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 <linux/refcount.h>
20 #include <net/if_inet6.h>
21 #include <net/ndisc.h>
23 #include <net/flow_dissector.h>
26 #define SIN6_LEN_RFC2133 24
28 #define IPV6_MAXPLEN 65535
31 * NextHeader field of IPv6 header
34 #define NEXTHDR_HOP 0 /* Hop-by-hop option header. */
35 #define NEXTHDR_TCP 6 /* TCP segment. */
36 #define NEXTHDR_UDP 17 /* UDP message. */
37 #define NEXTHDR_IPV6 41 /* IPv6 in IPv6 */
38 #define NEXTHDR_ROUTING 43 /* Routing header. */
39 #define NEXTHDR_FRAGMENT 44 /* Fragmentation/reassembly header. */
40 #define NEXTHDR_GRE 47 /* GRE header. */
41 #define NEXTHDR_ESP 50 /* Encapsulating security payload. */
42 #define NEXTHDR_AUTH 51 /* Authentication header. */
43 #define NEXTHDR_ICMP 58 /* ICMP for IPv6. */
44 #define NEXTHDR_NONE 59 /* No next header */
45 #define NEXTHDR_DEST 60 /* Destination options header. */
46 #define NEXTHDR_SCTP 132 /* SCTP message. */
47 #define NEXTHDR_MOBILITY 135 /* Mobility header. */
49 #define NEXTHDR_MAX 255
51 #define IPV6_DEFAULT_HOPLIMIT 64
52 #define IPV6_DEFAULT_MCASTHOPS 1
54 /* Limits on Hop-by-Hop and Destination options.
56 * Per RFC8200 there is no limit on the maximum number or lengths of options in
57 * Hop-by-Hop or Destination options other then the packet must fit in an MTU.
58 * We allow configurable limits in order to mitigate potential denial of
61 * There are three limits that may be set:
62 * - Limit the number of options in a Hop-by-Hop or Destination options
64 * - Limit the byte length of a Hop-by-Hop or Destination options extension
66 * - Disallow unknown options
68 * The limits are expressed in corresponding sysctls:
70 * ipv6.sysctl.max_dst_opts_cnt
71 * ipv6.sysctl.max_hbh_opts_cnt
72 * ipv6.sysctl.max_dst_opts_len
73 * ipv6.sysctl.max_hbh_opts_len
75 * max_*_opts_cnt is the number of TLVs that are allowed for Destination
76 * options or Hop-by-Hop options. If the number is less than zero then unknown
77 * TLVs are disallowed and the number of known options that are allowed is the
78 * absolute value. Setting the value to INT_MAX indicates no limit.
80 * max_*_opts_len is the length limit in bytes of a Destination or
81 * Hop-by-Hop options extension header. Setting the value to INT_MAX
82 * indicates no length limit.
84 * If a limit is exceeded when processing an extension header the packet is
88 /* Default limits for Hop-by-Hop and Destination options */
89 #define IP6_DEFAULT_MAX_DST_OPTS_CNT 8
90 #define IP6_DEFAULT_MAX_HBH_OPTS_CNT 8
91 #define IP6_DEFAULT_MAX_DST_OPTS_LEN INT_MAX /* No limit */
92 #define IP6_DEFAULT_MAX_HBH_OPTS_LEN INT_MAX /* No limit */
97 * type - unicast | multicast
98 * scope - local | site | global
105 #define IPV6_ADDR_ANY 0x0000U
107 #define IPV6_ADDR_UNICAST 0x0001U
108 #define IPV6_ADDR_MULTICAST 0x0002U
110 #define IPV6_ADDR_LOOPBACK 0x0010U
111 #define IPV6_ADDR_LINKLOCAL 0x0020U
112 #define IPV6_ADDR_SITELOCAL 0x0040U
114 #define IPV6_ADDR_COMPATv4 0x0080U
116 #define IPV6_ADDR_SCOPE_MASK 0x00f0U
118 #define IPV6_ADDR_MAPPED 0x1000U
123 #define IPV6_ADDR_MC_SCOPE(a) \
124 ((a)->s6_addr[1] & 0x0f) /* nonstandard */
125 #define __IPV6_ADDR_SCOPE_INVALID -1
126 #define IPV6_ADDR_SCOPE_NODELOCAL 0x01
127 #define IPV6_ADDR_SCOPE_LINKLOCAL 0x02
128 #define IPV6_ADDR_SCOPE_SITELOCAL 0x05
129 #define IPV6_ADDR_SCOPE_ORGLOCAL 0x08
130 #define IPV6_ADDR_SCOPE_GLOBAL 0x0e
135 #define IPV6_ADDR_MC_FLAG_TRANSIENT(a) \
136 ((a)->s6_addr[1] & 0x10)
137 #define IPV6_ADDR_MC_FLAG_PREFIX(a) \
138 ((a)->s6_addr[1] & 0x20)
139 #define IPV6_ADDR_MC_FLAG_RENDEZVOUS(a) \
140 ((a)->s6_addr[1] & 0x40)
143 * fragmentation header
150 __be32 identification
;
153 #define IP6_MF 0x0001
154 #define IP6_OFFSET 0xFFF8
156 #define IP6_REPLY_MARK(net, mark) \
157 ((net)->ipv6.sysctl.fwmark_reflect ? (mark) : 0)
159 #include <net/sock.h>
162 extern int sysctl_mld_max_msf
;
163 extern int sysctl_mld_qrv
;
165 #define _DEVINC(net, statname, mod, idev, field) \
167 struct inet6_dev *_idev = (idev); \
168 if (likely(_idev != NULL)) \
169 mod##SNMP_INC_STATS64((_idev)->stats.statname, (field));\
170 mod##SNMP_INC_STATS64((net)->mib.statname##_statistics, (field));\
173 /* per device counters are atomic_long_t */
174 #define _DEVINCATOMIC(net, statname, mod, idev, field) \
176 struct inet6_dev *_idev = (idev); \
177 if (likely(_idev != NULL)) \
178 SNMP_INC_STATS_ATOMIC_LONG((_idev)->stats.statname##dev, (field)); \
179 mod##SNMP_INC_STATS((net)->mib.statname##_statistics, (field));\
182 /* per device and per net counters are atomic_long_t */
183 #define _DEVINC_ATOMIC_ATOMIC(net, statname, idev, field) \
185 struct inet6_dev *_idev = (idev); \
186 if (likely(_idev != NULL)) \
187 SNMP_INC_STATS_ATOMIC_LONG((_idev)->stats.statname##dev, (field)); \
188 SNMP_INC_STATS_ATOMIC_LONG((net)->mib.statname##_statistics, (field));\
191 #define _DEVADD(net, statname, mod, idev, field, val) \
193 struct inet6_dev *_idev = (idev); \
194 if (likely(_idev != NULL)) \
195 mod##SNMP_ADD_STATS((_idev)->stats.statname, (field), (val)); \
196 mod##SNMP_ADD_STATS((net)->mib.statname##_statistics, (field), (val));\
199 #define _DEVUPD(net, statname, mod, idev, field, val) \
201 struct inet6_dev *_idev = (idev); \
202 if (likely(_idev != NULL)) \
203 mod##SNMP_UPD_PO_STATS((_idev)->stats.statname, field, (val)); \
204 mod##SNMP_UPD_PO_STATS((net)->mib.statname##_statistics, field, (val));\
209 #define IP6_INC_STATS(net, idev,field) \
210 _DEVINC(net, ipv6, , idev, field)
211 #define __IP6_INC_STATS(net, idev,field) \
212 _DEVINC(net, ipv6, __, idev, field)
213 #define IP6_ADD_STATS(net, idev,field,val) \
214 _DEVADD(net, ipv6, , idev, field, val)
215 #define __IP6_ADD_STATS(net, idev,field,val) \
216 _DEVADD(net, ipv6, __, idev, field, val)
217 #define IP6_UPD_PO_STATS(net, idev,field,val) \
218 _DEVUPD(net, ipv6, , idev, field, val)
219 #define __IP6_UPD_PO_STATS(net, idev,field,val) \
220 _DEVUPD(net, ipv6, __, idev, field, val)
221 #define ICMP6_INC_STATS(net, idev, field) \
222 _DEVINCATOMIC(net, icmpv6, , idev, field)
223 #define __ICMP6_INC_STATS(net, idev, field) \
224 _DEVINCATOMIC(net, icmpv6, __, idev, field)
226 #define ICMP6MSGOUT_INC_STATS(net, idev, field) \
227 _DEVINC_ATOMIC_ATOMIC(net, icmpv6msg, idev, field +256)
228 #define ICMP6MSGIN_INC_STATS(net, idev, field) \
229 _DEVINC_ATOMIC_ATOMIC(net, icmpv6msg, idev, field)
231 struct ip6_ra_chain
{
232 struct ip6_ra_chain
*next
;
235 void (*destructor
)(struct sock
*);
238 extern struct ip6_ra_chain
*ip6_ra_chain
;
239 extern rwlock_t ip6_ra_lock
;
242 This structure is prepared by protocol, when parsing
243 ancillary data and passed to IPv6.
246 struct ipv6_txoptions
{
248 /* Length of this structure */
251 /* length of extension headers */
253 __u16 opt_flen
; /* after fragment hdr */
254 __u16 opt_nflen
; /* before fragment hdr */
256 struct ipv6_opt_hdr
*hopopt
;
257 struct ipv6_opt_hdr
*dst0opt
;
258 struct ipv6_rt_hdr
*srcrt
; /* Routing Header */
259 struct ipv6_opt_hdr
*dst1opt
;
261 /* Option buffer, as read by IPV6_PKTOPTIONS, starts here. */
264 struct ip6_flowlabel
{
265 struct ip6_flowlabel __rcu
*next
;
269 struct ipv6_txoptions
*opt
;
270 unsigned long linger
;
277 unsigned long lastuse
;
278 unsigned long expires
;
282 #define IPV6_FLOWINFO_MASK cpu_to_be32(0x0FFFFFFF)
283 #define IPV6_FLOWLABEL_MASK cpu_to_be32(0x000FFFFF)
284 #define IPV6_FLOWLABEL_STATELESS_FLAG cpu_to_be32(0x00080000)
286 #define IPV6_TCLASS_MASK (IPV6_FLOWINFO_MASK & ~IPV6_FLOWLABEL_MASK)
287 #define IPV6_TCLASS_SHIFT 20
289 struct ipv6_fl_socklist
{
290 struct ipv6_fl_socklist __rcu
*next
;
291 struct ip6_flowlabel
*fl
;
295 struct ipcm6_cookie
{
299 struct ipv6_txoptions
*opt
;
302 static inline struct ipv6_txoptions
*txopt_get(const struct ipv6_pinfo
*np
)
304 struct ipv6_txoptions
*opt
;
307 opt
= rcu_dereference(np
->opt
);
309 if (!refcount_inc_not_zero(&opt
->refcnt
))
312 opt
= rcu_pointer_handoff(opt
);
318 static inline void txopt_put(struct ipv6_txoptions
*opt
)
320 if (opt
&& refcount_dec_and_test(&opt
->refcnt
))
324 struct ip6_flowlabel
*fl6_sock_lookup(struct sock
*sk
, __be32 label
);
325 struct ipv6_txoptions
*fl6_merge_options(struct ipv6_txoptions
*opt_space
,
326 struct ip6_flowlabel
*fl
,
327 struct ipv6_txoptions
*fopt
);
328 void fl6_free_socklist(struct sock
*sk
);
329 int ipv6_flowlabel_opt(struct sock
*sk
, char __user
*optval
, int optlen
);
330 int ipv6_flowlabel_opt_get(struct sock
*sk
, struct in6_flowlabel_req
*freq
,
332 int ip6_flowlabel_init(void);
333 void ip6_flowlabel_cleanup(void);
334 bool ip6_autoflowlabel(struct net
*net
, const struct ipv6_pinfo
*np
);
336 static inline void fl6_sock_release(struct ip6_flowlabel
*fl
)
339 atomic_dec(&fl
->users
);
342 void icmpv6_notify(struct sk_buff
*skb
, u8 type
, u8 code
, __be32 info
);
344 void icmpv6_push_pending_frames(struct sock
*sk
, struct flowi6
*fl6
,
345 struct icmp6hdr
*thdr
, int len
);
347 int ip6_ra_control(struct sock
*sk
, int sel
);
349 int ipv6_parse_hopopts(struct sk_buff
*skb
);
351 struct ipv6_txoptions
*ipv6_dup_options(struct sock
*sk
,
352 struct ipv6_txoptions
*opt
);
353 struct ipv6_txoptions
*ipv6_renew_options(struct sock
*sk
,
354 struct ipv6_txoptions
*opt
,
356 struct ipv6_opt_hdr __user
*newopt
,
358 struct ipv6_txoptions
*
359 ipv6_renew_options_kern(struct sock
*sk
,
360 struct ipv6_txoptions
*opt
,
362 struct ipv6_opt_hdr
*newopt
,
364 struct ipv6_txoptions
*ipv6_fixup_options(struct ipv6_txoptions
*opt_space
,
365 struct ipv6_txoptions
*opt
);
367 bool ipv6_opt_accepted(const struct sock
*sk
, const struct sk_buff
*skb
,
368 const struct inet6_skb_parm
*opt
);
369 struct ipv6_txoptions
*ipv6_update_options(struct sock
*sk
,
370 struct ipv6_txoptions
*opt
);
372 static inline bool ipv6_accept_ra(struct inet6_dev
*idev
)
374 /* If forwarding is enabled, RA are not accepted unless the special
375 * hybrid mode (accept_ra=2) is enabled.
377 return idev
->cnf
.forwarding
? idev
->cnf
.accept_ra
== 2 :
381 #if IS_ENABLED(CONFIG_IPV6)
382 static inline int ip6_frag_mem(struct net
*net
)
384 return sum_frag_mem_limit(&net
->ipv6
.frags
);
388 #define IPV6_FRAG_HIGH_THRESH (4 * 1024*1024) /* 4194304 */
389 #define IPV6_FRAG_LOW_THRESH (3 * 1024*1024) /* 3145728 */
390 #define IPV6_FRAG_TIMEOUT (60 * HZ) /* 60 seconds */
392 int __ipv6_addr_type(const struct in6_addr
*addr
);
393 static inline int ipv6_addr_type(const struct in6_addr
*addr
)
395 return __ipv6_addr_type(addr
) & 0xffff;
398 static inline int ipv6_addr_scope(const struct in6_addr
*addr
)
400 return __ipv6_addr_type(addr
) & IPV6_ADDR_SCOPE_MASK
;
403 static inline int __ipv6_addr_src_scope(int type
)
405 return (type
== IPV6_ADDR_ANY
) ? __IPV6_ADDR_SCOPE_INVALID
: (type
>> 16);
408 static inline int ipv6_addr_src_scope(const struct in6_addr
*addr
)
410 return __ipv6_addr_src_scope(__ipv6_addr_type(addr
));
413 static inline bool __ipv6_addr_needs_scope_id(int type
)
415 return type
& IPV6_ADDR_LINKLOCAL
||
416 (type
& IPV6_ADDR_MULTICAST
&&
417 (type
& (IPV6_ADDR_LOOPBACK
|IPV6_ADDR_LINKLOCAL
)));
420 static inline __u32
ipv6_iface_scope_id(const struct in6_addr
*addr
, int iface
)
422 return __ipv6_addr_needs_scope_id(__ipv6_addr_type(addr
)) ? iface
: 0;
425 static inline int ipv6_addr_cmp(const struct in6_addr
*a1
, const struct in6_addr
*a2
)
427 return memcmp(a1
, a2
, sizeof(struct in6_addr
));
431 ipv6_masked_addr_cmp(const struct in6_addr
*a1
, const struct in6_addr
*m
,
432 const struct in6_addr
*a2
)
434 #if defined(CONFIG_HAVE_EFFICIENT_UNALIGNED_ACCESS) && BITS_PER_LONG == 64
435 const unsigned long *ul1
= (const unsigned long *)a1
;
436 const unsigned long *ulm
= (const unsigned long *)m
;
437 const unsigned long *ul2
= (const unsigned long *)a2
;
439 return !!(((ul1
[0] ^ ul2
[0]) & ulm
[0]) |
440 ((ul1
[1] ^ ul2
[1]) & ulm
[1]));
442 return !!(((a1
->s6_addr32
[0] ^ a2
->s6_addr32
[0]) & m
->s6_addr32
[0]) |
443 ((a1
->s6_addr32
[1] ^ a2
->s6_addr32
[1]) & m
->s6_addr32
[1]) |
444 ((a1
->s6_addr32
[2] ^ a2
->s6_addr32
[2]) & m
->s6_addr32
[2]) |
445 ((a1
->s6_addr32
[3] ^ a2
->s6_addr32
[3]) & m
->s6_addr32
[3]));
449 static inline void ipv6_addr_prefix(struct in6_addr
*pfx
,
450 const struct in6_addr
*addr
,
453 /* caller must guarantee 0 <= plen <= 128 */
457 memset(pfx
->s6_addr
, 0, sizeof(pfx
->s6_addr
));
458 memcpy(pfx
->s6_addr
, addr
, o
);
460 pfx
->s6_addr
[o
] = addr
->s6_addr
[o
] & (0xff00 >> b
);
463 static inline void ipv6_addr_prefix_copy(struct in6_addr
*addr
,
464 const struct in6_addr
*pfx
,
467 /* caller must guarantee 0 <= plen <= 128 */
471 memcpy(addr
->s6_addr
, pfx
, o
);
473 addr
->s6_addr
[o
] &= ~(0xff00 >> b
);
474 addr
->s6_addr
[o
] |= (pfx
->s6_addr
[o
] & (0xff00 >> b
));
478 static inline void __ipv6_addr_set_half(__be32
*addr
,
479 __be32 wh
, __be32 wl
)
481 #if defined(CONFIG_HAVE_EFFICIENT_UNALIGNED_ACCESS) && BITS_PER_LONG == 64
482 #if defined(__BIG_ENDIAN)
483 if (__builtin_constant_p(wh
) && __builtin_constant_p(wl
)) {
484 *(__force u64
*)addr
= ((__force u64
)(wh
) << 32 | (__force u64
)(wl
));
487 #elif defined(__LITTLE_ENDIAN)
488 if (__builtin_constant_p(wl
) && __builtin_constant_p(wh
)) {
489 *(__force u64
*)addr
= ((__force u64
)(wl
) << 32 | (__force u64
)(wh
));
498 static inline void ipv6_addr_set(struct in6_addr
*addr
,
499 __be32 w1
, __be32 w2
,
500 __be32 w3
, __be32 w4
)
502 __ipv6_addr_set_half(&addr
->s6_addr32
[0], w1
, w2
);
503 __ipv6_addr_set_half(&addr
->s6_addr32
[2], w3
, w4
);
506 static inline bool ipv6_addr_equal(const struct in6_addr
*a1
,
507 const struct in6_addr
*a2
)
509 #if defined(CONFIG_HAVE_EFFICIENT_UNALIGNED_ACCESS) && BITS_PER_LONG == 64
510 const unsigned long *ul1
= (const unsigned long *)a1
;
511 const unsigned long *ul2
= (const unsigned long *)a2
;
513 return ((ul1
[0] ^ ul2
[0]) | (ul1
[1] ^ ul2
[1])) == 0UL;
515 return ((a1
->s6_addr32
[0] ^ a2
->s6_addr32
[0]) |
516 (a1
->s6_addr32
[1] ^ a2
->s6_addr32
[1]) |
517 (a1
->s6_addr32
[2] ^ a2
->s6_addr32
[2]) |
518 (a1
->s6_addr32
[3] ^ a2
->s6_addr32
[3])) == 0;
522 #if defined(CONFIG_HAVE_EFFICIENT_UNALIGNED_ACCESS) && BITS_PER_LONG == 64
523 static inline bool __ipv6_prefix_equal64_half(const __be64
*a1
,
527 if (len
&& ((*a1
^ *a2
) & cpu_to_be64((~0UL) << (64 - len
))))
532 static inline bool ipv6_prefix_equal(const struct in6_addr
*addr1
,
533 const struct in6_addr
*addr2
,
534 unsigned int prefixlen
)
536 const __be64
*a1
= (const __be64
*)addr1
;
537 const __be64
*a2
= (const __be64
*)addr2
;
539 if (prefixlen
>= 64) {
542 return __ipv6_prefix_equal64_half(a1
+ 1, a2
+ 1, prefixlen
- 64);
544 return __ipv6_prefix_equal64_half(a1
, a2
, prefixlen
);
547 static inline bool ipv6_prefix_equal(const struct in6_addr
*addr1
,
548 const struct in6_addr
*addr2
,
549 unsigned int prefixlen
)
551 const __be32
*a1
= addr1
->s6_addr32
;
552 const __be32
*a2
= addr2
->s6_addr32
;
553 unsigned int pdw
, pbi
;
555 /* check complete u32 in prefix */
556 pdw
= prefixlen
>> 5;
557 if (pdw
&& memcmp(a1
, a2
, pdw
<< 2))
560 /* check incomplete u32 in prefix */
561 pbi
= prefixlen
& 0x1f;
562 if (pbi
&& ((a1
[pdw
] ^ a2
[pdw
]) & htonl((0xffffffff) << (32 - pbi
))))
569 struct inet_frag_queue
;
571 enum ip6_defrag_users
{
572 IP6_DEFRAG_LOCAL_DELIVER
,
573 IP6_DEFRAG_CONNTRACK_IN
,
574 __IP6_DEFRAG_CONNTRACK_IN
= IP6_DEFRAG_CONNTRACK_IN
+ USHRT_MAX
,
575 IP6_DEFRAG_CONNTRACK_OUT
,
576 __IP6_DEFRAG_CONNTRACK_OUT
= IP6_DEFRAG_CONNTRACK_OUT
+ USHRT_MAX
,
577 IP6_DEFRAG_CONNTRACK_BRIDGE_IN
,
578 __IP6_DEFRAG_CONNTRACK_BRIDGE_IN
= IP6_DEFRAG_CONNTRACK_BRIDGE_IN
+ USHRT_MAX
,
581 struct ip6_create_arg
{
584 const struct in6_addr
*src
;
585 const struct in6_addr
*dst
;
590 void ip6_frag_init(struct inet_frag_queue
*q
, const void *a
);
591 bool ip6_frag_match(const struct inet_frag_queue
*q
, const void *a
);
594 * Equivalent of ipv4 struct ip
597 struct inet_frag_queue q
;
599 __be32 id
; /* fragment id */
601 struct in6_addr saddr
;
602 struct in6_addr daddr
;
610 void ip6_expire_frag_queue(struct net
*net
, struct frag_queue
*fq
,
611 struct inet_frags
*frags
);
613 static inline bool ipv6_addr_any(const struct in6_addr
*a
)
615 #if defined(CONFIG_HAVE_EFFICIENT_UNALIGNED_ACCESS) && BITS_PER_LONG == 64
616 const unsigned long *ul
= (const unsigned long *)a
;
618 return (ul
[0] | ul
[1]) == 0UL;
620 return (a
->s6_addr32
[0] | a
->s6_addr32
[1] |
621 a
->s6_addr32
[2] | a
->s6_addr32
[3]) == 0;
625 static inline u32
ipv6_addr_hash(const struct in6_addr
*a
)
627 #if defined(CONFIG_HAVE_EFFICIENT_UNALIGNED_ACCESS) && BITS_PER_LONG == 64
628 const unsigned long *ul
= (const unsigned long *)a
;
629 unsigned long x
= ul
[0] ^ ul
[1];
631 return (u32
)(x
^ (x
>> 32));
633 return (__force u32
)(a
->s6_addr32
[0] ^ a
->s6_addr32
[1] ^
634 a
->s6_addr32
[2] ^ a
->s6_addr32
[3]);
638 /* more secured version of ipv6_addr_hash() */
639 static inline u32
__ipv6_addr_jhash(const struct in6_addr
*a
, const u32 initval
)
641 u32 v
= (__force u32
)a
->s6_addr32
[0] ^ (__force u32
)a
->s6_addr32
[1];
643 return jhash_3words(v
,
644 (__force u32
)a
->s6_addr32
[2],
645 (__force u32
)a
->s6_addr32
[3],
649 static inline bool ipv6_addr_loopback(const struct in6_addr
*a
)
651 #if defined(CONFIG_HAVE_EFFICIENT_UNALIGNED_ACCESS) && BITS_PER_LONG == 64
652 const __be64
*be
= (const __be64
*)a
;
654 return (be
[0] | (be
[1] ^ cpu_to_be64(1))) == 0UL;
656 return (a
->s6_addr32
[0] | a
->s6_addr32
[1] |
657 a
->s6_addr32
[2] | (a
->s6_addr32
[3] ^ cpu_to_be32(1))) == 0;
662 * Note that we must __force cast these to unsigned long to make sparse happy,
663 * since all of the endian-annotated types are fixed size regardless of arch.
665 static inline bool ipv6_addr_v4mapped(const struct in6_addr
*a
)
668 #if defined(CONFIG_HAVE_EFFICIENT_UNALIGNED_ACCESS) && BITS_PER_LONG == 64
669 *(unsigned long *)a
|
671 (__force
unsigned long)(a
->s6_addr32
[0] | a
->s6_addr32
[1]) |
673 (__force
unsigned long)(a
->s6_addr32
[2] ^
674 cpu_to_be32(0x0000ffff))) == 0UL;
678 * Check for a RFC 4843 ORCHID address
679 * (Overlay Routable Cryptographic Hash Identifiers)
681 static inline bool ipv6_addr_orchid(const struct in6_addr
*a
)
683 return (a
->s6_addr32
[0] & htonl(0xfffffff0)) == htonl(0x20010010);
686 static inline bool ipv6_addr_is_multicast(const struct in6_addr
*addr
)
688 return (addr
->s6_addr32
[0] & htonl(0xFF000000)) == htonl(0xFF000000);
691 static inline void ipv6_addr_set_v4mapped(const __be32 addr
,
692 struct in6_addr
*v4mapped
)
694 ipv6_addr_set(v4mapped
,
701 * find the first different bit between two addresses
702 * length of address must be a multiple of 32bits
704 static inline int __ipv6_addr_diff32(const void *token1
, const void *token2
, int addrlen
)
706 const __be32
*a1
= token1
, *a2
= token2
;
711 for (i
= 0; i
< addrlen
; i
++) {
712 __be32 xb
= a1
[i
] ^ a2
[i
];
714 return i
* 32 + 31 - __fls(ntohl(xb
));
718 * we should *never* get to this point since that
719 * would mean the addrs are equal
721 * However, we do get to it 8) And exacly, when
722 * addresses are equal 8)
724 * ip route add 1111::/128 via ...
725 * ip route add 1111::/64 via ...
728 * Ideally, this function should stop comparison
729 * at prefix length. It does not, but it is still OK,
730 * if returned value is greater than prefix length.
736 #if defined(CONFIG_HAVE_EFFICIENT_UNALIGNED_ACCESS) && BITS_PER_LONG == 64
737 static inline int __ipv6_addr_diff64(const void *token1
, const void *token2
, int addrlen
)
739 const __be64
*a1
= token1
, *a2
= token2
;
744 for (i
= 0; i
< addrlen
; i
++) {
745 __be64 xb
= a1
[i
] ^ a2
[i
];
747 return i
* 64 + 63 - __fls(be64_to_cpu(xb
));
754 static inline int __ipv6_addr_diff(const void *token1
, const void *token2
, int addrlen
)
756 #if defined(CONFIG_HAVE_EFFICIENT_UNALIGNED_ACCESS) && BITS_PER_LONG == 64
757 if (__builtin_constant_p(addrlen
) && !(addrlen
& 7))
758 return __ipv6_addr_diff64(token1
, token2
, addrlen
);
760 return __ipv6_addr_diff32(token1
, token2
, addrlen
);
763 static inline int ipv6_addr_diff(const struct in6_addr
*a1
, const struct in6_addr
*a2
)
765 return __ipv6_addr_diff(a1
, a2
, sizeof(struct in6_addr
));
768 __be32
ipv6_select_ident(struct net
*net
,
769 const struct in6_addr
*daddr
,
770 const struct in6_addr
*saddr
);
771 __be32
ipv6_proxy_select_ident(struct net
*net
, struct sk_buff
*skb
);
773 int ip6_dst_hoplimit(struct dst_entry
*dst
);
775 static inline int ip6_sk_dst_hoplimit(struct ipv6_pinfo
*np
, struct flowi6
*fl6
,
776 struct dst_entry
*dst
)
780 if (ipv6_addr_is_multicast(&fl6
->daddr
))
781 hlimit
= np
->mcast_hops
;
783 hlimit
= np
->hop_limit
;
785 hlimit
= ip6_dst_hoplimit(dst
);
789 /* copy IPv6 saddr & daddr to flow_keys, possibly using 64bit load/store
790 * Equivalent to : flow->v6addrs.src = iph->saddr;
791 * flow->v6addrs.dst = iph->daddr;
793 static inline void iph_to_flow_copy_v6addrs(struct flow_keys
*flow
,
794 const struct ipv6hdr
*iph
)
796 BUILD_BUG_ON(offsetof(typeof(flow
->addrs
), v6addrs
.dst
) !=
797 offsetof(typeof(flow
->addrs
), v6addrs
.src
) +
798 sizeof(flow
->addrs
.v6addrs
.src
));
799 memcpy(&flow
->addrs
.v6addrs
, &iph
->saddr
, sizeof(flow
->addrs
.v6addrs
));
800 flow
->control
.addr_type
= FLOW_DISSECTOR_KEY_IPV6_ADDRS
;
803 #if IS_ENABLED(CONFIG_IPV6)
805 /* Sysctl settings for net ipv6.auto_flowlabels */
806 #define IP6_AUTO_FLOW_LABEL_OFF 0
807 #define IP6_AUTO_FLOW_LABEL_OPTOUT 1
808 #define IP6_AUTO_FLOW_LABEL_OPTIN 2
809 #define IP6_AUTO_FLOW_LABEL_FORCED 3
811 #define IP6_AUTO_FLOW_LABEL_MAX IP6_AUTO_FLOW_LABEL_FORCED
813 #define IP6_DEFAULT_AUTO_FLOW_LABELS IP6_AUTO_FLOW_LABEL_OPTOUT
815 static inline __be32
ip6_make_flowlabel(struct net
*net
, struct sk_buff
*skb
,
816 __be32 flowlabel
, bool autolabel
,
821 /* @flowlabel may include more than a flow label, eg, the traffic class.
822 * Here we want only the flow label value.
824 flowlabel
&= IPV6_FLOWLABEL_MASK
;
827 net
->ipv6
.sysctl
.auto_flowlabels
== IP6_AUTO_FLOW_LABEL_OFF
||
829 net
->ipv6
.sysctl
.auto_flowlabels
!= IP6_AUTO_FLOW_LABEL_FORCED
))
832 hash
= skb_get_hash_flowi6(skb
, fl6
);
834 /* Since this is being sent on the wire obfuscate hash a bit
835 * to minimize possbility that any useful information to an
836 * attacker is leaked. Only lower 20 bits are relevant.
840 flowlabel
= (__force __be32
)hash
& IPV6_FLOWLABEL_MASK
;
842 if (net
->ipv6
.sysctl
.flowlabel_state_ranges
)
843 flowlabel
|= IPV6_FLOWLABEL_STATELESS_FLAG
;
848 static inline int ip6_default_np_autolabel(struct net
*net
)
850 switch (net
->ipv6
.sysctl
.auto_flowlabels
) {
851 case IP6_AUTO_FLOW_LABEL_OFF
:
852 case IP6_AUTO_FLOW_LABEL_OPTIN
:
855 case IP6_AUTO_FLOW_LABEL_OPTOUT
:
856 case IP6_AUTO_FLOW_LABEL_FORCED
:
861 static inline void ip6_set_txhash(struct sock
*sk
) { }
862 static inline __be32
ip6_make_flowlabel(struct net
*net
, struct sk_buff
*skb
,
863 __be32 flowlabel
, bool autolabel
,
868 static inline int ip6_default_np_autolabel(struct net
*net
)
876 * Header manipulation
878 static inline void ip6_flow_hdr(struct ipv6hdr
*hdr
, unsigned int tclass
,
881 *(__be32
*)hdr
= htonl(0x60000000 | (tclass
<< 20)) | flowlabel
;
884 static inline __be32
ip6_flowinfo(const struct ipv6hdr
*hdr
)
886 return *(__be32
*)hdr
& IPV6_FLOWINFO_MASK
;
889 static inline __be32
ip6_flowlabel(const struct ipv6hdr
*hdr
)
891 return *(__be32
*)hdr
& IPV6_FLOWLABEL_MASK
;
894 static inline u8
ip6_tclass(__be32 flowinfo
)
896 return ntohl(flowinfo
& IPV6_TCLASS_MASK
) >> IPV6_TCLASS_SHIFT
;
899 static inline __be32
ip6_make_flowinfo(unsigned int tclass
, __be32 flowlabel
)
901 return htonl(tclass
<< IPV6_TCLASS_SHIFT
) | flowlabel
;
905 * Prototypes exported by ipv6
909 * rcv function (called from netdevice level)
912 int ipv6_rcv(struct sk_buff
*skb
, struct net_device
*dev
,
913 struct packet_type
*pt
, struct net_device
*orig_dev
);
915 int ip6_rcv_finish(struct net
*net
, struct sock
*sk
, struct sk_buff
*skb
);
918 * upper-layer output functions
920 int ip6_xmit(const struct sock
*sk
, struct sk_buff
*skb
, struct flowi6
*fl6
,
921 __u32 mark
, struct ipv6_txoptions
*opt
, int tclass
);
923 int ip6_find_1stfragopt(struct sk_buff
*skb
, u8
**nexthdr
);
925 int ip6_append_data(struct sock
*sk
,
926 int getfrag(void *from
, char *to
, int offset
, int len
,
927 int odd
, struct sk_buff
*skb
),
928 void *from
, int length
, int transhdrlen
,
929 struct ipcm6_cookie
*ipc6
, struct flowi6
*fl6
,
930 struct rt6_info
*rt
, unsigned int flags
,
931 const struct sockcm_cookie
*sockc
);
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 const struct sockcm_cookie
*sockc
);
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
);
962 struct dst_entry
*ip6_blackhole_route(struct net
*net
,
963 struct dst_entry
*orig_dst
);
966 * skb processing functions
969 int ip6_output(struct net
*net
, struct sock
*sk
, struct sk_buff
*skb
);
970 int ip6_forward(struct sk_buff
*skb
);
971 int ip6_input(struct sk_buff
*skb
);
972 int ip6_mc_input(struct sk_buff
*skb
);
974 int __ip6_local_out(struct net
*net
, struct sock
*sk
, struct sk_buff
*skb
);
975 int ip6_local_out(struct net
*net
, struct sock
*sk
, struct sk_buff
*skb
);
978 * Extension header (options) processing
981 void ipv6_push_nfrag_opts(struct sk_buff
*skb
, struct ipv6_txoptions
*opt
,
982 u8
*proto
, struct in6_addr
**daddr_p
,
983 struct in6_addr
*saddr
);
984 void ipv6_push_frag_opts(struct sk_buff
*skb
, struct ipv6_txoptions
*opt
,
987 int ipv6_skip_exthdr(const struct sk_buff
*, int start
, u8
*nexthdrp
,
990 bool ipv6_ext_hdr(u8 nexthdr
);
993 IP6_FH_F_FRAG
= (1 << 0),
994 IP6_FH_F_AUTH
= (1 << 1),
995 IP6_FH_F_SKIP_RH
= (1 << 2),
998 /* find specified header and get offset to it */
999 int ipv6_find_hdr(const struct sk_buff
*skb
, unsigned int *offset
, int target
,
1000 unsigned short *fragoff
, int *fragflg
);
1002 int ipv6_find_tlv(const struct sk_buff
*skb
, int offset
, int type
);
1004 struct in6_addr
*fl6_update_dst(struct flowi6
*fl6
,
1005 const struct ipv6_txoptions
*opt
,
1006 struct in6_addr
*orig
);
1009 * socket options (ipv6_sockglue.c)
1012 int ipv6_setsockopt(struct sock
*sk
, int level
, int optname
,
1013 char __user
*optval
, unsigned int optlen
);
1014 int ipv6_getsockopt(struct sock
*sk
, int level
, int optname
,
1015 char __user
*optval
, int __user
*optlen
);
1016 int compat_ipv6_setsockopt(struct sock
*sk
, int level
, int optname
,
1017 char __user
*optval
, unsigned int optlen
);
1018 int compat_ipv6_getsockopt(struct sock
*sk
, int level
, int optname
,
1019 char __user
*optval
, int __user
*optlen
);
1021 int __ip6_datagram_connect(struct sock
*sk
, struct sockaddr
*addr
,
1023 int ip6_datagram_connect(struct sock
*sk
, struct sockaddr
*addr
, int addr_len
);
1024 int ip6_datagram_connect_v6_only(struct sock
*sk
, struct sockaddr
*addr
,
1026 int ip6_datagram_dst_update(struct sock
*sk
, bool fix_sk_saddr
);
1027 void ip6_datagram_release_cb(struct sock
*sk
);
1029 int ipv6_recv_error(struct sock
*sk
, struct msghdr
*msg
, int len
,
1031 int ipv6_recv_rxpmtu(struct sock
*sk
, struct msghdr
*msg
, int len
,
1033 void ipv6_icmp_error(struct sock
*sk
, struct sk_buff
*skb
, int err
, __be16 port
,
1034 u32 info
, u8
*payload
);
1035 void ipv6_local_error(struct sock
*sk
, int err
, struct flowi6
*fl6
, u32 info
);
1036 void ipv6_local_rxpmtu(struct sock
*sk
, struct flowi6
*fl6
, u32 mtu
);
1038 int inet6_release(struct socket
*sock
);
1039 int inet6_bind(struct socket
*sock
, struct sockaddr
*uaddr
, int addr_len
);
1040 int inet6_getname(struct socket
*sock
, struct sockaddr
*uaddr
, int *uaddr_len
,
1042 int inet6_ioctl(struct socket
*sock
, unsigned int cmd
, unsigned long arg
);
1044 int inet6_hash_connect(struct inet_timewait_death_row
*death_row
,
1050 extern const struct proto_ops inet6_stream_ops
;
1051 extern const struct proto_ops inet6_dgram_ops
;
1052 extern const struct proto_ops inet6_sockraw_ops
;
1054 struct group_source_req
;
1055 struct group_filter
;
1057 int ip6_mc_source(int add
, int omode
, struct sock
*sk
,
1058 struct group_source_req
*pgsr
);
1059 int ip6_mc_msfilter(struct sock
*sk
, struct group_filter
*gsf
);
1060 int ip6_mc_msfget(struct sock
*sk
, struct group_filter
*gsf
,
1061 struct group_filter __user
*optval
, int __user
*optlen
);
1063 #ifdef CONFIG_PROC_FS
1064 int ac6_proc_init(struct net
*net
);
1065 void ac6_proc_exit(struct net
*net
);
1066 int raw6_proc_init(void);
1067 void raw6_proc_exit(void);
1068 int tcp6_proc_init(struct net
*net
);
1069 void tcp6_proc_exit(struct net
*net
);
1070 int udp6_proc_init(struct net
*net
);
1071 void udp6_proc_exit(struct net
*net
);
1072 int udplite6_proc_init(void);
1073 void udplite6_proc_exit(void);
1074 int ipv6_misc_proc_init(void);
1075 void ipv6_misc_proc_exit(void);
1076 int snmp6_register_dev(struct inet6_dev
*idev
);
1077 int snmp6_unregister_dev(struct inet6_dev
*idev
);
1080 static inline int ac6_proc_init(struct net
*net
) { return 0; }
1081 static inline void ac6_proc_exit(struct net
*net
) { }
1082 static inline int snmp6_register_dev(struct inet6_dev
*idev
) { return 0; }
1083 static inline int snmp6_unregister_dev(struct inet6_dev
*idev
) { return 0; }
1086 #ifdef CONFIG_SYSCTL
1087 extern struct ctl_table ipv6_route_table_template
[];
1089 struct ctl_table
*ipv6_icmp_sysctl_init(struct net
*net
);
1090 struct ctl_table
*ipv6_route_sysctl_init(struct net
*net
);
1091 int ipv6_sysctl_register(void);
1092 void ipv6_sysctl_unregister(void);
1095 int ipv6_sock_mc_join(struct sock
*sk
, int ifindex
,
1096 const struct in6_addr
*addr
);
1097 int ipv6_sock_mc_drop(struct sock
*sk
, int ifindex
,
1098 const struct in6_addr
*addr
);
1099 #endif /* _NET_IPV6_H */