1 // SPDX-License-Identifier: GPL-2.0-or-later 2 /* 3 * UDP over IPv6 4 * Linux INET6 implementation 5 * 6 * Authors: 7 * Pedro Roque <roque@di.fc.ul.pt> 8 * 9 * Based on linux/ipv4/udp.c 10 * 11 * Fixes: 12 * Hideaki YOSHIFUJI : sin6_scope_id support 13 * YOSHIFUJI Hideaki @USAGI and: Support IPV6_V6ONLY socket option, which 14 * Alexey Kuznetsov allow both IPv4 and IPv6 sockets to bind 15 * a single port at the same time. 16 * Kazunori MIYAZAWA @USAGI: change process style to use ip6_append_data 17 * YOSHIFUJI Hideaki @USAGI: convert /proc/net/udp6 to seq_file. 18 */ 19 20 #include <linux/bpf-cgroup.h> 21 #include <linux/errno.h> 22 #include <linux/types.h> 23 #include <linux/socket.h> 24 #include <linux/sockios.h> 25 #include <linux/net.h> 26 #include <linux/in6.h> 27 #include <linux/netdevice.h> 28 #include <linux/if_arp.h> 29 #include <linux/ipv6.h> 30 #include <linux/icmpv6.h> 31 #include <linux/init.h> 32 #include <linux/module.h> 33 #include <linux/skbuff.h> 34 #include <linux/slab.h> 35 #include <linux/uaccess.h> 36 #include <linux/indirect_call_wrapper.h> 37 38 #include <net/addrconf.h> 39 #include <net/ndisc.h> 40 #include <net/protocol.h> 41 #include <net/transp_v6.h> 42 #include <net/ip6_route.h> 43 #include <net/raw.h> 44 #include <net/seg6.h> 45 #include <net/tcp_states.h> 46 #include <net/ip6_checksum.h> 47 #include <net/ip6_tunnel.h> 48 #include <trace/events/udp.h> 49 #include <net/xfrm.h> 50 #include <net/inet_hashtables.h> 51 #include <net/inet6_hashtables.h> 52 #include <net/busy_poll.h> 53 #include <net/sock_reuseport.h> 54 55 #include <linux/proc_fs.h> 56 #include <linux/seq_file.h> 57 #include <trace/events/skb.h> 58 #include "udp_impl.h" 59 60 static void udpv6_destruct_sock(struct sock *sk) 61 { 62 udp_destruct_common(sk); 63 inet6_sock_destruct(sk); 64 } 65 66 int udpv6_init_sock(struct sock *sk) 67 { 68 udp_lib_init_sock(sk); 69 sk->sk_destruct = udpv6_destruct_sock; 70 set_bit(SOCK_SUPPORT_ZC, &sk->sk_socket->flags); 71 return 0; 72 } 73 74 static u32 udp6_ehashfn(const struct net *net, 75 const struct in6_addr *laddr, 76 const u16 lport, 77 const struct in6_addr *faddr, 78 const __be16 fport) 79 { 80 static u32 udp6_ehash_secret __read_mostly; 81 static u32 udp_ipv6_hash_secret __read_mostly; 82 83 u32 lhash, fhash; 84 85 net_get_random_once(&udp6_ehash_secret, 86 sizeof(udp6_ehash_secret)); 87 net_get_random_once(&udp_ipv6_hash_secret, 88 sizeof(udp_ipv6_hash_secret)); 89 90 lhash = (__force u32)laddr->s6_addr32[3]; 91 fhash = __ipv6_addr_jhash(faddr, udp_ipv6_hash_secret); 92 93 return __inet6_ehashfn(lhash, lport, fhash, fport, 94 udp6_ehash_secret + net_hash_mix(net)); 95 } 96 97 int udp_v6_get_port(struct sock *sk, unsigned short snum) 98 { 99 unsigned int hash2_nulladdr = 100 ipv6_portaddr_hash(sock_net(sk), &in6addr_any, snum); 101 unsigned int hash2_partial = 102 ipv6_portaddr_hash(sock_net(sk), &sk->sk_v6_rcv_saddr, 0); 103 104 /* precompute partial secondary hash */ 105 udp_sk(sk)->udp_portaddr_hash = hash2_partial; 106 return udp_lib_get_port(sk, snum, hash2_nulladdr); 107 } 108 109 void udp_v6_rehash(struct sock *sk) 110 { 111 u16 new_hash = ipv6_portaddr_hash(sock_net(sk), 112 &sk->sk_v6_rcv_saddr, 113 inet_sk(sk)->inet_num); 114 115 udp_lib_rehash(sk, new_hash); 116 } 117 118 static int compute_score(struct sock *sk, struct net *net, 119 const struct in6_addr *saddr, __be16 sport, 120 const struct in6_addr *daddr, unsigned short hnum, 121 int dif, int sdif) 122 { 123 int bound_dev_if, score; 124 struct inet_sock *inet; 125 bool dev_match; 126 127 if (!net_eq(sock_net(sk), net) || 128 udp_sk(sk)->udp_port_hash != hnum || 129 sk->sk_family != PF_INET6) 130 return -1; 131 132 if (!ipv6_addr_equal(&sk->sk_v6_rcv_saddr, daddr)) 133 return -1; 134 135 score = 0; 136 inet = inet_sk(sk); 137 138 if (inet->inet_dport) { 139 if (inet->inet_dport != sport) 140 return -1; 141 score++; 142 } 143 144 if (!ipv6_addr_any(&sk->sk_v6_daddr)) { 145 if (!ipv6_addr_equal(&sk->sk_v6_daddr, saddr)) 146 return -1; 147 score++; 148 } 149 150 bound_dev_if = READ_ONCE(sk->sk_bound_dev_if); 151 dev_match = udp_sk_bound_dev_eq(net, bound_dev_if, dif, sdif); 152 if (!dev_match) 153 return -1; 154 if (bound_dev_if) 155 score++; 156 157 if (READ_ONCE(sk->sk_incoming_cpu) == raw_smp_processor_id()) 158 score++; 159 160 return score; 161 } 162 163 static struct sock *lookup_reuseport(struct net *net, struct sock *sk, 164 struct sk_buff *skb, 165 const struct in6_addr *saddr, 166 __be16 sport, 167 const struct in6_addr *daddr, 168 unsigned int hnum) 169 { 170 struct sock *reuse_sk = NULL; 171 u32 hash; 172 173 if (sk->sk_reuseport && sk->sk_state != TCP_ESTABLISHED) { 174 hash = udp6_ehashfn(net, daddr, hnum, saddr, sport); 175 reuse_sk = reuseport_select_sock(sk, hash, skb, 176 sizeof(struct udphdr)); 177 } 178 return reuse_sk; 179 } 180 181 /* called with rcu_read_lock() */ 182 static struct sock *udp6_lib_lookup2(struct net *net, 183 const struct in6_addr *saddr, __be16 sport, 184 const struct in6_addr *daddr, unsigned int hnum, 185 int dif, int sdif, struct udp_hslot *hslot2, 186 struct sk_buff *skb) 187 { 188 struct sock *sk, *result; 189 int score, badness; 190 191 result = NULL; 192 badness = -1; 193 udp_portaddr_for_each_entry_rcu(sk, &hslot2->head) { 194 score = compute_score(sk, net, saddr, sport, 195 daddr, hnum, dif, sdif); 196 if (score > badness) { 197 result = lookup_reuseport(net, sk, skb, 198 saddr, sport, daddr, hnum); 199 /* Fall back to scoring if group has connections */ 200 if (result && !reuseport_has_conns(sk)) 201 return result; 202 203 result = result ? : sk; 204 badness = score; 205 } 206 } 207 return result; 208 } 209 210 static inline struct sock *udp6_lookup_run_bpf(struct net *net, 211 struct udp_table *udptable, 212 struct sk_buff *skb, 213 const struct in6_addr *saddr, 214 __be16 sport, 215 const struct in6_addr *daddr, 216 u16 hnum, const int dif) 217 { 218 struct sock *sk, *reuse_sk; 219 bool no_reuseport; 220 221 if (udptable != net->ipv4.udp_table) 222 return NULL; /* only UDP is supported */ 223 224 no_reuseport = bpf_sk_lookup_run_v6(net, IPPROTO_UDP, saddr, sport, 225 daddr, hnum, dif, &sk); 226 if (no_reuseport || IS_ERR_OR_NULL(sk)) 227 return sk; 228 229 reuse_sk = lookup_reuseport(net, sk, skb, saddr, sport, daddr, hnum); 230 if (reuse_sk) 231 sk = reuse_sk; 232 return sk; 233 } 234 235 /* rcu_read_lock() must be held */ 236 struct sock *__udp6_lib_lookup(struct net *net, 237 const struct in6_addr *saddr, __be16 sport, 238 const struct in6_addr *daddr, __be16 dport, 239 int dif, int sdif, struct udp_table *udptable, 240 struct sk_buff *skb) 241 { 242 unsigned short hnum = ntohs(dport); 243 unsigned int hash2, slot2; 244 struct udp_hslot *hslot2; 245 struct sock *result, *sk; 246 247 hash2 = ipv6_portaddr_hash(net, daddr, hnum); 248 slot2 = hash2 & udptable->mask; 249 hslot2 = &udptable->hash2[slot2]; 250 251 /* Lookup connected or non-wildcard sockets */ 252 result = udp6_lib_lookup2(net, saddr, sport, 253 daddr, hnum, dif, sdif, 254 hslot2, skb); 255 if (!IS_ERR_OR_NULL(result) && result->sk_state == TCP_ESTABLISHED) 256 goto done; 257 258 /* Lookup redirect from BPF */ 259 if (static_branch_unlikely(&bpf_sk_lookup_enabled)) { 260 sk = udp6_lookup_run_bpf(net, udptable, skb, 261 saddr, sport, daddr, hnum, dif); 262 if (sk) { 263 result = sk; 264 goto done; 265 } 266 } 267 268 /* Got non-wildcard socket or error on first lookup */ 269 if (result) 270 goto done; 271 272 /* Lookup wildcard sockets */ 273 hash2 = ipv6_portaddr_hash(net, &in6addr_any, hnum); 274 slot2 = hash2 & udptable->mask; 275 hslot2 = &udptable->hash2[slot2]; 276 277 result = udp6_lib_lookup2(net, saddr, sport, 278 &in6addr_any, hnum, dif, sdif, 279 hslot2, skb); 280 done: 281 if (IS_ERR(result)) 282 return NULL; 283 return result; 284 } 285 EXPORT_SYMBOL_GPL(__udp6_lib_lookup); 286 287 static struct sock *__udp6_lib_lookup_skb(struct sk_buff *skb, 288 __be16 sport, __be16 dport, 289 struct udp_table *udptable) 290 { 291 const struct ipv6hdr *iph = ipv6_hdr(skb); 292 293 return __udp6_lib_lookup(dev_net(skb->dev), &iph->saddr, sport, 294 &iph->daddr, dport, inet6_iif(skb), 295 inet6_sdif(skb), udptable, skb); 296 } 297 298 struct sock *udp6_lib_lookup_skb(const struct sk_buff *skb, 299 __be16 sport, __be16 dport) 300 { 301 const struct ipv6hdr *iph = ipv6_hdr(skb); 302 struct net *net = dev_net(skb->dev); 303 304 return __udp6_lib_lookup(net, &iph->saddr, sport, 305 &iph->daddr, dport, inet6_iif(skb), 306 inet6_sdif(skb), net->ipv4.udp_table, NULL); 307 } 308 309 /* Must be called under rcu_read_lock(). 310 * Does increment socket refcount. 311 */ 312 #if IS_ENABLED(CONFIG_NF_TPROXY_IPV6) || IS_ENABLED(CONFIG_NF_SOCKET_IPV6) 313 struct sock *udp6_lib_lookup(struct net *net, const struct in6_addr *saddr, __be16 sport, 314 const struct in6_addr *daddr, __be16 dport, int dif) 315 { 316 struct sock *sk; 317 318 sk = __udp6_lib_lookup(net, saddr, sport, daddr, dport, 319 dif, 0, net->ipv4.udp_table, NULL); 320 if (sk && !refcount_inc_not_zero(&sk->sk_refcnt)) 321 sk = NULL; 322 return sk; 323 } 324 EXPORT_SYMBOL_GPL(udp6_lib_lookup); 325 #endif 326 327 /* do not use the scratch area len for jumbogram: their length execeeds the 328 * scratch area space; note that the IP6CB flags is still in the first 329 * cacheline, so checking for jumbograms is cheap 330 */ 331 static int udp6_skb_len(struct sk_buff *skb) 332 { 333 return unlikely(inet6_is_jumbogram(skb)) ? skb->len : udp_skb_len(skb); 334 } 335 336 /* 337 * This should be easy, if there is something there we 338 * return it, otherwise we block. 339 */ 340 341 int udpv6_recvmsg(struct sock *sk, struct msghdr *msg, size_t len, 342 int flags, int *addr_len) 343 { 344 struct ipv6_pinfo *np = inet6_sk(sk); 345 struct inet_sock *inet = inet_sk(sk); 346 struct sk_buff *skb; 347 unsigned int ulen, copied; 348 int off, err, peeking = flags & MSG_PEEK; 349 int is_udplite = IS_UDPLITE(sk); 350 struct udp_mib __percpu *mib; 351 bool checksum_valid = false; 352 int is_udp4; 353 354 if (flags & MSG_ERRQUEUE) 355 return ipv6_recv_error(sk, msg, len, addr_len); 356 357 if (np->rxpmtu && np->rxopt.bits.rxpmtu) 358 return ipv6_recv_rxpmtu(sk, msg, len, addr_len); 359 360 try_again: 361 off = sk_peek_offset(sk, flags); 362 skb = __skb_recv_udp(sk, flags, &off, &err); 363 if (!skb) 364 return err; 365 366 ulen = udp6_skb_len(skb); 367 copied = len; 368 if (copied > ulen - off) 369 copied = ulen - off; 370 else if (copied < ulen) 371 msg->msg_flags |= MSG_TRUNC; 372 373 is_udp4 = (skb->protocol == htons(ETH_P_IP)); 374 mib = __UDPX_MIB(sk, is_udp4); 375 376 /* 377 * If checksum is needed at all, try to do it while copying the 378 * data. If the data is truncated, or if we only want a partial 379 * coverage checksum (UDP-Lite), do it before the copy. 380 */ 381 382 if (copied < ulen || peeking || 383 (is_udplite && UDP_SKB_CB(skb)->partial_cov)) { 384 checksum_valid = udp_skb_csum_unnecessary(skb) || 385 !__udp_lib_checksum_complete(skb); 386 if (!checksum_valid) 387 goto csum_copy_err; 388 } 389 390 if (checksum_valid || udp_skb_csum_unnecessary(skb)) { 391 if (udp_skb_is_linear(skb)) 392 err = copy_linear_skb(skb, copied, off, &msg->msg_iter); 393 else 394 err = skb_copy_datagram_msg(skb, off, msg, copied); 395 } else { 396 err = skb_copy_and_csum_datagram_msg(skb, off, msg); 397 if (err == -EINVAL) 398 goto csum_copy_err; 399 } 400 if (unlikely(err)) { 401 if (!peeking) { 402 atomic_inc(&sk->sk_drops); 403 SNMP_INC_STATS(mib, UDP_MIB_INERRORS); 404 } 405 kfree_skb(skb); 406 return err; 407 } 408 if (!peeking) 409 SNMP_INC_STATS(mib, UDP_MIB_INDATAGRAMS); 410 411 sock_recv_cmsgs(msg, sk, skb); 412 413 /* Copy the address. */ 414 if (msg->msg_name) { 415 DECLARE_SOCKADDR(struct sockaddr_in6 *, sin6, msg->msg_name); 416 sin6->sin6_family = AF_INET6; 417 sin6->sin6_port = udp_hdr(skb)->source; 418 sin6->sin6_flowinfo = 0; 419 420 if (is_udp4) { 421 ipv6_addr_set_v4mapped(ip_hdr(skb)->saddr, 422 &sin6->sin6_addr); 423 sin6->sin6_scope_id = 0; 424 } else { 425 sin6->sin6_addr = ipv6_hdr(skb)->saddr; 426 sin6->sin6_scope_id = 427 ipv6_iface_scope_id(&sin6->sin6_addr, 428 inet6_iif(skb)); 429 } 430 *addr_len = sizeof(*sin6); 431 432 BPF_CGROUP_RUN_PROG_UDP6_RECVMSG_LOCK(sk, 433 (struct sockaddr *)sin6); 434 } 435 436 if (udp_sk(sk)->gro_enabled) 437 udp_cmsg_recv(msg, sk, skb); 438 439 if (np->rxopt.all) 440 ip6_datagram_recv_common_ctl(sk, msg, skb); 441 442 if (is_udp4) { 443 if (inet->cmsg_flags) 444 ip_cmsg_recv_offset(msg, sk, skb, 445 sizeof(struct udphdr), off); 446 } else { 447 if (np->rxopt.all) 448 ip6_datagram_recv_specific_ctl(sk, msg, skb); 449 } 450 451 err = copied; 452 if (flags & MSG_TRUNC) 453 err = ulen; 454 455 skb_consume_udp(sk, skb, peeking ? -err : err); 456 return err; 457 458 csum_copy_err: 459 if (!__sk_queue_drop_skb(sk, &udp_sk(sk)->reader_queue, skb, flags, 460 udp_skb_destructor)) { 461 SNMP_INC_STATS(mib, UDP_MIB_CSUMERRORS); 462 SNMP_INC_STATS(mib, UDP_MIB_INERRORS); 463 } 464 kfree_skb(skb); 465 466 /* starting over for a new packet, but check if we need to yield */ 467 cond_resched(); 468 msg->msg_flags &= ~MSG_TRUNC; 469 goto try_again; 470 } 471 472 DEFINE_STATIC_KEY_FALSE(udpv6_encap_needed_key); 473 void udpv6_encap_enable(void) 474 { 475 static_branch_inc(&udpv6_encap_needed_key); 476 } 477 EXPORT_SYMBOL(udpv6_encap_enable); 478 479 /* Handler for tunnels with arbitrary destination ports: no socket lookup, go 480 * through error handlers in encapsulations looking for a match. 481 */ 482 static int __udp6_lib_err_encap_no_sk(struct sk_buff *skb, 483 struct inet6_skb_parm *opt, 484 u8 type, u8 code, int offset, __be32 info) 485 { 486 int i; 487 488 for (i = 0; i < MAX_IPTUN_ENCAP_OPS; i++) { 489 int (*handler)(struct sk_buff *skb, struct inet6_skb_parm *opt, 490 u8 type, u8 code, int offset, __be32 info); 491 const struct ip6_tnl_encap_ops *encap; 492 493 encap = rcu_dereference(ip6tun_encaps[i]); 494 if (!encap) 495 continue; 496 handler = encap->err_handler; 497 if (handler && !handler(skb, opt, type, code, offset, info)) 498 return 0; 499 } 500 501 return -ENOENT; 502 } 503 504 /* Try to match ICMP errors to UDP tunnels by looking up a socket without 505 * reversing source and destination port: this will match tunnels that force the 506 * same destination port on both endpoints (e.g. VXLAN, GENEVE). Note that 507 * lwtunnels might actually break this assumption by being configured with 508 * different destination ports on endpoints, in this case we won't be able to 509 * trace ICMP messages back to them. 510 * 511 * If this doesn't match any socket, probe tunnels with arbitrary destination 512 * ports (e.g. FoU, GUE): there, the receiving socket is useless, as the port 513 * we've sent packets to won't necessarily match the local destination port. 514 * 515 * Then ask the tunnel implementation to match the error against a valid 516 * association. 517 * 518 * Return an error if we can't find a match, the socket if we need further 519 * processing, zero otherwise. 520 */ 521 static struct sock *__udp6_lib_err_encap(struct net *net, 522 const struct ipv6hdr *hdr, int offset, 523 struct udphdr *uh, 524 struct udp_table *udptable, 525 struct sock *sk, 526 struct sk_buff *skb, 527 struct inet6_skb_parm *opt, 528 u8 type, u8 code, __be32 info) 529 { 530 int (*lookup)(struct sock *sk, struct sk_buff *skb); 531 int network_offset, transport_offset; 532 struct udp_sock *up; 533 534 network_offset = skb_network_offset(skb); 535 transport_offset = skb_transport_offset(skb); 536 537 /* Network header needs to point to the outer IPv6 header inside ICMP */ 538 skb_reset_network_header(skb); 539 540 /* Transport header needs to point to the UDP header */ 541 skb_set_transport_header(skb, offset); 542 543 if (sk) { 544 up = udp_sk(sk); 545 546 lookup = READ_ONCE(up->encap_err_lookup); 547 if (lookup && lookup(sk, skb)) 548 sk = NULL; 549 550 goto out; 551 } 552 553 sk = __udp6_lib_lookup(net, &hdr->daddr, uh->source, 554 &hdr->saddr, uh->dest, 555 inet6_iif(skb), 0, udptable, skb); 556 if (sk) { 557 up = udp_sk(sk); 558 559 lookup = READ_ONCE(up->encap_err_lookup); 560 if (!lookup || lookup(sk, skb)) 561 sk = NULL; 562 } 563 564 out: 565 if (!sk) { 566 sk = ERR_PTR(__udp6_lib_err_encap_no_sk(skb, opt, type, code, 567 offset, info)); 568 } 569 570 skb_set_transport_header(skb, transport_offset); 571 skb_set_network_header(skb, network_offset); 572 573 return sk; 574 } 575 576 int __udp6_lib_err(struct sk_buff *skb, struct inet6_skb_parm *opt, 577 u8 type, u8 code, int offset, __be32 info, 578 struct udp_table *udptable) 579 { 580 struct ipv6_pinfo *np; 581 const struct ipv6hdr *hdr = (const struct ipv6hdr *)skb->data; 582 const struct in6_addr *saddr = &hdr->saddr; 583 const struct in6_addr *daddr = seg6_get_daddr(skb, opt) ? : &hdr->daddr; 584 struct udphdr *uh = (struct udphdr *)(skb->data+offset); 585 bool tunnel = false; 586 struct sock *sk; 587 int harderr; 588 int err; 589 struct net *net = dev_net(skb->dev); 590 591 sk = __udp6_lib_lookup(net, daddr, uh->dest, saddr, uh->source, 592 inet6_iif(skb), inet6_sdif(skb), udptable, NULL); 593 594 if (!sk || udp_sk(sk)->encap_type) { 595 /* No socket for error: try tunnels before discarding */ 596 if (static_branch_unlikely(&udpv6_encap_needed_key)) { 597 sk = __udp6_lib_err_encap(net, hdr, offset, uh, 598 udptable, sk, skb, 599 opt, type, code, info); 600 if (!sk) 601 return 0; 602 } else 603 sk = ERR_PTR(-ENOENT); 604 605 if (IS_ERR(sk)) { 606 __ICMP6_INC_STATS(net, __in6_dev_get(skb->dev), 607 ICMP6_MIB_INERRORS); 608 return PTR_ERR(sk); 609 } 610 611 tunnel = true; 612 } 613 614 harderr = icmpv6_err_convert(type, code, &err); 615 np = inet6_sk(sk); 616 617 if (type == ICMPV6_PKT_TOOBIG) { 618 if (!ip6_sk_accept_pmtu(sk)) 619 goto out; 620 ip6_sk_update_pmtu(skb, sk, info); 621 if (np->pmtudisc != IPV6_PMTUDISC_DONT) 622 harderr = 1; 623 } 624 if (type == NDISC_REDIRECT) { 625 if (tunnel) { 626 ip6_redirect(skb, sock_net(sk), inet6_iif(skb), 627 sk->sk_mark, sk->sk_uid); 628 } else { 629 ip6_sk_redirect(skb, sk); 630 } 631 goto out; 632 } 633 634 /* Tunnels don't have an application socket: don't pass errors back */ 635 if (tunnel) { 636 if (udp_sk(sk)->encap_err_rcv) 637 udp_sk(sk)->encap_err_rcv(sk, skb, err, uh->dest, 638 ntohl(info), (u8 *)(uh+1)); 639 goto out; 640 } 641 642 if (!np->recverr) { 643 if (!harderr || sk->sk_state != TCP_ESTABLISHED) 644 goto out; 645 } else { 646 ipv6_icmp_error(sk, skb, err, uh->dest, ntohl(info), (u8 *)(uh+1)); 647 } 648 649 sk->sk_err = err; 650 sk_error_report(sk); 651 out: 652 return 0; 653 } 654 655 static int __udpv6_queue_rcv_skb(struct sock *sk, struct sk_buff *skb) 656 { 657 int rc; 658 659 if (!ipv6_addr_any(&sk->sk_v6_daddr)) { 660 sock_rps_save_rxhash(sk, skb); 661 sk_mark_napi_id(sk, skb); 662 sk_incoming_cpu_update(sk); 663 } else { 664 sk_mark_napi_id_once(sk, skb); 665 } 666 667 rc = __udp_enqueue_schedule_skb(sk, skb); 668 if (rc < 0) { 669 int is_udplite = IS_UDPLITE(sk); 670 enum skb_drop_reason drop_reason; 671 672 /* Note that an ENOMEM error is charged twice */ 673 if (rc == -ENOMEM) { 674 UDP6_INC_STATS(sock_net(sk), 675 UDP_MIB_RCVBUFERRORS, is_udplite); 676 drop_reason = SKB_DROP_REASON_SOCKET_RCVBUFF; 677 } else { 678 UDP6_INC_STATS(sock_net(sk), 679 UDP_MIB_MEMERRORS, is_udplite); 680 drop_reason = SKB_DROP_REASON_PROTO_MEM; 681 } 682 UDP6_INC_STATS(sock_net(sk), UDP_MIB_INERRORS, is_udplite); 683 kfree_skb_reason(skb, drop_reason); 684 trace_udp_fail_queue_rcv_skb(rc, sk); 685 return -1; 686 } 687 688 return 0; 689 } 690 691 static __inline__ int udpv6_err(struct sk_buff *skb, 692 struct inet6_skb_parm *opt, u8 type, 693 u8 code, int offset, __be32 info) 694 { 695 return __udp6_lib_err(skb, opt, type, code, offset, info, 696 dev_net(skb->dev)->ipv4.udp_table); 697 } 698 699 static int udpv6_queue_rcv_one_skb(struct sock *sk, struct sk_buff *skb) 700 { 701 enum skb_drop_reason drop_reason = SKB_DROP_REASON_NOT_SPECIFIED; 702 struct udp_sock *up = udp_sk(sk); 703 int is_udplite = IS_UDPLITE(sk); 704 705 if (!xfrm6_policy_check(sk, XFRM_POLICY_IN, skb)) { 706 drop_reason = SKB_DROP_REASON_XFRM_POLICY; 707 goto drop; 708 } 709 nf_reset_ct(skb); 710 711 if (static_branch_unlikely(&udpv6_encap_needed_key) && up->encap_type) { 712 int (*encap_rcv)(struct sock *sk, struct sk_buff *skb); 713 714 /* 715 * This is an encapsulation socket so pass the skb to 716 * the socket's udp_encap_rcv() hook. Otherwise, just 717 * fall through and pass this up the UDP socket. 718 * up->encap_rcv() returns the following value: 719 * =0 if skb was successfully passed to the encap 720 * handler or was discarded by it. 721 * >0 if skb should be passed on to UDP. 722 * <0 if skb should be resubmitted as proto -N 723 */ 724 725 /* if we're overly short, let UDP handle it */ 726 encap_rcv = READ_ONCE(up->encap_rcv); 727 if (encap_rcv) { 728 int ret; 729 730 /* Verify checksum before giving to encap */ 731 if (udp_lib_checksum_complete(skb)) 732 goto csum_error; 733 734 ret = encap_rcv(sk, skb); 735 if (ret <= 0) { 736 __UDP6_INC_STATS(sock_net(sk), 737 UDP_MIB_INDATAGRAMS, 738 is_udplite); 739 return -ret; 740 } 741 } 742 743 /* FALLTHROUGH -- it's a UDP Packet */ 744 } 745 746 /* 747 * UDP-Lite specific tests, ignored on UDP sockets (see net/ipv4/udp.c). 748 */ 749 if ((up->pcflag & UDPLITE_RECV_CC) && UDP_SKB_CB(skb)->partial_cov) { 750 751 if (up->pcrlen == 0) { /* full coverage was set */ 752 net_dbg_ratelimited("UDPLITE6: partial coverage %d while full coverage %d requested\n", 753 UDP_SKB_CB(skb)->cscov, skb->len); 754 goto drop; 755 } 756 if (UDP_SKB_CB(skb)->cscov < up->pcrlen) { 757 net_dbg_ratelimited("UDPLITE6: coverage %d too small, need min %d\n", 758 UDP_SKB_CB(skb)->cscov, up->pcrlen); 759 goto drop; 760 } 761 } 762 763 prefetch(&sk->sk_rmem_alloc); 764 if (rcu_access_pointer(sk->sk_filter) && 765 udp_lib_checksum_complete(skb)) 766 goto csum_error; 767 768 if (sk_filter_trim_cap(sk, skb, sizeof(struct udphdr))) { 769 drop_reason = SKB_DROP_REASON_SOCKET_FILTER; 770 goto drop; 771 } 772 773 udp_csum_pull_header(skb); 774 775 skb_dst_drop(skb); 776 777 return __udpv6_queue_rcv_skb(sk, skb); 778 779 csum_error: 780 drop_reason = SKB_DROP_REASON_UDP_CSUM; 781 __UDP6_INC_STATS(sock_net(sk), UDP_MIB_CSUMERRORS, is_udplite); 782 drop: 783 __UDP6_INC_STATS(sock_net(sk), UDP_MIB_INERRORS, is_udplite); 784 atomic_inc(&sk->sk_drops); 785 kfree_skb_reason(skb, drop_reason); 786 return -1; 787 } 788 789 static int udpv6_queue_rcv_skb(struct sock *sk, struct sk_buff *skb) 790 { 791 struct sk_buff *next, *segs; 792 int ret; 793 794 if (likely(!udp_unexpected_gso(sk, skb))) 795 return udpv6_queue_rcv_one_skb(sk, skb); 796 797 __skb_push(skb, -skb_mac_offset(skb)); 798 segs = udp_rcv_segment(sk, skb, false); 799 skb_list_walk_safe(segs, skb, next) { 800 __skb_pull(skb, skb_transport_offset(skb)); 801 802 udp_post_segment_fix_csum(skb); 803 ret = udpv6_queue_rcv_one_skb(sk, skb); 804 if (ret > 0) 805 ip6_protocol_deliver_rcu(dev_net(skb->dev), skb, ret, 806 true); 807 } 808 return 0; 809 } 810 811 static bool __udp_v6_is_mcast_sock(struct net *net, const struct sock *sk, 812 __be16 loc_port, const struct in6_addr *loc_addr, 813 __be16 rmt_port, const struct in6_addr *rmt_addr, 814 int dif, int sdif, unsigned short hnum) 815 { 816 const struct inet_sock *inet = inet_sk(sk); 817 818 if (!net_eq(sock_net(sk), net)) 819 return false; 820 821 if (udp_sk(sk)->udp_port_hash != hnum || 822 sk->sk_family != PF_INET6 || 823 (inet->inet_dport && inet->inet_dport != rmt_port) || 824 (!ipv6_addr_any(&sk->sk_v6_daddr) && 825 !ipv6_addr_equal(&sk->sk_v6_daddr, rmt_addr)) || 826 !udp_sk_bound_dev_eq(net, READ_ONCE(sk->sk_bound_dev_if), dif, sdif) || 827 (!ipv6_addr_any(&sk->sk_v6_rcv_saddr) && 828 !ipv6_addr_equal(&sk->sk_v6_rcv_saddr, loc_addr))) 829 return false; 830 if (!inet6_mc_check(sk, loc_addr, rmt_addr)) 831 return false; 832 return true; 833 } 834 835 static void udp6_csum_zero_error(struct sk_buff *skb) 836 { 837 /* RFC 2460 section 8.1 says that we SHOULD log 838 * this error. Well, it is reasonable. 839 */ 840 net_dbg_ratelimited("IPv6: udp checksum is 0 for [%pI6c]:%u->[%pI6c]:%u\n", 841 &ipv6_hdr(skb)->saddr, ntohs(udp_hdr(skb)->source), 842 &ipv6_hdr(skb)->daddr, ntohs(udp_hdr(skb)->dest)); 843 } 844 845 /* 846 * Note: called only from the BH handler context, 847 * so we don't need to lock the hashes. 848 */ 849 static int __udp6_lib_mcast_deliver(struct net *net, struct sk_buff *skb, 850 const struct in6_addr *saddr, const struct in6_addr *daddr, 851 struct udp_table *udptable, int proto) 852 { 853 struct sock *sk, *first = NULL; 854 const struct udphdr *uh = udp_hdr(skb); 855 unsigned short hnum = ntohs(uh->dest); 856 struct udp_hslot *hslot = udp_hashslot(udptable, net, hnum); 857 unsigned int offset = offsetof(typeof(*sk), sk_node); 858 unsigned int hash2 = 0, hash2_any = 0, use_hash2 = (hslot->count > 10); 859 int dif = inet6_iif(skb); 860 int sdif = inet6_sdif(skb); 861 struct hlist_node *node; 862 struct sk_buff *nskb; 863 864 if (use_hash2) { 865 hash2_any = ipv6_portaddr_hash(net, &in6addr_any, hnum) & 866 udptable->mask; 867 hash2 = ipv6_portaddr_hash(net, daddr, hnum) & udptable->mask; 868 start_lookup: 869 hslot = &udptable->hash2[hash2]; 870 offset = offsetof(typeof(*sk), __sk_common.skc_portaddr_node); 871 } 872 873 sk_for_each_entry_offset_rcu(sk, node, &hslot->head, offset) { 874 if (!__udp_v6_is_mcast_sock(net, sk, uh->dest, daddr, 875 uh->source, saddr, dif, sdif, 876 hnum)) 877 continue; 878 /* If zero checksum and no_check is not on for 879 * the socket then skip it. 880 */ 881 if (!uh->check && !udp_sk(sk)->no_check6_rx) 882 continue; 883 if (!first) { 884 first = sk; 885 continue; 886 } 887 nskb = skb_clone(skb, GFP_ATOMIC); 888 if (unlikely(!nskb)) { 889 atomic_inc(&sk->sk_drops); 890 __UDP6_INC_STATS(net, UDP_MIB_RCVBUFERRORS, 891 IS_UDPLITE(sk)); 892 __UDP6_INC_STATS(net, UDP_MIB_INERRORS, 893 IS_UDPLITE(sk)); 894 continue; 895 } 896 897 if (udpv6_queue_rcv_skb(sk, nskb) > 0) 898 consume_skb(nskb); 899 } 900 901 /* Also lookup *:port if we are using hash2 and haven't done so yet. */ 902 if (use_hash2 && hash2 != hash2_any) { 903 hash2 = hash2_any; 904 goto start_lookup; 905 } 906 907 if (first) { 908 if (udpv6_queue_rcv_skb(first, skb) > 0) 909 consume_skb(skb); 910 } else { 911 kfree_skb(skb); 912 __UDP6_INC_STATS(net, UDP_MIB_IGNOREDMULTI, 913 proto == IPPROTO_UDPLITE); 914 } 915 return 0; 916 } 917 918 static void udp6_sk_rx_dst_set(struct sock *sk, struct dst_entry *dst) 919 { 920 if (udp_sk_rx_dst_set(sk, dst)) { 921 const struct rt6_info *rt = (const struct rt6_info *)dst; 922 923 sk->sk_rx_dst_cookie = rt6_get_cookie(rt); 924 } 925 } 926 927 /* wrapper for udp_queue_rcv_skb tacking care of csum conversion and 928 * return code conversion for ip layer consumption 929 */ 930 static int udp6_unicast_rcv_skb(struct sock *sk, struct sk_buff *skb, 931 struct udphdr *uh) 932 { 933 int ret; 934 935 if (inet_get_convert_csum(sk) && uh->check && !IS_UDPLITE(sk)) 936 skb_checksum_try_convert(skb, IPPROTO_UDP, ip6_compute_pseudo); 937 938 ret = udpv6_queue_rcv_skb(sk, skb); 939 940 /* a return value > 0 means to resubmit the input */ 941 if (ret > 0) 942 return ret; 943 return 0; 944 } 945 946 int __udp6_lib_rcv(struct sk_buff *skb, struct udp_table *udptable, 947 int proto) 948 { 949 enum skb_drop_reason reason = SKB_DROP_REASON_NOT_SPECIFIED; 950 const struct in6_addr *saddr, *daddr; 951 struct net *net = dev_net(skb->dev); 952 struct udphdr *uh; 953 struct sock *sk; 954 bool refcounted; 955 u32 ulen = 0; 956 957 if (!pskb_may_pull(skb, sizeof(struct udphdr))) 958 goto discard; 959 960 saddr = &ipv6_hdr(skb)->saddr; 961 daddr = &ipv6_hdr(skb)->daddr; 962 uh = udp_hdr(skb); 963 964 ulen = ntohs(uh->len); 965 if (ulen > skb->len) 966 goto short_packet; 967 968 if (proto == IPPROTO_UDP) { 969 /* UDP validates ulen. */ 970 971 /* Check for jumbo payload */ 972 if (ulen == 0) 973 ulen = skb->len; 974 975 if (ulen < sizeof(*uh)) 976 goto short_packet; 977 978 if (ulen < skb->len) { 979 if (pskb_trim_rcsum(skb, ulen)) 980 goto short_packet; 981 saddr = &ipv6_hdr(skb)->saddr; 982 daddr = &ipv6_hdr(skb)->daddr; 983 uh = udp_hdr(skb); 984 } 985 } 986 987 if (udp6_csum_init(skb, uh, proto)) 988 goto csum_error; 989 990 /* Check if the socket is already available, e.g. due to early demux */ 991 sk = skb_steal_sock(skb, &refcounted); 992 if (sk) { 993 struct dst_entry *dst = skb_dst(skb); 994 int ret; 995 996 if (unlikely(rcu_dereference(sk->sk_rx_dst) != dst)) 997 udp6_sk_rx_dst_set(sk, dst); 998 999 if (!uh->check && !udp_sk(sk)->no_check6_rx) { 1000 if (refcounted) 1001 sock_put(sk); 1002 goto report_csum_error; 1003 } 1004 1005 ret = udp6_unicast_rcv_skb(sk, skb, uh); 1006 if (refcounted) 1007 sock_put(sk); 1008 return ret; 1009 } 1010 1011 /* 1012 * Multicast receive code 1013 */ 1014 if (ipv6_addr_is_multicast(daddr)) 1015 return __udp6_lib_mcast_deliver(net, skb, 1016 saddr, daddr, udptable, proto); 1017 1018 /* Unicast */ 1019 sk = __udp6_lib_lookup_skb(skb, uh->source, uh->dest, udptable); 1020 if (sk) { 1021 if (!uh->check && !udp_sk(sk)->no_check6_rx) 1022 goto report_csum_error; 1023 return udp6_unicast_rcv_skb(sk, skb, uh); 1024 } 1025 1026 reason = SKB_DROP_REASON_NO_SOCKET; 1027 1028 if (!uh->check) 1029 goto report_csum_error; 1030 1031 if (!xfrm6_policy_check(NULL, XFRM_POLICY_IN, skb)) 1032 goto discard; 1033 nf_reset_ct(skb); 1034 1035 if (udp_lib_checksum_complete(skb)) 1036 goto csum_error; 1037 1038 __UDP6_INC_STATS(net, UDP_MIB_NOPORTS, proto == IPPROTO_UDPLITE); 1039 icmpv6_send(skb, ICMPV6_DEST_UNREACH, ICMPV6_PORT_UNREACH, 0); 1040 1041 kfree_skb_reason(skb, reason); 1042 return 0; 1043 1044 short_packet: 1045 if (reason == SKB_DROP_REASON_NOT_SPECIFIED) 1046 reason = SKB_DROP_REASON_PKT_TOO_SMALL; 1047 net_dbg_ratelimited("UDP%sv6: short packet: From [%pI6c]:%u %d/%d to [%pI6c]:%u\n", 1048 proto == IPPROTO_UDPLITE ? "-Lite" : "", 1049 saddr, ntohs(uh->source), 1050 ulen, skb->len, 1051 daddr, ntohs(uh->dest)); 1052 goto discard; 1053 1054 report_csum_error: 1055 udp6_csum_zero_error(skb); 1056 csum_error: 1057 if (reason == SKB_DROP_REASON_NOT_SPECIFIED) 1058 reason = SKB_DROP_REASON_UDP_CSUM; 1059 __UDP6_INC_STATS(net, UDP_MIB_CSUMERRORS, proto == IPPROTO_UDPLITE); 1060 discard: 1061 __UDP6_INC_STATS(net, UDP_MIB_INERRORS, proto == IPPROTO_UDPLITE); 1062 kfree_skb_reason(skb, reason); 1063 return 0; 1064 } 1065 1066 1067 static struct sock *__udp6_lib_demux_lookup(struct net *net, 1068 __be16 loc_port, const struct in6_addr *loc_addr, 1069 __be16 rmt_port, const struct in6_addr *rmt_addr, 1070 int dif, int sdif) 1071 { 1072 struct udp_table *udptable = net->ipv4.udp_table; 1073 unsigned short hnum = ntohs(loc_port); 1074 unsigned int hash2, slot2; 1075 struct udp_hslot *hslot2; 1076 __portpair ports; 1077 struct sock *sk; 1078 1079 hash2 = ipv6_portaddr_hash(net, loc_addr, hnum); 1080 slot2 = hash2 & udptable->mask; 1081 hslot2 = &udptable->hash2[slot2]; 1082 ports = INET_COMBINED_PORTS(rmt_port, hnum); 1083 1084 udp_portaddr_for_each_entry_rcu(sk, &hslot2->head) { 1085 if (sk->sk_state == TCP_ESTABLISHED && 1086 inet6_match(net, sk, rmt_addr, loc_addr, ports, dif, sdif)) 1087 return sk; 1088 /* Only check first socket in chain */ 1089 break; 1090 } 1091 return NULL; 1092 } 1093 1094 void udp_v6_early_demux(struct sk_buff *skb) 1095 { 1096 struct net *net = dev_net(skb->dev); 1097 const struct udphdr *uh; 1098 struct sock *sk; 1099 struct dst_entry *dst; 1100 int dif = skb->dev->ifindex; 1101 int sdif = inet6_sdif(skb); 1102 1103 if (!pskb_may_pull(skb, skb_transport_offset(skb) + 1104 sizeof(struct udphdr))) 1105 return; 1106 1107 uh = udp_hdr(skb); 1108 1109 if (skb->pkt_type == PACKET_HOST) 1110 sk = __udp6_lib_demux_lookup(net, uh->dest, 1111 &ipv6_hdr(skb)->daddr, 1112 uh->source, &ipv6_hdr(skb)->saddr, 1113 dif, sdif); 1114 else 1115 return; 1116 1117 if (!sk || !refcount_inc_not_zero(&sk->sk_refcnt)) 1118 return; 1119 1120 skb->sk = sk; 1121 skb->destructor = sock_efree; 1122 dst = rcu_dereference(sk->sk_rx_dst); 1123 1124 if (dst) 1125 dst = dst_check(dst, sk->sk_rx_dst_cookie); 1126 if (dst) { 1127 /* set noref for now. 1128 * any place which wants to hold dst has to call 1129 * dst_hold_safe() 1130 */ 1131 skb_dst_set_noref(skb, dst); 1132 } 1133 } 1134 1135 INDIRECT_CALLABLE_SCOPE int udpv6_rcv(struct sk_buff *skb) 1136 { 1137 return __udp6_lib_rcv(skb, dev_net(skb->dev)->ipv4.udp_table, IPPROTO_UDP); 1138 } 1139 1140 /* 1141 * Throw away all pending data and cancel the corking. Socket is locked. 1142 */ 1143 static void udp_v6_flush_pending_frames(struct sock *sk) 1144 { 1145 struct udp_sock *up = udp_sk(sk); 1146 1147 if (up->pending == AF_INET) 1148 udp_flush_pending_frames(sk); 1149 else if (up->pending) { 1150 up->len = 0; 1151 up->pending = 0; 1152 ip6_flush_pending_frames(sk); 1153 } 1154 } 1155 1156 static int udpv6_pre_connect(struct sock *sk, struct sockaddr *uaddr, 1157 int addr_len) 1158 { 1159 if (addr_len < offsetofend(struct sockaddr, sa_family)) 1160 return -EINVAL; 1161 /* The following checks are replicated from __ip6_datagram_connect() 1162 * and intended to prevent BPF program called below from accessing 1163 * bytes that are out of the bound specified by user in addr_len. 1164 */ 1165 if (uaddr->sa_family == AF_INET) { 1166 if (ipv6_only_sock(sk)) 1167 return -EAFNOSUPPORT; 1168 return udp_pre_connect(sk, uaddr, addr_len); 1169 } 1170 1171 if (addr_len < SIN6_LEN_RFC2133) 1172 return -EINVAL; 1173 1174 return BPF_CGROUP_RUN_PROG_INET6_CONNECT_LOCK(sk, uaddr); 1175 } 1176 1177 /** 1178 * udp6_hwcsum_outgoing - handle outgoing HW checksumming 1179 * @sk: socket we are sending on 1180 * @skb: sk_buff containing the filled-in UDP header 1181 * (checksum field must be zeroed out) 1182 * @saddr: source address 1183 * @daddr: destination address 1184 * @len: length of packet 1185 */ 1186 static void udp6_hwcsum_outgoing(struct sock *sk, struct sk_buff *skb, 1187 const struct in6_addr *saddr, 1188 const struct in6_addr *daddr, int len) 1189 { 1190 unsigned int offset; 1191 struct udphdr *uh = udp_hdr(skb); 1192 struct sk_buff *frags = skb_shinfo(skb)->frag_list; 1193 __wsum csum = 0; 1194 1195 if (!frags) { 1196 /* Only one fragment on the socket. */ 1197 skb->csum_start = skb_transport_header(skb) - skb->head; 1198 skb->csum_offset = offsetof(struct udphdr, check); 1199 uh->check = ~csum_ipv6_magic(saddr, daddr, len, IPPROTO_UDP, 0); 1200 } else { 1201 /* 1202 * HW-checksum won't work as there are two or more 1203 * fragments on the socket so that all csums of sk_buffs 1204 * should be together 1205 */ 1206 offset = skb_transport_offset(skb); 1207 skb->csum = skb_checksum(skb, offset, skb->len - offset, 0); 1208 csum = skb->csum; 1209 1210 skb->ip_summed = CHECKSUM_NONE; 1211 1212 do { 1213 csum = csum_add(csum, frags->csum); 1214 } while ((frags = frags->next)); 1215 1216 uh->check = csum_ipv6_magic(saddr, daddr, len, IPPROTO_UDP, 1217 csum); 1218 if (uh->check == 0) 1219 uh->check = CSUM_MANGLED_0; 1220 } 1221 } 1222 1223 /* 1224 * Sending 1225 */ 1226 1227 static int udp_v6_send_skb(struct sk_buff *skb, struct flowi6 *fl6, 1228 struct inet_cork *cork) 1229 { 1230 struct sock *sk = skb->sk; 1231 struct udphdr *uh; 1232 int err = 0; 1233 int is_udplite = IS_UDPLITE(sk); 1234 __wsum csum = 0; 1235 int offset = skb_transport_offset(skb); 1236 int len = skb->len - offset; 1237 int datalen = len - sizeof(*uh); 1238 1239 /* 1240 * Create a UDP header 1241 */ 1242 uh = udp_hdr(skb); 1243 uh->source = fl6->fl6_sport; 1244 uh->dest = fl6->fl6_dport; 1245 uh->len = htons(len); 1246 uh->check = 0; 1247 1248 if (cork->gso_size) { 1249 const int hlen = skb_network_header_len(skb) + 1250 sizeof(struct udphdr); 1251 1252 if (hlen + cork->gso_size > cork->fragsize) { 1253 kfree_skb(skb); 1254 return -EINVAL; 1255 } 1256 if (datalen > cork->gso_size * UDP_MAX_SEGMENTS) { 1257 kfree_skb(skb); 1258 return -EINVAL; 1259 } 1260 if (udp_sk(sk)->no_check6_tx) { 1261 kfree_skb(skb); 1262 return -EINVAL; 1263 } 1264 if (skb->ip_summed != CHECKSUM_PARTIAL || is_udplite || 1265 dst_xfrm(skb_dst(skb))) { 1266 kfree_skb(skb); 1267 return -EIO; 1268 } 1269 1270 if (datalen > cork->gso_size) { 1271 skb_shinfo(skb)->gso_size = cork->gso_size; 1272 skb_shinfo(skb)->gso_type = SKB_GSO_UDP_L4; 1273 skb_shinfo(skb)->gso_segs = DIV_ROUND_UP(datalen, 1274 cork->gso_size); 1275 } 1276 goto csum_partial; 1277 } 1278 1279 if (is_udplite) 1280 csum = udplite_csum(skb); 1281 else if (udp_sk(sk)->no_check6_tx) { /* UDP csum disabled */ 1282 skb->ip_summed = CHECKSUM_NONE; 1283 goto send; 1284 } else if (skb->ip_summed == CHECKSUM_PARTIAL) { /* UDP hardware csum */ 1285 csum_partial: 1286 udp6_hwcsum_outgoing(sk, skb, &fl6->saddr, &fl6->daddr, len); 1287 goto send; 1288 } else 1289 csum = udp_csum(skb); 1290 1291 /* add protocol-dependent pseudo-header */ 1292 uh->check = csum_ipv6_magic(&fl6->saddr, &fl6->daddr, 1293 len, fl6->flowi6_proto, csum); 1294 if (uh->check == 0) 1295 uh->check = CSUM_MANGLED_0; 1296 1297 send: 1298 err = ip6_send_skb(skb); 1299 if (err) { 1300 if (err == -ENOBUFS && !inet6_sk(sk)->recverr) { 1301 UDP6_INC_STATS(sock_net(sk), 1302 UDP_MIB_SNDBUFERRORS, is_udplite); 1303 err = 0; 1304 } 1305 } else { 1306 UDP6_INC_STATS(sock_net(sk), 1307 UDP_MIB_OUTDATAGRAMS, is_udplite); 1308 } 1309 return err; 1310 } 1311 1312 static int udp_v6_push_pending_frames(struct sock *sk) 1313 { 1314 struct sk_buff *skb; 1315 struct udp_sock *up = udp_sk(sk); 1316 int err = 0; 1317 1318 if (up->pending == AF_INET) 1319 return udp_push_pending_frames(sk); 1320 1321 skb = ip6_finish_skb(sk); 1322 if (!skb) 1323 goto out; 1324 1325 err = udp_v6_send_skb(skb, &inet_sk(sk)->cork.fl.u.ip6, 1326 &inet_sk(sk)->cork.base); 1327 out: 1328 up->len = 0; 1329 up->pending = 0; 1330 return err; 1331 } 1332 1333 int udpv6_sendmsg(struct sock *sk, struct msghdr *msg, size_t len) 1334 { 1335 struct ipv6_txoptions opt_space; 1336 struct udp_sock *up = udp_sk(sk); 1337 struct inet_sock *inet = inet_sk(sk); 1338 struct ipv6_pinfo *np = inet6_sk(sk); 1339 DECLARE_SOCKADDR(struct sockaddr_in6 *, sin6, msg->msg_name); 1340 struct in6_addr *daddr, *final_p, final; 1341 struct ipv6_txoptions *opt = NULL; 1342 struct ipv6_txoptions *opt_to_free = NULL; 1343 struct ip6_flowlabel *flowlabel = NULL; 1344 struct inet_cork_full cork; 1345 struct flowi6 *fl6 = &cork.fl.u.ip6; 1346 struct dst_entry *dst; 1347 struct ipcm6_cookie ipc6; 1348 int addr_len = msg->msg_namelen; 1349 bool connected = false; 1350 int ulen = len; 1351 int corkreq = READ_ONCE(up->corkflag) || msg->msg_flags&MSG_MORE; 1352 int err; 1353 int is_udplite = IS_UDPLITE(sk); 1354 int (*getfrag)(void *, char *, int, int, int, struct sk_buff *); 1355 1356 ipcm6_init(&ipc6); 1357 ipc6.gso_size = READ_ONCE(up->gso_size); 1358 ipc6.sockc.tsflags = sk->sk_tsflags; 1359 ipc6.sockc.mark = sk->sk_mark; 1360 1361 /* destination address check */ 1362 if (sin6) { 1363 if (addr_len < offsetof(struct sockaddr, sa_data)) 1364 return -EINVAL; 1365 1366 switch (sin6->sin6_family) { 1367 case AF_INET6: 1368 if (addr_len < SIN6_LEN_RFC2133) 1369 return -EINVAL; 1370 daddr = &sin6->sin6_addr; 1371 if (ipv6_addr_any(daddr) && 1372 ipv6_addr_v4mapped(&np->saddr)) 1373 ipv6_addr_set_v4mapped(htonl(INADDR_LOOPBACK), 1374 daddr); 1375 break; 1376 case AF_INET: 1377 goto do_udp_sendmsg; 1378 case AF_UNSPEC: 1379 msg->msg_name = sin6 = NULL; 1380 msg->msg_namelen = addr_len = 0; 1381 daddr = NULL; 1382 break; 1383 default: 1384 return -EINVAL; 1385 } 1386 } else if (!up->pending) { 1387 if (sk->sk_state != TCP_ESTABLISHED) 1388 return -EDESTADDRREQ; 1389 daddr = &sk->sk_v6_daddr; 1390 } else 1391 daddr = NULL; 1392 1393 if (daddr) { 1394 if (ipv6_addr_v4mapped(daddr)) { 1395 struct sockaddr_in sin; 1396 sin.sin_family = AF_INET; 1397 sin.sin_port = sin6 ? sin6->sin6_port : inet->inet_dport; 1398 sin.sin_addr.s_addr = daddr->s6_addr32[3]; 1399 msg->msg_name = &sin; 1400 msg->msg_namelen = sizeof(sin); 1401 do_udp_sendmsg: 1402 err = ipv6_only_sock(sk) ? 1403 -ENETUNREACH : udp_sendmsg(sk, msg, len); 1404 msg->msg_name = sin6; 1405 msg->msg_namelen = addr_len; 1406 return err; 1407 } 1408 } 1409 1410 /* Rough check on arithmetic overflow, 1411 better check is made in ip6_append_data(). 1412 */ 1413 if (len > INT_MAX - sizeof(struct udphdr)) 1414 return -EMSGSIZE; 1415 1416 getfrag = is_udplite ? udplite_getfrag : ip_generic_getfrag; 1417 if (up->pending) { 1418 if (up->pending == AF_INET) 1419 return udp_sendmsg(sk, msg, len); 1420 /* 1421 * There are pending frames. 1422 * The socket lock must be held while it's corked. 1423 */ 1424 lock_sock(sk); 1425 if (likely(up->pending)) { 1426 if (unlikely(up->pending != AF_INET6)) { 1427 release_sock(sk); 1428 return -EAFNOSUPPORT; 1429 } 1430 dst = NULL; 1431 goto do_append_data; 1432 } 1433 release_sock(sk); 1434 } 1435 ulen += sizeof(struct udphdr); 1436 1437 memset(fl6, 0, sizeof(*fl6)); 1438 1439 if (sin6) { 1440 if (sin6->sin6_port == 0) 1441 return -EINVAL; 1442 1443 fl6->fl6_dport = sin6->sin6_port; 1444 daddr = &sin6->sin6_addr; 1445 1446 if (np->sndflow) { 1447 fl6->flowlabel = sin6->sin6_flowinfo&IPV6_FLOWINFO_MASK; 1448 if (fl6->flowlabel & IPV6_FLOWLABEL_MASK) { 1449 flowlabel = fl6_sock_lookup(sk, fl6->flowlabel); 1450 if (IS_ERR(flowlabel)) 1451 return -EINVAL; 1452 } 1453 } 1454 1455 /* 1456 * Otherwise it will be difficult to maintain 1457 * sk->sk_dst_cache. 1458 */ 1459 if (sk->sk_state == TCP_ESTABLISHED && 1460 ipv6_addr_equal(daddr, &sk->sk_v6_daddr)) 1461 daddr = &sk->sk_v6_daddr; 1462 1463 if (addr_len >= sizeof(struct sockaddr_in6) && 1464 sin6->sin6_scope_id && 1465 __ipv6_addr_needs_scope_id(__ipv6_addr_type(daddr))) 1466 fl6->flowi6_oif = sin6->sin6_scope_id; 1467 } else { 1468 if (sk->sk_state != TCP_ESTABLISHED) 1469 return -EDESTADDRREQ; 1470 1471 fl6->fl6_dport = inet->inet_dport; 1472 daddr = &sk->sk_v6_daddr; 1473 fl6->flowlabel = np->flow_label; 1474 connected = true; 1475 } 1476 1477 if (!fl6->flowi6_oif) 1478 fl6->flowi6_oif = READ_ONCE(sk->sk_bound_dev_if); 1479 1480 if (!fl6->flowi6_oif) 1481 fl6->flowi6_oif = np->sticky_pktinfo.ipi6_ifindex; 1482 1483 fl6->flowi6_uid = sk->sk_uid; 1484 1485 if (msg->msg_controllen) { 1486 opt = &opt_space; 1487 memset(opt, 0, sizeof(struct ipv6_txoptions)); 1488 opt->tot_len = sizeof(*opt); 1489 ipc6.opt = opt; 1490 1491 err = udp_cmsg_send(sk, msg, &ipc6.gso_size); 1492 if (err > 0) 1493 err = ip6_datagram_send_ctl(sock_net(sk), sk, msg, fl6, 1494 &ipc6); 1495 if (err < 0) { 1496 fl6_sock_release(flowlabel); 1497 return err; 1498 } 1499 if ((fl6->flowlabel&IPV6_FLOWLABEL_MASK) && !flowlabel) { 1500 flowlabel = fl6_sock_lookup(sk, fl6->flowlabel); 1501 if (IS_ERR(flowlabel)) 1502 return -EINVAL; 1503 } 1504 if (!(opt->opt_nflen|opt->opt_flen)) 1505 opt = NULL; 1506 connected = false; 1507 } 1508 if (!opt) { 1509 opt = txopt_get(np); 1510 opt_to_free = opt; 1511 } 1512 if (flowlabel) 1513 opt = fl6_merge_options(&opt_space, flowlabel, opt); 1514 opt = ipv6_fixup_options(&opt_space, opt); 1515 ipc6.opt = opt; 1516 1517 fl6->flowi6_proto = sk->sk_protocol; 1518 fl6->flowi6_mark = ipc6.sockc.mark; 1519 fl6->daddr = *daddr; 1520 if (ipv6_addr_any(&fl6->saddr) && !ipv6_addr_any(&np->saddr)) 1521 fl6->saddr = np->saddr; 1522 fl6->fl6_sport = inet->inet_sport; 1523 1524 if (cgroup_bpf_enabled(CGROUP_UDP6_SENDMSG) && !connected) { 1525 err = BPF_CGROUP_RUN_PROG_UDP6_SENDMSG_LOCK(sk, 1526 (struct sockaddr *)sin6, 1527 &fl6->saddr); 1528 if (err) 1529 goto out_no_dst; 1530 if (sin6) { 1531 if (ipv6_addr_v4mapped(&sin6->sin6_addr)) { 1532 /* BPF program rewrote IPv6-only by IPv4-mapped 1533 * IPv6. It's currently unsupported. 1534 */ 1535 err = -ENOTSUPP; 1536 goto out_no_dst; 1537 } 1538 if (sin6->sin6_port == 0) { 1539 /* BPF program set invalid port. Reject it. */ 1540 err = -EINVAL; 1541 goto out_no_dst; 1542 } 1543 fl6->fl6_dport = sin6->sin6_port; 1544 fl6->daddr = sin6->sin6_addr; 1545 } 1546 } 1547 1548 if (ipv6_addr_any(&fl6->daddr)) 1549 fl6->daddr.s6_addr[15] = 0x1; /* :: means loopback (BSD'ism) */ 1550 1551 final_p = fl6_update_dst(fl6, opt, &final); 1552 if (final_p) 1553 connected = false; 1554 1555 if (!fl6->flowi6_oif && ipv6_addr_is_multicast(&fl6->daddr)) { 1556 fl6->flowi6_oif = np->mcast_oif; 1557 connected = false; 1558 } else if (!fl6->flowi6_oif) 1559 fl6->flowi6_oif = np->ucast_oif; 1560 1561 security_sk_classify_flow(sk, flowi6_to_flowi_common(fl6)); 1562 1563 if (ipc6.tclass < 0) 1564 ipc6.tclass = np->tclass; 1565 1566 fl6->flowlabel = ip6_make_flowinfo(ipc6.tclass, fl6->flowlabel); 1567 1568 dst = ip6_sk_dst_lookup_flow(sk, fl6, final_p, connected); 1569 if (IS_ERR(dst)) { 1570 err = PTR_ERR(dst); 1571 dst = NULL; 1572 goto out; 1573 } 1574 1575 if (ipc6.hlimit < 0) 1576 ipc6.hlimit = ip6_sk_dst_hoplimit(np, fl6, dst); 1577 1578 if (msg->msg_flags&MSG_CONFIRM) 1579 goto do_confirm; 1580 back_from_confirm: 1581 1582 /* Lockless fast path for the non-corking case */ 1583 if (!corkreq) { 1584 struct sk_buff *skb; 1585 1586 skb = ip6_make_skb(sk, getfrag, msg, ulen, 1587 sizeof(struct udphdr), &ipc6, 1588 (struct rt6_info *)dst, 1589 msg->msg_flags, &cork); 1590 err = PTR_ERR(skb); 1591 if (!IS_ERR_OR_NULL(skb)) 1592 err = udp_v6_send_skb(skb, fl6, &cork.base); 1593 /* ip6_make_skb steals dst reference */ 1594 goto out_no_dst; 1595 } 1596 1597 lock_sock(sk); 1598 if (unlikely(up->pending)) { 1599 /* The socket is already corked while preparing it. */ 1600 /* ... which is an evident application bug. --ANK */ 1601 release_sock(sk); 1602 1603 net_dbg_ratelimited("udp cork app bug 2\n"); 1604 err = -EINVAL; 1605 goto out; 1606 } 1607 1608 up->pending = AF_INET6; 1609 1610 do_append_data: 1611 if (ipc6.dontfrag < 0) 1612 ipc6.dontfrag = np->dontfrag; 1613 up->len += ulen; 1614 err = ip6_append_data(sk, getfrag, msg, ulen, sizeof(struct udphdr), 1615 &ipc6, fl6, (struct rt6_info *)dst, 1616 corkreq ? msg->msg_flags|MSG_MORE : msg->msg_flags); 1617 if (err) 1618 udp_v6_flush_pending_frames(sk); 1619 else if (!corkreq) 1620 err = udp_v6_push_pending_frames(sk); 1621 else if (unlikely(skb_queue_empty(&sk->sk_write_queue))) 1622 up->pending = 0; 1623 1624 if (err > 0) 1625 err = np->recverr ? net_xmit_errno(err) : 0; 1626 release_sock(sk); 1627 1628 out: 1629 dst_release(dst); 1630 out_no_dst: 1631 fl6_sock_release(flowlabel); 1632 txopt_put(opt_to_free); 1633 if (!err) 1634 return len; 1635 /* 1636 * ENOBUFS = no kernel mem, SOCK_NOSPACE = no sndbuf space. Reporting 1637 * ENOBUFS might not be good (it's not tunable per se), but otherwise 1638 * we don't have a good statistic (IpOutDiscards but it can be too many 1639 * things). We could add another new stat but at least for now that 1640 * seems like overkill. 1641 */ 1642 if (err == -ENOBUFS || test_bit(SOCK_NOSPACE, &sk->sk_socket->flags)) { 1643 UDP6_INC_STATS(sock_net(sk), 1644 UDP_MIB_SNDBUFERRORS, is_udplite); 1645 } 1646 return err; 1647 1648 do_confirm: 1649 if (msg->msg_flags & MSG_PROBE) 1650 dst_confirm_neigh(dst, &fl6->daddr); 1651 if (!(msg->msg_flags&MSG_PROBE) || len) 1652 goto back_from_confirm; 1653 err = 0; 1654 goto out; 1655 } 1656 EXPORT_SYMBOL(udpv6_sendmsg); 1657 1658 static void udpv6_splice_eof(struct socket *sock) 1659 { 1660 struct sock *sk = sock->sk; 1661 struct udp_sock *up = udp_sk(sk); 1662 1663 if (!up->pending || READ_ONCE(up->corkflag)) 1664 return; 1665 1666 lock_sock(sk); 1667 if (up->pending && !READ_ONCE(up->corkflag)) 1668 udp_v6_push_pending_frames(sk); 1669 release_sock(sk); 1670 } 1671 1672 void udpv6_destroy_sock(struct sock *sk) 1673 { 1674 struct udp_sock *up = udp_sk(sk); 1675 lock_sock(sk); 1676 1677 /* protects from races with udp_abort() */ 1678 sock_set_flag(sk, SOCK_DEAD); 1679 udp_v6_flush_pending_frames(sk); 1680 release_sock(sk); 1681 1682 if (static_branch_unlikely(&udpv6_encap_needed_key)) { 1683 if (up->encap_type) { 1684 void (*encap_destroy)(struct sock *sk); 1685 encap_destroy = READ_ONCE(up->encap_destroy); 1686 if (encap_destroy) 1687 encap_destroy(sk); 1688 } 1689 if (up->encap_enabled) { 1690 static_branch_dec(&udpv6_encap_needed_key); 1691 udp_encap_disable(); 1692 } 1693 } 1694 } 1695 1696 /* 1697 * Socket option code for UDP 1698 */ 1699 int udpv6_setsockopt(struct sock *sk, int level, int optname, sockptr_t optval, 1700 unsigned int optlen) 1701 { 1702 if (level == SOL_UDP || level == SOL_UDPLITE || level == SOL_SOCKET) 1703 return udp_lib_setsockopt(sk, level, optname, 1704 optval, optlen, 1705 udp_v6_push_pending_frames); 1706 return ipv6_setsockopt(sk, level, optname, optval, optlen); 1707 } 1708 1709 int udpv6_getsockopt(struct sock *sk, int level, int optname, 1710 char __user *optval, int __user *optlen) 1711 { 1712 if (level == SOL_UDP || level == SOL_UDPLITE) 1713 return udp_lib_getsockopt(sk, level, optname, optval, optlen); 1714 return ipv6_getsockopt(sk, level, optname, optval, optlen); 1715 } 1716 1717 static const struct inet6_protocol udpv6_protocol = { 1718 .handler = udpv6_rcv, 1719 .err_handler = udpv6_err, 1720 .flags = INET6_PROTO_NOPOLICY|INET6_PROTO_FINAL, 1721 }; 1722 1723 /* ------------------------------------------------------------------------ */ 1724 #ifdef CONFIG_PROC_FS 1725 int udp6_seq_show(struct seq_file *seq, void *v) 1726 { 1727 if (v == SEQ_START_TOKEN) { 1728 seq_puts(seq, IPV6_SEQ_DGRAM_HEADER); 1729 } else { 1730 int bucket = ((struct udp_iter_state *)seq->private)->bucket; 1731 const struct inet_sock *inet = inet_sk((const struct sock *)v); 1732 __u16 srcp = ntohs(inet->inet_sport); 1733 __u16 destp = ntohs(inet->inet_dport); 1734 __ip6_dgram_sock_seq_show(seq, v, srcp, destp, 1735 udp_rqueue_get(v), bucket); 1736 } 1737 return 0; 1738 } 1739 1740 const struct seq_operations udp6_seq_ops = { 1741 .start = udp_seq_start, 1742 .next = udp_seq_next, 1743 .stop = udp_seq_stop, 1744 .show = udp6_seq_show, 1745 }; 1746 EXPORT_SYMBOL(udp6_seq_ops); 1747 1748 static struct udp_seq_afinfo udp6_seq_afinfo = { 1749 .family = AF_INET6, 1750 .udp_table = NULL, 1751 }; 1752 1753 int __net_init udp6_proc_init(struct net *net) 1754 { 1755 if (!proc_create_net_data("udp6", 0444, net->proc_net, &udp6_seq_ops, 1756 sizeof(struct udp_iter_state), &udp6_seq_afinfo)) 1757 return -ENOMEM; 1758 return 0; 1759 } 1760 1761 void udp6_proc_exit(struct net *net) 1762 { 1763 remove_proc_entry("udp6", net->proc_net); 1764 } 1765 #endif /* CONFIG_PROC_FS */ 1766 1767 /* ------------------------------------------------------------------------ */ 1768 1769 struct proto udpv6_prot = { 1770 .name = "UDPv6", 1771 .owner = THIS_MODULE, 1772 .close = udp_lib_close, 1773 .pre_connect = udpv6_pre_connect, 1774 .connect = ip6_datagram_connect, 1775 .disconnect = udp_disconnect, 1776 .ioctl = udp_ioctl, 1777 .init = udpv6_init_sock, 1778 .destroy = udpv6_destroy_sock, 1779 .setsockopt = udpv6_setsockopt, 1780 .getsockopt = udpv6_getsockopt, 1781 .sendmsg = udpv6_sendmsg, 1782 .recvmsg = udpv6_recvmsg, 1783 .splice_eof = udpv6_splice_eof, 1784 .release_cb = ip6_datagram_release_cb, 1785 .hash = udp_lib_hash, 1786 .unhash = udp_lib_unhash, 1787 .rehash = udp_v6_rehash, 1788 .get_port = udp_v6_get_port, 1789 .put_port = udp_lib_unhash, 1790 #ifdef CONFIG_BPF_SYSCALL 1791 .psock_update_sk_prot = udp_bpf_update_proto, 1792 #endif 1793 1794 .memory_allocated = &udp_memory_allocated, 1795 .per_cpu_fw_alloc = &udp_memory_per_cpu_fw_alloc, 1796 1797 .sysctl_mem = sysctl_udp_mem, 1798 .sysctl_wmem_offset = offsetof(struct net, ipv4.sysctl_udp_wmem_min), 1799 .sysctl_rmem_offset = offsetof(struct net, ipv4.sysctl_udp_rmem_min), 1800 .obj_size = sizeof(struct udp6_sock), 1801 .h.udp_table = NULL, 1802 .diag_destroy = udp_abort, 1803 }; 1804 1805 static struct inet_protosw udpv6_protosw = { 1806 .type = SOCK_DGRAM, 1807 .protocol = IPPROTO_UDP, 1808 .prot = &udpv6_prot, 1809 .ops = &inet6_dgram_ops, 1810 .flags = INET_PROTOSW_PERMANENT, 1811 }; 1812 1813 int __init udpv6_init(void) 1814 { 1815 int ret; 1816 1817 ret = inet6_add_protocol(&udpv6_protocol, IPPROTO_UDP); 1818 if (ret) 1819 goto out; 1820 1821 ret = inet6_register_protosw(&udpv6_protosw); 1822 if (ret) 1823 goto out_udpv6_protocol; 1824 out: 1825 return ret; 1826 1827 out_udpv6_protocol: 1828 inet6_del_protocol(&udpv6_protocol, IPPROTO_UDP); 1829 goto out; 1830 } 1831 1832 void udpv6_exit(void) 1833 { 1834 inet6_unregister_protosw(&udpv6_protosw); 1835 inet6_del_protocol(&udpv6_protocol, IPPROTO_UDP); 1836 } 1837