1 /* Copyright (C) 2008-2013 Jozsef Kadlecsik <kadlec@blackhole.kfki.hu>
2 *
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.
6 */
7
8 /* Kernel module implementing an IP set type: the list:set type */
9
10 #include <linux/module.h>
11 #include <linux/ip.h>
12 #include <linux/rculist.h>
13 #include <linux/skbuff.h>
14 #include <linux/errno.h>
15
16 #include <linux/netfilter/ipset/ip_set.h>
17 #include <linux/netfilter/ipset/ip_set_list.h>
18
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 */
23
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");
28
29 /* Member elements */
30 struct set_elem {
31 struct rcu_head rcu;
32 struct list_head list;
33 ip_set_id_t id;
34 } __aligned(__alignof__(u64));
35
36 struct set_adt_elem {
37 ip_set_id_t id;
38 ip_set_id_t refid;
39 int before;
40 };
41
42 /* Type structure */
43 struct list_set {
44 u32 size; /* size of set list array */
45 struct timer_list gc; /* garbage collection */
46 struct net *net; /* namespace */
47 struct list_head members; /* the set members */
48 };
49
50 static int
51 list_set_ktest(struct ip_set *set, const struct sk_buff *skb,
52 const struct xt_action_param *par,
53 struct ip_set_adt_opt *opt, const struct ip_set_ext *ext)
54 {
55 struct list_set *map = set->data;
56 struct set_elem *e;
57 u32 cmdflags = opt->cmdflags;
58 int ret;
59
60 /* Don't lookup sub-counters at all */
61 opt->cmdflags &= ~IPSET_FLAG_MATCH_COUNTERS;
62 if (opt->cmdflags & IPSET_FLAG_SKIP_SUBCOUNTER_UPDATE)
63 opt->cmdflags |= IPSET_FLAG_SKIP_COUNTER_UPDATE;
64 list_for_each_entry_rcu(e, &map->members, list) {
65 if (SET_WITH_TIMEOUT(set) &&
66 ip_set_timeout_expired(ext_timeout(e, set)))
67 continue;
68 ret = ip_set_test(e->id, skb, par, opt);
69 if (ret > 0) {
70 if (SET_WITH_COUNTER(set))
71 ip_set_update_counter(ext_counter(e, set),
72 ext, &opt->ext,
73 cmdflags);
74 if (SET_WITH_SKBINFO(set))
75 ip_set_get_skbinfo(ext_skbinfo(e, set),
76 ext, &opt->ext,
77 cmdflags);
78 return ret;
79 }
80 }
81 return 0;
82 }
83
84 static int
85 list_set_kadd(struct ip_set *set, const struct sk_buff *skb,
86 const struct xt_action_param *par,
87 struct ip_set_adt_opt *opt, const struct ip_set_ext *ext)
88 {
89 struct list_set *map = set->data;
90 struct set_elem *e;
91 int ret;
92
93 list_for_each_entry(e, &map->members, list) {
94 if (SET_WITH_TIMEOUT(set) &&
95 ip_set_timeout_expired(ext_timeout(e, set)))
96 continue;
97 ret = ip_set_add(e->id, skb, par, opt);
98 if (ret == 0)
99 return ret;
100 }
101 return 0;
102 }
103
104 static int
105 list_set_kdel(struct ip_set *set, const struct sk_buff *skb,
106 const struct xt_action_param *par,
107 struct ip_set_adt_opt *opt, const struct ip_set_ext *ext)
108 {
109 struct list_set *map = set->data;
110 struct set_elem *e;
111 int ret;
112
113 list_for_each_entry(e, &map->members, list) {
114 if (SET_WITH_TIMEOUT(set) &&
115 ip_set_timeout_expired(ext_timeout(e, set)))
116 continue;
117 ret = ip_set_del(e->id, skb, par, opt);
118 if (ret == 0)
119 return ret;
120 }
121 return 0;
122 }
123
124 static int
125 list_set_kadt(struct ip_set *set, const struct sk_buff *skb,
126 const struct xt_action_param *par,
127 enum ipset_adt adt, struct ip_set_adt_opt *opt)
128 {
129 struct ip_set_ext ext = IP_SET_INIT_KEXT(skb, opt, set);
130 int ret = -EINVAL;
131
132 rcu_read_lock();
133 switch (adt) {
134 case IPSET_TEST:
135 ret = list_set_ktest(set, skb, par, opt, &ext);
136 break;
137 case IPSET_ADD:
138 ret = list_set_kadd(set, skb, par, opt, &ext);
139 break;
140 case IPSET_DEL:
141 ret = list_set_kdel(set, skb, par, opt, &ext);
142 break;
143 default:
144 break;
145 }
146 rcu_read_unlock();
147
148 return ret;
149 }
150
151 /* Userspace interfaces: we are protected by the nfnl mutex */
152
153 static void
154 __list_set_del(struct ip_set *set, struct set_elem *e)
155 {
156 struct list_set *map = set->data;
157
158 ip_set_put_byindex(map->net, e->id);
159 /* We may call it, because we don't have a to be destroyed
160 * extension which is used by the kernel.
161 */
162 ip_set_ext_destroy(set, e);
163 kfree_rcu(e, rcu);
164 }
165
166 static inline void
167 list_set_del(struct ip_set *set, struct set_elem *e)
168 {
169 list_del_rcu(&e->list);
170 __list_set_del(set, e);
171 }
172
173 static inline void
174 list_set_replace(struct ip_set *set, struct set_elem *e, struct set_elem *old)
175 {
176 list_replace_rcu(&old->list, &e->list);
177 __list_set_del(set, old);
178 }
179
180 static void
181 set_cleanup_entries(struct ip_set *set)
182 {
183 struct list_set *map = set->data;
184 struct set_elem *e, *n;
185
186 list_for_each_entry_safe(e, n, &map->members, list)
187 if (ip_set_timeout_expired(ext_timeout(e, set)))
188 list_set_del(set, e);
189 }
190
191 static int
192 list_set_utest(struct ip_set *set, void *value, const struct ip_set_ext *ext,
193 struct ip_set_ext *mext, u32 flags)
194 {
195 struct list_set *map = set->data;
196 struct set_adt_elem *d = value;
197 struct set_elem *e, *next, *prev = NULL;
198 int ret;
199
200 list_for_each_entry(e, &map->members, list) {
201 if (SET_WITH_TIMEOUT(set) &&
202 ip_set_timeout_expired(ext_timeout(e, set)))
203 continue;
204 else if (e->id != d->id) {
205 prev = e;
206 continue;
207 }
208
209 if (d->before == 0) {
210 ret = 1;
211 } else if (d->before > 0) {
212 next = list_next_entry(e, list);
213 ret = !list_is_last(&e->list, &map->members) &&
214 next->id == d->refid;
215 } else {
216 ret = prev && prev->id == d->refid;
217 }
218 return ret;
219 }
220 return 0;
221 }
222
223 static void
224 list_set_init_extensions(struct ip_set *set, const struct ip_set_ext *ext,
225 struct set_elem *e)
226 {
227 if (SET_WITH_COUNTER(set))
228 ip_set_init_counter(ext_counter(e, set), ext);
229 if (SET_WITH_COMMENT(set))
230 ip_set_init_comment(ext_comment(e, set), ext);
231 if (SET_WITH_SKBINFO(set))
232 ip_set_init_skbinfo(ext_skbinfo(e, set), ext);
233 /* Update timeout last */
234 if (SET_WITH_TIMEOUT(set))
235 ip_set_timeout_set(ext_timeout(e, set), ext->timeout);
236 }
237
238 static int
239 list_set_uadd(struct ip_set *set, void *value, const struct ip_set_ext *ext,
240 struct ip_set_ext *mext, u32 flags)
241 {
242 struct list_set *map = set->data;
243 struct set_adt_elem *d = value;
244 struct set_elem *e, *n, *prev, *next;
245 bool flag_exist = flags & IPSET_FLAG_EXIST;
246
247 if (SET_WITH_TIMEOUT(set))
248 set_cleanup_entries(set);
249
250 /* Find where to add the new entry */
251 n = prev = next = NULL;
252 list_for_each_entry(e, &map->members, list) {
253 if (SET_WITH_TIMEOUT(set) &&
254 ip_set_timeout_expired(ext_timeout(e, set)))
255 continue;
256 else if (d->id == e->id)
257 n = e;
258 else if (d->before == 0 || e->id != d->refid)
259 continue;
260 else if (d->before > 0)
261 next = e;
262 else
263 prev = e;
264 }
265 /* Re-add already existing element */
266 if (n) {
267 if ((d->before > 0 && !next) ||
268 (d->before < 0 && !prev))
269 return -IPSET_ERR_REF_EXIST;
270 if (!flag_exist)
271 return -IPSET_ERR_EXIST;
272 /* Update extensions */
273 ip_set_ext_destroy(set, n);
274 list_set_init_extensions(set, ext, n);
275
276 /* Set is already added to the list */
277 ip_set_put_byindex(map->net, d->id);
278 return 0;
279 }
280 /* Add new entry */
281 if (d->before == 0) {
282 /* Append */
283 n = list_empty(&map->members) ? NULL :
284 list_last_entry(&map->members, struct set_elem, list);
285 } else if (d->before > 0) {
286 /* Insert after next element */
287 if (!list_is_last(&next->list, &map->members))
288 n = list_next_entry(next, list);
289 } else {
290 /* Insert before prev element */
291 if (prev->list.prev != &map->members)
292 n = list_prev_entry(prev, list);
293 }
294 /* Can we replace a timed out entry? */
295 if (n &&
296 !(SET_WITH_TIMEOUT(set) &&
297 ip_set_timeout_expired(ext_timeout(n, set))))
298 n = NULL;
299
300 e = kzalloc(set->dsize, GFP_ATOMIC);
301 if (!e)
302 return -ENOMEM;
303 e->id = d->id;
304 INIT_LIST_HEAD(&e->list);
305 list_set_init_extensions(set, ext, e);
306 if (n)
307 list_set_replace(set, e, n);
308 else if (next)
309 list_add_tail_rcu(&e->list, &next->list);
310 else if (prev)
311 list_add_rcu(&e->list, &prev->list);
312 else
313 list_add_tail_rcu(&e->list, &map->members);
314
315 return 0;
316 }
317
318 static int
319 list_set_udel(struct ip_set *set, void *value, const struct ip_set_ext *ext,
320 struct ip_set_ext *mext, u32 flags)
321 {
322 struct list_set *map = set->data;
323 struct set_adt_elem *d = value;
324 struct set_elem *e, *next, *prev = NULL;
325
326 list_for_each_entry(e, &map->members, list) {
327 if (SET_WITH_TIMEOUT(set) &&
328 ip_set_timeout_expired(ext_timeout(e, set)))
329 continue;
330 else if (e->id != d->id) {
331 prev = e;
332 continue;
333 }
334
335 if (d->before > 0) {
336 next = list_next_entry(e, list);
337 if (list_is_last(&e->list, &map->members) ||
338 next->id != d->refid)
339 return -IPSET_ERR_REF_EXIST;
340 } else if (d->before < 0) {
341 if (!prev || prev->id != d->refid)
342 return -IPSET_ERR_REF_EXIST;
343 }
344 list_set_del(set, e);
345 return 0;
346 }
347 return d->before != 0 ? -IPSET_ERR_REF_EXIST : -IPSET_ERR_EXIST;
348 }
349
350 static int
351 list_set_uadt(struct ip_set *set, struct nlattr *tb[],
352 enum ipset_adt adt, u32 *lineno, u32 flags, bool retried)
353 {
354 struct list_set *map = set->data;
355 ipset_adtfn adtfn = set->variant->adt[adt];
356 struct set_adt_elem e = { .refid = IPSET_INVALID_ID };
357 struct ip_set_ext ext = IP_SET_INIT_UEXT(set);
358 struct ip_set *s;
359 int ret = 0;
360
361 if (tb[IPSET_ATTR_LINENO])
362 *lineno = nla_get_u32(tb[IPSET_ATTR_LINENO]);
363
364 if (unlikely(!tb[IPSET_ATTR_NAME] ||
365 !ip_set_optattr_netorder(tb, IPSET_ATTR_CADT_FLAGS)))
366 return -IPSET_ERR_PROTOCOL;
367
368 ret = ip_set_get_extensions(set, tb, &ext);
369 if (ret)
370 return ret;
371 e.id = ip_set_get_byname(map->net, nla_data(tb[IPSET_ATTR_NAME]), &s);
372 if (e.id == IPSET_INVALID_ID)
373 return -IPSET_ERR_NAME;
374 /* "Loop detection" */
375 if (s->type->features & IPSET_TYPE_NAME) {
376 ret = -IPSET_ERR_LOOP;
377 goto finish;
378 }
379
380 if (tb[IPSET_ATTR_CADT_FLAGS]) {
381 u32 f = ip_set_get_h32(tb[IPSET_ATTR_CADT_FLAGS]);
382
383 e.before = f & IPSET_FLAG_BEFORE;
384 }
385
386 if (e.before && !tb[IPSET_ATTR_NAMEREF]) {
387 ret = -IPSET_ERR_BEFORE;
388 goto finish;
389 }
390
391 if (tb[IPSET_ATTR_NAMEREF]) {
392 e.refid = ip_set_get_byname(map->net,
393 nla_data(tb[IPSET_ATTR_NAMEREF]),
394 &s);
395 if (e.refid == IPSET_INVALID_ID) {
396 ret = -IPSET_ERR_NAMEREF;
397 goto finish;
398 }
399 if (!e.before)
400 e.before = -1;
401 }
402 if (adt != IPSET_TEST && SET_WITH_TIMEOUT(set))
403 set_cleanup_entries(set);
404
405 ret = adtfn(set, &e, &ext, &ext, flags);
406
407 finish:
408 if (e.refid != IPSET_INVALID_ID)
409 ip_set_put_byindex(map->net, e.refid);
410 if (adt != IPSET_ADD || ret)
411 ip_set_put_byindex(map->net, e.id);
412
413 return ip_set_eexist(ret, flags) ? 0 : ret;
414 }
415
416 static void
417 list_set_flush(struct ip_set *set)
418 {
419 struct list_set *map = set->data;
420 struct set_elem *e, *n;
421
422 list_for_each_entry_safe(e, n, &map->members, list)
423 list_set_del(set, e);
424 }
425
426 static void
427 list_set_destroy(struct ip_set *set)
428 {
429 struct list_set *map = set->data;
430 struct set_elem *e, *n;
431
432 if (SET_WITH_TIMEOUT(set))
433 del_timer_sync(&map->gc);
434 list_for_each_entry_safe(e, n, &map->members, list) {
435 list_del(&e->list);
436 ip_set_put_byindex(map->net, e->id);
437 ip_set_ext_destroy(set, e);
438 kfree(e);
439 }
440 kfree(map);
441
442 set->data = NULL;
443 }
444
445 static int
446 list_set_head(struct ip_set *set, struct sk_buff *skb)
447 {
448 const struct list_set *map = set->data;
449 struct nlattr *nested;
450 struct set_elem *e;
451 u32 n = 0;
452
453 list_for_each_entry(e, &map->members, list)
454 n++;
455
456 nested = ipset_nest_start(skb, IPSET_ATTR_DATA);
457 if (!nested)
458 goto nla_put_failure;
459 if (nla_put_net32(skb, IPSET_ATTR_SIZE, htonl(map->size)) ||
460 nla_put_net32(skb, IPSET_ATTR_REFERENCES, htonl(set->ref - 1)) ||
461 nla_put_net32(skb, IPSET_ATTR_MEMSIZE,
462 htonl(sizeof(*map) + n * set->dsize)))
463 goto nla_put_failure;
464 if (unlikely(ip_set_put_flags(skb, set)))
465 goto nla_put_failure;
466 ipset_nest_end(skb, nested);
467
468 return 0;
469 nla_put_failure:
470 return -EMSGSIZE;
471 }
472
473 static int
474 list_set_list(const struct ip_set *set,
475 struct sk_buff *skb, struct netlink_callback *cb)
476 {
477 const struct list_set *map = set->data;
478 struct nlattr *atd, *nested;
479 u32 i = 0, first = cb->args[IPSET_CB_ARG0];
480 struct set_elem *e;
481 int ret = 0;
482
483 atd = ipset_nest_start(skb, IPSET_ATTR_ADT);
484 if (!atd)
485 return -EMSGSIZE;
486 list_for_each_entry(e, &map->members, list) {
487 if (i == first)
488 break;
489 i++;
490 }
491
492 rcu_read_lock();
493 list_for_each_entry_from(e, &map->members, list) {
494 i++;
495 if (SET_WITH_TIMEOUT(set) &&
496 ip_set_timeout_expired(ext_timeout(e, set)))
497 continue;
498 nested = ipset_nest_start(skb, IPSET_ATTR_DATA);
499 if (!nested) {
500 if (i == first) {
501 nla_nest_cancel(skb, atd);
502 ret = -EMSGSIZE;
503 goto out;
504 }
505 goto nla_put_failure;
506 }
507 if (nla_put_string(skb, IPSET_ATTR_NAME,
508 ip_set_name_byindex(map->net, e->id)))
509 goto nla_put_failure;
510 if (ip_set_put_extensions(skb, set, e, true))
511 goto nla_put_failure;
512 ipset_nest_end(skb, nested);
513 }
514
515 ipset_nest_end(skb, atd);
516 /* Set listing finished */
517 cb->args[IPSET_CB_ARG0] = 0;
518 goto out;
519
520 nla_put_failure:
521 nla_nest_cancel(skb, nested);
522 if (unlikely(i == first)) {
523 cb->args[IPSET_CB_ARG0] = 0;
524 ret = -EMSGSIZE;
525 }
526 cb->args[IPSET_CB_ARG0] = i - 1;
527 ipset_nest_end(skb, atd);
528 out:
529 rcu_read_unlock();
530 return ret;
531 }
532
533 static bool
534 list_set_same_set(const struct ip_set *a, const struct ip_set *b)
535 {
536 const struct list_set *x = a->data;
537 const struct list_set *y = b->data;
538
539 return x->size == y->size &&
540 a->timeout == b->timeout &&
541 a->extensions == b->extensions;
542 }
543
544 static const struct ip_set_type_variant set_variant = {
545 .kadt = list_set_kadt,
546 .uadt = list_set_uadt,
547 .adt = {
548 [IPSET_ADD] = list_set_uadd,
549 [IPSET_DEL] = list_set_udel,
550 [IPSET_TEST] = list_set_utest,
551 },
552 .destroy = list_set_destroy,
553 .flush = list_set_flush,
554 .head = list_set_head,
555 .list = list_set_list,
556 .same_set = list_set_same_set,
557 };
558
559 static void
560 list_set_gc(unsigned long ul_set)
561 {
562 struct ip_set *set = (struct ip_set *)ul_set;
563 struct list_set *map = set->data;
564
565 spin_lock_bh(&set->lock);
566 set_cleanup_entries(set);
567 spin_unlock_bh(&set->lock);
568
569 map->gc.expires = jiffies + IPSET_GC_PERIOD(set->timeout) * HZ;
570 add_timer(&map->gc);
571 }
572
573 static void
574 list_set_gc_init(struct ip_set *set, void (*gc)(unsigned long ul_set))
575 {
576 struct list_set *map = set->data;
577
578 init_timer(&map->gc);
579 map->gc.data = (unsigned long)set;
580 map->gc.function = gc;
581 map->gc.expires = jiffies + IPSET_GC_PERIOD(set->timeout) * HZ;
582 add_timer(&map->gc);
583 }
584
585 /* Create list:set type of sets */
586
587 static bool
588 init_list_set(struct net *net, struct ip_set *set, u32 size)
589 {
590 struct list_set *map;
591
592 map = kzalloc(sizeof(*map), GFP_KERNEL);
593 if (!map)
594 return false;
595
596 map->size = size;
597 map->net = net;
598 INIT_LIST_HEAD(&map->members);
599 set->data = map;
600
601 return true;
602 }
603
604 static int
605 list_set_create(struct net *net, struct ip_set *set, struct nlattr *tb[],
606 u32 flags)
607 {
608 u32 size = IP_SET_LIST_DEFAULT_SIZE;
609
610 if (unlikely(!ip_set_optattr_netorder(tb, IPSET_ATTR_SIZE) ||
611 !ip_set_optattr_netorder(tb, IPSET_ATTR_TIMEOUT) ||
612 !ip_set_optattr_netorder(tb, IPSET_ATTR_CADT_FLAGS)))
613 return -IPSET_ERR_PROTOCOL;
614
615 if (tb[IPSET_ATTR_SIZE])
616 size = ip_set_get_h32(tb[IPSET_ATTR_SIZE]);
617 if (size < IP_SET_LIST_MIN_SIZE)
618 size = IP_SET_LIST_MIN_SIZE;
619
620 set->variant = &set_variant;
621 set->dsize = ip_set_elem_len(set, tb, sizeof(struct set_elem),
622 __alignof__(struct set_elem));
623 if (!init_list_set(net, set, size))
624 return -ENOMEM;
625 if (tb[IPSET_ATTR_TIMEOUT]) {
626 set->timeout = ip_set_timeout_uget(tb[IPSET_ATTR_TIMEOUT]);
627 list_set_gc_init(set, list_set_gc);
628 }
629 return 0;
630 }
631
632 static struct ip_set_type list_set_type __read_mostly = {
633 .name = "list:set",
634 .protocol = IPSET_PROTOCOL,
635 .features = IPSET_TYPE_NAME | IPSET_DUMP_LAST,
636 .dimension = IPSET_DIM_ONE,
637 .family = NFPROTO_UNSPEC,
638 .revision_min = IPSET_TYPE_REV_MIN,
639 .revision_max = IPSET_TYPE_REV_MAX,
640 .create = list_set_create,
641 .create_policy = {
642 [IPSET_ATTR_SIZE] = { .type = NLA_U32 },
643 [IPSET_ATTR_TIMEOUT] = { .type = NLA_U32 },
644 [IPSET_ATTR_CADT_FLAGS] = { .type = NLA_U32 },
645 },
646 .adt_policy = {
647 [IPSET_ATTR_NAME] = { .type = NLA_STRING,
648 .len = IPSET_MAXNAMELEN },
649 [IPSET_ATTR_NAMEREF] = { .type = NLA_STRING,
650 .len = IPSET_MAXNAMELEN },
651 [IPSET_ATTR_TIMEOUT] = { .type = NLA_U32 },
652 [IPSET_ATTR_LINENO] = { .type = NLA_U32 },
653 [IPSET_ATTR_CADT_FLAGS] = { .type = NLA_U32 },
654 [IPSET_ATTR_BYTES] = { .type = NLA_U64 },
655 [IPSET_ATTR_PACKETS] = { .type = NLA_U64 },
656 [IPSET_ATTR_COMMENT] = { .type = NLA_NUL_STRING,
657 .len = IPSET_MAX_COMMENT_SIZE },
658 [IPSET_ATTR_SKBMARK] = { .type = NLA_U64 },
659 [IPSET_ATTR_SKBPRIO] = { .type = NLA_U32 },
660 [IPSET_ATTR_SKBQUEUE] = { .type = NLA_U16 },
661 },
662 .me = THIS_MODULE,
663 };
664
665 static int __init
666 list_set_init(void)
667 {
668 return ip_set_type_register(&list_set_type);
669 }
670
671 static void __exit
672 list_set_fini(void)
673 {
674 rcu_barrier();
675 ip_set_type_unregister(&list_set_type);
676 }
677
678 module_init(list_set_init);
679 module_exit(list_set_fini);
680
Linux® is a registered trademark of Linus Torvalds in the United States and other countries.
TOMOYO® is a registered trademark of NTT DATA CORPORATION.