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, mod, idev, field) \
126 struct inet6_dev *_idev = (idev); \
127 if (likely(_idev != NULL)) \
128 mod##SNMP_INC_STATS64((_idev)->stats.statname, (field));\
129 mod##SNMP_INC_STATS64((net)->mib.statname##_statistics, (field));\
132 /* per device counters are atomic_long_t */
133 #define _DEVINCATOMIC(net, statname, mod, 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 mod##SNMP_INC_STATS((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, mod, idev, field, val) \
152 struct inet6_dev *_idev = (idev); \
153 if (likely(_idev != NULL)) \
154 mod##SNMP_ADD_STATS((_idev)->stats.statname, (field), (val)); \
155 mod##SNMP_ADD_STATS((net)->mib.statname##_statistics, (field), (val));\
158 #define _DEVUPD(net, statname, mod, idev, field, val) \
160 struct inet6_dev *_idev = (idev); \
161 if (likely(_idev != NULL)) \
162 mod##SNMP_UPD_PO_STATS((_idev)->stats.statname, field, (val)); \
163 mod##SNMP_UPD_PO_STATS((net)->mib.statname##_statistics, field, (val));\
168 #define IP6_INC_STATS(net, idev,field) \
169 _DEVINC(net, ipv6, , idev, field)
170 #define __IP6_INC_STATS(net, idev,field) \
171 _DEVINC(net, ipv6, __, idev, field)
172 #define IP6_ADD_STATS(net, idev,field,val) \
173 _DEVADD(net, ipv6, , idev, field, val)
174 #define __IP6_ADD_STATS(net, idev,field,val) \
175 _DEVADD(net, ipv6, __, idev, field, val)
176 #define IP6_UPD_PO_STATS(net, idev,field,val) \
177 _DEVUPD(net, ipv6, , idev, field, val)
178 #define __IP6_UPD_PO_STATS(net, idev,field,val) \
179 _DEVUPD(net, ipv6, __, idev, field, val)
180 #define ICMP6_INC_STATS(net, idev, field) \
181 _DEVINCATOMIC(net, icmpv6, , idev, field)
182 #define __ICMP6_INC_STATS(net, idev, field) \
183 _DEVINCATOMIC(net, icmpv6, __, idev, field)
185 #define ICMP6MSGOUT_INC_STATS(net, idev, field) \
186 _DEVINC_ATOMIC_ATOMIC(net, icmpv6msg, idev, field +256)
187 #define ICMP6MSGIN_INC_STATS(net, idev, field) \
188 _DEVINC_ATOMIC_ATOMIC(net, icmpv6msg, idev, field)
190 struct ip6_ra_chain
{
191 struct ip6_ra_chain
*next
;
194 void (*destructor
)(struct sock
*);
197 extern struct ip6_ra_chain
*ip6_ra_chain
;
198 extern rwlock_t ip6_ra_lock
;
201 This structure is prepared by protocol, when parsing
202 ancillary data and passed to IPv6.
205 struct ipv6_txoptions
{
207 /* Length of this structure */
210 /* length of extension headers */
212 __u16 opt_flen
; /* after fragment hdr */
213 __u16 opt_nflen
; /* before fragment hdr */
215 struct ipv6_opt_hdr
*hopopt
;
216 struct ipv6_opt_hdr
*dst0opt
;
217 struct ipv6_rt_hdr
*srcrt
; /* Routing Header */
218 struct ipv6_opt_hdr
*dst1opt
;
220 /* Option buffer, as read by IPV6_PKTOPTIONS, starts here. */
223 struct ip6_flowlabel
{
224 struct ip6_flowlabel __rcu
*next
;
228 struct ipv6_txoptions
*opt
;
229 unsigned long linger
;
236 unsigned long lastuse
;
237 unsigned long expires
;
241 #define IPV6_FLOWINFO_MASK cpu_to_be32(0x0FFFFFFF)
242 #define IPV6_FLOWLABEL_MASK cpu_to_be32(0x000FFFFF)
243 #define IPV6_FLOWLABEL_STATELESS_FLAG cpu_to_be32(0x00080000)
245 #define IPV6_TCLASS_MASK (IPV6_FLOWINFO_MASK & ~IPV6_FLOWLABEL_MASK)
246 #define IPV6_TCLASS_SHIFT 20
248 struct ipv6_fl_socklist
{
249 struct ipv6_fl_socklist __rcu
*next
;
250 struct ip6_flowlabel
*fl
;
254 struct ipcm6_cookie
{
258 struct ipv6_txoptions
*opt
;
261 static inline struct ipv6_txoptions
*txopt_get(const struct ipv6_pinfo
*np
)
263 struct ipv6_txoptions
*opt
;
266 opt
= rcu_dereference(np
->opt
);
268 if (!atomic_inc_not_zero(&opt
->refcnt
))
271 opt
= rcu_pointer_handoff(opt
);
277 static inline void txopt_put(struct ipv6_txoptions
*opt
)
279 if (opt
&& atomic_dec_and_test(&opt
->refcnt
))
283 struct ip6_flowlabel
*fl6_sock_lookup(struct sock
*sk
, __be32 label
);
284 struct ipv6_txoptions
*fl6_merge_options(struct ipv6_txoptions
*opt_space
,
285 struct ip6_flowlabel
*fl
,
286 struct ipv6_txoptions
*fopt
);
287 void fl6_free_socklist(struct sock
*sk
);
288 int ipv6_flowlabel_opt(struct sock
*sk
, char __user
*optval
, int optlen
);
289 int ipv6_flowlabel_opt_get(struct sock
*sk
, struct in6_flowlabel_req
*freq
,
291 int ip6_flowlabel_init(void);
292 void ip6_flowlabel_cleanup(void);
293 bool ip6_autoflowlabel(struct net
*net
, const struct ipv6_pinfo
*np
);
295 static inline void fl6_sock_release(struct ip6_flowlabel
*fl
)
298 atomic_dec(&fl
->users
);
301 void icmpv6_notify(struct sk_buff
*skb
, u8 type
, u8 code
, __be32 info
);
303 int icmpv6_push_pending_frames(struct sock
*sk
, struct flowi6
*fl6
,
304 struct icmp6hdr
*thdr
, int len
);
306 int ip6_ra_control(struct sock
*sk
, int sel
);
308 int ipv6_parse_hopopts(struct sk_buff
*skb
);
310 struct ipv6_txoptions
*ipv6_dup_options(struct sock
*sk
,
311 struct ipv6_txoptions
*opt
);
312 struct ipv6_txoptions
*ipv6_renew_options(struct sock
*sk
,
313 struct ipv6_txoptions
*opt
,
315 struct ipv6_opt_hdr __user
*newopt
,
317 struct ipv6_txoptions
*
318 ipv6_renew_options_kern(struct sock
*sk
,
319 struct ipv6_txoptions
*opt
,
321 struct ipv6_opt_hdr
*newopt
,
323 struct ipv6_txoptions
*ipv6_fixup_options(struct ipv6_txoptions
*opt_space
,
324 struct ipv6_txoptions
*opt
);
326 bool ipv6_opt_accepted(const struct sock
*sk
, const struct sk_buff
*skb
,
327 const struct inet6_skb_parm
*opt
);
328 struct ipv6_txoptions
*ipv6_update_options(struct sock
*sk
,
329 struct ipv6_txoptions
*opt
);
331 static inline bool ipv6_accept_ra(struct inet6_dev
*idev
)
333 /* If forwarding is enabled, RA are not accepted unless the special
334 * hybrid mode (accept_ra=2) is enabled.
336 return idev
->cnf
.forwarding
? idev
->cnf
.accept_ra
== 2 :
340 #if IS_ENABLED(CONFIG_IPV6)
341 static inline int ip6_frag_mem(struct net
*net
)
343 return sum_frag_mem_limit(&net
->ipv6
.frags
);
347 #define IPV6_FRAG_HIGH_THRESH (4 * 1024*1024) /* 4194304 */
348 #define IPV6_FRAG_LOW_THRESH (3 * 1024*1024) /* 3145728 */
349 #define IPV6_FRAG_TIMEOUT (60 * HZ) /* 60 seconds */
351 int __ipv6_addr_type(const struct in6_addr
*addr
);
352 static inline int ipv6_addr_type(const struct in6_addr
*addr
)
354 return __ipv6_addr_type(addr
) & 0xffff;
357 static inline int ipv6_addr_scope(const struct in6_addr
*addr
)
359 return __ipv6_addr_type(addr
) & IPV6_ADDR_SCOPE_MASK
;
362 static inline int __ipv6_addr_src_scope(int type
)
364 return (type
== IPV6_ADDR_ANY
) ? __IPV6_ADDR_SCOPE_INVALID
: (type
>> 16);
367 static inline int ipv6_addr_src_scope(const struct in6_addr
*addr
)
369 return __ipv6_addr_src_scope(__ipv6_addr_type(addr
));
372 static inline bool __ipv6_addr_needs_scope_id(int type
)
374 return type
& IPV6_ADDR_LINKLOCAL
||
375 (type
& IPV6_ADDR_MULTICAST
&&
376 (type
& (IPV6_ADDR_LOOPBACK
|IPV6_ADDR_LINKLOCAL
)));
379 static inline __u32
ipv6_iface_scope_id(const struct in6_addr
*addr
, int iface
)
381 return __ipv6_addr_needs_scope_id(__ipv6_addr_type(addr
)) ? iface
: 0;
384 static inline int ipv6_addr_cmp(const struct in6_addr
*a1
, const struct in6_addr
*a2
)
386 return memcmp(a1
, a2
, sizeof(struct in6_addr
));
390 ipv6_masked_addr_cmp(const struct in6_addr
*a1
, const struct in6_addr
*m
,
391 const struct in6_addr
*a2
)
393 #if defined(CONFIG_HAVE_EFFICIENT_UNALIGNED_ACCESS) && BITS_PER_LONG == 64
394 const unsigned long *ul1
= (const unsigned long *)a1
;
395 const unsigned long *ulm
= (const unsigned long *)m
;
396 const unsigned long *ul2
= (const unsigned long *)a2
;
398 return !!(((ul1
[0] ^ ul2
[0]) & ulm
[0]) |
399 ((ul1
[1] ^ ul2
[1]) & ulm
[1]));
401 return !!(((a1
->s6_addr32
[0] ^ a2
->s6_addr32
[0]) & m
->s6_addr32
[0]) |
402 ((a1
->s6_addr32
[1] ^ a2
->s6_addr32
[1]) & m
->s6_addr32
[1]) |
403 ((a1
->s6_addr32
[2] ^ a2
->s6_addr32
[2]) & m
->s6_addr32
[2]) |
404 ((a1
->s6_addr32
[3] ^ a2
->s6_addr32
[3]) & m
->s6_addr32
[3]));
408 static inline void ipv6_addr_prefix(struct in6_addr
*pfx
,
409 const struct in6_addr
*addr
,
412 /* caller must guarantee 0 <= plen <= 128 */
416 memset(pfx
->s6_addr
, 0, sizeof(pfx
->s6_addr
));
417 memcpy(pfx
->s6_addr
, addr
, o
);
419 pfx
->s6_addr
[o
] = addr
->s6_addr
[o
] & (0xff00 >> b
);
422 static inline void ipv6_addr_prefix_copy(struct in6_addr
*addr
,
423 const struct in6_addr
*pfx
,
426 /* caller must guarantee 0 <= plen <= 128 */
430 memcpy(addr
->s6_addr
, pfx
, o
);
432 addr
->s6_addr
[o
] &= ~(0xff00 >> b
);
433 addr
->s6_addr
[o
] |= (pfx
->s6_addr
[o
] & (0xff00 >> b
));
437 static inline void __ipv6_addr_set_half(__be32
*addr
,
438 __be32 wh
, __be32 wl
)
440 #if defined(CONFIG_HAVE_EFFICIENT_UNALIGNED_ACCESS) && BITS_PER_LONG == 64
441 #if defined(__BIG_ENDIAN)
442 if (__builtin_constant_p(wh
) && __builtin_constant_p(wl
)) {
443 *(__force u64
*)addr
= ((__force u64
)(wh
) << 32 | (__force u64
)(wl
));
446 #elif defined(__LITTLE_ENDIAN)
447 if (__builtin_constant_p(wl
) && __builtin_constant_p(wh
)) {
448 *(__force u64
*)addr
= ((__force u64
)(wl
) << 32 | (__force u64
)(wh
));
457 static inline void ipv6_addr_set(struct in6_addr
*addr
,
458 __be32 w1
, __be32 w2
,
459 __be32 w3
, __be32 w4
)
461 __ipv6_addr_set_half(&addr
->s6_addr32
[0], w1
, w2
);
462 __ipv6_addr_set_half(&addr
->s6_addr32
[2], w3
, w4
);
465 static inline bool ipv6_addr_equal(const struct in6_addr
*a1
,
466 const struct in6_addr
*a2
)
468 #if defined(CONFIG_HAVE_EFFICIENT_UNALIGNED_ACCESS) && BITS_PER_LONG == 64
469 const unsigned long *ul1
= (const unsigned long *)a1
;
470 const unsigned long *ul2
= (const unsigned long *)a2
;
472 return ((ul1
[0] ^ ul2
[0]) | (ul1
[1] ^ ul2
[1])) == 0UL;
474 return ((a1
->s6_addr32
[0] ^ a2
->s6_addr32
[0]) |
475 (a1
->s6_addr32
[1] ^ a2
->s6_addr32
[1]) |
476 (a1
->s6_addr32
[2] ^ a2
->s6_addr32
[2]) |
477 (a1
->s6_addr32
[3] ^ a2
->s6_addr32
[3])) == 0;
481 #if defined(CONFIG_HAVE_EFFICIENT_UNALIGNED_ACCESS) && BITS_PER_LONG == 64
482 static inline bool __ipv6_prefix_equal64_half(const __be64
*a1
,
486 if (len
&& ((*a1
^ *a2
) & cpu_to_be64((~0UL) << (64 - len
))))
491 static inline bool ipv6_prefix_equal(const struct in6_addr
*addr1
,
492 const struct in6_addr
*addr2
,
493 unsigned int prefixlen
)
495 const __be64
*a1
= (const __be64
*)addr1
;
496 const __be64
*a2
= (const __be64
*)addr2
;
498 if (prefixlen
>= 64) {
501 return __ipv6_prefix_equal64_half(a1
+ 1, a2
+ 1, prefixlen
- 64);
503 return __ipv6_prefix_equal64_half(a1
, a2
, prefixlen
);
506 static inline bool ipv6_prefix_equal(const struct in6_addr
*addr1
,
507 const struct in6_addr
*addr2
,
508 unsigned int prefixlen
)
510 const __be32
*a1
= addr1
->s6_addr32
;
511 const __be32
*a2
= addr2
->s6_addr32
;
512 unsigned int pdw
, pbi
;
514 /* check complete u32 in prefix */
515 pdw
= prefixlen
>> 5;
516 if (pdw
&& memcmp(a1
, a2
, pdw
<< 2))
519 /* check incomplete u32 in prefix */
520 pbi
= prefixlen
& 0x1f;
521 if (pbi
&& ((a1
[pdw
] ^ a2
[pdw
]) & htonl((0xffffffff) << (32 - pbi
))))
528 struct inet_frag_queue
;
530 enum ip6_defrag_users
{
531 IP6_DEFRAG_LOCAL_DELIVER
,
532 IP6_DEFRAG_CONNTRACK_IN
,
533 __IP6_DEFRAG_CONNTRACK_IN
= IP6_DEFRAG_CONNTRACK_IN
+ USHRT_MAX
,
534 IP6_DEFRAG_CONNTRACK_OUT
,
535 __IP6_DEFRAG_CONNTRACK_OUT
= IP6_DEFRAG_CONNTRACK_OUT
+ USHRT_MAX
,
536 IP6_DEFRAG_CONNTRACK_BRIDGE_IN
,
537 __IP6_DEFRAG_CONNTRACK_BRIDGE_IN
= IP6_DEFRAG_CONNTRACK_BRIDGE_IN
+ USHRT_MAX
,
540 struct ip6_create_arg
{
543 const struct in6_addr
*src
;
544 const struct in6_addr
*dst
;
549 void ip6_frag_init(struct inet_frag_queue
*q
, const void *a
);
550 bool ip6_frag_match(const struct inet_frag_queue
*q
, const void *a
);
553 * Equivalent of ipv4 struct ip
556 struct inet_frag_queue q
;
558 __be32 id
; /* fragment id */
560 struct in6_addr saddr
;
561 struct in6_addr daddr
;
569 void ip6_expire_frag_queue(struct net
*net
, struct frag_queue
*fq
,
570 struct inet_frags
*frags
);
572 static inline bool ipv6_addr_any(const struct in6_addr
*a
)
574 #if defined(CONFIG_HAVE_EFFICIENT_UNALIGNED_ACCESS) && BITS_PER_LONG == 64
575 const unsigned long *ul
= (const unsigned long *)a
;
577 return (ul
[0] | ul
[1]) == 0UL;
579 return (a
->s6_addr32
[0] | a
->s6_addr32
[1] |
580 a
->s6_addr32
[2] | a
->s6_addr32
[3]) == 0;
584 static inline u32
ipv6_addr_hash(const struct in6_addr
*a
)
586 #if defined(CONFIG_HAVE_EFFICIENT_UNALIGNED_ACCESS) && BITS_PER_LONG == 64
587 const unsigned long *ul
= (const unsigned long *)a
;
588 unsigned long x
= ul
[0] ^ ul
[1];
590 return (u32
)(x
^ (x
>> 32));
592 return (__force u32
)(a
->s6_addr32
[0] ^ a
->s6_addr32
[1] ^
593 a
->s6_addr32
[2] ^ a
->s6_addr32
[3]);
597 /* more secured version of ipv6_addr_hash() */
598 static inline u32
__ipv6_addr_jhash(const struct in6_addr
*a
, const u32 initval
)
600 u32 v
= (__force u32
)a
->s6_addr32
[0] ^ (__force u32
)a
->s6_addr32
[1];
602 return jhash_3words(v
,
603 (__force u32
)a
->s6_addr32
[2],
604 (__force u32
)a
->s6_addr32
[3],
608 static inline bool ipv6_addr_loopback(const struct in6_addr
*a
)
610 #if defined(CONFIG_HAVE_EFFICIENT_UNALIGNED_ACCESS) && BITS_PER_LONG == 64
611 const __be64
*be
= (const __be64
*)a
;
613 return (be
[0] | (be
[1] ^ cpu_to_be64(1))) == 0UL;
615 return (a
->s6_addr32
[0] | a
->s6_addr32
[1] |
616 a
->s6_addr32
[2] | (a
->s6_addr32
[3] ^ cpu_to_be32(1))) == 0;
621 * Note that we must __force cast these to unsigned long to make sparse happy,
622 * since all of the endian-annotated types are fixed size regardless of arch.
624 static inline bool ipv6_addr_v4mapped(const struct in6_addr
*a
)
627 #if defined(CONFIG_HAVE_EFFICIENT_UNALIGNED_ACCESS) && BITS_PER_LONG == 64
628 *(unsigned long *)a
|
630 (__force
unsigned long)(a
->s6_addr32
[0] | a
->s6_addr32
[1]) |
632 (__force
unsigned long)(a
->s6_addr32
[2] ^
633 cpu_to_be32(0x0000ffff))) == 0UL;
637 * Check for a RFC 4843 ORCHID address
638 * (Overlay Routable Cryptographic Hash Identifiers)
640 static inline bool ipv6_addr_orchid(const struct in6_addr
*a
)
642 return (a
->s6_addr32
[0] & htonl(0xfffffff0)) == htonl(0x20010010);
645 static inline bool ipv6_addr_is_multicast(const struct in6_addr
*addr
)
647 return (addr
->s6_addr32
[0] & htonl(0xFF000000)) == htonl(0xFF000000);
650 static inline void ipv6_addr_set_v4mapped(const __be32 addr
,
651 struct in6_addr
*v4mapped
)
653 ipv6_addr_set(v4mapped
,
660 * find the first different bit between two addresses
661 * length of address must be a multiple of 32bits
663 static inline int __ipv6_addr_diff32(const void *token1
, const void *token2
, int addrlen
)
665 const __be32
*a1
= token1
, *a2
= token2
;
670 for (i
= 0; i
< addrlen
; i
++) {
671 __be32 xb
= a1
[i
] ^ a2
[i
];
673 return i
* 32 + 31 - __fls(ntohl(xb
));
677 * we should *never* get to this point since that
678 * would mean the addrs are equal
680 * However, we do get to it 8) And exacly, when
681 * addresses are equal 8)
683 * ip route add 1111::/128 via ...
684 * ip route add 1111::/64 via ...
687 * Ideally, this function should stop comparison
688 * at prefix length. It does not, but it is still OK,
689 * if returned value is greater than prefix length.
695 #if defined(CONFIG_HAVE_EFFICIENT_UNALIGNED_ACCESS) && BITS_PER_LONG == 64
696 static inline int __ipv6_addr_diff64(const void *token1
, const void *token2
, int addrlen
)
698 const __be64
*a1
= token1
, *a2
= token2
;
703 for (i
= 0; i
< addrlen
; i
++) {
704 __be64 xb
= a1
[i
] ^ a2
[i
];
706 return i
* 64 + 63 - __fls(be64_to_cpu(xb
));
713 static inline int __ipv6_addr_diff(const void *token1
, const void *token2
, int addrlen
)
715 #if defined(CONFIG_HAVE_EFFICIENT_UNALIGNED_ACCESS) && BITS_PER_LONG == 64
716 if (__builtin_constant_p(addrlen
) && !(addrlen
& 7))
717 return __ipv6_addr_diff64(token1
, token2
, addrlen
);
719 return __ipv6_addr_diff32(token1
, token2
, addrlen
);
722 static inline int ipv6_addr_diff(const struct in6_addr
*a1
, const struct in6_addr
*a2
)
724 return __ipv6_addr_diff(a1
, a2
, sizeof(struct in6_addr
));
727 __be32
ipv6_select_ident(struct net
*net
,
728 const struct in6_addr
*daddr
,
729 const struct in6_addr
*saddr
);
730 void ipv6_proxy_select_ident(struct net
*net
, struct sk_buff
*skb
);
732 int ip6_dst_hoplimit(struct dst_entry
*dst
);
734 static inline int ip6_sk_dst_hoplimit(struct ipv6_pinfo
*np
, struct flowi6
*fl6
,
735 struct dst_entry
*dst
)
739 if (ipv6_addr_is_multicast(&fl6
->daddr
))
740 hlimit
= np
->mcast_hops
;
742 hlimit
= np
->hop_limit
;
744 hlimit
= ip6_dst_hoplimit(dst
);
748 /* copy IPv6 saddr & daddr to flow_keys, possibly using 64bit load/store
749 * Equivalent to : flow->v6addrs.src = iph->saddr;
750 * flow->v6addrs.dst = iph->daddr;
752 static inline void iph_to_flow_copy_v6addrs(struct flow_keys
*flow
,
753 const struct ipv6hdr
*iph
)
755 BUILD_BUG_ON(offsetof(typeof(flow
->addrs
), v6addrs
.dst
) !=
756 offsetof(typeof(flow
->addrs
), v6addrs
.src
) +
757 sizeof(flow
->addrs
.v6addrs
.src
));
758 memcpy(&flow
->addrs
.v6addrs
, &iph
->saddr
, sizeof(flow
->addrs
.v6addrs
));
759 flow
->control
.addr_type
= FLOW_DISSECTOR_KEY_IPV6_ADDRS
;
762 #if IS_ENABLED(CONFIG_IPV6)
764 /* Sysctl settings for net ipv6.auto_flowlabels */
765 #define IP6_AUTO_FLOW_LABEL_OFF 0
766 #define IP6_AUTO_FLOW_LABEL_OPTOUT 1
767 #define IP6_AUTO_FLOW_LABEL_OPTIN 2
768 #define IP6_AUTO_FLOW_LABEL_FORCED 3
770 #define IP6_AUTO_FLOW_LABEL_MAX IP6_AUTO_FLOW_LABEL_FORCED
772 #define IP6_DEFAULT_AUTO_FLOW_LABELS IP6_AUTO_FLOW_LABEL_OPTOUT
774 static inline __be32
ip6_make_flowlabel(struct net
*net
, struct sk_buff
*skb
,
775 __be32 flowlabel
, bool autolabel
,
780 /* @flowlabel may include more than a flow label, eg, the traffic class.
781 * Here we want only the flow label value.
783 flowlabel
&= IPV6_FLOWLABEL_MASK
;
786 net
->ipv6
.sysctl
.auto_flowlabels
== IP6_AUTO_FLOW_LABEL_OFF
||
788 net
->ipv6
.sysctl
.auto_flowlabels
!= IP6_AUTO_FLOW_LABEL_FORCED
))
791 hash
= skb_get_hash_flowi6(skb
, fl6
);
793 /* Since this is being sent on the wire obfuscate hash a bit
794 * to minimize possbility that any useful information to an
795 * attacker is leaked. Only lower 20 bits are relevant.
799 flowlabel
= (__force __be32
)hash
& IPV6_FLOWLABEL_MASK
;
801 if (net
->ipv6
.sysctl
.flowlabel_state_ranges
)
802 flowlabel
|= IPV6_FLOWLABEL_STATELESS_FLAG
;
807 static inline int ip6_default_np_autolabel(struct net
*net
)
809 switch (net
->ipv6
.sysctl
.auto_flowlabels
) {
810 case IP6_AUTO_FLOW_LABEL_OFF
:
811 case IP6_AUTO_FLOW_LABEL_OPTIN
:
814 case IP6_AUTO_FLOW_LABEL_OPTOUT
:
815 case IP6_AUTO_FLOW_LABEL_FORCED
:
820 static inline void ip6_set_txhash(struct sock
*sk
) { }
821 static inline __be32
ip6_make_flowlabel(struct net
*net
, struct sk_buff
*skb
,
822 __be32 flowlabel
, bool autolabel
,
827 static inline int ip6_default_np_autolabel(struct net
*net
)
835 * Header manipulation
837 static inline void ip6_flow_hdr(struct ipv6hdr
*hdr
, unsigned int tclass
,
840 *(__be32
*)hdr
= htonl(0x60000000 | (tclass
<< 20)) | flowlabel
;
843 static inline __be32
ip6_flowinfo(const struct ipv6hdr
*hdr
)
845 return *(__be32
*)hdr
& IPV6_FLOWINFO_MASK
;
848 static inline __be32
ip6_flowlabel(const struct ipv6hdr
*hdr
)
850 return *(__be32
*)hdr
& IPV6_FLOWLABEL_MASK
;
853 static inline u8
ip6_tclass(__be32 flowinfo
)
855 return ntohl(flowinfo
& IPV6_TCLASS_MASK
) >> IPV6_TCLASS_SHIFT
;
858 static inline __be32
ip6_make_flowinfo(unsigned int tclass
, __be32 flowlabel
)
860 return htonl(tclass
<< IPV6_TCLASS_SHIFT
) | flowlabel
;
864 * Prototypes exported by ipv6
868 * rcv function (called from netdevice level)
871 int ipv6_rcv(struct sk_buff
*skb
, struct net_device
*dev
,
872 struct packet_type
*pt
, struct net_device
*orig_dev
);
874 int ip6_rcv_finish(struct net
*net
, struct sock
*sk
, struct sk_buff
*skb
);
877 * upper-layer output functions
879 int ip6_xmit(const struct sock
*sk
, struct sk_buff
*skb
, struct flowi6
*fl6
,
880 __u32 mark
, struct ipv6_txoptions
*opt
, int tclass
);
882 int ip6_find_1stfragopt(struct sk_buff
*skb
, u8
**nexthdr
);
884 int ip6_append_data(struct sock
*sk
,
885 int getfrag(void *from
, char *to
, int offset
, int len
,
886 int odd
, struct sk_buff
*skb
),
887 void *from
, int length
, int transhdrlen
,
888 struct ipcm6_cookie
*ipc6
, struct flowi6
*fl6
,
889 struct rt6_info
*rt
, unsigned int flags
,
890 const struct sockcm_cookie
*sockc
);
892 int ip6_push_pending_frames(struct sock
*sk
);
894 void ip6_flush_pending_frames(struct sock
*sk
);
896 int ip6_send_skb(struct sk_buff
*skb
);
898 struct sk_buff
*__ip6_make_skb(struct sock
*sk
, struct sk_buff_head
*queue
,
899 struct inet_cork_full
*cork
,
900 struct inet6_cork
*v6_cork
);
901 struct sk_buff
*ip6_make_skb(struct sock
*sk
,
902 int getfrag(void *from
, char *to
, int offset
,
903 int len
, int odd
, struct sk_buff
*skb
),
904 void *from
, int length
, int transhdrlen
,
905 struct ipcm6_cookie
*ipc6
, struct flowi6
*fl6
,
906 struct rt6_info
*rt
, unsigned int flags
,
907 const struct sockcm_cookie
*sockc
);
909 static inline struct sk_buff
*ip6_finish_skb(struct sock
*sk
)
911 return __ip6_make_skb(sk
, &sk
->sk_write_queue
, &inet_sk(sk
)->cork
,
912 &inet6_sk(sk
)->cork
);
915 int ip6_dst_lookup(struct net
*net
, struct sock
*sk
, struct dst_entry
**dst
,
917 struct dst_entry
*ip6_dst_lookup_flow(const struct sock
*sk
, struct flowi6
*fl6
,
918 const struct in6_addr
*final_dst
);
919 struct dst_entry
*ip6_sk_dst_lookup_flow(struct sock
*sk
, struct flowi6
*fl6
,
920 const struct in6_addr
*final_dst
);
921 struct dst_entry
*ip6_blackhole_route(struct net
*net
,
922 struct dst_entry
*orig_dst
);
925 * skb processing functions
928 int ip6_output(struct net
*net
, struct sock
*sk
, struct sk_buff
*skb
);
929 int ip6_forward(struct sk_buff
*skb
);
930 int ip6_input(struct sk_buff
*skb
);
931 int ip6_mc_input(struct sk_buff
*skb
);
933 int __ip6_local_out(struct net
*net
, struct sock
*sk
, struct sk_buff
*skb
);
934 int ip6_local_out(struct net
*net
, struct sock
*sk
, struct sk_buff
*skb
);
937 * Extension header (options) processing
940 void ipv6_push_nfrag_opts(struct sk_buff
*skb
, struct ipv6_txoptions
*opt
,
941 u8
*proto
, struct in6_addr
**daddr_p
);
942 void ipv6_push_frag_opts(struct sk_buff
*skb
, struct ipv6_txoptions
*opt
,
945 int ipv6_skip_exthdr(const struct sk_buff
*, int start
, u8
*nexthdrp
,
948 bool ipv6_ext_hdr(u8 nexthdr
);
951 IP6_FH_F_FRAG
= (1 << 0),
952 IP6_FH_F_AUTH
= (1 << 1),
953 IP6_FH_F_SKIP_RH
= (1 << 2),
956 /* find specified header and get offset to it */
957 int ipv6_find_hdr(const struct sk_buff
*skb
, unsigned int *offset
, int target
,
958 unsigned short *fragoff
, int *fragflg
);
960 int ipv6_find_tlv(const struct sk_buff
*skb
, int offset
, int type
);
962 struct in6_addr
*fl6_update_dst(struct flowi6
*fl6
,
963 const struct ipv6_txoptions
*opt
,
964 struct in6_addr
*orig
);
967 * socket options (ipv6_sockglue.c)
970 int ipv6_setsockopt(struct sock
*sk
, int level
, int optname
,
971 char __user
*optval
, unsigned int optlen
);
972 int ipv6_getsockopt(struct sock
*sk
, int level
, int optname
,
973 char __user
*optval
, int __user
*optlen
);
974 int compat_ipv6_setsockopt(struct sock
*sk
, int level
, int optname
,
975 char __user
*optval
, unsigned int optlen
);
976 int compat_ipv6_getsockopt(struct sock
*sk
, int level
, int optname
,
977 char __user
*optval
, int __user
*optlen
);
979 int __ip6_datagram_connect(struct sock
*sk
, struct sockaddr
*addr
,
981 int ip6_datagram_connect(struct sock
*sk
, struct sockaddr
*addr
, int addr_len
);
982 int ip6_datagram_connect_v6_only(struct sock
*sk
, struct sockaddr
*addr
,
984 int ip6_datagram_dst_update(struct sock
*sk
, bool fix_sk_saddr
);
985 void ip6_datagram_release_cb(struct sock
*sk
);
987 int ipv6_recv_error(struct sock
*sk
, struct msghdr
*msg
, int len
,
989 int ipv6_recv_rxpmtu(struct sock
*sk
, struct msghdr
*msg
, int len
,
991 void ipv6_icmp_error(struct sock
*sk
, struct sk_buff
*skb
, int err
, __be16 port
,
992 u32 info
, u8
*payload
);
993 void ipv6_local_error(struct sock
*sk
, int err
, struct flowi6
*fl6
, u32 info
);
994 void ipv6_local_rxpmtu(struct sock
*sk
, struct flowi6
*fl6
, u32 mtu
);
996 int inet6_release(struct socket
*sock
);
997 int inet6_bind(struct socket
*sock
, struct sockaddr
*uaddr
, int addr_len
);
998 int inet6_getname(struct socket
*sock
, struct sockaddr
*uaddr
, int *uaddr_len
,
1000 int inet6_ioctl(struct socket
*sock
, unsigned int cmd
, unsigned long arg
);
1002 int inet6_hash_connect(struct inet_timewait_death_row
*death_row
,
1008 extern const struct proto_ops inet6_stream_ops
;
1009 extern const struct proto_ops inet6_dgram_ops
;
1010 extern const struct proto_ops inet6_sockraw_ops
;
1012 struct group_source_req
;
1013 struct group_filter
;
1015 int ip6_mc_source(int add
, int omode
, struct sock
*sk
,
1016 struct group_source_req
*pgsr
);
1017 int ip6_mc_msfilter(struct sock
*sk
, struct group_filter
*gsf
);
1018 int ip6_mc_msfget(struct sock
*sk
, struct group_filter
*gsf
,
1019 struct group_filter __user
*optval
, int __user
*optlen
);
1021 #ifdef CONFIG_PROC_FS
1022 int ac6_proc_init(struct net
*net
);
1023 void ac6_proc_exit(struct net
*net
);
1024 int raw6_proc_init(void);
1025 void raw6_proc_exit(void);
1026 int tcp6_proc_init(struct net
*net
);
1027 void tcp6_proc_exit(struct net
*net
);
1028 int udp6_proc_init(struct net
*net
);
1029 void udp6_proc_exit(struct net
*net
);
1030 int udplite6_proc_init(void);
1031 void udplite6_proc_exit(void);
1032 int ipv6_misc_proc_init(void);
1033 void ipv6_misc_proc_exit(void);
1034 int snmp6_register_dev(struct inet6_dev
*idev
);
1035 int snmp6_unregister_dev(struct inet6_dev
*idev
);
1038 static inline int ac6_proc_init(struct net
*net
) { return 0; }
1039 static inline void ac6_proc_exit(struct net
*net
) { }
1040 static inline int snmp6_register_dev(struct inet6_dev
*idev
) { return 0; }
1041 static inline int snmp6_unregister_dev(struct inet6_dev
*idev
) { return 0; }
1044 #ifdef CONFIG_SYSCTL
1045 extern struct ctl_table ipv6_route_table_template
[];
1047 struct ctl_table
*ipv6_icmp_sysctl_init(struct net
*net
);
1048 struct ctl_table
*ipv6_route_sysctl_init(struct net
*net
);
1049 int ipv6_sysctl_register(void);
1050 void ipv6_sysctl_unregister(void);
1053 int ipv6_sock_mc_join(struct sock
*sk
, int ifindex
,
1054 const struct in6_addr
*addr
);
1055 int ipv6_sock_mc_drop(struct sock
*sk
, int ifindex
,
1056 const struct in6_addr
*addr
);
1057 #endif /* _NET_IPV6_H */