1 /*- 2 * SPDX-License-Identifier: BSD-2-Clause-FreeBSD 3 * 4 * Copyright (c) 2022 Alexander V. Chernikov <melifaro@FreeBSD.org> 5 * 6 * Redistribution and use in source and binary forms, with or without 7 * modification, are permitted provided that the following conditions 8 * are met: 9 * 1. Redistributions of source code must retain the above copyright 10 * notice, this list of conditions and the following disclaimer. 11 * 2. Redistributions in binary form must reproduce the above copyright 12 * notice, this list of conditions and the following disclaimer in the 13 * documentation and/or other materials provided with the distribution. 14 * 15 * THIS SOFTWARE IS PROVIDED BY THE AUTHOR AND CONTRIBUTORS ``AS IS'' AND 16 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE 17 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE 18 * ARE DISCLAIMED. IN NO EVENT SHALL THE AUTHOR OR CONTRIBUTORS BE LIABLE 19 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL 20 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS 21 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) 22 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT 23 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY 24 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF 25 * SUCH DAMAGE. 26 */ 27 28 #include <sys/cdefs.h> 29 __FBSDID("$FreeBSD$"); 30 #include "opt_inet.h" 31 #include "opt_inet6.h" 32 #include <sys/types.h> 33 #include <sys/eventhandler.h> 34 #include <sys/malloc.h> 35 #include <sys/socket.h> 36 #include <sys/syslog.h> 37 38 #include <net/if.h> 39 #include <net/if_llatbl.h> 40 #include <netlink/netlink.h> 41 #include <netlink/netlink_ctl.h> 42 #include <netlink/netlink_route.h> 43 #include <netlink/route/route_var.h> 44 45 #include <netinet6/in6_var.h> /* nd6.h requires this */ 46 #include <netinet6/nd6.h> /* nd6 state machine */ 47 #include <netinet6/scope6_var.h> /* scope deembedding */ 48 49 #define DEBUG_MOD_NAME nl_neigh 50 #define DEBUG_MAX_LEVEL LOG_DEBUG3 51 #include <netlink/netlink_debug.h> 52 _DECLARE_DEBUG(LOG_DEBUG); 53 54 static int lle_families[] = { AF_INET, AF_INET6 }; 55 56 static eventhandler_tag lle_event_p; 57 58 struct netlink_walkargs { 59 struct nl_writer *nw; 60 struct nlmsghdr hdr; 61 struct nlpcb *so; 62 struct ifnet *ifp; 63 int family; 64 int error; 65 int count; 66 int dumped; 67 }; 68 69 static int 70 lle_state_to_nl_state(int family, struct llentry *lle) 71 { 72 int state = lle->ln_state; 73 74 switch (family) { 75 case AF_INET: 76 if (lle->la_flags & (LLE_STATIC | LLE_IFADDR)) 77 state = 1; 78 switch (state) { 79 case 0: /* ARP_LLINFO_INCOMPLETE */ 80 return (NUD_INCOMPLETE); 81 case 1: /* ARP_LLINFO_REACHABLE */ 82 return (NUD_REACHABLE); 83 case 2: /* ARP_LLINFO_VERIFY */ 84 return (NUD_PROBE); 85 } 86 break; 87 case AF_INET6: 88 switch (state) { 89 case ND6_LLINFO_INCOMPLETE: 90 return (NUD_INCOMPLETE); 91 case ND6_LLINFO_REACHABLE: 92 return (NUD_REACHABLE); 93 case ND6_LLINFO_STALE: 94 return (NUD_STALE); 95 case ND6_LLINFO_DELAY: 96 return (NUD_DELAY); 97 case ND6_LLINFO_PROBE: 98 return (NUD_PROBE); 99 } 100 break; 101 } 102 103 return (NUD_NONE); 104 } 105 106 static uint32_t 107 lle_flags_to_nl_flags(const struct llentry *lle) 108 { 109 uint32_t nl_flags = 0; 110 111 if (lle->la_flags & LLE_IFADDR) 112 nl_flags |= NTF_SELF; 113 if (lle->la_flags & LLE_PUB) 114 nl_flags |= NTF_PROXY; 115 if (lle->la_flags & LLE_STATIC) 116 nl_flags |= NTF_STICKY; 117 if (lle->ln_router != 0) 118 nl_flags |= NTF_ROUTER; 119 120 return (nl_flags); 121 } 122 123 static int 124 dump_lle_locked(struct llentry *lle, void *arg) 125 { 126 struct netlink_walkargs *wa = (struct netlink_walkargs *)arg; 127 struct nlmsghdr *hdr = &wa->hdr; 128 struct nl_writer *nw = wa->nw; 129 struct ndmsg *ndm; 130 union { 131 struct in_addr in; 132 struct in6_addr in6; 133 } addr; 134 135 IF_DEBUG_LEVEL(LOG_DEBUG2) { 136 char llebuf[NHOP_PRINT_BUFSIZE]; 137 llentry_print_buf_lltable(lle, llebuf, sizeof(llebuf)); 138 NL_LOG(LOG_DEBUG2, "dumping %s", llebuf); 139 } 140 141 if (!nlmsg_reply(nw, hdr, sizeof(struct ndmsg))) 142 goto enomem; 143 144 ndm = nlmsg_reserve_object(nw, struct ndmsg); 145 ndm->ndm_family = wa->family; 146 ndm->ndm_ifindex = wa->ifp->if_index; 147 ndm->ndm_state = lle_state_to_nl_state(wa->family, lle); 148 ndm->ndm_flags = lle_flags_to_nl_flags(lle); 149 150 switch (wa->family) { 151 #ifdef INET 152 case AF_INET: 153 addr.in = lle->r_l3addr.addr4; 154 nlattr_add(nw, NDA_DST, 4, &addr); 155 break; 156 #endif 157 #ifdef INET6 158 case AF_INET6: 159 addr.in6 = lle->r_l3addr.addr6; 160 in6_clearscope(&addr.in6); 161 nlattr_add(nw, NDA_DST, 16, &addr); 162 break; 163 #endif 164 } 165 166 if (lle->r_flags & RLLE_VALID) { 167 /* Has L2 */ 168 int addrlen = wa->ifp->if_addrlen; 169 nlattr_add(nw, NDA_LLADDR, addrlen, lle->ll_addr); 170 } 171 172 nlattr_add_u32(nw, NDA_PROBES, lle->la_asked); 173 174 struct nda_cacheinfo *cache; 175 cache = nlmsg_reserve_attr(nw, NDA_CACHEINFO, struct nda_cacheinfo); 176 if (cache == NULL) 177 goto enomem; 178 /* TODO: provide confirmed/updated */ 179 cache->ndm_refcnt = lle->lle_refcnt; 180 181 if (nlmsg_end(nw)) 182 return (0); 183 enomem: 184 NL_LOG(LOG_DEBUG, "unable to dump lle state (ENOMEM)"); 185 nlmsg_abort(nw); 186 return (ENOMEM); 187 } 188 189 static int 190 dump_lle(struct lltable *llt, struct llentry *lle, void *arg) 191 { 192 int error; 193 194 LLE_RLOCK(lle); 195 error = dump_lle_locked(lle, arg); 196 LLE_RUNLOCK(lle); 197 return (error); 198 } 199 200 static bool 201 dump_llt(struct lltable *llt, struct netlink_walkargs *wa) 202 { 203 lltable_foreach_lle(llt, dump_lle, wa); 204 205 return (true); 206 } 207 208 static int 209 dump_llts_iface(struct netlink_walkargs *wa, struct ifnet *ifp, int family) 210 { 211 int error = 0; 212 213 wa->ifp = ifp; 214 for (int i = 0; i < sizeof(lle_families) / sizeof(int); i++) { 215 int fam = lle_families[i]; 216 struct lltable *llt = lltable_get(ifp, fam); 217 if (llt != NULL && (family == 0 || family == fam)) { 218 wa->count++; 219 wa->family = fam; 220 if (!dump_llt(llt, wa)) { 221 error = ENOMEM; 222 break; 223 } 224 wa->dumped++; 225 } 226 } 227 return (error); 228 } 229 230 static int 231 dump_llts(struct netlink_walkargs *wa, struct ifnet *ifp, int family) 232 { 233 NL_LOG(LOG_DEBUG, "Start dump ifp=%s family=%d", ifp ? if_name(ifp) : "NULL", family); 234 235 wa->hdr.nlmsg_flags |= NLM_F_MULTI; 236 237 if (ifp != NULL) { 238 dump_llts_iface(wa, ifp, family); 239 } else { 240 CK_STAILQ_FOREACH(ifp, &V_ifnet, if_link) { 241 dump_llts_iface(wa, ifp, family); 242 } 243 } 244 245 NL_LOG(LOG_DEBUG, "End dump, iterated %d dumped %d", wa->count, wa->dumped); 246 247 if (!nlmsg_end_dump(wa->nw, wa->error, &wa->hdr)) { 248 NL_LOG(LOG_DEBUG, "Unable to add new message"); 249 return (ENOMEM); 250 } 251 252 return (0); 253 } 254 255 static int 256 get_lle(struct netlink_walkargs *wa, struct ifnet *ifp, int family, struct sockaddr *dst) 257 { 258 struct lltable *llt = lltable_get(ifp, family); 259 if (llt == NULL) 260 return (ESRCH); 261 262 #ifdef INET6 263 if (dst->sa_family == AF_INET6) { 264 struct sockaddr_in6 *dst6 = (struct sockaddr_in6 *)dst; 265 266 if (IN6_IS_SCOPE_LINKLOCAL(&dst6->sin6_addr)) 267 in6_set_unicast_scopeid(&dst6->sin6_addr, ifp->if_index); 268 } 269 #endif 270 struct llentry *lle = lla_lookup(llt, LLE_UNLOCKED, dst); 271 if (lle == NULL) 272 return (ESRCH); 273 274 wa->ifp = ifp; 275 wa->family = family; 276 277 return (dump_lle(llt, lle, wa)); 278 } 279 280 struct nl_parsed_neigh { 281 struct sockaddr *nda_dst; 282 struct ifnet *nda_ifp; 283 struct nlattr *nda_lladdr; 284 uint32_t ndm_flags; 285 uint16_t ndm_state; 286 uint8_t ndm_family; 287 }; 288 289 #define _IN(_field) offsetof(struct ndmsg, _field) 290 #define _OUT(_field) offsetof(struct nl_parsed_neigh, _field) 291 static struct nlfield_parser nlf_p_neigh[] = { 292 { .off_in = _IN(ndm_family), .off_out = _OUT(ndm_family), .cb = nlf_get_u8 }, 293 { .off_in = _IN(ndm_flags), .off_out = _OUT(ndm_flags), .cb = nlf_get_u8_u32 }, 294 { .off_in = _IN(ndm_state), .off_out = _OUT(ndm_state), .cb = nlf_get_u16 }, 295 { .off_in = _IN(ndm_ifindex), .off_out = _OUT(nda_ifp), .cb = nlf_get_ifpz }, 296 }; 297 298 static struct nlattr_parser nla_p_neigh[] = { 299 { .type = NDA_DST, .off = _OUT(nda_dst), .cb = nlattr_get_ip }, 300 { .type = NDA_LLADDR, .off = _OUT(nda_lladdr), .cb = nlattr_get_nla }, 301 { .type = NDA_IFINDEX, .off = _OUT(nda_ifp), .cb = nlattr_get_ifp }, 302 { .type = NDA_FLAGS_EXT, .off = _OUT(ndm_flags), .cb = nlattr_get_uint32 }, 303 }; 304 #undef _IN 305 #undef _OUT 306 NL_DECLARE_PARSER(ndmsg_parser, struct ndmsg, nlf_p_neigh, nla_p_neigh); 307 308 309 /* 310 * type=RTM_NEWNEIGH, flags=NLM_F_REQUEST|NLM_F_ACK|NLM_F_EXCL|NLM_F_CREATE, seq=1661941473, pid=0}, 311 * {ndm_family=AF_INET6, ndm_ifindex=if_nametoindex("enp0s31f6"), ndm_state=NUD_PERMANENT, ndm_flags=0, ndm_type=RTN_UNSPEC}, 312 * [ 313 * {{nla_len=20, nla_type=NDA_DST}, inet_pton(AF_INET6, "2a01:4f8:13a:70c::3")}, 314 * {{nla_len=10, nla_type=NDA_LLADDR}, 20:4e:71:62:ae:f2}]}, iov_len=60} 315 */ 316 317 static int 318 rtnl_handle_newneigh(struct nlmsghdr *hdr, struct nlpcb *nlp, struct nl_pstate *npt) 319 { 320 int error; 321 322 struct nl_parsed_neigh attrs = {}; 323 error = nl_parse_nlmsg(hdr, &ndmsg_parser, npt, &attrs); 324 if (error != 0) 325 return (error); 326 327 if (attrs.nda_ifp == NULL || attrs.nda_dst == NULL || attrs.nda_lladdr == NULL) { 328 if (attrs.nda_ifp == NULL) 329 NLMSG_REPORT_ERR_MSG(npt, "NDA_IFINDEX / ndm_ifindex not set"); 330 if (attrs.nda_dst == NULL) 331 NLMSG_REPORT_ERR_MSG(npt, "NDA_DST not set"); 332 if (attrs.nda_lladdr == NULL) 333 NLMSG_REPORT_ERR_MSG(npt, "NDA_LLADDR not set"); 334 return (EINVAL); 335 } 336 337 if (attrs.nda_dst->sa_family != attrs.ndm_family) { 338 NLMSG_REPORT_ERR_MSG(npt, 339 "NDA_DST family (%d) is different from ndm_family (%d)", 340 attrs.nda_dst->sa_family, attrs.ndm_family); 341 return (EINVAL); 342 } 343 344 int addrlen = attrs.nda_ifp->if_addrlen; 345 if (attrs.nda_lladdr->nla_len != sizeof(struct nlattr) + addrlen) { 346 NLMSG_REPORT_ERR_MSG(npt, 347 "NDA_LLADDR address length (%d) is different from expected (%d)", 348 (int)attrs.nda_lladdr->nla_len - (int)sizeof(struct nlattr), addrlen); 349 return (EINVAL); 350 } 351 352 if (attrs.ndm_state != NUD_PERMANENT) { 353 NLMSG_REPORT_ERR_MSG(npt, "ndm_state %d not supported", attrs.ndm_state); 354 return (ENOTSUP); 355 } 356 357 const uint16_t supported_flags = NTF_PROXY | NTF_STICKY; 358 if ((attrs.ndm_flags & supported_flags) != attrs.ndm_flags) { 359 NLMSG_REPORT_ERR_MSG(npt, "ndm_flags %X not supported", 360 attrs.ndm_flags &~ supported_flags); 361 return (ENOTSUP); 362 } 363 364 /* Replacement requires new entry creation anyway */ 365 if ((hdr->nlmsg_flags & (NLM_F_CREATE | NLM_F_REPLACE)) == 0) 366 return (ENOTSUP); 367 368 struct lltable *llt = lltable_get(attrs.nda_ifp, attrs.ndm_family); 369 if (llt == NULL) 370 return (EAFNOSUPPORT); 371 372 373 uint8_t linkhdr[LLE_MAX_LINKHDR]; 374 size_t linkhdrsize = sizeof(linkhdr); 375 int lladdr_off = 0; 376 if (lltable_calc_llheader(attrs.nda_ifp, attrs.ndm_family, 377 (char *)(attrs.nda_lladdr + 1), linkhdr, &linkhdrsize, &lladdr_off) != 0) { 378 NLMSG_REPORT_ERR_MSG(npt, "unable to calculate lle prepend data"); 379 return (EINVAL); 380 } 381 382 int lle_flags = LLE_STATIC | ((attrs.ndm_flags & NTF_PROXY) ? LLE_PUB : 0); 383 struct llentry *lle = lltable_alloc_entry(llt, lle_flags, attrs.nda_dst); 384 if (lle == NULL) 385 return (ENOMEM); 386 lltable_set_entry_addr(attrs.nda_ifp, lle, linkhdr, linkhdrsize, lladdr_off); 387 388 /* llentry created, try to insert or update :*/ 389 IF_AFDATA_WLOCK(attrs.nda_ifp); 390 LLE_WLOCK(lle); 391 struct llentry *lle_tmp = lla_lookup(llt, LLE_EXCLUSIVE, attrs.nda_dst); 392 if (lle_tmp != NULL) { 393 if (hdr->nlmsg_flags & NLM_F_EXCL) { 394 LLE_WUNLOCK(lle_tmp); 395 lle_tmp = NULL; 396 error = EEXIST; 397 } else if (hdr->nlmsg_flags & NLM_F_REPLACE) { 398 lltable_unlink_entry(llt, lle_tmp); 399 lltable_link_entry(llt, lle); 400 } else 401 error = EEXIST; 402 } else { 403 if (hdr->nlmsg_flags & NLM_F_CREATE) 404 lltable_link_entry(llt, lle); 405 else 406 error = ENOENT; 407 } 408 IF_AFDATA_WUNLOCK(attrs.nda_ifp); 409 410 if (error != 0) { 411 if (lle != NULL) 412 llentry_free(lle); 413 return (error); 414 } 415 416 if (lle_tmp != NULL) 417 llentry_free(lle_tmp); 418 419 /* XXX: We're inside epoch */ 420 EVENTHANDLER_INVOKE(lle_event, lle, LLENTRY_RESOLVED); 421 LLE_WUNLOCK(lle); 422 423 return (0); 424 } 425 426 static int 427 rtnl_handle_delneigh(struct nlmsghdr *hdr, struct nlpcb *nlp, struct nl_pstate *npt) 428 { 429 int error; 430 431 struct nl_parsed_neigh attrs = {}; 432 error = nl_parse_nlmsg(hdr, &ndmsg_parser, npt, &attrs); 433 if (error != 0) 434 return (error); 435 436 if (attrs.nda_dst == NULL) { 437 NLMSG_REPORT_ERR_MSG(npt, "NDA_DST not set"); 438 return (EINVAL); 439 } 440 441 if (attrs.nda_ifp == NULL) { 442 NLMSG_REPORT_ERR_MSG(npt, "no ifindex provided"); 443 return (EINVAL); 444 } 445 446 struct lltable *llt = lltable_get(attrs.nda_ifp, attrs.ndm_family); 447 if (llt == NULL) 448 return (EAFNOSUPPORT); 449 450 IF_AFDATA_WLOCK(attrs.nda_ifp); 451 struct llentry *lle = lla_lookup(llt, LLE_EXCLUSIVE, attrs.nda_dst); 452 if (lle != NULL) { 453 if ((lle->la_flags & LLE_IFADDR) != 0) { 454 LLE_WUNLOCK(lle); 455 lle = NULL; 456 error = EPERM; 457 } else 458 lltable_unlink_entry(llt, lle); 459 } else 460 error = ENOENT; 461 IF_AFDATA_WUNLOCK(attrs.nda_ifp); 462 463 if (error == 0 && lle != NULL) 464 EVENTHANDLER_INVOKE(lle_event, lle, LLENTRY_DELETED); 465 466 if (lle != NULL) 467 llentry_free(lle); 468 469 return (error); 470 } 471 472 static int 473 rtnl_handle_getneigh(struct nlmsghdr *hdr, struct nlpcb *nlp, struct nl_pstate *npt) 474 { 475 int error; 476 477 struct nl_parsed_neigh attrs = {}; 478 error = nl_parse_nlmsg(hdr, &ndmsg_parser, npt, &attrs); 479 if (error != 0) 480 return (error); 481 482 if (attrs.nda_dst != NULL && attrs.nda_ifp == NULL) { 483 NLMSG_REPORT_ERR_MSG(npt, "has NDA_DST but no ifindex provided"); 484 return (EINVAL); 485 } 486 487 struct netlink_walkargs wa = { 488 .so = nlp, 489 .nw = npt->nw, 490 .hdr.nlmsg_pid = hdr->nlmsg_pid, 491 .hdr.nlmsg_seq = hdr->nlmsg_seq, 492 .hdr.nlmsg_flags = hdr->nlmsg_flags, 493 .hdr.nlmsg_type = NL_RTM_NEWNEIGH, 494 }; 495 496 if (attrs.nda_dst == NULL) 497 error = dump_llts(&wa, attrs.nda_ifp, attrs.ndm_family); 498 else 499 error = get_lle(&wa, attrs.nda_ifp, attrs.ndm_family, attrs.nda_dst); 500 501 return (error); 502 } 503 504 static const struct rtnl_cmd_handler cmd_handlers[] = { 505 { 506 .cmd = NL_RTM_NEWNEIGH, 507 .name = "RTM_NEWNEIGH", 508 .cb = &rtnl_handle_newneigh, 509 }, 510 { 511 .cmd = NL_RTM_DELNEIGH, 512 .name = "RTM_DELNEIGH", 513 .cb = &rtnl_handle_delneigh, 514 .priv = PRIV_NET_ROUTE, 515 }, 516 { 517 .cmd = NL_RTM_GETNEIGH, 518 .name = "RTM_GETNEIGH", 519 .cb = &rtnl_handle_getneigh, 520 .priv = PRIV_NET_ROUTE, 521 } 522 }; 523 524 static void 525 rtnl_lle_event(void *arg __unused, struct llentry *lle, int evt) 526 { 527 struct ifnet *ifp; 528 int family; 529 530 LLE_WLOCK_ASSERT(lle); 531 532 ifp = lltable_get_ifp(lle->lle_tbl); 533 family = lltable_get_af(lle->lle_tbl); 534 535 if (family != AF_INET && family != AF_INET6) 536 return; 537 538 int nlmsgs_type = evt == LLENTRY_RESOLVED ? NL_RTM_NEWNEIGH : NL_RTM_DELNEIGH; 539 540 struct nl_writer nw = {}; 541 if (!nlmsg_get_group_writer(&nw, NLMSG_SMALL, NETLINK_ROUTE, RTNLGRP_NEIGH)) { 542 NL_LOG(LOG_DEBUG, "error allocating group writer"); 543 return; 544 } 545 546 struct netlink_walkargs wa = { 547 .hdr.nlmsg_type = nlmsgs_type, 548 .nw = &nw, 549 .ifp = ifp, 550 .family = family, 551 }; 552 553 dump_lle_locked(lle, &wa); 554 nlmsg_flush(&nw); 555 } 556 557 static const struct nlhdr_parser *all_parsers[] = { &ndmsg_parser }; 558 559 void 560 rtnl_neighs_init() 561 { 562 NL_VERIFY_PARSERS(all_parsers); 563 rtnl_register_messages(cmd_handlers, NL_ARRAY_LEN(cmd_handlers)); 564 lle_event_p = EVENTHANDLER_REGISTER(lle_event, rtnl_lle_event, NULL, 565 EVENTHANDLER_PRI_ANY); 566 } 567 568 void 569 rtnl_neighs_destroy() 570 { 571 EVENTHANDLER_DEREGISTER(lle_event, lle_event_p); 572 } 573