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 1.29 2005/07/15 17:54:47 eirikn Exp $ 86 */ 87 88 #include "opt_sctp.h" 89 90 #include <sys/param.h> 91 #include <sys/systm.h> 92 #include <sys/kernel.h> 93 #include <sys/sysctl.h> 94 #include <sys/proc.h> 95 #include <sys/malloc.h> 96 #include <sys/mbuf.h> 97 #include <sys/protosw.h> 98 #include <sys/socket.h> 99 #include <sys/socketvar.h> 100 #include <sys/domain.h> 101 #include <sys/thread2.h> 102 103 #include <net/if.h> 104 #include <net/route.h> 105 #include <net/raw_cb.h> 106 107 #ifdef SCTP 108 extern void sctp_add_ip_address(struct ifaddr *ifa); 109 extern void sctp_delete_ip_address(struct ifaddr *ifa); 110 #endif /* SCTP */ 111 112 MALLOC_DEFINE(M_RTABLE, "routetbl", "routing tables"); 113 114 static struct route_cb { 115 int ip_count; 116 int ip6_count; 117 int ipx_count; 118 int ns_count; 119 int any_count; 120 } route_cb; 121 122 static const struct sockaddr route_src = { 2, PF_ROUTE, }; 123 124 struct walkarg { 125 int w_tmemsize; 126 int w_op, w_arg; 127 void *w_tmem; 128 struct sysctl_req *w_req; 129 }; 130 131 static struct mbuf * 132 rt_msg_mbuf (int, struct rt_addrinfo *); 133 static void rt_msg_buffer (int, struct rt_addrinfo *, void *buf, int len); 134 static int rt_msgsize (int type, struct rt_addrinfo *rtinfo); 135 static int rt_xaddrs (char *, char *, struct rt_addrinfo *); 136 static int sysctl_dumpentry (struct radix_node *rn, void *vw); 137 static int sysctl_iflist (int af, struct walkarg *w); 138 static int route_output(struct mbuf *, struct socket *, ...); 139 static void rt_setmetrics (u_long, struct rt_metrics *, 140 struct rt_metrics *); 141 142 /* 143 * It really doesn't make any sense at all for this code to share much 144 * with raw_usrreq.c, since its functionality is so restricted. XXX 145 */ 146 static int 147 rts_abort(struct socket *so) 148 { 149 int error; 150 151 crit_enter(); 152 error = raw_usrreqs.pru_abort(so); 153 crit_exit(); 154 return error; 155 } 156 157 /* pru_accept is EOPNOTSUPP */ 158 159 static int 160 rts_attach(struct socket *so, int proto, struct pru_attach_info *ai) 161 { 162 struct rawcb *rp; 163 int error; 164 165 if (sotorawcb(so) != NULL) 166 return EISCONN; /* XXX panic? */ 167 168 rp = malloc(sizeof *rp, M_PCB, M_WAITOK | M_ZERO); 169 if (rp == NULL) 170 return ENOBUFS; 171 172 /* 173 * The critical section is necessary to block protocols from sending 174 * error notifications (like RTM_REDIRECT or RTM_LOSING) while 175 * this PCB is extant but incompletely initialized. 176 * Probably we should try to do more of this work beforehand and 177 * eliminate the critical section. 178 */ 179 crit_enter(); 180 so->so_pcb = rp; 181 error = raw_attach(so, proto, ai->sb_rlimit); 182 rp = sotorawcb(so); 183 if (error) { 184 crit_exit(); 185 free(rp, M_PCB); 186 return error; 187 } 188 switch(rp->rcb_proto.sp_protocol) { 189 case AF_INET: 190 route_cb.ip_count++; 191 break; 192 case AF_INET6: 193 route_cb.ip6_count++; 194 break; 195 case AF_IPX: 196 route_cb.ipx_count++; 197 break; 198 case AF_NS: 199 route_cb.ns_count++; 200 break; 201 } 202 rp->rcb_faddr = &route_src; 203 route_cb.any_count++; 204 soisconnected(so); 205 so->so_options |= SO_USELOOPBACK; 206 crit_exit(); 207 return 0; 208 } 209 210 static int 211 rts_bind(struct socket *so, struct sockaddr *nam, struct thread *td) 212 { 213 int error; 214 215 crit_enter(); 216 error = raw_usrreqs.pru_bind(so, nam, td); /* xxx just EINVAL */ 217 crit_exit(); 218 return error; 219 } 220 221 static int 222 rts_connect(struct socket *so, struct sockaddr *nam, struct thread *td) 223 { 224 int error; 225 226 crit_enter(); 227 error = raw_usrreqs.pru_connect(so, nam, td); /* XXX just EINVAL */ 228 crit_exit(); 229 return error; 230 } 231 232 /* pru_connect2 is EOPNOTSUPP */ 233 /* pru_control is EOPNOTSUPP */ 234 235 static int 236 rts_detach(struct socket *so) 237 { 238 struct rawcb *rp = sotorawcb(so); 239 int error; 240 241 crit_enter(); 242 if (rp != NULL) { 243 switch(rp->rcb_proto.sp_protocol) { 244 case AF_INET: 245 route_cb.ip_count--; 246 break; 247 case AF_INET6: 248 route_cb.ip6_count--; 249 break; 250 case AF_IPX: 251 route_cb.ipx_count--; 252 break; 253 case AF_NS: 254 route_cb.ns_count--; 255 break; 256 } 257 route_cb.any_count--; 258 } 259 error = raw_usrreqs.pru_detach(so); 260 crit_exit(); 261 return error; 262 } 263 264 static int 265 rts_disconnect(struct socket *so) 266 { 267 int error; 268 269 crit_enter(); 270 error = raw_usrreqs.pru_disconnect(so); 271 crit_exit(); 272 return error; 273 } 274 275 /* pru_listen is EOPNOTSUPP */ 276 277 static int 278 rts_peeraddr(struct socket *so, struct sockaddr **nam) 279 { 280 int error; 281 282 crit_enter(); 283 error = raw_usrreqs.pru_peeraddr(so, nam); 284 crit_exit(); 285 return error; 286 } 287 288 /* pru_rcvd is EOPNOTSUPP */ 289 /* pru_rcvoob is EOPNOTSUPP */ 290 291 static int 292 rts_send(struct socket *so, int flags, struct mbuf *m, struct sockaddr *nam, 293 struct mbuf *control, struct thread *td) 294 { 295 int error; 296 297 crit_enter(); 298 error = raw_usrreqs.pru_send(so, flags, m, nam, control, td); 299 crit_exit(); 300 return error; 301 } 302 303 /* pru_sense is null */ 304 305 static int 306 rts_shutdown(struct socket *so) 307 { 308 int error; 309 310 crit_enter(); 311 error = raw_usrreqs.pru_shutdown(so); 312 crit_exit(); 313 return error; 314 } 315 316 static int 317 rts_sockaddr(struct socket *so, struct sockaddr **nam) 318 { 319 int error; 320 321 crit_enter(); 322 error = raw_usrreqs.pru_sockaddr(so, nam); 323 crit_exit(); 324 return error; 325 } 326 327 static struct pr_usrreqs route_usrreqs = { 328 rts_abort, pru_accept_notsupp, rts_attach, rts_bind, rts_connect, 329 pru_connect2_notsupp, pru_control_notsupp, rts_detach, rts_disconnect, 330 pru_listen_notsupp, rts_peeraddr, pru_rcvd_notsupp, pru_rcvoob_notsupp, 331 rts_send, pru_sense_null, rts_shutdown, rts_sockaddr, 332 sosend, soreceive, sopoll 333 }; 334 335 static __inline sa_family_t 336 familyof(struct sockaddr *sa) 337 { 338 return (sa != NULL ? sa->sa_family : 0); 339 } 340 341 static void 342 rts_input(struct mbuf *m, sa_family_t family) 343 { 344 static const struct sockaddr route_dst = { 2, PF_ROUTE, }; 345 struct sockproto route_proto = { PF_ROUTE, family }; 346 347 raw_input(m, &route_proto, &route_src, &route_dst); 348 } 349 350 static void * 351 reallocbuf(void *ptr, size_t len, size_t olen) 352 { 353 void *newptr; 354 355 newptr = malloc(len, M_RTABLE, M_INTWAIT | M_NULLOK); 356 if (newptr == NULL) 357 return NULL; 358 bcopy(ptr, newptr, olen); 359 free(ptr, M_RTABLE); 360 return (newptr); 361 } 362 363 /* 364 * Internal helper routine for route_output(). 365 */ 366 static int 367 fillrtmsg(struct rt_msghdr **prtm, struct rtentry *rt, 368 struct rt_addrinfo *rtinfo) 369 { 370 int msglen; 371 struct rt_msghdr *rtm = *prtm; 372 373 /* Fill in rt_addrinfo for call to rt_msg_buffer(). */ 374 rtinfo->rti_dst = rt_key(rt); 375 rtinfo->rti_gateway = rt->rt_gateway; 376 rtinfo->rti_netmask = rt_mask(rt); /* might be NULL */ 377 rtinfo->rti_genmask = rt->rt_genmask; /* might be NULL */ 378 if (rtm->rtm_addrs & (RTA_IFP | RTA_IFA)) { 379 if (rt->rt_ifp != NULL) { 380 rtinfo->rti_ifpaddr = 381 TAILQ_FIRST(&rt->rt_ifp->if_addrhead)->ifa_addr; 382 rtinfo->rti_ifaaddr = rt->rt_ifa->ifa_addr; 383 if (rt->rt_ifp->if_flags & IFF_POINTOPOINT) 384 rtinfo->rti_bcastaddr = rt->rt_ifa->ifa_dstaddr; 385 rtm->rtm_index = rt->rt_ifp->if_index; 386 } else { 387 rtinfo->rti_ifpaddr = NULL; 388 rtinfo->rti_ifaaddr = NULL; 389 } 390 } 391 392 msglen = rt_msgsize(rtm->rtm_type, rtinfo); 393 if (rtm->rtm_msglen < msglen) { 394 rtm = reallocbuf(rtm, msglen, rtm->rtm_msglen); 395 if (rtm == NULL) 396 return (ENOBUFS); 397 *prtm = rtm; 398 } 399 rt_msg_buffer(rtm->rtm_type, rtinfo, rtm, msglen); 400 401 rtm->rtm_flags = rt->rt_flags; 402 rtm->rtm_rmx = rt->rt_rmx; 403 rtm->rtm_addrs = rtinfo->rti_addrs; 404 405 return (0); 406 } 407 408 /*ARGSUSED*/ 409 static int 410 route_output(struct mbuf *m, struct socket *so, ...) 411 { 412 struct rt_msghdr *rtm = NULL; 413 struct rtentry *rt = NULL; 414 struct rtentry *saved_nrt = NULL; 415 struct radix_node_head *rnh; 416 struct ifaddr *ifa = NULL; 417 struct rawcb *rp = NULL; 418 struct pr_output_info *oi; 419 struct rt_addrinfo rtinfo; 420 int len, error = 0; 421 __va_list ap; 422 423 __va_start(ap, so); 424 oi = __va_arg(ap, struct pr_output_info *); 425 __va_end(ap); 426 427 #define gotoerr(e) { error = e; goto flush;} 428 429 if (m == NULL || 430 (m->m_len < sizeof(long) && 431 (m = m_pullup(m, sizeof(long))) == NULL)) 432 return (ENOBUFS); 433 if (!(m->m_flags & M_PKTHDR)) 434 panic("route_output"); 435 len = m->m_pkthdr.len; 436 if (len < sizeof(struct rt_msghdr) || 437 len != mtod(m, struct rt_msghdr *)->rtm_msglen) { 438 rtinfo.rti_dst = NULL; 439 gotoerr(EINVAL); 440 } 441 rtm = malloc(len, M_RTABLE, M_INTWAIT | M_NULLOK); 442 if (rtm == NULL) { 443 rtinfo.rti_dst = NULL; 444 gotoerr(ENOBUFS); 445 } 446 m_copydata(m, 0, len, (caddr_t)rtm); 447 if (rtm->rtm_version != RTM_VERSION) { 448 rtinfo.rti_dst = NULL; 449 gotoerr(EPROTONOSUPPORT); 450 } 451 rtm->rtm_pid = oi->p_pid; 452 bzero(&rtinfo, sizeof(struct rt_addrinfo)); 453 rtinfo.rti_addrs = rtm->rtm_addrs; 454 if (rt_xaddrs((char *)(rtm + 1), (char *)rtm + len, &rtinfo) != 0) { 455 rtinfo.rti_dst = NULL; 456 gotoerr(EINVAL); 457 } 458 rtinfo.rti_flags = rtm->rtm_flags; 459 if (rtinfo.rti_dst == NULL || rtinfo.rti_dst->sa_family >= AF_MAX || 460 (rtinfo.rti_gateway && rtinfo.rti_gateway->sa_family >= AF_MAX)) 461 gotoerr(EINVAL); 462 463 if (rtinfo.rti_genmask != NULL) { 464 struct radix_node *n; 465 466 #define clen(s) (*(u_char *)(s)) 467 n = rn_addmask((char *)rtinfo.rti_genmask, TRUE, 1); 468 if (n != NULL && 469 rtinfo.rti_genmask->sa_len >= clen(n->rn_key) && 470 bcmp((char *)rtinfo.rti_genmask + 1, 471 (char *)n->rn_key + 1, clen(n->rn_key) - 1) == 0) 472 rtinfo.rti_genmask = (struct sockaddr *)n->rn_key; 473 else 474 gotoerr(ENOBUFS); 475 } 476 477 /* 478 * Verify that the caller has the appropriate privilege; RTM_GET 479 * is the only operation the non-superuser is allowed. 480 */ 481 if (rtm->rtm_type != RTM_GET && suser_cred(so->so_cred, 0) != 0) 482 gotoerr(EPERM); 483 484 switch (rtm->rtm_type) { 485 case RTM_ADD: 486 if (rtinfo.rti_gateway == NULL) 487 gotoerr(EINVAL); 488 error = rtrequest1(RTM_ADD, &rtinfo, &saved_nrt); 489 if (error == 0 && saved_nrt != NULL) { 490 rt_setmetrics(rtm->rtm_inits, &rtm->rtm_rmx, 491 &saved_nrt->rt_rmx); 492 saved_nrt->rt_rmx.rmx_locks &= ~(rtm->rtm_inits); 493 saved_nrt->rt_rmx.rmx_locks |= 494 (rtm->rtm_inits & rtm->rtm_rmx.rmx_locks); 495 --saved_nrt->rt_refcnt; 496 saved_nrt->rt_genmask = rtinfo.rti_genmask; 497 } 498 break; 499 case RTM_DELETE: 500 error = rtrequest1(RTM_DELETE, &rtinfo, &saved_nrt); 501 if (error == 0) { 502 if ((rt = saved_nrt)) 503 rt->rt_refcnt++; 504 if (fillrtmsg(&rtm, rt, &rtinfo) != 0) 505 gotoerr(ENOBUFS); 506 } 507 break; 508 case RTM_GET: 509 case RTM_CHANGE: 510 case RTM_LOCK: 511 if ((rnh = rt_tables[rtinfo.rti_dst->sa_family]) == NULL) 512 gotoerr(EAFNOSUPPORT); 513 rt = (struct rtentry *) 514 rnh->rnh_lookup((char *)rtinfo.rti_dst, 515 (char *)rtinfo.rti_netmask, rnh); 516 if (rt == NULL) 517 gotoerr(ESRCH); 518 rt->rt_refcnt++; 519 520 switch(rtm->rtm_type) { 521 case RTM_GET: 522 if (fillrtmsg(&rtm, rt, &rtinfo) != 0) 523 gotoerr(ENOBUFS); 524 break; 525 case RTM_CHANGE: 526 /* 527 * new gateway could require new ifaddr, ifp; 528 * flags may also be different; ifp may be specified 529 * by ll sockaddr when protocol address is ambiguous 530 */ 531 if (((rt->rt_flags & RTF_GATEWAY) && 532 rtinfo.rti_gateway != NULL) || 533 rtinfo.rti_ifpaddr != NULL || 534 (rtinfo.rti_ifaaddr != NULL && 535 sa_equal(rtinfo.rti_ifaaddr, 536 rt->rt_ifa->ifa_addr))) { 537 error = rt_getifa(&rtinfo); 538 if (error != 0) 539 gotoerr(error); 540 } 541 if (rtinfo.rti_gateway != NULL) { 542 error = rt_setgate(rt, rt_key(rt), 543 rtinfo.rti_gateway); 544 if (error != 0) 545 gotoerr(error); 546 } 547 if ((ifa = rtinfo.rti_ifa) != NULL) { 548 struct ifaddr *oifa = rt->rt_ifa; 549 550 if (oifa != ifa) { 551 if (oifa && oifa->ifa_rtrequest) 552 oifa->ifa_rtrequest(RTM_DELETE, 553 rt, &rtinfo); 554 IFAFREE(rt->rt_ifa); 555 IFAREF(ifa); 556 rt->rt_ifa = ifa; 557 rt->rt_ifp = rtinfo.rti_ifp; 558 } 559 } 560 rt_setmetrics(rtm->rtm_inits, &rtm->rtm_rmx, 561 &rt->rt_rmx); 562 if (rt->rt_ifa && rt->rt_ifa->ifa_rtrequest) 563 rt->rt_ifa->ifa_rtrequest(RTM_ADD, rt, &rtinfo); 564 if (rtinfo.rti_genmask != NULL) 565 rt->rt_genmask = rtinfo.rti_genmask; 566 /* 567 * Fall into 568 */ 569 case RTM_LOCK: 570 rt->rt_rmx.rmx_locks &= ~(rtm->rtm_inits); 571 rt->rt_rmx.rmx_locks |= 572 (rtm->rtm_inits & rtm->rtm_rmx.rmx_locks); 573 break; 574 } 575 576 break; 577 default: 578 gotoerr(EOPNOTSUPP); 579 } 580 581 flush: 582 if (rtm != NULL) { 583 if (error != 0) 584 rtm->rtm_errno = error; 585 else 586 rtm->rtm_flags |= RTF_DONE; 587 } 588 if (rt != NULL) 589 rtfree(rt); 590 /* 591 * Check to see if we don't want our own messages. 592 */ 593 if (!(so->so_options & SO_USELOOPBACK)) { 594 if (route_cb.any_count <= 1) { 595 if (rtm != NULL) 596 free(rtm, M_RTABLE); 597 m_freem(m); 598 return (error); 599 } 600 /* There is another listener, so construct message */ 601 rp = sotorawcb(so); 602 } 603 if (rtm != NULL) { 604 m_copyback(m, 0, rtm->rtm_msglen, (caddr_t)rtm); 605 if (m->m_pkthdr.len < rtm->rtm_msglen) { 606 m_freem(m); 607 m = NULL; 608 } else if (m->m_pkthdr.len > rtm->rtm_msglen) 609 m_adj(m, rtm->rtm_msglen - m->m_pkthdr.len); 610 free(rtm, M_RTABLE); 611 } 612 if (rp != NULL) 613 rp->rcb_proto.sp_family = 0; /* Avoid us */ 614 if (m != NULL) 615 rts_input(m, familyof(rtinfo.rti_dst)); 616 if (rp != NULL) 617 rp->rcb_proto.sp_family = PF_ROUTE; 618 return (error); 619 } 620 621 static void 622 rt_setmetrics(u_long which, struct rt_metrics *in, struct rt_metrics *out) 623 { 624 #define setmetric(flag, elt) if (which & (flag)) out->elt = in->elt; 625 setmetric(RTV_RPIPE, rmx_recvpipe); 626 setmetric(RTV_SPIPE, rmx_sendpipe); 627 setmetric(RTV_SSTHRESH, rmx_ssthresh); 628 setmetric(RTV_RTT, rmx_rtt); 629 setmetric(RTV_RTTVAR, rmx_rttvar); 630 setmetric(RTV_HOPCOUNT, rmx_hopcount); 631 setmetric(RTV_MTU, rmx_mtu); 632 setmetric(RTV_EXPIRE, rmx_expire); 633 #undef setmetric 634 } 635 636 #define ROUNDUP(a) \ 637 ((a) > 0 ? (1 + (((a) - 1) | (sizeof(long) - 1))) : sizeof(long)) 638 639 /* 640 * Extract the addresses of the passed sockaddrs. 641 * Do a little sanity checking so as to avoid bad memory references. 642 * This data is derived straight from userland. 643 */ 644 static int 645 rt_xaddrs(char *cp, char *cplim, struct rt_addrinfo *rtinfo) 646 { 647 struct sockaddr *sa; 648 int i; 649 650 for (i = 0; (i < RTAX_MAX) && (cp < cplim); i++) { 651 if ((rtinfo->rti_addrs & (1 << i)) == 0) 652 continue; 653 sa = (struct sockaddr *)cp; 654 /* 655 * It won't fit. 656 */ 657 if ((cp + sa->sa_len) > cplim) { 658 return (EINVAL); 659 } 660 661 /* 662 * There are no more... Quit now. 663 * If there are more bits, they are in error. 664 * I've seen this. route(1) can evidently generate these. 665 * This causes kernel to core dump. 666 * For compatibility, if we see this, point to a safe address. 667 */ 668 if (sa->sa_len == 0) { 669 static struct sockaddr sa_zero = { 670 sizeof sa_zero, AF_INET, 671 }; 672 673 rtinfo->rti_info[i] = &sa_zero; 674 return (0); /* should be EINVAL but for compat */ 675 } 676 677 /* Accept the sockaddr. */ 678 rtinfo->rti_info[i] = sa; 679 cp += ROUNDUP(sa->sa_len); 680 } 681 return (0); 682 } 683 684 static int 685 rt_msghdrsize(int type) 686 { 687 switch (type) { 688 case RTM_DELADDR: 689 case RTM_NEWADDR: 690 return sizeof(struct ifa_msghdr); 691 case RTM_DELMADDR: 692 case RTM_NEWMADDR: 693 return sizeof(struct ifma_msghdr); 694 case RTM_IFINFO: 695 return sizeof(struct if_msghdr); 696 case RTM_IFANNOUNCE: 697 return sizeof(struct if_announcemsghdr); 698 default: 699 return sizeof(struct rt_msghdr); 700 } 701 } 702 703 static int 704 rt_msgsize(int type, struct rt_addrinfo *rtinfo) 705 { 706 int len, i; 707 708 len = rt_msghdrsize(type); 709 for (i = 0; i < RTAX_MAX; i++) { 710 if (rtinfo->rti_info[i] != NULL) 711 len += ROUNDUP(rtinfo->rti_info[i]->sa_len); 712 } 713 len = ALIGN(len); 714 return len; 715 } 716 717 /* 718 * Build a routing message in a buffer. 719 * Copy the addresses in the rtinfo->rti_info[] sockaddr array 720 * to the end of the buffer after the message header. 721 * 722 * Set the rtinfo->rti_addrs bitmask of addresses present in rtinfo->rti_info[]. 723 * This side-effect can be avoided if we reorder the addrs bitmask field in all 724 * the route messages to line up so we can set it here instead of back in the 725 * calling routine. 726 */ 727 static void 728 rt_msg_buffer(int type, struct rt_addrinfo *rtinfo, void *buf, int msglen) 729 { 730 struct rt_msghdr *rtm; 731 char *cp; 732 int dlen, i; 733 734 rtm = (struct rt_msghdr *) buf; 735 rtm->rtm_version = RTM_VERSION; 736 rtm->rtm_type = type; 737 rtm->rtm_msglen = msglen; 738 739 cp = (char *)buf + rt_msghdrsize(type); 740 rtinfo->rti_addrs = 0; 741 for (i = 0; i < RTAX_MAX; i++) { 742 struct sockaddr *sa; 743 744 if ((sa = rtinfo->rti_info[i]) == NULL) 745 continue; 746 rtinfo->rti_addrs |= (1 << i); 747 dlen = ROUNDUP(sa->sa_len); 748 bcopy(sa, cp, dlen); 749 cp += dlen; 750 } 751 } 752 753 /* 754 * Build a routing message in a mbuf chain. 755 * Copy the addresses in the rtinfo->rti_info[] sockaddr array 756 * to the end of the mbuf after the message header. 757 * 758 * Set the rtinfo->rti_addrs bitmask of addresses present in rtinfo->rti_info[]. 759 * This side-effect can be avoided if we reorder the addrs bitmask field in all 760 * the route messages to line up so we can set it here instead of back in the 761 * calling routine. 762 */ 763 static struct mbuf * 764 rt_msg_mbuf(int type, struct rt_addrinfo *rtinfo) 765 { 766 struct mbuf *m; 767 struct rt_msghdr *rtm; 768 int hlen, len; 769 int i; 770 771 hlen = rt_msghdrsize(type); 772 KASSERT(hlen <= MCLBYTES, ("rt_msg_mbuf: hlen %d doesn't fit", hlen)); 773 774 m = m_getl(hlen, MB_DONTWAIT, MT_DATA, M_PKTHDR, NULL); 775 if (m == NULL) 776 return (NULL); 777 m->m_pkthdr.len = m->m_len = hlen; 778 m->m_pkthdr.rcvif = NULL; 779 rtinfo->rti_addrs = 0; 780 len = hlen; 781 for (i = 0; i < RTAX_MAX; i++) { 782 struct sockaddr *sa; 783 int dlen; 784 785 if ((sa = rtinfo->rti_info[i]) == NULL) 786 continue; 787 rtinfo->rti_addrs |= (1 << i); 788 dlen = ROUNDUP(sa->sa_len); 789 m_copyback(m, len, dlen, (caddr_t)sa); /* can grow mbuf chain */ 790 len += dlen; 791 } 792 if (m->m_pkthdr.len != len) { /* one of the m_copyback() calls failed */ 793 m_freem(m); 794 return (NULL); 795 } 796 rtm = mtod(m, struct rt_msghdr *); 797 bzero(rtm, hlen); 798 rtm->rtm_msglen = len; 799 rtm->rtm_version = RTM_VERSION; 800 rtm->rtm_type = type; 801 return (m); 802 } 803 804 /* 805 * This routine is called to generate a message from the routing 806 * socket indicating that a redirect has occurred, a routing lookup 807 * has failed, or that a protocol has detected timeouts to a particular 808 * destination. 809 */ 810 void 811 rt_missmsg(int type, struct rt_addrinfo *rtinfo, int flags, int error) 812 { 813 struct sockaddr *dst = rtinfo->rti_info[RTAX_DST]; 814 struct rt_msghdr *rtm; 815 struct mbuf *m; 816 817 if (route_cb.any_count == 0) 818 return; 819 m = rt_msg_mbuf(type, rtinfo); 820 if (m == NULL) 821 return; 822 rtm = mtod(m, struct rt_msghdr *); 823 rtm->rtm_flags = RTF_DONE | flags; 824 rtm->rtm_errno = error; 825 rtm->rtm_addrs = rtinfo->rti_addrs; 826 rts_input(m, familyof(dst)); 827 } 828 829 void 830 rt_dstmsg(int type, struct sockaddr *dst, int error) 831 { 832 struct rt_msghdr *rtm; 833 struct rt_addrinfo addrs; 834 struct mbuf *m; 835 836 if (route_cb.any_count == 0) 837 return; 838 bzero(&addrs, sizeof(struct rt_addrinfo)); 839 addrs.rti_info[RTAX_DST] = dst; 840 m = rt_msg_mbuf(type, &addrs); 841 if (m == NULL) 842 return; 843 rtm = mtod(m, struct rt_msghdr *); 844 rtm->rtm_flags = RTF_DONE; 845 rtm->rtm_errno = error; 846 rtm->rtm_addrs = addrs.rti_addrs; 847 rts_input(m, familyof(dst)); 848 } 849 850 /* 851 * This routine is called to generate a message from the routing 852 * socket indicating that the status of a network interface has changed. 853 */ 854 void 855 rt_ifmsg(struct ifnet *ifp) 856 { 857 struct if_msghdr *ifm; 858 struct mbuf *m; 859 struct rt_addrinfo rtinfo; 860 861 if (route_cb.any_count == 0) 862 return; 863 bzero(&rtinfo, sizeof(struct rt_addrinfo)); 864 m = rt_msg_mbuf(RTM_IFINFO, &rtinfo); 865 if (m == NULL) 866 return; 867 ifm = mtod(m, struct if_msghdr *); 868 ifm->ifm_index = ifp->if_index; 869 ifm->ifm_flags = ifp->if_flags; 870 ifm->ifm_data = ifp->if_data; 871 ifm->ifm_addrs = 0; 872 rts_input(m, 0); 873 } 874 875 static void 876 rt_ifamsg(int cmd, struct ifaddr *ifa) 877 { 878 struct ifa_msghdr *ifam; 879 struct rt_addrinfo rtinfo; 880 struct mbuf *m; 881 struct ifnet *ifp = ifa->ifa_ifp; 882 883 bzero(&rtinfo, sizeof(struct rt_addrinfo)); 884 rtinfo.rti_ifaaddr = ifa->ifa_addr; 885 rtinfo.rti_ifpaddr = TAILQ_FIRST(&ifp->if_addrhead)->ifa_addr; 886 rtinfo.rti_netmask = ifa->ifa_netmask; 887 rtinfo.rti_bcastaddr = ifa->ifa_dstaddr; 888 889 m = rt_msg_mbuf(cmd, &rtinfo); 890 if (m == NULL) 891 return; 892 893 ifam = mtod(m, struct ifa_msghdr *); 894 ifam->ifam_index = ifp->if_index; 895 ifam->ifam_metric = ifa->ifa_metric; 896 ifam->ifam_flags = ifa->ifa_flags; 897 ifam->ifam_addrs = rtinfo.rti_addrs; 898 899 rts_input(m, familyof(ifa->ifa_addr)); 900 } 901 902 void 903 rt_rtmsg(int cmd, struct rtentry *rt, struct ifnet *ifp, int error) 904 { 905 struct rt_msghdr *rtm; 906 struct rt_addrinfo rtinfo; 907 struct mbuf *m; 908 struct sockaddr *dst; 909 910 if (rt == NULL) 911 return; 912 913 bzero(&rtinfo, sizeof(struct rt_addrinfo)); 914 rtinfo.rti_dst = dst = rt_key(rt); 915 rtinfo.rti_gateway = rt->rt_gateway; 916 rtinfo.rti_netmask = rt_mask(rt); 917 if (ifp != NULL) 918 rtinfo.rti_ifpaddr = TAILQ_FIRST(&ifp->if_addrhead)->ifa_addr; 919 rtinfo.rti_ifaaddr = rt->rt_ifa->ifa_addr; 920 921 m = rt_msg_mbuf(cmd, &rtinfo); 922 if (m == NULL) 923 return; 924 925 rtm = mtod(m, struct rt_msghdr *); 926 if (ifp != NULL) 927 rtm->rtm_index = ifp->if_index; 928 rtm->rtm_flags |= rt->rt_flags; 929 rtm->rtm_errno = error; 930 rtm->rtm_addrs = rtinfo.rti_addrs; 931 932 rts_input(m, familyof(dst)); 933 } 934 935 /* 936 * This is called to generate messages from the routing socket 937 * indicating a network interface has had addresses associated with it. 938 * if we ever reverse the logic and replace messages TO the routing 939 * socket indicate a request to configure interfaces, then it will 940 * be unnecessary as the routing socket will automatically generate 941 * copies of it. 942 */ 943 void 944 rt_newaddrmsg(int cmd, struct ifaddr *ifa, int error, struct rtentry *rt) 945 { 946 #ifdef SCTP 947 /* 948 * notify the SCTP stack 949 * this will only get called when an address is added/deleted 950 * XXX pass the ifaddr struct instead if ifa->ifa_addr... 951 */ 952 if (cmd == RTM_ADD) 953 sctp_add_ip_address(ifa); 954 else if (cmd == RTM_DELETE) 955 sctp_delete_ip_address(ifa); 956 #endif /* SCTP */ 957 958 if (route_cb.any_count == 0) 959 return; 960 961 if (cmd == RTM_ADD) { 962 rt_ifamsg(RTM_NEWADDR, ifa); 963 rt_rtmsg(RTM_ADD, rt, ifa->ifa_ifp, error); 964 } else { 965 KASSERT((cmd == RTM_DELETE), ("unknown cmd %d", cmd)); 966 rt_rtmsg(RTM_DELETE, rt, ifa->ifa_ifp, error); 967 rt_ifamsg(RTM_DELADDR, ifa); 968 } 969 } 970 971 /* 972 * This is the analogue to the rt_newaddrmsg which performs the same 973 * function but for multicast group memberhips. This is easier since 974 * there is no route state to worry about. 975 */ 976 void 977 rt_newmaddrmsg(int cmd, struct ifmultiaddr *ifma) 978 { 979 struct rt_addrinfo rtinfo; 980 struct mbuf *m = NULL; 981 struct ifnet *ifp = ifma->ifma_ifp; 982 struct ifma_msghdr *ifmam; 983 984 if (route_cb.any_count == 0) 985 return; 986 987 bzero(&rtinfo, sizeof(struct rt_addrinfo)); 988 rtinfo.rti_ifaaddr = ifma->ifma_addr; 989 if (ifp != NULL && !TAILQ_EMPTY(&ifp->if_addrhead)) 990 rtinfo.rti_ifpaddr = TAILQ_FIRST(&ifp->if_addrhead)->ifa_addr; 991 /* 992 * If a link-layer address is present, present it as a ``gateway'' 993 * (similarly to how ARP entries, e.g., are presented). 994 */ 995 rtinfo.rti_gateway = ifma->ifma_lladdr; 996 997 m = rt_msg_mbuf(cmd, &rtinfo); 998 if (m == NULL) 999 return; 1000 1001 ifmam = mtod(m, struct ifma_msghdr *); 1002 ifmam->ifmam_index = ifp->if_index; 1003 ifmam->ifmam_addrs = rtinfo.rti_addrs; 1004 1005 rts_input(m, familyof(ifma->ifma_addr)); 1006 } 1007 1008 /* 1009 * This is called to generate routing socket messages indicating 1010 * network interface arrival and departure. 1011 */ 1012 void 1013 rt_ifannouncemsg(struct ifnet *ifp, int what) 1014 { 1015 struct rt_addrinfo addrinfo; 1016 struct mbuf *m; 1017 struct if_announcemsghdr *ifan; 1018 1019 if (route_cb.any_count == 0) 1020 return; 1021 1022 bzero(&addrinfo, sizeof addrinfo); 1023 m = rt_msg_mbuf(RTM_IFANNOUNCE, &addrinfo); 1024 if (m == NULL) 1025 return; 1026 1027 ifan = mtod(m, struct if_announcemsghdr *); 1028 ifan->ifan_index = ifp->if_index; 1029 strlcpy(ifan->ifan_name, ifp->if_xname, sizeof ifan->ifan_name); 1030 ifan->ifan_what = what; 1031 1032 rts_input(m, 0); 1033 } 1034 1035 static int 1036 resizewalkarg(struct walkarg *w, int len) 1037 { 1038 void *newptr; 1039 1040 newptr = malloc(len, M_RTABLE, M_INTWAIT | M_NULLOK); 1041 if (newptr == NULL) 1042 return (ENOMEM); 1043 if (w->w_tmem != NULL) 1044 free(w->w_tmem, M_RTABLE); 1045 w->w_tmem = newptr; 1046 w->w_tmemsize = len; 1047 return (0); 1048 } 1049 1050 /* 1051 * This is used in dumping the kernel table via sysctl(). 1052 */ 1053 int 1054 sysctl_dumpentry(struct radix_node *rn, void *vw) 1055 { 1056 struct walkarg *w = vw; 1057 struct rtentry *rt = (struct rtentry *)rn; 1058 struct rt_addrinfo rtinfo; 1059 int error, msglen; 1060 1061 if (w->w_op == NET_RT_FLAGS && !(rt->rt_flags & w->w_arg)) 1062 return 0; 1063 1064 bzero(&rtinfo, sizeof(struct rt_addrinfo)); 1065 rtinfo.rti_dst = rt_key(rt); 1066 rtinfo.rti_gateway = rt->rt_gateway; 1067 rtinfo.rti_netmask = rt_mask(rt); 1068 rtinfo.rti_genmask = rt->rt_genmask; 1069 if (rt->rt_ifp != NULL) { 1070 rtinfo.rti_ifpaddr = 1071 TAILQ_FIRST(&rt->rt_ifp->if_addrhead)->ifa_addr; 1072 rtinfo.rti_ifaaddr = rt->rt_ifa->ifa_addr; 1073 if (rt->rt_ifp->if_flags & IFF_POINTOPOINT) 1074 rtinfo.rti_bcastaddr = rt->rt_ifa->ifa_dstaddr; 1075 } 1076 msglen = rt_msgsize(RTM_GET, &rtinfo); 1077 if (w->w_tmemsize < msglen && resizewalkarg(w, msglen) != 0) 1078 return (ENOMEM); 1079 rt_msg_buffer(RTM_GET, &rtinfo, w->w_tmem, msglen); 1080 if (w->w_req != NULL) { 1081 struct rt_msghdr *rtm = w->w_tmem; 1082 1083 rtm->rtm_flags = rt->rt_flags; 1084 rtm->rtm_use = rt->rt_use; 1085 rtm->rtm_rmx = rt->rt_rmx; 1086 rtm->rtm_index = rt->rt_ifp->if_index; 1087 rtm->rtm_errno = rtm->rtm_pid = rtm->rtm_seq = 0; 1088 rtm->rtm_addrs = rtinfo.rti_addrs; 1089 error = SYSCTL_OUT(w->w_req, rtm, msglen); 1090 return (error); 1091 } 1092 return (0); 1093 } 1094 1095 static int 1096 sysctl_iflist(int af, struct walkarg *w) 1097 { 1098 struct ifnet *ifp; 1099 struct ifaddr *ifa; 1100 struct rt_addrinfo rtinfo; 1101 int msglen, error; 1102 1103 bzero(&rtinfo, sizeof(struct rt_addrinfo)); 1104 TAILQ_FOREACH(ifp, &ifnet, if_link) { 1105 if (w->w_arg && w->w_arg != ifp->if_index) 1106 continue; 1107 ifa = TAILQ_FIRST(&ifp->if_addrhead); 1108 rtinfo.rti_ifpaddr = ifa->ifa_addr; 1109 msglen = rt_msgsize(RTM_IFINFO, &rtinfo); 1110 if (w->w_tmemsize < msglen && resizewalkarg(w, msglen) != 0) 1111 return (ENOMEM); 1112 rt_msg_buffer(RTM_IFINFO, &rtinfo, w->w_tmem, msglen); 1113 rtinfo.rti_ifpaddr = NULL; 1114 if (w->w_req != NULL && w->w_tmem != NULL) { 1115 struct if_msghdr *ifm = w->w_tmem; 1116 1117 ifm->ifm_index = ifp->if_index; 1118 ifm->ifm_flags = ifp->if_flags; 1119 ifm->ifm_data = ifp->if_data; 1120 ifm->ifm_addrs = rtinfo.rti_addrs; 1121 error = SYSCTL_OUT(w->w_req, ifm, msglen); 1122 if (error) 1123 return (error); 1124 } 1125 while ((ifa = TAILQ_NEXT(ifa, ifa_link)) != NULL) { 1126 if (af && af != ifa->ifa_addr->sa_family) 1127 continue; 1128 if (curproc->p_ucred->cr_prison && 1129 prison_if(curthread, ifa->ifa_addr)) 1130 continue; 1131 rtinfo.rti_ifaaddr = ifa->ifa_addr; 1132 rtinfo.rti_netmask = ifa->ifa_netmask; 1133 rtinfo.rti_bcastaddr = ifa->ifa_dstaddr; 1134 msglen = rt_msgsize(RTM_NEWADDR, &rtinfo); 1135 if (w->w_tmemsize < msglen && 1136 resizewalkarg(w, msglen) != 0) 1137 return (ENOMEM); 1138 rt_msg_buffer(RTM_NEWADDR, &rtinfo, w->w_tmem, msglen); 1139 if (w->w_req != NULL) { 1140 struct ifa_msghdr *ifam = w->w_tmem; 1141 1142 ifam->ifam_index = ifa->ifa_ifp->if_index; 1143 ifam->ifam_flags = ifa->ifa_flags; 1144 ifam->ifam_metric = ifa->ifa_metric; 1145 ifam->ifam_addrs = rtinfo.rti_addrs; 1146 error = SYSCTL_OUT(w->w_req, w->w_tmem, msglen); 1147 if (error) 1148 return (error); 1149 } 1150 } 1151 rtinfo.rti_netmask = NULL; 1152 rtinfo.rti_ifaaddr = NULL; 1153 rtinfo.rti_bcastaddr = NULL; 1154 } 1155 return (0); 1156 } 1157 1158 static int 1159 sysctl_rtsock(SYSCTL_HANDLER_ARGS) 1160 { 1161 int *name = (int *)arg1; 1162 u_int namelen = arg2; 1163 struct radix_node_head *rnh; 1164 int i, error = EINVAL; 1165 u_char af; 1166 struct walkarg w; 1167 1168 name ++; 1169 namelen--; 1170 if (req->newptr) 1171 return (EPERM); 1172 if (namelen != 3) 1173 return (EINVAL); 1174 af = name[0]; 1175 bzero(&w, sizeof w); 1176 w.w_op = name[1]; 1177 w.w_arg = name[2]; 1178 w.w_req = req; 1179 1180 crit_enter(); 1181 switch (w.w_op) { 1182 1183 case NET_RT_DUMP: 1184 case NET_RT_FLAGS: 1185 for (i = 1; i <= AF_MAX; i++) 1186 if ((rnh = rt_tables[i]) && (af == 0 || af == i) && 1187 (error = rnh->rnh_walktree(rnh, 1188 sysctl_dumpentry, &w))) 1189 break; 1190 break; 1191 1192 case NET_RT_IFLIST: 1193 error = sysctl_iflist(af, &w); 1194 } 1195 crit_exit(); 1196 if (w.w_tmem != NULL) 1197 free(w.w_tmem, M_RTABLE); 1198 return (error); 1199 } 1200 1201 SYSCTL_NODE(_net, PF_ROUTE, routetable, CTLFLAG_RD, sysctl_rtsock, ""); 1202 1203 /* 1204 * Definitions of protocols supported in the ROUTE domain. 1205 */ 1206 1207 extern struct domain routedomain; /* or at least forward */ 1208 1209 static struct protosw routesw[] = { 1210 { SOCK_RAW, &routedomain, 0, PR_ATOMIC|PR_ADDR, 1211 0, route_output, raw_ctlinput, 0, 1212 cpu0_soport, 1213 raw_init, 0, 0, 0, 1214 &route_usrreqs 1215 } 1216 }; 1217 1218 static struct domain routedomain = { 1219 PF_ROUTE, "route", NULL, NULL, NULL, 1220 routesw, &routesw[(sizeof routesw)/(sizeof routesw[0])], 1221 }; 1222 1223 DOMAIN_SET(route); 1224