1 /*- 2 * Copyright (c) 1982, 1986, 1988, 1993 3 * The Regents of the University of California. 4 * All rights reserved. 5 * 6 * Redistribution and use in source and binary forms, with or without 7 * modification, are permitted provided that the following conditions 8 * are met: 9 * 1. Redistributions of source code must retain the above copyright 10 * notice, this list of conditions and the following disclaimer. 11 * 2. Redistributions in binary form must reproduce the above copyright 12 * notice, this list of conditions and the following disclaimer in the 13 * documentation and/or other materials provided with the distribution. 14 * 4. Neither the name of the University nor the names of its contributors 15 * may be used to endorse or promote products derived from this software 16 * without specific prior written permission. 17 * 18 * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND 19 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE 20 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE 21 * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE 22 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL 23 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS 24 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) 25 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT 26 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY 27 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF 28 * SUCH DAMAGE. 29 * 30 * @(#)raw_ip.c 8.7 (Berkeley) 5/15/95 31 */ 32 33 #include <sys/cdefs.h> 34 __FBSDID("$FreeBSD$"); 35 36 #include "opt_inet6.h" 37 #include "opt_ipsec.h" 38 #include "opt_route.h" 39 40 #include <sys/param.h> 41 #include <sys/jail.h> 42 #include <sys/kernel.h> 43 #include <sys/lock.h> 44 #include <sys/malloc.h> 45 #include <sys/mbuf.h> 46 #include <sys/priv.h> 47 #include <sys/proc.h> 48 #include <sys/protosw.h> 49 #include <sys/rwlock.h> 50 #include <sys/signalvar.h> 51 #include <sys/socket.h> 52 #include <sys/socketvar.h> 53 #include <sys/sx.h> 54 #include <sys/sysctl.h> 55 #include <sys/systm.h> 56 #include <sys/vimage.h> 57 58 #include <vm/uma.h> 59 60 #include <net/if.h> 61 #include <net/route.h> 62 #include <net/vnet.h> 63 64 #include <netinet/in.h> 65 #include <netinet/in_systm.h> 66 #include <netinet/in_pcb.h> 67 #include <netinet/in_var.h> 68 #include <netinet/ip.h> 69 #include <netinet/ip_var.h> 70 #include <netinet/ip_mroute.h> 71 72 #include <netinet/vinet.h> 73 74 #ifdef IPSEC 75 #include <netipsec/ipsec.h> 76 #endif /*IPSEC*/ 77 78 #include <security/mac/mac_framework.h> 79 80 #ifdef VIMAGE_GLOBALS 81 struct inpcbhead ripcb; 82 struct inpcbinfo ripcbinfo; 83 #endif 84 85 /* 86 * Control and data hooks for ipfw and dummynet. 87 * The data hooks are not used here but it is convenient 88 * to keep them all in one place. 89 */ 90 int (*ip_fw_ctl_ptr)(struct sockopt *) = NULL; 91 int (*ip_dn_ctl_ptr)(struct sockopt *) = NULL; 92 int (*ip_fw_chk_ptr)(struct ip_fw_args *args) = NULL; 93 int (*ip_dn_io_ptr)(struct mbuf **m, int dir, struct ip_fw_args *fwa) = NULL; 94 95 /* 96 * Hooks for multicast routing. They all default to NULL, so leave them not 97 * initialized and rely on BSS being set to 0. 98 */ 99 100 /* 101 * The socket used to communicate with the multicast routing daemon. 102 */ 103 #ifdef VIMAGE_GLOBALS 104 struct socket *ip_mrouter; 105 #endif 106 107 /* 108 * The various mrouter and rsvp functions. 109 */ 110 int (*ip_mrouter_set)(struct socket *, struct sockopt *); 111 int (*ip_mrouter_get)(struct socket *, struct sockopt *); 112 int (*ip_mrouter_done)(void); 113 int (*ip_mforward)(struct ip *, struct ifnet *, struct mbuf *, 114 struct ip_moptions *); 115 int (*mrt_ioctl)(int, caddr_t, int); 116 int (*legal_vif_num)(int); 117 u_long (*ip_mcast_src)(int); 118 119 void (*rsvp_input_p)(struct mbuf *m, int off); 120 int (*ip_rsvp_vif)(struct socket *, struct sockopt *); 121 void (*ip_rsvp_force_done)(struct socket *); 122 123 /* 124 * Hash functions 125 */ 126 127 #define INP_PCBHASH_RAW_SIZE 256 128 #define INP_PCBHASH_RAW(proto, laddr, faddr, mask) \ 129 (((proto) + (laddr) + (faddr)) % (mask) + 1) 130 131 static void 132 rip_inshash(struct inpcb *inp) 133 { 134 struct inpcbinfo *pcbinfo = inp->inp_pcbinfo; 135 struct inpcbhead *pcbhash; 136 int hash; 137 138 INP_INFO_WLOCK_ASSERT(pcbinfo); 139 INP_WLOCK_ASSERT(inp); 140 141 if (inp->inp_ip_p != 0 && 142 inp->inp_laddr.s_addr != INADDR_ANY && 143 inp->inp_faddr.s_addr != INADDR_ANY) { 144 hash = INP_PCBHASH_RAW(inp->inp_ip_p, inp->inp_laddr.s_addr, 145 inp->inp_faddr.s_addr, pcbinfo->ipi_hashmask); 146 } else 147 hash = 0; 148 pcbhash = &pcbinfo->ipi_hashbase[hash]; 149 LIST_INSERT_HEAD(pcbhash, inp, inp_hash); 150 } 151 152 static void 153 rip_delhash(struct inpcb *inp) 154 { 155 156 INP_INFO_WLOCK_ASSERT(inp->inp_pcbinfo); 157 INP_WLOCK_ASSERT(inp); 158 159 LIST_REMOVE(inp, inp_hash); 160 } 161 162 /* 163 * Raw interface to IP protocol. 164 */ 165 166 /* 167 * Initialize raw connection block q. 168 */ 169 static void 170 rip_zone_change(void *tag) 171 { 172 INIT_VNET_INET(curvnet); 173 174 uma_zone_set_max(V_ripcbinfo.ipi_zone, maxsockets); 175 } 176 177 static int 178 rip_inpcb_init(void *mem, int size, int flags) 179 { 180 struct inpcb *inp = mem; 181 182 INP_LOCK_INIT(inp, "inp", "rawinp"); 183 return (0); 184 } 185 186 void 187 rip_init(void) 188 { 189 INIT_VNET_INET(curvnet); 190 191 INP_INFO_LOCK_INIT(&V_ripcbinfo, "rip"); 192 LIST_INIT(&V_ripcb); 193 #ifdef VIMAGE 194 V_ripcbinfo.ipi_vnet = curvnet; 195 #endif 196 V_ripcbinfo.ipi_listhead = &V_ripcb; 197 V_ripcbinfo.ipi_hashbase = 198 hashinit(INP_PCBHASH_RAW_SIZE, M_PCB, &V_ripcbinfo.ipi_hashmask); 199 V_ripcbinfo.ipi_porthashbase = 200 hashinit(1, M_PCB, &V_ripcbinfo.ipi_porthashmask); 201 V_ripcbinfo.ipi_zone = uma_zcreate("ripcb", sizeof(struct inpcb), 202 NULL, NULL, rip_inpcb_init, NULL, UMA_ALIGN_PTR, UMA_ZONE_NOFREE); 203 uma_zone_set_max(V_ripcbinfo.ipi_zone, maxsockets); 204 EVENTHANDLER_REGISTER(maxsockets_change, rip_zone_change, NULL, 205 EVENTHANDLER_PRI_ANY); 206 } 207 208 static int 209 rip_append(struct inpcb *last, struct ip *ip, struct mbuf *n, 210 struct sockaddr_in *ripsrc) 211 { 212 int policyfail = 0; 213 214 INP_RLOCK_ASSERT(last); 215 216 #ifdef IPSEC 217 /* check AH/ESP integrity. */ 218 if (ipsec4_in_reject(n, last)) { 219 policyfail = 1; 220 } 221 #endif /* IPSEC */ 222 #ifdef MAC 223 if (!policyfail && mac_inpcb_check_deliver(last, n) != 0) 224 policyfail = 1; 225 #endif 226 /* Check the minimum TTL for socket. */ 227 if (last->inp_ip_minttl && last->inp_ip_minttl > ip->ip_ttl) 228 policyfail = 1; 229 if (!policyfail) { 230 struct mbuf *opts = NULL; 231 struct socket *so; 232 233 so = last->inp_socket; 234 if ((last->inp_flags & INP_CONTROLOPTS) || 235 (so->so_options & (SO_TIMESTAMP | SO_BINTIME))) 236 ip_savecontrol(last, &opts, ip, n); 237 SOCKBUF_LOCK(&so->so_rcv); 238 if (sbappendaddr_locked(&so->so_rcv, 239 (struct sockaddr *)ripsrc, n, opts) == 0) { 240 /* should notify about lost packet */ 241 m_freem(n); 242 if (opts) 243 m_freem(opts); 244 SOCKBUF_UNLOCK(&so->so_rcv); 245 } else 246 sorwakeup_locked(so); 247 } else 248 m_freem(n); 249 return (policyfail); 250 } 251 252 /* 253 * Setup generic address and protocol structures for raw_input routine, then 254 * pass them along with mbuf chain. 255 */ 256 void 257 rip_input(struct mbuf *m, int off) 258 { 259 INIT_VNET_INET(curvnet); 260 struct ifnet *ifp; 261 struct ip *ip = mtod(m, struct ip *); 262 int proto = ip->ip_p; 263 struct inpcb *inp, *last; 264 struct sockaddr_in ripsrc; 265 int hash; 266 267 bzero(&ripsrc, sizeof(ripsrc)); 268 ripsrc.sin_len = sizeof(ripsrc); 269 ripsrc.sin_family = AF_INET; 270 ripsrc.sin_addr = ip->ip_src; 271 last = NULL; 272 273 ifp = m->m_pkthdr.rcvif; 274 275 hash = INP_PCBHASH_RAW(proto, ip->ip_src.s_addr, 276 ip->ip_dst.s_addr, V_ripcbinfo.ipi_hashmask); 277 INP_INFO_RLOCK(&V_ripcbinfo); 278 LIST_FOREACH(inp, &V_ripcbinfo.ipi_hashbase[hash], inp_hash) { 279 if (inp->inp_ip_p != proto) 280 continue; 281 #ifdef INET6 282 /* XXX inp locking */ 283 if ((inp->inp_vflag & INP_IPV4) == 0) 284 continue; 285 #endif 286 if (inp->inp_laddr.s_addr != ip->ip_dst.s_addr) 287 continue; 288 if (inp->inp_faddr.s_addr != ip->ip_src.s_addr) 289 continue; 290 if (jailed(inp->inp_cred)) { 291 /* 292 * XXX: If faddr was bound to multicast group, 293 * jailed raw socket will drop datagram. 294 */ 295 if (prison_check_ip4(inp->inp_cred, &ip->ip_dst) != 0) 296 continue; 297 } 298 if (last != NULL) { 299 struct mbuf *n; 300 301 n = m_copy(m, 0, (int)M_COPYALL); 302 if (n != NULL) 303 (void) rip_append(last, ip, n, &ripsrc); 304 /* XXX count dropped packet */ 305 INP_RUNLOCK(last); 306 } 307 INP_RLOCK(inp); 308 last = inp; 309 } 310 LIST_FOREACH(inp, &V_ripcbinfo.ipi_hashbase[0], inp_hash) { 311 if (inp->inp_ip_p && inp->inp_ip_p != proto) 312 continue; 313 #ifdef INET6 314 /* XXX inp locking */ 315 if ((inp->inp_vflag & INP_IPV4) == 0) 316 continue; 317 #endif 318 if (!in_nullhost(inp->inp_laddr) && 319 !in_hosteq(inp->inp_laddr, ip->ip_dst)) 320 continue; 321 if (!in_nullhost(inp->inp_faddr) && 322 !in_hosteq(inp->inp_faddr, ip->ip_src)) 323 continue; 324 if (jailed(inp->inp_cred)) { 325 /* 326 * Allow raw socket in jail to receive multicast; 327 * assume process had PRIV_NETINET_RAW at attach, 328 * and fall through into normal filter path if so. 329 */ 330 if (!IN_MULTICAST(ntohl(ip->ip_dst.s_addr)) && 331 prison_check_ip4(inp->inp_cred, &ip->ip_dst) != 0) 332 continue; 333 } 334 /* 335 * If this raw socket has multicast state, and we 336 * have received a multicast, check if this socket 337 * should receive it, as multicast filtering is now 338 * the responsibility of the transport layer. 339 */ 340 if (inp->inp_moptions != NULL && 341 IN_MULTICAST(ntohl(ip->ip_dst.s_addr))) { 342 struct sockaddr_in group; 343 int blocked; 344 345 bzero(&group, sizeof(struct sockaddr_in)); 346 group.sin_len = sizeof(struct sockaddr_in); 347 group.sin_family = AF_INET; 348 group.sin_addr = ip->ip_dst; 349 350 blocked = imo_multi_filter(inp->inp_moptions, ifp, 351 (struct sockaddr *)&group, 352 (struct sockaddr *)&ripsrc); 353 if (blocked != MCAST_PASS) { 354 IPSTAT_INC(ips_notmember); 355 continue; 356 } 357 } 358 if (last != NULL) { 359 struct mbuf *n; 360 361 n = m_copy(m, 0, (int)M_COPYALL); 362 if (n != NULL) 363 (void) rip_append(last, ip, n, &ripsrc); 364 /* XXX count dropped packet */ 365 INP_RUNLOCK(last); 366 } 367 INP_RLOCK(inp); 368 last = inp; 369 } 370 INP_INFO_RUNLOCK(&V_ripcbinfo); 371 if (last != NULL) { 372 if (rip_append(last, ip, m, &ripsrc) != 0) 373 IPSTAT_INC(ips_delivered); 374 INP_RUNLOCK(last); 375 } else { 376 m_freem(m); 377 IPSTAT_INC(ips_noproto); 378 IPSTAT_DEC(ips_delivered); 379 } 380 } 381 382 /* 383 * Generate IP header and pass packet to ip_output. Tack on options user may 384 * have setup with control call. 385 */ 386 int 387 rip_output(struct mbuf *m, struct socket *so, u_long dst) 388 { 389 INIT_VNET_INET(so->so_vnet); 390 struct ip *ip; 391 int error; 392 struct inpcb *inp = sotoinpcb(so); 393 int flags = ((so->so_options & SO_DONTROUTE) ? IP_ROUTETOIF : 0) | 394 IP_ALLOWBROADCAST; 395 396 /* 397 * If the user handed us a complete IP packet, use it. Otherwise, 398 * allocate an mbuf for a header and fill it in. 399 */ 400 if ((inp->inp_flags & INP_HDRINCL) == 0) { 401 if (m->m_pkthdr.len + sizeof(struct ip) > IP_MAXPACKET) { 402 m_freem(m); 403 return(EMSGSIZE); 404 } 405 M_PREPEND(m, sizeof(struct ip), M_DONTWAIT); 406 if (m == NULL) 407 return(ENOBUFS); 408 409 INP_RLOCK(inp); 410 ip = mtod(m, struct ip *); 411 ip->ip_tos = inp->inp_ip_tos; 412 if (inp->inp_flags & INP_DONTFRAG) 413 ip->ip_off = IP_DF; 414 else 415 ip->ip_off = 0; 416 ip->ip_p = inp->inp_ip_p; 417 ip->ip_len = m->m_pkthdr.len; 418 ip->ip_src = inp->inp_laddr; 419 error = prison_get_ip4(inp->inp_cred, &ip->ip_src); 420 if (error != 0) { 421 INP_RUNLOCK(inp); 422 m_freem(m); 423 return (error); 424 } 425 ip->ip_dst.s_addr = dst; 426 ip->ip_ttl = inp->inp_ip_ttl; 427 } else { 428 if (m->m_pkthdr.len > IP_MAXPACKET) { 429 m_freem(m); 430 return(EMSGSIZE); 431 } 432 INP_RLOCK(inp); 433 ip = mtod(m, struct ip *); 434 error = prison_check_ip4(inp->inp_cred, &ip->ip_src); 435 if (error != 0) { 436 INP_RUNLOCK(inp); 437 m_freem(m); 438 return (error); 439 } 440 441 /* 442 * Don't allow both user specified and setsockopt options, 443 * and don't allow packet length sizes that will crash. 444 */ 445 if (((ip->ip_hl != (sizeof (*ip) >> 2)) && inp->inp_options) 446 || (ip->ip_len > m->m_pkthdr.len) 447 || (ip->ip_len < (ip->ip_hl << 2))) { 448 INP_RUNLOCK(inp); 449 m_freem(m); 450 return (EINVAL); 451 } 452 if (ip->ip_id == 0) 453 ip->ip_id = ip_newid(); 454 455 /* 456 * XXX prevent ip_output from overwriting header fields. 457 */ 458 flags |= IP_RAWOUTPUT; 459 IPSTAT_INC(ips_rawout); 460 } 461 462 if (inp->inp_flags & INP_ONESBCAST) 463 flags |= IP_SENDONES; 464 465 #ifdef MAC 466 mac_inpcb_create_mbuf(inp, m); 467 #endif 468 469 error = ip_output(m, inp->inp_options, NULL, flags, 470 inp->inp_moptions, inp); 471 INP_RUNLOCK(inp); 472 return (error); 473 } 474 475 /* 476 * Raw IP socket option processing. 477 * 478 * IMPORTANT NOTE regarding access control: Traditionally, raw sockets could 479 * only be created by a privileged process, and as such, socket option 480 * operations to manage system properties on any raw socket were allowed to 481 * take place without explicit additional access control checks. However, 482 * raw sockets can now also be created in jail(), and therefore explicit 483 * checks are now required. Likewise, raw sockets can be used by a process 484 * after it gives up privilege, so some caution is required. For options 485 * passed down to the IP layer via ip_ctloutput(), checks are assumed to be 486 * performed in ip_ctloutput() and therefore no check occurs here. 487 * Unilaterally checking priv_check() here breaks normal IP socket option 488 * operations on raw sockets. 489 * 490 * When adding new socket options here, make sure to add access control 491 * checks here as necessary. 492 */ 493 int 494 rip_ctloutput(struct socket *so, struct sockopt *sopt) 495 { 496 struct inpcb *inp = sotoinpcb(so); 497 int error, optval; 498 499 if (sopt->sopt_level != IPPROTO_IP) { 500 if ((sopt->sopt_level == SOL_SOCKET) && 501 (sopt->sopt_name == SO_SETFIB)) { 502 inp->inp_inc.inc_fibnum = so->so_fibnum; 503 return (0); 504 } 505 return (EINVAL); 506 } 507 508 error = 0; 509 switch (sopt->sopt_dir) { 510 case SOPT_GET: 511 switch (sopt->sopt_name) { 512 case IP_HDRINCL: 513 optval = inp->inp_flags & INP_HDRINCL; 514 error = sooptcopyout(sopt, &optval, sizeof optval); 515 break; 516 517 case IP_FW_ADD: /* ADD actually returns the body... */ 518 case IP_FW_GET: 519 case IP_FW_TABLE_GETSIZE: 520 case IP_FW_TABLE_LIST: 521 case IP_FW_NAT_GET_CONFIG: 522 case IP_FW_NAT_GET_LOG: 523 if (ip_fw_ctl_ptr != NULL) 524 error = ip_fw_ctl_ptr(sopt); 525 else 526 error = ENOPROTOOPT; 527 break; 528 529 case IP_DUMMYNET_GET: 530 if (ip_dn_ctl_ptr != NULL) 531 error = ip_dn_ctl_ptr(sopt); 532 else 533 error = ENOPROTOOPT; 534 break ; 535 536 case MRT_INIT: 537 case MRT_DONE: 538 case MRT_ADD_VIF: 539 case MRT_DEL_VIF: 540 case MRT_ADD_MFC: 541 case MRT_DEL_MFC: 542 case MRT_VERSION: 543 case MRT_ASSERT: 544 case MRT_API_SUPPORT: 545 case MRT_API_CONFIG: 546 case MRT_ADD_BW_UPCALL: 547 case MRT_DEL_BW_UPCALL: 548 error = priv_check(curthread, PRIV_NETINET_MROUTE); 549 if (error != 0) 550 return (error); 551 error = ip_mrouter_get ? ip_mrouter_get(so, sopt) : 552 EOPNOTSUPP; 553 break; 554 555 default: 556 error = ip_ctloutput(so, sopt); 557 break; 558 } 559 break; 560 561 case SOPT_SET: 562 switch (sopt->sopt_name) { 563 case IP_HDRINCL: 564 error = sooptcopyin(sopt, &optval, sizeof optval, 565 sizeof optval); 566 if (error) 567 break; 568 if (optval) 569 inp->inp_flags |= INP_HDRINCL; 570 else 571 inp->inp_flags &= ~INP_HDRINCL; 572 break; 573 574 case IP_FW_ADD: 575 case IP_FW_DEL: 576 case IP_FW_FLUSH: 577 case IP_FW_ZERO: 578 case IP_FW_RESETLOG: 579 case IP_FW_TABLE_ADD: 580 case IP_FW_TABLE_DEL: 581 case IP_FW_TABLE_FLUSH: 582 case IP_FW_NAT_CFG: 583 case IP_FW_NAT_DEL: 584 if (ip_fw_ctl_ptr != NULL) 585 error = ip_fw_ctl_ptr(sopt); 586 else 587 error = ENOPROTOOPT; 588 break; 589 590 case IP_DUMMYNET_CONFIGURE: 591 case IP_DUMMYNET_DEL: 592 case IP_DUMMYNET_FLUSH: 593 if (ip_dn_ctl_ptr != NULL) 594 error = ip_dn_ctl_ptr(sopt); 595 else 596 error = ENOPROTOOPT ; 597 break ; 598 599 case IP_RSVP_ON: 600 error = priv_check(curthread, PRIV_NETINET_MROUTE); 601 if (error != 0) 602 return (error); 603 error = ip_rsvp_init(so); 604 break; 605 606 case IP_RSVP_OFF: 607 error = priv_check(curthread, PRIV_NETINET_MROUTE); 608 if (error != 0) 609 return (error); 610 error = ip_rsvp_done(); 611 break; 612 613 case IP_RSVP_VIF_ON: 614 case IP_RSVP_VIF_OFF: 615 error = priv_check(curthread, PRIV_NETINET_MROUTE); 616 if (error != 0) 617 return (error); 618 error = ip_rsvp_vif ? 619 ip_rsvp_vif(so, sopt) : EINVAL; 620 break; 621 622 case MRT_INIT: 623 case MRT_DONE: 624 case MRT_ADD_VIF: 625 case MRT_DEL_VIF: 626 case MRT_ADD_MFC: 627 case MRT_DEL_MFC: 628 case MRT_VERSION: 629 case MRT_ASSERT: 630 case MRT_API_SUPPORT: 631 case MRT_API_CONFIG: 632 case MRT_ADD_BW_UPCALL: 633 case MRT_DEL_BW_UPCALL: 634 error = priv_check(curthread, PRIV_NETINET_MROUTE); 635 if (error != 0) 636 return (error); 637 error = ip_mrouter_set ? ip_mrouter_set(so, sopt) : 638 EOPNOTSUPP; 639 break; 640 641 default: 642 error = ip_ctloutput(so, sopt); 643 break; 644 } 645 break; 646 } 647 648 return (error); 649 } 650 651 /* 652 * This function exists solely to receive the PRC_IFDOWN messages which are 653 * sent by if_down(). It looks for an ifaddr whose ifa_addr is sa, and calls 654 * in_ifadown() to remove all routes corresponding to that address. It also 655 * receives the PRC_IFUP messages from if_up() and reinstalls the interface 656 * routes. 657 */ 658 void 659 rip_ctlinput(int cmd, struct sockaddr *sa, void *vip) 660 { 661 INIT_VNET_INET(curvnet); 662 struct in_ifaddr *ia; 663 struct ifnet *ifp; 664 int err; 665 int flags; 666 667 switch (cmd) { 668 case PRC_IFDOWN: 669 TAILQ_FOREACH(ia, &V_in_ifaddrhead, ia_link) { 670 if (ia->ia_ifa.ifa_addr == sa 671 && (ia->ia_flags & IFA_ROUTE)) { 672 /* 673 * in_ifscrub kills the interface route. 674 */ 675 in_ifscrub(ia->ia_ifp, ia); 676 /* 677 * in_ifadown gets rid of all the rest of the 678 * routes. This is not quite the right thing 679 * to do, but at least if we are running a 680 * routing process they will come back. 681 */ 682 in_ifadown(&ia->ia_ifa, 0); 683 break; 684 } 685 } 686 break; 687 688 case PRC_IFUP: 689 TAILQ_FOREACH(ia, &V_in_ifaddrhead, ia_link) { 690 if (ia->ia_ifa.ifa_addr == sa) 691 break; 692 } 693 if (ia == 0 || (ia->ia_flags & IFA_ROUTE)) 694 return; 695 flags = RTF_UP; 696 ifp = ia->ia_ifa.ifa_ifp; 697 698 if ((ifp->if_flags & IFF_LOOPBACK) 699 || (ifp->if_flags & IFF_POINTOPOINT)) 700 flags |= RTF_HOST; 701 702 err = rtinit(&ia->ia_ifa, RTM_ADD, flags); 703 if (err == 0) 704 ia->ia_flags |= IFA_ROUTE; 705 break; 706 } 707 } 708 709 u_long rip_sendspace = 9216; 710 u_long rip_recvspace = 9216; 711 712 SYSCTL_ULONG(_net_inet_raw, OID_AUTO, maxdgram, CTLFLAG_RW, 713 &rip_sendspace, 0, "Maximum outgoing raw IP datagram size"); 714 SYSCTL_ULONG(_net_inet_raw, OID_AUTO, recvspace, CTLFLAG_RW, 715 &rip_recvspace, 0, "Maximum space for incoming raw IP datagrams"); 716 717 static int 718 rip_attach(struct socket *so, int proto, struct thread *td) 719 { 720 INIT_VNET_INET(so->so_vnet); 721 struct inpcb *inp; 722 int error; 723 724 inp = sotoinpcb(so); 725 KASSERT(inp == NULL, ("rip_attach: inp != NULL")); 726 727 error = priv_check(td, PRIV_NETINET_RAW); 728 if (error) 729 return (error); 730 if (proto >= IPPROTO_MAX || proto < 0) 731 return EPROTONOSUPPORT; 732 error = soreserve(so, rip_sendspace, rip_recvspace); 733 if (error) 734 return (error); 735 INP_INFO_WLOCK(&V_ripcbinfo); 736 error = in_pcballoc(so, &V_ripcbinfo); 737 if (error) { 738 INP_INFO_WUNLOCK(&V_ripcbinfo); 739 return (error); 740 } 741 inp = (struct inpcb *)so->so_pcb; 742 inp->inp_vflag |= INP_IPV4; 743 inp->inp_ip_p = proto; 744 inp->inp_ip_ttl = V_ip_defttl; 745 rip_inshash(inp); 746 INP_INFO_WUNLOCK(&V_ripcbinfo); 747 INP_WUNLOCK(inp); 748 return (0); 749 } 750 751 static void 752 rip_detach(struct socket *so) 753 { 754 INIT_VNET_INET(so->so_vnet); 755 struct inpcb *inp; 756 757 inp = sotoinpcb(so); 758 KASSERT(inp != NULL, ("rip_detach: inp == NULL")); 759 KASSERT(inp->inp_faddr.s_addr == INADDR_ANY, 760 ("rip_detach: not closed")); 761 762 INP_INFO_WLOCK(&V_ripcbinfo); 763 INP_WLOCK(inp); 764 rip_delhash(inp); 765 if (so == V_ip_mrouter && ip_mrouter_done) 766 ip_mrouter_done(); 767 if (ip_rsvp_force_done) 768 ip_rsvp_force_done(so); 769 if (so == V_ip_rsvpd) 770 ip_rsvp_done(); 771 in_pcbdetach(inp); 772 in_pcbfree(inp); 773 INP_INFO_WUNLOCK(&V_ripcbinfo); 774 } 775 776 static void 777 rip_dodisconnect(struct socket *so, struct inpcb *inp) 778 { 779 780 INP_INFO_WLOCK_ASSERT(inp->inp_pcbinfo); 781 INP_WLOCK_ASSERT(inp); 782 783 rip_delhash(inp); 784 inp->inp_faddr.s_addr = INADDR_ANY; 785 rip_inshash(inp); 786 SOCK_LOCK(so); 787 so->so_state &= ~SS_ISCONNECTED; 788 SOCK_UNLOCK(so); 789 } 790 791 static void 792 rip_abort(struct socket *so) 793 { 794 INIT_VNET_INET(so->so_vnet); 795 struct inpcb *inp; 796 797 inp = sotoinpcb(so); 798 KASSERT(inp != NULL, ("rip_abort: inp == NULL")); 799 800 INP_INFO_WLOCK(&V_ripcbinfo); 801 INP_WLOCK(inp); 802 rip_dodisconnect(so, inp); 803 INP_WUNLOCK(inp); 804 INP_INFO_WUNLOCK(&V_ripcbinfo); 805 } 806 807 static void 808 rip_close(struct socket *so) 809 { 810 INIT_VNET_INET(so->so_vnet); 811 struct inpcb *inp; 812 813 inp = sotoinpcb(so); 814 KASSERT(inp != NULL, ("rip_close: inp == NULL")); 815 816 INP_INFO_WLOCK(&V_ripcbinfo); 817 INP_WLOCK(inp); 818 rip_dodisconnect(so, inp); 819 INP_WUNLOCK(inp); 820 INP_INFO_WUNLOCK(&V_ripcbinfo); 821 } 822 823 static int 824 rip_disconnect(struct socket *so) 825 { 826 INIT_VNET_INET(so->so_vnet); 827 struct inpcb *inp; 828 829 if ((so->so_state & SS_ISCONNECTED) == 0) 830 return (ENOTCONN); 831 832 inp = sotoinpcb(so); 833 KASSERT(inp != NULL, ("rip_disconnect: inp == NULL")); 834 835 INP_INFO_WLOCK(&V_ripcbinfo); 836 INP_WLOCK(inp); 837 rip_dodisconnect(so, inp); 838 INP_WUNLOCK(inp); 839 INP_INFO_WUNLOCK(&V_ripcbinfo); 840 return (0); 841 } 842 843 static int 844 rip_bind(struct socket *so, struct sockaddr *nam, struct thread *td) 845 { 846 INIT_VNET_NET(so->so_vnet); 847 INIT_VNET_INET(so->so_vnet); 848 struct sockaddr_in *addr = (struct sockaddr_in *)nam; 849 struct inpcb *inp; 850 int error; 851 852 if (nam->sa_len != sizeof(*addr)) 853 return (EINVAL); 854 855 error = prison_check_ip4(td->td_ucred, &addr->sin_addr); 856 if (error != 0) 857 return (error); 858 859 inp = sotoinpcb(so); 860 KASSERT(inp != NULL, ("rip_bind: inp == NULL")); 861 862 if (TAILQ_EMPTY(&V_ifnet) || 863 (addr->sin_family != AF_INET && addr->sin_family != AF_IMPLINK) || 864 (addr->sin_addr.s_addr && 865 (inp->inp_flags & INP_BINDANY) == 0 && 866 ifa_ifwithaddr((struct sockaddr *)addr) == NULL)) 867 return (EADDRNOTAVAIL); 868 869 INP_INFO_WLOCK(&V_ripcbinfo); 870 INP_WLOCK(inp); 871 rip_delhash(inp); 872 inp->inp_laddr = addr->sin_addr; 873 rip_inshash(inp); 874 INP_WUNLOCK(inp); 875 INP_INFO_WUNLOCK(&V_ripcbinfo); 876 return (0); 877 } 878 879 static int 880 rip_connect(struct socket *so, struct sockaddr *nam, struct thread *td) 881 { 882 INIT_VNET_NET(so->so_vnet); 883 INIT_VNET_INET(so->so_vnet); 884 struct sockaddr_in *addr = (struct sockaddr_in *)nam; 885 struct inpcb *inp; 886 887 if (nam->sa_len != sizeof(*addr)) 888 return (EINVAL); 889 if (TAILQ_EMPTY(&V_ifnet)) 890 return (EADDRNOTAVAIL); 891 if (addr->sin_family != AF_INET && addr->sin_family != AF_IMPLINK) 892 return (EAFNOSUPPORT); 893 894 inp = sotoinpcb(so); 895 KASSERT(inp != NULL, ("rip_connect: inp == NULL")); 896 897 INP_INFO_WLOCK(&V_ripcbinfo); 898 INP_WLOCK(inp); 899 rip_delhash(inp); 900 inp->inp_faddr = addr->sin_addr; 901 rip_inshash(inp); 902 soisconnected(so); 903 INP_WUNLOCK(inp); 904 INP_INFO_WUNLOCK(&V_ripcbinfo); 905 return (0); 906 } 907 908 static int 909 rip_shutdown(struct socket *so) 910 { 911 struct inpcb *inp; 912 913 inp = sotoinpcb(so); 914 KASSERT(inp != NULL, ("rip_shutdown: inp == NULL")); 915 916 INP_WLOCK(inp); 917 socantsendmore(so); 918 INP_WUNLOCK(inp); 919 return (0); 920 } 921 922 static int 923 rip_send(struct socket *so, int flags, struct mbuf *m, struct sockaddr *nam, 924 struct mbuf *control, struct thread *td) 925 { 926 struct inpcb *inp; 927 u_long dst; 928 929 inp = sotoinpcb(so); 930 KASSERT(inp != NULL, ("rip_send: inp == NULL")); 931 932 /* 933 * Note: 'dst' reads below are unlocked. 934 */ 935 if (so->so_state & SS_ISCONNECTED) { 936 if (nam) { 937 m_freem(m); 938 return (EISCONN); 939 } 940 dst = inp->inp_faddr.s_addr; /* Unlocked read. */ 941 } else { 942 if (nam == NULL) { 943 m_freem(m); 944 return (ENOTCONN); 945 } 946 dst = ((struct sockaddr_in *)nam)->sin_addr.s_addr; 947 } 948 return (rip_output(m, so, dst)); 949 } 950 951 static int 952 rip_pcblist(SYSCTL_HANDLER_ARGS) 953 { 954 INIT_VNET_INET(curvnet); 955 int error, i, n; 956 struct inpcb *inp, **inp_list; 957 inp_gen_t gencnt; 958 struct xinpgen xig; 959 960 /* 961 * The process of preparing the TCB list is too time-consuming and 962 * resource-intensive to repeat twice on every request. 963 */ 964 if (req->oldptr == 0) { 965 n = V_ripcbinfo.ipi_count; 966 req->oldidx = 2 * (sizeof xig) 967 + (n + n/8) * sizeof(struct xinpcb); 968 return (0); 969 } 970 971 if (req->newptr != 0) 972 return (EPERM); 973 974 /* 975 * OK, now we're committed to doing something. 976 */ 977 INP_INFO_RLOCK(&V_ripcbinfo); 978 gencnt = V_ripcbinfo.ipi_gencnt; 979 n = V_ripcbinfo.ipi_count; 980 INP_INFO_RUNLOCK(&V_ripcbinfo); 981 982 xig.xig_len = sizeof xig; 983 xig.xig_count = n; 984 xig.xig_gen = gencnt; 985 xig.xig_sogen = so_gencnt; 986 error = SYSCTL_OUT(req, &xig, sizeof xig); 987 if (error) 988 return (error); 989 990 inp_list = malloc(n * sizeof *inp_list, M_TEMP, M_WAITOK); 991 if (inp_list == 0) 992 return (ENOMEM); 993 994 INP_INFO_RLOCK(&V_ripcbinfo); 995 for (inp = LIST_FIRST(V_ripcbinfo.ipi_listhead), i = 0; inp && i < n; 996 inp = LIST_NEXT(inp, inp_list)) { 997 INP_RLOCK(inp); 998 if (inp->inp_gencnt <= gencnt && 999 cr_canseeinpcb(req->td->td_ucred, inp) == 0) { 1000 /* XXX held references? */ 1001 inp_list[i++] = inp; 1002 } 1003 INP_RUNLOCK(inp); 1004 } 1005 INP_INFO_RUNLOCK(&V_ripcbinfo); 1006 n = i; 1007 1008 error = 0; 1009 for (i = 0; i < n; i++) { 1010 inp = inp_list[i]; 1011 INP_RLOCK(inp); 1012 if (inp->inp_gencnt <= gencnt) { 1013 struct xinpcb xi; 1014 1015 bzero(&xi, sizeof(xi)); 1016 xi.xi_len = sizeof xi; 1017 /* XXX should avoid extra copy */ 1018 bcopy(inp, &xi.xi_inp, sizeof *inp); 1019 if (inp->inp_socket) 1020 sotoxsocket(inp->inp_socket, &xi.xi_socket); 1021 INP_RUNLOCK(inp); 1022 error = SYSCTL_OUT(req, &xi, sizeof xi); 1023 } else 1024 INP_RUNLOCK(inp); 1025 } 1026 if (!error) { 1027 /* 1028 * Give the user an updated idea of our state. If the 1029 * generation differs from what we told her before, she knows 1030 * that something happened while we were processing this 1031 * request, and it might be necessary to retry. 1032 */ 1033 INP_INFO_RLOCK(&V_ripcbinfo); 1034 xig.xig_gen = V_ripcbinfo.ipi_gencnt; 1035 xig.xig_sogen = so_gencnt; 1036 xig.xig_count = V_ripcbinfo.ipi_count; 1037 INP_INFO_RUNLOCK(&V_ripcbinfo); 1038 error = SYSCTL_OUT(req, &xig, sizeof xig); 1039 } 1040 free(inp_list, M_TEMP); 1041 return (error); 1042 } 1043 1044 SYSCTL_PROC(_net_inet_raw, OID_AUTO/*XXX*/, pcblist, CTLFLAG_RD, 0, 0, 1045 rip_pcblist, "S,xinpcb", "List of active raw IP sockets"); 1046 1047 struct pr_usrreqs rip_usrreqs = { 1048 .pru_abort = rip_abort, 1049 .pru_attach = rip_attach, 1050 .pru_bind = rip_bind, 1051 .pru_connect = rip_connect, 1052 .pru_control = in_control, 1053 .pru_detach = rip_detach, 1054 .pru_disconnect = rip_disconnect, 1055 .pru_peeraddr = in_getpeeraddr, 1056 .pru_send = rip_send, 1057 .pru_shutdown = rip_shutdown, 1058 .pru_sockaddr = in_getsockaddr, 1059 .pru_sosetlabel = in_pcbsosetlabel, 1060 .pru_close = rip_close, 1061 }; 1062