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 * 3. 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("$FreeBSD$"); 34 35 #include "opt_inet.h" 36 #include "opt_ratelimit.h" 37 #include "opt_ipsec.h" 38 #include "opt_mbuf_stress_test.h" 39 #include "opt_mpath.h" 40 #include "opt_route.h" 41 #include "opt_sctp.h" 42 #include "opt_rss.h" 43 44 #include <sys/param.h> 45 #include <sys/systm.h> 46 #include <sys/kernel.h> 47 #include <sys/lock.h> 48 #include <sys/malloc.h> 49 #include <sys/mbuf.h> 50 #include <sys/priv.h> 51 #include <sys/proc.h> 52 #include <sys/protosw.h> 53 #include <sys/rmlock.h> 54 #include <sys/sdt.h> 55 #include <sys/socket.h> 56 #include <sys/socketvar.h> 57 #include <sys/sysctl.h> 58 #include <sys/ucred.h> 59 60 #include <net/if.h> 61 #include <net/if_var.h> 62 #include <net/if_llatbl.h> 63 #include <net/netisr.h> 64 #include <net/pfil.h> 65 #include <net/route.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 93 static int mbuf_frag_size = 0; 94 SYSCTL_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 98 static void ip_mloopback(struct ifnet *, const struct mbuf *, int); 99 100 101 extern int in_mcast_loop; 102 extern struct protosw inetsw[]; 103 104 static inline int 105 ip_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 */ 207 int 208 ip_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 if (opt) { 247 int len = 0; 248 m = ip_insertoptions(m, opt, &len); 249 if (len != 0) 250 hlen = len; /* ip->ip_hl is updated above */ 251 } 252 ip = mtod(m, struct ip *); 253 ip_len = ntohs(ip->ip_len); 254 ip_off = ntohs(ip->ip_off); 255 256 if ((flags & (IP_FORWARDING|IP_RAWOUTPUT)) == 0) { 257 ip->ip_v = IPVERSION; 258 ip->ip_hl = hlen >> 2; 259 ip_fillid(ip); 260 IPSTAT_INC(ips_localout); 261 } else { 262 /* Header already set, fetch hlen from there */ 263 hlen = ip->ip_hl << 2; 264 } 265 266 /* 267 * dst/gw handling: 268 * 269 * dst can be rewritten but always points to &ro->ro_dst. 270 * gw is readonly but can point either to dst OR rt_gateway, 271 * therefore we need restore gw if we're redoing lookup. 272 */ 273 gw = dst = (struct sockaddr_in *)&ro->ro_dst; 274 fibnum = (inp != NULL) ? inp->inp_inc.inc_fibnum : M_GETFIB(m); 275 rte = ro->ro_rt; 276 if (rte == NULL) { 277 bzero(dst, sizeof(*dst)); 278 dst->sin_family = AF_INET; 279 dst->sin_len = sizeof(*dst); 280 dst->sin_addr = ip->ip_dst; 281 } 282 again: 283 /* 284 * Validate route against routing table additions; 285 * a better/more specific route might have been added. 286 */ 287 if (inp) 288 RT_VALIDATE(ro, &inp->inp_rt_cookie, fibnum); 289 /* 290 * If there is a cached route, 291 * check that it is to the same destination 292 * and is still up. If not, free it and try again. 293 * The address family should also be checked in case of sharing the 294 * cache with IPv6. 295 * Also check whether routing cache needs invalidation. 296 */ 297 rte = ro->ro_rt; 298 if (rte && ((rte->rt_flags & RTF_UP) == 0 || 299 rte->rt_ifp == NULL || 300 !RT_LINK_IS_UP(rte->rt_ifp) || 301 dst->sin_family != AF_INET || 302 dst->sin_addr.s_addr != ip->ip_dst.s_addr)) { 303 RTFREE(rte); 304 rte = ro->ro_rt = (struct rtentry *)NULL; 305 if (ro->ro_lle) 306 LLE_FREE(ro->ro_lle); /* zeros ro_lle */ 307 ro->ro_lle = (struct llentry *)NULL; 308 } 309 ia = NULL; 310 have_ia_ref = 0; 311 /* 312 * If routing to interface only, short circuit routing lookup. 313 * The use of an all-ones broadcast address implies this; an 314 * interface is specified by the broadcast address of an interface, 315 * or the destination address of a ptp interface. 316 */ 317 if (flags & IP_SENDONES) { 318 if ((ia = ifatoia(ifa_ifwithbroadaddr(sintosa(dst), 319 M_GETFIB(m)))) == NULL && 320 (ia = ifatoia(ifa_ifwithdstaddr(sintosa(dst), 321 M_GETFIB(m)))) == NULL) { 322 IPSTAT_INC(ips_noroute); 323 error = ENETUNREACH; 324 goto bad; 325 } 326 have_ia_ref = 1; 327 ip->ip_dst.s_addr = INADDR_BROADCAST; 328 dst->sin_addr = ip->ip_dst; 329 ifp = ia->ia_ifp; 330 ip->ip_ttl = 1; 331 isbroadcast = 1; 332 } else if (flags & IP_ROUTETOIF) { 333 if ((ia = ifatoia(ifa_ifwithdstaddr(sintosa(dst), 334 M_GETFIB(m)))) == NULL && 335 (ia = ifatoia(ifa_ifwithnet(sintosa(dst), 0, 336 M_GETFIB(m)))) == NULL) { 337 IPSTAT_INC(ips_noroute); 338 error = ENETUNREACH; 339 goto bad; 340 } 341 have_ia_ref = 1; 342 ifp = ia->ia_ifp; 343 ip->ip_ttl = 1; 344 isbroadcast = ifp->if_flags & IFF_BROADCAST ? 345 in_ifaddr_broadcast(dst->sin_addr, ia) : 0; 346 } else if (IN_MULTICAST(ntohl(ip->ip_dst.s_addr)) && 347 imo != NULL && imo->imo_multicast_ifp != NULL) { 348 /* 349 * Bypass the normal routing lookup for multicast 350 * packets if the interface is specified. 351 */ 352 ifp = imo->imo_multicast_ifp; 353 IFP_TO_IA(ifp, ia, &in_ifa_tracker); 354 if (ia) 355 have_ia_ref = 1; 356 isbroadcast = 0; /* fool gcc */ 357 } else { 358 /* 359 * We want to do any cloning requested by the link layer, 360 * as this is probably required in all cases for correct 361 * operation (as it is for ARP). 362 */ 363 if (rte == NULL) { 364 #ifdef RADIX_MPATH 365 rtalloc_mpath_fib(ro, 366 ntohl(ip->ip_src.s_addr ^ ip->ip_dst.s_addr), 367 fibnum); 368 #else 369 in_rtalloc_ign(ro, 0, fibnum); 370 #endif 371 rte = ro->ro_rt; 372 } 373 if (rte == NULL || 374 (rte->rt_flags & RTF_UP) == 0 || 375 rte->rt_ifp == NULL || 376 !RT_LINK_IS_UP(rte->rt_ifp)) { 377 #if defined(IPSEC) || defined(IPSEC_SUPPORT) 378 /* 379 * There is no route for this packet, but it is 380 * possible that a matching SPD entry exists. 381 */ 382 no_route_but_check_spd = 1; 383 mtu = 0; /* Silence GCC warning. */ 384 goto sendit; 385 #endif 386 IPSTAT_INC(ips_noroute); 387 error = EHOSTUNREACH; 388 goto bad; 389 } 390 ia = ifatoia(rte->rt_ifa); 391 ifp = rte->rt_ifp; 392 counter_u64_add(rte->rt_pksent, 1); 393 rt_update_ro_flags(ro); 394 if (rte->rt_flags & RTF_GATEWAY) 395 gw = (struct sockaddr_in *)rte->rt_gateway; 396 if (rte->rt_flags & RTF_HOST) 397 isbroadcast = (rte->rt_flags & RTF_BROADCAST); 398 else if (ifp->if_flags & IFF_BROADCAST) 399 isbroadcast = in_ifaddr_broadcast(gw->sin_addr, ia); 400 else 401 isbroadcast = 0; 402 } 403 404 /* 405 * Calculate MTU. If we have a route that is up, use that, 406 * otherwise use the interface's MTU. 407 */ 408 if (rte != NULL && (rte->rt_flags & (RTF_UP|RTF_HOST))) 409 mtu = rte->rt_mtu; 410 else 411 mtu = ifp->if_mtu; 412 /* Catch a possible divide by zero later. */ 413 KASSERT(mtu > 0, ("%s: mtu %d <= 0, rte=%p (rt_flags=0x%08x) ifp=%p", 414 __func__, mtu, rte, (rte != NULL) ? rte->rt_flags : 0, ifp)); 415 416 if (IN_MULTICAST(ntohl(ip->ip_dst.s_addr))) { 417 m->m_flags |= M_MCAST; 418 /* 419 * IP destination address is multicast. Make sure "gw" 420 * still points to the address in "ro". (It may have been 421 * changed to point to a gateway address, above.) 422 */ 423 gw = dst; 424 /* 425 * See if the caller provided any multicast options 426 */ 427 if (imo != NULL) { 428 ip->ip_ttl = imo->imo_multicast_ttl; 429 if (imo->imo_multicast_vif != -1) 430 ip->ip_src.s_addr = 431 ip_mcast_src ? 432 ip_mcast_src(imo->imo_multicast_vif) : 433 INADDR_ANY; 434 } else 435 ip->ip_ttl = IP_DEFAULT_MULTICAST_TTL; 436 /* 437 * Confirm that the outgoing interface supports multicast. 438 */ 439 if ((imo == NULL) || (imo->imo_multicast_vif == -1)) { 440 if ((ifp->if_flags & IFF_MULTICAST) == 0) { 441 IPSTAT_INC(ips_noroute); 442 error = ENETUNREACH; 443 goto bad; 444 } 445 } 446 /* 447 * If source address not specified yet, use address 448 * of outgoing interface. 449 */ 450 if (ip->ip_src.s_addr == INADDR_ANY) { 451 /* Interface may have no addresses. */ 452 if (ia != NULL) 453 ip->ip_src = IA_SIN(ia)->sin_addr; 454 } 455 456 if ((imo == NULL && in_mcast_loop) || 457 (imo && imo->imo_multicast_loop)) { 458 /* 459 * Loop back multicast datagram if not expressly 460 * forbidden to do so, even if we are not a member 461 * of the group; ip_input() will filter it later, 462 * thus deferring a hash lookup and mutex acquisition 463 * at the expense of a cheap copy using m_copym(). 464 */ 465 ip_mloopback(ifp, m, hlen); 466 } else { 467 /* 468 * If we are acting as a multicast router, perform 469 * multicast forwarding as if the packet had just 470 * arrived on the interface to which we are about 471 * to send. The multicast forwarding function 472 * recursively calls this function, using the 473 * IP_FORWARDING flag to prevent infinite recursion. 474 * 475 * Multicasts that are looped back by ip_mloopback(), 476 * above, will be forwarded by the ip_input() routine, 477 * if necessary. 478 */ 479 if (V_ip_mrouter && (flags & IP_FORWARDING) == 0) { 480 /* 481 * If rsvp daemon is not running, do not 482 * set ip_moptions. This ensures that the packet 483 * is multicast and not just sent down one link 484 * as prescribed by rsvpd. 485 */ 486 if (!V_rsvp_on) 487 imo = NULL; 488 if (ip_mforward && 489 ip_mforward(ip, ifp, m, imo) != 0) { 490 m_freem(m); 491 goto done; 492 } 493 } 494 } 495 496 /* 497 * Multicasts with a time-to-live of zero may be looped- 498 * back, above, but must not be transmitted on a network. 499 * Also, multicasts addressed to the loopback interface 500 * are not sent -- the above call to ip_mloopback() will 501 * loop back a copy. ip_input() will drop the copy if 502 * this host does not belong to the destination group on 503 * the loopback interface. 504 */ 505 if (ip->ip_ttl == 0 || ifp->if_flags & IFF_LOOPBACK) { 506 m_freem(m); 507 goto done; 508 } 509 510 goto sendit; 511 } 512 513 /* 514 * If the source address is not specified yet, use the address 515 * of the outoing interface. 516 */ 517 if (ip->ip_src.s_addr == INADDR_ANY) { 518 /* Interface may have no addresses. */ 519 if (ia != NULL) { 520 ip->ip_src = IA_SIN(ia)->sin_addr; 521 } 522 } 523 524 /* 525 * Look for broadcast address and 526 * verify user is allowed to send 527 * such a packet. 528 */ 529 if (isbroadcast) { 530 if ((ifp->if_flags & IFF_BROADCAST) == 0) { 531 error = EADDRNOTAVAIL; 532 goto bad; 533 } 534 if ((flags & IP_ALLOWBROADCAST) == 0) { 535 error = EACCES; 536 goto bad; 537 } 538 /* don't allow broadcast messages to be fragmented */ 539 if (ip_len > mtu) { 540 error = EMSGSIZE; 541 goto bad; 542 } 543 m->m_flags |= M_BCAST; 544 } else { 545 m->m_flags &= ~M_BCAST; 546 } 547 548 sendit: 549 #if defined(IPSEC) || defined(IPSEC_SUPPORT) 550 if (IPSEC_ENABLED(ipv4)) { 551 if ((error = IPSEC_OUTPUT(ipv4, m, inp)) != 0) { 552 if (error == EINPROGRESS) 553 error = 0; 554 goto done; 555 } 556 } 557 /* 558 * Check if there was a route for this packet; return error if not. 559 */ 560 if (no_route_but_check_spd) { 561 IPSTAT_INC(ips_noroute); 562 error = EHOSTUNREACH; 563 goto bad; 564 } 565 /* Update variables that are affected by ipsec4_output(). */ 566 ip = mtod(m, struct ip *); 567 hlen = ip->ip_hl << 2; 568 #endif /* IPSEC */ 569 570 /* Jump over all PFIL processing if hooks are not active. */ 571 if (PFIL_HOOKED(&V_inet_pfil_hook)) { 572 switch (ip_output_pfil(&m, ifp, inp, dst, &fibnum, &error)) { 573 case 1: /* Finished */ 574 goto done; 575 576 case 0: /* Continue normally */ 577 ip = mtod(m, struct ip *); 578 break; 579 580 case -1: /* Need to try again */ 581 /* Reset everything for a new round */ 582 RO_RTFREE(ro); 583 if (have_ia_ref) 584 ifa_free(&ia->ia_ifa); 585 ro->ro_prepend = NULL; 586 rte = NULL; 587 gw = dst; 588 ip = mtod(m, struct ip *); 589 goto again; 590 591 } 592 } 593 594 /* 127/8 must not appear on wire - RFC1122. */ 595 if ((ntohl(ip->ip_dst.s_addr) >> IN_CLASSA_NSHIFT) == IN_LOOPBACKNET || 596 (ntohl(ip->ip_src.s_addr) >> IN_CLASSA_NSHIFT) == IN_LOOPBACKNET) { 597 if ((ifp->if_flags & IFF_LOOPBACK) == 0) { 598 IPSTAT_INC(ips_badaddr); 599 error = EADDRNOTAVAIL; 600 goto bad; 601 } 602 } 603 604 m->m_pkthdr.csum_flags |= CSUM_IP; 605 if (m->m_pkthdr.csum_flags & CSUM_DELAY_DATA & ~ifp->if_hwassist) { 606 in_delayed_cksum(m); 607 m->m_pkthdr.csum_flags &= ~CSUM_DELAY_DATA; 608 } 609 #ifdef SCTP 610 if (m->m_pkthdr.csum_flags & CSUM_SCTP & ~ifp->if_hwassist) { 611 sctp_delayed_cksum(m, (uint32_t)(ip->ip_hl << 2)); 612 m->m_pkthdr.csum_flags &= ~CSUM_SCTP; 613 } 614 #endif 615 616 /* 617 * If small enough for interface, or the interface will take 618 * care of the fragmentation for us, we can just send directly. 619 */ 620 if (ip_len <= mtu || 621 (m->m_pkthdr.csum_flags & ifp->if_hwassist & CSUM_TSO) != 0) { 622 ip->ip_sum = 0; 623 if (m->m_pkthdr.csum_flags & CSUM_IP & ~ifp->if_hwassist) { 624 ip->ip_sum = in_cksum(m, hlen); 625 m->m_pkthdr.csum_flags &= ~CSUM_IP; 626 } 627 628 /* 629 * Record statistics for this interface address. 630 * With CSUM_TSO the byte/packet count will be slightly 631 * incorrect because we count the IP+TCP headers only 632 * once instead of for every generated packet. 633 */ 634 if (!(flags & IP_FORWARDING) && ia) { 635 if (m->m_pkthdr.csum_flags & CSUM_TSO) 636 counter_u64_add(ia->ia_ifa.ifa_opackets, 637 m->m_pkthdr.len / m->m_pkthdr.tso_segsz); 638 else 639 counter_u64_add(ia->ia_ifa.ifa_opackets, 1); 640 641 counter_u64_add(ia->ia_ifa.ifa_obytes, m->m_pkthdr.len); 642 } 643 #ifdef MBUF_STRESS_TEST 644 if (mbuf_frag_size && m->m_pkthdr.len > mbuf_frag_size) 645 m = m_fragment(m, M_NOWAIT, mbuf_frag_size); 646 #endif 647 /* 648 * Reset layer specific mbuf flags 649 * to avoid confusing lower layers. 650 */ 651 m_clrprotoflags(m); 652 IP_PROBE(send, NULL, NULL, ip, ifp, ip, NULL); 653 #ifdef RATELIMIT 654 if (inp != NULL) { 655 if (inp->inp_flags2 & INP_RATE_LIMIT_CHANGED) 656 in_pcboutput_txrtlmt(inp, ifp, m); 657 /* stamp send tag on mbuf */ 658 m->m_pkthdr.snd_tag = inp->inp_snd_tag; 659 } else { 660 m->m_pkthdr.snd_tag = NULL; 661 } 662 #endif 663 error = (*ifp->if_output)(ifp, m, 664 (const struct sockaddr *)gw, ro); 665 #ifdef RATELIMIT 666 /* check for route change */ 667 if (error == EAGAIN) 668 in_pcboutput_eagain(inp); 669 #endif 670 goto done; 671 } 672 673 /* Balk when DF bit is set or the interface didn't support TSO. */ 674 if ((ip_off & IP_DF) || (m->m_pkthdr.csum_flags & CSUM_TSO)) { 675 error = EMSGSIZE; 676 IPSTAT_INC(ips_cantfrag); 677 goto bad; 678 } 679 680 /* 681 * Too large for interface; fragment if possible. If successful, 682 * on return, m will point to a list of packets to be sent. 683 */ 684 error = ip_fragment(ip, &m, mtu, ifp->if_hwassist); 685 if (error) 686 goto bad; 687 for (; m; m = m0) { 688 m0 = m->m_nextpkt; 689 m->m_nextpkt = 0; 690 if (error == 0) { 691 /* Record statistics for this interface address. */ 692 if (ia != NULL) { 693 counter_u64_add(ia->ia_ifa.ifa_opackets, 1); 694 counter_u64_add(ia->ia_ifa.ifa_obytes, 695 m->m_pkthdr.len); 696 } 697 /* 698 * Reset layer specific mbuf flags 699 * to avoid confusing upper layers. 700 */ 701 m_clrprotoflags(m); 702 703 IP_PROBE(send, NULL, NULL, mtod(m, struct ip *), ifp, 704 mtod(m, struct ip *), NULL); 705 #ifdef RATELIMIT 706 if (inp != NULL) { 707 if (inp->inp_flags2 & INP_RATE_LIMIT_CHANGED) 708 in_pcboutput_txrtlmt(inp, ifp, m); 709 /* stamp send tag on mbuf */ 710 m->m_pkthdr.snd_tag = inp->inp_snd_tag; 711 } else { 712 m->m_pkthdr.snd_tag = NULL; 713 } 714 #endif 715 error = (*ifp->if_output)(ifp, m, 716 (const struct sockaddr *)gw, ro); 717 #ifdef RATELIMIT 718 /* check for route change */ 719 if (error == EAGAIN) 720 in_pcboutput_eagain(inp); 721 #endif 722 } else 723 m_freem(m); 724 } 725 726 if (error == 0) 727 IPSTAT_INC(ips_fragmented); 728 729 done: 730 if (ro == &iproute) 731 RO_RTFREE(ro); 732 else if (rte == NULL) 733 /* 734 * If the caller supplied a route but somehow the reference 735 * to it has been released need to prevent the caller 736 * calling RTFREE on it again. 737 */ 738 ro->ro_rt = NULL; 739 if (have_ia_ref) 740 ifa_free(&ia->ia_ifa); 741 return (error); 742 bad: 743 m_freem(m); 744 goto done; 745 } 746 747 /* 748 * Create a chain of fragments which fit the given mtu. m_frag points to the 749 * mbuf to be fragmented; on return it points to the chain with the fragments. 750 * Return 0 if no error. If error, m_frag may contain a partially built 751 * chain of fragments that should be freed by the caller. 752 * 753 * if_hwassist_flags is the hw offload capabilities (see if_data.ifi_hwassist) 754 */ 755 int 756 ip_fragment(struct ip *ip, struct mbuf **m_frag, int mtu, 757 u_long if_hwassist_flags) 758 { 759 int error = 0; 760 int hlen = ip->ip_hl << 2; 761 int len = (mtu - hlen) & ~7; /* size of payload in each fragment */ 762 int off; 763 struct mbuf *m0 = *m_frag; /* the original packet */ 764 int firstlen; 765 struct mbuf **mnext; 766 int nfrags; 767 uint16_t ip_len, ip_off; 768 769 ip_len = ntohs(ip->ip_len); 770 ip_off = ntohs(ip->ip_off); 771 772 if (ip_off & IP_DF) { /* Fragmentation not allowed */ 773 IPSTAT_INC(ips_cantfrag); 774 return EMSGSIZE; 775 } 776 777 /* 778 * Must be able to put at least 8 bytes per fragment. 779 */ 780 if (len < 8) 781 return EMSGSIZE; 782 783 /* 784 * If the interface will not calculate checksums on 785 * fragmented packets, then do it here. 786 */ 787 if (m0->m_pkthdr.csum_flags & CSUM_DELAY_DATA) { 788 in_delayed_cksum(m0); 789 m0->m_pkthdr.csum_flags &= ~CSUM_DELAY_DATA; 790 } 791 #ifdef SCTP 792 if (m0->m_pkthdr.csum_flags & CSUM_SCTP) { 793 sctp_delayed_cksum(m0, hlen); 794 m0->m_pkthdr.csum_flags &= ~CSUM_SCTP; 795 } 796 #endif 797 if (len > PAGE_SIZE) { 798 /* 799 * Fragment large datagrams such that each segment 800 * contains a multiple of PAGE_SIZE amount of data, 801 * plus headers. This enables a receiver to perform 802 * page-flipping zero-copy optimizations. 803 * 804 * XXX When does this help given that sender and receiver 805 * could have different page sizes, and also mtu could 806 * be less than the receiver's page size ? 807 */ 808 int newlen; 809 810 off = MIN(mtu, m0->m_pkthdr.len); 811 812 /* 813 * firstlen (off - hlen) must be aligned on an 814 * 8-byte boundary 815 */ 816 if (off < hlen) 817 goto smart_frag_failure; 818 off = ((off - hlen) & ~7) + hlen; 819 newlen = (~PAGE_MASK) & mtu; 820 if ((newlen + sizeof (struct ip)) > mtu) { 821 /* we failed, go back the default */ 822 smart_frag_failure: 823 newlen = len; 824 off = hlen + len; 825 } 826 len = newlen; 827 828 } else { 829 off = hlen + len; 830 } 831 832 firstlen = off - hlen; 833 mnext = &m0->m_nextpkt; /* pointer to next packet */ 834 835 /* 836 * Loop through length of segment after first fragment, 837 * make new header and copy data of each part and link onto chain. 838 * Here, m0 is the original packet, m is the fragment being created. 839 * The fragments are linked off the m_nextpkt of the original 840 * packet, which after processing serves as the first fragment. 841 */ 842 for (nfrags = 1; off < ip_len; off += len, nfrags++) { 843 struct ip *mhip; /* ip header on the fragment */ 844 struct mbuf *m; 845 int mhlen = sizeof (struct ip); 846 847 m = m_gethdr(M_NOWAIT, MT_DATA); 848 if (m == NULL) { 849 error = ENOBUFS; 850 IPSTAT_INC(ips_odropped); 851 goto done; 852 } 853 /* 854 * Make sure the complete packet header gets copied 855 * from the originating mbuf to the newly created 856 * mbuf. This also ensures that existing firewall 857 * classification(s), VLAN tags and so on get copied 858 * to the resulting fragmented packet(s): 859 */ 860 if (m_dup_pkthdr(m, m0, M_NOWAIT) == 0) { 861 m_free(m); 862 error = ENOBUFS; 863 IPSTAT_INC(ips_odropped); 864 goto done; 865 } 866 /* 867 * In the first mbuf, leave room for the link header, then 868 * copy the original IP header including options. The payload 869 * goes into an additional mbuf chain returned by m_copym(). 870 */ 871 m->m_data += max_linkhdr; 872 mhip = mtod(m, struct ip *); 873 *mhip = *ip; 874 if (hlen > sizeof (struct ip)) { 875 mhlen = ip_optcopy(ip, mhip) + sizeof (struct ip); 876 mhip->ip_v = IPVERSION; 877 mhip->ip_hl = mhlen >> 2; 878 } 879 m->m_len = mhlen; 880 /* XXX do we need to add ip_off below ? */ 881 mhip->ip_off = ((off - hlen) >> 3) + ip_off; 882 if (off + len >= ip_len) 883 len = ip_len - off; 884 else 885 mhip->ip_off |= IP_MF; 886 mhip->ip_len = htons((u_short)(len + mhlen)); 887 m->m_next = m_copym(m0, off, len, M_NOWAIT); 888 if (m->m_next == NULL) { /* copy failed */ 889 m_free(m); 890 error = ENOBUFS; /* ??? */ 891 IPSTAT_INC(ips_odropped); 892 goto done; 893 } 894 m->m_pkthdr.len = mhlen + len; 895 #ifdef MAC 896 mac_netinet_fragment(m0, m); 897 #endif 898 mhip->ip_off = htons(mhip->ip_off); 899 mhip->ip_sum = 0; 900 if (m->m_pkthdr.csum_flags & CSUM_IP & ~if_hwassist_flags) { 901 mhip->ip_sum = in_cksum(m, mhlen); 902 m->m_pkthdr.csum_flags &= ~CSUM_IP; 903 } 904 *mnext = m; 905 mnext = &m->m_nextpkt; 906 } 907 IPSTAT_ADD(ips_ofragments, nfrags); 908 909 /* 910 * Update first fragment by trimming what's been copied out 911 * and updating header. 912 */ 913 m_adj(m0, hlen + firstlen - ip_len); 914 m0->m_pkthdr.len = hlen + firstlen; 915 ip->ip_len = htons((u_short)m0->m_pkthdr.len); 916 ip->ip_off = htons(ip_off | IP_MF); 917 ip->ip_sum = 0; 918 if (m0->m_pkthdr.csum_flags & CSUM_IP & ~if_hwassist_flags) { 919 ip->ip_sum = in_cksum(m0, hlen); 920 m0->m_pkthdr.csum_flags &= ~CSUM_IP; 921 } 922 923 done: 924 *m_frag = m0; 925 return error; 926 } 927 928 void 929 in_delayed_cksum(struct mbuf *m) 930 { 931 struct ip *ip; 932 uint16_t csum, offset, ip_len; 933 934 ip = mtod(m, struct ip *); 935 offset = ip->ip_hl << 2 ; 936 ip_len = ntohs(ip->ip_len); 937 csum = in_cksum_skip(m, ip_len, offset); 938 if (m->m_pkthdr.csum_flags & CSUM_UDP && csum == 0) 939 csum = 0xffff; 940 offset += m->m_pkthdr.csum_data; /* checksum offset */ 941 942 /* find the mbuf in the chain where the checksum starts*/ 943 while ((m != NULL) && (offset >= m->m_len)) { 944 offset -= m->m_len; 945 m = m->m_next; 946 } 947 KASSERT(m != NULL, ("in_delayed_cksum: checksum outside mbuf chain.")); 948 KASSERT(offset + sizeof(u_short) <= m->m_len, ("in_delayed_cksum: checksum split between mbufs.")); 949 *(u_short *)(m->m_data + offset) = csum; 950 } 951 952 /* 953 * IP socket option processing. 954 */ 955 int 956 ip_ctloutput(struct socket *so, struct sockopt *sopt) 957 { 958 struct inpcb *inp = sotoinpcb(so); 959 int error, optval; 960 #ifdef RSS 961 uint32_t rss_bucket; 962 int retval; 963 #endif 964 965 error = optval = 0; 966 if (sopt->sopt_level != IPPROTO_IP) { 967 error = EINVAL; 968 969 if (sopt->sopt_level == SOL_SOCKET && 970 sopt->sopt_dir == SOPT_SET) { 971 switch (sopt->sopt_name) { 972 case SO_REUSEADDR: 973 INP_WLOCK(inp); 974 if ((so->so_options & SO_REUSEADDR) != 0) 975 inp->inp_flags2 |= INP_REUSEADDR; 976 else 977 inp->inp_flags2 &= ~INP_REUSEADDR; 978 INP_WUNLOCK(inp); 979 error = 0; 980 break; 981 case SO_REUSEPORT: 982 INP_WLOCK(inp); 983 if ((so->so_options & SO_REUSEPORT) != 0) 984 inp->inp_flags2 |= INP_REUSEPORT; 985 else 986 inp->inp_flags2 &= ~INP_REUSEPORT; 987 INP_WUNLOCK(inp); 988 error = 0; 989 break; 990 case SO_SETFIB: 991 INP_WLOCK(inp); 992 inp->inp_inc.inc_fibnum = so->so_fibnum; 993 INP_WUNLOCK(inp); 994 error = 0; 995 break; 996 case SO_MAX_PACING_RATE: 997 #ifdef RATELIMIT 998 INP_WLOCK(inp); 999 inp->inp_flags2 |= INP_RATE_LIMIT_CHANGED; 1000 INP_WUNLOCK(inp); 1001 error = 0; 1002 #else 1003 error = EOPNOTSUPP; 1004 #endif 1005 break; 1006 default: 1007 break; 1008 } 1009 } 1010 return (error); 1011 } 1012 1013 switch (sopt->sopt_dir) { 1014 case SOPT_SET: 1015 switch (sopt->sopt_name) { 1016 case IP_OPTIONS: 1017 #ifdef notyet 1018 case IP_RETOPTS: 1019 #endif 1020 { 1021 struct mbuf *m; 1022 if (sopt->sopt_valsize > MLEN) { 1023 error = EMSGSIZE; 1024 break; 1025 } 1026 m = m_get(sopt->sopt_td ? M_WAITOK : M_NOWAIT, MT_DATA); 1027 if (m == NULL) { 1028 error = ENOBUFS; 1029 break; 1030 } 1031 m->m_len = sopt->sopt_valsize; 1032 error = sooptcopyin(sopt, mtod(m, char *), m->m_len, 1033 m->m_len); 1034 if (error) { 1035 m_free(m); 1036 break; 1037 } 1038 INP_WLOCK(inp); 1039 error = ip_pcbopts(inp, sopt->sopt_name, m); 1040 INP_WUNLOCK(inp); 1041 return (error); 1042 } 1043 1044 case IP_BINDANY: 1045 if (sopt->sopt_td != NULL) { 1046 error = priv_check(sopt->sopt_td, 1047 PRIV_NETINET_BINDANY); 1048 if (error) 1049 break; 1050 } 1051 /* FALLTHROUGH */ 1052 case IP_BINDMULTI: 1053 #ifdef RSS 1054 case IP_RSS_LISTEN_BUCKET: 1055 #endif 1056 case IP_TOS: 1057 case IP_TTL: 1058 case IP_MINTTL: 1059 case IP_RECVOPTS: 1060 case IP_RECVRETOPTS: 1061 case IP_ORIGDSTADDR: 1062 case IP_RECVDSTADDR: 1063 case IP_RECVTTL: 1064 case IP_RECVIF: 1065 case IP_ONESBCAST: 1066 case IP_DONTFRAG: 1067 case IP_RECVTOS: 1068 case IP_RECVFLOWID: 1069 #ifdef RSS 1070 case IP_RECVRSSBUCKETID: 1071 #endif 1072 error = sooptcopyin(sopt, &optval, sizeof optval, 1073 sizeof optval); 1074 if (error) 1075 break; 1076 1077 switch (sopt->sopt_name) { 1078 case IP_TOS: 1079 inp->inp_ip_tos = optval; 1080 break; 1081 1082 case IP_TTL: 1083 inp->inp_ip_ttl = optval; 1084 break; 1085 1086 case IP_MINTTL: 1087 if (optval >= 0 && optval <= MAXTTL) 1088 inp->inp_ip_minttl = optval; 1089 else 1090 error = EINVAL; 1091 break; 1092 1093 #define OPTSET(bit) do { \ 1094 INP_WLOCK(inp); \ 1095 if (optval) \ 1096 inp->inp_flags |= bit; \ 1097 else \ 1098 inp->inp_flags &= ~bit; \ 1099 INP_WUNLOCK(inp); \ 1100 } while (0) 1101 1102 #define OPTSET2(bit, val) do { \ 1103 INP_WLOCK(inp); \ 1104 if (val) \ 1105 inp->inp_flags2 |= bit; \ 1106 else \ 1107 inp->inp_flags2 &= ~bit; \ 1108 INP_WUNLOCK(inp); \ 1109 } while (0) 1110 1111 case IP_RECVOPTS: 1112 OPTSET(INP_RECVOPTS); 1113 break; 1114 1115 case IP_RECVRETOPTS: 1116 OPTSET(INP_RECVRETOPTS); 1117 break; 1118 1119 case IP_RECVDSTADDR: 1120 OPTSET(INP_RECVDSTADDR); 1121 break; 1122 1123 case IP_ORIGDSTADDR: 1124 OPTSET2(INP_ORIGDSTADDR, optval); 1125 break; 1126 1127 case IP_RECVTTL: 1128 OPTSET(INP_RECVTTL); 1129 break; 1130 1131 case IP_RECVIF: 1132 OPTSET(INP_RECVIF); 1133 break; 1134 1135 case IP_ONESBCAST: 1136 OPTSET(INP_ONESBCAST); 1137 break; 1138 case IP_DONTFRAG: 1139 OPTSET(INP_DONTFRAG); 1140 break; 1141 case IP_BINDANY: 1142 OPTSET(INP_BINDANY); 1143 break; 1144 case IP_RECVTOS: 1145 OPTSET(INP_RECVTOS); 1146 break; 1147 case IP_BINDMULTI: 1148 OPTSET2(INP_BINDMULTI, optval); 1149 break; 1150 case IP_RECVFLOWID: 1151 OPTSET2(INP_RECVFLOWID, optval); 1152 break; 1153 #ifdef RSS 1154 case IP_RSS_LISTEN_BUCKET: 1155 if ((optval >= 0) && 1156 (optval < rss_getnumbuckets())) { 1157 inp->inp_rss_listen_bucket = optval; 1158 OPTSET2(INP_RSS_BUCKET_SET, 1); 1159 } else { 1160 error = EINVAL; 1161 } 1162 break; 1163 case IP_RECVRSSBUCKETID: 1164 OPTSET2(INP_RECVRSSBUCKETID, optval); 1165 break; 1166 #endif 1167 } 1168 break; 1169 #undef OPTSET 1170 #undef OPTSET2 1171 1172 /* 1173 * Multicast socket options are processed by the in_mcast 1174 * module. 1175 */ 1176 case IP_MULTICAST_IF: 1177 case IP_MULTICAST_VIF: 1178 case IP_MULTICAST_TTL: 1179 case IP_MULTICAST_LOOP: 1180 case IP_ADD_MEMBERSHIP: 1181 case IP_DROP_MEMBERSHIP: 1182 case IP_ADD_SOURCE_MEMBERSHIP: 1183 case IP_DROP_SOURCE_MEMBERSHIP: 1184 case IP_BLOCK_SOURCE: 1185 case IP_UNBLOCK_SOURCE: 1186 case IP_MSFILTER: 1187 case MCAST_JOIN_GROUP: 1188 case MCAST_LEAVE_GROUP: 1189 case MCAST_JOIN_SOURCE_GROUP: 1190 case MCAST_LEAVE_SOURCE_GROUP: 1191 case MCAST_BLOCK_SOURCE: 1192 case MCAST_UNBLOCK_SOURCE: 1193 error = inp_setmoptions(inp, sopt); 1194 break; 1195 1196 case IP_PORTRANGE: 1197 error = sooptcopyin(sopt, &optval, sizeof optval, 1198 sizeof optval); 1199 if (error) 1200 break; 1201 1202 INP_WLOCK(inp); 1203 switch (optval) { 1204 case IP_PORTRANGE_DEFAULT: 1205 inp->inp_flags &= ~(INP_LOWPORT); 1206 inp->inp_flags &= ~(INP_HIGHPORT); 1207 break; 1208 1209 case IP_PORTRANGE_HIGH: 1210 inp->inp_flags &= ~(INP_LOWPORT); 1211 inp->inp_flags |= INP_HIGHPORT; 1212 break; 1213 1214 case IP_PORTRANGE_LOW: 1215 inp->inp_flags &= ~(INP_HIGHPORT); 1216 inp->inp_flags |= INP_LOWPORT; 1217 break; 1218 1219 default: 1220 error = EINVAL; 1221 break; 1222 } 1223 INP_WUNLOCK(inp); 1224 break; 1225 1226 #if defined(IPSEC) || defined(IPSEC_SUPPORT) 1227 case IP_IPSEC_POLICY: 1228 if (IPSEC_ENABLED(ipv4)) { 1229 error = IPSEC_PCBCTL(ipv4, inp, sopt); 1230 break; 1231 } 1232 /* FALLTHROUGH */ 1233 #endif /* IPSEC */ 1234 1235 default: 1236 error = ENOPROTOOPT; 1237 break; 1238 } 1239 break; 1240 1241 case SOPT_GET: 1242 switch (sopt->sopt_name) { 1243 case IP_OPTIONS: 1244 case IP_RETOPTS: 1245 if (inp->inp_options) 1246 error = sooptcopyout(sopt, 1247 mtod(inp->inp_options, 1248 char *), 1249 inp->inp_options->m_len); 1250 else 1251 sopt->sopt_valsize = 0; 1252 break; 1253 1254 case IP_TOS: 1255 case IP_TTL: 1256 case IP_MINTTL: 1257 case IP_RECVOPTS: 1258 case IP_RECVRETOPTS: 1259 case IP_ORIGDSTADDR: 1260 case IP_RECVDSTADDR: 1261 case IP_RECVTTL: 1262 case IP_RECVIF: 1263 case IP_PORTRANGE: 1264 case IP_ONESBCAST: 1265 case IP_DONTFRAG: 1266 case IP_BINDANY: 1267 case IP_RECVTOS: 1268 case IP_BINDMULTI: 1269 case IP_FLOWID: 1270 case IP_FLOWTYPE: 1271 case IP_RECVFLOWID: 1272 #ifdef RSS 1273 case IP_RSSBUCKETID: 1274 case IP_RECVRSSBUCKETID: 1275 #endif 1276 switch (sopt->sopt_name) { 1277 1278 case IP_TOS: 1279 optval = inp->inp_ip_tos; 1280 break; 1281 1282 case IP_TTL: 1283 optval = inp->inp_ip_ttl; 1284 break; 1285 1286 case IP_MINTTL: 1287 optval = inp->inp_ip_minttl; 1288 break; 1289 1290 #define OPTBIT(bit) (inp->inp_flags & bit ? 1 : 0) 1291 #define OPTBIT2(bit) (inp->inp_flags2 & bit ? 1 : 0) 1292 1293 case IP_RECVOPTS: 1294 optval = OPTBIT(INP_RECVOPTS); 1295 break; 1296 1297 case IP_RECVRETOPTS: 1298 optval = OPTBIT(INP_RECVRETOPTS); 1299 break; 1300 1301 case IP_RECVDSTADDR: 1302 optval = OPTBIT(INP_RECVDSTADDR); 1303 break; 1304 1305 case IP_ORIGDSTADDR: 1306 optval = OPTBIT2(INP_ORIGDSTADDR); 1307 break; 1308 1309 case IP_RECVTTL: 1310 optval = OPTBIT(INP_RECVTTL); 1311 break; 1312 1313 case IP_RECVIF: 1314 optval = OPTBIT(INP_RECVIF); 1315 break; 1316 1317 case IP_PORTRANGE: 1318 if (inp->inp_flags & INP_HIGHPORT) 1319 optval = IP_PORTRANGE_HIGH; 1320 else if (inp->inp_flags & INP_LOWPORT) 1321 optval = IP_PORTRANGE_LOW; 1322 else 1323 optval = 0; 1324 break; 1325 1326 case IP_ONESBCAST: 1327 optval = OPTBIT(INP_ONESBCAST); 1328 break; 1329 case IP_DONTFRAG: 1330 optval = OPTBIT(INP_DONTFRAG); 1331 break; 1332 case IP_BINDANY: 1333 optval = OPTBIT(INP_BINDANY); 1334 break; 1335 case IP_RECVTOS: 1336 optval = OPTBIT(INP_RECVTOS); 1337 break; 1338 case IP_FLOWID: 1339 optval = inp->inp_flowid; 1340 break; 1341 case IP_FLOWTYPE: 1342 optval = inp->inp_flowtype; 1343 break; 1344 case IP_RECVFLOWID: 1345 optval = OPTBIT2(INP_RECVFLOWID); 1346 break; 1347 #ifdef RSS 1348 case IP_RSSBUCKETID: 1349 retval = rss_hash2bucket(inp->inp_flowid, 1350 inp->inp_flowtype, 1351 &rss_bucket); 1352 if (retval == 0) 1353 optval = rss_bucket; 1354 else 1355 error = EINVAL; 1356 break; 1357 case IP_RECVRSSBUCKETID: 1358 optval = OPTBIT2(INP_RECVRSSBUCKETID); 1359 break; 1360 #endif 1361 case IP_BINDMULTI: 1362 optval = OPTBIT2(INP_BINDMULTI); 1363 break; 1364 } 1365 error = sooptcopyout(sopt, &optval, sizeof optval); 1366 break; 1367 1368 /* 1369 * Multicast socket options are processed by the in_mcast 1370 * module. 1371 */ 1372 case IP_MULTICAST_IF: 1373 case IP_MULTICAST_VIF: 1374 case IP_MULTICAST_TTL: 1375 case IP_MULTICAST_LOOP: 1376 case IP_MSFILTER: 1377 error = inp_getmoptions(inp, sopt); 1378 break; 1379 1380 #if defined(IPSEC) || defined(IPSEC_SUPPORT) 1381 case IP_IPSEC_POLICY: 1382 if (IPSEC_ENABLED(ipv4)) { 1383 error = IPSEC_PCBCTL(ipv4, inp, sopt); 1384 break; 1385 } 1386 /* FALLTHROUGH */ 1387 #endif /* IPSEC */ 1388 1389 default: 1390 error = ENOPROTOOPT; 1391 break; 1392 } 1393 break; 1394 } 1395 return (error); 1396 } 1397 1398 /* 1399 * Routine called from ip_output() to loop back a copy of an IP multicast 1400 * packet to the input queue of a specified interface. Note that this 1401 * calls the output routine of the loopback "driver", but with an interface 1402 * pointer that might NOT be a loopback interface -- evil, but easier than 1403 * replicating that code here. 1404 */ 1405 static void 1406 ip_mloopback(struct ifnet *ifp, const struct mbuf *m, int hlen) 1407 { 1408 struct ip *ip; 1409 struct mbuf *copym; 1410 1411 /* 1412 * Make a deep copy of the packet because we're going to 1413 * modify the pack in order to generate checksums. 1414 */ 1415 copym = m_dup(m, M_NOWAIT); 1416 if (copym != NULL && (!M_WRITABLE(copym) || copym->m_len < hlen)) 1417 copym = m_pullup(copym, hlen); 1418 if (copym != NULL) { 1419 /* If needed, compute the checksum and mark it as valid. */ 1420 if (copym->m_pkthdr.csum_flags & CSUM_DELAY_DATA) { 1421 in_delayed_cksum(copym); 1422 copym->m_pkthdr.csum_flags &= ~CSUM_DELAY_DATA; 1423 copym->m_pkthdr.csum_flags |= 1424 CSUM_DATA_VALID | CSUM_PSEUDO_HDR; 1425 copym->m_pkthdr.csum_data = 0xffff; 1426 } 1427 /* 1428 * We don't bother to fragment if the IP length is greater 1429 * than the interface's MTU. Can this possibly matter? 1430 */ 1431 ip = mtod(copym, struct ip *); 1432 ip->ip_sum = 0; 1433 ip->ip_sum = in_cksum(copym, hlen); 1434 if_simloop(ifp, copym, AF_INET, 0); 1435 } 1436 } 1437