1 /* Copyright (C) 2000-2002 Joakim Axelsson <gozem@linux.nu>
2  *                         Patrick Schaaf <bof@bof.de>
3  *                         Martin Josefsson <gandalf@wlug.westbo.se>
4  * Copyright (C) 2003-2013 Jozsef Kadlecsik <kadlec@blackhole.kfki.hu>
5  *
6  * This program is free software; you can redistribute it and/or modify
7  * it under the terms of the GNU General Public License version 2 as
8  * published by the Free Software Foundation.
9  */
10 #ifndef _IP_SET_H
11 #define _IP_SET_H
12 
13 #include <linux/ip.h>
14 #include <linux/ipv6.h>
15 #include <linux/netlink.h>
16 #include <linux/netfilter.h>
17 #include <linux/netfilter/x_tables.h>
18 #include <linux/stringify.h>
19 #include <linux/vmalloc.h>
20 #include <net/netlink.h>
21 #include <uapi/linux/netfilter/ipset/ip_set.h>
22 
23 #define _IP_SET_MODULE_DESC(a, b, c)		\
24 	MODULE_DESCRIPTION(a " type of IP sets, revisions " b "-" c)
25 #define IP_SET_MODULE_DESC(a, b, c)		\
26 	_IP_SET_MODULE_DESC(a, __stringify(b), __stringify(c))
27 
28 /* Set features */
29 enum ip_set_feature {
30 	IPSET_TYPE_IP_FLAG = 0,
31 	IPSET_TYPE_IP = (1 << IPSET_TYPE_IP_FLAG),
32 	IPSET_TYPE_PORT_FLAG = 1,
33 	IPSET_TYPE_PORT = (1 << IPSET_TYPE_PORT_FLAG),
34 	IPSET_TYPE_MAC_FLAG = 2,
35 	IPSET_TYPE_MAC = (1 << IPSET_TYPE_MAC_FLAG),
36 	IPSET_TYPE_IP2_FLAG = 3,
37 	IPSET_TYPE_IP2 = (1 << IPSET_TYPE_IP2_FLAG),
38 	IPSET_TYPE_NAME_FLAG = 4,
39 	IPSET_TYPE_NAME = (1 << IPSET_TYPE_NAME_FLAG),
40 	IPSET_TYPE_IFACE_FLAG = 5,
41 	IPSET_TYPE_IFACE = (1 << IPSET_TYPE_IFACE_FLAG),
42 	IPSET_TYPE_MARK_FLAG = 6,
43 	IPSET_TYPE_MARK = (1 << IPSET_TYPE_MARK_FLAG),
44 	IPSET_TYPE_NOMATCH_FLAG = 7,
45 	IPSET_TYPE_NOMATCH = (1 << IPSET_TYPE_NOMATCH_FLAG),
46 	/* Strictly speaking not a feature, but a flag for dumping:
47 	 * this settype must be dumped last */
48 	IPSET_DUMP_LAST_FLAG = 8,
49 	IPSET_DUMP_LAST = (1 << IPSET_DUMP_LAST_FLAG),
50 };
51 
52 /* Set extensions */
53 enum ip_set_extension {
54 	IPSET_EXT_BIT_TIMEOUT = 0,
55 	IPSET_EXT_TIMEOUT = (1 << IPSET_EXT_BIT_TIMEOUT),
56 	IPSET_EXT_BIT_COUNTER = 1,
57 	IPSET_EXT_COUNTER = (1 << IPSET_EXT_BIT_COUNTER),
58 	IPSET_EXT_BIT_COMMENT = 2,
59 	IPSET_EXT_COMMENT = (1 << IPSET_EXT_BIT_COMMENT),
60 	IPSET_EXT_BIT_SKBINFO = 3,
61 	IPSET_EXT_SKBINFO = (1 << IPSET_EXT_BIT_SKBINFO),
62 	/* Mark set with an extension which needs to call destroy */
63 	IPSET_EXT_BIT_DESTROY = 7,
64 	IPSET_EXT_DESTROY = (1 << IPSET_EXT_BIT_DESTROY),
65 };
66 
67 #define SET_WITH_TIMEOUT(s)	((s)->extensions & IPSET_EXT_TIMEOUT)
68 #define SET_WITH_COUNTER(s)	((s)->extensions & IPSET_EXT_COUNTER)
69 #define SET_WITH_COMMENT(s)	((s)->extensions & IPSET_EXT_COMMENT)
70 #define SET_WITH_SKBINFO(s)	((s)->extensions & IPSET_EXT_SKBINFO)
71 #define SET_WITH_FORCEADD(s)	((s)->flags & IPSET_CREATE_FLAG_FORCEADD)
72 
73 /* Extension id, in size order */
74 enum ip_set_ext_id {
75 	IPSET_EXT_ID_COUNTER = 0,
76 	IPSET_EXT_ID_TIMEOUT,
77 	IPSET_EXT_ID_SKBINFO,
78 	IPSET_EXT_ID_COMMENT,
79 	IPSET_EXT_ID_MAX,
80 };
81 
82 struct ip_set;
83 
84 /* Extension type */
85 struct ip_set_ext_type {
86 	/* Destroy extension private data (can be NULL) */
87 	void (*destroy)(struct ip_set *set, void *ext);
88 	enum ip_set_extension type;
89 	enum ipset_cadt_flags flag;
90 	/* Size and minimal alignment */
91 	u8 len;
92 	u8 align;
93 };
94 
95 extern const struct ip_set_ext_type ip_set_extensions[];
96 
97 struct ip_set_counter {
98 	atomic64_t bytes;
99 	atomic64_t packets;
100 };
101 
102 struct ip_set_comment_rcu {
103 	struct rcu_head rcu;
104 	char str[0];
105 };
106 
107 struct ip_set_comment {
108 	struct ip_set_comment_rcu __rcu *c;
109 };
110 
111 struct ip_set_skbinfo {
112 	u32 skbmark;
113 	u32 skbmarkmask;
114 	u32 skbprio;
115 	u16 skbqueue;
116 	u16 __pad;
117 };
118 
119 struct ip_set_ext {
120 	struct ip_set_skbinfo skbinfo;
121 	u64 packets;
122 	u64 bytes;
123 	char *comment;
124 	u32 timeout;
125 	u8 packets_op;
126 	u8 bytes_op;
127 };
128 
129 struct ip_set;
130 
131 #define ext_timeout(e, s)	\
132 ((unsigned long *)(((void *)(e)) + (s)->offset[IPSET_EXT_ID_TIMEOUT]))
133 #define ext_counter(e, s)	\
134 ((struct ip_set_counter *)(((void *)(e)) + (s)->offset[IPSET_EXT_ID_COUNTER]))
135 #define ext_comment(e, s)	\
136 ((struct ip_set_comment *)(((void *)(e)) + (s)->offset[IPSET_EXT_ID_COMMENT]))
137 #define ext_skbinfo(e, s)	\
138 ((struct ip_set_skbinfo *)(((void *)(e)) + (s)->offset[IPSET_EXT_ID_SKBINFO]))
139 
140 typedef int (*ipset_adtfn)(struct ip_set *set, void *value,
141 			   const struct ip_set_ext *ext,
142 			   struct ip_set_ext *mext, u32 cmdflags);
143 
144 /* Kernel API function options */
145 struct ip_set_adt_opt {
146 	u8 family;		/* Actual protocol family */
147 	u8 dim;			/* Dimension of match/target */
148 	u8 flags;		/* Direction and negation flags */
149 	u32 cmdflags;		/* Command-like flags */
150 	struct ip_set_ext ext;	/* Extensions */
151 };
152 
153 /* Set type, variant-specific part */
154 struct ip_set_type_variant {
155 	/* Kernelspace: test/add/del entries
156 	 *		returns negative error code,
157 	 *			zero for no match/success to add/delete
158 	 *			positive for matching element */
159 	int (*kadt)(struct ip_set *set, const struct sk_buff *skb,
160 		    const struct xt_action_param *par,
161 		    enum ipset_adt adt, struct ip_set_adt_opt *opt);
162 
163 	/* Userspace: test/add/del entries
164 	 *		returns negative error code,
165 	 *			zero for no match/success to add/delete
166 	 *			positive for matching element */
167 	int (*uadt)(struct ip_set *set, struct nlattr *tb[],
168 		    enum ipset_adt adt, u32 *lineno, u32 flags, bool retried);
169 
170 	/* Low level add/del/test functions */
171 	ipset_adtfn adt[IPSET_ADT_MAX];
172 
173 	/* When adding entries and set is full, try to resize the set */
174 	int (*resize)(struct ip_set *set, bool retried);
175 	/* Destroy the set */
176 	void (*destroy)(struct ip_set *set);
177 	/* Flush the elements */
178 	void (*flush)(struct ip_set *set);
179 	/* Expire entries before listing */
180 	void (*expire)(struct ip_set *set);
181 	/* List set header data */
182 	int (*head)(struct ip_set *set, struct sk_buff *skb);
183 	/* List elements */
184 	int (*list)(const struct ip_set *set, struct sk_buff *skb,
185 		    struct netlink_callback *cb);
186 	/* Keep listing private when resizing runs parallel */
187 	void (*uref)(struct ip_set *set, struct netlink_callback *cb,
188 		     bool start);
189 
190 	/* Return true if "b" set is the same as "a"
191 	 * according to the create set parameters */
192 	bool (*same_set)(const struct ip_set *a, const struct ip_set *b);
193 };
194 
195 /* The core set type structure */
196 struct ip_set_type {
197 	struct list_head list;
198 
199 	/* Typename */
200 	char name[IPSET_MAXNAMELEN];
201 	/* Protocol version */
202 	u8 protocol;
203 	/* Set type dimension */
204 	u8 dimension;
205 	/*
206 	 * Supported family: may be NFPROTO_UNSPEC for both
207 	 * NFPROTO_IPV4/NFPROTO_IPV6.
208 	 */
209 	u8 family;
210 	/* Type revisions */
211 	u8 revision_min, revision_max;
212 	/* Set features to control swapping */
213 	u16 features;
214 
215 	/* Create set */
216 	int (*create)(struct net *net, struct ip_set *set,
217 		      struct nlattr *tb[], u32 flags);
218 
219 	/* Attribute policies */
220 	const struct nla_policy create_policy[IPSET_ATTR_CREATE_MAX + 1];
221 	const struct nla_policy adt_policy[IPSET_ATTR_ADT_MAX + 1];
222 
223 	/* Set this to THIS_MODULE if you are a module, otherwise NULL */
224 	struct module *me;
225 };
226 
227 /* register and unregister set type */
228 extern int ip_set_type_register(struct ip_set_type *set_type);
229 extern void ip_set_type_unregister(struct ip_set_type *set_type);
230 
231 /* A generic IP set */
232 struct ip_set {
233 	/* The name of the set */
234 	char name[IPSET_MAXNAMELEN];
235 	/* Lock protecting the set data */
236 	spinlock_t lock;
237 	/* References to the set */
238 	u32 ref;
239 	/* References to the set for netlink events like dump,
240 	 * ref can be swapped out by ip_set_swap
241 	 */
242 	u32 ref_netlink;
243 	/* The core set type */
244 	struct ip_set_type *type;
245 	/* The type variant doing the real job */
246 	const struct ip_set_type_variant *variant;
247 	/* The actual INET family of the set */
248 	u8 family;
249 	/* The type revision */
250 	u8 revision;
251 	/* Extensions */
252 	u8 extensions;
253 	/* Create flags */
254 	u8 flags;
255 	/* Default timeout value, if enabled */
256 	u32 timeout;
257 	/* Number of elements (vs timeout) */
258 	u32 elements;
259 	/* Size of the dynamic extensions (vs timeout) */
260 	size_t ext_size;
261 	/* Element data size */
262 	size_t dsize;
263 	/* Offsets to extensions in elements */
264 	size_t offset[IPSET_EXT_ID_MAX];
265 	/* The type specific data */
266 	void *data;
267 };
268 
269 static inline void
ip_set_ext_destroy(struct ip_set * set,void * data)270 ip_set_ext_destroy(struct ip_set *set, void *data)
271 {
272 	/* Check that the extension is enabled for the set and
273 	 * call it's destroy function for its extension part in data.
274 	 */
275 	if (SET_WITH_COMMENT(set))
276 		ip_set_extensions[IPSET_EXT_ID_COMMENT].destroy(
277 			set, ext_comment(data, set));
278 }
279 
280 static inline int
ip_set_put_flags(struct sk_buff * skb,struct ip_set * set)281 ip_set_put_flags(struct sk_buff *skb, struct ip_set *set)
282 {
283 	u32 cadt_flags = 0;
284 
285 	if (SET_WITH_TIMEOUT(set))
286 		if (unlikely(nla_put_net32(skb, IPSET_ATTR_TIMEOUT,
287 					   htonl(set->timeout))))
288 			return -EMSGSIZE;
289 	if (SET_WITH_COUNTER(set))
290 		cadt_flags |= IPSET_FLAG_WITH_COUNTERS;
291 	if (SET_WITH_COMMENT(set))
292 		cadt_flags |= IPSET_FLAG_WITH_COMMENT;
293 	if (SET_WITH_SKBINFO(set))
294 		cadt_flags |= IPSET_FLAG_WITH_SKBINFO;
295 	if (SET_WITH_FORCEADD(set))
296 		cadt_flags |= IPSET_FLAG_WITH_FORCEADD;
297 
298 	if (!cadt_flags)
299 		return 0;
300 	return nla_put_net32(skb, IPSET_ATTR_CADT_FLAGS, htonl(cadt_flags));
301 }
302 
303 /* Netlink CB args */
304 enum {
305 	IPSET_CB_NET = 0,	/* net namespace */
306 	IPSET_CB_DUMP,		/* dump single set/all sets */
307 	IPSET_CB_INDEX,		/* set index */
308 	IPSET_CB_PRIVATE,	/* set private data */
309 	IPSET_CB_ARG0,		/* type specific */
310 	IPSET_CB_ARG1,
311 };
312 
313 /* register and unregister set references */
314 extern ip_set_id_t ip_set_get_byname(struct net *net,
315 				     const char *name, struct ip_set **set);
316 extern void ip_set_put_byindex(struct net *net, ip_set_id_t index);
317 extern void ip_set_name_byindex(struct net *net, ip_set_id_t index, char *name);
318 extern ip_set_id_t ip_set_nfnl_get_byindex(struct net *net, ip_set_id_t index);
319 extern void ip_set_nfnl_put(struct net *net, ip_set_id_t index);
320 
321 /* API for iptables set match, and SET target */
322 
323 extern int ip_set_add(ip_set_id_t id, const struct sk_buff *skb,
324 		      const struct xt_action_param *par,
325 		      struct ip_set_adt_opt *opt);
326 extern int ip_set_del(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 extern int ip_set_test(ip_set_id_t id, const struct sk_buff *skb,
330 		       const struct xt_action_param *par,
331 		       struct ip_set_adt_opt *opt);
332 
333 /* Utility functions */
334 extern void *ip_set_alloc(size_t size);
335 extern void ip_set_free(void *members);
336 extern int ip_set_get_ipaddr4(struct nlattr *nla,  __be32 *ipaddr);
337 extern int ip_set_get_ipaddr6(struct nlattr *nla, union nf_inet_addr *ipaddr);
338 extern size_t ip_set_elem_len(struct ip_set *set, struct nlattr *tb[],
339 			      size_t len, size_t align);
340 extern int ip_set_get_extensions(struct ip_set *set, struct nlattr *tb[],
341 				 struct ip_set_ext *ext);
342 extern int ip_set_put_extensions(struct sk_buff *skb, const struct ip_set *set,
343 				 const void *e, bool active);
344 extern bool ip_set_match_extensions(struct ip_set *set,
345 				    const struct ip_set_ext *ext,
346 				    struct ip_set_ext *mext,
347 				    u32 flags, void *data);
348 
349 static inline int
ip_set_get_hostipaddr4(struct nlattr * nla,u32 * ipaddr)350 ip_set_get_hostipaddr4(struct nlattr *nla, u32 *ipaddr)
351 {
352 	__be32 ip;
353 	int ret = ip_set_get_ipaddr4(nla, &ip);
354 
355 	if (ret)
356 		return ret;
357 	*ipaddr = ntohl(ip);
358 	return 0;
359 }
360 
361 /* Ignore IPSET_ERR_EXIST errors if asked to do so? */
362 static inline bool
ip_set_eexist(int ret,u32 flags)363 ip_set_eexist(int ret, u32 flags)
364 {
365 	return ret == -IPSET_ERR_EXIST && (flags & IPSET_FLAG_EXIST);
366 }
367 
368 /* Match elements marked with nomatch */
369 static inline bool
ip_set_enomatch(int ret,u32 flags,enum ipset_adt adt,struct ip_set * set)370 ip_set_enomatch(int ret, u32 flags, enum ipset_adt adt, struct ip_set *set)
371 {
372 	return adt == IPSET_TEST &&
373 	       (set->type->features & IPSET_TYPE_NOMATCH) &&
374 	       ((flags >> 16) & IPSET_FLAG_NOMATCH) &&
375 	       (ret > 0 || ret == -ENOTEMPTY);
376 }
377 
378 /* Check the NLA_F_NET_BYTEORDER flag */
379 static inline bool
ip_set_attr_netorder(struct nlattr * tb[],int type)380 ip_set_attr_netorder(struct nlattr *tb[], int type)
381 {
382 	return tb[type] && (tb[type]->nla_type & NLA_F_NET_BYTEORDER);
383 }
384 
385 static inline bool
ip_set_optattr_netorder(struct nlattr * tb[],int type)386 ip_set_optattr_netorder(struct nlattr *tb[], int type)
387 {
388 	return !tb[type] || (tb[type]->nla_type & NLA_F_NET_BYTEORDER);
389 }
390 
391 /* Useful converters */
392 static inline u32
ip_set_get_h32(const struct nlattr * attr)393 ip_set_get_h32(const struct nlattr *attr)
394 {
395 	return ntohl(nla_get_be32(attr));
396 }
397 
398 static inline u16
ip_set_get_h16(const struct nlattr * attr)399 ip_set_get_h16(const struct nlattr *attr)
400 {
401 	return ntohs(nla_get_be16(attr));
402 }
403 
404 #define ipset_nest_start(skb, attr) nla_nest_start(skb, attr | NLA_F_NESTED)
405 #define ipset_nest_end(skb, start)  nla_nest_end(skb, start)
406 
nla_put_ipaddr4(struct sk_buff * skb,int type,__be32 ipaddr)407 static inline int nla_put_ipaddr4(struct sk_buff *skb, int type, __be32 ipaddr)
408 {
409 	struct nlattr *__nested = ipset_nest_start(skb, type);
410 	int ret;
411 
412 	if (!__nested)
413 		return -EMSGSIZE;
414 	ret = nla_put_in_addr(skb, IPSET_ATTR_IPADDR_IPV4, ipaddr);
415 	if (!ret)
416 		ipset_nest_end(skb, __nested);
417 	return ret;
418 }
419 
nla_put_ipaddr6(struct sk_buff * skb,int type,const struct in6_addr * ipaddrptr)420 static inline int nla_put_ipaddr6(struct sk_buff *skb, int type,
421 				  const struct in6_addr *ipaddrptr)
422 {
423 	struct nlattr *__nested = ipset_nest_start(skb, type);
424 	int ret;
425 
426 	if (!__nested)
427 		return -EMSGSIZE;
428 	ret = nla_put_in6_addr(skb, IPSET_ATTR_IPADDR_IPV6, ipaddrptr);
429 	if (!ret)
430 		ipset_nest_end(skb, __nested);
431 	return ret;
432 }
433 
434 /* Get address from skbuff */
435 static inline __be32
ip4addr(const struct sk_buff * skb,bool src)436 ip4addr(const struct sk_buff *skb, bool src)
437 {
438 	return src ? ip_hdr(skb)->saddr : ip_hdr(skb)->daddr;
439 }
440 
441 static inline void
ip4addrptr(const struct sk_buff * skb,bool src,__be32 * addr)442 ip4addrptr(const struct sk_buff *skb, bool src, __be32 *addr)
443 {
444 	*addr = src ? ip_hdr(skb)->saddr : ip_hdr(skb)->daddr;
445 }
446 
447 static inline void
ip6addrptr(const struct sk_buff * skb,bool src,struct in6_addr * addr)448 ip6addrptr(const struct sk_buff *skb, bool src, struct in6_addr *addr)
449 {
450 	memcpy(addr, src ? &ipv6_hdr(skb)->saddr : &ipv6_hdr(skb)->daddr,
451 	       sizeof(*addr));
452 }
453 
454 #include <linux/netfilter/ipset/ip_set_timeout.h>
455 #include <linux/netfilter/ipset/ip_set_comment.h>
456 #include <linux/netfilter/ipset/ip_set_counter.h>
457 #include <linux/netfilter/ipset/ip_set_skbinfo.h>
458 
459 #define IP_SET_INIT_KEXT(skb, opt, set)			\
460 	{ .bytes = (skb)->len, .packets = 1,		\
461 	  .timeout = ip_set_adt_opt_timeout(opt, set) }
462 
463 #define IP_SET_INIT_UEXT(set)				\
464 	{ .bytes = ULLONG_MAX, .packets = ULLONG_MAX,	\
465 	  .timeout = (set)->timeout }
466 
467 #define IPSET_CONCAT(a, b)		a##b
468 #define IPSET_TOKEN(a, b)		IPSET_CONCAT(a, b)
469 
470 #endif /*_IP_SET_H */
471