ip_tables.c 54.7 KB
Newer Older
L
Linus Torvalds 已提交
1 2 3 4
/*
 * Packet matching code.
 *
 * 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>
L
Linus Torvalds 已提交
7 8 9 10 11
 *
 * This program is free software; you can redistribute it and/or modify
 * it under the terms of the GNU General Public License version 2 as
 * published by the Free Software Foundation.
 */
12
#define pr_fmt(fmt) KBUILD_MODNAME ": " fmt
L
Linus Torvalds 已提交
13
#include <linux/cache.h>
14
#include <linux/capability.h>
L
Linus Torvalds 已提交
15 16 17 18 19 20 21
#include <linux/skbuff.h>
#include <linux/kmod.h>
#include <linux/vmalloc.h>
#include <linux/netdevice.h>
#include <linux/module.h>
#include <linux/icmp.h>
#include <net/ip.h>
22
#include <net/compat.h>
L
Linus Torvalds 已提交
23
#include <asm/uaccess.h>
I
Ingo Molnar 已提交
24
#include <linux/mutex.h>
L
Linus Torvalds 已提交
25 26
#include <linux/proc_fs.h>
#include <linux/err.h>
27
#include <linux/cpumask.h>
L
Linus Torvalds 已提交
28

29
#include <linux/netfilter/x_tables.h>
L
Linus Torvalds 已提交
30
#include <linux/netfilter_ipv4/ip_tables.h>
31
#include <net/netfilter/nf_log.h>
32
#include "../../netfilter/xt_repldata.h"
L
Linus Torvalds 已提交
33 34 35 36 37 38 39 40 41 42

MODULE_LICENSE("GPL");
MODULE_AUTHOR("Netfilter Core Team <coreteam@netfilter.org>");
MODULE_DESCRIPTION("IPv4 packet filter");

/*#define DEBUG_IP_FIREWALL*/
/*#define DEBUG_ALLOW_ALL*/ /* Useful for remote debugging */
/*#define DEBUG_IP_FIREWALL_USER*/

#ifdef DEBUG_IP_FIREWALL
43
#define dprintf(format, args...) pr_info(format , ## args)
L
Linus Torvalds 已提交
44 45 46 47 48
#else
#define dprintf(format, args...)
#endif

#ifdef DEBUG_IP_FIREWALL_USER
49
#define duprintf(format, args...) pr_info(format , ## args)
L
Linus Torvalds 已提交
50 51 52 53 54
#else
#define duprintf(format, args...)
#endif

#ifdef CONFIG_NETFILTER_DEBUG
55
#define IP_NF_ASSERT(x)		WARN_ON(!(x))
L
Linus Torvalds 已提交
56 57 58 59 60 61 62 63 64 65
#else
#define IP_NF_ASSERT(x)
#endif

#if 0
/* All the better to debug you with... */
#define static
#define inline
#endif

66 67 68 69 70 71
void *ipt_alloc_initial_table(const struct xt_table *info)
{
	return xt_alloc_initial_table(ipt, IPT);
}
EXPORT_SYMBOL_GPL(ipt_alloc_initial_table);

L
Linus Torvalds 已提交
72
/* Returns whether matches rule or not. */
73
/* Performance critical - called for every packet */
74
static inline bool
L
Linus Torvalds 已提交
75 76 77 78 79 80 81 82
ip_packet_match(const struct iphdr *ip,
		const char *indev,
		const char *outdev,
		const struct ipt_ip *ipinfo,
		int isfrag)
{
	unsigned long ret;

83
#define FWINV(bool, invflg) ((bool) ^ !!(ipinfo->invflags & (invflg)))
L
Linus Torvalds 已提交
84 85

	if (FWINV((ip->saddr&ipinfo->smsk.s_addr) != ipinfo->src.s_addr,
86 87 88
		  IPT_INV_SRCIP) ||
	    FWINV((ip->daddr&ipinfo->dmsk.s_addr) != ipinfo->dst.s_addr,
		  IPT_INV_DSTIP)) {
L
Linus Torvalds 已提交
89 90
		dprintf("Source or dest mismatch.\n");

91 92
		dprintf("SRC: %pI4. Mask: %pI4. Target: %pI4.%s\n",
			&ip->saddr, &ipinfo->smsk.s_addr, &ipinfo->src.s_addr,
L
Linus Torvalds 已提交
93
			ipinfo->invflags & IPT_INV_SRCIP ? " (INV)" : "");
94 95
		dprintf("DST: %pI4 Mask: %pI4 Target: %pI4.%s\n",
			&ip->daddr, &ipinfo->dmsk.s_addr, &ipinfo->dst.s_addr,
L
Linus Torvalds 已提交
96
			ipinfo->invflags & IPT_INV_DSTIP ? " (INV)" : "");
97
		return false;
L
Linus Torvalds 已提交
98 99
	}

100
	ret = ifname_compare_aligned(indev, ipinfo->iniface, ipinfo->iniface_mask);
L
Linus Torvalds 已提交
101 102 103 104 105

	if (FWINV(ret != 0, IPT_INV_VIA_IN)) {
		dprintf("VIA in mismatch (%s vs %s).%s\n",
			indev, ipinfo->iniface,
			ipinfo->invflags&IPT_INV_VIA_IN ?" (INV)":"");
106
		return false;
L
Linus Torvalds 已提交
107 108
	}

109
	ret = ifname_compare_aligned(outdev, ipinfo->outiface, ipinfo->outiface_mask);
L
Linus Torvalds 已提交
110 111 112 113 114

	if (FWINV(ret != 0, IPT_INV_VIA_OUT)) {
		dprintf("VIA out mismatch (%s vs %s).%s\n",
			outdev, ipinfo->outiface,
			ipinfo->invflags&IPT_INV_VIA_OUT ?" (INV)":"");
115
		return false;
L
Linus Torvalds 已提交
116 117 118
	}

	/* Check specific protocol */
119 120
	if (ipinfo->proto &&
	    FWINV(ip->protocol != ipinfo->proto, IPT_INV_PROTO)) {
L
Linus Torvalds 已提交
121 122 123
		dprintf("Packet protocol %hi does not match %hi.%s\n",
			ip->protocol, ipinfo->proto,
			ipinfo->invflags&IPT_INV_PROTO ? " (INV)":"");
124
		return false;
L
Linus Torvalds 已提交
125 126 127 128 129 130 131
	}

	/* If we have a fragment rule but the packet is not a fragment
	 * then we return zero */
	if (FWINV((ipinfo->flags&IPT_F_FRAG) && !isfrag, IPT_INV_FRAG)) {
		dprintf("Fragment rule but not fragment.%s\n",
			ipinfo->invflags & IPT_INV_FRAG ? " (INV)" : "");
132
		return false;
L
Linus Torvalds 已提交
133 134
	}

135
	return true;
L
Linus Torvalds 已提交
136 137
}

138
static bool
L
Linus Torvalds 已提交
139 140 141 142 143
ip_checkentry(const struct ipt_ip *ip)
{
	if (ip->flags & ~IPT_F_MASK) {
		duprintf("Unknown flag bits set: %08X\n",
			 ip->flags & ~IPT_F_MASK);
144
		return false;
L
Linus Torvalds 已提交
145 146 147 148
	}
	if (ip->invflags & ~IPT_INV_MASK) {
		duprintf("Unknown invflag bits set: %08X\n",
			 ip->invflags & ~IPT_INV_MASK);
149
		return false;
L
Linus Torvalds 已提交
150
	}
151
	return true;
L
Linus Torvalds 已提交
152 153 154
}

static unsigned int
155
ipt_error(struct sk_buff *skb, const struct xt_action_param *par)
L
Linus Torvalds 已提交
156
{
157
	net_info_ratelimited("error: `%s'\n", (const char *)par->targinfo);
L
Linus Torvalds 已提交
158 159 160 161

	return NF_DROP;
}

162
/* Performance critical */
L
Linus Torvalds 已提交
163
static inline struct ipt_entry *
164
get_entry(const void *base, unsigned int offset)
L
Linus Torvalds 已提交
165 166 167 168
{
	return (struct ipt_entry *)(base + offset);
}

169
/* All zeroes == unconditional rule. */
170
/* Mildly perf critical (only if packet tracing is on) */
171
static inline bool unconditional(const struct ipt_ip *ip)
172
{
173
	static const struct ipt_ip uncond;
174

175
	return memcmp(ip, &uncond, sizeof(uncond)) == 0;
176
#undef FWINV
177 178
}

179
/* for const-correctness */
180
static inline const struct xt_entry_target *
181 182 183 184 185
ipt_get_target_c(const struct ipt_entry *e)
{
	return ipt_get_target((struct ipt_entry *)e);
}

186
#if IS_ENABLED(CONFIG_NETFILTER_XT_TARGET_TRACE)
187
static const char *const hooknames[] = {
188 189
	[NF_INET_PRE_ROUTING]		= "PREROUTING",
	[NF_INET_LOCAL_IN]		= "INPUT",
190
	[NF_INET_FORWARD]		= "FORWARD",
191 192
	[NF_INET_LOCAL_OUT]		= "OUTPUT",
	[NF_INET_POST_ROUTING]		= "POSTROUTING",
193 194 195 196 197 198 199 200
};

enum nf_ip_trace_comments {
	NF_IP_TRACE_COMMENT_RULE,
	NF_IP_TRACE_COMMENT_RETURN,
	NF_IP_TRACE_COMMENT_POLICY,
};

201
static const char *const comments[] = {
202 203 204 205 206 207 208 209 210 211 212 213 214 215 216
	[NF_IP_TRACE_COMMENT_RULE]	= "rule",
	[NF_IP_TRACE_COMMENT_RETURN]	= "return",
	[NF_IP_TRACE_COMMENT_POLICY]	= "policy",
};

static struct nf_loginfo trace_loginfo = {
	.type = NF_LOG_TYPE_LOG,
	.u = {
		.log = {
			.level = 4,
			.logflags = NF_LOG_MASK,
		},
	},
};

217
/* Mildly perf critical (only if packet tracing is on) */
218
static inline int
219
get_chainname_rulenum(const struct ipt_entry *s, const struct ipt_entry *e,
220 221
		      const char *hookname, const char **chainname,
		      const char **comment, unsigned int *rulenum)
222
{
223
	const struct xt_standard_target *t = (void *)ipt_get_target_c(s);
224

225
	if (strcmp(t->target.u.kernel.target->name, XT_ERROR_TARGET) == 0) {
226 227 228 229 230 231
		/* Head of user chain: ERROR target with chainname */
		*chainname = t->target.data;
		(*rulenum) = 0;
	} else if (s == e) {
		(*rulenum)++;

232 233
		if (s->target_offset == sizeof(struct ipt_entry) &&
		    strcmp(t->target.u.kernel.target->name,
234
			   XT_STANDARD_TARGET) == 0 &&
235 236
		   t->verdict < 0 &&
		   unconditional(&s->ip)) {
237 238
			/* Tail of chains: STANDARD target (return/policy) */
			*comment = *chainname == hookname
239 240
				? comments[NF_IP_TRACE_COMMENT_POLICY]
				: comments[NF_IP_TRACE_COMMENT_RETURN];
241 242 243 244 245 246 247 248
		}
		return 1;
	} else
		(*rulenum)++;

	return 0;
}

249
static void trace_packet(const struct sk_buff *skb,
250 251 252
			 unsigned int hook,
			 const struct net_device *in,
			 const struct net_device *out,
253
			 const char *tablename,
254 255
			 const struct xt_table_info *private,
			 const struct ipt_entry *e)
256
{
257
	const struct ipt_entry *root;
258
	const char *hookname, *chainname, *comment;
259
	const struct ipt_entry *iter;
260
	unsigned int rulenum = 0;
261
	struct net *net = dev_net(in ? in : out);
262

263
	root = get_entry(private->entries, private->hook_entry[hook]);
264

265 266
	hookname = chainname = hooknames[hook];
	comment = comments[NF_IP_TRACE_COMMENT_RULE];
267

268 269 270 271
	xt_entry_foreach(iter, root, private->size - private->hook_entry[hook])
		if (get_chainname_rulenum(iter, e, hookname,
		    &chainname, &comment, &rulenum) != 0)
			break;
272

273 274 275
	nf_log_trace(net, AF_INET, hook, skb, in, out, &trace_loginfo,
		     "TRACE: %s:%s:%s:%u ",
		     tablename, chainname, comment, rulenum);
276 277 278
}
#endif

279 280 281 282 283 284
static inline __pure
struct ipt_entry *ipt_next_entry(const struct ipt_entry *entry)
{
	return (void *)entry + entry->next_offset;
}

L
Linus Torvalds 已提交
285 286
/* Returns one of the generic firewall policies, like NF_ACCEPT. */
unsigned int
287
ipt_do_table(struct sk_buff *skb,
L
Linus Torvalds 已提交
288
	     unsigned int hook,
289
	     const struct nf_hook_state *state,
290
	     struct xt_table *table)
L
Linus Torvalds 已提交
291 292
{
	static const char nulldevname[IFNAMSIZ] __attribute__((aligned(sizeof(long))));
293
	const struct iphdr *ip;
L
Linus Torvalds 已提交
294 295 296
	/* Initializing verdict to NF_DROP keeps gcc happy. */
	unsigned int verdict = NF_DROP;
	const char *indev, *outdev;
297
	const void *table_base;
298 299
	struct ipt_entry *e, **jumpstack;
	unsigned int *stackptr, origptr, cpu;
300
	const struct xt_table_info *private;
301
	struct xt_action_param acpar;
302
	unsigned int addend;
L
Linus Torvalds 已提交
303 304

	/* Initialization */
305
	ip = ip_hdr(skb);
306 307
	indev = state->in ? state->in->name : nulldevname;
	outdev = state->out ? state->out->name : nulldevname;
L
Linus Torvalds 已提交
308 309 310 311 312 313
	/* We handle fragments by dealing with the first fragment as
	 * if it was a normal packet.  All other fragments are treated
	 * normally, except that they will NEVER match rules that ask
	 * things we don't know, ie. tcp syn flag or ports).  If the
	 * rule is also a fragment-specific rule, non-fragments won't
	 * match it. */
314 315
	acpar.fragoff = ntohs(ip->frag_off) & IP_OFFSET;
	acpar.thoff   = ip_hdrlen(skb);
316
	acpar.hotdrop = false;
317 318
	acpar.in      = state->in;
	acpar.out     = state->out;
319 320
	acpar.family  = NFPROTO_IPV4;
	acpar.hooknum = hook;
L
Linus Torvalds 已提交
321 322

	IP_NF_ASSERT(table->valid_hooks & (1 << hook));
323 324
	local_bh_disable();
	addend = xt_write_recseq_begin();
325
	private = table->private;
326
	cpu        = smp_processor_id();
327 328 329 330 331
	/*
	 * Ensure we load private-> members after we've fetched the base
	 * pointer.
	 */
	smp_read_barrier_depends();
332
	table_base = private->entries;
333
	jumpstack  = (struct ipt_entry **)private->jumpstack[cpu];
334
	stackptr   = per_cpu_ptr(private->stackptr, cpu);
335
	origptr    = *stackptr;
336

337
	e = get_entry(table_base, private->hook_entry[hook]);
L
Linus Torvalds 已提交
338

339
	pr_debug("Entering %s(hook %u); sp at %u (UF %p)\n",
340 341
		 table->name, hook, origptr,
		 get_entry(table_base, private->underflow[hook]));
L
Linus Torvalds 已提交
342 343

	do {
344
		const struct xt_entry_target *t;
345
		const struct xt_entry_match *ematch;
346
		struct xt_counters *counter;
347

L
Linus Torvalds 已提交
348
		IP_NF_ASSERT(e);
349
		if (!ip_packet_match(ip, indev, outdev,
350
		    &e->ip, acpar.fragoff)) {
351
 no_match:
352 353 354
			e = ipt_next_entry(e);
			continue;
		}
L
Linus Torvalds 已提交
355

356
		xt_ematch_foreach(ematch, e) {
357 358 359
			acpar.match     = ematch->u.kernel.match;
			acpar.matchinfo = ematch->data;
			if (!acpar.match->match(skb, &acpar))
360
				goto no_match;
361
		}
362

363 364
		counter = xt_get_this_cpu_counter(&e->counters);
		ADD_COUNTER(*counter, skb->len, 1);
L
Linus Torvalds 已提交
365

366 367
		t = ipt_get_target(e);
		IP_NF_ASSERT(t->u.kernel.target);
368

369
#if IS_ENABLED(CONFIG_NETFILTER_XT_TARGET_TRACE)
370 371
		/* The packet is traced: log it */
		if (unlikely(skb->nf_trace))
372
			trace_packet(skb, hook, state->in, state->out,
373
				     table->name, private, e);
374
#endif
375 376 377 378
		/* Standard target? */
		if (!t->u.kernel.target->target) {
			int v;

379
			v = ((struct xt_standard_target *)t)->verdict;
380 381
			if (v < 0) {
				/* Pop from stack? */
382
				if (v != XT_RETURN) {
383
					verdict = (unsigned int)(-v) - 1;
384
					break;
L
Linus Torvalds 已提交
385
				}
386
				if (*stackptr <= origptr) {
387 388
					e = get_entry(table_base,
					    private->underflow[hook]);
389
					pr_debug("Underflow (this is normal) "
390 391 392
						 "to %p\n", e);
				} else {
					e = jumpstack[--*stackptr];
393
					pr_debug("Pulled %p out from pos %u\n",
394 395 396
						 e, *stackptr);
					e = ipt_next_entry(e);
				}
397 398
				continue;
			}
399 400
			if (table_base + v != ipt_next_entry(e) &&
			    !(e->ip.flags & IPT_F_GOTO)) {
401 402 403 404 405
				if (*stackptr >= private->stacksize) {
					verdict = NF_DROP;
					break;
				}
				jumpstack[(*stackptr)++] = e;
406
				pr_debug("Pushed %p into pos %u\n",
407
					 e, *stackptr - 1);
408
			}
L
Linus Torvalds 已提交
409

410
			e = get_entry(table_base, v);
411 412 413
			continue;
		}

414 415
		acpar.target   = t->u.kernel.target;
		acpar.targinfo = t->data;
416

417
		verdict = t->u.kernel.target->target(skb, &acpar);
418 419
		/* Target might have changed stuff. */
		ip = ip_hdr(skb);
420
		if (verdict == XT_CONTINUE)
421 422 423 424
			e = ipt_next_entry(e);
		else
			/* Verdict */
			break;
425
	} while (!acpar.hotdrop);
426
	pr_debug("Exiting %s; resetting sp from %u to %u\n",
427 428
		 __func__, *stackptr, origptr);
	*stackptr = origptr;
429 430 431
 	xt_write_recseq_end(addend);
 	local_bh_enable();

L
Linus Torvalds 已提交
432 433 434
#ifdef DEBUG_ALLOW_ALL
	return NF_ACCEPT;
#else
435
	if (acpar.hotdrop)
L
Linus Torvalds 已提交
436 437 438 439 440 441
		return NF_DROP;
	else return verdict;
#endif
}

/* Figures out from what hook each rule can be called: returns 0 if
442 443 444 445
 * there are loops.  Puts hook bitmask in comefrom.
 *
 * Keeps track of largest call depth seen and stores it in newinfo->stacksize.
 */
L
Linus Torvalds 已提交
446
static int
447
mark_source_chains(struct xt_table_info *newinfo,
448
		   unsigned int valid_hooks, void *entry0)
L
Linus Torvalds 已提交
449
{
450
	unsigned int calldepth, max_calldepth = 0;
L
Linus Torvalds 已提交
451 452 453 454
	unsigned int hook;

	/* No recursion; use packet counter to save back ptrs (reset
	   to 0 as we leave), and comefrom to save source hook bitmask */
455
	for (hook = 0; hook < NF_INET_NUMHOOKS; hook++) {
L
Linus Torvalds 已提交
456
		unsigned int pos = newinfo->hook_entry[hook];
457
		struct ipt_entry *e = (struct ipt_entry *)(entry0 + pos);
L
Linus Torvalds 已提交
458 459 460 461 462 463

		if (!(valid_hooks & (1 << hook)))
			continue;

		/* Set initial back pointer. */
		e->counters.pcnt = pos;
464
		calldepth = 0;
L
Linus Torvalds 已提交
465 466

		for (;;) {
467
			const struct xt_standard_target *t
468
				= (void *)ipt_get_target_c(e);
469
			int visited = e->comefrom & (1 << hook);
L
Linus Torvalds 已提交
470

471
			if (e->comefrom & (1 << NF_INET_NUMHOOKS)) {
472
				pr_err("iptables: loop hook %u pos %u %08X.\n",
L
Linus Torvalds 已提交
473 474 475
				       hook, pos, e->comefrom);
				return 0;
			}
476
			e->comefrom |= ((1 << hook) | (1 << NF_INET_NUMHOOKS));
L
Linus Torvalds 已提交
477 478

			/* Unconditional return/END. */
479 480
			if ((e->target_offset == sizeof(struct ipt_entry) &&
			     (strcmp(t->target.u.user.name,
481
				     XT_STANDARD_TARGET) == 0) &&
482 483
			     t->verdict < 0 && unconditional(&e->ip)) ||
			    visited) {
L
Linus Torvalds 已提交
484 485
				unsigned int oldpos, size;

486
				if ((strcmp(t->target.u.user.name,
487
			    		    XT_STANDARD_TARGET) == 0) &&
488
				    t->verdict < -NF_MAX_VERDICT - 1) {
489 490 491 492 493 494
					duprintf("mark_source_chains: bad "
						"negative verdict (%i)\n",
								t->verdict);
					return 0;
				}

L
Linus Torvalds 已提交
495 496 497
				/* Return: backtrack through the last
				   big jump. */
				do {
498
					e->comefrom ^= (1<<NF_INET_NUMHOOKS);
L
Linus Torvalds 已提交
499 500
#ifdef DEBUG_IP_FIREWALL_USER
					if (e->comefrom
501
					    & (1 << NF_INET_NUMHOOKS)) {
L
Linus Torvalds 已提交
502 503 504 505 506 507 508 509 510 511 512 513 514 515 516
						duprintf("Back unset "
							 "on hook %u "
							 "rule %u\n",
							 hook, pos);
					}
#endif
					oldpos = pos;
					pos = e->counters.pcnt;
					e->counters.pcnt = 0;

					/* We're at the start. */
					if (pos == oldpos)
						goto next;

					e = (struct ipt_entry *)
517
						(entry0 + pos);
L
Linus Torvalds 已提交
518 519 520 521 522
				} while (oldpos == pos + e->next_offset);

				/* Move along one */
				size = e->next_offset;
				e = (struct ipt_entry *)
523
					(entry0 + pos + size);
L
Linus Torvalds 已提交
524 525
				e->counters.pcnt = pos;
				pos += size;
526 527 528
				WARN_ON_ONCE(calldepth == 0);
				if (calldepth > 0)
					--calldepth;
L
Linus Torvalds 已提交
529 530 531 532
			} else {
				int newpos = t->verdict;

				if (strcmp(t->target.u.user.name,
533
					   XT_STANDARD_TARGET) == 0 &&
534
				    newpos >= 0) {
535 536 537 538 539 540 541
					if (newpos > newinfo->size -
						sizeof(struct ipt_entry)) {
						duprintf("mark_source_chains: "
							"bad verdict (%i)\n",
								newpos);
						return 0;
					}
542 543 544 545 546
					if (entry0 + newpos != ipt_next_entry(e) &&
					    !(e->ip.flags & IPT_F_GOTO) &&
					    ++calldepth > max_calldepth)
						max_calldepth = calldepth;

L
Linus Torvalds 已提交
547
					/* This a jump; chase it. */
548 549
					duprintf("Jump rule %u -> %u, calldepth %d\n",
						 pos, newpos, calldepth);
L
Linus Torvalds 已提交
550 551 552 553 554
				} else {
					/* ... this is a fallthru */
					newpos = pos + e->next_offset;
				}
				e = (struct ipt_entry *)
555
					(entry0 + newpos);
L
Linus Torvalds 已提交
556 557 558 559 560 561 562
				e->counters.pcnt = pos;
				pos = newpos;
			}
		}
		next:
		duprintf("Finished chain %u\n", hook);
	}
563
	newinfo->stacksize = max_calldepth;
L
Linus Torvalds 已提交
564 565 566
	return 1;
}

567
static void cleanup_match(struct xt_entry_match *m, struct net *net)
L
Linus Torvalds 已提交
568
{
569 570
	struct xt_mtdtor_param par;

571
	par.net       = net;
572 573
	par.match     = m->u.kernel.match;
	par.matchinfo = m->data;
574
	par.family    = NFPROTO_IPV4;
575 576 577
	if (par.match->destroy != NULL)
		par.match->destroy(&par);
	module_put(par.match->me);
L
Linus Torvalds 已提交
578 579
}

580
static int
581
check_entry(const struct ipt_entry *e, const char *name)
582
{
583
	const struct xt_entry_target *t;
584 585

	if (!ip_checkentry(&e->ip)) {
586
		duprintf("ip check failed %p %s.\n", e, name);
587 588 589
		return -EINVAL;
	}

590
	if (e->target_offset + sizeof(struct xt_entry_target) >
591
	    e->next_offset)
592 593
		return -EINVAL;

594
	t = ipt_get_target_c(e);
595 596 597 598 599 600
	if (e->target_offset + t->u.target_size > e->next_offset)
		return -EINVAL;

	return 0;
}

601
static int
602
check_match(struct xt_entry_match *m, struct xt_mtchk_param *par)
603
{
604
	const struct ipt_ip *ip = par->entryinfo;
605 606
	int ret;

607 608 609
	par->match     = m->u.kernel.match;
	par->matchinfo = m->data;

610
	ret = xt_check_match(par, m->u.match_size - sizeof(*m),
611
	      ip->proto, ip->invflags & IPT_INV_PROTO);
612
	if (ret < 0) {
613
		duprintf("check failed for `%s'.\n", par->match->name);
614
		return ret;
615
	}
616
	return 0;
617 618
}

619
static int
620
find_check_match(struct xt_entry_match *m, struct xt_mtchk_param *par)
L
Linus Torvalds 已提交
621
{
622
	struct xt_match *match;
623
	int ret;
L
Linus Torvalds 已提交
624

625 626 627
	match = xt_request_find_match(NFPROTO_IPV4, m->u.user.name,
				      m->u.user.revision);
	if (IS_ERR(match)) {
628
		duprintf("find_check_match: `%s' not found\n", m->u.user.name);
629
		return PTR_ERR(match);
L
Linus Torvalds 已提交
630 631 632
	}
	m->u.kernel.match = match;

633
	ret = check_match(m, par);
634 635 636
	if (ret)
		goto err;

L
Linus Torvalds 已提交
637
	return 0;
638 639 640
err:
	module_put(m->u.kernel.match->me);
	return ret;
L
Linus Torvalds 已提交
641 642
}

643
static int check_target(struct ipt_entry *e, struct net *net, const char *name)
644
{
645
	struct xt_entry_target *t = ipt_get_target(e);
646
	struct xt_tgchk_param par = {
647
		.net       = net,
648 649 650 651 652
		.table     = name,
		.entryinfo = e,
		.target    = t->u.kernel.target,
		.targinfo  = t->data,
		.hook_mask = e->comefrom,
653
		.family    = NFPROTO_IPV4,
654
	};
655
	int ret;
656

657
	ret = xt_check_target(&par, t->u.target_size - sizeof(*t),
658
	      e->ip.proto, e->ip.invflags & IPT_INV_PROTO);
659
	if (ret < 0) {
660
		duprintf("check failed for `%s'.\n",
661
			 t->u.kernel.target->name);
662
		return ret;
663
	}
664
	return 0;
665
}
L
Linus Torvalds 已提交
666

667
static int
668
find_check_entry(struct ipt_entry *e, struct net *net, const char *name,
669
		 unsigned int size)
L
Linus Torvalds 已提交
670
{
671
	struct xt_entry_target *t;
672
	struct xt_target *target;
L
Linus Torvalds 已提交
673 674
	int ret;
	unsigned int j;
675
	struct xt_mtchk_param mtpar;
676
	struct xt_entry_match *ematch;
L
Linus Torvalds 已提交
677

678 679 680
	ret = check_entry(e, name);
	if (ret)
		return ret;
681

682 683 684 685
	e->counters.pcnt = xt_percpu_counter_alloc();
	if (IS_ERR_VALUE(e->counters.pcnt))
		return -ENOMEM;

L
Linus Torvalds 已提交
686
	j = 0;
687
	mtpar.net	= net;
688 689 690
	mtpar.table     = name;
	mtpar.entryinfo = &e->ip;
	mtpar.hook_mask = e->comefrom;
691
	mtpar.family    = NFPROTO_IPV4;
692
	xt_ematch_foreach(ematch, e) {
693
		ret = find_check_match(ematch, &mtpar);
694
		if (ret != 0)
695 696
			goto cleanup_matches;
		++j;
697
	}
L
Linus Torvalds 已提交
698 699

	t = ipt_get_target(e);
700 701 702
	target = xt_request_find_target(NFPROTO_IPV4, t->u.user.name,
					t->u.user.revision);
	if (IS_ERR(target)) {
703
		duprintf("find_check_entry: `%s' not found\n", t->u.user.name);
704
		ret = PTR_ERR(target);
L
Linus Torvalds 已提交
705 706 707 708
		goto cleanup_matches;
	}
	t->u.kernel.target = target;

709
	ret = check_target(e, net, name);
710 711
	if (ret)
		goto err;
712

L
Linus Torvalds 已提交
713
	return 0;
714 715
 err:
	module_put(t->u.kernel.target->me);
L
Linus Torvalds 已提交
716
 cleanup_matches:
717 718
	xt_ematch_foreach(ematch, e) {
		if (j-- == 0)
719
			break;
720 721
		cleanup_match(ematch, net);
	}
722 723 724

	xt_percpu_counter_free(e->counters.pcnt);

L
Linus Torvalds 已提交
725 726 727
	return ret;
}

728
static bool check_underflow(const struct ipt_entry *e)
729
{
730
	const struct xt_entry_target *t;
731 732 733 734
	unsigned int verdict;

	if (!unconditional(&e->ip))
		return false;
735
	t = ipt_get_target_c(e);
736 737
	if (strcmp(t->u.user.name, XT_STANDARD_TARGET) != 0)
		return false;
738
	verdict = ((struct xt_standard_target *)t)->verdict;
739 740 741 742
	verdict = -verdict - 1;
	return verdict == NF_DROP || verdict == NF_ACCEPT;
}

743
static int
L
Linus Torvalds 已提交
744
check_entry_size_and_hooks(struct ipt_entry *e,
745
			   struct xt_table_info *newinfo,
746 747
			   const unsigned char *base,
			   const unsigned char *limit,
L
Linus Torvalds 已提交
748 749
			   const unsigned int *hook_entries,
			   const unsigned int *underflows,
750
			   unsigned int valid_hooks)
L
Linus Torvalds 已提交
751 752 753
{
	unsigned int h;

754 755
	if ((unsigned long)e % __alignof__(struct ipt_entry) != 0 ||
	    (unsigned char *)e + sizeof(struct ipt_entry) >= limit) {
L
Linus Torvalds 已提交
756 757 758 759 760
		duprintf("Bad offset %p\n", e);
		return -EINVAL;
	}

	if (e->next_offset
761
	    < sizeof(struct ipt_entry) + sizeof(struct xt_entry_target)) {
L
Linus Torvalds 已提交
762 763 764 765 766 767
		duprintf("checking: element %p size %u\n",
			 e, e->next_offset);
		return -EINVAL;
	}

	/* Check hooks & underflows */
768
	for (h = 0; h < NF_INET_NUMHOOKS; h++) {
769 770
		if (!(valid_hooks & (1 << h)))
			continue;
L
Linus Torvalds 已提交
771 772
		if ((unsigned char *)e - base == hook_entries[h])
			newinfo->hook_entry[h] = hook_entries[h];
773
		if ((unsigned char *)e - base == underflows[h]) {
774 775 776 777
			if (!check_underflow(e)) {
				pr_err("Underflows must be unconditional and "
				       "use the STANDARD target with "
				       "ACCEPT/DROP\n");
778 779
				return -EINVAL;
			}
L
Linus Torvalds 已提交
780
			newinfo->underflow[h] = underflows[h];
781
		}
L
Linus Torvalds 已提交
782 783 784
	}

	/* Clear counters and comefrom */
785
	e->counters = ((struct xt_counters) { 0, 0 });
L
Linus Torvalds 已提交
786 787 788 789
	e->comefrom = 0;
	return 0;
}

790 791
static void
cleanup_entry(struct ipt_entry *e, struct net *net)
L
Linus Torvalds 已提交
792
{
793
	struct xt_tgdtor_param par;
794
	struct xt_entry_target *t;
795
	struct xt_entry_match *ematch;
L
Linus Torvalds 已提交
796 797

	/* Cleanup all matches */
798
	xt_ematch_foreach(ematch, e)
799
		cleanup_match(ematch, net);
L
Linus Torvalds 已提交
800
	t = ipt_get_target(e);
801

802
	par.net      = net;
803 804
	par.target   = t->u.kernel.target;
	par.targinfo = t->data;
805
	par.family   = NFPROTO_IPV4;
806 807 808
	if (par.target->destroy != NULL)
		par.target->destroy(&par);
	module_put(par.target->me);
809
	xt_percpu_counter_free(e->counters.pcnt);
L
Linus Torvalds 已提交
810 811 812 813 814
}

/* Checks and translates the user-supplied table segment (held in
   newinfo) */
static int
815 816
translate_table(struct net *net, struct xt_table_info *newinfo, void *entry0,
                const struct ipt_replace *repl)
L
Linus Torvalds 已提交
817
{
818
	struct ipt_entry *iter;
L
Linus Torvalds 已提交
819
	unsigned int i;
820
	int ret = 0;
L
Linus Torvalds 已提交
821

822 823
	newinfo->size = repl->size;
	newinfo->number = repl->num_entries;
L
Linus Torvalds 已提交
824 825

	/* Init all hooks to impossible value. */
826
	for (i = 0; i < NF_INET_NUMHOOKS; i++) {
L
Linus Torvalds 已提交
827 828 829 830 831 832 833
		newinfo->hook_entry[i] = 0xFFFFFFFF;
		newinfo->underflow[i] = 0xFFFFFFFF;
	}

	duprintf("translate_table: size %u\n", newinfo->size);
	i = 0;
	/* Walk through entries, checking offsets. */
834 835
	xt_entry_foreach(iter, entry0, newinfo->size) {
		ret = check_entry_size_and_hooks(iter, newinfo, entry0,
836 837 838 839
						 entry0 + repl->size,
						 repl->hook_entry,
						 repl->underflow,
						 repl->valid_hooks);
840
		if (ret != 0)
841 842
			return ret;
		++i;
843
	}
L
Linus Torvalds 已提交
844

845
	if (i != repl->num_entries) {
L
Linus Torvalds 已提交
846
		duprintf("translate_table: %u not %u entries\n",
847
			 i, repl->num_entries);
L
Linus Torvalds 已提交
848 849 850 851
		return -EINVAL;
	}

	/* Check hooks all assigned */
852
	for (i = 0; i < NF_INET_NUMHOOKS; i++) {
L
Linus Torvalds 已提交
853
		/* Only hooks which are valid */
854
		if (!(repl->valid_hooks & (1 << i)))
L
Linus Torvalds 已提交
855 856 857
			continue;
		if (newinfo->hook_entry[i] == 0xFFFFFFFF) {
			duprintf("Invalid hook entry %u %u\n",
858
				 i, repl->hook_entry[i]);
L
Linus Torvalds 已提交
859 860 861 862
			return -EINVAL;
		}
		if (newinfo->underflow[i] == 0xFFFFFFFF) {
			duprintf("Invalid underflow %u %u\n",
863
				 i, repl->underflow[i]);
L
Linus Torvalds 已提交
864 865 866 867
			return -EINVAL;
		}
	}

868
	if (!mark_source_chains(newinfo, repl->valid_hooks, entry0))
869 870
		return -ELOOP;

L
Linus Torvalds 已提交
871 872
	/* Finally, each sanity check must pass */
	i = 0;
873
	xt_entry_foreach(iter, entry0, newinfo->size) {
874
		ret = find_check_entry(iter, net, repl->name, repl->size);
875 876
		if (ret != 0)
			break;
877
		++i;
878
	}
L
Linus Torvalds 已提交
879

880
	if (ret != 0) {
881 882
		xt_entry_foreach(iter, entry0, newinfo->size) {
			if (i-- == 0)
883
				break;
884 885
			cleanup_entry(iter, net);
		}
886 887
		return ret;
	}
L
Linus Torvalds 已提交
888 889 890 891 892

	return ret;
}

static void
893 894
get_counters(const struct xt_table_info *t,
	     struct xt_counters counters[])
L
Linus Torvalds 已提交
895
{
896
	struct ipt_entry *iter;
L
Linus Torvalds 已提交
897 898 899
	unsigned int cpu;
	unsigned int i;

900
	for_each_possible_cpu(cpu) {
901
		seqcount_t *s = &per_cpu(xt_recseq, cpu);
902

L
Linus Torvalds 已提交
903
		i = 0;
904
		xt_entry_foreach(iter, t->entries, t->size) {
905
			struct xt_counters *tmp;
906 907 908
			u64 bcnt, pcnt;
			unsigned int start;

909
			tmp = xt_get_per_cpu_counter(&iter->counters, cpu);
910
			do {
911
				start = read_seqcount_begin(s);
912 913
				bcnt = tmp->bcnt;
				pcnt = tmp->pcnt;
914
			} while (read_seqcount_retry(s, start));
915 916

			ADD_COUNTER(counters[i], bcnt, pcnt);
917 918
			++i; /* macro does multi eval of i */
		}
L
Linus Torvalds 已提交
919
	}
920 921
}

922
static struct xt_counters *alloc_counters(const struct xt_table *table)
L
Linus Torvalds 已提交
923
{
924
	unsigned int countersize;
925
	struct xt_counters *counters;
926
	const struct xt_table_info *private = table->private;
L
Linus Torvalds 已提交
927 928 929 930

	/* We need atomic snapshot of counters: rest doesn't change
	   (other than comefrom, which userspace doesn't care
	   about). */
931
	countersize = sizeof(struct xt_counters) * private->number;
932
	counters = vzalloc(countersize);
L
Linus Torvalds 已提交
933 934

	if (counters == NULL)
935
		return ERR_PTR(-ENOMEM);
936

937
	get_counters(private, counters);
L
Linus Torvalds 已提交
938

939 940 941 942 943
	return counters;
}

static int
copy_entries_to_user(unsigned int total_size,
944
		     const struct xt_table *table,
945 946 947
		     void __user *userptr)
{
	unsigned int off, num;
948
	const struct ipt_entry *e;
949
	struct xt_counters *counters;
950
	const struct xt_table_info *private = table->private;
951
	int ret = 0;
952
	const void *loc_cpu_entry;
953 954 955 956 957

	counters = alloc_counters(table);
	if (IS_ERR(counters))
		return PTR_ERR(counters);

958
	loc_cpu_entry = private->entries;
959
	if (copy_to_user(userptr, loc_cpu_entry, total_size) != 0) {
L
Linus Torvalds 已提交
960 961 962 963 964 965 966 967
		ret = -EFAULT;
		goto free_counters;
	}

	/* FIXME: use iterator macros --RR */
	/* ... then go back and fix counters and names */
	for (off = 0, num = 0; off < total_size; off += e->next_offset, num++){
		unsigned int i;
968 969
		const struct xt_entry_match *m;
		const struct xt_entry_target *t;
L
Linus Torvalds 已提交
970

971
		e = (struct ipt_entry *)(loc_cpu_entry + off);
L
Linus Torvalds 已提交
972 973 974 975 976 977 978 979 980 981 982 983 984 985
		if (copy_to_user(userptr + off
				 + offsetof(struct ipt_entry, counters),
				 &counters[num],
				 sizeof(counters[num])) != 0) {
			ret = -EFAULT;
			goto free_counters;
		}

		for (i = sizeof(struct ipt_entry);
		     i < e->target_offset;
		     i += m->u.match_size) {
			m = (void *)e + i;

			if (copy_to_user(userptr + off + i
986
					 + offsetof(struct xt_entry_match,
L
Linus Torvalds 已提交
987 988 989 990 991 992 993 994 995
						    u.user.name),
					 m->u.kernel.match->name,
					 strlen(m->u.kernel.match->name)+1)
			    != 0) {
				ret = -EFAULT;
				goto free_counters;
			}
		}

996
		t = ipt_get_target_c(e);
L
Linus Torvalds 已提交
997
		if (copy_to_user(userptr + off + e->target_offset
998
				 + offsetof(struct xt_entry_target,
L
Linus Torvalds 已提交
999 1000 1001 1002 1003 1004 1005 1006 1007 1008 1009 1010 1011
					    u.user.name),
				 t->u.kernel.target->name,
				 strlen(t->u.kernel.target->name)+1) != 0) {
			ret = -EFAULT;
			goto free_counters;
		}
	}

 free_counters:
	vfree(counters);
	return ret;
}

1012
#ifdef CONFIG_COMPAT
1013
static void compat_standard_from_user(void *dst, const void *src)
1014
{
1015
	int v = *(compat_int_t *)src;
1016

1017
	if (v > 0)
1018
		v += xt_compat_calc_jump(AF_INET, v);
1019 1020
	memcpy(dst, &v, sizeof(v));
}
1021

1022
static int compat_standard_to_user(void __user *dst, const void *src)
1023
{
1024
	compat_int_t cv = *(int *)src;
1025

1026
	if (cv > 0)
1027
		cv -= xt_compat_calc_jump(AF_INET, cv);
1028
	return copy_to_user(dst, &cv, sizeof(cv)) ? -EFAULT : 0;
1029 1030
}

1031
static int compat_calc_entry(const struct ipt_entry *e,
1032
			     const struct xt_table_info *info,
1033
			     const void *base, struct xt_table_info *newinfo)
1034
{
1035
	const struct xt_entry_match *ematch;
1036
	const struct xt_entry_target *t;
1037
	unsigned int entry_offset;
1038 1039
	int off, i, ret;

1040
	off = sizeof(struct ipt_entry) - sizeof(struct compat_ipt_entry);
1041
	entry_offset = (void *)e - base;
1042
	xt_ematch_foreach(ematch, e)
1043
		off += xt_compat_match_offset(ematch->u.kernel.match);
1044
	t = ipt_get_target_c(e);
1045
	off += xt_compat_target_offset(t->u.kernel.target);
1046
	newinfo->size -= off;
1047
	ret = xt_compat_add_offset(AF_INET, entry_offset, off);
1048 1049 1050
	if (ret)
		return ret;

1051
	for (i = 0; i < NF_INET_NUMHOOKS; i++) {
1052 1053
		if (info->hook_entry[i] &&
		    (e < (struct ipt_entry *)(base + info->hook_entry[i])))
1054
			newinfo->hook_entry[i] -= off;
1055 1056
		if (info->underflow[i] &&
		    (e < (struct ipt_entry *)(base + info->underflow[i])))
1057 1058 1059 1060 1061
			newinfo->underflow[i] -= off;
	}
	return 0;
}

1062
static int compat_table_info(const struct xt_table_info *info,
1063
			     struct xt_table_info *newinfo)
1064
{
1065
	struct ipt_entry *iter;
1066
	const void *loc_cpu_entry;
1067
	int ret;
1068 1069 1070 1071

	if (!newinfo || !info)
		return -EINVAL;

1072
	/* we dont care about newinfo->entries */
1073 1074
	memcpy(newinfo, info, offsetof(struct xt_table_info, entries));
	newinfo->initial_entries = 0;
1075
	loc_cpu_entry = info->entries;
1076
	xt_compat_init_offsets(AF_INET, info->number);
1077 1078 1079
	xt_entry_foreach(iter, loc_cpu_entry, info->size) {
		ret = compat_calc_entry(iter, info, loc_cpu_entry, newinfo);
		if (ret != 0)
1080
			return ret;
1081
	}
1082
	return 0;
1083 1084 1085
}
#endif

1086 1087
static int get_info(struct net *net, void __user *user,
                    const int *len, int compat)
1088
{
1089
	char name[XT_TABLE_MAXNAMELEN];
1090
	struct xt_table *t;
1091 1092 1093
	int ret;

	if (*len != sizeof(struct ipt_getinfo)) {
1094 1095
		duprintf("length %u != %zu\n", *len,
			 sizeof(struct ipt_getinfo));
1096 1097 1098 1099 1100 1101
		return -EINVAL;
	}

	if (copy_from_user(name, user, sizeof(name)) != 0)
		return -EFAULT;

1102
	name[XT_TABLE_MAXNAMELEN-1] = '\0';
1103 1104 1105 1106
#ifdef CONFIG_COMPAT
	if (compat)
		xt_compat_lock(AF_INET);
#endif
1107
	t = try_then_request_module(xt_find_table_lock(net, AF_INET, name),
1108
				    "iptable_%s", name);
1109
	if (!IS_ERR_OR_NULL(t)) {
1110
		struct ipt_getinfo info;
1111
		const struct xt_table_info *private = t->private;
1112
#ifdef CONFIG_COMPAT
1113 1114
		struct xt_table_info tmp;

1115 1116
		if (compat) {
			ret = compat_table_info(private, &tmp);
1117
			xt_compat_flush_offsets(AF_INET);
1118
			private = &tmp;
1119 1120
		}
#endif
1121
		memset(&info, 0, sizeof(info));
1122 1123
		info.valid_hooks = t->valid_hooks;
		memcpy(info.hook_entry, private->hook_entry,
1124
		       sizeof(info.hook_entry));
1125
		memcpy(info.underflow, private->underflow,
1126
		       sizeof(info.underflow));
1127 1128 1129 1130 1131 1132 1133 1134 1135 1136 1137 1138 1139 1140 1141 1142 1143 1144 1145 1146 1147
		info.num_entries = private->number;
		info.size = private->size;
		strcpy(info.name, name);

		if (copy_to_user(user, &info, *len) != 0)
			ret = -EFAULT;
		else
			ret = 0;

		xt_table_unlock(t);
		module_put(t->me);
	} else
		ret = t ? PTR_ERR(t) : -ENOENT;
#ifdef CONFIG_COMPAT
	if (compat)
		xt_compat_unlock(AF_INET);
#endif
	return ret;
}

static int
1148 1149
get_entries(struct net *net, struct ipt_get_entries __user *uptr,
	    const int *len)
1150 1151 1152
{
	int ret;
	struct ipt_get_entries get;
1153
	struct xt_table *t;
1154 1155

	if (*len < sizeof(get)) {
1156
		duprintf("get_entries: %u < %zu\n", *len, sizeof(get));
1157 1158 1159 1160 1161
		return -EINVAL;
	}
	if (copy_from_user(&get, uptr, sizeof(get)) != 0)
		return -EFAULT;
	if (*len != sizeof(struct ipt_get_entries) + get.size) {
1162 1163
		duprintf("get_entries: %u != %zu\n",
			 *len, sizeof(get) + get.size);
1164 1165 1166
		return -EINVAL;
	}

1167
	t = xt_find_table_lock(net, AF_INET, get.name);
1168
	if (!IS_ERR_OR_NULL(t)) {
1169
		const struct xt_table_info *private = t->private;
1170
		duprintf("t->private->number = %u\n", private->number);
1171 1172 1173 1174 1175
		if (get.size == private->size)
			ret = copy_entries_to_user(private->size,
						   t, uptr->entrytable);
		else {
			duprintf("get_entries: I've got %u not %u!\n",
1176
				 private->size, get.size);
1177
			ret = -EAGAIN;
1178 1179 1180 1181 1182 1183 1184 1185 1186 1187
		}
		module_put(t->me);
		xt_table_unlock(t);
	} else
		ret = t ? PTR_ERR(t) : -ENOENT;

	return ret;
}

static int
1188
__do_replace(struct net *net, const char *name, unsigned int valid_hooks,
1189 1190
	     struct xt_table_info *newinfo, unsigned int num_counters,
	     void __user *counters_ptr)
1191 1192
{
	int ret;
1193
	struct xt_table *t;
1194 1195
	struct xt_table_info *oldinfo;
	struct xt_counters *counters;
1196
	struct ipt_entry *iter;
1197 1198

	ret = 0;
1199
	counters = vzalloc(num_counters * sizeof(struct xt_counters));
1200 1201 1202 1203 1204
	if (!counters) {
		ret = -ENOMEM;
		goto out;
	}

1205
	t = try_then_request_module(xt_find_table_lock(net, AF_INET, name),
1206
				    "iptable_%s", name);
1207
	if (IS_ERR_OR_NULL(t)) {
1208 1209 1210 1211 1212 1213 1214 1215 1216 1217 1218 1219 1220 1221 1222 1223 1224 1225 1226 1227 1228 1229 1230 1231 1232 1233
		ret = t ? PTR_ERR(t) : -ENOENT;
		goto free_newinfo_counters_untrans;
	}

	/* You lied! */
	if (valid_hooks != t->valid_hooks) {
		duprintf("Valid hook crap: %08X vs %08X\n",
			 valid_hooks, t->valid_hooks);
		ret = -EINVAL;
		goto put_module;
	}

	oldinfo = xt_replace_table(t, num_counters, newinfo, &ret);
	if (!oldinfo)
		goto put_module;

	/* Update module usage count based on number of rules */
	duprintf("do_replace: oldnum=%u, initnum=%u, newnum=%u\n",
		oldinfo->number, oldinfo->initial_entries, newinfo->number);
	if ((oldinfo->number > oldinfo->initial_entries) ||
	    (newinfo->number <= oldinfo->initial_entries))
		module_put(t->me);
	if ((oldinfo->number > oldinfo->initial_entries) &&
	    (newinfo->number <= oldinfo->initial_entries))
		module_put(t->me);

1234
	/* Get the old counters, and synchronize with replace */
1235
	get_counters(oldinfo, counters);
1236

1237
	/* Decrease module usage counts and free resource */
1238
	xt_entry_foreach(iter, oldinfo->entries, oldinfo->size)
1239
		cleanup_entry(iter, net);
1240

1241 1242
	xt_free_table_info(oldinfo);
	if (copy_to_user(counters_ptr, counters,
1243 1244 1245 1246
			 sizeof(struct xt_counters) * num_counters) != 0) {
		/* Silent error, can't fail, new table is already in place */
		net_warn_ratelimited("iptables: counters copy to user failed while replacing table\n");
	}
1247 1248 1249 1250 1251 1252 1253 1254 1255 1256 1257 1258 1259 1260
	vfree(counters);
	xt_table_unlock(t);
	return ret;

 put_module:
	module_put(t->me);
	xt_table_unlock(t);
 free_newinfo_counters_untrans:
	vfree(counters);
 out:
	return ret;
}

static int
1261
do_replace(struct net *net, const void __user *user, unsigned int len)
1262 1263 1264 1265 1266
{
	int ret;
	struct ipt_replace tmp;
	struct xt_table_info *newinfo;
	void *loc_cpu_entry;
1267
	struct ipt_entry *iter;
1268 1269 1270 1271 1272 1273 1274

	if (copy_from_user(&tmp, user, sizeof(tmp)) != 0)
		return -EFAULT;

	/* overflow check */
	if (tmp.num_counters >= INT_MAX / sizeof(struct xt_counters))
		return -ENOMEM;
1275 1276 1277
	if (tmp.num_counters == 0)
		return -EINVAL;

1278
	tmp.name[sizeof(tmp.name)-1] = 0;
1279 1280 1281 1282 1283

	newinfo = xt_alloc_table_info(tmp.size);
	if (!newinfo)
		return -ENOMEM;

1284
	loc_cpu_entry = newinfo->entries;
1285 1286 1287 1288 1289 1290
	if (copy_from_user(loc_cpu_entry, user + sizeof(tmp),
			   tmp.size) != 0) {
		ret = -EFAULT;
		goto free_newinfo;
	}

1291
	ret = translate_table(net, newinfo, loc_cpu_entry, &tmp);
1292 1293 1294
	if (ret != 0)
		goto free_newinfo;

1295
	duprintf("Translated table\n");
1296

1297
	ret = __do_replace(net, tmp.name, tmp.valid_hooks, newinfo,
1298
			   tmp.num_counters, tmp.counters);
1299 1300 1301 1302 1303
	if (ret)
		goto free_newinfo_untrans;
	return 0;

 free_newinfo_untrans:
1304
	xt_entry_foreach(iter, loc_cpu_entry, newinfo->size)
1305
		cleanup_entry(iter, net);
1306 1307 1308 1309 1310 1311
 free_newinfo:
	xt_free_table_info(newinfo);
	return ret;
}

static int
1312 1313
do_add_counters(struct net *net, const void __user *user,
                unsigned int len, int compat)
1314
{
1315
	unsigned int i;
1316 1317 1318
	struct xt_counters_info tmp;
	struct xt_counters *paddc;
	unsigned int num_counters;
1319
	const char *name;
1320 1321
	int size;
	void *ptmp;
1322
	struct xt_table *t;
1323
	const struct xt_table_info *private;
1324
	int ret = 0;
1325
	struct ipt_entry *iter;
1326
	unsigned int addend;
1327 1328 1329 1330 1331 1332 1333 1334 1335 1336 1337 1338 1339 1340 1341 1342 1343 1344 1345 1346 1347 1348 1349 1350 1351 1352 1353 1354 1355 1356
#ifdef CONFIG_COMPAT
	struct compat_xt_counters_info compat_tmp;

	if (compat) {
		ptmp = &compat_tmp;
		size = sizeof(struct compat_xt_counters_info);
	} else
#endif
	{
		ptmp = &tmp;
		size = sizeof(struct xt_counters_info);
	}

	if (copy_from_user(ptmp, user, size) != 0)
		return -EFAULT;

#ifdef CONFIG_COMPAT
	if (compat) {
		num_counters = compat_tmp.num_counters;
		name = compat_tmp.name;
	} else
#endif
	{
		num_counters = tmp.num_counters;
		name = tmp.name;
	}

	if (len != size + num_counters * sizeof(struct xt_counters))
		return -EINVAL;

E
Eric Dumazet 已提交
1357
	paddc = vmalloc(len - size);
1358 1359 1360 1361 1362 1363 1364 1365
	if (!paddc)
		return -ENOMEM;

	if (copy_from_user(paddc, user + size, len - size) != 0) {
		ret = -EFAULT;
		goto free;
	}

1366
	t = xt_find_table_lock(net, AF_INET, name);
1367
	if (IS_ERR_OR_NULL(t)) {
1368 1369 1370 1371
		ret = t ? PTR_ERR(t) : -ENOENT;
		goto free;
	}

1372
	local_bh_disable();
1373 1374 1375 1376 1377 1378 1379
	private = t->private;
	if (private->number != num_counters) {
		ret = -EINVAL;
		goto unlock_up_free;
	}

	i = 0;
1380
	addend = xt_write_recseq_begin();
1381
	xt_entry_foreach(iter, private->entries, private->size) {
1382 1383 1384 1385
		struct xt_counters *tmp;

		tmp = xt_get_this_cpu_counter(&iter->counters);
		ADD_COUNTER(*tmp, paddc[i].bcnt, paddc[i].pcnt);
1386 1387
		++i;
	}
1388
	xt_write_recseq_end(addend);
1389
 unlock_up_free:
1390
	local_bh_enable();
1391 1392 1393 1394 1395 1396 1397 1398 1399 1400
	xt_table_unlock(t);
	module_put(t->me);
 free:
	vfree(paddc);

	return ret;
}

#ifdef CONFIG_COMPAT
struct compat_ipt_replace {
1401
	char			name[XT_TABLE_MAXNAMELEN];
1402 1403 1404
	u32			valid_hooks;
	u32			num_entries;
	u32			size;
1405 1406
	u32			hook_entry[NF_INET_NUMHOOKS];
	u32			underflow[NF_INET_NUMHOOKS];
1407
	u32			num_counters;
1408
	compat_uptr_t		counters;	/* struct xt_counters * */
1409 1410 1411
	struct compat_ipt_entry	entries[0];
};

1412 1413
static int
compat_copy_entry_to_user(struct ipt_entry *e, void __user **dstptr,
1414
			  unsigned int *size, struct xt_counters *counters,
1415
			  unsigned int i)
1416
{
1417
	struct xt_entry_target *t;
1418 1419 1420
	struct compat_ipt_entry __user *ce;
	u_int16_t target_offset, next_offset;
	compat_uint_t origsize;
1421 1422
	const struct xt_entry_match *ematch;
	int ret = 0;
1423 1424 1425

	origsize = *size;
	ce = (struct compat_ipt_entry __user *)*dstptr;
1426 1427 1428 1429
	if (copy_to_user(ce, e, sizeof(struct ipt_entry)) != 0 ||
	    copy_to_user(&ce->counters, &counters[i],
	    sizeof(counters[i])) != 0)
		return -EFAULT;
1430

1431
	*dstptr += sizeof(struct compat_ipt_entry);
1432 1433
	*size -= sizeof(struct ipt_entry) - sizeof(struct compat_ipt_entry);

1434 1435 1436
	xt_ematch_foreach(ematch, e) {
		ret = xt_compat_match_to_user(ematch, dstptr, size);
		if (ret != 0)
1437
			return ret;
1438
	}
1439 1440
	target_offset = e->target_offset - (origsize - *size);
	t = ipt_get_target(e);
1441
	ret = xt_compat_target_to_user(t, dstptr, size);
1442
	if (ret)
1443
		return ret;
1444
	next_offset = e->next_offset - (origsize - *size);
1445 1446 1447
	if (put_user(target_offset, &ce->target_offset) != 0 ||
	    put_user(next_offset, &ce->next_offset) != 0)
		return -EFAULT;
1448 1449 1450
	return 0;
}

1451
static int
1452
compat_find_calc_match(struct xt_entry_match *m,
1453 1454
		       const char *name,
		       const struct ipt_ip *ip,
1455
		       int *size)
1456
{
1457
	struct xt_match *match;
1458

1459 1460 1461
	match = xt_request_find_match(NFPROTO_IPV4, m->u.user.name,
				      m->u.user.revision);
	if (IS_ERR(match)) {
1462
		duprintf("compat_check_calc_match: `%s' not found\n",
1463
			 m->u.user.name);
1464
		return PTR_ERR(match);
1465 1466
	}
	m->u.kernel.match = match;
1467
	*size += xt_compat_match_offset(match);
1468 1469 1470
	return 0;
}

1471
static void compat_release_entry(struct compat_ipt_entry *e)
1472
{
1473
	struct xt_entry_target *t;
1474
	struct xt_entry_match *ematch;
1475 1476

	/* Cleanup all matches */
1477
	xt_ematch_foreach(ematch, e)
1478
		module_put(ematch->u.kernel.match->me);
1479
	t = compat_ipt_get_target(e);
1480 1481 1482
	module_put(t->u.kernel.target->me);
}

1483
static int
1484
check_compat_entry_size_and_hooks(struct compat_ipt_entry *e,
1485 1486
				  struct xt_table_info *newinfo,
				  unsigned int *size,
1487 1488 1489 1490
				  const unsigned char *base,
				  const unsigned char *limit,
				  const unsigned int *hook_entries,
				  const unsigned int *underflows,
1491
				  const char *name)
1492
{
1493
	struct xt_entry_match *ematch;
1494
	struct xt_entry_target *t;
1495
	struct xt_target *target;
1496
	unsigned int entry_offset;
1497 1498
	unsigned int j;
	int ret, off, h;
1499 1500

	duprintf("check_compat_entry_size_and_hooks %p\n", e);
1501 1502
	if ((unsigned long)e % __alignof__(struct compat_ipt_entry) != 0 ||
	    (unsigned char *)e + sizeof(struct compat_ipt_entry) >= limit) {
1503 1504 1505 1506 1507
		duprintf("Bad offset %p, limit = %p\n", e, limit);
		return -EINVAL;
	}

	if (e->next_offset < sizeof(struct compat_ipt_entry) +
1508
			     sizeof(struct compat_xt_entry_target)) {
1509 1510 1511 1512 1513
		duprintf("checking: element %p size %u\n",
			 e, e->next_offset);
		return -EINVAL;
	}

1514 1515
	/* For purposes of check_entry casting the compat entry is fine */
	ret = check_entry((struct ipt_entry *)e, name);
1516 1517
	if (ret)
		return ret;
1518

1519
	off = sizeof(struct ipt_entry) - sizeof(struct compat_ipt_entry);
1520 1521
	entry_offset = (void *)e - (void *)base;
	j = 0;
1522
	xt_ematch_foreach(ematch, e) {
1523
		ret = compat_find_calc_match(ematch, name, &e->ip, &off);
1524
		if (ret != 0)
1525 1526
			goto release_matches;
		++j;
1527
	}
1528

1529
	t = compat_ipt_get_target(e);
1530 1531 1532
	target = xt_request_find_target(NFPROTO_IPV4, t->u.user.name,
					t->u.user.revision);
	if (IS_ERR(target)) {
1533
		duprintf("check_compat_entry_size_and_hooks: `%s' not found\n",
1534
			 t->u.user.name);
1535
		ret = PTR_ERR(target);
1536
		goto release_matches;
1537 1538 1539
	}
	t->u.kernel.target = target;

1540
	off += xt_compat_target_offset(target);
1541
	*size += off;
1542
	ret = xt_compat_add_offset(AF_INET, entry_offset, off);
1543 1544 1545 1546
	if (ret)
		goto out;

	/* Check hooks & underflows */
1547
	for (h = 0; h < NF_INET_NUMHOOKS; h++) {
1548 1549 1550 1551 1552 1553 1554
		if ((unsigned char *)e - base == hook_entries[h])
			newinfo->hook_entry[h] = hook_entries[h];
		if ((unsigned char *)e - base == underflows[h])
			newinfo->underflow[h] = underflows[h];
	}

	/* Clear counters and comefrom */
1555
	memset(&e->counters, 0, sizeof(e->counters));
1556 1557
	e->comefrom = 0;
	return 0;
1558

1559
out:
1560
	module_put(t->u.kernel.target->me);
1561
release_matches:
1562 1563
	xt_ematch_foreach(ematch, e) {
		if (j-- == 0)
1564
			break;
1565 1566
		module_put(ematch->u.kernel.match->me);
	}
1567 1568 1569
	return ret;
}

1570
static int
1571
compat_copy_entry_from_user(struct compat_ipt_entry *e, void **dstptr,
1572 1573
			    unsigned int *size, const char *name,
			    struct xt_table_info *newinfo, unsigned char *base)
1574
{
1575
	struct xt_entry_target *t;
1576
	struct xt_target *target;
1577 1578
	struct ipt_entry *de;
	unsigned int origsize;
1579
	int ret, h;
1580
	struct xt_entry_match *ematch;
1581 1582 1583 1584 1585

	ret = 0;
	origsize = *size;
	de = (struct ipt_entry *)*dstptr;
	memcpy(de, e, sizeof(struct ipt_entry));
1586
	memcpy(&de->counters, &e->counters, sizeof(e->counters));
1587

1588
	*dstptr += sizeof(struct ipt_entry);
1589 1590
	*size += sizeof(struct ipt_entry) - sizeof(struct compat_ipt_entry);

1591 1592 1593
	xt_ematch_foreach(ematch, e) {
		ret = xt_compat_match_from_user(ematch, dstptr, size);
		if (ret != 0)
1594
			return ret;
1595
	}
1596
	de->target_offset = e->target_offset - (origsize - *size);
1597
	t = compat_ipt_get_target(e);
1598
	target = t->u.kernel.target;
1599
	xt_compat_target_from_user(t, dstptr, size);
1600 1601

	de->next_offset = e->next_offset - (origsize - *size);
1602
	for (h = 0; h < NF_INET_NUMHOOKS; h++) {
1603 1604 1605 1606 1607
		if ((unsigned char *)de - base < newinfo->hook_entry[h])
			newinfo->hook_entry[h] -= origsize - *size;
		if ((unsigned char *)de - base < newinfo->underflow[h])
			newinfo->underflow[h] -= origsize - *size;
	}
1608 1609 1610
	return ret;
}

1611
static int
1612
compat_check_entry(struct ipt_entry *e, struct net *net, const char *name)
1613
{
1614
	struct xt_entry_match *ematch;
1615
	struct xt_mtchk_param mtpar;
1616
	unsigned int j;
1617
	int ret = 0;
1618

1619 1620 1621 1622
	e->counters.pcnt = xt_percpu_counter_alloc();
	if (IS_ERR_VALUE(e->counters.pcnt))
		return -ENOMEM;

1623
	j = 0;
1624
	mtpar.net	= net;
1625 1626 1627
	mtpar.table     = name;
	mtpar.entryinfo = &e->ip;
	mtpar.hook_mask = e->comefrom;
1628
	mtpar.family    = NFPROTO_IPV4;
1629
	xt_ematch_foreach(ematch, e) {
1630
		ret = check_match(ematch, &mtpar);
1631
		if (ret != 0)
1632 1633
			goto cleanup_matches;
		++j;
1634
	}
1635

1636
	ret = check_target(e, net, name);
1637 1638 1639 1640 1641
	if (ret)
		goto cleanup_matches;
	return 0;

 cleanup_matches:
1642 1643
	xt_ematch_foreach(ematch, e) {
		if (j-- == 0)
1644
			break;
1645 1646
		cleanup_match(ematch, net);
	}
1647 1648 1649

	xt_percpu_counter_free(e->counters.pcnt);

1650
	return ret;
1651 1652
}

L
Linus Torvalds 已提交
1653
static int
1654 1655
translate_compat_table(struct net *net,
		       const char *name,
1656 1657 1658 1659 1660 1661 1662
		       unsigned int valid_hooks,
		       struct xt_table_info **pinfo,
		       void **pentry0,
		       unsigned int total_size,
		       unsigned int number,
		       unsigned int *hook_entries,
		       unsigned int *underflows)
L
Linus Torvalds 已提交
1663
{
1664
	unsigned int i, j;
1665 1666
	struct xt_table_info *newinfo, *info;
	void *pos, *entry0, *entry1;
1667 1668
	struct compat_ipt_entry *iter0;
	struct ipt_entry *iter1;
1669
	unsigned int size;
1670
	int ret;
L
Linus Torvalds 已提交
1671

1672 1673 1674 1675 1676 1677
	info = *pinfo;
	entry0 = *pentry0;
	size = total_size;
	info->number = number;

	/* Init all hooks to impossible value. */
1678
	for (i = 0; i < NF_INET_NUMHOOKS; i++) {
1679 1680 1681 1682 1683
		info->hook_entry[i] = 0xFFFFFFFF;
		info->underflow[i] = 0xFFFFFFFF;
	}

	duprintf("translate_compat_table: size %u\n", info->size);
1684
	j = 0;
1685
	xt_compat_lock(AF_INET);
1686
	xt_compat_init_offsets(AF_INET, number);
1687
	/* Walk through entries, checking offsets. */
1688 1689
	xt_entry_foreach(iter0, entry0, total_size) {
		ret = check_compat_entry_size_and_hooks(iter0, info, &size,
1690 1691 1692 1693 1694
							entry0,
							entry0 + total_size,
							hook_entries,
							underflows,
							name);
1695
		if (ret != 0)
1696 1697
			goto out_unlock;
		++j;
1698
	}
1699 1700

	ret = -EINVAL;
1701
	if (j != number) {
1702
		duprintf("translate_compat_table: %u not %u entries\n",
1703
			 j, number);
1704 1705 1706 1707
		goto out_unlock;
	}

	/* Check hooks all assigned */
1708
	for (i = 0; i < NF_INET_NUMHOOKS; i++) {
1709 1710 1711 1712 1713 1714 1715
		/* Only hooks which are valid */
		if (!(valid_hooks & (1 << i)))
			continue;
		if (info->hook_entry[i] == 0xFFFFFFFF) {
			duprintf("Invalid hook entry %u %u\n",
				 i, hook_entries[i]);
			goto out_unlock;
L
Linus Torvalds 已提交
1716
		}
1717 1718 1719 1720 1721 1722 1723 1724 1725 1726 1727 1728 1729
		if (info->underflow[i] == 0xFFFFFFFF) {
			duprintf("Invalid underflow %u %u\n",
				 i, underflows[i]);
			goto out_unlock;
		}
	}

	ret = -ENOMEM;
	newinfo = xt_alloc_table_info(size);
	if (!newinfo)
		goto out_unlock;

	newinfo->number = number;
1730
	for (i = 0; i < NF_INET_NUMHOOKS; i++) {
1731 1732 1733
		newinfo->hook_entry[i] = info->hook_entry[i];
		newinfo->underflow[i] = info->underflow[i];
	}
1734
	entry1 = newinfo->entries;
1735
	pos = entry1;
1736
	size = total_size;
1737
	xt_entry_foreach(iter0, entry0, total_size) {
1738 1739
		ret = compat_copy_entry_from_user(iter0, &pos, &size,
						  name, newinfo, entry1);
1740 1741 1742
		if (ret != 0)
			break;
	}
1743
	xt_compat_flush_offsets(AF_INET);
1744 1745 1746 1747 1748 1749 1750 1751
	xt_compat_unlock(AF_INET);
	if (ret)
		goto free_newinfo;

	ret = -ELOOP;
	if (!mark_source_chains(newinfo, valid_hooks, entry1))
		goto free_newinfo;

1752
	i = 0;
1753
	xt_entry_foreach(iter1, entry1, newinfo->size) {
1754
		ret = compat_check_entry(iter1, net, name);
1755 1756
		if (ret != 0)
			break;
1757
		++i;
1758
	}
1759
	if (ret) {
1760 1761 1762 1763 1764 1765
		/*
		 * The first i matches need cleanup_entry (calls ->destroy)
		 * because they had called ->check already. The other j-i
		 * entries need only release.
		 */
		int skip = i;
1766
		j -= i;
1767 1768 1769
		xt_entry_foreach(iter0, entry0, newinfo->size) {
			if (skip-- > 0)
				continue;
1770
			if (j-- == 0)
1771
				break;
1772
			compat_release_entry(iter0);
1773
		}
1774 1775
		xt_entry_foreach(iter1, entry1, newinfo->size) {
			if (i-- == 0)
1776
				break;
1777 1778
			cleanup_entry(iter1, net);
		}
1779 1780 1781
		xt_free_table_info(newinfo);
		return ret;
	}
1782

1783 1784 1785 1786
	*pinfo = newinfo;
	*pentry0 = entry1;
	xt_free_table_info(info);
	return 0;
L
Linus Torvalds 已提交
1787

1788 1789 1790
free_newinfo:
	xt_free_table_info(newinfo);
out:
1791 1792
	xt_entry_foreach(iter0, entry0, total_size) {
		if (j-- == 0)
1793
			break;
1794 1795
		compat_release_entry(iter0);
	}
L
Linus Torvalds 已提交
1796
	return ret;
1797
out_unlock:
1798
	xt_compat_flush_offsets(AF_INET);
1799 1800
	xt_compat_unlock(AF_INET);
	goto out;
L
Linus Torvalds 已提交
1801 1802 1803
}

static int
1804
compat_do_replace(struct net *net, void __user *user, unsigned int len)
L
Linus Torvalds 已提交
1805 1806
{
	int ret;
1807 1808 1809
	struct compat_ipt_replace tmp;
	struct xt_table_info *newinfo;
	void *loc_cpu_entry;
1810
	struct ipt_entry *iter;
L
Linus Torvalds 已提交
1811 1812 1813 1814

	if (copy_from_user(&tmp, user, sizeof(tmp)) != 0)
		return -EFAULT;

1815
	/* overflow check */
1816
	if (tmp.size >= INT_MAX / num_possible_cpus())
1817 1818 1819
		return -ENOMEM;
	if (tmp.num_counters >= INT_MAX / sizeof(struct xt_counters))
		return -ENOMEM;
1820 1821 1822
	if (tmp.num_counters == 0)
		return -EINVAL;

1823
	tmp.name[sizeof(tmp.name)-1] = 0;
1824

1825
	newinfo = xt_alloc_table_info(tmp.size);
L
Linus Torvalds 已提交
1826 1827 1828
	if (!newinfo)
		return -ENOMEM;

1829
	loc_cpu_entry = newinfo->entries;
1830
	if (copy_from_user(loc_cpu_entry, user + sizeof(tmp),
L
Linus Torvalds 已提交
1831 1832 1833 1834 1835
			   tmp.size) != 0) {
		ret = -EFAULT;
		goto free_newinfo;
	}

1836
	ret = translate_compat_table(net, tmp.name, tmp.valid_hooks,
1837 1838 1839
				     &newinfo, &loc_cpu_entry, tmp.size,
				     tmp.num_entries, tmp.hook_entry,
				     tmp.underflow);
1840
	if (ret != 0)
L
Linus Torvalds 已提交
1841 1842
		goto free_newinfo;

1843
	duprintf("compat_do_replace: Translated table\n");
L
Linus Torvalds 已提交
1844

1845
	ret = __do_replace(net, tmp.name, tmp.valid_hooks, newinfo,
1846
			   tmp.num_counters, compat_ptr(tmp.counters));
1847 1848 1849
	if (ret)
		goto free_newinfo_untrans;
	return 0;
L
Linus Torvalds 已提交
1850

1851
 free_newinfo_untrans:
1852
	xt_entry_foreach(iter, loc_cpu_entry, newinfo->size)
1853
		cleanup_entry(iter, net);
1854 1855 1856 1857
 free_newinfo:
	xt_free_table_info(newinfo);
	return ret;
}
L
Linus Torvalds 已提交
1858

1859 1860
static int
compat_do_ipt_set_ctl(struct sock *sk,	int cmd, void __user *user,
1861
		      unsigned int len)
1862 1863
{
	int ret;
L
Linus Torvalds 已提交
1864

1865
	if (!ns_capable(sock_net(sk)->user_ns, CAP_NET_ADMIN))
1866
		return -EPERM;
L
Linus Torvalds 已提交
1867

1868 1869
	switch (cmd) {
	case IPT_SO_SET_REPLACE:
1870
		ret = compat_do_replace(sock_net(sk), user, len);
1871
		break;
L
Linus Torvalds 已提交
1872

1873
	case IPT_SO_SET_ADD_COUNTERS:
1874
		ret = do_add_counters(sock_net(sk), user, len, 1);
1875 1876 1877 1878 1879 1880
		break;

	default:
		duprintf("do_ipt_set_ctl:  unknown request %i\n", cmd);
		ret = -EINVAL;
	}
L
Linus Torvalds 已提交
1881 1882 1883 1884

	return ret;
}

1885
struct compat_ipt_get_entries {
1886
	char name[XT_TABLE_MAXNAMELEN];
1887 1888 1889
	compat_uint_t size;
	struct compat_ipt_entry entrytable[0];
};
L
Linus Torvalds 已提交
1890

1891 1892 1893
static int
compat_copy_entries_to_user(unsigned int total_size, struct xt_table *table,
			    void __user *userptr)
1894 1895
{
	struct xt_counters *counters;
1896
	const struct xt_table_info *private = table->private;
1897 1898 1899
	void __user *pos;
	unsigned int size;
	int ret = 0;
1900
	unsigned int i = 0;
1901
	struct ipt_entry *iter;
L
Linus Torvalds 已提交
1902

1903 1904 1905 1906 1907 1908
	counters = alloc_counters(table);
	if (IS_ERR(counters))
		return PTR_ERR(counters);

	pos = userptr;
	size = total_size;
1909
	xt_entry_foreach(iter, private->entries, total_size) {
1910
		ret = compat_copy_entry_to_user(iter, &pos,
1911
						&size, counters, i++);
1912 1913 1914
		if (ret != 0)
			break;
	}
1915 1916 1917

	vfree(counters);
	return ret;
L
Linus Torvalds 已提交
1918 1919 1920
}

static int
1921 1922
compat_get_entries(struct net *net, struct compat_ipt_get_entries __user *uptr,
		   int *len)
L
Linus Torvalds 已提交
1923
{
1924 1925
	int ret;
	struct compat_ipt_get_entries get;
1926
	struct xt_table *t;
L
Linus Torvalds 已提交
1927

1928
	if (*len < sizeof(get)) {
1929
		duprintf("compat_get_entries: %u < %zu\n", *len, sizeof(get));
L
Linus Torvalds 已提交
1930
		return -EINVAL;
1931
	}
L
Linus Torvalds 已提交
1932

1933 1934
	if (copy_from_user(&get, uptr, sizeof(get)) != 0)
		return -EFAULT;
L
Linus Torvalds 已提交
1935

1936
	if (*len != sizeof(struct compat_ipt_get_entries) + get.size) {
1937 1938
		duprintf("compat_get_entries: %u != %zu\n",
			 *len, sizeof(get) + get.size);
1939
		return -EINVAL;
L
Linus Torvalds 已提交
1940 1941
	}

1942
	xt_compat_lock(AF_INET);
1943
	t = xt_find_table_lock(net, AF_INET, get.name);
1944
	if (!IS_ERR_OR_NULL(t)) {
1945
		const struct xt_table_info *private = t->private;
1946
		struct xt_table_info info;
1947
		duprintf("t->private->number = %u\n", private->number);
1948 1949 1950
		ret = compat_table_info(private, &info);
		if (!ret && get.size == info.size) {
			ret = compat_copy_entries_to_user(private->size,
1951
							  t, uptr->entrytable);
1952 1953
		} else if (!ret) {
			duprintf("compat_get_entries: I've got %u not %u!\n",
1954
				 private->size, get.size);
1955
			ret = -EAGAIN;
1956
		}
1957
		xt_compat_flush_offsets(AF_INET);
1958 1959 1960
		module_put(t->me);
		xt_table_unlock(t);
	} else
L
Linus Torvalds 已提交
1961 1962
		ret = t ? PTR_ERR(t) : -ENOENT;

1963 1964 1965
	xt_compat_unlock(AF_INET);
	return ret;
}
L
Linus Torvalds 已提交
1966

1967 1968
static int do_ipt_get_ctl(struct sock *, int, void __user *, int *);

1969 1970 1971 1972
static int
compat_do_ipt_get_ctl(struct sock *sk, int cmd, void __user *user, int *len)
{
	int ret;
L
Linus Torvalds 已提交
1973

1974
	if (!ns_capable(sock_net(sk)->user_ns, CAP_NET_ADMIN))
1975 1976
		return -EPERM;

1977 1978
	switch (cmd) {
	case IPT_SO_GET_INFO:
1979
		ret = get_info(sock_net(sk), user, len, 1);
1980 1981
		break;
	case IPT_SO_GET_ENTRIES:
1982
		ret = compat_get_entries(sock_net(sk), user, len);
1983 1984
		break;
	default:
1985
		ret = do_ipt_get_ctl(sk, cmd, user, len);
1986
	}
L
Linus Torvalds 已提交
1987 1988
	return ret;
}
1989
#endif
L
Linus Torvalds 已提交
1990 1991

static int
1992
do_ipt_set_ctl(struct sock *sk, int cmd, void __user *user, unsigned int len)
L
Linus Torvalds 已提交
1993 1994 1995
{
	int ret;

1996
	if (!ns_capable(sock_net(sk)->user_ns, CAP_NET_ADMIN))
L
Linus Torvalds 已提交
1997 1998 1999 2000
		return -EPERM;

	switch (cmd) {
	case IPT_SO_SET_REPLACE:
2001
		ret = do_replace(sock_net(sk), user, len);
L
Linus Torvalds 已提交
2002 2003 2004
		break;

	case IPT_SO_SET_ADD_COUNTERS:
2005
		ret = do_add_counters(sock_net(sk), user, len, 0);
L
Linus Torvalds 已提交
2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017 2018 2019 2020
		break;

	default:
		duprintf("do_ipt_set_ctl:  unknown request %i\n", cmd);
		ret = -EINVAL;
	}

	return ret;
}

static int
do_ipt_get_ctl(struct sock *sk, int cmd, void __user *user, int *len)
{
	int ret;

2021
	if (!ns_capable(sock_net(sk)->user_ns, CAP_NET_ADMIN))
L
Linus Torvalds 已提交
2022 2023 2024
		return -EPERM;

	switch (cmd) {
2025
	case IPT_SO_GET_INFO:
2026
		ret = get_info(sock_net(sk), user, len, 0);
2027
		break;
L
Linus Torvalds 已提交
2028

2029
	case IPT_SO_GET_ENTRIES:
2030
		ret = get_entries(sock_net(sk), user, len);
L
Linus Torvalds 已提交
2031 2032 2033 2034
		break;

	case IPT_SO_GET_REVISION_MATCH:
	case IPT_SO_GET_REVISION_TARGET: {
2035
		struct xt_get_revision rev;
2036
		int target;
L
Linus Torvalds 已提交
2037 2038 2039 2040 2041 2042 2043 2044 2045

		if (*len != sizeof(rev)) {
			ret = -EINVAL;
			break;
		}
		if (copy_from_user(&rev, user, sizeof(rev)) != 0) {
			ret = -EFAULT;
			break;
		}
2046
		rev.name[sizeof(rev.name)-1] = 0;
L
Linus Torvalds 已提交
2047 2048

		if (cmd == IPT_SO_GET_REVISION_TARGET)
2049
			target = 1;
L
Linus Torvalds 已提交
2050
		else
2051
			target = 0;
L
Linus Torvalds 已提交
2052

2053 2054 2055
		try_then_request_module(xt_find_revision(AF_INET, rev.name,
							 rev.revision,
							 target, &ret),
L
Linus Torvalds 已提交
2056 2057 2058 2059 2060 2061 2062 2063 2064 2065 2066 2067
					"ipt_%s", rev.name);
		break;
	}

	default:
		duprintf("do_ipt_get_ctl: unknown request %i\n", cmd);
		ret = -EINVAL;
	}

	return ret;
}

2068 2069
struct xt_table *ipt_register_table(struct net *net,
				    const struct xt_table *table,
2070
				    const struct ipt_replace *repl)
L
Linus Torvalds 已提交
2071 2072
{
	int ret;
2073
	struct xt_table_info *newinfo;
2074
	struct xt_table_info bootstrap = {0};
2075
	void *loc_cpu_entry;
2076
	struct xt_table *new_table;
L
Linus Torvalds 已提交
2077

2078
	newinfo = xt_alloc_table_info(repl->size);
2079 2080 2081 2082
	if (!newinfo) {
		ret = -ENOMEM;
		goto out;
	}
L
Linus Torvalds 已提交
2083

2084
	loc_cpu_entry = newinfo->entries;
2085
	memcpy(loc_cpu_entry, repl->entries, repl->size);
L
Linus Torvalds 已提交
2086

2087
	ret = translate_table(net, newinfo, loc_cpu_entry, repl);
2088 2089
	if (ret != 0)
		goto out_free;
L
Linus Torvalds 已提交
2090

2091
	new_table = xt_register_table(net, table, &bootstrap, newinfo);
2092
	if (IS_ERR(new_table)) {
2093 2094
		ret = PTR_ERR(new_table);
		goto out_free;
L
Linus Torvalds 已提交
2095 2096
	}

2097 2098 2099 2100 2101 2102
	return new_table;

out_free:
	xt_free_table_info(newinfo);
out:
	return ERR_PTR(ret);
L
Linus Torvalds 已提交
2103 2104
}

2105
void ipt_unregister_table(struct net *net, struct xt_table *table)
L
Linus Torvalds 已提交
2106
{
2107
	struct xt_table_info *private;
2108
	void *loc_cpu_entry;
2109
	struct module *table_owner = table->me;
2110
	struct ipt_entry *iter;
2111

2112
	private = xt_unregister_table(table);
L
Linus Torvalds 已提交
2113 2114

	/* Decrease module usage counts and free resources */
2115
	loc_cpu_entry = private->entries;
2116
	xt_entry_foreach(iter, loc_cpu_entry, private->size)
2117
		cleanup_entry(iter, net);
2118 2119
	if (private->number > private->initial_entries)
		module_put(table_owner);
2120
	xt_free_table_info(private);
L
Linus Torvalds 已提交
2121 2122 2123
}

/* Returns 1 if the type and code is matched by the range, 0 otherwise */
2124
static inline bool
L
Linus Torvalds 已提交
2125 2126
icmp_type_code_match(u_int8_t test_type, u_int8_t min_code, u_int8_t max_code,
		     u_int8_t type, u_int8_t code,
2127
		     bool invert)
L
Linus Torvalds 已提交
2128
{
2129 2130
	return ((test_type == 0xFF) ||
		(type == test_type && code >= min_code && code <= max_code))
L
Linus Torvalds 已提交
2131 2132 2133
		^ invert;
}

2134
static bool
2135
icmp_match(const struct sk_buff *skb, struct xt_action_param *par)
L
Linus Torvalds 已提交
2136
{
2137 2138
	const struct icmphdr *ic;
	struct icmphdr _icmph;
2139
	const struct ipt_icmp *icmpinfo = par->matchinfo;
L
Linus Torvalds 已提交
2140 2141

	/* Must not be a fragment. */
2142
	if (par->fragoff != 0)
2143
		return false;
L
Linus Torvalds 已提交
2144

2145
	ic = skb_header_pointer(skb, par->thoff, sizeof(_icmph), &_icmph);
L
Linus Torvalds 已提交
2146 2147 2148 2149 2150
	if (ic == NULL) {
		/* We've been asked to examine this packet, and we
		 * can't.  Hence, no choice but to drop.
		 */
		duprintf("Dropping evil ICMP tinygram.\n");
2151
		par->hotdrop = true;
2152
		return false;
L
Linus Torvalds 已提交
2153 2154 2155 2156 2157 2158 2159 2160 2161
	}

	return icmp_type_code_match(icmpinfo->type,
				    icmpinfo->code[0],
				    icmpinfo->code[1],
				    ic->type, ic->code,
				    !!(icmpinfo->invflags&IPT_ICMP_INV));
}

2162
static int icmp_checkentry(const struct xt_mtchk_param *par)
L
Linus Torvalds 已提交
2163
{
2164
	const struct ipt_icmp *icmpinfo = par->matchinfo;
L
Linus Torvalds 已提交
2165

2166
	/* Must specify no unknown invflags */
2167
	return (icmpinfo->invflags & ~IPT_ICMP_INV) ? -EINVAL : 0;
L
Linus Torvalds 已提交
2168 2169
}

2170 2171
static struct xt_target ipt_builtin_tg[] __read_mostly = {
	{
2172
		.name             = XT_STANDARD_TARGET,
2173 2174
		.targetsize       = sizeof(int),
		.family           = NFPROTO_IPV4,
2175
#ifdef CONFIG_COMPAT
2176 2177 2178
		.compatsize       = sizeof(compat_int_t),
		.compat_from_user = compat_standard_from_user,
		.compat_to_user   = compat_standard_to_user,
2179
#endif
2180 2181
	},
	{
2182
		.name             = XT_ERROR_TARGET,
2183
		.target           = ipt_error,
2184
		.targetsize       = XT_FUNCTION_MAXNAMELEN,
2185 2186
		.family           = NFPROTO_IPV4,
	},
L
Linus Torvalds 已提交
2187 2188 2189 2190 2191 2192 2193
};

static struct nf_sockopt_ops ipt_sockopts = {
	.pf		= PF_INET,
	.set_optmin	= IPT_BASE_CTL,
	.set_optmax	= IPT_SO_SET_MAX+1,
	.set		= do_ipt_set_ctl,
2194 2195 2196
#ifdef CONFIG_COMPAT
	.compat_set	= compat_do_ipt_set_ctl,
#endif
L
Linus Torvalds 已提交
2197 2198 2199
	.get_optmin	= IPT_BASE_CTL,
	.get_optmax	= IPT_SO_GET_MAX+1,
	.get		= do_ipt_get_ctl,
2200 2201 2202
#ifdef CONFIG_COMPAT
	.compat_get	= compat_do_ipt_get_ctl,
#endif
2203
	.owner		= THIS_MODULE,
L
Linus Torvalds 已提交
2204 2205
};

2206 2207 2208 2209 2210 2211 2212 2213 2214
static struct xt_match ipt_builtin_mt[] __read_mostly = {
	{
		.name       = "icmp",
		.match      = icmp_match,
		.matchsize  = sizeof(struct ipt_icmp),
		.checkentry = icmp_checkentry,
		.proto      = IPPROTO_ICMP,
		.family     = NFPROTO_IPV4,
	},
L
Linus Torvalds 已提交
2215 2216
};

2217 2218
static int __net_init ip_tables_net_init(struct net *net)
{
2219
	return xt_proto_init(net, NFPROTO_IPV4);
2220 2221 2222 2223
}

static void __net_exit ip_tables_net_exit(struct net *net)
{
2224
	xt_proto_fini(net, NFPROTO_IPV4);
2225 2226 2227 2228 2229 2230 2231
}

static struct pernet_operations ip_tables_net_ops = {
	.init = ip_tables_net_init,
	.exit = ip_tables_net_exit,
};

2232
static int __init ip_tables_init(void)
L
Linus Torvalds 已提交
2233 2234 2235
{
	int ret;

2236
	ret = register_pernet_subsys(&ip_tables_net_ops);
2237 2238
	if (ret < 0)
		goto err1;
2239

L
Lucas De Marchi 已提交
2240
	/* No one else will be downing sem now, so we won't sleep */
2241
	ret = xt_register_targets(ipt_builtin_tg, ARRAY_SIZE(ipt_builtin_tg));
2242 2243
	if (ret < 0)
		goto err2;
2244
	ret = xt_register_matches(ipt_builtin_mt, ARRAY_SIZE(ipt_builtin_mt));
2245 2246
	if (ret < 0)
		goto err4;
L
Linus Torvalds 已提交
2247 2248 2249

	/* Register setsockopt */
	ret = nf_register_sockopt(&ipt_sockopts);
2250 2251
	if (ret < 0)
		goto err5;
L
Linus Torvalds 已提交
2252

2253
	pr_info("(C) 2000-2006 Netfilter Core Team\n");
L
Linus Torvalds 已提交
2254
	return 0;
2255 2256

err5:
2257
	xt_unregister_matches(ipt_builtin_mt, ARRAY_SIZE(ipt_builtin_mt));
2258
err4:
2259
	xt_unregister_targets(ipt_builtin_tg, ARRAY_SIZE(ipt_builtin_tg));
2260
err2:
2261
	unregister_pernet_subsys(&ip_tables_net_ops);
2262 2263
err1:
	return ret;
L
Linus Torvalds 已提交
2264 2265
}

2266
static void __exit ip_tables_fini(void)
L
Linus Torvalds 已提交
2267 2268
{
	nf_unregister_sockopt(&ipt_sockopts);
2269

2270 2271
	xt_unregister_matches(ipt_builtin_mt, ARRAY_SIZE(ipt_builtin_mt));
	xt_unregister_targets(ipt_builtin_tg, ARRAY_SIZE(ipt_builtin_tg));
2272
	unregister_pernet_subsys(&ip_tables_net_ops);
L
Linus Torvalds 已提交
2273 2274 2275 2276 2277
}

EXPORT_SYMBOL(ipt_register_table);
EXPORT_SYMBOL(ipt_unregister_table);
EXPORT_SYMBOL(ipt_do_table);
2278 2279
module_init(ip_tables_init);
module_exit(ip_tables_fini);