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; 353 354 pmsg = (void *)msg; 355 m = pmsg->nm_packet; 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 raw_input(m, &route_proto, &route_src, &route_dst); 361 } 362 363 static void 364 rts_input(struct mbuf *m, sa_family_t family) 365 { 366 struct netmsg_packet *pmsg; 367 lwkt_port_t port; 368 369 port = cpu0_soport(NULL, NULL, NULL, 0); 370 pmsg = &m->m_hdr.mh_netmsg; 371 netmsg_init(&pmsg->nm_netmsg, &netisr_apanic_rport, 372 0, rts_input_handler); 373 pmsg->nm_packet = m; 374 pmsg->nm_netmsg.nm_lmsg.u.ms_result = family; 375 lwkt_sendmsg(port, &pmsg->nm_netmsg.nm_lmsg); 376 } 377 378 static void * 379 reallocbuf(void *ptr, size_t len, size_t olen) 380 { 381 void *newptr; 382 383 newptr = kmalloc(len, M_RTABLE, M_INTWAIT | M_NULLOK); 384 if (newptr == NULL) 385 return NULL; 386 bcopy(ptr, newptr, olen); 387 kfree(ptr, M_RTABLE); 388 return (newptr); 389 } 390 391 /* 392 * Internal helper routine for route_output(). 393 */ 394 static int 395 fillrtmsg(struct rt_msghdr **prtm, struct rtentry *rt, 396 struct rt_addrinfo *rtinfo) 397 { 398 int msglen; 399 struct rt_msghdr *rtm = *prtm; 400 401 /* Fill in rt_addrinfo for call to rt_msg_buffer(). */ 402 rtinfo->rti_dst = rt_key(rt); 403 rtinfo->rti_gateway = rt->rt_gateway; 404 rtinfo->rti_netmask = rt_mask(rt); /* might be NULL */ 405 rtinfo->rti_genmask = rt->rt_genmask; /* might be NULL */ 406 if (rtm->rtm_addrs & (RTA_IFP | RTA_IFA)) { 407 if (rt->rt_ifp != NULL) { 408 rtinfo->rti_ifpaddr = 409 TAILQ_FIRST(&rt->rt_ifp->if_addrheads[mycpuid]) 410 ->ifa->ifa_addr; 411 rtinfo->rti_ifaaddr = rt->rt_ifa->ifa_addr; 412 if (rt->rt_ifp->if_flags & IFF_POINTOPOINT) 413 rtinfo->rti_bcastaddr = rt->rt_ifa->ifa_dstaddr; 414 rtm->rtm_index = rt->rt_ifp->if_index; 415 } else { 416 rtinfo->rti_ifpaddr = NULL; 417 rtinfo->rti_ifaaddr = NULL; 418 } 419 } 420 421 msglen = rt_msgsize(rtm->rtm_type, rtinfo); 422 if (rtm->rtm_msglen < msglen) { 423 rtm = reallocbuf(rtm, msglen, rtm->rtm_msglen); 424 if (rtm == NULL) 425 return (ENOBUFS); 426 *prtm = rtm; 427 } 428 rt_msg_buffer(rtm->rtm_type, rtinfo, rtm, msglen); 429 430 rtm->rtm_flags = rt->rt_flags; 431 rtm->rtm_rmx = rt->rt_rmx; 432 rtm->rtm_addrs = rtinfo->rti_addrs; 433 434 return (0); 435 } 436 437 static void route_output_add_callback(int, int, struct rt_addrinfo *, 438 struct rtentry *, void *); 439 static void route_output_delete_callback(int, int, struct rt_addrinfo *, 440 struct rtentry *, void *); 441 static void route_output_change_callback(int, int, struct rt_addrinfo *, 442 struct rtentry *, void *); 443 static void route_output_lock_callback(int, int, struct rt_addrinfo *, 444 struct rtentry *, void *); 445 446 /*ARGSUSED*/ 447 static int 448 route_output(struct mbuf *m, struct socket *so, ...) 449 { 450 struct rt_msghdr *rtm = NULL; 451 struct rtentry *rt; 452 struct radix_node_head *rnh; 453 struct rawcb *rp = NULL; 454 struct pr_output_info *oi; 455 struct rt_addrinfo rtinfo; 456 int len, error = 0; 457 __va_list ap; 458 459 __va_start(ap, so); 460 oi = __va_arg(ap, struct pr_output_info *); 461 __va_end(ap); 462 463 #define gotoerr(e) { error = e; goto flush;} 464 465 if (m == NULL || 466 (m->m_len < sizeof(long) && 467 (m = m_pullup(m, sizeof(long))) == NULL)) 468 return (ENOBUFS); 469 if (!(m->m_flags & M_PKTHDR)) 470 panic("route_output"); 471 len = m->m_pkthdr.len; 472 if (len < sizeof(struct rt_msghdr) || 473 len != mtod(m, struct rt_msghdr *)->rtm_msglen) { 474 rtinfo.rti_dst = NULL; 475 gotoerr(EINVAL); 476 } 477 rtm = kmalloc(len, M_RTABLE, M_INTWAIT | M_NULLOK); 478 if (rtm == NULL) { 479 rtinfo.rti_dst = NULL; 480 gotoerr(ENOBUFS); 481 } 482 m_copydata(m, 0, len, (caddr_t)rtm); 483 if (rtm->rtm_version != RTM_VERSION) { 484 rtinfo.rti_dst = NULL; 485 gotoerr(EPROTONOSUPPORT); 486 } 487 rtm->rtm_pid = oi->p_pid; 488 bzero(&rtinfo, sizeof(struct rt_addrinfo)); 489 rtinfo.rti_addrs = rtm->rtm_addrs; 490 if (rt_xaddrs((char *)(rtm + 1), (char *)rtm + len, &rtinfo) != 0) { 491 rtinfo.rti_dst = NULL; 492 gotoerr(EINVAL); 493 } 494 rtinfo.rti_flags = rtm->rtm_flags; 495 if (rtinfo.rti_dst == NULL || rtinfo.rti_dst->sa_family >= AF_MAX || 496 (rtinfo.rti_gateway && rtinfo.rti_gateway->sa_family >= AF_MAX)) 497 gotoerr(EINVAL); 498 499 if (rtinfo.rti_genmask != NULL) { 500 struct radix_node *n; 501 502 #define clen(s) (*(u_char *)(s)) 503 n = rn_addmask((char *)rtinfo.rti_genmask, TRUE, 1); 504 if (n != NULL && 505 rtinfo.rti_genmask->sa_len >= clen(n->rn_key) && 506 bcmp((char *)rtinfo.rti_genmask + 1, 507 (char *)n->rn_key + 1, clen(n->rn_key) - 1) == 0) 508 rtinfo.rti_genmask = (struct sockaddr *)n->rn_key; 509 else 510 gotoerr(ENOBUFS); 511 } 512 513 /* 514 * Verify that the caller has the appropriate privilege; RTM_GET 515 * is the only operation the non-superuser is allowed. 516 */ 517 if (rtm->rtm_type != RTM_GET && priv_check_cred(so->so_cred, PRIV_ROOT, 0) != 0) 518 gotoerr(EPERM); 519 520 switch (rtm->rtm_type) { 521 case RTM_ADD: 522 if (rtinfo.rti_gateway == NULL) { 523 error = EINVAL; 524 } else { 525 error = rtrequest1_global(RTM_ADD, &rtinfo, 526 route_output_add_callback, rtm); 527 } 528 break; 529 case RTM_DELETE: 530 /* 531 * note: &rtm passed as argument so 'rtm' can be replaced. 532 */ 533 error = rtrequest1_global(RTM_DELETE, &rtinfo, 534 route_output_delete_callback, &rtm); 535 break; 536 case RTM_GET: 537 rnh = rt_tables[mycpuid][rtinfo.rti_dst->sa_family]; 538 if (rnh == NULL) { 539 error = EAFNOSUPPORT; 540 break; 541 } 542 rt = (struct rtentry *) 543 rnh->rnh_lookup((char *)rtinfo.rti_dst, 544 (char *)rtinfo.rti_netmask, rnh); 545 if (rt == NULL) { 546 error = ESRCH; 547 break; 548 } 549 rt->rt_refcnt++; 550 if (fillrtmsg(&rtm, rt, &rtinfo) != 0) 551 gotoerr(ENOBUFS); 552 --rt->rt_refcnt; 553 break; 554 case RTM_CHANGE: 555 error = rtrequest1_global(RTM_GET, &rtinfo, 556 route_output_change_callback, rtm); 557 break; 558 case RTM_LOCK: 559 error = rtrequest1_global(RTM_GET, &rtinfo, 560 route_output_lock_callback, rtm); 561 break; 562 default: 563 error = EOPNOTSUPP; 564 break; 565 } 566 567 flush: 568 if (rtm != NULL) { 569 if (error != 0) 570 rtm->rtm_errno = error; 571 else 572 rtm->rtm_flags |= RTF_DONE; 573 } 574 575 /* 576 * Check to see if we don't want our own messages. 577 */ 578 if (!(so->so_options & SO_USELOOPBACK)) { 579 if (route_cb.any_count <= 1) { 580 if (rtm != NULL) 581 kfree(rtm, M_RTABLE); 582 m_freem(m); 583 return (error); 584 } 585 /* There is another listener, so construct message */ 586 rp = sotorawcb(so); 587 } 588 if (rtm != NULL) { 589 m_copyback(m, 0, rtm->rtm_msglen, (caddr_t)rtm); 590 if (m->m_pkthdr.len < rtm->rtm_msglen) { 591 m_freem(m); 592 m = NULL; 593 } else if (m->m_pkthdr.len > rtm->rtm_msglen) 594 m_adj(m, rtm->rtm_msglen - m->m_pkthdr.len); 595 kfree(rtm, M_RTABLE); 596 } 597 if (rp != NULL) 598 rp->rcb_proto.sp_family = 0; /* Avoid us */ 599 if (m != NULL) 600 rts_input(m, familyof(rtinfo.rti_dst)); 601 if (rp != NULL) 602 rp->rcb_proto.sp_family = PF_ROUTE; 603 return (error); 604 } 605 606 static void 607 route_output_add_callback(int cmd, int error, struct rt_addrinfo *rtinfo, 608 struct rtentry *rt, void *arg) 609 { 610 struct rt_msghdr *rtm = arg; 611 612 if (error == 0 && rt != NULL) { 613 rt_setmetrics(rtm->rtm_inits, &rtm->rtm_rmx, 614 &rt->rt_rmx); 615 rt->rt_rmx.rmx_locks &= ~(rtm->rtm_inits); 616 rt->rt_rmx.rmx_locks |= 617 (rtm->rtm_inits & rtm->rtm_rmx.rmx_locks); 618 rt->rt_genmask = rtinfo->rti_genmask; 619 } 620 } 621 622 static void 623 route_output_delete_callback(int cmd, int error, struct rt_addrinfo *rtinfo, 624 struct rtentry *rt, void *arg) 625 { 626 struct rt_msghdr **rtm = arg; 627 628 if (error == 0 && rt) { 629 ++rt->rt_refcnt; 630 if (fillrtmsg(rtm, rt, rtinfo) != 0) { 631 error = ENOBUFS; 632 /* XXX no way to return the error */ 633 } 634 --rt->rt_refcnt; 635 } 636 } 637 638 static void 639 route_output_change_callback(int cmd, int error, struct rt_addrinfo *rtinfo, 640 struct rtentry *rt, void *arg) 641 { 642 struct rt_msghdr *rtm = arg; 643 struct ifaddr *ifa; 644 645 if (error) 646 goto done; 647 648 /* 649 * new gateway could require new ifaddr, ifp; 650 * flags may also be different; ifp may be specified 651 * by ll sockaddr when protocol address is ambiguous 652 */ 653 if (((rt->rt_flags & RTF_GATEWAY) && rtinfo->rti_gateway != NULL) || 654 rtinfo->rti_ifpaddr != NULL || (rtinfo->rti_ifaaddr != NULL && 655 sa_equal(rtinfo->rti_ifaaddr, rt->rt_ifa->ifa_addr)) 656 ) { 657 error = rt_getifa(rtinfo); 658 if (error != 0) 659 goto done; 660 } 661 if (rtinfo->rti_gateway != NULL) { 662 error = rt_setgate(rt, rt_key(rt), rtinfo->rti_gateway); 663 if (error != 0) 664 goto done; 665 } 666 if ((ifa = rtinfo->rti_ifa) != NULL) { 667 struct ifaddr *oifa = rt->rt_ifa; 668 669 if (oifa != ifa) { 670 if (oifa && oifa->ifa_rtrequest) 671 oifa->ifa_rtrequest(RTM_DELETE, rt, rtinfo); 672 IFAFREE(rt->rt_ifa); 673 IFAREF(ifa); 674 rt->rt_ifa = ifa; 675 rt->rt_ifp = rtinfo->rti_ifp; 676 } 677 } 678 rt_setmetrics(rtm->rtm_inits, &rtm->rtm_rmx, &rt->rt_rmx); 679 if (rt->rt_ifa && rt->rt_ifa->ifa_rtrequest) 680 rt->rt_ifa->ifa_rtrequest(RTM_ADD, rt, rtinfo); 681 if (rtinfo->rti_genmask != NULL) 682 rt->rt_genmask = rtinfo->rti_genmask; 683 done: 684 /* XXX no way to return error */ 685 ; 686 } 687 688 static void 689 route_output_lock_callback(int cmd, int error, struct rt_addrinfo *rtinfo, 690 struct rtentry *rt, void *arg) 691 { 692 struct rt_msghdr *rtm = arg; 693 694 rt->rt_rmx.rmx_locks &= ~(rtm->rtm_inits); 695 rt->rt_rmx.rmx_locks |= 696 (rtm->rtm_inits & rtm->rtm_rmx.rmx_locks); 697 } 698 699 static void 700 rt_setmetrics(u_long which, struct rt_metrics *in, struct rt_metrics *out) 701 { 702 #define setmetric(flag, elt) if (which & (flag)) out->elt = in->elt; 703 setmetric(RTV_RPIPE, rmx_recvpipe); 704 setmetric(RTV_SPIPE, rmx_sendpipe); 705 setmetric(RTV_SSTHRESH, rmx_ssthresh); 706 setmetric(RTV_RTT, rmx_rtt); 707 setmetric(RTV_RTTVAR, rmx_rttvar); 708 setmetric(RTV_HOPCOUNT, rmx_hopcount); 709 setmetric(RTV_MTU, rmx_mtu); 710 setmetric(RTV_EXPIRE, rmx_expire); 711 #undef setmetric 712 } 713 714 #define ROUNDUP(a) \ 715 ((a) > 0 ? (1 + (((a) - 1) | (sizeof(long) - 1))) : sizeof(long)) 716 717 /* 718 * Extract the addresses of the passed sockaddrs. 719 * Do a little sanity checking so as to avoid bad memory references. 720 * This data is derived straight from userland. 721 */ 722 static int 723 rt_xaddrs(char *cp, char *cplim, struct rt_addrinfo *rtinfo) 724 { 725 struct sockaddr *sa; 726 int i; 727 728 for (i = 0; (i < RTAX_MAX) && (cp < cplim); i++) { 729 if ((rtinfo->rti_addrs & (1 << i)) == 0) 730 continue; 731 sa = (struct sockaddr *)cp; 732 /* 733 * It won't fit. 734 */ 735 if ((cp + sa->sa_len) > cplim) { 736 return (EINVAL); 737 } 738 739 /* 740 * There are no more... Quit now. 741 * If there are more bits, they are in error. 742 * I've seen this. route(1) can evidently generate these. 743 * This causes kernel to core dump. 744 * For compatibility, if we see this, point to a safe address. 745 */ 746 if (sa->sa_len == 0) { 747 static struct sockaddr sa_zero = { 748 sizeof sa_zero, AF_INET, 749 }; 750 751 rtinfo->rti_info[i] = &sa_zero; 752 kprintf("rtsock: received more addr bits than sockaddrs.\n"); 753 return (0); /* should be EINVAL but for compat */ 754 } 755 756 /* Accept the sockaddr. */ 757 rtinfo->rti_info[i] = sa; 758 cp += ROUNDUP(sa->sa_len); 759 } 760 return (0); 761 } 762 763 static int 764 rt_msghdrsize(int type) 765 { 766 switch (type) { 767 case RTM_DELADDR: 768 case RTM_NEWADDR: 769 return sizeof(struct ifa_msghdr); 770 case RTM_DELMADDR: 771 case RTM_NEWMADDR: 772 return sizeof(struct ifma_msghdr); 773 case RTM_IFINFO: 774 return sizeof(struct if_msghdr); 775 case RTM_IFANNOUNCE: 776 case RTM_IEEE80211: 777 return sizeof(struct if_announcemsghdr); 778 default: 779 return sizeof(struct rt_msghdr); 780 } 781 } 782 783 static int 784 rt_msgsize(int type, struct rt_addrinfo *rtinfo) 785 { 786 int len, i; 787 788 len = rt_msghdrsize(type); 789 for (i = 0; i < RTAX_MAX; i++) { 790 if (rtinfo->rti_info[i] != NULL) 791 len += ROUNDUP(rtinfo->rti_info[i]->sa_len); 792 } 793 len = ALIGN(len); 794 return len; 795 } 796 797 /* 798 * Build a routing message in a buffer. 799 * Copy the addresses in the rtinfo->rti_info[] sockaddr array 800 * to the end of the buffer after the message header. 801 * 802 * Set the rtinfo->rti_addrs bitmask of addresses present in rtinfo->rti_info[]. 803 * This side-effect can be avoided if we reorder the addrs bitmask field in all 804 * the route messages to line up so we can set it here instead of back in the 805 * calling routine. 806 */ 807 static void 808 rt_msg_buffer(int type, struct rt_addrinfo *rtinfo, void *buf, int msglen) 809 { 810 struct rt_msghdr *rtm; 811 char *cp; 812 int dlen, i; 813 814 rtm = (struct rt_msghdr *) buf; 815 rtm->rtm_version = RTM_VERSION; 816 rtm->rtm_type = type; 817 rtm->rtm_msglen = msglen; 818 819 cp = (char *)buf + rt_msghdrsize(type); 820 rtinfo->rti_addrs = 0; 821 for (i = 0; i < RTAX_MAX; i++) { 822 struct sockaddr *sa; 823 824 if ((sa = rtinfo->rti_info[i]) == NULL) 825 continue; 826 rtinfo->rti_addrs |= (1 << i); 827 dlen = ROUNDUP(sa->sa_len); 828 bcopy(sa, cp, dlen); 829 cp += dlen; 830 } 831 } 832 833 /* 834 * Build a routing message in a mbuf chain. 835 * Copy the addresses in the rtinfo->rti_info[] sockaddr array 836 * to the end of the mbuf after the message header. 837 * 838 * Set the rtinfo->rti_addrs bitmask of addresses present in rtinfo->rti_info[]. 839 * This side-effect can be avoided if we reorder the addrs bitmask field in all 840 * the route messages to line up so we can set it here instead of back in the 841 * calling routine. 842 */ 843 static struct mbuf * 844 rt_msg_mbuf(int type, struct rt_addrinfo *rtinfo) 845 { 846 struct mbuf *m; 847 struct rt_msghdr *rtm; 848 int hlen, len; 849 int i; 850 851 hlen = rt_msghdrsize(type); 852 KASSERT(hlen <= MCLBYTES, ("rt_msg_mbuf: hlen %d doesn't fit", hlen)); 853 854 m = m_getl(hlen, MB_DONTWAIT, MT_DATA, M_PKTHDR, NULL); 855 if (m == NULL) 856 return (NULL); 857 mbuftrackid(m, 32); 858 m->m_pkthdr.len = m->m_len = hlen; 859 m->m_pkthdr.rcvif = NULL; 860 rtinfo->rti_addrs = 0; 861 len = hlen; 862 for (i = 0; i < RTAX_MAX; i++) { 863 struct sockaddr *sa; 864 int dlen; 865 866 if ((sa = rtinfo->rti_info[i]) == NULL) 867 continue; 868 rtinfo->rti_addrs |= (1 << i); 869 dlen = ROUNDUP(sa->sa_len); 870 m_copyback(m, len, dlen, (caddr_t)sa); /* can grow mbuf chain */ 871 len += dlen; 872 } 873 if (m->m_pkthdr.len != len) { /* one of the m_copyback() calls failed */ 874 m_freem(m); 875 return (NULL); 876 } 877 rtm = mtod(m, struct rt_msghdr *); 878 bzero(rtm, hlen); 879 rtm->rtm_msglen = len; 880 rtm->rtm_version = RTM_VERSION; 881 rtm->rtm_type = type; 882 return (m); 883 } 884 885 /* 886 * This routine is called to generate a message from the routing 887 * socket indicating that a redirect has occurred, a routing lookup 888 * has failed, or that a protocol has detected timeouts to a particular 889 * destination. 890 */ 891 void 892 rt_missmsg(int type, struct rt_addrinfo *rtinfo, int flags, int error) 893 { 894 struct sockaddr *dst = rtinfo->rti_info[RTAX_DST]; 895 struct rt_msghdr *rtm; 896 struct mbuf *m; 897 898 if (route_cb.any_count == 0) 899 return; 900 m = rt_msg_mbuf(type, rtinfo); 901 if (m == NULL) 902 return; 903 rtm = mtod(m, struct rt_msghdr *); 904 rtm->rtm_flags = RTF_DONE | flags; 905 rtm->rtm_errno = error; 906 rtm->rtm_addrs = rtinfo->rti_addrs; 907 rts_input(m, familyof(dst)); 908 } 909 910 void 911 rt_dstmsg(int type, struct sockaddr *dst, int error) 912 { 913 struct rt_msghdr *rtm; 914 struct rt_addrinfo addrs; 915 struct mbuf *m; 916 917 if (route_cb.any_count == 0) 918 return; 919 bzero(&addrs, sizeof(struct rt_addrinfo)); 920 addrs.rti_info[RTAX_DST] = dst; 921 m = rt_msg_mbuf(type, &addrs); 922 if (m == NULL) 923 return; 924 rtm = mtod(m, struct rt_msghdr *); 925 rtm->rtm_flags = RTF_DONE; 926 rtm->rtm_errno = error; 927 rtm->rtm_addrs = addrs.rti_addrs; 928 rts_input(m, familyof(dst)); 929 } 930 931 /* 932 * This routine is called to generate a message from the routing 933 * socket indicating that the status of a network interface has changed. 934 */ 935 void 936 rt_ifmsg(struct ifnet *ifp) 937 { 938 struct if_msghdr *ifm; 939 struct mbuf *m; 940 struct rt_addrinfo rtinfo; 941 942 if (route_cb.any_count == 0) 943 return; 944 bzero(&rtinfo, sizeof(struct rt_addrinfo)); 945 m = rt_msg_mbuf(RTM_IFINFO, &rtinfo); 946 if (m == NULL) 947 return; 948 ifm = mtod(m, struct if_msghdr *); 949 ifm->ifm_index = ifp->if_index; 950 ifm->ifm_flags = ifp->if_flags; 951 ifm->ifm_data = ifp->if_data; 952 ifm->ifm_addrs = 0; 953 rts_input(m, 0); 954 } 955 956 static void 957 rt_ifamsg(int cmd, struct ifaddr *ifa) 958 { 959 struct ifa_msghdr *ifam; 960 struct rt_addrinfo rtinfo; 961 struct mbuf *m; 962 struct ifnet *ifp = ifa->ifa_ifp; 963 964 bzero(&rtinfo, sizeof(struct rt_addrinfo)); 965 rtinfo.rti_ifaaddr = ifa->ifa_addr; 966 rtinfo.rti_ifpaddr = 967 TAILQ_FIRST(&ifp->if_addrheads[mycpuid])->ifa->ifa_addr; 968 rtinfo.rti_netmask = ifa->ifa_netmask; 969 rtinfo.rti_bcastaddr = ifa->ifa_dstaddr; 970 971 m = rt_msg_mbuf(cmd, &rtinfo); 972 if (m == NULL) 973 return; 974 975 ifam = mtod(m, struct ifa_msghdr *); 976 ifam->ifam_index = ifp->if_index; 977 ifam->ifam_metric = ifa->ifa_metric; 978 ifam->ifam_flags = ifa->ifa_flags; 979 ifam->ifam_addrs = rtinfo.rti_addrs; 980 981 rts_input(m, familyof(ifa->ifa_addr)); 982 } 983 984 void 985 rt_rtmsg(int cmd, struct rtentry *rt, struct ifnet *ifp, int error) 986 { 987 struct rt_msghdr *rtm; 988 struct rt_addrinfo rtinfo; 989 struct mbuf *m; 990 struct sockaddr *dst; 991 992 if (rt == NULL) 993 return; 994 995 bzero(&rtinfo, sizeof(struct rt_addrinfo)); 996 rtinfo.rti_dst = dst = rt_key(rt); 997 rtinfo.rti_gateway = rt->rt_gateway; 998 rtinfo.rti_netmask = rt_mask(rt); 999 if (ifp != NULL) { 1000 rtinfo.rti_ifpaddr = 1001 TAILQ_FIRST(&ifp->if_addrheads[mycpuid])->ifa->ifa_addr; 1002 } 1003 rtinfo.rti_ifaaddr = rt->rt_ifa->ifa_addr; 1004 1005 m = rt_msg_mbuf(cmd, &rtinfo); 1006 if (m == NULL) 1007 return; 1008 1009 rtm = mtod(m, struct rt_msghdr *); 1010 if (ifp != NULL) 1011 rtm->rtm_index = ifp->if_index; 1012 rtm->rtm_flags |= rt->rt_flags; 1013 rtm->rtm_errno = error; 1014 rtm->rtm_addrs = rtinfo.rti_addrs; 1015 1016 rts_input(m, familyof(dst)); 1017 } 1018 1019 /* 1020 * This is called to generate messages from the routing socket 1021 * indicating a network interface has had addresses associated with it. 1022 * if we ever reverse the logic and replace messages TO the routing 1023 * socket indicate a request to configure interfaces, then it will 1024 * be unnecessary as the routing socket will automatically generate 1025 * copies of it. 1026 */ 1027 void 1028 rt_newaddrmsg(int cmd, struct ifaddr *ifa, int error, struct rtentry *rt) 1029 { 1030 #ifdef SCTP 1031 /* 1032 * notify the SCTP stack 1033 * this will only get called when an address is added/deleted 1034 * XXX pass the ifaddr struct instead if ifa->ifa_addr... 1035 */ 1036 if (cmd == RTM_ADD) 1037 sctp_add_ip_address(ifa); 1038 else if (cmd == RTM_DELETE) 1039 sctp_delete_ip_address(ifa); 1040 #endif /* SCTP */ 1041 1042 if (route_cb.any_count == 0) 1043 return; 1044 1045 if (cmd == RTM_ADD) { 1046 rt_ifamsg(RTM_NEWADDR, ifa); 1047 rt_rtmsg(RTM_ADD, rt, ifa->ifa_ifp, error); 1048 } else { 1049 KASSERT((cmd == RTM_DELETE), ("unknown cmd %d", cmd)); 1050 rt_rtmsg(RTM_DELETE, rt, ifa->ifa_ifp, error); 1051 rt_ifamsg(RTM_DELADDR, ifa); 1052 } 1053 } 1054 1055 /* 1056 * This is the analogue to the rt_newaddrmsg which performs the same 1057 * function but for multicast group memberhips. This is easier since 1058 * there is no route state to worry about. 1059 */ 1060 void 1061 rt_newmaddrmsg(int cmd, struct ifmultiaddr *ifma) 1062 { 1063 struct rt_addrinfo rtinfo; 1064 struct mbuf *m = NULL; 1065 struct ifnet *ifp = ifma->ifma_ifp; 1066 struct ifma_msghdr *ifmam; 1067 1068 if (route_cb.any_count == 0) 1069 return; 1070 1071 bzero(&rtinfo, sizeof(struct rt_addrinfo)); 1072 rtinfo.rti_ifaaddr = ifma->ifma_addr; 1073 if (ifp != NULL && !TAILQ_EMPTY(&ifp->if_addrheads[mycpuid])) { 1074 rtinfo.rti_ifpaddr = 1075 TAILQ_FIRST(&ifp->if_addrheads[mycpuid])->ifa->ifa_addr; 1076 } 1077 /* 1078 * If a link-layer address is present, present it as a ``gateway'' 1079 * (similarly to how ARP entries, e.g., are presented). 1080 */ 1081 rtinfo.rti_gateway = ifma->ifma_lladdr; 1082 1083 m = rt_msg_mbuf(cmd, &rtinfo); 1084 if (m == NULL) 1085 return; 1086 1087 ifmam = mtod(m, struct ifma_msghdr *); 1088 ifmam->ifmam_index = ifp->if_index; 1089 ifmam->ifmam_addrs = rtinfo.rti_addrs; 1090 1091 rts_input(m, familyof(ifma->ifma_addr)); 1092 } 1093 1094 static struct mbuf * 1095 rt_makeifannouncemsg(struct ifnet *ifp, int type, int what, 1096 struct rt_addrinfo *info) 1097 { 1098 struct if_announcemsghdr *ifan; 1099 struct mbuf *m; 1100 1101 if (route_cb.any_count == 0) 1102 return NULL; 1103 1104 bzero(info, sizeof(*info)); 1105 m = rt_msg_mbuf(type, info); 1106 if (m == NULL) 1107 return NULL; 1108 1109 ifan = mtod(m, struct if_announcemsghdr *); 1110 ifan->ifan_index = ifp->if_index; 1111 strlcpy(ifan->ifan_name, ifp->if_xname, sizeof ifan->ifan_name); 1112 ifan->ifan_what = what; 1113 return m; 1114 } 1115 1116 /* 1117 * This is called to generate routing socket messages indicating 1118 * IEEE80211 wireless events. 1119 * XXX we piggyback on the RTM_IFANNOUNCE msg format in a clumsy way. 1120 */ 1121 void 1122 rt_ieee80211msg(struct ifnet *ifp, int what, void *data, size_t data_len) 1123 { 1124 struct rt_addrinfo info; 1125 struct mbuf *m; 1126 1127 m = rt_makeifannouncemsg(ifp, RTM_IEEE80211, what, &info); 1128 if (m == NULL) 1129 return; 1130 1131 /* 1132 * Append the ieee80211 data. Try to stick it in the 1133 * mbuf containing the ifannounce msg; otherwise allocate 1134 * a new mbuf and append. 1135 * 1136 * NB: we assume m is a single mbuf. 1137 */ 1138 if (data_len > M_TRAILINGSPACE(m)) { 1139 struct mbuf *n = m_get(MB_DONTWAIT, MT_DATA); 1140 if (n == NULL) { 1141 m_freem(m); 1142 return; 1143 } 1144 bcopy(data, mtod(n, void *), data_len); 1145 n->m_len = data_len; 1146 m->m_next = n; 1147 } else if (data_len > 0) { 1148 bcopy(data, mtod(m, u_int8_t *) + m->m_len, data_len); 1149 m->m_len += data_len; 1150 } 1151 mbuftrackid(m, 33); 1152 if (m->m_flags & M_PKTHDR) 1153 m->m_pkthdr.len += data_len; 1154 mtod(m, struct if_announcemsghdr *)->ifan_msglen += data_len; 1155 rts_input(m, 0); 1156 } 1157 1158 /* 1159 * This is called to generate routing socket messages indicating 1160 * network interface arrival and departure. 1161 */ 1162 void 1163 rt_ifannouncemsg(struct ifnet *ifp, int what) 1164 { 1165 struct rt_addrinfo addrinfo; 1166 struct mbuf *m; 1167 1168 m = rt_makeifannouncemsg(ifp, RTM_IFANNOUNCE, what, &addrinfo); 1169 if (m != NULL) 1170 rts_input(m, 0); 1171 } 1172 1173 static int 1174 resizewalkarg(struct walkarg *w, int len) 1175 { 1176 void *newptr; 1177 1178 newptr = kmalloc(len, M_RTABLE, M_INTWAIT | M_NULLOK); 1179 if (newptr == NULL) 1180 return (ENOMEM); 1181 if (w->w_tmem != NULL) 1182 kfree(w->w_tmem, M_RTABLE); 1183 w->w_tmem = newptr; 1184 w->w_tmemsize = len; 1185 return (0); 1186 } 1187 1188 /* 1189 * This is used in dumping the kernel table via sysctl(). 1190 */ 1191 int 1192 sysctl_dumpentry(struct radix_node *rn, void *vw) 1193 { 1194 struct walkarg *w = vw; 1195 struct rtentry *rt = (struct rtentry *)rn; 1196 struct rt_addrinfo rtinfo; 1197 int error, msglen; 1198 1199 if (w->w_op == NET_RT_FLAGS && !(rt->rt_flags & w->w_arg)) 1200 return 0; 1201 1202 bzero(&rtinfo, sizeof(struct rt_addrinfo)); 1203 rtinfo.rti_dst = rt_key(rt); 1204 rtinfo.rti_gateway = rt->rt_gateway; 1205 rtinfo.rti_netmask = rt_mask(rt); 1206 rtinfo.rti_genmask = rt->rt_genmask; 1207 if (rt->rt_ifp != NULL) { 1208 rtinfo.rti_ifpaddr = 1209 TAILQ_FIRST(&rt->rt_ifp->if_addrheads[mycpuid])->ifa->ifa_addr; 1210 rtinfo.rti_ifaaddr = rt->rt_ifa->ifa_addr; 1211 if (rt->rt_ifp->if_flags & IFF_POINTOPOINT) 1212 rtinfo.rti_bcastaddr = rt->rt_ifa->ifa_dstaddr; 1213 } 1214 msglen = rt_msgsize(RTM_GET, &rtinfo); 1215 if (w->w_tmemsize < msglen && resizewalkarg(w, msglen) != 0) 1216 return (ENOMEM); 1217 rt_msg_buffer(RTM_GET, &rtinfo, w->w_tmem, msglen); 1218 if (w->w_req != NULL) { 1219 struct rt_msghdr *rtm = w->w_tmem; 1220 1221 rtm->rtm_flags = rt->rt_flags; 1222 rtm->rtm_use = rt->rt_use; 1223 rtm->rtm_rmx = rt->rt_rmx; 1224 rtm->rtm_index = rt->rt_ifp->if_index; 1225 rtm->rtm_errno = rtm->rtm_pid = rtm->rtm_seq = 0; 1226 rtm->rtm_addrs = rtinfo.rti_addrs; 1227 error = SYSCTL_OUT(w->w_req, rtm, msglen); 1228 return (error); 1229 } 1230 return (0); 1231 } 1232 1233 static int 1234 sysctl_iflist(int af, struct walkarg *w) 1235 { 1236 struct ifnet *ifp; 1237 struct rt_addrinfo rtinfo; 1238 int msglen, error; 1239 1240 bzero(&rtinfo, sizeof(struct rt_addrinfo)); 1241 TAILQ_FOREACH(ifp, &ifnet, if_link) { 1242 struct ifaddr_container *ifac; 1243 struct ifaddr *ifa; 1244 1245 if (w->w_arg && w->w_arg != ifp->if_index) 1246 continue; 1247 ifac = TAILQ_FIRST(&ifp->if_addrheads[mycpuid]); 1248 ifa = ifac->ifa; 1249 rtinfo.rti_ifpaddr = ifa->ifa_addr; 1250 msglen = rt_msgsize(RTM_IFINFO, &rtinfo); 1251 if (w->w_tmemsize < msglen && resizewalkarg(w, msglen) != 0) 1252 return (ENOMEM); 1253 rt_msg_buffer(RTM_IFINFO, &rtinfo, w->w_tmem, msglen); 1254 rtinfo.rti_ifpaddr = NULL; 1255 if (w->w_req != NULL && w->w_tmem != NULL) { 1256 struct if_msghdr *ifm = w->w_tmem; 1257 1258 ifm->ifm_index = ifp->if_index; 1259 ifm->ifm_flags = ifp->if_flags; 1260 ifm->ifm_data = ifp->if_data; 1261 ifm->ifm_addrs = rtinfo.rti_addrs; 1262 error = SYSCTL_OUT(w->w_req, ifm, msglen); 1263 if (error) 1264 return (error); 1265 } 1266 while ((ifac = TAILQ_NEXT(ifac, ifa_link)) != NULL) { 1267 ifa = ifac->ifa; 1268 1269 if (af && af != ifa->ifa_addr->sa_family) 1270 continue; 1271 if (curproc->p_ucred->cr_prison && 1272 prison_if(curproc->p_ucred, ifa->ifa_addr)) 1273 continue; 1274 rtinfo.rti_ifaaddr = ifa->ifa_addr; 1275 rtinfo.rti_netmask = ifa->ifa_netmask; 1276 rtinfo.rti_bcastaddr = ifa->ifa_dstaddr; 1277 msglen = rt_msgsize(RTM_NEWADDR, &rtinfo); 1278 if (w->w_tmemsize < msglen && 1279 resizewalkarg(w, msglen) != 0) 1280 return (ENOMEM); 1281 rt_msg_buffer(RTM_NEWADDR, &rtinfo, w->w_tmem, msglen); 1282 if (w->w_req != NULL) { 1283 struct ifa_msghdr *ifam = w->w_tmem; 1284 1285 ifam->ifam_index = ifa->ifa_ifp->if_index; 1286 ifam->ifam_flags = ifa->ifa_flags; 1287 ifam->ifam_metric = ifa->ifa_metric; 1288 ifam->ifam_addrs = rtinfo.rti_addrs; 1289 error = SYSCTL_OUT(w->w_req, w->w_tmem, msglen); 1290 if (error) 1291 return (error); 1292 } 1293 } 1294 rtinfo.rti_netmask = NULL; 1295 rtinfo.rti_ifaaddr = NULL; 1296 rtinfo.rti_bcastaddr = NULL; 1297 } 1298 return (0); 1299 } 1300 1301 static int 1302 sysctl_rtsock(SYSCTL_HANDLER_ARGS) 1303 { 1304 int *name = (int *)arg1; 1305 u_int namelen = arg2; 1306 struct radix_node_head *rnh; 1307 int i, error = EINVAL; 1308 int origcpu; 1309 u_char af; 1310 struct walkarg w; 1311 1312 name ++; 1313 namelen--; 1314 if (req->newptr) 1315 return (EPERM); 1316 if (namelen != 3 && namelen != 4) 1317 return (EINVAL); 1318 af = name[0]; 1319 bzero(&w, sizeof w); 1320 w.w_op = name[1]; 1321 w.w_arg = name[2]; 1322 w.w_req = req; 1323 1324 /* 1325 * Optional third argument specifies cpu, used primarily for 1326 * debugging the route table. 1327 */ 1328 if (namelen == 4) { 1329 if (name[3] < 0 || name[3] >= ncpus) 1330 return (EINVAL); 1331 origcpu = mycpuid; 1332 lwkt_migratecpu(name[3]); 1333 } else { 1334 origcpu = -1; 1335 } 1336 crit_enter(); 1337 switch (w.w_op) { 1338 case NET_RT_DUMP: 1339 case NET_RT_FLAGS: 1340 for (i = 1; i <= AF_MAX; i++) 1341 if ((rnh = rt_tables[mycpuid][i]) && 1342 (af == 0 || af == i) && 1343 (error = rnh->rnh_walktree(rnh, 1344 sysctl_dumpentry, &w))) 1345 break; 1346 break; 1347 1348 case NET_RT_IFLIST: 1349 error = sysctl_iflist(af, &w); 1350 } 1351 crit_exit(); 1352 if (w.w_tmem != NULL) 1353 kfree(w.w_tmem, M_RTABLE); 1354 if (origcpu >= 0) 1355 lwkt_migratecpu(origcpu); 1356 return (error); 1357 } 1358 1359 SYSCTL_NODE(_net, PF_ROUTE, routetable, CTLFLAG_RD, sysctl_rtsock, ""); 1360 1361 /* 1362 * Definitions of protocols supported in the ROUTE domain. 1363 */ 1364 1365 static struct domain routedomain; /* or at least forward */ 1366 1367 static struct protosw routesw[] = { 1368 { SOCK_RAW, &routedomain, 0, PR_ATOMIC|PR_ADDR, 1369 0, route_output, raw_ctlinput, 0, 1370 cpu0_soport, cpu0_ctlport, 1371 raw_init, 0, 0, 0, 1372 &route_usrreqs 1373 } 1374 }; 1375 1376 static struct domain routedomain = { 1377 PF_ROUTE, "route", NULL, NULL, NULL, 1378 routesw, &routesw[(sizeof routesw)/(sizeof routesw[0])], 1379 }; 1380 1381 DOMAIN_SET(route); 1382 1383