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 #include <net/netns/hash.h> 26 27 #define SIN6_LEN_RFC2133 24 28 29 #define IPV6_MAXPLEN 65535 30 31 /* 32 * NextHeader field of IPv6 header 33 */ 34 35 #define NEXTHDR_HOP 0 /* Hop-by-hop option header. */ 36 #define NEXTHDR_TCP 6 /* TCP segment. */ 37 #define NEXTHDR_UDP 17 /* UDP message. */ 38 #define NEXTHDR_IPV6 41 /* IPv6 in IPv6 */ 39 #define NEXTHDR_ROUTING 43 /* Routing header. */ 40 #define NEXTHDR_FRAGMENT 44 /* Fragmentation/reassembly header. */ 41 #define NEXTHDR_GRE 47 /* GRE header. */ 42 #define NEXTHDR_ESP 50 /* Encapsulating security payload. */ 43 #define NEXTHDR_AUTH 51 /* Authentication header. */ 44 #define NEXTHDR_ICMP 58 /* ICMP for IPv6. */ 45 #define NEXTHDR_NONE 59 /* No next header */ 46 #define NEXTHDR_DEST 60 /* Destination options header. */ 47 #define NEXTHDR_SCTP 132 /* SCTP message. */ 48 #define NEXTHDR_MOBILITY 135 /* Mobility header. */ 49 50 #define NEXTHDR_MAX 255 51 52 #define IPV6_DEFAULT_HOPLIMIT 64 53 #define IPV6_DEFAULT_MCASTHOPS 1 54 55 /* Limits on Hop-by-Hop and Destination options. 56 * 57 * Per RFC8200 there is no limit on the maximum number or lengths of options in 58 * Hop-by-Hop or Destination options other then the packet must fit in an MTU. 59 * We allow configurable limits in order to mitigate potential denial of 60 * service attacks. 61 * 62 * There are three limits that may be set: 63 * - Limit the number of options in a Hop-by-Hop or Destination options 64 * extension header 65 * - Limit the byte length of a Hop-by-Hop or Destination options extension 66 * header 67 * - Disallow unknown options 68 * 69 * The limits are expressed in corresponding sysctls: 70 * 71 * ipv6.sysctl.max_dst_opts_cnt 72 * ipv6.sysctl.max_hbh_opts_cnt 73 * ipv6.sysctl.max_dst_opts_len 74 * ipv6.sysctl.max_hbh_opts_len 75 * 76 * max_*_opts_cnt is the number of TLVs that are allowed for Destination 77 * options or Hop-by-Hop options. If the number is less than zero then unknown 78 * TLVs are disallowed and the number of known options that are allowed is the 79 * absolute value. Setting the value to INT_MAX indicates no limit. 80 * 81 * max_*_opts_len is the length limit in bytes of a Destination or 82 * Hop-by-Hop options extension header. Setting the value to INT_MAX 83 * indicates no length limit. 84 * 85 * If a limit is exceeded when processing an extension header the packet is 86 * silently discarded. 87 */ 88 89 /* Default limits for Hop-by-Hop and Destination options */ 90 #define IP6_DEFAULT_MAX_DST_OPTS_CNT 8 91 #define IP6_DEFAULT_MAX_HBH_OPTS_CNT 8 92 #define IP6_DEFAULT_MAX_DST_OPTS_LEN INT_MAX /* No limit */ 93 #define IP6_DEFAULT_MAX_HBH_OPTS_LEN INT_MAX /* No limit */ 94 95 /* 96 * Addr type 97 * 98 * type - unicast | multicast 99 * scope - local | site | global 100 * v4 - compat 101 * v4mapped 102 * any 103 * loopback 104 */ 105 106 #define IPV6_ADDR_ANY 0x0000U 107 108 #define IPV6_ADDR_UNICAST 0x0001U 109 #define IPV6_ADDR_MULTICAST 0x0002U 110 111 #define IPV6_ADDR_LOOPBACK 0x0010U 112 #define IPV6_ADDR_LINKLOCAL 0x0020U 113 #define IPV6_ADDR_SITELOCAL 0x0040U 114 115 #define IPV6_ADDR_COMPATv4 0x0080U 116 117 #define IPV6_ADDR_SCOPE_MASK 0x00f0U 118 119 #define IPV6_ADDR_MAPPED 0x1000U 120 121 /* 122 * Addr scopes 123 */ 124 #define IPV6_ADDR_MC_SCOPE(a) \ 125 ((a)->s6_addr[1] & 0x0f) /* nonstandard */ 126 #define __IPV6_ADDR_SCOPE_INVALID -1 127 #define IPV6_ADDR_SCOPE_NODELOCAL 0x01 128 #define IPV6_ADDR_SCOPE_LINKLOCAL 0x02 129 #define IPV6_ADDR_SCOPE_SITELOCAL 0x05 130 #define IPV6_ADDR_SCOPE_ORGLOCAL 0x08 131 #define IPV6_ADDR_SCOPE_GLOBAL 0x0e 132 133 /* 134 * Addr flags 135 */ 136 #define IPV6_ADDR_MC_FLAG_TRANSIENT(a) \ 137 ((a)->s6_addr[1] & 0x10) 138 #define IPV6_ADDR_MC_FLAG_PREFIX(a) \ 139 ((a)->s6_addr[1] & 0x20) 140 #define IPV6_ADDR_MC_FLAG_RENDEZVOUS(a) \ 141 ((a)->s6_addr[1] & 0x40) 142 143 /* 144 * fragmentation header 145 */ 146 147 struct frag_hdr { 148 __u8 nexthdr; 149 __u8 reserved; 150 __be16 frag_off; 151 __be32 identification; 152 }; 153 154 #define IP6_MF 0x0001 155 #define IP6_OFFSET 0xFFF8 156 157 struct ip6_fraglist_iter { 158 struct ipv6hdr *tmp_hdr; 159 struct sk_buff *frag_list; 160 struct sk_buff *frag; 161 int offset; 162 unsigned int hlen; 163 __be32 frag_id; 164 u8 nexthdr; 165 }; 166 167 int ip6_fraglist_init(struct sk_buff *skb, unsigned int hlen, u8 *prevhdr, 168 u8 nexthdr, __be32 frag_id, 169 struct ip6_fraglist_iter *iter); 170 void ip6_fraglist_prepare(struct sk_buff *skb, struct ip6_fraglist_iter *iter); 171 172 static inline struct sk_buff *ip6_fraglist_next(struct ip6_fraglist_iter *iter) 173 { 174 struct sk_buff *skb = iter->frag; 175 176 iter->frag = skb->next; 177 skb_mark_not_on_list(skb); 178 179 return skb; 180 } 181 182 struct ip6_frag_state { 183 u8 *prevhdr; 184 unsigned int hlen; 185 unsigned int mtu; 186 unsigned int left; 187 int offset; 188 int ptr; 189 int hroom; 190 int troom; 191 __be32 frag_id; 192 u8 nexthdr; 193 }; 194 195 void ip6_frag_init(struct sk_buff *skb, unsigned int hlen, unsigned int mtu, 196 unsigned short needed_tailroom, int hdr_room, u8 *prevhdr, 197 u8 nexthdr, __be32 frag_id, struct ip6_frag_state *state); 198 struct sk_buff *ip6_frag_next(struct sk_buff *skb, 199 struct ip6_frag_state *state); 200 201 #define IP6_REPLY_MARK(net, mark) \ 202 ((net)->ipv6.sysctl.fwmark_reflect ? (mark) : 0) 203 204 #include <net/sock.h> 205 206 /* sysctls */ 207 extern int sysctl_mld_max_msf; 208 extern int sysctl_mld_qrv; 209 210 #define _DEVINC(net, statname, mod, idev, field) \ 211 ({ \ 212 struct inet6_dev *_idev = (idev); \ 213 if (likely(_idev != NULL)) \ 214 mod##SNMP_INC_STATS64((_idev)->stats.statname, (field));\ 215 mod##SNMP_INC_STATS64((net)->mib.statname##_statistics, (field));\ 216 }) 217 218 /* per device counters are atomic_long_t */ 219 #define _DEVINCATOMIC(net, statname, mod, idev, field) \ 220 ({ \ 221 struct inet6_dev *_idev = (idev); \ 222 if (likely(_idev != NULL)) \ 223 SNMP_INC_STATS_ATOMIC_LONG((_idev)->stats.statname##dev, (field)); \ 224 mod##SNMP_INC_STATS((net)->mib.statname##_statistics, (field));\ 225 }) 226 227 /* per device and per net counters are atomic_long_t */ 228 #define _DEVINC_ATOMIC_ATOMIC(net, statname, idev, field) \ 229 ({ \ 230 struct inet6_dev *_idev = (idev); \ 231 if (likely(_idev != NULL)) \ 232 SNMP_INC_STATS_ATOMIC_LONG((_idev)->stats.statname##dev, (field)); \ 233 SNMP_INC_STATS_ATOMIC_LONG((net)->mib.statname##_statistics, (field));\ 234 }) 235 236 #define _DEVADD(net, statname, mod, idev, field, val) \ 237 ({ \ 238 struct inet6_dev *_idev = (idev); \ 239 if (likely(_idev != NULL)) \ 240 mod##SNMP_ADD_STATS((_idev)->stats.statname, (field), (val)); \ 241 mod##SNMP_ADD_STATS((net)->mib.statname##_statistics, (field), (val));\ 242 }) 243 244 #define _DEVUPD(net, statname, mod, idev, field, val) \ 245 ({ \ 246 struct inet6_dev *_idev = (idev); \ 247 if (likely(_idev != NULL)) \ 248 mod##SNMP_UPD_PO_STATS((_idev)->stats.statname, field, (val)); \ 249 mod##SNMP_UPD_PO_STATS((net)->mib.statname##_statistics, field, (val));\ 250 }) 251 252 /* MIBs */ 253 254 #define IP6_INC_STATS(net, idev,field) \ 255 _DEVINC(net, ipv6, , idev, field) 256 #define __IP6_INC_STATS(net, idev,field) \ 257 _DEVINC(net, ipv6, __, idev, field) 258 #define IP6_ADD_STATS(net, idev,field,val) \ 259 _DEVADD(net, ipv6, , idev, field, val) 260 #define __IP6_ADD_STATS(net, idev,field,val) \ 261 _DEVADD(net, ipv6, __, idev, field, val) 262 #define IP6_UPD_PO_STATS(net, idev,field,val) \ 263 _DEVUPD(net, ipv6, , idev, field, val) 264 #define __IP6_UPD_PO_STATS(net, idev,field,val) \ 265 _DEVUPD(net, ipv6, __, idev, field, val) 266 #define ICMP6_INC_STATS(net, idev, field) \ 267 _DEVINCATOMIC(net, icmpv6, , idev, field) 268 #define __ICMP6_INC_STATS(net, idev, field) \ 269 _DEVINCATOMIC(net, icmpv6, __, idev, field) 270 271 #define ICMP6MSGOUT_INC_STATS(net, idev, field) \ 272 _DEVINC_ATOMIC_ATOMIC(net, icmpv6msg, idev, field +256) 273 #define ICMP6MSGIN_INC_STATS(net, idev, field) \ 274 _DEVINC_ATOMIC_ATOMIC(net, icmpv6msg, idev, field) 275 276 struct ip6_ra_chain { 277 struct ip6_ra_chain *next; 278 struct sock *sk; 279 int sel; 280 void (*destructor)(struct sock *); 281 }; 282 283 extern struct ip6_ra_chain *ip6_ra_chain; 284 extern rwlock_t ip6_ra_lock; 285 286 /* 287 This structure is prepared by protocol, when parsing 288 ancillary data and passed to IPv6. 289 */ 290 291 struct ipv6_txoptions { 292 refcount_t refcnt; 293 /* Length of this structure */ 294 int tot_len; 295 296 /* length of extension headers */ 297 298 __u16 opt_flen; /* after fragment hdr */ 299 __u16 opt_nflen; /* before fragment hdr */ 300 301 struct ipv6_opt_hdr *hopopt; 302 struct ipv6_opt_hdr *dst0opt; 303 struct ipv6_rt_hdr *srcrt; /* Routing Header */ 304 struct ipv6_opt_hdr *dst1opt; 305 struct rcu_head rcu; 306 /* Option buffer, as read by IPV6_PKTOPTIONS, starts here. */ 307 }; 308 309 struct ip6_flowlabel { 310 struct ip6_flowlabel __rcu *next; 311 __be32 label; 312 atomic_t users; 313 struct in6_addr dst; 314 struct ipv6_txoptions *opt; 315 unsigned long linger; 316 struct rcu_head rcu; 317 u8 share; 318 union { 319 struct pid *pid; 320 kuid_t uid; 321 } owner; 322 unsigned long lastuse; 323 unsigned long expires; 324 struct net *fl_net; 325 }; 326 327 #define IPV6_FLOWINFO_MASK cpu_to_be32(0x0FFFFFFF) 328 #define IPV6_FLOWLABEL_MASK cpu_to_be32(0x000FFFFF) 329 #define IPV6_FLOWLABEL_STATELESS_FLAG cpu_to_be32(0x00080000) 330 331 #define IPV6_TCLASS_MASK (IPV6_FLOWINFO_MASK & ~IPV6_FLOWLABEL_MASK) 332 #define IPV6_TCLASS_SHIFT 20 333 334 struct ipv6_fl_socklist { 335 struct ipv6_fl_socklist __rcu *next; 336 struct ip6_flowlabel *fl; 337 struct rcu_head rcu; 338 }; 339 340 struct ipcm6_cookie { 341 struct sockcm_cookie sockc; 342 __s16 hlimit; 343 __s16 tclass; 344 __s8 dontfrag; 345 struct ipv6_txoptions *opt; 346 __u16 gso_size; 347 }; 348 349 static inline void ipcm6_init(struct ipcm6_cookie *ipc6) 350 { 351 *ipc6 = (struct ipcm6_cookie) { 352 .hlimit = -1, 353 .tclass = -1, 354 .dontfrag = -1, 355 }; 356 } 357 358 static inline void ipcm6_init_sk(struct ipcm6_cookie *ipc6, 359 const struct ipv6_pinfo *np) 360 { 361 *ipc6 = (struct ipcm6_cookie) { 362 .hlimit = -1, 363 .tclass = np->tclass, 364 .dontfrag = np->dontfrag, 365 }; 366 } 367 368 static inline struct ipv6_txoptions *txopt_get(const struct ipv6_pinfo *np) 369 { 370 struct ipv6_txoptions *opt; 371 372 rcu_read_lock(); 373 opt = rcu_dereference(np->opt); 374 if (opt) { 375 if (!refcount_inc_not_zero(&opt->refcnt)) 376 opt = NULL; 377 else 378 opt = rcu_pointer_handoff(opt); 379 } 380 rcu_read_unlock(); 381 return opt; 382 } 383 384 static inline void txopt_put(struct ipv6_txoptions *opt) 385 { 386 if (opt && refcount_dec_and_test(&opt->refcnt)) 387 kfree_rcu(opt, rcu); 388 } 389 390 struct ip6_flowlabel *fl6_sock_lookup(struct sock *sk, __be32 label); 391 struct ipv6_txoptions *fl6_merge_options(struct ipv6_txoptions *opt_space, 392 struct ip6_flowlabel *fl, 393 struct ipv6_txoptions *fopt); 394 void fl6_free_socklist(struct sock *sk); 395 int ipv6_flowlabel_opt(struct sock *sk, char __user *optval, int optlen); 396 int ipv6_flowlabel_opt_get(struct sock *sk, struct in6_flowlabel_req *freq, 397 int flags); 398 int ip6_flowlabel_init(void); 399 void ip6_flowlabel_cleanup(void); 400 bool ip6_autoflowlabel(struct net *net, const struct ipv6_pinfo *np); 401 402 static inline void fl6_sock_release(struct ip6_flowlabel *fl) 403 { 404 if (fl) 405 atomic_dec(&fl->users); 406 } 407 408 void icmpv6_notify(struct sk_buff *skb, u8 type, u8 code, __be32 info); 409 410 void icmpv6_push_pending_frames(struct sock *sk, struct flowi6 *fl6, 411 struct icmp6hdr *thdr, int len); 412 413 int ip6_ra_control(struct sock *sk, int sel); 414 415 int ipv6_parse_hopopts(struct sk_buff *skb); 416 417 struct ipv6_txoptions *ipv6_dup_options(struct sock *sk, 418 struct ipv6_txoptions *opt); 419 struct ipv6_txoptions *ipv6_renew_options(struct sock *sk, 420 struct ipv6_txoptions *opt, 421 int newtype, 422 struct ipv6_opt_hdr *newopt); 423 struct ipv6_txoptions *ipv6_fixup_options(struct ipv6_txoptions *opt_space, 424 struct ipv6_txoptions *opt); 425 426 bool ipv6_opt_accepted(const struct sock *sk, const struct sk_buff *skb, 427 const struct inet6_skb_parm *opt); 428 struct ipv6_txoptions *ipv6_update_options(struct sock *sk, 429 struct ipv6_txoptions *opt); 430 431 static inline bool ipv6_accept_ra(struct inet6_dev *idev) 432 { 433 /* If forwarding is enabled, RA are not accepted unless the special 434 * hybrid mode (accept_ra=2) is enabled. 435 */ 436 return idev->cnf.forwarding ? idev->cnf.accept_ra == 2 : 437 idev->cnf.accept_ra; 438 } 439 440 #define IPV6_FRAG_HIGH_THRESH (4 * 1024*1024) /* 4194304 */ 441 #define IPV6_FRAG_LOW_THRESH (3 * 1024*1024) /* 3145728 */ 442 #define IPV6_FRAG_TIMEOUT (60 * HZ) /* 60 seconds */ 443 444 int __ipv6_addr_type(const struct in6_addr *addr); 445 static inline int ipv6_addr_type(const struct in6_addr *addr) 446 { 447 return __ipv6_addr_type(addr) & 0xffff; 448 } 449 450 static inline int ipv6_addr_scope(const struct in6_addr *addr) 451 { 452 return __ipv6_addr_type(addr) & IPV6_ADDR_SCOPE_MASK; 453 } 454 455 static inline int __ipv6_addr_src_scope(int type) 456 { 457 return (type == IPV6_ADDR_ANY) ? __IPV6_ADDR_SCOPE_INVALID : (type >> 16); 458 } 459 460 static inline int ipv6_addr_src_scope(const struct in6_addr *addr) 461 { 462 return __ipv6_addr_src_scope(__ipv6_addr_type(addr)); 463 } 464 465 static inline bool __ipv6_addr_needs_scope_id(int type) 466 { 467 return type & IPV6_ADDR_LINKLOCAL || 468 (type & IPV6_ADDR_MULTICAST && 469 (type & (IPV6_ADDR_LOOPBACK|IPV6_ADDR_LINKLOCAL))); 470 } 471 472 static inline __u32 ipv6_iface_scope_id(const struct in6_addr *addr, int iface) 473 { 474 return __ipv6_addr_needs_scope_id(__ipv6_addr_type(addr)) ? iface : 0; 475 } 476 477 static inline int ipv6_addr_cmp(const struct in6_addr *a1, const struct in6_addr *a2) 478 { 479 return memcmp(a1, a2, sizeof(struct in6_addr)); 480 } 481 482 static inline bool 483 ipv6_masked_addr_cmp(const struct in6_addr *a1, const struct in6_addr *m, 484 const struct in6_addr *a2) 485 { 486 #if defined(CONFIG_HAVE_EFFICIENT_UNALIGNED_ACCESS) && BITS_PER_LONG == 64 487 const unsigned long *ul1 = (const unsigned long *)a1; 488 const unsigned long *ulm = (const unsigned long *)m; 489 const unsigned long *ul2 = (const unsigned long *)a2; 490 491 return !!(((ul1[0] ^ ul2[0]) & ulm[0]) | 492 ((ul1[1] ^ ul2[1]) & ulm[1])); 493 #else 494 return !!(((a1->s6_addr32[0] ^ a2->s6_addr32[0]) & m->s6_addr32[0]) | 495 ((a1->s6_addr32[1] ^ a2->s6_addr32[1]) & m->s6_addr32[1]) | 496 ((a1->s6_addr32[2] ^ a2->s6_addr32[2]) & m->s6_addr32[2]) | 497 ((a1->s6_addr32[3] ^ a2->s6_addr32[3]) & m->s6_addr32[3])); 498 #endif 499 } 500 501 static inline void ipv6_addr_prefix(struct in6_addr *pfx, 502 const struct in6_addr *addr, 503 int plen) 504 { 505 /* caller must guarantee 0 <= plen <= 128 */ 506 int o = plen >> 3, 507 b = plen & 0x7; 508 509 memset(pfx->s6_addr, 0, sizeof(pfx->s6_addr)); 510 memcpy(pfx->s6_addr, addr, o); 511 if (b != 0) 512 pfx->s6_addr[o] = addr->s6_addr[o] & (0xff00 >> b); 513 } 514 515 static inline void ipv6_addr_prefix_copy(struct in6_addr *addr, 516 const struct in6_addr *pfx, 517 int plen) 518 { 519 /* caller must guarantee 0 <= plen <= 128 */ 520 int o = plen >> 3, 521 b = plen & 0x7; 522 523 memcpy(addr->s6_addr, pfx, o); 524 if (b != 0) { 525 addr->s6_addr[o] &= ~(0xff00 >> b); 526 addr->s6_addr[o] |= (pfx->s6_addr[o] & (0xff00 >> b)); 527 } 528 } 529 530 static inline void __ipv6_addr_set_half(__be32 *addr, 531 __be32 wh, __be32 wl) 532 { 533 #if defined(CONFIG_HAVE_EFFICIENT_UNALIGNED_ACCESS) && BITS_PER_LONG == 64 534 #if defined(__BIG_ENDIAN) 535 if (__builtin_constant_p(wh) && __builtin_constant_p(wl)) { 536 *(__force u64 *)addr = ((__force u64)(wh) << 32 | (__force u64)(wl)); 537 return; 538 } 539 #elif defined(__LITTLE_ENDIAN) 540 if (__builtin_constant_p(wl) && __builtin_constant_p(wh)) { 541 *(__force u64 *)addr = ((__force u64)(wl) << 32 | (__force u64)(wh)); 542 return; 543 } 544 #endif 545 #endif 546 addr[0] = wh; 547 addr[1] = wl; 548 } 549 550 static inline void ipv6_addr_set(struct in6_addr *addr, 551 __be32 w1, __be32 w2, 552 __be32 w3, __be32 w4) 553 { 554 __ipv6_addr_set_half(&addr->s6_addr32[0], w1, w2); 555 __ipv6_addr_set_half(&addr->s6_addr32[2], w3, w4); 556 } 557 558 static inline bool ipv6_addr_equal(const struct in6_addr *a1, 559 const struct in6_addr *a2) 560 { 561 #if defined(CONFIG_HAVE_EFFICIENT_UNALIGNED_ACCESS) && BITS_PER_LONG == 64 562 const unsigned long *ul1 = (const unsigned long *)a1; 563 const unsigned long *ul2 = (const unsigned long *)a2; 564 565 return ((ul1[0] ^ ul2[0]) | (ul1[1] ^ ul2[1])) == 0UL; 566 #else 567 return ((a1->s6_addr32[0] ^ a2->s6_addr32[0]) | 568 (a1->s6_addr32[1] ^ a2->s6_addr32[1]) | 569 (a1->s6_addr32[2] ^ a2->s6_addr32[2]) | 570 (a1->s6_addr32[3] ^ a2->s6_addr32[3])) == 0; 571 #endif 572 } 573 574 #if defined(CONFIG_HAVE_EFFICIENT_UNALIGNED_ACCESS) && BITS_PER_LONG == 64 575 static inline bool __ipv6_prefix_equal64_half(const __be64 *a1, 576 const __be64 *a2, 577 unsigned int len) 578 { 579 if (len && ((*a1 ^ *a2) & cpu_to_be64((~0UL) << (64 - len)))) 580 return false; 581 return true; 582 } 583 584 static inline bool ipv6_prefix_equal(const struct in6_addr *addr1, 585 const struct in6_addr *addr2, 586 unsigned int prefixlen) 587 { 588 const __be64 *a1 = (const __be64 *)addr1; 589 const __be64 *a2 = (const __be64 *)addr2; 590 591 if (prefixlen >= 64) { 592 if (a1[0] ^ a2[0]) 593 return false; 594 return __ipv6_prefix_equal64_half(a1 + 1, a2 + 1, prefixlen - 64); 595 } 596 return __ipv6_prefix_equal64_half(a1, a2, prefixlen); 597 } 598 #else 599 static inline bool ipv6_prefix_equal(const struct in6_addr *addr1, 600 const struct in6_addr *addr2, 601 unsigned int prefixlen) 602 { 603 const __be32 *a1 = addr1->s6_addr32; 604 const __be32 *a2 = addr2->s6_addr32; 605 unsigned int pdw, pbi; 606 607 /* check complete u32 in prefix */ 608 pdw = prefixlen >> 5; 609 if (pdw && memcmp(a1, a2, pdw << 2)) 610 return false; 611 612 /* check incomplete u32 in prefix */ 613 pbi = prefixlen & 0x1f; 614 if (pbi && ((a1[pdw] ^ a2[pdw]) & htonl((0xffffffff) << (32 - pbi)))) 615 return false; 616 617 return true; 618 } 619 #endif 620 621 static inline bool ipv6_addr_any(const struct in6_addr *a) 622 { 623 #if defined(CONFIG_HAVE_EFFICIENT_UNALIGNED_ACCESS) && BITS_PER_LONG == 64 624 const unsigned long *ul = (const unsigned long *)a; 625 626 return (ul[0] | ul[1]) == 0UL; 627 #else 628 return (a->s6_addr32[0] | a->s6_addr32[1] | 629 a->s6_addr32[2] | a->s6_addr32[3]) == 0; 630 #endif 631 } 632 633 static inline u32 ipv6_addr_hash(const struct in6_addr *a) 634 { 635 #if defined(CONFIG_HAVE_EFFICIENT_UNALIGNED_ACCESS) && BITS_PER_LONG == 64 636 const unsigned long *ul = (const unsigned long *)a; 637 unsigned long x = ul[0] ^ ul[1]; 638 639 return (u32)(x ^ (x >> 32)); 640 #else 641 return (__force u32)(a->s6_addr32[0] ^ a->s6_addr32[1] ^ 642 a->s6_addr32[2] ^ a->s6_addr32[3]); 643 #endif 644 } 645 646 /* more secured version of ipv6_addr_hash() */ 647 static inline u32 __ipv6_addr_jhash(const struct in6_addr *a, const u32 initval) 648 { 649 u32 v = (__force u32)a->s6_addr32[0] ^ (__force u32)a->s6_addr32[1]; 650 651 return jhash_3words(v, 652 (__force u32)a->s6_addr32[2], 653 (__force u32)a->s6_addr32[3], 654 initval); 655 } 656 657 static inline bool ipv6_addr_loopback(const struct in6_addr *a) 658 { 659 #if defined(CONFIG_HAVE_EFFICIENT_UNALIGNED_ACCESS) && BITS_PER_LONG == 64 660 const __be64 *be = (const __be64 *)a; 661 662 return (be[0] | (be[1] ^ cpu_to_be64(1))) == 0UL; 663 #else 664 return (a->s6_addr32[0] | a->s6_addr32[1] | 665 a->s6_addr32[2] | (a->s6_addr32[3] ^ cpu_to_be32(1))) == 0; 666 #endif 667 } 668 669 /* 670 * Note that we must __force cast these to unsigned long to make sparse happy, 671 * since all of the endian-annotated types are fixed size regardless of arch. 672 */ 673 static inline bool ipv6_addr_v4mapped(const struct in6_addr *a) 674 { 675 return ( 676 #if defined(CONFIG_HAVE_EFFICIENT_UNALIGNED_ACCESS) && BITS_PER_LONG == 64 677 *(unsigned long *)a | 678 #else 679 (__force unsigned long)(a->s6_addr32[0] | a->s6_addr32[1]) | 680 #endif 681 (__force unsigned long)(a->s6_addr32[2] ^ 682 cpu_to_be32(0x0000ffff))) == 0UL; 683 } 684 685 static inline u32 ipv6_portaddr_hash(const struct net *net, 686 const struct in6_addr *addr6, 687 unsigned int port) 688 { 689 unsigned int hash, mix = net_hash_mix(net); 690 691 if (ipv6_addr_any(addr6)) 692 hash = jhash_1word(0, mix); 693 else if (ipv6_addr_v4mapped(addr6)) 694 hash = jhash_1word((__force u32)addr6->s6_addr32[3], mix); 695 else 696 hash = jhash2((__force u32 *)addr6->s6_addr32, 4, mix); 697 698 return hash ^ port; 699 } 700 701 /* 702 * Check for a RFC 4843 ORCHID address 703 * (Overlay Routable Cryptographic Hash Identifiers) 704 */ 705 static inline bool ipv6_addr_orchid(const struct in6_addr *a) 706 { 707 return (a->s6_addr32[0] & htonl(0xfffffff0)) == htonl(0x20010010); 708 } 709 710 static inline bool ipv6_addr_is_multicast(const struct in6_addr *addr) 711 { 712 return (addr->s6_addr32[0] & htonl(0xFF000000)) == htonl(0xFF000000); 713 } 714 715 static inline void ipv6_addr_set_v4mapped(const __be32 addr, 716 struct in6_addr *v4mapped) 717 { 718 ipv6_addr_set(v4mapped, 719 0, 0, 720 htonl(0x0000FFFF), 721 addr); 722 } 723 724 /* 725 * find the first different bit between two addresses 726 * length of address must be a multiple of 32bits 727 */ 728 static inline int __ipv6_addr_diff32(const void *token1, const void *token2, int addrlen) 729 { 730 const __be32 *a1 = token1, *a2 = token2; 731 int i; 732 733 addrlen >>= 2; 734 735 for (i = 0; i < addrlen; i++) { 736 __be32 xb = a1[i] ^ a2[i]; 737 if (xb) 738 return i * 32 + 31 - __fls(ntohl(xb)); 739 } 740 741 /* 742 * we should *never* get to this point since that 743 * would mean the addrs are equal 744 * 745 * However, we do get to it 8) And exacly, when 746 * addresses are equal 8) 747 * 748 * ip route add 1111::/128 via ... 749 * ip route add 1111::/64 via ... 750 * and we are here. 751 * 752 * Ideally, this function should stop comparison 753 * at prefix length. It does not, but it is still OK, 754 * if returned value is greater than prefix length. 755 * --ANK (980803) 756 */ 757 return addrlen << 5; 758 } 759 760 #if defined(CONFIG_HAVE_EFFICIENT_UNALIGNED_ACCESS) && BITS_PER_LONG == 64 761 static inline int __ipv6_addr_diff64(const void *token1, const void *token2, int addrlen) 762 { 763 const __be64 *a1 = token1, *a2 = token2; 764 int i; 765 766 addrlen >>= 3; 767 768 for (i = 0; i < addrlen; i++) { 769 __be64 xb = a1[i] ^ a2[i]; 770 if (xb) 771 return i * 64 + 63 - __fls(be64_to_cpu(xb)); 772 } 773 774 return addrlen << 6; 775 } 776 #endif 777 778 static inline int __ipv6_addr_diff(const void *token1, const void *token2, int addrlen) 779 { 780 #if defined(CONFIG_HAVE_EFFICIENT_UNALIGNED_ACCESS) && BITS_PER_LONG == 64 781 if (__builtin_constant_p(addrlen) && !(addrlen & 7)) 782 return __ipv6_addr_diff64(token1, token2, addrlen); 783 #endif 784 return __ipv6_addr_diff32(token1, token2, addrlen); 785 } 786 787 static inline int ipv6_addr_diff(const struct in6_addr *a1, const struct in6_addr *a2) 788 { 789 return __ipv6_addr_diff(a1, a2, sizeof(struct in6_addr)); 790 } 791 792 __be32 ipv6_select_ident(struct net *net, 793 const struct in6_addr *daddr, 794 const struct in6_addr *saddr); 795 __be32 ipv6_proxy_select_ident(struct net *net, struct sk_buff *skb); 796 797 int ip6_dst_hoplimit(struct dst_entry *dst); 798 799 static inline int ip6_sk_dst_hoplimit(struct ipv6_pinfo *np, struct flowi6 *fl6, 800 struct dst_entry *dst) 801 { 802 int hlimit; 803 804 if (ipv6_addr_is_multicast(&fl6->daddr)) 805 hlimit = np->mcast_hops; 806 else 807 hlimit = np->hop_limit; 808 if (hlimit < 0) 809 hlimit = ip6_dst_hoplimit(dst); 810 return hlimit; 811 } 812 813 /* copy IPv6 saddr & daddr to flow_keys, possibly using 64bit load/store 814 * Equivalent to : flow->v6addrs.src = iph->saddr; 815 * flow->v6addrs.dst = iph->daddr; 816 */ 817 static inline void iph_to_flow_copy_v6addrs(struct flow_keys *flow, 818 const struct ipv6hdr *iph) 819 { 820 BUILD_BUG_ON(offsetof(typeof(flow->addrs), v6addrs.dst) != 821 offsetof(typeof(flow->addrs), v6addrs.src) + 822 sizeof(flow->addrs.v6addrs.src)); 823 memcpy(&flow->addrs.v6addrs, &iph->saddr, sizeof(flow->addrs.v6addrs)); 824 flow->control.addr_type = FLOW_DISSECTOR_KEY_IPV6_ADDRS; 825 } 826 827 #if IS_ENABLED(CONFIG_IPV6) 828 829 static inline bool ipv6_can_nonlocal_bind(struct net *net, 830 struct inet_sock *inet) 831 { 832 return net->ipv6.sysctl.ip_nonlocal_bind || 833 inet->freebind || inet->transparent; 834 } 835 836 /* Sysctl settings for net ipv6.auto_flowlabels */ 837 #define IP6_AUTO_FLOW_LABEL_OFF 0 838 #define IP6_AUTO_FLOW_LABEL_OPTOUT 1 839 #define IP6_AUTO_FLOW_LABEL_OPTIN 2 840 #define IP6_AUTO_FLOW_LABEL_FORCED 3 841 842 #define IP6_AUTO_FLOW_LABEL_MAX IP6_AUTO_FLOW_LABEL_FORCED 843 844 #define IP6_DEFAULT_AUTO_FLOW_LABELS IP6_AUTO_FLOW_LABEL_OPTOUT 845 846 static inline __be32 ip6_make_flowlabel(struct net *net, struct sk_buff *skb, 847 __be32 flowlabel, bool autolabel, 848 struct flowi6 *fl6) 849 { 850 u32 hash; 851 852 /* @flowlabel may include more than a flow label, eg, the traffic class. 853 * Here we want only the flow label value. 854 */ 855 flowlabel &= IPV6_FLOWLABEL_MASK; 856 857 if (flowlabel || 858 net->ipv6.sysctl.auto_flowlabels == IP6_AUTO_FLOW_LABEL_OFF || 859 (!autolabel && 860 net->ipv6.sysctl.auto_flowlabels != IP6_AUTO_FLOW_LABEL_FORCED)) 861 return flowlabel; 862 863 hash = skb_get_hash_flowi6(skb, fl6); 864 865 /* Since this is being sent on the wire obfuscate hash a bit 866 * to minimize possbility that any useful information to an 867 * attacker is leaked. Only lower 20 bits are relevant. 868 */ 869 hash = rol32(hash, 16); 870 871 flowlabel = (__force __be32)hash & IPV6_FLOWLABEL_MASK; 872 873 if (net->ipv6.sysctl.flowlabel_state_ranges) 874 flowlabel |= IPV6_FLOWLABEL_STATELESS_FLAG; 875 876 return flowlabel; 877 } 878 879 static inline int ip6_default_np_autolabel(struct net *net) 880 { 881 switch (net->ipv6.sysctl.auto_flowlabels) { 882 case IP6_AUTO_FLOW_LABEL_OFF: 883 case IP6_AUTO_FLOW_LABEL_OPTIN: 884 default: 885 return 0; 886 case IP6_AUTO_FLOW_LABEL_OPTOUT: 887 case IP6_AUTO_FLOW_LABEL_FORCED: 888 return 1; 889 } 890 } 891 #else 892 static inline void ip6_set_txhash(struct sock *sk) { } 893 static inline __be32 ip6_make_flowlabel(struct net *net, struct sk_buff *skb, 894 __be32 flowlabel, bool autolabel, 895 struct flowi6 *fl6) 896 { 897 return flowlabel; 898 } 899 static inline int ip6_default_np_autolabel(struct net *net) 900 { 901 return 0; 902 } 903 #endif 904 905 #if IS_ENABLED(CONFIG_IPV6) 906 static inline int ip6_multipath_hash_policy(const struct net *net) 907 { 908 return net->ipv6.sysctl.multipath_hash_policy; 909 } 910 #else 911 static inline int ip6_multipath_hash_policy(const struct net *net) 912 { 913 return 0; 914 } 915 #endif 916 917 /* 918 * Header manipulation 919 */ 920 static inline void ip6_flow_hdr(struct ipv6hdr *hdr, unsigned int tclass, 921 __be32 flowlabel) 922 { 923 *(__be32 *)hdr = htonl(0x60000000 | (tclass << 20)) | flowlabel; 924 } 925 926 static inline __be32 ip6_flowinfo(const struct ipv6hdr *hdr) 927 { 928 return *(__be32 *)hdr & IPV6_FLOWINFO_MASK; 929 } 930 931 static inline __be32 ip6_flowlabel(const struct ipv6hdr *hdr) 932 { 933 return *(__be32 *)hdr & IPV6_FLOWLABEL_MASK; 934 } 935 936 static inline u8 ip6_tclass(__be32 flowinfo) 937 { 938 return ntohl(flowinfo & IPV6_TCLASS_MASK) >> IPV6_TCLASS_SHIFT; 939 } 940 941 static inline __be32 ip6_make_flowinfo(unsigned int tclass, __be32 flowlabel) 942 { 943 return htonl(tclass << IPV6_TCLASS_SHIFT) | flowlabel; 944 } 945 946 static inline __be32 flowi6_get_flowlabel(const struct flowi6 *fl6) 947 { 948 return fl6->flowlabel & IPV6_FLOWLABEL_MASK; 949 } 950 951 /* 952 * Prototypes exported by ipv6 953 */ 954 955 /* 956 * rcv function (called from netdevice level) 957 */ 958 959 int ipv6_rcv(struct sk_buff *skb, struct net_device *dev, 960 struct packet_type *pt, struct net_device *orig_dev); 961 void ipv6_list_rcv(struct list_head *head, struct packet_type *pt, 962 struct net_device *orig_dev); 963 964 int ip6_rcv_finish(struct net *net, struct sock *sk, struct sk_buff *skb); 965 966 /* 967 * upper-layer output functions 968 */ 969 int ip6_xmit(const struct sock *sk, struct sk_buff *skb, struct flowi6 *fl6, 970 __u32 mark, struct ipv6_txoptions *opt, int tclass); 971 972 int ip6_find_1stfragopt(struct sk_buff *skb, u8 **nexthdr); 973 974 int ip6_append_data(struct sock *sk, 975 int getfrag(void *from, char *to, int offset, int len, 976 int odd, struct sk_buff *skb), 977 void *from, int length, int transhdrlen, 978 struct ipcm6_cookie *ipc6, struct flowi6 *fl6, 979 struct rt6_info *rt, unsigned int flags); 980 981 int ip6_push_pending_frames(struct sock *sk); 982 983 void ip6_flush_pending_frames(struct sock *sk); 984 985 int ip6_send_skb(struct sk_buff *skb); 986 987 struct sk_buff *__ip6_make_skb(struct sock *sk, struct sk_buff_head *queue, 988 struct inet_cork_full *cork, 989 struct inet6_cork *v6_cork); 990 struct sk_buff *ip6_make_skb(struct sock *sk, 991 int getfrag(void *from, char *to, int offset, 992 int len, int odd, struct sk_buff *skb), 993 void *from, int length, int transhdrlen, 994 struct ipcm6_cookie *ipc6, struct flowi6 *fl6, 995 struct rt6_info *rt, unsigned int flags, 996 struct inet_cork_full *cork); 997 998 static inline struct sk_buff *ip6_finish_skb(struct sock *sk) 999 { 1000 return __ip6_make_skb(sk, &sk->sk_write_queue, &inet_sk(sk)->cork, 1001 &inet6_sk(sk)->cork); 1002 } 1003 1004 int ip6_dst_lookup(struct net *net, struct sock *sk, struct dst_entry **dst, 1005 struct flowi6 *fl6); 1006 struct dst_entry *ip6_dst_lookup_flow(const struct sock *sk, struct flowi6 *fl6, 1007 const struct in6_addr *final_dst); 1008 struct dst_entry *ip6_sk_dst_lookup_flow(struct sock *sk, struct flowi6 *fl6, 1009 const struct in6_addr *final_dst, 1010 bool connected); 1011 struct dst_entry *ip6_blackhole_route(struct net *net, 1012 struct dst_entry *orig_dst); 1013 1014 /* 1015 * skb processing functions 1016 */ 1017 1018 int ip6_output(struct net *net, struct sock *sk, struct sk_buff *skb); 1019 int ip6_forward(struct sk_buff *skb); 1020 int ip6_input(struct sk_buff *skb); 1021 int ip6_mc_input(struct sk_buff *skb); 1022 void ip6_protocol_deliver_rcu(struct net *net, struct sk_buff *skb, int nexthdr, 1023 bool have_final); 1024 1025 int __ip6_local_out(struct net *net, struct sock *sk, struct sk_buff *skb); 1026 int ip6_local_out(struct net *net, struct sock *sk, struct sk_buff *skb); 1027 1028 /* 1029 * Extension header (options) processing 1030 */ 1031 1032 void ipv6_push_nfrag_opts(struct sk_buff *skb, struct ipv6_txoptions *opt, 1033 u8 *proto, struct in6_addr **daddr_p, 1034 struct in6_addr *saddr); 1035 void ipv6_push_frag_opts(struct sk_buff *skb, struct ipv6_txoptions *opt, 1036 u8 *proto); 1037 1038 int ipv6_skip_exthdr(const struct sk_buff *, int start, u8 *nexthdrp, 1039 __be16 *frag_offp); 1040 1041 bool ipv6_ext_hdr(u8 nexthdr); 1042 1043 enum { 1044 IP6_FH_F_FRAG = (1 << 0), 1045 IP6_FH_F_AUTH = (1 << 1), 1046 IP6_FH_F_SKIP_RH = (1 << 2), 1047 }; 1048 1049 /* find specified header and get offset to it */ 1050 int ipv6_find_hdr(const struct sk_buff *skb, unsigned int *offset, int target, 1051 unsigned short *fragoff, int *fragflg); 1052 1053 int ipv6_find_tlv(const struct sk_buff *skb, int offset, int type); 1054 1055 struct in6_addr *fl6_update_dst(struct flowi6 *fl6, 1056 const struct ipv6_txoptions *opt, 1057 struct in6_addr *orig); 1058 1059 /* 1060 * socket options (ipv6_sockglue.c) 1061 */ 1062 1063 int ipv6_setsockopt(struct sock *sk, int level, int optname, 1064 char __user *optval, unsigned int optlen); 1065 int ipv6_getsockopt(struct sock *sk, int level, int optname, 1066 char __user *optval, int __user *optlen); 1067 int compat_ipv6_setsockopt(struct sock *sk, int level, int optname, 1068 char __user *optval, unsigned int optlen); 1069 int compat_ipv6_getsockopt(struct sock *sk, int level, int optname, 1070 char __user *optval, int __user *optlen); 1071 1072 int __ip6_datagram_connect(struct sock *sk, struct sockaddr *addr, 1073 int addr_len); 1074 int ip6_datagram_connect(struct sock *sk, struct sockaddr *addr, int addr_len); 1075 int ip6_datagram_connect_v6_only(struct sock *sk, struct sockaddr *addr, 1076 int addr_len); 1077 int ip6_datagram_dst_update(struct sock *sk, bool fix_sk_saddr); 1078 void ip6_datagram_release_cb(struct sock *sk); 1079 1080 int ipv6_recv_error(struct sock *sk, struct msghdr *msg, int len, 1081 int *addr_len); 1082 int ipv6_recv_rxpmtu(struct sock *sk, struct msghdr *msg, int len, 1083 int *addr_len); 1084 void ipv6_icmp_error(struct sock *sk, struct sk_buff *skb, int err, __be16 port, 1085 u32 info, u8 *payload); 1086 void ipv6_local_error(struct sock *sk, int err, struct flowi6 *fl6, u32 info); 1087 void ipv6_local_rxpmtu(struct sock *sk, struct flowi6 *fl6, u32 mtu); 1088 1089 int inet6_release(struct socket *sock); 1090 int inet6_bind(struct socket *sock, struct sockaddr *uaddr, int addr_len); 1091 int inet6_getname(struct socket *sock, struct sockaddr *uaddr, 1092 int peer); 1093 int inet6_ioctl(struct socket *sock, unsigned int cmd, unsigned long arg); 1094 1095 int inet6_hash_connect(struct inet_timewait_death_row *death_row, 1096 struct sock *sk); 1097 1098 /* 1099 * reassembly.c 1100 */ 1101 extern const struct proto_ops inet6_stream_ops; 1102 extern const struct proto_ops inet6_dgram_ops; 1103 extern const struct proto_ops inet6_sockraw_ops; 1104 1105 struct group_source_req; 1106 struct group_filter; 1107 1108 int ip6_mc_source(int add, int omode, struct sock *sk, 1109 struct group_source_req *pgsr); 1110 int ip6_mc_msfilter(struct sock *sk, struct group_filter *gsf); 1111 int ip6_mc_msfget(struct sock *sk, struct group_filter *gsf, 1112 struct group_filter __user *optval, int __user *optlen); 1113 1114 #ifdef CONFIG_PROC_FS 1115 int ac6_proc_init(struct net *net); 1116 void ac6_proc_exit(struct net *net); 1117 int raw6_proc_init(void); 1118 void raw6_proc_exit(void); 1119 int tcp6_proc_init(struct net *net); 1120 void tcp6_proc_exit(struct net *net); 1121 int udp6_proc_init(struct net *net); 1122 void udp6_proc_exit(struct net *net); 1123 int udplite6_proc_init(void); 1124 void udplite6_proc_exit(void); 1125 int ipv6_misc_proc_init(void); 1126 void ipv6_misc_proc_exit(void); 1127 int snmp6_register_dev(struct inet6_dev *idev); 1128 int snmp6_unregister_dev(struct inet6_dev *idev); 1129 1130 #else 1131 static inline int ac6_proc_init(struct net *net) { return 0; } 1132 static inline void ac6_proc_exit(struct net *net) { } 1133 static inline int snmp6_register_dev(struct inet6_dev *idev) { return 0; } 1134 static inline int snmp6_unregister_dev(struct inet6_dev *idev) { return 0; } 1135 #endif 1136 1137 #ifdef CONFIG_SYSCTL 1138 struct ctl_table *ipv6_icmp_sysctl_init(struct net *net); 1139 struct ctl_table *ipv6_route_sysctl_init(struct net *net); 1140 int ipv6_sysctl_register(void); 1141 void ipv6_sysctl_unregister(void); 1142 #endif 1143 1144 int ipv6_sock_mc_join(struct sock *sk, int ifindex, 1145 const struct in6_addr *addr); 1146 int ipv6_sock_mc_join_ssm(struct sock *sk, int ifindex, 1147 const struct in6_addr *addr, unsigned int mode); 1148 int ipv6_sock_mc_drop(struct sock *sk, int ifindex, 1149 const struct in6_addr *addr); 1150 #endif /* _NET_IPV6_H */ 1151