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;
159 struct list_set *map = set->data;
161 ip_set_put_byindex(map->net, e->id);
162 ip_set_ext_destroy(set, e);
167 list_set_del(struct ip_set *set, struct set_elem *e)
170 list_del_rcu(&e->list);
171 call_rcu(&e->rcu, __list_set_del_rcu);
175 list_set_replace(struct set_elem *e, struct set_elem *old)
177 list_replace_rcu(&old->list, &e->list);
178 call_rcu(&old->rcu, __list_set_del_rcu);
182 set_cleanup_entries(struct ip_set *set)
184 struct list_set *map = set->data;
185 struct set_elem *e, *n;
187 list_for_each_entry_safe(e, n, &map->members, list)
188 if (ip_set_timeout_expired(ext_timeout(e, set)))
189 list_set_del(set, e);
193 list_set_utest(struct ip_set *set, void *value, const struct ip_set_ext *ext,
194 struct ip_set_ext *mext, u32 flags)
196 struct list_set *map = set->data;
197 struct set_adt_elem *d = value;
198 struct set_elem *e, *next, *prev = NULL;
201 list_for_each_entry(e, &map->members, list) {
202 if (SET_WITH_TIMEOUT(set) &&
203 ip_set_timeout_expired(ext_timeout(e, set)))
205 else if (e->id != d->id) {
210 if (d->before == 0) {
212 } else if (d->before > 0) {
213 next = list_next_entry(e, list);
214 ret = !list_is_last(&e->list, &map->members) &&
215 next->id == d->refid;
217 ret = prev && prev->id == d->refid;
225 list_set_init_extensions(struct ip_set *set, const struct ip_set_ext *ext,
228 if (SET_WITH_COUNTER(set))
229 ip_set_init_counter(ext_counter(e, set), ext);
230 if (SET_WITH_COMMENT(set))
231 ip_set_init_comment(set, ext_comment(e, set), ext);
232 if (SET_WITH_SKBINFO(set))
233 ip_set_init_skbinfo(ext_skbinfo(e, set), ext);
234 /* Update timeout last */
235 if (SET_WITH_TIMEOUT(set))
236 ip_set_timeout_set(ext_timeout(e, set), ext->timeout);
240 list_set_uadd(struct ip_set *set, void *value, const struct ip_set_ext *ext,
241 struct ip_set_ext *mext, u32 flags)
243 struct list_set *map = set->data;
244 struct set_adt_elem *d = value;
245 struct set_elem *e, *n, *prev, *next;
246 bool flag_exist = flags & IPSET_FLAG_EXIST;
248 /* Find where to add the new entry */
249 n = prev = next = NULL;
250 list_for_each_entry(e, &map->members, list) {
251 if (SET_WITH_TIMEOUT(set) &&
252 ip_set_timeout_expired(ext_timeout(e, set)))
254 else if (d->id == e->id)
256 else if (d->before == 0 || e->id != d->refid)
258 else if (d->before > 0)
264 /* If before/after is used on an empty set */
265 if ((d->before > 0 && !next) ||
266 (d->before < 0 && !prev))
267 return -IPSET_ERR_REF_EXIST;
269 /* Re-add already existing element */
272 return -IPSET_ERR_EXIST;
273 /* Update extensions */
274 ip_set_ext_destroy(set, n);
275 list_set_init_extensions(set, ext, n);
277 /* Set is already added to the list */
278 ip_set_put_byindex(map->net, d->id);
282 if (d->before == 0) {
284 n = list_empty(&map->members) ? NULL :
285 list_last_entry(&map->members, struct set_elem, list);
286 } else if (d->before > 0) {
287 /* Insert after next element */
288 if (!list_is_last(&next->list, &map->members))
289 n = list_next_entry(next, list);
291 /* Insert before prev element */
292 if (prev->list.prev != &map->members)
293 n = list_prev_entry(prev, list);
295 /* Can we replace a timed out entry? */
297 !(SET_WITH_TIMEOUT(set) &&
298 ip_set_timeout_expired(ext_timeout(n, set))))
301 e = kzalloc(set->dsize, GFP_ATOMIC);
306 INIT_LIST_HEAD(&e->list);
307 list_set_init_extensions(set, ext, e);
309 list_set_replace(e, n);
311 list_add_tail_rcu(&e->list, &next->list);
313 list_add_rcu(&e->list, &prev->list);
315 list_add_tail_rcu(&e->list, &map->members);
322 list_set_udel(struct ip_set *set, void *value, const struct ip_set_ext *ext,
323 struct ip_set_ext *mext, u32 flags)
325 struct list_set *map = set->data;
326 struct set_adt_elem *d = value;
327 struct set_elem *e, *next, *prev = NULL;
329 list_for_each_entry(e, &map->members, list) {
330 if (SET_WITH_TIMEOUT(set) &&
331 ip_set_timeout_expired(ext_timeout(e, set)))
333 else if (e->id != d->id) {
339 next = list_next_entry(e, list);
340 if (list_is_last(&e->list, &map->members) ||
341 next->id != d->refid)
342 return -IPSET_ERR_REF_EXIST;
343 } else if (d->before < 0) {
344 if (!prev || prev->id != d->refid)
345 return -IPSET_ERR_REF_EXIST;
347 list_set_del(set, e);
350 return d->before != 0 ? -IPSET_ERR_REF_EXIST : -IPSET_ERR_EXIST;
354 list_set_uadt(struct ip_set *set, struct nlattr *tb[],
355 enum ipset_adt adt, u32 *lineno, u32 flags, bool retried)
357 struct list_set *map = set->data;
358 ipset_adtfn adtfn = set->variant->adt[adt];
359 struct set_adt_elem e = { .refid = IPSET_INVALID_ID };
360 struct ip_set_ext ext = IP_SET_INIT_UEXT(set);
364 if (tb[IPSET_ATTR_LINENO])
365 *lineno = nla_get_u32(tb[IPSET_ATTR_LINENO]);
367 if (unlikely(!tb[IPSET_ATTR_NAME] ||
368 !ip_set_optattr_netorder(tb, IPSET_ATTR_CADT_FLAGS)))
369 return -IPSET_ERR_PROTOCOL;
371 ret = ip_set_get_extensions(set, tb, &ext);
374 e.id = ip_set_get_byname(map->net, nla_data(tb[IPSET_ATTR_NAME]), &s);
375 if (e.id == IPSET_INVALID_ID)
376 return -IPSET_ERR_NAME;
377 /* "Loop detection" */
378 if (s->type->features & IPSET_TYPE_NAME) {
379 ret = -IPSET_ERR_LOOP;
383 if (tb[IPSET_ATTR_CADT_FLAGS]) {
384 u32 f = ip_set_get_h32(tb[IPSET_ATTR_CADT_FLAGS]);
386 e.before = f & IPSET_FLAG_BEFORE;
389 if (e.before && !tb[IPSET_ATTR_NAMEREF]) {
390 ret = -IPSET_ERR_BEFORE;
394 if (tb[IPSET_ATTR_NAMEREF]) {
395 e.refid = ip_set_get_byname(map->net,
396 nla_data(tb[IPSET_ATTR_NAMEREF]),
398 if (e.refid == IPSET_INVALID_ID) {
399 ret = -IPSET_ERR_NAMEREF;
405 if (adt != IPSET_TEST && SET_WITH_TIMEOUT(set))
406 set_cleanup_entries(set);
408 ret = adtfn(set, &e, &ext, &ext, flags);
411 if (e.refid != IPSET_INVALID_ID)
412 ip_set_put_byindex(map->net, e.refid);
413 if (adt != IPSET_ADD || ret)
414 ip_set_put_byindex(map->net, e.id);
416 return ip_set_eexist(ret, flags) ? 0 : ret;
420 list_set_flush(struct ip_set *set)
422 struct list_set *map = set->data;
423 struct set_elem *e, *n;
425 list_for_each_entry_safe(e, n, &map->members, list)
426 list_set_del(set, e);
432 list_set_destroy(struct ip_set *set)
434 struct list_set *map = set->data;
435 struct set_elem *e, *n;
437 if (SET_WITH_TIMEOUT(set))
438 del_timer_sync(&map->gc);
440 list_for_each_entry_safe(e, n, &map->members, list) {
442 ip_set_put_byindex(map->net, e->id);
443 ip_set_ext_destroy(set, e);
451 /* Calculate the actual memory size of the set data */
453 list_set_memsize(const struct list_set *map, size_t dsize)
460 list_for_each_entry_rcu(e, &map->members, list)
464 memsize = sizeof(*map) + n * dsize;
470 list_set_head(struct ip_set *set, struct sk_buff *skb)
472 const struct list_set *map = set->data;
473 struct nlattr *nested;
474 size_t memsize = list_set_memsize(map, set->dsize) + set->ext_size;
476 nested = ipset_nest_start(skb, IPSET_ATTR_DATA);
478 goto nla_put_failure;
479 if (nla_put_net32(skb, IPSET_ATTR_SIZE, htonl(map->size)) ||
480 nla_put_net32(skb, IPSET_ATTR_REFERENCES, htonl(set->ref)) ||
481 nla_put_net32(skb, IPSET_ATTR_MEMSIZE, htonl(memsize)) ||
482 nla_put_net32(skb, IPSET_ATTR_ELEMENTS, htonl(set->elements)))
483 goto nla_put_failure;
484 if (unlikely(ip_set_put_flags(skb, set)))
485 goto nla_put_failure;
486 ipset_nest_end(skb, nested);
494 list_set_list(const struct ip_set *set,
495 struct sk_buff *skb, struct netlink_callback *cb)
497 const struct list_set *map = set->data;
498 struct nlattr *atd, *nested;
499 u32 i = 0, first = cb->args[IPSET_CB_ARG0];
503 atd = ipset_nest_start(skb, IPSET_ATTR_ADT);
508 list_for_each_entry_rcu(e, &map->members, list) {
510 (SET_WITH_TIMEOUT(set) &&
511 ip_set_timeout_expired(ext_timeout(e, set)))) {
515 nested = ipset_nest_start(skb, IPSET_ATTR_DATA);
517 goto nla_put_failure;
518 if (nla_put_string(skb, IPSET_ATTR_NAME,
519 ip_set_name_byindex(map->net, e->id)))
520 goto nla_put_failure;
521 if (ip_set_put_extensions(skb, set, e, true))
522 goto nla_put_failure;
523 ipset_nest_end(skb, nested);
527 ipset_nest_end(skb, atd);
528 /* Set listing finished */
529 cb->args[IPSET_CB_ARG0] = 0;
533 nla_nest_cancel(skb, nested);
534 if (unlikely(i == first)) {
535 nla_nest_cancel(skb, atd);
536 cb->args[IPSET_CB_ARG0] = 0;
539 cb->args[IPSET_CB_ARG0] = i;
541 ipset_nest_end(skb, atd);
548 list_set_same_set(const struct ip_set *a, const struct ip_set *b)
550 const struct list_set *x = a->data;
551 const struct list_set *y = b->data;
553 return x->size == y->size &&
554 a->timeout == b->timeout &&
555 a->extensions == b->extensions;
558 static const struct ip_set_type_variant set_variant = {
559 .kadt = list_set_kadt,
560 .uadt = list_set_uadt,
562 [IPSET_ADD] = list_set_uadd,
563 [IPSET_DEL] = list_set_udel,
564 [IPSET_TEST] = list_set_utest,
566 .destroy = list_set_destroy,
567 .flush = list_set_flush,
568 .head = list_set_head,
569 .list = list_set_list,
570 .same_set = list_set_same_set,
574 list_set_gc(unsigned long ul_set)
576 struct ip_set *set = (struct ip_set *)ul_set;
577 struct list_set *map = set->data;
579 spin_lock_bh(&set->lock);
580 set_cleanup_entries(set);
581 spin_unlock_bh(&set->lock);
583 map->gc.expires = jiffies + IPSET_GC_PERIOD(set->timeout) * HZ;
588 list_set_gc_init(struct ip_set *set, void (*gc)(unsigned long ul_set))
590 struct list_set *map = set->data;
592 setup_timer(&map->gc, gc, (unsigned long)set);
593 mod_timer(&map->gc, jiffies + IPSET_GC_PERIOD(set->timeout) * HZ);
596 /* Create list:set type of sets */
599 init_list_set(struct net *net, struct ip_set *set, u32 size)
601 struct list_set *map;
603 map = kzalloc(sizeof(*map), GFP_KERNEL);
609 INIT_LIST_HEAD(&map->members);
616 list_set_create(struct net *net, struct ip_set *set, struct nlattr *tb[],
619 u32 size = IP_SET_LIST_DEFAULT_SIZE;
621 if (unlikely(!ip_set_optattr_netorder(tb, IPSET_ATTR_SIZE) ||
622 !ip_set_optattr_netorder(tb, IPSET_ATTR_TIMEOUT) ||
623 !ip_set_optattr_netorder(tb, IPSET_ATTR_CADT_FLAGS)))
624 return -IPSET_ERR_PROTOCOL;
626 if (tb[IPSET_ATTR_SIZE])
627 size = ip_set_get_h32(tb[IPSET_ATTR_SIZE]);
628 if (size < IP_SET_LIST_MIN_SIZE)
629 size = IP_SET_LIST_MIN_SIZE;
631 set->variant = &set_variant;
632 set->dsize = ip_set_elem_len(set, tb, sizeof(struct set_elem),
633 __alignof__(struct set_elem));
634 if (!init_list_set(net, set, size))
636 if (tb[IPSET_ATTR_TIMEOUT]) {
637 set->timeout = ip_set_timeout_uget(tb[IPSET_ATTR_TIMEOUT]);
638 list_set_gc_init(set, list_set_gc);
643 static struct ip_set_type list_set_type __read_mostly = {
645 .protocol = IPSET_PROTOCOL,
646 .features = IPSET_TYPE_NAME | IPSET_DUMP_LAST,
647 .dimension = IPSET_DIM_ONE,
648 .family = NFPROTO_UNSPEC,
649 .revision_min = IPSET_TYPE_REV_MIN,
650 .revision_max = IPSET_TYPE_REV_MAX,
651 .create = list_set_create,
653 [IPSET_ATTR_SIZE] = { .type = NLA_U32 },
654 [IPSET_ATTR_TIMEOUT] = { .type = NLA_U32 },
655 [IPSET_ATTR_CADT_FLAGS] = { .type = NLA_U32 },
658 [IPSET_ATTR_NAME] = { .type = NLA_STRING,
659 .len = IPSET_MAXNAMELEN },
660 [IPSET_ATTR_NAMEREF] = { .type = NLA_STRING,
661 .len = IPSET_MAXNAMELEN },
662 [IPSET_ATTR_TIMEOUT] = { .type = NLA_U32 },
663 [IPSET_ATTR_LINENO] = { .type = NLA_U32 },
664 [IPSET_ATTR_CADT_FLAGS] = { .type = NLA_U32 },
665 [IPSET_ATTR_BYTES] = { .type = NLA_U64 },
666 [IPSET_ATTR_PACKETS] = { .type = NLA_U64 },
667 [IPSET_ATTR_COMMENT] = { .type = NLA_NUL_STRING,
668 .len = IPSET_MAX_COMMENT_SIZE },
669 [IPSET_ATTR_SKBMARK] = { .type = NLA_U64 },
670 [IPSET_ATTR_SKBPRIO] = { .type = NLA_U32 },
671 [IPSET_ATTR_SKBQUEUE] = { .type = NLA_U16 },
679 return ip_set_type_register(&list_set_type);
686 ip_set_type_unregister(&list_set_type);
689 module_init(list_set_init);
690 module_exit(list_set_fini);