1 /* $OpenBSD: rtsock.c,v 1.365 2023/04/20 21:43:17 mvs Exp $ */ 2 /* $NetBSD: rtsock.c,v 1.18 1996/03/29 00:32:10 cgd Exp $ */ 3 4 /* 5 * Copyright (C) 1995, 1996, 1997, and 1998 WIDE Project. 6 * All rights reserved. 7 * 8 * Redistribution and use in source and binary forms, with or without 9 * modification, are permitted provided that the following conditions 10 * are met: 11 * 1. Redistributions of source code must retain the above copyright 12 * notice, this list of conditions and the following disclaimer. 13 * 2. Redistributions in binary form must reproduce the above copyright 14 * notice, this list of conditions and the following disclaimer in the 15 * documentation and/or other materials provided with the distribution. 16 * 3. Neither the name of the project nor the names of its contributors 17 * may be used to endorse or promote products derived from this software 18 * without specific prior written permission. 19 * 20 * THIS SOFTWARE IS PROVIDED BY THE PROJECT AND CONTRIBUTORS ``AS IS'' AND 21 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE 22 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE 23 * ARE DISCLAIMED. IN NO EVENT SHALL THE PROJECT OR CONTRIBUTORS BE LIABLE 24 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL 25 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS 26 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) 27 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT 28 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY 29 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF 30 * SUCH DAMAGE. 31 */ 32 33 /* 34 * Copyright (c) 1988, 1991, 1993 35 * The Regents of the University of California. All rights reserved. 36 * 37 * Redistribution and use in source and binary forms, with or without 38 * modification, are permitted provided that the following conditions 39 * are met: 40 * 1. Redistributions of source code must retain the above copyright 41 * notice, this list of conditions and the following disclaimer. 42 * 2. Redistributions in binary form must reproduce the above copyright 43 * notice, this list of conditions and the following disclaimer in the 44 * documentation and/or other materials provided with the distribution. 45 * 3. Neither the name of the University nor the names of its contributors 46 * may be used to endorse or promote products derived from this software 47 * without specific prior written permission. 48 * 49 * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND 50 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE 51 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE 52 * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE 53 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL 54 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS 55 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) 56 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT 57 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY 58 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF 59 * SUCH DAMAGE. 60 * 61 * @(#)rtsock.c 8.6 (Berkeley) 2/11/95 62 */ 63 64 #include <sys/param.h> 65 #include <sys/systm.h> 66 #include <sys/proc.h> 67 #include <sys/sysctl.h> 68 #include <sys/mbuf.h> 69 #include <sys/socket.h> 70 #include <sys/socketvar.h> 71 #include <sys/domain.h> 72 #include <sys/pool.h> 73 #include <sys/protosw.h> 74 #include <sys/srp.h> 75 76 #include <net/if.h> 77 #include <net/if_dl.h> 78 #include <net/if_var.h> 79 #include <net/route.h> 80 81 #include <netinet/in.h> 82 83 #ifdef MPLS 84 #include <netmpls/mpls.h> 85 #endif 86 #ifdef IPSEC 87 #include <netinet/ip_ipsp.h> 88 #include <net/if_enc.h> 89 #endif 90 #ifdef BFD 91 #include <net/bfd.h> 92 #endif 93 94 #include <sys/stdarg.h> 95 #include <sys/kernel.h> 96 #include <sys/timeout.h> 97 98 #define ROUTESNDQ 8192 99 #define ROUTERCVQ 8192 100 101 const struct sockaddr route_src = { 2, PF_ROUTE, }; 102 103 struct walkarg { 104 int w_op, w_arg, w_tmemsize; 105 size_t w_given, w_needed; 106 caddr_t w_where, w_tmem; 107 }; 108 109 void route_prinit(void); 110 void rcb_ref(void *, void *); 111 void rcb_unref(void *, void *); 112 int route_output(struct mbuf *, struct socket *); 113 int route_ctloutput(int, struct socket *, int, int, struct mbuf *); 114 int route_attach(struct socket *, int, int); 115 int route_detach(struct socket *); 116 int route_disconnect(struct socket *); 117 int route_shutdown(struct socket *); 118 void route_rcvd(struct socket *); 119 int route_send(struct socket *, struct mbuf *, struct mbuf *, 120 struct mbuf *); 121 int route_sockaddr(struct socket *, struct mbuf *); 122 int route_peeraddr(struct socket *, struct mbuf *); 123 void route_input(struct mbuf *m0, struct socket *, sa_family_t); 124 int route_arp_conflict(struct rtentry *, struct rt_addrinfo *); 125 int route_cleargateway(struct rtentry *, void *, unsigned int); 126 void rtm_senddesync_timer(void *); 127 void rtm_senddesync(struct socket *); 128 int rtm_sendup(struct socket *, struct mbuf *); 129 130 int rtm_getifa(struct rt_addrinfo *, unsigned int); 131 int rtm_output(struct rt_msghdr *, struct rtentry **, struct rt_addrinfo *, 132 uint8_t, unsigned int); 133 struct rt_msghdr *rtm_report(struct rtentry *, u_char, int, int); 134 struct mbuf *rtm_msg1(int, struct rt_addrinfo *); 135 int rtm_msg2(int, int, struct rt_addrinfo *, caddr_t, 136 struct walkarg *); 137 int rtm_xaddrs(caddr_t, caddr_t, struct rt_addrinfo *); 138 int rtm_validate_proposal(struct rt_addrinfo *); 139 void rtm_setmetrics(u_long, const struct rt_metrics *, 140 struct rt_kmetrics *); 141 void rtm_getmetrics(const struct rtentry *, 142 struct rt_metrics *); 143 144 int sysctl_iflist(int, struct walkarg *); 145 int sysctl_ifnames(struct walkarg *); 146 int sysctl_rtable_rtstat(void *, size_t *, void *); 147 148 int rt_setsource(unsigned int, struct sockaddr *); 149 150 /* 151 * Locks used to protect struct members 152 * I immutable after creation 153 * s solock 154 */ 155 struct rtpcb { 156 struct socket *rop_socket; /* [I] */ 157 158 SRPL_ENTRY(rtpcb) rop_list; 159 struct refcnt rop_refcnt; 160 struct timeout rop_timeout; 161 unsigned int rop_msgfilter; /* [s] */ 162 unsigned int rop_flagfilter; /* [s] */ 163 unsigned int rop_flags; /* [s] */ 164 u_int rop_rtableid; /* [s] */ 165 unsigned short rop_proto; /* [I] */ 166 u_char rop_priority; /* [s] */ 167 }; 168 #define sotortpcb(so) ((struct rtpcb *)(so)->so_pcb) 169 170 struct rtptable { 171 SRPL_HEAD(, rtpcb) rtp_list; 172 struct srpl_rc rtp_rc; 173 struct rwlock rtp_lk; 174 unsigned int rtp_count; 175 }; 176 177 struct pool rtpcb_pool; 178 struct rtptable rtptable; 179 180 /* 181 * These flags and timeout are used for indicating to userland (via a 182 * RTM_DESYNC msg) when the route socket has overflowed and messages 183 * have been lost. 184 */ 185 #define ROUTECB_FLAG_DESYNC 0x1 /* Route socket out of memory */ 186 #define ROUTECB_FLAG_FLUSH 0x2 /* Wait until socket is empty before 187 queueing more packets */ 188 189 #define ROUTE_DESYNC_RESEND_TIMEOUT 200 /* In ms */ 190 191 void 192 route_prinit(void) 193 { 194 srpl_rc_init(&rtptable.rtp_rc, rcb_ref, rcb_unref, NULL); 195 rw_init(&rtptable.rtp_lk, "rtsock"); 196 SRPL_INIT(&rtptable.rtp_list); 197 pool_init(&rtpcb_pool, sizeof(struct rtpcb), 0, 198 IPL_SOFTNET, PR_WAITOK, "rtpcb", NULL); 199 } 200 201 void 202 rcb_ref(void *null, void *v) 203 { 204 struct rtpcb *rop = v; 205 206 refcnt_take(&rop->rop_refcnt); 207 } 208 209 void 210 rcb_unref(void *null, void *v) 211 { 212 struct rtpcb *rop = v; 213 214 refcnt_rele_wake(&rop->rop_refcnt); 215 } 216 217 int 218 route_attach(struct socket *so, int proto, int wait) 219 { 220 struct rtpcb *rop; 221 int error; 222 223 error = soreserve(so, ROUTESNDQ, ROUTERCVQ); 224 if (error) 225 return (error); 226 /* 227 * use the rawcb but allocate a rtpcb, this 228 * code does not care about the additional fields 229 * and works directly on the raw socket. 230 */ 231 rop = pool_get(&rtpcb_pool, (wait == M_WAIT ? PR_WAITOK : PR_NOWAIT) | 232 PR_ZERO); 233 if (rop == NULL) 234 return (ENOBUFS); 235 so->so_pcb = rop; 236 /* Init the timeout structure */ 237 timeout_set_proc(&rop->rop_timeout, rtm_senddesync_timer, so); 238 refcnt_init(&rop->rop_refcnt); 239 240 rop->rop_socket = so; 241 rop->rop_proto = proto; 242 243 rop->rop_rtableid = curproc->p_p->ps_rtableid; 244 245 soisconnected(so); 246 so->so_options |= SO_USELOOPBACK; 247 248 rw_enter(&rtptable.rtp_lk, RW_WRITE); 249 SRPL_INSERT_HEAD_LOCKED(&rtptable.rtp_rc, &rtptable.rtp_list, rop, 250 rop_list); 251 rtptable.rtp_count++; 252 rw_exit(&rtptable.rtp_lk); 253 254 return (0); 255 } 256 257 int 258 route_detach(struct socket *so) 259 { 260 struct rtpcb *rop; 261 262 soassertlocked(so); 263 264 rop = sotortpcb(so); 265 if (rop == NULL) 266 return (EINVAL); 267 268 rw_enter(&rtptable.rtp_lk, RW_WRITE); 269 270 rtptable.rtp_count--; 271 SRPL_REMOVE_LOCKED(&rtptable.rtp_rc, &rtptable.rtp_list, rop, rtpcb, 272 rop_list); 273 rw_exit(&rtptable.rtp_lk); 274 275 sounlock(so); 276 277 /* wait for all references to drop */ 278 refcnt_finalize(&rop->rop_refcnt, "rtsockrefs"); 279 timeout_del_barrier(&rop->rop_timeout); 280 281 solock(so); 282 283 so->so_pcb = NULL; 284 KASSERT((so->so_state & SS_NOFDREF) == 0); 285 pool_put(&rtpcb_pool, rop); 286 287 return (0); 288 } 289 290 int 291 route_disconnect(struct socket *so) 292 { 293 soisdisconnected(so); 294 return (0); 295 } 296 297 int 298 route_shutdown(struct socket *so) 299 { 300 socantsendmore(so); 301 return (0); 302 } 303 304 void 305 route_rcvd(struct socket *so) 306 { 307 struct rtpcb *rop = sotortpcb(so); 308 309 soassertlocked(so); 310 311 /* 312 * If we are in a FLUSH state, check if the buffer is 313 * empty so that we can clear the flag. 314 */ 315 if (((rop->rop_flags & ROUTECB_FLAG_FLUSH) != 0) && 316 ((sbspace(rop->rop_socket, &rop->rop_socket->so_rcv) == 317 rop->rop_socket->so_rcv.sb_hiwat))) 318 rop->rop_flags &= ~ROUTECB_FLAG_FLUSH; 319 } 320 321 int 322 route_send(struct socket *so, struct mbuf *m, struct mbuf *nam, 323 struct mbuf *control) 324 { 325 int error; 326 327 soassertlocked(so); 328 329 if (control && control->m_len) { 330 error = EOPNOTSUPP; 331 goto out; 332 } 333 334 if (nam) { 335 error = EISCONN; 336 goto out; 337 } 338 339 error = route_output(m, so); 340 m = NULL; 341 342 out: 343 m_freem(control); 344 m_freem(m); 345 346 return (error); 347 } 348 349 int 350 route_sockaddr(struct socket *so, struct mbuf *nam) 351 { 352 return (EINVAL); 353 } 354 355 int 356 route_peeraddr(struct socket *so, struct mbuf *nam) 357 { 358 /* minimal support, just implement a fake peer address */ 359 bcopy(&route_src, mtod(nam, caddr_t), route_src.sa_len); 360 nam->m_len = route_src.sa_len; 361 return (0); 362 } 363 364 int 365 route_ctloutput(int op, struct socket *so, int level, int optname, 366 struct mbuf *m) 367 { 368 struct rtpcb *rop = sotortpcb(so); 369 int error = 0; 370 unsigned int tid, prio; 371 372 if (level != AF_ROUTE) 373 return (EINVAL); 374 375 switch (op) { 376 case PRCO_SETOPT: 377 switch (optname) { 378 case ROUTE_MSGFILTER: 379 if (m == NULL || m->m_len != sizeof(unsigned int)) 380 error = EINVAL; 381 else 382 rop->rop_msgfilter = *mtod(m, unsigned int *); 383 break; 384 case ROUTE_TABLEFILTER: 385 if (m == NULL || m->m_len != sizeof(unsigned int)) { 386 error = EINVAL; 387 break; 388 } 389 tid = *mtod(m, unsigned int *); 390 if (tid != RTABLE_ANY && !rtable_exists(tid)) 391 error = ENOENT; 392 else 393 rop->rop_rtableid = tid; 394 break; 395 case ROUTE_PRIOFILTER: 396 if (m == NULL || m->m_len != sizeof(unsigned int)) { 397 error = EINVAL; 398 break; 399 } 400 prio = *mtod(m, unsigned int *); 401 if (prio > RTP_MAX) 402 error = EINVAL; 403 else 404 rop->rop_priority = prio; 405 break; 406 case ROUTE_FLAGFILTER: 407 if (m == NULL || m->m_len != sizeof(unsigned int)) 408 error = EINVAL; 409 else 410 rop->rop_flagfilter = *mtod(m, unsigned int *); 411 break; 412 default: 413 error = ENOPROTOOPT; 414 break; 415 } 416 break; 417 case PRCO_GETOPT: 418 switch (optname) { 419 case ROUTE_MSGFILTER: 420 m->m_len = sizeof(unsigned int); 421 *mtod(m, unsigned int *) = rop->rop_msgfilter; 422 break; 423 case ROUTE_TABLEFILTER: 424 m->m_len = sizeof(unsigned int); 425 *mtod(m, unsigned int *) = rop->rop_rtableid; 426 break; 427 case ROUTE_PRIOFILTER: 428 m->m_len = sizeof(unsigned int); 429 *mtod(m, unsigned int *) = rop->rop_priority; 430 break; 431 case ROUTE_FLAGFILTER: 432 m->m_len = sizeof(unsigned int); 433 *mtod(m, unsigned int *) = rop->rop_flagfilter; 434 break; 435 default: 436 error = ENOPROTOOPT; 437 break; 438 } 439 } 440 return (error); 441 } 442 443 void 444 rtm_senddesync_timer(void *xso) 445 { 446 struct socket *so = xso; 447 448 solock(so); 449 rtm_senddesync(so); 450 sounlock(so); 451 } 452 453 void 454 rtm_senddesync(struct socket *so) 455 { 456 struct rtpcb *rop = sotortpcb(so); 457 struct mbuf *desync_mbuf; 458 459 soassertlocked(so); 460 461 /* 462 * Dying socket is disconnected by upper layer and there is 463 * no reason to send packet. Also we shouldn't reschedule 464 * timeout(9), otherwise timeout_del_barrier(9) can't help us. 465 */ 466 if ((so->so_state & SS_ISCONNECTED) == 0 || 467 (so->so_rcv.sb_state & SS_CANTRCVMORE)) 468 return; 469 470 /* If we are in a DESYNC state, try to send a RTM_DESYNC packet */ 471 if ((rop->rop_flags & ROUTECB_FLAG_DESYNC) == 0) 472 return; 473 474 /* 475 * If we fail to alloc memory or if sbappendaddr() 476 * fails, re-add timeout and try again. 477 */ 478 desync_mbuf = rtm_msg1(RTM_DESYNC, NULL); 479 if (desync_mbuf != NULL) { 480 if (sbappendaddr(so, &so->so_rcv, &route_src, 481 desync_mbuf, NULL) != 0) { 482 rop->rop_flags &= ~ROUTECB_FLAG_DESYNC; 483 sorwakeup(rop->rop_socket); 484 return; 485 } 486 m_freem(desync_mbuf); 487 } 488 /* Re-add timeout to try sending msg again */ 489 timeout_add_msec(&rop->rop_timeout, ROUTE_DESYNC_RESEND_TIMEOUT); 490 } 491 492 void 493 route_input(struct mbuf *m0, struct socket *so0, sa_family_t sa_family) 494 { 495 struct socket *so; 496 struct rtpcb *rop; 497 struct rt_msghdr *rtm; 498 struct mbuf *m = m0; 499 struct srp_ref sr; 500 501 /* ensure that we can access the rtm_type via mtod() */ 502 if (m->m_len < offsetof(struct rt_msghdr, rtm_type) + 1) { 503 m_freem(m); 504 return; 505 } 506 507 SRPL_FOREACH(rop, &sr, &rtptable.rtp_list, rop_list) { 508 /* 509 * If route socket is bound to an address family only send 510 * messages that match the address family. Address family 511 * agnostic messages are always sent. 512 */ 513 if (sa_family != AF_UNSPEC && rop->rop_proto != AF_UNSPEC && 514 rop->rop_proto != sa_family) 515 continue; 516 517 518 so = rop->rop_socket; 519 solock(so); 520 521 /* 522 * Check to see if we don't want our own messages and 523 * if we can receive anything. 524 */ 525 if ((so0 == so && !(so0->so_options & SO_USELOOPBACK)) || 526 !(so->so_state & SS_ISCONNECTED) || 527 (so->so_rcv.sb_state & SS_CANTRCVMORE)) 528 goto next; 529 530 /* filter messages that the process does not want */ 531 rtm = mtod(m, struct rt_msghdr *); 532 /* but RTM_DESYNC can't be filtered */ 533 if (rtm->rtm_type != RTM_DESYNC) { 534 if (rop->rop_msgfilter != 0 && 535 !(rop->rop_msgfilter & (1U << rtm->rtm_type))) 536 goto next; 537 if (ISSET(rop->rop_flagfilter, rtm->rtm_flags)) 538 goto next; 539 } 540 switch (rtm->rtm_type) { 541 case RTM_IFANNOUNCE: 542 case RTM_DESYNC: 543 /* no tableid */ 544 break; 545 case RTM_RESOLVE: 546 case RTM_NEWADDR: 547 case RTM_DELADDR: 548 case RTM_IFINFO: 549 case RTM_80211INFO: 550 case RTM_BFD: 551 /* check against rdomain id */ 552 if (rop->rop_rtableid != RTABLE_ANY && 553 rtable_l2(rop->rop_rtableid) != rtm->rtm_tableid) 554 goto next; 555 break; 556 default: 557 if (rop->rop_priority != 0 && 558 rop->rop_priority < rtm->rtm_priority) 559 goto next; 560 /* check against rtable id */ 561 if (rop->rop_rtableid != RTABLE_ANY && 562 rop->rop_rtableid != rtm->rtm_tableid) 563 goto next; 564 break; 565 } 566 567 /* 568 * Check to see if the flush flag is set. If so, don't queue 569 * any more messages until the flag is cleared. 570 */ 571 if ((rop->rop_flags & ROUTECB_FLAG_FLUSH) != 0) 572 goto next; 573 574 rtm_sendup(so, m); 575 next: 576 sounlock(so); 577 } 578 SRPL_LEAVE(&sr); 579 580 m_freem(m); 581 } 582 583 int 584 rtm_sendup(struct socket *so, struct mbuf *m0) 585 { 586 struct rtpcb *rop = sotortpcb(so); 587 struct mbuf *m; 588 589 soassertlocked(so); 590 591 m = m_copym(m0, 0, M_COPYALL, M_NOWAIT); 592 if (m == NULL) 593 return (ENOMEM); 594 595 if (sbspace(so, &so->so_rcv) < (2 * MSIZE) || 596 sbappendaddr(so, &so->so_rcv, &route_src, m, NULL) == 0) { 597 /* Flag socket as desync'ed and flush required */ 598 rop->rop_flags |= ROUTECB_FLAG_DESYNC | ROUTECB_FLAG_FLUSH; 599 rtm_senddesync(so); 600 m_freem(m); 601 return (ENOBUFS); 602 } 603 604 sorwakeup(so); 605 return (0); 606 } 607 608 struct rt_msghdr * 609 rtm_report(struct rtentry *rt, u_char type, int seq, int tableid) 610 { 611 struct rt_msghdr *rtm; 612 struct rt_addrinfo info; 613 struct sockaddr_rtlabel sa_rl; 614 struct sockaddr_in6 sa_mask; 615 #ifdef BFD 616 struct sockaddr_bfd sa_bfd; 617 #endif 618 struct ifnet *ifp = NULL; 619 int len; 620 621 bzero(&info, sizeof(info)); 622 info.rti_info[RTAX_DST] = rt_key(rt); 623 info.rti_info[RTAX_GATEWAY] = rt->rt_gateway; 624 info.rti_info[RTAX_NETMASK] = rt_plen2mask(rt, &sa_mask); 625 info.rti_info[RTAX_LABEL] = rtlabel_id2sa(rt->rt_labelid, &sa_rl); 626 #ifdef BFD 627 if (rt->rt_flags & RTF_BFD) { 628 KERNEL_LOCK(); 629 info.rti_info[RTAX_BFD] = bfd2sa(rt, &sa_bfd); 630 KERNEL_UNLOCK(); 631 } 632 #endif 633 #ifdef MPLS 634 if (rt->rt_flags & RTF_MPLS) { 635 struct sockaddr_mpls sa_mpls; 636 637 bzero(&sa_mpls, sizeof(sa_mpls)); 638 sa_mpls.smpls_family = AF_MPLS; 639 sa_mpls.smpls_len = sizeof(sa_mpls); 640 sa_mpls.smpls_label = ((struct rt_mpls *) 641 rt->rt_llinfo)->mpls_label; 642 info.rti_info[RTAX_SRC] = (struct sockaddr *)&sa_mpls; 643 info.rti_mpls = ((struct rt_mpls *) 644 rt->rt_llinfo)->mpls_operation; 645 } 646 #endif 647 ifp = if_get(rt->rt_ifidx); 648 if (ifp != NULL) { 649 info.rti_info[RTAX_IFP] = sdltosa(ifp->if_sadl); 650 info.rti_info[RTAX_IFA] = rtable_getsource(tableid, 651 info.rti_info[RTAX_DST]->sa_family); 652 if (info.rti_info[RTAX_IFA] == NULL) 653 info.rti_info[RTAX_IFA] = rt->rt_ifa->ifa_addr; 654 if (ifp->if_flags & IFF_POINTOPOINT) 655 info.rti_info[RTAX_BRD] = rt->rt_ifa->ifa_dstaddr; 656 } 657 if_put(ifp); 658 /* RTAX_GENMASK, RTAX_AUTHOR, RTAX_SRCMASK ignored */ 659 660 /* build new route message */ 661 len = rtm_msg2(type, RTM_VERSION, &info, NULL, NULL); 662 rtm = malloc(len, M_RTABLE, M_WAITOK | M_ZERO); 663 664 rtm_msg2(type, RTM_VERSION, &info, (caddr_t)rtm, NULL); 665 rtm->rtm_type = type; 666 rtm->rtm_index = rt->rt_ifidx; 667 rtm->rtm_tableid = tableid; 668 rtm->rtm_priority = rt->rt_priority & RTP_MASK; 669 rtm->rtm_flags = rt->rt_flags; 670 rtm->rtm_pid = curproc->p_p->ps_pid; 671 rtm->rtm_seq = seq; 672 rtm_getmetrics(rt, &rtm->rtm_rmx); 673 rtm->rtm_addrs = info.rti_addrs; 674 #ifdef MPLS 675 rtm->rtm_mpls = info.rti_mpls; 676 #endif 677 return rtm; 678 } 679 680 int 681 route_output(struct mbuf *m, struct socket *so) 682 { 683 struct rt_msghdr *rtm = NULL; 684 struct rtentry *rt = NULL; 685 struct rt_addrinfo info; 686 struct ifnet *ifp; 687 int len, seq, useloopback, error = 0; 688 u_int tableid; 689 u_int8_t prio; 690 u_char vers, type; 691 692 if (m == NULL || ((m->m_len < sizeof(int32_t)) && 693 (m = m_pullup(m, sizeof(int32_t))) == 0)) 694 return (ENOBUFS); 695 if ((m->m_flags & M_PKTHDR) == 0) 696 panic("route_output"); 697 698 useloopback = so->so_options & SO_USELOOPBACK; 699 700 /* 701 * The socket can't be closed concurrently because the file 702 * descriptor reference is still held. 703 */ 704 705 sounlock(so); 706 707 len = m->m_pkthdr.len; 708 if (len < offsetof(struct rt_msghdr, rtm_hdrlen) + 1 || 709 len != mtod(m, struct rt_msghdr *)->rtm_msglen) { 710 error = EINVAL; 711 goto fail; 712 } 713 vers = mtod(m, struct rt_msghdr *)->rtm_version; 714 switch (vers) { 715 case RTM_VERSION: 716 if (len < sizeof(struct rt_msghdr)) { 717 error = EINVAL; 718 goto fail; 719 } 720 if (len > RTM_MAXSIZE) { 721 error = EMSGSIZE; 722 goto fail; 723 } 724 rtm = malloc(len, M_RTABLE, M_WAITOK); 725 m_copydata(m, 0, len, rtm); 726 break; 727 default: 728 error = EPROTONOSUPPORT; 729 goto fail; 730 } 731 732 /* Verify that the caller is sending an appropriate message early */ 733 switch (rtm->rtm_type) { 734 case RTM_ADD: 735 case RTM_DELETE: 736 case RTM_GET: 737 case RTM_CHANGE: 738 case RTM_PROPOSAL: 739 case RTM_SOURCE: 740 break; 741 default: 742 error = EOPNOTSUPP; 743 goto fail; 744 } 745 /* 746 * Verify that the header length is valid. 747 * All messages from userland start with a struct rt_msghdr. 748 */ 749 if (rtm->rtm_hdrlen == 0) /* old client */ 750 rtm->rtm_hdrlen = sizeof(struct rt_msghdr); 751 if (rtm->rtm_hdrlen < sizeof(struct rt_msghdr) || 752 len < rtm->rtm_hdrlen) { 753 error = EINVAL; 754 goto fail; 755 } 756 757 rtm->rtm_pid = curproc->p_p->ps_pid; 758 759 /* 760 * Verify that the caller has the appropriate privilege; RTM_GET 761 * is the only operation the non-superuser is allowed. 762 */ 763 if (rtm->rtm_type != RTM_GET && suser(curproc) != 0) { 764 error = EACCES; 765 goto fail; 766 } 767 tableid = rtm->rtm_tableid; 768 if (!rtable_exists(tableid)) { 769 if (rtm->rtm_type == RTM_ADD) { 770 if ((error = rtable_add(tableid)) != 0) 771 goto fail; 772 } else { 773 error = EINVAL; 774 goto fail; 775 } 776 } 777 778 /* Do not let userland play with kernel-only flags. */ 779 if ((rtm->rtm_flags & (RTF_LOCAL|RTF_BROADCAST)) != 0) { 780 error = EINVAL; 781 goto fail; 782 } 783 784 /* make sure that kernel-only bits are not set */ 785 rtm->rtm_priority &= RTP_MASK; 786 rtm->rtm_flags &= ~(RTF_DONE|RTF_CLONED|RTF_CACHED); 787 rtm->rtm_fmask &= RTF_FMASK; 788 789 if (rtm->rtm_priority != 0) { 790 if (rtm->rtm_priority > RTP_MAX || 791 rtm->rtm_priority == RTP_LOCAL) { 792 error = EINVAL; 793 goto fail; 794 } 795 prio = rtm->rtm_priority; 796 } else if (rtm->rtm_type != RTM_ADD) 797 prio = RTP_ANY; 798 else if (rtm->rtm_flags & RTF_STATIC) 799 prio = 0; 800 else 801 prio = RTP_DEFAULT; 802 803 bzero(&info, sizeof(info)); 804 info.rti_addrs = rtm->rtm_addrs; 805 if ((error = rtm_xaddrs(rtm->rtm_hdrlen + (caddr_t)rtm, 806 len + (caddr_t)rtm, &info)) != 0) 807 goto fail; 808 809 info.rti_flags = rtm->rtm_flags; 810 811 if (rtm->rtm_type != RTM_SOURCE && 812 rtm->rtm_type != RTM_PROPOSAL && 813 (info.rti_info[RTAX_DST] == NULL || 814 info.rti_info[RTAX_DST]->sa_family >= AF_MAX || 815 (info.rti_info[RTAX_GATEWAY] != NULL && 816 info.rti_info[RTAX_GATEWAY]->sa_family >= AF_MAX) || 817 info.rti_info[RTAX_GENMASK] != NULL)) { 818 error = EINVAL; 819 goto fail; 820 } 821 #ifdef MPLS 822 info.rti_mpls = rtm->rtm_mpls; 823 #endif 824 825 if (info.rti_info[RTAX_GATEWAY] != NULL && 826 info.rti_info[RTAX_GATEWAY]->sa_family == AF_LINK && 827 (info.rti_flags & RTF_CLONING) == 0) { 828 info.rti_flags |= RTF_LLINFO; 829 } 830 831 /* 832 * Validate RTM_PROPOSAL and pass it along or error out. 833 */ 834 if (rtm->rtm_type == RTM_PROPOSAL) { 835 if (rtm_validate_proposal(&info) == -1) { 836 error = EINVAL; 837 goto fail; 838 } 839 /* 840 * If this is a solicitation proposal forward request to 841 * all interfaces. Most handlers will ignore it but at least 842 * umb(4) will send a response to this event. 843 */ 844 if (rtm->rtm_priority == RTP_PROPOSAL_SOLICIT) { 845 NET_LOCK(); 846 TAILQ_FOREACH(ifp, &ifnetlist, if_list) { 847 ifp->if_rtrequest(ifp, RTM_PROPOSAL, NULL); 848 } 849 NET_UNLOCK(); 850 } 851 } else if (rtm->rtm_type == RTM_SOURCE) { 852 if (info.rti_info[RTAX_IFA] == NULL) { 853 error = EINVAL; 854 goto fail; 855 } 856 NET_LOCK(); 857 error = rt_setsource(tableid, info.rti_info[RTAX_IFA]); 858 NET_UNLOCK(); 859 if (error) 860 goto fail; 861 } else { 862 error = rtm_output(rtm, &rt, &info, prio, tableid); 863 if (!error) { 864 type = rtm->rtm_type; 865 seq = rtm->rtm_seq; 866 free(rtm, M_RTABLE, len); 867 NET_LOCK_SHARED(); 868 rtm = rtm_report(rt, type, seq, tableid); 869 NET_UNLOCK_SHARED(); 870 len = rtm->rtm_msglen; 871 } 872 } 873 874 rtfree(rt); 875 if (error) { 876 rtm->rtm_errno = error; 877 } else { 878 rtm->rtm_flags |= RTF_DONE; 879 } 880 881 /* 882 * Check to see if we don't want our own messages. 883 */ 884 if (!useloopback) { 885 if (rtptable.rtp_count == 0) { 886 /* no other listener and no loopback of messages */ 887 goto fail; 888 } 889 } 890 if (m_copyback(m, 0, len, rtm, M_NOWAIT)) { 891 m_freem(m); 892 m = NULL; 893 } else if (m->m_pkthdr.len > len) 894 m_adj(m, len - m->m_pkthdr.len); 895 free(rtm, M_RTABLE, len); 896 if (m) 897 route_input(m, so, info.rti_info[RTAX_DST] ? 898 info.rti_info[RTAX_DST]->sa_family : AF_UNSPEC); 899 solock(so); 900 901 return (error); 902 fail: 903 free(rtm, M_RTABLE, len); 904 m_freem(m); 905 solock(so); 906 907 return (error); 908 } 909 910 int 911 rtm_output(struct rt_msghdr *rtm, struct rtentry **prt, 912 struct rt_addrinfo *info, uint8_t prio, unsigned int tableid) 913 { 914 struct rtentry *rt = *prt; 915 struct ifnet *ifp = NULL; 916 int plen, newgate = 0, error = 0; 917 918 switch (rtm->rtm_type) { 919 case RTM_ADD: 920 if (info->rti_info[RTAX_GATEWAY] == NULL) { 921 error = EINVAL; 922 break; 923 } 924 925 rt = rtable_match(tableid, info->rti_info[RTAX_DST], NULL); 926 if ((error = route_arp_conflict(rt, info))) { 927 rtfree(rt); 928 rt = NULL; 929 break; 930 } 931 932 /* 933 * We cannot go through a delete/create/insert cycle for 934 * cached route because this can lead to races in the 935 * receive path. Instead we update the L2 cache. 936 */ 937 if ((rt != NULL) && ISSET(rt->rt_flags, RTF_CACHED)) { 938 ifp = if_get(rt->rt_ifidx); 939 if (ifp == NULL) { 940 rtfree(rt); 941 rt = NULL; 942 error = ESRCH; 943 break; 944 } 945 946 goto change; 947 } 948 949 rtfree(rt); 950 rt = NULL; 951 952 NET_LOCK(); 953 if ((error = rtm_getifa(info, tableid)) != 0) { 954 NET_UNLOCK(); 955 break; 956 } 957 error = rtrequest(RTM_ADD, info, prio, &rt, tableid); 958 NET_UNLOCK(); 959 if (error == 0) 960 rtm_setmetrics(rtm->rtm_inits, &rtm->rtm_rmx, 961 &rt->rt_rmx); 962 break; 963 case RTM_DELETE: 964 rt = rtable_lookup(tableid, info->rti_info[RTAX_DST], 965 info->rti_info[RTAX_NETMASK], info->rti_info[RTAX_GATEWAY], 966 prio); 967 if (rt == NULL) { 968 error = ESRCH; 969 break; 970 } 971 972 /* 973 * If we got multipath routes, we require users to specify 974 * a matching gateway. 975 */ 976 if (ISSET(rt->rt_flags, RTF_MPATH) && 977 info->rti_info[RTAX_GATEWAY] == NULL) { 978 error = ESRCH; 979 break; 980 } 981 982 ifp = if_get(rt->rt_ifidx); 983 if (ifp == NULL) { 984 rtfree(rt); 985 rt = NULL; 986 error = ESRCH; 987 break; 988 } 989 990 /* 991 * Invalidate the cache of automagically created and 992 * referenced L2 entries to make sure that ``rt_gwroute'' 993 * pointer stays valid for other CPUs. 994 */ 995 if ((ISSET(rt->rt_flags, RTF_CACHED))) { 996 NET_LOCK(); 997 ifp->if_rtrequest(ifp, RTM_INVALIDATE, rt); 998 /* Reset the MTU of the gateway route. */ 999 rtable_walk(tableid, rt_key(rt)->sa_family, NULL, 1000 route_cleargateway, rt); 1001 NET_UNLOCK(); 1002 break; 1003 } 1004 1005 /* 1006 * Make sure that local routes are only modified by the 1007 * kernel. 1008 */ 1009 if (ISSET(rt->rt_flags, RTF_LOCAL|RTF_BROADCAST)) { 1010 error = EINVAL; 1011 break; 1012 } 1013 1014 rtfree(rt); 1015 rt = NULL; 1016 1017 NET_LOCK(); 1018 error = rtrequest_delete(info, prio, ifp, &rt, tableid); 1019 NET_UNLOCK(); 1020 break; 1021 case RTM_CHANGE: 1022 rt = rtable_lookup(tableid, info->rti_info[RTAX_DST], 1023 info->rti_info[RTAX_NETMASK], info->rti_info[RTAX_GATEWAY], 1024 prio); 1025 /* 1026 * If we got multipath routes, we require users to specify 1027 * a matching gateway. 1028 */ 1029 if ((rt != NULL) && ISSET(rt->rt_flags, RTF_MPATH) && 1030 (info->rti_info[RTAX_GATEWAY] == NULL)) { 1031 rtfree(rt); 1032 rt = NULL; 1033 } 1034 1035 /* 1036 * If RTAX_GATEWAY is the argument we're trying to 1037 * change, try to find a compatible route. 1038 */ 1039 if ((rt == NULL) && (info->rti_info[RTAX_GATEWAY] != NULL)) { 1040 rt = rtable_lookup(tableid, info->rti_info[RTAX_DST], 1041 info->rti_info[RTAX_NETMASK], NULL, prio); 1042 /* Ensure we don't pick a multipath one. */ 1043 if ((rt != NULL) && ISSET(rt->rt_flags, RTF_MPATH)) { 1044 rtfree(rt); 1045 rt = NULL; 1046 } 1047 } 1048 1049 if (rt == NULL) { 1050 error = ESRCH; 1051 break; 1052 } 1053 1054 /* 1055 * Make sure that local routes are only modified by the 1056 * kernel. 1057 */ 1058 if (ISSET(rt->rt_flags, RTF_LOCAL|RTF_BROADCAST)) { 1059 error = EINVAL; 1060 break; 1061 } 1062 1063 ifp = if_get(rt->rt_ifidx); 1064 if (ifp == NULL) { 1065 rtfree(rt); 1066 rt = NULL; 1067 error = ESRCH; 1068 break; 1069 } 1070 1071 /* 1072 * RTM_CHANGE needs a perfect match. 1073 */ 1074 plen = rtable_satoplen(info->rti_info[RTAX_DST]->sa_family, 1075 info->rti_info[RTAX_NETMASK]); 1076 if (rt_plen(rt) != plen) { 1077 error = ESRCH; 1078 break; 1079 } 1080 1081 if (info->rti_info[RTAX_GATEWAY] != NULL) 1082 if (rt->rt_gateway == NULL || 1083 bcmp(rt->rt_gateway, 1084 info->rti_info[RTAX_GATEWAY], 1085 info->rti_info[RTAX_GATEWAY]->sa_len)) { 1086 newgate = 1; 1087 } 1088 /* 1089 * Check reachable gateway before changing the route. 1090 * New gateway could require new ifaddr, ifp; 1091 * flags may also be different; ifp may be specified 1092 * by ll sockaddr when protocol address is ambiguous. 1093 */ 1094 if (newgate || info->rti_info[RTAX_IFP] != NULL || 1095 info->rti_info[RTAX_IFA] != NULL) { 1096 struct ifaddr *ifa = NULL; 1097 1098 NET_LOCK(); 1099 if ((error = rtm_getifa(info, tableid)) != 0) { 1100 NET_UNLOCK(); 1101 break; 1102 } 1103 ifa = info->rti_ifa; 1104 if (rt->rt_ifa != ifa) { 1105 ifp->if_rtrequest(ifp, RTM_DELETE, rt); 1106 ifafree(rt->rt_ifa); 1107 1108 rt->rt_ifa = ifaref(ifa); 1109 rt->rt_ifidx = ifa->ifa_ifp->if_index; 1110 /* recheck link state after ifp change */ 1111 rt_if_linkstate_change(rt, ifa->ifa_ifp, 1112 tableid); 1113 } 1114 NET_UNLOCK(); 1115 } 1116 change: 1117 if (info->rti_info[RTAX_GATEWAY] != NULL) { 1118 /* When updating the gateway, make sure it is valid. */ 1119 if (!newgate && rt->rt_gateway->sa_family != 1120 info->rti_info[RTAX_GATEWAY]->sa_family) { 1121 error = EINVAL; 1122 break; 1123 } 1124 1125 NET_LOCK(); 1126 error = rt_setgate(rt, 1127 info->rti_info[RTAX_GATEWAY], tableid); 1128 NET_UNLOCK(); 1129 if (error) 1130 break; 1131 } 1132 #ifdef MPLS 1133 if (rtm->rtm_flags & RTF_MPLS) { 1134 NET_LOCK(); 1135 error = rt_mpls_set(rt, 1136 info->rti_info[RTAX_SRC], info->rti_mpls); 1137 NET_UNLOCK(); 1138 if (error) 1139 break; 1140 } else if (newgate || (rtm->rtm_fmask & RTF_MPLS)) { 1141 NET_LOCK(); 1142 /* if gateway changed remove MPLS information */ 1143 rt_mpls_clear(rt); 1144 NET_UNLOCK(); 1145 } 1146 #endif 1147 1148 #ifdef BFD 1149 if (ISSET(rtm->rtm_flags, RTF_BFD)) { 1150 KERNEL_LOCK(); 1151 error = bfdset(rt); 1152 KERNEL_UNLOCK(); 1153 if (error) 1154 break; 1155 } else if (!ISSET(rtm->rtm_flags, RTF_BFD) && 1156 ISSET(rtm->rtm_fmask, RTF_BFD)) { 1157 KERNEL_LOCK(); 1158 bfdclear(rt); 1159 KERNEL_UNLOCK(); 1160 } 1161 #endif 1162 1163 NET_LOCK(); 1164 /* Hack to allow some flags to be toggled */ 1165 if (rtm->rtm_fmask) { 1166 /* MPLS flag it is set by rt_mpls_set() */ 1167 rtm->rtm_fmask &= ~RTF_MPLS; 1168 rtm->rtm_flags &= ~RTF_MPLS; 1169 rt->rt_flags = 1170 (rt->rt_flags & ~rtm->rtm_fmask) | 1171 (rtm->rtm_flags & rtm->rtm_fmask); 1172 } 1173 rtm_setmetrics(rtm->rtm_inits, &rtm->rtm_rmx, &rt->rt_rmx); 1174 1175 ifp->if_rtrequest(ifp, RTM_ADD, rt); 1176 1177 if (info->rti_info[RTAX_LABEL] != NULL) { 1178 char *rtlabel = ((struct sockaddr_rtlabel *) 1179 info->rti_info[RTAX_LABEL])->sr_label; 1180 rtlabel_unref(rt->rt_labelid); 1181 rt->rt_labelid = rtlabel_name2id(rtlabel); 1182 } 1183 if_group_routechange(info->rti_info[RTAX_DST], 1184 info->rti_info[RTAX_NETMASK]); 1185 rt->rt_locks &= ~(rtm->rtm_inits); 1186 rt->rt_locks |= (rtm->rtm_inits & rtm->rtm_rmx.rmx_locks); 1187 NET_UNLOCK(); 1188 break; 1189 case RTM_GET: 1190 rt = rtable_lookup(tableid, info->rti_info[RTAX_DST], 1191 info->rti_info[RTAX_NETMASK], info->rti_info[RTAX_GATEWAY], 1192 prio); 1193 if (rt == NULL) 1194 error = ESRCH; 1195 break; 1196 } 1197 1198 if_put(ifp); 1199 *prt = rt; 1200 return (error); 1201 } 1202 1203 struct ifaddr * 1204 ifa_ifwithroute(int flags, struct sockaddr *dst, struct sockaddr *gateway, 1205 unsigned int rtableid) 1206 { 1207 struct ifaddr *ifa; 1208 1209 if ((flags & RTF_GATEWAY) == 0) { 1210 /* 1211 * If we are adding a route to an interface, 1212 * and the interface is a pt to pt link 1213 * we should search for the destination 1214 * as our clue to the interface. Otherwise 1215 * we can use the local address. 1216 */ 1217 ifa = NULL; 1218 if (flags & RTF_HOST) 1219 ifa = ifa_ifwithdstaddr(dst, rtableid); 1220 if (ifa == NULL) 1221 ifa = ifa_ifwithaddr(gateway, rtableid); 1222 } else { 1223 /* 1224 * If we are adding a route to a remote net 1225 * or host, the gateway may still be on the 1226 * other end of a pt to pt link. 1227 */ 1228 ifa = ifa_ifwithdstaddr(gateway, rtableid); 1229 } 1230 if (ifa == NULL) { 1231 if (gateway->sa_family == AF_LINK) { 1232 struct sockaddr_dl *sdl = satosdl(gateway); 1233 struct ifnet *ifp = if_get(sdl->sdl_index); 1234 1235 if (ifp != NULL) 1236 ifa = ifaof_ifpforaddr(dst, ifp); 1237 if_put(ifp); 1238 } else { 1239 struct rtentry *rt; 1240 1241 rt = rtalloc(gateway, RT_RESOLVE, rtable_l2(rtableid)); 1242 if (rt != NULL) 1243 ifa = rt->rt_ifa; 1244 rtfree(rt); 1245 } 1246 } 1247 if (ifa == NULL) 1248 return (NULL); 1249 if (ifa->ifa_addr->sa_family != dst->sa_family) { 1250 struct ifaddr *oifa = ifa; 1251 ifa = ifaof_ifpforaddr(dst, ifa->ifa_ifp); 1252 if (ifa == NULL) 1253 ifa = oifa; 1254 } 1255 return (ifa); 1256 } 1257 1258 int 1259 rtm_getifa(struct rt_addrinfo *info, unsigned int rtid) 1260 { 1261 struct ifnet *ifp = NULL; 1262 1263 /* 1264 * The "returned" `ifa' is guaranteed to be alive only if 1265 * the NET_LOCK() is held. 1266 */ 1267 NET_ASSERT_LOCKED(); 1268 1269 /* 1270 * ifp may be specified by sockaddr_dl when protocol address 1271 * is ambiguous 1272 */ 1273 if (info->rti_info[RTAX_IFP] != NULL) { 1274 struct sockaddr_dl *sdl; 1275 1276 sdl = satosdl(info->rti_info[RTAX_IFP]); 1277 ifp = if_get(sdl->sdl_index); 1278 } 1279 1280 #ifdef IPSEC 1281 /* 1282 * If the destination is a PF_KEY address, we'll look 1283 * for the existence of a encap interface number or address 1284 * in the options list of the gateway. By default, we'll return 1285 * enc0. 1286 */ 1287 if (info->rti_info[RTAX_DST] && 1288 info->rti_info[RTAX_DST]->sa_family == PF_KEY) 1289 info->rti_ifa = enc_getifa(rtid, 0); 1290 #endif 1291 1292 if (info->rti_ifa == NULL && info->rti_info[RTAX_IFA] != NULL) 1293 info->rti_ifa = ifa_ifwithaddr(info->rti_info[RTAX_IFA], rtid); 1294 1295 if (info->rti_ifa == NULL) { 1296 struct sockaddr *sa; 1297 1298 if ((sa = info->rti_info[RTAX_IFA]) == NULL) 1299 if ((sa = info->rti_info[RTAX_GATEWAY]) == NULL) 1300 sa = info->rti_info[RTAX_DST]; 1301 1302 if (sa != NULL && ifp != NULL) 1303 info->rti_ifa = ifaof_ifpforaddr(sa, ifp); 1304 else if (info->rti_info[RTAX_DST] != NULL && 1305 info->rti_info[RTAX_GATEWAY] != NULL) 1306 info->rti_ifa = ifa_ifwithroute(info->rti_flags, 1307 info->rti_info[RTAX_DST], 1308 info->rti_info[RTAX_GATEWAY], 1309 rtid); 1310 else if (sa != NULL) 1311 info->rti_ifa = ifa_ifwithroute(info->rti_flags, 1312 sa, sa, rtid); 1313 } 1314 1315 if_put(ifp); 1316 1317 if (info->rti_ifa == NULL) 1318 return (ENETUNREACH); 1319 1320 return (0); 1321 } 1322 1323 int 1324 route_cleargateway(struct rtentry *rt, void *arg, unsigned int rtableid) 1325 { 1326 struct rtentry *nhrt = arg; 1327 1328 if (ISSET(rt->rt_flags, RTF_GATEWAY) && rt->rt_gwroute == nhrt && 1329 !ISSET(rt->rt_locks, RTV_MTU)) 1330 rt->rt_mtu = 0; 1331 1332 return (0); 1333 } 1334 1335 /* 1336 * Check if the user request to insert an ARP entry does not conflict 1337 * with existing ones. 1338 * 1339 * Only two entries are allowed for a given IP address: a private one 1340 * (priv) and a public one (pub). 1341 */ 1342 int 1343 route_arp_conflict(struct rtentry *rt, struct rt_addrinfo *info) 1344 { 1345 int proxy = (info->rti_flags & RTF_ANNOUNCE); 1346 1347 if ((info->rti_flags & RTF_LLINFO) == 0 || 1348 (info->rti_info[RTAX_DST]->sa_family != AF_INET)) 1349 return (0); 1350 1351 if (rt == NULL || !ISSET(rt->rt_flags, RTF_LLINFO)) 1352 return (0); 1353 1354 /* If the entry is cached, it can be updated. */ 1355 if (ISSET(rt->rt_flags, RTF_CACHED)) 1356 return (0); 1357 1358 /* 1359 * Same destination, not cached and both "priv" or "pub" conflict. 1360 * If a second entry exists, it always conflict. 1361 */ 1362 if ((ISSET(rt->rt_flags, RTF_ANNOUNCE) == proxy) || 1363 ISSET(rt->rt_flags, RTF_MPATH)) 1364 return (EEXIST); 1365 1366 /* No conflict but an entry exist so we need to force mpath. */ 1367 info->rti_flags |= RTF_MPATH; 1368 return (0); 1369 } 1370 1371 void 1372 rtm_setmetrics(u_long which, const struct rt_metrics *in, 1373 struct rt_kmetrics *out) 1374 { 1375 int64_t expire; 1376 1377 if (which & RTV_MTU) 1378 out->rmx_mtu = in->rmx_mtu; 1379 if (which & RTV_EXPIRE) { 1380 expire = in->rmx_expire; 1381 if (expire != 0) { 1382 expire -= gettime(); 1383 expire += getuptime(); 1384 } 1385 1386 out->rmx_expire = expire; 1387 } 1388 } 1389 1390 void 1391 rtm_getmetrics(const struct rtentry *rt, struct rt_metrics *out) 1392 { 1393 const struct rt_kmetrics *in = &rt->rt_rmx; 1394 int64_t expire; 1395 1396 expire = in->rmx_expire; 1397 if (expire == 0) 1398 expire = rt_timer_get_expire(rt); 1399 if (expire != 0) { 1400 expire -= getuptime(); 1401 expire += gettime(); 1402 } 1403 1404 bzero(out, sizeof(*out)); 1405 out->rmx_locks = in->rmx_locks; 1406 out->rmx_mtu = in->rmx_mtu; 1407 out->rmx_expire = expire; 1408 out->rmx_pksent = in->rmx_pksent; 1409 } 1410 1411 #define ROUNDUP(a) \ 1412 ((a) > 0 ? (1 + (((a) - 1) | (sizeof(long) - 1))) : sizeof(long)) 1413 #define ADVANCE(x, n) (x += ROUNDUP((n)->sa_len)) 1414 1415 int 1416 rtm_xaddrs(caddr_t cp, caddr_t cplim, struct rt_addrinfo *rtinfo) 1417 { 1418 struct sockaddr *sa; 1419 int i; 1420 1421 /* 1422 * Parse address bits, split address storage in chunks, and 1423 * set info pointers. Use sa_len for traversing the memory 1424 * and check that we stay within in the limit. 1425 */ 1426 bzero(rtinfo->rti_info, sizeof(rtinfo->rti_info)); 1427 for (i = 0; i < sizeof(rtinfo->rti_addrs) * 8; i++) { 1428 if ((rtinfo->rti_addrs & (1U << i)) == 0) 1429 continue; 1430 if (i >= RTAX_MAX || cp + sizeof(socklen_t) > cplim) 1431 return (EINVAL); 1432 sa = (struct sockaddr *)cp; 1433 if (cp + sa->sa_len > cplim) 1434 return (EINVAL); 1435 rtinfo->rti_info[i] = sa; 1436 ADVANCE(cp, sa); 1437 } 1438 /* 1439 * Check that the address family is suitable for the route address 1440 * type. Check that each address has a size that fits its family 1441 * and its length is within the size. Strings within addresses must 1442 * be NUL terminated. 1443 */ 1444 for (i = 0; i < RTAX_MAX; i++) { 1445 size_t len, maxlen, size; 1446 1447 sa = rtinfo->rti_info[i]; 1448 if (sa == NULL) 1449 continue; 1450 maxlen = size = 0; 1451 switch (i) { 1452 case RTAX_DST: 1453 case RTAX_GATEWAY: 1454 case RTAX_SRC: 1455 switch (sa->sa_family) { 1456 case AF_INET: 1457 size = sizeof(struct sockaddr_in); 1458 break; 1459 case AF_LINK: 1460 size = sizeof(struct sockaddr_dl); 1461 break; 1462 #ifdef INET6 1463 case AF_INET6: 1464 size = sizeof(struct sockaddr_in6); 1465 break; 1466 #endif 1467 #ifdef MPLS 1468 case AF_MPLS: 1469 size = sizeof(struct sockaddr_mpls); 1470 break; 1471 #endif 1472 } 1473 break; 1474 case RTAX_IFP: 1475 if (sa->sa_family != AF_LINK) 1476 return (EAFNOSUPPORT); 1477 /* 1478 * XXX Should be sizeof(struct sockaddr_dl), but 1479 * route(8) has a bug and provides less memory. 1480 * arp(8) has another bug and uses sizeof pointer. 1481 */ 1482 size = 4; 1483 break; 1484 case RTAX_IFA: 1485 switch (sa->sa_family) { 1486 case AF_INET: 1487 size = sizeof(struct sockaddr_in); 1488 break; 1489 #ifdef INET6 1490 case AF_INET6: 1491 size = sizeof(struct sockaddr_in6); 1492 break; 1493 #endif 1494 default: 1495 return (EAFNOSUPPORT); 1496 } 1497 break; 1498 case RTAX_LABEL: 1499 sa->sa_family = AF_UNSPEC; 1500 maxlen = RTLABEL_LEN; 1501 size = sizeof(struct sockaddr_rtlabel); 1502 break; 1503 #ifdef BFD 1504 case RTAX_BFD: 1505 sa->sa_family = AF_UNSPEC; 1506 size = sizeof(struct sockaddr_bfd); 1507 break; 1508 #endif 1509 case RTAX_DNS: 1510 /* more validation in rtm_validate_proposal */ 1511 if (sa->sa_len > sizeof(struct sockaddr_rtdns)) 1512 return (EINVAL); 1513 if (sa->sa_len < offsetof(struct sockaddr_rtdns, 1514 sr_dns)) 1515 return (EINVAL); 1516 switch (sa->sa_family) { 1517 case AF_INET: 1518 #ifdef INET6 1519 case AF_INET6: 1520 #endif 1521 break; 1522 default: 1523 return (EAFNOSUPPORT); 1524 } 1525 break; 1526 case RTAX_STATIC: 1527 sa->sa_family = AF_UNSPEC; 1528 maxlen = RTSTATIC_LEN; 1529 size = sizeof(struct sockaddr_rtstatic); 1530 break; 1531 case RTAX_SEARCH: 1532 sa->sa_family = AF_UNSPEC; 1533 maxlen = RTSEARCH_LEN; 1534 size = sizeof(struct sockaddr_rtsearch); 1535 break; 1536 } 1537 if (size) { 1538 /* memory for the full struct must be provided */ 1539 if (sa->sa_len < size) 1540 return (EINVAL); 1541 } 1542 if (maxlen) { 1543 /* this should not happen */ 1544 if (2 + maxlen > size) 1545 return (EINVAL); 1546 /* strings must be NUL terminated within the struct */ 1547 len = strnlen(sa->sa_data, maxlen); 1548 if (len >= maxlen || 2 + len >= sa->sa_len) 1549 return (EINVAL); 1550 break; 1551 } 1552 } 1553 return (0); 1554 } 1555 1556 struct mbuf * 1557 rtm_msg1(int type, struct rt_addrinfo *rtinfo) 1558 { 1559 struct rt_msghdr *rtm; 1560 struct mbuf *m; 1561 int i; 1562 struct sockaddr *sa; 1563 int len, dlen, hlen; 1564 1565 switch (type) { 1566 case RTM_DELADDR: 1567 case RTM_NEWADDR: 1568 hlen = sizeof(struct ifa_msghdr); 1569 break; 1570 case RTM_IFINFO: 1571 hlen = sizeof(struct if_msghdr); 1572 break; 1573 case RTM_IFANNOUNCE: 1574 hlen = sizeof(struct if_announcemsghdr); 1575 break; 1576 #ifdef BFD 1577 case RTM_BFD: 1578 hlen = sizeof(struct bfd_msghdr); 1579 break; 1580 #endif 1581 case RTM_80211INFO: 1582 hlen = sizeof(struct if_ieee80211_msghdr); 1583 break; 1584 default: 1585 hlen = sizeof(struct rt_msghdr); 1586 break; 1587 } 1588 len = hlen; 1589 for (i = 0; i < RTAX_MAX; i++) { 1590 if (rtinfo == NULL || (sa = rtinfo->rti_info[i]) == NULL) 1591 continue; 1592 len += ROUNDUP(sa->sa_len); 1593 } 1594 if (len > MCLBYTES) 1595 panic("rtm_msg1"); 1596 m = m_gethdr(M_DONTWAIT, MT_DATA); 1597 if (m && len > MHLEN) { 1598 MCLGET(m, M_DONTWAIT); 1599 if ((m->m_flags & M_EXT) == 0) { 1600 m_free(m); 1601 m = NULL; 1602 } 1603 } 1604 if (m == NULL) 1605 return (m); 1606 m->m_pkthdr.len = m->m_len = len; 1607 m->m_pkthdr.ph_ifidx = 0; 1608 rtm = mtod(m, struct rt_msghdr *); 1609 bzero(rtm, len); 1610 len = hlen; 1611 for (i = 0; i < RTAX_MAX; i++) { 1612 if (rtinfo == NULL || (sa = rtinfo->rti_info[i]) == NULL) 1613 continue; 1614 rtinfo->rti_addrs |= (1U << i); 1615 dlen = ROUNDUP(sa->sa_len); 1616 if (m_copyback(m, len, sa->sa_len, sa, M_NOWAIT)) { 1617 m_freem(m); 1618 return (NULL); 1619 } 1620 len += dlen; 1621 } 1622 rtm->rtm_msglen = len; 1623 rtm->rtm_hdrlen = hlen; 1624 rtm->rtm_version = RTM_VERSION; 1625 rtm->rtm_type = type; 1626 return (m); 1627 } 1628 1629 int 1630 rtm_msg2(int type, int vers, struct rt_addrinfo *rtinfo, caddr_t cp, 1631 struct walkarg *w) 1632 { 1633 int i; 1634 int len, dlen, hlen, second_time = 0; 1635 caddr_t cp0; 1636 1637 rtinfo->rti_addrs = 0; 1638 again: 1639 switch (type) { 1640 case RTM_DELADDR: 1641 case RTM_NEWADDR: 1642 len = sizeof(struct ifa_msghdr); 1643 break; 1644 case RTM_IFINFO: 1645 len = sizeof(struct if_msghdr); 1646 break; 1647 default: 1648 len = sizeof(struct rt_msghdr); 1649 break; 1650 } 1651 hlen = len; 1652 if ((cp0 = cp) != NULL) 1653 cp += len; 1654 for (i = 0; i < RTAX_MAX; i++) { 1655 struct sockaddr *sa; 1656 1657 if ((sa = rtinfo->rti_info[i]) == NULL) 1658 continue; 1659 rtinfo->rti_addrs |= (1U << i); 1660 dlen = ROUNDUP(sa->sa_len); 1661 if (cp) { 1662 bcopy(sa, cp, sa->sa_len); 1663 bzero(cp + sa->sa_len, dlen - sa->sa_len); 1664 cp += dlen; 1665 } 1666 len += dlen; 1667 } 1668 /* align message length to the next natural boundary */ 1669 len = ALIGN(len); 1670 if (cp == 0 && w != NULL && !second_time) { 1671 w->w_needed += len; 1672 if (w->w_needed <= w->w_given && w->w_where) { 1673 if (w->w_tmemsize < len) { 1674 free(w->w_tmem, M_RTABLE, w->w_tmemsize); 1675 w->w_tmem = malloc(len, M_RTABLE, 1676 M_NOWAIT | M_ZERO); 1677 if (w->w_tmem) 1678 w->w_tmemsize = len; 1679 } 1680 if (w->w_tmem) { 1681 cp = w->w_tmem; 1682 second_time = 1; 1683 goto again; 1684 } else 1685 w->w_where = 0; 1686 } 1687 } 1688 if (cp && w) /* clear the message header */ 1689 bzero(cp0, hlen); 1690 1691 if (cp) { 1692 struct rt_msghdr *rtm = (struct rt_msghdr *)cp0; 1693 1694 rtm->rtm_version = RTM_VERSION; 1695 rtm->rtm_type = type; 1696 rtm->rtm_msglen = len; 1697 rtm->rtm_hdrlen = hlen; 1698 } 1699 return (len); 1700 } 1701 1702 void 1703 rtm_send(struct rtentry *rt, int cmd, int error, unsigned int rtableid) 1704 { 1705 struct rt_addrinfo info; 1706 struct ifnet *ifp; 1707 struct sockaddr_rtlabel sa_rl; 1708 struct sockaddr_in6 sa_mask; 1709 1710 memset(&info, 0, sizeof(info)); 1711 info.rti_info[RTAX_DST] = rt_key(rt); 1712 info.rti_info[RTAX_GATEWAY] = rt->rt_gateway; 1713 if (!ISSET(rt->rt_flags, RTF_HOST)) 1714 info.rti_info[RTAX_NETMASK] = rt_plen2mask(rt, &sa_mask); 1715 info.rti_info[RTAX_LABEL] = rtlabel_id2sa(rt->rt_labelid, &sa_rl); 1716 ifp = if_get(rt->rt_ifidx); 1717 if (ifp != NULL) { 1718 info.rti_info[RTAX_IFP] = sdltosa(ifp->if_sadl); 1719 info.rti_info[RTAX_IFA] = rtable_getsource(rtableid, 1720 info.rti_info[RTAX_DST]->sa_family); 1721 if (info.rti_info[RTAX_IFA] == NULL) 1722 info.rti_info[RTAX_IFA] = rt->rt_ifa->ifa_addr; 1723 } 1724 1725 rtm_miss(cmd, &info, rt->rt_flags, rt->rt_priority, rt->rt_ifidx, error, 1726 rtableid); 1727 if_put(ifp); 1728 } 1729 1730 /* 1731 * This routine is called to generate a message from the routing 1732 * socket indicating that a redirect has occurred, a routing lookup 1733 * has failed, or that a protocol has detected timeouts to a particular 1734 * destination. 1735 */ 1736 void 1737 rtm_miss(int type, struct rt_addrinfo *rtinfo, int flags, uint8_t prio, 1738 u_int ifidx, int error, u_int tableid) 1739 { 1740 struct rt_msghdr *rtm; 1741 struct mbuf *m; 1742 struct sockaddr *sa = rtinfo->rti_info[RTAX_DST]; 1743 1744 if (rtptable.rtp_count == 0) 1745 return; 1746 m = rtm_msg1(type, rtinfo); 1747 if (m == NULL) 1748 return; 1749 rtm = mtod(m, struct rt_msghdr *); 1750 rtm->rtm_flags = RTF_DONE | flags; 1751 rtm->rtm_priority = prio; 1752 rtm->rtm_errno = error; 1753 rtm->rtm_tableid = tableid; 1754 rtm->rtm_addrs = rtinfo->rti_addrs; 1755 rtm->rtm_index = ifidx; 1756 route_input(m, NULL, sa ? sa->sa_family : AF_UNSPEC); 1757 } 1758 1759 /* 1760 * This routine is called to generate a message from the routing 1761 * socket indicating that the status of a network interface has changed. 1762 */ 1763 void 1764 rtm_ifchg(struct ifnet *ifp) 1765 { 1766 struct rt_addrinfo info; 1767 struct if_msghdr *ifm; 1768 struct mbuf *m; 1769 1770 if (rtptable.rtp_count == 0) 1771 return; 1772 memset(&info, 0, sizeof(info)); 1773 info.rti_info[RTAX_IFP] = sdltosa(ifp->if_sadl); 1774 m = rtm_msg1(RTM_IFINFO, &info); 1775 if (m == NULL) 1776 return; 1777 ifm = mtod(m, struct if_msghdr *); 1778 ifm->ifm_index = ifp->if_index; 1779 ifm->ifm_tableid = ifp->if_rdomain; 1780 ifm->ifm_flags = ifp->if_flags; 1781 ifm->ifm_xflags = ifp->if_xflags; 1782 if_getdata(ifp, &ifm->ifm_data); 1783 ifm->ifm_addrs = info.rti_addrs; 1784 route_input(m, NULL, AF_UNSPEC); 1785 } 1786 1787 /* 1788 * This is called to generate messages from the routing socket 1789 * indicating a network interface has had addresses associated with it. 1790 * if we ever reverse the logic and replace messages TO the routing 1791 * socket indicate a request to configure interfaces, then it will 1792 * be unnecessary as the routing socket will automatically generate 1793 * copies of it. 1794 */ 1795 void 1796 rtm_addr(int cmd, struct ifaddr *ifa) 1797 { 1798 struct ifnet *ifp = ifa->ifa_ifp; 1799 struct mbuf *m; 1800 struct rt_addrinfo info; 1801 struct ifa_msghdr *ifam; 1802 1803 if (rtptable.rtp_count == 0) 1804 return; 1805 1806 memset(&info, 0, sizeof(info)); 1807 info.rti_info[RTAX_IFA] = ifa->ifa_addr; 1808 info.rti_info[RTAX_IFP] = sdltosa(ifp->if_sadl); 1809 info.rti_info[RTAX_NETMASK] = ifa->ifa_netmask; 1810 info.rti_info[RTAX_BRD] = ifa->ifa_dstaddr; 1811 if ((m = rtm_msg1(cmd, &info)) == NULL) 1812 return; 1813 ifam = mtod(m, struct ifa_msghdr *); 1814 ifam->ifam_index = ifp->if_index; 1815 ifam->ifam_metric = ifa->ifa_metric; 1816 ifam->ifam_flags = ifa->ifa_flags; 1817 ifam->ifam_addrs = info.rti_addrs; 1818 ifam->ifam_tableid = ifp->if_rdomain; 1819 1820 route_input(m, NULL, 1821 ifa->ifa_addr ? ifa->ifa_addr->sa_family : AF_UNSPEC); 1822 } 1823 1824 /* 1825 * This is called to generate routing socket messages indicating 1826 * network interface arrival and departure. 1827 */ 1828 void 1829 rtm_ifannounce(struct ifnet *ifp, int what) 1830 { 1831 struct if_announcemsghdr *ifan; 1832 struct mbuf *m; 1833 1834 if (rtptable.rtp_count == 0) 1835 return; 1836 m = rtm_msg1(RTM_IFANNOUNCE, NULL); 1837 if (m == NULL) 1838 return; 1839 ifan = mtod(m, struct if_announcemsghdr *); 1840 ifan->ifan_index = ifp->if_index; 1841 strlcpy(ifan->ifan_name, ifp->if_xname, sizeof(ifan->ifan_name)); 1842 ifan->ifan_what = what; 1843 route_input(m, NULL, AF_UNSPEC); 1844 } 1845 1846 #ifdef BFD 1847 /* 1848 * This is used to generate routing socket messages indicating 1849 * the state of a BFD session. 1850 */ 1851 void 1852 rtm_bfd(struct bfd_config *bfd) 1853 { 1854 struct bfd_msghdr *bfdm; 1855 struct sockaddr_bfd sa_bfd; 1856 struct mbuf *m; 1857 struct rt_addrinfo info; 1858 1859 if (rtptable.rtp_count == 0) 1860 return; 1861 memset(&info, 0, sizeof(info)); 1862 info.rti_info[RTAX_DST] = rt_key(bfd->bc_rt); 1863 info.rti_info[RTAX_IFA] = bfd->bc_rt->rt_ifa->ifa_addr; 1864 1865 m = rtm_msg1(RTM_BFD, &info); 1866 if (m == NULL) 1867 return; 1868 bfdm = mtod(m, struct bfd_msghdr *); 1869 bfdm->bm_addrs = info.rti_addrs; 1870 1871 KERNEL_ASSERT_LOCKED(); 1872 bfd2sa(bfd->bc_rt, &sa_bfd); 1873 memcpy(&bfdm->bm_sa, &sa_bfd, sizeof(sa_bfd)); 1874 1875 route_input(m, NULL, info.rti_info[RTAX_DST]->sa_family); 1876 } 1877 #endif /* BFD */ 1878 1879 /* 1880 * This is used to generate routing socket messages indicating 1881 * the state of an ieee80211 interface. 1882 */ 1883 void 1884 rtm_80211info(struct ifnet *ifp, struct if_ieee80211_data *ifie) 1885 { 1886 struct if_ieee80211_msghdr *ifim; 1887 struct mbuf *m; 1888 1889 if (rtptable.rtp_count == 0) 1890 return; 1891 m = rtm_msg1(RTM_80211INFO, NULL); 1892 if (m == NULL) 1893 return; 1894 ifim = mtod(m, struct if_ieee80211_msghdr *); 1895 ifim->ifim_index = ifp->if_index; 1896 ifim->ifim_tableid = ifp->if_rdomain; 1897 1898 memcpy(&ifim->ifim_ifie, ifie, sizeof(ifim->ifim_ifie)); 1899 route_input(m, NULL, AF_UNSPEC); 1900 } 1901 1902 /* 1903 * This is used to generate routing socket messages indicating 1904 * the address selection proposal from an interface. 1905 */ 1906 void 1907 rtm_proposal(struct ifnet *ifp, struct rt_addrinfo *rtinfo, int flags, 1908 uint8_t prio) 1909 { 1910 struct rt_msghdr *rtm; 1911 struct mbuf *m; 1912 1913 m = rtm_msg1(RTM_PROPOSAL, rtinfo); 1914 if (m == NULL) 1915 return; 1916 rtm = mtod(m, struct rt_msghdr *); 1917 rtm->rtm_flags = RTF_DONE | flags; 1918 rtm->rtm_priority = prio; 1919 rtm->rtm_tableid = ifp->if_rdomain; 1920 rtm->rtm_index = ifp->if_index; 1921 rtm->rtm_addrs = rtinfo->rti_addrs; 1922 1923 route_input(m, NULL, rtinfo->rti_info[RTAX_DNS]->sa_family); 1924 } 1925 1926 /* 1927 * This is used in dumping the kernel table via sysctl(). 1928 */ 1929 int 1930 sysctl_dumpentry(struct rtentry *rt, void *v, unsigned int id) 1931 { 1932 struct walkarg *w = v; 1933 int error = 0, size; 1934 struct rt_addrinfo info; 1935 struct ifnet *ifp; 1936 #ifdef BFD 1937 struct sockaddr_bfd sa_bfd; 1938 #endif 1939 struct sockaddr_rtlabel sa_rl; 1940 struct sockaddr_in6 sa_mask; 1941 1942 if (w->w_op == NET_RT_FLAGS && !(rt->rt_flags & w->w_arg)) 1943 return 0; 1944 if (w->w_op == NET_RT_DUMP && w->w_arg) { 1945 u_int8_t prio = w->w_arg & RTP_MASK; 1946 if (w->w_arg < 0) { 1947 prio = (-w->w_arg) & RTP_MASK; 1948 /* Show all routes that are not this priority */ 1949 if (prio == (rt->rt_priority & RTP_MASK)) 1950 return 0; 1951 } else { 1952 if (prio != (rt->rt_priority & RTP_MASK) && 1953 prio != RTP_ANY) 1954 return 0; 1955 } 1956 } 1957 bzero(&info, sizeof(info)); 1958 info.rti_info[RTAX_DST] = rt_key(rt); 1959 info.rti_info[RTAX_GATEWAY] = rt->rt_gateway; 1960 info.rti_info[RTAX_NETMASK] = rt_plen2mask(rt, &sa_mask); 1961 ifp = if_get(rt->rt_ifidx); 1962 if (ifp != NULL) { 1963 info.rti_info[RTAX_IFP] = sdltosa(ifp->if_sadl); 1964 info.rti_info[RTAX_IFA] = 1965 rtable_getsource(id, info.rti_info[RTAX_DST]->sa_family); 1966 if (info.rti_info[RTAX_IFA] == NULL) 1967 info.rti_info[RTAX_IFA] = rt->rt_ifa->ifa_addr; 1968 if (ifp->if_flags & IFF_POINTOPOINT) 1969 info.rti_info[RTAX_BRD] = rt->rt_ifa->ifa_dstaddr; 1970 } 1971 if_put(ifp); 1972 info.rti_info[RTAX_LABEL] = rtlabel_id2sa(rt->rt_labelid, &sa_rl); 1973 #ifdef BFD 1974 if (rt->rt_flags & RTF_BFD) { 1975 KERNEL_ASSERT_LOCKED(); 1976 info.rti_info[RTAX_BFD] = bfd2sa(rt, &sa_bfd); 1977 } 1978 #endif 1979 #ifdef MPLS 1980 if (rt->rt_flags & RTF_MPLS) { 1981 struct sockaddr_mpls sa_mpls; 1982 1983 bzero(&sa_mpls, sizeof(sa_mpls)); 1984 sa_mpls.smpls_family = AF_MPLS; 1985 sa_mpls.smpls_len = sizeof(sa_mpls); 1986 sa_mpls.smpls_label = ((struct rt_mpls *) 1987 rt->rt_llinfo)->mpls_label; 1988 info.rti_info[RTAX_SRC] = (struct sockaddr *)&sa_mpls; 1989 info.rti_mpls = ((struct rt_mpls *) 1990 rt->rt_llinfo)->mpls_operation; 1991 } 1992 #endif 1993 1994 size = rtm_msg2(RTM_GET, RTM_VERSION, &info, NULL, w); 1995 if (w->w_where && w->w_tmem && w->w_needed <= w->w_given) { 1996 struct rt_msghdr *rtm = (struct rt_msghdr *)w->w_tmem; 1997 1998 rtm->rtm_pid = curproc->p_p->ps_pid; 1999 rtm->rtm_flags = RTF_DONE | rt->rt_flags; 2000 rtm->rtm_priority = rt->rt_priority & RTP_MASK; 2001 rtm_getmetrics(rt, &rtm->rtm_rmx); 2002 /* Do not account the routing table's reference. */ 2003 rtm->rtm_rmx.rmx_refcnt = refcnt_read(&rt->rt_refcnt) - 1; 2004 rtm->rtm_index = rt->rt_ifidx; 2005 rtm->rtm_addrs = info.rti_addrs; 2006 rtm->rtm_tableid = id; 2007 #ifdef MPLS 2008 rtm->rtm_mpls = info.rti_mpls; 2009 #endif 2010 if ((error = copyout(rtm, w->w_where, size)) != 0) 2011 w->w_where = NULL; 2012 else 2013 w->w_where += size; 2014 } 2015 return (error); 2016 } 2017 2018 int 2019 sysctl_iflist(int af, struct walkarg *w) 2020 { 2021 struct ifnet *ifp; 2022 struct ifaddr *ifa; 2023 struct rt_addrinfo info; 2024 int len, error = 0; 2025 2026 bzero(&info, sizeof(info)); 2027 TAILQ_FOREACH(ifp, &ifnetlist, if_list) { 2028 if (w->w_arg && w->w_arg != ifp->if_index) 2029 continue; 2030 /* Copy the link-layer address first */ 2031 info.rti_info[RTAX_IFP] = sdltosa(ifp->if_sadl); 2032 len = rtm_msg2(RTM_IFINFO, RTM_VERSION, &info, 0, w); 2033 if (w->w_where && w->w_tmem && w->w_needed <= w->w_given) { 2034 struct if_msghdr *ifm; 2035 2036 ifm = (struct if_msghdr *)w->w_tmem; 2037 ifm->ifm_index = ifp->if_index; 2038 ifm->ifm_tableid = ifp->if_rdomain; 2039 ifm->ifm_flags = ifp->if_flags; 2040 if_getdata(ifp, &ifm->ifm_data); 2041 ifm->ifm_addrs = info.rti_addrs; 2042 error = copyout(ifm, w->w_where, len); 2043 if (error) 2044 return (error); 2045 w->w_where += len; 2046 } 2047 info.rti_info[RTAX_IFP] = NULL; 2048 TAILQ_FOREACH(ifa, &ifp->if_addrlist, ifa_list) { 2049 KASSERT(ifa->ifa_addr->sa_family != AF_LINK); 2050 if (af && af != ifa->ifa_addr->sa_family) 2051 continue; 2052 info.rti_info[RTAX_IFA] = ifa->ifa_addr; 2053 info.rti_info[RTAX_NETMASK] = ifa->ifa_netmask; 2054 info.rti_info[RTAX_BRD] = ifa->ifa_dstaddr; 2055 len = rtm_msg2(RTM_NEWADDR, RTM_VERSION, &info, 0, w); 2056 if (w->w_where && w->w_tmem && 2057 w->w_needed <= w->w_given) { 2058 struct ifa_msghdr *ifam; 2059 2060 ifam = (struct ifa_msghdr *)w->w_tmem; 2061 ifam->ifam_index = ifa->ifa_ifp->if_index; 2062 ifam->ifam_flags = ifa->ifa_flags; 2063 ifam->ifam_metric = ifa->ifa_metric; 2064 ifam->ifam_addrs = info.rti_addrs; 2065 error = copyout(w->w_tmem, w->w_where, len); 2066 if (error) 2067 return (error); 2068 w->w_where += len; 2069 } 2070 } 2071 info.rti_info[RTAX_IFA] = info.rti_info[RTAX_NETMASK] = 2072 info.rti_info[RTAX_BRD] = NULL; 2073 } 2074 return (0); 2075 } 2076 2077 int 2078 sysctl_ifnames(struct walkarg *w) 2079 { 2080 struct if_nameindex_msg ifn; 2081 struct ifnet *ifp; 2082 int error = 0; 2083 2084 /* XXX ignore tableid for now */ 2085 TAILQ_FOREACH(ifp, &ifnetlist, if_list) { 2086 if (w->w_arg && w->w_arg != ifp->if_index) 2087 continue; 2088 w->w_needed += sizeof(ifn); 2089 if (w->w_where && w->w_needed <= w->w_given) { 2090 2091 memset(&ifn, 0, sizeof(ifn)); 2092 ifn.if_index = ifp->if_index; 2093 strlcpy(ifn.if_name, ifp->if_xname, 2094 sizeof(ifn.if_name)); 2095 error = copyout(&ifn, w->w_where, sizeof(ifn)); 2096 if (error) 2097 return (error); 2098 w->w_where += sizeof(ifn); 2099 } 2100 } 2101 2102 return (0); 2103 } 2104 2105 int 2106 sysctl_source(int af, u_int tableid, struct walkarg *w) 2107 { 2108 struct sockaddr *sa; 2109 int size, error = 0; 2110 2111 sa = rtable_getsource(tableid, af); 2112 if (sa) { 2113 switch (sa->sa_family) { 2114 case AF_INET: 2115 size = sizeof(struct sockaddr_in); 2116 break; 2117 #ifdef INET6 2118 case AF_INET6: 2119 size = sizeof(struct sockaddr_in6); 2120 break; 2121 #endif 2122 default: 2123 return (0); 2124 } 2125 w->w_needed += size; 2126 if (w->w_where && w->w_needed <= w->w_given) { 2127 if ((error = copyout(sa, w->w_where, size))) 2128 return (error); 2129 w->w_where += size; 2130 } 2131 } 2132 return (0); 2133 } 2134 2135 int 2136 sysctl_rtable(int *name, u_int namelen, void *where, size_t *given, void *new, 2137 size_t newlen) 2138 { 2139 int i, error = EINVAL; 2140 u_char af; 2141 struct walkarg w; 2142 struct rt_tableinfo tableinfo; 2143 u_int tableid = 0; 2144 2145 if (new) 2146 return (EPERM); 2147 if (namelen < 3 || namelen > 4) 2148 return (EINVAL); 2149 af = name[0]; 2150 bzero(&w, sizeof(w)); 2151 w.w_where = where; 2152 w.w_given = *given; 2153 w.w_op = name[1]; 2154 w.w_arg = name[2]; 2155 2156 if (namelen == 4) { 2157 tableid = name[3]; 2158 if (!rtable_exists(tableid)) 2159 return (ENOENT); 2160 } else 2161 tableid = curproc->p_p->ps_rtableid; 2162 2163 switch (w.w_op) { 2164 case NET_RT_DUMP: 2165 case NET_RT_FLAGS: 2166 NET_LOCK_SHARED(); 2167 for (i = 1; i <= AF_MAX; i++) { 2168 if (af != 0 && af != i) 2169 continue; 2170 2171 error = rtable_walk(tableid, i, NULL, sysctl_dumpentry, 2172 &w); 2173 if (error == EAFNOSUPPORT) 2174 error = 0; 2175 if (error) 2176 break; 2177 } 2178 NET_UNLOCK_SHARED(); 2179 break; 2180 2181 case NET_RT_IFLIST: 2182 NET_LOCK_SHARED(); 2183 error = sysctl_iflist(af, &w); 2184 NET_UNLOCK_SHARED(); 2185 break; 2186 2187 case NET_RT_STATS: 2188 return (sysctl_rtable_rtstat(where, given, new)); 2189 case NET_RT_TABLE: 2190 tableid = w.w_arg; 2191 if (!rtable_exists(tableid)) 2192 return (ENOENT); 2193 memset(&tableinfo, 0, sizeof tableinfo); 2194 tableinfo.rti_tableid = tableid; 2195 tableinfo.rti_domainid = rtable_l2(tableid); 2196 error = sysctl_rdstruct(where, given, new, 2197 &tableinfo, sizeof(tableinfo)); 2198 return (error); 2199 case NET_RT_IFNAMES: 2200 NET_LOCK_SHARED(); 2201 error = sysctl_ifnames(&w); 2202 NET_UNLOCK_SHARED(); 2203 break; 2204 case NET_RT_SOURCE: 2205 tableid = w.w_arg; 2206 if (!rtable_exists(tableid)) 2207 return (ENOENT); 2208 NET_LOCK_SHARED(); 2209 for (i = 1; i <= AF_MAX; i++) { 2210 if (af != 0 && af != i) 2211 continue; 2212 2213 error = sysctl_source(i, tableid, &w); 2214 if (error == EAFNOSUPPORT) 2215 error = 0; 2216 if (error) 2217 break; 2218 } 2219 NET_UNLOCK_SHARED(); 2220 break; 2221 } 2222 free(w.w_tmem, M_RTABLE, w.w_tmemsize); 2223 if (where) { 2224 *given = w.w_where - (caddr_t)where; 2225 if (w.w_needed > w.w_given) 2226 return (ENOMEM); 2227 } else if (w.w_needed == 0) { 2228 *given = 0; 2229 } else { 2230 *given = roundup(w.w_needed + MAX(w.w_needed / 10, 1024), 2231 PAGE_SIZE); 2232 } 2233 return (error); 2234 } 2235 2236 int 2237 sysctl_rtable_rtstat(void *oldp, size_t *oldlenp, void *newp) 2238 { 2239 extern struct cpumem *rtcounters; 2240 uint64_t counters[rts_ncounters]; 2241 struct rtstat rtstat; 2242 uint32_t *words = (uint32_t *)&rtstat; 2243 int i; 2244 2245 CTASSERT(sizeof(rtstat) == (nitems(counters) * sizeof(uint32_t))); 2246 memset(&rtstat, 0, sizeof rtstat); 2247 counters_read(rtcounters, counters, nitems(counters)); 2248 2249 for (i = 0; i < nitems(counters); i++) 2250 words[i] = (uint32_t)counters[i]; 2251 2252 return (sysctl_rdstruct(oldp, oldlenp, newp, &rtstat, sizeof(rtstat))); 2253 } 2254 2255 int 2256 rtm_validate_proposal(struct rt_addrinfo *info) 2257 { 2258 if (info->rti_addrs & ~(RTA_NETMASK | RTA_IFA | RTA_DNS | RTA_STATIC | 2259 RTA_SEARCH)) { 2260 return -1; 2261 } 2262 2263 if (ISSET(info->rti_addrs, RTA_NETMASK)) { 2264 struct sockaddr *sa = info->rti_info[RTAX_NETMASK]; 2265 if (sa == NULL) 2266 return -1; 2267 switch (sa->sa_family) { 2268 case AF_INET: 2269 if (sa->sa_len != sizeof(struct sockaddr_in)) 2270 return -1; 2271 break; 2272 case AF_INET6: 2273 if (sa->sa_len != sizeof(struct sockaddr_in6)) 2274 return -1; 2275 break; 2276 default: 2277 return -1; 2278 } 2279 } 2280 2281 if (ISSET(info->rti_addrs, RTA_IFA)) { 2282 struct sockaddr *sa = info->rti_info[RTAX_IFA]; 2283 if (sa == NULL) 2284 return -1; 2285 switch (sa->sa_family) { 2286 case AF_INET: 2287 if (sa->sa_len != sizeof(struct sockaddr_in)) 2288 return -1; 2289 break; 2290 case AF_INET6: 2291 if (sa->sa_len != sizeof(struct sockaddr_in6)) 2292 return -1; 2293 break; 2294 default: 2295 return -1; 2296 } 2297 } 2298 2299 if (ISSET(info->rti_addrs, RTA_DNS)) { 2300 struct sockaddr_rtdns *rtdns = 2301 (struct sockaddr_rtdns *)info->rti_info[RTAX_DNS]; 2302 if (rtdns == NULL) 2303 return -1; 2304 if (rtdns->sr_len > sizeof(*rtdns)) 2305 return -1; 2306 if (rtdns->sr_len < offsetof(struct sockaddr_rtdns, sr_dns)) 2307 return -1; 2308 switch (rtdns->sr_family) { 2309 case AF_INET: 2310 if ((rtdns->sr_len - offsetof(struct sockaddr_rtdns, 2311 sr_dns)) % sizeof(struct in_addr) != 0) 2312 return -1; 2313 break; 2314 #ifdef INET6 2315 case AF_INET6: 2316 if ((rtdns->sr_len - offsetof(struct sockaddr_rtdns, 2317 sr_dns)) % sizeof(struct in6_addr) != 0) 2318 return -1; 2319 break; 2320 #endif 2321 default: 2322 return -1; 2323 } 2324 } 2325 2326 if (ISSET(info->rti_addrs, RTA_STATIC)) { 2327 struct sockaddr_rtstatic *rtstatic = 2328 (struct sockaddr_rtstatic *)info->rti_info[RTAX_STATIC]; 2329 if (rtstatic == NULL) 2330 return -1; 2331 if (rtstatic->sr_len > sizeof(*rtstatic)) 2332 return -1; 2333 if (rtstatic->sr_len <= 2334 offsetof(struct sockaddr_rtstatic, sr_static)) 2335 return -1; 2336 } 2337 2338 if (ISSET(info->rti_addrs, RTA_SEARCH)) { 2339 struct sockaddr_rtsearch *rtsearch = 2340 (struct sockaddr_rtsearch *)info->rti_info[RTAX_SEARCH]; 2341 if (rtsearch == NULL) 2342 return -1; 2343 if (rtsearch->sr_len > sizeof(*rtsearch)) 2344 return -1; 2345 if (rtsearch->sr_len <= 2346 offsetof(struct sockaddr_rtsearch, sr_search)) 2347 return -1; 2348 } 2349 2350 return 0; 2351 } 2352 2353 int 2354 rt_setsource(unsigned int rtableid, struct sockaddr *src) 2355 { 2356 struct ifaddr *ifa; 2357 /* 2358 * If source address is 0.0.0.0 or :: 2359 * use automatic source selection 2360 */ 2361 switch(src->sa_family) { 2362 case AF_INET: 2363 if(satosin(src)->sin_addr.s_addr == INADDR_ANY) { 2364 rtable_setsource(rtableid, AF_INET, NULL); 2365 return (0); 2366 } 2367 break; 2368 #ifdef INET6 2369 case AF_INET6: 2370 if (IN6_IS_ADDR_UNSPECIFIED(&satosin6(src)->sin6_addr)) { 2371 rtable_setsource(rtableid, AF_INET6, NULL); 2372 return (0); 2373 } 2374 break; 2375 #endif 2376 default: 2377 return (EAFNOSUPPORT); 2378 } 2379 2380 /* 2381 * Check if source address is assigned to an interface in the 2382 * same rdomain 2383 */ 2384 if ((ifa = ifa_ifwithaddr(src, rtableid)) == NULL) 2385 return (EINVAL); 2386 2387 return rtable_setsource(rtableid, src->sa_family, ifa->ifa_addr); 2388 } 2389 2390 /* 2391 * Definitions of protocols supported in the ROUTE domain. 2392 */ 2393 2394 const struct pr_usrreqs route_usrreqs = { 2395 .pru_attach = route_attach, 2396 .pru_detach = route_detach, 2397 .pru_disconnect = route_disconnect, 2398 .pru_shutdown = route_shutdown, 2399 .pru_rcvd = route_rcvd, 2400 .pru_send = route_send, 2401 .pru_sockaddr = route_sockaddr, 2402 .pru_peeraddr = route_peeraddr, 2403 }; 2404 2405 const struct protosw routesw[] = { 2406 { 2407 .pr_type = SOCK_RAW, 2408 .pr_domain = &routedomain, 2409 .pr_flags = PR_ATOMIC|PR_ADDR|PR_WANTRCVD, 2410 .pr_ctloutput = route_ctloutput, 2411 .pr_usrreqs = &route_usrreqs, 2412 .pr_init = route_prinit, 2413 .pr_sysctl = sysctl_rtable 2414 } 2415 }; 2416 2417 const struct domain routedomain = { 2418 .dom_family = PF_ROUTE, 2419 .dom_name = "route", 2420 .dom_init = route_init, 2421 .dom_protosw = routesw, 2422 .dom_protoswNPROTOSW = &routesw[nitems(routesw)] 2423 }; 2424