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) 1988, 1991, 1993 35 * The Regents of the University of California. All rights reserved. 36 * 37 * Redistribution and use in source and binary forms, with or without 38 * modification, are permitted provided that the following conditions 39 * are met: 40 * 1. Redistributions of source code must retain the above copyright 41 * notice, this list of conditions and the following disclaimer. 42 * 2. Redistributions in binary form must reproduce the above copyright 43 * notice, this list of conditions and the following disclaimer in the 44 * documentation and/or other materials provided with the distribution. 45 * 3. All advertising materials mentioning features or use of this software 46 * must display the following acknowledgement: 47 * This product includes software developed by the University of 48 * California, Berkeley and its contributors. 49 * 4. Neither the name of the University nor the names of its contributors 50 * may be used to endorse or promote products derived from this software 51 * without specific prior written permission. 52 * 53 * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND 54 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE 55 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE 56 * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE 57 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL 58 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS 59 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) 60 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT 61 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY 62 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF 63 * SUCH DAMAGE. 64 * 65 * @(#)rtsock.c 8.7 (Berkeley) 10/12/95 66 * $FreeBSD: src/sys/net/rtsock.c,v 1.44.2.11 2002/12/04 14:05:41 ru Exp $ 67 * $DragonFly: src/sys/net/rtsock.c,v 1.45 2008/10/27 02:56:30 sephe Exp $ 68 */ 69 70 #include "opt_sctp.h" 71 72 #include <sys/param.h> 73 #include <sys/systm.h> 74 #include <sys/kernel.h> 75 #include <sys/sysctl.h> 76 #include <sys/proc.h> 77 #include <sys/priv.h> 78 #include <sys/malloc.h> 79 #include <sys/mbuf.h> 80 #include <sys/protosw.h> 81 #include <sys/socket.h> 82 #include <sys/socketvar.h> 83 #include <sys/domain.h> 84 #include <sys/thread2.h> 85 86 #include <net/if.h> 87 #include <net/route.h> 88 #include <net/raw_cb.h> 89 #include <net/netmsg2.h> 90 91 #ifdef SCTP 92 extern void sctp_add_ip_address(struct ifaddr *ifa); 93 extern void sctp_delete_ip_address(struct ifaddr *ifa); 94 #endif /* SCTP */ 95 96 MALLOC_DEFINE(M_RTABLE, "routetbl", "routing tables"); 97 98 static struct route_cb { 99 int ip_count; 100 int ip6_count; 101 int ipx_count; 102 int ns_count; 103 int any_count; 104 } route_cb; 105 106 static const struct sockaddr route_src = { 2, PF_ROUTE, }; 107 108 struct walkarg { 109 int w_tmemsize; 110 int w_op, w_arg; 111 void *w_tmem; 112 struct sysctl_req *w_req; 113 }; 114 115 static struct mbuf * 116 rt_msg_mbuf (int, struct rt_addrinfo *); 117 static void rt_msg_buffer (int, struct rt_addrinfo *, void *buf, int len); 118 static int rt_msgsize (int type, struct rt_addrinfo *rtinfo); 119 static int rt_xaddrs (char *, char *, struct rt_addrinfo *); 120 static int sysctl_dumpentry (struct radix_node *rn, void *vw); 121 static int sysctl_iflist (int af, struct walkarg *w); 122 static int route_output(struct mbuf *, struct socket *, ...); 123 static void rt_setmetrics (u_long, struct rt_metrics *, 124 struct rt_metrics *); 125 126 /* 127 * It really doesn't make any sense at all for this code to share much 128 * with raw_usrreq.c, since its functionality is so restricted. XXX 129 */ 130 static int 131 rts_abort(struct socket *so) 132 { 133 int error; 134 135 crit_enter(); 136 error = raw_usrreqs.pru_abort(so); 137 crit_exit(); 138 return error; 139 } 140 141 /* pru_accept is EOPNOTSUPP */ 142 143 static int 144 rts_attach(struct socket *so, int proto, struct pru_attach_info *ai) 145 { 146 struct rawcb *rp; 147 int error; 148 149 if (sotorawcb(so) != NULL) 150 return EISCONN; /* XXX panic? */ 151 152 rp = kmalloc(sizeof *rp, M_PCB, M_WAITOK | M_ZERO); 153 154 /* 155 * The critical section is necessary to block protocols from sending 156 * error notifications (like RTM_REDIRECT or RTM_LOSING) while 157 * this PCB is extant but incompletely initialized. 158 * Probably we should try to do more of this work beforehand and 159 * eliminate the critical section. 160 */ 161 crit_enter(); 162 so->so_pcb = rp; 163 error = raw_attach(so, proto, ai->sb_rlimit); 164 rp = sotorawcb(so); 165 if (error) { 166 crit_exit(); 167 kfree(rp, M_PCB); 168 return error; 169 } 170 switch(rp->rcb_proto.sp_protocol) { 171 case AF_INET: 172 route_cb.ip_count++; 173 break; 174 case AF_INET6: 175 route_cb.ip6_count++; 176 break; 177 case AF_IPX: 178 route_cb.ipx_count++; 179 break; 180 case AF_NS: 181 route_cb.ns_count++; 182 break; 183 } 184 rp->rcb_faddr = &route_src; 185 route_cb.any_count++; 186 soisconnected(so); 187 so->so_options |= SO_USELOOPBACK; 188 crit_exit(); 189 return 0; 190 } 191 192 static int 193 rts_bind(struct socket *so, struct sockaddr *nam, struct thread *td) 194 { 195 int error; 196 197 crit_enter(); 198 error = raw_usrreqs.pru_bind(so, nam, td); /* xxx just EINVAL */ 199 crit_exit(); 200 return error; 201 } 202 203 static int 204 rts_connect(struct socket *so, struct sockaddr *nam, struct thread *td) 205 { 206 int error; 207 208 crit_enter(); 209 error = raw_usrreqs.pru_connect(so, nam, td); /* XXX just EINVAL */ 210 crit_exit(); 211 return error; 212 } 213 214 /* pru_connect2 is EOPNOTSUPP */ 215 /* pru_control is EOPNOTSUPP */ 216 217 static int 218 rts_detach(struct socket *so) 219 { 220 struct rawcb *rp = sotorawcb(so); 221 int error; 222 223 crit_enter(); 224 if (rp != NULL) { 225 switch(rp->rcb_proto.sp_protocol) { 226 case AF_INET: 227 route_cb.ip_count--; 228 break; 229 case AF_INET6: 230 route_cb.ip6_count--; 231 break; 232 case AF_IPX: 233 route_cb.ipx_count--; 234 break; 235 case AF_NS: 236 route_cb.ns_count--; 237 break; 238 } 239 route_cb.any_count--; 240 } 241 error = raw_usrreqs.pru_detach(so); 242 crit_exit(); 243 return error; 244 } 245 246 static int 247 rts_disconnect(struct socket *so) 248 { 249 int error; 250 251 crit_enter(); 252 error = raw_usrreqs.pru_disconnect(so); 253 crit_exit(); 254 return error; 255 } 256 257 /* pru_listen is EOPNOTSUPP */ 258 259 static int 260 rts_peeraddr(struct socket *so, struct sockaddr **nam) 261 { 262 int error; 263 264 crit_enter(); 265 error = raw_usrreqs.pru_peeraddr(so, nam); 266 crit_exit(); 267 return error; 268 } 269 270 /* pru_rcvd is EOPNOTSUPP */ 271 /* pru_rcvoob is EOPNOTSUPP */ 272 273 static int 274 rts_send(struct socket *so, int flags, struct mbuf *m, struct sockaddr *nam, 275 struct mbuf *control, struct thread *td) 276 { 277 int error; 278 279 crit_enter(); 280 error = raw_usrreqs.pru_send(so, flags, m, nam, control, td); 281 crit_exit(); 282 return error; 283 } 284 285 /* pru_sense is null */ 286 287 static int 288 rts_shutdown(struct socket *so) 289 { 290 int error; 291 292 crit_enter(); 293 error = raw_usrreqs.pru_shutdown(so); 294 crit_exit(); 295 return error; 296 } 297 298 static int 299 rts_sockaddr(struct socket *so, struct sockaddr **nam) 300 { 301 int error; 302 303 crit_enter(); 304 error = raw_usrreqs.pru_sockaddr(so, nam); 305 crit_exit(); 306 return error; 307 } 308 309 static struct pr_usrreqs route_usrreqs = { 310 .pru_abort = rts_abort, 311 .pru_accept = pru_accept_notsupp, 312 .pru_attach = rts_attach, 313 .pru_bind = rts_bind, 314 .pru_connect = rts_connect, 315 .pru_connect2 = pru_connect2_notsupp, 316 .pru_control = pru_control_notsupp, 317 .pru_detach = rts_detach, 318 .pru_disconnect = rts_disconnect, 319 .pru_listen = pru_listen_notsupp, 320 .pru_peeraddr = rts_peeraddr, 321 .pru_rcvd = pru_rcvd_notsupp, 322 .pru_rcvoob = pru_rcvoob_notsupp, 323 .pru_send = rts_send, 324 .pru_sense = pru_sense_null, 325 .pru_shutdown = rts_shutdown, 326 .pru_sockaddr = rts_sockaddr, 327 .pru_sosend = sosend, 328 .pru_soreceive = soreceive, 329 .pru_sopoll = sopoll 330 }; 331 332 static __inline sa_family_t 333 familyof(struct sockaddr *sa) 334 { 335 return (sa != NULL ? sa->sa_family : 0); 336 } 337 338 /* 339 * Routing socket input function. The packet must be serialized onto cpu 0. 340 * We use the cpu0_soport() netisr processing loop to handle it. 341 * 342 * This looks messy but it means that anyone, including interrupt code, 343 * can send a message to the routing socket. 344 */ 345 static void 346 rts_input_handler(struct netmsg *msg) 347 { 348 static const struct sockaddr route_dst = { 2, PF_ROUTE, }; 349 struct sockproto route_proto; 350 struct netmsg_packet *pmsg; 351 struct mbuf *m; 352 sa_family_t family; 353 struct rawcb *skip; 354 355 pmsg = (void *)msg; 356 family = pmsg->nm_netmsg.nm_lmsg.u.ms_result; 357 route_proto.sp_family = PF_ROUTE; 358 route_proto.sp_protocol = family; 359 360 m = pmsg->nm_packet; 361 M_ASSERTPKTHDR(m); 362 363 skip = m->m_pkthdr.header; 364 m->m_pkthdr.header = NULL; 365 366 raw_input(m, &route_proto, &route_src, &route_dst, skip); 367 } 368 369 static void 370 rts_input_skip(struct mbuf *m, sa_family_t family, struct rawcb *skip) 371 { 372 struct netmsg_packet *pmsg; 373 lwkt_port_t port; 374 375 M_ASSERTPKTHDR(m); 376 377 port = cpu0_soport(NULL, NULL, NULL, 0); 378 pmsg = &m->m_hdr.mh_netmsg; 379 netmsg_init(&pmsg->nm_netmsg, &netisr_apanic_rport, 380 0, rts_input_handler); 381 pmsg->nm_packet = m; 382 pmsg->nm_netmsg.nm_lmsg.u.ms_result = family; 383 m->m_pkthdr.header = skip; /* XXX steal field in pkthdr */ 384 lwkt_sendmsg(port, &pmsg->nm_netmsg.nm_lmsg); 385 } 386 387 static __inline void 388 rts_input(struct mbuf *m, sa_family_t family) 389 { 390 rts_input_skip(m, family, NULL); 391 } 392 393 static void * 394 reallocbuf_nofree(void *ptr, size_t len, size_t olen) 395 { 396 void *newptr; 397 398 newptr = kmalloc(len, M_RTABLE, M_INTWAIT | M_NULLOK); 399 if (newptr == NULL) 400 return NULL; 401 bcopy(ptr, newptr, olen); 402 return (newptr); 403 } 404 405 /* 406 * Internal helper routine for route_output(). 407 */ 408 static int 409 _fillrtmsg(struct rt_msghdr **prtm, struct rtentry *rt, 410 struct rt_addrinfo *rtinfo) 411 { 412 int msglen; 413 struct rt_msghdr *rtm = *prtm; 414 415 /* Fill in rt_addrinfo for call to rt_msg_buffer(). */ 416 rtinfo->rti_dst = rt_key(rt); 417 rtinfo->rti_gateway = rt->rt_gateway; 418 rtinfo->rti_netmask = rt_mask(rt); /* might be NULL */ 419 rtinfo->rti_genmask = rt->rt_genmask; /* might be NULL */ 420 if (rtm->rtm_addrs & (RTA_IFP | RTA_IFA)) { 421 if (rt->rt_ifp != NULL) { 422 rtinfo->rti_ifpaddr = 423 TAILQ_FIRST(&rt->rt_ifp->if_addrheads[mycpuid]) 424 ->ifa->ifa_addr; 425 rtinfo->rti_ifaaddr = rt->rt_ifa->ifa_addr; 426 if (rt->rt_ifp->if_flags & IFF_POINTOPOINT) 427 rtinfo->rti_bcastaddr = rt->rt_ifa->ifa_dstaddr; 428 rtm->rtm_index = rt->rt_ifp->if_index; 429 } else { 430 rtinfo->rti_ifpaddr = NULL; 431 rtinfo->rti_ifaaddr = NULL; 432 } 433 } else if (rt->rt_ifp != NULL) { 434 rtm->rtm_index = rt->rt_ifp->if_index; 435 } 436 437 msglen = rt_msgsize(rtm->rtm_type, rtinfo); 438 if (rtm->rtm_msglen < msglen) { 439 /* NOTE: Caller will free the old rtm accordingly */ 440 rtm = reallocbuf_nofree(rtm, msglen, rtm->rtm_msglen); 441 if (rtm == NULL) 442 return (ENOBUFS); 443 *prtm = rtm; 444 } 445 rt_msg_buffer(rtm->rtm_type, rtinfo, rtm, msglen); 446 447 rtm->rtm_flags = rt->rt_flags; 448 rtm->rtm_rmx = rt->rt_rmx; 449 rtm->rtm_addrs = rtinfo->rti_addrs; 450 451 return (0); 452 } 453 454 struct rtm_arg { 455 struct rt_msghdr *bak_rtm; 456 struct rt_msghdr *new_rtm; 457 }; 458 459 static int 460 fillrtmsg(struct rtm_arg *arg, struct rtentry *rt, 461 struct rt_addrinfo *rtinfo) 462 { 463 struct rt_msghdr *rtm = arg->new_rtm; 464 int error; 465 466 error = _fillrtmsg(&rtm, rt, rtinfo); 467 if (!error) { 468 if (arg->new_rtm != rtm) { 469 /* 470 * _fillrtmsg() just allocated a new rtm; 471 * if the previously allocated rtm is not 472 * the backing rtm, it should be freed. 473 */ 474 if (arg->new_rtm != arg->bak_rtm) 475 kfree(arg->new_rtm, M_RTABLE); 476 arg->new_rtm = rtm; 477 } 478 } 479 return error; 480 } 481 482 static void route_output_add_callback(int, int, struct rt_addrinfo *, 483 struct rtentry *, void *); 484 static void route_output_delete_callback(int, int, struct rt_addrinfo *, 485 struct rtentry *, void *); 486 static int route_output_get_callback(int, struct rt_addrinfo *, 487 struct rtentry *, void *, int); 488 static int route_output_change_callback(int, struct rt_addrinfo *, 489 struct rtentry *, void *, int); 490 static int route_output_lock_callback(int, struct rt_addrinfo *, 491 struct rtentry *, void *, int); 492 493 /*ARGSUSED*/ 494 static int 495 route_output(struct mbuf *m, struct socket *so, ...) 496 { 497 struct rtm_arg arg; 498 struct rt_msghdr *rtm = NULL; 499 struct rawcb *rp = NULL; 500 struct pr_output_info *oi; 501 struct rt_addrinfo rtinfo; 502 sa_family_t family; 503 int len, error = 0; 504 __va_list ap; 505 506 M_ASSERTPKTHDR(m); 507 508 __va_start(ap, so); 509 oi = __va_arg(ap, struct pr_output_info *); 510 __va_end(ap); 511 512 family = familyof(NULL); 513 514 #define gotoerr(e) { error = e; goto flush;} 515 516 if (m == NULL || 517 (m->m_len < sizeof(long) && 518 (m = m_pullup(m, sizeof(long))) == NULL)) 519 return (ENOBUFS); 520 len = m->m_pkthdr.len; 521 if (len < sizeof(struct rt_msghdr) || 522 len != mtod(m, struct rt_msghdr *)->rtm_msglen) 523 gotoerr(EINVAL); 524 525 rtm = kmalloc(len, M_RTABLE, M_INTWAIT | M_NULLOK); 526 if (rtm == NULL) 527 gotoerr(ENOBUFS); 528 529 m_copydata(m, 0, len, (caddr_t)rtm); 530 if (rtm->rtm_version != RTM_VERSION) 531 gotoerr(EPROTONOSUPPORT); 532 533 rtm->rtm_pid = oi->p_pid; 534 bzero(&rtinfo, sizeof(struct rt_addrinfo)); 535 rtinfo.rti_addrs = rtm->rtm_addrs; 536 if (rt_xaddrs((char *)(rtm + 1), (char *)rtm + len, &rtinfo) != 0) 537 gotoerr(EINVAL); 538 539 rtinfo.rti_flags = rtm->rtm_flags; 540 if (rtinfo.rti_dst == NULL || rtinfo.rti_dst->sa_family >= AF_MAX || 541 (rtinfo.rti_gateway && rtinfo.rti_gateway->sa_family >= AF_MAX)) 542 gotoerr(EINVAL); 543 544 family = familyof(rtinfo.rti_dst); 545 546 if (rtinfo.rti_genmask != NULL) { 547 error = rtmask_add_global(rtinfo.rti_genmask); 548 if (error) 549 goto flush; 550 } 551 552 /* 553 * Verify that the caller has the appropriate privilege; RTM_GET 554 * is the only operation the non-superuser is allowed. 555 */ 556 if (rtm->rtm_type != RTM_GET && 557 priv_check_cred(so->so_cred, PRIV_ROOT, 0) != 0) 558 gotoerr(EPERM); 559 560 switch (rtm->rtm_type) { 561 case RTM_ADD: 562 if (rtinfo.rti_gateway == NULL) { 563 error = EINVAL; 564 } else { 565 error = rtrequest1_global(RTM_ADD, &rtinfo, 566 route_output_add_callback, rtm); 567 } 568 break; 569 case RTM_DELETE: 570 /* 571 * Backing rtm (bak_rtm) could _not_ be freed during 572 * rtrequest1_global or rtsearch_global, even if the 573 * callback reallocates the rtm due to its size changes, 574 * since rtinfo points to the backing rtm's memory area. 575 * After rtrequest1_global or rtsearch_global returns, 576 * it is safe to free the backing rtm, since rtinfo will 577 * not be used anymore. 578 * 579 * new_rtm will be used to save the new rtm allocated 580 * by rtrequest1_global or rtsearch_global. 581 */ 582 arg.bak_rtm = rtm; 583 arg.new_rtm = rtm; 584 error = rtrequest1_global(RTM_DELETE, &rtinfo, 585 route_output_delete_callback, &arg); 586 rtm = arg.new_rtm; 587 if (rtm != arg.bak_rtm) 588 kfree(arg.bak_rtm, M_RTABLE); 589 break; 590 case RTM_GET: 591 /* See the comment in RTM_DELETE */ 592 arg.bak_rtm = rtm; 593 arg.new_rtm = rtm; 594 error = rtsearch_global(RTM_GET, &rtinfo, 595 route_output_get_callback, &arg, 596 RTS_NOEXACTMATCH); 597 rtm = arg.new_rtm; 598 if (rtm != arg.bak_rtm) 599 kfree(arg.bak_rtm, M_RTABLE); 600 break; 601 case RTM_CHANGE: 602 error = rtsearch_global(RTM_CHANGE, &rtinfo, 603 route_output_change_callback, rtm, 604 RTS_EXACTMATCH); 605 break; 606 case RTM_LOCK: 607 error = rtsearch_global(RTM_LOCK, &rtinfo, 608 route_output_lock_callback, rtm, 609 RTS_EXACTMATCH); 610 break; 611 default: 612 error = EOPNOTSUPP; 613 break; 614 } 615 flush: 616 if (rtm != NULL) { 617 if (error != 0) 618 rtm->rtm_errno = error; 619 else 620 rtm->rtm_flags |= RTF_DONE; 621 } 622 623 /* 624 * Check to see if we don't want our own messages. 625 */ 626 if (!(so->so_options & SO_USELOOPBACK)) { 627 if (route_cb.any_count <= 1) { 628 if (rtm != NULL) 629 kfree(rtm, M_RTABLE); 630 m_freem(m); 631 return (error); 632 } 633 /* There is another listener, so construct message */ 634 rp = sotorawcb(so); 635 } 636 if (rtm != NULL) { 637 m_copyback(m, 0, rtm->rtm_msglen, (caddr_t)rtm); 638 if (m->m_pkthdr.len < rtm->rtm_msglen) { 639 m_freem(m); 640 m = NULL; 641 } else if (m->m_pkthdr.len > rtm->rtm_msglen) 642 m_adj(m, rtm->rtm_msglen - m->m_pkthdr.len); 643 kfree(rtm, M_RTABLE); 644 } 645 if (m != NULL) 646 rts_input_skip(m, family, rp); 647 return (error); 648 } 649 650 static void 651 route_output_add_callback(int cmd, int error, struct rt_addrinfo *rtinfo, 652 struct rtentry *rt, void *arg) 653 { 654 struct rt_msghdr *rtm = arg; 655 656 if (error == 0 && rt != NULL) { 657 rt_setmetrics(rtm->rtm_inits, &rtm->rtm_rmx, 658 &rt->rt_rmx); 659 rt->rt_rmx.rmx_locks &= ~(rtm->rtm_inits); 660 rt->rt_rmx.rmx_locks |= 661 (rtm->rtm_inits & rtm->rtm_rmx.rmx_locks); 662 if (rtinfo->rti_genmask != NULL) { 663 rt->rt_genmask = rtmask_purelookup(rtinfo->rti_genmask); 664 if (rt->rt_genmask == NULL) { 665 /* 666 * This should not happen, since we 667 * have already installed genmask 668 * on each CPU before we reach here. 669 */ 670 panic("genmask is gone!?"); 671 } 672 } else { 673 rt->rt_genmask = NULL; 674 } 675 rtm->rtm_index = rt->rt_ifp->if_index; 676 } 677 } 678 679 static void 680 route_output_delete_callback(int cmd, int error, struct rt_addrinfo *rtinfo, 681 struct rtentry *rt, void *arg) 682 { 683 if (error == 0 && rt) { 684 ++rt->rt_refcnt; 685 if (fillrtmsg(arg, rt, rtinfo) != 0) { 686 error = ENOBUFS; 687 /* XXX no way to return the error */ 688 } 689 --rt->rt_refcnt; 690 } 691 if (rt && rt->rt_refcnt == 0) { 692 ++rt->rt_refcnt; 693 rtfree(rt); 694 } 695 } 696 697 static int 698 route_output_get_callback(int cmd, struct rt_addrinfo *rtinfo, 699 struct rtentry *rt, void *arg, int found_cnt) 700 { 701 int error, found = 0; 702 703 if (((rtinfo->rti_flags ^ rt->rt_flags) & RTF_HOST) == 0) 704 found = 1; 705 706 error = fillrtmsg(arg, rt, rtinfo); 707 if (!error && found) { 708 /* Got the exact match, we could return now! */ 709 error = EJUSTRETURN; 710 } 711 return error; 712 } 713 714 static int 715 route_output_change_callback(int cmd, struct rt_addrinfo *rtinfo, 716 struct rtentry *rt, void *arg, int found_cnt) 717 { 718 struct rt_msghdr *rtm = arg; 719 struct ifaddr *ifa; 720 int error = 0; 721 722 /* 723 * new gateway could require new ifaddr, ifp; 724 * flags may also be different; ifp may be specified 725 * by ll sockaddr when protocol address is ambiguous 726 */ 727 if (((rt->rt_flags & RTF_GATEWAY) && rtinfo->rti_gateway != NULL) || 728 rtinfo->rti_ifpaddr != NULL || 729 (rtinfo->rti_ifaaddr != NULL && 730 !sa_equal(rtinfo->rti_ifaaddr, rt->rt_ifa->ifa_addr))) { 731 error = rt_getifa(rtinfo); 732 if (error != 0) 733 goto done; 734 } 735 if (rtinfo->rti_gateway != NULL) { 736 /* 737 * We only need to generate rtmsg upon the 738 * first route to be changed. 739 */ 740 error = rt_setgate(rt, rt_key(rt), rtinfo->rti_gateway, 741 found_cnt == 1 ? RTL_REPORTMSG : RTL_DONTREPORT); 742 if (error != 0) 743 goto done; 744 } 745 if ((ifa = rtinfo->rti_ifa) != NULL) { 746 struct ifaddr *oifa = rt->rt_ifa; 747 748 if (oifa != ifa) { 749 if (oifa && oifa->ifa_rtrequest) 750 oifa->ifa_rtrequest(RTM_DELETE, rt, rtinfo); 751 IFAFREE(rt->rt_ifa); 752 IFAREF(ifa); 753 rt->rt_ifa = ifa; 754 rt->rt_ifp = rtinfo->rti_ifp; 755 } 756 } 757 rt_setmetrics(rtm->rtm_inits, &rtm->rtm_rmx, &rt->rt_rmx); 758 if (rt->rt_ifa && rt->rt_ifa->ifa_rtrequest) 759 rt->rt_ifa->ifa_rtrequest(RTM_ADD, rt, rtinfo); 760 if (rtinfo->rti_genmask != NULL) { 761 rt->rt_genmask = rtmask_purelookup(rtinfo->rti_genmask); 762 if (rt->rt_genmask == NULL) { 763 /* 764 * This should not happen, since we 765 * have already installed genmask 766 * on each CPU before we reach here. 767 */ 768 panic("genmask is gone!?\n"); 769 } 770 } 771 rtm->rtm_index = rt->rt_ifp->if_index; 772 done: 773 return error; 774 } 775 776 static int 777 route_output_lock_callback(int cmd, struct rt_addrinfo *rtinfo, 778 struct rtentry *rt, void *arg, 779 int found_cnt __unused) 780 { 781 struct rt_msghdr *rtm = arg; 782 783 rt->rt_rmx.rmx_locks &= ~(rtm->rtm_inits); 784 rt->rt_rmx.rmx_locks |= 785 (rtm->rtm_inits & rtm->rtm_rmx.rmx_locks); 786 return 0; 787 } 788 789 static void 790 rt_setmetrics(u_long which, struct rt_metrics *in, struct rt_metrics *out) 791 { 792 #define setmetric(flag, elt) if (which & (flag)) out->elt = in->elt; 793 setmetric(RTV_RPIPE, rmx_recvpipe); 794 setmetric(RTV_SPIPE, rmx_sendpipe); 795 setmetric(RTV_SSTHRESH, rmx_ssthresh); 796 setmetric(RTV_RTT, rmx_rtt); 797 setmetric(RTV_RTTVAR, rmx_rttvar); 798 setmetric(RTV_HOPCOUNT, rmx_hopcount); 799 setmetric(RTV_MTU, rmx_mtu); 800 setmetric(RTV_EXPIRE, rmx_expire); 801 #undef setmetric 802 } 803 804 #define ROUNDUP(a) \ 805 ((a) > 0 ? (1 + (((a) - 1) | (sizeof(long) - 1))) : sizeof(long)) 806 807 /* 808 * Extract the addresses of the passed sockaddrs. 809 * Do a little sanity checking so as to avoid bad memory references. 810 * This data is derived straight from userland. 811 */ 812 static int 813 rt_xaddrs(char *cp, char *cplim, struct rt_addrinfo *rtinfo) 814 { 815 struct sockaddr *sa; 816 int i; 817 818 for (i = 0; (i < RTAX_MAX) && (cp < cplim); i++) { 819 if ((rtinfo->rti_addrs & (1 << i)) == 0) 820 continue; 821 sa = (struct sockaddr *)cp; 822 /* 823 * It won't fit. 824 */ 825 if ((cp + sa->sa_len) > cplim) { 826 return (EINVAL); 827 } 828 829 /* 830 * There are no more... Quit now. 831 * If there are more bits, they are in error. 832 * I've seen this. route(1) can evidently generate these. 833 * This causes kernel to core dump. 834 * For compatibility, if we see this, point to a safe address. 835 */ 836 if (sa->sa_len == 0) { 837 static struct sockaddr sa_zero = { 838 sizeof sa_zero, AF_INET, 839 }; 840 841 rtinfo->rti_info[i] = &sa_zero; 842 kprintf("rtsock: received more addr bits than sockaddrs.\n"); 843 return (0); /* should be EINVAL but for compat */ 844 } 845 846 /* Accept the sockaddr. */ 847 rtinfo->rti_info[i] = sa; 848 cp += ROUNDUP(sa->sa_len); 849 } 850 return (0); 851 } 852 853 static int 854 rt_msghdrsize(int type) 855 { 856 switch (type) { 857 case RTM_DELADDR: 858 case RTM_NEWADDR: 859 return sizeof(struct ifa_msghdr); 860 case RTM_DELMADDR: 861 case RTM_NEWMADDR: 862 return sizeof(struct ifma_msghdr); 863 case RTM_IFINFO: 864 return sizeof(struct if_msghdr); 865 case RTM_IFANNOUNCE: 866 case RTM_IEEE80211: 867 return sizeof(struct if_announcemsghdr); 868 default: 869 return sizeof(struct rt_msghdr); 870 } 871 } 872 873 static int 874 rt_msgsize(int type, struct rt_addrinfo *rtinfo) 875 { 876 int len, i; 877 878 len = rt_msghdrsize(type); 879 for (i = 0; i < RTAX_MAX; i++) { 880 if (rtinfo->rti_info[i] != NULL) 881 len += ROUNDUP(rtinfo->rti_info[i]->sa_len); 882 } 883 len = ALIGN(len); 884 return len; 885 } 886 887 /* 888 * Build a routing message in a buffer. 889 * Copy the addresses in the rtinfo->rti_info[] sockaddr array 890 * to the end of the buffer after the message header. 891 * 892 * Set the rtinfo->rti_addrs bitmask of addresses present in rtinfo->rti_info[]. 893 * This side-effect can be avoided if we reorder the addrs bitmask field in all 894 * the route messages to line up so we can set it here instead of back in the 895 * calling routine. 896 */ 897 static void 898 rt_msg_buffer(int type, struct rt_addrinfo *rtinfo, void *buf, int msglen) 899 { 900 struct rt_msghdr *rtm; 901 char *cp; 902 int dlen, i; 903 904 rtm = (struct rt_msghdr *) buf; 905 rtm->rtm_version = RTM_VERSION; 906 rtm->rtm_type = type; 907 rtm->rtm_msglen = msglen; 908 909 cp = (char *)buf + rt_msghdrsize(type); 910 rtinfo->rti_addrs = 0; 911 for (i = 0; i < RTAX_MAX; i++) { 912 struct sockaddr *sa; 913 914 if ((sa = rtinfo->rti_info[i]) == NULL) 915 continue; 916 rtinfo->rti_addrs |= (1 << i); 917 dlen = ROUNDUP(sa->sa_len); 918 bcopy(sa, cp, dlen); 919 cp += dlen; 920 } 921 } 922 923 /* 924 * Build a routing message in a mbuf chain. 925 * Copy the addresses in the rtinfo->rti_info[] sockaddr array 926 * to the end of the mbuf after the message header. 927 * 928 * Set the rtinfo->rti_addrs bitmask of addresses present in rtinfo->rti_info[]. 929 * This side-effect can be avoided if we reorder the addrs bitmask field in all 930 * the route messages to line up so we can set it here instead of back in the 931 * calling routine. 932 */ 933 static struct mbuf * 934 rt_msg_mbuf(int type, struct rt_addrinfo *rtinfo) 935 { 936 struct mbuf *m; 937 struct rt_msghdr *rtm; 938 int hlen, len; 939 int i; 940 941 hlen = rt_msghdrsize(type); 942 KASSERT(hlen <= MCLBYTES, ("rt_msg_mbuf: hlen %d doesn't fit", hlen)); 943 944 m = m_getl(hlen, MB_DONTWAIT, MT_DATA, M_PKTHDR, NULL); 945 if (m == NULL) 946 return (NULL); 947 mbuftrackid(m, 32); 948 m->m_pkthdr.len = m->m_len = hlen; 949 m->m_pkthdr.rcvif = NULL; 950 rtinfo->rti_addrs = 0; 951 len = hlen; 952 for (i = 0; i < RTAX_MAX; i++) { 953 struct sockaddr *sa; 954 int dlen; 955 956 if ((sa = rtinfo->rti_info[i]) == NULL) 957 continue; 958 rtinfo->rti_addrs |= (1 << i); 959 dlen = ROUNDUP(sa->sa_len); 960 m_copyback(m, len, dlen, (caddr_t)sa); /* can grow mbuf chain */ 961 len += dlen; 962 } 963 if (m->m_pkthdr.len != len) { /* one of the m_copyback() calls failed */ 964 m_freem(m); 965 return (NULL); 966 } 967 rtm = mtod(m, struct rt_msghdr *); 968 bzero(rtm, hlen); 969 rtm->rtm_msglen = len; 970 rtm->rtm_version = RTM_VERSION; 971 rtm->rtm_type = type; 972 return (m); 973 } 974 975 /* 976 * This routine is called to generate a message from the routing 977 * socket indicating that a redirect has occurred, a routing lookup 978 * has failed, or that a protocol has detected timeouts to a particular 979 * destination. 980 */ 981 void 982 rt_missmsg(int type, struct rt_addrinfo *rtinfo, int flags, int error) 983 { 984 struct sockaddr *dst = rtinfo->rti_info[RTAX_DST]; 985 struct rt_msghdr *rtm; 986 struct mbuf *m; 987 988 if (route_cb.any_count == 0) 989 return; 990 m = rt_msg_mbuf(type, rtinfo); 991 if (m == NULL) 992 return; 993 rtm = mtod(m, struct rt_msghdr *); 994 rtm->rtm_flags = RTF_DONE | flags; 995 rtm->rtm_errno = error; 996 rtm->rtm_addrs = rtinfo->rti_addrs; 997 rts_input(m, familyof(dst)); 998 } 999 1000 void 1001 rt_dstmsg(int type, struct sockaddr *dst, int error) 1002 { 1003 struct rt_msghdr *rtm; 1004 struct rt_addrinfo addrs; 1005 struct mbuf *m; 1006 1007 if (route_cb.any_count == 0) 1008 return; 1009 bzero(&addrs, sizeof(struct rt_addrinfo)); 1010 addrs.rti_info[RTAX_DST] = dst; 1011 m = rt_msg_mbuf(type, &addrs); 1012 if (m == NULL) 1013 return; 1014 rtm = mtod(m, struct rt_msghdr *); 1015 rtm->rtm_flags = RTF_DONE; 1016 rtm->rtm_errno = error; 1017 rtm->rtm_addrs = addrs.rti_addrs; 1018 rts_input(m, familyof(dst)); 1019 } 1020 1021 /* 1022 * This routine is called to generate a message from the routing 1023 * socket indicating that the status of a network interface has changed. 1024 */ 1025 void 1026 rt_ifmsg(struct ifnet *ifp) 1027 { 1028 struct if_msghdr *ifm; 1029 struct mbuf *m; 1030 struct rt_addrinfo rtinfo; 1031 1032 if (route_cb.any_count == 0) 1033 return; 1034 bzero(&rtinfo, sizeof(struct rt_addrinfo)); 1035 m = rt_msg_mbuf(RTM_IFINFO, &rtinfo); 1036 if (m == NULL) 1037 return; 1038 ifm = mtod(m, struct if_msghdr *); 1039 ifm->ifm_index = ifp->if_index; 1040 ifm->ifm_flags = ifp->if_flags; 1041 ifm->ifm_data = ifp->if_data; 1042 ifm->ifm_addrs = 0; 1043 rts_input(m, 0); 1044 } 1045 1046 static void 1047 rt_ifamsg(int cmd, struct ifaddr *ifa) 1048 { 1049 struct ifa_msghdr *ifam; 1050 struct rt_addrinfo rtinfo; 1051 struct mbuf *m; 1052 struct ifnet *ifp = ifa->ifa_ifp; 1053 1054 bzero(&rtinfo, sizeof(struct rt_addrinfo)); 1055 rtinfo.rti_ifaaddr = ifa->ifa_addr; 1056 rtinfo.rti_ifpaddr = 1057 TAILQ_FIRST(&ifp->if_addrheads[mycpuid])->ifa->ifa_addr; 1058 rtinfo.rti_netmask = ifa->ifa_netmask; 1059 rtinfo.rti_bcastaddr = ifa->ifa_dstaddr; 1060 1061 m = rt_msg_mbuf(cmd, &rtinfo); 1062 if (m == NULL) 1063 return; 1064 1065 ifam = mtod(m, struct ifa_msghdr *); 1066 ifam->ifam_index = ifp->if_index; 1067 ifam->ifam_metric = ifa->ifa_metric; 1068 ifam->ifam_flags = ifa->ifa_flags; 1069 ifam->ifam_addrs = rtinfo.rti_addrs; 1070 1071 rts_input(m, familyof(ifa->ifa_addr)); 1072 } 1073 1074 void 1075 rt_rtmsg(int cmd, struct rtentry *rt, struct ifnet *ifp, int error) 1076 { 1077 struct rt_msghdr *rtm; 1078 struct rt_addrinfo rtinfo; 1079 struct mbuf *m; 1080 struct sockaddr *dst; 1081 1082 if (rt == NULL) 1083 return; 1084 1085 bzero(&rtinfo, sizeof(struct rt_addrinfo)); 1086 rtinfo.rti_dst = dst = rt_key(rt); 1087 rtinfo.rti_gateway = rt->rt_gateway; 1088 rtinfo.rti_netmask = rt_mask(rt); 1089 if (ifp != NULL) { 1090 rtinfo.rti_ifpaddr = 1091 TAILQ_FIRST(&ifp->if_addrheads[mycpuid])->ifa->ifa_addr; 1092 } 1093 rtinfo.rti_ifaaddr = rt->rt_ifa->ifa_addr; 1094 1095 m = rt_msg_mbuf(cmd, &rtinfo); 1096 if (m == NULL) 1097 return; 1098 1099 rtm = mtod(m, struct rt_msghdr *); 1100 if (ifp != NULL) 1101 rtm->rtm_index = ifp->if_index; 1102 rtm->rtm_flags |= rt->rt_flags; 1103 rtm->rtm_errno = error; 1104 rtm->rtm_addrs = rtinfo.rti_addrs; 1105 1106 rts_input(m, familyof(dst)); 1107 } 1108 1109 /* 1110 * This is called to generate messages from the routing socket 1111 * indicating a network interface has had addresses associated with it. 1112 * if we ever reverse the logic and replace messages TO the routing 1113 * socket indicate a request to configure interfaces, then it will 1114 * be unnecessary as the routing socket will automatically generate 1115 * copies of it. 1116 */ 1117 void 1118 rt_newaddrmsg(int cmd, struct ifaddr *ifa, int error, struct rtentry *rt) 1119 { 1120 #ifdef SCTP 1121 /* 1122 * notify the SCTP stack 1123 * this will only get called when an address is added/deleted 1124 * XXX pass the ifaddr struct instead if ifa->ifa_addr... 1125 */ 1126 if (cmd == RTM_ADD) 1127 sctp_add_ip_address(ifa); 1128 else if (cmd == RTM_DELETE) 1129 sctp_delete_ip_address(ifa); 1130 #endif /* SCTP */ 1131 1132 if (route_cb.any_count == 0) 1133 return; 1134 1135 if (cmd == RTM_ADD) { 1136 rt_ifamsg(RTM_NEWADDR, ifa); 1137 rt_rtmsg(RTM_ADD, rt, ifa->ifa_ifp, error); 1138 } else { 1139 KASSERT((cmd == RTM_DELETE), ("unknown cmd %d", cmd)); 1140 rt_rtmsg(RTM_DELETE, rt, ifa->ifa_ifp, error); 1141 rt_ifamsg(RTM_DELADDR, ifa); 1142 } 1143 } 1144 1145 /* 1146 * This is the analogue to the rt_newaddrmsg which performs the same 1147 * function but for multicast group memberhips. This is easier since 1148 * there is no route state to worry about. 1149 */ 1150 void 1151 rt_newmaddrmsg(int cmd, struct ifmultiaddr *ifma) 1152 { 1153 struct rt_addrinfo rtinfo; 1154 struct mbuf *m = NULL; 1155 struct ifnet *ifp = ifma->ifma_ifp; 1156 struct ifma_msghdr *ifmam; 1157 1158 if (route_cb.any_count == 0) 1159 return; 1160 1161 bzero(&rtinfo, sizeof(struct rt_addrinfo)); 1162 rtinfo.rti_ifaaddr = ifma->ifma_addr; 1163 if (ifp != NULL && !TAILQ_EMPTY(&ifp->if_addrheads[mycpuid])) { 1164 rtinfo.rti_ifpaddr = 1165 TAILQ_FIRST(&ifp->if_addrheads[mycpuid])->ifa->ifa_addr; 1166 } 1167 /* 1168 * If a link-layer address is present, present it as a ``gateway'' 1169 * (similarly to how ARP entries, e.g., are presented). 1170 */ 1171 rtinfo.rti_gateway = ifma->ifma_lladdr; 1172 1173 m = rt_msg_mbuf(cmd, &rtinfo); 1174 if (m == NULL) 1175 return; 1176 1177 ifmam = mtod(m, struct ifma_msghdr *); 1178 ifmam->ifmam_index = ifp->if_index; 1179 ifmam->ifmam_addrs = rtinfo.rti_addrs; 1180 1181 rts_input(m, familyof(ifma->ifma_addr)); 1182 } 1183 1184 static struct mbuf * 1185 rt_makeifannouncemsg(struct ifnet *ifp, int type, int what, 1186 struct rt_addrinfo *info) 1187 { 1188 struct if_announcemsghdr *ifan; 1189 struct mbuf *m; 1190 1191 if (route_cb.any_count == 0) 1192 return NULL; 1193 1194 bzero(info, sizeof(*info)); 1195 m = rt_msg_mbuf(type, info); 1196 if (m == NULL) 1197 return NULL; 1198 1199 ifan = mtod(m, struct if_announcemsghdr *); 1200 ifan->ifan_index = ifp->if_index; 1201 strlcpy(ifan->ifan_name, ifp->if_xname, sizeof ifan->ifan_name); 1202 ifan->ifan_what = what; 1203 return m; 1204 } 1205 1206 /* 1207 * This is called to generate routing socket messages indicating 1208 * IEEE80211 wireless events. 1209 * XXX we piggyback on the RTM_IFANNOUNCE msg format in a clumsy way. 1210 */ 1211 void 1212 rt_ieee80211msg(struct ifnet *ifp, int what, void *data, size_t data_len) 1213 { 1214 struct rt_addrinfo info; 1215 struct mbuf *m; 1216 1217 m = rt_makeifannouncemsg(ifp, RTM_IEEE80211, what, &info); 1218 if (m == NULL) 1219 return; 1220 1221 /* 1222 * Append the ieee80211 data. Try to stick it in the 1223 * mbuf containing the ifannounce msg; otherwise allocate 1224 * a new mbuf and append. 1225 * 1226 * NB: we assume m is a single mbuf. 1227 */ 1228 if (data_len > M_TRAILINGSPACE(m)) { 1229 struct mbuf *n = m_get(MB_DONTWAIT, MT_DATA); 1230 if (n == NULL) { 1231 m_freem(m); 1232 return; 1233 } 1234 bcopy(data, mtod(n, void *), data_len); 1235 n->m_len = data_len; 1236 m->m_next = n; 1237 } else if (data_len > 0) { 1238 bcopy(data, mtod(m, u_int8_t *) + m->m_len, data_len); 1239 m->m_len += data_len; 1240 } 1241 mbuftrackid(m, 33); 1242 if (m->m_flags & M_PKTHDR) 1243 m->m_pkthdr.len += data_len; 1244 mtod(m, struct if_announcemsghdr *)->ifan_msglen += data_len; 1245 rts_input(m, 0); 1246 } 1247 1248 /* 1249 * This is called to generate routing socket messages indicating 1250 * network interface arrival and departure. 1251 */ 1252 void 1253 rt_ifannouncemsg(struct ifnet *ifp, int what) 1254 { 1255 struct rt_addrinfo addrinfo; 1256 struct mbuf *m; 1257 1258 m = rt_makeifannouncemsg(ifp, RTM_IFANNOUNCE, what, &addrinfo); 1259 if (m != NULL) 1260 rts_input(m, 0); 1261 } 1262 1263 static int 1264 resizewalkarg(struct walkarg *w, int len) 1265 { 1266 void *newptr; 1267 1268 newptr = kmalloc(len, M_RTABLE, M_INTWAIT | M_NULLOK); 1269 if (newptr == NULL) 1270 return (ENOMEM); 1271 if (w->w_tmem != NULL) 1272 kfree(w->w_tmem, M_RTABLE); 1273 w->w_tmem = newptr; 1274 w->w_tmemsize = len; 1275 return (0); 1276 } 1277 1278 /* 1279 * This is used in dumping the kernel table via sysctl(). 1280 */ 1281 int 1282 sysctl_dumpentry(struct radix_node *rn, void *vw) 1283 { 1284 struct walkarg *w = vw; 1285 struct rtentry *rt = (struct rtentry *)rn; 1286 struct rt_addrinfo rtinfo; 1287 int error, msglen; 1288 1289 if (w->w_op == NET_RT_FLAGS && !(rt->rt_flags & w->w_arg)) 1290 return 0; 1291 1292 bzero(&rtinfo, sizeof(struct rt_addrinfo)); 1293 rtinfo.rti_dst = rt_key(rt); 1294 rtinfo.rti_gateway = rt->rt_gateway; 1295 rtinfo.rti_netmask = rt_mask(rt); 1296 rtinfo.rti_genmask = rt->rt_genmask; 1297 if (rt->rt_ifp != NULL) { 1298 rtinfo.rti_ifpaddr = 1299 TAILQ_FIRST(&rt->rt_ifp->if_addrheads[mycpuid])->ifa->ifa_addr; 1300 rtinfo.rti_ifaaddr = rt->rt_ifa->ifa_addr; 1301 if (rt->rt_ifp->if_flags & IFF_POINTOPOINT) 1302 rtinfo.rti_bcastaddr = rt->rt_ifa->ifa_dstaddr; 1303 } 1304 msglen = rt_msgsize(RTM_GET, &rtinfo); 1305 if (w->w_tmemsize < msglen && resizewalkarg(w, msglen) != 0) 1306 return (ENOMEM); 1307 rt_msg_buffer(RTM_GET, &rtinfo, w->w_tmem, msglen); 1308 if (w->w_req != NULL) { 1309 struct rt_msghdr *rtm = w->w_tmem; 1310 1311 rtm->rtm_flags = rt->rt_flags; 1312 rtm->rtm_use = rt->rt_use; 1313 rtm->rtm_rmx = rt->rt_rmx; 1314 rtm->rtm_index = rt->rt_ifp->if_index; 1315 rtm->rtm_errno = rtm->rtm_pid = rtm->rtm_seq = 0; 1316 rtm->rtm_addrs = rtinfo.rti_addrs; 1317 error = SYSCTL_OUT(w->w_req, rtm, msglen); 1318 return (error); 1319 } 1320 return (0); 1321 } 1322 1323 static int 1324 sysctl_iflist(int af, struct walkarg *w) 1325 { 1326 struct ifnet *ifp; 1327 struct rt_addrinfo rtinfo; 1328 int msglen, error; 1329 1330 bzero(&rtinfo, sizeof(struct rt_addrinfo)); 1331 TAILQ_FOREACH(ifp, &ifnet, if_link) { 1332 struct ifaddr_container *ifac; 1333 struct ifaddr *ifa; 1334 1335 if (w->w_arg && w->w_arg != ifp->if_index) 1336 continue; 1337 ifac = TAILQ_FIRST(&ifp->if_addrheads[mycpuid]); 1338 ifa = ifac->ifa; 1339 rtinfo.rti_ifpaddr = ifa->ifa_addr; 1340 msglen = rt_msgsize(RTM_IFINFO, &rtinfo); 1341 if (w->w_tmemsize < msglen && resizewalkarg(w, msglen) != 0) 1342 return (ENOMEM); 1343 rt_msg_buffer(RTM_IFINFO, &rtinfo, w->w_tmem, msglen); 1344 rtinfo.rti_ifpaddr = NULL; 1345 if (w->w_req != NULL && w->w_tmem != NULL) { 1346 struct if_msghdr *ifm = w->w_tmem; 1347 1348 ifm->ifm_index = ifp->if_index; 1349 ifm->ifm_flags = ifp->if_flags; 1350 ifm->ifm_data = ifp->if_data; 1351 ifm->ifm_addrs = rtinfo.rti_addrs; 1352 error = SYSCTL_OUT(w->w_req, ifm, msglen); 1353 if (error) 1354 return (error); 1355 } 1356 while ((ifac = TAILQ_NEXT(ifac, ifa_link)) != NULL) { 1357 ifa = ifac->ifa; 1358 1359 if (af && af != ifa->ifa_addr->sa_family) 1360 continue; 1361 if (curproc->p_ucred->cr_prison && 1362 prison_if(curproc->p_ucred, ifa->ifa_addr)) 1363 continue; 1364 rtinfo.rti_ifaaddr = ifa->ifa_addr; 1365 rtinfo.rti_netmask = ifa->ifa_netmask; 1366 rtinfo.rti_bcastaddr = ifa->ifa_dstaddr; 1367 msglen = rt_msgsize(RTM_NEWADDR, &rtinfo); 1368 if (w->w_tmemsize < msglen && 1369 resizewalkarg(w, msglen) != 0) 1370 return (ENOMEM); 1371 rt_msg_buffer(RTM_NEWADDR, &rtinfo, w->w_tmem, msglen); 1372 if (w->w_req != NULL) { 1373 struct ifa_msghdr *ifam = w->w_tmem; 1374 1375 ifam->ifam_index = ifa->ifa_ifp->if_index; 1376 ifam->ifam_flags = ifa->ifa_flags; 1377 ifam->ifam_metric = ifa->ifa_metric; 1378 ifam->ifam_addrs = rtinfo.rti_addrs; 1379 error = SYSCTL_OUT(w->w_req, w->w_tmem, msglen); 1380 if (error) 1381 return (error); 1382 } 1383 } 1384 rtinfo.rti_netmask = NULL; 1385 rtinfo.rti_ifaaddr = NULL; 1386 rtinfo.rti_bcastaddr = NULL; 1387 } 1388 return (0); 1389 } 1390 1391 static int 1392 sysctl_rtsock(SYSCTL_HANDLER_ARGS) 1393 { 1394 int *name = (int *)arg1; 1395 u_int namelen = arg2; 1396 struct radix_node_head *rnh; 1397 int i, error = EINVAL; 1398 int origcpu; 1399 u_char af; 1400 struct walkarg w; 1401 1402 name ++; 1403 namelen--; 1404 if (req->newptr) 1405 return (EPERM); 1406 if (namelen != 3 && namelen != 4) 1407 return (EINVAL); 1408 af = name[0]; 1409 bzero(&w, sizeof w); 1410 w.w_op = name[1]; 1411 w.w_arg = name[2]; 1412 w.w_req = req; 1413 1414 /* 1415 * Optional third argument specifies cpu, used primarily for 1416 * debugging the route table. 1417 */ 1418 if (namelen == 4) { 1419 if (name[3] < 0 || name[3] >= ncpus) 1420 return (EINVAL); 1421 origcpu = mycpuid; 1422 lwkt_migratecpu(name[3]); 1423 } else { 1424 origcpu = -1; 1425 } 1426 crit_enter(); 1427 switch (w.w_op) { 1428 case NET_RT_DUMP: 1429 case NET_RT_FLAGS: 1430 for (i = 1; i <= AF_MAX; i++) 1431 if ((rnh = rt_tables[mycpuid][i]) && 1432 (af == 0 || af == i) && 1433 (error = rnh->rnh_walktree(rnh, 1434 sysctl_dumpentry, &w))) 1435 break; 1436 break; 1437 1438 case NET_RT_IFLIST: 1439 error = sysctl_iflist(af, &w); 1440 } 1441 crit_exit(); 1442 if (w.w_tmem != NULL) 1443 kfree(w.w_tmem, M_RTABLE); 1444 if (origcpu >= 0) 1445 lwkt_migratecpu(origcpu); 1446 return (error); 1447 } 1448 1449 SYSCTL_NODE(_net, PF_ROUTE, routetable, CTLFLAG_RD, sysctl_rtsock, ""); 1450 1451 /* 1452 * Definitions of protocols supported in the ROUTE domain. 1453 */ 1454 1455 static struct domain routedomain; /* or at least forward */ 1456 1457 static struct protosw routesw[] = { 1458 { SOCK_RAW, &routedomain, 0, PR_ATOMIC|PR_ADDR, 1459 0, route_output, raw_ctlinput, 0, 1460 cpu0_soport, cpu0_ctlport, 1461 raw_init, 0, 0, 0, 1462 &route_usrreqs 1463 } 1464 }; 1465 1466 static struct domain routedomain = { 1467 PF_ROUTE, "route", NULL, NULL, NULL, 1468 routesw, &routesw[(sizeof routesw)/(sizeof routesw[0])], 1469 }; 1470 1471 DOMAIN_SET(route); 1472 1473