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_ip *ip)
173 static const struct ipt_ip uncond;
175 return memcmp(ip, &uncond, sizeof(uncond)) == 0;
179 /* for const-correctness */
180 static inline const struct xt_entry_target *
181 ipt_get_target_c(const struct ipt_entry *e)
183 return ipt_get_target((struct ipt_entry *)e);
186 #if IS_ENABLED(CONFIG_NETFILTER_XT_TARGET_TRACE)
187 static const char *const hooknames[] = {
188 [NF_INET_PRE_ROUTING] = "PREROUTING",
189 [NF_INET_LOCAL_IN] = "INPUT",
190 [NF_INET_FORWARD] = "FORWARD",
191 [NF_INET_LOCAL_OUT] = "OUTPUT",
192 [NF_INET_POST_ROUTING] = "POSTROUTING",
195 enum nf_ip_trace_comments {
196 NF_IP_TRACE_COMMENT_RULE,
197 NF_IP_TRACE_COMMENT_RETURN,
198 NF_IP_TRACE_COMMENT_POLICY,
201 static const char *const comments[] = {
202 [NF_IP_TRACE_COMMENT_RULE] = "rule",
203 [NF_IP_TRACE_COMMENT_RETURN] = "return",
204 [NF_IP_TRACE_COMMENT_POLICY] = "policy",
207 static struct nf_loginfo trace_loginfo = {
208 .type = NF_LOG_TYPE_LOG,
212 .logflags = NF_LOG_MASK,
217 /* Mildly perf critical (only if packet tracing is on) */
219 get_chainname_rulenum(const struct ipt_entry *s, const struct ipt_entry *e,
220 const char *hookname, const char **chainname,
221 const char **comment, unsigned int *rulenum)
223 const struct xt_standard_target *t = (void *)ipt_get_target_c(s);
225 if (strcmp(t->target.u.kernel.target->name, XT_ERROR_TARGET) == 0) {
226 /* Head of user chain: ERROR target with chainname */
227 *chainname = t->target.data;
232 if (s->target_offset == sizeof(struct ipt_entry) &&
233 strcmp(t->target.u.kernel.target->name,
234 XT_STANDARD_TARGET) == 0 &&
236 unconditional(&s->ip)) {
237 /* Tail of chains: STANDARD target (return/policy) */
238 *comment = *chainname == hookname
239 ? comments[NF_IP_TRACE_COMMENT_POLICY]
240 : comments[NF_IP_TRACE_COMMENT_RETURN];
249 static void trace_packet(const struct sk_buff *skb,
251 const struct net_device *in,
252 const struct net_device *out,
253 const char *tablename,
254 const struct xt_table_info *private,
255 const struct ipt_entry *e)
257 const struct ipt_entry *root;
258 const char *hookname, *chainname, *comment;
259 const struct ipt_entry *iter;
260 unsigned int rulenum = 0;
261 struct net *net = dev_net(in ? in : out);
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,
289 const struct nf_hook_state *state,
290 struct xt_table *table)
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 *stackptr, origptr, cpu;
300 const struct xt_table_info *private;
301 struct xt_action_param acpar;
306 indev = state->in ? state->in->name : nulldevname;
307 outdev = state->out ? state->out->name : nulldevname;
308 /* We handle fragments by dealing with the first fragment as
309 * if it was a normal packet. All other fragments are treated
310 * normally, except that they will NEVER match rules that ask
311 * things we don't know, ie. tcp syn flag or ports). If the
312 * rule is also a fragment-specific rule, non-fragments won't
314 acpar.fragoff = ntohs(ip->frag_off) & IP_OFFSET;
315 acpar.thoff = ip_hdrlen(skb);
316 acpar.hotdrop = false;
317 acpar.in = state->in;
318 acpar.out = state->out;
319 acpar.family = NFPROTO_IPV4;
320 acpar.hooknum = hook;
322 IP_NF_ASSERT(table->valid_hooks & (1 << hook));
324 addend = xt_write_recseq_begin();
325 private = table->private;
326 cpu = smp_processor_id();
328 * Ensure we load private-> members after we've fetched the base
331 smp_read_barrier_depends();
332 table_base = private->entries;
333 jumpstack = (struct ipt_entry **)private->jumpstack[cpu];
334 stackptr = per_cpu_ptr(private->stackptr, cpu);
337 e = get_entry(table_base, private->hook_entry[hook]);
339 pr_debug("Entering %s(hook %u); sp at %u (UF %p)\n",
340 table->name, hook, origptr,
341 get_entry(table_base, private->underflow[hook]));
344 const struct xt_entry_target *t;
345 const struct xt_entry_match *ematch;
346 struct xt_counters *counter;
349 if (!ip_packet_match(ip, indev, outdev,
350 &e->ip, acpar.fragoff)) {
352 e = ipt_next_entry(e);
356 xt_ematch_foreach(ematch, e) {
357 acpar.match = ematch->u.kernel.match;
358 acpar.matchinfo = ematch->data;
359 if (!acpar.match->match(skb, &acpar))
363 counter = xt_get_this_cpu_counter(&e->counters);
364 ADD_COUNTER(*counter, skb->len, 1);
366 t = ipt_get_target(e);
367 IP_NF_ASSERT(t->u.kernel.target);
369 #if IS_ENABLED(CONFIG_NETFILTER_XT_TARGET_TRACE)
370 /* The packet is traced: log it */
371 if (unlikely(skb->nf_trace))
372 trace_packet(skb, hook, state->in, state->out,
373 table->name, private, e);
375 /* Standard target? */
376 if (!t->u.kernel.target->target) {
379 v = ((struct xt_standard_target *)t)->verdict;
381 /* Pop from stack? */
382 if (v != XT_RETURN) {
383 verdict = (unsigned int)(-v) - 1;
386 if (*stackptr <= origptr) {
387 e = get_entry(table_base,
388 private->underflow[hook]);
389 pr_debug("Underflow (this is normal) "
392 e = jumpstack[--*stackptr];
393 pr_debug("Pulled %p out from pos %u\n",
395 e = ipt_next_entry(e);
399 if (table_base + v != ipt_next_entry(e) &&
400 !(e->ip.flags & IPT_F_GOTO)) {
401 if (*stackptr >= private->stacksize) {
405 jumpstack[(*stackptr)++] = e;
406 pr_debug("Pushed %p into pos %u\n",
410 e = get_entry(table_base, v);
414 acpar.target = t->u.kernel.target;
415 acpar.targinfo = t->data;
417 verdict = t->u.kernel.target->target(skb, &acpar);
418 /* Target might have changed stuff. */
420 if (verdict == XT_CONTINUE)
421 e = ipt_next_entry(e);
425 } while (!acpar.hotdrop);
426 pr_debug("Exiting %s; resetting sp from %u to %u\n",
427 __func__, *stackptr, origptr);
429 xt_write_recseq_end(addend);
432 #ifdef DEBUG_ALLOW_ALL
441 /* Figures out from what hook each rule can be called: returns 0 if
442 * there are loops. Puts hook bitmask in comefrom.
444 * Keeps track of largest call depth seen and stores it in newinfo->stacksize.
447 mark_source_chains(struct xt_table_info *newinfo,
448 unsigned int valid_hooks, void *entry0)
450 unsigned int calldepth, max_calldepth = 0;
453 /* No recursion; use packet counter to save back ptrs (reset
454 to 0 as we leave), and comefrom to save source hook bitmask */
455 for (hook = 0; hook < NF_INET_NUMHOOKS; hook++) {
456 unsigned int pos = newinfo->hook_entry[hook];
457 struct ipt_entry *e = (struct ipt_entry *)(entry0 + pos);
459 if (!(valid_hooks & (1 << hook)))
462 /* Set initial back pointer. */
463 e->counters.pcnt = pos;
467 const struct xt_standard_target *t
468 = (void *)ipt_get_target_c(e);
469 int visited = e->comefrom & (1 << hook);
471 if (e->comefrom & (1 << NF_INET_NUMHOOKS)) {
472 pr_err("iptables: loop hook %u pos %u %08X.\n",
473 hook, pos, e->comefrom);
476 e->comefrom |= ((1 << hook) | (1 << NF_INET_NUMHOOKS));
478 /* Unconditional return/END. */
479 if ((e->target_offset == sizeof(struct ipt_entry) &&
480 (strcmp(t->target.u.user.name,
481 XT_STANDARD_TARGET) == 0) &&
482 t->verdict < 0 && unconditional(&e->ip)) ||
484 unsigned int oldpos, size;
486 if ((strcmp(t->target.u.user.name,
487 XT_STANDARD_TARGET) == 0) &&
488 t->verdict < -NF_MAX_VERDICT - 1) {
489 duprintf("mark_source_chains: bad "
490 "negative verdict (%i)\n",
495 /* Return: backtrack through the last
498 e->comefrom ^= (1<<NF_INET_NUMHOOKS);
499 #ifdef DEBUG_IP_FIREWALL_USER
501 & (1 << NF_INET_NUMHOOKS)) {
502 duprintf("Back unset "
509 pos = e->counters.pcnt;
510 e->counters.pcnt = 0;
512 /* We're at the start. */
516 e = (struct ipt_entry *)
518 } while (oldpos == pos + e->next_offset);
521 size = e->next_offset;
522 e = (struct ipt_entry *)
523 (entry0 + pos + size);
524 e->counters.pcnt = pos;
526 WARN_ON_ONCE(calldepth == 0);
530 int newpos = t->verdict;
532 if (strcmp(t->target.u.user.name,
533 XT_STANDARD_TARGET) == 0 &&
535 if (newpos > newinfo->size -
536 sizeof(struct ipt_entry)) {
537 duprintf("mark_source_chains: "
538 "bad verdict (%i)\n",
542 if (entry0 + newpos != ipt_next_entry(e) &&
543 !(e->ip.flags & IPT_F_GOTO) &&
544 ++calldepth > max_calldepth)
545 max_calldepth = calldepth;
547 /* This a jump; chase it. */
548 duprintf("Jump rule %u -> %u, calldepth %d\n",
549 pos, newpos, calldepth);
551 /* ... this is a fallthru */
552 newpos = pos + e->next_offset;
554 e = (struct ipt_entry *)
556 e->counters.pcnt = pos;
561 duprintf("Finished chain %u\n", hook);
563 newinfo->stacksize = max_calldepth;
567 static void cleanup_match(struct xt_entry_match *m, struct net *net)
569 struct xt_mtdtor_param par;
572 par.match = m->u.kernel.match;
573 par.matchinfo = m->data;
574 par.family = NFPROTO_IPV4;
575 if (par.match->destroy != NULL)
576 par.match->destroy(&par);
577 module_put(par.match->me);
581 check_entry(const struct ipt_entry *e, const char *name)
583 const struct xt_entry_target *t;
585 if (!ip_checkentry(&e->ip)) {
586 duprintf("ip check failed %p %s.\n", e, name);
590 if (e->target_offset + sizeof(struct xt_entry_target) >
594 t = ipt_get_target_c(e);
595 if (e->target_offset + t->u.target_size > e->next_offset)
602 check_match(struct xt_entry_match *m, struct xt_mtchk_param *par)
604 const struct ipt_ip *ip = par->entryinfo;
607 par->match = m->u.kernel.match;
608 par->matchinfo = m->data;
610 ret = xt_check_match(par, m->u.match_size - sizeof(*m),
611 ip->proto, ip->invflags & IPT_INV_PROTO);
613 duprintf("check failed for `%s'.\n", par->match->name);
620 find_check_match(struct xt_entry_match *m, struct xt_mtchk_param *par)
622 struct xt_match *match;
625 match = xt_request_find_match(NFPROTO_IPV4, m->u.user.name,
628 duprintf("find_check_match: `%s' not found\n", m->u.user.name);
629 return PTR_ERR(match);
631 m->u.kernel.match = match;
633 ret = check_match(m, par);
639 module_put(m->u.kernel.match->me);
643 static int check_target(struct ipt_entry *e, struct net *net, const char *name)
645 struct xt_entry_target *t = ipt_get_target(e);
646 struct xt_tgchk_param par = {
650 .target = t->u.kernel.target,
652 .hook_mask = e->comefrom,
653 .family = NFPROTO_IPV4,
657 ret = xt_check_target(&par, t->u.target_size - sizeof(*t),
658 e->ip.proto, e->ip.invflags & IPT_INV_PROTO);
660 duprintf("check failed for `%s'.\n",
661 t->u.kernel.target->name);
668 find_check_entry(struct ipt_entry *e, struct net *net, const char *name,
671 struct xt_entry_target *t;
672 struct xt_target *target;
675 struct xt_mtchk_param mtpar;
676 struct xt_entry_match *ematch;
678 ret = check_entry(e, name);
682 e->counters.pcnt = xt_percpu_counter_alloc();
683 if (IS_ERR_VALUE(e->counters.pcnt))
689 mtpar.entryinfo = &e->ip;
690 mtpar.hook_mask = e->comefrom;
691 mtpar.family = NFPROTO_IPV4;
692 xt_ematch_foreach(ematch, e) {
693 ret = find_check_match(ematch, &mtpar);
695 goto cleanup_matches;
699 t = ipt_get_target(e);
700 target = xt_request_find_target(NFPROTO_IPV4, t->u.user.name,
702 if (IS_ERR(target)) {
703 duprintf("find_check_entry: `%s' not found\n", t->u.user.name);
704 ret = PTR_ERR(target);
705 goto cleanup_matches;
707 t->u.kernel.target = target;
709 ret = check_target(e, net, name);
715 module_put(t->u.kernel.target->me);
717 xt_ematch_foreach(ematch, e) {
720 cleanup_match(ematch, net);
723 xt_percpu_counter_free(e->counters.pcnt);
728 static bool check_underflow(const struct ipt_entry *e)
730 const struct xt_entry_target *t;
731 unsigned int verdict;
733 if (!unconditional(&e->ip))
735 t = ipt_get_target_c(e);
736 if (strcmp(t->u.user.name, XT_STANDARD_TARGET) != 0)
738 verdict = ((struct xt_standard_target *)t)->verdict;
739 verdict = -verdict - 1;
740 return verdict == NF_DROP || verdict == NF_ACCEPT;
744 check_entry_size_and_hooks(struct ipt_entry *e,
745 struct xt_table_info *newinfo,
746 const unsigned char *base,
747 const unsigned char *limit,
748 const unsigned int *hook_entries,
749 const unsigned int *underflows,
750 unsigned int valid_hooks)
754 if ((unsigned long)e % __alignof__(struct ipt_entry) != 0 ||
755 (unsigned char *)e + sizeof(struct ipt_entry) >= limit) {
756 duprintf("Bad offset %p\n", e);
761 < sizeof(struct ipt_entry) + sizeof(struct xt_entry_target)) {
762 duprintf("checking: element %p size %u\n",
767 /* Check hooks & underflows */
768 for (h = 0; h < NF_INET_NUMHOOKS; h++) {
769 if (!(valid_hooks & (1 << h)))
771 if ((unsigned char *)e - base == hook_entries[h])
772 newinfo->hook_entry[h] = hook_entries[h];
773 if ((unsigned char *)e - base == underflows[h]) {
774 if (!check_underflow(e)) {
775 pr_err("Underflows must be unconditional and "
776 "use the STANDARD target with "
780 newinfo->underflow[h] = underflows[h];
784 /* Clear counters and comefrom */
785 e->counters = ((struct xt_counters) { 0, 0 });
791 cleanup_entry(struct ipt_entry *e, struct net *net)
793 struct xt_tgdtor_param par;
794 struct xt_entry_target *t;
795 struct xt_entry_match *ematch;
797 /* Cleanup all matches */
798 xt_ematch_foreach(ematch, e)
799 cleanup_match(ematch, net);
800 t = ipt_get_target(e);
803 par.target = t->u.kernel.target;
804 par.targinfo = t->data;
805 par.family = NFPROTO_IPV4;
806 if (par.target->destroy != NULL)
807 par.target->destroy(&par);
808 module_put(par.target->me);
809 xt_percpu_counter_free(e->counters.pcnt);
812 /* Checks and translates the user-supplied table segment (held in
815 translate_table(struct net *net, struct xt_table_info *newinfo, void *entry0,
816 const struct ipt_replace *repl)
818 struct ipt_entry *iter;
822 newinfo->size = repl->size;
823 newinfo->number = repl->num_entries;
825 /* Init all hooks to impossible value. */
826 for (i = 0; i < NF_INET_NUMHOOKS; i++) {
827 newinfo->hook_entry[i] = 0xFFFFFFFF;
828 newinfo->underflow[i] = 0xFFFFFFFF;
831 duprintf("translate_table: size %u\n", newinfo->size);
833 /* Walk through entries, checking offsets. */
834 xt_entry_foreach(iter, entry0, newinfo->size) {
835 ret = check_entry_size_and_hooks(iter, newinfo, entry0,
845 if (i != repl->num_entries) {
846 duprintf("translate_table: %u not %u entries\n",
847 i, repl->num_entries);
851 /* Check hooks all assigned */
852 for (i = 0; i < NF_INET_NUMHOOKS; i++) {
853 /* Only hooks which are valid */
854 if (!(repl->valid_hooks & (1 << i)))
856 if (newinfo->hook_entry[i] == 0xFFFFFFFF) {
857 duprintf("Invalid hook entry %u %u\n",
858 i, repl->hook_entry[i]);
861 if (newinfo->underflow[i] == 0xFFFFFFFF) {
862 duprintf("Invalid underflow %u %u\n",
863 i, repl->underflow[i]);
868 if (!mark_source_chains(newinfo, repl->valid_hooks, entry0))
871 /* Finally, each sanity check must pass */
873 xt_entry_foreach(iter, entry0, newinfo->size) {
874 ret = find_check_entry(iter, net, repl->name, repl->size);
881 xt_entry_foreach(iter, entry0, newinfo->size) {
884 cleanup_entry(iter, net);
893 get_counters(const struct xt_table_info *t,
894 struct xt_counters counters[])
896 struct ipt_entry *iter;
900 for_each_possible_cpu(cpu) {
901 seqcount_t *s = &per_cpu(xt_recseq, cpu);
904 xt_entry_foreach(iter, t->entries, t->size) {
905 struct xt_counters *tmp;
909 tmp = xt_get_per_cpu_counter(&iter->counters, cpu);
911 start = read_seqcount_begin(s);
914 } while (read_seqcount_retry(s, start));
916 ADD_COUNTER(counters[i], bcnt, pcnt);
917 ++i; /* macro does multi eval of i */
922 static struct xt_counters *alloc_counters(const struct xt_table *table)
924 unsigned int countersize;
925 struct xt_counters *counters;
926 const struct xt_table_info *private = table->private;
928 /* We need atomic snapshot of counters: rest doesn't change
929 (other than comefrom, which userspace doesn't care
931 countersize = sizeof(struct xt_counters) * private->number;
932 counters = vzalloc(countersize);
934 if (counters == NULL)
935 return ERR_PTR(-ENOMEM);
937 get_counters(private, counters);
943 copy_entries_to_user(unsigned int total_size,
944 const struct xt_table *table,
945 void __user *userptr)
947 unsigned int off, num;
948 const struct ipt_entry *e;
949 struct xt_counters *counters;
950 const struct xt_table_info *private = table->private;
952 const void *loc_cpu_entry;
954 counters = alloc_counters(table);
955 if (IS_ERR(counters))
956 return PTR_ERR(counters);
958 loc_cpu_entry = private->entries;
959 if (copy_to_user(userptr, loc_cpu_entry, total_size) != 0) {
964 /* FIXME: use iterator macros --RR */
965 /* ... then go back and fix counters and names */
966 for (off = 0, num = 0; off < total_size; off += e->next_offset, num++){
968 const struct xt_entry_match *m;
969 const struct xt_entry_target *t;
971 e = (struct ipt_entry *)(loc_cpu_entry + off);
972 if (copy_to_user(userptr + off
973 + offsetof(struct ipt_entry, counters),
975 sizeof(counters[num])) != 0) {
980 for (i = sizeof(struct ipt_entry);
981 i < e->target_offset;
982 i += m->u.match_size) {
985 if (copy_to_user(userptr + off + i
986 + offsetof(struct xt_entry_match,
988 m->u.kernel.match->name,
989 strlen(m->u.kernel.match->name)+1)
996 t = ipt_get_target_c(e);
997 if (copy_to_user(userptr + off + e->target_offset
998 + offsetof(struct xt_entry_target,
1000 t->u.kernel.target->name,
1001 strlen(t->u.kernel.target->name)+1) != 0) {
1012 #ifdef CONFIG_COMPAT
1013 static void compat_standard_from_user(void *dst, const void *src)
1015 int v = *(compat_int_t *)src;
1018 v += xt_compat_calc_jump(AF_INET, v);
1019 memcpy(dst, &v, sizeof(v));
1022 static int compat_standard_to_user(void __user *dst, const void *src)
1024 compat_int_t cv = *(int *)src;
1027 cv -= xt_compat_calc_jump(AF_INET, cv);
1028 return copy_to_user(dst, &cv, sizeof(cv)) ? -EFAULT : 0;
1031 static int compat_calc_entry(const struct ipt_entry *e,
1032 const struct xt_table_info *info,
1033 const void *base, struct xt_table_info *newinfo)
1035 const struct xt_entry_match *ematch;
1036 const struct xt_entry_target *t;
1037 unsigned int entry_offset;
1040 off = sizeof(struct ipt_entry) - sizeof(struct compat_ipt_entry);
1041 entry_offset = (void *)e - base;
1042 xt_ematch_foreach(ematch, e)
1043 off += xt_compat_match_offset(ematch->u.kernel.match);
1044 t = ipt_get_target_c(e);
1045 off += xt_compat_target_offset(t->u.kernel.target);
1046 newinfo->size -= off;
1047 ret = xt_compat_add_offset(AF_INET, entry_offset, off);
1051 for (i = 0; i < NF_INET_NUMHOOKS; i++) {
1052 if (info->hook_entry[i] &&
1053 (e < (struct ipt_entry *)(base + info->hook_entry[i])))
1054 newinfo->hook_entry[i] -= off;
1055 if (info->underflow[i] &&
1056 (e < (struct ipt_entry *)(base + info->underflow[i])))
1057 newinfo->underflow[i] -= off;
1062 static int compat_table_info(const struct xt_table_info *info,
1063 struct xt_table_info *newinfo)
1065 struct ipt_entry *iter;
1066 const void *loc_cpu_entry;
1069 if (!newinfo || !info)
1072 /* we dont care about newinfo->entries */
1073 memcpy(newinfo, info, offsetof(struct xt_table_info, entries));
1074 newinfo->initial_entries = 0;
1075 loc_cpu_entry = info->entries;
1076 xt_compat_init_offsets(AF_INET, info->number);
1077 xt_entry_foreach(iter, loc_cpu_entry, info->size) {
1078 ret = compat_calc_entry(iter, info, loc_cpu_entry, newinfo);
1086 static int get_info(struct net *net, void __user *user,
1087 const int *len, int compat)
1089 char name[XT_TABLE_MAXNAMELEN];
1093 if (*len != sizeof(struct ipt_getinfo)) {
1094 duprintf("length %u != %zu\n", *len,
1095 sizeof(struct ipt_getinfo));
1099 if (copy_from_user(name, user, sizeof(name)) != 0)
1102 name[XT_TABLE_MAXNAMELEN-1] = '\0';
1103 #ifdef CONFIG_COMPAT
1105 xt_compat_lock(AF_INET);
1107 t = try_then_request_module(xt_find_table_lock(net, AF_INET, name),
1108 "iptable_%s", name);
1109 if (!IS_ERR_OR_NULL(t)) {
1110 struct ipt_getinfo info;
1111 const struct xt_table_info *private = t->private;
1112 #ifdef CONFIG_COMPAT
1113 struct xt_table_info tmp;
1116 ret = compat_table_info(private, &tmp);
1117 xt_compat_flush_offsets(AF_INET);
1121 memset(&info, 0, sizeof(info));
1122 info.valid_hooks = t->valid_hooks;
1123 memcpy(info.hook_entry, private->hook_entry,
1124 sizeof(info.hook_entry));
1125 memcpy(info.underflow, private->underflow,
1126 sizeof(info.underflow));
1127 info.num_entries = private->number;
1128 info.size = private->size;
1129 strcpy(info.name, name);
1131 if (copy_to_user(user, &info, *len) != 0)
1139 ret = t ? PTR_ERR(t) : -ENOENT;
1140 #ifdef CONFIG_COMPAT
1142 xt_compat_unlock(AF_INET);
1148 get_entries(struct net *net, struct ipt_get_entries __user *uptr,
1152 struct ipt_get_entries get;
1155 if (*len < sizeof(get)) {
1156 duprintf("get_entries: %u < %zu\n", *len, sizeof(get));
1159 if (copy_from_user(&get, uptr, sizeof(get)) != 0)
1161 if (*len != sizeof(struct ipt_get_entries) + get.size) {
1162 duprintf("get_entries: %u != %zu\n",
1163 *len, sizeof(get) + get.size);
1167 t = xt_find_table_lock(net, AF_INET, get.name);
1168 if (!IS_ERR_OR_NULL(t)) {
1169 const struct xt_table_info *private = t->private;
1170 duprintf("t->private->number = %u\n", private->number);
1171 if (get.size == private->size)
1172 ret = copy_entries_to_user(private->size,
1173 t, uptr->entrytable);
1175 duprintf("get_entries: I've got %u not %u!\n",
1176 private->size, get.size);
1182 ret = t ? PTR_ERR(t) : -ENOENT;
1188 __do_replace(struct net *net, const char *name, unsigned int valid_hooks,
1189 struct xt_table_info *newinfo, unsigned int num_counters,
1190 void __user *counters_ptr)
1194 struct xt_table_info *oldinfo;
1195 struct xt_counters *counters;
1196 struct ipt_entry *iter;
1199 counters = vzalloc(num_counters * sizeof(struct xt_counters));
1205 t = try_then_request_module(xt_find_table_lock(net, AF_INET, name),
1206 "iptable_%s", name);
1207 if (IS_ERR_OR_NULL(t)) {
1208 ret = t ? PTR_ERR(t) : -ENOENT;
1209 goto free_newinfo_counters_untrans;
1213 if (valid_hooks != t->valid_hooks) {
1214 duprintf("Valid hook crap: %08X vs %08X\n",
1215 valid_hooks, t->valid_hooks);
1220 oldinfo = xt_replace_table(t, num_counters, newinfo, &ret);
1224 /* Update module usage count based on number of rules */
1225 duprintf("do_replace: oldnum=%u, initnum=%u, newnum=%u\n",
1226 oldinfo->number, oldinfo->initial_entries, newinfo->number);
1227 if ((oldinfo->number > oldinfo->initial_entries) ||
1228 (newinfo->number <= oldinfo->initial_entries))
1230 if ((oldinfo->number > oldinfo->initial_entries) &&
1231 (newinfo->number <= oldinfo->initial_entries))
1234 /* Get the old counters, and synchronize with replace */
1235 get_counters(oldinfo, counters);
1237 /* Decrease module usage counts and free resource */
1238 xt_entry_foreach(iter, oldinfo->entries, oldinfo->size)
1239 cleanup_entry(iter, net);
1241 xt_free_table_info(oldinfo);
1242 if (copy_to_user(counters_ptr, counters,
1243 sizeof(struct xt_counters) * num_counters) != 0) {
1244 /* Silent error, can't fail, new table is already in place */
1245 net_warn_ratelimited("iptables: counters copy to user failed while replacing table\n");
1254 free_newinfo_counters_untrans:
1261 do_replace(struct net *net, const void __user *user, unsigned int len)
1264 struct ipt_replace tmp;
1265 struct xt_table_info *newinfo;
1266 void *loc_cpu_entry;
1267 struct ipt_entry *iter;
1269 if (copy_from_user(&tmp, user, sizeof(tmp)) != 0)
1272 /* overflow check */
1273 if (tmp.num_counters >= INT_MAX / sizeof(struct xt_counters))
1275 if (tmp.num_counters == 0)
1278 tmp.name[sizeof(tmp.name)-1] = 0;
1280 newinfo = xt_alloc_table_info(tmp.size);
1284 loc_cpu_entry = newinfo->entries;
1285 if (copy_from_user(loc_cpu_entry, user + sizeof(tmp),
1291 ret = translate_table(net, newinfo, loc_cpu_entry, &tmp);
1295 duprintf("Translated table\n");
1297 ret = __do_replace(net, tmp.name, tmp.valid_hooks, newinfo,
1298 tmp.num_counters, tmp.counters);
1300 goto free_newinfo_untrans;
1303 free_newinfo_untrans:
1304 xt_entry_foreach(iter, loc_cpu_entry, newinfo->size)
1305 cleanup_entry(iter, net);
1307 xt_free_table_info(newinfo);
1312 do_add_counters(struct net *net, const void __user *user,
1313 unsigned int len, int compat)
1316 struct xt_counters_info tmp;
1317 struct xt_counters *paddc;
1318 unsigned int num_counters;
1323 const struct xt_table_info *private;
1325 struct ipt_entry *iter;
1326 unsigned int addend;
1327 #ifdef CONFIG_COMPAT
1328 struct compat_xt_counters_info compat_tmp;
1332 size = sizeof(struct compat_xt_counters_info);
1337 size = sizeof(struct xt_counters_info);
1340 if (copy_from_user(ptmp, user, size) != 0)
1343 #ifdef CONFIG_COMPAT
1345 num_counters = compat_tmp.num_counters;
1346 name = compat_tmp.name;
1350 num_counters = tmp.num_counters;
1354 if (len != size + num_counters * sizeof(struct xt_counters))
1357 paddc = vmalloc(len - size);
1361 if (copy_from_user(paddc, user + size, len - size) != 0) {
1366 t = xt_find_table_lock(net, AF_INET, name);
1367 if (IS_ERR_OR_NULL(t)) {
1368 ret = t ? PTR_ERR(t) : -ENOENT;
1373 private = t->private;
1374 if (private->number != num_counters) {
1376 goto unlock_up_free;
1380 addend = xt_write_recseq_begin();
1381 xt_entry_foreach(iter, private->entries, private->size) {
1382 struct xt_counters *tmp;
1384 tmp = xt_get_this_cpu_counter(&iter->counters);
1385 ADD_COUNTER(*tmp, paddc[i].bcnt, paddc[i].pcnt);
1388 xt_write_recseq_end(addend);
1399 #ifdef CONFIG_COMPAT
1400 struct compat_ipt_replace {
1401 char name[XT_TABLE_MAXNAMELEN];
1405 u32 hook_entry[NF_INET_NUMHOOKS];
1406 u32 underflow[NF_INET_NUMHOOKS];
1408 compat_uptr_t counters; /* struct xt_counters * */
1409 struct compat_ipt_entry entries[0];
1413 compat_copy_entry_to_user(struct ipt_entry *e, void __user **dstptr,
1414 unsigned int *size, struct xt_counters *counters,
1417 struct xt_entry_target *t;
1418 struct compat_ipt_entry __user *ce;
1419 u_int16_t target_offset, next_offset;
1420 compat_uint_t origsize;
1421 const struct xt_entry_match *ematch;
1425 ce = (struct compat_ipt_entry __user *)*dstptr;
1426 if (copy_to_user(ce, e, sizeof(struct ipt_entry)) != 0 ||
1427 copy_to_user(&ce->counters, &counters[i],
1428 sizeof(counters[i])) != 0)
1431 *dstptr += sizeof(struct compat_ipt_entry);
1432 *size -= sizeof(struct ipt_entry) - sizeof(struct compat_ipt_entry);
1434 xt_ematch_foreach(ematch, e) {
1435 ret = xt_compat_match_to_user(ematch, dstptr, size);
1439 target_offset = e->target_offset - (origsize - *size);
1440 t = ipt_get_target(e);
1441 ret = xt_compat_target_to_user(t, dstptr, size);
1444 next_offset = e->next_offset - (origsize - *size);
1445 if (put_user(target_offset, &ce->target_offset) != 0 ||
1446 put_user(next_offset, &ce->next_offset) != 0)
1452 compat_find_calc_match(struct xt_entry_match *m,
1454 const struct ipt_ip *ip,
1457 struct xt_match *match;
1459 match = xt_request_find_match(NFPROTO_IPV4, m->u.user.name,
1460 m->u.user.revision);
1461 if (IS_ERR(match)) {
1462 duprintf("compat_check_calc_match: `%s' not found\n",
1464 return PTR_ERR(match);
1466 m->u.kernel.match = match;
1467 *size += xt_compat_match_offset(match);
1471 static void compat_release_entry(struct compat_ipt_entry *e)
1473 struct xt_entry_target *t;
1474 struct xt_entry_match *ematch;
1476 /* Cleanup all matches */
1477 xt_ematch_foreach(ematch, e)
1478 module_put(ematch->u.kernel.match->me);
1479 t = compat_ipt_get_target(e);
1480 module_put(t->u.kernel.target->me);
1484 check_compat_entry_size_and_hooks(struct compat_ipt_entry *e,
1485 struct xt_table_info *newinfo,
1487 const unsigned char *base,
1488 const unsigned char *limit,
1489 const unsigned int *hook_entries,
1490 const unsigned int *underflows,
1493 struct xt_entry_match *ematch;
1494 struct xt_entry_target *t;
1495 struct xt_target *target;
1496 unsigned int entry_offset;
1500 duprintf("check_compat_entry_size_and_hooks %p\n", e);
1501 if ((unsigned long)e % __alignof__(struct compat_ipt_entry) != 0 ||
1502 (unsigned char *)e + sizeof(struct compat_ipt_entry) >= 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 /* For purposes of check_entry casting the compat entry is fine */
1515 ret = check_entry((struct ipt_entry *)e, name);
1519 off = sizeof(struct ipt_entry) - sizeof(struct compat_ipt_entry);
1520 entry_offset = (void *)e - (void *)base;
1522 xt_ematch_foreach(ematch, e) {
1523 ret = compat_find_calc_match(ematch, name, &e->ip, &off);
1525 goto release_matches;
1529 t = compat_ipt_get_target(e);
1530 target = xt_request_find_target(NFPROTO_IPV4, t->u.user.name,
1531 t->u.user.revision);
1532 if (IS_ERR(target)) {
1533 duprintf("check_compat_entry_size_and_hooks: `%s' not found\n",
1535 ret = PTR_ERR(target);
1536 goto release_matches;
1538 t->u.kernel.target = target;
1540 off += xt_compat_target_offset(target);
1542 ret = xt_compat_add_offset(AF_INET, entry_offset, off);
1546 /* Check hooks & underflows */
1547 for (h = 0; h < NF_INET_NUMHOOKS; h++) {
1548 if ((unsigned char *)e - base == hook_entries[h])
1549 newinfo->hook_entry[h] = hook_entries[h];
1550 if ((unsigned char *)e - base == underflows[h])
1551 newinfo->underflow[h] = underflows[h];
1554 /* Clear counters and comefrom */
1555 memset(&e->counters, 0, sizeof(e->counters));
1560 module_put(t->u.kernel.target->me);
1562 xt_ematch_foreach(ematch, e) {
1565 module_put(ematch->u.kernel.match->me);
1571 compat_copy_entry_from_user(struct compat_ipt_entry *e, void **dstptr,
1572 unsigned int *size, const char *name,
1573 struct xt_table_info *newinfo, unsigned char *base)
1575 struct xt_entry_target *t;
1576 struct xt_target *target;
1577 struct ipt_entry *de;
1578 unsigned int origsize;
1580 struct xt_entry_match *ematch;
1584 de = (struct ipt_entry *)*dstptr;
1585 memcpy(de, e, sizeof(struct ipt_entry));
1586 memcpy(&de->counters, &e->counters, sizeof(e->counters));
1588 *dstptr += sizeof(struct ipt_entry);
1589 *size += sizeof(struct ipt_entry) - sizeof(struct compat_ipt_entry);
1591 xt_ematch_foreach(ematch, e) {
1592 ret = xt_compat_match_from_user(ematch, dstptr, size);
1596 de->target_offset = e->target_offset - (origsize - *size);
1597 t = compat_ipt_get_target(e);
1598 target = t->u.kernel.target;
1599 xt_compat_target_from_user(t, dstptr, size);
1601 de->next_offset = e->next_offset - (origsize - *size);
1602 for (h = 0; h < NF_INET_NUMHOOKS; h++) {
1603 if ((unsigned char *)de - base < newinfo->hook_entry[h])
1604 newinfo->hook_entry[h] -= origsize - *size;
1605 if ((unsigned char *)de - base < newinfo->underflow[h])
1606 newinfo->underflow[h] -= origsize - *size;
1612 compat_check_entry(struct ipt_entry *e, struct net *net, const char *name)
1614 struct xt_entry_match *ematch;
1615 struct xt_mtchk_param mtpar;
1619 e->counters.pcnt = xt_percpu_counter_alloc();
1620 if (IS_ERR_VALUE(e->counters.pcnt))
1626 mtpar.entryinfo = &e->ip;
1627 mtpar.hook_mask = e->comefrom;
1628 mtpar.family = NFPROTO_IPV4;
1629 xt_ematch_foreach(ematch, e) {
1630 ret = check_match(ematch, &mtpar);
1632 goto cleanup_matches;
1636 ret = check_target(e, net, name);
1638 goto cleanup_matches;
1642 xt_ematch_foreach(ematch, e) {
1645 cleanup_match(ematch, net);
1648 xt_percpu_counter_free(e->counters.pcnt);
1654 translate_compat_table(struct net *net,
1656 unsigned int valid_hooks,
1657 struct xt_table_info **pinfo,
1659 unsigned int total_size,
1660 unsigned int number,
1661 unsigned int *hook_entries,
1662 unsigned int *underflows)
1665 struct xt_table_info *newinfo, *info;
1666 void *pos, *entry0, *entry1;
1667 struct compat_ipt_entry *iter0;
1668 struct ipt_entry *iter1;
1675 info->number = number;
1677 /* Init all hooks to impossible value. */
1678 for (i = 0; i < NF_INET_NUMHOOKS; i++) {
1679 info->hook_entry[i] = 0xFFFFFFFF;
1680 info->underflow[i] = 0xFFFFFFFF;
1683 duprintf("translate_compat_table: size %u\n", info->size);
1685 xt_compat_lock(AF_INET);
1686 xt_compat_init_offsets(AF_INET, number);
1687 /* Walk through entries, checking offsets. */
1688 xt_entry_foreach(iter0, entry0, total_size) {
1689 ret = check_compat_entry_size_and_hooks(iter0, info, &size,
1691 entry0 + total_size,
1702 duprintf("translate_compat_table: %u not %u entries\n",
1707 /* Check hooks all assigned */
1708 for (i = 0; i < NF_INET_NUMHOOKS; i++) {
1709 /* Only hooks which are valid */
1710 if (!(valid_hooks & (1 << i)))
1712 if (info->hook_entry[i] == 0xFFFFFFFF) {
1713 duprintf("Invalid hook entry %u %u\n",
1714 i, hook_entries[i]);
1717 if (info->underflow[i] == 0xFFFFFFFF) {
1718 duprintf("Invalid underflow %u %u\n",
1725 newinfo = xt_alloc_table_info(size);
1729 newinfo->number = number;
1730 for (i = 0; i < NF_INET_NUMHOOKS; i++) {
1731 newinfo->hook_entry[i] = info->hook_entry[i];
1732 newinfo->underflow[i] = info->underflow[i];
1734 entry1 = newinfo->entries;
1737 xt_entry_foreach(iter0, entry0, total_size) {
1738 ret = compat_copy_entry_from_user(iter0, &pos, &size,
1739 name, newinfo, entry1);
1743 xt_compat_flush_offsets(AF_INET);
1744 xt_compat_unlock(AF_INET);
1749 if (!mark_source_chains(newinfo, valid_hooks, entry1))
1753 xt_entry_foreach(iter1, entry1, newinfo->size) {
1754 ret = compat_check_entry(iter1, net, name);
1761 * The first i matches need cleanup_entry (calls ->destroy)
1762 * because they had called ->check already. The other j-i
1763 * entries need only release.
1767 xt_entry_foreach(iter0, entry0, newinfo->size) {
1772 compat_release_entry(iter0);
1774 xt_entry_foreach(iter1, entry1, newinfo->size) {
1777 cleanup_entry(iter1, net);
1779 xt_free_table_info(newinfo);
1785 xt_free_table_info(info);
1789 xt_free_table_info(newinfo);
1791 xt_entry_foreach(iter0, entry0, total_size) {
1794 compat_release_entry(iter0);
1798 xt_compat_flush_offsets(AF_INET);
1799 xt_compat_unlock(AF_INET);
1804 compat_do_replace(struct net *net, void __user *user, unsigned int len)
1807 struct compat_ipt_replace tmp;
1808 struct xt_table_info *newinfo;
1809 void *loc_cpu_entry;
1810 struct ipt_entry *iter;
1812 if (copy_from_user(&tmp, user, sizeof(tmp)) != 0)
1815 /* overflow check */
1816 if (tmp.size >= INT_MAX / num_possible_cpus())
1818 if (tmp.num_counters >= INT_MAX / sizeof(struct xt_counters))
1820 if (tmp.num_counters == 0)
1823 tmp.name[sizeof(tmp.name)-1] = 0;
1825 newinfo = xt_alloc_table_info(tmp.size);
1829 loc_cpu_entry = newinfo->entries;
1830 if (copy_from_user(loc_cpu_entry, user + sizeof(tmp),
1836 ret = translate_compat_table(net, tmp.name, tmp.valid_hooks,
1837 &newinfo, &loc_cpu_entry, tmp.size,
1838 tmp.num_entries, tmp.hook_entry,
1843 duprintf("compat_do_replace: Translated table\n");
1845 ret = __do_replace(net, tmp.name, tmp.valid_hooks, newinfo,
1846 tmp.num_counters, compat_ptr(tmp.counters));
1848 goto free_newinfo_untrans;
1851 free_newinfo_untrans:
1852 xt_entry_foreach(iter, loc_cpu_entry, newinfo->size)
1853 cleanup_entry(iter, net);
1855 xt_free_table_info(newinfo);
1860 compat_do_ipt_set_ctl(struct sock *sk, int cmd, void __user *user,
1865 if (!ns_capable(sock_net(sk)->user_ns, CAP_NET_ADMIN))
1869 case IPT_SO_SET_REPLACE:
1870 ret = compat_do_replace(sock_net(sk), user, len);
1873 case IPT_SO_SET_ADD_COUNTERS:
1874 ret = do_add_counters(sock_net(sk), user, len, 1);
1878 duprintf("do_ipt_set_ctl: unknown request %i\n", cmd);
1885 struct compat_ipt_get_entries {
1886 char name[XT_TABLE_MAXNAMELEN];
1888 struct compat_ipt_entry entrytable[0];
1892 compat_copy_entries_to_user(unsigned int total_size, struct xt_table *table,
1893 void __user *userptr)
1895 struct xt_counters *counters;
1896 const struct xt_table_info *private = table->private;
1901 struct ipt_entry *iter;
1903 counters = alloc_counters(table);
1904 if (IS_ERR(counters))
1905 return PTR_ERR(counters);
1909 xt_entry_foreach(iter, private->entries, total_size) {
1910 ret = compat_copy_entry_to_user(iter, &pos,
1911 &size, counters, i++);
1921 compat_get_entries(struct net *net, struct compat_ipt_get_entries __user *uptr,
1925 struct compat_ipt_get_entries get;
1928 if (*len < sizeof(get)) {
1929 duprintf("compat_get_entries: %u < %zu\n", *len, sizeof(get));
1933 if (copy_from_user(&get, uptr, sizeof(get)) != 0)
1936 if (*len != sizeof(struct compat_ipt_get_entries) + get.size) {
1937 duprintf("compat_get_entries: %u != %zu\n",
1938 *len, sizeof(get) + get.size);
1942 xt_compat_lock(AF_INET);
1943 t = xt_find_table_lock(net, AF_INET, get.name);
1944 if (!IS_ERR_OR_NULL(t)) {
1945 const struct xt_table_info *private = t->private;
1946 struct xt_table_info info;
1947 duprintf("t->private->number = %u\n", private->number);
1948 ret = compat_table_info(private, &info);
1949 if (!ret && get.size == info.size) {
1950 ret = compat_copy_entries_to_user(private->size,
1951 t, uptr->entrytable);
1953 duprintf("compat_get_entries: I've got %u not %u!\n",
1954 private->size, get.size);
1957 xt_compat_flush_offsets(AF_INET);
1961 ret = t ? PTR_ERR(t) : -ENOENT;
1963 xt_compat_unlock(AF_INET);
1967 static int do_ipt_get_ctl(struct sock *, int, void __user *, int *);
1970 compat_do_ipt_get_ctl(struct sock *sk, int cmd, void __user *user, int *len)
1974 if (!ns_capable(sock_net(sk)->user_ns, CAP_NET_ADMIN))
1978 case IPT_SO_GET_INFO:
1979 ret = get_info(sock_net(sk), user, len, 1);
1981 case IPT_SO_GET_ENTRIES:
1982 ret = compat_get_entries(sock_net(sk), user, len);
1985 ret = do_ipt_get_ctl(sk, cmd, user, len);
1992 do_ipt_set_ctl(struct sock *sk, int cmd, void __user *user, unsigned int len)
1996 if (!ns_capable(sock_net(sk)->user_ns, CAP_NET_ADMIN))
2000 case IPT_SO_SET_REPLACE:
2001 ret = do_replace(sock_net(sk), user, len);
2004 case IPT_SO_SET_ADD_COUNTERS:
2005 ret = do_add_counters(sock_net(sk), user, len, 0);
2009 duprintf("do_ipt_set_ctl: unknown request %i\n", cmd);
2017 do_ipt_get_ctl(struct sock *sk, int cmd, void __user *user, int *len)
2021 if (!ns_capable(sock_net(sk)->user_ns, CAP_NET_ADMIN))
2025 case IPT_SO_GET_INFO:
2026 ret = get_info(sock_net(sk), user, len, 0);
2029 case IPT_SO_GET_ENTRIES:
2030 ret = get_entries(sock_net(sk), user, len);
2033 case IPT_SO_GET_REVISION_MATCH:
2034 case IPT_SO_GET_REVISION_TARGET: {
2035 struct xt_get_revision rev;
2038 if (*len != sizeof(rev)) {
2042 if (copy_from_user(&rev, user, sizeof(rev)) != 0) {
2046 rev.name[sizeof(rev.name)-1] = 0;
2048 if (cmd == IPT_SO_GET_REVISION_TARGET)
2053 try_then_request_module(xt_find_revision(AF_INET, rev.name,
2056 "ipt_%s", rev.name);
2061 duprintf("do_ipt_get_ctl: unknown request %i\n", cmd);
2068 struct xt_table *ipt_register_table(struct net *net,
2069 const struct xt_table *table,
2070 const struct ipt_replace *repl)
2073 struct xt_table_info *newinfo;
2074 struct xt_table_info bootstrap = {0};
2075 void *loc_cpu_entry;
2076 struct xt_table *new_table;
2078 newinfo = xt_alloc_table_info(repl->size);
2084 loc_cpu_entry = newinfo->entries;
2085 memcpy(loc_cpu_entry, repl->entries, repl->size);
2087 ret = translate_table(net, newinfo, loc_cpu_entry, repl);
2091 new_table = xt_register_table(net, table, &bootstrap, newinfo);
2092 if (IS_ERR(new_table)) {
2093 ret = PTR_ERR(new_table);
2100 xt_free_table_info(newinfo);
2102 return ERR_PTR(ret);
2105 void ipt_unregister_table(struct net *net, struct xt_table *table)
2107 struct xt_table_info *private;
2108 void *loc_cpu_entry;
2109 struct module *table_owner = table->me;
2110 struct ipt_entry *iter;
2112 private = xt_unregister_table(table);
2114 /* Decrease module usage counts and free resources */
2115 loc_cpu_entry = private->entries;
2116 xt_entry_foreach(iter, loc_cpu_entry, private->size)
2117 cleanup_entry(iter, net);
2118 if (private->number > private->initial_entries)
2119 module_put(table_owner);
2120 xt_free_table_info(private);
2123 /* Returns 1 if the type and code is matched by the range, 0 otherwise */
2125 icmp_type_code_match(u_int8_t test_type, u_int8_t min_code, u_int8_t max_code,
2126 u_int8_t type, u_int8_t code,
2129 return ((test_type == 0xFF) ||
2130 (type == test_type && code >= min_code && code <= max_code))
2135 icmp_match(const struct sk_buff *skb, struct xt_action_param *par)
2137 const struct icmphdr *ic;
2138 struct icmphdr _icmph;
2139 const struct ipt_icmp *icmpinfo = par->matchinfo;
2141 /* Must not be a fragment. */
2142 if (par->fragoff != 0)
2145 ic = skb_header_pointer(skb, par->thoff, sizeof(_icmph), &_icmph);
2147 /* We've been asked to examine this packet, and we
2148 * can't. Hence, no choice but to drop.
2150 duprintf("Dropping evil ICMP tinygram.\n");
2151 par->hotdrop = true;
2155 return icmp_type_code_match(icmpinfo->type,
2159 !!(icmpinfo->invflags&IPT_ICMP_INV));
2162 static int icmp_checkentry(const struct xt_mtchk_param *par)
2164 const struct ipt_icmp *icmpinfo = par->matchinfo;
2166 /* Must specify no unknown invflags */
2167 return (icmpinfo->invflags & ~IPT_ICMP_INV) ? -EINVAL : 0;
2170 static struct xt_target ipt_builtin_tg[] __read_mostly = {
2172 .name = XT_STANDARD_TARGET,
2173 .targetsize = sizeof(int),
2174 .family = NFPROTO_IPV4,
2175 #ifdef CONFIG_COMPAT
2176 .compatsize = sizeof(compat_int_t),
2177 .compat_from_user = compat_standard_from_user,
2178 .compat_to_user = compat_standard_to_user,
2182 .name = XT_ERROR_TARGET,
2183 .target = ipt_error,
2184 .targetsize = XT_FUNCTION_MAXNAMELEN,
2185 .family = NFPROTO_IPV4,
2189 static struct nf_sockopt_ops ipt_sockopts = {
2191 .set_optmin = IPT_BASE_CTL,
2192 .set_optmax = IPT_SO_SET_MAX+1,
2193 .set = do_ipt_set_ctl,
2194 #ifdef CONFIG_COMPAT
2195 .compat_set = compat_do_ipt_set_ctl,
2197 .get_optmin = IPT_BASE_CTL,
2198 .get_optmax = IPT_SO_GET_MAX+1,
2199 .get = do_ipt_get_ctl,
2200 #ifdef CONFIG_COMPAT
2201 .compat_get = compat_do_ipt_get_ctl,
2203 .owner = THIS_MODULE,
2206 static struct xt_match ipt_builtin_mt[] __read_mostly = {
2209 .match = icmp_match,
2210 .matchsize = sizeof(struct ipt_icmp),
2211 .checkentry = icmp_checkentry,
2212 .proto = IPPROTO_ICMP,
2213 .family = NFPROTO_IPV4,
2217 static int __net_init ip_tables_net_init(struct net *net)
2219 return xt_proto_init(net, NFPROTO_IPV4);
2222 static void __net_exit ip_tables_net_exit(struct net *net)
2224 xt_proto_fini(net, NFPROTO_IPV4);
2227 static struct pernet_operations ip_tables_net_ops = {
2228 .init = ip_tables_net_init,
2229 .exit = ip_tables_net_exit,
2232 static int __init ip_tables_init(void)
2236 ret = register_pernet_subsys(&ip_tables_net_ops);
2240 /* No one else will be downing sem now, so we won't sleep */
2241 ret = xt_register_targets(ipt_builtin_tg, ARRAY_SIZE(ipt_builtin_tg));
2244 ret = xt_register_matches(ipt_builtin_mt, ARRAY_SIZE(ipt_builtin_mt));
2248 /* Register setsockopt */
2249 ret = nf_register_sockopt(&ipt_sockopts);
2253 pr_info("(C) 2000-2006 Netfilter Core Team\n");
2257 xt_unregister_matches(ipt_builtin_mt, ARRAY_SIZE(ipt_builtin_mt));
2259 xt_unregister_targets(ipt_builtin_tg, ARRAY_SIZE(ipt_builtin_tg));
2261 unregister_pernet_subsys(&ip_tables_net_ops);
2266 static void __exit ip_tables_fini(void)
2268 nf_unregister_sockopt(&ipt_sockopts);
2270 xt_unregister_matches(ipt_builtin_mt, ARRAY_SIZE(ipt_builtin_mt));
2271 xt_unregister_targets(ipt_builtin_tg, ARRAY_SIZE(ipt_builtin_tg));
2272 unregister_pernet_subsys(&ip_tables_net_ops);
2275 EXPORT_SYMBOL(ipt_register_table);
2276 EXPORT_SYMBOL(ipt_unregister_table);
2277 EXPORT_SYMBOL(ipt_do_table);
2278 module_init(ip_tables_init);
2279 module_exit(ip_tables_fini);