Merge branch 'for-linus' of git://openrisc.net/~jonas/linux
[linux-2.6/kvm.git] / net / netfilter / xt_recent.c
blobd2ff15a2412b3335b016e2598d10cb9e2d31eabe
1 /*
2 * Copyright (c) 2006 Patrick McHardy <kaber@trash.net>
3 * Copyright © CC Computer Consultants GmbH, 2007 - 2008
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.
9 * This is a replacement of the old ipt_recent module, which carried the
10 * following copyright notice:
12 * Author: Stephen Frost <sfrost@snowman.net>
13 * Copyright 2002-2003, Stephen Frost, 2.5.x port by laforge@netfilter.org
15 #define pr_fmt(fmt) KBUILD_MODNAME ": " fmt
16 #include <linux/init.h>
17 #include <linux/ip.h>
18 #include <linux/ipv6.h>
19 #include <linux/module.h>
20 #include <linux/moduleparam.h>
21 #include <linux/proc_fs.h>
22 #include <linux/seq_file.h>
23 #include <linux/string.h>
24 #include <linux/ctype.h>
25 #include <linux/list.h>
26 #include <linux/random.h>
27 #include <linux/jhash.h>
28 #include <linux/bitops.h>
29 #include <linux/skbuff.h>
30 #include <linux/inet.h>
31 #include <linux/slab.h>
32 #include <net/net_namespace.h>
33 #include <net/netns/generic.h>
35 #include <linux/netfilter/x_tables.h>
36 #include <linux/netfilter/xt_recent.h>
38 MODULE_AUTHOR("Patrick McHardy <kaber@trash.net>");
39 MODULE_AUTHOR("Jan Engelhardt <jengelh@medozas.de>");
40 MODULE_DESCRIPTION("Xtables: \"recently-seen\" host matching");
41 MODULE_LICENSE("GPL");
42 MODULE_ALIAS("ipt_recent");
43 MODULE_ALIAS("ip6t_recent");
45 static unsigned int ip_list_tot = 100;
46 static unsigned int ip_pkt_list_tot = 20;
47 static unsigned int ip_list_hash_size = 0;
48 static unsigned int ip_list_perms = 0644;
49 static unsigned int ip_list_uid = 0;
50 static unsigned int ip_list_gid = 0;
51 module_param(ip_list_tot, uint, 0400);
52 module_param(ip_pkt_list_tot, uint, 0400);
53 module_param(ip_list_hash_size, uint, 0400);
54 module_param(ip_list_perms, uint, 0400);
55 module_param(ip_list_uid, uint, S_IRUGO | S_IWUSR);
56 module_param(ip_list_gid, uint, S_IRUGO | S_IWUSR);
57 MODULE_PARM_DESC(ip_list_tot, "number of IPs to remember per list");
58 MODULE_PARM_DESC(ip_pkt_list_tot, "number of packets per IP address to remember (max. 255)");
59 MODULE_PARM_DESC(ip_list_hash_size, "size of hash table used to look up IPs");
60 MODULE_PARM_DESC(ip_list_perms, "permissions on /proc/net/xt_recent/* files");
61 MODULE_PARM_DESC(ip_list_uid, "default owner of /proc/net/xt_recent/* files");
62 MODULE_PARM_DESC(ip_list_gid, "default owning group of /proc/net/xt_recent/* files");
64 struct recent_entry {
65 struct list_head list;
66 struct list_head lru_list;
67 union nf_inet_addr addr;
68 u_int16_t family;
69 u_int8_t ttl;
70 u_int8_t index;
71 u_int16_t nstamps;
72 unsigned long stamps[0];
75 struct recent_table {
76 struct list_head list;
77 char name[XT_RECENT_NAME_LEN];
78 unsigned int refcnt;
79 unsigned int entries;
80 struct list_head lru_list;
81 struct list_head iphash[0];
84 struct recent_net {
85 struct list_head tables;
86 #ifdef CONFIG_PROC_FS
87 struct proc_dir_entry *xt_recent;
88 #endif
91 static int recent_net_id;
92 static inline struct recent_net *recent_pernet(struct net *net)
94 return net_generic(net, recent_net_id);
97 static DEFINE_SPINLOCK(recent_lock);
98 static DEFINE_MUTEX(recent_mutex);
100 #ifdef CONFIG_PROC_FS
101 static const struct file_operations recent_old_fops, recent_mt_fops;
102 #endif
104 static u_int32_t hash_rnd __read_mostly;
105 static bool hash_rnd_inited __read_mostly;
107 static inline unsigned int recent_entry_hash4(const union nf_inet_addr *addr)
109 return jhash_1word((__force u32)addr->ip, hash_rnd) &
110 (ip_list_hash_size - 1);
113 static inline unsigned int recent_entry_hash6(const union nf_inet_addr *addr)
115 return jhash2((u32 *)addr->ip6, ARRAY_SIZE(addr->ip6), hash_rnd) &
116 (ip_list_hash_size - 1);
119 static struct recent_entry *
120 recent_entry_lookup(const struct recent_table *table,
121 const union nf_inet_addr *addrp, u_int16_t family,
122 u_int8_t ttl)
124 struct recent_entry *e;
125 unsigned int h;
127 if (family == NFPROTO_IPV4)
128 h = recent_entry_hash4(addrp);
129 else
130 h = recent_entry_hash6(addrp);
132 list_for_each_entry(e, &table->iphash[h], list)
133 if (e->family == family &&
134 memcmp(&e->addr, addrp, sizeof(e->addr)) == 0 &&
135 (ttl == e->ttl || ttl == 0 || e->ttl == 0))
136 return e;
137 return NULL;
140 static void recent_entry_remove(struct recent_table *t, struct recent_entry *e)
142 list_del(&e->list);
143 list_del(&e->lru_list);
144 kfree(e);
145 t->entries--;
149 * Drop entries with timestamps older then 'time'.
151 static void recent_entry_reap(struct recent_table *t, unsigned long time)
153 struct recent_entry *e;
156 * The head of the LRU list is always the oldest entry.
158 e = list_entry(t->lru_list.next, struct recent_entry, lru_list);
161 * The last time stamp is the most recent.
163 if (time_after(time, e->stamps[e->index-1]))
164 recent_entry_remove(t, e);
167 static struct recent_entry *
168 recent_entry_init(struct recent_table *t, const union nf_inet_addr *addr,
169 u_int16_t family, u_int8_t ttl)
171 struct recent_entry *e;
173 if (t->entries >= ip_list_tot) {
174 e = list_entry(t->lru_list.next, struct recent_entry, lru_list);
175 recent_entry_remove(t, e);
177 e = kmalloc(sizeof(*e) + sizeof(e->stamps[0]) * ip_pkt_list_tot,
178 GFP_ATOMIC);
179 if (e == NULL)
180 return NULL;
181 memcpy(&e->addr, addr, sizeof(e->addr));
182 e->ttl = ttl;
183 e->stamps[0] = jiffies;
184 e->nstamps = 1;
185 e->index = 1;
186 e->family = family;
187 if (family == NFPROTO_IPV4)
188 list_add_tail(&e->list, &t->iphash[recent_entry_hash4(addr)]);
189 else
190 list_add_tail(&e->list, &t->iphash[recent_entry_hash6(addr)]);
191 list_add_tail(&e->lru_list, &t->lru_list);
192 t->entries++;
193 return e;
196 static void recent_entry_update(struct recent_table *t, struct recent_entry *e)
198 e->index %= ip_pkt_list_tot;
199 e->stamps[e->index++] = jiffies;
200 if (e->index > e->nstamps)
201 e->nstamps = e->index;
202 list_move_tail(&e->lru_list, &t->lru_list);
205 static struct recent_table *recent_table_lookup(struct recent_net *recent_net,
206 const char *name)
208 struct recent_table *t;
210 list_for_each_entry(t, &recent_net->tables, list)
211 if (!strcmp(t->name, name))
212 return t;
213 return NULL;
216 static void recent_table_flush(struct recent_table *t)
218 struct recent_entry *e, *next;
219 unsigned int i;
221 for (i = 0; i < ip_list_hash_size; i++)
222 list_for_each_entry_safe(e, next, &t->iphash[i], list)
223 recent_entry_remove(t, e);
226 static bool
227 recent_mt(const struct sk_buff *skb, struct xt_action_param *par)
229 struct net *net = dev_net(par->in ? par->in : par->out);
230 struct recent_net *recent_net = recent_pernet(net);
231 const struct xt_recent_mtinfo *info = par->matchinfo;
232 struct recent_table *t;
233 struct recent_entry *e;
234 union nf_inet_addr addr = {};
235 u_int8_t ttl;
236 bool ret = info->invert;
238 if (par->family == NFPROTO_IPV4) {
239 const struct iphdr *iph = ip_hdr(skb);
241 if (info->side == XT_RECENT_DEST)
242 addr.ip = iph->daddr;
243 else
244 addr.ip = iph->saddr;
246 ttl = iph->ttl;
247 } else {
248 const struct ipv6hdr *iph = ipv6_hdr(skb);
250 if (info->side == XT_RECENT_DEST)
251 memcpy(&addr.in6, &iph->daddr, sizeof(addr.in6));
252 else
253 memcpy(&addr.in6, &iph->saddr, sizeof(addr.in6));
255 ttl = iph->hop_limit;
258 /* use TTL as seen before forwarding */
259 if (par->out != NULL && skb->sk == NULL)
260 ttl++;
262 spin_lock_bh(&recent_lock);
263 t = recent_table_lookup(recent_net, info->name);
264 e = recent_entry_lookup(t, &addr, par->family,
265 (info->check_set & XT_RECENT_TTL) ? ttl : 0);
266 if (e == NULL) {
267 if (!(info->check_set & XT_RECENT_SET))
268 goto out;
269 e = recent_entry_init(t, &addr, par->family, ttl);
270 if (e == NULL)
271 par->hotdrop = true;
272 ret = !ret;
273 goto out;
276 if (info->check_set & XT_RECENT_SET)
277 ret = !ret;
278 else if (info->check_set & XT_RECENT_REMOVE) {
279 recent_entry_remove(t, e);
280 ret = !ret;
281 } else if (info->check_set & (XT_RECENT_CHECK | XT_RECENT_UPDATE)) {
282 unsigned long time = jiffies - info->seconds * HZ;
283 unsigned int i, hits = 0;
285 for (i = 0; i < e->nstamps; i++) {
286 if (info->seconds && time_after(time, e->stamps[i]))
287 continue;
288 if (!info->hit_count || ++hits >= info->hit_count) {
289 ret = !ret;
290 break;
294 /* info->seconds must be non-zero */
295 if (info->check_set & XT_RECENT_REAP)
296 recent_entry_reap(t, time);
299 if (info->check_set & XT_RECENT_SET ||
300 (info->check_set & XT_RECENT_UPDATE && ret)) {
301 recent_entry_update(t, e);
302 e->ttl = ttl;
304 out:
305 spin_unlock_bh(&recent_lock);
306 return ret;
309 static int recent_mt_check(const struct xt_mtchk_param *par)
311 struct recent_net *recent_net = recent_pernet(par->net);
312 const struct xt_recent_mtinfo *info = par->matchinfo;
313 struct recent_table *t;
314 #ifdef CONFIG_PROC_FS
315 struct proc_dir_entry *pde;
316 #endif
317 unsigned i;
318 int ret = -EINVAL;
320 if (unlikely(!hash_rnd_inited)) {
321 get_random_bytes(&hash_rnd, sizeof(hash_rnd));
322 hash_rnd_inited = true;
324 if (info->check_set & ~XT_RECENT_VALID_FLAGS) {
325 pr_info("Unsupported user space flags (%08x)\n",
326 info->check_set);
327 return -EINVAL;
329 if (hweight8(info->check_set &
330 (XT_RECENT_SET | XT_RECENT_REMOVE |
331 XT_RECENT_CHECK | XT_RECENT_UPDATE)) != 1)
332 return -EINVAL;
333 if ((info->check_set & (XT_RECENT_SET | XT_RECENT_REMOVE)) &&
334 (info->seconds || info->hit_count ||
335 (info->check_set & XT_RECENT_MODIFIERS)))
336 return -EINVAL;
337 if ((info->check_set & XT_RECENT_REAP) && !info->seconds)
338 return -EINVAL;
339 if (info->hit_count > ip_pkt_list_tot) {
340 pr_info("hitcount (%u) is larger than "
341 "packets to be remembered (%u)\n",
342 info->hit_count, ip_pkt_list_tot);
343 return -EINVAL;
345 if (info->name[0] == '\0' ||
346 strnlen(info->name, XT_RECENT_NAME_LEN) == XT_RECENT_NAME_LEN)
347 return -EINVAL;
349 mutex_lock(&recent_mutex);
350 t = recent_table_lookup(recent_net, info->name);
351 if (t != NULL) {
352 t->refcnt++;
353 ret = 0;
354 goto out;
357 t = kzalloc(sizeof(*t) + sizeof(t->iphash[0]) * ip_list_hash_size,
358 GFP_KERNEL);
359 if (t == NULL) {
360 ret = -ENOMEM;
361 goto out;
363 t->refcnt = 1;
364 strcpy(t->name, info->name);
365 INIT_LIST_HEAD(&t->lru_list);
366 for (i = 0; i < ip_list_hash_size; i++)
367 INIT_LIST_HEAD(&t->iphash[i]);
368 #ifdef CONFIG_PROC_FS
369 pde = proc_create_data(t->name, ip_list_perms, recent_net->xt_recent,
370 &recent_mt_fops, t);
371 if (pde == NULL) {
372 kfree(t);
373 ret = -ENOMEM;
374 goto out;
376 pde->uid = ip_list_uid;
377 pde->gid = ip_list_gid;
378 #endif
379 spin_lock_bh(&recent_lock);
380 list_add_tail(&t->list, &recent_net->tables);
381 spin_unlock_bh(&recent_lock);
382 ret = 0;
383 out:
384 mutex_unlock(&recent_mutex);
385 return ret;
388 static void recent_mt_destroy(const struct xt_mtdtor_param *par)
390 struct recent_net *recent_net = recent_pernet(par->net);
391 const struct xt_recent_mtinfo *info = par->matchinfo;
392 struct recent_table *t;
394 mutex_lock(&recent_mutex);
395 t = recent_table_lookup(recent_net, info->name);
396 if (--t->refcnt == 0) {
397 spin_lock_bh(&recent_lock);
398 list_del(&t->list);
399 spin_unlock_bh(&recent_lock);
400 #ifdef CONFIG_PROC_FS
401 remove_proc_entry(t->name, recent_net->xt_recent);
402 #endif
403 recent_table_flush(t);
404 kfree(t);
406 mutex_unlock(&recent_mutex);
409 #ifdef CONFIG_PROC_FS
410 struct recent_iter_state {
411 const struct recent_table *table;
412 unsigned int bucket;
415 static void *recent_seq_start(struct seq_file *seq, loff_t *pos)
416 __acquires(recent_lock)
418 struct recent_iter_state *st = seq->private;
419 const struct recent_table *t = st->table;
420 struct recent_entry *e;
421 loff_t p = *pos;
423 spin_lock_bh(&recent_lock);
425 for (st->bucket = 0; st->bucket < ip_list_hash_size; st->bucket++)
426 list_for_each_entry(e, &t->iphash[st->bucket], list)
427 if (p-- == 0)
428 return e;
429 return NULL;
432 static void *recent_seq_next(struct seq_file *seq, void *v, loff_t *pos)
434 struct recent_iter_state *st = seq->private;
435 const struct recent_table *t = st->table;
436 const struct recent_entry *e = v;
437 const struct list_head *head = e->list.next;
439 while (head == &t->iphash[st->bucket]) {
440 if (++st->bucket >= ip_list_hash_size)
441 return NULL;
442 head = t->iphash[st->bucket].next;
444 (*pos)++;
445 return list_entry(head, struct recent_entry, list);
448 static void recent_seq_stop(struct seq_file *s, void *v)
449 __releases(recent_lock)
451 spin_unlock_bh(&recent_lock);
454 static int recent_seq_show(struct seq_file *seq, void *v)
456 const struct recent_entry *e = v;
457 unsigned int i;
459 i = (e->index - 1) % ip_pkt_list_tot;
460 if (e->family == NFPROTO_IPV4)
461 seq_printf(seq, "src=%pI4 ttl: %u last_seen: %lu oldest_pkt: %u",
462 &e->addr.ip, e->ttl, e->stamps[i], e->index);
463 else
464 seq_printf(seq, "src=%pI6 ttl: %u last_seen: %lu oldest_pkt: %u",
465 &e->addr.in6, e->ttl, e->stamps[i], e->index);
466 for (i = 0; i < e->nstamps; i++)
467 seq_printf(seq, "%s %lu", i ? "," : "", e->stamps[i]);
468 seq_printf(seq, "\n");
469 return 0;
472 static const struct seq_operations recent_seq_ops = {
473 .start = recent_seq_start,
474 .next = recent_seq_next,
475 .stop = recent_seq_stop,
476 .show = recent_seq_show,
479 static int recent_seq_open(struct inode *inode, struct file *file)
481 struct proc_dir_entry *pde = PDE(inode);
482 struct recent_iter_state *st;
484 st = __seq_open_private(file, &recent_seq_ops, sizeof(*st));
485 if (st == NULL)
486 return -ENOMEM;
488 st->table = pde->data;
489 return 0;
492 static ssize_t
493 recent_mt_proc_write(struct file *file, const char __user *input,
494 size_t size, loff_t *loff)
496 const struct proc_dir_entry *pde = PDE(file->f_path.dentry->d_inode);
497 struct recent_table *t = pde->data;
498 struct recent_entry *e;
499 char buf[sizeof("+b335:1d35:1e55:dead:c0de:1715:5afe:c0de")];
500 const char *c = buf;
501 union nf_inet_addr addr = {};
502 u_int16_t family;
503 bool add, succ;
505 if (size == 0)
506 return 0;
507 if (size > sizeof(buf))
508 size = sizeof(buf);
509 if (copy_from_user(buf, input, size) != 0)
510 return -EFAULT;
512 /* Strict protocol! */
513 if (*loff != 0)
514 return -ESPIPE;
515 switch (*c) {
516 case '/': /* flush table */
517 spin_lock_bh(&recent_lock);
518 recent_table_flush(t);
519 spin_unlock_bh(&recent_lock);
520 return size;
521 case '-': /* remove address */
522 add = false;
523 break;
524 case '+': /* add address */
525 add = true;
526 break;
527 default:
528 pr_info("Need \"+ip\", \"-ip\" or \"/\"\n");
529 return -EINVAL;
532 ++c;
533 --size;
534 if (strnchr(c, size, ':') != NULL) {
535 family = NFPROTO_IPV6;
536 succ = in6_pton(c, size, (void *)&addr, '\n', NULL);
537 } else {
538 family = NFPROTO_IPV4;
539 succ = in4_pton(c, size, (void *)&addr, '\n', NULL);
542 if (!succ) {
543 pr_info("illegal address written to procfs\n");
544 return -EINVAL;
547 spin_lock_bh(&recent_lock);
548 e = recent_entry_lookup(t, &addr, family, 0);
549 if (e == NULL) {
550 if (add)
551 recent_entry_init(t, &addr, family, 0);
552 } else {
553 if (add)
554 recent_entry_update(t, e);
555 else
556 recent_entry_remove(t, e);
558 spin_unlock_bh(&recent_lock);
559 /* Note we removed one above */
560 *loff += size + 1;
561 return size + 1;
564 static const struct file_operations recent_mt_fops = {
565 .open = recent_seq_open,
566 .read = seq_read,
567 .write = recent_mt_proc_write,
568 .release = seq_release_private,
569 .owner = THIS_MODULE,
570 .llseek = seq_lseek,
573 static int __net_init recent_proc_net_init(struct net *net)
575 struct recent_net *recent_net = recent_pernet(net);
577 recent_net->xt_recent = proc_mkdir("xt_recent", net->proc_net);
578 if (!recent_net->xt_recent)
579 return -ENOMEM;
580 return 0;
583 static void __net_exit recent_proc_net_exit(struct net *net)
585 proc_net_remove(net, "xt_recent");
587 #else
588 static inline int recent_proc_net_init(struct net *net)
590 return 0;
593 static inline void recent_proc_net_exit(struct net *net)
596 #endif /* CONFIG_PROC_FS */
598 static int __net_init recent_net_init(struct net *net)
600 struct recent_net *recent_net = recent_pernet(net);
602 INIT_LIST_HEAD(&recent_net->tables);
603 return recent_proc_net_init(net);
606 static void __net_exit recent_net_exit(struct net *net)
608 struct recent_net *recent_net = recent_pernet(net);
610 BUG_ON(!list_empty(&recent_net->tables));
611 recent_proc_net_exit(net);
614 static struct pernet_operations recent_net_ops = {
615 .init = recent_net_init,
616 .exit = recent_net_exit,
617 .id = &recent_net_id,
618 .size = sizeof(struct recent_net),
621 static struct xt_match recent_mt_reg[] __read_mostly = {
623 .name = "recent",
624 .revision = 0,
625 .family = NFPROTO_IPV4,
626 .match = recent_mt,
627 .matchsize = sizeof(struct xt_recent_mtinfo),
628 .checkentry = recent_mt_check,
629 .destroy = recent_mt_destroy,
630 .me = THIS_MODULE,
633 .name = "recent",
634 .revision = 0,
635 .family = NFPROTO_IPV6,
636 .match = recent_mt,
637 .matchsize = sizeof(struct xt_recent_mtinfo),
638 .checkentry = recent_mt_check,
639 .destroy = recent_mt_destroy,
640 .me = THIS_MODULE,
644 static int __init recent_mt_init(void)
646 int err;
648 if (!ip_list_tot || !ip_pkt_list_tot || ip_pkt_list_tot > 255)
649 return -EINVAL;
650 ip_list_hash_size = 1 << fls(ip_list_tot);
652 err = register_pernet_subsys(&recent_net_ops);
653 if (err)
654 return err;
655 err = xt_register_matches(recent_mt_reg, ARRAY_SIZE(recent_mt_reg));
656 if (err)
657 unregister_pernet_subsys(&recent_net_ops);
658 return err;
661 static void __exit recent_mt_exit(void)
663 xt_unregister_matches(recent_mt_reg, ARRAY_SIZE(recent_mt_reg));
664 unregister_pernet_subsys(&recent_net_ops);
667 module_init(recent_mt_init);
668 module_exit(recent_mt_exit);