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