raw_ip.c revision 128903
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 128903 2004-05-04 00:10:16Z rwatson $ 31 */ 32 33#include "opt_inet6.h" 34#include "opt_ipsec.h" 35#include "opt_mac.h" 36#include "opt_random_ip_id.h" 37 38#include <sys/param.h> 39#include <sys/jail.h> 40#include <sys/kernel.h> 41#include <sys/lock.h> 42#include <sys/mac.h> 43#include <sys/malloc.h> 44#include <sys/mbuf.h> 45#include <sys/proc.h> 46#include <sys/protosw.h> 47#include <sys/signalvar.h> 48#include <sys/socket.h> 49#include <sys/socketvar.h> 50#include <sys/sx.h> 51#include <sys/sysctl.h> 52#include <sys/systm.h> 53 54#include <vm/uma.h> 55 56#include <net/if.h> 57#include <net/route.h> 58 59#include <netinet/in.h> 60#include <netinet/in_systm.h> 61#include <netinet/in_pcb.h> 62#include <netinet/in_var.h> 63#include <netinet/ip.h> 64#include <netinet/ip_var.h> 65#include <netinet/ip_mroute.h> 66 67#include <netinet/ip_fw.h> 68#include <netinet/ip_dummynet.h> 69 70#ifdef FAST_IPSEC 71#include <netipsec/ipsec.h> 72#endif /*FAST_IPSEC*/ 73 74#ifdef IPSEC 75#include <netinet6/ipsec.h> 76#endif /*IPSEC*/ 77 78struct inpcbhead ripcb; 79struct inpcbinfo ripcbinfo; 80 81/* control hooks for ipfw and dummynet */ 82ip_fw_ctl_t *ip_fw_ctl_ptr; 83ip_dn_ctl_t *ip_dn_ctl_ptr; 84 85/* 86 * hooks for multicast routing. They all default to NULL, 87 * so leave them not initialized and rely on BSS being set to 0. 88 */ 89 90/* The socket used to communicate with the multicast routing daemon. */ 91struct socket *ip_mrouter; 92 93/* The various mrouter and rsvp functions */ 94int (*ip_mrouter_set)(struct socket *, struct sockopt *); 95int (*ip_mrouter_get)(struct socket *, struct sockopt *); 96int (*ip_mrouter_done)(void); 97int (*ip_mforward)(struct ip *, struct ifnet *, struct mbuf *, 98 struct ip_moptions *); 99int (*mrt_ioctl)(int, caddr_t); 100int (*legal_vif_num)(int); 101u_long (*ip_mcast_src)(int); 102 103void (*rsvp_input_p)(struct mbuf *m, int off); 104int (*ip_rsvp_vif)(struct socket *, struct sockopt *); 105void (*ip_rsvp_force_done)(struct socket *); 106 107/* 108 * Nominal space allocated to a raw ip socket. 109 */ 110#define RIPSNDQ 8192 111#define RIPRCVQ 8192 112 113/* 114 * Raw interface to IP protocol. 115 */ 116 117/* 118 * Initialize raw connection block q. 119 */ 120void 121rip_init() 122{ 123 INP_INFO_LOCK_INIT(&ripcbinfo, "rip"); 124 LIST_INIT(&ripcb); 125 ripcbinfo.listhead = &ripcb; 126 /* 127 * XXX We don't use the hash list for raw IP, but it's easier 128 * to allocate a one entry hash list than it is to check all 129 * over the place for hashbase == NULL. 130 */ 131 ripcbinfo.hashbase = hashinit(1, M_PCB, &ripcbinfo.hashmask); 132 ripcbinfo.porthashbase = hashinit(1, M_PCB, &ripcbinfo.porthashmask); 133 ripcbinfo.ipi_zone = uma_zcreate("ripcb", sizeof(struct inpcb), 134 NULL, NULL, NULL, NULL, UMA_ALIGN_PTR, UMA_ZONE_NOFREE); 135 uma_zone_set_max(ripcbinfo.ipi_zone, maxsockets); 136} 137 138static struct sockaddr_in ripsrc = { sizeof(ripsrc), AF_INET }; 139 140static int 141raw_append(struct inpcb *last, struct ip *ip, struct mbuf *n) 142{ 143 int policyfail = 0; 144 145 INP_LOCK_ASSERT(last); 146 147#if defined(IPSEC) || defined(FAST_IPSEC) 148 /* check AH/ESP integrity. */ 149 if (ipsec4_in_reject(n, last)) { 150 policyfail = 1; 151#ifdef IPSEC 152 ipsecstat.in_polvio++; 153#endif /*IPSEC*/ 154 /* do not inject data to pcb */ 155 } 156#endif /*IPSEC || FAST_IPSEC*/ 157#ifdef MAC 158 if (!policyfail && mac_check_inpcb_deliver(last, n) != 0) 159 policyfail = 1; 160#endif 161 if (!policyfail) { 162 struct mbuf *opts = NULL; 163 164 if ((last->inp_flags & INP_CONTROLOPTS) || 165 (last->inp_socket->so_options & SO_TIMESTAMP)) 166 ip_savecontrol(last, &opts, ip, n); 167 if (sbappendaddr(&last->inp_socket->so_rcv, 168 (struct sockaddr *)&ripsrc, n, opts) == 0) { 169 /* should notify about lost packet */ 170 m_freem(n); 171 if (opts) 172 m_freem(opts); 173 } else 174 sorwakeup(last->inp_socket); 175 } else 176 m_freem(n); 177 return policyfail; 178} 179 180/* 181 * Setup generic address and protocol structures 182 * for raw_input routine, then pass them along with 183 * mbuf chain. 184 */ 185void 186rip_input(struct mbuf *m, int off) 187{ 188 struct ip *ip = mtod(m, struct ip *); 189 int proto = ip->ip_p; 190 struct inpcb *inp, *last; 191 192 INP_INFO_RLOCK(&ripcbinfo); 193 ripsrc.sin_addr = ip->ip_src; 194 last = NULL; 195 LIST_FOREACH(inp, &ripcb, inp_list) { 196 INP_LOCK(inp); 197 if (inp->inp_ip_p && inp->inp_ip_p != proto) { 198 docontinue: 199 INP_UNLOCK(inp); 200 continue; 201 } 202#ifdef INET6 203 if ((inp->inp_vflag & INP_IPV4) == 0) 204 goto docontinue; 205#endif 206 if (inp->inp_laddr.s_addr && 207 inp->inp_laddr.s_addr != ip->ip_dst.s_addr) 208 goto docontinue; 209 if (inp->inp_faddr.s_addr && 210 inp->inp_faddr.s_addr != ip->ip_src.s_addr) 211 goto docontinue; 212 if (jailed(inp->inp_socket->so_cred)) 213 if (htonl(prison_getip(inp->inp_socket->so_cred)) != 214 ip->ip_dst.s_addr) 215 goto docontinue; 216 if (last) { 217 struct mbuf *n; 218 219 n = m_copy(m, 0, (int)M_COPYALL); 220 if (n != NULL) 221 (void) raw_append(last, ip, n); 222 /* XXX count dropped packet */ 223 INP_UNLOCK(last); 224 } 225 last = inp; 226 } 227 if (last != NULL) { 228 if (raw_append(last, ip, m) != 0) 229 ipstat.ips_delivered--; 230 INP_UNLOCK(last); 231 } else { 232 m_freem(m); 233 ipstat.ips_noproto++; 234 ipstat.ips_delivered--; 235 } 236 INP_INFO_RUNLOCK(&ripcbinfo); 237} 238 239/* 240 * Generate IP header and pass packet to ip_output. 241 * Tack on options user may have setup with control call. 242 */ 243int 244rip_output(struct mbuf *m, struct socket *so, u_long dst) 245{ 246 struct ip *ip; 247 struct inpcb *inp = sotoinpcb(so); 248 int flags = (so->so_options & SO_DONTROUTE) | IP_ALLOWBROADCAST; 249 250#ifdef MAC 251 mac_create_mbuf_from_socket(so, m); 252#endif 253 254 /* 255 * If the user handed us a complete IP packet, use it. 256 * Otherwise, allocate an mbuf for a header and fill it in. 257 */ 258 if ((inp->inp_flags & INP_HDRINCL) == 0) { 259 if (m->m_pkthdr.len + sizeof(struct ip) > IP_MAXPACKET) { 260 m_freem(m); 261 return(EMSGSIZE); 262 } 263 M_PREPEND(m, sizeof(struct ip), M_TRYWAIT); 264 if (m == NULL) 265 return(ENOBUFS); 266 ip = mtod(m, struct ip *); 267 ip->ip_tos = inp->inp_ip_tos; 268 ip->ip_off = 0; 269 ip->ip_p = inp->inp_ip_p; 270 ip->ip_len = m->m_pkthdr.len; 271 if (jailed(inp->inp_socket->so_cred)) 272 ip->ip_src.s_addr = 273 htonl(prison_getip(inp->inp_socket->so_cred)); 274 else 275 ip->ip_src = inp->inp_laddr; 276 ip->ip_dst.s_addr = dst; 277 ip->ip_ttl = inp->inp_ip_ttl; 278 } else { 279 if (m->m_pkthdr.len > IP_MAXPACKET) { 280 m_freem(m); 281 return(EMSGSIZE); 282 } 283 ip = mtod(m, struct ip *); 284 if (jailed(inp->inp_socket->so_cred)) { 285 if (ip->ip_src.s_addr != 286 htonl(prison_getip(inp->inp_socket->so_cred))) { 287 m_freem(m); 288 return (EPERM); 289 } 290 } 291 /* don't allow both user specified and setsockopt options, 292 and don't allow packet length sizes that will crash */ 293 if (((ip->ip_hl != (sizeof (*ip) >> 2)) 294 && inp->inp_options) 295 || (ip->ip_len > m->m_pkthdr.len) 296 || (ip->ip_len < (ip->ip_hl << 2))) { 297 m_freem(m); 298 return EINVAL; 299 } 300 if (ip->ip_id == 0) 301#ifdef RANDOM_IP_ID 302 ip->ip_id = ip_randomid(); 303#else 304 ip->ip_id = htons(ip_id++); 305#endif 306 /* XXX prevent ip_output from overwriting header fields */ 307 flags |= IP_RAWOUTPUT; 308 ipstat.ips_rawout++; 309 } 310 311 if (inp->inp_flags & INP_ONESBCAST) 312 flags |= IP_SENDONES; 313 314 return (ip_output(m, inp->inp_options, NULL, flags, 315 inp->inp_moptions, inp)); 316} 317 318/* 319 * Raw IP socket option processing. 320 * 321 * Note that access to all of the IP administrative functions here is 322 * implicitly protected by suser() as gaining access to a raw socket 323 * requires either that the thread pass a suser() check, or that it be 324 * passed a raw socket by another thread that has passed a suser() check. 325 * If FreeBSD moves to a more fine-grained access control mechanism, 326 * additional checks will need to be placed here if the raw IP attachment 327 * check is not equivilent the the check required for these 328 * administrative operations; in some cases, these checks are already 329 * present. 330 */ 331int 332rip_ctloutput(struct socket *so, struct sockopt *sopt) 333{ 334 struct inpcb *inp = sotoinpcb(so); 335 int error, optval; 336 337 if (sopt->sopt_level != IPPROTO_IP) 338 return (EINVAL); 339 340 error = 0; 341 342 switch (sopt->sopt_dir) { 343 case SOPT_GET: 344 switch (sopt->sopt_name) { 345 case IP_HDRINCL: 346 optval = inp->inp_flags & INP_HDRINCL; 347 error = sooptcopyout(sopt, &optval, sizeof optval); 348 break; 349 350 case IP_FW_ADD: /* ADD actually returns the body... */ 351 case IP_FW_GET: 352 if (IPFW_LOADED) 353 error = ip_fw_ctl_ptr(sopt); 354 else 355 error = ENOPROTOOPT; 356 break; 357 358 case IP_DUMMYNET_GET: 359 if (DUMMYNET_LOADED) 360 error = ip_dn_ctl_ptr(sopt); 361 else 362 error = ENOPROTOOPT; 363 break ; 364 365 case MRT_INIT: 366 case MRT_DONE: 367 case MRT_ADD_VIF: 368 case MRT_DEL_VIF: 369 case MRT_ADD_MFC: 370 case MRT_DEL_MFC: 371 case MRT_VERSION: 372 case MRT_ASSERT: 373 case MRT_API_SUPPORT: 374 case MRT_API_CONFIG: 375 case MRT_ADD_BW_UPCALL: 376 case MRT_DEL_BW_UPCALL: 377 error = ip_mrouter_get ? ip_mrouter_get(so, sopt) : 378 EOPNOTSUPP; 379 break; 380 381 default: 382 error = ip_ctloutput(so, sopt); 383 break; 384 } 385 break; 386 387 case SOPT_SET: 388 switch (sopt->sopt_name) { 389 case IP_HDRINCL: 390 error = sooptcopyin(sopt, &optval, sizeof optval, 391 sizeof optval); 392 if (error) 393 break; 394 if (optval) 395 inp->inp_flags |= INP_HDRINCL; 396 else 397 inp->inp_flags &= ~INP_HDRINCL; 398 break; 399 400 case IP_FW_ADD: 401 case IP_FW_DEL: 402 case IP_FW_FLUSH: 403 case IP_FW_ZERO: 404 case IP_FW_RESETLOG: 405 if (IPFW_LOADED) 406 error = ip_fw_ctl_ptr(sopt); 407 else 408 error = ENOPROTOOPT; 409 break; 410 411 case IP_DUMMYNET_CONFIGURE: 412 case IP_DUMMYNET_DEL: 413 case IP_DUMMYNET_FLUSH: 414 if (DUMMYNET_LOADED) 415 error = ip_dn_ctl_ptr(sopt); 416 else 417 error = ENOPROTOOPT ; 418 break ; 419 420 case IP_RSVP_ON: 421 error = ip_rsvp_init(so); 422 break; 423 424 case IP_RSVP_OFF: 425 error = ip_rsvp_done(); 426 break; 427 428 case IP_RSVP_VIF_ON: 429 case IP_RSVP_VIF_OFF: 430 error = ip_rsvp_vif ? 431 ip_rsvp_vif(so, sopt) : EINVAL; 432 break; 433 434 case MRT_INIT: 435 case MRT_DONE: 436 case MRT_ADD_VIF: 437 case MRT_DEL_VIF: 438 case MRT_ADD_MFC: 439 case MRT_DEL_MFC: 440 case MRT_VERSION: 441 case MRT_ASSERT: 442 case MRT_API_SUPPORT: 443 case MRT_API_CONFIG: 444 case MRT_ADD_BW_UPCALL: 445 case MRT_DEL_BW_UPCALL: 446 error = ip_mrouter_set ? ip_mrouter_set(so, sopt) : 447 EOPNOTSUPP; 448 break; 449 450 default: 451 error = ip_ctloutput(so, sopt); 452 break; 453 } 454 break; 455 } 456 457 return (error); 458} 459 460/* 461 * This function exists solely to receive the PRC_IFDOWN messages which 462 * are sent by if_down(). It looks for an ifaddr whose ifa_addr is sa, 463 * and calls in_ifadown() to remove all routes corresponding to that address. 464 * It also receives the PRC_IFUP messages from if_up() and reinstalls the 465 * interface routes. 466 */ 467void 468rip_ctlinput(int cmd, struct sockaddr *sa, void *vip) 469{ 470 struct in_ifaddr *ia; 471 struct ifnet *ifp; 472 int err; 473 int flags; 474 475 switch (cmd) { 476 case PRC_IFDOWN: 477 TAILQ_FOREACH(ia, &in_ifaddrhead, ia_link) { 478 if (ia->ia_ifa.ifa_addr == sa 479 && (ia->ia_flags & IFA_ROUTE)) { 480 /* 481 * in_ifscrub kills the interface route. 482 */ 483 in_ifscrub(ia->ia_ifp, ia); 484 /* 485 * in_ifadown gets rid of all the rest of 486 * the routes. This is not quite the right 487 * thing to do, but at least if we are running 488 * a routing process they will come back. 489 */ 490 in_ifadown(&ia->ia_ifa, 0); 491 break; 492 } 493 } 494 break; 495 496 case PRC_IFUP: 497 TAILQ_FOREACH(ia, &in_ifaddrhead, ia_link) { 498 if (ia->ia_ifa.ifa_addr == sa) 499 break; 500 } 501 if (ia == 0 || (ia->ia_flags & IFA_ROUTE)) 502 return; 503 flags = RTF_UP; 504 ifp = ia->ia_ifa.ifa_ifp; 505 506 if ((ifp->if_flags & IFF_LOOPBACK) 507 || (ifp->if_flags & IFF_POINTOPOINT)) 508 flags |= RTF_HOST; 509 510 err = rtinit(&ia->ia_ifa, RTM_ADD, flags); 511 if (err == 0) 512 ia->ia_flags |= IFA_ROUTE; 513 break; 514 } 515} 516 517u_long rip_sendspace = RIPSNDQ; 518u_long rip_recvspace = RIPRCVQ; 519 520SYSCTL_INT(_net_inet_raw, OID_AUTO, maxdgram, CTLFLAG_RW, 521 &rip_sendspace, 0, "Maximum outgoing raw IP datagram size"); 522SYSCTL_INT(_net_inet_raw, OID_AUTO, recvspace, CTLFLAG_RW, 523 &rip_recvspace, 0, "Maximum space for incoming raw IP datagrams"); 524 525static int 526rip_attach(struct socket *so, int proto, struct thread *td) 527{ 528 struct inpcb *inp; 529 int error; 530 531 /* XXX why not lower? */ 532 INP_INFO_WLOCK(&ripcbinfo); 533 inp = sotoinpcb(so); 534 if (inp) { 535 /* XXX counter, printf */ 536 INP_INFO_WUNLOCK(&ripcbinfo); 537 return EINVAL; 538 } 539 if (td && jailed(td->td_ucred) && !jail_allow_raw_sockets) { 540 INP_INFO_WUNLOCK(&ripcbinfo); 541 return (EPERM); 542 } 543 if (td && (error = suser_cred(td->td_ucred, PRISON_ROOT)) != 0) { 544 INP_INFO_WUNLOCK(&ripcbinfo); 545 return error; 546 } 547 if (proto >= IPPROTO_MAX || proto < 0) { 548 INP_INFO_WUNLOCK(&ripcbinfo); 549 return EPROTONOSUPPORT; 550 } 551 552 error = soreserve(so, rip_sendspace, rip_recvspace); 553 if (error) { 554 INP_INFO_WUNLOCK(&ripcbinfo); 555 return error; 556 } 557 error = in_pcballoc(so, &ripcbinfo, "rawinp"); 558 if (error) { 559 INP_INFO_WUNLOCK(&ripcbinfo); 560 return error; 561 } 562 inp = (struct inpcb *)so->so_pcb; 563 INP_LOCK(inp); 564 INP_INFO_WUNLOCK(&ripcbinfo); 565 inp->inp_vflag |= INP_IPV4; 566 inp->inp_ip_p = proto; 567 inp->inp_ip_ttl = ip_defttl; 568 INP_UNLOCK(inp); 569 return 0; 570} 571 572static void 573rip_pcbdetach(struct socket *so, struct inpcb *inp) 574{ 575 INP_INFO_WLOCK_ASSERT(&ripcbinfo); 576 INP_LOCK_ASSERT(inp); 577 578 if (so == ip_mrouter && ip_mrouter_done) 579 ip_mrouter_done(); 580 if (ip_rsvp_force_done) 581 ip_rsvp_force_done(so); 582 if (so == ip_rsvpd) 583 ip_rsvp_done(); 584 in_pcbdetach(inp); 585} 586 587static int 588rip_detach(struct socket *so) 589{ 590 struct inpcb *inp; 591 592 INP_INFO_WLOCK(&ripcbinfo); 593 inp = sotoinpcb(so); 594 if (inp == 0) { 595 /* XXX counter, printf */ 596 INP_INFO_WUNLOCK(&ripcbinfo); 597 return EINVAL; 598 } 599 INP_LOCK(inp); 600 rip_pcbdetach(so, inp); 601 INP_INFO_WUNLOCK(&ripcbinfo); 602 return 0; 603} 604 605static int 606rip_abort(struct socket *so) 607{ 608 struct inpcb *inp; 609 610 INP_INFO_WLOCK(&ripcbinfo); 611 inp = sotoinpcb(so); 612 if (inp == 0) { 613 INP_INFO_WUNLOCK(&ripcbinfo); 614 return EINVAL; /* ??? possible? panic instead? */ 615 } 616 INP_LOCK(inp); 617 soisdisconnected(so); 618 if (so->so_state & SS_NOFDREF) 619 rip_pcbdetach(so, inp); 620 else 621 INP_UNLOCK(inp); 622 INP_INFO_WUNLOCK(&ripcbinfo); 623 return 0; 624} 625 626static int 627rip_disconnect(struct socket *so) 628{ 629 if ((so->so_state & SS_ISCONNECTED) == 0) 630 return ENOTCONN; 631 return rip_abort(so); 632} 633 634static int 635rip_bind(struct socket *so, struct sockaddr *nam, struct thread *td) 636{ 637 struct sockaddr_in *addr = (struct sockaddr_in *)nam; 638 struct inpcb *inp; 639 640 if (nam->sa_len != sizeof(*addr)) 641 return EINVAL; 642 643 if (jailed(td->td_ucred)) { 644 if (addr->sin_addr.s_addr == INADDR_ANY) 645 addr->sin_addr.s_addr = 646 htonl(prison_getip(td->td_ucred)); 647 if (htonl(prison_getip(td->td_ucred)) != addr->sin_addr.s_addr) 648 return (EADDRNOTAVAIL); 649 } 650 651 if (TAILQ_EMPTY(&ifnet) || 652 (addr->sin_family != AF_INET && addr->sin_family != AF_IMPLINK) || 653 (addr->sin_addr.s_addr && 654 ifa_ifwithaddr((struct sockaddr *)addr) == 0)) 655 return EADDRNOTAVAIL; 656 657 INP_INFO_WLOCK(&ripcbinfo); 658 inp = sotoinpcb(so); 659 if (inp == 0) { 660 INP_INFO_WUNLOCK(&ripcbinfo); 661 return EINVAL; 662 } 663 INP_LOCK(inp); 664 inp->inp_laddr = addr->sin_addr; 665 INP_UNLOCK(inp); 666 INP_INFO_WUNLOCK(&ripcbinfo); 667 return 0; 668} 669 670static int 671rip_connect(struct socket *so, struct sockaddr *nam, struct thread *td) 672{ 673 struct sockaddr_in *addr = (struct sockaddr_in *)nam; 674 struct inpcb *inp; 675 676 if (nam->sa_len != sizeof(*addr)) 677 return EINVAL; 678 if (TAILQ_EMPTY(&ifnet)) 679 return EADDRNOTAVAIL; 680 if (addr->sin_family != AF_INET && addr->sin_family != AF_IMPLINK) 681 return EAFNOSUPPORT; 682 683 INP_INFO_WLOCK(&ripcbinfo); 684 inp = sotoinpcb(so); 685 if (inp == 0) { 686 INP_INFO_WUNLOCK(&ripcbinfo); 687 return EINVAL; 688 } 689 INP_LOCK(inp); 690 inp->inp_faddr = addr->sin_addr; 691 soisconnected(so); 692 INP_UNLOCK(inp); 693 INP_INFO_WUNLOCK(&ripcbinfo); 694 return 0; 695} 696 697static int 698rip_shutdown(struct socket *so) 699{ 700 struct inpcb *inp; 701 702 INP_INFO_RLOCK(&ripcbinfo); 703 inp = sotoinpcb(so); 704 if (inp == 0) { 705 INP_INFO_RUNLOCK(&ripcbinfo); 706 return EINVAL; 707 } 708 INP_LOCK(inp); 709 INP_INFO_RUNLOCK(&ripcbinfo); 710 socantsendmore(so); 711 INP_UNLOCK(inp); 712 return 0; 713} 714 715static int 716rip_send(struct socket *so, int flags, struct mbuf *m, struct sockaddr *nam, 717 struct mbuf *control, struct thread *td) 718{ 719 struct inpcb *inp; 720 u_long dst; 721 int ret; 722 723 INP_INFO_WLOCK(&ripcbinfo); 724 inp = sotoinpcb(so); 725 if (so->so_state & SS_ISCONNECTED) { 726 if (nam) { 727 INP_INFO_WUNLOCK(&ripcbinfo); 728 m_freem(m); 729 return EISCONN; 730 } 731 dst = inp->inp_faddr.s_addr; 732 } else { 733 if (nam == NULL) { 734 INP_INFO_WUNLOCK(&ripcbinfo); 735 m_freem(m); 736 return ENOTCONN; 737 } 738 dst = ((struct sockaddr_in *)nam)->sin_addr.s_addr; 739 } 740 INP_LOCK(inp); 741 ret = rip_output(m, so, dst); 742 INP_UNLOCK(inp); 743 INP_INFO_WUNLOCK(&ripcbinfo); 744 return ret; 745} 746 747static int 748rip_pcblist(SYSCTL_HANDLER_ARGS) 749{ 750 int error, i, n; 751 struct inpcb *inp, **inp_list; 752 inp_gen_t gencnt; 753 struct xinpgen xig; 754 755 /* 756 * The process of preparing the TCB list is too time-consuming and 757 * resource-intensive to repeat twice on every request. 758 */ 759 if (req->oldptr == 0) { 760 n = ripcbinfo.ipi_count; 761 req->oldidx = 2 * (sizeof xig) 762 + (n + n/8) * sizeof(struct xinpcb); 763 return 0; 764 } 765 766 if (req->newptr != 0) 767 return EPERM; 768 769 /* 770 * OK, now we're committed to doing something. 771 */ 772 INP_INFO_RLOCK(&ripcbinfo); 773 gencnt = ripcbinfo.ipi_gencnt; 774 n = ripcbinfo.ipi_count; 775 INP_INFO_RUNLOCK(&ripcbinfo); 776 777 xig.xig_len = sizeof xig; 778 xig.xig_count = n; 779 xig.xig_gen = gencnt; 780 xig.xig_sogen = so_gencnt; 781 error = SYSCTL_OUT(req, &xig, sizeof xig); 782 if (error) 783 return error; 784 785 inp_list = malloc(n * sizeof *inp_list, M_TEMP, M_WAITOK); 786 if (inp_list == 0) 787 return ENOMEM; 788 789 INP_INFO_RLOCK(&ripcbinfo); 790 for (inp = LIST_FIRST(ripcbinfo.listhead), i = 0; inp && i < n; 791 inp = LIST_NEXT(inp, inp_list)) { 792 INP_LOCK(inp); 793 if (inp->inp_gencnt <= gencnt && 794 cr_canseesocket(req->td->td_ucred, inp->inp_socket) == 0) { 795 /* XXX held references? */ 796 inp_list[i++] = inp; 797 } 798 INP_UNLOCK(inp); 799 } 800 INP_INFO_RUNLOCK(&ripcbinfo); 801 n = i; 802 803 error = 0; 804 for (i = 0; i < n; i++) { 805 inp = inp_list[i]; 806 if (inp->inp_gencnt <= gencnt) { 807 struct xinpcb xi; 808 xi.xi_len = sizeof xi; 809 /* XXX should avoid extra copy */ 810 bcopy(inp, &xi.xi_inp, sizeof *inp); 811 if (inp->inp_socket) 812 sotoxsocket(inp->inp_socket, &xi.xi_socket); 813 error = SYSCTL_OUT(req, &xi, sizeof xi); 814 } 815 } 816 if (!error) { 817 /* 818 * Give the user an updated idea of our state. 819 * If the generation differs from what we told 820 * her before, she knows that something happened 821 * while we were processing this request, and it 822 * might be necessary to retry. 823 */ 824 INP_INFO_RLOCK(&ripcbinfo); 825 xig.xig_gen = ripcbinfo.ipi_gencnt; 826 xig.xig_sogen = so_gencnt; 827 xig.xig_count = ripcbinfo.ipi_count; 828 INP_INFO_RUNLOCK(&ripcbinfo); 829 error = SYSCTL_OUT(req, &xig, sizeof xig); 830 } 831 free(inp_list, M_TEMP); 832 return error; 833} 834 835/* 836 * This is the wrapper function for in_setsockaddr. We just pass down 837 * the pcbinfo for in_setpeeraddr to lock. 838 */ 839static int 840rip_sockaddr(struct socket *so, struct sockaddr **nam) 841{ 842 return (in_setsockaddr(so, nam, &ripcbinfo)); 843} 844 845/* 846 * This is the wrapper function for in_setpeeraddr. We just pass down 847 * the pcbinfo for in_setpeeraddr to lock. 848 */ 849static int 850rip_peeraddr(struct socket *so, struct sockaddr **nam) 851{ 852 return (in_setpeeraddr(so, nam, &ripcbinfo)); 853} 854 855 856SYSCTL_PROC(_net_inet_raw, OID_AUTO/*XXX*/, pcblist, CTLFLAG_RD, 0, 0, 857 rip_pcblist, "S,xinpcb", "List of active raw IP sockets"); 858 859struct pr_usrreqs rip_usrreqs = { 860 rip_abort, pru_accept_notsupp, rip_attach, rip_bind, rip_connect, 861 pru_connect2_notsupp, in_control, rip_detach, rip_disconnect, 862 pru_listen_notsupp, rip_peeraddr, pru_rcvd_notsupp, 863 pru_rcvoob_notsupp, rip_send, pru_sense_null, rip_shutdown, 864 rip_sockaddr, sosend, soreceive, sopoll, in_pcbsosetlabel 865}; 866