rtsock.c revision 1.115
1 /* $NetBSD: rtsock.c,v 1.115 2008/10/28 11:41:23 christos Exp $ */ 2 3 /* 4 * Copyright (C) 1995, 1996, 1997, and 1998 WIDE Project. 5 * All rights reserved. 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 project nor the names of its contributors 16 * may be used to endorse or promote products derived from this software 17 * without specific prior written permission. 18 * 19 * THIS SOFTWARE IS PROVIDED BY THE PROJECT AND CONTRIBUTORS ``AS IS'' AND 20 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE 21 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE 22 * ARE DISCLAIMED. IN NO EVENT SHALL THE PROJECT OR CONTRIBUTORS BE LIABLE 23 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL 24 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS 25 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) 26 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT 27 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY 28 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF 29 * SUCH DAMAGE. 30 */ 31 32 /* 33 * Copyright (c) 1988, 1991, 1993 34 * The Regents of the University of California. All rights reserved. 35 * 36 * Redistribution and use in source and binary forms, with or without 37 * modification, are permitted provided that the following conditions 38 * are met: 39 * 1. Redistributions of source code must retain the above copyright 40 * notice, this list of conditions and the following disclaimer. 41 * 2. Redistributions in binary form must reproduce the above copyright 42 * notice, this list of conditions and the following disclaimer in the 43 * documentation and/or other materials provided with the distribution. 44 * 3. Neither the name of the University nor the names of its contributors 45 * may be used to endorse or promote products derived from this software 46 * without specific prior written permission. 47 * 48 * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND 49 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE 50 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE 51 * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE 52 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL 53 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS 54 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) 55 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT 56 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY 57 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF 58 * SUCH DAMAGE. 59 * 60 * @(#)rtsock.c 8.7 (Berkeley) 10/12/95 61 */ 62 63 #include <sys/cdefs.h> 64 __KERNEL_RCSID(0, "$NetBSD: rtsock.c,v 1.115 2008/10/28 11:41:23 christos Exp $"); 65 66 #include "opt_inet.h" 67 68 #include <sys/param.h> 69 #include <sys/systm.h> 70 #include <sys/proc.h> 71 #include <sys/mbuf.h> 72 #include <sys/socket.h> 73 #include <sys/socketvar.h> 74 #include <sys/domain.h> 75 #include <sys/protosw.h> 76 #include <sys/sysctl.h> 77 #include <sys/kauth.h> 78 #include <sys/intr.h> 79 #ifdef RTSOCK_DEBUG 80 #include <netinet/in.h> 81 #endif /* RTSOCK_DEBUG */ 82 83 #include <net/if.h> 84 #include <net/route.h> 85 #include <net/raw_cb.h> 86 87 #include <machine/stdarg.h> 88 89 DOMAIN_DEFINE(routedomain); /* forward declare and add to link set */ 90 91 struct sockaddr route_dst = { .sa_len = 2, .sa_family = PF_ROUTE, }; 92 struct sockaddr route_src = { .sa_len = 2, .sa_family = PF_ROUTE, }; 93 94 int route_maxqlen = IFQ_MAXLEN; 95 static struct ifqueue route_intrq; 96 static void *route_sih; 97 98 struct walkarg { 99 int w_op; 100 int w_arg; 101 int w_given; 102 int w_needed; 103 void * w_where; 104 int w_tmemsize; 105 int w_tmemneeded; 106 void * w_tmem; 107 }; 108 109 static struct mbuf *rt_msg1(int, struct rt_addrinfo *, void *, int); 110 static int rt_msg2(int, struct rt_addrinfo *, void *, struct walkarg *, int *); 111 static int rt_xaddrs(u_char, const char *, const char *, struct rt_addrinfo *); 112 static struct mbuf *rt_makeifannouncemsg(struct ifnet *, int, int, 113 struct rt_addrinfo *); 114 static int sysctl_dumpentry(struct rtentry *, void *); 115 static int sysctl_iflist(int, struct walkarg *, int); 116 static int sysctl_rtable(SYSCTLFN_PROTO); 117 static inline void rt_adjustcount(int, int); 118 static void route_enqueue(struct mbuf *, int); 119 120 static inline void 121 rt_adjustcount(int af, int cnt) 122 { 123 route_cb.any_count += cnt; 124 switch (af) { 125 case AF_INET: 126 route_cb.ip_count += cnt; 127 return; 128 #ifdef INET6 129 case AF_INET6: 130 route_cb.ip6_count += cnt; 131 return; 132 #endif 133 case AF_IPX: 134 route_cb.ipx_count += cnt; 135 return; 136 case AF_NS: 137 route_cb.ns_count += cnt; 138 return; 139 case AF_ISO: 140 route_cb.iso_count += cnt; 141 return; 142 } 143 } 144 145 /*ARGSUSED*/ 146 int 147 route_usrreq(struct socket *so, int req, struct mbuf *m, struct mbuf *nam, 148 struct mbuf *control, struct lwp *l) 149 { 150 int error = 0; 151 struct rawcb *rp = sotorawcb(so); 152 int s; 153 154 if (req == PRU_ATTACH) { 155 sosetlock(so); 156 MALLOC(rp, struct rawcb *, sizeof(*rp), M_PCB, M_WAITOK|M_ZERO); 157 so->so_pcb = rp; 158 } 159 if (req == PRU_DETACH && rp) 160 rt_adjustcount(rp->rcb_proto.sp_protocol, -1); 161 s = splsoftnet(); 162 163 /* 164 * Don't call raw_usrreq() in the attach case, because 165 * we want to allow non-privileged processes to listen on 166 * and send "safe" commands to the routing socket. 167 */ 168 if (req == PRU_ATTACH) { 169 if (l == NULL) 170 error = EACCES; 171 else 172 error = raw_attach(so, (int)(long)nam); 173 } else 174 error = raw_usrreq(so, req, m, nam, control, l); 175 176 rp = sotorawcb(so); 177 if (req == PRU_ATTACH && rp) { 178 if (error) { 179 free(rp, M_PCB); 180 splx(s); 181 return error; 182 } 183 rt_adjustcount(rp->rcb_proto.sp_protocol, 1); 184 rp->rcb_laddr = &route_src; 185 rp->rcb_faddr = &route_dst; 186 soisconnected(so); 187 so->so_options |= SO_USELOOPBACK; 188 } 189 splx(s); 190 return error; 191 } 192 193 static const struct sockaddr * 194 intern_netmask(const struct sockaddr *mask) 195 { 196 struct radix_node *rn; 197 extern struct radix_node_head *mask_rnhead; 198 199 if (mask != NULL && 200 (rn = rn_search(mask, mask_rnhead->rnh_treetop))) 201 mask = (const struct sockaddr *)rn->rn_key; 202 203 return mask; 204 } 205 206 /*ARGSUSED*/ 207 int 208 route_output(struct mbuf *m, ...) 209 { 210 struct sockproto proto = { .sp_family = PF_ROUTE, }; 211 struct rt_msghdr *rtm = NULL; 212 struct rtentry *rt = NULL; 213 struct rtentry *saved_nrt = NULL; 214 struct rt_addrinfo info; 215 int len, error = 0; 216 struct ifnet *ifp = NULL; 217 struct ifaddr *ifa = NULL; 218 struct socket *so; 219 va_list ap; 220 sa_family_t family; 221 222 va_start(ap, m); 223 so = va_arg(ap, struct socket *); 224 va_end(ap); 225 226 #define senderr(e) do { error = e; goto flush;} while (/*CONSTCOND*/ 0) 227 if (m == NULL || ((m->m_len < sizeof(int32_t)) && 228 (m = m_pullup(m, sizeof(int32_t))) == NULL)) 229 return ENOBUFS; 230 if ((m->m_flags & M_PKTHDR) == 0) 231 panic("route_output"); 232 len = m->m_pkthdr.len; 233 if (len < sizeof(*rtm) || 234 len != mtod(m, struct rt_msghdr *)->rtm_msglen) { 235 info.rti_info[RTAX_DST] = NULL; 236 senderr(EINVAL); 237 } 238 R_Malloc(rtm, struct rt_msghdr *, len); 239 if (rtm == NULL) { 240 info.rti_info[RTAX_DST] = NULL; 241 senderr(ENOBUFS); 242 } 243 m_copydata(m, 0, len, rtm); 244 if (rtm->rtm_version != RTM_VERSION) { 245 info.rti_info[RTAX_DST] = NULL; 246 senderr(EPROTONOSUPPORT); 247 } 248 rtm->rtm_pid = curproc->p_pid; 249 memset(&info, 0, sizeof(info)); 250 info.rti_addrs = rtm->rtm_addrs; 251 if (rt_xaddrs(rtm->rtm_type, (const char *)(rtm + 1), len + (char *)rtm, 252 &info)) 253 senderr(EINVAL); 254 info.rti_flags = rtm->rtm_flags; 255 #ifdef RTSOCK_DEBUG 256 if (info.rti_info[RTAX_DST]->sa_family == AF_INET) { 257 printf("%s: extracted info.rti_info[RTAX_DST] %s\n", __func__, 258 inet_ntoa(((const struct sockaddr_in *) 259 info.rti_info[RTAX_DST])->sin_addr)); 260 } 261 #endif /* RTSOCK_DEBUG */ 262 if (info.rti_info[RTAX_DST] == NULL || 263 (info.rti_info[RTAX_DST]->sa_family >= AF_MAX)) 264 senderr(EINVAL); 265 if (info.rti_info[RTAX_GATEWAY] != NULL && 266 (info.rti_info[RTAX_GATEWAY]->sa_family >= AF_MAX)) 267 senderr(EINVAL); 268 269 /* 270 * Verify that the caller has the appropriate privilege; RTM_GET 271 * is the only operation the non-superuser is allowed. 272 */ 273 if (kauth_authorize_network(curlwp->l_cred, KAUTH_NETWORK_ROUTE, 274 0, rtm, NULL, NULL) != 0) 275 senderr(EACCES); 276 277 switch (rtm->rtm_type) { 278 279 case RTM_ADD: 280 if (info.rti_info[RTAX_GATEWAY] == NULL) 281 senderr(EINVAL); 282 error = rtrequest1(rtm->rtm_type, &info, &saved_nrt); 283 if (error == 0 && saved_nrt) { 284 rt_setmetrics(rtm->rtm_inits, 285 &rtm->rtm_rmx, &saved_nrt->rt_rmx); 286 saved_nrt->rt_refcnt--; 287 } 288 break; 289 290 case RTM_DELETE: 291 error = rtrequest1(rtm->rtm_type, &info, &saved_nrt); 292 if (error == 0) { 293 (rt = saved_nrt)->rt_refcnt++; 294 goto report; 295 } 296 break; 297 298 case RTM_GET: 299 case RTM_CHANGE: 300 case RTM_LOCK: 301 /* XXX This will mask info.rti_info[RTAX_DST] with 302 * info.rti_info[RTAX_NETMASK] before 303 * searching. It did not used to do that. --dyoung 304 */ 305 error = rtrequest1(RTM_GET, &info, &rt); 306 if (error != 0) 307 senderr(error); 308 if (rtm->rtm_type != RTM_GET) {/* XXX: too grotty */ 309 struct radix_node *rn; 310 311 if (memcmp(info.rti_info[RTAX_DST], rt_getkey(rt), 312 info.rti_info[RTAX_DST]->sa_len) != 0) 313 senderr(ESRCH); 314 info.rti_info[RTAX_NETMASK] = intern_netmask( 315 info.rti_info[RTAX_NETMASK]); 316 for (rn = rt->rt_nodes; rn; rn = rn->rn_dupedkey) 317 if (info.rti_info[RTAX_NETMASK] == 318 (const struct sockaddr *)rn->rn_mask) 319 break; 320 if (rn == NULL) 321 senderr(ETOOMANYREFS); 322 rt = (struct rtentry *)rn; 323 } 324 325 switch (rtm->rtm_type) { 326 case RTM_GET: 327 report: 328 info.rti_info[RTAX_DST] = rt_getkey(rt); 329 info.rti_info[RTAX_GATEWAY] = rt->rt_gateway; 330 info.rti_info[RTAX_NETMASK] = rt_mask(rt); 331 if ((rtm->rtm_addrs & (RTA_IFP | RTA_IFA)) == 0) 332 ; 333 else if ((ifp = rt->rt_ifp) != NULL) { 334 const struct ifaddr *rtifa; 335 info.rti_info[RTAX_IFP] = ifp->if_dl->ifa_addr; 336 /* rtifa used to be simply rt->rt_ifa. 337 * If rt->rt_ifa != NULL, then 338 * rt_get_ifa() != NULL. So this 339 * ought to still be safe. --dyoung 340 */ 341 rtifa = rt_get_ifa(rt); 342 info.rti_info[RTAX_IFA] = rtifa->ifa_addr; 343 #ifdef RTSOCK_DEBUG 344 if (info.rti_info[RTAX_IFA]->sa_family == 345 AF_INET) { 346 printf("%s: copying out RTAX_IFA %s ", 347 __func__, inet_ntoa( 348 (const struct sockaddr_in *) 349 info.rti_info[RTAX_IFA])->sin_addr); 350 printf("for info.rti_info[RTAX_DST] %s " 351 "ifa_getifa %p ifa_seqno %p\n", 352 inet_ntoa( 353 (const struct sockaddr_in *) 354 info.rti_info[RTAX_DST])->sin_addr), 355 (void *)rtifa->ifa_getifa, 356 rtifa->ifa_seqno); 357 } 358 #endif /* RTSOCK_DEBUG */ 359 if (ifp->if_flags & IFF_POINTOPOINT) { 360 info.rti_info[RTAX_BRD] = 361 rtifa->ifa_dstaddr; 362 } else 363 info.rti_info[RTAX_BRD] = NULL; 364 rtm->rtm_index = ifp->if_index; 365 } else { 366 info.rti_info[RTAX_IFP] = NULL; 367 info.rti_info[RTAX_IFA] = NULL; 368 } 369 (void)rt_msg2(rtm->rtm_type, &info, NULL, NULL, &len); 370 if (len > rtm->rtm_msglen) { 371 struct rt_msghdr *new_rtm; 372 R_Malloc(new_rtm, struct rt_msghdr *, len); 373 if (new_rtm == NULL) 374 senderr(ENOBUFS); 375 (void)memcpy(new_rtm, rtm, rtm->rtm_msglen); 376 Free(rtm); rtm = new_rtm; 377 } 378 (void)rt_msg2(rtm->rtm_type, &info, rtm, NULL, 0); 379 rtm->rtm_flags = rt->rt_flags; 380 rtm->rtm_rmx = rt->rt_rmx; 381 rtm->rtm_addrs = info.rti_addrs; 382 break; 383 384 case RTM_CHANGE: 385 /* 386 * new gateway could require new ifaddr, ifp; 387 * flags may also be different; ifp may be specified 388 * by ll sockaddr when protocol address is ambiguous 389 */ 390 if ((error = rt_getifa(&info)) != 0) 391 senderr(error); 392 if (info.rti_info[RTAX_GATEWAY] && 393 rt_setgate(rt, info.rti_info[RTAX_GATEWAY])) 394 senderr(EDQUOT); 395 /* new gateway could require new ifaddr, ifp; 396 flags may also be different; ifp may be specified 397 by ll sockaddr when protocol address is ambiguous */ 398 if (info.rti_info[RTAX_IFP] && 399 (ifa = ifa_ifwithnet(info.rti_info[RTAX_IFP])) && 400 (ifp = ifa->ifa_ifp) && (info.rti_info[RTAX_IFA] || 401 info.rti_info[RTAX_GATEWAY])) { 402 ifa = ifaof_ifpforaddr(info.rti_info[RTAX_IFA] ? 403 info.rti_info[RTAX_IFA] : 404 info.rti_info[RTAX_GATEWAY], ifp); 405 } else if ((info.rti_info[RTAX_IFA] && 406 (ifa = ifa_ifwithaddr(info.rti_info[RTAX_IFA]))) || 407 (info.rti_info[RTAX_GATEWAY] && 408 (ifa = ifa_ifwithroute(rt->rt_flags, 409 rt_getkey(rt), info.rti_info[RTAX_GATEWAY])))) { 410 ifp = ifa->ifa_ifp; 411 } 412 if (ifa) { 413 struct ifaddr *oifa = rt->rt_ifa; 414 if (oifa != ifa) { 415 if (oifa && oifa->ifa_rtrequest) { 416 oifa->ifa_rtrequest(RTM_DELETE, 417 rt, &info); 418 } 419 rt_replace_ifa(rt, ifa); 420 rt->rt_ifp = ifp; 421 } 422 } 423 rt_setmetrics(rtm->rtm_inits, &rtm->rtm_rmx, 424 &rt->rt_rmx); 425 if (rt->rt_ifa && rt->rt_ifa->ifa_rtrequest) 426 rt->rt_ifa->ifa_rtrequest(RTM_ADD, rt, &info); 427 /*FALLTHROUGH*/ 428 case RTM_LOCK: 429 rt->rt_rmx.rmx_locks &= ~(rtm->rtm_inits); 430 rt->rt_rmx.rmx_locks |= 431 (rtm->rtm_inits & rtm->rtm_rmx.rmx_locks); 432 break; 433 } 434 break; 435 436 default: 437 senderr(EOPNOTSUPP); 438 } 439 440 flush: 441 if (rtm) { 442 if (error) 443 rtm->rtm_errno = error; 444 else 445 rtm->rtm_flags |= RTF_DONE; 446 } 447 family = info.rti_info[RTAX_DST] ? info.rti_info[RTAX_DST]->sa_family : 448 0; 449 if (rt) 450 rtfree(rt); 451 { 452 struct rawcb *rp = NULL; 453 /* 454 * Check to see if we don't want our own messages. 455 */ 456 if ((so->so_options & SO_USELOOPBACK) == 0) { 457 if (route_cb.any_count <= 1) { 458 if (rtm) 459 Free(rtm); 460 m_freem(m); 461 return error; 462 } 463 /* There is another listener, so construct message */ 464 rp = sotorawcb(so); 465 } 466 if (rtm) { 467 m_copyback(m, 0, rtm->rtm_msglen, rtm); 468 if (m->m_pkthdr.len < rtm->rtm_msglen) { 469 m_freem(m); 470 m = NULL; 471 } else if (m->m_pkthdr.len > rtm->rtm_msglen) 472 m_adj(m, rtm->rtm_msglen - m->m_pkthdr.len); 473 Free(rtm); 474 } 475 if (rp) 476 rp->rcb_proto.sp_family = 0; /* Avoid us */ 477 if (family) 478 proto.sp_protocol = family; 479 if (m) 480 raw_input(m, &proto, &route_src, &route_dst); 481 if (rp) 482 rp->rcb_proto.sp_family = PF_ROUTE; 483 } 484 return error; 485 } 486 487 void 488 rt_setmetrics(u_long which, const struct rt_metrics *in, struct rt_metrics *out) 489 { 490 #define metric(f, e) if (which & (f)) out->e = in->e; 491 metric(RTV_RPIPE, rmx_recvpipe); 492 metric(RTV_SPIPE, rmx_sendpipe); 493 metric(RTV_SSTHRESH, rmx_ssthresh); 494 metric(RTV_RTT, rmx_rtt); 495 metric(RTV_RTTVAR, rmx_rttvar); 496 metric(RTV_HOPCOUNT, rmx_hopcount); 497 metric(RTV_MTU, rmx_mtu); 498 metric(RTV_EXPIRE, rmx_expire); 499 #undef metric 500 } 501 502 #define ROUNDUP(a) \ 503 ((a) > 0 ? (1 + (((a) - 1) | (sizeof(long) - 1))) : sizeof(long)) 504 #define ADVANCE(x, n) (x += ROUNDUP((n)->sa_len)) 505 506 static int 507 rt_xaddrs(u_char rtmtype, const char *cp, const char *cplim, 508 struct rt_addrinfo *rtinfo) 509 { 510 const struct sockaddr *sa = NULL; /* Quell compiler warning */ 511 int i; 512 513 #ifdef DIAGNOSTIC 514 if (rtinfo->rti_addrs & (1 << RTAX_GENMASK)) { 515 struct proc *p = curproc; 516 if (p != NULL) 517 printf("rt_xaddrs: process %d (%s) tried to alter " 518 "the route table with RTAX_GENMASK\n", 519 p->p_pid, p->p_comm); 520 } 521 #endif 522 for (i = 0; i < RTAX_MAX && cp < cplim; i++) { 523 if ((rtinfo->rti_addrs & (1 << i)) == 0) 524 continue; 525 if (i != RTAX_GENMASK) 526 rtinfo->rti_info[i] = sa = (const struct sockaddr *)cp; 527 ADVANCE(cp, sa); 528 } 529 530 /* 531 * Check for extra addresses specified, except RTM_GET asking 532 * for interface info. 533 */ 534 if (rtmtype == RTM_GET) { 535 if (((rtinfo->rti_addrs & 536 (~((1 << RTAX_IFP) | (1 << RTAX_IFA)))) & (~0 << i)) != 0) 537 return 1; 538 } else if ((rtinfo->rti_addrs & (~0 << i)) != 0) 539 return 1; 540 /* Check for bad data length. */ 541 if (cp != cplim) { 542 if (i == RTAX_NETMASK + 1 && sa != NULL && 543 cp - ROUNDUP(sa->sa_len) + sa->sa_len == cplim) 544 /* 545 * The last sockaddr was info.rti_info[RTAX_NETMASK]. 546 * We accept this for now for the sake of old 547 * binaries or third party softwares. 548 */ 549 ; 550 else 551 return 1; 552 } 553 return 0; 554 } 555 556 static struct mbuf * 557 rt_msg1(int type, struct rt_addrinfo *rtinfo, void *data, int datalen) 558 { 559 struct rt_msghdr *rtm; 560 struct mbuf *m; 561 int i; 562 const struct sockaddr *sa; 563 int len, dlen; 564 565 m = m_gethdr(M_DONTWAIT, MT_DATA); 566 if (m == NULL) 567 return m; 568 MCLAIM(m, &routedomain.dom_mowner); 569 switch (type) { 570 571 case RTM_DELADDR: 572 case RTM_NEWADDR: 573 len = sizeof(struct ifa_msghdr); 574 break; 575 576 #ifdef COMPAT_14 577 case RTM_OIFINFO: 578 len = sizeof(struct if_msghdr14); 579 break; 580 #endif 581 582 case RTM_IFINFO: 583 len = sizeof(struct if_msghdr); 584 break; 585 586 case RTM_IFANNOUNCE: 587 case RTM_IEEE80211: 588 len = sizeof(struct if_announcemsghdr); 589 break; 590 591 default: 592 len = sizeof(struct rt_msghdr); 593 } 594 if (len > MHLEN + MLEN) 595 panic("rt_msg1: message too long"); 596 else if (len > MHLEN) { 597 m->m_next = m_get(M_DONTWAIT, MT_DATA); 598 if (m->m_next == NULL) { 599 m_freem(m); 600 return NULL; 601 } 602 MCLAIM(m->m_next, m->m_owner); 603 m->m_pkthdr.len = len; 604 m->m_len = MHLEN; 605 m->m_next->m_len = len - MHLEN; 606 } else { 607 m->m_pkthdr.len = m->m_len = len; 608 } 609 m->m_pkthdr.rcvif = NULL; 610 m_copyback(m, 0, datalen, data); 611 if (len > datalen) 612 (void)memset(mtod(m, char *) + datalen, 0, len - datalen); 613 rtm = mtod(m, struct rt_msghdr *); 614 for (i = 0; i < RTAX_MAX; i++) { 615 if ((sa = rtinfo->rti_info[i]) == NULL) 616 continue; 617 rtinfo->rti_addrs |= (1 << i); 618 dlen = ROUNDUP(sa->sa_len); 619 m_copyback(m, len, dlen, sa); 620 len += dlen; 621 } 622 if (m->m_pkthdr.len != len) { 623 m_freem(m); 624 return NULL; 625 } 626 rtm->rtm_msglen = len; 627 rtm->rtm_version = RTM_VERSION; 628 rtm->rtm_type = type; 629 return m; 630 } 631 632 /* 633 * rt_msg2 634 * 635 * fills 'cp' or 'w'.w_tmem with the routing socket message and 636 * returns the length of the message in 'lenp'. 637 * 638 * if walkarg is 0, cp is expected to be 0 or a buffer large enough to hold 639 * the message 640 * otherwise walkarg's w_needed is updated and if the user buffer is 641 * specified and w_needed indicates space exists the information is copied 642 * into the temp space (w_tmem). w_tmem is [re]allocated if necessary, 643 * if the allocation fails ENOBUFS is returned. 644 */ 645 static int 646 rt_msg2(int type, struct rt_addrinfo *rtinfo, void *cpv, struct walkarg *w, 647 int *lenp) 648 { 649 int i; 650 int len, dlen, second_time = 0; 651 char *cp0, *cp = cpv; 652 653 rtinfo->rti_addrs = 0; 654 again: 655 switch (type) { 656 657 case RTM_DELADDR: 658 case RTM_NEWADDR: 659 len = sizeof(struct ifa_msghdr); 660 break; 661 #ifdef COMPAT_14 662 case RTM_OIFINFO: 663 len = sizeof(struct if_msghdr14); 664 break; 665 #endif 666 667 case RTM_IFINFO: 668 len = sizeof(struct if_msghdr); 669 break; 670 671 default: 672 len = sizeof(struct rt_msghdr); 673 } 674 if ((cp0 = cp) != NULL) 675 cp += len; 676 for (i = 0; i < RTAX_MAX; i++) { 677 const struct sockaddr *sa; 678 679 if ((sa = rtinfo->rti_info[i]) == NULL) 680 continue; 681 rtinfo->rti_addrs |= (1 << i); 682 dlen = ROUNDUP(sa->sa_len); 683 if (cp) { 684 (void)memcpy(cp, sa, (size_t)dlen); 685 cp += dlen; 686 } 687 len += dlen; 688 } 689 if (cp == NULL && w != NULL && !second_time) { 690 struct walkarg *rw = w; 691 692 rw->w_needed += len; 693 if (rw->w_needed <= 0 && rw->w_where) { 694 if (rw->w_tmemsize < len) { 695 if (rw->w_tmem) 696 free(rw->w_tmem, M_RTABLE); 697 rw->w_tmem = malloc(len, M_RTABLE, M_NOWAIT); 698 if (rw->w_tmem) 699 rw->w_tmemsize = len; 700 else 701 rw->w_tmemsize = 0; 702 } 703 if (rw->w_tmem) { 704 cp = rw->w_tmem; 705 second_time = 1; 706 goto again; 707 } else { 708 rw->w_tmemneeded = len; 709 return ENOBUFS; 710 } 711 } 712 } 713 if (cp) { 714 struct rt_msghdr *rtm = (struct rt_msghdr *)cp0; 715 716 rtm->rtm_version = RTM_VERSION; 717 rtm->rtm_type = type; 718 rtm->rtm_msglen = len; 719 } 720 if (lenp) 721 *lenp = len; 722 return 0; 723 } 724 725 /* 726 * This routine is called to generate a message from the routing 727 * socket indicating that a redirect has occurred, a routing lookup 728 * has failed, or that a protocol has detected timeouts to a particular 729 * destination. 730 */ 731 void 732 rt_missmsg(int type, struct rt_addrinfo *rtinfo, int flags, int error) 733 { 734 struct rt_msghdr rtm; 735 struct mbuf *m; 736 const struct sockaddr *sa = rtinfo->rti_info[RTAX_DST]; 737 738 if (route_cb.any_count == 0) 739 return; 740 memset(&rtm, 0, sizeof(rtm)); 741 rtm.rtm_flags = RTF_DONE | flags; 742 rtm.rtm_errno = error; 743 m = rt_msg1(type, rtinfo, &rtm, sizeof(rtm)); 744 if (m == NULL) 745 return; 746 mtod(m, struct rt_msghdr *)->rtm_addrs = rtinfo->rti_addrs; 747 route_enqueue(m, sa ? sa->sa_family : 0); 748 } 749 750 /* 751 * This routine is called to generate a message from the routing 752 * socket indicating that the status of a network interface has changed. 753 */ 754 void 755 rt_ifmsg(struct ifnet *ifp) 756 { 757 struct if_msghdr ifm; 758 #ifdef COMPAT_14 759 struct if_msghdr14 oifm; 760 #endif 761 struct mbuf *m; 762 struct rt_addrinfo info; 763 764 if (route_cb.any_count == 0) 765 return; 766 memset(&info, 0, sizeof(info)); 767 memset(&ifm, 0, sizeof(ifm)); 768 ifm.ifm_index = ifp->if_index; 769 ifm.ifm_flags = ifp->if_flags; 770 ifm.ifm_data = ifp->if_data; 771 ifm.ifm_addrs = 0; 772 m = rt_msg1(RTM_IFINFO, &info, &ifm, sizeof(ifm)); 773 if (m == NULL) 774 return; 775 route_enqueue(m, 0); 776 #ifdef COMPAT_14 777 memset(&info, 0, sizeof(info)); 778 memset(&oifm, 0, sizeof(oifm)); 779 oifm.ifm_index = ifp->if_index; 780 oifm.ifm_flags = ifp->if_flags; 781 oifm.ifm_data.ifi_type = ifp->if_data.ifi_type; 782 oifm.ifm_data.ifi_addrlen = ifp->if_data.ifi_addrlen; 783 oifm.ifm_data.ifi_hdrlen = ifp->if_data.ifi_hdrlen; 784 oifm.ifm_data.ifi_mtu = ifp->if_data.ifi_mtu; 785 oifm.ifm_data.ifi_metric = ifp->if_data.ifi_metric; 786 oifm.ifm_data.ifi_baudrate = ifp->if_data.ifi_baudrate; 787 oifm.ifm_data.ifi_ipackets = ifp->if_data.ifi_ipackets; 788 oifm.ifm_data.ifi_ierrors = ifp->if_data.ifi_ierrors; 789 oifm.ifm_data.ifi_opackets = ifp->if_data.ifi_opackets; 790 oifm.ifm_data.ifi_oerrors = ifp->if_data.ifi_oerrors; 791 oifm.ifm_data.ifi_collisions = ifp->if_data.ifi_collisions; 792 oifm.ifm_data.ifi_ibytes = ifp->if_data.ifi_ibytes; 793 oifm.ifm_data.ifi_obytes = ifp->if_data.ifi_obytes; 794 oifm.ifm_data.ifi_imcasts = ifp->if_data.ifi_imcasts; 795 oifm.ifm_data.ifi_omcasts = ifp->if_data.ifi_omcasts; 796 oifm.ifm_data.ifi_iqdrops = ifp->if_data.ifi_iqdrops; 797 oifm.ifm_data.ifi_noproto = ifp->if_data.ifi_noproto; 798 oifm.ifm_data.ifi_lastchange = ifp->if_data.ifi_lastchange; 799 oifm.ifm_addrs = 0; 800 m = rt_msg1(RTM_OIFINFO, &info, &oifm, sizeof(oifm)); 801 if (m == NULL) 802 return; 803 route_enqueue(m, 0); 804 #endif 805 } 806 807 /* 808 * This is called to generate messages from the routing socket 809 * indicating a network interface has had addresses associated with it. 810 * if we ever reverse the logic and replace messages TO the routing 811 * socket indicate a request to configure interfaces, then it will 812 * be unnecessary as the routing socket will automatically generate 813 * copies of it. 814 */ 815 void 816 rt_newaddrmsg(int cmd, struct ifaddr *ifa, int error, struct rtentry *rt) 817 { 818 struct rt_addrinfo info; 819 const struct sockaddr *sa = NULL; 820 int pass; 821 struct mbuf *m = NULL; 822 struct ifnet *ifp = ifa->ifa_ifp; 823 824 if (route_cb.any_count == 0) 825 return; 826 for (pass = 1; pass < 3; pass++) { 827 memset(&info, 0, sizeof(info)); 828 if ((cmd == RTM_ADD && pass == 1) || 829 (cmd == RTM_DELETE && pass == 2)) { 830 struct ifa_msghdr ifam; 831 int ncmd = cmd == RTM_ADD ? RTM_NEWADDR : RTM_DELADDR; 832 833 info.rti_info[RTAX_IFA] = sa = ifa->ifa_addr; 834 info.rti_info[RTAX_IFP] = ifp->if_dl->ifa_addr; 835 info.rti_info[RTAX_NETMASK] = ifa->ifa_netmask; 836 info.rti_info[RTAX_BRD] = ifa->ifa_dstaddr; 837 memset(&ifam, 0, sizeof(ifam)); 838 ifam.ifam_index = ifp->if_index; 839 ifam.ifam_metric = ifa->ifa_metric; 840 ifam.ifam_flags = ifa->ifa_flags; 841 m = rt_msg1(ncmd, &info, &ifam, sizeof(ifam)); 842 if (m == NULL) 843 continue; 844 mtod(m, struct ifa_msghdr *)->ifam_addrs = 845 info.rti_addrs; 846 } 847 if ((cmd == RTM_ADD && pass == 2) || 848 (cmd == RTM_DELETE && pass == 1)) { 849 struct rt_msghdr rtm; 850 851 if (rt == NULL) 852 continue; 853 info.rti_info[RTAX_NETMASK] = rt_mask(rt); 854 info.rti_info[RTAX_DST] = sa = rt_getkey(rt); 855 info.rti_info[RTAX_GATEWAY] = rt->rt_gateway; 856 memset(&rtm, 0, sizeof(rtm)); 857 rtm.rtm_index = ifp->if_index; 858 rtm.rtm_flags |= rt->rt_flags; 859 rtm.rtm_errno = error; 860 m = rt_msg1(cmd, &info, &rtm, sizeof(rtm)); 861 if (m == NULL) 862 continue; 863 mtod(m, struct rt_msghdr *)->rtm_addrs = info.rti_addrs; 864 } 865 #ifdef DIAGNOSTIC 866 if (m == NULL) 867 panic("%s: called with wrong command", __func__); 868 #endif 869 route_enqueue(m, sa ? sa->sa_family : 0); 870 } 871 } 872 873 static struct mbuf * 874 rt_makeifannouncemsg(struct ifnet *ifp, int type, int what, 875 struct rt_addrinfo *info) 876 { 877 struct if_announcemsghdr ifan; 878 879 memset(info, 0, sizeof(*info)); 880 memset(&ifan, 0, sizeof(ifan)); 881 ifan.ifan_index = ifp->if_index; 882 strlcpy(ifan.ifan_name, ifp->if_xname, sizeof(ifan.ifan_name)); 883 ifan.ifan_what = what; 884 return rt_msg1(type, info, &ifan, sizeof(ifan)); 885 } 886 887 /* 888 * This is called to generate routing socket messages indicating 889 * network interface arrival and departure. 890 */ 891 void 892 rt_ifannouncemsg(struct ifnet *ifp, int what) 893 { 894 struct mbuf *m; 895 struct rt_addrinfo info; 896 897 if (route_cb.any_count == 0) 898 return; 899 m = rt_makeifannouncemsg(ifp, RTM_IFANNOUNCE, what, &info); 900 if (m == NULL) 901 return; 902 route_enqueue(m, 0); 903 } 904 905 /* 906 * This is called to generate routing socket messages indicating 907 * IEEE80211 wireless events. 908 * XXX we piggyback on the RTM_IFANNOUNCE msg format in a clumsy way. 909 */ 910 void 911 rt_ieee80211msg(struct ifnet *ifp, int what, void *data, size_t data_len) 912 { 913 struct mbuf *m; 914 struct rt_addrinfo info; 915 916 if (route_cb.any_count == 0) 917 return; 918 m = rt_makeifannouncemsg(ifp, RTM_IEEE80211, what, &info); 919 if (m == NULL) 920 return; 921 /* 922 * Append the ieee80211 data. Try to stick it in the 923 * mbuf containing the ifannounce msg; otherwise allocate 924 * a new mbuf and append. 925 * 926 * NB: we assume m is a single mbuf. 927 */ 928 if (data_len > M_TRAILINGSPACE(m)) { 929 struct mbuf *n = m_get(M_NOWAIT, MT_DATA); 930 if (n == NULL) { 931 m_freem(m); 932 return; 933 } 934 (void)memcpy(mtod(n, void *), data, data_len); 935 n->m_len = data_len; 936 m->m_next = n; 937 } else if (data_len > 0) { 938 (void)memcpy(mtod(m, uint8_t *) + m->m_len, data, data_len); 939 m->m_len += data_len; 940 } 941 if (m->m_flags & M_PKTHDR) 942 m->m_pkthdr.len += data_len; 943 mtod(m, struct if_announcemsghdr *)->ifan_msglen += data_len; 944 route_enqueue(m, 0); 945 } 946 947 /* 948 * This is used in dumping the kernel table via sysctl(). 949 */ 950 static int 951 sysctl_dumpentry(struct rtentry *rt, void *v) 952 { 953 struct walkarg *w = v; 954 int error = 0, size; 955 struct rt_addrinfo info; 956 957 if (w->w_op == NET_RT_FLAGS && !(rt->rt_flags & w->w_arg)) 958 return 0; 959 memset(&info, 0, sizeof(info)); 960 info.rti_info[RTAX_DST] = rt_getkey(rt); 961 info.rti_info[RTAX_GATEWAY] = rt->rt_gateway; 962 info.rti_info[RTAX_NETMASK] = rt_mask(rt); 963 if (rt->rt_ifp) { 964 const struct ifaddr *rtifa; 965 info.rti_info[RTAX_IFP] = rt->rt_ifp->if_dl->ifa_addr; 966 /* rtifa used to be simply rt->rt_ifa. If rt->rt_ifa != NULL, 967 * then rt_get_ifa() != NULL. So this ought to still be safe. 968 * --dyoung 969 */ 970 rtifa = rt_get_ifa(rt); 971 info.rti_info[RTAX_IFA] = rtifa->ifa_addr; 972 if (rt->rt_ifp->if_flags & IFF_POINTOPOINT) 973 info.rti_info[RTAX_BRD] = rtifa->ifa_dstaddr; 974 } 975 if ((error = rt_msg2(RTM_GET, &info, 0, w, &size))) 976 return error; 977 if (w->w_where && w->w_tmem && w->w_needed <= 0) { 978 struct rt_msghdr *rtm = (struct rt_msghdr *)w->w_tmem; 979 980 rtm->rtm_flags = rt->rt_flags; 981 rtm->rtm_use = rt->rt_use; 982 rtm->rtm_rmx = rt->rt_rmx; 983 KASSERT(rt->rt_ifp != NULL); 984 rtm->rtm_index = rt->rt_ifp->if_index; 985 rtm->rtm_errno = rtm->rtm_pid = rtm->rtm_seq = 0; 986 rtm->rtm_addrs = info.rti_addrs; 987 if ((error = copyout(rtm, w->w_where, size)) != 0) 988 w->w_where = NULL; 989 else 990 w->w_where = (char *)w->w_where + size; 991 } 992 return error; 993 } 994 995 static int 996 sysctl_iflist(int af, struct walkarg *w, int type) 997 { 998 struct ifnet *ifp; 999 struct ifaddr *ifa; 1000 struct rt_addrinfo info; 1001 int len, error = 0; 1002 1003 memset(&info, 0, sizeof(info)); 1004 IFNET_FOREACH(ifp) { 1005 if (w->w_arg && w->w_arg != ifp->if_index) 1006 continue; 1007 if (IFADDR_EMPTY(ifp)) 1008 continue; 1009 info.rti_info[RTAX_IFP] = ifp->if_dl->ifa_addr; 1010 switch (type) { 1011 case NET_RT_IFLIST: 1012 error = rt_msg2(RTM_IFINFO, &info, NULL, w, &len); 1013 break; 1014 #ifdef COMPAT_14 1015 case NET_RT_OIFLIST: 1016 error = rt_msg2(RTM_OIFINFO, &info, NULL, w, &len); 1017 break; 1018 #endif 1019 default: 1020 panic("sysctl_iflist(1)"); 1021 } 1022 if (error) 1023 return error; 1024 info.rti_info[RTAX_IFP] = NULL; 1025 if (w->w_where && w->w_tmem && w->w_needed <= 0) { 1026 switch (type) { 1027 case NET_RT_IFLIST: { 1028 struct if_msghdr *ifm; 1029 1030 ifm = (struct if_msghdr *)w->w_tmem; 1031 ifm->ifm_index = ifp->if_index; 1032 ifm->ifm_flags = ifp->if_flags; 1033 ifm->ifm_data = ifp->if_data; 1034 ifm->ifm_addrs = info.rti_addrs; 1035 error = copyout(ifm, w->w_where, len); 1036 if (error) 1037 return error; 1038 w->w_where = (char *)w->w_where + len; 1039 break; 1040 } 1041 1042 #ifdef COMPAT_14 1043 case NET_RT_OIFLIST: { 1044 struct if_msghdr14 *ifm; 1045 1046 ifm = (struct if_msghdr14 *)w->w_tmem; 1047 ifm->ifm_index = ifp->if_index; 1048 ifm->ifm_flags = ifp->if_flags; 1049 ifm->ifm_data.ifi_type = ifp->if_data.ifi_type; 1050 ifm->ifm_data.ifi_addrlen = 1051 ifp->if_data.ifi_addrlen; 1052 ifm->ifm_data.ifi_hdrlen = 1053 ifp->if_data.ifi_hdrlen; 1054 ifm->ifm_data.ifi_mtu = ifp->if_data.ifi_mtu; 1055 ifm->ifm_data.ifi_metric = 1056 ifp->if_data.ifi_metric; 1057 ifm->ifm_data.ifi_baudrate = 1058 ifp->if_data.ifi_baudrate; 1059 ifm->ifm_data.ifi_ipackets = 1060 ifp->if_data.ifi_ipackets; 1061 ifm->ifm_data.ifi_ierrors = 1062 ifp->if_data.ifi_ierrors; 1063 ifm->ifm_data.ifi_opackets = 1064 ifp->if_data.ifi_opackets; 1065 ifm->ifm_data.ifi_oerrors = 1066 ifp->if_data.ifi_oerrors; 1067 ifm->ifm_data.ifi_collisions = 1068 ifp->if_data.ifi_collisions; 1069 ifm->ifm_data.ifi_ibytes = 1070 ifp->if_data.ifi_ibytes; 1071 ifm->ifm_data.ifi_obytes = 1072 ifp->if_data.ifi_obytes; 1073 ifm->ifm_data.ifi_imcasts = 1074 ifp->if_data.ifi_imcasts; 1075 ifm->ifm_data.ifi_omcasts = 1076 ifp->if_data.ifi_omcasts; 1077 ifm->ifm_data.ifi_iqdrops = 1078 ifp->if_data.ifi_iqdrops; 1079 ifm->ifm_data.ifi_noproto = 1080 ifp->if_data.ifi_noproto; 1081 ifm->ifm_data.ifi_lastchange = 1082 ifp->if_data.ifi_lastchange; 1083 ifm->ifm_addrs = info.rti_addrs; 1084 error = copyout(ifm, w->w_where, len); 1085 if (error) 1086 return error; 1087 w->w_where = (char *)w->w_where + len; 1088 break; 1089 } 1090 #endif 1091 default: 1092 panic("sysctl_iflist(2)"); 1093 } 1094 } 1095 IFADDR_FOREACH(ifa, ifp) { 1096 if (af && af != ifa->ifa_addr->sa_family) 1097 continue; 1098 info.rti_info[RTAX_IFA] = ifa->ifa_addr; 1099 info.rti_info[RTAX_NETMASK] = ifa->ifa_netmask; 1100 info.rti_info[RTAX_BRD] = ifa->ifa_dstaddr; 1101 if ((error = rt_msg2(RTM_NEWADDR, &info, 0, w, &len))) 1102 return error; 1103 if (w->w_where && w->w_tmem && w->w_needed <= 0) { 1104 struct ifa_msghdr *ifam; 1105 1106 ifam = (struct ifa_msghdr *)w->w_tmem; 1107 ifam->ifam_index = ifa->ifa_ifp->if_index; 1108 ifam->ifam_flags = ifa->ifa_flags; 1109 ifam->ifam_metric = ifa->ifa_metric; 1110 ifam->ifam_addrs = info.rti_addrs; 1111 error = copyout(w->w_tmem, w->w_where, len); 1112 if (error) 1113 return error; 1114 w->w_where = (char *)w->w_where + len; 1115 } 1116 } 1117 info.rti_info[RTAX_IFA] = info.rti_info[RTAX_NETMASK] = 1118 info.rti_info[RTAX_BRD] = NULL; 1119 } 1120 return 0; 1121 } 1122 1123 static int 1124 sysctl_rtable(SYSCTLFN_ARGS) 1125 { 1126 void *where = oldp; 1127 size_t *given = oldlenp; 1128 const void *new = newp; 1129 int i, s, error = EINVAL; 1130 u_char af; 1131 struct walkarg w; 1132 1133 if (namelen == 1 && name[0] == CTL_QUERY) 1134 return sysctl_query(SYSCTLFN_CALL(rnode)); 1135 1136 if (new) 1137 return EPERM; 1138 if (namelen != 3) 1139 return EINVAL; 1140 af = name[0]; 1141 w.w_tmemneeded = 0; 1142 w.w_tmemsize = 0; 1143 w.w_tmem = NULL; 1144 again: 1145 /* we may return here if a later [re]alloc of the t_mem buffer fails */ 1146 if (w.w_tmemneeded) { 1147 w.w_tmem = malloc(w.w_tmemneeded, M_RTABLE, M_WAITOK); 1148 w.w_tmemsize = w.w_tmemneeded; 1149 w.w_tmemneeded = 0; 1150 } 1151 w.w_op = name[1]; 1152 w.w_arg = name[2]; 1153 w.w_given = *given; 1154 w.w_needed = 0 - w.w_given; 1155 w.w_where = where; 1156 1157 s = splsoftnet(); 1158 switch (w.w_op) { 1159 1160 case NET_RT_DUMP: 1161 case NET_RT_FLAGS: 1162 for (i = 1; i <= AF_MAX; i++) 1163 if ((af == 0 || af == i) && 1164 (error = rt_walktree(i, sysctl_dumpentry, &w))) 1165 break; 1166 break; 1167 1168 #ifdef COMPAT_14 1169 case NET_RT_OIFLIST: 1170 error = sysctl_iflist(af, &w, w.w_op); 1171 break; 1172 #endif 1173 1174 case NET_RT_IFLIST: 1175 error = sysctl_iflist(af, &w, w.w_op); 1176 } 1177 splx(s); 1178 1179 /* check to see if we couldn't allocate memory with NOWAIT */ 1180 if (error == ENOBUFS && w.w_tmem == 0 && w.w_tmemneeded) 1181 goto again; 1182 1183 if (w.w_tmem) 1184 free(w.w_tmem, M_RTABLE); 1185 w.w_needed += w.w_given; 1186 if (where) { 1187 *given = (char *)w.w_where - (char *)where; 1188 if (*given < w.w_needed) 1189 return ENOMEM; 1190 } else { 1191 *given = (11 * w.w_needed) / 10; 1192 } 1193 return error; 1194 } 1195 1196 /* 1197 * Routing message software interrupt routine 1198 */ 1199 static void 1200 route_intr(void *cookie) 1201 { 1202 struct sockproto proto = { .sp_family = PF_ROUTE, }; 1203 struct mbuf *m; 1204 int s; 1205 1206 mutex_enter(softnet_lock); 1207 KERNEL_LOCK(1, NULL); 1208 while (!IF_IS_EMPTY(&route_intrq)) { 1209 s = splnet(); 1210 IF_DEQUEUE(&route_intrq, m); 1211 splx(s); 1212 if (m == NULL) 1213 break; 1214 proto.sp_protocol = M_GETCTX(m, uintptr_t); 1215 raw_input(m, &proto, &route_src, &route_dst); 1216 } 1217 KERNEL_UNLOCK_ONE(NULL); 1218 mutex_exit(softnet_lock); 1219 } 1220 1221 /* 1222 * Enqueue a message to the software interrupt routine. 1223 */ 1224 static void 1225 route_enqueue(struct mbuf *m, int family) 1226 { 1227 int s, wasempty; 1228 1229 s = splnet(); 1230 if (IF_QFULL(&route_intrq)) { 1231 IF_DROP(&route_intrq); 1232 m_freem(m); 1233 } else { 1234 wasempty = IF_IS_EMPTY(&route_intrq); 1235 M_SETCTX(m, (uintptr_t)family); 1236 IF_ENQUEUE(&route_intrq, m); 1237 if (wasempty) 1238 softint_schedule(route_sih); 1239 } 1240 splx(s); 1241 } 1242 1243 void 1244 rt_init(void) 1245 { 1246 1247 route_intrq.ifq_maxlen = route_maxqlen; 1248 route_sih = softint_establish(SOFTINT_NET | SOFTINT_MPSAFE, 1249 route_intr, NULL); 1250 } 1251 1252 /* 1253 * Definitions of protocols supported in the ROUTE domain. 1254 */ 1255 PR_WRAP_USRREQ(route_usrreq) 1256 #define route_usrreq route_usrreq_wrapper 1257 1258 const struct protosw routesw[] = { 1259 { 1260 .pr_type = SOCK_RAW, 1261 .pr_domain = &routedomain, 1262 .pr_flags = PR_ATOMIC|PR_ADDR, 1263 .pr_input = raw_input, 1264 .pr_output = route_output, 1265 .pr_ctlinput = raw_ctlinput, 1266 .pr_usrreq = route_usrreq, 1267 .pr_init = raw_init, 1268 }, 1269 }; 1270 1271 struct domain routedomain = { 1272 .dom_family = PF_ROUTE, 1273 .dom_name = "route", 1274 .dom_init = route_init, 1275 .dom_protosw = routesw, 1276 .dom_protoswNPROTOSW = &routesw[__arraycount(routesw)], 1277 }; 1278 1279 SYSCTL_SETUP(sysctl_net_route_setup, "sysctl net.route subtree setup") 1280 { 1281 const struct sysctlnode *rnode = NULL; 1282 1283 sysctl_createv(clog, 0, NULL, NULL, 1284 CTLFLAG_PERMANENT, 1285 CTLTYPE_NODE, "net", NULL, 1286 NULL, 0, NULL, 0, 1287 CTL_NET, CTL_EOL); 1288 1289 sysctl_createv(clog, 0, NULL, &rnode, 1290 CTLFLAG_PERMANENT, 1291 CTLTYPE_NODE, "route", 1292 SYSCTL_DESCR("PF_ROUTE information"), 1293 NULL, 0, NULL, 0, 1294 CTL_NET, PF_ROUTE, CTL_EOL); 1295 sysctl_createv(clog, 0, NULL, NULL, 1296 CTLFLAG_PERMANENT, 1297 CTLTYPE_NODE, "rtable", 1298 SYSCTL_DESCR("Routing table information"), 1299 sysctl_rtable, 0, NULL, 0, 1300 CTL_NET, PF_ROUTE, 0 /* any protocol */, CTL_EOL); 1301 sysctl_createv(clog, 0, &rnode, NULL, 1302 CTLFLAG_PERMANENT, 1303 CTLTYPE_STRUCT, "stats", 1304 SYSCTL_DESCR("Routing statistics"), 1305 NULL, 0, &rtstat, sizeof(rtstat), 1306 CTL_CREATE, CTL_EOL); 1307 } 1308
Indexes created Tue Sep 23 07:09:52 GMT 2025