perf evsel: Cache associated event_format
[linux-2.6.git] / net / netfilter / xt_recent.c
blobae2ad1eec8d0ccc03fa862cb9a752701c5cd9493
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 union nf_inet_addr mask;
79 unsigned int refcnt;
80 unsigned int entries;
81 struct list_head lru_list;
82 struct list_head iphash[0];
85 struct recent_net {
86 struct list_head tables;
87 #ifdef CONFIG_PROC_FS
88 struct proc_dir_entry *xt_recent;
89 #endif
92 static int recent_net_id;
93 static inline struct recent_net *recent_pernet(struct net *net)
95 return net_generic(net, recent_net_id);
98 static DEFINE_SPINLOCK(recent_lock);
99 static DEFINE_MUTEX(recent_mutex);
101 #ifdef CONFIG_PROC_FS
102 static const struct file_operations recent_old_fops, recent_mt_fops;
103 #endif
105 static u_int32_t hash_rnd __read_mostly;
106 static bool hash_rnd_inited __read_mostly;
108 static inline unsigned int recent_entry_hash4(const union nf_inet_addr *addr)
110 return jhash_1word((__force u32)addr->ip, hash_rnd) &
111 (ip_list_hash_size - 1);
114 static inline unsigned int recent_entry_hash6(const union nf_inet_addr *addr)
116 return jhash2((u32 *)addr->ip6, ARRAY_SIZE(addr->ip6), hash_rnd) &
117 (ip_list_hash_size - 1);
120 static struct recent_entry *
121 recent_entry_lookup(const struct recent_table *table,
122 const union nf_inet_addr *addrp, u_int16_t family,
123 u_int8_t ttl)
125 struct recent_entry *e;
126 unsigned int h;
128 if (family == NFPROTO_IPV4)
129 h = recent_entry_hash4(addrp);
130 else
131 h = recent_entry_hash6(addrp);
133 list_for_each_entry(e, &table->iphash[h], list)
134 if (e->family == family &&
135 memcmp(&e->addr, addrp, sizeof(e->addr)) == 0 &&
136 (ttl == e->ttl || ttl == 0 || e->ttl == 0))
137 return e;
138 return NULL;
141 static void recent_entry_remove(struct recent_table *t, struct recent_entry *e)
143 list_del(&e->list);
144 list_del(&e->lru_list);
145 kfree(e);
146 t->entries--;
150 * Drop entries with timestamps older then 'time'.
152 static void recent_entry_reap(struct recent_table *t, unsigned long time)
154 struct recent_entry *e;
157 * The head of the LRU list is always the oldest entry.
159 e = list_entry(t->lru_list.next, struct recent_entry, lru_list);
162 * The last time stamp is the most recent.
164 if (time_after(time, e->stamps[e->index-1]))
165 recent_entry_remove(t, e);
168 static struct recent_entry *
169 recent_entry_init(struct recent_table *t, const union nf_inet_addr *addr,
170 u_int16_t family, u_int8_t ttl)
172 struct recent_entry *e;
174 if (t->entries >= ip_list_tot) {
175 e = list_entry(t->lru_list.next, struct recent_entry, lru_list);
176 recent_entry_remove(t, e);
178 e = kmalloc(sizeof(*e) + sizeof(e->stamps[0]) * ip_pkt_list_tot,
179 GFP_ATOMIC);
180 if (e == NULL)
181 return NULL;
182 memcpy(&e->addr, addr, sizeof(e->addr));
183 e->ttl = ttl;
184 e->stamps[0] = jiffies;
185 e->nstamps = 1;
186 e->index = 1;
187 e->family = family;
188 if (family == NFPROTO_IPV4)
189 list_add_tail(&e->list, &t->iphash[recent_entry_hash4(addr)]);
190 else
191 list_add_tail(&e->list, &t->iphash[recent_entry_hash6(addr)]);
192 list_add_tail(&e->lru_list, &t->lru_list);
193 t->entries++;
194 return e;
197 static void recent_entry_update(struct recent_table *t, struct recent_entry *e)
199 e->index %= ip_pkt_list_tot;
200 e->stamps[e->index++] = jiffies;
201 if (e->index > e->nstamps)
202 e->nstamps = e->index;
203 list_move_tail(&e->lru_list, &t->lru_list);
206 static struct recent_table *recent_table_lookup(struct recent_net *recent_net,
207 const char *name)
209 struct recent_table *t;
211 list_for_each_entry(t, &recent_net->tables, list)
212 if (!strcmp(t->name, name))
213 return t;
214 return NULL;
217 static void recent_table_flush(struct recent_table *t)
219 struct recent_entry *e, *next;
220 unsigned int i;
222 for (i = 0; i < ip_list_hash_size; i++)
223 list_for_each_entry_safe(e, next, &t->iphash[i], list)
224 recent_entry_remove(t, e);
227 static bool
228 recent_mt(const struct sk_buff *skb, struct xt_action_param *par)
230 struct net *net = dev_net(par->in ? par->in : par->out);
231 struct recent_net *recent_net = recent_pernet(net);
232 const struct xt_recent_mtinfo_v1 *info = par->matchinfo;
233 struct recent_table *t;
234 struct recent_entry *e;
235 union nf_inet_addr addr = {}, addr_mask;
236 u_int8_t ttl;
237 bool ret = info->invert;
239 if (par->family == NFPROTO_IPV4) {
240 const struct iphdr *iph = ip_hdr(skb);
242 if (info->side == XT_RECENT_DEST)
243 addr.ip = iph->daddr;
244 else
245 addr.ip = iph->saddr;
247 ttl = iph->ttl;
248 } else {
249 const struct ipv6hdr *iph = ipv6_hdr(skb);
251 if (info->side == XT_RECENT_DEST)
252 memcpy(&addr.in6, &iph->daddr, sizeof(addr.in6));
253 else
254 memcpy(&addr.in6, &iph->saddr, sizeof(addr.in6));
256 ttl = iph->hop_limit;
259 /* use TTL as seen before forwarding */
260 if (par->out != NULL && skb->sk == NULL)
261 ttl++;
263 spin_lock_bh(&recent_lock);
264 t = recent_table_lookup(recent_net, info->name);
266 nf_inet_addr_mask(&addr, &addr_mask, &t->mask);
268 e = recent_entry_lookup(t, &addr_mask, par->family,
269 (info->check_set & XT_RECENT_TTL) ? ttl : 0);
270 if (e == NULL) {
271 if (!(info->check_set & XT_RECENT_SET))
272 goto out;
273 e = recent_entry_init(t, &addr_mask, par->family, ttl);
274 if (e == NULL)
275 par->hotdrop = true;
276 ret = !ret;
277 goto out;
280 if (info->check_set & XT_RECENT_SET)
281 ret = !ret;
282 else if (info->check_set & XT_RECENT_REMOVE) {
283 recent_entry_remove(t, e);
284 ret = !ret;
285 } else if (info->check_set & (XT_RECENT_CHECK | XT_RECENT_UPDATE)) {
286 unsigned long time = jiffies - info->seconds * HZ;
287 unsigned int i, hits = 0;
289 for (i = 0; i < e->nstamps; i++) {
290 if (info->seconds && time_after(time, e->stamps[i]))
291 continue;
292 if (!info->hit_count || ++hits >= info->hit_count) {
293 ret = !ret;
294 break;
298 /* info->seconds must be non-zero */
299 if (info->check_set & XT_RECENT_REAP)
300 recent_entry_reap(t, time);
303 if (info->check_set & XT_RECENT_SET ||
304 (info->check_set & XT_RECENT_UPDATE && ret)) {
305 recent_entry_update(t, e);
306 e->ttl = ttl;
308 out:
309 spin_unlock_bh(&recent_lock);
310 return ret;
313 static int recent_mt_check(const struct xt_mtchk_param *par,
314 const struct xt_recent_mtinfo_v1 *info)
316 struct recent_net *recent_net = recent_pernet(par->net);
317 struct recent_table *t;
318 #ifdef CONFIG_PROC_FS
319 struct proc_dir_entry *pde;
320 #endif
321 unsigned int i;
322 int ret = -EINVAL;
324 if (unlikely(!hash_rnd_inited)) {
325 get_random_bytes(&hash_rnd, sizeof(hash_rnd));
326 hash_rnd_inited = true;
328 if (info->check_set & ~XT_RECENT_VALID_FLAGS) {
329 pr_info("Unsupported user space flags (%08x)\n",
330 info->check_set);
331 return -EINVAL;
333 if (hweight8(info->check_set &
334 (XT_RECENT_SET | XT_RECENT_REMOVE |
335 XT_RECENT_CHECK | XT_RECENT_UPDATE)) != 1)
336 return -EINVAL;
337 if ((info->check_set & (XT_RECENT_SET | XT_RECENT_REMOVE)) &&
338 (info->seconds || info->hit_count ||
339 (info->check_set & XT_RECENT_MODIFIERS)))
340 return -EINVAL;
341 if ((info->check_set & XT_RECENT_REAP) && !info->seconds)
342 return -EINVAL;
343 if (info->hit_count > ip_pkt_list_tot) {
344 pr_info("hitcount (%u) is larger than "
345 "packets to be remembered (%u)\n",
346 info->hit_count, ip_pkt_list_tot);
347 return -EINVAL;
349 if (info->name[0] == '\0' ||
350 strnlen(info->name, XT_RECENT_NAME_LEN) == XT_RECENT_NAME_LEN)
351 return -EINVAL;
353 mutex_lock(&recent_mutex);
354 t = recent_table_lookup(recent_net, info->name);
355 if (t != NULL) {
356 t->refcnt++;
357 ret = 0;
358 goto out;
361 t = kzalloc(sizeof(*t) + sizeof(t->iphash[0]) * ip_list_hash_size,
362 GFP_KERNEL);
363 if (t == NULL) {
364 ret = -ENOMEM;
365 goto out;
367 t->refcnt = 1;
369 memcpy(&t->mask, &info->mask, sizeof(t->mask));
370 strcpy(t->name, info->name);
371 INIT_LIST_HEAD(&t->lru_list);
372 for (i = 0; i < ip_list_hash_size; i++)
373 INIT_LIST_HEAD(&t->iphash[i]);
374 #ifdef CONFIG_PROC_FS
375 pde = proc_create_data(t->name, ip_list_perms, recent_net->xt_recent,
376 &recent_mt_fops, t);
377 if (pde == NULL) {
378 kfree(t);
379 ret = -ENOMEM;
380 goto out;
382 pde->uid = ip_list_uid;
383 pde->gid = ip_list_gid;
384 #endif
385 spin_lock_bh(&recent_lock);
386 list_add_tail(&t->list, &recent_net->tables);
387 spin_unlock_bh(&recent_lock);
388 ret = 0;
389 out:
390 mutex_unlock(&recent_mutex);
391 return ret;
394 static int recent_mt_check_v0(const struct xt_mtchk_param *par)
396 const struct xt_recent_mtinfo_v0 *info_v0 = par->matchinfo;
397 struct xt_recent_mtinfo_v1 info_v1;
399 /* Copy revision 0 structure to revision 1 */
400 memcpy(&info_v1, info_v0, sizeof(struct xt_recent_mtinfo));
401 /* Set default mask to ensure backward compatible behaviour */
402 memset(info_v1.mask.all, 0xFF, sizeof(info_v1.mask.all));
404 return recent_mt_check(par, &info_v1);
407 static int recent_mt_check_v1(const struct xt_mtchk_param *par)
409 return recent_mt_check(par, par->matchinfo);
412 static void recent_mt_destroy(const struct xt_mtdtor_param *par)
414 struct recent_net *recent_net = recent_pernet(par->net);
415 const struct xt_recent_mtinfo_v1 *info = par->matchinfo;
416 struct recent_table *t;
418 mutex_lock(&recent_mutex);
419 t = recent_table_lookup(recent_net, info->name);
420 if (--t->refcnt == 0) {
421 spin_lock_bh(&recent_lock);
422 list_del(&t->list);
423 spin_unlock_bh(&recent_lock);
424 #ifdef CONFIG_PROC_FS
425 remove_proc_entry(t->name, recent_net->xt_recent);
426 #endif
427 recent_table_flush(t);
428 kfree(t);
430 mutex_unlock(&recent_mutex);
433 #ifdef CONFIG_PROC_FS
434 struct recent_iter_state {
435 const struct recent_table *table;
436 unsigned int bucket;
439 static void *recent_seq_start(struct seq_file *seq, loff_t *pos)
440 __acquires(recent_lock)
442 struct recent_iter_state *st = seq->private;
443 const struct recent_table *t = st->table;
444 struct recent_entry *e;
445 loff_t p = *pos;
447 spin_lock_bh(&recent_lock);
449 for (st->bucket = 0; st->bucket < ip_list_hash_size; st->bucket++)
450 list_for_each_entry(e, &t->iphash[st->bucket], list)
451 if (p-- == 0)
452 return e;
453 return NULL;
456 static void *recent_seq_next(struct seq_file *seq, void *v, loff_t *pos)
458 struct recent_iter_state *st = seq->private;
459 const struct recent_table *t = st->table;
460 const struct recent_entry *e = v;
461 const struct list_head *head = e->list.next;
463 while (head == &t->iphash[st->bucket]) {
464 if (++st->bucket >= ip_list_hash_size)
465 return NULL;
466 head = t->iphash[st->bucket].next;
468 (*pos)++;
469 return list_entry(head, struct recent_entry, list);
472 static void recent_seq_stop(struct seq_file *s, void *v)
473 __releases(recent_lock)
475 spin_unlock_bh(&recent_lock);
478 static int recent_seq_show(struct seq_file *seq, void *v)
480 const struct recent_entry *e = v;
481 unsigned int i;
483 i = (e->index - 1) % ip_pkt_list_tot;
484 if (e->family == NFPROTO_IPV4)
485 seq_printf(seq, "src=%pI4 ttl: %u last_seen: %lu oldest_pkt: %u",
486 &e->addr.ip, e->ttl, e->stamps[i], e->index);
487 else
488 seq_printf(seq, "src=%pI6 ttl: %u last_seen: %lu oldest_pkt: %u",
489 &e->addr.in6, e->ttl, e->stamps[i], e->index);
490 for (i = 0; i < e->nstamps; i++)
491 seq_printf(seq, "%s %lu", i ? "," : "", e->stamps[i]);
492 seq_printf(seq, "\n");
493 return 0;
496 static const struct seq_operations recent_seq_ops = {
497 .start = recent_seq_start,
498 .next = recent_seq_next,
499 .stop = recent_seq_stop,
500 .show = recent_seq_show,
503 static int recent_seq_open(struct inode *inode, struct file *file)
505 struct proc_dir_entry *pde = PDE(inode);
506 struct recent_iter_state *st;
508 st = __seq_open_private(file, &recent_seq_ops, sizeof(*st));
509 if (st == NULL)
510 return -ENOMEM;
512 st->table = pde->data;
513 return 0;
516 static ssize_t
517 recent_mt_proc_write(struct file *file, const char __user *input,
518 size_t size, loff_t *loff)
520 const struct proc_dir_entry *pde = PDE(file->f_path.dentry->d_inode);
521 struct recent_table *t = pde->data;
522 struct recent_entry *e;
523 char buf[sizeof("+b335:1d35:1e55:dead:c0de:1715:5afe:c0de")];
524 const char *c = buf;
525 union nf_inet_addr addr = {};
526 u_int16_t family;
527 bool add, succ;
529 if (size == 0)
530 return 0;
531 if (size > sizeof(buf))
532 size = sizeof(buf);
533 if (copy_from_user(buf, input, size) != 0)
534 return -EFAULT;
536 /* Strict protocol! */
537 if (*loff != 0)
538 return -ESPIPE;
539 switch (*c) {
540 case '/': /* flush table */
541 spin_lock_bh(&recent_lock);
542 recent_table_flush(t);
543 spin_unlock_bh(&recent_lock);
544 return size;
545 case '-': /* remove address */
546 add = false;
547 break;
548 case '+': /* add address */
549 add = true;
550 break;
551 default:
552 pr_info("Need \"+ip\", \"-ip\" or \"/\"\n");
553 return -EINVAL;
556 ++c;
557 --size;
558 if (strnchr(c, size, ':') != NULL) {
559 family = NFPROTO_IPV6;
560 succ = in6_pton(c, size, (void *)&addr, '\n', NULL);
561 } else {
562 family = NFPROTO_IPV4;
563 succ = in4_pton(c, size, (void *)&addr, '\n', NULL);
566 if (!succ) {
567 pr_info("illegal address written to procfs\n");
568 return -EINVAL;
571 spin_lock_bh(&recent_lock);
572 e = recent_entry_lookup(t, &addr, family, 0);
573 if (e == NULL) {
574 if (add)
575 recent_entry_init(t, &addr, family, 0);
576 } else {
577 if (add)
578 recent_entry_update(t, e);
579 else
580 recent_entry_remove(t, e);
582 spin_unlock_bh(&recent_lock);
583 /* Note we removed one above */
584 *loff += size + 1;
585 return size + 1;
588 static const struct file_operations recent_mt_fops = {
589 .open = recent_seq_open,
590 .read = seq_read,
591 .write = recent_mt_proc_write,
592 .release = seq_release_private,
593 .owner = THIS_MODULE,
594 .llseek = seq_lseek,
597 static int __net_init recent_proc_net_init(struct net *net)
599 struct recent_net *recent_net = recent_pernet(net);
601 recent_net->xt_recent = proc_mkdir("xt_recent", net->proc_net);
602 if (!recent_net->xt_recent)
603 return -ENOMEM;
604 return 0;
607 static void __net_exit recent_proc_net_exit(struct net *net)
609 proc_net_remove(net, "xt_recent");
611 #else
612 static inline int recent_proc_net_init(struct net *net)
614 return 0;
617 static inline void recent_proc_net_exit(struct net *net)
620 #endif /* CONFIG_PROC_FS */
622 static int __net_init recent_net_init(struct net *net)
624 struct recent_net *recent_net = recent_pernet(net);
626 INIT_LIST_HEAD(&recent_net->tables);
627 return recent_proc_net_init(net);
630 static void __net_exit recent_net_exit(struct net *net)
632 struct recent_net *recent_net = recent_pernet(net);
634 BUG_ON(!list_empty(&recent_net->tables));
635 recent_proc_net_exit(net);
638 static struct pernet_operations recent_net_ops = {
639 .init = recent_net_init,
640 .exit = recent_net_exit,
641 .id = &recent_net_id,
642 .size = sizeof(struct recent_net),
645 static struct xt_match recent_mt_reg[] __read_mostly = {
647 .name = "recent",
648 .revision = 0,
649 .family = NFPROTO_IPV4,
650 .match = recent_mt,
651 .matchsize = sizeof(struct xt_recent_mtinfo),
652 .checkentry = recent_mt_check_v0,
653 .destroy = recent_mt_destroy,
654 .me = THIS_MODULE,
657 .name = "recent",
658 .revision = 0,
659 .family = NFPROTO_IPV6,
660 .match = recent_mt,
661 .matchsize = sizeof(struct xt_recent_mtinfo),
662 .checkentry = recent_mt_check_v0,
663 .destroy = recent_mt_destroy,
664 .me = THIS_MODULE,
667 .name = "recent",
668 .revision = 1,
669 .family = NFPROTO_IPV4,
670 .match = recent_mt,
671 .matchsize = sizeof(struct xt_recent_mtinfo_v1),
672 .checkentry = recent_mt_check_v1,
673 .destroy = recent_mt_destroy,
674 .me = THIS_MODULE,
677 .name = "recent",
678 .revision = 1,
679 .family = NFPROTO_IPV6,
680 .match = recent_mt,
681 .matchsize = sizeof(struct xt_recent_mtinfo_v1),
682 .checkentry = recent_mt_check_v1,
683 .destroy = recent_mt_destroy,
684 .me = THIS_MODULE,
688 static int __init recent_mt_init(void)
690 int err;
692 if (!ip_list_tot || !ip_pkt_list_tot || ip_pkt_list_tot > 255)
693 return -EINVAL;
694 ip_list_hash_size = 1 << fls(ip_list_tot);
696 err = register_pernet_subsys(&recent_net_ops);
697 if (err)
698 return err;
699 err = xt_register_matches(recent_mt_reg, ARRAY_SIZE(recent_mt_reg));
700 if (err)
701 unregister_pernet_subsys(&recent_net_ops);
702 return err;
705 static void __exit recent_mt_exit(void)
707 xt_unregister_matches(recent_mt_reg, ARRAY_SIZE(recent_mt_reg));
708 unregister_pernet_subsys(&recent_net_ops);
711 module_init(recent_mt_init);
712 module_exit(recent_mt_exit);