raw_ip.c revision 134793
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 * 4. Neither the name of the University nor the names of its contributors 14 * may be used to endorse or promote products derived from this software 15 * without specific prior written permission. 16 * 17 * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND 18 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE 19 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE 20 * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE 21 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL 22 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS 23 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) 24 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT 25 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY 26 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF 27 * SUCH DAMAGE. 28 * 29 * @(#)raw_ip.c 8.7 (Berkeley) 5/15/95 30 * $FreeBSD: head/sys/netinet/raw_ip.c 134793 2004-09-05 02:34:12Z jmg $ 31 */ 32 33#include "opt_inet6.h" 34#include "opt_ipsec.h" 35#include "opt_mac.h" 36 37#include <sys/param.h> 38#include <sys/jail.h> 39#include <sys/kernel.h> 40#include <sys/lock.h> 41#include <sys/mac.h> 42#include <sys/malloc.h> 43#include <sys/mbuf.h> 44#include <sys/proc.h> 45#include <sys/protosw.h> 46#include <sys/signalvar.h> 47#include <sys/socket.h> 48#include <sys/socketvar.h> 49#include <sys/sx.h> 50#include <sys/sysctl.h> 51#include <sys/systm.h> 52 53#include <vm/uma.h> 54 55#include <net/if.h> 56#include <net/route.h> 57 58#include <netinet/in.h> 59#include <netinet/in_systm.h> 60#include <netinet/in_pcb.h> 61#include <netinet/in_var.h> 62#include <netinet/ip.h> 63#include <netinet/ip_var.h> 64#include <netinet/ip_mroute.h> 65 66#include <netinet/ip_fw.h> 67#include <netinet/ip_dummynet.h> 68 69#ifdef FAST_IPSEC 70#include <netipsec/ipsec.h> 71#endif /*FAST_IPSEC*/ 72 73#ifdef IPSEC 74#include <netinet6/ipsec.h> 75#endif /*IPSEC*/ 76 77struct inpcbhead ripcb; 78struct inpcbinfo ripcbinfo; 79 80/* control hooks for ipfw and dummynet */ 81ip_fw_ctl_t *ip_fw_ctl_ptr = NULL; 82ip_dn_ctl_t *ip_dn_ctl_ptr = NULL; 83 84/* 85 * hooks for multicast routing. They all default to NULL, 86 * so leave them not initialized and rely on BSS being set to 0. 87 */ 88 89/* The socket used to communicate with the multicast routing daemon. */ 90struct socket *ip_mrouter; 91 92/* The various mrouter and rsvp functions */ 93int (*ip_mrouter_set)(struct socket *, struct sockopt *); 94int (*ip_mrouter_get)(struct socket *, struct sockopt *); 95int (*ip_mrouter_done)(void); 96int (*ip_mforward)(struct ip *, struct ifnet *, struct mbuf *, 97 struct ip_moptions *); 98int (*mrt_ioctl)(int, caddr_t); 99int (*legal_vif_num)(int); 100u_long (*ip_mcast_src)(int); 101 102void (*rsvp_input_p)(struct mbuf *m, int off); 103int (*ip_rsvp_vif)(struct socket *, struct sockopt *); 104void (*ip_rsvp_force_done)(struct socket *); 105 106/* 107 * Nominal space allocated to a raw ip socket. 108 */ 109#define RIPSNDQ 8192 110#define RIPRCVQ 8192 111 112/* 113 * Raw interface to IP protocol. 114 */ 115 116/* 117 * Initialize raw connection block q. 118 */ 119void 120rip_init() 121{ 122 INP_INFO_LOCK_INIT(&ripcbinfo, "rip"); 123 LIST_INIT(&ripcb); 124 ripcbinfo.listhead = &ripcb; 125 /* 126 * XXX We don't use the hash list for raw IP, but it's easier 127 * to allocate a one entry hash list than it is to check all 128 * over the place for hashbase == NULL. 129 */ 130 ripcbinfo.hashbase = hashinit(1, M_PCB, &ripcbinfo.hashmask); 131 ripcbinfo.porthashbase = hashinit(1, M_PCB, &ripcbinfo.porthashmask); 132 ripcbinfo.ipi_zone = uma_zcreate("ripcb", sizeof(struct inpcb), 133 NULL, NULL, NULL, NULL, UMA_ALIGN_PTR, UMA_ZONE_NOFREE); 134 uma_zone_set_max(ripcbinfo.ipi_zone, maxsockets); 135} 136 137static struct sockaddr_in ripsrc = { sizeof(ripsrc), AF_INET }; 138 139static int 140raw_append(struct inpcb *last, struct ip *ip, struct mbuf *n) 141{ 142 int policyfail = 0; 143 144 INP_LOCK_ASSERT(last); 145 146#if defined(IPSEC) || defined(FAST_IPSEC) 147 /* check AH/ESP integrity. */ 148 if (ipsec4_in_reject(n, last)) { 149 policyfail = 1; 150#ifdef IPSEC 151 ipsecstat.in_polvio++; 152#endif /*IPSEC*/ 153 /* do not inject data to pcb */ 154 } 155#endif /*IPSEC || FAST_IPSEC*/ 156#ifdef MAC 157 if (!policyfail && mac_check_inpcb_deliver(last, n) != 0) 158 policyfail = 1; 159#endif 160 if (!policyfail) { 161 struct mbuf *opts = NULL; 162 struct socket *so; 163 164 so = last->inp_socket; 165 if ((last->inp_flags & INP_CONTROLOPTS) || 166 (so->so_options & SO_TIMESTAMP)) 167 ip_savecontrol(last, &opts, ip, n); 168 SOCKBUF_LOCK(&so->so_rcv); 169 if (sbappendaddr_locked(&so->so_rcv, 170 (struct sockaddr *)&ripsrc, n, opts) == 0) { 171 /* should notify about lost packet */ 172 m_freem(n); 173 if (opts) 174 m_freem(opts); 175 SOCKBUF_UNLOCK(&so->so_rcv); 176 } else 177 sorwakeup_locked(so); 178 } else 179 m_freem(n); 180 return policyfail; 181} 182 183/* 184 * Setup generic address and protocol structures 185 * for raw_input routine, then pass them along with 186 * mbuf chain. 187 */ 188void 189rip_input(struct mbuf *m, int off) 190{ 191 struct ip *ip = mtod(m, struct ip *); 192 int proto = ip->ip_p; 193 struct inpcb *inp, *last; 194 195 INP_INFO_RLOCK(&ripcbinfo); 196 ripsrc.sin_addr = ip->ip_src; 197 last = NULL; 198 LIST_FOREACH(inp, &ripcb, inp_list) { 199 INP_LOCK(inp); 200 if (inp->inp_ip_p && inp->inp_ip_p != proto) { 201 docontinue: 202 INP_UNLOCK(inp); 203 continue; 204 } 205#ifdef INET6 206 if ((inp->inp_vflag & INP_IPV4) == 0) 207 goto docontinue; 208#endif 209 if (inp->inp_laddr.s_addr && 210 inp->inp_laddr.s_addr != ip->ip_dst.s_addr) 211 goto docontinue; 212 if (inp->inp_faddr.s_addr && 213 inp->inp_faddr.s_addr != ip->ip_src.s_addr) 214 goto docontinue; 215 if (jailed(inp->inp_socket->so_cred)) 216 if (htonl(prison_getip(inp->inp_socket->so_cred)) != 217 ip->ip_dst.s_addr) 218 goto docontinue; 219 if (last) { 220 struct mbuf *n; 221 222 n = m_copy(m, 0, (int)M_COPYALL); 223 if (n != NULL) 224 (void) raw_append(last, ip, n); 225 /* XXX count dropped packet */ 226 INP_UNLOCK(last); 227 } 228 last = inp; 229 } 230 if (last != NULL) { 231 if (raw_append(last, ip, m) != 0) 232 ipstat.ips_delivered--; 233 INP_UNLOCK(last); 234 } else { 235 m_freem(m); 236 ipstat.ips_noproto++; 237 ipstat.ips_delivered--; 238 } 239 INP_INFO_RUNLOCK(&ripcbinfo); 240} 241 242/* 243 * Generate IP header and pass packet to ip_output. 244 * Tack on options user may have setup with control call. 245 */ 246int 247rip_output(struct mbuf *m, struct socket *so, u_long dst) 248{ 249 struct ip *ip; 250 int error; 251 struct inpcb *inp = sotoinpcb(so); 252 int flags = ((so->so_options & SO_DONTROUTE) ? IP_ROUTETOIF : 0) | 253 IP_ALLOWBROADCAST; 254 255 /* 256 * If the user handed us a complete IP packet, use it. 257 * Otherwise, allocate an mbuf for a header and fill it in. 258 */ 259 if ((inp->inp_flags & INP_HDRINCL) == 0) { 260 if (m->m_pkthdr.len + sizeof(struct ip) > IP_MAXPACKET) { 261 m_freem(m); 262 return(EMSGSIZE); 263 } 264 M_PREPEND(m, sizeof(struct ip), M_DONTWAIT); 265 if (m == NULL) 266 return(ENOBUFS); 267 268 INP_LOCK(inp); 269 ip = mtod(m, struct ip *); 270 ip->ip_tos = inp->inp_ip_tos; 271 ip->ip_off = 0; 272 ip->ip_p = inp->inp_ip_p; 273 ip->ip_len = m->m_pkthdr.len; 274 if (jailed(inp->inp_socket->so_cred)) 275 ip->ip_src.s_addr = 276 htonl(prison_getip(inp->inp_socket->so_cred)); 277 else 278 ip->ip_src = inp->inp_laddr; 279 ip->ip_dst.s_addr = dst; 280 ip->ip_ttl = inp->inp_ip_ttl; 281 } else { 282 if (m->m_pkthdr.len > IP_MAXPACKET) { 283 m_freem(m); 284 return(EMSGSIZE); 285 } 286 INP_LOCK(inp); 287 ip = mtod(m, struct ip *); 288 if (jailed(inp->inp_socket->so_cred)) { 289 if (ip->ip_src.s_addr != 290 htonl(prison_getip(inp->inp_socket->so_cred))) { 291 INP_UNLOCK(inp); 292 m_freem(m); 293 return (EPERM); 294 } 295 } 296 /* don't allow both user specified and setsockopt options, 297 and don't allow packet length sizes that will crash */ 298 if (((ip->ip_hl != (sizeof (*ip) >> 2)) 299 && inp->inp_options) 300 || (ip->ip_len > m->m_pkthdr.len) 301 || (ip->ip_len < (ip->ip_hl << 2))) { 302 INP_UNLOCK(inp); 303 m_freem(m); 304 return EINVAL; 305 } 306 if (ip->ip_id == 0) 307 ip->ip_id = ip_newid(); 308 /* XXX prevent ip_output from overwriting header fields */ 309 flags |= IP_RAWOUTPUT; 310 ipstat.ips_rawout++; 311 } 312 313 if (inp->inp_flags & INP_ONESBCAST) 314 flags |= IP_SENDONES; 315 316#ifdef MAC 317 mac_create_mbuf_from_inpcb(inp, m); 318#endif 319 320 error = ip_output(m, inp->inp_options, NULL, flags, 321 inp->inp_moptions, inp); 322 INP_UNLOCK(inp); 323 return error; 324} 325 326/* 327 * Raw IP socket option processing. 328 * 329 * Note that access to all of the IP administrative functions here is 330 * implicitly protected by suser() as gaining access to a raw socket 331 * requires either that the thread pass a suser() check, or that it be 332 * passed a raw socket by another thread that has passed a suser() check. 333 * If FreeBSD moves to a more fine-grained access control mechanism, 334 * additional checks will need to be placed here if the raw IP attachment 335 * check is not equivilent the the check required for these 336 * administrative operations; in some cases, these checks are already 337 * present. 338 */ 339int 340rip_ctloutput(struct socket *so, struct sockopt *sopt) 341{ 342 struct inpcb *inp = sotoinpcb(so); 343 int error, optval; 344 345 if (sopt->sopt_level != IPPROTO_IP) 346 return (EINVAL); 347 348 /* 349 * Even though super-user is required to create a raw socket, the 350 * calling cred could be prison root. If so we want to restrict the 351 * access to IP_HDRINCL only. 352 */ 353 if (sopt->sopt_name != IP_HDRINCL) { 354 error = suser(curthread); 355 if (error != 0) 356 return (error); 357 } 358 error = 0; 359 360 switch (sopt->sopt_dir) { 361 case SOPT_GET: 362 switch (sopt->sopt_name) { 363 case IP_HDRINCL: 364 optval = inp->inp_flags & INP_HDRINCL; 365 error = sooptcopyout(sopt, &optval, sizeof optval); 366 break; 367 368 case IP_FW_ADD: /* ADD actually returns the body... */ 369 case IP_FW_GET: 370 case IP_FW_TABLE_GETSIZE: 371 case IP_FW_TABLE_LIST: 372 if (ip_fw_ctl_ptr != NULL) 373 error = ip_fw_ctl_ptr(sopt); 374 else 375 error = ENOPROTOOPT; 376 break; 377 378 case IP_DUMMYNET_GET: 379 if (ip_dn_ctl_ptr != NULL) 380 error = ip_dn_ctl_ptr(sopt); 381 else 382 error = ENOPROTOOPT; 383 break ; 384 385 case MRT_INIT: 386 case MRT_DONE: 387 case MRT_ADD_VIF: 388 case MRT_DEL_VIF: 389 case MRT_ADD_MFC: 390 case MRT_DEL_MFC: 391 case MRT_VERSION: 392 case MRT_ASSERT: 393 case MRT_API_SUPPORT: 394 case MRT_API_CONFIG: 395 case MRT_ADD_BW_UPCALL: 396 case MRT_DEL_BW_UPCALL: 397 error = ip_mrouter_get ? ip_mrouter_get(so, sopt) : 398 EOPNOTSUPP; 399 break; 400 401 default: 402 error = ip_ctloutput(so, sopt); 403 break; 404 } 405 break; 406 407 case SOPT_SET: 408 switch (sopt->sopt_name) { 409 case IP_HDRINCL: 410 error = sooptcopyin(sopt, &optval, sizeof optval, 411 sizeof optval); 412 if (error) 413 break; 414 if (optval) 415 inp->inp_flags |= INP_HDRINCL; 416 else 417 inp->inp_flags &= ~INP_HDRINCL; 418 break; 419 420 case IP_FW_ADD: 421 case IP_FW_DEL: 422 case IP_FW_FLUSH: 423 case IP_FW_ZERO: 424 case IP_FW_RESETLOG: 425 case IP_FW_TABLE_ADD: 426 case IP_FW_TABLE_DEL: 427 case IP_FW_TABLE_FLUSH: 428 if (ip_fw_ctl_ptr != NULL) 429 error = ip_fw_ctl_ptr(sopt); 430 else 431 error = ENOPROTOOPT; 432 break; 433 434 case IP_DUMMYNET_CONFIGURE: 435 case IP_DUMMYNET_DEL: 436 case IP_DUMMYNET_FLUSH: 437 if (ip_dn_ctl_ptr != NULL) 438 error = ip_dn_ctl_ptr(sopt); 439 else 440 error = ENOPROTOOPT ; 441 break ; 442 443 case IP_RSVP_ON: 444 error = ip_rsvp_init(so); 445 break; 446 447 case IP_RSVP_OFF: 448 error = ip_rsvp_done(); 449 break; 450 451 case IP_RSVP_VIF_ON: 452 case IP_RSVP_VIF_OFF: 453 error = ip_rsvp_vif ? 454 ip_rsvp_vif(so, sopt) : EINVAL; 455 break; 456 457 case MRT_INIT: 458 case MRT_DONE: 459 case MRT_ADD_VIF: 460 case MRT_DEL_VIF: 461 case MRT_ADD_MFC: 462 case MRT_DEL_MFC: 463 case MRT_VERSION: 464 case MRT_ASSERT: 465 case MRT_API_SUPPORT: 466 case MRT_API_CONFIG: 467 case MRT_ADD_BW_UPCALL: 468 case MRT_DEL_BW_UPCALL: 469 error = ip_mrouter_set ? ip_mrouter_set(so, sopt) : 470 EOPNOTSUPP; 471 break; 472 473 default: 474 error = ip_ctloutput(so, sopt); 475 break; 476 } 477 break; 478 } 479 480 return (error); 481} 482 483/* 484 * This function exists solely to receive the PRC_IFDOWN messages which 485 * are sent by if_down(). It looks for an ifaddr whose ifa_addr is sa, 486 * and calls in_ifadown() to remove all routes corresponding to that address. 487 * It also receives the PRC_IFUP messages from if_up() and reinstalls the 488 * interface routes. 489 */ 490void 491rip_ctlinput(int cmd, struct sockaddr *sa, void *vip) 492{ 493 struct in_ifaddr *ia; 494 struct ifnet *ifp; 495 int err; 496 int flags; 497 498 switch (cmd) { 499 case PRC_IFDOWN: 500 TAILQ_FOREACH(ia, &in_ifaddrhead, ia_link) { 501 if (ia->ia_ifa.ifa_addr == sa 502 && (ia->ia_flags & IFA_ROUTE)) { 503 /* 504 * in_ifscrub kills the interface route. 505 */ 506 in_ifscrub(ia->ia_ifp, ia); 507 /* 508 * in_ifadown gets rid of all the rest of 509 * the routes. This is not quite the right 510 * thing to do, but at least if we are running 511 * a routing process they will come back. 512 */ 513 in_ifadown(&ia->ia_ifa, 0); 514 break; 515 } 516 } 517 break; 518 519 case PRC_IFUP: 520 TAILQ_FOREACH(ia, &in_ifaddrhead, ia_link) { 521 if (ia->ia_ifa.ifa_addr == sa) 522 break; 523 } 524 if (ia == 0 || (ia->ia_flags & IFA_ROUTE)) 525 return; 526 flags = RTF_UP; 527 ifp = ia->ia_ifa.ifa_ifp; 528 529 if ((ifp->if_flags & IFF_LOOPBACK) 530 || (ifp->if_flags & IFF_POINTOPOINT)) 531 flags |= RTF_HOST; 532 533 err = rtinit(&ia->ia_ifa, RTM_ADD, flags); 534 if (err == 0) 535 ia->ia_flags |= IFA_ROUTE; 536 break; 537 } 538} 539 540u_long rip_sendspace = RIPSNDQ; 541u_long rip_recvspace = RIPRCVQ; 542 543SYSCTL_INT(_net_inet_raw, OID_AUTO, maxdgram, CTLFLAG_RW, 544 &rip_sendspace, 0, "Maximum outgoing raw IP datagram size"); 545SYSCTL_INT(_net_inet_raw, OID_AUTO, recvspace, CTLFLAG_RW, 546 &rip_recvspace, 0, "Maximum space for incoming raw IP datagrams"); 547 548static int 549rip_attach(struct socket *so, int proto, struct thread *td) 550{ 551 struct inpcb *inp; 552 int error; 553 554 /* XXX why not lower? */ 555 INP_INFO_WLOCK(&ripcbinfo); 556 inp = sotoinpcb(so); 557 if (inp) { 558 /* XXX counter, printf */ 559 INP_INFO_WUNLOCK(&ripcbinfo); 560 return EINVAL; 561 } 562 if (td && jailed(td->td_ucred) && !jail_allow_raw_sockets) { 563 INP_INFO_WUNLOCK(&ripcbinfo); 564 return (EPERM); 565 } 566 if (td && (error = suser_cred(td->td_ucred, SUSER_ALLOWJAIL)) != 0) { 567 INP_INFO_WUNLOCK(&ripcbinfo); 568 return error; 569 } 570 if (proto >= IPPROTO_MAX || proto < 0) { 571 INP_INFO_WUNLOCK(&ripcbinfo); 572 return EPROTONOSUPPORT; 573 } 574 575 error = soreserve(so, rip_sendspace, rip_recvspace); 576 if (error) { 577 INP_INFO_WUNLOCK(&ripcbinfo); 578 return error; 579 } 580 error = in_pcballoc(so, &ripcbinfo, "rawinp"); 581 if (error) { 582 INP_INFO_WUNLOCK(&ripcbinfo); 583 return error; 584 } 585 inp = (struct inpcb *)so->so_pcb; 586 INP_LOCK(inp); 587 INP_INFO_WUNLOCK(&ripcbinfo); 588 inp->inp_vflag |= INP_IPV4; 589 inp->inp_ip_p = proto; 590 inp->inp_ip_ttl = ip_defttl; 591 INP_UNLOCK(inp); 592 return 0; 593} 594 595static void 596rip_pcbdetach(struct socket *so, struct inpcb *inp) 597{ 598 INP_INFO_WLOCK_ASSERT(&ripcbinfo); 599 INP_LOCK_ASSERT(inp); 600 601 if (so == ip_mrouter && ip_mrouter_done) 602 ip_mrouter_done(); 603 if (ip_rsvp_force_done) 604 ip_rsvp_force_done(so); 605 if (so == ip_rsvpd) 606 ip_rsvp_done(); 607 in_pcbdetach(inp); 608} 609 610static int 611rip_detach(struct socket *so) 612{ 613 struct inpcb *inp; 614 615 INP_INFO_WLOCK(&ripcbinfo); 616 inp = sotoinpcb(so); 617 if (inp == 0) { 618 /* XXX counter, printf */ 619 INP_INFO_WUNLOCK(&ripcbinfo); 620 return EINVAL; 621 } 622 INP_LOCK(inp); 623 rip_pcbdetach(so, inp); 624 INP_INFO_WUNLOCK(&ripcbinfo); 625 return 0; 626} 627 628static int 629rip_abort(struct socket *so) 630{ 631 struct inpcb *inp; 632 633 INP_INFO_WLOCK(&ripcbinfo); 634 inp = sotoinpcb(so); 635 if (inp == 0) { 636 INP_INFO_WUNLOCK(&ripcbinfo); 637 return EINVAL; /* ??? possible? panic instead? */ 638 } 639 INP_LOCK(inp); 640 soisdisconnected(so); 641 if (so->so_state & SS_NOFDREF) 642 rip_pcbdetach(so, inp); 643 else 644 INP_UNLOCK(inp); 645 INP_INFO_WUNLOCK(&ripcbinfo); 646 return 0; 647} 648 649static int 650rip_disconnect(struct socket *so) 651{ 652 if ((so->so_state & SS_ISCONNECTED) == 0) 653 return ENOTCONN; 654 return rip_abort(so); 655} 656 657static int 658rip_bind(struct socket *so, struct sockaddr *nam, struct thread *td) 659{ 660 struct sockaddr_in *addr = (struct sockaddr_in *)nam; 661 struct inpcb *inp; 662 663 if (nam->sa_len != sizeof(*addr)) 664 return EINVAL; 665 666 if (jailed(td->td_ucred)) { 667 if (addr->sin_addr.s_addr == INADDR_ANY) 668 addr->sin_addr.s_addr = 669 htonl(prison_getip(td->td_ucred)); 670 if (htonl(prison_getip(td->td_ucred)) != addr->sin_addr.s_addr) 671 return (EADDRNOTAVAIL); 672 } 673 674 if (TAILQ_EMPTY(&ifnet) || 675 (addr->sin_family != AF_INET && addr->sin_family != AF_IMPLINK) || 676 (addr->sin_addr.s_addr && 677 ifa_ifwithaddr((struct sockaddr *)addr) == 0)) 678 return EADDRNOTAVAIL; 679 680 INP_INFO_WLOCK(&ripcbinfo); 681 inp = sotoinpcb(so); 682 if (inp == 0) { 683 INP_INFO_WUNLOCK(&ripcbinfo); 684 return EINVAL; 685 } 686 INP_LOCK(inp); 687 inp->inp_laddr = addr->sin_addr; 688 INP_UNLOCK(inp); 689 INP_INFO_WUNLOCK(&ripcbinfo); 690 return 0; 691} 692 693static int 694rip_connect(struct socket *so, struct sockaddr *nam, struct thread *td) 695{ 696 struct sockaddr_in *addr = (struct sockaddr_in *)nam; 697 struct inpcb *inp; 698 699 if (nam->sa_len != sizeof(*addr)) 700 return EINVAL; 701 if (TAILQ_EMPTY(&ifnet)) 702 return EADDRNOTAVAIL; 703 if (addr->sin_family != AF_INET && addr->sin_family != AF_IMPLINK) 704 return EAFNOSUPPORT; 705 706 INP_INFO_WLOCK(&ripcbinfo); 707 inp = sotoinpcb(so); 708 if (inp == 0) { 709 INP_INFO_WUNLOCK(&ripcbinfo); 710 return EINVAL; 711 } 712 INP_LOCK(inp); 713 inp->inp_faddr = addr->sin_addr; 714 soisconnected(so); 715 INP_UNLOCK(inp); 716 INP_INFO_WUNLOCK(&ripcbinfo); 717 return 0; 718} 719 720static int 721rip_shutdown(struct socket *so) 722{ 723 struct inpcb *inp; 724 725 INP_INFO_RLOCK(&ripcbinfo); 726 inp = sotoinpcb(so); 727 if (inp == 0) { 728 INP_INFO_RUNLOCK(&ripcbinfo); 729 return EINVAL; 730 } 731 INP_LOCK(inp); 732 INP_INFO_RUNLOCK(&ripcbinfo); 733 socantsendmore(so); 734 INP_UNLOCK(inp); 735 return 0; 736} 737 738static int 739rip_send(struct socket *so, int flags, struct mbuf *m, struct sockaddr *nam, 740 struct mbuf *control, struct thread *td) 741{ 742 struct inpcb *inp; 743 u_long dst; 744 int ret; 745 746 INP_INFO_WLOCK(&ripcbinfo); 747 inp = sotoinpcb(so); 748 if (so->so_state & SS_ISCONNECTED) { 749 if (nam) { 750 INP_INFO_WUNLOCK(&ripcbinfo); 751 m_freem(m); 752 return EISCONN; 753 } 754 dst = inp->inp_faddr.s_addr; 755 } else { 756 if (nam == NULL) { 757 INP_INFO_WUNLOCK(&ripcbinfo); 758 m_freem(m); 759 return ENOTCONN; 760 } 761 dst = ((struct sockaddr_in *)nam)->sin_addr.s_addr; 762 } 763 ret = rip_output(m, so, dst); 764 INP_INFO_WUNLOCK(&ripcbinfo); 765 return ret; 766} 767 768static int 769rip_pcblist(SYSCTL_HANDLER_ARGS) 770{ 771 int error, i, n; 772 struct inpcb *inp, **inp_list; 773 inp_gen_t gencnt; 774 struct xinpgen xig; 775 776 /* 777 * The process of preparing the TCB list is too time-consuming and 778 * resource-intensive to repeat twice on every request. 779 */ 780 if (req->oldptr == 0) { 781 n = ripcbinfo.ipi_count; 782 req->oldidx = 2 * (sizeof xig) 783 + (n + n/8) * sizeof(struct xinpcb); 784 return 0; 785 } 786 787 if (req->newptr != 0) 788 return EPERM; 789 790 /* 791 * OK, now we're committed to doing something. 792 */ 793 INP_INFO_RLOCK(&ripcbinfo); 794 gencnt = ripcbinfo.ipi_gencnt; 795 n = ripcbinfo.ipi_count; 796 INP_INFO_RUNLOCK(&ripcbinfo); 797 798 xig.xig_len = sizeof xig; 799 xig.xig_count = n; 800 xig.xig_gen = gencnt; 801 xig.xig_sogen = so_gencnt; 802 error = SYSCTL_OUT(req, &xig, sizeof xig); 803 if (error) 804 return error; 805 806 inp_list = malloc(n * sizeof *inp_list, M_TEMP, M_WAITOK); 807 if (inp_list == 0) 808 return ENOMEM; 809 810 INP_INFO_RLOCK(&ripcbinfo); 811 for (inp = LIST_FIRST(ripcbinfo.listhead), i = 0; inp && i < n; 812 inp = LIST_NEXT(inp, inp_list)) { 813 INP_LOCK(inp); 814 if (inp->inp_gencnt <= gencnt && 815 cr_canseesocket(req->td->td_ucred, inp->inp_socket) == 0) { 816 /* XXX held references? */ 817 inp_list[i++] = inp; 818 } 819 INP_UNLOCK(inp); 820 } 821 INP_INFO_RUNLOCK(&ripcbinfo); 822 n = i; 823 824 error = 0; 825 for (i = 0; i < n; i++) { 826 inp = inp_list[i]; 827 if (inp->inp_gencnt <= gencnt) { 828 struct xinpcb xi; 829 xi.xi_len = sizeof xi; 830 /* XXX should avoid extra copy */ 831 bcopy(inp, &xi.xi_inp, sizeof *inp); 832 if (inp->inp_socket) 833 sotoxsocket(inp->inp_socket, &xi.xi_socket); 834 error = SYSCTL_OUT(req, &xi, sizeof xi); 835 } 836 } 837 if (!error) { 838 /* 839 * Give the user an updated idea of our state. 840 * If the generation differs from what we told 841 * her before, she knows that something happened 842 * while we were processing this request, and it 843 * might be necessary to retry. 844 */ 845 INP_INFO_RLOCK(&ripcbinfo); 846 xig.xig_gen = ripcbinfo.ipi_gencnt; 847 xig.xig_sogen = so_gencnt; 848 xig.xig_count = ripcbinfo.ipi_count; 849 INP_INFO_RUNLOCK(&ripcbinfo); 850 error = SYSCTL_OUT(req, &xig, sizeof xig); 851 } 852 free(inp_list, M_TEMP); 853 return error; 854} 855 856/* 857 * This is the wrapper function for in_setsockaddr. We just pass down 858 * the pcbinfo for in_setpeeraddr to lock. 859 */ 860static int 861rip_sockaddr(struct socket *so, struct sockaddr **nam) 862{ 863 return (in_setsockaddr(so, nam, &ripcbinfo)); 864} 865 866/* 867 * This is the wrapper function for in_setpeeraddr. We just pass down 868 * the pcbinfo for in_setpeeraddr to lock. 869 */ 870static int 871rip_peeraddr(struct socket *so, struct sockaddr **nam) 872{ 873 return (in_setpeeraddr(so, nam, &ripcbinfo)); 874} 875 876 877SYSCTL_PROC(_net_inet_raw, OID_AUTO/*XXX*/, pcblist, CTLFLAG_RD, 0, 0, 878 rip_pcblist, "S,xinpcb", "List of active raw IP sockets"); 879 880struct pr_usrreqs rip_usrreqs = { 881 rip_abort, pru_accept_notsupp, rip_attach, rip_bind, rip_connect, 882 pru_connect2_notsupp, in_control, rip_detach, rip_disconnect, 883 pru_listen_notsupp, rip_peeraddr, pru_rcvd_notsupp, 884 pru_rcvoob_notsupp, rip_send, pru_sense_null, rip_shutdown, 885 rip_sockaddr, sosend, soreceive, sopoll, in_pcbsosetlabel 886}; 887