raw_ip.c revision 98115
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: head/sys/netinet/raw_ip.c 98115 2002-06-11 10:58:57Z hsu $ 35 */ 36 37#include "opt_inet6.h" 38#include "opt_ipsec.h" 39#include "opt_random_ip_id.h" 40 41#include <sys/param.h> 42#include <sys/kernel.h> 43#include <sys/lock.h> 44#include <sys/malloc.h> 45#include <sys/mbuf.h> 46#include <sys/proc.h> 47#include <sys/protosw.h> 48#include <sys/signalvar.h> 49#include <sys/socket.h> 50#include <sys/socketvar.h> 51#include <sys/sx.h> 52#include <sys/sysctl.h> 53#include <sys/systm.h> 54 55#include <vm/uma.h> 56 57#include <net/if.h> 58#include <net/route.h> 59 60#define _IP_VHL 61#include <netinet/in.h> 62#include <netinet/in_systm.h> 63#include <netinet/in_pcb.h> 64#include <netinet/in_var.h> 65#include <netinet/ip.h> 66#include <netinet/ip_var.h> 67#include <netinet/ip_mroute.h> 68 69#include <netinet/ip_fw.h> 70#include <netinet/ip_dummynet.h> 71 72#ifdef IPSEC 73#include <netinet6/ipsec.h> 74#endif /*IPSEC*/ 75 76struct inpcbhead ripcb; 77struct inpcbinfo ripcbinfo; 78 79/* control hooks for ipfw and dummynet */ 80ip_fw_ctl_t *ip_fw_ctl_ptr; 81ip_dn_ctl_t *ip_dn_ctl_ptr; 82 83/* 84 * Nominal space allocated to a raw ip socket. 85 */ 86#define RIPSNDQ 8192 87#define RIPRCVQ 8192 88 89/* 90 * Raw interface to IP protocol. 91 */ 92 93/* 94 * Initialize raw connection block q. 95 */ 96void 97rip_init() 98{ 99 INP_INFO_LOCK_INIT(&ripcbinfo, "rip"); 100 LIST_INIT(&ripcb); 101 ripcbinfo.listhead = &ripcb; 102 /* 103 * XXX We don't use the hash list for raw IP, but it's easier 104 * to allocate a one entry hash list than it is to check all 105 * over the place for hashbase == NULL. 106 */ 107 ripcbinfo.hashbase = hashinit(1, M_PCB, &ripcbinfo.hashmask); 108 ripcbinfo.porthashbase = hashinit(1, M_PCB, &ripcbinfo.porthashmask); 109 ripcbinfo.ipi_zone = uma_zcreate("ripcb", sizeof(struct inpcb), 110 NULL, NULL, NULL, NULL, UMA_ALIGN_PTR, UMA_ZONE_NOFREE); 111 uma_zone_set_max(ripcbinfo.ipi_zone, maxsockets); 112} 113 114static struct sockaddr_in ripsrc = { sizeof(ripsrc), AF_INET }; 115/* 116 * Setup generic address and protocol structures 117 * for raw_input routine, then pass them along with 118 * mbuf chain. 119 */ 120void 121rip_input(m, off) 122 struct mbuf *m; 123 int off; 124{ 125 register struct ip *ip = mtod(m, struct ip *); 126 register struct inpcb *inp; 127 struct inpcb *last = 0; 128 struct mbuf *opts = 0; 129 int proto = ip->ip_p; 130 131 ripsrc.sin_addr = ip->ip_src; 132 LIST_FOREACH(inp, &ripcb, inp_list) { 133#ifdef INET6 134 if ((inp->inp_vflag & INP_IPV4) == 0) 135 continue; 136#endif 137 if (inp->inp_ip_p && inp->inp_ip_p != proto) 138 continue; 139 if (inp->inp_laddr.s_addr && 140 inp->inp_laddr.s_addr != ip->ip_dst.s_addr) 141 continue; 142 if (inp->inp_faddr.s_addr && 143 inp->inp_faddr.s_addr != ip->ip_src.s_addr) 144 continue; 145 if (last) { 146 struct mbuf *n = m_copy(m, 0, (int)M_COPYALL); 147 148#ifdef IPSEC 149 /* check AH/ESP integrity. */ 150 if (n && ipsec4_in_reject_so(n, last->inp_socket)) { 151 m_freem(n); 152 ipsecstat.in_polvio++; 153 /* do not inject data to pcb */ 154 } else 155#endif /*IPSEC*/ 156 if (n) { 157 if (last->inp_flags & INP_CONTROLOPTS || 158 last->inp_socket->so_options & SO_TIMESTAMP) 159 ip_savecontrol(last, &opts, ip, n); 160 if (sbappendaddr(&last->inp_socket->so_rcv, 161 (struct sockaddr *)&ripsrc, n, 162 opts) == 0) { 163 /* should notify about lost packet */ 164 m_freem(n); 165 if (opts) 166 m_freem(opts); 167 } else 168 sorwakeup(last->inp_socket); 169 opts = 0; 170 } 171 } 172 last = inp; 173 } 174#ifdef IPSEC 175 /* check AH/ESP integrity. */ 176 if (last && ipsec4_in_reject_so(m, last->inp_socket)) { 177 m_freem(m); 178 ipsecstat.in_polvio++; 179 ipstat.ips_delivered--; 180 /* do not inject data to pcb */ 181 } else 182#endif /*IPSEC*/ 183 if (last) { 184 if (last->inp_flags & INP_CONTROLOPTS || 185 last->inp_socket->so_options & SO_TIMESTAMP) 186 ip_savecontrol(last, &opts, ip, m); 187 if (sbappendaddr(&last->inp_socket->so_rcv, 188 (struct sockaddr *)&ripsrc, m, opts) == 0) { 189 m_freem(m); 190 if (opts) 191 m_freem(opts); 192 } else 193 sorwakeup(last->inp_socket); 194 } else { 195 m_freem(m); 196 ipstat.ips_noproto++; 197 ipstat.ips_delivered--; 198 } 199} 200 201/* 202 * Generate IP header and pass packet to ip_output. 203 * Tack on options user may have setup with control call. 204 */ 205int 206rip_output(m, so, dst) 207 struct mbuf *m; 208 struct socket *so; 209 u_long dst; 210{ 211 register struct ip *ip; 212 register struct inpcb *inp = sotoinpcb(so); 213 int flags = (so->so_options & SO_DONTROUTE) | IP_ALLOWBROADCAST; 214 215 /* 216 * If the user handed us a complete IP packet, use it. 217 * Otherwise, allocate an mbuf for a header and fill it in. 218 */ 219 if ((inp->inp_flags & INP_HDRINCL) == 0) { 220 if (m->m_pkthdr.len + sizeof(struct ip) > IP_MAXPACKET) { 221 m_freem(m); 222 return(EMSGSIZE); 223 } 224 M_PREPEND(m, sizeof(struct ip), M_TRYWAIT); 225 ip = mtod(m, struct ip *); 226 ip->ip_tos = inp->inp_ip_tos; 227 ip->ip_off = 0; 228 ip->ip_p = inp->inp_ip_p; 229 ip->ip_len = m->m_pkthdr.len; 230 ip->ip_src = inp->inp_laddr; 231 ip->ip_dst.s_addr = dst; 232 ip->ip_ttl = inp->inp_ip_ttl; 233 } else { 234 if (m->m_pkthdr.len > IP_MAXPACKET) { 235 m_freem(m); 236 return(EMSGSIZE); 237 } 238 ip = mtod(m, struct ip *); 239 /* don't allow both user specified and setsockopt options, 240 and don't allow packet length sizes that will crash */ 241 if (((IP_VHL_HL(ip->ip_vhl) != (sizeof (*ip) >> 2)) 242 && inp->inp_options) 243 || (ip->ip_len > m->m_pkthdr.len) 244 || (ip->ip_len < (IP_VHL_HL(ip->ip_vhl) << 2))) { 245 m_freem(m); 246 return EINVAL; 247 } 248 if (ip->ip_id == 0) 249#ifdef RANDOM_IP_ID 250 ip->ip_id = ip_randomid(); 251#else 252 ip->ip_id = htons(ip_id++); 253#endif 254 /* XXX prevent ip_output from overwriting header fields */ 255 flags |= IP_RAWOUTPUT; 256 ipstat.ips_rawout++; 257 } 258 259#ifdef IPSEC 260 if (ipsec_setsocket(m, so) != 0) { 261 m_freem(m); 262 return ENOBUFS; 263 } 264#endif /*IPSEC*/ 265 266 return (ip_output(m, inp->inp_options, &inp->inp_route, flags, 267 inp->inp_moptions)); 268} 269 270/* 271 * Raw IP socket option processing. 272 */ 273int 274rip_ctloutput(so, sopt) 275 struct socket *so; 276 struct sockopt *sopt; 277{ 278 struct inpcb *inp = sotoinpcb(so); 279 int error, optval; 280 281 if (sopt->sopt_level != IPPROTO_IP) 282 return (EINVAL); 283 284 error = 0; 285 286 switch (sopt->sopt_dir) { 287 case SOPT_GET: 288 switch (sopt->sopt_name) { 289 case IP_HDRINCL: 290 optval = inp->inp_flags & INP_HDRINCL; 291 error = sooptcopyout(sopt, &optval, sizeof optval); 292 break; 293 294 case IP_FW_ADD: /* ADD actually returns the body... */ 295 case IP_FW_GET: 296 if (IPFW_LOADED) 297 error = ip_fw_ctl_ptr(sopt); 298 else 299 error = ENOPROTOOPT; 300 break; 301 302 case IP_DUMMYNET_GET: 303 if (DUMMYNET_LOADED) 304 error = ip_dn_ctl_ptr(sopt); 305 else 306 error = ENOPROTOOPT; 307 break ; 308 309 case MRT_INIT: 310 case MRT_DONE: 311 case MRT_ADD_VIF: 312 case MRT_DEL_VIF: 313 case MRT_ADD_MFC: 314 case MRT_DEL_MFC: 315 case MRT_VERSION: 316 case MRT_ASSERT: 317 error = ip_mrouter_get(so, sopt); 318 break; 319 320 default: 321 error = ip_ctloutput(so, sopt); 322 break; 323 } 324 break; 325 326 case SOPT_SET: 327 switch (sopt->sopt_name) { 328 case IP_HDRINCL: 329 error = sooptcopyin(sopt, &optval, sizeof optval, 330 sizeof optval); 331 if (error) 332 break; 333 if (optval) 334 inp->inp_flags |= INP_HDRINCL; 335 else 336 inp->inp_flags &= ~INP_HDRINCL; 337 break; 338 339 case IP_FW_ADD: 340 case IP_FW_DEL: 341 case IP_FW_FLUSH: 342 case IP_FW_ZERO: 343 case IP_FW_RESETLOG: 344 if (IPFW_LOADED) 345 error = ip_fw_ctl_ptr(sopt); 346 else 347 error = ENOPROTOOPT; 348 break; 349 350 case IP_DUMMYNET_CONFIGURE: 351 case IP_DUMMYNET_DEL: 352 case IP_DUMMYNET_FLUSH: 353 if (DUMMYNET_LOADED) 354 error = ip_dn_ctl_ptr(sopt); 355 else 356 error = ENOPROTOOPT ; 357 break ; 358 359 case IP_RSVP_ON: 360 error = ip_rsvp_init(so); 361 break; 362 363 case IP_RSVP_OFF: 364 error = ip_rsvp_done(); 365 break; 366 367 /* XXX - should be combined */ 368 case IP_RSVP_VIF_ON: 369 error = ip_rsvp_vif_init(so, sopt); 370 break; 371 372 case IP_RSVP_VIF_OFF: 373 error = ip_rsvp_vif_done(so, sopt); 374 break; 375 376 case MRT_INIT: 377 case MRT_DONE: 378 case MRT_ADD_VIF: 379 case MRT_DEL_VIF: 380 case MRT_ADD_MFC: 381 case MRT_DEL_MFC: 382 case MRT_VERSION: 383 case MRT_ASSERT: 384 error = ip_mrouter_set(so, sopt); 385 break; 386 387 default: 388 error = ip_ctloutput(so, sopt); 389 break; 390 } 391 break; 392 } 393 394 return (error); 395} 396 397/* 398 * This function exists solely to receive the PRC_IFDOWN messages which 399 * are sent by if_down(). It looks for an ifaddr whose ifa_addr is sa, 400 * and calls in_ifadown() to remove all routes corresponding to that address. 401 * It also receives the PRC_IFUP messages from if_up() and reinstalls the 402 * interface routes. 403 */ 404void 405rip_ctlinput(cmd, sa, vip) 406 int cmd; 407 struct sockaddr *sa; 408 void *vip; 409{ 410 struct in_ifaddr *ia; 411 struct ifnet *ifp; 412 int err; 413 int flags; 414 415 switch (cmd) { 416 case PRC_IFDOWN: 417 TAILQ_FOREACH(ia, &in_ifaddrhead, ia_link) { 418 if (ia->ia_ifa.ifa_addr == sa 419 && (ia->ia_flags & IFA_ROUTE)) { 420 /* 421 * in_ifscrub kills the interface route. 422 */ 423 in_ifscrub(ia->ia_ifp, ia); 424 /* 425 * in_ifadown gets rid of all the rest of 426 * the routes. This is not quite the right 427 * thing to do, but at least if we are running 428 * a routing process they will come back. 429 */ 430 in_ifadown(&ia->ia_ifa, 0); 431 break; 432 } 433 } 434 break; 435 436 case PRC_IFUP: 437 TAILQ_FOREACH(ia, &in_ifaddrhead, ia_link) { 438 if (ia->ia_ifa.ifa_addr == sa) 439 break; 440 } 441 if (ia == 0 || (ia->ia_flags & IFA_ROUTE)) 442 return; 443 flags = RTF_UP; 444 ifp = ia->ia_ifa.ifa_ifp; 445 446 if ((ifp->if_flags & IFF_LOOPBACK) 447 || (ifp->if_flags & IFF_POINTOPOINT)) 448 flags |= RTF_HOST; 449 450 err = rtinit(&ia->ia_ifa, RTM_ADD, flags); 451 if (err == 0) 452 ia->ia_flags |= IFA_ROUTE; 453 break; 454 } 455} 456 457u_long rip_sendspace = RIPSNDQ; 458u_long rip_recvspace = RIPRCVQ; 459 460SYSCTL_INT(_net_inet_raw, OID_AUTO, maxdgram, CTLFLAG_RW, 461 &rip_sendspace, 0, "Maximum outgoing raw IP datagram size"); 462SYSCTL_INT(_net_inet_raw, OID_AUTO, recvspace, CTLFLAG_RW, 463 &rip_recvspace, 0, "Maximum incoming raw IP datagram size"); 464 465static int 466rip_attach(struct socket *so, int proto, struct thread *td) 467{ 468 struct inpcb *inp; 469 int error, s; 470 471 inp = sotoinpcb(so); 472 if (inp) 473 panic("rip_attach"); 474 if (td && (error = suser(td)) != 0) 475 return error; 476 477 error = soreserve(so, rip_sendspace, rip_recvspace); 478 if (error) 479 return error; 480 s = splnet(); 481 error = in_pcballoc(so, &ripcbinfo, td); 482 splx(s); 483 if (error) 484 return error; 485 inp = (struct inpcb *)so->so_pcb; 486 inp->inp_vflag |= INP_IPV4; 487 inp->inp_ip_p = proto; 488 inp->inp_ip_ttl = ip_defttl; 489 return 0; 490} 491 492static int 493rip_detach(struct socket *so) 494{ 495 struct inpcb *inp; 496 497 inp = sotoinpcb(so); 498 if (inp == 0) 499 panic("rip_detach"); 500 if (so == ip_mrouter) 501 ip_mrouter_done(); 502 ip_rsvp_force_done(so); 503 if (so == ip_rsvpd) 504 ip_rsvp_done(); 505 in_pcbdetach(inp); 506 return 0; 507} 508 509static int 510rip_abort(struct socket *so) 511{ 512 soisdisconnected(so); 513 return rip_detach(so); 514} 515 516static int 517rip_disconnect(struct socket *so) 518{ 519 if ((so->so_state & SS_ISCONNECTED) == 0) 520 return ENOTCONN; 521 return rip_abort(so); 522} 523 524static int 525rip_bind(struct socket *so, struct sockaddr *nam, struct thread *td) 526{ 527 struct inpcb *inp = sotoinpcb(so); 528 struct sockaddr_in *addr = (struct sockaddr_in *)nam; 529 530 if (nam->sa_len != sizeof(*addr)) 531 return EINVAL; 532 533 if (TAILQ_EMPTY(&ifnet) || ((addr->sin_family != AF_INET) && 534 (addr->sin_family != AF_IMPLINK)) || 535 (addr->sin_addr.s_addr && 536 ifa_ifwithaddr((struct sockaddr *)addr) == 0)) 537 return EADDRNOTAVAIL; 538 inp->inp_laddr = addr->sin_addr; 539 return 0; 540} 541 542static int 543rip_connect(struct socket *so, struct sockaddr *nam, struct thread *td) 544{ 545 struct inpcb *inp = sotoinpcb(so); 546 struct sockaddr_in *addr = (struct sockaddr_in *)nam; 547 548 if (nam->sa_len != sizeof(*addr)) 549 return EINVAL; 550 if (TAILQ_EMPTY(&ifnet)) 551 return EADDRNOTAVAIL; 552 if ((addr->sin_family != AF_INET) && 553 (addr->sin_family != AF_IMPLINK)) 554 return EAFNOSUPPORT; 555 inp->inp_faddr = addr->sin_addr; 556 soisconnected(so); 557 return 0; 558} 559 560static int 561rip_shutdown(struct socket *so) 562{ 563 socantsendmore(so); 564 return 0; 565} 566 567static int 568rip_send(struct socket *so, int flags, struct mbuf *m, struct sockaddr *nam, 569 struct mbuf *control, struct thread *td) 570{ 571 struct inpcb *inp = sotoinpcb(so); 572 register u_long dst; 573 574 if (so->so_state & SS_ISCONNECTED) { 575 if (nam) { 576 m_freem(m); 577 return EISCONN; 578 } 579 dst = inp->inp_faddr.s_addr; 580 } else { 581 if (nam == NULL) { 582 m_freem(m); 583 return ENOTCONN; 584 } 585 dst = ((struct sockaddr_in *)nam)->sin_addr.s_addr; 586 } 587 return rip_output(m, so, dst); 588} 589 590static int 591rip_pcblist(SYSCTL_HANDLER_ARGS) 592{ 593 int error, i, n, s; 594 struct inpcb *inp, **inp_list; 595 inp_gen_t gencnt; 596 struct xinpgen xig; 597 598 /* 599 * The process of preparing the TCB list is too time-consuming and 600 * resource-intensive to repeat twice on every request. 601 */ 602 if (req->oldptr == 0) { 603 n = ripcbinfo.ipi_count; 604 req->oldidx = 2 * (sizeof xig) 605 + (n + n/8) * sizeof(struct xinpcb); 606 return 0; 607 } 608 609 if (req->newptr != 0) 610 return EPERM; 611 612 /* 613 * OK, now we're committed to doing something. 614 */ 615 s = splnet(); 616 gencnt = ripcbinfo.ipi_gencnt; 617 n = ripcbinfo.ipi_count; 618 splx(s); 619 620 xig.xig_len = sizeof xig; 621 xig.xig_count = n; 622 xig.xig_gen = gencnt; 623 xig.xig_sogen = so_gencnt; 624 error = SYSCTL_OUT(req, &xig, sizeof xig); 625 if (error) 626 return error; 627 628 inp_list = malloc(n * sizeof *inp_list, M_TEMP, M_WAITOK); 629 if (inp_list == 0) 630 return ENOMEM; 631 632 s = splnet(); 633 for (inp = LIST_FIRST(ripcbinfo.listhead), i = 0; inp && i < n; 634 inp = LIST_NEXT(inp, inp_list)) { 635 if (inp->inp_gencnt <= gencnt) { 636 if (cr_canseesocket(req->td->td_ucred, 637 inp->inp_socket)) 638 continue; 639 inp_list[i++] = inp; 640 } 641 } 642 splx(s); 643 n = i; 644 645 error = 0; 646 for (i = 0; i < n; i++) { 647 inp = inp_list[i]; 648 if (inp->inp_gencnt <= gencnt) { 649 struct xinpcb xi; 650 xi.xi_len = sizeof xi; 651 /* XXX should avoid extra copy */ 652 bcopy(inp, &xi.xi_inp, sizeof *inp); 653 if (inp->inp_socket) 654 sotoxsocket(inp->inp_socket, &xi.xi_socket); 655 error = SYSCTL_OUT(req, &xi, sizeof xi); 656 } 657 } 658 if (!error) { 659 /* 660 * Give the user an updated idea of our state. 661 * If the generation differs from what we told 662 * her before, she knows that something happened 663 * while we were processing this request, and it 664 * might be necessary to retry. 665 */ 666 s = splnet(); 667 xig.xig_gen = ripcbinfo.ipi_gencnt; 668 xig.xig_sogen = so_gencnt; 669 xig.xig_count = ripcbinfo.ipi_count; 670 splx(s); 671 error = SYSCTL_OUT(req, &xig, sizeof xig); 672 } 673 free(inp_list, M_TEMP); 674 return error; 675} 676 677/* 678 * This is the wrapper function for in_setsockaddr. We just pass down 679 * the pcbinfo for in_setpeeraddr to lock. 680 */ 681static int 682rip_sockaddr(struct socket *so, struct sockaddr **nam) 683{ 684 return (in_setsockaddr(so, nam, &ripcbinfo)); 685} 686 687/* 688 * This is the wrapper function for in_setpeeraddr. We just pass down 689 * the pcbinfo for in_setpeeraddr to lock. 690 */ 691static int 692rip_peeraddr(struct socket *so, struct sockaddr **nam) 693{ 694 return (in_setpeeraddr(so, nam, &ripcbinfo)); 695} 696 697 698SYSCTL_PROC(_net_inet_raw, OID_AUTO/*XXX*/, pcblist, CTLFLAG_RD, 0, 0, 699 rip_pcblist, "S,xinpcb", "List of active raw IP sockets"); 700 701struct pr_usrreqs rip_usrreqs = { 702 rip_abort, pru_accept_notsupp, rip_attach, rip_bind, rip_connect, 703 pru_connect2_notsupp, in_control, rip_detach, rip_disconnect, 704 pru_listen_notsupp, rip_peeraddr, pru_rcvd_notsupp, 705 pru_rcvoob_notsupp, rip_send, pru_sense_null, rip_shutdown, 706 rip_sockaddr, sosend, soreceive, sopoll 707}; 708