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 <net/if_inet6.h>
17 #include <net/ndisc.h>
19 #include <net/flow_dissector.h>
21 #include <net/netns/hash.h>
23 #define SIN6_LEN_RFC2133 24
25 #define IPV6_MAXPLEN 65535
28 * NextHeader field of IPv6 header
31 #define NEXTHDR_HOP 0 /* Hop-by-hop option header. */
32 #define NEXTHDR_TCP 6 /* TCP segment. */
33 #define NEXTHDR_UDP 17 /* UDP message. */
34 #define NEXTHDR_IPV6 41 /* IPv6 in IPv6 */
35 #define NEXTHDR_ROUTING 43 /* Routing header. */
36 #define NEXTHDR_FRAGMENT 44 /* Fragmentation/reassembly header. */
37 #define NEXTHDR_GRE 47 /* GRE header. */
38 #define NEXTHDR_ESP 50 /* Encapsulating security payload. */
39 #define NEXTHDR_AUTH 51 /* Authentication header. */
40 #define NEXTHDR_ICMP 58 /* ICMP for IPv6. */
41 #define NEXTHDR_NONE 59 /* No next header */
42 #define NEXTHDR_DEST 60 /* Destination options header. */
43 #define NEXTHDR_SCTP 132 /* SCTP message. */
44 #define NEXTHDR_MOBILITY 135 /* Mobility header. */
46 #define NEXTHDR_MAX 255
48 #define IPV6_DEFAULT_HOPLIMIT 64
49 #define IPV6_DEFAULT_MCASTHOPS 1
51 /* Limits on Hop-by-Hop and Destination options.
53 * Per RFC8200 there is no limit on the maximum number or lengths of options in
54 * Hop-by-Hop or Destination options other then the packet must fit in an MTU.
55 * We allow configurable limits in order to mitigate potential denial of
58 * There are three limits that may be set:
59 * - Limit the number of options in a Hop-by-Hop or Destination options
61 * - Limit the byte length of a Hop-by-Hop or Destination options extension
63 * - Disallow unknown options
65 * The limits are expressed in corresponding sysctls:
67 * ipv6.sysctl.max_dst_opts_cnt
68 * ipv6.sysctl.max_hbh_opts_cnt
69 * ipv6.sysctl.max_dst_opts_len
70 * ipv6.sysctl.max_hbh_opts_len
72 * max_*_opts_cnt is the number of TLVs that are allowed for Destination
73 * options or Hop-by-Hop options. If the number is less than zero then unknown
74 * TLVs are disallowed and the number of known options that are allowed is the
75 * absolute value. Setting the value to INT_MAX indicates no limit.
77 * max_*_opts_len is the length limit in bytes of a Destination or
78 * Hop-by-Hop options extension header. Setting the value to INT_MAX
79 * indicates no length limit.
81 * If a limit is exceeded when processing an extension header the packet is
85 /* Default limits for Hop-by-Hop and Destination options */
86 #define IP6_DEFAULT_MAX_DST_OPTS_CNT 8
87 #define IP6_DEFAULT_MAX_HBH_OPTS_CNT 8
88 #define IP6_DEFAULT_MAX_DST_OPTS_LEN INT_MAX /* No limit */
89 #define IP6_DEFAULT_MAX_HBH_OPTS_LEN INT_MAX /* No limit */
94 * type - unicast | multicast
95 * scope - local | site | global
102 #define IPV6_ADDR_ANY 0x0000U
104 #define IPV6_ADDR_UNICAST 0x0001U
105 #define IPV6_ADDR_MULTICAST 0x0002U
107 #define IPV6_ADDR_LOOPBACK 0x0010U
108 #define IPV6_ADDR_LINKLOCAL 0x0020U
109 #define IPV6_ADDR_SITELOCAL 0x0040U
111 #define IPV6_ADDR_COMPATv4 0x0080U
113 #define IPV6_ADDR_SCOPE_MASK 0x00f0U
115 #define IPV6_ADDR_MAPPED 0x1000U
120 #define IPV6_ADDR_MC_SCOPE(a) \
121 ((a)->s6_addr[1] & 0x0f) /* nonstandard */
122 #define __IPV6_ADDR_SCOPE_INVALID -1
123 #define IPV6_ADDR_SCOPE_NODELOCAL 0x01
124 #define IPV6_ADDR_SCOPE_LINKLOCAL 0x02
125 #define IPV6_ADDR_SCOPE_SITELOCAL 0x05
126 #define IPV6_ADDR_SCOPE_ORGLOCAL 0x08
127 #define IPV6_ADDR_SCOPE_GLOBAL 0x0e
132 #define IPV6_ADDR_MC_FLAG_TRANSIENT(a) \
133 ((a)->s6_addr[1] & 0x10)
134 #define IPV6_ADDR_MC_FLAG_PREFIX(a) \
135 ((a)->s6_addr[1] & 0x20)
136 #define IPV6_ADDR_MC_FLAG_RENDEZVOUS(a) \
137 ((a)->s6_addr[1] & 0x40)
140 * fragmentation header
147 __be32 identification;
150 #define IP6_MF 0x0001
151 #define IP6_OFFSET 0xFFF8
153 #define IP6_REPLY_MARK(net, mark) \
154 ((net)->ipv6.sysctl.fwmark_reflect ? (mark) : 0)
156 #include <net/sock.h>
159 extern int sysctl_mld_max_msf;
160 extern int sysctl_mld_qrv;
162 #define _DEVINC(net, statname, mod, idev, field) \
164 struct inet6_dev *_idev = (idev); \
165 if (likely(_idev != NULL)) \
166 mod##SNMP_INC_STATS64((_idev)->stats.statname, (field));\
167 mod##SNMP_INC_STATS64((net)->mib.statname##_statistics, (field));\
170 /* per device counters are atomic_long_t */
171 #define _DEVINCATOMIC(net, statname, mod, idev, field) \
173 struct inet6_dev *_idev = (idev); \
174 if (likely(_idev != NULL)) \
175 SNMP_INC_STATS_ATOMIC_LONG((_idev)->stats.statname##dev, (field)); \
176 mod##SNMP_INC_STATS((net)->mib.statname##_statistics, (field));\
179 /* per device and per net counters are atomic_long_t */
180 #define _DEVINC_ATOMIC_ATOMIC(net, statname, idev, field) \
182 struct inet6_dev *_idev = (idev); \
183 if (likely(_idev != NULL)) \
184 SNMP_INC_STATS_ATOMIC_LONG((_idev)->stats.statname##dev, (field)); \
185 SNMP_INC_STATS_ATOMIC_LONG((net)->mib.statname##_statistics, (field));\
188 #define _DEVADD(net, statname, mod, idev, field, val) \
190 struct inet6_dev *_idev = (idev); \
191 if (likely(_idev != NULL)) \
192 mod##SNMP_ADD_STATS((_idev)->stats.statname, (field), (val)); \
193 mod##SNMP_ADD_STATS((net)->mib.statname##_statistics, (field), (val));\
196 #define _DEVUPD(net, statname, mod, idev, field, val) \
198 struct inet6_dev *_idev = (idev); \
199 if (likely(_idev != NULL)) \
200 mod##SNMP_UPD_PO_STATS((_idev)->stats.statname, field, (val)); \
201 mod##SNMP_UPD_PO_STATS((net)->mib.statname##_statistics, field, (val));\
206 #define IP6_INC_STATS(net, idev,field) \
207 _DEVINC(net, ipv6, , idev, field)
208 #define __IP6_INC_STATS(net, idev,field) \
209 _DEVINC(net, ipv6, __, idev, field)
210 #define IP6_ADD_STATS(net, idev,field,val) \
211 _DEVADD(net, ipv6, , idev, field, val)
212 #define __IP6_ADD_STATS(net, idev,field,val) \
213 _DEVADD(net, ipv6, __, idev, field, val)
214 #define IP6_UPD_PO_STATS(net, idev,field,val) \
215 _DEVUPD(net, ipv6, , idev, field, val)
216 #define __IP6_UPD_PO_STATS(net, idev,field,val) \
217 _DEVUPD(net, ipv6, __, idev, field, val)
218 #define ICMP6_INC_STATS(net, idev, field) \
219 _DEVINCATOMIC(net, icmpv6, , idev, field)
220 #define __ICMP6_INC_STATS(net, idev, field) \
221 _DEVINCATOMIC(net, icmpv6, __, idev, field)
223 #define ICMP6MSGOUT_INC_STATS(net, idev, field) \
224 _DEVINC_ATOMIC_ATOMIC(net, icmpv6msg, idev, field +256)
225 #define ICMP6MSGIN_INC_STATS(net, idev, field) \
226 _DEVINC_ATOMIC_ATOMIC(net, icmpv6msg, idev, field)
228 struct ip6_ra_chain {
229 struct ip6_ra_chain *next;
232 void (*destructor)(struct sock *);
235 extern struct ip6_ra_chain *ip6_ra_chain;
236 extern rwlock_t ip6_ra_lock;
239 This structure is prepared by protocol, when parsing
240 ancillary data and passed to IPv6.
243 struct ipv6_txoptions {
245 /* Length of this structure */
248 /* length of extension headers */
250 __u16 opt_flen; /* after fragment hdr */
251 __u16 opt_nflen; /* before fragment hdr */
253 struct ipv6_opt_hdr *hopopt;
254 struct ipv6_opt_hdr *dst0opt;
255 struct ipv6_rt_hdr *srcrt; /* Routing Header */
256 struct ipv6_opt_hdr *dst1opt;
258 /* Option buffer, as read by IPV6_PKTOPTIONS, starts here. */
261 struct ip6_flowlabel {
262 struct ip6_flowlabel __rcu *next;
266 struct ipv6_txoptions *opt;
267 unsigned long linger;
274 unsigned long lastuse;
275 unsigned long expires;
279 #define IPV6_FLOWINFO_MASK cpu_to_be32(0x0FFFFFFF)
280 #define IPV6_FLOWLABEL_MASK cpu_to_be32(0x000FFFFF)
281 #define IPV6_FLOWLABEL_STATELESS_FLAG cpu_to_be32(0x00080000)
283 #define IPV6_TCLASS_MASK (IPV6_FLOWINFO_MASK & ~IPV6_FLOWLABEL_MASK)
284 #define IPV6_TCLASS_SHIFT 20
286 struct ipv6_fl_socklist {
287 struct ipv6_fl_socklist __rcu *next;
288 struct ip6_flowlabel *fl;
292 struct ipcm6_cookie {
293 struct sockcm_cookie sockc;
297 struct ipv6_txoptions *opt;
301 static inline void ipcm6_init(struct ipcm6_cookie *ipc6)
303 *ipc6 = (struct ipcm6_cookie) {
310 static inline void ipcm6_init_sk(struct ipcm6_cookie *ipc6,
311 const struct ipv6_pinfo *np)
313 *ipc6 = (struct ipcm6_cookie) {
315 .tclass = np->tclass,
316 .dontfrag = np->dontfrag,
320 static inline struct ipv6_txoptions *txopt_get(const struct ipv6_pinfo *np)
322 struct ipv6_txoptions *opt;
325 opt = rcu_dereference(np->opt);
327 if (!refcount_inc_not_zero(&opt->refcnt))
330 opt = rcu_pointer_handoff(opt);
336 static inline void txopt_put(struct ipv6_txoptions *opt)
338 if (opt && refcount_dec_and_test(&opt->refcnt))
342 struct ip6_flowlabel *fl6_sock_lookup(struct sock *sk, __be32 label);
343 struct ipv6_txoptions *fl6_merge_options(struct ipv6_txoptions *opt_space,
344 struct ip6_flowlabel *fl,
345 struct ipv6_txoptions *fopt);
346 void fl6_free_socklist(struct sock *sk);
347 int ipv6_flowlabel_opt(struct sock *sk, char __user *optval, int optlen);
348 int ipv6_flowlabel_opt_get(struct sock *sk, struct in6_flowlabel_req *freq,
350 int ip6_flowlabel_init(void);
351 void ip6_flowlabel_cleanup(void);
352 bool ip6_autoflowlabel(struct net *net, const struct ipv6_pinfo *np);
354 static inline void fl6_sock_release(struct ip6_flowlabel *fl)
357 atomic_dec(&fl->users);
360 void icmpv6_notify(struct sk_buff *skb, u8 type, u8 code, __be32 info);
362 void icmpv6_push_pending_frames(struct sock *sk, struct flowi6 *fl6,
363 struct icmp6hdr *thdr, int len);
365 int ip6_ra_control(struct sock *sk, int sel);
367 int ipv6_parse_hopopts(struct sk_buff *skb);
369 struct ipv6_txoptions *ipv6_dup_options(struct sock *sk,
370 struct ipv6_txoptions *opt);
371 struct ipv6_txoptions *ipv6_renew_options(struct sock *sk,
372 struct ipv6_txoptions *opt,
374 struct ipv6_opt_hdr *newopt);
375 struct ipv6_txoptions *ipv6_fixup_options(struct ipv6_txoptions *opt_space,
376 struct ipv6_txoptions *opt);
378 bool ipv6_opt_accepted(const struct sock *sk, const struct sk_buff *skb,
379 const struct inet6_skb_parm *opt);
380 struct ipv6_txoptions *ipv6_update_options(struct sock *sk,
381 struct ipv6_txoptions *opt);
383 static inline bool ipv6_accept_ra(struct inet6_dev *idev)
385 /* If forwarding is enabled, RA are not accepted unless the special
386 * hybrid mode (accept_ra=2) is enabled.
388 return idev->cnf.forwarding ? idev->cnf.accept_ra == 2 :
392 #define IPV6_FRAG_HIGH_THRESH (4 * 1024*1024) /* 4194304 */
393 #define IPV6_FRAG_LOW_THRESH (3 * 1024*1024) /* 3145728 */
394 #define IPV6_FRAG_TIMEOUT (60 * HZ) /* 60 seconds */
396 int __ipv6_addr_type(const struct in6_addr *addr);
397 static inline int ipv6_addr_type(const struct in6_addr *addr)
399 return __ipv6_addr_type(addr) & 0xffff;
402 static inline int ipv6_addr_scope(const struct in6_addr *addr)
404 return __ipv6_addr_type(addr) & IPV6_ADDR_SCOPE_MASK;
407 static inline int __ipv6_addr_src_scope(int type)
409 return (type == IPV6_ADDR_ANY) ? __IPV6_ADDR_SCOPE_INVALID : (type >> 16);
412 static inline int ipv6_addr_src_scope(const struct in6_addr *addr)
414 return __ipv6_addr_src_scope(__ipv6_addr_type(addr));
417 static inline bool __ipv6_addr_needs_scope_id(int type)
419 return type & IPV6_ADDR_LINKLOCAL ||
420 (type & IPV6_ADDR_MULTICAST &&
421 (type & (IPV6_ADDR_LOOPBACK|IPV6_ADDR_LINKLOCAL)));
424 static inline __u32 ipv6_iface_scope_id(const struct in6_addr *addr, int iface)
426 return __ipv6_addr_needs_scope_id(__ipv6_addr_type(addr)) ? iface : 0;
429 static inline int ipv6_addr_cmp(const struct in6_addr *a1, const struct in6_addr *a2)
431 return memcmp(a1, a2, sizeof(struct in6_addr));
435 ipv6_masked_addr_cmp(const struct in6_addr *a1, const struct in6_addr *m,
436 const struct in6_addr *a2)
438 #if defined(CONFIG_HAVE_EFFICIENT_UNALIGNED_ACCESS) && BITS_PER_LONG == 64
439 const unsigned long *ul1 = (const unsigned long *)a1;
440 const unsigned long *ulm = (const unsigned long *)m;
441 const unsigned long *ul2 = (const unsigned long *)a2;
443 return !!(((ul1[0] ^ ul2[0]) & ulm[0]) |
444 ((ul1[1] ^ ul2[1]) & ulm[1]));
446 return !!(((a1->s6_addr32[0] ^ a2->s6_addr32[0]) & m->s6_addr32[0]) |
447 ((a1->s6_addr32[1] ^ a2->s6_addr32[1]) & m->s6_addr32[1]) |
448 ((a1->s6_addr32[2] ^ a2->s6_addr32[2]) & m->s6_addr32[2]) |
449 ((a1->s6_addr32[3] ^ a2->s6_addr32[3]) & m->s6_addr32[3]));
453 static inline void ipv6_addr_prefix(struct in6_addr *pfx,
454 const struct in6_addr *addr,
457 /* caller must guarantee 0 <= plen <= 128 */
461 memset(pfx->s6_addr, 0, sizeof(pfx->s6_addr));
462 memcpy(pfx->s6_addr, addr, o);
464 pfx->s6_addr[o] = addr->s6_addr[o] & (0xff00 >> b);
467 static inline void ipv6_addr_prefix_copy(struct in6_addr *addr,
468 const struct in6_addr *pfx,
471 /* caller must guarantee 0 <= plen <= 128 */
475 memcpy(addr->s6_addr, pfx, o);
477 addr->s6_addr[o] &= ~(0xff00 >> b);
478 addr->s6_addr[o] |= (pfx->s6_addr[o] & (0xff00 >> b));
482 static inline void __ipv6_addr_set_half(__be32 *addr,
483 __be32 wh, __be32 wl)
485 #if defined(CONFIG_HAVE_EFFICIENT_UNALIGNED_ACCESS) && BITS_PER_LONG == 64
486 #if defined(__BIG_ENDIAN)
487 if (__builtin_constant_p(wh) && __builtin_constant_p(wl)) {
488 *(__force u64 *)addr = ((__force u64)(wh) << 32 | (__force u64)(wl));
491 #elif defined(__LITTLE_ENDIAN)
492 if (__builtin_constant_p(wl) && __builtin_constant_p(wh)) {
493 *(__force u64 *)addr = ((__force u64)(wl) << 32 | (__force u64)(wh));
502 static inline void ipv6_addr_set(struct in6_addr *addr,
503 __be32 w1, __be32 w2,
504 __be32 w3, __be32 w4)
506 __ipv6_addr_set_half(&addr->s6_addr32[0], w1, w2);
507 __ipv6_addr_set_half(&addr->s6_addr32[2], w3, w4);
510 static inline bool ipv6_addr_equal(const struct in6_addr *a1,
511 const struct in6_addr *a2)
513 #if defined(CONFIG_HAVE_EFFICIENT_UNALIGNED_ACCESS) && BITS_PER_LONG == 64
514 const unsigned long *ul1 = (const unsigned long *)a1;
515 const unsigned long *ul2 = (const unsigned long *)a2;
517 return ((ul1[0] ^ ul2[0]) | (ul1[1] ^ ul2[1])) == 0UL;
519 return ((a1->s6_addr32[0] ^ a2->s6_addr32[0]) |
520 (a1->s6_addr32[1] ^ a2->s6_addr32[1]) |
521 (a1->s6_addr32[2] ^ a2->s6_addr32[2]) |
522 (a1->s6_addr32[3] ^ a2->s6_addr32[3])) == 0;
526 #if defined(CONFIG_HAVE_EFFICIENT_UNALIGNED_ACCESS) && BITS_PER_LONG == 64
527 static inline bool __ipv6_prefix_equal64_half(const __be64 *a1,
531 if (len && ((*a1 ^ *a2) & cpu_to_be64((~0UL) << (64 - len))))
536 static inline bool ipv6_prefix_equal(const struct in6_addr *addr1,
537 const struct in6_addr *addr2,
538 unsigned int prefixlen)
540 const __be64 *a1 = (const __be64 *)addr1;
541 const __be64 *a2 = (const __be64 *)addr2;
543 if (prefixlen >= 64) {
546 return __ipv6_prefix_equal64_half(a1 + 1, a2 + 1, prefixlen - 64);
548 return __ipv6_prefix_equal64_half(a1, a2, prefixlen);
551 static inline bool ipv6_prefix_equal(const struct in6_addr *addr1,
552 const struct in6_addr *addr2,
553 unsigned int prefixlen)
555 const __be32 *a1 = addr1->s6_addr32;
556 const __be32 *a2 = addr2->s6_addr32;
557 unsigned int pdw, pbi;
559 /* check complete u32 in prefix */
560 pdw = prefixlen >> 5;
561 if (pdw && memcmp(a1, a2, pdw << 2))
564 /* check incomplete u32 in prefix */
565 pbi = prefixlen & 0x1f;
566 if (pbi && ((a1[pdw] ^ a2[pdw]) & htonl((0xffffffff) << (32 - pbi))))
573 static inline bool ipv6_addr_any(const struct in6_addr *a)
575 #if defined(CONFIG_HAVE_EFFICIENT_UNALIGNED_ACCESS) && BITS_PER_LONG == 64
576 const unsigned long *ul = (const unsigned long *)a;
578 return (ul[0] | ul[1]) == 0UL;
580 return (a->s6_addr32[0] | a->s6_addr32[1] |
581 a->s6_addr32[2] | a->s6_addr32[3]) == 0;
585 static inline u32 ipv6_addr_hash(const struct in6_addr *a)
587 #if defined(CONFIG_HAVE_EFFICIENT_UNALIGNED_ACCESS) && BITS_PER_LONG == 64
588 const unsigned long *ul = (const unsigned long *)a;
589 unsigned long x = ul[0] ^ ul[1];
591 return (u32)(x ^ (x >> 32));
593 return (__force u32)(a->s6_addr32[0] ^ a->s6_addr32[1] ^
594 a->s6_addr32[2] ^ a->s6_addr32[3]);
598 /* more secured version of ipv6_addr_hash() */
599 static inline u32 __ipv6_addr_jhash(const struct in6_addr *a, const u32 initval)
601 u32 v = (__force u32)a->s6_addr32[0] ^ (__force u32)a->s6_addr32[1];
603 return jhash_3words(v,
604 (__force u32)a->s6_addr32[2],
605 (__force u32)a->s6_addr32[3],
609 static inline bool ipv6_addr_loopback(const struct in6_addr *a)
611 #if defined(CONFIG_HAVE_EFFICIENT_UNALIGNED_ACCESS) && BITS_PER_LONG == 64
612 const __be64 *be = (const __be64 *)a;
614 return (be[0] | (be[1] ^ cpu_to_be64(1))) == 0UL;
616 return (a->s6_addr32[0] | a->s6_addr32[1] |
617 a->s6_addr32[2] | (a->s6_addr32[3] ^ cpu_to_be32(1))) == 0;
622 * Note that we must __force cast these to unsigned long to make sparse happy,
623 * since all of the endian-annotated types are fixed size regardless of arch.
625 static inline bool ipv6_addr_v4mapped(const struct in6_addr *a)
628 #if defined(CONFIG_HAVE_EFFICIENT_UNALIGNED_ACCESS) && BITS_PER_LONG == 64
629 *(unsigned long *)a |
631 (__force unsigned long)(a->s6_addr32[0] | a->s6_addr32[1]) |
633 (__force unsigned long)(a->s6_addr32[2] ^
634 cpu_to_be32(0x0000ffff))) == 0UL;
637 static inline u32 ipv6_portaddr_hash(const struct net *net,
638 const struct in6_addr *addr6,
641 unsigned int hash, mix = net_hash_mix(net);
643 if (ipv6_addr_any(addr6))
644 hash = jhash_1word(0, mix);
645 else if (ipv6_addr_v4mapped(addr6))
646 hash = jhash_1word((__force u32)addr6->s6_addr32[3], mix);
648 hash = jhash2((__force u32 *)addr6->s6_addr32, 4, mix);
654 * Check for a RFC 4843 ORCHID address
655 * (Overlay Routable Cryptographic Hash Identifiers)
657 static inline bool ipv6_addr_orchid(const struct in6_addr *a)
659 return (a->s6_addr32[0] & htonl(0xfffffff0)) == htonl(0x20010010);
662 static inline bool ipv6_addr_is_multicast(const struct in6_addr *addr)
664 return (addr->s6_addr32[0] & htonl(0xFF000000)) == htonl(0xFF000000);
667 static inline void ipv6_addr_set_v4mapped(const __be32 addr,
668 struct in6_addr *v4mapped)
670 ipv6_addr_set(v4mapped,
677 * find the first different bit between two addresses
678 * length of address must be a multiple of 32bits
680 static inline int __ipv6_addr_diff32(const void *token1, const void *token2, int addrlen)
682 const __be32 *a1 = token1, *a2 = token2;
687 for (i = 0; i < addrlen; i++) {
688 __be32 xb = a1[i] ^ a2[i];
690 return i * 32 + 31 - __fls(ntohl(xb));
694 * we should *never* get to this point since that
695 * would mean the addrs are equal
697 * However, we do get to it 8) And exacly, when
698 * addresses are equal 8)
700 * ip route add 1111::/128 via ...
701 * ip route add 1111::/64 via ...
704 * Ideally, this function should stop comparison
705 * at prefix length. It does not, but it is still OK,
706 * if returned value is greater than prefix length.
712 #if defined(CONFIG_HAVE_EFFICIENT_UNALIGNED_ACCESS) && BITS_PER_LONG == 64
713 static inline int __ipv6_addr_diff64(const void *token1, const void *token2, int addrlen)
715 const __be64 *a1 = token1, *a2 = token2;
720 for (i = 0; i < addrlen; i++) {
721 __be64 xb = a1[i] ^ a2[i];
723 return i * 64 + 63 - __fls(be64_to_cpu(xb));
730 static inline int __ipv6_addr_diff(const void *token1, const void *token2, int addrlen)
732 #if defined(CONFIG_HAVE_EFFICIENT_UNALIGNED_ACCESS) && BITS_PER_LONG == 64
733 if (__builtin_constant_p(addrlen) && !(addrlen & 7))
734 return __ipv6_addr_diff64(token1, token2, addrlen);
736 return __ipv6_addr_diff32(token1, token2, addrlen);
739 static inline int ipv6_addr_diff(const struct in6_addr *a1, const struct in6_addr *a2)
741 return __ipv6_addr_diff(a1, a2, sizeof(struct in6_addr));
744 __be32 ipv6_select_ident(struct net *net,
745 const struct in6_addr *daddr,
746 const struct in6_addr *saddr);
747 __be32 ipv6_proxy_select_ident(struct net *net, struct sk_buff *skb);
749 int ip6_dst_hoplimit(struct dst_entry *dst);
751 static inline int ip6_sk_dst_hoplimit(struct ipv6_pinfo *np, struct flowi6 *fl6,
752 struct dst_entry *dst)
756 if (ipv6_addr_is_multicast(&fl6->daddr))
757 hlimit = np->mcast_hops;
759 hlimit = np->hop_limit;
761 hlimit = ip6_dst_hoplimit(dst);
765 /* copy IPv6 saddr & daddr to flow_keys, possibly using 64bit load/store
766 * Equivalent to : flow->v6addrs.src = iph->saddr;
767 * flow->v6addrs.dst = iph->daddr;
769 static inline void iph_to_flow_copy_v6addrs(struct flow_keys *flow,
770 const struct ipv6hdr *iph)
772 BUILD_BUG_ON(offsetof(typeof(flow->addrs), v6addrs.dst) !=
773 offsetof(typeof(flow->addrs), v6addrs.src) +
774 sizeof(flow->addrs.v6addrs.src));
775 memcpy(&flow->addrs.v6addrs, &iph->saddr, sizeof(flow->addrs.v6addrs));
776 flow->control.addr_type = FLOW_DISSECTOR_KEY_IPV6_ADDRS;
779 #if IS_ENABLED(CONFIG_IPV6)
781 static inline bool ipv6_can_nonlocal_bind(struct net *net,
782 struct inet_sock *inet)
784 return net->ipv6.sysctl.ip_nonlocal_bind ||
785 inet->freebind || inet->transparent;
788 /* Sysctl settings for net ipv6.auto_flowlabels */
789 #define IP6_AUTO_FLOW_LABEL_OFF 0
790 #define IP6_AUTO_FLOW_LABEL_OPTOUT 1
791 #define IP6_AUTO_FLOW_LABEL_OPTIN 2
792 #define IP6_AUTO_FLOW_LABEL_FORCED 3
794 #define IP6_AUTO_FLOW_LABEL_MAX IP6_AUTO_FLOW_LABEL_FORCED
796 #define IP6_DEFAULT_AUTO_FLOW_LABELS IP6_AUTO_FLOW_LABEL_OPTOUT
798 static inline __be32 ip6_make_flowlabel(struct net *net, struct sk_buff *skb,
799 __be32 flowlabel, bool autolabel,
804 /* @flowlabel may include more than a flow label, eg, the traffic class.
805 * Here we want only the flow label value.
807 flowlabel &= IPV6_FLOWLABEL_MASK;
810 net->ipv6.sysctl.auto_flowlabels == IP6_AUTO_FLOW_LABEL_OFF ||
812 net->ipv6.sysctl.auto_flowlabels != IP6_AUTO_FLOW_LABEL_FORCED))
815 hash = skb_get_hash_flowi6(skb, fl6);
817 /* Since this is being sent on the wire obfuscate hash a bit
818 * to minimize possbility that any useful information to an
819 * attacker is leaked. Only lower 20 bits are relevant.
821 hash = rol32(hash, 16);
823 flowlabel = (__force __be32)hash & IPV6_FLOWLABEL_MASK;
825 if (net->ipv6.sysctl.flowlabel_state_ranges)
826 flowlabel |= IPV6_FLOWLABEL_STATELESS_FLAG;
831 static inline int ip6_default_np_autolabel(struct net *net)
833 switch (net->ipv6.sysctl.auto_flowlabels) {
834 case IP6_AUTO_FLOW_LABEL_OFF:
835 case IP6_AUTO_FLOW_LABEL_OPTIN:
838 case IP6_AUTO_FLOW_LABEL_OPTOUT:
839 case IP6_AUTO_FLOW_LABEL_FORCED:
844 static inline void ip6_set_txhash(struct sock *sk) { }
845 static inline __be32 ip6_make_flowlabel(struct net *net, struct sk_buff *skb,
846 __be32 flowlabel, bool autolabel,
851 static inline int ip6_default_np_autolabel(struct net *net)
857 #if IS_ENABLED(CONFIG_IPV6)
858 static inline int ip6_multipath_hash_policy(const struct net *net)
860 return net->ipv6.sysctl.multipath_hash_policy;
863 static inline int ip6_multipath_hash_policy(const struct net *net)
870 * Header manipulation
872 static inline void ip6_flow_hdr(struct ipv6hdr *hdr, unsigned int tclass,
875 *(__be32 *)hdr = htonl(0x60000000 | (tclass << 20)) | flowlabel;
878 static inline __be32 ip6_flowinfo(const struct ipv6hdr *hdr)
880 return *(__be32 *)hdr & IPV6_FLOWINFO_MASK;
883 static inline __be32 ip6_flowlabel(const struct ipv6hdr *hdr)
885 return *(__be32 *)hdr & IPV6_FLOWLABEL_MASK;
888 static inline u8 ip6_tclass(__be32 flowinfo)
890 return ntohl(flowinfo & IPV6_TCLASS_MASK) >> IPV6_TCLASS_SHIFT;
893 static inline __be32 ip6_make_flowinfo(unsigned int tclass, __be32 flowlabel)
895 return htonl(tclass << IPV6_TCLASS_SHIFT) | flowlabel;
898 static inline __be32 flowi6_get_flowlabel(const struct flowi6 *fl6)
900 return fl6->flowlabel & IPV6_FLOWLABEL_MASK;
904 * Prototypes exported by ipv6
908 * rcv function (called from netdevice level)
911 int ipv6_rcv(struct sk_buff *skb, struct net_device *dev,
912 struct packet_type *pt, struct net_device *orig_dev);
913 void ipv6_list_rcv(struct list_head *head, struct packet_type *pt,
914 struct net_device *orig_dev);
916 int ip6_rcv_finish(struct net *net, struct sock *sk, struct sk_buff *skb);
919 * upper-layer output functions
921 int ip6_xmit(const struct sock *sk, struct sk_buff *skb, struct flowi6 *fl6,
922 __u32 mark, struct ipv6_txoptions *opt, int tclass);
924 int ip6_find_1stfragopt(struct sk_buff *skb, u8 **nexthdr);
926 int ip6_append_data(struct sock *sk,
927 int getfrag(void *from, char *to, int offset, int len,
928 int odd, struct sk_buff *skb),
929 void *from, int length, int transhdrlen,
930 struct ipcm6_cookie *ipc6, struct flowi6 *fl6,
931 struct rt6_info *rt, unsigned int flags);
933 int ip6_push_pending_frames(struct sock *sk);
935 void ip6_flush_pending_frames(struct sock *sk);
937 int ip6_send_skb(struct sk_buff *skb);
939 struct sk_buff *__ip6_make_skb(struct sock *sk, struct sk_buff_head *queue,
940 struct inet_cork_full *cork,
941 struct inet6_cork *v6_cork);
942 struct sk_buff *ip6_make_skb(struct sock *sk,
943 int getfrag(void *from, char *to, int offset,
944 int len, int odd, struct sk_buff *skb),
945 void *from, int length, int transhdrlen,
946 struct ipcm6_cookie *ipc6, struct flowi6 *fl6,
947 struct rt6_info *rt, unsigned int flags,
948 struct inet_cork_full *cork);
950 static inline struct sk_buff *ip6_finish_skb(struct sock *sk)
952 return __ip6_make_skb(sk, &sk->sk_write_queue, &inet_sk(sk)->cork,
953 &inet6_sk(sk)->cork);
956 int ip6_dst_lookup(struct net *net, struct sock *sk, struct dst_entry **dst,
958 struct dst_entry *ip6_dst_lookup_flow(const struct sock *sk, struct flowi6 *fl6,
959 const struct in6_addr *final_dst);
960 struct dst_entry *ip6_sk_dst_lookup_flow(struct sock *sk, struct flowi6 *fl6,
961 const struct in6_addr *final_dst,
963 struct dst_entry *ip6_blackhole_route(struct net *net,
964 struct dst_entry *orig_dst);
967 * skb processing functions
970 int ip6_output(struct net *net, struct sock *sk, struct sk_buff *skb);
971 int ip6_forward(struct sk_buff *skb);
972 int ip6_input(struct sk_buff *skb);
973 int ip6_mc_input(struct sk_buff *skb);
974 void ip6_protocol_deliver_rcu(struct net *net, struct sk_buff *skb, int nexthdr,
977 int __ip6_local_out(struct net *net, struct sock *sk, struct sk_buff *skb);
978 int ip6_local_out(struct net *net, struct sock *sk, struct sk_buff *skb);
981 * Extension header (options) processing
984 void ipv6_push_nfrag_opts(struct sk_buff *skb, struct ipv6_txoptions *opt,
985 u8 *proto, struct in6_addr **daddr_p,
986 struct in6_addr *saddr);
987 void ipv6_push_frag_opts(struct sk_buff *skb, struct ipv6_txoptions *opt,
990 int ipv6_skip_exthdr(const struct sk_buff *, int start, u8 *nexthdrp,
993 bool ipv6_ext_hdr(u8 nexthdr);
996 IP6_FH_F_FRAG = (1 << 0),
997 IP6_FH_F_AUTH = (1 << 1),
998 IP6_FH_F_SKIP_RH = (1 << 2),
1001 /* find specified header and get offset to it */
1002 int ipv6_find_hdr(const struct sk_buff *skb, unsigned int *offset, int target,
1003 unsigned short *fragoff, int *fragflg);
1005 int ipv6_find_tlv(const struct sk_buff *skb, int offset, int type);
1007 struct in6_addr *fl6_update_dst(struct flowi6 *fl6,
1008 const struct ipv6_txoptions *opt,
1009 struct in6_addr *orig);
1012 * socket options (ipv6_sockglue.c)
1015 int ipv6_setsockopt(struct sock *sk, int level, int optname,
1016 char __user *optval, unsigned int optlen);
1017 int ipv6_getsockopt(struct sock *sk, int level, int optname,
1018 char __user *optval, int __user *optlen);
1019 int compat_ipv6_setsockopt(struct sock *sk, int level, int optname,
1020 char __user *optval, unsigned int optlen);
1021 int compat_ipv6_getsockopt(struct sock *sk, int level, int optname,
1022 char __user *optval, int __user *optlen);
1024 int __ip6_datagram_connect(struct sock *sk, struct sockaddr *addr,
1026 int ip6_datagram_connect(struct sock *sk, struct sockaddr *addr, int addr_len);
1027 int ip6_datagram_connect_v6_only(struct sock *sk, struct sockaddr *addr,
1029 int ip6_datagram_dst_update(struct sock *sk, bool fix_sk_saddr);
1030 void ip6_datagram_release_cb(struct sock *sk);
1032 int ipv6_recv_error(struct sock *sk, struct msghdr *msg, int len,
1034 int ipv6_recv_rxpmtu(struct sock *sk, struct msghdr *msg, int len,
1036 void ipv6_icmp_error(struct sock *sk, struct sk_buff *skb, int err, __be16 port,
1037 u32 info, u8 *payload);
1038 void ipv6_local_error(struct sock *sk, int err, struct flowi6 *fl6, u32 info);
1039 void ipv6_local_rxpmtu(struct sock *sk, struct flowi6 *fl6, u32 mtu);
1041 int inet6_release(struct socket *sock);
1042 int inet6_bind(struct socket *sock, struct sockaddr *uaddr, int addr_len);
1043 int inet6_getname(struct socket *sock, struct sockaddr *uaddr,
1045 int inet6_ioctl(struct socket *sock, unsigned int cmd, unsigned long arg);
1047 int inet6_hash_connect(struct inet_timewait_death_row *death_row,
1053 extern const struct proto_ops inet6_stream_ops;
1054 extern const struct proto_ops inet6_dgram_ops;
1055 extern const struct proto_ops inet6_sockraw_ops;
1057 struct group_source_req;
1058 struct group_filter;
1060 int ip6_mc_source(int add, int omode, struct sock *sk,
1061 struct group_source_req *pgsr);
1062 int ip6_mc_msfilter(struct sock *sk, struct group_filter *gsf);
1063 int ip6_mc_msfget(struct sock *sk, struct group_filter *gsf,
1064 struct group_filter __user *optval, int __user *optlen);
1066 #ifdef CONFIG_PROC_FS
1067 int ac6_proc_init(struct net *net);
1068 void ac6_proc_exit(struct net *net);
1069 int raw6_proc_init(void);
1070 void raw6_proc_exit(void);
1071 int tcp6_proc_init(struct net *net);
1072 void tcp6_proc_exit(struct net *net);
1073 int udp6_proc_init(struct net *net);
1074 void udp6_proc_exit(struct net *net);
1075 int udplite6_proc_init(void);
1076 void udplite6_proc_exit(void);
1077 int ipv6_misc_proc_init(void);
1078 void ipv6_misc_proc_exit(void);
1079 int snmp6_register_dev(struct inet6_dev *idev);
1080 int snmp6_unregister_dev(struct inet6_dev *idev);
1083 static inline int ac6_proc_init(struct net *net) { return 0; }
1084 static inline void ac6_proc_exit(struct net *net) { }
1085 static inline int snmp6_register_dev(struct inet6_dev *idev) { return 0; }
1086 static inline int snmp6_unregister_dev(struct inet6_dev *idev) { return 0; }
1089 #ifdef CONFIG_SYSCTL
1090 struct ctl_table *ipv6_icmp_sysctl_init(struct net *net);
1091 struct ctl_table *ipv6_route_sysctl_init(struct net *net);
1092 int ipv6_sysctl_register(void);
1093 void ipv6_sysctl_unregister(void);
1096 int ipv6_sock_mc_join(struct sock *sk, int ifindex,
1097 const struct in6_addr *addr);
1098 int ipv6_sock_mc_join_ssm(struct sock *sk, int ifindex,
1099 const struct in6_addr *addr, unsigned int mode);
1100 int ipv6_sock_mc_drop(struct sock *sk, int ifindex,
1101 const struct in6_addr *addr);
1102 #endif /* _NET_IPV6_H */