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 #define IP6_REPLY_MARK(net, mark) \ 158 ((net)->ipv6.sysctl.fwmark_reflect ? (mark) : 0) 159 160 #include <net/sock.h> 161 162 /* sysctls */ 163 extern int sysctl_mld_max_msf; 164 extern int sysctl_mld_qrv; 165 166 #define _DEVINC(net, statname, mod, idev, field) \ 167 ({ \ 168 struct inet6_dev *_idev = (idev); \ 169 if (likely(_idev != NULL)) \ 170 mod##SNMP_INC_STATS64((_idev)->stats.statname, (field));\ 171 mod##SNMP_INC_STATS64((net)->mib.statname##_statistics, (field));\ 172 }) 173 174 /* per device counters are atomic_long_t */ 175 #define _DEVINCATOMIC(net, statname, mod, idev, field) \ 176 ({ \ 177 struct inet6_dev *_idev = (idev); \ 178 if (likely(_idev != NULL)) \ 179 SNMP_INC_STATS_ATOMIC_LONG((_idev)->stats.statname##dev, (field)); \ 180 mod##SNMP_INC_STATS((net)->mib.statname##_statistics, (field));\ 181 }) 182 183 /* per device and per net counters are atomic_long_t */ 184 #define _DEVINC_ATOMIC_ATOMIC(net, statname, idev, field) \ 185 ({ \ 186 struct inet6_dev *_idev = (idev); \ 187 if (likely(_idev != NULL)) \ 188 SNMP_INC_STATS_ATOMIC_LONG((_idev)->stats.statname##dev, (field)); \ 189 SNMP_INC_STATS_ATOMIC_LONG((net)->mib.statname##_statistics, (field));\ 190 }) 191 192 #define _DEVADD(net, statname, mod, idev, field, val) \ 193 ({ \ 194 struct inet6_dev *_idev = (idev); \ 195 if (likely(_idev != NULL)) \ 196 mod##SNMP_ADD_STATS((_idev)->stats.statname, (field), (val)); \ 197 mod##SNMP_ADD_STATS((net)->mib.statname##_statistics, (field), (val));\ 198 }) 199 200 #define _DEVUPD(net, statname, mod, idev, field, val) \ 201 ({ \ 202 struct inet6_dev *_idev = (idev); \ 203 if (likely(_idev != NULL)) \ 204 mod##SNMP_UPD_PO_STATS((_idev)->stats.statname, field, (val)); \ 205 mod##SNMP_UPD_PO_STATS((net)->mib.statname##_statistics, field, (val));\ 206 }) 207 208 /* MIBs */ 209 210 #define IP6_INC_STATS(net, idev,field) \ 211 _DEVINC(net, ipv6, , idev, field) 212 #define __IP6_INC_STATS(net, idev,field) \ 213 _DEVINC(net, ipv6, __, idev, field) 214 #define IP6_ADD_STATS(net, idev,field,val) \ 215 _DEVADD(net, ipv6, , idev, field, val) 216 #define __IP6_ADD_STATS(net, idev,field,val) \ 217 _DEVADD(net, ipv6, __, idev, field, val) 218 #define IP6_UPD_PO_STATS(net, idev,field,val) \ 219 _DEVUPD(net, ipv6, , idev, field, val) 220 #define __IP6_UPD_PO_STATS(net, idev,field,val) \ 221 _DEVUPD(net, ipv6, __, idev, field, val) 222 #define ICMP6_INC_STATS(net, idev, field) \ 223 _DEVINCATOMIC(net, icmpv6, , idev, field) 224 #define __ICMP6_INC_STATS(net, idev, field) \ 225 _DEVINCATOMIC(net, icmpv6, __, idev, field) 226 227 #define ICMP6MSGOUT_INC_STATS(net, idev, field) \ 228 _DEVINC_ATOMIC_ATOMIC(net, icmpv6msg, idev, field +256) 229 #define ICMP6MSGIN_INC_STATS(net, idev, field) \ 230 _DEVINC_ATOMIC_ATOMIC(net, icmpv6msg, idev, field) 231 232 struct ip6_ra_chain { 233 struct ip6_ra_chain *next; 234 struct sock *sk; 235 int sel; 236 void (*destructor)(struct sock *); 237 }; 238 239 extern struct ip6_ra_chain *ip6_ra_chain; 240 extern rwlock_t ip6_ra_lock; 241 242 /* 243 This structure is prepared by protocol, when parsing 244 ancillary data and passed to IPv6. 245 */ 246 247 struct ipv6_txoptions { 248 refcount_t refcnt; 249 /* Length of this structure */ 250 int tot_len; 251 252 /* length of extension headers */ 253 254 __u16 opt_flen; /* after fragment hdr */ 255 __u16 opt_nflen; /* before fragment hdr */ 256 257 struct ipv6_opt_hdr *hopopt; 258 struct ipv6_opt_hdr *dst0opt; 259 struct ipv6_rt_hdr *srcrt; /* Routing Header */ 260 struct ipv6_opt_hdr *dst1opt; 261 struct rcu_head rcu; 262 /* Option buffer, as read by IPV6_PKTOPTIONS, starts here. */ 263 }; 264 265 struct ip6_flowlabel { 266 struct ip6_flowlabel __rcu *next; 267 __be32 label; 268 atomic_t users; 269 struct in6_addr dst; 270 struct ipv6_txoptions *opt; 271 unsigned long linger; 272 struct rcu_head rcu; 273 u8 share; 274 union { 275 struct pid *pid; 276 kuid_t uid; 277 } owner; 278 unsigned long lastuse; 279 unsigned long expires; 280 struct net *fl_net; 281 }; 282 283 #define IPV6_FLOWINFO_MASK cpu_to_be32(0x0FFFFFFF) 284 #define IPV6_FLOWLABEL_MASK cpu_to_be32(0x000FFFFF) 285 #define IPV6_FLOWLABEL_STATELESS_FLAG cpu_to_be32(0x00080000) 286 287 #define IPV6_TCLASS_MASK (IPV6_FLOWINFO_MASK & ~IPV6_FLOWLABEL_MASK) 288 #define IPV6_TCLASS_SHIFT 20 289 290 struct ipv6_fl_socklist { 291 struct ipv6_fl_socklist __rcu *next; 292 struct ip6_flowlabel *fl; 293 struct rcu_head rcu; 294 }; 295 296 struct ipcm6_cookie { 297 __s16 hlimit; 298 __s16 tclass; 299 __s8 dontfrag; 300 struct ipv6_txoptions *opt; 301 }; 302 303 static inline struct ipv6_txoptions *txopt_get(const struct ipv6_pinfo *np) 304 { 305 struct ipv6_txoptions *opt; 306 307 rcu_read_lock(); 308 opt = rcu_dereference(np->opt); 309 if (opt) { 310 if (!refcount_inc_not_zero(&opt->refcnt)) 311 opt = NULL; 312 else 313 opt = rcu_pointer_handoff(opt); 314 } 315 rcu_read_unlock(); 316 return opt; 317 } 318 319 static inline void txopt_put(struct ipv6_txoptions *opt) 320 { 321 if (opt && refcount_dec_and_test(&opt->refcnt)) 322 kfree_rcu(opt, rcu); 323 } 324 325 struct ip6_flowlabel *fl6_sock_lookup(struct sock *sk, __be32 label); 326 struct ipv6_txoptions *fl6_merge_options(struct ipv6_txoptions *opt_space, 327 struct ip6_flowlabel *fl, 328 struct ipv6_txoptions *fopt); 329 void fl6_free_socklist(struct sock *sk); 330 int ipv6_flowlabel_opt(struct sock *sk, char __user *optval, int optlen); 331 int ipv6_flowlabel_opt_get(struct sock *sk, struct in6_flowlabel_req *freq, 332 int flags); 333 int ip6_flowlabel_init(void); 334 void ip6_flowlabel_cleanup(void); 335 bool ip6_autoflowlabel(struct net *net, const struct ipv6_pinfo *np); 336 337 static inline void fl6_sock_release(struct ip6_flowlabel *fl) 338 { 339 if (fl) 340 atomic_dec(&fl->users); 341 } 342 343 void icmpv6_notify(struct sk_buff *skb, u8 type, u8 code, __be32 info); 344 345 void icmpv6_push_pending_frames(struct sock *sk, struct flowi6 *fl6, 346 struct icmp6hdr *thdr, int len); 347 348 int ip6_ra_control(struct sock *sk, int sel); 349 350 int ipv6_parse_hopopts(struct sk_buff *skb); 351 352 struct ipv6_txoptions *ipv6_dup_options(struct sock *sk, 353 struct ipv6_txoptions *opt); 354 struct ipv6_txoptions *ipv6_renew_options(struct sock *sk, 355 struct ipv6_txoptions *opt, 356 int newtype, 357 struct ipv6_opt_hdr __user *newopt, 358 int newoptlen); 359 struct ipv6_txoptions * 360 ipv6_renew_options_kern(struct sock *sk, 361 struct ipv6_txoptions *opt, 362 int newtype, 363 struct ipv6_opt_hdr *newopt, 364 int newoptlen); 365 struct ipv6_txoptions *ipv6_fixup_options(struct ipv6_txoptions *opt_space, 366 struct ipv6_txoptions *opt); 367 368 bool ipv6_opt_accepted(const struct sock *sk, const struct sk_buff *skb, 369 const struct inet6_skb_parm *opt); 370 struct ipv6_txoptions *ipv6_update_options(struct sock *sk, 371 struct ipv6_txoptions *opt); 372 373 static inline bool ipv6_accept_ra(struct inet6_dev *idev) 374 { 375 /* If forwarding is enabled, RA are not accepted unless the special 376 * hybrid mode (accept_ra=2) is enabled. 377 */ 378 return idev->cnf.forwarding ? idev->cnf.accept_ra == 2 : 379 idev->cnf.accept_ra; 380 } 381 382 #if IS_ENABLED(CONFIG_IPV6) 383 static inline int ip6_frag_mem(struct net *net) 384 { 385 return sum_frag_mem_limit(&net->ipv6.frags); 386 } 387 #endif 388 389 #define IPV6_FRAG_HIGH_THRESH (4 * 1024*1024) /* 4194304 */ 390 #define IPV6_FRAG_LOW_THRESH (3 * 1024*1024) /* 3145728 */ 391 #define IPV6_FRAG_TIMEOUT (60 * HZ) /* 60 seconds */ 392 393 int __ipv6_addr_type(const struct in6_addr *addr); 394 static inline int ipv6_addr_type(const struct in6_addr *addr) 395 { 396 return __ipv6_addr_type(addr) & 0xffff; 397 } 398 399 static inline int ipv6_addr_scope(const struct in6_addr *addr) 400 { 401 return __ipv6_addr_type(addr) & IPV6_ADDR_SCOPE_MASK; 402 } 403 404 static inline int __ipv6_addr_src_scope(int type) 405 { 406 return (type == IPV6_ADDR_ANY) ? __IPV6_ADDR_SCOPE_INVALID : (type >> 16); 407 } 408 409 static inline int ipv6_addr_src_scope(const struct in6_addr *addr) 410 { 411 return __ipv6_addr_src_scope(__ipv6_addr_type(addr)); 412 } 413 414 static inline bool __ipv6_addr_needs_scope_id(int type) 415 { 416 return type & IPV6_ADDR_LINKLOCAL || 417 (type & IPV6_ADDR_MULTICAST && 418 (type & (IPV6_ADDR_LOOPBACK|IPV6_ADDR_LINKLOCAL))); 419 } 420 421 static inline __u32 ipv6_iface_scope_id(const struct in6_addr *addr, int iface) 422 { 423 return __ipv6_addr_needs_scope_id(__ipv6_addr_type(addr)) ? iface : 0; 424 } 425 426 static inline int ipv6_addr_cmp(const struct in6_addr *a1, const struct in6_addr *a2) 427 { 428 return memcmp(a1, a2, sizeof(struct in6_addr)); 429 } 430 431 static inline bool 432 ipv6_masked_addr_cmp(const struct in6_addr *a1, const struct in6_addr *m, 433 const struct in6_addr *a2) 434 { 435 #if defined(CONFIG_HAVE_EFFICIENT_UNALIGNED_ACCESS) && BITS_PER_LONG == 64 436 const unsigned long *ul1 = (const unsigned long *)a1; 437 const unsigned long *ulm = (const unsigned long *)m; 438 const unsigned long *ul2 = (const unsigned long *)a2; 439 440 return !!(((ul1[0] ^ ul2[0]) & ulm[0]) | 441 ((ul1[1] ^ ul2[1]) & ulm[1])); 442 #else 443 return !!(((a1->s6_addr32[0] ^ a2->s6_addr32[0]) & m->s6_addr32[0]) | 444 ((a1->s6_addr32[1] ^ a2->s6_addr32[1]) & m->s6_addr32[1]) | 445 ((a1->s6_addr32[2] ^ a2->s6_addr32[2]) & m->s6_addr32[2]) | 446 ((a1->s6_addr32[3] ^ a2->s6_addr32[3]) & m->s6_addr32[3])); 447 #endif 448 } 449 450 static inline void ipv6_addr_prefix(struct in6_addr *pfx, 451 const struct in6_addr *addr, 452 int plen) 453 { 454 /* caller must guarantee 0 <= plen <= 128 */ 455 int o = plen >> 3, 456 b = plen & 0x7; 457 458 memset(pfx->s6_addr, 0, sizeof(pfx->s6_addr)); 459 memcpy(pfx->s6_addr, addr, o); 460 if (b != 0) 461 pfx->s6_addr[o] = addr->s6_addr[o] & (0xff00 >> b); 462 } 463 464 static inline void ipv6_addr_prefix_copy(struct in6_addr *addr, 465 const struct in6_addr *pfx, 466 int plen) 467 { 468 /* caller must guarantee 0 <= plen <= 128 */ 469 int o = plen >> 3, 470 b = plen & 0x7; 471 472 memcpy(addr->s6_addr, pfx, o); 473 if (b != 0) { 474 addr->s6_addr[o] &= ~(0xff00 >> b); 475 addr->s6_addr[o] |= (pfx->s6_addr[o] & (0xff00 >> b)); 476 } 477 } 478 479 static inline void __ipv6_addr_set_half(__be32 *addr, 480 __be32 wh, __be32 wl) 481 { 482 #if defined(CONFIG_HAVE_EFFICIENT_UNALIGNED_ACCESS) && BITS_PER_LONG == 64 483 #if defined(__BIG_ENDIAN) 484 if (__builtin_constant_p(wh) && __builtin_constant_p(wl)) { 485 *(__force u64 *)addr = ((__force u64)(wh) << 32 | (__force u64)(wl)); 486 return; 487 } 488 #elif defined(__LITTLE_ENDIAN) 489 if (__builtin_constant_p(wl) && __builtin_constant_p(wh)) { 490 *(__force u64 *)addr = ((__force u64)(wl) << 32 | (__force u64)(wh)); 491 return; 492 } 493 #endif 494 #endif 495 addr[0] = wh; 496 addr[1] = wl; 497 } 498 499 static inline void ipv6_addr_set(struct in6_addr *addr, 500 __be32 w1, __be32 w2, 501 __be32 w3, __be32 w4) 502 { 503 __ipv6_addr_set_half(&addr->s6_addr32[0], w1, w2); 504 __ipv6_addr_set_half(&addr->s6_addr32[2], w3, w4); 505 } 506 507 static inline bool ipv6_addr_equal(const struct in6_addr *a1, 508 const struct in6_addr *a2) 509 { 510 #if defined(CONFIG_HAVE_EFFICIENT_UNALIGNED_ACCESS) && BITS_PER_LONG == 64 511 const unsigned long *ul1 = (const unsigned long *)a1; 512 const unsigned long *ul2 = (const unsigned long *)a2; 513 514 return ((ul1[0] ^ ul2[0]) | (ul1[1] ^ ul2[1])) == 0UL; 515 #else 516 return ((a1->s6_addr32[0] ^ a2->s6_addr32[0]) | 517 (a1->s6_addr32[1] ^ a2->s6_addr32[1]) | 518 (a1->s6_addr32[2] ^ a2->s6_addr32[2]) | 519 (a1->s6_addr32[3] ^ a2->s6_addr32[3])) == 0; 520 #endif 521 } 522 523 #if defined(CONFIG_HAVE_EFFICIENT_UNALIGNED_ACCESS) && BITS_PER_LONG == 64 524 static inline bool __ipv6_prefix_equal64_half(const __be64 *a1, 525 const __be64 *a2, 526 unsigned int len) 527 { 528 if (len && ((*a1 ^ *a2) & cpu_to_be64((~0UL) << (64 - len)))) 529 return false; 530 return true; 531 } 532 533 static inline bool ipv6_prefix_equal(const struct in6_addr *addr1, 534 const struct in6_addr *addr2, 535 unsigned int prefixlen) 536 { 537 const __be64 *a1 = (const __be64 *)addr1; 538 const __be64 *a2 = (const __be64 *)addr2; 539 540 if (prefixlen >= 64) { 541 if (a1[0] ^ a2[0]) 542 return false; 543 return __ipv6_prefix_equal64_half(a1 + 1, a2 + 1, prefixlen - 64); 544 } 545 return __ipv6_prefix_equal64_half(a1, a2, prefixlen); 546 } 547 #else 548 static inline bool ipv6_prefix_equal(const struct in6_addr *addr1, 549 const struct in6_addr *addr2, 550 unsigned int prefixlen) 551 { 552 const __be32 *a1 = addr1->s6_addr32; 553 const __be32 *a2 = addr2->s6_addr32; 554 unsigned int pdw, pbi; 555 556 /* check complete u32 in prefix */ 557 pdw = prefixlen >> 5; 558 if (pdw && memcmp(a1, a2, pdw << 2)) 559 return false; 560 561 /* check incomplete u32 in prefix */ 562 pbi = prefixlen & 0x1f; 563 if (pbi && ((a1[pdw] ^ a2[pdw]) & htonl((0xffffffff) << (32 - pbi)))) 564 return false; 565 566 return true; 567 } 568 #endif 569 570 struct inet_frag_queue; 571 572 enum ip6_defrag_users { 573 IP6_DEFRAG_LOCAL_DELIVER, 574 IP6_DEFRAG_CONNTRACK_IN, 575 __IP6_DEFRAG_CONNTRACK_IN = IP6_DEFRAG_CONNTRACK_IN + USHRT_MAX, 576 IP6_DEFRAG_CONNTRACK_OUT, 577 __IP6_DEFRAG_CONNTRACK_OUT = IP6_DEFRAG_CONNTRACK_OUT + USHRT_MAX, 578 IP6_DEFRAG_CONNTRACK_BRIDGE_IN, 579 __IP6_DEFRAG_CONNTRACK_BRIDGE_IN = IP6_DEFRAG_CONNTRACK_BRIDGE_IN + USHRT_MAX, 580 }; 581 582 struct ip6_create_arg { 583 __be32 id; 584 u32 user; 585 const struct in6_addr *src; 586 const struct in6_addr *dst; 587 int iif; 588 u8 ecn; 589 }; 590 591 void ip6_frag_init(struct inet_frag_queue *q, const void *a); 592 bool ip6_frag_match(const struct inet_frag_queue *q, const void *a); 593 594 /* 595 * Equivalent of ipv4 struct ip 596 */ 597 struct frag_queue { 598 struct inet_frag_queue q; 599 600 __be32 id; /* fragment id */ 601 u32 user; 602 struct in6_addr saddr; 603 struct in6_addr daddr; 604 605 int iif; 606 __u16 nhoffset; 607 u8 ecn; 608 }; 609 610 void ip6_expire_frag_queue(struct net *net, struct frag_queue *fq); 611 612 static inline bool ipv6_addr_any(const struct in6_addr *a) 613 { 614 #if defined(CONFIG_HAVE_EFFICIENT_UNALIGNED_ACCESS) && BITS_PER_LONG == 64 615 const unsigned long *ul = (const unsigned long *)a; 616 617 return (ul[0] | ul[1]) == 0UL; 618 #else 619 return (a->s6_addr32[0] | a->s6_addr32[1] | 620 a->s6_addr32[2] | a->s6_addr32[3]) == 0; 621 #endif 622 } 623 624 static inline u32 ipv6_addr_hash(const struct in6_addr *a) 625 { 626 #if defined(CONFIG_HAVE_EFFICIENT_UNALIGNED_ACCESS) && BITS_PER_LONG == 64 627 const unsigned long *ul = (const unsigned long *)a; 628 unsigned long x = ul[0] ^ ul[1]; 629 630 return (u32)(x ^ (x >> 32)); 631 #else 632 return (__force u32)(a->s6_addr32[0] ^ a->s6_addr32[1] ^ 633 a->s6_addr32[2] ^ a->s6_addr32[3]); 634 #endif 635 } 636 637 /* more secured version of ipv6_addr_hash() */ 638 static inline u32 __ipv6_addr_jhash(const struct in6_addr *a, const u32 initval) 639 { 640 u32 v = (__force u32)a->s6_addr32[0] ^ (__force u32)a->s6_addr32[1]; 641 642 return jhash_3words(v, 643 (__force u32)a->s6_addr32[2], 644 (__force u32)a->s6_addr32[3], 645 initval); 646 } 647 648 static inline bool ipv6_addr_loopback(const struct in6_addr *a) 649 { 650 #if defined(CONFIG_HAVE_EFFICIENT_UNALIGNED_ACCESS) && BITS_PER_LONG == 64 651 const __be64 *be = (const __be64 *)a; 652 653 return (be[0] | (be[1] ^ cpu_to_be64(1))) == 0UL; 654 #else 655 return (a->s6_addr32[0] | a->s6_addr32[1] | 656 a->s6_addr32[2] | (a->s6_addr32[3] ^ cpu_to_be32(1))) == 0; 657 #endif 658 } 659 660 /* 661 * Note that we must __force cast these to unsigned long to make sparse happy, 662 * since all of the endian-annotated types are fixed size regardless of arch. 663 */ 664 static inline bool ipv6_addr_v4mapped(const struct in6_addr *a) 665 { 666 return ( 667 #if defined(CONFIG_HAVE_EFFICIENT_UNALIGNED_ACCESS) && BITS_PER_LONG == 64 668 *(unsigned long *)a | 669 #else 670 (__force unsigned long)(a->s6_addr32[0] | a->s6_addr32[1]) | 671 #endif 672 (__force unsigned long)(a->s6_addr32[2] ^ 673 cpu_to_be32(0x0000ffff))) == 0UL; 674 } 675 676 static inline u32 ipv6_portaddr_hash(const struct net *net, 677 const struct in6_addr *addr6, 678 unsigned int port) 679 { 680 unsigned int hash, mix = net_hash_mix(net); 681 682 if (ipv6_addr_any(addr6)) 683 hash = jhash_1word(0, mix); 684 else if (ipv6_addr_v4mapped(addr6)) 685 hash = jhash_1word((__force u32)addr6->s6_addr32[3], mix); 686 else 687 hash = jhash2((__force u32 *)addr6->s6_addr32, 4, mix); 688 689 return hash ^ port; 690 } 691 692 /* 693 * Check for a RFC 4843 ORCHID address 694 * (Overlay Routable Cryptographic Hash Identifiers) 695 */ 696 static inline bool ipv6_addr_orchid(const struct in6_addr *a) 697 { 698 return (a->s6_addr32[0] & htonl(0xfffffff0)) == htonl(0x20010010); 699 } 700 701 static inline bool ipv6_addr_is_multicast(const struct in6_addr *addr) 702 { 703 return (addr->s6_addr32[0] & htonl(0xFF000000)) == htonl(0xFF000000); 704 } 705 706 static inline void ipv6_addr_set_v4mapped(const __be32 addr, 707 struct in6_addr *v4mapped) 708 { 709 ipv6_addr_set(v4mapped, 710 0, 0, 711 htonl(0x0000FFFF), 712 addr); 713 } 714 715 /* 716 * find the first different bit between two addresses 717 * length of address must be a multiple of 32bits 718 */ 719 static inline int __ipv6_addr_diff32(const void *token1, const void *token2, int addrlen) 720 { 721 const __be32 *a1 = token1, *a2 = token2; 722 int i; 723 724 addrlen >>= 2; 725 726 for (i = 0; i < addrlen; i++) { 727 __be32 xb = a1[i] ^ a2[i]; 728 if (xb) 729 return i * 32 + 31 - __fls(ntohl(xb)); 730 } 731 732 /* 733 * we should *never* get to this point since that 734 * would mean the addrs are equal 735 * 736 * However, we do get to it 8) And exacly, when 737 * addresses are equal 8) 738 * 739 * ip route add 1111::/128 via ... 740 * ip route add 1111::/64 via ... 741 * and we are here. 742 * 743 * Ideally, this function should stop comparison 744 * at prefix length. It does not, but it is still OK, 745 * if returned value is greater than prefix length. 746 * --ANK (980803) 747 */ 748 return addrlen << 5; 749 } 750 751 #if defined(CONFIG_HAVE_EFFICIENT_UNALIGNED_ACCESS) && BITS_PER_LONG == 64 752 static inline int __ipv6_addr_diff64(const void *token1, const void *token2, int addrlen) 753 { 754 const __be64 *a1 = token1, *a2 = token2; 755 int i; 756 757 addrlen >>= 3; 758 759 for (i = 0; i < addrlen; i++) { 760 __be64 xb = a1[i] ^ a2[i]; 761 if (xb) 762 return i * 64 + 63 - __fls(be64_to_cpu(xb)); 763 } 764 765 return addrlen << 6; 766 } 767 #endif 768 769 static inline int __ipv6_addr_diff(const void *token1, const void *token2, int addrlen) 770 { 771 #if defined(CONFIG_HAVE_EFFICIENT_UNALIGNED_ACCESS) && BITS_PER_LONG == 64 772 if (__builtin_constant_p(addrlen) && !(addrlen & 7)) 773 return __ipv6_addr_diff64(token1, token2, addrlen); 774 #endif 775 return __ipv6_addr_diff32(token1, token2, addrlen); 776 } 777 778 static inline int ipv6_addr_diff(const struct in6_addr *a1, const struct in6_addr *a2) 779 { 780 return __ipv6_addr_diff(a1, a2, sizeof(struct in6_addr)); 781 } 782 783 __be32 ipv6_select_ident(struct net *net, 784 const struct in6_addr *daddr, 785 const struct in6_addr *saddr); 786 __be32 ipv6_proxy_select_ident(struct net *net, struct sk_buff *skb); 787 788 int ip6_dst_hoplimit(struct dst_entry *dst); 789 790 static inline int ip6_sk_dst_hoplimit(struct ipv6_pinfo *np, struct flowi6 *fl6, 791 struct dst_entry *dst) 792 { 793 int hlimit; 794 795 if (ipv6_addr_is_multicast(&fl6->daddr)) 796 hlimit = np->mcast_hops; 797 else 798 hlimit = np->hop_limit; 799 if (hlimit < 0) 800 hlimit = ip6_dst_hoplimit(dst); 801 return hlimit; 802 } 803 804 /* copy IPv6 saddr & daddr to flow_keys, possibly using 64bit load/store 805 * Equivalent to : flow->v6addrs.src = iph->saddr; 806 * flow->v6addrs.dst = iph->daddr; 807 */ 808 static inline void iph_to_flow_copy_v6addrs(struct flow_keys *flow, 809 const struct ipv6hdr *iph) 810 { 811 BUILD_BUG_ON(offsetof(typeof(flow->addrs), v6addrs.dst) != 812 offsetof(typeof(flow->addrs), v6addrs.src) + 813 sizeof(flow->addrs.v6addrs.src)); 814 memcpy(&flow->addrs.v6addrs, &iph->saddr, sizeof(flow->addrs.v6addrs)); 815 flow->control.addr_type = FLOW_DISSECTOR_KEY_IPV6_ADDRS; 816 } 817 818 #if IS_ENABLED(CONFIG_IPV6) 819 820 /* Sysctl settings for net ipv6.auto_flowlabels */ 821 #define IP6_AUTO_FLOW_LABEL_OFF 0 822 #define IP6_AUTO_FLOW_LABEL_OPTOUT 1 823 #define IP6_AUTO_FLOW_LABEL_OPTIN 2 824 #define IP6_AUTO_FLOW_LABEL_FORCED 3 825 826 #define IP6_AUTO_FLOW_LABEL_MAX IP6_AUTO_FLOW_LABEL_FORCED 827 828 #define IP6_DEFAULT_AUTO_FLOW_LABELS IP6_AUTO_FLOW_LABEL_OPTOUT 829 830 static inline __be32 ip6_make_flowlabel(struct net *net, struct sk_buff *skb, 831 __be32 flowlabel, bool autolabel, 832 struct flowi6 *fl6) 833 { 834 u32 hash; 835 836 /* @flowlabel may include more than a flow label, eg, the traffic class. 837 * Here we want only the flow label value. 838 */ 839 flowlabel &= IPV6_FLOWLABEL_MASK; 840 841 if (flowlabel || 842 net->ipv6.sysctl.auto_flowlabels == IP6_AUTO_FLOW_LABEL_OFF || 843 (!autolabel && 844 net->ipv6.sysctl.auto_flowlabels != IP6_AUTO_FLOW_LABEL_FORCED)) 845 return flowlabel; 846 847 hash = skb_get_hash_flowi6(skb, fl6); 848 849 /* Since this is being sent on the wire obfuscate hash a bit 850 * to minimize possbility that any useful information to an 851 * attacker is leaked. Only lower 20 bits are relevant. 852 */ 853 rol32(hash, 16); 854 855 flowlabel = (__force __be32)hash & IPV6_FLOWLABEL_MASK; 856 857 if (net->ipv6.sysctl.flowlabel_state_ranges) 858 flowlabel |= IPV6_FLOWLABEL_STATELESS_FLAG; 859 860 return flowlabel; 861 } 862 863 static inline int ip6_default_np_autolabel(struct net *net) 864 { 865 switch (net->ipv6.sysctl.auto_flowlabels) { 866 case IP6_AUTO_FLOW_LABEL_OFF: 867 case IP6_AUTO_FLOW_LABEL_OPTIN: 868 default: 869 return 0; 870 case IP6_AUTO_FLOW_LABEL_OPTOUT: 871 case IP6_AUTO_FLOW_LABEL_FORCED: 872 return 1; 873 } 874 } 875 #else 876 static inline void ip6_set_txhash(struct sock *sk) { } 877 static inline __be32 ip6_make_flowlabel(struct net *net, struct sk_buff *skb, 878 __be32 flowlabel, bool autolabel, 879 struct flowi6 *fl6) 880 { 881 return flowlabel; 882 } 883 static inline int ip6_default_np_autolabel(struct net *net) 884 { 885 return 0; 886 } 887 #endif 888 889 #if IS_ENABLED(CONFIG_IPV6) 890 static inline int ip6_multipath_hash_policy(const struct net *net) 891 { 892 return net->ipv6.sysctl.multipath_hash_policy; 893 } 894 #else 895 static inline int ip6_multipath_hash_policy(const struct net *net) 896 { 897 return 0; 898 } 899 #endif 900 901 /* 902 * Header manipulation 903 */ 904 static inline void ip6_flow_hdr(struct ipv6hdr *hdr, unsigned int tclass, 905 __be32 flowlabel) 906 { 907 *(__be32 *)hdr = htonl(0x60000000 | (tclass << 20)) | flowlabel; 908 } 909 910 static inline __be32 ip6_flowinfo(const struct ipv6hdr *hdr) 911 { 912 return *(__be32 *)hdr & IPV6_FLOWINFO_MASK; 913 } 914 915 static inline __be32 ip6_flowlabel(const struct ipv6hdr *hdr) 916 { 917 return *(__be32 *)hdr & IPV6_FLOWLABEL_MASK; 918 } 919 920 static inline u8 ip6_tclass(__be32 flowinfo) 921 { 922 return ntohl(flowinfo & IPV6_TCLASS_MASK) >> IPV6_TCLASS_SHIFT; 923 } 924 925 static inline __be32 ip6_make_flowinfo(unsigned int tclass, __be32 flowlabel) 926 { 927 return htonl(tclass << IPV6_TCLASS_SHIFT) | flowlabel; 928 } 929 930 /* 931 * Prototypes exported by ipv6 932 */ 933 934 /* 935 * rcv function (called from netdevice level) 936 */ 937 938 int ipv6_rcv(struct sk_buff *skb, struct net_device *dev, 939 struct packet_type *pt, struct net_device *orig_dev); 940 941 int ip6_rcv_finish(struct net *net, struct sock *sk, struct sk_buff *skb); 942 943 /* 944 * upper-layer output functions 945 */ 946 int ip6_xmit(const struct sock *sk, struct sk_buff *skb, struct flowi6 *fl6, 947 __u32 mark, struct ipv6_txoptions *opt, int tclass); 948 949 int ip6_find_1stfragopt(struct sk_buff *skb, u8 **nexthdr); 950 951 int ip6_append_data(struct sock *sk, 952 int getfrag(void *from, char *to, int offset, int len, 953 int odd, struct sk_buff *skb), 954 void *from, int length, int transhdrlen, 955 struct ipcm6_cookie *ipc6, struct flowi6 *fl6, 956 struct rt6_info *rt, unsigned int flags, 957 const struct sockcm_cookie *sockc); 958 959 int ip6_push_pending_frames(struct sock *sk); 960 961 void ip6_flush_pending_frames(struct sock *sk); 962 963 int ip6_send_skb(struct sk_buff *skb); 964 965 struct sk_buff *__ip6_make_skb(struct sock *sk, struct sk_buff_head *queue, 966 struct inet_cork_full *cork, 967 struct inet6_cork *v6_cork); 968 struct sk_buff *ip6_make_skb(struct sock *sk, 969 int getfrag(void *from, char *to, int offset, 970 int len, int odd, struct sk_buff *skb), 971 void *from, int length, int transhdrlen, 972 struct ipcm6_cookie *ipc6, struct flowi6 *fl6, 973 struct rt6_info *rt, unsigned int flags, 974 const struct sockcm_cookie *sockc); 975 976 static inline struct sk_buff *ip6_finish_skb(struct sock *sk) 977 { 978 return __ip6_make_skb(sk, &sk->sk_write_queue, &inet_sk(sk)->cork, 979 &inet6_sk(sk)->cork); 980 } 981 982 unsigned int ip6_dst_mtu_forward(const struct dst_entry *dst); 983 984 int ip6_dst_lookup(struct net *net, struct sock *sk, struct dst_entry **dst, 985 struct flowi6 *fl6); 986 struct dst_entry *ip6_dst_lookup_flow(const struct sock *sk, struct flowi6 *fl6, 987 const struct in6_addr *final_dst); 988 struct dst_entry *ip6_sk_dst_lookup_flow(struct sock *sk, struct flowi6 *fl6, 989 const struct in6_addr *final_dst); 990 struct dst_entry *ip6_blackhole_route(struct net *net, 991 struct dst_entry *orig_dst); 992 993 /* 994 * skb processing functions 995 */ 996 997 int ip6_output(struct net *net, struct sock *sk, struct sk_buff *skb); 998 int ip6_forward(struct sk_buff *skb); 999 int ip6_input(struct sk_buff *skb); 1000 int ip6_mc_input(struct sk_buff *skb); 1001 1002 int __ip6_local_out(struct net *net, struct sock *sk, struct sk_buff *skb); 1003 int ip6_local_out(struct net *net, struct sock *sk, struct sk_buff *skb); 1004 1005 /* 1006 * Extension header (options) processing 1007 */ 1008 1009 void ipv6_push_nfrag_opts(struct sk_buff *skb, struct ipv6_txoptions *opt, 1010 u8 *proto, struct in6_addr **daddr_p, 1011 struct in6_addr *saddr); 1012 void ipv6_push_frag_opts(struct sk_buff *skb, struct ipv6_txoptions *opt, 1013 u8 *proto); 1014 1015 int ipv6_skip_exthdr(const struct sk_buff *, int start, u8 *nexthdrp, 1016 __be16 *frag_offp); 1017 1018 bool ipv6_ext_hdr(u8 nexthdr); 1019 1020 enum { 1021 IP6_FH_F_FRAG = (1 << 0), 1022 IP6_FH_F_AUTH = (1 << 1), 1023 IP6_FH_F_SKIP_RH = (1 << 2), 1024 }; 1025 1026 /* find specified header and get offset to it */ 1027 int ipv6_find_hdr(const struct sk_buff *skb, unsigned int *offset, int target, 1028 unsigned short *fragoff, int *fragflg); 1029 1030 int ipv6_find_tlv(const struct sk_buff *skb, int offset, int type); 1031 1032 struct in6_addr *fl6_update_dst(struct flowi6 *fl6, 1033 const struct ipv6_txoptions *opt, 1034 struct in6_addr *orig); 1035 1036 /* 1037 * socket options (ipv6_sockglue.c) 1038 */ 1039 1040 int ipv6_setsockopt(struct sock *sk, int level, int optname, 1041 char __user *optval, unsigned int optlen); 1042 int ipv6_getsockopt(struct sock *sk, int level, int optname, 1043 char __user *optval, int __user *optlen); 1044 int compat_ipv6_setsockopt(struct sock *sk, int level, int optname, 1045 char __user *optval, unsigned int optlen); 1046 int compat_ipv6_getsockopt(struct sock *sk, int level, int optname, 1047 char __user *optval, int __user *optlen); 1048 1049 int __ip6_datagram_connect(struct sock *sk, struct sockaddr *addr, 1050 int addr_len); 1051 int ip6_datagram_connect(struct sock *sk, struct sockaddr *addr, int addr_len); 1052 int ip6_datagram_connect_v6_only(struct sock *sk, struct sockaddr *addr, 1053 int addr_len); 1054 int ip6_datagram_dst_update(struct sock *sk, bool fix_sk_saddr); 1055 void ip6_datagram_release_cb(struct sock *sk); 1056 1057 int ipv6_recv_error(struct sock *sk, struct msghdr *msg, int len, 1058 int *addr_len); 1059 int ipv6_recv_rxpmtu(struct sock *sk, struct msghdr *msg, int len, 1060 int *addr_len); 1061 void ipv6_icmp_error(struct sock *sk, struct sk_buff *skb, int err, __be16 port, 1062 u32 info, u8 *payload); 1063 void ipv6_local_error(struct sock *sk, int err, struct flowi6 *fl6, u32 info); 1064 void ipv6_local_rxpmtu(struct sock *sk, struct flowi6 *fl6, u32 mtu); 1065 1066 int inet6_release(struct socket *sock); 1067 int inet6_bind(struct socket *sock, struct sockaddr *uaddr, int addr_len); 1068 int inet6_getname(struct socket *sock, struct sockaddr *uaddr, 1069 int peer); 1070 int inet6_ioctl(struct socket *sock, unsigned int cmd, unsigned long arg); 1071 1072 int inet6_hash_connect(struct inet_timewait_death_row *death_row, 1073 struct sock *sk); 1074 1075 /* 1076 * reassembly.c 1077 */ 1078 extern const struct proto_ops inet6_stream_ops; 1079 extern const struct proto_ops inet6_dgram_ops; 1080 extern const struct proto_ops inet6_sockraw_ops; 1081 1082 struct group_source_req; 1083 struct group_filter; 1084 1085 int ip6_mc_source(int add, int omode, struct sock *sk, 1086 struct group_source_req *pgsr); 1087 int ip6_mc_msfilter(struct sock *sk, struct group_filter *gsf); 1088 int ip6_mc_msfget(struct sock *sk, struct group_filter *gsf, 1089 struct group_filter __user *optval, int __user *optlen); 1090 1091 #ifdef CONFIG_PROC_FS 1092 int ac6_proc_init(struct net *net); 1093 void ac6_proc_exit(struct net *net); 1094 int raw6_proc_init(void); 1095 void raw6_proc_exit(void); 1096 int tcp6_proc_init(struct net *net); 1097 void tcp6_proc_exit(struct net *net); 1098 int udp6_proc_init(struct net *net); 1099 void udp6_proc_exit(struct net *net); 1100 int udplite6_proc_init(void); 1101 void udplite6_proc_exit(void); 1102 int ipv6_misc_proc_init(void); 1103 void ipv6_misc_proc_exit(void); 1104 int snmp6_register_dev(struct inet6_dev *idev); 1105 int snmp6_unregister_dev(struct inet6_dev *idev); 1106 1107 #else 1108 static inline int ac6_proc_init(struct net *net) { return 0; } 1109 static inline void ac6_proc_exit(struct net *net) { } 1110 static inline int snmp6_register_dev(struct inet6_dev *idev) { return 0; } 1111 static inline int snmp6_unregister_dev(struct inet6_dev *idev) { return 0; } 1112 #endif 1113 1114 #ifdef CONFIG_SYSCTL 1115 extern struct ctl_table ipv6_route_table_template[]; 1116 1117 struct ctl_table *ipv6_icmp_sysctl_init(struct net *net); 1118 struct ctl_table *ipv6_route_sysctl_init(struct net *net); 1119 int ipv6_sysctl_register(void); 1120 void ipv6_sysctl_unregister(void); 1121 #endif 1122 1123 int ipv6_sock_mc_join(struct sock *sk, int ifindex, 1124 const struct in6_addr *addr); 1125 int ipv6_sock_mc_drop(struct sock *sk, int ifindex, 1126 const struct in6_addr *addr); 1127 #endif /* _NET_IPV6_H */ 1128