2 * Packet matching code.
4 * Copyright (C) 1999 Paul `Rusty' Russell & Michael J. Neuling
5 * Copyright (C) 2000-2005 Netfilter Core Team <coreteam@netfilter.org>
7 * This program is free software; you can redistribute it and/or modify
8 * it under the terms of the GNU General Public License version 2 as
9 * published by the Free Software Foundation.
11 * 19 Jan 2002 Harald Welte <laforge@gnumonks.org>
12 * - increase module usage count as soon as we have rules inside
14 * 06 Jun 2002 Andras Kis-Szabo <kisza@sch.bme.hu>
15 * - new extension header parser code
16 * 15 Oct 2005 Harald Welte <laforge@netfilter.org>
17 * - Unification of {ip,ip6}_tables into x_tables
18 * - Removed tcp and udp code, since it's not ipv6 specific
21 #include <linux/capability.h>
23 #include <linux/skbuff.h>
24 #include <linux/kmod.h>
25 #include <linux/vmalloc.h>
26 #include <linux/netdevice.h>
27 #include <linux/module.h>
28 #include <linux/poison.h>
29 #include <linux/icmpv6.h>
31 #include <asm/uaccess.h>
32 #include <linux/mutex.h>
33 #include <linux/proc_fs.h>
34 #include <linux/cpumask.h>
36 #include <linux/netfilter_ipv6/ip6_tables.h>
37 #include <linux/netfilter/x_tables.h>
39 MODULE_LICENSE("GPL");
40 MODULE_AUTHOR("Netfilter Core Team <coreteam@netfilter.org>");
41 MODULE_DESCRIPTION("IPv6 packet filter");
43 #define IPV6_HDR_LEN (sizeof(struct ipv6hdr))
44 #define IPV6_OPTHDR_LEN (sizeof(struct ipv6_opt_hdr))
46 /*#define DEBUG_IP_FIREWALL*/
47 /*#define DEBUG_ALLOW_ALL*/ /* Useful for remote debugging */
48 /*#define DEBUG_IP_FIREWALL_USER*/
50 #ifdef DEBUG_IP_FIREWALL
51 #define dprintf(format, args...) printk(format , ## args)
53 #define dprintf(format, args...)
56 #ifdef DEBUG_IP_FIREWALL_USER
57 #define duprintf(format, args...) printk(format , ## args)
59 #define duprintf(format, args...)
62 #ifdef CONFIG_NETFILTER_DEBUG
63 #define IP_NF_ASSERT(x) \
66 printk("IP_NF_ASSERT: %s:%s:%u\n", \
67 __FUNCTION__, __FILE__, __LINE__); \
70 #define IP_NF_ASSERT(x)
74 /* All the better to debug you with... */
80 We keep a set of rules for each CPU, so we can avoid write-locking
81 them in the softirq when updating the counters and therefore
82 only need to read-lock in the softirq; doing a write_lock_bh() in user
83 context stops packets coming through and allows user context to read
84 the counters or update the rules.
86 Hence the start of any table is given by get_table() below. */
89 #define down(x) do { printk("DOWN:%u:" #x "\n", __LINE__); down(x); } while(0)
90 #define down_interruptible(x) ({ int __r; printk("DOWNi:%u:" #x "\n", __LINE__); __r = down_interruptible(x); if (__r != 0) printk("ABORT-DOWNi:%u\n", __LINE__); __r; })
91 #define up(x) do { printk("UP:%u:" #x "\n", __LINE__); up(x); } while(0)
94 /* Check for an extension */
96 ip6t_ext_hdr(u8 nexthdr
)
98 return ( (nexthdr
== IPPROTO_HOPOPTS
) ||
99 (nexthdr
== IPPROTO_ROUTING
) ||
100 (nexthdr
== IPPROTO_FRAGMENT
) ||
101 (nexthdr
== IPPROTO_ESP
) ||
102 (nexthdr
== IPPROTO_AH
) ||
103 (nexthdr
== IPPROTO_NONE
) ||
104 (nexthdr
== IPPROTO_DSTOPTS
) );
107 /* Returns whether matches rule or not. */
109 ip6_packet_match(const struct sk_buff
*skb
,
112 const struct ip6t_ip6
*ip6info
,
113 unsigned int *protoff
,
114 int *fragoff
, int *hotdrop
)
118 const struct ipv6hdr
*ipv6
= skb
->nh
.ipv6h
;
120 #define FWINV(bool,invflg) ((bool) ^ !!(ip6info->invflags & invflg))
122 if (FWINV(ipv6_masked_addr_cmp(&ipv6
->saddr
, &ip6info
->smsk
,
123 &ip6info
->src
), IP6T_INV_SRCIP
)
124 || FWINV(ipv6_masked_addr_cmp(&ipv6
->daddr
, &ip6info
->dmsk
,
125 &ip6info
->dst
), IP6T_INV_DSTIP
)) {
126 dprintf("Source or dest mismatch.\n");
128 dprintf("SRC: %u. Mask: %u. Target: %u.%s\n", ip->saddr,
129 ipinfo->smsk.s_addr, ipinfo->src.s_addr,
130 ipinfo->invflags & IP6T_INV_SRCIP ? " (INV)" : "");
131 dprintf("DST: %u. Mask: %u. Target: %u.%s\n", ip->daddr,
132 ipinfo->dmsk.s_addr, ipinfo->dst.s_addr,
133 ipinfo->invflags & IP6T_INV_DSTIP ? " (INV)" : "");*/
137 /* Look for ifname matches; this should unroll nicely. */
138 for (i
= 0, ret
= 0; i
< IFNAMSIZ
/sizeof(unsigned long); i
++) {
139 ret
|= (((const unsigned long *)indev
)[i
]
140 ^ ((const unsigned long *)ip6info
->iniface
)[i
])
141 & ((const unsigned long *)ip6info
->iniface_mask
)[i
];
144 if (FWINV(ret
!= 0, IP6T_INV_VIA_IN
)) {
145 dprintf("VIA in mismatch (%s vs %s).%s\n",
146 indev
, ip6info
->iniface
,
147 ip6info
->invflags
&IP6T_INV_VIA_IN
?" (INV)":"");
151 for (i
= 0, ret
= 0; i
< IFNAMSIZ
/sizeof(unsigned long); i
++) {
152 ret
|= (((const unsigned long *)outdev
)[i
]
153 ^ ((const unsigned long *)ip6info
->outiface
)[i
])
154 & ((const unsigned long *)ip6info
->outiface_mask
)[i
];
157 if (FWINV(ret
!= 0, IP6T_INV_VIA_OUT
)) {
158 dprintf("VIA out mismatch (%s vs %s).%s\n",
159 outdev
, ip6info
->outiface
,
160 ip6info
->invflags
&IP6T_INV_VIA_OUT
?" (INV)":"");
164 /* ... might want to do something with class and flowlabel here ... */
166 /* look for the desired protocol header */
167 if((ip6info
->flags
& IP6T_F_PROTO
)) {
169 unsigned short _frag_off
;
171 protohdr
= ipv6_find_hdr(skb
, protoff
, -1, &_frag_off
);
177 *fragoff
= _frag_off
;
179 dprintf("Packet protocol %hi ?= %s%hi.\n",
181 ip6info
->invflags
& IP6T_INV_PROTO
? "!":"",
184 if (ip6info
->proto
== protohdr
) {
185 if(ip6info
->invflags
& IP6T_INV_PROTO
) {
191 /* We need match for the '-p all', too! */
192 if ((ip6info
->proto
!= 0) &&
193 !(ip6info
->invflags
& IP6T_INV_PROTO
))
199 /* should be ip6 safe */
201 ip6_checkentry(const struct ip6t_ip6
*ipv6
)
203 if (ipv6
->flags
& ~IP6T_F_MASK
) {
204 duprintf("Unknown flag bits set: %08X\n",
205 ipv6
->flags
& ~IP6T_F_MASK
);
208 if (ipv6
->invflags
& ~IP6T_INV_MASK
) {
209 duprintf("Unknown invflag bits set: %08X\n",
210 ipv6
->invflags
& ~IP6T_INV_MASK
);
217 ip6t_error(struct sk_buff
**pskb
,
218 const struct net_device
*in
,
219 const struct net_device
*out
,
220 unsigned int hooknum
,
221 const struct xt_target
*target
,
222 const void *targinfo
)
225 printk("ip6_tables: error: `%s'\n", (char *)targinfo
);
231 int do_match(struct ip6t_entry_match
*m
,
232 const struct sk_buff
*skb
,
233 const struct net_device
*in
,
234 const struct net_device
*out
,
236 unsigned int protoff
,
239 /* Stop iteration if it doesn't match */
240 if (!m
->u
.kernel
.match
->match(skb
, in
, out
, m
->u
.kernel
.match
, m
->data
,
241 offset
, protoff
, hotdrop
))
247 static inline struct ip6t_entry
*
248 get_entry(void *base
, unsigned int offset
)
250 return (struct ip6t_entry
*)(base
+ offset
);
253 /* Returns one of the generic firewall policies, like NF_ACCEPT. */
255 ip6t_do_table(struct sk_buff
**pskb
,
257 const struct net_device
*in
,
258 const struct net_device
*out
,
259 struct xt_table
*table
)
261 static const char nulldevname
[IFNAMSIZ
] __attribute__((aligned(sizeof(long))));
263 unsigned int protoff
= 0;
265 /* Initializing verdict to NF_DROP keeps gcc happy. */
266 unsigned int verdict
= NF_DROP
;
267 const char *indev
, *outdev
;
269 struct ip6t_entry
*e
, *back
;
270 struct xt_table_info
*private;
273 indev
= in
? in
->name
: nulldevname
;
274 outdev
= out
? out
->name
: nulldevname
;
275 /* We handle fragments by dealing with the first fragment as
276 * if it was a normal packet. All other fragments are treated
277 * normally, except that they will NEVER match rules that ask
278 * things we don't know, ie. tcp syn flag or ports). If the
279 * rule is also a fragment-specific rule, non-fragments won't
282 read_lock_bh(&table
->lock
);
283 private = table
->private;
284 IP_NF_ASSERT(table
->valid_hooks
& (1 << hook
));
285 table_base
= (void *)private->entries
[smp_processor_id()];
286 e
= get_entry(table_base
, private->hook_entry
[hook
]);
288 /* For return from builtin chain */
289 back
= get_entry(table_base
, private->underflow
[hook
]);
294 if (ip6_packet_match(*pskb
, indev
, outdev
, &e
->ipv6
,
295 &protoff
, &offset
, &hotdrop
)) {
296 struct ip6t_entry_target
*t
;
298 if (IP6T_MATCH_ITERATE(e
, do_match
,
300 offset
, protoff
, &hotdrop
) != 0)
303 ADD_COUNTER(e
->counters
,
304 ntohs((*pskb
)->nh
.ipv6h
->payload_len
)
308 t
= ip6t_get_target(e
);
309 IP_NF_ASSERT(t
->u
.kernel
.target
);
310 /* Standard target? */
311 if (!t
->u
.kernel
.target
->target
) {
314 v
= ((struct ip6t_standard_target
*)t
)->verdict
;
316 /* Pop from stack? */
317 if (v
!= IP6T_RETURN
) {
318 verdict
= (unsigned)(-v
) - 1;
322 back
= get_entry(table_base
,
326 if (table_base
+ v
!= (void *)e
+ e
->next_offset
327 && !(e
->ipv6
.flags
& IP6T_F_GOTO
)) {
328 /* Save old back ptr in next entry */
329 struct ip6t_entry
*next
330 = (void *)e
+ e
->next_offset
;
332 = (void *)back
- table_base
;
333 /* set back pointer to next entry */
337 e
= get_entry(table_base
, v
);
339 /* Targets which reenter must return
341 #ifdef CONFIG_NETFILTER_DEBUG
342 ((struct ip6t_entry
*)table_base
)->comefrom
345 verdict
= t
->u
.kernel
.target
->target(pskb
,
351 #ifdef CONFIG_NETFILTER_DEBUG
352 if (((struct ip6t_entry
*)table_base
)->comefrom
354 && verdict
== IP6T_CONTINUE
) {
355 printk("Target %s reentered!\n",
356 t
->u
.kernel
.target
->name
);
359 ((struct ip6t_entry
*)table_base
)->comefrom
362 if (verdict
== IP6T_CONTINUE
)
363 e
= (void *)e
+ e
->next_offset
;
371 e
= (void *)e
+ e
->next_offset
;
375 #ifdef CONFIG_NETFILTER_DEBUG
376 ((struct ip6t_entry
*)table_base
)->comefrom
= NETFILTER_LINK_POISON
;
378 read_unlock_bh(&table
->lock
);
380 #ifdef DEBUG_ALLOW_ALL
389 /* All zeroes == unconditional rule. */
391 unconditional(const struct ip6t_ip6
*ipv6
)
395 for (i
= 0; i
< sizeof(*ipv6
); i
++)
396 if (((char *)ipv6
)[i
])
399 return (i
== sizeof(*ipv6
));
402 /* Figures out from what hook each rule can be called: returns 0 if
403 there are loops. Puts hook bitmask in comefrom. */
405 mark_source_chains(struct xt_table_info
*newinfo
,
406 unsigned int valid_hooks
, void *entry0
)
410 /* No recursion; use packet counter to save back ptrs (reset
411 to 0 as we leave), and comefrom to save source hook bitmask */
412 for (hook
= 0; hook
< NF_IP6_NUMHOOKS
; hook
++) {
413 unsigned int pos
= newinfo
->hook_entry
[hook
];
415 = (struct ip6t_entry
*)(entry0
+ pos
);
417 if (!(valid_hooks
& (1 << hook
)))
420 /* Set initial back pointer. */
421 e
->counters
.pcnt
= pos
;
424 struct ip6t_standard_target
*t
425 = (void *)ip6t_get_target(e
);
427 if (e
->comefrom
& (1 << NF_IP6_NUMHOOKS
)) {
428 printk("iptables: loop hook %u pos %u %08X.\n",
429 hook
, pos
, e
->comefrom
);
433 |= ((1 << hook
) | (1 << NF_IP6_NUMHOOKS
));
435 /* Unconditional return/END. */
436 if (e
->target_offset
== sizeof(struct ip6t_entry
)
437 && (strcmp(t
->target
.u
.user
.name
,
438 IP6T_STANDARD_TARGET
) == 0)
440 && unconditional(&e
->ipv6
)) {
441 unsigned int oldpos
, size
;
443 /* Return: backtrack through the last
446 e
->comefrom
^= (1<<NF_IP6_NUMHOOKS
);
447 #ifdef DEBUG_IP_FIREWALL_USER
449 & (1 << NF_IP6_NUMHOOKS
)) {
450 duprintf("Back unset "
457 pos
= e
->counters
.pcnt
;
458 e
->counters
.pcnt
= 0;
460 /* We're at the start. */
464 e
= (struct ip6t_entry
*)
466 } while (oldpos
== pos
+ e
->next_offset
);
469 size
= e
->next_offset
;
470 e
= (struct ip6t_entry
*)
471 (entry0
+ pos
+ size
);
472 e
->counters
.pcnt
= pos
;
475 int newpos
= t
->verdict
;
477 if (strcmp(t
->target
.u
.user
.name
,
478 IP6T_STANDARD_TARGET
) == 0
480 /* This a jump; chase it. */
481 duprintf("Jump rule %u -> %u\n",
484 /* ... this is a fallthru */
485 newpos
= pos
+ e
->next_offset
;
487 e
= (struct ip6t_entry
*)
489 e
->counters
.pcnt
= pos
;
494 duprintf("Finished chain %u\n", hook
);
500 cleanup_match(struct ip6t_entry_match
*m
, unsigned int *i
)
502 if (i
&& (*i
)-- == 0)
505 if (m
->u
.kernel
.match
->destroy
)
506 m
->u
.kernel
.match
->destroy(m
->u
.kernel
.match
, m
->data
);
507 module_put(m
->u
.kernel
.match
->me
);
512 standard_check(const struct ip6t_entry_target
*t
,
513 unsigned int max_offset
)
515 struct ip6t_standard_target
*targ
= (void *)t
;
517 /* Check standard info. */
518 if (targ
->verdict
>= 0
519 && targ
->verdict
> max_offset
- sizeof(struct ip6t_entry
)) {
520 duprintf("ip6t_standard_check: bad verdict (%i)\n",
524 if (targ
->verdict
< -NF_MAX_VERDICT
- 1) {
525 duprintf("ip6t_standard_check: bad negative verdict (%i)\n",
533 check_match(struct ip6t_entry_match
*m
,
535 const struct ip6t_ip6
*ipv6
,
536 unsigned int hookmask
,
539 struct ip6t_match
*match
;
542 match
= try_then_request_module(xt_find_match(AF_INET6
, m
->u
.user
.name
,
544 "ip6t_%s", m
->u
.user
.name
);
545 if (IS_ERR(match
) || !match
) {
546 duprintf("check_match: `%s' not found\n", m
->u
.user
.name
);
547 return match
? PTR_ERR(match
) : -ENOENT
;
549 m
->u
.kernel
.match
= match
;
551 ret
= xt_check_match(match
, AF_INET6
, m
->u
.match_size
- sizeof(*m
),
552 name
, hookmask
, ipv6
->proto
,
553 ipv6
->invflags
& IP6T_INV_PROTO
);
557 if (m
->u
.kernel
.match
->checkentry
558 && !m
->u
.kernel
.match
->checkentry(name
, ipv6
, match
, m
->data
,
560 duprintf("ip_tables: check failed for `%s'.\n",
561 m
->u
.kernel
.match
->name
);
569 module_put(m
->u
.kernel
.match
->me
);
573 static struct ip6t_target ip6t_standard_target
;
576 check_entry(struct ip6t_entry
*e
, const char *name
, unsigned int size
,
579 struct ip6t_entry_target
*t
;
580 struct ip6t_target
*target
;
584 if (!ip6_checkentry(&e
->ipv6
)) {
585 duprintf("ip_tables: ip check failed %p %s.\n", e
, name
);
589 if (e
->target_offset
+ sizeof(struct ip6t_entry_target
) >
594 ret
= IP6T_MATCH_ITERATE(e
, check_match
, name
, &e
->ipv6
, e
->comefrom
, &j
);
596 goto cleanup_matches
;
598 t
= ip6t_get_target(e
);
600 if (e
->target_offset
+ t
->u
.target_size
> e
->next_offset
)
601 goto cleanup_matches
;
602 target
= try_then_request_module(xt_find_target(AF_INET6
,
605 "ip6t_%s", t
->u
.user
.name
);
606 if (IS_ERR(target
) || !target
) {
607 duprintf("check_entry: `%s' not found\n", t
->u
.user
.name
);
608 ret
= target
? PTR_ERR(target
) : -ENOENT
;
609 goto cleanup_matches
;
611 t
->u
.kernel
.target
= target
;
613 ret
= xt_check_target(target
, AF_INET6
, t
->u
.target_size
- sizeof(*t
),
614 name
, e
->comefrom
, e
->ipv6
.proto
,
615 e
->ipv6
.invflags
& IP6T_INV_PROTO
);
619 if (t
->u
.kernel
.target
== &ip6t_standard_target
) {
620 if (!standard_check(t
, size
)) {
624 } else if (t
->u
.kernel
.target
->checkentry
625 && !t
->u
.kernel
.target
->checkentry(name
, e
, target
, t
->data
,
627 duprintf("ip_tables: check failed for `%s'.\n",
628 t
->u
.kernel
.target
->name
);
636 module_put(t
->u
.kernel
.target
->me
);
638 IP6T_MATCH_ITERATE(e
, cleanup_match
, &j
);
643 check_entry_size_and_hooks(struct ip6t_entry
*e
,
644 struct xt_table_info
*newinfo
,
646 unsigned char *limit
,
647 const unsigned int *hook_entries
,
648 const unsigned int *underflows
,
653 if ((unsigned long)e
% __alignof__(struct ip6t_entry
) != 0
654 || (unsigned char *)e
+ sizeof(struct ip6t_entry
) >= limit
) {
655 duprintf("Bad offset %p\n", e
);
660 < sizeof(struct ip6t_entry
) + sizeof(struct ip6t_entry_target
)) {
661 duprintf("checking: element %p size %u\n",
666 /* Check hooks & underflows */
667 for (h
= 0; h
< NF_IP6_NUMHOOKS
; h
++) {
668 if ((unsigned char *)e
- base
== hook_entries
[h
])
669 newinfo
->hook_entry
[h
] = hook_entries
[h
];
670 if ((unsigned char *)e
- base
== underflows
[h
])
671 newinfo
->underflow
[h
] = underflows
[h
];
674 /* FIXME: underflows must be unconditional, standard verdicts
675 < 0 (not IP6T_RETURN). --RR */
677 /* Clear counters and comefrom */
678 e
->counters
= ((struct xt_counters
) { 0, 0 });
686 cleanup_entry(struct ip6t_entry
*e
, unsigned int *i
)
688 struct ip6t_entry_target
*t
;
690 if (i
&& (*i
)-- == 0)
693 /* Cleanup all matches */
694 IP6T_MATCH_ITERATE(e
, cleanup_match
, NULL
);
695 t
= ip6t_get_target(e
);
696 if (t
->u
.kernel
.target
->destroy
)
697 t
->u
.kernel
.target
->destroy(t
->u
.kernel
.target
, t
->data
);
698 module_put(t
->u
.kernel
.target
->me
);
702 /* Checks and translates the user-supplied table segment (held in
705 translate_table(const char *name
,
706 unsigned int valid_hooks
,
707 struct xt_table_info
*newinfo
,
711 const unsigned int *hook_entries
,
712 const unsigned int *underflows
)
717 newinfo
->size
= size
;
718 newinfo
->number
= number
;
720 /* Init all hooks to impossible value. */
721 for (i
= 0; i
< NF_IP6_NUMHOOKS
; i
++) {
722 newinfo
->hook_entry
[i
] = 0xFFFFFFFF;
723 newinfo
->underflow
[i
] = 0xFFFFFFFF;
726 duprintf("translate_table: size %u\n", newinfo
->size
);
728 /* Walk through entries, checking offsets. */
729 ret
= IP6T_ENTRY_ITERATE(entry0
, newinfo
->size
,
730 check_entry_size_and_hooks
,
734 hook_entries
, underflows
, &i
);
739 duprintf("translate_table: %u not %u entries\n",
744 /* Check hooks all assigned */
745 for (i
= 0; i
< NF_IP6_NUMHOOKS
; i
++) {
746 /* Only hooks which are valid */
747 if (!(valid_hooks
& (1 << i
)))
749 if (newinfo
->hook_entry
[i
] == 0xFFFFFFFF) {
750 duprintf("Invalid hook entry %u %u\n",
754 if (newinfo
->underflow
[i
] == 0xFFFFFFFF) {
755 duprintf("Invalid underflow %u %u\n",
761 /* Finally, each sanity check must pass */
763 ret
= IP6T_ENTRY_ITERATE(entry0
, newinfo
->size
,
764 check_entry
, name
, size
, &i
);
770 if (!mark_source_chains(newinfo
, valid_hooks
, entry0
))
773 /* And one copy for every other CPU */
774 for_each_possible_cpu(i
) {
775 if (newinfo
->entries
[i
] && newinfo
->entries
[i
] != entry0
)
776 memcpy(newinfo
->entries
[i
], entry0
, newinfo
->size
);
781 IP6T_ENTRY_ITERATE(entry0
, newinfo
->size
, cleanup_entry
, &i
);
787 add_entry_to_counter(const struct ip6t_entry
*e
,
788 struct xt_counters total
[],
791 ADD_COUNTER(total
[*i
], e
->counters
.bcnt
, e
->counters
.pcnt
);
798 set_entry_to_counter(const struct ip6t_entry
*e
,
799 struct ip6t_counters total
[],
802 SET_COUNTER(total
[*i
], e
->counters
.bcnt
, e
->counters
.pcnt
);
809 get_counters(const struct xt_table_info
*t
,
810 struct xt_counters counters
[])
816 /* Instead of clearing (by a previous call to memset())
817 * the counters and using adds, we set the counters
818 * with data used by 'current' CPU
819 * We dont care about preemption here.
821 curcpu
= raw_smp_processor_id();
824 IP6T_ENTRY_ITERATE(t
->entries
[curcpu
],
826 set_entry_to_counter
,
830 for_each_possible_cpu(cpu
) {
834 IP6T_ENTRY_ITERATE(t
->entries
[cpu
],
836 add_entry_to_counter
,
843 copy_entries_to_user(unsigned int total_size
,
844 struct xt_table
*table
,
845 void __user
*userptr
)
847 unsigned int off
, num
, countersize
;
848 struct ip6t_entry
*e
;
849 struct xt_counters
*counters
;
850 struct xt_table_info
*private = table
->private;
854 /* We need atomic snapshot of counters: rest doesn't change
855 (other than comefrom, which userspace doesn't care
857 countersize
= sizeof(struct xt_counters
) * private->number
;
858 counters
= vmalloc(countersize
);
860 if (counters
== NULL
)
863 /* First, sum counters... */
864 write_lock_bh(&table
->lock
);
865 get_counters(private, counters
);
866 write_unlock_bh(&table
->lock
);
868 /* choose the copy that is on ourc node/cpu */
869 loc_cpu_entry
= private->entries
[raw_smp_processor_id()];
870 if (copy_to_user(userptr
, loc_cpu_entry
, total_size
) != 0) {
875 /* FIXME: use iterator macros --RR */
876 /* ... then go back and fix counters and names */
877 for (off
= 0, num
= 0; off
< total_size
; off
+= e
->next_offset
, num
++){
879 struct ip6t_entry_match
*m
;
880 struct ip6t_entry_target
*t
;
882 e
= (struct ip6t_entry
*)(loc_cpu_entry
+ off
);
883 if (copy_to_user(userptr
+ off
884 + offsetof(struct ip6t_entry
, counters
),
886 sizeof(counters
[num
])) != 0) {
891 for (i
= sizeof(struct ip6t_entry
);
892 i
< e
->target_offset
;
893 i
+= m
->u
.match_size
) {
896 if (copy_to_user(userptr
+ off
+ i
897 + offsetof(struct ip6t_entry_match
,
899 m
->u
.kernel
.match
->name
,
900 strlen(m
->u
.kernel
.match
->name
)+1)
907 t
= ip6t_get_target(e
);
908 if (copy_to_user(userptr
+ off
+ e
->target_offset
909 + offsetof(struct ip6t_entry_target
,
911 t
->u
.kernel
.target
->name
,
912 strlen(t
->u
.kernel
.target
->name
)+1) != 0) {
924 get_entries(const struct ip6t_get_entries
*entries
,
925 struct ip6t_get_entries __user
*uptr
)
930 t
= xt_find_table_lock(AF_INET6
, entries
->name
);
931 if (t
&& !IS_ERR(t
)) {
932 struct xt_table_info
*private = t
->private;
933 duprintf("t->private->number = %u\n", private->number
);
934 if (entries
->size
== private->size
)
935 ret
= copy_entries_to_user(private->size
,
936 t
, uptr
->entrytable
);
938 duprintf("get_entries: I've got %u not %u!\n",
939 private->size
, entries
->size
);
945 ret
= t
? PTR_ERR(t
) : -ENOENT
;
951 do_replace(void __user
*user
, unsigned int len
)
954 struct ip6t_replace tmp
;
956 struct xt_table_info
*newinfo
, *oldinfo
;
957 struct xt_counters
*counters
;
958 void *loc_cpu_entry
, *loc_cpu_old_entry
;
960 if (copy_from_user(&tmp
, user
, sizeof(tmp
)) != 0)
964 if (tmp
.size
>= (INT_MAX
- sizeof(struct xt_table_info
)) / NR_CPUS
-
967 if (tmp
.num_counters
>= INT_MAX
/ sizeof(struct xt_counters
))
970 newinfo
= xt_alloc_table_info(tmp
.size
);
974 /* choose the copy that is on our node/cpu */
975 loc_cpu_entry
= newinfo
->entries
[raw_smp_processor_id()];
976 if (copy_from_user(loc_cpu_entry
, user
+ sizeof(tmp
),
982 counters
= vmalloc(tmp
.num_counters
* sizeof(struct xt_counters
));
988 ret
= translate_table(tmp
.name
, tmp
.valid_hooks
,
989 newinfo
, loc_cpu_entry
, tmp
.size
, tmp
.num_entries
,
990 tmp
.hook_entry
, tmp
.underflow
);
992 goto free_newinfo_counters
;
994 duprintf("ip_tables: Translated table\n");
996 t
= try_then_request_module(xt_find_table_lock(AF_INET6
, tmp
.name
),
997 "ip6table_%s", tmp
.name
);
998 if (!t
|| IS_ERR(t
)) {
999 ret
= t
? PTR_ERR(t
) : -ENOENT
;
1000 goto free_newinfo_counters_untrans
;
1004 if (tmp
.valid_hooks
!= t
->valid_hooks
) {
1005 duprintf("Valid hook crap: %08X vs %08X\n",
1006 tmp
.valid_hooks
, t
->valid_hooks
);
1011 oldinfo
= xt_replace_table(t
, tmp
.num_counters
, newinfo
, &ret
);
1015 /* Update module usage count based on number of rules */
1016 duprintf("do_replace: oldnum=%u, initnum=%u, newnum=%u\n",
1017 oldinfo
->number
, oldinfo
->initial_entries
, newinfo
->number
);
1018 if ((oldinfo
->number
> oldinfo
->initial_entries
) ||
1019 (newinfo
->number
<= oldinfo
->initial_entries
))
1021 if ((oldinfo
->number
> oldinfo
->initial_entries
) &&
1022 (newinfo
->number
<= oldinfo
->initial_entries
))
1025 /* Get the old counters. */
1026 get_counters(oldinfo
, counters
);
1027 /* Decrease module usage counts and free resource */
1028 loc_cpu_old_entry
= oldinfo
->entries
[raw_smp_processor_id()];
1029 IP6T_ENTRY_ITERATE(loc_cpu_old_entry
, oldinfo
->size
, cleanup_entry
,NULL
);
1030 xt_free_table_info(oldinfo
);
1031 if (copy_to_user(tmp
.counters
, counters
,
1032 sizeof(struct xt_counters
) * tmp
.num_counters
) != 0)
1041 free_newinfo_counters_untrans
:
1042 IP6T_ENTRY_ITERATE(loc_cpu_entry
, newinfo
->size
, cleanup_entry
,NULL
);
1043 free_newinfo_counters
:
1046 xt_free_table_info(newinfo
);
1050 /* We're lazy, and add to the first CPU; overflow works its fey magic
1051 * and everything is OK. */
1053 add_counter_to_entry(struct ip6t_entry
*e
,
1054 const struct xt_counters addme
[],
1058 duprintf("add_counter: Entry %u %lu/%lu + %lu/%lu\n",
1060 (long unsigned int)e
->counters
.pcnt
,
1061 (long unsigned int)e
->counters
.bcnt
,
1062 (long unsigned int)addme
[*i
].pcnt
,
1063 (long unsigned int)addme
[*i
].bcnt
);
1066 ADD_COUNTER(e
->counters
, addme
[*i
].bcnt
, addme
[*i
].pcnt
);
1073 do_add_counters(void __user
*user
, unsigned int len
)
1076 struct xt_counters_info tmp
, *paddc
;
1077 struct xt_table_info
*private;
1080 void *loc_cpu_entry
;
1082 if (copy_from_user(&tmp
, user
, sizeof(tmp
)) != 0)
1085 if (len
!= sizeof(tmp
) + tmp
.num_counters
*sizeof(struct xt_counters
))
1088 paddc
= vmalloc(len
);
1092 if (copy_from_user(paddc
, user
, len
) != 0) {
1097 t
= xt_find_table_lock(AF_INET6
, tmp
.name
);
1098 if (!t
|| IS_ERR(t
)) {
1099 ret
= t
? PTR_ERR(t
) : -ENOENT
;
1103 write_lock_bh(&t
->lock
);
1104 private = t
->private;
1105 if (private->number
!= tmp
.num_counters
) {
1107 goto unlock_up_free
;
1111 /* Choose the copy that is on our node */
1112 loc_cpu_entry
= private->entries
[smp_processor_id()];
1113 IP6T_ENTRY_ITERATE(loc_cpu_entry
,
1115 add_counter_to_entry
,
1119 write_unlock_bh(&t
->lock
);
1129 do_ip6t_set_ctl(struct sock
*sk
, int cmd
, void __user
*user
, unsigned int len
)
1133 if (!capable(CAP_NET_ADMIN
))
1137 case IP6T_SO_SET_REPLACE
:
1138 ret
= do_replace(user
, len
);
1141 case IP6T_SO_SET_ADD_COUNTERS
:
1142 ret
= do_add_counters(user
, len
);
1146 duprintf("do_ip6t_set_ctl: unknown request %i\n", cmd
);
1154 do_ip6t_get_ctl(struct sock
*sk
, int cmd
, void __user
*user
, int *len
)
1158 if (!capable(CAP_NET_ADMIN
))
1162 case IP6T_SO_GET_INFO
: {
1163 char name
[IP6T_TABLE_MAXNAMELEN
];
1166 if (*len
!= sizeof(struct ip6t_getinfo
)) {
1167 duprintf("length %u != %u\n", *len
,
1168 sizeof(struct ip6t_getinfo
));
1173 if (copy_from_user(name
, user
, sizeof(name
)) != 0) {
1177 name
[IP6T_TABLE_MAXNAMELEN
-1] = '\0';
1179 t
= try_then_request_module(xt_find_table_lock(AF_INET6
, name
),
1180 "ip6table_%s", name
);
1181 if (t
&& !IS_ERR(t
)) {
1182 struct ip6t_getinfo info
;
1183 struct xt_table_info
*private = t
->private;
1185 info
.valid_hooks
= t
->valid_hooks
;
1186 memcpy(info
.hook_entry
, private->hook_entry
,
1187 sizeof(info
.hook_entry
));
1188 memcpy(info
.underflow
, private->underflow
,
1189 sizeof(info
.underflow
));
1190 info
.num_entries
= private->number
;
1191 info
.size
= private->size
;
1192 memcpy(info
.name
, name
, sizeof(info
.name
));
1194 if (copy_to_user(user
, &info
, *len
) != 0)
1201 ret
= t
? PTR_ERR(t
) : -ENOENT
;
1205 case IP6T_SO_GET_ENTRIES
: {
1206 struct ip6t_get_entries get
;
1208 if (*len
< sizeof(get
)) {
1209 duprintf("get_entries: %u < %u\n", *len
, sizeof(get
));
1211 } else if (copy_from_user(&get
, user
, sizeof(get
)) != 0) {
1213 } else if (*len
!= sizeof(struct ip6t_get_entries
) + get
.size
) {
1214 duprintf("get_entries: %u != %u\n", *len
,
1215 sizeof(struct ip6t_get_entries
) + get
.size
);
1218 ret
= get_entries(&get
, user
);
1222 case IP6T_SO_GET_REVISION_MATCH
:
1223 case IP6T_SO_GET_REVISION_TARGET
: {
1224 struct ip6t_get_revision rev
;
1227 if (*len
!= sizeof(rev
)) {
1231 if (copy_from_user(&rev
, user
, sizeof(rev
)) != 0) {
1236 if (cmd
== IP6T_SO_GET_REVISION_TARGET
)
1241 try_then_request_module(xt_find_revision(AF_INET6
, rev
.name
,
1244 "ip6t_%s", rev
.name
);
1249 duprintf("do_ip6t_get_ctl: unknown request %i\n", cmd
);
1256 int ip6t_register_table(struct xt_table
*table
,
1257 const struct ip6t_replace
*repl
)
1260 struct xt_table_info
*newinfo
;
1261 static struct xt_table_info bootstrap
1262 = { 0, 0, 0, { 0 }, { 0 }, { } };
1263 void *loc_cpu_entry
;
1265 newinfo
= xt_alloc_table_info(repl
->size
);
1269 /* choose the copy on our node/cpu */
1270 loc_cpu_entry
= newinfo
->entries
[raw_smp_processor_id()];
1271 memcpy(loc_cpu_entry
, repl
->entries
, repl
->size
);
1273 ret
= translate_table(table
->name
, table
->valid_hooks
,
1274 newinfo
, loc_cpu_entry
, repl
->size
,
1279 xt_free_table_info(newinfo
);
1283 ret
= xt_register_table(table
, &bootstrap
, newinfo
);
1285 xt_free_table_info(newinfo
);
1292 void ip6t_unregister_table(struct xt_table
*table
)
1294 struct xt_table_info
*private;
1295 void *loc_cpu_entry
;
1297 private = xt_unregister_table(table
);
1299 /* Decrease module usage counts and free resources */
1300 loc_cpu_entry
= private->entries
[raw_smp_processor_id()];
1301 IP6T_ENTRY_ITERATE(loc_cpu_entry
, private->size
, cleanup_entry
, NULL
);
1302 xt_free_table_info(private);
1305 /* Returns 1 if the type and code is matched by the range, 0 otherwise */
1307 icmp6_type_code_match(u_int8_t test_type
, u_int8_t min_code
, u_int8_t max_code
,
1308 u_int8_t type
, u_int8_t code
,
1311 return (type
== test_type
&& code
>= min_code
&& code
<= max_code
)
1316 icmp6_match(const struct sk_buff
*skb
,
1317 const struct net_device
*in
,
1318 const struct net_device
*out
,
1319 const struct xt_match
*match
,
1320 const void *matchinfo
,
1322 unsigned int protoff
,
1325 struct icmp6hdr _icmp
, *ic
;
1326 const struct ip6t_icmp
*icmpinfo
= matchinfo
;
1328 /* Must not be a fragment. */
1332 ic
= skb_header_pointer(skb
, protoff
, sizeof(_icmp
), &_icmp
);
1334 /* We've been asked to examine this packet, and we
1335 can't. Hence, no choice but to drop. */
1336 duprintf("Dropping evil ICMP tinygram.\n");
1341 return icmp6_type_code_match(icmpinfo
->type
,
1344 ic
->icmp6_type
, ic
->icmp6_code
,
1345 !!(icmpinfo
->invflags
&IP6T_ICMP_INV
));
1348 /* Called when user tries to insert an entry of this type. */
1350 icmp6_checkentry(const char *tablename
,
1352 const struct xt_match
*match
,
1354 unsigned int hook_mask
)
1356 const struct ip6t_icmp
*icmpinfo
= matchinfo
;
1358 /* Must specify no unknown invflags */
1359 return !(icmpinfo
->invflags
& ~IP6T_ICMP_INV
);
1362 /* The built-in targets: standard (NULL) and error. */
1363 static struct ip6t_target ip6t_standard_target
= {
1364 .name
= IP6T_STANDARD_TARGET
,
1365 .targetsize
= sizeof(int),
1369 static struct ip6t_target ip6t_error_target
= {
1370 .name
= IP6T_ERROR_TARGET
,
1371 .target
= ip6t_error
,
1372 .targetsize
= IP6T_FUNCTION_MAXNAMELEN
,
1376 static struct nf_sockopt_ops ip6t_sockopts
= {
1378 .set_optmin
= IP6T_BASE_CTL
,
1379 .set_optmax
= IP6T_SO_SET_MAX
+1,
1380 .set
= do_ip6t_set_ctl
,
1381 .get_optmin
= IP6T_BASE_CTL
,
1382 .get_optmax
= IP6T_SO_GET_MAX
+1,
1383 .get
= do_ip6t_get_ctl
,
1386 static struct ip6t_match icmp6_matchstruct
= {
1388 .match
= &icmp6_match
,
1389 .matchsize
= sizeof(struct ip6t_icmp
),
1390 .checkentry
= icmp6_checkentry
,
1391 .proto
= IPPROTO_ICMPV6
,
1395 static int __init
ip6_tables_init(void)
1399 ret
= xt_proto_init(AF_INET6
);
1403 /* Noone else will be downing sem now, so we won't sleep */
1404 ret
= xt_register_target(&ip6t_standard_target
);
1407 ret
= xt_register_target(&ip6t_error_target
);
1410 ret
= xt_register_match(&icmp6_matchstruct
);
1414 /* Register setsockopt */
1415 ret
= nf_register_sockopt(&ip6t_sockopts
);
1419 printk("ip6_tables: (C) 2000-2006 Netfilter Core Team\n");
1423 xt_unregister_match(&icmp6_matchstruct
);
1425 xt_unregister_target(&ip6t_error_target
);
1427 xt_unregister_target(&ip6t_standard_target
);
1429 xt_proto_fini(AF_INET6
);
1434 static void __exit
ip6_tables_fini(void)
1436 nf_unregister_sockopt(&ip6t_sockopts
);
1437 xt_unregister_match(&icmp6_matchstruct
);
1438 xt_unregister_target(&ip6t_error_target
);
1439 xt_unregister_target(&ip6t_standard_target
);
1440 xt_proto_fini(AF_INET6
);
1444 * find the offset to specified header or the protocol number of last header
1445 * if target < 0. "last header" is transport protocol header, ESP, or
1448 * If target header is found, its offset is set in *offset and return protocol
1449 * number. Otherwise, return -1.
1451 * If the first fragment doesn't contain the final protocol header or
1452 * NEXTHDR_NONE it is considered invalid.
1454 * Note that non-1st fragment is special case that "the protocol number
1455 * of last header" is "next header" field in Fragment header. In this case,
1456 * *offset is meaningless and fragment offset is stored in *fragoff if fragoff
1460 int ipv6_find_hdr(const struct sk_buff
*skb
, unsigned int *offset
,
1461 int target
, unsigned short *fragoff
)
1463 unsigned int start
= (u8
*)(skb
->nh
.ipv6h
+ 1) - skb
->data
;
1464 u8 nexthdr
= skb
->nh
.ipv6h
->nexthdr
;
1465 unsigned int len
= skb
->len
- start
;
1470 while (nexthdr
!= target
) {
1471 struct ipv6_opt_hdr _hdr
, *hp
;
1472 unsigned int hdrlen
;
1474 if ((!ipv6_ext_hdr(nexthdr
)) || nexthdr
== NEXTHDR_NONE
) {
1480 hp
= skb_header_pointer(skb
, start
, sizeof(_hdr
), &_hdr
);
1483 if (nexthdr
== NEXTHDR_FRAGMENT
) {
1484 unsigned short _frag_off
, *fp
;
1485 fp
= skb_header_pointer(skb
,
1486 start
+offsetof(struct frag_hdr
,
1493 _frag_off
= ntohs(*fp
) & ~0x7;
1496 ((!ipv6_ext_hdr(hp
->nexthdr
)) ||
1497 hp
->nexthdr
== NEXTHDR_NONE
)) {
1499 *fragoff
= _frag_off
;
1505 } else if (nexthdr
== NEXTHDR_AUTH
)
1506 hdrlen
= (hp
->hdrlen
+ 2) << 2;
1508 hdrlen
= ipv6_optlen(hp
);
1510 nexthdr
= hp
->nexthdr
;
1519 EXPORT_SYMBOL(ip6t_register_table
);
1520 EXPORT_SYMBOL(ip6t_unregister_table
);
1521 EXPORT_SYMBOL(ip6t_do_table
);
1522 EXPORT_SYMBOL(ip6t_ext_hdr
);
1523 EXPORT_SYMBOL(ipv6_find_hdr
);
1525 module_init(ip6_tables_init
);
1526 module_exit(ip6_tables_fini
);