1 /* Copyright (C) 2008-2013 Jozsef Kadlecsik <kadlec@blackhole.kfki.hu>
3 * This program is free software; you can redistribute it and/or modify
4 * it under the terms of the GNU General Public License version 2 as
5 * published by the Free Software Foundation.
8 /* Kernel module implementing an IP set type: the list:set type */
10 #include <linux/module.h>
12 #include <linux/rculist.h>
13 #include <linux/skbuff.h>
14 #include <linux/errno.h>
16 #include <linux/netfilter/ipset/ip_set.h>
17 #include <linux/netfilter/ipset/ip_set_list.h>
19 #define IPSET_TYPE_REV_MIN 0
20 /* 1 Counters support added */
21 /* 2 Comments support added */
22 #define IPSET_TYPE_REV_MAX 3 /* skbinfo support added */
24 MODULE_LICENSE("GPL");
25 MODULE_AUTHOR("Jozsef Kadlecsik <kadlec@blackhole.kfki.hu>");
26 IP_SET_MODULE_DESC("list:set", IPSET_TYPE_REV_MIN, IPSET_TYPE_REV_MAX);
27 MODULE_ALIAS("ip_set_list:set");
32 struct list_head list;
33 struct ip_set *set; /* Sigh, in order to cleanup reference */
35 } __aligned(__alignof__(u64));
45 u32 size; /* size of set list array */
46 struct timer_list gc; /* garbage collection */
47 struct net *net; /* namespace */
48 struct list_head members; /* the set members */
52 list_set_ktest(struct ip_set *set, const struct sk_buff *skb,
53 const struct xt_action_param *par,
54 struct ip_set_adt_opt *opt, const struct ip_set_ext *ext)
56 struct list_set *map = set->data;
58 u32 cmdflags = opt->cmdflags;
61 /* Don't lookup sub-counters at all */
62 opt->cmdflags &= ~IPSET_FLAG_MATCH_COUNTERS;
63 if (opt->cmdflags & IPSET_FLAG_SKIP_SUBCOUNTER_UPDATE)
64 opt->cmdflags |= IPSET_FLAG_SKIP_COUNTER_UPDATE;
65 list_for_each_entry_rcu(e, &map->members, list) {
66 if (SET_WITH_TIMEOUT(set) &&
67 ip_set_timeout_expired(ext_timeout(e, set)))
69 ret = ip_set_test(e->id, skb, par, opt);
71 if (SET_WITH_COUNTER(set))
72 ip_set_update_counter(ext_counter(e, set),
75 if (SET_WITH_SKBINFO(set))
76 ip_set_get_skbinfo(ext_skbinfo(e, set),
86 list_set_kadd(struct ip_set *set, const struct sk_buff *skb,
87 const struct xt_action_param *par,
88 struct ip_set_adt_opt *opt, const struct ip_set_ext *ext)
90 struct list_set *map = set->data;
94 list_for_each_entry(e, &map->members, list) {
95 if (SET_WITH_TIMEOUT(set) &&
96 ip_set_timeout_expired(ext_timeout(e, set)))
98 ret = ip_set_add(e->id, skb, par, opt);
106 list_set_kdel(struct ip_set *set, const struct sk_buff *skb,
107 const struct xt_action_param *par,
108 struct ip_set_adt_opt *opt, const struct ip_set_ext *ext)
110 struct list_set *map = set->data;
114 list_for_each_entry(e, &map->members, list) {
115 if (SET_WITH_TIMEOUT(set) &&
116 ip_set_timeout_expired(ext_timeout(e, set)))
118 ret = ip_set_del(e->id, skb, par, opt);
126 list_set_kadt(struct ip_set *set, const struct sk_buff *skb,
127 const struct xt_action_param *par,
128 enum ipset_adt adt, struct ip_set_adt_opt *opt)
130 struct ip_set_ext ext = IP_SET_INIT_KEXT(skb, opt, set);
136 ret = list_set_ktest(set, skb, par, opt, &ext);
139 ret = list_set_kadd(set, skb, par, opt, &ext);
142 ret = list_set_kdel(set, skb, par, opt, &ext);
152 /* Userspace interfaces: we are protected by the nfnl mutex */
155 __list_set_del_rcu(struct rcu_head * rcu)
157 struct set_elem *e = container_of(rcu, struct set_elem, rcu);
158 struct ip_set *set = e->set;
160 ip_set_ext_destroy(set, e);
165 list_set_del(struct ip_set *set, struct set_elem *e)
167 struct list_set *map = set->data;
170 list_del_rcu(&e->list);
171 ip_set_put_byindex(map->net, e->id);
172 call_rcu(&e->rcu, __list_set_del_rcu);
176 list_set_replace(struct ip_set *set, struct set_elem *e, struct set_elem *old)
178 struct list_set *map = set->data;
180 list_replace_rcu(&old->list, &e->list);
181 ip_set_put_byindex(map->net, old->id);
182 call_rcu(&old->rcu, __list_set_del_rcu);
186 set_cleanup_entries(struct ip_set *set)
188 struct list_set *map = set->data;
189 struct set_elem *e, *n;
191 list_for_each_entry_safe(e, n, &map->members, list)
192 if (ip_set_timeout_expired(ext_timeout(e, set)))
193 list_set_del(set, e);
197 list_set_utest(struct ip_set *set, void *value, const struct ip_set_ext *ext,
198 struct ip_set_ext *mext, u32 flags)
200 struct list_set *map = set->data;
201 struct set_adt_elem *d = value;
202 struct set_elem *e, *next, *prev = NULL;
205 list_for_each_entry(e, &map->members, list) {
206 if (SET_WITH_TIMEOUT(set) &&
207 ip_set_timeout_expired(ext_timeout(e, set)))
209 else if (e->id != d->id) {
214 if (d->before == 0) {
216 } else if (d->before > 0) {
217 next = list_next_entry(e, list);
218 ret = !list_is_last(&e->list, &map->members) &&
219 next->id == d->refid;
221 ret = prev && prev->id == d->refid;
229 list_set_init_extensions(struct ip_set *set, const struct ip_set_ext *ext,
232 if (SET_WITH_COUNTER(set))
233 ip_set_init_counter(ext_counter(e, set), ext);
234 if (SET_WITH_COMMENT(set))
235 ip_set_init_comment(set, ext_comment(e, set), ext);
236 if (SET_WITH_SKBINFO(set))
237 ip_set_init_skbinfo(ext_skbinfo(e, set), ext);
238 /* Update timeout last */
239 if (SET_WITH_TIMEOUT(set))
240 ip_set_timeout_set(ext_timeout(e, set), ext->timeout);
244 list_set_uadd(struct ip_set *set, void *value, const struct ip_set_ext *ext,
245 struct ip_set_ext *mext, u32 flags)
247 struct list_set *map = set->data;
248 struct set_adt_elem *d = value;
249 struct set_elem *e, *n, *prev, *next;
250 bool flag_exist = flags & IPSET_FLAG_EXIST;
252 /* Find where to add the new entry */
253 n = prev = next = NULL;
254 list_for_each_entry(e, &map->members, list) {
255 if (SET_WITH_TIMEOUT(set) &&
256 ip_set_timeout_expired(ext_timeout(e, set)))
258 else if (d->id == e->id)
260 else if (d->before == 0 || e->id != d->refid)
262 else if (d->before > 0)
268 /* If before/after is used on an empty set */
269 if ((d->before > 0 && !next) ||
270 (d->before < 0 && !prev))
271 return -IPSET_ERR_REF_EXIST;
273 /* Re-add already existing element */
276 return -IPSET_ERR_EXIST;
277 /* Update extensions */
278 ip_set_ext_destroy(set, n);
279 list_set_init_extensions(set, ext, n);
281 /* Set is already added to the list */
282 ip_set_put_byindex(map->net, d->id);
286 if (d->before == 0) {
288 n = list_empty(&map->members) ? NULL :
289 list_last_entry(&map->members, struct set_elem, list);
290 } else if (d->before > 0) {
291 /* Insert after next element */
292 if (!list_is_last(&next->list, &map->members))
293 n = list_next_entry(next, list);
295 /* Insert before prev element */
296 if (prev->list.prev != &map->members)
297 n = list_prev_entry(prev, list);
299 /* Can we replace a timed out entry? */
301 !(SET_WITH_TIMEOUT(set) &&
302 ip_set_timeout_expired(ext_timeout(n, set))))
305 e = kzalloc(set->dsize, GFP_ATOMIC);
310 INIT_LIST_HEAD(&e->list);
311 list_set_init_extensions(set, ext, e);
313 list_set_replace(set, e, n);
315 list_add_tail_rcu(&e->list, &next->list);
317 list_add_rcu(&e->list, &prev->list);
319 list_add_tail_rcu(&e->list, &map->members);
326 list_set_udel(struct ip_set *set, void *value, const struct ip_set_ext *ext,
327 struct ip_set_ext *mext, u32 flags)
329 struct list_set *map = set->data;
330 struct set_adt_elem *d = value;
331 struct set_elem *e, *next, *prev = NULL;
333 list_for_each_entry(e, &map->members, list) {
334 if (SET_WITH_TIMEOUT(set) &&
335 ip_set_timeout_expired(ext_timeout(e, set)))
337 else if (e->id != d->id) {
343 next = list_next_entry(e, list);
344 if (list_is_last(&e->list, &map->members) ||
345 next->id != d->refid)
346 return -IPSET_ERR_REF_EXIST;
347 } else if (d->before < 0) {
348 if (!prev || prev->id != d->refid)
349 return -IPSET_ERR_REF_EXIST;
351 list_set_del(set, e);
354 return d->before != 0 ? -IPSET_ERR_REF_EXIST : -IPSET_ERR_EXIST;
358 list_set_uadt(struct ip_set *set, struct nlattr *tb[],
359 enum ipset_adt adt, u32 *lineno, u32 flags, bool retried)
361 struct list_set *map = set->data;
362 ipset_adtfn adtfn = set->variant->adt[adt];
363 struct set_adt_elem e = { .refid = IPSET_INVALID_ID };
364 struct ip_set_ext ext = IP_SET_INIT_UEXT(set);
368 if (tb[IPSET_ATTR_LINENO])
369 *lineno = nla_get_u32(tb[IPSET_ATTR_LINENO]);
371 if (unlikely(!tb[IPSET_ATTR_NAME] ||
372 !ip_set_optattr_netorder(tb, IPSET_ATTR_CADT_FLAGS)))
373 return -IPSET_ERR_PROTOCOL;
375 ret = ip_set_get_extensions(set, tb, &ext);
378 e.id = ip_set_get_byname(map->net, nla_data(tb[IPSET_ATTR_NAME]), &s);
379 if (e.id == IPSET_INVALID_ID)
380 return -IPSET_ERR_NAME;
381 /* "Loop detection" */
382 if (s->type->features & IPSET_TYPE_NAME) {
383 ret = -IPSET_ERR_LOOP;
387 if (tb[IPSET_ATTR_CADT_FLAGS]) {
388 u32 f = ip_set_get_h32(tb[IPSET_ATTR_CADT_FLAGS]);
390 e.before = f & IPSET_FLAG_BEFORE;
393 if (e.before && !tb[IPSET_ATTR_NAMEREF]) {
394 ret = -IPSET_ERR_BEFORE;
398 if (tb[IPSET_ATTR_NAMEREF]) {
399 e.refid = ip_set_get_byname(map->net,
400 nla_data(tb[IPSET_ATTR_NAMEREF]),
402 if (e.refid == IPSET_INVALID_ID) {
403 ret = -IPSET_ERR_NAMEREF;
409 if (adt != IPSET_TEST && SET_WITH_TIMEOUT(set))
410 set_cleanup_entries(set);
412 ret = adtfn(set, &e, &ext, &ext, flags);
415 if (e.refid != IPSET_INVALID_ID)
416 ip_set_put_byindex(map->net, e.refid);
417 if (adt != IPSET_ADD || ret)
418 ip_set_put_byindex(map->net, e.id);
420 return ip_set_eexist(ret, flags) ? 0 : ret;
424 list_set_flush(struct ip_set *set)
426 struct list_set *map = set->data;
427 struct set_elem *e, *n;
429 list_for_each_entry_safe(e, n, &map->members, list)
430 list_set_del(set, e);
436 list_set_destroy(struct ip_set *set)
438 struct list_set *map = set->data;
439 struct set_elem *e, *n;
441 if (SET_WITH_TIMEOUT(set))
442 del_timer_sync(&map->gc);
444 list_for_each_entry_safe(e, n, &map->members, list) {
446 ip_set_put_byindex(map->net, e->id);
447 ip_set_ext_destroy(set, e);
455 /* Calculate the actual memory size of the set data */
457 list_set_memsize(const struct list_set *map, size_t dsize)
464 list_for_each_entry_rcu(e, &map->members, list)
468 memsize = sizeof(*map) + n * dsize;
474 list_set_head(struct ip_set *set, struct sk_buff *skb)
476 const struct list_set *map = set->data;
477 struct nlattr *nested;
478 size_t memsize = list_set_memsize(map, set->dsize) + set->ext_size;
480 nested = ipset_nest_start(skb, IPSET_ATTR_DATA);
482 goto nla_put_failure;
483 if (nla_put_net32(skb, IPSET_ATTR_SIZE, htonl(map->size)) ||
484 nla_put_net32(skb, IPSET_ATTR_REFERENCES, htonl(set->ref)) ||
485 nla_put_net32(skb, IPSET_ATTR_MEMSIZE, htonl(memsize)) ||
486 nla_put_net32(skb, IPSET_ATTR_ELEMENTS, htonl(set->elements)))
487 goto nla_put_failure;
488 if (unlikely(ip_set_put_flags(skb, set)))
489 goto nla_put_failure;
490 ipset_nest_end(skb, nested);
498 list_set_list(const struct ip_set *set,
499 struct sk_buff *skb, struct netlink_callback *cb)
501 const struct list_set *map = set->data;
502 struct nlattr *atd, *nested;
503 u32 i = 0, first = cb->args[IPSET_CB_ARG0];
504 char name[IPSET_MAXNAMELEN];
508 atd = ipset_nest_start(skb, IPSET_ATTR_ADT);
513 list_for_each_entry_rcu(e, &map->members, list) {
515 (SET_WITH_TIMEOUT(set) &&
516 ip_set_timeout_expired(ext_timeout(e, set)))) {
520 nested = ipset_nest_start(skb, IPSET_ATTR_DATA);
522 goto nla_put_failure;
523 ip_set_name_byindex(map->net, e->id, name);
524 if (nla_put_string(skb, IPSET_ATTR_NAME, name))
525 goto nla_put_failure;
526 if (ip_set_put_extensions(skb, set, e, true))
527 goto nla_put_failure;
528 ipset_nest_end(skb, nested);
532 ipset_nest_end(skb, atd);
533 /* Set listing finished */
534 cb->args[IPSET_CB_ARG0] = 0;
538 nla_nest_cancel(skb, nested);
539 if (unlikely(i == first)) {
540 nla_nest_cancel(skb, atd);
541 cb->args[IPSET_CB_ARG0] = 0;
544 cb->args[IPSET_CB_ARG0] = i;
545 ipset_nest_end(skb, atd);
553 list_set_same_set(const struct ip_set *a, const struct ip_set *b)
555 const struct list_set *x = a->data;
556 const struct list_set *y = b->data;
558 return x->size == y->size &&
559 a->timeout == b->timeout &&
560 a->extensions == b->extensions;
563 static const struct ip_set_type_variant set_variant = {
564 .kadt = list_set_kadt,
565 .uadt = list_set_uadt,
567 [IPSET_ADD] = list_set_uadd,
568 [IPSET_DEL] = list_set_udel,
569 [IPSET_TEST] = list_set_utest,
571 .destroy = list_set_destroy,
572 .flush = list_set_flush,
573 .head = list_set_head,
574 .list = list_set_list,
575 .same_set = list_set_same_set,
579 list_set_gc(unsigned long ul_set)
581 struct ip_set *set = (struct ip_set *)ul_set;
582 struct list_set *map = set->data;
584 spin_lock_bh(&set->lock);
585 set_cleanup_entries(set);
586 spin_unlock_bh(&set->lock);
588 map->gc.expires = jiffies + IPSET_GC_PERIOD(set->timeout) * HZ;
593 list_set_gc_init(struct ip_set *set, void (*gc)(unsigned long ul_set))
595 struct list_set *map = set->data;
597 setup_timer(&map->gc, gc, (unsigned long)set);
598 mod_timer(&map->gc, jiffies + IPSET_GC_PERIOD(set->timeout) * HZ);
601 /* Create list:set type of sets */
604 init_list_set(struct net *net, struct ip_set *set, u32 size)
606 struct list_set *map;
608 map = kzalloc(sizeof(*map), GFP_KERNEL);
614 INIT_LIST_HEAD(&map->members);
621 list_set_create(struct net *net, struct ip_set *set, struct nlattr *tb[],
624 u32 size = IP_SET_LIST_DEFAULT_SIZE;
626 if (unlikely(!ip_set_optattr_netorder(tb, IPSET_ATTR_SIZE) ||
627 !ip_set_optattr_netorder(tb, IPSET_ATTR_TIMEOUT) ||
628 !ip_set_optattr_netorder(tb, IPSET_ATTR_CADT_FLAGS)))
629 return -IPSET_ERR_PROTOCOL;
631 if (tb[IPSET_ATTR_SIZE])
632 size = ip_set_get_h32(tb[IPSET_ATTR_SIZE]);
633 if (size < IP_SET_LIST_MIN_SIZE)
634 size = IP_SET_LIST_MIN_SIZE;
636 set->variant = &set_variant;
637 set->dsize = ip_set_elem_len(set, tb, sizeof(struct set_elem),
638 __alignof__(struct set_elem));
639 if (!init_list_set(net, set, size))
641 if (tb[IPSET_ATTR_TIMEOUT]) {
642 set->timeout = ip_set_timeout_uget(tb[IPSET_ATTR_TIMEOUT]);
643 list_set_gc_init(set, list_set_gc);
648 static struct ip_set_type list_set_type __read_mostly = {
650 .protocol = IPSET_PROTOCOL,
651 .features = IPSET_TYPE_NAME | IPSET_DUMP_LAST,
652 .dimension = IPSET_DIM_ONE,
653 .family = NFPROTO_UNSPEC,
654 .revision_min = IPSET_TYPE_REV_MIN,
655 .revision_max = IPSET_TYPE_REV_MAX,
656 .create = list_set_create,
658 [IPSET_ATTR_SIZE] = { .type = NLA_U32 },
659 [IPSET_ATTR_TIMEOUT] = { .type = NLA_U32 },
660 [IPSET_ATTR_CADT_FLAGS] = { .type = NLA_U32 },
663 [IPSET_ATTR_NAME] = { .type = NLA_STRING,
664 .len = IPSET_MAXNAMELEN },
665 [IPSET_ATTR_NAMEREF] = { .type = NLA_STRING,
666 .len = IPSET_MAXNAMELEN },
667 [IPSET_ATTR_TIMEOUT] = { .type = NLA_U32 },
668 [IPSET_ATTR_LINENO] = { .type = NLA_U32 },
669 [IPSET_ATTR_CADT_FLAGS] = { .type = NLA_U32 },
670 [IPSET_ATTR_BYTES] = { .type = NLA_U64 },
671 [IPSET_ATTR_PACKETS] = { .type = NLA_U64 },
672 [IPSET_ATTR_COMMENT] = { .type = NLA_NUL_STRING,
673 .len = IPSET_MAX_COMMENT_SIZE },
674 [IPSET_ATTR_SKBMARK] = { .type = NLA_U64 },
675 [IPSET_ATTR_SKBPRIO] = { .type = NLA_U32 },
676 [IPSET_ATTR_SKBQUEUE] = { .type = NLA_U16 },
684 return ip_set_type_register(&list_set_type);
691 ip_set_type_unregister(&list_set_type);
694 module_init(list_set_init);
695 module_exit(list_set_fini);