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 #define IPV6_FRAG_HIGH_THRESH (4 * 1024*1024) /* 4194304 */ 383 #define IPV6_FRAG_LOW_THRESH (3 * 1024*1024) /* 3145728 */ 384 #define IPV6_FRAG_TIMEOUT (60 * HZ) /* 60 seconds */ 385 386 int __ipv6_addr_type(const struct in6_addr *addr); 387 static inline int ipv6_addr_type(const struct in6_addr *addr) 388 { 389 return __ipv6_addr_type(addr) & 0xffff; 390 } 391 392 static inline int ipv6_addr_scope(const struct in6_addr *addr) 393 { 394 return __ipv6_addr_type(addr) & IPV6_ADDR_SCOPE_MASK; 395 } 396 397 static inline int __ipv6_addr_src_scope(int type) 398 { 399 return (type == IPV6_ADDR_ANY) ? __IPV6_ADDR_SCOPE_INVALID : (type >> 16); 400 } 401 402 static inline int ipv6_addr_src_scope(const struct in6_addr *addr) 403 { 404 return __ipv6_addr_src_scope(__ipv6_addr_type(addr)); 405 } 406 407 static inline bool __ipv6_addr_needs_scope_id(int type) 408 { 409 return type & IPV6_ADDR_LINKLOCAL || 410 (type & IPV6_ADDR_MULTICAST && 411 (type & (IPV6_ADDR_LOOPBACK|IPV6_ADDR_LINKLOCAL))); 412 } 413 414 static inline __u32 ipv6_iface_scope_id(const struct in6_addr *addr, int iface) 415 { 416 return __ipv6_addr_needs_scope_id(__ipv6_addr_type(addr)) ? iface : 0; 417 } 418 419 static inline int ipv6_addr_cmp(const struct in6_addr *a1, const struct in6_addr *a2) 420 { 421 return memcmp(a1, a2, sizeof(struct in6_addr)); 422 } 423 424 static inline bool 425 ipv6_masked_addr_cmp(const struct in6_addr *a1, const struct in6_addr *m, 426 const struct in6_addr *a2) 427 { 428 #if defined(CONFIG_HAVE_EFFICIENT_UNALIGNED_ACCESS) && BITS_PER_LONG == 64 429 const unsigned long *ul1 = (const unsigned long *)a1; 430 const unsigned long *ulm = (const unsigned long *)m; 431 const unsigned long *ul2 = (const unsigned long *)a2; 432 433 return !!(((ul1[0] ^ ul2[0]) & ulm[0]) | 434 ((ul1[1] ^ ul2[1]) & ulm[1])); 435 #else 436 return !!(((a1->s6_addr32[0] ^ a2->s6_addr32[0]) & m->s6_addr32[0]) | 437 ((a1->s6_addr32[1] ^ a2->s6_addr32[1]) & m->s6_addr32[1]) | 438 ((a1->s6_addr32[2] ^ a2->s6_addr32[2]) & m->s6_addr32[2]) | 439 ((a1->s6_addr32[3] ^ a2->s6_addr32[3]) & m->s6_addr32[3])); 440 #endif 441 } 442 443 static inline void ipv6_addr_prefix(struct in6_addr *pfx, 444 const struct in6_addr *addr, 445 int plen) 446 { 447 /* caller must guarantee 0 <= plen <= 128 */ 448 int o = plen >> 3, 449 b = plen & 0x7; 450 451 memset(pfx->s6_addr, 0, sizeof(pfx->s6_addr)); 452 memcpy(pfx->s6_addr, addr, o); 453 if (b != 0) 454 pfx->s6_addr[o] = addr->s6_addr[o] & (0xff00 >> b); 455 } 456 457 static inline void ipv6_addr_prefix_copy(struct in6_addr *addr, 458 const struct in6_addr *pfx, 459 int plen) 460 { 461 /* caller must guarantee 0 <= plen <= 128 */ 462 int o = plen >> 3, 463 b = plen & 0x7; 464 465 memcpy(addr->s6_addr, pfx, o); 466 if (b != 0) { 467 addr->s6_addr[o] &= ~(0xff00 >> b); 468 addr->s6_addr[o] |= (pfx->s6_addr[o] & (0xff00 >> b)); 469 } 470 } 471 472 static inline void __ipv6_addr_set_half(__be32 *addr, 473 __be32 wh, __be32 wl) 474 { 475 #if defined(CONFIG_HAVE_EFFICIENT_UNALIGNED_ACCESS) && BITS_PER_LONG == 64 476 #if defined(__BIG_ENDIAN) 477 if (__builtin_constant_p(wh) && __builtin_constant_p(wl)) { 478 *(__force u64 *)addr = ((__force u64)(wh) << 32 | (__force u64)(wl)); 479 return; 480 } 481 #elif defined(__LITTLE_ENDIAN) 482 if (__builtin_constant_p(wl) && __builtin_constant_p(wh)) { 483 *(__force u64 *)addr = ((__force u64)(wl) << 32 | (__force u64)(wh)); 484 return; 485 } 486 #endif 487 #endif 488 addr[0] = wh; 489 addr[1] = wl; 490 } 491 492 static inline void ipv6_addr_set(struct in6_addr *addr, 493 __be32 w1, __be32 w2, 494 __be32 w3, __be32 w4) 495 { 496 __ipv6_addr_set_half(&addr->s6_addr32[0], w1, w2); 497 __ipv6_addr_set_half(&addr->s6_addr32[2], w3, w4); 498 } 499 500 static inline bool ipv6_addr_equal(const struct in6_addr *a1, 501 const struct in6_addr *a2) 502 { 503 #if defined(CONFIG_HAVE_EFFICIENT_UNALIGNED_ACCESS) && BITS_PER_LONG == 64 504 const unsigned long *ul1 = (const unsigned long *)a1; 505 const unsigned long *ul2 = (const unsigned long *)a2; 506 507 return ((ul1[0] ^ ul2[0]) | (ul1[1] ^ ul2[1])) == 0UL; 508 #else 509 return ((a1->s6_addr32[0] ^ a2->s6_addr32[0]) | 510 (a1->s6_addr32[1] ^ a2->s6_addr32[1]) | 511 (a1->s6_addr32[2] ^ a2->s6_addr32[2]) | 512 (a1->s6_addr32[3] ^ a2->s6_addr32[3])) == 0; 513 #endif 514 } 515 516 #if defined(CONFIG_HAVE_EFFICIENT_UNALIGNED_ACCESS) && BITS_PER_LONG == 64 517 static inline bool __ipv6_prefix_equal64_half(const __be64 *a1, 518 const __be64 *a2, 519 unsigned int len) 520 { 521 if (len && ((*a1 ^ *a2) & cpu_to_be64((~0UL) << (64 - len)))) 522 return false; 523 return true; 524 } 525 526 static inline bool ipv6_prefix_equal(const struct in6_addr *addr1, 527 const struct in6_addr *addr2, 528 unsigned int prefixlen) 529 { 530 const __be64 *a1 = (const __be64 *)addr1; 531 const __be64 *a2 = (const __be64 *)addr2; 532 533 if (prefixlen >= 64) { 534 if (a1[0] ^ a2[0]) 535 return false; 536 return __ipv6_prefix_equal64_half(a1 + 1, a2 + 1, prefixlen - 64); 537 } 538 return __ipv6_prefix_equal64_half(a1, a2, prefixlen); 539 } 540 #else 541 static inline bool ipv6_prefix_equal(const struct in6_addr *addr1, 542 const struct in6_addr *addr2, 543 unsigned int prefixlen) 544 { 545 const __be32 *a1 = addr1->s6_addr32; 546 const __be32 *a2 = addr2->s6_addr32; 547 unsigned int pdw, pbi; 548 549 /* check complete u32 in prefix */ 550 pdw = prefixlen >> 5; 551 if (pdw && memcmp(a1, a2, pdw << 2)) 552 return false; 553 554 /* check incomplete u32 in prefix */ 555 pbi = prefixlen & 0x1f; 556 if (pbi && ((a1[pdw] ^ a2[pdw]) & htonl((0xffffffff) << (32 - pbi)))) 557 return false; 558 559 return true; 560 } 561 #endif 562 563 struct inet_frag_queue; 564 565 enum ip6_defrag_users { 566 IP6_DEFRAG_LOCAL_DELIVER, 567 IP6_DEFRAG_CONNTRACK_IN, 568 __IP6_DEFRAG_CONNTRACK_IN = IP6_DEFRAG_CONNTRACK_IN + USHRT_MAX, 569 IP6_DEFRAG_CONNTRACK_OUT, 570 __IP6_DEFRAG_CONNTRACK_OUT = IP6_DEFRAG_CONNTRACK_OUT + USHRT_MAX, 571 IP6_DEFRAG_CONNTRACK_BRIDGE_IN, 572 __IP6_DEFRAG_CONNTRACK_BRIDGE_IN = IP6_DEFRAG_CONNTRACK_BRIDGE_IN + USHRT_MAX, 573 }; 574 575 void ip6_frag_init(struct inet_frag_queue *q, const void *a); 576 extern const struct rhashtable_params ip6_rhash_params; 577 578 /* 579 * Equivalent of ipv4 struct ip 580 */ 581 struct frag_queue { 582 struct inet_frag_queue q; 583 584 int iif; 585 __u16 nhoffset; 586 u8 ecn; 587 }; 588 589 void ip6_expire_frag_queue(struct net *net, struct frag_queue *fq); 590 591 static inline bool ipv6_addr_any(const struct in6_addr *a) 592 { 593 #if defined(CONFIG_HAVE_EFFICIENT_UNALIGNED_ACCESS) && BITS_PER_LONG == 64 594 const unsigned long *ul = (const unsigned long *)a; 595 596 return (ul[0] | ul[1]) == 0UL; 597 #else 598 return (a->s6_addr32[0] | a->s6_addr32[1] | 599 a->s6_addr32[2] | a->s6_addr32[3]) == 0; 600 #endif 601 } 602 603 static inline u32 ipv6_addr_hash(const struct in6_addr *a) 604 { 605 #if defined(CONFIG_HAVE_EFFICIENT_UNALIGNED_ACCESS) && BITS_PER_LONG == 64 606 const unsigned long *ul = (const unsigned long *)a; 607 unsigned long x = ul[0] ^ ul[1]; 608 609 return (u32)(x ^ (x >> 32)); 610 #else 611 return (__force u32)(a->s6_addr32[0] ^ a->s6_addr32[1] ^ 612 a->s6_addr32[2] ^ a->s6_addr32[3]); 613 #endif 614 } 615 616 /* more secured version of ipv6_addr_hash() */ 617 static inline u32 __ipv6_addr_jhash(const struct in6_addr *a, const u32 initval) 618 { 619 u32 v = (__force u32)a->s6_addr32[0] ^ (__force u32)a->s6_addr32[1]; 620 621 return jhash_3words(v, 622 (__force u32)a->s6_addr32[2], 623 (__force u32)a->s6_addr32[3], 624 initval); 625 } 626 627 static inline bool ipv6_addr_loopback(const struct in6_addr *a) 628 { 629 #if defined(CONFIG_HAVE_EFFICIENT_UNALIGNED_ACCESS) && BITS_PER_LONG == 64 630 const __be64 *be = (const __be64 *)a; 631 632 return (be[0] | (be[1] ^ cpu_to_be64(1))) == 0UL; 633 #else 634 return (a->s6_addr32[0] | a->s6_addr32[1] | 635 a->s6_addr32[2] | (a->s6_addr32[3] ^ cpu_to_be32(1))) == 0; 636 #endif 637 } 638 639 /* 640 * Note that we must __force cast these to unsigned long to make sparse happy, 641 * since all of the endian-annotated types are fixed size regardless of arch. 642 */ 643 static inline bool ipv6_addr_v4mapped(const struct in6_addr *a) 644 { 645 return ( 646 #if defined(CONFIG_HAVE_EFFICIENT_UNALIGNED_ACCESS) && BITS_PER_LONG == 64 647 *(unsigned long *)a | 648 #else 649 (__force unsigned long)(a->s6_addr32[0] | a->s6_addr32[1]) | 650 #endif 651 (__force unsigned long)(a->s6_addr32[2] ^ 652 cpu_to_be32(0x0000ffff))) == 0UL; 653 } 654 655 static inline u32 ipv6_portaddr_hash(const struct net *net, 656 const struct in6_addr *addr6, 657 unsigned int port) 658 { 659 unsigned int hash, mix = net_hash_mix(net); 660 661 if (ipv6_addr_any(addr6)) 662 hash = jhash_1word(0, mix); 663 else if (ipv6_addr_v4mapped(addr6)) 664 hash = jhash_1word((__force u32)addr6->s6_addr32[3], mix); 665 else 666 hash = jhash2((__force u32 *)addr6->s6_addr32, 4, mix); 667 668 return hash ^ port; 669 } 670 671 /* 672 * Check for a RFC 4843 ORCHID address 673 * (Overlay Routable Cryptographic Hash Identifiers) 674 */ 675 static inline bool ipv6_addr_orchid(const struct in6_addr *a) 676 { 677 return (a->s6_addr32[0] & htonl(0xfffffff0)) == htonl(0x20010010); 678 } 679 680 static inline bool ipv6_addr_is_multicast(const struct in6_addr *addr) 681 { 682 return (addr->s6_addr32[0] & htonl(0xFF000000)) == htonl(0xFF000000); 683 } 684 685 static inline void ipv6_addr_set_v4mapped(const __be32 addr, 686 struct in6_addr *v4mapped) 687 { 688 ipv6_addr_set(v4mapped, 689 0, 0, 690 htonl(0x0000FFFF), 691 addr); 692 } 693 694 /* 695 * find the first different bit between two addresses 696 * length of address must be a multiple of 32bits 697 */ 698 static inline int __ipv6_addr_diff32(const void *token1, const void *token2, int addrlen) 699 { 700 const __be32 *a1 = token1, *a2 = token2; 701 int i; 702 703 addrlen >>= 2; 704 705 for (i = 0; i < addrlen; i++) { 706 __be32 xb = a1[i] ^ a2[i]; 707 if (xb) 708 return i * 32 + 31 - __fls(ntohl(xb)); 709 } 710 711 /* 712 * we should *never* get to this point since that 713 * would mean the addrs are equal 714 * 715 * However, we do get to it 8) And exacly, when 716 * addresses are equal 8) 717 * 718 * ip route add 1111::/128 via ... 719 * ip route add 1111::/64 via ... 720 * and we are here. 721 * 722 * Ideally, this function should stop comparison 723 * at prefix length. It does not, but it is still OK, 724 * if returned value is greater than prefix length. 725 * --ANK (980803) 726 */ 727 return addrlen << 5; 728 } 729 730 #if defined(CONFIG_HAVE_EFFICIENT_UNALIGNED_ACCESS) && BITS_PER_LONG == 64 731 static inline int __ipv6_addr_diff64(const void *token1, const void *token2, int addrlen) 732 { 733 const __be64 *a1 = token1, *a2 = token2; 734 int i; 735 736 addrlen >>= 3; 737 738 for (i = 0; i < addrlen; i++) { 739 __be64 xb = a1[i] ^ a2[i]; 740 if (xb) 741 return i * 64 + 63 - __fls(be64_to_cpu(xb)); 742 } 743 744 return addrlen << 6; 745 } 746 #endif 747 748 static inline int __ipv6_addr_diff(const void *token1, const void *token2, int addrlen) 749 { 750 #if defined(CONFIG_HAVE_EFFICIENT_UNALIGNED_ACCESS) && BITS_PER_LONG == 64 751 if (__builtin_constant_p(addrlen) && !(addrlen & 7)) 752 return __ipv6_addr_diff64(token1, token2, addrlen); 753 #endif 754 return __ipv6_addr_diff32(token1, token2, addrlen); 755 } 756 757 static inline int ipv6_addr_diff(const struct in6_addr *a1, const struct in6_addr *a2) 758 { 759 return __ipv6_addr_diff(a1, a2, sizeof(struct in6_addr)); 760 } 761 762 __be32 ipv6_select_ident(struct net *net, 763 const struct in6_addr *daddr, 764 const struct in6_addr *saddr); 765 __be32 ipv6_proxy_select_ident(struct net *net, struct sk_buff *skb); 766 767 int ip6_dst_hoplimit(struct dst_entry *dst); 768 769 static inline int ip6_sk_dst_hoplimit(struct ipv6_pinfo *np, struct flowi6 *fl6, 770 struct dst_entry *dst) 771 { 772 int hlimit; 773 774 if (ipv6_addr_is_multicast(&fl6->daddr)) 775 hlimit = np->mcast_hops; 776 else 777 hlimit = np->hop_limit; 778 if (hlimit < 0) 779 hlimit = ip6_dst_hoplimit(dst); 780 return hlimit; 781 } 782 783 /* copy IPv6 saddr & daddr to flow_keys, possibly using 64bit load/store 784 * Equivalent to : flow->v6addrs.src = iph->saddr; 785 * flow->v6addrs.dst = iph->daddr; 786 */ 787 static inline void iph_to_flow_copy_v6addrs(struct flow_keys *flow, 788 const struct ipv6hdr *iph) 789 { 790 BUILD_BUG_ON(offsetof(typeof(flow->addrs), v6addrs.dst) != 791 offsetof(typeof(flow->addrs), v6addrs.src) + 792 sizeof(flow->addrs.v6addrs.src)); 793 memcpy(&flow->addrs.v6addrs, &iph->saddr, sizeof(flow->addrs.v6addrs)); 794 flow->control.addr_type = FLOW_DISSECTOR_KEY_IPV6_ADDRS; 795 } 796 797 #if IS_ENABLED(CONFIG_IPV6) 798 799 /* Sysctl settings for net ipv6.auto_flowlabels */ 800 #define IP6_AUTO_FLOW_LABEL_OFF 0 801 #define IP6_AUTO_FLOW_LABEL_OPTOUT 1 802 #define IP6_AUTO_FLOW_LABEL_OPTIN 2 803 #define IP6_AUTO_FLOW_LABEL_FORCED 3 804 805 #define IP6_AUTO_FLOW_LABEL_MAX IP6_AUTO_FLOW_LABEL_FORCED 806 807 #define IP6_DEFAULT_AUTO_FLOW_LABELS IP6_AUTO_FLOW_LABEL_OPTOUT 808 809 static inline __be32 ip6_make_flowlabel(struct net *net, struct sk_buff *skb, 810 __be32 flowlabel, bool autolabel, 811 struct flowi6 *fl6) 812 { 813 u32 hash; 814 815 /* @flowlabel may include more than a flow label, eg, the traffic class. 816 * Here we want only the flow label value. 817 */ 818 flowlabel &= IPV6_FLOWLABEL_MASK; 819 820 if (flowlabel || 821 net->ipv6.sysctl.auto_flowlabels == IP6_AUTO_FLOW_LABEL_OFF || 822 (!autolabel && 823 net->ipv6.sysctl.auto_flowlabels != IP6_AUTO_FLOW_LABEL_FORCED)) 824 return flowlabel; 825 826 hash = skb_get_hash_flowi6(skb, fl6); 827 828 /* Since this is being sent on the wire obfuscate hash a bit 829 * to minimize possbility that any useful information to an 830 * attacker is leaked. Only lower 20 bits are relevant. 831 */ 832 rol32(hash, 16); 833 834 flowlabel = (__force __be32)hash & IPV6_FLOWLABEL_MASK; 835 836 if (net->ipv6.sysctl.flowlabel_state_ranges) 837 flowlabel |= IPV6_FLOWLABEL_STATELESS_FLAG; 838 839 return flowlabel; 840 } 841 842 static inline int ip6_default_np_autolabel(struct net *net) 843 { 844 switch (net->ipv6.sysctl.auto_flowlabels) { 845 case IP6_AUTO_FLOW_LABEL_OFF: 846 case IP6_AUTO_FLOW_LABEL_OPTIN: 847 default: 848 return 0; 849 case IP6_AUTO_FLOW_LABEL_OPTOUT: 850 case IP6_AUTO_FLOW_LABEL_FORCED: 851 return 1; 852 } 853 } 854 #else 855 static inline void ip6_set_txhash(struct sock *sk) { } 856 static inline __be32 ip6_make_flowlabel(struct net *net, struct sk_buff *skb, 857 __be32 flowlabel, bool autolabel, 858 struct flowi6 *fl6) 859 { 860 return flowlabel; 861 } 862 static inline int ip6_default_np_autolabel(struct net *net) 863 { 864 return 0; 865 } 866 #endif 867 868 #if IS_ENABLED(CONFIG_IPV6) 869 static inline int ip6_multipath_hash_policy(const struct net *net) 870 { 871 return net->ipv6.sysctl.multipath_hash_policy; 872 } 873 #else 874 static inline int ip6_multipath_hash_policy(const struct net *net) 875 { 876 return 0; 877 } 878 #endif 879 880 /* 881 * Header manipulation 882 */ 883 static inline void ip6_flow_hdr(struct ipv6hdr *hdr, unsigned int tclass, 884 __be32 flowlabel) 885 { 886 *(__be32 *)hdr = htonl(0x60000000 | (tclass << 20)) | flowlabel; 887 } 888 889 static inline __be32 ip6_flowinfo(const struct ipv6hdr *hdr) 890 { 891 return *(__be32 *)hdr & IPV6_FLOWINFO_MASK; 892 } 893 894 static inline __be32 ip6_flowlabel(const struct ipv6hdr *hdr) 895 { 896 return *(__be32 *)hdr & IPV6_FLOWLABEL_MASK; 897 } 898 899 static inline u8 ip6_tclass(__be32 flowinfo) 900 { 901 return ntohl(flowinfo & IPV6_TCLASS_MASK) >> IPV6_TCLASS_SHIFT; 902 } 903 904 static inline __be32 ip6_make_flowinfo(unsigned int tclass, __be32 flowlabel) 905 { 906 return htonl(tclass << IPV6_TCLASS_SHIFT) | flowlabel; 907 } 908 909 /* 910 * Prototypes exported by ipv6 911 */ 912 913 /* 914 * rcv function (called from netdevice level) 915 */ 916 917 int ipv6_rcv(struct sk_buff *skb, struct net_device *dev, 918 struct packet_type *pt, struct net_device *orig_dev); 919 920 int ip6_rcv_finish(struct net *net, struct sock *sk, struct sk_buff *skb); 921 922 /* 923 * upper-layer output functions 924 */ 925 int ip6_xmit(const struct sock *sk, struct sk_buff *skb, struct flowi6 *fl6, 926 __u32 mark, struct ipv6_txoptions *opt, int tclass); 927 928 int ip6_find_1stfragopt(struct sk_buff *skb, u8 **nexthdr); 929 930 int ip6_append_data(struct sock *sk, 931 int getfrag(void *from, char *to, int offset, int len, 932 int odd, struct sk_buff *skb), 933 void *from, int length, int transhdrlen, 934 struct ipcm6_cookie *ipc6, struct flowi6 *fl6, 935 struct rt6_info *rt, unsigned int flags, 936 const struct sockcm_cookie *sockc); 937 938 int ip6_push_pending_frames(struct sock *sk); 939 940 void ip6_flush_pending_frames(struct sock *sk); 941 942 int ip6_send_skb(struct sk_buff *skb); 943 944 struct sk_buff *__ip6_make_skb(struct sock *sk, struct sk_buff_head *queue, 945 struct inet_cork_full *cork, 946 struct inet6_cork *v6_cork); 947 struct sk_buff *ip6_make_skb(struct sock *sk, 948 int getfrag(void *from, char *to, int offset, 949 int len, int odd, struct sk_buff *skb), 950 void *from, int length, int transhdrlen, 951 struct ipcm6_cookie *ipc6, struct flowi6 *fl6, 952 struct rt6_info *rt, unsigned int flags, 953 const struct sockcm_cookie *sockc); 954 955 static inline struct sk_buff *ip6_finish_skb(struct sock *sk) 956 { 957 return __ip6_make_skb(sk, &sk->sk_write_queue, &inet_sk(sk)->cork, 958 &inet6_sk(sk)->cork); 959 } 960 961 unsigned int ip6_dst_mtu_forward(const struct dst_entry *dst); 962 963 int ip6_dst_lookup(struct net *net, struct sock *sk, struct dst_entry **dst, 964 struct flowi6 *fl6); 965 struct dst_entry *ip6_dst_lookup_flow(const struct sock *sk, struct flowi6 *fl6, 966 const struct in6_addr *final_dst); 967 struct dst_entry *ip6_sk_dst_lookup_flow(struct sock *sk, struct flowi6 *fl6, 968 const struct in6_addr *final_dst, 969 bool connected); 970 struct dst_entry *ip6_blackhole_route(struct net *net, 971 struct dst_entry *orig_dst); 972 973 /* 974 * skb processing functions 975 */ 976 977 int ip6_output(struct net *net, struct sock *sk, struct sk_buff *skb); 978 int ip6_forward(struct sk_buff *skb); 979 int ip6_input(struct sk_buff *skb); 980 int ip6_mc_input(struct sk_buff *skb); 981 982 int __ip6_local_out(struct net *net, struct sock *sk, struct sk_buff *skb); 983 int ip6_local_out(struct net *net, struct sock *sk, struct sk_buff *skb); 984 985 /* 986 * Extension header (options) processing 987 */ 988 989 void ipv6_push_nfrag_opts(struct sk_buff *skb, struct ipv6_txoptions *opt, 990 u8 *proto, struct in6_addr **daddr_p, 991 struct in6_addr *saddr); 992 void ipv6_push_frag_opts(struct sk_buff *skb, struct ipv6_txoptions *opt, 993 u8 *proto); 994 995 int ipv6_skip_exthdr(const struct sk_buff *, int start, u8 *nexthdrp, 996 __be16 *frag_offp); 997 998 bool ipv6_ext_hdr(u8 nexthdr); 999 1000 enum { 1001 IP6_FH_F_FRAG = (1 << 0), 1002 IP6_FH_F_AUTH = (1 << 1), 1003 IP6_FH_F_SKIP_RH = (1 << 2), 1004 }; 1005 1006 /* find specified header and get offset to it */ 1007 int ipv6_find_hdr(const struct sk_buff *skb, unsigned int *offset, int target, 1008 unsigned short *fragoff, int *fragflg); 1009 1010 int ipv6_find_tlv(const struct sk_buff *skb, int offset, int type); 1011 1012 struct in6_addr *fl6_update_dst(struct flowi6 *fl6, 1013 const struct ipv6_txoptions *opt, 1014 struct in6_addr *orig); 1015 1016 /* 1017 * socket options (ipv6_sockglue.c) 1018 */ 1019 1020 int ipv6_setsockopt(struct sock *sk, int level, int optname, 1021 char __user *optval, unsigned int optlen); 1022 int ipv6_getsockopt(struct sock *sk, int level, int optname, 1023 char __user *optval, int __user *optlen); 1024 int compat_ipv6_setsockopt(struct sock *sk, int level, int optname, 1025 char __user *optval, unsigned int optlen); 1026 int compat_ipv6_getsockopt(struct sock *sk, int level, int optname, 1027 char __user *optval, int __user *optlen); 1028 1029 int __ip6_datagram_connect(struct sock *sk, struct sockaddr *addr, 1030 int addr_len); 1031 int ip6_datagram_connect(struct sock *sk, struct sockaddr *addr, int addr_len); 1032 int ip6_datagram_connect_v6_only(struct sock *sk, struct sockaddr *addr, 1033 int addr_len); 1034 int ip6_datagram_dst_update(struct sock *sk, bool fix_sk_saddr); 1035 void ip6_datagram_release_cb(struct sock *sk); 1036 1037 int ipv6_recv_error(struct sock *sk, struct msghdr *msg, int len, 1038 int *addr_len); 1039 int ipv6_recv_rxpmtu(struct sock *sk, struct msghdr *msg, int len, 1040 int *addr_len); 1041 void ipv6_icmp_error(struct sock *sk, struct sk_buff *skb, int err, __be16 port, 1042 u32 info, u8 *payload); 1043 void ipv6_local_error(struct sock *sk, int err, struct flowi6 *fl6, u32 info); 1044 void ipv6_local_rxpmtu(struct sock *sk, struct flowi6 *fl6, u32 mtu); 1045 1046 int inet6_release(struct socket *sock); 1047 int __inet6_bind(struct sock *sock, struct sockaddr *uaddr, int addr_len, 1048 bool force_bind_address_no_port, bool with_lock); 1049 int inet6_bind(struct socket *sock, struct sockaddr *uaddr, int addr_len); 1050 int inet6_getname(struct socket *sock, struct sockaddr *uaddr, 1051 int peer); 1052 int inet6_ioctl(struct socket *sock, unsigned int cmd, unsigned long arg); 1053 1054 int inet6_hash_connect(struct inet_timewait_death_row *death_row, 1055 struct sock *sk); 1056 1057 /* 1058 * reassembly.c 1059 */ 1060 extern const struct proto_ops inet6_stream_ops; 1061 extern const struct proto_ops inet6_dgram_ops; 1062 extern const struct proto_ops inet6_sockraw_ops; 1063 1064 struct group_source_req; 1065 struct group_filter; 1066 1067 int ip6_mc_source(int add, int omode, struct sock *sk, 1068 struct group_source_req *pgsr); 1069 int ip6_mc_msfilter(struct sock *sk, struct group_filter *gsf); 1070 int ip6_mc_msfget(struct sock *sk, struct group_filter *gsf, 1071 struct group_filter __user *optval, int __user *optlen); 1072 1073 #ifdef CONFIG_PROC_FS 1074 int ac6_proc_init(struct net *net); 1075 void ac6_proc_exit(struct net *net); 1076 int raw6_proc_init(void); 1077 void raw6_proc_exit(void); 1078 int tcp6_proc_init(struct net *net); 1079 void tcp6_proc_exit(struct net *net); 1080 int udp6_proc_init(struct net *net); 1081 void udp6_proc_exit(struct net *net); 1082 int udplite6_proc_init(void); 1083 void udplite6_proc_exit(void); 1084 int ipv6_misc_proc_init(void); 1085 void ipv6_misc_proc_exit(void); 1086 int snmp6_register_dev(struct inet6_dev *idev); 1087 int snmp6_unregister_dev(struct inet6_dev *idev); 1088 1089 #else 1090 static inline int ac6_proc_init(struct net *net) { return 0; } 1091 static inline void ac6_proc_exit(struct net *net) { } 1092 static inline int snmp6_register_dev(struct inet6_dev *idev) { return 0; } 1093 static inline int snmp6_unregister_dev(struct inet6_dev *idev) { return 0; } 1094 #endif 1095 1096 #ifdef CONFIG_SYSCTL 1097 extern struct ctl_table ipv6_route_table_template[]; 1098 1099 struct ctl_table *ipv6_icmp_sysctl_init(struct net *net); 1100 struct ctl_table *ipv6_route_sysctl_init(struct net *net); 1101 int ipv6_sysctl_register(void); 1102 void ipv6_sysctl_unregister(void); 1103 #endif 1104 1105 int ipv6_sock_mc_join(struct sock *sk, int ifindex, 1106 const struct in6_addr *addr); 1107 int ipv6_sock_mc_drop(struct sock *sk, int ifindex, 1108 const struct in6_addr *addr); 1109 #endif /* _NET_IPV6_H */ 1110