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 struct ip_set;
128
129 #define ext_timeout(e, s) \
130 ((unsigned long *)(((void *)(e)) + (s)->offset[IPSET_EXT_ID_TIMEOUT]))
131 #define ext_counter(e, s) \
132 ((struct ip_set_counter *)(((void *)(e)) + (s)->offset[IPSET_EXT_ID_COUNTER]))
133 #define ext_comment(e, s) \
134 ((struct ip_set_comment *)(((void *)(e)) + (s)->offset[IPSET_EXT_ID_COMMENT]))
135 #define ext_skbinfo(e, s) \
136 ((struct ip_set_skbinfo *)(((void *)(e)) + (s)->offset[IPSET_EXT_ID_SKBINFO]))
137
138 typedef int (*ipset_adtfn)(struct ip_set *set, void *value,
139 const struct ip_set_ext *ext,
140 struct ip_set_ext *mext, u32 cmdflags);
141
142 /* Kernel API function options */
143 struct ip_set_adt_opt {
144 u8 family; /* Actual protocol family */
145 u8 dim; /* Dimension of match/target */
146 u8 flags; /* Direction and negation flags */
147 u32 cmdflags; /* Command-like flags */
148 struct ip_set_ext ext; /* Extensions */
149 };
150
151 /* Set type, variant-specific part */
152 struct ip_set_type_variant {
153 /* Kernelspace: test/add/del entries
154 * returns negative error code,
155 * zero for no match/success to add/delete
156 * positive for matching element */
157 int (*kadt)(struct ip_set *set, const struct sk_buff *skb,
158 const struct xt_action_param *par,
159 enum ipset_adt adt, struct ip_set_adt_opt *opt);
160
161 /* Userspace: test/add/del entries
162 * returns negative error code,
163 * zero for no match/success to add/delete
164 * positive for matching element */
165 int (*uadt)(struct ip_set *set, struct nlattr *tb[],
166 enum ipset_adt adt, u32 *lineno, u32 flags, bool retried);
167
168 /* Low level add/del/test functions */
169 ipset_adtfn adt[IPSET_ADT_MAX];
170
171 /* When adding entries and set is full, try to resize the set */
172 int (*resize)(struct ip_set *set, bool retried);
173 /* Destroy the set */
174 void (*destroy)(struct ip_set *set);
175 /* Flush the elements */
176 void (*flush)(struct ip_set *set);
177 /* Expire entries before listing */
178 void (*expire)(struct ip_set *set);
179 /* List set header data */
180 int (*head)(struct ip_set *set, struct sk_buff *skb);
181 /* List elements */
182 int (*list)(const struct ip_set *set, struct sk_buff *skb,
183 struct netlink_callback *cb);
184 /* Keep listing private when resizing runs parallel */
185 void (*uref)(struct ip_set *set, struct netlink_callback *cb,
186 bool start);
187
188 /* Return true if "b" set is the same as "a"
189 * according to the create set parameters */
190 bool (*same_set)(const struct ip_set *a, const struct ip_set *b);
191 /* Region-locking is used */
192 bool region_lock;
193 };
194
195 struct 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 /* The core set type structure */
202 struct ip_set_type {
203 struct list_head list;
204
205 /* Typename */
206 char name[IPSET_MAXNAMELEN];
207 /* Protocol version */
208 u8 protocol;
209 /* Set type dimension */
210 u8 dimension;
211 /*
212 * Supported family: may be NFPROTO_UNSPEC for both
213 * NFPROTO_IPV4/NFPROTO_IPV6.
214 */
215 u8 family;
216 /* Type revisions */
217 u8 revision_min, revision_max;
218 /* Set features to control swapping */
219 u16 features;
220
221 /* Create set */
222 int (*create)(struct net *net, struct ip_set *set,
223 struct nlattr *tb[], u32 flags);
224
225 /* Attribute policies */
226 const struct nla_policy create_policy[IPSET_ATTR_CREATE_MAX + 1];
227 const struct nla_policy adt_policy[IPSET_ATTR_ADT_MAX + 1];
228
229 /* Set this to THIS_MODULE if you are a module, otherwise NULL */
230 struct module *me;
231 };
232
233 /* register and unregister set type */
234 extern int ip_set_type_register(struct ip_set_type *set_type);
235 extern void ip_set_type_unregister(struct ip_set_type *set_type);
236
237 /* A generic IP set */
238 struct ip_set {
239 /* The name of the set */
240 char name[IPSET_MAXNAMELEN];
241 /* Lock protecting the set data */
242 spinlock_t lock;
243 /* References to the set */
244 u32 ref;
245 /* References to the set for netlink events like dump,
246 * ref can be swapped out by ip_set_swap
247 */
248 u32 ref_netlink;
249 /* The core set type */
250 struct ip_set_type *type;
251 /* The type variant doing the real job */
252 const struct ip_set_type_variant *variant;
253 /* The actual INET family of the set */
254 u8 family;
255 /* The type revision */
256 u8 revision;
257 /* Extensions */
258 u8 extensions;
259 /* Create flags */
260 u8 flags;
261 /* Default timeout value, if enabled */
262 u32 timeout;
263 /* Number of elements (vs timeout) */
264 u32 elements;
265 /* Size of the dynamic extensions (vs timeout) */
266 size_t ext_size;
267 /* Element data size */
268 size_t dsize;
269 /* Offsets to extensions in elements */
270 size_t offset[IPSET_EXT_ID_MAX];
271 /* The type specific data */
272 void *data;
273 };
274
275 static inline void
ip_set_ext_destroy(struct ip_set * set,void * data)276 ip_set_ext_destroy(struct ip_set *set, void *data)
277 {
278 /* Check that the extension is enabled for the set and
279 * call it's destroy function for its extension part in data.
280 */
281 if (SET_WITH_COMMENT(set)) {
282 struct ip_set_comment *c = ext_comment(data, set);
283
284 ip_set_extensions[IPSET_EXT_ID_COMMENT].destroy(set, c);
285 }
286 }
287
288 int ip_set_put_flags(struct sk_buff *skb, struct ip_set *set);
289
290 /* Netlink CB args */
291 enum {
292 IPSET_CB_NET = 0, /* net namespace */
293 IPSET_CB_PROTO, /* ipset protocol */
294 IPSET_CB_DUMP, /* dump single set/all sets */
295 IPSET_CB_INDEX, /* set index */
296 IPSET_CB_PRIVATE, /* set private data */
297 IPSET_CB_ARG0, /* type specific */
298 };
299
300 /* register and unregister set references */
301 extern ip_set_id_t ip_set_get_byname(struct net *net,
302 const char *name, struct ip_set **set);
303 extern void ip_set_put_byindex(struct net *net, ip_set_id_t index);
304 extern void ip_set_name_byindex(struct net *net, ip_set_id_t index, char *name);
305 extern ip_set_id_t ip_set_nfnl_get_byindex(struct net *net, ip_set_id_t index);
306 extern void ip_set_nfnl_put(struct net *net, ip_set_id_t index);
307
308 /* API for iptables set match, and SET target */
309
310 extern int ip_set_add(ip_set_id_t id, const struct sk_buff *skb,
311 const struct xt_action_param *par,
312 struct ip_set_adt_opt *opt);
313 extern int ip_set_del(ip_set_id_t id, const struct sk_buff *skb,
314 const struct xt_action_param *par,
315 struct ip_set_adt_opt *opt);
316 extern int ip_set_test(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
320 /* Utility functions */
321 extern void *ip_set_alloc(size_t size);
322 extern void ip_set_free(void *members);
323 extern int ip_set_get_ipaddr4(struct nlattr *nla, __be32 *ipaddr);
324 extern int ip_set_get_ipaddr6(struct nlattr *nla, union nf_inet_addr *ipaddr);
325 extern size_t ip_set_elem_len(struct ip_set *set, struct nlattr *tb[],
326 size_t len, size_t align);
327 extern int ip_set_get_extensions(struct ip_set *set, struct nlattr *tb[],
328 struct ip_set_ext *ext);
329 extern int ip_set_put_extensions(struct sk_buff *skb, const struct ip_set *set,
330 const void *e, bool active);
331 extern bool ip_set_match_extensions(struct ip_set *set,
332 const struct ip_set_ext *ext,
333 struct ip_set_ext *mext,
334 u32 flags, void *data);
335
336 static inline int
ip_set_get_hostipaddr4(struct nlattr * nla,u32 * ipaddr)337 ip_set_get_hostipaddr4(struct nlattr *nla, u32 *ipaddr)
338 {
339 __be32 ip;
340 int ret = ip_set_get_ipaddr4(nla, &ip);
341
342 if (ret)
343 return ret;
344 *ipaddr = ntohl(ip);
345 return 0;
346 }
347
348 /* Ignore IPSET_ERR_EXIST errors if asked to do so? */
349 static inline bool
ip_set_eexist(int ret,u32 flags)350 ip_set_eexist(int ret, u32 flags)
351 {
352 return ret == -IPSET_ERR_EXIST && (flags & IPSET_FLAG_EXIST);
353 }
354
355 /* Match elements marked with nomatch */
356 static inline bool
ip_set_enomatch(int ret,u32 flags,enum ipset_adt adt,struct ip_set * set)357 ip_set_enomatch(int ret, u32 flags, enum ipset_adt adt, struct ip_set *set)
358 {
359 return adt == IPSET_TEST &&
360 (set->type->features & IPSET_TYPE_NOMATCH) &&
361 ((flags >> 16) & IPSET_FLAG_NOMATCH) &&
362 (ret > 0 || ret == -ENOTEMPTY);
363 }
364
365 /* Check the NLA_F_NET_BYTEORDER flag */
366 static inline bool
ip_set_attr_netorder(struct nlattr * tb[],int type)367 ip_set_attr_netorder(struct nlattr *tb[], int type)
368 {
369 return tb[type] && (tb[type]->nla_type & NLA_F_NET_BYTEORDER);
370 }
371
372 static inline bool
ip_set_optattr_netorder(struct nlattr * tb[],int type)373 ip_set_optattr_netorder(struct nlattr *tb[], int type)
374 {
375 return !tb[type] || (tb[type]->nla_type & NLA_F_NET_BYTEORDER);
376 }
377
378 /* Useful converters */
379 static inline u32
ip_set_get_h32(const struct nlattr * attr)380 ip_set_get_h32(const struct nlattr *attr)
381 {
382 return ntohl(nla_get_be32(attr));
383 }
384
385 static inline u16
ip_set_get_h16(const struct nlattr * attr)386 ip_set_get_h16(const struct nlattr *attr)
387 {
388 return ntohs(nla_get_be16(attr));
389 }
390
nla_put_ipaddr4(struct sk_buff * skb,int type,__be32 ipaddr)391 static inline int nla_put_ipaddr4(struct sk_buff *skb, int type, __be32 ipaddr)
392 {
393 struct nlattr *__nested = nla_nest_start(skb, type);
394 int ret;
395
396 if (!__nested)
397 return -EMSGSIZE;
398 ret = nla_put_in_addr(skb, IPSET_ATTR_IPADDR_IPV4, ipaddr);
399 if (!ret)
400 nla_nest_end(skb, __nested);
401 return ret;
402 }
403
nla_put_ipaddr6(struct sk_buff * skb,int type,const struct in6_addr * ipaddrptr)404 static inline int nla_put_ipaddr6(struct sk_buff *skb, int type,
405 const struct in6_addr *ipaddrptr)
406 {
407 struct nlattr *__nested = nla_nest_start(skb, type);
408 int ret;
409
410 if (!__nested)
411 return -EMSGSIZE;
412 ret = nla_put_in6_addr(skb, IPSET_ATTR_IPADDR_IPV6, ipaddrptr);
413 if (!ret)
414 nla_nest_end(skb, __nested);
415 return ret;
416 }
417
418 /* Get address from skbuff */
419 static inline __be32
ip4addr(const struct sk_buff * skb,bool src)420 ip4addr(const struct sk_buff *skb, bool src)
421 {
422 return src ? ip_hdr(skb)->saddr : ip_hdr(skb)->daddr;
423 }
424
425 static inline void
ip4addrptr(const struct sk_buff * skb,bool src,__be32 * addr)426 ip4addrptr(const struct sk_buff *skb, bool src, __be32 *addr)
427 {
428 *addr = src ? ip_hdr(skb)->saddr : ip_hdr(skb)->daddr;
429 }
430
431 static inline void
ip6addrptr(const struct sk_buff * skb,bool src,struct in6_addr * addr)432 ip6addrptr(const struct sk_buff *skb, bool src, struct in6_addr *addr)
433 {
434 memcpy(addr, src ? &ipv6_hdr(skb)->saddr : &ipv6_hdr(skb)->daddr,
435 sizeof(*addr));
436 }
437
438 /* How often should the gc be run by default */
439 #define IPSET_GC_TIME (3 * 60)
440
441 /* Timeout period depending on the timeout value of the given set */
442 #define IPSET_GC_PERIOD(timeout) \
443 ((timeout/3) ? min_t(u32, (timeout)/3, IPSET_GC_TIME) : 1)
444
445 /* Entry is set with no timeout value */
446 #define IPSET_ELEM_PERMANENT 0
447
448 /* Set is defined with timeout support: timeout value may be 0 */
449 #define IPSET_NO_TIMEOUT UINT_MAX
450
451 /* Max timeout value, see msecs_to_jiffies() in jiffies.h */
452 #define IPSET_MAX_TIMEOUT (UINT_MAX >> 1)/MSEC_PER_SEC
453
454 #define ip_set_adt_opt_timeout(opt, set) \
455 ((opt)->ext.timeout != IPSET_NO_TIMEOUT ? (opt)->ext.timeout : (set)->timeout)
456
457 static inline unsigned int
ip_set_timeout_uget(struct nlattr * tb)458 ip_set_timeout_uget(struct nlattr *tb)
459 {
460 unsigned int timeout = ip_set_get_h32(tb);
461
462 /* Normalize to fit into jiffies */
463 if (timeout > IPSET_MAX_TIMEOUT)
464 timeout = IPSET_MAX_TIMEOUT;
465
466 return timeout;
467 }
468
469 static inline bool
ip_set_timeout_expired(const unsigned long * t)470 ip_set_timeout_expired(const unsigned long *t)
471 {
472 return *t != IPSET_ELEM_PERMANENT && time_is_before_jiffies(*t);
473 }
474
475 static inline void
ip_set_timeout_set(unsigned long * timeout,u32 value)476 ip_set_timeout_set(unsigned long *timeout, u32 value)
477 {
478 unsigned long t;
479
480 if (!value) {
481 *timeout = IPSET_ELEM_PERMANENT;
482 return;
483 }
484
485 t = msecs_to_jiffies(value * MSEC_PER_SEC) + jiffies;
486 if (t == IPSET_ELEM_PERMANENT)
487 /* Bingo! :-) */
488 t--;
489 *timeout = t;
490 }
491
492 void ip_set_init_comment(struct ip_set *set, struct ip_set_comment *comment,
493 const struct ip_set_ext *ext);
494
495 static inline void
ip_set_init_counter(struct ip_set_counter * counter,const struct ip_set_ext * ext)496 ip_set_init_counter(struct ip_set_counter *counter,
497 const struct ip_set_ext *ext)
498 {
499 if (ext->bytes != ULLONG_MAX)
500 atomic64_set(&(counter)->bytes, (long long)(ext->bytes));
501 if (ext->packets != ULLONG_MAX)
502 atomic64_set(&(counter)->packets, (long long)(ext->packets));
503 }
504
505 static inline void
ip_set_init_skbinfo(struct ip_set_skbinfo * skbinfo,const struct ip_set_ext * ext)506 ip_set_init_skbinfo(struct ip_set_skbinfo *skbinfo,
507 const struct ip_set_ext *ext)
508 {
509 *skbinfo = ext->skbinfo;
510 }
511
512 #define IP_SET_INIT_KEXT(skb, opt, set) \
513 { .bytes = (skb)->len, .packets = 1, .target = true,\
514 .timeout = ip_set_adt_opt_timeout(opt, set) }
515
516 #define IP_SET_INIT_UEXT(set) \
517 { .bytes = ULLONG_MAX, .packets = ULLONG_MAX, \
518 .timeout = (set)->timeout }
519
520 #define IPSET_CONCAT(a, b) a##b
521 #define IPSET_TOKEN(a, b) IPSET_CONCAT(a, b)
522
523 #endif /*_IP_SET_H */
524