2 * Linux INET6 implementation
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.
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>
23 #include <net/flow_dissector.h>
25 #include <net/netns/hash.h>
27 #define SIN6_LEN_RFC2133 24
29 #define IPV6_MAXPLEN 65535
32 * NextHeader field of IPv6 header
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. */
50 #define NEXTHDR_MAX 255
52 #define IPV6_DEFAULT_HOPLIMIT 64
53 #define IPV6_DEFAULT_MCASTHOPS 1
55 /* Limits on Hop-by-Hop and Destination options.
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
62 * There are three limits that may be set:
63 * - Limit the number of options in a Hop-by-Hop or Destination options
65 * - Limit the byte length of a Hop-by-Hop or Destination options extension
67 * - Disallow unknown options
69 * The limits are expressed in corresponding sysctls:
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
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.
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.
85 * If a limit is exceeded when processing an extension header the packet is
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 */
98 * type - unicast | multicast
99 * scope - local | site | global
106 #define IPV6_ADDR_ANY 0x0000U
108 #define IPV6_ADDR_UNICAST 0x0001U
109 #define IPV6_ADDR_MULTICAST 0x0002U
111 #define IPV6_ADDR_LOOPBACK 0x0010U
112 #define IPV6_ADDR_LINKLOCAL 0x0020U
113 #define IPV6_ADDR_SITELOCAL 0x0040U
115 #define IPV6_ADDR_COMPATv4 0x0080U
117 #define IPV6_ADDR_SCOPE_MASK 0x00f0U
119 #define IPV6_ADDR_MAPPED 0x1000U
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
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)
144 * fragmentation header
151 __be32 identification;
154 #define IP6_MF 0x0001
155 #define IP6_OFFSET 0xFFF8
157 #define IP6_REPLY_MARK(net, mark) \
158 ((net)->ipv6.sysctl.fwmark_reflect ? (mark) : 0)
160 #include <net/sock.h>
163 extern int sysctl_mld_max_msf;
164 extern int sysctl_mld_qrv;
166 #define _DEVINC(net, statname, mod, idev, field) \
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));\
174 /* per device counters are atomic_long_t */
175 #define _DEVINCATOMIC(net, statname, mod, idev, field) \
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));\
183 /* per device and per net counters are atomic_long_t */
184 #define _DEVINC_ATOMIC_ATOMIC(net, statname, idev, field) \
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));\
192 #define _DEVADD(net, statname, mod, idev, field, val) \
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));\
200 #define _DEVUPD(net, statname, mod, idev, field, val) \
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));\
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)
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)
232 struct ip6_ra_chain {
233 struct ip6_ra_chain *next;
236 void (*destructor)(struct sock *);
239 extern struct ip6_ra_chain *ip6_ra_chain;
240 extern rwlock_t ip6_ra_lock;
243 This structure is prepared by protocol, when parsing
244 ancillary data and passed to IPv6.
247 struct ipv6_txoptions {
249 /* Length of this structure */
252 /* length of extension headers */
254 __u16 opt_flen; /* after fragment hdr */
255 __u16 opt_nflen; /* before fragment hdr */
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;
262 /* Option buffer, as read by IPV6_PKTOPTIONS, starts here. */
265 struct ip6_flowlabel {
266 struct ip6_flowlabel __rcu *next;
270 struct ipv6_txoptions *opt;
271 unsigned long linger;
278 unsigned long lastuse;
279 unsigned long expires;
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)
287 #define IPV6_TCLASS_MASK (IPV6_FLOWINFO_MASK & ~IPV6_FLOWLABEL_MASK)
288 #define IPV6_TCLASS_SHIFT 20
290 struct ipv6_fl_socklist {
291 struct ipv6_fl_socklist __rcu *next;
292 struct ip6_flowlabel *fl;
296 struct ipcm6_cookie {
300 struct ipv6_txoptions *opt;
304 static inline struct ipv6_txoptions *txopt_get(const struct ipv6_pinfo *np)
306 struct ipv6_txoptions *opt;
309 opt = rcu_dereference(np->opt);
311 if (!refcount_inc_not_zero(&opt->refcnt))
314 opt = rcu_pointer_handoff(opt);
320 static inline void txopt_put(struct ipv6_txoptions *opt)
322 if (opt && refcount_dec_and_test(&opt->refcnt))
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,
334 int ip6_flowlabel_init(void);
335 void ip6_flowlabel_cleanup(void);
336 bool ip6_autoflowlabel(struct net *net, const struct ipv6_pinfo *np);
338 static inline void fl6_sock_release(struct ip6_flowlabel *fl)
341 atomic_dec(&fl->users);
344 void icmpv6_notify(struct sk_buff *skb, u8 type, u8 code, __be32 info);
346 void icmpv6_push_pending_frames(struct sock *sk, struct flowi6 *fl6,
347 struct icmp6hdr *thdr, int len);
349 int ip6_ra_control(struct sock *sk, int sel);
351 int ipv6_parse_hopopts(struct sk_buff *skb);
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,
358 struct ipv6_opt_hdr __user *newopt,
360 struct ipv6_txoptions *
361 ipv6_renew_options_kern(struct sock *sk,
362 struct ipv6_txoptions *opt,
364 struct ipv6_opt_hdr *newopt,
366 struct ipv6_txoptions *ipv6_fixup_options(struct ipv6_txoptions *opt_space,
367 struct ipv6_txoptions *opt);
369 bool ipv6_opt_accepted(const struct sock *sk, const struct sk_buff *skb,
370 const struct inet6_skb_parm *opt);
371 struct ipv6_txoptions *ipv6_update_options(struct sock *sk,
372 struct ipv6_txoptions *opt);
374 static inline bool ipv6_accept_ra(struct inet6_dev *idev)
376 /* If forwarding is enabled, RA are not accepted unless the special
377 * hybrid mode (accept_ra=2) is enabled.
379 return idev->cnf.forwarding ? idev->cnf.accept_ra == 2 :
383 #define IPV6_FRAG_HIGH_THRESH (4 * 1024*1024) /* 4194304 */
384 #define IPV6_FRAG_LOW_THRESH (3 * 1024*1024) /* 3145728 */
385 #define IPV6_FRAG_TIMEOUT (60 * HZ) /* 60 seconds */
387 int __ipv6_addr_type(const struct in6_addr *addr);
388 static inline int ipv6_addr_type(const struct in6_addr *addr)
390 return __ipv6_addr_type(addr) & 0xffff;
393 static inline int ipv6_addr_scope(const struct in6_addr *addr)
395 return __ipv6_addr_type(addr) & IPV6_ADDR_SCOPE_MASK;
398 static inline int __ipv6_addr_src_scope(int type)
400 return (type == IPV6_ADDR_ANY) ? __IPV6_ADDR_SCOPE_INVALID : (type >> 16);
403 static inline int ipv6_addr_src_scope(const struct in6_addr *addr)
405 return __ipv6_addr_src_scope(__ipv6_addr_type(addr));
408 static inline bool __ipv6_addr_needs_scope_id(int type)
410 return type & IPV6_ADDR_LINKLOCAL ||
411 (type & IPV6_ADDR_MULTICAST &&
412 (type & (IPV6_ADDR_LOOPBACK|IPV6_ADDR_LINKLOCAL)));
415 static inline __u32 ipv6_iface_scope_id(const struct in6_addr *addr, int iface)
417 return __ipv6_addr_needs_scope_id(__ipv6_addr_type(addr)) ? iface : 0;
420 static inline int ipv6_addr_cmp(const struct in6_addr *a1, const struct in6_addr *a2)
422 return memcmp(a1, a2, sizeof(struct in6_addr));
426 ipv6_masked_addr_cmp(const struct in6_addr *a1, const struct in6_addr *m,
427 const struct in6_addr *a2)
429 #if defined(CONFIG_HAVE_EFFICIENT_UNALIGNED_ACCESS) && BITS_PER_LONG == 64
430 const unsigned long *ul1 = (const unsigned long *)a1;
431 const unsigned long *ulm = (const unsigned long *)m;
432 const unsigned long *ul2 = (const unsigned long *)a2;
434 return !!(((ul1[0] ^ ul2[0]) & ulm[0]) |
435 ((ul1[1] ^ ul2[1]) & ulm[1]));
437 return !!(((a1->s6_addr32[0] ^ a2->s6_addr32[0]) & m->s6_addr32[0]) |
438 ((a1->s6_addr32[1] ^ a2->s6_addr32[1]) & m->s6_addr32[1]) |
439 ((a1->s6_addr32[2] ^ a2->s6_addr32[2]) & m->s6_addr32[2]) |
440 ((a1->s6_addr32[3] ^ a2->s6_addr32[3]) & m->s6_addr32[3]));
444 static inline void ipv6_addr_prefix(struct in6_addr *pfx,
445 const struct in6_addr *addr,
448 /* caller must guarantee 0 <= plen <= 128 */
452 memset(pfx->s6_addr, 0, sizeof(pfx->s6_addr));
453 memcpy(pfx->s6_addr, addr, o);
455 pfx->s6_addr[o] = addr->s6_addr[o] & (0xff00 >> b);
458 static inline void ipv6_addr_prefix_copy(struct in6_addr *addr,
459 const struct in6_addr *pfx,
462 /* caller must guarantee 0 <= plen <= 128 */
466 memcpy(addr->s6_addr, pfx, o);
468 addr->s6_addr[o] &= ~(0xff00 >> b);
469 addr->s6_addr[o] |= (pfx->s6_addr[o] & (0xff00 >> b));
473 static inline void __ipv6_addr_set_half(__be32 *addr,
474 __be32 wh, __be32 wl)
476 #if defined(CONFIG_HAVE_EFFICIENT_UNALIGNED_ACCESS) && BITS_PER_LONG == 64
477 #if defined(__BIG_ENDIAN)
478 if (__builtin_constant_p(wh) && __builtin_constant_p(wl)) {
479 *(__force u64 *)addr = ((__force u64)(wh) << 32 | (__force u64)(wl));
482 #elif defined(__LITTLE_ENDIAN)
483 if (__builtin_constant_p(wl) && __builtin_constant_p(wh)) {
484 *(__force u64 *)addr = ((__force u64)(wl) << 32 | (__force u64)(wh));
493 static inline void ipv6_addr_set(struct in6_addr *addr,
494 __be32 w1, __be32 w2,
495 __be32 w3, __be32 w4)
497 __ipv6_addr_set_half(&addr->s6_addr32[0], w1, w2);
498 __ipv6_addr_set_half(&addr->s6_addr32[2], w3, w4);
501 static inline bool ipv6_addr_equal(const struct in6_addr *a1,
502 const struct in6_addr *a2)
504 #if defined(CONFIG_HAVE_EFFICIENT_UNALIGNED_ACCESS) && BITS_PER_LONG == 64
505 const unsigned long *ul1 = (const unsigned long *)a1;
506 const unsigned long *ul2 = (const unsigned long *)a2;
508 return ((ul1[0] ^ ul2[0]) | (ul1[1] ^ ul2[1])) == 0UL;
510 return ((a1->s6_addr32[0] ^ a2->s6_addr32[0]) |
511 (a1->s6_addr32[1] ^ a2->s6_addr32[1]) |
512 (a1->s6_addr32[2] ^ a2->s6_addr32[2]) |
513 (a1->s6_addr32[3] ^ a2->s6_addr32[3])) == 0;
517 #if defined(CONFIG_HAVE_EFFICIENT_UNALIGNED_ACCESS) && BITS_PER_LONG == 64
518 static inline bool __ipv6_prefix_equal64_half(const __be64 *a1,
522 if (len && ((*a1 ^ *a2) & cpu_to_be64((~0UL) << (64 - len))))
527 static inline bool ipv6_prefix_equal(const struct in6_addr *addr1,
528 const struct in6_addr *addr2,
529 unsigned int prefixlen)
531 const __be64 *a1 = (const __be64 *)addr1;
532 const __be64 *a2 = (const __be64 *)addr2;
534 if (prefixlen >= 64) {
537 return __ipv6_prefix_equal64_half(a1 + 1, a2 + 1, prefixlen - 64);
539 return __ipv6_prefix_equal64_half(a1, a2, prefixlen);
542 static inline bool ipv6_prefix_equal(const struct in6_addr *addr1,
543 const struct in6_addr *addr2,
544 unsigned int prefixlen)
546 const __be32 *a1 = addr1->s6_addr32;
547 const __be32 *a2 = addr2->s6_addr32;
548 unsigned int pdw, pbi;
550 /* check complete u32 in prefix */
551 pdw = prefixlen >> 5;
552 if (pdw && memcmp(a1, a2, pdw << 2))
555 /* check incomplete u32 in prefix */
556 pbi = prefixlen & 0x1f;
557 if (pbi && ((a1[pdw] ^ a2[pdw]) & htonl((0xffffffff) << (32 - pbi))))
564 struct inet_frag_queue;
566 enum ip6_defrag_users {
567 IP6_DEFRAG_LOCAL_DELIVER,
568 IP6_DEFRAG_CONNTRACK_IN,
569 __IP6_DEFRAG_CONNTRACK_IN = IP6_DEFRAG_CONNTRACK_IN + USHRT_MAX,
570 IP6_DEFRAG_CONNTRACK_OUT,
571 __IP6_DEFRAG_CONNTRACK_OUT = IP6_DEFRAG_CONNTRACK_OUT + USHRT_MAX,
572 IP6_DEFRAG_CONNTRACK_BRIDGE_IN,
573 __IP6_DEFRAG_CONNTRACK_BRIDGE_IN = IP6_DEFRAG_CONNTRACK_BRIDGE_IN + USHRT_MAX,
576 void ip6_frag_init(struct inet_frag_queue *q, const void *a);
577 extern const struct rhashtable_params ip6_rhash_params;
580 * Equivalent of ipv4 struct ip
583 struct inet_frag_queue q;
590 void ip6_expire_frag_queue(struct net *net, struct frag_queue *fq);
592 static inline bool ipv6_addr_any(const struct in6_addr *a)
594 #if defined(CONFIG_HAVE_EFFICIENT_UNALIGNED_ACCESS) && BITS_PER_LONG == 64
595 const unsigned long *ul = (const unsigned long *)a;
597 return (ul[0] | ul[1]) == 0UL;
599 return (a->s6_addr32[0] | a->s6_addr32[1] |
600 a->s6_addr32[2] | a->s6_addr32[3]) == 0;
604 static inline u32 ipv6_addr_hash(const struct in6_addr *a)
606 #if defined(CONFIG_HAVE_EFFICIENT_UNALIGNED_ACCESS) && BITS_PER_LONG == 64
607 const unsigned long *ul = (const unsigned long *)a;
608 unsigned long x = ul[0] ^ ul[1];
610 return (u32)(x ^ (x >> 32));
612 return (__force u32)(a->s6_addr32[0] ^ a->s6_addr32[1] ^
613 a->s6_addr32[2] ^ a->s6_addr32[3]);
617 /* more secured version of ipv6_addr_hash() */
618 static inline u32 __ipv6_addr_jhash(const struct in6_addr *a, const u32 initval)
620 u32 v = (__force u32)a->s6_addr32[0] ^ (__force u32)a->s6_addr32[1];
622 return jhash_3words(v,
623 (__force u32)a->s6_addr32[2],
624 (__force u32)a->s6_addr32[3],
628 static inline bool ipv6_addr_loopback(const struct in6_addr *a)
630 #if defined(CONFIG_HAVE_EFFICIENT_UNALIGNED_ACCESS) && BITS_PER_LONG == 64
631 const __be64 *be = (const __be64 *)a;
633 return (be[0] | (be[1] ^ cpu_to_be64(1))) == 0UL;
635 return (a->s6_addr32[0] | a->s6_addr32[1] |
636 a->s6_addr32[2] | (a->s6_addr32[3] ^ cpu_to_be32(1))) == 0;
641 * Note that we must __force cast these to unsigned long to make sparse happy,
642 * since all of the endian-annotated types are fixed size regardless of arch.
644 static inline bool ipv6_addr_v4mapped(const struct in6_addr *a)
647 #if defined(CONFIG_HAVE_EFFICIENT_UNALIGNED_ACCESS) && BITS_PER_LONG == 64
648 *(unsigned long *)a |
650 (__force unsigned long)(a->s6_addr32[0] | a->s6_addr32[1]) |
652 (__force unsigned long)(a->s6_addr32[2] ^
653 cpu_to_be32(0x0000ffff))) == 0UL;
656 static inline u32 ipv6_portaddr_hash(const struct net *net,
657 const struct in6_addr *addr6,
660 unsigned int hash, mix = net_hash_mix(net);
662 if (ipv6_addr_any(addr6))
663 hash = jhash_1word(0, mix);
664 else if (ipv6_addr_v4mapped(addr6))
665 hash = jhash_1word((__force u32)addr6->s6_addr32[3], mix);
667 hash = jhash2((__force u32 *)addr6->s6_addr32, 4, mix);
673 * Check for a RFC 4843 ORCHID address
674 * (Overlay Routable Cryptographic Hash Identifiers)
676 static inline bool ipv6_addr_orchid(const struct in6_addr *a)
678 return (a->s6_addr32[0] & htonl(0xfffffff0)) == htonl(0x20010010);
681 static inline bool ipv6_addr_is_multicast(const struct in6_addr *addr)
683 return (addr->s6_addr32[0] & htonl(0xFF000000)) == htonl(0xFF000000);
686 static inline void ipv6_addr_set_v4mapped(const __be32 addr,
687 struct in6_addr *v4mapped)
689 ipv6_addr_set(v4mapped,
696 * find the first different bit between two addresses
697 * length of address must be a multiple of 32bits
699 static inline int __ipv6_addr_diff32(const void *token1, const void *token2, int addrlen)
701 const __be32 *a1 = token1, *a2 = token2;
706 for (i = 0; i < addrlen; i++) {
707 __be32 xb = a1[i] ^ a2[i];
709 return i * 32 + 31 - __fls(ntohl(xb));
713 * we should *never* get to this point since that
714 * would mean the addrs are equal
716 * However, we do get to it 8) And exacly, when
717 * addresses are equal 8)
719 * ip route add 1111::/128 via ...
720 * ip route add 1111::/64 via ...
723 * Ideally, this function should stop comparison
724 * at prefix length. It does not, but it is still OK,
725 * if returned value is greater than prefix length.
731 #if defined(CONFIG_HAVE_EFFICIENT_UNALIGNED_ACCESS) && BITS_PER_LONG == 64
732 static inline int __ipv6_addr_diff64(const void *token1, const void *token2, int addrlen)
734 const __be64 *a1 = token1, *a2 = token2;
739 for (i = 0; i < addrlen; i++) {
740 __be64 xb = a1[i] ^ a2[i];
742 return i * 64 + 63 - __fls(be64_to_cpu(xb));
749 static inline int __ipv6_addr_diff(const void *token1, const void *token2, int addrlen)
751 #if defined(CONFIG_HAVE_EFFICIENT_UNALIGNED_ACCESS) && BITS_PER_LONG == 64
752 if (__builtin_constant_p(addrlen) && !(addrlen & 7))
753 return __ipv6_addr_diff64(token1, token2, addrlen);
755 return __ipv6_addr_diff32(token1, token2, addrlen);
758 static inline int ipv6_addr_diff(const struct in6_addr *a1, const struct in6_addr *a2)
760 return __ipv6_addr_diff(a1, a2, sizeof(struct in6_addr));
763 __be32 ipv6_select_ident(struct net *net,
764 const struct in6_addr *daddr,
765 const struct in6_addr *saddr);
766 __be32 ipv6_proxy_select_ident(struct net *net, struct sk_buff *skb);
768 int ip6_dst_hoplimit(struct dst_entry *dst);
770 static inline int ip6_sk_dst_hoplimit(struct ipv6_pinfo *np, struct flowi6 *fl6,
771 struct dst_entry *dst)
775 if (ipv6_addr_is_multicast(&fl6->daddr))
776 hlimit = np->mcast_hops;
778 hlimit = np->hop_limit;
780 hlimit = ip6_dst_hoplimit(dst);
784 /* copy IPv6 saddr & daddr to flow_keys, possibly using 64bit load/store
785 * Equivalent to : flow->v6addrs.src = iph->saddr;
786 * flow->v6addrs.dst = iph->daddr;
788 static inline void iph_to_flow_copy_v6addrs(struct flow_keys *flow,
789 const struct ipv6hdr *iph)
791 BUILD_BUG_ON(offsetof(typeof(flow->addrs), v6addrs.dst) !=
792 offsetof(typeof(flow->addrs), v6addrs.src) +
793 sizeof(flow->addrs.v6addrs.src));
794 memcpy(&flow->addrs.v6addrs, &iph->saddr, sizeof(flow->addrs.v6addrs));
795 flow->control.addr_type = FLOW_DISSECTOR_KEY_IPV6_ADDRS;
798 #if IS_ENABLED(CONFIG_IPV6)
800 /* Sysctl settings for net ipv6.auto_flowlabels */
801 #define IP6_AUTO_FLOW_LABEL_OFF 0
802 #define IP6_AUTO_FLOW_LABEL_OPTOUT 1
803 #define IP6_AUTO_FLOW_LABEL_OPTIN 2
804 #define IP6_AUTO_FLOW_LABEL_FORCED 3
806 #define IP6_AUTO_FLOW_LABEL_MAX IP6_AUTO_FLOW_LABEL_FORCED
808 #define IP6_DEFAULT_AUTO_FLOW_LABELS IP6_AUTO_FLOW_LABEL_OPTOUT
810 static inline __be32 ip6_make_flowlabel(struct net *net, struct sk_buff *skb,
811 __be32 flowlabel, bool autolabel,
816 /* @flowlabel may include more than a flow label, eg, the traffic class.
817 * Here we want only the flow label value.
819 flowlabel &= IPV6_FLOWLABEL_MASK;
822 net->ipv6.sysctl.auto_flowlabels == IP6_AUTO_FLOW_LABEL_OFF ||
824 net->ipv6.sysctl.auto_flowlabels != IP6_AUTO_FLOW_LABEL_FORCED))
827 hash = skb_get_hash_flowi6(skb, fl6);
829 /* Since this is being sent on the wire obfuscate hash a bit
830 * to minimize possbility that any useful information to an
831 * attacker is leaked. Only lower 20 bits are relevant.
835 flowlabel = (__force __be32)hash & IPV6_FLOWLABEL_MASK;
837 if (net->ipv6.sysctl.flowlabel_state_ranges)
838 flowlabel |= IPV6_FLOWLABEL_STATELESS_FLAG;
843 static inline int ip6_default_np_autolabel(struct net *net)
845 switch (net->ipv6.sysctl.auto_flowlabels) {
846 case IP6_AUTO_FLOW_LABEL_OFF:
847 case IP6_AUTO_FLOW_LABEL_OPTIN:
850 case IP6_AUTO_FLOW_LABEL_OPTOUT:
851 case IP6_AUTO_FLOW_LABEL_FORCED:
856 static inline void ip6_set_txhash(struct sock *sk) { }
857 static inline __be32 ip6_make_flowlabel(struct net *net, struct sk_buff *skb,
858 __be32 flowlabel, bool autolabel,
863 static inline int ip6_default_np_autolabel(struct net *net)
869 #if IS_ENABLED(CONFIG_IPV6)
870 static inline int ip6_multipath_hash_policy(const struct net *net)
872 return net->ipv6.sysctl.multipath_hash_policy;
875 static inline int ip6_multipath_hash_policy(const struct net *net)
882 * Header manipulation
884 static inline void ip6_flow_hdr(struct ipv6hdr *hdr, unsigned int tclass,
887 *(__be32 *)hdr = htonl(0x60000000 | (tclass << 20)) | flowlabel;
890 static inline __be32 ip6_flowinfo(const struct ipv6hdr *hdr)
892 return *(__be32 *)hdr & IPV6_FLOWINFO_MASK;
895 static inline __be32 ip6_flowlabel(const struct ipv6hdr *hdr)
897 return *(__be32 *)hdr & IPV6_FLOWLABEL_MASK;
900 static inline u8 ip6_tclass(__be32 flowinfo)
902 return ntohl(flowinfo & IPV6_TCLASS_MASK) >> IPV6_TCLASS_SHIFT;
905 static inline __be32 ip6_make_flowinfo(unsigned int tclass, __be32 flowlabel)
907 return htonl(tclass << IPV6_TCLASS_SHIFT) | flowlabel;
910 static inline __be32 flowi6_get_flowlabel(const struct flowi6 *fl6)
912 return fl6->flowlabel & IPV6_FLOWLABEL_MASK;
916 * Prototypes exported by ipv6
920 * rcv function (called from netdevice level)
923 int ipv6_rcv(struct sk_buff *skb, struct net_device *dev,
924 struct packet_type *pt, struct net_device *orig_dev);
926 int ip6_rcv_finish(struct net *net, struct sock *sk, struct sk_buff *skb);
929 * upper-layer output functions
931 int ip6_xmit(const struct sock *sk, struct sk_buff *skb, struct flowi6 *fl6,
932 __u32 mark, struct ipv6_txoptions *opt, int tclass);
934 int ip6_find_1stfragopt(struct sk_buff *skb, u8 **nexthdr);
936 int ip6_append_data(struct sock *sk,
937 int getfrag(void *from, char *to, int offset, int len,
938 int odd, struct sk_buff *skb),
939 void *from, int length, int transhdrlen,
940 struct ipcm6_cookie *ipc6, struct flowi6 *fl6,
941 struct rt6_info *rt, unsigned int flags,
942 const struct sockcm_cookie *sockc);
944 int ip6_push_pending_frames(struct sock *sk);
946 void ip6_flush_pending_frames(struct sock *sk);
948 int ip6_send_skb(struct sk_buff *skb);
950 struct sk_buff *__ip6_make_skb(struct sock *sk, struct sk_buff_head *queue,
951 struct inet_cork_full *cork,
952 struct inet6_cork *v6_cork);
953 struct sk_buff *ip6_make_skb(struct sock *sk,
954 int getfrag(void *from, char *to, int offset,
955 int len, int odd, struct sk_buff *skb),
956 void *from, int length, int transhdrlen,
957 struct ipcm6_cookie *ipc6, struct flowi6 *fl6,
958 struct rt6_info *rt, unsigned int flags,
959 struct inet_cork_full *cork,
960 const struct sockcm_cookie *sockc);
962 static inline struct sk_buff *ip6_finish_skb(struct sock *sk)
964 return __ip6_make_skb(sk, &sk->sk_write_queue, &inet_sk(sk)->cork,
965 &inet6_sk(sk)->cork);
968 int ip6_dst_lookup(struct net *net, struct sock *sk, struct dst_entry **dst,
970 struct dst_entry *ip6_dst_lookup_flow(const struct sock *sk, struct flowi6 *fl6,
971 const struct in6_addr *final_dst);
972 struct dst_entry *ip6_sk_dst_lookup_flow(struct sock *sk, struct flowi6 *fl6,
973 const struct in6_addr *final_dst,
975 struct dst_entry *ip6_blackhole_route(struct net *net,
976 struct dst_entry *orig_dst);
979 * skb processing functions
982 int ip6_output(struct net *net, struct sock *sk, struct sk_buff *skb);
983 int ip6_forward(struct sk_buff *skb);
984 int ip6_input(struct sk_buff *skb);
985 int ip6_mc_input(struct sk_buff *skb);
987 int __ip6_local_out(struct net *net, struct sock *sk, struct sk_buff *skb);
988 int ip6_local_out(struct net *net, struct sock *sk, struct sk_buff *skb);
991 * Extension header (options) processing
994 void ipv6_push_nfrag_opts(struct sk_buff *skb, struct ipv6_txoptions *opt,
995 u8 *proto, struct in6_addr **daddr_p,
996 struct in6_addr *saddr);
997 void ipv6_push_frag_opts(struct sk_buff *skb, struct ipv6_txoptions *opt,
1000 int ipv6_skip_exthdr(const struct sk_buff *, int start, u8 *nexthdrp,
1003 bool ipv6_ext_hdr(u8 nexthdr);
1006 IP6_FH_F_FRAG = (1 << 0),
1007 IP6_FH_F_AUTH = (1 << 1),
1008 IP6_FH_F_SKIP_RH = (1 << 2),
1011 /* find specified header and get offset to it */
1012 int ipv6_find_hdr(const struct sk_buff *skb, unsigned int *offset, int target,
1013 unsigned short *fragoff, int *fragflg);
1015 int ipv6_find_tlv(const struct sk_buff *skb, int offset, int type);
1017 struct in6_addr *fl6_update_dst(struct flowi6 *fl6,
1018 const struct ipv6_txoptions *opt,
1019 struct in6_addr *orig);
1022 * socket options (ipv6_sockglue.c)
1025 int ipv6_setsockopt(struct sock *sk, int level, int optname,
1026 char __user *optval, unsigned int optlen);
1027 int ipv6_getsockopt(struct sock *sk, int level, int optname,
1028 char __user *optval, int __user *optlen);
1029 int compat_ipv6_setsockopt(struct sock *sk, int level, int optname,
1030 char __user *optval, unsigned int optlen);
1031 int compat_ipv6_getsockopt(struct sock *sk, int level, int optname,
1032 char __user *optval, int __user *optlen);
1034 int __ip6_datagram_connect(struct sock *sk, struct sockaddr *addr,
1036 int ip6_datagram_connect(struct sock *sk, struct sockaddr *addr, int addr_len);
1037 int ip6_datagram_connect_v6_only(struct sock *sk, struct sockaddr *addr,
1039 int ip6_datagram_dst_update(struct sock *sk, bool fix_sk_saddr);
1040 void ip6_datagram_release_cb(struct sock *sk);
1042 int ipv6_recv_error(struct sock *sk, struct msghdr *msg, int len,
1044 int ipv6_recv_rxpmtu(struct sock *sk, struct msghdr *msg, int len,
1046 void ipv6_icmp_error(struct sock *sk, struct sk_buff *skb, int err, __be16 port,
1047 u32 info, u8 *payload);
1048 void ipv6_local_error(struct sock *sk, int err, struct flowi6 *fl6, u32 info);
1049 void ipv6_local_rxpmtu(struct sock *sk, struct flowi6 *fl6, u32 mtu);
1051 int inet6_release(struct socket *sock);
1052 int inet6_bind(struct socket *sock, struct sockaddr *uaddr, int addr_len);
1053 int inet6_getname(struct socket *sock, struct sockaddr *uaddr,
1055 int inet6_ioctl(struct socket *sock, unsigned int cmd, unsigned long arg);
1057 int inet6_hash_connect(struct inet_timewait_death_row *death_row,
1063 extern const struct proto_ops inet6_stream_ops;
1064 extern const struct proto_ops inet6_dgram_ops;
1065 extern const struct proto_ops inet6_sockraw_ops;
1067 struct group_source_req;
1068 struct group_filter;
1070 int ip6_mc_source(int add, int omode, struct sock *sk,
1071 struct group_source_req *pgsr);
1072 int ip6_mc_msfilter(struct sock *sk, struct group_filter *gsf);
1073 int ip6_mc_msfget(struct sock *sk, struct group_filter *gsf,
1074 struct group_filter __user *optval, int __user *optlen);
1076 #ifdef CONFIG_PROC_FS
1077 int ac6_proc_init(struct net *net);
1078 void ac6_proc_exit(struct net *net);
1079 int raw6_proc_init(void);
1080 void raw6_proc_exit(void);
1081 int tcp6_proc_init(struct net *net);
1082 void tcp6_proc_exit(struct net *net);
1083 int udp6_proc_init(struct net *net);
1084 void udp6_proc_exit(struct net *net);
1085 int udplite6_proc_init(void);
1086 void udplite6_proc_exit(void);
1087 int ipv6_misc_proc_init(void);
1088 void ipv6_misc_proc_exit(void);
1089 int snmp6_register_dev(struct inet6_dev *idev);
1090 int snmp6_unregister_dev(struct inet6_dev *idev);
1093 static inline int ac6_proc_init(struct net *net) { return 0; }
1094 static inline void ac6_proc_exit(struct net *net) { }
1095 static inline int snmp6_register_dev(struct inet6_dev *idev) { return 0; }
1096 static inline int snmp6_unregister_dev(struct inet6_dev *idev) { return 0; }
1099 #ifdef CONFIG_SYSCTL
1100 extern struct ctl_table ipv6_route_table_template[];
1102 struct ctl_table *ipv6_icmp_sysctl_init(struct net *net);
1103 struct ctl_table *ipv6_route_sysctl_init(struct net *net);
1104 int ipv6_sysctl_register(void);
1105 void ipv6_sysctl_unregister(void);
1108 int ipv6_sock_mc_join(struct sock *sk, int ifindex,
1109 const struct in6_addr *addr);
1110 int ipv6_sock_mc_drop(struct sock *sk, int ifindex,
1111 const struct in6_addr *addr);
1112 #endif /* _NET_IPV6_H */