1 /*- 2 * SPDX-License-Identifier: BSD-3-Clause 3 * 4 * Copyright (C) 1995, 1996, 1997, and 1998 WIDE Project. 5 * Copyright (c) 2010-2011 Juniper Networks, Inc. 6 * All rights reserved. 7 * 8 * Portions of this software were developed by Robert N. M. Watson under 9 * contract to Juniper Networks, Inc. 10 * 11 * Redistribution and use in source and binary forms, with or without 12 * modification, are permitted provided that the following conditions 13 * are met: 14 * 1. Redistributions of source code must retain the above copyright 15 * notice, this list of conditions and the following disclaimer. 16 * 2. Redistributions in binary form must reproduce the above copyright 17 * notice, this list of conditions and the following disclaimer in the 18 * documentation and/or other materials provided with the distribution. 19 * 3. Neither the name of the project nor the names of its contributors 20 * may be used to endorse or promote products derived from this software 21 * without specific prior written permission. 22 * 23 * THIS SOFTWARE IS PROVIDED BY THE PROJECT AND CONTRIBUTORS ``AS IS'' AND 24 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE 25 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE 26 * ARE DISCLAIMED. IN NO EVENT SHALL THE PROJECT OR CONTRIBUTORS BE LIABLE 27 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL 28 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS 29 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) 30 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT 31 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY 32 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF 33 * SUCH DAMAGE. 34 * 35 * $KAME: in6_pcb.c,v 1.31 2001/05/21 05:45:10 jinmei Exp $ 36 */ 37 38 /*- 39 * Copyright (c) 1982, 1986, 1991, 1993 40 * The Regents of the University of California. All rights reserved. 41 * 42 * Redistribution and use in source and binary forms, with or without 43 * modification, are permitted provided that the following conditions 44 * are met: 45 * 1. Redistributions of source code must retain the above copyright 46 * notice, this list of conditions and the following disclaimer. 47 * 2. Redistributions in binary form must reproduce the above copyright 48 * notice, this list of conditions and the following disclaimer in the 49 * documentation and/or other materials provided with the distribution. 50 * 3. Neither the name of the University nor the names of its contributors 51 * may be used to endorse or promote products derived from this software 52 * without specific prior written permission. 53 * 54 * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND 55 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE 56 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE 57 * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE 58 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL 59 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS 60 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) 61 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT 62 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY 63 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF 64 * SUCH DAMAGE. 65 * 66 * @(#)in_pcb.c 8.2 (Berkeley) 1/4/94 67 */ 68 69 #include <sys/cdefs.h> 70 __FBSDID("$FreeBSD$"); 71 72 #include "opt_inet.h" 73 #include "opt_inet6.h" 74 #include "opt_ipsec.h" 75 #include "opt_route.h" 76 #include "opt_rss.h" 77 78 #include <sys/hash.h> 79 #include <sys/param.h> 80 #include <sys/systm.h> 81 #include <sys/malloc.h> 82 #include <sys/mbuf.h> 83 #include <sys/domain.h> 84 #include <sys/protosw.h> 85 #include <sys/socket.h> 86 #include <sys/socketvar.h> 87 #include <sys/sockio.h> 88 #include <sys/errno.h> 89 #include <sys/time.h> 90 #include <sys/priv.h> 91 #include <sys/proc.h> 92 #include <sys/jail.h> 93 94 #include <vm/uma.h> 95 96 #include <net/if.h> 97 #include <net/if_var.h> 98 #include <net/if_llatbl.h> 99 #include <net/if_types.h> 100 #include <net/route.h> 101 #include <net/route/nhop.h> 102 103 #include <netinet/in.h> 104 #include <netinet/in_var.h> 105 #include <netinet/in_systm.h> 106 #include <netinet/ip6.h> 107 #include <netinet/ip_var.h> 108 109 #include <netinet6/ip6_var.h> 110 #include <netinet6/nd6.h> 111 #include <netinet/in_pcb.h> 112 #include <netinet/in_pcb_var.h> 113 #include <netinet6/in6_pcb.h> 114 #include <netinet6/in6_fib.h> 115 #include <netinet6/scope6_var.h> 116 117 int 118 in6_pcbsetport(struct in6_addr *laddr, struct inpcb *inp, struct ucred *cred) 119 { 120 struct socket *so = inp->inp_socket; 121 u_int16_t lport = 0; 122 int error, lookupflags = 0; 123 #ifdef INVARIANTS 124 struct inpcbinfo *pcbinfo = inp->inp_pcbinfo; 125 #endif 126 127 INP_WLOCK_ASSERT(inp); 128 INP_HASH_WLOCK_ASSERT(pcbinfo); 129 130 error = prison_local_ip6(cred, laddr, 131 ((inp->inp_flags & IN6P_IPV6_V6ONLY) != 0)); 132 if (error) 133 return(error); 134 135 /* XXX: this is redundant when called from in6_pcbbind */ 136 if ((so->so_options & (SO_REUSEADDR|SO_REUSEPORT|SO_REUSEPORT_LB)) == 0) 137 lookupflags = INPLOOKUP_WILDCARD; 138 139 inp->inp_flags |= INP_ANONPORT; 140 141 error = in_pcb_lport(inp, NULL, &lport, cred, lookupflags); 142 if (error != 0) 143 return (error); 144 145 inp->inp_lport = lport; 146 if (in_pcbinshash(inp) != 0) { 147 inp->in6p_laddr = in6addr_any; 148 inp->inp_lport = 0; 149 return (EAGAIN); 150 } 151 152 return (0); 153 } 154 155 int 156 in6_pcbbind(struct inpcb *inp, struct sockaddr_in6 *sin6, struct ucred *cred) 157 { 158 struct socket *so = inp->inp_socket; 159 struct inpcbinfo *pcbinfo = inp->inp_pcbinfo; 160 u_short lport = 0; 161 int error, lookupflags = 0; 162 int reuseport = (so->so_options & SO_REUSEPORT); 163 164 /* 165 * XXX: Maybe we could let SO_REUSEPORT_LB set SO_REUSEPORT bit here 166 * so that we don't have to add to the (already messy) code below. 167 */ 168 int reuseport_lb = (so->so_options & SO_REUSEPORT_LB); 169 170 INP_WLOCK_ASSERT(inp); 171 INP_HASH_WLOCK_ASSERT(pcbinfo); 172 173 if (inp->inp_lport || !IN6_IS_ADDR_UNSPECIFIED(&inp->in6p_laddr)) 174 return (EINVAL); 175 if ((so->so_options & (SO_REUSEADDR|SO_REUSEPORT|SO_REUSEPORT_LB)) == 0) 176 lookupflags = INPLOOKUP_WILDCARD; 177 if (sin6 == NULL) { 178 if ((error = prison_local_ip6(cred, &inp->in6p_laddr, 179 ((inp->inp_flags & IN6P_IPV6_V6ONLY) != 0))) != 0) 180 return (error); 181 } else { 182 KASSERT(sin6->sin6_family == AF_INET6, 183 ("%s: invalid address family for %p", __func__, sin6)); 184 KASSERT(sin6->sin6_len == sizeof(*sin6), 185 ("%s: invalid address length for %p", __func__, sin6)); 186 187 if ((error = sa6_embedscope(sin6, V_ip6_use_defzone)) != 0) 188 return(error); 189 190 if ((error = prison_local_ip6(cred, &sin6->sin6_addr, 191 ((inp->inp_flags & IN6P_IPV6_V6ONLY) != 0))) != 0) 192 return (error); 193 194 lport = sin6->sin6_port; 195 if (IN6_IS_ADDR_MULTICAST(&sin6->sin6_addr)) { 196 /* 197 * Treat SO_REUSEADDR as SO_REUSEPORT for multicast; 198 * allow compepte duplication of binding if 199 * SO_REUSEPORT is set, or if SO_REUSEADDR is set 200 * and a multicast address is bound on both 201 * new and duplicated sockets. 202 */ 203 if ((so->so_options & (SO_REUSEADDR|SO_REUSEPORT)) != 0) 204 reuseport = SO_REUSEADDR|SO_REUSEPORT; 205 /* 206 * XXX: How to deal with SO_REUSEPORT_LB here? 207 * Treat same as SO_REUSEPORT for now. 208 */ 209 if ((so->so_options & 210 (SO_REUSEADDR|SO_REUSEPORT_LB)) != 0) 211 reuseport_lb = SO_REUSEADDR|SO_REUSEPORT_LB; 212 } else if (!IN6_IS_ADDR_UNSPECIFIED(&sin6->sin6_addr)) { 213 struct epoch_tracker et; 214 struct ifaddr *ifa; 215 216 sin6->sin6_port = 0; /* yech... */ 217 NET_EPOCH_ENTER(et); 218 if ((ifa = ifa_ifwithaddr((struct sockaddr *)sin6)) == 219 NULL && 220 (inp->inp_flags & INP_BINDANY) == 0) { 221 NET_EPOCH_EXIT(et); 222 return (EADDRNOTAVAIL); 223 } 224 225 /* 226 * XXX: bind to an anycast address might accidentally 227 * cause sending a packet with anycast source address. 228 * We should allow to bind to a deprecated address, since 229 * the application dares to use it. 230 */ 231 if (ifa != NULL && 232 ((struct in6_ifaddr *)ifa)->ia6_flags & 233 (IN6_IFF_ANYCAST|IN6_IFF_NOTREADY|IN6_IFF_DETACHED)) { 234 NET_EPOCH_EXIT(et); 235 return (EADDRNOTAVAIL); 236 } 237 NET_EPOCH_EXIT(et); 238 } 239 if (lport) { 240 struct inpcb *t; 241 242 /* GROSS */ 243 if (ntohs(lport) <= V_ipport_reservedhigh && 244 ntohs(lport) >= V_ipport_reservedlow && 245 priv_check_cred(cred, PRIV_NETINET_RESERVEDPORT)) 246 return (EACCES); 247 if (!IN6_IS_ADDR_MULTICAST(&sin6->sin6_addr) && 248 priv_check_cred(inp->inp_cred, PRIV_NETINET_REUSEPORT) != 0) { 249 t = in6_pcblookup_local(pcbinfo, 250 &sin6->sin6_addr, lport, 251 INPLOOKUP_WILDCARD, cred); 252 if (t != NULL && 253 (so->so_type != SOCK_STREAM || 254 IN6_IS_ADDR_UNSPECIFIED(&t->in6p_faddr)) && 255 (!IN6_IS_ADDR_UNSPECIFIED(&sin6->sin6_addr) || 256 !IN6_IS_ADDR_UNSPECIFIED(&t->in6p_laddr) || 257 (t->inp_flags2 & INP_REUSEPORT) || 258 (t->inp_flags2 & INP_REUSEPORT_LB) == 0) && 259 (inp->inp_cred->cr_uid != 260 t->inp_cred->cr_uid)) 261 return (EADDRINUSE); 262 263 #ifdef INET 264 if ((inp->inp_flags & IN6P_IPV6_V6ONLY) == 0 && 265 IN6_IS_ADDR_UNSPECIFIED(&sin6->sin6_addr)) { 266 struct sockaddr_in sin; 267 268 in6_sin6_2_sin(&sin, sin6); 269 t = in_pcblookup_local(pcbinfo, 270 sin.sin_addr, lport, 271 INPLOOKUP_WILDCARD, cred); 272 if (t != NULL && 273 (so->so_type != SOCK_STREAM || 274 ntohl(t->inp_faddr.s_addr) == 275 INADDR_ANY) && 276 (inp->inp_cred->cr_uid != 277 t->inp_cred->cr_uid)) 278 return (EADDRINUSE); 279 } 280 #endif 281 } 282 t = in6_pcblookup_local(pcbinfo, &sin6->sin6_addr, 283 lport, lookupflags, cred); 284 if (t && (reuseport & inp_so_options(t)) == 0 && 285 (reuseport_lb & inp_so_options(t)) == 0) { 286 return (EADDRINUSE); 287 } 288 #ifdef INET 289 if ((inp->inp_flags & IN6P_IPV6_V6ONLY) == 0 && 290 IN6_IS_ADDR_UNSPECIFIED(&sin6->sin6_addr)) { 291 struct sockaddr_in sin; 292 293 in6_sin6_2_sin(&sin, sin6); 294 t = in_pcblookup_local(pcbinfo, sin.sin_addr, 295 lport, lookupflags, cred); 296 if (t && 297 (reuseport & inp_so_options(t)) == 0 && 298 (reuseport_lb & inp_so_options(t)) == 0 && 299 (ntohl(t->inp_laddr.s_addr) != INADDR_ANY || 300 (t->inp_vflag & INP_IPV6PROTO) != 0)) { 301 return (EADDRINUSE); 302 } 303 } 304 #endif 305 } 306 inp->in6p_laddr = sin6->sin6_addr; 307 } 308 if (lport == 0) { 309 if ((error = in6_pcbsetport(&inp->in6p_laddr, inp, cred)) != 0) { 310 /* Undo an address bind that may have occurred. */ 311 inp->in6p_laddr = in6addr_any; 312 return (error); 313 } 314 } else { 315 inp->inp_lport = lport; 316 if (in_pcbinshash(inp) != 0) { 317 inp->in6p_laddr = in6addr_any; 318 inp->inp_lport = 0; 319 return (EAGAIN); 320 } 321 } 322 return (0); 323 } 324 325 /* 326 * Transform old in6_pcbconnect() into an inner subroutine for new 327 * in6_pcbconnect(): Do some validity-checking on the remote 328 * address (in mbuf 'nam') and then determine local host address 329 * (i.e., which interface) to use to access that remote host. 330 * 331 * This preserves definition of in6_pcbconnect(), while supporting a 332 * slightly different version for T/TCP. (This is more than 333 * a bit of a kludge, but cleaning up the internal interfaces would 334 * have forced minor changes in every protocol). 335 */ 336 static int 337 in6_pcbladdr(struct inpcb *inp, struct sockaddr_in6 *sin6, 338 struct in6_addr *plocal_addr6) 339 { 340 int error = 0; 341 int scope_ambiguous = 0; 342 struct in6_addr in6a; 343 344 NET_EPOCH_ASSERT(); 345 INP_WLOCK_ASSERT(inp); 346 INP_HASH_WLOCK_ASSERT(inp->inp_pcbinfo); /* XXXRW: why? */ 347 348 if (sin6->sin6_port == 0) 349 return (EADDRNOTAVAIL); 350 351 if (sin6->sin6_scope_id == 0 && !V_ip6_use_defzone) 352 scope_ambiguous = 1; 353 if ((error = sa6_embedscope(sin6, V_ip6_use_defzone)) != 0) 354 return(error); 355 356 if (!CK_STAILQ_EMPTY(&V_in6_ifaddrhead)) { 357 /* 358 * If the destination address is UNSPECIFIED addr, 359 * use the loopback addr, e.g ::1. 360 */ 361 if (IN6_IS_ADDR_UNSPECIFIED(&sin6->sin6_addr)) 362 sin6->sin6_addr = in6addr_loopback; 363 } 364 if ((error = prison_remote_ip6(inp->inp_cred, &sin6->sin6_addr)) != 0) 365 return (error); 366 367 error = in6_selectsrc_socket(sin6, inp->in6p_outputopts, 368 inp, inp->inp_cred, scope_ambiguous, &in6a, NULL); 369 if (error) 370 return (error); 371 372 /* 373 * Do not update this earlier, in case we return with an error. 374 * 375 * XXX: this in6_selectsrc_socket result might replace the bound local 376 * address with the address specified by setsockopt(IPV6_PKTINFO). 377 * Is it the intended behavior? 378 */ 379 *plocal_addr6 = in6a; 380 381 /* 382 * Don't do pcblookup call here; return interface in 383 * plocal_addr6 384 * and exit to caller, that will do the lookup. 385 */ 386 387 return (0); 388 } 389 390 /* 391 * Outer subroutine: 392 * Connect from a socket to a specified address. 393 * Both address and port must be specified in argument sin. 394 * If don't have a local address for this socket yet, 395 * then pick one. 396 */ 397 int 398 in6_pcbconnect(struct inpcb *inp, struct sockaddr_in6 *sin6, struct ucred *cred, 399 bool rehash) 400 { 401 struct inpcbinfo *pcbinfo = inp->inp_pcbinfo; 402 struct sockaddr_in6 laddr6; 403 int error; 404 405 NET_EPOCH_ASSERT(); 406 INP_WLOCK_ASSERT(inp); 407 INP_HASH_WLOCK_ASSERT(pcbinfo); 408 KASSERT(sin6->sin6_family == AF_INET6, 409 ("%s: invalid address family for %p", __func__, sin6)); 410 KASSERT(sin6->sin6_len == sizeof(*sin6), 411 ("%s: invalid address length for %p", __func__, sin6)); 412 413 bzero(&laddr6, sizeof(laddr6)); 414 laddr6.sin6_family = AF_INET6; 415 416 #ifdef ROUTE_MPATH 417 if (CALC_FLOWID_OUTBOUND) { 418 uint32_t hash_type, hash_val; 419 420 hash_val = fib6_calc_software_hash(&inp->in6p_laddr, 421 &sin6->sin6_addr, 0, sin6->sin6_port, 422 inp->inp_socket->so_proto->pr_protocol, &hash_type); 423 inp->inp_flowid = hash_val; 424 inp->inp_flowtype = hash_type; 425 } 426 #endif 427 /* 428 * Call inner routine, to assign local interface address. 429 * in6_pcbladdr() may automatically fill in sin6_scope_id. 430 */ 431 if ((error = in6_pcbladdr(inp, sin6, &laddr6.sin6_addr)) != 0) 432 return (error); 433 434 if (in6_pcblookup_hash_locked(pcbinfo, &sin6->sin6_addr, 435 sin6->sin6_port, IN6_IS_ADDR_UNSPECIFIED(&inp->in6p_laddr) ? 436 &laddr6.sin6_addr : &inp->in6p_laddr, inp->inp_lport, 0, 437 M_NODOM) != NULL) 438 return (EADDRINUSE); 439 if (IN6_IS_ADDR_UNSPECIFIED(&inp->in6p_laddr)) { 440 if (inp->inp_lport == 0) { 441 /* 442 * rehash was required to be true in the past for 443 * this case; retain that convention. However, 444 * we now call in_pcb_lport_dest rather than 445 * in6_pcbbind; the former does not insert into 446 * the hash table, the latter does. Change rehash 447 * to false to do the in_pcbinshash below. 448 */ 449 KASSERT(rehash == true, 450 ("Rehashing required for unbound inps")); 451 rehash = false; 452 error = in_pcb_lport_dest(inp, 453 (struct sockaddr *) &laddr6, &inp->inp_lport, 454 (struct sockaddr *) sin6, sin6->sin6_port, cred, 455 INPLOOKUP_WILDCARD); 456 if (error) 457 return (error); 458 } 459 inp->in6p_laddr = laddr6.sin6_addr; 460 } 461 inp->in6p_faddr = sin6->sin6_addr; 462 inp->inp_fport = sin6->sin6_port; 463 /* update flowinfo - draft-itojun-ipv6-flowlabel-api-00 */ 464 inp->inp_flow &= ~IPV6_FLOWLABEL_MASK; 465 if (inp->inp_flags & IN6P_AUTOFLOWLABEL) 466 inp->inp_flow |= 467 (htonl(ip6_randomflowlabel()) & IPV6_FLOWLABEL_MASK); 468 469 if (rehash) { 470 in_pcbrehash(inp); 471 } else { 472 in_pcbinshash(inp); 473 } 474 475 return (0); 476 } 477 478 void 479 in6_pcbdisconnect(struct inpcb *inp) 480 { 481 482 INP_WLOCK_ASSERT(inp); 483 INP_HASH_WLOCK_ASSERT(inp->inp_pcbinfo); 484 485 memset(&inp->in6p_laddr, 0, sizeof(inp->in6p_laddr)); 486 memset(&inp->in6p_faddr, 0, sizeof(inp->in6p_faddr)); 487 inp->inp_fport = 0; 488 /* clear flowinfo - draft-itojun-ipv6-flowlabel-api-00 */ 489 inp->inp_flow &= ~IPV6_FLOWLABEL_MASK; 490 in_pcbrehash(inp); 491 } 492 493 struct sockaddr * 494 in6_sockaddr(in_port_t port, struct in6_addr *addr_p) 495 { 496 struct sockaddr_in6 *sin6; 497 498 sin6 = malloc(sizeof *sin6, M_SONAME, M_WAITOK); 499 bzero(sin6, sizeof *sin6); 500 sin6->sin6_family = AF_INET6; 501 sin6->sin6_len = sizeof(*sin6); 502 sin6->sin6_port = port; 503 sin6->sin6_addr = *addr_p; 504 (void)sa6_recoverscope(sin6); /* XXX: should catch errors */ 505 506 return (struct sockaddr *)sin6; 507 } 508 509 struct sockaddr * 510 in6_v4mapsin6_sockaddr(in_port_t port, struct in_addr *addr_p) 511 { 512 struct sockaddr_in sin; 513 struct sockaddr_in6 *sin6_p; 514 515 bzero(&sin, sizeof sin); 516 sin.sin_family = AF_INET; 517 sin.sin_len = sizeof(sin); 518 sin.sin_port = port; 519 sin.sin_addr = *addr_p; 520 521 sin6_p = malloc(sizeof *sin6_p, M_SONAME, 522 M_WAITOK); 523 in6_sin_2_v4mapsin6(&sin, sin6_p); 524 525 return (struct sockaddr *)sin6_p; 526 } 527 528 int 529 in6_getsockaddr(struct socket *so, struct sockaddr **nam) 530 { 531 struct inpcb *inp; 532 struct in6_addr addr; 533 in_port_t port; 534 535 inp = sotoinpcb(so); 536 KASSERT(inp != NULL, ("in6_getsockaddr: inp == NULL")); 537 538 INP_RLOCK(inp); 539 port = inp->inp_lport; 540 addr = inp->in6p_laddr; 541 INP_RUNLOCK(inp); 542 543 *nam = in6_sockaddr(port, &addr); 544 return 0; 545 } 546 547 int 548 in6_getpeeraddr(struct socket *so, struct sockaddr **nam) 549 { 550 struct inpcb *inp; 551 struct in6_addr addr; 552 in_port_t port; 553 554 inp = sotoinpcb(so); 555 KASSERT(inp != NULL, ("in6_getpeeraddr: inp == NULL")); 556 557 INP_RLOCK(inp); 558 port = inp->inp_fport; 559 addr = inp->in6p_faddr; 560 INP_RUNLOCK(inp); 561 562 *nam = in6_sockaddr(port, &addr); 563 return 0; 564 } 565 566 int 567 in6_mapped_sockaddr(struct socket *so, struct sockaddr **nam) 568 { 569 struct inpcb *inp; 570 int error; 571 572 inp = sotoinpcb(so); 573 KASSERT(inp != NULL, ("in6_mapped_sockaddr: inp == NULL")); 574 575 #ifdef INET 576 if ((inp->inp_vflag & (INP_IPV4 | INP_IPV6)) == INP_IPV4) { 577 error = in_getsockaddr(so, nam); 578 if (error == 0) 579 in6_sin_2_v4mapsin6_in_sock(nam); 580 } else 581 #endif 582 { 583 /* scope issues will be handled in in6_getsockaddr(). */ 584 error = in6_getsockaddr(so, nam); 585 } 586 587 return error; 588 } 589 590 int 591 in6_mapped_peeraddr(struct socket *so, struct sockaddr **nam) 592 { 593 struct inpcb *inp; 594 int error; 595 596 inp = sotoinpcb(so); 597 KASSERT(inp != NULL, ("in6_mapped_peeraddr: inp == NULL")); 598 599 #ifdef INET 600 if ((inp->inp_vflag & (INP_IPV4 | INP_IPV6)) == INP_IPV4) { 601 error = in_getpeeraddr(so, nam); 602 if (error == 0) 603 in6_sin_2_v4mapsin6_in_sock(nam); 604 } else 605 #endif 606 /* scope issues will be handled in in6_getpeeraddr(). */ 607 error = in6_getpeeraddr(so, nam); 608 609 return error; 610 } 611 612 /* 613 * Pass some notification to all connections of a protocol 614 * associated with address dst. The local address and/or port numbers 615 * may be specified to limit the search. The "usual action" will be 616 * taken, depending on the ctlinput cmd. The caller must filter any 617 * cmds that are uninteresting (e.g., no error in the map). 618 * Call the protocol specific routine (if any) to report 619 * any errors for each matching socket. 620 */ 621 static bool 622 inp_match6(const struct inpcb *inp, void *v __unused) 623 { 624 625 return ((inp->inp_vflag & INP_IPV6) != 0); 626 } 627 628 void 629 in6_pcbnotify(struct inpcbinfo *pcbinfo, struct sockaddr_in6 *sa6_dst, 630 u_int fport_arg, const struct sockaddr_in6 *src, u_int lport_arg, 631 int errno, void *cmdarg, 632 struct inpcb *(*notify)(struct inpcb *, int)) 633 { 634 struct inpcb_iterator inpi = INP_ITERATOR(pcbinfo, INPLOOKUP_WLOCKPCB, 635 inp_match6, NULL); 636 struct inpcb *inp; 637 struct sockaddr_in6 sa6_src; 638 u_short fport = fport_arg, lport = lport_arg; 639 u_int32_t flowinfo; 640 641 if (IN6_IS_ADDR_UNSPECIFIED(&sa6_dst->sin6_addr)) 642 return; 643 644 /* 645 * note that src can be NULL when we get notify by local fragmentation. 646 */ 647 sa6_src = (src == NULL) ? sa6_any : *src; 648 flowinfo = sa6_src.sin6_flowinfo; 649 650 while ((inp = inp_next(&inpi)) != NULL) { 651 INP_WLOCK_ASSERT(inp); 652 /* 653 * If the error designates a new path MTU for a destination 654 * and the application (associated with this socket) wanted to 655 * know the value, notify. 656 * XXX: should we avoid to notify the value to TCP sockets? 657 */ 658 if (errno == EMSGSIZE && cmdarg != NULL) 659 ip6_notify_pmtu(inp, sa6_dst, *(uint32_t *)cmdarg); 660 661 /* 662 * Detect if we should notify the error. If no source and 663 * destination ports are specified, but non-zero flowinfo and 664 * local address match, notify the error. This is the case 665 * when the error is delivered with an encrypted buffer 666 * by ESP. Otherwise, just compare addresses and ports 667 * as usual. 668 */ 669 if (lport == 0 && fport == 0 && flowinfo && 670 inp->inp_socket != NULL && 671 flowinfo == (inp->inp_flow & IPV6_FLOWLABEL_MASK) && 672 IN6_ARE_ADDR_EQUAL(&inp->in6p_laddr, &sa6_src.sin6_addr)) 673 goto do_notify; 674 else if (!IN6_ARE_ADDR_EQUAL(&inp->in6p_faddr, 675 &sa6_dst->sin6_addr) || 676 inp->inp_socket == 0 || 677 (lport && inp->inp_lport != lport) || 678 (!IN6_IS_ADDR_UNSPECIFIED(&sa6_src.sin6_addr) && 679 !IN6_ARE_ADDR_EQUAL(&inp->in6p_laddr, 680 &sa6_src.sin6_addr)) || 681 (fport && inp->inp_fport != fport)) { 682 continue; 683 } 684 685 do_notify: 686 if (notify) 687 (*notify)(inp, errno); 688 } 689 } 690 691 /* 692 * Lookup a PCB based on the local address and port. Caller must hold the 693 * hash lock. No inpcb locks or references are acquired. 694 */ 695 struct inpcb * 696 in6_pcblookup_local(struct inpcbinfo *pcbinfo, struct in6_addr *laddr, 697 u_short lport, int lookupflags, struct ucred *cred) 698 { 699 struct inpcb *inp; 700 int matchwild = 3, wildcard; 701 702 KASSERT((lookupflags & ~(INPLOOKUP_WILDCARD)) == 0, 703 ("%s: invalid lookup flags %d", __func__, lookupflags)); 704 705 INP_HASH_LOCK_ASSERT(pcbinfo); 706 707 if ((lookupflags & INPLOOKUP_WILDCARD) == 0) { 708 struct inpcbhead *head; 709 /* 710 * Look for an unconnected (wildcard foreign addr) PCB that 711 * matches the local address and port we're looking for. 712 */ 713 head = &pcbinfo->ipi_hashbase[INP_PCBHASH_WILD(lport, 714 pcbinfo->ipi_hashmask)]; 715 CK_LIST_FOREACH(inp, head, inp_hash) { 716 /* XXX inp locking */ 717 if ((inp->inp_vflag & INP_IPV6) == 0) 718 continue; 719 if (IN6_IS_ADDR_UNSPECIFIED(&inp->in6p_faddr) && 720 IN6_ARE_ADDR_EQUAL(&inp->in6p_laddr, laddr) && 721 inp->inp_lport == lport) { 722 /* Found. */ 723 if (prison_equal_ip6(cred->cr_prison, 724 inp->inp_cred->cr_prison)) 725 return (inp); 726 } 727 } 728 /* 729 * Not found. 730 */ 731 return (NULL); 732 } else { 733 struct inpcbporthead *porthash; 734 struct inpcbport *phd; 735 struct inpcb *match = NULL; 736 /* 737 * Best fit PCB lookup. 738 * 739 * First see if this local port is in use by looking on the 740 * port hash list. 741 */ 742 porthash = &pcbinfo->ipi_porthashbase[INP_PCBPORTHASH(lport, 743 pcbinfo->ipi_porthashmask)]; 744 CK_LIST_FOREACH(phd, porthash, phd_hash) { 745 if (phd->phd_port == lport) 746 break; 747 } 748 if (phd != NULL) { 749 /* 750 * Port is in use by one or more PCBs. Look for best 751 * fit. 752 */ 753 CK_LIST_FOREACH(inp, &phd->phd_pcblist, inp_portlist) { 754 wildcard = 0; 755 if (!prison_equal_ip6(cred->cr_prison, 756 inp->inp_cred->cr_prison)) 757 continue; 758 /* XXX inp locking */ 759 if ((inp->inp_vflag & INP_IPV6) == 0) 760 continue; 761 if (!IN6_IS_ADDR_UNSPECIFIED(&inp->in6p_faddr)) 762 wildcard++; 763 if (!IN6_IS_ADDR_UNSPECIFIED( 764 &inp->in6p_laddr)) { 765 if (IN6_IS_ADDR_UNSPECIFIED(laddr)) 766 wildcard++; 767 else if (!IN6_ARE_ADDR_EQUAL( 768 &inp->in6p_laddr, laddr)) 769 continue; 770 } else { 771 if (!IN6_IS_ADDR_UNSPECIFIED(laddr)) 772 wildcard++; 773 } 774 if (wildcard < matchwild) { 775 match = inp; 776 matchwild = wildcard; 777 if (matchwild == 0) 778 break; 779 } 780 } 781 } 782 return (match); 783 } 784 } 785 786 static bool 787 in6_multi_match(const struct inpcb *inp, void *v __unused) 788 { 789 790 if ((inp->inp_vflag & INP_IPV6) && inp->in6p_moptions != NULL) 791 return (true); 792 else 793 return (false); 794 } 795 796 void 797 in6_pcbpurgeif0(struct inpcbinfo *pcbinfo, struct ifnet *ifp) 798 { 799 struct inpcb_iterator inpi = INP_ITERATOR(pcbinfo, INPLOOKUP_RLOCKPCB, 800 in6_multi_match, NULL); 801 struct inpcb *inp; 802 struct in6_multi *inm; 803 struct in6_mfilter *imf; 804 struct ip6_moptions *im6o; 805 806 IN6_MULTI_LOCK_ASSERT(); 807 808 while ((inp = inp_next(&inpi)) != NULL) { 809 INP_RLOCK_ASSERT(inp); 810 811 im6o = inp->in6p_moptions; 812 /* 813 * Unselect the outgoing ifp for multicast if it 814 * is being detached. 815 */ 816 if (im6o->im6o_multicast_ifp == ifp) 817 im6o->im6o_multicast_ifp = NULL; 818 /* 819 * Drop multicast group membership if we joined 820 * through the interface being detached. 821 */ 822 restart: 823 IP6_MFILTER_FOREACH(imf, &im6o->im6o_head) { 824 if ((inm = imf->im6f_in6m) == NULL) 825 continue; 826 if (inm->in6m_ifp != ifp) 827 continue; 828 ip6_mfilter_remove(&im6o->im6o_head, imf); 829 in6_leavegroup_locked(inm, NULL); 830 ip6_mfilter_free(imf); 831 goto restart; 832 } 833 } 834 } 835 836 /* 837 * Check for alternatives when higher level complains 838 * about service problems. For now, invalidate cached 839 * routing information. If the route was created dynamically 840 * (by a redirect), time to try a default gateway again. 841 */ 842 void 843 in6_losing(struct inpcb *inp) 844 { 845 846 RO_INVALIDATE_CACHE(&inp->inp_route6); 847 } 848 849 /* 850 * After a routing change, flush old routing 851 * and allocate a (hopefully) better one. 852 */ 853 struct inpcb * 854 in6_rtchange(struct inpcb *inp, int errno __unused) 855 { 856 857 RO_INVALIDATE_CACHE(&inp->inp_route6); 858 return inp; 859 } 860 861 static bool 862 in6_pcblookup_lb_numa_match(const struct inpcblbgroup *grp, int domain) 863 { 864 return (domain == M_NODOM || domain == grp->il_numa_domain); 865 } 866 867 static struct inpcb * 868 in6_pcblookup_lbgroup(const struct inpcbinfo *pcbinfo, 869 const struct in6_addr *faddr, uint16_t fport, const struct in6_addr *laddr, 870 uint16_t lport, uint8_t domain) 871 { 872 const struct inpcblbgrouphead *hdr; 873 struct inpcblbgroup *grp; 874 struct inpcblbgroup *jail_exact, *jail_wild, *local_exact, *local_wild; 875 876 INP_HASH_LOCK_ASSERT(pcbinfo); 877 878 hdr = &pcbinfo->ipi_lbgrouphashbase[ 879 INP_PCBPORTHASH(lport, pcbinfo->ipi_lbgrouphashmask)]; 880 881 /* 882 * Search for an LB group match based on the following criteria: 883 * - prefer jailed groups to non-jailed groups 884 * - prefer exact source address matches to wildcard matches 885 * - prefer groups bound to the specified NUMA domain 886 */ 887 jail_exact = jail_wild = local_exact = local_wild = NULL; 888 CK_LIST_FOREACH(grp, hdr, il_list) { 889 bool injail; 890 891 #ifdef INET 892 if (!(grp->il_vflag & INP_IPV6)) 893 continue; 894 #endif 895 if (grp->il_lport != lport) 896 continue; 897 898 injail = prison_flag(grp->il_cred, PR_IP6) != 0; 899 if (injail && prison_check_ip6_locked(grp->il_cred->cr_prison, 900 laddr) != 0) 901 continue; 902 903 if (IN6_ARE_ADDR_EQUAL(&grp->il6_laddr, laddr)) { 904 if (injail) { 905 jail_exact = grp; 906 if (in6_pcblookup_lb_numa_match(grp, domain)) 907 /* This is a perfect match. */ 908 goto out; 909 } else if (local_exact == NULL || 910 in6_pcblookup_lb_numa_match(grp, domain)) { 911 local_exact = grp; 912 } 913 } else if (IN6_IS_ADDR_UNSPECIFIED(&grp->il6_laddr)) { 914 if (injail) { 915 if (jail_wild == NULL || 916 in6_pcblookup_lb_numa_match(grp, domain)) 917 jail_wild = grp; 918 } else if (local_wild == NULL || 919 in6_pcblookup_lb_numa_match(grp, domain)) { 920 local_wild = grp; 921 } 922 } 923 } 924 925 if (jail_exact != NULL) 926 grp = jail_exact; 927 else if (jail_wild != NULL) 928 grp = jail_wild; 929 else if (local_exact != NULL) 930 grp = local_exact; 931 else 932 grp = local_wild; 933 if (grp == NULL) 934 return (NULL); 935 out: 936 return (grp->il_inp[INP6_PCBLBGROUP_PKTHASH(faddr, lport, fport) % 937 grp->il_inpcnt]); 938 } 939 940 static struct inpcb * 941 in6_pcblookup_hash_exact(struct inpcbinfo *pcbinfo, struct in6_addr *faddr, 942 u_short fport, struct in6_addr *laddr, u_short lport) 943 { 944 struct inpcbhead *head; 945 struct inpcb *inp, *match; 946 947 INP_HASH_LOCK_ASSERT(pcbinfo); 948 949 /* 950 * First look for an exact match. 951 */ 952 match = NULL; 953 head = &pcbinfo->ipi_hashbase[INP6_PCBHASH(faddr, lport, fport, 954 pcbinfo->ipi_hashmask)]; 955 CK_LIST_FOREACH(inp, head, inp_hash) { 956 /* XXX inp locking */ 957 if ((inp->inp_vflag & INP_IPV6) == 0) 958 continue; 959 if (IN6_ARE_ADDR_EQUAL(&inp->in6p_faddr, faddr) && 960 IN6_ARE_ADDR_EQUAL(&inp->in6p_laddr, laddr) && 961 inp->inp_fport == fport && 962 inp->inp_lport == lport) 963 return (inp); 964 } 965 return (match); 966 } 967 968 static struct inpcb * 969 in6_pcblookup_hash_wild_locked(struct inpcbinfo *pcbinfo, 970 struct in6_addr *faddr, u_short fport, struct in6_addr *laddr, 971 u_short lport) 972 { 973 struct inpcbhead *head; 974 struct inpcb *inp, *jail_wild, *local_exact, *local_wild; 975 976 /* 977 * Order of socket selection - we always prefer jails. 978 * 1. jailed, non-wild. 979 * 2. jailed, wild. 980 * 3. non-jailed, non-wild. 981 * 4. non-jailed, wild. 982 */ 983 head = &pcbinfo->ipi_hashbase[INP_PCBHASH_WILD(lport, 984 pcbinfo->ipi_hashmask)]; 985 local_wild = local_exact = jail_wild = NULL; 986 CK_LIST_FOREACH(inp, head, inp_hash) { 987 bool injail; 988 989 /* XXX inp locking */ 990 if ((inp->inp_vflag & INP_IPV6) == 0) 991 continue; 992 993 if (!IN6_IS_ADDR_UNSPECIFIED(&inp->in6p_faddr) || 994 inp->inp_lport != lport) { 995 continue; 996 } 997 998 injail = prison_flag(inp->inp_cred, PR_IP6) != 0; 999 if (injail) { 1000 if (prison_check_ip6_locked( 1001 inp->inp_cred->cr_prison, laddr) != 0) 1002 continue; 1003 } else { 1004 if (local_exact != NULL) 1005 continue; 1006 } 1007 1008 if (IN6_ARE_ADDR_EQUAL(&inp->in6p_laddr, laddr)) { 1009 if (injail) 1010 return (inp); 1011 else 1012 local_exact = inp; 1013 } else if (IN6_IS_ADDR_UNSPECIFIED(&inp->in6p_laddr)) { 1014 if (injail) 1015 jail_wild = inp; 1016 else 1017 local_wild = inp; 1018 } 1019 } 1020 1021 if (jail_wild != NULL) 1022 return (jail_wild); 1023 if (local_exact != NULL) 1024 return (local_exact); 1025 if (local_wild != NULL) 1026 return (local_wild); 1027 return (NULL); 1028 } 1029 1030 struct inpcb * 1031 in6_pcblookup_hash_locked(struct inpcbinfo *pcbinfo, struct in6_addr *faddr, 1032 u_int fport_arg, struct in6_addr *laddr, u_int lport_arg, 1033 int lookupflags, uint8_t numa_domain) 1034 { 1035 struct inpcb *inp; 1036 u_short fport = fport_arg, lport = lport_arg; 1037 1038 KASSERT((lookupflags & ~INPLOOKUP_WILDCARD) == 0, 1039 ("%s: invalid lookup flags %d", __func__, lookupflags)); 1040 KASSERT(!IN6_IS_ADDR_UNSPECIFIED(faddr), 1041 ("%s: invalid foreign address", __func__)); 1042 KASSERT(!IN6_IS_ADDR_UNSPECIFIED(laddr), 1043 ("%s: invalid local address", __func__)); 1044 1045 INP_HASH_LOCK_ASSERT(pcbinfo); 1046 1047 inp = in6_pcblookup_hash_exact(pcbinfo, faddr, fport, laddr, lport); 1048 if (inp != NULL) 1049 return (inp); 1050 1051 if ((lookupflags & INPLOOKUP_WILDCARD) != 0) { 1052 inp = in6_pcblookup_lbgroup(pcbinfo, faddr, fport, laddr, 1053 lport, numa_domain); 1054 if (inp == NULL) { 1055 inp = in6_pcblookup_hash_wild_locked(pcbinfo, faddr, 1056 fport, laddr, lport); 1057 } 1058 } 1059 return (inp); 1060 } 1061 1062 static struct inpcb * 1063 in6_pcblookup_hash_smr(struct inpcbinfo *pcbinfo, struct in6_addr *faddr, 1064 u_int fport, struct in6_addr *laddr, u_int lport, int lookupflags, 1065 uint8_t numa_domain) 1066 { 1067 struct inpcb *inp; 1068 1069 KASSERT((lookupflags & ~INPLOOKUP_MASK) == 0, 1070 ("%s: invalid lookup flags %d", __func__, lookupflags)); 1071 KASSERT((lookupflags & (INPLOOKUP_RLOCKPCB | INPLOOKUP_WLOCKPCB)) != 0, 1072 ("%s: LOCKPCB not set", __func__)); 1073 1074 smr_enter(pcbinfo->ipi_smr); 1075 inp = in6_pcblookup_hash_locked(pcbinfo, faddr, fport, laddr, lport, 1076 lookupflags & INPLOOKUP_WILDCARD, numa_domain); 1077 if (inp != NULL) { 1078 if (__predict_false(inp_smr_lock(inp, 1079 (lookupflags & INPLOOKUP_LOCKMASK)) == false)) 1080 inp = NULL; 1081 } else 1082 smr_exit(pcbinfo->ipi_smr); 1083 1084 return (inp); 1085 } 1086 1087 /* 1088 * Public inpcb lookup routines, accepting a 4-tuple, and optionally, an mbuf 1089 * from which a pre-calculated hash value may be extracted. 1090 */ 1091 struct inpcb * 1092 in6_pcblookup(struct inpcbinfo *pcbinfo, struct in6_addr *faddr, u_int fport, 1093 struct in6_addr *laddr, u_int lport, int lookupflags, 1094 struct ifnet *ifp __unused) 1095 { 1096 return (in6_pcblookup_hash_smr(pcbinfo, faddr, fport, laddr, lport, 1097 lookupflags, M_NODOM)); 1098 } 1099 1100 struct inpcb * 1101 in6_pcblookup_mbuf(struct inpcbinfo *pcbinfo, struct in6_addr *faddr, 1102 u_int fport, struct in6_addr *laddr, u_int lport, int lookupflags, 1103 struct ifnet *ifp __unused, struct mbuf *m) 1104 { 1105 return (in6_pcblookup_hash_smr(pcbinfo, faddr, fport, laddr, lport, 1106 lookupflags, m->m_pkthdr.numa_domain)); 1107 } 1108 1109 void 1110 init_sin6(struct sockaddr_in6 *sin6, struct mbuf *m, int srcordst) 1111 { 1112 struct ip6_hdr *ip; 1113 1114 ip = mtod(m, struct ip6_hdr *); 1115 bzero(sin6, sizeof(*sin6)); 1116 sin6->sin6_len = sizeof(*sin6); 1117 sin6->sin6_family = AF_INET6; 1118 sin6->sin6_addr = srcordst ? ip->ip6_dst : ip->ip6_src; 1119 1120 (void)sa6_recoverscope(sin6); /* XXX: should catch errors... */ 1121 1122 return; 1123 } 1124