raw_ip.c revision 149371
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 149371 2005-08-22 16:13:08Z andre $ 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 /* Check the minimum TTL for socket. */ 161 if (last->inp_ip_minttl && last->inp_ip_minttl > ip->ip_ttl) 162 policyfail = 1; 163 if (!policyfail) { 164 struct mbuf *opts = NULL; 165 struct socket *so; 166 167 so = last->inp_socket; 168 if ((last->inp_flags & INP_CONTROLOPTS) || 169 (so->so_options & SO_TIMESTAMP)) 170 ip_savecontrol(last, &opts, ip, n); 171 SOCKBUF_LOCK(&so->so_rcv); 172 if (sbappendaddr_locked(&so->so_rcv, 173 (struct sockaddr *)&ripsrc, n, opts) == 0) { 174 /* should notify about lost packet */ 175 m_freem(n); 176 if (opts) 177 m_freem(opts); 178 SOCKBUF_UNLOCK(&so->so_rcv); 179 } else 180 sorwakeup_locked(so); 181 } else 182 m_freem(n); 183 return policyfail; 184} 185 186/* 187 * Setup generic address and protocol structures 188 * for raw_input routine, then pass them along with 189 * mbuf chain. 190 */ 191void 192rip_input(struct mbuf *m, int off) 193{ 194 struct ip *ip = mtod(m, struct ip *); 195 int proto = ip->ip_p; 196 struct inpcb *inp, *last; 197 198 INP_INFO_RLOCK(&ripcbinfo); 199 ripsrc.sin_addr = ip->ip_src; 200 last = NULL; 201 LIST_FOREACH(inp, &ripcb, inp_list) { 202 INP_LOCK(inp); 203 if (inp->inp_ip_p && inp->inp_ip_p != proto) { 204 docontinue: 205 INP_UNLOCK(inp); 206 continue; 207 } 208#ifdef INET6 209 if ((inp->inp_vflag & INP_IPV4) == 0) 210 goto docontinue; 211#endif 212 if (inp->inp_laddr.s_addr && 213 inp->inp_laddr.s_addr != ip->ip_dst.s_addr) 214 goto docontinue; 215 if (inp->inp_faddr.s_addr && 216 inp->inp_faddr.s_addr != ip->ip_src.s_addr) 217 goto docontinue; 218 if (jailed(inp->inp_socket->so_cred)) 219 if (htonl(prison_getip(inp->inp_socket->so_cred)) != 220 ip->ip_dst.s_addr) 221 goto docontinue; 222 if (last) { 223 struct mbuf *n; 224 225 n = m_copy(m, 0, (int)M_COPYALL); 226 if (n != NULL) 227 (void) raw_append(last, ip, n); 228 /* XXX count dropped packet */ 229 INP_UNLOCK(last); 230 } 231 last = inp; 232 } 233 if (last != NULL) { 234 if (raw_append(last, ip, m) != 0) 235 ipstat.ips_delivered--; 236 INP_UNLOCK(last); 237 } else { 238 m_freem(m); 239 ipstat.ips_noproto++; 240 ipstat.ips_delivered--; 241 } 242 INP_INFO_RUNLOCK(&ripcbinfo); 243} 244 245/* 246 * Generate IP header and pass packet to ip_output. 247 * Tack on options user may have setup with control call. 248 */ 249int 250rip_output(struct mbuf *m, struct socket *so, u_long dst) 251{ 252 struct ip *ip; 253 int error; 254 struct inpcb *inp = sotoinpcb(so); 255 int flags = ((so->so_options & SO_DONTROUTE) ? IP_ROUTETOIF : 0) | 256 IP_ALLOWBROADCAST; 257 258 /* 259 * If the user handed us a complete IP packet, use it. 260 * Otherwise, allocate an mbuf for a header and fill it in. 261 */ 262 if ((inp->inp_flags & INP_HDRINCL) == 0) { 263 if (m->m_pkthdr.len + sizeof(struct ip) > IP_MAXPACKET) { 264 m_freem(m); 265 return(EMSGSIZE); 266 } 267 M_PREPEND(m, sizeof(struct ip), M_DONTWAIT); 268 if (m == NULL) 269 return(ENOBUFS); 270 271 INP_LOCK(inp); 272 ip = mtod(m, struct ip *); 273 ip->ip_tos = inp->inp_ip_tos; 274 ip->ip_off = 0; 275 ip->ip_p = inp->inp_ip_p; 276 ip->ip_len = m->m_pkthdr.len; 277 if (jailed(inp->inp_socket->so_cred)) 278 ip->ip_src.s_addr = 279 htonl(prison_getip(inp->inp_socket->so_cred)); 280 else 281 ip->ip_src = inp->inp_laddr; 282 ip->ip_dst.s_addr = dst; 283 ip->ip_ttl = inp->inp_ip_ttl; 284 } else { 285 if (m->m_pkthdr.len > IP_MAXPACKET) { 286 m_freem(m); 287 return(EMSGSIZE); 288 } 289 INP_LOCK(inp); 290 ip = mtod(m, struct ip *); 291 if (jailed(inp->inp_socket->so_cred)) { 292 if (ip->ip_src.s_addr != 293 htonl(prison_getip(inp->inp_socket->so_cred))) { 294 INP_UNLOCK(inp); 295 m_freem(m); 296 return (EPERM); 297 } 298 } 299 /* don't allow both user specified and setsockopt options, 300 and don't allow packet length sizes that will crash */ 301 if (((ip->ip_hl != (sizeof (*ip) >> 2)) 302 && inp->inp_options) 303 || (ip->ip_len > m->m_pkthdr.len) 304 || (ip->ip_len < (ip->ip_hl << 2))) { 305 INP_UNLOCK(inp); 306 m_freem(m); 307 return EINVAL; 308 } 309 if (ip->ip_id == 0) 310 ip->ip_id = ip_newid(); 311 /* XXX prevent ip_output from overwriting header fields */ 312 flags |= IP_RAWOUTPUT; 313 ipstat.ips_rawout++; 314 } 315 316 if (inp->inp_flags & INP_ONESBCAST) 317 flags |= IP_SENDONES; 318 319#ifdef MAC 320 mac_create_mbuf_from_inpcb(inp, m); 321#endif 322 323 error = ip_output(m, inp->inp_options, NULL, flags, 324 inp->inp_moptions, inp); 325 INP_UNLOCK(inp); 326 return error; 327} 328 329/* 330 * Raw IP socket option processing. 331 * 332 * IMPORTANT NOTE regarding access control: Traditionally, raw sockets could 333 * only be created by a privileged process, and as such, socket option 334 * operations to manage system properties on any raw socket were allowed to 335 * take place without explicit additional access control checks. However, 336 * raw sockets can now also be created in jail(), and therefore explicit 337 * checks are now required. Likewise, raw sockets can be used by a process 338 * after it gives up privilege, so some caution is required. For options 339 * passed down to the IP layer via ip_ctloutput(), checks are assumed to be 340 * performed in ip_ctloutput() and therefore no check occurs here. 341 * Unilaterally checking suser() here breaks normal IP socket option 342 * operations on raw sockets. 343 * 344 * When adding new socket options here, make sure to add access control 345 * checks here as necessary. 346 */ 347int 348rip_ctloutput(struct socket *so, struct sockopt *sopt) 349{ 350 struct inpcb *inp = sotoinpcb(so); 351 int error, optval; 352 353 if (sopt->sopt_level != IPPROTO_IP) 354 return (EINVAL); 355 356 error = 0; 357 switch (sopt->sopt_dir) { 358 case SOPT_GET: 359 switch (sopt->sopt_name) { 360 case IP_HDRINCL: 361 optval = inp->inp_flags & INP_HDRINCL; 362 error = sooptcopyout(sopt, &optval, sizeof optval); 363 break; 364 365 case IP_FW_ADD: /* ADD actually returns the body... */ 366 case IP_FW_GET: 367 case IP_FW_TABLE_GETSIZE: 368 case IP_FW_TABLE_LIST: 369 error = suser(curthread); 370 if (error != 0) 371 return (error); 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 error = suser(curthread); 380 if (error != 0) 381 return (error); 382 if (ip_dn_ctl_ptr != NULL) 383 error = ip_dn_ctl_ptr(sopt); 384 else 385 error = ENOPROTOOPT; 386 break ; 387 388 case MRT_INIT: 389 case MRT_DONE: 390 case MRT_ADD_VIF: 391 case MRT_DEL_VIF: 392 case MRT_ADD_MFC: 393 case MRT_DEL_MFC: 394 case MRT_VERSION: 395 case MRT_ASSERT: 396 case MRT_API_SUPPORT: 397 case MRT_API_CONFIG: 398 case MRT_ADD_BW_UPCALL: 399 case MRT_DEL_BW_UPCALL: 400 error = suser(curthread); 401 if (error != 0) 402 return (error); 403 error = ip_mrouter_get ? ip_mrouter_get(so, sopt) : 404 EOPNOTSUPP; 405 break; 406 407 default: 408 error = ip_ctloutput(so, sopt); 409 break; 410 } 411 break; 412 413 case SOPT_SET: 414 switch (sopt->sopt_name) { 415 case IP_HDRINCL: 416 error = sooptcopyin(sopt, &optval, sizeof optval, 417 sizeof optval); 418 if (error) 419 break; 420 if (optval) 421 inp->inp_flags |= INP_HDRINCL; 422 else 423 inp->inp_flags &= ~INP_HDRINCL; 424 break; 425 426 case IP_FW_ADD: 427 case IP_FW_DEL: 428 case IP_FW_FLUSH: 429 case IP_FW_ZERO: 430 case IP_FW_RESETLOG: 431 case IP_FW_TABLE_ADD: 432 case IP_FW_TABLE_DEL: 433 case IP_FW_TABLE_FLUSH: 434 error = suser(curthread); 435 if (error != 0) 436 return (error); 437 if (ip_fw_ctl_ptr != NULL) 438 error = ip_fw_ctl_ptr(sopt); 439 else 440 error = ENOPROTOOPT; 441 break; 442 443 case IP_DUMMYNET_CONFIGURE: 444 case IP_DUMMYNET_DEL: 445 case IP_DUMMYNET_FLUSH: 446 error = suser(curthread); 447 if (error != 0) 448 return (error); 449 if (ip_dn_ctl_ptr != NULL) 450 error = ip_dn_ctl_ptr(sopt); 451 else 452 error = ENOPROTOOPT ; 453 break ; 454 455 case IP_RSVP_ON: 456 error = suser(curthread); 457 if (error != 0) 458 return (error); 459 error = ip_rsvp_init(so); 460 break; 461 462 case IP_RSVP_OFF: 463 error = suser(curthread); 464 if (error != 0) 465 return (error); 466 error = ip_rsvp_done(); 467 break; 468 469 case IP_RSVP_VIF_ON: 470 case IP_RSVP_VIF_OFF: 471 error = suser(curthread); 472 if (error != 0) 473 return (error); 474 error = ip_rsvp_vif ? 475 ip_rsvp_vif(so, sopt) : EINVAL; 476 break; 477 478 case MRT_INIT: 479 case MRT_DONE: 480 case MRT_ADD_VIF: 481 case MRT_DEL_VIF: 482 case MRT_ADD_MFC: 483 case MRT_DEL_MFC: 484 case MRT_VERSION: 485 case MRT_ASSERT: 486 case MRT_API_SUPPORT: 487 case MRT_API_CONFIG: 488 case MRT_ADD_BW_UPCALL: 489 case MRT_DEL_BW_UPCALL: 490 error = suser(curthread); 491 if (error != 0) 492 return (error); 493 error = ip_mrouter_set ? ip_mrouter_set(so, sopt) : 494 EOPNOTSUPP; 495 break; 496 497 default: 498 error = ip_ctloutput(so, sopt); 499 break; 500 } 501 break; 502 } 503 504 return (error); 505} 506 507/* 508 * This function exists solely to receive the PRC_IFDOWN messages which 509 * are sent by if_down(). It looks for an ifaddr whose ifa_addr is sa, 510 * and calls in_ifadown() to remove all routes corresponding to that address. 511 * It also receives the PRC_IFUP messages from if_up() and reinstalls the 512 * interface routes. 513 */ 514void 515rip_ctlinput(int cmd, struct sockaddr *sa, void *vip) 516{ 517 struct in_ifaddr *ia; 518 struct ifnet *ifp; 519 int err; 520 int flags; 521 522 switch (cmd) { 523 case PRC_IFDOWN: 524 TAILQ_FOREACH(ia, &in_ifaddrhead, ia_link) { 525 if (ia->ia_ifa.ifa_addr == sa 526 && (ia->ia_flags & IFA_ROUTE)) { 527 /* 528 * in_ifscrub kills the interface route. 529 */ 530 in_ifscrub(ia->ia_ifp, ia); 531 /* 532 * in_ifadown gets rid of all the rest of 533 * the routes. This is not quite the right 534 * thing to do, but at least if we are running 535 * a routing process they will come back. 536 */ 537 in_ifadown(&ia->ia_ifa, 0); 538 break; 539 } 540 } 541 break; 542 543 case PRC_IFUP: 544 TAILQ_FOREACH(ia, &in_ifaddrhead, ia_link) { 545 if (ia->ia_ifa.ifa_addr == sa) 546 break; 547 } 548 if (ia == 0 || (ia->ia_flags & IFA_ROUTE)) 549 return; 550 flags = RTF_UP; 551 ifp = ia->ia_ifa.ifa_ifp; 552 553 if ((ifp->if_flags & IFF_LOOPBACK) 554 || (ifp->if_flags & IFF_POINTOPOINT)) 555 flags |= RTF_HOST; 556 557 err = rtinit(&ia->ia_ifa, RTM_ADD, flags); 558 if (err == 0) 559 ia->ia_flags |= IFA_ROUTE; 560 break; 561 } 562} 563 564u_long rip_sendspace = RIPSNDQ; 565u_long rip_recvspace = RIPRCVQ; 566 567SYSCTL_INT(_net_inet_raw, OID_AUTO, maxdgram, CTLFLAG_RW, 568 &rip_sendspace, 0, "Maximum outgoing raw IP datagram size"); 569SYSCTL_INT(_net_inet_raw, OID_AUTO, recvspace, CTLFLAG_RW, 570 &rip_recvspace, 0, "Maximum space for incoming raw IP datagrams"); 571 572static int 573rip_attach(struct socket *so, int proto, struct thread *td) 574{ 575 struct inpcb *inp; 576 int error; 577 578 /* XXX why not lower? */ 579 INP_INFO_WLOCK(&ripcbinfo); 580 inp = sotoinpcb(so); 581 if (inp) { 582 /* XXX counter, printf */ 583 INP_INFO_WUNLOCK(&ripcbinfo); 584 return EINVAL; 585 } 586 if (jailed(td->td_ucred) && !jail_allow_raw_sockets) { 587 INP_INFO_WUNLOCK(&ripcbinfo); 588 return (EPERM); 589 } 590 if ((error = suser_cred(td->td_ucred, SUSER_ALLOWJAIL)) != 0) { 591 INP_INFO_WUNLOCK(&ripcbinfo); 592 return error; 593 } 594 if (proto >= IPPROTO_MAX || proto < 0) { 595 INP_INFO_WUNLOCK(&ripcbinfo); 596 return EPROTONOSUPPORT; 597 } 598 599 error = soreserve(so, rip_sendspace, rip_recvspace); 600 if (error) { 601 INP_INFO_WUNLOCK(&ripcbinfo); 602 return error; 603 } 604 error = in_pcballoc(so, &ripcbinfo, "rawinp"); 605 if (error) { 606 INP_INFO_WUNLOCK(&ripcbinfo); 607 return error; 608 } 609 inp = (struct inpcb *)so->so_pcb; 610 INP_LOCK(inp); 611 INP_INFO_WUNLOCK(&ripcbinfo); 612 inp->inp_vflag |= INP_IPV4; 613 inp->inp_ip_p = proto; 614 inp->inp_ip_ttl = ip_defttl; 615 INP_UNLOCK(inp); 616 return 0; 617} 618 619static void 620rip_pcbdetach(struct socket *so, struct inpcb *inp) 621{ 622 623 INP_INFO_WLOCK_ASSERT(&ripcbinfo); 624 INP_LOCK_ASSERT(inp); 625 626 if (so == ip_mrouter && ip_mrouter_done) 627 ip_mrouter_done(); 628 if (ip_rsvp_force_done) 629 ip_rsvp_force_done(so); 630 if (so == ip_rsvpd) 631 ip_rsvp_done(); 632 in_pcbdetach(inp); 633} 634 635static int 636rip_detach(struct socket *so) 637{ 638 struct inpcb *inp; 639 640 INP_INFO_WLOCK(&ripcbinfo); 641 inp = sotoinpcb(so); 642 if (inp == 0) { 643 /* XXX counter, printf */ 644 INP_INFO_WUNLOCK(&ripcbinfo); 645 return EINVAL; 646 } 647 INP_LOCK(inp); 648 rip_pcbdetach(so, inp); 649 INP_INFO_WUNLOCK(&ripcbinfo); 650 return 0; 651} 652 653static int 654rip_abort(struct socket *so) 655{ 656 struct inpcb *inp; 657 658 INP_INFO_WLOCK(&ripcbinfo); 659 inp = sotoinpcb(so); 660 if (inp == 0) { 661 INP_INFO_WUNLOCK(&ripcbinfo); 662 return EINVAL; /* ??? possible? panic instead? */ 663 } 664 INP_LOCK(inp); 665 soisdisconnected(so); 666 if (so->so_state & SS_NOFDREF) 667 rip_pcbdetach(so, inp); 668 else 669 INP_UNLOCK(inp); 670 INP_INFO_WUNLOCK(&ripcbinfo); 671 return 0; 672} 673 674static int 675rip_disconnect(struct socket *so) 676{ 677 if ((so->so_state & SS_ISCONNECTED) == 0) 678 return ENOTCONN; 679 return rip_abort(so); 680} 681 682static int 683rip_bind(struct socket *so, struct sockaddr *nam, struct thread *td) 684{ 685 struct sockaddr_in *addr = (struct sockaddr_in *)nam; 686 struct inpcb *inp; 687 688 if (nam->sa_len != sizeof(*addr)) 689 return EINVAL; 690 691 if (jailed(td->td_ucred)) { 692 if (addr->sin_addr.s_addr == INADDR_ANY) 693 addr->sin_addr.s_addr = 694 htonl(prison_getip(td->td_ucred)); 695 if (htonl(prison_getip(td->td_ucred)) != addr->sin_addr.s_addr) 696 return (EADDRNOTAVAIL); 697 } 698 699 if (TAILQ_EMPTY(&ifnet) || 700 (addr->sin_family != AF_INET && addr->sin_family != AF_IMPLINK) || 701 (addr->sin_addr.s_addr && 702 ifa_ifwithaddr((struct sockaddr *)addr) == 0)) 703 return EADDRNOTAVAIL; 704 705 INP_INFO_WLOCK(&ripcbinfo); 706 inp = sotoinpcb(so); 707 if (inp == 0) { 708 INP_INFO_WUNLOCK(&ripcbinfo); 709 return EINVAL; 710 } 711 INP_LOCK(inp); 712 inp->inp_laddr = addr->sin_addr; 713 INP_UNLOCK(inp); 714 INP_INFO_WUNLOCK(&ripcbinfo); 715 return 0; 716} 717 718static int 719rip_connect(struct socket *so, struct sockaddr *nam, struct thread *td) 720{ 721 struct sockaddr_in *addr = (struct sockaddr_in *)nam; 722 struct inpcb *inp; 723 724 if (nam->sa_len != sizeof(*addr)) 725 return EINVAL; 726 if (TAILQ_EMPTY(&ifnet)) 727 return EADDRNOTAVAIL; 728 if (addr->sin_family != AF_INET && addr->sin_family != AF_IMPLINK) 729 return EAFNOSUPPORT; 730 731 INP_INFO_WLOCK(&ripcbinfo); 732 inp = sotoinpcb(so); 733 if (inp == 0) { 734 INP_INFO_WUNLOCK(&ripcbinfo); 735 return EINVAL; 736 } 737 INP_LOCK(inp); 738 inp->inp_faddr = addr->sin_addr; 739 soisconnected(so); 740 INP_UNLOCK(inp); 741 INP_INFO_WUNLOCK(&ripcbinfo); 742 return 0; 743} 744 745static int 746rip_shutdown(struct socket *so) 747{ 748 struct inpcb *inp; 749 750 INP_INFO_RLOCK(&ripcbinfo); 751 inp = sotoinpcb(so); 752 if (inp == 0) { 753 INP_INFO_RUNLOCK(&ripcbinfo); 754 return EINVAL; 755 } 756 INP_LOCK(inp); 757 INP_INFO_RUNLOCK(&ripcbinfo); 758 socantsendmore(so); 759 INP_UNLOCK(inp); 760 return 0; 761} 762 763static int 764rip_send(struct socket *so, int flags, struct mbuf *m, struct sockaddr *nam, 765 struct mbuf *control, struct thread *td) 766{ 767 struct inpcb *inp; 768 u_long dst; 769 int ret; 770 771 INP_INFO_WLOCK(&ripcbinfo); 772 inp = sotoinpcb(so); 773 if (so->so_state & SS_ISCONNECTED) { 774 if (nam) { 775 INP_INFO_WUNLOCK(&ripcbinfo); 776 m_freem(m); 777 return EISCONN; 778 } 779 dst = inp->inp_faddr.s_addr; 780 } else { 781 if (nam == NULL) { 782 INP_INFO_WUNLOCK(&ripcbinfo); 783 m_freem(m); 784 return ENOTCONN; 785 } 786 dst = ((struct sockaddr_in *)nam)->sin_addr.s_addr; 787 } 788 ret = rip_output(m, so, dst); 789 INP_INFO_WUNLOCK(&ripcbinfo); 790 return ret; 791} 792 793static int 794rip_pcblist(SYSCTL_HANDLER_ARGS) 795{ 796 int error, i, n; 797 struct inpcb *inp, **inp_list; 798 inp_gen_t gencnt; 799 struct xinpgen xig; 800 801 /* 802 * The process of preparing the TCB list is too time-consuming and 803 * resource-intensive to repeat twice on every request. 804 */ 805 if (req->oldptr == 0) { 806 n = ripcbinfo.ipi_count; 807 req->oldidx = 2 * (sizeof xig) 808 + (n + n/8) * sizeof(struct xinpcb); 809 return 0; 810 } 811 812 if (req->newptr != 0) 813 return EPERM; 814 815 /* 816 * OK, now we're committed to doing something. 817 */ 818 INP_INFO_RLOCK(&ripcbinfo); 819 gencnt = ripcbinfo.ipi_gencnt; 820 n = ripcbinfo.ipi_count; 821 INP_INFO_RUNLOCK(&ripcbinfo); 822 823 xig.xig_len = sizeof xig; 824 xig.xig_count = n; 825 xig.xig_gen = gencnt; 826 xig.xig_sogen = so_gencnt; 827 error = SYSCTL_OUT(req, &xig, sizeof xig); 828 if (error) 829 return error; 830 831 inp_list = malloc(n * sizeof *inp_list, M_TEMP, M_WAITOK); 832 if (inp_list == 0) 833 return ENOMEM; 834 835 INP_INFO_RLOCK(&ripcbinfo); 836 for (inp = LIST_FIRST(ripcbinfo.listhead), i = 0; inp && i < n; 837 inp = LIST_NEXT(inp, inp_list)) { 838 INP_LOCK(inp); 839 if (inp->inp_gencnt <= gencnt && 840 cr_canseesocket(req->td->td_ucred, inp->inp_socket) == 0) { 841 /* XXX held references? */ 842 inp_list[i++] = inp; 843 } 844 INP_UNLOCK(inp); 845 } 846 INP_INFO_RUNLOCK(&ripcbinfo); 847 n = i; 848 849 error = 0; 850 for (i = 0; i < n; i++) { 851 inp = inp_list[i]; 852 if (inp->inp_gencnt <= gencnt) { 853 struct xinpcb xi; 854 bzero(&xi, sizeof(xi)); 855 xi.xi_len = sizeof xi; 856 /* XXX should avoid extra copy */ 857 bcopy(inp, &xi.xi_inp, sizeof *inp); 858 if (inp->inp_socket) 859 sotoxsocket(inp->inp_socket, &xi.xi_socket); 860 error = SYSCTL_OUT(req, &xi, sizeof xi); 861 } 862 } 863 if (!error) { 864 /* 865 * Give the user an updated idea of our state. 866 * If the generation differs from what we told 867 * her before, she knows that something happened 868 * while we were processing this request, and it 869 * might be necessary to retry. 870 */ 871 INP_INFO_RLOCK(&ripcbinfo); 872 xig.xig_gen = ripcbinfo.ipi_gencnt; 873 xig.xig_sogen = so_gencnt; 874 xig.xig_count = ripcbinfo.ipi_count; 875 INP_INFO_RUNLOCK(&ripcbinfo); 876 error = SYSCTL_OUT(req, &xig, sizeof xig); 877 } 878 free(inp_list, M_TEMP); 879 return error; 880} 881 882/* 883 * This is the wrapper function for in_setsockaddr. We just pass down 884 * the pcbinfo for in_setpeeraddr to lock. 885 */ 886static int 887rip_sockaddr(struct socket *so, struct sockaddr **nam) 888{ 889 return (in_setsockaddr(so, nam, &ripcbinfo)); 890} 891 892/* 893 * This is the wrapper function for in_setpeeraddr. We just pass down 894 * the pcbinfo for in_setpeeraddr to lock. 895 */ 896static int 897rip_peeraddr(struct socket *so, struct sockaddr **nam) 898{ 899 return (in_setpeeraddr(so, nam, &ripcbinfo)); 900} 901 902 903SYSCTL_PROC(_net_inet_raw, OID_AUTO/*XXX*/, pcblist, CTLFLAG_RD, 0, 0, 904 rip_pcblist, "S,xinpcb", "List of active raw IP sockets"); 905 906struct pr_usrreqs rip_usrreqs = { 907 .pru_abort = rip_abort, 908 .pru_attach = rip_attach, 909 .pru_bind = rip_bind, 910 .pru_connect = rip_connect, 911 .pru_control = in_control, 912 .pru_detach = rip_detach, 913 .pru_disconnect = rip_disconnect, 914 .pru_peeraddr = rip_peeraddr, 915 .pru_send = rip_send, 916 .pru_shutdown = rip_shutdown, 917 .pru_sockaddr = rip_sockaddr, 918 .pru_sosetlabel = in_pcbsosetlabel 919}; 920