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