ip_input.c revision 1.347.2.1
1 /* $NetBSD: ip_input.c,v 1.347.2.1 2017/04/21 16:54:06 bouyer 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 1.347.2.1 2017/04/21 16:54:06 bouyer 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 percpu_t *ipforward_rt_percpu __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 * Used 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 struct ip_srcrt { 263 int isr_nhops; /* number of hops */ 264 struct in_addr isr_dst; /* final destination */ 265 char isr_nop; /* one NOP to align */ 266 char isr_hdr[IPOPT_OFFSET + 1]; /* OPTVAL, OLEN & OFFSET */ 267 struct in_addr isr_routes[MAX_IPOPTLEN/sizeof(struct in_addr)]; 268 }; 269 270 static int ip_drainwanted; 271 272 static void save_rte(struct mbuf *, u_char *, struct in_addr); 273 274 #ifdef MBUFTRACE 275 struct mowner ip_rx_mowner = MOWNER_INIT("internet", "rx"); 276 struct mowner ip_tx_mowner = MOWNER_INIT("internet", "tx"); 277 #endif 278 279 static void ipintr(void *); 280 static void ip_input(struct mbuf *); 281 static void ip_forward(struct mbuf *, int, struct ifnet *); 282 static bool ip_dooptions(struct mbuf *); 283 static struct in_ifaddr *ip_rtaddr(struct in_addr, struct psref *); 284 static void sysctl_net_inet_ip_setup(struct sysctllog **); 285 286 static struct in_ifaddr *ip_match_our_address(struct ifnet *, struct ip *, 287 int *); 288 static struct in_ifaddr *ip_match_our_address_broadcast(struct ifnet *, 289 struct ip *); 290 291 #ifdef NET_MPSAFE 292 #define SOFTNET_LOCK() mutex_enter(softnet_lock) 293 #define SOFTNET_UNLOCK() mutex_exit(softnet_lock) 294 #else 295 #define SOFTNET_LOCK() KASSERT(mutex_owned(softnet_lock)) 296 #define SOFTNET_UNLOCK() KASSERT(mutex_owned(softnet_lock)) 297 #endif 298 299 /* 300 * IP initialization: fill in IP protocol switch table. 301 * All protocols not implemented in kernel go to raw IP protocol handler. 302 */ 303 void 304 ip_init(void) 305 { 306 const struct protosw *pr; 307 308 in_init(); 309 sysctl_net_inet_ip_setup(NULL); 310 311 pr = pffindproto(PF_INET, IPPROTO_RAW, SOCK_RAW); 312 KASSERT(pr != NULL); 313 314 ip_pktq = pktq_create(IFQ_MAXLEN, ipintr, NULL); 315 KASSERT(ip_pktq != NULL); 316 317 for (u_int i = 0; i < IPPROTO_MAX; i++) { 318 ip_protox[i] = pr - inetsw; 319 } 320 for (pr = inetdomain.dom_protosw; 321 pr < inetdomain.dom_protoswNPROTOSW; pr++) 322 if (pr->pr_domain->dom_family == PF_INET && 323 pr->pr_protocol && pr->pr_protocol != IPPROTO_RAW) 324 ip_protox[pr->pr_protocol] = pr - inetsw; 325 326 ip_reass_init(); 327 328 ip_ids = ip_id_init(); 329 ip_id = time_uptime & 0xfffff; 330 331 #ifdef GATEWAY 332 ipflow_init(); 333 #endif 334 335 /* Register our Packet Filter hook. */ 336 inet_pfil_hook = pfil_head_create(PFIL_TYPE_AF, (void *)AF_INET); 337 KASSERT(inet_pfil_hook != NULL); 338 339 #ifdef MBUFTRACE 340 MOWNER_ATTACH(&ip_tx_mowner); 341 MOWNER_ATTACH(&ip_rx_mowner); 342 #endif /* MBUFTRACE */ 343 344 ipstat_percpu = percpu_alloc(sizeof(uint64_t) * IP_NSTATS); 345 346 ipforward_rt_percpu = percpu_alloc(sizeof(struct route)); 347 if (ipforward_rt_percpu == NULL) 348 panic("failed to allocate ipforward_rt_percpu"); 349 350 ip_mtudisc_timeout_q = rt_timer_queue_create(ip_mtudisc_timeout); 351 } 352 353 static struct in_ifaddr * 354 ip_match_our_address(struct ifnet *ifp, struct ip *ip, int *downmatch) 355 { 356 struct in_ifaddr *ia = NULL; 357 int checkif; 358 359 /* 360 * Enable a consistency check between the destination address 361 * and the arrival interface for a unicast packet (the RFC 1122 362 * strong ES model) if IP forwarding is disabled and the packet 363 * is not locally generated. 364 * 365 * XXX - Checking also should be disabled if the destination 366 * address is ipnat'ed to a different interface. 367 * 368 * XXX - Checking is incompatible with IP aliases added 369 * to the loopback interface instead of the interface where 370 * the packets are received. 371 * 372 * XXX - We need to add a per ifaddr flag for this so that 373 * we get finer grain control. 374 */ 375 checkif = ip_checkinterface && (ipforwarding == 0) && 376 (ifp->if_flags & IFF_LOOPBACK) == 0; 377 378 IN_ADDRHASH_READER_FOREACH(ia, ip->ip_dst.s_addr) { 379 if (in_hosteq(ia->ia_addr.sin_addr, ip->ip_dst)) { 380 if (ia->ia4_flags & IN_IFF_NOTREADY) 381 continue; 382 if (checkif && ia->ia_ifp != ifp) 383 continue; 384 if ((ia->ia_ifp->if_flags & IFF_UP) != 0 && 385 (ia->ia4_flags & IN_IFF_DETACHED) == 0) 386 break; 387 else 388 (*downmatch)++; 389 } 390 } 391 392 return ia; 393 } 394 395 static struct in_ifaddr * 396 ip_match_our_address_broadcast(struct ifnet *ifp, struct ip *ip) 397 { 398 struct in_ifaddr *ia = NULL; 399 struct ifaddr *ifa; 400 401 IFADDR_READER_FOREACH(ifa, ifp) { 402 if (ifa->ifa_addr->sa_family != AF_INET) 403 continue; 404 ia = ifatoia(ifa); 405 if (ia->ia4_flags & (IN_IFF_NOTREADY | IN_IFF_DETACHED)) 406 continue; 407 if (in_hosteq(ip->ip_dst, ia->ia_broadaddr.sin_addr) || 408 in_hosteq(ip->ip_dst, ia->ia_netbroadcast) || 409 /* 410 * Look for all-0's host part (old broadcast addr), 411 * either for subnet or net. 412 */ 413 ip->ip_dst.s_addr == ia->ia_subnet || 414 ip->ip_dst.s_addr == ia->ia_net) 415 goto matched; 416 /* 417 * An interface with IP address zero accepts 418 * all packets that arrive on that interface. 419 */ 420 if (in_nullhost(ia->ia_addr.sin_addr)) 421 goto matched; 422 } 423 ia = NULL; 424 425 matched: 426 return ia; 427 } 428 429 /* 430 * IP software interrupt routine. 431 */ 432 static void 433 ipintr(void *arg __unused) 434 { 435 struct mbuf *m; 436 437 KASSERT(cpu_softintr_p()); 438 439 #ifndef NET_MPSAFE 440 mutex_enter(softnet_lock); 441 #endif 442 while ((m = pktq_dequeue(ip_pktq)) != NULL) { 443 ip_input(m); 444 } 445 #ifndef NET_MPSAFE 446 mutex_exit(softnet_lock); 447 #endif 448 } 449 450 /* 451 * IP input routine. Checksum and byte swap header. If fragmented 452 * try to reassemble. Process options. Pass to next level. 453 */ 454 static void 455 ip_input(struct mbuf *m) 456 { 457 struct ip *ip = NULL; 458 struct in_ifaddr *ia = NULL; 459 int hlen = 0, len; 460 int downmatch; 461 int srcrt = 0; 462 ifnet_t *ifp; 463 struct psref psref; 464 int s; 465 466 KASSERTMSG(cpu_softintr_p(), "ip_input: not in the software " 467 "interrupt handler; synchronization assumptions violated"); 468 469 MCLAIM(m, &ip_rx_mowner); 470 KASSERT((m->m_flags & M_PKTHDR) != 0); 471 472 ifp = m_get_rcvif_psref(m, &psref); 473 if (__predict_false(ifp == NULL)) 474 goto out; 475 476 /* 477 * If no IP addresses have been set yet but the interfaces 478 * are receiving, can't do anything with incoming packets yet. 479 * Note: we pre-check without locks held. 480 */ 481 if (IN_ADDRLIST_READER_EMPTY()) 482 goto out; 483 IP_STATINC(IP_STAT_TOTAL); 484 485 /* 486 * If the IP header is not aligned, slurp it up into a new 487 * mbuf with space for link headers, in the event we forward 488 * it. Otherwise, if it is aligned, make sure the entire 489 * base IP header is in the first mbuf of the chain. 490 */ 491 if (IP_HDR_ALIGNED_P(mtod(m, void *)) == 0) { 492 if ((m = m_copyup(m, sizeof(struct ip), 493 (max_linkhdr + 3) & ~3)) == NULL) { 494 /* XXXJRT new stat, please */ 495 IP_STATINC(IP_STAT_TOOSMALL); 496 goto out; 497 } 498 } else if (__predict_false(m->m_len < sizeof (struct ip))) { 499 if ((m = m_pullup(m, sizeof (struct ip))) == NULL) { 500 IP_STATINC(IP_STAT_TOOSMALL); 501 goto out; 502 } 503 } 504 ip = mtod(m, struct ip *); 505 if (ip->ip_v != IPVERSION) { 506 IP_STATINC(IP_STAT_BADVERS); 507 goto out; 508 } 509 hlen = ip->ip_hl << 2; 510 if (hlen < sizeof(struct ip)) { /* minimum header length */ 511 IP_STATINC(IP_STAT_BADHLEN); 512 goto out; 513 } 514 if (hlen > m->m_len) { 515 if ((m = m_pullup(m, hlen)) == NULL) { 516 IP_STATINC(IP_STAT_BADHLEN); 517 goto out; 518 } 519 ip = mtod(m, struct ip *); 520 } 521 522 /* 523 * RFC1122: packets with a multicast source address are 524 * not allowed. 525 */ 526 if (IN_MULTICAST(ip->ip_src.s_addr)) { 527 IP_STATINC(IP_STAT_BADADDR); 528 goto out; 529 } 530 531 /* 127/8 must not appear on wire - RFC1122 */ 532 if ((ntohl(ip->ip_dst.s_addr) >> IN_CLASSA_NSHIFT) == IN_LOOPBACKNET || 533 (ntohl(ip->ip_src.s_addr) >> IN_CLASSA_NSHIFT) == IN_LOOPBACKNET) { 534 if ((ifp->if_flags & IFF_LOOPBACK) == 0) { 535 IP_STATINC(IP_STAT_BADADDR); 536 goto out; 537 } 538 } 539 540 switch (m->m_pkthdr.csum_flags & 541 ((ifp->if_csum_flags_rx & M_CSUM_IPv4) | 542 M_CSUM_IPv4_BAD)) { 543 case M_CSUM_IPv4|M_CSUM_IPv4_BAD: 544 INET_CSUM_COUNTER_INCR(&ip_hwcsum_bad); 545 IP_STATINC(IP_STAT_BADSUM); 546 goto out; 547 548 case M_CSUM_IPv4: 549 /* Checksum was okay. */ 550 INET_CSUM_COUNTER_INCR(&ip_hwcsum_ok); 551 break; 552 553 default: 554 /* 555 * Must compute it ourselves. Maybe skip checksum on 556 * loopback interfaces. 557 */ 558 if (__predict_true(!(ifp->if_flags & IFF_LOOPBACK) || 559 ip_do_loopback_cksum)) { 560 INET_CSUM_COUNTER_INCR(&ip_swcsum); 561 if (in_cksum(m, hlen) != 0) { 562 IP_STATINC(IP_STAT_BADSUM); 563 goto out; 564 } 565 } 566 break; 567 } 568 569 /* Retrieve the packet length. */ 570 len = ntohs(ip->ip_len); 571 572 /* 573 * Check for additional length bogosity 574 */ 575 if (len < hlen) { 576 IP_STATINC(IP_STAT_BADLEN); 577 goto out; 578 } 579 580 /* 581 * Check that the amount of data in the buffers 582 * is as at least much as the IP header would have us expect. 583 * Trim mbufs if longer than we expect. 584 * Drop packet if shorter than we expect. 585 */ 586 if (m->m_pkthdr.len < len) { 587 IP_STATINC(IP_STAT_TOOSHORT); 588 goto out; 589 } 590 if (m->m_pkthdr.len > len) { 591 if (m->m_len == m->m_pkthdr.len) { 592 m->m_len = len; 593 m->m_pkthdr.len = len; 594 } else 595 m_adj(m, len - m->m_pkthdr.len); 596 } 597 598 /* 599 * Assume that we can create a fast-forward IP flow entry 600 * based on this packet. 601 */ 602 m->m_flags |= M_CANFASTFWD; 603 604 /* 605 * Run through list of hooks for input packets. If there are any 606 * filters which require that additional packets in the flow are 607 * not fast-forwarded, they must clear the M_CANFASTFWD flag. 608 * Note that filters must _never_ set this flag, as another filter 609 * in the list may have previously cleared it. 610 */ 611 #if defined(IPSEC) 612 if (!ipsec_used || !ipsec_indone(m)) 613 #else 614 if (1) 615 #endif 616 { 617 struct in_addr odst = ip->ip_dst; 618 bool freed; 619 620 freed = pfil_run_hooks(inet_pfil_hook, &m, ifp, PFIL_IN) != 0; 621 if (freed || m == NULL) { 622 m = NULL; 623 goto out; 624 } 625 ip = mtod(m, struct ip *); 626 hlen = ip->ip_hl << 2; 627 628 /* 629 * XXX The setting of "srcrt" here is to prevent ip_forward() 630 * from generating ICMP redirects for packets that have 631 * been redirected by a hook back out on to the same LAN that 632 * they came from and is not an indication that the packet 633 * is being inffluenced by source routing options. This 634 * allows things like 635 * "rdr tlp0 0/0 port 80 -> 1.1.1.200 3128 tcp" 636 * where tlp0 is both on the 1.1.1.0/24 network and is the 637 * default route for hosts on 1.1.1.0/24. Of course this 638 * also requires a "map tlp0 ..." to complete the story. 639 * One might argue whether or not this kind of network config. 640 * should be supported in this manner... 641 */ 642 srcrt = (odst.s_addr != ip->ip_dst.s_addr); 643 } 644 645 #ifdef ALTQ 646 /* XXX Temporary until ALTQ is changed to use a pfil hook */ 647 if (altq_input) { 648 SOFTNET_LOCK(); 649 if ((*altq_input)(m, AF_INET) == 0) { 650 /* Packet dropped by traffic conditioner. */ 651 SOFTNET_UNLOCK(); 652 m = NULL; 653 goto out; 654 } 655 SOFTNET_UNLOCK(); 656 } 657 #endif 658 659 /* 660 * Process options and, if not destined for us, 661 * ship it on. ip_dooptions returns 1 when an 662 * error was detected (causing an icmp message 663 * to be sent and the original packet to be freed). 664 */ 665 if (hlen > sizeof (struct ip) && ip_dooptions(m)) { 666 m = NULL; 667 goto out; 668 } 669 670 /* 671 * Check our list of addresses, to see if the packet is for us. 672 * 673 * Traditional 4.4BSD did not consult IFF_UP at all. 674 * The behavior here is to treat addresses on !IFF_UP interface 675 * or IN_IFF_NOTREADY addresses as not mine. 676 */ 677 downmatch = 0; 678 s = pserialize_read_enter(); 679 ia = ip_match_our_address(ifp, ip, &downmatch); 680 if (ia != NULL) { 681 pserialize_read_exit(s); 682 goto ours; 683 } 684 685 if (ifp->if_flags & IFF_BROADCAST) { 686 ia = ip_match_our_address_broadcast(ifp, ip); 687 if (ia != NULL) { 688 pserialize_read_exit(s); 689 goto ours; 690 } 691 } 692 pserialize_read_exit(s); 693 694 if (IN_MULTICAST(ip->ip_dst.s_addr)) { 695 #ifdef MROUTING 696 extern struct socket *ip_mrouter; 697 698 if (ip_mrouter) { 699 /* 700 * If we are acting as a multicast router, all 701 * incoming multicast packets are passed to the 702 * kernel-level multicast forwarding function. 703 * The packet is returned (relatively) intact; if 704 * ip_mforward() returns a non-zero value, the packet 705 * must be discarded, else it may be accepted below. 706 * 707 * (The IP ident field is put in the same byte order 708 * as expected when ip_mforward() is called from 709 * ip_output().) 710 */ 711 SOFTNET_LOCK(); 712 if (ip_mforward(m, ifp) != 0) { 713 SOFTNET_UNLOCK(); 714 IP_STATINC(IP_STAT_CANTFORWARD); 715 goto out; 716 } 717 SOFTNET_UNLOCK(); 718 719 /* 720 * The process-level routing demon needs to receive 721 * all multicast IGMP packets, whether or not this 722 * host belongs to their destination groups. 723 */ 724 if (ip->ip_p == IPPROTO_IGMP) { 725 goto ours; 726 } 727 IP_STATINC(IP_STAT_CANTFORWARD); 728 } 729 #endif 730 /* 731 * See if we belong to the destination multicast group on the 732 * arrival interface. 733 */ 734 if (!in_multi_group(ip->ip_dst, ifp, 0)) { 735 IP_STATINC(IP_STAT_CANTFORWARD); 736 goto out; 737 } 738 goto ours; 739 } 740 if (ip->ip_dst.s_addr == INADDR_BROADCAST || 741 in_nullhost(ip->ip_dst)) 742 goto ours; 743 744 /* 745 * Not for us; forward if possible and desirable. 746 */ 747 if (ipforwarding == 0) { 748 m_put_rcvif_psref(ifp, &psref); 749 IP_STATINC(IP_STAT_CANTFORWARD); 750 m_freem(m); 751 } else { 752 /* 753 * If ip_dst matched any of my address on !IFF_UP interface, 754 * and there's no IFF_UP interface that matches ip_dst, 755 * send icmp unreach. Forwarding it will result in in-kernel 756 * forwarding loop till TTL goes to 0. 757 */ 758 if (downmatch) { 759 m_put_rcvif_psref(ifp, &psref); 760 icmp_error(m, ICMP_UNREACH, ICMP_UNREACH_HOST, 0, 0); 761 IP_STATINC(IP_STAT_CANTFORWARD); 762 return; 763 } 764 #ifdef IPSEC 765 /* Perform IPsec, if any. */ 766 if (ipsec_used) { 767 SOFTNET_LOCK(); 768 if (ipsec4_input(m, IP_FORWARDING | 769 (ip_directedbcast ? IP_ALLOWBROADCAST : 0)) != 0) { 770 SOFTNET_UNLOCK(); 771 goto out; 772 } 773 SOFTNET_UNLOCK(); 774 } 775 #endif 776 ip_forward(m, srcrt, ifp); 777 m_put_rcvif_psref(ifp, &psref); 778 } 779 return; 780 781 ours: 782 m_put_rcvif_psref(ifp, &psref); 783 ifp = NULL; 784 785 /* 786 * If offset or IP_MF are set, must reassemble. 787 */ 788 if (ip->ip_off & ~htons(IP_DF|IP_RF)) { 789 /* 790 * Pass to IP reassembly mechanism. 791 */ 792 if (ip_reass_packet(&m, ip) != 0) { 793 /* Failed; invalid fragment(s) or packet. */ 794 goto out; 795 } 796 if (m == NULL) { 797 /* More fragments should come; silently return. */ 798 goto out; 799 } 800 /* 801 * Reassembly is done, we have the final packet. 802 * Updated cached data in local variable(s). 803 */ 804 ip = mtod(m, struct ip *); 805 hlen = ip->ip_hl << 2; 806 } 807 808 #ifdef IPSEC 809 /* 810 * Enforce IPsec policy checking if we are seeing last header. 811 * Note that we do not visit this with protocols with PCB layer 812 * code - like UDP/TCP/raw IP. 813 */ 814 if (ipsec_used && 815 (inetsw[ip_protox[ip->ip_p]].pr_flags & PR_LASTHDR) != 0) { 816 SOFTNET_LOCK(); 817 if (ipsec4_input(m, 0) != 0) { 818 SOFTNET_UNLOCK(); 819 goto out; 820 } 821 SOFTNET_UNLOCK(); 822 } 823 #endif 824 825 /* 826 * Switch out to protocol's input routine. 827 */ 828 #if IFA_STATS 829 if (ia && ip) { 830 struct in_ifaddr *_ia; 831 /* 832 * Keep a reference from ip_match_our_address with psref 833 * is expensive, so explore ia here again. 834 */ 835 s = pserialize_read_enter(); 836 _ia = in_get_ia(ip->ip_dst); 837 _ia->ia_ifa.ifa_data.ifad_inbytes += ntohs(ip->ip_len); 838 pserialize_read_exit(s); 839 } 840 #endif 841 IP_STATINC(IP_STAT_DELIVERED); 842 843 const int off = hlen, nh = ip->ip_p; 844 845 SOFTNET_LOCK(); 846 (*inetsw[ip_protox[nh]].pr_input)(m, off, nh); 847 SOFTNET_UNLOCK(); 848 return; 849 850 out: 851 m_put_rcvif_psref(ifp, &psref); 852 if (m != NULL) 853 m_freem(m); 854 } 855 856 /* 857 * IP timer processing. 858 */ 859 void 860 ip_slowtimo(void) 861 { 862 863 #ifndef NET_MPSAFE 864 mutex_enter(softnet_lock); 865 KERNEL_LOCK(1, NULL); 866 #endif 867 868 ip_reass_slowtimo(); 869 870 #ifndef NET_MPSAFE 871 KERNEL_UNLOCK_ONE(NULL); 872 mutex_exit(softnet_lock); 873 #endif 874 } 875 876 /* 877 * IP drain processing. 878 */ 879 void 880 ip_drain(void) 881 { 882 883 KERNEL_LOCK(1, NULL); 884 ip_reass_drain(); 885 KERNEL_UNLOCK_ONE(NULL); 886 } 887 888 /* 889 * ip_dooptions: perform option processing on a datagram, possibly discarding 890 * it if bad options are encountered, or forwarding it if source-routed. 891 * 892 * => Returns true if packet has been forwarded/freed. 893 * => Returns false if the packet should be processed further. 894 */ 895 static bool 896 ip_dooptions(struct mbuf *m) 897 { 898 struct ip *ip = mtod(m, struct ip *); 899 u_char *cp, *cp0; 900 struct ip_timestamp *ipt; 901 struct in_ifaddr *ia; 902 int opt, optlen, cnt, off, code, type = ICMP_PARAMPROB, forward = 0; 903 struct in_addr dst; 904 n_time ntime; 905 struct ifaddr *ifa = NULL; 906 int s; 907 908 dst = ip->ip_dst; 909 cp = (u_char *)(ip + 1); 910 cnt = (ip->ip_hl << 2) - sizeof (struct ip); 911 for (; cnt > 0; cnt -= optlen, cp += optlen) { 912 opt = cp[IPOPT_OPTVAL]; 913 if (opt == IPOPT_EOL) 914 break; 915 if (opt == IPOPT_NOP) 916 optlen = 1; 917 else { 918 if (cnt < IPOPT_OLEN + sizeof(*cp)) { 919 code = &cp[IPOPT_OLEN] - (u_char *)ip; 920 goto bad; 921 } 922 optlen = cp[IPOPT_OLEN]; 923 if (optlen < IPOPT_OLEN + sizeof(*cp) || optlen > cnt) { 924 code = &cp[IPOPT_OLEN] - (u_char *)ip; 925 goto bad; 926 } 927 } 928 switch (opt) { 929 930 default: 931 break; 932 933 /* 934 * Source routing with record. 935 * Find interface with current destination address. 936 * If none on this machine then drop if strictly routed, 937 * or do nothing if loosely routed. 938 * Record interface address and bring up next address 939 * component. If strictly routed make sure next 940 * address is on directly accessible net. 941 */ 942 case IPOPT_LSRR: 943 case IPOPT_SSRR: { 944 struct psref psref; 945 struct sockaddr_in ipaddr = { 946 .sin_len = sizeof(ipaddr), 947 .sin_family = AF_INET, 948 }; 949 950 if (ip_allowsrcrt == 0) { 951 type = ICMP_UNREACH; 952 code = ICMP_UNREACH_NET_PROHIB; 953 goto bad; 954 } 955 if (optlen < IPOPT_OFFSET + sizeof(*cp)) { 956 code = &cp[IPOPT_OLEN] - (u_char *)ip; 957 goto bad; 958 } 959 if ((off = cp[IPOPT_OFFSET]) < IPOPT_MINOFF) { 960 code = &cp[IPOPT_OFFSET] - (u_char *)ip; 961 goto bad; 962 } 963 ipaddr.sin_addr = ip->ip_dst; 964 965 s = pserialize_read_enter(); 966 ifa = ifa_ifwithaddr(sintosa(&ipaddr)); 967 if (ifa == NULL) { 968 pserialize_read_exit(s); 969 if (opt == IPOPT_SSRR) { 970 type = ICMP_UNREACH; 971 code = ICMP_UNREACH_SRCFAIL; 972 goto bad; 973 } 974 /* 975 * Loose routing, and not at next destination 976 * yet; nothing to do except forward. 977 */ 978 break; 979 } 980 pserialize_read_exit(s); 981 982 off--; /* 0 origin */ 983 if ((off + sizeof(struct in_addr)) > optlen) { 984 /* 985 * End of source route. Should be for us. 986 */ 987 save_rte(m, cp, ip->ip_src); 988 break; 989 } 990 /* 991 * locate outgoing interface 992 */ 993 memcpy((void *)&ipaddr.sin_addr, (void *)(cp + off), 994 sizeof(ipaddr.sin_addr)); 995 if (opt == IPOPT_SSRR) { 996 ifa = ifa_ifwithladdr_psref(sintosa(&ipaddr), 997 &psref); 998 if (ifa != NULL) 999 ia = ifatoia(ifa); 1000 else 1001 ia = NULL; 1002 } else { 1003 ia = ip_rtaddr(ipaddr.sin_addr, &psref); 1004 } 1005 if (ia == NULL) { 1006 type = ICMP_UNREACH; 1007 code = ICMP_UNREACH_SRCFAIL; 1008 goto bad; 1009 } 1010 ip->ip_dst = ipaddr.sin_addr; 1011 bcopy((void *)&ia->ia_addr.sin_addr, 1012 (void *)(cp + off), sizeof(struct in_addr)); 1013 ia4_release(ia, &psref); 1014 cp[IPOPT_OFFSET] += sizeof(struct in_addr); 1015 /* 1016 * Let ip_intr's mcast routing check handle mcast pkts 1017 */ 1018 forward = !IN_MULTICAST(ip->ip_dst.s_addr); 1019 break; 1020 } 1021 1022 case IPOPT_RR: { 1023 struct psref psref; 1024 struct sockaddr_in ipaddr = { 1025 .sin_len = sizeof(ipaddr), 1026 .sin_family = AF_INET, 1027 }; 1028 1029 if (optlen < IPOPT_OFFSET + sizeof(*cp)) { 1030 code = &cp[IPOPT_OLEN] - (u_char *)ip; 1031 goto bad; 1032 } 1033 if ((off = cp[IPOPT_OFFSET]) < IPOPT_MINOFF) { 1034 code = &cp[IPOPT_OFFSET] - (u_char *)ip; 1035 goto bad; 1036 } 1037 /* 1038 * If no space remains, ignore. 1039 */ 1040 off--; /* 0 origin */ 1041 if ((off + sizeof(struct in_addr)) > optlen) 1042 break; 1043 memcpy((void *)&ipaddr.sin_addr, (void *)(&ip->ip_dst), 1044 sizeof(ipaddr.sin_addr)); 1045 /* 1046 * locate outgoing interface; if we're the destination, 1047 * use the incoming interface (should be same). 1048 */ 1049 ifa = ifa_ifwithaddr_psref(sintosa(&ipaddr), &psref); 1050 if (ifa == NULL) { 1051 ia = ip_rtaddr(ipaddr.sin_addr, &psref); 1052 if (ia == NULL) { 1053 type = ICMP_UNREACH; 1054 code = ICMP_UNREACH_HOST; 1055 goto bad; 1056 } 1057 } else { 1058 ia = ifatoia(ifa); 1059 } 1060 bcopy((void *)&ia->ia_addr.sin_addr, 1061 (void *)(cp + off), sizeof(struct in_addr)); 1062 ia4_release(ia, &psref); 1063 cp[IPOPT_OFFSET] += sizeof(struct in_addr); 1064 break; 1065 } 1066 1067 case IPOPT_TS: 1068 code = cp - (u_char *)ip; 1069 ipt = (struct ip_timestamp *)cp; 1070 if (ipt->ipt_len < 4 || ipt->ipt_len > 40) { 1071 code = (u_char *)&ipt->ipt_len - (u_char *)ip; 1072 goto bad; 1073 } 1074 if (ipt->ipt_ptr < 5) { 1075 code = (u_char *)&ipt->ipt_ptr - (u_char *)ip; 1076 goto bad; 1077 } 1078 if (ipt->ipt_ptr > ipt->ipt_len - sizeof (int32_t)) { 1079 if (++ipt->ipt_oflw == 0) { 1080 code = (u_char *)&ipt->ipt_ptr - 1081 (u_char *)ip; 1082 goto bad; 1083 } 1084 break; 1085 } 1086 cp0 = (cp + ipt->ipt_ptr - 1); 1087 switch (ipt->ipt_flg) { 1088 1089 case IPOPT_TS_TSONLY: 1090 break; 1091 1092 case IPOPT_TS_TSANDADDR: { 1093 struct ifnet *rcvif; 1094 int _s, _ss; 1095 struct sockaddr_in ipaddr = { 1096 .sin_len = sizeof(ipaddr), 1097 .sin_family = AF_INET, 1098 }; 1099 1100 if (ipt->ipt_ptr - 1 + sizeof(n_time) + 1101 sizeof(struct in_addr) > ipt->ipt_len) { 1102 code = (u_char *)&ipt->ipt_ptr - 1103 (u_char *)ip; 1104 goto bad; 1105 } 1106 ipaddr.sin_addr = dst; 1107 _ss = pserialize_read_enter(); 1108 rcvif = m_get_rcvif(m, &_s); 1109 if (__predict_true(rcvif != NULL)) { 1110 ifa = ifaof_ifpforaddr(sintosa(&ipaddr), 1111 rcvif); 1112 } 1113 m_put_rcvif(rcvif, &_s); 1114 if (ifa == NULL) { 1115 pserialize_read_exit(_ss); 1116 break; 1117 } 1118 ia = ifatoia(ifa); 1119 bcopy(&ia->ia_addr.sin_addr, 1120 cp0, sizeof(struct in_addr)); 1121 pserialize_read_exit(_ss); 1122 ipt->ipt_ptr += sizeof(struct in_addr); 1123 break; 1124 } 1125 1126 case IPOPT_TS_PRESPEC: { 1127 struct sockaddr_in ipaddr = { 1128 .sin_len = sizeof(ipaddr), 1129 .sin_family = AF_INET, 1130 }; 1131 1132 if (ipt->ipt_ptr - 1 + sizeof(n_time) + 1133 sizeof(struct in_addr) > ipt->ipt_len) { 1134 code = (u_char *)&ipt->ipt_ptr - 1135 (u_char *)ip; 1136 goto bad; 1137 } 1138 memcpy(&ipaddr.sin_addr, cp0, 1139 sizeof(struct in_addr)); 1140 s = pserialize_read_enter(); 1141 ifa = ifa_ifwithaddr(sintosa(&ipaddr)); 1142 if (ifa == NULL) { 1143 pserialize_read_exit(s); 1144 continue; 1145 } 1146 pserialize_read_exit(s); 1147 ipt->ipt_ptr += sizeof(struct in_addr); 1148 break; 1149 } 1150 1151 default: 1152 /* XXX can't take &ipt->ipt_flg */ 1153 code = (u_char *)&ipt->ipt_ptr - 1154 (u_char *)ip + 1; 1155 goto bad; 1156 } 1157 ntime = iptime(); 1158 cp0 = (u_char *) &ntime; /* XXX grumble, GCC... */ 1159 memmove((char *)cp + ipt->ipt_ptr - 1, cp0, 1160 sizeof(n_time)); 1161 ipt->ipt_ptr += sizeof(n_time); 1162 } 1163 } 1164 if (forward) { 1165 struct ifnet *rcvif; 1166 struct psref _psref; 1167 1168 if (ip_forwsrcrt == 0) { 1169 type = ICMP_UNREACH; 1170 code = ICMP_UNREACH_SRCFAIL; 1171 goto bad; 1172 } 1173 1174 rcvif = m_get_rcvif_psref(m, &_psref); 1175 if (__predict_false(rcvif == NULL)) { 1176 type = ICMP_UNREACH; 1177 code = ICMP_UNREACH_HOST; 1178 goto bad; 1179 } 1180 ip_forward(m, 1, rcvif); 1181 m_put_rcvif_psref(rcvif, &_psref); 1182 return true; 1183 } 1184 return false; 1185 bad: 1186 icmp_error(m, type, code, 0, 0); 1187 IP_STATINC(IP_STAT_BADOPTIONS); 1188 return true; 1189 } 1190 1191 /* 1192 * ip_rtaddr: given address of next destination (final or next hop), 1193 * return internet address info of interface to be used to get there. 1194 */ 1195 static struct in_ifaddr * 1196 ip_rtaddr(struct in_addr dst, struct psref *psref) 1197 { 1198 struct rtentry *rt; 1199 union { 1200 struct sockaddr dst; 1201 struct sockaddr_in dst4; 1202 } u; 1203 struct route *ro; 1204 1205 sockaddr_in_init(&u.dst4, &dst, 0); 1206 1207 ro = percpu_getref(ipforward_rt_percpu); 1208 rt = rtcache_lookup(ro, &u.dst); 1209 if (rt == NULL) { 1210 percpu_putref(ipforward_rt_percpu); 1211 return NULL; 1212 } 1213 1214 ia4_acquire(ifatoia(rt->rt_ifa), psref); 1215 rtcache_unref(rt, ro); 1216 percpu_putref(ipforward_rt_percpu); 1217 1218 return ifatoia(rt->rt_ifa); 1219 } 1220 1221 /* 1222 * save_rte: save incoming source route for use in replies, to be picked 1223 * up later by ip_srcroute if the receiver is interested. 1224 */ 1225 static void 1226 save_rte(struct mbuf *m, u_char *option, struct in_addr dst) 1227 { 1228 struct ip_srcrt *isr; 1229 struct m_tag *mtag; 1230 unsigned olen; 1231 1232 olen = option[IPOPT_OLEN]; 1233 if (olen > sizeof(isr->isr_hdr) + sizeof(isr->isr_routes)) 1234 return; 1235 1236 mtag = m_tag_get(PACKET_TAG_SRCROUTE, sizeof(*isr), M_NOWAIT); 1237 if (mtag == NULL) 1238 return; 1239 isr = (struct ip_srcrt *)(mtag + 1); 1240 1241 memcpy(isr->isr_hdr, option, olen); 1242 isr->isr_nhops = (olen - IPOPT_OFFSET - 1) / sizeof(struct in_addr); 1243 isr->isr_dst = dst; 1244 m_tag_prepend(m, mtag); 1245 } 1246 1247 /* 1248 * Retrieve incoming source route for use in replies, 1249 * in the same form used by setsockopt. 1250 * The first hop is placed before the options, will be removed later. 1251 */ 1252 struct mbuf * 1253 ip_srcroute(struct mbuf *m0) 1254 { 1255 struct in_addr *p, *q; 1256 struct mbuf *m; 1257 struct ip_srcrt *isr; 1258 struct m_tag *mtag; 1259 1260 mtag = m_tag_find(m0, PACKET_TAG_SRCROUTE, NULL); 1261 if (mtag == NULL) 1262 return NULL; 1263 isr = (struct ip_srcrt *)(mtag + 1); 1264 1265 if (isr->isr_nhops == 0) 1266 return NULL; 1267 1268 m = m_get(M_DONTWAIT, MT_SOOPTS); 1269 if (m == NULL) 1270 return NULL; 1271 1272 MCLAIM(m, &inetdomain.dom_mowner); 1273 #define OPTSIZ (sizeof(isr->isr_nop) + sizeof(isr->isr_hdr)) 1274 1275 /* length is (nhops+1)*sizeof(addr) + sizeof(nop + header) */ 1276 m->m_len = (isr->isr_nhops + 1) * sizeof(struct in_addr) + OPTSIZ; 1277 1278 /* 1279 * First save first hop for return route 1280 */ 1281 p = &(isr->isr_routes[isr->isr_nhops - 1]); 1282 *(mtod(m, struct in_addr *)) = *p--; 1283 1284 /* 1285 * Copy option fields and padding (nop) to mbuf. 1286 */ 1287 isr->isr_nop = IPOPT_NOP; 1288 isr->isr_hdr[IPOPT_OFFSET] = IPOPT_MINOFF; 1289 memmove(mtod(m, char *) + sizeof(struct in_addr), &isr->isr_nop, 1290 OPTSIZ); 1291 q = (struct in_addr *)(mtod(m, char *) + 1292 sizeof(struct in_addr) + OPTSIZ); 1293 #undef OPTSIZ 1294 /* 1295 * Record return path as an IP source route, 1296 * reversing the path (pointers are now aligned). 1297 */ 1298 while (p >= isr->isr_routes) { 1299 *q++ = *p--; 1300 } 1301 /* 1302 * Last hop goes to final destination. 1303 */ 1304 *q = isr->isr_dst; 1305 m_tag_delete(m0, mtag); 1306 return m; 1307 } 1308 1309 const int inetctlerrmap[PRC_NCMDS] = { 1310 [PRC_MSGSIZE] = EMSGSIZE, 1311 [PRC_HOSTDEAD] = EHOSTDOWN, 1312 [PRC_HOSTUNREACH] = EHOSTUNREACH, 1313 [PRC_UNREACH_NET] = EHOSTUNREACH, 1314 [PRC_UNREACH_HOST] = EHOSTUNREACH, 1315 [PRC_UNREACH_PROTOCOL] = ECONNREFUSED, 1316 [PRC_UNREACH_PORT] = ECONNREFUSED, 1317 [PRC_UNREACH_SRCFAIL] = EHOSTUNREACH, 1318 [PRC_PARAMPROB] = ENOPROTOOPT, 1319 }; 1320 1321 void 1322 ip_fasttimo(void) 1323 { 1324 if (ip_drainwanted) { 1325 ip_drain(); 1326 ip_drainwanted = 0; 1327 } 1328 } 1329 1330 void 1331 ip_drainstub(void) 1332 { 1333 ip_drainwanted = 1; 1334 } 1335 1336 /* 1337 * Forward a packet. If some error occurs return the sender 1338 * an icmp packet. Note we can't always generate a meaningful 1339 * icmp message because icmp doesn't have a large enough repertoire 1340 * of codes and types. 1341 * 1342 * If not forwarding, just drop the packet. This could be confusing 1343 * if ipforwarding was zero but some routing protocol was advancing 1344 * us as a gateway to somewhere. However, we must let the routing 1345 * protocol deal with that. 1346 * 1347 * The srcrt parameter indicates whether the packet is being forwarded 1348 * via a source route. 1349 */ 1350 static void 1351 ip_forward(struct mbuf *m, int srcrt, struct ifnet *rcvif) 1352 { 1353 struct ip *ip = mtod(m, struct ip *); 1354 struct rtentry *rt; 1355 int error, type = 0, code = 0, destmtu = 0; 1356 struct mbuf *mcopy; 1357 n_long dest; 1358 union { 1359 struct sockaddr dst; 1360 struct sockaddr_in dst4; 1361 } u; 1362 uint64_t *ips; 1363 struct route *ro; 1364 1365 KASSERTMSG(cpu_softintr_p(), "ip_forward: not in the software " 1366 "interrupt handler; synchronization assumptions violated"); 1367 1368 /* 1369 * We are now in the output path. 1370 */ 1371 MCLAIM(m, &ip_tx_mowner); 1372 1373 /* 1374 * Clear any in-bound checksum flags for this packet. 1375 */ 1376 m->m_pkthdr.csum_flags = 0; 1377 1378 dest = 0; 1379 if (m->m_flags & (M_BCAST|M_MCAST) || in_canforward(ip->ip_dst) == 0) { 1380 IP_STATINC(IP_STAT_CANTFORWARD); 1381 m_freem(m); 1382 return; 1383 } 1384 1385 if (ip->ip_ttl <= IPTTLDEC) { 1386 icmp_error(m, ICMP_TIMXCEED, ICMP_TIMXCEED_INTRANS, dest, 0); 1387 return; 1388 } 1389 1390 sockaddr_in_init(&u.dst4, &ip->ip_dst, 0); 1391 1392 ro = percpu_getref(ipforward_rt_percpu); 1393 rt = rtcache_lookup(ro, &u.dst); 1394 if (rt == NULL) { 1395 percpu_putref(ipforward_rt_percpu); 1396 icmp_error(m, ICMP_UNREACH, ICMP_UNREACH_NET, dest, 0); 1397 return; 1398 } 1399 1400 /* 1401 * Save at most 68 bytes of the packet in case 1402 * we need to generate an ICMP message to the src. 1403 * Pullup to avoid sharing mbuf cluster between m and mcopy. 1404 */ 1405 mcopy = m_copym(m, 0, imin(ntohs(ip->ip_len), 68), M_DONTWAIT); 1406 if (mcopy) 1407 mcopy = m_pullup(mcopy, ip->ip_hl << 2); 1408 1409 ip->ip_ttl -= IPTTLDEC; 1410 1411 /* 1412 * If forwarding packet using same interface that it came in on, 1413 * perhaps should send a redirect to sender to shortcut a hop. 1414 * Only send redirect if source is sending directly to us, 1415 * and if packet was not source routed (or has any options). 1416 * Also, don't send redirect if forwarding using a default route 1417 * or a route modified by a redirect. 1418 */ 1419 if (rt->rt_ifp == rcvif && 1420 (rt->rt_flags & (RTF_DYNAMIC|RTF_MODIFIED)) == 0 && 1421 !in_nullhost(satocsin(rt_getkey(rt))->sin_addr) && 1422 ipsendredirects && !srcrt) { 1423 if (rt->rt_ifa && 1424 (ip->ip_src.s_addr & ifatoia(rt->rt_ifa)->ia_subnetmask) == 1425 ifatoia(rt->rt_ifa)->ia_subnet) { 1426 if (rt->rt_flags & RTF_GATEWAY) 1427 dest = satosin(rt->rt_gateway)->sin_addr.s_addr; 1428 else 1429 dest = ip->ip_dst.s_addr; 1430 /* 1431 * Router requirements says to only send host 1432 * redirects. 1433 */ 1434 type = ICMP_REDIRECT; 1435 code = ICMP_REDIRECT_HOST; 1436 } 1437 } 1438 rtcache_unref(rt, ro); 1439 1440 error = ip_output(m, NULL, ro, 1441 (IP_FORWARDING | (ip_directedbcast ? IP_ALLOWBROADCAST : 0)), 1442 NULL, NULL); 1443 1444 if (error) { 1445 IP_STATINC(IP_STAT_CANTFORWARD); 1446 goto error; 1447 } 1448 1449 ips = IP_STAT_GETREF(); 1450 ips[IP_STAT_FORWARD]++; 1451 1452 if (type) { 1453 ips[IP_STAT_REDIRECTSENT]++; 1454 IP_STAT_PUTREF(); 1455 goto redirect; 1456 } 1457 1458 IP_STAT_PUTREF(); 1459 if (mcopy) { 1460 #ifdef GATEWAY 1461 if (mcopy->m_flags & M_CANFASTFWD) 1462 ipflow_create(ro, mcopy); 1463 #endif 1464 m_freem(mcopy); 1465 } 1466 1467 percpu_putref(ipforward_rt_percpu); 1468 return; 1469 1470 redirect: 1471 error: 1472 if (mcopy == NULL) { 1473 percpu_putref(ipforward_rt_percpu); 1474 return; 1475 } 1476 1477 switch (error) { 1478 1479 case 0: /* forwarded, but need redirect */ 1480 /* type, code set above */ 1481 break; 1482 1483 case ENETUNREACH: /* shouldn't happen, checked above */ 1484 case EHOSTUNREACH: 1485 case ENETDOWN: 1486 case EHOSTDOWN: 1487 default: 1488 type = ICMP_UNREACH; 1489 code = ICMP_UNREACH_HOST; 1490 break; 1491 1492 case EMSGSIZE: 1493 type = ICMP_UNREACH; 1494 code = ICMP_UNREACH_NEEDFRAG; 1495 1496 if ((rt = rtcache_validate(ro)) != NULL) { 1497 destmtu = rt->rt_ifp->if_mtu; 1498 rtcache_unref(rt, ro); 1499 } 1500 #ifdef IPSEC 1501 if (ipsec_used) 1502 (void)ipsec4_forward(mcopy, &destmtu); 1503 #endif 1504 IP_STATINC(IP_STAT_CANTFRAG); 1505 break; 1506 1507 case ENOBUFS: 1508 /* 1509 * Do not generate ICMP_SOURCEQUENCH as required in RFC 1812, 1510 * Requirements for IP Version 4 Routers. Source quench can 1511 * big problem under DoS attacks or if the underlying 1512 * interface is rate-limited. 1513 */ 1514 if (mcopy) 1515 m_freem(mcopy); 1516 percpu_putref(ipforward_rt_percpu); 1517 return; 1518 } 1519 icmp_error(mcopy, type, code, dest, destmtu); 1520 percpu_putref(ipforward_rt_percpu); 1521 } 1522 1523 void 1524 ip_savecontrol(struct inpcb *inp, struct mbuf **mp, struct ip *ip, 1525 struct mbuf *m) 1526 { 1527 struct socket *so = inp->inp_socket; 1528 ifnet_t *ifp; 1529 int inpflags = inp->inp_flags; 1530 struct psref psref; 1531 1532 ifp = m_get_rcvif_psref(m, &psref); 1533 if (__predict_false(ifp == NULL)) 1534 return; /* XXX should report error? */ 1535 1536 if (so->so_options & SO_TIMESTAMP 1537 #ifdef SO_OTIMESTAMP 1538 || so->so_options & SO_OTIMESTAMP 1539 #endif 1540 ) { 1541 struct timeval tv; 1542 1543 microtime(&tv); 1544 #ifdef SO_OTIMESTAMP 1545 if (so->so_options & SO_OTIMESTAMP) { 1546 struct timeval50 tv50; 1547 timeval_to_timeval50(&tv, &tv50); 1548 *mp = sbcreatecontrol((void *) &tv50, sizeof(tv50), 1549 SCM_OTIMESTAMP, SOL_SOCKET); 1550 } else 1551 #endif 1552 *mp = sbcreatecontrol((void *) &tv, sizeof(tv), 1553 SCM_TIMESTAMP, SOL_SOCKET); 1554 if (*mp) 1555 mp = &(*mp)->m_next; 1556 } 1557 if (inpflags & INP_RECVDSTADDR) { 1558 *mp = sbcreatecontrol((void *) &ip->ip_dst, 1559 sizeof(struct in_addr), IP_RECVDSTADDR, IPPROTO_IP); 1560 if (*mp) 1561 mp = &(*mp)->m_next; 1562 } 1563 if (inpflags & INP_RECVPKTINFO) { 1564 struct in_pktinfo ipi; 1565 ipi.ipi_addr = ip->ip_src; 1566 ipi.ipi_ifindex = ifp->if_index; 1567 *mp = sbcreatecontrol((void *) &ipi, 1568 sizeof(ipi), IP_RECVPKTINFO, IPPROTO_IP); 1569 if (*mp) 1570 mp = &(*mp)->m_next; 1571 } 1572 if (inpflags & INP_PKTINFO) { 1573 struct in_pktinfo ipi; 1574 ipi.ipi_addr = ip->ip_dst; 1575 ipi.ipi_ifindex = ifp->if_index; 1576 *mp = sbcreatecontrol((void *) &ipi, 1577 sizeof(ipi), IP_PKTINFO, IPPROTO_IP); 1578 if (*mp) 1579 mp = &(*mp)->m_next; 1580 } 1581 if (inpflags & INP_RECVIF) { 1582 struct sockaddr_dl sdl; 1583 1584 sockaddr_dl_init(&sdl, sizeof(sdl), ifp->if_index, 0, NULL, 0, 1585 NULL, 0); 1586 *mp = sbcreatecontrol(&sdl, sdl.sdl_len, IP_RECVIF, IPPROTO_IP); 1587 if (*mp) 1588 mp = &(*mp)->m_next; 1589 } 1590 if (inpflags & INP_RECVTTL) { 1591 *mp = sbcreatecontrol((void *) &ip->ip_ttl, 1592 sizeof(uint8_t), IP_RECVTTL, IPPROTO_IP); 1593 if (*mp) 1594 mp = &(*mp)->m_next; 1595 } 1596 m_put_rcvif_psref(ifp, &psref); 1597 } 1598 1599 /* 1600 * sysctl helper routine for net.inet.ip.forwsrcrt. 1601 */ 1602 static int 1603 sysctl_net_inet_ip_forwsrcrt(SYSCTLFN_ARGS) 1604 { 1605 int error, tmp; 1606 struct sysctlnode node; 1607 1608 node = *rnode; 1609 tmp = ip_forwsrcrt; 1610 node.sysctl_data = &tmp; 1611 error = sysctl_lookup(SYSCTLFN_CALL(&node)); 1612 if (error || newp == NULL) 1613 return (error); 1614 1615 error = kauth_authorize_network(l->l_cred, KAUTH_NETWORK_FORWSRCRT, 1616 0, NULL, NULL, NULL); 1617 if (error) 1618 return (error); 1619 1620 ip_forwsrcrt = tmp; 1621 1622 return (0); 1623 } 1624 1625 /* 1626 * sysctl helper routine for net.inet.ip.mtudisctimeout. checks the 1627 * range of the new value and tweaks timers if it changes. 1628 */ 1629 static int 1630 sysctl_net_inet_ip_pmtudto(SYSCTLFN_ARGS) 1631 { 1632 int error, tmp; 1633 struct sysctlnode node; 1634 1635 icmp_mtudisc_lock(); 1636 1637 node = *rnode; 1638 tmp = ip_mtudisc_timeout; 1639 node.sysctl_data = &tmp; 1640 error = sysctl_lookup(SYSCTLFN_CALL(&node)); 1641 if (error || newp == NULL) 1642 goto out; 1643 if (tmp < 0) { 1644 error = EINVAL; 1645 goto out; 1646 } 1647 1648 ip_mtudisc_timeout = tmp; 1649 rt_timer_queue_change(ip_mtudisc_timeout_q, ip_mtudisc_timeout); 1650 error = 0; 1651 out: 1652 icmp_mtudisc_unlock(); 1653 return error; 1654 } 1655 1656 static int 1657 sysctl_net_inet_ip_stats(SYSCTLFN_ARGS) 1658 { 1659 1660 return (NETSTAT_SYSCTL(ipstat_percpu, IP_NSTATS)); 1661 } 1662 1663 static void 1664 sysctl_net_inet_ip_setup(struct sysctllog **clog) 1665 { 1666 sysctl_createv(clog, 0, NULL, NULL, 1667 CTLFLAG_PERMANENT, 1668 CTLTYPE_NODE, "inet", 1669 SYSCTL_DESCR("PF_INET related settings"), 1670 NULL, 0, NULL, 0, 1671 CTL_NET, PF_INET, CTL_EOL); 1672 sysctl_createv(clog, 0, NULL, NULL, 1673 CTLFLAG_PERMANENT, 1674 CTLTYPE_NODE, "ip", 1675 SYSCTL_DESCR("IPv4 related settings"), 1676 NULL, 0, NULL, 0, 1677 CTL_NET, PF_INET, IPPROTO_IP, CTL_EOL); 1678 1679 sysctl_createv(clog, 0, NULL, NULL, 1680 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1681 CTLTYPE_INT, "forwarding", 1682 SYSCTL_DESCR("Enable forwarding of INET datagrams"), 1683 NULL, 0, &ipforwarding, 0, 1684 CTL_NET, PF_INET, IPPROTO_IP, 1685 IPCTL_FORWARDING, CTL_EOL); 1686 sysctl_createv(clog, 0, NULL, NULL, 1687 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1688 CTLTYPE_INT, "redirect", 1689 SYSCTL_DESCR("Enable sending of ICMP redirect messages"), 1690 NULL, 0, &ipsendredirects, 0, 1691 CTL_NET, PF_INET, IPPROTO_IP, 1692 IPCTL_SENDREDIRECTS, CTL_EOL); 1693 sysctl_createv(clog, 0, NULL, NULL, 1694 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1695 CTLTYPE_INT, "ttl", 1696 SYSCTL_DESCR("Default TTL for an INET datagram"), 1697 NULL, 0, &ip_defttl, 0, 1698 CTL_NET, PF_INET, IPPROTO_IP, 1699 IPCTL_DEFTTL, CTL_EOL); 1700 #ifdef IPCTL_DEFMTU 1701 sysctl_createv(clog, 0, NULL, NULL, 1702 CTLFLAG_PERMANENT /* |CTLFLAG_READWRITE? */, 1703 CTLTYPE_INT, "mtu", 1704 SYSCTL_DESCR("Default MTA for an INET route"), 1705 NULL, 0, &ip_mtu, 0, 1706 CTL_NET, PF_INET, IPPROTO_IP, 1707 IPCTL_DEFMTU, CTL_EOL); 1708 #endif /* IPCTL_DEFMTU */ 1709 sysctl_createv(clog, 0, NULL, NULL, 1710 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1711 CTLTYPE_INT, "forwsrcrt", 1712 SYSCTL_DESCR("Enable forwarding of source-routed " 1713 "datagrams"), 1714 sysctl_net_inet_ip_forwsrcrt, 0, &ip_forwsrcrt, 0, 1715 CTL_NET, PF_INET, IPPROTO_IP, 1716 IPCTL_FORWSRCRT, CTL_EOL); 1717 sysctl_createv(clog, 0, NULL, NULL, 1718 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1719 CTLTYPE_INT, "directed-broadcast", 1720 SYSCTL_DESCR("Enable forwarding of broadcast datagrams"), 1721 NULL, 0, &ip_directedbcast, 0, 1722 CTL_NET, PF_INET, IPPROTO_IP, 1723 IPCTL_DIRECTEDBCAST, CTL_EOL); 1724 sysctl_createv(clog, 0, NULL, NULL, 1725 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1726 CTLTYPE_INT, "allowsrcrt", 1727 SYSCTL_DESCR("Accept source-routed datagrams"), 1728 NULL, 0, &ip_allowsrcrt, 0, 1729 CTL_NET, PF_INET, IPPROTO_IP, 1730 IPCTL_ALLOWSRCRT, CTL_EOL); 1731 1732 sysctl_createv(clog, 0, NULL, NULL, 1733 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1734 CTLTYPE_INT, "mtudisc", 1735 SYSCTL_DESCR("Use RFC1191 Path MTU Discovery"), 1736 NULL, 0, &ip_mtudisc, 0, 1737 CTL_NET, PF_INET, IPPROTO_IP, 1738 IPCTL_MTUDISC, CTL_EOL); 1739 sysctl_createv(clog, 0, NULL, NULL, 1740 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1741 CTLTYPE_INT, "anonportmin", 1742 SYSCTL_DESCR("Lowest ephemeral port number to assign"), 1743 sysctl_net_inet_ip_ports, 0, &anonportmin, 0, 1744 CTL_NET, PF_INET, IPPROTO_IP, 1745 IPCTL_ANONPORTMIN, CTL_EOL); 1746 sysctl_createv(clog, 0, NULL, NULL, 1747 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1748 CTLTYPE_INT, "anonportmax", 1749 SYSCTL_DESCR("Highest ephemeral port number to assign"), 1750 sysctl_net_inet_ip_ports, 0, &anonportmax, 0, 1751 CTL_NET, PF_INET, IPPROTO_IP, 1752 IPCTL_ANONPORTMAX, CTL_EOL); 1753 sysctl_createv(clog, 0, NULL, NULL, 1754 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1755 CTLTYPE_INT, "mtudisctimeout", 1756 SYSCTL_DESCR("Lifetime of a Path MTU Discovered route"), 1757 sysctl_net_inet_ip_pmtudto, 0, (void *)&ip_mtudisc_timeout, 0, 1758 CTL_NET, PF_INET, IPPROTO_IP, 1759 IPCTL_MTUDISCTIMEOUT, CTL_EOL); 1760 #ifndef IPNOPRIVPORTS 1761 sysctl_createv(clog, 0, NULL, NULL, 1762 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1763 CTLTYPE_INT, "lowportmin", 1764 SYSCTL_DESCR("Lowest privileged ephemeral port number " 1765 "to assign"), 1766 sysctl_net_inet_ip_ports, 0, &lowportmin, 0, 1767 CTL_NET, PF_INET, IPPROTO_IP, 1768 IPCTL_LOWPORTMIN, CTL_EOL); 1769 sysctl_createv(clog, 0, NULL, NULL, 1770 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1771 CTLTYPE_INT, "lowportmax", 1772 SYSCTL_DESCR("Highest privileged ephemeral port number " 1773 "to assign"), 1774 sysctl_net_inet_ip_ports, 0, &lowportmax, 0, 1775 CTL_NET, PF_INET, IPPROTO_IP, 1776 IPCTL_LOWPORTMAX, CTL_EOL); 1777 #endif /* IPNOPRIVPORTS */ 1778 #if NGRE > 0 1779 sysctl_createv(clog, 0, NULL, NULL, 1780 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1781 CTLTYPE_INT, "grettl", 1782 SYSCTL_DESCR("Default TTL for a gre tunnel datagram"), 1783 NULL, 0, &ip_gre_ttl, 0, 1784 CTL_NET, PF_INET, IPPROTO_IP, 1785 IPCTL_GRE_TTL, CTL_EOL); 1786 #endif /* NGRE */ 1787 sysctl_createv(clog, 0, NULL, NULL, 1788 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1789 CTLTYPE_INT, "checkinterface", 1790 SYSCTL_DESCR("Enable receive side of Strong ES model " 1791 "from RFC1122"), 1792 NULL, 0, &ip_checkinterface, 0, 1793 CTL_NET, PF_INET, IPPROTO_IP, 1794 IPCTL_CHECKINTERFACE, CTL_EOL); 1795 sysctl_createv(clog, 0, NULL, NULL, 1796 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1797 CTLTYPE_INT, "random_id", 1798 SYSCTL_DESCR("Assign random ip_id values"), 1799 NULL, 0, &ip_do_randomid, 0, 1800 CTL_NET, PF_INET, IPPROTO_IP, 1801 IPCTL_RANDOMID, CTL_EOL); 1802 sysctl_createv(clog, 0, NULL, NULL, 1803 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1804 CTLTYPE_INT, "do_loopback_cksum", 1805 SYSCTL_DESCR("Perform IP checksum on loopback"), 1806 NULL, 0, &ip_do_loopback_cksum, 0, 1807 CTL_NET, PF_INET, IPPROTO_IP, 1808 IPCTL_LOOPBACKCKSUM, CTL_EOL); 1809 sysctl_createv(clog, 0, NULL, NULL, 1810 CTLFLAG_PERMANENT, 1811 CTLTYPE_STRUCT, "stats", 1812 SYSCTL_DESCR("IP statistics"), 1813 sysctl_net_inet_ip_stats, 0, NULL, 0, 1814 CTL_NET, PF_INET, IPPROTO_IP, IPCTL_STATS, 1815 CTL_EOL); 1816 #if NARP 1817 sysctl_createv(clog, 0, NULL, NULL, 1818 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1819 CTLTYPE_INT, "dad_count", 1820 SYSCTL_DESCR("Number of Duplicate Address Detection " 1821 "probes to send"), 1822 NULL, 0, &ip_dad_count, 0, 1823 CTL_NET, PF_INET, IPPROTO_IP, 1824 IPCTL_DAD_COUNT, CTL_EOL); 1825 #endif 1826 1827 /* anonportalgo RFC6056 subtree */ 1828 const struct sysctlnode *portalgo_node; 1829 sysctl_createv(clog, 0, NULL, &portalgo_node, 1830 CTLFLAG_PERMANENT, 1831 CTLTYPE_NODE, "anonportalgo", 1832 SYSCTL_DESCR("Anonymous Port Algorithm Selection (RFC 6056)"), 1833 NULL, 0, NULL, 0, 1834 CTL_NET, PF_INET, IPPROTO_IP, CTL_CREATE, CTL_EOL); 1835 sysctl_createv(clog, 0, &portalgo_node, NULL, 1836 CTLFLAG_PERMANENT, 1837 CTLTYPE_STRING, "available", 1838 SYSCTL_DESCR("available algorithms"), 1839 sysctl_portalgo_available, 0, NULL, PORTALGO_MAXLEN, 1840 CTL_CREATE, CTL_EOL); 1841 sysctl_createv(clog, 0, &portalgo_node, NULL, 1842 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1843 CTLTYPE_STRING, "selected", 1844 SYSCTL_DESCR("selected algorithm"), 1845 sysctl_portalgo_selected4, 0, NULL, PORTALGO_MAXLEN, 1846 CTL_CREATE, CTL_EOL); 1847 sysctl_createv(clog, 0, &portalgo_node, NULL, 1848 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1849 CTLTYPE_STRUCT, "reserve", 1850 SYSCTL_DESCR("bitmap of reserved ports"), 1851 sysctl_portalgo_reserve4, 0, NULL, 0, 1852 CTL_CREATE, CTL_EOL); 1853 } 1854 1855 void 1856 ip_statinc(u_int stat) 1857 { 1858 1859 KASSERT(stat < IP_NSTATS); 1860 IP_STATINC(stat); 1861 } 1862
Indexes created Sat Oct 18 23:09:54 GMT 2025