| 1 | // SPDX-License-Identifier: GPL-2.0-or-later |
| 2 | /* |
| 3 | * PF_INET6 socket protocol family |
| 4 | * Linux INET6 implementation |
| 5 | * |
| 6 | * Authors: |
| 7 | * Pedro Roque <roque@di.fc.ul.pt> |
| 8 | * |
| 9 | * Adapted from linux/net/ipv4/af_inet.c |
| 10 | * |
| 11 | * Fixes: |
| 12 | * piggy, Karl Knutson : Socket protocol table |
| 13 | * Hideaki YOSHIFUJI : sin6_scope_id support |
| 14 | * Arnaldo Melo : check proc_net_create return, cleanups |
| 15 | */ |
| 16 | |
| 17 | #define pr_fmt(fmt) "IPv6: " fmt |
| 18 | |
| 19 | #include <linux/module.h> |
| 20 | #include <linux/capability.h> |
| 21 | #include <linux/errno.h> |
| 22 | #include <linux/types.h> |
| 23 | #include <linux/socket.h> |
| 24 | #include <linux/in.h> |
| 25 | #include <linux/kernel.h> |
| 26 | #include <linux/timer.h> |
| 27 | #include <linux/string.h> |
| 28 | #include <linux/sockios.h> |
| 29 | #include <linux/net.h> |
| 30 | #include <linux/fcntl.h> |
| 31 | #include <linux/mm.h> |
| 32 | #include <linux/interrupt.h> |
| 33 | #include <linux/proc_fs.h> |
| 34 | #include <linux/stat.h> |
| 35 | #include <linux/init.h> |
| 36 | #include <linux/slab.h> |
| 37 | |
| 38 | #include <linux/inet.h> |
| 39 | #include <linux/netdevice.h> |
| 40 | #include <linux/icmpv6.h> |
| 41 | #include <linux/netfilter_ipv6.h> |
| 42 | |
| 43 | #include <net/ip.h> |
| 44 | #include <net/ipv6.h> |
| 45 | #include <net/udp.h> |
| 46 | #include <net/udplite.h> |
| 47 | #include <net/tcp.h> |
| 48 | #include <net/ping.h> |
| 49 | #include <net/protocol.h> |
| 50 | #include <net/inet_common.h> |
| 51 | #include <net/route.h> |
| 52 | #include <net/transp_v6.h> |
| 53 | #include <net/ip6_route.h> |
| 54 | #include <net/addrconf.h> |
| 55 | #include <net/ipv6_stubs.h> |
| 56 | #include <net/ndisc.h> |
| 57 | #ifdef CONFIG_IPV6_TUNNEL |
| 58 | #include <net/ip6_tunnel.h> |
| 59 | #endif |
| 60 | #include <net/calipso.h> |
| 61 | #include <net/seg6.h> |
| 62 | #include <net/rpl.h> |
| 63 | #include <net/compat.h> |
| 64 | #include <net/xfrm.h> |
| 65 | #include <net/ioam6.h> |
| 66 | #include <net/rawv6.h> |
| 67 | #include <net/rps.h> |
| 68 | |
| 69 | #include <linux/uaccess.h> |
| 70 | #include <linux/mroute6.h> |
| 71 | |
| 72 | #include "ip6_offload.h" |
| 73 | |
| 74 | MODULE_AUTHOR("Cast of dozens" ); |
| 75 | MODULE_DESCRIPTION("IPv6 protocol stack for Linux" ); |
| 76 | MODULE_LICENSE("GPL" ); |
| 77 | |
| 78 | /* The inetsw6 table contains everything that inet6_create needs to |
| 79 | * build a new socket. |
| 80 | */ |
| 81 | static struct list_head inetsw6[SOCK_MAX]; |
| 82 | static DEFINE_SPINLOCK(inetsw6_lock); |
| 83 | |
| 84 | struct ipv6_params ipv6_defaults = { |
| 85 | .disable_ipv6 = 0, |
| 86 | .autoconf = 1, |
| 87 | }; |
| 88 | |
| 89 | static int disable_ipv6_mod; |
| 90 | |
| 91 | module_param_named(disable, disable_ipv6_mod, int, 0444); |
| 92 | MODULE_PARM_DESC(disable, "Disable IPv6 module such that it is non-functional" ); |
| 93 | |
| 94 | module_param_named(disable_ipv6, ipv6_defaults.disable_ipv6, int, 0444); |
| 95 | MODULE_PARM_DESC(disable_ipv6, "Disable IPv6 on all interfaces" ); |
| 96 | |
| 97 | module_param_named(autoconf, ipv6_defaults.autoconf, int, 0444); |
| 98 | MODULE_PARM_DESC(autoconf, "Enable IPv6 address autoconfiguration on all interfaces" ); |
| 99 | |
| 100 | bool ipv6_mod_enabled(void) |
| 101 | { |
| 102 | return disable_ipv6_mod == 0; |
| 103 | } |
| 104 | EXPORT_SYMBOL_GPL(ipv6_mod_enabled); |
| 105 | |
| 106 | static struct ipv6_pinfo *inet6_sk_generic(struct sock *sk) |
| 107 | { |
| 108 | const int offset = sk->sk_prot->ipv6_pinfo_offset; |
| 109 | |
| 110 | return (struct ipv6_pinfo *)(((u8 *)sk) + offset); |
| 111 | } |
| 112 | |
| 113 | void inet6_sock_destruct(struct sock *sk) |
| 114 | { |
| 115 | inet6_cleanup_sock(sk); |
| 116 | inet_sock_destruct(sk); |
| 117 | } |
| 118 | EXPORT_SYMBOL_GPL(inet6_sock_destruct); |
| 119 | |
| 120 | static int inet6_create(struct net *net, struct socket *sock, int protocol, |
| 121 | int kern) |
| 122 | { |
| 123 | struct inet_sock *inet; |
| 124 | struct ipv6_pinfo *np; |
| 125 | struct sock *sk; |
| 126 | struct inet_protosw *answer; |
| 127 | struct proto *answer_prot; |
| 128 | unsigned char answer_flags; |
| 129 | int try_loading_module = 0; |
| 130 | int err; |
| 131 | |
| 132 | if (protocol < 0 || protocol >= IPPROTO_MAX) |
| 133 | return -EINVAL; |
| 134 | |
| 135 | /* Look for the requested type/protocol pair. */ |
| 136 | lookup_protocol: |
| 137 | err = -ESOCKTNOSUPPORT; |
| 138 | rcu_read_lock(); |
| 139 | list_for_each_entry_rcu(answer, &inetsw6[sock->type], list) { |
| 140 | |
| 141 | err = 0; |
| 142 | /* Check the non-wild match. */ |
| 143 | if (protocol == answer->protocol) { |
| 144 | if (protocol != IPPROTO_IP) |
| 145 | break; |
| 146 | } else { |
| 147 | /* Check for the two wild cases. */ |
| 148 | if (IPPROTO_IP == protocol) { |
| 149 | protocol = answer->protocol; |
| 150 | break; |
| 151 | } |
| 152 | if (IPPROTO_IP == answer->protocol) |
| 153 | break; |
| 154 | } |
| 155 | err = -EPROTONOSUPPORT; |
| 156 | } |
| 157 | |
| 158 | if (err) { |
| 159 | if (try_loading_module < 2) { |
| 160 | rcu_read_unlock(); |
| 161 | /* |
| 162 | * Be more specific, e.g. net-pf-10-proto-132-type-1 |
| 163 | * (net-pf-PF_INET6-proto-IPPROTO_SCTP-type-SOCK_STREAM) |
| 164 | */ |
| 165 | if (++try_loading_module == 1) |
| 166 | request_module("net-pf-%d-proto-%d-type-%d" , |
| 167 | PF_INET6, protocol, sock->type); |
| 168 | /* |
| 169 | * Fall back to generic, e.g. net-pf-10-proto-132 |
| 170 | * (net-pf-PF_INET6-proto-IPPROTO_SCTP) |
| 171 | */ |
| 172 | else |
| 173 | request_module("net-pf-%d-proto-%d" , |
| 174 | PF_INET6, protocol); |
| 175 | goto lookup_protocol; |
| 176 | } else |
| 177 | goto out_rcu_unlock; |
| 178 | } |
| 179 | |
| 180 | err = -EPERM; |
| 181 | if (sock->type == SOCK_RAW && !kern && |
| 182 | !ns_capable(ns: net->user_ns, CAP_NET_RAW)) |
| 183 | goto out_rcu_unlock; |
| 184 | |
| 185 | sock->ops = answer->ops; |
| 186 | answer_prot = answer->prot; |
| 187 | answer_flags = answer->flags; |
| 188 | rcu_read_unlock(); |
| 189 | |
| 190 | WARN_ON(!answer_prot->slab); |
| 191 | |
| 192 | err = -ENOBUFS; |
| 193 | sk = sk_alloc(net, PF_INET6, GFP_KERNEL, prot: answer_prot, kern); |
| 194 | if (!sk) |
| 195 | goto out; |
| 196 | |
| 197 | sock_init_data(sock, sk); |
| 198 | |
| 199 | err = 0; |
| 200 | if (INET_PROTOSW_REUSE & answer_flags) |
| 201 | sk->sk_reuse = SK_CAN_REUSE; |
| 202 | |
| 203 | if (INET_PROTOSW_ICSK & answer_flags) |
| 204 | inet_init_csk_locks(sk); |
| 205 | |
| 206 | inet = inet_sk(sk); |
| 207 | inet_assign_bit(IS_ICSK, sk, INET_PROTOSW_ICSK & answer_flags); |
| 208 | |
| 209 | if (SOCK_RAW == sock->type) { |
| 210 | inet->inet_num = protocol; |
| 211 | if (IPPROTO_RAW == protocol) |
| 212 | inet_set_bit(HDRINCL, sk); |
| 213 | } |
| 214 | |
| 215 | sk->sk_destruct = inet6_sock_destruct; |
| 216 | sk->sk_family = PF_INET6; |
| 217 | sk->sk_protocol = protocol; |
| 218 | |
| 219 | sk->sk_backlog_rcv = answer->prot->backlog_rcv; |
| 220 | |
| 221 | inet_sk(sk)->pinet6 = np = inet6_sk_generic(sk); |
| 222 | np->hop_limit = -1; |
| 223 | np->mcast_hops = IPV6_DEFAULT_MCASTHOPS; |
| 224 | inet6_set_bit(MC6_LOOP, sk); |
| 225 | inet6_set_bit(MC6_ALL, sk); |
| 226 | np->pmtudisc = IPV6_PMTUDISC_WANT; |
| 227 | inet6_assign_bit(REPFLOW, sk, net->ipv6.sysctl.flowlabel_reflect & |
| 228 | FLOWLABEL_REFLECT_ESTABLISHED); |
| 229 | sk->sk_ipv6only = net->ipv6.sysctl.bindv6only; |
| 230 | sk->sk_txrehash = READ_ONCE(net->core.sysctl_txrehash); |
| 231 | |
| 232 | /* Init the ipv4 part of the socket since we can have sockets |
| 233 | * using v6 API for ipv4. |
| 234 | */ |
| 235 | inet->uc_ttl = -1; |
| 236 | |
| 237 | inet_set_bit(MC_LOOP, sk); |
| 238 | inet->mc_ttl = 1; |
| 239 | inet->mc_index = 0; |
| 240 | RCU_INIT_POINTER(inet->mc_list, NULL); |
| 241 | inet->rcv_tos = 0; |
| 242 | |
| 243 | if (READ_ONCE(net->ipv4.sysctl_ip_no_pmtu_disc)) |
| 244 | inet->pmtudisc = IP_PMTUDISC_DONT; |
| 245 | else |
| 246 | inet->pmtudisc = IP_PMTUDISC_WANT; |
| 247 | |
| 248 | if (inet->inet_num) { |
| 249 | /* It assumes that any protocol which allows |
| 250 | * the user to assign a number at socket |
| 251 | * creation time automatically shares. |
| 252 | */ |
| 253 | inet->inet_sport = htons(inet->inet_num); |
| 254 | err = sk->sk_prot->hash(sk); |
| 255 | if (err) |
| 256 | goto out_sk_release; |
| 257 | } |
| 258 | if (sk->sk_prot->init) { |
| 259 | err = sk->sk_prot->init(sk); |
| 260 | if (err) |
| 261 | goto out_sk_release; |
| 262 | } |
| 263 | |
| 264 | if (!kern) { |
| 265 | err = BPF_CGROUP_RUN_PROG_INET_SOCK(sk); |
| 266 | if (err) |
| 267 | goto out_sk_release; |
| 268 | } |
| 269 | out: |
| 270 | return err; |
| 271 | out_rcu_unlock: |
| 272 | rcu_read_unlock(); |
| 273 | goto out; |
| 274 | out_sk_release: |
| 275 | sk_common_release(sk); |
| 276 | sock->sk = NULL; |
| 277 | goto out; |
| 278 | } |
| 279 | |
| 280 | static int __inet6_bind(struct sock *sk, struct sockaddr *uaddr, int addr_len, |
| 281 | u32 flags) |
| 282 | { |
| 283 | struct sockaddr_in6 *addr = (struct sockaddr_in6 *)uaddr; |
| 284 | struct inet_sock *inet = inet_sk(sk); |
| 285 | struct ipv6_pinfo *np = inet6_sk(sk: sk); |
| 286 | struct net *net = sock_net(sk); |
| 287 | __be32 v4addr = 0; |
| 288 | unsigned short snum; |
| 289 | bool saved_ipv6only; |
| 290 | int addr_type = 0; |
| 291 | int err = 0; |
| 292 | |
| 293 | if (addr->sin6_family != AF_INET6) |
| 294 | return -EAFNOSUPPORT; |
| 295 | |
| 296 | addr_type = ipv6_addr_type(addr: &addr->sin6_addr); |
| 297 | if ((addr_type & IPV6_ADDR_MULTICAST) && sk->sk_type == SOCK_STREAM) |
| 298 | return -EINVAL; |
| 299 | |
| 300 | snum = ntohs(addr->sin6_port); |
| 301 | if (!(flags & BIND_NO_CAP_NET_BIND_SERVICE) && |
| 302 | snum && inet_port_requires_bind_service(net, port: snum) && |
| 303 | !ns_capable(ns: net->user_ns, CAP_NET_BIND_SERVICE)) |
| 304 | return -EACCES; |
| 305 | |
| 306 | if (flags & BIND_WITH_LOCK) |
| 307 | lock_sock(sk); |
| 308 | |
| 309 | /* Check these errors (active socket, double bind). */ |
| 310 | if (sk->sk_state != TCP_CLOSE || inet->inet_num) { |
| 311 | err = -EINVAL; |
| 312 | goto out; |
| 313 | } |
| 314 | |
| 315 | /* Check if the address belongs to the host. */ |
| 316 | if (addr_type == IPV6_ADDR_MAPPED) { |
| 317 | struct net_device *dev = NULL; |
| 318 | int chk_addr_ret; |
| 319 | |
| 320 | /* Binding to v4-mapped address on a v6-only socket |
| 321 | * makes no sense |
| 322 | */ |
| 323 | if (ipv6_only_sock(sk)) { |
| 324 | err = -EINVAL; |
| 325 | goto out; |
| 326 | } |
| 327 | |
| 328 | rcu_read_lock(); |
| 329 | if (sk->sk_bound_dev_if) { |
| 330 | dev = dev_get_by_index_rcu(net, ifindex: sk->sk_bound_dev_if); |
| 331 | if (!dev) { |
| 332 | err = -ENODEV; |
| 333 | goto out_unlock; |
| 334 | } |
| 335 | } |
| 336 | |
| 337 | /* Reproduce AF_INET checks to make the bindings consistent */ |
| 338 | v4addr = addr->sin6_addr.s6_addr32[3]; |
| 339 | chk_addr_ret = inet_addr_type_dev_table(net, dev, addr: v4addr); |
| 340 | rcu_read_unlock(); |
| 341 | |
| 342 | if (!inet_addr_valid_or_nonlocal(net, inet, addr: v4addr, |
| 343 | addr_type: chk_addr_ret)) { |
| 344 | err = -EADDRNOTAVAIL; |
| 345 | goto out; |
| 346 | } |
| 347 | } else { |
| 348 | if (addr_type != IPV6_ADDR_ANY) { |
| 349 | struct net_device *dev = NULL; |
| 350 | |
| 351 | rcu_read_lock(); |
| 352 | if (__ipv6_addr_needs_scope_id(type: addr_type)) { |
| 353 | if (addr_len >= sizeof(struct sockaddr_in6) && |
| 354 | addr->sin6_scope_id) { |
| 355 | /* Override any existing binding, if another one |
| 356 | * is supplied by user. |
| 357 | */ |
| 358 | sk->sk_bound_dev_if = addr->sin6_scope_id; |
| 359 | } |
| 360 | |
| 361 | /* Binding to link-local address requires an interface */ |
| 362 | if (!sk->sk_bound_dev_if) { |
| 363 | err = -EINVAL; |
| 364 | goto out_unlock; |
| 365 | } |
| 366 | } |
| 367 | |
| 368 | if (sk->sk_bound_dev_if) { |
| 369 | dev = dev_get_by_index_rcu(net, ifindex: sk->sk_bound_dev_if); |
| 370 | if (!dev) { |
| 371 | err = -ENODEV; |
| 372 | goto out_unlock; |
| 373 | } |
| 374 | } |
| 375 | |
| 376 | /* ipv4 addr of the socket is invalid. Only the |
| 377 | * unspecified and mapped address have a v4 equivalent. |
| 378 | */ |
| 379 | v4addr = LOOPBACK4_IPV6; |
| 380 | if (!(addr_type & IPV6_ADDR_MULTICAST)) { |
| 381 | if (!ipv6_can_nonlocal_bind(net, inet) && |
| 382 | !ipv6_chk_addr(net, addr: &addr->sin6_addr, |
| 383 | dev, strict: 0)) { |
| 384 | err = -EADDRNOTAVAIL; |
| 385 | goto out_unlock; |
| 386 | } |
| 387 | } |
| 388 | rcu_read_unlock(); |
| 389 | } |
| 390 | } |
| 391 | |
| 392 | inet->inet_rcv_saddr = v4addr; |
| 393 | inet->inet_saddr = v4addr; |
| 394 | |
| 395 | sk->sk_v6_rcv_saddr = addr->sin6_addr; |
| 396 | |
| 397 | if (!(addr_type & IPV6_ADDR_MULTICAST)) |
| 398 | np->saddr = addr->sin6_addr; |
| 399 | |
| 400 | saved_ipv6only = sk->sk_ipv6only; |
| 401 | if (addr_type != IPV6_ADDR_ANY && addr_type != IPV6_ADDR_MAPPED) |
| 402 | sk->sk_ipv6only = 1; |
| 403 | |
| 404 | /* Make sure we are allowed to bind here. */ |
| 405 | if (snum || !(inet_test_bit(BIND_ADDRESS_NO_PORT, sk) || |
| 406 | (flags & BIND_FORCE_ADDRESS_NO_PORT))) { |
| 407 | err = sk->sk_prot->get_port(sk, snum); |
| 408 | if (err) { |
| 409 | sk->sk_ipv6only = saved_ipv6only; |
| 410 | inet_reset_saddr(sk); |
| 411 | goto out; |
| 412 | } |
| 413 | if (!(flags & BIND_FROM_BPF)) { |
| 414 | err = BPF_CGROUP_RUN_PROG_INET6_POST_BIND(sk); |
| 415 | if (err) { |
| 416 | sk->sk_ipv6only = saved_ipv6only; |
| 417 | inet_reset_saddr(sk); |
| 418 | if (sk->sk_prot->put_port) |
| 419 | sk->sk_prot->put_port(sk); |
| 420 | goto out; |
| 421 | } |
| 422 | } |
| 423 | } |
| 424 | |
| 425 | if (addr_type != IPV6_ADDR_ANY) |
| 426 | sk->sk_userlocks |= SOCK_BINDADDR_LOCK; |
| 427 | if (snum) |
| 428 | sk->sk_userlocks |= SOCK_BINDPORT_LOCK; |
| 429 | inet->inet_sport = htons(inet->inet_num); |
| 430 | inet->inet_dport = 0; |
| 431 | inet->inet_daddr = 0; |
| 432 | out: |
| 433 | if (flags & BIND_WITH_LOCK) |
| 434 | release_sock(sk); |
| 435 | return err; |
| 436 | out_unlock: |
| 437 | rcu_read_unlock(); |
| 438 | goto out; |
| 439 | } |
| 440 | |
| 441 | int inet6_bind_sk(struct sock *sk, struct sockaddr *uaddr, int addr_len) |
| 442 | { |
| 443 | u32 flags = BIND_WITH_LOCK; |
| 444 | const struct proto *prot; |
| 445 | int err = 0; |
| 446 | |
| 447 | /* IPV6_ADDRFORM can change sk->sk_prot under us. */ |
| 448 | prot = READ_ONCE(sk->sk_prot); |
| 449 | /* If the socket has its own bind function then use it. */ |
| 450 | if (prot->bind) |
| 451 | return prot->bind(sk, uaddr, addr_len); |
| 452 | |
| 453 | if (addr_len < SIN6_LEN_RFC2133) |
| 454 | return -EINVAL; |
| 455 | |
| 456 | /* BPF prog is run before any checks are done so that if the prog |
| 457 | * changes context in a wrong way it will be caught. |
| 458 | */ |
| 459 | err = BPF_CGROUP_RUN_PROG_INET_BIND_LOCK(sk, uaddr, &addr_len, |
| 460 | CGROUP_INET6_BIND, &flags); |
| 461 | if (err) |
| 462 | return err; |
| 463 | |
| 464 | return __inet6_bind(sk, uaddr, addr_len, flags); |
| 465 | } |
| 466 | |
| 467 | /* bind for INET6 API */ |
| 468 | int inet6_bind(struct socket *sock, struct sockaddr *uaddr, int addr_len) |
| 469 | { |
| 470 | return inet6_bind_sk(sk: sock->sk, uaddr, addr_len); |
| 471 | } |
| 472 | EXPORT_SYMBOL(inet6_bind); |
| 473 | |
| 474 | int inet6_release(struct socket *sock) |
| 475 | { |
| 476 | struct sock *sk = sock->sk; |
| 477 | |
| 478 | if (!sk) |
| 479 | return -EINVAL; |
| 480 | |
| 481 | /* Free mc lists */ |
| 482 | ipv6_sock_mc_close(sk); |
| 483 | |
| 484 | /* Free ac lists */ |
| 485 | ipv6_sock_ac_close(sk); |
| 486 | |
| 487 | return inet_release(sock); |
| 488 | } |
| 489 | EXPORT_SYMBOL(inet6_release); |
| 490 | |
| 491 | void inet6_cleanup_sock(struct sock *sk) |
| 492 | { |
| 493 | struct ipv6_pinfo *np = inet6_sk(sk: sk); |
| 494 | struct sk_buff *skb; |
| 495 | struct ipv6_txoptions *opt; |
| 496 | |
| 497 | /* Release rx options */ |
| 498 | |
| 499 | skb = xchg(&np->pktoptions, NULL); |
| 500 | kfree_skb(skb); |
| 501 | |
| 502 | skb = xchg(&np->rxpmtu, NULL); |
| 503 | kfree_skb(skb); |
| 504 | |
| 505 | /* Free flowlabels */ |
| 506 | fl6_free_socklist(sk); |
| 507 | |
| 508 | /* Free tx options */ |
| 509 | |
| 510 | opt = unrcu_pointer(xchg(&np->opt, NULL)); |
| 511 | if (opt) { |
| 512 | atomic_sub(i: opt->tot_len, v: &sk->sk_omem_alloc); |
| 513 | txopt_put(opt); |
| 514 | } |
| 515 | } |
| 516 | EXPORT_SYMBOL_GPL(inet6_cleanup_sock); |
| 517 | |
| 518 | /* |
| 519 | * This does both peername and sockname. |
| 520 | */ |
| 521 | int inet6_getname(struct socket *sock, struct sockaddr *uaddr, |
| 522 | int peer) |
| 523 | { |
| 524 | struct sockaddr_in6 *sin = (struct sockaddr_in6 *)uaddr; |
| 525 | int sin_addr_len = sizeof(*sin); |
| 526 | struct sock *sk = sock->sk; |
| 527 | struct inet_sock *inet = inet_sk(sk); |
| 528 | struct ipv6_pinfo *np = inet6_sk(sk: sk); |
| 529 | |
| 530 | sin->sin6_family = AF_INET6; |
| 531 | sin->sin6_flowinfo = 0; |
| 532 | sin->sin6_scope_id = 0; |
| 533 | lock_sock(sk); |
| 534 | if (peer) { |
| 535 | if (!inet->inet_dport || |
| 536 | (((1 << sk->sk_state) & (TCPF_CLOSE | TCPF_SYN_SENT)) && |
| 537 | peer == 1)) { |
| 538 | release_sock(sk); |
| 539 | return -ENOTCONN; |
| 540 | } |
| 541 | sin->sin6_port = inet->inet_dport; |
| 542 | sin->sin6_addr = sk->sk_v6_daddr; |
| 543 | if (inet6_test_bit(SNDFLOW, sk)) |
| 544 | sin->sin6_flowinfo = np->flow_label; |
| 545 | BPF_CGROUP_RUN_SA_PROG(sk, (struct sockaddr *)sin, &sin_addr_len, |
| 546 | CGROUP_INET6_GETPEERNAME); |
| 547 | } else { |
| 548 | if (ipv6_addr_any(a: &sk->sk_v6_rcv_saddr)) |
| 549 | sin->sin6_addr = np->saddr; |
| 550 | else |
| 551 | sin->sin6_addr = sk->sk_v6_rcv_saddr; |
| 552 | sin->sin6_port = inet->inet_sport; |
| 553 | BPF_CGROUP_RUN_SA_PROG(sk, (struct sockaddr *)sin, &sin_addr_len, |
| 554 | CGROUP_INET6_GETSOCKNAME); |
| 555 | } |
| 556 | sin->sin6_scope_id = ipv6_iface_scope_id(addr: &sin->sin6_addr, |
| 557 | iface: sk->sk_bound_dev_if); |
| 558 | release_sock(sk); |
| 559 | return sin_addr_len; |
| 560 | } |
| 561 | EXPORT_SYMBOL(inet6_getname); |
| 562 | |
| 563 | int inet6_ioctl(struct socket *sock, unsigned int cmd, unsigned long arg) |
| 564 | { |
| 565 | void __user *argp = (void __user *)arg; |
| 566 | struct sock *sk = sock->sk; |
| 567 | struct net *net = sock_net(sk); |
| 568 | const struct proto *prot; |
| 569 | |
| 570 | switch (cmd) { |
| 571 | case SIOCADDRT: |
| 572 | case SIOCDELRT: { |
| 573 | struct in6_rtmsg rtmsg; |
| 574 | |
| 575 | if (copy_from_user(to: &rtmsg, from: argp, n: sizeof(rtmsg))) |
| 576 | return -EFAULT; |
| 577 | return ipv6_route_ioctl(net, cmd, rtmsg: &rtmsg); |
| 578 | } |
| 579 | case SIOCSIFADDR: |
| 580 | return addrconf_add_ifaddr(net, arg: argp); |
| 581 | case SIOCDIFADDR: |
| 582 | return addrconf_del_ifaddr(net, arg: argp); |
| 583 | case SIOCSIFDSTADDR: |
| 584 | return addrconf_set_dstaddr(net, arg: argp); |
| 585 | default: |
| 586 | /* IPV6_ADDRFORM can change sk->sk_prot under us. */ |
| 587 | prot = READ_ONCE(sk->sk_prot); |
| 588 | if (!prot->ioctl) |
| 589 | return -ENOIOCTLCMD; |
| 590 | return sk_ioctl(sk, cmd, arg: (void __user *)arg); |
| 591 | } |
| 592 | /*NOTREACHED*/ |
| 593 | return 0; |
| 594 | } |
| 595 | EXPORT_SYMBOL(inet6_ioctl); |
| 596 | |
| 597 | #ifdef CONFIG_COMPAT |
| 598 | struct compat_in6_rtmsg { |
| 599 | struct in6_addr rtmsg_dst; |
| 600 | struct in6_addr rtmsg_src; |
| 601 | struct in6_addr rtmsg_gateway; |
| 602 | u32 rtmsg_type; |
| 603 | u16 rtmsg_dst_len; |
| 604 | u16 rtmsg_src_len; |
| 605 | u32 rtmsg_metric; |
| 606 | u32 rtmsg_info; |
| 607 | u32 rtmsg_flags; |
| 608 | s32 rtmsg_ifindex; |
| 609 | }; |
| 610 | |
| 611 | static int inet6_compat_routing_ioctl(struct sock *sk, unsigned int cmd, |
| 612 | struct compat_in6_rtmsg __user *ur) |
| 613 | { |
| 614 | struct in6_rtmsg rt; |
| 615 | |
| 616 | if (copy_from_user(to: &rt.rtmsg_dst, from: &ur->rtmsg_dst, |
| 617 | n: 3 * sizeof(struct in6_addr)) || |
| 618 | get_user(rt.rtmsg_type, &ur->rtmsg_type) || |
| 619 | get_user(rt.rtmsg_dst_len, &ur->rtmsg_dst_len) || |
| 620 | get_user(rt.rtmsg_src_len, &ur->rtmsg_src_len) || |
| 621 | get_user(rt.rtmsg_metric, &ur->rtmsg_metric) || |
| 622 | get_user(rt.rtmsg_info, &ur->rtmsg_info) || |
| 623 | get_user(rt.rtmsg_flags, &ur->rtmsg_flags) || |
| 624 | get_user(rt.rtmsg_ifindex, &ur->rtmsg_ifindex)) |
| 625 | return -EFAULT; |
| 626 | |
| 627 | |
| 628 | return ipv6_route_ioctl(net: sock_net(sk), cmd, rtmsg: &rt); |
| 629 | } |
| 630 | |
| 631 | int inet6_compat_ioctl(struct socket *sock, unsigned int cmd, unsigned long arg) |
| 632 | { |
| 633 | void __user *argp = compat_ptr(uptr: arg); |
| 634 | struct sock *sk = sock->sk; |
| 635 | |
| 636 | switch (cmd) { |
| 637 | case SIOCADDRT: |
| 638 | case SIOCDELRT: |
| 639 | return inet6_compat_routing_ioctl(sk, cmd, ur: argp); |
| 640 | default: |
| 641 | return -ENOIOCTLCMD; |
| 642 | } |
| 643 | } |
| 644 | EXPORT_SYMBOL_GPL(inet6_compat_ioctl); |
| 645 | #endif /* CONFIG_COMPAT */ |
| 646 | |
| 647 | INDIRECT_CALLABLE_DECLARE(int udpv6_sendmsg(struct sock *, struct msghdr *, |
| 648 | size_t)); |
| 649 | int inet6_sendmsg(struct socket *sock, struct msghdr *msg, size_t size) |
| 650 | { |
| 651 | struct sock *sk = sock->sk; |
| 652 | const struct proto *prot; |
| 653 | |
| 654 | if (unlikely(inet_send_prepare(sk))) |
| 655 | return -EAGAIN; |
| 656 | |
| 657 | /* IPV6_ADDRFORM can change sk->sk_prot under us. */ |
| 658 | prot = READ_ONCE(sk->sk_prot); |
| 659 | return INDIRECT_CALL_2(prot->sendmsg, tcp_sendmsg, udpv6_sendmsg, |
| 660 | sk, msg, size); |
| 661 | } |
| 662 | |
| 663 | INDIRECT_CALLABLE_DECLARE(int udpv6_recvmsg(struct sock *, struct msghdr *, |
| 664 | size_t, int, int *)); |
| 665 | int inet6_recvmsg(struct socket *sock, struct msghdr *msg, size_t size, |
| 666 | int flags) |
| 667 | { |
| 668 | struct sock *sk = sock->sk; |
| 669 | const struct proto *prot; |
| 670 | int addr_len = 0; |
| 671 | int err; |
| 672 | |
| 673 | if (likely(!(flags & MSG_ERRQUEUE))) |
| 674 | sock_rps_record_flow(sk); |
| 675 | |
| 676 | /* IPV6_ADDRFORM can change sk->sk_prot under us. */ |
| 677 | prot = READ_ONCE(sk->sk_prot); |
| 678 | err = INDIRECT_CALL_2(prot->recvmsg, tcp_recvmsg, udpv6_recvmsg, |
| 679 | sk, msg, size, flags, &addr_len); |
| 680 | if (err >= 0) |
| 681 | msg->msg_namelen = addr_len; |
| 682 | return err; |
| 683 | } |
| 684 | |
| 685 | const struct proto_ops inet6_stream_ops = { |
| 686 | .family = PF_INET6, |
| 687 | .owner = THIS_MODULE, |
| 688 | .release = inet6_release, |
| 689 | .bind = inet6_bind, |
| 690 | .connect = inet_stream_connect, /* ok */ |
| 691 | .socketpair = sock_no_socketpair, /* a do nothing */ |
| 692 | .accept = inet_accept, /* ok */ |
| 693 | .getname = inet6_getname, |
| 694 | .poll = tcp_poll, /* ok */ |
| 695 | .ioctl = inet6_ioctl, /* must change */ |
| 696 | .gettstamp = sock_gettstamp, |
| 697 | .listen = inet_listen, /* ok */ |
| 698 | .shutdown = inet_shutdown, /* ok */ |
| 699 | .setsockopt = sock_common_setsockopt, /* ok */ |
| 700 | .getsockopt = sock_common_getsockopt, /* ok */ |
| 701 | .sendmsg = inet6_sendmsg, /* retpoline's sake */ |
| 702 | .recvmsg = inet6_recvmsg, /* retpoline's sake */ |
| 703 | #ifdef CONFIG_MMU |
| 704 | .mmap = tcp_mmap, |
| 705 | #endif |
| 706 | .splice_eof = inet_splice_eof, |
| 707 | .sendmsg_locked = tcp_sendmsg_locked, |
| 708 | .splice_read = tcp_splice_read, |
| 709 | .set_peek_off = sk_set_peek_off, |
| 710 | .read_sock = tcp_read_sock, |
| 711 | .read_skb = tcp_read_skb, |
| 712 | .peek_len = tcp_peek_len, |
| 713 | #ifdef CONFIG_COMPAT |
| 714 | .compat_ioctl = inet6_compat_ioctl, |
| 715 | #endif |
| 716 | .set_rcvlowat = tcp_set_rcvlowat, |
| 717 | }; |
| 718 | EXPORT_SYMBOL_GPL(inet6_stream_ops); |
| 719 | |
| 720 | const struct proto_ops inet6_dgram_ops = { |
| 721 | .family = PF_INET6, |
| 722 | .owner = THIS_MODULE, |
| 723 | .release = inet6_release, |
| 724 | .bind = inet6_bind, |
| 725 | .connect = inet_dgram_connect, /* ok */ |
| 726 | .socketpair = sock_no_socketpair, /* a do nothing */ |
| 727 | .accept = sock_no_accept, /* a do nothing */ |
| 728 | .getname = inet6_getname, |
| 729 | .poll = udp_poll, /* ok */ |
| 730 | .ioctl = inet6_ioctl, /* must change */ |
| 731 | .gettstamp = sock_gettstamp, |
| 732 | .listen = sock_no_listen, /* ok */ |
| 733 | .shutdown = inet_shutdown, /* ok */ |
| 734 | .setsockopt = sock_common_setsockopt, /* ok */ |
| 735 | .getsockopt = sock_common_getsockopt, /* ok */ |
| 736 | .sendmsg = inet6_sendmsg, /* retpoline's sake */ |
| 737 | .recvmsg = inet6_recvmsg, /* retpoline's sake */ |
| 738 | .read_skb = udp_read_skb, |
| 739 | .mmap = sock_no_mmap, |
| 740 | .set_peek_off = udp_set_peek_off, |
| 741 | #ifdef CONFIG_COMPAT |
| 742 | .compat_ioctl = inet6_compat_ioctl, |
| 743 | #endif |
| 744 | }; |
| 745 | |
| 746 | static const struct net_proto_family inet6_family_ops = { |
| 747 | .family = PF_INET6, |
| 748 | .create = inet6_create, |
| 749 | .owner = THIS_MODULE, |
| 750 | }; |
| 751 | |
| 752 | int inet6_register_protosw(struct inet_protosw *p) |
| 753 | { |
| 754 | struct list_head *lh; |
| 755 | struct inet_protosw *answer; |
| 756 | struct list_head *last_perm; |
| 757 | int protocol = p->protocol; |
| 758 | int ret; |
| 759 | |
| 760 | spin_lock_bh(lock: &inetsw6_lock); |
| 761 | |
| 762 | ret = -EINVAL; |
| 763 | if (p->type >= SOCK_MAX) |
| 764 | goto out_illegal; |
| 765 | |
| 766 | /* If we are trying to override a permanent protocol, bail. */ |
| 767 | answer = NULL; |
| 768 | ret = -EPERM; |
| 769 | last_perm = &inetsw6[p->type]; |
| 770 | list_for_each(lh, &inetsw6[p->type]) { |
| 771 | answer = list_entry(lh, struct inet_protosw, list); |
| 772 | |
| 773 | /* Check only the non-wild match. */ |
| 774 | if (INET_PROTOSW_PERMANENT & answer->flags) { |
| 775 | if (protocol == answer->protocol) |
| 776 | break; |
| 777 | last_perm = lh; |
| 778 | } |
| 779 | |
| 780 | answer = NULL; |
| 781 | } |
| 782 | if (answer) |
| 783 | goto out_permanent; |
| 784 | |
| 785 | /* Add the new entry after the last permanent entry if any, so that |
| 786 | * the new entry does not override a permanent entry when matched with |
| 787 | * a wild-card protocol. But it is allowed to override any existing |
| 788 | * non-permanent entry. This means that when we remove this entry, the |
| 789 | * system automatically returns to the old behavior. |
| 790 | */ |
| 791 | list_add_rcu(new: &p->list, head: last_perm); |
| 792 | ret = 0; |
| 793 | out: |
| 794 | spin_unlock_bh(lock: &inetsw6_lock); |
| 795 | return ret; |
| 796 | |
| 797 | out_permanent: |
| 798 | pr_err("Attempt to override permanent protocol %d\n" , protocol); |
| 799 | goto out; |
| 800 | |
| 801 | out_illegal: |
| 802 | pr_err("Ignoring attempt to register invalid socket type %d\n" , |
| 803 | p->type); |
| 804 | goto out; |
| 805 | } |
| 806 | EXPORT_SYMBOL(inet6_register_protosw); |
| 807 | |
| 808 | void |
| 809 | inet6_unregister_protosw(struct inet_protosw *p) |
| 810 | { |
| 811 | if (INET_PROTOSW_PERMANENT & p->flags) { |
| 812 | pr_err("Attempt to unregister permanent protocol %d\n" , |
| 813 | p->protocol); |
| 814 | } else { |
| 815 | spin_lock_bh(lock: &inetsw6_lock); |
| 816 | list_del_rcu(entry: &p->list); |
| 817 | spin_unlock_bh(lock: &inetsw6_lock); |
| 818 | |
| 819 | synchronize_net(); |
| 820 | } |
| 821 | } |
| 822 | EXPORT_SYMBOL(inet6_unregister_protosw); |
| 823 | |
| 824 | int (struct sock *sk) |
| 825 | { |
| 826 | struct ipv6_pinfo *np = inet6_sk(sk: sk); |
| 827 | struct dst_entry *dst; |
| 828 | |
| 829 | dst = __sk_dst_check(sk, cookie: np->dst_cookie); |
| 830 | |
| 831 | if (!dst) { |
| 832 | struct inet_sock *inet = inet_sk(sk); |
| 833 | struct in6_addr *final_p, final; |
| 834 | struct flowi6 fl6; |
| 835 | |
| 836 | memset(&fl6, 0, sizeof(fl6)); |
| 837 | fl6.flowi6_proto = sk->sk_protocol; |
| 838 | fl6.daddr = sk->sk_v6_daddr; |
| 839 | fl6.saddr = np->saddr; |
| 840 | fl6.flowlabel = np->flow_label; |
| 841 | fl6.flowi6_oif = sk->sk_bound_dev_if; |
| 842 | fl6.flowi6_mark = sk->sk_mark; |
| 843 | fl6.fl6_dport = inet->inet_dport; |
| 844 | fl6.fl6_sport = inet->inet_sport; |
| 845 | fl6.flowi6_uid = sk->sk_uid; |
| 846 | security_sk_classify_flow(sk, flic: flowi6_to_flowi_common(fl6: &fl6)); |
| 847 | |
| 848 | rcu_read_lock(); |
| 849 | final_p = fl6_update_dst(fl6: &fl6, rcu_dereference(np->opt), |
| 850 | orig: &final); |
| 851 | rcu_read_unlock(); |
| 852 | |
| 853 | dst = ip6_dst_lookup_flow(net: sock_net(sk), sk, fl6: &fl6, final_dst: final_p); |
| 854 | if (IS_ERR(ptr: dst)) { |
| 855 | sk->sk_route_caps = 0; |
| 856 | WRITE_ONCE(sk->sk_err_soft, -PTR_ERR(dst)); |
| 857 | return PTR_ERR(ptr: dst); |
| 858 | } |
| 859 | |
| 860 | ip6_dst_store(sk, dst, NULL, NULL); |
| 861 | } |
| 862 | |
| 863 | return 0; |
| 864 | } |
| 865 | EXPORT_SYMBOL_GPL(inet6_sk_rebuild_header); |
| 866 | |
| 867 | bool ipv6_opt_accepted(const struct sock *sk, const struct sk_buff *skb, |
| 868 | const struct inet6_skb_parm *opt) |
| 869 | { |
| 870 | const struct ipv6_pinfo *np = inet6_sk(sk: sk); |
| 871 | |
| 872 | if (np->rxopt.all) { |
| 873 | if (((opt->flags & IP6SKB_HOPBYHOP) && |
| 874 | (np->rxopt.bits.hopopts || np->rxopt.bits.ohopopts)) || |
| 875 | (ip6_flowinfo(hdr: (struct ipv6hdr *) skb_network_header(skb)) && |
| 876 | np->rxopt.bits.rxflow) || |
| 877 | (opt->srcrt && (np->rxopt.bits.srcrt || |
| 878 | np->rxopt.bits.osrcrt)) || |
| 879 | ((opt->dst1 || opt->dst0) && |
| 880 | (np->rxopt.bits.dstopts || np->rxopt.bits.odstopts))) |
| 881 | return true; |
| 882 | } |
| 883 | return false; |
| 884 | } |
| 885 | |
| 886 | static struct packet_type ipv6_packet_type __read_mostly = { |
| 887 | .type = cpu_to_be16(ETH_P_IPV6), |
| 888 | .func = ipv6_rcv, |
| 889 | .list_func = ipv6_list_rcv, |
| 890 | }; |
| 891 | |
| 892 | static int __init ipv6_packet_init(void) |
| 893 | { |
| 894 | dev_add_pack(pt: &ipv6_packet_type); |
| 895 | return 0; |
| 896 | } |
| 897 | |
| 898 | static void ipv6_packet_cleanup(void) |
| 899 | { |
| 900 | dev_remove_pack(pt: &ipv6_packet_type); |
| 901 | } |
| 902 | |
| 903 | static int __net_init ipv6_init_mibs(struct net *net) |
| 904 | { |
| 905 | int i; |
| 906 | |
| 907 | net->mib.udp_stats_in6 = alloc_percpu(struct udp_mib); |
| 908 | if (!net->mib.udp_stats_in6) |
| 909 | return -ENOMEM; |
| 910 | net->mib.udplite_stats_in6 = alloc_percpu(struct udp_mib); |
| 911 | if (!net->mib.udplite_stats_in6) |
| 912 | goto err_udplite_mib; |
| 913 | net->mib.ipv6_statistics = alloc_percpu(struct ipstats_mib); |
| 914 | if (!net->mib.ipv6_statistics) |
| 915 | goto err_ip_mib; |
| 916 | |
| 917 | for_each_possible_cpu(i) { |
| 918 | struct ipstats_mib *af_inet6_stats; |
| 919 | af_inet6_stats = per_cpu_ptr(net->mib.ipv6_statistics, i); |
| 920 | u64_stats_init(syncp: &af_inet6_stats->syncp); |
| 921 | } |
| 922 | |
| 923 | |
| 924 | net->mib.icmpv6_statistics = alloc_percpu(struct icmpv6_mib); |
| 925 | if (!net->mib.icmpv6_statistics) |
| 926 | goto err_icmp_mib; |
| 927 | net->mib.icmpv6msg_statistics = kzalloc(sizeof(struct icmpv6msg_mib), |
| 928 | GFP_KERNEL); |
| 929 | if (!net->mib.icmpv6msg_statistics) |
| 930 | goto err_icmpmsg_mib; |
| 931 | return 0; |
| 932 | |
| 933 | err_icmpmsg_mib: |
| 934 | free_percpu(pdata: net->mib.icmpv6_statistics); |
| 935 | err_icmp_mib: |
| 936 | free_percpu(pdata: net->mib.ipv6_statistics); |
| 937 | err_ip_mib: |
| 938 | free_percpu(pdata: net->mib.udplite_stats_in6); |
| 939 | err_udplite_mib: |
| 940 | free_percpu(pdata: net->mib.udp_stats_in6); |
| 941 | return -ENOMEM; |
| 942 | } |
| 943 | |
| 944 | static void ipv6_cleanup_mibs(struct net *net) |
| 945 | { |
| 946 | free_percpu(pdata: net->mib.udp_stats_in6); |
| 947 | free_percpu(pdata: net->mib.udplite_stats_in6); |
| 948 | free_percpu(pdata: net->mib.ipv6_statistics); |
| 949 | free_percpu(pdata: net->mib.icmpv6_statistics); |
| 950 | kfree(objp: net->mib.icmpv6msg_statistics); |
| 951 | } |
| 952 | |
| 953 | static int __net_init inet6_net_init(struct net *net) |
| 954 | { |
| 955 | int err = 0; |
| 956 | |
| 957 | net->ipv6.sysctl.bindv6only = 0; |
| 958 | net->ipv6.sysctl.icmpv6_time = 1*HZ; |
| 959 | net->ipv6.sysctl.icmpv6_echo_ignore_all = 0; |
| 960 | net->ipv6.sysctl.icmpv6_echo_ignore_multicast = 0; |
| 961 | net->ipv6.sysctl.icmpv6_echo_ignore_anycast = 0; |
| 962 | net->ipv6.sysctl.icmpv6_error_anycast_as_unicast = 0; |
| 963 | |
| 964 | /* By default, rate limit error messages. |
| 965 | * Except for pmtu discovery, it would break it. |
| 966 | * proc_do_large_bitmap needs pointer to the bitmap. |
| 967 | */ |
| 968 | bitmap_set(map: net->ipv6.sysctl.icmpv6_ratemask, start: 0, ICMPV6_ERRMSG_MAX + 1); |
| 969 | bitmap_clear(map: net->ipv6.sysctl.icmpv6_ratemask, ICMPV6_PKT_TOOBIG, nbits: 1); |
| 970 | net->ipv6.sysctl.icmpv6_ratemask_ptr = net->ipv6.sysctl.icmpv6_ratemask; |
| 971 | |
| 972 | net->ipv6.sysctl.flowlabel_consistency = 1; |
| 973 | net->ipv6.sysctl.auto_flowlabels = IP6_DEFAULT_AUTO_FLOW_LABELS; |
| 974 | net->ipv6.sysctl.idgen_retries = 3; |
| 975 | net->ipv6.sysctl.idgen_delay = 1 * HZ; |
| 976 | net->ipv6.sysctl.flowlabel_state_ranges = 0; |
| 977 | net->ipv6.sysctl.max_dst_opts_cnt = IP6_DEFAULT_MAX_DST_OPTS_CNT; |
| 978 | net->ipv6.sysctl.max_hbh_opts_cnt = IP6_DEFAULT_MAX_HBH_OPTS_CNT; |
| 979 | net->ipv6.sysctl.max_dst_opts_len = IP6_DEFAULT_MAX_DST_OPTS_LEN; |
| 980 | net->ipv6.sysctl.max_hbh_opts_len = IP6_DEFAULT_MAX_HBH_OPTS_LEN; |
| 981 | net->ipv6.sysctl.fib_notify_on_flag_change = 0; |
| 982 | atomic_set(v: &net->ipv6.fib6_sernum, i: 1); |
| 983 | |
| 984 | net->ipv6.sysctl.ioam6_id = IOAM6_DEFAULT_ID; |
| 985 | net->ipv6.sysctl.ioam6_id_wide = IOAM6_DEFAULT_ID_WIDE; |
| 986 | |
| 987 | err = ipv6_init_mibs(net); |
| 988 | if (err) |
| 989 | return err; |
| 990 | #ifdef CONFIG_PROC_FS |
| 991 | err = udp6_proc_init(net); |
| 992 | if (err) |
| 993 | goto out; |
| 994 | err = tcp6_proc_init(net); |
| 995 | if (err) |
| 996 | goto proc_tcp6_fail; |
| 997 | err = ac6_proc_init(net); |
| 998 | if (err) |
| 999 | goto proc_ac6_fail; |
| 1000 | #endif |
| 1001 | return err; |
| 1002 | |
| 1003 | #ifdef CONFIG_PROC_FS |
| 1004 | proc_ac6_fail: |
| 1005 | tcp6_proc_exit(net); |
| 1006 | proc_tcp6_fail: |
| 1007 | udp6_proc_exit(net); |
| 1008 | out: |
| 1009 | ipv6_cleanup_mibs(net); |
| 1010 | return err; |
| 1011 | #endif |
| 1012 | } |
| 1013 | |
| 1014 | static void __net_exit inet6_net_exit(struct net *net) |
| 1015 | { |
| 1016 | #ifdef CONFIG_PROC_FS |
| 1017 | udp6_proc_exit(net); |
| 1018 | tcp6_proc_exit(net); |
| 1019 | ac6_proc_exit(net); |
| 1020 | #endif |
| 1021 | ipv6_cleanup_mibs(net); |
| 1022 | } |
| 1023 | |
| 1024 | static struct pernet_operations inet6_net_ops = { |
| 1025 | .init = inet6_net_init, |
| 1026 | .exit = inet6_net_exit, |
| 1027 | }; |
| 1028 | |
| 1029 | static int ipv6_route_input(struct sk_buff *skb) |
| 1030 | { |
| 1031 | ip6_route_input(skb); |
| 1032 | return skb_dst(skb)->error; |
| 1033 | } |
| 1034 | |
| 1035 | static const struct ipv6_stub ipv6_stub_impl = { |
| 1036 | .ipv6_sock_mc_join = ipv6_sock_mc_join, |
| 1037 | .ipv6_sock_mc_drop = ipv6_sock_mc_drop, |
| 1038 | .ipv6_dst_lookup_flow = ip6_dst_lookup_flow, |
| 1039 | .ipv6_route_input = ipv6_route_input, |
| 1040 | .fib6_get_table = fib6_get_table, |
| 1041 | .fib6_table_lookup = fib6_table_lookup, |
| 1042 | .fib6_lookup = fib6_lookup, |
| 1043 | .fib6_select_path = fib6_select_path, |
| 1044 | .ip6_mtu_from_fib6 = ip6_mtu_from_fib6, |
| 1045 | .fib6_nh_init = fib6_nh_init, |
| 1046 | .fib6_nh_release = fib6_nh_release, |
| 1047 | .fib6_nh_release_dsts = fib6_nh_release_dsts, |
| 1048 | .fib6_update_sernum = fib6_update_sernum_stub, |
| 1049 | .fib6_rt_update = fib6_rt_update, |
| 1050 | .ip6_del_rt = ip6_del_rt, |
| 1051 | .udpv6_encap_enable = udpv6_encap_enable, |
| 1052 | .ndisc_send_na = ndisc_send_na, |
| 1053 | #if IS_ENABLED(CONFIG_XFRM) |
| 1054 | .xfrm6_local_rxpmtu = xfrm6_local_rxpmtu, |
| 1055 | .xfrm6_udp_encap_rcv = xfrm6_udp_encap_rcv, |
| 1056 | .xfrm6_gro_udp_encap_rcv = xfrm6_gro_udp_encap_rcv, |
| 1057 | .xfrm6_rcv_encap = xfrm6_rcv_encap, |
| 1058 | #endif |
| 1059 | .nd_tbl = &nd_tbl, |
| 1060 | .ipv6_fragment = ip6_fragment, |
| 1061 | .ipv6_dev_find = ipv6_dev_find, |
| 1062 | .ip6_xmit = ip6_xmit, |
| 1063 | }; |
| 1064 | |
| 1065 | static const struct ipv6_bpf_stub ipv6_bpf_stub_impl = { |
| 1066 | .inet6_bind = __inet6_bind, |
| 1067 | .udp6_lib_lookup = __udp6_lib_lookup, |
| 1068 | .ipv6_setsockopt = do_ipv6_setsockopt, |
| 1069 | .ipv6_getsockopt = do_ipv6_getsockopt, |
| 1070 | .ipv6_dev_get_saddr = ipv6_dev_get_saddr, |
| 1071 | }; |
| 1072 | |
| 1073 | static int __init inet6_init(void) |
| 1074 | { |
| 1075 | struct list_head *r; |
| 1076 | int err = 0; |
| 1077 | |
| 1078 | sock_skb_cb_check_size(sizeof(struct inet6_skb_parm)); |
| 1079 | |
| 1080 | /* Register the socket-side information for inet6_create. */ |
| 1081 | for (r = &inetsw6[0]; r < &inetsw6[SOCK_MAX]; ++r) |
| 1082 | INIT_LIST_HEAD(list: r); |
| 1083 | |
| 1084 | raw_hashinfo_init(hashinfo: &raw_v6_hashinfo); |
| 1085 | |
| 1086 | if (disable_ipv6_mod) { |
| 1087 | pr_info("Loaded, but administratively disabled, reboot required to enable\n" ); |
| 1088 | goto out; |
| 1089 | } |
| 1090 | |
| 1091 | err = proto_register(prot: &tcpv6_prot, alloc_slab: 1); |
| 1092 | if (err) |
| 1093 | goto out; |
| 1094 | |
| 1095 | err = proto_register(prot: &udpv6_prot, alloc_slab: 1); |
| 1096 | if (err) |
| 1097 | goto out_unregister_tcp_proto; |
| 1098 | |
| 1099 | err = proto_register(prot: &udplitev6_prot, alloc_slab: 1); |
| 1100 | if (err) |
| 1101 | goto out_unregister_udp_proto; |
| 1102 | |
| 1103 | err = proto_register(prot: &rawv6_prot, alloc_slab: 1); |
| 1104 | if (err) |
| 1105 | goto out_unregister_udplite_proto; |
| 1106 | |
| 1107 | err = proto_register(prot: &pingv6_prot, alloc_slab: 1); |
| 1108 | if (err) |
| 1109 | goto out_unregister_raw_proto; |
| 1110 | |
| 1111 | /* We MUST register RAW sockets before we create the ICMP6, |
| 1112 | * IGMP6, or NDISC control sockets. |
| 1113 | */ |
| 1114 | err = rawv6_init(); |
| 1115 | if (err) |
| 1116 | goto out_unregister_ping_proto; |
| 1117 | |
| 1118 | /* Register the family here so that the init calls below will |
| 1119 | * be able to create sockets. (?? is this dangerous ??) |
| 1120 | */ |
| 1121 | err = sock_register(fam: &inet6_family_ops); |
| 1122 | if (err) |
| 1123 | goto out_sock_register_fail; |
| 1124 | |
| 1125 | /* |
| 1126 | * ipngwg API draft makes clear that the correct semantics |
| 1127 | * for TCP and UDP is to consider one TCP and UDP instance |
| 1128 | * in a host available by both INET and INET6 APIs and |
| 1129 | * able to communicate via both network protocols. |
| 1130 | */ |
| 1131 | |
| 1132 | err = register_pernet_subsys(&inet6_net_ops); |
| 1133 | if (err) |
| 1134 | goto register_pernet_fail; |
| 1135 | err = ip6_mr_init(); |
| 1136 | if (err) |
| 1137 | goto ipmr_fail; |
| 1138 | err = icmpv6_init(); |
| 1139 | if (err) |
| 1140 | goto icmp_fail; |
| 1141 | err = ndisc_init(); |
| 1142 | if (err) |
| 1143 | goto ndisc_fail; |
| 1144 | err = igmp6_init(); |
| 1145 | if (err) |
| 1146 | goto igmp_fail; |
| 1147 | |
| 1148 | err = ipv6_netfilter_init(); |
| 1149 | if (err) |
| 1150 | goto netfilter_fail; |
| 1151 | /* Create /proc/foo6 entries. */ |
| 1152 | #ifdef CONFIG_PROC_FS |
| 1153 | err = -ENOMEM; |
| 1154 | if (raw6_proc_init()) |
| 1155 | goto proc_raw6_fail; |
| 1156 | if (udplite6_proc_init()) |
| 1157 | goto proc_udplite6_fail; |
| 1158 | if (ipv6_misc_proc_init()) |
| 1159 | goto proc_misc6_fail; |
| 1160 | if (if6_proc_init()) |
| 1161 | goto proc_if6_fail; |
| 1162 | #endif |
| 1163 | err = ip6_route_init(); |
| 1164 | if (err) |
| 1165 | goto ip6_route_fail; |
| 1166 | err = ndisc_late_init(); |
| 1167 | if (err) |
| 1168 | goto ndisc_late_fail; |
| 1169 | err = ip6_flowlabel_init(); |
| 1170 | if (err) |
| 1171 | goto ip6_flowlabel_fail; |
| 1172 | err = ipv6_anycast_init(); |
| 1173 | if (err) |
| 1174 | goto ipv6_anycast_fail; |
| 1175 | err = addrconf_init(); |
| 1176 | if (err) |
| 1177 | goto addrconf_fail; |
| 1178 | |
| 1179 | /* Init v6 extension headers. */ |
| 1180 | err = ipv6_exthdrs_init(); |
| 1181 | if (err) |
| 1182 | goto ipv6_exthdrs_fail; |
| 1183 | |
| 1184 | err = ipv6_frag_init(); |
| 1185 | if (err) |
| 1186 | goto ipv6_frag_fail; |
| 1187 | |
| 1188 | /* Init v6 transport protocols. */ |
| 1189 | err = udpv6_init(); |
| 1190 | if (err) |
| 1191 | goto udpv6_fail; |
| 1192 | |
| 1193 | err = udplitev6_init(); |
| 1194 | if (err) |
| 1195 | goto udplitev6_fail; |
| 1196 | |
| 1197 | err = udpv6_offload_init(); |
| 1198 | if (err) |
| 1199 | goto udpv6_offload_fail; |
| 1200 | |
| 1201 | err = tcpv6_init(); |
| 1202 | if (err) |
| 1203 | goto tcpv6_fail; |
| 1204 | |
| 1205 | err = ipv6_packet_init(); |
| 1206 | if (err) |
| 1207 | goto ipv6_packet_fail; |
| 1208 | |
| 1209 | err = pingv6_init(); |
| 1210 | if (err) |
| 1211 | goto pingv6_fail; |
| 1212 | |
| 1213 | err = calipso_init(); |
| 1214 | if (err) |
| 1215 | goto calipso_fail; |
| 1216 | |
| 1217 | err = seg6_init(); |
| 1218 | if (err) |
| 1219 | goto seg6_fail; |
| 1220 | |
| 1221 | err = rpl_init(); |
| 1222 | if (err) |
| 1223 | goto rpl_fail; |
| 1224 | |
| 1225 | err = ioam6_init(); |
| 1226 | if (err) |
| 1227 | goto ioam6_fail; |
| 1228 | |
| 1229 | err = igmp6_late_init(); |
| 1230 | if (err) |
| 1231 | goto igmp6_late_err; |
| 1232 | |
| 1233 | #ifdef CONFIG_SYSCTL |
| 1234 | err = ipv6_sysctl_register(); |
| 1235 | if (err) |
| 1236 | goto sysctl_fail; |
| 1237 | #endif |
| 1238 | |
| 1239 | /* ensure that ipv6 stubs are visible only after ipv6 is ready */ |
| 1240 | wmb(); |
| 1241 | ipv6_stub = &ipv6_stub_impl; |
| 1242 | ipv6_bpf_stub = &ipv6_bpf_stub_impl; |
| 1243 | out: |
| 1244 | return err; |
| 1245 | |
| 1246 | #ifdef CONFIG_SYSCTL |
| 1247 | sysctl_fail: |
| 1248 | igmp6_late_cleanup(); |
| 1249 | #endif |
| 1250 | igmp6_late_err: |
| 1251 | ioam6_exit(); |
| 1252 | ioam6_fail: |
| 1253 | rpl_exit(); |
| 1254 | rpl_fail: |
| 1255 | seg6_exit(); |
| 1256 | seg6_fail: |
| 1257 | calipso_exit(); |
| 1258 | calipso_fail: |
| 1259 | pingv6_exit(); |
| 1260 | pingv6_fail: |
| 1261 | ipv6_packet_cleanup(); |
| 1262 | ipv6_packet_fail: |
| 1263 | tcpv6_exit(); |
| 1264 | tcpv6_fail: |
| 1265 | udpv6_offload_exit(); |
| 1266 | udpv6_offload_fail: |
| 1267 | udplitev6_exit(); |
| 1268 | udplitev6_fail: |
| 1269 | udpv6_exit(); |
| 1270 | udpv6_fail: |
| 1271 | ipv6_frag_exit(); |
| 1272 | ipv6_frag_fail: |
| 1273 | ipv6_exthdrs_exit(); |
| 1274 | ipv6_exthdrs_fail: |
| 1275 | addrconf_cleanup(); |
| 1276 | addrconf_fail: |
| 1277 | ipv6_anycast_cleanup(); |
| 1278 | ipv6_anycast_fail: |
| 1279 | ip6_flowlabel_cleanup(); |
| 1280 | ip6_flowlabel_fail: |
| 1281 | ndisc_late_cleanup(); |
| 1282 | ndisc_late_fail: |
| 1283 | ip6_route_cleanup(); |
| 1284 | ip6_route_fail: |
| 1285 | #ifdef CONFIG_PROC_FS |
| 1286 | if6_proc_exit(); |
| 1287 | proc_if6_fail: |
| 1288 | ipv6_misc_proc_exit(); |
| 1289 | proc_misc6_fail: |
| 1290 | udplite6_proc_exit(); |
| 1291 | proc_udplite6_fail: |
| 1292 | raw6_proc_exit(); |
| 1293 | proc_raw6_fail: |
| 1294 | #endif |
| 1295 | ipv6_netfilter_fini(); |
| 1296 | netfilter_fail: |
| 1297 | igmp6_cleanup(); |
| 1298 | igmp_fail: |
| 1299 | ndisc_cleanup(); |
| 1300 | ndisc_fail: |
| 1301 | icmpv6_cleanup(); |
| 1302 | icmp_fail: |
| 1303 | ip6_mr_cleanup(); |
| 1304 | ipmr_fail: |
| 1305 | unregister_pernet_subsys(&inet6_net_ops); |
| 1306 | register_pernet_fail: |
| 1307 | sock_unregister(PF_INET6); |
| 1308 | rtnl_unregister_all(PF_INET6); |
| 1309 | out_sock_register_fail: |
| 1310 | rawv6_exit(); |
| 1311 | out_unregister_ping_proto: |
| 1312 | proto_unregister(prot: &pingv6_prot); |
| 1313 | out_unregister_raw_proto: |
| 1314 | proto_unregister(prot: &rawv6_prot); |
| 1315 | out_unregister_udplite_proto: |
| 1316 | proto_unregister(prot: &udplitev6_prot); |
| 1317 | out_unregister_udp_proto: |
| 1318 | proto_unregister(prot: &udpv6_prot); |
| 1319 | out_unregister_tcp_proto: |
| 1320 | proto_unregister(prot: &tcpv6_prot); |
| 1321 | goto out; |
| 1322 | } |
| 1323 | module_init(inet6_init); |
| 1324 | |
| 1325 | MODULE_ALIAS_NETPROTO(PF_INET6); |
| 1326 | |