1 /* 2 * Linux INET6 implementation 3 * 4 * Authors: 5 * Pedro Roque <roque@di.fc.ul.pt> 6 * 7 * This program is free software; you can redistribute it and/or 8 * modify it under the terms of the GNU General Public License 9 * as published by the Free Software Foundation; either version 10 * 2 of the License, or (at your option) any later version. 11 */ 12 13 #ifndef _NET_IPV6_H 14 #define _NET_IPV6_H 15 16 #include <linux/ipv6.h> 17 #include <linux/hardirq.h> 18 #include <linux/jhash.h> 19 #include <linux/refcount.h> 20 #include <net/if_inet6.h> 21 #include <net/ndisc.h> 22 #include <net/flow.h> 23 #include <net/flow_dissector.h> 24 #include <net/snmp.h> 25 26 #define SIN6_LEN_RFC2133 24 27 28 #define IPV6_MAXPLEN 65535 29 30 /* 31 * NextHeader field of IPv6 header 32 */ 33 34 #define NEXTHDR_HOP 0 /* Hop-by-hop option header. */ 35 #define NEXTHDR_TCP 6 /* TCP segment. */ 36 #define NEXTHDR_UDP 17 /* UDP message. */ 37 #define NEXTHDR_IPV6 41 /* IPv6 in IPv6 */ 38 #define NEXTHDR_ROUTING 43 /* Routing header. */ 39 #define NEXTHDR_FRAGMENT 44 /* Fragmentation/reassembly header. */ 40 #define NEXTHDR_GRE 47 /* GRE header. */ 41 #define NEXTHDR_ESP 50 /* Encapsulating security payload. */ 42 #define NEXTHDR_AUTH 51 /* Authentication header. */ 43 #define NEXTHDR_ICMP 58 /* ICMP for IPv6. */ 44 #define NEXTHDR_NONE 59 /* No next header */ 45 #define NEXTHDR_DEST 60 /* Destination options header. */ 46 #define NEXTHDR_SCTP 132 /* SCTP message. */ 47 #define NEXTHDR_MOBILITY 135 /* Mobility header. */ 48 49 #define NEXTHDR_MAX 255 50 51 #define IPV6_DEFAULT_HOPLIMIT 64 52 #define IPV6_DEFAULT_MCASTHOPS 1 53 54 /* Limits on Hop-by-Hop and Destination options. 55 * 56 * Per RFC8200 there is no limit on the maximum number or lengths of options in 57 * Hop-by-Hop or Destination options other then the packet must fit in an MTU. 58 * We allow configurable limits in order to mitigate potential denial of 59 * service attacks. 60 * 61 * There are three limits that may be set: 62 * - Limit the number of options in a Hop-by-Hop or Destination options 63 * extension header 64 * - Limit the byte length of a Hop-by-Hop or Destination options extension 65 * header 66 * - Disallow unknown options 67 * 68 * The limits are expressed in corresponding sysctls: 69 * 70 * ipv6.sysctl.max_dst_opts_cnt 71 * ipv6.sysctl.max_hbh_opts_cnt 72 * ipv6.sysctl.max_dst_opts_len 73 * ipv6.sysctl.max_hbh_opts_len 74 * 75 * max_*_opts_cnt is the number of TLVs that are allowed for Destination 76 * options or Hop-by-Hop options. If the number is less than zero then unknown 77 * TLVs are disallowed and the number of known options that are allowed is the 78 * absolute value. Setting the value to INT_MAX indicates no limit. 79 * 80 * max_*_opts_len is the length limit in bytes of a Destination or 81 * Hop-by-Hop options extension header. Setting the value to INT_MAX 82 * indicates no length limit. 83 * 84 * If a limit is exceeded when processing an extension header the packet is 85 * silently discarded. 86 */ 87 88 /* Default limits for Hop-by-Hop and Destination options */ 89 #define IP6_DEFAULT_MAX_DST_OPTS_CNT 8 90 #define IP6_DEFAULT_MAX_HBH_OPTS_CNT 8 91 #define IP6_DEFAULT_MAX_DST_OPTS_LEN INT_MAX /* No limit */ 92 #define IP6_DEFAULT_MAX_HBH_OPTS_LEN INT_MAX /* No limit */ 93 94 /* 95 * Addr type 96 * 97 * type - unicast | multicast 98 * scope - local | site | global 99 * v4 - compat 100 * v4mapped 101 * any 102 * loopback 103 */ 104 105 #define IPV6_ADDR_ANY 0x0000U 106 107 #define IPV6_ADDR_UNICAST 0x0001U 108 #define IPV6_ADDR_MULTICAST 0x0002U 109 110 #define IPV6_ADDR_LOOPBACK 0x0010U 111 #define IPV6_ADDR_LINKLOCAL 0x0020U 112 #define IPV6_ADDR_SITELOCAL 0x0040U 113 114 #define IPV6_ADDR_COMPATv4 0x0080U 115 116 #define IPV6_ADDR_SCOPE_MASK 0x00f0U 117 118 #define IPV6_ADDR_MAPPED 0x1000U 119 120 /* 121 * Addr scopes 122 */ 123 #define IPV6_ADDR_MC_SCOPE(a) \ 124 ((a)->s6_addr[1] & 0x0f) /* nonstandard */ 125 #define __IPV6_ADDR_SCOPE_INVALID -1 126 #define IPV6_ADDR_SCOPE_NODELOCAL 0x01 127 #define IPV6_ADDR_SCOPE_LINKLOCAL 0x02 128 #define IPV6_ADDR_SCOPE_SITELOCAL 0x05 129 #define IPV6_ADDR_SCOPE_ORGLOCAL 0x08 130 #define IPV6_ADDR_SCOPE_GLOBAL 0x0e 131 132 /* 133 * Addr flags 134 */ 135 #define IPV6_ADDR_MC_FLAG_TRANSIENT(a) \ 136 ((a)->s6_addr[1] & 0x10) 137 #define IPV6_ADDR_MC_FLAG_PREFIX(a) \ 138 ((a)->s6_addr[1] & 0x20) 139 #define IPV6_ADDR_MC_FLAG_RENDEZVOUS(a) \ 140 ((a)->s6_addr[1] & 0x40) 141 142 /* 143 * fragmentation header 144 */ 145 146 struct frag_hdr { 147 __u8 nexthdr; 148 __u8 reserved; 149 __be16 frag_off; 150 __be32 identification; 151 }; 152 153 #define IP6_MF 0x0001 154 #define IP6_OFFSET 0xFFF8 155 156 #define IP6_REPLY_MARK(net, mark) \ 157 ((net)->ipv6.sysctl.fwmark_reflect ? (mark) : 0) 158 159 #include <net/sock.h> 160 161 /* sysctls */ 162 extern int sysctl_mld_max_msf; 163 extern int sysctl_mld_qrv; 164 165 #define _DEVINC(net, statname, mod, idev, field) \ 166 ({ \ 167 struct inet6_dev *_idev = (idev); \ 168 if (likely(_idev != NULL)) \ 169 mod##SNMP_INC_STATS64((_idev)->stats.statname, (field));\ 170 mod##SNMP_INC_STATS64((net)->mib.statname##_statistics, (field));\ 171 }) 172 173 /* per device counters are atomic_long_t */ 174 #define _DEVINCATOMIC(net, statname, mod, idev, field) \ 175 ({ \ 176 struct inet6_dev *_idev = (idev); \ 177 if (likely(_idev != NULL)) \ 178 SNMP_INC_STATS_ATOMIC_LONG((_idev)->stats.statname##dev, (field)); \ 179 mod##SNMP_INC_STATS((net)->mib.statname##_statistics, (field));\ 180 }) 181 182 /* per device and per net counters are atomic_long_t */ 183 #define _DEVINC_ATOMIC_ATOMIC(net, statname, idev, field) \ 184 ({ \ 185 struct inet6_dev *_idev = (idev); \ 186 if (likely(_idev != NULL)) \ 187 SNMP_INC_STATS_ATOMIC_LONG((_idev)->stats.statname##dev, (field)); \ 188 SNMP_INC_STATS_ATOMIC_LONG((net)->mib.statname##_statistics, (field));\ 189 }) 190 191 #define _DEVADD(net, statname, mod, idev, field, val) \ 192 ({ \ 193 struct inet6_dev *_idev = (idev); \ 194 if (likely(_idev != NULL)) \ 195 mod##SNMP_ADD_STATS((_idev)->stats.statname, (field), (val)); \ 196 mod##SNMP_ADD_STATS((net)->mib.statname##_statistics, (field), (val));\ 197 }) 198 199 #define _DEVUPD(net, statname, mod, idev, field, val) \ 200 ({ \ 201 struct inet6_dev *_idev = (idev); \ 202 if (likely(_idev != NULL)) \ 203 mod##SNMP_UPD_PO_STATS((_idev)->stats.statname, field, (val)); \ 204 mod##SNMP_UPD_PO_STATS((net)->mib.statname##_statistics, field, (val));\ 205 }) 206 207 /* MIBs */ 208 209 #define IP6_INC_STATS(net, idev,field) \ 210 _DEVINC(net, ipv6, , idev, field) 211 #define __IP6_INC_STATS(net, idev,field) \ 212 _DEVINC(net, ipv6, __, idev, field) 213 #define IP6_ADD_STATS(net, idev,field,val) \ 214 _DEVADD(net, ipv6, , idev, field, val) 215 #define __IP6_ADD_STATS(net, idev,field,val) \ 216 _DEVADD(net, ipv6, __, idev, field, val) 217 #define IP6_UPD_PO_STATS(net, idev,field,val) \ 218 _DEVUPD(net, ipv6, , idev, field, val) 219 #define __IP6_UPD_PO_STATS(net, idev,field,val) \ 220 _DEVUPD(net, ipv6, __, idev, field, val) 221 #define ICMP6_INC_STATS(net, idev, field) \ 222 _DEVINCATOMIC(net, icmpv6, , idev, field) 223 #define __ICMP6_INC_STATS(net, idev, field) \ 224 _DEVINCATOMIC(net, icmpv6, __, idev, field) 225 226 #define ICMP6MSGOUT_INC_STATS(net, idev, field) \ 227 _DEVINC_ATOMIC_ATOMIC(net, icmpv6msg, idev, field +256) 228 #define ICMP6MSGIN_INC_STATS(net, idev, field) \ 229 _DEVINC_ATOMIC_ATOMIC(net, icmpv6msg, idev, field) 230 231 struct ip6_ra_chain { 232 struct ip6_ra_chain *next; 233 struct sock *sk; 234 int sel; 235 void (*destructor)(struct sock *); 236 }; 237 238 extern struct ip6_ra_chain *ip6_ra_chain; 239 extern rwlock_t ip6_ra_lock; 240 241 /* 242 This structure is prepared by protocol, when parsing 243 ancillary data and passed to IPv6. 244 */ 245 246 struct ipv6_txoptions { 247 refcount_t refcnt; 248 /* Length of this structure */ 249 int tot_len; 250 251 /* length of extension headers */ 252 253 __u16 opt_flen; /* after fragment hdr */ 254 __u16 opt_nflen; /* before fragment hdr */ 255 256 struct ipv6_opt_hdr *hopopt; 257 struct ipv6_opt_hdr *dst0opt; 258 struct ipv6_rt_hdr *srcrt; /* Routing Header */ 259 struct ipv6_opt_hdr *dst1opt; 260 struct rcu_head rcu; 261 /* Option buffer, as read by IPV6_PKTOPTIONS, starts here. */ 262 }; 263 264 struct ip6_flowlabel { 265 struct ip6_flowlabel __rcu *next; 266 __be32 label; 267 atomic_t users; 268 struct in6_addr dst; 269 struct ipv6_txoptions *opt; 270 unsigned long linger; 271 struct rcu_head rcu; 272 u8 share; 273 union { 274 struct pid *pid; 275 kuid_t uid; 276 } owner; 277 unsigned long lastuse; 278 unsigned long expires; 279 struct net *fl_net; 280 }; 281 282 #define IPV6_FLOWINFO_MASK cpu_to_be32(0x0FFFFFFF) 283 #define IPV6_FLOWLABEL_MASK cpu_to_be32(0x000FFFFF) 284 #define IPV6_FLOWLABEL_STATELESS_FLAG cpu_to_be32(0x00080000) 285 286 #define IPV6_TCLASS_MASK (IPV6_FLOWINFO_MASK & ~IPV6_FLOWLABEL_MASK) 287 #define IPV6_TCLASS_SHIFT 20 288 289 struct ipv6_fl_socklist { 290 struct ipv6_fl_socklist __rcu *next; 291 struct ip6_flowlabel *fl; 292 struct rcu_head rcu; 293 }; 294 295 struct ipcm6_cookie { 296 __s16 hlimit; 297 __s16 tclass; 298 __s8 dontfrag; 299 struct ipv6_txoptions *opt; 300 }; 301 302 static inline struct ipv6_txoptions *txopt_get(const struct ipv6_pinfo *np) 303 { 304 struct ipv6_txoptions *opt; 305 306 rcu_read_lock(); 307 opt = rcu_dereference(np->opt); 308 if (opt) { 309 if (!refcount_inc_not_zero(&opt->refcnt)) 310 opt = NULL; 311 else 312 opt = rcu_pointer_handoff(opt); 313 } 314 rcu_read_unlock(); 315 return opt; 316 } 317 318 static inline void txopt_put(struct ipv6_txoptions *opt) 319 { 320 if (opt && refcount_dec_and_test(&opt->refcnt)) 321 kfree_rcu(opt, rcu); 322 } 323 324 struct ip6_flowlabel *fl6_sock_lookup(struct sock *sk, __be32 label); 325 struct ipv6_txoptions *fl6_merge_options(struct ipv6_txoptions *opt_space, 326 struct ip6_flowlabel *fl, 327 struct ipv6_txoptions *fopt); 328 void fl6_free_socklist(struct sock *sk); 329 int ipv6_flowlabel_opt(struct sock *sk, char __user *optval, int optlen); 330 int ipv6_flowlabel_opt_get(struct sock *sk, struct in6_flowlabel_req *freq, 331 int flags); 332 int ip6_flowlabel_init(void); 333 void ip6_flowlabel_cleanup(void); 334 335 static inline void fl6_sock_release(struct ip6_flowlabel *fl) 336 { 337 if (fl) 338 atomic_dec(&fl->users); 339 } 340 341 void icmpv6_notify(struct sk_buff *skb, u8 type, u8 code, __be32 info); 342 343 void icmpv6_push_pending_frames(struct sock *sk, struct flowi6 *fl6, 344 struct icmp6hdr *thdr, int len); 345 346 int ip6_ra_control(struct sock *sk, int sel); 347 348 int ipv6_parse_hopopts(struct sk_buff *skb); 349 350 struct ipv6_txoptions *ipv6_dup_options(struct sock *sk, 351 struct ipv6_txoptions *opt); 352 struct ipv6_txoptions *ipv6_renew_options(struct sock *sk, 353 struct ipv6_txoptions *opt, 354 int newtype, 355 struct ipv6_opt_hdr __user *newopt, 356 int newoptlen); 357 struct ipv6_txoptions * 358 ipv6_renew_options_kern(struct sock *sk, 359 struct ipv6_txoptions *opt, 360 int newtype, 361 struct ipv6_opt_hdr *newopt, 362 int newoptlen); 363 struct ipv6_txoptions *ipv6_fixup_options(struct ipv6_txoptions *opt_space, 364 struct ipv6_txoptions *opt); 365 366 bool ipv6_opt_accepted(const struct sock *sk, const struct sk_buff *skb, 367 const struct inet6_skb_parm *opt); 368 struct ipv6_txoptions *ipv6_update_options(struct sock *sk, 369 struct ipv6_txoptions *opt); 370 371 static inline bool ipv6_accept_ra(struct inet6_dev *idev) 372 { 373 /* If forwarding is enabled, RA are not accepted unless the special 374 * hybrid mode (accept_ra=2) is enabled. 375 */ 376 return idev->cnf.forwarding ? idev->cnf.accept_ra == 2 : 377 idev->cnf.accept_ra; 378 } 379 380 #if IS_ENABLED(CONFIG_IPV6) 381 static inline int ip6_frag_mem(struct net *net) 382 { 383 return sum_frag_mem_limit(&net->ipv6.frags); 384 } 385 #endif 386 387 #define IPV6_FRAG_HIGH_THRESH (4 * 1024*1024) /* 4194304 */ 388 #define IPV6_FRAG_LOW_THRESH (3 * 1024*1024) /* 3145728 */ 389 #define IPV6_FRAG_TIMEOUT (60 * HZ) /* 60 seconds */ 390 391 int __ipv6_addr_type(const struct in6_addr *addr); 392 static inline int ipv6_addr_type(const struct in6_addr *addr) 393 { 394 return __ipv6_addr_type(addr) & 0xffff; 395 } 396 397 static inline int ipv6_addr_scope(const struct in6_addr *addr) 398 { 399 return __ipv6_addr_type(addr) & IPV6_ADDR_SCOPE_MASK; 400 } 401 402 static inline int __ipv6_addr_src_scope(int type) 403 { 404 return (type == IPV6_ADDR_ANY) ? __IPV6_ADDR_SCOPE_INVALID : (type >> 16); 405 } 406 407 static inline int ipv6_addr_src_scope(const struct in6_addr *addr) 408 { 409 return __ipv6_addr_src_scope(__ipv6_addr_type(addr)); 410 } 411 412 static inline bool __ipv6_addr_needs_scope_id(int type) 413 { 414 return type & IPV6_ADDR_LINKLOCAL || 415 (type & IPV6_ADDR_MULTICAST && 416 (type & (IPV6_ADDR_LOOPBACK|IPV6_ADDR_LINKLOCAL))); 417 } 418 419 static inline __u32 ipv6_iface_scope_id(const struct in6_addr *addr, int iface) 420 { 421 return __ipv6_addr_needs_scope_id(__ipv6_addr_type(addr)) ? iface : 0; 422 } 423 424 static inline int ipv6_addr_cmp(const struct in6_addr *a1, const struct in6_addr *a2) 425 { 426 return memcmp(a1, a2, sizeof(struct in6_addr)); 427 } 428 429 static inline bool 430 ipv6_masked_addr_cmp(const struct in6_addr *a1, const struct in6_addr *m, 431 const struct in6_addr *a2) 432 { 433 #if defined(CONFIG_HAVE_EFFICIENT_UNALIGNED_ACCESS) && BITS_PER_LONG == 64 434 const unsigned long *ul1 = (const unsigned long *)a1; 435 const unsigned long *ulm = (const unsigned long *)m; 436 const unsigned long *ul2 = (const unsigned long *)a2; 437 438 return !!(((ul1[0] ^ ul2[0]) & ulm[0]) | 439 ((ul1[1] ^ ul2[1]) & ulm[1])); 440 #else 441 return !!(((a1->s6_addr32[0] ^ a2->s6_addr32[0]) & m->s6_addr32[0]) | 442 ((a1->s6_addr32[1] ^ a2->s6_addr32[1]) & m->s6_addr32[1]) | 443 ((a1->s6_addr32[2] ^ a2->s6_addr32[2]) & m->s6_addr32[2]) | 444 ((a1->s6_addr32[3] ^ a2->s6_addr32[3]) & m->s6_addr32[3])); 445 #endif 446 } 447 448 static inline void ipv6_addr_prefix(struct in6_addr *pfx, 449 const struct in6_addr *addr, 450 int plen) 451 { 452 /* caller must guarantee 0 <= plen <= 128 */ 453 int o = plen >> 3, 454 b = plen & 0x7; 455 456 memset(pfx->s6_addr, 0, sizeof(pfx->s6_addr)); 457 memcpy(pfx->s6_addr, addr, o); 458 if (b != 0) 459 pfx->s6_addr[o] = addr->s6_addr[o] & (0xff00 >> b); 460 } 461 462 static inline void ipv6_addr_prefix_copy(struct in6_addr *addr, 463 const struct in6_addr *pfx, 464 int plen) 465 { 466 /* caller must guarantee 0 <= plen <= 128 */ 467 int o = plen >> 3, 468 b = plen & 0x7; 469 470 memcpy(addr->s6_addr, pfx, o); 471 if (b != 0) { 472 addr->s6_addr[o] &= ~(0xff00 >> b); 473 addr->s6_addr[o] |= (pfx->s6_addr[o] & (0xff00 >> b)); 474 } 475 } 476 477 static inline void __ipv6_addr_set_half(__be32 *addr, 478 __be32 wh, __be32 wl) 479 { 480 #if defined(CONFIG_HAVE_EFFICIENT_UNALIGNED_ACCESS) && BITS_PER_LONG == 64 481 #if defined(__BIG_ENDIAN) 482 if (__builtin_constant_p(wh) && __builtin_constant_p(wl)) { 483 *(__force u64 *)addr = ((__force u64)(wh) << 32 | (__force u64)(wl)); 484 return; 485 } 486 #elif defined(__LITTLE_ENDIAN) 487 if (__builtin_constant_p(wl) && __builtin_constant_p(wh)) { 488 *(__force u64 *)addr = ((__force u64)(wl) << 32 | (__force u64)(wh)); 489 return; 490 } 491 #endif 492 #endif 493 addr[0] = wh; 494 addr[1] = wl; 495 } 496 497 static inline void ipv6_addr_set(struct in6_addr *addr, 498 __be32 w1, __be32 w2, 499 __be32 w3, __be32 w4) 500 { 501 __ipv6_addr_set_half(&addr->s6_addr32[0], w1, w2); 502 __ipv6_addr_set_half(&addr->s6_addr32[2], w3, w4); 503 } 504 505 static inline bool ipv6_addr_equal(const struct in6_addr *a1, 506 const struct in6_addr *a2) 507 { 508 #if defined(CONFIG_HAVE_EFFICIENT_UNALIGNED_ACCESS) && BITS_PER_LONG == 64 509 const unsigned long *ul1 = (const unsigned long *)a1; 510 const unsigned long *ul2 = (const unsigned long *)a2; 511 512 return ((ul1[0] ^ ul2[0]) | (ul1[1] ^ ul2[1])) == 0UL; 513 #else 514 return ((a1->s6_addr32[0] ^ a2->s6_addr32[0]) | 515 (a1->s6_addr32[1] ^ a2->s6_addr32[1]) | 516 (a1->s6_addr32[2] ^ a2->s6_addr32[2]) | 517 (a1->s6_addr32[3] ^ a2->s6_addr32[3])) == 0; 518 #endif 519 } 520 521 #if defined(CONFIG_HAVE_EFFICIENT_UNALIGNED_ACCESS) && BITS_PER_LONG == 64 522 static inline bool __ipv6_prefix_equal64_half(const __be64 *a1, 523 const __be64 *a2, 524 unsigned int len) 525 { 526 if (len && ((*a1 ^ *a2) & cpu_to_be64((~0UL) << (64 - len)))) 527 return false; 528 return true; 529 } 530 531 static inline bool ipv6_prefix_equal(const struct in6_addr *addr1, 532 const struct in6_addr *addr2, 533 unsigned int prefixlen) 534 { 535 const __be64 *a1 = (const __be64 *)addr1; 536 const __be64 *a2 = (const __be64 *)addr2; 537 538 if (prefixlen >= 64) { 539 if (a1[0] ^ a2[0]) 540 return false; 541 return __ipv6_prefix_equal64_half(a1 + 1, a2 + 1, prefixlen - 64); 542 } 543 return __ipv6_prefix_equal64_half(a1, a2, prefixlen); 544 } 545 #else 546 static inline bool ipv6_prefix_equal(const struct in6_addr *addr1, 547 const struct in6_addr *addr2, 548 unsigned int prefixlen) 549 { 550 const __be32 *a1 = addr1->s6_addr32; 551 const __be32 *a2 = addr2->s6_addr32; 552 unsigned int pdw, pbi; 553 554 /* check complete u32 in prefix */ 555 pdw = prefixlen >> 5; 556 if (pdw && memcmp(a1, a2, pdw << 2)) 557 return false; 558 559 /* check incomplete u32 in prefix */ 560 pbi = prefixlen & 0x1f; 561 if (pbi && ((a1[pdw] ^ a2[pdw]) & htonl((0xffffffff) << (32 - pbi)))) 562 return false; 563 564 return true; 565 } 566 #endif 567 568 struct inet_frag_queue; 569 570 enum ip6_defrag_users { 571 IP6_DEFRAG_LOCAL_DELIVER, 572 IP6_DEFRAG_CONNTRACK_IN, 573 __IP6_DEFRAG_CONNTRACK_IN = IP6_DEFRAG_CONNTRACK_IN + USHRT_MAX, 574 IP6_DEFRAG_CONNTRACK_OUT, 575 __IP6_DEFRAG_CONNTRACK_OUT = IP6_DEFRAG_CONNTRACK_OUT + USHRT_MAX, 576 IP6_DEFRAG_CONNTRACK_BRIDGE_IN, 577 __IP6_DEFRAG_CONNTRACK_BRIDGE_IN = IP6_DEFRAG_CONNTRACK_BRIDGE_IN + USHRT_MAX, 578 }; 579 580 struct ip6_create_arg { 581 __be32 id; 582 u32 user; 583 const struct in6_addr *src; 584 const struct in6_addr *dst; 585 int iif; 586 u8 ecn; 587 }; 588 589 void ip6_frag_init(struct inet_frag_queue *q, const void *a); 590 bool ip6_frag_match(const struct inet_frag_queue *q, const void *a); 591 592 /* 593 * Equivalent of ipv4 struct ip 594 */ 595 struct frag_queue { 596 struct inet_frag_queue q; 597 598 __be32 id; /* fragment id */ 599 u32 user; 600 struct in6_addr saddr; 601 struct in6_addr daddr; 602 603 int iif; 604 unsigned int csum; 605 __u16 nhoffset; 606 u8 ecn; 607 }; 608 609 void ip6_expire_frag_queue(struct net *net, struct frag_queue *fq, 610 struct inet_frags *frags); 611 612 static inline bool ipv6_addr_any(const struct in6_addr *a) 613 { 614 #if defined(CONFIG_HAVE_EFFICIENT_UNALIGNED_ACCESS) && BITS_PER_LONG == 64 615 const unsigned long *ul = (const unsigned long *)a; 616 617 return (ul[0] | ul[1]) == 0UL; 618 #else 619 return (a->s6_addr32[0] | a->s6_addr32[1] | 620 a->s6_addr32[2] | a->s6_addr32[3]) == 0; 621 #endif 622 } 623 624 static inline u32 ipv6_addr_hash(const struct in6_addr *a) 625 { 626 #if defined(CONFIG_HAVE_EFFICIENT_UNALIGNED_ACCESS) && BITS_PER_LONG == 64 627 const unsigned long *ul = (const unsigned long *)a; 628 unsigned long x = ul[0] ^ ul[1]; 629 630 return (u32)(x ^ (x >> 32)); 631 #else 632 return (__force u32)(a->s6_addr32[0] ^ a->s6_addr32[1] ^ 633 a->s6_addr32[2] ^ a->s6_addr32[3]); 634 #endif 635 } 636 637 /* more secured version of ipv6_addr_hash() */ 638 static inline u32 __ipv6_addr_jhash(const struct in6_addr *a, const u32 initval) 639 { 640 u32 v = (__force u32)a->s6_addr32[0] ^ (__force u32)a->s6_addr32[1]; 641 642 return jhash_3words(v, 643 (__force u32)a->s6_addr32[2], 644 (__force u32)a->s6_addr32[3], 645 initval); 646 } 647 648 static inline bool ipv6_addr_loopback(const struct in6_addr *a) 649 { 650 #if defined(CONFIG_HAVE_EFFICIENT_UNALIGNED_ACCESS) && BITS_PER_LONG == 64 651 const __be64 *be = (const __be64 *)a; 652 653 return (be[0] | (be[1] ^ cpu_to_be64(1))) == 0UL; 654 #else 655 return (a->s6_addr32[0] | a->s6_addr32[1] | 656 a->s6_addr32[2] | (a->s6_addr32[3] ^ cpu_to_be32(1))) == 0; 657 #endif 658 } 659 660 /* 661 * Note that we must __force cast these to unsigned long to make sparse happy, 662 * since all of the endian-annotated types are fixed size regardless of arch. 663 */ 664 static inline bool ipv6_addr_v4mapped(const struct in6_addr *a) 665 { 666 return ( 667 #if defined(CONFIG_HAVE_EFFICIENT_UNALIGNED_ACCESS) && BITS_PER_LONG == 64 668 *(unsigned long *)a | 669 #else 670 (__force unsigned long)(a->s6_addr32[0] | a->s6_addr32[1]) | 671 #endif 672 (__force unsigned long)(a->s6_addr32[2] ^ 673 cpu_to_be32(0x0000ffff))) == 0UL; 674 } 675 676 /* 677 * Check for a RFC 4843 ORCHID address 678 * (Overlay Routable Cryptographic Hash Identifiers) 679 */ 680 static inline bool ipv6_addr_orchid(const struct in6_addr *a) 681 { 682 return (a->s6_addr32[0] & htonl(0xfffffff0)) == htonl(0x20010010); 683 } 684 685 static inline bool ipv6_addr_is_multicast(const struct in6_addr *addr) 686 { 687 return (addr->s6_addr32[0] & htonl(0xFF000000)) == htonl(0xFF000000); 688 } 689 690 static inline void ipv6_addr_set_v4mapped(const __be32 addr, 691 struct in6_addr *v4mapped) 692 { 693 ipv6_addr_set(v4mapped, 694 0, 0, 695 htonl(0x0000FFFF), 696 addr); 697 } 698 699 /* 700 * find the first different bit between two addresses 701 * length of address must be a multiple of 32bits 702 */ 703 static inline int __ipv6_addr_diff32(const void *token1, const void *token2, int addrlen) 704 { 705 const __be32 *a1 = token1, *a2 = token2; 706 int i; 707 708 addrlen >>= 2; 709 710 for (i = 0; i < addrlen; i++) { 711 __be32 xb = a1[i] ^ a2[i]; 712 if (xb) 713 return i * 32 + 31 - __fls(ntohl(xb)); 714 } 715 716 /* 717 * we should *never* get to this point since that 718 * would mean the addrs are equal 719 * 720 * However, we do get to it 8) And exacly, when 721 * addresses are equal 8) 722 * 723 * ip route add 1111::/128 via ... 724 * ip route add 1111::/64 via ... 725 * and we are here. 726 * 727 * Ideally, this function should stop comparison 728 * at prefix length. It does not, but it is still OK, 729 * if returned value is greater than prefix length. 730 * --ANK (980803) 731 */ 732 return addrlen << 5; 733 } 734 735 #if defined(CONFIG_HAVE_EFFICIENT_UNALIGNED_ACCESS) && BITS_PER_LONG == 64 736 static inline int __ipv6_addr_diff64(const void *token1, const void *token2, int addrlen) 737 { 738 const __be64 *a1 = token1, *a2 = token2; 739 int i; 740 741 addrlen >>= 3; 742 743 for (i = 0; i < addrlen; i++) { 744 __be64 xb = a1[i] ^ a2[i]; 745 if (xb) 746 return i * 64 + 63 - __fls(be64_to_cpu(xb)); 747 } 748 749 return addrlen << 6; 750 } 751 #endif 752 753 static inline int __ipv6_addr_diff(const void *token1, const void *token2, int addrlen) 754 { 755 #if defined(CONFIG_HAVE_EFFICIENT_UNALIGNED_ACCESS) && BITS_PER_LONG == 64 756 if (__builtin_constant_p(addrlen) && !(addrlen & 7)) 757 return __ipv6_addr_diff64(token1, token2, addrlen); 758 #endif 759 return __ipv6_addr_diff32(token1, token2, addrlen); 760 } 761 762 static inline int ipv6_addr_diff(const struct in6_addr *a1, const struct in6_addr *a2) 763 { 764 return __ipv6_addr_diff(a1, a2, sizeof(struct in6_addr)); 765 } 766 767 __be32 ipv6_select_ident(struct net *net, 768 const struct in6_addr *daddr, 769 const struct in6_addr *saddr); 770 __be32 ipv6_proxy_select_ident(struct net *net, struct sk_buff *skb); 771 772 int ip6_dst_hoplimit(struct dst_entry *dst); 773 774 static inline int ip6_sk_dst_hoplimit(struct ipv6_pinfo *np, struct flowi6 *fl6, 775 struct dst_entry *dst) 776 { 777 int hlimit; 778 779 if (ipv6_addr_is_multicast(&fl6->daddr)) 780 hlimit = np->mcast_hops; 781 else 782 hlimit = np->hop_limit; 783 if (hlimit < 0) 784 hlimit = ip6_dst_hoplimit(dst); 785 return hlimit; 786 } 787 788 /* copy IPv6 saddr & daddr to flow_keys, possibly using 64bit load/store 789 * Equivalent to : flow->v6addrs.src = iph->saddr; 790 * flow->v6addrs.dst = iph->daddr; 791 */ 792 static inline void iph_to_flow_copy_v6addrs(struct flow_keys *flow, 793 const struct ipv6hdr *iph) 794 { 795 BUILD_BUG_ON(offsetof(typeof(flow->addrs), v6addrs.dst) != 796 offsetof(typeof(flow->addrs), v6addrs.src) + 797 sizeof(flow->addrs.v6addrs.src)); 798 memcpy(&flow->addrs.v6addrs, &iph->saddr, sizeof(flow->addrs.v6addrs)); 799 flow->control.addr_type = FLOW_DISSECTOR_KEY_IPV6_ADDRS; 800 } 801 802 #if IS_ENABLED(CONFIG_IPV6) 803 804 /* Sysctl settings for net ipv6.auto_flowlabels */ 805 #define IP6_AUTO_FLOW_LABEL_OFF 0 806 #define IP6_AUTO_FLOW_LABEL_OPTOUT 1 807 #define IP6_AUTO_FLOW_LABEL_OPTIN 2 808 #define IP6_AUTO_FLOW_LABEL_FORCED 3 809 810 #define IP6_AUTO_FLOW_LABEL_MAX IP6_AUTO_FLOW_LABEL_FORCED 811 812 #define IP6_DEFAULT_AUTO_FLOW_LABELS IP6_AUTO_FLOW_LABEL_OPTOUT 813 814 static inline __be32 ip6_make_flowlabel(struct net *net, struct sk_buff *skb, 815 __be32 flowlabel, bool autolabel, 816 struct flowi6 *fl6) 817 { 818 u32 hash; 819 820 /* @flowlabel may include more than a flow label, eg, the traffic class. 821 * Here we want only the flow label value. 822 */ 823 flowlabel &= IPV6_FLOWLABEL_MASK; 824 825 if (flowlabel || 826 net->ipv6.sysctl.auto_flowlabels == IP6_AUTO_FLOW_LABEL_OFF || 827 (!autolabel && 828 net->ipv6.sysctl.auto_flowlabels != IP6_AUTO_FLOW_LABEL_FORCED)) 829 return flowlabel; 830 831 hash = skb_get_hash_flowi6(skb, fl6); 832 833 /* Since this is being sent on the wire obfuscate hash a bit 834 * to minimize possbility that any useful information to an 835 * attacker is leaked. Only lower 20 bits are relevant. 836 */ 837 rol32(hash, 16); 838 839 flowlabel = (__force __be32)hash & IPV6_FLOWLABEL_MASK; 840 841 if (net->ipv6.sysctl.flowlabel_state_ranges) 842 flowlabel |= IPV6_FLOWLABEL_STATELESS_FLAG; 843 844 return flowlabel; 845 } 846 847 static inline int ip6_default_np_autolabel(struct net *net) 848 { 849 switch (net->ipv6.sysctl.auto_flowlabels) { 850 case IP6_AUTO_FLOW_LABEL_OFF: 851 case IP6_AUTO_FLOW_LABEL_OPTIN: 852 default: 853 return 0; 854 case IP6_AUTO_FLOW_LABEL_OPTOUT: 855 case IP6_AUTO_FLOW_LABEL_FORCED: 856 return 1; 857 } 858 } 859 #else 860 static inline void ip6_set_txhash(struct sock *sk) { } 861 static inline __be32 ip6_make_flowlabel(struct net *net, struct sk_buff *skb, 862 __be32 flowlabel, bool autolabel, 863 struct flowi6 *fl6) 864 { 865 return flowlabel; 866 } 867 static inline int ip6_default_np_autolabel(struct net *net) 868 { 869 return 0; 870 } 871 #endif 872 873 874 /* 875 * Header manipulation 876 */ 877 static inline void ip6_flow_hdr(struct ipv6hdr *hdr, unsigned int tclass, 878 __be32 flowlabel) 879 { 880 *(__be32 *)hdr = htonl(0x60000000 | (tclass << 20)) | flowlabel; 881 } 882 883 static inline __be32 ip6_flowinfo(const struct ipv6hdr *hdr) 884 { 885 return *(__be32 *)hdr & IPV6_FLOWINFO_MASK; 886 } 887 888 static inline __be32 ip6_flowlabel(const struct ipv6hdr *hdr) 889 { 890 return *(__be32 *)hdr & IPV6_FLOWLABEL_MASK; 891 } 892 893 static inline u8 ip6_tclass(__be32 flowinfo) 894 { 895 return ntohl(flowinfo & IPV6_TCLASS_MASK) >> IPV6_TCLASS_SHIFT; 896 } 897 898 static inline __be32 ip6_make_flowinfo(unsigned int tclass, __be32 flowlabel) 899 { 900 return htonl(tclass << IPV6_TCLASS_SHIFT) | flowlabel; 901 } 902 903 /* 904 * Prototypes exported by ipv6 905 */ 906 907 /* 908 * rcv function (called from netdevice level) 909 */ 910 911 int ipv6_rcv(struct sk_buff *skb, struct net_device *dev, 912 struct packet_type *pt, struct net_device *orig_dev); 913 914 int ip6_rcv_finish(struct net *net, struct sock *sk, struct sk_buff *skb); 915 916 /* 917 * upper-layer output functions 918 */ 919 int ip6_xmit(const struct sock *sk, struct sk_buff *skb, struct flowi6 *fl6, 920 __u32 mark, struct ipv6_txoptions *opt, int tclass); 921 922 int ip6_find_1stfragopt(struct sk_buff *skb, u8 **nexthdr); 923 924 int ip6_append_data(struct sock *sk, 925 int getfrag(void *from, char *to, int offset, int len, 926 int odd, struct sk_buff *skb), 927 void *from, int length, int transhdrlen, 928 struct ipcm6_cookie *ipc6, struct flowi6 *fl6, 929 struct rt6_info *rt, unsigned int flags, 930 const struct sockcm_cookie *sockc); 931 932 int ip6_push_pending_frames(struct sock *sk); 933 934 void ip6_flush_pending_frames(struct sock *sk); 935 936 int ip6_send_skb(struct sk_buff *skb); 937 938 struct sk_buff *__ip6_make_skb(struct sock *sk, struct sk_buff_head *queue, 939 struct inet_cork_full *cork, 940 struct inet6_cork *v6_cork); 941 struct sk_buff *ip6_make_skb(struct sock *sk, 942 int getfrag(void *from, char *to, int offset, 943 int len, int odd, struct sk_buff *skb), 944 void *from, int length, int transhdrlen, 945 struct ipcm6_cookie *ipc6, struct flowi6 *fl6, 946 struct rt6_info *rt, unsigned int flags, 947 const struct sockcm_cookie *sockc); 948 949 static inline struct sk_buff *ip6_finish_skb(struct sock *sk) 950 { 951 return __ip6_make_skb(sk, &sk->sk_write_queue, &inet_sk(sk)->cork, 952 &inet6_sk(sk)->cork); 953 } 954 955 int ip6_dst_lookup(struct net *net, struct sock *sk, struct dst_entry **dst, 956 struct flowi6 *fl6); 957 struct dst_entry *ip6_dst_lookup_flow(const struct sock *sk, struct flowi6 *fl6, 958 const struct in6_addr *final_dst); 959 struct dst_entry *ip6_sk_dst_lookup_flow(struct sock *sk, struct flowi6 *fl6, 960 const struct in6_addr *final_dst); 961 struct dst_entry *ip6_blackhole_route(struct net *net, 962 struct dst_entry *orig_dst); 963 964 /* 965 * skb processing functions 966 */ 967 968 int ip6_output(struct net *net, struct sock *sk, struct sk_buff *skb); 969 int ip6_forward(struct sk_buff *skb); 970 int ip6_input(struct sk_buff *skb); 971 int ip6_mc_input(struct sk_buff *skb); 972 973 int __ip6_local_out(struct net *net, struct sock *sk, struct sk_buff *skb); 974 int ip6_local_out(struct net *net, struct sock *sk, struct sk_buff *skb); 975 976 /* 977 * Extension header (options) processing 978 */ 979 980 void ipv6_push_nfrag_opts(struct sk_buff *skb, struct ipv6_txoptions *opt, 981 u8 *proto, struct in6_addr **daddr_p, 982 struct in6_addr *saddr); 983 void ipv6_push_frag_opts(struct sk_buff *skb, struct ipv6_txoptions *opt, 984 u8 *proto); 985 986 int ipv6_skip_exthdr(const struct sk_buff *, int start, u8 *nexthdrp, 987 __be16 *frag_offp); 988 989 bool ipv6_ext_hdr(u8 nexthdr); 990 991 enum { 992 IP6_FH_F_FRAG = (1 << 0), 993 IP6_FH_F_AUTH = (1 << 1), 994 IP6_FH_F_SKIP_RH = (1 << 2), 995 }; 996 997 /* find specified header and get offset to it */ 998 int ipv6_find_hdr(const struct sk_buff *skb, unsigned int *offset, int target, 999 unsigned short *fragoff, int *fragflg); 1000 1001 int ipv6_find_tlv(const struct sk_buff *skb, int offset, int type); 1002 1003 struct in6_addr *fl6_update_dst(struct flowi6 *fl6, 1004 const struct ipv6_txoptions *opt, 1005 struct in6_addr *orig); 1006 1007 /* 1008 * socket options (ipv6_sockglue.c) 1009 */ 1010 1011 int ipv6_setsockopt(struct sock *sk, int level, int optname, 1012 char __user *optval, unsigned int optlen); 1013 int ipv6_getsockopt(struct sock *sk, int level, int optname, 1014 char __user *optval, int __user *optlen); 1015 int compat_ipv6_setsockopt(struct sock *sk, int level, int optname, 1016 char __user *optval, unsigned int optlen); 1017 int compat_ipv6_getsockopt(struct sock *sk, int level, int optname, 1018 char __user *optval, int __user *optlen); 1019 1020 int __ip6_datagram_connect(struct sock *sk, struct sockaddr *addr, 1021 int addr_len); 1022 int ip6_datagram_connect(struct sock *sk, struct sockaddr *addr, int addr_len); 1023 int ip6_datagram_connect_v6_only(struct sock *sk, struct sockaddr *addr, 1024 int addr_len); 1025 int ip6_datagram_dst_update(struct sock *sk, bool fix_sk_saddr); 1026 void ip6_datagram_release_cb(struct sock *sk); 1027 1028 int ipv6_recv_error(struct sock *sk, struct msghdr *msg, int len, 1029 int *addr_len); 1030 int ipv6_recv_rxpmtu(struct sock *sk, struct msghdr *msg, int len, 1031 int *addr_len); 1032 void ipv6_icmp_error(struct sock *sk, struct sk_buff *skb, int err, __be16 port, 1033 u32 info, u8 *payload); 1034 void ipv6_local_error(struct sock *sk, int err, struct flowi6 *fl6, u32 info); 1035 void ipv6_local_rxpmtu(struct sock *sk, struct flowi6 *fl6, u32 mtu); 1036 1037 int inet6_release(struct socket *sock); 1038 int inet6_bind(struct socket *sock, struct sockaddr *uaddr, int addr_len); 1039 int inet6_getname(struct socket *sock, struct sockaddr *uaddr, int *uaddr_len, 1040 int peer); 1041 int inet6_ioctl(struct socket *sock, unsigned int cmd, unsigned long arg); 1042 1043 int inet6_hash_connect(struct inet_timewait_death_row *death_row, 1044 struct sock *sk); 1045 1046 /* 1047 * reassembly.c 1048 */ 1049 extern const struct proto_ops inet6_stream_ops; 1050 extern const struct proto_ops inet6_dgram_ops; 1051 extern const struct proto_ops inet6_sockraw_ops; 1052 1053 struct group_source_req; 1054 struct group_filter; 1055 1056 int ip6_mc_source(int add, int omode, struct sock *sk, 1057 struct group_source_req *pgsr); 1058 int ip6_mc_msfilter(struct sock *sk, struct group_filter *gsf); 1059 int ip6_mc_msfget(struct sock *sk, struct group_filter *gsf, 1060 struct group_filter __user *optval, int __user *optlen); 1061 1062 #ifdef CONFIG_PROC_FS 1063 int ac6_proc_init(struct net *net); 1064 void ac6_proc_exit(struct net *net); 1065 int raw6_proc_init(void); 1066 void raw6_proc_exit(void); 1067 int tcp6_proc_init(struct net *net); 1068 void tcp6_proc_exit(struct net *net); 1069 int udp6_proc_init(struct net *net); 1070 void udp6_proc_exit(struct net *net); 1071 int udplite6_proc_init(void); 1072 void udplite6_proc_exit(void); 1073 int ipv6_misc_proc_init(void); 1074 void ipv6_misc_proc_exit(void); 1075 int snmp6_register_dev(struct inet6_dev *idev); 1076 int snmp6_unregister_dev(struct inet6_dev *idev); 1077 1078 #else 1079 static inline int ac6_proc_init(struct net *net) { return 0; } 1080 static inline void ac6_proc_exit(struct net *net) { } 1081 static inline int snmp6_register_dev(struct inet6_dev *idev) { return 0; } 1082 static inline int snmp6_unregister_dev(struct inet6_dev *idev) { return 0; } 1083 #endif 1084 1085 #ifdef CONFIG_SYSCTL 1086 extern struct ctl_table ipv6_route_table_template[]; 1087 1088 struct ctl_table *ipv6_icmp_sysctl_init(struct net *net); 1089 struct ctl_table *ipv6_route_sysctl_init(struct net *net); 1090 int ipv6_sysctl_register(void); 1091 void ipv6_sysctl_unregister(void); 1092 #endif 1093 1094 int ipv6_sock_mc_join(struct sock *sk, int ifindex, 1095 const struct in6_addr *addr); 1096 int ipv6_sock_mc_drop(struct sock *sk, int ifindex, 1097 const struct in6_addr *addr); 1098 #endif /* _NET_IPV6_H */ 1099