Commit | Line | Data |
---|---|---|
1da177e4 LT |
1 | /* |
2 | * Packet matching code. | |
3 | * | |
4 | * Copyright (C) 1999 Paul `Rusty' Russell & Michael J. Neuling | |
2e4e6a17 | 5 | * Copyright (C) 2000-2005 Netfilter Core Team <coreteam@netfilter.org> |
1da177e4 LT |
6 | * |
7 | * This program is free software; you can redistribute it and/or modify | |
8 | * it under the terms of the GNU General Public License version 2 as | |
9 | * published by the Free Software Foundation. | |
10 | * | |
11 | * 19 Jan 2002 Harald Welte <laforge@gnumonks.org> | |
12 | * - increase module usage count as soon as we have rules inside | |
13 | * a table | |
2e4e6a17 HW |
14 | * 08 Oct 2005 Harald Welte <lafore@netfilter.org> |
15 | * - Generalize into "x_tables" layer and "{ip,ip6,arp}_tables" | |
1da177e4 LT |
16 | */ |
17 | #include <linux/config.h> | |
18 | #include <linux/cache.h> | |
4fc268d2 | 19 | #include <linux/capability.h> |
1da177e4 LT |
20 | #include <linux/skbuff.h> |
21 | #include <linux/kmod.h> | |
22 | #include <linux/vmalloc.h> | |
23 | #include <linux/netdevice.h> | |
24 | #include <linux/module.h> | |
1da177e4 LT |
25 | #include <linux/icmp.h> |
26 | #include <net/ip.h> | |
27 | #include <asm/uaccess.h> | |
57b47a53 | 28 | #include <linux/mutex.h> |
1da177e4 LT |
29 | #include <linux/proc_fs.h> |
30 | #include <linux/err.h> | |
c8923c6b | 31 | #include <linux/cpumask.h> |
1da177e4 | 32 | |
2e4e6a17 | 33 | #include <linux/netfilter/x_tables.h> |
1da177e4 LT |
34 | #include <linux/netfilter_ipv4/ip_tables.h> |
35 | ||
36 | MODULE_LICENSE("GPL"); | |
37 | MODULE_AUTHOR("Netfilter Core Team <coreteam@netfilter.org>"); | |
38 | MODULE_DESCRIPTION("IPv4 packet filter"); | |
39 | ||
40 | /*#define DEBUG_IP_FIREWALL*/ | |
41 | /*#define DEBUG_ALLOW_ALL*/ /* Useful for remote debugging */ | |
42 | /*#define DEBUG_IP_FIREWALL_USER*/ | |
43 | ||
44 | #ifdef DEBUG_IP_FIREWALL | |
45 | #define dprintf(format, args...) printk(format , ## args) | |
46 | #else | |
47 | #define dprintf(format, args...) | |
48 | #endif | |
49 | ||
50 | #ifdef DEBUG_IP_FIREWALL_USER | |
51 | #define duprintf(format, args...) printk(format , ## args) | |
52 | #else | |
53 | #define duprintf(format, args...) | |
54 | #endif | |
55 | ||
56 | #ifdef CONFIG_NETFILTER_DEBUG | |
57 | #define IP_NF_ASSERT(x) \ | |
58 | do { \ | |
59 | if (!(x)) \ | |
60 | printk("IP_NF_ASSERT: %s:%s:%u\n", \ | |
61 | __FUNCTION__, __FILE__, __LINE__); \ | |
62 | } while(0) | |
63 | #else | |
64 | #define IP_NF_ASSERT(x) | |
65 | #endif | |
1da177e4 LT |
66 | |
67 | #if 0 | |
68 | /* All the better to debug you with... */ | |
69 | #define static | |
70 | #define inline | |
71 | #endif | |
72 | ||
73 | /* | |
74 | We keep a set of rules for each CPU, so we can avoid write-locking | |
75 | them in the softirq when updating the counters and therefore | |
76 | only need to read-lock in the softirq; doing a write_lock_bh() in user | |
77 | context stops packets coming through and allows user context to read | |
78 | the counters or update the rules. | |
79 | ||
1da177e4 LT |
80 | Hence the start of any table is given by get_table() below. */ |
81 | ||
1da177e4 LT |
82 | /* Returns whether matches rule or not. */ |
83 | static inline int | |
84 | ip_packet_match(const struct iphdr *ip, | |
85 | const char *indev, | |
86 | const char *outdev, | |
87 | const struct ipt_ip *ipinfo, | |
88 | int isfrag) | |
89 | { | |
90 | size_t i; | |
91 | unsigned long ret; | |
92 | ||
93 | #define FWINV(bool,invflg) ((bool) ^ !!(ipinfo->invflags & invflg)) | |
94 | ||
95 | if (FWINV((ip->saddr&ipinfo->smsk.s_addr) != ipinfo->src.s_addr, | |
96 | IPT_INV_SRCIP) | |
97 | || FWINV((ip->daddr&ipinfo->dmsk.s_addr) != ipinfo->dst.s_addr, | |
98 | IPT_INV_DSTIP)) { | |
99 | dprintf("Source or dest mismatch.\n"); | |
100 | ||
101 | dprintf("SRC: %u.%u.%u.%u. Mask: %u.%u.%u.%u. Target: %u.%u.%u.%u.%s\n", | |
102 | NIPQUAD(ip->saddr), | |
103 | NIPQUAD(ipinfo->smsk.s_addr), | |
104 | NIPQUAD(ipinfo->src.s_addr), | |
105 | ipinfo->invflags & IPT_INV_SRCIP ? " (INV)" : ""); | |
106 | dprintf("DST: %u.%u.%u.%u Mask: %u.%u.%u.%u Target: %u.%u.%u.%u.%s\n", | |
107 | NIPQUAD(ip->daddr), | |
108 | NIPQUAD(ipinfo->dmsk.s_addr), | |
109 | NIPQUAD(ipinfo->dst.s_addr), | |
110 | ipinfo->invflags & IPT_INV_DSTIP ? " (INV)" : ""); | |
111 | return 0; | |
112 | } | |
113 | ||
114 | /* Look for ifname matches; this should unroll nicely. */ | |
115 | for (i = 0, ret = 0; i < IFNAMSIZ/sizeof(unsigned long); i++) { | |
116 | ret |= (((const unsigned long *)indev)[i] | |
117 | ^ ((const unsigned long *)ipinfo->iniface)[i]) | |
118 | & ((const unsigned long *)ipinfo->iniface_mask)[i]; | |
119 | } | |
120 | ||
121 | if (FWINV(ret != 0, IPT_INV_VIA_IN)) { | |
122 | dprintf("VIA in mismatch (%s vs %s).%s\n", | |
123 | indev, ipinfo->iniface, | |
124 | ipinfo->invflags&IPT_INV_VIA_IN ?" (INV)":""); | |
125 | return 0; | |
126 | } | |
127 | ||
128 | for (i = 0, ret = 0; i < IFNAMSIZ/sizeof(unsigned long); i++) { | |
129 | ret |= (((const unsigned long *)outdev)[i] | |
130 | ^ ((const unsigned long *)ipinfo->outiface)[i]) | |
131 | & ((const unsigned long *)ipinfo->outiface_mask)[i]; | |
132 | } | |
133 | ||
134 | if (FWINV(ret != 0, IPT_INV_VIA_OUT)) { | |
135 | dprintf("VIA out mismatch (%s vs %s).%s\n", | |
136 | outdev, ipinfo->outiface, | |
137 | ipinfo->invflags&IPT_INV_VIA_OUT ?" (INV)":""); | |
138 | return 0; | |
139 | } | |
140 | ||
141 | /* Check specific protocol */ | |
142 | if (ipinfo->proto | |
143 | && FWINV(ip->protocol != ipinfo->proto, IPT_INV_PROTO)) { | |
144 | dprintf("Packet protocol %hi does not match %hi.%s\n", | |
145 | ip->protocol, ipinfo->proto, | |
146 | ipinfo->invflags&IPT_INV_PROTO ? " (INV)":""); | |
147 | return 0; | |
148 | } | |
149 | ||
150 | /* If we have a fragment rule but the packet is not a fragment | |
151 | * then we return zero */ | |
152 | if (FWINV((ipinfo->flags&IPT_F_FRAG) && !isfrag, IPT_INV_FRAG)) { | |
153 | dprintf("Fragment rule but not fragment.%s\n", | |
154 | ipinfo->invflags & IPT_INV_FRAG ? " (INV)" : ""); | |
155 | return 0; | |
156 | } | |
157 | ||
158 | return 1; | |
159 | } | |
160 | ||
161 | static inline int | |
162 | ip_checkentry(const struct ipt_ip *ip) | |
163 | { | |
164 | if (ip->flags & ~IPT_F_MASK) { | |
165 | duprintf("Unknown flag bits set: %08X\n", | |
166 | ip->flags & ~IPT_F_MASK); | |
167 | return 0; | |
168 | } | |
169 | if (ip->invflags & ~IPT_INV_MASK) { | |
170 | duprintf("Unknown invflag bits set: %08X\n", | |
171 | ip->invflags & ~IPT_INV_MASK); | |
172 | return 0; | |
173 | } | |
174 | return 1; | |
175 | } | |
176 | ||
177 | static unsigned int | |
178 | ipt_error(struct sk_buff **pskb, | |
179 | const struct net_device *in, | |
180 | const struct net_device *out, | |
181 | unsigned int hooknum, | |
c4986734 | 182 | const struct xt_target *target, |
1da177e4 LT |
183 | const void *targinfo, |
184 | void *userinfo) | |
185 | { | |
186 | if (net_ratelimit()) | |
187 | printk("ip_tables: error: `%s'\n", (char *)targinfo); | |
188 | ||
189 | return NF_DROP; | |
190 | } | |
191 | ||
192 | static inline | |
193 | int do_match(struct ipt_entry_match *m, | |
194 | const struct sk_buff *skb, | |
195 | const struct net_device *in, | |
196 | const struct net_device *out, | |
197 | int offset, | |
198 | int *hotdrop) | |
199 | { | |
200 | /* Stop iteration if it doesn't match */ | |
1c524830 PM |
201 | if (!m->u.kernel.match->match(skb, in, out, m->u.kernel.match, m->data, |
202 | offset, skb->nh.iph->ihl*4, hotdrop)) | |
1da177e4 LT |
203 | return 1; |
204 | else | |
205 | return 0; | |
206 | } | |
207 | ||
208 | static inline struct ipt_entry * | |
209 | get_entry(void *base, unsigned int offset) | |
210 | { | |
211 | return (struct ipt_entry *)(base + offset); | |
212 | } | |
213 | ||
214 | /* Returns one of the generic firewall policies, like NF_ACCEPT. */ | |
215 | unsigned int | |
216 | ipt_do_table(struct sk_buff **pskb, | |
217 | unsigned int hook, | |
218 | const struct net_device *in, | |
219 | const struct net_device *out, | |
220 | struct ipt_table *table, | |
221 | void *userdata) | |
222 | { | |
223 | static const char nulldevname[IFNAMSIZ] __attribute__((aligned(sizeof(long)))); | |
224 | u_int16_t offset; | |
225 | struct iphdr *ip; | |
226 | u_int16_t datalen; | |
227 | int hotdrop = 0; | |
228 | /* Initializing verdict to NF_DROP keeps gcc happy. */ | |
229 | unsigned int verdict = NF_DROP; | |
230 | const char *indev, *outdev; | |
231 | void *table_base; | |
232 | struct ipt_entry *e, *back; | |
2e4e6a17 | 233 | struct xt_table_info *private = table->private; |
1da177e4 LT |
234 | |
235 | /* Initialization */ | |
236 | ip = (*pskb)->nh.iph; | |
237 | datalen = (*pskb)->len - ip->ihl * 4; | |
238 | indev = in ? in->name : nulldevname; | |
239 | outdev = out ? out->name : nulldevname; | |
240 | /* We handle fragments by dealing with the first fragment as | |
241 | * if it was a normal packet. All other fragments are treated | |
242 | * normally, except that they will NEVER match rules that ask | |
243 | * things we don't know, ie. tcp syn flag or ports). If the | |
244 | * rule is also a fragment-specific rule, non-fragments won't | |
245 | * match it. */ | |
246 | offset = ntohs(ip->frag_off) & IP_OFFSET; | |
247 | ||
248 | read_lock_bh(&table->lock); | |
249 | IP_NF_ASSERT(table->valid_hooks & (1 << hook)); | |
2e4e6a17 HW |
250 | table_base = (void *)private->entries[smp_processor_id()]; |
251 | e = get_entry(table_base, private->hook_entry[hook]); | |
1da177e4 LT |
252 | |
253 | /* For return from builtin chain */ | |
2e4e6a17 | 254 | back = get_entry(table_base, private->underflow[hook]); |
1da177e4 LT |
255 | |
256 | do { | |
257 | IP_NF_ASSERT(e); | |
258 | IP_NF_ASSERT(back); | |
1da177e4 LT |
259 | if (ip_packet_match(ip, indev, outdev, &e->ip, offset)) { |
260 | struct ipt_entry_target *t; | |
261 | ||
262 | if (IPT_MATCH_ITERATE(e, do_match, | |
263 | *pskb, in, out, | |
264 | offset, &hotdrop) != 0) | |
265 | goto no_match; | |
266 | ||
267 | ADD_COUNTER(e->counters, ntohs(ip->tot_len), 1); | |
268 | ||
269 | t = ipt_get_target(e); | |
270 | IP_NF_ASSERT(t->u.kernel.target); | |
271 | /* Standard target? */ | |
272 | if (!t->u.kernel.target->target) { | |
273 | int v; | |
274 | ||
275 | v = ((struct ipt_standard_target *)t)->verdict; | |
276 | if (v < 0) { | |
277 | /* Pop from stack? */ | |
278 | if (v != IPT_RETURN) { | |
279 | verdict = (unsigned)(-v) - 1; | |
280 | break; | |
281 | } | |
282 | e = back; | |
283 | back = get_entry(table_base, | |
284 | back->comefrom); | |
285 | continue; | |
286 | } | |
05465343 PM |
287 | if (table_base + v != (void *)e + e->next_offset |
288 | && !(e->ip.flags & IPT_F_GOTO)) { | |
1da177e4 LT |
289 | /* Save old back ptr in next entry */ |
290 | struct ipt_entry *next | |
291 | = (void *)e + e->next_offset; | |
292 | next->comefrom | |
293 | = (void *)back - table_base; | |
294 | /* set back pointer to next entry */ | |
295 | back = next; | |
296 | } | |
297 | ||
298 | e = get_entry(table_base, v); | |
299 | } else { | |
300 | /* Targets which reenter must return | |
301 | abs. verdicts */ | |
302 | #ifdef CONFIG_NETFILTER_DEBUG | |
303 | ((struct ipt_entry *)table_base)->comefrom | |
304 | = 0xeeeeeeec; | |
305 | #endif | |
306 | verdict = t->u.kernel.target->target(pskb, | |
307 | in, out, | |
308 | hook, | |
1c524830 | 309 | t->u.kernel.target, |
1da177e4 LT |
310 | t->data, |
311 | userdata); | |
312 | ||
313 | #ifdef CONFIG_NETFILTER_DEBUG | |
314 | if (((struct ipt_entry *)table_base)->comefrom | |
315 | != 0xeeeeeeec | |
316 | && verdict == IPT_CONTINUE) { | |
317 | printk("Target %s reentered!\n", | |
318 | t->u.kernel.target->name); | |
319 | verdict = NF_DROP; | |
320 | } | |
321 | ((struct ipt_entry *)table_base)->comefrom | |
322 | = 0x57acc001; | |
323 | #endif | |
324 | /* Target might have changed stuff. */ | |
325 | ip = (*pskb)->nh.iph; | |
326 | datalen = (*pskb)->len - ip->ihl * 4; | |
327 | ||
328 | if (verdict == IPT_CONTINUE) | |
329 | e = (void *)e + e->next_offset; | |
330 | else | |
331 | /* Verdict */ | |
332 | break; | |
333 | } | |
334 | } else { | |
335 | ||
336 | no_match: | |
337 | e = (void *)e + e->next_offset; | |
338 | } | |
339 | } while (!hotdrop); | |
340 | ||
1da177e4 LT |
341 | read_unlock_bh(&table->lock); |
342 | ||
343 | #ifdef DEBUG_ALLOW_ALL | |
344 | return NF_ACCEPT; | |
345 | #else | |
346 | if (hotdrop) | |
347 | return NF_DROP; | |
348 | else return verdict; | |
349 | #endif | |
350 | } | |
351 | ||
1da177e4 LT |
352 | /* All zeroes == unconditional rule. */ |
353 | static inline int | |
354 | unconditional(const struct ipt_ip *ip) | |
355 | { | |
356 | unsigned int i; | |
357 | ||
358 | for (i = 0; i < sizeof(*ip)/sizeof(__u32); i++) | |
359 | if (((__u32 *)ip)[i]) | |
360 | return 0; | |
361 | ||
362 | return 1; | |
363 | } | |
364 | ||
365 | /* Figures out from what hook each rule can be called: returns 0 if | |
366 | there are loops. Puts hook bitmask in comefrom. */ | |
367 | static int | |
2e4e6a17 | 368 | mark_source_chains(struct xt_table_info *newinfo, |
31836064 | 369 | unsigned int valid_hooks, void *entry0) |
1da177e4 LT |
370 | { |
371 | unsigned int hook; | |
372 | ||
373 | /* No recursion; use packet counter to save back ptrs (reset | |
374 | to 0 as we leave), and comefrom to save source hook bitmask */ | |
375 | for (hook = 0; hook < NF_IP_NUMHOOKS; hook++) { | |
376 | unsigned int pos = newinfo->hook_entry[hook]; | |
377 | struct ipt_entry *e | |
31836064 | 378 | = (struct ipt_entry *)(entry0 + pos); |
1da177e4 LT |
379 | |
380 | if (!(valid_hooks & (1 << hook))) | |
381 | continue; | |
382 | ||
383 | /* Set initial back pointer. */ | |
384 | e->counters.pcnt = pos; | |
385 | ||
386 | for (;;) { | |
387 | struct ipt_standard_target *t | |
388 | = (void *)ipt_get_target(e); | |
389 | ||
390 | if (e->comefrom & (1 << NF_IP_NUMHOOKS)) { | |
391 | printk("iptables: loop hook %u pos %u %08X.\n", | |
392 | hook, pos, e->comefrom); | |
393 | return 0; | |
394 | } | |
395 | e->comefrom | |
396 | |= ((1 << hook) | (1 << NF_IP_NUMHOOKS)); | |
397 | ||
398 | /* Unconditional return/END. */ | |
399 | if (e->target_offset == sizeof(struct ipt_entry) | |
400 | && (strcmp(t->target.u.user.name, | |
401 | IPT_STANDARD_TARGET) == 0) | |
402 | && t->verdict < 0 | |
403 | && unconditional(&e->ip)) { | |
404 | unsigned int oldpos, size; | |
405 | ||
406 | /* Return: backtrack through the last | |
407 | big jump. */ | |
408 | do { | |
409 | e->comefrom ^= (1<<NF_IP_NUMHOOKS); | |
410 | #ifdef DEBUG_IP_FIREWALL_USER | |
411 | if (e->comefrom | |
412 | & (1 << NF_IP_NUMHOOKS)) { | |
413 | duprintf("Back unset " | |
414 | "on hook %u " | |
415 | "rule %u\n", | |
416 | hook, pos); | |
417 | } | |
418 | #endif | |
419 | oldpos = pos; | |
420 | pos = e->counters.pcnt; | |
421 | e->counters.pcnt = 0; | |
422 | ||
423 | /* We're at the start. */ | |
424 | if (pos == oldpos) | |
425 | goto next; | |
426 | ||
427 | e = (struct ipt_entry *) | |
31836064 | 428 | (entry0 + pos); |
1da177e4 LT |
429 | } while (oldpos == pos + e->next_offset); |
430 | ||
431 | /* Move along one */ | |
432 | size = e->next_offset; | |
433 | e = (struct ipt_entry *) | |
31836064 | 434 | (entry0 + pos + size); |
1da177e4 LT |
435 | e->counters.pcnt = pos; |
436 | pos += size; | |
437 | } else { | |
438 | int newpos = t->verdict; | |
439 | ||
440 | if (strcmp(t->target.u.user.name, | |
441 | IPT_STANDARD_TARGET) == 0 | |
442 | && newpos >= 0) { | |
443 | /* This a jump; chase it. */ | |
444 | duprintf("Jump rule %u -> %u\n", | |
445 | pos, newpos); | |
446 | } else { | |
447 | /* ... this is a fallthru */ | |
448 | newpos = pos + e->next_offset; | |
449 | } | |
450 | e = (struct ipt_entry *) | |
31836064 | 451 | (entry0 + newpos); |
1da177e4 LT |
452 | e->counters.pcnt = pos; |
453 | pos = newpos; | |
454 | } | |
455 | } | |
456 | next: | |
457 | duprintf("Finished chain %u\n", hook); | |
458 | } | |
459 | return 1; | |
460 | } | |
461 | ||
462 | static inline int | |
463 | cleanup_match(struct ipt_entry_match *m, unsigned int *i) | |
464 | { | |
465 | if (i && (*i)-- == 0) | |
466 | return 1; | |
467 | ||
468 | if (m->u.kernel.match->destroy) | |
1c524830 | 469 | m->u.kernel.match->destroy(m->u.kernel.match, m->data, |
1da177e4 LT |
470 | m->u.match_size - sizeof(*m)); |
471 | module_put(m->u.kernel.match->me); | |
472 | return 0; | |
473 | } | |
474 | ||
475 | static inline int | |
476 | standard_check(const struct ipt_entry_target *t, | |
477 | unsigned int max_offset) | |
478 | { | |
479 | struct ipt_standard_target *targ = (void *)t; | |
480 | ||
481 | /* Check standard info. */ | |
1da177e4 LT |
482 | if (targ->verdict >= 0 |
483 | && targ->verdict > max_offset - sizeof(struct ipt_entry)) { | |
484 | duprintf("ipt_standard_check: bad verdict (%i)\n", | |
485 | targ->verdict); | |
486 | return 0; | |
487 | } | |
1da177e4 LT |
488 | if (targ->verdict < -NF_MAX_VERDICT - 1) { |
489 | duprintf("ipt_standard_check: bad negative verdict (%i)\n", | |
490 | targ->verdict); | |
491 | return 0; | |
492 | } | |
493 | return 1; | |
494 | } | |
495 | ||
496 | static inline int | |
497 | check_match(struct ipt_entry_match *m, | |
498 | const char *name, | |
499 | const struct ipt_ip *ip, | |
500 | unsigned int hookmask, | |
501 | unsigned int *i) | |
502 | { | |
503 | struct ipt_match *match; | |
3cdc7c95 | 504 | int ret; |
1da177e4 | 505 | |
2e4e6a17 | 506 | match = try_then_request_module(xt_find_match(AF_INET, m->u.user.name, |
1da177e4 LT |
507 | m->u.user.revision), |
508 | "ipt_%s", m->u.user.name); | |
509 | if (IS_ERR(match) || !match) { | |
510 | duprintf("check_match: `%s' not found\n", m->u.user.name); | |
511 | return match ? PTR_ERR(match) : -ENOENT; | |
512 | } | |
513 | m->u.kernel.match = match; | |
514 | ||
3cdc7c95 PM |
515 | ret = xt_check_match(match, AF_INET, m->u.match_size - sizeof(*m), |
516 | name, hookmask, ip->proto, | |
517 | ip->invflags & IPT_INV_PROTO); | |
518 | if (ret) | |
519 | goto err; | |
520 | ||
1da177e4 | 521 | if (m->u.kernel.match->checkentry |
1c524830 | 522 | && !m->u.kernel.match->checkentry(name, ip, match, m->data, |
1da177e4 LT |
523 | m->u.match_size - sizeof(*m), |
524 | hookmask)) { | |
1da177e4 LT |
525 | duprintf("ip_tables: check failed for `%s'.\n", |
526 | m->u.kernel.match->name); | |
3cdc7c95 PM |
527 | ret = -EINVAL; |
528 | goto err; | |
1da177e4 LT |
529 | } |
530 | ||
531 | (*i)++; | |
532 | return 0; | |
3cdc7c95 PM |
533 | err: |
534 | module_put(m->u.kernel.match->me); | |
535 | return ret; | |
1da177e4 LT |
536 | } |
537 | ||
538 | static struct ipt_target ipt_standard_target; | |
539 | ||
540 | static inline int | |
541 | check_entry(struct ipt_entry *e, const char *name, unsigned int size, | |
542 | unsigned int *i) | |
543 | { | |
544 | struct ipt_entry_target *t; | |
545 | struct ipt_target *target; | |
546 | int ret; | |
547 | unsigned int j; | |
548 | ||
549 | if (!ip_checkentry(&e->ip)) { | |
550 | duprintf("ip_tables: ip check failed %p %s.\n", e, name); | |
551 | return -EINVAL; | |
552 | } | |
553 | ||
554 | j = 0; | |
555 | ret = IPT_MATCH_ITERATE(e, check_match, name, &e->ip, e->comefrom, &j); | |
556 | if (ret != 0) | |
557 | goto cleanup_matches; | |
558 | ||
559 | t = ipt_get_target(e); | |
2e4e6a17 HW |
560 | target = try_then_request_module(xt_find_target(AF_INET, |
561 | t->u.user.name, | |
1da177e4 LT |
562 | t->u.user.revision), |
563 | "ipt_%s", t->u.user.name); | |
564 | if (IS_ERR(target) || !target) { | |
565 | duprintf("check_entry: `%s' not found\n", t->u.user.name); | |
566 | ret = target ? PTR_ERR(target) : -ENOENT; | |
567 | goto cleanup_matches; | |
568 | } | |
569 | t->u.kernel.target = target; | |
570 | ||
3cdc7c95 PM |
571 | ret = xt_check_target(target, AF_INET, t->u.target_size - sizeof(*t), |
572 | name, e->comefrom, e->ip.proto, | |
573 | e->ip.invflags & IPT_INV_PROTO); | |
574 | if (ret) | |
575 | goto err; | |
576 | ||
1da177e4 LT |
577 | if (t->u.kernel.target == &ipt_standard_target) { |
578 | if (!standard_check(t, size)) { | |
579 | ret = -EINVAL; | |
580 | goto cleanup_matches; | |
581 | } | |
582 | } else if (t->u.kernel.target->checkentry | |
1c524830 | 583 | && !t->u.kernel.target->checkentry(name, e, target, t->data, |
1da177e4 LT |
584 | t->u.target_size |
585 | - sizeof(*t), | |
586 | e->comefrom)) { | |
1da177e4 LT |
587 | duprintf("ip_tables: check failed for `%s'.\n", |
588 | t->u.kernel.target->name); | |
589 | ret = -EINVAL; | |
3cdc7c95 | 590 | goto err; |
1da177e4 LT |
591 | } |
592 | ||
593 | (*i)++; | |
594 | return 0; | |
3cdc7c95 PM |
595 | err: |
596 | module_put(t->u.kernel.target->me); | |
1da177e4 LT |
597 | cleanup_matches: |
598 | IPT_MATCH_ITERATE(e, cleanup_match, &j); | |
599 | return ret; | |
600 | } | |
601 | ||
602 | static inline int | |
603 | check_entry_size_and_hooks(struct ipt_entry *e, | |
2e4e6a17 | 604 | struct xt_table_info *newinfo, |
1da177e4 LT |
605 | unsigned char *base, |
606 | unsigned char *limit, | |
607 | const unsigned int *hook_entries, | |
608 | const unsigned int *underflows, | |
609 | unsigned int *i) | |
610 | { | |
611 | unsigned int h; | |
612 | ||
613 | if ((unsigned long)e % __alignof__(struct ipt_entry) != 0 | |
614 | || (unsigned char *)e + sizeof(struct ipt_entry) >= limit) { | |
615 | duprintf("Bad offset %p\n", e); | |
616 | return -EINVAL; | |
617 | } | |
618 | ||
619 | if (e->next_offset | |
620 | < sizeof(struct ipt_entry) + sizeof(struct ipt_entry_target)) { | |
621 | duprintf("checking: element %p size %u\n", | |
622 | e, e->next_offset); | |
623 | return -EINVAL; | |
624 | } | |
625 | ||
626 | /* Check hooks & underflows */ | |
627 | for (h = 0; h < NF_IP_NUMHOOKS; h++) { | |
628 | if ((unsigned char *)e - base == hook_entries[h]) | |
629 | newinfo->hook_entry[h] = hook_entries[h]; | |
630 | if ((unsigned char *)e - base == underflows[h]) | |
631 | newinfo->underflow[h] = underflows[h]; | |
632 | } | |
633 | ||
634 | /* FIXME: underflows must be unconditional, standard verdicts | |
635 | < 0 (not IPT_RETURN). --RR */ | |
636 | ||
637 | /* Clear counters and comefrom */ | |
2e4e6a17 | 638 | e->counters = ((struct xt_counters) { 0, 0 }); |
1da177e4 LT |
639 | e->comefrom = 0; |
640 | ||
641 | (*i)++; | |
642 | return 0; | |
643 | } | |
644 | ||
645 | static inline int | |
646 | cleanup_entry(struct ipt_entry *e, unsigned int *i) | |
647 | { | |
648 | struct ipt_entry_target *t; | |
649 | ||
650 | if (i && (*i)-- == 0) | |
651 | return 1; | |
652 | ||
653 | /* Cleanup all matches */ | |
654 | IPT_MATCH_ITERATE(e, cleanup_match, NULL); | |
655 | t = ipt_get_target(e); | |
656 | if (t->u.kernel.target->destroy) | |
1c524830 | 657 | t->u.kernel.target->destroy(t->u.kernel.target, t->data, |
1da177e4 LT |
658 | t->u.target_size - sizeof(*t)); |
659 | module_put(t->u.kernel.target->me); | |
660 | return 0; | |
661 | } | |
662 | ||
663 | /* Checks and translates the user-supplied table segment (held in | |
664 | newinfo) */ | |
665 | static int | |
666 | translate_table(const char *name, | |
667 | unsigned int valid_hooks, | |
2e4e6a17 | 668 | struct xt_table_info *newinfo, |
31836064 | 669 | void *entry0, |
1da177e4 LT |
670 | unsigned int size, |
671 | unsigned int number, | |
672 | const unsigned int *hook_entries, | |
673 | const unsigned int *underflows) | |
674 | { | |
675 | unsigned int i; | |
676 | int ret; | |
677 | ||
678 | newinfo->size = size; | |
679 | newinfo->number = number; | |
680 | ||
681 | /* Init all hooks to impossible value. */ | |
682 | for (i = 0; i < NF_IP_NUMHOOKS; i++) { | |
683 | newinfo->hook_entry[i] = 0xFFFFFFFF; | |
684 | newinfo->underflow[i] = 0xFFFFFFFF; | |
685 | } | |
686 | ||
687 | duprintf("translate_table: size %u\n", newinfo->size); | |
688 | i = 0; | |
689 | /* Walk through entries, checking offsets. */ | |
31836064 | 690 | ret = IPT_ENTRY_ITERATE(entry0, newinfo->size, |
1da177e4 LT |
691 | check_entry_size_and_hooks, |
692 | newinfo, | |
31836064 ED |
693 | entry0, |
694 | entry0 + size, | |
1da177e4 LT |
695 | hook_entries, underflows, &i); |
696 | if (ret != 0) | |
697 | return ret; | |
698 | ||
699 | if (i != number) { | |
700 | duprintf("translate_table: %u not %u entries\n", | |
701 | i, number); | |
702 | return -EINVAL; | |
703 | } | |
704 | ||
705 | /* Check hooks all assigned */ | |
706 | for (i = 0; i < NF_IP_NUMHOOKS; i++) { | |
707 | /* Only hooks which are valid */ | |
708 | if (!(valid_hooks & (1 << i))) | |
709 | continue; | |
710 | if (newinfo->hook_entry[i] == 0xFFFFFFFF) { | |
711 | duprintf("Invalid hook entry %u %u\n", | |
712 | i, hook_entries[i]); | |
713 | return -EINVAL; | |
714 | } | |
715 | if (newinfo->underflow[i] == 0xFFFFFFFF) { | |
716 | duprintf("Invalid underflow %u %u\n", | |
717 | i, underflows[i]); | |
718 | return -EINVAL; | |
719 | } | |
720 | } | |
721 | ||
31836064 | 722 | if (!mark_source_chains(newinfo, valid_hooks, entry0)) |
1da177e4 LT |
723 | return -ELOOP; |
724 | ||
725 | /* Finally, each sanity check must pass */ | |
726 | i = 0; | |
31836064 | 727 | ret = IPT_ENTRY_ITERATE(entry0, newinfo->size, |
1da177e4 LT |
728 | check_entry, name, size, &i); |
729 | ||
730 | if (ret != 0) { | |
31836064 | 731 | IPT_ENTRY_ITERATE(entry0, newinfo->size, |
1da177e4 LT |
732 | cleanup_entry, &i); |
733 | return ret; | |
734 | } | |
735 | ||
736 | /* And one copy for every other CPU */ | |
c8923c6b | 737 | for_each_cpu(i) { |
31836064 ED |
738 | if (newinfo->entries[i] && newinfo->entries[i] != entry0) |
739 | memcpy(newinfo->entries[i], entry0, newinfo->size); | |
1da177e4 LT |
740 | } |
741 | ||
742 | return ret; | |
743 | } | |
744 | ||
1da177e4 LT |
745 | /* Gets counters. */ |
746 | static inline int | |
747 | add_entry_to_counter(const struct ipt_entry *e, | |
2e4e6a17 | 748 | struct xt_counters total[], |
1da177e4 LT |
749 | unsigned int *i) |
750 | { | |
751 | ADD_COUNTER(total[*i], e->counters.bcnt, e->counters.pcnt); | |
752 | ||
753 | (*i)++; | |
754 | return 0; | |
755 | } | |
756 | ||
31836064 ED |
757 | static inline int |
758 | set_entry_to_counter(const struct ipt_entry *e, | |
759 | struct ipt_counters total[], | |
760 | unsigned int *i) | |
761 | { | |
762 | SET_COUNTER(total[*i], e->counters.bcnt, e->counters.pcnt); | |
763 | ||
764 | (*i)++; | |
765 | return 0; | |
766 | } | |
767 | ||
1da177e4 | 768 | static void |
2e4e6a17 HW |
769 | get_counters(const struct xt_table_info *t, |
770 | struct xt_counters counters[]) | |
1da177e4 LT |
771 | { |
772 | unsigned int cpu; | |
773 | unsigned int i; | |
31836064 ED |
774 | unsigned int curcpu; |
775 | ||
776 | /* Instead of clearing (by a previous call to memset()) | |
777 | * the counters and using adds, we set the counters | |
778 | * with data used by 'current' CPU | |
779 | * We dont care about preemption here. | |
780 | */ | |
781 | curcpu = raw_smp_processor_id(); | |
782 | ||
783 | i = 0; | |
784 | IPT_ENTRY_ITERATE(t->entries[curcpu], | |
785 | t->size, | |
786 | set_entry_to_counter, | |
787 | counters, | |
788 | &i); | |
1da177e4 | 789 | |
c8923c6b | 790 | for_each_cpu(cpu) { |
31836064 ED |
791 | if (cpu == curcpu) |
792 | continue; | |
1da177e4 | 793 | i = 0; |
31836064 | 794 | IPT_ENTRY_ITERATE(t->entries[cpu], |
1da177e4 LT |
795 | t->size, |
796 | add_entry_to_counter, | |
797 | counters, | |
798 | &i); | |
799 | } | |
800 | } | |
801 | ||
802 | static int | |
803 | copy_entries_to_user(unsigned int total_size, | |
804 | struct ipt_table *table, | |
805 | void __user *userptr) | |
806 | { | |
807 | unsigned int off, num, countersize; | |
808 | struct ipt_entry *e; | |
2e4e6a17 HW |
809 | struct xt_counters *counters; |
810 | struct xt_table_info *private = table->private; | |
1da177e4 | 811 | int ret = 0; |
31836064 | 812 | void *loc_cpu_entry; |
1da177e4 LT |
813 | |
814 | /* We need atomic snapshot of counters: rest doesn't change | |
815 | (other than comefrom, which userspace doesn't care | |
816 | about). */ | |
2e4e6a17 | 817 | countersize = sizeof(struct xt_counters) * private->number; |
31836064 | 818 | counters = vmalloc_node(countersize, numa_node_id()); |
1da177e4 LT |
819 | |
820 | if (counters == NULL) | |
821 | return -ENOMEM; | |
822 | ||
823 | /* First, sum counters... */ | |
1da177e4 | 824 | write_lock_bh(&table->lock); |
2e4e6a17 | 825 | get_counters(private, counters); |
1da177e4 LT |
826 | write_unlock_bh(&table->lock); |
827 | ||
31836064 ED |
828 | /* choose the copy that is on our node/cpu, ... |
829 | * This choice is lazy (because current thread is | |
830 | * allowed to migrate to another cpu) | |
831 | */ | |
2e4e6a17 | 832 | loc_cpu_entry = private->entries[raw_smp_processor_id()]; |
31836064 ED |
833 | /* ... then copy entire thing ... */ |
834 | if (copy_to_user(userptr, loc_cpu_entry, total_size) != 0) { | |
1da177e4 LT |
835 | ret = -EFAULT; |
836 | goto free_counters; | |
837 | } | |
838 | ||
839 | /* FIXME: use iterator macros --RR */ | |
840 | /* ... then go back and fix counters and names */ | |
841 | for (off = 0, num = 0; off < total_size; off += e->next_offset, num++){ | |
842 | unsigned int i; | |
843 | struct ipt_entry_match *m; | |
844 | struct ipt_entry_target *t; | |
845 | ||
31836064 | 846 | e = (struct ipt_entry *)(loc_cpu_entry + off); |
1da177e4 LT |
847 | if (copy_to_user(userptr + off |
848 | + offsetof(struct ipt_entry, counters), | |
849 | &counters[num], | |
850 | sizeof(counters[num])) != 0) { | |
851 | ret = -EFAULT; | |
852 | goto free_counters; | |
853 | } | |
854 | ||
855 | for (i = sizeof(struct ipt_entry); | |
856 | i < e->target_offset; | |
857 | i += m->u.match_size) { | |
858 | m = (void *)e + i; | |
859 | ||
860 | if (copy_to_user(userptr + off + i | |
861 | + offsetof(struct ipt_entry_match, | |
862 | u.user.name), | |
863 | m->u.kernel.match->name, | |
864 | strlen(m->u.kernel.match->name)+1) | |
865 | != 0) { | |
866 | ret = -EFAULT; | |
867 | goto free_counters; | |
868 | } | |
869 | } | |
870 | ||
871 | t = ipt_get_target(e); | |
872 | if (copy_to_user(userptr + off + e->target_offset | |
873 | + offsetof(struct ipt_entry_target, | |
874 | u.user.name), | |
875 | t->u.kernel.target->name, | |
876 | strlen(t->u.kernel.target->name)+1) != 0) { | |
877 | ret = -EFAULT; | |
878 | goto free_counters; | |
879 | } | |
880 | } | |
881 | ||
882 | free_counters: | |
883 | vfree(counters); | |
884 | return ret; | |
885 | } | |
886 | ||
887 | static int | |
888 | get_entries(const struct ipt_get_entries *entries, | |
889 | struct ipt_get_entries __user *uptr) | |
890 | { | |
891 | int ret; | |
892 | struct ipt_table *t; | |
893 | ||
2e4e6a17 | 894 | t = xt_find_table_lock(AF_INET, entries->name); |
1da177e4 | 895 | if (t && !IS_ERR(t)) { |
2e4e6a17 | 896 | struct xt_table_info *private = t->private; |
1da177e4 | 897 | duprintf("t->private->number = %u\n", |
2e4e6a17 HW |
898 | private->number); |
899 | if (entries->size == private->size) | |
900 | ret = copy_entries_to_user(private->size, | |
1da177e4 LT |
901 | t, uptr->entrytable); |
902 | else { | |
903 | duprintf("get_entries: I've got %u not %u!\n", | |
2e4e6a17 | 904 | private->size, |
1da177e4 LT |
905 | entries->size); |
906 | ret = -EINVAL; | |
907 | } | |
908 | module_put(t->me); | |
2e4e6a17 | 909 | xt_table_unlock(t); |
1da177e4 LT |
910 | } else |
911 | ret = t ? PTR_ERR(t) : -ENOENT; | |
912 | ||
913 | return ret; | |
914 | } | |
915 | ||
916 | static int | |
917 | do_replace(void __user *user, unsigned int len) | |
918 | { | |
919 | int ret; | |
920 | struct ipt_replace tmp; | |
921 | struct ipt_table *t; | |
2e4e6a17 HW |
922 | struct xt_table_info *newinfo, *oldinfo; |
923 | struct xt_counters *counters; | |
31836064 | 924 | void *loc_cpu_entry, *loc_cpu_old_entry; |
1da177e4 LT |
925 | |
926 | if (copy_from_user(&tmp, user, sizeof(tmp)) != 0) | |
927 | return -EFAULT; | |
928 | ||
929 | /* Hack: Causes ipchains to give correct error msg --RR */ | |
930 | if (len != sizeof(tmp) + tmp.size) | |
931 | return -ENOPROTOOPT; | |
932 | ||
ee4bb818 KK |
933 | /* overflow check */ |
934 | if (tmp.size >= (INT_MAX - sizeof(struct xt_table_info)) / NR_CPUS - | |
935 | SMP_CACHE_BYTES) | |
936 | return -ENOMEM; | |
937 | if (tmp.num_counters >= INT_MAX / sizeof(struct xt_counters)) | |
938 | return -ENOMEM; | |
939 | ||
2e4e6a17 | 940 | newinfo = xt_alloc_table_info(tmp.size); |
1da177e4 LT |
941 | if (!newinfo) |
942 | return -ENOMEM; | |
943 | ||
31836064 ED |
944 | /* choose the copy that is our node/cpu */ |
945 | loc_cpu_entry = newinfo->entries[raw_smp_processor_id()]; | |
946 | if (copy_from_user(loc_cpu_entry, user + sizeof(tmp), | |
1da177e4 LT |
947 | tmp.size) != 0) { |
948 | ret = -EFAULT; | |
949 | goto free_newinfo; | |
950 | } | |
951 | ||
2e4e6a17 | 952 | counters = vmalloc(tmp.num_counters * sizeof(struct xt_counters)); |
1da177e4 LT |
953 | if (!counters) { |
954 | ret = -ENOMEM; | |
955 | goto free_newinfo; | |
956 | } | |
1da177e4 LT |
957 | |
958 | ret = translate_table(tmp.name, tmp.valid_hooks, | |
31836064 | 959 | newinfo, loc_cpu_entry, tmp.size, tmp.num_entries, |
1da177e4 LT |
960 | tmp.hook_entry, tmp.underflow); |
961 | if (ret != 0) | |
962 | goto free_newinfo_counters; | |
963 | ||
964 | duprintf("ip_tables: Translated table\n"); | |
965 | ||
2e4e6a17 | 966 | t = try_then_request_module(xt_find_table_lock(AF_INET, tmp.name), |
1da177e4 LT |
967 | "iptable_%s", tmp.name); |
968 | if (!t || IS_ERR(t)) { | |
969 | ret = t ? PTR_ERR(t) : -ENOENT; | |
970 | goto free_newinfo_counters_untrans; | |
971 | } | |
972 | ||
973 | /* You lied! */ | |
974 | if (tmp.valid_hooks != t->valid_hooks) { | |
975 | duprintf("Valid hook crap: %08X vs %08X\n", | |
976 | tmp.valid_hooks, t->valid_hooks); | |
977 | ret = -EINVAL; | |
978 | goto put_module; | |
979 | } | |
980 | ||
2e4e6a17 | 981 | oldinfo = xt_replace_table(t, tmp.num_counters, newinfo, &ret); |
1da177e4 LT |
982 | if (!oldinfo) |
983 | goto put_module; | |
984 | ||
985 | /* Update module usage count based on number of rules */ | |
986 | duprintf("do_replace: oldnum=%u, initnum=%u, newnum=%u\n", | |
987 | oldinfo->number, oldinfo->initial_entries, newinfo->number); | |
988 | if ((oldinfo->number > oldinfo->initial_entries) || | |
989 | (newinfo->number <= oldinfo->initial_entries)) | |
990 | module_put(t->me); | |
991 | if ((oldinfo->number > oldinfo->initial_entries) && | |
992 | (newinfo->number <= oldinfo->initial_entries)) | |
993 | module_put(t->me); | |
994 | ||
995 | /* Get the old counters. */ | |
996 | get_counters(oldinfo, counters); | |
997 | /* Decrease module usage counts and free resource */ | |
31836064 ED |
998 | loc_cpu_old_entry = oldinfo->entries[raw_smp_processor_id()]; |
999 | IPT_ENTRY_ITERATE(loc_cpu_old_entry, oldinfo->size, cleanup_entry,NULL); | |
2e4e6a17 | 1000 | xt_free_table_info(oldinfo); |
1da177e4 | 1001 | if (copy_to_user(tmp.counters, counters, |
2e4e6a17 | 1002 | sizeof(struct xt_counters) * tmp.num_counters) != 0) |
1da177e4 LT |
1003 | ret = -EFAULT; |
1004 | vfree(counters); | |
2e4e6a17 | 1005 | xt_table_unlock(t); |
1da177e4 LT |
1006 | return ret; |
1007 | ||
1008 | put_module: | |
1009 | module_put(t->me); | |
2e4e6a17 | 1010 | xt_table_unlock(t); |
1da177e4 | 1011 | free_newinfo_counters_untrans: |
31836064 | 1012 | IPT_ENTRY_ITERATE(loc_cpu_entry, newinfo->size, cleanup_entry,NULL); |
1da177e4 LT |
1013 | free_newinfo_counters: |
1014 | vfree(counters); | |
1015 | free_newinfo: | |
2e4e6a17 | 1016 | xt_free_table_info(newinfo); |
1da177e4 LT |
1017 | return ret; |
1018 | } | |
1019 | ||
1020 | /* We're lazy, and add to the first CPU; overflow works its fey magic | |
1021 | * and everything is OK. */ | |
1022 | static inline int | |
1023 | add_counter_to_entry(struct ipt_entry *e, | |
2e4e6a17 | 1024 | const struct xt_counters addme[], |
1da177e4 LT |
1025 | unsigned int *i) |
1026 | { | |
1027 | #if 0 | |
1028 | duprintf("add_counter: Entry %u %lu/%lu + %lu/%lu\n", | |
1029 | *i, | |
1030 | (long unsigned int)e->counters.pcnt, | |
1031 | (long unsigned int)e->counters.bcnt, | |
1032 | (long unsigned int)addme[*i].pcnt, | |
1033 | (long unsigned int)addme[*i].bcnt); | |
1034 | #endif | |
1035 | ||
1036 | ADD_COUNTER(e->counters, addme[*i].bcnt, addme[*i].pcnt); | |
1037 | ||
1038 | (*i)++; | |
1039 | return 0; | |
1040 | } | |
1041 | ||
1042 | static int | |
1043 | do_add_counters(void __user *user, unsigned int len) | |
1044 | { | |
1045 | unsigned int i; | |
2e4e6a17 | 1046 | struct xt_counters_info tmp, *paddc; |
1da177e4 | 1047 | struct ipt_table *t; |
2e4e6a17 | 1048 | struct xt_table_info *private; |
1da177e4 | 1049 | int ret = 0; |
31836064 | 1050 | void *loc_cpu_entry; |
1da177e4 LT |
1051 | |
1052 | if (copy_from_user(&tmp, user, sizeof(tmp)) != 0) | |
1053 | return -EFAULT; | |
1054 | ||
2e4e6a17 | 1055 | if (len != sizeof(tmp) + tmp.num_counters*sizeof(struct xt_counters)) |
1da177e4 LT |
1056 | return -EINVAL; |
1057 | ||
31836064 | 1058 | paddc = vmalloc_node(len, numa_node_id()); |
1da177e4 LT |
1059 | if (!paddc) |
1060 | return -ENOMEM; | |
1061 | ||
1062 | if (copy_from_user(paddc, user, len) != 0) { | |
1063 | ret = -EFAULT; | |
1064 | goto free; | |
1065 | } | |
1066 | ||
2e4e6a17 | 1067 | t = xt_find_table_lock(AF_INET, tmp.name); |
1da177e4 LT |
1068 | if (!t || IS_ERR(t)) { |
1069 | ret = t ? PTR_ERR(t) : -ENOENT; | |
1070 | goto free; | |
1071 | } | |
1072 | ||
1073 | write_lock_bh(&t->lock); | |
2e4e6a17 HW |
1074 | private = t->private; |
1075 | if (private->number != paddc->num_counters) { | |
1da177e4 LT |
1076 | ret = -EINVAL; |
1077 | goto unlock_up_free; | |
1078 | } | |
1079 | ||
1080 | i = 0; | |
31836064 | 1081 | /* Choose the copy that is on our node */ |
2e4e6a17 | 1082 | loc_cpu_entry = private->entries[raw_smp_processor_id()]; |
31836064 | 1083 | IPT_ENTRY_ITERATE(loc_cpu_entry, |
2e4e6a17 | 1084 | private->size, |
1da177e4 LT |
1085 | add_counter_to_entry, |
1086 | paddc->counters, | |
1087 | &i); | |
1088 | unlock_up_free: | |
1089 | write_unlock_bh(&t->lock); | |
2e4e6a17 | 1090 | xt_table_unlock(t); |
1da177e4 LT |
1091 | module_put(t->me); |
1092 | free: | |
1093 | vfree(paddc); | |
1094 | ||
1095 | return ret; | |
1096 | } | |
1097 | ||
1098 | static int | |
1099 | do_ipt_set_ctl(struct sock *sk, int cmd, void __user *user, unsigned int len) | |
1100 | { | |
1101 | int ret; | |
1102 | ||
1103 | if (!capable(CAP_NET_ADMIN)) | |
1104 | return -EPERM; | |
1105 | ||
1106 | switch (cmd) { | |
1107 | case IPT_SO_SET_REPLACE: | |
1108 | ret = do_replace(user, len); | |
1109 | break; | |
1110 | ||
1111 | case IPT_SO_SET_ADD_COUNTERS: | |
1112 | ret = do_add_counters(user, len); | |
1113 | break; | |
1114 | ||
1115 | default: | |
1116 | duprintf("do_ipt_set_ctl: unknown request %i\n", cmd); | |
1117 | ret = -EINVAL; | |
1118 | } | |
1119 | ||
1120 | return ret; | |
1121 | } | |
1122 | ||
1123 | static int | |
1124 | do_ipt_get_ctl(struct sock *sk, int cmd, void __user *user, int *len) | |
1125 | { | |
1126 | int ret; | |
1127 | ||
1128 | if (!capable(CAP_NET_ADMIN)) | |
1129 | return -EPERM; | |
1130 | ||
1131 | switch (cmd) { | |
1132 | case IPT_SO_GET_INFO: { | |
1133 | char name[IPT_TABLE_MAXNAMELEN]; | |
1134 | struct ipt_table *t; | |
1135 | ||
1136 | if (*len != sizeof(struct ipt_getinfo)) { | |
1137 | duprintf("length %u != %u\n", *len, | |
1138 | sizeof(struct ipt_getinfo)); | |
1139 | ret = -EINVAL; | |
1140 | break; | |
1141 | } | |
1142 | ||
1143 | if (copy_from_user(name, user, sizeof(name)) != 0) { | |
1144 | ret = -EFAULT; | |
1145 | break; | |
1146 | } | |
1147 | name[IPT_TABLE_MAXNAMELEN-1] = '\0'; | |
1148 | ||
2e4e6a17 | 1149 | t = try_then_request_module(xt_find_table_lock(AF_INET, name), |
1da177e4 LT |
1150 | "iptable_%s", name); |
1151 | if (t && !IS_ERR(t)) { | |
1152 | struct ipt_getinfo info; | |
2e4e6a17 | 1153 | struct xt_table_info *private = t->private; |
1da177e4 LT |
1154 | |
1155 | info.valid_hooks = t->valid_hooks; | |
2e4e6a17 | 1156 | memcpy(info.hook_entry, private->hook_entry, |
1da177e4 | 1157 | sizeof(info.hook_entry)); |
2e4e6a17 | 1158 | memcpy(info.underflow, private->underflow, |
1da177e4 | 1159 | sizeof(info.underflow)); |
2e4e6a17 HW |
1160 | info.num_entries = private->number; |
1161 | info.size = private->size; | |
1da177e4 LT |
1162 | memcpy(info.name, name, sizeof(info.name)); |
1163 | ||
1164 | if (copy_to_user(user, &info, *len) != 0) | |
1165 | ret = -EFAULT; | |
1166 | else | |
1167 | ret = 0; | |
2e4e6a17 | 1168 | xt_table_unlock(t); |
1da177e4 LT |
1169 | module_put(t->me); |
1170 | } else | |
1171 | ret = t ? PTR_ERR(t) : -ENOENT; | |
1172 | } | |
1173 | break; | |
1174 | ||
1175 | case IPT_SO_GET_ENTRIES: { | |
1176 | struct ipt_get_entries get; | |
1177 | ||
1178 | if (*len < sizeof(get)) { | |
1179 | duprintf("get_entries: %u < %u\n", *len, sizeof(get)); | |
1180 | ret = -EINVAL; | |
1181 | } else if (copy_from_user(&get, user, sizeof(get)) != 0) { | |
1182 | ret = -EFAULT; | |
1183 | } else if (*len != sizeof(struct ipt_get_entries) + get.size) { | |
1184 | duprintf("get_entries: %u != %u\n", *len, | |
1185 | sizeof(struct ipt_get_entries) + get.size); | |
1186 | ret = -EINVAL; | |
1187 | } else | |
1188 | ret = get_entries(&get, user); | |
1189 | break; | |
1190 | } | |
1191 | ||
1192 | case IPT_SO_GET_REVISION_MATCH: | |
1193 | case IPT_SO_GET_REVISION_TARGET: { | |
1194 | struct ipt_get_revision rev; | |
2e4e6a17 | 1195 | int target; |
1da177e4 LT |
1196 | |
1197 | if (*len != sizeof(rev)) { | |
1198 | ret = -EINVAL; | |
1199 | break; | |
1200 | } | |
1201 | if (copy_from_user(&rev, user, sizeof(rev)) != 0) { | |
1202 | ret = -EFAULT; | |
1203 | break; | |
1204 | } | |
1205 | ||
1206 | if (cmd == IPT_SO_GET_REVISION_TARGET) | |
2e4e6a17 | 1207 | target = 1; |
1da177e4 | 1208 | else |
2e4e6a17 | 1209 | target = 0; |
1da177e4 | 1210 | |
2e4e6a17 HW |
1211 | try_then_request_module(xt_find_revision(AF_INET, rev.name, |
1212 | rev.revision, | |
1213 | target, &ret), | |
1da177e4 LT |
1214 | "ipt_%s", rev.name); |
1215 | break; | |
1216 | } | |
1217 | ||
1218 | default: | |
1219 | duprintf("do_ipt_get_ctl: unknown request %i\n", cmd); | |
1220 | ret = -EINVAL; | |
1221 | } | |
1222 | ||
1223 | return ret; | |
1224 | } | |
1225 | ||
2e4e6a17 | 1226 | int ipt_register_table(struct xt_table *table, const struct ipt_replace *repl) |
1da177e4 LT |
1227 | { |
1228 | int ret; | |
2e4e6a17 HW |
1229 | struct xt_table_info *newinfo; |
1230 | static struct xt_table_info bootstrap | |
1da177e4 | 1231 | = { 0, 0, 0, { 0 }, { 0 }, { } }; |
31836064 | 1232 | void *loc_cpu_entry; |
1da177e4 | 1233 | |
2e4e6a17 | 1234 | newinfo = xt_alloc_table_info(repl->size); |
1da177e4 LT |
1235 | if (!newinfo) |
1236 | return -ENOMEM; | |
1237 | ||
31836064 ED |
1238 | /* choose the copy on our node/cpu |
1239 | * but dont care of preemption | |
1240 | */ | |
1241 | loc_cpu_entry = newinfo->entries[raw_smp_processor_id()]; | |
1242 | memcpy(loc_cpu_entry, repl->entries, repl->size); | |
1da177e4 LT |
1243 | |
1244 | ret = translate_table(table->name, table->valid_hooks, | |
31836064 | 1245 | newinfo, loc_cpu_entry, repl->size, |
1da177e4 LT |
1246 | repl->num_entries, |
1247 | repl->hook_entry, | |
1248 | repl->underflow); | |
1249 | if (ret != 0) { | |
2e4e6a17 | 1250 | xt_free_table_info(newinfo); |
1da177e4 LT |
1251 | return ret; |
1252 | } | |
1253 | ||
2e4e6a17 HW |
1254 | if (xt_register_table(table, &bootstrap, newinfo) != 0) { |
1255 | xt_free_table_info(newinfo); | |
1da177e4 LT |
1256 | return ret; |
1257 | } | |
1258 | ||
2e4e6a17 | 1259 | return 0; |
1da177e4 LT |
1260 | } |
1261 | ||
1262 | void ipt_unregister_table(struct ipt_table *table) | |
1263 | { | |
2e4e6a17 | 1264 | struct xt_table_info *private; |
31836064 ED |
1265 | void *loc_cpu_entry; |
1266 | ||
2e4e6a17 | 1267 | private = xt_unregister_table(table); |
1da177e4 LT |
1268 | |
1269 | /* Decrease module usage counts and free resources */ | |
2e4e6a17 HW |
1270 | loc_cpu_entry = private->entries[raw_smp_processor_id()]; |
1271 | IPT_ENTRY_ITERATE(loc_cpu_entry, private->size, cleanup_entry, NULL); | |
1272 | xt_free_table_info(private); | |
1da177e4 LT |
1273 | } |
1274 | ||
1275 | /* Returns 1 if the type and code is matched by the range, 0 otherwise */ | |
1276 | static inline int | |
1277 | icmp_type_code_match(u_int8_t test_type, u_int8_t min_code, u_int8_t max_code, | |
1278 | u_int8_t type, u_int8_t code, | |
1279 | int invert) | |
1280 | { | |
1281 | return ((test_type == 0xFF) || (type == test_type && code >= min_code && code <= max_code)) | |
1282 | ^ invert; | |
1283 | } | |
1284 | ||
1285 | static int | |
1286 | icmp_match(const struct sk_buff *skb, | |
1287 | const struct net_device *in, | |
1288 | const struct net_device *out, | |
c4986734 | 1289 | const struct xt_match *match, |
1da177e4 LT |
1290 | const void *matchinfo, |
1291 | int offset, | |
2e4e6a17 | 1292 | unsigned int protoff, |
1da177e4 LT |
1293 | int *hotdrop) |
1294 | { | |
1295 | struct icmphdr _icmph, *ic; | |
1296 | const struct ipt_icmp *icmpinfo = matchinfo; | |
1297 | ||
1298 | /* Must not be a fragment. */ | |
1299 | if (offset) | |
1300 | return 0; | |
1301 | ||
2e4e6a17 | 1302 | ic = skb_header_pointer(skb, protoff, sizeof(_icmph), &_icmph); |
1da177e4 LT |
1303 | if (ic == NULL) { |
1304 | /* We've been asked to examine this packet, and we | |
1305 | * can't. Hence, no choice but to drop. | |
1306 | */ | |
1307 | duprintf("Dropping evil ICMP tinygram.\n"); | |
1308 | *hotdrop = 1; | |
1309 | return 0; | |
1310 | } | |
1311 | ||
1312 | return icmp_type_code_match(icmpinfo->type, | |
1313 | icmpinfo->code[0], | |
1314 | icmpinfo->code[1], | |
1315 | ic->type, ic->code, | |
1316 | !!(icmpinfo->invflags&IPT_ICMP_INV)); | |
1317 | } | |
1318 | ||
1319 | /* Called when user tries to insert an entry of this type. */ | |
1320 | static int | |
1321 | icmp_checkentry(const char *tablename, | |
2e4e6a17 | 1322 | const void *info, |
c4986734 | 1323 | const struct xt_match *match, |
1da177e4 LT |
1324 | void *matchinfo, |
1325 | unsigned int matchsize, | |
1326 | unsigned int hook_mask) | |
1327 | { | |
1328 | const struct ipt_icmp *icmpinfo = matchinfo; | |
1329 | ||
1d5cd909 PM |
1330 | /* Must specify no unknown invflags */ |
1331 | return !(icmpinfo->invflags & ~IPT_ICMP_INV); | |
1da177e4 LT |
1332 | } |
1333 | ||
1334 | /* The built-in targets: standard (NULL) and error. */ | |
1335 | static struct ipt_target ipt_standard_target = { | |
1336 | .name = IPT_STANDARD_TARGET, | |
1d5cd909 | 1337 | .targetsize = sizeof(int), |
a45049c5 | 1338 | .family = AF_INET, |
1da177e4 LT |
1339 | }; |
1340 | ||
1341 | static struct ipt_target ipt_error_target = { | |
1342 | .name = IPT_ERROR_TARGET, | |
1343 | .target = ipt_error, | |
1d5cd909 | 1344 | .targetsize = IPT_FUNCTION_MAXNAMELEN, |
a45049c5 | 1345 | .family = AF_INET, |
1da177e4 LT |
1346 | }; |
1347 | ||
1348 | static struct nf_sockopt_ops ipt_sockopts = { | |
1349 | .pf = PF_INET, | |
1350 | .set_optmin = IPT_BASE_CTL, | |
1351 | .set_optmax = IPT_SO_SET_MAX+1, | |
1352 | .set = do_ipt_set_ctl, | |
1353 | .get_optmin = IPT_BASE_CTL, | |
1354 | .get_optmax = IPT_SO_GET_MAX+1, | |
1355 | .get = do_ipt_get_ctl, | |
1356 | }; | |
1357 | ||
1da177e4 LT |
1358 | static struct ipt_match icmp_matchstruct = { |
1359 | .name = "icmp", | |
1d5cd909 PM |
1360 | .match = icmp_match, |
1361 | .matchsize = sizeof(struct ipt_icmp), | |
1362 | .proto = IPPROTO_ICMP, | |
a45049c5 | 1363 | .family = AF_INET, |
1d5cd909 | 1364 | .checkentry = icmp_checkentry, |
1da177e4 LT |
1365 | }; |
1366 | ||
65b4b4e8 | 1367 | static int __init ip_tables_init(void) |
1da177e4 LT |
1368 | { |
1369 | int ret; | |
1370 | ||
2e4e6a17 HW |
1371 | xt_proto_init(AF_INET); |
1372 | ||
1da177e4 | 1373 | /* Noone else will be downing sem now, so we won't sleep */ |
a45049c5 PNA |
1374 | xt_register_target(&ipt_standard_target); |
1375 | xt_register_target(&ipt_error_target); | |
1376 | xt_register_match(&icmp_matchstruct); | |
1da177e4 LT |
1377 | |
1378 | /* Register setsockopt */ | |
1379 | ret = nf_register_sockopt(&ipt_sockopts); | |
1380 | if (ret < 0) { | |
1381 | duprintf("Unable to register sockopts.\n"); | |
1382 | return ret; | |
1383 | } | |
1384 | ||
2e4e6a17 | 1385 | printk("ip_tables: (C) 2000-2006 Netfilter Core Team\n"); |
1da177e4 LT |
1386 | return 0; |
1387 | } | |
1388 | ||
65b4b4e8 | 1389 | static void __exit ip_tables_fini(void) |
1da177e4 LT |
1390 | { |
1391 | nf_unregister_sockopt(&ipt_sockopts); | |
2e4e6a17 | 1392 | |
a45049c5 PNA |
1393 | xt_unregister_match(&icmp_matchstruct); |
1394 | xt_unregister_target(&ipt_error_target); | |
1395 | xt_unregister_target(&ipt_standard_target); | |
2e4e6a17 HW |
1396 | |
1397 | xt_proto_fini(AF_INET); | |
1da177e4 LT |
1398 | } |
1399 | ||
1400 | EXPORT_SYMBOL(ipt_register_table); | |
1401 | EXPORT_SYMBOL(ipt_unregister_table); | |
1da177e4 | 1402 | EXPORT_SYMBOL(ipt_do_table); |
65b4b4e8 AM |
1403 | module_init(ip_tables_init); |
1404 | module_exit(ip_tables_fini); |