]> git.proxmox.com Git - mirror_ubuntu-bionic-kernel.git/blob - include/net/ipv6.h
inet: make sure to grab rcu_read_lock before using ireq->ireq_opt
[mirror_ubuntu-bionic-kernel.git] / include / net / ipv6.h
1 /*
2 * Linux INET6 implementation
3 *
4 * Authors:
5 * Pedro Roque <roque@di.fc.ul.pt>
6 *
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.
11 */
12
13 #ifndef _NET_IPV6_H
14 #define _NET_IPV6_H
15
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>
22 #include <net/flow.h>
23 #include <net/flow_dissector.h>
24 #include <net/snmp.h>
25
26 #define SIN6_LEN_RFC2133 24
27
28 #define IPV6_MAXPLEN 65535
29
30 /*
31 * NextHeader field of IPv6 header
32 */
33
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. */
48
49 #define NEXTHDR_MAX 255
50
51 #define IPV6_DEFAULT_HOPLIMIT 64
52 #define IPV6_DEFAULT_MCASTHOPS 1
53
54 /* Limits on Hop-by-Hop and Destination options.
55 *
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
59 * service attacks.
60 *
61 * There are three limits that may be set:
62 * - Limit the number of options in a Hop-by-Hop or Destination options
63 * extension header
64 * - Limit the byte length of a Hop-by-Hop or Destination options extension
65 * header
66 * - Disallow unknown options
67 *
68 * The limits are expressed in corresponding sysctls:
69 *
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
74 *
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.
79 *
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.
83 *
84 * If a limit is exceeded when processing an extension header the packet is
85 * silently discarded.
86 */
87
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 */
93
94 /*
95 * Addr type
96 *
97 * type - unicast | multicast
98 * scope - local | site | global
99 * v4 - compat
100 * v4mapped
101 * any
102 * loopback
103 */
104
105 #define IPV6_ADDR_ANY 0x0000U
106
107 #define IPV6_ADDR_UNICAST 0x0001U
108 #define IPV6_ADDR_MULTICAST 0x0002U
109
110 #define IPV6_ADDR_LOOPBACK 0x0010U
111 #define IPV6_ADDR_LINKLOCAL 0x0020U
112 #define IPV6_ADDR_SITELOCAL 0x0040U
113
114 #define IPV6_ADDR_COMPATv4 0x0080U
115
116 #define IPV6_ADDR_SCOPE_MASK 0x00f0U
117
118 #define IPV6_ADDR_MAPPED 0x1000U
119
120 /*
121 * Addr scopes
122 */
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
131
132 /*
133 * Addr flags
134 */
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)
141
142 /*
143 * fragmentation header
144 */
145
146 struct frag_hdr {
147 __u8 nexthdr;
148 __u8 reserved;
149 __be16 frag_off;
150 __be32 identification;
151 };
152
153 #define IP6_MF 0x0001
154 #define IP6_OFFSET 0xFFF8
155
156 #define IP6_REPLY_MARK(net, mark) \
157 ((net)->ipv6.sysctl.fwmark_reflect ? (mark) : 0)
158
159 #include <net/sock.h>
160
161 /* sysctls */
162 extern int sysctl_mld_max_msf;
163 extern int sysctl_mld_qrv;
164
165 #define _DEVINC(net, statname, mod, idev, field) \
166 ({ \
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));\
171 })
172
173 /* per device counters are atomic_long_t */
174 #define _DEVINCATOMIC(net, statname, mod, idev, field) \
175 ({ \
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));\
180 })
181
182 /* per device and per net counters are atomic_long_t */
183 #define _DEVINC_ATOMIC_ATOMIC(net, statname, idev, field) \
184 ({ \
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));\
189 })
190
191 #define _DEVADD(net, statname, mod, idev, field, val) \
192 ({ \
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));\
197 })
198
199 #define _DEVUPD(net, statname, mod, idev, field, val) \
200 ({ \
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));\
205 })
206
207 /* MIBs */
208
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)
225
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)
230
231 struct ip6_ra_chain {
232 struct ip6_ra_chain *next;
233 struct sock *sk;
234 int sel;
235 void (*destructor)(struct sock *);
236 };
237
238 extern struct ip6_ra_chain *ip6_ra_chain;
239 extern rwlock_t ip6_ra_lock;
240
241 /*
242 This structure is prepared by protocol, when parsing
243 ancillary data and passed to IPv6.
244 */
245
246 struct ipv6_txoptions {
247 refcount_t refcnt;
248 /* Length of this structure */
249 int tot_len;
250
251 /* length of extension headers */
252
253 __u16 opt_flen; /* after fragment hdr */
254 __u16 opt_nflen; /* before fragment hdr */
255
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;
260 struct rcu_head rcu;
261 /* Option buffer, as read by IPV6_PKTOPTIONS, starts here. */
262 };
263
264 struct ip6_flowlabel {
265 struct ip6_flowlabel __rcu *next;
266 __be32 label;
267 atomic_t users;
268 struct in6_addr dst;
269 struct ipv6_txoptions *opt;
270 unsigned long linger;
271 struct rcu_head rcu;
272 u8 share;
273 union {
274 struct pid *pid;
275 kuid_t uid;
276 } owner;
277 unsigned long lastuse;
278 unsigned long expires;
279 struct net *fl_net;
280 };
281
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)
285
286 #define IPV6_TCLASS_MASK (IPV6_FLOWINFO_MASK & ~IPV6_FLOWLABEL_MASK)
287 #define IPV6_TCLASS_SHIFT 20
288
289 struct ipv6_fl_socklist {
290 struct ipv6_fl_socklist __rcu *next;
291 struct ip6_flowlabel *fl;
292 struct rcu_head rcu;
293 };
294
295 struct ipcm6_cookie {
296 __s16 hlimit;
297 __s16 tclass;
298 __s8 dontfrag;
299 struct ipv6_txoptions *opt;
300 };
301
302 static inline struct ipv6_txoptions *txopt_get(const struct ipv6_pinfo *np)
303 {
304 struct ipv6_txoptions *opt;
305
306 rcu_read_lock();
307 opt = rcu_dereference(np->opt);
308 if (opt) {
309 if (!refcount_inc_not_zero(&opt->refcnt))
310 opt = NULL;
311 else
312 opt = rcu_pointer_handoff(opt);
313 }
314 rcu_read_unlock();
315 return opt;
316 }
317
318 static inline void txopt_put(struct ipv6_txoptions *opt)
319 {
320 if (opt && refcount_dec_and_test(&opt->refcnt))
321 kfree_rcu(opt, rcu);
322 }
323
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,
331 int flags);
332 int ip6_flowlabel_init(void);
333 void ip6_flowlabel_cleanup(void);
334 bool ip6_autoflowlabel(struct net *net, const struct ipv6_pinfo *np);
335
336 static inline void fl6_sock_release(struct ip6_flowlabel *fl)
337 {
338 if (fl)
339 atomic_dec(&fl->users);
340 }
341
342 void icmpv6_notify(struct sk_buff *skb, u8 type, u8 code, __be32 info);
343
344 void icmpv6_push_pending_frames(struct sock *sk, struct flowi6 *fl6,
345 struct icmp6hdr *thdr, int len);
346
347 int ip6_ra_control(struct sock *sk, int sel);
348
349 int ipv6_parse_hopopts(struct sk_buff *skb);
350
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,
355 int newtype,
356 struct ipv6_opt_hdr *newopt);
357 struct ipv6_txoptions *ipv6_fixup_options(struct ipv6_txoptions *opt_space,
358 struct ipv6_txoptions *opt);
359
360 bool ipv6_opt_accepted(const struct sock *sk, const struct sk_buff *skb,
361 const struct inet6_skb_parm *opt);
362 struct ipv6_txoptions *ipv6_update_options(struct sock *sk,
363 struct ipv6_txoptions *opt);
364
365 static inline bool ipv6_accept_ra(struct inet6_dev *idev)
366 {
367 /* If forwarding is enabled, RA are not accepted unless the special
368 * hybrid mode (accept_ra=2) is enabled.
369 */
370 return idev->cnf.forwarding ? idev->cnf.accept_ra == 2 :
371 idev->cnf.accept_ra;
372 }
373
374 #define IPV6_FRAG_HIGH_THRESH (256 * 1024) /* 262144 */
375 #define IPV6_FRAG_LOW_THRESH (192 * 1024) /* 196608 */
376 #define IPV6_FRAG_TIMEOUT (60 * HZ) /* 60 seconds */
377
378 int __ipv6_addr_type(const struct in6_addr *addr);
379 static inline int ipv6_addr_type(const struct in6_addr *addr)
380 {
381 return __ipv6_addr_type(addr) & 0xffff;
382 }
383
384 static inline int ipv6_addr_scope(const struct in6_addr *addr)
385 {
386 return __ipv6_addr_type(addr) & IPV6_ADDR_SCOPE_MASK;
387 }
388
389 static inline int __ipv6_addr_src_scope(int type)
390 {
391 return (type == IPV6_ADDR_ANY) ? __IPV6_ADDR_SCOPE_INVALID : (type >> 16);
392 }
393
394 static inline int ipv6_addr_src_scope(const struct in6_addr *addr)
395 {
396 return __ipv6_addr_src_scope(__ipv6_addr_type(addr));
397 }
398
399 static inline bool __ipv6_addr_needs_scope_id(int type)
400 {
401 return type & IPV6_ADDR_LINKLOCAL ||
402 (type & IPV6_ADDR_MULTICAST &&
403 (type & (IPV6_ADDR_LOOPBACK|IPV6_ADDR_LINKLOCAL)));
404 }
405
406 static inline __u32 ipv6_iface_scope_id(const struct in6_addr *addr, int iface)
407 {
408 return __ipv6_addr_needs_scope_id(__ipv6_addr_type(addr)) ? iface : 0;
409 }
410
411 static inline int ipv6_addr_cmp(const struct in6_addr *a1, const struct in6_addr *a2)
412 {
413 return memcmp(a1, a2, sizeof(struct in6_addr));
414 }
415
416 static inline bool
417 ipv6_masked_addr_cmp(const struct in6_addr *a1, const struct in6_addr *m,
418 const struct in6_addr *a2)
419 {
420 #if defined(CONFIG_HAVE_EFFICIENT_UNALIGNED_ACCESS) && BITS_PER_LONG == 64
421 const unsigned long *ul1 = (const unsigned long *)a1;
422 const unsigned long *ulm = (const unsigned long *)m;
423 const unsigned long *ul2 = (const unsigned long *)a2;
424
425 return !!(((ul1[0] ^ ul2[0]) & ulm[0]) |
426 ((ul1[1] ^ ul2[1]) & ulm[1]));
427 #else
428 return !!(((a1->s6_addr32[0] ^ a2->s6_addr32[0]) & m->s6_addr32[0]) |
429 ((a1->s6_addr32[1] ^ a2->s6_addr32[1]) & m->s6_addr32[1]) |
430 ((a1->s6_addr32[2] ^ a2->s6_addr32[2]) & m->s6_addr32[2]) |
431 ((a1->s6_addr32[3] ^ a2->s6_addr32[3]) & m->s6_addr32[3]));
432 #endif
433 }
434
435 static inline void ipv6_addr_prefix(struct in6_addr *pfx,
436 const struct in6_addr *addr,
437 int plen)
438 {
439 /* caller must guarantee 0 <= plen <= 128 */
440 int o = plen >> 3,
441 b = plen & 0x7;
442
443 memset(pfx->s6_addr, 0, sizeof(pfx->s6_addr));
444 memcpy(pfx->s6_addr, addr, o);
445 if (b != 0)
446 pfx->s6_addr[o] = addr->s6_addr[o] & (0xff00 >> b);
447 }
448
449 static inline void ipv6_addr_prefix_copy(struct in6_addr *addr,
450 const struct in6_addr *pfx,
451 int plen)
452 {
453 /* caller must guarantee 0 <= plen <= 128 */
454 int o = plen >> 3,
455 b = plen & 0x7;
456
457 memcpy(addr->s6_addr, pfx, o);
458 if (b != 0) {
459 addr->s6_addr[o] &= ~(0xff00 >> b);
460 addr->s6_addr[o] |= (pfx->s6_addr[o] & (0xff00 >> b));
461 }
462 }
463
464 static inline void __ipv6_addr_set_half(__be32 *addr,
465 __be32 wh, __be32 wl)
466 {
467 #if defined(CONFIG_HAVE_EFFICIENT_UNALIGNED_ACCESS) && BITS_PER_LONG == 64
468 #if defined(__BIG_ENDIAN)
469 if (__builtin_constant_p(wh) && __builtin_constant_p(wl)) {
470 *(__force u64 *)addr = ((__force u64)(wh) << 32 | (__force u64)(wl));
471 return;
472 }
473 #elif defined(__LITTLE_ENDIAN)
474 if (__builtin_constant_p(wl) && __builtin_constant_p(wh)) {
475 *(__force u64 *)addr = ((__force u64)(wl) << 32 | (__force u64)(wh));
476 return;
477 }
478 #endif
479 #endif
480 addr[0] = wh;
481 addr[1] = wl;
482 }
483
484 static inline void ipv6_addr_set(struct in6_addr *addr,
485 __be32 w1, __be32 w2,
486 __be32 w3, __be32 w4)
487 {
488 __ipv6_addr_set_half(&addr->s6_addr32[0], w1, w2);
489 __ipv6_addr_set_half(&addr->s6_addr32[2], w3, w4);
490 }
491
492 static inline bool ipv6_addr_equal(const struct in6_addr *a1,
493 const struct in6_addr *a2)
494 {
495 #if defined(CONFIG_HAVE_EFFICIENT_UNALIGNED_ACCESS) && BITS_PER_LONG == 64
496 const unsigned long *ul1 = (const unsigned long *)a1;
497 const unsigned long *ul2 = (const unsigned long *)a2;
498
499 return ((ul1[0] ^ ul2[0]) | (ul1[1] ^ ul2[1])) == 0UL;
500 #else
501 return ((a1->s6_addr32[0] ^ a2->s6_addr32[0]) |
502 (a1->s6_addr32[1] ^ a2->s6_addr32[1]) |
503 (a1->s6_addr32[2] ^ a2->s6_addr32[2]) |
504 (a1->s6_addr32[3] ^ a2->s6_addr32[3])) == 0;
505 #endif
506 }
507
508 #if defined(CONFIG_HAVE_EFFICIENT_UNALIGNED_ACCESS) && BITS_PER_LONG == 64
509 static inline bool __ipv6_prefix_equal64_half(const __be64 *a1,
510 const __be64 *a2,
511 unsigned int len)
512 {
513 if (len && ((*a1 ^ *a2) & cpu_to_be64((~0UL) << (64 - len))))
514 return false;
515 return true;
516 }
517
518 static inline bool ipv6_prefix_equal(const struct in6_addr *addr1,
519 const struct in6_addr *addr2,
520 unsigned int prefixlen)
521 {
522 const __be64 *a1 = (const __be64 *)addr1;
523 const __be64 *a2 = (const __be64 *)addr2;
524
525 if (prefixlen >= 64) {
526 if (a1[0] ^ a2[0])
527 return false;
528 return __ipv6_prefix_equal64_half(a1 + 1, a2 + 1, prefixlen - 64);
529 }
530 return __ipv6_prefix_equal64_half(a1, a2, prefixlen);
531 }
532 #else
533 static inline bool ipv6_prefix_equal(const struct in6_addr *addr1,
534 const struct in6_addr *addr2,
535 unsigned int prefixlen)
536 {
537 const __be32 *a1 = addr1->s6_addr32;
538 const __be32 *a2 = addr2->s6_addr32;
539 unsigned int pdw, pbi;
540
541 /* check complete u32 in prefix */
542 pdw = prefixlen >> 5;
543 if (pdw && memcmp(a1, a2, pdw << 2))
544 return false;
545
546 /* check incomplete u32 in prefix */
547 pbi = prefixlen & 0x1f;
548 if (pbi && ((a1[pdw] ^ a2[pdw]) & htonl((0xffffffff) << (32 - pbi))))
549 return false;
550
551 return true;
552 }
553 #endif
554
555 struct inet_frag_queue;
556
557 enum ip6_defrag_users {
558 IP6_DEFRAG_LOCAL_DELIVER,
559 IP6_DEFRAG_CONNTRACK_IN,
560 __IP6_DEFRAG_CONNTRACK_IN = IP6_DEFRAG_CONNTRACK_IN + USHRT_MAX,
561 IP6_DEFRAG_CONNTRACK_OUT,
562 __IP6_DEFRAG_CONNTRACK_OUT = IP6_DEFRAG_CONNTRACK_OUT + USHRT_MAX,
563 IP6_DEFRAG_CONNTRACK_BRIDGE_IN,
564 __IP6_DEFRAG_CONNTRACK_BRIDGE_IN = IP6_DEFRAG_CONNTRACK_BRIDGE_IN + USHRT_MAX,
565 };
566
567 void ip6_frag_init(struct inet_frag_queue *q, const void *a);
568 extern const struct rhashtable_params ip6_rhash_params;
569
570 /*
571 * Equivalent of ipv4 struct ip
572 */
573 struct frag_queue {
574 struct inet_frag_queue q;
575
576 int iif;
577 unsigned int csum;
578 __u16 nhoffset;
579 u8 ecn;
580 };
581
582 void ip6_expire_frag_queue(struct net *net, struct frag_queue *fq);
583
584 static inline bool ipv6_addr_any(const struct in6_addr *a)
585 {
586 #if defined(CONFIG_HAVE_EFFICIENT_UNALIGNED_ACCESS) && BITS_PER_LONG == 64
587 const unsigned long *ul = (const unsigned long *)a;
588
589 return (ul[0] | ul[1]) == 0UL;
590 #else
591 return (a->s6_addr32[0] | a->s6_addr32[1] |
592 a->s6_addr32[2] | a->s6_addr32[3]) == 0;
593 #endif
594 }
595
596 static inline u32 ipv6_addr_hash(const struct in6_addr *a)
597 {
598 #if defined(CONFIG_HAVE_EFFICIENT_UNALIGNED_ACCESS) && BITS_PER_LONG == 64
599 const unsigned long *ul = (const unsigned long *)a;
600 unsigned long x = ul[0] ^ ul[1];
601
602 return (u32)(x ^ (x >> 32));
603 #else
604 return (__force u32)(a->s6_addr32[0] ^ a->s6_addr32[1] ^
605 a->s6_addr32[2] ^ a->s6_addr32[3]);
606 #endif
607 }
608
609 /* more secured version of ipv6_addr_hash() */
610 static inline u32 __ipv6_addr_jhash(const struct in6_addr *a, const u32 initval)
611 {
612 u32 v = (__force u32)a->s6_addr32[0] ^ (__force u32)a->s6_addr32[1];
613
614 return jhash_3words(v,
615 (__force u32)a->s6_addr32[2],
616 (__force u32)a->s6_addr32[3],
617 initval);
618 }
619
620 static inline bool ipv6_addr_loopback(const struct in6_addr *a)
621 {
622 #if defined(CONFIG_HAVE_EFFICIENT_UNALIGNED_ACCESS) && BITS_PER_LONG == 64
623 const __be64 *be = (const __be64 *)a;
624
625 return (be[0] | (be[1] ^ cpu_to_be64(1))) == 0UL;
626 #else
627 return (a->s6_addr32[0] | a->s6_addr32[1] |
628 a->s6_addr32[2] | (a->s6_addr32[3] ^ cpu_to_be32(1))) == 0;
629 #endif
630 }
631
632 /*
633 * Note that we must __force cast these to unsigned long to make sparse happy,
634 * since all of the endian-annotated types are fixed size regardless of arch.
635 */
636 static inline bool ipv6_addr_v4mapped(const struct in6_addr *a)
637 {
638 return (
639 #if defined(CONFIG_HAVE_EFFICIENT_UNALIGNED_ACCESS) && BITS_PER_LONG == 64
640 *(unsigned long *)a |
641 #else
642 (__force unsigned long)(a->s6_addr32[0] | a->s6_addr32[1]) |
643 #endif
644 (__force unsigned long)(a->s6_addr32[2] ^
645 cpu_to_be32(0x0000ffff))) == 0UL;
646 }
647
648 /*
649 * Check for a RFC 4843 ORCHID address
650 * (Overlay Routable Cryptographic Hash Identifiers)
651 */
652 static inline bool ipv6_addr_orchid(const struct in6_addr *a)
653 {
654 return (a->s6_addr32[0] & htonl(0xfffffff0)) == htonl(0x20010010);
655 }
656
657 static inline bool ipv6_addr_is_multicast(const struct in6_addr *addr)
658 {
659 return (addr->s6_addr32[0] & htonl(0xFF000000)) == htonl(0xFF000000);
660 }
661
662 static inline void ipv6_addr_set_v4mapped(const __be32 addr,
663 struct in6_addr *v4mapped)
664 {
665 ipv6_addr_set(v4mapped,
666 0, 0,
667 htonl(0x0000FFFF),
668 addr);
669 }
670
671 /*
672 * find the first different bit between two addresses
673 * length of address must be a multiple of 32bits
674 */
675 static inline int __ipv6_addr_diff32(const void *token1, const void *token2, int addrlen)
676 {
677 const __be32 *a1 = token1, *a2 = token2;
678 int i;
679
680 addrlen >>= 2;
681
682 for (i = 0; i < addrlen; i++) {
683 __be32 xb = a1[i] ^ a2[i];
684 if (xb)
685 return i * 32 + 31 - __fls(ntohl(xb));
686 }
687
688 /*
689 * we should *never* get to this point since that
690 * would mean the addrs are equal
691 *
692 * However, we do get to it 8) And exacly, when
693 * addresses are equal 8)
694 *
695 * ip route add 1111::/128 via ...
696 * ip route add 1111::/64 via ...
697 * and we are here.
698 *
699 * Ideally, this function should stop comparison
700 * at prefix length. It does not, but it is still OK,
701 * if returned value is greater than prefix length.
702 * --ANK (980803)
703 */
704 return addrlen << 5;
705 }
706
707 #if defined(CONFIG_HAVE_EFFICIENT_UNALIGNED_ACCESS) && BITS_PER_LONG == 64
708 static inline int __ipv6_addr_diff64(const void *token1, const void *token2, int addrlen)
709 {
710 const __be64 *a1 = token1, *a2 = token2;
711 int i;
712
713 addrlen >>= 3;
714
715 for (i = 0; i < addrlen; i++) {
716 __be64 xb = a1[i] ^ a2[i];
717 if (xb)
718 return i * 64 + 63 - __fls(be64_to_cpu(xb));
719 }
720
721 return addrlen << 6;
722 }
723 #endif
724
725 static inline int __ipv6_addr_diff(const void *token1, const void *token2, int addrlen)
726 {
727 #if defined(CONFIG_HAVE_EFFICIENT_UNALIGNED_ACCESS) && BITS_PER_LONG == 64
728 if (__builtin_constant_p(addrlen) && !(addrlen & 7))
729 return __ipv6_addr_diff64(token1, token2, addrlen);
730 #endif
731 return __ipv6_addr_diff32(token1, token2, addrlen);
732 }
733
734 static inline int ipv6_addr_diff(const struct in6_addr *a1, const struct in6_addr *a2)
735 {
736 return __ipv6_addr_diff(a1, a2, sizeof(struct in6_addr));
737 }
738
739 __be32 ipv6_select_ident(struct net *net,
740 const struct in6_addr *daddr,
741 const struct in6_addr *saddr);
742 __be32 ipv6_proxy_select_ident(struct net *net, struct sk_buff *skb);
743
744 int ip6_dst_hoplimit(struct dst_entry *dst);
745
746 static inline int ip6_sk_dst_hoplimit(struct ipv6_pinfo *np, struct flowi6 *fl6,
747 struct dst_entry *dst)
748 {
749 int hlimit;
750
751 if (ipv6_addr_is_multicast(&fl6->daddr))
752 hlimit = np->mcast_hops;
753 else
754 hlimit = np->hop_limit;
755 if (hlimit < 0)
756 hlimit = ip6_dst_hoplimit(dst);
757 return hlimit;
758 }
759
760 /* copy IPv6 saddr & daddr to flow_keys, possibly using 64bit load/store
761 * Equivalent to : flow->v6addrs.src = iph->saddr;
762 * flow->v6addrs.dst = iph->daddr;
763 */
764 static inline void iph_to_flow_copy_v6addrs(struct flow_keys *flow,
765 const struct ipv6hdr *iph)
766 {
767 BUILD_BUG_ON(offsetof(typeof(flow->addrs), v6addrs.dst) !=
768 offsetof(typeof(flow->addrs), v6addrs.src) +
769 sizeof(flow->addrs.v6addrs.src));
770 memcpy(&flow->addrs.v6addrs, &iph->saddr, sizeof(flow->addrs.v6addrs));
771 flow->control.addr_type = FLOW_DISSECTOR_KEY_IPV6_ADDRS;
772 }
773
774 #if IS_ENABLED(CONFIG_IPV6)
775
776 /* Sysctl settings for net ipv6.auto_flowlabels */
777 #define IP6_AUTO_FLOW_LABEL_OFF 0
778 #define IP6_AUTO_FLOW_LABEL_OPTOUT 1
779 #define IP6_AUTO_FLOW_LABEL_OPTIN 2
780 #define IP6_AUTO_FLOW_LABEL_FORCED 3
781
782 #define IP6_AUTO_FLOW_LABEL_MAX IP6_AUTO_FLOW_LABEL_FORCED
783
784 #define IP6_DEFAULT_AUTO_FLOW_LABELS IP6_AUTO_FLOW_LABEL_OPTOUT
785
786 static inline __be32 ip6_make_flowlabel(struct net *net, struct sk_buff *skb,
787 __be32 flowlabel, bool autolabel,
788 struct flowi6 *fl6)
789 {
790 u32 hash;
791
792 /* @flowlabel may include more than a flow label, eg, the traffic class.
793 * Here we want only the flow label value.
794 */
795 flowlabel &= IPV6_FLOWLABEL_MASK;
796
797 if (flowlabel ||
798 net->ipv6.sysctl.auto_flowlabels == IP6_AUTO_FLOW_LABEL_OFF ||
799 (!autolabel &&
800 net->ipv6.sysctl.auto_flowlabels != IP6_AUTO_FLOW_LABEL_FORCED))
801 return flowlabel;
802
803 hash = skb_get_hash_flowi6(skb, fl6);
804
805 /* Since this is being sent on the wire obfuscate hash a bit
806 * to minimize possbility that any useful information to an
807 * attacker is leaked. Only lower 20 bits are relevant.
808 */
809 hash = rol32(hash, 16);
810
811 flowlabel = (__force __be32)hash & IPV6_FLOWLABEL_MASK;
812
813 if (net->ipv6.sysctl.flowlabel_state_ranges)
814 flowlabel |= IPV6_FLOWLABEL_STATELESS_FLAG;
815
816 return flowlabel;
817 }
818
819 static inline int ip6_default_np_autolabel(struct net *net)
820 {
821 switch (net->ipv6.sysctl.auto_flowlabels) {
822 case IP6_AUTO_FLOW_LABEL_OFF:
823 case IP6_AUTO_FLOW_LABEL_OPTIN:
824 default:
825 return 0;
826 case IP6_AUTO_FLOW_LABEL_OPTOUT:
827 case IP6_AUTO_FLOW_LABEL_FORCED:
828 return 1;
829 }
830 }
831 #else
832 static inline void ip6_set_txhash(struct sock *sk) { }
833 static inline __be32 ip6_make_flowlabel(struct net *net, struct sk_buff *skb,
834 __be32 flowlabel, bool autolabel,
835 struct flowi6 *fl6)
836 {
837 return flowlabel;
838 }
839 static inline int ip6_default_np_autolabel(struct net *net)
840 {
841 return 0;
842 }
843 #endif
844
845
846 /*
847 * Header manipulation
848 */
849 static inline void ip6_flow_hdr(struct ipv6hdr *hdr, unsigned int tclass,
850 __be32 flowlabel)
851 {
852 *(__be32 *)hdr = htonl(0x60000000 | (tclass << 20)) | flowlabel;
853 }
854
855 static inline __be32 ip6_flowinfo(const struct ipv6hdr *hdr)
856 {
857 return *(__be32 *)hdr & IPV6_FLOWINFO_MASK;
858 }
859
860 static inline __be32 ip6_flowlabel(const struct ipv6hdr *hdr)
861 {
862 return *(__be32 *)hdr & IPV6_FLOWLABEL_MASK;
863 }
864
865 static inline u8 ip6_tclass(__be32 flowinfo)
866 {
867 return ntohl(flowinfo & IPV6_TCLASS_MASK) >> IPV6_TCLASS_SHIFT;
868 }
869
870 static inline __be32 ip6_make_flowinfo(unsigned int tclass, __be32 flowlabel)
871 {
872 return htonl(tclass << IPV6_TCLASS_SHIFT) | flowlabel;
873 }
874
875 static inline __be32 flowi6_get_flowlabel(const struct flowi6 *fl6)
876 {
877 return fl6->flowlabel & IPV6_FLOWLABEL_MASK;
878 }
879
880 /*
881 * Prototypes exported by ipv6
882 */
883
884 /*
885 * rcv function (called from netdevice level)
886 */
887
888 int ipv6_rcv(struct sk_buff *skb, struct net_device *dev,
889 struct packet_type *pt, struct net_device *orig_dev);
890
891 int ip6_rcv_finish(struct net *net, struct sock *sk, struct sk_buff *skb);
892
893 /*
894 * upper-layer output functions
895 */
896 int ip6_xmit(const struct sock *sk, struct sk_buff *skb, struct flowi6 *fl6,
897 __u32 mark, struct ipv6_txoptions *opt, int tclass);
898
899 int ip6_find_1stfragopt(struct sk_buff *skb, u8 **nexthdr);
900
901 int ip6_append_data(struct sock *sk,
902 int getfrag(void *from, char *to, int offset, int len,
903 int odd, struct sk_buff *skb),
904 void *from, int length, int transhdrlen,
905 struct ipcm6_cookie *ipc6, struct flowi6 *fl6,
906 struct rt6_info *rt, unsigned int flags,
907 const struct sockcm_cookie *sockc);
908
909 int ip6_push_pending_frames(struct sock *sk);
910
911 void ip6_flush_pending_frames(struct sock *sk);
912
913 int ip6_send_skb(struct sk_buff *skb);
914
915 struct sk_buff *__ip6_make_skb(struct sock *sk, struct sk_buff_head *queue,
916 struct inet_cork_full *cork,
917 struct inet6_cork *v6_cork);
918 struct sk_buff *ip6_make_skb(struct sock *sk,
919 int getfrag(void *from, char *to, int offset,
920 int len, int odd, struct sk_buff *skb),
921 void *from, int length, int transhdrlen,
922 struct ipcm6_cookie *ipc6, struct flowi6 *fl6,
923 struct rt6_info *rt, unsigned int flags,
924 const struct sockcm_cookie *sockc);
925
926 static inline struct sk_buff *ip6_finish_skb(struct sock *sk)
927 {
928 return __ip6_make_skb(sk, &sk->sk_write_queue, &inet_sk(sk)->cork,
929 &inet6_sk(sk)->cork);
930 }
931
932 int ip6_dst_lookup(struct net *net, struct sock *sk, struct dst_entry **dst,
933 struct flowi6 *fl6);
934 struct dst_entry *ip6_dst_lookup_flow(const struct sock *sk, struct flowi6 *fl6,
935 const struct in6_addr *final_dst);
936 struct dst_entry *ip6_sk_dst_lookup_flow(struct sock *sk, struct flowi6 *fl6,
937 const struct in6_addr *final_dst);
938 struct dst_entry *ip6_blackhole_route(struct net *net,
939 struct dst_entry *orig_dst);
940
941 /*
942 * skb processing functions
943 */
944
945 int ip6_output(struct net *net, struct sock *sk, struct sk_buff *skb);
946 int ip6_forward(struct sk_buff *skb);
947 int ip6_input(struct sk_buff *skb);
948 int ip6_mc_input(struct sk_buff *skb);
949
950 int __ip6_local_out(struct net *net, struct sock *sk, struct sk_buff *skb);
951 int ip6_local_out(struct net *net, struct sock *sk, struct sk_buff *skb);
952
953 /*
954 * Extension header (options) processing
955 */
956
957 void ipv6_push_nfrag_opts(struct sk_buff *skb, struct ipv6_txoptions *opt,
958 u8 *proto, struct in6_addr **daddr_p,
959 struct in6_addr *saddr);
960 void ipv6_push_frag_opts(struct sk_buff *skb, struct ipv6_txoptions *opt,
961 u8 *proto);
962
963 int ipv6_skip_exthdr(const struct sk_buff *, int start, u8 *nexthdrp,
964 __be16 *frag_offp);
965
966 bool ipv6_ext_hdr(u8 nexthdr);
967
968 enum {
969 IP6_FH_F_FRAG = (1 << 0),
970 IP6_FH_F_AUTH = (1 << 1),
971 IP6_FH_F_SKIP_RH = (1 << 2),
972 };
973
974 /* find specified header and get offset to it */
975 int ipv6_find_hdr(const struct sk_buff *skb, unsigned int *offset, int target,
976 unsigned short *fragoff, int *fragflg);
977
978 int ipv6_find_tlv(const struct sk_buff *skb, int offset, int type);
979
980 struct in6_addr *fl6_update_dst(struct flowi6 *fl6,
981 const struct ipv6_txoptions *opt,
982 struct in6_addr *orig);
983
984 /*
985 * socket options (ipv6_sockglue.c)
986 */
987
988 int ipv6_setsockopt(struct sock *sk, int level, int optname,
989 char __user *optval, unsigned int optlen);
990 int ipv6_getsockopt(struct sock *sk, int level, int optname,
991 char __user *optval, int __user *optlen);
992 int compat_ipv6_setsockopt(struct sock *sk, int level, int optname,
993 char __user *optval, unsigned int optlen);
994 int compat_ipv6_getsockopt(struct sock *sk, int level, int optname,
995 char __user *optval, int __user *optlen);
996
997 int __ip6_datagram_connect(struct sock *sk, struct sockaddr *addr,
998 int addr_len);
999 int ip6_datagram_connect(struct sock *sk, struct sockaddr *addr, int addr_len);
1000 int ip6_datagram_connect_v6_only(struct sock *sk, struct sockaddr *addr,
1001 int addr_len);
1002 int ip6_datagram_dst_update(struct sock *sk, bool fix_sk_saddr);
1003 void ip6_datagram_release_cb(struct sock *sk);
1004
1005 int ipv6_recv_error(struct sock *sk, struct msghdr *msg, int len,
1006 int *addr_len);
1007 int ipv6_recv_rxpmtu(struct sock *sk, struct msghdr *msg, int len,
1008 int *addr_len);
1009 void ipv6_icmp_error(struct sock *sk, struct sk_buff *skb, int err, __be16 port,
1010 u32 info, u8 *payload);
1011 void ipv6_local_error(struct sock *sk, int err, struct flowi6 *fl6, u32 info);
1012 void ipv6_local_rxpmtu(struct sock *sk, struct flowi6 *fl6, u32 mtu);
1013
1014 int inet6_release(struct socket *sock);
1015 int inet6_bind(struct socket *sock, struct sockaddr *uaddr, int addr_len);
1016 int inet6_getname(struct socket *sock, struct sockaddr *uaddr, int *uaddr_len,
1017 int peer);
1018 int inet6_ioctl(struct socket *sock, unsigned int cmd, unsigned long arg);
1019
1020 int inet6_hash_connect(struct inet_timewait_death_row *death_row,
1021 struct sock *sk);
1022
1023 /*
1024 * reassembly.c
1025 */
1026 extern const struct proto_ops inet6_stream_ops;
1027 extern const struct proto_ops inet6_dgram_ops;
1028 extern const struct proto_ops inet6_sockraw_ops;
1029
1030 struct group_source_req;
1031 struct group_filter;
1032
1033 int ip6_mc_source(int add, int omode, struct sock *sk,
1034 struct group_source_req *pgsr);
1035 int ip6_mc_msfilter(struct sock *sk, struct group_filter *gsf);
1036 int ip6_mc_msfget(struct sock *sk, struct group_filter *gsf,
1037 struct group_filter __user *optval, int __user *optlen);
1038
1039 #ifdef CONFIG_PROC_FS
1040 int ac6_proc_init(struct net *net);
1041 void ac6_proc_exit(struct net *net);
1042 int raw6_proc_init(void);
1043 void raw6_proc_exit(void);
1044 int tcp6_proc_init(struct net *net);
1045 void tcp6_proc_exit(struct net *net);
1046 int udp6_proc_init(struct net *net);
1047 void udp6_proc_exit(struct net *net);
1048 int udplite6_proc_init(void);
1049 void udplite6_proc_exit(void);
1050 int ipv6_misc_proc_init(void);
1051 void ipv6_misc_proc_exit(void);
1052 int snmp6_register_dev(struct inet6_dev *idev);
1053 int snmp6_unregister_dev(struct inet6_dev *idev);
1054
1055 #else
1056 static inline int ac6_proc_init(struct net *net) { return 0; }
1057 static inline void ac6_proc_exit(struct net *net) { }
1058 static inline int snmp6_register_dev(struct inet6_dev *idev) { return 0; }
1059 static inline int snmp6_unregister_dev(struct inet6_dev *idev) { return 0; }
1060 #endif
1061
1062 #ifdef CONFIG_SYSCTL
1063 extern struct ctl_table ipv6_route_table_template[];
1064
1065 struct ctl_table *ipv6_icmp_sysctl_init(struct net *net);
1066 struct ctl_table *ipv6_route_sysctl_init(struct net *net);
1067 int ipv6_sysctl_register(void);
1068 void ipv6_sysctl_unregister(void);
1069 #endif
1070
1071 int ipv6_sock_mc_join(struct sock *sk, int ifindex,
1072 const struct in6_addr *addr);
1073 int ipv6_sock_mc_drop(struct sock *sk, int ifindex,
1074 const struct in6_addr *addr);
1075 #endif /* _NET_IPV6_H */