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 __u16 gso_size; 302 }; 303 304 static inline struct ipv6_txoptions *txopt_get(const struct ipv6_pinfo *np) 305 { 306 struct ipv6_txoptions *opt; 307 308 rcu_read_lock(); 309 opt = rcu_dereference(np->opt); 310 if (opt) { 311 if (!refcount_inc_not_zero(&opt->refcnt)) 312 opt = NULL; 313 else 314 opt = rcu_pointer_handoff(opt); 315 } 316 rcu_read_unlock(); 317 return opt; 318 } 319 320 static inline void txopt_put(struct ipv6_txoptions *opt) 321 { 322 if (opt && refcount_dec_and_test(&opt->refcnt)) 323 kfree_rcu(opt, rcu); 324 } 325 326 struct ip6_flowlabel *fl6_sock_lookup(struct sock *sk, __be32 label); 327 struct ipv6_txoptions *fl6_merge_options(struct ipv6_txoptions *opt_space, 328 struct ip6_flowlabel *fl, 329 struct ipv6_txoptions *fopt); 330 void fl6_free_socklist(struct sock *sk); 331 int ipv6_flowlabel_opt(struct sock *sk, char __user *optval, int optlen); 332 int ipv6_flowlabel_opt_get(struct sock *sk, struct in6_flowlabel_req *freq, 333 int flags); 334 int ip6_flowlabel_init(void); 335 void ip6_flowlabel_cleanup(void); 336 bool ip6_autoflowlabel(struct net *net, const struct ipv6_pinfo *np); 337 338 static inline void fl6_sock_release(struct ip6_flowlabel *fl) 339 { 340 if (fl) 341 atomic_dec(&fl->users); 342 } 343 344 void icmpv6_notify(struct sk_buff *skb, u8 type, u8 code, __be32 info); 345 346 void icmpv6_push_pending_frames(struct sock *sk, struct flowi6 *fl6, 347 struct icmp6hdr *thdr, int len); 348 349 int ip6_ra_control(struct sock *sk, int sel); 350 351 int ipv6_parse_hopopts(struct sk_buff *skb); 352 353 struct ipv6_txoptions *ipv6_dup_options(struct sock *sk, 354 struct ipv6_txoptions *opt); 355 struct ipv6_txoptions *ipv6_renew_options(struct sock *sk, 356 struct ipv6_txoptions *opt, 357 int newtype, 358 struct ipv6_opt_hdr *newopt); 359 struct ipv6_txoptions *ipv6_fixup_options(struct ipv6_txoptions *opt_space, 360 struct ipv6_txoptions *opt); 361 362 bool ipv6_opt_accepted(const struct sock *sk, const struct sk_buff *skb, 363 const struct inet6_skb_parm *opt); 364 struct ipv6_txoptions *ipv6_update_options(struct sock *sk, 365 struct ipv6_txoptions *opt); 366 367 static inline bool ipv6_accept_ra(struct inet6_dev *idev) 368 { 369 /* If forwarding is enabled, RA are not accepted unless the special 370 * hybrid mode (accept_ra=2) is enabled. 371 */ 372 return idev->cnf.forwarding ? idev->cnf.accept_ra == 2 : 373 idev->cnf.accept_ra; 374 } 375 376 #define IPV6_FRAG_HIGH_THRESH (4 * 1024*1024) /* 4194304 */ 377 #define IPV6_FRAG_LOW_THRESH (3 * 1024*1024) /* 3145728 */ 378 #define IPV6_FRAG_TIMEOUT (60 * HZ) /* 60 seconds */ 379 380 int __ipv6_addr_type(const struct in6_addr *addr); 381 static inline int ipv6_addr_type(const struct in6_addr *addr) 382 { 383 return __ipv6_addr_type(addr) & 0xffff; 384 } 385 386 static inline int ipv6_addr_scope(const struct in6_addr *addr) 387 { 388 return __ipv6_addr_type(addr) & IPV6_ADDR_SCOPE_MASK; 389 } 390 391 static inline int __ipv6_addr_src_scope(int type) 392 { 393 return (type == IPV6_ADDR_ANY) ? __IPV6_ADDR_SCOPE_INVALID : (type >> 16); 394 } 395 396 static inline int ipv6_addr_src_scope(const struct in6_addr *addr) 397 { 398 return __ipv6_addr_src_scope(__ipv6_addr_type(addr)); 399 } 400 401 static inline bool __ipv6_addr_needs_scope_id(int type) 402 { 403 return type & IPV6_ADDR_LINKLOCAL || 404 (type & IPV6_ADDR_MULTICAST && 405 (type & (IPV6_ADDR_LOOPBACK|IPV6_ADDR_LINKLOCAL))); 406 } 407 408 static inline __u32 ipv6_iface_scope_id(const struct in6_addr *addr, int iface) 409 { 410 return __ipv6_addr_needs_scope_id(__ipv6_addr_type(addr)) ? iface : 0; 411 } 412 413 static inline int ipv6_addr_cmp(const struct in6_addr *a1, const struct in6_addr *a2) 414 { 415 return memcmp(a1, a2, sizeof(struct in6_addr)); 416 } 417 418 static inline bool 419 ipv6_masked_addr_cmp(const struct in6_addr *a1, const struct in6_addr *m, 420 const struct in6_addr *a2) 421 { 422 #if defined(CONFIG_HAVE_EFFICIENT_UNALIGNED_ACCESS) && BITS_PER_LONG == 64 423 const unsigned long *ul1 = (const unsigned long *)a1; 424 const unsigned long *ulm = (const unsigned long *)m; 425 const unsigned long *ul2 = (const unsigned long *)a2; 426 427 return !!(((ul1[0] ^ ul2[0]) & ulm[0]) | 428 ((ul1[1] ^ ul2[1]) & ulm[1])); 429 #else 430 return !!(((a1->s6_addr32[0] ^ a2->s6_addr32[0]) & m->s6_addr32[0]) | 431 ((a1->s6_addr32[1] ^ a2->s6_addr32[1]) & m->s6_addr32[1]) | 432 ((a1->s6_addr32[2] ^ a2->s6_addr32[2]) & m->s6_addr32[2]) | 433 ((a1->s6_addr32[3] ^ a2->s6_addr32[3]) & m->s6_addr32[3])); 434 #endif 435 } 436 437 static inline void ipv6_addr_prefix(struct in6_addr *pfx, 438 const struct in6_addr *addr, 439 int plen) 440 { 441 /* caller must guarantee 0 <= plen <= 128 */ 442 int o = plen >> 3, 443 b = plen & 0x7; 444 445 memset(pfx->s6_addr, 0, sizeof(pfx->s6_addr)); 446 memcpy(pfx->s6_addr, addr, o); 447 if (b != 0) 448 pfx->s6_addr[o] = addr->s6_addr[o] & (0xff00 >> b); 449 } 450 451 static inline void ipv6_addr_prefix_copy(struct in6_addr *addr, 452 const struct in6_addr *pfx, 453 int plen) 454 { 455 /* caller must guarantee 0 <= plen <= 128 */ 456 int o = plen >> 3, 457 b = plen & 0x7; 458 459 memcpy(addr->s6_addr, pfx, o); 460 if (b != 0) { 461 addr->s6_addr[o] &= ~(0xff00 >> b); 462 addr->s6_addr[o] |= (pfx->s6_addr[o] & (0xff00 >> b)); 463 } 464 } 465 466 static inline void __ipv6_addr_set_half(__be32 *addr, 467 __be32 wh, __be32 wl) 468 { 469 #if defined(CONFIG_HAVE_EFFICIENT_UNALIGNED_ACCESS) && BITS_PER_LONG == 64 470 #if defined(__BIG_ENDIAN) 471 if (__builtin_constant_p(wh) && __builtin_constant_p(wl)) { 472 *(__force u64 *)addr = ((__force u64)(wh) << 32 | (__force u64)(wl)); 473 return; 474 } 475 #elif defined(__LITTLE_ENDIAN) 476 if (__builtin_constant_p(wl) && __builtin_constant_p(wh)) { 477 *(__force u64 *)addr = ((__force u64)(wl) << 32 | (__force u64)(wh)); 478 return; 479 } 480 #endif 481 #endif 482 addr[0] = wh; 483 addr[1] = wl; 484 } 485 486 static inline void ipv6_addr_set(struct in6_addr *addr, 487 __be32 w1, __be32 w2, 488 __be32 w3, __be32 w4) 489 { 490 __ipv6_addr_set_half(&addr->s6_addr32[0], w1, w2); 491 __ipv6_addr_set_half(&addr->s6_addr32[2], w3, w4); 492 } 493 494 static inline bool ipv6_addr_equal(const struct in6_addr *a1, 495 const struct in6_addr *a2) 496 { 497 #if defined(CONFIG_HAVE_EFFICIENT_UNALIGNED_ACCESS) && BITS_PER_LONG == 64 498 const unsigned long *ul1 = (const unsigned long *)a1; 499 const unsigned long *ul2 = (const unsigned long *)a2; 500 501 return ((ul1[0] ^ ul2[0]) | (ul1[1] ^ ul2[1])) == 0UL; 502 #else 503 return ((a1->s6_addr32[0] ^ a2->s6_addr32[0]) | 504 (a1->s6_addr32[1] ^ a2->s6_addr32[1]) | 505 (a1->s6_addr32[2] ^ a2->s6_addr32[2]) | 506 (a1->s6_addr32[3] ^ a2->s6_addr32[3])) == 0; 507 #endif 508 } 509 510 #if defined(CONFIG_HAVE_EFFICIENT_UNALIGNED_ACCESS) && BITS_PER_LONG == 64 511 static inline bool __ipv6_prefix_equal64_half(const __be64 *a1, 512 const __be64 *a2, 513 unsigned int len) 514 { 515 if (len && ((*a1 ^ *a2) & cpu_to_be64((~0UL) << (64 - len)))) 516 return false; 517 return true; 518 } 519 520 static inline bool ipv6_prefix_equal(const struct in6_addr *addr1, 521 const struct in6_addr *addr2, 522 unsigned int prefixlen) 523 { 524 const __be64 *a1 = (const __be64 *)addr1; 525 const __be64 *a2 = (const __be64 *)addr2; 526 527 if (prefixlen >= 64) { 528 if (a1[0] ^ a2[0]) 529 return false; 530 return __ipv6_prefix_equal64_half(a1 + 1, a2 + 1, prefixlen - 64); 531 } 532 return __ipv6_prefix_equal64_half(a1, a2, prefixlen); 533 } 534 #else 535 static inline bool ipv6_prefix_equal(const struct in6_addr *addr1, 536 const struct in6_addr *addr2, 537 unsigned int prefixlen) 538 { 539 const __be32 *a1 = addr1->s6_addr32; 540 const __be32 *a2 = addr2->s6_addr32; 541 unsigned int pdw, pbi; 542 543 /* check complete u32 in prefix */ 544 pdw = prefixlen >> 5; 545 if (pdw && memcmp(a1, a2, pdw << 2)) 546 return false; 547 548 /* check incomplete u32 in prefix */ 549 pbi = prefixlen & 0x1f; 550 if (pbi && ((a1[pdw] ^ a2[pdw]) & htonl((0xffffffff) << (32 - pbi)))) 551 return false; 552 553 return true; 554 } 555 #endif 556 557 struct inet_frag_queue; 558 559 enum ip6_defrag_users { 560 IP6_DEFRAG_LOCAL_DELIVER, 561 IP6_DEFRAG_CONNTRACK_IN, 562 __IP6_DEFRAG_CONNTRACK_IN = IP6_DEFRAG_CONNTRACK_IN + USHRT_MAX, 563 IP6_DEFRAG_CONNTRACK_OUT, 564 __IP6_DEFRAG_CONNTRACK_OUT = IP6_DEFRAG_CONNTRACK_OUT + USHRT_MAX, 565 IP6_DEFRAG_CONNTRACK_BRIDGE_IN, 566 __IP6_DEFRAG_CONNTRACK_BRIDGE_IN = IP6_DEFRAG_CONNTRACK_BRIDGE_IN + USHRT_MAX, 567 }; 568 569 void ip6_frag_init(struct inet_frag_queue *q, const void *a); 570 extern const struct rhashtable_params ip6_rhash_params; 571 572 /* 573 * Equivalent of ipv4 struct ip 574 */ 575 struct frag_queue { 576 struct inet_frag_queue q; 577 578 int iif; 579 __u16 nhoffset; 580 u8 ecn; 581 }; 582 583 void ip6_expire_frag_queue(struct net *net, struct frag_queue *fq); 584 585 static inline bool ipv6_addr_any(const struct in6_addr *a) 586 { 587 #if defined(CONFIG_HAVE_EFFICIENT_UNALIGNED_ACCESS) && BITS_PER_LONG == 64 588 const unsigned long *ul = (const unsigned long *)a; 589 590 return (ul[0] | ul[1]) == 0UL; 591 #else 592 return (a->s6_addr32[0] | a->s6_addr32[1] | 593 a->s6_addr32[2] | a->s6_addr32[3]) == 0; 594 #endif 595 } 596 597 static inline u32 ipv6_addr_hash(const struct in6_addr *a) 598 { 599 #if defined(CONFIG_HAVE_EFFICIENT_UNALIGNED_ACCESS) && BITS_PER_LONG == 64 600 const unsigned long *ul = (const unsigned long *)a; 601 unsigned long x = ul[0] ^ ul[1]; 602 603 return (u32)(x ^ (x >> 32)); 604 #else 605 return (__force u32)(a->s6_addr32[0] ^ a->s6_addr32[1] ^ 606 a->s6_addr32[2] ^ a->s6_addr32[3]); 607 #endif 608 } 609 610 /* more secured version of ipv6_addr_hash() */ 611 static inline u32 __ipv6_addr_jhash(const struct in6_addr *a, const u32 initval) 612 { 613 u32 v = (__force u32)a->s6_addr32[0] ^ (__force u32)a->s6_addr32[1]; 614 615 return jhash_3words(v, 616 (__force u32)a->s6_addr32[2], 617 (__force u32)a->s6_addr32[3], 618 initval); 619 } 620 621 static inline bool ipv6_addr_loopback(const struct in6_addr *a) 622 { 623 #if defined(CONFIG_HAVE_EFFICIENT_UNALIGNED_ACCESS) && BITS_PER_LONG == 64 624 const __be64 *be = (const __be64 *)a; 625 626 return (be[0] | (be[1] ^ cpu_to_be64(1))) == 0UL; 627 #else 628 return (a->s6_addr32[0] | a->s6_addr32[1] | 629 a->s6_addr32[2] | (a->s6_addr32[3] ^ cpu_to_be32(1))) == 0; 630 #endif 631 } 632 633 /* 634 * Note that we must __force cast these to unsigned long to make sparse happy, 635 * since all of the endian-annotated types are fixed size regardless of arch. 636 */ 637 static inline bool ipv6_addr_v4mapped(const struct in6_addr *a) 638 { 639 return ( 640 #if defined(CONFIG_HAVE_EFFICIENT_UNALIGNED_ACCESS) && BITS_PER_LONG == 64 641 *(unsigned long *)a | 642 #else 643 (__force unsigned long)(a->s6_addr32[0] | a->s6_addr32[1]) | 644 #endif 645 (__force unsigned long)(a->s6_addr32[2] ^ 646 cpu_to_be32(0x0000ffff))) == 0UL; 647 } 648 649 static inline u32 ipv6_portaddr_hash(const struct net *net, 650 const struct in6_addr *addr6, 651 unsigned int port) 652 { 653 unsigned int hash, mix = net_hash_mix(net); 654 655 if (ipv6_addr_any(addr6)) 656 hash = jhash_1word(0, mix); 657 else if (ipv6_addr_v4mapped(addr6)) 658 hash = jhash_1word((__force u32)addr6->s6_addr32[3], mix); 659 else 660 hash = jhash2((__force u32 *)addr6->s6_addr32, 4, mix); 661 662 return hash ^ port; 663 } 664 665 /* 666 * Check for a RFC 4843 ORCHID address 667 * (Overlay Routable Cryptographic Hash Identifiers) 668 */ 669 static inline bool ipv6_addr_orchid(const struct in6_addr *a) 670 { 671 return (a->s6_addr32[0] & htonl(0xfffffff0)) == htonl(0x20010010); 672 } 673 674 static inline bool ipv6_addr_is_multicast(const struct in6_addr *addr) 675 { 676 return (addr->s6_addr32[0] & htonl(0xFF000000)) == htonl(0xFF000000); 677 } 678 679 static inline void ipv6_addr_set_v4mapped(const __be32 addr, 680 struct in6_addr *v4mapped) 681 { 682 ipv6_addr_set(v4mapped, 683 0, 0, 684 htonl(0x0000FFFF), 685 addr); 686 } 687 688 /* 689 * find the first different bit between two addresses 690 * length of address must be a multiple of 32bits 691 */ 692 static inline int __ipv6_addr_diff32(const void *token1, const void *token2, int addrlen) 693 { 694 const __be32 *a1 = token1, *a2 = token2; 695 int i; 696 697 addrlen >>= 2; 698 699 for (i = 0; i < addrlen; i++) { 700 __be32 xb = a1[i] ^ a2[i]; 701 if (xb) 702 return i * 32 + 31 - __fls(ntohl(xb)); 703 } 704 705 /* 706 * we should *never* get to this point since that 707 * would mean the addrs are equal 708 * 709 * However, we do get to it 8) And exacly, when 710 * addresses are equal 8) 711 * 712 * ip route add 1111::/128 via ... 713 * ip route add 1111::/64 via ... 714 * and we are here. 715 * 716 * Ideally, this function should stop comparison 717 * at prefix length. It does not, but it is still OK, 718 * if returned value is greater than prefix length. 719 * --ANK (980803) 720 */ 721 return addrlen << 5; 722 } 723 724 #if defined(CONFIG_HAVE_EFFICIENT_UNALIGNED_ACCESS) && BITS_PER_LONG == 64 725 static inline int __ipv6_addr_diff64(const void *token1, const void *token2, int addrlen) 726 { 727 const __be64 *a1 = token1, *a2 = token2; 728 int i; 729 730 addrlen >>= 3; 731 732 for (i = 0; i < addrlen; i++) { 733 __be64 xb = a1[i] ^ a2[i]; 734 if (xb) 735 return i * 64 + 63 - __fls(be64_to_cpu(xb)); 736 } 737 738 return addrlen << 6; 739 } 740 #endif 741 742 static inline int __ipv6_addr_diff(const void *token1, const void *token2, int addrlen) 743 { 744 #if defined(CONFIG_HAVE_EFFICIENT_UNALIGNED_ACCESS) && BITS_PER_LONG == 64 745 if (__builtin_constant_p(addrlen) && !(addrlen & 7)) 746 return __ipv6_addr_diff64(token1, token2, addrlen); 747 #endif 748 return __ipv6_addr_diff32(token1, token2, addrlen); 749 } 750 751 static inline int ipv6_addr_diff(const struct in6_addr *a1, const struct in6_addr *a2) 752 { 753 return __ipv6_addr_diff(a1, a2, sizeof(struct in6_addr)); 754 } 755 756 __be32 ipv6_select_ident(struct net *net, 757 const struct in6_addr *daddr, 758 const struct in6_addr *saddr); 759 __be32 ipv6_proxy_select_ident(struct net *net, struct sk_buff *skb); 760 761 int ip6_dst_hoplimit(struct dst_entry *dst); 762 763 static inline int ip6_sk_dst_hoplimit(struct ipv6_pinfo *np, struct flowi6 *fl6, 764 struct dst_entry *dst) 765 { 766 int hlimit; 767 768 if (ipv6_addr_is_multicast(&fl6->daddr)) 769 hlimit = np->mcast_hops; 770 else 771 hlimit = np->hop_limit; 772 if (hlimit < 0) 773 hlimit = ip6_dst_hoplimit(dst); 774 return hlimit; 775 } 776 777 /* copy IPv6 saddr & daddr to flow_keys, possibly using 64bit load/store 778 * Equivalent to : flow->v6addrs.src = iph->saddr; 779 * flow->v6addrs.dst = iph->daddr; 780 */ 781 static inline void iph_to_flow_copy_v6addrs(struct flow_keys *flow, 782 const struct ipv6hdr *iph) 783 { 784 BUILD_BUG_ON(offsetof(typeof(flow->addrs), v6addrs.dst) != 785 offsetof(typeof(flow->addrs), v6addrs.src) + 786 sizeof(flow->addrs.v6addrs.src)); 787 memcpy(&flow->addrs.v6addrs, &iph->saddr, sizeof(flow->addrs.v6addrs)); 788 flow->control.addr_type = FLOW_DISSECTOR_KEY_IPV6_ADDRS; 789 } 790 791 #if IS_ENABLED(CONFIG_IPV6) 792 793 /* Sysctl settings for net ipv6.auto_flowlabels */ 794 #define IP6_AUTO_FLOW_LABEL_OFF 0 795 #define IP6_AUTO_FLOW_LABEL_OPTOUT 1 796 #define IP6_AUTO_FLOW_LABEL_OPTIN 2 797 #define IP6_AUTO_FLOW_LABEL_FORCED 3 798 799 #define IP6_AUTO_FLOW_LABEL_MAX IP6_AUTO_FLOW_LABEL_FORCED 800 801 #define IP6_DEFAULT_AUTO_FLOW_LABELS IP6_AUTO_FLOW_LABEL_OPTOUT 802 803 static inline __be32 ip6_make_flowlabel(struct net *net, struct sk_buff *skb, 804 __be32 flowlabel, bool autolabel, 805 struct flowi6 *fl6) 806 { 807 u32 hash; 808 809 /* @flowlabel may include more than a flow label, eg, the traffic class. 810 * Here we want only the flow label value. 811 */ 812 flowlabel &= IPV6_FLOWLABEL_MASK; 813 814 if (flowlabel || 815 net->ipv6.sysctl.auto_flowlabels == IP6_AUTO_FLOW_LABEL_OFF || 816 (!autolabel && 817 net->ipv6.sysctl.auto_flowlabels != IP6_AUTO_FLOW_LABEL_FORCED)) 818 return flowlabel; 819 820 hash = skb_get_hash_flowi6(skb, fl6); 821 822 /* Since this is being sent on the wire obfuscate hash a bit 823 * to minimize possbility that any useful information to an 824 * attacker is leaked. Only lower 20 bits are relevant. 825 */ 826 hash = rol32(hash, 16); 827 828 flowlabel = (__force __be32)hash & IPV6_FLOWLABEL_MASK; 829 830 if (net->ipv6.sysctl.flowlabel_state_ranges) 831 flowlabel |= IPV6_FLOWLABEL_STATELESS_FLAG; 832 833 return flowlabel; 834 } 835 836 static inline int ip6_default_np_autolabel(struct net *net) 837 { 838 switch (net->ipv6.sysctl.auto_flowlabels) { 839 case IP6_AUTO_FLOW_LABEL_OFF: 840 case IP6_AUTO_FLOW_LABEL_OPTIN: 841 default: 842 return 0; 843 case IP6_AUTO_FLOW_LABEL_OPTOUT: 844 case IP6_AUTO_FLOW_LABEL_FORCED: 845 return 1; 846 } 847 } 848 #else 849 static inline void ip6_set_txhash(struct sock *sk) { } 850 static inline __be32 ip6_make_flowlabel(struct net *net, struct sk_buff *skb, 851 __be32 flowlabel, bool autolabel, 852 struct flowi6 *fl6) 853 { 854 return flowlabel; 855 } 856 static inline int ip6_default_np_autolabel(struct net *net) 857 { 858 return 0; 859 } 860 #endif 861 862 #if IS_ENABLED(CONFIG_IPV6) 863 static inline int ip6_multipath_hash_policy(const struct net *net) 864 { 865 return net->ipv6.sysctl.multipath_hash_policy; 866 } 867 #else 868 static inline int ip6_multipath_hash_policy(const struct net *net) 869 { 870 return 0; 871 } 872 #endif 873 874 /* 875 * Header manipulation 876 */ 877 static inline void ip6_flow_hdr(struct ipv6hdr *hdr, unsigned int tclass, 878 __be32 flowlabel) 879 { 880 *(__be32 *)hdr = htonl(0x60000000 | (tclass << 20)) | flowlabel; 881 } 882 883 static inline __be32 ip6_flowinfo(const struct ipv6hdr *hdr) 884 { 885 return *(__be32 *)hdr & IPV6_FLOWINFO_MASK; 886 } 887 888 static inline __be32 ip6_flowlabel(const struct ipv6hdr *hdr) 889 { 890 return *(__be32 *)hdr & IPV6_FLOWLABEL_MASK; 891 } 892 893 static inline u8 ip6_tclass(__be32 flowinfo) 894 { 895 return ntohl(flowinfo & IPV6_TCLASS_MASK) >> IPV6_TCLASS_SHIFT; 896 } 897 898 static inline __be32 ip6_make_flowinfo(unsigned int tclass, __be32 flowlabel) 899 { 900 return htonl(tclass << IPV6_TCLASS_SHIFT) | flowlabel; 901 } 902 903 static inline __be32 flowi6_get_flowlabel(const struct flowi6 *fl6) 904 { 905 return fl6->flowlabel & IPV6_FLOWLABEL_MASK; 906 } 907 908 /* 909 * Prototypes exported by ipv6 910 */ 911 912 /* 913 * rcv function (called from netdevice level) 914 */ 915 916 int ipv6_rcv(struct sk_buff *skb, struct net_device *dev, 917 struct packet_type *pt, struct net_device *orig_dev); 918 919 int ip6_rcv_finish(struct net *net, struct sock *sk, struct sk_buff *skb); 920 921 /* 922 * upper-layer output functions 923 */ 924 int ip6_xmit(const struct sock *sk, struct sk_buff *skb, struct flowi6 *fl6, 925 __u32 mark, struct ipv6_txoptions *opt, int tclass); 926 927 int ip6_find_1stfragopt(struct sk_buff *skb, u8 **nexthdr); 928 929 int ip6_append_data(struct sock *sk, 930 int getfrag(void *from, char *to, int offset, int len, 931 int odd, struct sk_buff *skb), 932 void *from, int length, int transhdrlen, 933 struct ipcm6_cookie *ipc6, struct flowi6 *fl6, 934 struct rt6_info *rt, unsigned int flags, 935 const struct sockcm_cookie *sockc); 936 937 int ip6_push_pending_frames(struct sock *sk); 938 939 void ip6_flush_pending_frames(struct sock *sk); 940 941 int ip6_send_skb(struct sk_buff *skb); 942 943 struct sk_buff *__ip6_make_skb(struct sock *sk, struct sk_buff_head *queue, 944 struct inet_cork_full *cork, 945 struct inet6_cork *v6_cork); 946 struct sk_buff *ip6_make_skb(struct sock *sk, 947 int getfrag(void *from, char *to, int offset, 948 int len, int odd, struct sk_buff *skb), 949 void *from, int length, int transhdrlen, 950 struct ipcm6_cookie *ipc6, struct flowi6 *fl6, 951 struct rt6_info *rt, unsigned int flags, 952 struct inet_cork_full *cork, 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 int ip6_dst_lookup(struct net *net, struct sock *sk, struct dst_entry **dst, 962 struct flowi6 *fl6); 963 struct dst_entry *ip6_dst_lookup_flow(const struct sock *sk, struct flowi6 *fl6, 964 const struct in6_addr *final_dst); 965 struct dst_entry *ip6_sk_dst_lookup_flow(struct sock *sk, struct flowi6 *fl6, 966 const struct in6_addr *final_dst, 967 bool connected); 968 struct dst_entry *ip6_blackhole_route(struct net *net, 969 struct dst_entry *orig_dst); 970 971 /* 972 * skb processing functions 973 */ 974 975 int ip6_output(struct net *net, struct sock *sk, struct sk_buff *skb); 976 int ip6_forward(struct sk_buff *skb); 977 int ip6_input(struct sk_buff *skb); 978 int ip6_mc_input(struct sk_buff *skb); 979 980 int __ip6_local_out(struct net *net, struct sock *sk, struct sk_buff *skb); 981 int ip6_local_out(struct net *net, struct sock *sk, struct sk_buff *skb); 982 983 /* 984 * Extension header (options) processing 985 */ 986 987 void ipv6_push_nfrag_opts(struct sk_buff *skb, struct ipv6_txoptions *opt, 988 u8 *proto, struct in6_addr **daddr_p, 989 struct in6_addr *saddr); 990 void ipv6_push_frag_opts(struct sk_buff *skb, struct ipv6_txoptions *opt, 991 u8 *proto); 992 993 int ipv6_skip_exthdr(const struct sk_buff *, int start, u8 *nexthdrp, 994 __be16 *frag_offp); 995 996 bool ipv6_ext_hdr(u8 nexthdr); 997 998 enum { 999 IP6_FH_F_FRAG = (1 << 0), 1000 IP6_FH_F_AUTH = (1 << 1), 1001 IP6_FH_F_SKIP_RH = (1 << 2), 1002 }; 1003 1004 /* find specified header and get offset to it */ 1005 int ipv6_find_hdr(const struct sk_buff *skb, unsigned int *offset, int target, 1006 unsigned short *fragoff, int *fragflg); 1007 1008 int ipv6_find_tlv(const struct sk_buff *skb, int offset, int type); 1009 1010 struct in6_addr *fl6_update_dst(struct flowi6 *fl6, 1011 const struct ipv6_txoptions *opt, 1012 struct in6_addr *orig); 1013 1014 /* 1015 * socket options (ipv6_sockglue.c) 1016 */ 1017 1018 int ipv6_setsockopt(struct sock *sk, int level, int optname, 1019 char __user *optval, unsigned int optlen); 1020 int ipv6_getsockopt(struct sock *sk, int level, int optname, 1021 char __user *optval, int __user *optlen); 1022 int compat_ipv6_setsockopt(struct sock *sk, int level, int optname, 1023 char __user *optval, unsigned int optlen); 1024 int compat_ipv6_getsockopt(struct sock *sk, int level, int optname, 1025 char __user *optval, int __user *optlen); 1026 1027 int __ip6_datagram_connect(struct sock *sk, struct sockaddr *addr, 1028 int addr_len); 1029 int ip6_datagram_connect(struct sock *sk, struct sockaddr *addr, int addr_len); 1030 int ip6_datagram_connect_v6_only(struct sock *sk, struct sockaddr *addr, 1031 int addr_len); 1032 int ip6_datagram_dst_update(struct sock *sk, bool fix_sk_saddr); 1033 void ip6_datagram_release_cb(struct sock *sk); 1034 1035 int ipv6_recv_error(struct sock *sk, struct msghdr *msg, int len, 1036 int *addr_len); 1037 int ipv6_recv_rxpmtu(struct sock *sk, struct msghdr *msg, int len, 1038 int *addr_len); 1039 void ipv6_icmp_error(struct sock *sk, struct sk_buff *skb, int err, __be16 port, 1040 u32 info, u8 *payload); 1041 void ipv6_local_error(struct sock *sk, int err, struct flowi6 *fl6, u32 info); 1042 void ipv6_local_rxpmtu(struct sock *sk, struct flowi6 *fl6, u32 mtu); 1043 1044 int inet6_release(struct socket *sock); 1045 int inet6_bind(struct socket *sock, struct sockaddr *uaddr, int addr_len); 1046 int inet6_getname(struct socket *sock, struct sockaddr *uaddr, 1047 int peer); 1048 int inet6_ioctl(struct socket *sock, unsigned int cmd, unsigned long arg); 1049 1050 int inet6_hash_connect(struct inet_timewait_death_row *death_row, 1051 struct sock *sk); 1052 1053 /* 1054 * reassembly.c 1055 */ 1056 extern const struct proto_ops inet6_stream_ops; 1057 extern const struct proto_ops inet6_dgram_ops; 1058 extern const struct proto_ops inet6_sockraw_ops; 1059 1060 struct group_source_req; 1061 struct group_filter; 1062 1063 int ip6_mc_source(int add, int omode, struct sock *sk, 1064 struct group_source_req *pgsr); 1065 int ip6_mc_msfilter(struct sock *sk, struct group_filter *gsf); 1066 int ip6_mc_msfget(struct sock *sk, struct group_filter *gsf, 1067 struct group_filter __user *optval, int __user *optlen); 1068 1069 #ifdef CONFIG_PROC_FS 1070 int ac6_proc_init(struct net *net); 1071 void ac6_proc_exit(struct net *net); 1072 int raw6_proc_init(void); 1073 void raw6_proc_exit(void); 1074 int tcp6_proc_init(struct net *net); 1075 void tcp6_proc_exit(struct net *net); 1076 int udp6_proc_init(struct net *net); 1077 void udp6_proc_exit(struct net *net); 1078 int udplite6_proc_init(void); 1079 void udplite6_proc_exit(void); 1080 int ipv6_misc_proc_init(void); 1081 void ipv6_misc_proc_exit(void); 1082 int snmp6_register_dev(struct inet6_dev *idev); 1083 int snmp6_unregister_dev(struct inet6_dev *idev); 1084 1085 #else 1086 static inline int ac6_proc_init(struct net *net) { return 0; } 1087 static inline void ac6_proc_exit(struct net *net) { } 1088 static inline int snmp6_register_dev(struct inet6_dev *idev) { return 0; } 1089 static inline int snmp6_unregister_dev(struct inet6_dev *idev) { return 0; } 1090 #endif 1091 1092 #ifdef CONFIG_SYSCTL 1093 extern struct ctl_table ipv6_route_table_template[]; 1094 1095 struct ctl_table *ipv6_icmp_sysctl_init(struct net *net); 1096 struct ctl_table *ipv6_route_sysctl_init(struct net *net); 1097 int ipv6_sysctl_register(void); 1098 void ipv6_sysctl_unregister(void); 1099 #endif 1100 1101 int ipv6_sock_mc_join(struct sock *sk, int ifindex, 1102 const struct in6_addr *addr); 1103 int ipv6_sock_mc_join_ssm(struct sock *sk, int ifindex, 1104 const struct in6_addr *addr, unsigned int mode); 1105 int ipv6_sock_mc_drop(struct sock *sk, int ifindex, 1106 const struct in6_addr *addr); 1107 #endif /* _NET_IPV6_H */ 1108