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_IPV4 4 /* IPv4 in IPv6 */
34 #define NEXTHDR_TCP 6 /* TCP segment. */
35 #define NEXTHDR_UDP 17 /* UDP message. */
36 #define NEXTHDR_IPV6 41 /* IPv6 in IPv6 */
37 #define NEXTHDR_ROUTING 43 /* Routing header. */
38 #define NEXTHDR_FRAGMENT 44 /* Fragmentation/reassembly header. */
39 #define NEXTHDR_GRE 47 /* GRE header. */
40 #define NEXTHDR_ESP 50 /* Encapsulating security payload. */
41 #define NEXTHDR_AUTH 51 /* Authentication header. */
42 #define NEXTHDR_ICMP 58 /* ICMP for IPv6. */
43 #define NEXTHDR_NONE 59 /* No next header */
44 #define NEXTHDR_DEST 60 /* Destination options header. */
45 #define NEXTHDR_SCTP 132 /* SCTP message. */
46 #define NEXTHDR_MOBILITY 135 /* Mobility header. */
48 #define NEXTHDR_MAX 255
50 #define IPV6_DEFAULT_HOPLIMIT 64
51 #define IPV6_DEFAULT_MCASTHOPS 1
53 /* Limits on Hop-by-Hop and Destination options.
55 * Per RFC8200 there is no limit on the maximum number or lengths of options in
56 * Hop-by-Hop or Destination options other then the packet must fit in an MTU.
57 * We allow configurable limits in order to mitigate potential denial of
60 * There are three limits that may be set:
61 * - Limit the number of options in a Hop-by-Hop or Destination options
63 * - Limit the byte length of a Hop-by-Hop or Destination options extension
65 * - Disallow unknown options
67 * The limits are expressed in corresponding sysctls:
69 * ipv6.sysctl.max_dst_opts_cnt
70 * ipv6.sysctl.max_hbh_opts_cnt
71 * ipv6.sysctl.max_dst_opts_len
72 * ipv6.sysctl.max_hbh_opts_len
74 * max_*_opts_cnt is the number of TLVs that are allowed for Destination
75 * options or Hop-by-Hop options. If the number is less than zero then unknown
76 * TLVs are disallowed and the number of known options that are allowed is the
77 * absolute value. Setting the value to INT_MAX indicates no limit.
79 * max_*_opts_len is the length limit in bytes of a Destination or
80 * Hop-by-Hop options extension header. Setting the value to INT_MAX
81 * indicates no length limit.
83 * If a limit is exceeded when processing an extension header the packet is
87 /* Default limits for Hop-by-Hop and Destination options */
88 #define IP6_DEFAULT_MAX_DST_OPTS_CNT 8
89 #define IP6_DEFAULT_MAX_HBH_OPTS_CNT 8
90 #define IP6_DEFAULT_MAX_DST_OPTS_LEN INT_MAX /* No limit */
91 #define IP6_DEFAULT_MAX_HBH_OPTS_LEN INT_MAX /* No limit */
96 * type - unicast | multicast
97 * scope - local | site | global
104 #define IPV6_ADDR_ANY 0x0000U
106 #define IPV6_ADDR_UNICAST 0x0001U
107 #define IPV6_ADDR_MULTICAST 0x0002U
109 #define IPV6_ADDR_LOOPBACK 0x0010U
110 #define IPV6_ADDR_LINKLOCAL 0x0020U
111 #define IPV6_ADDR_SITELOCAL 0x0040U
113 #define IPV6_ADDR_COMPATv4 0x0080U
115 #define IPV6_ADDR_SCOPE_MASK 0x00f0U
117 #define IPV6_ADDR_MAPPED 0x1000U
122 #define IPV6_ADDR_MC_SCOPE(a) \
123 ((a)->s6_addr[1] & 0x0f) /* nonstandard */
124 #define __IPV6_ADDR_SCOPE_INVALID -1
125 #define IPV6_ADDR_SCOPE_NODELOCAL 0x01
126 #define IPV6_ADDR_SCOPE_LINKLOCAL 0x02
127 #define IPV6_ADDR_SCOPE_SITELOCAL 0x05
128 #define IPV6_ADDR_SCOPE_ORGLOCAL 0x08
129 #define IPV6_ADDR_SCOPE_GLOBAL 0x0e
134 #define IPV6_ADDR_MC_FLAG_TRANSIENT(a) \
135 ((a)->s6_addr[1] & 0x10)
136 #define IPV6_ADDR_MC_FLAG_PREFIX(a) \
137 ((a)->s6_addr[1] & 0x20)
138 #define IPV6_ADDR_MC_FLAG_RENDEZVOUS(a) \
139 ((a)->s6_addr[1] & 0x40)
142 * fragmentation header
149 __be32 identification;
152 #define IP6_MF 0x0001
153 #define IP6_OFFSET 0xFFF8
155 struct ip6_fraglist_iter {
156 struct ipv6hdr *tmp_hdr;
157 struct sk_buff *frag;
164 int ip6_fraglist_init(struct sk_buff *skb, unsigned int hlen, u8 *prevhdr,
165 u8 nexthdr, __be32 frag_id,
166 struct ip6_fraglist_iter *iter);
167 void ip6_fraglist_prepare(struct sk_buff *skb, struct ip6_fraglist_iter *iter);
169 static inline struct sk_buff *ip6_fraglist_next(struct ip6_fraglist_iter *iter)
171 struct sk_buff *skb = iter->frag;
173 iter->frag = skb->next;
174 skb_mark_not_on_list(skb);
179 struct ip6_frag_state {
192 void ip6_frag_init(struct sk_buff *skb, unsigned int hlen, unsigned int mtu,
193 unsigned short needed_tailroom, int hdr_room, u8 *prevhdr,
194 u8 nexthdr, __be32 frag_id, struct ip6_frag_state *state);
195 struct sk_buff *ip6_frag_next(struct sk_buff *skb,
196 struct ip6_frag_state *state);
198 #define IP6_REPLY_MARK(net, mark) \
199 ((net)->ipv6.sysctl.fwmark_reflect ? (mark) : 0)
201 #include <net/sock.h>
204 extern int sysctl_mld_max_msf;
205 extern int sysctl_mld_qrv;
207 #define _DEVINC(net, statname, mod, idev, field) \
209 struct inet6_dev *_idev = (idev); \
210 if (likely(_idev != NULL)) \
211 mod##SNMP_INC_STATS64((_idev)->stats.statname, (field));\
212 mod##SNMP_INC_STATS64((net)->mib.statname##_statistics, (field));\
215 /* per device counters are atomic_long_t */
216 #define _DEVINCATOMIC(net, statname, mod, idev, field) \
218 struct inet6_dev *_idev = (idev); \
219 if (likely(_idev != NULL)) \
220 SNMP_INC_STATS_ATOMIC_LONG((_idev)->stats.statname##dev, (field)); \
221 mod##SNMP_INC_STATS((net)->mib.statname##_statistics, (field));\
224 /* per device and per net counters are atomic_long_t */
225 #define _DEVINC_ATOMIC_ATOMIC(net, statname, idev, field) \
227 struct inet6_dev *_idev = (idev); \
228 if (likely(_idev != NULL)) \
229 SNMP_INC_STATS_ATOMIC_LONG((_idev)->stats.statname##dev, (field)); \
230 SNMP_INC_STATS_ATOMIC_LONG((net)->mib.statname##_statistics, (field));\
233 #define _DEVADD(net, statname, mod, idev, field, val) \
235 struct inet6_dev *_idev = (idev); \
236 if (likely(_idev != NULL)) \
237 mod##SNMP_ADD_STATS((_idev)->stats.statname, (field), (val)); \
238 mod##SNMP_ADD_STATS((net)->mib.statname##_statistics, (field), (val));\
241 #define _DEVUPD(net, statname, mod, idev, field, val) \
243 struct inet6_dev *_idev = (idev); \
244 if (likely(_idev != NULL)) \
245 mod##SNMP_UPD_PO_STATS((_idev)->stats.statname, field, (val)); \
246 mod##SNMP_UPD_PO_STATS((net)->mib.statname##_statistics, field, (val));\
251 #define IP6_INC_STATS(net, idev,field) \
252 _DEVINC(net, ipv6, , idev, field)
253 #define __IP6_INC_STATS(net, idev,field) \
254 _DEVINC(net, ipv6, __, idev, field)
255 #define IP6_ADD_STATS(net, idev,field,val) \
256 _DEVADD(net, ipv6, , idev, field, val)
257 #define __IP6_ADD_STATS(net, idev,field,val) \
258 _DEVADD(net, ipv6, __, idev, field, val)
259 #define IP6_UPD_PO_STATS(net, idev,field,val) \
260 _DEVUPD(net, ipv6, , idev, field, val)
261 #define __IP6_UPD_PO_STATS(net, idev,field,val) \
262 _DEVUPD(net, ipv6, __, idev, field, val)
263 #define ICMP6_INC_STATS(net, idev, field) \
264 _DEVINCATOMIC(net, icmpv6, , idev, field)
265 #define __ICMP6_INC_STATS(net, idev, field) \
266 _DEVINCATOMIC(net, icmpv6, __, idev, field)
268 #define ICMP6MSGOUT_INC_STATS(net, idev, field) \
269 _DEVINC_ATOMIC_ATOMIC(net, icmpv6msg, idev, field +256)
270 #define ICMP6MSGIN_INC_STATS(net, idev, field) \
271 _DEVINC_ATOMIC_ATOMIC(net, icmpv6msg, idev, field)
273 struct ip6_ra_chain {
274 struct ip6_ra_chain *next;
277 void (*destructor)(struct sock *);
280 extern struct ip6_ra_chain *ip6_ra_chain;
281 extern rwlock_t ip6_ra_lock;
284 This structure is prepared by protocol, when parsing
285 ancillary data and passed to IPv6.
288 struct ipv6_txoptions {
290 /* Length of this structure */
293 /* length of extension headers */
295 __u16 opt_flen; /* after fragment hdr */
296 __u16 opt_nflen; /* before fragment hdr */
298 struct ipv6_opt_hdr *hopopt;
299 struct ipv6_opt_hdr *dst0opt;
300 struct ipv6_rt_hdr *srcrt; /* Routing Header */
301 struct ipv6_opt_hdr *dst1opt;
303 /* Option buffer, as read by IPV6_PKTOPTIONS, starts here. */
306 /* flowlabel_reflect sysctl values */
307 enum flowlabel_reflect {
308 FLOWLABEL_REFLECT_ESTABLISHED = 1,
309 FLOWLABEL_REFLECT_TCP_RESET = 2,
310 FLOWLABEL_REFLECT_ICMPV6_ECHO_REPLIES = 4,
313 struct ip6_flowlabel {
314 struct ip6_flowlabel __rcu *next;
318 struct ipv6_txoptions *opt;
319 unsigned long linger;
326 unsigned long lastuse;
327 unsigned long expires;
331 #define IPV6_FLOWINFO_MASK cpu_to_be32(0x0FFFFFFF)
332 #define IPV6_FLOWLABEL_MASK cpu_to_be32(0x000FFFFF)
333 #define IPV6_FLOWLABEL_STATELESS_FLAG cpu_to_be32(0x00080000)
335 #define IPV6_TCLASS_MASK (IPV6_FLOWINFO_MASK & ~IPV6_FLOWLABEL_MASK)
336 #define IPV6_TCLASS_SHIFT 20
338 struct ipv6_fl_socklist {
339 struct ipv6_fl_socklist __rcu *next;
340 struct ip6_flowlabel *fl;
344 struct ipcm6_cookie {
345 struct sockcm_cookie sockc;
349 struct ipv6_txoptions *opt;
353 static inline void ipcm6_init(struct ipcm6_cookie *ipc6)
355 *ipc6 = (struct ipcm6_cookie) {
362 static inline void ipcm6_init_sk(struct ipcm6_cookie *ipc6,
363 const struct ipv6_pinfo *np)
365 *ipc6 = (struct ipcm6_cookie) {
367 .tclass = np->tclass,
368 .dontfrag = np->dontfrag,
372 static inline struct ipv6_txoptions *txopt_get(const struct ipv6_pinfo *np)
374 struct ipv6_txoptions *opt;
377 opt = rcu_dereference(np->opt);
379 if (!refcount_inc_not_zero(&opt->refcnt))
382 opt = rcu_pointer_handoff(opt);
388 static inline void txopt_put(struct ipv6_txoptions *opt)
390 if (opt && refcount_dec_and_test(&opt->refcnt))
394 #if IS_ENABLED(CONFIG_IPV6)
395 struct ip6_flowlabel *__fl6_sock_lookup(struct sock *sk, __be32 label);
397 extern struct static_key_false_deferred ipv6_flowlabel_exclusive;
398 static inline struct ip6_flowlabel *fl6_sock_lookup(struct sock *sk,
401 if (static_branch_unlikely(&ipv6_flowlabel_exclusive.key) &&
402 READ_ONCE(sock_net(sk)->ipv6.flowlabel_has_excl))
403 return __fl6_sock_lookup(sk, label) ? : ERR_PTR(-ENOENT);
409 struct ipv6_txoptions *fl6_merge_options(struct ipv6_txoptions *opt_space,
410 struct ip6_flowlabel *fl,
411 struct ipv6_txoptions *fopt);
412 void fl6_free_socklist(struct sock *sk);
413 int ipv6_flowlabel_opt(struct sock *sk, sockptr_t optval, int optlen);
414 int ipv6_flowlabel_opt_get(struct sock *sk, struct in6_flowlabel_req *freq,
416 int ip6_flowlabel_init(void);
417 void ip6_flowlabel_cleanup(void);
418 bool ip6_autoflowlabel(struct net *net, const struct ipv6_pinfo *np);
420 static inline void fl6_sock_release(struct ip6_flowlabel *fl)
423 atomic_dec(&fl->users);
426 void icmpv6_notify(struct sk_buff *skb, u8 type, u8 code, __be32 info);
428 void icmpv6_push_pending_frames(struct sock *sk, struct flowi6 *fl6,
429 struct icmp6hdr *thdr, int len);
431 int ip6_ra_control(struct sock *sk, int sel);
433 int ipv6_parse_hopopts(struct sk_buff *skb);
435 struct ipv6_txoptions *ipv6_dup_options(struct sock *sk,
436 struct ipv6_txoptions *opt);
437 struct ipv6_txoptions *ipv6_renew_options(struct sock *sk,
438 struct ipv6_txoptions *opt,
440 struct ipv6_opt_hdr *newopt);
441 struct ipv6_txoptions *ipv6_fixup_options(struct ipv6_txoptions *opt_space,
442 struct ipv6_txoptions *opt);
444 bool ipv6_opt_accepted(const struct sock *sk, const struct sk_buff *skb,
445 const struct inet6_skb_parm *opt);
446 struct ipv6_txoptions *ipv6_update_options(struct sock *sk,
447 struct ipv6_txoptions *opt);
449 static inline bool ipv6_accept_ra(struct inet6_dev *idev)
451 /* If forwarding is enabled, RA are not accepted unless the special
452 * hybrid mode (accept_ra=2) is enabled.
454 return idev->cnf.forwarding ? idev->cnf.accept_ra == 2 :
458 #define IPV6_FRAG_HIGH_THRESH (4 * 1024*1024) /* 4194304 */
459 #define IPV6_FRAG_LOW_THRESH (3 * 1024*1024) /* 3145728 */
460 #define IPV6_FRAG_TIMEOUT (60 * HZ) /* 60 seconds */
462 int __ipv6_addr_type(const struct in6_addr *addr);
463 static inline int ipv6_addr_type(const struct in6_addr *addr)
465 return __ipv6_addr_type(addr) & 0xffff;
468 static inline int ipv6_addr_scope(const struct in6_addr *addr)
470 return __ipv6_addr_type(addr) & IPV6_ADDR_SCOPE_MASK;
473 static inline int __ipv6_addr_src_scope(int type)
475 return (type == IPV6_ADDR_ANY) ? __IPV6_ADDR_SCOPE_INVALID : (type >> 16);
478 static inline int ipv6_addr_src_scope(const struct in6_addr *addr)
480 return __ipv6_addr_src_scope(__ipv6_addr_type(addr));
483 static inline bool __ipv6_addr_needs_scope_id(int type)
485 return type & IPV6_ADDR_LINKLOCAL ||
486 (type & IPV6_ADDR_MULTICAST &&
487 (type & (IPV6_ADDR_LOOPBACK|IPV6_ADDR_LINKLOCAL)));
490 static inline __u32 ipv6_iface_scope_id(const struct in6_addr *addr, int iface)
492 return __ipv6_addr_needs_scope_id(__ipv6_addr_type(addr)) ? iface : 0;
495 static inline int ipv6_addr_cmp(const struct in6_addr *a1, const struct in6_addr *a2)
497 return memcmp(a1, a2, sizeof(struct in6_addr));
501 ipv6_masked_addr_cmp(const struct in6_addr *a1, const struct in6_addr *m,
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 *ulm = (const unsigned long *)m;
507 const unsigned long *ul2 = (const unsigned long *)a2;
509 return !!(((ul1[0] ^ ul2[0]) & ulm[0]) |
510 ((ul1[1] ^ ul2[1]) & ulm[1]));
512 return !!(((a1->s6_addr32[0] ^ a2->s6_addr32[0]) & m->s6_addr32[0]) |
513 ((a1->s6_addr32[1] ^ a2->s6_addr32[1]) & m->s6_addr32[1]) |
514 ((a1->s6_addr32[2] ^ a2->s6_addr32[2]) & m->s6_addr32[2]) |
515 ((a1->s6_addr32[3] ^ a2->s6_addr32[3]) & m->s6_addr32[3]));
519 static inline void ipv6_addr_prefix(struct in6_addr *pfx,
520 const struct in6_addr *addr,
523 /* caller must guarantee 0 <= plen <= 128 */
527 memset(pfx->s6_addr, 0, sizeof(pfx->s6_addr));
528 memcpy(pfx->s6_addr, addr, o);
530 pfx->s6_addr[o] = addr->s6_addr[o] & (0xff00 >> b);
533 static inline void ipv6_addr_prefix_copy(struct in6_addr *addr,
534 const struct in6_addr *pfx,
537 /* caller must guarantee 0 <= plen <= 128 */
541 memcpy(addr->s6_addr, pfx, o);
543 addr->s6_addr[o] &= ~(0xff00 >> b);
544 addr->s6_addr[o] |= (pfx->s6_addr[o] & (0xff00 >> b));
548 static inline void __ipv6_addr_set_half(__be32 *addr,
549 __be32 wh, __be32 wl)
551 #if defined(CONFIG_HAVE_EFFICIENT_UNALIGNED_ACCESS) && BITS_PER_LONG == 64
552 #if defined(__BIG_ENDIAN)
553 if (__builtin_constant_p(wh) && __builtin_constant_p(wl)) {
554 *(__force u64 *)addr = ((__force u64)(wh) << 32 | (__force u64)(wl));
557 #elif defined(__LITTLE_ENDIAN)
558 if (__builtin_constant_p(wl) && __builtin_constant_p(wh)) {
559 *(__force u64 *)addr = ((__force u64)(wl) << 32 | (__force u64)(wh));
568 static inline void ipv6_addr_set(struct in6_addr *addr,
569 __be32 w1, __be32 w2,
570 __be32 w3, __be32 w4)
572 __ipv6_addr_set_half(&addr->s6_addr32[0], w1, w2);
573 __ipv6_addr_set_half(&addr->s6_addr32[2], w3, w4);
576 static inline bool ipv6_addr_equal(const struct in6_addr *a1,
577 const struct in6_addr *a2)
579 #if defined(CONFIG_HAVE_EFFICIENT_UNALIGNED_ACCESS) && BITS_PER_LONG == 64
580 const unsigned long *ul1 = (const unsigned long *)a1;
581 const unsigned long *ul2 = (const unsigned long *)a2;
583 return ((ul1[0] ^ ul2[0]) | (ul1[1] ^ ul2[1])) == 0UL;
585 return ((a1->s6_addr32[0] ^ a2->s6_addr32[0]) |
586 (a1->s6_addr32[1] ^ a2->s6_addr32[1]) |
587 (a1->s6_addr32[2] ^ a2->s6_addr32[2]) |
588 (a1->s6_addr32[3] ^ a2->s6_addr32[3])) == 0;
592 #if defined(CONFIG_HAVE_EFFICIENT_UNALIGNED_ACCESS) && BITS_PER_LONG == 64
593 static inline bool __ipv6_prefix_equal64_half(const __be64 *a1,
597 if (len && ((*a1 ^ *a2) & cpu_to_be64((~0UL) << (64 - len))))
602 static inline bool ipv6_prefix_equal(const struct in6_addr *addr1,
603 const struct in6_addr *addr2,
604 unsigned int prefixlen)
606 const __be64 *a1 = (const __be64 *)addr1;
607 const __be64 *a2 = (const __be64 *)addr2;
609 if (prefixlen >= 64) {
612 return __ipv6_prefix_equal64_half(a1 + 1, a2 + 1, prefixlen - 64);
614 return __ipv6_prefix_equal64_half(a1, a2, prefixlen);
617 static inline bool ipv6_prefix_equal(const struct in6_addr *addr1,
618 const struct in6_addr *addr2,
619 unsigned int prefixlen)
621 const __be32 *a1 = addr1->s6_addr32;
622 const __be32 *a2 = addr2->s6_addr32;
623 unsigned int pdw, pbi;
625 /* check complete u32 in prefix */
626 pdw = prefixlen >> 5;
627 if (pdw && memcmp(a1, a2, pdw << 2))
630 /* check incomplete u32 in prefix */
631 pbi = prefixlen & 0x1f;
632 if (pbi && ((a1[pdw] ^ a2[pdw]) & htonl((0xffffffff) << (32 - pbi))))
639 static inline bool ipv6_addr_any(const struct in6_addr *a)
641 #if defined(CONFIG_HAVE_EFFICIENT_UNALIGNED_ACCESS) && BITS_PER_LONG == 64
642 const unsigned long *ul = (const unsigned long *)a;
644 return (ul[0] | ul[1]) == 0UL;
646 return (a->s6_addr32[0] | a->s6_addr32[1] |
647 a->s6_addr32[2] | a->s6_addr32[3]) == 0;
651 static inline u32 ipv6_addr_hash(const struct in6_addr *a)
653 #if defined(CONFIG_HAVE_EFFICIENT_UNALIGNED_ACCESS) && BITS_PER_LONG == 64
654 const unsigned long *ul = (const unsigned long *)a;
655 unsigned long x = ul[0] ^ ul[1];
657 return (u32)(x ^ (x >> 32));
659 return (__force u32)(a->s6_addr32[0] ^ a->s6_addr32[1] ^
660 a->s6_addr32[2] ^ a->s6_addr32[3]);
664 /* more secured version of ipv6_addr_hash() */
665 static inline u32 __ipv6_addr_jhash(const struct in6_addr *a, const u32 initval)
667 u32 v = (__force u32)a->s6_addr32[0] ^ (__force u32)a->s6_addr32[1];
669 return jhash_3words(v,
670 (__force u32)a->s6_addr32[2],
671 (__force u32)a->s6_addr32[3],
675 static inline bool ipv6_addr_loopback(const struct in6_addr *a)
677 #if defined(CONFIG_HAVE_EFFICIENT_UNALIGNED_ACCESS) && BITS_PER_LONG == 64
678 const __be64 *be = (const __be64 *)a;
680 return (be[0] | (be[1] ^ cpu_to_be64(1))) == 0UL;
682 return (a->s6_addr32[0] | a->s6_addr32[1] |
683 a->s6_addr32[2] | (a->s6_addr32[3] ^ cpu_to_be32(1))) == 0;
688 * Note that we must __force cast these to unsigned long to make sparse happy,
689 * since all of the endian-annotated types are fixed size regardless of arch.
691 static inline bool ipv6_addr_v4mapped(const struct in6_addr *a)
694 #if defined(CONFIG_HAVE_EFFICIENT_UNALIGNED_ACCESS) && BITS_PER_LONG == 64
695 *(unsigned long *)a |
697 (__force unsigned long)(a->s6_addr32[0] | a->s6_addr32[1]) |
699 (__force unsigned long)(a->s6_addr32[2] ^
700 cpu_to_be32(0x0000ffff))) == 0UL;
703 static inline bool ipv6_addr_v4mapped_loopback(const struct in6_addr *a)
705 return ipv6_addr_v4mapped(a) && ipv4_is_loopback(a->s6_addr32[3]);
708 static inline u32 ipv6_portaddr_hash(const struct net *net,
709 const struct in6_addr *addr6,
712 unsigned int hash, mix = net_hash_mix(net);
714 if (ipv6_addr_any(addr6))
715 hash = jhash_1word(0, mix);
716 else if (ipv6_addr_v4mapped(addr6))
717 hash = jhash_1word((__force u32)addr6->s6_addr32[3], mix);
719 hash = jhash2((__force u32 *)addr6->s6_addr32, 4, mix);
725 * Check for a RFC 4843 ORCHID address
726 * (Overlay Routable Cryptographic Hash Identifiers)
728 static inline bool ipv6_addr_orchid(const struct in6_addr *a)
730 return (a->s6_addr32[0] & htonl(0xfffffff0)) == htonl(0x20010010);
733 static inline bool ipv6_addr_is_multicast(const struct in6_addr *addr)
735 return (addr->s6_addr32[0] & htonl(0xFF000000)) == htonl(0xFF000000);
738 static inline void ipv6_addr_set_v4mapped(const __be32 addr,
739 struct in6_addr *v4mapped)
741 ipv6_addr_set(v4mapped,
748 * find the first different bit between two addresses
749 * length of address must be a multiple of 32bits
751 static inline int __ipv6_addr_diff32(const void *token1, const void *token2, int addrlen)
753 const __be32 *a1 = token1, *a2 = token2;
758 for (i = 0; i < addrlen; i++) {
759 __be32 xb = a1[i] ^ a2[i];
761 return i * 32 + 31 - __fls(ntohl(xb));
765 * we should *never* get to this point since that
766 * would mean the addrs are equal
768 * However, we do get to it 8) And exacly, when
769 * addresses are equal 8)
771 * ip route add 1111::/128 via ...
772 * ip route add 1111::/64 via ...
775 * Ideally, this function should stop comparison
776 * at prefix length. It does not, but it is still OK,
777 * if returned value is greater than prefix length.
783 #if defined(CONFIG_HAVE_EFFICIENT_UNALIGNED_ACCESS) && BITS_PER_LONG == 64
784 static inline int __ipv6_addr_diff64(const void *token1, const void *token2, int addrlen)
786 const __be64 *a1 = token1, *a2 = token2;
791 for (i = 0; i < addrlen; i++) {
792 __be64 xb = a1[i] ^ a2[i];
794 return i * 64 + 63 - __fls(be64_to_cpu(xb));
801 static inline int __ipv6_addr_diff(const void *token1, const void *token2, int addrlen)
803 #if defined(CONFIG_HAVE_EFFICIENT_UNALIGNED_ACCESS) && BITS_PER_LONG == 64
804 if (__builtin_constant_p(addrlen) && !(addrlen & 7))
805 return __ipv6_addr_diff64(token1, token2, addrlen);
807 return __ipv6_addr_diff32(token1, token2, addrlen);
810 static inline int ipv6_addr_diff(const struct in6_addr *a1, const struct in6_addr *a2)
812 return __ipv6_addr_diff(a1, a2, sizeof(struct in6_addr));
815 __be32 ipv6_select_ident(struct net *net,
816 const struct in6_addr *daddr,
817 const struct in6_addr *saddr);
818 __be32 ipv6_proxy_select_ident(struct net *net, struct sk_buff *skb);
820 int ip6_dst_hoplimit(struct dst_entry *dst);
822 static inline int ip6_sk_dst_hoplimit(struct ipv6_pinfo *np, struct flowi6 *fl6,
823 struct dst_entry *dst)
827 if (ipv6_addr_is_multicast(&fl6->daddr))
828 hlimit = np->mcast_hops;
830 hlimit = np->hop_limit;
832 hlimit = ip6_dst_hoplimit(dst);
836 /* copy IPv6 saddr & daddr to flow_keys, possibly using 64bit load/store
837 * Equivalent to : flow->v6addrs.src = iph->saddr;
838 * flow->v6addrs.dst = iph->daddr;
840 static inline void iph_to_flow_copy_v6addrs(struct flow_keys *flow,
841 const struct ipv6hdr *iph)
843 BUILD_BUG_ON(offsetof(typeof(flow->addrs), v6addrs.dst) !=
844 offsetof(typeof(flow->addrs), v6addrs.src) +
845 sizeof(flow->addrs.v6addrs.src));
846 memcpy(&flow->addrs.v6addrs, &iph->addrs, sizeof(flow->addrs.v6addrs));
847 flow->control.addr_type = FLOW_DISSECTOR_KEY_IPV6_ADDRS;
850 #if IS_ENABLED(CONFIG_IPV6)
852 static inline bool ipv6_can_nonlocal_bind(struct net *net,
853 struct inet_sock *inet)
855 return net->ipv6.sysctl.ip_nonlocal_bind ||
856 inet->freebind || inet->transparent;
859 /* Sysctl settings for net ipv6.auto_flowlabels */
860 #define IP6_AUTO_FLOW_LABEL_OFF 0
861 #define IP6_AUTO_FLOW_LABEL_OPTOUT 1
862 #define IP6_AUTO_FLOW_LABEL_OPTIN 2
863 #define IP6_AUTO_FLOW_LABEL_FORCED 3
865 #define IP6_AUTO_FLOW_LABEL_MAX IP6_AUTO_FLOW_LABEL_FORCED
867 #define IP6_DEFAULT_AUTO_FLOW_LABELS IP6_AUTO_FLOW_LABEL_OPTOUT
869 static inline __be32 ip6_make_flowlabel(struct net *net, struct sk_buff *skb,
870 __be32 flowlabel, bool autolabel,
875 /* @flowlabel may include more than a flow label, eg, the traffic class.
876 * Here we want only the flow label value.
878 flowlabel &= IPV6_FLOWLABEL_MASK;
881 net->ipv6.sysctl.auto_flowlabels == IP6_AUTO_FLOW_LABEL_OFF ||
883 net->ipv6.sysctl.auto_flowlabels != IP6_AUTO_FLOW_LABEL_FORCED))
886 hash = skb_get_hash_flowi6(skb, fl6);
888 /* Since this is being sent on the wire obfuscate hash a bit
889 * to minimize possbility that any useful information to an
890 * attacker is leaked. Only lower 20 bits are relevant.
892 hash = rol32(hash, 16);
894 flowlabel = (__force __be32)hash & IPV6_FLOWLABEL_MASK;
896 if (net->ipv6.sysctl.flowlabel_state_ranges)
897 flowlabel |= IPV6_FLOWLABEL_STATELESS_FLAG;
902 static inline int ip6_default_np_autolabel(struct net *net)
904 switch (net->ipv6.sysctl.auto_flowlabels) {
905 case IP6_AUTO_FLOW_LABEL_OFF:
906 case IP6_AUTO_FLOW_LABEL_OPTIN:
909 case IP6_AUTO_FLOW_LABEL_OPTOUT:
910 case IP6_AUTO_FLOW_LABEL_FORCED:
915 static inline __be32 ip6_make_flowlabel(struct net *net, struct sk_buff *skb,
916 __be32 flowlabel, bool autolabel,
921 static inline int ip6_default_np_autolabel(struct net *net)
927 #if IS_ENABLED(CONFIG_IPV6)
928 static inline int ip6_multipath_hash_policy(const struct net *net)
930 return net->ipv6.sysctl.multipath_hash_policy;
932 static inline u32 ip6_multipath_hash_fields(const struct net *net)
934 return net->ipv6.sysctl.multipath_hash_fields;
937 static inline int ip6_multipath_hash_policy(const struct net *net)
941 static inline u32 ip6_multipath_hash_fields(const struct net *net)
948 * Header manipulation
950 static inline void ip6_flow_hdr(struct ipv6hdr *hdr, unsigned int tclass,
953 *(__be32 *)hdr = htonl(0x60000000 | (tclass << 20)) | flowlabel;
956 static inline __be32 ip6_flowinfo(const struct ipv6hdr *hdr)
958 return *(__be32 *)hdr & IPV6_FLOWINFO_MASK;
961 static inline __be32 ip6_flowlabel(const struct ipv6hdr *hdr)
963 return *(__be32 *)hdr & IPV6_FLOWLABEL_MASK;
966 static inline u8 ip6_tclass(__be32 flowinfo)
968 return ntohl(flowinfo & IPV6_TCLASS_MASK) >> IPV6_TCLASS_SHIFT;
971 static inline __be32 ip6_make_flowinfo(unsigned int tclass, __be32 flowlabel)
973 return htonl(tclass << IPV6_TCLASS_SHIFT) | flowlabel;
976 static inline __be32 flowi6_get_flowlabel(const struct flowi6 *fl6)
978 return fl6->flowlabel & IPV6_FLOWLABEL_MASK;
982 * Prototypes exported by ipv6
986 * rcv function (called from netdevice level)
989 int ipv6_rcv(struct sk_buff *skb, struct net_device *dev,
990 struct packet_type *pt, struct net_device *orig_dev);
991 void ipv6_list_rcv(struct list_head *head, struct packet_type *pt,
992 struct net_device *orig_dev);
994 int ip6_rcv_finish(struct net *net, struct sock *sk, struct sk_buff *skb);
997 * upper-layer output functions
999 int ip6_xmit(const struct sock *sk, struct sk_buff *skb, struct flowi6 *fl6,
1000 __u32 mark, struct ipv6_txoptions *opt, int tclass, u32 priority);
1002 int ip6_find_1stfragopt(struct sk_buff *skb, u8 **nexthdr);
1004 int ip6_append_data(struct sock *sk,
1005 int getfrag(void *from, char *to, int offset, int len,
1006 int odd, struct sk_buff *skb),
1007 void *from, int length, int transhdrlen,
1008 struct ipcm6_cookie *ipc6, struct flowi6 *fl6,
1009 struct rt6_info *rt, unsigned int flags);
1011 int ip6_push_pending_frames(struct sock *sk);
1013 void ip6_flush_pending_frames(struct sock *sk);
1015 int ip6_send_skb(struct sk_buff *skb);
1017 struct sk_buff *__ip6_make_skb(struct sock *sk, struct sk_buff_head *queue,
1018 struct inet_cork_full *cork,
1019 struct inet6_cork *v6_cork);
1020 struct sk_buff *ip6_make_skb(struct sock *sk,
1021 int getfrag(void *from, char *to, int offset,
1022 int len, int odd, struct sk_buff *skb),
1023 void *from, int length, int transhdrlen,
1024 struct ipcm6_cookie *ipc6, struct flowi6 *fl6,
1025 struct rt6_info *rt, unsigned int flags,
1026 struct inet_cork_full *cork);
1028 static inline struct sk_buff *ip6_finish_skb(struct sock *sk)
1030 return __ip6_make_skb(sk, &sk->sk_write_queue, &inet_sk(sk)->cork,
1031 &inet6_sk(sk)->cork);
1034 int ip6_dst_lookup(struct net *net, struct sock *sk, struct dst_entry **dst,
1035 struct flowi6 *fl6);
1036 struct dst_entry *ip6_dst_lookup_flow(struct net *net, const struct sock *sk, struct flowi6 *fl6,
1037 const struct in6_addr *final_dst);
1038 struct dst_entry *ip6_sk_dst_lookup_flow(struct sock *sk, struct flowi6 *fl6,
1039 const struct in6_addr *final_dst,
1041 struct dst_entry *ip6_dst_lookup_tunnel(struct sk_buff *skb,
1042 struct net_device *dev,
1043 struct net *net, struct socket *sock,
1044 struct in6_addr *saddr,
1045 const struct ip_tunnel_info *info,
1046 u8 protocol, bool use_cache);
1047 struct dst_entry *ip6_blackhole_route(struct net *net,
1048 struct dst_entry *orig_dst);
1051 * skb processing functions
1054 int ip6_output(struct net *net, struct sock *sk, struct sk_buff *skb);
1055 int ip6_forward(struct sk_buff *skb);
1056 int ip6_input(struct sk_buff *skb);
1057 int ip6_mc_input(struct sk_buff *skb);
1058 void ip6_protocol_deliver_rcu(struct net *net, struct sk_buff *skb, int nexthdr,
1061 int __ip6_local_out(struct net *net, struct sock *sk, struct sk_buff *skb);
1062 int ip6_local_out(struct net *net, struct sock *sk, struct sk_buff *skb);
1065 * Extension header (options) processing
1068 void ipv6_push_nfrag_opts(struct sk_buff *skb, struct ipv6_txoptions *opt,
1069 u8 *proto, struct in6_addr **daddr_p,
1070 struct in6_addr *saddr);
1071 void ipv6_push_frag_opts(struct sk_buff *skb, struct ipv6_txoptions *opt,
1074 int ipv6_skip_exthdr(const struct sk_buff *, int start, u8 *nexthdrp,
1077 bool ipv6_ext_hdr(u8 nexthdr);
1080 IP6_FH_F_FRAG = (1 << 0),
1081 IP6_FH_F_AUTH = (1 << 1),
1082 IP6_FH_F_SKIP_RH = (1 << 2),
1085 /* find specified header and get offset to it */
1086 int ipv6_find_hdr(const struct sk_buff *skb, unsigned int *offset, int target,
1087 unsigned short *fragoff, int *fragflg);
1089 int ipv6_find_tlv(const struct sk_buff *skb, int offset, int type);
1091 struct in6_addr *fl6_update_dst(struct flowi6 *fl6,
1092 const struct ipv6_txoptions *opt,
1093 struct in6_addr *orig);
1096 * socket options (ipv6_sockglue.c)
1099 int ipv6_setsockopt(struct sock *sk, int level, int optname, sockptr_t optval,
1100 unsigned int optlen);
1101 int ipv6_getsockopt(struct sock *sk, int level, int optname,
1102 char __user *optval, int __user *optlen);
1104 int __ip6_datagram_connect(struct sock *sk, struct sockaddr *addr,
1106 int ip6_datagram_connect(struct sock *sk, struct sockaddr *addr, int addr_len);
1107 int ip6_datagram_connect_v6_only(struct sock *sk, struct sockaddr *addr,
1109 int ip6_datagram_dst_update(struct sock *sk, bool fix_sk_saddr);
1110 void ip6_datagram_release_cb(struct sock *sk);
1112 int ipv6_recv_error(struct sock *sk, struct msghdr *msg, int len,
1114 int ipv6_recv_rxpmtu(struct sock *sk, struct msghdr *msg, int len,
1116 void ipv6_icmp_error(struct sock *sk, struct sk_buff *skb, int err, __be16 port,
1117 u32 info, u8 *payload);
1118 void ipv6_local_error(struct sock *sk, int err, struct flowi6 *fl6, u32 info);
1119 void ipv6_local_rxpmtu(struct sock *sk, struct flowi6 *fl6, u32 mtu);
1121 int inet6_release(struct socket *sock);
1122 int inet6_bind(struct socket *sock, struct sockaddr *uaddr, int addr_len);
1123 int inet6_getname(struct socket *sock, struct sockaddr *uaddr,
1125 int inet6_ioctl(struct socket *sock, unsigned int cmd, unsigned long arg);
1126 int inet6_compat_ioctl(struct socket *sock, unsigned int cmd,
1129 int inet6_hash_connect(struct inet_timewait_death_row *death_row,
1131 int inet6_sendmsg(struct socket *sock, struct msghdr *msg, size_t size);
1132 int inet6_recvmsg(struct socket *sock, struct msghdr *msg, size_t size,
1138 extern const struct proto_ops inet6_stream_ops;
1139 extern const struct proto_ops inet6_dgram_ops;
1140 extern const struct proto_ops inet6_sockraw_ops;
1142 struct group_source_req;
1143 struct group_filter;
1145 int ip6_mc_source(int add, int omode, struct sock *sk,
1146 struct group_source_req *pgsr);
1147 int ip6_mc_msfilter(struct sock *sk, struct group_filter *gsf,
1148 struct sockaddr_storage *list);
1149 int ip6_mc_msfget(struct sock *sk, struct group_filter *gsf,
1150 struct sockaddr_storage __user *p);
1152 #ifdef CONFIG_PROC_FS
1153 int ac6_proc_init(struct net *net);
1154 void ac6_proc_exit(struct net *net);
1155 int raw6_proc_init(void);
1156 void raw6_proc_exit(void);
1157 int tcp6_proc_init(struct net *net);
1158 void tcp6_proc_exit(struct net *net);
1159 int udp6_proc_init(struct net *net);
1160 void udp6_proc_exit(struct net *net);
1161 int udplite6_proc_init(void);
1162 void udplite6_proc_exit(void);
1163 int ipv6_misc_proc_init(void);
1164 void ipv6_misc_proc_exit(void);
1165 int snmp6_register_dev(struct inet6_dev *idev);
1166 int snmp6_unregister_dev(struct inet6_dev *idev);
1169 static inline int ac6_proc_init(struct net *net) { return 0; }
1170 static inline void ac6_proc_exit(struct net *net) { }
1171 static inline int snmp6_register_dev(struct inet6_dev *idev) { return 0; }
1172 static inline int snmp6_unregister_dev(struct inet6_dev *idev) { return 0; }
1175 #ifdef CONFIG_SYSCTL
1176 struct ctl_table *ipv6_icmp_sysctl_init(struct net *net);
1177 struct ctl_table *ipv6_route_sysctl_init(struct net *net);
1178 int ipv6_sysctl_register(void);
1179 void ipv6_sysctl_unregister(void);
1182 int ipv6_sock_mc_join(struct sock *sk, int ifindex,
1183 const struct in6_addr *addr);
1184 int ipv6_sock_mc_join_ssm(struct sock *sk, int ifindex,
1185 const struct in6_addr *addr, unsigned int mode);
1186 int ipv6_sock_mc_drop(struct sock *sk, int ifindex,
1187 const struct in6_addr *addr);
1189 static inline int ip6_sock_set_v6only(struct sock *sk)
1191 if (inet_sk(sk)->inet_num)
1194 sk->sk_ipv6only = true;
1199 static inline void ip6_sock_set_recverr(struct sock *sk)
1202 inet6_sk(sk)->recverr = true;
1206 static inline int __ip6_sock_set_addr_preferences(struct sock *sk, int val)
1208 unsigned int pref = 0;
1209 unsigned int prefmask = ~0;
1211 /* check PUBLIC/TMP/PUBTMP_DEFAULT conflicts */
1212 switch (val & (IPV6_PREFER_SRC_PUBLIC |
1213 IPV6_PREFER_SRC_TMP |
1214 IPV6_PREFER_SRC_PUBTMP_DEFAULT)) {
1215 case IPV6_PREFER_SRC_PUBLIC:
1216 pref |= IPV6_PREFER_SRC_PUBLIC;
1217 prefmask &= ~(IPV6_PREFER_SRC_PUBLIC |
1218 IPV6_PREFER_SRC_TMP);
1220 case IPV6_PREFER_SRC_TMP:
1221 pref |= IPV6_PREFER_SRC_TMP;
1222 prefmask &= ~(IPV6_PREFER_SRC_PUBLIC |
1223 IPV6_PREFER_SRC_TMP);
1225 case IPV6_PREFER_SRC_PUBTMP_DEFAULT:
1226 prefmask &= ~(IPV6_PREFER_SRC_PUBLIC |
1227 IPV6_PREFER_SRC_TMP);
1235 /* check HOME/COA conflicts */
1236 switch (val & (IPV6_PREFER_SRC_HOME | IPV6_PREFER_SRC_COA)) {
1237 case IPV6_PREFER_SRC_HOME:
1238 prefmask &= ~IPV6_PREFER_SRC_COA;
1240 case IPV6_PREFER_SRC_COA:
1241 pref |= IPV6_PREFER_SRC_COA;
1249 /* check CGA/NONCGA conflicts */
1250 switch (val & (IPV6_PREFER_SRC_CGA|IPV6_PREFER_SRC_NONCGA)) {
1251 case IPV6_PREFER_SRC_CGA:
1252 case IPV6_PREFER_SRC_NONCGA:
1259 inet6_sk(sk)->srcprefs = (inet6_sk(sk)->srcprefs & prefmask) | pref;
1263 static inline int ip6_sock_set_addr_preferences(struct sock *sk, bool val)
1268 ret = __ip6_sock_set_addr_preferences(sk, val);
1273 static inline void ip6_sock_set_recvpktinfo(struct sock *sk)
1276 inet6_sk(sk)->rxopt.bits.rxinfo = true;
1280 #endif /* _NET_IPV6_H */