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(struct net
*net
,
250 const struct sk_buff
*skb
,
252 const struct net_device
*in
,
253 const struct net_device
*out
,
254 const char *tablename
,
255 const struct xt_table_info
*private,
256 const struct ipt_entry
*e
)
258 const struct ipt_entry
*root
;
259 const char *hookname
, *chainname
, *comment
;
260 const struct ipt_entry
*iter
;
261 unsigned int rulenum
= 0;
263 root
= get_entry(private->entries
, private->hook_entry
[hook
]);
265 hookname
= chainname
= hooknames
[hook
];
266 comment
= comments
[NF_IP_TRACE_COMMENT_RULE
];
268 xt_entry_foreach(iter
, root
, private->size
- private->hook_entry
[hook
])
269 if (get_chainname_rulenum(iter
, e
, hookname
,
270 &chainname
, &comment
, &rulenum
) != 0)
273 nf_log_trace(net
, AF_INET
, hook
, skb
, in
, out
, &trace_loginfo
,
274 "TRACE: %s:%s:%s:%u ",
275 tablename
, chainname
, comment
, rulenum
);
280 struct ipt_entry
*ipt_next_entry(const struct ipt_entry
*entry
)
282 return (void *)entry
+ entry
->next_offset
;
285 /* Returns one of the generic firewall policies, like NF_ACCEPT. */
287 ipt_do_table(struct sk_buff
*skb
,
288 const struct nf_hook_state
*state
,
289 struct xt_table
*table
)
291 unsigned int hook
= state
->hook
;
292 static const char nulldevname
[IFNAMSIZ
] __attribute__((aligned(sizeof(long))));
293 const struct iphdr
*ip
;
294 /* Initializing verdict to NF_DROP keeps gcc happy. */
295 unsigned int verdict
= NF_DROP
;
296 const char *indev
, *outdev
;
297 const void *table_base
;
298 struct ipt_entry
*e
, **jumpstack
;
299 unsigned int stackidx
, cpu
;
300 const struct xt_table_info
*private;
301 struct xt_action_param acpar
;
307 indev
= state
->in
? state
->in
->name
: nulldevname
;
308 outdev
= state
->out
? state
->out
->name
: nulldevname
;
309 /* We handle fragments by dealing with the first fragment as
310 * if it was a normal packet. All other fragments are treated
311 * normally, except that they will NEVER match rules that ask
312 * things we don't know, ie. tcp syn flag or ports). If the
313 * rule is also a fragment-specific rule, non-fragments won't
315 acpar
.fragoff
= ntohs(ip
->frag_off
) & IP_OFFSET
;
316 acpar
.thoff
= ip_hdrlen(skb
);
317 acpar
.hotdrop
= false;
318 acpar
.net
= state
->net
;
319 acpar
.in
= state
->in
;
320 acpar
.out
= state
->out
;
321 acpar
.family
= NFPROTO_IPV4
;
322 acpar
.hooknum
= hook
;
324 IP_NF_ASSERT(table
->valid_hooks
& (1 << hook
));
326 addend
= xt_write_recseq_begin();
327 private = table
->private;
328 cpu
= smp_processor_id();
330 * Ensure we load private-> members after we've fetched the base
333 smp_read_barrier_depends();
334 table_base
= private->entries
;
335 jumpstack
= (struct ipt_entry
**)private->jumpstack
[cpu
];
337 /* Switch to alternate jumpstack if we're being invoked via TEE.
338 * TEE issues XT_CONTINUE verdict on original skb so we must not
339 * clobber the jumpstack.
341 * For recursion via REJECT or SYNPROXY the stack will be clobbered
342 * but it is no problem since absolute verdict is issued by these.
344 if (static_key_false(&xt_tee_enabled
))
345 jumpstack
+= private->stacksize
* __this_cpu_read(nf_skb_duplicated
);
347 e
= get_entry(table_base
, private->hook_entry
[hook
]);
349 pr_debug("Entering %s(hook %u), UF %p\n",
351 get_entry(table_base
, private->underflow
[hook
]));
354 const struct xt_entry_target
*t
;
355 const struct xt_entry_match
*ematch
;
356 struct xt_counters
*counter
;
359 if (!ip_packet_match(ip
, indev
, outdev
,
360 &e
->ip
, acpar
.fragoff
)) {
362 e
= ipt_next_entry(e
);
366 xt_ematch_foreach(ematch
, e
) {
367 acpar
.match
= ematch
->u
.kernel
.match
;
368 acpar
.matchinfo
= ematch
->data
;
369 if (!acpar
.match
->match(skb
, &acpar
))
373 counter
= xt_get_this_cpu_counter(&e
->counters
);
374 ADD_COUNTER(*counter
, skb
->len
, 1);
376 t
= ipt_get_target(e
);
377 IP_NF_ASSERT(t
->u
.kernel
.target
);
379 #if IS_ENABLED(CONFIG_NETFILTER_XT_TARGET_TRACE)
380 /* The packet is traced: log it */
381 if (unlikely(skb
->nf_trace
))
382 trace_packet(state
->net
, skb
, hook
, state
->in
,
383 state
->out
, table
->name
, private, e
);
385 /* Standard target? */
386 if (!t
->u
.kernel
.target
->target
) {
389 v
= ((struct xt_standard_target
*)t
)->verdict
;
391 /* Pop from stack? */
392 if (v
!= XT_RETURN
) {
393 verdict
= (unsigned int)(-v
) - 1;
397 e
= get_entry(table_base
,
398 private->underflow
[hook
]);
399 pr_debug("Underflow (this is normal) "
402 e
= jumpstack
[--stackidx
];
403 pr_debug("Pulled %p out from pos %u\n",
405 e
= ipt_next_entry(e
);
409 if (table_base
+ v
!= ipt_next_entry(e
) &&
410 !(e
->ip
.flags
& IPT_F_GOTO
)) {
411 jumpstack
[stackidx
++] = e
;
412 pr_debug("Pushed %p into pos %u\n",
416 e
= get_entry(table_base
, v
);
420 acpar
.target
= t
->u
.kernel
.target
;
421 acpar
.targinfo
= t
->data
;
423 verdict
= t
->u
.kernel
.target
->target(skb
, &acpar
);
424 /* Target might have changed stuff. */
426 if (verdict
== XT_CONTINUE
)
427 e
= ipt_next_entry(e
);
431 } while (!acpar
.hotdrop
);
432 pr_debug("Exiting %s; sp at %u\n", __func__
, stackidx
);
434 xt_write_recseq_end(addend
);
437 #ifdef DEBUG_ALLOW_ALL
446 static bool find_jump_target(const struct xt_table_info
*t
,
447 const struct ipt_entry
*target
)
449 struct ipt_entry
*iter
;
451 xt_entry_foreach(iter
, t
->entries
, t
->size
) {
458 /* Figures out from what hook each rule can be called: returns 0 if
459 there are loops. Puts hook bitmask in comefrom. */
461 mark_source_chains(const struct xt_table_info
*newinfo
,
462 unsigned int valid_hooks
, void *entry0
)
466 /* No recursion; use packet counter to save back ptrs (reset
467 to 0 as we leave), and comefrom to save source hook bitmask */
468 for (hook
= 0; hook
< NF_INET_NUMHOOKS
; hook
++) {
469 unsigned int pos
= newinfo
->hook_entry
[hook
];
470 struct ipt_entry
*e
= (struct ipt_entry
*)(entry0
+ pos
);
472 if (!(valid_hooks
& (1 << hook
)))
475 /* Set initial back pointer. */
476 e
->counters
.pcnt
= pos
;
479 const struct xt_standard_target
*t
480 = (void *)ipt_get_target_c(e
);
481 int visited
= e
->comefrom
& (1 << hook
);
483 if (e
->comefrom
& (1 << NF_INET_NUMHOOKS
)) {
484 pr_err("iptables: loop hook %u pos %u %08X.\n",
485 hook
, pos
, e
->comefrom
);
488 e
->comefrom
|= ((1 << hook
) | (1 << NF_INET_NUMHOOKS
));
490 /* Unconditional return/END. */
491 if ((unconditional(e
) &&
492 (strcmp(t
->target
.u
.user
.name
,
493 XT_STANDARD_TARGET
) == 0) &&
494 t
->verdict
< 0) || visited
) {
495 unsigned int oldpos
, size
;
497 if ((strcmp(t
->target
.u
.user
.name
,
498 XT_STANDARD_TARGET
) == 0) &&
499 t
->verdict
< -NF_MAX_VERDICT
- 1) {
500 duprintf("mark_source_chains: bad "
501 "negative verdict (%i)\n",
506 /* Return: backtrack through the last
509 e
->comefrom
^= (1<<NF_INET_NUMHOOKS
);
510 #ifdef DEBUG_IP_FIREWALL_USER
512 & (1 << NF_INET_NUMHOOKS
)) {
513 duprintf("Back unset "
520 pos
= e
->counters
.pcnt
;
521 e
->counters
.pcnt
= 0;
523 /* We're at the start. */
527 e
= (struct ipt_entry
*)
529 } while (oldpos
== pos
+ e
->next_offset
);
532 size
= e
->next_offset
;
533 e
= (struct ipt_entry
*)
534 (entry0
+ pos
+ size
);
535 if (pos
+ size
>= newinfo
->size
)
537 e
->counters
.pcnt
= pos
;
540 int newpos
= t
->verdict
;
542 if (strcmp(t
->target
.u
.user
.name
,
543 XT_STANDARD_TARGET
) == 0 &&
545 if (newpos
> newinfo
->size
-
546 sizeof(struct ipt_entry
)) {
547 duprintf("mark_source_chains: "
548 "bad verdict (%i)\n",
552 /* This a jump; chase it. */
553 duprintf("Jump rule %u -> %u\n",
555 e
= (struct ipt_entry
*)
557 if (!find_jump_target(newinfo
, e
))
560 /* ... this is a fallthru */
561 newpos
= pos
+ e
->next_offset
;
562 if (newpos
>= newinfo
->size
)
565 e
= (struct ipt_entry
*)
567 e
->counters
.pcnt
= pos
;
572 duprintf("Finished chain %u\n", hook
);
577 static void cleanup_match(struct xt_entry_match
*m
, struct net
*net
)
579 struct xt_mtdtor_param par
;
582 par
.match
= m
->u
.kernel
.match
;
583 par
.matchinfo
= m
->data
;
584 par
.family
= NFPROTO_IPV4
;
585 if (par
.match
->destroy
!= NULL
)
586 par
.match
->destroy(&par
);
587 module_put(par
.match
->me
);
591 check_match(struct xt_entry_match
*m
, struct xt_mtchk_param
*par
)
593 const struct ipt_ip
*ip
= par
->entryinfo
;
596 par
->match
= m
->u
.kernel
.match
;
597 par
->matchinfo
= m
->data
;
599 ret
= xt_check_match(par
, m
->u
.match_size
- sizeof(*m
),
600 ip
->proto
, ip
->invflags
& IPT_INV_PROTO
);
602 duprintf("check failed for `%s'.\n", par
->match
->name
);
609 find_check_match(struct xt_entry_match
*m
, struct xt_mtchk_param
*par
)
611 struct xt_match
*match
;
614 match
= xt_request_find_match(NFPROTO_IPV4
, m
->u
.user
.name
,
617 duprintf("find_check_match: `%s' not found\n", m
->u
.user
.name
);
618 return PTR_ERR(match
);
620 m
->u
.kernel
.match
= match
;
622 ret
= check_match(m
, par
);
628 module_put(m
->u
.kernel
.match
->me
);
632 static int check_target(struct ipt_entry
*e
, struct net
*net
, const char *name
)
634 struct xt_entry_target
*t
= ipt_get_target(e
);
635 struct xt_tgchk_param par
= {
639 .target
= t
->u
.kernel
.target
,
641 .hook_mask
= e
->comefrom
,
642 .family
= NFPROTO_IPV4
,
646 ret
= xt_check_target(&par
, t
->u
.target_size
- sizeof(*t
),
647 e
->ip
.proto
, e
->ip
.invflags
& IPT_INV_PROTO
);
649 duprintf("check failed for `%s'.\n",
650 t
->u
.kernel
.target
->name
);
657 find_check_entry(struct ipt_entry
*e
, struct net
*net
, const char *name
,
660 struct xt_entry_target
*t
;
661 struct xt_target
*target
;
664 struct xt_mtchk_param mtpar
;
665 struct xt_entry_match
*ematch
;
667 e
->counters
.pcnt
= xt_percpu_counter_alloc();
668 if (IS_ERR_VALUE(e
->counters
.pcnt
))
674 mtpar
.entryinfo
= &e
->ip
;
675 mtpar
.hook_mask
= e
->comefrom
;
676 mtpar
.family
= NFPROTO_IPV4
;
677 xt_ematch_foreach(ematch
, e
) {
678 ret
= find_check_match(ematch
, &mtpar
);
680 goto cleanup_matches
;
684 t
= ipt_get_target(e
);
685 target
= xt_request_find_target(NFPROTO_IPV4
, t
->u
.user
.name
,
687 if (IS_ERR(target
)) {
688 duprintf("find_check_entry: `%s' not found\n", t
->u
.user
.name
);
689 ret
= PTR_ERR(target
);
690 goto cleanup_matches
;
692 t
->u
.kernel
.target
= target
;
694 ret
= check_target(e
, net
, name
);
700 module_put(t
->u
.kernel
.target
->me
);
702 xt_ematch_foreach(ematch
, e
) {
705 cleanup_match(ematch
, net
);
708 xt_percpu_counter_free(e
->counters
.pcnt
);
713 static bool check_underflow(const struct ipt_entry
*e
)
715 const struct xt_entry_target
*t
;
716 unsigned int verdict
;
718 if (!unconditional(e
))
720 t
= ipt_get_target_c(e
);
721 if (strcmp(t
->u
.user
.name
, XT_STANDARD_TARGET
) != 0)
723 verdict
= ((struct xt_standard_target
*)t
)->verdict
;
724 verdict
= -verdict
- 1;
725 return verdict
== NF_DROP
|| verdict
== NF_ACCEPT
;
729 check_entry_size_and_hooks(struct ipt_entry
*e
,
730 struct xt_table_info
*newinfo
,
731 const unsigned char *base
,
732 const unsigned char *limit
,
733 const unsigned int *hook_entries
,
734 const unsigned int *underflows
,
735 unsigned int valid_hooks
)
740 if ((unsigned long)e
% __alignof__(struct ipt_entry
) != 0 ||
741 (unsigned char *)e
+ sizeof(struct ipt_entry
) >= limit
||
742 (unsigned char *)e
+ e
->next_offset
> limit
) {
743 duprintf("Bad offset %p\n", e
);
748 < sizeof(struct ipt_entry
) + sizeof(struct xt_entry_target
)) {
749 duprintf("checking: element %p size %u\n",
754 if (!ip_checkentry(&e
->ip
))
757 err
= xt_check_entry_offsets(e
, e
->elems
, e
->target_offset
,
762 /* Check hooks & underflows */
763 for (h
= 0; h
< NF_INET_NUMHOOKS
; h
++) {
764 if (!(valid_hooks
& (1 << h
)))
766 if ((unsigned char *)e
- base
== hook_entries
[h
])
767 newinfo
->hook_entry
[h
] = hook_entries
[h
];
768 if ((unsigned char *)e
- base
== underflows
[h
]) {
769 if (!check_underflow(e
)) {
770 pr_debug("Underflows must be unconditional and "
771 "use the STANDARD target with "
775 newinfo
->underflow
[h
] = underflows
[h
];
779 /* Clear counters and comefrom */
780 e
->counters
= ((struct xt_counters
) { 0, 0 });
786 cleanup_entry(struct ipt_entry
*e
, struct net
*net
)
788 struct xt_tgdtor_param par
;
789 struct xt_entry_target
*t
;
790 struct xt_entry_match
*ematch
;
792 /* Cleanup all matches */
793 xt_ematch_foreach(ematch
, e
)
794 cleanup_match(ematch
, net
);
795 t
= ipt_get_target(e
);
798 par
.target
= t
->u
.kernel
.target
;
799 par
.targinfo
= t
->data
;
800 par
.family
= NFPROTO_IPV4
;
801 if (par
.target
->destroy
!= NULL
)
802 par
.target
->destroy(&par
);
803 module_put(par
.target
->me
);
804 xt_percpu_counter_free(e
->counters
.pcnt
);
807 /* Checks and translates the user-supplied table segment (held in
810 translate_table(struct net
*net
, struct xt_table_info
*newinfo
, void *entry0
,
811 const struct ipt_replace
*repl
)
813 struct ipt_entry
*iter
;
817 newinfo
->size
= repl
->size
;
818 newinfo
->number
= repl
->num_entries
;
820 /* Init all hooks to impossible value. */
821 for (i
= 0; i
< NF_INET_NUMHOOKS
; i
++) {
822 newinfo
->hook_entry
[i
] = 0xFFFFFFFF;
823 newinfo
->underflow
[i
] = 0xFFFFFFFF;
826 duprintf("translate_table: size %u\n", newinfo
->size
);
828 /* Walk through entries, checking offsets. */
829 xt_entry_foreach(iter
, entry0
, newinfo
->size
) {
830 ret
= check_entry_size_and_hooks(iter
, newinfo
, entry0
,
838 if (strcmp(ipt_get_target(iter
)->u
.user
.name
,
839 XT_ERROR_TARGET
) == 0)
840 ++newinfo
->stacksize
;
843 if (i
!= repl
->num_entries
) {
844 duprintf("translate_table: %u not %u entries\n",
845 i
, repl
->num_entries
);
849 /* Check hooks all assigned */
850 for (i
= 0; i
< NF_INET_NUMHOOKS
; i
++) {
851 /* Only hooks which are valid */
852 if (!(repl
->valid_hooks
& (1 << i
)))
854 if (newinfo
->hook_entry
[i
] == 0xFFFFFFFF) {
855 duprintf("Invalid hook entry %u %u\n",
856 i
, repl
->hook_entry
[i
]);
859 if (newinfo
->underflow
[i
] == 0xFFFFFFFF) {
860 duprintf("Invalid underflow %u %u\n",
861 i
, repl
->underflow
[i
]);
866 if (!mark_source_chains(newinfo
, repl
->valid_hooks
, entry0
))
869 /* Finally, each sanity check must pass */
871 xt_entry_foreach(iter
, entry0
, newinfo
->size
) {
872 ret
= find_check_entry(iter
, net
, repl
->name
, repl
->size
);
879 xt_entry_foreach(iter
, entry0
, newinfo
->size
) {
882 cleanup_entry(iter
, net
);
891 get_counters(const struct xt_table_info
*t
,
892 struct xt_counters counters
[])
894 struct ipt_entry
*iter
;
898 for_each_possible_cpu(cpu
) {
899 seqcount_t
*s
= &per_cpu(xt_recseq
, cpu
);
902 xt_entry_foreach(iter
, t
->entries
, t
->size
) {
903 struct xt_counters
*tmp
;
907 tmp
= xt_get_per_cpu_counter(&iter
->counters
, cpu
);
909 start
= read_seqcount_begin(s
);
912 } while (read_seqcount_retry(s
, start
));
914 ADD_COUNTER(counters
[i
], bcnt
, pcnt
);
915 ++i
; /* macro does multi eval of i */
920 static struct xt_counters
*alloc_counters(const struct xt_table
*table
)
922 unsigned int countersize
;
923 struct xt_counters
*counters
;
924 const struct xt_table_info
*private = table
->private;
926 /* We need atomic snapshot of counters: rest doesn't change
927 (other than comefrom, which userspace doesn't care
929 countersize
= sizeof(struct xt_counters
) * private->number
;
930 counters
= vzalloc(countersize
);
932 if (counters
== NULL
)
933 return ERR_PTR(-ENOMEM
);
935 get_counters(private, counters
);
941 copy_entries_to_user(unsigned int total_size
,
942 const struct xt_table
*table
,
943 void __user
*userptr
)
945 unsigned int off
, num
;
946 const struct ipt_entry
*e
;
947 struct xt_counters
*counters
;
948 const struct xt_table_info
*private = table
->private;
950 const void *loc_cpu_entry
;
952 counters
= alloc_counters(table
);
953 if (IS_ERR(counters
))
954 return PTR_ERR(counters
);
956 loc_cpu_entry
= private->entries
;
957 if (copy_to_user(userptr
, loc_cpu_entry
, total_size
) != 0) {
962 /* FIXME: use iterator macros --RR */
963 /* ... then go back and fix counters and names */
964 for (off
= 0, num
= 0; off
< total_size
; off
+= e
->next_offset
, num
++){
966 const struct xt_entry_match
*m
;
967 const struct xt_entry_target
*t
;
969 e
= (struct ipt_entry
*)(loc_cpu_entry
+ off
);
970 if (copy_to_user(userptr
+ off
971 + offsetof(struct ipt_entry
, counters
),
973 sizeof(counters
[num
])) != 0) {
978 for (i
= sizeof(struct ipt_entry
);
979 i
< e
->target_offset
;
980 i
+= m
->u
.match_size
) {
983 if (copy_to_user(userptr
+ off
+ i
984 + offsetof(struct xt_entry_match
,
986 m
->u
.kernel
.match
->name
,
987 strlen(m
->u
.kernel
.match
->name
)+1)
994 t
= ipt_get_target_c(e
);
995 if (copy_to_user(userptr
+ off
+ e
->target_offset
996 + offsetof(struct xt_entry_target
,
998 t
->u
.kernel
.target
->name
,
999 strlen(t
->u
.kernel
.target
->name
)+1) != 0) {
1010 #ifdef CONFIG_COMPAT
1011 static void compat_standard_from_user(void *dst
, const void *src
)
1013 int v
= *(compat_int_t
*)src
;
1016 v
+= xt_compat_calc_jump(AF_INET
, v
);
1017 memcpy(dst
, &v
, sizeof(v
));
1020 static int compat_standard_to_user(void __user
*dst
, const void *src
)
1022 compat_int_t cv
= *(int *)src
;
1025 cv
-= xt_compat_calc_jump(AF_INET
, cv
);
1026 return copy_to_user(dst
, &cv
, sizeof(cv
)) ? -EFAULT
: 0;
1029 static int compat_calc_entry(const struct ipt_entry
*e
,
1030 const struct xt_table_info
*info
,
1031 const void *base
, struct xt_table_info
*newinfo
)
1033 const struct xt_entry_match
*ematch
;
1034 const struct xt_entry_target
*t
;
1035 unsigned int entry_offset
;
1038 off
= sizeof(struct ipt_entry
) - sizeof(struct compat_ipt_entry
);
1039 entry_offset
= (void *)e
- base
;
1040 xt_ematch_foreach(ematch
, e
)
1041 off
+= xt_compat_match_offset(ematch
->u
.kernel
.match
);
1042 t
= ipt_get_target_c(e
);
1043 off
+= xt_compat_target_offset(t
->u
.kernel
.target
);
1044 newinfo
->size
-= off
;
1045 ret
= xt_compat_add_offset(AF_INET
, entry_offset
, off
);
1049 for (i
= 0; i
< NF_INET_NUMHOOKS
; i
++) {
1050 if (info
->hook_entry
[i
] &&
1051 (e
< (struct ipt_entry
*)(base
+ info
->hook_entry
[i
])))
1052 newinfo
->hook_entry
[i
] -= off
;
1053 if (info
->underflow
[i
] &&
1054 (e
< (struct ipt_entry
*)(base
+ info
->underflow
[i
])))
1055 newinfo
->underflow
[i
] -= off
;
1060 static int compat_table_info(const struct xt_table_info
*info
,
1061 struct xt_table_info
*newinfo
)
1063 struct ipt_entry
*iter
;
1064 const void *loc_cpu_entry
;
1067 if (!newinfo
|| !info
)
1070 /* we dont care about newinfo->entries */
1071 memcpy(newinfo
, info
, offsetof(struct xt_table_info
, entries
));
1072 newinfo
->initial_entries
= 0;
1073 loc_cpu_entry
= info
->entries
;
1074 xt_compat_init_offsets(AF_INET
, info
->number
);
1075 xt_entry_foreach(iter
, loc_cpu_entry
, info
->size
) {
1076 ret
= compat_calc_entry(iter
, info
, loc_cpu_entry
, newinfo
);
1084 static int get_info(struct net
*net
, void __user
*user
,
1085 const int *len
, int compat
)
1087 char name
[XT_TABLE_MAXNAMELEN
];
1091 if (*len
!= sizeof(struct ipt_getinfo
)) {
1092 duprintf("length %u != %zu\n", *len
,
1093 sizeof(struct ipt_getinfo
));
1097 if (copy_from_user(name
, user
, sizeof(name
)) != 0)
1100 name
[XT_TABLE_MAXNAMELEN
-1] = '\0';
1101 #ifdef CONFIG_COMPAT
1103 xt_compat_lock(AF_INET
);
1105 t
= try_then_request_module(xt_find_table_lock(net
, AF_INET
, name
),
1106 "iptable_%s", name
);
1107 if (!IS_ERR_OR_NULL(t
)) {
1108 struct ipt_getinfo info
;
1109 const struct xt_table_info
*private = t
->private;
1110 #ifdef CONFIG_COMPAT
1111 struct xt_table_info tmp
;
1114 ret
= compat_table_info(private, &tmp
);
1115 xt_compat_flush_offsets(AF_INET
);
1119 memset(&info
, 0, sizeof(info
));
1120 info
.valid_hooks
= t
->valid_hooks
;
1121 memcpy(info
.hook_entry
, private->hook_entry
,
1122 sizeof(info
.hook_entry
));
1123 memcpy(info
.underflow
, private->underflow
,
1124 sizeof(info
.underflow
));
1125 info
.num_entries
= private->number
;
1126 info
.size
= private->size
;
1127 strcpy(info
.name
, name
);
1129 if (copy_to_user(user
, &info
, *len
) != 0)
1137 ret
= t
? PTR_ERR(t
) : -ENOENT
;
1138 #ifdef CONFIG_COMPAT
1140 xt_compat_unlock(AF_INET
);
1146 get_entries(struct net
*net
, struct ipt_get_entries __user
*uptr
,
1150 struct ipt_get_entries get
;
1153 if (*len
< sizeof(get
)) {
1154 duprintf("get_entries: %u < %zu\n", *len
, sizeof(get
));
1157 if (copy_from_user(&get
, uptr
, sizeof(get
)) != 0)
1159 if (*len
!= sizeof(struct ipt_get_entries
) + get
.size
) {
1160 duprintf("get_entries: %u != %zu\n",
1161 *len
, sizeof(get
) + get
.size
);
1164 get
.name
[sizeof(get
.name
) - 1] = '\0';
1166 t
= xt_find_table_lock(net
, AF_INET
, get
.name
);
1167 if (!IS_ERR_OR_NULL(t
)) {
1168 const struct xt_table_info
*private = t
->private;
1169 duprintf("t->private->number = %u\n", private->number
);
1170 if (get
.size
== private->size
)
1171 ret
= copy_entries_to_user(private->size
,
1172 t
, uptr
->entrytable
);
1174 duprintf("get_entries: I've got %u not %u!\n",
1175 private->size
, get
.size
);
1181 ret
= t
? PTR_ERR(t
) : -ENOENT
;
1187 __do_replace(struct net
*net
, const char *name
, unsigned int valid_hooks
,
1188 struct xt_table_info
*newinfo
, unsigned int num_counters
,
1189 void __user
*counters_ptr
)
1193 struct xt_table_info
*oldinfo
;
1194 struct xt_counters
*counters
;
1195 struct ipt_entry
*iter
;
1198 counters
= vzalloc(num_counters
* sizeof(struct xt_counters
));
1204 t
= try_then_request_module(xt_find_table_lock(net
, AF_INET
, name
),
1205 "iptable_%s", name
);
1206 if (IS_ERR_OR_NULL(t
)) {
1207 ret
= t
? PTR_ERR(t
) : -ENOENT
;
1208 goto free_newinfo_counters_untrans
;
1212 if (valid_hooks
!= t
->valid_hooks
) {
1213 duprintf("Valid hook crap: %08X vs %08X\n",
1214 valid_hooks
, t
->valid_hooks
);
1219 oldinfo
= xt_replace_table(t
, num_counters
, newinfo
, &ret
);
1223 /* Update module usage count based on number of rules */
1224 duprintf("do_replace: oldnum=%u, initnum=%u, newnum=%u\n",
1225 oldinfo
->number
, oldinfo
->initial_entries
, newinfo
->number
);
1226 if ((oldinfo
->number
> oldinfo
->initial_entries
) ||
1227 (newinfo
->number
<= oldinfo
->initial_entries
))
1229 if ((oldinfo
->number
> oldinfo
->initial_entries
) &&
1230 (newinfo
->number
<= oldinfo
->initial_entries
))
1233 /* Get the old counters, and synchronize with replace */
1234 get_counters(oldinfo
, counters
);
1236 /* Decrease module usage counts and free resource */
1237 xt_entry_foreach(iter
, oldinfo
->entries
, oldinfo
->size
)
1238 cleanup_entry(iter
, net
);
1240 xt_free_table_info(oldinfo
);
1241 if (copy_to_user(counters_ptr
, counters
,
1242 sizeof(struct xt_counters
) * num_counters
) != 0) {
1243 /* Silent error, can't fail, new table is already in place */
1244 net_warn_ratelimited("iptables: counters copy to user failed while replacing table\n");
1253 free_newinfo_counters_untrans
:
1260 do_replace(struct net
*net
, const void __user
*user
, unsigned int len
)
1263 struct ipt_replace tmp
;
1264 struct xt_table_info
*newinfo
;
1265 void *loc_cpu_entry
;
1266 struct ipt_entry
*iter
;
1268 if (copy_from_user(&tmp
, user
, sizeof(tmp
)) != 0)
1271 /* overflow check */
1272 if (tmp
.num_counters
>= INT_MAX
/ sizeof(struct xt_counters
))
1274 if (tmp
.num_counters
== 0)
1277 tmp
.name
[sizeof(tmp
.name
)-1] = 0;
1279 newinfo
= xt_alloc_table_info(tmp
.size
);
1283 loc_cpu_entry
= newinfo
->entries
;
1284 if (copy_from_user(loc_cpu_entry
, user
+ sizeof(tmp
),
1290 ret
= translate_table(net
, newinfo
, loc_cpu_entry
, &tmp
);
1294 duprintf("Translated table\n");
1296 ret
= __do_replace(net
, tmp
.name
, tmp
.valid_hooks
, newinfo
,
1297 tmp
.num_counters
, tmp
.counters
);
1299 goto free_newinfo_untrans
;
1302 free_newinfo_untrans
:
1303 xt_entry_foreach(iter
, loc_cpu_entry
, newinfo
->size
)
1304 cleanup_entry(iter
, net
);
1306 xt_free_table_info(newinfo
);
1311 do_add_counters(struct net
*net
, const void __user
*user
,
1312 unsigned int len
, int compat
)
1315 struct xt_counters_info tmp
;
1316 struct xt_counters
*paddc
;
1317 unsigned int num_counters
;
1322 const struct xt_table_info
*private;
1324 struct ipt_entry
*iter
;
1325 unsigned int addend
;
1326 #ifdef CONFIG_COMPAT
1327 struct compat_xt_counters_info compat_tmp
;
1331 size
= sizeof(struct compat_xt_counters_info
);
1336 size
= sizeof(struct xt_counters_info
);
1339 if (copy_from_user(ptmp
, user
, size
) != 0)
1342 #ifdef CONFIG_COMPAT
1344 num_counters
= compat_tmp
.num_counters
;
1345 name
= compat_tmp
.name
;
1349 num_counters
= tmp
.num_counters
;
1353 if (len
!= size
+ num_counters
* sizeof(struct xt_counters
))
1356 paddc
= vmalloc(len
- size
);
1360 if (copy_from_user(paddc
, user
+ size
, len
- size
) != 0) {
1365 t
= xt_find_table_lock(net
, AF_INET
, name
);
1366 if (IS_ERR_OR_NULL(t
)) {
1367 ret
= t
? PTR_ERR(t
) : -ENOENT
;
1372 private = t
->private;
1373 if (private->number
!= num_counters
) {
1375 goto unlock_up_free
;
1379 addend
= xt_write_recseq_begin();
1380 xt_entry_foreach(iter
, private->entries
, private->size
) {
1381 struct xt_counters
*tmp
;
1383 tmp
= xt_get_this_cpu_counter(&iter
->counters
);
1384 ADD_COUNTER(*tmp
, paddc
[i
].bcnt
, paddc
[i
].pcnt
);
1387 xt_write_recseq_end(addend
);
1398 #ifdef CONFIG_COMPAT
1399 struct compat_ipt_replace
{
1400 char name
[XT_TABLE_MAXNAMELEN
];
1404 u32 hook_entry
[NF_INET_NUMHOOKS
];
1405 u32 underflow
[NF_INET_NUMHOOKS
];
1407 compat_uptr_t counters
; /* struct xt_counters * */
1408 struct compat_ipt_entry entries
[0];
1412 compat_copy_entry_to_user(struct ipt_entry
*e
, void __user
**dstptr
,
1413 unsigned int *size
, struct xt_counters
*counters
,
1416 struct xt_entry_target
*t
;
1417 struct compat_ipt_entry __user
*ce
;
1418 u_int16_t target_offset
, next_offset
;
1419 compat_uint_t origsize
;
1420 const struct xt_entry_match
*ematch
;
1424 ce
= (struct compat_ipt_entry __user
*)*dstptr
;
1425 if (copy_to_user(ce
, e
, sizeof(struct ipt_entry
)) != 0 ||
1426 copy_to_user(&ce
->counters
, &counters
[i
],
1427 sizeof(counters
[i
])) != 0)
1430 *dstptr
+= sizeof(struct compat_ipt_entry
);
1431 *size
-= sizeof(struct ipt_entry
) - sizeof(struct compat_ipt_entry
);
1433 xt_ematch_foreach(ematch
, e
) {
1434 ret
= xt_compat_match_to_user(ematch
, dstptr
, size
);
1438 target_offset
= e
->target_offset
- (origsize
- *size
);
1439 t
= ipt_get_target(e
);
1440 ret
= xt_compat_target_to_user(t
, dstptr
, size
);
1443 next_offset
= e
->next_offset
- (origsize
- *size
);
1444 if (put_user(target_offset
, &ce
->target_offset
) != 0 ||
1445 put_user(next_offset
, &ce
->next_offset
) != 0)
1451 compat_find_calc_match(struct xt_entry_match
*m
,
1453 const struct ipt_ip
*ip
,
1456 struct xt_match
*match
;
1458 match
= xt_request_find_match(NFPROTO_IPV4
, m
->u
.user
.name
,
1459 m
->u
.user
.revision
);
1460 if (IS_ERR(match
)) {
1461 duprintf("compat_check_calc_match: `%s' not found\n",
1463 return PTR_ERR(match
);
1465 m
->u
.kernel
.match
= match
;
1466 *size
+= xt_compat_match_offset(match
);
1470 static void compat_release_entry(struct compat_ipt_entry
*e
)
1472 struct xt_entry_target
*t
;
1473 struct xt_entry_match
*ematch
;
1475 /* Cleanup all matches */
1476 xt_ematch_foreach(ematch
, e
)
1477 module_put(ematch
->u
.kernel
.match
->me
);
1478 t
= compat_ipt_get_target(e
);
1479 module_put(t
->u
.kernel
.target
->me
);
1483 check_compat_entry_size_and_hooks(struct compat_ipt_entry
*e
,
1484 struct xt_table_info
*newinfo
,
1486 const unsigned char *base
,
1487 const unsigned char *limit
,
1488 const unsigned int *hook_entries
,
1489 const unsigned int *underflows
,
1492 struct xt_entry_match
*ematch
;
1493 struct xt_entry_target
*t
;
1494 struct xt_target
*target
;
1495 unsigned int entry_offset
;
1499 duprintf("check_compat_entry_size_and_hooks %p\n", e
);
1500 if ((unsigned long)e
% __alignof__(struct compat_ipt_entry
) != 0 ||
1501 (unsigned char *)e
+ sizeof(struct compat_ipt_entry
) >= limit
||
1502 (unsigned char *)e
+ e
->next_offset
> limit
) {
1503 duprintf("Bad offset %p, limit = %p\n", e
, limit
);
1507 if (e
->next_offset
< sizeof(struct compat_ipt_entry
) +
1508 sizeof(struct compat_xt_entry_target
)) {
1509 duprintf("checking: element %p size %u\n",
1514 if (!ip_checkentry(&e
->ip
))
1517 ret
= xt_compat_check_entry_offsets(e
, e
->elems
,
1518 e
->target_offset
, e
->next_offset
);
1522 off
= sizeof(struct ipt_entry
) - sizeof(struct compat_ipt_entry
);
1523 entry_offset
= (void *)e
- (void *)base
;
1525 xt_ematch_foreach(ematch
, e
) {
1526 ret
= compat_find_calc_match(ematch
, name
, &e
->ip
, &off
);
1528 goto release_matches
;
1532 t
= compat_ipt_get_target(e
);
1533 target
= xt_request_find_target(NFPROTO_IPV4
, t
->u
.user
.name
,
1534 t
->u
.user
.revision
);
1535 if (IS_ERR(target
)) {
1536 duprintf("check_compat_entry_size_and_hooks: `%s' not found\n",
1538 ret
= PTR_ERR(target
);
1539 goto release_matches
;
1541 t
->u
.kernel
.target
= target
;
1543 off
+= xt_compat_target_offset(target
);
1545 ret
= xt_compat_add_offset(AF_INET
, entry_offset
, off
);
1549 /* Check hooks & underflows */
1550 for (h
= 0; h
< NF_INET_NUMHOOKS
; h
++) {
1551 if ((unsigned char *)e
- base
== hook_entries
[h
])
1552 newinfo
->hook_entry
[h
] = hook_entries
[h
];
1553 if ((unsigned char *)e
- base
== underflows
[h
])
1554 newinfo
->underflow
[h
] = underflows
[h
];
1557 /* Clear counters and comefrom */
1558 memset(&e
->counters
, 0, sizeof(e
->counters
));
1563 module_put(t
->u
.kernel
.target
->me
);
1565 xt_ematch_foreach(ematch
, e
) {
1568 module_put(ematch
->u
.kernel
.match
->me
);
1574 compat_copy_entry_from_user(struct compat_ipt_entry
*e
, void **dstptr
,
1575 unsigned int *size
, const char *name
,
1576 struct xt_table_info
*newinfo
, unsigned char *base
)
1578 struct xt_entry_target
*t
;
1579 struct xt_target
*target
;
1580 struct ipt_entry
*de
;
1581 unsigned int origsize
;
1583 struct xt_entry_match
*ematch
;
1587 de
= (struct ipt_entry
*)*dstptr
;
1588 memcpy(de
, e
, sizeof(struct ipt_entry
));
1589 memcpy(&de
->counters
, &e
->counters
, sizeof(e
->counters
));
1591 *dstptr
+= sizeof(struct ipt_entry
);
1592 *size
+= sizeof(struct ipt_entry
) - sizeof(struct compat_ipt_entry
);
1594 xt_ematch_foreach(ematch
, e
) {
1595 ret
= xt_compat_match_from_user(ematch
, dstptr
, size
);
1599 de
->target_offset
= e
->target_offset
- (origsize
- *size
);
1600 t
= compat_ipt_get_target(e
);
1601 target
= t
->u
.kernel
.target
;
1602 xt_compat_target_from_user(t
, dstptr
, size
);
1604 de
->next_offset
= e
->next_offset
- (origsize
- *size
);
1605 for (h
= 0; h
< NF_INET_NUMHOOKS
; h
++) {
1606 if ((unsigned char *)de
- base
< newinfo
->hook_entry
[h
])
1607 newinfo
->hook_entry
[h
] -= origsize
- *size
;
1608 if ((unsigned char *)de
- base
< newinfo
->underflow
[h
])
1609 newinfo
->underflow
[h
] -= origsize
- *size
;
1615 compat_check_entry(struct ipt_entry
*e
, struct net
*net
, const char *name
)
1617 struct xt_entry_match
*ematch
;
1618 struct xt_mtchk_param mtpar
;
1622 e
->counters
.pcnt
= xt_percpu_counter_alloc();
1623 if (IS_ERR_VALUE(e
->counters
.pcnt
))
1629 mtpar
.entryinfo
= &e
->ip
;
1630 mtpar
.hook_mask
= e
->comefrom
;
1631 mtpar
.family
= NFPROTO_IPV4
;
1632 xt_ematch_foreach(ematch
, e
) {
1633 ret
= check_match(ematch
, &mtpar
);
1635 goto cleanup_matches
;
1639 ret
= check_target(e
, net
, name
);
1641 goto cleanup_matches
;
1645 xt_ematch_foreach(ematch
, e
) {
1648 cleanup_match(ematch
, net
);
1651 xt_percpu_counter_free(e
->counters
.pcnt
);
1657 translate_compat_table(struct net
*net
,
1659 unsigned int valid_hooks
,
1660 struct xt_table_info
**pinfo
,
1662 unsigned int total_size
,
1663 unsigned int number
,
1664 unsigned int *hook_entries
,
1665 unsigned int *underflows
)
1668 struct xt_table_info
*newinfo
, *info
;
1669 void *pos
, *entry0
, *entry1
;
1670 struct compat_ipt_entry
*iter0
;
1671 struct ipt_entry
*iter1
;
1678 info
->number
= number
;
1680 /* Init all hooks to impossible value. */
1681 for (i
= 0; i
< NF_INET_NUMHOOKS
; i
++) {
1682 info
->hook_entry
[i
] = 0xFFFFFFFF;
1683 info
->underflow
[i
] = 0xFFFFFFFF;
1686 duprintf("translate_compat_table: size %u\n", info
->size
);
1688 xt_compat_lock(AF_INET
);
1689 xt_compat_init_offsets(AF_INET
, number
);
1690 /* Walk through entries, checking offsets. */
1691 xt_entry_foreach(iter0
, entry0
, total_size
) {
1692 ret
= check_compat_entry_size_and_hooks(iter0
, info
, &size
,
1694 entry0
+ total_size
,
1705 duprintf("translate_compat_table: %u not %u entries\n",
1710 /* Check hooks all assigned */
1711 for (i
= 0; i
< NF_INET_NUMHOOKS
; i
++) {
1712 /* Only hooks which are valid */
1713 if (!(valid_hooks
& (1 << i
)))
1715 if (info
->hook_entry
[i
] == 0xFFFFFFFF) {
1716 duprintf("Invalid hook entry %u %u\n",
1717 i
, hook_entries
[i
]);
1720 if (info
->underflow
[i
] == 0xFFFFFFFF) {
1721 duprintf("Invalid underflow %u %u\n",
1728 newinfo
= xt_alloc_table_info(size
);
1732 newinfo
->number
= number
;
1733 for (i
= 0; i
< NF_INET_NUMHOOKS
; i
++) {
1734 newinfo
->hook_entry
[i
] = info
->hook_entry
[i
];
1735 newinfo
->underflow
[i
] = info
->underflow
[i
];
1737 entry1
= newinfo
->entries
;
1740 xt_entry_foreach(iter0
, entry0
, total_size
) {
1741 ret
= compat_copy_entry_from_user(iter0
, &pos
, &size
,
1742 name
, newinfo
, entry1
);
1746 xt_compat_flush_offsets(AF_INET
);
1747 xt_compat_unlock(AF_INET
);
1752 if (!mark_source_chains(newinfo
, valid_hooks
, entry1
))
1756 xt_entry_foreach(iter1
, entry1
, newinfo
->size
) {
1757 ret
= compat_check_entry(iter1
, net
, name
);
1761 if (strcmp(ipt_get_target(iter1
)->u
.user
.name
,
1762 XT_ERROR_TARGET
) == 0)
1763 ++newinfo
->stacksize
;
1767 * The first i matches need cleanup_entry (calls ->destroy)
1768 * because they had called ->check already. The other j-i
1769 * entries need only release.
1773 xt_entry_foreach(iter0
, entry0
, newinfo
->size
) {
1778 compat_release_entry(iter0
);
1780 xt_entry_foreach(iter1
, entry1
, newinfo
->size
) {
1783 cleanup_entry(iter1
, net
);
1785 xt_free_table_info(newinfo
);
1791 xt_free_table_info(info
);
1795 xt_free_table_info(newinfo
);
1797 xt_entry_foreach(iter0
, entry0
, total_size
) {
1800 compat_release_entry(iter0
);
1804 xt_compat_flush_offsets(AF_INET
);
1805 xt_compat_unlock(AF_INET
);
1810 compat_do_replace(struct net
*net
, void __user
*user
, unsigned int len
)
1813 struct compat_ipt_replace tmp
;
1814 struct xt_table_info
*newinfo
;
1815 void *loc_cpu_entry
;
1816 struct ipt_entry
*iter
;
1818 if (copy_from_user(&tmp
, user
, sizeof(tmp
)) != 0)
1821 /* overflow check */
1822 if (tmp
.size
>= INT_MAX
/ num_possible_cpus())
1824 if (tmp
.num_counters
>= INT_MAX
/ sizeof(struct xt_counters
))
1826 if (tmp
.num_counters
== 0)
1829 tmp
.name
[sizeof(tmp
.name
)-1] = 0;
1831 newinfo
= xt_alloc_table_info(tmp
.size
);
1835 loc_cpu_entry
= newinfo
->entries
;
1836 if (copy_from_user(loc_cpu_entry
, user
+ sizeof(tmp
),
1842 ret
= translate_compat_table(net
, tmp
.name
, tmp
.valid_hooks
,
1843 &newinfo
, &loc_cpu_entry
, tmp
.size
,
1844 tmp
.num_entries
, tmp
.hook_entry
,
1849 duprintf("compat_do_replace: Translated table\n");
1851 ret
= __do_replace(net
, tmp
.name
, tmp
.valid_hooks
, newinfo
,
1852 tmp
.num_counters
, compat_ptr(tmp
.counters
));
1854 goto free_newinfo_untrans
;
1857 free_newinfo_untrans
:
1858 xt_entry_foreach(iter
, loc_cpu_entry
, newinfo
->size
)
1859 cleanup_entry(iter
, net
);
1861 xt_free_table_info(newinfo
);
1866 compat_do_ipt_set_ctl(struct sock
*sk
, int cmd
, void __user
*user
,
1871 if (!ns_capable(sock_net(sk
)->user_ns
, CAP_NET_ADMIN
))
1875 case IPT_SO_SET_REPLACE
:
1876 ret
= compat_do_replace(sock_net(sk
), user
, len
);
1879 case IPT_SO_SET_ADD_COUNTERS
:
1880 ret
= do_add_counters(sock_net(sk
), user
, len
, 1);
1884 duprintf("do_ipt_set_ctl: unknown request %i\n", cmd
);
1891 struct compat_ipt_get_entries
{
1892 char name
[XT_TABLE_MAXNAMELEN
];
1894 struct compat_ipt_entry entrytable
[0];
1898 compat_copy_entries_to_user(unsigned int total_size
, struct xt_table
*table
,
1899 void __user
*userptr
)
1901 struct xt_counters
*counters
;
1902 const struct xt_table_info
*private = table
->private;
1907 struct ipt_entry
*iter
;
1909 counters
= alloc_counters(table
);
1910 if (IS_ERR(counters
))
1911 return PTR_ERR(counters
);
1915 xt_entry_foreach(iter
, private->entries
, total_size
) {
1916 ret
= compat_copy_entry_to_user(iter
, &pos
,
1917 &size
, counters
, i
++);
1927 compat_get_entries(struct net
*net
, struct compat_ipt_get_entries __user
*uptr
,
1931 struct compat_ipt_get_entries get
;
1934 if (*len
< sizeof(get
)) {
1935 duprintf("compat_get_entries: %u < %zu\n", *len
, sizeof(get
));
1939 if (copy_from_user(&get
, uptr
, sizeof(get
)) != 0)
1942 if (*len
!= sizeof(struct compat_ipt_get_entries
) + get
.size
) {
1943 duprintf("compat_get_entries: %u != %zu\n",
1944 *len
, sizeof(get
) + get
.size
);
1947 get
.name
[sizeof(get
.name
) - 1] = '\0';
1949 xt_compat_lock(AF_INET
);
1950 t
= xt_find_table_lock(net
, AF_INET
, get
.name
);
1951 if (!IS_ERR_OR_NULL(t
)) {
1952 const struct xt_table_info
*private = t
->private;
1953 struct xt_table_info info
;
1954 duprintf("t->private->number = %u\n", private->number
);
1955 ret
= compat_table_info(private, &info
);
1956 if (!ret
&& get
.size
== info
.size
) {
1957 ret
= compat_copy_entries_to_user(private->size
,
1958 t
, uptr
->entrytable
);
1960 duprintf("compat_get_entries: I've got %u not %u!\n",
1961 private->size
, get
.size
);
1964 xt_compat_flush_offsets(AF_INET
);
1968 ret
= t
? PTR_ERR(t
) : -ENOENT
;
1970 xt_compat_unlock(AF_INET
);
1974 static int do_ipt_get_ctl(struct sock
*, int, void __user
*, int *);
1977 compat_do_ipt_get_ctl(struct sock
*sk
, int cmd
, void __user
*user
, int *len
)
1981 if (!ns_capable(sock_net(sk
)->user_ns
, CAP_NET_ADMIN
))
1985 case IPT_SO_GET_INFO
:
1986 ret
= get_info(sock_net(sk
), user
, len
, 1);
1988 case IPT_SO_GET_ENTRIES
:
1989 ret
= compat_get_entries(sock_net(sk
), user
, len
);
1992 ret
= do_ipt_get_ctl(sk
, cmd
, user
, len
);
1999 do_ipt_set_ctl(struct sock
*sk
, int cmd
, void __user
*user
, unsigned int len
)
2003 if (!ns_capable(sock_net(sk
)->user_ns
, CAP_NET_ADMIN
))
2007 case IPT_SO_SET_REPLACE
:
2008 ret
= do_replace(sock_net(sk
), user
, len
);
2011 case IPT_SO_SET_ADD_COUNTERS
:
2012 ret
= do_add_counters(sock_net(sk
), user
, len
, 0);
2016 duprintf("do_ipt_set_ctl: unknown request %i\n", cmd
);
2024 do_ipt_get_ctl(struct sock
*sk
, int cmd
, void __user
*user
, int *len
)
2028 if (!ns_capable(sock_net(sk
)->user_ns
, CAP_NET_ADMIN
))
2032 case IPT_SO_GET_INFO
:
2033 ret
= get_info(sock_net(sk
), user
, len
, 0);
2036 case IPT_SO_GET_ENTRIES
:
2037 ret
= get_entries(sock_net(sk
), user
, len
);
2040 case IPT_SO_GET_REVISION_MATCH
:
2041 case IPT_SO_GET_REVISION_TARGET
: {
2042 struct xt_get_revision rev
;
2045 if (*len
!= sizeof(rev
)) {
2049 if (copy_from_user(&rev
, user
, sizeof(rev
)) != 0) {
2053 rev
.name
[sizeof(rev
.name
)-1] = 0;
2055 if (cmd
== IPT_SO_GET_REVISION_TARGET
)
2060 try_then_request_module(xt_find_revision(AF_INET
, rev
.name
,
2063 "ipt_%s", rev
.name
);
2068 duprintf("do_ipt_get_ctl: unknown request %i\n", cmd
);
2075 static void __ipt_unregister_table(struct net
*net
, struct xt_table
*table
)
2077 struct xt_table_info
*private;
2078 void *loc_cpu_entry
;
2079 struct module
*table_owner
= table
->me
;
2080 struct ipt_entry
*iter
;
2082 private = xt_unregister_table(table
);
2084 /* Decrease module usage counts and free resources */
2085 loc_cpu_entry
= private->entries
;
2086 xt_entry_foreach(iter
, loc_cpu_entry
, private->size
)
2087 cleanup_entry(iter
, net
);
2088 if (private->number
> private->initial_entries
)
2089 module_put(table_owner
);
2090 xt_free_table_info(private);
2093 int ipt_register_table(struct net
*net
, const struct xt_table
*table
,
2094 const struct ipt_replace
*repl
,
2095 const struct nf_hook_ops
*ops
, struct xt_table
**res
)
2098 struct xt_table_info
*newinfo
;
2099 struct xt_table_info bootstrap
= {0};
2100 void *loc_cpu_entry
;
2101 struct xt_table
*new_table
;
2103 newinfo
= xt_alloc_table_info(repl
->size
);
2107 loc_cpu_entry
= newinfo
->entries
;
2108 memcpy(loc_cpu_entry
, repl
->entries
, repl
->size
);
2110 ret
= translate_table(net
, newinfo
, loc_cpu_entry
, repl
);
2114 new_table
= xt_register_table(net
, table
, &bootstrap
, newinfo
);
2115 if (IS_ERR(new_table
)) {
2116 ret
= PTR_ERR(new_table
);
2120 /* set res now, will see skbs right after nf_register_net_hooks */
2121 WRITE_ONCE(*res
, new_table
);
2123 ret
= nf_register_net_hooks(net
, ops
, hweight32(table
->valid_hooks
));
2125 __ipt_unregister_table(net
, new_table
);
2132 xt_free_table_info(newinfo
);
2136 void ipt_unregister_table(struct net
*net
, struct xt_table
*table
,
2137 const struct nf_hook_ops
*ops
)
2139 nf_unregister_net_hooks(net
, ops
, hweight32(table
->valid_hooks
));
2140 __ipt_unregister_table(net
, table
);
2143 /* Returns 1 if the type and code is matched by the range, 0 otherwise */
2145 icmp_type_code_match(u_int8_t test_type
, u_int8_t min_code
, u_int8_t max_code
,
2146 u_int8_t type
, u_int8_t code
,
2149 return ((test_type
== 0xFF) ||
2150 (type
== test_type
&& code
>= min_code
&& code
<= max_code
))
2155 icmp_match(const struct sk_buff
*skb
, struct xt_action_param
*par
)
2157 const struct icmphdr
*ic
;
2158 struct icmphdr _icmph
;
2159 const struct ipt_icmp
*icmpinfo
= par
->matchinfo
;
2161 /* Must not be a fragment. */
2162 if (par
->fragoff
!= 0)
2165 ic
= skb_header_pointer(skb
, par
->thoff
, sizeof(_icmph
), &_icmph
);
2167 /* We've been asked to examine this packet, and we
2168 * can't. Hence, no choice but to drop.
2170 duprintf("Dropping evil ICMP tinygram.\n");
2171 par
->hotdrop
= true;
2175 return icmp_type_code_match(icmpinfo
->type
,
2179 !!(icmpinfo
->invflags
&IPT_ICMP_INV
));
2182 static int icmp_checkentry(const struct xt_mtchk_param
*par
)
2184 const struct ipt_icmp
*icmpinfo
= par
->matchinfo
;
2186 /* Must specify no unknown invflags */
2187 return (icmpinfo
->invflags
& ~IPT_ICMP_INV
) ? -EINVAL
: 0;
2190 static struct xt_target ipt_builtin_tg
[] __read_mostly
= {
2192 .name
= XT_STANDARD_TARGET
,
2193 .targetsize
= sizeof(int),
2194 .family
= NFPROTO_IPV4
,
2195 #ifdef CONFIG_COMPAT
2196 .compatsize
= sizeof(compat_int_t
),
2197 .compat_from_user
= compat_standard_from_user
,
2198 .compat_to_user
= compat_standard_to_user
,
2202 .name
= XT_ERROR_TARGET
,
2203 .target
= ipt_error
,
2204 .targetsize
= XT_FUNCTION_MAXNAMELEN
,
2205 .family
= NFPROTO_IPV4
,
2209 static struct nf_sockopt_ops ipt_sockopts
= {
2211 .set_optmin
= IPT_BASE_CTL
,
2212 .set_optmax
= IPT_SO_SET_MAX
+1,
2213 .set
= do_ipt_set_ctl
,
2214 #ifdef CONFIG_COMPAT
2215 .compat_set
= compat_do_ipt_set_ctl
,
2217 .get_optmin
= IPT_BASE_CTL
,
2218 .get_optmax
= IPT_SO_GET_MAX
+1,
2219 .get
= do_ipt_get_ctl
,
2220 #ifdef CONFIG_COMPAT
2221 .compat_get
= compat_do_ipt_get_ctl
,
2223 .owner
= THIS_MODULE
,
2226 static struct xt_match ipt_builtin_mt
[] __read_mostly
= {
2229 .match
= icmp_match
,
2230 .matchsize
= sizeof(struct ipt_icmp
),
2231 .checkentry
= icmp_checkentry
,
2232 .proto
= IPPROTO_ICMP
,
2233 .family
= NFPROTO_IPV4
,
2237 static int __net_init
ip_tables_net_init(struct net
*net
)
2239 return xt_proto_init(net
, NFPROTO_IPV4
);
2242 static void __net_exit
ip_tables_net_exit(struct net
*net
)
2244 xt_proto_fini(net
, NFPROTO_IPV4
);
2247 static struct pernet_operations ip_tables_net_ops
= {
2248 .init
= ip_tables_net_init
,
2249 .exit
= ip_tables_net_exit
,
2252 static int __init
ip_tables_init(void)
2256 ret
= register_pernet_subsys(&ip_tables_net_ops
);
2260 /* No one else will be downing sem now, so we won't sleep */
2261 ret
= xt_register_targets(ipt_builtin_tg
, ARRAY_SIZE(ipt_builtin_tg
));
2264 ret
= xt_register_matches(ipt_builtin_mt
, ARRAY_SIZE(ipt_builtin_mt
));
2268 /* Register setsockopt */
2269 ret
= nf_register_sockopt(&ipt_sockopts
);
2273 pr_info("(C) 2000-2006 Netfilter Core Team\n");
2277 xt_unregister_matches(ipt_builtin_mt
, ARRAY_SIZE(ipt_builtin_mt
));
2279 xt_unregister_targets(ipt_builtin_tg
, ARRAY_SIZE(ipt_builtin_tg
));
2281 unregister_pernet_subsys(&ip_tables_net_ops
);
2286 static void __exit
ip_tables_fini(void)
2288 nf_unregister_sockopt(&ipt_sockopts
);
2290 xt_unregister_matches(ipt_builtin_mt
, ARRAY_SIZE(ipt_builtin_mt
));
2291 xt_unregister_targets(ipt_builtin_tg
, ARRAY_SIZE(ipt_builtin_tg
));
2292 unregister_pernet_subsys(&ip_tables_net_ops
);
2295 EXPORT_SYMBOL(ipt_register_table
);
2296 EXPORT_SYMBOL(ipt_unregister_table
);
2297 EXPORT_SYMBOL(ipt_do_table
);
2298 module_init(ip_tables_init
);
2299 module_exit(ip_tables_fini
);