raw_ip.c revision 188144
1/*- 2 * Copyright (c) 1982, 1986, 1988, 1993 3 * The Regents of the University of California. 4 * All rights reserved. 5 * 6 * Redistribution and use in source and binary forms, with or without 7 * modification, are permitted provided that the following conditions 8 * are met: 9 * 1. Redistributions of source code must retain the above copyright 10 * notice, this list of conditions and the following disclaimer. 11 * 2. Redistributions in binary form must reproduce the above copyright 12 * notice, this list of conditions and the following disclaimer in the 13 * documentation and/or other materials provided with the distribution. 14 * 4. Neither the name of the University nor the names of its contributors 15 * may be used to endorse or promote products derived from this software 16 * without specific prior written permission. 17 * 18 * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND 19 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE 20 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE 21 * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE 22 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL 23 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS 24 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) 25 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT 26 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY 27 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF 28 * SUCH DAMAGE. 29 * 30 * @(#)raw_ip.c 8.7 (Berkeley) 5/15/95 31 */ 32 33#include <sys/cdefs.h> 34__FBSDID("$FreeBSD: head/sys/netinet/raw_ip.c 188144 2009-02-05 14:06:09Z jamie $"); 35 36#include "opt_inet6.h" 37#include "opt_ipsec.h" 38#include "opt_mac.h" 39 40#include <sys/param.h> 41#include <sys/jail.h> 42#include <sys/kernel.h> 43#include <sys/lock.h> 44#include <sys/malloc.h> 45#include <sys/mbuf.h> 46#include <sys/priv.h> 47#include <sys/proc.h> 48#include <sys/protosw.h> 49#include <sys/rwlock.h> 50#include <sys/signalvar.h> 51#include <sys/socket.h> 52#include <sys/socketvar.h> 53#include <sys/sx.h> 54#include <sys/sysctl.h> 55#include <sys/systm.h> 56#include <sys/vimage.h> 57 58#include <vm/uma.h> 59 60#include <net/if.h> 61#include <net/route.h> 62#include <net/vnet.h> 63 64#include <netinet/in.h> 65#include <netinet/in_systm.h> 66#include <netinet/in_pcb.h> 67#include <netinet/in_var.h> 68#include <netinet/ip.h> 69#include <netinet/ip_var.h> 70#include <netinet/ip_mroute.h> 71 72#include <netinet/ip_fw.h> 73#include <netinet/ip_dummynet.h> 74#include <netinet/vinet.h> 75 76#ifdef IPSEC 77#include <netipsec/ipsec.h> 78#endif /*IPSEC*/ 79 80#include <security/mac/mac_framework.h> 81 82#ifdef VIMAGE_GLOBALS 83struct inpcbhead ripcb; 84struct inpcbinfo ripcbinfo; 85#endif 86 87/* control hooks for ipfw and dummynet */ 88ip_fw_ctl_t *ip_fw_ctl_ptr = NULL; 89ip_dn_ctl_t *ip_dn_ctl_ptr = NULL; 90 91/* 92 * Hooks for multicast routing. They all default to NULL, so leave them not 93 * initialized and rely on BSS being set to 0. 94 */ 95 96/* 97 * The socket used to communicate with the multicast routing daemon. 98 */ 99#ifdef VIMAGE_GLOBALS 100struct socket *ip_mrouter; 101#endif 102 103/* 104 * The various mrouter and rsvp functions. 105 */ 106int (*ip_mrouter_set)(struct socket *, struct sockopt *); 107int (*ip_mrouter_get)(struct socket *, struct sockopt *); 108int (*ip_mrouter_done)(void); 109int (*ip_mforward)(struct ip *, struct ifnet *, struct mbuf *, 110 struct ip_moptions *); 111int (*mrt_ioctl)(int, caddr_t, int); 112int (*legal_vif_num)(int); 113u_long (*ip_mcast_src)(int); 114 115void (*rsvp_input_p)(struct mbuf *m, int off); 116int (*ip_rsvp_vif)(struct socket *, struct sockopt *); 117void (*ip_rsvp_force_done)(struct socket *); 118 119/* 120 * Hash functions 121 */ 122 123#define INP_PCBHASH_RAW_SIZE 256 124#define INP_PCBHASH_RAW(proto, laddr, faddr, mask) \ 125 (((proto) + (laddr) + (faddr)) % (mask) + 1) 126 127static void 128rip_inshash(struct inpcb *inp) 129{ 130 struct inpcbinfo *pcbinfo = inp->inp_pcbinfo; 131 struct inpcbhead *pcbhash; 132 int hash; 133 134 INP_INFO_WLOCK_ASSERT(pcbinfo); 135 INP_WLOCK_ASSERT(inp); 136 137 if (inp->inp_ip_p != 0 && 138 inp->inp_laddr.s_addr != INADDR_ANY && 139 inp->inp_faddr.s_addr != INADDR_ANY) { 140 hash = INP_PCBHASH_RAW(inp->inp_ip_p, inp->inp_laddr.s_addr, 141 inp->inp_faddr.s_addr, pcbinfo->ipi_hashmask); 142 } else 143 hash = 0; 144 pcbhash = &pcbinfo->ipi_hashbase[hash]; 145 LIST_INSERT_HEAD(pcbhash, inp, inp_hash); 146} 147 148static void 149rip_delhash(struct inpcb *inp) 150{ 151 152 INP_INFO_WLOCK_ASSERT(inp->inp_pcbinfo); 153 INP_WLOCK_ASSERT(inp); 154 155 LIST_REMOVE(inp, inp_hash); 156} 157 158/* 159 * Raw interface to IP protocol. 160 */ 161 162/* 163 * Initialize raw connection block q. 164 */ 165static void 166rip_zone_change(void *tag) 167{ 168 INIT_VNET_INET(curvnet); 169 170 uma_zone_set_max(V_ripcbinfo.ipi_zone, maxsockets); 171} 172 173static int 174rip_inpcb_init(void *mem, int size, int flags) 175{ 176 struct inpcb *inp = mem; 177 178 INP_LOCK_INIT(inp, "inp", "rawinp"); 179 return (0); 180} 181 182void 183rip_init(void) 184{ 185 INIT_VNET_INET(curvnet); 186 187 INP_INFO_LOCK_INIT(&V_ripcbinfo, "rip"); 188 LIST_INIT(&V_ripcb); 189 V_ripcbinfo.ipi_listhead = &V_ripcb; 190 V_ripcbinfo.ipi_hashbase = 191 hashinit(INP_PCBHASH_RAW_SIZE, M_PCB, &V_ripcbinfo.ipi_hashmask); 192 V_ripcbinfo.ipi_porthashbase = 193 hashinit(1, M_PCB, &V_ripcbinfo.ipi_porthashmask); 194 V_ripcbinfo.ipi_zone = uma_zcreate("ripcb", sizeof(struct inpcb), 195 NULL, NULL, rip_inpcb_init, NULL, UMA_ALIGN_PTR, UMA_ZONE_NOFREE); 196 uma_zone_set_max(V_ripcbinfo.ipi_zone, maxsockets); 197 EVENTHANDLER_REGISTER(maxsockets_change, rip_zone_change, NULL, 198 EVENTHANDLER_PRI_ANY); 199} 200 201static int 202rip_append(struct inpcb *last, struct ip *ip, struct mbuf *n, 203 struct sockaddr_in *ripsrc) 204{ 205 int policyfail = 0; 206 207 INP_RLOCK_ASSERT(last); 208 209#ifdef IPSEC 210 /* check AH/ESP integrity. */ 211 if (ipsec4_in_reject(n, last)) { 212 policyfail = 1; 213 } 214#endif /* IPSEC */ 215#ifdef MAC 216 if (!policyfail && mac_inpcb_check_deliver(last, n) != 0) 217 policyfail = 1; 218#endif 219 /* Check the minimum TTL for socket. */ 220 if (last->inp_ip_minttl && last->inp_ip_minttl > ip->ip_ttl) 221 policyfail = 1; 222 if (!policyfail) { 223 struct mbuf *opts = NULL; 224 struct socket *so; 225 226 so = last->inp_socket; 227 if ((last->inp_flags & INP_CONTROLOPTS) || 228 (so->so_options & (SO_TIMESTAMP | SO_BINTIME))) 229 ip_savecontrol(last, &opts, ip, n); 230 SOCKBUF_LOCK(&so->so_rcv); 231 if (sbappendaddr_locked(&so->so_rcv, 232 (struct sockaddr *)ripsrc, n, opts) == 0) { 233 /* should notify about lost packet */ 234 m_freem(n); 235 if (opts) 236 m_freem(opts); 237 SOCKBUF_UNLOCK(&so->so_rcv); 238 } else 239 sorwakeup_locked(so); 240 } else 241 m_freem(n); 242 return (policyfail); 243} 244 245/* 246 * Setup generic address and protocol structures for raw_input routine, then 247 * pass them along with mbuf chain. 248 */ 249void 250rip_input(struct mbuf *m, int off) 251{ 252 INIT_VNET_INET(curvnet); 253 struct ip *ip = mtod(m, struct ip *); 254 int proto = ip->ip_p; 255 struct inpcb *inp, *last; 256 struct sockaddr_in ripsrc; 257 int hash; 258 259 bzero(&ripsrc, sizeof(ripsrc)); 260 ripsrc.sin_len = sizeof(ripsrc); 261 ripsrc.sin_family = AF_INET; 262 ripsrc.sin_addr = ip->ip_src; 263 last = NULL; 264 hash = INP_PCBHASH_RAW(proto, ip->ip_src.s_addr, 265 ip->ip_dst.s_addr, V_ripcbinfo.ipi_hashmask); 266 INP_INFO_RLOCK(&V_ripcbinfo); 267 LIST_FOREACH(inp, &V_ripcbinfo.ipi_hashbase[hash], inp_hash) { 268 if (inp->inp_ip_p != proto) 269 continue; 270#ifdef INET6 271 /* XXX inp locking */ 272 if ((inp->inp_vflag & INP_IPV4) == 0) 273 continue; 274#endif 275 if (inp->inp_laddr.s_addr != ip->ip_dst.s_addr) 276 continue; 277 if (inp->inp_faddr.s_addr != ip->ip_src.s_addr) 278 continue; 279 if (prison_check_ip4(inp->inp_cred, &ip->ip_dst) != 0) 280 continue; 281 if (last != NULL) { 282 struct mbuf *n; 283 284 n = m_copy(m, 0, (int)M_COPYALL); 285 if (n != NULL) 286 (void) rip_append(last, ip, n, &ripsrc); 287 /* XXX count dropped packet */ 288 INP_RUNLOCK(last); 289 } 290 INP_RLOCK(inp); 291 last = inp; 292 } 293 LIST_FOREACH(inp, &V_ripcbinfo.ipi_hashbase[0], inp_hash) { 294 if (inp->inp_ip_p && inp->inp_ip_p != proto) 295 continue; 296#ifdef INET6 297 /* XXX inp locking */ 298 if ((inp->inp_vflag & INP_IPV4) == 0) 299 continue; 300#endif 301 if (inp->inp_laddr.s_addr && 302 inp->inp_laddr.s_addr != ip->ip_dst.s_addr) 303 continue; 304 if (inp->inp_faddr.s_addr && 305 inp->inp_faddr.s_addr != ip->ip_src.s_addr) 306 continue; 307 if (prison_check_ip4(inp->inp_cred, &ip->ip_dst) != 0) 308 continue; 309 if (last != NULL) { 310 struct mbuf *n; 311 312 n = m_copy(m, 0, (int)M_COPYALL); 313 if (n != NULL) 314 (void) rip_append(last, ip, n, &ripsrc); 315 /* XXX count dropped packet */ 316 INP_RUNLOCK(last); 317 } 318 INP_RLOCK(inp); 319 last = inp; 320 } 321 INP_INFO_RUNLOCK(&V_ripcbinfo); 322 if (last != NULL) { 323 if (rip_append(last, ip, m, &ripsrc) != 0) 324 V_ipstat.ips_delivered--; 325 INP_RUNLOCK(last); 326 } else { 327 m_freem(m); 328 V_ipstat.ips_noproto++; 329 V_ipstat.ips_delivered--; 330 } 331} 332 333/* 334 * Generate IP header and pass packet to ip_output. Tack on options user may 335 * have setup with control call. 336 */ 337int 338rip_output(struct mbuf *m, struct socket *so, u_long dst) 339{ 340 INIT_VNET_INET(so->so_vnet); 341 struct ip *ip; 342 int error; 343 struct inpcb *inp = sotoinpcb(so); 344 int flags = ((so->so_options & SO_DONTROUTE) ? IP_ROUTETOIF : 0) | 345 IP_ALLOWBROADCAST; 346 347 /* 348 * If the user handed us a complete IP packet, use it. Otherwise, 349 * allocate an mbuf for a header and fill it in. 350 */ 351 if ((inp->inp_flags & INP_HDRINCL) == 0) { 352 if (m->m_pkthdr.len + sizeof(struct ip) > IP_MAXPACKET) { 353 m_freem(m); 354 return(EMSGSIZE); 355 } 356 M_PREPEND(m, sizeof(struct ip), M_DONTWAIT); 357 if (m == NULL) 358 return(ENOBUFS); 359 360 INP_RLOCK(inp); 361 ip = mtod(m, struct ip *); 362 ip->ip_tos = inp->inp_ip_tos; 363 if (inp->inp_flags & INP_DONTFRAG) 364 ip->ip_off = IP_DF; 365 else 366 ip->ip_off = 0; 367 ip->ip_p = inp->inp_ip_p; 368 ip->ip_len = m->m_pkthdr.len; 369 ip->ip_src = inp->inp_laddr; 370 error = prison_get_ip4(inp->inp_cred, &ip->ip_src); 371 if (error != 0) { 372 INP_RUNLOCK(inp); 373 m_freem(m); 374 return (error); 375 } 376 ip->ip_dst.s_addr = dst; 377 ip->ip_ttl = inp->inp_ip_ttl; 378 } else { 379 if (m->m_pkthdr.len > IP_MAXPACKET) { 380 m_freem(m); 381 return(EMSGSIZE); 382 } 383 INP_RLOCK(inp); 384 ip = mtod(m, struct ip *); 385 error = prison_check_ip4(inp->inp_cred, &ip->ip_src); 386 if (error != 0) { 387 INP_RUNLOCK(inp); 388 m_freem(m); 389 return (error); 390 } 391 392 /* 393 * Don't allow both user specified and setsockopt options, 394 * and don't allow packet length sizes that will crash. 395 */ 396 if (((ip->ip_hl != (sizeof (*ip) >> 2)) && inp->inp_options) 397 || (ip->ip_len > m->m_pkthdr.len) 398 || (ip->ip_len < (ip->ip_hl << 2))) { 399 INP_RUNLOCK(inp); 400 m_freem(m); 401 return (EINVAL); 402 } 403 if (ip->ip_id == 0) 404 ip->ip_id = ip_newid(); 405 406 /* 407 * XXX prevent ip_output from overwriting header fields. 408 */ 409 flags |= IP_RAWOUTPUT; 410 V_ipstat.ips_rawout++; 411 } 412 413 if (inp->inp_flags & INP_ONESBCAST) 414 flags |= IP_SENDONES; 415 416#ifdef MAC 417 mac_inpcb_create_mbuf(inp, m); 418#endif 419 420 error = ip_output(m, inp->inp_options, NULL, flags, 421 inp->inp_moptions, inp); 422 INP_RUNLOCK(inp); 423 return (error); 424} 425 426/* 427 * Raw IP socket option processing. 428 * 429 * IMPORTANT NOTE regarding access control: Traditionally, raw sockets could 430 * only be created by a privileged process, and as such, socket option 431 * operations to manage system properties on any raw socket were allowed to 432 * take place without explicit additional access control checks. However, 433 * raw sockets can now also be created in jail(), and therefore explicit 434 * checks are now required. Likewise, raw sockets can be used by a process 435 * after it gives up privilege, so some caution is required. For options 436 * passed down to the IP layer via ip_ctloutput(), checks are assumed to be 437 * performed in ip_ctloutput() and therefore no check occurs here. 438 * Unilaterally checking priv_check() here breaks normal IP socket option 439 * operations on raw sockets. 440 * 441 * When adding new socket options here, make sure to add access control 442 * checks here as necessary. 443 */ 444int 445rip_ctloutput(struct socket *so, struct sockopt *sopt) 446{ 447 struct inpcb *inp = sotoinpcb(so); 448 int error, optval; 449 450 if (sopt->sopt_level != IPPROTO_IP) { 451 if ((sopt->sopt_level == SOL_SOCKET) && 452 (sopt->sopt_name == SO_SETFIB)) { 453 inp->inp_inc.inc_fibnum = so->so_fibnum; 454 return (0); 455 } 456 return (EINVAL); 457 } 458 459 error = 0; 460 switch (sopt->sopt_dir) { 461 case SOPT_GET: 462 switch (sopt->sopt_name) { 463 case IP_HDRINCL: 464 optval = inp->inp_flags & INP_HDRINCL; 465 error = sooptcopyout(sopt, &optval, sizeof optval); 466 break; 467 468 case IP_FW_ADD: /* ADD actually returns the body... */ 469 case IP_FW_GET: 470 case IP_FW_TABLE_GETSIZE: 471 case IP_FW_TABLE_LIST: 472 case IP_FW_NAT_GET_CONFIG: 473 case IP_FW_NAT_GET_LOG: 474 if (ip_fw_ctl_ptr != NULL) 475 error = ip_fw_ctl_ptr(sopt); 476 else 477 error = ENOPROTOOPT; 478 break; 479 480 case IP_DUMMYNET_GET: 481 if (ip_dn_ctl_ptr != NULL) 482 error = ip_dn_ctl_ptr(sopt); 483 else 484 error = ENOPROTOOPT; 485 break ; 486 487 case MRT_INIT: 488 case MRT_DONE: 489 case MRT_ADD_VIF: 490 case MRT_DEL_VIF: 491 case MRT_ADD_MFC: 492 case MRT_DEL_MFC: 493 case MRT_VERSION: 494 case MRT_ASSERT: 495 case MRT_API_SUPPORT: 496 case MRT_API_CONFIG: 497 case MRT_ADD_BW_UPCALL: 498 case MRT_DEL_BW_UPCALL: 499 error = priv_check(curthread, PRIV_NETINET_MROUTE); 500 if (error != 0) 501 return (error); 502 error = ip_mrouter_get ? ip_mrouter_get(so, sopt) : 503 EOPNOTSUPP; 504 break; 505 506 default: 507 error = ip_ctloutput(so, sopt); 508 break; 509 } 510 break; 511 512 case SOPT_SET: 513 switch (sopt->sopt_name) { 514 case IP_HDRINCL: 515 error = sooptcopyin(sopt, &optval, sizeof optval, 516 sizeof optval); 517 if (error) 518 break; 519 if (optval) 520 inp->inp_flags |= INP_HDRINCL; 521 else 522 inp->inp_flags &= ~INP_HDRINCL; 523 break; 524 525 case IP_FW_ADD: 526 case IP_FW_DEL: 527 case IP_FW_FLUSH: 528 case IP_FW_ZERO: 529 case IP_FW_RESETLOG: 530 case IP_FW_TABLE_ADD: 531 case IP_FW_TABLE_DEL: 532 case IP_FW_TABLE_FLUSH: 533 case IP_FW_NAT_CFG: 534 case IP_FW_NAT_DEL: 535 if (ip_fw_ctl_ptr != NULL) 536 error = ip_fw_ctl_ptr(sopt); 537 else 538 error = ENOPROTOOPT; 539 break; 540 541 case IP_DUMMYNET_CONFIGURE: 542 case IP_DUMMYNET_DEL: 543 case IP_DUMMYNET_FLUSH: 544 if (ip_dn_ctl_ptr != NULL) 545 error = ip_dn_ctl_ptr(sopt); 546 else 547 error = ENOPROTOOPT ; 548 break ; 549 550 case IP_RSVP_ON: 551 error = priv_check(curthread, PRIV_NETINET_MROUTE); 552 if (error != 0) 553 return (error); 554 error = ip_rsvp_init(so); 555 break; 556 557 case IP_RSVP_OFF: 558 error = priv_check(curthread, PRIV_NETINET_MROUTE); 559 if (error != 0) 560 return (error); 561 error = ip_rsvp_done(); 562 break; 563 564 case IP_RSVP_VIF_ON: 565 case IP_RSVP_VIF_OFF: 566 error = priv_check(curthread, PRIV_NETINET_MROUTE); 567 if (error != 0) 568 return (error); 569 error = ip_rsvp_vif ? 570 ip_rsvp_vif(so, sopt) : EINVAL; 571 break; 572 573 case MRT_INIT: 574 case MRT_DONE: 575 case MRT_ADD_VIF: 576 case MRT_DEL_VIF: 577 case MRT_ADD_MFC: 578 case MRT_DEL_MFC: 579 case MRT_VERSION: 580 case MRT_ASSERT: 581 case MRT_API_SUPPORT: 582 case MRT_API_CONFIG: 583 case MRT_ADD_BW_UPCALL: 584 case MRT_DEL_BW_UPCALL: 585 error = priv_check(curthread, PRIV_NETINET_MROUTE); 586 if (error != 0) 587 return (error); 588 error = ip_mrouter_set ? ip_mrouter_set(so, sopt) : 589 EOPNOTSUPP; 590 break; 591 592 default: 593 error = ip_ctloutput(so, sopt); 594 break; 595 } 596 break; 597 } 598 599 return (error); 600} 601 602/* 603 * This function exists solely to receive the PRC_IFDOWN messages which are 604 * sent by if_down(). It looks for an ifaddr whose ifa_addr is sa, and calls 605 * in_ifadown() to remove all routes corresponding to that address. It also 606 * receives the PRC_IFUP messages from if_up() and reinstalls the interface 607 * routes. 608 */ 609void 610rip_ctlinput(int cmd, struct sockaddr *sa, void *vip) 611{ 612 INIT_VNET_INET(curvnet); 613 struct in_ifaddr *ia; 614 struct ifnet *ifp; 615 int err; 616 int flags; 617 618 switch (cmd) { 619 case PRC_IFDOWN: 620 TAILQ_FOREACH(ia, &V_in_ifaddrhead, ia_link) { 621 if (ia->ia_ifa.ifa_addr == sa 622 && (ia->ia_flags & IFA_ROUTE)) { 623 /* 624 * in_ifscrub kills the interface route. 625 */ 626 in_ifscrub(ia->ia_ifp, ia); 627 /* 628 * in_ifadown gets rid of all the rest of the 629 * routes. This is not quite the right thing 630 * to do, but at least if we are running a 631 * routing process they will come back. 632 */ 633 in_ifadown(&ia->ia_ifa, 0); 634 break; 635 } 636 } 637 break; 638 639 case PRC_IFUP: 640 TAILQ_FOREACH(ia, &V_in_ifaddrhead, ia_link) { 641 if (ia->ia_ifa.ifa_addr == sa) 642 break; 643 } 644 if (ia == 0 || (ia->ia_flags & IFA_ROUTE)) 645 return; 646 flags = RTF_UP; 647 ifp = ia->ia_ifa.ifa_ifp; 648 649 if ((ifp->if_flags & IFF_LOOPBACK) 650 || (ifp->if_flags & IFF_POINTOPOINT)) 651 flags |= RTF_HOST; 652 653 err = rtinit(&ia->ia_ifa, RTM_ADD, flags); 654 if (err == 0) 655 ia->ia_flags |= IFA_ROUTE; 656 break; 657 } 658} 659 660u_long rip_sendspace = 9216; 661u_long rip_recvspace = 9216; 662 663SYSCTL_ULONG(_net_inet_raw, OID_AUTO, maxdgram, CTLFLAG_RW, 664 &rip_sendspace, 0, "Maximum outgoing raw IP datagram size"); 665SYSCTL_ULONG(_net_inet_raw, OID_AUTO, recvspace, CTLFLAG_RW, 666 &rip_recvspace, 0, "Maximum space for incoming raw IP datagrams"); 667 668static int 669rip_attach(struct socket *so, int proto, struct thread *td) 670{ 671 INIT_VNET_INET(so->so_vnet); 672 struct inpcb *inp; 673 int error; 674 675 inp = sotoinpcb(so); 676 KASSERT(inp == NULL, ("rip_attach: inp != NULL")); 677 678 error = priv_check(td, PRIV_NETINET_RAW); 679 if (error) 680 return (error); 681 if (proto >= IPPROTO_MAX || proto < 0) 682 return EPROTONOSUPPORT; 683 error = soreserve(so, rip_sendspace, rip_recvspace); 684 if (error) 685 return (error); 686 INP_INFO_WLOCK(&V_ripcbinfo); 687 error = in_pcballoc(so, &V_ripcbinfo); 688 if (error) { 689 INP_INFO_WUNLOCK(&V_ripcbinfo); 690 return (error); 691 } 692 inp = (struct inpcb *)so->so_pcb; 693 inp->inp_vflag |= INP_IPV4; 694 inp->inp_ip_p = proto; 695 inp->inp_ip_ttl = V_ip_defttl; 696 rip_inshash(inp); 697 INP_INFO_WUNLOCK(&V_ripcbinfo); 698 INP_WUNLOCK(inp); 699 return (0); 700} 701 702static void 703rip_detach(struct socket *so) 704{ 705 INIT_VNET_INET(so->so_vnet); 706 struct inpcb *inp; 707 708 inp = sotoinpcb(so); 709 KASSERT(inp != NULL, ("rip_detach: inp == NULL")); 710 KASSERT(inp->inp_faddr.s_addr == INADDR_ANY, 711 ("rip_detach: not closed")); 712 713 INP_INFO_WLOCK(&V_ripcbinfo); 714 INP_WLOCK(inp); 715 rip_delhash(inp); 716 if (so == V_ip_mrouter && ip_mrouter_done) 717 ip_mrouter_done(); 718 if (ip_rsvp_force_done) 719 ip_rsvp_force_done(so); 720 if (so == V_ip_rsvpd) 721 ip_rsvp_done(); 722 in_pcbdetach(inp); 723 in_pcbfree(inp); 724 INP_INFO_WUNLOCK(&V_ripcbinfo); 725} 726 727static void 728rip_dodisconnect(struct socket *so, struct inpcb *inp) 729{ 730 731 INP_INFO_WLOCK_ASSERT(inp->inp_pcbinfo); 732 INP_WLOCK_ASSERT(inp); 733 734 rip_delhash(inp); 735 inp->inp_faddr.s_addr = INADDR_ANY; 736 rip_inshash(inp); 737 SOCK_LOCK(so); 738 so->so_state &= ~SS_ISCONNECTED; 739 SOCK_UNLOCK(so); 740} 741 742static void 743rip_abort(struct socket *so) 744{ 745 INIT_VNET_INET(so->so_vnet); 746 struct inpcb *inp; 747 748 inp = sotoinpcb(so); 749 KASSERT(inp != NULL, ("rip_abort: inp == NULL")); 750 751 INP_INFO_WLOCK(&V_ripcbinfo); 752 INP_WLOCK(inp); 753 rip_dodisconnect(so, inp); 754 INP_WUNLOCK(inp); 755 INP_INFO_WUNLOCK(&V_ripcbinfo); 756} 757 758static void 759rip_close(struct socket *so) 760{ 761 INIT_VNET_INET(so->so_vnet); 762 struct inpcb *inp; 763 764 inp = sotoinpcb(so); 765 KASSERT(inp != NULL, ("rip_close: inp == NULL")); 766 767 INP_INFO_WLOCK(&V_ripcbinfo); 768 INP_WLOCK(inp); 769 rip_dodisconnect(so, inp); 770 INP_WUNLOCK(inp); 771 INP_INFO_WUNLOCK(&V_ripcbinfo); 772} 773 774static int 775rip_disconnect(struct socket *so) 776{ 777 INIT_VNET_INET(so->so_vnet); 778 struct inpcb *inp; 779 780 if ((so->so_state & SS_ISCONNECTED) == 0) 781 return (ENOTCONN); 782 783 inp = sotoinpcb(so); 784 KASSERT(inp != NULL, ("rip_disconnect: inp == NULL")); 785 786 INP_INFO_WLOCK(&V_ripcbinfo); 787 INP_WLOCK(inp); 788 rip_dodisconnect(so, inp); 789 INP_WUNLOCK(inp); 790 INP_INFO_WUNLOCK(&V_ripcbinfo); 791 return (0); 792} 793 794static int 795rip_bind(struct socket *so, struct sockaddr *nam, struct thread *td) 796{ 797 INIT_VNET_NET(so->so_vnet); 798 INIT_VNET_INET(so->so_vnet); 799 struct sockaddr_in *addr = (struct sockaddr_in *)nam; 800 struct inpcb *inp; 801 int error; 802 803 if (nam->sa_len != sizeof(*addr)) 804 return (EINVAL); 805 806 error = prison_check_ip4(td->td_ucred, &addr->sin_addr); 807 if (error != 0) 808 return (error); 809 810 if (TAILQ_EMPTY(&V_ifnet) || 811 (addr->sin_family != AF_INET && addr->sin_family != AF_IMPLINK) || 812 (addr->sin_addr.s_addr && 813 ifa_ifwithaddr((struct sockaddr *)addr) == 0)) 814 return (EADDRNOTAVAIL); 815 816 inp = sotoinpcb(so); 817 KASSERT(inp != NULL, ("rip_bind: inp == NULL")); 818 819 INP_INFO_WLOCK(&V_ripcbinfo); 820 INP_WLOCK(inp); 821 rip_delhash(inp); 822 inp->inp_laddr = addr->sin_addr; 823 rip_inshash(inp); 824 INP_WUNLOCK(inp); 825 INP_INFO_WUNLOCK(&V_ripcbinfo); 826 return (0); 827} 828 829static int 830rip_connect(struct socket *so, struct sockaddr *nam, struct thread *td) 831{ 832 INIT_VNET_NET(so->so_vnet); 833 INIT_VNET_INET(so->so_vnet); 834 struct sockaddr_in *addr = (struct sockaddr_in *)nam; 835 struct inpcb *inp; 836 837 if (nam->sa_len != sizeof(*addr)) 838 return (EINVAL); 839 if (TAILQ_EMPTY(&V_ifnet)) 840 return (EADDRNOTAVAIL); 841 if (addr->sin_family != AF_INET && addr->sin_family != AF_IMPLINK) 842 return (EAFNOSUPPORT); 843 844 inp = sotoinpcb(so); 845 KASSERT(inp != NULL, ("rip_connect: inp == NULL")); 846 847 INP_INFO_WLOCK(&V_ripcbinfo); 848 INP_WLOCK(inp); 849 rip_delhash(inp); 850 inp->inp_faddr = addr->sin_addr; 851 rip_inshash(inp); 852 soisconnected(so); 853 INP_WUNLOCK(inp); 854 INP_INFO_WUNLOCK(&V_ripcbinfo); 855 return (0); 856} 857 858static int 859rip_shutdown(struct socket *so) 860{ 861 struct inpcb *inp; 862 863 inp = sotoinpcb(so); 864 KASSERT(inp != NULL, ("rip_shutdown: inp == NULL")); 865 866 INP_WLOCK(inp); 867 socantsendmore(so); 868 INP_WUNLOCK(inp); 869 return (0); 870} 871 872static int 873rip_send(struct socket *so, int flags, struct mbuf *m, struct sockaddr *nam, 874 struct mbuf *control, struct thread *td) 875{ 876 struct inpcb *inp; 877 u_long dst; 878 879 inp = sotoinpcb(so); 880 KASSERT(inp != NULL, ("rip_send: inp == NULL")); 881 882 /* 883 * Note: 'dst' reads below are unlocked. 884 */ 885 if (so->so_state & SS_ISCONNECTED) { 886 if (nam) { 887 m_freem(m); 888 return (EISCONN); 889 } 890 dst = inp->inp_faddr.s_addr; /* Unlocked read. */ 891 } else { 892 if (nam == NULL) { 893 m_freem(m); 894 return (ENOTCONN); 895 } 896 dst = ((struct sockaddr_in *)nam)->sin_addr.s_addr; 897 } 898 return (rip_output(m, so, dst)); 899} 900 901static int 902rip_pcblist(SYSCTL_HANDLER_ARGS) 903{ 904 INIT_VNET_INET(curvnet); 905 int error, i, n; 906 struct inpcb *inp, **inp_list; 907 inp_gen_t gencnt; 908 struct xinpgen xig; 909 910 /* 911 * The process of preparing the TCB list is too time-consuming and 912 * resource-intensive to repeat twice on every request. 913 */ 914 if (req->oldptr == 0) { 915 n = V_ripcbinfo.ipi_count; 916 req->oldidx = 2 * (sizeof xig) 917 + (n + n/8) * sizeof(struct xinpcb); 918 return (0); 919 } 920 921 if (req->newptr != 0) 922 return (EPERM); 923 924 /* 925 * OK, now we're committed to doing something. 926 */ 927 INP_INFO_RLOCK(&V_ripcbinfo); 928 gencnt = V_ripcbinfo.ipi_gencnt; 929 n = V_ripcbinfo.ipi_count; 930 INP_INFO_RUNLOCK(&V_ripcbinfo); 931 932 xig.xig_len = sizeof xig; 933 xig.xig_count = n; 934 xig.xig_gen = gencnt; 935 xig.xig_sogen = so_gencnt; 936 error = SYSCTL_OUT(req, &xig, sizeof xig); 937 if (error) 938 return (error); 939 940 inp_list = malloc(n * sizeof *inp_list, M_TEMP, M_WAITOK); 941 if (inp_list == 0) 942 return (ENOMEM); 943 944 INP_INFO_RLOCK(&V_ripcbinfo); 945 for (inp = LIST_FIRST(V_ripcbinfo.ipi_listhead), i = 0; inp && i < n; 946 inp = LIST_NEXT(inp, inp_list)) { 947 INP_RLOCK(inp); 948 if (inp->inp_gencnt <= gencnt && 949 cr_canseeinpcb(req->td->td_ucred, inp) == 0) { 950 /* XXX held references? */ 951 inp_list[i++] = inp; 952 } 953 INP_RUNLOCK(inp); 954 } 955 INP_INFO_RUNLOCK(&V_ripcbinfo); 956 n = i; 957 958 error = 0; 959 for (i = 0; i < n; i++) { 960 inp = inp_list[i]; 961 INP_RLOCK(inp); 962 if (inp->inp_gencnt <= gencnt) { 963 struct xinpcb xi; 964 965 bzero(&xi, sizeof(xi)); 966 xi.xi_len = sizeof xi; 967 /* XXX should avoid extra copy */ 968 bcopy(inp, &xi.xi_inp, sizeof *inp); 969 if (inp->inp_socket) 970 sotoxsocket(inp->inp_socket, &xi.xi_socket); 971 INP_RUNLOCK(inp); 972 error = SYSCTL_OUT(req, &xi, sizeof xi); 973 } else 974 INP_RUNLOCK(inp); 975 } 976 if (!error) { 977 /* 978 * Give the user an updated idea of our state. If the 979 * generation differs from what we told her before, she knows 980 * that something happened while we were processing this 981 * request, and it might be necessary to retry. 982 */ 983 INP_INFO_RLOCK(&V_ripcbinfo); 984 xig.xig_gen = V_ripcbinfo.ipi_gencnt; 985 xig.xig_sogen = so_gencnt; 986 xig.xig_count = V_ripcbinfo.ipi_count; 987 INP_INFO_RUNLOCK(&V_ripcbinfo); 988 error = SYSCTL_OUT(req, &xig, sizeof xig); 989 } 990 free(inp_list, M_TEMP); 991 return (error); 992} 993 994SYSCTL_PROC(_net_inet_raw, OID_AUTO/*XXX*/, pcblist, CTLFLAG_RD, 0, 0, 995 rip_pcblist, "S,xinpcb", "List of active raw IP sockets"); 996 997struct pr_usrreqs rip_usrreqs = { 998 .pru_abort = rip_abort, 999 .pru_attach = rip_attach, 1000 .pru_bind = rip_bind, 1001 .pru_connect = rip_connect, 1002 .pru_control = in_control, 1003 .pru_detach = rip_detach, 1004 .pru_disconnect = rip_disconnect, 1005 .pru_peeraddr = in_getpeeraddr, 1006 .pru_send = rip_send, 1007 .pru_shutdown = rip_shutdown, 1008 .pru_sockaddr = in_getsockaddr, 1009 .pru_sosetlabel = in_pcbsosetlabel, 1010 .pru_close = rip_close, 1011}; 1012