1/*- 2 * Copyright (c) 1988, 1991, 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 * @(#)rtsock.c 8.7 (Berkeley) 10/12/95 30 * $FreeBSD$ 31 */ 32#include "opt_compat.h" 33#include "opt_mpath.h" 34#include "opt_inet.h" 35#include "opt_inet6.h" 36 37#include <sys/param.h> 38#include <sys/jail.h> 39#include <sys/kernel.h> 40#include <sys/domain.h> 41#include <sys/lock.h> 42#include <sys/malloc.h> 43#include <sys/mbuf.h> 44#include <sys/priv.h> 45#include <sys/proc.h> 46#include <sys/protosw.h> 47#include <sys/rwlock.h> 48#include <sys/signalvar.h> 49#include <sys/socket.h> 50#include <sys/socketvar.h> 51#include <sys/sysctl.h> 52#include <sys/systm.h> 53 54#include <net/if.h> 55#include <net/if_dl.h> 56#include <net/if_llatbl.h> 57#include <net/if_types.h> 58#include <net/netisr.h> 59#include <net/raw_cb.h> 60#include <net/route.h> 61#include <net/vnet.h> 62 63#include <netinet/in.h> 64#include <netinet/if_ether.h> 65#ifdef INET6 66#include <netinet6/scope6_var.h> 67#endif 68 69#ifdef COMPAT_FREEBSD32 70#include <sys/mount.h> 71#include <compat/freebsd32/freebsd32.h> 72 73struct if_data32 { 74 uint8_t ifi_type; 75 uint8_t ifi_physical; 76 uint8_t ifi_addrlen; 77 uint8_t ifi_hdrlen; 78 uint8_t ifi_link_state; 79 uint8_t ifi_spare_char1; 80 uint8_t ifi_spare_char2; 81 uint8_t ifi_datalen; 82 uint32_t ifi_mtu; 83 uint32_t ifi_metric; 84 uint32_t ifi_baudrate; 85 uint32_t ifi_ipackets; 86 uint32_t ifi_ierrors; 87 uint32_t ifi_opackets; 88 uint32_t ifi_oerrors; 89 uint32_t ifi_collisions; 90 uint32_t ifi_ibytes; 91 uint32_t ifi_obytes; 92 uint32_t ifi_imcasts; 93 uint32_t ifi_omcasts; 94 uint32_t ifi_iqdrops; 95 uint32_t ifi_noproto; 96 uint32_t ifi_hwassist; 97 int32_t ifi_epoch; 98 struct timeval32 ifi_lastchange; 99}; 100 101struct if_msghdr32 { 102 uint16_t ifm_msglen; 103 uint8_t ifm_version; 104 uint8_t ifm_type; 105 int32_t ifm_addrs; 106 int32_t ifm_flags; 107 uint16_t ifm_index; 108 struct if_data32 ifm_data; 109}; 110 111struct if_msghdrl32 { 112 uint16_t ifm_msglen; 113 uint8_t ifm_version; 114 uint8_t ifm_type; 115 int32_t ifm_addrs; 116 int32_t ifm_flags; 117 uint16_t ifm_index; 118 uint16_t _ifm_spare1; 119 uint16_t ifm_len; 120 uint16_t ifm_data_off; 121 struct if_data32 ifm_data; 122}; 123 124struct ifa_msghdrl32 { 125 uint16_t ifam_msglen; 126 uint8_t ifam_version; 127 uint8_t ifam_type; 128 int32_t ifam_addrs; 129 int32_t ifam_flags; 130 uint16_t ifam_index; 131 uint16_t _ifam_spare1; 132 uint16_t ifam_len; 133 uint16_t ifam_data_off; 134 int32_t ifam_metric; 135 struct if_data32 ifam_data; 136}; 137#endif /* COMPAT_FREEBSD32 */ 138 139MALLOC_DEFINE(M_RTABLE, "routetbl", "routing tables"); 140 141/* NB: these are not modified */ 142static struct sockaddr route_src = { 2, PF_ROUTE, }; 143static struct sockaddr sa_zero = { sizeof(sa_zero), AF_INET, }; 144 145/* 146 * Used by rtsock/raw_input callback code to decide whether to filter the update 147 * notification to a socket bound to a particular FIB. 148 */ 149#define RTS_FILTER_FIB M_PROTO8 150 151static struct { 152 int ip_count; /* attached w/ AF_INET */ 153 int ip6_count; /* attached w/ AF_INET6 */ 154 int ipx_count; /* attached w/ AF_IPX */ 155 int any_count; /* total attached */ 156} route_cb; 157 158struct mtx rtsock_mtx; 159MTX_SYSINIT(rtsock, &rtsock_mtx, "rtsock route_cb lock", MTX_DEF); 160 161#define RTSOCK_LOCK() mtx_lock(&rtsock_mtx) 162#define RTSOCK_UNLOCK() mtx_unlock(&rtsock_mtx) 163#define RTSOCK_LOCK_ASSERT() mtx_assert(&rtsock_mtx, MA_OWNED) 164 165static SYSCTL_NODE(_net, OID_AUTO, route, CTLFLAG_RD, 0, ""); 166 167struct walkarg { 168 int w_tmemsize; 169 int w_op, w_arg; 170 caddr_t w_tmem; 171 struct sysctl_req *w_req; 172}; 173 174static void rts_input(struct mbuf *m); 175static struct mbuf *rt_msg1(int type, struct rt_addrinfo *rtinfo); 176static int rt_msg2(int type, struct rt_addrinfo *rtinfo, 177 caddr_t cp, struct walkarg *w); 178static int rt_xaddrs(caddr_t cp, caddr_t cplim, 179 struct rt_addrinfo *rtinfo); 180static int sysctl_dumpentry(struct radix_node *rn, void *vw); 181static int sysctl_iflist(int af, struct walkarg *w); 182static int sysctl_ifmalist(int af, struct walkarg *w); 183static int route_output(struct mbuf *m, struct socket *so); 184static void rt_setmetrics(u_long which, const struct rt_metrics *in, 185 struct rt_metrics_lite *out); 186static void rt_getmetrics(const struct rt_metrics_lite *in, 187 struct rt_metrics *out); 188static void rt_dispatch(struct mbuf *, sa_family_t); 189 190static struct netisr_handler rtsock_nh = { 191 .nh_name = "rtsock", 192 .nh_handler = rts_input, 193 .nh_proto = NETISR_ROUTE, 194 .nh_policy = NETISR_POLICY_SOURCE, 195}; 196 197static int 198sysctl_route_netisr_maxqlen(SYSCTL_HANDLER_ARGS) 199{ 200 int error, qlimit; 201 202 netisr_getqlimit(&rtsock_nh, &qlimit); 203 error = sysctl_handle_int(oidp, &qlimit, 0, req); 204 if (error || !req->newptr) 205 return (error); 206 if (qlimit < 1) 207 return (EINVAL); 208 return (netisr_setqlimit(&rtsock_nh, qlimit)); 209} 210SYSCTL_PROC(_net_route, OID_AUTO, netisr_maxqlen, CTLTYPE_INT|CTLFLAG_RW, 211 0, 0, sysctl_route_netisr_maxqlen, "I", 212 "maximum routing socket dispatch queue length"); 213 214static void 215rts_init(void) 216{ 217 int tmp; 218 219 if (TUNABLE_INT_FETCH("net.route.netisr_maxqlen", &tmp)) 220 rtsock_nh.nh_qlimit = tmp; 221 netisr_register(&rtsock_nh); 222} 223SYSINIT(rtsock, SI_SUB_PROTO_DOMAIN, SI_ORDER_THIRD, rts_init, 0); 224 225static int 226raw_input_rts_cb(struct mbuf *m, struct sockproto *proto, struct sockaddr *src, 227 struct rawcb *rp) 228{ 229 int fibnum; 230 231 KASSERT(m != NULL, ("%s: m is NULL", __func__)); 232 KASSERT(proto != NULL, ("%s: proto is NULL", __func__)); 233 KASSERT(rp != NULL, ("%s: rp is NULL", __func__)); 234 235 /* No filtering requested. */ 236 if ((m->m_flags & RTS_FILTER_FIB) == 0) 237 return (0); 238 239 /* Check if it is a rts and the fib matches the one of the socket. */ 240 fibnum = M_GETFIB(m); 241 if (proto->sp_family != PF_ROUTE || 242 rp->rcb_socket == NULL || 243 rp->rcb_socket->so_fibnum == fibnum) 244 return (0); 245 246 /* Filtering requested and no match, the socket shall be skipped. */ 247 return (1); 248} 249 250static void 251rts_input(struct mbuf *m) 252{ 253 struct sockproto route_proto; 254 unsigned short *family; 255 struct m_tag *tag; 256 257 route_proto.sp_family = PF_ROUTE; 258 tag = m_tag_find(m, PACKET_TAG_RTSOCKFAM, NULL); 259 if (tag != NULL) { 260 family = (unsigned short *)(tag + 1); 261 route_proto.sp_protocol = *family; 262 m_tag_delete(m, tag); 263 } else 264 route_proto.sp_protocol = 0; 265 266 raw_input_ext(m, &route_proto, &route_src, raw_input_rts_cb); 267} 268 269/* 270 * It really doesn't make any sense at all for this code to share much 271 * with raw_usrreq.c, since its functionality is so restricted. XXX 272 */ 273static void 274rts_abort(struct socket *so) 275{ 276 277 raw_usrreqs.pru_abort(so); 278} 279 280static void 281rts_close(struct socket *so) 282{ 283 284 raw_usrreqs.pru_close(so); 285} 286 287/* pru_accept is EOPNOTSUPP */ 288 289static int 290rts_attach(struct socket *so, int proto, struct thread *td) 291{ 292 struct rawcb *rp; 293 int s, error; 294 295 KASSERT(so->so_pcb == NULL, ("rts_attach: so_pcb != NULL")); 296 297 /* XXX */ 298 rp = malloc(sizeof *rp, M_PCB, M_WAITOK | M_ZERO); 299 if (rp == NULL) 300 return ENOBUFS; 301 302 /* 303 * The splnet() is necessary to block protocols from sending 304 * error notifications (like RTM_REDIRECT or RTM_LOSING) while 305 * this PCB is extant but incompletely initialized. 306 * Probably we should try to do more of this work beforehand and 307 * eliminate the spl. 308 */ 309 s = splnet(); 310 so->so_pcb = (caddr_t)rp; 311 so->so_fibnum = td->td_proc->p_fibnum; 312 error = raw_attach(so, proto); 313 rp = sotorawcb(so); 314 if (error) { 315 splx(s); 316 so->so_pcb = NULL; 317 free(rp, M_PCB); 318 return error; 319 } 320 RTSOCK_LOCK(); 321 switch(rp->rcb_proto.sp_protocol) { 322 case AF_INET: 323 route_cb.ip_count++; 324 break; 325 case AF_INET6: 326 route_cb.ip6_count++; 327 break; 328 case AF_IPX: 329 route_cb.ipx_count++; 330 break; 331 } 332 route_cb.any_count++; 333 RTSOCK_UNLOCK(); 334 soisconnected(so); 335 so->so_options |= SO_USELOOPBACK; 336 splx(s); 337 return 0; 338} 339 340static int 341rts_bind(struct socket *so, struct sockaddr *nam, struct thread *td) 342{ 343 344 return (raw_usrreqs.pru_bind(so, nam, td)); /* xxx just EINVAL */ 345} 346 347static int 348rts_connect(struct socket *so, struct sockaddr *nam, struct thread *td) 349{ 350 351 return (raw_usrreqs.pru_connect(so, nam, td)); /* XXX just EINVAL */ 352} 353 354/* pru_connect2 is EOPNOTSUPP */ 355/* pru_control is EOPNOTSUPP */ 356 357static void 358rts_detach(struct socket *so) 359{ 360 struct rawcb *rp = sotorawcb(so); 361 362 KASSERT(rp != NULL, ("rts_detach: rp == NULL")); 363 364 RTSOCK_LOCK(); 365 switch(rp->rcb_proto.sp_protocol) { 366 case AF_INET: 367 route_cb.ip_count--; 368 break; 369 case AF_INET6: 370 route_cb.ip6_count--; 371 break; 372 case AF_IPX: 373 route_cb.ipx_count--; 374 break; 375 } 376 route_cb.any_count--; 377 RTSOCK_UNLOCK(); 378 raw_usrreqs.pru_detach(so); 379} 380 381static int 382rts_disconnect(struct socket *so) 383{ 384 385 return (raw_usrreqs.pru_disconnect(so)); 386} 387 388/* pru_listen is EOPNOTSUPP */ 389 390static int 391rts_peeraddr(struct socket *so, struct sockaddr **nam) 392{ 393 394 return (raw_usrreqs.pru_peeraddr(so, nam)); 395} 396 397/* pru_rcvd is EOPNOTSUPP */ 398/* pru_rcvoob is EOPNOTSUPP */ 399 400static int 401rts_send(struct socket *so, int flags, struct mbuf *m, struct sockaddr *nam, 402 struct mbuf *control, struct thread *td) 403{ 404 405 return (raw_usrreqs.pru_send(so, flags, m, nam, control, td)); 406} 407 408/* pru_sense is null */ 409 410static int 411rts_shutdown(struct socket *so) 412{ 413 414 return (raw_usrreqs.pru_shutdown(so)); 415} 416 417static int 418rts_sockaddr(struct socket *so, struct sockaddr **nam) 419{ 420 421 return (raw_usrreqs.pru_sockaddr(so, nam)); 422} 423 424static struct pr_usrreqs route_usrreqs = { 425 .pru_abort = rts_abort, 426 .pru_attach = rts_attach, 427 .pru_bind = rts_bind, 428 .pru_connect = rts_connect, 429 .pru_detach = rts_detach, 430 .pru_disconnect = rts_disconnect, 431 .pru_peeraddr = rts_peeraddr, 432 .pru_send = rts_send, 433 .pru_shutdown = rts_shutdown, 434 .pru_sockaddr = rts_sockaddr, 435 .pru_close = rts_close, 436}; 437 438#ifndef _SOCKADDR_UNION_DEFINED 439#define _SOCKADDR_UNION_DEFINED 440/* 441 * The union of all possible address formats we handle. 442 */ 443union sockaddr_union { 444 struct sockaddr sa; 445 struct sockaddr_in sin; 446 struct sockaddr_in6 sin6; 447}; 448#endif /* _SOCKADDR_UNION_DEFINED */ 449 450static int 451rtm_get_jailed(struct rt_addrinfo *info, struct ifnet *ifp, 452 struct rtentry *rt, union sockaddr_union *saun, struct ucred *cred) 453{ 454 455 /* First, see if the returned address is part of the jail. */ 456 if (prison_if(cred, rt->rt_ifa->ifa_addr) == 0) { 457 info->rti_info[RTAX_IFA] = rt->rt_ifa->ifa_addr; 458 return (0); 459 } 460 461 switch (info->rti_info[RTAX_DST]->sa_family) { 462#ifdef INET 463 case AF_INET: 464 { 465 struct in_addr ia; 466 struct ifaddr *ifa; 467 int found; 468 469 found = 0; 470 /* 471 * Try to find an address on the given outgoing interface 472 * that belongs to the jail. 473 */ 474 IF_ADDR_RLOCK(ifp); 475 TAILQ_FOREACH(ifa, &ifp->if_addrhead, ifa_link) { 476 struct sockaddr *sa; 477 sa = ifa->ifa_addr; 478 if (sa->sa_family != AF_INET) 479 continue; 480 ia = ((struct sockaddr_in *)sa)->sin_addr; 481 if (prison_check_ip4(cred, &ia) == 0) { 482 found = 1; 483 break; 484 } 485 } 486 IF_ADDR_RUNLOCK(ifp); 487 if (!found) { 488 /* 489 * As a last resort return the 'default' jail address. 490 */ 491 ia = ((struct sockaddr_in *)rt->rt_ifa->ifa_addr)-> 492 sin_addr; 493 if (prison_get_ip4(cred, &ia) != 0) 494 return (ESRCH); 495 } 496 bzero(&saun->sin, sizeof(struct sockaddr_in)); 497 saun->sin.sin_len = sizeof(struct sockaddr_in); 498 saun->sin.sin_family = AF_INET; 499 saun->sin.sin_addr.s_addr = ia.s_addr; 500 info->rti_info[RTAX_IFA] = (struct sockaddr *)&saun->sin; 501 break; 502 } 503#endif 504#ifdef INET6 505 case AF_INET6: 506 { 507 struct in6_addr ia6; 508 struct ifaddr *ifa; 509 int found; 510 511 found = 0; 512 /* 513 * Try to find an address on the given outgoing interface 514 * that belongs to the jail. 515 */ 516 IF_ADDR_RLOCK(ifp); 517 TAILQ_FOREACH(ifa, &ifp->if_addrhead, ifa_link) { 518 struct sockaddr *sa; 519 sa = ifa->ifa_addr; 520 if (sa->sa_family != AF_INET6) 521 continue; 522 bcopy(&((struct sockaddr_in6 *)sa)->sin6_addr, 523 &ia6, sizeof(struct in6_addr)); 524 if (prison_check_ip6(cred, &ia6) == 0) { 525 found = 1; 526 break; 527 } 528 } 529 IF_ADDR_RUNLOCK(ifp); 530 if (!found) { 531 /* 532 * As a last resort return the 'default' jail address. 533 */ 534 ia6 = ((struct sockaddr_in6 *)rt->rt_ifa->ifa_addr)-> 535 sin6_addr; 536 if (prison_get_ip6(cred, &ia6) != 0) 537 return (ESRCH); 538 } 539 bzero(&saun->sin6, sizeof(struct sockaddr_in6)); 540 saun->sin6.sin6_len = sizeof(struct sockaddr_in6); 541 saun->sin6.sin6_family = AF_INET6; 542 bcopy(&ia6, &saun->sin6.sin6_addr, sizeof(struct in6_addr)); 543 if (sa6_recoverscope(&saun->sin6) != 0) 544 return (ESRCH); 545 info->rti_info[RTAX_IFA] = (struct sockaddr *)&saun->sin6; 546 break; 547 } 548#endif 549 default: 550 return (ESRCH); 551 } 552 return (0); 553} 554 555/*ARGSUSED*/ 556static int 557route_output(struct mbuf *m, struct socket *so) 558{ 559#define sa_equal(a1, a2) (bcmp((a1), (a2), (a1)->sa_len) == 0) 560 struct rt_msghdr *rtm = NULL; 561 struct rtentry *rt = NULL; 562 struct radix_node_head *rnh; 563 struct rt_addrinfo info; 564 int len, error = 0; 565 struct ifnet *ifp = NULL; 566 union sockaddr_union saun; 567 sa_family_t saf = AF_UNSPEC; 568 569#define senderr(e) { error = e; goto flush;} 570 if (m == NULL || ((m->m_len < sizeof(long)) && 571 (m = m_pullup(m, sizeof(long))) == NULL)) 572 return (ENOBUFS); 573 if ((m->m_flags & M_PKTHDR) == 0) 574 panic("route_output"); 575 len = m->m_pkthdr.len; 576 if (len < sizeof(*rtm) || 577 len != mtod(m, struct rt_msghdr *)->rtm_msglen) { 578 info.rti_info[RTAX_DST] = NULL; 579 senderr(EINVAL); 580 } 581 R_Malloc(rtm, struct rt_msghdr *, len); 582 if (rtm == NULL) { 583 info.rti_info[RTAX_DST] = NULL; 584 senderr(ENOBUFS); 585 } 586 m_copydata(m, 0, len, (caddr_t)rtm); 587 if (rtm->rtm_version != RTM_VERSION) { 588 info.rti_info[RTAX_DST] = NULL; 589 senderr(EPROTONOSUPPORT); 590 } 591 rtm->rtm_pid = curproc->p_pid; 592 bzero(&info, sizeof(info)); 593 info.rti_addrs = rtm->rtm_addrs; 594 if (rt_xaddrs((caddr_t)(rtm + 1), len + (caddr_t)rtm, &info)) { 595 info.rti_info[RTAX_DST] = NULL; 596 senderr(EINVAL); 597 } 598 info.rti_flags = rtm->rtm_flags; 599 if (info.rti_info[RTAX_DST] == NULL || 600 info.rti_info[RTAX_DST]->sa_family >= AF_MAX || 601 (info.rti_info[RTAX_GATEWAY] != NULL && 602 info.rti_info[RTAX_GATEWAY]->sa_family >= AF_MAX)) 603 senderr(EINVAL); 604 saf = info.rti_info[RTAX_DST]->sa_family; 605 /* 606 * Verify that the caller has the appropriate privilege; RTM_GET 607 * is the only operation the non-superuser is allowed. 608 */ 609 if (rtm->rtm_type != RTM_GET) { 610 error = priv_check(curthread, PRIV_NET_ROUTE); 611 if (error) 612 senderr(error); 613 } 614 615 /* 616 * The given gateway address may be an interface address. 617 * For example, issuing a "route change" command on a route 618 * entry that was created from a tunnel, and the gateway 619 * address given is the local end point. In this case the 620 * RTF_GATEWAY flag must be cleared or the destination will 621 * not be reachable even though there is no error message. 622 */ 623 if (info.rti_info[RTAX_GATEWAY] != NULL && 624 info.rti_info[RTAX_GATEWAY]->sa_family != AF_LINK) { 625 struct route gw_ro; 626 627 bzero(&gw_ro, sizeof(gw_ro)); 628 gw_ro.ro_dst = *info.rti_info[RTAX_GATEWAY]; 629 rtalloc_ign_fib(&gw_ro, 0, so->so_fibnum); 630 /* 631 * A host route through the loopback interface is 632 * installed for each interface adddress. In pre 8.0 633 * releases the interface address of a PPP link type 634 * is not reachable locally. This behavior is fixed as 635 * part of the new L2/L3 redesign and rewrite work. The 636 * signature of this interface address route is the 637 * AF_LINK sa_family type of the rt_gateway, and the 638 * rt_ifp has the IFF_LOOPBACK flag set. 639 */ 640 if (gw_ro.ro_rt != NULL && 641 gw_ro.ro_rt->rt_gateway->sa_family == AF_LINK && 642 gw_ro.ro_rt->rt_ifp->if_flags & IFF_LOOPBACK) 643 info.rti_flags &= ~RTF_GATEWAY; 644 if (gw_ro.ro_rt != NULL) 645 RTFREE(gw_ro.ro_rt); 646 } 647 648 switch (rtm->rtm_type) { 649 struct rtentry *saved_nrt; 650 651 case RTM_ADD: 652 if (info.rti_info[RTAX_GATEWAY] == NULL) 653 senderr(EINVAL); 654 saved_nrt = NULL; 655 656 /* support for new ARP code */ 657 if (info.rti_info[RTAX_GATEWAY]->sa_family == AF_LINK && 658 (rtm->rtm_flags & RTF_LLDATA) != 0) { 659 error = lla_rt_output(rtm, &info); 660 break; 661 } 662 error = rtrequest1_fib(RTM_ADD, &info, &saved_nrt, 663 so->so_fibnum); 664 if (error == 0 && saved_nrt) { 665 RT_LOCK(saved_nrt); 666 rt_setmetrics(rtm->rtm_inits, 667 &rtm->rtm_rmx, &saved_nrt->rt_rmx); 668 rtm->rtm_index = saved_nrt->rt_ifp->if_index; 669 RT_REMREF(saved_nrt); 670 RT_UNLOCK(saved_nrt); 671 } 672 break; 673 674 case RTM_DELETE: 675 saved_nrt = NULL; 676 /* support for new ARP code */ 677 if (info.rti_info[RTAX_GATEWAY] && 678 (info.rti_info[RTAX_GATEWAY]->sa_family == AF_LINK) && 679 (rtm->rtm_flags & RTF_LLDATA) != 0) { 680 error = lla_rt_output(rtm, &info); 681 break; 682 } 683 error = rtrequest1_fib(RTM_DELETE, &info, &saved_nrt, 684 so->so_fibnum); 685 if (error == 0) { 686 RT_LOCK(saved_nrt); 687 rt = saved_nrt; 688 goto report; 689 } 690 break; 691 692 case RTM_GET: 693 case RTM_CHANGE: 694 case RTM_LOCK: 695 rnh = rt_tables_get_rnh(so->so_fibnum, 696 info.rti_info[RTAX_DST]->sa_family); 697 if (rnh == NULL) 698 senderr(EAFNOSUPPORT); 699 700 RADIX_NODE_HEAD_RLOCK(rnh); 701 702 if (info.rti_info[RTAX_NETMASK] == NULL && 703 rtm->rtm_type == RTM_GET) { 704 /* 705 * Provide logest prefix match for 706 * address lookup (no mask). 707 * 'route -n get addr' 708 */ 709 rt = (struct rtentry *) rnh->rnh_matchaddr( 710 info.rti_info[RTAX_DST], rnh); 711 } else 712 rt = (struct rtentry *) rnh->rnh_lookup( 713 info.rti_info[RTAX_DST], 714 info.rti_info[RTAX_NETMASK], rnh); 715 716 if (rt == NULL) { 717 RADIX_NODE_HEAD_RUNLOCK(rnh); 718 senderr(ESRCH); 719 } 720#ifdef RADIX_MPATH 721 /* 722 * for RTM_CHANGE/LOCK, if we got multipath routes, 723 * we require users to specify a matching RTAX_GATEWAY. 724 * 725 * for RTM_GET, gate is optional even with multipath. 726 * if gate == NULL the first match is returned. 727 * (no need to call rt_mpath_matchgate if gate == NULL) 728 */ 729 if (rn_mpath_capable(rnh) && 730 (rtm->rtm_type != RTM_GET || info.rti_info[RTAX_GATEWAY])) { 731 rt = rt_mpath_matchgate(rt, info.rti_info[RTAX_GATEWAY]); 732 if (!rt) { 733 RADIX_NODE_HEAD_RUNLOCK(rnh); 734 senderr(ESRCH); 735 } 736 } 737#endif 738 /* 739 * If performing proxied L2 entry insertion, and 740 * the actual PPP host entry is found, perform 741 * another search to retrieve the prefix route of 742 * the local end point of the PPP link. 743 */ 744 if (rtm->rtm_flags & RTF_ANNOUNCE) { 745 struct sockaddr laddr; 746 747 if (rt->rt_ifp != NULL && 748 rt->rt_ifp->if_type == IFT_PROPVIRTUAL) { 749 struct ifaddr *ifa; 750 751 ifa = ifa_ifwithnet(info.rti_info[RTAX_DST], 1); 752 if (ifa != NULL) 753 rt_maskedcopy(ifa->ifa_addr, 754 &laddr, 755 ifa->ifa_netmask); 756 } else 757 rt_maskedcopy(rt->rt_ifa->ifa_addr, 758 &laddr, 759 rt->rt_ifa->ifa_netmask); 760 /* 761 * refactor rt and no lock operation necessary 762 */ 763 rt = (struct rtentry *)rnh->rnh_matchaddr(&laddr, rnh); 764 if (rt == NULL) { 765 RADIX_NODE_HEAD_RUNLOCK(rnh); 766 senderr(ESRCH); 767 } 768 } 769 RT_LOCK(rt); 770 RT_ADDREF(rt); 771 RADIX_NODE_HEAD_RUNLOCK(rnh); 772 773 switch(rtm->rtm_type) { 774 775 case RTM_GET: 776 report: 777 RT_LOCK_ASSERT(rt); 778 if ((rt->rt_flags & RTF_HOST) == 0 779 ? jailed_without_vnet(curthread->td_ucred) 780 : prison_if(curthread->td_ucred, 781 rt_key(rt)) != 0) { 782 RT_UNLOCK(rt); 783 senderr(ESRCH); 784 } 785 info.rti_info[RTAX_DST] = rt_key(rt); 786 info.rti_info[RTAX_GATEWAY] = rt->rt_gateway; 787 info.rti_info[RTAX_NETMASK] = rt_mask(rt); 788 info.rti_info[RTAX_GENMASK] = 0; 789 if (rtm->rtm_addrs & (RTA_IFP | RTA_IFA)) { 790 ifp = rt->rt_ifp; 791 if (ifp) { 792 info.rti_info[RTAX_IFP] = 793 ifp->if_addr->ifa_addr; 794 error = rtm_get_jailed(&info, ifp, rt, 795 &saun, curthread->td_ucred); 796 if (error != 0) { 797 RT_UNLOCK(rt); 798 senderr(error); 799 } 800 if (ifp->if_flags & IFF_POINTOPOINT) 801 info.rti_info[RTAX_BRD] = 802 rt->rt_ifa->ifa_dstaddr; 803 rtm->rtm_index = ifp->if_index; 804 } else { 805 info.rti_info[RTAX_IFP] = NULL; 806 info.rti_info[RTAX_IFA] = NULL; 807 } 808 } else if ((ifp = rt->rt_ifp) != NULL) { 809 rtm->rtm_index = ifp->if_index; 810 } 811 len = rt_msg2(rtm->rtm_type, &info, NULL, NULL); 812 if (len > rtm->rtm_msglen) { 813 struct rt_msghdr *new_rtm; 814 R_Malloc(new_rtm, struct rt_msghdr *, len); 815 if (new_rtm == NULL) { 816 RT_UNLOCK(rt); 817 senderr(ENOBUFS); 818 } 819 bcopy(rtm, new_rtm, rtm->rtm_msglen); 820 Free(rtm); rtm = new_rtm; 821 } 822 (void)rt_msg2(rtm->rtm_type, &info, (caddr_t)rtm, NULL); 823 rtm->rtm_flags = rt->rt_flags; 824 rt_getmetrics(&rt->rt_rmx, &rtm->rtm_rmx); 825 rtm->rtm_addrs = info.rti_addrs; 826 break; 827 828 case RTM_CHANGE: 829 /* 830 * New gateway could require new ifaddr, ifp; 831 * flags may also be different; ifp may be specified 832 * by ll sockaddr when protocol address is ambiguous 833 */ 834 if (((rt->rt_flags & RTF_GATEWAY) && 835 info.rti_info[RTAX_GATEWAY] != NULL) || 836 info.rti_info[RTAX_IFP] != NULL || 837 (info.rti_info[RTAX_IFA] != NULL && 838 !sa_equal(info.rti_info[RTAX_IFA], 839 rt->rt_ifa->ifa_addr))) { 840 RT_UNLOCK(rt); 841 RADIX_NODE_HEAD_LOCK(rnh); 842 error = rt_getifa_fib(&info, rt->rt_fibnum); 843 /* 844 * XXXRW: Really we should release this 845 * reference later, but this maintains 846 * historical behavior. 847 */ 848 if (info.rti_ifa != NULL) 849 ifa_free(info.rti_ifa); 850 RADIX_NODE_HEAD_UNLOCK(rnh); 851 if (error != 0) 852 senderr(error); 853 RT_LOCK(rt); 854 } 855 if (info.rti_ifa != NULL && 856 info.rti_ifa != rt->rt_ifa && 857 rt->rt_ifa != NULL && 858 rt->rt_ifa->ifa_rtrequest != NULL) { 859 rt->rt_ifa->ifa_rtrequest(RTM_DELETE, rt, 860 &info); 861 ifa_free(rt->rt_ifa); 862 } 863 if (info.rti_info[RTAX_GATEWAY] != NULL) { 864 RT_UNLOCK(rt); 865 RADIX_NODE_HEAD_LOCK(rnh); 866 RT_LOCK(rt); 867 868 error = rt_setgate(rt, rt_key(rt), 869 info.rti_info[RTAX_GATEWAY]); 870 RADIX_NODE_HEAD_UNLOCK(rnh); 871 if (error != 0) { 872 RT_UNLOCK(rt); 873 senderr(error); 874 } 875 rt->rt_flags |= (RTF_GATEWAY & info.rti_flags); 876 } 877 if (info.rti_ifa != NULL && 878 info.rti_ifa != rt->rt_ifa) { 879 ifa_ref(info.rti_ifa); 880 rt->rt_ifa = info.rti_ifa; 881 rt->rt_ifp = info.rti_ifp; 882 } 883 /* Allow some flags to be toggled on change. */ 884 rt->rt_flags = (rt->rt_flags & ~RTF_FMASK) | 885 (rtm->rtm_flags & RTF_FMASK); 886 rt_setmetrics(rtm->rtm_inits, &rtm->rtm_rmx, 887 &rt->rt_rmx); 888 rtm->rtm_index = rt->rt_ifp->if_index; 889 if (rt->rt_ifa && rt->rt_ifa->ifa_rtrequest) 890 rt->rt_ifa->ifa_rtrequest(RTM_ADD, rt, &info); 891 /* FALLTHROUGH */ 892 case RTM_LOCK: 893 /* We don't support locks anymore */ 894 break; 895 } 896 RT_UNLOCK(rt); 897 break; 898 899 default: 900 senderr(EOPNOTSUPP); 901 } 902 903flush: 904 if (rtm) { 905 if (error) 906 rtm->rtm_errno = error; 907 else 908 rtm->rtm_flags |= RTF_DONE; 909 } 910 if (rt) /* XXX can this be true? */ 911 RTFREE(rt); 912 { 913 struct rawcb *rp = NULL; 914 /* 915 * Check to see if we don't want our own messages. 916 */ 917 if ((so->so_options & SO_USELOOPBACK) == 0) { 918 if (route_cb.any_count <= 1) { 919 if (rtm) 920 Free(rtm); 921 m_freem(m); 922 return (error); 923 } 924 /* There is another listener, so construct message */ 925 rp = sotorawcb(so); 926 } 927 if (rtm) { 928 m_copyback(m, 0, rtm->rtm_msglen, (caddr_t)rtm); 929 if (m->m_pkthdr.len < rtm->rtm_msglen) { 930 m_freem(m); 931 m = NULL; 932 } else if (m->m_pkthdr.len > rtm->rtm_msglen) 933 m_adj(m, rtm->rtm_msglen - m->m_pkthdr.len); 934 } 935 if (m) { 936 M_SETFIB(m, so->so_fibnum); 937 m->m_flags |= RTS_FILTER_FIB; 938 if (rp) { 939 /* 940 * XXX insure we don't get a copy by 941 * invalidating our protocol 942 */ 943 unsigned short family = rp->rcb_proto.sp_family; 944 rp->rcb_proto.sp_family = 0; 945 rt_dispatch(m, saf); 946 rp->rcb_proto.sp_family = family; 947 } else 948 rt_dispatch(m, saf); 949 } 950 /* info.rti_info[RTAX_DST] (used above) can point inside of rtm */ 951 if (rtm) 952 Free(rtm); 953 } 954 return (error); 955#undef sa_equal 956} 957 958static void 959rt_setmetrics(u_long which, const struct rt_metrics *in, 960 struct rt_metrics_lite *out) 961{ 962#define metric(f, e) if (which & (f)) out->e = in->e; 963 /* 964 * Only these are stored in the routing entry since introduction 965 * of tcp hostcache. The rest is ignored. 966 */ 967 metric(RTV_MTU, rmx_mtu); 968 metric(RTV_WEIGHT, rmx_weight); 969 /* Userland -> kernel timebase conversion. */ 970 if (which & RTV_EXPIRE) 971 out->rmx_expire = in->rmx_expire ? 972 in->rmx_expire - time_second + time_uptime : 0; 973#undef metric 974} 975 976static void 977rt_getmetrics(const struct rt_metrics_lite *in, struct rt_metrics *out) 978{ 979#define metric(e) out->e = in->e; 980 bzero(out, sizeof(*out)); 981 metric(rmx_mtu); 982 metric(rmx_weight); 983 /* Kernel -> userland timebase conversion. */ 984 out->rmx_expire = in->rmx_expire ? 985 in->rmx_expire - time_uptime + time_second : 0; 986#undef metric 987} 988 989/* 990 * Extract the addresses of the passed sockaddrs. 991 * Do a little sanity checking so as to avoid bad memory references. 992 * This data is derived straight from userland. 993 */ 994static int 995rt_xaddrs(caddr_t cp, caddr_t cplim, struct rt_addrinfo *rtinfo) 996{ 997 struct sockaddr *sa; 998 int i; 999 1000 for (i = 0; i < RTAX_MAX && cp < cplim; i++) { 1001 if ((rtinfo->rti_addrs & (1 << i)) == 0) 1002 continue; 1003 sa = (struct sockaddr *)cp; 1004 /* 1005 * It won't fit. 1006 */ 1007 if (cp + sa->sa_len > cplim) 1008 return (EINVAL); 1009 /* 1010 * there are no more.. quit now 1011 * If there are more bits, they are in error. 1012 * I've seen this. route(1) can evidently generate these. 1013 * This causes kernel to core dump. 1014 * for compatibility, If we see this, point to a safe address. 1015 */ 1016 if (sa->sa_len == 0) { 1017 rtinfo->rti_info[i] = &sa_zero; 1018 return (0); /* should be EINVAL but for compat */ 1019 } 1020 /* accept it */ 1021 rtinfo->rti_info[i] = sa; 1022 cp += SA_SIZE(sa); 1023 } 1024 return (0); 1025} 1026 1027/* 1028 * Used by the routing socket. 1029 */ 1030static struct mbuf * 1031rt_msg1(int type, struct rt_addrinfo *rtinfo) 1032{ 1033 struct rt_msghdr *rtm; 1034 struct mbuf *m; 1035 int i; 1036 struct sockaddr *sa; 1037 int len, dlen; 1038 1039 switch (type) { 1040 1041 case RTM_DELADDR: 1042 case RTM_NEWADDR: 1043 len = sizeof(struct ifa_msghdr); 1044 break; 1045 1046 case RTM_DELMADDR: 1047 case RTM_NEWMADDR: 1048 len = sizeof(struct ifma_msghdr); 1049 break; 1050 1051 case RTM_IFINFO: 1052 len = sizeof(struct if_msghdr); 1053 break; 1054 1055 case RTM_IFANNOUNCE: 1056 case RTM_IEEE80211: 1057 len = sizeof(struct if_announcemsghdr); 1058 break; 1059 1060 default: 1061 len = sizeof(struct rt_msghdr); 1062 } 1063 if (len > MCLBYTES) 1064 panic("rt_msg1"); 1065 m = m_gethdr(M_DONTWAIT, MT_DATA); 1066 if (m && len > MHLEN) { 1067 MCLGET(m, M_DONTWAIT); 1068 if ((m->m_flags & M_EXT) == 0) { 1069 m_free(m); 1070 m = NULL; 1071 } 1072 } 1073 if (m == NULL) 1074 return (m); 1075 m->m_pkthdr.len = m->m_len = len; 1076 m->m_pkthdr.rcvif = NULL; 1077 rtm = mtod(m, struct rt_msghdr *); 1078 bzero((caddr_t)rtm, len); 1079 for (i = 0; i < RTAX_MAX; i++) { 1080 if ((sa = rtinfo->rti_info[i]) == NULL) 1081 continue; 1082 rtinfo->rti_addrs |= (1 << i); 1083 dlen = SA_SIZE(sa); 1084 m_copyback(m, len, dlen, (caddr_t)sa); 1085 len += dlen; 1086 } 1087 if (m->m_pkthdr.len != len) { 1088 m_freem(m); 1089 return (NULL); 1090 } 1091 rtm->rtm_msglen = len; 1092 rtm->rtm_version = RTM_VERSION; 1093 rtm->rtm_type = type; 1094 return (m); 1095} 1096 1097/* 1098 * Used by the sysctl code and routing socket. 1099 */ 1100static int 1101rt_msg2(int type, struct rt_addrinfo *rtinfo, caddr_t cp, struct walkarg *w) 1102{ 1103 int i; 1104 int len, dlen, second_time = 0; 1105 caddr_t cp0; 1106 1107 rtinfo->rti_addrs = 0; 1108again: 1109 switch (type) { 1110 1111 case RTM_DELADDR: 1112 case RTM_NEWADDR: 1113 if (w != NULL && w->w_op == NET_RT_IFLISTL) { 1114#ifdef COMPAT_FREEBSD32 1115 if (w->w_req->flags & SCTL_MASK32) 1116 len = sizeof(struct ifa_msghdrl32); 1117 else 1118#endif 1119 len = sizeof(struct ifa_msghdrl); 1120 } else 1121 len = sizeof(struct ifa_msghdr); 1122 break; 1123 1124 case RTM_IFINFO: 1125#ifdef COMPAT_FREEBSD32 1126 if (w != NULL && w->w_req->flags & SCTL_MASK32) { 1127 if (w->w_op == NET_RT_IFLISTL) 1128 len = sizeof(struct if_msghdrl32); 1129 else 1130 len = sizeof(struct if_msghdr32); 1131 break; 1132 } 1133#endif 1134 if (w != NULL && w->w_op == NET_RT_IFLISTL) 1135 len = sizeof(struct if_msghdrl); 1136 else 1137 len = sizeof(struct if_msghdr); 1138 break; 1139 1140 case RTM_NEWMADDR: 1141 len = sizeof(struct ifma_msghdr); 1142 break; 1143 1144 default: 1145 len = sizeof(struct rt_msghdr); 1146 } 1147 cp0 = cp; 1148 if (cp0) 1149 cp += len; 1150 for (i = 0; i < RTAX_MAX; i++) { 1151 struct sockaddr *sa; 1152 1153 if ((sa = rtinfo->rti_info[i]) == NULL) 1154 continue; 1155 rtinfo->rti_addrs |= (1 << i); 1156 dlen = SA_SIZE(sa); 1157 if (cp) { 1158 bcopy((caddr_t)sa, cp, (unsigned)dlen); 1159 cp += dlen; 1160 } 1161 len += dlen; 1162 } 1163 len = ALIGN(len); 1164 if (cp == NULL && w != NULL && !second_time) { 1165 struct walkarg *rw = w; 1166 1167 if (rw->w_req) { 1168 if (rw->w_tmemsize < len) { 1169 if (rw->w_tmem) 1170 free(rw->w_tmem, M_RTABLE); 1171 rw->w_tmem = (caddr_t) 1172 malloc(len, M_RTABLE, M_NOWAIT); 1173 if (rw->w_tmem) 1174 rw->w_tmemsize = len; 1175 } 1176 if (rw->w_tmem) { 1177 cp = rw->w_tmem; 1178 second_time = 1; 1179 goto again; 1180 } 1181 } 1182 } 1183 if (cp) { 1184 struct rt_msghdr *rtm = (struct rt_msghdr *)cp0; 1185 1186 rtm->rtm_version = RTM_VERSION; 1187 rtm->rtm_type = type; 1188 rtm->rtm_msglen = len; 1189 } 1190 return (len); 1191} 1192 1193/* 1194 * This routine is called to generate a message from the routing 1195 * socket indicating that a redirect has occured, a routing lookup 1196 * has failed, or that a protocol has detected timeouts to a particular 1197 * destination. 1198 */ 1199void 1200rt_missmsg_fib(int type, struct rt_addrinfo *rtinfo, int flags, int error, 1201 int fibnum) 1202{ 1203 struct rt_msghdr *rtm; 1204 struct mbuf *m; 1205 struct sockaddr *sa = rtinfo->rti_info[RTAX_DST]; 1206 1207 if (route_cb.any_count == 0) 1208 return; 1209 m = rt_msg1(type, rtinfo); 1210 if (m == NULL) 1211 return; 1212 1213 if (fibnum != RT_ALL_FIBS) { 1214 KASSERT(fibnum >= 0 && fibnum < rt_numfibs, ("%s: fibnum out " 1215 "of range 0 <= %d < %d", __func__, fibnum, rt_numfibs)); 1216 M_SETFIB(m, fibnum); 1217 m->m_flags |= RTS_FILTER_FIB; 1218 } 1219 1220 rtm = mtod(m, struct rt_msghdr *); 1221 rtm->rtm_flags = RTF_DONE | flags; 1222 rtm->rtm_errno = error; 1223 rtm->rtm_addrs = rtinfo->rti_addrs; 1224 rt_dispatch(m, sa ? sa->sa_family : AF_UNSPEC); 1225} 1226 1227void 1228rt_missmsg(int type, struct rt_addrinfo *rtinfo, int flags, int error) 1229{ 1230 1231 rt_missmsg_fib(type, rtinfo, flags, error, RT_ALL_FIBS); 1232} 1233 1234/* 1235 * This routine is called to generate a message from the routing 1236 * socket indicating that the status of a network interface has changed. 1237 */ 1238void 1239rt_ifmsg(struct ifnet *ifp) 1240{ 1241 struct if_msghdr *ifm; 1242 struct mbuf *m; 1243 struct rt_addrinfo info; 1244 1245 if (route_cb.any_count == 0) 1246 return; 1247 bzero((caddr_t)&info, sizeof(info)); 1248 m = rt_msg1(RTM_IFINFO, &info); 1249 if (m == NULL) 1250 return; 1251 ifm = mtod(m, struct if_msghdr *); 1252 ifm->ifm_index = ifp->if_index; 1253 ifm->ifm_flags = ifp->if_flags | ifp->if_drv_flags; 1254 ifm->ifm_data = ifp->if_data; 1255 ifm->ifm_addrs = 0; 1256 rt_dispatch(m, AF_UNSPEC); 1257} 1258 1259/* 1260 * Announce interface address arrival/withdraw. 1261 * Please do not call directly, use rt_addrmsg(). 1262 * Assume input data to be valid. 1263 * Returns 0 on success. 1264 */ 1265int 1266rtsock_addrmsg(int cmd, struct ifaddr *ifa, int fibnum) 1267{ 1268 struct rt_addrinfo info; 1269 struct sockaddr *sa; 1270 int ncmd; 1271 struct mbuf *m; 1272 struct ifa_msghdr *ifam; 1273 struct ifnet *ifp = ifa->ifa_ifp; 1274 1275 if (route_cb.any_count == 0) 1276 return (0); 1277 1278 ncmd = cmd == RTM_ADD ? RTM_NEWADDR : RTM_DELADDR; 1279 1280 bzero((caddr_t)&info, sizeof(info)); 1281 info.rti_info[RTAX_IFA] = sa = ifa->ifa_addr; 1282 info.rti_info[RTAX_IFP] = ifp->if_addr->ifa_addr; 1283 info.rti_info[RTAX_NETMASK] = ifa->ifa_netmask; 1284 info.rti_info[RTAX_BRD] = ifa->ifa_dstaddr; 1285 if ((m = rt_msg1(ncmd, &info)) == NULL) 1286 return (ENOBUFS); 1287 ifam = mtod(m, struct ifa_msghdr *); 1288 ifam->ifam_index = ifp->if_index; 1289 ifam->ifam_metric = ifa->ifa_metric; 1290 ifam->ifam_flags = ifa->ifa_flags; 1291 ifam->ifam_addrs = info.rti_addrs; 1292 1293 if (fibnum != RT_ALL_FIBS) { 1294 M_SETFIB(m, fibnum); 1295 m->m_flags |= RTS_FILTER_FIB; 1296 } 1297 1298 rt_dispatch(m, sa ? sa->sa_family : AF_UNSPEC); 1299 1300 return (0); 1301} 1302 1303/* 1304 * Announce route addition/removal. 1305 * Please do not call directly, use rt_routemsg(). 1306 * Note that @rt data MAY be inconsistent/invalid: 1307 * if some userland app sends us "invalid" route message (invalid mask, 1308 * no dst, wrong address families, etc...) we need to pass it back 1309 * to app (and any other rtsock consumers) with rtm_errno field set to 1310 * non-zero value. 1311 * 1312 * Returns 0 on success. 1313 */ 1314int 1315rtsock_routemsg(int cmd, struct ifnet *ifp, int error, struct rtentry *rt, 1316 int fibnum) 1317{ 1318 struct rt_addrinfo info; 1319 struct sockaddr *sa; 1320 struct mbuf *m; 1321 struct rt_msghdr *rtm; 1322 1323 if (route_cb.any_count == 0) 1324 return (0); 1325 1326 bzero((caddr_t)&info, sizeof(info)); 1327 info.rti_info[RTAX_NETMASK] = rt_mask(rt); 1328 info.rti_info[RTAX_DST] = sa = rt_key(rt); 1329 info.rti_info[RTAX_GATEWAY] = rt->rt_gateway; 1330 if ((m = rt_msg1(cmd, &info)) == NULL) 1331 return (ENOBUFS); 1332 rtm = mtod(m, struct rt_msghdr *); 1333 rtm->rtm_index = ifp->if_index; 1334 rtm->rtm_flags |= rt->rt_flags; 1335 rtm->rtm_errno = error; 1336 rtm->rtm_addrs = info.rti_addrs; 1337 1338 if (fibnum != RT_ALL_FIBS) { 1339 M_SETFIB(m, fibnum); 1340 m->m_flags |= RTS_FILTER_FIB; 1341 } 1342 1343 rt_dispatch(m, sa ? sa->sa_family : AF_UNSPEC); 1344 1345 return (0); 1346} 1347 1348/* 1349 * This is the analogue to the rt_newaddrmsg which performs the same 1350 * function but for multicast group memberhips. This is easier since 1351 * there is no route state to worry about. 1352 */ 1353void 1354rt_newmaddrmsg(int cmd, struct ifmultiaddr *ifma) 1355{ 1356 struct rt_addrinfo info; 1357 struct mbuf *m = NULL; 1358 struct ifnet *ifp = ifma->ifma_ifp; 1359 struct ifma_msghdr *ifmam; 1360 1361 if (route_cb.any_count == 0) 1362 return; 1363 1364 bzero((caddr_t)&info, sizeof(info)); 1365 info.rti_info[RTAX_IFA] = ifma->ifma_addr; 1366 info.rti_info[RTAX_IFP] = ifp ? ifp->if_addr->ifa_addr : NULL; 1367 /* 1368 * If a link-layer address is present, present it as a ``gateway'' 1369 * (similarly to how ARP entries, e.g., are presented). 1370 */ 1371 info.rti_info[RTAX_GATEWAY] = ifma->ifma_lladdr; 1372 m = rt_msg1(cmd, &info); 1373 if (m == NULL) 1374 return; 1375 ifmam = mtod(m, struct ifma_msghdr *); 1376 KASSERT(ifp != NULL, ("%s: link-layer multicast address w/o ifp\n", 1377 __func__)); 1378 ifmam->ifmam_index = ifp->if_index; 1379 ifmam->ifmam_addrs = info.rti_addrs; 1380 rt_dispatch(m, ifma->ifma_addr ? ifma->ifma_addr->sa_family : AF_UNSPEC); 1381} 1382 1383static struct mbuf * 1384rt_makeifannouncemsg(struct ifnet *ifp, int type, int what, 1385 struct rt_addrinfo *info) 1386{ 1387 struct if_announcemsghdr *ifan; 1388 struct mbuf *m; 1389 1390 if (route_cb.any_count == 0) 1391 return NULL; 1392 bzero((caddr_t)info, sizeof(*info)); 1393 m = rt_msg1(type, info); 1394 if (m != NULL) { 1395 ifan = mtod(m, struct if_announcemsghdr *); 1396 ifan->ifan_index = ifp->if_index; 1397 strlcpy(ifan->ifan_name, ifp->if_xname, 1398 sizeof(ifan->ifan_name)); 1399 ifan->ifan_what = what; 1400 } 1401 return m; 1402} 1403 1404/* 1405 * This is called to generate routing socket messages indicating 1406 * IEEE80211 wireless events. 1407 * XXX we piggyback on the RTM_IFANNOUNCE msg format in a clumsy way. 1408 */ 1409void 1410rt_ieee80211msg(struct ifnet *ifp, int what, void *data, size_t data_len) 1411{ 1412 struct mbuf *m; 1413 struct rt_addrinfo info; 1414 1415 m = rt_makeifannouncemsg(ifp, RTM_IEEE80211, what, &info); 1416 if (m != NULL) { 1417 /* 1418 * Append the ieee80211 data. Try to stick it in the 1419 * mbuf containing the ifannounce msg; otherwise allocate 1420 * a new mbuf and append. 1421 * 1422 * NB: we assume m is a single mbuf. 1423 */ 1424 if (data_len > M_TRAILINGSPACE(m)) { 1425 struct mbuf *n = m_get(M_NOWAIT, MT_DATA); 1426 if (n == NULL) { 1427 m_freem(m); 1428 return; 1429 } 1430 bcopy(data, mtod(n, void *), data_len); 1431 n->m_len = data_len; 1432 m->m_next = n; 1433 } else if (data_len > 0) { 1434 bcopy(data, mtod(m, u_int8_t *) + m->m_len, data_len); 1435 m->m_len += data_len; 1436 } 1437 if (m->m_flags & M_PKTHDR) 1438 m->m_pkthdr.len += data_len; 1439 mtod(m, struct if_announcemsghdr *)->ifan_msglen += data_len; 1440 rt_dispatch(m, AF_UNSPEC); 1441 } 1442} 1443 1444/* 1445 * This is called to generate routing socket messages indicating 1446 * network interface arrival and departure. 1447 */ 1448void 1449rt_ifannouncemsg(struct ifnet *ifp, int what) 1450{ 1451 struct mbuf *m; 1452 struct rt_addrinfo info; 1453 1454 m = rt_makeifannouncemsg(ifp, RTM_IFANNOUNCE, what, &info); 1455 if (m != NULL) 1456 rt_dispatch(m, AF_UNSPEC); 1457} 1458 1459static void 1460rt_dispatch(struct mbuf *m, sa_family_t saf) 1461{ 1462 struct m_tag *tag; 1463 1464 /* 1465 * Preserve the family from the sockaddr, if any, in an m_tag for 1466 * use when injecting the mbuf into the routing socket buffer from 1467 * the netisr. 1468 */ 1469 if (saf != AF_UNSPEC) { 1470 tag = m_tag_get(PACKET_TAG_RTSOCKFAM, sizeof(unsigned short), 1471 M_NOWAIT); 1472 if (tag == NULL) { 1473 m_freem(m); 1474 return; 1475 } 1476 *(unsigned short *)(tag + 1) = saf; 1477 m_tag_prepend(m, tag); 1478 } 1479#ifdef VIMAGE 1480 if (V_loif) 1481 m->m_pkthdr.rcvif = V_loif; 1482 else { 1483 m_freem(m); 1484 return; 1485 } 1486#endif 1487 netisr_queue(NETISR_ROUTE, m); /* mbuf is free'd on failure. */ 1488} 1489 1490/* 1491 * This is used in dumping the kernel table via sysctl(). 1492 */ 1493static int 1494sysctl_dumpentry(struct radix_node *rn, void *vw) 1495{ 1496 struct walkarg *w = vw; 1497 struct rtentry *rt = (struct rtentry *)rn; 1498 int error = 0, size; 1499 struct rt_addrinfo info; 1500 1501 if (w->w_op == NET_RT_FLAGS && !(rt->rt_flags & w->w_arg)) 1502 return 0; 1503 if ((rt->rt_flags & RTF_HOST) == 0 1504 ? jailed_without_vnet(w->w_req->td->td_ucred) 1505 : prison_if(w->w_req->td->td_ucred, rt_key(rt)) != 0) 1506 return (0); 1507 bzero((caddr_t)&info, sizeof(info)); 1508 info.rti_info[RTAX_DST] = rt_key(rt); 1509 info.rti_info[RTAX_GATEWAY] = rt->rt_gateway; 1510 info.rti_info[RTAX_NETMASK] = rt_mask(rt); 1511 info.rti_info[RTAX_GENMASK] = 0; 1512 if (rt->rt_ifp) { 1513 info.rti_info[RTAX_IFP] = rt->rt_ifp->if_addr->ifa_addr; 1514 info.rti_info[RTAX_IFA] = rt->rt_ifa->ifa_addr; 1515 if (rt->rt_ifp->if_flags & IFF_POINTOPOINT) 1516 info.rti_info[RTAX_BRD] = rt->rt_ifa->ifa_dstaddr; 1517 } 1518 size = rt_msg2(RTM_GET, &info, NULL, w); 1519 if (w->w_req && w->w_tmem) { 1520 struct rt_msghdr *rtm = (struct rt_msghdr *)w->w_tmem; 1521 1522 rtm->rtm_flags = rt->rt_flags; 1523 /* 1524 * let's be honest about this being a retarded hack 1525 */ 1526 rtm->rtm_fmask = rt->rt_rmx.rmx_pksent; 1527 rt_getmetrics(&rt->rt_rmx, &rtm->rtm_rmx); 1528 rtm->rtm_index = rt->rt_ifp->if_index; 1529 rtm->rtm_errno = rtm->rtm_pid = rtm->rtm_seq = 0; 1530 rtm->rtm_addrs = info.rti_addrs; 1531 error = SYSCTL_OUT(w->w_req, (caddr_t)rtm, size); 1532 return (error); 1533 } 1534 return (error); 1535} 1536 1537#ifdef COMPAT_FREEBSD32 1538static void 1539copy_ifdata32(struct if_data *src, struct if_data32 *dst) 1540{ 1541 1542 bzero(dst, sizeof(*dst)); 1543 CP(*src, *dst, ifi_type); 1544 CP(*src, *dst, ifi_physical); 1545 CP(*src, *dst, ifi_addrlen); 1546 CP(*src, *dst, ifi_hdrlen); 1547 CP(*src, *dst, ifi_link_state); 1548 dst->ifi_datalen = sizeof(struct if_data32); 1549 CP(*src, *dst, ifi_mtu); 1550 CP(*src, *dst, ifi_metric); 1551 CP(*src, *dst, ifi_baudrate); 1552 CP(*src, *dst, ifi_ipackets); 1553 CP(*src, *dst, ifi_ierrors); 1554 CP(*src, *dst, ifi_opackets); 1555 CP(*src, *dst, ifi_oerrors); 1556 CP(*src, *dst, ifi_collisions); 1557 CP(*src, *dst, ifi_ibytes); 1558 CP(*src, *dst, ifi_obytes); 1559 CP(*src, *dst, ifi_imcasts); 1560 CP(*src, *dst, ifi_omcasts); 1561 CP(*src, *dst, ifi_iqdrops); 1562 CP(*src, *dst, ifi_noproto); 1563 CP(*src, *dst, ifi_hwassist); 1564 CP(*src, *dst, ifi_epoch); 1565 TV_CP(*src, *dst, ifi_lastchange); 1566} 1567#endif 1568 1569static int 1570sysctl_iflist_ifml(struct ifnet *ifp, struct rt_addrinfo *info, 1571 struct walkarg *w, int len) 1572{ 1573 struct if_msghdrl *ifm; 1574 1575#ifdef COMPAT_FREEBSD32 1576 if (w->w_req->flags & SCTL_MASK32) { 1577 struct if_msghdrl32 *ifm32; 1578 1579 ifm32 = (struct if_msghdrl32 *)w->w_tmem; 1580 ifm32->ifm_addrs = info->rti_addrs; 1581 ifm32->ifm_flags = ifp->if_flags | ifp->if_drv_flags; 1582 ifm32->ifm_index = ifp->if_index; 1583 ifm32->_ifm_spare1 = 0; 1584 ifm32->ifm_len = sizeof(*ifm32); 1585 ifm32->ifm_data_off = offsetof(struct if_msghdrl32, ifm_data); 1586 1587 copy_ifdata32(&ifp->if_data, &ifm32->ifm_data); 1588 1589 return (SYSCTL_OUT(w->w_req, (caddr_t)ifm32, len)); 1590 } 1591#endif 1592 ifm = (struct if_msghdrl *)w->w_tmem; 1593 ifm->ifm_addrs = info->rti_addrs; 1594 ifm->ifm_flags = ifp->if_flags | ifp->if_drv_flags; 1595 ifm->ifm_index = ifp->if_index; 1596 ifm->_ifm_spare1 = 0; 1597 ifm->ifm_len = sizeof(*ifm); 1598 ifm->ifm_data_off = offsetof(struct if_msghdrl, ifm_data); 1599 1600 ifm->ifm_data = ifp->if_data; 1601 1602 return (SYSCTL_OUT(w->w_req, (caddr_t)ifm, len)); 1603} 1604 1605static int 1606sysctl_iflist_ifm(struct ifnet *ifp, struct rt_addrinfo *info, 1607 struct walkarg *w, int len) 1608{ 1609 struct if_msghdr *ifm; 1610 1611#ifdef COMPAT_FREEBSD32 1612 if (w->w_req->flags & SCTL_MASK32) { 1613 struct if_msghdr32 *ifm32; 1614 1615 ifm32 = (struct if_msghdr32 *)w->w_tmem; 1616 ifm32->ifm_addrs = info->rti_addrs; 1617 ifm32->ifm_flags = ifp->if_flags | ifp->if_drv_flags; 1618 ifm32->ifm_index = ifp->if_index; 1619 1620 copy_ifdata32(&ifp->if_data, &ifm32->ifm_data); 1621 1622 return (SYSCTL_OUT(w->w_req, (caddr_t)ifm32, len)); 1623 } 1624#endif 1625 ifm = (struct if_msghdr *)w->w_tmem; 1626 ifm->ifm_addrs = info->rti_addrs; 1627 ifm->ifm_flags = ifp->if_flags | ifp->if_drv_flags; 1628 ifm->ifm_index = ifp->if_index; 1629 1630 ifm->ifm_data = ifp->if_data; 1631 1632 return (SYSCTL_OUT(w->w_req, (caddr_t)ifm, len)); 1633} 1634 1635static int 1636sysctl_iflist_ifaml(struct ifaddr *ifa, struct rt_addrinfo *info, 1637 struct walkarg *w, int len) 1638{ 1639 struct ifa_msghdrl *ifam; 1640 1641#ifdef COMPAT_FREEBSD32 1642 if (w->w_req->flags & SCTL_MASK32) { 1643 struct ifa_msghdrl32 *ifam32; 1644 1645 ifam32 = (struct ifa_msghdrl32 *)w->w_tmem; 1646 ifam32->ifam_addrs = info->rti_addrs; 1647 ifam32->ifam_flags = ifa->ifa_flags; 1648 ifam32->ifam_index = ifa->ifa_ifp->if_index; 1649 ifam32->_ifam_spare1 = 0; 1650 ifam32->ifam_len = sizeof(*ifam32); 1651 ifam32->ifam_data_off = 1652 offsetof(struct ifa_msghdrl32, ifam_data); 1653 ifam32->ifam_metric = ifa->ifa_metric; 1654 1655 copy_ifdata32(&ifa->ifa_ifp->if_data, &ifam32->ifam_data); 1656 1657 return (SYSCTL_OUT(w->w_req, (caddr_t)ifam32, len)); 1658 } 1659#endif 1660 1661 ifam = (struct ifa_msghdrl *)w->w_tmem; 1662 ifam->ifam_addrs = info->rti_addrs; 1663 ifam->ifam_flags = ifa->ifa_flags; 1664 ifam->ifam_index = ifa->ifa_ifp->if_index; 1665 ifam->_ifam_spare1 = 0; 1666 ifam->ifam_len = sizeof(*ifam); 1667 ifam->ifam_data_off = offsetof(struct ifa_msghdrl, ifam_data); 1668 ifam->ifam_metric = ifa->ifa_metric; 1669 1670 ifam->ifam_data = ifa->if_data; 1671 1672 return (SYSCTL_OUT(w->w_req, w->w_tmem, len)); 1673} 1674 1675static int 1676sysctl_iflist_ifam(struct ifaddr *ifa, struct rt_addrinfo *info, 1677 struct walkarg *w, int len) 1678{ 1679 struct ifa_msghdr *ifam; 1680 1681 ifam = (struct ifa_msghdr *)w->w_tmem; 1682 ifam->ifam_addrs = info->rti_addrs; 1683 ifam->ifam_flags = ifa->ifa_flags; 1684 ifam->ifam_index = ifa->ifa_ifp->if_index; 1685 ifam->ifam_metric = ifa->ifa_metric; 1686 1687 return (SYSCTL_OUT(w->w_req, w->w_tmem, len)); 1688} 1689 1690static int 1691sysctl_iflist(int af, struct walkarg *w) 1692{ 1693 struct ifnet *ifp; 1694 struct ifaddr *ifa; 1695 struct rt_addrinfo info; 1696 int len, error = 0; 1697 1698 bzero((caddr_t)&info, sizeof(info)); 1699 IFNET_RLOCK(); 1700 TAILQ_FOREACH(ifp, &V_ifnet, if_link) { 1701 if (w->w_arg && w->w_arg != ifp->if_index) 1702 continue; 1703 IF_ADDR_RLOCK(ifp); 1704 ifa = ifp->if_addr; 1705 info.rti_info[RTAX_IFP] = ifa->ifa_addr; 1706 len = rt_msg2(RTM_IFINFO, &info, NULL, w); 1707 info.rti_info[RTAX_IFP] = NULL; 1708 if (w->w_req && w->w_tmem) { 1709 if (w->w_op == NET_RT_IFLISTL) 1710 error = sysctl_iflist_ifml(ifp, &info, w, len); 1711 else 1712 error = sysctl_iflist_ifm(ifp, &info, w, len); 1713 if (error) 1714 goto done; 1715 } 1716 while ((ifa = TAILQ_NEXT(ifa, ifa_link)) != NULL) { 1717 if (af && af != ifa->ifa_addr->sa_family) 1718 continue; 1719 if (prison_if(w->w_req->td->td_ucred, 1720 ifa->ifa_addr) != 0) 1721 continue; 1722 info.rti_info[RTAX_IFA] = ifa->ifa_addr; 1723 info.rti_info[RTAX_NETMASK] = ifa->ifa_netmask; 1724 info.rti_info[RTAX_BRD] = ifa->ifa_dstaddr; 1725 len = rt_msg2(RTM_NEWADDR, &info, NULL, w); 1726 if (w->w_req && w->w_tmem) { 1727 if (w->w_op == NET_RT_IFLISTL) 1728 error = sysctl_iflist_ifaml(ifa, &info, 1729 w, len); 1730 else 1731 error = sysctl_iflist_ifam(ifa, &info, 1732 w, len); 1733 if (error) 1734 goto done; 1735 } 1736 } 1737 IF_ADDR_RUNLOCK(ifp); 1738 info.rti_info[RTAX_IFA] = info.rti_info[RTAX_NETMASK] = 1739 info.rti_info[RTAX_BRD] = NULL; 1740 } 1741done: 1742 if (ifp != NULL) 1743 IF_ADDR_RUNLOCK(ifp); 1744 IFNET_RUNLOCK(); 1745 return (error); 1746} 1747 1748static int 1749sysctl_ifmalist(int af, struct walkarg *w) 1750{ 1751 struct ifnet *ifp; 1752 struct ifmultiaddr *ifma; 1753 struct rt_addrinfo info; 1754 int len, error = 0; 1755 struct ifaddr *ifa; 1756 1757 bzero((caddr_t)&info, sizeof(info)); 1758 IFNET_RLOCK(); 1759 TAILQ_FOREACH(ifp, &V_ifnet, if_link) { 1760 if (w->w_arg && w->w_arg != ifp->if_index) 1761 continue; 1762 ifa = ifp->if_addr; 1763 info.rti_info[RTAX_IFP] = ifa ? ifa->ifa_addr : NULL; 1764 IF_ADDR_RLOCK(ifp); 1765 TAILQ_FOREACH(ifma, &ifp->if_multiaddrs, ifma_link) { 1766 if (af && af != ifma->ifma_addr->sa_family) 1767 continue; 1768 if (prison_if(w->w_req->td->td_ucred, 1769 ifma->ifma_addr) != 0) 1770 continue; 1771 info.rti_info[RTAX_IFA] = ifma->ifma_addr; 1772 info.rti_info[RTAX_GATEWAY] = 1773 (ifma->ifma_addr->sa_family != AF_LINK) ? 1774 ifma->ifma_lladdr : NULL; 1775 len = rt_msg2(RTM_NEWMADDR, &info, NULL, w); 1776 if (w->w_req && w->w_tmem) { 1777 struct ifma_msghdr *ifmam; 1778 1779 ifmam = (struct ifma_msghdr *)w->w_tmem; 1780 ifmam->ifmam_index = ifma->ifma_ifp->if_index; 1781 ifmam->ifmam_flags = 0; 1782 ifmam->ifmam_addrs = info.rti_addrs; 1783 error = SYSCTL_OUT(w->w_req, w->w_tmem, len); 1784 if (error) { 1785 IF_ADDR_RUNLOCK(ifp); 1786 goto done; 1787 } 1788 } 1789 } 1790 IF_ADDR_RUNLOCK(ifp); 1791 } 1792done: 1793 IFNET_RUNLOCK(); 1794 return (error); 1795} 1796 1797static int 1798sysctl_rtsock(SYSCTL_HANDLER_ARGS) 1799{ 1800 int *name = (int *)arg1; 1801 u_int namelen = arg2; 1802 struct radix_node_head *rnh = NULL; /* silence compiler. */ 1803 int i, lim, error = EINVAL; 1804 int fib = 0; 1805 u_char af; 1806 struct walkarg w; 1807 1808 name ++; 1809 namelen--; 1810 if (req->newptr) 1811 return (EPERM); 1812 if (name[1] == NET_RT_DUMP) { 1813 if (namelen == 3) 1814 fib = req->td->td_proc->p_fibnum; 1815 else if (namelen == 4) 1816 fib = (name[3] == RT_ALL_FIBS) ? 1817 req->td->td_proc->p_fibnum : name[3]; 1818 else 1819 return ((namelen < 3) ? EISDIR : ENOTDIR); 1820 if (fib < 0 || fib >= rt_numfibs) 1821 return (EINVAL); 1822 } else if (namelen != 3) 1823 return ((namelen < 3) ? EISDIR : ENOTDIR); 1824 af = name[0]; 1825 if (af > AF_MAX) 1826 return (EINVAL); 1827 bzero(&w, sizeof(w)); 1828 w.w_op = name[1]; 1829 w.w_arg = name[2]; 1830 w.w_req = req; 1831 1832 error = sysctl_wire_old_buffer(req, 0); 1833 if (error) 1834 return (error); 1835 switch (w.w_op) { 1836 1837 case NET_RT_DUMP: 1838 case NET_RT_FLAGS: 1839 if (af == 0) { /* dump all tables */ 1840 i = 1; 1841 lim = AF_MAX; 1842 } else /* dump only one table */ 1843 i = lim = af; 1844 1845 /* 1846 * take care of llinfo entries, the caller must 1847 * specify an AF 1848 */ 1849 if (w.w_op == NET_RT_FLAGS && 1850 (w.w_arg == 0 || w.w_arg & RTF_LLINFO)) { 1851 if (af != 0) 1852 error = lltable_sysctl_dumparp(af, w.w_req); 1853 else 1854 error = EINVAL; 1855 break; 1856 } 1857 /* 1858 * take care of routing entries 1859 */ 1860 for (error = 0; error == 0 && i <= lim; i++) { 1861 rnh = rt_tables_get_rnh(fib, i); 1862 if (rnh != NULL) { 1863 RADIX_NODE_HEAD_RLOCK(rnh); 1864 error = rnh->rnh_walktree(rnh, 1865 sysctl_dumpentry, &w); 1866 RADIX_NODE_HEAD_RUNLOCK(rnh); 1867 } else if (af != 0) 1868 error = EAFNOSUPPORT; 1869 } 1870 break; 1871 1872 case NET_RT_IFLIST: 1873 case NET_RT_IFLISTL: 1874 error = sysctl_iflist(af, &w); 1875 break; 1876 1877 case NET_RT_IFMALIST: 1878 error = sysctl_ifmalist(af, &w); 1879 break; 1880 } 1881 if (w.w_tmem) 1882 free(w.w_tmem, M_RTABLE); 1883 return (error); 1884} 1885 1886static SYSCTL_NODE(_net, PF_ROUTE, routetable, CTLFLAG_RD, sysctl_rtsock, ""); 1887 1888/* 1889 * Definitions of protocols supported in the ROUTE domain. 1890 */ 1891 1892static struct domain routedomain; /* or at least forward */ 1893 1894static struct protosw routesw[] = { 1895{ 1896 .pr_type = SOCK_RAW, 1897 .pr_domain = &routedomain, 1898 .pr_flags = PR_ATOMIC|PR_ADDR, 1899 .pr_output = route_output, 1900 .pr_ctlinput = raw_ctlinput, 1901 .pr_init = raw_init, 1902 .pr_usrreqs = &route_usrreqs 1903} 1904}; 1905 1906static struct domain routedomain = { 1907 .dom_family = PF_ROUTE, 1908 .dom_name = "route", 1909 .dom_protosw = routesw, 1910 .dom_protoswNPROTOSW = &routesw[sizeof(routesw)/sizeof(routesw[0])] 1911}; 1912 1913VNET_DOMAIN_SET(route); 1914