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 33 #include <sys/param.h> 34 #include <sys/domain.h> 35 #include <sys/kernel.h> 36 #include <sys/jail.h> 37 #include <sys/malloc.h> 38 #include <sys/mbuf.h> 39 #include <sys/proc.h> 40 #include <sys/protosw.h> 41 #include <sys/signalvar.h> 42 #include <sys/socket.h> 43 #include <sys/socketvar.h> 44 #include <sys/sysctl.h> 45 #include <sys/systm.h> 46 47 #include <net/if.h> 48 #include <net/netisr.h> 49 #include <net/raw_cb.h> 50 #include <net/route.h> 51 52 #include <netinet/in.h> 53 54 MALLOC_DEFINE(M_RTABLE, "routetbl", "routing tables"); 55 56 /* NB: these are not modified */ 57 static struct sockaddr route_dst = { 2, PF_ROUTE, }; 58 static struct sockaddr route_src = { 2, PF_ROUTE, }; 59 static struct sockaddr sa_zero = { sizeof(sa_zero), AF_INET, }; 60 61 static struct { 62 int ip_count; /* attached w/ AF_INET */ 63 int ip6_count; /* attached w/ AF_INET6 */ 64 int ipx_count; /* attached w/ AF_IPX */ 65 int any_count; /* total attached */ 66 } route_cb; 67 68 struct mtx rtsock_mtx; 69 MTX_SYSINIT(rtsock, &rtsock_mtx, "rtsock route_cb lock", MTX_DEF); 70 71 #define RTSOCK_LOCK() mtx_lock(&rtsock_mtx) 72 #define RTSOCK_UNLOCK() mtx_unlock(&rtsock_mtx) 73 #define RTSOCK_LOCK_ASSERT() mtx_assert(&rtsock_mtx, MA_OWNED) 74 75 static struct ifqueue rtsintrq; 76 77 SYSCTL_NODE(_net, OID_AUTO, route, CTLFLAG_RD, 0, ""); 78 SYSCTL_INT(_net_route, OID_AUTO, netisr_maxqlen, CTLFLAG_RW, 79 &rtsintrq.ifq_maxlen, 0, "maximum routing socket dispatch queue length"); 80 81 struct walkarg { 82 int w_tmemsize; 83 int w_op, w_arg; 84 caddr_t w_tmem; 85 struct sysctl_req *w_req; 86 }; 87 88 static void rts_input(struct mbuf *m); 89 static struct mbuf *rt_msg1(int type, struct rt_addrinfo *rtinfo); 90 static int rt_msg2(int type, struct rt_addrinfo *rtinfo, 91 caddr_t cp, struct walkarg *w); 92 static int rt_xaddrs(caddr_t cp, caddr_t cplim, 93 struct rt_addrinfo *rtinfo); 94 static int sysctl_dumpentry(struct radix_node *rn, void *vw); 95 static int sysctl_iflist(int af, struct walkarg *w); 96 static int sysctl_ifmalist(int af, struct walkarg *w); 97 static int route_output(struct mbuf *m, struct socket *so); 98 static void rt_setmetrics(u_long which, const struct rt_metrics *in, 99 struct rt_metrics_lite *out); 100 static void rt_getmetrics(const struct rt_metrics_lite *in, 101 struct rt_metrics *out); 102 static void rt_dispatch(struct mbuf *, const struct sockaddr *); 103 104 static void 105 rts_init(void) 106 { 107 int tmp; 108 109 rtsintrq.ifq_maxlen = 256; 110 if (TUNABLE_INT_FETCH("net.route.netisr_maxqlen", &tmp)) 111 rtsintrq.ifq_maxlen = tmp; 112 mtx_init(&rtsintrq.ifq_mtx, "rts_inq", NULL, MTX_DEF); 113 netisr_register(NETISR_ROUTE, rts_input, &rtsintrq, NETISR_MPSAFE); 114 } 115 SYSINIT(rtsock, SI_SUB_PROTO_DOMAIN, SI_ORDER_THIRD, rts_init, 0) 116 117 static void 118 rts_input(struct mbuf *m) 119 { 120 struct sockproto route_proto; 121 unsigned short *family; 122 struct m_tag *tag; 123 124 route_proto.sp_family = PF_ROUTE; 125 tag = m_tag_find(m, PACKET_TAG_RTSOCKFAM, NULL); 126 if (tag != NULL) { 127 family = (unsigned short *)(tag + 1); 128 route_proto.sp_protocol = *family; 129 m_tag_delete(m, tag); 130 } else 131 route_proto.sp_protocol = 0; 132 133 raw_input(m, &route_proto, &route_src, &route_dst); 134 } 135 136 /* 137 * It really doesn't make any sense at all for this code to share much 138 * with raw_usrreq.c, since its functionality is so restricted. XXX 139 */ 140 static int 141 rts_abort(struct socket *so) 142 { 143 144 return (raw_usrreqs.pru_abort(so)); 145 } 146 147 /* pru_accept is EOPNOTSUPP */ 148 149 static int 150 rts_attach(struct socket *so, int proto, struct thread *td) 151 { 152 struct rawcb *rp; 153 int s, error; 154 155 if (sotorawcb(so) != NULL) 156 return EISCONN; /* XXX panic? */ 157 /* XXX */ 158 MALLOC(rp, struct rawcb *, sizeof *rp, M_PCB, M_WAITOK | M_ZERO); 159 if (rp == NULL) 160 return ENOBUFS; 161 162 /* 163 * The splnet() is necessary to block protocols from sending 164 * error notifications (like RTM_REDIRECT or RTM_LOSING) while 165 * this PCB is extant but incompletely initialized. 166 * Probably we should try to do more of this work beforehand and 167 * eliminate the spl. 168 */ 169 s = splnet(); 170 so->so_pcb = (caddr_t)rp; 171 error = raw_attach(so, proto); 172 rp = sotorawcb(so); 173 if (error) { 174 splx(s); 175 so->so_pcb = NULL; 176 free(rp, M_PCB); 177 return error; 178 } 179 RTSOCK_LOCK(); 180 switch(rp->rcb_proto.sp_protocol) { 181 case AF_INET: 182 route_cb.ip_count++; 183 break; 184 case AF_INET6: 185 route_cb.ip6_count++; 186 break; 187 case AF_IPX: 188 route_cb.ipx_count++; 189 break; 190 } 191 rp->rcb_faddr = &route_src; 192 route_cb.any_count++; 193 RTSOCK_UNLOCK(); 194 soisconnected(so); 195 so->so_options |= SO_USELOOPBACK; 196 splx(s); 197 return 0; 198 } 199 200 static int 201 rts_bind(struct socket *so, struct sockaddr *nam, struct thread *td) 202 { 203 204 return (raw_usrreqs.pru_bind(so, nam, td)); /* xxx just EINVAL */ 205 } 206 207 static int 208 rts_connect(struct socket *so, struct sockaddr *nam, struct thread *td) 209 { 210 211 return (raw_usrreqs.pru_connect(so, nam, td)); /* XXX just EINVAL */ 212 } 213 214 /* pru_connect2 is EOPNOTSUPP */ 215 /* pru_control is EOPNOTSUPP */ 216 217 static int 218 rts_detach(struct socket *so) 219 { 220 struct rawcb *rp = sotorawcb(so); 221 int s, error; 222 223 s = splnet(); 224 if (rp != NULL) { 225 RTSOCK_LOCK(); 226 switch(rp->rcb_proto.sp_protocol) { 227 case AF_INET: 228 route_cb.ip_count--; 229 break; 230 case AF_INET6: 231 route_cb.ip6_count--; 232 break; 233 case AF_IPX: 234 route_cb.ipx_count--; 235 break; 236 } 237 route_cb.any_count--; 238 RTSOCK_UNLOCK(); 239 } 240 error = raw_usrreqs.pru_detach(so); 241 splx(s); 242 return error; 243 } 244 245 static int 246 rts_disconnect(struct socket *so) 247 { 248 249 return (raw_usrreqs.pru_disconnect(so)); 250 } 251 252 /* pru_listen is EOPNOTSUPP */ 253 254 static int 255 rts_peeraddr(struct socket *so, struct sockaddr **nam) 256 { 257 258 return (raw_usrreqs.pru_peeraddr(so, nam)); 259 } 260 261 /* pru_rcvd is EOPNOTSUPP */ 262 /* pru_rcvoob is EOPNOTSUPP */ 263 264 static int 265 rts_send(struct socket *so, int flags, struct mbuf *m, struct sockaddr *nam, 266 struct mbuf *control, struct thread *td) 267 { 268 269 return (raw_usrreqs.pru_send(so, flags, m, nam, control, td)); 270 } 271 272 /* pru_sense is null */ 273 274 static int 275 rts_shutdown(struct socket *so) 276 { 277 278 return (raw_usrreqs.pru_shutdown(so)); 279 } 280 281 static int 282 rts_sockaddr(struct socket *so, struct sockaddr **nam) 283 { 284 285 return (raw_usrreqs.pru_sockaddr(so, nam)); 286 } 287 288 static struct pr_usrreqs route_usrreqs = { 289 .pru_abort = rts_abort, 290 .pru_attach = rts_attach, 291 .pru_bind = rts_bind, 292 .pru_connect = rts_connect, 293 .pru_detach = rts_detach, 294 .pru_disconnect = rts_disconnect, 295 .pru_peeraddr = rts_peeraddr, 296 .pru_send = rts_send, 297 .pru_shutdown = rts_shutdown, 298 .pru_sockaddr = rts_sockaddr, 299 }; 300 301 /*ARGSUSED*/ 302 static int 303 route_output(struct mbuf *m, struct socket *so) 304 { 305 #define sa_equal(a1, a2) (bcmp((a1), (a2), (a1)->sa_len) == 0) 306 struct rt_msghdr *rtm = NULL; 307 struct rtentry *rt = NULL; 308 struct radix_node_head *rnh; 309 struct rt_addrinfo info; 310 int len, error = 0; 311 struct ifnet *ifp = NULL; 312 struct ifaddr *ifa = NULL; 313 struct sockaddr_in jail; 314 315 #define senderr(e) { error = e; goto flush;} 316 if (m == NULL || ((m->m_len < sizeof(long)) && 317 (m = m_pullup(m, sizeof(long))) == NULL)) 318 return (ENOBUFS); 319 if ((m->m_flags & M_PKTHDR) == 0) 320 panic("route_output"); 321 len = m->m_pkthdr.len; 322 if (len < sizeof(*rtm) || 323 len != mtod(m, struct rt_msghdr *)->rtm_msglen) { 324 info.rti_info[RTAX_DST] = NULL; 325 senderr(EINVAL); 326 } 327 R_Malloc(rtm, struct rt_msghdr *, len); 328 if (rtm == NULL) { 329 info.rti_info[RTAX_DST] = NULL; 330 senderr(ENOBUFS); 331 } 332 m_copydata(m, 0, len, (caddr_t)rtm); 333 if (rtm->rtm_version != RTM_VERSION) { 334 info.rti_info[RTAX_DST] = NULL; 335 senderr(EPROTONOSUPPORT); 336 } 337 rtm->rtm_pid = curproc->p_pid; 338 bzero(&info, sizeof(info)); 339 info.rti_addrs = rtm->rtm_addrs; 340 if (rt_xaddrs((caddr_t)(rtm + 1), len + (caddr_t)rtm, &info)) { 341 info.rti_info[RTAX_DST] = NULL; 342 senderr(EINVAL); 343 } 344 info.rti_flags = rtm->rtm_flags; 345 if (info.rti_info[RTAX_DST] == NULL || 346 info.rti_info[RTAX_DST]->sa_family >= AF_MAX || 347 (info.rti_info[RTAX_GATEWAY] != NULL && 348 info.rti_info[RTAX_GATEWAY]->sa_family >= AF_MAX)) 349 senderr(EINVAL); 350 if (info.rti_info[RTAX_GENMASK]) { 351 struct radix_node *t; 352 t = rn_addmask((caddr_t) info.rti_info[RTAX_GENMASK], 0, 1); 353 if (t != NULL && 354 bcmp((char *)(void *)info.rti_info[RTAX_GENMASK] + 1, 355 (char *)(void *)t->rn_key + 1, 356 ((struct sockaddr *)t->rn_key)->sa_len - 1) == 0) 357 info.rti_info[RTAX_GENMASK] = 358 (struct sockaddr *)t->rn_key; 359 else 360 senderr(ENOBUFS); 361 } 362 363 /* 364 * Verify that the caller has the appropriate privilege; RTM_GET 365 * is the only operation the non-superuser is allowed. 366 */ 367 if (rtm->rtm_type != RTM_GET && (error = suser(curthread)) != 0) 368 senderr(error); 369 370 switch (rtm->rtm_type) { 371 struct rtentry *saved_nrt; 372 373 case RTM_ADD: 374 if (info.rti_info[RTAX_GATEWAY] == NULL) 375 senderr(EINVAL); 376 saved_nrt = NULL; 377 error = rtrequest1(RTM_ADD, &info, &saved_nrt); 378 if (error == 0 && saved_nrt) { 379 RT_LOCK(saved_nrt); 380 rt_setmetrics(rtm->rtm_inits, 381 &rtm->rtm_rmx, &saved_nrt->rt_rmx); 382 RT_REMREF(saved_nrt); 383 saved_nrt->rt_genmask = info.rti_info[RTAX_GENMASK]; 384 RT_UNLOCK(saved_nrt); 385 } 386 break; 387 388 case RTM_DELETE: 389 saved_nrt = NULL; 390 error = rtrequest1(RTM_DELETE, &info, &saved_nrt); 391 if (error == 0) { 392 RT_LOCK(saved_nrt); 393 rt = saved_nrt; 394 goto report; 395 } 396 break; 397 398 case RTM_GET: 399 case RTM_CHANGE: 400 case RTM_LOCK: 401 rnh = rt_tables[info.rti_info[RTAX_DST]->sa_family]; 402 if (rnh == NULL) 403 senderr(EAFNOSUPPORT); 404 RADIX_NODE_HEAD_LOCK(rnh); 405 rt = (struct rtentry *) rnh->rnh_lookup(info.rti_info[RTAX_DST], 406 info.rti_info[RTAX_NETMASK], rnh); 407 if (rt == NULL) { /* XXX looks bogus */ 408 RADIX_NODE_HEAD_UNLOCK(rnh); 409 senderr(ESRCH); 410 } 411 RT_LOCK(rt); 412 RT_ADDREF(rt); 413 RADIX_NODE_HEAD_UNLOCK(rnh); 414 415 /* 416 * Fix for PR: 82974 417 * 418 * RTM_CHANGE/LOCK need a perfect match, rn_lookup() 419 * returns a perfect match in case a netmask is 420 * specified. For host routes only a longest prefix 421 * match is returned so it is necessary to compare the 422 * existence of the netmask. If both have a netmask 423 * rnh_lookup() did a perfect match and if none of them 424 * have a netmask both are host routes which is also a 425 * perfect match. 426 */ 427 428 if (rtm->rtm_type != RTM_GET && 429 (!rt_mask(rt) != !info.rti_info[RTAX_NETMASK])) { 430 RT_UNLOCK(rt); 431 senderr(ESRCH); 432 } 433 434 switch(rtm->rtm_type) { 435 436 case RTM_GET: 437 report: 438 RT_LOCK_ASSERT(rt); 439 info.rti_info[RTAX_DST] = rt_key(rt); 440 info.rti_info[RTAX_GATEWAY] = rt->rt_gateway; 441 info.rti_info[RTAX_NETMASK] = rt_mask(rt); 442 info.rti_info[RTAX_GENMASK] = rt->rt_genmask; 443 if (rtm->rtm_addrs & (RTA_IFP | RTA_IFA)) { 444 ifp = rt->rt_ifp; 445 if (ifp) { 446 info.rti_info[RTAX_IFP] = 447 ifaddr_byindex(ifp->if_index)->ifa_addr; 448 if (jailed(so->so_cred)) { 449 bzero(&jail, sizeof(jail)); 450 jail.sin_family = PF_INET; 451 jail.sin_len = sizeof(jail); 452 jail.sin_addr.s_addr = 453 htonl(prison_getip(so->so_cred)); 454 info.rti_info[RTAX_IFA] = 455 (struct sockaddr *)&jail; 456 } else 457 info.rti_info[RTAX_IFA] = 458 rt->rt_ifa->ifa_addr; 459 if (ifp->if_flags & IFF_POINTOPOINT) 460 info.rti_info[RTAX_BRD] = 461 rt->rt_ifa->ifa_dstaddr; 462 rtm->rtm_index = ifp->if_index; 463 } else { 464 info.rti_info[RTAX_IFP] = NULL; 465 info.rti_info[RTAX_IFA] = NULL; 466 } 467 } else if ((ifp = rt->rt_ifp) != NULL) { 468 rtm->rtm_index = ifp->if_index; 469 } 470 len = rt_msg2(rtm->rtm_type, &info, NULL, NULL); 471 if (len > rtm->rtm_msglen) { 472 struct rt_msghdr *new_rtm; 473 R_Malloc(new_rtm, struct rt_msghdr *, len); 474 if (new_rtm == NULL) { 475 RT_UNLOCK(rt); 476 senderr(ENOBUFS); 477 } 478 bcopy(rtm, new_rtm, rtm->rtm_msglen); 479 Free(rtm); rtm = new_rtm; 480 } 481 (void)rt_msg2(rtm->rtm_type, &info, (caddr_t)rtm, NULL); 482 rtm->rtm_flags = rt->rt_flags; 483 rt_getmetrics(&rt->rt_rmx, &rtm->rtm_rmx); 484 rtm->rtm_addrs = info.rti_addrs; 485 break; 486 487 case RTM_CHANGE: 488 /* 489 * New gateway could require new ifaddr, ifp; 490 * flags may also be different; ifp may be specified 491 * by ll sockaddr when protocol address is ambiguous 492 */ 493 if (((rt->rt_flags & RTF_GATEWAY) && 494 info.rti_info[RTAX_GATEWAY] != NULL) || 495 info.rti_info[RTAX_IFP] != NULL || 496 (info.rti_info[RTAX_IFA] != NULL && 497 !sa_equal(info.rti_info[RTAX_IFA], 498 rt->rt_ifa->ifa_addr))) { 499 RT_UNLOCK(rt); 500 if ((error = rt_getifa(&info)) != 0) 501 senderr(error); 502 RT_LOCK(rt); 503 } 504 if (info.rti_info[RTAX_GATEWAY] != NULL && 505 (error = rt_setgate(rt, rt_key(rt), 506 info.rti_info[RTAX_GATEWAY])) != 0) { 507 RT_UNLOCK(rt); 508 senderr(error); 509 } 510 if ((ifa = info.rti_ifa) != NULL) { 511 struct ifaddr *oifa = rt->rt_ifa; 512 if (oifa != ifa) { 513 if (oifa) { 514 if (oifa->ifa_rtrequest) 515 oifa->ifa_rtrequest( 516 RTM_DELETE, rt, 517 &info); 518 IFAFREE(oifa); 519 } 520 IFAREF(ifa); 521 rt->rt_ifa = ifa; 522 rt->rt_ifp = info.rti_ifp; 523 } 524 } 525 rt_setmetrics(rtm->rtm_inits, &rtm->rtm_rmx, 526 &rt->rt_rmx); 527 if (rt->rt_ifa && rt->rt_ifa->ifa_rtrequest) 528 rt->rt_ifa->ifa_rtrequest(RTM_ADD, rt, &info); 529 if (info.rti_info[RTAX_GENMASK]) 530 rt->rt_genmask = info.rti_info[RTAX_GENMASK]; 531 /* FALLTHROUGH */ 532 case RTM_LOCK: 533 /* We don't support locks anymore */ 534 break; 535 } 536 RT_UNLOCK(rt); 537 break; 538 539 default: 540 senderr(EOPNOTSUPP); 541 } 542 543 flush: 544 if (rtm) { 545 if (error) 546 rtm->rtm_errno = error; 547 else 548 rtm->rtm_flags |= RTF_DONE; 549 } 550 if (rt) /* XXX can this be true? */ 551 RTFREE(rt); 552 { 553 struct rawcb *rp = NULL; 554 /* 555 * Check to see if we don't want our own messages. 556 */ 557 if ((so->so_options & SO_USELOOPBACK) == 0) { 558 if (route_cb.any_count <= 1) { 559 if (rtm) 560 Free(rtm); 561 m_freem(m); 562 return (error); 563 } 564 /* There is another listener, so construct message */ 565 rp = sotorawcb(so); 566 } 567 if (rtm) { 568 m_copyback(m, 0, rtm->rtm_msglen, (caddr_t)rtm); 569 if (m->m_pkthdr.len < rtm->rtm_msglen) { 570 m_freem(m); 571 m = NULL; 572 } else if (m->m_pkthdr.len > rtm->rtm_msglen) 573 m_adj(m, rtm->rtm_msglen - m->m_pkthdr.len); 574 Free(rtm); 575 } 576 if (m) { 577 if (rp) { 578 /* 579 * XXX insure we don't get a copy by 580 * invalidating our protocol 581 */ 582 unsigned short family = rp->rcb_proto.sp_family; 583 rp->rcb_proto.sp_family = 0; 584 rt_dispatch(m, info.rti_info[RTAX_DST]); 585 rp->rcb_proto.sp_family = family; 586 } else 587 rt_dispatch(m, info.rti_info[RTAX_DST]); 588 } 589 } 590 return (error); 591 #undef sa_equal 592 } 593 594 static void 595 rt_setmetrics(u_long which, const struct rt_metrics *in, 596 struct rt_metrics_lite *out) 597 { 598 #define metric(f, e) if (which & (f)) out->e = in->e; 599 /* 600 * Only these are stored in the routing entry since introduction 601 * of tcp hostcache. The rest is ignored. 602 */ 603 metric(RTV_MTU, rmx_mtu); 604 metric(RTV_EXPIRE, rmx_expire); 605 #undef metric 606 } 607 608 static void 609 rt_getmetrics(const struct rt_metrics_lite *in, struct rt_metrics *out) 610 { 611 #define metric(e) out->e = in->e; 612 bzero(out, sizeof(*out)); 613 metric(rmx_mtu); 614 metric(rmx_expire); 615 #undef metric 616 } 617 618 /* 619 * Extract the addresses of the passed sockaddrs. 620 * Do a little sanity checking so as to avoid bad memory references. 621 * This data is derived straight from userland. 622 */ 623 static int 624 rt_xaddrs(caddr_t cp, caddr_t cplim, struct rt_addrinfo *rtinfo) 625 { 626 struct sockaddr *sa; 627 int i; 628 629 for (i = 0; i < RTAX_MAX && cp < cplim; i++) { 630 if ((rtinfo->rti_addrs & (1 << i)) == 0) 631 continue; 632 sa = (struct sockaddr *)cp; 633 /* 634 * It won't fit. 635 */ 636 if (cp + sa->sa_len > cplim) 637 return (EINVAL); 638 /* 639 * there are no more.. quit now 640 * If there are more bits, they are in error. 641 * I've seen this. route(1) can evidently generate these. 642 * This causes kernel to core dump. 643 * for compatibility, If we see this, point to a safe address. 644 */ 645 if (sa->sa_len == 0) { 646 rtinfo->rti_info[i] = &sa_zero; 647 return (0); /* should be EINVAL but for compat */ 648 } 649 /* accept it */ 650 rtinfo->rti_info[i] = sa; 651 cp += SA_SIZE(sa); 652 } 653 return (0); 654 } 655 656 static struct mbuf * 657 rt_msg1(int type, struct rt_addrinfo *rtinfo) 658 { 659 struct rt_msghdr *rtm; 660 struct mbuf *m; 661 int i; 662 struct sockaddr *sa; 663 int len, dlen; 664 665 switch (type) { 666 667 case RTM_DELADDR: 668 case RTM_NEWADDR: 669 len = sizeof(struct ifa_msghdr); 670 break; 671 672 case RTM_DELMADDR: 673 case RTM_NEWMADDR: 674 len = sizeof(struct ifma_msghdr); 675 break; 676 677 case RTM_IFINFO: 678 len = sizeof(struct if_msghdr); 679 break; 680 681 case RTM_IFANNOUNCE: 682 case RTM_IEEE80211: 683 len = sizeof(struct if_announcemsghdr); 684 break; 685 686 default: 687 len = sizeof(struct rt_msghdr); 688 } 689 if (len > MCLBYTES) 690 panic("rt_msg1"); 691 m = m_gethdr(M_DONTWAIT, MT_DATA); 692 if (m && len > MHLEN) { 693 MCLGET(m, M_DONTWAIT); 694 if ((m->m_flags & M_EXT) == 0) { 695 m_free(m); 696 m = NULL; 697 } 698 } 699 if (m == NULL) 700 return (m); 701 m->m_pkthdr.len = m->m_len = len; 702 m->m_pkthdr.rcvif = NULL; 703 rtm = mtod(m, struct rt_msghdr *); 704 bzero((caddr_t)rtm, len); 705 for (i = 0; i < RTAX_MAX; i++) { 706 if ((sa = rtinfo->rti_info[i]) == NULL) 707 continue; 708 rtinfo->rti_addrs |= (1 << i); 709 dlen = SA_SIZE(sa); 710 m_copyback(m, len, dlen, (caddr_t)sa); 711 len += dlen; 712 } 713 if (m->m_pkthdr.len != len) { 714 m_freem(m); 715 return (NULL); 716 } 717 rtm->rtm_msglen = len; 718 rtm->rtm_version = RTM_VERSION; 719 rtm->rtm_type = type; 720 return (m); 721 } 722 723 static int 724 rt_msg2(int type, struct rt_addrinfo *rtinfo, caddr_t cp, struct walkarg *w) 725 { 726 int i; 727 int len, dlen, second_time = 0; 728 caddr_t cp0; 729 730 rtinfo->rti_addrs = 0; 731 again: 732 switch (type) { 733 734 case RTM_DELADDR: 735 case RTM_NEWADDR: 736 len = sizeof(struct ifa_msghdr); 737 break; 738 739 case RTM_IFINFO: 740 len = sizeof(struct if_msghdr); 741 break; 742 743 case RTM_NEWMADDR: 744 len = sizeof(struct ifma_msghdr); 745 break; 746 747 default: 748 len = sizeof(struct rt_msghdr); 749 } 750 cp0 = cp; 751 if (cp0) 752 cp += len; 753 for (i = 0; i < RTAX_MAX; i++) { 754 struct sockaddr *sa; 755 756 if ((sa = rtinfo->rti_info[i]) == NULL) 757 continue; 758 rtinfo->rti_addrs |= (1 << i); 759 dlen = SA_SIZE(sa); 760 if (cp) { 761 bcopy((caddr_t)sa, cp, (unsigned)dlen); 762 cp += dlen; 763 } 764 len += dlen; 765 } 766 len = ALIGN(len); 767 if (cp == NULL && w != NULL && !second_time) { 768 struct walkarg *rw = w; 769 770 if (rw->w_req) { 771 if (rw->w_tmemsize < len) { 772 if (rw->w_tmem) 773 free(rw->w_tmem, M_RTABLE); 774 rw->w_tmem = (caddr_t) 775 malloc(len, M_RTABLE, M_NOWAIT); 776 if (rw->w_tmem) 777 rw->w_tmemsize = len; 778 } 779 if (rw->w_tmem) { 780 cp = rw->w_tmem; 781 second_time = 1; 782 goto again; 783 } 784 } 785 } 786 if (cp) { 787 struct rt_msghdr *rtm = (struct rt_msghdr *)cp0; 788 789 rtm->rtm_version = RTM_VERSION; 790 rtm->rtm_type = type; 791 rtm->rtm_msglen = len; 792 } 793 return (len); 794 } 795 796 /* 797 * This routine is called to generate a message from the routing 798 * socket indicating that a redirect has occured, a routing lookup 799 * has failed, or that a protocol has detected timeouts to a particular 800 * destination. 801 */ 802 void 803 rt_missmsg(int type, struct rt_addrinfo *rtinfo, int flags, int error) 804 { 805 struct rt_msghdr *rtm; 806 struct mbuf *m; 807 struct sockaddr *sa = rtinfo->rti_info[RTAX_DST]; 808 809 if (route_cb.any_count == 0) 810 return; 811 m = rt_msg1(type, rtinfo); 812 if (m == NULL) 813 return; 814 rtm = mtod(m, struct rt_msghdr *); 815 rtm->rtm_flags = RTF_DONE | flags; 816 rtm->rtm_errno = error; 817 rtm->rtm_addrs = rtinfo->rti_addrs; 818 rt_dispatch(m, sa); 819 } 820 821 /* 822 * This routine is called to generate a message from the routing 823 * socket indicating that the status of a network interface has changed. 824 */ 825 void 826 rt_ifmsg(struct ifnet *ifp) 827 { 828 struct if_msghdr *ifm; 829 struct mbuf *m; 830 struct rt_addrinfo info; 831 832 if (route_cb.any_count == 0) 833 return; 834 bzero((caddr_t)&info, sizeof(info)); 835 m = rt_msg1(RTM_IFINFO, &info); 836 if (m == NULL) 837 return; 838 ifm = mtod(m, struct if_msghdr *); 839 ifm->ifm_index = ifp->if_index; 840 ifm->ifm_flags = ifp->if_flags | ifp->if_drv_flags; 841 ifm->ifm_data = ifp->if_data; 842 ifm->ifm_addrs = 0; 843 rt_dispatch(m, NULL); 844 } 845 846 /* 847 * This is called to generate messages from the routing socket 848 * indicating a network interface has had addresses associated with it. 849 * if we ever reverse the logic and replace messages TO the routing 850 * socket indicate a request to configure interfaces, then it will 851 * be unnecessary as the routing socket will automatically generate 852 * copies of it. 853 */ 854 void 855 rt_newaddrmsg(int cmd, struct ifaddr *ifa, int error, struct rtentry *rt) 856 { 857 struct rt_addrinfo info; 858 struct sockaddr *sa = NULL; 859 int pass; 860 struct mbuf *m = NULL; 861 struct ifnet *ifp = ifa->ifa_ifp; 862 863 KASSERT(cmd == RTM_ADD || cmd == RTM_DELETE, 864 ("unexpected cmd %u", cmd)); 865 866 if (route_cb.any_count == 0) 867 return; 868 for (pass = 1; pass < 3; pass++) { 869 bzero((caddr_t)&info, sizeof(info)); 870 if ((cmd == RTM_ADD && pass == 1) || 871 (cmd == RTM_DELETE && pass == 2)) { 872 struct ifa_msghdr *ifam; 873 int ncmd = cmd == RTM_ADD ? RTM_NEWADDR : RTM_DELADDR; 874 875 info.rti_info[RTAX_IFA] = sa = ifa->ifa_addr; 876 info.rti_info[RTAX_IFP] = 877 ifaddr_byindex(ifp->if_index)->ifa_addr; 878 info.rti_info[RTAX_NETMASK] = ifa->ifa_netmask; 879 info.rti_info[RTAX_BRD] = ifa->ifa_dstaddr; 880 if ((m = rt_msg1(ncmd, &info)) == NULL) 881 continue; 882 ifam = mtod(m, struct ifa_msghdr *); 883 ifam->ifam_index = ifp->if_index; 884 ifam->ifam_metric = ifa->ifa_metric; 885 ifam->ifam_flags = ifa->ifa_flags; 886 ifam->ifam_addrs = info.rti_addrs; 887 } 888 if ((cmd == RTM_ADD && pass == 2) || 889 (cmd == RTM_DELETE && pass == 1)) { 890 struct rt_msghdr *rtm; 891 892 if (rt == NULL) 893 continue; 894 info.rti_info[RTAX_NETMASK] = rt_mask(rt); 895 info.rti_info[RTAX_DST] = sa = rt_key(rt); 896 info.rti_info[RTAX_GATEWAY] = rt->rt_gateway; 897 if ((m = rt_msg1(cmd, &info)) == NULL) 898 continue; 899 rtm = mtod(m, struct rt_msghdr *); 900 rtm->rtm_index = ifp->if_index; 901 rtm->rtm_flags |= rt->rt_flags; 902 rtm->rtm_errno = error; 903 rtm->rtm_addrs = info.rti_addrs; 904 } 905 rt_dispatch(m, sa); 906 } 907 } 908 909 /* 910 * This is the analogue to the rt_newaddrmsg which performs the same 911 * function but for multicast group memberhips. This is easier since 912 * there is no route state to worry about. 913 */ 914 void 915 rt_newmaddrmsg(int cmd, struct ifmultiaddr *ifma) 916 { 917 struct rt_addrinfo info; 918 struct mbuf *m = NULL; 919 struct ifnet *ifp = ifma->ifma_ifp; 920 struct ifma_msghdr *ifmam; 921 922 if (route_cb.any_count == 0) 923 return; 924 925 bzero((caddr_t)&info, sizeof(info)); 926 info.rti_info[RTAX_IFA] = ifma->ifma_addr; 927 info.rti_info[RTAX_IFP] = 928 ifp ? ifaddr_byindex(ifp->if_index)->ifa_addr : NULL; 929 /* 930 * If a link-layer address is present, present it as a ``gateway'' 931 * (similarly to how ARP entries, e.g., are presented). 932 */ 933 info.rti_info[RTAX_GATEWAY] = ifma->ifma_lladdr; 934 m = rt_msg1(cmd, &info); 935 if (m == NULL) 936 return; 937 ifmam = mtod(m, struct ifma_msghdr *); 938 ifmam->ifmam_index = ifp->if_index; 939 ifmam->ifmam_addrs = info.rti_addrs; 940 rt_dispatch(m, ifma->ifma_addr); 941 } 942 943 static struct mbuf * 944 rt_makeifannouncemsg(struct ifnet *ifp, int type, int what, 945 struct rt_addrinfo *info) 946 { 947 struct if_announcemsghdr *ifan; 948 struct mbuf *m; 949 950 if (route_cb.any_count == 0) 951 return NULL; 952 bzero((caddr_t)info, sizeof(*info)); 953 m = rt_msg1(type, info); 954 if (m != NULL) { 955 ifan = mtod(m, struct if_announcemsghdr *); 956 ifan->ifan_index = ifp->if_index; 957 strlcpy(ifan->ifan_name, ifp->if_xname, 958 sizeof(ifan->ifan_name)); 959 ifan->ifan_what = what; 960 } 961 return m; 962 } 963 964 /* 965 * This is called to generate routing socket messages indicating 966 * IEEE80211 wireless events. 967 * XXX we piggyback on the RTM_IFANNOUNCE msg format in a clumsy way. 968 */ 969 void 970 rt_ieee80211msg(struct ifnet *ifp, int what, void *data, size_t data_len) 971 { 972 struct mbuf *m; 973 struct rt_addrinfo info; 974 975 m = rt_makeifannouncemsg(ifp, RTM_IEEE80211, what, &info); 976 if (m != NULL) { 977 /* 978 * Append the ieee80211 data. Try to stick it in the 979 * mbuf containing the ifannounce msg; otherwise allocate 980 * a new mbuf and append. 981 * 982 * NB: we assume m is a single mbuf. 983 */ 984 if (data_len > M_TRAILINGSPACE(m)) { 985 struct mbuf *n = m_get(M_NOWAIT, MT_DATA); 986 if (n == NULL) { 987 m_freem(m); 988 return; 989 } 990 bcopy(data, mtod(n, void *), data_len); 991 n->m_len = data_len; 992 m->m_next = n; 993 } else if (data_len > 0) { 994 bcopy(data, mtod(m, u_int8_t *) + m->m_len, data_len); 995 m->m_len += data_len; 996 } 997 if (m->m_flags & M_PKTHDR) 998 m->m_pkthdr.len += data_len; 999 mtod(m, struct if_announcemsghdr *)->ifan_msglen += data_len; 1000 rt_dispatch(m, NULL); 1001 } 1002 } 1003 1004 /* 1005 * This is called to generate routing socket messages indicating 1006 * network interface arrival and departure. 1007 */ 1008 void 1009 rt_ifannouncemsg(struct ifnet *ifp, int what) 1010 { 1011 struct mbuf *m; 1012 struct rt_addrinfo info; 1013 1014 m = rt_makeifannouncemsg(ifp, RTM_IFANNOUNCE, what, &info); 1015 if (m != NULL) 1016 rt_dispatch(m, NULL); 1017 } 1018 1019 static void 1020 rt_dispatch(struct mbuf *m, const struct sockaddr *sa) 1021 { 1022 struct m_tag *tag; 1023 1024 /* 1025 * Preserve the family from the sockaddr, if any, in an m_tag for 1026 * use when injecting the mbuf into the routing socket buffer from 1027 * the netisr. 1028 */ 1029 if (sa != NULL) { 1030 tag = m_tag_get(PACKET_TAG_RTSOCKFAM, sizeof(unsigned short), 1031 M_NOWAIT); 1032 if (tag == NULL) { 1033 m_freem(m); 1034 return; 1035 } 1036 *(unsigned short *)(tag + 1) = sa->sa_family; 1037 m_tag_prepend(m, tag); 1038 } 1039 netisr_queue(NETISR_ROUTE, m); /* mbuf is free'd on failure. */ 1040 } 1041 1042 /* 1043 * This is used in dumping the kernel table via sysctl(). 1044 */ 1045 static int 1046 sysctl_dumpentry(struct radix_node *rn, void *vw) 1047 { 1048 struct walkarg *w = vw; 1049 struct rtentry *rt = (struct rtentry *)rn; 1050 int error = 0, size; 1051 struct rt_addrinfo info; 1052 1053 if (w->w_op == NET_RT_FLAGS && !(rt->rt_flags & w->w_arg)) 1054 return 0; 1055 bzero((caddr_t)&info, sizeof(info)); 1056 info.rti_info[RTAX_DST] = rt_key(rt); 1057 info.rti_info[RTAX_GATEWAY] = rt->rt_gateway; 1058 info.rti_info[RTAX_NETMASK] = rt_mask(rt); 1059 info.rti_info[RTAX_GENMASK] = rt->rt_genmask; 1060 if (rt->rt_ifp) { 1061 info.rti_info[RTAX_IFP] = 1062 ifaddr_byindex(rt->rt_ifp->if_index)->ifa_addr; 1063 info.rti_info[RTAX_IFA] = rt->rt_ifa->ifa_addr; 1064 if (rt->rt_ifp->if_flags & IFF_POINTOPOINT) 1065 info.rti_info[RTAX_BRD] = rt->rt_ifa->ifa_dstaddr; 1066 } 1067 size = rt_msg2(RTM_GET, &info, NULL, w); 1068 if (w->w_req && w->w_tmem) { 1069 struct rt_msghdr *rtm = (struct rt_msghdr *)w->w_tmem; 1070 1071 rtm->rtm_flags = rt->rt_flags; 1072 rtm->rtm_use = rt->rt_rmx.rmx_pksent; 1073 rt_getmetrics(&rt->rt_rmx, &rtm->rtm_rmx); 1074 rtm->rtm_index = rt->rt_ifp->if_index; 1075 rtm->rtm_errno = rtm->rtm_pid = rtm->rtm_seq = 0; 1076 rtm->rtm_addrs = info.rti_addrs; 1077 error = SYSCTL_OUT(w->w_req, (caddr_t)rtm, size); 1078 return (error); 1079 } 1080 return (error); 1081 } 1082 1083 static int 1084 sysctl_iflist(int af, struct walkarg *w) 1085 { 1086 struct ifnet *ifp; 1087 struct ifaddr *ifa; 1088 struct rt_addrinfo info; 1089 int len, error = 0; 1090 1091 bzero((caddr_t)&info, sizeof(info)); 1092 IFNET_RLOCK(); 1093 TAILQ_FOREACH(ifp, &ifnet, if_link) { 1094 if (w->w_arg && w->w_arg != ifp->if_index) 1095 continue; 1096 ifa = ifaddr_byindex(ifp->if_index); 1097 info.rti_info[RTAX_IFP] = ifa->ifa_addr; 1098 len = rt_msg2(RTM_IFINFO, &info, NULL, w); 1099 info.rti_info[RTAX_IFP] = NULL; 1100 if (w->w_req && w->w_tmem) { 1101 struct if_msghdr *ifm; 1102 1103 ifm = (struct if_msghdr *)w->w_tmem; 1104 ifm->ifm_index = ifp->if_index; 1105 ifm->ifm_flags = ifp->if_flags | ifp->if_drv_flags; 1106 ifm->ifm_data = ifp->if_data; 1107 ifm->ifm_addrs = info.rti_addrs; 1108 error = SYSCTL_OUT(w->w_req,(caddr_t)ifm, len); 1109 if (error) 1110 goto done; 1111 } 1112 while ((ifa = TAILQ_NEXT(ifa, ifa_link)) != NULL) { 1113 if (af && af != ifa->ifa_addr->sa_family) 1114 continue; 1115 if (jailed(curthread->td_ucred) && 1116 prison_if(curthread->td_ucred, ifa->ifa_addr)) 1117 continue; 1118 info.rti_info[RTAX_IFA] = ifa->ifa_addr; 1119 info.rti_info[RTAX_NETMASK] = ifa->ifa_netmask; 1120 info.rti_info[RTAX_BRD] = ifa->ifa_dstaddr; 1121 len = rt_msg2(RTM_NEWADDR, &info, NULL, w); 1122 if (w->w_req && w->w_tmem) { 1123 struct ifa_msghdr *ifam; 1124 1125 ifam = (struct ifa_msghdr *)w->w_tmem; 1126 ifam->ifam_index = ifa->ifa_ifp->if_index; 1127 ifam->ifam_flags = ifa->ifa_flags; 1128 ifam->ifam_metric = ifa->ifa_metric; 1129 ifam->ifam_addrs = info.rti_addrs; 1130 error = SYSCTL_OUT(w->w_req, w->w_tmem, len); 1131 if (error) 1132 goto done; 1133 } 1134 } 1135 info.rti_info[RTAX_IFA] = info.rti_info[RTAX_NETMASK] = 1136 info.rti_info[RTAX_BRD] = NULL; 1137 } 1138 done: 1139 IFNET_RUNLOCK(); 1140 return (error); 1141 } 1142 1143 int 1144 sysctl_ifmalist(int af, struct walkarg *w) 1145 { 1146 struct ifnet *ifp; 1147 struct ifmultiaddr *ifma; 1148 struct rt_addrinfo info; 1149 int len, error = 0; 1150 struct ifaddr *ifa; 1151 1152 bzero((caddr_t)&info, sizeof(info)); 1153 IFNET_RLOCK(); 1154 TAILQ_FOREACH(ifp, &ifnet, if_link) { 1155 if (w->w_arg && w->w_arg != ifp->if_index) 1156 continue; 1157 ifa = ifaddr_byindex(ifp->if_index); 1158 info.rti_info[RTAX_IFP] = ifa ? ifa->ifa_addr : NULL; 1159 IF_ADDR_LOCK(ifp); 1160 TAILQ_FOREACH(ifma, &ifp->if_multiaddrs, ifma_link) { 1161 if (af && af != ifma->ifma_addr->sa_family) 1162 continue; 1163 if (jailed(curproc->p_ucred) && 1164 prison_if(curproc->p_ucred, ifma->ifma_addr)) 1165 continue; 1166 info.rti_info[RTAX_IFA] = ifma->ifma_addr; 1167 info.rti_info[RTAX_GATEWAY] = 1168 (ifma->ifma_addr->sa_family != AF_LINK) ? 1169 ifma->ifma_lladdr : NULL; 1170 len = rt_msg2(RTM_NEWMADDR, &info, NULL, w); 1171 if (w->w_req && w->w_tmem) { 1172 struct ifma_msghdr *ifmam; 1173 1174 ifmam = (struct ifma_msghdr *)w->w_tmem; 1175 ifmam->ifmam_index = ifma->ifma_ifp->if_index; 1176 ifmam->ifmam_flags = 0; 1177 ifmam->ifmam_addrs = info.rti_addrs; 1178 error = SYSCTL_OUT(w->w_req, w->w_tmem, len); 1179 if (error) { 1180 IF_ADDR_UNLOCK(ifp); 1181 goto done; 1182 } 1183 } 1184 } 1185 IF_ADDR_UNLOCK(ifp); 1186 } 1187 done: 1188 IFNET_RUNLOCK(); 1189 return (error); 1190 } 1191 1192 static int 1193 sysctl_rtsock(SYSCTL_HANDLER_ARGS) 1194 { 1195 int *name = (int *)arg1; 1196 u_int namelen = arg2; 1197 struct radix_node_head *rnh; 1198 int i, lim, error = EINVAL; 1199 u_char af; 1200 struct walkarg w; 1201 1202 name ++; 1203 namelen--; 1204 if (req->newptr) 1205 return (EPERM); 1206 if (namelen != 3) 1207 return ((namelen < 3) ? EISDIR : ENOTDIR); 1208 af = name[0]; 1209 if (af > AF_MAX) 1210 return (EINVAL); 1211 bzero(&w, sizeof(w)); 1212 w.w_op = name[1]; 1213 w.w_arg = name[2]; 1214 w.w_req = req; 1215 1216 error = sysctl_wire_old_buffer(req, 0); 1217 if (error) 1218 return (error); 1219 switch (w.w_op) { 1220 1221 case NET_RT_DUMP: 1222 case NET_RT_FLAGS: 1223 if (af == 0) { /* dump all tables */ 1224 i = 1; 1225 lim = AF_MAX; 1226 } else /* dump only one table */ 1227 i = lim = af; 1228 for (error = 0; error == 0 && i <= lim; i++) 1229 if ((rnh = rt_tables[i]) != NULL) { 1230 RADIX_NODE_HEAD_LOCK(rnh); 1231 error = rnh->rnh_walktree(rnh, 1232 sysctl_dumpentry, &w); 1233 RADIX_NODE_HEAD_UNLOCK(rnh); 1234 } else if (af != 0) 1235 error = EAFNOSUPPORT; 1236 break; 1237 1238 case NET_RT_IFLIST: 1239 error = sysctl_iflist(af, &w); 1240 break; 1241 1242 case NET_RT_IFMALIST: 1243 error = sysctl_ifmalist(af, &w); 1244 break; 1245 } 1246 if (w.w_tmem) 1247 free(w.w_tmem, M_RTABLE); 1248 return (error); 1249 } 1250 1251 SYSCTL_NODE(_net, PF_ROUTE, routetable, CTLFLAG_RD, sysctl_rtsock, ""); 1252 1253 /* 1254 * Definitions of protocols supported in the ROUTE domain. 1255 */ 1256 1257 static struct domain routedomain; /* or at least forward */ 1258 1259 static struct protosw routesw[] = { 1260 { SOCK_RAW, &routedomain, 0, PR_ATOMIC|PR_ADDR, 1261 0, route_output, raw_ctlinput, 0, 1262 0, 1263 raw_init, 0, 0, 0, 1264 &route_usrreqs 1265 } 1266 }; 1267 1268 static struct domain routedomain = 1269 { PF_ROUTE, "route", 0, 0, 0, 1270 routesw, &routesw[sizeof(routesw)/sizeof(routesw[0])] }; 1271 1272 DOMAIN_SET(route); 1273