1 /*- 2 * SPDX-License-Identifier: BSD-3-Clause 3 * 4 * Copyright (c) 1982, 1986, 1988, 1990, 1993 5 * The Regents of the University of California. 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 University 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 REGENTS 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 REGENTS 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 * @(#)ip_output.c 8.3 (Berkeley) 1/21/94 32 */ 33 34 #include <sys/cdefs.h> 35 __FBSDID("$FreeBSD$"); 36 37 #include "opt_inet.h" 38 #include "opt_ipsec.h" 39 #include "opt_kern_tls.h" 40 #include "opt_mbuf_stress_test.h" 41 #include "opt_ratelimit.h" 42 #include "opt_route.h" 43 #include "opt_rss.h" 44 #include "opt_sctp.h" 45 46 #include <sys/param.h> 47 #include <sys/systm.h> 48 #include <sys/kernel.h> 49 #include <sys/ktls.h> 50 #include <sys/lock.h> 51 #include <sys/malloc.h> 52 #include <sys/mbuf.h> 53 #include <sys/priv.h> 54 #include <sys/proc.h> 55 #include <sys/protosw.h> 56 #include <sys/sdt.h> 57 #include <sys/socket.h> 58 #include <sys/socketvar.h> 59 #include <sys/sysctl.h> 60 #include <sys/ucred.h> 61 62 #include <net/if.h> 63 #include <net/if_var.h> 64 #include <net/if_vlan_var.h> 65 #include <net/if_llatbl.h> 66 #include <net/ethernet.h> 67 #include <net/netisr.h> 68 #include <net/pfil.h> 69 #include <net/route.h> 70 #include <net/route/nhop.h> 71 #include <net/rss_config.h> 72 #include <net/vnet.h> 73 74 #include <netinet/in.h> 75 #include <netinet/in_fib.h> 76 #include <netinet/in_kdtrace.h> 77 #include <netinet/in_systm.h> 78 #include <netinet/ip.h> 79 #include <netinet/in_fib.h> 80 #include <netinet/in_pcb.h> 81 #include <netinet/in_rss.h> 82 #include <netinet/in_var.h> 83 #include <netinet/ip_var.h> 84 #include <netinet/ip_options.h> 85 #include <netinet/ip_mroute.h> 86 87 #include <netinet/udp.h> 88 #include <netinet/udp_var.h> 89 90 #if defined(SCTP) || defined(SCTP_SUPPORT) 91 #include <netinet/sctp.h> 92 #include <netinet/sctp_crc32.h> 93 #endif 94 95 #include <netipsec/ipsec_support.h> 96 97 #include <machine/in_cksum.h> 98 99 #include <security/mac/mac_framework.h> 100 101 #ifdef MBUF_STRESS_TEST 102 static int mbuf_frag_size = 0; 103 SYSCTL_INT(_net_inet_ip, OID_AUTO, mbuf_frag_size, CTLFLAG_RW, 104 &mbuf_frag_size, 0, "Fragment outgoing mbufs to this size"); 105 #endif 106 107 static void ip_mloopback(struct ifnet *, const struct mbuf *, int); 108 109 extern int in_mcast_loop; 110 extern struct protosw inetsw[]; 111 112 static inline int 113 ip_output_pfil(struct mbuf **mp, struct ifnet *ifp, int flags, 114 struct inpcb *inp, struct sockaddr_in *dst, int *fibnum, int *error) 115 { 116 struct m_tag *fwd_tag = NULL; 117 struct mbuf *m; 118 struct in_addr odst; 119 struct ip *ip; 120 int pflags = PFIL_OUT; 121 122 if (flags & IP_FORWARDING) 123 pflags |= PFIL_FWD; 124 125 m = *mp; 126 ip = mtod(m, struct ip *); 127 128 /* Run through list of hooks for output packets. */ 129 odst.s_addr = ip->ip_dst.s_addr; 130 switch (pfil_run_hooks(V_inet_pfil_head, mp, ifp, pflags, inp)) { 131 case PFIL_DROPPED: 132 *error = EACCES; 133 /* FALLTHROUGH */ 134 case PFIL_CONSUMED: 135 return 1; /* Finished */ 136 case PFIL_PASS: 137 *error = 0; 138 } 139 m = *mp; 140 ip = mtod(m, struct ip *); 141 142 /* See if destination IP address was changed by packet filter. */ 143 if (odst.s_addr != ip->ip_dst.s_addr) { 144 m->m_flags |= M_SKIP_FIREWALL; 145 /* If destination is now ourself drop to ip_input(). */ 146 if (in_localip(ip->ip_dst)) { 147 m->m_flags |= M_FASTFWD_OURS; 148 if (m->m_pkthdr.rcvif == NULL) 149 m->m_pkthdr.rcvif = V_loif; 150 if (m->m_pkthdr.csum_flags & CSUM_DELAY_DATA) { 151 m->m_pkthdr.csum_flags |= 152 CSUM_DATA_VALID | CSUM_PSEUDO_HDR; 153 m->m_pkthdr.csum_data = 0xffff; 154 } 155 m->m_pkthdr.csum_flags |= 156 CSUM_IP_CHECKED | CSUM_IP_VALID; 157 #if defined(SCTP) || defined(SCTP_SUPPORT) 158 if (m->m_pkthdr.csum_flags & CSUM_SCTP) 159 m->m_pkthdr.csum_flags |= CSUM_SCTP_VALID; 160 #endif 161 *error = netisr_queue(NETISR_IP, m); 162 return 1; /* Finished */ 163 } 164 165 bzero(dst, sizeof(*dst)); 166 dst->sin_family = AF_INET; 167 dst->sin_len = sizeof(*dst); 168 dst->sin_addr = ip->ip_dst; 169 170 return -1; /* Reloop */ 171 } 172 /* See if fib was changed by packet filter. */ 173 if ((*fibnum) != M_GETFIB(m)) { 174 m->m_flags |= M_SKIP_FIREWALL; 175 *fibnum = M_GETFIB(m); 176 return -1; /* Reloop for FIB change */ 177 } 178 179 /* See if local, if yes, send it to netisr with IP_FASTFWD_OURS. */ 180 if (m->m_flags & M_FASTFWD_OURS) { 181 if (m->m_pkthdr.rcvif == NULL) 182 m->m_pkthdr.rcvif = V_loif; 183 if (m->m_pkthdr.csum_flags & CSUM_DELAY_DATA) { 184 m->m_pkthdr.csum_flags |= 185 CSUM_DATA_VALID | CSUM_PSEUDO_HDR; 186 m->m_pkthdr.csum_data = 0xffff; 187 } 188 #if defined(SCTP) || defined(SCTP_SUPPORT) 189 if (m->m_pkthdr.csum_flags & CSUM_SCTP) 190 m->m_pkthdr.csum_flags |= CSUM_SCTP_VALID; 191 #endif 192 m->m_pkthdr.csum_flags |= 193 CSUM_IP_CHECKED | CSUM_IP_VALID; 194 195 *error = netisr_queue(NETISR_IP, m); 196 return 1; /* Finished */ 197 } 198 /* Or forward to some other address? */ 199 if ((m->m_flags & M_IP_NEXTHOP) && 200 ((fwd_tag = m_tag_find(m, PACKET_TAG_IPFORWARD, NULL)) != NULL)) { 201 bcopy((fwd_tag+1), dst, sizeof(struct sockaddr_in)); 202 m->m_flags |= M_SKIP_FIREWALL; 203 m->m_flags &= ~M_IP_NEXTHOP; 204 m_tag_delete(m, fwd_tag); 205 206 return -1; /* Reloop for CHANGE of dst */ 207 } 208 209 return 0; 210 } 211 212 static int 213 ip_output_send(struct inpcb *inp, struct ifnet *ifp, struct mbuf *m, 214 const struct sockaddr *gw, struct route *ro, bool stamp_tag) 215 { 216 #ifdef KERN_TLS 217 struct ktls_session *tls = NULL; 218 #endif 219 struct m_snd_tag *mst; 220 int error; 221 222 MPASS((m->m_pkthdr.csum_flags & CSUM_SND_TAG) == 0); 223 mst = NULL; 224 225 #ifdef KERN_TLS 226 /* 227 * If this is an unencrypted TLS record, save a reference to 228 * the record. This local reference is used to call 229 * ktls_output_eagain after the mbuf has been freed (thus 230 * dropping the mbuf's reference) in if_output. 231 */ 232 if (m->m_next != NULL && mbuf_has_tls_session(m->m_next)) { 233 tls = ktls_hold(m->m_next->m_epg_tls); 234 mst = tls->snd_tag; 235 236 /* 237 * If a TLS session doesn't have a valid tag, it must 238 * have had an earlier ifp mismatch, so drop this 239 * packet. 240 */ 241 if (mst == NULL) { 242 error = EAGAIN; 243 goto done; 244 } 245 /* 246 * Always stamp tags that include NIC ktls. 247 */ 248 stamp_tag = true; 249 } 250 #endif 251 #ifdef RATELIMIT 252 if (inp != NULL && mst == NULL) { 253 if ((inp->inp_flags2 & INP_RATE_LIMIT_CHANGED) != 0 || 254 (inp->inp_snd_tag != NULL && 255 inp->inp_snd_tag->ifp != ifp)) 256 in_pcboutput_txrtlmt(inp, ifp, m); 257 258 if (inp->inp_snd_tag != NULL) 259 mst = inp->inp_snd_tag; 260 } 261 #endif 262 if (stamp_tag && mst != NULL) { 263 KASSERT(m->m_pkthdr.rcvif == NULL, 264 ("trying to add a send tag to a forwarded packet")); 265 if (mst->ifp != ifp) { 266 error = EAGAIN; 267 goto done; 268 } 269 270 /* stamp send tag on mbuf */ 271 m->m_pkthdr.snd_tag = m_snd_tag_ref(mst); 272 m->m_pkthdr.csum_flags |= CSUM_SND_TAG; 273 } 274 275 error = (*ifp->if_output)(ifp, m, gw, ro); 276 277 done: 278 /* Check for route change invalidating send tags. */ 279 #ifdef KERN_TLS 280 if (tls != NULL) { 281 if (error == EAGAIN) 282 error = ktls_output_eagain(inp, tls); 283 ktls_free(tls); 284 } 285 #endif 286 #ifdef RATELIMIT 287 if (error == EAGAIN) 288 in_pcboutput_eagain(inp); 289 #endif 290 return (error); 291 } 292 293 /* rte<>ro_flags translation */ 294 static inline void 295 rt_update_ro_flags(struct route *ro, const struct nhop_object *nh) 296 { 297 int nh_flags = nh->nh_flags; 298 299 ro->ro_flags &= ~ (RT_REJECT|RT_BLACKHOLE|RT_HAS_GW); 300 301 ro->ro_flags |= (nh_flags & NHF_REJECT) ? RT_REJECT : 0; 302 ro->ro_flags |= (nh_flags & NHF_BLACKHOLE) ? RT_BLACKHOLE : 0; 303 ro->ro_flags |= (nh_flags & NHF_GATEWAY) ? RT_HAS_GW : 0; 304 } 305 306 /* 307 * IP output. The packet in mbuf chain m contains a skeletal IP 308 * header (with len, off, ttl, proto, tos, src, dst). 309 * The mbuf chain containing the packet will be freed. 310 * The mbuf opt, if present, will not be freed. 311 * If route ro is present and has ro_rt initialized, route lookup would be 312 * skipped and ro->ro_rt would be used. If ro is present but ro->ro_rt is NULL, 313 * then result of route lookup is stored in ro->ro_rt. 314 * 315 * In the IP forwarding case, the packet will arrive with options already 316 * inserted, so must have a NULL opt pointer. 317 */ 318 int 319 ip_output(struct mbuf *m, struct mbuf *opt, struct route *ro, int flags, 320 struct ip_moptions *imo, struct inpcb *inp) 321 { 322 MROUTER_RLOCK_TRACKER; 323 struct ip *ip; 324 struct ifnet *ifp = NULL; /* keep compiler happy */ 325 struct mbuf *m0; 326 int hlen = sizeof (struct ip); 327 int mtu = 0; 328 int error = 0; 329 int vlan_pcp = -1; 330 struct sockaddr_in *dst; 331 const struct sockaddr *gw; 332 struct in_ifaddr *ia = NULL; 333 struct in_addr src; 334 int isbroadcast; 335 uint16_t ip_len, ip_off; 336 struct route iproute; 337 uint32_t fibnum; 338 #if defined(IPSEC) || defined(IPSEC_SUPPORT) 339 int no_route_but_check_spd = 0; 340 #endif 341 342 M_ASSERTPKTHDR(m); 343 NET_EPOCH_ASSERT(); 344 345 if (inp != NULL) { 346 INP_LOCK_ASSERT(inp); 347 M_SETFIB(m, inp->inp_inc.inc_fibnum); 348 if ((flags & IP_NODEFAULTFLOWID) == 0) { 349 m->m_pkthdr.flowid = inp->inp_flowid; 350 M_HASHTYPE_SET(m, inp->inp_flowtype); 351 } 352 if ((inp->inp_flags2 & INP_2PCP_SET) != 0) 353 vlan_pcp = (inp->inp_flags2 & INP_2PCP_MASK) >> 354 INP_2PCP_SHIFT; 355 #ifdef NUMA 356 m->m_pkthdr.numa_domain = inp->inp_numa_domain; 357 #endif 358 } 359 360 if (opt) { 361 int len = 0; 362 m = ip_insertoptions(m, opt, &len); 363 if (len != 0) 364 hlen = len; /* ip->ip_hl is updated above */ 365 } 366 ip = mtod(m, struct ip *); 367 ip_len = ntohs(ip->ip_len); 368 ip_off = ntohs(ip->ip_off); 369 370 if ((flags & (IP_FORWARDING|IP_RAWOUTPUT)) == 0) { 371 ip->ip_v = IPVERSION; 372 ip->ip_hl = hlen >> 2; 373 ip_fillid(ip); 374 } else { 375 /* Header already set, fetch hlen from there */ 376 hlen = ip->ip_hl << 2; 377 } 378 if ((flags & IP_FORWARDING) == 0) 379 IPSTAT_INC(ips_localout); 380 381 /* 382 * dst/gw handling: 383 * 384 * gw is readonly but can point either to dst OR rt_gateway, 385 * therefore we need restore gw if we're redoing lookup. 386 */ 387 fibnum = (inp != NULL) ? inp->inp_inc.inc_fibnum : M_GETFIB(m); 388 if (ro == NULL) { 389 ro = &iproute; 390 bzero(ro, sizeof (*ro)); 391 } 392 dst = (struct sockaddr_in *)&ro->ro_dst; 393 if (ro->ro_nh == NULL) { 394 dst->sin_family = AF_INET; 395 dst->sin_len = sizeof(*dst); 396 dst->sin_addr = ip->ip_dst; 397 } 398 gw = (const struct sockaddr *)dst; 399 again: 400 /* 401 * Validate route against routing table additions; 402 * a better/more specific route might have been added. 403 */ 404 if (inp != NULL && ro->ro_nh != NULL) 405 NH_VALIDATE(ro, &inp->inp_rt_cookie, fibnum); 406 /* 407 * If there is a cached route, 408 * check that it is to the same destination 409 * and is still up. If not, free it and try again. 410 * The address family should also be checked in case of sharing the 411 * cache with IPv6. 412 * Also check whether routing cache needs invalidation. 413 */ 414 if (ro->ro_nh != NULL && 415 ((!NH_IS_VALID(ro->ro_nh)) || dst->sin_family != AF_INET || 416 dst->sin_addr.s_addr != ip->ip_dst.s_addr)) 417 RO_INVALIDATE_CACHE(ro); 418 ia = NULL; 419 /* 420 * If routing to interface only, short circuit routing lookup. 421 * The use of an all-ones broadcast address implies this; an 422 * interface is specified by the broadcast address of an interface, 423 * or the destination address of a ptp interface. 424 */ 425 if (flags & IP_SENDONES) { 426 if ((ia = ifatoia(ifa_ifwithbroadaddr(sintosa(dst), 427 M_GETFIB(m)))) == NULL && 428 (ia = ifatoia(ifa_ifwithdstaddr(sintosa(dst), 429 M_GETFIB(m)))) == NULL) { 430 IPSTAT_INC(ips_noroute); 431 error = ENETUNREACH; 432 goto bad; 433 } 434 ip->ip_dst.s_addr = INADDR_BROADCAST; 435 dst->sin_addr = ip->ip_dst; 436 ifp = ia->ia_ifp; 437 mtu = ifp->if_mtu; 438 ip->ip_ttl = 1; 439 isbroadcast = 1; 440 src = IA_SIN(ia)->sin_addr; 441 } else if (flags & IP_ROUTETOIF) { 442 if ((ia = ifatoia(ifa_ifwithdstaddr(sintosa(dst), 443 M_GETFIB(m)))) == NULL && 444 (ia = ifatoia(ifa_ifwithnet(sintosa(dst), 0, 445 M_GETFIB(m)))) == NULL) { 446 IPSTAT_INC(ips_noroute); 447 error = ENETUNREACH; 448 goto bad; 449 } 450 ifp = ia->ia_ifp; 451 mtu = ifp->if_mtu; 452 ip->ip_ttl = 1; 453 isbroadcast = ifp->if_flags & IFF_BROADCAST ? 454 in_ifaddr_broadcast(dst->sin_addr, ia) : 0; 455 src = IA_SIN(ia)->sin_addr; 456 } else if (IN_MULTICAST(ntohl(ip->ip_dst.s_addr)) && 457 imo != NULL && imo->imo_multicast_ifp != NULL) { 458 /* 459 * Bypass the normal routing lookup for multicast 460 * packets if the interface is specified. 461 */ 462 ifp = imo->imo_multicast_ifp; 463 mtu = ifp->if_mtu; 464 IFP_TO_IA(ifp, ia); 465 isbroadcast = 0; /* fool gcc */ 466 /* Interface may have no addresses. */ 467 if (ia != NULL) 468 src = IA_SIN(ia)->sin_addr; 469 else 470 src.s_addr = INADDR_ANY; 471 } else if (ro != &iproute) { 472 if (ro->ro_nh == NULL) { 473 /* 474 * We want to do any cloning requested by the link 475 * layer, as this is probably required in all cases 476 * for correct operation (as it is for ARP). 477 */ 478 uint32_t flowid; 479 flowid = m->m_pkthdr.flowid; 480 ro->ro_nh = fib4_lookup(fibnum, dst->sin_addr, 0, 481 NHR_REF, flowid); 482 483 if (ro->ro_nh == NULL || (!NH_IS_VALID(ro->ro_nh))) { 484 #if defined(IPSEC) || defined(IPSEC_SUPPORT) 485 /* 486 * There is no route for this packet, but it is 487 * possible that a matching SPD entry exists. 488 */ 489 no_route_but_check_spd = 1; 490 goto sendit; 491 #endif 492 IPSTAT_INC(ips_noroute); 493 error = EHOSTUNREACH; 494 goto bad; 495 } 496 } 497 struct nhop_object *nh = ro->ro_nh; 498 499 ia = ifatoia(nh->nh_ifa); 500 ifp = nh->nh_ifp; 501 counter_u64_add(nh->nh_pksent, 1); 502 rt_update_ro_flags(ro, nh); 503 if (nh->nh_flags & NHF_GATEWAY) 504 gw = &nh->gw_sa; 505 if (nh->nh_flags & NHF_HOST) 506 isbroadcast = (nh->nh_flags & NHF_BROADCAST); 507 else if ((ifp->if_flags & IFF_BROADCAST) && (gw->sa_family == AF_INET)) 508 isbroadcast = in_ifaddr_broadcast(((const struct sockaddr_in *)gw)->sin_addr, ia); 509 else 510 isbroadcast = 0; 511 mtu = nh->nh_mtu; 512 src = IA_SIN(ia)->sin_addr; 513 } else { 514 struct nhop_object *nh; 515 516 nh = fib4_lookup(M_GETFIB(m), ip->ip_dst, 0, NHR_NONE, 517 m->m_pkthdr.flowid); 518 if (nh == NULL) { 519 #if defined(IPSEC) || defined(IPSEC_SUPPORT) 520 /* 521 * There is no route for this packet, but it is 522 * possible that a matching SPD entry exists. 523 */ 524 no_route_but_check_spd = 1; 525 goto sendit; 526 #endif 527 IPSTAT_INC(ips_noroute); 528 error = EHOSTUNREACH; 529 goto bad; 530 } 531 ifp = nh->nh_ifp; 532 mtu = nh->nh_mtu; 533 rt_update_ro_flags(ro, nh); 534 if (nh->nh_flags & NHF_GATEWAY) 535 gw = &nh->gw_sa; 536 ia = ifatoia(nh->nh_ifa); 537 src = IA_SIN(ia)->sin_addr; 538 isbroadcast = (((nh->nh_flags & (NHF_HOST | NHF_BROADCAST)) == 539 (NHF_HOST | NHF_BROADCAST)) || 540 ((ifp->if_flags & IFF_BROADCAST) && 541 (gw->sa_family == AF_INET) && 542 in_ifaddr_broadcast(((const struct sockaddr_in *)gw)->sin_addr, ia))); 543 } 544 545 /* Catch a possible divide by zero later. */ 546 KASSERT(mtu > 0, ("%s: mtu %d <= 0, ro=%p (nh_flags=0x%08x) ifp=%p", 547 __func__, mtu, ro, 548 (ro != NULL && ro->ro_nh != NULL) ? ro->ro_nh->nh_flags : 0, ifp)); 549 550 if (IN_MULTICAST(ntohl(ip->ip_dst.s_addr))) { 551 m->m_flags |= M_MCAST; 552 /* 553 * IP destination address is multicast. Make sure "gw" 554 * still points to the address in "ro". (It may have been 555 * changed to point to a gateway address, above.) 556 */ 557 gw = (const struct sockaddr *)dst; 558 /* 559 * See if the caller provided any multicast options 560 */ 561 if (imo != NULL) { 562 ip->ip_ttl = imo->imo_multicast_ttl; 563 if (imo->imo_multicast_vif != -1) 564 ip->ip_src.s_addr = 565 ip_mcast_src ? 566 ip_mcast_src(imo->imo_multicast_vif) : 567 INADDR_ANY; 568 } else 569 ip->ip_ttl = IP_DEFAULT_MULTICAST_TTL; 570 /* 571 * Confirm that the outgoing interface supports multicast. 572 */ 573 if ((imo == NULL) || (imo->imo_multicast_vif == -1)) { 574 if ((ifp->if_flags & IFF_MULTICAST) == 0) { 575 IPSTAT_INC(ips_noroute); 576 error = ENETUNREACH; 577 goto bad; 578 } 579 } 580 /* 581 * If source address not specified yet, use address 582 * of outgoing interface. 583 */ 584 if (ip->ip_src.s_addr == INADDR_ANY) 585 ip->ip_src = src; 586 587 if ((imo == NULL && in_mcast_loop) || 588 (imo && imo->imo_multicast_loop)) { 589 /* 590 * Loop back multicast datagram if not expressly 591 * forbidden to do so, even if we are not a member 592 * of the group; ip_input() will filter it later, 593 * thus deferring a hash lookup and mutex acquisition 594 * at the expense of a cheap copy using m_copym(). 595 */ 596 ip_mloopback(ifp, m, hlen); 597 } else { 598 /* 599 * If we are acting as a multicast router, perform 600 * multicast forwarding as if the packet had just 601 * arrived on the interface to which we are about 602 * to send. The multicast forwarding function 603 * recursively calls this function, using the 604 * IP_FORWARDING flag to prevent infinite recursion. 605 * 606 * Multicasts that are looped back by ip_mloopback(), 607 * above, will be forwarded by the ip_input() routine, 608 * if necessary. 609 */ 610 MROUTER_RLOCK(); 611 if (V_ip_mrouter && (flags & IP_FORWARDING) == 0) { 612 /* 613 * If rsvp daemon is not running, do not 614 * set ip_moptions. This ensures that the packet 615 * is multicast and not just sent down one link 616 * as prescribed by rsvpd. 617 */ 618 if (!V_rsvp_on) 619 imo = NULL; 620 if (ip_mforward && 621 ip_mforward(ip, ifp, m, imo) != 0) { 622 MROUTER_RUNLOCK(); 623 m_freem(m); 624 goto done; 625 } 626 } 627 MROUTER_RUNLOCK(); 628 } 629 630 /* 631 * Multicasts with a time-to-live of zero may be looped- 632 * back, above, but must not be transmitted on a network. 633 * Also, multicasts addressed to the loopback interface 634 * are not sent -- the above call to ip_mloopback() will 635 * loop back a copy. ip_input() will drop the copy if 636 * this host does not belong to the destination group on 637 * the loopback interface. 638 */ 639 if (ip->ip_ttl == 0 || ifp->if_flags & IFF_LOOPBACK) { 640 m_freem(m); 641 goto done; 642 } 643 644 goto sendit; 645 } 646 647 /* 648 * If the source address is not specified yet, use the address 649 * of the outoing interface. 650 */ 651 if (ip->ip_src.s_addr == INADDR_ANY) 652 ip->ip_src = src; 653 654 /* 655 * Look for broadcast address and 656 * verify user is allowed to send 657 * such a packet. 658 */ 659 if (isbroadcast) { 660 if ((ifp->if_flags & IFF_BROADCAST) == 0) { 661 error = EADDRNOTAVAIL; 662 goto bad; 663 } 664 if ((flags & IP_ALLOWBROADCAST) == 0) { 665 error = EACCES; 666 goto bad; 667 } 668 /* don't allow broadcast messages to be fragmented */ 669 if (ip_len > mtu) { 670 error = EMSGSIZE; 671 goto bad; 672 } 673 m->m_flags |= M_BCAST; 674 } else { 675 m->m_flags &= ~M_BCAST; 676 } 677 678 sendit: 679 #if defined(IPSEC) || defined(IPSEC_SUPPORT) 680 if (IPSEC_ENABLED(ipv4)) { 681 if ((error = IPSEC_OUTPUT(ipv4, m, inp)) != 0) { 682 if (error == EINPROGRESS) 683 error = 0; 684 goto done; 685 } 686 } 687 /* 688 * Check if there was a route for this packet; return error if not. 689 */ 690 if (no_route_but_check_spd) { 691 IPSTAT_INC(ips_noroute); 692 error = EHOSTUNREACH; 693 goto bad; 694 } 695 /* Update variables that are affected by ipsec4_output(). */ 696 ip = mtod(m, struct ip *); 697 hlen = ip->ip_hl << 2; 698 #endif /* IPSEC */ 699 700 /* Jump over all PFIL processing if hooks are not active. */ 701 if (PFIL_HOOKED_OUT(V_inet_pfil_head)) { 702 switch (ip_output_pfil(&m, ifp, flags, inp, dst, &fibnum, 703 &error)) { 704 case 1: /* Finished */ 705 goto done; 706 707 case 0: /* Continue normally */ 708 ip = mtod(m, struct ip *); 709 break; 710 711 case -1: /* Need to try again */ 712 /* Reset everything for a new round */ 713 if (ro != NULL) { 714 RO_NHFREE(ro); 715 ro->ro_prepend = NULL; 716 } 717 gw = (const struct sockaddr *)dst; 718 ip = mtod(m, struct ip *); 719 goto again; 720 } 721 } 722 723 if (vlan_pcp > -1) 724 EVL_APPLY_PRI(m, vlan_pcp); 725 726 /* IN_LOOPBACK must not appear on the wire - RFC1122. */ 727 if (IN_LOOPBACK(ntohl(ip->ip_dst.s_addr)) || 728 IN_LOOPBACK(ntohl(ip->ip_src.s_addr))) { 729 if ((ifp->if_flags & IFF_LOOPBACK) == 0) { 730 IPSTAT_INC(ips_badaddr); 731 error = EADDRNOTAVAIL; 732 goto bad; 733 } 734 } 735 736 m->m_pkthdr.csum_flags |= CSUM_IP; 737 if (m->m_pkthdr.csum_flags & CSUM_DELAY_DATA & ~ifp->if_hwassist) { 738 m = mb_unmapped_to_ext(m); 739 if (m == NULL) { 740 IPSTAT_INC(ips_odropped); 741 error = ENOBUFS; 742 goto bad; 743 } 744 in_delayed_cksum(m); 745 m->m_pkthdr.csum_flags &= ~CSUM_DELAY_DATA; 746 } else if ((ifp->if_capenable & IFCAP_MEXTPG) == 0) { 747 m = mb_unmapped_to_ext(m); 748 if (m == NULL) { 749 IPSTAT_INC(ips_odropped); 750 error = ENOBUFS; 751 goto bad; 752 } 753 } 754 #if defined(SCTP) || defined(SCTP_SUPPORT) 755 if (m->m_pkthdr.csum_flags & CSUM_SCTP & ~ifp->if_hwassist) { 756 sctp_delayed_cksum(m, (uint32_t)(ip->ip_hl << 2)); 757 m->m_pkthdr.csum_flags &= ~CSUM_SCTP; 758 } 759 #endif 760 761 /* 762 * If small enough for interface, or the interface will take 763 * care of the fragmentation for us, we can just send directly. 764 * Note that if_vxlan could have requested TSO even though the outer 765 * frame is UDP. It is correct to not fragment such datagrams and 766 * instead just pass them on to the driver. 767 */ 768 if (ip_len <= mtu || 769 (m->m_pkthdr.csum_flags & ifp->if_hwassist & 770 (CSUM_TSO | CSUM_INNER_TSO)) != 0) { 771 ip->ip_sum = 0; 772 if (m->m_pkthdr.csum_flags & CSUM_IP & ~ifp->if_hwassist) { 773 ip->ip_sum = in_cksum(m, hlen); 774 m->m_pkthdr.csum_flags &= ~CSUM_IP; 775 } 776 777 /* 778 * Record statistics for this interface address. 779 * With CSUM_TSO the byte/packet count will be slightly 780 * incorrect because we count the IP+TCP headers only 781 * once instead of for every generated packet. 782 */ 783 if (!(flags & IP_FORWARDING) && ia) { 784 if (m->m_pkthdr.csum_flags & 785 (CSUM_TSO | CSUM_INNER_TSO)) 786 counter_u64_add(ia->ia_ifa.ifa_opackets, 787 m->m_pkthdr.len / m->m_pkthdr.tso_segsz); 788 else 789 counter_u64_add(ia->ia_ifa.ifa_opackets, 1); 790 791 counter_u64_add(ia->ia_ifa.ifa_obytes, m->m_pkthdr.len); 792 } 793 #ifdef MBUF_STRESS_TEST 794 if (mbuf_frag_size && m->m_pkthdr.len > mbuf_frag_size) 795 m = m_fragment(m, M_NOWAIT, mbuf_frag_size); 796 #endif 797 /* 798 * Reset layer specific mbuf flags 799 * to avoid confusing lower layers. 800 */ 801 m_clrprotoflags(m); 802 IP_PROBE(send, NULL, NULL, ip, ifp, ip, NULL); 803 error = ip_output_send(inp, ifp, m, gw, ro, 804 (flags & IP_NO_SND_TAG_RL) ? false : true); 805 goto done; 806 } 807 808 /* Balk when DF bit is set or the interface didn't support TSO. */ 809 if ((ip_off & IP_DF) || 810 (m->m_pkthdr.csum_flags & (CSUM_TSO | CSUM_INNER_TSO))) { 811 error = EMSGSIZE; 812 IPSTAT_INC(ips_cantfrag); 813 goto bad; 814 } 815 816 /* 817 * Too large for interface; fragment if possible. If successful, 818 * on return, m will point to a list of packets to be sent. 819 */ 820 error = ip_fragment(ip, &m, mtu, ifp->if_hwassist); 821 if (error) 822 goto bad; 823 for (; m; m = m0) { 824 m0 = m->m_nextpkt; 825 m->m_nextpkt = 0; 826 if (error == 0) { 827 /* Record statistics for this interface address. */ 828 if (ia != NULL) { 829 counter_u64_add(ia->ia_ifa.ifa_opackets, 1); 830 counter_u64_add(ia->ia_ifa.ifa_obytes, 831 m->m_pkthdr.len); 832 } 833 /* 834 * Reset layer specific mbuf flags 835 * to avoid confusing upper layers. 836 */ 837 m_clrprotoflags(m); 838 839 IP_PROBE(send, NULL, NULL, mtod(m, struct ip *), ifp, 840 mtod(m, struct ip *), NULL); 841 error = ip_output_send(inp, ifp, m, gw, ro, true); 842 } else 843 m_freem(m); 844 } 845 846 if (error == 0) 847 IPSTAT_INC(ips_fragmented); 848 849 done: 850 return (error); 851 bad: 852 m_freem(m); 853 goto done; 854 } 855 856 /* 857 * Create a chain of fragments which fit the given mtu. m_frag points to the 858 * mbuf to be fragmented; on return it points to the chain with the fragments. 859 * Return 0 if no error. If error, m_frag may contain a partially built 860 * chain of fragments that should be freed by the caller. 861 * 862 * if_hwassist_flags is the hw offload capabilities (see if_data.ifi_hwassist) 863 */ 864 int 865 ip_fragment(struct ip *ip, struct mbuf **m_frag, int mtu, 866 u_long if_hwassist_flags) 867 { 868 int error = 0; 869 int hlen = ip->ip_hl << 2; 870 int len = (mtu - hlen) & ~7; /* size of payload in each fragment */ 871 int off; 872 struct mbuf *m0 = *m_frag; /* the original packet */ 873 int firstlen; 874 struct mbuf **mnext; 875 int nfrags; 876 uint16_t ip_len, ip_off; 877 878 ip_len = ntohs(ip->ip_len); 879 ip_off = ntohs(ip->ip_off); 880 881 if (ip_off & IP_DF) { /* Fragmentation not allowed */ 882 IPSTAT_INC(ips_cantfrag); 883 return EMSGSIZE; 884 } 885 886 /* 887 * Must be able to put at least 8 bytes per fragment. 888 */ 889 if (len < 8) 890 return EMSGSIZE; 891 892 /* 893 * If the interface will not calculate checksums on 894 * fragmented packets, then do it here. 895 */ 896 if (m0->m_pkthdr.csum_flags & CSUM_DELAY_DATA) { 897 m0 = mb_unmapped_to_ext(m0); 898 if (m0 == NULL) { 899 error = ENOBUFS; 900 IPSTAT_INC(ips_odropped); 901 goto done; 902 } 903 in_delayed_cksum(m0); 904 m0->m_pkthdr.csum_flags &= ~CSUM_DELAY_DATA; 905 } 906 #if defined(SCTP) || defined(SCTP_SUPPORT) 907 if (m0->m_pkthdr.csum_flags & CSUM_SCTP) { 908 sctp_delayed_cksum(m0, hlen); 909 m0->m_pkthdr.csum_flags &= ~CSUM_SCTP; 910 } 911 #endif 912 if (len > PAGE_SIZE) { 913 /* 914 * Fragment large datagrams such that each segment 915 * contains a multiple of PAGE_SIZE amount of data, 916 * plus headers. This enables a receiver to perform 917 * page-flipping zero-copy optimizations. 918 * 919 * XXX When does this help given that sender and receiver 920 * could have different page sizes, and also mtu could 921 * be less than the receiver's page size ? 922 */ 923 int newlen; 924 925 off = MIN(mtu, m0->m_pkthdr.len); 926 927 /* 928 * firstlen (off - hlen) must be aligned on an 929 * 8-byte boundary 930 */ 931 if (off < hlen) 932 goto smart_frag_failure; 933 off = ((off - hlen) & ~7) + hlen; 934 newlen = (~PAGE_MASK) & mtu; 935 if ((newlen + sizeof (struct ip)) > mtu) { 936 /* we failed, go back the default */ 937 smart_frag_failure: 938 newlen = len; 939 off = hlen + len; 940 } 941 len = newlen; 942 943 } else { 944 off = hlen + len; 945 } 946 947 firstlen = off - hlen; 948 mnext = &m0->m_nextpkt; /* pointer to next packet */ 949 950 /* 951 * Loop through length of segment after first fragment, 952 * make new header and copy data of each part and link onto chain. 953 * Here, m0 is the original packet, m is the fragment being created. 954 * The fragments are linked off the m_nextpkt of the original 955 * packet, which after processing serves as the first fragment. 956 */ 957 for (nfrags = 1; off < ip_len; off += len, nfrags++) { 958 struct ip *mhip; /* ip header on the fragment */ 959 struct mbuf *m; 960 int mhlen = sizeof (struct ip); 961 962 m = m_gethdr(M_NOWAIT, MT_DATA); 963 if (m == NULL) { 964 error = ENOBUFS; 965 IPSTAT_INC(ips_odropped); 966 goto done; 967 } 968 /* 969 * Make sure the complete packet header gets copied 970 * from the originating mbuf to the newly created 971 * mbuf. This also ensures that existing firewall 972 * classification(s), VLAN tags and so on get copied 973 * to the resulting fragmented packet(s): 974 */ 975 if (m_dup_pkthdr(m, m0, M_NOWAIT) == 0) { 976 m_free(m); 977 error = ENOBUFS; 978 IPSTAT_INC(ips_odropped); 979 goto done; 980 } 981 /* 982 * In the first mbuf, leave room for the link header, then 983 * copy the original IP header including options. The payload 984 * goes into an additional mbuf chain returned by m_copym(). 985 */ 986 m->m_data += max_linkhdr; 987 mhip = mtod(m, struct ip *); 988 *mhip = *ip; 989 if (hlen > sizeof (struct ip)) { 990 mhlen = ip_optcopy(ip, mhip) + sizeof (struct ip); 991 mhip->ip_v = IPVERSION; 992 mhip->ip_hl = mhlen >> 2; 993 } 994 m->m_len = mhlen; 995 /* XXX do we need to add ip_off below ? */ 996 mhip->ip_off = ((off - hlen) >> 3) + ip_off; 997 if (off + len >= ip_len) 998 len = ip_len - off; 999 else 1000 mhip->ip_off |= IP_MF; 1001 mhip->ip_len = htons((u_short)(len + mhlen)); 1002 m->m_next = m_copym(m0, off, len, M_NOWAIT); 1003 if (m->m_next == NULL) { /* copy failed */ 1004 m_free(m); 1005 error = ENOBUFS; /* ??? */ 1006 IPSTAT_INC(ips_odropped); 1007 goto done; 1008 } 1009 m->m_pkthdr.len = mhlen + len; 1010 #ifdef MAC 1011 mac_netinet_fragment(m0, m); 1012 #endif 1013 mhip->ip_off = htons(mhip->ip_off); 1014 mhip->ip_sum = 0; 1015 if (m->m_pkthdr.csum_flags & CSUM_IP & ~if_hwassist_flags) { 1016 mhip->ip_sum = in_cksum(m, mhlen); 1017 m->m_pkthdr.csum_flags &= ~CSUM_IP; 1018 } 1019 *mnext = m; 1020 mnext = &m->m_nextpkt; 1021 } 1022 IPSTAT_ADD(ips_ofragments, nfrags); 1023 1024 /* 1025 * Update first fragment by trimming what's been copied out 1026 * and updating header. 1027 */ 1028 m_adj(m0, hlen + firstlen - ip_len); 1029 m0->m_pkthdr.len = hlen + firstlen; 1030 ip->ip_len = htons((u_short)m0->m_pkthdr.len); 1031 ip->ip_off = htons(ip_off | IP_MF); 1032 ip->ip_sum = 0; 1033 if (m0->m_pkthdr.csum_flags & CSUM_IP & ~if_hwassist_flags) { 1034 ip->ip_sum = in_cksum(m0, hlen); 1035 m0->m_pkthdr.csum_flags &= ~CSUM_IP; 1036 } 1037 1038 done: 1039 *m_frag = m0; 1040 return error; 1041 } 1042 1043 void 1044 in_delayed_cksum(struct mbuf *m) 1045 { 1046 struct ip *ip; 1047 struct udphdr *uh; 1048 uint16_t cklen, csum, offset; 1049 1050 ip = mtod(m, struct ip *); 1051 offset = ip->ip_hl << 2 ; 1052 1053 if (m->m_pkthdr.csum_flags & CSUM_UDP) { 1054 /* if udp header is not in the first mbuf copy udplen */ 1055 if (offset + sizeof(struct udphdr) > m->m_len) { 1056 m_copydata(m, offset + offsetof(struct udphdr, 1057 uh_ulen), sizeof(cklen), (caddr_t)&cklen); 1058 cklen = ntohs(cklen); 1059 } else { 1060 uh = (struct udphdr *)mtodo(m, offset); 1061 cklen = ntohs(uh->uh_ulen); 1062 } 1063 csum = in_cksum_skip(m, cklen + offset, offset); 1064 if (csum == 0) 1065 csum = 0xffff; 1066 } else { 1067 cklen = ntohs(ip->ip_len); 1068 csum = in_cksum_skip(m, cklen, offset); 1069 } 1070 offset += m->m_pkthdr.csum_data; /* checksum offset */ 1071 1072 if (offset + sizeof(csum) > m->m_len) 1073 m_copyback(m, offset, sizeof(csum), (caddr_t)&csum); 1074 else 1075 *(u_short *)mtodo(m, offset) = csum; 1076 } 1077 1078 /* 1079 * IP socket option processing. 1080 */ 1081 int 1082 ip_ctloutput(struct socket *so, struct sockopt *sopt) 1083 { 1084 struct inpcb *inp = sotoinpcb(so); 1085 int error, optval; 1086 #ifdef RSS 1087 uint32_t rss_bucket; 1088 int retval; 1089 #endif 1090 1091 error = optval = 0; 1092 if (sopt->sopt_level != IPPROTO_IP) { 1093 error = EINVAL; 1094 1095 if (sopt->sopt_level == SOL_SOCKET && 1096 sopt->sopt_dir == SOPT_SET) { 1097 switch (sopt->sopt_name) { 1098 case SO_REUSEADDR: 1099 INP_WLOCK(inp); 1100 if ((so->so_options & SO_REUSEADDR) != 0) 1101 inp->inp_flags2 |= INP_REUSEADDR; 1102 else 1103 inp->inp_flags2 &= ~INP_REUSEADDR; 1104 INP_WUNLOCK(inp); 1105 error = 0; 1106 break; 1107 case SO_REUSEPORT: 1108 INP_WLOCK(inp); 1109 if ((so->so_options & SO_REUSEPORT) != 0) 1110 inp->inp_flags2 |= INP_REUSEPORT; 1111 else 1112 inp->inp_flags2 &= ~INP_REUSEPORT; 1113 INP_WUNLOCK(inp); 1114 error = 0; 1115 break; 1116 case SO_REUSEPORT_LB: 1117 INP_WLOCK(inp); 1118 if ((so->so_options & SO_REUSEPORT_LB) != 0) 1119 inp->inp_flags2 |= INP_REUSEPORT_LB; 1120 else 1121 inp->inp_flags2 &= ~INP_REUSEPORT_LB; 1122 INP_WUNLOCK(inp); 1123 error = 0; 1124 break; 1125 case SO_SETFIB: 1126 INP_WLOCK(inp); 1127 inp->inp_inc.inc_fibnum = so->so_fibnum; 1128 INP_WUNLOCK(inp); 1129 error = 0; 1130 break; 1131 case SO_MAX_PACING_RATE: 1132 #ifdef RATELIMIT 1133 INP_WLOCK(inp); 1134 inp->inp_flags2 |= INP_RATE_LIMIT_CHANGED; 1135 INP_WUNLOCK(inp); 1136 error = 0; 1137 #else 1138 error = EOPNOTSUPP; 1139 #endif 1140 break; 1141 default: 1142 break; 1143 } 1144 } 1145 return (error); 1146 } 1147 1148 switch (sopt->sopt_dir) { 1149 case SOPT_SET: 1150 switch (sopt->sopt_name) { 1151 case IP_OPTIONS: 1152 #ifdef notyet 1153 case IP_RETOPTS: 1154 #endif 1155 { 1156 struct mbuf *m; 1157 if (sopt->sopt_valsize > MLEN) { 1158 error = EMSGSIZE; 1159 break; 1160 } 1161 m = m_get(sopt->sopt_td ? M_WAITOK : M_NOWAIT, MT_DATA); 1162 if (m == NULL) { 1163 error = ENOBUFS; 1164 break; 1165 } 1166 m->m_len = sopt->sopt_valsize; 1167 error = sooptcopyin(sopt, mtod(m, char *), m->m_len, 1168 m->m_len); 1169 if (error) { 1170 m_free(m); 1171 break; 1172 } 1173 INP_WLOCK(inp); 1174 error = ip_pcbopts(inp, sopt->sopt_name, m); 1175 INP_WUNLOCK(inp); 1176 return (error); 1177 } 1178 1179 case IP_BINDANY: 1180 if (sopt->sopt_td != NULL) { 1181 error = priv_check(sopt->sopt_td, 1182 PRIV_NETINET_BINDANY); 1183 if (error) 1184 break; 1185 } 1186 /* FALLTHROUGH */ 1187 case IP_BINDMULTI: 1188 #ifdef RSS 1189 case IP_RSS_LISTEN_BUCKET: 1190 #endif 1191 case IP_TOS: 1192 case IP_TTL: 1193 case IP_MINTTL: 1194 case IP_RECVOPTS: 1195 case IP_RECVRETOPTS: 1196 case IP_ORIGDSTADDR: 1197 case IP_RECVDSTADDR: 1198 case IP_RECVTTL: 1199 case IP_RECVIF: 1200 case IP_ONESBCAST: 1201 case IP_DONTFRAG: 1202 case IP_RECVTOS: 1203 case IP_RECVFLOWID: 1204 #ifdef RSS 1205 case IP_RECVRSSBUCKETID: 1206 #endif 1207 case IP_VLAN_PCP: 1208 error = sooptcopyin(sopt, &optval, sizeof optval, 1209 sizeof optval); 1210 if (error) 1211 break; 1212 1213 switch (sopt->sopt_name) { 1214 case IP_TOS: 1215 inp->inp_ip_tos = optval; 1216 break; 1217 1218 case IP_TTL: 1219 inp->inp_ip_ttl = optval; 1220 break; 1221 1222 case IP_MINTTL: 1223 if (optval >= 0 && optval <= MAXTTL) 1224 inp->inp_ip_minttl = optval; 1225 else 1226 error = EINVAL; 1227 break; 1228 1229 #define OPTSET(bit) do { \ 1230 INP_WLOCK(inp); \ 1231 if (optval) \ 1232 inp->inp_flags |= bit; \ 1233 else \ 1234 inp->inp_flags &= ~bit; \ 1235 INP_WUNLOCK(inp); \ 1236 } while (0) 1237 1238 #define OPTSET2(bit, val) do { \ 1239 INP_WLOCK(inp); \ 1240 if (val) \ 1241 inp->inp_flags2 |= bit; \ 1242 else \ 1243 inp->inp_flags2 &= ~bit; \ 1244 INP_WUNLOCK(inp); \ 1245 } while (0) 1246 1247 case IP_RECVOPTS: 1248 OPTSET(INP_RECVOPTS); 1249 break; 1250 1251 case IP_RECVRETOPTS: 1252 OPTSET(INP_RECVRETOPTS); 1253 break; 1254 1255 case IP_RECVDSTADDR: 1256 OPTSET(INP_RECVDSTADDR); 1257 break; 1258 1259 case IP_ORIGDSTADDR: 1260 OPTSET2(INP_ORIGDSTADDR, optval); 1261 break; 1262 1263 case IP_RECVTTL: 1264 OPTSET(INP_RECVTTL); 1265 break; 1266 1267 case IP_RECVIF: 1268 OPTSET(INP_RECVIF); 1269 break; 1270 1271 case IP_ONESBCAST: 1272 OPTSET(INP_ONESBCAST); 1273 break; 1274 case IP_DONTFRAG: 1275 OPTSET(INP_DONTFRAG); 1276 break; 1277 case IP_BINDANY: 1278 OPTSET(INP_BINDANY); 1279 break; 1280 case IP_RECVTOS: 1281 OPTSET(INP_RECVTOS); 1282 break; 1283 case IP_BINDMULTI: 1284 OPTSET2(INP_BINDMULTI, optval); 1285 break; 1286 case IP_RECVFLOWID: 1287 OPTSET2(INP_RECVFLOWID, optval); 1288 break; 1289 #ifdef RSS 1290 case IP_RSS_LISTEN_BUCKET: 1291 if ((optval >= 0) && 1292 (optval < rss_getnumbuckets())) { 1293 inp->inp_rss_listen_bucket = optval; 1294 OPTSET2(INP_RSS_BUCKET_SET, 1); 1295 } else { 1296 error = EINVAL; 1297 } 1298 break; 1299 case IP_RECVRSSBUCKETID: 1300 OPTSET2(INP_RECVRSSBUCKETID, optval); 1301 break; 1302 #endif 1303 case IP_VLAN_PCP: 1304 if ((optval >= -1) && (optval <= 1305 (INP_2PCP_MASK >> INP_2PCP_SHIFT))) { 1306 if (optval == -1) { 1307 INP_WLOCK(inp); 1308 inp->inp_flags2 &= 1309 ~(INP_2PCP_SET | 1310 INP_2PCP_MASK); 1311 INP_WUNLOCK(inp); 1312 } else { 1313 INP_WLOCK(inp); 1314 inp->inp_flags2 |= 1315 INP_2PCP_SET; 1316 inp->inp_flags2 &= 1317 ~INP_2PCP_MASK; 1318 inp->inp_flags2 |= 1319 optval << INP_2PCP_SHIFT; 1320 INP_WUNLOCK(inp); 1321 } 1322 } else 1323 error = EINVAL; 1324 break; 1325 } 1326 break; 1327 #undef OPTSET 1328 #undef OPTSET2 1329 1330 /* 1331 * Multicast socket options are processed by the in_mcast 1332 * module. 1333 */ 1334 case IP_MULTICAST_IF: 1335 case IP_MULTICAST_VIF: 1336 case IP_MULTICAST_TTL: 1337 case IP_MULTICAST_LOOP: 1338 case IP_ADD_MEMBERSHIP: 1339 case IP_DROP_MEMBERSHIP: 1340 case IP_ADD_SOURCE_MEMBERSHIP: 1341 case IP_DROP_SOURCE_MEMBERSHIP: 1342 case IP_BLOCK_SOURCE: 1343 case IP_UNBLOCK_SOURCE: 1344 case IP_MSFILTER: 1345 case MCAST_JOIN_GROUP: 1346 case MCAST_LEAVE_GROUP: 1347 case MCAST_JOIN_SOURCE_GROUP: 1348 case MCAST_LEAVE_SOURCE_GROUP: 1349 case MCAST_BLOCK_SOURCE: 1350 case MCAST_UNBLOCK_SOURCE: 1351 error = inp_setmoptions(inp, sopt); 1352 break; 1353 1354 case IP_PORTRANGE: 1355 error = sooptcopyin(sopt, &optval, sizeof optval, 1356 sizeof optval); 1357 if (error) 1358 break; 1359 1360 INP_WLOCK(inp); 1361 switch (optval) { 1362 case IP_PORTRANGE_DEFAULT: 1363 inp->inp_flags &= ~(INP_LOWPORT); 1364 inp->inp_flags &= ~(INP_HIGHPORT); 1365 break; 1366 1367 case IP_PORTRANGE_HIGH: 1368 inp->inp_flags &= ~(INP_LOWPORT); 1369 inp->inp_flags |= INP_HIGHPORT; 1370 break; 1371 1372 case IP_PORTRANGE_LOW: 1373 inp->inp_flags &= ~(INP_HIGHPORT); 1374 inp->inp_flags |= INP_LOWPORT; 1375 break; 1376 1377 default: 1378 error = EINVAL; 1379 break; 1380 } 1381 INP_WUNLOCK(inp); 1382 break; 1383 1384 #if defined(IPSEC) || defined(IPSEC_SUPPORT) 1385 case IP_IPSEC_POLICY: 1386 if (IPSEC_ENABLED(ipv4)) { 1387 error = IPSEC_PCBCTL(ipv4, inp, sopt); 1388 break; 1389 } 1390 /* FALLTHROUGH */ 1391 #endif /* IPSEC */ 1392 1393 default: 1394 error = ENOPROTOOPT; 1395 break; 1396 } 1397 break; 1398 1399 case SOPT_GET: 1400 switch (sopt->sopt_name) { 1401 case IP_OPTIONS: 1402 case IP_RETOPTS: 1403 INP_RLOCK(inp); 1404 if (inp->inp_options) { 1405 struct mbuf *options; 1406 1407 options = m_copym(inp->inp_options, 0, 1408 M_COPYALL, M_NOWAIT); 1409 INP_RUNLOCK(inp); 1410 if (options != NULL) { 1411 error = sooptcopyout(sopt, 1412 mtod(options, char *), 1413 options->m_len); 1414 m_freem(options); 1415 } else 1416 error = ENOMEM; 1417 } else { 1418 INP_RUNLOCK(inp); 1419 sopt->sopt_valsize = 0; 1420 } 1421 break; 1422 1423 case IP_TOS: 1424 case IP_TTL: 1425 case IP_MINTTL: 1426 case IP_RECVOPTS: 1427 case IP_RECVRETOPTS: 1428 case IP_ORIGDSTADDR: 1429 case IP_RECVDSTADDR: 1430 case IP_RECVTTL: 1431 case IP_RECVIF: 1432 case IP_PORTRANGE: 1433 case IP_ONESBCAST: 1434 case IP_DONTFRAG: 1435 case IP_BINDANY: 1436 case IP_RECVTOS: 1437 case IP_BINDMULTI: 1438 case IP_FLOWID: 1439 case IP_FLOWTYPE: 1440 case IP_RECVFLOWID: 1441 #ifdef RSS 1442 case IP_RSSBUCKETID: 1443 case IP_RECVRSSBUCKETID: 1444 #endif 1445 case IP_VLAN_PCP: 1446 switch (sopt->sopt_name) { 1447 case IP_TOS: 1448 optval = inp->inp_ip_tos; 1449 break; 1450 1451 case IP_TTL: 1452 optval = inp->inp_ip_ttl; 1453 break; 1454 1455 case IP_MINTTL: 1456 optval = inp->inp_ip_minttl; 1457 break; 1458 1459 #define OPTBIT(bit) (inp->inp_flags & bit ? 1 : 0) 1460 #define OPTBIT2(bit) (inp->inp_flags2 & bit ? 1 : 0) 1461 1462 case IP_RECVOPTS: 1463 optval = OPTBIT(INP_RECVOPTS); 1464 break; 1465 1466 case IP_RECVRETOPTS: 1467 optval = OPTBIT(INP_RECVRETOPTS); 1468 break; 1469 1470 case IP_RECVDSTADDR: 1471 optval = OPTBIT(INP_RECVDSTADDR); 1472 break; 1473 1474 case IP_ORIGDSTADDR: 1475 optval = OPTBIT2(INP_ORIGDSTADDR); 1476 break; 1477 1478 case IP_RECVTTL: 1479 optval = OPTBIT(INP_RECVTTL); 1480 break; 1481 1482 case IP_RECVIF: 1483 optval = OPTBIT(INP_RECVIF); 1484 break; 1485 1486 case IP_PORTRANGE: 1487 if (inp->inp_flags & INP_HIGHPORT) 1488 optval = IP_PORTRANGE_HIGH; 1489 else if (inp->inp_flags & INP_LOWPORT) 1490 optval = IP_PORTRANGE_LOW; 1491 else 1492 optval = 0; 1493 break; 1494 1495 case IP_ONESBCAST: 1496 optval = OPTBIT(INP_ONESBCAST); 1497 break; 1498 case IP_DONTFRAG: 1499 optval = OPTBIT(INP_DONTFRAG); 1500 break; 1501 case IP_BINDANY: 1502 optval = OPTBIT(INP_BINDANY); 1503 break; 1504 case IP_RECVTOS: 1505 optval = OPTBIT(INP_RECVTOS); 1506 break; 1507 case IP_FLOWID: 1508 optval = inp->inp_flowid; 1509 break; 1510 case IP_FLOWTYPE: 1511 optval = inp->inp_flowtype; 1512 break; 1513 case IP_RECVFLOWID: 1514 optval = OPTBIT2(INP_RECVFLOWID); 1515 break; 1516 #ifdef RSS 1517 case IP_RSSBUCKETID: 1518 retval = rss_hash2bucket(inp->inp_flowid, 1519 inp->inp_flowtype, 1520 &rss_bucket); 1521 if (retval == 0) 1522 optval = rss_bucket; 1523 else 1524 error = EINVAL; 1525 break; 1526 case IP_RECVRSSBUCKETID: 1527 optval = OPTBIT2(INP_RECVRSSBUCKETID); 1528 break; 1529 #endif 1530 case IP_BINDMULTI: 1531 optval = OPTBIT2(INP_BINDMULTI); 1532 break; 1533 case IP_VLAN_PCP: 1534 if (OPTBIT2(INP_2PCP_SET)) { 1535 optval = (inp->inp_flags2 & 1536 INP_2PCP_MASK) >> INP_2PCP_SHIFT; 1537 } else { 1538 optval = -1; 1539 } 1540 break; 1541 } 1542 error = sooptcopyout(sopt, &optval, sizeof optval); 1543 break; 1544 1545 /* 1546 * Multicast socket options are processed by the in_mcast 1547 * module. 1548 */ 1549 case IP_MULTICAST_IF: 1550 case IP_MULTICAST_VIF: 1551 case IP_MULTICAST_TTL: 1552 case IP_MULTICAST_LOOP: 1553 case IP_MSFILTER: 1554 error = inp_getmoptions(inp, sopt); 1555 break; 1556 1557 #if defined(IPSEC) || defined(IPSEC_SUPPORT) 1558 case IP_IPSEC_POLICY: 1559 if (IPSEC_ENABLED(ipv4)) { 1560 error = IPSEC_PCBCTL(ipv4, inp, sopt); 1561 break; 1562 } 1563 /* FALLTHROUGH */ 1564 #endif /* IPSEC */ 1565 1566 default: 1567 error = ENOPROTOOPT; 1568 break; 1569 } 1570 break; 1571 } 1572 return (error); 1573 } 1574 1575 /* 1576 * Routine called from ip_output() to loop back a copy of an IP multicast 1577 * packet to the input queue of a specified interface. Note that this 1578 * calls the output routine of the loopback "driver", but with an interface 1579 * pointer that might NOT be a loopback interface -- evil, but easier than 1580 * replicating that code here. 1581 */ 1582 static void 1583 ip_mloopback(struct ifnet *ifp, const struct mbuf *m, int hlen) 1584 { 1585 struct ip *ip; 1586 struct mbuf *copym; 1587 1588 /* 1589 * Make a deep copy of the packet because we're going to 1590 * modify the pack in order to generate checksums. 1591 */ 1592 copym = m_dup(m, M_NOWAIT); 1593 if (copym != NULL && (!M_WRITABLE(copym) || copym->m_len < hlen)) 1594 copym = m_pullup(copym, hlen); 1595 if (copym != NULL) { 1596 /* If needed, compute the checksum and mark it as valid. */ 1597 if (copym->m_pkthdr.csum_flags & CSUM_DELAY_DATA) { 1598 in_delayed_cksum(copym); 1599 copym->m_pkthdr.csum_flags &= ~CSUM_DELAY_DATA; 1600 copym->m_pkthdr.csum_flags |= 1601 CSUM_DATA_VALID | CSUM_PSEUDO_HDR; 1602 copym->m_pkthdr.csum_data = 0xffff; 1603 } 1604 /* 1605 * We don't bother to fragment if the IP length is greater 1606 * than the interface's MTU. Can this possibly matter? 1607 */ 1608 ip = mtod(copym, struct ip *); 1609 ip->ip_sum = 0; 1610 ip->ip_sum = in_cksum(copym, hlen); 1611 if_simloop(ifp, copym, AF_INET, 0); 1612 } 1613 } 1614