1 /* SPDX-License-Identifier: GPL-2.0-or-later */ 2 /* 3 * Linux INET6 implementation 4 * 5 * Authors: 6 * Pedro Roque <roque@di.fc.ul.pt> 7 */ 8 9 #ifndef _NET_IPV6_H 10 #define _NET_IPV6_H 11 12 #include <linux/ipv6.h> 13 #include <linux/hardirq.h> 14 #include <linux/jhash.h> 15 #include <linux/refcount.h> 16 #include <net/if_inet6.h> 17 #include <net/ndisc.h> 18 #include <net/flow.h> 19 #include <net/flow_dissector.h> 20 #include <net/snmp.h> 21 #include <net/netns/hash.h> 22 23 #define SIN6_LEN_RFC2133 24 24 25 #define IPV6_MAXPLEN 65535 26 27 /* 28 * NextHeader field of IPv6 header 29 */ 30 31 #define NEXTHDR_HOP 0 /* Hop-by-hop option header. */ 32 #define NEXTHDR_TCP 6 /* TCP segment. */ 33 #define NEXTHDR_UDP 17 /* UDP message. */ 34 #define NEXTHDR_IPV6 41 /* IPv6 in IPv6 */ 35 #define NEXTHDR_ROUTING 43 /* Routing header. */ 36 #define NEXTHDR_FRAGMENT 44 /* Fragmentation/reassembly header. */ 37 #define NEXTHDR_GRE 47 /* GRE header. */ 38 #define NEXTHDR_ESP 50 /* Encapsulating security payload. */ 39 #define NEXTHDR_AUTH 51 /* Authentication header. */ 40 #define NEXTHDR_ICMP 58 /* ICMP for IPv6. */ 41 #define NEXTHDR_NONE 59 /* No next header */ 42 #define NEXTHDR_DEST 60 /* Destination options header. */ 43 #define NEXTHDR_SCTP 132 /* SCTP message. */ 44 #define NEXTHDR_MOBILITY 135 /* Mobility header. */ 45 46 #define NEXTHDR_MAX 255 47 48 #define IPV6_DEFAULT_HOPLIMIT 64 49 #define IPV6_DEFAULT_MCASTHOPS 1 50 51 /* Limits on Hop-by-Hop and Destination options. 52 * 53 * Per RFC8200 there is no limit on the maximum number or lengths of options in 54 * Hop-by-Hop or Destination options other then the packet must fit in an MTU. 55 * We allow configurable limits in order to mitigate potential denial of 56 * service attacks. 57 * 58 * There are three limits that may be set: 59 * - Limit the number of options in a Hop-by-Hop or Destination options 60 * extension header 61 * - Limit the byte length of a Hop-by-Hop or Destination options extension 62 * header 63 * - Disallow unknown options 64 * 65 * The limits are expressed in corresponding sysctls: 66 * 67 * ipv6.sysctl.max_dst_opts_cnt 68 * ipv6.sysctl.max_hbh_opts_cnt 69 * ipv6.sysctl.max_dst_opts_len 70 * ipv6.sysctl.max_hbh_opts_len 71 * 72 * max_*_opts_cnt is the number of TLVs that are allowed for Destination 73 * options or Hop-by-Hop options. If the number is less than zero then unknown 74 * TLVs are disallowed and the number of known options that are allowed is the 75 * absolute value. Setting the value to INT_MAX indicates no limit. 76 * 77 * max_*_opts_len is the length limit in bytes of a Destination or 78 * Hop-by-Hop options extension header. Setting the value to INT_MAX 79 * indicates no length limit. 80 * 81 * If a limit is exceeded when processing an extension header the packet is 82 * silently discarded. 83 */ 84 85 /* Default limits for Hop-by-Hop and Destination options */ 86 #define IP6_DEFAULT_MAX_DST_OPTS_CNT 8 87 #define IP6_DEFAULT_MAX_HBH_OPTS_CNT 8 88 #define IP6_DEFAULT_MAX_DST_OPTS_LEN INT_MAX /* No limit */ 89 #define IP6_DEFAULT_MAX_HBH_OPTS_LEN INT_MAX /* No limit */ 90 91 /* 92 * Addr type 93 * 94 * type - unicast | multicast 95 * scope - local | site | global 96 * v4 - compat 97 * v4mapped 98 * any 99 * loopback 100 */ 101 102 #define IPV6_ADDR_ANY 0x0000U 103 104 #define IPV6_ADDR_UNICAST 0x0001U 105 #define IPV6_ADDR_MULTICAST 0x0002U 106 107 #define IPV6_ADDR_LOOPBACK 0x0010U 108 #define IPV6_ADDR_LINKLOCAL 0x0020U 109 #define IPV6_ADDR_SITELOCAL 0x0040U 110 111 #define IPV6_ADDR_COMPATv4 0x0080U 112 113 #define IPV6_ADDR_SCOPE_MASK 0x00f0U 114 115 #define IPV6_ADDR_MAPPED 0x1000U 116 117 /* 118 * Addr scopes 119 */ 120 #define IPV6_ADDR_MC_SCOPE(a) \ 121 ((a)->s6_addr[1] & 0x0f) /* nonstandard */ 122 #define __IPV6_ADDR_SCOPE_INVALID -1 123 #define IPV6_ADDR_SCOPE_NODELOCAL 0x01 124 #define IPV6_ADDR_SCOPE_LINKLOCAL 0x02 125 #define IPV6_ADDR_SCOPE_SITELOCAL 0x05 126 #define IPV6_ADDR_SCOPE_ORGLOCAL 0x08 127 #define IPV6_ADDR_SCOPE_GLOBAL 0x0e 128 129 /* 130 * Addr flags 131 */ 132 #define IPV6_ADDR_MC_FLAG_TRANSIENT(a) \ 133 ((a)->s6_addr[1] & 0x10) 134 #define IPV6_ADDR_MC_FLAG_PREFIX(a) \ 135 ((a)->s6_addr[1] & 0x20) 136 #define IPV6_ADDR_MC_FLAG_RENDEZVOUS(a) \ 137 ((a)->s6_addr[1] & 0x40) 138 139 /* 140 * fragmentation header 141 */ 142 143 struct frag_hdr { 144 __u8 nexthdr; 145 __u8 reserved; 146 __be16 frag_off; 147 __be32 identification; 148 }; 149 150 #define IP6_MF 0x0001 151 #define IP6_OFFSET 0xFFF8 152 153 struct ip6_fraglist_iter { 154 struct ipv6hdr *tmp_hdr; 155 struct sk_buff *frag; 156 int offset; 157 unsigned int hlen; 158 __be32 frag_id; 159 u8 nexthdr; 160 }; 161 162 int ip6_fraglist_init(struct sk_buff *skb, unsigned int hlen, u8 *prevhdr, 163 u8 nexthdr, __be32 frag_id, 164 struct ip6_fraglist_iter *iter); 165 void ip6_fraglist_prepare(struct sk_buff *skb, struct ip6_fraglist_iter *iter); 166 167 static inline struct sk_buff *ip6_fraglist_next(struct ip6_fraglist_iter *iter) 168 { 169 struct sk_buff *skb = iter->frag; 170 171 iter->frag = skb->next; 172 skb_mark_not_on_list(skb); 173 174 return skb; 175 } 176 177 struct ip6_frag_state { 178 u8 *prevhdr; 179 unsigned int hlen; 180 unsigned int mtu; 181 unsigned int left; 182 int offset; 183 int ptr; 184 int hroom; 185 int troom; 186 __be32 frag_id; 187 u8 nexthdr; 188 }; 189 190 void ip6_frag_init(struct sk_buff *skb, unsigned int hlen, unsigned int mtu, 191 unsigned short needed_tailroom, int hdr_room, u8 *prevhdr, 192 u8 nexthdr, __be32 frag_id, struct ip6_frag_state *state); 193 struct sk_buff *ip6_frag_next(struct sk_buff *skb, 194 struct ip6_frag_state *state); 195 196 #define IP6_REPLY_MARK(net, mark) \ 197 ((net)->ipv6.sysctl.fwmark_reflect ? (mark) : 0) 198 199 #include <net/sock.h> 200 201 /* sysctls */ 202 extern int sysctl_mld_max_msf; 203 extern int sysctl_mld_qrv; 204 205 #define _DEVINC(net, statname, mod, idev, field) \ 206 ({ \ 207 struct inet6_dev *_idev = (idev); \ 208 if (likely(_idev != NULL)) \ 209 mod##SNMP_INC_STATS64((_idev)->stats.statname, (field));\ 210 mod##SNMP_INC_STATS64((net)->mib.statname##_statistics, (field));\ 211 }) 212 213 /* per device counters are atomic_long_t */ 214 #define _DEVINCATOMIC(net, statname, mod, idev, field) \ 215 ({ \ 216 struct inet6_dev *_idev = (idev); \ 217 if (likely(_idev != NULL)) \ 218 SNMP_INC_STATS_ATOMIC_LONG((_idev)->stats.statname##dev, (field)); \ 219 mod##SNMP_INC_STATS((net)->mib.statname##_statistics, (field));\ 220 }) 221 222 /* per device and per net counters are atomic_long_t */ 223 #define _DEVINC_ATOMIC_ATOMIC(net, statname, idev, field) \ 224 ({ \ 225 struct inet6_dev *_idev = (idev); \ 226 if (likely(_idev != NULL)) \ 227 SNMP_INC_STATS_ATOMIC_LONG((_idev)->stats.statname##dev, (field)); \ 228 SNMP_INC_STATS_ATOMIC_LONG((net)->mib.statname##_statistics, (field));\ 229 }) 230 231 #define _DEVADD(net, statname, mod, idev, field, val) \ 232 ({ \ 233 struct inet6_dev *_idev = (idev); \ 234 if (likely(_idev != NULL)) \ 235 mod##SNMP_ADD_STATS((_idev)->stats.statname, (field), (val)); \ 236 mod##SNMP_ADD_STATS((net)->mib.statname##_statistics, (field), (val));\ 237 }) 238 239 #define _DEVUPD(net, statname, mod, idev, field, val) \ 240 ({ \ 241 struct inet6_dev *_idev = (idev); \ 242 if (likely(_idev != NULL)) \ 243 mod##SNMP_UPD_PO_STATS((_idev)->stats.statname, field, (val)); \ 244 mod##SNMP_UPD_PO_STATS((net)->mib.statname##_statistics, field, (val));\ 245 }) 246 247 /* MIBs */ 248 249 #define IP6_INC_STATS(net, idev,field) \ 250 _DEVINC(net, ipv6, , idev, field) 251 #define __IP6_INC_STATS(net, idev,field) \ 252 _DEVINC(net, ipv6, __, idev, field) 253 #define IP6_ADD_STATS(net, idev,field,val) \ 254 _DEVADD(net, ipv6, , idev, field, val) 255 #define __IP6_ADD_STATS(net, idev,field,val) \ 256 _DEVADD(net, ipv6, __, idev, field, val) 257 #define IP6_UPD_PO_STATS(net, idev,field,val) \ 258 _DEVUPD(net, ipv6, , idev, field, val) 259 #define __IP6_UPD_PO_STATS(net, idev,field,val) \ 260 _DEVUPD(net, ipv6, __, idev, field, val) 261 #define ICMP6_INC_STATS(net, idev, field) \ 262 _DEVINCATOMIC(net, icmpv6, , idev, field) 263 #define __ICMP6_INC_STATS(net, idev, field) \ 264 _DEVINCATOMIC(net, icmpv6, __, idev, field) 265 266 #define ICMP6MSGOUT_INC_STATS(net, idev, field) \ 267 _DEVINC_ATOMIC_ATOMIC(net, icmpv6msg, idev, field +256) 268 #define ICMP6MSGIN_INC_STATS(net, idev, field) \ 269 _DEVINC_ATOMIC_ATOMIC(net, icmpv6msg, idev, field) 270 271 struct ip6_ra_chain { 272 struct ip6_ra_chain *next; 273 struct sock *sk; 274 int sel; 275 void (*destructor)(struct sock *); 276 }; 277 278 extern struct ip6_ra_chain *ip6_ra_chain; 279 extern rwlock_t ip6_ra_lock; 280 281 /* 282 This structure is prepared by protocol, when parsing 283 ancillary data and passed to IPv6. 284 */ 285 286 struct ipv6_txoptions { 287 refcount_t refcnt; 288 /* Length of this structure */ 289 int tot_len; 290 291 /* length of extension headers */ 292 293 __u16 opt_flen; /* after fragment hdr */ 294 __u16 opt_nflen; /* before fragment hdr */ 295 296 struct ipv6_opt_hdr *hopopt; 297 struct ipv6_opt_hdr *dst0opt; 298 struct ipv6_rt_hdr *srcrt; /* Routing Header */ 299 struct ipv6_opt_hdr *dst1opt; 300 struct rcu_head rcu; 301 /* Option buffer, as read by IPV6_PKTOPTIONS, starts here. */ 302 }; 303 304 struct ip6_flowlabel { 305 struct ip6_flowlabel __rcu *next; 306 __be32 label; 307 atomic_t users; 308 struct in6_addr dst; 309 struct ipv6_txoptions *opt; 310 unsigned long linger; 311 struct rcu_head rcu; 312 u8 share; 313 union { 314 struct pid *pid; 315 kuid_t uid; 316 } owner; 317 unsigned long lastuse; 318 unsigned long expires; 319 struct net *fl_net; 320 }; 321 322 #define IPV6_FLOWINFO_MASK cpu_to_be32(0x0FFFFFFF) 323 #define IPV6_FLOWLABEL_MASK cpu_to_be32(0x000FFFFF) 324 #define IPV6_FLOWLABEL_STATELESS_FLAG cpu_to_be32(0x00080000) 325 326 #define IPV6_TCLASS_MASK (IPV6_FLOWINFO_MASK & ~IPV6_FLOWLABEL_MASK) 327 #define IPV6_TCLASS_SHIFT 20 328 329 struct ipv6_fl_socklist { 330 struct ipv6_fl_socklist __rcu *next; 331 struct ip6_flowlabel *fl; 332 struct rcu_head rcu; 333 }; 334 335 struct ipcm6_cookie { 336 struct sockcm_cookie sockc; 337 __s16 hlimit; 338 __s16 tclass; 339 __s8 dontfrag; 340 struct ipv6_txoptions *opt; 341 __u16 gso_size; 342 }; 343 344 static inline void ipcm6_init(struct ipcm6_cookie *ipc6) 345 { 346 *ipc6 = (struct ipcm6_cookie) { 347 .hlimit = -1, 348 .tclass = -1, 349 .dontfrag = -1, 350 }; 351 } 352 353 static inline void ipcm6_init_sk(struct ipcm6_cookie *ipc6, 354 const struct ipv6_pinfo *np) 355 { 356 *ipc6 = (struct ipcm6_cookie) { 357 .hlimit = -1, 358 .tclass = np->tclass, 359 .dontfrag = np->dontfrag, 360 }; 361 } 362 363 static inline struct ipv6_txoptions *txopt_get(const struct ipv6_pinfo *np) 364 { 365 struct ipv6_txoptions *opt; 366 367 rcu_read_lock(); 368 opt = rcu_dereference(np->opt); 369 if (opt) { 370 if (!refcount_inc_not_zero(&opt->refcnt)) 371 opt = NULL; 372 else 373 opt = rcu_pointer_handoff(opt); 374 } 375 rcu_read_unlock(); 376 return opt; 377 } 378 379 static inline void txopt_put(struct ipv6_txoptions *opt) 380 { 381 if (opt && refcount_dec_and_test(&opt->refcnt)) 382 kfree_rcu(opt, rcu); 383 } 384 385 struct ip6_flowlabel *fl6_sock_lookup(struct sock *sk, __be32 label); 386 struct ipv6_txoptions *fl6_merge_options(struct ipv6_txoptions *opt_space, 387 struct ip6_flowlabel *fl, 388 struct ipv6_txoptions *fopt); 389 void fl6_free_socklist(struct sock *sk); 390 int ipv6_flowlabel_opt(struct sock *sk, char __user *optval, int optlen); 391 int ipv6_flowlabel_opt_get(struct sock *sk, struct in6_flowlabel_req *freq, 392 int flags); 393 int ip6_flowlabel_init(void); 394 void ip6_flowlabel_cleanup(void); 395 bool ip6_autoflowlabel(struct net *net, const struct ipv6_pinfo *np); 396 397 static inline void fl6_sock_release(struct ip6_flowlabel *fl) 398 { 399 if (fl) 400 atomic_dec(&fl->users); 401 } 402 403 void icmpv6_notify(struct sk_buff *skb, u8 type, u8 code, __be32 info); 404 405 void icmpv6_push_pending_frames(struct sock *sk, struct flowi6 *fl6, 406 struct icmp6hdr *thdr, int len); 407 408 int ip6_ra_control(struct sock *sk, int sel); 409 410 int ipv6_parse_hopopts(struct sk_buff *skb); 411 412 struct ipv6_txoptions *ipv6_dup_options(struct sock *sk, 413 struct ipv6_txoptions *opt); 414 struct ipv6_txoptions *ipv6_renew_options(struct sock *sk, 415 struct ipv6_txoptions *opt, 416 int newtype, 417 struct ipv6_opt_hdr *newopt); 418 struct ipv6_txoptions *ipv6_fixup_options(struct ipv6_txoptions *opt_space, 419 struct ipv6_txoptions *opt); 420 421 bool ipv6_opt_accepted(const struct sock *sk, const struct sk_buff *skb, 422 const struct inet6_skb_parm *opt); 423 struct ipv6_txoptions *ipv6_update_options(struct sock *sk, 424 struct ipv6_txoptions *opt); 425 426 static inline bool ipv6_accept_ra(struct inet6_dev *idev) 427 { 428 /* If forwarding is enabled, RA are not accepted unless the special 429 * hybrid mode (accept_ra=2) is enabled. 430 */ 431 return idev->cnf.forwarding ? idev->cnf.accept_ra == 2 : 432 idev->cnf.accept_ra; 433 } 434 435 #define IPV6_FRAG_HIGH_THRESH (4 * 1024*1024) /* 4194304 */ 436 #define IPV6_FRAG_LOW_THRESH (3 * 1024*1024) /* 3145728 */ 437 #define IPV6_FRAG_TIMEOUT (60 * HZ) /* 60 seconds */ 438 439 int __ipv6_addr_type(const struct in6_addr *addr); 440 static inline int ipv6_addr_type(const struct in6_addr *addr) 441 { 442 return __ipv6_addr_type(addr) & 0xffff; 443 } 444 445 static inline int ipv6_addr_scope(const struct in6_addr *addr) 446 { 447 return __ipv6_addr_type(addr) & IPV6_ADDR_SCOPE_MASK; 448 } 449 450 static inline int __ipv6_addr_src_scope(int type) 451 { 452 return (type == IPV6_ADDR_ANY) ? __IPV6_ADDR_SCOPE_INVALID : (type >> 16); 453 } 454 455 static inline int ipv6_addr_src_scope(const struct in6_addr *addr) 456 { 457 return __ipv6_addr_src_scope(__ipv6_addr_type(addr)); 458 } 459 460 static inline bool __ipv6_addr_needs_scope_id(int type) 461 { 462 return type & IPV6_ADDR_LINKLOCAL || 463 (type & IPV6_ADDR_MULTICAST && 464 (type & (IPV6_ADDR_LOOPBACK|IPV6_ADDR_LINKLOCAL))); 465 } 466 467 static inline __u32 ipv6_iface_scope_id(const struct in6_addr *addr, int iface) 468 { 469 return __ipv6_addr_needs_scope_id(__ipv6_addr_type(addr)) ? iface : 0; 470 } 471 472 static inline int ipv6_addr_cmp(const struct in6_addr *a1, const struct in6_addr *a2) 473 { 474 return memcmp(a1, a2, sizeof(struct in6_addr)); 475 } 476 477 static inline bool 478 ipv6_masked_addr_cmp(const struct in6_addr *a1, const struct in6_addr *m, 479 const struct in6_addr *a2) 480 { 481 #if defined(CONFIG_HAVE_EFFICIENT_UNALIGNED_ACCESS) && BITS_PER_LONG == 64 482 const unsigned long *ul1 = (const unsigned long *)a1; 483 const unsigned long *ulm = (const unsigned long *)m; 484 const unsigned long *ul2 = (const unsigned long *)a2; 485 486 return !!(((ul1[0] ^ ul2[0]) & ulm[0]) | 487 ((ul1[1] ^ ul2[1]) & ulm[1])); 488 #else 489 return !!(((a1->s6_addr32[0] ^ a2->s6_addr32[0]) & m->s6_addr32[0]) | 490 ((a1->s6_addr32[1] ^ a2->s6_addr32[1]) & m->s6_addr32[1]) | 491 ((a1->s6_addr32[2] ^ a2->s6_addr32[2]) & m->s6_addr32[2]) | 492 ((a1->s6_addr32[3] ^ a2->s6_addr32[3]) & m->s6_addr32[3])); 493 #endif 494 } 495 496 static inline void ipv6_addr_prefix(struct in6_addr *pfx, 497 const struct in6_addr *addr, 498 int plen) 499 { 500 /* caller must guarantee 0 <= plen <= 128 */ 501 int o = plen >> 3, 502 b = plen & 0x7; 503 504 memset(pfx->s6_addr, 0, sizeof(pfx->s6_addr)); 505 memcpy(pfx->s6_addr, addr, o); 506 if (b != 0) 507 pfx->s6_addr[o] = addr->s6_addr[o] & (0xff00 >> b); 508 } 509 510 static inline void ipv6_addr_prefix_copy(struct in6_addr *addr, 511 const struct in6_addr *pfx, 512 int plen) 513 { 514 /* caller must guarantee 0 <= plen <= 128 */ 515 int o = plen >> 3, 516 b = plen & 0x7; 517 518 memcpy(addr->s6_addr, pfx, o); 519 if (b != 0) { 520 addr->s6_addr[o] &= ~(0xff00 >> b); 521 addr->s6_addr[o] |= (pfx->s6_addr[o] & (0xff00 >> b)); 522 } 523 } 524 525 static inline void __ipv6_addr_set_half(__be32 *addr, 526 __be32 wh, __be32 wl) 527 { 528 #if defined(CONFIG_HAVE_EFFICIENT_UNALIGNED_ACCESS) && BITS_PER_LONG == 64 529 #if defined(__BIG_ENDIAN) 530 if (__builtin_constant_p(wh) && __builtin_constant_p(wl)) { 531 *(__force u64 *)addr = ((__force u64)(wh) << 32 | (__force u64)(wl)); 532 return; 533 } 534 #elif defined(__LITTLE_ENDIAN) 535 if (__builtin_constant_p(wl) && __builtin_constant_p(wh)) { 536 *(__force u64 *)addr = ((__force u64)(wl) << 32 | (__force u64)(wh)); 537 return; 538 } 539 #endif 540 #endif 541 addr[0] = wh; 542 addr[1] = wl; 543 } 544 545 static inline void ipv6_addr_set(struct in6_addr *addr, 546 __be32 w1, __be32 w2, 547 __be32 w3, __be32 w4) 548 { 549 __ipv6_addr_set_half(&addr->s6_addr32[0], w1, w2); 550 __ipv6_addr_set_half(&addr->s6_addr32[2], w3, w4); 551 } 552 553 static inline bool ipv6_addr_equal(const struct in6_addr *a1, 554 const struct in6_addr *a2) 555 { 556 #if defined(CONFIG_HAVE_EFFICIENT_UNALIGNED_ACCESS) && BITS_PER_LONG == 64 557 const unsigned long *ul1 = (const unsigned long *)a1; 558 const unsigned long *ul2 = (const unsigned long *)a2; 559 560 return ((ul1[0] ^ ul2[0]) | (ul1[1] ^ ul2[1])) == 0UL; 561 #else 562 return ((a1->s6_addr32[0] ^ a2->s6_addr32[0]) | 563 (a1->s6_addr32[1] ^ a2->s6_addr32[1]) | 564 (a1->s6_addr32[2] ^ a2->s6_addr32[2]) | 565 (a1->s6_addr32[3] ^ a2->s6_addr32[3])) == 0; 566 #endif 567 } 568 569 #if defined(CONFIG_HAVE_EFFICIENT_UNALIGNED_ACCESS) && BITS_PER_LONG == 64 570 static inline bool __ipv6_prefix_equal64_half(const __be64 *a1, 571 const __be64 *a2, 572 unsigned int len) 573 { 574 if (len && ((*a1 ^ *a2) & cpu_to_be64((~0UL) << (64 - len)))) 575 return false; 576 return true; 577 } 578 579 static inline bool ipv6_prefix_equal(const struct in6_addr *addr1, 580 const struct in6_addr *addr2, 581 unsigned int prefixlen) 582 { 583 const __be64 *a1 = (const __be64 *)addr1; 584 const __be64 *a2 = (const __be64 *)addr2; 585 586 if (prefixlen >= 64) { 587 if (a1[0] ^ a2[0]) 588 return false; 589 return __ipv6_prefix_equal64_half(a1 + 1, a2 + 1, prefixlen - 64); 590 } 591 return __ipv6_prefix_equal64_half(a1, a2, prefixlen); 592 } 593 #else 594 static inline bool ipv6_prefix_equal(const struct in6_addr *addr1, 595 const struct in6_addr *addr2, 596 unsigned int prefixlen) 597 { 598 const __be32 *a1 = addr1->s6_addr32; 599 const __be32 *a2 = addr2->s6_addr32; 600 unsigned int pdw, pbi; 601 602 /* check complete u32 in prefix */ 603 pdw = prefixlen >> 5; 604 if (pdw && memcmp(a1, a2, pdw << 2)) 605 return false; 606 607 /* check incomplete u32 in prefix */ 608 pbi = prefixlen & 0x1f; 609 if (pbi && ((a1[pdw] ^ a2[pdw]) & htonl((0xffffffff) << (32 - pbi)))) 610 return false; 611 612 return true; 613 } 614 #endif 615 616 static inline bool ipv6_addr_any(const struct in6_addr *a) 617 { 618 #if defined(CONFIG_HAVE_EFFICIENT_UNALIGNED_ACCESS) && BITS_PER_LONG == 64 619 const unsigned long *ul = (const unsigned long *)a; 620 621 return (ul[0] | ul[1]) == 0UL; 622 #else 623 return (a->s6_addr32[0] | a->s6_addr32[1] | 624 a->s6_addr32[2] | a->s6_addr32[3]) == 0; 625 #endif 626 } 627 628 static inline u32 ipv6_addr_hash(const struct in6_addr *a) 629 { 630 #if defined(CONFIG_HAVE_EFFICIENT_UNALIGNED_ACCESS) && BITS_PER_LONG == 64 631 const unsigned long *ul = (const unsigned long *)a; 632 unsigned long x = ul[0] ^ ul[1]; 633 634 return (u32)(x ^ (x >> 32)); 635 #else 636 return (__force u32)(a->s6_addr32[0] ^ a->s6_addr32[1] ^ 637 a->s6_addr32[2] ^ a->s6_addr32[3]); 638 #endif 639 } 640 641 /* more secured version of ipv6_addr_hash() */ 642 static inline u32 __ipv6_addr_jhash(const struct in6_addr *a, const u32 initval) 643 { 644 u32 v = (__force u32)a->s6_addr32[0] ^ (__force u32)a->s6_addr32[1]; 645 646 return jhash_3words(v, 647 (__force u32)a->s6_addr32[2], 648 (__force u32)a->s6_addr32[3], 649 initval); 650 } 651 652 static inline bool ipv6_addr_loopback(const struct in6_addr *a) 653 { 654 #if defined(CONFIG_HAVE_EFFICIENT_UNALIGNED_ACCESS) && BITS_PER_LONG == 64 655 const __be64 *be = (const __be64 *)a; 656 657 return (be[0] | (be[1] ^ cpu_to_be64(1))) == 0UL; 658 #else 659 return (a->s6_addr32[0] | a->s6_addr32[1] | 660 a->s6_addr32[2] | (a->s6_addr32[3] ^ cpu_to_be32(1))) == 0; 661 #endif 662 } 663 664 /* 665 * Note that we must __force cast these to unsigned long to make sparse happy, 666 * since all of the endian-annotated types are fixed size regardless of arch. 667 */ 668 static inline bool ipv6_addr_v4mapped(const struct in6_addr *a) 669 { 670 return ( 671 #if defined(CONFIG_HAVE_EFFICIENT_UNALIGNED_ACCESS) && BITS_PER_LONG == 64 672 *(unsigned long *)a | 673 #else 674 (__force unsigned long)(a->s6_addr32[0] | a->s6_addr32[1]) | 675 #endif 676 (__force unsigned long)(a->s6_addr32[2] ^ 677 cpu_to_be32(0x0000ffff))) == 0UL; 678 } 679 680 static inline u32 ipv6_portaddr_hash(const struct net *net, 681 const struct in6_addr *addr6, 682 unsigned int port) 683 { 684 unsigned int hash, mix = net_hash_mix(net); 685 686 if (ipv6_addr_any(addr6)) 687 hash = jhash_1word(0, mix); 688 else if (ipv6_addr_v4mapped(addr6)) 689 hash = jhash_1word((__force u32)addr6->s6_addr32[3], mix); 690 else 691 hash = jhash2((__force u32 *)addr6->s6_addr32, 4, mix); 692 693 return hash ^ port; 694 } 695 696 /* 697 * Check for a RFC 4843 ORCHID address 698 * (Overlay Routable Cryptographic Hash Identifiers) 699 */ 700 static inline bool ipv6_addr_orchid(const struct in6_addr *a) 701 { 702 return (a->s6_addr32[0] & htonl(0xfffffff0)) == htonl(0x20010010); 703 } 704 705 static inline bool ipv6_addr_is_multicast(const struct in6_addr *addr) 706 { 707 return (addr->s6_addr32[0] & htonl(0xFF000000)) == htonl(0xFF000000); 708 } 709 710 static inline void ipv6_addr_set_v4mapped(const __be32 addr, 711 struct in6_addr *v4mapped) 712 { 713 ipv6_addr_set(v4mapped, 714 0, 0, 715 htonl(0x0000FFFF), 716 addr); 717 } 718 719 /* 720 * find the first different bit between two addresses 721 * length of address must be a multiple of 32bits 722 */ 723 static inline int __ipv6_addr_diff32(const void *token1, const void *token2, int addrlen) 724 { 725 const __be32 *a1 = token1, *a2 = token2; 726 int i; 727 728 addrlen >>= 2; 729 730 for (i = 0; i < addrlen; i++) { 731 __be32 xb = a1[i] ^ a2[i]; 732 if (xb) 733 return i * 32 + 31 - __fls(ntohl(xb)); 734 } 735 736 /* 737 * we should *never* get to this point since that 738 * would mean the addrs are equal 739 * 740 * However, we do get to it 8) And exacly, when 741 * addresses are equal 8) 742 * 743 * ip route add 1111::/128 via ... 744 * ip route add 1111::/64 via ... 745 * and we are here. 746 * 747 * Ideally, this function should stop comparison 748 * at prefix length. It does not, but it is still OK, 749 * if returned value is greater than prefix length. 750 * --ANK (980803) 751 */ 752 return addrlen << 5; 753 } 754 755 #if defined(CONFIG_HAVE_EFFICIENT_UNALIGNED_ACCESS) && BITS_PER_LONG == 64 756 static inline int __ipv6_addr_diff64(const void *token1, const void *token2, int addrlen) 757 { 758 const __be64 *a1 = token1, *a2 = token2; 759 int i; 760 761 addrlen >>= 3; 762 763 for (i = 0; i < addrlen; i++) { 764 __be64 xb = a1[i] ^ a2[i]; 765 if (xb) 766 return i * 64 + 63 - __fls(be64_to_cpu(xb)); 767 } 768 769 return addrlen << 6; 770 } 771 #endif 772 773 static inline int __ipv6_addr_diff(const void *token1, const void *token2, int addrlen) 774 { 775 #if defined(CONFIG_HAVE_EFFICIENT_UNALIGNED_ACCESS) && BITS_PER_LONG == 64 776 if (__builtin_constant_p(addrlen) && !(addrlen & 7)) 777 return __ipv6_addr_diff64(token1, token2, addrlen); 778 #endif 779 return __ipv6_addr_diff32(token1, token2, addrlen); 780 } 781 782 static inline int ipv6_addr_diff(const struct in6_addr *a1, const struct in6_addr *a2) 783 { 784 return __ipv6_addr_diff(a1, a2, sizeof(struct in6_addr)); 785 } 786 787 __be32 ipv6_select_ident(struct net *net, 788 const struct in6_addr *daddr, 789 const struct in6_addr *saddr); 790 __be32 ipv6_proxy_select_ident(struct net *net, struct sk_buff *skb); 791 792 int ip6_dst_hoplimit(struct dst_entry *dst); 793 794 static inline int ip6_sk_dst_hoplimit(struct ipv6_pinfo *np, struct flowi6 *fl6, 795 struct dst_entry *dst) 796 { 797 int hlimit; 798 799 if (ipv6_addr_is_multicast(&fl6->daddr)) 800 hlimit = np->mcast_hops; 801 else 802 hlimit = np->hop_limit; 803 if (hlimit < 0) 804 hlimit = ip6_dst_hoplimit(dst); 805 return hlimit; 806 } 807 808 /* copy IPv6 saddr & daddr to flow_keys, possibly using 64bit load/store 809 * Equivalent to : flow->v6addrs.src = iph->saddr; 810 * flow->v6addrs.dst = iph->daddr; 811 */ 812 static inline void iph_to_flow_copy_v6addrs(struct flow_keys *flow, 813 const struct ipv6hdr *iph) 814 { 815 BUILD_BUG_ON(offsetof(typeof(flow->addrs), v6addrs.dst) != 816 offsetof(typeof(flow->addrs), v6addrs.src) + 817 sizeof(flow->addrs.v6addrs.src)); 818 memcpy(&flow->addrs.v6addrs, &iph->saddr, sizeof(flow->addrs.v6addrs)); 819 flow->control.addr_type = FLOW_DISSECTOR_KEY_IPV6_ADDRS; 820 } 821 822 #if IS_ENABLED(CONFIG_IPV6) 823 824 static inline bool ipv6_can_nonlocal_bind(struct net *net, 825 struct inet_sock *inet) 826 { 827 return net->ipv6.sysctl.ip_nonlocal_bind || 828 inet->freebind || inet->transparent; 829 } 830 831 /* Sysctl settings for net ipv6.auto_flowlabels */ 832 #define IP6_AUTO_FLOW_LABEL_OFF 0 833 #define IP6_AUTO_FLOW_LABEL_OPTOUT 1 834 #define IP6_AUTO_FLOW_LABEL_OPTIN 2 835 #define IP6_AUTO_FLOW_LABEL_FORCED 3 836 837 #define IP6_AUTO_FLOW_LABEL_MAX IP6_AUTO_FLOW_LABEL_FORCED 838 839 #define IP6_DEFAULT_AUTO_FLOW_LABELS IP6_AUTO_FLOW_LABEL_OPTOUT 840 841 static inline __be32 ip6_make_flowlabel(struct net *net, struct sk_buff *skb, 842 __be32 flowlabel, bool autolabel, 843 struct flowi6 *fl6) 844 { 845 u32 hash; 846 847 /* @flowlabel may include more than a flow label, eg, the traffic class. 848 * Here we want only the flow label value. 849 */ 850 flowlabel &= IPV6_FLOWLABEL_MASK; 851 852 if (flowlabel || 853 net->ipv6.sysctl.auto_flowlabels == IP6_AUTO_FLOW_LABEL_OFF || 854 (!autolabel && 855 net->ipv6.sysctl.auto_flowlabels != IP6_AUTO_FLOW_LABEL_FORCED)) 856 return flowlabel; 857 858 hash = skb_get_hash_flowi6(skb, fl6); 859 860 /* Since this is being sent on the wire obfuscate hash a bit 861 * to minimize possbility that any useful information to an 862 * attacker is leaked. Only lower 20 bits are relevant. 863 */ 864 hash = rol32(hash, 16); 865 866 flowlabel = (__force __be32)hash & IPV6_FLOWLABEL_MASK; 867 868 if (net->ipv6.sysctl.flowlabel_state_ranges) 869 flowlabel |= IPV6_FLOWLABEL_STATELESS_FLAG; 870 871 return flowlabel; 872 } 873 874 static inline int ip6_default_np_autolabel(struct net *net) 875 { 876 switch (net->ipv6.sysctl.auto_flowlabels) { 877 case IP6_AUTO_FLOW_LABEL_OFF: 878 case IP6_AUTO_FLOW_LABEL_OPTIN: 879 default: 880 return 0; 881 case IP6_AUTO_FLOW_LABEL_OPTOUT: 882 case IP6_AUTO_FLOW_LABEL_FORCED: 883 return 1; 884 } 885 } 886 #else 887 static inline void ip6_set_txhash(struct sock *sk) { } 888 static inline __be32 ip6_make_flowlabel(struct net *net, struct sk_buff *skb, 889 __be32 flowlabel, bool autolabel, 890 struct flowi6 *fl6) 891 { 892 return flowlabel; 893 } 894 static inline int ip6_default_np_autolabel(struct net *net) 895 { 896 return 0; 897 } 898 #endif 899 900 #if IS_ENABLED(CONFIG_IPV6) 901 static inline int ip6_multipath_hash_policy(const struct net *net) 902 { 903 return net->ipv6.sysctl.multipath_hash_policy; 904 } 905 #else 906 static inline int ip6_multipath_hash_policy(const struct net *net) 907 { 908 return 0; 909 } 910 #endif 911 912 /* 913 * Header manipulation 914 */ 915 static inline void ip6_flow_hdr(struct ipv6hdr *hdr, unsigned int tclass, 916 __be32 flowlabel) 917 { 918 *(__be32 *)hdr = htonl(0x60000000 | (tclass << 20)) | flowlabel; 919 } 920 921 static inline __be32 ip6_flowinfo(const struct ipv6hdr *hdr) 922 { 923 return *(__be32 *)hdr & IPV6_FLOWINFO_MASK; 924 } 925 926 static inline __be32 ip6_flowlabel(const struct ipv6hdr *hdr) 927 { 928 return *(__be32 *)hdr & IPV6_FLOWLABEL_MASK; 929 } 930 931 static inline u8 ip6_tclass(__be32 flowinfo) 932 { 933 return ntohl(flowinfo & IPV6_TCLASS_MASK) >> IPV6_TCLASS_SHIFT; 934 } 935 936 static inline __be32 ip6_make_flowinfo(unsigned int tclass, __be32 flowlabel) 937 { 938 return htonl(tclass << IPV6_TCLASS_SHIFT) | flowlabel; 939 } 940 941 static inline __be32 flowi6_get_flowlabel(const struct flowi6 *fl6) 942 { 943 return fl6->flowlabel & IPV6_FLOWLABEL_MASK; 944 } 945 946 /* 947 * Prototypes exported by ipv6 948 */ 949 950 /* 951 * rcv function (called from netdevice level) 952 */ 953 954 int ipv6_rcv(struct sk_buff *skb, struct net_device *dev, 955 struct packet_type *pt, struct net_device *orig_dev); 956 void ipv6_list_rcv(struct list_head *head, struct packet_type *pt, 957 struct net_device *orig_dev); 958 959 int ip6_rcv_finish(struct net *net, struct sock *sk, struct sk_buff *skb); 960 961 /* 962 * upper-layer output functions 963 */ 964 int ip6_xmit(const struct sock *sk, struct sk_buff *skb, struct flowi6 *fl6, 965 __u32 mark, struct ipv6_txoptions *opt, int tclass); 966 967 int ip6_find_1stfragopt(struct sk_buff *skb, u8 **nexthdr); 968 969 int ip6_append_data(struct sock *sk, 970 int getfrag(void *from, char *to, int offset, int len, 971 int odd, struct sk_buff *skb), 972 void *from, int length, int transhdrlen, 973 struct ipcm6_cookie *ipc6, struct flowi6 *fl6, 974 struct rt6_info *rt, unsigned int flags); 975 976 int ip6_push_pending_frames(struct sock *sk); 977 978 void ip6_flush_pending_frames(struct sock *sk); 979 980 int ip6_send_skb(struct sk_buff *skb); 981 982 struct sk_buff *__ip6_make_skb(struct sock *sk, struct sk_buff_head *queue, 983 struct inet_cork_full *cork, 984 struct inet6_cork *v6_cork); 985 struct sk_buff *ip6_make_skb(struct sock *sk, 986 int getfrag(void *from, char *to, int offset, 987 int len, int odd, struct sk_buff *skb), 988 void *from, int length, int transhdrlen, 989 struct ipcm6_cookie *ipc6, struct flowi6 *fl6, 990 struct rt6_info *rt, unsigned int flags, 991 struct inet_cork_full *cork); 992 993 static inline struct sk_buff *ip6_finish_skb(struct sock *sk) 994 { 995 return __ip6_make_skb(sk, &sk->sk_write_queue, &inet_sk(sk)->cork, 996 &inet6_sk(sk)->cork); 997 } 998 999 int ip6_dst_lookup(struct net *net, struct sock *sk, struct dst_entry **dst, 1000 struct flowi6 *fl6); 1001 struct dst_entry *ip6_dst_lookup_flow(const struct sock *sk, struct flowi6 *fl6, 1002 const struct in6_addr *final_dst); 1003 struct dst_entry *ip6_sk_dst_lookup_flow(struct sock *sk, struct flowi6 *fl6, 1004 const struct in6_addr *final_dst, 1005 bool connected); 1006 struct dst_entry *ip6_blackhole_route(struct net *net, 1007 struct dst_entry *orig_dst); 1008 1009 /* 1010 * skb processing functions 1011 */ 1012 1013 int ip6_output(struct net *net, struct sock *sk, struct sk_buff *skb); 1014 int ip6_forward(struct sk_buff *skb); 1015 int ip6_input(struct sk_buff *skb); 1016 int ip6_mc_input(struct sk_buff *skb); 1017 void ip6_protocol_deliver_rcu(struct net *net, struct sk_buff *skb, int nexthdr, 1018 bool have_final); 1019 1020 int __ip6_local_out(struct net *net, struct sock *sk, struct sk_buff *skb); 1021 int ip6_local_out(struct net *net, struct sock *sk, struct sk_buff *skb); 1022 1023 /* 1024 * Extension header (options) processing 1025 */ 1026 1027 void ipv6_push_nfrag_opts(struct sk_buff *skb, struct ipv6_txoptions *opt, 1028 u8 *proto, struct in6_addr **daddr_p, 1029 struct in6_addr *saddr); 1030 void ipv6_push_frag_opts(struct sk_buff *skb, struct ipv6_txoptions *opt, 1031 u8 *proto); 1032 1033 int ipv6_skip_exthdr(const struct sk_buff *, int start, u8 *nexthdrp, 1034 __be16 *frag_offp); 1035 1036 bool ipv6_ext_hdr(u8 nexthdr); 1037 1038 enum { 1039 IP6_FH_F_FRAG = (1 << 0), 1040 IP6_FH_F_AUTH = (1 << 1), 1041 IP6_FH_F_SKIP_RH = (1 << 2), 1042 }; 1043 1044 /* find specified header and get offset to it */ 1045 int ipv6_find_hdr(const struct sk_buff *skb, unsigned int *offset, int target, 1046 unsigned short *fragoff, int *fragflg); 1047 1048 int ipv6_find_tlv(const struct sk_buff *skb, int offset, int type); 1049 1050 struct in6_addr *fl6_update_dst(struct flowi6 *fl6, 1051 const struct ipv6_txoptions *opt, 1052 struct in6_addr *orig); 1053 1054 /* 1055 * socket options (ipv6_sockglue.c) 1056 */ 1057 1058 int ipv6_setsockopt(struct sock *sk, int level, int optname, 1059 char __user *optval, unsigned int optlen); 1060 int ipv6_getsockopt(struct sock *sk, int level, int optname, 1061 char __user *optval, int __user *optlen); 1062 int compat_ipv6_setsockopt(struct sock *sk, int level, int optname, 1063 char __user *optval, unsigned int optlen); 1064 int compat_ipv6_getsockopt(struct sock *sk, int level, int optname, 1065 char __user *optval, int __user *optlen); 1066 1067 int __ip6_datagram_connect(struct sock *sk, struct sockaddr *addr, 1068 int addr_len); 1069 int ip6_datagram_connect(struct sock *sk, struct sockaddr *addr, int addr_len); 1070 int ip6_datagram_connect_v6_only(struct sock *sk, struct sockaddr *addr, 1071 int addr_len); 1072 int ip6_datagram_dst_update(struct sock *sk, bool fix_sk_saddr); 1073 void ip6_datagram_release_cb(struct sock *sk); 1074 1075 int ipv6_recv_error(struct sock *sk, struct msghdr *msg, int len, 1076 int *addr_len); 1077 int ipv6_recv_rxpmtu(struct sock *sk, struct msghdr *msg, int len, 1078 int *addr_len); 1079 void ipv6_icmp_error(struct sock *sk, struct sk_buff *skb, int err, __be16 port, 1080 u32 info, u8 *payload); 1081 void ipv6_local_error(struct sock *sk, int err, struct flowi6 *fl6, u32 info); 1082 void ipv6_local_rxpmtu(struct sock *sk, struct flowi6 *fl6, u32 mtu); 1083 1084 int inet6_release(struct socket *sock); 1085 int inet6_bind(struct socket *sock, struct sockaddr *uaddr, int addr_len); 1086 int inet6_getname(struct socket *sock, struct sockaddr *uaddr, 1087 int peer); 1088 int inet6_ioctl(struct socket *sock, unsigned int cmd, unsigned long arg); 1089 1090 int inet6_hash_connect(struct inet_timewait_death_row *death_row, 1091 struct sock *sk); 1092 1093 /* 1094 * reassembly.c 1095 */ 1096 extern const struct proto_ops inet6_stream_ops; 1097 extern const struct proto_ops inet6_dgram_ops; 1098 extern const struct proto_ops inet6_sockraw_ops; 1099 1100 struct group_source_req; 1101 struct group_filter; 1102 1103 int ip6_mc_source(int add, int omode, struct sock *sk, 1104 struct group_source_req *pgsr); 1105 int ip6_mc_msfilter(struct sock *sk, struct group_filter *gsf); 1106 int ip6_mc_msfget(struct sock *sk, struct group_filter *gsf, 1107 struct group_filter __user *optval, int __user *optlen); 1108 1109 #ifdef CONFIG_PROC_FS 1110 int ac6_proc_init(struct net *net); 1111 void ac6_proc_exit(struct net *net); 1112 int raw6_proc_init(void); 1113 void raw6_proc_exit(void); 1114 int tcp6_proc_init(struct net *net); 1115 void tcp6_proc_exit(struct net *net); 1116 int udp6_proc_init(struct net *net); 1117 void udp6_proc_exit(struct net *net); 1118 int udplite6_proc_init(void); 1119 void udplite6_proc_exit(void); 1120 int ipv6_misc_proc_init(void); 1121 void ipv6_misc_proc_exit(void); 1122 int snmp6_register_dev(struct inet6_dev *idev); 1123 int snmp6_unregister_dev(struct inet6_dev *idev); 1124 1125 #else 1126 static inline int ac6_proc_init(struct net *net) { return 0; } 1127 static inline void ac6_proc_exit(struct net *net) { } 1128 static inline int snmp6_register_dev(struct inet6_dev *idev) { return 0; } 1129 static inline int snmp6_unregister_dev(struct inet6_dev *idev) { return 0; } 1130 #endif 1131 1132 #ifdef CONFIG_SYSCTL 1133 struct ctl_table *ipv6_icmp_sysctl_init(struct net *net); 1134 struct ctl_table *ipv6_route_sysctl_init(struct net *net); 1135 int ipv6_sysctl_register(void); 1136 void ipv6_sysctl_unregister(void); 1137 #endif 1138 1139 int ipv6_sock_mc_join(struct sock *sk, int ifindex, 1140 const struct in6_addr *addr); 1141 int ipv6_sock_mc_join_ssm(struct sock *sk, int ifindex, 1142 const struct in6_addr *addr, unsigned int mode); 1143 int ipv6_sock_mc_drop(struct sock *sk, int ifindex, 1144 const struct in6_addr *addr); 1145 #endif /* _NET_IPV6_H */ 1146