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 */
26 enum 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 */
50 enum 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 */
71 enum 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 
79 struct ip_set;
80 
81 /* Extension type */
82 struct 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 
92 extern const struct ip_set_ext_type ip_set_extensions[];
93 
94 struct ip_set_counter {
95 	atomic64_t bytes;
96 	atomic64_t packets;
97 };
98 
99 struct ip_set_comment_rcu {
100 	struct rcu_head rcu;
101 	char str[];
102 };
103 
104 struct ip_set_comment {
105 	struct ip_set_comment_rcu __rcu *c;
106 };
107 
108 struct ip_set_skbinfo {
109 	u32 skbmark;
110 	u32 skbmarkmask;
111 	u32 skbprio;
112 	u16 skbqueue;
113 	u16 __pad;
114 };
115 
116 struct 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 
136 typedef 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 */
141 struct 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 */
150 struct 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 	/* Region-locking is used */
190 	bool region_lock;
191 };
192 
193 struct ip_set_region {
194 	spinlock_t lock;	/* Region lock */
195 	size_t ext_size;	/* Size of the dynamic extensions */
196 	u32 elements;		/* Number of elements vs timeout */
197 };
198 
199 /* Max range where every element is added/deleted in one step */
200 #define IPSET_MAX_RANGE		(1<<20)
201 
202 /* The max revision number supported by any set type + 1 */
203 #define IPSET_REVISION_MAX	9
204 
205 /* The core set type structure */
206 struct ip_set_type {
207 	struct list_head list;
208 
209 	/* Typename */
210 	char name[IPSET_MAXNAMELEN];
211 	/* Protocol version */
212 	u8 protocol;
213 	/* Set type dimension */
214 	u8 dimension;
215 	/*
216 	 * Supported family: may be NFPROTO_UNSPEC for both
217 	 * NFPROTO_IPV4/NFPROTO_IPV6.
218 	 */
219 	u8 family;
220 	/* Type revisions */
221 	u8 revision_min, revision_max;
222 	/* Revision-specific supported (create) flags */
223 	u8 create_flags[IPSET_REVISION_MAX+1];
224 	/* Set features to control swapping */
225 	u16 features;
226 
227 	/* Create set */
228 	int (*create)(struct net *net, struct ip_set *set,
229 		      struct nlattr *tb[], u32 flags);
230 
231 	/* Attribute policies */
232 	const struct nla_policy create_policy[IPSET_ATTR_CREATE_MAX + 1];
233 	const struct nla_policy adt_policy[IPSET_ATTR_ADT_MAX + 1];
234 
235 	/* Set this to THIS_MODULE if you are a module, otherwise NULL */
236 	struct module *me;
237 };
238 
239 /* register and unregister set type */
240 extern int ip_set_type_register(struct ip_set_type *set_type);
241 extern void ip_set_type_unregister(struct ip_set_type *set_type);
242 
243 /* A generic IP set */
244 struct ip_set {
245 	/* The name of the set */
246 	char name[IPSET_MAXNAMELEN];
247 	/* Lock protecting the set data */
248 	spinlock_t lock;
249 	/* References to the set */
250 	u32 ref;
251 	/* References to the set for netlink events like dump,
252 	 * ref can be swapped out by ip_set_swap
253 	 */
254 	u32 ref_netlink;
255 	/* The core set type */
256 	struct ip_set_type *type;
257 	/* The type variant doing the real job */
258 	const struct ip_set_type_variant *variant;
259 	/* The actual INET family of the set */
260 	u8 family;
261 	/* The type revision */
262 	u8 revision;
263 	/* Extensions */
264 	u8 extensions;
265 	/* Create flags */
266 	u8 flags;
267 	/* Default timeout value, if enabled */
268 	u32 timeout;
269 	/* Number of elements (vs timeout) */
270 	u32 elements;
271 	/* Size of the dynamic extensions (vs timeout) */
272 	size_t ext_size;
273 	/* Element data size */
274 	size_t dsize;
275 	/* Offsets to extensions in elements */
276 	size_t offset[IPSET_EXT_ID_MAX];
277 	/* The type specific data */
278 	void *data;
279 };
280 
281 static inline void
ip_set_ext_destroy(struct ip_set * set,void * data)282 ip_set_ext_destroy(struct ip_set *set, void *data)
283 {
284 	/* Check that the extension is enabled for the set and
285 	 * call it's destroy function for its extension part in data.
286 	 */
287 	if (SET_WITH_COMMENT(set)) {
288 		struct ip_set_comment *c = ext_comment(data, set);
289 
290 		ip_set_extensions[IPSET_EXT_ID_COMMENT].destroy(set, c);
291 	}
292 }
293 
294 int ip_set_put_flags(struct sk_buff *skb, struct ip_set *set);
295 
296 /* Netlink CB args */
297 enum {
298 	IPSET_CB_NET = 0,	/* net namespace */
299 	IPSET_CB_PROTO,		/* ipset protocol */
300 	IPSET_CB_DUMP,		/* dump single set/all sets */
301 	IPSET_CB_INDEX,		/* set index */
302 	IPSET_CB_PRIVATE,	/* set private data */
303 	IPSET_CB_ARG0,		/* type specific */
304 };
305 
306 /* register and unregister set references */
307 extern ip_set_id_t ip_set_get_byname(struct net *net,
308 				     const char *name, struct ip_set **set);
309 extern void ip_set_put_byindex(struct net *net, ip_set_id_t index);
310 extern void ip_set_name_byindex(struct net *net, ip_set_id_t index, char *name);
311 extern ip_set_id_t ip_set_nfnl_get_byindex(struct net *net, ip_set_id_t index);
312 extern void ip_set_nfnl_put(struct net *net, ip_set_id_t index);
313 
314 /* API for iptables set match, and SET target */
315 
316 extern int ip_set_add(ip_set_id_t id, const struct sk_buff *skb,
317 		      const struct xt_action_param *par,
318 		      struct ip_set_adt_opt *opt);
319 extern int ip_set_del(ip_set_id_t id, const struct sk_buff *skb,
320 		      const struct xt_action_param *par,
321 		      struct ip_set_adt_opt *opt);
322 extern int ip_set_test(ip_set_id_t id, const struct sk_buff *skb,
323 		       const struct xt_action_param *par,
324 		       struct ip_set_adt_opt *opt);
325 
326 /* Utility functions */
327 extern void *ip_set_alloc(size_t size);
328 extern void ip_set_free(void *members);
329 extern int ip_set_get_ipaddr4(struct nlattr *nla,  __be32 *ipaddr);
330 extern int ip_set_get_ipaddr6(struct nlattr *nla, union nf_inet_addr *ipaddr);
331 extern size_t ip_set_elem_len(struct ip_set *set, struct nlattr *tb[],
332 			      size_t len, size_t align);
333 extern int ip_set_get_extensions(struct ip_set *set, struct nlattr *tb[],
334 				 struct ip_set_ext *ext);
335 extern int ip_set_put_extensions(struct sk_buff *skb, const struct ip_set *set,
336 				 const void *e, bool active);
337 extern bool ip_set_match_extensions(struct ip_set *set,
338 				    const struct ip_set_ext *ext,
339 				    struct ip_set_ext *mext,
340 				    u32 flags, void *data);
341 
342 static inline int
ip_set_get_hostipaddr4(struct nlattr * nla,u32 * ipaddr)343 ip_set_get_hostipaddr4(struct nlattr *nla, u32 *ipaddr)
344 {
345 	__be32 ip;
346 	int ret = ip_set_get_ipaddr4(nla, &ip);
347 
348 	if (ret)
349 		return ret;
350 	*ipaddr = ntohl(ip);
351 	return 0;
352 }
353 
354 /* Ignore IPSET_ERR_EXIST errors if asked to do so? */
355 static inline bool
ip_set_eexist(int ret,u32 flags)356 ip_set_eexist(int ret, u32 flags)
357 {
358 	return ret == -IPSET_ERR_EXIST && (flags & IPSET_FLAG_EXIST);
359 }
360 
361 /* Match elements marked with nomatch */
362 static inline bool
ip_set_enomatch(int ret,u32 flags,enum ipset_adt adt,struct ip_set * set)363 ip_set_enomatch(int ret, u32 flags, enum ipset_adt adt, struct ip_set *set)
364 {
365 	return adt == IPSET_TEST &&
366 	       (set->type->features & IPSET_TYPE_NOMATCH) &&
367 	       ((flags >> 16) & IPSET_FLAG_NOMATCH) &&
368 	       (ret > 0 || ret == -ENOTEMPTY);
369 }
370 
371 /* Check the NLA_F_NET_BYTEORDER flag */
372 static inline bool
ip_set_attr_netorder(struct nlattr * tb[],int type)373 ip_set_attr_netorder(struct nlattr *tb[], int type)
374 {
375 	return tb[type] && (tb[type]->nla_type & NLA_F_NET_BYTEORDER);
376 }
377 
378 static inline bool
ip_set_optattr_netorder(struct nlattr * tb[],int type)379 ip_set_optattr_netorder(struct nlattr *tb[], int type)
380 {
381 	return !tb[type] || (tb[type]->nla_type & NLA_F_NET_BYTEORDER);
382 }
383 
384 /* Useful converters */
385 static inline u32
ip_set_get_h32(const struct nlattr * attr)386 ip_set_get_h32(const struct nlattr *attr)
387 {
388 	return ntohl(nla_get_be32(attr));
389 }
390 
391 static inline u16
ip_set_get_h16(const struct nlattr * attr)392 ip_set_get_h16(const struct nlattr *attr)
393 {
394 	return ntohs(nla_get_be16(attr));
395 }
396 
nla_put_ipaddr4(struct sk_buff * skb,int type,__be32 ipaddr)397 static inline int nla_put_ipaddr4(struct sk_buff *skb, int type, __be32 ipaddr)
398 {
399 	struct nlattr *__nested = nla_nest_start(skb, type);
400 	int ret;
401 
402 	if (!__nested)
403 		return -EMSGSIZE;
404 	ret = nla_put_in_addr(skb, IPSET_ATTR_IPADDR_IPV4, ipaddr);
405 	if (!ret)
406 		nla_nest_end(skb, __nested);
407 	return ret;
408 }
409 
nla_put_ipaddr6(struct sk_buff * skb,int type,const struct in6_addr * ipaddrptr)410 static inline int nla_put_ipaddr6(struct sk_buff *skb, int type,
411 				  const struct in6_addr *ipaddrptr)
412 {
413 	struct nlattr *__nested = nla_nest_start(skb, type);
414 	int ret;
415 
416 	if (!__nested)
417 		return -EMSGSIZE;
418 	ret = nla_put_in6_addr(skb, IPSET_ATTR_IPADDR_IPV6, ipaddrptr);
419 	if (!ret)
420 		nla_nest_end(skb, __nested);
421 	return ret;
422 }
423 
424 /* Get address from skbuff */
425 static inline __be32
ip4addr(const struct sk_buff * skb,bool src)426 ip4addr(const struct sk_buff *skb, bool src)
427 {
428 	return src ? ip_hdr(skb)->saddr : ip_hdr(skb)->daddr;
429 }
430 
431 static inline void
ip4addrptr(const struct sk_buff * skb,bool src,__be32 * addr)432 ip4addrptr(const struct sk_buff *skb, bool src, __be32 *addr)
433 {
434 	*addr = src ? ip_hdr(skb)->saddr : ip_hdr(skb)->daddr;
435 }
436 
437 static inline void
ip6addrptr(const struct sk_buff * skb,bool src,struct in6_addr * addr)438 ip6addrptr(const struct sk_buff *skb, bool src, struct in6_addr *addr)
439 {
440 	memcpy(addr, src ? &ipv6_hdr(skb)->saddr : &ipv6_hdr(skb)->daddr,
441 	       sizeof(*addr));
442 }
443 
444 /* How often should the gc be run by default */
445 #define IPSET_GC_TIME			(3 * 60)
446 
447 /* Timeout period depending on the timeout value of the given set */
448 #define IPSET_GC_PERIOD(timeout) \
449 	((timeout/3) ? min_t(u32, (timeout)/3, IPSET_GC_TIME) : 1)
450 
451 /* Entry is set with no timeout value */
452 #define IPSET_ELEM_PERMANENT	0
453 
454 /* Set is defined with timeout support: timeout value may be 0 */
455 #define IPSET_NO_TIMEOUT	UINT_MAX
456 
457 /* Max timeout value, see msecs_to_jiffies() in jiffies.h */
458 #define IPSET_MAX_TIMEOUT	(UINT_MAX >> 1)/MSEC_PER_SEC
459 
460 #define ip_set_adt_opt_timeout(opt, set)	\
461 ((opt)->ext.timeout != IPSET_NO_TIMEOUT ? (opt)->ext.timeout : (set)->timeout)
462 
463 static inline unsigned int
ip_set_timeout_uget(struct nlattr * tb)464 ip_set_timeout_uget(struct nlattr *tb)
465 {
466 	unsigned int timeout = ip_set_get_h32(tb);
467 
468 	/* Normalize to fit into jiffies */
469 	if (timeout > IPSET_MAX_TIMEOUT)
470 		timeout = IPSET_MAX_TIMEOUT;
471 
472 	return timeout;
473 }
474 
475 static inline bool
ip_set_timeout_expired(const unsigned long * t)476 ip_set_timeout_expired(const unsigned long *t)
477 {
478 	return *t != IPSET_ELEM_PERMANENT && time_is_before_jiffies(*t);
479 }
480 
481 static inline void
ip_set_timeout_set(unsigned long * timeout,u32 value)482 ip_set_timeout_set(unsigned long *timeout, u32 value)
483 {
484 	unsigned long t;
485 
486 	if (!value) {
487 		*timeout = IPSET_ELEM_PERMANENT;
488 		return;
489 	}
490 
491 	t = msecs_to_jiffies(value * MSEC_PER_SEC) + jiffies;
492 	if (t == IPSET_ELEM_PERMANENT)
493 		/* Bingo! :-) */
494 		t--;
495 	*timeout = t;
496 }
497 
498 void ip_set_init_comment(struct ip_set *set, struct ip_set_comment *comment,
499 			 const struct ip_set_ext *ext);
500 
501 static inline void
ip_set_init_counter(struct ip_set_counter * counter,const struct ip_set_ext * ext)502 ip_set_init_counter(struct ip_set_counter *counter,
503 		    const struct ip_set_ext *ext)
504 {
505 	if (ext->bytes != ULLONG_MAX)
506 		atomic64_set(&(counter)->bytes, (long long)(ext->bytes));
507 	if (ext->packets != ULLONG_MAX)
508 		atomic64_set(&(counter)->packets, (long long)(ext->packets));
509 }
510 
511 static inline void
ip_set_init_skbinfo(struct ip_set_skbinfo * skbinfo,const struct ip_set_ext * ext)512 ip_set_init_skbinfo(struct ip_set_skbinfo *skbinfo,
513 		    const struct ip_set_ext *ext)
514 {
515 	*skbinfo = ext->skbinfo;
516 }
517 
518 #define IP_SET_INIT_KEXT(skb, opt, set)			\
519 	{ .bytes = (skb)->len, .packets = 1, .target = true,\
520 	  .timeout = ip_set_adt_opt_timeout(opt, set) }
521 
522 #define IP_SET_INIT_UEXT(set)				\
523 	{ .bytes = ULLONG_MAX, .packets = ULLONG_MAX,	\
524 	  .timeout = (set)->timeout }
525 
526 #define IPSET_CONCAT(a, b)		a##b
527 #define IPSET_TOKEN(a, b)		IPSET_CONCAT(a, b)
528 
529 #endif /*_IP_SET_H */
530