Now available: The Design and Implementation of the FreeBSD Operating System (Second Edition)
[ source navigation ] [ diff markup ] [ identifier search ] [ freetext search ] [ file search ] [ list types ] [ track identifier ]
FreeBSD/Linux Kernel Cross Reference
sys/bsd/net/rtsock.c
Version:
- FREEBSD - FREEBSD-13-STABLE - FREEBSD-13-0 - FREEBSD-12-STABLE - FREEBSD-12-0 - FREEBSD-11-STABLE - FREEBSD-11-0 - FREEBSD-10-STABLE - FREEBSD-10-0 - FREEBSD-9-STABLE - FREEBSD-9-0 - FREEBSD-8-STABLE - FREEBSD-8-0 - FREEBSD-7-STABLE - FREEBSD-7-0 - FREEBSD-6-STABLE - FREEBSD-6-0 - FREEBSD-5-STABLE - FREEBSD-5-0 - FREEBSD-4-STABLE - FREEBSD-3-STABLE - FREEBSD22 - l41 - OPENBSD - linux-2.6 - MK84 - PLAN9 - xnu-8792
SearchContext: - none - 3 - 10
SearchContext: - none - 3 - 10
1 /* 2 * Copyright (c) 2000 Apple Computer, Inc. All rights reserved. 3 * 4 * @APPLE_LICENSE_HEADER_START@ 5 * 6 * Copyright (c) 1999-2003 Apple Computer, Inc. All Rights Reserved. 7 * 8 * This file contains Original Code and/or Modifications of Original Code 9 * as defined in and that are subject to the Apple Public Source License 10 * Version 2.0 (the 'License'). You may not use this file except in 11 * compliance with the License. Please obtain a copy of the License at 12 * http://www.opensource.apple.com/apsl/ and read it before using this 13 * file. 14 * 15 * The Original Code and all software distributed under the License are 16 * distributed on an 'AS IS' basis, WITHOUT WARRANTY OF ANY KIND, EITHER 17 * EXPRESS OR IMPLIED, AND APPLE HEREBY DISCLAIMS ALL SUCH WARRANTIES, 18 * INCLUDING WITHOUT LIMITATION, ANY WARRANTIES OF MERCHANTABILITY, 19 * FITNESS FOR A PARTICULAR PURPOSE, QUIET ENJOYMENT OR NON-INFRINGEMENT. 20 * Please see the License for the specific language governing rights and 21 * limitations under the License. 22 * 23 * @APPLE_LICENSE_HEADER_END@ 24 */ 25 /* 26 * Copyright (c) 1988, 1991, 1993 27 * The Regents of the University of California. All rights reserved. 28 * 29 * Redistribution and use in source and binary forms, with or without 30 * modification, are permitted provided that the following conditions 31 * are met: 32 * 1. Redistributions of source code must retain the above copyright 33 * notice, this list of conditions and the following disclaimer. 34 * 2. Redistributions in binary form must reproduce the above copyright 35 * notice, this list of conditions and the following disclaimer in the 36 * documentation and/or other materials provided with the distribution. 37 * 3. All advertising materials mentioning features or use of this software 38 * must display the following acknowledgement: 39 * This product includes software developed by the University of 40 * California, Berkeley and its contributors. 41 * 4. Neither the name of the University nor the names of its contributors 42 * may be used to endorse or promote products derived from this software 43 * without specific prior written permission. 44 * 45 * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND 46 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE 47 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE 48 * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE 49 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL 50 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS 51 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) 52 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT 53 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY 54 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF 55 * SUCH DAMAGE. 56 * 57 * @(#)rtsock.c 8.5 (Berkeley) 11/2/94 58 */ 59 60 61 #include <sys/param.h> 62 #include <sys/systm.h> 63 #include <sys/kernel.h> 64 #include <sys/sysctl.h> 65 #include <sys/proc.h> 66 #include <sys/malloc.h> 67 #include <sys/mbuf.h> 68 #include <sys/socket.h> 69 #include <sys/socketvar.h> 70 #include <sys/domain.h> 71 #include <sys/protosw.h> 72 #include <sys/syslog.h> 73 74 #include <net/if.h> 75 #include <net/route.h> 76 #include <net/raw_cb.h> 77 #include <netinet/in.h> 78 79 MALLOC_DEFINE(M_RTABLE, "routetbl", "routing tables"); 80 81 static struct sockaddr route_dst = { 2, PF_ROUTE, }; 82 static struct sockaddr route_src = { 2, PF_ROUTE, }; 83 static struct sockaddr sa_zero = { sizeof(sa_zero), AF_INET, }; 84 static struct sockproto route_proto = { PF_ROUTE, }; 85 86 struct walkarg { 87 int w_tmemsize; 88 int w_op, w_arg; 89 caddr_t w_tmem; 90 struct sysctl_req *w_req; 91 }; 92 93 static struct mbuf * 94 rt_msg1 __P((int, struct rt_addrinfo *)); 95 static int rt_msg2 __P((int, 96 struct rt_addrinfo *, caddr_t, struct walkarg *)); 97 static int rt_xaddrs __P((caddr_t, caddr_t, struct rt_addrinfo *)); 98 static int sysctl_dumpentry __P((struct radix_node *rn, void *vw)); 99 static int sysctl_iflist __P((int af, struct walkarg *w)); 100 static int route_output __P((struct mbuf *, struct socket *)); 101 static void rt_setmetrics __P((u_long, struct rt_metrics *, struct rt_metrics *)); 102 static void rt_setif __P((struct rtentry *, struct sockaddr *, struct sockaddr *, 103 struct sockaddr *)); 104 105 /* Sleazy use of local variables throughout file, warning!!!! */ 106 #define dst info.rti_info[RTAX_DST] 107 #define gate info.rti_info[RTAX_GATEWAY] 108 #define netmask info.rti_info[RTAX_NETMASK] 109 #define genmask info.rti_info[RTAX_GENMASK] 110 #define ifpaddr info.rti_info[RTAX_IFP] 111 #define ifaaddr info.rti_info[RTAX_IFA] 112 #define brdaddr info.rti_info[RTAX_BRD] 113 114 /* 115 * It really doesn't make any sense at all for this code to share much 116 * with raw_usrreq.c, since its functionality is so restricted. XXX 117 */ 118 static int 119 rts_abort(struct socket *so) 120 { 121 int s, error; 122 s = splnet(); 123 error = raw_usrreqs.pru_abort(so); 124 splx(s); 125 return error; 126 } 127 128 /* pru_accept is EOPNOTSUPP */ 129 130 static int 131 rts_attach(struct socket *so, int proto, struct proc *p) 132 { 133 struct rawcb *rp; 134 int s, error; 135 136 if (sotorawcb(so) != 0) 137 return EISCONN; /* XXX panic? */ 138 MALLOC(rp, struct rawcb *, sizeof *rp, M_PCB, M_WAITOK); /* XXX */ 139 if (rp == 0) 140 return ENOBUFS; 141 bzero(rp, sizeof *rp); 142 143 /* 144 * The splnet() is necessary to block protocols from sending 145 * error notifications (like RTM_REDIRECT or RTM_LOSING) while 146 * this PCB is extant but incompletely initialized. 147 * Probably we should try to do more of this work beforehand and 148 * eliminate the spl. 149 */ 150 s = splnet(); 151 so->so_pcb = (caddr_t)rp; 152 error = raw_attach(so, proto); /* don't use raw_usrreqs.pru_attach, it checks for SS_PRIV */ 153 rp = sotorawcb(so); 154 if (error) { 155 splx(s); 156 FREE(rp, M_PCB); 157 so->so_pcb = 0; 158 return error; 159 } 160 switch(rp->rcb_proto.sp_protocol) { 161 case AF_INET: 162 route_cb.ip_count++; 163 break; 164 case AF_INET6: 165 route_cb.ip6_count++; 166 break; 167 case AF_IPX: 168 route_cb.ipx_count++; 169 break; 170 case AF_NS: 171 route_cb.ns_count++; 172 break; 173 } 174 rp->rcb_faddr = &route_src; 175 route_cb.any_count++; 176 soisconnected(so); 177 so->so_options |= SO_USELOOPBACK; 178 splx(s); 179 return 0; 180 } 181 182 static int 183 rts_bind(struct socket *so, struct sockaddr *nam, struct proc *p) 184 { 185 int s, error; 186 s = splnet(); 187 error = raw_usrreqs.pru_bind(so, nam, p); /* xxx just EINVAL */ 188 splx(s); 189 return error; 190 } 191 192 static int 193 rts_connect(struct socket *so, struct sockaddr *nam, struct proc *p) 194 { 195 int s, error; 196 s = splnet(); 197 error = raw_usrreqs.pru_connect(so, nam, p); /* XXX just EINVAL */ 198 splx(s); 199 return error; 200 } 201 202 /* pru_connect2 is EOPNOTSUPP */ 203 /* pru_control is EOPNOTSUPP */ 204 205 static int 206 rts_detach(struct socket *so) 207 { 208 struct rawcb *rp = sotorawcb(so); 209 int s, error; 210 211 s = splnet(); 212 if (rp != 0) { 213 switch(rp->rcb_proto.sp_protocol) { 214 case AF_INET: 215 route_cb.ip_count--; 216 break; 217 case AF_INET6: 218 route_cb.ip6_count--; 219 break; 220 case AF_IPX: 221 route_cb.ipx_count--; 222 break; 223 case AF_NS: 224 route_cb.ns_count--; 225 break; 226 } 227 route_cb.any_count--; 228 } 229 error = raw_usrreqs.pru_detach(so); 230 splx(s); 231 return error; 232 } 233 234 static int 235 rts_disconnect(struct socket *so) 236 { 237 int s, error; 238 s = splnet(); 239 error = raw_usrreqs.pru_disconnect(so); 240 splx(s); 241 return error; 242 } 243 244 /* pru_listen is EOPNOTSUPP */ 245 246 static int 247 rts_peeraddr(struct socket *so, struct sockaddr **nam) 248 { 249 int s, error; 250 s = splnet(); 251 error = raw_usrreqs.pru_peeraddr(so, nam); 252 splx(s); 253 return error; 254 } 255 256 /* pru_rcvd is EOPNOTSUPP */ 257 /* pru_rcvoob is EOPNOTSUPP */ 258 259 static int 260 rts_send(struct socket *so, int flags, struct mbuf *m, struct sockaddr *nam, 261 struct mbuf *control, struct proc *p) 262 { 263 int s, error; 264 s = splnet(); 265 error = raw_usrreqs.pru_send(so, flags, m, nam, control, p); 266 splx(s); 267 return error; 268 } 269 270 /* pru_sense is null */ 271 272 static int 273 rts_shutdown(struct socket *so) 274 { 275 int s, error; 276 s = splnet(); 277 error = raw_usrreqs.pru_shutdown(so); 278 splx(s); 279 return error; 280 } 281 282 static int 283 rts_sockaddr(struct socket *so, struct sockaddr **nam) 284 { 285 int s, error; 286 s = splnet(); 287 error = raw_usrreqs.pru_sockaddr(so, nam); 288 splx(s); 289 return error; 290 } 291 292 static struct pr_usrreqs route_usrreqs = { 293 rts_abort, pru_accept_notsupp, rts_attach, rts_bind, rts_connect, 294 pru_connect2_notsupp, pru_control_notsupp, rts_detach, rts_disconnect, 295 pru_listen_notsupp, rts_peeraddr, pru_rcvd_notsupp, pru_rcvoob_notsupp, 296 rts_send, pru_sense_null, rts_shutdown, rts_sockaddr, 297 sosend, soreceive, sopoll 298 }; 299 300 /*ARGSUSED*/ 301 static int 302 route_output(m, so) 303 register struct mbuf *m; 304 struct socket *so; 305 { 306 register struct rt_msghdr *rtm = 0; 307 register struct rtentry *rt = 0; 308 struct rtentry *saved_nrt = 0; 309 struct radix_node_head *rnh; 310 struct rt_addrinfo info; 311 int len, error = 0; 312 struct ifnet *ifp = 0; 313 struct ifaddr *ifa = 0; 314 struct proc *curproc = current_proc(); 315 int sendonlytoself = 0; 316 317 #define senderr(e) { error = e; goto flush;} 318 if (m == 0 || ((m->m_len < sizeof(long)) && 319 (m = m_pullup(m, sizeof(long))) == 0)) 320 return (ENOBUFS); 321 if ((m->m_flags & M_PKTHDR) == 0) 322 panic("route_output"); 323 len = m->m_pkthdr.len; 324 if (len < sizeof(*rtm) || 325 len != mtod(m, struct rt_msghdr *)->rtm_msglen) { 326 dst = 0; 327 senderr(EINVAL); 328 } 329 R_Malloc(rtm, struct rt_msghdr *, len); 330 if (rtm == 0) { 331 dst = 0; 332 senderr(ENOBUFS); 333 } 334 m_copydata(m, 0, len, (caddr_t)rtm); 335 if (rtm->rtm_version != RTM_VERSION) { 336 dst = 0; 337 senderr(EPROTONOSUPPORT); 338 } 339 340 /* 341 * Silent version of RTM_GET for Reachabiltiy APIs. We may change 342 * all RTM_GETs to be silent in the future, so this is private for now. 343 */ 344 if (rtm->rtm_type == RTM_GET_SILENT) { 345 if ((so->so_options & SO_USELOOPBACK) == 0) 346 senderr(EINVAL); 347 sendonlytoself = 1; 348 rtm->rtm_type = RTM_GET; 349 } 350 351 /* 352 * Perform permission checking, only privileged sockets 353 * may perform operations other than RTM_GET 354 */ 355 if (rtm->rtm_type != RTM_GET && (so->so_state & SS_PRIV) == 0) { 356 dst = 0; 357 senderr(EPERM); 358 } 359 rtm->rtm_pid = curproc->p_pid; 360 info.rti_addrs = rtm->rtm_addrs; 361 if (rt_xaddrs((caddr_t)(rtm + 1), len + (caddr_t)rtm, &info)) { 362 dst = 0; 363 senderr(EINVAL); 364 } 365 if (dst == 0 || (dst->sa_family >= AF_MAX) 366 || (gate != 0 && (gate->sa_family >= AF_MAX))) 367 senderr(EINVAL); 368 if (genmask) { 369 struct radix_node *t; 370 t = rn_addmask((caddr_t)genmask, 0, 1); 371 if (t && Bcmp(genmask, t->rn_key, *(u_char *)genmask) == 0) 372 genmask = (struct sockaddr *)(t->rn_key); 373 else 374 senderr(ENOBUFS); 375 } 376 switch (rtm->rtm_type) { 377 378 case RTM_ADD: 379 if (gate == 0) 380 senderr(EINVAL); 381 382 #ifdef __APPLE__ 383 /* XXX LD11JUL02 Special case for AOL 5.1.2 connectivity issue to AirPort BS (Radar 2969954) 384 * AOL is adding a circular route ("10.0.1.1/32 10.0.1.1") when establishing its ppp tunnel 385 * to the AP BaseStation by removing the default gateway and replacing it with their tunnel entry point. 386 * There is no apparent reason to add this route as there is a valid 10.0.1.1/24 route to the BS. 387 * That circular route was ignored on previous version of MacOS X because of a routing bug 388 * corrected with the merge to FreeBSD4.4 (a route generated from an RTF_CLONING route had the RTF_WASCLONED 389 * flag set but did not have a reference to the parent route) and that entry was left in the RT. This workaround is 390 * made in order to provide binary compatibility with AOL. 391 * If we catch a process adding a circular route with a /32 from the routing socket, we error it out instead of 392 * confusing the routing table with a wrong route to the previous default gateway 393 */ 394 { 395 extern int check_routeselfref; 396 #define satosinaddr(sa) (((struct sockaddr_in *)sa)->sin_addr.s_addr) 397 398 if (check_routeselfref && (dst && dst->sa_family == AF_INET) && 399 (netmask && satosinaddr(netmask) == INADDR_BROADCAST) && 400 (gate && satosinaddr(dst) == satosinaddr(gate))) { 401 log(LOG_WARNING, "route_output: circular route %ld.%ld.%ld.%ld/32 ignored\n", 402 (ntohl(satosinaddr(gate)>>24))&0xff, 403 (ntohl(satosinaddr(gate)>>16))&0xff, 404 (ntohl(satosinaddr(gate)>>8))&0xff, 405 (ntohl(satosinaddr(gate)))&0xff); 406 407 senderr(EINVAL); 408 } 409 } 410 #endif 411 error = rtrequest(RTM_ADD, dst, gate, netmask, 412 rtm->rtm_flags, &saved_nrt); 413 if (error == 0 && saved_nrt) { 414 #ifdef __APPLE__ 415 /* 416 * If the route request specified an interface with 417 * IFA and/or IFP, we set the requested interface on 418 * the route with rt_setif. It would be much better 419 * to do this inside rtrequest, but that would 420 * require passing the desired interface, in some 421 * form, to rtrequest. Since rtrequest is called in 422 * so many places (roughly 40 in our source), adding 423 * a parameter is to much for us to swallow; this is 424 * something for the FreeBSD developers to tackle. 425 * Instead, we let rtrequest compute whatever 426 * interface it wants, then come in behind it and 427 * stick in the interface that we really want. This 428 * works reasonably well except when rtrequest can't 429 * figure out what interface to use (with 430 * ifa_withroute) and returns ENETUNREACH. Ideally 431 * it shouldn't matter if rtrequest can't figure out 432 * the interface if we're going to explicitly set it 433 * ourselves anyway. But practically we can't 434 * recover here because rtrequest will not do any of 435 * the work necessary to add the route if it can't 436 * find an interface. As long as there is a default 437 * route that leads to some interface, rtrequest will 438 * find an interface, so this problem should be 439 * rarely encountered. 440 * dwiggins@bbn.com 441 */ 442 443 rt_setif(saved_nrt, ifpaddr, ifaaddr, gate); 444 #endif 445 rt_setmetrics(rtm->rtm_inits, 446 &rtm->rtm_rmx, &saved_nrt->rt_rmx); 447 saved_nrt->rt_rmx.rmx_locks &= ~(rtm->rtm_inits); 448 saved_nrt->rt_rmx.rmx_locks |= 449 (rtm->rtm_inits & rtm->rtm_rmx.rmx_locks); 450 rtunref(saved_nrt); 451 saved_nrt->rt_genmask = genmask; 452 } 453 break; 454 455 case RTM_DELETE: 456 error = rtrequest(RTM_DELETE, dst, gate, netmask, 457 rtm->rtm_flags, &saved_nrt); 458 if (error == 0) { 459 if ((rt = saved_nrt)) 460 rtref(rt); 461 goto report; 462 } 463 break; 464 465 case RTM_GET: 466 case RTM_CHANGE: 467 case RTM_LOCK: 468 if ((rnh = rt_tables[dst->sa_family]) == 0) { 469 senderr(EAFNOSUPPORT); 470 } else if ((rt = (struct rtentry *) 471 rnh->rnh_lookup(dst, netmask, rnh)) != NULL) 472 rtref(rt); 473 else 474 senderr(ESRCH); 475 switch(rtm->rtm_type) { 476 477 case RTM_GET: 478 report: 479 dst = rt_key(rt); 480 gate = rt->rt_gateway; 481 netmask = rt_mask(rt); 482 genmask = rt->rt_genmask; 483 if (rtm->rtm_addrs & (RTA_IFP | RTA_IFA)) { 484 ifp = rt->rt_ifp; 485 if (ifp) { 486 ifpaddr = ifp->if_addrhead.tqh_first->ifa_addr; 487 ifaaddr = rt->rt_ifa->ifa_addr; 488 rtm->rtm_index = ifp->if_index; 489 } else { 490 ifpaddr = 0; 491 ifaaddr = 0; 492 } 493 } 494 len = rt_msg2(rtm->rtm_type, &info, (caddr_t)0, 495 (struct walkarg *)0); 496 if (len > rtm->rtm_msglen) { 497 struct rt_msghdr *new_rtm; 498 R_Malloc(new_rtm, struct rt_msghdr *, len); 499 if (new_rtm == 0) 500 senderr(ENOBUFS); 501 Bcopy(rtm, new_rtm, rtm->rtm_msglen); 502 Free(rtm); rtm = new_rtm; 503 } 504 (void)rt_msg2(rtm->rtm_type, &info, (caddr_t)rtm, 505 (struct walkarg *)0); 506 rtm->rtm_flags = rt->rt_flags; 507 rtm->rtm_rmx = rt->rt_rmx; 508 rtm->rtm_addrs = info.rti_addrs; 509 break; 510 511 case RTM_CHANGE: 512 if (gate && (error = rt_setgate(rt, rt_key(rt), gate))) 513 senderr(error); 514 515 /* 516 * If they tried to change things but didn't specify 517 * the required gateway, then just use the old one. 518 * This can happen if the user tries to change the 519 * flags on the default route without changing the 520 * default gateway. Changing flags still doesn't work. 521 */ 522 if ((rt->rt_flags & RTF_GATEWAY) && !gate) 523 gate = rt->rt_gateway; 524 525 #ifdef __APPLE__ 526 /* 527 * On Darwin, we call rt_setif which contains the 528 * equivalent to the code found at this very spot 529 * in BSD. 530 */ 531 rt_setif(rt, ifpaddr, ifaaddr, gate); 532 #endif 533 534 rt_setmetrics(rtm->rtm_inits, &rtm->rtm_rmx, 535 &rt->rt_rmx); 536 #ifndef __APPLE__ 537 /* rt_setif, called above does this for us on darwin */ 538 if (rt->rt_ifa && rt->rt_ifa->ifa_rtrequest) 539 rt->rt_ifa->ifa_rtrequest(RTM_ADD, rt, gate); 540 #endif 541 if (genmask) 542 rt->rt_genmask = genmask; 543 /* 544 * Fall into 545 */ 546 case RTM_LOCK: 547 rt->rt_rmx.rmx_locks &= ~(rtm->rtm_inits); 548 rt->rt_rmx.rmx_locks |= 549 (rtm->rtm_inits & rtm->rtm_rmx.rmx_locks); 550 break; 551 } 552 break; 553 554 default: 555 senderr(EOPNOTSUPP); 556 } 557 558 flush: 559 if (rtm) { 560 if (error) 561 rtm->rtm_errno = error; 562 else 563 rtm->rtm_flags |= RTF_DONE; 564 } 565 if (rt) 566 rtfree(rt); 567 { 568 register struct rawcb *rp = 0; 569 /* 570 * Check to see if we don't want our own messages. 571 */ 572 if ((so->so_options & SO_USELOOPBACK) == 0) { 573 if (route_cb.any_count <= 1) { 574 if (rtm) 575 Free(rtm); 576 m_freem(m); 577 return (error); 578 } 579 /* There is another listener, so construct message */ 580 rp = sotorawcb(so); 581 } 582 if (rtm) { 583 m_copyback(m, 0, rtm->rtm_msglen, (caddr_t)rtm); 584 if (m->m_pkthdr.len < rtm->rtm_msglen) { 585 m_freem(m); 586 m = NULL; 587 } else if (m->m_pkthdr.len > rtm->rtm_msglen) 588 m_adj(m, rtm->rtm_msglen - m->m_pkthdr.len); 589 Free(rtm); 590 } 591 if (sendonlytoself && m) { 592 if (sbappendaddr(&so->so_rcv, &route_src, m, (struct mbuf*)0) == 0) { 593 m_freem(m); 594 error = ENOBUFS; 595 } else { 596 sorwakeup(so); 597 } 598 } else { 599 if (rp) 600 rp->rcb_proto.sp_family = 0; /* Avoid us */ 601 if (dst) 602 route_proto.sp_protocol = dst->sa_family; 603 if (m) 604 raw_input(m, &route_proto, &route_src, &route_dst); 605 if (rp) 606 rp->rcb_proto.sp_family = PF_ROUTE; 607 } 608 } 609 return (error); 610 } 611 612 static void 613 rt_setmetrics(which, in, out) 614 u_long which; 615 register struct rt_metrics *in, *out; 616 { 617 #define metric(f, e) if (which & (f)) out->e = in->e; 618 metric(RTV_RPIPE, rmx_recvpipe); 619 metric(RTV_SPIPE, rmx_sendpipe); 620 metric(RTV_SSTHRESH, rmx_ssthresh); 621 metric(RTV_RTT, rmx_rtt); 622 metric(RTV_RTTVAR, rmx_rttvar); 623 metric(RTV_HOPCOUNT, rmx_hopcount); 624 metric(RTV_MTU, rmx_mtu); 625 metric(RTV_EXPIRE, rmx_expire); 626 #undef metric 627 } 628 629 /* 630 * Set route's interface given ifpaddr, ifaaddr, and gateway. 631 */ 632 static void 633 rt_setif(rt, Ifpaddr, Ifaaddr, Gate) 634 struct rtentry *rt; 635 struct sockaddr *Ifpaddr, *Ifaaddr, *Gate; 636 { 637 struct ifaddr *ifa = 0; 638 struct ifnet *ifp = 0; 639 640 /* new gateway could require new ifaddr, ifp; 641 flags may also be different; ifp may be specified 642 by ll sockaddr when protocol address is ambiguous */ 643 if (Ifpaddr && (ifa = ifa_ifwithnet(Ifpaddr)) && 644 (ifp = ifa->ifa_ifp) && (Ifaaddr || Gate)) 645 ifa = ifaof_ifpforaddr(Ifaaddr ? Ifaaddr : Gate, 646 ifp); 647 else if (Ifpaddr && (ifp = if_withname(Ifpaddr)) ) { 648 ifa = Gate ? ifaof_ifpforaddr(Gate, ifp) : 649 TAILQ_FIRST(&ifp->if_addrhead); 650 } 651 else if ((Ifaaddr && (ifa = ifa_ifwithaddr(Ifaaddr))) || 652 (Gate && (ifa = ifa_ifwithroute(rt->rt_flags, 653 rt_key(rt), Gate)))) 654 ifp = ifa->ifa_ifp; 655 if (ifa) { 656 register struct ifaddr *oifa = rt->rt_ifa; 657 if (oifa != ifa) { 658 if (oifa && oifa->ifa_rtrequest) 659 oifa->ifa_rtrequest(RTM_DELETE, 660 rt, Gate); 661 rtsetifa(rt, ifa); 662 rt->rt_ifp = ifp; 663 rt->rt_rmx.rmx_mtu = ifp->if_mtu; 664 if (rt->rt_ifa && rt->rt_ifa->ifa_rtrequest) 665 rt->rt_ifa->ifa_rtrequest(RTM_ADD, rt, Gate); 666 } else 667 goto call_ifareq; 668 return; 669 } 670 call_ifareq: 671 /* XXX: to reset gateway to correct value, at RTM_CHANGE */ 672 if (rt->rt_ifa && rt->rt_ifa->ifa_rtrequest) 673 rt->rt_ifa->ifa_rtrequest(RTM_ADD, rt, Gate); 674 } 675 676 677 #define ROUNDUP(a) \ 678 ((a) > 0 ? (1 + (((a) - 1) | (sizeof(long) - 1))) : sizeof(long)) 679 #define ADVANCE(x, n) (x += ROUNDUP((n)->sa_len)) 680 681 682 /* 683 * Extract the addresses of the passed sockaddrs. 684 * Do a little sanity checking so as to avoid bad memory references. 685 * This data is derived straight from userland. 686 */ 687 static int 688 rt_xaddrs(cp, cplim, rtinfo) 689 register caddr_t cp, cplim; 690 register struct rt_addrinfo *rtinfo; 691 { 692 register struct sockaddr *sa; 693 register int i; 694 695 bzero(rtinfo->rti_info, sizeof(rtinfo->rti_info)); 696 for (i = 0; (i < RTAX_MAX) && (cp < cplim); i++) { 697 if ((rtinfo->rti_addrs & (1 << i)) == 0) 698 continue; 699 sa = (struct sockaddr *)cp; 700 /* 701 * It won't fit. 702 */ 703 if ( (cp + sa->sa_len) > cplim ) { 704 return (EINVAL); 705 } 706 707 /* 708 * there are no more.. quit now 709 * If there are more bits, they are in error. 710 * I've seen this. route(1) can evidently generate these. 711 * This causes kernel to core dump. 712 * for compatibility, If we see this, point to a safe address. 713 */ 714 if (sa->sa_len == 0) { 715 rtinfo->rti_info[i] = &sa_zero; 716 return (0); /* should be EINVAL but for compat */ 717 } 718 719 /* accept it */ 720 rtinfo->rti_info[i] = sa; 721 ADVANCE(cp, sa); 722 } 723 return (0); 724 } 725 726 static struct mbuf * 727 rt_msg1(type, rtinfo) 728 int type; 729 register struct rt_addrinfo *rtinfo; 730 { 731 register struct rt_msghdr *rtm; 732 register struct mbuf *m; 733 register int i; 734 register struct sockaddr *sa; 735 int len, dlen; 736 737 switch (type) { 738 739 case RTM_DELADDR: 740 case RTM_NEWADDR: 741 len = sizeof(struct ifa_msghdr); 742 break; 743 744 case RTM_DELMADDR: 745 case RTM_NEWMADDR: 746 len = sizeof(struct ifma_msghdr); 747 break; 748 749 case RTM_IFINFO: 750 len = sizeof(struct if_msghdr); 751 break; 752 753 default: 754 len = sizeof(struct rt_msghdr); 755 } 756 if (len > MCLBYTES) 757 panic("rt_msg1"); 758 m = m_gethdr(M_DONTWAIT, MT_DATA); 759 if (m && len > MHLEN) { 760 MCLGET(m, M_DONTWAIT); 761 if ((m->m_flags & M_EXT) == 0) { 762 m_free(m); 763 m = NULL; 764 } 765 } 766 if (m == 0) 767 return (m); 768 m->m_pkthdr.len = m->m_len = len; 769 m->m_pkthdr.rcvif = 0; 770 rtm = mtod(m, struct rt_msghdr *); 771 bzero((caddr_t)rtm, len); 772 for (i = 0; i < RTAX_MAX; i++) { 773 if ((sa = rtinfo->rti_info[i]) == NULL) 774 continue; 775 rtinfo->rti_addrs |= (1 << i); 776 dlen = ROUNDUP(sa->sa_len); 777 m_copyback(m, len, dlen, (caddr_t)sa); 778 len += dlen; 779 } 780 if (m->m_pkthdr.len != len) { 781 m_freem(m); 782 return (NULL); 783 } 784 rtm->rtm_msglen = len; 785 rtm->rtm_version = RTM_VERSION; 786 rtm->rtm_type = type; 787 return (m); 788 } 789 790 static int 791 rt_msg2(type, rtinfo, cp, w) 792 int type; 793 register struct rt_addrinfo *rtinfo; 794 caddr_t cp; 795 struct walkarg *w; 796 { 797 register int i; 798 int len, dlen, second_time = 0; 799 caddr_t cp0; 800 801 rtinfo->rti_addrs = 0; 802 again: 803 switch (type) { 804 805 case RTM_DELADDR: 806 case RTM_NEWADDR: 807 len = sizeof(struct ifa_msghdr); 808 break; 809 810 case RTM_IFINFO: 811 len = sizeof(struct if_msghdr); 812 break; 813 814 default: 815 len = sizeof(struct rt_msghdr); 816 } 817 cp0 = cp; 818 if (cp0) 819 cp += len; 820 for (i = 0; i < RTAX_MAX; i++) { 821 register struct sockaddr *sa; 822 823 if ((sa = rtinfo->rti_info[i]) == 0) 824 continue; 825 rtinfo->rti_addrs |= (1 << i); 826 dlen = ROUNDUP(sa->sa_len); 827 if (cp) { 828 bcopy((caddr_t)sa, cp, (unsigned)dlen); 829 cp += dlen; 830 } 831 len += dlen; 832 } 833 if (cp == 0 && w != NULL && !second_time) { 834 register struct walkarg *rw = w; 835 836 if (rw->w_req) { 837 if (rw->w_tmemsize < len) { 838 if (rw->w_tmem) 839 FREE(rw->w_tmem, M_RTABLE); 840 rw->w_tmem = (caddr_t) 841 _MALLOC(len, M_RTABLE, M_WAITOK); /*###LD0412 was NOWAIT */ 842 if (rw->w_tmem) 843 rw->w_tmemsize = len; 844 } 845 if (rw->w_tmem) { 846 cp = rw->w_tmem; 847 second_time = 1; 848 goto again; 849 } 850 } 851 } 852 if (cp) { 853 register struct rt_msghdr *rtm = (struct rt_msghdr *)cp0; 854 855 rtm->rtm_version = RTM_VERSION; 856 rtm->rtm_type = type; 857 rtm->rtm_msglen = len; 858 } 859 return (len); 860 } 861 862 /* 863 * This routine is called to generate a message from the routing 864 * socket indicating that a redirect has occured, a routing lookup 865 * has failed, or that a protocol has detected timeouts to a particular 866 * destination. 867 */ 868 void 869 rt_missmsg(type, rtinfo, flags, error) 870 int type, flags, error; 871 register struct rt_addrinfo *rtinfo; 872 { 873 register struct rt_msghdr *rtm; 874 register struct mbuf *m; 875 struct sockaddr *sa = rtinfo->rti_info[RTAX_DST]; 876 877 if (route_cb.any_count == 0) 878 return; 879 m = rt_msg1(type, rtinfo); 880 if (m == 0) 881 return; 882 rtm = mtod(m, struct rt_msghdr *); 883 rtm->rtm_flags = RTF_DONE | flags; 884 rtm->rtm_errno = error; 885 rtm->rtm_addrs = rtinfo->rti_addrs; 886 route_proto.sp_protocol = sa ? sa->sa_family : 0; 887 raw_input(m, &route_proto, &route_src, &route_dst); 888 } 889 890 /* 891 * This routine is called to generate a message from the routing 892 * socket indicating that the status of a network interface has changed. 893 */ 894 void 895 rt_ifmsg(ifp) 896 register struct ifnet *ifp; 897 { 898 register struct if_msghdr *ifm; 899 struct mbuf *m; 900 struct rt_addrinfo info; 901 902 if (route_cb.any_count == 0) 903 return; 904 bzero((caddr_t)&info, sizeof(info)); 905 m = rt_msg1(RTM_IFINFO, &info); 906 if (m == 0) 907 return; 908 ifm = mtod(m, struct if_msghdr *); 909 ifm->ifm_index = ifp->if_index; 910 ifm->ifm_flags = (u_short)ifp->if_flags; 911 ifm->ifm_data = ifp->if_data; 912 ifm->ifm_addrs = 0; 913 route_proto.sp_protocol = 0; 914 raw_input(m, &route_proto, &route_src, &route_dst); 915 } 916 917 /* 918 * This is called to generate messages from the routing socket 919 * indicating a network interface has had addresses associated with it. 920 * if we ever reverse the logic and replace messages TO the routing 921 * socket indicate a request to configure interfaces, then it will 922 * be unnecessary as the routing socket will automatically generate 923 * copies of it. 924 */ 925 void 926 rt_newaddrmsg(cmd, ifa, error, rt) 927 int cmd, error; 928 register struct ifaddr *ifa; 929 register struct rtentry *rt; 930 { 931 struct rt_addrinfo info; 932 struct sockaddr *sa = 0; 933 int pass; 934 struct mbuf *m = 0; 935 struct ifnet *ifp = ifa->ifa_ifp; 936 937 if (route_cb.any_count == 0) 938 return; 939 for (pass = 1; pass < 3; pass++) { 940 bzero((caddr_t)&info, sizeof(info)); 941 if ((cmd == RTM_ADD && pass == 1) || 942 (cmd == RTM_DELETE && pass == 2)) { 943 register struct ifa_msghdr *ifam; 944 int ncmd = cmd == RTM_ADD ? RTM_NEWADDR : RTM_DELADDR; 945 946 ifaaddr = sa = ifa->ifa_addr; 947 ifpaddr = ifp->if_addrhead.tqh_first->ifa_addr; 948 netmask = ifa->ifa_netmask; 949 brdaddr = ifa->ifa_dstaddr; 950 if ((m = rt_msg1(ncmd, &info)) == NULL) 951 continue; 952 ifam = mtod(m, struct ifa_msghdr *); 953 ifam->ifam_index = ifp->if_index; 954 ifam->ifam_metric = ifa->ifa_metric; 955 ifam->ifam_flags = ifa->ifa_flags; 956 ifam->ifam_addrs = info.rti_addrs; 957 } 958 if ((cmd == RTM_ADD && pass == 2) || 959 (cmd == RTM_DELETE && pass == 1)) { 960 register struct rt_msghdr *rtm; 961 962 if (rt == 0) 963 continue; 964 netmask = rt_mask(rt); 965 dst = sa = rt_key(rt); 966 gate = rt->rt_gateway; 967 if ((m = rt_msg1(cmd, &info)) == NULL) 968 continue; 969 rtm = mtod(m, struct rt_msghdr *); 970 rtm->rtm_index = ifp->if_index; 971 rtm->rtm_flags |= rt->rt_flags; 972 rtm->rtm_errno = error; 973 rtm->rtm_addrs = info.rti_addrs; 974 } 975 route_proto.sp_protocol = sa ? sa->sa_family : 0; 976 raw_input(m, &route_proto, &route_src, &route_dst); 977 } 978 } 979 980 /* 981 * This is the analogue to the rt_newaddrmsg which performs the same 982 * function but for multicast group memberhips. This is easier since 983 * there is no route state to worry about. 984 */ 985 void 986 rt_newmaddrmsg(cmd, ifma) 987 int cmd; 988 struct ifmultiaddr *ifma; 989 { 990 struct rt_addrinfo info; 991 struct mbuf *m = 0; 992 struct ifnet *ifp = ifma->ifma_ifp; 993 struct ifma_msghdr *ifmam; 994 995 if (route_cb.any_count == 0) 996 return; 997 998 bzero((caddr_t)&info, sizeof(info)); 999 ifaaddr = ifma->ifma_addr; 1000 if (ifp && ifp->if_addrhead.tqh_first) 1001 ifpaddr = ifp->if_addrhead.tqh_first->ifa_addr; 1002 else 1003 ifpaddr = NULL; 1004 /* 1005 * If a link-layer address is present, present it as a ``gateway'' 1006 * (similarly to how ARP entries, e.g., are presented). 1007 */ 1008 gate = ifma->ifma_lladdr; 1009 if ((m = rt_msg1(cmd, &info)) == NULL) 1010 return; 1011 ifmam = mtod(m, struct ifma_msghdr *); 1012 ifmam->ifmam_index = ifp->if_index; 1013 ifmam->ifmam_addrs = info.rti_addrs; 1014 route_proto.sp_protocol = ifma->ifma_addr->sa_family; 1015 raw_input(m, &route_proto, &route_src, &route_dst); 1016 } 1017 1018 /* 1019 * This is used in dumping the kernel table via sysctl(). 1020 */ 1021 int 1022 sysctl_dumpentry(rn, vw) 1023 struct radix_node *rn; 1024 void *vw; 1025 { 1026 register struct walkarg *w = vw; 1027 register struct rtentry *rt = (struct rtentry *)rn; 1028 int error = 0, size; 1029 struct rt_addrinfo info; 1030 1031 if (w->w_op == NET_RT_FLAGS && !(rt->rt_flags & w->w_arg)) 1032 return 0; 1033 bzero((caddr_t)&info, sizeof(info)); 1034 dst = rt_key(rt); 1035 gate = rt->rt_gateway; 1036 netmask = rt_mask(rt); 1037 genmask = rt->rt_genmask; 1038 size = rt_msg2(RTM_GET, &info, 0, w); 1039 if (w->w_req && w->w_tmem) { 1040 register struct rt_msghdr *rtm = (struct rt_msghdr *)w->w_tmem; 1041 1042 rtm->rtm_flags = rt->rt_flags; 1043 rtm->rtm_use = rt->rt_use; 1044 rtm->rtm_rmx = rt->rt_rmx; 1045 rtm->rtm_index = rt->rt_ifp->if_index; 1046 rtm->rtm_errno = rtm->rtm_pid = rtm->rtm_seq = 0; 1047 rtm->rtm_addrs = info.rti_addrs; 1048 error = SYSCTL_OUT(w->w_req, (caddr_t)rtm, size); 1049 return (error); 1050 } 1051 return (error); 1052 } 1053 1054 int 1055 sysctl_iflist(af, w) 1056 int af; 1057 register struct walkarg *w; 1058 { 1059 register struct ifnet *ifp; 1060 register struct ifaddr *ifa; 1061 struct rt_addrinfo info; 1062 int len, error = 0; 1063 1064 bzero((caddr_t)&info, sizeof(info)); 1065 for (ifp = ifnet.tqh_first; ifp; ifp = ifp->if_link.tqe_next) { 1066 if (w->w_arg && w->w_arg != ifp->if_index) 1067 continue; 1068 ifa = ifp->if_addrhead.tqh_first; 1069 ifpaddr = ifa->ifa_addr; 1070 len = rt_msg2(RTM_IFINFO, &info, (caddr_t)0, w); 1071 ifpaddr = 0; 1072 if (w->w_req && w->w_tmem) { 1073 register struct if_msghdr *ifm; 1074 1075 ifm = (struct if_msghdr *)w->w_tmem; 1076 ifm->ifm_index = ifp->if_index; 1077 ifm->ifm_flags = (u_short)ifp->if_flags; 1078 ifm->ifm_data = ifp->if_data; 1079 ifm->ifm_addrs = info.rti_addrs; 1080 error = SYSCTL_OUT(w->w_req,(caddr_t)ifm, len); 1081 if (error) 1082 return (error); 1083 } 1084 while ((ifa = ifa->ifa_link.tqe_next) != 0) { 1085 if (af && af != ifa->ifa_addr->sa_family) 1086 continue; 1087 #ifndef __APPLE__ 1088 if (curproc->p_prison && prison_if(curproc, ifa->ifa_addr)) 1089 continue; 1090 #endif 1091 ifaaddr = ifa->ifa_addr; 1092 netmask = ifa->ifa_netmask; 1093 brdaddr = ifa->ifa_dstaddr; 1094 len = rt_msg2(RTM_NEWADDR, &info, 0, w); 1095 if (w->w_req && w->w_tmem) { 1096 register struct ifa_msghdr *ifam; 1097 1098 ifam = (struct ifa_msghdr *)w->w_tmem; 1099 ifam->ifam_index = ifa->ifa_ifp->if_index; 1100 ifam->ifam_flags = ifa->ifa_flags; 1101 ifam->ifam_metric = ifa->ifa_metric; 1102 ifam->ifam_addrs = info.rti_addrs; 1103 error = SYSCTL_OUT(w->w_req, w->w_tmem, len); 1104 if (error) 1105 return (error); 1106 } 1107 } 1108 ifaaddr = netmask = brdaddr = 0; 1109 } 1110 return (0); 1111 } 1112 1113 static int 1114 sysctl_rtsock SYSCTL_HANDLER_ARGS 1115 { 1116 int *name = (int *)arg1; 1117 u_int namelen = arg2; 1118 register struct radix_node_head *rnh; 1119 int i, s, error = EINVAL; 1120 u_char af; 1121 struct walkarg w; 1122 1123 name ++; 1124 namelen--; 1125 if (req->newptr) 1126 return (EPERM); 1127 if (namelen != 3) 1128 return (EINVAL); 1129 af = name[0]; 1130 Bzero(&w, sizeof(w)); 1131 w.w_op = name[1]; 1132 w.w_arg = name[2]; 1133 w.w_req = req; 1134 1135 s = splnet(); 1136 switch (w.w_op) { 1137 1138 case NET_RT_DUMP: 1139 case NET_RT_FLAGS: 1140 for (i = 1; i <= AF_MAX; i++) 1141 if ((rnh = rt_tables[i]) && (af == 0 || af == i) && 1142 (error = rnh->rnh_walktree(rnh, 1143 sysctl_dumpentry, &w))) 1144 break; 1145 break; 1146 1147 case NET_RT_IFLIST: 1148 error = sysctl_iflist(af, &w); 1149 } 1150 splx(s); 1151 if (w.w_tmem) 1152 FREE(w.w_tmem, M_RTABLE); 1153 return (error); 1154 } 1155 1156 SYSCTL_NODE(_net, PF_ROUTE, routetable, CTLFLAG_RD, sysctl_rtsock, ""); 1157 1158 /* 1159 * Definitions of protocols supported in the ROUTE domain. 1160 */ 1161 1162 struct domain routedomain; /* or at least forward */ 1163 1164 static struct protosw routesw[] = { 1165 { SOCK_RAW, &routedomain, 0, PR_ATOMIC|PR_ADDR, 1166 0, route_output, raw_ctlinput, 0, 1167 0, 1168 raw_init, 0, 0, 0, 1169 0, &route_usrreqs, 0, 0 1170 } 1171 }; 1172 1173 struct domain routedomain = 1174 { PF_ROUTE, "route", route_init, 0, 0, 1175 routesw}; 1176 1177 DOMAIN_SET(route); 1178
Cache object: db8289701eb48c0555ace2e1ce7c000c
[ source navigation ] [ diff markup ] [ identifier search ] [ freetext search ] [ file search ] [ list types ] [ track identifier ]