ip_output.c revision 331722
1/*- 2 * Copyright (c) 1982, 1986, 1988, 1990, 1993 3 * The Regents of the University of California. All rights reserved. 4 * 5 * Redistribution and use in source and binary forms, with or without 6 * modification, are permitted provided that the following conditions 7 * are met: 8 * 1. Redistributions of source code must retain the above copyright 9 * notice, this list of conditions and the following disclaimer. 10 * 2. Redistributions in binary form must reproduce the above copyright 11 * notice, this list of conditions and the following disclaimer in the 12 * documentation and/or other materials provided with the distribution. 13 * 4. Neither the name of the University nor the names of its contributors 14 * may be used to endorse or promote products derived from this software 15 * without specific prior written permission. 16 * 17 * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND 18 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE 19 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE 20 * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE 21 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL 22 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS 23 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) 24 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT 25 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY 26 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF 27 * SUCH DAMAGE. 28 * 29 * @(#)ip_output.c 8.3 (Berkeley) 1/21/94 30 */ 31 32#include <sys/cdefs.h> 33__FBSDID("$FreeBSD: stable/11/sys/netinet/ip_output.c 331722 2018-03-29 02:50:57Z eadler $"); 34 35#include "opt_inet.h" 36#include "opt_ipsec.h" 37#include "opt_mbuf_stress_test.h" 38#include "opt_mpath.h" 39#include "opt_route.h" 40#include "opt_sctp.h" 41#include "opt_rss.h" 42 43#include <sys/param.h> 44#include <sys/systm.h> 45#include <sys/kernel.h> 46#include <sys/lock.h> 47#include <sys/malloc.h> 48#include <sys/mbuf.h> 49#include <sys/priv.h> 50#include <sys/proc.h> 51#include <sys/protosw.h> 52#include <sys/rmlock.h> 53#include <sys/sdt.h> 54#include <sys/socket.h> 55#include <sys/socketvar.h> 56#include <sys/sysctl.h> 57#include <sys/ucred.h> 58 59#include <net/if.h> 60#include <net/if_var.h> 61#include <net/if_llatbl.h> 62#include <net/netisr.h> 63#include <net/pfil.h> 64#include <net/route.h> 65#include <net/flowtable.h> 66#ifdef RADIX_MPATH 67#include <net/radix_mpath.h> 68#endif 69#include <net/rss_config.h> 70#include <net/vnet.h> 71 72#include <netinet/in.h> 73#include <netinet/in_kdtrace.h> 74#include <netinet/in_systm.h> 75#include <netinet/ip.h> 76#include <netinet/in_pcb.h> 77#include <netinet/in_rss.h> 78#include <netinet/in_var.h> 79#include <netinet/ip_var.h> 80#include <netinet/ip_options.h> 81#ifdef SCTP 82#include <netinet/sctp.h> 83#include <netinet/sctp_crc32.h> 84#endif 85 86#include <netipsec/ipsec_support.h> 87 88#include <machine/in_cksum.h> 89 90#include <security/mac/mac_framework.h> 91 92#ifdef MBUF_STRESS_TEST 93static int mbuf_frag_size = 0; 94SYSCTL_INT(_net_inet_ip, OID_AUTO, mbuf_frag_size, CTLFLAG_RW, 95 &mbuf_frag_size, 0, "Fragment outgoing mbufs to this size"); 96#endif 97 98static void ip_mloopback(struct ifnet *, const struct mbuf *, int); 99 100 101extern int in_mcast_loop; 102extern struct protosw inetsw[]; 103 104static inline int 105ip_output_pfil(struct mbuf **mp, struct ifnet *ifp, struct inpcb *inp, 106 struct sockaddr_in *dst, int *fibnum, int *error) 107{ 108 struct m_tag *fwd_tag = NULL; 109 struct mbuf *m; 110 struct in_addr odst; 111 struct ip *ip; 112 113 m = *mp; 114 ip = mtod(m, struct ip *); 115 116 /* Run through list of hooks for output packets. */ 117 odst.s_addr = ip->ip_dst.s_addr; 118 *error = pfil_run_hooks(&V_inet_pfil_hook, mp, ifp, PFIL_OUT, inp); 119 m = *mp; 120 if ((*error) != 0 || m == NULL) 121 return 1; /* Finished */ 122 123 ip = mtod(m, struct ip *); 124 125 /* See if destination IP address was changed by packet filter. */ 126 if (odst.s_addr != ip->ip_dst.s_addr) { 127 m->m_flags |= M_SKIP_FIREWALL; 128 /* If destination is now ourself drop to ip_input(). */ 129 if (in_localip(ip->ip_dst)) { 130 m->m_flags |= M_FASTFWD_OURS; 131 if (m->m_pkthdr.rcvif == NULL) 132 m->m_pkthdr.rcvif = V_loif; 133 if (m->m_pkthdr.csum_flags & CSUM_DELAY_DATA) { 134 m->m_pkthdr.csum_flags |= 135 CSUM_DATA_VALID | CSUM_PSEUDO_HDR; 136 m->m_pkthdr.csum_data = 0xffff; 137 } 138 m->m_pkthdr.csum_flags |= 139 CSUM_IP_CHECKED | CSUM_IP_VALID; 140#ifdef SCTP 141 if (m->m_pkthdr.csum_flags & CSUM_SCTP) 142 m->m_pkthdr.csum_flags |= CSUM_SCTP_VALID; 143#endif 144 *error = netisr_queue(NETISR_IP, m); 145 return 1; /* Finished */ 146 } 147 148 bzero(dst, sizeof(*dst)); 149 dst->sin_family = AF_INET; 150 dst->sin_len = sizeof(*dst); 151 dst->sin_addr = ip->ip_dst; 152 153 return -1; /* Reloop */ 154 } 155 /* See if fib was changed by packet filter. */ 156 if ((*fibnum) != M_GETFIB(m)) { 157 m->m_flags |= M_SKIP_FIREWALL; 158 *fibnum = M_GETFIB(m); 159 return -1; /* Reloop for FIB change */ 160 } 161 162 /* See if local, if yes, send it to netisr with IP_FASTFWD_OURS. */ 163 if (m->m_flags & M_FASTFWD_OURS) { 164 if (m->m_pkthdr.rcvif == NULL) 165 m->m_pkthdr.rcvif = V_loif; 166 if (m->m_pkthdr.csum_flags & CSUM_DELAY_DATA) { 167 m->m_pkthdr.csum_flags |= 168 CSUM_DATA_VALID | CSUM_PSEUDO_HDR; 169 m->m_pkthdr.csum_data = 0xffff; 170 } 171#ifdef SCTP 172 if (m->m_pkthdr.csum_flags & CSUM_SCTP) 173 m->m_pkthdr.csum_flags |= CSUM_SCTP_VALID; 174#endif 175 m->m_pkthdr.csum_flags |= 176 CSUM_IP_CHECKED | CSUM_IP_VALID; 177 178 *error = netisr_queue(NETISR_IP, m); 179 return 1; /* Finished */ 180 } 181 /* Or forward to some other address? */ 182 if ((m->m_flags & M_IP_NEXTHOP) && 183 ((fwd_tag = m_tag_find(m, PACKET_TAG_IPFORWARD, NULL)) != NULL)) { 184 bcopy((fwd_tag+1), dst, sizeof(struct sockaddr_in)); 185 m->m_flags |= M_SKIP_FIREWALL; 186 m->m_flags &= ~M_IP_NEXTHOP; 187 m_tag_delete(m, fwd_tag); 188 189 return -1; /* Reloop for CHANGE of dst */ 190 } 191 192 return 0; 193} 194 195/* 196 * IP output. The packet in mbuf chain m contains a skeletal IP 197 * header (with len, off, ttl, proto, tos, src, dst). 198 * The mbuf chain containing the packet will be freed. 199 * The mbuf opt, if present, will not be freed. 200 * If route ro is present and has ro_rt initialized, route lookup would be 201 * skipped and ro->ro_rt would be used. If ro is present but ro->ro_rt is NULL, 202 * then result of route lookup is stored in ro->ro_rt. 203 * 204 * In the IP forwarding case, the packet will arrive with options already 205 * inserted, so must have a NULL opt pointer. 206 */ 207int 208ip_output(struct mbuf *m, struct mbuf *opt, struct route *ro, int flags, 209 struct ip_moptions *imo, struct inpcb *inp) 210{ 211 struct rm_priotracker in_ifa_tracker; 212 struct ip *ip; 213 struct ifnet *ifp = NULL; /* keep compiler happy */ 214 struct mbuf *m0; 215 int hlen = sizeof (struct ip); 216 int mtu; 217 int error = 0; 218 struct sockaddr_in *dst; 219 const struct sockaddr_in *gw; 220 struct in_ifaddr *ia; 221 int isbroadcast; 222 uint16_t ip_len, ip_off; 223 struct route iproute; 224 struct rtentry *rte; /* cache for ro->ro_rt */ 225 uint32_t fibnum; 226 int have_ia_ref; 227#if defined(IPSEC) || defined(IPSEC_SUPPORT) 228 int no_route_but_check_spd = 0; 229#endif 230 M_ASSERTPKTHDR(m); 231 232 if (inp != NULL) { 233 INP_LOCK_ASSERT(inp); 234 M_SETFIB(m, inp->inp_inc.inc_fibnum); 235 if ((flags & IP_NODEFAULTFLOWID) == 0) { 236 m->m_pkthdr.flowid = inp->inp_flowid; 237 M_HASHTYPE_SET(m, inp->inp_flowtype); 238 } 239 } 240 241 if (ro == NULL) { 242 ro = &iproute; 243 bzero(ro, sizeof (*ro)); 244 } 245 246#ifdef FLOWTABLE 247 if (ro->ro_rt == NULL) 248 (void )flowtable_lookup(AF_INET, m, ro); 249#endif 250 251 if (opt) { 252 int len = 0; 253 m = ip_insertoptions(m, opt, &len); 254 if (len != 0) 255 hlen = len; /* ip->ip_hl is updated above */ 256 } 257 ip = mtod(m, struct ip *); 258 ip_len = ntohs(ip->ip_len); 259 ip_off = ntohs(ip->ip_off); 260 261 if ((flags & (IP_FORWARDING|IP_RAWOUTPUT)) == 0) { 262 ip->ip_v = IPVERSION; 263 ip->ip_hl = hlen >> 2; 264 ip_fillid(ip); 265 IPSTAT_INC(ips_localout); 266 } else { 267 /* Header already set, fetch hlen from there */ 268 hlen = ip->ip_hl << 2; 269 } 270 271 /* 272 * dst/gw handling: 273 * 274 * dst can be rewritten but always points to &ro->ro_dst. 275 * gw is readonly but can point either to dst OR rt_gateway, 276 * therefore we need restore gw if we're redoing lookup. 277 */ 278 gw = dst = (struct sockaddr_in *)&ro->ro_dst; 279 fibnum = (inp != NULL) ? inp->inp_inc.inc_fibnum : M_GETFIB(m); 280 rte = ro->ro_rt; 281 if (rte == NULL) { 282 bzero(dst, sizeof(*dst)); 283 dst->sin_family = AF_INET; 284 dst->sin_len = sizeof(*dst); 285 dst->sin_addr = ip->ip_dst; 286 } 287again: 288 /* 289 * Validate route against routing table additions; 290 * a better/more specific route might have been added. 291 */ 292 if (inp) 293 RT_VALIDATE(ro, &inp->inp_rt_cookie, fibnum); 294 /* 295 * If there is a cached route, 296 * check that it is to the same destination 297 * and is still up. If not, free it and try again. 298 * The address family should also be checked in case of sharing the 299 * cache with IPv6. 300 * Also check whether routing cache needs invalidation. 301 */ 302 rte = ro->ro_rt; 303 if (rte && ((rte->rt_flags & RTF_UP) == 0 || 304 rte->rt_ifp == NULL || 305 !RT_LINK_IS_UP(rte->rt_ifp) || 306 dst->sin_family != AF_INET || 307 dst->sin_addr.s_addr != ip->ip_dst.s_addr)) { 308 RTFREE(rte); 309 rte = ro->ro_rt = (struct rtentry *)NULL; 310 if (ro->ro_lle) 311 LLE_FREE(ro->ro_lle); /* zeros ro_lle */ 312 ro->ro_lle = (struct llentry *)NULL; 313 } 314 ia = NULL; 315 have_ia_ref = 0; 316 /* 317 * If routing to interface only, short circuit routing lookup. 318 * The use of an all-ones broadcast address implies this; an 319 * interface is specified by the broadcast address of an interface, 320 * or the destination address of a ptp interface. 321 */ 322 if (flags & IP_SENDONES) { 323 if ((ia = ifatoia(ifa_ifwithbroadaddr(sintosa(dst), 324 M_GETFIB(m)))) == NULL && 325 (ia = ifatoia(ifa_ifwithdstaddr(sintosa(dst), 326 M_GETFIB(m)))) == NULL) { 327 IPSTAT_INC(ips_noroute); 328 error = ENETUNREACH; 329 goto bad; 330 } 331 have_ia_ref = 1; 332 ip->ip_dst.s_addr = INADDR_BROADCAST; 333 dst->sin_addr = ip->ip_dst; 334 ifp = ia->ia_ifp; 335 ip->ip_ttl = 1; 336 isbroadcast = 1; 337 } else if (flags & IP_ROUTETOIF) { 338 if ((ia = ifatoia(ifa_ifwithdstaddr(sintosa(dst), 339 M_GETFIB(m)))) == NULL && 340 (ia = ifatoia(ifa_ifwithnet(sintosa(dst), 0, 341 M_GETFIB(m)))) == NULL) { 342 IPSTAT_INC(ips_noroute); 343 error = ENETUNREACH; 344 goto bad; 345 } 346 have_ia_ref = 1; 347 ifp = ia->ia_ifp; 348 ip->ip_ttl = 1; 349 isbroadcast = in_broadcast(dst->sin_addr, ifp); 350 } else if (IN_MULTICAST(ntohl(ip->ip_dst.s_addr)) && 351 imo != NULL && imo->imo_multicast_ifp != NULL) { 352 /* 353 * Bypass the normal routing lookup for multicast 354 * packets if the interface is specified. 355 */ 356 ifp = imo->imo_multicast_ifp; 357 IFP_TO_IA(ifp, ia, &in_ifa_tracker); 358 if (ia) 359 have_ia_ref = 1; 360 isbroadcast = 0; /* fool gcc */ 361 } else { 362 /* 363 * We want to do any cloning requested by the link layer, 364 * as this is probably required in all cases for correct 365 * operation (as it is for ARP). 366 */ 367 if (rte == NULL) { 368#ifdef RADIX_MPATH 369 rtalloc_mpath_fib(ro, 370 ntohl(ip->ip_src.s_addr ^ ip->ip_dst.s_addr), 371 fibnum); 372#else 373 in_rtalloc_ign(ro, 0, fibnum); 374#endif 375 rte = ro->ro_rt; 376 } 377 if (rte == NULL || 378 (rte->rt_flags & RTF_UP) == 0 || 379 rte->rt_ifp == NULL || 380 !RT_LINK_IS_UP(rte->rt_ifp)) { 381#if defined(IPSEC) || defined(IPSEC_SUPPORT) 382 /* 383 * There is no route for this packet, but it is 384 * possible that a matching SPD entry exists. 385 */ 386 no_route_but_check_spd = 1; 387 mtu = 0; /* Silence GCC warning. */ 388 goto sendit; 389#endif 390 IPSTAT_INC(ips_noroute); 391 error = EHOSTUNREACH; 392 goto bad; 393 } 394 ia = ifatoia(rte->rt_ifa); 395 ifp = rte->rt_ifp; 396 counter_u64_add(rte->rt_pksent, 1); 397 rt_update_ro_flags(ro); 398 if (rte->rt_flags & RTF_GATEWAY) 399 gw = (struct sockaddr_in *)rte->rt_gateway; 400 if (rte->rt_flags & RTF_HOST) 401 isbroadcast = (rte->rt_flags & RTF_BROADCAST); 402 else 403 isbroadcast = in_broadcast(gw->sin_addr, ifp); 404 } 405 406 /* 407 * Calculate MTU. If we have a route that is up, use that, 408 * otherwise use the interface's MTU. 409 */ 410 if (rte != NULL && (rte->rt_flags & (RTF_UP|RTF_HOST))) 411 mtu = rte->rt_mtu; 412 else 413 mtu = ifp->if_mtu; 414 /* Catch a possible divide by zero later. */ 415 KASSERT(mtu > 0, ("%s: mtu %d <= 0, rte=%p (rt_flags=0x%08x) ifp=%p", 416 __func__, mtu, rte, (rte != NULL) ? rte->rt_flags : 0, ifp)); 417 418 if (IN_MULTICAST(ntohl(ip->ip_dst.s_addr))) { 419 m->m_flags |= M_MCAST; 420 /* 421 * IP destination address is multicast. Make sure "gw" 422 * still points to the address in "ro". (It may have been 423 * changed to point to a gateway address, above.) 424 */ 425 gw = dst; 426 /* 427 * See if the caller provided any multicast options 428 */ 429 if (imo != NULL) { 430 ip->ip_ttl = imo->imo_multicast_ttl; 431 if (imo->imo_multicast_vif != -1) 432 ip->ip_src.s_addr = 433 ip_mcast_src ? 434 ip_mcast_src(imo->imo_multicast_vif) : 435 INADDR_ANY; 436 } else 437 ip->ip_ttl = IP_DEFAULT_MULTICAST_TTL; 438 /* 439 * Confirm that the outgoing interface supports multicast. 440 */ 441 if ((imo == NULL) || (imo->imo_multicast_vif == -1)) { 442 if ((ifp->if_flags & IFF_MULTICAST) == 0) { 443 IPSTAT_INC(ips_noroute); 444 error = ENETUNREACH; 445 goto bad; 446 } 447 } 448 /* 449 * If source address not specified yet, use address 450 * of outgoing interface. 451 */ 452 if (ip->ip_src.s_addr == INADDR_ANY) { 453 /* Interface may have no addresses. */ 454 if (ia != NULL) 455 ip->ip_src = IA_SIN(ia)->sin_addr; 456 } 457 458 if ((imo == NULL && in_mcast_loop) || 459 (imo && imo->imo_multicast_loop)) { 460 /* 461 * Loop back multicast datagram if not expressly 462 * forbidden to do so, even if we are not a member 463 * of the group; ip_input() will filter it later, 464 * thus deferring a hash lookup and mutex acquisition 465 * at the expense of a cheap copy using m_copym(). 466 */ 467 ip_mloopback(ifp, m, hlen); 468 } else { 469 /* 470 * If we are acting as a multicast router, perform 471 * multicast forwarding as if the packet had just 472 * arrived on the interface to which we are about 473 * to send. The multicast forwarding function 474 * recursively calls this function, using the 475 * IP_FORWARDING flag to prevent infinite recursion. 476 * 477 * Multicasts that are looped back by ip_mloopback(), 478 * above, will be forwarded by the ip_input() routine, 479 * if necessary. 480 */ 481 if (V_ip_mrouter && (flags & IP_FORWARDING) == 0) { 482 /* 483 * If rsvp daemon is not running, do not 484 * set ip_moptions. This ensures that the packet 485 * is multicast and not just sent down one link 486 * as prescribed by rsvpd. 487 */ 488 if (!V_rsvp_on) 489 imo = NULL; 490 if (ip_mforward && 491 ip_mforward(ip, ifp, m, imo) != 0) { 492 m_freem(m); 493 goto done; 494 } 495 } 496 } 497 498 /* 499 * Multicasts with a time-to-live of zero may be looped- 500 * back, above, but must not be transmitted on a network. 501 * Also, multicasts addressed to the loopback interface 502 * are not sent -- the above call to ip_mloopback() will 503 * loop back a copy. ip_input() will drop the copy if 504 * this host does not belong to the destination group on 505 * the loopback interface. 506 */ 507 if (ip->ip_ttl == 0 || ifp->if_flags & IFF_LOOPBACK) { 508 m_freem(m); 509 goto done; 510 } 511 512 goto sendit; 513 } 514 515 /* 516 * If the source address is not specified yet, use the address 517 * of the outoing interface. 518 */ 519 if (ip->ip_src.s_addr == INADDR_ANY) { 520 /* Interface may have no addresses. */ 521 if (ia != NULL) { 522 ip->ip_src = IA_SIN(ia)->sin_addr; 523 } 524 } 525 526 /* 527 * Look for broadcast address and 528 * verify user is allowed to send 529 * such a packet. 530 */ 531 if (isbroadcast) { 532 if ((ifp->if_flags & IFF_BROADCAST) == 0) { 533 error = EADDRNOTAVAIL; 534 goto bad; 535 } 536 if ((flags & IP_ALLOWBROADCAST) == 0) { 537 error = EACCES; 538 goto bad; 539 } 540 /* don't allow broadcast messages to be fragmented */ 541 if (ip_len > mtu) { 542 error = EMSGSIZE; 543 goto bad; 544 } 545 m->m_flags |= M_BCAST; 546 } else { 547 m->m_flags &= ~M_BCAST; 548 } 549 550sendit: 551#if defined(IPSEC) || defined(IPSEC_SUPPORT) 552 if (IPSEC_ENABLED(ipv4)) { 553 if ((error = IPSEC_OUTPUT(ipv4, m, inp)) != 0) { 554 if (error == EINPROGRESS) 555 error = 0; 556 goto done; 557 } 558 } 559 /* 560 * Check if there was a route for this packet; return error if not. 561 */ 562 if (no_route_but_check_spd) { 563 IPSTAT_INC(ips_noroute); 564 error = EHOSTUNREACH; 565 goto bad; 566 } 567 /* Update variables that are affected by ipsec4_output(). */ 568 ip = mtod(m, struct ip *); 569 hlen = ip->ip_hl << 2; 570#endif /* IPSEC */ 571 572 /* Jump over all PFIL processing if hooks are not active. */ 573 if (PFIL_HOOKED(&V_inet_pfil_hook)) { 574 switch (ip_output_pfil(&m, ifp, inp, dst, &fibnum, &error)) { 575 case 1: /* Finished */ 576 goto done; 577 578 case 0: /* Continue normally */ 579 ip = mtod(m, struct ip *); 580 break; 581 582 case -1: /* Need to try again */ 583 /* Reset everything for a new round */ 584 RO_RTFREE(ro); 585 if (have_ia_ref) 586 ifa_free(&ia->ia_ifa); 587 ro->ro_prepend = NULL; 588 rte = NULL; 589 gw = dst; 590 ip = mtod(m, struct ip *); 591 goto again; 592 593 } 594 } 595 596 /* 127/8 must not appear on wire - RFC1122. */ 597 if ((ntohl(ip->ip_dst.s_addr) >> IN_CLASSA_NSHIFT) == IN_LOOPBACKNET || 598 (ntohl(ip->ip_src.s_addr) >> IN_CLASSA_NSHIFT) == IN_LOOPBACKNET) { 599 if ((ifp->if_flags & IFF_LOOPBACK) == 0) { 600 IPSTAT_INC(ips_badaddr); 601 error = EADDRNOTAVAIL; 602 goto bad; 603 } 604 } 605 606 m->m_pkthdr.csum_flags |= CSUM_IP; 607 if (m->m_pkthdr.csum_flags & CSUM_DELAY_DATA & ~ifp->if_hwassist) { 608 in_delayed_cksum(m); 609 m->m_pkthdr.csum_flags &= ~CSUM_DELAY_DATA; 610 } 611#ifdef SCTP 612 if (m->m_pkthdr.csum_flags & CSUM_SCTP & ~ifp->if_hwassist) { 613 sctp_delayed_cksum(m, (uint32_t)(ip->ip_hl << 2)); 614 m->m_pkthdr.csum_flags &= ~CSUM_SCTP; 615 } 616#endif 617 618 /* 619 * If small enough for interface, or the interface will take 620 * care of the fragmentation for us, we can just send directly. 621 */ 622 if (ip_len <= mtu || 623 (m->m_pkthdr.csum_flags & ifp->if_hwassist & CSUM_TSO) != 0) { 624 ip->ip_sum = 0; 625 if (m->m_pkthdr.csum_flags & CSUM_IP & ~ifp->if_hwassist) { 626 ip->ip_sum = in_cksum(m, hlen); 627 m->m_pkthdr.csum_flags &= ~CSUM_IP; 628 } 629 630 /* 631 * Record statistics for this interface address. 632 * With CSUM_TSO the byte/packet count will be slightly 633 * incorrect because we count the IP+TCP headers only 634 * once instead of for every generated packet. 635 */ 636 if (!(flags & IP_FORWARDING) && ia) { 637 if (m->m_pkthdr.csum_flags & CSUM_TSO) 638 counter_u64_add(ia->ia_ifa.ifa_opackets, 639 m->m_pkthdr.len / m->m_pkthdr.tso_segsz); 640 else 641 counter_u64_add(ia->ia_ifa.ifa_opackets, 1); 642 643 counter_u64_add(ia->ia_ifa.ifa_obytes, m->m_pkthdr.len); 644 } 645#ifdef MBUF_STRESS_TEST 646 if (mbuf_frag_size && m->m_pkthdr.len > mbuf_frag_size) 647 m = m_fragment(m, M_NOWAIT, mbuf_frag_size); 648#endif 649 /* 650 * Reset layer specific mbuf flags 651 * to avoid confusing lower layers. 652 */ 653 m_clrprotoflags(m); 654 IP_PROBE(send, NULL, NULL, ip, ifp, ip, NULL); 655 error = (*ifp->if_output)(ifp, m, 656 (const struct sockaddr *)gw, ro); 657 goto done; 658 } 659 660 /* Balk when DF bit is set or the interface didn't support TSO. */ 661 if ((ip_off & IP_DF) || (m->m_pkthdr.csum_flags & CSUM_TSO)) { 662 error = EMSGSIZE; 663 IPSTAT_INC(ips_cantfrag); 664 goto bad; 665 } 666 667 /* 668 * Too large for interface; fragment if possible. If successful, 669 * on return, m will point to a list of packets to be sent. 670 */ 671 error = ip_fragment(ip, &m, mtu, ifp->if_hwassist); 672 if (error) 673 goto bad; 674 for (; m; m = m0) { 675 m0 = m->m_nextpkt; 676 m->m_nextpkt = 0; 677 if (error == 0) { 678 /* Record statistics for this interface address. */ 679 if (ia != NULL) { 680 counter_u64_add(ia->ia_ifa.ifa_opackets, 1); 681 counter_u64_add(ia->ia_ifa.ifa_obytes, 682 m->m_pkthdr.len); 683 } 684 /* 685 * Reset layer specific mbuf flags 686 * to avoid confusing upper layers. 687 */ 688 m_clrprotoflags(m); 689 690 IP_PROBE(send, NULL, NULL, mtod(m, struct ip *), ifp, 691 mtod(m, struct ip *), NULL); 692 error = (*ifp->if_output)(ifp, m, 693 (const struct sockaddr *)gw, ro); 694 } else 695 m_freem(m); 696 } 697 698 if (error == 0) 699 IPSTAT_INC(ips_fragmented); 700 701done: 702 /* 703 * Release the route if using our private route, or if 704 * (with flowtable) we don't have our own reference. 705 */ 706 if (ro == &iproute || ro->ro_flags & RT_NORTREF) 707 RO_RTFREE(ro); 708 else if (rte == NULL) 709 /* 710 * If the caller supplied a route but somehow the reference 711 * to it has been released need to prevent the caller 712 * calling RTFREE on it again. 713 */ 714 ro->ro_rt = NULL; 715 if (have_ia_ref) 716 ifa_free(&ia->ia_ifa); 717 return (error); 718bad: 719 m_freem(m); 720 goto done; 721} 722 723/* 724 * Create a chain of fragments which fit the given mtu. m_frag points to the 725 * mbuf to be fragmented; on return it points to the chain with the fragments. 726 * Return 0 if no error. If error, m_frag may contain a partially built 727 * chain of fragments that should be freed by the caller. 728 * 729 * if_hwassist_flags is the hw offload capabilities (see if_data.ifi_hwassist) 730 */ 731int 732ip_fragment(struct ip *ip, struct mbuf **m_frag, int mtu, 733 u_long if_hwassist_flags) 734{ 735 int error = 0; 736 int hlen = ip->ip_hl << 2; 737 int len = (mtu - hlen) & ~7; /* size of payload in each fragment */ 738 int off; 739 struct mbuf *m0 = *m_frag; /* the original packet */ 740 int firstlen; 741 struct mbuf **mnext; 742 int nfrags; 743 uint16_t ip_len, ip_off; 744 745 ip_len = ntohs(ip->ip_len); 746 ip_off = ntohs(ip->ip_off); 747 748 if (ip_off & IP_DF) { /* Fragmentation not allowed */ 749 IPSTAT_INC(ips_cantfrag); 750 return EMSGSIZE; 751 } 752 753 /* 754 * Must be able to put at least 8 bytes per fragment. 755 */ 756 if (len < 8) 757 return EMSGSIZE; 758 759 /* 760 * If the interface will not calculate checksums on 761 * fragmented packets, then do it here. 762 */ 763 if (m0->m_pkthdr.csum_flags & CSUM_DELAY_DATA) { 764 in_delayed_cksum(m0); 765 m0->m_pkthdr.csum_flags &= ~CSUM_DELAY_DATA; 766 } 767#ifdef SCTP 768 if (m0->m_pkthdr.csum_flags & CSUM_SCTP) { 769 sctp_delayed_cksum(m0, hlen); 770 m0->m_pkthdr.csum_flags &= ~CSUM_SCTP; 771 } 772#endif 773 if (len > PAGE_SIZE) { 774 /* 775 * Fragment large datagrams such that each segment 776 * contains a multiple of PAGE_SIZE amount of data, 777 * plus headers. This enables a receiver to perform 778 * page-flipping zero-copy optimizations. 779 * 780 * XXX When does this help given that sender and receiver 781 * could have different page sizes, and also mtu could 782 * be less than the receiver's page size ? 783 */ 784 int newlen; 785 786 off = MIN(mtu, m0->m_pkthdr.len); 787 788 /* 789 * firstlen (off - hlen) must be aligned on an 790 * 8-byte boundary 791 */ 792 if (off < hlen) 793 goto smart_frag_failure; 794 off = ((off - hlen) & ~7) + hlen; 795 newlen = (~PAGE_MASK) & mtu; 796 if ((newlen + sizeof (struct ip)) > mtu) { 797 /* we failed, go back the default */ 798smart_frag_failure: 799 newlen = len; 800 off = hlen + len; 801 } 802 len = newlen; 803 804 } else { 805 off = hlen + len; 806 } 807 808 firstlen = off - hlen; 809 mnext = &m0->m_nextpkt; /* pointer to next packet */ 810 811 /* 812 * Loop through length of segment after first fragment, 813 * make new header and copy data of each part and link onto chain. 814 * Here, m0 is the original packet, m is the fragment being created. 815 * The fragments are linked off the m_nextpkt of the original 816 * packet, which after processing serves as the first fragment. 817 */ 818 for (nfrags = 1; off < ip_len; off += len, nfrags++) { 819 struct ip *mhip; /* ip header on the fragment */ 820 struct mbuf *m; 821 int mhlen = sizeof (struct ip); 822 823 m = m_gethdr(M_NOWAIT, MT_DATA); 824 if (m == NULL) { 825 error = ENOBUFS; 826 IPSTAT_INC(ips_odropped); 827 goto done; 828 } 829 /* 830 * Make sure the complete packet header gets copied 831 * from the originating mbuf to the newly created 832 * mbuf. This also ensures that existing firewall 833 * classification(s), VLAN tags and so on get copied 834 * to the resulting fragmented packet(s): 835 */ 836 if (m_dup_pkthdr(m, m0, M_NOWAIT) == 0) { 837 m_free(m); 838 error = ENOBUFS; 839 IPSTAT_INC(ips_odropped); 840 goto done; 841 } 842 /* 843 * In the first mbuf, leave room for the link header, then 844 * copy the original IP header including options. The payload 845 * goes into an additional mbuf chain returned by m_copym(). 846 */ 847 m->m_data += max_linkhdr; 848 mhip = mtod(m, struct ip *); 849 *mhip = *ip; 850 if (hlen > sizeof (struct ip)) { 851 mhlen = ip_optcopy(ip, mhip) + sizeof (struct ip); 852 mhip->ip_v = IPVERSION; 853 mhip->ip_hl = mhlen >> 2; 854 } 855 m->m_len = mhlen; 856 /* XXX do we need to add ip_off below ? */ 857 mhip->ip_off = ((off - hlen) >> 3) + ip_off; 858 if (off + len >= ip_len) 859 len = ip_len - off; 860 else 861 mhip->ip_off |= IP_MF; 862 mhip->ip_len = htons((u_short)(len + mhlen)); 863 m->m_next = m_copym(m0, off, len, M_NOWAIT); 864 if (m->m_next == NULL) { /* copy failed */ 865 m_free(m); 866 error = ENOBUFS; /* ??? */ 867 IPSTAT_INC(ips_odropped); 868 goto done; 869 } 870 m->m_pkthdr.len = mhlen + len; 871#ifdef MAC 872 mac_netinet_fragment(m0, m); 873#endif 874 mhip->ip_off = htons(mhip->ip_off); 875 mhip->ip_sum = 0; 876 if (m->m_pkthdr.csum_flags & CSUM_IP & ~if_hwassist_flags) { 877 mhip->ip_sum = in_cksum(m, mhlen); 878 m->m_pkthdr.csum_flags &= ~CSUM_IP; 879 } 880 *mnext = m; 881 mnext = &m->m_nextpkt; 882 } 883 IPSTAT_ADD(ips_ofragments, nfrags); 884 885 /* 886 * Update first fragment by trimming what's been copied out 887 * and updating header. 888 */ 889 m_adj(m0, hlen + firstlen - ip_len); 890 m0->m_pkthdr.len = hlen + firstlen; 891 ip->ip_len = htons((u_short)m0->m_pkthdr.len); 892 ip->ip_off = htons(ip_off | IP_MF); 893 ip->ip_sum = 0; 894 if (m0->m_pkthdr.csum_flags & CSUM_IP & ~if_hwassist_flags) { 895 ip->ip_sum = in_cksum(m0, hlen); 896 m0->m_pkthdr.csum_flags &= ~CSUM_IP; 897 } 898 899done: 900 *m_frag = m0; 901 return error; 902} 903 904void 905in_delayed_cksum(struct mbuf *m) 906{ 907 struct ip *ip; 908 uint16_t csum, offset, ip_len; 909 910 ip = mtod(m, struct ip *); 911 offset = ip->ip_hl << 2 ; 912 ip_len = ntohs(ip->ip_len); 913 csum = in_cksum_skip(m, ip_len, offset); 914 if (m->m_pkthdr.csum_flags & CSUM_UDP && csum == 0) 915 csum = 0xffff; 916 offset += m->m_pkthdr.csum_data; /* checksum offset */ 917 918 /* find the mbuf in the chain where the checksum starts*/ 919 while ((m != NULL) && (offset >= m->m_len)) { 920 offset -= m->m_len; 921 m = m->m_next; 922 } 923 KASSERT(m != NULL, ("in_delayed_cksum: checksum outside mbuf chain.")); 924 KASSERT(offset + sizeof(u_short) <= m->m_len, ("in_delayed_cksum: checksum split between mbufs.")); 925 *(u_short *)(m->m_data + offset) = csum; 926} 927 928/* 929 * IP socket option processing. 930 */ 931int 932ip_ctloutput(struct socket *so, struct sockopt *sopt) 933{ 934 struct inpcb *inp = sotoinpcb(so); 935 int error, optval; 936#ifdef RSS 937 uint32_t rss_bucket; 938 int retval; 939#endif 940 941 error = optval = 0; 942 if (sopt->sopt_level != IPPROTO_IP) { 943 error = EINVAL; 944 945 if (sopt->sopt_level == SOL_SOCKET && 946 sopt->sopt_dir == SOPT_SET) { 947 switch (sopt->sopt_name) { 948 case SO_REUSEADDR: 949 INP_WLOCK(inp); 950 if ((so->so_options & SO_REUSEADDR) != 0) 951 inp->inp_flags2 |= INP_REUSEADDR; 952 else 953 inp->inp_flags2 &= ~INP_REUSEADDR; 954 INP_WUNLOCK(inp); 955 error = 0; 956 break; 957 case SO_REUSEPORT: 958 INP_WLOCK(inp); 959 if ((so->so_options & SO_REUSEPORT) != 0) 960 inp->inp_flags2 |= INP_REUSEPORT; 961 else 962 inp->inp_flags2 &= ~INP_REUSEPORT; 963 INP_WUNLOCK(inp); 964 error = 0; 965 break; 966 case SO_SETFIB: 967 INP_WLOCK(inp); 968 inp->inp_inc.inc_fibnum = so->so_fibnum; 969 INP_WUNLOCK(inp); 970 error = 0; 971 break; 972 default: 973 break; 974 } 975 } 976 return (error); 977 } 978 979 switch (sopt->sopt_dir) { 980 case SOPT_SET: 981 switch (sopt->sopt_name) { 982 case IP_OPTIONS: 983#ifdef notyet 984 case IP_RETOPTS: 985#endif 986 { 987 struct mbuf *m; 988 if (sopt->sopt_valsize > MLEN) { 989 error = EMSGSIZE; 990 break; 991 } 992 m = m_get(sopt->sopt_td ? M_WAITOK : M_NOWAIT, MT_DATA); 993 if (m == NULL) { 994 error = ENOBUFS; 995 break; 996 } 997 m->m_len = sopt->sopt_valsize; 998 error = sooptcopyin(sopt, mtod(m, char *), m->m_len, 999 m->m_len); 1000 if (error) { 1001 m_free(m); 1002 break; 1003 } 1004 INP_WLOCK(inp); 1005 error = ip_pcbopts(inp, sopt->sopt_name, m); 1006 INP_WUNLOCK(inp); 1007 return (error); 1008 } 1009 1010 case IP_BINDANY: 1011 if (sopt->sopt_td != NULL) { 1012 error = priv_check(sopt->sopt_td, 1013 PRIV_NETINET_BINDANY); 1014 if (error) 1015 break; 1016 } 1017 /* FALLTHROUGH */ 1018 case IP_BINDMULTI: 1019#ifdef RSS 1020 case IP_RSS_LISTEN_BUCKET: 1021#endif 1022 case IP_TOS: 1023 case IP_TTL: 1024 case IP_MINTTL: 1025 case IP_RECVOPTS: 1026 case IP_RECVRETOPTS: 1027 case IP_RECVDSTADDR: 1028 case IP_RECVTTL: 1029 case IP_RECVIF: 1030 case IP_ONESBCAST: 1031 case IP_DONTFRAG: 1032 case IP_RECVTOS: 1033 case IP_RECVFLOWID: 1034#ifdef RSS 1035 case IP_RECVRSSBUCKETID: 1036#endif 1037 error = sooptcopyin(sopt, &optval, sizeof optval, 1038 sizeof optval); 1039 if (error) 1040 break; 1041 1042 switch (sopt->sopt_name) { 1043 case IP_TOS: 1044 inp->inp_ip_tos = optval; 1045 break; 1046 1047 case IP_TTL: 1048 inp->inp_ip_ttl = optval; 1049 break; 1050 1051 case IP_MINTTL: 1052 if (optval >= 0 && optval <= MAXTTL) 1053 inp->inp_ip_minttl = optval; 1054 else 1055 error = EINVAL; 1056 break; 1057 1058#define OPTSET(bit) do { \ 1059 INP_WLOCK(inp); \ 1060 if (optval) \ 1061 inp->inp_flags |= bit; \ 1062 else \ 1063 inp->inp_flags &= ~bit; \ 1064 INP_WUNLOCK(inp); \ 1065} while (0) 1066 1067#define OPTSET2(bit, val) do { \ 1068 INP_WLOCK(inp); \ 1069 if (val) \ 1070 inp->inp_flags2 |= bit; \ 1071 else \ 1072 inp->inp_flags2 &= ~bit; \ 1073 INP_WUNLOCK(inp); \ 1074} while (0) 1075 1076 case IP_RECVOPTS: 1077 OPTSET(INP_RECVOPTS); 1078 break; 1079 1080 case IP_RECVRETOPTS: 1081 OPTSET(INP_RECVRETOPTS); 1082 break; 1083 1084 case IP_RECVDSTADDR: 1085 OPTSET(INP_RECVDSTADDR); 1086 break; 1087 1088 case IP_RECVTTL: 1089 OPTSET(INP_RECVTTL); 1090 break; 1091 1092 case IP_RECVIF: 1093 OPTSET(INP_RECVIF); 1094 break; 1095 1096 case IP_ONESBCAST: 1097 OPTSET(INP_ONESBCAST); 1098 break; 1099 case IP_DONTFRAG: 1100 OPTSET(INP_DONTFRAG); 1101 break; 1102 case IP_BINDANY: 1103 OPTSET(INP_BINDANY); 1104 break; 1105 case IP_RECVTOS: 1106 OPTSET(INP_RECVTOS); 1107 break; 1108 case IP_BINDMULTI: 1109 OPTSET2(INP_BINDMULTI, optval); 1110 break; 1111 case IP_RECVFLOWID: 1112 OPTSET2(INP_RECVFLOWID, optval); 1113 break; 1114#ifdef RSS 1115 case IP_RSS_LISTEN_BUCKET: 1116 if ((optval >= 0) && 1117 (optval < rss_getnumbuckets())) { 1118 inp->inp_rss_listen_bucket = optval; 1119 OPTSET2(INP_RSS_BUCKET_SET, 1); 1120 } else { 1121 error = EINVAL; 1122 } 1123 break; 1124 case IP_RECVRSSBUCKETID: 1125 OPTSET2(INP_RECVRSSBUCKETID, optval); 1126 break; 1127#endif 1128 } 1129 break; 1130#undef OPTSET 1131#undef OPTSET2 1132 1133 /* 1134 * Multicast socket options are processed by the in_mcast 1135 * module. 1136 */ 1137 case IP_MULTICAST_IF: 1138 case IP_MULTICAST_VIF: 1139 case IP_MULTICAST_TTL: 1140 case IP_MULTICAST_LOOP: 1141 case IP_ADD_MEMBERSHIP: 1142 case IP_DROP_MEMBERSHIP: 1143 case IP_ADD_SOURCE_MEMBERSHIP: 1144 case IP_DROP_SOURCE_MEMBERSHIP: 1145 case IP_BLOCK_SOURCE: 1146 case IP_UNBLOCK_SOURCE: 1147 case IP_MSFILTER: 1148 case MCAST_JOIN_GROUP: 1149 case MCAST_LEAVE_GROUP: 1150 case MCAST_JOIN_SOURCE_GROUP: 1151 case MCAST_LEAVE_SOURCE_GROUP: 1152 case MCAST_BLOCK_SOURCE: 1153 case MCAST_UNBLOCK_SOURCE: 1154 error = inp_setmoptions(inp, sopt); 1155 break; 1156 1157 case IP_PORTRANGE: 1158 error = sooptcopyin(sopt, &optval, sizeof optval, 1159 sizeof optval); 1160 if (error) 1161 break; 1162 1163 INP_WLOCK(inp); 1164 switch (optval) { 1165 case IP_PORTRANGE_DEFAULT: 1166 inp->inp_flags &= ~(INP_LOWPORT); 1167 inp->inp_flags &= ~(INP_HIGHPORT); 1168 break; 1169 1170 case IP_PORTRANGE_HIGH: 1171 inp->inp_flags &= ~(INP_LOWPORT); 1172 inp->inp_flags |= INP_HIGHPORT; 1173 break; 1174 1175 case IP_PORTRANGE_LOW: 1176 inp->inp_flags &= ~(INP_HIGHPORT); 1177 inp->inp_flags |= INP_LOWPORT; 1178 break; 1179 1180 default: 1181 error = EINVAL; 1182 break; 1183 } 1184 INP_WUNLOCK(inp); 1185 break; 1186 1187#if defined(IPSEC) || defined(IPSEC_SUPPORT) 1188 case IP_IPSEC_POLICY: 1189 if (IPSEC_ENABLED(ipv4)) { 1190 error = IPSEC_PCBCTL(ipv4, inp, sopt); 1191 break; 1192 } 1193 /* FALLTHROUGH */ 1194#endif /* IPSEC */ 1195 1196 default: 1197 error = ENOPROTOOPT; 1198 break; 1199 } 1200 break; 1201 1202 case SOPT_GET: 1203 switch (sopt->sopt_name) { 1204 case IP_OPTIONS: 1205 case IP_RETOPTS: 1206 if (inp->inp_options) 1207 error = sooptcopyout(sopt, 1208 mtod(inp->inp_options, 1209 char *), 1210 inp->inp_options->m_len); 1211 else 1212 sopt->sopt_valsize = 0; 1213 break; 1214 1215 case IP_TOS: 1216 case IP_TTL: 1217 case IP_MINTTL: 1218 case IP_RECVOPTS: 1219 case IP_RECVRETOPTS: 1220 case IP_RECVDSTADDR: 1221 case IP_RECVTTL: 1222 case IP_RECVIF: 1223 case IP_PORTRANGE: 1224 case IP_ONESBCAST: 1225 case IP_DONTFRAG: 1226 case IP_BINDANY: 1227 case IP_RECVTOS: 1228 case IP_BINDMULTI: 1229 case IP_FLOWID: 1230 case IP_FLOWTYPE: 1231 case IP_RECVFLOWID: 1232#ifdef RSS 1233 case IP_RSSBUCKETID: 1234 case IP_RECVRSSBUCKETID: 1235#endif 1236 switch (sopt->sopt_name) { 1237 1238 case IP_TOS: 1239 optval = inp->inp_ip_tos; 1240 break; 1241 1242 case IP_TTL: 1243 optval = inp->inp_ip_ttl; 1244 break; 1245 1246 case IP_MINTTL: 1247 optval = inp->inp_ip_minttl; 1248 break; 1249 1250#define OPTBIT(bit) (inp->inp_flags & bit ? 1 : 0) 1251#define OPTBIT2(bit) (inp->inp_flags2 & bit ? 1 : 0) 1252 1253 case IP_RECVOPTS: 1254 optval = OPTBIT(INP_RECVOPTS); 1255 break; 1256 1257 case IP_RECVRETOPTS: 1258 optval = OPTBIT(INP_RECVRETOPTS); 1259 break; 1260 1261 case IP_RECVDSTADDR: 1262 optval = OPTBIT(INP_RECVDSTADDR); 1263 break; 1264 1265 case IP_RECVTTL: 1266 optval = OPTBIT(INP_RECVTTL); 1267 break; 1268 1269 case IP_RECVIF: 1270 optval = OPTBIT(INP_RECVIF); 1271 break; 1272 1273 case IP_PORTRANGE: 1274 if (inp->inp_flags & INP_HIGHPORT) 1275 optval = IP_PORTRANGE_HIGH; 1276 else if (inp->inp_flags & INP_LOWPORT) 1277 optval = IP_PORTRANGE_LOW; 1278 else 1279 optval = 0; 1280 break; 1281 1282 case IP_ONESBCAST: 1283 optval = OPTBIT(INP_ONESBCAST); 1284 break; 1285 case IP_DONTFRAG: 1286 optval = OPTBIT(INP_DONTFRAG); 1287 break; 1288 case IP_BINDANY: 1289 optval = OPTBIT(INP_BINDANY); 1290 break; 1291 case IP_RECVTOS: 1292 optval = OPTBIT(INP_RECVTOS); 1293 break; 1294 case IP_FLOWID: 1295 optval = inp->inp_flowid; 1296 break; 1297 case IP_FLOWTYPE: 1298 optval = inp->inp_flowtype; 1299 break; 1300 case IP_RECVFLOWID: 1301 optval = OPTBIT2(INP_RECVFLOWID); 1302 break; 1303#ifdef RSS 1304 case IP_RSSBUCKETID: 1305 retval = rss_hash2bucket(inp->inp_flowid, 1306 inp->inp_flowtype, 1307 &rss_bucket); 1308 if (retval == 0) 1309 optval = rss_bucket; 1310 else 1311 error = EINVAL; 1312 break; 1313 case IP_RECVRSSBUCKETID: 1314 optval = OPTBIT2(INP_RECVRSSBUCKETID); 1315 break; 1316#endif 1317 case IP_BINDMULTI: 1318 optval = OPTBIT2(INP_BINDMULTI); 1319 break; 1320 } 1321 error = sooptcopyout(sopt, &optval, sizeof optval); 1322 break; 1323 1324 /* 1325 * Multicast socket options are processed by the in_mcast 1326 * module. 1327 */ 1328 case IP_MULTICAST_IF: 1329 case IP_MULTICAST_VIF: 1330 case IP_MULTICAST_TTL: 1331 case IP_MULTICAST_LOOP: 1332 case IP_MSFILTER: 1333 error = inp_getmoptions(inp, sopt); 1334 break; 1335 1336#if defined(IPSEC) || defined(IPSEC_SUPPORT) 1337 case IP_IPSEC_POLICY: 1338 if (IPSEC_ENABLED(ipv4)) { 1339 error = IPSEC_PCBCTL(ipv4, inp, sopt); 1340 break; 1341 } 1342 /* FALLTHROUGH */ 1343#endif /* IPSEC */ 1344 1345 default: 1346 error = ENOPROTOOPT; 1347 break; 1348 } 1349 break; 1350 } 1351 return (error); 1352} 1353 1354/* 1355 * Routine called from ip_output() to loop back a copy of an IP multicast 1356 * packet to the input queue of a specified interface. Note that this 1357 * calls the output routine of the loopback "driver", but with an interface 1358 * pointer that might NOT be a loopback interface -- evil, but easier than 1359 * replicating that code here. 1360 */ 1361static void 1362ip_mloopback(struct ifnet *ifp, const struct mbuf *m, int hlen) 1363{ 1364 struct ip *ip; 1365 struct mbuf *copym; 1366 1367 /* 1368 * Make a deep copy of the packet because we're going to 1369 * modify the pack in order to generate checksums. 1370 */ 1371 copym = m_dup(m, M_NOWAIT); 1372 if (copym != NULL && (!M_WRITABLE(copym) || copym->m_len < hlen)) 1373 copym = m_pullup(copym, hlen); 1374 if (copym != NULL) { 1375 /* If needed, compute the checksum and mark it as valid. */ 1376 if (copym->m_pkthdr.csum_flags & CSUM_DELAY_DATA) { 1377 in_delayed_cksum(copym); 1378 copym->m_pkthdr.csum_flags &= ~CSUM_DELAY_DATA; 1379 copym->m_pkthdr.csum_flags |= 1380 CSUM_DATA_VALID | CSUM_PSEUDO_HDR; 1381 copym->m_pkthdr.csum_data = 0xffff; 1382 } 1383 /* 1384 * We don't bother to fragment if the IP length is greater 1385 * than the interface's MTU. Can this possibly matter? 1386 */ 1387 ip = mtod(copym, struct ip *); 1388 ip->ip_sum = 0; 1389 ip->ip_sum = in_cksum(copym, hlen); 1390 if_simloop(ifp, copym, AF_INET, 0); 1391 } 1392} 1393