2 * Linux INET6 implementation
5 * Pedro Roque <roque@di.fc.ul.pt>
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 <net/if_inet6.h>
20 #include <net/ndisc.h>
22 #include <net/flow_dissector.h>
25 #define SIN6_LEN_RFC2133 24
27 #define IPV6_MAXPLEN 65535
30 * NextHeader field of IPv6 header
33 #define NEXTHDR_HOP 0 /* Hop-by-hop option header. */
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
56 * type - unicast | multicast
57 * scope - local | site | global
64 #define IPV6_ADDR_ANY 0x0000U
66 #define IPV6_ADDR_UNICAST 0x0001U
67 #define IPV6_ADDR_MULTICAST 0x0002U
69 #define IPV6_ADDR_LOOPBACK 0x0010U
70 #define IPV6_ADDR_LINKLOCAL 0x0020U
71 #define IPV6_ADDR_SITELOCAL 0x0040U
73 #define IPV6_ADDR_COMPATv4 0x0080U
75 #define IPV6_ADDR_SCOPE_MASK 0x00f0U
77 #define IPV6_ADDR_MAPPED 0x1000U
82 #define IPV6_ADDR_MC_SCOPE(a) \
83 ((a)->s6_addr[1] & 0x0f) /* nonstandard */
84 #define __IPV6_ADDR_SCOPE_INVALID -1
85 #define IPV6_ADDR_SCOPE_NODELOCAL 0x01
86 #define IPV6_ADDR_SCOPE_LINKLOCAL 0x02
87 #define IPV6_ADDR_SCOPE_SITELOCAL 0x05
88 #define IPV6_ADDR_SCOPE_ORGLOCAL 0x08
89 #define IPV6_ADDR_SCOPE_GLOBAL 0x0e
94 #define IPV6_ADDR_MC_FLAG_TRANSIENT(a) \
95 ((a)->s6_addr[1] & 0x10)
96 #define IPV6_ADDR_MC_FLAG_PREFIX(a) \
97 ((a)->s6_addr[1] & 0x20)
98 #define IPV6_ADDR_MC_FLAG_RENDEZVOUS(a) \
99 ((a)->s6_addr[1] & 0x40)
102 * fragmentation header
109 __be32 identification
;
112 #define IP6_MF 0x0001
113 #define IP6_OFFSET 0xFFF8
115 #define IP6_REPLY_MARK(net, mark) \
116 ((net)->ipv6.sysctl.fwmark_reflect ? (mark) : 0)
118 #include <net/sock.h>
121 extern int sysctl_mld_max_msf
;
122 extern int sysctl_mld_qrv
;
124 #define _DEVINC(net, statname, modifier, idev, field) \
126 struct inet6_dev *_idev = (idev); \
127 if (likely(_idev != NULL)) \
128 SNMP_INC_STATS##modifier((_idev)->stats.statname, (field)); \
129 SNMP_INC_STATS##modifier((net)->mib.statname##_statistics, (field));\
132 /* per device counters are atomic_long_t */
133 #define _DEVINCATOMIC(net, statname, modifier, idev, field) \
135 struct inet6_dev *_idev = (idev); \
136 if (likely(_idev != NULL)) \
137 SNMP_INC_STATS_ATOMIC_LONG((_idev)->stats.statname##dev, (field)); \
138 SNMP_INC_STATS##modifier((net)->mib.statname##_statistics, (field));\
141 /* per device and per net counters are atomic_long_t */
142 #define _DEVINC_ATOMIC_ATOMIC(net, statname, idev, field) \
144 struct inet6_dev *_idev = (idev); \
145 if (likely(_idev != NULL)) \
146 SNMP_INC_STATS_ATOMIC_LONG((_idev)->stats.statname##dev, (field)); \
147 SNMP_INC_STATS_ATOMIC_LONG((net)->mib.statname##_statistics, (field));\
150 #define _DEVADD(net, statname, modifier, idev, field, val) \
152 struct inet6_dev *_idev = (idev); \
153 if (likely(_idev != NULL)) \
154 SNMP_ADD_STATS##modifier((_idev)->stats.statname, (field), (val)); \
155 SNMP_ADD_STATS##modifier((net)->mib.statname##_statistics, (field), (val));\
158 #define _DEVUPD(net, statname, modifier, idev, field, val) \
160 struct inet6_dev *_idev = (idev); \
161 if (likely(_idev != NULL)) \
162 SNMP_UPD_PO_STATS##modifier((_idev)->stats.statname, field, (val)); \
163 SNMP_UPD_PO_STATS##modifier((net)->mib.statname##_statistics, field, (val));\
168 #define IP6_INC_STATS(net, idev,field) \
169 _DEVINC(net, ipv6, 64, idev, field)
170 #define IP6_INC_STATS_BH(net, idev,field) \
171 _DEVINC(net, ipv6, 64_BH, idev, field)
172 #define IP6_ADD_STATS(net, idev,field,val) \
173 _DEVADD(net, ipv6, 64, idev, field, val)
174 #define IP6_ADD_STATS_BH(net, idev,field,val) \
175 _DEVADD(net, ipv6, 64_BH, idev, field, val)
176 #define IP6_UPD_PO_STATS(net, idev,field,val) \
177 _DEVUPD(net, ipv6, 64, idev, field, val)
178 #define IP6_UPD_PO_STATS_BH(net, idev,field,val) \
179 _DEVUPD(net, ipv6, 64_BH, idev, field, val)
180 #define ICMP6_INC_STATS(net, idev, field) \
181 _DEVINCATOMIC(net, icmpv6, , idev, field)
182 #define ICMP6_INC_STATS_BH(net, idev, field) \
183 _DEVINCATOMIC(net, icmpv6, _BH, idev, field)
185 #define ICMP6MSGOUT_INC_STATS(net, idev, field) \
186 _DEVINC_ATOMIC_ATOMIC(net, icmpv6msg, idev, field +256)
187 #define ICMP6MSGOUT_INC_STATS_BH(net, idev, field) \
188 _DEVINC_ATOMIC_ATOMIC(net, icmpv6msg, idev, field +256)
189 #define ICMP6MSGIN_INC_STATS_BH(net, idev, field) \
190 _DEVINC_ATOMIC_ATOMIC(net, icmpv6msg, idev, field)
192 struct ip6_ra_chain
{
193 struct ip6_ra_chain
*next
;
196 void (*destructor
)(struct sock
*);
199 extern struct ip6_ra_chain
*ip6_ra_chain
;
200 extern rwlock_t ip6_ra_lock
;
203 This structure is prepared by protocol, when parsing
204 ancillary data and passed to IPv6.
207 struct ipv6_txoptions
{
209 /* Length of this structure */
212 /* length of extension headers */
214 __u16 opt_flen
; /* after fragment hdr */
215 __u16 opt_nflen
; /* before fragment hdr */
217 struct ipv6_opt_hdr
*hopopt
;
218 struct ipv6_opt_hdr
*dst0opt
;
219 struct ipv6_rt_hdr
*srcrt
; /* Routing Header */
220 struct ipv6_opt_hdr
*dst1opt
;
222 /* Option buffer, as read by IPV6_PKTOPTIONS, starts here. */
225 struct ip6_flowlabel
{
226 struct ip6_flowlabel __rcu
*next
;
230 struct ipv6_txoptions
*opt
;
231 unsigned long linger
;
238 unsigned long lastuse
;
239 unsigned long expires
;
243 #define IPV6_FLOWINFO_MASK cpu_to_be32(0x0FFFFFFF)
244 #define IPV6_FLOWLABEL_MASK cpu_to_be32(0x000FFFFF)
245 #define IPV6_FLOWLABEL_STATELESS_FLAG cpu_to_be32(0x00080000)
247 #define IPV6_TCLASS_MASK (IPV6_FLOWINFO_MASK & ~IPV6_FLOWLABEL_MASK)
248 #define IPV6_TCLASS_SHIFT 20
250 struct ipv6_fl_socklist
{
251 struct ipv6_fl_socklist __rcu
*next
;
252 struct ip6_flowlabel
*fl
;
256 static inline struct ipv6_txoptions
*txopt_get(const struct ipv6_pinfo
*np
)
258 struct ipv6_txoptions
*opt
;
261 opt
= rcu_dereference(np
->opt
);
263 if (!atomic_inc_not_zero(&opt
->refcnt
))
266 opt
= rcu_pointer_handoff(opt
);
272 static inline void txopt_put(struct ipv6_txoptions
*opt
)
274 if (opt
&& atomic_dec_and_test(&opt
->refcnt
))
278 struct ip6_flowlabel
*fl6_sock_lookup(struct sock
*sk
, __be32 label
);
279 struct ipv6_txoptions
*fl6_merge_options(struct ipv6_txoptions
*opt_space
,
280 struct ip6_flowlabel
*fl
,
281 struct ipv6_txoptions
*fopt
);
282 void fl6_free_socklist(struct sock
*sk
);
283 int ipv6_flowlabel_opt(struct sock
*sk
, char __user
*optval
, int optlen
);
284 int ipv6_flowlabel_opt_get(struct sock
*sk
, struct in6_flowlabel_req
*freq
,
286 int ip6_flowlabel_init(void);
287 void ip6_flowlabel_cleanup(void);
289 static inline void fl6_sock_release(struct ip6_flowlabel
*fl
)
292 atomic_dec(&fl
->users
);
295 void icmpv6_notify(struct sk_buff
*skb
, u8 type
, u8 code
, __be32 info
);
297 int icmpv6_push_pending_frames(struct sock
*sk
, struct flowi6
*fl6
,
298 struct icmp6hdr
*thdr
, int len
);
300 int ip6_ra_control(struct sock
*sk
, int sel
);
302 int ipv6_parse_hopopts(struct sk_buff
*skb
);
304 struct ipv6_txoptions
*ipv6_dup_options(struct sock
*sk
,
305 struct ipv6_txoptions
*opt
);
306 struct ipv6_txoptions
*ipv6_renew_options(struct sock
*sk
,
307 struct ipv6_txoptions
*opt
,
309 struct ipv6_opt_hdr __user
*newopt
,
311 struct ipv6_txoptions
*
312 ipv6_renew_options_kern(struct sock
*sk
,
313 struct ipv6_txoptions
*opt
,
315 struct ipv6_opt_hdr
*newopt
,
317 struct ipv6_txoptions
*ipv6_fixup_options(struct ipv6_txoptions
*opt_space
,
318 struct ipv6_txoptions
*opt
);
320 bool ipv6_opt_accepted(const struct sock
*sk
, const struct sk_buff
*skb
,
321 const struct inet6_skb_parm
*opt
);
322 struct ipv6_txoptions
*ipv6_update_options(struct sock
*sk
,
323 struct ipv6_txoptions
*opt
);
325 static inline bool ipv6_accept_ra(struct inet6_dev
*idev
)
327 /* If forwarding is enabled, RA are not accepted unless the special
328 * hybrid mode (accept_ra=2) is enabled.
330 return idev
->cnf
.forwarding
? idev
->cnf
.accept_ra
== 2 :
334 #if IS_ENABLED(CONFIG_IPV6)
335 static inline int ip6_frag_mem(struct net
*net
)
337 return sum_frag_mem_limit(&net
->ipv6
.frags
);
341 #define IPV6_FRAG_HIGH_THRESH (4 * 1024*1024) /* 4194304 */
342 #define IPV6_FRAG_LOW_THRESH (3 * 1024*1024) /* 3145728 */
343 #define IPV6_FRAG_TIMEOUT (60 * HZ) /* 60 seconds */
345 int __ipv6_addr_type(const struct in6_addr
*addr
);
346 static inline int ipv6_addr_type(const struct in6_addr
*addr
)
348 return __ipv6_addr_type(addr
) & 0xffff;
351 static inline int ipv6_addr_scope(const struct in6_addr
*addr
)
353 return __ipv6_addr_type(addr
) & IPV6_ADDR_SCOPE_MASK
;
356 static inline int __ipv6_addr_src_scope(int type
)
358 return (type
== IPV6_ADDR_ANY
) ? __IPV6_ADDR_SCOPE_INVALID
: (type
>> 16);
361 static inline int ipv6_addr_src_scope(const struct in6_addr
*addr
)
363 return __ipv6_addr_src_scope(__ipv6_addr_type(addr
));
366 static inline bool __ipv6_addr_needs_scope_id(int type
)
368 return type
& IPV6_ADDR_LINKLOCAL
||
369 (type
& IPV6_ADDR_MULTICAST
&&
370 (type
& (IPV6_ADDR_LOOPBACK
|IPV6_ADDR_LINKLOCAL
)));
373 static inline __u32
ipv6_iface_scope_id(const struct in6_addr
*addr
, int iface
)
375 return __ipv6_addr_needs_scope_id(__ipv6_addr_type(addr
)) ? iface
: 0;
378 static inline int ipv6_addr_cmp(const struct in6_addr
*a1
, const struct in6_addr
*a2
)
380 return memcmp(a1
, a2
, sizeof(struct in6_addr
));
384 ipv6_masked_addr_cmp(const struct in6_addr
*a1
, const struct in6_addr
*m
,
385 const struct in6_addr
*a2
)
387 #if defined(CONFIG_HAVE_EFFICIENT_UNALIGNED_ACCESS) && BITS_PER_LONG == 64
388 const unsigned long *ul1
= (const unsigned long *)a1
;
389 const unsigned long *ulm
= (const unsigned long *)m
;
390 const unsigned long *ul2
= (const unsigned long *)a2
;
392 return !!(((ul1
[0] ^ ul2
[0]) & ulm
[0]) |
393 ((ul1
[1] ^ ul2
[1]) & ulm
[1]));
395 return !!(((a1
->s6_addr32
[0] ^ a2
->s6_addr32
[0]) & m
->s6_addr32
[0]) |
396 ((a1
->s6_addr32
[1] ^ a2
->s6_addr32
[1]) & m
->s6_addr32
[1]) |
397 ((a1
->s6_addr32
[2] ^ a2
->s6_addr32
[2]) & m
->s6_addr32
[2]) |
398 ((a1
->s6_addr32
[3] ^ a2
->s6_addr32
[3]) & m
->s6_addr32
[3]));
402 static inline void ipv6_addr_prefix(struct in6_addr
*pfx
,
403 const struct in6_addr
*addr
,
406 /* caller must guarantee 0 <= plen <= 128 */
410 memset(pfx
->s6_addr
, 0, sizeof(pfx
->s6_addr
));
411 memcpy(pfx
->s6_addr
, addr
, o
);
413 pfx
->s6_addr
[o
] = addr
->s6_addr
[o
] & (0xff00 >> b
);
416 static inline void ipv6_addr_prefix_copy(struct in6_addr
*addr
,
417 const struct in6_addr
*pfx
,
420 /* caller must guarantee 0 <= plen <= 128 */
424 memcpy(addr
->s6_addr
, pfx
, o
);
426 addr
->s6_addr
[o
] &= ~(0xff00 >> b
);
427 addr
->s6_addr
[o
] |= (pfx
->s6_addr
[o
] & (0xff00 >> b
));
431 static inline void __ipv6_addr_set_half(__be32
*addr
,
432 __be32 wh
, __be32 wl
)
434 #if defined(CONFIG_HAVE_EFFICIENT_UNALIGNED_ACCESS) && BITS_PER_LONG == 64
435 #if defined(__BIG_ENDIAN)
436 if (__builtin_constant_p(wh
) && __builtin_constant_p(wl
)) {
437 *(__force u64
*)addr
= ((__force u64
)(wh
) << 32 | (__force u64
)(wl
));
440 #elif defined(__LITTLE_ENDIAN)
441 if (__builtin_constant_p(wl
) && __builtin_constant_p(wh
)) {
442 *(__force u64
*)addr
= ((__force u64
)(wl
) << 32 | (__force u64
)(wh
));
451 static inline void ipv6_addr_set(struct in6_addr
*addr
,
452 __be32 w1
, __be32 w2
,
453 __be32 w3
, __be32 w4
)
455 __ipv6_addr_set_half(&addr
->s6_addr32
[0], w1
, w2
);
456 __ipv6_addr_set_half(&addr
->s6_addr32
[2], w3
, w4
);
459 static inline bool ipv6_addr_equal(const struct in6_addr
*a1
,
460 const struct in6_addr
*a2
)
462 #if defined(CONFIG_HAVE_EFFICIENT_UNALIGNED_ACCESS) && BITS_PER_LONG == 64
463 const unsigned long *ul1
= (const unsigned long *)a1
;
464 const unsigned long *ul2
= (const unsigned long *)a2
;
466 return ((ul1
[0] ^ ul2
[0]) | (ul1
[1] ^ ul2
[1])) == 0UL;
468 return ((a1
->s6_addr32
[0] ^ a2
->s6_addr32
[0]) |
469 (a1
->s6_addr32
[1] ^ a2
->s6_addr32
[1]) |
470 (a1
->s6_addr32
[2] ^ a2
->s6_addr32
[2]) |
471 (a1
->s6_addr32
[3] ^ a2
->s6_addr32
[3])) == 0;
475 #if defined(CONFIG_HAVE_EFFICIENT_UNALIGNED_ACCESS) && BITS_PER_LONG == 64
476 static inline bool __ipv6_prefix_equal64_half(const __be64
*a1
,
480 if (len
&& ((*a1
^ *a2
) & cpu_to_be64((~0UL) << (64 - len
))))
485 static inline bool ipv6_prefix_equal(const struct in6_addr
*addr1
,
486 const struct in6_addr
*addr2
,
487 unsigned int prefixlen
)
489 const __be64
*a1
= (const __be64
*)addr1
;
490 const __be64
*a2
= (const __be64
*)addr2
;
492 if (prefixlen
>= 64) {
495 return __ipv6_prefix_equal64_half(a1
+ 1, a2
+ 1, prefixlen
- 64);
497 return __ipv6_prefix_equal64_half(a1
, a2
, prefixlen
);
500 static inline bool ipv6_prefix_equal(const struct in6_addr
*addr1
,
501 const struct in6_addr
*addr2
,
502 unsigned int prefixlen
)
504 const __be32
*a1
= addr1
->s6_addr32
;
505 const __be32
*a2
= addr2
->s6_addr32
;
506 unsigned int pdw
, pbi
;
508 /* check complete u32 in prefix */
509 pdw
= prefixlen
>> 5;
510 if (pdw
&& memcmp(a1
, a2
, pdw
<< 2))
513 /* check incomplete u32 in prefix */
514 pbi
= prefixlen
& 0x1f;
515 if (pbi
&& ((a1
[pdw
] ^ a2
[pdw
]) & htonl((0xffffffff) << (32 - pbi
))))
522 struct inet_frag_queue
;
524 enum ip6_defrag_users
{
525 IP6_DEFRAG_LOCAL_DELIVER
,
526 IP6_DEFRAG_CONNTRACK_IN
,
527 __IP6_DEFRAG_CONNTRACK_IN
= IP6_DEFRAG_CONNTRACK_IN
+ USHRT_MAX
,
528 IP6_DEFRAG_CONNTRACK_OUT
,
529 __IP6_DEFRAG_CONNTRACK_OUT
= IP6_DEFRAG_CONNTRACK_OUT
+ USHRT_MAX
,
530 IP6_DEFRAG_CONNTRACK_BRIDGE_IN
,
531 __IP6_DEFRAG_CONNTRACK_BRIDGE_IN
= IP6_DEFRAG_CONNTRACK_BRIDGE_IN
+ USHRT_MAX
,
534 struct ip6_create_arg
{
537 const struct in6_addr
*src
;
538 const struct in6_addr
*dst
;
543 void ip6_frag_init(struct inet_frag_queue
*q
, const void *a
);
544 bool ip6_frag_match(const struct inet_frag_queue
*q
, const void *a
);
547 * Equivalent of ipv4 struct ip
550 struct inet_frag_queue q
;
552 __be32 id
; /* fragment id */
554 struct in6_addr saddr
;
555 struct in6_addr daddr
;
563 void ip6_expire_frag_queue(struct net
*net
, struct frag_queue
*fq
,
564 struct inet_frags
*frags
);
566 static inline bool ipv6_addr_any(const struct in6_addr
*a
)
568 #if defined(CONFIG_HAVE_EFFICIENT_UNALIGNED_ACCESS) && BITS_PER_LONG == 64
569 const unsigned long *ul
= (const unsigned long *)a
;
571 return (ul
[0] | ul
[1]) == 0UL;
573 return (a
->s6_addr32
[0] | a
->s6_addr32
[1] |
574 a
->s6_addr32
[2] | a
->s6_addr32
[3]) == 0;
578 static inline u32
ipv6_addr_hash(const struct in6_addr
*a
)
580 #if defined(CONFIG_HAVE_EFFICIENT_UNALIGNED_ACCESS) && BITS_PER_LONG == 64
581 const unsigned long *ul
= (const unsigned long *)a
;
582 unsigned long x
= ul
[0] ^ ul
[1];
584 return (u32
)(x
^ (x
>> 32));
586 return (__force u32
)(a
->s6_addr32
[0] ^ a
->s6_addr32
[1] ^
587 a
->s6_addr32
[2] ^ a
->s6_addr32
[3]);
591 /* more secured version of ipv6_addr_hash() */
592 static inline u32
__ipv6_addr_jhash(const struct in6_addr
*a
, const u32 initval
)
594 u32 v
= (__force u32
)a
->s6_addr32
[0] ^ (__force u32
)a
->s6_addr32
[1];
596 return jhash_3words(v
,
597 (__force u32
)a
->s6_addr32
[2],
598 (__force u32
)a
->s6_addr32
[3],
602 static inline bool ipv6_addr_loopback(const struct in6_addr
*a
)
604 #if defined(CONFIG_HAVE_EFFICIENT_UNALIGNED_ACCESS) && BITS_PER_LONG == 64
605 const __be64
*be
= (const __be64
*)a
;
607 return (be
[0] | (be
[1] ^ cpu_to_be64(1))) == 0UL;
609 return (a
->s6_addr32
[0] | a
->s6_addr32
[1] |
610 a
->s6_addr32
[2] | (a
->s6_addr32
[3] ^ cpu_to_be32(1))) == 0;
615 * Note that we must __force cast these to unsigned long to make sparse happy,
616 * since all of the endian-annotated types are fixed size regardless of arch.
618 static inline bool ipv6_addr_v4mapped(const struct in6_addr
*a
)
621 #if defined(CONFIG_HAVE_EFFICIENT_UNALIGNED_ACCESS) && BITS_PER_LONG == 64
622 *(unsigned long *)a
|
624 (__force
unsigned long)(a
->s6_addr32
[0] | a
->s6_addr32
[1]) |
626 (__force
unsigned long)(a
->s6_addr32
[2] ^
627 cpu_to_be32(0x0000ffff))) == 0UL;
631 * Check for a RFC 4843 ORCHID address
632 * (Overlay Routable Cryptographic Hash Identifiers)
634 static inline bool ipv6_addr_orchid(const struct in6_addr
*a
)
636 return (a
->s6_addr32
[0] & htonl(0xfffffff0)) == htonl(0x20010010);
639 static inline bool ipv6_addr_is_multicast(const struct in6_addr
*addr
)
641 return (addr
->s6_addr32
[0] & htonl(0xFF000000)) == htonl(0xFF000000);
644 static inline void ipv6_addr_set_v4mapped(const __be32 addr
,
645 struct in6_addr
*v4mapped
)
647 ipv6_addr_set(v4mapped
,
654 * find the first different bit between two addresses
655 * length of address must be a multiple of 32bits
657 static inline int __ipv6_addr_diff32(const void *token1
, const void *token2
, int addrlen
)
659 const __be32
*a1
= token1
, *a2
= token2
;
664 for (i
= 0; i
< addrlen
; i
++) {
665 __be32 xb
= a1
[i
] ^ a2
[i
];
667 return i
* 32 + 31 - __fls(ntohl(xb
));
671 * we should *never* get to this point since that
672 * would mean the addrs are equal
674 * However, we do get to it 8) And exacly, when
675 * addresses are equal 8)
677 * ip route add 1111::/128 via ...
678 * ip route add 1111::/64 via ...
681 * Ideally, this function should stop comparison
682 * at prefix length. It does not, but it is still OK,
683 * if returned value is greater than prefix length.
689 #if defined(CONFIG_HAVE_EFFICIENT_UNALIGNED_ACCESS) && BITS_PER_LONG == 64
690 static inline int __ipv6_addr_diff64(const void *token1
, const void *token2
, int addrlen
)
692 const __be64
*a1
= token1
, *a2
= token2
;
697 for (i
= 0; i
< addrlen
; i
++) {
698 __be64 xb
= a1
[i
] ^ a2
[i
];
700 return i
* 64 + 63 - __fls(be64_to_cpu(xb
));
707 static inline int __ipv6_addr_diff(const void *token1
, const void *token2
, int addrlen
)
709 #if defined(CONFIG_HAVE_EFFICIENT_UNALIGNED_ACCESS) && BITS_PER_LONG == 64
710 if (__builtin_constant_p(addrlen
) && !(addrlen
& 7))
711 return __ipv6_addr_diff64(token1
, token2
, addrlen
);
713 return __ipv6_addr_diff32(token1
, token2
, addrlen
);
716 static inline int ipv6_addr_diff(const struct in6_addr
*a1
, const struct in6_addr
*a2
)
718 return __ipv6_addr_diff(a1
, a2
, sizeof(struct in6_addr
));
721 __be32
ipv6_select_ident(struct net
*net
,
722 const struct in6_addr
*daddr
,
723 const struct in6_addr
*saddr
);
724 void ipv6_proxy_select_ident(struct net
*net
, struct sk_buff
*skb
);
726 int ip6_dst_hoplimit(struct dst_entry
*dst
);
728 static inline int ip6_sk_dst_hoplimit(struct ipv6_pinfo
*np
, struct flowi6
*fl6
,
729 struct dst_entry
*dst
)
733 if (ipv6_addr_is_multicast(&fl6
->daddr
))
734 hlimit
= np
->mcast_hops
;
736 hlimit
= np
->hop_limit
;
738 hlimit
= ip6_dst_hoplimit(dst
);
742 /* copy IPv6 saddr & daddr to flow_keys, possibly using 64bit load/store
743 * Equivalent to : flow->v6addrs.src = iph->saddr;
744 * flow->v6addrs.dst = iph->daddr;
746 static inline void iph_to_flow_copy_v6addrs(struct flow_keys
*flow
,
747 const struct ipv6hdr
*iph
)
749 BUILD_BUG_ON(offsetof(typeof(flow
->addrs
), v6addrs
.dst
) !=
750 offsetof(typeof(flow
->addrs
), v6addrs
.src
) +
751 sizeof(flow
->addrs
.v6addrs
.src
));
752 memcpy(&flow
->addrs
.v6addrs
, &iph
->saddr
, sizeof(flow
->addrs
.v6addrs
));
753 flow
->control
.addr_type
= FLOW_DISSECTOR_KEY_IPV6_ADDRS
;
756 #if IS_ENABLED(CONFIG_IPV6)
758 /* Sysctl settings for net ipv6.auto_flowlabels */
759 #define IP6_AUTO_FLOW_LABEL_OFF 0
760 #define IP6_AUTO_FLOW_LABEL_OPTOUT 1
761 #define IP6_AUTO_FLOW_LABEL_OPTIN 2
762 #define IP6_AUTO_FLOW_LABEL_FORCED 3
764 #define IP6_AUTO_FLOW_LABEL_MAX IP6_AUTO_FLOW_LABEL_FORCED
766 #define IP6_DEFAULT_AUTO_FLOW_LABELS IP6_AUTO_FLOW_LABEL_OPTOUT
768 static inline __be32
ip6_make_flowlabel(struct net
*net
, struct sk_buff
*skb
,
769 __be32 flowlabel
, bool autolabel
,
775 net
->ipv6
.sysctl
.auto_flowlabels
== IP6_AUTO_FLOW_LABEL_OFF
||
777 net
->ipv6
.sysctl
.auto_flowlabels
!= IP6_AUTO_FLOW_LABEL_FORCED
))
780 hash
= skb_get_hash_flowi6(skb
, fl6
);
782 /* Since this is being sent on the wire obfuscate hash a bit
783 * to minimize possbility that any useful information to an
784 * attacker is leaked. Only lower 20 bits are relevant.
788 flowlabel
= (__force __be32
)hash
& IPV6_FLOWLABEL_MASK
;
790 if (net
->ipv6
.sysctl
.flowlabel_state_ranges
)
791 flowlabel
|= IPV6_FLOWLABEL_STATELESS_FLAG
;
796 static inline int ip6_default_np_autolabel(struct net
*net
)
798 switch (net
->ipv6
.sysctl
.auto_flowlabels
) {
799 case IP6_AUTO_FLOW_LABEL_OFF
:
800 case IP6_AUTO_FLOW_LABEL_OPTIN
:
803 case IP6_AUTO_FLOW_LABEL_OPTOUT
:
804 case IP6_AUTO_FLOW_LABEL_FORCED
:
809 static inline void ip6_set_txhash(struct sock
*sk
) { }
810 static inline __be32
ip6_make_flowlabel(struct net
*net
, struct sk_buff
*skb
,
811 __be32 flowlabel
, bool autolabel
,
816 static inline int ip6_default_np_autolabel(struct net
*net
)
824 * Header manipulation
826 static inline void ip6_flow_hdr(struct ipv6hdr
*hdr
, unsigned int tclass
,
829 *(__be32
*)hdr
= htonl(0x60000000 | (tclass
<< 20)) | flowlabel
;
832 static inline __be32
ip6_flowinfo(const struct ipv6hdr
*hdr
)
834 return *(__be32
*)hdr
& IPV6_FLOWINFO_MASK
;
837 static inline __be32
ip6_flowlabel(const struct ipv6hdr
*hdr
)
839 return *(__be32
*)hdr
& IPV6_FLOWLABEL_MASK
;
842 static inline u8
ip6_tclass(__be32 flowinfo
)
844 return ntohl(flowinfo
& IPV6_TCLASS_MASK
) >> IPV6_TCLASS_SHIFT
;
847 static inline __be32
ip6_make_flowinfo(unsigned int tclass
, __be32 flowlabel
)
849 return htonl(tclass
<< IPV6_TCLASS_SHIFT
) | flowlabel
;
853 * Prototypes exported by ipv6
857 * rcv function (called from netdevice level)
860 int ipv6_rcv(struct sk_buff
*skb
, struct net_device
*dev
,
861 struct packet_type
*pt
, struct net_device
*orig_dev
);
863 int ip6_rcv_finish(struct net
*net
, struct sock
*sk
, struct sk_buff
*skb
);
866 * upper-layer output functions
868 int ip6_xmit(const struct sock
*sk
, struct sk_buff
*skb
, struct flowi6
*fl6
,
869 struct ipv6_txoptions
*opt
, int tclass
);
871 int ip6_find_1stfragopt(struct sk_buff
*skb
, u8
**nexthdr
);
873 int ip6_append_data(struct sock
*sk
,
874 int getfrag(void *from
, char *to
, int offset
, int len
,
875 int odd
, struct sk_buff
*skb
),
876 void *from
, int length
, int transhdrlen
, int hlimit
,
877 int tclass
, struct ipv6_txoptions
*opt
, struct flowi6
*fl6
,
878 struct rt6_info
*rt
, unsigned int flags
, int dontfrag
);
880 int ip6_push_pending_frames(struct sock
*sk
);
882 void ip6_flush_pending_frames(struct sock
*sk
);
884 int ip6_send_skb(struct sk_buff
*skb
);
886 struct sk_buff
*__ip6_make_skb(struct sock
*sk
, struct sk_buff_head
*queue
,
887 struct inet_cork_full
*cork
,
888 struct inet6_cork
*v6_cork
);
889 struct sk_buff
*ip6_make_skb(struct sock
*sk
,
890 int getfrag(void *from
, char *to
, int offset
,
891 int len
, int odd
, struct sk_buff
*skb
),
892 void *from
, int length
, int transhdrlen
,
893 int hlimit
, int tclass
, struct ipv6_txoptions
*opt
,
894 struct flowi6
*fl6
, struct rt6_info
*rt
,
895 unsigned int flags
, int dontfrag
);
897 static inline struct sk_buff
*ip6_finish_skb(struct sock
*sk
)
899 return __ip6_make_skb(sk
, &sk
->sk_write_queue
, &inet_sk(sk
)->cork
,
900 &inet6_sk(sk
)->cork
);
903 int ip6_dst_lookup(struct net
*net
, struct sock
*sk
, struct dst_entry
**dst
,
905 struct dst_entry
*ip6_dst_lookup_flow(const struct sock
*sk
, struct flowi6
*fl6
,
906 const struct in6_addr
*final_dst
);
907 struct dst_entry
*ip6_sk_dst_lookup_flow(struct sock
*sk
, struct flowi6
*fl6
,
908 const struct in6_addr
*final_dst
);
909 struct dst_entry
*ip6_blackhole_route(struct net
*net
,
910 struct dst_entry
*orig_dst
);
913 * skb processing functions
916 int ip6_output(struct net
*net
, struct sock
*sk
, struct sk_buff
*skb
);
917 int ip6_forward(struct sk_buff
*skb
);
918 int ip6_input(struct sk_buff
*skb
);
919 int ip6_mc_input(struct sk_buff
*skb
);
921 int __ip6_local_out(struct net
*net
, struct sock
*sk
, struct sk_buff
*skb
);
922 int ip6_local_out(struct net
*net
, struct sock
*sk
, struct sk_buff
*skb
);
925 * Extension header (options) processing
928 void ipv6_push_nfrag_opts(struct sk_buff
*skb
, struct ipv6_txoptions
*opt
,
929 u8
*proto
, struct in6_addr
**daddr_p
);
930 void ipv6_push_frag_opts(struct sk_buff
*skb
, struct ipv6_txoptions
*opt
,
933 int ipv6_skip_exthdr(const struct sk_buff
*, int start
, u8
*nexthdrp
,
936 bool ipv6_ext_hdr(u8 nexthdr
);
939 IP6_FH_F_FRAG
= (1 << 0),
940 IP6_FH_F_AUTH
= (1 << 1),
941 IP6_FH_F_SKIP_RH
= (1 << 2),
944 /* find specified header and get offset to it */
945 int ipv6_find_hdr(const struct sk_buff
*skb
, unsigned int *offset
, int target
,
946 unsigned short *fragoff
, int *fragflg
);
948 int ipv6_find_tlv(const struct sk_buff
*skb
, int offset
, int type
);
950 struct in6_addr
*fl6_update_dst(struct flowi6
*fl6
,
951 const struct ipv6_txoptions
*opt
,
952 struct in6_addr
*orig
);
955 * socket options (ipv6_sockglue.c)
958 int ipv6_setsockopt(struct sock
*sk
, int level
, int optname
,
959 char __user
*optval
, unsigned int optlen
);
960 int ipv6_getsockopt(struct sock
*sk
, int level
, int optname
,
961 char __user
*optval
, int __user
*optlen
);
962 int compat_ipv6_setsockopt(struct sock
*sk
, int level
, int optname
,
963 char __user
*optval
, unsigned int optlen
);
964 int compat_ipv6_getsockopt(struct sock
*sk
, int level
, int optname
,
965 char __user
*optval
, int __user
*optlen
);
967 int ip6_datagram_connect(struct sock
*sk
, struct sockaddr
*addr
, int addr_len
);
968 int ip6_datagram_connect_v6_only(struct sock
*sk
, struct sockaddr
*addr
,
971 int ipv6_recv_error(struct sock
*sk
, struct msghdr
*msg
, int len
,
973 int ipv6_recv_rxpmtu(struct sock
*sk
, struct msghdr
*msg
, int len
,
975 void ipv6_icmp_error(struct sock
*sk
, struct sk_buff
*skb
, int err
, __be16 port
,
976 u32 info
, u8
*payload
);
977 void ipv6_local_error(struct sock
*sk
, int err
, struct flowi6
*fl6
, u32 info
);
978 void ipv6_local_rxpmtu(struct sock
*sk
, struct flowi6
*fl6
, u32 mtu
);
980 int inet6_release(struct socket
*sock
);
981 int inet6_bind(struct socket
*sock
, struct sockaddr
*uaddr
, int addr_len
);
982 int inet6_getname(struct socket
*sock
, struct sockaddr
*uaddr
, int *uaddr_len
,
984 int inet6_ioctl(struct socket
*sock
, unsigned int cmd
, unsigned long arg
);
986 int inet6_hash_connect(struct inet_timewait_death_row
*death_row
,
992 extern const struct proto_ops inet6_stream_ops
;
993 extern const struct proto_ops inet6_dgram_ops
;
995 struct group_source_req
;
998 int ip6_mc_source(int add
, int omode
, struct sock
*sk
,
999 struct group_source_req
*pgsr
);
1000 int ip6_mc_msfilter(struct sock
*sk
, struct group_filter
*gsf
);
1001 int ip6_mc_msfget(struct sock
*sk
, struct group_filter
*gsf
,
1002 struct group_filter __user
*optval
, int __user
*optlen
);
1004 #ifdef CONFIG_PROC_FS
1005 int ac6_proc_init(struct net
*net
);
1006 void ac6_proc_exit(struct net
*net
);
1007 int raw6_proc_init(void);
1008 void raw6_proc_exit(void);
1009 int tcp6_proc_init(struct net
*net
);
1010 void tcp6_proc_exit(struct net
*net
);
1011 int udp6_proc_init(struct net
*net
);
1012 void udp6_proc_exit(struct net
*net
);
1013 int udplite6_proc_init(void);
1014 void udplite6_proc_exit(void);
1015 int ipv6_misc_proc_init(void);
1016 void ipv6_misc_proc_exit(void);
1017 int snmp6_register_dev(struct inet6_dev
*idev
);
1018 int snmp6_unregister_dev(struct inet6_dev
*idev
);
1021 static inline int ac6_proc_init(struct net
*net
) { return 0; }
1022 static inline void ac6_proc_exit(struct net
*net
) { }
1023 static inline int snmp6_register_dev(struct inet6_dev
*idev
) { return 0; }
1024 static inline int snmp6_unregister_dev(struct inet6_dev
*idev
) { return 0; }
1027 #ifdef CONFIG_SYSCTL
1028 extern struct ctl_table ipv6_route_table_template
[];
1030 struct ctl_table
*ipv6_icmp_sysctl_init(struct net
*net
);
1031 struct ctl_table
*ipv6_route_sysctl_init(struct net
*net
);
1032 int ipv6_sysctl_register(void);
1033 void ipv6_sysctl_unregister(void);
1036 int ipv6_sock_mc_join(struct sock
*sk
, int ifindex
,
1037 const struct in6_addr
*addr
);
1038 int ipv6_sock_mc_drop(struct sock
*sk
, int ifindex
,
1039 const struct in6_addr
*addr
);
1040 #endif /* _NET_IPV6_H */