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 *nam, 157 struct ucred *cred) 158 { 159 struct socket *so = inp->inp_socket; 160 struct sockaddr_in6 *sin6 = (struct sockaddr_in6 *)NULL; 161 struct inpcbinfo *pcbinfo = inp->inp_pcbinfo; 162 u_short lport = 0; 163 int error, lookupflags = 0; 164 int reuseport = (so->so_options & SO_REUSEPORT); 165 166 /* 167 * XXX: Maybe we could let SO_REUSEPORT_LB set SO_REUSEPORT bit here 168 * so that we don't have to add to the (already messy) code below. 169 */ 170 int reuseport_lb = (so->so_options & SO_REUSEPORT_LB); 171 172 INP_WLOCK_ASSERT(inp); 173 INP_HASH_WLOCK_ASSERT(pcbinfo); 174 175 if (inp->inp_lport || !IN6_IS_ADDR_UNSPECIFIED(&inp->in6p_laddr)) 176 return (EINVAL); 177 if ((so->so_options & (SO_REUSEADDR|SO_REUSEPORT|SO_REUSEPORT_LB)) == 0) 178 lookupflags = INPLOOKUP_WILDCARD; 179 if (nam == NULL) { 180 if ((error = prison_local_ip6(cred, &inp->in6p_laddr, 181 ((inp->inp_flags & IN6P_IPV6_V6ONLY) != 0))) != 0) 182 return (error); 183 } else { 184 sin6 = (struct sockaddr_in6 *)nam; 185 KASSERT(sin6->sin6_family == AF_INET6, 186 ("%s: invalid address family for %p", __func__, sin6)); 187 KASSERT(sin6->sin6_len == sizeof(*sin6), 188 ("%s: invalid address length for %p", __func__, sin6)); 189 190 if ((error = sa6_embedscope(sin6, V_ip6_use_defzone)) != 0) 191 return(error); 192 193 if ((error = prison_local_ip6(cred, &sin6->sin6_addr, 194 ((inp->inp_flags & IN6P_IPV6_V6ONLY) != 0))) != 0) 195 return (error); 196 197 lport = sin6->sin6_port; 198 if (IN6_IS_ADDR_MULTICAST(&sin6->sin6_addr)) { 199 /* 200 * Treat SO_REUSEADDR as SO_REUSEPORT for multicast; 201 * allow compepte duplication of binding if 202 * SO_REUSEPORT is set, or if SO_REUSEADDR is set 203 * and a multicast address is bound on both 204 * new and duplicated sockets. 205 */ 206 if ((so->so_options & (SO_REUSEADDR|SO_REUSEPORT)) != 0) 207 reuseport = SO_REUSEADDR|SO_REUSEPORT; 208 /* 209 * XXX: How to deal with SO_REUSEPORT_LB here? 210 * Treat same as SO_REUSEPORT for now. 211 */ 212 if ((so->so_options & 213 (SO_REUSEADDR|SO_REUSEPORT_LB)) != 0) 214 reuseport_lb = SO_REUSEADDR|SO_REUSEPORT_LB; 215 } else if (!IN6_IS_ADDR_UNSPECIFIED(&sin6->sin6_addr)) { 216 struct epoch_tracker et; 217 struct ifaddr *ifa; 218 219 sin6->sin6_port = 0; /* yech... */ 220 NET_EPOCH_ENTER(et); 221 if ((ifa = ifa_ifwithaddr((struct sockaddr *)sin6)) == 222 NULL && 223 (inp->inp_flags & INP_BINDANY) == 0) { 224 NET_EPOCH_EXIT(et); 225 return (EADDRNOTAVAIL); 226 } 227 228 /* 229 * XXX: bind to an anycast address might accidentally 230 * cause sending a packet with anycast source address. 231 * We should allow to bind to a deprecated address, since 232 * the application dares to use it. 233 */ 234 if (ifa != NULL && 235 ((struct in6_ifaddr *)ifa)->ia6_flags & 236 (IN6_IFF_ANYCAST|IN6_IFF_NOTREADY|IN6_IFF_DETACHED)) { 237 NET_EPOCH_EXIT(et); 238 return (EADDRNOTAVAIL); 239 } 240 NET_EPOCH_EXIT(et); 241 } 242 if (lport) { 243 struct inpcb *t; 244 245 /* GROSS */ 246 if (ntohs(lport) <= V_ipport_reservedhigh && 247 ntohs(lport) >= V_ipport_reservedlow && 248 priv_check_cred(cred, PRIV_NETINET_RESERVEDPORT)) 249 return (EACCES); 250 if (!IN6_IS_ADDR_MULTICAST(&sin6->sin6_addr) && 251 priv_check_cred(inp->inp_cred, PRIV_NETINET_REUSEPORT) != 0) { 252 t = in6_pcblookup_local(pcbinfo, 253 &sin6->sin6_addr, lport, 254 INPLOOKUP_WILDCARD, cred); 255 if (t && 256 ((inp->inp_flags2 & INP_BINDMULTI) == 0) && 257 (so->so_type != SOCK_STREAM || 258 IN6_IS_ADDR_UNSPECIFIED(&t->in6p_faddr)) && 259 (!IN6_IS_ADDR_UNSPECIFIED(&sin6->sin6_addr) || 260 !IN6_IS_ADDR_UNSPECIFIED(&t->in6p_laddr) || 261 (t->inp_flags2 & INP_REUSEPORT) || 262 (t->inp_flags2 & INP_REUSEPORT_LB) == 0) && 263 (inp->inp_cred->cr_uid != 264 t->inp_cred->cr_uid)) 265 return (EADDRINUSE); 266 267 /* 268 * If the socket is a BINDMULTI socket, then 269 * the credentials need to match and the 270 * original socket also has to have been bound 271 * with BINDMULTI. 272 */ 273 if (t && (! in_pcbbind_check_bindmulti(inp, t))) 274 return (EADDRINUSE); 275 276 #ifdef INET 277 if ((inp->inp_flags & IN6P_IPV6_V6ONLY) == 0 && 278 IN6_IS_ADDR_UNSPECIFIED(&sin6->sin6_addr)) { 279 struct sockaddr_in sin; 280 281 in6_sin6_2_sin(&sin, sin6); 282 t = in_pcblookup_local(pcbinfo, 283 sin.sin_addr, lport, 284 INPLOOKUP_WILDCARD, cred); 285 if (t && 286 ((inp->inp_flags2 & INP_BINDMULTI) == 0) && 287 (so->so_type != SOCK_STREAM || 288 ntohl(t->inp_faddr.s_addr) == 289 INADDR_ANY) && 290 (inp->inp_cred->cr_uid != 291 t->inp_cred->cr_uid)) 292 return (EADDRINUSE); 293 294 if (t && (! in_pcbbind_check_bindmulti(inp, t))) 295 return (EADDRINUSE); 296 } 297 #endif 298 } 299 t = in6_pcblookup_local(pcbinfo, &sin6->sin6_addr, 300 lport, lookupflags, cred); 301 if (t && (reuseport & inp_so_options(t)) == 0 && 302 (reuseport_lb & inp_so_options(t)) == 0) { 303 return (EADDRINUSE); 304 } 305 #ifdef INET 306 if ((inp->inp_flags & IN6P_IPV6_V6ONLY) == 0 && 307 IN6_IS_ADDR_UNSPECIFIED(&sin6->sin6_addr)) { 308 struct sockaddr_in sin; 309 310 in6_sin6_2_sin(&sin, sin6); 311 t = in_pcblookup_local(pcbinfo, sin.sin_addr, 312 lport, lookupflags, cred); 313 if (t && 314 (reuseport & inp_so_options(t)) == 0 && 315 (reuseport_lb & inp_so_options(t)) == 0 && 316 (ntohl(t->inp_laddr.s_addr) != INADDR_ANY || 317 (t->inp_vflag & INP_IPV6PROTO) != 0)) { 318 return (EADDRINUSE); 319 } 320 } 321 #endif 322 } 323 inp->in6p_laddr = sin6->sin6_addr; 324 } 325 if (lport == 0) { 326 if ((error = in6_pcbsetport(&inp->in6p_laddr, inp, cred)) != 0) { 327 /* Undo an address bind that may have occurred. */ 328 inp->in6p_laddr = in6addr_any; 329 return (error); 330 } 331 } else { 332 inp->inp_lport = lport; 333 if (in_pcbinshash(inp) != 0) { 334 inp->in6p_laddr = in6addr_any; 335 inp->inp_lport = 0; 336 return (EAGAIN); 337 } 338 } 339 return (0); 340 } 341 342 /* 343 * Transform old in6_pcbconnect() into an inner subroutine for new 344 * in6_pcbconnect(): Do some validity-checking on the remote 345 * address (in mbuf 'nam') and then determine local host address 346 * (i.e., which interface) to use to access that remote host. 347 * 348 * This preserves definition of in6_pcbconnect(), while supporting a 349 * slightly different version for T/TCP. (This is more than 350 * a bit of a kludge, but cleaning up the internal interfaces would 351 * have forced minor changes in every protocol). 352 */ 353 static int 354 in6_pcbladdr(struct inpcb *inp, struct sockaddr_in6 *sin6, 355 struct in6_addr *plocal_addr6) 356 { 357 int error = 0; 358 int scope_ambiguous = 0; 359 struct in6_addr in6a; 360 361 NET_EPOCH_ASSERT(); 362 INP_WLOCK_ASSERT(inp); 363 INP_HASH_WLOCK_ASSERT(inp->inp_pcbinfo); /* XXXRW: why? */ 364 365 if (sin6->sin6_port == 0) 366 return (EADDRNOTAVAIL); 367 368 if (sin6->sin6_scope_id == 0 && !V_ip6_use_defzone) 369 scope_ambiguous = 1; 370 if ((error = sa6_embedscope(sin6, V_ip6_use_defzone)) != 0) 371 return(error); 372 373 if (!CK_STAILQ_EMPTY(&V_in6_ifaddrhead)) { 374 /* 375 * If the destination address is UNSPECIFIED addr, 376 * use the loopback addr, e.g ::1. 377 */ 378 if (IN6_IS_ADDR_UNSPECIFIED(&sin6->sin6_addr)) 379 sin6->sin6_addr = in6addr_loopback; 380 } 381 if ((error = prison_remote_ip6(inp->inp_cred, &sin6->sin6_addr)) != 0) 382 return (error); 383 384 error = in6_selectsrc_socket(sin6, inp->in6p_outputopts, 385 inp, inp->inp_cred, scope_ambiguous, &in6a, NULL); 386 if (error) 387 return (error); 388 389 /* 390 * Do not update this earlier, in case we return with an error. 391 * 392 * XXX: this in6_selectsrc_socket result might replace the bound local 393 * address with the address specified by setsockopt(IPV6_PKTINFO). 394 * Is it the intended behavior? 395 */ 396 *plocal_addr6 = in6a; 397 398 /* 399 * Don't do pcblookup call here; return interface in 400 * plocal_addr6 401 * and exit to caller, that will do the lookup. 402 */ 403 404 return (0); 405 } 406 407 /* 408 * Outer subroutine: 409 * Connect from a socket to a specified address. 410 * Both address and port must be specified in argument sin. 411 * If don't have a local address for this socket yet, 412 * then pick one. 413 */ 414 int 415 in6_pcbconnect(struct inpcb *inp, struct sockaddr_in6 *sin6, struct ucred *cred, 416 bool rehash) 417 { 418 struct inpcbinfo *pcbinfo = inp->inp_pcbinfo; 419 struct sockaddr_in6 laddr6; 420 int error; 421 422 NET_EPOCH_ASSERT(); 423 INP_WLOCK_ASSERT(inp); 424 INP_HASH_WLOCK_ASSERT(pcbinfo); 425 KASSERT(sin6->sin6_family == AF_INET6, 426 ("%s: invalid address family for %p", __func__, sin6)); 427 KASSERT(sin6->sin6_len == sizeof(*sin6), 428 ("%s: invalid address length for %p", __func__, sin6)); 429 430 bzero(&laddr6, sizeof(laddr6)); 431 laddr6.sin6_family = AF_INET6; 432 433 #ifdef ROUTE_MPATH 434 if (CALC_FLOWID_OUTBOUND) { 435 uint32_t hash_type, hash_val; 436 437 hash_val = fib6_calc_software_hash(&inp->in6p_laddr, 438 &sin6->sin6_addr, 0, sin6->sin6_port, 439 inp->inp_socket->so_proto->pr_protocol, &hash_type); 440 inp->inp_flowid = hash_val; 441 inp->inp_flowtype = hash_type; 442 } 443 #endif 444 /* 445 * Call inner routine, to assign local interface address. 446 * in6_pcbladdr() may automatically fill in sin6_scope_id. 447 */ 448 if ((error = in6_pcbladdr(inp, sin6, &laddr6.sin6_addr)) != 0) 449 return (error); 450 451 if (in6_pcblookup_hash_locked(pcbinfo, &sin6->sin6_addr, 452 sin6->sin6_port, 453 IN6_IS_ADDR_UNSPECIFIED(&inp->in6p_laddr) 454 ? &laddr6.sin6_addr : &inp->in6p_laddr, 455 inp->inp_lport, 0, NULL, M_NODOM) != NULL) { 456 return (EADDRINUSE); 457 } 458 if (IN6_IS_ADDR_UNSPECIFIED(&inp->in6p_laddr)) { 459 if (inp->inp_lport == 0) { 460 /* 461 * rehash was required to be true in the past for 462 * this case; retain that convention. However, 463 * we now call in_pcb_lport_dest rather than 464 * in6_pcbbind; the former does not insert into 465 * the hash table, the latter does. Change rehash 466 * to false to do the in_pcbinshash below. 467 */ 468 KASSERT(rehash == true, 469 ("Rehashing required for unbound inps")); 470 rehash = false; 471 error = in_pcb_lport_dest(inp, 472 (struct sockaddr *) &laddr6, &inp->inp_lport, 473 (struct sockaddr *) sin6, sin6->sin6_port, cred, 474 INPLOOKUP_WILDCARD); 475 if (error) 476 return (error); 477 } 478 inp->in6p_laddr = laddr6.sin6_addr; 479 } 480 inp->in6p_faddr = sin6->sin6_addr; 481 inp->inp_fport = sin6->sin6_port; 482 /* update flowinfo - draft-itojun-ipv6-flowlabel-api-00 */ 483 inp->inp_flow &= ~IPV6_FLOWLABEL_MASK; 484 if (inp->inp_flags & IN6P_AUTOFLOWLABEL) 485 inp->inp_flow |= 486 (htonl(ip6_randomflowlabel()) & IPV6_FLOWLABEL_MASK); 487 488 if (rehash) { 489 in_pcbrehash(inp); 490 } else { 491 in_pcbinshash(inp); 492 } 493 494 return (0); 495 } 496 497 void 498 in6_pcbdisconnect(struct inpcb *inp) 499 { 500 501 INP_WLOCK_ASSERT(inp); 502 INP_HASH_WLOCK_ASSERT(inp->inp_pcbinfo); 503 504 memset(&inp->in6p_laddr, 0, sizeof(inp->in6p_laddr)); 505 memset(&inp->in6p_faddr, 0, sizeof(inp->in6p_faddr)); 506 inp->inp_fport = 0; 507 /* clear flowinfo - draft-itojun-ipv6-flowlabel-api-00 */ 508 inp->inp_flow &= ~IPV6_FLOWLABEL_MASK; 509 in_pcbrehash(inp); 510 } 511 512 struct sockaddr * 513 in6_sockaddr(in_port_t port, struct in6_addr *addr_p) 514 { 515 struct sockaddr_in6 *sin6; 516 517 sin6 = malloc(sizeof *sin6, M_SONAME, M_WAITOK); 518 bzero(sin6, sizeof *sin6); 519 sin6->sin6_family = AF_INET6; 520 sin6->sin6_len = sizeof(*sin6); 521 sin6->sin6_port = port; 522 sin6->sin6_addr = *addr_p; 523 (void)sa6_recoverscope(sin6); /* XXX: should catch errors */ 524 525 return (struct sockaddr *)sin6; 526 } 527 528 struct sockaddr * 529 in6_v4mapsin6_sockaddr(in_port_t port, struct in_addr *addr_p) 530 { 531 struct sockaddr_in sin; 532 struct sockaddr_in6 *sin6_p; 533 534 bzero(&sin, sizeof sin); 535 sin.sin_family = AF_INET; 536 sin.sin_len = sizeof(sin); 537 sin.sin_port = port; 538 sin.sin_addr = *addr_p; 539 540 sin6_p = malloc(sizeof *sin6_p, M_SONAME, 541 M_WAITOK); 542 in6_sin_2_v4mapsin6(&sin, sin6_p); 543 544 return (struct sockaddr *)sin6_p; 545 } 546 547 int 548 in6_getsockaddr(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_getsockaddr: inp == NULL")); 556 557 INP_RLOCK(inp); 558 port = inp->inp_lport; 559 addr = inp->in6p_laddr; 560 INP_RUNLOCK(inp); 561 562 *nam = in6_sockaddr(port, &addr); 563 return 0; 564 } 565 566 int 567 in6_getpeeraddr(struct socket *so, struct sockaddr **nam) 568 { 569 struct inpcb *inp; 570 struct in6_addr addr; 571 in_port_t port; 572 573 inp = sotoinpcb(so); 574 KASSERT(inp != NULL, ("in6_getpeeraddr: inp == NULL")); 575 576 INP_RLOCK(inp); 577 port = inp->inp_fport; 578 addr = inp->in6p_faddr; 579 INP_RUNLOCK(inp); 580 581 *nam = in6_sockaddr(port, &addr); 582 return 0; 583 } 584 585 int 586 in6_mapped_sockaddr(struct socket *so, struct sockaddr **nam) 587 { 588 struct inpcb *inp; 589 int error; 590 591 inp = sotoinpcb(so); 592 KASSERT(inp != NULL, ("in6_mapped_sockaddr: inp == NULL")); 593 594 #ifdef INET 595 if ((inp->inp_vflag & (INP_IPV4 | INP_IPV6)) == INP_IPV4) { 596 error = in_getsockaddr(so, nam); 597 if (error == 0) 598 in6_sin_2_v4mapsin6_in_sock(nam); 599 } else 600 #endif 601 { 602 /* scope issues will be handled in in6_getsockaddr(). */ 603 error = in6_getsockaddr(so, nam); 604 } 605 606 return error; 607 } 608 609 int 610 in6_mapped_peeraddr(struct socket *so, struct sockaddr **nam) 611 { 612 struct inpcb *inp; 613 int error; 614 615 inp = sotoinpcb(so); 616 KASSERT(inp != NULL, ("in6_mapped_peeraddr: inp == NULL")); 617 618 #ifdef INET 619 if ((inp->inp_vflag & (INP_IPV4 | INP_IPV6)) == INP_IPV4) { 620 error = in_getpeeraddr(so, nam); 621 if (error == 0) 622 in6_sin_2_v4mapsin6_in_sock(nam); 623 } else 624 #endif 625 /* scope issues will be handled in in6_getpeeraddr(). */ 626 error = in6_getpeeraddr(so, nam); 627 628 return error; 629 } 630 631 /* 632 * Pass some notification to all connections of a protocol 633 * associated with address dst. The local address and/or port numbers 634 * may be specified to limit the search. The "usual action" will be 635 * taken, depending on the ctlinput cmd. The caller must filter any 636 * cmds that are uninteresting (e.g., no error in the map). 637 * Call the protocol specific routine (if any) to report 638 * any errors for each matching socket. 639 */ 640 static bool 641 inp_match6(const struct inpcb *inp, void *v __unused) 642 { 643 644 return ((inp->inp_vflag & INP_IPV6) != 0); 645 } 646 647 void 648 in6_pcbnotify(struct inpcbinfo *pcbinfo, struct sockaddr_in6 *sa6_dst, 649 u_int fport_arg, const struct sockaddr_in6 *src, u_int lport_arg, 650 int errno, void *cmdarg, 651 struct inpcb *(*notify)(struct inpcb *, int)) 652 { 653 struct inpcb_iterator inpi = INP_ITERATOR(pcbinfo, INPLOOKUP_WLOCKPCB, 654 inp_match6, NULL); 655 struct inpcb *inp; 656 struct sockaddr_in6 sa6_src; 657 u_short fport = fport_arg, lport = lport_arg; 658 u_int32_t flowinfo; 659 660 if (IN6_IS_ADDR_UNSPECIFIED(&sa6_dst->sin6_addr)) 661 return; 662 663 /* 664 * note that src can be NULL when we get notify by local fragmentation. 665 */ 666 sa6_src = (src == NULL) ? sa6_any : *src; 667 flowinfo = sa6_src.sin6_flowinfo; 668 669 while ((inp = inp_next(&inpi)) != NULL) { 670 INP_WLOCK_ASSERT(inp); 671 /* 672 * If the error designates a new path MTU for a destination 673 * and the application (associated with this socket) wanted to 674 * know the value, notify. 675 * XXX: should we avoid to notify the value to TCP sockets? 676 */ 677 if (errno == EMSGSIZE && cmdarg != NULL) 678 ip6_notify_pmtu(inp, sa6_dst, *(uint32_t *)cmdarg); 679 680 /* 681 * Detect if we should notify the error. If no source and 682 * destination ports are specified, but non-zero flowinfo and 683 * local address match, notify the error. This is the case 684 * when the error is delivered with an encrypted buffer 685 * by ESP. Otherwise, just compare addresses and ports 686 * as usual. 687 */ 688 if (lport == 0 && fport == 0 && flowinfo && 689 inp->inp_socket != NULL && 690 flowinfo == (inp->inp_flow & IPV6_FLOWLABEL_MASK) && 691 IN6_ARE_ADDR_EQUAL(&inp->in6p_laddr, &sa6_src.sin6_addr)) 692 goto do_notify; 693 else if (!IN6_ARE_ADDR_EQUAL(&inp->in6p_faddr, 694 &sa6_dst->sin6_addr) || 695 inp->inp_socket == 0 || 696 (lport && inp->inp_lport != lport) || 697 (!IN6_IS_ADDR_UNSPECIFIED(&sa6_src.sin6_addr) && 698 !IN6_ARE_ADDR_EQUAL(&inp->in6p_laddr, 699 &sa6_src.sin6_addr)) || 700 (fport && inp->inp_fport != fport)) { 701 continue; 702 } 703 704 do_notify: 705 if (notify) 706 (*notify)(inp, errno); 707 } 708 } 709 710 /* 711 * Lookup a PCB based on the local address and port. Caller must hold the 712 * hash lock. No inpcb locks or references are acquired. 713 */ 714 struct inpcb * 715 in6_pcblookup_local(struct inpcbinfo *pcbinfo, struct in6_addr *laddr, 716 u_short lport, int lookupflags, struct ucred *cred) 717 { 718 struct inpcb *inp; 719 int matchwild = 3, wildcard; 720 721 KASSERT((lookupflags & ~(INPLOOKUP_WILDCARD)) == 0, 722 ("%s: invalid lookup flags %d", __func__, lookupflags)); 723 724 INP_HASH_LOCK_ASSERT(pcbinfo); 725 726 if ((lookupflags & INPLOOKUP_WILDCARD) == 0) { 727 struct inpcbhead *head; 728 /* 729 * Look for an unconnected (wildcard foreign addr) PCB that 730 * matches the local address and port we're looking for. 731 */ 732 head = &pcbinfo->ipi_hashbase[INP_PCBHASH_WILD(lport, 733 pcbinfo->ipi_hashmask)]; 734 CK_LIST_FOREACH(inp, head, inp_hash) { 735 /* XXX inp locking */ 736 if ((inp->inp_vflag & INP_IPV6) == 0) 737 continue; 738 if (IN6_IS_ADDR_UNSPECIFIED(&inp->in6p_faddr) && 739 IN6_ARE_ADDR_EQUAL(&inp->in6p_laddr, laddr) && 740 inp->inp_lport == lport) { 741 /* Found. */ 742 if (prison_equal_ip6(cred->cr_prison, 743 inp->inp_cred->cr_prison)) 744 return (inp); 745 } 746 } 747 /* 748 * Not found. 749 */ 750 return (NULL); 751 } else { 752 struct inpcbporthead *porthash; 753 struct inpcbport *phd; 754 struct inpcb *match = NULL; 755 /* 756 * Best fit PCB lookup. 757 * 758 * First see if this local port is in use by looking on the 759 * port hash list. 760 */ 761 porthash = &pcbinfo->ipi_porthashbase[INP_PCBPORTHASH(lport, 762 pcbinfo->ipi_porthashmask)]; 763 CK_LIST_FOREACH(phd, porthash, phd_hash) { 764 if (phd->phd_port == lport) 765 break; 766 } 767 if (phd != NULL) { 768 /* 769 * Port is in use by one or more PCBs. Look for best 770 * fit. 771 */ 772 CK_LIST_FOREACH(inp, &phd->phd_pcblist, inp_portlist) { 773 wildcard = 0; 774 if (!prison_equal_ip6(cred->cr_prison, 775 inp->inp_cred->cr_prison)) 776 continue; 777 /* XXX inp locking */ 778 if ((inp->inp_vflag & INP_IPV6) == 0) 779 continue; 780 if (!IN6_IS_ADDR_UNSPECIFIED(&inp->in6p_faddr)) 781 wildcard++; 782 if (!IN6_IS_ADDR_UNSPECIFIED( 783 &inp->in6p_laddr)) { 784 if (IN6_IS_ADDR_UNSPECIFIED(laddr)) 785 wildcard++; 786 else if (!IN6_ARE_ADDR_EQUAL( 787 &inp->in6p_laddr, laddr)) 788 continue; 789 } else { 790 if (!IN6_IS_ADDR_UNSPECIFIED(laddr)) 791 wildcard++; 792 } 793 if (wildcard < matchwild) { 794 match = inp; 795 matchwild = wildcard; 796 if (matchwild == 0) 797 break; 798 } 799 } 800 } 801 return (match); 802 } 803 } 804 805 static bool 806 in6_multi_match(const struct inpcb *inp, void *v __unused) 807 { 808 809 if ((inp->inp_vflag & INP_IPV6) && inp->in6p_moptions != NULL) 810 return (true); 811 else 812 return (false); 813 } 814 815 void 816 in6_pcbpurgeif0(struct inpcbinfo *pcbinfo, struct ifnet *ifp) 817 { 818 struct inpcb_iterator inpi = INP_ITERATOR(pcbinfo, INPLOOKUP_RLOCKPCB, 819 in6_multi_match, NULL); 820 struct inpcb *inp; 821 struct in6_multi *inm; 822 struct in6_mfilter *imf; 823 struct ip6_moptions *im6o; 824 825 IN6_MULTI_LOCK_ASSERT(); 826 827 while ((inp = inp_next(&inpi)) != NULL) { 828 INP_RLOCK_ASSERT(inp); 829 830 im6o = inp->in6p_moptions; 831 /* 832 * Unselect the outgoing ifp for multicast if it 833 * is being detached. 834 */ 835 if (im6o->im6o_multicast_ifp == ifp) 836 im6o->im6o_multicast_ifp = NULL; 837 /* 838 * Drop multicast group membership if we joined 839 * through the interface being detached. 840 */ 841 restart: 842 IP6_MFILTER_FOREACH(imf, &im6o->im6o_head) { 843 if ((inm = imf->im6f_in6m) == NULL) 844 continue; 845 if (inm->in6m_ifp != ifp) 846 continue; 847 ip6_mfilter_remove(&im6o->im6o_head, imf); 848 in6_leavegroup_locked(inm, NULL); 849 ip6_mfilter_free(imf); 850 goto restart; 851 } 852 } 853 } 854 855 /* 856 * Check for alternatives when higher level complains 857 * about service problems. For now, invalidate cached 858 * routing information. If the route was created dynamically 859 * (by a redirect), time to try a default gateway again. 860 */ 861 void 862 in6_losing(struct inpcb *inp) 863 { 864 865 RO_INVALIDATE_CACHE(&inp->inp_route6); 866 } 867 868 /* 869 * After a routing change, flush old routing 870 * and allocate a (hopefully) better one. 871 */ 872 struct inpcb * 873 in6_rtchange(struct inpcb *inp, int errno __unused) 874 { 875 876 RO_INVALIDATE_CACHE(&inp->inp_route6); 877 return inp; 878 } 879 880 static bool 881 in6_pcblookup_lb_numa_match(const struct inpcblbgroup *grp, int domain) 882 { 883 return (domain == M_NODOM || domain == grp->il_numa_domain); 884 } 885 886 static struct inpcb * 887 in6_pcblookup_lbgroup(const struct inpcbinfo *pcbinfo, 888 const struct in6_addr *laddr, uint16_t lport, const struct in6_addr *faddr, 889 uint16_t fport, int lookupflags, uint8_t domain) 890 { 891 const struct inpcblbgrouphead *hdr; 892 struct inpcblbgroup *grp; 893 struct inpcblbgroup *jail_exact, *jail_wild, *local_exact, *local_wild; 894 895 INP_HASH_LOCK_ASSERT(pcbinfo); 896 897 hdr = &pcbinfo->ipi_lbgrouphashbase[ 898 INP_PCBPORTHASH(lport, pcbinfo->ipi_lbgrouphashmask)]; 899 900 /* 901 * Search for an LB group match based on the following criteria: 902 * - prefer jailed groups to non-jailed groups 903 * - prefer exact source address matches to wildcard matches 904 * - prefer groups bound to the specified NUMA domain 905 */ 906 jail_exact = jail_wild = local_exact = local_wild = NULL; 907 CK_LIST_FOREACH(grp, hdr, il_list) { 908 bool injail; 909 910 #ifdef INET 911 if (!(grp->il_vflag & INP_IPV6)) 912 continue; 913 #endif 914 if (grp->il_lport != lport) 915 continue; 916 917 injail = prison_flag(grp->il_cred, PR_IP6) != 0; 918 if (injail && prison_check_ip6_locked(grp->il_cred->cr_prison, 919 laddr) != 0) 920 continue; 921 922 if (IN6_ARE_ADDR_EQUAL(&grp->il6_laddr, laddr)) { 923 if (injail) { 924 jail_exact = grp; 925 if (in6_pcblookup_lb_numa_match(grp, domain)) 926 /* This is a perfect match. */ 927 goto out; 928 } else if (local_exact == NULL || 929 in6_pcblookup_lb_numa_match(grp, domain)) { 930 local_exact = grp; 931 } 932 } else if (IN6_IS_ADDR_UNSPECIFIED(&grp->il6_laddr) && 933 (lookupflags & INPLOOKUP_WILDCARD) != 0) { 934 if (injail) { 935 if (jail_wild == NULL || 936 in6_pcblookup_lb_numa_match(grp, domain)) 937 jail_wild = grp; 938 } else if (local_wild == NULL || 939 in6_pcblookup_lb_numa_match(grp, domain)) { 940 local_wild = grp; 941 } 942 } 943 } 944 945 if (jail_exact != NULL) 946 grp = jail_exact; 947 else if (jail_wild != NULL) 948 grp = jail_wild; 949 else if (local_exact != NULL) 950 grp = local_exact; 951 else 952 grp = local_wild; 953 if (grp == NULL) 954 return (NULL); 955 out: 956 return (grp->il_inp[INP6_PCBLBGROUP_PKTHASH(faddr, lport, fport) % 957 grp->il_inpcnt]); 958 } 959 960 /* 961 * Lookup PCB in hash list. Used in in_pcb.c as well as here. 962 */ 963 struct inpcb * 964 in6_pcblookup_hash_locked(struct inpcbinfo *pcbinfo, struct in6_addr *faddr, 965 u_int fport_arg, struct in6_addr *laddr, u_int lport_arg, 966 int lookupflags, struct ifnet *ifp, uint8_t numa_domain) 967 { 968 struct inpcbhead *head; 969 struct inpcb *inp; 970 u_short fport = fport_arg, lport = lport_arg; 971 972 KASSERT((lookupflags & ~(INPLOOKUP_WILDCARD)) == 0, 973 ("%s: invalid lookup flags %d", __func__, lookupflags)); 974 KASSERT(!IN6_IS_ADDR_UNSPECIFIED(faddr), 975 ("%s: invalid foreign address", __func__)); 976 KASSERT(!IN6_IS_ADDR_UNSPECIFIED(laddr), 977 ("%s: invalid local address", __func__)); 978 979 INP_HASH_LOCK_ASSERT(pcbinfo); 980 981 /* 982 * First look for an exact match. 983 */ 984 head = &pcbinfo->ipi_hashbase[INP6_PCBHASH(faddr, lport, fport, 985 pcbinfo->ipi_hashmask)]; 986 CK_LIST_FOREACH(inp, head, inp_hash) { 987 /* XXX inp locking */ 988 if ((inp->inp_vflag & INP_IPV6) == 0) 989 continue; 990 if (IN6_ARE_ADDR_EQUAL(&inp->in6p_faddr, faddr) && 991 IN6_ARE_ADDR_EQUAL(&inp->in6p_laddr, laddr) && 992 inp->inp_fport == fport && 993 inp->inp_lport == lport) 994 return (inp); 995 } 996 997 /* 998 * Then look for a wildcard match, if requested. 999 */ 1000 if ((lookupflags & INPLOOKUP_WILDCARD) != 0) { 1001 struct inpcb *local_wild = NULL, *local_exact = NULL; 1002 struct inpcb *jail_wild = NULL; 1003 int injail; 1004 1005 /* 1006 * First see if an LB group matches the request before scanning 1007 * all sockets on this port. 1008 */ 1009 inp = in6_pcblookup_lbgroup(pcbinfo, laddr, lport, faddr, 1010 fport, lookupflags, numa_domain); 1011 if (inp != NULL) 1012 return (inp); 1013 1014 /* 1015 * Order of socket selection - we always prefer jails. 1016 * 1. jailed, non-wild. 1017 * 2. jailed, wild. 1018 * 3. non-jailed, non-wild. 1019 * 4. non-jailed, wild. 1020 */ 1021 head = &pcbinfo->ipi_hashbase[INP_PCBHASH_WILD(lport, 1022 pcbinfo->ipi_hashmask)]; 1023 CK_LIST_FOREACH(inp, head, inp_hash) { 1024 /* XXX inp locking */ 1025 if ((inp->inp_vflag & INP_IPV6) == 0) 1026 continue; 1027 1028 if (!IN6_IS_ADDR_UNSPECIFIED(&inp->in6p_faddr) || 1029 inp->inp_lport != lport) { 1030 continue; 1031 } 1032 1033 injail = prison_flag(inp->inp_cred, PR_IP6); 1034 if (injail) { 1035 if (prison_check_ip6_locked( 1036 inp->inp_cred->cr_prison, laddr) != 0) 1037 continue; 1038 } else { 1039 if (local_exact != NULL) 1040 continue; 1041 } 1042 1043 if (IN6_ARE_ADDR_EQUAL(&inp->in6p_laddr, laddr)) { 1044 if (injail) 1045 return (inp); 1046 else 1047 local_exact = inp; 1048 } else if (IN6_IS_ADDR_UNSPECIFIED(&inp->in6p_laddr)) { 1049 if (injail) 1050 jail_wild = inp; 1051 else 1052 local_wild = inp; 1053 } 1054 } /* LIST_FOREACH */ 1055 1056 if (jail_wild != NULL) 1057 return (jail_wild); 1058 if (local_exact != NULL) 1059 return (local_exact); 1060 if (local_wild != NULL) 1061 return (local_wild); 1062 } /* if ((lookupflags & INPLOOKUP_WILDCARD) != 0) */ 1063 1064 /* 1065 * Not found. 1066 */ 1067 return (NULL); 1068 } 1069 1070 /* 1071 * Lookup PCB in hash list, using pcbinfo tables. This variation locks the 1072 * hash list lock, and will return the inpcb locked (i.e., requires 1073 * INPLOOKUP_LOCKPCB). 1074 */ 1075 static struct inpcb * 1076 in6_pcblookup_hash(struct inpcbinfo *pcbinfo, struct in6_addr *faddr, 1077 u_int fport, struct in6_addr *laddr, u_int lport, int lookupflags, 1078 struct ifnet *ifp, uint8_t numa_domain) 1079 { 1080 struct inpcb *inp; 1081 1082 KASSERT((lookupflags & ~INPLOOKUP_MASK) == 0, 1083 ("%s: invalid lookup flags %d", __func__, lookupflags)); 1084 KASSERT((lookupflags & (INPLOOKUP_RLOCKPCB | INPLOOKUP_WLOCKPCB)) != 0, 1085 ("%s: LOCKPCB not set", __func__)); 1086 1087 smr_enter(pcbinfo->ipi_smr); 1088 inp = in6_pcblookup_hash_locked(pcbinfo, faddr, fport, laddr, lport, 1089 lookupflags & INPLOOKUP_WILDCARD, ifp, numa_domain); 1090 if (inp != NULL) { 1091 if (__predict_false(inp_smr_lock(inp, 1092 (lookupflags & INPLOOKUP_LOCKMASK)) == false)) 1093 inp = NULL; 1094 } else 1095 smr_exit(pcbinfo->ipi_smr); 1096 1097 return (inp); 1098 } 1099 1100 /* 1101 * Public inpcb lookup routines, accepting a 4-tuple, and optionally, an mbuf 1102 * from which a pre-calculated hash value may be extracted. 1103 */ 1104 struct inpcb * 1105 in6_pcblookup(struct inpcbinfo *pcbinfo, struct in6_addr *faddr, u_int fport, 1106 struct in6_addr *laddr, u_int lport, int lookupflags, struct ifnet *ifp) 1107 { 1108 return (in6_pcblookup_hash(pcbinfo, faddr, fport, laddr, lport, 1109 lookupflags, ifp, M_NODOM)); 1110 } 1111 1112 struct inpcb * 1113 in6_pcblookup_mbuf(struct inpcbinfo *pcbinfo, struct in6_addr *faddr, 1114 u_int fport, struct in6_addr *laddr, u_int lport, int lookupflags, 1115 struct ifnet *ifp, struct mbuf *m) 1116 { 1117 return (in6_pcblookup_hash(pcbinfo, faddr, fport, laddr, lport, 1118 lookupflags, ifp, m->m_pkthdr.numa_domain)); 1119 } 1120 1121 void 1122 init_sin6(struct sockaddr_in6 *sin6, struct mbuf *m, int srcordst) 1123 { 1124 struct ip6_hdr *ip; 1125 1126 ip = mtod(m, struct ip6_hdr *); 1127 bzero(sin6, sizeof(*sin6)); 1128 sin6->sin6_len = sizeof(*sin6); 1129 sin6->sin6_family = AF_INET6; 1130 sin6->sin6_addr = srcordst ? ip->ip6_dst : ip->ip6_src; 1131 1132 (void)sa6_recoverscope(sin6); /* XXX: should catch errors... */ 1133 1134 return; 1135 } 1136