1 /* SPDX-License-Identifier: GPL-2.0-or-later */
3 * Linux INET6 implementation
6 * Pedro Roque <roque@di.fc.ul.pt>
12 #include <linux/ipv6.h>
13 #include <linux/hardirq.h>
14 #include <linux/jhash.h>
15 #include <linux/refcount.h>
16 #include <linux/jump_label_ratelimit.h>
17 #include <net/if_inet6.h>
19 #include <net/flow_dissector.h>
20 #include <net/inet_dscp.h>
22 #include <net/netns/hash.h>
24 struct ip_tunnel_info;
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_IPV4 4 /* IPv4 in IPv6 */
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;
155 * Jumbo payload option, as described in RFC 2675 2.
157 struct hop_jumbo_hdr {
160 u8 tlv_type; /* IPV6_TLV_JUMBO, 0xC2 */
162 __be32 jumbo_payload_len;
165 #define IP6_MF 0x0001
166 #define IP6_OFFSET 0xFFF8
168 struct ip6_fraglist_iter {
169 struct ipv6hdr *tmp_hdr;
170 struct sk_buff *frag;
177 int ip6_fraglist_init(struct sk_buff *skb, unsigned int hlen, u8 *prevhdr,
178 u8 nexthdr, __be32 frag_id,
179 struct ip6_fraglist_iter *iter);
180 void ip6_fraglist_prepare(struct sk_buff *skb, struct ip6_fraglist_iter *iter);
182 static inline struct sk_buff *ip6_fraglist_next(struct ip6_fraglist_iter *iter)
184 struct sk_buff *skb = iter->frag;
186 iter->frag = skb->next;
187 skb_mark_not_on_list(skb);
192 struct ip6_frag_state {
205 void ip6_frag_init(struct sk_buff *skb, unsigned int hlen, unsigned int mtu,
206 unsigned short needed_tailroom, int hdr_room, u8 *prevhdr,
207 u8 nexthdr, __be32 frag_id, struct ip6_frag_state *state);
208 struct sk_buff *ip6_frag_next(struct sk_buff *skb,
209 struct ip6_frag_state *state);
211 #define IP6_REPLY_MARK(net, mark) \
212 ((net)->ipv6.sysctl.fwmark_reflect ? (mark) : 0)
214 #include <net/sock.h>
217 extern int sysctl_mld_max_msf;
218 extern int sysctl_mld_qrv;
220 #define _DEVINC(net, statname, mod, idev, field) \
222 struct inet6_dev *_idev = (idev); \
223 if (likely(_idev != NULL)) \
224 mod##SNMP_INC_STATS64((_idev)->stats.statname, (field));\
225 mod##SNMP_INC_STATS64((net)->mib.statname##_statistics, (field));\
228 /* per device counters are atomic_long_t */
229 #define _DEVINCATOMIC(net, statname, mod, idev, field) \
231 struct inet6_dev *_idev = (idev); \
232 if (likely(_idev != NULL)) \
233 SNMP_INC_STATS_ATOMIC_LONG((_idev)->stats.statname##dev, (field)); \
234 mod##SNMP_INC_STATS((net)->mib.statname##_statistics, (field));\
237 /* per device and per net counters are atomic_long_t */
238 #define _DEVINC_ATOMIC_ATOMIC(net, statname, idev, field) \
240 struct inet6_dev *_idev = (idev); \
241 if (likely(_idev != NULL)) \
242 SNMP_INC_STATS_ATOMIC_LONG((_idev)->stats.statname##dev, (field)); \
243 SNMP_INC_STATS_ATOMIC_LONG((net)->mib.statname##_statistics, (field));\
246 #define _DEVADD(net, statname, mod, idev, field, val) \
248 struct inet6_dev *_idev = (idev); \
249 if (likely(_idev != NULL)) \
250 mod##SNMP_ADD_STATS((_idev)->stats.statname, (field), (val)); \
251 mod##SNMP_ADD_STATS((net)->mib.statname##_statistics, (field), (val));\
254 #define _DEVUPD(net, statname, mod, idev, field, val) \
256 struct inet6_dev *_idev = (idev); \
257 if (likely(_idev != NULL)) \
258 mod##SNMP_UPD_PO_STATS((_idev)->stats.statname, field, (val)); \
259 mod##SNMP_UPD_PO_STATS((net)->mib.statname##_statistics, field, (val));\
264 #define IP6_INC_STATS(net, idev,field) \
265 _DEVINC(net, ipv6, , idev, field)
266 #define __IP6_INC_STATS(net, idev,field) \
267 _DEVINC(net, ipv6, __, idev, field)
268 #define IP6_ADD_STATS(net, idev,field,val) \
269 _DEVADD(net, ipv6, , idev, field, val)
270 #define __IP6_ADD_STATS(net, idev,field,val) \
271 _DEVADD(net, ipv6, __, idev, field, val)
272 #define IP6_UPD_PO_STATS(net, idev,field,val) \
273 _DEVUPD(net, ipv6, , idev, field, val)
274 #define __IP6_UPD_PO_STATS(net, idev,field,val) \
275 _DEVUPD(net, ipv6, __, idev, field, val)
276 #define ICMP6_INC_STATS(net, idev, field) \
277 _DEVINCATOMIC(net, icmpv6, , idev, field)
278 #define __ICMP6_INC_STATS(net, idev, field) \
279 _DEVINCATOMIC(net, icmpv6, __, idev, field)
281 #define ICMP6MSGOUT_INC_STATS(net, idev, field) \
282 _DEVINC_ATOMIC_ATOMIC(net, icmpv6msg, idev, field +256)
283 #define ICMP6MSGIN_INC_STATS(net, idev, field) \
284 _DEVINC_ATOMIC_ATOMIC(net, icmpv6msg, idev, field)
286 struct ip6_ra_chain {
287 struct ip6_ra_chain *next;
290 void (*destructor)(struct sock *);
293 extern struct ip6_ra_chain *ip6_ra_chain;
294 extern rwlock_t ip6_ra_lock;
297 This structure is prepared by protocol, when parsing
298 ancillary data and passed to IPv6.
301 struct ipv6_txoptions {
303 /* Length of this structure */
306 /* length of extension headers */
308 __u16 opt_flen; /* after fragment hdr */
309 __u16 opt_nflen; /* before fragment hdr */
311 struct ipv6_opt_hdr *hopopt;
312 struct ipv6_opt_hdr *dst0opt;
313 struct ipv6_rt_hdr *srcrt; /* Routing Header */
314 struct ipv6_opt_hdr *dst1opt;
316 /* Option buffer, as read by IPV6_PKTOPTIONS, starts here. */
319 /* flowlabel_reflect sysctl values */
320 enum flowlabel_reflect {
321 FLOWLABEL_REFLECT_ESTABLISHED = 1,
322 FLOWLABEL_REFLECT_TCP_RESET = 2,
323 FLOWLABEL_REFLECT_ICMPV6_ECHO_REPLIES = 4,
326 struct ip6_flowlabel {
327 struct ip6_flowlabel __rcu *next;
331 struct ipv6_txoptions *opt;
332 unsigned long linger;
339 unsigned long lastuse;
340 unsigned long expires;
344 #define IPV6_FLOWINFO_MASK cpu_to_be32(0x0FFFFFFF)
345 #define IPV6_FLOWLABEL_MASK cpu_to_be32(0x000FFFFF)
346 #define IPV6_FLOWLABEL_STATELESS_FLAG cpu_to_be32(0x00080000)
348 #define IPV6_TCLASS_MASK (IPV6_FLOWINFO_MASK & ~IPV6_FLOWLABEL_MASK)
349 #define IPV6_TCLASS_SHIFT 20
351 struct ipv6_fl_socklist {
352 struct ipv6_fl_socklist __rcu *next;
353 struct ip6_flowlabel *fl;
357 struct ipcm6_cookie {
358 struct sockcm_cookie sockc;
363 struct ipv6_txoptions *opt;
366 static inline void ipcm6_init(struct ipcm6_cookie *ipc6)
368 *ipc6 = (struct ipcm6_cookie) {
375 static inline void ipcm6_init_sk(struct ipcm6_cookie *ipc6,
376 const struct ipv6_pinfo *np)
378 *ipc6 = (struct ipcm6_cookie) {
380 .tclass = np->tclass,
381 .dontfrag = np->dontfrag,
385 static inline struct ipv6_txoptions *txopt_get(const struct ipv6_pinfo *np)
387 struct ipv6_txoptions *opt;
390 opt = rcu_dereference(np->opt);
392 if (!refcount_inc_not_zero(&opt->refcnt))
395 opt = rcu_pointer_handoff(opt);
401 static inline void txopt_put(struct ipv6_txoptions *opt)
403 if (opt && refcount_dec_and_test(&opt->refcnt))
407 #if IS_ENABLED(CONFIG_IPV6)
408 struct ip6_flowlabel *__fl6_sock_lookup(struct sock *sk, __be32 label);
410 extern struct static_key_false_deferred ipv6_flowlabel_exclusive;
411 static inline struct ip6_flowlabel *fl6_sock_lookup(struct sock *sk,
414 if (static_branch_unlikely(&ipv6_flowlabel_exclusive.key) &&
415 READ_ONCE(sock_net(sk)->ipv6.flowlabel_has_excl))
416 return __fl6_sock_lookup(sk, label) ? : ERR_PTR(-ENOENT);
422 struct ipv6_txoptions *fl6_merge_options(struct ipv6_txoptions *opt_space,
423 struct ip6_flowlabel *fl,
424 struct ipv6_txoptions *fopt);
425 void fl6_free_socklist(struct sock *sk);
426 int ipv6_flowlabel_opt(struct sock *sk, sockptr_t optval, int optlen);
427 int ipv6_flowlabel_opt_get(struct sock *sk, struct in6_flowlabel_req *freq,
429 int ip6_flowlabel_init(void);
430 void ip6_flowlabel_cleanup(void);
431 bool ip6_autoflowlabel(struct net *net, const struct ipv6_pinfo *np);
433 static inline void fl6_sock_release(struct ip6_flowlabel *fl)
436 atomic_dec(&fl->users);
439 enum skb_drop_reason icmpv6_notify(struct sk_buff *skb, u8 type,
440 u8 code, __be32 info);
442 void icmpv6_push_pending_frames(struct sock *sk, struct flowi6 *fl6,
443 struct icmp6hdr *thdr, int len);
445 int ip6_ra_control(struct sock *sk, int sel);
447 int ipv6_parse_hopopts(struct sk_buff *skb);
449 struct ipv6_txoptions *ipv6_dup_options(struct sock *sk,
450 struct ipv6_txoptions *opt);
451 struct ipv6_txoptions *ipv6_renew_options(struct sock *sk,
452 struct ipv6_txoptions *opt,
454 struct ipv6_opt_hdr *newopt);
455 struct ipv6_txoptions *__ipv6_fixup_options(struct ipv6_txoptions *opt_space,
456 struct ipv6_txoptions *opt);
458 static inline struct ipv6_txoptions *
459 ipv6_fixup_options(struct ipv6_txoptions *opt_space, struct ipv6_txoptions *opt)
463 return __ipv6_fixup_options(opt_space, opt);
466 bool ipv6_opt_accepted(const struct sock *sk, const struct sk_buff *skb,
467 const struct inet6_skb_parm *opt);
468 struct ipv6_txoptions *ipv6_update_options(struct sock *sk,
469 struct ipv6_txoptions *opt);
471 /* This helper is specialized for BIG TCP needs.
472 * It assumes the hop_jumbo_hdr will immediately follow the IPV6 header.
473 * It assumes headers are already in skb->head.
474 * Returns 0, or IPPROTO_TCP if a BIG TCP packet is there.
476 static inline int ipv6_has_hopopt_jumbo(const struct sk_buff *skb)
478 const struct hop_jumbo_hdr *jhdr;
479 const struct ipv6hdr *nhdr;
481 if (likely(skb->len <= GRO_LEGACY_MAX_SIZE))
484 if (skb->protocol != htons(ETH_P_IPV6))
487 if (skb_network_offset(skb) +
488 sizeof(struct ipv6hdr) +
489 sizeof(struct hop_jumbo_hdr) > skb_headlen(skb))
492 nhdr = ipv6_hdr(skb);
494 if (nhdr->nexthdr != NEXTHDR_HOP)
497 jhdr = (const struct hop_jumbo_hdr *) (nhdr + 1);
498 if (jhdr->tlv_type != IPV6_TLV_JUMBO || jhdr->hdrlen != 0 ||
499 jhdr->nexthdr != IPPROTO_TCP)
501 return jhdr->nexthdr;
504 /* Return 0 if HBH header is successfully removed
505 * Or if HBH removal is unnecessary (packet is not big TCP)
506 * Return error to indicate dropping the packet
508 static inline int ipv6_hopopt_jumbo_remove(struct sk_buff *skb)
510 const int hophdr_len = sizeof(struct hop_jumbo_hdr);
511 int nexthdr = ipv6_has_hopopt_jumbo(skb);
517 if (skb_cow_head(skb, 0))
520 /* Remove the HBH header.
521 * Layout: [Ethernet header][IPv6 header][HBH][L4 Header]
523 memmove(skb_mac_header(skb) + hophdr_len, skb_mac_header(skb),
524 skb_network_header(skb) - skb_mac_header(skb) +
525 sizeof(struct ipv6hdr));
527 __skb_pull(skb, hophdr_len);
528 skb->network_header += hophdr_len;
529 skb->mac_header += hophdr_len;
532 h6->nexthdr = nexthdr;
537 static inline bool ipv6_accept_ra(struct inet6_dev *idev)
539 /* If forwarding is enabled, RA are not accepted unless the special
540 * hybrid mode (accept_ra=2) is enabled.
542 return idev->cnf.forwarding ? idev->cnf.accept_ra == 2 :
546 #define IPV6_FRAG_HIGH_THRESH (4 * 1024*1024) /* 4194304 */
547 #define IPV6_FRAG_LOW_THRESH (3 * 1024*1024) /* 3145728 */
548 #define IPV6_FRAG_TIMEOUT (60 * HZ) /* 60 seconds */
550 int __ipv6_addr_type(const struct in6_addr *addr);
551 static inline int ipv6_addr_type(const struct in6_addr *addr)
553 return __ipv6_addr_type(addr) & 0xffff;
556 static inline int ipv6_addr_scope(const struct in6_addr *addr)
558 return __ipv6_addr_type(addr) & IPV6_ADDR_SCOPE_MASK;
561 static inline int __ipv6_addr_src_scope(int type)
563 return (type == IPV6_ADDR_ANY) ? __IPV6_ADDR_SCOPE_INVALID : (type >> 16);
566 static inline int ipv6_addr_src_scope(const struct in6_addr *addr)
568 return __ipv6_addr_src_scope(__ipv6_addr_type(addr));
571 static inline bool __ipv6_addr_needs_scope_id(int type)
573 return type & IPV6_ADDR_LINKLOCAL ||
574 (type & IPV6_ADDR_MULTICAST &&
575 (type & (IPV6_ADDR_LOOPBACK|IPV6_ADDR_LINKLOCAL)));
578 static inline __u32 ipv6_iface_scope_id(const struct in6_addr *addr, int iface)
580 return __ipv6_addr_needs_scope_id(__ipv6_addr_type(addr)) ? iface : 0;
583 static inline int ipv6_addr_cmp(const struct in6_addr *a1, const struct in6_addr *a2)
585 return memcmp(a1, a2, sizeof(struct in6_addr));
589 ipv6_masked_addr_cmp(const struct in6_addr *a1, const struct in6_addr *m,
590 const struct in6_addr *a2)
592 #if defined(CONFIG_HAVE_EFFICIENT_UNALIGNED_ACCESS) && BITS_PER_LONG == 64
593 const unsigned long *ul1 = (const unsigned long *)a1;
594 const unsigned long *ulm = (const unsigned long *)m;
595 const unsigned long *ul2 = (const unsigned long *)a2;
597 return !!(((ul1[0] ^ ul2[0]) & ulm[0]) |
598 ((ul1[1] ^ ul2[1]) & ulm[1]));
600 return !!(((a1->s6_addr32[0] ^ a2->s6_addr32[0]) & m->s6_addr32[0]) |
601 ((a1->s6_addr32[1] ^ a2->s6_addr32[1]) & m->s6_addr32[1]) |
602 ((a1->s6_addr32[2] ^ a2->s6_addr32[2]) & m->s6_addr32[2]) |
603 ((a1->s6_addr32[3] ^ a2->s6_addr32[3]) & m->s6_addr32[3]));
607 static inline void ipv6_addr_prefix(struct in6_addr *pfx,
608 const struct in6_addr *addr,
611 /* caller must guarantee 0 <= plen <= 128 */
615 memset(pfx->s6_addr, 0, sizeof(pfx->s6_addr));
616 memcpy(pfx->s6_addr, addr, o);
618 pfx->s6_addr[o] = addr->s6_addr[o] & (0xff00 >> b);
621 static inline void ipv6_addr_prefix_copy(struct in6_addr *addr,
622 const struct in6_addr *pfx,
625 /* caller must guarantee 0 <= plen <= 128 */
629 memcpy(addr->s6_addr, pfx, o);
631 addr->s6_addr[o] &= ~(0xff00 >> b);
632 addr->s6_addr[o] |= (pfx->s6_addr[o] & (0xff00 >> b));
636 static inline void __ipv6_addr_set_half(__be32 *addr,
637 __be32 wh, __be32 wl)
639 #if defined(CONFIG_HAVE_EFFICIENT_UNALIGNED_ACCESS) && BITS_PER_LONG == 64
640 #if defined(__BIG_ENDIAN)
641 if (__builtin_constant_p(wh) && __builtin_constant_p(wl)) {
642 *(__force u64 *)addr = ((__force u64)(wh) << 32 | (__force u64)(wl));
645 #elif defined(__LITTLE_ENDIAN)
646 if (__builtin_constant_p(wl) && __builtin_constant_p(wh)) {
647 *(__force u64 *)addr = ((__force u64)(wl) << 32 | (__force u64)(wh));
656 static inline void ipv6_addr_set(struct in6_addr *addr,
657 __be32 w1, __be32 w2,
658 __be32 w3, __be32 w4)
660 __ipv6_addr_set_half(&addr->s6_addr32[0], w1, w2);
661 __ipv6_addr_set_half(&addr->s6_addr32[2], w3, w4);
664 static inline bool ipv6_addr_equal(const struct in6_addr *a1,
665 const struct in6_addr *a2)
667 #if defined(CONFIG_HAVE_EFFICIENT_UNALIGNED_ACCESS) && BITS_PER_LONG == 64
668 const unsigned long *ul1 = (const unsigned long *)a1;
669 const unsigned long *ul2 = (const unsigned long *)a2;
671 return ((ul1[0] ^ ul2[0]) | (ul1[1] ^ ul2[1])) == 0UL;
673 return ((a1->s6_addr32[0] ^ a2->s6_addr32[0]) |
674 (a1->s6_addr32[1] ^ a2->s6_addr32[1]) |
675 (a1->s6_addr32[2] ^ a2->s6_addr32[2]) |
676 (a1->s6_addr32[3] ^ a2->s6_addr32[3])) == 0;
680 #if defined(CONFIG_HAVE_EFFICIENT_UNALIGNED_ACCESS) && BITS_PER_LONG == 64
681 static inline bool __ipv6_prefix_equal64_half(const __be64 *a1,
685 if (len && ((*a1 ^ *a2) & cpu_to_be64((~0UL) << (64 - len))))
690 static inline bool ipv6_prefix_equal(const struct in6_addr *addr1,
691 const struct in6_addr *addr2,
692 unsigned int prefixlen)
694 const __be64 *a1 = (const __be64 *)addr1;
695 const __be64 *a2 = (const __be64 *)addr2;
697 if (prefixlen >= 64) {
700 return __ipv6_prefix_equal64_half(a1 + 1, a2 + 1, prefixlen - 64);
702 return __ipv6_prefix_equal64_half(a1, a2, prefixlen);
705 static inline bool ipv6_prefix_equal(const struct in6_addr *addr1,
706 const struct in6_addr *addr2,
707 unsigned int prefixlen)
709 const __be32 *a1 = addr1->s6_addr32;
710 const __be32 *a2 = addr2->s6_addr32;
711 unsigned int pdw, pbi;
713 /* check complete u32 in prefix */
714 pdw = prefixlen >> 5;
715 if (pdw && memcmp(a1, a2, pdw << 2))
718 /* check incomplete u32 in prefix */
719 pbi = prefixlen & 0x1f;
720 if (pbi && ((a1[pdw] ^ a2[pdw]) & htonl((0xffffffff) << (32 - pbi))))
727 static inline bool ipv6_addr_any(const struct in6_addr *a)
729 #if defined(CONFIG_HAVE_EFFICIENT_UNALIGNED_ACCESS) && BITS_PER_LONG == 64
730 const unsigned long *ul = (const unsigned long *)a;
732 return (ul[0] | ul[1]) == 0UL;
734 return (a->s6_addr32[0] | a->s6_addr32[1] |
735 a->s6_addr32[2] | a->s6_addr32[3]) == 0;
739 static inline u32 ipv6_addr_hash(const struct in6_addr *a)
741 #if defined(CONFIG_HAVE_EFFICIENT_UNALIGNED_ACCESS) && BITS_PER_LONG == 64
742 const unsigned long *ul = (const unsigned long *)a;
743 unsigned long x = ul[0] ^ ul[1];
745 return (u32)(x ^ (x >> 32));
747 return (__force u32)(a->s6_addr32[0] ^ a->s6_addr32[1] ^
748 a->s6_addr32[2] ^ a->s6_addr32[3]);
752 /* more secured version of ipv6_addr_hash() */
753 static inline u32 __ipv6_addr_jhash(const struct in6_addr *a, const u32 initval)
755 return jhash2((__force const u32 *)a->s6_addr32,
756 ARRAY_SIZE(a->s6_addr32), initval);
759 static inline bool ipv6_addr_loopback(const struct in6_addr *a)
761 #if defined(CONFIG_HAVE_EFFICIENT_UNALIGNED_ACCESS) && BITS_PER_LONG == 64
762 const __be64 *be = (const __be64 *)a;
764 return (be[0] | (be[1] ^ cpu_to_be64(1))) == 0UL;
766 return (a->s6_addr32[0] | a->s6_addr32[1] |
767 a->s6_addr32[2] | (a->s6_addr32[3] ^ cpu_to_be32(1))) == 0;
772 * Note that we must __force cast these to unsigned long to make sparse happy,
773 * since all of the endian-annotated types are fixed size regardless of arch.
775 static inline bool ipv6_addr_v4mapped(const struct in6_addr *a)
778 #if defined(CONFIG_HAVE_EFFICIENT_UNALIGNED_ACCESS) && BITS_PER_LONG == 64
779 *(unsigned long *)a |
781 (__force unsigned long)(a->s6_addr32[0] | a->s6_addr32[1]) |
783 (__force unsigned long)(a->s6_addr32[2] ^
784 cpu_to_be32(0x0000ffff))) == 0UL;
787 static inline bool ipv6_addr_v4mapped_any(const struct in6_addr *a)
789 return ipv6_addr_v4mapped(a) && ipv4_is_zeronet(a->s6_addr32[3]);
792 static inline bool ipv6_addr_v4mapped_loopback(const struct in6_addr *a)
794 return ipv6_addr_v4mapped(a) && ipv4_is_loopback(a->s6_addr32[3]);
797 static inline u32 ipv6_portaddr_hash(const struct net *net,
798 const struct in6_addr *addr6,
801 unsigned int hash, mix = net_hash_mix(net);
803 if (ipv6_addr_any(addr6))
804 hash = jhash_1word(0, mix);
805 else if (ipv6_addr_v4mapped(addr6))
806 hash = jhash_1word((__force u32)addr6->s6_addr32[3], mix);
808 hash = jhash2((__force u32 *)addr6->s6_addr32, 4, mix);
814 * Check for a RFC 4843 ORCHID address
815 * (Overlay Routable Cryptographic Hash Identifiers)
817 static inline bool ipv6_addr_orchid(const struct in6_addr *a)
819 return (a->s6_addr32[0] & htonl(0xfffffff0)) == htonl(0x20010010);
822 static inline bool ipv6_addr_is_multicast(const struct in6_addr *addr)
824 return (addr->s6_addr32[0] & htonl(0xFF000000)) == htonl(0xFF000000);
827 static inline void ipv6_addr_set_v4mapped(const __be32 addr,
828 struct in6_addr *v4mapped)
830 ipv6_addr_set(v4mapped,
837 * find the first different bit between two addresses
838 * length of address must be a multiple of 32bits
840 static inline int __ipv6_addr_diff32(const void *token1, const void *token2, int addrlen)
842 const __be32 *a1 = token1, *a2 = token2;
847 for (i = 0; i < addrlen; i++) {
848 __be32 xb = a1[i] ^ a2[i];
850 return i * 32 + 31 - __fls(ntohl(xb));
854 * we should *never* get to this point since that
855 * would mean the addrs are equal
857 * However, we do get to it 8) And exacly, when
858 * addresses are equal 8)
860 * ip route add 1111::/128 via ...
861 * ip route add 1111::/64 via ...
864 * Ideally, this function should stop comparison
865 * at prefix length. It does not, but it is still OK,
866 * if returned value is greater than prefix length.
872 #if defined(CONFIG_HAVE_EFFICIENT_UNALIGNED_ACCESS) && BITS_PER_LONG == 64
873 static inline int __ipv6_addr_diff64(const void *token1, const void *token2, int addrlen)
875 const __be64 *a1 = token1, *a2 = token2;
880 for (i = 0; i < addrlen; i++) {
881 __be64 xb = a1[i] ^ a2[i];
883 return i * 64 + 63 - __fls(be64_to_cpu(xb));
890 static inline int __ipv6_addr_diff(const void *token1, const void *token2, int addrlen)
892 #if defined(CONFIG_HAVE_EFFICIENT_UNALIGNED_ACCESS) && BITS_PER_LONG == 64
893 if (__builtin_constant_p(addrlen) && !(addrlen & 7))
894 return __ipv6_addr_diff64(token1, token2, addrlen);
896 return __ipv6_addr_diff32(token1, token2, addrlen);
899 static inline int ipv6_addr_diff(const struct in6_addr *a1, const struct in6_addr *a2)
901 return __ipv6_addr_diff(a1, a2, sizeof(struct in6_addr));
904 __be32 ipv6_select_ident(struct net *net,
905 const struct in6_addr *daddr,
906 const struct in6_addr *saddr);
907 __be32 ipv6_proxy_select_ident(struct net *net, struct sk_buff *skb);
909 int ip6_dst_hoplimit(struct dst_entry *dst);
911 static inline int ip6_sk_dst_hoplimit(struct ipv6_pinfo *np, struct flowi6 *fl6,
912 struct dst_entry *dst)
916 if (ipv6_addr_is_multicast(&fl6->daddr))
917 hlimit = np->mcast_hops;
919 hlimit = np->hop_limit;
921 hlimit = ip6_dst_hoplimit(dst);
925 /* copy IPv6 saddr & daddr to flow_keys, possibly using 64bit load/store
926 * Equivalent to : flow->v6addrs.src = iph->saddr;
927 * flow->v6addrs.dst = iph->daddr;
929 static inline void iph_to_flow_copy_v6addrs(struct flow_keys *flow,
930 const struct ipv6hdr *iph)
932 BUILD_BUG_ON(offsetof(typeof(flow->addrs), v6addrs.dst) !=
933 offsetof(typeof(flow->addrs), v6addrs.src) +
934 sizeof(flow->addrs.v6addrs.src));
935 memcpy(&flow->addrs.v6addrs, &iph->addrs, sizeof(flow->addrs.v6addrs));
936 flow->control.addr_type = FLOW_DISSECTOR_KEY_IPV6_ADDRS;
939 #if IS_ENABLED(CONFIG_IPV6)
941 static inline bool ipv6_can_nonlocal_bind(struct net *net,
942 struct inet_sock *inet)
944 return net->ipv6.sysctl.ip_nonlocal_bind ||
945 test_bit(INET_FLAGS_FREEBIND, &inet->inet_flags) ||
946 test_bit(INET_FLAGS_TRANSPARENT, &inet->inet_flags);
949 /* Sysctl settings for net ipv6.auto_flowlabels */
950 #define IP6_AUTO_FLOW_LABEL_OFF 0
951 #define IP6_AUTO_FLOW_LABEL_OPTOUT 1
952 #define IP6_AUTO_FLOW_LABEL_OPTIN 2
953 #define IP6_AUTO_FLOW_LABEL_FORCED 3
955 #define IP6_AUTO_FLOW_LABEL_MAX IP6_AUTO_FLOW_LABEL_FORCED
957 #define IP6_DEFAULT_AUTO_FLOW_LABELS IP6_AUTO_FLOW_LABEL_OPTOUT
959 static inline __be32 ip6_make_flowlabel(struct net *net, struct sk_buff *skb,
960 __be32 flowlabel, bool autolabel,
965 /* @flowlabel may include more than a flow label, eg, the traffic class.
966 * Here we want only the flow label value.
968 flowlabel &= IPV6_FLOWLABEL_MASK;
971 net->ipv6.sysctl.auto_flowlabels == IP6_AUTO_FLOW_LABEL_OFF ||
973 net->ipv6.sysctl.auto_flowlabels != IP6_AUTO_FLOW_LABEL_FORCED))
976 hash = skb_get_hash_flowi6(skb, fl6);
978 /* Since this is being sent on the wire obfuscate hash a bit
979 * to minimize possbility that any useful information to an
980 * attacker is leaked. Only lower 20 bits are relevant.
982 hash = rol32(hash, 16);
984 flowlabel = (__force __be32)hash & IPV6_FLOWLABEL_MASK;
986 if (net->ipv6.sysctl.flowlabel_state_ranges)
987 flowlabel |= IPV6_FLOWLABEL_STATELESS_FLAG;
992 static inline int ip6_default_np_autolabel(struct net *net)
994 switch (net->ipv6.sysctl.auto_flowlabels) {
995 case IP6_AUTO_FLOW_LABEL_OFF:
996 case IP6_AUTO_FLOW_LABEL_OPTIN:
999 case IP6_AUTO_FLOW_LABEL_OPTOUT:
1000 case IP6_AUTO_FLOW_LABEL_FORCED:
1005 static inline __be32 ip6_make_flowlabel(struct net *net, struct sk_buff *skb,
1006 __be32 flowlabel, bool autolabel,
1011 static inline int ip6_default_np_autolabel(struct net *net)
1017 #if IS_ENABLED(CONFIG_IPV6)
1018 static inline int ip6_multipath_hash_policy(const struct net *net)
1020 return net->ipv6.sysctl.multipath_hash_policy;
1022 static inline u32 ip6_multipath_hash_fields(const struct net *net)
1024 return net->ipv6.sysctl.multipath_hash_fields;
1027 static inline int ip6_multipath_hash_policy(const struct net *net)
1031 static inline u32 ip6_multipath_hash_fields(const struct net *net)
1038 * Header manipulation
1040 static inline void ip6_flow_hdr(struct ipv6hdr *hdr, unsigned int tclass,
1043 *(__be32 *)hdr = htonl(0x60000000 | (tclass << 20)) | flowlabel;
1046 static inline __be32 ip6_flowinfo(const struct ipv6hdr *hdr)
1048 return *(__be32 *)hdr & IPV6_FLOWINFO_MASK;
1051 static inline __be32 ip6_flowlabel(const struct ipv6hdr *hdr)
1053 return *(__be32 *)hdr & IPV6_FLOWLABEL_MASK;
1056 static inline u8 ip6_tclass(__be32 flowinfo)
1058 return ntohl(flowinfo & IPV6_TCLASS_MASK) >> IPV6_TCLASS_SHIFT;
1061 static inline dscp_t ip6_dscp(__be32 flowinfo)
1063 return inet_dsfield_to_dscp(ip6_tclass(flowinfo));
1066 static inline __be32 ip6_make_flowinfo(unsigned int tclass, __be32 flowlabel)
1068 return htonl(tclass << IPV6_TCLASS_SHIFT) | flowlabel;
1071 static inline __be32 flowi6_get_flowlabel(const struct flowi6 *fl6)
1073 return fl6->flowlabel & IPV6_FLOWLABEL_MASK;
1077 * Prototypes exported by ipv6
1081 * rcv function (called from netdevice level)
1084 int ipv6_rcv(struct sk_buff *skb, struct net_device *dev,
1085 struct packet_type *pt, struct net_device *orig_dev);
1086 void ipv6_list_rcv(struct list_head *head, struct packet_type *pt,
1087 struct net_device *orig_dev);
1089 int ip6_rcv_finish(struct net *net, struct sock *sk, struct sk_buff *skb);
1092 * upper-layer output functions
1094 int ip6_xmit(const struct sock *sk, struct sk_buff *skb, struct flowi6 *fl6,
1095 __u32 mark, struct ipv6_txoptions *opt, int tclass, u32 priority);
1097 int ip6_find_1stfragopt(struct sk_buff *skb, u8 **nexthdr);
1099 int ip6_append_data(struct sock *sk,
1100 int getfrag(void *from, char *to, int offset, int len,
1101 int odd, struct sk_buff *skb),
1102 void *from, size_t length, int transhdrlen,
1103 struct ipcm6_cookie *ipc6, struct flowi6 *fl6,
1104 struct rt6_info *rt, unsigned int flags);
1106 int ip6_push_pending_frames(struct sock *sk);
1108 void ip6_flush_pending_frames(struct sock *sk);
1110 int ip6_send_skb(struct sk_buff *skb);
1112 struct sk_buff *__ip6_make_skb(struct sock *sk, struct sk_buff_head *queue,
1113 struct inet_cork_full *cork,
1114 struct inet6_cork *v6_cork);
1115 struct sk_buff *ip6_make_skb(struct sock *sk,
1116 int getfrag(void *from, char *to, int offset,
1117 int len, int odd, struct sk_buff *skb),
1118 void *from, size_t length, int transhdrlen,
1119 struct ipcm6_cookie *ipc6,
1120 struct rt6_info *rt, unsigned int flags,
1121 struct inet_cork_full *cork);
1123 static inline struct sk_buff *ip6_finish_skb(struct sock *sk)
1125 return __ip6_make_skb(sk, &sk->sk_write_queue, &inet_sk(sk)->cork,
1126 &inet6_sk(sk)->cork);
1129 int ip6_dst_lookup(struct net *net, struct sock *sk, struct dst_entry **dst,
1130 struct flowi6 *fl6);
1131 struct dst_entry *ip6_dst_lookup_flow(struct net *net, const struct sock *sk, struct flowi6 *fl6,
1132 const struct in6_addr *final_dst);
1133 struct dst_entry *ip6_sk_dst_lookup_flow(struct sock *sk, struct flowi6 *fl6,
1134 const struct in6_addr *final_dst,
1136 struct dst_entry *ip6_dst_lookup_tunnel(struct sk_buff *skb,
1137 struct net_device *dev,
1138 struct net *net, struct socket *sock,
1139 struct in6_addr *saddr,
1140 const struct ip_tunnel_info *info,
1141 u8 protocol, bool use_cache);
1142 struct dst_entry *ip6_blackhole_route(struct net *net,
1143 struct dst_entry *orig_dst);
1146 * skb processing functions
1149 int ip6_output(struct net *net, struct sock *sk, struct sk_buff *skb);
1150 int ip6_forward(struct sk_buff *skb);
1151 int ip6_input(struct sk_buff *skb);
1152 int ip6_mc_input(struct sk_buff *skb);
1153 void ip6_protocol_deliver_rcu(struct net *net, struct sk_buff *skb, int nexthdr,
1156 int __ip6_local_out(struct net *net, struct sock *sk, struct sk_buff *skb);
1157 int ip6_local_out(struct net *net, struct sock *sk, struct sk_buff *skb);
1160 * Extension header (options) processing
1163 void ipv6_push_nfrag_opts(struct sk_buff *skb, struct ipv6_txoptions *opt,
1164 u8 *proto, struct in6_addr **daddr_p,
1165 struct in6_addr *saddr);
1166 void ipv6_push_frag_opts(struct sk_buff *skb, struct ipv6_txoptions *opt,
1169 int ipv6_skip_exthdr(const struct sk_buff *, int start, u8 *nexthdrp,
1172 bool ipv6_ext_hdr(u8 nexthdr);
1175 IP6_FH_F_FRAG = (1 << 0),
1176 IP6_FH_F_AUTH = (1 << 1),
1177 IP6_FH_F_SKIP_RH = (1 << 2),
1180 /* find specified header and get offset to it */
1181 int ipv6_find_hdr(const struct sk_buff *skb, unsigned int *offset, int target,
1182 unsigned short *fragoff, int *fragflg);
1184 int ipv6_find_tlv(const struct sk_buff *skb, int offset, int type);
1186 struct in6_addr *fl6_update_dst(struct flowi6 *fl6,
1187 const struct ipv6_txoptions *opt,
1188 struct in6_addr *orig);
1191 * socket options (ipv6_sockglue.c)
1193 DECLARE_STATIC_KEY_FALSE(ip6_min_hopcount);
1195 int do_ipv6_setsockopt(struct sock *sk, int level, int optname, sockptr_t optval,
1196 unsigned int optlen);
1197 int ipv6_setsockopt(struct sock *sk, int level, int optname, sockptr_t optval,
1198 unsigned int optlen);
1199 int do_ipv6_getsockopt(struct sock *sk, int level, int optname,
1200 sockptr_t optval, sockptr_t optlen);
1201 int ipv6_getsockopt(struct sock *sk, int level, int optname,
1202 char __user *optval, int __user *optlen);
1204 int __ip6_datagram_connect(struct sock *sk, struct sockaddr *addr,
1206 int ip6_datagram_connect(struct sock *sk, struct sockaddr *addr, int addr_len);
1207 int ip6_datagram_connect_v6_only(struct sock *sk, struct sockaddr *addr,
1209 int ip6_datagram_dst_update(struct sock *sk, bool fix_sk_saddr);
1210 void ip6_datagram_release_cb(struct sock *sk);
1212 int ipv6_recv_error(struct sock *sk, struct msghdr *msg, int len,
1214 int ipv6_recv_rxpmtu(struct sock *sk, struct msghdr *msg, int len,
1216 void ipv6_icmp_error(struct sock *sk, struct sk_buff *skb, int err, __be16 port,
1217 u32 info, u8 *payload);
1218 void ipv6_local_error(struct sock *sk, int err, struct flowi6 *fl6, u32 info);
1219 void ipv6_local_rxpmtu(struct sock *sk, struct flowi6 *fl6, u32 mtu);
1221 void inet6_cleanup_sock(struct sock *sk);
1222 void inet6_sock_destruct(struct sock *sk);
1223 int inet6_release(struct socket *sock);
1224 int inet6_bind(struct socket *sock, struct sockaddr *uaddr, int addr_len);
1225 int inet6_bind_sk(struct sock *sk, struct sockaddr *uaddr, int addr_len);
1226 int inet6_getname(struct socket *sock, struct sockaddr *uaddr,
1228 int inet6_ioctl(struct socket *sock, unsigned int cmd, unsigned long arg);
1229 int inet6_compat_ioctl(struct socket *sock, unsigned int cmd,
1232 int inet6_hash_connect(struct inet_timewait_death_row *death_row,
1234 int inet6_sendmsg(struct socket *sock, struct msghdr *msg, size_t size);
1235 int inet6_recvmsg(struct socket *sock, struct msghdr *msg, size_t size,
1241 extern const struct proto_ops inet6_stream_ops;
1242 extern const struct proto_ops inet6_dgram_ops;
1243 extern const struct proto_ops inet6_sockraw_ops;
1245 struct group_source_req;
1246 struct group_filter;
1248 int ip6_mc_source(int add, int omode, struct sock *sk,
1249 struct group_source_req *pgsr);
1250 int ip6_mc_msfilter(struct sock *sk, struct group_filter *gsf,
1251 struct sockaddr_storage *list);
1252 int ip6_mc_msfget(struct sock *sk, struct group_filter *gsf,
1253 sockptr_t optval, size_t ss_offset);
1255 #ifdef CONFIG_PROC_FS
1256 int ac6_proc_init(struct net *net);
1257 void ac6_proc_exit(struct net *net);
1258 int raw6_proc_init(void);
1259 void raw6_proc_exit(void);
1260 int tcp6_proc_init(struct net *net);
1261 void tcp6_proc_exit(struct net *net);
1262 int udp6_proc_init(struct net *net);
1263 void udp6_proc_exit(struct net *net);
1264 int udplite6_proc_init(void);
1265 void udplite6_proc_exit(void);
1266 int ipv6_misc_proc_init(void);
1267 void ipv6_misc_proc_exit(void);
1268 int snmp6_register_dev(struct inet6_dev *idev);
1269 int snmp6_unregister_dev(struct inet6_dev *idev);
1272 static inline int ac6_proc_init(struct net *net) { return 0; }
1273 static inline void ac6_proc_exit(struct net *net) { }
1274 static inline int snmp6_register_dev(struct inet6_dev *idev) { return 0; }
1275 static inline int snmp6_unregister_dev(struct inet6_dev *idev) { return 0; }
1278 #ifdef CONFIG_SYSCTL
1279 struct ctl_table *ipv6_icmp_sysctl_init(struct net *net);
1280 size_t ipv6_icmp_sysctl_table_size(void);
1281 struct ctl_table *ipv6_route_sysctl_init(struct net *net);
1282 size_t ipv6_route_sysctl_table_size(struct net *net);
1283 int ipv6_sysctl_register(void);
1284 void ipv6_sysctl_unregister(void);
1287 int ipv6_sock_mc_join(struct sock *sk, int ifindex,
1288 const struct in6_addr *addr);
1289 int ipv6_sock_mc_join_ssm(struct sock *sk, int ifindex,
1290 const struct in6_addr *addr, unsigned int mode);
1291 int ipv6_sock_mc_drop(struct sock *sk, int ifindex,
1292 const struct in6_addr *addr);
1294 static inline int ip6_sock_set_v6only(struct sock *sk)
1296 if (inet_sk(sk)->inet_num)
1299 sk->sk_ipv6only = true;
1304 static inline void ip6_sock_set_recverr(struct sock *sk)
1307 inet6_sk(sk)->recverr = true;
1311 static inline int __ip6_sock_set_addr_preferences(struct sock *sk, int val)
1313 unsigned int pref = 0;
1314 unsigned int prefmask = ~0;
1316 /* check PUBLIC/TMP/PUBTMP_DEFAULT conflicts */
1317 switch (val & (IPV6_PREFER_SRC_PUBLIC |
1318 IPV6_PREFER_SRC_TMP |
1319 IPV6_PREFER_SRC_PUBTMP_DEFAULT)) {
1320 case IPV6_PREFER_SRC_PUBLIC:
1321 pref |= IPV6_PREFER_SRC_PUBLIC;
1322 prefmask &= ~(IPV6_PREFER_SRC_PUBLIC |
1323 IPV6_PREFER_SRC_TMP);
1325 case IPV6_PREFER_SRC_TMP:
1326 pref |= IPV6_PREFER_SRC_TMP;
1327 prefmask &= ~(IPV6_PREFER_SRC_PUBLIC |
1328 IPV6_PREFER_SRC_TMP);
1330 case IPV6_PREFER_SRC_PUBTMP_DEFAULT:
1331 prefmask &= ~(IPV6_PREFER_SRC_PUBLIC |
1332 IPV6_PREFER_SRC_TMP);
1340 /* check HOME/COA conflicts */
1341 switch (val & (IPV6_PREFER_SRC_HOME | IPV6_PREFER_SRC_COA)) {
1342 case IPV6_PREFER_SRC_HOME:
1343 prefmask &= ~IPV6_PREFER_SRC_COA;
1345 case IPV6_PREFER_SRC_COA:
1346 pref |= IPV6_PREFER_SRC_COA;
1354 /* check CGA/NONCGA conflicts */
1355 switch (val & (IPV6_PREFER_SRC_CGA|IPV6_PREFER_SRC_NONCGA)) {
1356 case IPV6_PREFER_SRC_CGA:
1357 case IPV6_PREFER_SRC_NONCGA:
1364 inet6_sk(sk)->srcprefs = (inet6_sk(sk)->srcprefs & prefmask) | pref;
1368 static inline int ip6_sock_set_addr_preferences(struct sock *sk, int val)
1373 ret = __ip6_sock_set_addr_preferences(sk, val);
1378 static inline void ip6_sock_set_recvpktinfo(struct sock *sk)
1381 inet6_sk(sk)->rxopt.bits.rxinfo = true;
1385 #endif /* _NET_IPV6_H */