1 /* 2 * Copyright (c) 1982, 1986, 1988, 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. All advertising materials mentioning features or use of this software 14 * must display the following acknowledgement: 15 * This product includes software developed by the University of 16 * California, Berkeley and its contributors. 17 * 4. Neither the name of the University nor the names of its contributors 18 * may be used to endorse or promote products derived from this software 19 * without specific prior written permission. 20 * 21 * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND 22 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE 23 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE 24 * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE 25 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL 26 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS 27 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) 28 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT 29 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY 30 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF 31 * SUCH DAMAGE. 32 * 33 * @(#)raw_ip.c 8.7 (Berkeley) 5/15/95 34 * $FreeBSD: src/sys/netinet/raw_ip.c,v 1.64.2.16 2003/08/24 08:24:38 hsu Exp $ 35 * $DragonFly: src/sys/netinet/raw_ip.c,v 1.32 2008/07/28 15:07:28 sephe Exp $ 36 */ 37 38 #include "opt_inet6.h" 39 #include "opt_ipsec.h" 40 41 #include <sys/param.h> 42 #include <sys/systm.h> 43 #include <sys/kernel.h> 44 #include <sys/jail.h> 45 #include <sys/malloc.h> 46 #include <sys/mbuf.h> 47 #include <sys/proc.h> 48 #include <sys/protosw.h> 49 #include <sys/socket.h> 50 #include <sys/socketvar.h> 51 #include <sys/sysctl.h> 52 #include <sys/thread2.h> 53 54 #include <machine/stdarg.h> 55 56 #include <vm/vm_zone.h> 57 58 #include <net/if.h> 59 #include <net/route.h> 60 61 #define _IP_VHL 62 #include <netinet/in.h> 63 #include <netinet/in_systm.h> 64 #include <netinet/ip.h> 65 #include <netinet/in_pcb.h> 66 #include <netinet/in_var.h> 67 #include <netinet/ip_var.h> 68 69 #include <net/ip_mroute/ip_mroute.h> 70 #include <net/ipfw/ip_fw.h> 71 #include <net/dummynet/ip_dummynet.h> 72 73 #ifdef FAST_IPSEC 74 #include <netproto/ipsec/ipsec.h> 75 #endif /*FAST_IPSEC*/ 76 77 #ifdef IPSEC 78 #include <netinet6/ipsec.h> 79 #endif /*IPSEC*/ 80 81 struct inpcbinfo ripcbinfo; 82 83 /* control hooks for ipfw and dummynet */ 84 ip_fw_ctl_t *ip_fw_ctl_ptr; 85 ip_dn_ctl_t *ip_dn_ctl_ptr; 86 87 /* 88 * hooks for multicast routing. They all default to NULL, 89 * so leave them not initialized and rely on BSS being set to 0. 90 */ 91 92 /* The socket used to communicate with the multicast routing daemon. */ 93 struct socket *ip_mrouter; 94 95 /* The various mrouter and rsvp functions */ 96 int (*ip_mrouter_set)(struct socket *, struct sockopt *); 97 int (*ip_mrouter_get)(struct socket *, struct sockopt *); 98 int (*ip_mrouter_done)(void); 99 int (*ip_mforward)(struct ip *, struct ifnet *, struct mbuf *, 100 struct ip_moptions *); 101 int (*mrt_ioctl)(int, caddr_t); 102 int (*legal_vif_num)(int); 103 u_long (*ip_mcast_src)(int); 104 105 void (*rsvp_input_p)(struct mbuf *m, ...); 106 int (*ip_rsvp_vif)(struct socket *, struct sockopt *); 107 void (*ip_rsvp_force_done)(struct socket *); 108 109 /* 110 * Nominal space allocated to a raw ip socket. 111 */ 112 #define RIPSNDQ 8192 113 #define RIPRCVQ 8192 114 115 /* 116 * Raw interface to IP protocol. 117 */ 118 119 /* 120 * Initialize raw connection block queue. 121 */ 122 void 123 rip_init(void) 124 { 125 in_pcbinfo_init(&ripcbinfo); 126 /* 127 * XXX We don't use the hash list for raw IP, but it's easier 128 * to allocate a one entry hash list than it is to check all 129 * over the place for hashbase == NULL. 130 */ 131 ripcbinfo.hashbase = hashinit(1, M_PCB, &ripcbinfo.hashmask); 132 ripcbinfo.porthashbase = hashinit(1, M_PCB, &ripcbinfo.porthashmask); 133 ripcbinfo.wildcardhashbase = hashinit(1, M_PCB, 134 &ripcbinfo.wildcardhashmask); 135 ripcbinfo.ipi_zone = zinit("ripcb", sizeof(struct inpcb), 136 maxsockets, ZONE_INTERRUPT, 0); 137 } 138 139 /* 140 * Setup generic address and protocol structures 141 * for raw_input routine, then pass them along with 142 * mbuf chain. 143 */ 144 void 145 rip_input(struct mbuf *m, ...) 146 { 147 struct sockaddr_in ripsrc = { sizeof ripsrc, AF_INET }; 148 struct ip *ip = mtod(m, struct ip *); 149 struct inpcb *inp; 150 struct inpcb *last = NULL; 151 struct mbuf *opts = NULL; 152 int off, proto; 153 __va_list ap; 154 155 __va_start(ap, m); 156 off = __va_arg(ap, int); 157 proto = __va_arg(ap, int); 158 __va_end(ap); 159 160 ripsrc.sin_addr = ip->ip_src; 161 LIST_FOREACH(inp, &ripcbinfo.pcblisthead, inp_list) { 162 if (inp->inp_flags & INP_PLACEMARKER) 163 continue; 164 #ifdef INET6 165 if ((inp->inp_vflag & INP_IPV4) == 0) 166 continue; 167 #endif 168 if (inp->inp_ip_p && inp->inp_ip_p != proto) 169 continue; 170 if (inp->inp_laddr.s_addr != INADDR_ANY && 171 inp->inp_laddr.s_addr != ip->ip_dst.s_addr) 172 continue; 173 if (inp->inp_faddr.s_addr != INADDR_ANY && 174 inp->inp_faddr.s_addr != ip->ip_src.s_addr) 175 continue; 176 if (last) { 177 struct mbuf *n = m_copypacket(m, MB_DONTWAIT); 178 179 #ifdef IPSEC 180 /* check AH/ESP integrity. */ 181 if (n && ipsec4_in_reject_so(n, last->inp_socket)) { 182 m_freem(n); 183 ipsecstat.in_polvio++; 184 /* do not inject data to pcb */ 185 } else 186 #endif /*IPSEC*/ 187 #ifdef FAST_IPSEC 188 /* check AH/ESP integrity. */ 189 if (ipsec4_in_reject(n, last)) { 190 m_freem(n); 191 /* do not inject data to pcb */ 192 } else 193 #endif /*FAST_IPSEC*/ 194 if (n) { 195 if (last->inp_flags & INP_CONTROLOPTS || 196 last->inp_socket->so_options & SO_TIMESTAMP) 197 ip_savecontrol(last, &opts, ip, n); 198 if (ssb_appendaddr(&last->inp_socket->so_rcv, 199 (struct sockaddr *)&ripsrc, n, 200 opts) == 0) { 201 /* should notify about lost packet */ 202 m_freem(n); 203 if (opts) 204 m_freem(opts); 205 } else 206 sorwakeup(last->inp_socket); 207 opts = 0; 208 } 209 } 210 last = inp; 211 } 212 #ifdef IPSEC 213 /* check AH/ESP integrity. */ 214 if (last && ipsec4_in_reject_so(m, last->inp_socket)) { 215 m_freem(m); 216 ipsecstat.in_polvio++; 217 ipstat.ips_delivered--; 218 /* do not inject data to pcb */ 219 } else 220 #endif /*IPSEC*/ 221 #ifdef FAST_IPSEC 222 /* check AH/ESP integrity. */ 223 if (last && ipsec4_in_reject(m, last)) { 224 m_freem(m); 225 ipstat.ips_delivered--; 226 /* do not inject data to pcb */ 227 } else 228 #endif /*FAST_IPSEC*/ 229 /* Check the minimum TTL for socket. */ 230 if (last && ip->ip_ttl < last->inp_ip_minttl) { 231 m_freem(opts); 232 ipstat.ips_delivered--; 233 } else if (last) { 234 if (last->inp_flags & INP_CONTROLOPTS || 235 last->inp_socket->so_options & SO_TIMESTAMP) 236 ip_savecontrol(last, &opts, ip, m); 237 if (ssb_appendaddr(&last->inp_socket->so_rcv, 238 (struct sockaddr *)&ripsrc, m, opts) == 0) { 239 m_freem(m); 240 if (opts) 241 m_freem(opts); 242 } else 243 sorwakeup(last->inp_socket); 244 } else { 245 m_freem(m); 246 ipstat.ips_noproto++; 247 ipstat.ips_delivered--; 248 } 249 } 250 251 /* 252 * Generate IP header and pass packet to ip_output. 253 * Tack on options user may have setup with control call. 254 */ 255 int 256 rip_output(struct mbuf *m, struct socket *so, ...) 257 { 258 struct ip *ip; 259 struct inpcb *inp = so->so_pcb; 260 __va_list ap; 261 int flags = (so->so_options & SO_DONTROUTE) | IP_ALLOWBROADCAST; 262 u_long dst; 263 264 __va_start(ap, so); 265 dst = __va_arg(ap, u_long); 266 __va_end(ap); 267 268 /* 269 * If the user handed us a complete IP packet, use it. 270 * Otherwise, allocate an mbuf for a header and fill it in. 271 */ 272 if ((inp->inp_flags & INP_HDRINCL) == 0) { 273 if (m->m_pkthdr.len + sizeof(struct ip) > IP_MAXPACKET) { 274 m_freem(m); 275 return(EMSGSIZE); 276 } 277 M_PREPEND(m, sizeof(struct ip), MB_WAIT); 278 if (m == NULL) 279 return(ENOBUFS); 280 ip = mtod(m, struct ip *); 281 ip->ip_tos = inp->inp_ip_tos; 282 ip->ip_off = 0; 283 ip->ip_p = inp->inp_ip_p; 284 ip->ip_len = m->m_pkthdr.len; 285 ip->ip_src = inp->inp_laddr; 286 ip->ip_dst.s_addr = dst; 287 ip->ip_ttl = inp->inp_ip_ttl; 288 } else { 289 if (m->m_pkthdr.len > IP_MAXPACKET) { 290 m_freem(m); 291 return(EMSGSIZE); 292 } 293 ip = mtod(m, struct ip *); 294 /* don't allow both user specified and setsockopt options, 295 and don't allow packet length sizes that will crash */ 296 if (((IP_VHL_HL(ip->ip_vhl) != (sizeof (*ip) >> 2)) && 297 inp->inp_options) || 298 (ip->ip_len > m->m_pkthdr.len) || 299 (ip->ip_len < (IP_VHL_HL(ip->ip_vhl) << 2))) { 300 m_freem(m); 301 return EINVAL; 302 } 303 if (ip->ip_id == 0) 304 ip->ip_id = ip_newid(); 305 /* XXX prevent ip_output from overwriting header fields */ 306 flags |= IP_RAWOUTPUT; 307 ipstat.ips_rawout++; 308 } 309 310 return (ip_output(m, inp->inp_options, &inp->inp_route, flags, 311 inp->inp_moptions, inp)); 312 } 313 314 /* 315 * Raw IP socket option processing. 316 */ 317 int 318 rip_ctloutput(struct socket *so, struct sockopt *sopt) 319 { 320 struct inpcb *inp = so->so_pcb; 321 int error, optval; 322 323 if (sopt->sopt_level != IPPROTO_IP) 324 return (EINVAL); 325 326 error = 0; 327 328 switch (sopt->sopt_dir) { 329 case SOPT_GET: 330 switch (sopt->sopt_name) { 331 case IP_HDRINCL: 332 optval = inp->inp_flags & INP_HDRINCL; 333 soopt_from_kbuf(sopt, &optval, sizeof optval); 334 break; 335 336 case IP_FW_ADD: /* ADD actually returns the body... */ 337 case IP_FW_GET: 338 if (IPFW_LOADED) 339 error = ip_fw_sockopt(sopt); 340 else 341 error = ENOPROTOOPT; 342 break; 343 344 case IP_DUMMYNET_GET: 345 error = ip_dn_sockopt(sopt); 346 break ; 347 348 case MRT_INIT: 349 case MRT_DONE: 350 case MRT_ADD_VIF: 351 case MRT_DEL_VIF: 352 case MRT_ADD_MFC: 353 case MRT_DEL_MFC: 354 case MRT_VERSION: 355 case MRT_ASSERT: 356 case MRT_API_SUPPORT: 357 case MRT_API_CONFIG: 358 case MRT_ADD_BW_UPCALL: 359 case MRT_DEL_BW_UPCALL: 360 error = ip_mrouter_get ? ip_mrouter_get(so, sopt) : 361 EOPNOTSUPP; 362 break; 363 364 default: 365 error = ip_ctloutput(so, sopt); 366 break; 367 } 368 break; 369 370 case SOPT_SET: 371 switch (sopt->sopt_name) { 372 case IP_HDRINCL: 373 error = soopt_to_kbuf(sopt, &optval, sizeof optval, 374 sizeof optval); 375 if (error) 376 break; 377 if (optval) 378 inp->inp_flags |= INP_HDRINCL; 379 else 380 inp->inp_flags &= ~INP_HDRINCL; 381 break; 382 383 case IP_FW_ADD: 384 case IP_FW_DEL: 385 case IP_FW_FLUSH: 386 case IP_FW_ZERO: 387 case IP_FW_RESETLOG: 388 if (IPFW_LOADED) 389 error = ip_fw_ctl_ptr(sopt); 390 else 391 error = ENOPROTOOPT; 392 break; 393 394 case IP_DUMMYNET_CONFIGURE: 395 case IP_DUMMYNET_DEL: 396 case IP_DUMMYNET_FLUSH: 397 error = ip_dn_sockopt(sopt); 398 break ; 399 400 case IP_RSVP_ON: 401 error = ip_rsvp_init(so); 402 break; 403 404 case IP_RSVP_OFF: 405 error = ip_rsvp_done(); 406 break; 407 408 case IP_RSVP_VIF_ON: 409 case IP_RSVP_VIF_OFF: 410 error = ip_rsvp_vif ? 411 ip_rsvp_vif(so, sopt) : EINVAL; 412 break; 413 414 case MRT_INIT: 415 case MRT_DONE: 416 case MRT_ADD_VIF: 417 case MRT_DEL_VIF: 418 case MRT_ADD_MFC: 419 case MRT_DEL_MFC: 420 case MRT_VERSION: 421 case MRT_ASSERT: 422 case MRT_API_SUPPORT: 423 case MRT_API_CONFIG: 424 case MRT_ADD_BW_UPCALL: 425 case MRT_DEL_BW_UPCALL: 426 error = ip_mrouter_set ? ip_mrouter_set(so, sopt) : 427 EOPNOTSUPP; 428 break; 429 430 default: 431 error = ip_ctloutput(so, sopt); 432 break; 433 } 434 break; 435 } 436 437 return (error); 438 } 439 440 /* 441 * This function exists solely to receive the PRC_IFDOWN messages which 442 * are sent by if_down(). It looks for an ifaddr whose ifa_addr is sa, 443 * and calls in_ifadown() to remove all routes corresponding to that address. 444 * It also receives the PRC_IFUP messages from if_up() and reinstalls the 445 * interface routes. 446 */ 447 void 448 rip_ctlinput(int cmd, struct sockaddr *sa, void *vip) 449 { 450 struct in_ifaddr *ia; 451 struct in_ifaddr_container *iac; 452 struct ifnet *ifp; 453 int err; 454 int flags; 455 456 switch (cmd) { 457 case PRC_IFDOWN: 458 TAILQ_FOREACH(iac, &in_ifaddrheads[mycpuid], ia_link) { 459 ia = iac->ia; 460 461 if (ia->ia_ifa.ifa_addr == sa && 462 (ia->ia_flags & IFA_ROUTE)) { 463 /* 464 * in_ifscrub kills the interface route. 465 */ 466 in_ifscrub(ia->ia_ifp, ia); 467 /* 468 * in_ifadown gets rid of all the rest of 469 * the routes. This is not quite the right 470 * thing to do, but at least if we are running 471 * a routing process they will come back. 472 */ 473 in_ifadown(&ia->ia_ifa, 0); 474 break; 475 } 476 } 477 break; 478 479 case PRC_IFUP: 480 ia = NULL; 481 TAILQ_FOREACH(iac, &in_ifaddrheads[mycpuid], ia_link) { 482 if (iac->ia->ia_ifa.ifa_addr == sa) { 483 ia = iac->ia; 484 break; 485 } 486 } 487 if (ia == NULL || (ia->ia_flags & IFA_ROUTE)) 488 return; 489 flags = RTF_UP; 490 ifp = ia->ia_ifa.ifa_ifp; 491 492 if ((ifp->if_flags & IFF_LOOPBACK) || 493 (ifp->if_flags & IFF_POINTOPOINT)) 494 flags |= RTF_HOST; 495 496 err = rtinit(&ia->ia_ifa, RTM_ADD, flags); 497 if (err == 0) 498 ia->ia_flags |= IFA_ROUTE; 499 break; 500 } 501 } 502 503 u_long rip_sendspace = RIPSNDQ; 504 u_long rip_recvspace = RIPRCVQ; 505 506 SYSCTL_INT(_net_inet_raw, OID_AUTO, maxdgram, CTLFLAG_RW, 507 &rip_sendspace, 0, "Maximum outgoing raw IP datagram size"); 508 SYSCTL_INT(_net_inet_raw, OID_AUTO, recvspace, CTLFLAG_RW, 509 &rip_recvspace, 0, "Maximum incoming raw IP datagram size"); 510 511 static int 512 rip_attach(struct socket *so, int proto, struct pru_attach_info *ai) 513 { 514 struct inpcb *inp; 515 int error; 516 int flag; 517 518 if (jailed(ai->p_ucred) && jail_allow_raw_sockets) 519 flag = NULL_CRED_OKAY | PRISON_ROOT; 520 else 521 flag = NULL_CRED_OKAY; 522 523 inp = so->so_pcb; 524 if (inp) 525 panic("rip_attach"); 526 if ((error = suser_cred(ai->p_ucred, flag)) != 0) 527 return error; 528 529 error = soreserve(so, rip_sendspace, rip_recvspace, ai->sb_rlimit); 530 if (error) 531 return error; 532 crit_enter(); 533 error = in_pcballoc(so, &ripcbinfo); 534 crit_exit(); 535 if (error) 536 return error; 537 inp = (struct inpcb *)so->so_pcb; 538 inp->inp_vflag |= INP_IPV4; 539 inp->inp_ip_p = proto; 540 inp->inp_ip_ttl = ip_defttl; 541 return 0; 542 } 543 544 static int 545 rip_detach(struct socket *so) 546 { 547 struct inpcb *inp; 548 549 inp = so->so_pcb; 550 if (inp == 0) 551 panic("rip_detach"); 552 if (so == ip_mrouter && ip_mrouter_done) 553 ip_mrouter_done(); 554 if (ip_rsvp_force_done) 555 ip_rsvp_force_done(so); 556 if (so == ip_rsvpd) 557 ip_rsvp_done(); 558 in_pcbdetach(inp); 559 return 0; 560 } 561 562 static int 563 rip_abort(struct socket *so) 564 { 565 soisdisconnected(so); 566 if (so->so_state & SS_NOFDREF) 567 return rip_detach(so); 568 return 0; 569 } 570 571 static int 572 rip_disconnect(struct socket *so) 573 { 574 if ((so->so_state & SS_ISCONNECTED) == 0) 575 return ENOTCONN; 576 return rip_abort(so); 577 } 578 579 static int 580 rip_bind(struct socket *so, struct sockaddr *nam, struct thread *td) 581 { 582 struct inpcb *inp = so->so_pcb; 583 struct sockaddr_in *addr = (struct sockaddr_in *)nam; 584 585 if (nam->sa_len != sizeof(*addr)) 586 return EINVAL; 587 588 if (TAILQ_EMPTY(&ifnet) || ((addr->sin_family != AF_INET) && 589 (addr->sin_family != AF_IMPLINK)) || 590 (addr->sin_addr.s_addr != INADDR_ANY && 591 ifa_ifwithaddr((struct sockaddr *)addr) == 0)) 592 return EADDRNOTAVAIL; 593 inp->inp_laddr = addr->sin_addr; 594 return 0; 595 } 596 597 static int 598 rip_connect(struct socket *so, struct sockaddr *nam, struct thread *td) 599 { 600 struct inpcb *inp = so->so_pcb; 601 struct sockaddr_in *addr = (struct sockaddr_in *)nam; 602 603 if (nam->sa_len != sizeof(*addr)) 604 return EINVAL; 605 if (TAILQ_EMPTY(&ifnet)) 606 return EADDRNOTAVAIL; 607 if ((addr->sin_family != AF_INET) && 608 (addr->sin_family != AF_IMPLINK)) 609 return EAFNOSUPPORT; 610 inp->inp_faddr = addr->sin_addr; 611 soisconnected(so); 612 return 0; 613 } 614 615 static int 616 rip_shutdown(struct socket *so) 617 { 618 socantsendmore(so); 619 return 0; 620 } 621 622 static int 623 rip_send(struct socket *so, int flags, struct mbuf *m, struct sockaddr *nam, 624 struct mbuf *control, struct thread *td) 625 { 626 struct inpcb *inp = so->so_pcb; 627 u_long dst; 628 629 if (so->so_state & SS_ISCONNECTED) { 630 if (nam) { 631 m_freem(m); 632 return EISCONN; 633 } 634 dst = inp->inp_faddr.s_addr; 635 } else { 636 if (nam == NULL) { 637 m_freem(m); 638 return ENOTCONN; 639 } 640 dst = ((struct sockaddr_in *)nam)->sin_addr.s_addr; 641 } 642 return rip_output(m, so, dst); 643 } 644 645 SYSCTL_PROC(_net_inet_raw, OID_AUTO/*XXX*/, pcblist, CTLFLAG_RD, &ripcbinfo, 0, 646 in_pcblist_global, "S,xinpcb", "List of active raw IP sockets"); 647 648 struct pr_usrreqs rip_usrreqs = { 649 .pru_abort = rip_abort, 650 .pru_accept = pru_accept_notsupp, 651 .pru_attach = rip_attach, 652 .pru_bind = rip_bind, 653 .pru_connect = rip_connect, 654 .pru_connect2 = pru_connect2_notsupp, 655 .pru_control = in_control, 656 .pru_detach = rip_detach, 657 .pru_disconnect = rip_disconnect, 658 .pru_listen = pru_listen_notsupp, 659 .pru_peeraddr = in_setpeeraddr, 660 .pru_rcvd = pru_rcvd_notsupp, 661 .pru_rcvoob = pru_rcvoob_notsupp, 662 .pru_send = rip_send, 663 .pru_sense = pru_sense_null, 664 .pru_shutdown = rip_shutdown, 665 .pru_sockaddr = in_setsockaddr, 666 .pru_sosend = sosend, 667 .pru_soreceive = soreceive, 668 .pru_sopoll = sopoll 669 }; 670