nfnetlink.c 17.3 KB
Newer Older
H
Harald Welte 已提交
1 2 3 4 5
/* Netfilter messages via netlink socket. Allows for user space
 * protocol helpers and general trouble making from userspace.
 *
 * (C) 2001 by Jay Schulist <jschlst@samba.org>,
 * (C) 2002-2005 by Harald Welte <laforge@gnumonks.org>
6
 * (C) 2005-2017 by Pablo Neira Ayuso <pablo@netfilter.org>
H
Harald Welte 已提交
7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24
 *
 * Initial netfilter messages via netlink development funded and
 * generally made possible by Network Robots, Inc. (www.networkrobots.com)
 *
 * Further development of this code funded by Astaro AG (http://www.astaro.com)
 *
 * This software may be used and distributed according to the terms
 * of the GNU General Public License, incorporated herein by reference.
 */

#include <linux/module.h>
#include <linux/types.h>
#include <linux/socket.h>
#include <linux/kernel.h>
#include <linux/string.h>
#include <linux/sockios.h>
#include <linux/net.h>
#include <linux/skbuff.h>
25
#include <linux/uaccess.h>
H
Harald Welte 已提交
26 27
#include <net/sock.h>
#include <linux/init.h>
28
#include <linux/sched/signal.h>
H
Harald Welte 已提交
29

30
#include <net/netlink.h>
31
#include <net/netns/generic.h>
H
Harald Welte 已提交
32 33 34
#include <linux/netfilter/nfnetlink.h>

MODULE_LICENSE("GPL");
35 36
MODULE_AUTHOR("Harald Welte <laforge@netfilter.org>");
MODULE_ALIAS_NET_PF_PROTO(PF_NETLINK, NETLINK_NETFILTER);
37
MODULE_DESCRIPTION("Netfilter messages via netlink socket");
H
Harald Welte 已提交
38

39 40 41 42
#define nfnl_dereference_protected(id) \
	rcu_dereference_protected(table[(id)].subsys, \
				  lockdep_nfnl_is_held((id)))

43 44
#define NFNL_MAX_ATTR_COUNT	32

45 46 47 48 49 50
static unsigned int nfnetlink_pernet_id __read_mostly;

struct nfnl_net {
	struct sock *nfnl;
};

51 52 53 54
static struct {
	struct mutex				mutex;
	const struct nfnetlink_subsystem __rcu	*subsys;
} table[NFNL_SUBSYS_COUNT];
H
Harald Welte 已提交
55

56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72
static struct lock_class_key nfnl_lockdep_keys[NFNL_SUBSYS_COUNT];

static const char *const nfnl_lockdep_names[NFNL_SUBSYS_COUNT] = {
	[NFNL_SUBSYS_NONE] = "nfnl_subsys_none",
	[NFNL_SUBSYS_CTNETLINK] = "nfnl_subsys_ctnetlink",
	[NFNL_SUBSYS_CTNETLINK_EXP] = "nfnl_subsys_ctnetlink_exp",
	[NFNL_SUBSYS_QUEUE] = "nfnl_subsys_queue",
	[NFNL_SUBSYS_ULOG] = "nfnl_subsys_ulog",
	[NFNL_SUBSYS_OSF] = "nfnl_subsys_osf",
	[NFNL_SUBSYS_IPSET] = "nfnl_subsys_ipset",
	[NFNL_SUBSYS_ACCT] = "nfnl_subsys_acct",
	[NFNL_SUBSYS_CTNETLINK_TIMEOUT] = "nfnl_subsys_cttimeout",
	[NFNL_SUBSYS_CTHELPER] = "nfnl_subsys_cthelper",
	[NFNL_SUBSYS_NFTABLES] = "nfnl_subsys_nftables",
	[NFNL_SUBSYS_NFT_COMPAT] = "nfnl_subsys_nftcompat",
};

73 74 75 76 77 78 79
static const int nfnl_group2type[NFNLGRP_MAX+1] = {
	[NFNLGRP_CONNTRACK_NEW]		= NFNL_SUBSYS_CTNETLINK,
	[NFNLGRP_CONNTRACK_UPDATE]	= NFNL_SUBSYS_CTNETLINK,
	[NFNLGRP_CONNTRACK_DESTROY]	= NFNL_SUBSYS_CTNETLINK,
	[NFNLGRP_CONNTRACK_EXP_NEW]	= NFNL_SUBSYS_CTNETLINK_EXP,
	[NFNLGRP_CONNTRACK_EXP_UPDATE]	= NFNL_SUBSYS_CTNETLINK_EXP,
	[NFNLGRP_CONNTRACK_EXP_DESTROY] = NFNL_SUBSYS_CTNETLINK_EXP,
80 81
	[NFNLGRP_NFTABLES]		= NFNL_SUBSYS_NFTABLES,
	[NFNLGRP_ACCT_QUOTA]		= NFNL_SUBSYS_ACCT,
82
	[NFNLGRP_NFTRACE]		= NFNL_SUBSYS_NFTABLES,
83 84
};

85 86 87 88 89
static struct nfnl_net *nfnl_pernet(struct net *net)
{
	return net_generic(net, nfnetlink_pernet_id);
}

90
void nfnl_lock(__u8 subsys_id)
H
Harald Welte 已提交
91
{
92
	mutex_lock(&table[subsys_id].mutex);
H
Harald Welte 已提交
93
}
94
EXPORT_SYMBOL_GPL(nfnl_lock);
H
Harald Welte 已提交
95

96
void nfnl_unlock(__u8 subsys_id)
97
{
98
	mutex_unlock(&table[subsys_id].mutex);
H
Harald Welte 已提交
99
}
100
EXPORT_SYMBOL_GPL(nfnl_unlock);
H
Harald Welte 已提交
101

102
#ifdef CONFIG_PROVE_LOCKING
103
bool lockdep_nfnl_is_held(u8 subsys_id)
104 105 106 107 108 109
{
	return lockdep_is_held(&table[subsys_id].mutex);
}
EXPORT_SYMBOL_GPL(lockdep_nfnl_is_held);
#endif

110
int nfnetlink_subsys_register(const struct nfnetlink_subsystem *n)
H
Harald Welte 已提交
111
{
112 113 114 115 116 117 118
	u8 cb_id;

	/* Sanity-check attr_count size to avoid stack buffer overflow. */
	for (cb_id = 0; cb_id < n->cb_count; cb_id++)
		if (WARN_ON(n->cb[cb_id].attr_count > NFNL_MAX_ATTR_COUNT))
			return -EINVAL;

119 120 121
	nfnl_lock(n->subsys_id);
	if (table[n->subsys_id].subsys) {
		nfnl_unlock(n->subsys_id);
122 123
		return -EBUSY;
	}
124 125
	rcu_assign_pointer(table[n->subsys_id].subsys, n);
	nfnl_unlock(n->subsys_id);
H
Harald Welte 已提交
126 127 128

	return 0;
}
129
EXPORT_SYMBOL_GPL(nfnetlink_subsys_register);
H
Harald Welte 已提交
130

131
int nfnetlink_subsys_unregister(const struct nfnetlink_subsystem *n)
H
Harald Welte 已提交
132
{
133 134 135
	nfnl_lock(n->subsys_id);
	table[n->subsys_id].subsys = NULL;
	nfnl_unlock(n->subsys_id);
136
	synchronize_rcu();
H
Harald Welte 已提交
137 138
	return 0;
}
139
EXPORT_SYMBOL_GPL(nfnetlink_subsys_unregister);
H
Harald Welte 已提交
140

141
static inline const struct nfnetlink_subsystem *nfnetlink_get_subsys(u16 type)
H
Harald Welte 已提交
142
{
143
	u8 subsys_id = NFNL_SUBSYS_ID(type);
H
Harald Welte 已提交
144

145
	if (subsys_id >= NFNL_SUBSYS_COUNT)
H
Harald Welte 已提交
146 147
		return NULL;

148
	return rcu_dereference(table[subsys_id].subsys);
H
Harald Welte 已提交
149 150
}

151
static inline const struct nfnl_callback *
152
nfnetlink_find_client(u16 type, const struct nfnetlink_subsystem *ss)
H
Harald Welte 已提交
153
{
154
	u8 cb_id = NFNL_MSG_TYPE(type);
155

156
	if (cb_id >= ss->cb_count)
H
Harald Welte 已提交
157 158 159 160 161
		return NULL;

	return &ss->cb[cb_id];
}

162
int nfnetlink_has_listeners(struct net *net, unsigned int group)
163
{
164 165 166
	struct nfnl_net *nfnlnet = nfnl_pernet(net);

	return netlink_has_listeners(nfnlnet->nfnl, group);
167 168 169
}
EXPORT_SYMBOL_GPL(nfnetlink_has_listeners);

170
int nfnetlink_send(struct sk_buff *skb, struct net *net, u32 portid,
171
		   unsigned int group, int echo, gfp_t flags)
H
Harald Welte 已提交
172
{
173 174 175
	struct nfnl_net *nfnlnet = nfnl_pernet(net);

	return nlmsg_notify(nfnlnet->nfnl, skb, portid, group, echo, flags);
H
Harald Welte 已提交
176
}
177
EXPORT_SYMBOL_GPL(nfnetlink_send);
H
Harald Welte 已提交
178

179
int nfnetlink_set_err(struct net *net, u32 portid, u32 group, int error)
180
{
181 182 183
	struct nfnl_net *nfnlnet = nfnl_pernet(net);

	return netlink_set_err(nfnlnet->nfnl, portid, group, error);
184 185 186
}
EXPORT_SYMBOL_GPL(nfnetlink_set_err);

187
int nfnetlink_unicast(struct sk_buff *skb, struct net *net, u32 portid)
H
Harald Welte 已提交
188
{
189
	struct nfnl_net *nfnlnet = nfnl_pernet(net);
190 191
	int err;

192
	err = nlmsg_unicast(nfnlnet->nfnl, skb, portid);
193 194 195 196
	if (err == -EAGAIN)
		err = -ENOBUFS;

	return err;
H
Harald Welte 已提交
197
}
198
EXPORT_SYMBOL_GPL(nfnetlink_unicast);
H
Harald Welte 已提交
199

200 201 202
void nfnetlink_broadcast(struct net *net, struct sk_buff *skb, __u32 portid,
			 __u32 group, gfp_t allocation)
{
203 204 205
	struct nfnl_net *nfnlnet = nfnl_pernet(net);

	netlink_broadcast(nfnlnet->nfnl, skb, portid, group, allocation);
206 207 208
}
EXPORT_SYMBOL_GPL(nfnetlink_broadcast);

H
Harald Welte 已提交
209
/* Process one complete nfnetlink message. */
J
Johannes Berg 已提交
210 211
static int nfnetlink_rcv_msg(struct sk_buff *skb, struct nlmsghdr *nlh,
			     struct netlink_ext_ack *extack)
H
Harald Welte 已提交
212
{
213
	struct net *net = sock_net(skb->sk);
214 215
	const struct nfnl_callback *nc;
	const struct nfnetlink_subsystem *ss;
216
	int type, err;
H
Harald Welte 已提交
217 218

	/* All the messages must at least contain nfgenmsg */
219
	if (nlmsg_len(nlh) < sizeof(struct nfgenmsg))
H
Harald Welte 已提交
220 221 222
		return 0;

	type = nlh->nlmsg_type;
223
replay:
224
	rcu_read_lock();
225

H
Harald Welte 已提交
226
	ss = nfnetlink_get_subsys(type);
227
	if (!ss) {
228
#ifdef CONFIG_MODULES
229
		rcu_read_unlock();
230
		request_module("nfnetlink-subsys-%d", NFNL_SUBSYS_ID(type));
231
		rcu_read_lock();
232
		ss = nfnetlink_get_subsys(type);
233 234
		if (!ss)
#endif
235 236
		{
			rcu_read_unlock();
237
			return -EINVAL;
238
		}
239
	}
H
Harald Welte 已提交
240 241

	nc = nfnetlink_find_client(type, ss);
242 243
	if (!nc) {
		rcu_read_unlock();
244
		return -EINVAL;
245
	}
H
Harald Welte 已提交
246 247

	{
248
		int min_len = nlmsg_total_size(sizeof(struct nfgenmsg));
249
		struct nfnl_net *nfnlnet = nfnl_pernet(net);
250
		u8 cb_id = NFNL_MSG_TYPE(nlh->nlmsg_type);
251
		struct nlattr *cda[NFNL_MAX_ATTR_COUNT + 1];
252 253
		struct nlattr *attr = (void *)nlh + min_len;
		int attrlen = nlh->nlmsg_len - min_len;
254
		__u8 subsys_id = NFNL_SUBSYS_ID(type);
255 256 257 258 259 260
		struct nfnl_info info = {
			.net	= net,
			.sk	= nfnlnet->nfnl,
			.nlh	= nlh,
			.extack	= extack,
		};
261

262 263 264 265 266 267
		/* Sanity-check NFNL_MAX_ATTR_COUNT */
		if (ss->cb[cb_id].attr_count > NFNL_MAX_ATTR_COUNT) {
			rcu_read_unlock();
			return -ENOMEM;
		}

268 269 270
		err = nla_parse_deprecated(cda, ss->cb[cb_id].attr_count,
					   attr, attrlen,
					   ss->cb[cb_id].policy, extack);
271 272
		if (err < 0) {
			rcu_read_unlock();
273
			return err;
274
		}
275

276
		if (nc->call_rcu) {
277 278
			err = nc->call_rcu(skb, &info,
					   (const struct nlattr **)cda);
279 280 281
			rcu_read_unlock();
		} else {
			rcu_read_unlock();
282
			nfnl_lock(subsys_id);
283
			if (nfnl_dereference_protected(subsys_id) != ss ||
284
			    nfnetlink_find_client(type, ss) != nc) {
285
				err = -EAGAIN;
286 287 288 289
			} else if (nc->call) {
				err = nc->call(skb, &info,
					       (const struct nlattr **)cda);
			} else {
290
				err = -EINVAL;
291
			}
292
			nfnl_unlock(subsys_id);
293
		}
294 295 296
		if (err == -EAGAIN)
			goto replay;
		return err;
H
Harald Welte 已提交
297 298 299
	}
}

300 301 302 303
struct nfnl_err {
	struct list_head	head;
	struct nlmsghdr		*nlh;
	int			err;
304
	struct netlink_ext_ack	extack;
305 306
};

307 308
static int nfnl_err_add(struct list_head *list, struct nlmsghdr *nlh, int err,
			const struct netlink_ext_ack *extack)
309 310 311 312 313 314 315 316 317
{
	struct nfnl_err *nfnl_err;

	nfnl_err = kmalloc(sizeof(struct nfnl_err), GFP_KERNEL);
	if (nfnl_err == NULL)
		return -ENOMEM;

	nfnl_err->nlh = nlh;
	nfnl_err->err = err;
318
	nfnl_err->extack = *extack;
319 320 321 322 323 324 325 326 327 328 329 330 331 332 333 334 335 336 337 338 339 340 341 342
	list_add_tail(&nfnl_err->head, list);

	return 0;
}

static void nfnl_err_del(struct nfnl_err *nfnl_err)
{
	list_del(&nfnl_err->head);
	kfree(nfnl_err);
}

static void nfnl_err_reset(struct list_head *err_list)
{
	struct nfnl_err *nfnl_err, *next;

	list_for_each_entry_safe(nfnl_err, next, err_list, head)
		nfnl_err_del(nfnl_err);
}

static void nfnl_err_deliver(struct list_head *err_list, struct sk_buff *skb)
{
	struct nfnl_err *nfnl_err, *next;

	list_for_each_entry_safe(nfnl_err, next, err_list, head) {
343 344
		netlink_ack(skb, nfnl_err->nlh, nfnl_err->err,
			    &nfnl_err->extack);
345 346 347 348
		nfnl_err_del(nfnl_err);
	}
}

349 350 351 352 353 354
enum {
	NFNL_BATCH_FAILURE	= (1 << 0),
	NFNL_BATCH_DONE		= (1 << 1),
	NFNL_BATCH_REPLAY	= (1 << 2),
};

355
static void nfnetlink_rcv_batch(struct sk_buff *skb, struct nlmsghdr *nlh,
356
				u16 subsys_id, u32 genid)
357
{
358
	struct sk_buff *oskb = skb;
359 360 361
	struct net *net = sock_net(skb->sk);
	const struct nfnetlink_subsystem *ss;
	const struct nfnl_callback *nc;
362
	struct netlink_ext_ack extack;
363
	LIST_HEAD(err_list);
364
	u32 status;
365 366 367
	int err;

	if (subsys_id >= NFNL_SUBSYS_COUNT)
J
Johannes Berg 已提交
368
		return netlink_ack(skb, nlh, -EINVAL, NULL);
369
replay:
370
	status = 0;
371
replay_abort:
372 373
	skb = netlink_skb_clone(oskb, GFP_KERNEL);
	if (!skb)
J
Johannes Berg 已提交
374
		return netlink_ack(oskb, nlh, -ENOMEM, NULL);
375 376

	nfnl_lock(subsys_id);
377
	ss = nfnl_dereference_protected(subsys_id);
378 379 380 381 382
	if (!ss) {
#ifdef CONFIG_MODULES
		nfnl_unlock(subsys_id);
		request_module("nfnetlink-subsys-%d", subsys_id);
		nfnl_lock(subsys_id);
383
		ss = nfnl_dereference_protected(subsys_id);
384 385 386 387
		if (!ss)
#endif
		{
			nfnl_unlock(subsys_id);
J
Johannes Berg 已提交
388
			netlink_ack(oskb, nlh, -EOPNOTSUPP, NULL);
389
			return kfree_skb(skb);
390 391 392
		}
	}

393
	if (!ss->valid_genid || !ss->commit || !ss->abort) {
394
		nfnl_unlock(subsys_id);
J
Johannes Berg 已提交
395
		netlink_ack(oskb, nlh, -EOPNOTSUPP, NULL);
396
		return kfree_skb(skb);
397 398
	}

399 400 401 402 403 404
	if (!try_module_get(ss->owner)) {
		nfnl_unlock(subsys_id);
		netlink_ack(oskb, nlh, -EOPNOTSUPP, NULL);
		return kfree_skb(skb);
	}

405
	if (!ss->valid_genid(net, genid)) {
406
		module_put(ss->owner);
407
		nfnl_unlock(subsys_id);
J
Johannes Berg 已提交
408
		netlink_ack(oskb, nlh, -ERESTART, NULL);
409 410 411
		return kfree_skb(skb);
	}

412 413
	nfnl_unlock(subsys_id);

414 415 416
	while (skb->len >= nlmsg_total_size(0)) {
		int msglen, type;

417 418 419 420 421 422 423
		if (fatal_signal_pending(current)) {
			nfnl_err_reset(&err_list);
			err = -EINTR;
			status = NFNL_BATCH_FAILURE;
			goto done;
		}

424
		memset(&extack, 0, sizeof(extack));
425 426 427
		nlh = nlmsg_hdr(skb);
		err = 0;

428 429 430 431 432 433
		if (nlh->nlmsg_len < NLMSG_HDRLEN ||
		    skb->len < nlh->nlmsg_len ||
		    nlmsg_len(nlh) < sizeof(struct nfgenmsg)) {
			nfnl_err_reset(&err_list);
			status |= NFNL_BATCH_FAILURE;
			goto done;
434 435 436 437 438 439 440 441 442 443 444
		}

		/* Only requests are handled by the kernel */
		if (!(nlh->nlmsg_flags & NLM_F_REQUEST)) {
			err = -EINVAL;
			goto ack;
		}

		type = nlh->nlmsg_type;
		if (type == NFNL_MSG_BATCH_BEGIN) {
			/* Malformed: Batch begin twice */
445
			nfnl_err_reset(&err_list);
446
			status |= NFNL_BATCH_FAILURE;
447 448
			goto done;
		} else if (type == NFNL_MSG_BATCH_END) {
449
			status |= NFNL_BATCH_DONE;
450 451 452 453 454 455 456 457 458 459 460 461 462 463 464 465 466 467 468 469 470 471
			goto done;
		} else if (type < NLMSG_MIN_TYPE) {
			err = -EINVAL;
			goto ack;
		}

		/* We only accept a batch with messages for the same
		 * subsystem.
		 */
		if (NFNL_SUBSYS_ID(type) != subsys_id) {
			err = -EINVAL;
			goto ack;
		}

		nc = nfnetlink_find_client(type, ss);
		if (!nc) {
			err = -EINVAL;
			goto ack;
		}

		{
			int min_len = nlmsg_total_size(sizeof(struct nfgenmsg));
472
			u8 cb_id = NFNL_MSG_TYPE(nlh->nlmsg_type);
473
			struct nlattr *cda[NFNL_MAX_ATTR_COUNT + 1];
474 475 476
			struct nlattr *attr = (void *)nlh + min_len;
			int attrlen = nlh->nlmsg_len - min_len;

477 478 479 480 481 482
			/* Sanity-check NFTA_MAX_ATTR */
			if (ss->cb[cb_id].attr_count > NFNL_MAX_ATTR_COUNT) {
				err = -ENOMEM;
				goto ack;
			}

483 484 485 486
			err = nla_parse_deprecated(cda,
						   ss->cb[cb_id].attr_count,
						   attr, attrlen,
						   ss->cb[cb_id].policy, NULL);
487 488 489 490
			if (err < 0)
				goto ack;

			if (nc->call_batch) {
491 492 493
				struct nfnl_net *nfnlnet = nfnl_pernet(net);

				err = nc->call_batch(net, nfnlnet->nfnl, skb, nlh,
494 495
						     (const struct nlattr **)cda,
						     &extack);
496 497 498 499 500 501 502
			}

			/* The lock was released to autoload some module, we
			 * have to abort and start from scratch using the
			 * original skb.
			 */
			if (err == -EAGAIN) {
503
				status |= NFNL_BATCH_REPLAY;
504
				goto done;
505 506 507 508
			}
		}
ack:
		if (nlh->nlmsg_flags & NLM_F_ACK || err) {
509 510 511 512
			/* Errors are delivered once the full batch has been
			 * processed, this avoids that the same error is
			 * reported several times when replaying the batch.
			 */
513
			if (nfnl_err_add(&err_list, nlh, err, &extack) < 0) {
514 515 516 517 518
				/* We failed to enqueue an error, reset the
				 * list of errors and send OOM to userspace
				 * pointing to the batch header.
				 */
				nfnl_err_reset(&err_list);
J
Johannes Berg 已提交
519 520
				netlink_ack(oskb, nlmsg_hdr(oskb), -ENOMEM,
					    NULL);
521
				status |= NFNL_BATCH_FAILURE;
522 523
				goto done;
			}
524 525 526 527 528
			/* We don't stop processing the batch on errors, thus,
			 * userspace gets all the errors that the batch
			 * triggers.
			 */
			if (err)
529
				status |= NFNL_BATCH_FAILURE;
530
		}
531

532 533 534 535 536 537
		msglen = NLMSG_ALIGN(nlh->nlmsg_len);
		if (msglen > skb->len)
			msglen = skb->len;
		skb_pull(skb, msglen);
	}
done:
538
	if (status & NFNL_BATCH_REPLAY) {
539
		ss->abort(net, oskb, NFNL_ABORT_AUTOLOAD);
540 541
		nfnl_err_reset(&err_list);
		kfree_skb(skb);
542
		module_put(ss->owner);
543 544
		goto replay;
	} else if (status == NFNL_BATCH_DONE) {
545 546 547 548 549
		err = ss->commit(net, oskb);
		if (err == -EAGAIN) {
			status |= NFNL_BATCH_REPLAY;
			goto done;
		} else if (err) {
550
			ss->abort(net, oskb, NFNL_ABORT_NONE);
551 552
			netlink_ack(oskb, nlmsg_hdr(oskb), err, NULL);
		}
553
	} else {
554 555 556 557 558 559 560 561 562 563 564 565 566 567 568
		enum nfnl_abort_action abort_action;

		if (status & NFNL_BATCH_FAILURE)
			abort_action = NFNL_ABORT_NONE;
		else
			abort_action = NFNL_ABORT_VALIDATE;

		err = ss->abort(net, oskb, abort_action);
		if (err == -EAGAIN) {
			nfnl_err_reset(&err_list);
			kfree_skb(skb);
			module_put(ss->owner);
			status |= NFNL_BATCH_FAILURE;
			goto replay_abort;
		}
569
	}
570 571
	if (ss->cleanup)
		ss->cleanup(net);
572

573
	nfnl_err_deliver(&err_list, oskb);
574
	kfree_skb(skb);
575
	module_put(ss->owner);
576 577
}

578 579 580 581
static const struct nla_policy nfnl_batch_policy[NFNL_BATCH_MAX + 1] = {
	[NFNL_BATCH_GENID]	= { .type = NLA_U32 },
};

582
static void nfnetlink_rcv_skb_batch(struct sk_buff *skb, struct nlmsghdr *nlh)
H
Harald Welte 已提交
583
{
584 585 586 587
	int min_len = nlmsg_total_size(sizeof(struct nfgenmsg));
	struct nlattr *attr = (void *)nlh + min_len;
	struct nlattr *cda[NFNL_BATCH_MAX + 1];
	int attrlen = nlh->nlmsg_len - min_len;
588
	struct nfgenmsg *nfgenmsg;
589 590
	int msglen, err;
	u32 gen_id = 0;
591
	u16 res_id;
592

593 594 595 596
	msglen = NLMSG_ALIGN(nlh->nlmsg_len);
	if (msglen > skb->len)
		msglen = skb->len;

597
	if (skb->len < NLMSG_HDRLEN + sizeof(struct nfgenmsg))
598 599
		return;

600 601
	err = nla_parse_deprecated(cda, NFNL_BATCH_MAX, attr, attrlen,
				   nfnl_batch_policy, NULL);
602
	if (err < 0) {
J
Johannes Berg 已提交
603
		netlink_ack(skb, nlh, err, NULL);
604 605 606 607 608
		return;
	}
	if (cda[NFNL_BATCH_GENID])
		gen_id = ntohl(nla_get_be32(cda[NFNL_BATCH_GENID]));

609 610 611 612 613 614 615 616
	nfgenmsg = nlmsg_data(nlh);
	skb_pull(skb, msglen);
	/* Work around old nft using host byte order */
	if (nfgenmsg->res_id == NFNL_SUBSYS_NFTABLES)
		res_id = NFNL_SUBSYS_NFTABLES;
	else
		res_id = ntohs(nfgenmsg->res_id);

617
	nfnetlink_rcv_batch(skb, nlh, res_id, gen_id);
618 619 620 621 622 623
}

static void nfnetlink_rcv(struct sk_buff *skb)
{
	struct nlmsghdr *nlh = nlmsg_hdr(skb);

624 625
	if (skb->len < NLMSG_HDRLEN ||
	    nlh->nlmsg_len < NLMSG_HDRLEN ||
626 627 628
	    skb->len < nlh->nlmsg_len)
		return;

629
	if (!netlink_net_capable(skb, CAP_NET_ADMIN)) {
J
Johannes Berg 已提交
630
		netlink_ack(skb, nlh, -EPERM, NULL);
631 632 633
		return;
	}

634 635 636
	if (nlh->nlmsg_type == NFNL_MSG_BATCH_BEGIN)
		nfnetlink_rcv_skb_batch(skb, nlh);
	else
637
		netlink_rcv_skb(skb, nfnetlink_rcv_msg);
H
Harald Welte 已提交
638 639
}

640
#ifdef CONFIG_MODULES
641
static int nfnetlink_bind(struct net *net, int group)
642 643
{
	const struct nfnetlink_subsystem *ss;
644 645 646
	int type;

	if (group <= NFNLGRP_NONE || group > NFNLGRP_MAX)
647
		return 0;
648 649

	type = nfnl_group2type[group];
650 651

	rcu_read_lock();
652
	ss = nfnetlink_get_subsys(type << 8);
653
	rcu_read_unlock();
654
	if (!ss)
655
		request_module_nowait("nfnetlink-subsys-%d", type);
656
	return 0;
657 658 659
}
#endif

660
static int __net_init nfnetlink_net_init(struct net *net)
H
Harald Welte 已提交
661
{
662
	struct nfnl_net *nfnlnet = nfnl_pernet(net);
663 664 665
	struct netlink_kernel_cfg cfg = {
		.groups	= NFNLGRP_MAX,
		.input	= nfnetlink_rcv,
666 667 668
#ifdef CONFIG_MODULES
		.bind	= nfnetlink_bind,
#endif
669
	};
670

671 672
	nfnlnet->nfnl = netlink_kernel_create(net, NETLINK_NETFILTER, &cfg);
	if (!nfnlnet->nfnl)
673 674
		return -ENOMEM;
	return 0;
H
Harald Welte 已提交
675 676
}

677
static void __net_exit nfnetlink_net_exit_batch(struct list_head *net_exit_list)
H
Harald Welte 已提交
678
{
679
	struct nfnl_net *nfnlnet;
680
	struct net *net;
H
Harald Welte 已提交
681

682 683 684 685 686
	list_for_each_entry(net, net_exit_list, exit_list) {
		nfnlnet = nfnl_pernet(net);

		netlink_kernel_release(nfnlnet->nfnl);
	}
687
}
H
Harald Welte 已提交
688

689 690 691
static struct pernet_operations nfnetlink_net_ops = {
	.init		= nfnetlink_net_init,
	.exit_batch	= nfnetlink_net_exit_batch,
692 693
	.id		= &nfnetlink_pernet_id,
	.size		= sizeof(struct nfnl_net),
694 695 696 697
};

static int __init nfnetlink_init(void)
{
698 699
	int i;

700 701 702
	for (i = NFNLGRP_NONE + 1; i <= NFNLGRP_MAX; i++)
		BUG_ON(nfnl_group2type[i] == NFNL_SUBSYS_NONE);

703
	for (i=0; i<NFNL_SUBSYS_COUNT; i++)
704
		__mutex_init(&table[i].mutex, nfnl_lockdep_names[i], &nfnl_lockdep_keys[i]);
705

706
	return register_pernet_subsys(&nfnetlink_net_ops);
H
Harald Welte 已提交
707 708
}

709 710 711 712
static void __exit nfnetlink_exit(void)
{
	unregister_pernet_subsys(&nfnetlink_net_ops);
}
H
Harald Welte 已提交
713 714
module_init(nfnetlink_init);
module_exit(nfnetlink_exit);