raw_ip.c revision 134122
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 134122 2004-08-21 17:38:57Z csjp $ 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_ALLOWBROADCAST; 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_DONTWAIT); 264 if (m == NULL) 265 return(ENOBUFS); 266 267 INP_LOCK(inp); 268 ip = mtod(m, struct ip *); 269 ip->ip_tos = inp->inp_ip_tos; 270 ip->ip_off = 0; 271 ip->ip_p = inp->inp_ip_p; 272 ip->ip_len = m->m_pkthdr.len; 273 if (jailed(inp->inp_socket->so_cred)) 274 ip->ip_src.s_addr = 275 htonl(prison_getip(inp->inp_socket->so_cred)); 276 else 277 ip->ip_src = inp->inp_laddr; 278 ip->ip_dst.s_addr = dst; 279 ip->ip_ttl = inp->inp_ip_ttl; 280 } else { 281 if (m->m_pkthdr.len > IP_MAXPACKET) { 282 m_freem(m); 283 return(EMSGSIZE); 284 } 285 INP_LOCK(inp); 286 ip = mtod(m, struct ip *); 287 if (jailed(inp->inp_socket->so_cred)) { 288 if (ip->ip_src.s_addr != 289 htonl(prison_getip(inp->inp_socket->so_cred))) { 290 INP_UNLOCK(inp); 291 m_freem(m); 292 return (EPERM); 293 } 294 } 295 /* don't allow both user specified and setsockopt options, 296 and don't allow packet length sizes that will crash */ 297 if (((ip->ip_hl != (sizeof (*ip) >> 2)) 298 && inp->inp_options) 299 || (ip->ip_len > m->m_pkthdr.len) 300 || (ip->ip_len < (ip->ip_hl << 2))) { 301 INP_UNLOCK(inp); 302 m_freem(m); 303 return EINVAL; 304 } 305 if (ip->ip_id == 0) 306 ip->ip_id = ip_newid(); 307 /* XXX prevent ip_output from overwriting header fields */ 308 flags |= IP_RAWOUTPUT; 309 ipstat.ips_rawout++; 310 } 311 312 if (inp->inp_flags & INP_ONESBCAST) 313 flags |= IP_SENDONES; 314 315#ifdef MAC 316 mac_create_mbuf_from_inpcb(inp, m); 317#endif 318 319 error = ip_output(m, inp->inp_options, NULL, flags, 320 inp->inp_moptions, inp); 321 INP_UNLOCK(inp); 322 return error; 323} 324 325/* 326 * Raw IP socket option processing. 327 * 328 * Note that access to all of the IP administrative functions here is 329 * implicitly protected by suser() as gaining access to a raw socket 330 * requires either that the thread pass a suser() check, or that it be 331 * passed a raw socket by another thread that has passed a suser() check. 332 * If FreeBSD moves to a more fine-grained access control mechanism, 333 * additional checks will need to be placed here if the raw IP attachment 334 * check is not equivilent the the check required for these 335 * administrative operations; in some cases, these checks are already 336 * present. 337 */ 338int 339rip_ctloutput(struct socket *so, struct sockopt *sopt) 340{ 341 struct inpcb *inp = sotoinpcb(so); 342 int error, optval; 343 344 if (sopt->sopt_level != IPPROTO_IP) 345 return (EINVAL); 346 347 /* 348 * Even though super-user is required to create a raw socket, the 349 * calling cred could be prison root. If so we want to restrict the 350 * access to IP_HDRINCL only. 351 */ 352 if (sopt->sopt_name != IP_HDRINCL) { 353 error = suser(curthread); 354 if (error != 0) 355 return (error); 356 } 357 error = 0; 358 359 switch (sopt->sopt_dir) { 360 case SOPT_GET: 361 switch (sopt->sopt_name) { 362 case IP_HDRINCL: 363 optval = inp->inp_flags & INP_HDRINCL; 364 error = sooptcopyout(sopt, &optval, sizeof optval); 365 break; 366 367 case IP_FW_ADD: /* ADD actually returns the body... */ 368 case IP_FW_GET: 369 case IP_FW_TABLE_GETSIZE: 370 case IP_FW_TABLE_LIST: 371 if (ip_fw_ctl_ptr != NULL) 372 error = ip_fw_ctl_ptr(sopt); 373 else 374 error = ENOPROTOOPT; 375 break; 376 377 case IP_DUMMYNET_GET: 378 if (ip_dn_ctl_ptr != NULL) 379 error = ip_dn_ctl_ptr(sopt); 380 else 381 error = ENOPROTOOPT; 382 break ; 383 384 case MRT_INIT: 385 case MRT_DONE: 386 case MRT_ADD_VIF: 387 case MRT_DEL_VIF: 388 case MRT_ADD_MFC: 389 case MRT_DEL_MFC: 390 case MRT_VERSION: 391 case MRT_ASSERT: 392 case MRT_API_SUPPORT: 393 case MRT_API_CONFIG: 394 case MRT_ADD_BW_UPCALL: 395 case MRT_DEL_BW_UPCALL: 396 error = ip_mrouter_get ? ip_mrouter_get(so, sopt) : 397 EOPNOTSUPP; 398 break; 399 400 default: 401 error = ip_ctloutput(so, sopt); 402 break; 403 } 404 break; 405 406 case SOPT_SET: 407 switch (sopt->sopt_name) { 408 case IP_HDRINCL: 409 error = sooptcopyin(sopt, &optval, sizeof optval, 410 sizeof optval); 411 if (error) 412 break; 413 if (optval) 414 inp->inp_flags |= INP_HDRINCL; 415 else 416 inp->inp_flags &= ~INP_HDRINCL; 417 break; 418 419 case IP_FW_ADD: 420 case IP_FW_DEL: 421 case IP_FW_FLUSH: 422 case IP_FW_ZERO: 423 case IP_FW_RESETLOG: 424 case IP_FW_TABLE_ADD: 425 case IP_FW_TABLE_DEL: 426 case IP_FW_TABLE_FLUSH: 427 if (ip_fw_ctl_ptr != NULL) 428 error = ip_fw_ctl_ptr(sopt); 429 else 430 error = ENOPROTOOPT; 431 break; 432 433 case IP_DUMMYNET_CONFIGURE: 434 case IP_DUMMYNET_DEL: 435 case IP_DUMMYNET_FLUSH: 436 if (ip_dn_ctl_ptr != NULL) 437 error = ip_dn_ctl_ptr(sopt); 438 else 439 error = ENOPROTOOPT ; 440 break ; 441 442 case IP_RSVP_ON: 443 error = ip_rsvp_init(so); 444 break; 445 446 case IP_RSVP_OFF: 447 error = ip_rsvp_done(); 448 break; 449 450 case IP_RSVP_VIF_ON: 451 case IP_RSVP_VIF_OFF: 452 error = ip_rsvp_vif ? 453 ip_rsvp_vif(so, sopt) : EINVAL; 454 break; 455 456 case MRT_INIT: 457 case MRT_DONE: 458 case MRT_ADD_VIF: 459 case MRT_DEL_VIF: 460 case MRT_ADD_MFC: 461 case MRT_DEL_MFC: 462 case MRT_VERSION: 463 case MRT_ASSERT: 464 case MRT_API_SUPPORT: 465 case MRT_API_CONFIG: 466 case MRT_ADD_BW_UPCALL: 467 case MRT_DEL_BW_UPCALL: 468 error = ip_mrouter_set ? ip_mrouter_set(so, sopt) : 469 EOPNOTSUPP; 470 break; 471 472 default: 473 error = ip_ctloutput(so, sopt); 474 break; 475 } 476 break; 477 } 478 479 return (error); 480} 481 482/* 483 * This function exists solely to receive the PRC_IFDOWN messages which 484 * are sent by if_down(). It looks for an ifaddr whose ifa_addr is sa, 485 * and calls in_ifadown() to remove all routes corresponding to that address. 486 * It also receives the PRC_IFUP messages from if_up() and reinstalls the 487 * interface routes. 488 */ 489void 490rip_ctlinput(int cmd, struct sockaddr *sa, void *vip) 491{ 492 struct in_ifaddr *ia; 493 struct ifnet *ifp; 494 int err; 495 int flags; 496 497 switch (cmd) { 498 case PRC_IFDOWN: 499 TAILQ_FOREACH(ia, &in_ifaddrhead, ia_link) { 500 if (ia->ia_ifa.ifa_addr == sa 501 && (ia->ia_flags & IFA_ROUTE)) { 502 /* 503 * in_ifscrub kills the interface route. 504 */ 505 in_ifscrub(ia->ia_ifp, ia); 506 /* 507 * in_ifadown gets rid of all the rest of 508 * the routes. This is not quite the right 509 * thing to do, but at least if we are running 510 * a routing process they will come back. 511 */ 512 in_ifadown(&ia->ia_ifa, 0); 513 break; 514 } 515 } 516 break; 517 518 case PRC_IFUP: 519 TAILQ_FOREACH(ia, &in_ifaddrhead, ia_link) { 520 if (ia->ia_ifa.ifa_addr == sa) 521 break; 522 } 523 if (ia == 0 || (ia->ia_flags & IFA_ROUTE)) 524 return; 525 flags = RTF_UP; 526 ifp = ia->ia_ifa.ifa_ifp; 527 528 if ((ifp->if_flags & IFF_LOOPBACK) 529 || (ifp->if_flags & IFF_POINTOPOINT)) 530 flags |= RTF_HOST; 531 532 err = rtinit(&ia->ia_ifa, RTM_ADD, flags); 533 if (err == 0) 534 ia->ia_flags |= IFA_ROUTE; 535 break; 536 } 537} 538 539u_long rip_sendspace = RIPSNDQ; 540u_long rip_recvspace = RIPRCVQ; 541 542SYSCTL_INT(_net_inet_raw, OID_AUTO, maxdgram, CTLFLAG_RW, 543 &rip_sendspace, 0, "Maximum outgoing raw IP datagram size"); 544SYSCTL_INT(_net_inet_raw, OID_AUTO, recvspace, CTLFLAG_RW, 545 &rip_recvspace, 0, "Maximum space for incoming raw IP datagrams"); 546 547static int 548rip_attach(struct socket *so, int proto, struct thread *td) 549{ 550 struct inpcb *inp; 551 int error; 552 553 /* XXX why not lower? */ 554 INP_INFO_WLOCK(&ripcbinfo); 555 inp = sotoinpcb(so); 556 if (inp) { 557 /* XXX counter, printf */ 558 INP_INFO_WUNLOCK(&ripcbinfo); 559 return EINVAL; 560 } 561 if (td && jailed(td->td_ucred) && !jail_allow_raw_sockets) { 562 INP_INFO_WUNLOCK(&ripcbinfo); 563 return (EPERM); 564 } 565 if (td && (error = suser_cred(td->td_ucred, SUSER_ALLOWJAIL)) != 0) { 566 INP_INFO_WUNLOCK(&ripcbinfo); 567 return error; 568 } 569 if (proto >= IPPROTO_MAX || proto < 0) { 570 INP_INFO_WUNLOCK(&ripcbinfo); 571 return EPROTONOSUPPORT; 572 } 573 574 error = soreserve(so, rip_sendspace, rip_recvspace); 575 if (error) { 576 INP_INFO_WUNLOCK(&ripcbinfo); 577 return error; 578 } 579 error = in_pcballoc(so, &ripcbinfo, "rawinp"); 580 if (error) { 581 INP_INFO_WUNLOCK(&ripcbinfo); 582 return error; 583 } 584 inp = (struct inpcb *)so->so_pcb; 585 INP_LOCK(inp); 586 INP_INFO_WUNLOCK(&ripcbinfo); 587 inp->inp_vflag |= INP_IPV4; 588 inp->inp_ip_p = proto; 589 inp->inp_ip_ttl = ip_defttl; 590 INP_UNLOCK(inp); 591 return 0; 592} 593 594static void 595rip_pcbdetach(struct socket *so, struct inpcb *inp) 596{ 597 INP_INFO_WLOCK_ASSERT(&ripcbinfo); 598 INP_LOCK_ASSERT(inp); 599 600 if (so == ip_mrouter && ip_mrouter_done) 601 ip_mrouter_done(); 602 if (ip_rsvp_force_done) 603 ip_rsvp_force_done(so); 604 if (so == ip_rsvpd) 605 ip_rsvp_done(); 606 in_pcbdetach(inp); 607} 608 609static int 610rip_detach(struct socket *so) 611{ 612 struct inpcb *inp; 613 614 INP_INFO_WLOCK(&ripcbinfo); 615 inp = sotoinpcb(so); 616 if (inp == 0) { 617 /* XXX counter, printf */ 618 INP_INFO_WUNLOCK(&ripcbinfo); 619 return EINVAL; 620 } 621 INP_LOCK(inp); 622 rip_pcbdetach(so, inp); 623 INP_INFO_WUNLOCK(&ripcbinfo); 624 return 0; 625} 626 627static int 628rip_abort(struct socket *so) 629{ 630 struct inpcb *inp; 631 632 INP_INFO_WLOCK(&ripcbinfo); 633 inp = sotoinpcb(so); 634 if (inp == 0) { 635 INP_INFO_WUNLOCK(&ripcbinfo); 636 return EINVAL; /* ??? possible? panic instead? */ 637 } 638 INP_LOCK(inp); 639 soisdisconnected(so); 640 if (so->so_state & SS_NOFDREF) 641 rip_pcbdetach(so, inp); 642 else 643 INP_UNLOCK(inp); 644 INP_INFO_WUNLOCK(&ripcbinfo); 645 return 0; 646} 647 648static int 649rip_disconnect(struct socket *so) 650{ 651 if ((so->so_state & SS_ISCONNECTED) == 0) 652 return ENOTCONN; 653 return rip_abort(so); 654} 655 656static int 657rip_bind(struct socket *so, struct sockaddr *nam, struct thread *td) 658{ 659 struct sockaddr_in *addr = (struct sockaddr_in *)nam; 660 struct inpcb *inp; 661 662 if (nam->sa_len != sizeof(*addr)) 663 return EINVAL; 664 665 if (jailed(td->td_ucred)) { 666 if (addr->sin_addr.s_addr == INADDR_ANY) 667 addr->sin_addr.s_addr = 668 htonl(prison_getip(td->td_ucred)); 669 if (htonl(prison_getip(td->td_ucred)) != addr->sin_addr.s_addr) 670 return (EADDRNOTAVAIL); 671 } 672 673 if (TAILQ_EMPTY(&ifnet) || 674 (addr->sin_family != AF_INET && addr->sin_family != AF_IMPLINK) || 675 (addr->sin_addr.s_addr && 676 ifa_ifwithaddr((struct sockaddr *)addr) == 0)) 677 return EADDRNOTAVAIL; 678 679 INP_INFO_WLOCK(&ripcbinfo); 680 inp = sotoinpcb(so); 681 if (inp == 0) { 682 INP_INFO_WUNLOCK(&ripcbinfo); 683 return EINVAL; 684 } 685 INP_LOCK(inp); 686 inp->inp_laddr = addr->sin_addr; 687 INP_UNLOCK(inp); 688 INP_INFO_WUNLOCK(&ripcbinfo); 689 return 0; 690} 691 692static int 693rip_connect(struct socket *so, struct sockaddr *nam, struct thread *td) 694{ 695 struct sockaddr_in *addr = (struct sockaddr_in *)nam; 696 struct inpcb *inp; 697 698 if (nam->sa_len != sizeof(*addr)) 699 return EINVAL; 700 if (TAILQ_EMPTY(&ifnet)) 701 return EADDRNOTAVAIL; 702 if (addr->sin_family != AF_INET && addr->sin_family != AF_IMPLINK) 703 return EAFNOSUPPORT; 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_faddr = addr->sin_addr; 713 soisconnected(so); 714 INP_UNLOCK(inp); 715 INP_INFO_WUNLOCK(&ripcbinfo); 716 return 0; 717} 718 719static int 720rip_shutdown(struct socket *so) 721{ 722 struct inpcb *inp; 723 724 INP_INFO_RLOCK(&ripcbinfo); 725 inp = sotoinpcb(so); 726 if (inp == 0) { 727 INP_INFO_RUNLOCK(&ripcbinfo); 728 return EINVAL; 729 } 730 INP_LOCK(inp); 731 INP_INFO_RUNLOCK(&ripcbinfo); 732 socantsendmore(so); 733 INP_UNLOCK(inp); 734 return 0; 735} 736 737static int 738rip_send(struct socket *so, int flags, struct mbuf *m, struct sockaddr *nam, 739 struct mbuf *control, struct thread *td) 740{ 741 struct inpcb *inp; 742 u_long dst; 743 int ret; 744 745 INP_INFO_WLOCK(&ripcbinfo); 746 inp = sotoinpcb(so); 747 if (so->so_state & SS_ISCONNECTED) { 748 if (nam) { 749 INP_INFO_WUNLOCK(&ripcbinfo); 750 m_freem(m); 751 return EISCONN; 752 } 753 dst = inp->inp_faddr.s_addr; 754 } else { 755 if (nam == NULL) { 756 INP_INFO_WUNLOCK(&ripcbinfo); 757 m_freem(m); 758 return ENOTCONN; 759 } 760 dst = ((struct sockaddr_in *)nam)->sin_addr.s_addr; 761 } 762 ret = rip_output(m, so, dst); 763 INP_INFO_WUNLOCK(&ripcbinfo); 764 return ret; 765} 766 767static int 768rip_pcblist(SYSCTL_HANDLER_ARGS) 769{ 770 int error, i, n; 771 struct inpcb *inp, **inp_list; 772 inp_gen_t gencnt; 773 struct xinpgen xig; 774 775 /* 776 * The process of preparing the TCB list is too time-consuming and 777 * resource-intensive to repeat twice on every request. 778 */ 779 if (req->oldptr == 0) { 780 n = ripcbinfo.ipi_count; 781 req->oldidx = 2 * (sizeof xig) 782 + (n + n/8) * sizeof(struct xinpcb); 783 return 0; 784 } 785 786 if (req->newptr != 0) 787 return EPERM; 788 789 /* 790 * OK, now we're committed to doing something. 791 */ 792 INP_INFO_RLOCK(&ripcbinfo); 793 gencnt = ripcbinfo.ipi_gencnt; 794 n = ripcbinfo.ipi_count; 795 INP_INFO_RUNLOCK(&ripcbinfo); 796 797 xig.xig_len = sizeof xig; 798 xig.xig_count = n; 799 xig.xig_gen = gencnt; 800 xig.xig_sogen = so_gencnt; 801 error = SYSCTL_OUT(req, &xig, sizeof xig); 802 if (error) 803 return error; 804 805 inp_list = malloc(n * sizeof *inp_list, M_TEMP, M_WAITOK); 806 if (inp_list == 0) 807 return ENOMEM; 808 809 INP_INFO_RLOCK(&ripcbinfo); 810 for (inp = LIST_FIRST(ripcbinfo.listhead), i = 0; inp && i < n; 811 inp = LIST_NEXT(inp, inp_list)) { 812 INP_LOCK(inp); 813 if (inp->inp_gencnt <= gencnt && 814 cr_canseesocket(req->td->td_ucred, inp->inp_socket) == 0) { 815 /* XXX held references? */ 816 inp_list[i++] = inp; 817 } 818 INP_UNLOCK(inp); 819 } 820 INP_INFO_RUNLOCK(&ripcbinfo); 821 n = i; 822 823 error = 0; 824 for (i = 0; i < n; i++) { 825 inp = inp_list[i]; 826 if (inp->inp_gencnt <= gencnt) { 827 struct xinpcb xi; 828 xi.xi_len = sizeof xi; 829 /* XXX should avoid extra copy */ 830 bcopy(inp, &xi.xi_inp, sizeof *inp); 831 if (inp->inp_socket) 832 sotoxsocket(inp->inp_socket, &xi.xi_socket); 833 error = SYSCTL_OUT(req, &xi, sizeof xi); 834 } 835 } 836 if (!error) { 837 /* 838 * Give the user an updated idea of our state. 839 * If the generation differs from what we told 840 * her before, she knows that something happened 841 * while we were processing this request, and it 842 * might be necessary to retry. 843 */ 844 INP_INFO_RLOCK(&ripcbinfo); 845 xig.xig_gen = ripcbinfo.ipi_gencnt; 846 xig.xig_sogen = so_gencnt; 847 xig.xig_count = ripcbinfo.ipi_count; 848 INP_INFO_RUNLOCK(&ripcbinfo); 849 error = SYSCTL_OUT(req, &xig, sizeof xig); 850 } 851 free(inp_list, M_TEMP); 852 return error; 853} 854 855/* 856 * This is the wrapper function for in_setsockaddr. We just pass down 857 * the pcbinfo for in_setpeeraddr to lock. 858 */ 859static int 860rip_sockaddr(struct socket *so, struct sockaddr **nam) 861{ 862 return (in_setsockaddr(so, nam, &ripcbinfo)); 863} 864 865/* 866 * This is the wrapper function for in_setpeeraddr. We just pass down 867 * the pcbinfo for in_setpeeraddr to lock. 868 */ 869static int 870rip_peeraddr(struct socket *so, struct sockaddr **nam) 871{ 872 return (in_setpeeraddr(so, nam, &ripcbinfo)); 873} 874 875 876SYSCTL_PROC(_net_inet_raw, OID_AUTO/*XXX*/, pcblist, CTLFLAG_RD, 0, 0, 877 rip_pcblist, "S,xinpcb", "List of active raw IP sockets"); 878 879struct pr_usrreqs rip_usrreqs = { 880 rip_abort, pru_accept_notsupp, rip_attach, rip_bind, rip_connect, 881 pru_connect2_notsupp, in_control, rip_detach, rip_disconnect, 882 pru_listen_notsupp, rip_peeraddr, pru_rcvd_notsupp, 883 pru_rcvoob_notsupp, rip_send, pru_sense_null, rip_shutdown, 884 rip_sockaddr, sosend, soreceive, sopoll, in_pcbsosetlabel 885}; 886