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_sctp.h" 34 #include "opt_mpath.h" 35 #include "opt_inet.h" 36 #include "opt_inet6.h" 37 38 #include <sys/param.h> 39 #include <sys/jail.h> 40 #include <sys/kernel.h> 41 #include <sys/domain.h> 42 #include <sys/lock.h> 43 #include <sys/malloc.h> 44 #include <sys/mbuf.h> 45 #include <sys/priv.h> 46 #include <sys/proc.h> 47 #include <sys/protosw.h> 48 #include <sys/rwlock.h> 49 #include <sys/signalvar.h> 50 #include <sys/socket.h> 51 #include <sys/socketvar.h> 52 #include <sys/sysctl.h> 53 #include <sys/systm.h> 54 55 #include <net/if.h> 56 #include <net/if_var.h> 57 #include <net/if_dl.h> 58 #include <net/if_llatbl.h> 59 #include <net/if_types.h> 60 #include <net/netisr.h> 61 #include <net/raw_cb.h> 62 #include <net/route.h> 63 #include <net/vnet.h> 64 65 #include <netinet/in.h> 66 #include <netinet/if_ether.h> 67 #include <netinet/ip_carp.h> 68 #ifdef INET6 69 #include <netinet6/ip6_var.h> 70 #include <netinet6/scope6_var.h> 71 #endif 72 73 #if defined(INET) || defined(INET6) 74 #ifdef SCTP 75 extern void sctp_addr_change(struct ifaddr *ifa, int cmd); 76 #endif /* SCTP */ 77 #endif 78 79 #ifdef COMPAT_FREEBSD32 80 #include <sys/mount.h> 81 #include <compat/freebsd32/freebsd32.h> 82 83 struct if_data32 { 84 uint8_t ifi_type; 85 uint8_t ifi_physical; 86 uint8_t ifi_addrlen; 87 uint8_t ifi_hdrlen; 88 uint8_t ifi_link_state; 89 uint8_t ifi_vhid; 90 uint8_t ifi_baudrate_pf; 91 uint8_t ifi_datalen; 92 uint32_t ifi_mtu; 93 uint32_t ifi_metric; 94 uint32_t ifi_baudrate; 95 uint32_t ifi_ipackets; 96 uint32_t ifi_ierrors; 97 uint32_t ifi_opackets; 98 uint32_t ifi_oerrors; 99 uint32_t ifi_collisions; 100 uint32_t ifi_ibytes; 101 uint32_t ifi_obytes; 102 uint32_t ifi_imcasts; 103 uint32_t ifi_omcasts; 104 uint32_t ifi_iqdrops; 105 uint32_t ifi_noproto; 106 uint32_t ifi_hwassist; 107 int32_t ifi_epoch; 108 struct timeval32 ifi_lastchange; 109 }; 110 111 struct if_msghdr32 { 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 struct if_data32 ifm_data; 119 }; 120 121 struct if_msghdrl32 { 122 uint16_t ifm_msglen; 123 uint8_t ifm_version; 124 uint8_t ifm_type; 125 int32_t ifm_addrs; 126 int32_t ifm_flags; 127 uint16_t ifm_index; 128 uint16_t _ifm_spare1; 129 uint16_t ifm_len; 130 uint16_t ifm_data_off; 131 struct if_data32 ifm_data; 132 }; 133 134 struct ifa_msghdrl32 { 135 uint16_t ifam_msglen; 136 uint8_t ifam_version; 137 uint8_t ifam_type; 138 int32_t ifam_addrs; 139 int32_t ifam_flags; 140 uint16_t ifam_index; 141 uint16_t _ifam_spare1; 142 uint16_t ifam_len; 143 uint16_t ifam_data_off; 144 int32_t ifam_metric; 145 struct if_data32 ifam_data; 146 }; 147 #endif /* COMPAT_FREEBSD32 */ 148 149 MALLOC_DEFINE(M_RTABLE, "routetbl", "routing tables"); 150 151 /* NB: these are not modified */ 152 static struct sockaddr route_src = { 2, PF_ROUTE, }; 153 static struct sockaddr sa_zero = { sizeof(sa_zero), AF_INET, }; 154 155 /* These are external hooks for CARP. */ 156 int (*carp_get_vhid_p)(struct ifaddr *); 157 158 /* 159 * Used by rtsock/raw_input callback code to decide whether to filter the update 160 * notification to a socket bound to a particular FIB. 161 */ 162 #define RTS_FILTER_FIB M_PROTO8 163 #define RTS_ALLFIBS -1 164 165 static struct { 166 int ip_count; /* attached w/ AF_INET */ 167 int ip6_count; /* attached w/ AF_INET6 */ 168 int ipx_count; /* attached w/ AF_IPX */ 169 int any_count; /* total attached */ 170 } route_cb; 171 172 struct mtx rtsock_mtx; 173 MTX_SYSINIT(rtsock, &rtsock_mtx, "rtsock route_cb lock", MTX_DEF); 174 175 #define RTSOCK_LOCK() mtx_lock(&rtsock_mtx) 176 #define RTSOCK_UNLOCK() mtx_unlock(&rtsock_mtx) 177 #define RTSOCK_LOCK_ASSERT() mtx_assert(&rtsock_mtx, MA_OWNED) 178 179 static SYSCTL_NODE(_net, OID_AUTO, route, CTLFLAG_RD, 0, ""); 180 181 struct walkarg { 182 int w_tmemsize; 183 int w_op, w_arg; 184 caddr_t w_tmem; 185 struct sysctl_req *w_req; 186 }; 187 188 static void rts_input(struct mbuf *m); 189 static struct mbuf *rt_msg1(int type, struct rt_addrinfo *rtinfo); 190 static int rt_msg2(int type, struct rt_addrinfo *rtinfo, 191 caddr_t cp, struct walkarg *w); 192 static int rt_xaddrs(caddr_t cp, caddr_t cplim, 193 struct rt_addrinfo *rtinfo); 194 static int sysctl_dumpentry(struct radix_node *rn, void *vw); 195 static int sysctl_iflist(int af, struct walkarg *w); 196 static int sysctl_ifmalist(int af, struct walkarg *w); 197 static int route_output(struct mbuf *m, struct socket *so); 198 static void rt_setmetrics(u_long which, const struct rt_metrics *in, 199 struct rt_metrics_lite *out); 200 static void rt_getmetrics(const struct rt_metrics_lite *in, 201 struct rt_metrics *out); 202 static void rt_dispatch(struct mbuf *, sa_family_t); 203 204 static struct netisr_handler rtsock_nh = { 205 .nh_name = "rtsock", 206 .nh_handler = rts_input, 207 .nh_proto = NETISR_ROUTE, 208 .nh_policy = NETISR_POLICY_SOURCE, 209 }; 210 211 static int 212 sysctl_route_netisr_maxqlen(SYSCTL_HANDLER_ARGS) 213 { 214 int error, qlimit; 215 216 netisr_getqlimit(&rtsock_nh, &qlimit); 217 error = sysctl_handle_int(oidp, &qlimit, 0, req); 218 if (error || !req->newptr) 219 return (error); 220 if (qlimit < 1) 221 return (EINVAL); 222 return (netisr_setqlimit(&rtsock_nh, qlimit)); 223 } 224 SYSCTL_PROC(_net_route, OID_AUTO, netisr_maxqlen, CTLTYPE_INT|CTLFLAG_RW, 225 0, 0, sysctl_route_netisr_maxqlen, "I", 226 "maximum routing socket dispatch queue length"); 227 228 static void 229 rts_init(void) 230 { 231 int tmp; 232 233 if (TUNABLE_INT_FETCH("net.route.netisr_maxqlen", &tmp)) 234 rtsock_nh.nh_qlimit = tmp; 235 netisr_register(&rtsock_nh); 236 } 237 SYSINIT(rtsock, SI_SUB_PROTO_DOMAIN, SI_ORDER_THIRD, rts_init, 0); 238 239 static int 240 raw_input_rts_cb(struct mbuf *m, struct sockproto *proto, struct sockaddr *src, 241 struct rawcb *rp) 242 { 243 int fibnum; 244 245 KASSERT(m != NULL, ("%s: m is NULL", __func__)); 246 KASSERT(proto != NULL, ("%s: proto is NULL", __func__)); 247 KASSERT(rp != NULL, ("%s: rp is NULL", __func__)); 248 249 /* No filtering requested. */ 250 if ((m->m_flags & RTS_FILTER_FIB) == 0) 251 return (0); 252 253 /* Check if it is a rts and the fib matches the one of the socket. */ 254 fibnum = M_GETFIB(m); 255 if (proto->sp_family != PF_ROUTE || 256 rp->rcb_socket == NULL || 257 rp->rcb_socket->so_fibnum == fibnum) 258 return (0); 259 260 /* Filtering requested and no match, the socket shall be skipped. */ 261 return (1); 262 } 263 264 static void 265 rts_input(struct mbuf *m) 266 { 267 struct sockproto route_proto; 268 unsigned short *family; 269 struct m_tag *tag; 270 271 route_proto.sp_family = PF_ROUTE; 272 tag = m_tag_find(m, PACKET_TAG_RTSOCKFAM, NULL); 273 if (tag != NULL) { 274 family = (unsigned short *)(tag + 1); 275 route_proto.sp_protocol = *family; 276 m_tag_delete(m, tag); 277 } else 278 route_proto.sp_protocol = 0; 279 280 raw_input_ext(m, &route_proto, &route_src, raw_input_rts_cb); 281 } 282 283 /* 284 * It really doesn't make any sense at all for this code to share much 285 * with raw_usrreq.c, since its functionality is so restricted. XXX 286 */ 287 static void 288 rts_abort(struct socket *so) 289 { 290 291 raw_usrreqs.pru_abort(so); 292 } 293 294 static void 295 rts_close(struct socket *so) 296 { 297 298 raw_usrreqs.pru_close(so); 299 } 300 301 /* pru_accept is EOPNOTSUPP */ 302 303 static int 304 rts_attach(struct socket *so, int proto, struct thread *td) 305 { 306 struct rawcb *rp; 307 int error; 308 309 KASSERT(so->so_pcb == NULL, ("rts_attach: so_pcb != NULL")); 310 311 /* XXX */ 312 rp = malloc(sizeof *rp, M_PCB, M_WAITOK | M_ZERO); 313 if (rp == NULL) 314 return ENOBUFS; 315 316 so->so_pcb = (caddr_t)rp; 317 so->so_fibnum = td->td_proc->p_fibnum; 318 error = raw_attach(so, proto); 319 rp = sotorawcb(so); 320 if (error) { 321 so->so_pcb = NULL; 322 free(rp, M_PCB); 323 return error; 324 } 325 RTSOCK_LOCK(); 326 switch(rp->rcb_proto.sp_protocol) { 327 case AF_INET: 328 route_cb.ip_count++; 329 break; 330 case AF_INET6: 331 route_cb.ip6_count++; 332 break; 333 case AF_IPX: 334 route_cb.ipx_count++; 335 break; 336 } 337 route_cb.any_count++; 338 RTSOCK_UNLOCK(); 339 soisconnected(so); 340 so->so_options |= SO_USELOOPBACK; 341 return 0; 342 } 343 344 static int 345 rts_bind(struct socket *so, struct sockaddr *nam, struct thread *td) 346 { 347 348 return (raw_usrreqs.pru_bind(so, nam, td)); /* xxx just EINVAL */ 349 } 350 351 static int 352 rts_connect(struct socket *so, struct sockaddr *nam, struct thread *td) 353 { 354 355 return (raw_usrreqs.pru_connect(so, nam, td)); /* XXX just EINVAL */ 356 } 357 358 /* pru_connect2 is EOPNOTSUPP */ 359 /* pru_control is EOPNOTSUPP */ 360 361 static void 362 rts_detach(struct socket *so) 363 { 364 struct rawcb *rp = sotorawcb(so); 365 366 KASSERT(rp != NULL, ("rts_detach: rp == NULL")); 367 368 RTSOCK_LOCK(); 369 switch(rp->rcb_proto.sp_protocol) { 370 case AF_INET: 371 route_cb.ip_count--; 372 break; 373 case AF_INET6: 374 route_cb.ip6_count--; 375 break; 376 case AF_IPX: 377 route_cb.ipx_count--; 378 break; 379 } 380 route_cb.any_count--; 381 RTSOCK_UNLOCK(); 382 raw_usrreqs.pru_detach(so); 383 } 384 385 static int 386 rts_disconnect(struct socket *so) 387 { 388 389 return (raw_usrreqs.pru_disconnect(so)); 390 } 391 392 /* pru_listen is EOPNOTSUPP */ 393 394 static int 395 rts_peeraddr(struct socket *so, struct sockaddr **nam) 396 { 397 398 return (raw_usrreqs.pru_peeraddr(so, nam)); 399 } 400 401 /* pru_rcvd is EOPNOTSUPP */ 402 /* pru_rcvoob is EOPNOTSUPP */ 403 404 static int 405 rts_send(struct socket *so, int flags, struct mbuf *m, struct sockaddr *nam, 406 struct mbuf *control, struct thread *td) 407 { 408 409 return (raw_usrreqs.pru_send(so, flags, m, nam, control, td)); 410 } 411 412 /* pru_sense is null */ 413 414 static int 415 rts_shutdown(struct socket *so) 416 { 417 418 return (raw_usrreqs.pru_shutdown(so)); 419 } 420 421 static int 422 rts_sockaddr(struct socket *so, struct sockaddr **nam) 423 { 424 425 return (raw_usrreqs.pru_sockaddr(so, nam)); 426 } 427 428 static struct pr_usrreqs route_usrreqs = { 429 .pru_abort = rts_abort, 430 .pru_attach = rts_attach, 431 .pru_bind = rts_bind, 432 .pru_connect = rts_connect, 433 .pru_detach = rts_detach, 434 .pru_disconnect = rts_disconnect, 435 .pru_peeraddr = rts_peeraddr, 436 .pru_send = rts_send, 437 .pru_shutdown = rts_shutdown, 438 .pru_sockaddr = rts_sockaddr, 439 .pru_close = rts_close, 440 }; 441 442 #ifndef _SOCKADDR_UNION_DEFINED 443 #define _SOCKADDR_UNION_DEFINED 444 /* 445 * The union of all possible address formats we handle. 446 */ 447 union sockaddr_union { 448 struct sockaddr sa; 449 struct sockaddr_in sin; 450 struct sockaddr_in6 sin6; 451 }; 452 #endif /* _SOCKADDR_UNION_DEFINED */ 453 454 static int 455 rtm_get_jailed(struct rt_addrinfo *info, struct ifnet *ifp, 456 struct rtentry *rt, union sockaddr_union *saun, struct ucred *cred) 457 { 458 459 /* First, see if the returned address is part of the jail. */ 460 if (prison_if(cred, rt->rt_ifa->ifa_addr) == 0) { 461 info->rti_info[RTAX_IFA] = rt->rt_ifa->ifa_addr; 462 return (0); 463 } 464 465 switch (info->rti_info[RTAX_DST]->sa_family) { 466 #ifdef INET 467 case AF_INET: 468 { 469 struct in_addr ia; 470 struct ifaddr *ifa; 471 int found; 472 473 found = 0; 474 /* 475 * Try to find an address on the given outgoing interface 476 * that belongs to the jail. 477 */ 478 IF_ADDR_RLOCK(ifp); 479 TAILQ_FOREACH(ifa, &ifp->if_addrhead, ifa_link) { 480 struct sockaddr *sa; 481 sa = ifa->ifa_addr; 482 if (sa->sa_family != AF_INET) 483 continue; 484 ia = ((struct sockaddr_in *)sa)->sin_addr; 485 if (prison_check_ip4(cred, &ia) == 0) { 486 found = 1; 487 break; 488 } 489 } 490 IF_ADDR_RUNLOCK(ifp); 491 if (!found) { 492 /* 493 * As a last resort return the 'default' jail address. 494 */ 495 ia = ((struct sockaddr_in *)rt->rt_ifa->ifa_addr)-> 496 sin_addr; 497 if (prison_get_ip4(cred, &ia) != 0) 498 return (ESRCH); 499 } 500 bzero(&saun->sin, sizeof(struct sockaddr_in)); 501 saun->sin.sin_len = sizeof(struct sockaddr_in); 502 saun->sin.sin_family = AF_INET; 503 saun->sin.sin_addr.s_addr = ia.s_addr; 504 info->rti_info[RTAX_IFA] = (struct sockaddr *)&saun->sin; 505 break; 506 } 507 #endif 508 #ifdef INET6 509 case AF_INET6: 510 { 511 struct in6_addr ia6; 512 struct ifaddr *ifa; 513 int found; 514 515 found = 0; 516 /* 517 * Try to find an address on the given outgoing interface 518 * that belongs to the jail. 519 */ 520 IF_ADDR_RLOCK(ifp); 521 TAILQ_FOREACH(ifa, &ifp->if_addrhead, ifa_link) { 522 struct sockaddr *sa; 523 sa = ifa->ifa_addr; 524 if (sa->sa_family != AF_INET6) 525 continue; 526 bcopy(&((struct sockaddr_in6 *)sa)->sin6_addr, 527 &ia6, sizeof(struct in6_addr)); 528 if (prison_check_ip6(cred, &ia6) == 0) { 529 found = 1; 530 break; 531 } 532 } 533 IF_ADDR_RUNLOCK(ifp); 534 if (!found) { 535 /* 536 * As a last resort return the 'default' jail address. 537 */ 538 ia6 = ((struct sockaddr_in6 *)rt->rt_ifa->ifa_addr)-> 539 sin6_addr; 540 if (prison_get_ip6(cred, &ia6) != 0) 541 return (ESRCH); 542 } 543 bzero(&saun->sin6, sizeof(struct sockaddr_in6)); 544 saun->sin6.sin6_len = sizeof(struct sockaddr_in6); 545 saun->sin6.sin6_family = AF_INET6; 546 bcopy(&ia6, &saun->sin6.sin6_addr, sizeof(struct in6_addr)); 547 if (sa6_recoverscope(&saun->sin6) != 0) 548 return (ESRCH); 549 info->rti_info[RTAX_IFA] = (struct sockaddr *)&saun->sin6; 550 break; 551 } 552 #endif 553 default: 554 return (ESRCH); 555 } 556 return (0); 557 } 558 559 /*ARGSUSED*/ 560 static int 561 route_output(struct mbuf *m, struct socket *so) 562 { 563 #define sa_equal(a1, a2) (bcmp((a1), (a2), (a1)->sa_len) == 0) 564 struct rt_msghdr *rtm = NULL; 565 struct rtentry *rt = NULL; 566 struct radix_node_head *rnh; 567 struct rt_addrinfo info; 568 #ifdef INET6 569 struct sockaddr_storage ss; 570 struct sockaddr_in6 *sin6; 571 int i, rti_need_deembed = 0; 572 #endif 573 int len, error = 0; 574 struct ifnet *ifp = NULL; 575 union sockaddr_union saun; 576 sa_family_t saf = AF_UNSPEC; 577 578 #define senderr(e) { error = e; goto flush;} 579 if (m == NULL || ((m->m_len < sizeof(long)) && 580 (m = m_pullup(m, sizeof(long))) == NULL)) 581 return (ENOBUFS); 582 if ((m->m_flags & M_PKTHDR) == 0) 583 panic("route_output"); 584 len = m->m_pkthdr.len; 585 if (len < sizeof(*rtm) || 586 len != mtod(m, struct rt_msghdr *)->rtm_msglen) { 587 info.rti_info[RTAX_DST] = NULL; 588 senderr(EINVAL); 589 } 590 R_Malloc(rtm, struct rt_msghdr *, len); 591 if (rtm == NULL) { 592 info.rti_info[RTAX_DST] = NULL; 593 senderr(ENOBUFS); 594 } 595 m_copydata(m, 0, len, (caddr_t)rtm); 596 if (rtm->rtm_version != RTM_VERSION) { 597 info.rti_info[RTAX_DST] = NULL; 598 senderr(EPROTONOSUPPORT); 599 } 600 rtm->rtm_pid = curproc->p_pid; 601 bzero(&info, sizeof(info)); 602 info.rti_addrs = rtm->rtm_addrs; 603 /* 604 * rt_xaddrs() performs s6_addr[2] := sin6_scope_id for AF_INET6 605 * link-local address because rtrequest requires addresses with 606 * embedded scope id. 607 */ 608 if (rt_xaddrs((caddr_t)(rtm + 1), len + (caddr_t)rtm, &info)) { 609 info.rti_info[RTAX_DST] = NULL; 610 senderr(EINVAL); 611 } 612 info.rti_flags = rtm->rtm_flags; 613 if (info.rti_info[RTAX_DST] == NULL || 614 info.rti_info[RTAX_DST]->sa_family >= AF_MAX || 615 (info.rti_info[RTAX_GATEWAY] != NULL && 616 info.rti_info[RTAX_GATEWAY]->sa_family >= AF_MAX)) 617 senderr(EINVAL); 618 saf = info.rti_info[RTAX_DST]->sa_family; 619 /* 620 * Verify that the caller has the appropriate privilege; RTM_GET 621 * is the only operation the non-superuser is allowed. 622 */ 623 if (rtm->rtm_type != RTM_GET) { 624 error = priv_check(curthread, PRIV_NET_ROUTE); 625 if (error) 626 senderr(error); 627 } 628 629 /* 630 * The given gateway address may be an interface address. 631 * For example, issuing a "route change" command on a route 632 * entry that was created from a tunnel, and the gateway 633 * address given is the local end point. In this case the 634 * RTF_GATEWAY flag must be cleared or the destination will 635 * not be reachable even though there is no error message. 636 */ 637 if (info.rti_info[RTAX_GATEWAY] != NULL && 638 info.rti_info[RTAX_GATEWAY]->sa_family != AF_LINK) { 639 struct route gw_ro; 640 641 bzero(&gw_ro, sizeof(gw_ro)); 642 gw_ro.ro_dst = *info.rti_info[RTAX_GATEWAY]; 643 rtalloc_ign_fib(&gw_ro, 0, so->so_fibnum); 644 /* 645 * A host route through the loopback interface is 646 * installed for each interface adddress. In pre 8.0 647 * releases the interface address of a PPP link type 648 * is not reachable locally. This behavior is fixed as 649 * part of the new L2/L3 redesign and rewrite work. The 650 * signature of this interface address route is the 651 * AF_LINK sa_family type of the rt_gateway, and the 652 * rt_ifp has the IFF_LOOPBACK flag set. 653 */ 654 if (gw_ro.ro_rt != NULL && 655 gw_ro.ro_rt->rt_gateway->sa_family == AF_LINK && 656 gw_ro.ro_rt->rt_ifp->if_flags & IFF_LOOPBACK) { 657 info.rti_flags &= ~RTF_GATEWAY; 658 info.rti_flags |= RTF_GWFLAG_COMPAT; 659 } 660 if (gw_ro.ro_rt != NULL) 661 RTFREE(gw_ro.ro_rt); 662 } 663 664 switch (rtm->rtm_type) { 665 struct rtentry *saved_nrt; 666 667 case RTM_ADD: 668 if (info.rti_info[RTAX_GATEWAY] == NULL) 669 senderr(EINVAL); 670 saved_nrt = NULL; 671 672 /* support for new ARP code */ 673 if (info.rti_info[RTAX_GATEWAY]->sa_family == AF_LINK && 674 (rtm->rtm_flags & RTF_LLDATA) != 0) { 675 error = lla_rt_output(rtm, &info); 676 #ifdef INET6 677 if (error == 0) 678 rti_need_deembed = (V_deembed_scopeid) ? 1 : 0; 679 #endif 680 break; 681 } 682 error = rtrequest1_fib(RTM_ADD, &info, &saved_nrt, 683 so->so_fibnum); 684 if (error == 0 && saved_nrt) { 685 #ifdef INET6 686 rti_need_deembed = (V_deembed_scopeid) ? 1 : 0; 687 #endif 688 RT_LOCK(saved_nrt); 689 rt_setmetrics(rtm->rtm_inits, 690 &rtm->rtm_rmx, &saved_nrt->rt_rmx); 691 rtm->rtm_index = saved_nrt->rt_ifp->if_index; 692 RT_REMREF(saved_nrt); 693 RT_UNLOCK(saved_nrt); 694 } 695 break; 696 697 case RTM_DELETE: 698 saved_nrt = NULL; 699 /* support for new ARP code */ 700 if (info.rti_info[RTAX_GATEWAY] && 701 (info.rti_info[RTAX_GATEWAY]->sa_family == AF_LINK) && 702 (rtm->rtm_flags & RTF_LLDATA) != 0) { 703 error = lla_rt_output(rtm, &info); 704 #ifdef INET6 705 if (error == 0) 706 rti_need_deembed = (V_deembed_scopeid) ? 1 : 0; 707 #endif 708 break; 709 } 710 error = rtrequest1_fib(RTM_DELETE, &info, &saved_nrt, 711 so->so_fibnum); 712 if (error == 0) { 713 RT_LOCK(saved_nrt); 714 rt = saved_nrt; 715 goto report; 716 } 717 #ifdef INET6 718 /* rt_msg2() will not be used when RTM_DELETE fails. */ 719 rti_need_deembed = (V_deembed_scopeid) ? 1 : 0; 720 #endif 721 break; 722 723 case RTM_GET: 724 case RTM_CHANGE: 725 case RTM_LOCK: 726 rnh = rt_tables_get_rnh(so->so_fibnum, 727 info.rti_info[RTAX_DST]->sa_family); 728 if (rnh == NULL) 729 senderr(EAFNOSUPPORT); 730 731 RADIX_NODE_HEAD_RLOCK(rnh); 732 733 if (info.rti_info[RTAX_NETMASK] == NULL && 734 rtm->rtm_type == RTM_GET) { 735 /* 736 * Provide logest prefix match for 737 * address lookup (no mask). 738 * 'route -n get addr' 739 */ 740 rt = (struct rtentry *) rnh->rnh_matchaddr( 741 info.rti_info[RTAX_DST], rnh); 742 } else 743 rt = (struct rtentry *) rnh->rnh_lookup( 744 info.rti_info[RTAX_DST], 745 info.rti_info[RTAX_NETMASK], rnh); 746 747 if (rt == NULL) { 748 RADIX_NODE_HEAD_RUNLOCK(rnh); 749 senderr(ESRCH); 750 } 751 #ifdef RADIX_MPATH 752 /* 753 * for RTM_CHANGE/LOCK, if we got multipath routes, 754 * we require users to specify a matching RTAX_GATEWAY. 755 * 756 * for RTM_GET, gate is optional even with multipath. 757 * if gate == NULL the first match is returned. 758 * (no need to call rt_mpath_matchgate if gate == NULL) 759 */ 760 if (rn_mpath_capable(rnh) && 761 (rtm->rtm_type != RTM_GET || info.rti_info[RTAX_GATEWAY])) { 762 rt = rt_mpath_matchgate(rt, info.rti_info[RTAX_GATEWAY]); 763 if (!rt) { 764 RADIX_NODE_HEAD_RUNLOCK(rnh); 765 senderr(ESRCH); 766 } 767 } 768 #endif 769 /* 770 * If performing proxied L2 entry insertion, and 771 * the actual PPP host entry is found, perform 772 * another search to retrieve the prefix route of 773 * the local end point of the PPP link. 774 */ 775 if (rtm->rtm_flags & RTF_ANNOUNCE) { 776 struct sockaddr laddr; 777 778 if (rt->rt_ifp != NULL && 779 rt->rt_ifp->if_type == IFT_PROPVIRTUAL) { 780 struct ifaddr *ifa; 781 782 ifa = ifa_ifwithnet(info.rti_info[RTAX_DST], 1); 783 if (ifa != NULL) 784 rt_maskedcopy(ifa->ifa_addr, 785 &laddr, 786 ifa->ifa_netmask); 787 } else 788 rt_maskedcopy(rt->rt_ifa->ifa_addr, 789 &laddr, 790 rt->rt_ifa->ifa_netmask); 791 /* 792 * refactor rt and no lock operation necessary 793 */ 794 rt = (struct rtentry *)rnh->rnh_matchaddr(&laddr, rnh); 795 if (rt == NULL) { 796 RADIX_NODE_HEAD_RUNLOCK(rnh); 797 senderr(ESRCH); 798 } 799 } 800 RT_LOCK(rt); 801 RT_ADDREF(rt); 802 RADIX_NODE_HEAD_RUNLOCK(rnh); 803 804 switch(rtm->rtm_type) { 805 806 case RTM_GET: 807 report: 808 RT_LOCK_ASSERT(rt); 809 if ((rt->rt_flags & RTF_HOST) == 0 810 ? jailed_without_vnet(curthread->td_ucred) 811 : prison_if(curthread->td_ucred, 812 rt_key(rt)) != 0) { 813 RT_UNLOCK(rt); 814 senderr(ESRCH); 815 } 816 info.rti_info[RTAX_DST] = rt_key(rt); 817 info.rti_info[RTAX_GATEWAY] = rt->rt_gateway; 818 info.rti_info[RTAX_NETMASK] = rt_mask(rt); 819 info.rti_info[RTAX_GENMASK] = 0; 820 if (rtm->rtm_addrs & (RTA_IFP | RTA_IFA)) { 821 ifp = rt->rt_ifp; 822 if (ifp) { 823 info.rti_info[RTAX_IFP] = 824 ifp->if_addr->ifa_addr; 825 error = rtm_get_jailed(&info, ifp, rt, 826 &saun, curthread->td_ucred); 827 if (error != 0) { 828 RT_UNLOCK(rt); 829 senderr(error); 830 } 831 if (ifp->if_flags & IFF_POINTOPOINT) 832 info.rti_info[RTAX_BRD] = 833 rt->rt_ifa->ifa_dstaddr; 834 rtm->rtm_index = ifp->if_index; 835 } else { 836 info.rti_info[RTAX_IFP] = NULL; 837 info.rti_info[RTAX_IFA] = NULL; 838 } 839 } else if ((ifp = rt->rt_ifp) != NULL) { 840 rtm->rtm_index = ifp->if_index; 841 } 842 len = rt_msg2(rtm->rtm_type, &info, NULL, NULL); 843 if (len > rtm->rtm_msglen) { 844 struct rt_msghdr *new_rtm; 845 R_Malloc(new_rtm, struct rt_msghdr *, len); 846 if (new_rtm == NULL) { 847 RT_UNLOCK(rt); 848 senderr(ENOBUFS); 849 } 850 bcopy(rtm, new_rtm, rtm->rtm_msglen); 851 Free(rtm); rtm = new_rtm; 852 } 853 (void)rt_msg2(rtm->rtm_type, &info, (caddr_t)rtm, NULL); 854 if (rt->rt_flags & RTF_GWFLAG_COMPAT) 855 rtm->rtm_flags = RTF_GATEWAY | 856 (rt->rt_flags & ~RTF_GWFLAG_COMPAT); 857 else 858 rtm->rtm_flags = rt->rt_flags; 859 rt_getmetrics(&rt->rt_rmx, &rtm->rtm_rmx); 860 rtm->rtm_addrs = info.rti_addrs; 861 break; 862 863 case RTM_CHANGE: 864 /* 865 * New gateway could require new ifaddr, ifp; 866 * flags may also be different; ifp may be specified 867 * by ll sockaddr when protocol address is ambiguous 868 */ 869 if (((rt->rt_flags & RTF_GATEWAY) && 870 info.rti_info[RTAX_GATEWAY] != NULL) || 871 info.rti_info[RTAX_IFP] != NULL || 872 (info.rti_info[RTAX_IFA] != NULL && 873 !sa_equal(info.rti_info[RTAX_IFA], 874 rt->rt_ifa->ifa_addr))) { 875 RT_UNLOCK(rt); 876 RADIX_NODE_HEAD_LOCK(rnh); 877 error = rt_getifa_fib(&info, rt->rt_fibnum); 878 /* 879 * XXXRW: Really we should release this 880 * reference later, but this maintains 881 * historical behavior. 882 */ 883 if (info.rti_ifa != NULL) 884 ifa_free(info.rti_ifa); 885 RADIX_NODE_HEAD_UNLOCK(rnh); 886 if (error != 0) 887 senderr(error); 888 RT_LOCK(rt); 889 } 890 if (info.rti_ifa != NULL && 891 info.rti_ifa != rt->rt_ifa && 892 rt->rt_ifa != NULL && 893 rt->rt_ifa->ifa_rtrequest != NULL) { 894 rt->rt_ifa->ifa_rtrequest(RTM_DELETE, rt, 895 &info); 896 ifa_free(rt->rt_ifa); 897 } 898 if (info.rti_info[RTAX_GATEWAY] != NULL) { 899 RT_UNLOCK(rt); 900 RADIX_NODE_HEAD_LOCK(rnh); 901 RT_LOCK(rt); 902 903 error = rt_setgate(rt, rt_key(rt), 904 info.rti_info[RTAX_GATEWAY]); 905 RADIX_NODE_HEAD_UNLOCK(rnh); 906 if (error != 0) { 907 RT_UNLOCK(rt); 908 senderr(error); 909 } 910 rt->rt_flags &= ~RTF_GATEWAY; 911 rt->rt_flags |= (RTF_GATEWAY & info.rti_flags); 912 } 913 if (info.rti_ifa != NULL && 914 info.rti_ifa != rt->rt_ifa) { 915 ifa_ref(info.rti_ifa); 916 rt->rt_ifa = info.rti_ifa; 917 rt->rt_ifp = info.rti_ifp; 918 } 919 /* Allow some flags to be toggled on change. */ 920 rt->rt_flags = (rt->rt_flags & ~RTF_FMASK) | 921 (rtm->rtm_flags & RTF_FMASK); 922 rt_setmetrics(rtm->rtm_inits, &rtm->rtm_rmx, 923 &rt->rt_rmx); 924 rtm->rtm_index = rt->rt_ifp->if_index; 925 if (rt->rt_ifa && rt->rt_ifa->ifa_rtrequest) 926 rt->rt_ifa->ifa_rtrequest(RTM_ADD, rt, &info); 927 /* FALLTHROUGH */ 928 case RTM_LOCK: 929 /* We don't support locks anymore */ 930 break; 931 } 932 RT_UNLOCK(rt); 933 break; 934 935 default: 936 senderr(EOPNOTSUPP); 937 } 938 939 flush: 940 if (rtm) { 941 if (error) 942 rtm->rtm_errno = error; 943 else 944 rtm->rtm_flags |= RTF_DONE; 945 } 946 if (rt) /* XXX can this be true? */ 947 RTFREE(rt); 948 { 949 struct rawcb *rp = NULL; 950 /* 951 * Check to see if we don't want our own messages. 952 */ 953 if ((so->so_options & SO_USELOOPBACK) == 0) { 954 if (route_cb.any_count <= 1) { 955 if (rtm) 956 Free(rtm); 957 m_freem(m); 958 return (error); 959 } 960 /* There is another listener, so construct message */ 961 rp = sotorawcb(so); 962 } 963 if (rtm) { 964 #ifdef INET6 965 if (rti_need_deembed) { 966 /* sin6_scope_id is recovered before sending rtm. */ 967 sin6 = (struct sockaddr_in6 *)&ss; 968 for (i = 0; i < RTAX_MAX; i++) { 969 if (info.rti_info[i] == NULL) 970 continue; 971 if (info.rti_info[i]->sa_family != AF_INET6) 972 continue; 973 bcopy(info.rti_info[i], sin6, sizeof(*sin6)); 974 if (sa6_recoverscope(sin6) == 0) 975 bcopy(sin6, info.rti_info[i], 976 sizeof(*sin6)); 977 } 978 } 979 #endif 980 m_copyback(m, 0, rtm->rtm_msglen, (caddr_t)rtm); 981 if (m->m_pkthdr.len < rtm->rtm_msglen) { 982 m_freem(m); 983 m = NULL; 984 } else if (m->m_pkthdr.len > rtm->rtm_msglen) 985 m_adj(m, rtm->rtm_msglen - m->m_pkthdr.len); 986 } 987 if (m) { 988 M_SETFIB(m, so->so_fibnum); 989 m->m_flags |= RTS_FILTER_FIB; 990 if (rp) { 991 /* 992 * XXX insure we don't get a copy by 993 * invalidating our protocol 994 */ 995 unsigned short family = rp->rcb_proto.sp_family; 996 rp->rcb_proto.sp_family = 0; 997 rt_dispatch(m, saf); 998 rp->rcb_proto.sp_family = family; 999 } else 1000 rt_dispatch(m, saf); 1001 } 1002 /* info.rti_info[RTAX_DST] (used above) can point inside of rtm */ 1003 if (rtm) 1004 Free(rtm); 1005 } 1006 return (error); 1007 #undef sa_equal 1008 } 1009 1010 static void 1011 rt_setmetrics(u_long which, const struct rt_metrics *in, 1012 struct rt_metrics_lite *out) 1013 { 1014 #define metric(f, e) if (which & (f)) out->e = in->e; 1015 /* 1016 * Only these are stored in the routing entry since introduction 1017 * of tcp hostcache. The rest is ignored. 1018 */ 1019 metric(RTV_MTU, rmx_mtu); 1020 metric(RTV_WEIGHT, rmx_weight); 1021 /* Userland -> kernel timebase conversion. */ 1022 if (which & RTV_EXPIRE) 1023 out->rmx_expire = in->rmx_expire ? 1024 in->rmx_expire - time_second + time_uptime : 0; 1025 #undef metric 1026 } 1027 1028 static void 1029 rt_getmetrics(const struct rt_metrics_lite *in, struct rt_metrics *out) 1030 { 1031 #define metric(e) out->e = in->e; 1032 bzero(out, sizeof(*out)); 1033 metric(rmx_mtu); 1034 metric(rmx_weight); 1035 /* Kernel -> userland timebase conversion. */ 1036 out->rmx_expire = in->rmx_expire ? 1037 in->rmx_expire - time_uptime + time_second : 0; 1038 #undef metric 1039 } 1040 1041 /* 1042 * Extract the addresses of the passed sockaddrs. 1043 * Do a little sanity checking so as to avoid bad memory references. 1044 * This data is derived straight from userland. 1045 */ 1046 static int 1047 rt_xaddrs(caddr_t cp, caddr_t cplim, struct rt_addrinfo *rtinfo) 1048 { 1049 struct sockaddr *sa; 1050 int i; 1051 1052 for (i = 0; i < RTAX_MAX && cp < cplim; i++) { 1053 if ((rtinfo->rti_addrs & (1 << i)) == 0) 1054 continue; 1055 sa = (struct sockaddr *)cp; 1056 /* 1057 * It won't fit. 1058 */ 1059 if (cp + sa->sa_len > cplim) 1060 return (EINVAL); 1061 /* 1062 * there are no more.. quit now 1063 * If there are more bits, they are in error. 1064 * I've seen this. route(1) can evidently generate these. 1065 * This causes kernel to core dump. 1066 * for compatibility, If we see this, point to a safe address. 1067 */ 1068 if (sa->sa_len == 0) { 1069 rtinfo->rti_info[i] = &sa_zero; 1070 return (0); /* should be EINVAL but for compat */ 1071 } 1072 /* accept it */ 1073 #ifdef INET6 1074 if (sa->sa_family == AF_INET6) 1075 sa6_embedscope((struct sockaddr_in6 *)sa, 1076 V_ip6_use_defzone); 1077 #endif 1078 rtinfo->rti_info[i] = sa; 1079 cp += SA_SIZE(sa); 1080 } 1081 return (0); 1082 } 1083 1084 /* 1085 * Used by the routing socket. 1086 */ 1087 static struct mbuf * 1088 rt_msg1(int type, struct rt_addrinfo *rtinfo) 1089 { 1090 struct rt_msghdr *rtm; 1091 struct mbuf *m; 1092 int i; 1093 struct sockaddr *sa; 1094 #ifdef INET6 1095 struct sockaddr_storage ss; 1096 struct sockaddr_in6 *sin6; 1097 #endif 1098 int len, dlen; 1099 1100 switch (type) { 1101 1102 case RTM_DELADDR: 1103 case RTM_NEWADDR: 1104 len = sizeof(struct ifa_msghdr); 1105 break; 1106 1107 case RTM_DELMADDR: 1108 case RTM_NEWMADDR: 1109 len = sizeof(struct ifma_msghdr); 1110 break; 1111 1112 case RTM_IFINFO: 1113 len = sizeof(struct if_msghdr); 1114 break; 1115 1116 case RTM_IFANNOUNCE: 1117 case RTM_IEEE80211: 1118 len = sizeof(struct if_announcemsghdr); 1119 break; 1120 1121 default: 1122 len = sizeof(struct rt_msghdr); 1123 } 1124 1125 /* XXXGL: can we use MJUMPAGESIZE cluster here? */ 1126 KASSERT(len <= MCLBYTES, ("%s: message too big", __func__)); 1127 if (len > MHLEN) 1128 m = m_getcl(M_NOWAIT, MT_DATA, M_PKTHDR); 1129 else 1130 m = m_gethdr(M_NOWAIT, MT_DATA); 1131 if (m == NULL) 1132 return (m); 1133 1134 m->m_pkthdr.len = m->m_len = len; 1135 rtm = mtod(m, struct rt_msghdr *); 1136 bzero((caddr_t)rtm, len); 1137 for (i = 0; i < RTAX_MAX; i++) { 1138 if ((sa = rtinfo->rti_info[i]) == NULL) 1139 continue; 1140 rtinfo->rti_addrs |= (1 << i); 1141 dlen = SA_SIZE(sa); 1142 #ifdef INET6 1143 if (V_deembed_scopeid && sa->sa_family == AF_INET6) { 1144 sin6 = (struct sockaddr_in6 *)&ss; 1145 bcopy(sa, sin6, sizeof(*sin6)); 1146 if (sa6_recoverscope(sin6) == 0) 1147 sa = (struct sockaddr *)sin6; 1148 } 1149 #endif 1150 m_copyback(m, len, dlen, (caddr_t)sa); 1151 len += dlen; 1152 } 1153 if (m->m_pkthdr.len != len) { 1154 m_freem(m); 1155 return (NULL); 1156 } 1157 rtm->rtm_msglen = len; 1158 rtm->rtm_version = RTM_VERSION; 1159 rtm->rtm_type = type; 1160 return (m); 1161 } 1162 1163 /* 1164 * Used by the sysctl code and routing socket. 1165 */ 1166 static int 1167 rt_msg2(int type, struct rt_addrinfo *rtinfo, caddr_t cp, struct walkarg *w) 1168 { 1169 int i; 1170 int len, dlen, second_time = 0; 1171 caddr_t cp0; 1172 #ifdef INET6 1173 struct sockaddr_storage ss; 1174 struct sockaddr_in6 *sin6; 1175 #endif 1176 1177 rtinfo->rti_addrs = 0; 1178 again: 1179 switch (type) { 1180 1181 case RTM_DELADDR: 1182 case RTM_NEWADDR: 1183 if (w != NULL && w->w_op == NET_RT_IFLISTL) { 1184 #ifdef COMPAT_FREEBSD32 1185 if (w->w_req->flags & SCTL_MASK32) 1186 len = sizeof(struct ifa_msghdrl32); 1187 else 1188 #endif 1189 len = sizeof(struct ifa_msghdrl); 1190 } else 1191 len = sizeof(struct ifa_msghdr); 1192 break; 1193 1194 case RTM_IFINFO: 1195 #ifdef COMPAT_FREEBSD32 1196 if (w != NULL && w->w_req->flags & SCTL_MASK32) { 1197 if (w->w_op == NET_RT_IFLISTL) 1198 len = sizeof(struct if_msghdrl32); 1199 else 1200 len = sizeof(struct if_msghdr32); 1201 break; 1202 } 1203 #endif 1204 if (w != NULL && w->w_op == NET_RT_IFLISTL) 1205 len = sizeof(struct if_msghdrl); 1206 else 1207 len = sizeof(struct if_msghdr); 1208 break; 1209 1210 case RTM_NEWMADDR: 1211 len = sizeof(struct ifma_msghdr); 1212 break; 1213 1214 default: 1215 len = sizeof(struct rt_msghdr); 1216 } 1217 cp0 = cp; 1218 if (cp0) 1219 cp += len; 1220 for (i = 0; i < RTAX_MAX; i++) { 1221 struct sockaddr *sa; 1222 1223 if ((sa = rtinfo->rti_info[i]) == NULL) 1224 continue; 1225 rtinfo->rti_addrs |= (1 << i); 1226 dlen = SA_SIZE(sa); 1227 if (cp) { 1228 #ifdef INET6 1229 if (V_deembed_scopeid && sa->sa_family == AF_INET6) { 1230 sin6 = (struct sockaddr_in6 *)&ss; 1231 bcopy(sa, sin6, sizeof(*sin6)); 1232 if (sa6_recoverscope(sin6) == 0) 1233 sa = (struct sockaddr *)sin6; 1234 } 1235 #endif 1236 bcopy((caddr_t)sa, cp, (unsigned)dlen); 1237 cp += dlen; 1238 } 1239 len += dlen; 1240 } 1241 len = ALIGN(len); 1242 if (cp == NULL && w != NULL && !second_time) { 1243 struct walkarg *rw = w; 1244 1245 if (rw->w_req) { 1246 if (rw->w_tmemsize < len) { 1247 if (rw->w_tmem) 1248 free(rw->w_tmem, M_RTABLE); 1249 rw->w_tmem = (caddr_t) 1250 malloc(len, M_RTABLE, M_NOWAIT); 1251 if (rw->w_tmem) 1252 rw->w_tmemsize = len; 1253 } 1254 if (rw->w_tmem) { 1255 cp = rw->w_tmem; 1256 second_time = 1; 1257 goto again; 1258 } 1259 } 1260 } 1261 if (cp) { 1262 struct rt_msghdr *rtm = (struct rt_msghdr *)cp0; 1263 1264 rtm->rtm_version = RTM_VERSION; 1265 rtm->rtm_type = type; 1266 rtm->rtm_msglen = len; 1267 } 1268 return (len); 1269 } 1270 1271 /* 1272 * This routine is called to generate a message from the routing 1273 * socket indicating that a redirect has occured, a routing lookup 1274 * has failed, or that a protocol has detected timeouts to a particular 1275 * destination. 1276 */ 1277 void 1278 rt_missmsg_fib(int type, struct rt_addrinfo *rtinfo, int flags, int error, 1279 int fibnum) 1280 { 1281 struct rt_msghdr *rtm; 1282 struct mbuf *m; 1283 struct sockaddr *sa = rtinfo->rti_info[RTAX_DST]; 1284 1285 if (route_cb.any_count == 0) 1286 return; 1287 m = rt_msg1(type, rtinfo); 1288 if (m == NULL) 1289 return; 1290 1291 if (fibnum != RTS_ALLFIBS) { 1292 KASSERT(fibnum >= 0 && fibnum < rt_numfibs, ("%s: fibnum out " 1293 "of range 0 <= %d < %d", __func__, fibnum, rt_numfibs)); 1294 M_SETFIB(m, fibnum); 1295 m->m_flags |= RTS_FILTER_FIB; 1296 } 1297 1298 rtm = mtod(m, struct rt_msghdr *); 1299 rtm->rtm_flags = RTF_DONE | flags; 1300 rtm->rtm_errno = error; 1301 rtm->rtm_addrs = rtinfo->rti_addrs; 1302 rt_dispatch(m, sa ? sa->sa_family : AF_UNSPEC); 1303 } 1304 1305 void 1306 rt_missmsg(int type, struct rt_addrinfo *rtinfo, int flags, int error) 1307 { 1308 1309 rt_missmsg_fib(type, rtinfo, flags, error, RTS_ALLFIBS); 1310 } 1311 1312 /* 1313 * This routine is called to generate a message from the routing 1314 * socket indicating that the status of a network interface has changed. 1315 */ 1316 void 1317 rt_ifmsg(struct ifnet *ifp) 1318 { 1319 struct if_msghdr *ifm; 1320 struct mbuf *m; 1321 struct rt_addrinfo info; 1322 1323 if (route_cb.any_count == 0) 1324 return; 1325 bzero((caddr_t)&info, sizeof(info)); 1326 m = rt_msg1(RTM_IFINFO, &info); 1327 if (m == NULL) 1328 return; 1329 ifm = mtod(m, struct if_msghdr *); 1330 ifm->ifm_index = ifp->if_index; 1331 ifm->ifm_flags = ifp->if_flags | ifp->if_drv_flags; 1332 ifm->ifm_data = ifp->if_data; 1333 ifm->ifm_addrs = 0; 1334 rt_dispatch(m, AF_UNSPEC); 1335 } 1336 1337 /* 1338 * This is called to generate messages from the routing socket 1339 * indicating a network interface has had addresses associated with it. 1340 * if we ever reverse the logic and replace messages TO the routing 1341 * socket indicate a request to configure interfaces, then it will 1342 * be unnecessary as the routing socket will automatically generate 1343 * copies of it. 1344 */ 1345 void 1346 rt_newaddrmsg_fib(int cmd, struct ifaddr *ifa, int error, struct rtentry *rt, 1347 int fibnum) 1348 { 1349 struct rt_addrinfo info; 1350 struct sockaddr *sa = NULL; 1351 int pass; 1352 struct mbuf *m = NULL; 1353 struct ifnet *ifp = ifa->ifa_ifp; 1354 1355 KASSERT(cmd == RTM_ADD || cmd == RTM_DELETE, 1356 ("unexpected cmd %u", cmd)); 1357 #if defined(INET) || defined(INET6) 1358 #ifdef SCTP 1359 /* 1360 * notify the SCTP stack 1361 * this will only get called when an address is added/deleted 1362 * XXX pass the ifaddr struct instead if ifa->ifa_addr... 1363 */ 1364 sctp_addr_change(ifa, cmd); 1365 #endif /* SCTP */ 1366 #endif 1367 if (route_cb.any_count == 0) 1368 return; 1369 for (pass = 1; pass < 3; pass++) { 1370 bzero((caddr_t)&info, sizeof(info)); 1371 if ((cmd == RTM_ADD && pass == 1) || 1372 (cmd == RTM_DELETE && pass == 2)) { 1373 struct ifa_msghdr *ifam; 1374 int ncmd = cmd == RTM_ADD ? RTM_NEWADDR : RTM_DELADDR; 1375 1376 info.rti_info[RTAX_IFA] = sa = ifa->ifa_addr; 1377 info.rti_info[RTAX_IFP] = ifp->if_addr->ifa_addr; 1378 info.rti_info[RTAX_NETMASK] = ifa->ifa_netmask; 1379 info.rti_info[RTAX_BRD] = ifa->ifa_dstaddr; 1380 if ((m = rt_msg1(ncmd, &info)) == NULL) 1381 continue; 1382 ifam = mtod(m, struct ifa_msghdr *); 1383 ifam->ifam_index = ifp->if_index; 1384 ifam->ifam_metric = ifa->ifa_metric; 1385 ifam->ifam_flags = ifa->ifa_flags; 1386 ifam->ifam_addrs = info.rti_addrs; 1387 } 1388 if ((cmd == RTM_ADD && pass == 2) || 1389 (cmd == RTM_DELETE && pass == 1)) { 1390 struct rt_msghdr *rtm; 1391 1392 if (rt == NULL) 1393 continue; 1394 info.rti_info[RTAX_NETMASK] = rt_mask(rt); 1395 info.rti_info[RTAX_DST] = sa = rt_key(rt); 1396 info.rti_info[RTAX_GATEWAY] = rt->rt_gateway; 1397 if ((m = rt_msg1(cmd, &info)) == NULL) 1398 continue; 1399 rtm = mtod(m, struct rt_msghdr *); 1400 rtm->rtm_index = ifp->if_index; 1401 rtm->rtm_flags |= rt->rt_flags; 1402 rtm->rtm_errno = error; 1403 rtm->rtm_addrs = info.rti_addrs; 1404 } 1405 if (fibnum != RTS_ALLFIBS) { 1406 KASSERT(fibnum >= 0 && fibnum < rt_numfibs, ("%s: " 1407 "fibnum out of range 0 <= %d < %d", __func__, 1408 fibnum, rt_numfibs)); 1409 M_SETFIB(m, fibnum); 1410 m->m_flags |= RTS_FILTER_FIB; 1411 } 1412 rt_dispatch(m, sa ? sa->sa_family : AF_UNSPEC); 1413 } 1414 } 1415 1416 void 1417 rt_newaddrmsg(int cmd, struct ifaddr *ifa, int error, struct rtentry *rt) 1418 { 1419 1420 rt_newaddrmsg_fib(cmd, ifa, error, rt, RTS_ALLFIBS); 1421 } 1422 1423 /* 1424 * This is the analogue to the rt_newaddrmsg which performs the same 1425 * function but for multicast group memberhips. This is easier since 1426 * there is no route state to worry about. 1427 */ 1428 void 1429 rt_newmaddrmsg(int cmd, struct ifmultiaddr *ifma) 1430 { 1431 struct rt_addrinfo info; 1432 struct mbuf *m = NULL; 1433 struct ifnet *ifp = ifma->ifma_ifp; 1434 struct ifma_msghdr *ifmam; 1435 1436 if (route_cb.any_count == 0) 1437 return; 1438 1439 bzero((caddr_t)&info, sizeof(info)); 1440 info.rti_info[RTAX_IFA] = ifma->ifma_addr; 1441 info.rti_info[RTAX_IFP] = ifp ? ifp->if_addr->ifa_addr : NULL; 1442 /* 1443 * If a link-layer address is present, present it as a ``gateway'' 1444 * (similarly to how ARP entries, e.g., are presented). 1445 */ 1446 info.rti_info[RTAX_GATEWAY] = ifma->ifma_lladdr; 1447 m = rt_msg1(cmd, &info); 1448 if (m == NULL) 1449 return; 1450 ifmam = mtod(m, struct ifma_msghdr *); 1451 KASSERT(ifp != NULL, ("%s: link-layer multicast address w/o ifp\n", 1452 __func__)); 1453 ifmam->ifmam_index = ifp->if_index; 1454 ifmam->ifmam_addrs = info.rti_addrs; 1455 rt_dispatch(m, ifma->ifma_addr ? ifma->ifma_addr->sa_family : AF_UNSPEC); 1456 } 1457 1458 static struct mbuf * 1459 rt_makeifannouncemsg(struct ifnet *ifp, int type, int what, 1460 struct rt_addrinfo *info) 1461 { 1462 struct if_announcemsghdr *ifan; 1463 struct mbuf *m; 1464 1465 if (route_cb.any_count == 0) 1466 return NULL; 1467 bzero((caddr_t)info, sizeof(*info)); 1468 m = rt_msg1(type, info); 1469 if (m != NULL) { 1470 ifan = mtod(m, struct if_announcemsghdr *); 1471 ifan->ifan_index = ifp->if_index; 1472 strlcpy(ifan->ifan_name, ifp->if_xname, 1473 sizeof(ifan->ifan_name)); 1474 ifan->ifan_what = what; 1475 } 1476 return m; 1477 } 1478 1479 /* 1480 * This is called to generate routing socket messages indicating 1481 * IEEE80211 wireless events. 1482 * XXX we piggyback on the RTM_IFANNOUNCE msg format in a clumsy way. 1483 */ 1484 void 1485 rt_ieee80211msg(struct ifnet *ifp, int what, void *data, size_t data_len) 1486 { 1487 struct mbuf *m; 1488 struct rt_addrinfo info; 1489 1490 m = rt_makeifannouncemsg(ifp, RTM_IEEE80211, what, &info); 1491 if (m != NULL) { 1492 /* 1493 * Append the ieee80211 data. Try to stick it in the 1494 * mbuf containing the ifannounce msg; otherwise allocate 1495 * a new mbuf and append. 1496 * 1497 * NB: we assume m is a single mbuf. 1498 */ 1499 if (data_len > M_TRAILINGSPACE(m)) { 1500 struct mbuf *n = m_get(M_NOWAIT, MT_DATA); 1501 if (n == NULL) { 1502 m_freem(m); 1503 return; 1504 } 1505 bcopy(data, mtod(n, void *), data_len); 1506 n->m_len = data_len; 1507 m->m_next = n; 1508 } else if (data_len > 0) { 1509 bcopy(data, mtod(m, u_int8_t *) + m->m_len, data_len); 1510 m->m_len += data_len; 1511 } 1512 if (m->m_flags & M_PKTHDR) 1513 m->m_pkthdr.len += data_len; 1514 mtod(m, struct if_announcemsghdr *)->ifan_msglen += data_len; 1515 rt_dispatch(m, AF_UNSPEC); 1516 } 1517 } 1518 1519 /* 1520 * This is called to generate routing socket messages indicating 1521 * network interface arrival and departure. 1522 */ 1523 void 1524 rt_ifannouncemsg(struct ifnet *ifp, int what) 1525 { 1526 struct mbuf *m; 1527 struct rt_addrinfo info; 1528 1529 m = rt_makeifannouncemsg(ifp, RTM_IFANNOUNCE, what, &info); 1530 if (m != NULL) 1531 rt_dispatch(m, AF_UNSPEC); 1532 } 1533 1534 static void 1535 rt_dispatch(struct mbuf *m, sa_family_t saf) 1536 { 1537 struct m_tag *tag; 1538 1539 /* 1540 * Preserve the family from the sockaddr, if any, in an m_tag for 1541 * use when injecting the mbuf into the routing socket buffer from 1542 * the netisr. 1543 */ 1544 if (saf != AF_UNSPEC) { 1545 tag = m_tag_get(PACKET_TAG_RTSOCKFAM, sizeof(unsigned short), 1546 M_NOWAIT); 1547 if (tag == NULL) { 1548 m_freem(m); 1549 return; 1550 } 1551 *(unsigned short *)(tag + 1) = saf; 1552 m_tag_prepend(m, tag); 1553 } 1554 #ifdef VIMAGE 1555 if (V_loif) 1556 m->m_pkthdr.rcvif = V_loif; 1557 else { 1558 m_freem(m); 1559 return; 1560 } 1561 #endif 1562 netisr_queue(NETISR_ROUTE, m); /* mbuf is free'd on failure. */ 1563 } 1564 1565 /* 1566 * This is used in dumping the kernel table via sysctl(). 1567 */ 1568 static int 1569 sysctl_dumpentry(struct radix_node *rn, void *vw) 1570 { 1571 struct walkarg *w = vw; 1572 struct rtentry *rt = (struct rtentry *)rn; 1573 int error = 0, size; 1574 struct rt_addrinfo info; 1575 1576 if (w->w_op == NET_RT_FLAGS && !(rt->rt_flags & w->w_arg)) 1577 return 0; 1578 if ((rt->rt_flags & RTF_HOST) == 0 1579 ? jailed_without_vnet(w->w_req->td->td_ucred) 1580 : prison_if(w->w_req->td->td_ucred, rt_key(rt)) != 0) 1581 return (0); 1582 bzero((caddr_t)&info, sizeof(info)); 1583 info.rti_info[RTAX_DST] = rt_key(rt); 1584 info.rti_info[RTAX_GATEWAY] = rt->rt_gateway; 1585 info.rti_info[RTAX_NETMASK] = rt_mask(rt); 1586 info.rti_info[RTAX_GENMASK] = 0; 1587 if (rt->rt_ifp) { 1588 info.rti_info[RTAX_IFP] = rt->rt_ifp->if_addr->ifa_addr; 1589 info.rti_info[RTAX_IFA] = rt->rt_ifa->ifa_addr; 1590 if (rt->rt_ifp->if_flags & IFF_POINTOPOINT) 1591 info.rti_info[RTAX_BRD] = rt->rt_ifa->ifa_dstaddr; 1592 } 1593 size = rt_msg2(RTM_GET, &info, NULL, w); 1594 if (w->w_req && w->w_tmem) { 1595 struct rt_msghdr *rtm = (struct rt_msghdr *)w->w_tmem; 1596 1597 if (rt->rt_flags & RTF_GWFLAG_COMPAT) 1598 rtm->rtm_flags = RTF_GATEWAY | 1599 (rt->rt_flags & ~RTF_GWFLAG_COMPAT); 1600 else 1601 rtm->rtm_flags = rt->rt_flags; 1602 /* 1603 * let's be honest about this being a retarded hack 1604 */ 1605 rtm->rtm_fmask = rt->rt_rmx.rmx_pksent; 1606 rt_getmetrics(&rt->rt_rmx, &rtm->rtm_rmx); 1607 rtm->rtm_index = rt->rt_ifp->if_index; 1608 rtm->rtm_errno = rtm->rtm_pid = rtm->rtm_seq = 0; 1609 rtm->rtm_addrs = info.rti_addrs; 1610 error = SYSCTL_OUT(w->w_req, (caddr_t)rtm, size); 1611 return (error); 1612 } 1613 return (error); 1614 } 1615 1616 #ifdef COMPAT_FREEBSD32 1617 static void 1618 copy_ifdata32(struct if_data *src, struct if_data32 *dst) 1619 { 1620 1621 bzero(dst, sizeof(*dst)); 1622 CP(*src, *dst, ifi_type); 1623 CP(*src, *dst, ifi_physical); 1624 CP(*src, *dst, ifi_addrlen); 1625 CP(*src, *dst, ifi_hdrlen); 1626 CP(*src, *dst, ifi_link_state); 1627 CP(*src, *dst, ifi_vhid); 1628 CP(*src, *dst, ifi_baudrate_pf); 1629 dst->ifi_datalen = sizeof(struct if_data32); 1630 CP(*src, *dst, ifi_mtu); 1631 CP(*src, *dst, ifi_metric); 1632 CP(*src, *dst, ifi_baudrate); 1633 CP(*src, *dst, ifi_ipackets); 1634 CP(*src, *dst, ifi_ierrors); 1635 CP(*src, *dst, ifi_opackets); 1636 CP(*src, *dst, ifi_oerrors); 1637 CP(*src, *dst, ifi_collisions); 1638 CP(*src, *dst, ifi_ibytes); 1639 CP(*src, *dst, ifi_obytes); 1640 CP(*src, *dst, ifi_imcasts); 1641 CP(*src, *dst, ifi_omcasts); 1642 CP(*src, *dst, ifi_iqdrops); 1643 CP(*src, *dst, ifi_noproto); 1644 CP(*src, *dst, ifi_hwassist); 1645 CP(*src, *dst, ifi_epoch); 1646 TV_CP(*src, *dst, ifi_lastchange); 1647 } 1648 #endif 1649 1650 static int 1651 sysctl_iflist_ifml(struct ifnet *ifp, struct rt_addrinfo *info, 1652 struct walkarg *w, int len) 1653 { 1654 struct if_msghdrl *ifm; 1655 1656 #ifdef COMPAT_FREEBSD32 1657 if (w->w_req->flags & SCTL_MASK32) { 1658 struct if_msghdrl32 *ifm32; 1659 1660 ifm32 = (struct if_msghdrl32 *)w->w_tmem; 1661 ifm32->ifm_addrs = info->rti_addrs; 1662 ifm32->ifm_flags = ifp->if_flags | ifp->if_drv_flags; 1663 ifm32->ifm_index = ifp->if_index; 1664 ifm32->_ifm_spare1 = 0; 1665 ifm32->ifm_len = sizeof(*ifm32); 1666 ifm32->ifm_data_off = offsetof(struct if_msghdrl32, ifm_data); 1667 1668 copy_ifdata32(&ifp->if_data, &ifm32->ifm_data); 1669 /* Fixup if_data carp(4) vhid. */ 1670 if (carp_get_vhid_p != NULL) 1671 ifm32->ifm_data.ifi_vhid = 1672 (*carp_get_vhid_p)(ifp->if_addr); 1673 1674 return (SYSCTL_OUT(w->w_req, (caddr_t)ifm32, len)); 1675 } 1676 #endif 1677 ifm = (struct if_msghdrl *)w->w_tmem; 1678 ifm->ifm_addrs = info->rti_addrs; 1679 ifm->ifm_flags = ifp->if_flags | ifp->if_drv_flags; 1680 ifm->ifm_index = ifp->if_index; 1681 ifm->_ifm_spare1 = 0; 1682 ifm->ifm_len = sizeof(*ifm); 1683 ifm->ifm_data_off = offsetof(struct if_msghdrl, ifm_data); 1684 1685 ifm->ifm_data = ifp->if_data; 1686 /* Fixup if_data carp(4) vhid. */ 1687 if (carp_get_vhid_p != NULL) 1688 ifm->ifm_data.ifi_vhid = (*carp_get_vhid_p)(ifp->if_addr); 1689 1690 return (SYSCTL_OUT(w->w_req, (caddr_t)ifm, len)); 1691 } 1692 1693 static int 1694 sysctl_iflist_ifm(struct ifnet *ifp, struct rt_addrinfo *info, 1695 struct walkarg *w, int len) 1696 { 1697 struct if_msghdr *ifm; 1698 1699 #ifdef COMPAT_FREEBSD32 1700 if (w->w_req->flags & SCTL_MASK32) { 1701 struct if_msghdr32 *ifm32; 1702 1703 ifm32 = (struct if_msghdr32 *)w->w_tmem; 1704 ifm32->ifm_addrs = info->rti_addrs; 1705 ifm32->ifm_flags = ifp->if_flags | ifp->if_drv_flags; 1706 ifm32->ifm_index = ifp->if_index; 1707 1708 copy_ifdata32(&ifp->if_data, &ifm32->ifm_data); 1709 /* Fixup if_data carp(4) vhid. */ 1710 if (carp_get_vhid_p != NULL) 1711 ifm32->ifm_data.ifi_vhid = 1712 (*carp_get_vhid_p)(ifp->if_addr); 1713 1714 return (SYSCTL_OUT(w->w_req, (caddr_t)ifm32, len)); 1715 } 1716 #endif 1717 ifm = (struct if_msghdr *)w->w_tmem; 1718 ifm->ifm_addrs = info->rti_addrs; 1719 ifm->ifm_flags = ifp->if_flags | ifp->if_drv_flags; 1720 ifm->ifm_index = ifp->if_index; 1721 1722 ifm->ifm_data = ifp->if_data; 1723 /* Fixup if_data carp(4) vhid. */ 1724 if (carp_get_vhid_p != NULL) 1725 ifm->ifm_data.ifi_vhid = (*carp_get_vhid_p)(ifp->if_addr); 1726 1727 return (SYSCTL_OUT(w->w_req, (caddr_t)ifm, len)); 1728 } 1729 1730 static int 1731 sysctl_iflist_ifaml(struct ifaddr *ifa, struct rt_addrinfo *info, 1732 struct walkarg *w, int len) 1733 { 1734 struct ifa_msghdrl *ifam; 1735 1736 #ifdef COMPAT_FREEBSD32 1737 if (w->w_req->flags & SCTL_MASK32) { 1738 struct ifa_msghdrl32 *ifam32; 1739 1740 ifam32 = (struct ifa_msghdrl32 *)w->w_tmem; 1741 ifam32->ifam_addrs = info->rti_addrs; 1742 ifam32->ifam_flags = ifa->ifa_flags; 1743 ifam32->ifam_index = ifa->ifa_ifp->if_index; 1744 ifam32->_ifam_spare1 = 0; 1745 ifam32->ifam_len = sizeof(*ifam32); 1746 ifam32->ifam_data_off = 1747 offsetof(struct ifa_msghdrl32, ifam_data); 1748 ifam32->ifam_metric = ifa->ifa_metric; 1749 1750 bzero(&ifam32->ifam_data, sizeof(ifam32->ifam_data)); 1751 ifam32->ifam_data.ifi_datalen = sizeof(struct if_data32); 1752 ifam32->ifam_data.ifi_ipackets = 1753 counter_u64_fetch(ifa->ifa_ipackets); 1754 ifam32->ifam_data.ifi_opackets = 1755 counter_u64_fetch(ifa->ifa_opackets); 1756 ifam32->ifam_data.ifi_ibytes = 1757 counter_u64_fetch(ifa->ifa_ibytes); 1758 ifam32->ifam_data.ifi_obytes = 1759 counter_u64_fetch(ifa->ifa_obytes); 1760 1761 /* Fixup if_data carp(4) vhid. */ 1762 if (carp_get_vhid_p != NULL) 1763 ifam32->ifam_data.ifi_vhid = (*carp_get_vhid_p)(ifa); 1764 1765 return (SYSCTL_OUT(w->w_req, (caddr_t)ifam32, len)); 1766 } 1767 #endif 1768 1769 ifam = (struct ifa_msghdrl *)w->w_tmem; 1770 ifam->ifam_addrs = info->rti_addrs; 1771 ifam->ifam_flags = ifa->ifa_flags; 1772 ifam->ifam_index = ifa->ifa_ifp->if_index; 1773 ifam->_ifam_spare1 = 0; 1774 ifam->ifam_len = sizeof(*ifam); 1775 ifam->ifam_data_off = offsetof(struct ifa_msghdrl, ifam_data); 1776 ifam->ifam_metric = ifa->ifa_metric; 1777 1778 bzero(&ifam->ifam_data, sizeof(ifam->ifam_data)); 1779 ifam->ifam_data.ifi_datalen = sizeof(struct if_data); 1780 ifam->ifam_data.ifi_ipackets = counter_u64_fetch(ifa->ifa_ipackets); 1781 ifam->ifam_data.ifi_opackets = counter_u64_fetch(ifa->ifa_opackets); 1782 ifam->ifam_data.ifi_ibytes = counter_u64_fetch(ifa->ifa_ibytes); 1783 ifam->ifam_data.ifi_obytes = counter_u64_fetch(ifa->ifa_obytes); 1784 1785 /* Fixup if_data carp(4) vhid. */ 1786 if (carp_get_vhid_p != NULL) 1787 ifam->ifam_data.ifi_vhid = (*carp_get_vhid_p)(ifa); 1788 1789 return (SYSCTL_OUT(w->w_req, w->w_tmem, len)); 1790 } 1791 1792 static int 1793 sysctl_iflist_ifam(struct ifaddr *ifa, struct rt_addrinfo *info, 1794 struct walkarg *w, int len) 1795 { 1796 struct ifa_msghdr *ifam; 1797 1798 ifam = (struct ifa_msghdr *)w->w_tmem; 1799 ifam->ifam_addrs = info->rti_addrs; 1800 ifam->ifam_flags = ifa->ifa_flags; 1801 ifam->ifam_index = ifa->ifa_ifp->if_index; 1802 ifam->ifam_metric = ifa->ifa_metric; 1803 1804 return (SYSCTL_OUT(w->w_req, w->w_tmem, len)); 1805 } 1806 1807 static int 1808 sysctl_iflist(int af, struct walkarg *w) 1809 { 1810 struct ifnet *ifp; 1811 struct ifaddr *ifa; 1812 struct rt_addrinfo info; 1813 int len, error = 0; 1814 1815 bzero((caddr_t)&info, sizeof(info)); 1816 IFNET_RLOCK_NOSLEEP(); 1817 TAILQ_FOREACH(ifp, &V_ifnet, if_link) { 1818 if (w->w_arg && w->w_arg != ifp->if_index) 1819 continue; 1820 IF_ADDR_RLOCK(ifp); 1821 ifa = ifp->if_addr; 1822 info.rti_info[RTAX_IFP] = ifa->ifa_addr; 1823 len = rt_msg2(RTM_IFINFO, &info, NULL, w); 1824 info.rti_info[RTAX_IFP] = NULL; 1825 if (w->w_req && w->w_tmem) { 1826 if (w->w_op == NET_RT_IFLISTL) 1827 error = sysctl_iflist_ifml(ifp, &info, w, len); 1828 else 1829 error = sysctl_iflist_ifm(ifp, &info, w, len); 1830 if (error) 1831 goto done; 1832 } 1833 while ((ifa = TAILQ_NEXT(ifa, ifa_link)) != NULL) { 1834 if (af && af != ifa->ifa_addr->sa_family) 1835 continue; 1836 if (prison_if(w->w_req->td->td_ucred, 1837 ifa->ifa_addr) != 0) 1838 continue; 1839 info.rti_info[RTAX_IFA] = ifa->ifa_addr; 1840 info.rti_info[RTAX_NETMASK] = ifa->ifa_netmask; 1841 info.rti_info[RTAX_BRD] = ifa->ifa_dstaddr; 1842 len = rt_msg2(RTM_NEWADDR, &info, NULL, w); 1843 if (w->w_req && w->w_tmem) { 1844 if (w->w_op == NET_RT_IFLISTL) 1845 error = sysctl_iflist_ifaml(ifa, &info, 1846 w, len); 1847 else 1848 error = sysctl_iflist_ifam(ifa, &info, 1849 w, len); 1850 if (error) 1851 goto done; 1852 } 1853 } 1854 IF_ADDR_RUNLOCK(ifp); 1855 info.rti_info[RTAX_IFA] = info.rti_info[RTAX_NETMASK] = 1856 info.rti_info[RTAX_BRD] = NULL; 1857 } 1858 done: 1859 if (ifp != NULL) 1860 IF_ADDR_RUNLOCK(ifp); 1861 IFNET_RUNLOCK_NOSLEEP(); 1862 return (error); 1863 } 1864 1865 static int 1866 sysctl_ifmalist(int af, struct walkarg *w) 1867 { 1868 struct ifnet *ifp; 1869 struct ifmultiaddr *ifma; 1870 struct rt_addrinfo info; 1871 int len, error = 0; 1872 struct ifaddr *ifa; 1873 1874 bzero((caddr_t)&info, sizeof(info)); 1875 IFNET_RLOCK_NOSLEEP(); 1876 TAILQ_FOREACH(ifp, &V_ifnet, if_link) { 1877 if (w->w_arg && w->w_arg != ifp->if_index) 1878 continue; 1879 ifa = ifp->if_addr; 1880 info.rti_info[RTAX_IFP] = ifa ? ifa->ifa_addr : NULL; 1881 IF_ADDR_RLOCK(ifp); 1882 TAILQ_FOREACH(ifma, &ifp->if_multiaddrs, ifma_link) { 1883 if (af && af != ifma->ifma_addr->sa_family) 1884 continue; 1885 if (prison_if(w->w_req->td->td_ucred, 1886 ifma->ifma_addr) != 0) 1887 continue; 1888 info.rti_info[RTAX_IFA] = ifma->ifma_addr; 1889 info.rti_info[RTAX_GATEWAY] = 1890 (ifma->ifma_addr->sa_family != AF_LINK) ? 1891 ifma->ifma_lladdr : NULL; 1892 len = rt_msg2(RTM_NEWMADDR, &info, NULL, w); 1893 if (w->w_req && w->w_tmem) { 1894 struct ifma_msghdr *ifmam; 1895 1896 ifmam = (struct ifma_msghdr *)w->w_tmem; 1897 ifmam->ifmam_index = ifma->ifma_ifp->if_index; 1898 ifmam->ifmam_flags = 0; 1899 ifmam->ifmam_addrs = info.rti_addrs; 1900 error = SYSCTL_OUT(w->w_req, w->w_tmem, len); 1901 if (error) { 1902 IF_ADDR_RUNLOCK(ifp); 1903 goto done; 1904 } 1905 } 1906 } 1907 IF_ADDR_RUNLOCK(ifp); 1908 } 1909 done: 1910 IFNET_RUNLOCK_NOSLEEP(); 1911 return (error); 1912 } 1913 1914 static int 1915 sysctl_rtsock(SYSCTL_HANDLER_ARGS) 1916 { 1917 int *name = (int *)arg1; 1918 u_int namelen = arg2; 1919 struct radix_node_head *rnh = NULL; /* silence compiler. */ 1920 int i, lim, error = EINVAL; 1921 int fib = 0; 1922 u_char af; 1923 struct walkarg w; 1924 1925 name ++; 1926 namelen--; 1927 if (req->newptr) 1928 return (EPERM); 1929 if (name[1] == NET_RT_DUMP) { 1930 if (namelen == 3) 1931 fib = req->td->td_proc->p_fibnum; 1932 else if (namelen == 4) 1933 fib = (name[3] == -1) ? 1934 req->td->td_proc->p_fibnum : name[3]; 1935 else 1936 return ((namelen < 3) ? EISDIR : ENOTDIR); 1937 if (fib < 0 || fib >= rt_numfibs) 1938 return (EINVAL); 1939 } else if (namelen != 3) 1940 return ((namelen < 3) ? EISDIR : ENOTDIR); 1941 af = name[0]; 1942 if (af > AF_MAX) 1943 return (EINVAL); 1944 bzero(&w, sizeof(w)); 1945 w.w_op = name[1]; 1946 w.w_arg = name[2]; 1947 w.w_req = req; 1948 1949 error = sysctl_wire_old_buffer(req, 0); 1950 if (error) 1951 return (error); 1952 switch (w.w_op) { 1953 1954 case NET_RT_DUMP: 1955 case NET_RT_FLAGS: 1956 if (af == 0) { /* dump all tables */ 1957 i = 1; 1958 lim = AF_MAX; 1959 } else /* dump only one table */ 1960 i = lim = af; 1961 1962 /* 1963 * take care of llinfo entries, the caller must 1964 * specify an AF 1965 */ 1966 if (w.w_op == NET_RT_FLAGS && 1967 (w.w_arg == 0 || w.w_arg & RTF_LLINFO)) { 1968 if (af != 0) 1969 error = lltable_sysctl_dumparp(af, w.w_req); 1970 else 1971 error = EINVAL; 1972 break; 1973 } 1974 /* 1975 * take care of routing entries 1976 */ 1977 for (error = 0; error == 0 && i <= lim; i++) { 1978 rnh = rt_tables_get_rnh(fib, i); 1979 if (rnh != NULL) { 1980 RADIX_NODE_HEAD_RLOCK(rnh); 1981 error = rnh->rnh_walktree(rnh, 1982 sysctl_dumpentry, &w); 1983 RADIX_NODE_HEAD_RUNLOCK(rnh); 1984 } else if (af != 0) 1985 error = EAFNOSUPPORT; 1986 } 1987 break; 1988 1989 case NET_RT_IFLIST: 1990 case NET_RT_IFLISTL: 1991 error = sysctl_iflist(af, &w); 1992 break; 1993 1994 case NET_RT_IFMALIST: 1995 error = sysctl_ifmalist(af, &w); 1996 break; 1997 } 1998 if (w.w_tmem) 1999 free(w.w_tmem, M_RTABLE); 2000 return (error); 2001 } 2002 2003 static SYSCTL_NODE(_net, PF_ROUTE, routetable, CTLFLAG_RD, sysctl_rtsock, ""); 2004 2005 /* 2006 * Definitions of protocols supported in the ROUTE domain. 2007 */ 2008 2009 static struct domain routedomain; /* or at least forward */ 2010 2011 static struct protosw routesw[] = { 2012 { 2013 .pr_type = SOCK_RAW, 2014 .pr_domain = &routedomain, 2015 .pr_flags = PR_ATOMIC|PR_ADDR, 2016 .pr_output = route_output, 2017 .pr_ctlinput = raw_ctlinput, 2018 .pr_init = raw_init, 2019 .pr_usrreqs = &route_usrreqs 2020 } 2021 }; 2022 2023 static struct domain routedomain = { 2024 .dom_family = PF_ROUTE, 2025 .dom_name = "route", 2026 .dom_protosw = routesw, 2027 .dom_protoswNPROTOSW = &routesw[sizeof(routesw)/sizeof(routesw[0])] 2028 }; 2029 2030 VNET_DOMAIN_SET(route); 2031