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 1.32 2006/05/18 13:51:45 sephe 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 static void route_output_add_callback(int, int, struct rt_addrinfo *, 409 struct rtentry *, void *); 410 static void route_output_delete_callback(int, int, struct rt_addrinfo *, 411 struct rtentry *, void *); 412 static void route_output_change_callback(int, int, struct rt_addrinfo *, 413 struct rtentry *, void *); 414 static void route_output_lock_callback(int, int, struct rt_addrinfo *, 415 struct rtentry *, void *); 416 417 /*ARGSUSED*/ 418 static int 419 route_output(struct mbuf *m, struct socket *so, ...) 420 { 421 struct rt_msghdr *rtm = NULL; 422 struct rtentry *rt; 423 struct radix_node_head *rnh; 424 struct rawcb *rp = NULL; 425 struct pr_output_info *oi; 426 struct rt_addrinfo rtinfo; 427 int len, error = 0; 428 __va_list ap; 429 430 __va_start(ap, so); 431 oi = __va_arg(ap, struct pr_output_info *); 432 __va_end(ap); 433 434 #define gotoerr(e) { error = e; goto flush;} 435 436 if (m == NULL || 437 (m->m_len < sizeof(long) && 438 (m = m_pullup(m, sizeof(long))) == NULL)) 439 return (ENOBUFS); 440 if (!(m->m_flags & M_PKTHDR)) 441 panic("route_output"); 442 len = m->m_pkthdr.len; 443 if (len < sizeof(struct rt_msghdr) || 444 len != mtod(m, struct rt_msghdr *)->rtm_msglen) { 445 rtinfo.rti_dst = NULL; 446 gotoerr(EINVAL); 447 } 448 rtm = malloc(len, M_RTABLE, M_INTWAIT | M_NULLOK); 449 if (rtm == NULL) { 450 rtinfo.rti_dst = NULL; 451 gotoerr(ENOBUFS); 452 } 453 m_copydata(m, 0, len, (caddr_t)rtm); 454 if (rtm->rtm_version != RTM_VERSION) { 455 rtinfo.rti_dst = NULL; 456 gotoerr(EPROTONOSUPPORT); 457 } 458 rtm->rtm_pid = oi->p_pid; 459 bzero(&rtinfo, sizeof(struct rt_addrinfo)); 460 rtinfo.rti_addrs = rtm->rtm_addrs; 461 if (rt_xaddrs((char *)(rtm + 1), (char *)rtm + len, &rtinfo) != 0) { 462 rtinfo.rti_dst = NULL; 463 gotoerr(EINVAL); 464 } 465 rtinfo.rti_flags = rtm->rtm_flags; 466 if (rtinfo.rti_dst == NULL || rtinfo.rti_dst->sa_family >= AF_MAX || 467 (rtinfo.rti_gateway && rtinfo.rti_gateway->sa_family >= AF_MAX)) 468 gotoerr(EINVAL); 469 470 if (rtinfo.rti_genmask != NULL) { 471 struct radix_node *n; 472 473 #define clen(s) (*(u_char *)(s)) 474 n = rn_addmask((char *)rtinfo.rti_genmask, TRUE, 1); 475 if (n != NULL && 476 rtinfo.rti_genmask->sa_len >= clen(n->rn_key) && 477 bcmp((char *)rtinfo.rti_genmask + 1, 478 (char *)n->rn_key + 1, clen(n->rn_key) - 1) == 0) 479 rtinfo.rti_genmask = (struct sockaddr *)n->rn_key; 480 else 481 gotoerr(ENOBUFS); 482 } 483 484 /* 485 * Verify that the caller has the appropriate privilege; RTM_GET 486 * is the only operation the non-superuser is allowed. 487 */ 488 if (rtm->rtm_type != RTM_GET && suser_cred(so->so_cred, 0) != 0) 489 gotoerr(EPERM); 490 491 switch (rtm->rtm_type) { 492 case RTM_ADD: 493 if (rtinfo.rti_gateway == NULL) { 494 error = EINVAL; 495 } else { 496 error = rtrequest1_global(RTM_ADD, &rtinfo, 497 route_output_add_callback, rtm); 498 } 499 break; 500 case RTM_DELETE: 501 /* 502 * note: &rtm passed as argument so 'rtm' can be replaced. 503 */ 504 error = rtrequest1_global(RTM_DELETE, &rtinfo, 505 route_output_delete_callback, &rtm); 506 break; 507 case RTM_GET: 508 rnh = rt_tables[mycpuid][rtinfo.rti_dst->sa_family]; 509 if (rnh == NULL) { 510 error = EAFNOSUPPORT; 511 break; 512 } 513 rt = (struct rtentry *) 514 rnh->rnh_lookup((char *)rtinfo.rti_dst, 515 (char *)rtinfo.rti_netmask, rnh); 516 if (rt == NULL) { 517 error = ESRCH; 518 break; 519 } 520 rt->rt_refcnt++; 521 if (fillrtmsg(&rtm, rt, &rtinfo) != 0) 522 gotoerr(ENOBUFS); 523 --rt->rt_refcnt; 524 break; 525 case RTM_CHANGE: 526 error = rtrequest1_global(RTM_GET, &rtinfo, 527 route_output_change_callback, rtm); 528 break; 529 case RTM_LOCK: 530 error = rtrequest1_global(RTM_GET, &rtinfo, 531 route_output_lock_callback, rtm); 532 break; 533 default: 534 error = EOPNOTSUPP; 535 break; 536 } 537 538 flush: 539 if (rtm != NULL) { 540 if (error != 0) 541 rtm->rtm_errno = error; 542 else 543 rtm->rtm_flags |= RTF_DONE; 544 } 545 546 /* 547 * Check to see if we don't want our own messages. 548 */ 549 if (!(so->so_options & SO_USELOOPBACK)) { 550 if (route_cb.any_count <= 1) { 551 if (rtm != NULL) 552 free(rtm, M_RTABLE); 553 m_freem(m); 554 return (error); 555 } 556 /* There is another listener, so construct message */ 557 rp = sotorawcb(so); 558 } 559 if (rtm != NULL) { 560 m_copyback(m, 0, rtm->rtm_msglen, (caddr_t)rtm); 561 if (m->m_pkthdr.len < rtm->rtm_msglen) { 562 m_freem(m); 563 m = NULL; 564 } else if (m->m_pkthdr.len > rtm->rtm_msglen) 565 m_adj(m, rtm->rtm_msglen - m->m_pkthdr.len); 566 free(rtm, M_RTABLE); 567 } 568 if (rp != NULL) 569 rp->rcb_proto.sp_family = 0; /* Avoid us */ 570 if (m != NULL) 571 rts_input(m, familyof(rtinfo.rti_dst)); 572 if (rp != NULL) 573 rp->rcb_proto.sp_family = PF_ROUTE; 574 return (error); 575 } 576 577 static void 578 route_output_add_callback(int cmd, int error, struct rt_addrinfo *rtinfo, 579 struct rtentry *rt, void *arg) 580 { 581 struct rt_msghdr *rtm = arg; 582 583 if (error == 0 && rt != NULL) { 584 rt_setmetrics(rtm->rtm_inits, &rtm->rtm_rmx, 585 &rt->rt_rmx); 586 rt->rt_rmx.rmx_locks &= ~(rtm->rtm_inits); 587 rt->rt_rmx.rmx_locks |= 588 (rtm->rtm_inits & rtm->rtm_rmx.rmx_locks); 589 rt->rt_genmask = rtinfo->rti_genmask; 590 } 591 } 592 593 static void 594 route_output_delete_callback(int cmd, int error, struct rt_addrinfo *rtinfo, 595 struct rtentry *rt, void *arg) 596 { 597 struct rt_msghdr **rtm = arg; 598 599 if (error == 0 && rt) { 600 ++rt->rt_refcnt; 601 if (fillrtmsg(rtm, rt, rtinfo) != 0) { 602 error = ENOBUFS; 603 /* XXX no way to return the error */ 604 } 605 --rt->rt_refcnt; 606 } 607 } 608 609 static void 610 route_output_change_callback(int cmd, int error, struct rt_addrinfo *rtinfo, 611 struct rtentry *rt, void *arg) 612 { 613 struct rt_msghdr *rtm = arg; 614 struct ifaddr *ifa; 615 616 if (error) 617 goto done; 618 619 /* 620 * new gateway could require new ifaddr, ifp; 621 * flags may also be different; ifp may be specified 622 * by ll sockaddr when protocol address is ambiguous 623 */ 624 if (((rt->rt_flags & RTF_GATEWAY) && rtinfo->rti_gateway != NULL) || 625 rtinfo->rti_ifpaddr != NULL || (rtinfo->rti_ifaaddr != NULL && 626 sa_equal(rtinfo->rti_ifaaddr, rt->rt_ifa->ifa_addr)) 627 ) { 628 error = rt_getifa(rtinfo); 629 if (error != 0) 630 goto done; 631 } 632 if (rtinfo->rti_gateway != NULL) { 633 error = rt_setgate(rt, rt_key(rt), rtinfo->rti_gateway); 634 if (error != 0) 635 goto done; 636 } 637 if ((ifa = rtinfo->rti_ifa) != NULL) { 638 struct ifaddr *oifa = rt->rt_ifa; 639 640 if (oifa != ifa) { 641 if (oifa && oifa->ifa_rtrequest) 642 oifa->ifa_rtrequest(RTM_DELETE, rt, rtinfo); 643 IFAFREE(rt->rt_ifa); 644 IFAREF(ifa); 645 rt->rt_ifa = ifa; 646 rt->rt_ifp = rtinfo->rti_ifp; 647 } 648 } 649 rt_setmetrics(rtm->rtm_inits, &rtm->rtm_rmx, &rt->rt_rmx); 650 if (rt->rt_ifa && rt->rt_ifa->ifa_rtrequest) 651 rt->rt_ifa->ifa_rtrequest(RTM_ADD, rt, rtinfo); 652 if (rtinfo->rti_genmask != NULL) 653 rt->rt_genmask = rtinfo->rti_genmask; 654 done: 655 /* XXX no way to return error */ 656 ; 657 } 658 659 static void 660 route_output_lock_callback(int cmd, int error, struct rt_addrinfo *rtinfo, 661 struct rtentry *rt, void *arg) 662 { 663 struct rt_msghdr *rtm = arg; 664 665 rt->rt_rmx.rmx_locks &= ~(rtm->rtm_inits); 666 rt->rt_rmx.rmx_locks |= 667 (rtm->rtm_inits & rtm->rtm_rmx.rmx_locks); 668 } 669 670 static void 671 rt_setmetrics(u_long which, struct rt_metrics *in, struct rt_metrics *out) 672 { 673 #define setmetric(flag, elt) if (which & (flag)) out->elt = in->elt; 674 setmetric(RTV_RPIPE, rmx_recvpipe); 675 setmetric(RTV_SPIPE, rmx_sendpipe); 676 setmetric(RTV_SSTHRESH, rmx_ssthresh); 677 setmetric(RTV_RTT, rmx_rtt); 678 setmetric(RTV_RTTVAR, rmx_rttvar); 679 setmetric(RTV_HOPCOUNT, rmx_hopcount); 680 setmetric(RTV_MTU, rmx_mtu); 681 setmetric(RTV_EXPIRE, rmx_expire); 682 #undef setmetric 683 } 684 685 #define ROUNDUP(a) \ 686 ((a) > 0 ? (1 + (((a) - 1) | (sizeof(long) - 1))) : sizeof(long)) 687 688 /* 689 * Extract the addresses of the passed sockaddrs. 690 * Do a little sanity checking so as to avoid bad memory references. 691 * This data is derived straight from userland. 692 */ 693 static int 694 rt_xaddrs(char *cp, char *cplim, struct rt_addrinfo *rtinfo) 695 { 696 struct sockaddr *sa; 697 int i; 698 699 for (i = 0; (i < RTAX_MAX) && (cp < cplim); i++) { 700 if ((rtinfo->rti_addrs & (1 << i)) == 0) 701 continue; 702 sa = (struct sockaddr *)cp; 703 /* 704 * It won't fit. 705 */ 706 if ((cp + sa->sa_len) > cplim) { 707 return (EINVAL); 708 } 709 710 /* 711 * There are no more... Quit now. 712 * If there are more bits, they are in error. 713 * I've seen this. route(1) can evidently generate these. 714 * This causes kernel to core dump. 715 * For compatibility, if we see this, point to a safe address. 716 */ 717 if (sa->sa_len == 0) { 718 static struct sockaddr sa_zero = { 719 sizeof sa_zero, AF_INET, 720 }; 721 722 rtinfo->rti_info[i] = &sa_zero; 723 return (0); /* should be EINVAL but for compat */ 724 } 725 726 /* Accept the sockaddr. */ 727 rtinfo->rti_info[i] = sa; 728 cp += ROUNDUP(sa->sa_len); 729 } 730 return (0); 731 } 732 733 static int 734 rt_msghdrsize(int type) 735 { 736 switch (type) { 737 case RTM_DELADDR: 738 case RTM_NEWADDR: 739 return sizeof(struct ifa_msghdr); 740 case RTM_DELMADDR: 741 case RTM_NEWMADDR: 742 return sizeof(struct ifma_msghdr); 743 case RTM_IFINFO: 744 return sizeof(struct if_msghdr); 745 case RTM_IFANNOUNCE: 746 case RTM_IEEE80211: 747 return sizeof(struct if_announcemsghdr); 748 default: 749 return sizeof(struct rt_msghdr); 750 } 751 } 752 753 static int 754 rt_msgsize(int type, struct rt_addrinfo *rtinfo) 755 { 756 int len, i; 757 758 len = rt_msghdrsize(type); 759 for (i = 0; i < RTAX_MAX; i++) { 760 if (rtinfo->rti_info[i] != NULL) 761 len += ROUNDUP(rtinfo->rti_info[i]->sa_len); 762 } 763 len = ALIGN(len); 764 return len; 765 } 766 767 /* 768 * Build a routing message in a buffer. 769 * Copy the addresses in the rtinfo->rti_info[] sockaddr array 770 * to the end of the buffer after the message header. 771 * 772 * Set the rtinfo->rti_addrs bitmask of addresses present in rtinfo->rti_info[]. 773 * This side-effect can be avoided if we reorder the addrs bitmask field in all 774 * the route messages to line up so we can set it here instead of back in the 775 * calling routine. 776 */ 777 static void 778 rt_msg_buffer(int type, struct rt_addrinfo *rtinfo, void *buf, int msglen) 779 { 780 struct rt_msghdr *rtm; 781 char *cp; 782 int dlen, i; 783 784 rtm = (struct rt_msghdr *) buf; 785 rtm->rtm_version = RTM_VERSION; 786 rtm->rtm_type = type; 787 rtm->rtm_msglen = msglen; 788 789 cp = (char *)buf + rt_msghdrsize(type); 790 rtinfo->rti_addrs = 0; 791 for (i = 0; i < RTAX_MAX; i++) { 792 struct sockaddr *sa; 793 794 if ((sa = rtinfo->rti_info[i]) == NULL) 795 continue; 796 rtinfo->rti_addrs |= (1 << i); 797 dlen = ROUNDUP(sa->sa_len); 798 bcopy(sa, cp, dlen); 799 cp += dlen; 800 } 801 } 802 803 /* 804 * Build a routing message in a mbuf chain. 805 * Copy the addresses in the rtinfo->rti_info[] sockaddr array 806 * to the end of the mbuf after the message header. 807 * 808 * Set the rtinfo->rti_addrs bitmask of addresses present in rtinfo->rti_info[]. 809 * This side-effect can be avoided if we reorder the addrs bitmask field in all 810 * the route messages to line up so we can set it here instead of back in the 811 * calling routine. 812 */ 813 static struct mbuf * 814 rt_msg_mbuf(int type, struct rt_addrinfo *rtinfo) 815 { 816 struct mbuf *m; 817 struct rt_msghdr *rtm; 818 int hlen, len; 819 int i; 820 821 hlen = rt_msghdrsize(type); 822 KASSERT(hlen <= MCLBYTES, ("rt_msg_mbuf: hlen %d doesn't fit", hlen)); 823 824 m = m_getl(hlen, MB_DONTWAIT, MT_DATA, M_PKTHDR, NULL); 825 if (m == NULL) 826 return (NULL); 827 m->m_pkthdr.len = m->m_len = hlen; 828 m->m_pkthdr.rcvif = NULL; 829 rtinfo->rti_addrs = 0; 830 len = hlen; 831 for (i = 0; i < RTAX_MAX; i++) { 832 struct sockaddr *sa; 833 int dlen; 834 835 if ((sa = rtinfo->rti_info[i]) == NULL) 836 continue; 837 rtinfo->rti_addrs |= (1 << i); 838 dlen = ROUNDUP(sa->sa_len); 839 m_copyback(m, len, dlen, (caddr_t)sa); /* can grow mbuf chain */ 840 len += dlen; 841 } 842 if (m->m_pkthdr.len != len) { /* one of the m_copyback() calls failed */ 843 m_freem(m); 844 return (NULL); 845 } 846 rtm = mtod(m, struct rt_msghdr *); 847 bzero(rtm, hlen); 848 rtm->rtm_msglen = len; 849 rtm->rtm_version = RTM_VERSION; 850 rtm->rtm_type = type; 851 return (m); 852 } 853 854 /* 855 * This routine is called to generate a message from the routing 856 * socket indicating that a redirect has occurred, a routing lookup 857 * has failed, or that a protocol has detected timeouts to a particular 858 * destination. 859 */ 860 void 861 rt_missmsg(int type, struct rt_addrinfo *rtinfo, int flags, int error) 862 { 863 struct sockaddr *dst = rtinfo->rti_info[RTAX_DST]; 864 struct rt_msghdr *rtm; 865 struct mbuf *m; 866 867 if (route_cb.any_count == 0) 868 return; 869 m = rt_msg_mbuf(type, rtinfo); 870 if (m == NULL) 871 return; 872 rtm = mtod(m, struct rt_msghdr *); 873 rtm->rtm_flags = RTF_DONE | flags; 874 rtm->rtm_errno = error; 875 rtm->rtm_addrs = rtinfo->rti_addrs; 876 rts_input(m, familyof(dst)); 877 } 878 879 void 880 rt_dstmsg(int type, struct sockaddr *dst, int error) 881 { 882 struct rt_msghdr *rtm; 883 struct rt_addrinfo addrs; 884 struct mbuf *m; 885 886 if (route_cb.any_count == 0) 887 return; 888 bzero(&addrs, sizeof(struct rt_addrinfo)); 889 addrs.rti_info[RTAX_DST] = dst; 890 m = rt_msg_mbuf(type, &addrs); 891 if (m == NULL) 892 return; 893 rtm = mtod(m, struct rt_msghdr *); 894 rtm->rtm_flags = RTF_DONE; 895 rtm->rtm_errno = error; 896 rtm->rtm_addrs = addrs.rti_addrs; 897 rts_input(m, familyof(dst)); 898 } 899 900 /* 901 * This routine is called to generate a message from the routing 902 * socket indicating that the status of a network interface has changed. 903 */ 904 void 905 rt_ifmsg(struct ifnet *ifp) 906 { 907 struct if_msghdr *ifm; 908 struct mbuf *m; 909 struct rt_addrinfo rtinfo; 910 911 if (route_cb.any_count == 0) 912 return; 913 bzero(&rtinfo, sizeof(struct rt_addrinfo)); 914 m = rt_msg_mbuf(RTM_IFINFO, &rtinfo); 915 if (m == NULL) 916 return; 917 ifm = mtod(m, struct if_msghdr *); 918 ifm->ifm_index = ifp->if_index; 919 ifm->ifm_flags = ifp->if_flags; 920 ifm->ifm_data = ifp->if_data; 921 ifm->ifm_addrs = 0; 922 rts_input(m, 0); 923 } 924 925 static void 926 rt_ifamsg(int cmd, struct ifaddr *ifa) 927 { 928 struct ifa_msghdr *ifam; 929 struct rt_addrinfo rtinfo; 930 struct mbuf *m; 931 struct ifnet *ifp = ifa->ifa_ifp; 932 933 bzero(&rtinfo, sizeof(struct rt_addrinfo)); 934 rtinfo.rti_ifaaddr = ifa->ifa_addr; 935 rtinfo.rti_ifpaddr = TAILQ_FIRST(&ifp->if_addrhead)->ifa_addr; 936 rtinfo.rti_netmask = ifa->ifa_netmask; 937 rtinfo.rti_bcastaddr = ifa->ifa_dstaddr; 938 939 m = rt_msg_mbuf(cmd, &rtinfo); 940 if (m == NULL) 941 return; 942 943 ifam = mtod(m, struct ifa_msghdr *); 944 ifam->ifam_index = ifp->if_index; 945 ifam->ifam_metric = ifa->ifa_metric; 946 ifam->ifam_flags = ifa->ifa_flags; 947 ifam->ifam_addrs = rtinfo.rti_addrs; 948 949 rts_input(m, familyof(ifa->ifa_addr)); 950 } 951 952 void 953 rt_rtmsg(int cmd, struct rtentry *rt, struct ifnet *ifp, int error) 954 { 955 struct rt_msghdr *rtm; 956 struct rt_addrinfo rtinfo; 957 struct mbuf *m; 958 struct sockaddr *dst; 959 960 if (rt == NULL) 961 return; 962 963 bzero(&rtinfo, sizeof(struct rt_addrinfo)); 964 rtinfo.rti_dst = dst = rt_key(rt); 965 rtinfo.rti_gateway = rt->rt_gateway; 966 rtinfo.rti_netmask = rt_mask(rt); 967 if (ifp != NULL) 968 rtinfo.rti_ifpaddr = TAILQ_FIRST(&ifp->if_addrhead)->ifa_addr; 969 rtinfo.rti_ifaaddr = rt->rt_ifa->ifa_addr; 970 971 m = rt_msg_mbuf(cmd, &rtinfo); 972 if (m == NULL) 973 return; 974 975 rtm = mtod(m, struct rt_msghdr *); 976 if (ifp != NULL) 977 rtm->rtm_index = ifp->if_index; 978 rtm->rtm_flags |= rt->rt_flags; 979 rtm->rtm_errno = error; 980 rtm->rtm_addrs = rtinfo.rti_addrs; 981 982 rts_input(m, familyof(dst)); 983 } 984 985 /* 986 * This is called to generate messages from the routing socket 987 * indicating a network interface has had addresses associated with it. 988 * if we ever reverse the logic and replace messages TO the routing 989 * socket indicate a request to configure interfaces, then it will 990 * be unnecessary as the routing socket will automatically generate 991 * copies of it. 992 */ 993 void 994 rt_newaddrmsg(int cmd, struct ifaddr *ifa, int error, struct rtentry *rt) 995 { 996 #ifdef SCTP 997 /* 998 * notify the SCTP stack 999 * this will only get called when an address is added/deleted 1000 * XXX pass the ifaddr struct instead if ifa->ifa_addr... 1001 */ 1002 if (cmd == RTM_ADD) 1003 sctp_add_ip_address(ifa); 1004 else if (cmd == RTM_DELETE) 1005 sctp_delete_ip_address(ifa); 1006 #endif /* SCTP */ 1007 1008 if (route_cb.any_count == 0) 1009 return; 1010 1011 if (cmd == RTM_ADD) { 1012 rt_ifamsg(RTM_NEWADDR, ifa); 1013 rt_rtmsg(RTM_ADD, rt, ifa->ifa_ifp, error); 1014 } else { 1015 KASSERT((cmd == RTM_DELETE), ("unknown cmd %d", cmd)); 1016 rt_rtmsg(RTM_DELETE, rt, ifa->ifa_ifp, error); 1017 rt_ifamsg(RTM_DELADDR, ifa); 1018 } 1019 } 1020 1021 /* 1022 * This is the analogue to the rt_newaddrmsg which performs the same 1023 * function but for multicast group memberhips. This is easier since 1024 * there is no route state to worry about. 1025 */ 1026 void 1027 rt_newmaddrmsg(int cmd, struct ifmultiaddr *ifma) 1028 { 1029 struct rt_addrinfo rtinfo; 1030 struct mbuf *m = NULL; 1031 struct ifnet *ifp = ifma->ifma_ifp; 1032 struct ifma_msghdr *ifmam; 1033 1034 if (route_cb.any_count == 0) 1035 return; 1036 1037 bzero(&rtinfo, sizeof(struct rt_addrinfo)); 1038 rtinfo.rti_ifaaddr = ifma->ifma_addr; 1039 if (ifp != NULL && !TAILQ_EMPTY(&ifp->if_addrhead)) 1040 rtinfo.rti_ifpaddr = TAILQ_FIRST(&ifp->if_addrhead)->ifa_addr; 1041 /* 1042 * If a link-layer address is present, present it as a ``gateway'' 1043 * (similarly to how ARP entries, e.g., are presented). 1044 */ 1045 rtinfo.rti_gateway = ifma->ifma_lladdr; 1046 1047 m = rt_msg_mbuf(cmd, &rtinfo); 1048 if (m == NULL) 1049 return; 1050 1051 ifmam = mtod(m, struct ifma_msghdr *); 1052 ifmam->ifmam_index = ifp->if_index; 1053 ifmam->ifmam_addrs = rtinfo.rti_addrs; 1054 1055 rts_input(m, familyof(ifma->ifma_addr)); 1056 } 1057 1058 static struct mbuf * 1059 rt_makeifannouncemsg(struct ifnet *ifp, int type, int what, 1060 struct rt_addrinfo *info) 1061 { 1062 struct if_announcemsghdr *ifan; 1063 struct mbuf *m; 1064 1065 if (route_cb.any_count == 0) 1066 return NULL; 1067 1068 bzero(info, sizeof(*info)); 1069 m = rt_msg_mbuf(type, info); 1070 if (m == NULL) 1071 return NULL; 1072 1073 ifan = mtod(m, struct if_announcemsghdr *); 1074 ifan->ifan_index = ifp->if_index; 1075 strlcpy(ifan->ifan_name, ifp->if_xname, sizeof ifan->ifan_name); 1076 ifan->ifan_what = what; 1077 return m; 1078 } 1079 1080 /* 1081 * This is called to generate routing socket messages indicating 1082 * IEEE80211 wireless events. 1083 * XXX we piggyback on the RTM_IFANNOUNCE msg format in a clumsy way. 1084 */ 1085 void 1086 rt_ieee80211msg(struct ifnet *ifp, int what, void *data, size_t data_len) 1087 { 1088 struct rt_addrinfo info; 1089 struct mbuf *m; 1090 1091 m = rt_makeifannouncemsg(ifp, RTM_IEEE80211, what, &info); 1092 if (m == NULL) 1093 return; 1094 1095 /* 1096 * Append the ieee80211 data. Try to stick it in the 1097 * mbuf containing the ifannounce msg; otherwise allocate 1098 * a new mbuf and append. 1099 * 1100 * NB: we assume m is a single mbuf. 1101 */ 1102 if (data_len > M_TRAILINGSPACE(m)) { 1103 struct mbuf *n = m_get(M_NOWAIT, MT_DATA); 1104 if (n == NULL) { 1105 m_freem(m); 1106 return; 1107 } 1108 bcopy(data, mtod(n, void *), data_len); 1109 n->m_len = data_len; 1110 m->m_next = n; 1111 } else if (data_len > 0) { 1112 bcopy(data, mtod(m, u_int8_t *) + m->m_len, data_len); 1113 m->m_len += data_len; 1114 } 1115 if (m->m_flags & M_PKTHDR) 1116 m->m_pkthdr.len += data_len; 1117 mtod(m, struct if_announcemsghdr *)->ifan_msglen += data_len; 1118 rts_input(m, 0); 1119 } 1120 1121 /* 1122 * This is called to generate routing socket messages indicating 1123 * network interface arrival and departure. 1124 */ 1125 void 1126 rt_ifannouncemsg(struct ifnet *ifp, int what) 1127 { 1128 struct rt_addrinfo addrinfo; 1129 struct mbuf *m; 1130 1131 m = rt_makeifannouncemsg(ifp, RTM_IFANNOUNCE, what, &addrinfo); 1132 if (m != NULL) 1133 rts_input(m, 0); 1134 } 1135 1136 static int 1137 resizewalkarg(struct walkarg *w, int len) 1138 { 1139 void *newptr; 1140 1141 newptr = malloc(len, M_RTABLE, M_INTWAIT | M_NULLOK); 1142 if (newptr == NULL) 1143 return (ENOMEM); 1144 if (w->w_tmem != NULL) 1145 free(w->w_tmem, M_RTABLE); 1146 w->w_tmem = newptr; 1147 w->w_tmemsize = len; 1148 return (0); 1149 } 1150 1151 /* 1152 * This is used in dumping the kernel table via sysctl(). 1153 */ 1154 int 1155 sysctl_dumpentry(struct radix_node *rn, void *vw) 1156 { 1157 struct walkarg *w = vw; 1158 struct rtentry *rt = (struct rtentry *)rn; 1159 struct rt_addrinfo rtinfo; 1160 int error, msglen; 1161 1162 if (w->w_op == NET_RT_FLAGS && !(rt->rt_flags & w->w_arg)) 1163 return 0; 1164 1165 bzero(&rtinfo, sizeof(struct rt_addrinfo)); 1166 rtinfo.rti_dst = rt_key(rt); 1167 rtinfo.rti_gateway = rt->rt_gateway; 1168 rtinfo.rti_netmask = rt_mask(rt); 1169 rtinfo.rti_genmask = rt->rt_genmask; 1170 if (rt->rt_ifp != NULL) { 1171 rtinfo.rti_ifpaddr = 1172 TAILQ_FIRST(&rt->rt_ifp->if_addrhead)->ifa_addr; 1173 rtinfo.rti_ifaaddr = rt->rt_ifa->ifa_addr; 1174 if (rt->rt_ifp->if_flags & IFF_POINTOPOINT) 1175 rtinfo.rti_bcastaddr = rt->rt_ifa->ifa_dstaddr; 1176 } 1177 msglen = rt_msgsize(RTM_GET, &rtinfo); 1178 if (w->w_tmemsize < msglen && resizewalkarg(w, msglen) != 0) 1179 return (ENOMEM); 1180 rt_msg_buffer(RTM_GET, &rtinfo, w->w_tmem, msglen); 1181 if (w->w_req != NULL) { 1182 struct rt_msghdr *rtm = w->w_tmem; 1183 1184 rtm->rtm_flags = rt->rt_flags; 1185 rtm->rtm_use = rt->rt_use; 1186 rtm->rtm_rmx = rt->rt_rmx; 1187 rtm->rtm_index = rt->rt_ifp->if_index; 1188 rtm->rtm_errno = rtm->rtm_pid = rtm->rtm_seq = 0; 1189 rtm->rtm_addrs = rtinfo.rti_addrs; 1190 error = SYSCTL_OUT(w->w_req, rtm, msglen); 1191 return (error); 1192 } 1193 return (0); 1194 } 1195 1196 static int 1197 sysctl_iflist(int af, struct walkarg *w) 1198 { 1199 struct ifnet *ifp; 1200 struct ifaddr *ifa; 1201 struct rt_addrinfo rtinfo; 1202 int msglen, error; 1203 1204 bzero(&rtinfo, sizeof(struct rt_addrinfo)); 1205 TAILQ_FOREACH(ifp, &ifnet, if_link) { 1206 if (w->w_arg && w->w_arg != ifp->if_index) 1207 continue; 1208 ifa = TAILQ_FIRST(&ifp->if_addrhead); 1209 rtinfo.rti_ifpaddr = ifa->ifa_addr; 1210 msglen = rt_msgsize(RTM_IFINFO, &rtinfo); 1211 if (w->w_tmemsize < msglen && resizewalkarg(w, msglen) != 0) 1212 return (ENOMEM); 1213 rt_msg_buffer(RTM_IFINFO, &rtinfo, w->w_tmem, msglen); 1214 rtinfo.rti_ifpaddr = NULL; 1215 if (w->w_req != NULL && w->w_tmem != NULL) { 1216 struct if_msghdr *ifm = w->w_tmem; 1217 1218 ifm->ifm_index = ifp->if_index; 1219 ifm->ifm_flags = ifp->if_flags; 1220 ifm->ifm_data = ifp->if_data; 1221 ifm->ifm_addrs = rtinfo.rti_addrs; 1222 error = SYSCTL_OUT(w->w_req, ifm, msglen); 1223 if (error) 1224 return (error); 1225 } 1226 while ((ifa = TAILQ_NEXT(ifa, ifa_link)) != NULL) { 1227 if (af && af != ifa->ifa_addr->sa_family) 1228 continue; 1229 if (curproc->p_ucred->cr_prison && 1230 prison_if(curproc->p_ucred, ifa->ifa_addr)) 1231 continue; 1232 rtinfo.rti_ifaaddr = ifa->ifa_addr; 1233 rtinfo.rti_netmask = ifa->ifa_netmask; 1234 rtinfo.rti_bcastaddr = ifa->ifa_dstaddr; 1235 msglen = rt_msgsize(RTM_NEWADDR, &rtinfo); 1236 if (w->w_tmemsize < msglen && 1237 resizewalkarg(w, msglen) != 0) 1238 return (ENOMEM); 1239 rt_msg_buffer(RTM_NEWADDR, &rtinfo, w->w_tmem, msglen); 1240 if (w->w_req != NULL) { 1241 struct ifa_msghdr *ifam = w->w_tmem; 1242 1243 ifam->ifam_index = ifa->ifa_ifp->if_index; 1244 ifam->ifam_flags = ifa->ifa_flags; 1245 ifam->ifam_metric = ifa->ifa_metric; 1246 ifam->ifam_addrs = rtinfo.rti_addrs; 1247 error = SYSCTL_OUT(w->w_req, w->w_tmem, msglen); 1248 if (error) 1249 return (error); 1250 } 1251 } 1252 rtinfo.rti_netmask = NULL; 1253 rtinfo.rti_ifaaddr = NULL; 1254 rtinfo.rti_bcastaddr = NULL; 1255 } 1256 return (0); 1257 } 1258 1259 static int 1260 sysctl_rtsock(SYSCTL_HANDLER_ARGS) 1261 { 1262 int *name = (int *)arg1; 1263 u_int namelen = arg2; 1264 struct radix_node_head *rnh; 1265 int i, error = EINVAL; 1266 int origcpu; 1267 u_char af; 1268 struct walkarg w; 1269 1270 name ++; 1271 namelen--; 1272 if (req->newptr) 1273 return (EPERM); 1274 if (namelen != 3 && namelen != 4) 1275 return (EINVAL); 1276 af = name[0]; 1277 bzero(&w, sizeof w); 1278 w.w_op = name[1]; 1279 w.w_arg = name[2]; 1280 w.w_req = req; 1281 1282 /* 1283 * Optional third argument specifies cpu, used primarily for 1284 * debugging the route table. 1285 */ 1286 if (namelen == 4) { 1287 if (name[3] < 0 || name[3] >= ncpus) 1288 return (EINVAL); 1289 origcpu = mycpuid; 1290 lwkt_migratecpu(name[3]); 1291 } else { 1292 origcpu = -1; 1293 } 1294 crit_enter(); 1295 switch (w.w_op) { 1296 case NET_RT_DUMP: 1297 case NET_RT_FLAGS: 1298 for (i = 1; i <= AF_MAX; i++) 1299 if ((rnh = rt_tables[mycpuid][i]) && 1300 (af == 0 || af == i) && 1301 (error = rnh->rnh_walktree(rnh, 1302 sysctl_dumpentry, &w))) 1303 break; 1304 break; 1305 1306 case NET_RT_IFLIST: 1307 error = sysctl_iflist(af, &w); 1308 } 1309 crit_exit(); 1310 if (w.w_tmem != NULL) 1311 free(w.w_tmem, M_RTABLE); 1312 if (origcpu >= 0) 1313 lwkt_migratecpu(origcpu); 1314 return (error); 1315 } 1316 1317 SYSCTL_NODE(_net, PF_ROUTE, routetable, CTLFLAG_RD, sysctl_rtsock, ""); 1318 1319 /* 1320 * Definitions of protocols supported in the ROUTE domain. 1321 */ 1322 1323 extern struct domain routedomain; /* or at least forward */ 1324 1325 static struct protosw routesw[] = { 1326 { SOCK_RAW, &routedomain, 0, PR_ATOMIC|PR_ADDR, 1327 0, route_output, raw_ctlinput, 0, 1328 cpu0_soport, 1329 raw_init, 0, 0, 0, 1330 &route_usrreqs 1331 } 1332 }; 1333 1334 static struct domain routedomain = { 1335 PF_ROUTE, "route", NULL, NULL, NULL, 1336 routesw, &routesw[(sizeof routesw)/(sizeof routesw[0])], 1337 }; 1338 1339 DOMAIN_SET(route); 1340