767 /* Perform IPsec, if any. */ 768 if (ipsec_used) { 769 SOFTNET_LOCK(); 770 if (ipsec4_input(m, IP_FORWARDING | 771 (ip_directedbcast ? IP_ALLOWBROADCAST : 0)) != 0) { 772 SOFTNET_UNLOCK(); 773 goto out; 774 } 775 SOFTNET_UNLOCK(); 776 } 777 #endif 778 ip_forward(m, srcrt, ifp); 779 m_put_rcvif_psref(ifp, &psref); 780 } 781 return; 782 783 ours: 784 m_put_rcvif_psref(ifp, &psref); 785 ifp = NULL; 786 787 /* 788 * If offset or IP_MF are set, must reassemble. 789 */ 790 if (ip->ip_off & ~htons(IP_DF|IP_RF)) { 791 /* 792 * Pass to IP reassembly mechanism. 793 */ 794 if (ip_reass_packet(&m, ip) != 0) { 795 /* Failed; invalid fragment(s) or packet. */ 796 goto out; 797 } 798 if (m == NULL) { 799 /* More fragments should come; silently return. */ 800 goto out; 801 } 802 /* 803 * Reassembly is done, we have the final packet. 804 * Updated cached data in local variable(s). 805 */ 806 ip = mtod(m, struct ip *); 807 hlen = ip->ip_hl << 2; 808 } 809 810 #ifdef IPSEC 811 /* 812 * Enforce IPsec policy checking if we are seeing last header. 813 * Note that we do not visit this with protocols with PCB layer 814 * code - like UDP/TCP/raw IP. 815 */ 816 if (ipsec_used && 817 (inetsw[ip_protox[ip->ip_p]].pr_flags & PR_LASTHDR) != 0) { 818 SOFTNET_LOCK(); 819 if (ipsec4_input(m, 0) != 0) { 820 SOFTNET_UNLOCK(); 821 goto out; 822 } 823 SOFTNET_UNLOCK(); 824 } 825 #endif 826 827 /* 828 * Switch out to protocol's input routine. 829 */ 830 #if IFA_STATS 831 if (ia && ip) { 832 struct in_ifaddr *_ia; 833 /* 834 * Keep a reference from ip_match_our_address with psref 835 * is expensive, so explore ia here again. 836 */ 837 s = pserialize_read_enter(); 838 _ia = in_get_ia(ip->ip_dst); 839 _ia->ia_ifa.ifa_data.ifad_inbytes += ntohs(ip->ip_len); 840 pserialize_read_exit(s); 841 } 842 #endif 843 IP_STATINC(IP_STAT_DELIVERED); 844 845 const int off = hlen, nh = ip->ip_p; 846 847 SOFTNET_LOCK(); 848 (*inetsw[ip_protox[nh]].pr_input)(m, off, nh); 849 SOFTNET_UNLOCK(); 850 return; 851 852 out: 853 m_put_rcvif_psref(ifp, &psref); 854 if (m != NULL) 855 m_freem(m); 856 } 857 858 /* 859 * IP timer processing. 860 */ 861 void 862 ip_slowtimo(void) 863 { 864 865 mutex_enter(softnet_lock); 866 KERNEL_LOCK(1, NULL); 867 868 ip_reass_slowtimo(); 869 870 KERNEL_UNLOCK_ONE(NULL); 871 mutex_exit(softnet_lock); 872 } 873 874 /* 875 * IP drain processing. 876 */ 877 void 878 ip_drain(void) 879 { 880 881 KERNEL_LOCK(1, NULL); 882 ip_reass_drain(); 883 KERNEL_UNLOCK_ONE(NULL); 884 } 885 886 /* 887 * ip_dooptions: perform option processing on a datagram, possibly discarding 888 * it if bad options are encountered, or forwarding it if source-routed. 889 * 890 * => Returns true if packet has been forwarded/freed. 891 * => Returns false if the packet should be processed further. 892 */ 893 static bool 894 ip_dooptions(struct mbuf *m) 895 { 896 struct ip *ip = mtod(m, struct ip *); 897 u_char *cp, *cp0; 898 struct ip_timestamp *ipt; 899 struct in_ifaddr *ia; 900 int opt, optlen, cnt, off, code, type = ICMP_PARAMPROB, forward = 0; 901 struct in_addr dst; 902 n_time ntime; 903 struct ifaddr *ifa; 904 int s; 905 906 dst = ip->ip_dst; 907 cp = (u_char *)(ip + 1); 908 cnt = (ip->ip_hl << 2) - sizeof (struct ip); 909 for (; cnt > 0; cnt -= optlen, cp += optlen) { 910 opt = cp[IPOPT_OPTVAL]; 911 if (opt == IPOPT_EOL) 912 break; 913 if (opt == IPOPT_NOP) 914 optlen = 1; 915 else { 916 if (cnt < IPOPT_OLEN + sizeof(*cp)) { 917 code = &cp[IPOPT_OLEN] - (u_char *)ip; 918 goto bad; 919 } 920 optlen = cp[IPOPT_OLEN]; 921 if (optlen < IPOPT_OLEN + sizeof(*cp) || optlen > cnt) { 922 code = &cp[IPOPT_OLEN] - (u_char *)ip; 923 goto bad; 924 } 925 } 926 switch (opt) { 927 928 default: 929 break; 930 931 /* 932 * Source routing with record. 933 * Find interface with current destination address. 934 * If none on this machine then drop if strictly routed, 935 * or do nothing if loosely routed. 936 * Record interface address and bring up next address 937 * component. If strictly routed make sure next 938 * address is on directly accessible net. 939 */ 940 case IPOPT_LSRR: 941 case IPOPT_SSRR: 942 if (ip_allowsrcrt == 0) { 943 type = ICMP_UNREACH; 944 code = ICMP_UNREACH_NET_PROHIB; 945 goto bad; 946 } 947 if (optlen < IPOPT_OFFSET + sizeof(*cp)) { 948 code = &cp[IPOPT_OLEN] - (u_char *)ip; 949 goto bad; 950 } 951 if ((off = cp[IPOPT_OFFSET]) < IPOPT_MINOFF) { 952 code = &cp[IPOPT_OFFSET] - (u_char *)ip; 953 goto bad; 954 } 955 ipaddr.sin_addr = ip->ip_dst; 956 957 s = pserialize_read_enter(); 958 ifa = ifa_ifwithaddr(sintosa(&ipaddr)); 959 if (ifa == NULL) { 960 pserialize_read_exit(s); 961 if (opt == IPOPT_SSRR) { 962 type = ICMP_UNREACH; 963 code = ICMP_UNREACH_SRCFAIL; 964 goto bad; 965 } 966 /* 967 * Loose routing, and not at next destination 968 * yet; nothing to do except forward. 969 */ 970 break; 971 } 972 pserialize_read_exit(s); 973 974 off--; /* 0 origin */ 975 if ((off + sizeof(struct in_addr)) > optlen) { 976 /* 977 * End of source route. Should be for us. 978 */ 979 save_rte(cp, ip->ip_src); 980 break; 981 } 982 /* 983 * locate outgoing interface 984 */ 985 memcpy((void *)&ipaddr.sin_addr, (void *)(cp + off), 986 sizeof(ipaddr.sin_addr)); 987 s = pserialize_read_enter(); 988 if (opt == IPOPT_SSRR) { 989 ifa = ifa_ifwithladdr(sintosa(&ipaddr)); 990 if (ifa != NULL) 991 ia = ifatoia(ifa); 992 else 993 ia = NULL; 994 } else { 995 ia = ip_rtaddr(ipaddr.sin_addr); 996 } 997 if (ia == NULL) { 998 type = ICMP_UNREACH; 999 code = ICMP_UNREACH_SRCFAIL; 1000 pserialize_read_exit(s); 1001 goto bad; 1002 } 1003 ip->ip_dst = ipaddr.sin_addr; 1004 bcopy((void *)&ia->ia_addr.sin_addr, 1005 (void *)(cp + off), sizeof(struct in_addr)); 1006 pserialize_read_exit(s); 1007 cp[IPOPT_OFFSET] += sizeof(struct in_addr); 1008 /* 1009 * Let ip_intr's mcast routing check handle mcast pkts 1010 */ 1011 forward = !IN_MULTICAST(ip->ip_dst.s_addr); 1012 break; 1013 1014 case IPOPT_RR: 1015 if (optlen < IPOPT_OFFSET + sizeof(*cp)) { 1016 code = &cp[IPOPT_OLEN] - (u_char *)ip; 1017 goto bad; 1018 } 1019 if ((off = cp[IPOPT_OFFSET]) < IPOPT_MINOFF) { 1020 code = &cp[IPOPT_OFFSET] - (u_char *)ip; 1021 goto bad; 1022 } 1023 /* 1024 * If no space remains, ignore. 1025 */ 1026 off--; /* 0 origin */ 1027 if ((off + sizeof(struct in_addr)) > optlen) 1028 break; 1029 memcpy((void *)&ipaddr.sin_addr, (void *)(&ip->ip_dst), 1030 sizeof(ipaddr.sin_addr)); 1031 /* 1032 * locate outgoing interface; if we're the destination, 1033 * use the incoming interface (should be same). 1034 */ 1035 s = pserialize_read_enter(); 1036 ifa = ifa_ifwithaddr(sintosa(&ipaddr)); 1037 if (ifa == NULL) { 1038 ia = ip_rtaddr(ipaddr.sin_addr); 1039 if (ia == NULL) { 1040 pserialize_read_exit(s); 1041 type = ICMP_UNREACH; 1042 code = ICMP_UNREACH_HOST; 1043 goto bad; 1044 } 1045 } else { 1046 ia = ifatoia(ifa); 1047 } 1048 bcopy((void *)&ia->ia_addr.sin_addr, 1049 (void *)(cp + off), sizeof(struct in_addr)); 1050 pserialize_read_exit(s); 1051 cp[IPOPT_OFFSET] += sizeof(struct in_addr); 1052 break; 1053 1054 case IPOPT_TS: 1055 code = cp - (u_char *)ip; 1056 ipt = (struct ip_timestamp *)cp; 1057 if (ipt->ipt_len < 4 || ipt->ipt_len > 40) { 1058 code = (u_char *)&ipt->ipt_len - (u_char *)ip; 1059 goto bad; 1060 } 1061 if (ipt->ipt_ptr < 5) { 1062 code = (u_char *)&ipt->ipt_ptr - (u_char *)ip; 1063 goto bad; 1064 } 1065 if (ipt->ipt_ptr > ipt->ipt_len - sizeof (int32_t)) { 1066 if (++ipt->ipt_oflw == 0) { 1067 code = (u_char *)&ipt->ipt_ptr - 1068 (u_char *)ip; 1069 goto bad; 1070 } 1071 break; 1072 } 1073 cp0 = (cp + ipt->ipt_ptr - 1); 1074 switch (ipt->ipt_flg) { 1075 1076 case IPOPT_TS_TSONLY: 1077 break; 1078 1079 case IPOPT_TS_TSANDADDR: { 1080 struct ifnet *rcvif; 1081 int _s, _ss; 1082 1083 if (ipt->ipt_ptr - 1 + sizeof(n_time) + 1084 sizeof(struct in_addr) > ipt->ipt_len) { 1085 code = (u_char *)&ipt->ipt_ptr - 1086 (u_char *)ip; 1087 goto bad; 1088 } 1089 ipaddr.sin_addr = dst; 1090 _ss = pserialize_read_enter(); 1091 rcvif = m_get_rcvif(m, &_s); 1092 ifa = ifaof_ifpforaddr(sintosa(&ipaddr), rcvif); 1093 m_put_rcvif(rcvif, &_s); 1094 if (ifa == NULL) { 1095 pserialize_read_exit(_ss); 1096 break; 1097 } 1098 ia = ifatoia(ifa); 1099 bcopy(&ia->ia_addr.sin_addr, 1100 cp0, sizeof(struct in_addr)); 1101 pserialize_read_exit(_ss); 1102 ipt->ipt_ptr += sizeof(struct in_addr); 1103 break; 1104 } 1105 1106 case IPOPT_TS_PRESPEC: 1107 if (ipt->ipt_ptr - 1 + sizeof(n_time) + 1108 sizeof(struct in_addr) > ipt->ipt_len) { 1109 code = (u_char *)&ipt->ipt_ptr - 1110 (u_char *)ip; 1111 goto bad; 1112 } 1113 memcpy(&ipaddr.sin_addr, cp0, 1114 sizeof(struct in_addr)); 1115 s = pserialize_read_enter(); 1116 ifa = ifa_ifwithaddr(sintosa(&ipaddr)); 1117 if (ifa == NULL) { 1118 pserialize_read_exit(s); 1119 continue; 1120 } 1121 pserialize_read_exit(s); 1122 ipt->ipt_ptr += sizeof(struct in_addr); 1123 break; 1124 1125 default: 1126 /* XXX can't take &ipt->ipt_flg */ 1127 code = (u_char *)&ipt->ipt_ptr - 1128 (u_char *)ip + 1; 1129 goto bad; 1130 } 1131 ntime = iptime(); 1132 cp0 = (u_char *) &ntime; /* XXX grumble, GCC... */ 1133 memmove((char *)cp + ipt->ipt_ptr - 1, cp0, 1134 sizeof(n_time)); 1135 ipt->ipt_ptr += sizeof(n_time); 1136 } 1137 } 1138 if (forward) { 1139 struct ifnet *rcvif; 1140 struct psref _psref; 1141 1142 if (ip_forwsrcrt == 0) { 1143 type = ICMP_UNREACH; 1144 code = ICMP_UNREACH_SRCFAIL; 1145 goto bad; 1146 } 1147 1148 rcvif = m_get_rcvif_psref(m, &_psref); 1149 if (__predict_false(rcvif == NULL)) { 1150 type = ICMP_UNREACH; 1151 code = ICMP_UNREACH_HOST; 1152 goto bad; 1153 } 1154 ip_forward(m, 1, rcvif); 1155 m_put_rcvif_psref(rcvif, &_psref); 1156 return true; 1157 } 1158 return false; 1159 bad: 1160 icmp_error(m, type, code, 0, 0); 1161 IP_STATINC(IP_STAT_BADOPTIONS); 1162 return true; 1163 } 1164 1165 /* 1166 * ip_rtaddr: given address of next destination (final or next hop), 1167 * return internet address info of interface to be used to get there. 1168 */ 1169 static struct in_ifaddr * 1170 ip_rtaddr(struct in_addr dst) 1171 { 1172 struct rtentry *rt; 1173 union { 1174 struct sockaddr dst; 1175 struct sockaddr_in dst4; 1176 } u; 1177 1178 sockaddr_in_init(&u.dst4, &dst, 0); 1179 1180 SOFTNET_LOCK(); 1181 rt = rtcache_lookup(&ipforward_rt, &u.dst); 1182 SOFTNET_UNLOCK(); 1183 if (rt == NULL) 1184 return NULL; 1185 1186 return ifatoia(rt->rt_ifa); 1187 } 1188 1189 /* 1190 * save_rte: save incoming source route for use in replies, to be picked 1191 * up later by ip_srcroute if the receiver is interested. 1192 */ 1193 static void 1194 save_rte(u_char *option, struct in_addr dst) 1195 { 1196 unsigned olen; 1197 1198 olen = option[IPOPT_OLEN]; 1199 if (olen > sizeof(ip_srcrt) - (1 + sizeof(dst))) 1200 return; 1201 memcpy((void *)ip_srcrt.srcopt, (void *)option, olen); 1202 ip_nhops = (olen - IPOPT_OFFSET - 1) / sizeof(struct in_addr); 1203 ip_srcrt.dst = dst; 1204 } 1205 1206 /* 1207 * Retrieve incoming source route for use in replies, 1208 * in the same form used by setsockopt. 1209 * The first hop is placed before the options, will be removed later. 1210 */ 1211 struct mbuf * 1212 ip_srcroute(void) 1213 { 1214 struct in_addr *p, *q; 1215 struct mbuf *m; 1216 1217 if (ip_nhops == 0) 1218 return NULL; 1219 m = m_get(M_DONTWAIT, MT_SOOPTS); 1220 if (m == 0) 1221 return NULL; 1222 1223 MCLAIM(m, &inetdomain.dom_mowner); 1224 #define OPTSIZ (sizeof(ip_srcrt.nop) + sizeof(ip_srcrt.srcopt)) 1225 1226 /* length is (nhops+1)*sizeof(addr) + sizeof(nop + srcrt header) */ 1227 m->m_len = ip_nhops * sizeof(struct in_addr) + sizeof(struct in_addr) + 1228 OPTSIZ; 1229 1230 /* 1231 * First save first hop for return route 1232 */ 1233 p = &ip_srcrt.route[ip_nhops - 1]; 1234 *(mtod(m, struct in_addr *)) = *p--; 1235 1236 /* 1237 * Copy option fields and padding (nop) to mbuf. 1238 */ 1239 ip_srcrt.nop = IPOPT_NOP; 1240 ip_srcrt.srcopt[IPOPT_OFFSET] = IPOPT_MINOFF; 1241 memmove(mtod(m, char *) + sizeof(struct in_addr), &ip_srcrt.nop, 1242 OPTSIZ); 1243 q = (struct in_addr *)(mtod(m, char *) + 1244 sizeof(struct in_addr) + OPTSIZ); 1245 #undef OPTSIZ 1246 /* 1247 * Record return path as an IP source route, 1248 * reversing the path (pointers are now aligned). 1249 */ 1250 while (p >= ip_srcrt.route) { 1251 *q++ = *p--; 1252 } 1253 /* 1254 * Last hop goes to final destination. 1255 */ 1256 *q = ip_srcrt.dst; 1257 return (m); 1258 } 1259 1260 const int inetctlerrmap[PRC_NCMDS] = { 1261 [PRC_MSGSIZE] = EMSGSIZE, 1262 [PRC_HOSTDEAD] = EHOSTDOWN, 1263 [PRC_HOSTUNREACH] = EHOSTUNREACH, 1264 [PRC_UNREACH_NET] = EHOSTUNREACH, 1265 [PRC_UNREACH_HOST] = EHOSTUNREACH, 1266 [PRC_UNREACH_PROTOCOL] = ECONNREFUSED, 1267 [PRC_UNREACH_PORT] = ECONNREFUSED, 1268 [PRC_UNREACH_SRCFAIL] = EHOSTUNREACH, 1269 [PRC_PARAMPROB] = ENOPROTOOPT, 1270 }; 1271 1272 void 1273 ip_fasttimo(void) 1274 { 1275 if (ip_drainwanted) { 1276 ip_drain(); 1277 ip_drainwanted = 0; 1278 } 1279 } 1280 1281 void 1282 ip_drainstub(void) 1283 { 1284 ip_drainwanted = 1; 1285 } 1286 1287 /* 1288 * Forward a packet. If some error occurs return the sender 1289 * an icmp packet. Note we can't always generate a meaningful 1290 * icmp message because icmp doesn't have a large enough repertoire 1291 * of codes and types. 1292 * 1293 * If not forwarding, just drop the packet. This could be confusing 1294 * if ipforwarding was zero but some routing protocol was advancing 1295 * us as a gateway to somewhere. However, we must let the routing 1296 * protocol deal with that. 1297 * 1298 * The srcrt parameter indicates whether the packet is being forwarded 1299 * via a source route. 1300 */ 1301 static void 1302 ip_forward(struct mbuf *m, int srcrt, struct ifnet *rcvif) 1303 { 1304 struct ip *ip = mtod(m, struct ip *); 1305 struct rtentry *rt; 1306 int error, type = 0, code = 0, destmtu = 0; 1307 struct mbuf *mcopy; 1308 n_long dest; 1309 union { 1310 struct sockaddr dst; 1311 struct sockaddr_in dst4; 1312 } u; 1313 uint64_t *ips; 1314 1315 KASSERTMSG(cpu_softintr_p(), "ip_forward: not in the software " 1316 "interrupt handler; synchronization assumptions violated"); 1317 1318 /* 1319 * We are now in the output path. 1320 */ 1321 MCLAIM(m, &ip_tx_mowner); 1322 1323 /* 1324 * Clear any in-bound checksum flags for this packet. 1325 */ 1326 m->m_pkthdr.csum_flags = 0; 1327 1328 dest = 0; 1329 if (m->m_flags & (M_BCAST|M_MCAST) || in_canforward(ip->ip_dst) == 0) { 1330 IP_STATINC(IP_STAT_CANTFORWARD); 1331 m_freem(m); 1332 return; 1333 } 1334 1335 if (ip->ip_ttl <= IPTTLDEC) { 1336 icmp_error(m, ICMP_TIMXCEED, ICMP_TIMXCEED_INTRANS, dest, 0); 1337 return; 1338 } 1339 1340 sockaddr_in_init(&u.dst4, &ip->ip_dst, 0); 1341 1342 if ((rt = rtcache_lookup(&ipforward_rt, &u.dst)) == NULL) { 1343 icmp_error(m, ICMP_UNREACH, ICMP_UNREACH_NET, dest, 0); 1344 return; 1345 } 1346 1347 /* 1348 * Save at most 68 bytes of the packet in case 1349 * we need to generate an ICMP message to the src. 1350 * Pullup to avoid sharing mbuf cluster between m and mcopy. 1351 */ 1352 mcopy = m_copym(m, 0, imin(ntohs(ip->ip_len), 68), M_DONTWAIT); 1353 if (mcopy) 1354 mcopy = m_pullup(mcopy, ip->ip_hl << 2); 1355 1356 ip->ip_ttl -= IPTTLDEC; 1357 1358 /* 1359 * If forwarding packet using same interface that it came in on, 1360 * perhaps should send a redirect to sender to shortcut a hop. 1361 * Only send redirect if source is sending directly to us, 1362 * and if packet was not source routed (or has any options). 1363 * Also, don't send redirect if forwarding using a default route 1364 * or a route modified by a redirect. 1365 */ 1366 if (rt->rt_ifp == rcvif && 1367 (rt->rt_flags & (RTF_DYNAMIC|RTF_MODIFIED)) == 0 && 1368 !in_nullhost(satocsin(rt_getkey(rt))->sin_addr) && 1369 ipsendredirects && !srcrt) { 1370 if (rt->rt_ifa && 1371 (ip->ip_src.s_addr & ifatoia(rt->rt_ifa)->ia_subnetmask) == 1372 ifatoia(rt->rt_ifa)->ia_subnet) { 1373 if (rt->rt_flags & RTF_GATEWAY) 1374 dest = satosin(rt->rt_gateway)->sin_addr.s_addr; 1375 else 1376 dest = ip->ip_dst.s_addr; 1377 /* 1378 * Router requirements says to only send host 1379 * redirects. 1380 */ 1381 type = ICMP_REDIRECT; 1382 code = ICMP_REDIRECT_HOST; 1383 } 1384 } 1385 1386 error = ip_output(m, NULL, &ipforward_rt, 1387 (IP_FORWARDING | (ip_directedbcast ? IP_ALLOWBROADCAST : 0)), 1388 NULL, NULL); 1389 1390 if (error) { 1391 IP_STATINC(IP_STAT_CANTFORWARD); 1392 goto error; 1393 } 1394 1395 ips = IP_STAT_GETREF(); 1396 ips[IP_STAT_FORWARD]++; 1397 1398 if (type) { 1399 ips[IP_STAT_REDIRECTSENT]++; 1400 IP_STAT_PUTREF(); 1401 goto redirect; 1402 } 1403 1404 IP_STAT_PUTREF(); 1405 if (mcopy) { 1406 #ifdef GATEWAY 1407 if (mcopy->m_flags & M_CANFASTFWD) 1408 ipflow_create(&ipforward_rt, mcopy); 1409 #endif 1410 m_freem(mcopy); 1411 } 1412 1413 percpu_putref(ipforward_rt_percpu); 1414 return; 1415 1416 redirect: 1417 error: 1418 if (mcopy == NULL) { 1419 percpu_putref(ipforward_rt_percpu); 1420 return; 1421 } 1422 1423 switch (error) { 1424 1425 case 0: /* forwarded, but need redirect */ 1426 /* type, code set above */ 1427 break; 1428 1429 case ENETUNREACH: /* shouldn't happen, checked above */ 1430 case EHOSTUNREACH: 1431 case ENETDOWN: 1432 case EHOSTDOWN: 1433 default: 1434 type = ICMP_UNREACH; 1435 code = ICMP_UNREACH_HOST; 1436 break; 1437 1438 case EMSGSIZE: 1439 type = ICMP_UNREACH; 1440 code = ICMP_UNREACH_NEEDFRAG; 1441 1442 if ((rt = rtcache_validate(&ipforward_rt)) != NULL) 1443 destmtu = rt->rt_ifp->if_mtu; 1444 #ifdef IPSEC 1445 if (ipsec_used) 1446 (void)ipsec4_forward(mcopy, &destmtu); 1447 #endif 1448 IP_STATINC(IP_STAT_CANTFRAG); 1449 break; 1450 1451 case ENOBUFS: 1452 /* 1453 * Do not generate ICMP_SOURCEQUENCH as required in RFC 1812, 1454 * Requirements for IP Version 4 Routers. Source quench can 1455 * big problem under DoS attacks or if the underlying 1456 * interface is rate-limited. 1457 */ 1458 if (mcopy) 1459 m_freem(mcopy); 1460 percpu_putref(ipforward_rt_percpu); 1461 return; 1462 } 1463 icmp_error(mcopy, type, code, dest, destmtu); 1464 percpu_putref(ipforward_rt_percpu); 1465 } 1466 1467 void 1468 ip_savecontrol(struct inpcb *inp, struct mbuf **mp, struct ip *ip, 1469 struct mbuf *m) 1470 { 1471 struct socket *so = inp->inp_socket; 1472 ifnet_t *ifp; 1473 int inpflags = inp->inp_flags; 1474 struct psref psref; 1475 1476 ifp = m_get_rcvif_psref(m, &psref); 1477 if (__predict_false(ifp == NULL)) 1478 return; /* XXX should report error? */ 1479 1480 if (so->so_options & SO_TIMESTAMP 1481 #ifdef SO_OTIMESTAMP 1482 || so->so_options & SO_OTIMESTAMP 1483 #endif 1484 ) { 1485 struct timeval tv; 1486 1487 microtime(&tv); 1488 #ifdef SO_OTIMESTAMP 1489 if (so->so_options & SO_OTIMESTAMP) { 1490 struct timeval50 tv50; 1491 timeval_to_timeval50(&tv, &tv50); 1492 *mp = sbcreatecontrol((void *) &tv50, sizeof(tv50), 1493 SCM_OTIMESTAMP, SOL_SOCKET); 1494 } else 1495 #endif 1496 *mp = sbcreatecontrol((void *) &tv, sizeof(tv), 1497 SCM_TIMESTAMP, SOL_SOCKET); 1498 if (*mp) 1499 mp = &(*mp)->m_next; 1500 } 1501 if (inpflags & INP_RECVDSTADDR) { 1502 *mp = sbcreatecontrol((void *) &ip->ip_dst, 1503 sizeof(struct in_addr), IP_RECVDSTADDR, IPPROTO_IP); 1504 if (*mp) 1505 mp = &(*mp)->m_next; 1506 } 1507 if (inpflags & INP_RECVPKTINFO) { 1508 struct in_pktinfo ipi; 1509 ipi.ipi_addr = ip->ip_src; 1510 ipi.ipi_ifindex = ifp->if_index; 1511 *mp = sbcreatecontrol((void *) &ipi, 1512 sizeof(ipi), IP_RECVPKTINFO, IPPROTO_IP); 1513 if (*mp) 1514 mp = &(*mp)->m_next; 1515 } 1516 if (inpflags & INP_PKTINFO) { 1517 struct in_pktinfo ipi; 1518 ipi.ipi_addr = ip->ip_dst; 1519 ipi.ipi_ifindex = ifp->if_index; 1520 *mp = sbcreatecontrol((void *) &ipi, 1521 sizeof(ipi), IP_PKTINFO, IPPROTO_IP); 1522 if (*mp) 1523 mp = &(*mp)->m_next; 1524 } 1525 if (inpflags & INP_RECVIF) { 1526 struct sockaddr_dl sdl; 1527 1528 sockaddr_dl_init(&sdl, sizeof(sdl), ifp->if_index, 0, NULL, 0, 1529 NULL, 0); 1530 *mp = sbcreatecontrol(&sdl, sdl.sdl_len, IP_RECVIF, IPPROTO_IP); 1531 if (*mp) 1532 mp = &(*mp)->m_next; 1533 } 1534 if (inpflags & INP_RECVTTL) { 1535 *mp = sbcreatecontrol((void *) &ip->ip_ttl, 1536 sizeof(uint8_t), IP_RECVTTL, IPPROTO_IP); 1537 if (*mp) 1538 mp = &(*mp)->m_next; 1539 } 1540 m_put_rcvif_psref(ifp, &psref); 1541 } 1542 1543 /* 1544 * sysctl helper routine for net.inet.ip.forwsrcrt. 1545 */ 1546 static int 1547 sysctl_net_inet_ip_forwsrcrt(SYSCTLFN_ARGS) 1548 { 1549 int error, tmp; 1550 struct sysctlnode node; 1551 1552 node = *rnode; 1553 tmp = ip_forwsrcrt; 1554 node.sysctl_data = &tmp; 1555 error = sysctl_lookup(SYSCTLFN_CALL(&node)); 1556 if (error || newp == NULL) 1557 return (error); 1558 1559 error = kauth_authorize_network(l->l_cred, KAUTH_NETWORK_FORWSRCRT, 1560 0, NULL, NULL, NULL); 1561 if (error) 1562 return (error); 1563 1564 ip_forwsrcrt = tmp; 1565 1566 return (0); 1567 } 1568 1569 /* 1570 * sysctl helper routine for net.inet.ip.mtudisctimeout. checks the 1571 * range of the new value and tweaks timers if it changes. 1572 */ 1573 static int 1574 sysctl_net_inet_ip_pmtudto(SYSCTLFN_ARGS) 1575 { 1576 int error, tmp; 1577 struct sysctlnode node; 1578 1579 node = *rnode; 1580 tmp = ip_mtudisc_timeout; 1581 node.sysctl_data = &tmp; 1582 error = sysctl_lookup(SYSCTLFN_CALL(&node)); 1583 if (error || newp == NULL) 1584 return (error); 1585 if (tmp < 0) 1586 return (EINVAL); 1587 1588 mutex_enter(softnet_lock); 1589 1590 ip_mtudisc_timeout = tmp; 1591 rt_timer_queue_change(ip_mtudisc_timeout_q, ip_mtudisc_timeout); 1592 1593 mutex_exit(softnet_lock); 1594 1595 return (0); 1596 } 1597 1598 static int 1599 sysctl_net_inet_ip_stats(SYSCTLFN_ARGS) 1600 { 1601 1602 return (NETSTAT_SYSCTL(ipstat_percpu, IP_NSTATS)); 1603 } 1604 1605 static void 1606 sysctl_net_inet_ip_setup(struct sysctllog **clog) 1607 { 1608 sysctl_createv(clog, 0, NULL, NULL, 1609 CTLFLAG_PERMANENT, 1610 CTLTYPE_NODE, "inet", 1611 SYSCTL_DESCR("PF_INET related settings"), 1612 NULL, 0, NULL, 0, 1613 CTL_NET, PF_INET, CTL_EOL); 1614 sysctl_createv(clog, 0, NULL, NULL, 1615 CTLFLAG_PERMANENT, 1616 CTLTYPE_NODE, "ip", 1617 SYSCTL_DESCR("IPv4 related settings"), 1618 NULL, 0, NULL, 0, 1619 CTL_NET, PF_INET, IPPROTO_IP, CTL_EOL); 1620 1621 sysctl_createv(clog, 0, NULL, NULL, 1622 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1623 CTLTYPE_INT, "forwarding", 1624 SYSCTL_DESCR("Enable forwarding of INET datagrams"), 1625 NULL, 0, &ipforwarding, 0, 1626 CTL_NET, PF_INET, IPPROTO_IP, 1627 IPCTL_FORWARDING, CTL_EOL); 1628 sysctl_createv(clog, 0, NULL, NULL, 1629 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1630 CTLTYPE_INT, "redirect", 1631 SYSCTL_DESCR("Enable sending of ICMP redirect messages"), 1632 NULL, 0, &ipsendredirects, 0, 1633 CTL_NET, PF_INET, IPPROTO_IP, 1634 IPCTL_SENDREDIRECTS, CTL_EOL); 1635 sysctl_createv(clog, 0, NULL, NULL, 1636 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1637 CTLTYPE_INT, "ttl", 1638 SYSCTL_DESCR("Default TTL for an INET datagram"), 1639 NULL, 0, &ip_defttl, 0, 1640 CTL_NET, PF_INET, IPPROTO_IP, 1641 IPCTL_DEFTTL, CTL_EOL); 1642 #ifdef IPCTL_DEFMTU 1643 sysctl_createv(clog, 0, NULL, NULL, 1644 CTLFLAG_PERMANENT /* |CTLFLAG_READWRITE? */, 1645 CTLTYPE_INT, "mtu", 1646 SYSCTL_DESCR("Default MTA for an INET route"), 1647 NULL, 0, &ip_mtu, 0, 1648 CTL_NET, PF_INET, IPPROTO_IP, 1649 IPCTL_DEFMTU, CTL_EOL); 1650 #endif /* IPCTL_DEFMTU */ 1651 sysctl_createv(clog, 0, NULL, NULL, 1652 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1653 CTLTYPE_INT, "forwsrcrt", 1654 SYSCTL_DESCR("Enable forwarding of source-routed " 1655 "datagrams"), 1656 sysctl_net_inet_ip_forwsrcrt, 0, &ip_forwsrcrt, 0, 1657 CTL_NET, PF_INET, IPPROTO_IP, 1658 IPCTL_FORWSRCRT, CTL_EOL); 1659 sysctl_createv(clog, 0, NULL, NULL, 1660 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1661 CTLTYPE_INT, "directed-broadcast", 1662 SYSCTL_DESCR("Enable forwarding of broadcast datagrams"), 1663 NULL, 0, &ip_directedbcast, 0, 1664 CTL_NET, PF_INET, IPPROTO_IP, 1665 IPCTL_DIRECTEDBCAST, CTL_EOL); 1666 sysctl_createv(clog, 0, NULL, NULL, 1667 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1668 CTLTYPE_INT, "allowsrcrt", 1669 SYSCTL_DESCR("Accept source-routed datagrams"), 1670 NULL, 0, &ip_allowsrcrt, 0, 1671 CTL_NET, PF_INET, IPPROTO_IP, 1672 IPCTL_ALLOWSRCRT, CTL_EOL); 1673 1674 sysctl_createv(clog, 0, NULL, NULL, 1675 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1676 CTLTYPE_INT, "mtudisc", 1677 SYSCTL_DESCR("Use RFC1191 Path MTU Discovery"), 1678 NULL, 0, &ip_mtudisc, 0, 1679 CTL_NET, PF_INET, IPPROTO_IP, 1680 IPCTL_MTUDISC, CTL_EOL); 1681 sysctl_createv(clog, 0, NULL, NULL, 1682 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1683 CTLTYPE_INT, "anonportmin", 1684 SYSCTL_DESCR("Lowest ephemeral port number to assign"), 1685 sysctl_net_inet_ip_ports, 0, &anonportmin, 0, 1686 CTL_NET, PF_INET, IPPROTO_IP, 1687 IPCTL_ANONPORTMIN, CTL_EOL); 1688 sysctl_createv(clog, 0, NULL, NULL, 1689 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1690 CTLTYPE_INT, "anonportmax", 1691 SYSCTL_DESCR("Highest ephemeral port number to assign"), 1692 sysctl_net_inet_ip_ports, 0, &anonportmax, 0, 1693 CTL_NET, PF_INET, IPPROTO_IP, 1694 IPCTL_ANONPORTMAX, CTL_EOL); 1695 sysctl_createv(clog, 0, NULL, NULL, 1696 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1697 CTLTYPE_INT, "mtudisctimeout", 1698 SYSCTL_DESCR("Lifetime of a Path MTU Discovered route"), 1699 sysctl_net_inet_ip_pmtudto, 0, (void *)&ip_mtudisc_timeout, 0, 1700 CTL_NET, PF_INET, IPPROTO_IP, 1701 IPCTL_MTUDISCTIMEOUT, CTL_EOL); 1702 #ifndef IPNOPRIVPORTS 1703 sysctl_createv(clog, 0, NULL, NULL, 1704 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1705 CTLTYPE_INT, "lowportmin", 1706 SYSCTL_DESCR("Lowest privileged ephemeral port number " 1707 "to assign"), 1708 sysctl_net_inet_ip_ports, 0, &lowportmin, 0, 1709 CTL_NET, PF_INET, IPPROTO_IP, 1710 IPCTL_LOWPORTMIN, CTL_EOL); 1711 sysctl_createv(clog, 0, NULL, NULL, 1712 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1713 CTLTYPE_INT, "lowportmax", 1714 SYSCTL_DESCR("Highest privileged ephemeral port number " 1715 "to assign"), 1716 sysctl_net_inet_ip_ports, 0, &lowportmax, 0, 1717 CTL_NET, PF_INET, IPPROTO_IP, 1718 IPCTL_LOWPORTMAX, CTL_EOL); 1719 #endif /* IPNOPRIVPORTS */ 1720 #if NGRE > 0 1721 sysctl_createv(clog, 0, NULL, NULL, 1722 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1723 CTLTYPE_INT, "grettl", 1724 SYSCTL_DESCR("Default TTL for a gre tunnel datagram"), 1725 NULL, 0, &ip_gre_ttl, 0, 1726 CTL_NET, PF_INET, IPPROTO_IP, 1727 IPCTL_GRE_TTL, CTL_EOL); 1728 #endif /* NGRE */ 1729 sysctl_createv(clog, 0, NULL, NULL, 1730 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1731 CTLTYPE_INT, "checkinterface", 1732 SYSCTL_DESCR("Enable receive side of Strong ES model " 1733 "from RFC1122"), 1734 NULL, 0, &ip_checkinterface, 0, 1735 CTL_NET, PF_INET, IPPROTO_IP, 1736 IPCTL_CHECKINTERFACE, CTL_EOL); 1737 sysctl_createv(clog, 0, NULL, NULL, 1738 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1739 CTLTYPE_INT, "random_id", 1740 SYSCTL_DESCR("Assign random ip_id values"), 1741 NULL, 0, &ip_do_randomid, 0, 1742 CTL_NET, PF_INET, IPPROTO_IP, 1743 IPCTL_RANDOMID, CTL_EOL); 1744 sysctl_createv(clog, 0, NULL, NULL, 1745 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1746 CTLTYPE_INT, "do_loopback_cksum", 1747 SYSCTL_DESCR("Perform IP checksum on loopback"), 1748 NULL, 0, &ip_do_loopback_cksum, 0, 1749 CTL_NET, PF_INET, IPPROTO_IP, 1750 IPCTL_LOOPBACKCKSUM, CTL_EOL); 1751 sysctl_createv(clog, 0, NULL, NULL, 1752 CTLFLAG_PERMANENT, 1753 CTLTYPE_STRUCT, "stats", 1754 SYSCTL_DESCR("IP statistics"), 1755 sysctl_net_inet_ip_stats, 0, NULL, 0, 1756 CTL_NET, PF_INET, IPPROTO_IP, IPCTL_STATS, 1757 CTL_EOL); 1758 #if NARP 1759 sysctl_createv(clog, 0, NULL, NULL, 1760 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1761 CTLTYPE_INT, "dad_count", 1762 SYSCTL_DESCR("Number of Duplicate Address Detection " 1763 "probes to send"), 1764 NULL, 0, &ip_dad_count, 0, 1765 CTL_NET, PF_INET, IPPROTO_IP, 1766 IPCTL_DAD_COUNT, CTL_EOL); 1767 #endif 1768 1769 /* anonportalgo RFC6056 subtree */ 1770 const struct sysctlnode *portalgo_node; 1771 sysctl_createv(clog, 0, NULL, &portalgo_node, 1772 CTLFLAG_PERMANENT, 1773 CTLTYPE_NODE, "anonportalgo", 1774 SYSCTL_DESCR("Anonymous Port Algorithm Selection (RFC 6056)"), 1775 NULL, 0, NULL, 0, 1776 CTL_NET, PF_INET, IPPROTO_IP, CTL_CREATE, CTL_EOL); 1777 sysctl_createv(clog, 0, &portalgo_node, NULL, 1778 CTLFLAG_PERMANENT, 1779 CTLTYPE_STRING, "available", 1780 SYSCTL_DESCR("available algorithms"), 1781 sysctl_portalgo_available, 0, NULL, PORTALGO_MAXLEN, 1782 CTL_CREATE, CTL_EOL); 1783 sysctl_createv(clog, 0, &portalgo_node, NULL, 1784 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1785 CTLTYPE_STRING, "selected", 1786 SYSCTL_DESCR("selected algorithm"), 1787 sysctl_portalgo_selected4, 0, NULL, PORTALGO_MAXLEN, 1788 CTL_CREATE, CTL_EOL); 1789 sysctl_createv(clog, 0, &portalgo_node, NULL, 1790 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1791 CTLTYPE_STRUCT, "reserve", 1792 SYSCTL_DESCR("bitmap of reserved ports"), 1793 sysctl_portalgo_reserve4, 0, NULL, 0, 1794 CTL_CREATE, CTL_EOL); 1795 } 1796 1797 void 1798 ip_statinc(u_int stat) 1799 { 1800 1801 KASSERT(stat < IP_NSTATS); 1802 IP_STATINC(stat); 1803 } 1804
Indexes created Tue Oct 21 19:09:51 GMT 2025