1/* SPDX-License-Identifier: GPL-2.0-only */
2/* Copyright (C) 2000-2002 Joakim Axelsson <gozem@linux.nu>
3 *                         Patrick Schaaf <bof@bof.de>
4 *                         Martin Josefsson <gandalf@wlug.westbo.se>
5 * Copyright (C) 2003-2013 Jozsef Kadlecsik <kadlec@netfilter.org>
6 */
7#ifndef _IP_SET_H
8#define _IP_SET_H
9
10#include <linux/ip.h>
11#include <linux/ipv6.h>
12#include <linux/netlink.h>
13#include <linux/netfilter.h>
14#include <linux/netfilter/x_tables.h>
15#include <linux/stringify.h>
16#include <linux/vmalloc.h>
17#include <net/netlink.h>
18#include <uapi/linux/netfilter/ipset/ip_set.h>
19
20#define _IP_SET_MODULE_DESC(a, b, c)		\
21	MODULE_DESCRIPTION(a " type of IP sets, revisions " b "-" c)
22#define IP_SET_MODULE_DESC(a, b, c)		\
23	_IP_SET_MODULE_DESC(a, __stringify(b), __stringify(c))
24
25/* Set features */
26enum ip_set_feature {
27	IPSET_TYPE_IP_FLAG = 0,
28	IPSET_TYPE_IP = (1 << IPSET_TYPE_IP_FLAG),
29	IPSET_TYPE_PORT_FLAG = 1,
30	IPSET_TYPE_PORT = (1 << IPSET_TYPE_PORT_FLAG),
31	IPSET_TYPE_MAC_FLAG = 2,
32	IPSET_TYPE_MAC = (1 << IPSET_TYPE_MAC_FLAG),
33	IPSET_TYPE_IP2_FLAG = 3,
34	IPSET_TYPE_IP2 = (1 << IPSET_TYPE_IP2_FLAG),
35	IPSET_TYPE_NAME_FLAG = 4,
36	IPSET_TYPE_NAME = (1 << IPSET_TYPE_NAME_FLAG),
37	IPSET_TYPE_IFACE_FLAG = 5,
38	IPSET_TYPE_IFACE = (1 << IPSET_TYPE_IFACE_FLAG),
39	IPSET_TYPE_MARK_FLAG = 6,
40	IPSET_TYPE_MARK = (1 << IPSET_TYPE_MARK_FLAG),
41	IPSET_TYPE_NOMATCH_FLAG = 7,
42	IPSET_TYPE_NOMATCH = (1 << IPSET_TYPE_NOMATCH_FLAG),
43	/* Strictly speaking not a feature, but a flag for dumping:
44	 * this settype must be dumped last */
45	IPSET_DUMP_LAST_FLAG = 8,
46	IPSET_DUMP_LAST = (1 << IPSET_DUMP_LAST_FLAG),
47};
48
49/* Set extensions */
50enum ip_set_extension {
51	IPSET_EXT_BIT_TIMEOUT = 0,
52	IPSET_EXT_TIMEOUT = (1 << IPSET_EXT_BIT_TIMEOUT),
53	IPSET_EXT_BIT_COUNTER = 1,
54	IPSET_EXT_COUNTER = (1 << IPSET_EXT_BIT_COUNTER),
55	IPSET_EXT_BIT_COMMENT = 2,
56	IPSET_EXT_COMMENT = (1 << IPSET_EXT_BIT_COMMENT),
57	IPSET_EXT_BIT_SKBINFO = 3,
58	IPSET_EXT_SKBINFO = (1 << IPSET_EXT_BIT_SKBINFO),
59	/* Mark set with an extension which needs to call destroy */
60	IPSET_EXT_BIT_DESTROY = 7,
61	IPSET_EXT_DESTROY = (1 << IPSET_EXT_BIT_DESTROY),
62};
63
64#define SET_WITH_TIMEOUT(s)	((s)->extensions & IPSET_EXT_TIMEOUT)
65#define SET_WITH_COUNTER(s)	((s)->extensions & IPSET_EXT_COUNTER)
66#define SET_WITH_COMMENT(s)	((s)->extensions & IPSET_EXT_COMMENT)
67#define SET_WITH_SKBINFO(s)	((s)->extensions & IPSET_EXT_SKBINFO)
68#define SET_WITH_FORCEADD(s)	((s)->flags & IPSET_CREATE_FLAG_FORCEADD)
69
70/* Extension id, in size order */
71enum ip_set_ext_id {
72	IPSET_EXT_ID_COUNTER = 0,
73	IPSET_EXT_ID_TIMEOUT,
74	IPSET_EXT_ID_SKBINFO,
75	IPSET_EXT_ID_COMMENT,
76	IPSET_EXT_ID_MAX,
77};
78
79struct ip_set;
80
81/* Extension type */
82struct ip_set_ext_type {
83	/* Destroy extension private data (can be NULL) */
84	void (*destroy)(struct ip_set *set, void *ext);
85	enum ip_set_extension type;
86	enum ipset_cadt_flags flag;
87	/* Size and minimal alignment */
88	u8 len;
89	u8 align;
90};
91
92extern const struct ip_set_ext_type ip_set_extensions[];
93
94struct ip_set_counter {
95	atomic64_t bytes;
96	atomic64_t packets;
97};
98
99struct ip_set_comment_rcu {
100	struct rcu_head rcu;
101	char str[];
102};
103
104struct ip_set_comment {
105	struct ip_set_comment_rcu __rcu *c;
106};
107
108struct ip_set_skbinfo {
109	u32 skbmark;
110	u32 skbmarkmask;
111	u32 skbprio;
112	u16 skbqueue;
113	u16 __pad;
114};
115
116struct ip_set_ext {
117	struct ip_set_skbinfo skbinfo;
118	u64 packets;
119	u64 bytes;
120	char *comment;
121	u32 timeout;
122	u8 packets_op;
123	u8 bytes_op;
124	bool target;
125};
126
127#define ext_timeout(e, s)	\
128((unsigned long *)(((void *)(e)) + (s)->offset[IPSET_EXT_ID_TIMEOUT]))
129#define ext_counter(e, s)	\
130((struct ip_set_counter *)(((void *)(e)) + (s)->offset[IPSET_EXT_ID_COUNTER]))
131#define ext_comment(e, s)	\
132((struct ip_set_comment *)(((void *)(e)) + (s)->offset[IPSET_EXT_ID_COMMENT]))
133#define ext_skbinfo(e, s)	\
134((struct ip_set_skbinfo *)(((void *)(e)) + (s)->offset[IPSET_EXT_ID_SKBINFO]))
135
136typedef int (*ipset_adtfn)(struct ip_set *set, void *value,
137			   const struct ip_set_ext *ext,
138			   struct ip_set_ext *mext, u32 cmdflags);
139
140/* Kernel API function options */
141struct ip_set_adt_opt {
142	u8 family;		/* Actual protocol family */
143	u8 dim;			/* Dimension of match/target */
144	u8 flags;		/* Direction and negation flags */
145	u32 cmdflags;		/* Command-like flags */
146	struct ip_set_ext ext;	/* Extensions */
147};
148
149/* Set type, variant-specific part */
150struct ip_set_type_variant {
151	/* Kernelspace: test/add/del entries
152	 *		returns negative error code,
153	 *			zero for no match/success to add/delete
154	 *			positive for matching element */
155	int (*kadt)(struct ip_set *set, const struct sk_buff *skb,
156		    const struct xt_action_param *par,
157		    enum ipset_adt adt, struct ip_set_adt_opt *opt);
158
159	/* Userspace: test/add/del entries
160	 *		returns negative error code,
161	 *			zero for no match/success to add/delete
162	 *			positive for matching element */
163	int (*uadt)(struct ip_set *set, struct nlattr *tb[],
164		    enum ipset_adt adt, u32 *lineno, u32 flags, bool retried);
165
166	/* Low level add/del/test functions */
167	ipset_adtfn adt[IPSET_ADT_MAX];
168
169	/* When adding entries and set is full, try to resize the set */
170	int (*resize)(struct ip_set *set, bool retried);
171	/* Destroy the set */
172	void (*destroy)(struct ip_set *set);
173	/* Flush the elements */
174	void (*flush)(struct ip_set *set);
175	/* Expire entries before listing */
176	void (*expire)(struct ip_set *set);
177	/* List set header data */
178	int (*head)(struct ip_set *set, struct sk_buff *skb);
179	/* List elements */
180	int (*list)(const struct ip_set *set, struct sk_buff *skb,
181		    struct netlink_callback *cb);
182	/* Keep listing private when resizing runs parallel */
183	void (*uref)(struct ip_set *set, struct netlink_callback *cb,
184		     bool start);
185
186	/* Return true if "b" set is the same as "a"
187	 * according to the create set parameters */
188	bool (*same_set)(const struct ip_set *a, const struct ip_set *b);
189	/* Cancel ongoing garbage collectors before destroying the set*/
190	void (*cancel_gc)(struct ip_set *set);
191	/* Region-locking is used */
192	bool region_lock;
193};
194
195struct ip_set_region {
196	spinlock_t lock;	/* Region lock */
197	size_t ext_size;	/* Size of the dynamic extensions */
198	u32 elements;		/* Number of elements vs timeout */
199};
200
201/* Max range where every element is added/deleted in one step */
202#define IPSET_MAX_RANGE		(1<<14)
203
204/* The max revision number supported by any set type + 1 */
205#define IPSET_REVISION_MAX	9
206
207/* The core set type structure */
208struct ip_set_type {
209	struct list_head list;
210
211	/* Typename */
212	char name[IPSET_MAXNAMELEN];
213	/* Protocol version */
214	u8 protocol;
215	/* Set type dimension */
216	u8 dimension;
217	/*
218	 * Supported family: may be NFPROTO_UNSPEC for both
219	 * NFPROTO_IPV4/NFPROTO_IPV6.
220	 */
221	u8 family;
222	/* Type revisions */
223	u8 revision_min, revision_max;
224	/* Revision-specific supported (create) flags */
225	u8 create_flags[IPSET_REVISION_MAX+1];
226	/* Set features to control swapping */
227	u16 features;
228
229	/* Create set */
230	int (*create)(struct net *net, struct ip_set *set,
231		      struct nlattr *tb[], u32 flags);
232
233	/* Attribute policies */
234	const struct nla_policy create_policy[IPSET_ATTR_CREATE_MAX + 1];
235	const struct nla_policy adt_policy[IPSET_ATTR_ADT_MAX + 1];
236
237	/* Set this to THIS_MODULE if you are a module, otherwise NULL */
238	struct module *me;
239};
240
241/* register and unregister set type */
242extern int ip_set_type_register(struct ip_set_type *set_type);
243extern void ip_set_type_unregister(struct ip_set_type *set_type);
244
245/* A generic IP set */
246struct ip_set {
247	/* For call_cru in destroy */
248	struct rcu_head rcu;
249	/* The name of the set */
250	char name[IPSET_MAXNAMELEN];
251	/* Lock protecting the set data */
252	spinlock_t lock;
253	/* References to the set */
254	u32 ref;
255	/* References to the set for netlink events like dump,
256	 * ref can be swapped out by ip_set_swap
257	 */
258	u32 ref_netlink;
259	/* The core set type */
260	struct ip_set_type *type;
261	/* The type variant doing the real job */
262	const struct ip_set_type_variant *variant;
263	/* The actual INET family of the set */
264	u8 family;
265	/* The type revision */
266	u8 revision;
267	/* Extensions */
268	u8 extensions;
269	/* Create flags */
270	u8 flags;
271	/* Default timeout value, if enabled */
272	u32 timeout;
273	/* Number of elements (vs timeout) */
274	u32 elements;
275	/* Size of the dynamic extensions (vs timeout) */
276	size_t ext_size;
277	/* Element data size */
278	size_t dsize;
279	/* Offsets to extensions in elements */
280	size_t offset[IPSET_EXT_ID_MAX];
281	/* The type specific data */
282	void *data;
283};
284
285static inline void
286ip_set_ext_destroy(struct ip_set *set, void *data)
287{
288	/* Check that the extension is enabled for the set and
289	 * call it's destroy function for its extension part in data.
290	 */
291	if (SET_WITH_COMMENT(set)) {
292		struct ip_set_comment *c = ext_comment(data, set);
293
294		ip_set_extensions[IPSET_EXT_ID_COMMENT].destroy(set, c);
295	}
296}
297
298int ip_set_put_flags(struct sk_buff *skb, struct ip_set *set);
299
300/* Netlink CB args */
301enum {
302	IPSET_CB_NET = 0,	/* net namespace */
303	IPSET_CB_PROTO,		/* ipset protocol */
304	IPSET_CB_DUMP,		/* dump single set/all sets */
305	IPSET_CB_INDEX,		/* set index */
306	IPSET_CB_PRIVATE,	/* set private data */
307	IPSET_CB_ARG0,		/* type specific */
308};
309
310/* register and unregister set references */
311extern ip_set_id_t ip_set_get_byname(struct net *net,
312				     const char *name, struct ip_set **set);
313extern void ip_set_put_byindex(struct net *net, ip_set_id_t index);
314extern void ip_set_name_byindex(struct net *net, ip_set_id_t index, char *name);
315extern ip_set_id_t ip_set_nfnl_get_byindex(struct net *net, ip_set_id_t index);
316extern void ip_set_nfnl_put(struct net *net, ip_set_id_t index);
317
318/* API for iptables set match, and SET target */
319
320extern int ip_set_add(ip_set_id_t id, const struct sk_buff *skb,
321		      const struct xt_action_param *par,
322		      struct ip_set_adt_opt *opt);
323extern int ip_set_del(ip_set_id_t id, const struct sk_buff *skb,
324		      const struct xt_action_param *par,
325		      struct ip_set_adt_opt *opt);
326extern int ip_set_test(ip_set_id_t id, const struct sk_buff *skb,
327		       const struct xt_action_param *par,
328		       struct ip_set_adt_opt *opt);
329
330/* Utility functions */
331extern void *ip_set_alloc(size_t size);
332extern void ip_set_free(void *members);
333extern int ip_set_get_ipaddr4(struct nlattr *nla,  __be32 *ipaddr);
334extern int ip_set_get_ipaddr6(struct nlattr *nla, union nf_inet_addr *ipaddr);
335extern size_t ip_set_elem_len(struct ip_set *set, struct nlattr *tb[],
336			      size_t len, size_t align);
337extern int ip_set_get_extensions(struct ip_set *set, struct nlattr *tb[],
338				 struct ip_set_ext *ext);
339extern int ip_set_put_extensions(struct sk_buff *skb, const struct ip_set *set,
340				 const void *e, bool active);
341extern bool ip_set_match_extensions(struct ip_set *set,
342				    const struct ip_set_ext *ext,
343				    struct ip_set_ext *mext,
344				    u32 flags, void *data);
345
346static inline int
347ip_set_get_hostipaddr4(struct nlattr *nla, u32 *ipaddr)
348{
349	__be32 ip;
350	int ret = ip_set_get_ipaddr4(nla, &ip);
351
352	if (ret)
353		return ret;
354	*ipaddr = ntohl(ip);
355	return 0;
356}
357
358/* Ignore IPSET_ERR_EXIST errors if asked to do so? */
359static inline bool
360ip_set_eexist(int ret, u32 flags)
361{
362	return ret == -IPSET_ERR_EXIST && (flags & IPSET_FLAG_EXIST);
363}
364
365/* Match elements marked with nomatch */
366static inline bool
367ip_set_enomatch(int ret, u32 flags, enum ipset_adt adt, struct ip_set *set)
368{
369	return adt == IPSET_TEST &&
370	       (set->type->features & IPSET_TYPE_NOMATCH) &&
371	       ((flags >> 16) & IPSET_FLAG_NOMATCH) &&
372	       (ret > 0 || ret == -ENOTEMPTY);
373}
374
375/* Check the NLA_F_NET_BYTEORDER flag */
376static inline bool
377ip_set_attr_netorder(struct nlattr *tb[], int type)
378{
379	return tb[type] && (tb[type]->nla_type & NLA_F_NET_BYTEORDER);
380}
381
382static inline bool
383ip_set_optattr_netorder(struct nlattr *tb[], int type)
384{
385	return !tb[type] || (tb[type]->nla_type & NLA_F_NET_BYTEORDER);
386}
387
388/* Useful converters */
389static inline u32
390ip_set_get_h32(const struct nlattr *attr)
391{
392	return ntohl(nla_get_be32(attr));
393}
394
395static inline u16
396ip_set_get_h16(const struct nlattr *attr)
397{
398	return ntohs(nla_get_be16(attr));
399}
400
401static inline int nla_put_ipaddr4(struct sk_buff *skb, int type, __be32 ipaddr)
402{
403	struct nlattr *__nested = nla_nest_start(skb, type);
404	int ret;
405
406	if (!__nested)
407		return -EMSGSIZE;
408	ret = nla_put_in_addr(skb, IPSET_ATTR_IPADDR_IPV4, ipaddr);
409	if (!ret)
410		nla_nest_end(skb, __nested);
411	return ret;
412}
413
414static inline int nla_put_ipaddr6(struct sk_buff *skb, int type,
415				  const struct in6_addr *ipaddrptr)
416{
417	struct nlattr *__nested = nla_nest_start(skb, type);
418	int ret;
419
420	if (!__nested)
421		return -EMSGSIZE;
422	ret = nla_put_in6_addr(skb, IPSET_ATTR_IPADDR_IPV6, ipaddrptr);
423	if (!ret)
424		nla_nest_end(skb, __nested);
425	return ret;
426}
427
428/* Get address from skbuff */
429static inline __be32
430ip4addr(const struct sk_buff *skb, bool src)
431{
432	return src ? ip_hdr(skb)->saddr : ip_hdr(skb)->daddr;
433}
434
435static inline void
436ip4addrptr(const struct sk_buff *skb, bool src, __be32 *addr)
437{
438	*addr = src ? ip_hdr(skb)->saddr : ip_hdr(skb)->daddr;
439}
440
441static inline void
442ip6addrptr(const struct sk_buff *skb, bool src, struct in6_addr *addr)
443{
444	memcpy(addr, src ? &ipv6_hdr(skb)->saddr : &ipv6_hdr(skb)->daddr,
445	       sizeof(*addr));
446}
447
448/* How often should the gc be run by default */
449#define IPSET_GC_TIME			(3 * 60)
450
451/* Timeout period depending on the timeout value of the given set */
452#define IPSET_GC_PERIOD(timeout) \
453	((timeout/3) ? min_t(u32, (timeout)/3, IPSET_GC_TIME) : 1)
454
455/* Entry is set with no timeout value */
456#define IPSET_ELEM_PERMANENT	0
457
458/* Set is defined with timeout support: timeout value may be 0 */
459#define IPSET_NO_TIMEOUT	UINT_MAX
460
461/* Max timeout value, see msecs_to_jiffies() in jiffies.h */
462#define IPSET_MAX_TIMEOUT	(UINT_MAX >> 1)/MSEC_PER_SEC
463
464#define ip_set_adt_opt_timeout(opt, set)	\
465((opt)->ext.timeout != IPSET_NO_TIMEOUT ? (opt)->ext.timeout : (set)->timeout)
466
467static inline unsigned int
468ip_set_timeout_uget(struct nlattr *tb)
469{
470	unsigned int timeout = ip_set_get_h32(tb);
471
472	/* Normalize to fit into jiffies */
473	if (timeout > IPSET_MAX_TIMEOUT)
474		timeout = IPSET_MAX_TIMEOUT;
475
476	return timeout;
477}
478
479static inline bool
480ip_set_timeout_expired(const unsigned long *t)
481{
482	return *t != IPSET_ELEM_PERMANENT && time_is_before_jiffies(*t);
483}
484
485static inline void
486ip_set_timeout_set(unsigned long *timeout, u32 value)
487{
488	unsigned long t;
489
490	if (!value) {
491		*timeout = IPSET_ELEM_PERMANENT;
492		return;
493	}
494
495	t = msecs_to_jiffies(value * MSEC_PER_SEC) + jiffies;
496	if (t == IPSET_ELEM_PERMANENT)
497		/* Bingo! :-) */
498		t--;
499	*timeout = t;
500}
501
502void ip_set_init_comment(struct ip_set *set, struct ip_set_comment *comment,
503			 const struct ip_set_ext *ext);
504
505static inline void
506ip_set_init_counter(struct ip_set_counter *counter,
507		    const struct ip_set_ext *ext)
508{
509	if (ext->bytes != ULLONG_MAX)
510		atomic64_set(&(counter)->bytes, (long long)(ext->bytes));
511	if (ext->packets != ULLONG_MAX)
512		atomic64_set(&(counter)->packets, (long long)(ext->packets));
513}
514
515static inline void
516ip_set_init_skbinfo(struct ip_set_skbinfo *skbinfo,
517		    const struct ip_set_ext *ext)
518{
519	*skbinfo = ext->skbinfo;
520}
521
522static inline void
523nf_inet_addr_mask_inplace(union nf_inet_addr *a1,
524			  const union nf_inet_addr *mask)
525{
526	a1->all[0] &= mask->all[0];
527	a1->all[1] &= mask->all[1];
528	a1->all[2] &= mask->all[2];
529	a1->all[3] &= mask->all[3];
530}
531
532#define IP_SET_INIT_KEXT(skb, opt, set)			\
533	{ .bytes = (skb)->len, .packets = 1, .target = true,\
534	  .timeout = ip_set_adt_opt_timeout(opt, set) }
535
536#define IP_SET_INIT_UEXT(set)				\
537	{ .bytes = ULLONG_MAX, .packets = ULLONG_MAX,	\
538	  .timeout = (set)->timeout }
539
540#define IPSET_CONCAT(a, b)		a##b
541#define IPSET_TOKEN(a, b)		IPSET_CONCAT(a, b)
542
543#endif /*_IP_SET_H */
544