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>
25 #include <net/netns/hash.h>
27 #define SIN6_LEN_RFC2133 24
29 #define IPV6_MAXPLEN 65535
32 * NextHeader field of IPv6 header
35 #define NEXTHDR_HOP 0 /* Hop-by-hop option header. */
36 #define NEXTHDR_TCP 6 /* TCP segment. */
37 #define NEXTHDR_UDP 17 /* UDP message. */
38 #define NEXTHDR_IPV6 41 /* IPv6 in IPv6 */
39 #define NEXTHDR_ROUTING 43 /* Routing header. */
40 #define NEXTHDR_FRAGMENT 44 /* Fragmentation/reassembly header. */
41 #define NEXTHDR_GRE 47 /* GRE header. */
42 #define NEXTHDR_ESP 50 /* Encapsulating security payload. */
43 #define NEXTHDR_AUTH 51 /* Authentication header. */
44 #define NEXTHDR_ICMP 58 /* ICMP for IPv6. */
45 #define NEXTHDR_NONE 59 /* No next header */
46 #define NEXTHDR_DEST 60 /* Destination options header. */
47 #define NEXTHDR_SCTP 132 /* SCTP message. */
48 #define NEXTHDR_MOBILITY 135 /* Mobility header. */
50 #define NEXTHDR_MAX 255
52 #define IPV6_DEFAULT_HOPLIMIT 64
53 #define IPV6_DEFAULT_MCASTHOPS 1
55 /* Limits on Hop-by-Hop and Destination options.
57 * Per RFC8200 there is no limit on the maximum number or lengths of options in
58 * Hop-by-Hop or Destination options other then the packet must fit in an MTU.
59 * We allow configurable limits in order to mitigate potential denial of
62 * There are three limits that may be set:
63 * - Limit the number of options in a Hop-by-Hop or Destination options
65 * - Limit the byte length of a Hop-by-Hop or Destination options extension
67 * - Disallow unknown options
69 * The limits are expressed in corresponding sysctls:
71 * ipv6.sysctl.max_dst_opts_cnt
72 * ipv6.sysctl.max_hbh_opts_cnt
73 * ipv6.sysctl.max_dst_opts_len
74 * ipv6.sysctl.max_hbh_opts_len
76 * max_*_opts_cnt is the number of TLVs that are allowed for Destination
77 * options or Hop-by-Hop options. If the number is less than zero then unknown
78 * TLVs are disallowed and the number of known options that are allowed is the
79 * absolute value. Setting the value to INT_MAX indicates no limit.
81 * max_*_opts_len is the length limit in bytes of a Destination or
82 * Hop-by-Hop options extension header. Setting the value to INT_MAX
83 * indicates no length limit.
85 * If a limit is exceeded when processing an extension header the packet is
89 /* Default limits for Hop-by-Hop and Destination options */
90 #define IP6_DEFAULT_MAX_DST_OPTS_CNT 8
91 #define IP6_DEFAULT_MAX_HBH_OPTS_CNT 8
92 #define IP6_DEFAULT_MAX_DST_OPTS_LEN INT_MAX /* No limit */
93 #define IP6_DEFAULT_MAX_HBH_OPTS_LEN INT_MAX /* No limit */
98 * type - unicast | multicast
99 * scope - local | site | global
106 #define IPV6_ADDR_ANY 0x0000U
108 #define IPV6_ADDR_UNICAST 0x0001U
109 #define IPV6_ADDR_MULTICAST 0x0002U
111 #define IPV6_ADDR_LOOPBACK 0x0010U
112 #define IPV6_ADDR_LINKLOCAL 0x0020U
113 #define IPV6_ADDR_SITELOCAL 0x0040U
115 #define IPV6_ADDR_COMPATv4 0x0080U
117 #define IPV6_ADDR_SCOPE_MASK 0x00f0U
119 #define IPV6_ADDR_MAPPED 0x1000U
124 #define IPV6_ADDR_MC_SCOPE(a) \
125 ((a)->s6_addr[1] & 0x0f) /* nonstandard */
126 #define __IPV6_ADDR_SCOPE_INVALID -1
127 #define IPV6_ADDR_SCOPE_NODELOCAL 0x01
128 #define IPV6_ADDR_SCOPE_LINKLOCAL 0x02
129 #define IPV6_ADDR_SCOPE_SITELOCAL 0x05
130 #define IPV6_ADDR_SCOPE_ORGLOCAL 0x08
131 #define IPV6_ADDR_SCOPE_GLOBAL 0x0e
136 #define IPV6_ADDR_MC_FLAG_TRANSIENT(a) \
137 ((a)->s6_addr[1] & 0x10)
138 #define IPV6_ADDR_MC_FLAG_PREFIX(a) \
139 ((a)->s6_addr[1] & 0x20)
140 #define IPV6_ADDR_MC_FLAG_RENDEZVOUS(a) \
141 ((a)->s6_addr[1] & 0x40)
144 * fragmentation header
151 __be32 identification;
154 #define IP6_MF 0x0001
155 #define IP6_OFFSET 0xFFF8
157 #define IP6_REPLY_MARK(net, mark) \
158 ((net)->ipv6.sysctl.fwmark_reflect ? (mark) : 0)
160 #include <net/sock.h>
163 extern int sysctl_mld_max_msf;
164 extern int sysctl_mld_qrv;
166 #define _DEVINC(net, statname, mod, idev, field) \
168 struct inet6_dev *_idev = (idev); \
169 if (likely(_idev != NULL)) \
170 mod##SNMP_INC_STATS64((_idev)->stats.statname, (field));\
171 mod##SNMP_INC_STATS64((net)->mib.statname##_statistics, (field));\
174 /* per device counters are atomic_long_t */
175 #define _DEVINCATOMIC(net, statname, mod, idev, field) \
177 struct inet6_dev *_idev = (idev); \
178 if (likely(_idev != NULL)) \
179 SNMP_INC_STATS_ATOMIC_LONG((_idev)->stats.statname##dev, (field)); \
180 mod##SNMP_INC_STATS((net)->mib.statname##_statistics, (field));\
183 /* per device and per net counters are atomic_long_t */
184 #define _DEVINC_ATOMIC_ATOMIC(net, statname, idev, field) \
186 struct inet6_dev *_idev = (idev); \
187 if (likely(_idev != NULL)) \
188 SNMP_INC_STATS_ATOMIC_LONG((_idev)->stats.statname##dev, (field)); \
189 SNMP_INC_STATS_ATOMIC_LONG((net)->mib.statname##_statistics, (field));\
192 #define _DEVADD(net, statname, mod, idev, field, val) \
194 struct inet6_dev *_idev = (idev); \
195 if (likely(_idev != NULL)) \
196 mod##SNMP_ADD_STATS((_idev)->stats.statname, (field), (val)); \
197 mod##SNMP_ADD_STATS((net)->mib.statname##_statistics, (field), (val));\
200 #define _DEVUPD(net, statname, mod, idev, field, val) \
202 struct inet6_dev *_idev = (idev); \
203 if (likely(_idev != NULL)) \
204 mod##SNMP_UPD_PO_STATS((_idev)->stats.statname, field, (val)); \
205 mod##SNMP_UPD_PO_STATS((net)->mib.statname##_statistics, field, (val));\
210 #define IP6_INC_STATS(net, idev,field) \
211 _DEVINC(net, ipv6, , idev, field)
212 #define __IP6_INC_STATS(net, idev,field) \
213 _DEVINC(net, ipv6, __, idev, field)
214 #define IP6_ADD_STATS(net, idev,field,val) \
215 _DEVADD(net, ipv6, , idev, field, val)
216 #define __IP6_ADD_STATS(net, idev,field,val) \
217 _DEVADD(net, ipv6, __, idev, field, val)
218 #define IP6_UPD_PO_STATS(net, idev,field,val) \
219 _DEVUPD(net, ipv6, , idev, field, val)
220 #define __IP6_UPD_PO_STATS(net, idev,field,val) \
221 _DEVUPD(net, ipv6, __, idev, field, val)
222 #define ICMP6_INC_STATS(net, idev, field) \
223 _DEVINCATOMIC(net, icmpv6, , idev, field)
224 #define __ICMP6_INC_STATS(net, idev, field) \
225 _DEVINCATOMIC(net, icmpv6, __, idev, field)
227 #define ICMP6MSGOUT_INC_STATS(net, idev, field) \
228 _DEVINC_ATOMIC_ATOMIC(net, icmpv6msg, idev, field +256)
229 #define ICMP6MSGIN_INC_STATS(net, idev, field) \
230 _DEVINC_ATOMIC_ATOMIC(net, icmpv6msg, idev, field)
232 struct ip6_ra_chain {
233 struct ip6_ra_chain *next;
236 void (*destructor)(struct sock *);
239 extern struct ip6_ra_chain *ip6_ra_chain;
240 extern rwlock_t ip6_ra_lock;
243 This structure is prepared by protocol, when parsing
244 ancillary data and passed to IPv6.
247 struct ipv6_txoptions {
249 /* Length of this structure */
252 /* length of extension headers */
254 __u16 opt_flen; /* after fragment hdr */
255 __u16 opt_nflen; /* before fragment hdr */
257 struct ipv6_opt_hdr *hopopt;
258 struct ipv6_opt_hdr *dst0opt;
259 struct ipv6_rt_hdr *srcrt; /* Routing Header */
260 struct ipv6_opt_hdr *dst1opt;
262 /* Option buffer, as read by IPV6_PKTOPTIONS, starts here. */
265 struct ip6_flowlabel {
266 struct ip6_flowlabel __rcu *next;
270 struct ipv6_txoptions *opt;
271 unsigned long linger;
278 unsigned long lastuse;
279 unsigned long expires;
283 #define IPV6_FLOWINFO_MASK cpu_to_be32(0x0FFFFFFF)
284 #define IPV6_FLOWLABEL_MASK cpu_to_be32(0x000FFFFF)
285 #define IPV6_FLOWLABEL_STATELESS_FLAG cpu_to_be32(0x00080000)
287 #define IPV6_TCLASS_MASK (IPV6_FLOWINFO_MASK & ~IPV6_FLOWLABEL_MASK)
288 #define IPV6_TCLASS_SHIFT 20
290 struct ipv6_fl_socklist {
291 struct ipv6_fl_socklist __rcu *next;
292 struct ip6_flowlabel *fl;
296 struct ipcm6_cookie {
297 struct sockcm_cookie sockc;
301 struct ipv6_txoptions *opt;
305 static inline void ipcm6_init(struct ipcm6_cookie *ipc6)
307 *ipc6 = (struct ipcm6_cookie) {
314 static inline void ipcm6_init_sk(struct ipcm6_cookie *ipc6,
315 const struct ipv6_pinfo *np)
317 *ipc6 = (struct ipcm6_cookie) {
319 .tclass = np->tclass,
320 .dontfrag = np->dontfrag,
324 static inline struct ipv6_txoptions *txopt_get(const struct ipv6_pinfo *np)
326 struct ipv6_txoptions *opt;
329 opt = rcu_dereference(np->opt);
331 if (!refcount_inc_not_zero(&opt->refcnt))
334 opt = rcu_pointer_handoff(opt);
340 static inline void txopt_put(struct ipv6_txoptions *opt)
342 if (opt && refcount_dec_and_test(&opt->refcnt))
346 struct ip6_flowlabel *fl6_sock_lookup(struct sock *sk, __be32 label);
347 struct ipv6_txoptions *fl6_merge_options(struct ipv6_txoptions *opt_space,
348 struct ip6_flowlabel *fl,
349 struct ipv6_txoptions *fopt);
350 void fl6_free_socklist(struct sock *sk);
351 int ipv6_flowlabel_opt(struct sock *sk, char __user *optval, int optlen);
352 int ipv6_flowlabel_opt_get(struct sock *sk, struct in6_flowlabel_req *freq,
354 int ip6_flowlabel_init(void);
355 void ip6_flowlabel_cleanup(void);
356 bool ip6_autoflowlabel(struct net *net, const struct ipv6_pinfo *np);
358 static inline void fl6_sock_release(struct ip6_flowlabel *fl)
361 atomic_dec(&fl->users);
364 void icmpv6_notify(struct sk_buff *skb, u8 type, u8 code, __be32 info);
366 void icmpv6_push_pending_frames(struct sock *sk, struct flowi6 *fl6,
367 struct icmp6hdr *thdr, int len);
369 int ip6_ra_control(struct sock *sk, int sel);
371 int ipv6_parse_hopopts(struct sk_buff *skb);
373 struct ipv6_txoptions *ipv6_dup_options(struct sock *sk,
374 struct ipv6_txoptions *opt);
375 struct ipv6_txoptions *ipv6_renew_options(struct sock *sk,
376 struct ipv6_txoptions *opt,
378 struct ipv6_opt_hdr *newopt);
379 struct ipv6_txoptions *ipv6_fixup_options(struct ipv6_txoptions *opt_space,
380 struct ipv6_txoptions *opt);
382 bool ipv6_opt_accepted(const struct sock *sk, const struct sk_buff *skb,
383 const struct inet6_skb_parm *opt);
384 struct ipv6_txoptions *ipv6_update_options(struct sock *sk,
385 struct ipv6_txoptions *opt);
387 static inline bool ipv6_accept_ra(struct inet6_dev *idev)
389 /* If forwarding is enabled, RA are not accepted unless the special
390 * hybrid mode (accept_ra=2) is enabled.
392 return idev->cnf.forwarding ? idev->cnf.accept_ra == 2 :
396 #define IPV6_FRAG_HIGH_THRESH (4 * 1024*1024) /* 4194304 */
397 #define IPV6_FRAG_LOW_THRESH (3 * 1024*1024) /* 3145728 */
398 #define IPV6_FRAG_TIMEOUT (60 * HZ) /* 60 seconds */
400 int __ipv6_addr_type(const struct in6_addr *addr);
401 static inline int ipv6_addr_type(const struct in6_addr *addr)
403 return __ipv6_addr_type(addr) & 0xffff;
406 static inline int ipv6_addr_scope(const struct in6_addr *addr)
408 return __ipv6_addr_type(addr) & IPV6_ADDR_SCOPE_MASK;
411 static inline int __ipv6_addr_src_scope(int type)
413 return (type == IPV6_ADDR_ANY) ? __IPV6_ADDR_SCOPE_INVALID : (type >> 16);
416 static inline int ipv6_addr_src_scope(const struct in6_addr *addr)
418 return __ipv6_addr_src_scope(__ipv6_addr_type(addr));
421 static inline bool __ipv6_addr_needs_scope_id(int type)
423 return type & IPV6_ADDR_LINKLOCAL ||
424 (type & IPV6_ADDR_MULTICAST &&
425 (type & (IPV6_ADDR_LOOPBACK|IPV6_ADDR_LINKLOCAL)));
428 static inline __u32 ipv6_iface_scope_id(const struct in6_addr *addr, int iface)
430 return __ipv6_addr_needs_scope_id(__ipv6_addr_type(addr)) ? iface : 0;
433 static inline int ipv6_addr_cmp(const struct in6_addr *a1, const struct in6_addr *a2)
435 return memcmp(a1, a2, sizeof(struct in6_addr));
439 ipv6_masked_addr_cmp(const struct in6_addr *a1, const struct in6_addr *m,
440 const struct in6_addr *a2)
442 #if defined(CONFIG_HAVE_EFFICIENT_UNALIGNED_ACCESS) && BITS_PER_LONG == 64
443 const unsigned long *ul1 = (const unsigned long *)a1;
444 const unsigned long *ulm = (const unsigned long *)m;
445 const unsigned long *ul2 = (const unsigned long *)a2;
447 return !!(((ul1[0] ^ ul2[0]) & ulm[0]) |
448 ((ul1[1] ^ ul2[1]) & ulm[1]));
450 return !!(((a1->s6_addr32[0] ^ a2->s6_addr32[0]) & m->s6_addr32[0]) |
451 ((a1->s6_addr32[1] ^ a2->s6_addr32[1]) & m->s6_addr32[1]) |
452 ((a1->s6_addr32[2] ^ a2->s6_addr32[2]) & m->s6_addr32[2]) |
453 ((a1->s6_addr32[3] ^ a2->s6_addr32[3]) & m->s6_addr32[3]));
457 static inline void ipv6_addr_prefix(struct in6_addr *pfx,
458 const struct in6_addr *addr,
461 /* caller must guarantee 0 <= plen <= 128 */
465 memset(pfx->s6_addr, 0, sizeof(pfx->s6_addr));
466 memcpy(pfx->s6_addr, addr, o);
468 pfx->s6_addr[o] = addr->s6_addr[o] & (0xff00 >> b);
471 static inline void ipv6_addr_prefix_copy(struct in6_addr *addr,
472 const struct in6_addr *pfx,
475 /* caller must guarantee 0 <= plen <= 128 */
479 memcpy(addr->s6_addr, pfx, o);
481 addr->s6_addr[o] &= ~(0xff00 >> b);
482 addr->s6_addr[o] |= (pfx->s6_addr[o] & (0xff00 >> b));
486 static inline void __ipv6_addr_set_half(__be32 *addr,
487 __be32 wh, __be32 wl)
489 #if defined(CONFIG_HAVE_EFFICIENT_UNALIGNED_ACCESS) && BITS_PER_LONG == 64
490 #if defined(__BIG_ENDIAN)
491 if (__builtin_constant_p(wh) && __builtin_constant_p(wl)) {
492 *(__force u64 *)addr = ((__force u64)(wh) << 32 | (__force u64)(wl));
495 #elif defined(__LITTLE_ENDIAN)
496 if (__builtin_constant_p(wl) && __builtin_constant_p(wh)) {
497 *(__force u64 *)addr = ((__force u64)(wl) << 32 | (__force u64)(wh));
506 static inline void ipv6_addr_set(struct in6_addr *addr,
507 __be32 w1, __be32 w2,
508 __be32 w3, __be32 w4)
510 __ipv6_addr_set_half(&addr->s6_addr32[0], w1, w2);
511 __ipv6_addr_set_half(&addr->s6_addr32[2], w3, w4);
514 static inline bool ipv6_addr_equal(const struct in6_addr *a1,
515 const struct in6_addr *a2)
517 #if defined(CONFIG_HAVE_EFFICIENT_UNALIGNED_ACCESS) && BITS_PER_LONG == 64
518 const unsigned long *ul1 = (const unsigned long *)a1;
519 const unsigned long *ul2 = (const unsigned long *)a2;
521 return ((ul1[0] ^ ul2[0]) | (ul1[1] ^ ul2[1])) == 0UL;
523 return ((a1->s6_addr32[0] ^ a2->s6_addr32[0]) |
524 (a1->s6_addr32[1] ^ a2->s6_addr32[1]) |
525 (a1->s6_addr32[2] ^ a2->s6_addr32[2]) |
526 (a1->s6_addr32[3] ^ a2->s6_addr32[3])) == 0;
530 #if defined(CONFIG_HAVE_EFFICIENT_UNALIGNED_ACCESS) && BITS_PER_LONG == 64
531 static inline bool __ipv6_prefix_equal64_half(const __be64 *a1,
535 if (len && ((*a1 ^ *a2) & cpu_to_be64((~0UL) << (64 - len))))
540 static inline bool ipv6_prefix_equal(const struct in6_addr *addr1,
541 const struct in6_addr *addr2,
542 unsigned int prefixlen)
544 const __be64 *a1 = (const __be64 *)addr1;
545 const __be64 *a2 = (const __be64 *)addr2;
547 if (prefixlen >= 64) {
550 return __ipv6_prefix_equal64_half(a1 + 1, a2 + 1, prefixlen - 64);
552 return __ipv6_prefix_equal64_half(a1, a2, prefixlen);
555 static inline bool ipv6_prefix_equal(const struct in6_addr *addr1,
556 const struct in6_addr *addr2,
557 unsigned int prefixlen)
559 const __be32 *a1 = addr1->s6_addr32;
560 const __be32 *a2 = addr2->s6_addr32;
561 unsigned int pdw, pbi;
563 /* check complete u32 in prefix */
564 pdw = prefixlen >> 5;
565 if (pdw && memcmp(a1, a2, pdw << 2))
568 /* check incomplete u32 in prefix */
569 pbi = prefixlen & 0x1f;
570 if (pbi && ((a1[pdw] ^ a2[pdw]) & htonl((0xffffffff) << (32 - pbi))))
577 static inline bool ipv6_addr_any(const struct in6_addr *a)
579 #if defined(CONFIG_HAVE_EFFICIENT_UNALIGNED_ACCESS) && BITS_PER_LONG == 64
580 const unsigned long *ul = (const unsigned long *)a;
582 return (ul[0] | ul[1]) == 0UL;
584 return (a->s6_addr32[0] | a->s6_addr32[1] |
585 a->s6_addr32[2] | a->s6_addr32[3]) == 0;
589 static inline u32 ipv6_addr_hash(const struct in6_addr *a)
591 #if defined(CONFIG_HAVE_EFFICIENT_UNALIGNED_ACCESS) && BITS_PER_LONG == 64
592 const unsigned long *ul = (const unsigned long *)a;
593 unsigned long x = ul[0] ^ ul[1];
595 return (u32)(x ^ (x >> 32));
597 return (__force u32)(a->s6_addr32[0] ^ a->s6_addr32[1] ^
598 a->s6_addr32[2] ^ a->s6_addr32[3]);
602 /* more secured version of ipv6_addr_hash() */
603 static inline u32 __ipv6_addr_jhash(const struct in6_addr *a, const u32 initval)
605 u32 v = (__force u32)a->s6_addr32[0] ^ (__force u32)a->s6_addr32[1];
607 return jhash_3words(v,
608 (__force u32)a->s6_addr32[2],
609 (__force u32)a->s6_addr32[3],
613 static inline bool ipv6_addr_loopback(const struct in6_addr *a)
615 #if defined(CONFIG_HAVE_EFFICIENT_UNALIGNED_ACCESS) && BITS_PER_LONG == 64
616 const __be64 *be = (const __be64 *)a;
618 return (be[0] | (be[1] ^ cpu_to_be64(1))) == 0UL;
620 return (a->s6_addr32[0] | a->s6_addr32[1] |
621 a->s6_addr32[2] | (a->s6_addr32[3] ^ cpu_to_be32(1))) == 0;
626 * Note that we must __force cast these to unsigned long to make sparse happy,
627 * since all of the endian-annotated types are fixed size regardless of arch.
629 static inline bool ipv6_addr_v4mapped(const struct in6_addr *a)
632 #if defined(CONFIG_HAVE_EFFICIENT_UNALIGNED_ACCESS) && BITS_PER_LONG == 64
633 *(unsigned long *)a |
635 (__force unsigned long)(a->s6_addr32[0] | a->s6_addr32[1]) |
637 (__force unsigned long)(a->s6_addr32[2] ^
638 cpu_to_be32(0x0000ffff))) == 0UL;
641 static inline u32 ipv6_portaddr_hash(const struct net *net,
642 const struct in6_addr *addr6,
645 unsigned int hash, mix = net_hash_mix(net);
647 if (ipv6_addr_any(addr6))
648 hash = jhash_1word(0, mix);
649 else if (ipv6_addr_v4mapped(addr6))
650 hash = jhash_1word((__force u32)addr6->s6_addr32[3], mix);
652 hash = jhash2((__force u32 *)addr6->s6_addr32, 4, mix);
658 * Check for a RFC 4843 ORCHID address
659 * (Overlay Routable Cryptographic Hash Identifiers)
661 static inline bool ipv6_addr_orchid(const struct in6_addr *a)
663 return (a->s6_addr32[0] & htonl(0xfffffff0)) == htonl(0x20010010);
666 static inline bool ipv6_addr_is_multicast(const struct in6_addr *addr)
668 return (addr->s6_addr32[0] & htonl(0xFF000000)) == htonl(0xFF000000);
671 static inline void ipv6_addr_set_v4mapped(const __be32 addr,
672 struct in6_addr *v4mapped)
674 ipv6_addr_set(v4mapped,
681 * find the first different bit between two addresses
682 * length of address must be a multiple of 32bits
684 static inline int __ipv6_addr_diff32(const void *token1, const void *token2, int addrlen)
686 const __be32 *a1 = token1, *a2 = token2;
691 for (i = 0; i < addrlen; i++) {
692 __be32 xb = a1[i] ^ a2[i];
694 return i * 32 + 31 - __fls(ntohl(xb));
698 * we should *never* get to this point since that
699 * would mean the addrs are equal
701 * However, we do get to it 8) And exacly, when
702 * addresses are equal 8)
704 * ip route add 1111::/128 via ...
705 * ip route add 1111::/64 via ...
708 * Ideally, this function should stop comparison
709 * at prefix length. It does not, but it is still OK,
710 * if returned value is greater than prefix length.
716 #if defined(CONFIG_HAVE_EFFICIENT_UNALIGNED_ACCESS) && BITS_PER_LONG == 64
717 static inline int __ipv6_addr_diff64(const void *token1, const void *token2, int addrlen)
719 const __be64 *a1 = token1, *a2 = token2;
724 for (i = 0; i < addrlen; i++) {
725 __be64 xb = a1[i] ^ a2[i];
727 return i * 64 + 63 - __fls(be64_to_cpu(xb));
734 static inline int __ipv6_addr_diff(const void *token1, const void *token2, int addrlen)
736 #if defined(CONFIG_HAVE_EFFICIENT_UNALIGNED_ACCESS) && BITS_PER_LONG == 64
737 if (__builtin_constant_p(addrlen) && !(addrlen & 7))
738 return __ipv6_addr_diff64(token1, token2, addrlen);
740 return __ipv6_addr_diff32(token1, token2, addrlen);
743 static inline int ipv6_addr_diff(const struct in6_addr *a1, const struct in6_addr *a2)
745 return __ipv6_addr_diff(a1, a2, sizeof(struct in6_addr));
748 __be32 ipv6_select_ident(struct net *net,
749 const struct in6_addr *daddr,
750 const struct in6_addr *saddr);
751 __be32 ipv6_proxy_select_ident(struct net *net, struct sk_buff *skb);
753 int ip6_dst_hoplimit(struct dst_entry *dst);
755 static inline int ip6_sk_dst_hoplimit(struct ipv6_pinfo *np, struct flowi6 *fl6,
756 struct dst_entry *dst)
760 if (ipv6_addr_is_multicast(&fl6->daddr))
761 hlimit = np->mcast_hops;
763 hlimit = np->hop_limit;
765 hlimit = ip6_dst_hoplimit(dst);
769 /* copy IPv6 saddr & daddr to flow_keys, possibly using 64bit load/store
770 * Equivalent to : flow->v6addrs.src = iph->saddr;
771 * flow->v6addrs.dst = iph->daddr;
773 static inline void iph_to_flow_copy_v6addrs(struct flow_keys *flow,
774 const struct ipv6hdr *iph)
776 BUILD_BUG_ON(offsetof(typeof(flow->addrs), v6addrs.dst) !=
777 offsetof(typeof(flow->addrs), v6addrs.src) +
778 sizeof(flow->addrs.v6addrs.src));
779 memcpy(&flow->addrs.v6addrs, &iph->saddr, sizeof(flow->addrs.v6addrs));
780 flow->control.addr_type = FLOW_DISSECTOR_KEY_IPV6_ADDRS;
783 #if IS_ENABLED(CONFIG_IPV6)
785 static inline bool ipv6_can_nonlocal_bind(struct net *net,
786 struct inet_sock *inet)
788 return net->ipv6.sysctl.ip_nonlocal_bind ||
789 inet->freebind || inet->transparent;
792 /* Sysctl settings for net ipv6.auto_flowlabels */
793 #define IP6_AUTO_FLOW_LABEL_OFF 0
794 #define IP6_AUTO_FLOW_LABEL_OPTOUT 1
795 #define IP6_AUTO_FLOW_LABEL_OPTIN 2
796 #define IP6_AUTO_FLOW_LABEL_FORCED 3
798 #define IP6_AUTO_FLOW_LABEL_MAX IP6_AUTO_FLOW_LABEL_FORCED
800 #define IP6_DEFAULT_AUTO_FLOW_LABELS IP6_AUTO_FLOW_LABEL_OPTOUT
802 static inline __be32 ip6_make_flowlabel(struct net *net, struct sk_buff *skb,
803 __be32 flowlabel, bool autolabel,
808 /* @flowlabel may include more than a flow label, eg, the traffic class.
809 * Here we want only the flow label value.
811 flowlabel &= IPV6_FLOWLABEL_MASK;
814 net->ipv6.sysctl.auto_flowlabels == IP6_AUTO_FLOW_LABEL_OFF ||
816 net->ipv6.sysctl.auto_flowlabels != IP6_AUTO_FLOW_LABEL_FORCED))
819 hash = skb_get_hash_flowi6(skb, fl6);
821 /* Since this is being sent on the wire obfuscate hash a bit
822 * to minimize possbility that any useful information to an
823 * attacker is leaked. Only lower 20 bits are relevant.
825 hash = rol32(hash, 16);
827 flowlabel = (__force __be32)hash & IPV6_FLOWLABEL_MASK;
829 if (net->ipv6.sysctl.flowlabel_state_ranges)
830 flowlabel |= IPV6_FLOWLABEL_STATELESS_FLAG;
835 static inline int ip6_default_np_autolabel(struct net *net)
837 switch (net->ipv6.sysctl.auto_flowlabels) {
838 case IP6_AUTO_FLOW_LABEL_OFF:
839 case IP6_AUTO_FLOW_LABEL_OPTIN:
842 case IP6_AUTO_FLOW_LABEL_OPTOUT:
843 case IP6_AUTO_FLOW_LABEL_FORCED:
848 static inline void ip6_set_txhash(struct sock *sk) { }
849 static inline __be32 ip6_make_flowlabel(struct net *net, struct sk_buff *skb,
850 __be32 flowlabel, bool autolabel,
855 static inline int ip6_default_np_autolabel(struct net *net)
861 #if IS_ENABLED(CONFIG_IPV6)
862 static inline int ip6_multipath_hash_policy(const struct net *net)
864 return net->ipv6.sysctl.multipath_hash_policy;
867 static inline int ip6_multipath_hash_policy(const struct net *net)
874 * Header manipulation
876 static inline void ip6_flow_hdr(struct ipv6hdr *hdr, unsigned int tclass,
879 *(__be32 *)hdr = htonl(0x60000000 | (tclass << 20)) | flowlabel;
882 static inline __be32 ip6_flowinfo(const struct ipv6hdr *hdr)
884 return *(__be32 *)hdr & IPV6_FLOWINFO_MASK;
887 static inline __be32 ip6_flowlabel(const struct ipv6hdr *hdr)
889 return *(__be32 *)hdr & IPV6_FLOWLABEL_MASK;
892 static inline u8 ip6_tclass(__be32 flowinfo)
894 return ntohl(flowinfo & IPV6_TCLASS_MASK) >> IPV6_TCLASS_SHIFT;
897 static inline __be32 ip6_make_flowinfo(unsigned int tclass, __be32 flowlabel)
899 return htonl(tclass << IPV6_TCLASS_SHIFT) | flowlabel;
902 static inline __be32 flowi6_get_flowlabel(const struct flowi6 *fl6)
904 return fl6->flowlabel & IPV6_FLOWLABEL_MASK;
908 * Prototypes exported by ipv6
912 * rcv function (called from netdevice level)
915 int ipv6_rcv(struct sk_buff *skb, struct net_device *dev,
916 struct packet_type *pt, struct net_device *orig_dev);
917 void ipv6_list_rcv(struct list_head *head, struct packet_type *pt,
918 struct net_device *orig_dev);
920 int ip6_rcv_finish(struct net *net, struct sock *sk, struct sk_buff *skb);
923 * upper-layer output functions
925 int ip6_xmit(const struct sock *sk, struct sk_buff *skb, struct flowi6 *fl6,
926 __u32 mark, struct ipv6_txoptions *opt, int tclass);
928 int ip6_find_1stfragopt(struct sk_buff *skb, u8 **nexthdr);
930 int ip6_append_data(struct sock *sk,
931 int getfrag(void *from, char *to, int offset, int len,
932 int odd, struct sk_buff *skb),
933 void *from, int length, int transhdrlen,
934 struct ipcm6_cookie *ipc6, struct flowi6 *fl6,
935 struct rt6_info *rt, unsigned int flags);
937 int ip6_push_pending_frames(struct sock *sk);
939 void ip6_flush_pending_frames(struct sock *sk);
941 int ip6_send_skb(struct sk_buff *skb);
943 struct sk_buff *__ip6_make_skb(struct sock *sk, struct sk_buff_head *queue,
944 struct inet_cork_full *cork,
945 struct inet6_cork *v6_cork);
946 struct sk_buff *ip6_make_skb(struct sock *sk,
947 int getfrag(void *from, char *to, int offset,
948 int len, int odd, struct sk_buff *skb),
949 void *from, int length, int transhdrlen,
950 struct ipcm6_cookie *ipc6, struct flowi6 *fl6,
951 struct rt6_info *rt, unsigned int flags,
952 struct inet_cork_full *cork);
954 static inline struct sk_buff *ip6_finish_skb(struct sock *sk)
956 return __ip6_make_skb(sk, &sk->sk_write_queue, &inet_sk(sk)->cork,
957 &inet6_sk(sk)->cork);
960 int ip6_dst_lookup(struct net *net, struct sock *sk, struct dst_entry **dst,
962 struct dst_entry *ip6_dst_lookup_flow(const struct sock *sk, struct flowi6 *fl6,
963 const struct in6_addr *final_dst);
964 struct dst_entry *ip6_sk_dst_lookup_flow(struct sock *sk, struct flowi6 *fl6,
965 const struct in6_addr *final_dst,
967 struct dst_entry *ip6_blackhole_route(struct net *net,
968 struct dst_entry *orig_dst);
971 * skb processing functions
974 int ip6_output(struct net *net, struct sock *sk, struct sk_buff *skb);
975 int ip6_forward(struct sk_buff *skb);
976 int ip6_input(struct sk_buff *skb);
977 int ip6_mc_input(struct sk_buff *skb);
979 int __ip6_local_out(struct net *net, struct sock *sk, struct sk_buff *skb);
980 int ip6_local_out(struct net *net, struct sock *sk, struct sk_buff *skb);
983 * Extension header (options) processing
986 void ipv6_push_nfrag_opts(struct sk_buff *skb, struct ipv6_txoptions *opt,
987 u8 *proto, struct in6_addr **daddr_p,
988 struct in6_addr *saddr);
989 void ipv6_push_frag_opts(struct sk_buff *skb, struct ipv6_txoptions *opt,
992 int ipv6_skip_exthdr(const struct sk_buff *, int start, u8 *nexthdrp,
995 bool ipv6_ext_hdr(u8 nexthdr);
998 IP6_FH_F_FRAG = (1 << 0),
999 IP6_FH_F_AUTH = (1 << 1),
1000 IP6_FH_F_SKIP_RH = (1 << 2),
1003 /* find specified header and get offset to it */
1004 int ipv6_find_hdr(const struct sk_buff *skb, unsigned int *offset, int target,
1005 unsigned short *fragoff, int *fragflg);
1007 int ipv6_find_tlv(const struct sk_buff *skb, int offset, int type);
1009 struct in6_addr *fl6_update_dst(struct flowi6 *fl6,
1010 const struct ipv6_txoptions *opt,
1011 struct in6_addr *orig);
1014 * socket options (ipv6_sockglue.c)
1017 int ipv6_setsockopt(struct sock *sk, int level, int optname,
1018 char __user *optval, unsigned int optlen);
1019 int ipv6_getsockopt(struct sock *sk, int level, int optname,
1020 char __user *optval, int __user *optlen);
1021 int compat_ipv6_setsockopt(struct sock *sk, int level, int optname,
1022 char __user *optval, unsigned int optlen);
1023 int compat_ipv6_getsockopt(struct sock *sk, int level, int optname,
1024 char __user *optval, int __user *optlen);
1026 int __ip6_datagram_connect(struct sock *sk, struct sockaddr *addr,
1028 int ip6_datagram_connect(struct sock *sk, struct sockaddr *addr, int addr_len);
1029 int ip6_datagram_connect_v6_only(struct sock *sk, struct sockaddr *addr,
1031 int ip6_datagram_dst_update(struct sock *sk, bool fix_sk_saddr);
1032 void ip6_datagram_release_cb(struct sock *sk);
1034 int ipv6_recv_error(struct sock *sk, struct msghdr *msg, int len,
1036 int ipv6_recv_rxpmtu(struct sock *sk, struct msghdr *msg, int len,
1038 void ipv6_icmp_error(struct sock *sk, struct sk_buff *skb, int err, __be16 port,
1039 u32 info, u8 *payload);
1040 void ipv6_local_error(struct sock *sk, int err, struct flowi6 *fl6, u32 info);
1041 void ipv6_local_rxpmtu(struct sock *sk, struct flowi6 *fl6, u32 mtu);
1043 int inet6_release(struct socket *sock);
1044 int inet6_bind(struct socket *sock, struct sockaddr *uaddr, int addr_len);
1045 int inet6_getname(struct socket *sock, struct sockaddr *uaddr,
1047 int inet6_ioctl(struct socket *sock, unsigned int cmd, unsigned long arg);
1049 int inet6_hash_connect(struct inet_timewait_death_row *death_row,
1055 extern const struct proto_ops inet6_stream_ops;
1056 extern const struct proto_ops inet6_dgram_ops;
1057 extern const struct proto_ops inet6_sockraw_ops;
1059 struct group_source_req;
1060 struct group_filter;
1062 int ip6_mc_source(int add, int omode, struct sock *sk,
1063 struct group_source_req *pgsr);
1064 int ip6_mc_msfilter(struct sock *sk, struct group_filter *gsf);
1065 int ip6_mc_msfget(struct sock *sk, struct group_filter *gsf,
1066 struct group_filter __user *optval, int __user *optlen);
1068 #ifdef CONFIG_PROC_FS
1069 int ac6_proc_init(struct net *net);
1070 void ac6_proc_exit(struct net *net);
1071 int raw6_proc_init(void);
1072 void raw6_proc_exit(void);
1073 int tcp6_proc_init(struct net *net);
1074 void tcp6_proc_exit(struct net *net);
1075 int udp6_proc_init(struct net *net);
1076 void udp6_proc_exit(struct net *net);
1077 int udplite6_proc_init(void);
1078 void udplite6_proc_exit(void);
1079 int ipv6_misc_proc_init(void);
1080 void ipv6_misc_proc_exit(void);
1081 int snmp6_register_dev(struct inet6_dev *idev);
1082 int snmp6_unregister_dev(struct inet6_dev *idev);
1085 static inline int ac6_proc_init(struct net *net) { return 0; }
1086 static inline void ac6_proc_exit(struct net *net) { }
1087 static inline int snmp6_register_dev(struct inet6_dev *idev) { return 0; }
1088 static inline int snmp6_unregister_dev(struct inet6_dev *idev) { return 0; }
1091 #ifdef CONFIG_SYSCTL
1092 extern struct ctl_table ipv6_route_table_template[];
1094 struct ctl_table *ipv6_icmp_sysctl_init(struct net *net);
1095 struct ctl_table *ipv6_route_sysctl_init(struct net *net);
1096 int ipv6_sysctl_register(void);
1097 void ipv6_sysctl_unregister(void);
1100 int ipv6_sock_mc_join(struct sock *sk, int ifindex,
1101 const struct in6_addr *addr);
1102 int ipv6_sock_mc_join_ssm(struct sock *sk, int ifindex,
1103 const struct in6_addr *addr, unsigned int mode);
1104 int ipv6_sock_mc_drop(struct sock *sk, int ifindex,
1105 const struct in6_addr *addr);
1106 #endif /* _NET_IPV6_H */