1 // SPDX-License-Identifier: GPL-2.0-only
3 * Copyright (c) 2006 Patrick McHardy <kaber@trash.net>
4 * Copyright © CC Computer Consultants GmbH, 2007 - 2008
6 * This is a replacement of the old ipt_recent module, which carried the
7 * following copyright notice:
9 * Author: Stephen Frost <sfrost@snowman.net>
10 * Copyright 2002-2003, Stephen Frost, 2.5.x port by laforge@netfilter.org
12 #define pr_fmt(fmt) KBUILD_MODNAME ": " fmt
13 #include <linux/init.h>
15 #include <linux/ipv6.h>
16 #include <linux/module.h>
17 #include <linux/moduleparam.h>
18 #include <linux/proc_fs.h>
19 #include <linux/seq_file.h>
20 #include <linux/string.h>
21 #include <linux/ctype.h>
22 #include <linux/list.h>
23 #include <linux/random.h>
24 #include <linux/jhash.h>
25 #include <linux/bitops.h>
26 #include <linux/skbuff.h>
27 #include <linux/inet.h>
28 #include <linux/slab.h>
29 #include <linux/vmalloc.h>
30 #include <net/net_namespace.h>
31 #include <net/netns/generic.h>
33 #include <linux/netfilter/x_tables.h>
34 #include <linux/netfilter/xt_recent.h>
36 MODULE_AUTHOR("Patrick McHardy <kaber@trash.net>");
37 MODULE_AUTHOR("Jan Engelhardt <jengelh@medozas.de>");
38 MODULE_DESCRIPTION("Xtables: \"recently-seen\" host matching");
39 MODULE_LICENSE("GPL");
40 MODULE_ALIAS("ipt_recent");
41 MODULE_ALIAS("ip6t_recent");
43 static unsigned int ip_list_tot __read_mostly
= 100;
44 static unsigned int ip_list_hash_size __read_mostly
;
45 static unsigned int ip_list_perms __read_mostly
= 0644;
46 static unsigned int ip_list_uid __read_mostly
;
47 static unsigned int ip_list_gid __read_mostly
;
48 module_param(ip_list_tot
, uint
, 0400);
49 module_param(ip_list_hash_size
, uint
, 0400);
50 module_param(ip_list_perms
, uint
, 0400);
51 module_param(ip_list_uid
, uint
, 0644);
52 module_param(ip_list_gid
, uint
, 0644);
53 MODULE_PARM_DESC(ip_list_tot
, "number of IPs to remember per list");
54 MODULE_PARM_DESC(ip_list_hash_size
, "size of hash table used to look up IPs");
55 MODULE_PARM_DESC(ip_list_perms
, "permissions on /proc/net/xt_recent/* files");
56 MODULE_PARM_DESC(ip_list_uid
, "default owner of /proc/net/xt_recent/* files");
57 MODULE_PARM_DESC(ip_list_gid
, "default owning group of /proc/net/xt_recent/* files");
59 /* retained for backwards compatibility */
60 static unsigned int ip_pkt_list_tot __read_mostly
;
61 module_param(ip_pkt_list_tot
, uint
, 0400);
62 MODULE_PARM_DESC(ip_pkt_list_tot
, "number of packets per IP address to remember (max. 65535)");
64 #define XT_RECENT_MAX_NSTAMPS 65536
67 struct list_head list
;
68 struct list_head lru_list
;
69 union nf_inet_addr addr
;
74 unsigned long stamps
[];
78 struct list_head list
;
79 char name
[XT_RECENT_NAME_LEN
];
80 union nf_inet_addr mask
;
83 u_int16_t nstamps_max_mask
;
84 struct list_head lru_list
;
85 struct list_head iphash
[];
89 struct list_head tables
;
91 struct proc_dir_entry
*xt_recent
;
95 static unsigned int recent_net_id __read_mostly
;
97 static inline struct recent_net
*recent_pernet(struct net
*net
)
99 return net_generic(net
, recent_net_id
);
102 static DEFINE_SPINLOCK(recent_lock
);
103 static DEFINE_MUTEX(recent_mutex
);
105 #ifdef CONFIG_PROC_FS
106 static const struct proc_ops recent_mt_proc_ops
;
109 static u_int32_t hash_rnd __read_mostly
;
111 static inline unsigned int recent_entry_hash4(const union nf_inet_addr
*addr
)
113 return jhash_1word((__force u32
)addr
->ip
, hash_rnd
) &
114 (ip_list_hash_size
- 1);
117 static inline unsigned int recent_entry_hash6(const union nf_inet_addr
*addr
)
119 return jhash2((u32
*)addr
->ip6
, ARRAY_SIZE(addr
->ip6
), hash_rnd
) &
120 (ip_list_hash_size
- 1);
123 static struct recent_entry
*
124 recent_entry_lookup(const struct recent_table
*table
,
125 const union nf_inet_addr
*addrp
, u_int16_t family
,
128 struct recent_entry
*e
;
131 if (family
== NFPROTO_IPV4
)
132 h
= recent_entry_hash4(addrp
);
134 h
= recent_entry_hash6(addrp
);
136 list_for_each_entry(e
, &table
->iphash
[h
], list
)
137 if (e
->family
== family
&&
138 memcmp(&e
->addr
, addrp
, sizeof(e
->addr
)) == 0 &&
139 (ttl
== e
->ttl
|| ttl
== 0 || e
->ttl
== 0))
144 static void recent_entry_remove(struct recent_table
*t
, struct recent_entry
*e
)
147 list_del(&e
->lru_list
);
153 * Drop entries with timestamps older then 'time'.
155 static void recent_entry_reap(struct recent_table
*t
, unsigned long time
,
156 struct recent_entry
*working
, bool update
)
158 struct recent_entry
*e
;
161 * The head of the LRU list is always the oldest entry.
163 e
= list_entry(t
->lru_list
.next
, struct recent_entry
, lru_list
);
166 * Do not reap the entry which are going to be updated.
168 if (e
== working
&& update
)
172 * The last time stamp is the most recent.
174 if (time_after(time
, e
->stamps
[e
->index
-1]))
175 recent_entry_remove(t
, e
);
178 static struct recent_entry
*
179 recent_entry_init(struct recent_table
*t
, const union nf_inet_addr
*addr
,
180 u_int16_t family
, u_int8_t ttl
)
182 struct recent_entry
*e
;
183 unsigned int nstamps_max
= t
->nstamps_max_mask
;
185 if (t
->entries
>= ip_list_tot
) {
186 e
= list_entry(t
->lru_list
.next
, struct recent_entry
, lru_list
);
187 recent_entry_remove(t
, e
);
191 e
= kmalloc(struct_size(e
, stamps
, nstamps_max
), GFP_ATOMIC
);
194 memcpy(&e
->addr
, addr
, sizeof(e
->addr
));
196 e
->stamps
[0] = jiffies
;
200 if (family
== NFPROTO_IPV4
)
201 list_add_tail(&e
->list
, &t
->iphash
[recent_entry_hash4(addr
)]);
203 list_add_tail(&e
->list
, &t
->iphash
[recent_entry_hash6(addr
)]);
204 list_add_tail(&e
->lru_list
, &t
->lru_list
);
209 static void recent_entry_update(struct recent_table
*t
, struct recent_entry
*e
)
211 e
->index
&= t
->nstamps_max_mask
;
212 e
->stamps
[e
->index
++] = jiffies
;
213 if (e
->index
> e
->nstamps
)
214 e
->nstamps
= e
->index
;
215 list_move_tail(&e
->lru_list
, &t
->lru_list
);
218 static struct recent_table
*recent_table_lookup(struct recent_net
*recent_net
,
221 struct recent_table
*t
;
223 list_for_each_entry(t
, &recent_net
->tables
, list
)
224 if (!strcmp(t
->name
, name
))
229 static void recent_table_flush(struct recent_table
*t
)
231 struct recent_entry
*e
, *next
;
234 for (i
= 0; i
< ip_list_hash_size
; i
++)
235 list_for_each_entry_safe(e
, next
, &t
->iphash
[i
], list
)
236 recent_entry_remove(t
, e
);
240 recent_mt(const struct sk_buff
*skb
, struct xt_action_param
*par
)
242 struct net
*net
= xt_net(par
);
243 struct recent_net
*recent_net
= recent_pernet(net
);
244 const struct xt_recent_mtinfo_v1
*info
= par
->matchinfo
;
245 struct recent_table
*t
;
246 struct recent_entry
*e
;
247 union nf_inet_addr addr
= {}, addr_mask
;
249 bool ret
= info
->invert
;
251 if (xt_family(par
) == NFPROTO_IPV4
) {
252 const struct iphdr
*iph
= ip_hdr(skb
);
254 if (info
->side
== XT_RECENT_DEST
)
255 addr
.ip
= iph
->daddr
;
257 addr
.ip
= iph
->saddr
;
261 const struct ipv6hdr
*iph
= ipv6_hdr(skb
);
263 if (info
->side
== XT_RECENT_DEST
)
264 memcpy(&addr
.in6
, &iph
->daddr
, sizeof(addr
.in6
));
266 memcpy(&addr
.in6
, &iph
->saddr
, sizeof(addr
.in6
));
268 ttl
= iph
->hop_limit
;
271 /* use TTL as seen before forwarding */
272 if (xt_out(par
) != NULL
&&
273 (!skb
->sk
|| !net_eq(net
, sock_net(skb
->sk
))))
276 spin_lock_bh(&recent_lock
);
277 t
= recent_table_lookup(recent_net
, info
->name
);
279 nf_inet_addr_mask(&addr
, &addr_mask
, &t
->mask
);
281 e
= recent_entry_lookup(t
, &addr_mask
, xt_family(par
),
282 (info
->check_set
& XT_RECENT_TTL
) ? ttl
: 0);
284 if (!(info
->check_set
& XT_RECENT_SET
))
286 e
= recent_entry_init(t
, &addr_mask
, xt_family(par
), ttl
);
293 if (info
->check_set
& XT_RECENT_SET
)
295 else if (info
->check_set
& XT_RECENT_REMOVE
) {
296 recent_entry_remove(t
, e
);
298 } else if (info
->check_set
& (XT_RECENT_CHECK
| XT_RECENT_UPDATE
)) {
299 unsigned long time
= jiffies
- info
->seconds
* HZ
;
300 unsigned int i
, hits
= 0;
302 for (i
= 0; i
< e
->nstamps
; i
++) {
303 if (info
->seconds
&& time_after(time
, e
->stamps
[i
]))
305 if (!info
->hit_count
|| ++hits
>= info
->hit_count
) {
311 /* info->seconds must be non-zero */
312 if (info
->check_set
& XT_RECENT_REAP
)
313 recent_entry_reap(t
, time
, e
,
314 info
->check_set
& XT_RECENT_UPDATE
&& ret
);
317 if (info
->check_set
& XT_RECENT_SET
||
318 (info
->check_set
& XT_RECENT_UPDATE
&& ret
)) {
319 recent_entry_update(t
, e
);
323 spin_unlock_bh(&recent_lock
);
327 static void recent_table_free(void *addr
)
332 static int recent_mt_check(const struct xt_mtchk_param
*par
,
333 const struct xt_recent_mtinfo_v1
*info
)
335 struct recent_net
*recent_net
= recent_pernet(par
->net
);
336 struct recent_table
*t
;
337 #ifdef CONFIG_PROC_FS
338 struct proc_dir_entry
*pde
;
342 unsigned int nstamp_mask
;
346 net_get_random_once(&hash_rnd
, sizeof(hash_rnd
));
348 if (info
->check_set
& ~XT_RECENT_VALID_FLAGS
) {
349 pr_info_ratelimited("Unsupported userspace flags (%08x)\n",
353 if (hweight8(info
->check_set
&
354 (XT_RECENT_SET
| XT_RECENT_REMOVE
|
355 XT_RECENT_CHECK
| XT_RECENT_UPDATE
)) != 1)
357 if ((info
->check_set
& (XT_RECENT_SET
| XT_RECENT_REMOVE
)) &&
358 (info
->seconds
|| info
->hit_count
||
359 (info
->check_set
& XT_RECENT_MODIFIERS
)))
361 if ((info
->check_set
& XT_RECENT_REAP
) && !info
->seconds
)
363 if (info
->hit_count
>= XT_RECENT_MAX_NSTAMPS
) {
364 pr_info_ratelimited("hitcount (%u) is larger than allowed maximum (%u)\n",
365 info
->hit_count
, XT_RECENT_MAX_NSTAMPS
- 1);
368 ret
= xt_check_proc_name(info
->name
, sizeof(info
->name
));
372 if (ip_pkt_list_tot
&& info
->hit_count
< ip_pkt_list_tot
)
373 nstamp_mask
= roundup_pow_of_two(ip_pkt_list_tot
) - 1;
374 else if (info
->hit_count
)
375 nstamp_mask
= roundup_pow_of_two(info
->hit_count
) - 1;
377 nstamp_mask
= 32 - 1;
379 mutex_lock(&recent_mutex
);
380 t
= recent_table_lookup(recent_net
, info
->name
);
382 if (nstamp_mask
> t
->nstamps_max_mask
) {
383 spin_lock_bh(&recent_lock
);
384 recent_table_flush(t
);
385 t
->nstamps_max_mask
= nstamp_mask
;
386 spin_unlock_bh(&recent_lock
);
394 t
= kvzalloc(struct_size(t
, iphash
, ip_list_hash_size
), GFP_KERNEL
);
400 t
->nstamps_max_mask
= nstamp_mask
;
402 memcpy(&t
->mask
, &info
->mask
, sizeof(t
->mask
));
403 strcpy(t
->name
, info
->name
);
404 INIT_LIST_HEAD(&t
->lru_list
);
405 for (i
= 0; i
< ip_list_hash_size
; i
++)
406 INIT_LIST_HEAD(&t
->iphash
[i
]);
407 #ifdef CONFIG_PROC_FS
408 uid
= make_kuid(&init_user_ns
, ip_list_uid
);
409 gid
= make_kgid(&init_user_ns
, ip_list_gid
);
410 if (!uid_valid(uid
) || !gid_valid(gid
)) {
411 recent_table_free(t
);
415 pde
= proc_create_data(t
->name
, ip_list_perms
, recent_net
->xt_recent
,
416 &recent_mt_proc_ops
, t
);
418 recent_table_free(t
);
422 proc_set_user(pde
, uid
, gid
);
424 spin_lock_bh(&recent_lock
);
425 list_add_tail(&t
->list
, &recent_net
->tables
);
426 spin_unlock_bh(&recent_lock
);
429 mutex_unlock(&recent_mutex
);
433 static int recent_mt_check_v0(const struct xt_mtchk_param
*par
)
435 const struct xt_recent_mtinfo_v0
*info_v0
= par
->matchinfo
;
436 struct xt_recent_mtinfo_v1 info_v1
;
438 /* Copy revision 0 structure to revision 1 */
439 memcpy(&info_v1
, info_v0
, sizeof(struct xt_recent_mtinfo
));
440 /* Set default mask to ensure backward compatible behaviour */
441 memset(info_v1
.mask
.all
, 0xFF, sizeof(info_v1
.mask
.all
));
443 return recent_mt_check(par
, &info_v1
);
446 static int recent_mt_check_v1(const struct xt_mtchk_param
*par
)
448 return recent_mt_check(par
, par
->matchinfo
);
451 static void recent_mt_destroy(const struct xt_mtdtor_param
*par
)
453 struct recent_net
*recent_net
= recent_pernet(par
->net
);
454 const struct xt_recent_mtinfo_v1
*info
= par
->matchinfo
;
455 struct recent_table
*t
;
457 mutex_lock(&recent_mutex
);
458 t
= recent_table_lookup(recent_net
, info
->name
);
459 if (--t
->refcnt
== 0) {
460 spin_lock_bh(&recent_lock
);
462 spin_unlock_bh(&recent_lock
);
463 #ifdef CONFIG_PROC_FS
464 if (recent_net
->xt_recent
!= NULL
)
465 remove_proc_entry(t
->name
, recent_net
->xt_recent
);
467 recent_table_flush(t
);
468 recent_table_free(t
);
470 mutex_unlock(&recent_mutex
);
473 #ifdef CONFIG_PROC_FS
474 struct recent_iter_state
{
475 const struct recent_table
*table
;
479 static void *recent_seq_start(struct seq_file
*seq
, loff_t
*pos
)
480 __acquires(recent_lock
)
482 struct recent_iter_state
*st
= seq
->private;
483 const struct recent_table
*t
= st
->table
;
484 struct recent_entry
*e
;
487 spin_lock_bh(&recent_lock
);
489 for (st
->bucket
= 0; st
->bucket
< ip_list_hash_size
; st
->bucket
++)
490 list_for_each_entry(e
, &t
->iphash
[st
->bucket
], list
)
496 static void *recent_seq_next(struct seq_file
*seq
, void *v
, loff_t
*pos
)
498 struct recent_iter_state
*st
= seq
->private;
499 const struct recent_table
*t
= st
->table
;
500 const struct recent_entry
*e
= v
;
501 const struct list_head
*head
= e
->list
.next
;
504 while (head
== &t
->iphash
[st
->bucket
]) {
505 if (++st
->bucket
>= ip_list_hash_size
)
507 head
= t
->iphash
[st
->bucket
].next
;
509 return list_entry(head
, struct recent_entry
, list
);
512 static void recent_seq_stop(struct seq_file
*s
, void *v
)
513 __releases(recent_lock
)
515 spin_unlock_bh(&recent_lock
);
518 static int recent_seq_show(struct seq_file
*seq
, void *v
)
520 const struct recent_entry
*e
= v
;
521 struct recent_iter_state
*st
= seq
->private;
522 const struct recent_table
*t
= st
->table
;
525 i
= (e
->index
- 1) & t
->nstamps_max_mask
;
527 if (e
->family
== NFPROTO_IPV4
)
528 seq_printf(seq
, "src=%pI4 ttl: %u last_seen: %lu oldest_pkt: %u",
529 &e
->addr
.ip
, e
->ttl
, e
->stamps
[i
], e
->index
);
531 seq_printf(seq
, "src=%pI6 ttl: %u last_seen: %lu oldest_pkt: %u",
532 &e
->addr
.in6
, e
->ttl
, e
->stamps
[i
], e
->index
);
533 for (i
= 0; i
< e
->nstamps
; i
++)
534 seq_printf(seq
, "%s %lu", i
? "," : "", e
->stamps
[i
]);
539 static const struct seq_operations recent_seq_ops
= {
540 .start
= recent_seq_start
,
541 .next
= recent_seq_next
,
542 .stop
= recent_seq_stop
,
543 .show
= recent_seq_show
,
546 static int recent_seq_open(struct inode
*inode
, struct file
*file
)
548 struct recent_iter_state
*st
;
550 st
= __seq_open_private(file
, &recent_seq_ops
, sizeof(*st
));
554 st
->table
= pde_data(inode
);
559 recent_mt_proc_write(struct file
*file
, const char __user
*input
,
560 size_t size
, loff_t
*loff
)
562 struct recent_table
*t
= pde_data(file_inode(file
));
563 struct recent_entry
*e
;
564 char buf
[sizeof("+b335:1d35:1e55:dead:c0de:1715:255.255.255.255")];
566 union nf_inet_addr addr
= {};
572 if (size
> sizeof(buf
))
574 if (copy_from_user(buf
, input
, size
) != 0)
577 /* Strict protocol! */
581 case '/': /* flush table */
582 spin_lock_bh(&recent_lock
);
583 recent_table_flush(t
);
584 spin_unlock_bh(&recent_lock
);
586 case '-': /* remove address */
589 case '+': /* add address */
593 pr_info_ratelimited("Need \"+ip\", \"-ip\" or \"/\"\n");
599 if (strnchr(c
, size
, ':') != NULL
) {
600 family
= NFPROTO_IPV6
;
601 succ
= in6_pton(c
, size
, (void *)&addr
, '\n', NULL
);
603 family
= NFPROTO_IPV4
;
604 succ
= in4_pton(c
, size
, (void *)&addr
, '\n', NULL
);
610 spin_lock_bh(&recent_lock
);
611 e
= recent_entry_lookup(t
, &addr
, family
, 0);
614 recent_entry_init(t
, &addr
, family
, 0);
617 recent_entry_update(t
, e
);
619 recent_entry_remove(t
, e
);
621 spin_unlock_bh(&recent_lock
);
622 /* Note we removed one above */
627 static const struct proc_ops recent_mt_proc_ops
= {
628 .proc_open
= recent_seq_open
,
629 .proc_read
= seq_read
,
630 .proc_write
= recent_mt_proc_write
,
631 .proc_release
= seq_release_private
,
632 .proc_lseek
= seq_lseek
,
635 static int __net_init
recent_proc_net_init(struct net
*net
)
637 struct recent_net
*recent_net
= recent_pernet(net
);
639 recent_net
->xt_recent
= proc_mkdir("xt_recent", net
->proc_net
);
640 if (!recent_net
->xt_recent
)
645 static void __net_exit
recent_proc_net_exit(struct net
*net
)
647 struct recent_net
*recent_net
= recent_pernet(net
);
648 struct recent_table
*t
;
650 /* recent_net_exit() is called before recent_mt_destroy(). Make sure
651 * that the parent xt_recent proc entry is empty before trying to
654 spin_lock_bh(&recent_lock
);
655 list_for_each_entry(t
, &recent_net
->tables
, list
)
656 remove_proc_entry(t
->name
, recent_net
->xt_recent
);
658 recent_net
->xt_recent
= NULL
;
659 spin_unlock_bh(&recent_lock
);
661 remove_proc_entry("xt_recent", net
->proc_net
);
664 static inline int recent_proc_net_init(struct net
*net
)
669 static inline void recent_proc_net_exit(struct net
*net
)
672 #endif /* CONFIG_PROC_FS */
674 static int __net_init
recent_net_init(struct net
*net
)
676 struct recent_net
*recent_net
= recent_pernet(net
);
678 INIT_LIST_HEAD(&recent_net
->tables
);
679 return recent_proc_net_init(net
);
682 static void __net_exit
recent_net_exit(struct net
*net
)
684 recent_proc_net_exit(net
);
687 static struct pernet_operations recent_net_ops
= {
688 .init
= recent_net_init
,
689 .exit
= recent_net_exit
,
690 .id
= &recent_net_id
,
691 .size
= sizeof(struct recent_net
),
694 static struct xt_match recent_mt_reg
[] __read_mostly
= {
698 .family
= NFPROTO_IPV4
,
700 .matchsize
= sizeof(struct xt_recent_mtinfo
),
701 .checkentry
= recent_mt_check_v0
,
702 .destroy
= recent_mt_destroy
,
708 .family
= NFPROTO_IPV6
,
710 .matchsize
= sizeof(struct xt_recent_mtinfo
),
711 .checkentry
= recent_mt_check_v0
,
712 .destroy
= recent_mt_destroy
,
718 .family
= NFPROTO_IPV4
,
720 .matchsize
= sizeof(struct xt_recent_mtinfo_v1
),
721 .checkentry
= recent_mt_check_v1
,
722 .destroy
= recent_mt_destroy
,
728 .family
= NFPROTO_IPV6
,
730 .matchsize
= sizeof(struct xt_recent_mtinfo_v1
),
731 .checkentry
= recent_mt_check_v1
,
732 .destroy
= recent_mt_destroy
,
737 static int __init
recent_mt_init(void)
741 BUILD_BUG_ON_NOT_POWER_OF_2(XT_RECENT_MAX_NSTAMPS
);
743 if (!ip_list_tot
|| ip_pkt_list_tot
>= XT_RECENT_MAX_NSTAMPS
)
745 ip_list_hash_size
= 1 << fls(ip_list_tot
);
747 err
= register_pernet_subsys(&recent_net_ops
);
750 err
= xt_register_matches(recent_mt_reg
, ARRAY_SIZE(recent_mt_reg
));
752 unregister_pernet_subsys(&recent_net_ops
);
756 static void __exit
recent_mt_exit(void)
758 xt_unregister_matches(recent_mt_reg
, ARRAY_SIZE(recent_mt_reg
));
759 unregister_pernet_subsys(&recent_net_ops
);
762 module_init(recent_mt_init
);
763 module_exit(recent_mt_exit
);