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 if (unlikely(stackidx >= private->stacksize)) {
415 jumpstack[stackidx++] = e;
416 pr_debug("Pushed %p into pos %u\n",
420 e = get_entry(table_base, v);
424 acpar.target = t->u.kernel.target;
425 acpar.targinfo = t->data;
427 verdict = t->u.kernel.target->target(skb, &acpar);
428 /* Target might have changed stuff. */
430 if (verdict == XT_CONTINUE)
431 e = ipt_next_entry(e);
435 } while (!acpar.hotdrop);
436 pr_debug("Exiting %s; sp at %u\n", __func__, stackidx);
438 xt_write_recseq_end(addend);
441 #ifdef DEBUG_ALLOW_ALL
450 /* Figures out from what hook each rule can be called: returns 0 if
451 there are loops. Puts hook bitmask in comefrom. */
453 mark_source_chains(const struct xt_table_info *newinfo,
454 unsigned int valid_hooks, void *entry0,
455 unsigned int *offsets)
459 /* No recursion; use packet counter to save back ptrs (reset
460 to 0 as we leave), and comefrom to save source hook bitmask */
461 for (hook = 0; hook < NF_INET_NUMHOOKS; hook++) {
462 unsigned int pos = newinfo->hook_entry[hook];
463 struct ipt_entry *e = (struct ipt_entry *)(entry0 + pos);
465 if (!(valid_hooks & (1 << hook)))
468 /* Set initial back pointer. */
469 e->counters.pcnt = pos;
472 const struct xt_standard_target *t
473 = (void *)ipt_get_target_c(e);
474 int visited = e->comefrom & (1 << hook);
476 if (e->comefrom & (1 << NF_INET_NUMHOOKS)) {
477 pr_err("iptables: loop hook %u pos %u %08X.\n",
478 hook, pos, e->comefrom);
481 e->comefrom |= ((1 << hook) | (1 << NF_INET_NUMHOOKS));
483 /* Unconditional return/END. */
484 if ((unconditional(e) &&
485 (strcmp(t->target.u.user.name,
486 XT_STANDARD_TARGET) == 0) &&
487 t->verdict < 0) || visited) {
488 unsigned int oldpos, size;
490 if ((strcmp(t->target.u.user.name,
491 XT_STANDARD_TARGET) == 0) &&
492 t->verdict < -NF_MAX_VERDICT - 1) {
493 duprintf("mark_source_chains: bad "
494 "negative verdict (%i)\n",
499 /* Return: backtrack through the last
502 e->comefrom ^= (1<<NF_INET_NUMHOOKS);
503 #ifdef DEBUG_IP_FIREWALL_USER
505 & (1 << NF_INET_NUMHOOKS)) {
506 duprintf("Back unset "
513 pos = e->counters.pcnt;
514 e->counters.pcnt = 0;
516 /* We're at the start. */
520 e = (struct ipt_entry *)
522 } while (oldpos == pos + e->next_offset);
525 size = e->next_offset;
526 e = (struct ipt_entry *)
527 (entry0 + pos + size);
528 if (pos + size >= newinfo->size)
530 e->counters.pcnt = pos;
533 int newpos = t->verdict;
535 if (strcmp(t->target.u.user.name,
536 XT_STANDARD_TARGET) == 0 &&
538 if (newpos > newinfo->size -
539 sizeof(struct ipt_entry)) {
540 duprintf("mark_source_chains: "
541 "bad verdict (%i)\n",
545 /* This a jump; chase it. */
546 duprintf("Jump rule %u -> %u\n",
548 if (!xt_find_jump_offset(offsets, newpos,
551 e = (struct ipt_entry *)
554 /* ... this is a fallthru */
555 newpos = pos + e->next_offset;
556 if (newpos >= newinfo->size)
559 e = (struct ipt_entry *)
561 e->counters.pcnt = pos;
566 duprintf("Finished chain %u\n", hook);
571 static void cleanup_match(struct xt_entry_match *m, struct net *net)
573 struct xt_mtdtor_param par;
576 par.match = m->u.kernel.match;
577 par.matchinfo = m->data;
578 par.family = NFPROTO_IPV4;
579 if (par.match->destroy != NULL)
580 par.match->destroy(&par);
581 module_put(par.match->me);
585 check_match(struct xt_entry_match *m, struct xt_mtchk_param *par)
587 const struct ipt_ip *ip = par->entryinfo;
590 par->match = m->u.kernel.match;
591 par->matchinfo = m->data;
593 ret = xt_check_match(par, m->u.match_size - sizeof(*m),
594 ip->proto, ip->invflags & IPT_INV_PROTO);
596 duprintf("check failed for `%s'.\n", par->match->name);
603 find_check_match(struct xt_entry_match *m, struct xt_mtchk_param *par)
605 struct xt_match *match;
608 match = xt_request_find_match(NFPROTO_IPV4, m->u.user.name,
611 duprintf("find_check_match: `%s' not found\n", m->u.user.name);
612 return PTR_ERR(match);
614 m->u.kernel.match = match;
616 ret = check_match(m, par);
622 module_put(m->u.kernel.match->me);
626 static int check_target(struct ipt_entry *e, struct net *net, const char *name)
628 struct xt_entry_target *t = ipt_get_target(e);
629 struct xt_tgchk_param par = {
633 .target = t->u.kernel.target,
635 .hook_mask = e->comefrom,
636 .family = NFPROTO_IPV4,
640 ret = xt_check_target(&par, t->u.target_size - sizeof(*t),
641 e->ip.proto, e->ip.invflags & IPT_INV_PROTO);
643 duprintf("check failed for `%s'.\n",
644 t->u.kernel.target->name);
651 find_check_entry(struct ipt_entry *e, struct net *net, const char *name,
653 struct xt_percpu_counter_alloc_state *alloc_state)
655 struct xt_entry_target *t;
656 struct xt_target *target;
659 struct xt_mtchk_param mtpar;
660 struct xt_entry_match *ematch;
662 if (!xt_percpu_counter_alloc(alloc_state, &e->counters))
666 memset(&mtpar, 0, sizeof(mtpar));
669 mtpar.entryinfo = &e->ip;
670 mtpar.hook_mask = e->comefrom;
671 mtpar.family = NFPROTO_IPV4;
672 xt_ematch_foreach(ematch, e) {
673 ret = find_check_match(ematch, &mtpar);
675 goto cleanup_matches;
679 t = ipt_get_target(e);
680 target = xt_request_find_target(NFPROTO_IPV4, t->u.user.name,
682 if (IS_ERR(target)) {
683 duprintf("find_check_entry: `%s' not found\n", t->u.user.name);
684 ret = PTR_ERR(target);
685 goto cleanup_matches;
687 t->u.kernel.target = target;
689 ret = check_target(e, net, name);
695 module_put(t->u.kernel.target->me);
697 xt_ematch_foreach(ematch, e) {
700 cleanup_match(ematch, net);
703 xt_percpu_counter_free(&e->counters);
708 static bool check_underflow(const struct ipt_entry *e)
710 const struct xt_entry_target *t;
711 unsigned int verdict;
713 if (!unconditional(e))
715 t = ipt_get_target_c(e);
716 if (strcmp(t->u.user.name, XT_STANDARD_TARGET) != 0)
718 verdict = ((struct xt_standard_target *)t)->verdict;
719 verdict = -verdict - 1;
720 return verdict == NF_DROP || verdict == NF_ACCEPT;
724 check_entry_size_and_hooks(struct ipt_entry *e,
725 struct xt_table_info *newinfo,
726 const unsigned char *base,
727 const unsigned char *limit,
728 const unsigned int *hook_entries,
729 const unsigned int *underflows,
730 unsigned int valid_hooks)
735 if ((unsigned long)e % __alignof__(struct ipt_entry) != 0 ||
736 (unsigned char *)e + sizeof(struct ipt_entry) >= limit ||
737 (unsigned char *)e + e->next_offset > limit) {
738 duprintf("Bad offset %p\n", e);
743 < sizeof(struct ipt_entry) + sizeof(struct xt_entry_target)) {
744 duprintf("checking: element %p size %u\n",
749 if (!ip_checkentry(&e->ip))
752 err = xt_check_entry_offsets(e, e->elems, e->target_offset,
757 /* Check hooks & underflows */
758 for (h = 0; h < NF_INET_NUMHOOKS; h++) {
759 if (!(valid_hooks & (1 << h)))
761 if ((unsigned char *)e - base == hook_entries[h])
762 newinfo->hook_entry[h] = hook_entries[h];
763 if ((unsigned char *)e - base == underflows[h]) {
764 if (!check_underflow(e)) {
765 pr_debug("Underflows must be unconditional and "
766 "use the STANDARD target with "
770 newinfo->underflow[h] = underflows[h];
774 /* Clear counters and comefrom */
775 e->counters = ((struct xt_counters) { 0, 0 });
781 cleanup_entry(struct ipt_entry *e, struct net *net)
783 struct xt_tgdtor_param par;
784 struct xt_entry_target *t;
785 struct xt_entry_match *ematch;
787 /* Cleanup all matches */
788 xt_ematch_foreach(ematch, e)
789 cleanup_match(ematch, net);
790 t = ipt_get_target(e);
793 par.target = t->u.kernel.target;
794 par.targinfo = t->data;
795 par.family = NFPROTO_IPV4;
796 if (par.target->destroy != NULL)
797 par.target->destroy(&par);
798 module_put(par.target->me);
799 xt_percpu_counter_free(&e->counters);
802 /* Checks and translates the user-supplied table segment (held in
805 translate_table(struct net *net, struct xt_table_info *newinfo, void *entry0,
806 const struct ipt_replace *repl)
808 struct xt_percpu_counter_alloc_state alloc_state = { 0 };
809 struct ipt_entry *iter;
810 unsigned int *offsets;
814 newinfo->size = repl->size;
815 newinfo->number = repl->num_entries;
817 /* Init all hooks to impossible value. */
818 for (i = 0; i < NF_INET_NUMHOOKS; i++) {
819 newinfo->hook_entry[i] = 0xFFFFFFFF;
820 newinfo->underflow[i] = 0xFFFFFFFF;
823 duprintf("translate_table: size %u\n", newinfo->size);
824 offsets = xt_alloc_entry_offsets(newinfo->number);
828 /* Walk through entries, checking offsets. */
829 xt_entry_foreach(iter, entry0, newinfo->size) {
830 ret = check_entry_size_and_hooks(iter, newinfo, entry0,
837 if (i < repl->num_entries)
838 offsets[i] = (void *)iter - entry0;
840 if (strcmp(ipt_get_target(iter)->u.user.name,
841 XT_ERROR_TARGET) == 0)
842 ++newinfo->stacksize;
846 if (i != repl->num_entries) {
847 duprintf("translate_table: %u not %u entries\n",
848 i, repl->num_entries);
852 /* Check hooks all assigned */
853 for (i = 0; i < NF_INET_NUMHOOKS; i++) {
854 /* Only hooks which are valid */
855 if (!(repl->valid_hooks & (1 << i)))
857 if (newinfo->hook_entry[i] == 0xFFFFFFFF) {
858 duprintf("Invalid hook entry %u %u\n",
859 i, repl->hook_entry[i]);
862 if (newinfo->underflow[i] == 0xFFFFFFFF) {
863 duprintf("Invalid underflow %u %u\n",
864 i, repl->underflow[i]);
869 if (!mark_source_chains(newinfo, repl->valid_hooks, entry0, offsets)) {
875 /* Finally, each sanity check must pass */
877 xt_entry_foreach(iter, entry0, newinfo->size) {
878 ret = find_check_entry(iter, net, repl->name, repl->size,
886 xt_entry_foreach(iter, entry0, newinfo->size) {
889 cleanup_entry(iter, net);
901 get_counters(const struct xt_table_info *t,
902 struct xt_counters counters[])
904 struct ipt_entry *iter;
908 for_each_possible_cpu(cpu) {
909 seqcount_t *s = &per_cpu(xt_recseq, cpu);
912 xt_entry_foreach(iter, t->entries, t->size) {
913 struct xt_counters *tmp;
917 tmp = xt_get_per_cpu_counter(&iter->counters, cpu);
919 start = read_seqcount_begin(s);
922 } while (read_seqcount_retry(s, start));
924 ADD_COUNTER(counters[i], bcnt, pcnt);
925 ++i; /* macro does multi eval of i */
930 static struct xt_counters *alloc_counters(const struct xt_table *table)
932 unsigned int countersize;
933 struct xt_counters *counters;
934 const struct xt_table_info *private = table->private;
936 /* We need atomic snapshot of counters: rest doesn't change
937 (other than comefrom, which userspace doesn't care
939 countersize = sizeof(struct xt_counters) * private->number;
940 counters = vzalloc(countersize);
942 if (counters == NULL)
943 return ERR_PTR(-ENOMEM);
945 get_counters(private, counters);
951 copy_entries_to_user(unsigned int total_size,
952 const struct xt_table *table,
953 void __user *userptr)
955 unsigned int off, num;
956 const struct ipt_entry *e;
957 struct xt_counters *counters;
958 const struct xt_table_info *private = table->private;
960 const void *loc_cpu_entry;
962 counters = alloc_counters(table);
963 if (IS_ERR(counters))
964 return PTR_ERR(counters);
966 loc_cpu_entry = private->entries;
967 if (copy_to_user(userptr, loc_cpu_entry, total_size) != 0) {
972 /* FIXME: use iterator macros --RR */
973 /* ... then go back and fix counters and names */
974 for (off = 0, num = 0; off < total_size; off += e->next_offset, num++){
976 const struct xt_entry_match *m;
977 const struct xt_entry_target *t;
979 e = (struct ipt_entry *)(loc_cpu_entry + off);
980 if (copy_to_user(userptr + off
981 + offsetof(struct ipt_entry, counters),
983 sizeof(counters[num])) != 0) {
988 for (i = sizeof(struct ipt_entry);
989 i < e->target_offset;
990 i += m->u.match_size) {
993 if (copy_to_user(userptr + off + i
994 + offsetof(struct xt_entry_match,
996 m->u.kernel.match->name,
997 strlen(m->u.kernel.match->name)+1)
1004 t = ipt_get_target_c(e);
1005 if (copy_to_user(userptr + off + e->target_offset
1006 + offsetof(struct xt_entry_target,
1008 t->u.kernel.target->name,
1009 strlen(t->u.kernel.target->name)+1) != 0) {
1020 #ifdef CONFIG_COMPAT
1021 static void compat_standard_from_user(void *dst, const void *src)
1023 int v = *(compat_int_t *)src;
1026 v += xt_compat_calc_jump(AF_INET, v);
1027 memcpy(dst, &v, sizeof(v));
1030 static int compat_standard_to_user(void __user *dst, const void *src)
1032 compat_int_t cv = *(int *)src;
1035 cv -= xt_compat_calc_jump(AF_INET, cv);
1036 return copy_to_user(dst, &cv, sizeof(cv)) ? -EFAULT : 0;
1039 static int compat_calc_entry(const struct ipt_entry *e,
1040 const struct xt_table_info *info,
1041 const void *base, struct xt_table_info *newinfo)
1043 const struct xt_entry_match *ematch;
1044 const struct xt_entry_target *t;
1045 unsigned int entry_offset;
1048 off = sizeof(struct ipt_entry) - sizeof(struct compat_ipt_entry);
1049 entry_offset = (void *)e - base;
1050 xt_ematch_foreach(ematch, e)
1051 off += xt_compat_match_offset(ematch->u.kernel.match);
1052 t = ipt_get_target_c(e);
1053 off += xt_compat_target_offset(t->u.kernel.target);
1054 newinfo->size -= off;
1055 ret = xt_compat_add_offset(AF_INET, entry_offset, off);
1059 for (i = 0; i < NF_INET_NUMHOOKS; i++) {
1060 if (info->hook_entry[i] &&
1061 (e < (struct ipt_entry *)(base + info->hook_entry[i])))
1062 newinfo->hook_entry[i] -= off;
1063 if (info->underflow[i] &&
1064 (e < (struct ipt_entry *)(base + info->underflow[i])))
1065 newinfo->underflow[i] -= off;
1070 static int compat_table_info(const struct xt_table_info *info,
1071 struct xt_table_info *newinfo)
1073 struct ipt_entry *iter;
1074 const void *loc_cpu_entry;
1077 if (!newinfo || !info)
1080 /* we dont care about newinfo->entries */
1081 memcpy(newinfo, info, offsetof(struct xt_table_info, entries));
1082 newinfo->initial_entries = 0;
1083 loc_cpu_entry = info->entries;
1084 xt_compat_init_offsets(AF_INET, info->number);
1085 xt_entry_foreach(iter, loc_cpu_entry, info->size) {
1086 ret = compat_calc_entry(iter, info, loc_cpu_entry, newinfo);
1094 static int get_info(struct net *net, void __user *user,
1095 const int *len, int compat)
1097 char name[XT_TABLE_MAXNAMELEN];
1101 if (*len != sizeof(struct ipt_getinfo)) {
1102 duprintf("length %u != %zu\n", *len,
1103 sizeof(struct ipt_getinfo));
1107 if (copy_from_user(name, user, sizeof(name)) != 0)
1110 name[XT_TABLE_MAXNAMELEN-1] = '\0';
1111 #ifdef CONFIG_COMPAT
1113 xt_compat_lock(AF_INET);
1115 t = try_then_request_module(xt_find_table_lock(net, AF_INET, name),
1116 "iptable_%s", name);
1117 if (!IS_ERR_OR_NULL(t)) {
1118 struct ipt_getinfo info;
1119 const struct xt_table_info *private = t->private;
1120 #ifdef CONFIG_COMPAT
1121 struct xt_table_info tmp;
1124 ret = compat_table_info(private, &tmp);
1125 xt_compat_flush_offsets(AF_INET);
1129 memset(&info, 0, sizeof(info));
1130 info.valid_hooks = t->valid_hooks;
1131 memcpy(info.hook_entry, private->hook_entry,
1132 sizeof(info.hook_entry));
1133 memcpy(info.underflow, private->underflow,
1134 sizeof(info.underflow));
1135 info.num_entries = private->number;
1136 info.size = private->size;
1137 strcpy(info.name, name);
1139 if (copy_to_user(user, &info, *len) != 0)
1147 ret = t ? PTR_ERR(t) : -ENOENT;
1148 #ifdef CONFIG_COMPAT
1150 xt_compat_unlock(AF_INET);
1156 get_entries(struct net *net, struct ipt_get_entries __user *uptr,
1160 struct ipt_get_entries get;
1163 if (*len < sizeof(get)) {
1164 duprintf("get_entries: %u < %zu\n", *len, sizeof(get));
1167 if (copy_from_user(&get, uptr, sizeof(get)) != 0)
1169 if (*len != sizeof(struct ipt_get_entries) + get.size) {
1170 duprintf("get_entries: %u != %zu\n",
1171 *len, sizeof(get) + get.size);
1174 get.name[sizeof(get.name) - 1] = '\0';
1176 t = xt_find_table_lock(net, AF_INET, get.name);
1177 if (!IS_ERR_OR_NULL(t)) {
1178 const struct xt_table_info *private = t->private;
1179 duprintf("t->private->number = %u\n", private->number);
1180 if (get.size == private->size)
1181 ret = copy_entries_to_user(private->size,
1182 t, uptr->entrytable);
1184 duprintf("get_entries: I've got %u not %u!\n",
1185 private->size, get.size);
1191 ret = t ? PTR_ERR(t) : -ENOENT;
1197 __do_replace(struct net *net, const char *name, unsigned int valid_hooks,
1198 struct xt_table_info *newinfo, unsigned int num_counters,
1199 void __user *counters_ptr)
1203 struct xt_table_info *oldinfo;
1204 struct xt_counters *counters;
1205 struct ipt_entry *iter;
1208 counters = vzalloc(num_counters * sizeof(struct xt_counters));
1214 t = try_then_request_module(xt_find_table_lock(net, AF_INET, name),
1215 "iptable_%s", name);
1216 if (IS_ERR_OR_NULL(t)) {
1217 ret = t ? PTR_ERR(t) : -ENOENT;
1218 goto free_newinfo_counters_untrans;
1222 if (valid_hooks != t->valid_hooks) {
1223 duprintf("Valid hook crap: %08X vs %08X\n",
1224 valid_hooks, t->valid_hooks);
1229 oldinfo = xt_replace_table(t, num_counters, newinfo, &ret);
1233 /* Update module usage count based on number of rules */
1234 duprintf("do_replace: oldnum=%u, initnum=%u, newnum=%u\n",
1235 oldinfo->number, oldinfo->initial_entries, newinfo->number);
1236 if ((oldinfo->number > oldinfo->initial_entries) ||
1237 (newinfo->number <= oldinfo->initial_entries))
1239 if ((oldinfo->number > oldinfo->initial_entries) &&
1240 (newinfo->number <= oldinfo->initial_entries))
1243 /* Get the old counters, and synchronize with replace */
1244 get_counters(oldinfo, counters);
1246 /* Decrease module usage counts and free resource */
1247 xt_entry_foreach(iter, oldinfo->entries, oldinfo->size)
1248 cleanup_entry(iter, net);
1250 xt_free_table_info(oldinfo);
1251 if (copy_to_user(counters_ptr, counters,
1252 sizeof(struct xt_counters) * num_counters) != 0) {
1253 /* Silent error, can't fail, new table is already in place */
1254 net_warn_ratelimited("iptables: counters copy to user failed while replacing table\n");
1263 free_newinfo_counters_untrans:
1270 do_replace(struct net *net, const void __user *user, unsigned int len)
1273 struct ipt_replace tmp;
1274 struct xt_table_info *newinfo;
1275 void *loc_cpu_entry;
1276 struct ipt_entry *iter;
1278 if (copy_from_user(&tmp, user, sizeof(tmp)) != 0)
1281 /* overflow check */
1282 if (tmp.num_counters >= INT_MAX / sizeof(struct xt_counters))
1284 if (tmp.num_counters == 0)
1287 tmp.name[sizeof(tmp.name)-1] = 0;
1289 newinfo = xt_alloc_table_info(tmp.size);
1293 loc_cpu_entry = newinfo->entries;
1294 if (copy_from_user(loc_cpu_entry, user + sizeof(tmp),
1300 ret = translate_table(net, newinfo, loc_cpu_entry, &tmp);
1304 duprintf("Translated table\n");
1306 ret = __do_replace(net, tmp.name, tmp.valid_hooks, newinfo,
1307 tmp.num_counters, tmp.counters);
1309 goto free_newinfo_untrans;
1312 free_newinfo_untrans:
1313 xt_entry_foreach(iter, loc_cpu_entry, newinfo->size)
1314 cleanup_entry(iter, net);
1316 xt_free_table_info(newinfo);
1321 do_add_counters(struct net *net, const void __user *user,
1322 unsigned int len, int compat)
1325 struct xt_counters_info tmp;
1326 struct xt_counters *paddc;
1328 const struct xt_table_info *private;
1330 struct ipt_entry *iter;
1331 unsigned int addend;
1333 paddc = xt_copy_counters_from_user(user, len, &tmp, compat);
1335 return PTR_ERR(paddc);
1337 t = xt_find_table_lock(net, AF_INET, tmp.name);
1338 if (IS_ERR_OR_NULL(t)) {
1339 ret = t ? PTR_ERR(t) : -ENOENT;
1344 private = t->private;
1345 if (private->number != tmp.num_counters) {
1347 goto unlock_up_free;
1351 addend = xt_write_recseq_begin();
1352 xt_entry_foreach(iter, private->entries, private->size) {
1353 struct xt_counters *tmp;
1355 tmp = xt_get_this_cpu_counter(&iter->counters);
1356 ADD_COUNTER(*tmp, paddc[i].bcnt, paddc[i].pcnt);
1359 xt_write_recseq_end(addend);
1370 #ifdef CONFIG_COMPAT
1371 struct compat_ipt_replace {
1372 char name[XT_TABLE_MAXNAMELEN];
1376 u32 hook_entry[NF_INET_NUMHOOKS];
1377 u32 underflow[NF_INET_NUMHOOKS];
1379 compat_uptr_t counters; /* struct xt_counters * */
1380 struct compat_ipt_entry entries[0];
1384 compat_copy_entry_to_user(struct ipt_entry *e, void __user **dstptr,
1385 unsigned int *size, struct xt_counters *counters,
1388 struct xt_entry_target *t;
1389 struct compat_ipt_entry __user *ce;
1390 u_int16_t target_offset, next_offset;
1391 compat_uint_t origsize;
1392 const struct xt_entry_match *ematch;
1396 ce = (struct compat_ipt_entry __user *)*dstptr;
1397 if (copy_to_user(ce, e, sizeof(struct ipt_entry)) != 0 ||
1398 copy_to_user(&ce->counters, &counters[i],
1399 sizeof(counters[i])) != 0)
1402 *dstptr += sizeof(struct compat_ipt_entry);
1403 *size -= sizeof(struct ipt_entry) - sizeof(struct compat_ipt_entry);
1405 xt_ematch_foreach(ematch, e) {
1406 ret = xt_compat_match_to_user(ematch, dstptr, size);
1410 target_offset = e->target_offset - (origsize - *size);
1411 t = ipt_get_target(e);
1412 ret = xt_compat_target_to_user(t, dstptr, size);
1415 next_offset = e->next_offset - (origsize - *size);
1416 if (put_user(target_offset, &ce->target_offset) != 0 ||
1417 put_user(next_offset, &ce->next_offset) != 0)
1423 compat_find_calc_match(struct xt_entry_match *m,
1424 const struct ipt_ip *ip,
1427 struct xt_match *match;
1429 match = xt_request_find_match(NFPROTO_IPV4, m->u.user.name,
1430 m->u.user.revision);
1431 if (IS_ERR(match)) {
1432 duprintf("compat_check_calc_match: `%s' not found\n",
1434 return PTR_ERR(match);
1436 m->u.kernel.match = match;
1437 *size += xt_compat_match_offset(match);
1441 static void compat_release_entry(struct compat_ipt_entry *e)
1443 struct xt_entry_target *t;
1444 struct xt_entry_match *ematch;
1446 /* Cleanup all matches */
1447 xt_ematch_foreach(ematch, e)
1448 module_put(ematch->u.kernel.match->me);
1449 t = compat_ipt_get_target(e);
1450 module_put(t->u.kernel.target->me);
1454 check_compat_entry_size_and_hooks(struct compat_ipt_entry *e,
1455 struct xt_table_info *newinfo,
1457 const unsigned char *base,
1458 const unsigned char *limit)
1460 struct xt_entry_match *ematch;
1461 struct xt_entry_target *t;
1462 struct xt_target *target;
1463 unsigned int entry_offset;
1467 duprintf("check_compat_entry_size_and_hooks %p\n", e);
1468 if ((unsigned long)e % __alignof__(struct compat_ipt_entry) != 0 ||
1469 (unsigned char *)e + sizeof(struct compat_ipt_entry) >= limit ||
1470 (unsigned char *)e + e->next_offset > limit) {
1471 duprintf("Bad offset %p, limit = %p\n", e, limit);
1475 if (e->next_offset < sizeof(struct compat_ipt_entry) +
1476 sizeof(struct compat_xt_entry_target)) {
1477 duprintf("checking: element %p size %u\n",
1482 if (!ip_checkentry(&e->ip))
1485 ret = xt_compat_check_entry_offsets(e, e->elems,
1486 e->target_offset, e->next_offset);
1490 off = sizeof(struct ipt_entry) - sizeof(struct compat_ipt_entry);
1491 entry_offset = (void *)e - (void *)base;
1493 xt_ematch_foreach(ematch, e) {
1494 ret = compat_find_calc_match(ematch, &e->ip, &off);
1496 goto release_matches;
1500 t = compat_ipt_get_target(e);
1501 target = xt_request_find_target(NFPROTO_IPV4, t->u.user.name,
1502 t->u.user.revision);
1503 if (IS_ERR(target)) {
1504 duprintf("check_compat_entry_size_and_hooks: `%s' not found\n",
1506 ret = PTR_ERR(target);
1507 goto release_matches;
1509 t->u.kernel.target = target;
1511 off += xt_compat_target_offset(target);
1513 ret = xt_compat_add_offset(AF_INET, entry_offset, off);
1520 module_put(t->u.kernel.target->me);
1522 xt_ematch_foreach(ematch, e) {
1525 module_put(ematch->u.kernel.match->me);
1531 compat_copy_entry_from_user(struct compat_ipt_entry *e, void **dstptr,
1533 struct xt_table_info *newinfo, unsigned char *base)
1535 struct xt_entry_target *t;
1536 struct xt_target *target;
1537 struct ipt_entry *de;
1538 unsigned int origsize;
1540 struct xt_entry_match *ematch;
1543 de = (struct ipt_entry *)*dstptr;
1544 memcpy(de, e, sizeof(struct ipt_entry));
1545 memcpy(&de->counters, &e->counters, sizeof(e->counters));
1547 *dstptr += sizeof(struct ipt_entry);
1548 *size += sizeof(struct ipt_entry) - sizeof(struct compat_ipt_entry);
1550 xt_ematch_foreach(ematch, e)
1551 xt_compat_match_from_user(ematch, dstptr, size);
1553 de->target_offset = e->target_offset - (origsize - *size);
1554 t = compat_ipt_get_target(e);
1555 target = t->u.kernel.target;
1556 xt_compat_target_from_user(t, dstptr, size);
1558 de->next_offset = e->next_offset - (origsize - *size);
1560 for (h = 0; h < NF_INET_NUMHOOKS; h++) {
1561 if ((unsigned char *)de - base < newinfo->hook_entry[h])
1562 newinfo->hook_entry[h] -= origsize - *size;
1563 if ((unsigned char *)de - base < newinfo->underflow[h])
1564 newinfo->underflow[h] -= origsize - *size;
1569 translate_compat_table(struct net *net,
1570 struct xt_table_info **pinfo,
1572 const struct compat_ipt_replace *compatr)
1575 struct xt_table_info *newinfo, *info;
1576 void *pos, *entry0, *entry1;
1577 struct compat_ipt_entry *iter0;
1578 struct ipt_replace repl;
1584 size = compatr->size;
1585 info->number = compatr->num_entries;
1587 duprintf("translate_compat_table: size %u\n", info->size);
1589 xt_compat_lock(AF_INET);
1590 xt_compat_init_offsets(AF_INET, compatr->num_entries);
1591 /* Walk through entries, checking offsets. */
1592 xt_entry_foreach(iter0, entry0, compatr->size) {
1593 ret = check_compat_entry_size_and_hooks(iter0, info, &size,
1595 entry0 + compatr->size);
1602 if (j != compatr->num_entries) {
1603 duprintf("translate_compat_table: %u not %u entries\n",
1604 j, compatr->num_entries);
1609 newinfo = xt_alloc_table_info(size);
1613 memset(newinfo->entries, 0, size);
1615 newinfo->number = compatr->num_entries;
1616 for (i = 0; i < NF_INET_NUMHOOKS; i++) {
1617 newinfo->hook_entry[i] = compatr->hook_entry[i];
1618 newinfo->underflow[i] = compatr->underflow[i];
1620 entry1 = newinfo->entries;
1622 size = compatr->size;
1623 xt_entry_foreach(iter0, entry0, compatr->size)
1624 compat_copy_entry_from_user(iter0, &pos, &size,
1627 /* all module references in entry0 are now gone.
1628 * entry1/newinfo contains a 64bit ruleset that looks exactly as
1629 * generated by 64bit userspace.
1631 * Call standard translate_table() to validate all hook_entrys,
1632 * underflows, check for loops, etc.
1634 xt_compat_flush_offsets(AF_INET);
1635 xt_compat_unlock(AF_INET);
1637 memcpy(&repl, compatr, sizeof(*compatr));
1639 for (i = 0; i < NF_INET_NUMHOOKS; i++) {
1640 repl.hook_entry[i] = newinfo->hook_entry[i];
1641 repl.underflow[i] = newinfo->underflow[i];
1644 repl.num_counters = 0;
1645 repl.counters = NULL;
1646 repl.size = newinfo->size;
1647 ret = translate_table(net, newinfo, entry1, &repl);
1653 xt_free_table_info(info);
1657 xt_free_table_info(newinfo);
1660 xt_compat_flush_offsets(AF_INET);
1661 xt_compat_unlock(AF_INET);
1662 xt_entry_foreach(iter0, entry0, compatr->size) {
1665 compat_release_entry(iter0);
1671 compat_do_replace(struct net *net, void __user *user, unsigned int len)
1674 struct compat_ipt_replace tmp;
1675 struct xt_table_info *newinfo;
1676 void *loc_cpu_entry;
1677 struct ipt_entry *iter;
1679 if (copy_from_user(&tmp, user, sizeof(tmp)) != 0)
1682 /* overflow check */
1683 if (tmp.size >= INT_MAX / num_possible_cpus())
1685 if (tmp.num_counters >= INT_MAX / sizeof(struct xt_counters))
1687 if (tmp.num_counters == 0)
1690 tmp.name[sizeof(tmp.name)-1] = 0;
1692 newinfo = xt_alloc_table_info(tmp.size);
1696 loc_cpu_entry = newinfo->entries;
1697 if (copy_from_user(loc_cpu_entry, user + sizeof(tmp),
1703 ret = translate_compat_table(net, &newinfo, &loc_cpu_entry, &tmp);
1707 duprintf("compat_do_replace: Translated table\n");
1709 ret = __do_replace(net, tmp.name, tmp.valid_hooks, newinfo,
1710 tmp.num_counters, compat_ptr(tmp.counters));
1712 goto free_newinfo_untrans;
1715 free_newinfo_untrans:
1716 xt_entry_foreach(iter, loc_cpu_entry, newinfo->size)
1717 cleanup_entry(iter, net);
1719 xt_free_table_info(newinfo);
1724 compat_do_ipt_set_ctl(struct sock *sk, int cmd, void __user *user,
1729 if (!ns_capable(sock_net(sk)->user_ns, CAP_NET_ADMIN))
1733 case IPT_SO_SET_REPLACE:
1734 ret = compat_do_replace(sock_net(sk), user, len);
1737 case IPT_SO_SET_ADD_COUNTERS:
1738 ret = do_add_counters(sock_net(sk), user, len, 1);
1742 duprintf("do_ipt_set_ctl: unknown request %i\n", cmd);
1749 struct compat_ipt_get_entries {
1750 char name[XT_TABLE_MAXNAMELEN];
1752 struct compat_ipt_entry entrytable[0];
1756 compat_copy_entries_to_user(unsigned int total_size, struct xt_table *table,
1757 void __user *userptr)
1759 struct xt_counters *counters;
1760 const struct xt_table_info *private = table->private;
1765 struct ipt_entry *iter;
1767 counters = alloc_counters(table);
1768 if (IS_ERR(counters))
1769 return PTR_ERR(counters);
1773 xt_entry_foreach(iter, private->entries, total_size) {
1774 ret = compat_copy_entry_to_user(iter, &pos,
1775 &size, counters, i++);
1785 compat_get_entries(struct net *net, struct compat_ipt_get_entries __user *uptr,
1789 struct compat_ipt_get_entries get;
1792 if (*len < sizeof(get)) {
1793 duprintf("compat_get_entries: %u < %zu\n", *len, sizeof(get));
1797 if (copy_from_user(&get, uptr, sizeof(get)) != 0)
1800 if (*len != sizeof(struct compat_ipt_get_entries) + get.size) {
1801 duprintf("compat_get_entries: %u != %zu\n",
1802 *len, sizeof(get) + get.size);
1805 get.name[sizeof(get.name) - 1] = '\0';
1807 xt_compat_lock(AF_INET);
1808 t = xt_find_table_lock(net, AF_INET, get.name);
1809 if (!IS_ERR_OR_NULL(t)) {
1810 const struct xt_table_info *private = t->private;
1811 struct xt_table_info info;
1812 duprintf("t->private->number = %u\n", private->number);
1813 ret = compat_table_info(private, &info);
1814 if (!ret && get.size == info.size) {
1815 ret = compat_copy_entries_to_user(private->size,
1816 t, uptr->entrytable);
1818 duprintf("compat_get_entries: I've got %u not %u!\n",
1819 private->size, get.size);
1822 xt_compat_flush_offsets(AF_INET);
1826 ret = t ? PTR_ERR(t) : -ENOENT;
1828 xt_compat_unlock(AF_INET);
1832 static int do_ipt_get_ctl(struct sock *, int, void __user *, int *);
1835 compat_do_ipt_get_ctl(struct sock *sk, int cmd, void __user *user, int *len)
1839 if (!ns_capable(sock_net(sk)->user_ns, CAP_NET_ADMIN))
1843 case IPT_SO_GET_INFO:
1844 ret = get_info(sock_net(sk), user, len, 1);
1846 case IPT_SO_GET_ENTRIES:
1847 ret = compat_get_entries(sock_net(sk), user, len);
1850 ret = do_ipt_get_ctl(sk, cmd, user, len);
1857 do_ipt_set_ctl(struct sock *sk, int cmd, void __user *user, unsigned int len)
1861 if (!ns_capable(sock_net(sk)->user_ns, CAP_NET_ADMIN))
1865 case IPT_SO_SET_REPLACE:
1866 ret = do_replace(sock_net(sk), user, len);
1869 case IPT_SO_SET_ADD_COUNTERS:
1870 ret = do_add_counters(sock_net(sk), user, len, 0);
1874 duprintf("do_ipt_set_ctl: unknown request %i\n", cmd);
1882 do_ipt_get_ctl(struct sock *sk, int cmd, void __user *user, int *len)
1886 if (!ns_capable(sock_net(sk)->user_ns, CAP_NET_ADMIN))
1890 case IPT_SO_GET_INFO:
1891 ret = get_info(sock_net(sk), user, len, 0);
1894 case IPT_SO_GET_ENTRIES:
1895 ret = get_entries(sock_net(sk), user, len);
1898 case IPT_SO_GET_REVISION_MATCH:
1899 case IPT_SO_GET_REVISION_TARGET: {
1900 struct xt_get_revision rev;
1903 if (*len != sizeof(rev)) {
1907 if (copy_from_user(&rev, user, sizeof(rev)) != 0) {
1911 rev.name[sizeof(rev.name)-1] = 0;
1913 if (cmd == IPT_SO_GET_REVISION_TARGET)
1918 try_then_request_module(xt_find_revision(AF_INET, rev.name,
1921 "ipt_%s", rev.name);
1926 duprintf("do_ipt_get_ctl: unknown request %i\n", cmd);
1933 struct xt_table *ipt_register_table(struct net *net,
1934 const struct xt_table *table,
1935 const struct ipt_replace *repl)
1938 struct xt_table_info *newinfo;
1939 struct xt_table_info bootstrap = {0};
1940 void *loc_cpu_entry;
1941 struct xt_table *new_table;
1943 newinfo = xt_alloc_table_info(repl->size);
1949 loc_cpu_entry = newinfo->entries;
1950 memcpy(loc_cpu_entry, repl->entries, repl->size);
1952 ret = translate_table(net, newinfo, loc_cpu_entry, repl);
1956 new_table = xt_register_table(net, table, &bootstrap, newinfo);
1957 if (IS_ERR(new_table)) {
1958 ret = PTR_ERR(new_table);
1965 xt_free_table_info(newinfo);
1967 return ERR_PTR(ret);
1970 void ipt_unregister_table(struct net *net, struct xt_table *table)
1972 struct xt_table_info *private;
1973 void *loc_cpu_entry;
1974 struct module *table_owner = table->me;
1975 struct ipt_entry *iter;
1977 private = xt_unregister_table(table);
1979 /* Decrease module usage counts and free resources */
1980 loc_cpu_entry = private->entries;
1981 xt_entry_foreach(iter, loc_cpu_entry, private->size)
1982 cleanup_entry(iter, net);
1983 if (private->number > private->initial_entries)
1984 module_put(table_owner);
1985 xt_free_table_info(private);
1988 /* Returns 1 if the type and code is matched by the range, 0 otherwise */
1990 icmp_type_code_match(u_int8_t test_type, u_int8_t min_code, u_int8_t max_code,
1991 u_int8_t type, u_int8_t code,
1994 return ((test_type == 0xFF) ||
1995 (type == test_type && code >= min_code && code <= max_code))
2000 icmp_match(const struct sk_buff *skb, struct xt_action_param *par)
2002 const struct icmphdr *ic;
2003 struct icmphdr _icmph;
2004 const struct ipt_icmp *icmpinfo = par->matchinfo;
2006 /* Must not be a fragment. */
2007 if (par->fragoff != 0)
2010 ic = skb_header_pointer(skb, par->thoff, sizeof(_icmph), &_icmph);
2012 /* We've been asked to examine this packet, and we
2013 * can't. Hence, no choice but to drop.
2015 duprintf("Dropping evil ICMP tinygram.\n");
2016 par->hotdrop = true;
2020 return icmp_type_code_match(icmpinfo->type,
2024 !!(icmpinfo->invflags&IPT_ICMP_INV));
2027 static int icmp_checkentry(const struct xt_mtchk_param *par)
2029 const struct ipt_icmp *icmpinfo = par->matchinfo;
2031 /* Must specify no unknown invflags */
2032 return (icmpinfo->invflags & ~IPT_ICMP_INV) ? -EINVAL : 0;
2035 static struct xt_target ipt_builtin_tg[] __read_mostly = {
2037 .name = XT_STANDARD_TARGET,
2038 .targetsize = sizeof(int),
2039 .family = NFPROTO_IPV4,
2040 #ifdef CONFIG_COMPAT
2041 .compatsize = sizeof(compat_int_t),
2042 .compat_from_user = compat_standard_from_user,
2043 .compat_to_user = compat_standard_to_user,
2047 .name = XT_ERROR_TARGET,
2048 .target = ipt_error,
2049 .targetsize = XT_FUNCTION_MAXNAMELEN,
2050 .family = NFPROTO_IPV4,
2054 static struct nf_sockopt_ops ipt_sockopts = {
2056 .set_optmin = IPT_BASE_CTL,
2057 .set_optmax = IPT_SO_SET_MAX+1,
2058 .set = do_ipt_set_ctl,
2059 #ifdef CONFIG_COMPAT
2060 .compat_set = compat_do_ipt_set_ctl,
2062 .get_optmin = IPT_BASE_CTL,
2063 .get_optmax = IPT_SO_GET_MAX+1,
2064 .get = do_ipt_get_ctl,
2065 #ifdef CONFIG_COMPAT
2066 .compat_get = compat_do_ipt_get_ctl,
2068 .owner = THIS_MODULE,
2071 static struct xt_match ipt_builtin_mt[] __read_mostly = {
2074 .match = icmp_match,
2075 .matchsize = sizeof(struct ipt_icmp),
2076 .checkentry = icmp_checkentry,
2077 .proto = IPPROTO_ICMP,
2078 .family = NFPROTO_IPV4,
2083 static int __net_init ip_tables_net_init(struct net *net)
2085 return xt_proto_init(net, NFPROTO_IPV4);
2088 static void __net_exit ip_tables_net_exit(struct net *net)
2090 xt_proto_fini(net, NFPROTO_IPV4);
2093 static struct pernet_operations ip_tables_net_ops = {
2094 .init = ip_tables_net_init,
2095 .exit = ip_tables_net_exit,
2098 static int __init ip_tables_init(void)
2102 ret = register_pernet_subsys(&ip_tables_net_ops);
2106 /* No one else will be downing sem now, so we won't sleep */
2107 ret = xt_register_targets(ipt_builtin_tg, ARRAY_SIZE(ipt_builtin_tg));
2110 ret = xt_register_matches(ipt_builtin_mt, ARRAY_SIZE(ipt_builtin_mt));
2114 /* Register setsockopt */
2115 ret = nf_register_sockopt(&ipt_sockopts);
2119 pr_info("(C) 2000-2006 Netfilter Core Team\n");
2123 xt_unregister_matches(ipt_builtin_mt, ARRAY_SIZE(ipt_builtin_mt));
2125 xt_unregister_targets(ipt_builtin_tg, ARRAY_SIZE(ipt_builtin_tg));
2127 unregister_pernet_subsys(&ip_tables_net_ops);
2132 static void __exit ip_tables_fini(void)
2134 nf_unregister_sockopt(&ipt_sockopts);
2136 xt_unregister_matches(ipt_builtin_mt, ARRAY_SIZE(ipt_builtin_mt));
2137 xt_unregister_targets(ipt_builtin_tg, ARRAY_SIZE(ipt_builtin_tg));
2138 unregister_pernet_subsys(&ip_tables_net_ops);
2141 EXPORT_SYMBOL(ipt_register_table);
2142 EXPORT_SYMBOL(ipt_unregister_table);
2143 EXPORT_SYMBOL(ipt_do_table);
2144 module_init(ip_tables_init);
2145 module_exit(ip_tables_fini);