1 /* 2 * Copyright (c) 2004, 2005 The DragonFly Project. All rights reserved. 3 * 4 * This code is derived from software contributed to The DragonFly Project 5 * by Jeffrey M. Hsu. 6 * 7 * Redistribution and use in source and binary forms, with or without 8 * modification, are permitted provided that the following conditions 9 * are met: 10 * 1. Redistributions of source code must retain the above copyright 11 * notice, this list of conditions and the following disclaimer. 12 * 2. Redistributions in binary form must reproduce the above copyright 13 * notice, this list of conditions and the following disclaimer in the 14 * documentation and/or other materials provided with the distribution. 15 * 3. Neither the name of The DragonFly Project nor the names of its 16 * contributors may be used to endorse or promote products derived 17 * from this software without specific, prior written permission. 18 * 19 * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS 20 * ``AS IS'' AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT 21 * LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS 22 * FOR A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE 23 * COPYRIGHT HOLDERS OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, 24 * INCIDENTAL, SPECIAL, EXEMPLARY OR CONSEQUENTIAL DAMAGES (INCLUDING, 25 * BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; 26 * LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED 27 * AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, 28 * OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT 29 * OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF 30 * SUCH DAMAGE. 31 */ 32 33 /* 34 * Copyright (c) 2004, 2005 Jeffrey M. Hsu. All rights reserved. 35 * 36 * License terms: all terms for the DragonFly license above plus the following: 37 * 38 * 4. All advertising materials mentioning features or use of this software 39 * must display the following acknowledgement: 40 * 41 * This product includes software developed by Jeffrey M. Hsu 42 * for the DragonFly Project. 43 * 44 * This requirement may be waived with permission from Jeffrey Hsu. 45 * Permission will be granted to any DragonFly user for free. 46 * This requirement will sunset and may be removed on Jan 31, 2006, 47 * after which the standard DragonFly license (as shown above) will 48 * apply. 49 */ 50 51 /* 52 * Copyright (c) 1988, 1991, 1993 53 * The Regents of the University of California. All rights reserved. 54 * 55 * Redistribution and use in source and binary forms, with or without 56 * modification, are permitted provided that the following conditions 57 * are met: 58 * 1. Redistributions of source code must retain the above copyright 59 * notice, this list of conditions and the following disclaimer. 60 * 2. Redistributions in binary form must reproduce the above copyright 61 * notice, this list of conditions and the following disclaimer in the 62 * documentation and/or other materials provided with the distribution. 63 * 3. All advertising materials mentioning features or use of this software 64 * must display the following acknowledgement: 65 * This product includes software developed by the University of 66 * California, Berkeley and its contributors. 67 * 4. Neither the name of the University nor the names of its contributors 68 * may be used to endorse or promote products derived from this software 69 * without specific prior written permission. 70 * 71 * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND 72 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE 73 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE 74 * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE 75 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL 76 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS 77 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) 78 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT 79 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY 80 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF 81 * SUCH DAMAGE. 82 * 83 * @(#)rtsock.c 8.7 (Berkeley) 10/12/95 84 * $FreeBSD: src/sys/net/rtsock.c,v 1.44.2.11 2002/12/04 14:05:41 ru Exp $ 85 * $DragonFly: src/sys/net/rtsock.c,v 1.28 2005/06/14 19:47:30 joerg Exp $ 86 */ 87 88 #include <sys/param.h> 89 #include <sys/systm.h> 90 #include <sys/kernel.h> 91 #include <sys/sysctl.h> 92 #include <sys/proc.h> 93 #include <sys/malloc.h> 94 #include <sys/mbuf.h> 95 #include <sys/protosw.h> 96 #include <sys/socket.h> 97 #include <sys/socketvar.h> 98 #include <sys/domain.h> 99 #include <sys/thread2.h> 100 101 #include <net/if.h> 102 #include <net/route.h> 103 #include <net/raw_cb.h> 104 105 MALLOC_DEFINE(M_RTABLE, "routetbl", "routing tables"); 106 107 static struct route_cb { 108 int ip_count; 109 int ip6_count; 110 int ipx_count; 111 int ns_count; 112 int any_count; 113 } route_cb; 114 115 static const struct sockaddr route_src = { 2, PF_ROUTE, }; 116 117 struct walkarg { 118 int w_tmemsize; 119 int w_op, w_arg; 120 void *w_tmem; 121 struct sysctl_req *w_req; 122 }; 123 124 static struct mbuf * 125 rt_msg_mbuf (int, struct rt_addrinfo *); 126 static void rt_msg_buffer (int, struct rt_addrinfo *, void *buf, int len); 127 static int rt_msgsize (int type, struct rt_addrinfo *rtinfo); 128 static int rt_xaddrs (char *, char *, struct rt_addrinfo *); 129 static int sysctl_dumpentry (struct radix_node *rn, void *vw); 130 static int sysctl_iflist (int af, struct walkarg *w); 131 static int route_output(struct mbuf *, struct socket *, ...); 132 static void rt_setmetrics (u_long, struct rt_metrics *, 133 struct rt_metrics *); 134 135 /* 136 * It really doesn't make any sense at all for this code to share much 137 * with raw_usrreq.c, since its functionality is so restricted. XXX 138 */ 139 static int 140 rts_abort(struct socket *so) 141 { 142 int error; 143 144 crit_enter(); 145 error = raw_usrreqs.pru_abort(so); 146 crit_exit(); 147 return error; 148 } 149 150 /* pru_accept is EOPNOTSUPP */ 151 152 static int 153 rts_attach(struct socket *so, int proto, struct pru_attach_info *ai) 154 { 155 struct rawcb *rp; 156 int error; 157 158 if (sotorawcb(so) != NULL) 159 return EISCONN; /* XXX panic? */ 160 161 rp = malloc(sizeof *rp, M_PCB, M_WAITOK | M_ZERO); 162 if (rp == NULL) 163 return ENOBUFS; 164 165 /* 166 * The critical section 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 critical section. 171 */ 172 crit_enter(); 173 so->so_pcb = rp; 174 error = raw_attach(so, proto, ai->sb_rlimit); 175 rp = sotorawcb(so); 176 if (error) { 177 crit_exit(); 178 free(rp, M_PCB); 179 return error; 180 } 181 switch(rp->rcb_proto.sp_protocol) { 182 case AF_INET: 183 route_cb.ip_count++; 184 break; 185 case AF_INET6: 186 route_cb.ip6_count++; 187 break; 188 case AF_IPX: 189 route_cb.ipx_count++; 190 break; 191 case AF_NS: 192 route_cb.ns_count++; 193 break; 194 } 195 rp->rcb_faddr = &route_src; 196 route_cb.any_count++; 197 soisconnected(so); 198 so->so_options |= SO_USELOOPBACK; 199 crit_exit(); 200 return 0; 201 } 202 203 static int 204 rts_bind(struct socket *so, struct sockaddr *nam, struct thread *td) 205 { 206 int error; 207 208 crit_enter(); 209 error = raw_usrreqs.pru_bind(so, nam, td); /* xxx just EINVAL */ 210 crit_exit(); 211 return error; 212 } 213 214 static int 215 rts_connect(struct socket *so, struct sockaddr *nam, struct thread *td) 216 { 217 int error; 218 219 crit_enter(); 220 error = raw_usrreqs.pru_connect(so, nam, td); /* XXX just EINVAL */ 221 crit_exit(); 222 return error; 223 } 224 225 /* pru_connect2 is EOPNOTSUPP */ 226 /* pru_control is EOPNOTSUPP */ 227 228 static int 229 rts_detach(struct socket *so) 230 { 231 struct rawcb *rp = sotorawcb(so); 232 int error; 233 234 crit_enter(); 235 if (rp != NULL) { 236 switch(rp->rcb_proto.sp_protocol) { 237 case AF_INET: 238 route_cb.ip_count--; 239 break; 240 case AF_INET6: 241 route_cb.ip6_count--; 242 break; 243 case AF_IPX: 244 route_cb.ipx_count--; 245 break; 246 case AF_NS: 247 route_cb.ns_count--; 248 break; 249 } 250 route_cb.any_count--; 251 } 252 error = raw_usrreqs.pru_detach(so); 253 crit_exit(); 254 return error; 255 } 256 257 static int 258 rts_disconnect(struct socket *so) 259 { 260 int error; 261 262 crit_enter(); 263 error = raw_usrreqs.pru_disconnect(so); 264 crit_exit(); 265 return error; 266 } 267 268 /* pru_listen is EOPNOTSUPP */ 269 270 static int 271 rts_peeraddr(struct socket *so, struct sockaddr **nam) 272 { 273 int error; 274 275 crit_enter(); 276 error = raw_usrreqs.pru_peeraddr(so, nam); 277 crit_exit(); 278 return error; 279 } 280 281 /* pru_rcvd is EOPNOTSUPP */ 282 /* pru_rcvoob is EOPNOTSUPP */ 283 284 static int 285 rts_send(struct socket *so, int flags, struct mbuf *m, struct sockaddr *nam, 286 struct mbuf *control, struct thread *td) 287 { 288 int error; 289 290 crit_enter(); 291 error = raw_usrreqs.pru_send(so, flags, m, nam, control, td); 292 crit_exit(); 293 return error; 294 } 295 296 /* pru_sense is null */ 297 298 static int 299 rts_shutdown(struct socket *so) 300 { 301 int error; 302 303 crit_enter(); 304 error = raw_usrreqs.pru_shutdown(so); 305 crit_exit(); 306 return error; 307 } 308 309 static int 310 rts_sockaddr(struct socket *so, struct sockaddr **nam) 311 { 312 int error; 313 314 crit_enter(); 315 error = raw_usrreqs.pru_sockaddr(so, nam); 316 crit_exit(); 317 return error; 318 } 319 320 static struct pr_usrreqs route_usrreqs = { 321 rts_abort, pru_accept_notsupp, rts_attach, rts_bind, rts_connect, 322 pru_connect2_notsupp, pru_control_notsupp, rts_detach, rts_disconnect, 323 pru_listen_notsupp, rts_peeraddr, pru_rcvd_notsupp, pru_rcvoob_notsupp, 324 rts_send, pru_sense_null, rts_shutdown, rts_sockaddr, 325 sosend, soreceive, sopoll 326 }; 327 328 static __inline sa_family_t 329 familyof(struct sockaddr *sa) 330 { 331 return (sa != NULL ? sa->sa_family : 0); 332 } 333 334 static void 335 rts_input(struct mbuf *m, sa_family_t family) 336 { 337 static const struct sockaddr route_dst = { 2, PF_ROUTE, }; 338 struct sockproto route_proto = { PF_ROUTE, family }; 339 340 raw_input(m, &route_proto, &route_src, &route_dst); 341 } 342 343 static void * 344 reallocbuf(void *ptr, size_t len, size_t olen) 345 { 346 void *newptr; 347 348 newptr = malloc(len, M_RTABLE, M_INTWAIT | M_NULLOK); 349 if (newptr == NULL) 350 return NULL; 351 bcopy(ptr, newptr, olen); 352 free(ptr, M_RTABLE); 353 return (newptr); 354 } 355 356 /* 357 * Internal helper routine for route_output(). 358 */ 359 static int 360 fillrtmsg(struct rt_msghdr **prtm, struct rtentry *rt, 361 struct rt_addrinfo *rtinfo) 362 { 363 int msglen; 364 struct rt_msghdr *rtm = *prtm; 365 366 /* Fill in rt_addrinfo for call to rt_msg_buffer(). */ 367 rtinfo->rti_dst = rt_key(rt); 368 rtinfo->rti_gateway = rt->rt_gateway; 369 rtinfo->rti_netmask = rt_mask(rt); /* might be NULL */ 370 rtinfo->rti_genmask = rt->rt_genmask; /* might be NULL */ 371 if (rtm->rtm_addrs & (RTA_IFP | RTA_IFA)) { 372 if (rt->rt_ifp != NULL) { 373 rtinfo->rti_ifpaddr = 374 TAILQ_FIRST(&rt->rt_ifp->if_addrhead)->ifa_addr; 375 rtinfo->rti_ifaaddr = rt->rt_ifa->ifa_addr; 376 if (rt->rt_ifp->if_flags & IFF_POINTOPOINT) 377 rtinfo->rti_bcastaddr = rt->rt_ifa->ifa_dstaddr; 378 rtm->rtm_index = rt->rt_ifp->if_index; 379 } else { 380 rtinfo->rti_ifpaddr = NULL; 381 rtinfo->rti_ifaaddr = NULL; 382 } 383 } 384 385 msglen = rt_msgsize(rtm->rtm_type, rtinfo); 386 if (rtm->rtm_msglen < msglen) { 387 rtm = reallocbuf(rtm, msglen, rtm->rtm_msglen); 388 if (rtm == NULL) 389 return (ENOBUFS); 390 *prtm = rtm; 391 } 392 rt_msg_buffer(rtm->rtm_type, rtinfo, rtm, msglen); 393 394 rtm->rtm_flags = rt->rt_flags; 395 rtm->rtm_rmx = rt->rt_rmx; 396 rtm->rtm_addrs = rtinfo->rti_addrs; 397 398 return (0); 399 } 400 401 /*ARGSUSED*/ 402 static int 403 route_output(struct mbuf *m, struct socket *so, ...) 404 { 405 struct rt_msghdr *rtm = NULL; 406 struct rtentry *rt = NULL; 407 struct rtentry *saved_nrt = NULL; 408 struct radix_node_head *rnh; 409 struct ifaddr *ifa = NULL; 410 struct rawcb *rp = NULL; 411 struct pr_output_info *oi; 412 struct rt_addrinfo rtinfo; 413 int len, error = 0; 414 __va_list ap; 415 416 __va_start(ap, so); 417 oi = __va_arg(ap, struct pr_output_info *); 418 __va_end(ap); 419 420 #define gotoerr(e) { error = e; goto flush;} 421 422 if (m == NULL || 423 (m->m_len < sizeof(long) && 424 (m = m_pullup(m, sizeof(long))) == NULL)) 425 return (ENOBUFS); 426 if (!(m->m_flags & M_PKTHDR)) 427 panic("route_output"); 428 len = m->m_pkthdr.len; 429 if (len < sizeof(struct rt_msghdr) || 430 len != mtod(m, struct rt_msghdr *)->rtm_msglen) { 431 rtinfo.rti_dst = NULL; 432 gotoerr(EINVAL); 433 } 434 rtm = malloc(len, M_RTABLE, M_INTWAIT | M_NULLOK); 435 if (rtm == NULL) { 436 rtinfo.rti_dst = NULL; 437 gotoerr(ENOBUFS); 438 } 439 m_copydata(m, 0, len, (caddr_t)rtm); 440 if (rtm->rtm_version != RTM_VERSION) { 441 rtinfo.rti_dst = NULL; 442 gotoerr(EPROTONOSUPPORT); 443 } 444 rtm->rtm_pid = oi->p_pid; 445 bzero(&rtinfo, sizeof(struct rt_addrinfo)); 446 rtinfo.rti_addrs = rtm->rtm_addrs; 447 if (rt_xaddrs((char *)(rtm + 1), (char *)rtm + len, &rtinfo) != 0) { 448 rtinfo.rti_dst = NULL; 449 gotoerr(EINVAL); 450 } 451 rtinfo.rti_flags = rtm->rtm_flags; 452 if (rtinfo.rti_dst == NULL || rtinfo.rti_dst->sa_family >= AF_MAX || 453 (rtinfo.rti_gateway && rtinfo.rti_gateway->sa_family >= AF_MAX)) 454 gotoerr(EINVAL); 455 456 if (rtinfo.rti_genmask != NULL) { 457 struct radix_node *n; 458 459 #define clen(s) (*(u_char *)(s)) 460 n = rn_addmask((char *)rtinfo.rti_genmask, TRUE, 1); 461 if (n != NULL && 462 rtinfo.rti_genmask->sa_len >= clen(n->rn_key) && 463 bcmp((char *)rtinfo.rti_genmask + 1, 464 (char *)n->rn_key + 1, clen(n->rn_key) - 1) == 0) 465 rtinfo.rti_genmask = (struct sockaddr *)n->rn_key; 466 else 467 gotoerr(ENOBUFS); 468 } 469 470 /* 471 * Verify that the caller has the appropriate privilege; RTM_GET 472 * is the only operation the non-superuser is allowed. 473 */ 474 if (rtm->rtm_type != RTM_GET && suser_cred(so->so_cred, 0) != 0) 475 gotoerr(EPERM); 476 477 switch (rtm->rtm_type) { 478 case RTM_ADD: 479 if (rtinfo.rti_gateway == NULL) 480 gotoerr(EINVAL); 481 error = rtrequest1(RTM_ADD, &rtinfo, &saved_nrt); 482 if (error == 0 && saved_nrt != NULL) { 483 rt_setmetrics(rtm->rtm_inits, &rtm->rtm_rmx, 484 &saved_nrt->rt_rmx); 485 saved_nrt->rt_rmx.rmx_locks &= ~(rtm->rtm_inits); 486 saved_nrt->rt_rmx.rmx_locks |= 487 (rtm->rtm_inits & rtm->rtm_rmx.rmx_locks); 488 --saved_nrt->rt_refcnt; 489 saved_nrt->rt_genmask = rtinfo.rti_genmask; 490 } 491 break; 492 case RTM_DELETE: 493 error = rtrequest1(RTM_DELETE, &rtinfo, &saved_nrt); 494 if (error == 0) { 495 if ((rt = saved_nrt)) 496 rt->rt_refcnt++; 497 if (fillrtmsg(&rtm, rt, &rtinfo) != 0) 498 gotoerr(ENOBUFS); 499 } 500 break; 501 case RTM_GET: 502 case RTM_CHANGE: 503 case RTM_LOCK: 504 if ((rnh = rt_tables[rtinfo.rti_dst->sa_family]) == NULL) 505 gotoerr(EAFNOSUPPORT); 506 rt = (struct rtentry *) 507 rnh->rnh_lookup((char *)rtinfo.rti_dst, 508 (char *)rtinfo.rti_netmask, rnh); 509 if (rt == NULL) 510 gotoerr(ESRCH); 511 rt->rt_refcnt++; 512 513 switch(rtm->rtm_type) { 514 case RTM_GET: 515 if (fillrtmsg(&rtm, rt, &rtinfo) != 0) 516 gotoerr(ENOBUFS); 517 break; 518 case RTM_CHANGE: 519 /* 520 * new gateway could require new ifaddr, ifp; 521 * flags may also be different; ifp may be specified 522 * by ll sockaddr when protocol address is ambiguous 523 */ 524 if (((rt->rt_flags & RTF_GATEWAY) && 525 rtinfo.rti_gateway != NULL) || 526 rtinfo.rti_ifpaddr != NULL || 527 (rtinfo.rti_ifaaddr != NULL && 528 sa_equal(rtinfo.rti_ifaaddr, 529 rt->rt_ifa->ifa_addr))) { 530 error = rt_getifa(&rtinfo); 531 if (error != 0) 532 gotoerr(error); 533 } 534 if (rtinfo.rti_gateway != NULL) { 535 error = rt_setgate(rt, rt_key(rt), 536 rtinfo.rti_gateway); 537 if (error != 0) 538 gotoerr(error); 539 } 540 if ((ifa = rtinfo.rti_ifa) != NULL) { 541 struct ifaddr *oifa = rt->rt_ifa; 542 543 if (oifa != ifa) { 544 if (oifa && oifa->ifa_rtrequest) 545 oifa->ifa_rtrequest(RTM_DELETE, 546 rt, &rtinfo); 547 IFAFREE(rt->rt_ifa); 548 IFAREF(ifa); 549 rt->rt_ifa = ifa; 550 rt->rt_ifp = rtinfo.rti_ifp; 551 } 552 } 553 rt_setmetrics(rtm->rtm_inits, &rtm->rtm_rmx, 554 &rt->rt_rmx); 555 if (rt->rt_ifa && rt->rt_ifa->ifa_rtrequest) 556 rt->rt_ifa->ifa_rtrequest(RTM_ADD, rt, &rtinfo); 557 if (rtinfo.rti_genmask != NULL) 558 rt->rt_genmask = rtinfo.rti_genmask; 559 /* 560 * Fall into 561 */ 562 case RTM_LOCK: 563 rt->rt_rmx.rmx_locks &= ~(rtm->rtm_inits); 564 rt->rt_rmx.rmx_locks |= 565 (rtm->rtm_inits & rtm->rtm_rmx.rmx_locks); 566 break; 567 } 568 569 break; 570 default: 571 gotoerr(EOPNOTSUPP); 572 } 573 574 flush: 575 if (rtm != NULL) { 576 if (error != 0) 577 rtm->rtm_errno = error; 578 else 579 rtm->rtm_flags |= RTF_DONE; 580 } 581 if (rt != NULL) 582 rtfree(rt); 583 /* 584 * Check to see if we don't want our own messages. 585 */ 586 if (!(so->so_options & SO_USELOOPBACK)) { 587 if (route_cb.any_count <= 1) { 588 if (rtm != NULL) 589 free(rtm, M_RTABLE); 590 m_freem(m); 591 return (error); 592 } 593 /* There is another listener, so construct message */ 594 rp = sotorawcb(so); 595 } 596 if (rtm != NULL) { 597 m_copyback(m, 0, rtm->rtm_msglen, (caddr_t)rtm); 598 if (m->m_pkthdr.len < rtm->rtm_msglen) { 599 m_freem(m); 600 m = NULL; 601 } else if (m->m_pkthdr.len > rtm->rtm_msglen) 602 m_adj(m, rtm->rtm_msglen - m->m_pkthdr.len); 603 free(rtm, M_RTABLE); 604 } 605 if (rp != NULL) 606 rp->rcb_proto.sp_family = 0; /* Avoid us */ 607 if (m != NULL) 608 rts_input(m, familyof(rtinfo.rti_dst)); 609 if (rp != NULL) 610 rp->rcb_proto.sp_family = PF_ROUTE; 611 return (error); 612 } 613 614 static void 615 rt_setmetrics(u_long which, struct rt_metrics *in, struct rt_metrics *out) 616 { 617 #define setmetric(flag, elt) if (which & (flag)) out->elt = in->elt; 618 setmetric(RTV_RPIPE, rmx_recvpipe); 619 setmetric(RTV_SPIPE, rmx_sendpipe); 620 setmetric(RTV_SSTHRESH, rmx_ssthresh); 621 setmetric(RTV_RTT, rmx_rtt); 622 setmetric(RTV_RTTVAR, rmx_rttvar); 623 setmetric(RTV_HOPCOUNT, rmx_hopcount); 624 setmetric(RTV_MTU, rmx_mtu); 625 setmetric(RTV_EXPIRE, rmx_expire); 626 #undef setmetric 627 } 628 629 #define ROUNDUP(a) \ 630 ((a) > 0 ? (1 + (((a) - 1) | (sizeof(long) - 1))) : sizeof(long)) 631 632 /* 633 * Extract the addresses of the passed sockaddrs. 634 * Do a little sanity checking so as to avoid bad memory references. 635 * This data is derived straight from userland. 636 */ 637 static int 638 rt_xaddrs(char *cp, char *cplim, struct rt_addrinfo *rtinfo) 639 { 640 struct sockaddr *sa; 641 int i; 642 643 for (i = 0; (i < RTAX_MAX) && (cp < cplim); i++) { 644 if ((rtinfo->rti_addrs & (1 << i)) == 0) 645 continue; 646 sa = (struct sockaddr *)cp; 647 /* 648 * It won't fit. 649 */ 650 if ((cp + sa->sa_len) > cplim) { 651 return (EINVAL); 652 } 653 654 /* 655 * There are no more... Quit now. 656 * If there are more bits, they are in error. 657 * I've seen this. route(1) can evidently generate these. 658 * This causes kernel to core dump. 659 * For compatibility, if we see this, point to a safe address. 660 */ 661 if (sa->sa_len == 0) { 662 static struct sockaddr sa_zero = { 663 sizeof sa_zero, AF_INET, 664 }; 665 666 rtinfo->rti_info[i] = &sa_zero; 667 return (0); /* should be EINVAL but for compat */ 668 } 669 670 /* Accept the sockaddr. */ 671 rtinfo->rti_info[i] = sa; 672 cp += ROUNDUP(sa->sa_len); 673 } 674 return (0); 675 } 676 677 static int 678 rt_msghdrsize(int type) 679 { 680 switch (type) { 681 case RTM_DELADDR: 682 case RTM_NEWADDR: 683 return sizeof(struct ifa_msghdr); 684 case RTM_DELMADDR: 685 case RTM_NEWMADDR: 686 return sizeof(struct ifma_msghdr); 687 case RTM_IFINFO: 688 return sizeof(struct if_msghdr); 689 case RTM_IFANNOUNCE: 690 return sizeof(struct if_announcemsghdr); 691 default: 692 return sizeof(struct rt_msghdr); 693 } 694 } 695 696 static int 697 rt_msgsize(int type, struct rt_addrinfo *rtinfo) 698 { 699 int len, i; 700 701 len = rt_msghdrsize(type); 702 for (i = 0; i < RTAX_MAX; i++) { 703 if (rtinfo->rti_info[i] != NULL) 704 len += ROUNDUP(rtinfo->rti_info[i]->sa_len); 705 } 706 len = ALIGN(len); 707 return len; 708 } 709 710 /* 711 * Build a routing message in a buffer. 712 * Copy the addresses in the rtinfo->rti_info[] sockaddr array 713 * to the end of the buffer after the message header. 714 * 715 * Set the rtinfo->rti_addrs bitmask of addresses present in rtinfo->rti_info[]. 716 * This side-effect can be avoided if we reorder the addrs bitmask field in all 717 * the route messages to line up so we can set it here instead of back in the 718 * calling routine. 719 */ 720 static void 721 rt_msg_buffer(int type, struct rt_addrinfo *rtinfo, void *buf, int msglen) 722 { 723 struct rt_msghdr *rtm; 724 char *cp; 725 int dlen, i; 726 727 rtm = (struct rt_msghdr *) buf; 728 rtm->rtm_version = RTM_VERSION; 729 rtm->rtm_type = type; 730 rtm->rtm_msglen = msglen; 731 732 cp = (char *)buf + rt_msghdrsize(type); 733 rtinfo->rti_addrs = 0; 734 for (i = 0; i < RTAX_MAX; i++) { 735 struct sockaddr *sa; 736 737 if ((sa = rtinfo->rti_info[i]) == NULL) 738 continue; 739 rtinfo->rti_addrs |= (1 << i); 740 dlen = ROUNDUP(sa->sa_len); 741 bcopy(sa, cp, dlen); 742 cp += dlen; 743 } 744 } 745 746 /* 747 * Build a routing message in a mbuf chain. 748 * Copy the addresses in the rtinfo->rti_info[] sockaddr array 749 * to the end of the mbuf after the message header. 750 * 751 * Set the rtinfo->rti_addrs bitmask of addresses present in rtinfo->rti_info[]. 752 * This side-effect can be avoided if we reorder the addrs bitmask field in all 753 * the route messages to line up so we can set it here instead of back in the 754 * calling routine. 755 */ 756 static struct mbuf * 757 rt_msg_mbuf(int type, struct rt_addrinfo *rtinfo) 758 { 759 struct mbuf *m; 760 struct rt_msghdr *rtm; 761 int hlen, len; 762 int i; 763 764 hlen = rt_msghdrsize(type); 765 KASSERT(hlen <= MCLBYTES, ("rt_msg_mbuf: hlen %d doesn't fit", hlen)); 766 767 m = m_getl(hlen, MB_DONTWAIT, MT_DATA, M_PKTHDR, NULL); 768 if (m == NULL) 769 return (NULL); 770 m->m_pkthdr.len = m->m_len = hlen; 771 m->m_pkthdr.rcvif = NULL; 772 rtinfo->rti_addrs = 0; 773 len = hlen; 774 for (i = 0; i < RTAX_MAX; i++) { 775 struct sockaddr *sa; 776 int dlen; 777 778 if ((sa = rtinfo->rti_info[i]) == NULL) 779 continue; 780 rtinfo->rti_addrs |= (1 << i); 781 dlen = ROUNDUP(sa->sa_len); 782 m_copyback(m, len, dlen, (caddr_t)sa); /* can grow mbuf chain */ 783 len += dlen; 784 } 785 if (m->m_pkthdr.len != len) { /* one of the m_copyback() calls failed */ 786 m_freem(m); 787 return (NULL); 788 } 789 rtm = mtod(m, struct rt_msghdr *); 790 bzero(rtm, hlen); 791 rtm->rtm_msglen = len; 792 rtm->rtm_version = RTM_VERSION; 793 rtm->rtm_type = type; 794 return (m); 795 } 796 797 /* 798 * This routine is called to generate a message from the routing 799 * socket indicating that a redirect has occurred, 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 sockaddr *dst = rtinfo->rti_info[RTAX_DST]; 807 struct rt_msghdr *rtm; 808 struct mbuf *m; 809 810 if (route_cb.any_count == 0) 811 return; 812 m = rt_msg_mbuf(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 rts_input(m, familyof(dst)); 820 } 821 822 void 823 rt_dstmsg(int type, struct sockaddr *dst, int error) 824 { 825 struct rt_msghdr *rtm; 826 struct rt_addrinfo addrs; 827 struct mbuf *m; 828 829 if (route_cb.any_count == 0) 830 return; 831 bzero(&addrs, sizeof(struct rt_addrinfo)); 832 addrs.rti_info[RTAX_DST] = dst; 833 m = rt_msg_mbuf(type, &addrs); 834 if (m == NULL) 835 return; 836 rtm = mtod(m, struct rt_msghdr *); 837 rtm->rtm_flags = RTF_DONE; 838 rtm->rtm_errno = error; 839 rtm->rtm_addrs = addrs.rti_addrs; 840 rts_input(m, familyof(dst)); 841 } 842 843 /* 844 * This routine is called to generate a message from the routing 845 * socket indicating that the status of a network interface has changed. 846 */ 847 void 848 rt_ifmsg(struct ifnet *ifp) 849 { 850 struct if_msghdr *ifm; 851 struct mbuf *m; 852 struct rt_addrinfo rtinfo; 853 854 if (route_cb.any_count == 0) 855 return; 856 bzero(&rtinfo, sizeof(struct rt_addrinfo)); 857 m = rt_msg_mbuf(RTM_IFINFO, &rtinfo); 858 if (m == NULL) 859 return; 860 ifm = mtod(m, struct if_msghdr *); 861 ifm->ifm_index = ifp->if_index; 862 ifm->ifm_flags = ifp->if_flags; 863 ifm->ifm_data = ifp->if_data; 864 ifm->ifm_addrs = 0; 865 rts_input(m, 0); 866 } 867 868 static void 869 rt_ifamsg(int cmd, struct ifaddr *ifa) 870 { 871 struct ifa_msghdr *ifam; 872 struct rt_addrinfo rtinfo; 873 struct mbuf *m; 874 struct ifnet *ifp = ifa->ifa_ifp; 875 876 bzero(&rtinfo, sizeof(struct rt_addrinfo)); 877 rtinfo.rti_ifaaddr = ifa->ifa_addr; 878 rtinfo.rti_ifpaddr = TAILQ_FIRST(&ifp->if_addrhead)->ifa_addr; 879 rtinfo.rti_netmask = ifa->ifa_netmask; 880 rtinfo.rti_bcastaddr = ifa->ifa_dstaddr; 881 882 m = rt_msg_mbuf(cmd, &rtinfo); 883 if (m == NULL) 884 return; 885 886 ifam = mtod(m, struct ifa_msghdr *); 887 ifam->ifam_index = ifp->if_index; 888 ifam->ifam_metric = ifa->ifa_metric; 889 ifam->ifam_flags = ifa->ifa_flags; 890 ifam->ifam_addrs = rtinfo.rti_addrs; 891 892 rts_input(m, familyof(ifa->ifa_addr)); 893 } 894 895 void 896 rt_rtmsg(int cmd, struct rtentry *rt, struct ifnet *ifp, int error) 897 { 898 struct rt_msghdr *rtm; 899 struct rt_addrinfo rtinfo; 900 struct mbuf *m; 901 struct sockaddr *dst; 902 903 if (rt == NULL) 904 return; 905 906 bzero(&rtinfo, sizeof(struct rt_addrinfo)); 907 rtinfo.rti_dst = dst = rt_key(rt); 908 rtinfo.rti_gateway = rt->rt_gateway; 909 rtinfo.rti_netmask = rt_mask(rt); 910 if (ifp != NULL) 911 rtinfo.rti_ifpaddr = TAILQ_FIRST(&ifp->if_addrhead)->ifa_addr; 912 rtinfo.rti_ifaaddr = rt->rt_ifa->ifa_addr; 913 914 m = rt_msg_mbuf(cmd, &rtinfo); 915 if (m == NULL) 916 return; 917 918 rtm = mtod(m, struct rt_msghdr *); 919 if (ifp != NULL) 920 rtm->rtm_index = ifp->if_index; 921 rtm->rtm_flags |= rt->rt_flags; 922 rtm->rtm_errno = error; 923 rtm->rtm_addrs = rtinfo.rti_addrs; 924 925 rts_input(m, familyof(dst)); 926 } 927 928 /* 929 * This is called to generate messages from the routing socket 930 * indicating a network interface has had addresses associated with it. 931 * if we ever reverse the logic and replace messages TO the routing 932 * socket indicate a request to configure interfaces, then it will 933 * be unnecessary as the routing socket will automatically generate 934 * copies of it. 935 */ 936 void 937 rt_newaddrmsg(int cmd, struct ifaddr *ifa, int error, struct rtentry *rt) 938 { 939 if (route_cb.any_count == 0) 940 return; 941 942 if (cmd == RTM_ADD) { 943 rt_ifamsg(RTM_NEWADDR, ifa); 944 rt_rtmsg(RTM_ADD, rt, ifa->ifa_ifp, error); 945 } else { 946 KASSERT((cmd == RTM_DELETE), ("unknown cmd %d", cmd)); 947 rt_rtmsg(RTM_DELETE, rt, ifa->ifa_ifp, error); 948 rt_ifamsg(RTM_DELADDR, ifa); 949 } 950 } 951 952 /* 953 * This is the analogue to the rt_newaddrmsg which performs the same 954 * function but for multicast group memberhips. This is easier since 955 * there is no route state to worry about. 956 */ 957 void 958 rt_newmaddrmsg(int cmd, struct ifmultiaddr *ifma) 959 { 960 struct rt_addrinfo rtinfo; 961 struct mbuf *m = NULL; 962 struct ifnet *ifp = ifma->ifma_ifp; 963 struct ifma_msghdr *ifmam; 964 965 if (route_cb.any_count == 0) 966 return; 967 968 bzero(&rtinfo, sizeof(struct rt_addrinfo)); 969 rtinfo.rti_ifaaddr = ifma->ifma_addr; 970 if (ifp != NULL && !TAILQ_EMPTY(&ifp->if_addrhead)) 971 rtinfo.rti_ifpaddr = TAILQ_FIRST(&ifp->if_addrhead)->ifa_addr; 972 /* 973 * If a link-layer address is present, present it as a ``gateway'' 974 * (similarly to how ARP entries, e.g., are presented). 975 */ 976 rtinfo.rti_gateway = ifma->ifma_lladdr; 977 978 m = rt_msg_mbuf(cmd, &rtinfo); 979 if (m == NULL) 980 return; 981 982 ifmam = mtod(m, struct ifma_msghdr *); 983 ifmam->ifmam_index = ifp->if_index; 984 ifmam->ifmam_addrs = rtinfo.rti_addrs; 985 986 rts_input(m, familyof(ifma->ifma_addr)); 987 } 988 989 /* 990 * This is called to generate routing socket messages indicating 991 * network interface arrival and departure. 992 */ 993 void 994 rt_ifannouncemsg(struct ifnet *ifp, int what) 995 { 996 struct rt_addrinfo addrinfo; 997 struct mbuf *m; 998 struct if_announcemsghdr *ifan; 999 1000 if (route_cb.any_count == 0) 1001 return; 1002 1003 bzero(&addrinfo, sizeof addrinfo); 1004 m = rt_msg_mbuf(RTM_IFANNOUNCE, &addrinfo); 1005 if (m == NULL) 1006 return; 1007 1008 ifan = mtod(m, struct if_announcemsghdr *); 1009 ifan->ifan_index = ifp->if_index; 1010 strlcpy(ifan->ifan_name, ifp->if_xname, sizeof ifan->ifan_name); 1011 ifan->ifan_what = what; 1012 1013 rts_input(m, 0); 1014 } 1015 1016 static int 1017 resizewalkarg(struct walkarg *w, int len) 1018 { 1019 void *newptr; 1020 1021 newptr = malloc(len, M_RTABLE, M_INTWAIT | M_NULLOK); 1022 if (newptr == NULL) 1023 return (ENOMEM); 1024 if (w->w_tmem != NULL) 1025 free(w->w_tmem, M_RTABLE); 1026 w->w_tmem = newptr; 1027 w->w_tmemsize = len; 1028 return (0); 1029 } 1030 1031 /* 1032 * This is used in dumping the kernel table via sysctl(). 1033 */ 1034 int 1035 sysctl_dumpentry(struct radix_node *rn, void *vw) 1036 { 1037 struct walkarg *w = vw; 1038 struct rtentry *rt = (struct rtentry *)rn; 1039 struct rt_addrinfo rtinfo; 1040 int error, msglen; 1041 1042 if (w->w_op == NET_RT_FLAGS && !(rt->rt_flags & w->w_arg)) 1043 return 0; 1044 1045 bzero(&rtinfo, sizeof(struct rt_addrinfo)); 1046 rtinfo.rti_dst = rt_key(rt); 1047 rtinfo.rti_gateway = rt->rt_gateway; 1048 rtinfo.rti_netmask = rt_mask(rt); 1049 rtinfo.rti_genmask = rt->rt_genmask; 1050 if (rt->rt_ifp != NULL) { 1051 rtinfo.rti_ifpaddr = 1052 TAILQ_FIRST(&rt->rt_ifp->if_addrhead)->ifa_addr; 1053 rtinfo.rti_ifaaddr = rt->rt_ifa->ifa_addr; 1054 if (rt->rt_ifp->if_flags & IFF_POINTOPOINT) 1055 rtinfo.rti_bcastaddr = rt->rt_ifa->ifa_dstaddr; 1056 } 1057 msglen = rt_msgsize(RTM_GET, &rtinfo); 1058 if (w->w_tmemsize < msglen && resizewalkarg(w, msglen) != 0) 1059 return (ENOMEM); 1060 rt_msg_buffer(RTM_GET, &rtinfo, w->w_tmem, msglen); 1061 if (w->w_req != NULL) { 1062 struct rt_msghdr *rtm = w->w_tmem; 1063 1064 rtm->rtm_flags = rt->rt_flags; 1065 rtm->rtm_use = rt->rt_use; 1066 rtm->rtm_rmx = rt->rt_rmx; 1067 rtm->rtm_index = rt->rt_ifp->if_index; 1068 rtm->rtm_errno = rtm->rtm_pid = rtm->rtm_seq = 0; 1069 rtm->rtm_addrs = rtinfo.rti_addrs; 1070 error = SYSCTL_OUT(w->w_req, rtm, msglen); 1071 return (error); 1072 } 1073 return (0); 1074 } 1075 1076 static int 1077 sysctl_iflist(int af, struct walkarg *w) 1078 { 1079 struct ifnet *ifp; 1080 struct ifaddr *ifa; 1081 struct rt_addrinfo rtinfo; 1082 int msglen, error; 1083 1084 bzero(&rtinfo, sizeof(struct rt_addrinfo)); 1085 TAILQ_FOREACH(ifp, &ifnet, if_link) { 1086 if (w->w_arg && w->w_arg != ifp->if_index) 1087 continue; 1088 ifa = TAILQ_FIRST(&ifp->if_addrhead); 1089 rtinfo.rti_ifpaddr = ifa->ifa_addr; 1090 msglen = rt_msgsize(RTM_IFINFO, &rtinfo); 1091 if (w->w_tmemsize < msglen && resizewalkarg(w, msglen) != 0) 1092 return (ENOMEM); 1093 rt_msg_buffer(RTM_IFINFO, &rtinfo, w->w_tmem, msglen); 1094 rtinfo.rti_ifpaddr = NULL; 1095 if (w->w_req != NULL && w->w_tmem != NULL) { 1096 struct if_msghdr *ifm = w->w_tmem; 1097 1098 ifm->ifm_index = ifp->if_index; 1099 ifm->ifm_flags = ifp->if_flags; 1100 ifm->ifm_data = ifp->if_data; 1101 ifm->ifm_addrs = rtinfo.rti_addrs; 1102 error = SYSCTL_OUT(w->w_req, ifm, msglen); 1103 if (error) 1104 return (error); 1105 } 1106 while ((ifa = TAILQ_NEXT(ifa, ifa_link)) != NULL) { 1107 if (af && af != ifa->ifa_addr->sa_family) 1108 continue; 1109 if (curproc->p_ucred->cr_prison && 1110 prison_if(curthread, ifa->ifa_addr)) 1111 continue; 1112 rtinfo.rti_ifaaddr = ifa->ifa_addr; 1113 rtinfo.rti_netmask = ifa->ifa_netmask; 1114 rtinfo.rti_bcastaddr = ifa->ifa_dstaddr; 1115 msglen = rt_msgsize(RTM_NEWADDR, &rtinfo); 1116 if (w->w_tmemsize < msglen && 1117 resizewalkarg(w, msglen) != 0) 1118 return (ENOMEM); 1119 rt_msg_buffer(RTM_NEWADDR, &rtinfo, w->w_tmem, msglen); 1120 if (w->w_req != NULL) { 1121 struct ifa_msghdr *ifam = w->w_tmem; 1122 1123 ifam->ifam_index = ifa->ifa_ifp->if_index; 1124 ifam->ifam_flags = ifa->ifa_flags; 1125 ifam->ifam_metric = ifa->ifa_metric; 1126 ifam->ifam_addrs = rtinfo.rti_addrs; 1127 error = SYSCTL_OUT(w->w_req, w->w_tmem, msglen); 1128 if (error) 1129 return (error); 1130 } 1131 } 1132 rtinfo.rti_netmask = NULL; 1133 rtinfo.rti_ifaaddr = NULL; 1134 rtinfo.rti_bcastaddr = NULL; 1135 } 1136 return (0); 1137 } 1138 1139 static int 1140 sysctl_rtsock(SYSCTL_HANDLER_ARGS) 1141 { 1142 int *name = (int *)arg1; 1143 u_int namelen = arg2; 1144 struct radix_node_head *rnh; 1145 int i, error = EINVAL; 1146 u_char af; 1147 struct walkarg w; 1148 1149 name ++; 1150 namelen--; 1151 if (req->newptr) 1152 return (EPERM); 1153 if (namelen != 3) 1154 return (EINVAL); 1155 af = name[0]; 1156 bzero(&w, sizeof w); 1157 w.w_op = name[1]; 1158 w.w_arg = name[2]; 1159 w.w_req = req; 1160 1161 crit_enter(); 1162 switch (w.w_op) { 1163 1164 case NET_RT_DUMP: 1165 case NET_RT_FLAGS: 1166 for (i = 1; i <= AF_MAX; i++) 1167 if ((rnh = rt_tables[i]) && (af == 0 || af == i) && 1168 (error = rnh->rnh_walktree(rnh, 1169 sysctl_dumpentry, &w))) 1170 break; 1171 break; 1172 1173 case NET_RT_IFLIST: 1174 error = sysctl_iflist(af, &w); 1175 } 1176 crit_exit(); 1177 if (w.w_tmem != NULL) 1178 free(w.w_tmem, M_RTABLE); 1179 return (error); 1180 } 1181 1182 SYSCTL_NODE(_net, PF_ROUTE, routetable, CTLFLAG_RD, sysctl_rtsock, ""); 1183 1184 /* 1185 * Definitions of protocols supported in the ROUTE domain. 1186 */ 1187 1188 extern struct domain routedomain; /* or at least forward */ 1189 1190 static struct protosw routesw[] = { 1191 { SOCK_RAW, &routedomain, 0, PR_ATOMIC|PR_ADDR, 1192 0, route_output, raw_ctlinput, 0, 1193 cpu0_soport, 1194 raw_init, 0, 0, 0, 1195 &route_usrreqs 1196 } 1197 }; 1198 1199 static struct domain routedomain = { 1200 PF_ROUTE, "route", NULL, NULL, NULL, 1201 routesw, &routesw[(sizeof routesw)/(sizeof routesw[0])], 1202 }; 1203 1204 DOMAIN_SET(route); 1205