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>
6 * Copyright (C) 2006-2010 Patrick McHardy <kaber@trash.net>
8 * This program is free software; you can redistribute it and/or modify
9 * it under the terms of the GNU General Public License version 2 as
10 * published by the Free Software Foundation.
12 #define pr_fmt(fmt) KBUILD_MODNAME ": " fmt
13 #include <linux/cache.h>
14 #include <linux/capability.h>
15 #include <linux/skbuff.h>
16 #include <linux/kmod.h>
17 #include <linux/vmalloc.h>
18 #include <linux/netdevice.h>
19 #include <linux/module.h>
20 #include <linux/icmp.h>
22 #include <net/compat.h>
23 #include <asm/uaccess.h>
24 #include <linux/mutex.h>
25 #include <linux/proc_fs.h>
26 #include <linux/err.h>
27 #include <linux/cpumask.h>
29 #include <linux/netfilter/x_tables.h>
30 #include <linux/netfilter_ipv4/ip_tables.h>
31 #include <net/netfilter/nf_log.h>
32 #include "../../netfilter/xt_repldata.h"
34 MODULE_LICENSE("GPL");
35 MODULE_AUTHOR("Netfilter Core Team <coreteam@netfilter.org>");
36 MODULE_DESCRIPTION("IPv4 packet filter");
38 /*#define DEBUG_IP_FIREWALL*/
39 /*#define DEBUG_ALLOW_ALL*/ /* Useful for remote debugging */
40 /*#define DEBUG_IP_FIREWALL_USER*/
42 #ifdef DEBUG_IP_FIREWALL
43 #define dprintf(format, args...) pr_info(format , ## args)
45 #define dprintf(format, args...)
48 #ifdef DEBUG_IP_FIREWALL_USER
49 #define duprintf(format, args...) pr_info(format , ## args)
51 #define duprintf(format, args...)
54 #ifdef CONFIG_NETFILTER_DEBUG
55 #define IP_NF_ASSERT(x) WARN_ON(!(x))
57 #define IP_NF_ASSERT(x)
61 /* All the better to debug you with... */
66 void *ipt_alloc_initial_table(const struct xt_table
*info
)
68 return xt_alloc_initial_table(ipt
, IPT
);
70 EXPORT_SYMBOL_GPL(ipt_alloc_initial_table
);
72 /* Returns whether matches rule or not. */
73 /* Performance critical - called for every packet */
75 ip_packet_match(const struct iphdr
*ip
,
78 const struct ipt_ip
*ipinfo
,
83 #define FWINV(bool, invflg) ((bool) ^ !!(ipinfo->invflags & (invflg)))
85 if (FWINV((ip
->saddr
&ipinfo
->smsk
.s_addr
) != ipinfo
->src
.s_addr
,
87 FWINV((ip
->daddr
&ipinfo
->dmsk
.s_addr
) != ipinfo
->dst
.s_addr
,
89 dprintf("Source or dest mismatch.\n");
91 dprintf("SRC: %pI4. Mask: %pI4. Target: %pI4.%s\n",
92 &ip
->saddr
, &ipinfo
->smsk
.s_addr
, &ipinfo
->src
.s_addr
,
93 ipinfo
->invflags
& IPT_INV_SRCIP
? " (INV)" : "");
94 dprintf("DST: %pI4 Mask: %pI4 Target: %pI4.%s\n",
95 &ip
->daddr
, &ipinfo
->dmsk
.s_addr
, &ipinfo
->dst
.s_addr
,
96 ipinfo
->invflags
& IPT_INV_DSTIP
? " (INV)" : "");
100 ret
= ifname_compare_aligned(indev
, ipinfo
->iniface
, ipinfo
->iniface_mask
);
102 if (FWINV(ret
!= 0, IPT_INV_VIA_IN
)) {
103 dprintf("VIA in mismatch (%s vs %s).%s\n",
104 indev
, ipinfo
->iniface
,
105 ipinfo
->invflags
&IPT_INV_VIA_IN
?" (INV)":"");
109 ret
= ifname_compare_aligned(outdev
, ipinfo
->outiface
, ipinfo
->outiface_mask
);
111 if (FWINV(ret
!= 0, IPT_INV_VIA_OUT
)) {
112 dprintf("VIA out mismatch (%s vs %s).%s\n",
113 outdev
, ipinfo
->outiface
,
114 ipinfo
->invflags
&IPT_INV_VIA_OUT
?" (INV)":"");
118 /* Check specific protocol */
120 FWINV(ip
->protocol
!= ipinfo
->proto
, IPT_INV_PROTO
)) {
121 dprintf("Packet protocol %hi does not match %hi.%s\n",
122 ip
->protocol
, ipinfo
->proto
,
123 ipinfo
->invflags
&IPT_INV_PROTO
? " (INV)":"");
127 /* If we have a fragment rule but the packet is not a fragment
128 * then we return zero */
129 if (FWINV((ipinfo
->flags
&IPT_F_FRAG
) && !isfrag
, IPT_INV_FRAG
)) {
130 dprintf("Fragment rule but not fragment.%s\n",
131 ipinfo
->invflags
& IPT_INV_FRAG
? " (INV)" : "");
139 ip_checkentry(const struct ipt_ip
*ip
)
141 if (ip
->flags
& ~IPT_F_MASK
) {
142 duprintf("Unknown flag bits set: %08X\n",
143 ip
->flags
& ~IPT_F_MASK
);
146 if (ip
->invflags
& ~IPT_INV_MASK
) {
147 duprintf("Unknown invflag bits set: %08X\n",
148 ip
->invflags
& ~IPT_INV_MASK
);
155 ipt_error(struct sk_buff
*skb
, const struct xt_action_param
*par
)
157 net_info_ratelimited("error: `%s'\n", (const char *)par
->targinfo
);
162 /* Performance critical */
163 static inline struct ipt_entry
*
164 get_entry(const void *base
, unsigned int offset
)
166 return (struct ipt_entry
*)(base
+ offset
);
169 /* All zeroes == unconditional rule. */
170 /* Mildly perf critical (only if packet tracing is on) */
171 static inline bool unconditional(const struct ipt_entry
*e
)
173 static const struct ipt_ip uncond
;
175 return e
->target_offset
== sizeof(struct ipt_entry
) &&
176 memcmp(&e
->ip
, &uncond
, sizeof(uncond
)) == 0;
180 /* for const-correctness */
181 static inline const struct xt_entry_target
*
182 ipt_get_target_c(const struct ipt_entry
*e
)
184 return ipt_get_target((struct ipt_entry
*)e
);
187 #if IS_ENABLED(CONFIG_NETFILTER_XT_TARGET_TRACE)
188 static const char *const hooknames
[] = {
189 [NF_INET_PRE_ROUTING
] = "PREROUTING",
190 [NF_INET_LOCAL_IN
] = "INPUT",
191 [NF_INET_FORWARD
] = "FORWARD",
192 [NF_INET_LOCAL_OUT
] = "OUTPUT",
193 [NF_INET_POST_ROUTING
] = "POSTROUTING",
196 enum nf_ip_trace_comments
{
197 NF_IP_TRACE_COMMENT_RULE
,
198 NF_IP_TRACE_COMMENT_RETURN
,
199 NF_IP_TRACE_COMMENT_POLICY
,
202 static const char *const comments
[] = {
203 [NF_IP_TRACE_COMMENT_RULE
] = "rule",
204 [NF_IP_TRACE_COMMENT_RETURN
] = "return",
205 [NF_IP_TRACE_COMMENT_POLICY
] = "policy",
208 static struct nf_loginfo trace_loginfo
= {
209 .type
= NF_LOG_TYPE_LOG
,
213 .logflags
= NF_LOG_MASK
,
218 /* Mildly perf critical (only if packet tracing is on) */
220 get_chainname_rulenum(const struct ipt_entry
*s
, const struct ipt_entry
*e
,
221 const char *hookname
, const char **chainname
,
222 const char **comment
, unsigned int *rulenum
)
224 const struct xt_standard_target
*t
= (void *)ipt_get_target_c(s
);
226 if (strcmp(t
->target
.u
.kernel
.target
->name
, XT_ERROR_TARGET
) == 0) {
227 /* Head of user chain: ERROR target with chainname */
228 *chainname
= t
->target
.data
;
233 if (unconditional(s
) &&
234 strcmp(t
->target
.u
.kernel
.target
->name
,
235 XT_STANDARD_TARGET
) == 0 &&
237 /* Tail of chains: STANDARD target (return/policy) */
238 *comment
= *chainname
== hookname
239 ? comments
[NF_IP_TRACE_COMMENT_POLICY
]
240 : comments
[NF_IP_TRACE_COMMENT_RETURN
];
249 static void trace_packet(const struct sk_buff
*skb
,
251 const struct net_device
*in
,
252 const struct net_device
*out
,
253 const char *tablename
,
254 const struct xt_table_info
*private,
255 const struct ipt_entry
*e
)
257 const void *table_base
;
258 const struct ipt_entry
*root
;
259 const char *hookname
, *chainname
, *comment
;
260 const struct ipt_entry
*iter
;
261 unsigned int rulenum
= 0;
262 struct net
*net
= dev_net(in
? in
: out
);
264 table_base
= private->entries
[smp_processor_id()];
265 root
= get_entry(table_base
, private->hook_entry
[hook
]);
267 hookname
= chainname
= hooknames
[hook
];
268 comment
= comments
[NF_IP_TRACE_COMMENT_RULE
];
270 xt_entry_foreach(iter
, root
, private->size
- private->hook_entry
[hook
])
271 if (get_chainname_rulenum(iter
, e
, hookname
,
272 &chainname
, &comment
, &rulenum
) != 0)
275 nf_log_packet(net
, AF_INET
, hook
, skb
, in
, out
, &trace_loginfo
,
276 "TRACE: %s:%s:%s:%u ",
277 tablename
, chainname
, comment
, rulenum
);
282 struct ipt_entry
*ipt_next_entry(const struct ipt_entry
*entry
)
284 return (void *)entry
+ entry
->next_offset
;
287 /* Returns one of the generic firewall policies, like NF_ACCEPT. */
289 ipt_do_table(struct sk_buff
*skb
,
291 const struct net_device
*in
,
292 const struct net_device
*out
,
293 struct xt_table
*table
)
295 static const char nulldevname
[IFNAMSIZ
] __attribute__((aligned(sizeof(long))));
296 const struct iphdr
*ip
;
297 /* Initializing verdict to NF_DROP keeps gcc happy. */
298 unsigned int verdict
= NF_DROP
;
299 const char *indev
, *outdev
;
300 const void *table_base
;
301 struct ipt_entry
*e
, **jumpstack
;
302 unsigned int *stackptr
, origptr
, cpu
;
303 const struct xt_table_info
*private;
304 struct xt_action_param acpar
;
309 indev
= in
? in
->name
: nulldevname
;
310 outdev
= out
? out
->name
: nulldevname
;
311 /* We handle fragments by dealing with the first fragment as
312 * if it was a normal packet. All other fragments are treated
313 * normally, except that they will NEVER match rules that ask
314 * things we don't know, ie. tcp syn flag or ports). If the
315 * rule is also a fragment-specific rule, non-fragments won't
317 acpar
.fragoff
= ntohs(ip
->frag_off
) & IP_OFFSET
;
318 acpar
.thoff
= ip_hdrlen(skb
);
319 acpar
.hotdrop
= false;
322 acpar
.family
= NFPROTO_IPV4
;
323 acpar
.hooknum
= hook
;
325 IP_NF_ASSERT(table
->valid_hooks
& (1 << hook
));
327 addend
= xt_write_recseq_begin();
328 private = table
->private;
329 cpu
= smp_processor_id();
330 table_base
= private->entries
[cpu
];
331 jumpstack
= (struct ipt_entry
**)private->jumpstack
[cpu
];
332 stackptr
= per_cpu_ptr(private->stackptr
, cpu
);
335 e
= get_entry(table_base
, private->hook_entry
[hook
]);
337 pr_debug("Entering %s(hook %u); sp at %u (UF %p)\n",
338 table
->name
, hook
, origptr
,
339 get_entry(table_base
, private->underflow
[hook
]));
342 const struct xt_entry_target
*t
;
343 const struct xt_entry_match
*ematch
;
346 if (!ip_packet_match(ip
, indev
, outdev
,
347 &e
->ip
, acpar
.fragoff
)) {
349 e
= ipt_next_entry(e
);
353 xt_ematch_foreach(ematch
, e
) {
354 acpar
.match
= ematch
->u
.kernel
.match
;
355 acpar
.matchinfo
= ematch
->data
;
356 if (!acpar
.match
->match(skb
, &acpar
))
360 ADD_COUNTER(e
->counters
, skb
->len
, 1);
362 t
= ipt_get_target(e
);
363 IP_NF_ASSERT(t
->u
.kernel
.target
);
365 #if IS_ENABLED(CONFIG_NETFILTER_XT_TARGET_TRACE)
366 /* The packet is traced: log it */
367 if (unlikely(skb
->nf_trace
))
368 trace_packet(skb
, hook
, in
, out
,
369 table
->name
, private, e
);
371 /* Standard target? */
372 if (!t
->u
.kernel
.target
->target
) {
375 v
= ((struct xt_standard_target
*)t
)->verdict
;
377 /* Pop from stack? */
378 if (v
!= XT_RETURN
) {
379 verdict
= (unsigned int)(-v
) - 1;
382 if (*stackptr
<= origptr
) {
383 e
= get_entry(table_base
,
384 private->underflow
[hook
]);
385 pr_debug("Underflow (this is normal) "
388 e
= jumpstack
[--*stackptr
];
389 pr_debug("Pulled %p out from pos %u\n",
391 e
= ipt_next_entry(e
);
395 if (table_base
+ v
!= ipt_next_entry(e
) &&
396 !(e
->ip
.flags
& IPT_F_GOTO
)) {
397 if (*stackptr
>= private->stacksize
) {
401 jumpstack
[(*stackptr
)++] = e
;
402 pr_debug("Pushed %p into pos %u\n",
406 e
= get_entry(table_base
, v
);
410 acpar
.target
= t
->u
.kernel
.target
;
411 acpar
.targinfo
= t
->data
;
413 verdict
= t
->u
.kernel
.target
->target(skb
, &acpar
);
414 /* Target might have changed stuff. */
416 if (verdict
== XT_CONTINUE
)
417 e
= ipt_next_entry(e
);
421 } while (!acpar
.hotdrop
);
422 pr_debug("Exiting %s; resetting sp from %u to %u\n",
423 __func__
, *stackptr
, origptr
);
425 xt_write_recseq_end(addend
);
428 #ifdef DEBUG_ALLOW_ALL
437 /* Figures out from what hook each rule can be called: returns 0 if
438 there are loops. Puts hook bitmask in comefrom. */
440 mark_source_chains(const struct xt_table_info
*newinfo
,
441 unsigned int valid_hooks
, void *entry0
)
445 /* No recursion; use packet counter to save back ptrs (reset
446 to 0 as we leave), and comefrom to save source hook bitmask */
447 for (hook
= 0; hook
< NF_INET_NUMHOOKS
; hook
++) {
448 unsigned int pos
= newinfo
->hook_entry
[hook
];
449 struct ipt_entry
*e
= (struct ipt_entry
*)(entry0
+ pos
);
451 if (!(valid_hooks
& (1 << hook
)))
454 /* Set initial back pointer. */
455 e
->counters
.pcnt
= pos
;
458 const struct xt_standard_target
*t
459 = (void *)ipt_get_target_c(e
);
460 int visited
= e
->comefrom
& (1 << hook
);
462 if (e
->comefrom
& (1 << NF_INET_NUMHOOKS
)) {
463 pr_err("iptables: loop hook %u pos %u %08X.\n",
464 hook
, pos
, e
->comefrom
);
467 e
->comefrom
|= ((1 << hook
) | (1 << NF_INET_NUMHOOKS
));
469 /* Unconditional return/END. */
470 if ((unconditional(e
) &&
471 (strcmp(t
->target
.u
.user
.name
,
472 XT_STANDARD_TARGET
) == 0) &&
473 t
->verdict
< 0) || visited
) {
474 unsigned int oldpos
, size
;
476 if ((strcmp(t
->target
.u
.user
.name
,
477 XT_STANDARD_TARGET
) == 0) &&
478 t
->verdict
< -NF_MAX_VERDICT
- 1) {
479 duprintf("mark_source_chains: bad "
480 "negative verdict (%i)\n",
485 /* Return: backtrack through the last
488 e
->comefrom
^= (1<<NF_INET_NUMHOOKS
);
489 #ifdef DEBUG_IP_FIREWALL_USER
491 & (1 << NF_INET_NUMHOOKS
)) {
492 duprintf("Back unset "
499 pos
= e
->counters
.pcnt
;
500 e
->counters
.pcnt
= 0;
502 /* We're at the start. */
506 e
= (struct ipt_entry
*)
508 } while (oldpos
== pos
+ e
->next_offset
);
511 size
= e
->next_offset
;
512 e
= (struct ipt_entry
*)
513 (entry0
+ pos
+ size
);
514 if (pos
+ size
>= newinfo
->size
)
516 e
->counters
.pcnt
= pos
;
519 int newpos
= t
->verdict
;
521 if (strcmp(t
->target
.u
.user
.name
,
522 XT_STANDARD_TARGET
) == 0 &&
524 if (newpos
> newinfo
->size
-
525 sizeof(struct ipt_entry
)) {
526 duprintf("mark_source_chains: "
527 "bad verdict (%i)\n",
531 /* This a jump; chase it. */
532 duprintf("Jump rule %u -> %u\n",
535 /* ... this is a fallthru */
536 newpos
= pos
+ e
->next_offset
;
537 if (newpos
>= newinfo
->size
)
540 e
= (struct ipt_entry
*)
542 e
->counters
.pcnt
= pos
;
547 duprintf("Finished chain %u\n", hook
);
552 static void cleanup_match(struct xt_entry_match
*m
, struct net
*net
)
554 struct xt_mtdtor_param par
;
557 par
.match
= m
->u
.kernel
.match
;
558 par
.matchinfo
= m
->data
;
559 par
.family
= NFPROTO_IPV4
;
560 if (par
.match
->destroy
!= NULL
)
561 par
.match
->destroy(&par
);
562 module_put(par
.match
->me
);
566 check_entry(const struct ipt_entry
*e
)
568 const struct xt_entry_target
*t
;
570 if (!ip_checkentry(&e
->ip
))
573 if (e
->target_offset
+ sizeof(struct xt_entry_target
) >
577 t
= ipt_get_target_c(e
);
578 if (e
->target_offset
+ t
->u
.target_size
> e
->next_offset
)
586 check_match(struct xt_entry_match
*m
, struct xt_mtchk_param
*par
)
588 const struct ipt_ip
*ip
= par
->entryinfo
;
591 par
->match
= m
->u
.kernel
.match
;
592 par
->matchinfo
= m
->data
;
594 ret
= xt_check_match(par
, m
->u
.match_size
- sizeof(*m
),
595 ip
->proto
, ip
->invflags
& IPT_INV_PROTO
);
597 duprintf("check failed for `%s'.\n", par
->match
->name
);
604 find_check_match(struct xt_entry_match
*m
, struct xt_mtchk_param
*par
)
606 struct xt_match
*match
;
609 match
= xt_request_find_match(NFPROTO_IPV4
, m
->u
.user
.name
,
612 duprintf("find_check_match: `%s' not found\n", m
->u
.user
.name
);
613 return PTR_ERR(match
);
615 m
->u
.kernel
.match
= match
;
617 ret
= check_match(m
, par
);
623 module_put(m
->u
.kernel
.match
->me
);
627 static int check_target(struct ipt_entry
*e
, struct net
*net
, const char *name
)
629 struct xt_entry_target
*t
= ipt_get_target(e
);
630 struct xt_tgchk_param par
= {
634 .target
= t
->u
.kernel
.target
,
636 .hook_mask
= e
->comefrom
,
637 .family
= NFPROTO_IPV4
,
641 ret
= xt_check_target(&par
, t
->u
.target_size
- sizeof(*t
),
642 e
->ip
.proto
, e
->ip
.invflags
& IPT_INV_PROTO
);
644 duprintf("check failed for `%s'.\n",
645 t
->u
.kernel
.target
->name
);
652 find_check_entry(struct ipt_entry
*e
, struct net
*net
, const char *name
,
655 struct xt_entry_target
*t
;
656 struct xt_target
*target
;
659 struct xt_mtchk_param mtpar
;
660 struct xt_entry_match
*ematch
;
665 mtpar
.entryinfo
= &e
->ip
;
666 mtpar
.hook_mask
= e
->comefrom
;
667 mtpar
.family
= NFPROTO_IPV4
;
668 xt_ematch_foreach(ematch
, e
) {
669 ret
= find_check_match(ematch
, &mtpar
);
671 goto cleanup_matches
;
675 t
= ipt_get_target(e
);
676 target
= xt_request_find_target(NFPROTO_IPV4
, t
->u
.user
.name
,
678 if (IS_ERR(target
)) {
679 duprintf("find_check_entry: `%s' not found\n", t
->u
.user
.name
);
680 ret
= PTR_ERR(target
);
681 goto cleanup_matches
;
683 t
->u
.kernel
.target
= target
;
685 ret
= check_target(e
, net
, name
);
690 module_put(t
->u
.kernel
.target
->me
);
692 xt_ematch_foreach(ematch
, e
) {
695 cleanup_match(ematch
, net
);
700 static bool check_underflow(const struct ipt_entry
*e
)
702 const struct xt_entry_target
*t
;
703 unsigned int verdict
;
705 if (!unconditional(e
))
707 t
= ipt_get_target_c(e
);
708 if (strcmp(t
->u
.user
.name
, XT_STANDARD_TARGET
) != 0)
710 verdict
= ((struct xt_standard_target
*)t
)->verdict
;
711 verdict
= -verdict
- 1;
712 return verdict
== NF_DROP
|| verdict
== NF_ACCEPT
;
716 check_entry_size_and_hooks(struct ipt_entry
*e
,
717 struct xt_table_info
*newinfo
,
718 const unsigned char *base
,
719 const unsigned char *limit
,
720 const unsigned int *hook_entries
,
721 const unsigned int *underflows
,
722 unsigned int valid_hooks
)
727 if ((unsigned long)e
% __alignof__(struct ipt_entry
) != 0 ||
728 (unsigned char *)e
+ sizeof(struct ipt_entry
) >= limit
||
729 (unsigned char *)e
+ e
->next_offset
> limit
) {
730 duprintf("Bad offset %p\n", e
);
735 < sizeof(struct ipt_entry
) + sizeof(struct xt_entry_target
)) {
736 duprintf("checking: element %p size %u\n",
741 err
= check_entry(e
);
746 /* Check hooks & underflows */
747 for (h
= 0; h
< NF_INET_NUMHOOKS
; h
++) {
748 if (!(valid_hooks
& (1 << h
)))
750 if ((unsigned char *)e
- base
== hook_entries
[h
])
751 newinfo
->hook_entry
[h
] = hook_entries
[h
];
752 if ((unsigned char *)e
- base
== underflows
[h
]) {
753 if (!check_underflow(e
)) {
754 pr_debug("Underflows must be unconditional and "
755 "use the STANDARD target with "
759 newinfo
->underflow
[h
] = underflows
[h
];
763 /* Clear counters and comefrom */
764 e
->counters
= ((struct xt_counters
) { 0, 0 });
770 cleanup_entry(struct ipt_entry
*e
, struct net
*net
)
772 struct xt_tgdtor_param par
;
773 struct xt_entry_target
*t
;
774 struct xt_entry_match
*ematch
;
776 /* Cleanup all matches */
777 xt_ematch_foreach(ematch
, e
)
778 cleanup_match(ematch
, net
);
779 t
= ipt_get_target(e
);
782 par
.target
= t
->u
.kernel
.target
;
783 par
.targinfo
= t
->data
;
784 par
.family
= NFPROTO_IPV4
;
785 if (par
.target
->destroy
!= NULL
)
786 par
.target
->destroy(&par
);
787 module_put(par
.target
->me
);
790 /* Checks and translates the user-supplied table segment (held in
793 translate_table(struct net
*net
, struct xt_table_info
*newinfo
, void *entry0
,
794 const struct ipt_replace
*repl
)
796 struct ipt_entry
*iter
;
800 newinfo
->size
= repl
->size
;
801 newinfo
->number
= repl
->num_entries
;
803 /* Init all hooks to impossible value. */
804 for (i
= 0; i
< NF_INET_NUMHOOKS
; i
++) {
805 newinfo
->hook_entry
[i
] = 0xFFFFFFFF;
806 newinfo
->underflow
[i
] = 0xFFFFFFFF;
809 duprintf("translate_table: size %u\n", newinfo
->size
);
811 /* Walk through entries, checking offsets. */
812 xt_entry_foreach(iter
, entry0
, newinfo
->size
) {
813 ret
= check_entry_size_and_hooks(iter
, newinfo
, entry0
,
821 if (strcmp(ipt_get_target(iter
)->u
.user
.name
,
822 XT_ERROR_TARGET
) == 0)
823 ++newinfo
->stacksize
;
826 if (i
!= repl
->num_entries
) {
827 duprintf("translate_table: %u not %u entries\n",
828 i
, repl
->num_entries
);
832 /* Check hooks all assigned */
833 for (i
= 0; i
< NF_INET_NUMHOOKS
; i
++) {
834 /* Only hooks which are valid */
835 if (!(repl
->valid_hooks
& (1 << i
)))
837 if (newinfo
->hook_entry
[i
] == 0xFFFFFFFF) {
838 duprintf("Invalid hook entry %u %u\n",
839 i
, repl
->hook_entry
[i
]);
842 if (newinfo
->underflow
[i
] == 0xFFFFFFFF) {
843 duprintf("Invalid underflow %u %u\n",
844 i
, repl
->underflow
[i
]);
849 if (!mark_source_chains(newinfo
, repl
->valid_hooks
, entry0
))
852 /* Finally, each sanity check must pass */
854 xt_entry_foreach(iter
, entry0
, newinfo
->size
) {
855 ret
= find_check_entry(iter
, net
, repl
->name
, repl
->size
);
862 xt_entry_foreach(iter
, entry0
, newinfo
->size
) {
865 cleanup_entry(iter
, net
);
870 /* And one copy for every other CPU */
871 for_each_possible_cpu(i
) {
872 if (newinfo
->entries
[i
] && newinfo
->entries
[i
] != entry0
)
873 memcpy(newinfo
->entries
[i
], entry0
, newinfo
->size
);
880 get_counters(const struct xt_table_info
*t
,
881 struct xt_counters counters
[])
883 struct ipt_entry
*iter
;
887 for_each_possible_cpu(cpu
) {
888 seqcount_t
*s
= &per_cpu(xt_recseq
, cpu
);
891 xt_entry_foreach(iter
, t
->entries
[cpu
], t
->size
) {
896 start
= read_seqcount_begin(s
);
897 bcnt
= iter
->counters
.bcnt
;
898 pcnt
= iter
->counters
.pcnt
;
899 } while (read_seqcount_retry(s
, start
));
901 ADD_COUNTER(counters
[i
], bcnt
, pcnt
);
902 ++i
; /* macro does multi eval of i */
907 static struct xt_counters
*alloc_counters(const struct xt_table
*table
)
909 unsigned int countersize
;
910 struct xt_counters
*counters
;
911 const struct xt_table_info
*private = table
->private;
913 /* We need atomic snapshot of counters: rest doesn't change
914 (other than comefrom, which userspace doesn't care
916 countersize
= sizeof(struct xt_counters
) * private->number
;
917 counters
= vzalloc(countersize
);
919 if (counters
== NULL
)
920 return ERR_PTR(-ENOMEM
);
922 get_counters(private, counters
);
928 copy_entries_to_user(unsigned int total_size
,
929 const struct xt_table
*table
,
930 void __user
*userptr
)
932 unsigned int off
, num
;
933 const struct ipt_entry
*e
;
934 struct xt_counters
*counters
;
935 const struct xt_table_info
*private = table
->private;
937 const void *loc_cpu_entry
;
939 counters
= alloc_counters(table
);
940 if (IS_ERR(counters
))
941 return PTR_ERR(counters
);
943 /* choose the copy that is on our node/cpu, ...
944 * This choice is lazy (because current thread is
945 * allowed to migrate to another cpu)
947 loc_cpu_entry
= private->entries
[raw_smp_processor_id()];
948 if (copy_to_user(userptr
, loc_cpu_entry
, total_size
) != 0) {
953 /* FIXME: use iterator macros --RR */
954 /* ... then go back and fix counters and names */
955 for (off
= 0, num
= 0; off
< total_size
; off
+= e
->next_offset
, num
++){
957 const struct xt_entry_match
*m
;
958 const struct xt_entry_target
*t
;
960 e
= (struct ipt_entry
*)(loc_cpu_entry
+ off
);
961 if (copy_to_user(userptr
+ off
962 + offsetof(struct ipt_entry
, counters
),
964 sizeof(counters
[num
])) != 0) {
969 for (i
= sizeof(struct ipt_entry
);
970 i
< e
->target_offset
;
971 i
+= m
->u
.match_size
) {
974 if (copy_to_user(userptr
+ off
+ i
975 + offsetof(struct xt_entry_match
,
977 m
->u
.kernel
.match
->name
,
978 strlen(m
->u
.kernel
.match
->name
)+1)
985 t
= ipt_get_target_c(e
);
986 if (copy_to_user(userptr
+ off
+ e
->target_offset
987 + offsetof(struct xt_entry_target
,
989 t
->u
.kernel
.target
->name
,
990 strlen(t
->u
.kernel
.target
->name
)+1) != 0) {
1001 #ifdef CONFIG_COMPAT
1002 static void compat_standard_from_user(void *dst
, const void *src
)
1004 int v
= *(compat_int_t
*)src
;
1007 v
+= xt_compat_calc_jump(AF_INET
, v
);
1008 memcpy(dst
, &v
, sizeof(v
));
1011 static int compat_standard_to_user(void __user
*dst
, const void *src
)
1013 compat_int_t cv
= *(int *)src
;
1016 cv
-= xt_compat_calc_jump(AF_INET
, cv
);
1017 return copy_to_user(dst
, &cv
, sizeof(cv
)) ? -EFAULT
: 0;
1020 static int compat_calc_entry(const struct ipt_entry
*e
,
1021 const struct xt_table_info
*info
,
1022 const void *base
, struct xt_table_info
*newinfo
)
1024 const struct xt_entry_match
*ematch
;
1025 const struct xt_entry_target
*t
;
1026 unsigned int entry_offset
;
1029 off
= sizeof(struct ipt_entry
) - sizeof(struct compat_ipt_entry
);
1030 entry_offset
= (void *)e
- base
;
1031 xt_ematch_foreach(ematch
, e
)
1032 off
+= xt_compat_match_offset(ematch
->u
.kernel
.match
);
1033 t
= ipt_get_target_c(e
);
1034 off
+= xt_compat_target_offset(t
->u
.kernel
.target
);
1035 newinfo
->size
-= off
;
1036 ret
= xt_compat_add_offset(AF_INET
, entry_offset
, off
);
1040 for (i
= 0; i
< NF_INET_NUMHOOKS
; i
++) {
1041 if (info
->hook_entry
[i
] &&
1042 (e
< (struct ipt_entry
*)(base
+ info
->hook_entry
[i
])))
1043 newinfo
->hook_entry
[i
] -= off
;
1044 if (info
->underflow
[i
] &&
1045 (e
< (struct ipt_entry
*)(base
+ info
->underflow
[i
])))
1046 newinfo
->underflow
[i
] -= off
;
1051 static int compat_table_info(const struct xt_table_info
*info
,
1052 struct xt_table_info
*newinfo
)
1054 struct ipt_entry
*iter
;
1055 void *loc_cpu_entry
;
1058 if (!newinfo
|| !info
)
1061 /* we dont care about newinfo->entries[] */
1062 memcpy(newinfo
, info
, offsetof(struct xt_table_info
, entries
));
1063 newinfo
->initial_entries
= 0;
1064 loc_cpu_entry
= info
->entries
[raw_smp_processor_id()];
1065 xt_compat_init_offsets(AF_INET
, info
->number
);
1066 xt_entry_foreach(iter
, loc_cpu_entry
, info
->size
) {
1067 ret
= compat_calc_entry(iter
, info
, loc_cpu_entry
, newinfo
);
1075 static int get_info(struct net
*net
, void __user
*user
,
1076 const int *len
, int compat
)
1078 char name
[XT_TABLE_MAXNAMELEN
];
1082 if (*len
!= sizeof(struct ipt_getinfo
)) {
1083 duprintf("length %u != %zu\n", *len
,
1084 sizeof(struct ipt_getinfo
));
1088 if (copy_from_user(name
, user
, sizeof(name
)) != 0)
1091 name
[XT_TABLE_MAXNAMELEN
-1] = '\0';
1092 #ifdef CONFIG_COMPAT
1094 xt_compat_lock(AF_INET
);
1096 t
= try_then_request_module(xt_find_table_lock(net
, AF_INET
, name
),
1097 "iptable_%s", name
);
1098 if (!IS_ERR_OR_NULL(t
)) {
1099 struct ipt_getinfo info
;
1100 const struct xt_table_info
*private = t
->private;
1101 #ifdef CONFIG_COMPAT
1102 struct xt_table_info tmp
;
1105 ret
= compat_table_info(private, &tmp
);
1106 xt_compat_flush_offsets(AF_INET
);
1110 memset(&info
, 0, sizeof(info
));
1111 info
.valid_hooks
= t
->valid_hooks
;
1112 memcpy(info
.hook_entry
, private->hook_entry
,
1113 sizeof(info
.hook_entry
));
1114 memcpy(info
.underflow
, private->underflow
,
1115 sizeof(info
.underflow
));
1116 info
.num_entries
= private->number
;
1117 info
.size
= private->size
;
1118 strcpy(info
.name
, name
);
1120 if (copy_to_user(user
, &info
, *len
) != 0)
1128 ret
= t
? PTR_ERR(t
) : -ENOENT
;
1129 #ifdef CONFIG_COMPAT
1131 xt_compat_unlock(AF_INET
);
1137 get_entries(struct net
*net
, struct ipt_get_entries __user
*uptr
,
1141 struct ipt_get_entries get
;
1144 if (*len
< sizeof(get
)) {
1145 duprintf("get_entries: %u < %zu\n", *len
, sizeof(get
));
1148 if (copy_from_user(&get
, uptr
, sizeof(get
)) != 0)
1150 if (*len
!= sizeof(struct ipt_get_entries
) + get
.size
) {
1151 duprintf("get_entries: %u != %zu\n",
1152 *len
, sizeof(get
) + get
.size
);
1156 t
= xt_find_table_lock(net
, AF_INET
, get
.name
);
1157 if (!IS_ERR_OR_NULL(t
)) {
1158 const struct xt_table_info
*private = t
->private;
1159 duprintf("t->private->number = %u\n", private->number
);
1160 if (get
.size
== private->size
)
1161 ret
= copy_entries_to_user(private->size
,
1162 t
, uptr
->entrytable
);
1164 duprintf("get_entries: I've got %u not %u!\n",
1165 private->size
, get
.size
);
1171 ret
= t
? PTR_ERR(t
) : -ENOENT
;
1177 __do_replace(struct net
*net
, const char *name
, unsigned int valid_hooks
,
1178 struct xt_table_info
*newinfo
, unsigned int num_counters
,
1179 void __user
*counters_ptr
)
1183 struct xt_table_info
*oldinfo
;
1184 struct xt_counters
*counters
;
1185 void *loc_cpu_old_entry
;
1186 struct ipt_entry
*iter
;
1189 counters
= vzalloc(num_counters
* sizeof(struct xt_counters
));
1195 t
= try_then_request_module(xt_find_table_lock(net
, AF_INET
, name
),
1196 "iptable_%s", name
);
1197 if (IS_ERR_OR_NULL(t
)) {
1198 ret
= t
? PTR_ERR(t
) : -ENOENT
;
1199 goto free_newinfo_counters_untrans
;
1203 if (valid_hooks
!= t
->valid_hooks
) {
1204 duprintf("Valid hook crap: %08X vs %08X\n",
1205 valid_hooks
, t
->valid_hooks
);
1210 oldinfo
= xt_replace_table(t
, num_counters
, newinfo
, &ret
);
1214 /* Update module usage count based on number of rules */
1215 duprintf("do_replace: oldnum=%u, initnum=%u, newnum=%u\n",
1216 oldinfo
->number
, oldinfo
->initial_entries
, newinfo
->number
);
1217 if ((oldinfo
->number
> oldinfo
->initial_entries
) ||
1218 (newinfo
->number
<= oldinfo
->initial_entries
))
1220 if ((oldinfo
->number
> oldinfo
->initial_entries
) &&
1221 (newinfo
->number
<= oldinfo
->initial_entries
))
1224 /* Get the old counters, and synchronize with replace */
1225 get_counters(oldinfo
, counters
);
1227 /* Decrease module usage counts and free resource */
1228 loc_cpu_old_entry
= oldinfo
->entries
[raw_smp_processor_id()];
1229 xt_entry_foreach(iter
, loc_cpu_old_entry
, oldinfo
->size
)
1230 cleanup_entry(iter
, net
);
1232 xt_free_table_info(oldinfo
);
1233 if (copy_to_user(counters_ptr
, counters
,
1234 sizeof(struct xt_counters
) * num_counters
) != 0) {
1235 /* Silent error, can't fail, new table is already in place */
1236 net_warn_ratelimited("iptables: counters copy to user failed while replacing table\n");
1245 free_newinfo_counters_untrans
:
1252 do_replace(struct net
*net
, const void __user
*user
, unsigned int len
)
1255 struct ipt_replace tmp
;
1256 struct xt_table_info
*newinfo
;
1257 void *loc_cpu_entry
;
1258 struct ipt_entry
*iter
;
1260 if (copy_from_user(&tmp
, user
, sizeof(tmp
)) != 0)
1263 /* overflow check */
1264 if (tmp
.num_counters
>= INT_MAX
/ sizeof(struct xt_counters
))
1266 if (tmp
.num_counters
== 0)
1269 tmp
.name
[sizeof(tmp
.name
)-1] = 0;
1271 newinfo
= xt_alloc_table_info(tmp
.size
);
1275 /* choose the copy that is on our node/cpu */
1276 loc_cpu_entry
= newinfo
->entries
[raw_smp_processor_id()];
1277 if (copy_from_user(loc_cpu_entry
, user
+ sizeof(tmp
),
1283 ret
= translate_table(net
, newinfo
, loc_cpu_entry
, &tmp
);
1287 duprintf("Translated table\n");
1289 ret
= __do_replace(net
, tmp
.name
, tmp
.valid_hooks
, newinfo
,
1290 tmp
.num_counters
, tmp
.counters
);
1292 goto free_newinfo_untrans
;
1295 free_newinfo_untrans
:
1296 xt_entry_foreach(iter
, loc_cpu_entry
, newinfo
->size
)
1297 cleanup_entry(iter
, net
);
1299 xt_free_table_info(newinfo
);
1304 do_add_counters(struct net
*net
, const void __user
*user
,
1305 unsigned int len
, int compat
)
1307 unsigned int i
, curcpu
;
1308 struct xt_counters_info tmp
;
1309 struct xt_counters
*paddc
;
1311 const struct xt_table_info
*private;
1313 void *loc_cpu_entry
;
1314 struct ipt_entry
*iter
;
1315 unsigned int addend
;
1317 paddc
= xt_copy_counters_from_user(user
, len
, &tmp
, compat
);
1319 return PTR_ERR(paddc
);
1321 t
= xt_find_table_lock(net
, AF_INET
, tmp
.name
);
1322 if (IS_ERR_OR_NULL(t
)) {
1323 ret
= t
? PTR_ERR(t
) : -ENOENT
;
1328 private = t
->private;
1329 if (private->number
!= tmp
.num_counters
) {
1331 goto unlock_up_free
;
1335 /* Choose the copy that is on our node */
1336 curcpu
= smp_processor_id();
1337 loc_cpu_entry
= private->entries
[curcpu
];
1338 addend
= xt_write_recseq_begin();
1339 xt_entry_foreach(iter
, loc_cpu_entry
, private->size
) {
1340 ADD_COUNTER(iter
->counters
, paddc
[i
].bcnt
, paddc
[i
].pcnt
);
1343 xt_write_recseq_end(addend
);
1354 #ifdef CONFIG_COMPAT
1355 struct compat_ipt_replace
{
1356 char name
[XT_TABLE_MAXNAMELEN
];
1360 u32 hook_entry
[NF_INET_NUMHOOKS
];
1361 u32 underflow
[NF_INET_NUMHOOKS
];
1363 compat_uptr_t counters
; /* struct xt_counters * */
1364 struct compat_ipt_entry entries
[0];
1368 compat_copy_entry_to_user(struct ipt_entry
*e
, void __user
**dstptr
,
1369 unsigned int *size
, struct xt_counters
*counters
,
1372 struct xt_entry_target
*t
;
1373 struct compat_ipt_entry __user
*ce
;
1374 u_int16_t target_offset
, next_offset
;
1375 compat_uint_t origsize
;
1376 const struct xt_entry_match
*ematch
;
1380 ce
= (struct compat_ipt_entry __user
*)*dstptr
;
1381 if (copy_to_user(ce
, e
, sizeof(struct ipt_entry
)) != 0 ||
1382 copy_to_user(&ce
->counters
, &counters
[i
],
1383 sizeof(counters
[i
])) != 0)
1386 *dstptr
+= sizeof(struct compat_ipt_entry
);
1387 *size
-= sizeof(struct ipt_entry
) - sizeof(struct compat_ipt_entry
);
1389 xt_ematch_foreach(ematch
, e
) {
1390 ret
= xt_compat_match_to_user(ematch
, dstptr
, size
);
1394 target_offset
= e
->target_offset
- (origsize
- *size
);
1395 t
= ipt_get_target(e
);
1396 ret
= xt_compat_target_to_user(t
, dstptr
, size
);
1399 next_offset
= e
->next_offset
- (origsize
- *size
);
1400 if (put_user(target_offset
, &ce
->target_offset
) != 0 ||
1401 put_user(next_offset
, &ce
->next_offset
) != 0)
1407 compat_find_calc_match(struct xt_entry_match
*m
,
1408 const struct ipt_ip
*ip
,
1409 unsigned int hookmask
,
1412 struct xt_match
*match
;
1414 match
= xt_request_find_match(NFPROTO_IPV4
, m
->u
.user
.name
,
1415 m
->u
.user
.revision
);
1416 if (IS_ERR(match
)) {
1417 duprintf("compat_check_calc_match: `%s' not found\n",
1419 return PTR_ERR(match
);
1421 m
->u
.kernel
.match
= match
;
1422 *size
+= xt_compat_match_offset(match
);
1426 static void compat_release_entry(struct compat_ipt_entry
*e
)
1428 struct xt_entry_target
*t
;
1429 struct xt_entry_match
*ematch
;
1431 /* Cleanup all matches */
1432 xt_ematch_foreach(ematch
, e
)
1433 module_put(ematch
->u
.kernel
.match
->me
);
1434 t
= compat_ipt_get_target(e
);
1435 module_put(t
->u
.kernel
.target
->me
);
1439 check_compat_entry_size_and_hooks(struct compat_ipt_entry
*e
,
1440 struct xt_table_info
*newinfo
,
1442 const unsigned char *base
,
1443 const unsigned char *limit
)
1445 struct xt_entry_match
*ematch
;
1446 struct xt_entry_target
*t
;
1447 struct xt_target
*target
;
1448 unsigned int entry_offset
;
1452 duprintf("check_compat_entry_size_and_hooks %p\n", e
);
1453 if ((unsigned long)e
% __alignof__(struct compat_ipt_entry
) != 0 ||
1454 (unsigned char *)e
+ sizeof(struct compat_ipt_entry
) >= limit
||
1455 (unsigned char *)e
+ e
->next_offset
> limit
) {
1456 duprintf("Bad offset %p, limit = %p\n", e
, limit
);
1460 if (e
->next_offset
< sizeof(struct compat_ipt_entry
) +
1461 sizeof(struct compat_xt_entry_target
)) {
1462 duprintf("checking: element %p size %u\n",
1467 /* For purposes of check_entry casting the compat entry is fine */
1468 ret
= check_entry((struct ipt_entry
*)e
);
1473 off
= sizeof(struct ipt_entry
) - sizeof(struct compat_ipt_entry
);
1474 entry_offset
= (void *)e
- (void *)base
;
1476 xt_ematch_foreach(ematch
, e
) {
1477 ret
= compat_find_calc_match(ematch
, &e
->ip
, e
->comefrom
,
1480 goto release_matches
;
1484 t
= compat_ipt_get_target(e
);
1485 target
= xt_request_find_target(NFPROTO_IPV4
, t
->u
.user
.name
,
1486 t
->u
.user
.revision
);
1487 if (IS_ERR(target
)) {
1488 duprintf("check_compat_entry_size_and_hooks: `%s' not found\n",
1490 ret
= PTR_ERR(target
);
1491 goto release_matches
;
1493 t
->u
.kernel
.target
= target
;
1495 off
+= xt_compat_target_offset(target
);
1497 ret
= xt_compat_add_offset(AF_INET
, entry_offset
, off
);
1504 module_put(t
->u
.kernel
.target
->me
);
1506 xt_ematch_foreach(ematch
, e
) {
1509 module_put(ematch
->u
.kernel
.match
->me
);
1515 compat_copy_entry_from_user(struct compat_ipt_entry
*e
, void **dstptr
,
1517 struct xt_table_info
*newinfo
, unsigned char *base
)
1519 struct xt_entry_target
*t
;
1520 struct xt_target
*target
;
1521 struct ipt_entry
*de
;
1522 unsigned int origsize
;
1524 struct xt_entry_match
*ematch
;
1527 de
= (struct ipt_entry
*)*dstptr
;
1528 memcpy(de
, e
, sizeof(struct ipt_entry
));
1529 memcpy(&de
->counters
, &e
->counters
, sizeof(e
->counters
));
1531 *dstptr
+= sizeof(struct ipt_entry
);
1532 *size
+= sizeof(struct ipt_entry
) - sizeof(struct compat_ipt_entry
);
1534 xt_ematch_foreach(ematch
, e
)
1535 xt_compat_match_from_user(ematch
, dstptr
, size
);
1537 de
->target_offset
= e
->target_offset
- (origsize
- *size
);
1538 t
= compat_ipt_get_target(e
);
1539 target
= t
->u
.kernel
.target
;
1540 xt_compat_target_from_user(t
, dstptr
, size
);
1542 de
->next_offset
= e
->next_offset
- (origsize
- *size
);
1544 for (h
= 0; h
< NF_INET_NUMHOOKS
; h
++) {
1545 if ((unsigned char *)de
- base
< newinfo
->hook_entry
[h
])
1546 newinfo
->hook_entry
[h
] -= origsize
- *size
;
1547 if ((unsigned char *)de
- base
< newinfo
->underflow
[h
])
1548 newinfo
->underflow
[h
] -= origsize
- *size
;
1553 translate_compat_table(struct net
*net
,
1554 struct xt_table_info
**pinfo
,
1556 const struct compat_ipt_replace
*compatr
)
1559 struct xt_table_info
*newinfo
, *info
;
1560 void *pos
, *entry0
, *entry1
;
1561 struct compat_ipt_entry
*iter0
;
1562 struct ipt_replace repl
;
1568 size
= compatr
->size
;
1569 info
->number
= compatr
->num_entries
;
1571 duprintf("translate_compat_table: size %u\n", info
->size
);
1573 xt_compat_lock(AF_INET
);
1574 xt_compat_init_offsets(AF_INET
, compatr
->num_entries
);
1575 /* Walk through entries, checking offsets. */
1576 xt_entry_foreach(iter0
, entry0
, compatr
->size
) {
1577 ret
= check_compat_entry_size_and_hooks(iter0
, info
, &size
,
1579 entry0
+ compatr
->size
);
1586 if (j
!= compatr
->num_entries
) {
1587 duprintf("translate_compat_table: %u not %u entries\n",
1588 j
, compatr
->num_entries
);
1593 newinfo
= xt_alloc_table_info(size
);
1597 newinfo
->number
= compatr
->num_entries
;
1598 for (i
= 0; i
< NF_INET_NUMHOOKS
; i
++) {
1599 newinfo
->hook_entry
[i
] = compatr
->hook_entry
[i
];
1600 newinfo
->underflow
[i
] = compatr
->underflow
[i
];
1602 entry1
= newinfo
->entries
[raw_smp_processor_id()];
1604 size
= compatr
->size
;
1605 xt_entry_foreach(iter0
, entry0
, compatr
->size
)
1606 compat_copy_entry_from_user(iter0
, &pos
, &size
,
1609 /* all module references in entry0 are now gone.
1610 * entry1/newinfo contains a 64bit ruleset that looks exactly as
1611 * generated by 64bit userspace.
1613 * Call standard translate_table() to validate all hook_entrys,
1614 * underflows, check for loops, etc.
1616 xt_compat_flush_offsets(AF_INET
);
1617 xt_compat_unlock(AF_INET
);
1619 memcpy(&repl
, compatr
, sizeof(*compatr
));
1621 for (i
= 0; i
< NF_INET_NUMHOOKS
; i
++) {
1622 repl
.hook_entry
[i
] = newinfo
->hook_entry
[i
];
1623 repl
.underflow
[i
] = newinfo
->underflow
[i
];
1626 repl
.num_counters
= 0;
1627 repl
.counters
= NULL
;
1628 repl
.size
= newinfo
->size
;
1629 ret
= translate_table(net
, newinfo
, entry1
, &repl
);
1635 xt_free_table_info(info
);
1639 xt_free_table_info(newinfo
);
1642 xt_compat_flush_offsets(AF_INET
);
1643 xt_compat_unlock(AF_INET
);
1644 xt_entry_foreach(iter0
, entry0
, compatr
->size
) {
1647 compat_release_entry(iter0
);
1653 compat_do_replace(struct net
*net
, void __user
*user
, unsigned int len
)
1656 struct compat_ipt_replace tmp
;
1657 struct xt_table_info
*newinfo
;
1658 void *loc_cpu_entry
;
1659 struct ipt_entry
*iter
;
1661 if (copy_from_user(&tmp
, user
, sizeof(tmp
)) != 0)
1664 /* overflow check */
1665 if (tmp
.size
>= INT_MAX
/ num_possible_cpus())
1667 if (tmp
.num_counters
>= INT_MAX
/ sizeof(struct xt_counters
))
1669 if (tmp
.num_counters
== 0)
1672 tmp
.name
[sizeof(tmp
.name
)-1] = 0;
1674 newinfo
= xt_alloc_table_info(tmp
.size
);
1678 /* choose the copy that is on our node/cpu */
1679 loc_cpu_entry
= newinfo
->entries
[raw_smp_processor_id()];
1680 if (copy_from_user(loc_cpu_entry
, user
+ sizeof(tmp
),
1686 ret
= translate_compat_table(net
, &newinfo
, &loc_cpu_entry
, &tmp
);
1690 duprintf("compat_do_replace: Translated table\n");
1692 ret
= __do_replace(net
, tmp
.name
, tmp
.valid_hooks
, newinfo
,
1693 tmp
.num_counters
, compat_ptr(tmp
.counters
));
1695 goto free_newinfo_untrans
;
1698 free_newinfo_untrans
:
1699 xt_entry_foreach(iter
, loc_cpu_entry
, newinfo
->size
)
1700 cleanup_entry(iter
, net
);
1702 xt_free_table_info(newinfo
);
1707 compat_do_ipt_set_ctl(struct sock
*sk
, int cmd
, void __user
*user
,
1712 if (!ns_capable(sock_net(sk
)->user_ns
, CAP_NET_ADMIN
))
1716 case IPT_SO_SET_REPLACE
:
1717 ret
= compat_do_replace(sock_net(sk
), user
, len
);
1720 case IPT_SO_SET_ADD_COUNTERS
:
1721 ret
= do_add_counters(sock_net(sk
), user
, len
, 1);
1725 duprintf("do_ipt_set_ctl: unknown request %i\n", cmd
);
1732 struct compat_ipt_get_entries
{
1733 char name
[XT_TABLE_MAXNAMELEN
];
1735 struct compat_ipt_entry entrytable
[0];
1739 compat_copy_entries_to_user(unsigned int total_size
, struct xt_table
*table
,
1740 void __user
*userptr
)
1742 struct xt_counters
*counters
;
1743 const struct xt_table_info
*private = table
->private;
1747 const void *loc_cpu_entry
;
1749 struct ipt_entry
*iter
;
1751 counters
= alloc_counters(table
);
1752 if (IS_ERR(counters
))
1753 return PTR_ERR(counters
);
1755 /* choose the copy that is on our node/cpu, ...
1756 * This choice is lazy (because current thread is
1757 * allowed to migrate to another cpu)
1759 loc_cpu_entry
= private->entries
[raw_smp_processor_id()];
1762 xt_entry_foreach(iter
, loc_cpu_entry
, total_size
) {
1763 ret
= compat_copy_entry_to_user(iter
, &pos
,
1764 &size
, counters
, i
++);
1774 compat_get_entries(struct net
*net
, struct compat_ipt_get_entries __user
*uptr
,
1778 struct compat_ipt_get_entries get
;
1781 if (*len
< sizeof(get
)) {
1782 duprintf("compat_get_entries: %u < %zu\n", *len
, sizeof(get
));
1786 if (copy_from_user(&get
, uptr
, sizeof(get
)) != 0)
1789 if (*len
!= sizeof(struct compat_ipt_get_entries
) + get
.size
) {
1790 duprintf("compat_get_entries: %u != %zu\n",
1791 *len
, sizeof(get
) + get
.size
);
1795 xt_compat_lock(AF_INET
);
1796 t
= xt_find_table_lock(net
, AF_INET
, get
.name
);
1797 if (!IS_ERR_OR_NULL(t
)) {
1798 const struct xt_table_info
*private = t
->private;
1799 struct xt_table_info info
;
1800 duprintf("t->private->number = %u\n", private->number
);
1801 ret
= compat_table_info(private, &info
);
1802 if (!ret
&& get
.size
== info
.size
) {
1803 ret
= compat_copy_entries_to_user(private->size
,
1804 t
, uptr
->entrytable
);
1806 duprintf("compat_get_entries: I've got %u not %u!\n",
1807 private->size
, get
.size
);
1810 xt_compat_flush_offsets(AF_INET
);
1814 ret
= t
? PTR_ERR(t
) : -ENOENT
;
1816 xt_compat_unlock(AF_INET
);
1820 static int do_ipt_get_ctl(struct sock
*, int, void __user
*, int *);
1823 compat_do_ipt_get_ctl(struct sock
*sk
, int cmd
, void __user
*user
, int *len
)
1827 if (!ns_capable(sock_net(sk
)->user_ns
, CAP_NET_ADMIN
))
1831 case IPT_SO_GET_INFO
:
1832 ret
= get_info(sock_net(sk
), user
, len
, 1);
1834 case IPT_SO_GET_ENTRIES
:
1835 ret
= compat_get_entries(sock_net(sk
), user
, len
);
1838 ret
= do_ipt_get_ctl(sk
, cmd
, user
, len
);
1845 do_ipt_set_ctl(struct sock
*sk
, int cmd
, void __user
*user
, unsigned int len
)
1849 if (!ns_capable(sock_net(sk
)->user_ns
, CAP_NET_ADMIN
))
1853 case IPT_SO_SET_REPLACE
:
1854 ret
= do_replace(sock_net(sk
), user
, len
);
1857 case IPT_SO_SET_ADD_COUNTERS
:
1858 ret
= do_add_counters(sock_net(sk
), user
, len
, 0);
1862 duprintf("do_ipt_set_ctl: unknown request %i\n", cmd
);
1870 do_ipt_get_ctl(struct sock
*sk
, int cmd
, void __user
*user
, int *len
)
1874 if (!ns_capable(sock_net(sk
)->user_ns
, CAP_NET_ADMIN
))
1878 case IPT_SO_GET_INFO
:
1879 ret
= get_info(sock_net(sk
), user
, len
, 0);
1882 case IPT_SO_GET_ENTRIES
:
1883 ret
= get_entries(sock_net(sk
), user
, len
);
1886 case IPT_SO_GET_REVISION_MATCH
:
1887 case IPT_SO_GET_REVISION_TARGET
: {
1888 struct xt_get_revision rev
;
1891 if (*len
!= sizeof(rev
)) {
1895 if (copy_from_user(&rev
, user
, sizeof(rev
)) != 0) {
1899 rev
.name
[sizeof(rev
.name
)-1] = 0;
1901 if (cmd
== IPT_SO_GET_REVISION_TARGET
)
1906 try_then_request_module(xt_find_revision(AF_INET
, rev
.name
,
1909 "ipt_%s", rev
.name
);
1914 duprintf("do_ipt_get_ctl: unknown request %i\n", cmd
);
1921 struct xt_table
*ipt_register_table(struct net
*net
,
1922 const struct xt_table
*table
,
1923 const struct ipt_replace
*repl
)
1926 struct xt_table_info
*newinfo
;
1927 struct xt_table_info bootstrap
= {0};
1928 void *loc_cpu_entry
;
1929 struct xt_table
*new_table
;
1931 newinfo
= xt_alloc_table_info(repl
->size
);
1937 /* choose the copy on our node/cpu, but dont care about preemption */
1938 loc_cpu_entry
= newinfo
->entries
[raw_smp_processor_id()];
1939 memcpy(loc_cpu_entry
, repl
->entries
, repl
->size
);
1941 ret
= translate_table(net
, newinfo
, loc_cpu_entry
, repl
);
1945 new_table
= xt_register_table(net
, table
, &bootstrap
, newinfo
);
1946 if (IS_ERR(new_table
)) {
1947 ret
= PTR_ERR(new_table
);
1954 xt_free_table_info(newinfo
);
1956 return ERR_PTR(ret
);
1959 void ipt_unregister_table(struct net
*net
, struct xt_table
*table
)
1961 struct xt_table_info
*private;
1962 void *loc_cpu_entry
;
1963 struct module
*table_owner
= table
->me
;
1964 struct ipt_entry
*iter
;
1966 private = xt_unregister_table(table
);
1968 /* Decrease module usage counts and free resources */
1969 loc_cpu_entry
= private->entries
[raw_smp_processor_id()];
1970 xt_entry_foreach(iter
, loc_cpu_entry
, private->size
)
1971 cleanup_entry(iter
, net
);
1972 if (private->number
> private->initial_entries
)
1973 module_put(table_owner
);
1974 xt_free_table_info(private);
1977 /* Returns 1 if the type and code is matched by the range, 0 otherwise */
1979 icmp_type_code_match(u_int8_t test_type
, u_int8_t min_code
, u_int8_t max_code
,
1980 u_int8_t type
, u_int8_t code
,
1983 return ((test_type
== 0xFF) ||
1984 (type
== test_type
&& code
>= min_code
&& code
<= max_code
))
1989 icmp_match(const struct sk_buff
*skb
, struct xt_action_param
*par
)
1991 const struct icmphdr
*ic
;
1992 struct icmphdr _icmph
;
1993 const struct ipt_icmp
*icmpinfo
= par
->matchinfo
;
1995 /* Must not be a fragment. */
1996 if (par
->fragoff
!= 0)
1999 ic
= skb_header_pointer(skb
, par
->thoff
, sizeof(_icmph
), &_icmph
);
2001 /* We've been asked to examine this packet, and we
2002 * can't. Hence, no choice but to drop.
2004 duprintf("Dropping evil ICMP tinygram.\n");
2005 par
->hotdrop
= true;
2009 return icmp_type_code_match(icmpinfo
->type
,
2013 !!(icmpinfo
->invflags
&IPT_ICMP_INV
));
2016 static int icmp_checkentry(const struct xt_mtchk_param
*par
)
2018 const struct ipt_icmp
*icmpinfo
= par
->matchinfo
;
2020 /* Must specify no unknown invflags */
2021 return (icmpinfo
->invflags
& ~IPT_ICMP_INV
) ? -EINVAL
: 0;
2024 static struct xt_target ipt_builtin_tg
[] __read_mostly
= {
2026 .name
= XT_STANDARD_TARGET
,
2027 .targetsize
= sizeof(int),
2028 .family
= NFPROTO_IPV4
,
2029 #ifdef CONFIG_COMPAT
2030 .compatsize
= sizeof(compat_int_t
),
2031 .compat_from_user
= compat_standard_from_user
,
2032 .compat_to_user
= compat_standard_to_user
,
2036 .name
= XT_ERROR_TARGET
,
2037 .target
= ipt_error
,
2038 .targetsize
= XT_FUNCTION_MAXNAMELEN
,
2039 .family
= NFPROTO_IPV4
,
2043 static struct nf_sockopt_ops ipt_sockopts
= {
2045 .set_optmin
= IPT_BASE_CTL
,
2046 .set_optmax
= IPT_SO_SET_MAX
+1,
2047 .set
= do_ipt_set_ctl
,
2048 #ifdef CONFIG_COMPAT
2049 .compat_set
= compat_do_ipt_set_ctl
,
2051 .get_optmin
= IPT_BASE_CTL
,
2052 .get_optmax
= IPT_SO_GET_MAX
+1,
2053 .get
= do_ipt_get_ctl
,
2054 #ifdef CONFIG_COMPAT
2055 .compat_get
= compat_do_ipt_get_ctl
,
2057 .owner
= THIS_MODULE
,
2060 static struct xt_match ipt_builtin_mt
[] __read_mostly
= {
2063 .match
= icmp_match
,
2064 .matchsize
= sizeof(struct ipt_icmp
),
2065 .checkentry
= icmp_checkentry
,
2066 .proto
= IPPROTO_ICMP
,
2067 .family
= NFPROTO_IPV4
,
2071 static int __net_init
ip_tables_net_init(struct net
*net
)
2073 return xt_proto_init(net
, NFPROTO_IPV4
);
2076 static void __net_exit
ip_tables_net_exit(struct net
*net
)
2078 xt_proto_fini(net
, NFPROTO_IPV4
);
2081 static struct pernet_operations ip_tables_net_ops
= {
2082 .init
= ip_tables_net_init
,
2083 .exit
= ip_tables_net_exit
,
2086 static int __init
ip_tables_init(void)
2090 ret
= register_pernet_subsys(&ip_tables_net_ops
);
2094 /* No one else will be downing sem now, so we won't sleep */
2095 ret
= xt_register_targets(ipt_builtin_tg
, ARRAY_SIZE(ipt_builtin_tg
));
2098 ret
= xt_register_matches(ipt_builtin_mt
, ARRAY_SIZE(ipt_builtin_mt
));
2102 /* Register setsockopt */
2103 ret
= nf_register_sockopt(&ipt_sockopts
);
2107 pr_info("(C) 2000-2006 Netfilter Core Team\n");
2111 xt_unregister_matches(ipt_builtin_mt
, ARRAY_SIZE(ipt_builtin_mt
));
2113 xt_unregister_targets(ipt_builtin_tg
, ARRAY_SIZE(ipt_builtin_tg
));
2115 unregister_pernet_subsys(&ip_tables_net_ops
);
2120 static void __exit
ip_tables_fini(void)
2122 nf_unregister_sockopt(&ipt_sockopts
);
2124 xt_unregister_matches(ipt_builtin_mt
, ARRAY_SIZE(ipt_builtin_mt
));
2125 xt_unregister_targets(ipt_builtin_tg
, ARRAY_SIZE(ipt_builtin_tg
));
2126 unregister_pernet_subsys(&ip_tables_net_ops
);
2129 EXPORT_SYMBOL(ipt_register_table
);
2130 EXPORT_SYMBOL(ipt_unregister_table
);
2131 EXPORT_SYMBOL(ipt_do_table
);
2132 module_init(ip_tables_init
);
2133 module_exit(ip_tables_fini
);