xref: /openbmc/linux/include/net/ipv6.h (revision 9dae47aba0a055f761176d9297371d5bb24289ec)
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 #include <net/netns/hash.h>
26 
27 #define SIN6_LEN_RFC2133	24
28 
29 #define IPV6_MAXPLEN		65535
30 
31 /*
32  *	NextHeader field of IPv6 header
33  */
34 
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. */
49 
50 #define NEXTHDR_MAX		255
51 
52 #define IPV6_DEFAULT_HOPLIMIT   64
53 #define IPV6_DEFAULT_MCASTHOPS	1
54 
55 /* Limits on Hop-by-Hop and Destination options.
56  *
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
60  * service attacks.
61  *
62  * There are three limits that may be set:
63  *   - Limit the number of options in a Hop-by-Hop or Destination options
64  *     extension header
65  *   - Limit the byte length of a Hop-by-Hop or Destination options extension
66  *     header
67  *   - Disallow unknown options
68  *
69  * The limits are expressed in corresponding sysctls:
70  *
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
75  *
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.
80  *
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.
84  *
85  * If a limit is exceeded when processing an extension header the packet is
86  * silently discarded.
87  */
88 
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 */
94 
95 /*
96  *	Addr type
97  *
98  *	type	-	unicast | multicast
99  *	scope	-	local	| site	    | global
100  *	v4	-	compat
101  *	v4mapped
102  *	any
103  *	loopback
104  */
105 
106 #define IPV6_ADDR_ANY		0x0000U
107 
108 #define IPV6_ADDR_UNICAST      	0x0001U
109 #define IPV6_ADDR_MULTICAST    	0x0002U
110 
111 #define IPV6_ADDR_LOOPBACK	0x0010U
112 #define IPV6_ADDR_LINKLOCAL	0x0020U
113 #define IPV6_ADDR_SITELOCAL	0x0040U
114 
115 #define IPV6_ADDR_COMPATv4	0x0080U
116 
117 #define IPV6_ADDR_SCOPE_MASK	0x00f0U
118 
119 #define IPV6_ADDR_MAPPED	0x1000U
120 
121 /*
122  *	Addr scopes
123  */
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
132 
133 /*
134  *	Addr flags
135  */
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)
142 
143 /*
144  *	fragmentation header
145  */
146 
147 struct frag_hdr {
148 	__u8	nexthdr;
149 	__u8	reserved;
150 	__be16	frag_off;
151 	__be32	identification;
152 };
153 
154 #define	IP6_MF		0x0001
155 #define	IP6_OFFSET	0xFFF8
156 
157 #define IP6_REPLY_MARK(net, mark) \
158 	((net)->ipv6.sysctl.fwmark_reflect ? (mark) : 0)
159 
160 #include <net/sock.h>
161 
162 /* sysctls */
163 extern int sysctl_mld_max_msf;
164 extern int sysctl_mld_qrv;
165 
166 #define _DEVINC(net, statname, mod, idev, field)			\
167 ({									\
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));\
172 })
173 
174 /* per device counters are atomic_long_t */
175 #define _DEVINCATOMIC(net, statname, mod, idev, field)			\
176 ({									\
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));\
181 })
182 
183 /* per device and per net counters are atomic_long_t */
184 #define _DEVINC_ATOMIC_ATOMIC(net, statname, idev, field)		\
185 ({									\
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));\
190 })
191 
192 #define _DEVADD(net, statname, mod, idev, field, val)			\
193 ({									\
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));\
198 })
199 
200 #define _DEVUPD(net, statname, mod, idev, field, val)			\
201 ({									\
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));\
206 })
207 
208 /* MIBs */
209 
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)
226 
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)
231 
232 struct ip6_ra_chain {
233 	struct ip6_ra_chain	*next;
234 	struct sock		*sk;
235 	int			sel;
236 	void			(*destructor)(struct sock *);
237 };
238 
239 extern struct ip6_ra_chain	*ip6_ra_chain;
240 extern rwlock_t ip6_ra_lock;
241 
242 /*
243    This structure is prepared by protocol, when parsing
244    ancillary data and passed to IPv6.
245  */
246 
247 struct ipv6_txoptions {
248 	refcount_t		refcnt;
249 	/* Length of this structure */
250 	int			tot_len;
251 
252 	/* length of extension headers   */
253 
254 	__u16			opt_flen;	/* after fragment hdr */
255 	__u16			opt_nflen;	/* before fragment hdr */
256 
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;
261 	struct rcu_head		rcu;
262 	/* Option buffer, as read by IPV6_PKTOPTIONS, starts here. */
263 };
264 
265 struct ip6_flowlabel {
266 	struct ip6_flowlabel __rcu *next;
267 	__be32			label;
268 	atomic_t		users;
269 	struct in6_addr		dst;
270 	struct ipv6_txoptions	*opt;
271 	unsigned long		linger;
272 	struct rcu_head		rcu;
273 	u8			share;
274 	union {
275 		struct pid *pid;
276 		kuid_t uid;
277 	} owner;
278 	unsigned long		lastuse;
279 	unsigned long		expires;
280 	struct net		*fl_net;
281 };
282 
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)
286 
287 #define IPV6_TCLASS_MASK (IPV6_FLOWINFO_MASK & ~IPV6_FLOWLABEL_MASK)
288 #define IPV6_TCLASS_SHIFT	20
289 
290 struct ipv6_fl_socklist {
291 	struct ipv6_fl_socklist	__rcu	*next;
292 	struct ip6_flowlabel		*fl;
293 	struct rcu_head			rcu;
294 };
295 
296 struct ipcm6_cookie {
297 	__s16 hlimit;
298 	__s16 tclass;
299 	__s8  dontfrag;
300 	struct ipv6_txoptions *opt;
301 };
302 
303 static inline struct ipv6_txoptions *txopt_get(const struct ipv6_pinfo *np)
304 {
305 	struct ipv6_txoptions *opt;
306 
307 	rcu_read_lock();
308 	opt = rcu_dereference(np->opt);
309 	if (opt) {
310 		if (!refcount_inc_not_zero(&opt->refcnt))
311 			opt = NULL;
312 		else
313 			opt = rcu_pointer_handoff(opt);
314 	}
315 	rcu_read_unlock();
316 	return opt;
317 }
318 
319 static inline void txopt_put(struct ipv6_txoptions *opt)
320 {
321 	if (opt && refcount_dec_and_test(&opt->refcnt))
322 		kfree_rcu(opt, rcu);
323 }
324 
325 struct ip6_flowlabel *fl6_sock_lookup(struct sock *sk, __be32 label);
326 struct ipv6_txoptions *fl6_merge_options(struct ipv6_txoptions *opt_space,
327 					 struct ip6_flowlabel *fl,
328 					 struct ipv6_txoptions *fopt);
329 void fl6_free_socklist(struct sock *sk);
330 int ipv6_flowlabel_opt(struct sock *sk, char __user *optval, int optlen);
331 int ipv6_flowlabel_opt_get(struct sock *sk, struct in6_flowlabel_req *freq,
332 			   int flags);
333 int ip6_flowlabel_init(void);
334 void ip6_flowlabel_cleanup(void);
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 __user *newopt,
357 					  int newoptlen);
358 struct ipv6_txoptions *
359 ipv6_renew_options_kern(struct sock *sk,
360 			struct ipv6_txoptions *opt,
361 			int newtype,
362 			struct ipv6_opt_hdr *newopt,
363 			int newoptlen);
364 struct ipv6_txoptions *ipv6_fixup_options(struct ipv6_txoptions *opt_space,
365 					  struct ipv6_txoptions *opt);
366 
367 bool ipv6_opt_accepted(const struct sock *sk, const struct sk_buff *skb,
368 		       const struct inet6_skb_parm *opt);
369 struct ipv6_txoptions *ipv6_update_options(struct sock *sk,
370 					   struct ipv6_txoptions *opt);
371 
372 static inline bool ipv6_accept_ra(struct inet6_dev *idev)
373 {
374 	/* If forwarding is enabled, RA are not accepted unless the special
375 	 * hybrid mode (accept_ra=2) is enabled.
376 	 */
377 	return idev->cnf.forwarding ? idev->cnf.accept_ra == 2 :
378 	    idev->cnf.accept_ra;
379 }
380 
381 #if IS_ENABLED(CONFIG_IPV6)
382 static inline int ip6_frag_mem(struct net *net)
383 {
384 	return sum_frag_mem_limit(&net->ipv6.frags);
385 }
386 #endif
387 
388 #define IPV6_FRAG_HIGH_THRESH	(4 * 1024*1024)	/* 4194304 */
389 #define IPV6_FRAG_LOW_THRESH	(3 * 1024*1024)	/* 3145728 */
390 #define IPV6_FRAG_TIMEOUT	(60 * HZ)	/* 60 seconds */
391 
392 int __ipv6_addr_type(const struct in6_addr *addr);
393 static inline int ipv6_addr_type(const struct in6_addr *addr)
394 {
395 	return __ipv6_addr_type(addr) & 0xffff;
396 }
397 
398 static inline int ipv6_addr_scope(const struct in6_addr *addr)
399 {
400 	return __ipv6_addr_type(addr) & IPV6_ADDR_SCOPE_MASK;
401 }
402 
403 static inline int __ipv6_addr_src_scope(int type)
404 {
405 	return (type == IPV6_ADDR_ANY) ? __IPV6_ADDR_SCOPE_INVALID : (type >> 16);
406 }
407 
408 static inline int ipv6_addr_src_scope(const struct in6_addr *addr)
409 {
410 	return __ipv6_addr_src_scope(__ipv6_addr_type(addr));
411 }
412 
413 static inline bool __ipv6_addr_needs_scope_id(int type)
414 {
415 	return type & IPV6_ADDR_LINKLOCAL ||
416 	       (type & IPV6_ADDR_MULTICAST &&
417 		(type & (IPV6_ADDR_LOOPBACK|IPV6_ADDR_LINKLOCAL)));
418 }
419 
420 static inline __u32 ipv6_iface_scope_id(const struct in6_addr *addr, int iface)
421 {
422 	return __ipv6_addr_needs_scope_id(__ipv6_addr_type(addr)) ? iface : 0;
423 }
424 
425 static inline int ipv6_addr_cmp(const struct in6_addr *a1, const struct in6_addr *a2)
426 {
427 	return memcmp(a1, a2, sizeof(struct in6_addr));
428 }
429 
430 static inline bool
431 ipv6_masked_addr_cmp(const struct in6_addr *a1, const struct in6_addr *m,
432 		     const struct in6_addr *a2)
433 {
434 #if defined(CONFIG_HAVE_EFFICIENT_UNALIGNED_ACCESS) && BITS_PER_LONG == 64
435 	const unsigned long *ul1 = (const unsigned long *)a1;
436 	const unsigned long *ulm = (const unsigned long *)m;
437 	const unsigned long *ul2 = (const unsigned long *)a2;
438 
439 	return !!(((ul1[0] ^ ul2[0]) & ulm[0]) |
440 		  ((ul1[1] ^ ul2[1]) & ulm[1]));
441 #else
442 	return !!(((a1->s6_addr32[0] ^ a2->s6_addr32[0]) & m->s6_addr32[0]) |
443 		  ((a1->s6_addr32[1] ^ a2->s6_addr32[1]) & m->s6_addr32[1]) |
444 		  ((a1->s6_addr32[2] ^ a2->s6_addr32[2]) & m->s6_addr32[2]) |
445 		  ((a1->s6_addr32[3] ^ a2->s6_addr32[3]) & m->s6_addr32[3]));
446 #endif
447 }
448 
449 static inline void ipv6_addr_prefix(struct in6_addr *pfx,
450 				    const struct in6_addr *addr,
451 				    int plen)
452 {
453 	/* caller must guarantee 0 <= plen <= 128 */
454 	int o = plen >> 3,
455 	    b = plen & 0x7;
456 
457 	memset(pfx->s6_addr, 0, sizeof(pfx->s6_addr));
458 	memcpy(pfx->s6_addr, addr, o);
459 	if (b != 0)
460 		pfx->s6_addr[o] = addr->s6_addr[o] & (0xff00 >> b);
461 }
462 
463 static inline void ipv6_addr_prefix_copy(struct in6_addr *addr,
464 					 const struct in6_addr *pfx,
465 					 int plen)
466 {
467 	/* caller must guarantee 0 <= plen <= 128 */
468 	int o = plen >> 3,
469 	    b = plen & 0x7;
470 
471 	memcpy(addr->s6_addr, pfx, o);
472 	if (b != 0) {
473 		addr->s6_addr[o] &= ~(0xff00 >> b);
474 		addr->s6_addr[o] |= (pfx->s6_addr[o] & (0xff00 >> b));
475 	}
476 }
477 
478 static inline void __ipv6_addr_set_half(__be32 *addr,
479 					__be32 wh, __be32 wl)
480 {
481 #if defined(CONFIG_HAVE_EFFICIENT_UNALIGNED_ACCESS) && BITS_PER_LONG == 64
482 #if defined(__BIG_ENDIAN)
483 	if (__builtin_constant_p(wh) && __builtin_constant_p(wl)) {
484 		*(__force u64 *)addr = ((__force u64)(wh) << 32 | (__force u64)(wl));
485 		return;
486 	}
487 #elif defined(__LITTLE_ENDIAN)
488 	if (__builtin_constant_p(wl) && __builtin_constant_p(wh)) {
489 		*(__force u64 *)addr = ((__force u64)(wl) << 32 | (__force u64)(wh));
490 		return;
491 	}
492 #endif
493 #endif
494 	addr[0] = wh;
495 	addr[1] = wl;
496 }
497 
498 static inline void ipv6_addr_set(struct in6_addr *addr,
499 				     __be32 w1, __be32 w2,
500 				     __be32 w3, __be32 w4)
501 {
502 	__ipv6_addr_set_half(&addr->s6_addr32[0], w1, w2);
503 	__ipv6_addr_set_half(&addr->s6_addr32[2], w3, w4);
504 }
505 
506 static inline bool ipv6_addr_equal(const struct in6_addr *a1,
507 				   const struct in6_addr *a2)
508 {
509 #if defined(CONFIG_HAVE_EFFICIENT_UNALIGNED_ACCESS) && BITS_PER_LONG == 64
510 	const unsigned long *ul1 = (const unsigned long *)a1;
511 	const unsigned long *ul2 = (const unsigned long *)a2;
512 
513 	return ((ul1[0] ^ ul2[0]) | (ul1[1] ^ ul2[1])) == 0UL;
514 #else
515 	return ((a1->s6_addr32[0] ^ a2->s6_addr32[0]) |
516 		(a1->s6_addr32[1] ^ a2->s6_addr32[1]) |
517 		(a1->s6_addr32[2] ^ a2->s6_addr32[2]) |
518 		(a1->s6_addr32[3] ^ a2->s6_addr32[3])) == 0;
519 #endif
520 }
521 
522 #if defined(CONFIG_HAVE_EFFICIENT_UNALIGNED_ACCESS) && BITS_PER_LONG == 64
523 static inline bool __ipv6_prefix_equal64_half(const __be64 *a1,
524 					      const __be64 *a2,
525 					      unsigned int len)
526 {
527 	if (len && ((*a1 ^ *a2) & cpu_to_be64((~0UL) << (64 - len))))
528 		return false;
529 	return true;
530 }
531 
532 static inline bool ipv6_prefix_equal(const struct in6_addr *addr1,
533 				     const struct in6_addr *addr2,
534 				     unsigned int prefixlen)
535 {
536 	const __be64 *a1 = (const __be64 *)addr1;
537 	const __be64 *a2 = (const __be64 *)addr2;
538 
539 	if (prefixlen >= 64) {
540 		if (a1[0] ^ a2[0])
541 			return false;
542 		return __ipv6_prefix_equal64_half(a1 + 1, a2 + 1, prefixlen - 64);
543 	}
544 	return __ipv6_prefix_equal64_half(a1, a2, prefixlen);
545 }
546 #else
547 static inline bool ipv6_prefix_equal(const struct in6_addr *addr1,
548 				     const struct in6_addr *addr2,
549 				     unsigned int prefixlen)
550 {
551 	const __be32 *a1 = addr1->s6_addr32;
552 	const __be32 *a2 = addr2->s6_addr32;
553 	unsigned int pdw, pbi;
554 
555 	/* check complete u32 in prefix */
556 	pdw = prefixlen >> 5;
557 	if (pdw && memcmp(a1, a2, pdw << 2))
558 		return false;
559 
560 	/* check incomplete u32 in prefix */
561 	pbi = prefixlen & 0x1f;
562 	if (pbi && ((a1[pdw] ^ a2[pdw]) & htonl((0xffffffff) << (32 - pbi))))
563 		return false;
564 
565 	return true;
566 }
567 #endif
568 
569 struct inet_frag_queue;
570 
571 enum ip6_defrag_users {
572 	IP6_DEFRAG_LOCAL_DELIVER,
573 	IP6_DEFRAG_CONNTRACK_IN,
574 	__IP6_DEFRAG_CONNTRACK_IN	= IP6_DEFRAG_CONNTRACK_IN + USHRT_MAX,
575 	IP6_DEFRAG_CONNTRACK_OUT,
576 	__IP6_DEFRAG_CONNTRACK_OUT	= IP6_DEFRAG_CONNTRACK_OUT + USHRT_MAX,
577 	IP6_DEFRAG_CONNTRACK_BRIDGE_IN,
578 	__IP6_DEFRAG_CONNTRACK_BRIDGE_IN = IP6_DEFRAG_CONNTRACK_BRIDGE_IN + USHRT_MAX,
579 };
580 
581 struct ip6_create_arg {
582 	__be32 id;
583 	u32 user;
584 	const struct in6_addr *src;
585 	const struct in6_addr *dst;
586 	int iif;
587 	u8 ecn;
588 };
589 
590 void ip6_frag_init(struct inet_frag_queue *q, const void *a);
591 bool ip6_frag_match(const struct inet_frag_queue *q, const void *a);
592 
593 /*
594  *	Equivalent of ipv4 struct ip
595  */
596 struct frag_queue {
597 	struct inet_frag_queue	q;
598 
599 	__be32			id;		/* fragment id		*/
600 	u32			user;
601 	struct in6_addr		saddr;
602 	struct in6_addr		daddr;
603 
604 	int			iif;
605 	unsigned int		csum;
606 	__u16			nhoffset;
607 	u8			ecn;
608 };
609 
610 void ip6_expire_frag_queue(struct net *net, struct frag_queue *fq,
611 			   struct inet_frags *frags);
612 
613 static inline bool ipv6_addr_any(const struct in6_addr *a)
614 {
615 #if defined(CONFIG_HAVE_EFFICIENT_UNALIGNED_ACCESS) && BITS_PER_LONG == 64
616 	const unsigned long *ul = (const unsigned long *)a;
617 
618 	return (ul[0] | ul[1]) == 0UL;
619 #else
620 	return (a->s6_addr32[0] | a->s6_addr32[1] |
621 		a->s6_addr32[2] | a->s6_addr32[3]) == 0;
622 #endif
623 }
624 
625 static inline u32 ipv6_addr_hash(const struct in6_addr *a)
626 {
627 #if defined(CONFIG_HAVE_EFFICIENT_UNALIGNED_ACCESS) && BITS_PER_LONG == 64
628 	const unsigned long *ul = (const unsigned long *)a;
629 	unsigned long x = ul[0] ^ ul[1];
630 
631 	return (u32)(x ^ (x >> 32));
632 #else
633 	return (__force u32)(a->s6_addr32[0] ^ a->s6_addr32[1] ^
634 			     a->s6_addr32[2] ^ a->s6_addr32[3]);
635 #endif
636 }
637 
638 /* more secured version of ipv6_addr_hash() */
639 static inline u32 __ipv6_addr_jhash(const struct in6_addr *a, const u32 initval)
640 {
641 	u32 v = (__force u32)a->s6_addr32[0] ^ (__force u32)a->s6_addr32[1];
642 
643 	return jhash_3words(v,
644 			    (__force u32)a->s6_addr32[2],
645 			    (__force u32)a->s6_addr32[3],
646 			    initval);
647 }
648 
649 static inline bool ipv6_addr_loopback(const struct in6_addr *a)
650 {
651 #if defined(CONFIG_HAVE_EFFICIENT_UNALIGNED_ACCESS) && BITS_PER_LONG == 64
652 	const __be64 *be = (const __be64 *)a;
653 
654 	return (be[0] | (be[1] ^ cpu_to_be64(1))) == 0UL;
655 #else
656 	return (a->s6_addr32[0] | a->s6_addr32[1] |
657 		a->s6_addr32[2] | (a->s6_addr32[3] ^ cpu_to_be32(1))) == 0;
658 #endif
659 }
660 
661 /*
662  * Note that we must __force cast these to unsigned long to make sparse happy,
663  * since all of the endian-annotated types are fixed size regardless of arch.
664  */
665 static inline bool ipv6_addr_v4mapped(const struct in6_addr *a)
666 {
667 	return (
668 #if defined(CONFIG_HAVE_EFFICIENT_UNALIGNED_ACCESS) && BITS_PER_LONG == 64
669 		*(unsigned long *)a |
670 #else
671 		(__force unsigned long)(a->s6_addr32[0] | a->s6_addr32[1]) |
672 #endif
673 		(__force unsigned long)(a->s6_addr32[2] ^
674 					cpu_to_be32(0x0000ffff))) == 0UL;
675 }
676 
677 static inline u32 ipv6_portaddr_hash(const struct net *net,
678 				     const struct in6_addr *addr6,
679 				     unsigned int port)
680 {
681 	unsigned int hash, mix = net_hash_mix(net);
682 
683 	if (ipv6_addr_any(addr6))
684 		hash = jhash_1word(0, mix);
685 	else if (ipv6_addr_v4mapped(addr6))
686 		hash = jhash_1word((__force u32)addr6->s6_addr32[3], mix);
687 	else
688 		hash = jhash2((__force u32 *)addr6->s6_addr32, 4, mix);
689 
690 	return hash ^ port;
691 }
692 
693 /*
694  * Check for a RFC 4843 ORCHID address
695  * (Overlay Routable Cryptographic Hash Identifiers)
696  */
697 static inline bool ipv6_addr_orchid(const struct in6_addr *a)
698 {
699 	return (a->s6_addr32[0] & htonl(0xfffffff0)) == htonl(0x20010010);
700 }
701 
702 static inline bool ipv6_addr_is_multicast(const struct in6_addr *addr)
703 {
704 	return (addr->s6_addr32[0] & htonl(0xFF000000)) == htonl(0xFF000000);
705 }
706 
707 static inline void ipv6_addr_set_v4mapped(const __be32 addr,
708 					  struct in6_addr *v4mapped)
709 {
710 	ipv6_addr_set(v4mapped,
711 			0, 0,
712 			htonl(0x0000FFFF),
713 			addr);
714 }
715 
716 /*
717  * find the first different bit between two addresses
718  * length of address must be a multiple of 32bits
719  */
720 static inline int __ipv6_addr_diff32(const void *token1, const void *token2, int addrlen)
721 {
722 	const __be32 *a1 = token1, *a2 = token2;
723 	int i;
724 
725 	addrlen >>= 2;
726 
727 	for (i = 0; i < addrlen; i++) {
728 		__be32 xb = a1[i] ^ a2[i];
729 		if (xb)
730 			return i * 32 + 31 - __fls(ntohl(xb));
731 	}
732 
733 	/*
734 	 *	we should *never* get to this point since that
735 	 *	would mean the addrs are equal
736 	 *
737 	 *	However, we do get to it 8) And exacly, when
738 	 *	addresses are equal 8)
739 	 *
740 	 *	ip route add 1111::/128 via ...
741 	 *	ip route add 1111::/64 via ...
742 	 *	and we are here.
743 	 *
744 	 *	Ideally, this function should stop comparison
745 	 *	at prefix length. It does not, but it is still OK,
746 	 *	if returned value is greater than prefix length.
747 	 *					--ANK (980803)
748 	 */
749 	return addrlen << 5;
750 }
751 
752 #if defined(CONFIG_HAVE_EFFICIENT_UNALIGNED_ACCESS) && BITS_PER_LONG == 64
753 static inline int __ipv6_addr_diff64(const void *token1, const void *token2, int addrlen)
754 {
755 	const __be64 *a1 = token1, *a2 = token2;
756 	int i;
757 
758 	addrlen >>= 3;
759 
760 	for (i = 0; i < addrlen; i++) {
761 		__be64 xb = a1[i] ^ a2[i];
762 		if (xb)
763 			return i * 64 + 63 - __fls(be64_to_cpu(xb));
764 	}
765 
766 	return addrlen << 6;
767 }
768 #endif
769 
770 static inline int __ipv6_addr_diff(const void *token1, const void *token2, int addrlen)
771 {
772 #if defined(CONFIG_HAVE_EFFICIENT_UNALIGNED_ACCESS) && BITS_PER_LONG == 64
773 	if (__builtin_constant_p(addrlen) && !(addrlen & 7))
774 		return __ipv6_addr_diff64(token1, token2, addrlen);
775 #endif
776 	return __ipv6_addr_diff32(token1, token2, addrlen);
777 }
778 
779 static inline int ipv6_addr_diff(const struct in6_addr *a1, const struct in6_addr *a2)
780 {
781 	return __ipv6_addr_diff(a1, a2, sizeof(struct in6_addr));
782 }
783 
784 __be32 ipv6_select_ident(struct net *net,
785 			 const struct in6_addr *daddr,
786 			 const struct in6_addr *saddr);
787 __be32 ipv6_proxy_select_ident(struct net *net, struct sk_buff *skb);
788 
789 int ip6_dst_hoplimit(struct dst_entry *dst);
790 
791 static inline int ip6_sk_dst_hoplimit(struct ipv6_pinfo *np, struct flowi6 *fl6,
792 				      struct dst_entry *dst)
793 {
794 	int hlimit;
795 
796 	if (ipv6_addr_is_multicast(&fl6->daddr))
797 		hlimit = np->mcast_hops;
798 	else
799 		hlimit = np->hop_limit;
800 	if (hlimit < 0)
801 		hlimit = ip6_dst_hoplimit(dst);
802 	return hlimit;
803 }
804 
805 /* copy IPv6 saddr & daddr to flow_keys, possibly using 64bit load/store
806  * Equivalent to :	flow->v6addrs.src = iph->saddr;
807  *			flow->v6addrs.dst = iph->daddr;
808  */
809 static inline void iph_to_flow_copy_v6addrs(struct flow_keys *flow,
810 					    const struct ipv6hdr *iph)
811 {
812 	BUILD_BUG_ON(offsetof(typeof(flow->addrs), v6addrs.dst) !=
813 		     offsetof(typeof(flow->addrs), v6addrs.src) +
814 		     sizeof(flow->addrs.v6addrs.src));
815 	memcpy(&flow->addrs.v6addrs, &iph->saddr, sizeof(flow->addrs.v6addrs));
816 	flow->control.addr_type = FLOW_DISSECTOR_KEY_IPV6_ADDRS;
817 }
818 
819 #if IS_ENABLED(CONFIG_IPV6)
820 
821 /* Sysctl settings for net ipv6.auto_flowlabels */
822 #define IP6_AUTO_FLOW_LABEL_OFF		0
823 #define IP6_AUTO_FLOW_LABEL_OPTOUT	1
824 #define IP6_AUTO_FLOW_LABEL_OPTIN	2
825 #define IP6_AUTO_FLOW_LABEL_FORCED	3
826 
827 #define IP6_AUTO_FLOW_LABEL_MAX		IP6_AUTO_FLOW_LABEL_FORCED
828 
829 #define IP6_DEFAULT_AUTO_FLOW_LABELS	IP6_AUTO_FLOW_LABEL_OPTOUT
830 
831 static inline __be32 ip6_make_flowlabel(struct net *net, struct sk_buff *skb,
832 					__be32 flowlabel, bool autolabel,
833 					struct flowi6 *fl6)
834 {
835 	u32 hash;
836 
837 	/* @flowlabel may include more than a flow label, eg, the traffic class.
838 	 * Here we want only the flow label value.
839 	 */
840 	flowlabel &= IPV6_FLOWLABEL_MASK;
841 
842 	if (flowlabel ||
843 	    net->ipv6.sysctl.auto_flowlabels == IP6_AUTO_FLOW_LABEL_OFF ||
844 	    (!autolabel &&
845 	     net->ipv6.sysctl.auto_flowlabels != IP6_AUTO_FLOW_LABEL_FORCED))
846 		return flowlabel;
847 
848 	hash = skb_get_hash_flowi6(skb, fl6);
849 
850 	/* Since this is being sent on the wire obfuscate hash a bit
851 	 * to minimize possbility that any useful information to an
852 	 * attacker is leaked. Only lower 20 bits are relevant.
853 	 */
854 	rol32(hash, 16);
855 
856 	flowlabel = (__force __be32)hash & IPV6_FLOWLABEL_MASK;
857 
858 	if (net->ipv6.sysctl.flowlabel_state_ranges)
859 		flowlabel |= IPV6_FLOWLABEL_STATELESS_FLAG;
860 
861 	return flowlabel;
862 }
863 
864 static inline int ip6_default_np_autolabel(struct net *net)
865 {
866 	switch (net->ipv6.sysctl.auto_flowlabels) {
867 	case IP6_AUTO_FLOW_LABEL_OFF:
868 	case IP6_AUTO_FLOW_LABEL_OPTIN:
869 	default:
870 		return 0;
871 	case IP6_AUTO_FLOW_LABEL_OPTOUT:
872 	case IP6_AUTO_FLOW_LABEL_FORCED:
873 		return 1;
874 	}
875 }
876 #else
877 static inline void ip6_set_txhash(struct sock *sk) { }
878 static inline __be32 ip6_make_flowlabel(struct net *net, struct sk_buff *skb,
879 					__be32 flowlabel, bool autolabel,
880 					struct flowi6 *fl6)
881 {
882 	return flowlabel;
883 }
884 static inline int ip6_default_np_autolabel(struct net *net)
885 {
886 	return 0;
887 }
888 #endif
889 
890 
891 /*
892  *	Header manipulation
893  */
894 static inline void ip6_flow_hdr(struct ipv6hdr *hdr, unsigned int tclass,
895 				__be32 flowlabel)
896 {
897 	*(__be32 *)hdr = htonl(0x60000000 | (tclass << 20)) | flowlabel;
898 }
899 
900 static inline __be32 ip6_flowinfo(const struct ipv6hdr *hdr)
901 {
902 	return *(__be32 *)hdr & IPV6_FLOWINFO_MASK;
903 }
904 
905 static inline __be32 ip6_flowlabel(const struct ipv6hdr *hdr)
906 {
907 	return *(__be32 *)hdr & IPV6_FLOWLABEL_MASK;
908 }
909 
910 static inline u8 ip6_tclass(__be32 flowinfo)
911 {
912 	return ntohl(flowinfo & IPV6_TCLASS_MASK) >> IPV6_TCLASS_SHIFT;
913 }
914 
915 static inline __be32 ip6_make_flowinfo(unsigned int tclass, __be32 flowlabel)
916 {
917 	return htonl(tclass << IPV6_TCLASS_SHIFT) | flowlabel;
918 }
919 
920 /*
921  *	Prototypes exported by ipv6
922  */
923 
924 /*
925  *	rcv function (called from netdevice level)
926  */
927 
928 int ipv6_rcv(struct sk_buff *skb, struct net_device *dev,
929 	     struct packet_type *pt, struct net_device *orig_dev);
930 
931 int ip6_rcv_finish(struct net *net, struct sock *sk, struct sk_buff *skb);
932 
933 /*
934  *	upper-layer output functions
935  */
936 int ip6_xmit(const struct sock *sk, struct sk_buff *skb, struct flowi6 *fl6,
937 	     __u32 mark, struct ipv6_txoptions *opt, int tclass);
938 
939 int ip6_find_1stfragopt(struct sk_buff *skb, u8 **nexthdr);
940 
941 int ip6_append_data(struct sock *sk,
942 		    int getfrag(void *from, char *to, int offset, int len,
943 				int odd, struct sk_buff *skb),
944 		    void *from, int length, int transhdrlen,
945 		    struct ipcm6_cookie *ipc6, struct flowi6 *fl6,
946 		    struct rt6_info *rt, unsigned int flags,
947 		    const struct sockcm_cookie *sockc);
948 
949 int ip6_push_pending_frames(struct sock *sk);
950 
951 void ip6_flush_pending_frames(struct sock *sk);
952 
953 int ip6_send_skb(struct sk_buff *skb);
954 
955 struct sk_buff *__ip6_make_skb(struct sock *sk, struct sk_buff_head *queue,
956 			       struct inet_cork_full *cork,
957 			       struct inet6_cork *v6_cork);
958 struct sk_buff *ip6_make_skb(struct sock *sk,
959 			     int getfrag(void *from, char *to, int offset,
960 					 int len, int odd, struct sk_buff *skb),
961 			     void *from, int length, int transhdrlen,
962 			     struct ipcm6_cookie *ipc6, struct flowi6 *fl6,
963 			     struct rt6_info *rt, unsigned int flags,
964 			     const struct sockcm_cookie *sockc);
965 
966 static inline struct sk_buff *ip6_finish_skb(struct sock *sk)
967 {
968 	return __ip6_make_skb(sk, &sk->sk_write_queue, &inet_sk(sk)->cork,
969 			      &inet6_sk(sk)->cork);
970 }
971 
972 int ip6_dst_lookup(struct net *net, struct sock *sk, struct dst_entry **dst,
973 		   struct flowi6 *fl6);
974 struct dst_entry *ip6_dst_lookup_flow(const struct sock *sk, struct flowi6 *fl6,
975 				      const struct in6_addr *final_dst);
976 struct dst_entry *ip6_sk_dst_lookup_flow(struct sock *sk, struct flowi6 *fl6,
977 					 const struct in6_addr *final_dst);
978 struct dst_entry *ip6_blackhole_route(struct net *net,
979 				      struct dst_entry *orig_dst);
980 
981 /*
982  *	skb processing functions
983  */
984 
985 int ip6_output(struct net *net, struct sock *sk, struct sk_buff *skb);
986 int ip6_forward(struct sk_buff *skb);
987 int ip6_input(struct sk_buff *skb);
988 int ip6_mc_input(struct sk_buff *skb);
989 
990 int __ip6_local_out(struct net *net, struct sock *sk, struct sk_buff *skb);
991 int ip6_local_out(struct net *net, struct sock *sk, struct sk_buff *skb);
992 
993 /*
994  *	Extension header (options) processing
995  */
996 
997 void ipv6_push_nfrag_opts(struct sk_buff *skb, struct ipv6_txoptions *opt,
998 			  u8 *proto, struct in6_addr **daddr_p,
999 			  struct in6_addr *saddr);
1000 void ipv6_push_frag_opts(struct sk_buff *skb, struct ipv6_txoptions *opt,
1001 			 u8 *proto);
1002 
1003 int ipv6_skip_exthdr(const struct sk_buff *, int start, u8 *nexthdrp,
1004 		     __be16 *frag_offp);
1005 
1006 bool ipv6_ext_hdr(u8 nexthdr);
1007 
1008 enum {
1009 	IP6_FH_F_FRAG		= (1 << 0),
1010 	IP6_FH_F_AUTH		= (1 << 1),
1011 	IP6_FH_F_SKIP_RH	= (1 << 2),
1012 };
1013 
1014 /* find specified header and get offset to it */
1015 int ipv6_find_hdr(const struct sk_buff *skb, unsigned int *offset, int target,
1016 		  unsigned short *fragoff, int *fragflg);
1017 
1018 int ipv6_find_tlv(const struct sk_buff *skb, int offset, int type);
1019 
1020 struct in6_addr *fl6_update_dst(struct flowi6 *fl6,
1021 				const struct ipv6_txoptions *opt,
1022 				struct in6_addr *orig);
1023 
1024 /*
1025  *	socket options (ipv6_sockglue.c)
1026  */
1027 
1028 int ipv6_setsockopt(struct sock *sk, int level, int optname,
1029 		    char __user *optval, unsigned int optlen);
1030 int ipv6_getsockopt(struct sock *sk, int level, int optname,
1031 		    char __user *optval, int __user *optlen);
1032 int compat_ipv6_setsockopt(struct sock *sk, int level, int optname,
1033 			   char __user *optval, unsigned int optlen);
1034 int compat_ipv6_getsockopt(struct sock *sk, int level, int optname,
1035 			   char __user *optval, int __user *optlen);
1036 
1037 int __ip6_datagram_connect(struct sock *sk, struct sockaddr *addr,
1038 			   int addr_len);
1039 int ip6_datagram_connect(struct sock *sk, struct sockaddr *addr, int addr_len);
1040 int ip6_datagram_connect_v6_only(struct sock *sk, struct sockaddr *addr,
1041 				 int addr_len);
1042 int ip6_datagram_dst_update(struct sock *sk, bool fix_sk_saddr);
1043 void ip6_datagram_release_cb(struct sock *sk);
1044 
1045 int ipv6_recv_error(struct sock *sk, struct msghdr *msg, int len,
1046 		    int *addr_len);
1047 int ipv6_recv_rxpmtu(struct sock *sk, struct msghdr *msg, int len,
1048 		     int *addr_len);
1049 void ipv6_icmp_error(struct sock *sk, struct sk_buff *skb, int err, __be16 port,
1050 		     u32 info, u8 *payload);
1051 void ipv6_local_error(struct sock *sk, int err, struct flowi6 *fl6, u32 info);
1052 void ipv6_local_rxpmtu(struct sock *sk, struct flowi6 *fl6, u32 mtu);
1053 
1054 int inet6_release(struct socket *sock);
1055 int inet6_bind(struct socket *sock, struct sockaddr *uaddr, int addr_len);
1056 int inet6_getname(struct socket *sock, struct sockaddr *uaddr, int *uaddr_len,
1057 		  int peer);
1058 int inet6_ioctl(struct socket *sock, unsigned int cmd, unsigned long arg);
1059 
1060 int inet6_hash_connect(struct inet_timewait_death_row *death_row,
1061 			      struct sock *sk);
1062 
1063 /*
1064  * reassembly.c
1065  */
1066 extern const struct proto_ops inet6_stream_ops;
1067 extern const struct proto_ops inet6_dgram_ops;
1068 extern const struct proto_ops inet6_sockraw_ops;
1069 
1070 struct group_source_req;
1071 struct group_filter;
1072 
1073 int ip6_mc_source(int add, int omode, struct sock *sk,
1074 		  struct group_source_req *pgsr);
1075 int ip6_mc_msfilter(struct sock *sk, struct group_filter *gsf);
1076 int ip6_mc_msfget(struct sock *sk, struct group_filter *gsf,
1077 		  struct group_filter __user *optval, int __user *optlen);
1078 
1079 #ifdef CONFIG_PROC_FS
1080 int ac6_proc_init(struct net *net);
1081 void ac6_proc_exit(struct net *net);
1082 int raw6_proc_init(void);
1083 void raw6_proc_exit(void);
1084 int tcp6_proc_init(struct net *net);
1085 void tcp6_proc_exit(struct net *net);
1086 int udp6_proc_init(struct net *net);
1087 void udp6_proc_exit(struct net *net);
1088 int udplite6_proc_init(void);
1089 void udplite6_proc_exit(void);
1090 int ipv6_misc_proc_init(void);
1091 void ipv6_misc_proc_exit(void);
1092 int snmp6_register_dev(struct inet6_dev *idev);
1093 int snmp6_unregister_dev(struct inet6_dev *idev);
1094 
1095 #else
1096 static inline int ac6_proc_init(struct net *net) { return 0; }
1097 static inline void ac6_proc_exit(struct net *net) { }
1098 static inline int snmp6_register_dev(struct inet6_dev *idev) { return 0; }
1099 static inline int snmp6_unregister_dev(struct inet6_dev *idev) { return 0; }
1100 #endif
1101 
1102 #ifdef CONFIG_SYSCTL
1103 extern struct ctl_table ipv6_route_table_template[];
1104 
1105 struct ctl_table *ipv6_icmp_sysctl_init(struct net *net);
1106 struct ctl_table *ipv6_route_sysctl_init(struct net *net);
1107 int ipv6_sysctl_register(void);
1108 void ipv6_sysctl_unregister(void);
1109 #endif
1110 
1111 int ipv6_sock_mc_join(struct sock *sk, int ifindex,
1112 		      const struct in6_addr *addr);
1113 int ipv6_sock_mc_drop(struct sock *sk, int ifindex,
1114 		      const struct in6_addr *addr);
1115 #endif /* _NET_IPV6_H */
1116