rtsock.c revision 1.118
1 /* $NetBSD: rtsock.c,v 1.118 2008/12/17 20:51:37 cegger 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 1.118 2008/12/17 20:51:37 cegger 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 rp = malloc(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 rt_msghdr *old_rtm = NULL; 213 struct rtentry *rt = NULL; 214 struct rtentry *saved_nrt = NULL; 215 struct rt_addrinfo info; 216 int len, error = 0; 217 struct ifnet *ifp = NULL; 218 struct ifaddr *ifa = NULL; 219 struct socket *so; 220 va_list ap; 221 sa_family_t family; 222 223 va_start(ap, m); 224 so = va_arg(ap, struct socket *); 225 va_end(ap); 226 227 #define senderr(e) do { error = e; goto flush;} while (/*CONSTCOND*/ 0) 228 if (m == NULL || ((m->m_len < sizeof(int32_t)) && 229 (m = m_pullup(m, sizeof(int32_t))) == NULL)) 230 return ENOBUFS; 231 if ((m->m_flags & M_PKTHDR) == 0) 232 panic("route_output"); 233 len = m->m_pkthdr.len; 234 if (len < sizeof(*rtm) || 235 len != mtod(m, struct rt_msghdr *)->rtm_msglen) { 236 info.rti_info[RTAX_DST] = NULL; 237 senderr(EINVAL); 238 } 239 R_Malloc(rtm, struct rt_msghdr *, len); 240 if (rtm == NULL) { 241 info.rti_info[RTAX_DST] = NULL; 242 senderr(ENOBUFS); 243 } 244 m_copydata(m, 0, len, rtm); 245 if (rtm->rtm_version != RTM_VERSION) { 246 info.rti_info[RTAX_DST] = NULL; 247 senderr(EPROTONOSUPPORT); 248 } 249 rtm->rtm_pid = curproc->p_pid; 250 memset(&info, 0, sizeof(info)); 251 info.rti_addrs = rtm->rtm_addrs; 252 if (rt_xaddrs(rtm->rtm_type, (const char *)(rtm + 1), len + (char *)rtm, 253 &info)) 254 senderr(EINVAL); 255 info.rti_flags = rtm->rtm_flags; 256 #ifdef RTSOCK_DEBUG 257 if (info.rti_info[RTAX_DST]->sa_family == AF_INET) { 258 printf("%s: extracted info.rti_info[RTAX_DST] %s\n", __func__, 259 inet_ntoa(((const struct sockaddr_in *) 260 info.rti_info[RTAX_DST])->sin_addr)); 261 } 262 #endif /* RTSOCK_DEBUG */ 263 if (info.rti_info[RTAX_DST] == NULL || 264 (info.rti_info[RTAX_DST]->sa_family >= AF_MAX)) 265 senderr(EINVAL); 266 if (info.rti_info[RTAX_GATEWAY] != NULL && 267 (info.rti_info[RTAX_GATEWAY]->sa_family >= AF_MAX)) 268 senderr(EINVAL); 269 270 /* 271 * Verify that the caller has the appropriate privilege; RTM_GET 272 * is the only operation the non-superuser is allowed. 273 */ 274 if (kauth_authorize_network(curlwp->l_cred, KAUTH_NETWORK_ROUTE, 275 0, rtm, NULL, NULL) != 0) 276 senderr(EACCES); 277 278 switch (rtm->rtm_type) { 279 280 case RTM_ADD: 281 if (info.rti_info[RTAX_GATEWAY] == NULL) 282 senderr(EINVAL); 283 error = rtrequest1(rtm->rtm_type, &info, &saved_nrt); 284 if (error == 0 && saved_nrt) { 285 rt_setmetrics(rtm->rtm_inits, 286 &rtm->rtm_rmx, &saved_nrt->rt_rmx); 287 saved_nrt->rt_refcnt--; 288 } 289 break; 290 291 case RTM_DELETE: 292 error = rtrequest1(rtm->rtm_type, &info, &saved_nrt); 293 if (error == 0) { 294 (rt = saved_nrt)->rt_refcnt++; 295 goto report; 296 } 297 break; 298 299 case RTM_GET: 300 case RTM_CHANGE: 301 case RTM_LOCK: 302 /* XXX This will mask info.rti_info[RTAX_DST] with 303 * info.rti_info[RTAX_NETMASK] before 304 * searching. It did not used to do that. --dyoung 305 */ 306 error = rtrequest1(RTM_GET, &info, &rt); 307 if (error != 0) 308 senderr(error); 309 if (rtm->rtm_type != RTM_GET) {/* XXX: too grotty */ 310 struct radix_node *rn; 311 312 if (memcmp(info.rti_info[RTAX_DST], rt_getkey(rt), 313 info.rti_info[RTAX_DST]->sa_len) != 0) 314 senderr(ESRCH); 315 info.rti_info[RTAX_NETMASK] = intern_netmask( 316 info.rti_info[RTAX_NETMASK]); 317 for (rn = rt->rt_nodes; rn; rn = rn->rn_dupedkey) 318 if (info.rti_info[RTAX_NETMASK] == 319 (const struct sockaddr *)rn->rn_mask) 320 break; 321 if (rn == NULL) 322 senderr(ETOOMANYREFS); 323 rt = (struct rtentry *)rn; 324 } 325 326 switch (rtm->rtm_type) { 327 case RTM_GET: 328 report: 329 info.rti_info[RTAX_DST] = rt_getkey(rt); 330 info.rti_info[RTAX_GATEWAY] = rt->rt_gateway; 331 info.rti_info[RTAX_NETMASK] = rt_mask(rt); 332 if ((rtm->rtm_addrs & (RTA_IFP | RTA_IFA)) == 0) 333 ; 334 else if ((ifp = rt->rt_ifp) != NULL) { 335 const struct ifaddr *rtifa; 336 info.rti_info[RTAX_IFP] = ifp->if_dl->ifa_addr; 337 /* rtifa used to be simply rt->rt_ifa. 338 * If rt->rt_ifa != NULL, then 339 * rt_get_ifa() != NULL. So this 340 * ought to still be safe. --dyoung 341 */ 342 rtifa = rt_get_ifa(rt); 343 info.rti_info[RTAX_IFA] = rtifa->ifa_addr; 344 #ifdef RTSOCK_DEBUG 345 if (info.rti_info[RTAX_IFA]->sa_family == 346 AF_INET) { 347 printf("%s: copying out RTAX_IFA %s ", 348 __func__, inet_ntoa( 349 (const struct sockaddr_in *) 350 info.rti_info[RTAX_IFA])->sin_addr); 351 printf("for info.rti_info[RTAX_DST] %s " 352 "ifa_getifa %p ifa_seqno %p\n", 353 inet_ntoa( 354 (const struct sockaddr_in *) 355 info.rti_info[RTAX_DST])->sin_addr), 356 (void *)rtifa->ifa_getifa, 357 rtifa->ifa_seqno); 358 } 359 #endif /* RTSOCK_DEBUG */ 360 if (ifp->if_flags & IFF_POINTOPOINT) { 361 info.rti_info[RTAX_BRD] = 362 rtifa->ifa_dstaddr; 363 } else 364 info.rti_info[RTAX_BRD] = NULL; 365 rtm->rtm_index = ifp->if_index; 366 } else { 367 info.rti_info[RTAX_IFP] = NULL; 368 info.rti_info[RTAX_IFA] = NULL; 369 } 370 (void)rt_msg2(rtm->rtm_type, &info, NULL, NULL, &len); 371 if (len > rtm->rtm_msglen) { 372 old_rtm = rtm; 373 R_Malloc(rtm, struct rt_msghdr *, len); 374 if (rtm == NULL) 375 senderr(ENOBUFS); 376 (void)memcpy(rtm, old_rtm, old_rtm->rtm_msglen); 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 /* We cannot free old_rtm until we have stopped using the 450 * pointers in info, some of which may point to sockaddrs 451 * in old_rtm. 452 */ 453 if (old_rtm != NULL) 454 Free(old_rtm); 455 if (rt) 456 rtfree(rt); 457 { 458 struct rawcb *rp = NULL; 459 /* 460 * Check to see if we don't want our own messages. 461 */ 462 if ((so->so_options & SO_USELOOPBACK) == 0) { 463 if (route_cb.any_count <= 1) { 464 if (rtm) 465 Free(rtm); 466 m_freem(m); 467 return error; 468 } 469 /* There is another listener, so construct message */ 470 rp = sotorawcb(so); 471 } 472 if (rtm) { 473 m_copyback(m, 0, rtm->rtm_msglen, rtm); 474 if (m->m_pkthdr.len < rtm->rtm_msglen) { 475 m_freem(m); 476 m = NULL; 477 } else if (m->m_pkthdr.len > rtm->rtm_msglen) 478 m_adj(m, rtm->rtm_msglen - m->m_pkthdr.len); 479 Free(rtm); 480 } 481 if (rp) 482 rp->rcb_proto.sp_family = 0; /* Avoid us */ 483 if (family) 484 proto.sp_protocol = family; 485 if (m) 486 raw_input(m, &proto, &route_src, &route_dst); 487 if (rp) 488 rp->rcb_proto.sp_family = PF_ROUTE; 489 } 490 return error; 491 } 492 493 void 494 rt_setmetrics(u_long which, const struct rt_metrics *in, struct rt_metrics *out) 495 { 496 #define metric(f, e) if (which & (f)) out->e = in->e; 497 metric(RTV_RPIPE, rmx_recvpipe); 498 metric(RTV_SPIPE, rmx_sendpipe); 499 metric(RTV_SSTHRESH, rmx_ssthresh); 500 metric(RTV_RTT, rmx_rtt); 501 metric(RTV_RTTVAR, rmx_rttvar); 502 metric(RTV_HOPCOUNT, rmx_hopcount); 503 metric(RTV_MTU, rmx_mtu); 504 metric(RTV_EXPIRE, rmx_expire); 505 #undef metric 506 } 507 508 #define ROUNDUP(a) \ 509 ((a) > 0 ? (1 + (((a) - 1) | (sizeof(long) - 1))) : sizeof(long)) 510 #define ADVANCE(x, n) (x += ROUNDUP((n)->sa_len)) 511 512 static int 513 rt_xaddrs(u_char rtmtype, const char *cp, const char *cplim, 514 struct rt_addrinfo *rtinfo) 515 { 516 const struct sockaddr *sa = NULL; /* Quell compiler warning */ 517 int i; 518 519 for (i = 0; i < RTAX_MAX && cp < cplim; i++) { 520 if ((rtinfo->rti_addrs & (1 << i)) == 0) 521 continue; 522 rtinfo->rti_info[i] = sa = (const struct sockaddr *)cp; 523 ADVANCE(cp, sa); 524 } 525 526 /* 527 * Check for extra addresses specified, except RTM_GET asking 528 * for interface info. 529 */ 530 if (rtmtype == RTM_GET) { 531 if (((rtinfo->rti_addrs & 532 (~((1 << RTAX_IFP) | (1 << RTAX_IFA)))) & (~0 << i)) != 0) 533 return 1; 534 } else if ((rtinfo->rti_addrs & (~0 << i)) != 0) 535 return 1; 536 /* Check for bad data length. */ 537 if (cp != cplim) { 538 if (i == RTAX_NETMASK + 1 && sa != NULL && 539 cp - ROUNDUP(sa->sa_len) + sa->sa_len == cplim) 540 /* 541 * The last sockaddr was info.rti_info[RTAX_NETMASK]. 542 * We accept this for now for the sake of old 543 * binaries or third party softwares. 544 */ 545 ; 546 else 547 return 1; 548 } 549 return 0; 550 } 551 552 static struct mbuf * 553 rt_msg1(int type, struct rt_addrinfo *rtinfo, void *data, int datalen) 554 { 555 struct rt_msghdr *rtm; 556 struct mbuf *m; 557 int i; 558 const struct sockaddr *sa; 559 int len, dlen; 560 561 m = m_gethdr(M_DONTWAIT, MT_DATA); 562 if (m == NULL) 563 return m; 564 MCLAIM(m, &routedomain.dom_mowner); 565 switch (type) { 566 567 case RTM_DELADDR: 568 case RTM_NEWADDR: 569 len = sizeof(struct ifa_msghdr); 570 break; 571 572 #ifdef COMPAT_14 573 case RTM_OIFINFO: 574 len = sizeof(struct if_msghdr14); 575 break; 576 #endif 577 578 case RTM_IFINFO: 579 len = sizeof(struct if_msghdr); 580 break; 581 582 case RTM_IFANNOUNCE: 583 case RTM_IEEE80211: 584 len = sizeof(struct if_announcemsghdr); 585 break; 586 587 default: 588 len = sizeof(struct rt_msghdr); 589 } 590 if (len > MHLEN + MLEN) 591 panic("rt_msg1: message too long"); 592 else if (len > MHLEN) { 593 m->m_next = m_get(M_DONTWAIT, MT_DATA); 594 if (m->m_next == NULL) { 595 m_freem(m); 596 return NULL; 597 } 598 MCLAIM(m->m_next, m->m_owner); 599 m->m_pkthdr.len = len; 600 m->m_len = MHLEN; 601 m->m_next->m_len = len - MHLEN; 602 } else { 603 m->m_pkthdr.len = m->m_len = len; 604 } 605 m->m_pkthdr.rcvif = NULL; 606 m_copyback(m, 0, datalen, data); 607 if (len > datalen) 608 (void)memset(mtod(m, char *) + datalen, 0, len - datalen); 609 rtm = mtod(m, struct rt_msghdr *); 610 for (i = 0; i < RTAX_MAX; i++) { 611 if ((sa = rtinfo->rti_info[i]) == NULL) 612 continue; 613 rtinfo->rti_addrs |= (1 << i); 614 dlen = ROUNDUP(sa->sa_len); 615 m_copyback(m, len, dlen, sa); 616 len += dlen; 617 } 618 if (m->m_pkthdr.len != len) { 619 m_freem(m); 620 return NULL; 621 } 622 rtm->rtm_msglen = len; 623 rtm->rtm_version = RTM_VERSION; 624 rtm->rtm_type = type; 625 return m; 626 } 627 628 /* 629 * rt_msg2 630 * 631 * fills 'cp' or 'w'.w_tmem with the routing socket message and 632 * returns the length of the message in 'lenp'. 633 * 634 * if walkarg is 0, cp is expected to be 0 or a buffer large enough to hold 635 * the message 636 * otherwise walkarg's w_needed is updated and if the user buffer is 637 * specified and w_needed indicates space exists the information is copied 638 * into the temp space (w_tmem). w_tmem is [re]allocated if necessary, 639 * if the allocation fails ENOBUFS is returned. 640 */ 641 static int 642 rt_msg2(int type, struct rt_addrinfo *rtinfo, void *cpv, struct walkarg *w, 643 int *lenp) 644 { 645 int i; 646 int len, dlen, second_time = 0; 647 char *cp0, *cp = cpv; 648 649 rtinfo->rti_addrs = 0; 650 again: 651 switch (type) { 652 653 case RTM_DELADDR: 654 case RTM_NEWADDR: 655 len = sizeof(struct ifa_msghdr); 656 break; 657 #ifdef COMPAT_14 658 case RTM_OIFINFO: 659 len = sizeof(struct if_msghdr14); 660 break; 661 #endif 662 663 case RTM_IFINFO: 664 len = sizeof(struct if_msghdr); 665 break; 666 667 default: 668 len = sizeof(struct rt_msghdr); 669 } 670 if ((cp0 = cp) != NULL) 671 cp += len; 672 for (i = 0; i < RTAX_MAX; i++) { 673 const struct sockaddr *sa; 674 675 if ((sa = rtinfo->rti_info[i]) == NULL) 676 continue; 677 rtinfo->rti_addrs |= (1 << i); 678 dlen = ROUNDUP(sa->sa_len); 679 if (cp) { 680 (void)memcpy(cp, sa, (size_t)dlen); 681 cp += dlen; 682 } 683 len += dlen; 684 } 685 if (cp == NULL && w != NULL && !second_time) { 686 struct walkarg *rw = w; 687 688 rw->w_needed += len; 689 if (rw->w_needed <= 0 && rw->w_where) { 690 if (rw->w_tmemsize < len) { 691 if (rw->w_tmem) 692 free(rw->w_tmem, M_RTABLE); 693 rw->w_tmem = malloc(len, M_RTABLE, M_NOWAIT); 694 if (rw->w_tmem) 695 rw->w_tmemsize = len; 696 else 697 rw->w_tmemsize = 0; 698 } 699 if (rw->w_tmem) { 700 cp = rw->w_tmem; 701 second_time = 1; 702 goto again; 703 } else { 704 rw->w_tmemneeded = len; 705 return ENOBUFS; 706 } 707 } 708 } 709 if (cp) { 710 struct rt_msghdr *rtm = (struct rt_msghdr *)cp0; 711 712 rtm->rtm_version = RTM_VERSION; 713 rtm->rtm_type = type; 714 rtm->rtm_msglen = len; 715 } 716 if (lenp) 717 *lenp = len; 718 return 0; 719 } 720 721 /* 722 * This routine is called to generate a message from the routing 723 * socket indicating that a redirect has occurred, a routing lookup 724 * has failed, or that a protocol has detected timeouts to a particular 725 * destination. 726 */ 727 void 728 rt_missmsg(int type, struct rt_addrinfo *rtinfo, int flags, int error) 729 { 730 struct rt_msghdr rtm; 731 struct mbuf *m; 732 const struct sockaddr *sa = rtinfo->rti_info[RTAX_DST]; 733 734 if (route_cb.any_count == 0) 735 return; 736 memset(&rtm, 0, sizeof(rtm)); 737 rtm.rtm_flags = RTF_DONE | flags; 738 rtm.rtm_errno = error; 739 m = rt_msg1(type, rtinfo, &rtm, sizeof(rtm)); 740 if (m == NULL) 741 return; 742 mtod(m, struct rt_msghdr *)->rtm_addrs = rtinfo->rti_addrs; 743 route_enqueue(m, sa ? sa->sa_family : 0); 744 } 745 746 /* 747 * This routine is called to generate a message from the routing 748 * socket indicating that the status of a network interface has changed. 749 */ 750 void 751 rt_ifmsg(struct ifnet *ifp) 752 { 753 struct if_msghdr ifm; 754 #ifdef COMPAT_14 755 struct if_msghdr14 oifm; 756 #endif 757 struct mbuf *m; 758 struct rt_addrinfo info; 759 760 if (route_cb.any_count == 0) 761 return; 762 memset(&info, 0, sizeof(info)); 763 memset(&ifm, 0, sizeof(ifm)); 764 ifm.ifm_index = ifp->if_index; 765 ifm.ifm_flags = ifp->if_flags; 766 ifm.ifm_data = ifp->if_data; 767 ifm.ifm_addrs = 0; 768 m = rt_msg1(RTM_IFINFO, &info, &ifm, sizeof(ifm)); 769 if (m == NULL) 770 return; 771 route_enqueue(m, 0); 772 #ifdef COMPAT_14 773 memset(&info, 0, sizeof(info)); 774 memset(&oifm, 0, sizeof(oifm)); 775 oifm.ifm_index = ifp->if_index; 776 oifm.ifm_flags = ifp->if_flags; 777 oifm.ifm_data.ifi_type = ifp->if_data.ifi_type; 778 oifm.ifm_data.ifi_addrlen = ifp->if_data.ifi_addrlen; 779 oifm.ifm_data.ifi_hdrlen = ifp->if_data.ifi_hdrlen; 780 oifm.ifm_data.ifi_mtu = ifp->if_data.ifi_mtu; 781 oifm.ifm_data.ifi_metric = ifp->if_data.ifi_metric; 782 oifm.ifm_data.ifi_baudrate = ifp->if_data.ifi_baudrate; 783 oifm.ifm_data.ifi_ipackets = ifp->if_data.ifi_ipackets; 784 oifm.ifm_data.ifi_ierrors = ifp->if_data.ifi_ierrors; 785 oifm.ifm_data.ifi_opackets = ifp->if_data.ifi_opackets; 786 oifm.ifm_data.ifi_oerrors = ifp->if_data.ifi_oerrors; 787 oifm.ifm_data.ifi_collisions = ifp->if_data.ifi_collisions; 788 oifm.ifm_data.ifi_ibytes = ifp->if_data.ifi_ibytes; 789 oifm.ifm_data.ifi_obytes = ifp->if_data.ifi_obytes; 790 oifm.ifm_data.ifi_imcasts = ifp->if_data.ifi_imcasts; 791 oifm.ifm_data.ifi_omcasts = ifp->if_data.ifi_omcasts; 792 oifm.ifm_data.ifi_iqdrops = ifp->if_data.ifi_iqdrops; 793 oifm.ifm_data.ifi_noproto = ifp->if_data.ifi_noproto; 794 oifm.ifm_data.ifi_lastchange = ifp->if_data.ifi_lastchange; 795 oifm.ifm_addrs = 0; 796 m = rt_msg1(RTM_OIFINFO, &info, &oifm, sizeof(oifm)); 797 if (m == NULL) 798 return; 799 route_enqueue(m, 0); 800 #endif 801 } 802 803 /* 804 * This is called to generate messages from the routing socket 805 * indicating a network interface has had addresses associated with it. 806 * if we ever reverse the logic and replace messages TO the routing 807 * socket indicate a request to configure interfaces, then it will 808 * be unnecessary as the routing socket will automatically generate 809 * copies of it. 810 */ 811 void 812 rt_newaddrmsg(int cmd, struct ifaddr *ifa, int error, struct rtentry *rt) 813 { 814 #define cmdpass(__cmd, __pass) (((__cmd) << 2) | (__pass)) 815 struct rt_addrinfo info; 816 const struct sockaddr *sa; 817 int pass; 818 struct mbuf *m; 819 struct ifnet *ifp = ifa->ifa_ifp; 820 struct rt_msghdr rtm; 821 struct ifa_msghdr ifam; 822 int ncmd; 823 824 if (route_cb.any_count == 0) 825 return; 826 for (pass = 1; pass < 3; pass++) { 827 memset(&info, 0, sizeof(info)); 828 switch (cmdpass(cmd, pass)) { 829 case cmdpass(RTM_ADD, 1): 830 case cmdpass(RTM_CHANGE, 1): 831 case cmdpass(RTM_DELETE, 2): 832 if (cmd == RTM_ADD) 833 ncmd = RTM_NEWADDR; 834 else 835 ncmd = RTM_DELADDR; 836 837 info.rti_info[RTAX_IFA] = sa = ifa->ifa_addr; 838 info.rti_info[RTAX_IFP] = ifp->if_dl->ifa_addr; 839 info.rti_info[RTAX_NETMASK] = ifa->ifa_netmask; 840 info.rti_info[RTAX_BRD] = ifa->ifa_dstaddr; 841 memset(&ifam, 0, sizeof(ifam)); 842 ifam.ifam_index = ifp->if_index; 843 ifam.ifam_metric = ifa->ifa_metric; 844 ifam.ifam_flags = ifa->ifa_flags; 845 m = rt_msg1(ncmd, &info, &ifam, sizeof(ifam)); 846 if (m == NULL) 847 continue; 848 mtod(m, struct ifa_msghdr *)->ifam_addrs = 849 info.rti_addrs; 850 break; 851 case cmdpass(RTM_ADD, 2): 852 case cmdpass(RTM_CHANGE, 2): 853 case cmdpass(RTM_DELETE, 1): 854 if (rt == NULL) 855 continue; 856 info.rti_info[RTAX_NETMASK] = rt_mask(rt); 857 info.rti_info[RTAX_DST] = sa = rt_getkey(rt); 858 info.rti_info[RTAX_GATEWAY] = rt->rt_gateway; 859 memset(&rtm, 0, sizeof(rtm)); 860 rtm.rtm_index = ifp->if_index; 861 rtm.rtm_flags |= rt->rt_flags; 862 rtm.rtm_errno = error; 863 m = rt_msg1(cmd, &info, &rtm, sizeof(rtm)); 864 if (m == NULL) 865 continue; 866 mtod(m, struct rt_msghdr *)->rtm_addrs = info.rti_addrs; 867 break; 868 default: 869 continue; 870 } 871 #ifdef DIAGNOSTIC 872 if (m == NULL) 873 panic("%s: called with wrong command", __func__); 874 #endif 875 route_enqueue(m, sa ? sa->sa_family : 0); 876 } 877 #undef cmdpass 878 } 879 880 static struct mbuf * 881 rt_makeifannouncemsg(struct ifnet *ifp, int type, int what, 882 struct rt_addrinfo *info) 883 { 884 struct if_announcemsghdr ifan; 885 886 memset(info, 0, sizeof(*info)); 887 memset(&ifan, 0, sizeof(ifan)); 888 ifan.ifan_index = ifp->if_index; 889 strlcpy(ifan.ifan_name, ifp->if_xname, sizeof(ifan.ifan_name)); 890 ifan.ifan_what = what; 891 return rt_msg1(type, info, &ifan, sizeof(ifan)); 892 } 893 894 /* 895 * This is called to generate routing socket messages indicating 896 * network interface arrival and departure. 897 */ 898 void 899 rt_ifannouncemsg(struct ifnet *ifp, int what) 900 { 901 struct mbuf *m; 902 struct rt_addrinfo info; 903 904 if (route_cb.any_count == 0) 905 return; 906 m = rt_makeifannouncemsg(ifp, RTM_IFANNOUNCE, what, &info); 907 if (m == NULL) 908 return; 909 route_enqueue(m, 0); 910 } 911 912 /* 913 * This is called to generate routing socket messages indicating 914 * IEEE80211 wireless events. 915 * XXX we piggyback on the RTM_IFANNOUNCE msg format in a clumsy way. 916 */ 917 void 918 rt_ieee80211msg(struct ifnet *ifp, int what, void *data, size_t data_len) 919 { 920 struct mbuf *m; 921 struct rt_addrinfo info; 922 923 if (route_cb.any_count == 0) 924 return; 925 m = rt_makeifannouncemsg(ifp, RTM_IEEE80211, what, &info); 926 if (m == NULL) 927 return; 928 /* 929 * Append the ieee80211 data. Try to stick it in the 930 * mbuf containing the ifannounce msg; otherwise allocate 931 * a new mbuf and append. 932 * 933 * NB: we assume m is a single mbuf. 934 */ 935 if (data_len > M_TRAILINGSPACE(m)) { 936 struct mbuf *n = m_get(M_NOWAIT, MT_DATA); 937 if (n == NULL) { 938 m_freem(m); 939 return; 940 } 941 (void)memcpy(mtod(n, void *), data, data_len); 942 n->m_len = data_len; 943 m->m_next = n; 944 } else if (data_len > 0) { 945 (void)memcpy(mtod(m, uint8_t *) + m->m_len, data, data_len); 946 m->m_len += data_len; 947 } 948 if (m->m_flags & M_PKTHDR) 949 m->m_pkthdr.len += data_len; 950 mtod(m, struct if_announcemsghdr *)->ifan_msglen += data_len; 951 route_enqueue(m, 0); 952 } 953 954 /* 955 * This is used in dumping the kernel table via sysctl(). 956 */ 957 static int 958 sysctl_dumpentry(struct rtentry *rt, void *v) 959 { 960 struct walkarg *w = v; 961 int error = 0, size; 962 struct rt_addrinfo info; 963 964 if (w->w_op == NET_RT_FLAGS && !(rt->rt_flags & w->w_arg)) 965 return 0; 966 memset(&info, 0, sizeof(info)); 967 info.rti_info[RTAX_DST] = rt_getkey(rt); 968 info.rti_info[RTAX_GATEWAY] = rt->rt_gateway; 969 info.rti_info[RTAX_NETMASK] = rt_mask(rt); 970 if (rt->rt_ifp) { 971 const struct ifaddr *rtifa; 972 info.rti_info[RTAX_IFP] = rt->rt_ifp->if_dl->ifa_addr; 973 /* rtifa used to be simply rt->rt_ifa. If rt->rt_ifa != NULL, 974 * then rt_get_ifa() != NULL. So this ought to still be safe. 975 * --dyoung 976 */ 977 rtifa = rt_get_ifa(rt); 978 info.rti_info[RTAX_IFA] = rtifa->ifa_addr; 979 if (rt->rt_ifp->if_flags & IFF_POINTOPOINT) 980 info.rti_info[RTAX_BRD] = rtifa->ifa_dstaddr; 981 } 982 if ((error = rt_msg2(RTM_GET, &info, 0, w, &size))) 983 return error; 984 if (w->w_where && w->w_tmem && w->w_needed <= 0) { 985 struct rt_msghdr *rtm = (struct rt_msghdr *)w->w_tmem; 986 987 rtm->rtm_flags = rt->rt_flags; 988 rtm->rtm_use = rt->rt_use; 989 rtm->rtm_rmx = rt->rt_rmx; 990 KASSERT(rt->rt_ifp != NULL); 991 rtm->rtm_index = rt->rt_ifp->if_index; 992 rtm->rtm_errno = rtm->rtm_pid = rtm->rtm_seq = 0; 993 rtm->rtm_addrs = info.rti_addrs; 994 if ((error = copyout(rtm, w->w_where, size)) != 0) 995 w->w_where = NULL; 996 else 997 w->w_where = (char *)w->w_where + size; 998 } 999 return error; 1000 } 1001 1002 static int 1003 sysctl_iflist(int af, struct walkarg *w, int type) 1004 { 1005 struct ifnet *ifp; 1006 struct ifaddr *ifa; 1007 struct rt_addrinfo info; 1008 int len, error = 0; 1009 1010 memset(&info, 0, sizeof(info)); 1011 IFNET_FOREACH(ifp) { 1012 if (w->w_arg && w->w_arg != ifp->if_index) 1013 continue; 1014 if (IFADDR_EMPTY(ifp)) 1015 continue; 1016 info.rti_info[RTAX_IFP] = ifp->if_dl->ifa_addr; 1017 switch (type) { 1018 case NET_RT_IFLIST: 1019 error = rt_msg2(RTM_IFINFO, &info, NULL, w, &len); 1020 break; 1021 #ifdef COMPAT_14 1022 case NET_RT_OIFLIST: 1023 error = rt_msg2(RTM_OIFINFO, &info, NULL, w, &len); 1024 break; 1025 #endif 1026 default: 1027 panic("sysctl_iflist(1)"); 1028 } 1029 if (error) 1030 return error; 1031 info.rti_info[RTAX_IFP] = NULL; 1032 if (w->w_where && w->w_tmem && w->w_needed <= 0) { 1033 switch (type) { 1034 case NET_RT_IFLIST: { 1035 struct if_msghdr *ifm; 1036 1037 ifm = (struct if_msghdr *)w->w_tmem; 1038 ifm->ifm_index = ifp->if_index; 1039 ifm->ifm_flags = ifp->if_flags; 1040 ifm->ifm_data = ifp->if_data; 1041 ifm->ifm_addrs = info.rti_addrs; 1042 error = copyout(ifm, w->w_where, len); 1043 if (error) 1044 return error; 1045 w->w_where = (char *)w->w_where + len; 1046 break; 1047 } 1048 1049 #ifdef COMPAT_14 1050 case NET_RT_OIFLIST: { 1051 struct if_msghdr14 *ifm; 1052 1053 ifm = (struct if_msghdr14 *)w->w_tmem; 1054 ifm->ifm_index = ifp->if_index; 1055 ifm->ifm_flags = ifp->if_flags; 1056 ifm->ifm_data.ifi_type = ifp->if_data.ifi_type; 1057 ifm->ifm_data.ifi_addrlen = 1058 ifp->if_data.ifi_addrlen; 1059 ifm->ifm_data.ifi_hdrlen = 1060 ifp->if_data.ifi_hdrlen; 1061 ifm->ifm_data.ifi_mtu = ifp->if_data.ifi_mtu; 1062 ifm->ifm_data.ifi_metric = 1063 ifp->if_data.ifi_metric; 1064 ifm->ifm_data.ifi_baudrate = 1065 ifp->if_data.ifi_baudrate; 1066 ifm->ifm_data.ifi_ipackets = 1067 ifp->if_data.ifi_ipackets; 1068 ifm->ifm_data.ifi_ierrors = 1069 ifp->if_data.ifi_ierrors; 1070 ifm->ifm_data.ifi_opackets = 1071 ifp->if_data.ifi_opackets; 1072 ifm->ifm_data.ifi_oerrors = 1073 ifp->if_data.ifi_oerrors; 1074 ifm->ifm_data.ifi_collisions = 1075 ifp->if_data.ifi_collisions; 1076 ifm->ifm_data.ifi_ibytes = 1077 ifp->if_data.ifi_ibytes; 1078 ifm->ifm_data.ifi_obytes = 1079 ifp->if_data.ifi_obytes; 1080 ifm->ifm_data.ifi_imcasts = 1081 ifp->if_data.ifi_imcasts; 1082 ifm->ifm_data.ifi_omcasts = 1083 ifp->if_data.ifi_omcasts; 1084 ifm->ifm_data.ifi_iqdrops = 1085 ifp->if_data.ifi_iqdrops; 1086 ifm->ifm_data.ifi_noproto = 1087 ifp->if_data.ifi_noproto; 1088 ifm->ifm_data.ifi_lastchange = 1089 ifp->if_data.ifi_lastchange; 1090 ifm->ifm_addrs = info.rti_addrs; 1091 error = copyout(ifm, w->w_where, len); 1092 if (error) 1093 return error; 1094 w->w_where = (char *)w->w_where + len; 1095 break; 1096 } 1097 #endif 1098 default: 1099 panic("sysctl_iflist(2)"); 1100 } 1101 } 1102 IFADDR_FOREACH(ifa, ifp) { 1103 if (af && af != ifa->ifa_addr->sa_family) 1104 continue; 1105 info.rti_info[RTAX_IFA] = ifa->ifa_addr; 1106 info.rti_info[RTAX_NETMASK] = ifa->ifa_netmask; 1107 info.rti_info[RTAX_BRD] = ifa->ifa_dstaddr; 1108 if ((error = rt_msg2(RTM_NEWADDR, &info, 0, w, &len))) 1109 return error; 1110 if (w->w_where && w->w_tmem && w->w_needed <= 0) { 1111 struct ifa_msghdr *ifam; 1112 1113 ifam = (struct ifa_msghdr *)w->w_tmem; 1114 ifam->ifam_index = ifa->ifa_ifp->if_index; 1115 ifam->ifam_flags = ifa->ifa_flags; 1116 ifam->ifam_metric = ifa->ifa_metric; 1117 ifam->ifam_addrs = info.rti_addrs; 1118 error = copyout(w->w_tmem, w->w_where, len); 1119 if (error) 1120 return error; 1121 w->w_where = (char *)w->w_where + len; 1122 } 1123 } 1124 info.rti_info[RTAX_IFA] = info.rti_info[RTAX_NETMASK] = 1125 info.rti_info[RTAX_BRD] = NULL; 1126 } 1127 return 0; 1128 } 1129 1130 static int 1131 sysctl_rtable(SYSCTLFN_ARGS) 1132 { 1133 void *where = oldp; 1134 size_t *given = oldlenp; 1135 const void *new = newp; 1136 int i, s, error = EINVAL; 1137 u_char af; 1138 struct walkarg w; 1139 1140 if (namelen == 1 && name[0] == CTL_QUERY) 1141 return sysctl_query(SYSCTLFN_CALL(rnode)); 1142 1143 if (new) 1144 return EPERM; 1145 if (namelen != 3) 1146 return EINVAL; 1147 af = name[0]; 1148 w.w_tmemneeded = 0; 1149 w.w_tmemsize = 0; 1150 w.w_tmem = NULL; 1151 again: 1152 /* we may return here if a later [re]alloc of the t_mem buffer fails */ 1153 if (w.w_tmemneeded) { 1154 w.w_tmem = malloc(w.w_tmemneeded, M_RTABLE, M_WAITOK); 1155 w.w_tmemsize = w.w_tmemneeded; 1156 w.w_tmemneeded = 0; 1157 } 1158 w.w_op = name[1]; 1159 w.w_arg = name[2]; 1160 w.w_given = *given; 1161 w.w_needed = 0 - w.w_given; 1162 w.w_where = where; 1163 1164 s = splsoftnet(); 1165 switch (w.w_op) { 1166 1167 case NET_RT_DUMP: 1168 case NET_RT_FLAGS: 1169 for (i = 1; i <= AF_MAX; i++) 1170 if ((af == 0 || af == i) && 1171 (error = rt_walktree(i, sysctl_dumpentry, &w))) 1172 break; 1173 break; 1174 1175 #ifdef COMPAT_14 1176 case NET_RT_OIFLIST: 1177 error = sysctl_iflist(af, &w, w.w_op); 1178 break; 1179 #endif 1180 1181 case NET_RT_IFLIST: 1182 error = sysctl_iflist(af, &w, w.w_op); 1183 } 1184 splx(s); 1185 1186 /* check to see if we couldn't allocate memory with NOWAIT */ 1187 if (error == ENOBUFS && w.w_tmem == 0 && w.w_tmemneeded) 1188 goto again; 1189 1190 if (w.w_tmem) 1191 free(w.w_tmem, M_RTABLE); 1192 w.w_needed += w.w_given; 1193 if (where) { 1194 *given = (char *)w.w_where - (char *)where; 1195 if (*given < w.w_needed) 1196 return ENOMEM; 1197 } else { 1198 *given = (11 * w.w_needed) / 10; 1199 } 1200 return error; 1201 } 1202 1203 /* 1204 * Routing message software interrupt routine 1205 */ 1206 static void 1207 route_intr(void *cookie) 1208 { 1209 struct sockproto proto = { .sp_family = PF_ROUTE, }; 1210 struct mbuf *m; 1211 int s; 1212 1213 mutex_enter(softnet_lock); 1214 KERNEL_LOCK(1, NULL); 1215 while (!IF_IS_EMPTY(&route_intrq)) { 1216 s = splnet(); 1217 IF_DEQUEUE(&route_intrq, m); 1218 splx(s); 1219 if (m == NULL) 1220 break; 1221 proto.sp_protocol = M_GETCTX(m, uintptr_t); 1222 raw_input(m, &proto, &route_src, &route_dst); 1223 } 1224 KERNEL_UNLOCK_ONE(NULL); 1225 mutex_exit(softnet_lock); 1226 } 1227 1228 /* 1229 * Enqueue a message to the software interrupt routine. 1230 */ 1231 static void 1232 route_enqueue(struct mbuf *m, int family) 1233 { 1234 int s, wasempty; 1235 1236 s = splnet(); 1237 if (IF_QFULL(&route_intrq)) { 1238 IF_DROP(&route_intrq); 1239 m_freem(m); 1240 } else { 1241 wasempty = IF_IS_EMPTY(&route_intrq); 1242 M_SETCTX(m, (uintptr_t)family); 1243 IF_ENQUEUE(&route_intrq, m); 1244 if (wasempty) 1245 softint_schedule(route_sih); 1246 } 1247 splx(s); 1248 } 1249 1250 void 1251 rt_init(void) 1252 { 1253 1254 route_intrq.ifq_maxlen = route_maxqlen; 1255 route_sih = softint_establish(SOFTINT_NET | SOFTINT_MPSAFE, 1256 route_intr, NULL); 1257 } 1258 1259 /* 1260 * Definitions of protocols supported in the ROUTE domain. 1261 */ 1262 PR_WRAP_USRREQ(route_usrreq) 1263 #define route_usrreq route_usrreq_wrapper 1264 1265 const struct protosw routesw[] = { 1266 { 1267 .pr_type = SOCK_RAW, 1268 .pr_domain = &routedomain, 1269 .pr_flags = PR_ATOMIC|PR_ADDR, 1270 .pr_input = raw_input, 1271 .pr_output = route_output, 1272 .pr_ctlinput = raw_ctlinput, 1273 .pr_usrreq = route_usrreq, 1274 .pr_init = raw_init, 1275 }, 1276 }; 1277 1278 struct domain routedomain = { 1279 .dom_family = PF_ROUTE, 1280 .dom_name = "route", 1281 .dom_init = route_init, 1282 .dom_protosw = routesw, 1283 .dom_protoswNPROTOSW = &routesw[__arraycount(routesw)], 1284 }; 1285 1286 SYSCTL_SETUP(sysctl_net_route_setup, "sysctl net.route subtree setup") 1287 { 1288 const struct sysctlnode *rnode = NULL; 1289 1290 sysctl_createv(clog, 0, NULL, NULL, 1291 CTLFLAG_PERMANENT, 1292 CTLTYPE_NODE, "net", NULL, 1293 NULL, 0, NULL, 0, 1294 CTL_NET, CTL_EOL); 1295 1296 sysctl_createv(clog, 0, NULL, &rnode, 1297 CTLFLAG_PERMANENT, 1298 CTLTYPE_NODE, "route", 1299 SYSCTL_DESCR("PF_ROUTE information"), 1300 NULL, 0, NULL, 0, 1301 CTL_NET, PF_ROUTE, CTL_EOL); 1302 sysctl_createv(clog, 0, NULL, NULL, 1303 CTLFLAG_PERMANENT, 1304 CTLTYPE_NODE, "rtable", 1305 SYSCTL_DESCR("Routing table information"), 1306 sysctl_rtable, 0, NULL, 0, 1307 CTL_NET, PF_ROUTE, 0 /* any protocol */, CTL_EOL); 1308 sysctl_createv(clog, 0, &rnode, NULL, 1309 CTLFLAG_PERMANENT, 1310 CTLTYPE_STRUCT, "stats", 1311 SYSCTL_DESCR("Routing statistics"), 1312 NULL, 0, &rtstat, sizeof(rtstat), 1313 CTL_CREATE, CTL_EOL); 1314 } 1315
Indexes created Wed Sep 24 15:09:51 GMT 2025