~ [ source navigation ] ~ [ diff markup ] ~ [ identifier search ] ~

TOMOYO Linux Cross Reference
Linux/net/ipv4/netfilter/ip_tables.c

Version: ~ [ linux-5.9.1 ] ~ [ linux-5.8.16 ] ~ [ linux-5.7.19 ] ~ [ linux-5.6.19 ] ~ [ linux-5.5.19 ] ~ [ linux-5.4.72 ] ~ [ linux-5.3.18 ] ~ [ linux-5.2.21 ] ~ [ linux-5.1.21 ] ~ [ linux-5.0.21 ] ~ [ linux-4.20.17 ] ~ [ linux-4.19.152 ] ~ [ linux-4.18.20 ] ~ [ linux-4.17.19 ] ~ [ linux-4.16.18 ] ~ [ linux-4.15.18 ] ~ [ linux-4.14.202 ] ~ [ linux-4.13.16 ] ~ [ linux-4.12.14 ] ~ [ linux-4.11.12 ] ~ [ linux-4.10.17 ] ~ [ linux-4.9.240 ] ~ [ linux-4.8.17 ] ~ [ linux-4.7.10 ] ~ [ linux-4.6.7 ] ~ [ linux-4.5.7 ] ~ [ linux-4.4.240 ] ~ [ linux-4.3.6 ] ~ [ linux-4.2.8 ] ~ [ linux-4.1.52 ] ~ [ linux-4.0.9 ] ~ [ linux-3.19.8 ] ~ [ linux-3.18.140 ] ~ [ linux-3.17.8 ] ~ [ linux-3.16.85 ] ~ [ linux-3.15.10 ] ~ [ linux-3.14.79 ] ~ [ linux-3.13.11 ] ~ [ linux-3.12.74 ] ~ [ linux-3.11.10 ] ~ [ linux-3.10.108 ] ~ [ linux-2.6.32.71 ] ~ [ linux-2.6.0 ] ~ [ linux-2.4.37.11 ] ~ [ unix-v6-master ] ~ [ ccs-tools-1.8.5 ] ~ [ policy-sample ] ~
Architecture: ~ [ i386 ] ~ [ alpha ] ~ [ m68k ] ~ [ mips ] ~ [ ppc ] ~ [ sparc ] ~ [ sparc64 ] ~

  1 /*
  2  * Packet matching code.
  3  *
  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>
  7  *
  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.
 11  */
 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>
 21 #include <net/ip.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>
 28 
 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"
 33 
 34 MODULE_LICENSE("GPL");
 35 MODULE_AUTHOR("Netfilter Core Team <coreteam@netfilter.org>");
 36 MODULE_DESCRIPTION("IPv4 packet filter");
 37 
 38 /*#define DEBUG_IP_FIREWALL*/
 39 /*#define DEBUG_ALLOW_ALL*/ /* Useful for remote debugging */
 40 /*#define DEBUG_IP_FIREWALL_USER*/
 41 
 42 #ifdef DEBUG_IP_FIREWALL
 43 #define dprintf(format, args...) pr_info(format , ## args)
 44 #else
 45 #define dprintf(format, args...)
 46 #endif
 47 
 48 #ifdef DEBUG_IP_FIREWALL_USER
 49 #define duprintf(format, args...) pr_info(format , ## args)
 50 #else
 51 #define duprintf(format, args...)
 52 #endif
 53 
 54 #ifdef CONFIG_NETFILTER_DEBUG
 55 #define IP_NF_ASSERT(x)         WARN_ON(!(x))
 56 #else
 57 #define IP_NF_ASSERT(x)
 58 #endif
 59 
 60 #if 0
 61 /* All the better to debug you with... */
 62 #define static
 63 #define inline
 64 #endif
 65 
 66 void *ipt_alloc_initial_table(const struct xt_table *info)
 67 {
 68         return xt_alloc_initial_table(ipt, IPT);
 69 }
 70 EXPORT_SYMBOL_GPL(ipt_alloc_initial_table);
 71 
 72 /* Returns whether matches rule or not. */
 73 /* Performance critical - called for every packet */
 74 static inline bool
 75 ip_packet_match(const struct iphdr *ip,
 76                 const char *indev,
 77                 const char *outdev,
 78                 const struct ipt_ip *ipinfo,
 79                 int isfrag)
 80 {
 81         unsigned long ret;
 82 
 83 #define FWINV(bool, invflg) ((bool) ^ !!(ipinfo->invflags & (invflg)))
 84 
 85         if (FWINV((ip->saddr&ipinfo->smsk.s_addr) != ipinfo->src.s_addr,
 86                   IPT_INV_SRCIP) ||
 87             FWINV((ip->daddr&ipinfo->dmsk.s_addr) != ipinfo->dst.s_addr,
 88                   IPT_INV_DSTIP)) {
 89                 dprintf("Source or dest mismatch.\n");
 90 
 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)" : "");
 97                 return false;
 98         }
 99 
100         ret = ifname_compare_aligned(indev, ipinfo->iniface, ipinfo->iniface_mask);
101 
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)":"");
106                 return false;
107         }
108 
109         ret = ifname_compare_aligned(outdev, ipinfo->outiface, ipinfo->outiface_mask);
110 
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)":"");
115                 return false;
116         }
117 
118         /* Check specific protocol */
119         if (ipinfo->proto &&
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)":"");
124                 return false;
125         }
126 
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)" : "");
132                 return false;
133         }
134 
135         return true;
136 }
137 
138 static bool
139 ip_checkentry(const struct ipt_ip *ip)
140 {
141         if (ip->flags & ~IPT_F_MASK) {
142                 duprintf("Unknown flag bits set: %08X\n",
143                          ip->flags & ~IPT_F_MASK);
144                 return false;
145         }
146         if (ip->invflags & ~IPT_INV_MASK) {
147                 duprintf("Unknown invflag bits set: %08X\n",
148                          ip->invflags & ~IPT_INV_MASK);
149                 return false;
150         }
151         return true;
152 }
153 
154 static unsigned int
155 ipt_error(struct sk_buff *skb, const struct xt_action_param *par)
156 {
157         net_info_ratelimited("error: `%s'\n", (const char *)par->targinfo);
158 
159         return NF_DROP;
160 }
161 
162 /* Performance critical */
163 static inline struct ipt_entry *
164 get_entry(const void *base, unsigned int offset)
165 {
166         return (struct ipt_entry *)(base + offset);
167 }
168 
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)
172 {
173         static const struct ipt_ip uncond;
174 
175         return e->target_offset == sizeof(struct ipt_entry) &&
176                memcmp(&e->ip, &uncond, sizeof(uncond)) == 0;
177 #undef FWINV
178 }
179 
180 /* for const-correctness */
181 static inline const struct xt_entry_target *
182 ipt_get_target_c(const struct ipt_entry *e)
183 {
184         return ipt_get_target((struct ipt_entry *)e);
185 }
186 
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",
194 };
195 
196 enum nf_ip_trace_comments {
197         NF_IP_TRACE_COMMENT_RULE,
198         NF_IP_TRACE_COMMENT_RETURN,
199         NF_IP_TRACE_COMMENT_POLICY,
200 };
201 
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",
206 };
207 
208 static struct nf_loginfo trace_loginfo = {
209         .type = NF_LOG_TYPE_LOG,
210         .u = {
211                 .log = {
212                         .level = 4,
213                         .logflags = NF_LOG_MASK,
214                 },
215         },
216 };
217 
218 /* Mildly perf critical (only if packet tracing is on) */
219 static inline int
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)
223 {
224         const struct xt_standard_target *t = (void *)ipt_get_target_c(s);
225 
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;
229                 (*rulenum) = 0;
230         } else if (s == e) {
231                 (*rulenum)++;
232 
233                 if (unconditional(s) &&
234                     strcmp(t->target.u.kernel.target->name,
235                            XT_STANDARD_TARGET) == 0 &&
236                    t->verdict < 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];
241                 }
242                 return 1;
243         } else
244                 (*rulenum)++;
245 
246         return 0;
247 }
248 
249 static void trace_packet(const struct sk_buff *skb,
250                          unsigned int hook,
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)
256 {
257         const void *table_base;
258         const struct ipt_entry *root;
259         const char *hookname, *chainname, *comment;
260         const struct ipt_entry *iter;
261         unsigned int rulenum = 0;
262         struct net *net = dev_net(in ? in : out);
263 
264         table_base = private->entries[smp_processor_id()];
265         root = get_entry(table_base, private->hook_entry[hook]);
266 
267         hookname = chainname = hooknames[hook];
268         comment = comments[NF_IP_TRACE_COMMENT_RULE];
269 
270         xt_entry_foreach(iter, root, private->size - private->hook_entry[hook])
271                 if (get_chainname_rulenum(iter, e, hookname,
272                     &chainname, &comment, &rulenum) != 0)
273                         break;
274 
275         nf_log_packet(net, AF_INET, hook, skb, in, out, &trace_loginfo,
276                       "TRACE: %s:%s:%s:%u ",
277                       tablename, chainname, comment, rulenum);
278 }
279 #endif
280 
281 static inline __pure
282 struct ipt_entry *ipt_next_entry(const struct ipt_entry *entry)
283 {
284         return (void *)entry + entry->next_offset;
285 }
286 
287 /* Returns one of the generic firewall policies, like NF_ACCEPT. */
288 unsigned int
289 ipt_do_table(struct sk_buff *skb,
290              unsigned int hook,
291              const struct net_device *in,
292              const struct net_device *out,
293              struct xt_table *table)
294 {
295         static const char nulldevname[IFNAMSIZ] __attribute__((aligned(sizeof(long))));
296         const struct iphdr *ip;
297         /* Initializing verdict to NF_DROP keeps gcc happy. */
298         unsigned int verdict = NF_DROP;
299         const char *indev, *outdev;
300         const void *table_base;
301         struct ipt_entry *e, **jumpstack;
302         unsigned int *stackptr, origptr, cpu;
303         const struct xt_table_info *private;
304         struct xt_action_param acpar;
305         unsigned int addend;
306 
307         /* Initialization */
308         ip = ip_hdr(skb);
309         indev = in ? in->name : nulldevname;
310         outdev = out ? out->name : nulldevname;
311         /* We handle fragments by dealing with the first fragment as
312          * if it was a normal packet.  All other fragments are treated
313          * normally, except that they will NEVER match rules that ask
314          * things we don't know, ie. tcp syn flag or ports).  If the
315          * rule is also a fragment-specific rule, non-fragments won't
316          * match it. */
317         acpar.fragoff = ntohs(ip->frag_off) & IP_OFFSET;
318         acpar.thoff   = ip_hdrlen(skb);
319         acpar.hotdrop = false;
320         acpar.in      = in;
321         acpar.out     = out;
322         acpar.family  = NFPROTO_IPV4;
323         acpar.hooknum = hook;
324 
325         IP_NF_ASSERT(table->valid_hooks & (1 << hook));
326         local_bh_disable();
327         addend = xt_write_recseq_begin();
328         private = table->private;
329         cpu        = smp_processor_id();
330         table_base = private->entries[cpu];
331         jumpstack  = (struct ipt_entry **)private->jumpstack[cpu];
332         stackptr   = per_cpu_ptr(private->stackptr, cpu);
333         origptr    = *stackptr;
334 
335         e = get_entry(table_base, private->hook_entry[hook]);
336 
337         pr_debug("Entering %s(hook %u); sp at %u (UF %p)\n",
338                  table->name, hook, origptr,
339                  get_entry(table_base, private->underflow[hook]));
340 
341         do {
342                 const struct xt_entry_target *t;
343                 const struct xt_entry_match *ematch;
344 
345                 IP_NF_ASSERT(e);
346                 if (!ip_packet_match(ip, indev, outdev,
347                     &e->ip, acpar.fragoff)) {
348  no_match:
349                         e = ipt_next_entry(e);
350                         continue;
351                 }
352 
353                 xt_ematch_foreach(ematch, e) {
354                         acpar.match     = ematch->u.kernel.match;
355                         acpar.matchinfo = ematch->data;
356                         if (!acpar.match->match(skb, &acpar))
357                                 goto no_match;
358                 }
359 
360                 ADD_COUNTER(e->counters, skb->len, 1);
361 
362                 t = ipt_get_target(e);
363                 IP_NF_ASSERT(t->u.kernel.target);
364 
365 #if IS_ENABLED(CONFIG_NETFILTER_XT_TARGET_TRACE)
366                 /* The packet is traced: log it */
367                 if (unlikely(skb->nf_trace))
368                         trace_packet(skb, hook, in, out,
369                                      table->name, private, e);
370 #endif
371                 /* Standard target? */
372                 if (!t->u.kernel.target->target) {
373                         int v;
374 
375                         v = ((struct xt_standard_target *)t)->verdict;
376                         if (v < 0) {
377                                 /* Pop from stack? */
378                                 if (v != XT_RETURN) {
379                                         verdict = (unsigned int)(-v) - 1;
380                                         break;
381                                 }
382                                 if (*stackptr <= origptr) {
383                                         e = get_entry(table_base,
384                                             private->underflow[hook]);
385                                         pr_debug("Underflow (this is normal) "
386                                                  "to %p\n", e);
387                                 } else {
388                                         e = jumpstack[--*stackptr];
389                                         pr_debug("Pulled %p out from pos %u\n",
390                                                  e, *stackptr);
391                                         e = ipt_next_entry(e);
392                                 }
393                                 continue;
394                         }
395                         if (table_base + v != ipt_next_entry(e) &&
396                             !(e->ip.flags & IPT_F_GOTO)) {
397                                 if (*stackptr >= private->stacksize) {
398                                         verdict = NF_DROP;
399                                         break;
400                                 }
401                                 jumpstack[(*stackptr)++] = e;
402                                 pr_debug("Pushed %p into pos %u\n",
403                                          e, *stackptr - 1);
404                         }
405 
406                         e = get_entry(table_base, v);
407                         continue;
408                 }
409 
410                 acpar.target   = t->u.kernel.target;
411                 acpar.targinfo = t->data;
412 
413                 verdict = t->u.kernel.target->target(skb, &acpar);
414                 /* Target might have changed stuff. */
415                 ip = ip_hdr(skb);
416                 if (verdict == XT_CONTINUE)
417                         e = ipt_next_entry(e);
418                 else
419                         /* Verdict */
420                         break;
421         } while (!acpar.hotdrop);
422         pr_debug("Exiting %s; resetting sp from %u to %u\n",
423                  __func__, *stackptr, origptr);
424         *stackptr = origptr;
425         xt_write_recseq_end(addend);
426         local_bh_enable();
427 
428 #ifdef DEBUG_ALLOW_ALL
429         return NF_ACCEPT;
430 #else
431         if (acpar.hotdrop)
432                 return NF_DROP;
433         else return verdict;
434 #endif
435 }
436 
437 /* Figures out from what hook each rule can be called: returns 0 if
438    there are loops.  Puts hook bitmask in comefrom. */
439 static int
440 mark_source_chains(const struct xt_table_info *newinfo,
441                    unsigned int valid_hooks, void *entry0)
442 {
443         unsigned int hook;
444 
445         /* No recursion; use packet counter to save back ptrs (reset
446            to 0 as we leave), and comefrom to save source hook bitmask */
447         for (hook = 0; hook < NF_INET_NUMHOOKS; hook++) {
448                 unsigned int pos = newinfo->hook_entry[hook];
449                 struct ipt_entry *e = (struct ipt_entry *)(entry0 + pos);
450 
451                 if (!(valid_hooks & (1 << hook)))
452                         continue;
453 
454                 /* Set initial back pointer. */
455                 e->counters.pcnt = pos;
456 
457                 for (;;) {
458                         const struct xt_standard_target *t
459                                 = (void *)ipt_get_target_c(e);
460                         int visited = e->comefrom & (1 << hook);
461 
462                         if (e->comefrom & (1 << NF_INET_NUMHOOKS)) {
463                                 pr_err("iptables: loop hook %u pos %u %08X.\n",
464                                        hook, pos, e->comefrom);
465                                 return 0;
466                         }
467                         e->comefrom |= ((1 << hook) | (1 << NF_INET_NUMHOOKS));
468 
469                         /* Unconditional return/END. */
470                         if ((unconditional(e) &&
471                              (strcmp(t->target.u.user.name,
472                                      XT_STANDARD_TARGET) == 0) &&
473                              t->verdict < 0) || visited) {
474                                 unsigned int oldpos, size;
475 
476                                 if ((strcmp(t->target.u.user.name,
477                                             XT_STANDARD_TARGET) == 0) &&
478                                     t->verdict < -NF_MAX_VERDICT - 1) {
479                                         duprintf("mark_source_chains: bad "
480                                                 "negative verdict (%i)\n",
481                                                                 t->verdict);
482                                         return 0;
483                                 }
484 
485                                 /* Return: backtrack through the last
486                                    big jump. */
487                                 do {
488                                         e->comefrom ^= (1<<NF_INET_NUMHOOKS);
489 #ifdef DEBUG_IP_FIREWALL_USER
490                                         if (e->comefrom
491                                             & (1 << NF_INET_NUMHOOKS)) {
492                                                 duprintf("Back unset "
493                                                          "on hook %u "
494                                                          "rule %u\n",
495                                                          hook, pos);
496                                         }
497 #endif
498                                         oldpos = pos;
499                                         pos = e->counters.pcnt;
500                                         e->counters.pcnt = 0;
501 
502                                         /* We're at the start. */
503                                         if (pos == oldpos)
504                                                 goto next;
505 
506                                         e = (struct ipt_entry *)
507                                                 (entry0 + pos);
508                                 } while (oldpos == pos + e->next_offset);
509 
510                                 /* Move along one */
511                                 size = e->next_offset;
512                                 e = (struct ipt_entry *)
513                                         (entry0 + pos + size);
514                                 if (pos + size >= newinfo->size)
515                                         return 0;
516                                 e->counters.pcnt = pos;
517                                 pos += size;
518                         } else {
519                                 int newpos = t->verdict;
520 
521                                 if (strcmp(t->target.u.user.name,
522                                            XT_STANDARD_TARGET) == 0 &&
523                                     newpos >= 0) {
524                                         if (newpos > newinfo->size -
525                                                 sizeof(struct ipt_entry)) {
526                                                 duprintf("mark_source_chains: "
527                                                         "bad verdict (%i)\n",
528                                                                 newpos);
529                                                 return 0;
530                                         }
531                                         /* This a jump; chase it. */
532                                         duprintf("Jump rule %u -> %u\n",
533                                                  pos, newpos);
534                                 } else {
535                                         /* ... this is a fallthru */
536                                         newpos = pos + e->next_offset;
537                                         if (newpos >= newinfo->size)
538                                                 return 0;
539                                 }
540                                 e = (struct ipt_entry *)
541                                         (entry0 + newpos);
542                                 e->counters.pcnt = pos;
543                                 pos = newpos;
544                         }
545                 }
546                 next:
547                 duprintf("Finished chain %u\n", hook);
548         }
549         return 1;
550 }
551 
552 static void cleanup_match(struct xt_entry_match *m, struct net *net)
553 {
554         struct xt_mtdtor_param par;
555 
556         par.net       = net;
557         par.match     = m->u.kernel.match;
558         par.matchinfo = m->data;
559         par.family    = NFPROTO_IPV4;
560         if (par.match->destroy != NULL)
561                 par.match->destroy(&par);
562         module_put(par.match->me);
563 }
564 
565 static int
566 check_match(struct xt_entry_match *m, struct xt_mtchk_param *par)
567 {
568         const struct ipt_ip *ip = par->entryinfo;
569         int ret;
570 
571         par->match     = m->u.kernel.match;
572         par->matchinfo = m->data;
573 
574         ret = xt_check_match(par, m->u.match_size - sizeof(*m),
575               ip->proto, ip->invflags & IPT_INV_PROTO);
576         if (ret < 0) {
577                 duprintf("check failed for `%s'.\n", par->match->name);
578                 return ret;
579         }
580         return 0;
581 }
582 
583 static int
584 find_check_match(struct xt_entry_match *m, struct xt_mtchk_param *par)
585 {
586         struct xt_match *match;
587         int ret;
588 
589         match = xt_request_find_match(NFPROTO_IPV4, m->u.user.name,
590                                       m->u.user.revision);
591         if (IS_ERR(match)) {
592                 duprintf("find_check_match: `%s' not found\n", m->u.user.name);
593                 return PTR_ERR(match);
594         }
595         m->u.kernel.match = match;
596 
597         ret = check_match(m, par);
598         if (ret)
599                 goto err;
600 
601         return 0;
602 err:
603         module_put(m->u.kernel.match->me);
604         return ret;
605 }
606 
607 static int check_target(struct ipt_entry *e, struct net *net, const char *name)
608 {
609         struct xt_entry_target *t = ipt_get_target(e);
610         struct xt_tgchk_param par = {
611                 .net       = net,
612                 .table     = name,
613                 .entryinfo = e,
614                 .target    = t->u.kernel.target,
615                 .targinfo  = t->data,
616                 .hook_mask = e->comefrom,
617                 .family    = NFPROTO_IPV4,
618         };
619         int ret;
620 
621         ret = xt_check_target(&par, t->u.target_size - sizeof(*t),
622               e->ip.proto, e->ip.invflags & IPT_INV_PROTO);
623         if (ret < 0) {
624                 duprintf("check failed for `%s'.\n",
625                          t->u.kernel.target->name);
626                 return ret;
627         }
628         return 0;
629 }
630 
631 static int
632 find_check_entry(struct ipt_entry *e, struct net *net, const char *name,
633                  unsigned int size)
634 {
635         struct xt_entry_target *t;
636         struct xt_target *target;
637         int ret;
638         unsigned int j;
639         struct xt_mtchk_param mtpar;
640         struct xt_entry_match *ematch;
641 
642         j = 0;
643         mtpar.net       = net;
644         mtpar.table     = name;
645         mtpar.entryinfo = &e->ip;
646         mtpar.hook_mask = e->comefrom;
647         mtpar.family    = NFPROTO_IPV4;
648         xt_ematch_foreach(ematch, e) {
649                 ret = find_check_match(ematch, &mtpar);
650                 if (ret != 0)
651                         goto cleanup_matches;
652                 ++j;
653         }
654 
655         t = ipt_get_target(e);
656         target = xt_request_find_target(NFPROTO_IPV4, t->u.user.name,
657                                         t->u.user.revision);
658         if (IS_ERR(target)) {
659                 duprintf("find_check_entry: `%s' not found\n", t->u.user.name);
660                 ret = PTR_ERR(target);
661                 goto cleanup_matches;
662         }
663         t->u.kernel.target = target;
664 
665         ret = check_target(e, net, name);
666         if (ret)
667                 goto err;
668         return 0;
669  err:
670         module_put(t->u.kernel.target->me);
671  cleanup_matches:
672         xt_ematch_foreach(ematch, e) {
673                 if (j-- == 0)
674                         break;
675                 cleanup_match(ematch, net);
676         }
677         return ret;
678 }
679 
680 static bool check_underflow(const struct ipt_entry *e)
681 {
682         const struct xt_entry_target *t;
683         unsigned int verdict;
684 
685         if (!unconditional(e))
686                 return false;
687         t = ipt_get_target_c(e);
688         if (strcmp(t->u.user.name, XT_STANDARD_TARGET) != 0)
689                 return false;
690         verdict = ((struct xt_standard_target *)t)->verdict;
691         verdict = -verdict - 1;
692         return verdict == NF_DROP || verdict == NF_ACCEPT;
693 }
694 
695 static int
696 check_entry_size_and_hooks(struct ipt_entry *e,
697                            struct xt_table_info *newinfo,
698                            const unsigned char *base,
699                            const unsigned char *limit,
700                            const unsigned int *hook_entries,
701                            const unsigned int *underflows,
702                            unsigned int valid_hooks)
703 {
704         unsigned int h;
705         int err;
706 
707         if ((unsigned long)e % __alignof__(struct ipt_entry) != 0 ||
708             (unsigned char *)e + sizeof(struct ipt_entry) >= limit ||
709             (unsigned char *)e + e->next_offset > limit) {
710                 duprintf("Bad offset %p\n", e);
711                 return -EINVAL;
712         }
713 
714         if (e->next_offset
715             < sizeof(struct ipt_entry) + sizeof(struct xt_entry_target)) {
716                 duprintf("checking: element %p size %u\n",
717                          e, e->next_offset);
718                 return -EINVAL;
719         }
720 
721         if (!ip_checkentry(&e->ip))
722                 return -EINVAL;
723 
724         err = xt_check_entry_offsets(e, e->elems, e->target_offset,
725                                      e->next_offset);
726         if (err)
727                 return err;
728 
729         /* Check hooks & underflows */
730         for (h = 0; h < NF_INET_NUMHOOKS; h++) {
731                 if (!(valid_hooks & (1 << h)))
732                         continue;
733                 if ((unsigned char *)e - base == hook_entries[h])
734                         newinfo->hook_entry[h] = hook_entries[h];
735                 if ((unsigned char *)e - base == underflows[h]) {
736                         if (!check_underflow(e)) {
737                                 pr_debug("Underflows must be unconditional and "
738                                          "use the STANDARD target with "
739                                          "ACCEPT/DROP\n");
740                                 return -EINVAL;
741                         }
742                         newinfo->underflow[h] = underflows[h];
743                 }
744         }
745 
746         /* Clear counters and comefrom */
747         e->counters = ((struct xt_counters) { 0, 0 });
748         e->comefrom = 0;
749         return 0;
750 }
751 
752 static void
753 cleanup_entry(struct ipt_entry *e, struct net *net)
754 {
755         struct xt_tgdtor_param par;
756         struct xt_entry_target *t;
757         struct xt_entry_match *ematch;
758 
759         /* Cleanup all matches */
760         xt_ematch_foreach(ematch, e)
761                 cleanup_match(ematch, net);
762         t = ipt_get_target(e);
763 
764         par.net      = net;
765         par.target   = t->u.kernel.target;
766         par.targinfo = t->data;
767         par.family   = NFPROTO_IPV4;
768         if (par.target->destroy != NULL)
769                 par.target->destroy(&par);
770         module_put(par.target->me);
771 }
772 
773 /* Checks and translates the user-supplied table segment (held in
774    newinfo) */
775 static int
776 translate_table(struct net *net, struct xt_table_info *newinfo, void *entry0,
777                 const struct ipt_replace *repl)
778 {
779         struct ipt_entry *iter;
780         unsigned int i;
781         int ret = 0;
782 
783         newinfo->size = repl->size;
784         newinfo->number = repl->num_entries;
785 
786         /* Init all hooks to impossible value. */
787         for (i = 0; i < NF_INET_NUMHOOKS; i++) {
788                 newinfo->hook_entry[i] = 0xFFFFFFFF;
789                 newinfo->underflow[i] = 0xFFFFFFFF;
790         }
791 
792         duprintf("translate_table: size %u\n", newinfo->size);
793         i = 0;
794         /* Walk through entries, checking offsets. */
795         xt_entry_foreach(iter, entry0, newinfo->size) {
796                 ret = check_entry_size_and_hooks(iter, newinfo, entry0,
797                                                  entry0 + repl->size,
798                                                  repl->hook_entry,
799                                                  repl->underflow,
800                                                  repl->valid_hooks);
801                 if (ret != 0)
802                         return ret;
803                 ++i;
804                 if (strcmp(ipt_get_target(iter)->u.user.name,
805                     XT_ERROR_TARGET) == 0)
806                         ++newinfo->stacksize;
807         }
808 
809         if (i != repl->num_entries) {
810                 duprintf("translate_table: %u not %u entries\n",
811                          i, repl->num_entries);
812                 return -EINVAL;
813         }
814 
815         /* Check hooks all assigned */
816         for (i = 0; i < NF_INET_NUMHOOKS; i++) {
817                 /* Only hooks which are valid */
818                 if (!(repl->valid_hooks & (1 << i)))
819                         continue;
820                 if (newinfo->hook_entry[i] == 0xFFFFFFFF) {
821                         duprintf("Invalid hook entry %u %u\n",
822                                  i, repl->hook_entry[i]);
823                         return -EINVAL;
824                 }
825                 if (newinfo->underflow[i] == 0xFFFFFFFF) {
826                         duprintf("Invalid underflow %u %u\n",
827                                  i, repl->underflow[i]);
828                         return -EINVAL;
829                 }
830         }
831 
832         if (!mark_source_chains(newinfo, repl->valid_hooks, entry0))
833                 return -ELOOP;
834 
835         /* Finally, each sanity check must pass */
836         i = 0;
837         xt_entry_foreach(iter, entry0, newinfo->size) {
838                 ret = find_check_entry(iter, net, repl->name, repl->size);
839                 if (ret != 0)
840                         break;
841                 ++i;
842         }
843 
844         if (ret != 0) {
845                 xt_entry_foreach(iter, entry0, newinfo->size) {
846                         if (i-- == 0)
847                                 break;
848                         cleanup_entry(iter, net);
849                 }
850                 return ret;
851         }
852 
853         /* And one copy for every other CPU */
854         for_each_possible_cpu(i) {
855                 if (newinfo->entries[i] && newinfo->entries[i] != entry0)
856                         memcpy(newinfo->entries[i], entry0, newinfo->size);
857         }
858 
859         return ret;
860 }
861 
862 static void
863 get_counters(const struct xt_table_info *t,
864              struct xt_counters counters[])
865 {
866         struct ipt_entry *iter;
867         unsigned int cpu;
868         unsigned int i;
869 
870         for_each_possible_cpu(cpu) {
871                 seqcount_t *s = &per_cpu(xt_recseq, cpu);
872 
873                 i = 0;
874                 xt_entry_foreach(iter, t->entries[cpu], t->size) {
875                         u64 bcnt, pcnt;
876                         unsigned int start;
877 
878                         do {
879                                 start = read_seqcount_begin(s);
880                                 bcnt = iter->counters.bcnt;
881                                 pcnt = iter->counters.pcnt;
882                         } while (read_seqcount_retry(s, start));
883 
884                         ADD_COUNTER(counters[i], bcnt, pcnt);
885                         ++i; /* macro does multi eval of i */
886                 }
887         }
888 }
889 
890 static struct xt_counters *alloc_counters(const struct xt_table *table)
891 {
892         unsigned int countersize;
893         struct xt_counters *counters;
894         const struct xt_table_info *private = table->private;
895 
896         /* We need atomic snapshot of counters: rest doesn't change
897            (other than comefrom, which userspace doesn't care
898            about). */
899         countersize = sizeof(struct xt_counters) * private->number;
900         counters = vzalloc(countersize);
901 
902         if (counters == NULL)
903                 return ERR_PTR(-ENOMEM);
904 
905         get_counters(private, counters);
906 
907         return counters;
908 }
909 
910 static int
911 copy_entries_to_user(unsigned int total_size,
912                      const struct xt_table *table,
913                      void __user *userptr)
914 {
915         unsigned int off, num;
916         const struct ipt_entry *e;
917         struct xt_counters *counters;
918         const struct xt_table_info *private = table->private;
919         int ret = 0;
920         const void *loc_cpu_entry;
921 
922         counters = alloc_counters(table);
923         if (IS_ERR(counters))
924                 return PTR_ERR(counters);
925 
926         /* choose the copy that is on our node/cpu, ...
927          * This choice is lazy (because current thread is
928          * allowed to migrate to another cpu)
929          */
930         loc_cpu_entry = private->entries[raw_smp_processor_id()];
931         if (copy_to_user(userptr, loc_cpu_entry, total_size) != 0) {
932                 ret = -EFAULT;
933                 goto free_counters;
934         }
935 
936         /* FIXME: use iterator macros --RR */
937         /* ... then go back and fix counters and names */
938         for (off = 0, num = 0; off < total_size; off += e->next_offset, num++){
939                 unsigned int i;
940                 const struct xt_entry_match *m;
941                 const struct xt_entry_target *t;
942 
943                 e = (struct ipt_entry *)(loc_cpu_entry + off);
944                 if (copy_to_user(userptr + off
945                                  + offsetof(struct ipt_entry, counters),
946                                  &counters[num],
947                                  sizeof(counters[num])) != 0) {
948                         ret = -EFAULT;
949                         goto free_counters;
950                 }
951 
952                 for (i = sizeof(struct ipt_entry);
953                      i < e->target_offset;
954                      i += m->u.match_size) {
955                         m = (void *)e + i;
956 
957                         if (copy_to_user(userptr + off + i
958                                          + offsetof(struct xt_entry_match,
959                                                     u.user.name),
960                                          m->u.kernel.match->name,
961                                          strlen(m->u.kernel.match->name)+1)
962                             != 0) {
963                                 ret = -EFAULT;
964                                 goto free_counters;
965                         }
966                 }
967 
968                 t = ipt_get_target_c(e);
969                 if (copy_to_user(userptr + off + e->target_offset
970                                  + offsetof(struct xt_entry_target,
971                                             u.user.name),
972                                  t->u.kernel.target->name,
973                                  strlen(t->u.kernel.target->name)+1) != 0) {
974                         ret = -EFAULT;
975                         goto free_counters;
976                 }
977         }
978 
979  free_counters:
980         vfree(counters);
981         return ret;
982 }
983 
984 #ifdef CONFIG_COMPAT
985 static void compat_standard_from_user(void *dst, const void *src)
986 {
987         int v = *(compat_int_t *)src;
988 
989         if (v > 0)
990                 v += xt_compat_calc_jump(AF_INET, v);
991         memcpy(dst, &v, sizeof(v));
992 }
993 
994 static int compat_standard_to_user(void __user *dst, const void *src)
995 {
996         compat_int_t cv = *(int *)src;
997 
998         if (cv > 0)
999                 cv -= xt_compat_calc_jump(AF_INET, cv);
1000         return copy_to_user(dst, &cv, sizeof(cv)) ? -EFAULT : 0;
1001 }
1002 
1003 static int compat_calc_entry(const struct ipt_entry *e,
1004                              const struct xt_table_info *info,
1005                              const void *base, struct xt_table_info *newinfo)
1006 {
1007         const struct xt_entry_match *ematch;
1008         const struct xt_entry_target *t;
1009         unsigned int entry_offset;
1010         int off, i, ret;
1011 
1012         off = sizeof(struct ipt_entry) - sizeof(struct compat_ipt_entry);
1013         entry_offset = (void *)e - base;
1014         xt_ematch_foreach(ematch, e)
1015                 off += xt_compat_match_offset(ematch->u.kernel.match);
1016         t = ipt_get_target_c(e);
1017         off += xt_compat_target_offset(t->u.kernel.target);
1018         newinfo->size -= off;
1019         ret = xt_compat_add_offset(AF_INET, entry_offset, off);
1020         if (ret)
1021                 return ret;
1022 
1023         for (i = 0; i < NF_INET_NUMHOOKS; i++) {
1024                 if (info->hook_entry[i] &&
1025                     (e < (struct ipt_entry *)(base + info->hook_entry[i])))
1026                         newinfo->hook_entry[i] -= off;
1027                 if (info->underflow[i] &&
1028                     (e < (struct ipt_entry *)(base + info->underflow[i])))
1029                         newinfo->underflow[i] -= off;
1030         }
1031         return 0;
1032 }
1033 
1034 static int compat_table_info(const struct xt_table_info *info,
1035                              struct xt_table_info *newinfo)
1036 {
1037         struct ipt_entry *iter;
1038         void *loc_cpu_entry;
1039         int ret;
1040 
1041         if (!newinfo || !info)
1042                 return -EINVAL;
1043 
1044         /* we dont care about newinfo->entries[] */
1045         memcpy(newinfo, info, offsetof(struct xt_table_info, entries));
1046         newinfo->initial_entries = 0;
1047         loc_cpu_entry = info->entries[raw_smp_processor_id()];
1048         xt_compat_init_offsets(AF_INET, info->number);
1049         xt_entry_foreach(iter, loc_cpu_entry, info->size) {
1050                 ret = compat_calc_entry(iter, info, loc_cpu_entry, newinfo);
1051                 if (ret != 0)
1052                         return ret;
1053         }
1054         return 0;
1055 }
1056 #endif
1057 
1058 static int get_info(struct net *net, void __user *user,
1059                     const int *len, int compat)
1060 {
1061         char name[XT_TABLE_MAXNAMELEN];
1062         struct xt_table *t;
1063         int ret;
1064 
1065         if (*len != sizeof(struct ipt_getinfo)) {
1066                 duprintf("length %u != %zu\n", *len,
1067                          sizeof(struct ipt_getinfo));
1068                 return -EINVAL;
1069         }
1070 
1071         if (copy_from_user(name, user, sizeof(name)) != 0)
1072                 return -EFAULT;
1073 
1074         name[XT_TABLE_MAXNAMELEN-1] = '\0';
1075 #ifdef CONFIG_COMPAT
1076         if (compat)
1077                 xt_compat_lock(AF_INET);
1078 #endif
1079         t = try_then_request_module(xt_find_table_lock(net, AF_INET, name),
1080                                     "iptable_%s", name);
1081         if (!IS_ERR_OR_NULL(t)) {
1082                 struct ipt_getinfo info;
1083                 const struct xt_table_info *private = t->private;
1084 #ifdef CONFIG_COMPAT
1085                 struct xt_table_info tmp;
1086 
1087                 if (compat) {
1088                         ret = compat_table_info(private, &tmp);
1089                         xt_compat_flush_offsets(AF_INET);
1090                         private = &tmp;
1091                 }
1092 #endif
1093                 memset(&info, 0, sizeof(info));
1094                 info.valid_hooks = t->valid_hooks;
1095                 memcpy(info.hook_entry, private->hook_entry,
1096                        sizeof(info.hook_entry));
1097                 memcpy(info.underflow, private->underflow,
1098                        sizeof(info.underflow));
1099                 info.num_entries = private->number;
1100                 info.size = private->size;
1101                 strcpy(info.name, name);
1102 
1103                 if (copy_to_user(user, &info, *len) != 0)
1104                         ret = -EFAULT;
1105                 else
1106                         ret = 0;
1107 
1108                 xt_table_unlock(t);
1109                 module_put(t->me);
1110         } else
1111                 ret = t ? PTR_ERR(t) : -ENOENT;
1112 #ifdef CONFIG_COMPAT
1113         if (compat)
1114                 xt_compat_unlock(AF_INET);
1115 #endif
1116         return ret;
1117 }
1118 
1119 static int
1120 get_entries(struct net *net, struct ipt_get_entries __user *uptr,
1121             const int *len)
1122 {
1123         int ret;
1124         struct ipt_get_entries get;
1125         struct xt_table *t;
1126 
1127         if (*len < sizeof(get)) {
1128                 duprintf("get_entries: %u < %zu\n", *len, sizeof(get));
1129                 return -EINVAL;
1130         }
1131         if (copy_from_user(&get, uptr, sizeof(get)) != 0)
1132                 return -EFAULT;
1133         if (*len != sizeof(struct ipt_get_entries) + get.size) {
1134                 duprintf("get_entries: %u != %zu\n",
1135                          *len, sizeof(get) + get.size);
1136                 return -EINVAL;
1137         }
1138 
1139         t = xt_find_table_lock(net, AF_INET, get.name);
1140         if (!IS_ERR_OR_NULL(t)) {
1141                 const struct xt_table_info *private = t->private;
1142                 duprintf("t->private->number = %u\n", private->number);
1143                 if (get.size == private->size)
1144                         ret = copy_entries_to_user(private->size,
1145                                                    t, uptr->entrytable);
1146                 else {
1147                         duprintf("get_entries: I've got %u not %u!\n",
1148                                  private->size, get.size);
1149                         ret = -EAGAIN;
1150                 }
1151                 module_put(t->me);
1152                 xt_table_unlock(t);
1153         } else
1154                 ret = t ? PTR_ERR(t) : -ENOENT;
1155 
1156         return ret;
1157 }
1158 
1159 static int
1160 __do_replace(struct net *net, const char *name, unsigned int valid_hooks,
1161              struct xt_table_info *newinfo, unsigned int num_counters,
1162              void __user *counters_ptr)
1163 {
1164         int ret;
1165         struct xt_table *t;
1166         struct xt_table_info *oldinfo;
1167         struct xt_counters *counters;
1168         void *loc_cpu_old_entry;
1169         struct ipt_entry *iter;
1170 
1171         ret = 0;
1172         counters = vzalloc(num_counters * sizeof(struct xt_counters));
1173         if (!counters) {
1174                 ret = -ENOMEM;
1175                 goto out;
1176         }
1177 
1178         t = try_then_request_module(xt_find_table_lock(net, AF_INET, name),
1179                                     "iptable_%s", name);
1180         if (IS_ERR_OR_NULL(t)) {
1181                 ret = t ? PTR_ERR(t) : -ENOENT;
1182                 goto free_newinfo_counters_untrans;
1183         }
1184 
1185         /* You lied! */
1186         if (valid_hooks != t->valid_hooks) {
1187                 duprintf("Valid hook crap: %08X vs %08X\n",
1188                          valid_hooks, t->valid_hooks);
1189                 ret = -EINVAL;
1190                 goto put_module;
1191         }
1192 
1193         oldinfo = xt_replace_table(t, num_counters, newinfo, &ret);
1194         if (!oldinfo)
1195                 goto put_module;
1196 
1197         /* Update module usage count based on number of rules */
1198         duprintf("do_replace: oldnum=%u, initnum=%u, newnum=%u\n",
1199                 oldinfo->number, oldinfo->initial_entries, newinfo->number);
1200         if ((oldinfo->number > oldinfo->initial_entries) ||
1201             (newinfo->number <= oldinfo->initial_entries))
1202                 module_put(t->me);
1203         if ((oldinfo->number > oldinfo->initial_entries) &&
1204             (newinfo->number <= oldinfo->initial_entries))
1205                 module_put(t->me);
1206 
1207         /* Get the old counters, and synchronize with replace */
1208         get_counters(oldinfo, counters);
1209 
1210         /* Decrease module usage counts and free resource */
1211         loc_cpu_old_entry = oldinfo->entries[raw_smp_processor_id()];
1212         xt_entry_foreach(iter, loc_cpu_old_entry, oldinfo->size)
1213                 cleanup_entry(iter, net);
1214 
1215         xt_free_table_info(oldinfo);
1216         if (copy_to_user(counters_ptr, counters,
1217                          sizeof(struct xt_counters) * num_counters) != 0) {
1218                 /* Silent error, can't fail, new table is already in place */
1219                 net_warn_ratelimited("iptables: counters copy to user failed while replacing table\n");
1220         }
1221         vfree(counters);
1222         xt_table_unlock(t);
1223         return ret;
1224 
1225  put_module:
1226         module_put(t->me);
1227         xt_table_unlock(t);
1228  free_newinfo_counters_untrans:
1229         vfree(counters);
1230  out:
1231         return ret;
1232 }
1233 
1234 static int
1235 do_replace(struct net *net, const void __user *user, unsigned int len)
1236 {
1237         int ret;
1238         struct ipt_replace tmp;
1239         struct xt_table_info *newinfo;
1240         void *loc_cpu_entry;
1241         struct ipt_entry *iter;
1242 
1243         if (copy_from_user(&tmp, user, sizeof(tmp)) != 0)
1244                 return -EFAULT;
1245 
1246         /* overflow check */
1247         if (tmp.num_counters >= INT_MAX / sizeof(struct xt_counters))
1248                 return -ENOMEM;
1249         if (tmp.num_counters == 0)
1250                 return -EINVAL;
1251 
1252         tmp.name[sizeof(tmp.name)-1] = 0;
1253 
1254         newinfo = xt_alloc_table_info(tmp.size);
1255         if (!newinfo)
1256                 return -ENOMEM;
1257 
1258         /* choose the copy that is on our node/cpu */
1259         loc_cpu_entry = newinfo->entries[raw_smp_processor_id()];
1260         if (copy_from_user(loc_cpu_entry, user + sizeof(tmp),
1261                            tmp.size) != 0) {
1262                 ret = -EFAULT;
1263                 goto free_newinfo;
1264         }
1265 
1266         ret = translate_table(net, newinfo, loc_cpu_entry, &tmp);
1267         if (ret != 0)
1268                 goto free_newinfo;
1269 
1270         duprintf("Translated table\n");
1271 
1272         ret = __do_replace(net, tmp.name, tmp.valid_hooks, newinfo,
1273                            tmp.num_counters, tmp.counters);
1274         if (ret)
1275                 goto free_newinfo_untrans;
1276         return 0;
1277 
1278  free_newinfo_untrans:
1279         xt_entry_foreach(iter, loc_cpu_entry, newinfo->size)
1280                 cleanup_entry(iter, net);
1281  free_newinfo:
1282         xt_free_table_info(newinfo);
1283         return ret;
1284 }
1285 
1286 static int
1287 do_add_counters(struct net *net, const void __user *user,
1288                 unsigned int len, int compat)
1289 {
1290         unsigned int i, curcpu;
1291         struct xt_counters_info tmp;
1292         struct xt_counters *paddc;
1293         struct xt_table *t;
1294         const struct xt_table_info *private;
1295         int ret = 0;
1296         void *loc_cpu_entry;
1297         struct ipt_entry *iter;
1298         unsigned int addend;
1299 
1300         paddc = xt_copy_counters_from_user(user, len, &tmp, compat);
1301         if (IS_ERR(paddc))
1302                 return PTR_ERR(paddc);
1303 
1304         t = xt_find_table_lock(net, AF_INET, tmp.name);
1305         if (IS_ERR_OR_NULL(t)) {
1306                 ret = t ? PTR_ERR(t) : -ENOENT;
1307                 goto free;
1308         }
1309 
1310         local_bh_disable();
1311         private = t->private;
1312         if (private->number != tmp.num_counters) {
1313                 ret = -EINVAL;
1314                 goto unlock_up_free;
1315         }
1316 
1317         i = 0;
1318         /* Choose the copy that is on our node */
1319         curcpu = smp_processor_id();
1320         loc_cpu_entry = private->entries[curcpu];
1321         addend = xt_write_recseq_begin();
1322         xt_entry_foreach(iter, loc_cpu_entry, private->size) {
1323                 ADD_COUNTER(iter->counters, paddc[i].bcnt, paddc[i].pcnt);
1324                 ++i;
1325         }
1326         xt_write_recseq_end(addend);
1327  unlock_up_free:
1328         local_bh_enable();
1329         xt_table_unlock(t);
1330         module_put(t->me);
1331  free:
1332         vfree(paddc);
1333 
1334         return ret;
1335 }
1336 
1337 #ifdef CONFIG_COMPAT
1338 struct compat_ipt_replace {
1339         char                    name[XT_TABLE_MAXNAMELEN];
1340         u32                     valid_hooks;
1341         u32                     num_entries;
1342         u32                     size;
1343         u32                     hook_entry[NF_INET_NUMHOOKS];
1344         u32                     underflow[NF_INET_NUMHOOKS];
1345         u32                     num_counters;
1346         compat_uptr_t           counters;       /* struct xt_counters * */
1347         struct compat_ipt_entry entries[0];
1348 };
1349 
1350 static int
1351 compat_copy_entry_to_user(struct ipt_entry *e, void __user **dstptr,
1352                           unsigned int *size, struct xt_counters *counters,
1353                           unsigned int i)
1354 {
1355         struct xt_entry_target *t;
1356         struct compat_ipt_entry __user *ce;
1357         u_int16_t target_offset, next_offset;
1358         compat_uint_t origsize;
1359         const struct xt_entry_match *ematch;
1360         int ret = 0;
1361 
1362         origsize = *size;
1363         ce = (struct compat_ipt_entry __user *)*dstptr;
1364         if (copy_to_user(ce, e, sizeof(struct ipt_entry)) != 0 ||
1365             copy_to_user(&ce->counters, &counters[i],
1366             sizeof(counters[i])) != 0)
1367                 return -EFAULT;
1368 
1369         *dstptr += sizeof(struct compat_ipt_entry);
1370         *size -= sizeof(struct ipt_entry) - sizeof(struct compat_ipt_entry);
1371 
1372         xt_ematch_foreach(ematch, e) {
1373                 ret = xt_compat_match_to_user(ematch, dstptr, size);
1374                 if (ret != 0)
1375                         return ret;
1376         }
1377         target_offset = e->target_offset - (origsize - *size);
1378         t = ipt_get_target(e);
1379         ret = xt_compat_target_to_user(t, dstptr, size);
1380         if (ret)
1381                 return ret;
1382         next_offset = e->next_offset - (origsize - *size);
1383         if (put_user(target_offset, &ce->target_offset) != 0 ||
1384             put_user(next_offset, &ce->next_offset) != 0)
1385                 return -EFAULT;
1386         return 0;
1387 }
1388 
1389 static int
1390 compat_find_calc_match(struct xt_entry_match *m,
1391                        const struct ipt_ip *ip,
1392                        unsigned int hookmask,
1393                        int *size)
1394 {
1395         struct xt_match *match;
1396 
1397         match = xt_request_find_match(NFPROTO_IPV4, m->u.user.name,
1398                                       m->u.user.revision);
1399         if (IS_ERR(match)) {
1400                 duprintf("compat_check_calc_match: `%s' not found\n",
1401                          m->u.user.name);
1402                 return PTR_ERR(match);
1403         }
1404         m->u.kernel.match = match;
1405         *size += xt_compat_match_offset(match);
1406         return 0;
1407 }
1408 
1409 static void compat_release_entry(struct compat_ipt_entry *e)
1410 {
1411         struct xt_entry_target *t;
1412         struct xt_entry_match *ematch;
1413 
1414         /* Cleanup all matches */
1415         xt_ematch_foreach(ematch, e)
1416                 module_put(ematch->u.kernel.match->me);
1417         t = compat_ipt_get_target(e);
1418         module_put(t->u.kernel.target->me);
1419 }
1420 
1421 static int
1422 check_compat_entry_size_and_hooks(struct compat_ipt_entry *e,
1423                                   struct xt_table_info *newinfo,
1424                                   unsigned int *size,
1425                                   const unsigned char *base,
1426                                   const unsigned char *limit)
1427 {
1428         struct xt_entry_match *ematch;
1429         struct xt_entry_target *t;
1430         struct xt_target *target;
1431         unsigned int entry_offset;
1432         unsigned int j;
1433         int ret, off;
1434 
1435         duprintf("check_compat_entry_size_and_hooks %p\n", e);
1436         if ((unsigned long)e % __alignof__(struct compat_ipt_entry) != 0 ||
1437             (unsigned char *)e + sizeof(struct compat_ipt_entry) >= limit ||
1438             (unsigned char *)e + e->next_offset > limit) {
1439                 duprintf("Bad offset %p, limit = %p\n", e, limit);
1440                 return -EINVAL;
1441         }
1442 
1443         if (e->next_offset < sizeof(struct compat_ipt_entry) +
1444                              sizeof(struct compat_xt_entry_target)) {
1445                 duprintf("checking: element %p size %u\n",
1446                          e, e->next_offset);
1447                 return -EINVAL;
1448         }
1449 
1450         if (!ip_checkentry(&e->ip))
1451                 return -EINVAL;
1452 
1453         ret = xt_compat_check_entry_offsets(e, e->elems,
1454                                             e->target_offset, e->next_offset);
1455         if (ret)
1456                 return ret;
1457 
1458         off = sizeof(struct ipt_entry) - sizeof(struct compat_ipt_entry);
1459         entry_offset = (void *)e - (void *)base;
1460         j = 0;
1461         xt_ematch_foreach(ematch, e) {
1462                 ret = compat_find_calc_match(ematch, &e->ip, e->comefrom,
1463                                              &off);
1464                 if (ret != 0)
1465                         goto release_matches;
1466                 ++j;
1467         }
1468 
1469         t = compat_ipt_get_target(e);
1470         target = xt_request_find_target(NFPROTO_IPV4, t->u.user.name,
1471                                         t->u.user.revision);
1472         if (IS_ERR(target)) {
1473                 duprintf("check_compat_entry_size_and_hooks: `%s' not found\n",
1474                          t->u.user.name);
1475                 ret = PTR_ERR(target);
1476                 goto release_matches;
1477         }
1478         t->u.kernel.target = target;
1479 
1480         off += xt_compat_target_offset(target);
1481         *size += off;
1482         ret = xt_compat_add_offset(AF_INET, entry_offset, off);
1483         if (ret)
1484                 goto out;
1485 
1486         return 0;
1487 
1488 out:
1489         module_put(t->u.kernel.target->me);
1490 release_matches:
1491         xt_ematch_foreach(ematch, e) {
1492                 if (j-- == 0)
1493                         break;
1494                 module_put(ematch->u.kernel.match->me);
1495         }
1496         return ret;
1497 }
1498 
1499 static void
1500 compat_copy_entry_from_user(struct compat_ipt_entry *e, void **dstptr,
1501                             unsigned int *size,
1502                             struct xt_table_info *newinfo, unsigned char *base)
1503 {
1504         struct xt_entry_target *t;
1505         struct xt_target *target;
1506         struct ipt_entry *de;
1507         unsigned int origsize;
1508         int h;
1509         struct xt_entry_match *ematch;
1510 
1511         origsize = *size;
1512         de = (struct ipt_entry *)*dstptr;
1513         memcpy(de, e, sizeof(struct ipt_entry));
1514         memcpy(&de->counters, &e->counters, sizeof(e->counters));
1515 
1516         *dstptr += sizeof(struct ipt_entry);
1517         *size += sizeof(struct ipt_entry) - sizeof(struct compat_ipt_entry);
1518 
1519         xt_ematch_foreach(ematch, e)
1520                 xt_compat_match_from_user(ematch, dstptr, size);
1521 
1522         de->target_offset = e->target_offset - (origsize - *size);
1523         t = compat_ipt_get_target(e);
1524         target = t->u.kernel.target;
1525         xt_compat_target_from_user(t, dstptr, size);
1526 
1527         de->next_offset = e->next_offset - (origsize - *size);
1528 
1529         for (h = 0; h < NF_INET_NUMHOOKS; h++) {
1530                 if ((unsigned char *)de - base < newinfo->hook_entry[h])
1531                         newinfo->hook_entry[h] -= origsize - *size;
1532                 if ((unsigned char *)de - base < newinfo->underflow[h])
1533                         newinfo->underflow[h] -= origsize - *size;
1534         }
1535 }
1536 
1537 static int
1538 translate_compat_table(struct net *net,
1539                        struct xt_table_info **pinfo,
1540                        void **pentry0,
1541                        const struct compat_ipt_replace *compatr)
1542 {
1543         unsigned int i, j;
1544         struct xt_table_info *newinfo, *info;
1545         void *pos, *entry0, *entry1;
1546         struct compat_ipt_entry *iter0;
1547         struct ipt_replace repl;
1548         unsigned int size;
1549         int ret;
1550 
1551         info = *pinfo;
1552         entry0 = *pentry0;
1553         size = compatr->size;
1554         info->number = compatr->num_entries;
1555 
1556         duprintf("translate_compat_table: size %u\n", info->size);
1557         j = 0;
1558         xt_compat_lock(AF_INET);
1559         xt_compat_init_offsets(AF_INET, compatr->num_entries);
1560         /* Walk through entries, checking offsets. */
1561         xt_entry_foreach(iter0, entry0, compatr->size) {
1562                 ret = check_compat_entry_size_and_hooks(iter0, info, &size,
1563                                                         entry0,
1564                                                         entry0 + compatr->size);
1565                 if (ret != 0)
1566                         goto out_unlock;
1567                 ++j;
1568         }
1569 
1570         ret = -EINVAL;
1571         if (j != compatr->num_entries) {
1572                 duprintf("translate_compat_table: %u not %u entries\n",
1573                          j, compatr->num_entries);
1574                 goto out_unlock;
1575         }
1576 
1577         ret = -ENOMEM;
1578         newinfo = xt_alloc_table_info(size);
1579         if (!newinfo)
1580                 goto out_unlock;
1581 
1582         newinfo->number = compatr->num_entries;
1583         for (i = 0; i < NF_INET_NUMHOOKS; i++) {
1584                 newinfo->hook_entry[i] = compatr->hook_entry[i];
1585                 newinfo->underflow[i] = compatr->underflow[i];
1586         }
1587         entry1 = newinfo->entries[raw_smp_processor_id()];
1588         pos = entry1;
1589         size = compatr->size;
1590         xt_entry_foreach(iter0, entry0, compatr->size)
1591                 compat_copy_entry_from_user(iter0, &pos, &size,
1592                                             newinfo, entry1);
1593 
1594         /* all module references in entry0 are now gone.
1595          * entry1/newinfo contains a 64bit ruleset that looks exactly as
1596          * generated by 64bit userspace.
1597          *
1598          * Call standard translate_table() to validate all hook_entrys,
1599          * underflows, check for loops, etc.
1600          */
1601         xt_compat_flush_offsets(AF_INET);
1602         xt_compat_unlock(AF_INET);
1603 
1604         memcpy(&repl, compatr, sizeof(*compatr));
1605 
1606         for (i = 0; i < NF_INET_NUMHOOKS; i++) {
1607                 repl.hook_entry[i] = newinfo->hook_entry[i];
1608                 repl.underflow[i] = newinfo->underflow[i];
1609         }
1610 
1611         repl.num_counters = 0;
1612         repl.counters = NULL;
1613         repl.size = newinfo->size;
1614         ret = translate_table(net, newinfo, entry1, &repl);
1615         if (ret)
1616                 goto free_newinfo;
1617 
1618         *pinfo = newinfo;
1619         *pentry0 = entry1;
1620         xt_free_table_info(info);
1621         return 0;
1622 
1623 free_newinfo:
1624         xt_free_table_info(newinfo);
1625         return ret;
1626 out_unlock:
1627         xt_compat_flush_offsets(AF_INET);
1628         xt_compat_unlock(AF_INET);
1629         xt_entry_foreach(iter0, entry0, compatr->size) {
1630                 if (j-- == 0)
1631                         break;
1632                 compat_release_entry(iter0);
1633         }
1634         return ret;
1635 }
1636 
1637 static int
1638 compat_do_replace(struct net *net, void __user *user, unsigned int len)
1639 {
1640         int ret;
1641         struct compat_ipt_replace tmp;
1642         struct xt_table_info *newinfo;
1643         void *loc_cpu_entry;
1644         struct ipt_entry *iter;
1645 
1646         if (copy_from_user(&tmp, user, sizeof(tmp)) != 0)
1647                 return -EFAULT;
1648 
1649         /* overflow check */
1650         if (tmp.size >= INT_MAX / num_possible_cpus())
1651                 return -ENOMEM;
1652         if (tmp.num_counters >= INT_MAX / sizeof(struct xt_counters))
1653                 return -ENOMEM;
1654         if (tmp.num_counters == 0)
1655                 return -EINVAL;
1656 
1657         tmp.name[sizeof(tmp.name)-1] = 0;
1658 
1659         newinfo = xt_alloc_table_info(tmp.size);
1660         if (!newinfo)
1661                 return -ENOMEM;
1662 
1663         /* choose the copy that is on our node/cpu */
1664         loc_cpu_entry = newinfo->entries[raw_smp_processor_id()];
1665         if (copy_from_user(loc_cpu_entry, user + sizeof(tmp),
1666                            tmp.size) != 0) {
1667                 ret = -EFAULT;
1668                 goto free_newinfo;
1669         }
1670 
1671         ret = translate_compat_table(net, &newinfo, &loc_cpu_entry, &tmp);
1672         if (ret != 0)
1673                 goto free_newinfo;
1674 
1675         duprintf("compat_do_replace: Translated table\n");
1676 
1677         ret = __do_replace(net, tmp.name, tmp.valid_hooks, newinfo,
1678                            tmp.num_counters, compat_ptr(tmp.counters));
1679         if (ret)
1680                 goto free_newinfo_untrans;
1681         return 0;
1682 
1683  free_newinfo_untrans:
1684         xt_entry_foreach(iter, loc_cpu_entry, newinfo->size)
1685                 cleanup_entry(iter, net);
1686  free_newinfo:
1687         xt_free_table_info(newinfo);
1688         return ret;
1689 }
1690 
1691 static int
1692 compat_do_ipt_set_ctl(struct sock *sk,  int cmd, void __user *user,
1693                       unsigned int len)
1694 {
1695         int ret;
1696 
1697         if (!ns_capable(sock_net(sk)->user_ns, CAP_NET_ADMIN))
1698                 return -EPERM;
1699 
1700         switch (cmd) {
1701         case IPT_SO_SET_REPLACE:
1702                 ret = compat_do_replace(sock_net(sk), user, len);
1703                 break;
1704 
1705         case IPT_SO_SET_ADD_COUNTERS:
1706                 ret = do_add_counters(sock_net(sk), user, len, 1);
1707                 break;
1708 
1709         default:
1710                 duprintf("do_ipt_set_ctl:  unknown request %i\n", cmd);
1711                 ret = -EINVAL;
1712         }
1713 
1714         return ret;
1715 }
1716 
1717 struct compat_ipt_get_entries {
1718         char name[XT_TABLE_MAXNAMELEN];
1719         compat_uint_t size;
1720         struct compat_ipt_entry entrytable[0];
1721 };
1722 
1723 static int
1724 compat_copy_entries_to_user(unsigned int total_size, struct xt_table *table,
1725                             void __user *userptr)
1726 {
1727         struct xt_counters *counters;
1728         const struct xt_table_info *private = table->private;
1729         void __user *pos;
1730         unsigned int size;
1731         int ret = 0;
1732         const void *loc_cpu_entry;
1733         unsigned int i = 0;
1734         struct ipt_entry *iter;
1735 
1736         counters = alloc_counters(table);
1737         if (IS_ERR(counters))
1738                 return PTR_ERR(counters);
1739 
1740         /* choose the copy that is on our node/cpu, ...
1741          * This choice is lazy (because current thread is
1742          * allowed to migrate to another cpu)
1743          */
1744         loc_cpu_entry = private->entries[raw_smp_processor_id()];
1745         pos = userptr;
1746         size = total_size;
1747         xt_entry_foreach(iter, loc_cpu_entry, total_size) {
1748                 ret = compat_copy_entry_to_user(iter, &pos,
1749                                                 &size, counters, i++);
1750                 if (ret != 0)
1751                         break;
1752         }
1753 
1754         vfree(counters);
1755         return ret;
1756 }
1757 
1758 static int
1759 compat_get_entries(struct net *net, struct compat_ipt_get_entries __user *uptr,
1760                    int *len)
1761 {
1762         int ret;
1763         struct compat_ipt_get_entries get;
1764         struct xt_table *t;
1765 
1766         if (*len < sizeof(get)) {
1767                 duprintf("compat_get_entries: %u < %zu\n", *len, sizeof(get));
1768                 return -EINVAL;
1769         }
1770 
1771         if (copy_from_user(&get, uptr, sizeof(get)) != 0)
1772                 return -EFAULT;
1773 
1774         if (*len != sizeof(struct compat_ipt_get_entries) + get.size) {
1775                 duprintf("compat_get_entries: %u != %zu\n",
1776                          *len, sizeof(get) + get.size);
1777                 return -EINVAL;
1778         }
1779 
1780         xt_compat_lock(AF_INET);
1781         t = xt_find_table_lock(net, AF_INET, get.name);
1782         if (!IS_ERR_OR_NULL(t)) {
1783                 const struct xt_table_info *private = t->private;
1784                 struct xt_table_info info;
1785                 duprintf("t->private->number = %u\n", private->number);
1786                 ret = compat_table_info(private, &info);
1787                 if (!ret && get.size == info.size) {
1788                         ret = compat_copy_entries_to_user(private->size,
1789                                                           t, uptr->entrytable);
1790                 } else if (!ret) {
1791                         duprintf("compat_get_entries: I've got %u not %u!\n",
1792                                  private->size, get.size);
1793                         ret = -EAGAIN;
1794                 }
1795                 xt_compat_flush_offsets(AF_INET);
1796                 module_put(t->me);
1797                 xt_table_unlock(t);
1798         } else
1799                 ret = t ? PTR_ERR(t) : -ENOENT;
1800 
1801         xt_compat_unlock(AF_INET);
1802         return ret;
1803 }
1804 
1805 static int do_ipt_get_ctl(struct sock *, int, void __user *, int *);
1806 
1807 static int
1808 compat_do_ipt_get_ctl(struct sock *sk, int cmd, void __user *user, int *len)
1809 {
1810         int ret;
1811 
1812         if (!ns_capable(sock_net(sk)->user_ns, CAP_NET_ADMIN))
1813                 return -EPERM;
1814 
1815         switch (cmd) {
1816         case IPT_SO_GET_INFO:
1817                 ret = get_info(sock_net(sk), user, len, 1);
1818                 break;
1819         case IPT_SO_GET_ENTRIES:
1820                 ret = compat_get_entries(sock_net(sk), user, len);
1821                 break;
1822         default:
1823                 ret = do_ipt_get_ctl(sk, cmd, user, len);
1824         }
1825         return ret;
1826 }
1827 #endif
1828 
1829 static int
1830 do_ipt_set_ctl(struct sock *sk, int cmd, void __user *user, unsigned int len)
1831 {
1832         int ret;
1833 
1834         if (!ns_capable(sock_net(sk)->user_ns, CAP_NET_ADMIN))
1835                 return -EPERM;
1836 
1837         switch (cmd) {
1838         case IPT_SO_SET_REPLACE:
1839                 ret = do_replace(sock_net(sk), user, len);
1840                 break;
1841 
1842         case IPT_SO_SET_ADD_COUNTERS:
1843                 ret = do_add_counters(sock_net(sk), user, len, 0);
1844                 break;
1845 
1846         default:
1847                 duprintf("do_ipt_set_ctl:  unknown request %i\n", cmd);
1848                 ret = -EINVAL;
1849         }
1850 
1851         return ret;
1852 }
1853 
1854 static int
1855 do_ipt_get_ctl(struct sock *sk, int cmd, void __user *user, int *len)
1856 {
1857         int ret;
1858 
1859         if (!ns_capable(sock_net(sk)->user_ns, CAP_NET_ADMIN))
1860                 return -EPERM;
1861 
1862         switch (cmd) {
1863         case IPT_SO_GET_INFO:
1864                 ret = get_info(sock_net(sk), user, len, 0);
1865                 break;
1866 
1867         case IPT_SO_GET_ENTRIES:
1868                 ret = get_entries(sock_net(sk), user, len);
1869                 break;
1870 
1871         case IPT_SO_GET_REVISION_MATCH:
1872         case IPT_SO_GET_REVISION_TARGET: {
1873                 struct xt_get_revision rev;
1874                 int target;
1875 
1876                 if (*len != sizeof(rev)) {
1877                         ret = -EINVAL;
1878                         break;
1879                 }
1880                 if (copy_from_user(&rev, user, sizeof(rev)) != 0) {
1881                         ret = -EFAULT;
1882                         break;
1883                 }
1884                 rev.name[sizeof(rev.name)-1] = 0;
1885 
1886                 if (cmd == IPT_SO_GET_REVISION_TARGET)
1887                         target = 1;
1888                 else
1889                         target = 0;
1890 
1891                 try_then_request_module(xt_find_revision(AF_INET, rev.name,
1892                                                          rev.revision,
1893                                                          target, &ret),
1894                                         "ipt_%s", rev.name);
1895                 break;
1896         }
1897 
1898         default:
1899                 duprintf("do_ipt_get_ctl: unknown request %i\n", cmd);
1900                 ret = -EINVAL;
1901         }
1902 
1903         return ret;
1904 }
1905 
1906 struct xt_table *ipt_register_table(struct net *net,
1907                                     const struct xt_table *table,
1908                                     const struct ipt_replace *repl)
1909 {
1910         int ret;
1911         struct xt_table_info *newinfo;
1912         struct xt_table_info bootstrap = {0};
1913         void *loc_cpu_entry;
1914         struct xt_table *new_table;
1915 
1916         newinfo = xt_alloc_table_info(repl->size);
1917         if (!newinfo) {
1918                 ret = -ENOMEM;
1919                 goto out;
1920         }
1921 
1922         /* choose the copy on our node/cpu, but dont care about preemption */
1923         loc_cpu_entry = newinfo->entries[raw_smp_processor_id()];
1924         memcpy(loc_cpu_entry, repl->entries, repl->size);
1925 
1926         ret = translate_table(net, newinfo, loc_cpu_entry, repl);
1927         if (ret != 0)
1928                 goto out_free;
1929 
1930         new_table = xt_register_table(net, table, &bootstrap, newinfo);
1931         if (IS_ERR(new_table)) {
1932                 ret = PTR_ERR(new_table);
1933                 goto out_free;
1934         }
1935 
1936         return new_table;
1937 
1938 out_free:
1939         xt_free_table_info(newinfo);
1940 out:
1941         return ERR_PTR(ret);
1942 }
1943 
1944 void ipt_unregister_table(struct net *net, struct xt_table *table)
1945 {
1946         struct xt_table_info *private;
1947         void *loc_cpu_entry;
1948         struct module *table_owner = table->me;
1949         struct ipt_entry *iter;
1950 
1951         private = xt_unregister_table(table);
1952 
1953         /* Decrease module usage counts and free resources */
1954         loc_cpu_entry = private->entries[raw_smp_processor_id()];
1955         xt_entry_foreach(iter, loc_cpu_entry, private->size)
1956                 cleanup_entry(iter, net);
1957         if (private->number > private->initial_entries)
1958                 module_put(table_owner);
1959         xt_free_table_info(private);
1960 }
1961 
1962 /* Returns 1 if the type and code is matched by the range, 0 otherwise */
1963 static inline bool
1964 icmp_type_code_match(u_int8_t test_type, u_int8_t min_code, u_int8_t max_code,
1965                      u_int8_t type, u_int8_t code,
1966                      bool invert)
1967 {
1968         return ((test_type == 0xFF) ||
1969                 (type == test_type && code >= min_code && code <= max_code))
1970                 ^ invert;
1971 }
1972 
1973 static bool
1974 icmp_match(const struct sk_buff *skb, struct xt_action_param *par)
1975 {
1976         const struct icmphdr *ic;
1977         struct icmphdr _icmph;
1978         const struct ipt_icmp *icmpinfo = par->matchinfo;
1979 
1980         /* Must not be a fragment. */
1981         if (par->fragoff != 0)
1982                 return false;
1983 
1984         ic = skb_header_pointer(skb, par->thoff, sizeof(_icmph), &_icmph);
1985         if (ic == NULL) {
1986                 /* We've been asked to examine this packet, and we
1987                  * can't.  Hence, no choice but to drop.
1988                  */
1989                 duprintf("Dropping evil ICMP tinygram.\n");
1990                 par->hotdrop = true;
1991                 return false;
1992         }
1993 
1994         return icmp_type_code_match(icmpinfo->type,
1995                                     icmpinfo->code[0],
1996                                     icmpinfo->code[1],
1997                                     ic->type, ic->code,
1998                                     !!(icmpinfo->invflags&IPT_ICMP_INV));
1999 }
2000 
2001 static int icmp_checkentry(const struct xt_mtchk_param *par)
2002 {
2003         const struct ipt_icmp *icmpinfo = par->matchinfo;
2004 
2005         /* Must specify no unknown invflags */
2006         return (icmpinfo->invflags & ~IPT_ICMP_INV) ? -EINVAL : 0;
2007 }
2008 
2009 static struct xt_target ipt_builtin_tg[] __read_mostly = {
2010         {
2011                 .name             = XT_STANDARD_TARGET,
2012                 .targetsize       = sizeof(int),
2013                 .family           = NFPROTO_IPV4,
2014 #ifdef CONFIG_COMPAT
2015                 .compatsize       = sizeof(compat_int_t),
2016                 .compat_from_user = compat_standard_from_user,
2017                 .compat_to_user   = compat_standard_to_user,
2018 #endif
2019         },
2020         {
2021                 .name             = XT_ERROR_TARGET,
2022                 .target           = ipt_error,
2023                 .targetsize       = XT_FUNCTION_MAXNAMELEN,
2024                 .family           = NFPROTO_IPV4,
2025         },
2026 };
2027 
2028 static struct nf_sockopt_ops ipt_sockopts = {
2029         .pf             = PF_INET,
2030         .set_optmin     = IPT_BASE_CTL,
2031         .set_optmax     = IPT_SO_SET_MAX+1,
2032         .set            = do_ipt_set_ctl,
2033 #ifdef CONFIG_COMPAT
2034         .compat_set     = compat_do_ipt_set_ctl,
2035 #endif
2036         .get_optmin     = IPT_BASE_CTL,
2037         .get_optmax     = IPT_SO_GET_MAX+1,
2038         .get            = do_ipt_get_ctl,
2039 #ifdef CONFIG_COMPAT
2040         .compat_get     = compat_do_ipt_get_ctl,
2041 #endif
2042         .owner          = THIS_MODULE,
2043 };
2044 
2045 static struct xt_match ipt_builtin_mt[] __read_mostly = {
2046         {
2047                 .name       = "icmp",
2048                 .match      = icmp_match,
2049                 .matchsize  = sizeof(struct ipt_icmp),
2050                 .checkentry = icmp_checkentry,
2051                 .proto      = IPPROTO_ICMP,
2052                 .family     = NFPROTO_IPV4,
2053         },
2054 };
2055 
2056 static int __net_init ip_tables_net_init(struct net *net)
2057 {
2058         return xt_proto_init(net, NFPROTO_IPV4);
2059 }
2060 
2061 static void __net_exit ip_tables_net_exit(struct net *net)
2062 {
2063         xt_proto_fini(net, NFPROTO_IPV4);
2064 }
2065 
2066 static struct pernet_operations ip_tables_net_ops = {
2067         .init = ip_tables_net_init,
2068         .exit = ip_tables_net_exit,
2069 };
2070 
2071 static int __init ip_tables_init(void)
2072 {
2073         int ret;
2074 
2075         ret = register_pernet_subsys(&ip_tables_net_ops);
2076         if (ret < 0)
2077                 goto err1;
2078 
2079         /* No one else will be downing sem now, so we won't sleep */
2080         ret = xt_register_targets(ipt_builtin_tg, ARRAY_SIZE(ipt_builtin_tg));
2081         if (ret < 0)
2082                 goto err2;
2083         ret = xt_register_matches(ipt_builtin_mt, ARRAY_SIZE(ipt_builtin_mt));
2084         if (ret < 0)
2085                 goto err4;
2086 
2087         /* Register setsockopt */
2088         ret = nf_register_sockopt(&ipt_sockopts);
2089         if (ret < 0)
2090                 goto err5;
2091 
2092         pr_info("(C) 2000-2006 Netfilter Core Team\n");
2093         return 0;
2094 
2095 err5:
2096         xt_unregister_matches(ipt_builtin_mt, ARRAY_SIZE(ipt_builtin_mt));
2097 err4:
2098         xt_unregister_targets(ipt_builtin_tg, ARRAY_SIZE(ipt_builtin_tg));
2099 err2:
2100         unregister_pernet_subsys(&ip_tables_net_ops);
2101 err1:
2102         return ret;
2103 }
2104 
2105 static void __exit ip_tables_fini(void)
2106 {
2107         nf_unregister_sockopt(&ipt_sockopts);
2108 
2109         xt_unregister_matches(ipt_builtin_mt, ARRAY_SIZE(ipt_builtin_mt));
2110         xt_unregister_targets(ipt_builtin_tg, ARRAY_SIZE(ipt_builtin_tg));
2111         unregister_pernet_subsys(&ip_tables_net_ops);
2112 }
2113 
2114 EXPORT_SYMBOL(ipt_register_table);
2115 EXPORT_SYMBOL(ipt_unregister_table);
2116 EXPORT_SYMBOL(ipt_do_table);
2117 module_init(ip_tables_init);
2118 module_exit(ip_tables_fini);
2119 

~ [ source navigation ] ~ [ diff markup ] ~ [ identifier search ] ~

kernel.org | git.kernel.org | LWN.net | Project Home | Wiki (Japanese) | Wiki (English) | SVN repository | Mail admin

Linux® is a registered trademark of Linus Torvalds in the United States and other countries.
TOMOYO® is a registered trademark of NTT DATA CORPORATION.

osdn.jp