1 /* SPDX-License-Identifier: GPL-2.0-or-later */
3 * Linux INET6 implementation
6 * Pedro Roque <roque@di.fc.ul.pt>
12 #include <linux/ipv6.h>
13 #include <linux/hardirq.h>
14 #include <linux/jhash.h>
15 #include <linux/refcount.h>
16 #include <linux/jump_label_ratelimit.h>
17 #include <net/if_inet6.h>
18 #include <net/ndisc.h>
20 #include <net/flow_dissector.h>
22 #include <net/netns/hash.h>
24 #define SIN6_LEN_RFC2133 24
26 #define IPV6_MAXPLEN 65535
29 * NextHeader field of IPv6 header
32 #define NEXTHDR_HOP 0 /* Hop-by-hop option header. */
33 #define NEXTHDR_TCP 6 /* TCP segment. */
34 #define NEXTHDR_UDP 17 /* UDP message. */
35 #define NEXTHDR_IPV6 41 /* IPv6 in IPv6 */
36 #define NEXTHDR_ROUTING 43 /* Routing header. */
37 #define NEXTHDR_FRAGMENT 44 /* Fragmentation/reassembly header. */
38 #define NEXTHDR_GRE 47 /* GRE header. */
39 #define NEXTHDR_ESP 50 /* Encapsulating security payload. */
40 #define NEXTHDR_AUTH 51 /* Authentication header. */
41 #define NEXTHDR_ICMP 58 /* ICMP for IPv6. */
42 #define NEXTHDR_NONE 59 /* No next header */
43 #define NEXTHDR_DEST 60 /* Destination options header. */
44 #define NEXTHDR_SCTP 132 /* SCTP message. */
45 #define NEXTHDR_MOBILITY 135 /* Mobility header. */
47 #define NEXTHDR_MAX 255
49 #define IPV6_DEFAULT_HOPLIMIT 64
50 #define IPV6_DEFAULT_MCASTHOPS 1
52 /* Limits on Hop-by-Hop and Destination options.
54 * Per RFC8200 there is no limit on the maximum number or lengths of options in
55 * Hop-by-Hop or Destination options other then the packet must fit in an MTU.
56 * We allow configurable limits in order to mitigate potential denial of
59 * There are three limits that may be set:
60 * - Limit the number of options in a Hop-by-Hop or Destination options
62 * - Limit the byte length of a Hop-by-Hop or Destination options extension
64 * - Disallow unknown options
66 * The limits are expressed in corresponding sysctls:
68 * ipv6.sysctl.max_dst_opts_cnt
69 * ipv6.sysctl.max_hbh_opts_cnt
70 * ipv6.sysctl.max_dst_opts_len
71 * ipv6.sysctl.max_hbh_opts_len
73 * max_*_opts_cnt is the number of TLVs that are allowed for Destination
74 * options or Hop-by-Hop options. If the number is less than zero then unknown
75 * TLVs are disallowed and the number of known options that are allowed is the
76 * absolute value. Setting the value to INT_MAX indicates no limit.
78 * max_*_opts_len is the length limit in bytes of a Destination or
79 * Hop-by-Hop options extension header. Setting the value to INT_MAX
80 * indicates no length limit.
82 * If a limit is exceeded when processing an extension header the packet is
86 /* Default limits for Hop-by-Hop and Destination options */
87 #define IP6_DEFAULT_MAX_DST_OPTS_CNT 8
88 #define IP6_DEFAULT_MAX_HBH_OPTS_CNT 8
89 #define IP6_DEFAULT_MAX_DST_OPTS_LEN INT_MAX /* No limit */
90 #define IP6_DEFAULT_MAX_HBH_OPTS_LEN INT_MAX /* No limit */
95 * type - unicast | multicast
96 * scope - local | site | global
103 #define IPV6_ADDR_ANY 0x0000U
105 #define IPV6_ADDR_UNICAST 0x0001U
106 #define IPV6_ADDR_MULTICAST 0x0002U
108 #define IPV6_ADDR_LOOPBACK 0x0010U
109 #define IPV6_ADDR_LINKLOCAL 0x0020U
110 #define IPV6_ADDR_SITELOCAL 0x0040U
112 #define IPV6_ADDR_COMPATv4 0x0080U
114 #define IPV6_ADDR_SCOPE_MASK 0x00f0U
116 #define IPV6_ADDR_MAPPED 0x1000U
121 #define IPV6_ADDR_MC_SCOPE(a) \
122 ((a)->s6_addr[1] & 0x0f) /* nonstandard */
123 #define __IPV6_ADDR_SCOPE_INVALID -1
124 #define IPV6_ADDR_SCOPE_NODELOCAL 0x01
125 #define IPV6_ADDR_SCOPE_LINKLOCAL 0x02
126 #define IPV6_ADDR_SCOPE_SITELOCAL 0x05
127 #define IPV6_ADDR_SCOPE_ORGLOCAL 0x08
128 #define IPV6_ADDR_SCOPE_GLOBAL 0x0e
133 #define IPV6_ADDR_MC_FLAG_TRANSIENT(a) \
134 ((a)->s6_addr[1] & 0x10)
135 #define IPV6_ADDR_MC_FLAG_PREFIX(a) \
136 ((a)->s6_addr[1] & 0x20)
137 #define IPV6_ADDR_MC_FLAG_RENDEZVOUS(a) \
138 ((a)->s6_addr[1] & 0x40)
141 * fragmentation header
148 __be32 identification;
151 #define IP6_MF 0x0001
152 #define IP6_OFFSET 0xFFF8
154 struct ip6_fraglist_iter {
155 struct ipv6hdr *tmp_hdr;
156 struct sk_buff *frag;
163 int ip6_fraglist_init(struct sk_buff *skb, unsigned int hlen, u8 *prevhdr,
164 u8 nexthdr, __be32 frag_id,
165 struct ip6_fraglist_iter *iter);
166 void ip6_fraglist_prepare(struct sk_buff *skb, struct ip6_fraglist_iter *iter);
168 static inline struct sk_buff *ip6_fraglist_next(struct ip6_fraglist_iter *iter)
170 struct sk_buff *skb = iter->frag;
172 iter->frag = skb->next;
173 skb_mark_not_on_list(skb);
178 struct ip6_frag_state {
191 void ip6_frag_init(struct sk_buff *skb, unsigned int hlen, unsigned int mtu,
192 unsigned short needed_tailroom, int hdr_room, u8 *prevhdr,
193 u8 nexthdr, __be32 frag_id, struct ip6_frag_state *state);
194 struct sk_buff *ip6_frag_next(struct sk_buff *skb,
195 struct ip6_frag_state *state);
197 #define IP6_REPLY_MARK(net, mark) \
198 ((net)->ipv6.sysctl.fwmark_reflect ? (mark) : 0)
200 #include <net/sock.h>
203 extern int sysctl_mld_max_msf;
204 extern int sysctl_mld_qrv;
206 #define _DEVINC(net, statname, mod, idev, field) \
208 struct inet6_dev *_idev = (idev); \
209 if (likely(_idev != NULL)) \
210 mod##SNMP_INC_STATS64((_idev)->stats.statname, (field));\
211 mod##SNMP_INC_STATS64((net)->mib.statname##_statistics, (field));\
214 /* per device counters are atomic_long_t */
215 #define _DEVINCATOMIC(net, statname, mod, idev, field) \
217 struct inet6_dev *_idev = (idev); \
218 if (likely(_idev != NULL)) \
219 SNMP_INC_STATS_ATOMIC_LONG((_idev)->stats.statname##dev, (field)); \
220 mod##SNMP_INC_STATS((net)->mib.statname##_statistics, (field));\
223 /* per device and per net counters are atomic_long_t */
224 #define _DEVINC_ATOMIC_ATOMIC(net, statname, idev, field) \
226 struct inet6_dev *_idev = (idev); \
227 if (likely(_idev != NULL)) \
228 SNMP_INC_STATS_ATOMIC_LONG((_idev)->stats.statname##dev, (field)); \
229 SNMP_INC_STATS_ATOMIC_LONG((net)->mib.statname##_statistics, (field));\
232 #define _DEVADD(net, statname, mod, idev, field, val) \
234 struct inet6_dev *_idev = (idev); \
235 if (likely(_idev != NULL)) \
236 mod##SNMP_ADD_STATS((_idev)->stats.statname, (field), (val)); \
237 mod##SNMP_ADD_STATS((net)->mib.statname##_statistics, (field), (val));\
240 #define _DEVUPD(net, statname, mod, idev, field, val) \
242 struct inet6_dev *_idev = (idev); \
243 if (likely(_idev != NULL)) \
244 mod##SNMP_UPD_PO_STATS((_idev)->stats.statname, field, (val)); \
245 mod##SNMP_UPD_PO_STATS((net)->mib.statname##_statistics, field, (val));\
250 #define IP6_INC_STATS(net, idev,field) \
251 _DEVINC(net, ipv6, , idev, field)
252 #define __IP6_INC_STATS(net, idev,field) \
253 _DEVINC(net, ipv6, __, idev, field)
254 #define IP6_ADD_STATS(net, idev,field,val) \
255 _DEVADD(net, ipv6, , idev, field, val)
256 #define __IP6_ADD_STATS(net, idev,field,val) \
257 _DEVADD(net, ipv6, __, idev, field, val)
258 #define IP6_UPD_PO_STATS(net, idev,field,val) \
259 _DEVUPD(net, ipv6, , idev, field, val)
260 #define __IP6_UPD_PO_STATS(net, idev,field,val) \
261 _DEVUPD(net, ipv6, __, idev, field, val)
262 #define ICMP6_INC_STATS(net, idev, field) \
263 _DEVINCATOMIC(net, icmpv6, , idev, field)
264 #define __ICMP6_INC_STATS(net, idev, field) \
265 _DEVINCATOMIC(net, icmpv6, __, idev, field)
267 #define ICMP6MSGOUT_INC_STATS(net, idev, field) \
268 _DEVINC_ATOMIC_ATOMIC(net, icmpv6msg, idev, field +256)
269 #define ICMP6MSGIN_INC_STATS(net, idev, field) \
270 _DEVINC_ATOMIC_ATOMIC(net, icmpv6msg, idev, field)
272 struct ip6_ra_chain {
273 struct ip6_ra_chain *next;
276 void (*destructor)(struct sock *);
279 extern struct ip6_ra_chain *ip6_ra_chain;
280 extern rwlock_t ip6_ra_lock;
283 This structure is prepared by protocol, when parsing
284 ancillary data and passed to IPv6.
287 struct ipv6_txoptions {
289 /* Length of this structure */
292 /* length of extension headers */
294 __u16 opt_flen; /* after fragment hdr */
295 __u16 opt_nflen; /* before fragment hdr */
297 struct ipv6_opt_hdr *hopopt;
298 struct ipv6_opt_hdr *dst0opt;
299 struct ipv6_rt_hdr *srcrt; /* Routing Header */
300 struct ipv6_opt_hdr *dst1opt;
302 /* Option buffer, as read by IPV6_PKTOPTIONS, starts here. */
305 /* flowlabel_reflect sysctl values */
306 enum flowlabel_reflect {
307 FLOWLABEL_REFLECT_ESTABLISHED = 1,
308 FLOWLABEL_REFLECT_TCP_RESET = 2,
309 FLOWLABEL_REFLECT_ICMPV6_ECHO_REPLIES = 4,
312 struct ip6_flowlabel {
313 struct ip6_flowlabel __rcu *next;
317 struct ipv6_txoptions *opt;
318 unsigned long linger;
325 unsigned long lastuse;
326 unsigned long expires;
330 #define IPV6_FLOWINFO_MASK cpu_to_be32(0x0FFFFFFF)
331 #define IPV6_FLOWLABEL_MASK cpu_to_be32(0x000FFFFF)
332 #define IPV6_FLOWLABEL_STATELESS_FLAG cpu_to_be32(0x00080000)
334 #define IPV6_TCLASS_MASK (IPV6_FLOWINFO_MASK & ~IPV6_FLOWLABEL_MASK)
335 #define IPV6_TCLASS_SHIFT 20
337 struct ipv6_fl_socklist {
338 struct ipv6_fl_socklist __rcu *next;
339 struct ip6_flowlabel *fl;
343 struct ipcm6_cookie {
344 struct sockcm_cookie sockc;
348 struct ipv6_txoptions *opt;
352 static inline void ipcm6_init(struct ipcm6_cookie *ipc6)
354 *ipc6 = (struct ipcm6_cookie) {
361 static inline void ipcm6_init_sk(struct ipcm6_cookie *ipc6,
362 const struct ipv6_pinfo *np)
364 *ipc6 = (struct ipcm6_cookie) {
366 .tclass = np->tclass,
367 .dontfrag = np->dontfrag,
371 static inline struct ipv6_txoptions *txopt_get(const struct ipv6_pinfo *np)
373 struct ipv6_txoptions *opt;
376 opt = rcu_dereference(np->opt);
378 if (!refcount_inc_not_zero(&opt->refcnt))
381 opt = rcu_pointer_handoff(opt);
387 static inline void txopt_put(struct ipv6_txoptions *opt)
389 if (opt && refcount_dec_and_test(&opt->refcnt))
393 struct ip6_flowlabel *__fl6_sock_lookup(struct sock *sk, __be32 label);
395 extern struct static_key_false_deferred ipv6_flowlabel_exclusive;
396 static inline struct ip6_flowlabel *fl6_sock_lookup(struct sock *sk,
399 if (static_branch_unlikely(&ipv6_flowlabel_exclusive.key))
400 return __fl6_sock_lookup(sk, label) ? : ERR_PTR(-ENOENT);
405 struct ipv6_txoptions *fl6_merge_options(struct ipv6_txoptions *opt_space,
406 struct ip6_flowlabel *fl,
407 struct ipv6_txoptions *fopt);
408 void fl6_free_socklist(struct sock *sk);
409 int ipv6_flowlabel_opt(struct sock *sk, char __user *optval, int optlen);
410 int ipv6_flowlabel_opt_get(struct sock *sk, struct in6_flowlabel_req *freq,
412 int ip6_flowlabel_init(void);
413 void ip6_flowlabel_cleanup(void);
414 bool ip6_autoflowlabel(struct net *net, const struct ipv6_pinfo *np);
416 static inline void fl6_sock_release(struct ip6_flowlabel *fl)
419 atomic_dec(&fl->users);
422 void icmpv6_notify(struct sk_buff *skb, u8 type, u8 code, __be32 info);
424 void icmpv6_push_pending_frames(struct sock *sk, struct flowi6 *fl6,
425 struct icmp6hdr *thdr, int len);
427 int ip6_ra_control(struct sock *sk, int sel);
429 int ipv6_parse_hopopts(struct sk_buff *skb);
431 struct ipv6_txoptions *ipv6_dup_options(struct sock *sk,
432 struct ipv6_txoptions *opt);
433 struct ipv6_txoptions *ipv6_renew_options(struct sock *sk,
434 struct ipv6_txoptions *opt,
436 struct ipv6_opt_hdr *newopt);
437 struct ipv6_txoptions *ipv6_fixup_options(struct ipv6_txoptions *opt_space,
438 struct ipv6_txoptions *opt);
440 bool ipv6_opt_accepted(const struct sock *sk, const struct sk_buff *skb,
441 const struct inet6_skb_parm *opt);
442 struct ipv6_txoptions *ipv6_update_options(struct sock *sk,
443 struct ipv6_txoptions *opt);
445 static inline bool ipv6_accept_ra(struct inet6_dev *idev)
447 /* If forwarding is enabled, RA are not accepted unless the special
448 * hybrid mode (accept_ra=2) is enabled.
450 return idev->cnf.forwarding ? idev->cnf.accept_ra == 2 :
454 #define IPV6_FRAG_HIGH_THRESH (4 * 1024*1024) /* 4194304 */
455 #define IPV6_FRAG_LOW_THRESH (3 * 1024*1024) /* 3145728 */
456 #define IPV6_FRAG_TIMEOUT (60 * HZ) /* 60 seconds */
458 int __ipv6_addr_type(const struct in6_addr *addr);
459 static inline int ipv6_addr_type(const struct in6_addr *addr)
461 return __ipv6_addr_type(addr) & 0xffff;
464 static inline int ipv6_addr_scope(const struct in6_addr *addr)
466 return __ipv6_addr_type(addr) & IPV6_ADDR_SCOPE_MASK;
469 static inline int __ipv6_addr_src_scope(int type)
471 return (type == IPV6_ADDR_ANY) ? __IPV6_ADDR_SCOPE_INVALID : (type >> 16);
474 static inline int ipv6_addr_src_scope(const struct in6_addr *addr)
476 return __ipv6_addr_src_scope(__ipv6_addr_type(addr));
479 static inline bool __ipv6_addr_needs_scope_id(int type)
481 return type & IPV6_ADDR_LINKLOCAL ||
482 (type & IPV6_ADDR_MULTICAST &&
483 (type & (IPV6_ADDR_LOOPBACK|IPV6_ADDR_LINKLOCAL)));
486 static inline __u32 ipv6_iface_scope_id(const struct in6_addr *addr, int iface)
488 return __ipv6_addr_needs_scope_id(__ipv6_addr_type(addr)) ? iface : 0;
491 static inline int ipv6_addr_cmp(const struct in6_addr *a1, const struct in6_addr *a2)
493 return memcmp(a1, a2, sizeof(struct in6_addr));
497 ipv6_masked_addr_cmp(const struct in6_addr *a1, const struct in6_addr *m,
498 const struct in6_addr *a2)
500 #if defined(CONFIG_HAVE_EFFICIENT_UNALIGNED_ACCESS) && BITS_PER_LONG == 64
501 const unsigned long *ul1 = (const unsigned long *)a1;
502 const unsigned long *ulm = (const unsigned long *)m;
503 const unsigned long *ul2 = (const unsigned long *)a2;
505 return !!(((ul1[0] ^ ul2[0]) & ulm[0]) |
506 ((ul1[1] ^ ul2[1]) & ulm[1]));
508 return !!(((a1->s6_addr32[0] ^ a2->s6_addr32[0]) & m->s6_addr32[0]) |
509 ((a1->s6_addr32[1] ^ a2->s6_addr32[1]) & m->s6_addr32[1]) |
510 ((a1->s6_addr32[2] ^ a2->s6_addr32[2]) & m->s6_addr32[2]) |
511 ((a1->s6_addr32[3] ^ a2->s6_addr32[3]) & m->s6_addr32[3]));
515 static inline void ipv6_addr_prefix(struct in6_addr *pfx,
516 const struct in6_addr *addr,
519 /* caller must guarantee 0 <= plen <= 128 */
523 memset(pfx->s6_addr, 0, sizeof(pfx->s6_addr));
524 memcpy(pfx->s6_addr, addr, o);
526 pfx->s6_addr[o] = addr->s6_addr[o] & (0xff00 >> b);
529 static inline void ipv6_addr_prefix_copy(struct in6_addr *addr,
530 const struct in6_addr *pfx,
533 /* caller must guarantee 0 <= plen <= 128 */
537 memcpy(addr->s6_addr, pfx, o);
539 addr->s6_addr[o] &= ~(0xff00 >> b);
540 addr->s6_addr[o] |= (pfx->s6_addr[o] & (0xff00 >> b));
544 static inline void __ipv6_addr_set_half(__be32 *addr,
545 __be32 wh, __be32 wl)
547 #if defined(CONFIG_HAVE_EFFICIENT_UNALIGNED_ACCESS) && BITS_PER_LONG == 64
548 #if defined(__BIG_ENDIAN)
549 if (__builtin_constant_p(wh) && __builtin_constant_p(wl)) {
550 *(__force u64 *)addr = ((__force u64)(wh) << 32 | (__force u64)(wl));
553 #elif defined(__LITTLE_ENDIAN)
554 if (__builtin_constant_p(wl) && __builtin_constant_p(wh)) {
555 *(__force u64 *)addr = ((__force u64)(wl) << 32 | (__force u64)(wh));
564 static inline void ipv6_addr_set(struct in6_addr *addr,
565 __be32 w1, __be32 w2,
566 __be32 w3, __be32 w4)
568 __ipv6_addr_set_half(&addr->s6_addr32[0], w1, w2);
569 __ipv6_addr_set_half(&addr->s6_addr32[2], w3, w4);
572 static inline bool ipv6_addr_equal(const struct in6_addr *a1,
573 const struct in6_addr *a2)
575 #if defined(CONFIG_HAVE_EFFICIENT_UNALIGNED_ACCESS) && BITS_PER_LONG == 64
576 const unsigned long *ul1 = (const unsigned long *)a1;
577 const unsigned long *ul2 = (const unsigned long *)a2;
579 return ((ul1[0] ^ ul2[0]) | (ul1[1] ^ ul2[1])) == 0UL;
581 return ((a1->s6_addr32[0] ^ a2->s6_addr32[0]) |
582 (a1->s6_addr32[1] ^ a2->s6_addr32[1]) |
583 (a1->s6_addr32[2] ^ a2->s6_addr32[2]) |
584 (a1->s6_addr32[3] ^ a2->s6_addr32[3])) == 0;
588 #if defined(CONFIG_HAVE_EFFICIENT_UNALIGNED_ACCESS) && BITS_PER_LONG == 64
589 static inline bool __ipv6_prefix_equal64_half(const __be64 *a1,
593 if (len && ((*a1 ^ *a2) & cpu_to_be64((~0UL) << (64 - len))))
598 static inline bool ipv6_prefix_equal(const struct in6_addr *addr1,
599 const struct in6_addr *addr2,
600 unsigned int prefixlen)
602 const __be64 *a1 = (const __be64 *)addr1;
603 const __be64 *a2 = (const __be64 *)addr2;
605 if (prefixlen >= 64) {
608 return __ipv6_prefix_equal64_half(a1 + 1, a2 + 1, prefixlen - 64);
610 return __ipv6_prefix_equal64_half(a1, a2, prefixlen);
613 static inline bool ipv6_prefix_equal(const struct in6_addr *addr1,
614 const struct in6_addr *addr2,
615 unsigned int prefixlen)
617 const __be32 *a1 = addr1->s6_addr32;
618 const __be32 *a2 = addr2->s6_addr32;
619 unsigned int pdw, pbi;
621 /* check complete u32 in prefix */
622 pdw = prefixlen >> 5;
623 if (pdw && memcmp(a1, a2, pdw << 2))
626 /* check incomplete u32 in prefix */
627 pbi = prefixlen & 0x1f;
628 if (pbi && ((a1[pdw] ^ a2[pdw]) & htonl((0xffffffff) << (32 - pbi))))
635 static inline bool ipv6_addr_any(const struct in6_addr *a)
637 #if defined(CONFIG_HAVE_EFFICIENT_UNALIGNED_ACCESS) && BITS_PER_LONG == 64
638 const unsigned long *ul = (const unsigned long *)a;
640 return (ul[0] | ul[1]) == 0UL;
642 return (a->s6_addr32[0] | a->s6_addr32[1] |
643 a->s6_addr32[2] | a->s6_addr32[3]) == 0;
647 static inline u32 ipv6_addr_hash(const struct in6_addr *a)
649 #if defined(CONFIG_HAVE_EFFICIENT_UNALIGNED_ACCESS) && BITS_PER_LONG == 64
650 const unsigned long *ul = (const unsigned long *)a;
651 unsigned long x = ul[0] ^ ul[1];
653 return (u32)(x ^ (x >> 32));
655 return (__force u32)(a->s6_addr32[0] ^ a->s6_addr32[1] ^
656 a->s6_addr32[2] ^ a->s6_addr32[3]);
660 /* more secured version of ipv6_addr_hash() */
661 static inline u32 __ipv6_addr_jhash(const struct in6_addr *a, const u32 initval)
663 u32 v = (__force u32)a->s6_addr32[0] ^ (__force u32)a->s6_addr32[1];
665 return jhash_3words(v,
666 (__force u32)a->s6_addr32[2],
667 (__force u32)a->s6_addr32[3],
671 static inline bool ipv6_addr_loopback(const struct in6_addr *a)
673 #if defined(CONFIG_HAVE_EFFICIENT_UNALIGNED_ACCESS) && BITS_PER_LONG == 64
674 const __be64 *be = (const __be64 *)a;
676 return (be[0] | (be[1] ^ cpu_to_be64(1))) == 0UL;
678 return (a->s6_addr32[0] | a->s6_addr32[1] |
679 a->s6_addr32[2] | (a->s6_addr32[3] ^ cpu_to_be32(1))) == 0;
684 * Note that we must __force cast these to unsigned long to make sparse happy,
685 * since all of the endian-annotated types are fixed size regardless of arch.
687 static inline bool ipv6_addr_v4mapped(const struct in6_addr *a)
690 #if defined(CONFIG_HAVE_EFFICIENT_UNALIGNED_ACCESS) && BITS_PER_LONG == 64
691 *(unsigned long *)a |
693 (__force unsigned long)(a->s6_addr32[0] | a->s6_addr32[1]) |
695 (__force unsigned long)(a->s6_addr32[2] ^
696 cpu_to_be32(0x0000ffff))) == 0UL;
699 static inline bool ipv6_addr_v4mapped_loopback(const struct in6_addr *a)
701 return ipv6_addr_v4mapped(a) && ipv4_is_loopback(a->s6_addr32[3]);
704 static inline u32 ipv6_portaddr_hash(const struct net *net,
705 const struct in6_addr *addr6,
708 unsigned int hash, mix = net_hash_mix(net);
710 if (ipv6_addr_any(addr6))
711 hash = jhash_1word(0, mix);
712 else if (ipv6_addr_v4mapped(addr6))
713 hash = jhash_1word((__force u32)addr6->s6_addr32[3], mix);
715 hash = jhash2((__force u32 *)addr6->s6_addr32, 4, mix);
721 * Check for a RFC 4843 ORCHID address
722 * (Overlay Routable Cryptographic Hash Identifiers)
724 static inline bool ipv6_addr_orchid(const struct in6_addr *a)
726 return (a->s6_addr32[0] & htonl(0xfffffff0)) == htonl(0x20010010);
729 static inline bool ipv6_addr_is_multicast(const struct in6_addr *addr)
731 return (addr->s6_addr32[0] & htonl(0xFF000000)) == htonl(0xFF000000);
734 static inline void ipv6_addr_set_v4mapped(const __be32 addr,
735 struct in6_addr *v4mapped)
737 ipv6_addr_set(v4mapped,
744 * find the first different bit between two addresses
745 * length of address must be a multiple of 32bits
747 static inline int __ipv6_addr_diff32(const void *token1, const void *token2, int addrlen)
749 const __be32 *a1 = token1, *a2 = token2;
754 for (i = 0; i < addrlen; i++) {
755 __be32 xb = a1[i] ^ a2[i];
757 return i * 32 + 31 - __fls(ntohl(xb));
761 * we should *never* get to this point since that
762 * would mean the addrs are equal
764 * However, we do get to it 8) And exacly, when
765 * addresses are equal 8)
767 * ip route add 1111::/128 via ...
768 * ip route add 1111::/64 via ...
771 * Ideally, this function should stop comparison
772 * at prefix length. It does not, but it is still OK,
773 * if returned value is greater than prefix length.
779 #if defined(CONFIG_HAVE_EFFICIENT_UNALIGNED_ACCESS) && BITS_PER_LONG == 64
780 static inline int __ipv6_addr_diff64(const void *token1, const void *token2, int addrlen)
782 const __be64 *a1 = token1, *a2 = token2;
787 for (i = 0; i < addrlen; i++) {
788 __be64 xb = a1[i] ^ a2[i];
790 return i * 64 + 63 - __fls(be64_to_cpu(xb));
797 static inline int __ipv6_addr_diff(const void *token1, const void *token2, int addrlen)
799 #if defined(CONFIG_HAVE_EFFICIENT_UNALIGNED_ACCESS) && BITS_PER_LONG == 64
800 if (__builtin_constant_p(addrlen) && !(addrlen & 7))
801 return __ipv6_addr_diff64(token1, token2, addrlen);
803 return __ipv6_addr_diff32(token1, token2, addrlen);
806 static inline int ipv6_addr_diff(const struct in6_addr *a1, const struct in6_addr *a2)
808 return __ipv6_addr_diff(a1, a2, sizeof(struct in6_addr));
811 __be32 ipv6_select_ident(struct net *net,
812 const struct in6_addr *daddr,
813 const struct in6_addr *saddr);
814 __be32 ipv6_proxy_select_ident(struct net *net, struct sk_buff *skb);
816 int ip6_dst_hoplimit(struct dst_entry *dst);
818 static inline int ip6_sk_dst_hoplimit(struct ipv6_pinfo *np, struct flowi6 *fl6,
819 struct dst_entry *dst)
823 if (ipv6_addr_is_multicast(&fl6->daddr))
824 hlimit = np->mcast_hops;
826 hlimit = np->hop_limit;
828 hlimit = ip6_dst_hoplimit(dst);
832 /* copy IPv6 saddr & daddr to flow_keys, possibly using 64bit load/store
833 * Equivalent to : flow->v6addrs.src = iph->saddr;
834 * flow->v6addrs.dst = iph->daddr;
836 static inline void iph_to_flow_copy_v6addrs(struct flow_keys *flow,
837 const struct ipv6hdr *iph)
839 BUILD_BUG_ON(offsetof(typeof(flow->addrs), v6addrs.dst) !=
840 offsetof(typeof(flow->addrs), v6addrs.src) +
841 sizeof(flow->addrs.v6addrs.src));
842 memcpy(&flow->addrs.v6addrs, &iph->saddr, sizeof(flow->addrs.v6addrs));
843 flow->control.addr_type = FLOW_DISSECTOR_KEY_IPV6_ADDRS;
846 #if IS_ENABLED(CONFIG_IPV6)
848 static inline bool ipv6_can_nonlocal_bind(struct net *net,
849 struct inet_sock *inet)
851 return net->ipv6.sysctl.ip_nonlocal_bind ||
852 inet->freebind || inet->transparent;
855 /* Sysctl settings for net ipv6.auto_flowlabels */
856 #define IP6_AUTO_FLOW_LABEL_OFF 0
857 #define IP6_AUTO_FLOW_LABEL_OPTOUT 1
858 #define IP6_AUTO_FLOW_LABEL_OPTIN 2
859 #define IP6_AUTO_FLOW_LABEL_FORCED 3
861 #define IP6_AUTO_FLOW_LABEL_MAX IP6_AUTO_FLOW_LABEL_FORCED
863 #define IP6_DEFAULT_AUTO_FLOW_LABELS IP6_AUTO_FLOW_LABEL_OPTOUT
865 static inline __be32 ip6_make_flowlabel(struct net *net, struct sk_buff *skb,
866 __be32 flowlabel, bool autolabel,
871 /* @flowlabel may include more than a flow label, eg, the traffic class.
872 * Here we want only the flow label value.
874 flowlabel &= IPV6_FLOWLABEL_MASK;
877 net->ipv6.sysctl.auto_flowlabels == IP6_AUTO_FLOW_LABEL_OFF ||
879 net->ipv6.sysctl.auto_flowlabels != IP6_AUTO_FLOW_LABEL_FORCED))
882 hash = skb_get_hash_flowi6(skb, fl6);
884 /* Since this is being sent on the wire obfuscate hash a bit
885 * to minimize possbility that any useful information to an
886 * attacker is leaked. Only lower 20 bits are relevant.
888 hash = rol32(hash, 16);
890 flowlabel = (__force __be32)hash & IPV6_FLOWLABEL_MASK;
892 if (net->ipv6.sysctl.flowlabel_state_ranges)
893 flowlabel |= IPV6_FLOWLABEL_STATELESS_FLAG;
898 static inline int ip6_default_np_autolabel(struct net *net)
900 switch (net->ipv6.sysctl.auto_flowlabels) {
901 case IP6_AUTO_FLOW_LABEL_OFF:
902 case IP6_AUTO_FLOW_LABEL_OPTIN:
905 case IP6_AUTO_FLOW_LABEL_OPTOUT:
906 case IP6_AUTO_FLOW_LABEL_FORCED:
911 static inline void ip6_set_txhash(struct sock *sk) { }
912 static inline __be32 ip6_make_flowlabel(struct net *net, struct sk_buff *skb,
913 __be32 flowlabel, bool autolabel,
918 static inline int ip6_default_np_autolabel(struct net *net)
924 #if IS_ENABLED(CONFIG_IPV6)
925 static inline int ip6_multipath_hash_policy(const struct net *net)
927 return net->ipv6.sysctl.multipath_hash_policy;
930 static inline int ip6_multipath_hash_policy(const struct net *net)
937 * Header manipulation
939 static inline void ip6_flow_hdr(struct ipv6hdr *hdr, unsigned int tclass,
942 *(__be32 *)hdr = htonl(0x60000000 | (tclass << 20)) | flowlabel;
945 static inline __be32 ip6_flowinfo(const struct ipv6hdr *hdr)
947 return *(__be32 *)hdr & IPV6_FLOWINFO_MASK;
950 static inline __be32 ip6_flowlabel(const struct ipv6hdr *hdr)
952 return *(__be32 *)hdr & IPV6_FLOWLABEL_MASK;
955 static inline u8 ip6_tclass(__be32 flowinfo)
957 return ntohl(flowinfo & IPV6_TCLASS_MASK) >> IPV6_TCLASS_SHIFT;
960 static inline __be32 ip6_make_flowinfo(unsigned int tclass, __be32 flowlabel)
962 return htonl(tclass << IPV6_TCLASS_SHIFT) | flowlabel;
965 static inline __be32 flowi6_get_flowlabel(const struct flowi6 *fl6)
967 return fl6->flowlabel & IPV6_FLOWLABEL_MASK;
971 * Prototypes exported by ipv6
975 * rcv function (called from netdevice level)
978 int ipv6_rcv(struct sk_buff *skb, struct net_device *dev,
979 struct packet_type *pt, struct net_device *orig_dev);
980 void ipv6_list_rcv(struct list_head *head, struct packet_type *pt,
981 struct net_device *orig_dev);
983 int ip6_rcv_finish(struct net *net, struct sock *sk, struct sk_buff *skb);
986 * upper-layer output functions
988 int ip6_xmit(const struct sock *sk, struct sk_buff *skb, struct flowi6 *fl6,
989 __u32 mark, struct ipv6_txoptions *opt, int tclass, u32 priority);
991 int ip6_find_1stfragopt(struct sk_buff *skb, u8 **nexthdr);
993 int ip6_append_data(struct sock *sk,
994 int getfrag(void *from, char *to, int offset, int len,
995 int odd, struct sk_buff *skb),
996 void *from, int length, int transhdrlen,
997 struct ipcm6_cookie *ipc6, struct flowi6 *fl6,
998 struct rt6_info *rt, unsigned int flags);
1000 int ip6_push_pending_frames(struct sock *sk);
1002 void ip6_flush_pending_frames(struct sock *sk);
1004 int ip6_send_skb(struct sk_buff *skb);
1006 struct sk_buff *__ip6_make_skb(struct sock *sk, struct sk_buff_head *queue,
1007 struct inet_cork_full *cork,
1008 struct inet6_cork *v6_cork);
1009 struct sk_buff *ip6_make_skb(struct sock *sk,
1010 int getfrag(void *from, char *to, int offset,
1011 int len, int odd, struct sk_buff *skb),
1012 void *from, int length, int transhdrlen,
1013 struct ipcm6_cookie *ipc6, struct flowi6 *fl6,
1014 struct rt6_info *rt, unsigned int flags,
1015 struct inet_cork_full *cork);
1017 static inline struct sk_buff *ip6_finish_skb(struct sock *sk)
1019 return __ip6_make_skb(sk, &sk->sk_write_queue, &inet_sk(sk)->cork,
1020 &inet6_sk(sk)->cork);
1023 int ip6_dst_lookup(struct net *net, struct sock *sk, struct dst_entry **dst,
1024 struct flowi6 *fl6);
1025 struct dst_entry *ip6_dst_lookup_flow(struct net *net, const struct sock *sk, struct flowi6 *fl6,
1026 const struct in6_addr *final_dst);
1027 struct dst_entry *ip6_sk_dst_lookup_flow(struct sock *sk, struct flowi6 *fl6,
1028 const struct in6_addr *final_dst,
1030 struct dst_entry *ip6_dst_lookup_tunnel(struct sk_buff *skb,
1031 struct net_device *dev,
1032 struct net *net, struct socket *sock,
1033 struct in6_addr *saddr,
1034 const struct ip_tunnel_info *info,
1035 u8 protocol, bool use_cache);
1036 struct dst_entry *ip6_blackhole_route(struct net *net,
1037 struct dst_entry *orig_dst);
1040 * skb processing functions
1043 int ip6_output(struct net *net, struct sock *sk, struct sk_buff *skb);
1044 int ip6_forward(struct sk_buff *skb);
1045 int ip6_input(struct sk_buff *skb);
1046 int ip6_mc_input(struct sk_buff *skb);
1047 void ip6_protocol_deliver_rcu(struct net *net, struct sk_buff *skb, int nexthdr,
1050 int __ip6_local_out(struct net *net, struct sock *sk, struct sk_buff *skb);
1051 int ip6_local_out(struct net *net, struct sock *sk, struct sk_buff *skb);
1054 * Extension header (options) processing
1057 void ipv6_push_nfrag_opts(struct sk_buff *skb, struct ipv6_txoptions *opt,
1058 u8 *proto, struct in6_addr **daddr_p,
1059 struct in6_addr *saddr);
1060 void ipv6_push_frag_opts(struct sk_buff *skb, struct ipv6_txoptions *opt,
1063 int ipv6_skip_exthdr(const struct sk_buff *, int start, u8 *nexthdrp,
1066 bool ipv6_ext_hdr(u8 nexthdr);
1069 IP6_FH_F_FRAG = (1 << 0),
1070 IP6_FH_F_AUTH = (1 << 1),
1071 IP6_FH_F_SKIP_RH = (1 << 2),
1074 /* find specified header and get offset to it */
1075 int ipv6_find_hdr(const struct sk_buff *skb, unsigned int *offset, int target,
1076 unsigned short *fragoff, int *fragflg);
1078 int ipv6_find_tlv(const struct sk_buff *skb, int offset, int type);
1080 struct in6_addr *fl6_update_dst(struct flowi6 *fl6,
1081 const struct ipv6_txoptions *opt,
1082 struct in6_addr *orig);
1085 * socket options (ipv6_sockglue.c)
1088 int ipv6_setsockopt(struct sock *sk, int level, int optname,
1089 char __user *optval, unsigned int optlen);
1090 int ipv6_getsockopt(struct sock *sk, int level, int optname,
1091 char __user *optval, int __user *optlen);
1092 int compat_ipv6_setsockopt(struct sock *sk, int level, int optname,
1093 char __user *optval, unsigned int optlen);
1094 int compat_ipv6_getsockopt(struct sock *sk, int level, int optname,
1095 char __user *optval, int __user *optlen);
1097 int __ip6_datagram_connect(struct sock *sk, struct sockaddr *addr,
1099 int ip6_datagram_connect(struct sock *sk, struct sockaddr *addr, int addr_len);
1100 int ip6_datagram_connect_v6_only(struct sock *sk, struct sockaddr *addr,
1102 int ip6_datagram_dst_update(struct sock *sk, bool fix_sk_saddr);
1103 void ip6_datagram_release_cb(struct sock *sk);
1105 int ipv6_recv_error(struct sock *sk, struct msghdr *msg, int len,
1107 int ipv6_recv_rxpmtu(struct sock *sk, struct msghdr *msg, int len,
1109 void ipv6_icmp_error(struct sock *sk, struct sk_buff *skb, int err, __be16 port,
1110 u32 info, u8 *payload);
1111 void ipv6_local_error(struct sock *sk, int err, struct flowi6 *fl6, u32 info);
1112 void ipv6_local_rxpmtu(struct sock *sk, struct flowi6 *fl6, u32 mtu);
1114 int inet6_release(struct socket *sock);
1115 int inet6_bind(struct socket *sock, struct sockaddr *uaddr, int addr_len);
1116 int inet6_getname(struct socket *sock, struct sockaddr *uaddr,
1118 int inet6_ioctl(struct socket *sock, unsigned int cmd, unsigned long arg);
1120 int inet6_hash_connect(struct inet_timewait_death_row *death_row,
1122 int inet6_sendmsg(struct socket *sock, struct msghdr *msg, size_t size);
1123 int inet6_recvmsg(struct socket *sock, struct msghdr *msg, size_t size,
1129 extern const struct proto_ops inet6_stream_ops;
1130 extern const struct proto_ops inet6_dgram_ops;
1131 extern const struct proto_ops inet6_sockraw_ops;
1133 struct group_source_req;
1134 struct group_filter;
1136 int ip6_mc_source(int add, int omode, struct sock *sk,
1137 struct group_source_req *pgsr);
1138 int ip6_mc_msfilter(struct sock *sk, struct group_filter *gsf);
1139 int ip6_mc_msfget(struct sock *sk, struct group_filter *gsf,
1140 struct group_filter __user *optval, int __user *optlen);
1142 #ifdef CONFIG_PROC_FS
1143 int ac6_proc_init(struct net *net);
1144 void ac6_proc_exit(struct net *net);
1145 int raw6_proc_init(void);
1146 void raw6_proc_exit(void);
1147 int tcp6_proc_init(struct net *net);
1148 void tcp6_proc_exit(struct net *net);
1149 int udp6_proc_init(struct net *net);
1150 void udp6_proc_exit(struct net *net);
1151 int udplite6_proc_init(void);
1152 void udplite6_proc_exit(void);
1153 int ipv6_misc_proc_init(void);
1154 void ipv6_misc_proc_exit(void);
1155 int snmp6_register_dev(struct inet6_dev *idev);
1156 int snmp6_unregister_dev(struct inet6_dev *idev);
1159 static inline int ac6_proc_init(struct net *net) { return 0; }
1160 static inline void ac6_proc_exit(struct net *net) { }
1161 static inline int snmp6_register_dev(struct inet6_dev *idev) { return 0; }
1162 static inline int snmp6_unregister_dev(struct inet6_dev *idev) { return 0; }
1165 #ifdef CONFIG_SYSCTL
1166 struct ctl_table *ipv6_icmp_sysctl_init(struct net *net);
1167 struct ctl_table *ipv6_route_sysctl_init(struct net *net);
1168 int ipv6_sysctl_register(void);
1169 void ipv6_sysctl_unregister(void);
1172 int ipv6_sock_mc_join(struct sock *sk, int ifindex,
1173 const struct in6_addr *addr);
1174 int ipv6_sock_mc_join_ssm(struct sock *sk, int ifindex,
1175 const struct in6_addr *addr, unsigned int mode);
1176 int ipv6_sock_mc_drop(struct sock *sk, int ifindex,
1177 const struct in6_addr *addr);
1178 #endif /* _NET_IPV6_H */