rtsock.c revision 1.216
1 /* $NetBSD: rtsock.c,v 1.216 2017/06/16 02:26:17 ozaki-r 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 1.216 2017/06/16 02:26:17 ozaki-r Exp $"); 65 66 #ifdef _KERNEL_OPT 67 #include "opt_inet.h" 68 #include "opt_mpls.h" 69 #include "opt_compat_netbsd.h" 70 #include "opt_sctp.h" 71 #include "opt_net_mpsafe.h" 72 #endif 73 74 #include <sys/param.h> 75 #include <sys/systm.h> 76 #include <sys/proc.h> 77 #include <sys/socket.h> 78 #include <sys/socketvar.h> 79 #include <sys/domain.h> 80 #include <sys/protosw.h> 81 #include <sys/sysctl.h> 82 #include <sys/kauth.h> 83 #include <sys/kmem.h> 84 #include <sys/intr.h> 85 86 #include <net/if.h> 87 #include <net/if_llatbl.h> 88 #include <net/if_types.h> 89 #include <net/route.h> 90 #include <net/raw_cb.h> 91 92 #include <netinet/in_var.h> 93 #include <netinet/if_inarp.h> 94 95 #include <netmpls/mpls.h> 96 97 #ifdef SCTP 98 extern void sctp_add_ip_address(struct ifaddr *); 99 extern void sctp_delete_ip_address(struct ifaddr *); 100 #endif 101 102 #if defined(COMPAT_14) || defined(COMPAT_50) || defined(COMPAT_70) 103 #include <compat/net/if.h> 104 #include <compat/net/route.h> 105 #endif 106 #ifdef COMPAT_RTSOCK 107 #define RTM_XVERSION RTM_OVERSION 108 #define RTM_XNEWADDR RTM_ONEWADDR 109 #define RTM_XDELADDR RTM_ODELADDR 110 #define RTM_XCHGADDR RTM_OCHGADDR 111 #define RT_XADVANCE(a,b) RT_OADVANCE(a,b) 112 #define RT_XROUNDUP(n) RT_OROUNDUP(n) 113 #define PF_XROUTE PF_OROUTE 114 #define rt_xmsghdr rt_msghdr50 115 #define if_xmsghdr if_msghdr /* if_msghdr50 is for RTM_OIFINFO */ 116 #define ifa_xmsghdr ifa_msghdr50 117 #define if_xannouncemsghdr if_announcemsghdr50 118 #define COMPATNAME(x) compat_50_ ## x 119 #define DOMAINNAME "oroute" 120 CTASSERT(sizeof(struct ifa_xmsghdr) == 20); 121 DOMAIN_DEFINE(compat_50_routedomain); /* forward declare and add to link set */ 122 #undef COMPAT_70 123 #else /* COMPAT_RTSOCK */ 124 #define RTM_XVERSION RTM_VERSION 125 #define RTM_XNEWADDR RTM_NEWADDR 126 #define RTM_XDELADDR RTM_DELADDR 127 #define RTM_XCHGADDR RTM_CHGADDR 128 #define RT_XADVANCE(a,b) RT_ADVANCE(a,b) 129 #define RT_XROUNDUP(n) RT_ROUNDUP(n) 130 #define PF_XROUTE PF_ROUTE 131 #define rt_xmsghdr rt_msghdr 132 #define if_xmsghdr if_msghdr 133 #define ifa_xmsghdr ifa_msghdr 134 #define if_xannouncemsghdr if_announcemsghdr 135 #define COMPATNAME(x) x 136 #define DOMAINNAME "route" 137 CTASSERT(sizeof(struct ifa_xmsghdr) == 32); 138 #ifdef COMPAT_50 139 #define COMPATCALL(name, args) compat_50_ ## name args 140 #endif 141 DOMAIN_DEFINE(routedomain); /* forward declare and add to link set */ 142 #undef COMPAT_50 143 #undef COMPAT_14 144 #endif /* COMPAT_RTSOCK */ 145 146 #ifndef COMPATCALL 147 #define COMPATCALL(name, args) do { } while (/*CONSTCOND*/ 0) 148 #endif 149 150 #ifdef RTSOCK_DEBUG 151 #define RT_IN_PRINT(info, b, a) (in_print((b), sizeof(b), \ 152 &((const struct sockaddr_in *)(info)->rti_info[(a)])->sin_addr), (b)) 153 #endif /* RTSOCK_DEBUG */ 154 155 struct route_info COMPATNAME(route_info) = { 156 .ri_dst = { .sa_len = 2, .sa_family = PF_XROUTE, }, 157 .ri_src = { .sa_len = 2, .sa_family = PF_XROUTE, }, 158 .ri_maxqlen = IFQ_MAXLEN, 159 }; 160 161 #define PRESERVED_RTF (RTF_UP | RTF_GATEWAY | RTF_HOST | RTF_DONE | RTF_MASK) 162 163 static void COMPATNAME(route_init)(void); 164 static int COMPATNAME(route_output)(struct mbuf *, struct socket *); 165 166 static int rt_xaddrs(u_char, const char *, const char *, struct rt_addrinfo *); 167 static struct mbuf *rt_makeifannouncemsg(struct ifnet *, int, int, 168 struct rt_addrinfo *); 169 static int rt_msg2(int, struct rt_addrinfo *, void *, struct rt_walkarg *, int *); 170 static void rt_setmetrics(int, const struct rt_xmsghdr *, struct rtentry *); 171 static void rtm_setmetrics(const struct rtentry *, struct rt_xmsghdr *); 172 static void sysctl_net_route_setup(struct sysctllog **); 173 static int sysctl_dumpentry(struct rtentry *, void *); 174 static int sysctl_iflist(int, struct rt_walkarg *, int); 175 static int sysctl_rtable(SYSCTLFN_PROTO); 176 static void rt_adjustcount(int, int); 177 178 static const struct protosw COMPATNAME(route_protosw)[]; 179 180 struct routecb { 181 struct rawcb rocb_rcb; 182 unsigned int rocb_msgfilter; 183 #define RTMSGFILTER(m) (1U << (m)) 184 }; 185 #define sotoroutecb(so) ((struct routecb *)(so)->so_pcb) 186 187 static void 188 rt_adjustcount(int af, int cnt) 189 { 190 struct route_cb * const cb = &COMPATNAME(route_info).ri_cb; 191 192 cb->any_count += cnt; 193 194 switch (af) { 195 case AF_INET: 196 cb->ip_count += cnt; 197 return; 198 #ifdef INET6 199 case AF_INET6: 200 cb->ip6_count += cnt; 201 return; 202 #endif 203 case AF_MPLS: 204 cb->mpls_count += cnt; 205 return; 206 } 207 } 208 209 static int 210 COMPATNAME(route_filter)(struct mbuf *m, struct sockproto *proto, 211 struct rawcb *rp) 212 { 213 struct routecb *rop = (struct routecb *)rp; 214 struct rt_xmsghdr *rtm; 215 216 KASSERT(m != NULL); 217 KASSERT(proto != NULL); 218 KASSERT(rp != NULL); 219 220 /* Wrong family for this socket. */ 221 if (proto->sp_family != PF_ROUTE) 222 return ENOPROTOOPT; 223 224 /* If no filter set, just return. */ 225 if (rop->rocb_msgfilter == 0) 226 return 0; 227 228 /* Ensure we can access rtm_type */ 229 if (m->m_len < 230 offsetof(struct rt_xmsghdr, rtm_type) + sizeof(rtm->rtm_type)) 231 return EINVAL; 232 233 rtm = mtod(m, struct rt_xmsghdr *); 234 /* If the rtm type is filtered out, return a positive. */ 235 if (!(rop->rocb_msgfilter & RTMSGFILTER(rtm->rtm_type))) 236 return EEXIST; 237 238 /* Passed the filter. */ 239 return 0; 240 } 241 242 static int 243 COMPATNAME(route_attach)(struct socket *so, int proto) 244 { 245 struct rawcb *rp; 246 struct routecb *rop; 247 int s, error; 248 249 KASSERT(sotorawcb(so) == NULL); 250 rop = kmem_zalloc(sizeof(*rop), KM_SLEEP); 251 rp = &rop->rocb_rcb; 252 rp->rcb_len = sizeof(*rop); 253 so->so_pcb = rp; 254 255 s = splsoftnet(); 256 if ((error = raw_attach(so, proto)) == 0) { 257 rt_adjustcount(rp->rcb_proto.sp_protocol, 1); 258 rp->rcb_laddr = &COMPATNAME(route_info).ri_src; 259 rp->rcb_faddr = &COMPATNAME(route_info).ri_dst; 260 rp->rcb_filter = COMPATNAME(route_filter); 261 } 262 splx(s); 263 264 if (error) { 265 kmem_free(rop, sizeof(*rop)); 266 so->so_pcb = NULL; 267 return error; 268 } 269 270 soisconnected(so); 271 so->so_options |= SO_USELOOPBACK; 272 KASSERT(solocked(so)); 273 274 return error; 275 } 276 277 static void 278 COMPATNAME(route_detach)(struct socket *so) 279 { 280 struct rawcb *rp = sotorawcb(so); 281 int s; 282 283 KASSERT(rp != NULL); 284 KASSERT(solocked(so)); 285 286 s = splsoftnet(); 287 rt_adjustcount(rp->rcb_proto.sp_protocol, -1); 288 raw_detach(so); 289 splx(s); 290 } 291 292 static int 293 COMPATNAME(route_accept)(struct socket *so, struct sockaddr *nam) 294 { 295 KASSERT(solocked(so)); 296 297 panic("route_accept"); 298 299 return EOPNOTSUPP; 300 } 301 302 static int 303 COMPATNAME(route_bind)(struct socket *so, struct sockaddr *nam, struct lwp *l) 304 { 305 KASSERT(solocked(so)); 306 307 return EOPNOTSUPP; 308 } 309 310 static int 311 COMPATNAME(route_listen)(struct socket *so, struct lwp *l) 312 { 313 KASSERT(solocked(so)); 314 315 return EOPNOTSUPP; 316 } 317 318 static int 319 COMPATNAME(route_connect)(struct socket *so, struct sockaddr *nam, struct lwp *l) 320 { 321 KASSERT(solocked(so)); 322 323 return EOPNOTSUPP; 324 } 325 326 static int 327 COMPATNAME(route_connect2)(struct socket *so, struct socket *so2) 328 { 329 KASSERT(solocked(so)); 330 331 return EOPNOTSUPP; 332 } 333 334 static int 335 COMPATNAME(route_disconnect)(struct socket *so) 336 { 337 struct rawcb *rp = sotorawcb(so); 338 int s; 339 340 KASSERT(solocked(so)); 341 KASSERT(rp != NULL); 342 343 s = splsoftnet(); 344 soisdisconnected(so); 345 raw_disconnect(rp); 346 splx(s); 347 348 return 0; 349 } 350 351 static int 352 COMPATNAME(route_shutdown)(struct socket *so) 353 { 354 int s; 355 356 KASSERT(solocked(so)); 357 358 /* 359 * Mark the connection as being incapable of further input. 360 */ 361 s = splsoftnet(); 362 socantsendmore(so); 363 splx(s); 364 return 0; 365 } 366 367 static int 368 COMPATNAME(route_abort)(struct socket *so) 369 { 370 KASSERT(solocked(so)); 371 372 panic("route_abort"); 373 374 return EOPNOTSUPP; 375 } 376 377 static int 378 COMPATNAME(route_ioctl)(struct socket *so, u_long cmd, void *nam, 379 struct ifnet * ifp) 380 { 381 return EOPNOTSUPP; 382 } 383 384 static int 385 COMPATNAME(route_stat)(struct socket *so, struct stat *ub) 386 { 387 KASSERT(solocked(so)); 388 389 return 0; 390 } 391 392 static int 393 COMPATNAME(route_peeraddr)(struct socket *so, struct sockaddr *nam) 394 { 395 struct rawcb *rp = sotorawcb(so); 396 397 KASSERT(solocked(so)); 398 KASSERT(rp != NULL); 399 KASSERT(nam != NULL); 400 401 if (rp->rcb_faddr == NULL) 402 return ENOTCONN; 403 404 raw_setpeeraddr(rp, nam); 405 return 0; 406 } 407 408 static int 409 COMPATNAME(route_sockaddr)(struct socket *so, struct sockaddr *nam) 410 { 411 struct rawcb *rp = sotorawcb(so); 412 413 KASSERT(solocked(so)); 414 KASSERT(rp != NULL); 415 KASSERT(nam != NULL); 416 417 if (rp->rcb_faddr == NULL) 418 return ENOTCONN; 419 420 raw_setsockaddr(rp, nam); 421 return 0; 422 } 423 424 static int 425 COMPATNAME(route_rcvd)(struct socket *so, int flags, struct lwp *l) 426 { 427 KASSERT(solocked(so)); 428 429 return EOPNOTSUPP; 430 } 431 432 static int 433 COMPATNAME(route_recvoob)(struct socket *so, struct mbuf *m, int flags) 434 { 435 KASSERT(solocked(so)); 436 437 return EOPNOTSUPP; 438 } 439 440 static int 441 COMPATNAME(route_send)(struct socket *so, struct mbuf *m, 442 struct sockaddr *nam, struct mbuf *control, struct lwp *l) 443 { 444 int error = 0; 445 int s; 446 447 KASSERT(solocked(so)); 448 KASSERT(so->so_proto == &COMPATNAME(route_protosw)[0]); 449 450 s = splsoftnet(); 451 error = raw_send(so, m, nam, control, l, &COMPATNAME(route_output)); 452 splx(s); 453 454 return error; 455 } 456 457 static int 458 COMPATNAME(route_sendoob)(struct socket *so, struct mbuf *m, 459 struct mbuf *control) 460 { 461 KASSERT(solocked(so)); 462 463 m_freem(m); 464 m_freem(control); 465 466 return EOPNOTSUPP; 467 } 468 static int 469 COMPATNAME(route_purgeif)(struct socket *so, struct ifnet *ifp) 470 { 471 472 panic("route_purgeif"); 473 474 return EOPNOTSUPP; 475 } 476 477 #ifdef INET 478 static int 479 route_get_sdl_index(struct rt_addrinfo *info, int *sdl_index) 480 { 481 struct rtentry *nrt; 482 int error; 483 484 error = rtrequest1(RTM_GET, info, &nrt); 485 if (error != 0) 486 return error; 487 /* 488 * nrt->rt_ifp->if_index may not be correct 489 * due to changing to ifplo0. 490 */ 491 *sdl_index = satosdl(nrt->rt_gateway)->sdl_index; 492 rt_unref(nrt); 493 494 return 0; 495 } 496 #endif /* INET */ 497 498 static void 499 route_get_sdl(const struct ifnet *ifp, const struct sockaddr *dst, 500 struct sockaddr_dl *sdl, int *flags) 501 { 502 struct llentry *la; 503 504 KASSERT(ifp != NULL); 505 506 IF_AFDATA_RLOCK(ifp); 507 switch (dst->sa_family) { 508 case AF_INET: 509 la = lla_lookup(LLTABLE(ifp), 0, dst); 510 break; 511 case AF_INET6: 512 la = lla_lookup(LLTABLE6(ifp), 0, dst); 513 break; 514 default: 515 la = NULL; 516 KASSERTMSG(0, "Invalid AF=%d\n", dst->sa_family); 517 break; 518 } 519 IF_AFDATA_RUNLOCK(ifp); 520 521 void *a = (LLE_IS_VALID(la) && (la->la_flags & LLE_VALID) == LLE_VALID) 522 ? &la->ll_addr : NULL; 523 524 a = sockaddr_dl_init(sdl, sizeof(*sdl), ifp->if_index, ifp->if_type, 525 NULL, 0, a, ifp->if_addrlen); 526 KASSERT(a != NULL); 527 528 if (la != NULL) { 529 *flags = la->la_flags; 530 LLE_RUNLOCK(la); 531 } 532 } 533 534 static int 535 route_output_report(struct rtentry *rt, struct rt_addrinfo *info, 536 struct rt_xmsghdr *rtm, struct rt_xmsghdr **new_rtm) 537 { 538 int len; 539 540 if (rtm->rtm_addrs & (RTA_IFP | RTA_IFA)) { 541 const struct ifaddr *rtifa; 542 const struct ifnet *ifp = rt->rt_ifp; 543 544 info->rti_info[RTAX_IFP] = ifp->if_dl->ifa_addr; 545 /* rtifa used to be simply rt->rt_ifa. 546 * If rt->rt_ifa != NULL, then 547 * rt_get_ifa() != NULL. So this 548 * ought to still be safe. --dyoung 549 */ 550 rtifa = rt_get_ifa(rt); 551 info->rti_info[RTAX_IFA] = rtifa->ifa_addr; 552 #ifdef RTSOCK_DEBUG 553 if (info->rti_info[RTAX_IFA]->sa_family == AF_INET) { 554 char ibuf[INET_ADDRSTRLEN]; 555 char abuf[INET_ADDRSTRLEN]; 556 printf("%s: copying out RTAX_IFA %s " 557 "for info->rti_info[RTAX_DST] %s " 558 "ifa_getifa %p ifa_seqno %p\n", 559 __func__, 560 RT_IN_PRINT(info, ibuf, RTAX_IFA), 561 RT_IN_PRINT(info, abuf, RTAX_DST), 562 (void *)rtifa->ifa_getifa, 563 rtifa->ifa_seqno); 564 } 565 #endif /* RTSOCK_DEBUG */ 566 if (ifp->if_flags & IFF_POINTOPOINT) 567 info->rti_info[RTAX_BRD] = rtifa->ifa_dstaddr; 568 else 569 info->rti_info[RTAX_BRD] = NULL; 570 rtm->rtm_index = ifp->if_index; 571 } 572 (void)rt_msg2(rtm->rtm_type, info, NULL, NULL, &len); 573 if (len > rtm->rtm_msglen) { 574 struct rt_xmsghdr *old_rtm = rtm; 575 R_Malloc(*new_rtm, struct rt_xmsghdr *, len); 576 if (*new_rtm == NULL) 577 return ENOBUFS; 578 (void)memcpy(*new_rtm, old_rtm, old_rtm->rtm_msglen); 579 rtm = *new_rtm; 580 } 581 (void)rt_msg2(rtm->rtm_type, info, rtm, NULL, 0); 582 rtm->rtm_flags = rt->rt_flags; 583 rtm_setmetrics(rt, rtm); 584 rtm->rtm_addrs = info->rti_addrs; 585 586 return 0; 587 } 588 589 static struct ifaddr * 590 route_output_get_ifa(const struct rt_addrinfo info, const struct rtentry *rt, 591 struct ifnet **ifp, struct psref *psref) 592 { 593 struct ifaddr *ifa = NULL; 594 595 *ifp = NULL; 596 if (info.rti_info[RTAX_IFP] != NULL) { 597 ifa = ifa_ifwithnet_psref(info.rti_info[RTAX_IFP], psref); 598 if (ifa == NULL) 599 goto next; 600 *ifp = ifa->ifa_ifp; 601 if (info.rti_info[RTAX_IFA] == NULL && 602 info.rti_info[RTAX_GATEWAY] == NULL) 603 goto next; 604 if (info.rti_info[RTAX_IFA] == NULL) { 605 /* route change <dst> <gw> -ifp <if> */ 606 ifa = ifaof_ifpforaddr_psref(info.rti_info[RTAX_GATEWAY], 607 *ifp, psref); 608 } else { 609 /* route change <dst> -ifp <if> -ifa <addr> */ 610 ifa = ifa_ifwithaddr_psref(info.rti_info[RTAX_IFA], psref); 611 if (ifa != NULL) 612 goto out; 613 ifa = ifaof_ifpforaddr_psref(info.rti_info[RTAX_IFA], 614 *ifp, psref); 615 } 616 goto out; 617 } 618 next: 619 if (info.rti_info[RTAX_IFA] != NULL) { 620 /* route change <dst> <gw> -ifa <addr> */ 621 ifa = ifa_ifwithaddr_psref(info.rti_info[RTAX_IFA], psref); 622 if (ifa != NULL) 623 goto out; 624 } 625 if (info.rti_info[RTAX_GATEWAY] != NULL) { 626 /* route change <dst> <gw> */ 627 ifa = ifa_ifwithroute_psref(rt->rt_flags, rt_getkey(rt), 628 info.rti_info[RTAX_GATEWAY], psref); 629 } 630 out: 631 if (ifa != NULL && *ifp == NULL) 632 *ifp = ifa->ifa_ifp; 633 return ifa; 634 } 635 636 static int 637 route_output_change(struct rtentry *rt, struct rt_addrinfo *info, 638 struct rt_xmsghdr *rtm) 639 { 640 int error = 0; 641 struct ifnet *ifp = NULL, *new_ifp; 642 struct ifaddr *ifa = NULL, *new_ifa; 643 struct psref psref_ifa, psref_new_ifa, psref_ifp; 644 bool newgw; 645 646 /* 647 * New gateway could require new ifaddr, ifp; 648 * flags may also be different; ifp may be specified 649 * by ll sockaddr when protocol address is ambiguous 650 */ 651 newgw = info->rti_info[RTAX_GATEWAY] != NULL && 652 sockaddr_cmp(info->rti_info[RTAX_GATEWAY], rt->rt_gateway) != 0; 653 654 if (newgw || info->rti_info[RTAX_IFP] != NULL || 655 info->rti_info[RTAX_IFA] != NULL) { 656 ifp = rt_getifp(info, &psref_ifp); 657 ifa = rt_getifa(info, &psref_ifa); 658 if (ifa == NULL) { 659 error = ENETUNREACH; 660 goto out; 661 } 662 } 663 if (newgw) { 664 error = rt_setgate(rt, info->rti_info[RTAX_GATEWAY]); 665 if (error != 0) 666 goto out; 667 } 668 if (info->rti_info[RTAX_TAG]) { 669 const struct sockaddr *tag; 670 tag = rt_settag(rt, info->rti_info[RTAX_TAG]); 671 if (tag == NULL) { 672 error = ENOBUFS; 673 goto out; 674 } 675 } 676 /* 677 * New gateway could require new ifaddr, ifp; 678 * flags may also be different; ifp may be specified 679 * by ll sockaddr when protocol address is ambiguous 680 */ 681 new_ifa = route_output_get_ifa(*info, rt, &new_ifp, &psref_new_ifa); 682 if (new_ifa != NULL) { 683 ifa_release(ifa, &psref_ifa); 684 ifa = new_ifa; 685 } 686 if (ifa) { 687 struct ifaddr *oifa = rt->rt_ifa; 688 if (oifa != ifa && !ifa_is_destroying(ifa) && 689 new_ifp != NULL && !if_is_deactivated(new_ifp)) { 690 if (oifa && oifa->ifa_rtrequest) 691 oifa->ifa_rtrequest(RTM_DELETE, rt, info); 692 rt_replace_ifa(rt, ifa); 693 rt->rt_ifp = new_ifp; 694 } 695 if (new_ifa == NULL) 696 ifa_release(ifa, &psref_ifa); 697 } 698 ifa_release(new_ifa, &psref_new_ifa); 699 if (new_ifp && rt->rt_ifp != new_ifp && !if_is_deactivated(new_ifp)) 700 rt->rt_ifp = new_ifp; 701 rt_setmetrics(rtm->rtm_inits, rtm, rt); 702 if (rt->rt_flags != info->rti_flags) { 703 rt->rt_flags = (info->rti_flags & ~PRESERVED_RTF) | 704 (rt->rt_flags & PRESERVED_RTF); 705 } 706 if (rt->rt_ifa && rt->rt_ifa->ifa_rtrequest) 707 rt->rt_ifa->ifa_rtrequest(RTM_ADD, rt, info); 708 out: 709 if_put(ifp, &psref_ifp); 710 711 return error; 712 } 713 714 /*ARGSUSED*/ 715 int 716 COMPATNAME(route_output)(struct mbuf *m, struct socket *so) 717 { 718 struct sockproto proto = { .sp_family = PF_XROUTE, }; 719 struct rt_xmsghdr *rtm = NULL; 720 struct rt_xmsghdr *old_rtm = NULL, *new_rtm = NULL; 721 struct rtentry *rt = NULL; 722 struct rtentry *saved_nrt = NULL; 723 struct rt_addrinfo info; 724 int len, error = 0; 725 sa_family_t family; 726 struct sockaddr_dl sdl; 727 int bound = curlwp_bind(); 728 bool do_rt_free = false; 729 730 #define senderr(e) do { error = e; goto flush;} while (/*CONSTCOND*/ 0) 731 if (m == NULL || ((m->m_len < sizeof(int32_t)) && 732 (m = m_pullup(m, sizeof(int32_t))) == NULL)) { 733 error = ENOBUFS; 734 goto out; 735 } 736 if ((m->m_flags & M_PKTHDR) == 0) 737 panic("%s", __func__); 738 len = m->m_pkthdr.len; 739 if (len < sizeof(*rtm) || 740 len != mtod(m, struct rt_xmsghdr *)->rtm_msglen) { 741 info.rti_info[RTAX_DST] = NULL; 742 senderr(EINVAL); 743 } 744 R_Malloc(rtm, struct rt_xmsghdr *, len); 745 if (rtm == NULL) { 746 info.rti_info[RTAX_DST] = NULL; 747 senderr(ENOBUFS); 748 } 749 m_copydata(m, 0, len, rtm); 750 if (rtm->rtm_version != RTM_XVERSION) { 751 info.rti_info[RTAX_DST] = NULL; 752 senderr(EPROTONOSUPPORT); 753 } 754 rtm->rtm_pid = curproc->p_pid; 755 memset(&info, 0, sizeof(info)); 756 info.rti_addrs = rtm->rtm_addrs; 757 if (rt_xaddrs(rtm->rtm_type, (const char *)(rtm + 1), len + (char *)rtm, 758 &info)) { 759 senderr(EINVAL); 760 } 761 info.rti_flags = rtm->rtm_flags; 762 #ifdef RTSOCK_DEBUG 763 if (info.rti_info[RTAX_DST]->sa_family == AF_INET) { 764 char abuf[INET_ADDRSTRLEN]; 765 printf("%s: extracted info.rti_info[RTAX_DST] %s\n", __func__, 766 RT_IN_PRINT(&info, abuf, RTAX_DST)); 767 } 768 #endif /* RTSOCK_DEBUG */ 769 if (info.rti_info[RTAX_DST] == NULL || 770 (info.rti_info[RTAX_DST]->sa_family >= AF_MAX)) { 771 senderr(EINVAL); 772 } 773 if (info.rti_info[RTAX_GATEWAY] != NULL && 774 (info.rti_info[RTAX_GATEWAY]->sa_family >= AF_MAX)) { 775 senderr(EINVAL); 776 } 777 778 /* 779 * Verify that the caller has the appropriate privilege; RTM_GET 780 * is the only operation the non-superuser is allowed. 781 */ 782 if (kauth_authorize_network(curlwp->l_cred, KAUTH_NETWORK_ROUTE, 783 0, rtm, NULL, NULL) != 0) 784 senderr(EACCES); 785 786 switch (rtm->rtm_type) { 787 788 case RTM_ADD: 789 if (info.rti_info[RTAX_GATEWAY] == NULL) { 790 senderr(EINVAL); 791 } 792 #ifdef INET 793 /* support for new ARP code with keeping backcompat */ 794 if (info.rti_info[RTAX_GATEWAY]->sa_family == AF_LINK) { 795 const struct sockaddr_dl *sdlp = 796 satocsdl(info.rti_info[RTAX_GATEWAY]); 797 798 /* Allow routing requests by interface index */ 799 if (sdlp->sdl_nlen == 0 && sdlp->sdl_alen == 0 800 && sdlp->sdl_slen == 0) 801 goto fallback; 802 /* 803 * Old arp binaries don't set the sdl_index 804 * so we have to complement it. 805 */ 806 int sdl_index = sdlp->sdl_index; 807 if (sdl_index == 0) { 808 error = route_get_sdl_index(&info, &sdl_index); 809 if (error != 0) 810 goto fallback; 811 } else if ( 812 info.rti_info[RTAX_DST]->sa_family == AF_INET) { 813 /* 814 * XXX workaround for SIN_PROXY case; proxy arp 815 * entry should be in an interface that has 816 * a network route including the destination, 817 * not a local (link) route that may not be a 818 * desired place, for example a tap. 819 */ 820 const struct sockaddr_inarp *sina = 821 (const struct sockaddr_inarp *) 822 info.rti_info[RTAX_DST]; 823 if (sina->sin_other & SIN_PROXY) { 824 error = route_get_sdl_index(&info, 825 &sdl_index); 826 if (error != 0) 827 goto fallback; 828 } 829 } 830 error = lla_rt_output(rtm->rtm_type, rtm->rtm_flags, 831 rtm->rtm_rmx.rmx_expire, &info, sdl_index); 832 break; 833 } 834 fallback: 835 #endif /* INET */ 836 error = rtrequest1(rtm->rtm_type, &info, &saved_nrt); 837 if (error == 0) { 838 rt_setmetrics(rtm->rtm_inits, rtm, saved_nrt); 839 rt_unref(saved_nrt); 840 } 841 break; 842 843 case RTM_DELETE: 844 #ifdef INET 845 /* support for new ARP code */ 846 if (info.rti_info[RTAX_GATEWAY] && 847 (info.rti_info[RTAX_GATEWAY]->sa_family == AF_LINK) && 848 (rtm->rtm_flags & RTF_LLDATA) != 0) { 849 error = lla_rt_output(rtm->rtm_type, rtm->rtm_flags, 850 rtm->rtm_rmx.rmx_expire, &info, 0); 851 break; 852 } 853 #endif /* INET */ 854 error = rtrequest1(rtm->rtm_type, &info, &saved_nrt); 855 if (error != 0) 856 break; 857 858 rt = saved_nrt; 859 do_rt_free = true; 860 info.rti_info[RTAX_DST] = rt_getkey(rt); 861 info.rti_info[RTAX_GATEWAY] = rt->rt_gateway; 862 info.rti_info[RTAX_NETMASK] = rt_mask(rt); 863 info.rti_info[RTAX_TAG] = rt_gettag(rt); 864 error = route_output_report(rt, &info, rtm, &new_rtm); 865 if (error) 866 senderr(error); 867 if (new_rtm != NULL) { 868 old_rtm = rtm; 869 rtm = new_rtm; 870 } 871 break; 872 873 case RTM_GET: 874 case RTM_CHANGE: 875 case RTM_LOCK: 876 /* XXX This will mask info.rti_info[RTAX_DST] with 877 * info.rti_info[RTAX_NETMASK] before 878 * searching. It did not used to do that. --dyoung 879 */ 880 rt = NULL; 881 error = rtrequest1(RTM_GET, &info, &rt); 882 if (error != 0) 883 senderr(error); 884 if (rtm->rtm_type != RTM_GET) {/* XXX: too grotty */ 885 if (memcmp(info.rti_info[RTAX_DST], rt_getkey(rt), 886 info.rti_info[RTAX_DST]->sa_len) != 0) 887 senderr(ESRCH); 888 if (info.rti_info[RTAX_NETMASK] == NULL && 889 rt_mask(rt) != NULL) 890 senderr(ETOOMANYREFS); 891 } 892 893 /* 894 * XXX if arp/ndp requests an L2 entry, we have to obtain 895 * it from lltable while for the route command we have to 896 * return a route as it is. How to distinguish them? 897 * For newer arp/ndp, RTF_LLDATA flag set by arp/ndp 898 * indicates an L2 entry is requested. For old arp/ndp 899 * binaries, we check RTF_UP flag is NOT set; it works 900 * by the fact that arp/ndp don't set it while the route 901 * command sets it. 902 */ 903 if (((rtm->rtm_flags & RTF_LLDATA) != 0 || 904 (rtm->rtm_flags & RTF_UP) == 0) && 905 rtm->rtm_type == RTM_GET && 906 sockaddr_cmp(rt_getkey(rt), info.rti_info[RTAX_DST]) != 0) { 907 int ll_flags = 0; 908 route_get_sdl(rt->rt_ifp, info.rti_info[RTAX_DST], &sdl, 909 &ll_flags); 910 info.rti_info[RTAX_GATEWAY] = sstocsa(&sdl); 911 error = route_output_report(rt, &info, rtm, &new_rtm); 912 if (error) 913 senderr(error); 914 if (new_rtm != NULL) { 915 old_rtm = rtm; 916 rtm = new_rtm; 917 } 918 rtm->rtm_flags |= RTF_LLDATA; 919 rtm->rtm_flags &= ~RTF_CONNECTED; 920 rtm->rtm_flags |= (ll_flags & LLE_STATIC) ? RTF_STATIC : 0; 921 break; 922 } 923 924 switch (rtm->rtm_type) { 925 case RTM_GET: 926 info.rti_info[RTAX_DST] = rt_getkey(rt); 927 info.rti_info[RTAX_GATEWAY] = rt->rt_gateway; 928 info.rti_info[RTAX_NETMASK] = rt_mask(rt); 929 info.rti_info[RTAX_TAG] = rt_gettag(rt); 930 error = route_output_report(rt, &info, rtm, &new_rtm); 931 if (error) 932 senderr(error); 933 if (new_rtm != NULL) { 934 old_rtm = rtm; 935 rtm = new_rtm; 936 } 937 break; 938 939 case RTM_CHANGE: 940 #ifdef NET_MPSAFE 941 error = rt_update_prepare(rt); 942 if (error == 0) { 943 error = route_output_change(rt, &info, rtm); 944 rt_update_finish(rt); 945 } 946 #else 947 error = route_output_change(rt, &info, rtm); 948 #endif 949 if (error != 0) 950 goto flush; 951 /*FALLTHROUGH*/ 952 case RTM_LOCK: 953 rt->rt_rmx.rmx_locks &= ~(rtm->rtm_inits); 954 rt->rt_rmx.rmx_locks |= 955 (rtm->rtm_inits & rtm->rtm_rmx.rmx_locks); 956 break; 957 } 958 break; 959 960 default: 961 senderr(EOPNOTSUPP); 962 } 963 964 flush: 965 if (rtm) { 966 if (error) 967 rtm->rtm_errno = error; 968 else 969 rtm->rtm_flags |= RTF_DONE; 970 } 971 family = info.rti_info[RTAX_DST] ? info.rti_info[RTAX_DST]->sa_family : 972 0; 973 /* We cannot free old_rtm until we have stopped using the 974 * pointers in info, some of which may point to sockaddrs 975 * in old_rtm. 976 */ 977 if (old_rtm != NULL) 978 Free(old_rtm); 979 if (rt) { 980 if (do_rt_free) 981 rt_free(rt); 982 else 983 rt_unref(rt); 984 } 985 { 986 struct rawcb *rp = NULL; 987 /* 988 * Check to see if we don't want our own messages. 989 */ 990 if ((so->so_options & SO_USELOOPBACK) == 0) { 991 if (COMPATNAME(route_info).ri_cb.any_count <= 1) { 992 if (rtm) 993 Free(rtm); 994 m_freem(m); 995 goto out; 996 } 997 /* There is another listener, so construct message */ 998 rp = sotorawcb(so); 999 } 1000 if (rtm) { 1001 m_copyback(m, 0, rtm->rtm_msglen, rtm); 1002 if (m->m_pkthdr.len < rtm->rtm_msglen) { 1003 m_freem(m); 1004 m = NULL; 1005 } else if (m->m_pkthdr.len > rtm->rtm_msglen) 1006 m_adj(m, rtm->rtm_msglen - m->m_pkthdr.len); 1007 Free(rtm); 1008 } 1009 if (rp) 1010 rp->rcb_proto.sp_family = 0; /* Avoid us */ 1011 if (family) 1012 proto.sp_protocol = family; 1013 if (m) 1014 raw_input(m, &proto, &COMPATNAME(route_info).ri_src, 1015 &COMPATNAME(route_info).ri_dst); 1016 if (rp) 1017 rp->rcb_proto.sp_family = PF_XROUTE; 1018 } 1019 out: 1020 curlwp_bindx(bound); 1021 return error; 1022 } 1023 1024 static int 1025 route_ctloutput(int op, struct socket *so, struct sockopt *sopt) 1026 { 1027 struct routecb *rop = sotoroutecb(so); 1028 int error = 0; 1029 unsigned char *rtm_type; 1030 size_t len; 1031 unsigned int msgfilter; 1032 1033 KASSERT(solocked(so)); 1034 1035 if (sopt->sopt_level != AF_ROUTE) { 1036 error = ENOPROTOOPT; 1037 } else switch (op) { 1038 case PRCO_SETOPT: 1039 switch (sopt->sopt_name) { 1040 case RO_MSGFILTER: 1041 msgfilter = 0; 1042 for (rtm_type = sopt->sopt_data, len = sopt->sopt_size; 1043 len != 0; 1044 rtm_type++, len -= sizeof(*rtm_type)) 1045 { 1046 /* Guard against overflowing our storage. */ 1047 if (*rtm_type >= sizeof(msgfilter) * CHAR_BIT) { 1048 error = EOVERFLOW; 1049 break; 1050 } 1051 msgfilter |= RTMSGFILTER(*rtm_type); 1052 } 1053 if (error == 0) 1054 rop->rocb_msgfilter = msgfilter; 1055 break; 1056 default: 1057 error = ENOPROTOOPT; 1058 break; 1059 } 1060 break; 1061 case PRCO_GETOPT: 1062 switch (sopt->sopt_name) { 1063 case RO_MSGFILTER: 1064 error = ENOTSUP; 1065 break; 1066 default: 1067 error = ENOPROTOOPT; 1068 break; 1069 } 1070 } 1071 return error; 1072 } 1073 1074 static void 1075 rt_setmetrics(int which, const struct rt_xmsghdr *in, struct rtentry *out) 1076 { 1077 #define metric(f, e) if (which & (f)) out->rt_rmx.e = in->rtm_rmx.e; 1078 metric(RTV_RPIPE, rmx_recvpipe); 1079 metric(RTV_SPIPE, rmx_sendpipe); 1080 metric(RTV_SSTHRESH, rmx_ssthresh); 1081 metric(RTV_RTT, rmx_rtt); 1082 metric(RTV_RTTVAR, rmx_rttvar); 1083 metric(RTV_HOPCOUNT, rmx_hopcount); 1084 metric(RTV_MTU, rmx_mtu); 1085 #undef metric 1086 if (which & RTV_EXPIRE) { 1087 out->rt_rmx.rmx_expire = in->rtm_rmx.rmx_expire ? 1088 time_wall_to_mono(in->rtm_rmx.rmx_expire) : 0; 1089 } 1090 } 1091 1092 static void 1093 rtm_setmetrics(const struct rtentry *in, struct rt_xmsghdr *out) 1094 { 1095 #define metric(e) out->rtm_rmx.e = in->rt_rmx.e; 1096 metric(rmx_recvpipe); 1097 metric(rmx_sendpipe); 1098 metric(rmx_ssthresh); 1099 metric(rmx_rtt); 1100 metric(rmx_rttvar); 1101 metric(rmx_hopcount); 1102 metric(rmx_mtu); 1103 metric(rmx_locks); 1104 #undef metric 1105 out->rtm_rmx.rmx_expire = in->rt_rmx.rmx_expire ? 1106 time_mono_to_wall(in->rt_rmx.rmx_expire) : 0; 1107 } 1108 1109 static int 1110 rt_xaddrs(u_char rtmtype, const char *cp, const char *cplim, 1111 struct rt_addrinfo *rtinfo) 1112 { 1113 const struct sockaddr *sa = NULL; /* Quell compiler warning */ 1114 int i; 1115 1116 for (i = 0; i < RTAX_MAX && cp < cplim; i++) { 1117 if ((rtinfo->rti_addrs & (1 << i)) == 0) 1118 continue; 1119 rtinfo->rti_info[i] = sa = (const struct sockaddr *)cp; 1120 RT_XADVANCE(cp, sa); 1121 } 1122 1123 /* 1124 * Check for extra addresses specified, except RTM_GET asking 1125 * for interface info. 1126 */ 1127 if (rtmtype == RTM_GET) { 1128 if (((rtinfo->rti_addrs & 1129 (~((1 << RTAX_IFP) | (1 << RTAX_IFA)))) & (~0U << i)) != 0) 1130 return 1; 1131 } else if ((rtinfo->rti_addrs & (~0U << i)) != 0) 1132 return 1; 1133 /* Check for bad data length. */ 1134 if (cp != cplim) { 1135 if (i == RTAX_NETMASK + 1 && sa != NULL && 1136 cp - RT_XROUNDUP(sa->sa_len) + sa->sa_len == cplim) 1137 /* 1138 * The last sockaddr was info.rti_info[RTAX_NETMASK]. 1139 * We accept this for now for the sake of old 1140 * binaries or third party softwares. 1141 */ 1142 ; 1143 else 1144 return 1; 1145 } 1146 return 0; 1147 } 1148 1149 static int 1150 rt_getlen(int type) 1151 { 1152 #ifndef COMPAT_RTSOCK 1153 CTASSERT(__alignof(struct ifa_msghdr) >= sizeof(uint64_t)); 1154 CTASSERT(__alignof(struct if_msghdr) >= sizeof(uint64_t)); 1155 CTASSERT(__alignof(struct if_announcemsghdr) >= sizeof(uint64_t)); 1156 CTASSERT(__alignof(struct rt_msghdr) >= sizeof(uint64_t)); 1157 #endif 1158 1159 switch (type) { 1160 case RTM_ODELADDR: 1161 case RTM_ONEWADDR: 1162 case RTM_OCHGADDR: 1163 #ifdef COMPAT_70 1164 return sizeof(struct ifa_msghdr70); 1165 #else 1166 #ifdef RTSOCK_DEBUG 1167 printf("%s: unsupported RTM type %d\n", __func__, type); 1168 #endif 1169 return -1; 1170 #endif 1171 case RTM_DELADDR: 1172 case RTM_NEWADDR: 1173 case RTM_CHGADDR: 1174 return sizeof(struct ifa_xmsghdr); 1175 1176 case RTM_OOIFINFO: 1177 #ifdef COMPAT_14 1178 return sizeof(struct if_msghdr14); 1179 #else 1180 #ifdef RTSOCK_DEBUG 1181 printf("%s: unsupported RTM type RTM_OOIFINFO\n", __func__); 1182 #endif 1183 return -1; 1184 #endif 1185 case RTM_OIFINFO: 1186 #ifdef COMPAT_50 1187 return sizeof(struct if_msghdr50); 1188 #else 1189 #ifdef RTSOCK_DEBUG 1190 printf("%s: unsupported RTM type RTM_OIFINFO\n", __func__); 1191 #endif 1192 return -1; 1193 #endif 1194 1195 case RTM_IFINFO: 1196 return sizeof(struct if_xmsghdr); 1197 1198 case RTM_IFANNOUNCE: 1199 case RTM_IEEE80211: 1200 return sizeof(struct if_xannouncemsghdr); 1201 1202 default: 1203 return sizeof(struct rt_xmsghdr); 1204 } 1205 } 1206 1207 1208 struct mbuf * 1209 COMPATNAME(rt_msg1)(int type, struct rt_addrinfo *rtinfo, void *data, int datalen) 1210 { 1211 struct rt_xmsghdr *rtm; 1212 struct mbuf *m; 1213 int i; 1214 const struct sockaddr *sa; 1215 int len, dlen; 1216 1217 m = m_gethdr(M_DONTWAIT, MT_DATA); 1218 if (m == NULL) 1219 return m; 1220 MCLAIM(m, &COMPATNAME(routedomain).dom_mowner); 1221 1222 if ((len = rt_getlen(type)) == -1) 1223 goto out; 1224 if (len > MHLEN + MLEN) 1225 panic("%s: message too long", __func__); 1226 else if (len > MHLEN) { 1227 m->m_next = m_get(M_DONTWAIT, MT_DATA); 1228 if (m->m_next == NULL) 1229 goto out; 1230 MCLAIM(m->m_next, m->m_owner); 1231 m->m_pkthdr.len = len; 1232 m->m_len = MHLEN; 1233 m->m_next->m_len = len - MHLEN; 1234 } else { 1235 m->m_pkthdr.len = m->m_len = len; 1236 } 1237 m_reset_rcvif(m); 1238 m_copyback(m, 0, datalen, data); 1239 if (len > datalen) 1240 (void)memset(mtod(m, char *) + datalen, 0, len - datalen); 1241 rtm = mtod(m, struct rt_xmsghdr *); 1242 for (i = 0; i < RTAX_MAX; i++) { 1243 if ((sa = rtinfo->rti_info[i]) == NULL) 1244 continue; 1245 rtinfo->rti_addrs |= (1 << i); 1246 dlen = RT_XROUNDUP(sa->sa_len); 1247 m_copyback(m, len, sa->sa_len, sa); 1248 if (dlen != sa->sa_len) { 1249 /* 1250 * Up to 6 + 1 nul's since roundup is to 1251 * sizeof(uint64_t) (8 bytes) 1252 */ 1253 m_copyback(m, len + sa->sa_len, 1254 dlen - sa->sa_len, "\0\0\0\0\0\0"); 1255 } 1256 len += dlen; 1257 } 1258 if (m->m_pkthdr.len != len) 1259 goto out; 1260 rtm->rtm_msglen = len; 1261 rtm->rtm_version = RTM_XVERSION; 1262 rtm->rtm_type = type; 1263 return m; 1264 out: 1265 m_freem(m); 1266 return NULL; 1267 } 1268 1269 /* 1270 * rt_msg2 1271 * 1272 * fills 'cp' or 'w'.w_tmem with the routing socket message and 1273 * returns the length of the message in 'lenp'. 1274 * 1275 * if walkarg is 0, cp is expected to be 0 or a buffer large enough to hold 1276 * the message 1277 * otherwise walkarg's w_needed is updated and if the user buffer is 1278 * specified and w_needed indicates space exists the information is copied 1279 * into the temp space (w_tmem). w_tmem is [re]allocated if necessary, 1280 * if the allocation fails ENOBUFS is returned. 1281 */ 1282 static int 1283 rt_msg2(int type, struct rt_addrinfo *rtinfo, void *cpv, struct rt_walkarg *w, 1284 int *lenp) 1285 { 1286 int i; 1287 int len, dlen, second_time = 0; 1288 char *cp0, *cp = cpv; 1289 1290 rtinfo->rti_addrs = 0; 1291 again: 1292 if ((len = rt_getlen(type)) == -1) 1293 return EINVAL; 1294 1295 if ((cp0 = cp) != NULL) 1296 cp += len; 1297 for (i = 0; i < RTAX_MAX; i++) { 1298 const struct sockaddr *sa; 1299 1300 if ((sa = rtinfo->rti_info[i]) == NULL) 1301 continue; 1302 rtinfo->rti_addrs |= (1 << i); 1303 dlen = RT_XROUNDUP(sa->sa_len); 1304 if (cp) { 1305 int diff = dlen - sa->sa_len; 1306 (void)memcpy(cp, sa, (size_t)sa->sa_len); 1307 cp += sa->sa_len; 1308 if (diff > 0) { 1309 (void)memset(cp, 0, (size_t)diff); 1310 cp += diff; 1311 } 1312 } 1313 len += dlen; 1314 } 1315 if (cp == NULL && w != NULL && !second_time) { 1316 struct rt_walkarg *rw = w; 1317 1318 rw->w_needed += len; 1319 if (rw->w_needed <= 0 && rw->w_where) { 1320 if (rw->w_tmemsize < len) { 1321 if (rw->w_tmem) 1322 kmem_free(rw->w_tmem, rw->w_tmemsize); 1323 rw->w_tmem = kmem_alloc(len, KM_SLEEP); 1324 rw->w_tmemsize = len; 1325 } 1326 if (rw->w_tmem) { 1327 cp = rw->w_tmem; 1328 second_time = 1; 1329 goto again; 1330 } else { 1331 rw->w_tmemneeded = len; 1332 return ENOBUFS; 1333 } 1334 } 1335 } 1336 if (cp) { 1337 struct rt_xmsghdr *rtm = (struct rt_xmsghdr *)cp0; 1338 1339 rtm->rtm_version = RTM_XVERSION; 1340 rtm->rtm_type = type; 1341 rtm->rtm_msglen = len; 1342 } 1343 if (lenp) 1344 *lenp = len; 1345 return 0; 1346 } 1347 1348 #ifndef COMPAT_RTSOCK 1349 int 1350 rt_msg3(int type, struct rt_addrinfo *rtinfo, void *cpv, struct rt_walkarg *w, 1351 int *lenp) 1352 { 1353 return rt_msg2(type, rtinfo, cpv, w, lenp); 1354 } 1355 #endif 1356 1357 /* 1358 * This routine is called to generate a message from the routing 1359 * socket indicating that a redirect has occurred, a routing lookup 1360 * has failed, or that a protocol has detected timeouts to a particular 1361 * destination. 1362 */ 1363 void 1364 COMPATNAME(rt_missmsg)(int type, const struct rt_addrinfo *rtinfo, int flags, 1365 int error) 1366 { 1367 struct rt_xmsghdr rtm; 1368 struct mbuf *m; 1369 const struct sockaddr *sa = rtinfo->rti_info[RTAX_DST]; 1370 struct rt_addrinfo info = *rtinfo; 1371 1372 COMPATCALL(rt_missmsg, (type, rtinfo, flags, error)); 1373 if (COMPATNAME(route_info).ri_cb.any_count == 0) 1374 return; 1375 memset(&rtm, 0, sizeof(rtm)); 1376 rtm.rtm_pid = curproc->p_pid; 1377 rtm.rtm_flags = RTF_DONE | flags; 1378 rtm.rtm_errno = error; 1379 m = COMPATNAME(rt_msg1)(type, &info, &rtm, sizeof(rtm)); 1380 if (m == NULL) 1381 return; 1382 mtod(m, struct rt_xmsghdr *)->rtm_addrs = info.rti_addrs; 1383 COMPATNAME(route_enqueue)(m, sa ? sa->sa_family : 0); 1384 } 1385 1386 /* 1387 * This routine is called to generate a message from the routing 1388 * socket indicating that the status of a network interface has changed. 1389 */ 1390 void 1391 COMPATNAME(rt_ifmsg)(struct ifnet *ifp) 1392 { 1393 struct if_xmsghdr ifm; 1394 struct mbuf *m; 1395 struct rt_addrinfo info; 1396 1397 COMPATCALL(rt_ifmsg, (ifp)); 1398 if (COMPATNAME(route_info).ri_cb.any_count == 0) 1399 return; 1400 (void)memset(&info, 0, sizeof(info)); 1401 (void)memset(&ifm, 0, sizeof(ifm)); 1402 ifm.ifm_index = ifp->if_index; 1403 ifm.ifm_flags = ifp->if_flags; 1404 ifm.ifm_data = ifp->if_data; 1405 ifm.ifm_addrs = 0; 1406 m = COMPATNAME(rt_msg1)(RTM_IFINFO, &info, &ifm, sizeof(ifm)); 1407 if (m == NULL) 1408 return; 1409 COMPATNAME(route_enqueue)(m, 0); 1410 #ifdef COMPAT_14 1411 compat_14_rt_oifmsg(ifp); 1412 #endif 1413 #ifdef COMPAT_50 1414 compat_50_rt_oifmsg(ifp); 1415 #endif 1416 } 1417 1418 #ifndef COMPAT_RTSOCK 1419 static int 1420 if_addrflags(struct ifaddr *ifa) 1421 { 1422 1423 switch (ifa->ifa_addr->sa_family) { 1424 #ifdef INET 1425 case AF_INET: 1426 return ((struct in_ifaddr *)ifa)->ia4_flags; 1427 #endif 1428 #ifdef INET6 1429 case AF_INET6: 1430 return ((struct in6_ifaddr *)ifa)->ia6_flags; 1431 #endif 1432 default: 1433 return 0; 1434 } 1435 } 1436 #endif 1437 1438 /* 1439 * This is called to generate messages from the routing socket 1440 * indicating a network interface has had addresses associated with it. 1441 * if we ever reverse the logic and replace messages TO the routing 1442 * socket indicate a request to configure interfaces, then it will 1443 * be unnecessary as the routing socket will automatically generate 1444 * copies of it. 1445 */ 1446 void 1447 COMPATNAME(rt_newaddrmsg)(int cmd, struct ifaddr *ifa, int error, 1448 struct rtentry *rt) 1449 { 1450 #define cmdpass(__cmd, __pass) (((__cmd) << 2) | (__pass)) 1451 struct rt_addrinfo info; 1452 const struct sockaddr *sa; 1453 int pass; 1454 struct mbuf *m; 1455 struct ifnet *ifp; 1456 struct rt_xmsghdr rtm; 1457 struct ifa_xmsghdr ifam; 1458 int ncmd; 1459 1460 KASSERT(ifa != NULL); 1461 KASSERT(ifa->ifa_addr != NULL); 1462 ifp = ifa->ifa_ifp; 1463 #ifdef SCTP 1464 if (cmd == RTM_ADD) { 1465 sctp_add_ip_address(ifa); 1466 } else if (cmd == RTM_DELETE) { 1467 sctp_delete_ip_address(ifa); 1468 } 1469 #endif 1470 1471 COMPATCALL(rt_newaddrmsg, (cmd, ifa, error, rt)); 1472 if (COMPATNAME(route_info).ri_cb.any_count == 0) 1473 return; 1474 for (pass = 1; pass < 3; pass++) { 1475 memset(&info, 0, sizeof(info)); 1476 switch (cmdpass(cmd, pass)) { 1477 case cmdpass(RTM_ADD, 1): 1478 case cmdpass(RTM_CHANGE, 1): 1479 case cmdpass(RTM_DELETE, 2): 1480 case cmdpass(RTM_NEWADDR, 1): 1481 case cmdpass(RTM_DELADDR, 1): 1482 case cmdpass(RTM_CHGADDR, 1): 1483 switch (cmd) { 1484 case RTM_ADD: 1485 ncmd = RTM_XNEWADDR; 1486 break; 1487 case RTM_DELETE: 1488 ncmd = RTM_XDELADDR; 1489 break; 1490 case RTM_CHANGE: 1491 ncmd = RTM_XCHGADDR; 1492 break; 1493 case RTM_NEWADDR: 1494 ncmd = RTM_XNEWADDR; 1495 break; 1496 case RTM_DELADDR: 1497 ncmd = RTM_XDELADDR; 1498 break; 1499 case RTM_CHGADDR: 1500 ncmd = RTM_XCHGADDR; 1501 break; 1502 default: 1503 panic("%s: unknown command %d", __func__, cmd); 1504 } 1505 #ifdef COMPAT_70 1506 compat_70_rt_newaddrmsg1(ncmd, ifa); 1507 #endif 1508 info.rti_info[RTAX_IFA] = sa = ifa->ifa_addr; 1509 KASSERT(ifp->if_dl != NULL); 1510 info.rti_info[RTAX_IFP] = ifp->if_dl->ifa_addr; 1511 info.rti_info[RTAX_NETMASK] = ifa->ifa_netmask; 1512 info.rti_info[RTAX_BRD] = ifa->ifa_dstaddr; 1513 memset(&ifam, 0, sizeof(ifam)); 1514 ifam.ifam_index = ifp->if_index; 1515 ifam.ifam_metric = ifa->ifa_metric; 1516 ifam.ifam_flags = ifa->ifa_flags; 1517 #ifndef COMPAT_RTSOCK 1518 ifam.ifam_pid = curproc->p_pid; 1519 ifam.ifam_addrflags = if_addrflags(ifa); 1520 #endif 1521 m = COMPATNAME(rt_msg1)(ncmd, &info, &ifam, sizeof(ifam)); 1522 if (m == NULL) 1523 continue; 1524 mtod(m, struct ifa_xmsghdr *)->ifam_addrs = 1525 info.rti_addrs; 1526 break; 1527 case cmdpass(RTM_ADD, 2): 1528 case cmdpass(RTM_CHANGE, 2): 1529 case cmdpass(RTM_DELETE, 1): 1530 if (rt == NULL) 1531 continue; 1532 info.rti_info[RTAX_NETMASK] = rt_mask(rt); 1533 info.rti_info[RTAX_DST] = sa = rt_getkey(rt); 1534 info.rti_info[RTAX_GATEWAY] = rt->rt_gateway; 1535 memset(&rtm, 0, sizeof(rtm)); 1536 rtm.rtm_pid = curproc->p_pid; 1537 rtm.rtm_index = ifp->if_index; 1538 rtm.rtm_flags |= rt->rt_flags; 1539 rtm.rtm_errno = error; 1540 m = COMPATNAME(rt_msg1)(cmd, &info, &rtm, sizeof(rtm)); 1541 if (m == NULL) 1542 continue; 1543 mtod(m, struct rt_xmsghdr *)->rtm_addrs = info.rti_addrs; 1544 break; 1545 default: 1546 continue; 1547 } 1548 KASSERTMSG(m != NULL, "called with wrong command"); 1549 COMPATNAME(route_enqueue)(m, sa ? sa->sa_family : 0); 1550 } 1551 #undef cmdpass 1552 1553 } 1554 1555 static struct mbuf * 1556 rt_makeifannouncemsg(struct ifnet *ifp, int type, int what, 1557 struct rt_addrinfo *info) 1558 { 1559 struct if_xannouncemsghdr ifan; 1560 1561 memset(info, 0, sizeof(*info)); 1562 memset(&ifan, 0, sizeof(ifan)); 1563 ifan.ifan_index = ifp->if_index; 1564 strlcpy(ifan.ifan_name, ifp->if_xname, sizeof(ifan.ifan_name)); 1565 ifan.ifan_what = what; 1566 return COMPATNAME(rt_msg1)(type, info, &ifan, sizeof(ifan)); 1567 } 1568 1569 /* 1570 * This is called to generate routing socket messages indicating 1571 * network interface arrival and departure. 1572 */ 1573 void 1574 COMPATNAME(rt_ifannouncemsg)(struct ifnet *ifp, int what) 1575 { 1576 struct mbuf *m; 1577 struct rt_addrinfo info; 1578 1579 COMPATCALL(rt_ifannouncemsg, (ifp, what)); 1580 if (COMPATNAME(route_info).ri_cb.any_count == 0) 1581 return; 1582 m = rt_makeifannouncemsg(ifp, RTM_IFANNOUNCE, what, &info); 1583 if (m == NULL) 1584 return; 1585 COMPATNAME(route_enqueue)(m, 0); 1586 } 1587 1588 /* 1589 * This is called to generate routing socket messages indicating 1590 * IEEE80211 wireless events. 1591 * XXX we piggyback on the RTM_IFANNOUNCE msg format in a clumsy way. 1592 */ 1593 void 1594 COMPATNAME(rt_ieee80211msg)(struct ifnet *ifp, int what, void *data, 1595 size_t data_len) 1596 { 1597 struct mbuf *m; 1598 struct rt_addrinfo info; 1599 1600 COMPATCALL(rt_ieee80211msg, (ifp, what, data, data_len)); 1601 if (COMPATNAME(route_info).ri_cb.any_count == 0) 1602 return; 1603 m = rt_makeifannouncemsg(ifp, RTM_IEEE80211, what, &info); 1604 if (m == NULL) 1605 return; 1606 /* 1607 * Append the ieee80211 data. Try to stick it in the 1608 * mbuf containing the ifannounce msg; otherwise allocate 1609 * a new mbuf and append. 1610 * 1611 * NB: we assume m is a single mbuf. 1612 */ 1613 if (data_len > M_TRAILINGSPACE(m)) { 1614 struct mbuf *n = m_get(M_NOWAIT, MT_DATA); 1615 if (n == NULL) { 1616 m_freem(m); 1617 return; 1618 } 1619 (void)memcpy(mtod(n, void *), data, data_len); 1620 n->m_len = data_len; 1621 m->m_next = n; 1622 } else if (data_len > 0) { 1623 (void)memcpy(mtod(m, uint8_t *) + m->m_len, data, data_len); 1624 m->m_len += data_len; 1625 } 1626 if (m->m_flags & M_PKTHDR) 1627 m->m_pkthdr.len += data_len; 1628 mtod(m, struct if_xannouncemsghdr *)->ifan_msglen += data_len; 1629 COMPATNAME(route_enqueue)(m, 0); 1630 } 1631 1632 #ifndef COMPAT_RTSOCK 1633 /* 1634 * Send a routing message as mimicing that a cloned route is added. 1635 */ 1636 void 1637 rt_clonedmsg(const struct sockaddr *dst, const struct ifnet *ifp, 1638 const struct rtentry *rt) 1639 { 1640 struct rt_addrinfo info; 1641 /* Mimic flags exactly */ 1642 #define RTF_LLINFO 0x400 1643 #define RTF_CLONED 0x2000 1644 int flags = RTF_UP | RTF_HOST | RTF_DONE | RTF_LLINFO | RTF_CLONED; 1645 union { 1646 struct sockaddr sa; 1647 struct sockaddr_storage ss; 1648 struct sockaddr_dl sdl; 1649 } u; 1650 uint8_t namelen = strlen(ifp->if_xname); 1651 uint8_t addrlen = ifp->if_addrlen; 1652 1653 if (rt == NULL) 1654 return; /* XXX */ 1655 1656 memset(&info, 0, sizeof(info)); 1657 info.rti_info[RTAX_DST] = dst; 1658 sockaddr_dl_init(&u.sdl, sizeof(u.ss), ifp->if_index, ifp->if_type, 1659 NULL, namelen, NULL, addrlen); 1660 info.rti_info[RTAX_GATEWAY] = &u.sa; 1661 info.rti_info[RTAX_IFP] = rt->rt_ifp->if_dl->ifa_addr; 1662 info.rti_info[RTAX_IFA] = rt->rt_ifa->ifa_addr; 1663 1664 rt_missmsg(RTM_ADD, &info, flags, 0); 1665 #undef RTF_LLINFO 1666 #undef RTF_CLONED 1667 } 1668 #endif /* COMPAT_RTSOCK */ 1669 1670 /* 1671 * This is used in dumping the kernel table via sysctl(). 1672 */ 1673 static int 1674 sysctl_dumpentry(struct rtentry *rt, void *v) 1675 { 1676 struct rt_walkarg *w = v; 1677 int error = 0, size; 1678 struct rt_addrinfo info; 1679 1680 if (w->w_op == NET_RT_FLAGS && !(rt->rt_flags & w->w_arg)) 1681 return 0; 1682 memset(&info, 0, sizeof(info)); 1683 info.rti_info[RTAX_DST] = rt_getkey(rt); 1684 info.rti_info[RTAX_GATEWAY] = rt->rt_gateway; 1685 info.rti_info[RTAX_NETMASK] = rt_mask(rt); 1686 info.rti_info[RTAX_TAG] = rt_gettag(rt); 1687 if (rt->rt_ifp) { 1688 const struct ifaddr *rtifa; 1689 info.rti_info[RTAX_IFP] = rt->rt_ifp->if_dl->ifa_addr; 1690 /* rtifa used to be simply rt->rt_ifa. If rt->rt_ifa != NULL, 1691 * then rt_get_ifa() != NULL. So this ought to still be safe. 1692 * --dyoung 1693 */ 1694 rtifa = rt_get_ifa(rt); 1695 info.rti_info[RTAX_IFA] = rtifa->ifa_addr; 1696 if (rt->rt_ifp->if_flags & IFF_POINTOPOINT) 1697 info.rti_info[RTAX_BRD] = rtifa->ifa_dstaddr; 1698 } 1699 if ((error = rt_msg2(RTM_GET, &info, 0, w, &size))) 1700 return error; 1701 if (w->w_where && w->w_tmem && w->w_needed <= 0) { 1702 struct rt_xmsghdr *rtm = (struct rt_xmsghdr *)w->w_tmem; 1703 1704 rtm->rtm_flags = rt->rt_flags; 1705 rtm->rtm_use = rt->rt_use; 1706 rtm_setmetrics(rt, rtm); 1707 KASSERT(rt->rt_ifp != NULL); 1708 rtm->rtm_index = rt->rt_ifp->if_index; 1709 rtm->rtm_errno = rtm->rtm_pid = rtm->rtm_seq = 0; 1710 rtm->rtm_addrs = info.rti_addrs; 1711 if ((error = copyout(rtm, w->w_where, size)) != 0) 1712 w->w_where = NULL; 1713 else 1714 w->w_where = (char *)w->w_where + size; 1715 } 1716 return error; 1717 } 1718 1719 static int 1720 sysctl_iflist_if(struct ifnet *ifp, struct rt_walkarg *w, 1721 struct rt_addrinfo *info, size_t len) 1722 { 1723 struct if_xmsghdr *ifm; 1724 int error; 1725 1726 ifm = (struct if_xmsghdr *)w->w_tmem; 1727 ifm->ifm_index = ifp->if_index; 1728 ifm->ifm_flags = ifp->if_flags; 1729 ifm->ifm_data = ifp->if_data; 1730 ifm->ifm_addrs = info->rti_addrs; 1731 if ((error = copyout(ifm, w->w_where, len)) == 0) 1732 w->w_where = (char *)w->w_where + len; 1733 return error; 1734 } 1735 1736 static int 1737 sysctl_iflist_addr(struct rt_walkarg *w, struct ifaddr *ifa, 1738 struct rt_addrinfo *info) 1739 { 1740 int len, error; 1741 1742 if ((error = rt_msg2(RTM_XNEWADDR, info, 0, w, &len))) 1743 return error; 1744 if (w->w_where && w->w_tmem && w->w_needed <= 0) { 1745 struct ifa_xmsghdr *ifam; 1746 1747 ifam = (struct ifa_xmsghdr *)w->w_tmem; 1748 ifam->ifam_index = ifa->ifa_ifp->if_index; 1749 ifam->ifam_flags = ifa->ifa_flags; 1750 ifam->ifam_metric = ifa->ifa_metric; 1751 ifam->ifam_addrs = info->rti_addrs; 1752 #ifndef COMPAT_RTSOCK 1753 ifam->ifam_pid = 0; 1754 ifam->ifam_addrflags = if_addrflags(ifa); 1755 #endif 1756 if ((error = copyout(w->w_tmem, w->w_where, len)) == 0) 1757 w->w_where = (char *)w->w_where + len; 1758 } 1759 return error; 1760 } 1761 1762 static int 1763 sysctl_iflist(int af, struct rt_walkarg *w, int type) 1764 { 1765 struct ifnet *ifp; 1766 struct ifaddr *ifa; 1767 struct rt_addrinfo info; 1768 int cmd, len, error = 0; 1769 int (*iflist_if)(struct ifnet *, struct rt_walkarg *, 1770 struct rt_addrinfo *, size_t); 1771 int (*iflist_addr)(struct rt_walkarg *, struct ifaddr *, 1772 struct rt_addrinfo *); 1773 int s; 1774 struct psref psref; 1775 int bound; 1776 1777 switch (type) { 1778 case NET_RT_IFLIST: 1779 cmd = RTM_IFINFO; 1780 iflist_if = sysctl_iflist_if; 1781 iflist_addr = sysctl_iflist_addr; 1782 break; 1783 #ifdef COMPAT_14 1784 case NET_RT_OOOIFLIST: 1785 cmd = RTM_OOIFINFO; 1786 iflist_if = compat_14_iflist; 1787 iflist_addr = compat_70_iflist_addr; 1788 break; 1789 #endif 1790 #ifdef COMPAT_50 1791 case NET_RT_OOIFLIST: 1792 cmd = RTM_OIFINFO; 1793 iflist_if = compat_50_iflist; 1794 iflist_addr = compat_70_iflist_addr; 1795 break; 1796 #endif 1797 #ifdef COMPAT_70 1798 case NET_RT_OIFLIST: 1799 cmd = RTM_IFINFO; 1800 iflist_if = sysctl_iflist_if; 1801 iflist_addr = compat_70_iflist_addr; 1802 break; 1803 #endif 1804 default: 1805 #ifdef RTSOCK_DEBUG 1806 printf("%s: unsupported IFLIST type %d\n", __func__, type); 1807 #endif 1808 return EINVAL; 1809 } 1810 1811 memset(&info, 0, sizeof(info)); 1812 1813 bound = curlwp_bind(); 1814 s = pserialize_read_enter(); 1815 IFNET_READER_FOREACH(ifp) { 1816 int _s; 1817 if (w->w_arg && w->w_arg != ifp->if_index) 1818 continue; 1819 if (IFADDR_READER_EMPTY(ifp)) 1820 continue; 1821 1822 if_acquire(ifp, &psref); 1823 pserialize_read_exit(s); 1824 1825 info.rti_info[RTAX_IFP] = ifp->if_dl->ifa_addr; 1826 if ((error = rt_msg2(cmd, &info, NULL, w, &len)) != 0) 1827 goto release_exit; 1828 info.rti_info[RTAX_IFP] = NULL; 1829 if (w->w_where && w->w_tmem && w->w_needed <= 0) { 1830 if ((error = iflist_if(ifp, w, &info, len)) != 0) 1831 goto release_exit; 1832 } 1833 _s = pserialize_read_enter(); 1834 IFADDR_READER_FOREACH(ifa, ifp) { 1835 struct psref _psref; 1836 if (af && af != ifa->ifa_addr->sa_family) 1837 continue; 1838 ifa_acquire(ifa, &_psref); 1839 pserialize_read_exit(_s); 1840 1841 info.rti_info[RTAX_IFA] = ifa->ifa_addr; 1842 info.rti_info[RTAX_NETMASK] = ifa->ifa_netmask; 1843 info.rti_info[RTAX_BRD] = ifa->ifa_dstaddr; 1844 error = iflist_addr(w, ifa, &info); 1845 1846 _s = pserialize_read_enter(); 1847 ifa_release(ifa, &_psref); 1848 if (error != 0) { 1849 pserialize_read_exit(_s); 1850 goto release_exit; 1851 } 1852 } 1853 pserialize_read_exit(_s); 1854 info.rti_info[RTAX_IFA] = info.rti_info[RTAX_NETMASK] = 1855 info.rti_info[RTAX_BRD] = NULL; 1856 1857 s = pserialize_read_enter(); 1858 if_release(ifp, &psref); 1859 } 1860 pserialize_read_exit(s); 1861 curlwp_bindx(bound); 1862 1863 return 0; 1864 1865 release_exit: 1866 if_release(ifp, &psref); 1867 curlwp_bindx(bound); 1868 return error; 1869 } 1870 1871 static int 1872 sysctl_rtable(SYSCTLFN_ARGS) 1873 { 1874 void *where = oldp; 1875 size_t *given = oldlenp; 1876 int i, s, error = EINVAL; 1877 u_char af; 1878 struct rt_walkarg w; 1879 1880 if (namelen == 1 && name[0] == CTL_QUERY) 1881 return sysctl_query(SYSCTLFN_CALL(rnode)); 1882 1883 if (newp) 1884 return EPERM; 1885 if (namelen != 3) 1886 return EINVAL; 1887 af = name[0]; 1888 w.w_tmemneeded = 0; 1889 w.w_tmemsize = 0; 1890 w.w_tmem = NULL; 1891 again: 1892 /* we may return here if a later [re]alloc of the t_mem buffer fails */ 1893 if (w.w_tmemneeded) { 1894 w.w_tmem = kmem_alloc(w.w_tmemneeded, KM_SLEEP); 1895 w.w_tmemsize = w.w_tmemneeded; 1896 w.w_tmemneeded = 0; 1897 } 1898 w.w_op = name[1]; 1899 w.w_arg = name[2]; 1900 w.w_given = *given; 1901 w.w_needed = 0 - w.w_given; 1902 w.w_where = where; 1903 1904 s = splsoftnet(); 1905 switch (w.w_op) { 1906 1907 case NET_RT_DUMP: 1908 case NET_RT_FLAGS: 1909 #ifdef INET 1910 /* 1911 * take care of llinfo entries, the caller must 1912 * specify an AF 1913 */ 1914 if (w.w_op == NET_RT_FLAGS && 1915 (w.w_arg == 0 || w.w_arg & RTF_LLDATA)) { 1916 if (af != 0) 1917 error = lltable_sysctl_dumparp(af, &w); 1918 else 1919 error = EINVAL; 1920 break; 1921 } 1922 #endif /* INET */ 1923 1924 for (i = 1; i <= AF_MAX; i++) 1925 if ((af == 0 || af == i) && 1926 (error = rt_walktree(i, sysctl_dumpentry, &w))) 1927 break; 1928 break; 1929 1930 #ifdef COMPAT_14 1931 case NET_RT_OOOIFLIST: 1932 error = sysctl_iflist(af, &w, w.w_op); 1933 break; 1934 #endif 1935 #ifdef COMPAT_50 1936 case NET_RT_OOIFLIST: 1937 error = sysctl_iflist(af, &w, w.w_op); 1938 break; 1939 #endif 1940 #ifdef COMPAT_70 1941 case NET_RT_OIFLIST: 1942 error = sysctl_iflist(af, &w, w.w_op); 1943 break; 1944 #endif 1945 case NET_RT_IFLIST: 1946 error = sysctl_iflist(af, &w, w.w_op); 1947 break; 1948 } 1949 splx(s); 1950 1951 /* check to see if we couldn't allocate memory with NOWAIT */ 1952 if (error == ENOBUFS && w.w_tmem == 0 && w.w_tmemneeded) 1953 goto again; 1954 1955 if (w.w_tmem) 1956 kmem_free(w.w_tmem, w.w_tmemsize); 1957 w.w_needed += w.w_given; 1958 if (where) { 1959 *given = (char *)w.w_where - (char *)where; 1960 if (*given < w.w_needed) 1961 return ENOMEM; 1962 } else { 1963 *given = (11 * w.w_needed) / 10; 1964 } 1965 return error; 1966 } 1967 1968 /* 1969 * Routing message software interrupt routine 1970 */ 1971 static void 1972 COMPATNAME(route_intr)(void *cookie) 1973 { 1974 struct sockproto proto = { .sp_family = PF_XROUTE, }; 1975 struct route_info * const ri = &COMPATNAME(route_info); 1976 struct mbuf *m; 1977 1978 mutex_enter(softnet_lock); 1979 KERNEL_LOCK(1, NULL); 1980 for (;;) { 1981 IFQ_LOCK(&ri->ri_intrq); 1982 IF_DEQUEUE(&ri->ri_intrq, m); 1983 IFQ_UNLOCK(&ri->ri_intrq); 1984 if (m == NULL) 1985 break; 1986 proto.sp_protocol = M_GETCTX(m, uintptr_t); 1987 raw_input(m, &proto, &ri->ri_src, &ri->ri_dst); 1988 } 1989 KERNEL_UNLOCK_ONE(NULL); 1990 mutex_exit(softnet_lock); 1991 } 1992 1993 /* 1994 * Enqueue a message to the software interrupt routine. 1995 */ 1996 void 1997 COMPATNAME(route_enqueue)(struct mbuf *m, int family) 1998 { 1999 struct route_info * const ri = &COMPATNAME(route_info); 2000 int wasempty; 2001 2002 IFQ_LOCK(&ri->ri_intrq); 2003 if (IF_QFULL(&ri->ri_intrq)) { 2004 IF_DROP(&ri->ri_intrq); 2005 IFQ_UNLOCK(&ri->ri_intrq); 2006 m_freem(m); 2007 } else { 2008 wasempty = IF_IS_EMPTY(&ri->ri_intrq); 2009 M_SETCTX(m, (uintptr_t)family); 2010 IF_ENQUEUE(&ri->ri_intrq, m); 2011 IFQ_UNLOCK(&ri->ri_intrq); 2012 if (wasempty) { 2013 kpreempt_disable(); 2014 softint_schedule(ri->ri_sih); 2015 kpreempt_enable(); 2016 } 2017 } 2018 } 2019 2020 static void 2021 COMPATNAME(route_init)(void) 2022 { 2023 struct route_info * const ri = &COMPATNAME(route_info); 2024 2025 #ifndef COMPAT_RTSOCK 2026 rt_init(); 2027 #endif 2028 2029 sysctl_net_route_setup(NULL); 2030 ri->ri_intrq.ifq_maxlen = ri->ri_maxqlen; 2031 ri->ri_sih = softint_establish(SOFTINT_NET | SOFTINT_MPSAFE, 2032 COMPATNAME(route_intr), NULL); 2033 IFQ_LOCK_INIT(&ri->ri_intrq); 2034 } 2035 2036 /* 2037 * Definitions of protocols supported in the ROUTE domain. 2038 */ 2039 #ifndef COMPAT_RTSOCK 2040 PR_WRAP_USRREQS(route); 2041 #else 2042 PR_WRAP_USRREQS(compat_50_route); 2043 #endif 2044 2045 static const struct pr_usrreqs route_usrreqs = { 2046 .pr_attach = COMPATNAME(route_attach_wrapper), 2047 .pr_detach = COMPATNAME(route_detach_wrapper), 2048 .pr_accept = COMPATNAME(route_accept_wrapper), 2049 .pr_bind = COMPATNAME(route_bind_wrapper), 2050 .pr_listen = COMPATNAME(route_listen_wrapper), 2051 .pr_connect = COMPATNAME(route_connect_wrapper), 2052 .pr_connect2 = COMPATNAME(route_connect2_wrapper), 2053 .pr_disconnect = COMPATNAME(route_disconnect_wrapper), 2054 .pr_shutdown = COMPATNAME(route_shutdown_wrapper), 2055 .pr_abort = COMPATNAME(route_abort_wrapper), 2056 .pr_ioctl = COMPATNAME(route_ioctl_wrapper), 2057 .pr_stat = COMPATNAME(route_stat_wrapper), 2058 .pr_peeraddr = COMPATNAME(route_peeraddr_wrapper), 2059 .pr_sockaddr = COMPATNAME(route_sockaddr_wrapper), 2060 .pr_rcvd = COMPATNAME(route_rcvd_wrapper), 2061 .pr_recvoob = COMPATNAME(route_recvoob_wrapper), 2062 .pr_send = COMPATNAME(route_send_wrapper), 2063 .pr_sendoob = COMPATNAME(route_sendoob_wrapper), 2064 .pr_purgeif = COMPATNAME(route_purgeif_wrapper), 2065 }; 2066 2067 static const struct protosw COMPATNAME(route_protosw)[] = { 2068 { 2069 .pr_type = SOCK_RAW, 2070 .pr_domain = &COMPATNAME(routedomain), 2071 .pr_flags = PR_ATOMIC|PR_ADDR, 2072 .pr_input = raw_input, 2073 .pr_ctlinput = raw_ctlinput, 2074 .pr_ctloutput = route_ctloutput, 2075 .pr_usrreqs = &route_usrreqs, 2076 .pr_init = raw_init, 2077 }, 2078 }; 2079 2080 struct domain COMPATNAME(routedomain) = { 2081 .dom_family = PF_XROUTE, 2082 .dom_name = DOMAINNAME, 2083 .dom_init = COMPATNAME(route_init), 2084 .dom_protosw = COMPATNAME(route_protosw), 2085 .dom_protoswNPROTOSW = 2086 &COMPATNAME(route_protosw)[__arraycount(COMPATNAME(route_protosw))], 2087 }; 2088 2089 static void 2090 sysctl_net_route_setup(struct sysctllog **clog) 2091 { 2092 const struct sysctlnode *rnode = NULL; 2093 2094 sysctl_createv(clog, 0, NULL, &rnode, 2095 CTLFLAG_PERMANENT, 2096 CTLTYPE_NODE, DOMAINNAME, 2097 SYSCTL_DESCR("PF_ROUTE information"), 2098 NULL, 0, NULL, 0, 2099 CTL_NET, PF_XROUTE, CTL_EOL); 2100 2101 sysctl_createv(clog, 0, NULL, NULL, 2102 CTLFLAG_PERMANENT, 2103 CTLTYPE_NODE, "rtable", 2104 SYSCTL_DESCR("Routing table information"), 2105 sysctl_rtable, 0, NULL, 0, 2106 CTL_NET, PF_XROUTE, 0 /* any protocol */, CTL_EOL); 2107 2108 sysctl_createv(clog, 0, &rnode, NULL, 2109 CTLFLAG_PERMANENT, 2110 CTLTYPE_STRUCT, "stats", 2111 SYSCTL_DESCR("Routing statistics"), 2112 NULL, 0, &rtstat, sizeof(rtstat), 2113 CTL_CREATE, CTL_EOL); 2114 } 2115
Indexes created Tue Sep 23 14:10:03 GMT 2025