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>
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 <linux/vmalloc.h>
33 #include <net/net_namespace.h>
34 #include <net/netns/generic.h>
36 #include <linux/netfilter/x_tables.h>
37 #include <linux/netfilter/xt_recent.h>
39 MODULE_AUTHOR("Patrick McHardy <kaber@trash.net>");
40 MODULE_AUTHOR("Jan Engelhardt <jengelh@medozas.de>");
41 MODULE_DESCRIPTION("Xtables: \"recently-seen\" host matching");
42 MODULE_LICENSE("GPL");
43 MODULE_ALIAS("ipt_recent");
44 MODULE_ALIAS("ip6t_recent");
46 static unsigned int ip_list_tot __read_mostly
= 100;
47 static unsigned int ip_list_hash_size __read_mostly
;
48 static unsigned int ip_list_perms __read_mostly
= 0644;
49 static unsigned int ip_list_uid __read_mostly
;
50 static unsigned int ip_list_gid __read_mostly
;
51 module_param(ip_list_tot
, uint
, 0400);
52 module_param(ip_list_hash_size
, uint
, 0400);
53 module_param(ip_list_perms
, uint
, 0400);
54 module_param(ip_list_uid
, uint
, S_IRUGO
| S_IWUSR
);
55 module_param(ip_list_gid
, uint
, S_IRUGO
| S_IWUSR
);
56 MODULE_PARM_DESC(ip_list_tot
, "number of IPs to remember per list");
57 MODULE_PARM_DESC(ip_list_hash_size
, "size of hash table used to look up IPs");
58 MODULE_PARM_DESC(ip_list_perms
, "permissions on /proc/net/xt_recent/* files");
59 MODULE_PARM_DESC(ip_list_uid
, "default owner of /proc/net/xt_recent/* files");
60 MODULE_PARM_DESC(ip_list_gid
, "default owning group of /proc/net/xt_recent/* files");
62 /* retained for backwards compatibility */
63 static unsigned int ip_pkt_list_tot __read_mostly
;
64 module_param(ip_pkt_list_tot
, uint
, 0400);
65 MODULE_PARM_DESC(ip_pkt_list_tot
, "number of packets per IP address to remember (max. 255)");
67 #define XT_RECENT_MAX_NSTAMPS 256
70 struct list_head list
;
71 struct list_head lru_list
;
72 union nf_inet_addr addr
;
77 unsigned long stamps
[0];
81 struct list_head list
;
82 char name
[XT_RECENT_NAME_LEN
];
83 union nf_inet_addr mask
;
87 struct list_head lru_list
;
88 struct list_head iphash
[0];
92 struct list_head tables
;
94 struct proc_dir_entry
*xt_recent
;
98 static int recent_net_id __read_mostly
;
100 static inline struct recent_net
*recent_pernet(struct net
*net
)
102 return net_generic(net
, recent_net_id
);
105 static DEFINE_SPINLOCK(recent_lock
);
106 static DEFINE_MUTEX(recent_mutex
);
108 #ifdef CONFIG_PROC_FS
109 static const struct file_operations recent_old_fops
, recent_mt_fops
;
112 static u_int32_t hash_rnd __read_mostly
;
113 static bool hash_rnd_inited __read_mostly
;
115 static inline unsigned int recent_entry_hash4(const union nf_inet_addr
*addr
)
117 return jhash_1word((__force u32
)addr
->ip
, hash_rnd
) &
118 (ip_list_hash_size
- 1);
121 static inline unsigned int recent_entry_hash6(const union nf_inet_addr
*addr
)
123 return jhash2((u32
*)addr
->ip6
, ARRAY_SIZE(addr
->ip6
), hash_rnd
) &
124 (ip_list_hash_size
- 1);
127 static struct recent_entry
*
128 recent_entry_lookup(const struct recent_table
*table
,
129 const union nf_inet_addr
*addrp
, u_int16_t family
,
132 struct recent_entry
*e
;
135 if (family
== NFPROTO_IPV4
)
136 h
= recent_entry_hash4(addrp
);
138 h
= recent_entry_hash6(addrp
);
140 list_for_each_entry(e
, &table
->iphash
[h
], list
)
141 if (e
->family
== family
&&
142 memcmp(&e
->addr
, addrp
, sizeof(e
->addr
)) == 0 &&
143 (ttl
== e
->ttl
|| ttl
== 0 || e
->ttl
== 0))
148 static void recent_entry_remove(struct recent_table
*t
, struct recent_entry
*e
)
151 list_del(&e
->lru_list
);
157 * Drop entries with timestamps older then 'time'.
159 static void recent_entry_reap(struct recent_table
*t
, unsigned long time
)
161 struct recent_entry
*e
;
164 * The head of the LRU list is always the oldest entry.
166 e
= list_entry(t
->lru_list
.next
, struct recent_entry
, lru_list
);
169 * The last time stamp is the most recent.
171 if (time_after(time
, e
->stamps
[e
->index
-1]))
172 recent_entry_remove(t
, e
);
175 static struct recent_entry
*
176 recent_entry_init(struct recent_table
*t
, const union nf_inet_addr
*addr
,
177 u_int16_t family
, u_int8_t ttl
)
179 struct recent_entry
*e
;
180 unsigned int nstamps_max
= t
->nstamps_max_mask
;
182 if (t
->entries
>= ip_list_tot
) {
183 e
= list_entry(t
->lru_list
.next
, struct recent_entry
, lru_list
);
184 recent_entry_remove(t
, e
);
188 e
= kmalloc(sizeof(*e
) + sizeof(e
->stamps
[0]) * nstamps_max
,
192 memcpy(&e
->addr
, addr
, sizeof(e
->addr
));
194 e
->stamps
[0] = jiffies
;
198 if (family
== NFPROTO_IPV4
)
199 list_add_tail(&e
->list
, &t
->iphash
[recent_entry_hash4(addr
)]);
201 list_add_tail(&e
->list
, &t
->iphash
[recent_entry_hash6(addr
)]);
202 list_add_tail(&e
->lru_list
, &t
->lru_list
);
207 static void recent_entry_update(struct recent_table
*t
, struct recent_entry
*e
)
209 e
->index
&= t
->nstamps_max_mask
;
210 e
->stamps
[e
->index
++] = jiffies
;
211 if (e
->index
> e
->nstamps
)
212 e
->nstamps
= e
->index
;
213 list_move_tail(&e
->lru_list
, &t
->lru_list
);
216 static struct recent_table
*recent_table_lookup(struct recent_net
*recent_net
,
219 struct recent_table
*t
;
221 list_for_each_entry(t
, &recent_net
->tables
, list
)
222 if (!strcmp(t
->name
, name
))
227 static void recent_table_flush(struct recent_table
*t
)
229 struct recent_entry
*e
, *next
;
232 for (i
= 0; i
< ip_list_hash_size
; i
++)
233 list_for_each_entry_safe(e
, next
, &t
->iphash
[i
], list
)
234 recent_entry_remove(t
, e
);
238 recent_mt(const struct sk_buff
*skb
, struct xt_action_param
*par
)
240 struct net
*net
= dev_net(par
->in
? par
->in
: par
->out
);
241 struct recent_net
*recent_net
= recent_pernet(net
);
242 const struct xt_recent_mtinfo_v1
*info
= par
->matchinfo
;
243 struct recent_table
*t
;
244 struct recent_entry
*e
;
245 union nf_inet_addr addr
= {}, addr_mask
;
247 bool ret
= info
->invert
;
249 if (par
->family
== NFPROTO_IPV4
) {
250 const struct iphdr
*iph
= ip_hdr(skb
);
252 if (info
->side
== XT_RECENT_DEST
)
253 addr
.ip
= iph
->daddr
;
255 addr
.ip
= iph
->saddr
;
259 const struct ipv6hdr
*iph
= ipv6_hdr(skb
);
261 if (info
->side
== XT_RECENT_DEST
)
262 memcpy(&addr
.in6
, &iph
->daddr
, sizeof(addr
.in6
));
264 memcpy(&addr
.in6
, &iph
->saddr
, sizeof(addr
.in6
));
266 ttl
= iph
->hop_limit
;
269 /* use TTL as seen before forwarding */
270 if (par
->out
!= NULL
&& skb
->sk
== NULL
)
273 spin_lock_bh(&recent_lock
);
274 t
= recent_table_lookup(recent_net
, info
->name
);
276 nf_inet_addr_mask(&addr
, &addr_mask
, &t
->mask
);
278 e
= recent_entry_lookup(t
, &addr_mask
, par
->family
,
279 (info
->check_set
& XT_RECENT_TTL
) ? ttl
: 0);
281 if (!(info
->check_set
& XT_RECENT_SET
))
283 e
= recent_entry_init(t
, &addr_mask
, par
->family
, ttl
);
290 if (info
->check_set
& XT_RECENT_SET
)
292 else if (info
->check_set
& XT_RECENT_REMOVE
) {
293 recent_entry_remove(t
, e
);
295 } else if (info
->check_set
& (XT_RECENT_CHECK
| XT_RECENT_UPDATE
)) {
296 unsigned long time
= jiffies
- info
->seconds
* HZ
;
297 unsigned int i
, hits
= 0;
299 for (i
= 0; i
< e
->nstamps
; i
++) {
300 if (info
->seconds
&& time_after(time
, e
->stamps
[i
]))
302 if (!info
->hit_count
|| ++hits
>= info
->hit_count
) {
308 /* info->seconds must be non-zero */
309 if (info
->check_set
& XT_RECENT_REAP
)
310 recent_entry_reap(t
, time
);
313 if (info
->check_set
& XT_RECENT_SET
||
314 (info
->check_set
& XT_RECENT_UPDATE
&& ret
)) {
315 recent_entry_update(t
, e
);
319 spin_unlock_bh(&recent_lock
);
323 static void recent_table_free(void *addr
)
328 static int recent_mt_check(const struct xt_mtchk_param
*par
,
329 const struct xt_recent_mtinfo_v1
*info
)
331 struct recent_net
*recent_net
= recent_pernet(par
->net
);
332 struct recent_table
*t
;
333 #ifdef CONFIG_PROC_FS
334 struct proc_dir_entry
*pde
;
338 unsigned int nstamp_mask
;
343 if (unlikely(!hash_rnd_inited
)) {
344 get_random_bytes(&hash_rnd
, sizeof(hash_rnd
));
345 hash_rnd_inited
= true;
347 if (info
->check_set
& ~XT_RECENT_VALID_FLAGS
) {
348 pr_info("Unsupported user space flags (%08x)\n",
352 if (hweight8(info
->check_set
&
353 (XT_RECENT_SET
| XT_RECENT_REMOVE
|
354 XT_RECENT_CHECK
| XT_RECENT_UPDATE
)) != 1)
356 if ((info
->check_set
& (XT_RECENT_SET
| XT_RECENT_REMOVE
)) &&
357 (info
->seconds
|| info
->hit_count
||
358 (info
->check_set
& XT_RECENT_MODIFIERS
)))
360 if ((info
->check_set
& XT_RECENT_REAP
) && !info
->seconds
)
362 if (info
->hit_count
>= XT_RECENT_MAX_NSTAMPS
) {
363 pr_info("hitcount (%u) is larger than allowed maximum (%u)\n",
364 info
->hit_count
, XT_RECENT_MAX_NSTAMPS
- 1);
367 if (info
->name
[0] == '\0' ||
368 strnlen(info
->name
, XT_RECENT_NAME_LEN
) == XT_RECENT_NAME_LEN
)
371 if (ip_pkt_list_tot
&& info
->hit_count
< ip_pkt_list_tot
)
372 nstamp_mask
= roundup_pow_of_two(ip_pkt_list_tot
) - 1;
373 else if (info
->hit_count
)
374 nstamp_mask
= roundup_pow_of_two(info
->hit_count
) - 1;
376 nstamp_mask
= 32 - 1;
378 mutex_lock(&recent_mutex
);
379 t
= recent_table_lookup(recent_net
, info
->name
);
381 if (info
->hit_count
> t
->nstamps_max_mask
) {
382 pr_info("hitcount (%u) is larger than packets to be remembered (%u) for table %s\n",
383 info
->hit_count
, t
->nstamps_max_mask
+ 1,
394 sz
= sizeof(*t
) + sizeof(t
->iphash
[0]) * ip_list_hash_size
;
396 t
= kzalloc(sz
, GFP_KERNEL
);
404 t
->nstamps_max_mask
= nstamp_mask
;
406 memcpy(&t
->mask
, &info
->mask
, sizeof(t
->mask
));
407 strcpy(t
->name
, info
->name
);
408 INIT_LIST_HEAD(&t
->lru_list
);
409 for (i
= 0; i
< ip_list_hash_size
; i
++)
410 INIT_LIST_HEAD(&t
->iphash
[i
]);
411 #ifdef CONFIG_PROC_FS
412 uid
= make_kuid(&init_user_ns
, ip_list_uid
);
413 gid
= make_kgid(&init_user_ns
, ip_list_gid
);
414 if (!uid_valid(uid
) || !gid_valid(gid
)) {
415 recent_table_free(t
);
419 pde
= proc_create_data(t
->name
, ip_list_perms
, recent_net
->xt_recent
,
422 recent_table_free(t
);
426 proc_set_user(pde
, uid
, gid
);
428 spin_lock_bh(&recent_lock
);
429 list_add_tail(&t
->list
, &recent_net
->tables
);
430 spin_unlock_bh(&recent_lock
);
433 mutex_unlock(&recent_mutex
);
437 static int recent_mt_check_v0(const struct xt_mtchk_param
*par
)
439 const struct xt_recent_mtinfo_v0
*info_v0
= par
->matchinfo
;
440 struct xt_recent_mtinfo_v1 info_v1
;
442 /* Copy revision 0 structure to revision 1 */
443 memcpy(&info_v1
, info_v0
, sizeof(struct xt_recent_mtinfo
));
444 /* Set default mask to ensure backward compatible behaviour */
445 memset(info_v1
.mask
.all
, 0xFF, sizeof(info_v1
.mask
.all
));
447 return recent_mt_check(par
, &info_v1
);
450 static int recent_mt_check_v1(const struct xt_mtchk_param
*par
)
452 return recent_mt_check(par
, par
->matchinfo
);
455 static void recent_mt_destroy(const struct xt_mtdtor_param
*par
)
457 struct recent_net
*recent_net
= recent_pernet(par
->net
);
458 const struct xt_recent_mtinfo_v1
*info
= par
->matchinfo
;
459 struct recent_table
*t
;
461 mutex_lock(&recent_mutex
);
462 t
= recent_table_lookup(recent_net
, info
->name
);
463 if (--t
->refcnt
== 0) {
464 spin_lock_bh(&recent_lock
);
466 spin_unlock_bh(&recent_lock
);
467 #ifdef CONFIG_PROC_FS
468 if (recent_net
->xt_recent
!= NULL
)
469 remove_proc_entry(t
->name
, recent_net
->xt_recent
);
471 recent_table_flush(t
);
472 recent_table_free(t
);
474 mutex_unlock(&recent_mutex
);
477 #ifdef CONFIG_PROC_FS
478 struct recent_iter_state
{
479 const struct recent_table
*table
;
483 static void *recent_seq_start(struct seq_file
*seq
, loff_t
*pos
)
484 __acquires(recent_lock
)
486 struct recent_iter_state
*st
= seq
->private;
487 const struct recent_table
*t
= st
->table
;
488 struct recent_entry
*e
;
491 spin_lock_bh(&recent_lock
);
493 for (st
->bucket
= 0; st
->bucket
< ip_list_hash_size
; st
->bucket
++)
494 list_for_each_entry(e
, &t
->iphash
[st
->bucket
], list
)
500 static void *recent_seq_next(struct seq_file
*seq
, void *v
, loff_t
*pos
)
502 struct recent_iter_state
*st
= seq
->private;
503 const struct recent_table
*t
= st
->table
;
504 const struct recent_entry
*e
= v
;
505 const struct list_head
*head
= e
->list
.next
;
507 while (head
== &t
->iphash
[st
->bucket
]) {
508 if (++st
->bucket
>= ip_list_hash_size
)
510 head
= t
->iphash
[st
->bucket
].next
;
513 return list_entry(head
, struct recent_entry
, list
);
516 static void recent_seq_stop(struct seq_file
*s
, void *v
)
517 __releases(recent_lock
)
519 spin_unlock_bh(&recent_lock
);
522 static int recent_seq_show(struct seq_file
*seq
, void *v
)
524 const struct recent_entry
*e
= v
;
525 struct recent_iter_state
*st
= seq
->private;
526 const struct recent_table
*t
= st
->table
;
529 i
= (e
->index
- 1) & t
->nstamps_max_mask
;
531 if (e
->family
== NFPROTO_IPV4
)
532 seq_printf(seq
, "src=%pI4 ttl: %u last_seen: %lu oldest_pkt: %u",
533 &e
->addr
.ip
, e
->ttl
, e
->stamps
[i
], e
->index
);
535 seq_printf(seq
, "src=%pI6 ttl: %u last_seen: %lu oldest_pkt: %u",
536 &e
->addr
.in6
, e
->ttl
, e
->stamps
[i
], e
->index
);
537 for (i
= 0; i
< e
->nstamps
; i
++)
538 seq_printf(seq
, "%s %lu", i
? "," : "", e
->stamps
[i
]);
539 seq_printf(seq
, "\n");
543 static const struct seq_operations recent_seq_ops
= {
544 .start
= recent_seq_start
,
545 .next
= recent_seq_next
,
546 .stop
= recent_seq_stop
,
547 .show
= recent_seq_show
,
550 static int recent_seq_open(struct inode
*inode
, struct file
*file
)
552 struct recent_iter_state
*st
;
554 st
= __seq_open_private(file
, &recent_seq_ops
, sizeof(*st
));
558 st
->table
= PDE_DATA(inode
);
563 recent_mt_proc_write(struct file
*file
, const char __user
*input
,
564 size_t size
, loff_t
*loff
)
566 struct recent_table
*t
= PDE_DATA(file_inode(file
));
567 struct recent_entry
*e
;
568 char buf
[sizeof("+b335:1d35:1e55:dead:c0de:1715:5afe:c0de")];
570 union nf_inet_addr addr
= {};
576 if (size
> sizeof(buf
))
578 if (copy_from_user(buf
, input
, size
) != 0)
581 /* Strict protocol! */
585 case '/': /* flush table */
586 spin_lock_bh(&recent_lock
);
587 recent_table_flush(t
);
588 spin_unlock_bh(&recent_lock
);
590 case '-': /* remove address */
593 case '+': /* add address */
597 pr_info("Need \"+ip\", \"-ip\" or \"/\"\n");
603 if (strnchr(c
, size
, ':') != NULL
) {
604 family
= NFPROTO_IPV6
;
605 succ
= in6_pton(c
, size
, (void *)&addr
, '\n', NULL
);
607 family
= NFPROTO_IPV4
;
608 succ
= in4_pton(c
, size
, (void *)&addr
, '\n', NULL
);
612 pr_info("illegal address written to procfs\n");
616 spin_lock_bh(&recent_lock
);
617 e
= recent_entry_lookup(t
, &addr
, family
, 0);
620 recent_entry_init(t
, &addr
, family
, 0);
623 recent_entry_update(t
, e
);
625 recent_entry_remove(t
, e
);
627 spin_unlock_bh(&recent_lock
);
628 /* Note we removed one above */
633 static const struct file_operations recent_mt_fops
= {
634 .open
= recent_seq_open
,
636 .write
= recent_mt_proc_write
,
637 .release
= seq_release_private
,
638 .owner
= THIS_MODULE
,
642 static int __net_init
recent_proc_net_init(struct net
*net
)
644 struct recent_net
*recent_net
= recent_pernet(net
);
646 recent_net
->xt_recent
= proc_mkdir("xt_recent", net
->proc_net
);
647 if (!recent_net
->xt_recent
)
652 static void __net_exit
recent_proc_net_exit(struct net
*net
)
654 struct recent_net
*recent_net
= recent_pernet(net
);
655 struct recent_table
*t
;
657 /* recent_net_exit() is called before recent_mt_destroy(). Make sure
658 * that the parent xt_recent proc entry is is empty before trying to
661 spin_lock_bh(&recent_lock
);
662 list_for_each_entry(t
, &recent_net
->tables
, list
)
663 remove_proc_entry(t
->name
, recent_net
->xt_recent
);
665 recent_net
->xt_recent
= NULL
;
666 spin_unlock_bh(&recent_lock
);
668 remove_proc_entry("xt_recent", net
->proc_net
);
671 static inline int recent_proc_net_init(struct net
*net
)
676 static inline void recent_proc_net_exit(struct net
*net
)
679 #endif /* CONFIG_PROC_FS */
681 static int __net_init
recent_net_init(struct net
*net
)
683 struct recent_net
*recent_net
= recent_pernet(net
);
685 INIT_LIST_HEAD(&recent_net
->tables
);
686 return recent_proc_net_init(net
);
689 static void __net_exit
recent_net_exit(struct net
*net
)
691 recent_proc_net_exit(net
);
694 static struct pernet_operations recent_net_ops
= {
695 .init
= recent_net_init
,
696 .exit
= recent_net_exit
,
697 .id
= &recent_net_id
,
698 .size
= sizeof(struct recent_net
),
701 static struct xt_match recent_mt_reg
[] __read_mostly
= {
705 .family
= NFPROTO_IPV4
,
707 .matchsize
= sizeof(struct xt_recent_mtinfo
),
708 .checkentry
= recent_mt_check_v0
,
709 .destroy
= recent_mt_destroy
,
715 .family
= NFPROTO_IPV6
,
717 .matchsize
= sizeof(struct xt_recent_mtinfo
),
718 .checkentry
= recent_mt_check_v0
,
719 .destroy
= recent_mt_destroy
,
725 .family
= NFPROTO_IPV4
,
727 .matchsize
= sizeof(struct xt_recent_mtinfo_v1
),
728 .checkentry
= recent_mt_check_v1
,
729 .destroy
= recent_mt_destroy
,
735 .family
= NFPROTO_IPV6
,
737 .matchsize
= sizeof(struct xt_recent_mtinfo_v1
),
738 .checkentry
= recent_mt_check_v1
,
739 .destroy
= recent_mt_destroy
,
744 static int __init
recent_mt_init(void)
748 BUILD_BUG_ON_NOT_POWER_OF_2(XT_RECENT_MAX_NSTAMPS
);
750 if (!ip_list_tot
|| ip_pkt_list_tot
>= XT_RECENT_MAX_NSTAMPS
)
752 ip_list_hash_size
= 1 << fls(ip_list_tot
);
754 err
= register_pernet_subsys(&recent_net_ops
);
757 err
= xt_register_matches(recent_mt_reg
, ARRAY_SIZE(recent_mt_reg
));
759 unregister_pernet_subsys(&recent_net_ops
);
763 static void __exit
recent_mt_exit(void)
765 xt_unregister_matches(recent_mt_reg
, ARRAY_SIZE(recent_mt_reg
));
766 unregister_pernet_subsys(&recent_net_ops
);
769 module_init(recent_mt_init
);
770 module_exit(recent_mt_exit
);