1 /* Copyright (C) 2000-2002 Joakim Axelsson <gozem@linux.nu>
2 * Patrick Schaaf <bof@bof.de>
3 * Copyright (C) 2003-2011 Jozsef Kadlecsik <kadlec@blackhole.kfki.hu>
5 * This program is free software; you can redistribute it and/or modify
6 * it under the terms of the GNU General Public License version 2 as
7 * published by the Free Software Foundation.
10 /* Kernel module for IP set management */
12 #include <linux/init.h>
13 #include <linux/module.h>
14 #include <linux/moduleparam.h>
16 #include <linux/skbuff.h>
17 #include <linux/spinlock.h>
18 #include <linux/netlink.h>
19 #include <linux/rculist.h>
20 #include <net/netlink.h>
22 #include <linux/netfilter.h>
23 #include <linux/netfilter/x_tables.h>
24 #include <linux/netfilter/nfnetlink.h>
25 #include <linux/netfilter/ipset/ip_set.h>
27 static LIST_HEAD(ip_set_type_list); /* all registered set types */
28 static DEFINE_MUTEX(ip_set_type_mutex); /* protects ip_set_type_list */
29 static DEFINE_RWLOCK(ip_set_ref_lock); /* protects the set refs */
31 static struct ip_set **ip_set_list; /* all individual sets */
32 static ip_set_id_t ip_set_max = CONFIG_IP_SET_MAX; /* max number of sets */
34 #define STREQ(a, b) (strncmp(a, b, IPSET_MAXNAMELEN) == 0)
36 static unsigned int max_sets;
38 module_param(max_sets, int, 0600);
39 MODULE_PARM_DESC(max_sets, "maximal number of sets");
40 MODULE_LICENSE("GPL");
41 MODULE_AUTHOR("Jozsef Kadlecsik <kadlec@blackhole.kfki.hu>");
42 MODULE_DESCRIPTION("core IP set support");
43 MODULE_ALIAS_NFNL_SUBSYS(NFNL_SUBSYS_IPSET);
46 * The set types are implemented in modules and registered set types
47 * can be found in ip_set_type_list. Adding/deleting types is
48 * serialized by ip_set_type_mutex.
52 ip_set_type_lock(void)
54 mutex_lock(&ip_set_type_mutex);
58 ip_set_type_unlock(void)
60 mutex_unlock(&ip_set_type_mutex);
63 /* Register and deregister settype */
65 static struct ip_set_type *
66 find_set_type(const char *name, u8 family, u8 revision)
68 struct ip_set_type *type;
70 list_for_each_entry_rcu(type, &ip_set_type_list, list)
71 if (STREQ(type->name, name) &&
72 (type->family == family || type->family == NFPROTO_UNSPEC) &&
73 revision >= type->revision_min &&
74 revision <= type->revision_max)
79 /* Unlock, try to load a set type module and lock again */
81 load_settype(const char *name)
84 pr_debug("try to load ip_set_%s\n", name);
85 if (request_module("ip_set_%s", name) < 0) {
86 pr_warning("Can't find ip_set type %s\n", name);
94 /* Find a set type and reference it */
95 #define find_set_type_get(name, family, revision, found) \
96 __find_set_type_get(name, family, revision, found, false)
99 __find_set_type_get(const char *name, u8 family, u8 revision,
100 struct ip_set_type **found, bool retry)
102 struct ip_set_type *type;
105 if (retry && !load_settype(name))
106 return -IPSET_ERR_FIND_TYPE;
109 *found = find_set_type(name, family, revision);
111 err = !try_module_get((*found)->me) ? -EFAULT : 0;
114 /* Make sure the type is already loaded
115 * but we don't support the revision */
116 list_for_each_entry_rcu(type, &ip_set_type_list, list)
117 if (STREQ(type->name, name)) {
118 err = -IPSET_ERR_FIND_TYPE;
123 return retry ? -IPSET_ERR_FIND_TYPE :
124 __find_set_type_get(name, family, revision, found, true);
131 /* Find a given set type by name and family.
132 * If we succeeded, the supported minimal and maximum revisions are
135 #define find_set_type_minmax(name, family, min, max) \
136 __find_set_type_minmax(name, family, min, max, false)
139 __find_set_type_minmax(const char *name, u8 family, u8 *min, u8 *max,
142 struct ip_set_type *type;
145 if (retry && !load_settype(name))
146 return -IPSET_ERR_FIND_TYPE;
148 *min = 255; *max = 0;
150 list_for_each_entry_rcu(type, &ip_set_type_list, list)
151 if (STREQ(type->name, name) &&
152 (type->family == family || type->family == NFPROTO_UNSPEC)) {
154 if (type->revision_min < *min)
155 *min = type->revision_min;
156 if (type->revision_max > *max)
157 *max = type->revision_max;
163 return retry ? -IPSET_ERR_FIND_TYPE :
164 __find_set_type_minmax(name, family, min, max, true);
167 #define family_name(f) ((f) == NFPROTO_IPV4 ? "inet" : \
168 (f) == NFPROTO_IPV6 ? "inet6" : "any")
170 /* Register a set type structure. The type is identified by
171 * the unique triple of name, family and revision.
174 ip_set_type_register(struct ip_set_type *type)
178 if (type->protocol != IPSET_PROTOCOL) {
179 pr_warning("ip_set type %s, family %s, revision %u:%u uses "
180 "wrong protocol version %u (want %u)\n",
181 type->name, family_name(type->family),
182 type->revision_min, type->revision_max,
183 type->protocol, IPSET_PROTOCOL);
188 if (find_set_type(type->name, type->family, type->revision_min)) {
190 pr_warning("ip_set type %s, family %s with revision min %u "
191 "already registered!\n", type->name,
192 family_name(type->family), type->revision_min);
196 list_add_rcu(&type->list, &ip_set_type_list);
197 pr_debug("type %s, family %s, revision %u:%u registered.\n",
198 type->name, family_name(type->family),
199 type->revision_min, type->revision_max);
201 ip_set_type_unlock();
204 EXPORT_SYMBOL_GPL(ip_set_type_register);
206 /* Unregister a set type. There's a small race with ip_set_create */
208 ip_set_type_unregister(struct ip_set_type *type)
211 if (!find_set_type(type->name, type->family, type->revision_min)) {
212 pr_warning("ip_set type %s, family %s with revision min %u "
213 "not registered\n", type->name,
214 family_name(type->family), type->revision_min);
217 list_del_rcu(&type->list);
218 pr_debug("type %s, family %s with revision min %u unregistered.\n",
219 type->name, family_name(type->family), type->revision_min);
221 ip_set_type_unlock();
225 EXPORT_SYMBOL_GPL(ip_set_type_unregister);
227 /* Utility functions */
229 ip_set_alloc(size_t size)
231 void *members = NULL;
233 if (size < KMALLOC_MAX_SIZE)
234 members = kzalloc(size, GFP_KERNEL | __GFP_NOWARN);
237 pr_debug("%p: allocated with kmalloc\n", members);
241 members = vzalloc(size);
244 pr_debug("%p: allocated with vmalloc\n", members);
248 EXPORT_SYMBOL_GPL(ip_set_alloc);
251 ip_set_free(void *members)
253 pr_debug("%p: free with %s\n", members,
254 is_vmalloc_addr(members) ? "vfree" : "kfree");
255 if (is_vmalloc_addr(members))
260 EXPORT_SYMBOL_GPL(ip_set_free);
263 flag_nested(const struct nlattr *nla)
265 return nla->nla_type & NLA_F_NESTED;
268 static const struct nla_policy ipaddr_policy[IPSET_ATTR_IPADDR_MAX + 1] = {
269 [IPSET_ATTR_IPADDR_IPV4] = { .type = NLA_U32 },
270 [IPSET_ATTR_IPADDR_IPV6] = { .type = NLA_BINARY,
271 .len = sizeof(struct in6_addr) },
275 ip_set_get_ipaddr4(struct nlattr *nla, __be32 *ipaddr)
277 struct nlattr *tb[IPSET_ATTR_IPADDR_MAX+1];
279 if (unlikely(!flag_nested(nla)))
280 return -IPSET_ERR_PROTOCOL;
281 if (nla_parse_nested(tb, IPSET_ATTR_IPADDR_MAX, nla, ipaddr_policy))
282 return -IPSET_ERR_PROTOCOL;
283 if (unlikely(!ip_set_attr_netorder(tb, IPSET_ATTR_IPADDR_IPV4)))
284 return -IPSET_ERR_PROTOCOL;
286 *ipaddr = nla_get_be32(tb[IPSET_ATTR_IPADDR_IPV4]);
289 EXPORT_SYMBOL_GPL(ip_set_get_ipaddr4);
292 ip_set_get_ipaddr6(struct nlattr *nla, union nf_inet_addr *ipaddr)
294 struct nlattr *tb[IPSET_ATTR_IPADDR_MAX+1];
296 if (unlikely(!flag_nested(nla)))
297 return -IPSET_ERR_PROTOCOL;
299 if (nla_parse_nested(tb, IPSET_ATTR_IPADDR_MAX, nla, ipaddr_policy))
300 return -IPSET_ERR_PROTOCOL;
301 if (unlikely(!ip_set_attr_netorder(tb, IPSET_ATTR_IPADDR_IPV6)))
302 return -IPSET_ERR_PROTOCOL;
304 memcpy(ipaddr, nla_data(tb[IPSET_ATTR_IPADDR_IPV6]),
305 sizeof(struct in6_addr));
308 EXPORT_SYMBOL_GPL(ip_set_get_ipaddr6);
311 * Creating/destroying/renaming/swapping affect the existence and
312 * the properties of a set. All of these can be executed from userspace
313 * only and serialized by the nfnl mutex indirectly from nfnetlink.
315 * Sets are identified by their index in ip_set_list and the index
316 * is used by the external references (set/SET netfilter modules).
318 * The set behind an index may change by swapping only, from userspace.
322 __ip_set_get(ip_set_id_t index)
324 write_lock_bh(&ip_set_ref_lock);
325 ip_set_list[index]->ref++;
326 write_unlock_bh(&ip_set_ref_lock);
330 __ip_set_put(ip_set_id_t index)
332 write_lock_bh(&ip_set_ref_lock);
333 BUG_ON(ip_set_list[index]->ref == 0);
334 ip_set_list[index]->ref--;
335 write_unlock_bh(&ip_set_ref_lock);
339 * Add, del and test set entries from kernel.
341 * The set behind the index must exist and must be referenced
342 * so it can't be destroyed (or changed) under our foot.
346 ip_set_test(ip_set_id_t index, const struct sk_buff *skb,
347 const struct xt_action_param *par,
348 const struct ip_set_adt_opt *opt)
350 struct ip_set *set = ip_set_list[index];
354 pr_debug("set %s, index %u\n", set->name, index);
356 if (opt->dim < set->type->dimension ||
357 !(opt->family == set->family || set->family == NFPROTO_UNSPEC))
360 read_lock_bh(&set->lock);
361 ret = set->variant->kadt(set, skb, par, IPSET_TEST, opt);
362 read_unlock_bh(&set->lock);
364 if (ret == -EAGAIN) {
365 /* Type requests element to be completed */
366 pr_debug("element must be competed, ADD is triggered\n");
367 write_lock_bh(&set->lock);
368 set->variant->kadt(set, skb, par, IPSET_ADD, opt);
369 write_unlock_bh(&set->lock);
373 /* Convert error codes to nomatch */
374 return (ret < 0 ? 0 : ret);
376 EXPORT_SYMBOL_GPL(ip_set_test);
379 ip_set_add(ip_set_id_t index, const struct sk_buff *skb,
380 const struct xt_action_param *par,
381 const struct ip_set_adt_opt *opt)
383 struct ip_set *set = ip_set_list[index];
387 pr_debug("set %s, index %u\n", set->name, index);
389 if (opt->dim < set->type->dimension ||
390 !(opt->family == set->family || set->family == NFPROTO_UNSPEC))
393 write_lock_bh(&set->lock);
394 ret = set->variant->kadt(set, skb, par, IPSET_ADD, opt);
395 write_unlock_bh(&set->lock);
399 EXPORT_SYMBOL_GPL(ip_set_add);
402 ip_set_del(ip_set_id_t index, const struct sk_buff *skb,
403 const struct xt_action_param *par,
404 const struct ip_set_adt_opt *opt)
406 struct ip_set *set = ip_set_list[index];
410 pr_debug("set %s, index %u\n", set->name, index);
412 if (opt->dim < set->type->dimension ||
413 !(opt->family == set->family || set->family == NFPROTO_UNSPEC))
416 write_lock_bh(&set->lock);
417 ret = set->variant->kadt(set, skb, par, IPSET_DEL, opt);
418 write_unlock_bh(&set->lock);
422 EXPORT_SYMBOL_GPL(ip_set_del);
425 * Find set by name, reference it once. The reference makes sure the
426 * thing pointed to, does not go away under our feet.
430 ip_set_get_byname(const char *name, struct ip_set **set)
432 ip_set_id_t i, index = IPSET_INVALID_ID;
435 for (i = 0; i < ip_set_max; i++) {
437 if (s != NULL && STREQ(s->name, name)) {
446 EXPORT_SYMBOL_GPL(ip_set_get_byname);
449 * If the given set pointer points to a valid set, decrement
450 * reference count by 1. The caller shall not assume the index
451 * to be valid, after calling this function.
455 ip_set_put_byindex(ip_set_id_t index)
457 if (ip_set_list[index] != NULL)
460 EXPORT_SYMBOL_GPL(ip_set_put_byindex);
463 * Get the name of a set behind a set index.
464 * We assume the set is referenced, so it does exist and
465 * can't be destroyed. The set cannot be renamed due to
466 * the referencing either.
470 ip_set_name_byindex(ip_set_id_t index)
472 const struct ip_set *set = ip_set_list[index];
475 BUG_ON(set->ref == 0);
477 /* Referenced, so it's safe */
480 EXPORT_SYMBOL_GPL(ip_set_name_byindex);
483 * Routines to call by external subsystems, which do not
484 * call nfnl_lock for us.
488 * Find set by name, reference it once. The reference makes sure the
489 * thing pointed to, does not go away under our feet.
491 * The nfnl mutex is used in the function.
494 ip_set_nfnl_get(const char *name)
500 index = ip_set_get_byname(name, &s);
505 EXPORT_SYMBOL_GPL(ip_set_nfnl_get);
508 * Find set by index, reference it once. The reference makes sure the
509 * thing pointed to, does not go away under our feet.
511 * The nfnl mutex is used in the function.
514 ip_set_nfnl_get_byindex(ip_set_id_t index)
516 if (index > ip_set_max)
517 return IPSET_INVALID_ID;
520 if (ip_set_list[index])
523 index = IPSET_INVALID_ID;
528 EXPORT_SYMBOL_GPL(ip_set_nfnl_get_byindex);
531 * If the given set pointer points to a valid set, decrement
532 * reference count by 1. The caller shall not assume the index
533 * to be valid, after calling this function.
535 * The nfnl mutex is used in the function.
538 ip_set_nfnl_put(ip_set_id_t index)
541 ip_set_put_byindex(index);
544 EXPORT_SYMBOL_GPL(ip_set_nfnl_put);
547 * Communication protocol with userspace over netlink.
549 * The commands are serialized by the nfnl mutex.
553 protocol_failed(const struct nlattr * const tb[])
555 return !tb[IPSET_ATTR_PROTOCOL] ||
556 nla_get_u8(tb[IPSET_ATTR_PROTOCOL]) != IPSET_PROTOCOL;
560 flag_exist(const struct nlmsghdr *nlh)
562 return nlh->nlmsg_flags & NLM_F_EXCL ? 0 : IPSET_FLAG_EXIST;
565 static struct nlmsghdr *
566 start_msg(struct sk_buff *skb, u32 pid, u32 seq, unsigned int flags,
569 struct nlmsghdr *nlh;
570 struct nfgenmsg *nfmsg;
572 nlh = nlmsg_put(skb, pid, seq, cmd | (NFNL_SUBSYS_IPSET << 8),
573 sizeof(*nfmsg), flags);
577 nfmsg = nlmsg_data(nlh);
578 nfmsg->nfgen_family = NFPROTO_IPV4;
579 nfmsg->version = NFNETLINK_V0;
587 static const struct nla_policy ip_set_create_policy[IPSET_ATTR_CMD_MAX + 1] = {
588 [IPSET_ATTR_PROTOCOL] = { .type = NLA_U8 },
589 [IPSET_ATTR_SETNAME] = { .type = NLA_NUL_STRING,
590 .len = IPSET_MAXNAMELEN - 1 },
591 [IPSET_ATTR_TYPENAME] = { .type = NLA_NUL_STRING,
592 .len = IPSET_MAXNAMELEN - 1},
593 [IPSET_ATTR_REVISION] = { .type = NLA_U8 },
594 [IPSET_ATTR_FAMILY] = { .type = NLA_U8 },
595 [IPSET_ATTR_DATA] = { .type = NLA_NESTED },
599 find_set_id(const char *name)
601 ip_set_id_t i, index = IPSET_INVALID_ID;
602 const struct ip_set *set;
604 for (i = 0; index == IPSET_INVALID_ID && i < ip_set_max; i++) {
605 set = ip_set_list[i];
606 if (set != NULL && STREQ(set->name, name))
612 static inline struct ip_set *
613 find_set(const char *name)
615 ip_set_id_t index = find_set_id(name);
617 return index == IPSET_INVALID_ID ? NULL : ip_set_list[index];
621 find_free_id(const char *name, ip_set_id_t *index, struct ip_set **set)
625 *index = IPSET_INVALID_ID;
626 for (i = 0; i < ip_set_max; i++) {
627 if (ip_set_list[i] == NULL) {
628 if (*index == IPSET_INVALID_ID)
630 } else if (STREQ(name, ip_set_list[i]->name)) {
632 *set = ip_set_list[i];
636 if (*index == IPSET_INVALID_ID)
637 /* No free slot remained */
638 return -IPSET_ERR_MAX_SETS;
643 ip_set_create(struct sock *ctnl, struct sk_buff *skb,
644 const struct nlmsghdr *nlh,
645 const struct nlattr * const attr[])
647 struct ip_set *set, *clash = NULL;
648 ip_set_id_t index = IPSET_INVALID_ID;
649 struct nlattr *tb[IPSET_ATTR_CREATE_MAX+1] = {};
650 const char *name, *typename;
652 u32 flags = flag_exist(nlh);
655 if (unlikely(protocol_failed(attr) ||
656 attr[IPSET_ATTR_SETNAME] == NULL ||
657 attr[IPSET_ATTR_TYPENAME] == NULL ||
658 attr[IPSET_ATTR_REVISION] == NULL ||
659 attr[IPSET_ATTR_FAMILY] == NULL ||
660 (attr[IPSET_ATTR_DATA] != NULL &&
661 !flag_nested(attr[IPSET_ATTR_DATA]))))
662 return -IPSET_ERR_PROTOCOL;
664 name = nla_data(attr[IPSET_ATTR_SETNAME]);
665 typename = nla_data(attr[IPSET_ATTR_TYPENAME]);
666 family = nla_get_u8(attr[IPSET_ATTR_FAMILY]);
667 revision = nla_get_u8(attr[IPSET_ATTR_REVISION]);
668 pr_debug("setname: %s, typename: %s, family: %s, revision: %u\n",
669 name, typename, family_name(family), revision);
672 * First, and without any locks, allocate and initialize
673 * a normal base set structure.
675 set = kzalloc(sizeof(struct ip_set), GFP_KERNEL);
678 rwlock_init(&set->lock);
679 strlcpy(set->name, name, IPSET_MAXNAMELEN);
680 set->family = family;
681 set->revision = revision;
684 * Next, check that we know the type, and take
685 * a reference on the type, to make sure it stays available
686 * while constructing our new set.
688 * After referencing the type, we try to create the type
689 * specific part of the set without holding any locks.
691 ret = find_set_type_get(typename, family, revision, &(set->type));
696 * Without holding any locks, create private part.
698 if (attr[IPSET_ATTR_DATA] &&
699 nla_parse_nested(tb, IPSET_ATTR_CREATE_MAX, attr[IPSET_ATTR_DATA],
700 set->type->create_policy)) {
701 ret = -IPSET_ERR_PROTOCOL;
705 ret = set->type->create(set, tb, flags);
709 /* BTW, ret==0 here. */
712 * Here, we have a valid, constructed set and we are protected
713 * by the nfnl mutex. Find the first free index in ip_set_list
714 * and check clashing.
716 if ((ret = find_free_id(set->name, &index, &clash)) != 0) {
717 /* If this is the same set and requested, ignore error */
718 if (ret == -EEXIST &&
719 (flags & IPSET_FLAG_EXIST) &&
720 STREQ(set->type->name, clash->type->name) &&
721 set->type->family == clash->type->family &&
722 set->type->revision_min == clash->type->revision_min &&
723 set->type->revision_max == clash->type->revision_max &&
724 set->variant->same_set(set, clash))
730 * Finally! Add our shiny new set to the list, and be done.
732 pr_debug("create: '%s' created with index %u!\n", set->name, index);
733 ip_set_list[index] = set;
738 set->variant->destroy(set);
740 module_put(set->type->me);
748 static const struct nla_policy
749 ip_set_setname_policy[IPSET_ATTR_CMD_MAX + 1] = {
750 [IPSET_ATTR_PROTOCOL] = { .type = NLA_U8 },
751 [IPSET_ATTR_SETNAME] = { .type = NLA_NUL_STRING,
752 .len = IPSET_MAXNAMELEN - 1 },
756 ip_set_destroy_set(ip_set_id_t index)
758 struct ip_set *set = ip_set_list[index];
760 pr_debug("set: %s\n", set->name);
761 ip_set_list[index] = NULL;
763 /* Must call it without holding any lock */
764 set->variant->destroy(set);
765 module_put(set->type->me);
770 ip_set_destroy(struct sock *ctnl, struct sk_buff *skb,
771 const struct nlmsghdr *nlh,
772 const struct nlattr * const attr[])
777 if (unlikely(protocol_failed(attr)))
778 return -IPSET_ERR_PROTOCOL;
780 /* Commands are serialized and references are
781 * protected by the ip_set_ref_lock.
782 * External systems (i.e. xt_set) must call
783 * ip_set_put|get_nfnl_* functions, that way we
784 * can safely check references here.
786 * list:set timer can only decrement the reference
787 * counter, so if it's already zero, we can proceed
788 * without holding the lock.
790 read_lock_bh(&ip_set_ref_lock);
791 if (!attr[IPSET_ATTR_SETNAME]) {
792 for (i = 0; i < ip_set_max; i++) {
793 if (ip_set_list[i] != NULL && ip_set_list[i]->ref) {
794 ret = -IPSET_ERR_BUSY;
798 read_unlock_bh(&ip_set_ref_lock);
799 for (i = 0; i < ip_set_max; i++) {
800 if (ip_set_list[i] != NULL)
801 ip_set_destroy_set(i);
804 i = find_set_id(nla_data(attr[IPSET_ATTR_SETNAME]));
805 if (i == IPSET_INVALID_ID) {
808 } else if (ip_set_list[i]->ref) {
809 ret = -IPSET_ERR_BUSY;
812 read_unlock_bh(&ip_set_ref_lock);
814 ip_set_destroy_set(i);
818 read_unlock_bh(&ip_set_ref_lock);
825 ip_set_flush_set(struct ip_set *set)
827 pr_debug("set: %s\n", set->name);
829 write_lock_bh(&set->lock);
830 set->variant->flush(set);
831 write_unlock_bh(&set->lock);
835 ip_set_flush(struct sock *ctnl, struct sk_buff *skb,
836 const struct nlmsghdr *nlh,
837 const struct nlattr * const attr[])
841 if (unlikely(protocol_failed(attr)))
842 return -IPSET_ERR_PROTOCOL;
844 if (!attr[IPSET_ATTR_SETNAME]) {
845 for (i = 0; i < ip_set_max; i++)
846 if (ip_set_list[i] != NULL)
847 ip_set_flush_set(ip_set_list[i]);
849 i = find_set_id(nla_data(attr[IPSET_ATTR_SETNAME]));
850 if (i == IPSET_INVALID_ID)
853 ip_set_flush_set(ip_set_list[i]);
861 static const struct nla_policy
862 ip_set_setname2_policy[IPSET_ATTR_CMD_MAX + 1] = {
863 [IPSET_ATTR_PROTOCOL] = { .type = NLA_U8 },
864 [IPSET_ATTR_SETNAME] = { .type = NLA_NUL_STRING,
865 .len = IPSET_MAXNAMELEN - 1 },
866 [IPSET_ATTR_SETNAME2] = { .type = NLA_NUL_STRING,
867 .len = IPSET_MAXNAMELEN - 1 },
871 ip_set_rename(struct sock *ctnl, struct sk_buff *skb,
872 const struct nlmsghdr *nlh,
873 const struct nlattr * const attr[])
880 if (unlikely(protocol_failed(attr) ||
881 attr[IPSET_ATTR_SETNAME] == NULL ||
882 attr[IPSET_ATTR_SETNAME2] == NULL))
883 return -IPSET_ERR_PROTOCOL;
885 set = find_set(nla_data(attr[IPSET_ATTR_SETNAME]));
889 read_lock_bh(&ip_set_ref_lock);
891 ret = -IPSET_ERR_REFERENCED;
895 name2 = nla_data(attr[IPSET_ATTR_SETNAME2]);
896 for (i = 0; i < ip_set_max; i++) {
897 if (ip_set_list[i] != NULL &&
898 STREQ(ip_set_list[i]->name, name2)) {
899 ret = -IPSET_ERR_EXIST_SETNAME2;
903 strncpy(set->name, name2, IPSET_MAXNAMELEN);
906 read_unlock_bh(&ip_set_ref_lock);
910 /* Swap two sets so that name/index points to the other.
911 * References and set names are also swapped.
913 * The commands are serialized by the nfnl mutex and references are
914 * protected by the ip_set_ref_lock. The kernel interfaces
915 * do not hold the mutex but the pointer settings are atomic
916 * so the ip_set_list always contains valid pointers to the sets.
920 ip_set_swap(struct sock *ctnl, struct sk_buff *skb,
921 const struct nlmsghdr *nlh,
922 const struct nlattr * const attr[])
924 struct ip_set *from, *to;
925 ip_set_id_t from_id, to_id;
926 char from_name[IPSET_MAXNAMELEN];
928 if (unlikely(protocol_failed(attr) ||
929 attr[IPSET_ATTR_SETNAME] == NULL ||
930 attr[IPSET_ATTR_SETNAME2] == NULL))
931 return -IPSET_ERR_PROTOCOL;
933 from_id = find_set_id(nla_data(attr[IPSET_ATTR_SETNAME]));
934 if (from_id == IPSET_INVALID_ID)
937 to_id = find_set_id(nla_data(attr[IPSET_ATTR_SETNAME2]));
938 if (to_id == IPSET_INVALID_ID)
939 return -IPSET_ERR_EXIST_SETNAME2;
941 from = ip_set_list[from_id];
942 to = ip_set_list[to_id];
944 /* Features must not change.
945 * Not an artificial restriction anymore, as we must prevent
946 * possible loops created by swapping in setlist type of sets. */
947 if (!(from->type->features == to->type->features &&
948 from->type->family == to->type->family))
949 return -IPSET_ERR_TYPE_MISMATCH;
951 strncpy(from_name, from->name, IPSET_MAXNAMELEN);
952 strncpy(from->name, to->name, IPSET_MAXNAMELEN);
953 strncpy(to->name, from_name, IPSET_MAXNAMELEN);
955 write_lock_bh(&ip_set_ref_lock);
956 swap(from->ref, to->ref);
957 ip_set_list[from_id] = to;
958 ip_set_list[to_id] = from;
959 write_unlock_bh(&ip_set_ref_lock);
964 /* List/save set data */
971 #define DUMP_TYPE(arg) (((u32)(arg)) & 0x0000FFFF)
972 #define DUMP_FLAGS(arg) (((u32)(arg)) >> 16)
975 ip_set_dump_done(struct netlink_callback *cb)
978 pr_debug("release set %s\n", ip_set_list[cb->args[1]]->name);
979 ip_set_put_byindex((ip_set_id_t) cb->args[1]);
985 dump_attrs(struct nlmsghdr *nlh)
987 const struct nlattr *attr;
990 pr_debug("dump nlmsg\n");
991 nlmsg_for_each_attr(attr, nlh, sizeof(struct nfgenmsg), rem) {
992 pr_debug("type: %u, len %u\n", nla_type(attr), attr->nla_len);
997 dump_init(struct netlink_callback *cb)
999 struct nlmsghdr *nlh = nlmsg_hdr(cb->skb);
1000 int min_len = NLMSG_SPACE(sizeof(struct nfgenmsg));
1001 struct nlattr *cda[IPSET_ATTR_CMD_MAX+1];
1002 struct nlattr *attr = (void *)nlh + min_len;
1006 /* Second pass, so parser can't fail */
1007 nla_parse(cda, IPSET_ATTR_CMD_MAX,
1008 attr, nlh->nlmsg_len - min_len, ip_set_setname_policy);
1010 /* cb->args[0] : dump single set/all sets
1012 * [..]: type specific
1015 if (cda[IPSET_ATTR_SETNAME]) {
1016 index = find_set_id(nla_data(cda[IPSET_ATTR_SETNAME]));
1017 if (index == IPSET_INVALID_ID)
1020 dump_type = DUMP_ONE;
1021 cb->args[1] = index;
1023 dump_type = DUMP_ALL;
1025 if (cda[IPSET_ATTR_FLAGS]) {
1026 u32 f = ip_set_get_h32(cda[IPSET_ATTR_FLAGS]);
1027 dump_type |= (f << 16);
1029 cb->args[0] = dump_type;
1035 ip_set_dump_start(struct sk_buff *skb, struct netlink_callback *cb)
1037 ip_set_id_t index = IPSET_INVALID_ID, max;
1038 struct ip_set *set = NULL;
1039 struct nlmsghdr *nlh = NULL;
1040 unsigned int flags = NETLINK_CB(cb->skb).pid ? NLM_F_MULTI : 0;
1041 u32 dump_type, dump_flags;
1045 ret = dump_init(cb);
1047 nlh = nlmsg_hdr(cb->skb);
1048 /* We have to create and send the error message
1050 if (nlh->nlmsg_flags & NLM_F_ACK)
1051 netlink_ack(cb->skb, nlh, ret);
1056 if (cb->args[1] >= ip_set_max)
1059 dump_type = DUMP_TYPE(cb->args[0]);
1060 dump_flags = DUMP_FLAGS(cb->args[0]);
1061 max = dump_type == DUMP_ONE ? cb->args[1] + 1 : ip_set_max;
1063 pr_debug("args[0]: %u %u args[1]: %ld\n",
1064 dump_type, dump_flags, cb->args[1]);
1065 for (; cb->args[1] < max; cb->args[1]++) {
1066 index = (ip_set_id_t) cb->args[1];
1067 set = ip_set_list[index];
1069 if (dump_type == DUMP_ONE) {
1075 /* When dumping all sets, we must dump "sorted"
1076 * so that lists (unions of sets) are dumped last.
1078 if (dump_type != DUMP_ONE &&
1079 ((dump_type == DUMP_ALL) ==
1080 !!(set->type->features & IPSET_DUMP_LAST)))
1082 pr_debug("List set: %s\n", set->name);
1084 /* Start listing: make sure set won't be destroyed */
1085 pr_debug("reference set\n");
1086 __ip_set_get(index);
1088 nlh = start_msg(skb, NETLINK_CB(cb->skb).pid,
1089 cb->nlh->nlmsg_seq, flags,
1093 goto release_refcount;
1095 if (nla_put_u8(skb, IPSET_ATTR_PROTOCOL, IPSET_PROTOCOL) ||
1096 nla_put_string(skb, IPSET_ATTR_SETNAME, set->name))
1097 goto nla_put_failure;
1098 if (dump_flags & IPSET_FLAG_LIST_SETNAME)
1100 switch (cb->args[2]) {
1102 /* Core header data */
1103 if (nla_put_string(skb, IPSET_ATTR_TYPENAME,
1105 nla_put_u8(skb, IPSET_ATTR_FAMILY,
1107 nla_put_u8(skb, IPSET_ATTR_REVISION,
1109 goto nla_put_failure;
1110 ret = set->variant->head(set, skb);
1112 goto release_refcount;
1113 if (dump_flags & IPSET_FLAG_LIST_HEADER)
1115 /* Fall through and add elements */
1117 read_lock_bh(&set->lock);
1118 ret = set->variant->list(set, skb, cb);
1119 read_unlock_bh(&set->lock);
1121 /* Set is done, proceed with next one */
1123 goto release_refcount;
1126 /* If we dump all sets, continue with dumping last ones */
1127 if (dump_type == DUMP_ALL) {
1128 dump_type = DUMP_LAST;
1129 cb->args[0] = dump_type | (dump_flags << 16);
1138 if (dump_type == DUMP_ONE)
1139 cb->args[1] = IPSET_INVALID_ID;
1143 /* If there was an error or set is done, release set */
1144 if (ret || !cb->args[2]) {
1145 pr_debug("release set %s\n", ip_set_list[index]->name);
1146 ip_set_put_byindex(index);
1151 nlmsg_end(skb, nlh);
1152 pr_debug("nlmsg_len: %u\n", nlh->nlmsg_len);
1156 return ret < 0 ? ret : skb->len;
1160 ip_set_dump(struct sock *ctnl, struct sk_buff *skb,
1161 const struct nlmsghdr *nlh,
1162 const struct nlattr * const attr[])
1164 if (unlikely(protocol_failed(attr)))
1165 return -IPSET_ERR_PROTOCOL;
1168 struct netlink_dump_control c = {
1169 .dump = ip_set_dump_start,
1170 .done = ip_set_dump_done,
1172 return netlink_dump_start(ctnl, skb, nlh, &c);
1176 /* Add, del and test */
1178 static const struct nla_policy ip_set_adt_policy[IPSET_ATTR_CMD_MAX + 1] = {
1179 [IPSET_ATTR_PROTOCOL] = { .type = NLA_U8 },
1180 [IPSET_ATTR_SETNAME] = { .type = NLA_NUL_STRING,
1181 .len = IPSET_MAXNAMELEN - 1 },
1182 [IPSET_ATTR_LINENO] = { .type = NLA_U32 },
1183 [IPSET_ATTR_DATA] = { .type = NLA_NESTED },
1184 [IPSET_ATTR_ADT] = { .type = NLA_NESTED },
1188 call_ad(struct sock *ctnl, struct sk_buff *skb, struct ip_set *set,
1189 struct nlattr *tb[], enum ipset_adt adt,
1190 u32 flags, bool use_lineno)
1194 bool eexist = flags & IPSET_FLAG_EXIST, retried = false;
1197 write_lock_bh(&set->lock);
1198 ret = set->variant->uadt(set, tb, adt, &lineno, flags, retried);
1199 write_unlock_bh(&set->lock);
1201 } while (ret == -EAGAIN &&
1202 set->variant->resize &&
1203 (ret = set->variant->resize(set, retried)) == 0);
1205 if (!ret || (ret == -IPSET_ERR_EXIST && eexist))
1207 if (lineno && use_lineno) {
1208 /* Error in restore/batch mode: send back lineno */
1209 struct nlmsghdr *rep, *nlh = nlmsg_hdr(skb);
1210 struct sk_buff *skb2;
1211 struct nlmsgerr *errmsg;
1212 size_t payload = sizeof(*errmsg) + nlmsg_len(nlh);
1213 int min_len = NLMSG_SPACE(sizeof(struct nfgenmsg));
1214 struct nlattr *cda[IPSET_ATTR_CMD_MAX+1];
1215 struct nlattr *cmdattr;
1218 skb2 = nlmsg_new(payload, GFP_KERNEL);
1221 rep = __nlmsg_put(skb2, NETLINK_CB(skb).pid,
1222 nlh->nlmsg_seq, NLMSG_ERROR, payload, 0);
1223 errmsg = nlmsg_data(rep);
1224 errmsg->error = ret;
1225 memcpy(&errmsg->msg, nlh, nlh->nlmsg_len);
1226 cmdattr = (void *)&errmsg->msg + min_len;
1228 nla_parse(cda, IPSET_ATTR_CMD_MAX,
1229 cmdattr, nlh->nlmsg_len - min_len,
1232 errline = nla_data(cda[IPSET_ATTR_LINENO]);
1236 netlink_unicast(ctnl, skb2, NETLINK_CB(skb).pid, MSG_DONTWAIT);
1237 /* Signal netlink not to send its ACK/errmsg. */
1245 ip_set_uadd(struct sock *ctnl, struct sk_buff *skb,
1246 const struct nlmsghdr *nlh,
1247 const struct nlattr * const attr[])
1250 struct nlattr *tb[IPSET_ATTR_ADT_MAX+1] = {};
1251 const struct nlattr *nla;
1252 u32 flags = flag_exist(nlh);
1256 if (unlikely(protocol_failed(attr) ||
1257 attr[IPSET_ATTR_SETNAME] == NULL ||
1258 !((attr[IPSET_ATTR_DATA] != NULL) ^
1259 (attr[IPSET_ATTR_ADT] != NULL)) ||
1260 (attr[IPSET_ATTR_DATA] != NULL &&
1261 !flag_nested(attr[IPSET_ATTR_DATA])) ||
1262 (attr[IPSET_ATTR_ADT] != NULL &&
1263 (!flag_nested(attr[IPSET_ATTR_ADT]) ||
1264 attr[IPSET_ATTR_LINENO] == NULL))))
1265 return -IPSET_ERR_PROTOCOL;
1267 set = find_set(nla_data(attr[IPSET_ATTR_SETNAME]));
1271 use_lineno = !!attr[IPSET_ATTR_LINENO];
1272 if (attr[IPSET_ATTR_DATA]) {
1273 if (nla_parse_nested(tb, IPSET_ATTR_ADT_MAX,
1274 attr[IPSET_ATTR_DATA],
1275 set->type->adt_policy))
1276 return -IPSET_ERR_PROTOCOL;
1277 ret = call_ad(ctnl, skb, set, tb, IPSET_ADD, flags,
1282 nla_for_each_nested(nla, attr[IPSET_ATTR_ADT], nla_rem) {
1283 memset(tb, 0, sizeof(tb));
1284 if (nla_type(nla) != IPSET_ATTR_DATA ||
1285 !flag_nested(nla) ||
1286 nla_parse_nested(tb, IPSET_ATTR_ADT_MAX, nla,
1287 set->type->adt_policy))
1288 return -IPSET_ERR_PROTOCOL;
1289 ret = call_ad(ctnl, skb, set, tb, IPSET_ADD,
1299 ip_set_udel(struct sock *ctnl, struct sk_buff *skb,
1300 const struct nlmsghdr *nlh,
1301 const struct nlattr * const attr[])
1304 struct nlattr *tb[IPSET_ATTR_ADT_MAX+1] = {};
1305 const struct nlattr *nla;
1306 u32 flags = flag_exist(nlh);
1310 if (unlikely(protocol_failed(attr) ||
1311 attr[IPSET_ATTR_SETNAME] == NULL ||
1312 !((attr[IPSET_ATTR_DATA] != NULL) ^
1313 (attr[IPSET_ATTR_ADT] != NULL)) ||
1314 (attr[IPSET_ATTR_DATA] != NULL &&
1315 !flag_nested(attr[IPSET_ATTR_DATA])) ||
1316 (attr[IPSET_ATTR_ADT] != NULL &&
1317 (!flag_nested(attr[IPSET_ATTR_ADT]) ||
1318 attr[IPSET_ATTR_LINENO] == NULL))))
1319 return -IPSET_ERR_PROTOCOL;
1321 set = find_set(nla_data(attr[IPSET_ATTR_SETNAME]));
1325 use_lineno = !!attr[IPSET_ATTR_LINENO];
1326 if (attr[IPSET_ATTR_DATA]) {
1327 if (nla_parse_nested(tb, IPSET_ATTR_ADT_MAX,
1328 attr[IPSET_ATTR_DATA],
1329 set->type->adt_policy))
1330 return -IPSET_ERR_PROTOCOL;
1331 ret = call_ad(ctnl, skb, set, tb, IPSET_DEL, flags,
1336 nla_for_each_nested(nla, attr[IPSET_ATTR_ADT], nla_rem) {
1337 memset(tb, 0, sizeof(*tb));
1338 if (nla_type(nla) != IPSET_ATTR_DATA ||
1339 !flag_nested(nla) ||
1340 nla_parse_nested(tb, IPSET_ATTR_ADT_MAX, nla,
1341 set->type->adt_policy))
1342 return -IPSET_ERR_PROTOCOL;
1343 ret = call_ad(ctnl, skb, set, tb, IPSET_DEL,
1353 ip_set_utest(struct sock *ctnl, struct sk_buff *skb,
1354 const struct nlmsghdr *nlh,
1355 const struct nlattr * const attr[])
1358 struct nlattr *tb[IPSET_ATTR_ADT_MAX+1] = {};
1361 if (unlikely(protocol_failed(attr) ||
1362 attr[IPSET_ATTR_SETNAME] == NULL ||
1363 attr[IPSET_ATTR_DATA] == NULL ||
1364 !flag_nested(attr[IPSET_ATTR_DATA])))
1365 return -IPSET_ERR_PROTOCOL;
1367 set = find_set(nla_data(attr[IPSET_ATTR_SETNAME]));
1371 if (nla_parse_nested(tb, IPSET_ATTR_ADT_MAX, attr[IPSET_ATTR_DATA],
1372 set->type->adt_policy))
1373 return -IPSET_ERR_PROTOCOL;
1375 read_lock_bh(&set->lock);
1376 ret = set->variant->uadt(set, tb, IPSET_TEST, NULL, 0, 0);
1377 read_unlock_bh(&set->lock);
1378 /* Userspace can't trigger element to be re-added */
1382 return ret < 0 ? ret : ret > 0 ? 0 : -IPSET_ERR_EXIST;
1385 /* Get headed data of a set */
1388 ip_set_header(struct sock *ctnl, struct sk_buff *skb,
1389 const struct nlmsghdr *nlh,
1390 const struct nlattr * const attr[])
1392 const struct ip_set *set;
1393 struct sk_buff *skb2;
1394 struct nlmsghdr *nlh2;
1398 if (unlikely(protocol_failed(attr) ||
1399 attr[IPSET_ATTR_SETNAME] == NULL))
1400 return -IPSET_ERR_PROTOCOL;
1402 index = find_set_id(nla_data(attr[IPSET_ATTR_SETNAME]));
1403 if (index == IPSET_INVALID_ID)
1405 set = ip_set_list[index];
1407 skb2 = nlmsg_new(NLMSG_DEFAULT_SIZE, GFP_KERNEL);
1411 nlh2 = start_msg(skb2, NETLINK_CB(skb).pid, nlh->nlmsg_seq, 0,
1415 if (nla_put_u8(skb2, IPSET_ATTR_PROTOCOL, IPSET_PROTOCOL) ||
1416 nla_put_string(skb2, IPSET_ATTR_SETNAME, set->name) ||
1417 nla_put_string(skb2, IPSET_ATTR_TYPENAME, set->type->name) ||
1418 nla_put_u8(skb2, IPSET_ATTR_FAMILY, set->family) ||
1419 nla_put_u8(skb2, IPSET_ATTR_REVISION, set->revision))
1420 goto nla_put_failure;
1421 nlmsg_end(skb2, nlh2);
1423 ret = netlink_unicast(ctnl, skb2, NETLINK_CB(skb).pid, MSG_DONTWAIT);
1430 nlmsg_cancel(skb2, nlh2);
1438 static const struct nla_policy ip_set_type_policy[IPSET_ATTR_CMD_MAX + 1] = {
1439 [IPSET_ATTR_PROTOCOL] = { .type = NLA_U8 },
1440 [IPSET_ATTR_TYPENAME] = { .type = NLA_NUL_STRING,
1441 .len = IPSET_MAXNAMELEN - 1 },
1442 [IPSET_ATTR_FAMILY] = { .type = NLA_U8 },
1446 ip_set_type(struct sock *ctnl, struct sk_buff *skb,
1447 const struct nlmsghdr *nlh,
1448 const struct nlattr * const attr[])
1450 struct sk_buff *skb2;
1451 struct nlmsghdr *nlh2;
1452 u8 family, min, max;
1453 const char *typename;
1456 if (unlikely(protocol_failed(attr) ||
1457 attr[IPSET_ATTR_TYPENAME] == NULL ||
1458 attr[IPSET_ATTR_FAMILY] == NULL))
1459 return -IPSET_ERR_PROTOCOL;
1461 family = nla_get_u8(attr[IPSET_ATTR_FAMILY]);
1462 typename = nla_data(attr[IPSET_ATTR_TYPENAME]);
1463 ret = find_set_type_minmax(typename, family, &min, &max);
1467 skb2 = nlmsg_new(NLMSG_DEFAULT_SIZE, GFP_KERNEL);
1471 nlh2 = start_msg(skb2, NETLINK_CB(skb).pid, nlh->nlmsg_seq, 0,
1475 if (nla_put_u8(skb2, IPSET_ATTR_PROTOCOL, IPSET_PROTOCOL) ||
1476 nla_put_string(skb2, IPSET_ATTR_TYPENAME, typename) ||
1477 nla_put_u8(skb2, IPSET_ATTR_FAMILY, family) ||
1478 nla_put_u8(skb2, IPSET_ATTR_REVISION, max) ||
1479 nla_put_u8(skb2, IPSET_ATTR_REVISION_MIN, min))
1480 goto nla_put_failure;
1481 nlmsg_end(skb2, nlh2);
1483 pr_debug("Send TYPE, nlmsg_len: %u\n", nlh2->nlmsg_len);
1484 ret = netlink_unicast(ctnl, skb2, NETLINK_CB(skb).pid, MSG_DONTWAIT);
1491 nlmsg_cancel(skb2, nlh2);
1497 /* Get protocol version */
1499 static const struct nla_policy
1500 ip_set_protocol_policy[IPSET_ATTR_CMD_MAX + 1] = {
1501 [IPSET_ATTR_PROTOCOL] = { .type = NLA_U8 },
1505 ip_set_protocol(struct sock *ctnl, struct sk_buff *skb,
1506 const struct nlmsghdr *nlh,
1507 const struct nlattr * const attr[])
1509 struct sk_buff *skb2;
1510 struct nlmsghdr *nlh2;
1513 if (unlikely(attr[IPSET_ATTR_PROTOCOL] == NULL))
1514 return -IPSET_ERR_PROTOCOL;
1516 skb2 = nlmsg_new(NLMSG_DEFAULT_SIZE, GFP_KERNEL);
1520 nlh2 = start_msg(skb2, NETLINK_CB(skb).pid, nlh->nlmsg_seq, 0,
1521 IPSET_CMD_PROTOCOL);
1524 if (nla_put_u8(skb2, IPSET_ATTR_PROTOCOL, IPSET_PROTOCOL))
1525 goto nla_put_failure;
1526 nlmsg_end(skb2, nlh2);
1528 ret = netlink_unicast(ctnl, skb2, NETLINK_CB(skb).pid, MSG_DONTWAIT);
1535 nlmsg_cancel(skb2, nlh2);
1541 static const struct nfnl_callback ip_set_netlink_subsys_cb[IPSET_MSG_MAX] = {
1542 [IPSET_CMD_CREATE] = {
1543 .call = ip_set_create,
1544 .attr_count = IPSET_ATTR_CMD_MAX,
1545 .policy = ip_set_create_policy,
1547 [IPSET_CMD_DESTROY] = {
1548 .call = ip_set_destroy,
1549 .attr_count = IPSET_ATTR_CMD_MAX,
1550 .policy = ip_set_setname_policy,
1552 [IPSET_CMD_FLUSH] = {
1553 .call = ip_set_flush,
1554 .attr_count = IPSET_ATTR_CMD_MAX,
1555 .policy = ip_set_setname_policy,
1557 [IPSET_CMD_RENAME] = {
1558 .call = ip_set_rename,
1559 .attr_count = IPSET_ATTR_CMD_MAX,
1560 .policy = ip_set_setname2_policy,
1562 [IPSET_CMD_SWAP] = {
1563 .call = ip_set_swap,
1564 .attr_count = IPSET_ATTR_CMD_MAX,
1565 .policy = ip_set_setname2_policy,
1567 [IPSET_CMD_LIST] = {
1568 .call = ip_set_dump,
1569 .attr_count = IPSET_ATTR_CMD_MAX,
1570 .policy = ip_set_setname_policy,
1572 [IPSET_CMD_SAVE] = {
1573 .call = ip_set_dump,
1574 .attr_count = IPSET_ATTR_CMD_MAX,
1575 .policy = ip_set_setname_policy,
1578 .call = ip_set_uadd,
1579 .attr_count = IPSET_ATTR_CMD_MAX,
1580 .policy = ip_set_adt_policy,
1583 .call = ip_set_udel,
1584 .attr_count = IPSET_ATTR_CMD_MAX,
1585 .policy = ip_set_adt_policy,
1587 [IPSET_CMD_TEST] = {
1588 .call = ip_set_utest,
1589 .attr_count = IPSET_ATTR_CMD_MAX,
1590 .policy = ip_set_adt_policy,
1592 [IPSET_CMD_HEADER] = {
1593 .call = ip_set_header,
1594 .attr_count = IPSET_ATTR_CMD_MAX,
1595 .policy = ip_set_setname_policy,
1597 [IPSET_CMD_TYPE] = {
1598 .call = ip_set_type,
1599 .attr_count = IPSET_ATTR_CMD_MAX,
1600 .policy = ip_set_type_policy,
1602 [IPSET_CMD_PROTOCOL] = {
1603 .call = ip_set_protocol,
1604 .attr_count = IPSET_ATTR_CMD_MAX,
1605 .policy = ip_set_protocol_policy,
1609 static struct nfnetlink_subsystem ip_set_netlink_subsys __read_mostly = {
1611 .subsys_id = NFNL_SUBSYS_IPSET,
1612 .cb_count = IPSET_MSG_MAX,
1613 .cb = ip_set_netlink_subsys_cb,
1616 /* Interface to iptables/ip6tables */
1619 ip_set_sockfn_get(struct sock *sk, int optval, void __user *user, int *len)
1623 int copylen = *len, ret = 0;
1625 if (!capable(CAP_NET_ADMIN))
1627 if (optval != SO_IP_SET)
1629 if (*len < sizeof(unsigned int))
1632 data = vmalloc(*len);
1635 if (copy_from_user(data, user, *len) != 0) {
1639 op = (unsigned int *) data;
1641 if (*op < IP_SET_OP_VERSION) {
1642 /* Check the version at the beginning of operations */
1643 struct ip_set_req_version *req_version = data;
1644 if (req_version->version != IPSET_PROTOCOL) {
1651 case IP_SET_OP_VERSION: {
1652 struct ip_set_req_version *req_version = data;
1654 if (*len != sizeof(struct ip_set_req_version)) {
1659 req_version->version = IPSET_PROTOCOL;
1660 ret = copy_to_user(user, req_version,
1661 sizeof(struct ip_set_req_version));
1664 case IP_SET_OP_GET_BYNAME: {
1665 struct ip_set_req_get_set *req_get = data;
1667 if (*len != sizeof(struct ip_set_req_get_set)) {
1671 req_get->set.name[IPSET_MAXNAMELEN - 1] = '\0';
1673 req_get->set.index = find_set_id(req_get->set.name);
1677 case IP_SET_OP_GET_BYINDEX: {
1678 struct ip_set_req_get_set *req_get = data;
1680 if (*len != sizeof(struct ip_set_req_get_set) ||
1681 req_get->set.index >= ip_set_max) {
1686 strncpy(req_get->set.name,
1687 ip_set_list[req_get->set.index]
1688 ? ip_set_list[req_get->set.index]->name : "",
1696 } /* end of switch(op) */
1699 ret = copy_to_user(user, data, copylen);
1708 static struct nf_sockopt_ops so_set __read_mostly = {
1710 .get_optmin = SO_IP_SET,
1711 .get_optmax = SO_IP_SET + 1,
1712 .get = &ip_set_sockfn_get,
1713 .owner = THIS_MODULE,
1722 ip_set_max = max_sets;
1723 if (ip_set_max >= IPSET_INVALID_ID)
1724 ip_set_max = IPSET_INVALID_ID - 1;
1726 ip_set_list = kzalloc(sizeof(struct ip_set *) * ip_set_max,
1731 ret = nfnetlink_subsys_register(&ip_set_netlink_subsys);
1733 pr_err("ip_set: cannot register with nfnetlink.\n");
1737 ret = nf_register_sockopt(&so_set);
1739 pr_err("SO_SET registry failed: %d\n", ret);
1740 nfnetlink_subsys_unregister(&ip_set_netlink_subsys);
1745 pr_notice("ip_set: protocol %u\n", IPSET_PROTOCOL);
1752 /* There can't be any existing set */
1753 nf_unregister_sockopt(&so_set);
1754 nfnetlink_subsys_unregister(&ip_set_netlink_subsys);
1756 pr_debug("these are the famous last words\n");
1759 module_init(ip_set_init);
1760 module_exit(ip_set_fini);