xfrm_state.c 53.6 KB
Newer Older
L
Linus Torvalds 已提交
1 2 3 4 5 6 7 8 9 10 11 12
/*
 * xfrm_state.c
 *
 * Changes:
 *	Mitsuru KANDA @USAGI
 * 	Kazunori MIYAZAWA @USAGI
 * 	Kunihiro Ishiguro <kunihiro@ipinfusion.com>
 * 		IPv6 support
 * 	YOSHIFUJI Hideaki @USAGI
 * 		Split up af-specific functions
 *	Derek Atkins <derek@ihtfp.com>
 *		Add UDP Encapsulation
13
 *
L
Linus Torvalds 已提交
14 15 16 17 18 19 20
 */

#include <linux/workqueue.h>
#include <net/xfrm.h>
#include <linux/pfkeyv2.h>
#include <linux/ipsec.h>
#include <linux/module.h>
21
#include <linux/cache.h>
P
Paul Moore 已提交
22
#include <linux/audit.h>
23
#include <asm/uaccess.h>
L
Linus Torvalds 已提交
24

25 26
#include "xfrm_hash.h"

L
Linus Torvalds 已提交
27 28 29
/* Each xfrm_state may be linked to two tables:

   1. Hash table by (spi,daddr,ah/esp) to find SA by SPI. (input,ctl)
30
   2. Hash table by (daddr,family,reqid) to find what SAs exist for given
L
Linus Torvalds 已提交
31 32 33 34 35
      destination/tunnel endpoint. (output)
 */

static DEFINE_SPINLOCK(xfrm_state_lock);

36
static unsigned int xfrm_state_hashmax __read_mostly = 1 * 1024 * 1024;
37
static unsigned int xfrm_state_genid;
38

39 40 41
static struct xfrm_state_afinfo *xfrm_state_get_afinfo(unsigned int family);
static void xfrm_state_put_afinfo(struct xfrm_state_afinfo *afinfo);

P
Paul Moore 已提交
42 43 44 45 46 47 48
#ifdef CONFIG_AUDITSYSCALL
static void xfrm_audit_state_replay(struct xfrm_state *x,
				    struct sk_buff *skb, __be32 net_seq);
#else
#define xfrm_audit_state_replay(x, s, sq)	do { ; } while (0)
#endif /* CONFIG_AUDITSYSCALL */

49 50
static inline unsigned int xfrm_dst_hash(struct net *net,
					 xfrm_address_t *daddr,
51 52
					 xfrm_address_t *saddr,
					 u32 reqid,
53
					 unsigned short family)
54
{
55
	return __xfrm_dst_hash(daddr, saddr, reqid, family, net->xfrm.state_hmask);
56 57
}

58 59
static inline unsigned int xfrm_src_hash(struct net *net,
					 xfrm_address_t *daddr,
60
					 xfrm_address_t *saddr,
61
					 unsigned short family)
62
{
63
	return __xfrm_src_hash(daddr, saddr, family, net->xfrm.state_hmask);
64 65 66
}

static inline unsigned int
67
xfrm_spi_hash(struct net *net, xfrm_address_t *daddr, __be32 spi, u8 proto, unsigned short family)
68
{
69
	return __xfrm_spi_hash(daddr, spi, proto, family, net->xfrm.state_hmask);
70 71 72 73 74 75 76 77 78 79 80 81 82 83
}

static void xfrm_hash_transfer(struct hlist_head *list,
			       struct hlist_head *ndsttable,
			       struct hlist_head *nsrctable,
			       struct hlist_head *nspitable,
			       unsigned int nhashmask)
{
	struct hlist_node *entry, *tmp;
	struct xfrm_state *x;

	hlist_for_each_entry_safe(x, entry, tmp, list, bydst) {
		unsigned int h;

84 85 86
		h = __xfrm_dst_hash(&x->id.daddr, &x->props.saddr,
				    x->props.reqid, x->props.family,
				    nhashmask);
87 88
		hlist_add_head(&x->bydst, ndsttable+h);

89 90
		h = __xfrm_src_hash(&x->id.daddr, &x->props.saddr,
				    x->props.family,
91 92 93
				    nhashmask);
		hlist_add_head(&x->bysrc, nsrctable+h);

94 95 96 97 98 99
		if (x->id.spi) {
			h = __xfrm_spi_hash(&x->id.daddr, x->id.spi,
					    x->id.proto, x->props.family,
					    nhashmask);
			hlist_add_head(&x->byspi, nspitable+h);
		}
100 101 102
	}
}

103
static unsigned long xfrm_hash_new_size(unsigned int state_hmask)
104
{
105
	return ((state_hmask + 1) << 1) * sizeof(struct hlist_head);
106 107 108 109
}

static DEFINE_MUTEX(hash_resize_mutex);

110
static void xfrm_hash_resize(struct work_struct *work)
111
{
112
	struct net *net = container_of(work, struct net, xfrm.state_hash_work);
113 114 115 116 117 118 119
	struct hlist_head *ndst, *nsrc, *nspi, *odst, *osrc, *ospi;
	unsigned long nsize, osize;
	unsigned int nhashmask, ohashmask;
	int i;

	mutex_lock(&hash_resize_mutex);

120
	nsize = xfrm_hash_new_size(net->xfrm.state_hmask);
121
	ndst = xfrm_hash_alloc(nsize);
122 123
	if (!ndst)
		goto out_unlock;
124
	nsrc = xfrm_hash_alloc(nsize);
125
	if (!nsrc) {
126
		xfrm_hash_free(ndst, nsize);
127 128
		goto out_unlock;
	}
129
	nspi = xfrm_hash_alloc(nsize);
130
	if (!nspi) {
131 132
		xfrm_hash_free(ndst, nsize);
		xfrm_hash_free(nsrc, nsize);
133 134 135 136 137 138
		goto out_unlock;
	}

	spin_lock_bh(&xfrm_state_lock);

	nhashmask = (nsize / sizeof(struct hlist_head)) - 1U;
139 140
	for (i = net->xfrm.state_hmask; i >= 0; i--)
		xfrm_hash_transfer(net->xfrm.state_bydst+i, ndst, nsrc, nspi,
141 142
				   nhashmask);

143 144 145 146
	odst = net->xfrm.state_bydst;
	osrc = net->xfrm.state_bysrc;
	ospi = net->xfrm.state_byspi;
	ohashmask = net->xfrm.state_hmask;
147

148 149 150 151
	net->xfrm.state_bydst = ndst;
	net->xfrm.state_bysrc = nsrc;
	net->xfrm.state_byspi = nspi;
	net->xfrm.state_hmask = nhashmask;
152 153 154 155

	spin_unlock_bh(&xfrm_state_lock);

	osize = (ohashmask + 1) * sizeof(struct hlist_head);
156 157 158
	xfrm_hash_free(odst, osize);
	xfrm_hash_free(osrc, osize);
	xfrm_hash_free(ospi, osize);
159 160 161 162 163

out_unlock:
	mutex_unlock(&hash_resize_mutex);
}

L
Linus Torvalds 已提交
164 165 166 167 168
static DEFINE_RWLOCK(xfrm_state_afinfo_lock);
static struct xfrm_state_afinfo *xfrm_state_afinfo[NPROTO];

static DEFINE_SPINLOCK(xfrm_state_gc_lock);

169
int __xfrm_state_delete(struct xfrm_state *x);
L
Linus Torvalds 已提交
170

171
int km_query(struct xfrm_state *x, struct xfrm_tmpl *t, struct xfrm_policy *pol);
172
void km_state_expired(struct xfrm_state *x, int hard, u32 pid);
L
Linus Torvalds 已提交
173

174 175 176 177 178 179 180 181 182 183 184 185 186
static struct xfrm_state_afinfo *xfrm_state_lock_afinfo(unsigned int family)
{
	struct xfrm_state_afinfo *afinfo;
	if (unlikely(family >= NPROTO))
		return NULL;
	write_lock_bh(&xfrm_state_afinfo_lock);
	afinfo = xfrm_state_afinfo[family];
	if (unlikely(!afinfo))
		write_unlock_bh(&xfrm_state_afinfo_lock);
	return afinfo;
}

static void xfrm_state_unlock_afinfo(struct xfrm_state_afinfo *afinfo)
187
	__releases(xfrm_state_afinfo_lock)
188 189 190 191
{
	write_unlock_bh(&xfrm_state_afinfo_lock);
}

192
int xfrm_register_type(const struct xfrm_type *type, unsigned short family)
193 194
{
	struct xfrm_state_afinfo *afinfo = xfrm_state_lock_afinfo(family);
195
	const struct xfrm_type **typemap;
196 197 198 199 200 201 202 203 204 205 206 207 208 209 210
	int err = 0;

	if (unlikely(afinfo == NULL))
		return -EAFNOSUPPORT;
	typemap = afinfo->type_map;

	if (likely(typemap[type->proto] == NULL))
		typemap[type->proto] = type;
	else
		err = -EEXIST;
	xfrm_state_unlock_afinfo(afinfo);
	return err;
}
EXPORT_SYMBOL(xfrm_register_type);

211
int xfrm_unregister_type(const struct xfrm_type *type, unsigned short family)
212 213
{
	struct xfrm_state_afinfo *afinfo = xfrm_state_lock_afinfo(family);
214
	const struct xfrm_type **typemap;
215 216 217 218 219 220 221 222 223 224 225 226 227 228 229
	int err = 0;

	if (unlikely(afinfo == NULL))
		return -EAFNOSUPPORT;
	typemap = afinfo->type_map;

	if (unlikely(typemap[type->proto] != type))
		err = -ENOENT;
	else
		typemap[type->proto] = NULL;
	xfrm_state_unlock_afinfo(afinfo);
	return err;
}
EXPORT_SYMBOL(xfrm_unregister_type);

230
static const struct xfrm_type *xfrm_get_type(u8 proto, unsigned short family)
231 232
{
	struct xfrm_state_afinfo *afinfo;
233 234
	const struct xfrm_type **typemap;
	const struct xfrm_type *type;
235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256
	int modload_attempted = 0;

retry:
	afinfo = xfrm_state_get_afinfo(family);
	if (unlikely(afinfo == NULL))
		return NULL;
	typemap = afinfo->type_map;

	type = typemap[proto];
	if (unlikely(type && !try_module_get(type->owner)))
		type = NULL;
	if (!type && !modload_attempted) {
		xfrm_state_put_afinfo(afinfo);
		request_module("xfrm-type-%d-%d", family, proto);
		modload_attempted = 1;
		goto retry;
	}

	xfrm_state_put_afinfo(afinfo);
	return type;
}

257
static void xfrm_put_type(const struct xfrm_type *type)
258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276
{
	module_put(type->owner);
}

int xfrm_register_mode(struct xfrm_mode *mode, int family)
{
	struct xfrm_state_afinfo *afinfo;
	struct xfrm_mode **modemap;
	int err;

	if (unlikely(mode->encap >= XFRM_MODE_MAX))
		return -EINVAL;

	afinfo = xfrm_state_lock_afinfo(family);
	if (unlikely(afinfo == NULL))
		return -EAFNOSUPPORT;

	err = -EEXIST;
	modemap = afinfo->mode_map;
277 278
	if (modemap[mode->encap])
		goto out;
279

280 281 282 283 284 285 286 287 288
	err = -ENOENT;
	if (!try_module_get(afinfo->owner))
		goto out;

	mode->afinfo = afinfo;
	modemap[mode->encap] = mode;
	err = 0;

out:
289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310
	xfrm_state_unlock_afinfo(afinfo);
	return err;
}
EXPORT_SYMBOL(xfrm_register_mode);

int xfrm_unregister_mode(struct xfrm_mode *mode, int family)
{
	struct xfrm_state_afinfo *afinfo;
	struct xfrm_mode **modemap;
	int err;

	if (unlikely(mode->encap >= XFRM_MODE_MAX))
		return -EINVAL;

	afinfo = xfrm_state_lock_afinfo(family);
	if (unlikely(afinfo == NULL))
		return -EAFNOSUPPORT;

	err = -ENOENT;
	modemap = afinfo->mode_map;
	if (likely(modemap[mode->encap] == mode)) {
		modemap[mode->encap] = NULL;
311
		module_put(mode->afinfo->owner);
312 313 314 315 316 317 318 319 320 321 322 323 324 325 326 327 328 329 330 331 332 333 334 335 336 337 338 339 340 341 342 343 344 345 346 347 348 349 350 351 352
		err = 0;
	}

	xfrm_state_unlock_afinfo(afinfo);
	return err;
}
EXPORT_SYMBOL(xfrm_unregister_mode);

static struct xfrm_mode *xfrm_get_mode(unsigned int encap, int family)
{
	struct xfrm_state_afinfo *afinfo;
	struct xfrm_mode *mode;
	int modload_attempted = 0;

	if (unlikely(encap >= XFRM_MODE_MAX))
		return NULL;

retry:
	afinfo = xfrm_state_get_afinfo(family);
	if (unlikely(afinfo == NULL))
		return NULL;

	mode = afinfo->mode_map[encap];
	if (unlikely(mode && !try_module_get(mode->owner)))
		mode = NULL;
	if (!mode && !modload_attempted) {
		xfrm_state_put_afinfo(afinfo);
		request_module("xfrm-mode-%d-%d", family, encap);
		modload_attempted = 1;
		goto retry;
	}

	xfrm_state_put_afinfo(afinfo);
	return mode;
}

static void xfrm_put_mode(struct xfrm_mode *mode)
{
	module_put(mode->owner);
}

L
Linus Torvalds 已提交
353 354
static void xfrm_state_gc_destroy(struct xfrm_state *x)
{
355 356
	del_timer_sync(&x->timer);
	del_timer_sync(&x->rtimer);
J
Jesper Juhl 已提交
357 358 359 360
	kfree(x->aalg);
	kfree(x->ealg);
	kfree(x->calg);
	kfree(x->encap);
361
	kfree(x->coaddr);
362 363
	if (x->inner_mode)
		xfrm_put_mode(x->inner_mode);
364 365
	if (x->inner_mode_iaf)
		xfrm_put_mode(x->inner_mode_iaf);
366 367
	if (x->outer_mode)
		xfrm_put_mode(x->outer_mode);
L
Linus Torvalds 已提交
368 369 370 371
	if (x->type) {
		x->type->destructor(x);
		xfrm_put_type(x->type);
	}
372
	security_xfrm_state_free(x);
L
Linus Torvalds 已提交
373 374 375
	kfree(x);
}

376
static void xfrm_state_gc_task(struct work_struct *work)
L
Linus Torvalds 已提交
377
{
378
	struct net *net = container_of(work, struct net, xfrm.state_gc_work);
H
Herbert Xu 已提交
379 380 381
	struct xfrm_state *x;
	struct hlist_node *entry, *tmp;
	struct hlist_head gc_list;
L
Linus Torvalds 已提交
382 383

	spin_lock_bh(&xfrm_state_gc_lock);
384
	hlist_move_list(&net->xfrm.state_gc_list, &gc_list);
L
Linus Torvalds 已提交
385 386
	spin_unlock_bh(&xfrm_state_gc_lock);

H
Herbert Xu 已提交
387
	hlist_for_each_entry_safe(x, entry, tmp, &gc_list, gclist)
L
Linus Torvalds 已提交
388
		xfrm_state_gc_destroy(x);
389

A
Alexey Dobriyan 已提交
390
	wake_up(&net->xfrm.km_waitq);
L
Linus Torvalds 已提交
391 392 393 394 395 396 397
}

static inline unsigned long make_jiffies(long secs)
{
	if (secs >= (MAX_SCHEDULE_TIMEOUT-1)/HZ)
		return MAX_SCHEDULE_TIMEOUT-1;
	else
398
		return secs*HZ;
L
Linus Torvalds 已提交
399 400 401 402 403
}

static void xfrm_timer_handler(unsigned long data)
{
	struct xfrm_state *x = (struct xfrm_state*)data;
404
	struct net *net = xs_net(x);
405
	unsigned long now = get_seconds();
L
Linus Torvalds 已提交
406 407
	long next = LONG_MAX;
	int warn = 0;
J
Joy Latten 已提交
408
	int err = 0;
L
Linus Torvalds 已提交
409 410 411 412 413 414 415 416 417 418 419 420 421 422 423 424 425 426 427 428 429 430 431 432 433 434 435 436 437 438 439 440 441 442 443 444 445 446 447 448 449

	spin_lock(&x->lock);
	if (x->km.state == XFRM_STATE_DEAD)
		goto out;
	if (x->km.state == XFRM_STATE_EXPIRED)
		goto expired;
	if (x->lft.hard_add_expires_seconds) {
		long tmo = x->lft.hard_add_expires_seconds +
			x->curlft.add_time - now;
		if (tmo <= 0)
			goto expired;
		if (tmo < next)
			next = tmo;
	}
	if (x->lft.hard_use_expires_seconds) {
		long tmo = x->lft.hard_use_expires_seconds +
			(x->curlft.use_time ? : now) - now;
		if (tmo <= 0)
			goto expired;
		if (tmo < next)
			next = tmo;
	}
	if (x->km.dying)
		goto resched;
	if (x->lft.soft_add_expires_seconds) {
		long tmo = x->lft.soft_add_expires_seconds +
			x->curlft.add_time - now;
		if (tmo <= 0)
			warn = 1;
		else if (tmo < next)
			next = tmo;
	}
	if (x->lft.soft_use_expires_seconds) {
		long tmo = x->lft.soft_use_expires_seconds +
			(x->curlft.use_time ? : now) - now;
		if (tmo <= 0)
			warn = 1;
		else if (tmo < next)
			next = tmo;
	}

450
	x->km.dying = warn;
L
Linus Torvalds 已提交
451
	if (warn)
452
		km_state_expired(x, 0, 0);
L
Linus Torvalds 已提交
453
resched:
454 455 456
	if (next != LONG_MAX)
		mod_timer(&x->timer, jiffies + make_jiffies(next));

L
Linus Torvalds 已提交
457 458 459 460 461
	goto out;

expired:
	if (x->km.state == XFRM_STATE_ACQ && x->id.spi == 0) {
		x->km.state = XFRM_STATE_EXPIRED;
462
		wake_up(&net->xfrm.km_waitq);
L
Linus Torvalds 已提交
463 464 465
		next = 2;
		goto resched;
	}
J
Joy Latten 已提交
466 467 468

	err = __xfrm_state_delete(x);
	if (!err && x->id.spi)
469
		km_state_expired(x, 1, 0);
L
Linus Torvalds 已提交
470

J
Joy Latten 已提交
471
	xfrm_audit_state_delete(x, err ? 0 : 1,
472 473
				audit_get_loginuid(current),
				audit_get_sessionid(current), 0);
J
Joy Latten 已提交
474

L
Linus Torvalds 已提交
475 476 477 478
out:
	spin_unlock(&x->lock);
}

479 480
static void xfrm_replay_timer_handler(unsigned long data);

481
struct xfrm_state *xfrm_state_alloc(struct net *net)
L
Linus Torvalds 已提交
482 483 484
{
	struct xfrm_state *x;

485
	x = kzalloc(sizeof(struct xfrm_state), GFP_ATOMIC);
L
Linus Torvalds 已提交
486 487

	if (x) {
488
		write_pnet(&x->xs_net, net);
L
Linus Torvalds 已提交
489 490
		atomic_set(&x->refcnt, 1);
		atomic_set(&x->tunnel_users, 0);
H
Herbert Xu 已提交
491
		INIT_LIST_HEAD(&x->km.all);
492 493 494
		INIT_HLIST_NODE(&x->bydst);
		INIT_HLIST_NODE(&x->bysrc);
		INIT_HLIST_NODE(&x->byspi);
495 496 497
		setup_timer(&x->timer, xfrm_timer_handler, (unsigned long)x);
		setup_timer(&x->rtimer, xfrm_replay_timer_handler,
				(unsigned long)x);
498
		x->curlft.add_time = get_seconds();
L
Linus Torvalds 已提交
499 500 501 502
		x->lft.soft_byte_limit = XFRM_INF;
		x->lft.soft_packet_limit = XFRM_INF;
		x->lft.hard_byte_limit = XFRM_INF;
		x->lft.hard_packet_limit = XFRM_INF;
503 504
		x->replay_maxage = 0;
		x->replay_maxdiff = 0;
505 506
		x->inner_mode = NULL;
		x->inner_mode_iaf = NULL;
L
Linus Torvalds 已提交
507 508 509 510 511 512 513 514
		spin_lock_init(&x->lock);
	}
	return x;
}
EXPORT_SYMBOL(xfrm_state_alloc);

void __xfrm_state_destroy(struct xfrm_state *x)
{
515 516
	struct net *net = xs_net(x);

517
	WARN_ON(x->km.state != XFRM_STATE_DEAD);
L
Linus Torvalds 已提交
518 519

	spin_lock_bh(&xfrm_state_gc_lock);
520
	hlist_add_head(&x->gclist, &net->xfrm.state_gc_list);
L
Linus Torvalds 已提交
521
	spin_unlock_bh(&xfrm_state_gc_lock);
522
	schedule_work(&net->xfrm.state_gc_work);
L
Linus Torvalds 已提交
523 524 525
}
EXPORT_SYMBOL(__xfrm_state_destroy);

526
int __xfrm_state_delete(struct xfrm_state *x)
L
Linus Torvalds 已提交
527
{
528
	struct net *net = xs_net(x);
529 530
	int err = -ESRCH;

L
Linus Torvalds 已提交
531 532 533
	if (x->km.state != XFRM_STATE_DEAD) {
		x->km.state = XFRM_STATE_DEAD;
		spin_lock(&xfrm_state_lock);
H
Herbert Xu 已提交
534
		list_del(&x->km.all);
535 536
		hlist_del(&x->bydst);
		hlist_del(&x->bysrc);
537
		if (x->id.spi)
538
			hlist_del(&x->byspi);
539
		net->xfrm.state_num--;
L
Linus Torvalds 已提交
540 541 542 543 544 545
		spin_unlock(&xfrm_state_lock);

		/* All xfrm_state objects are created by xfrm_state_alloc.
		 * The xfrm_state_alloc call gives a reference, and that
		 * is what we are dropping here.
		 */
546
		xfrm_state_put(x);
547
		err = 0;
L
Linus Torvalds 已提交
548
	}
549 550

	return err;
L
Linus Torvalds 已提交
551
}
552
EXPORT_SYMBOL(__xfrm_state_delete);
L
Linus Torvalds 已提交
553

554
int xfrm_state_delete(struct xfrm_state *x)
L
Linus Torvalds 已提交
555
{
556 557
	int err;

L
Linus Torvalds 已提交
558
	spin_lock_bh(&x->lock);
559
	err = __xfrm_state_delete(x);
L
Linus Torvalds 已提交
560
	spin_unlock_bh(&x->lock);
561 562

	return err;
L
Linus Torvalds 已提交
563 564 565
}
EXPORT_SYMBOL(xfrm_state_delete);

566 567
#ifdef CONFIG_SECURITY_NETWORK_XFRM
static inline int
A
Alexey Dobriyan 已提交
568
xfrm_state_flush_secctx_check(struct net *net, u8 proto, struct xfrm_audit *audit_info)
L
Linus Torvalds 已提交
569
{
570 571
	int i, err = 0;

A
Alexey Dobriyan 已提交
572
	for (i = 0; i <= net->xfrm.state_hmask; i++) {
573 574 575
		struct hlist_node *entry;
		struct xfrm_state *x;

A
Alexey Dobriyan 已提交
576
		hlist_for_each_entry(x, entry, net->xfrm.state_bydst+i, bydst) {
577 578
			if (xfrm_id_proto_match(x->id.proto, proto) &&
			   (err = security_xfrm_state_delete(x)) != 0) {
J
Joy Latten 已提交
579 580
				xfrm_audit_state_delete(x, 0,
							audit_info->loginuid,
581
							audit_info->sessionid,
J
Joy Latten 已提交
582
							audit_info->secid);
583 584 585 586 587 588 589 590 591
				return err;
			}
		}
	}

	return err;
}
#else
static inline int
A
Alexey Dobriyan 已提交
592
xfrm_state_flush_secctx_check(struct net *net, u8 proto, struct xfrm_audit *audit_info)
593 594 595 596 597
{
	return 0;
}
#endif

A
Alexey Dobriyan 已提交
598
int xfrm_state_flush(struct net *net, u8 proto, struct xfrm_audit *audit_info)
599 600
{
	int i, err = 0;
L
Linus Torvalds 已提交
601 602

	spin_lock_bh(&xfrm_state_lock);
A
Alexey Dobriyan 已提交
603
	err = xfrm_state_flush_secctx_check(net, proto, audit_info);
604 605 606
	if (err)
		goto out;

A
Alexey Dobriyan 已提交
607
	for (i = 0; i <= net->xfrm.state_hmask; i++) {
608 609
		struct hlist_node *entry;
		struct xfrm_state *x;
L
Linus Torvalds 已提交
610
restart:
A
Alexey Dobriyan 已提交
611
		hlist_for_each_entry(x, entry, net->xfrm.state_bydst+i, bydst) {
L
Linus Torvalds 已提交
612
			if (!xfrm_state_kern(x) &&
613
			    xfrm_id_proto_match(x->id.proto, proto)) {
L
Linus Torvalds 已提交
614 615 616
				xfrm_state_hold(x);
				spin_unlock_bh(&xfrm_state_lock);

J
Joy Latten 已提交
617
				err = xfrm_state_delete(x);
J
Joy Latten 已提交
618 619
				xfrm_audit_state_delete(x, err ? 0 : 1,
							audit_info->loginuid,
620
							audit_info->sessionid,
J
Joy Latten 已提交
621
							audit_info->secid);
L
Linus Torvalds 已提交
622 623 624 625 626 627 628
				xfrm_state_put(x);

				spin_lock_bh(&xfrm_state_lock);
				goto restart;
			}
		}
	}
629 630 631
	err = 0;

out:
L
Linus Torvalds 已提交
632
	spin_unlock_bh(&xfrm_state_lock);
A
Alexey Dobriyan 已提交
633
	wake_up(&net->xfrm.km_waitq);
634
	return err;
L
Linus Torvalds 已提交
635 636 637
}
EXPORT_SYMBOL(xfrm_state_flush);

638
void xfrm_sad_getinfo(struct xfrmk_sadinfo *si)
J
Jamal Hadi Salim 已提交
639 640
{
	spin_lock_bh(&xfrm_state_lock);
641
	si->sadcnt = init_net.xfrm.state_num;
642
	si->sadhcnt = init_net.xfrm.state_hmask;
J
Jamal Hadi Salim 已提交
643 644 645 646 647
	si->sadhmcnt = xfrm_state_hashmax;
	spin_unlock_bh(&xfrm_state_lock);
}
EXPORT_SYMBOL(xfrm_sad_getinfo);

L
Linus Torvalds 已提交
648 649 650 651 652 653 654 655 656 657 658 659 660 661
static int
xfrm_init_tempsel(struct xfrm_state *x, struct flowi *fl,
		  struct xfrm_tmpl *tmpl,
		  xfrm_address_t *daddr, xfrm_address_t *saddr,
		  unsigned short family)
{
	struct xfrm_state_afinfo *afinfo = xfrm_state_get_afinfo(family);
	if (!afinfo)
		return -1;
	afinfo->init_tempsel(x, fl, tmpl, daddr, saddr);
	xfrm_state_put_afinfo(afinfo);
	return 0;
}

662
static struct xfrm_state *__xfrm_state_lookup(struct net *net, xfrm_address_t *daddr, __be32 spi, u8 proto, unsigned short family)
663
{
664
	unsigned int h = xfrm_spi_hash(net, daddr, spi, proto, family);
665
	struct xfrm_state *x;
666
	struct hlist_node *entry;
667

668
	hlist_for_each_entry(x, entry, net->xfrm.state_byspi+h, byspi) {
669 670 671 672 673 674 675 676 677 678 679 680 681 682 683 684
		if (x->props.family != family ||
		    x->id.spi       != spi ||
		    x->id.proto     != proto)
			continue;

		switch (family) {
		case AF_INET:
			if (x->id.daddr.a4 != daddr->a4)
				continue;
			break;
		case AF_INET6:
			if (!ipv6_addr_equal((struct in6_addr *)daddr,
					     (struct in6_addr *)
					     x->id.daddr.a6))
				continue;
			break;
685
		}
686 687 688 689 690 691 692 693

		xfrm_state_hold(x);
		return x;
	}

	return NULL;
}

694
static struct xfrm_state *__xfrm_state_lookup_byaddr(struct net *net, xfrm_address_t *daddr, xfrm_address_t *saddr, u8 proto, unsigned short family)
695
{
696
	unsigned int h = xfrm_src_hash(net, daddr, saddr, family);
697
	struct xfrm_state *x;
698
	struct hlist_node *entry;
699

700
	hlist_for_each_entry(x, entry, net->xfrm.state_bysrc+h, bysrc) {
701 702 703 704 705 706 707 708 709 710 711 712 713 714 715 716 717 718 719
		if (x->props.family != family ||
		    x->id.proto     != proto)
			continue;

		switch (family) {
		case AF_INET:
			if (x->id.daddr.a4 != daddr->a4 ||
			    x->props.saddr.a4 != saddr->a4)
				continue;
			break;
		case AF_INET6:
			if (!ipv6_addr_equal((struct in6_addr *)daddr,
					     (struct in6_addr *)
					     x->id.daddr.a6) ||
			    !ipv6_addr_equal((struct in6_addr *)saddr,
					     (struct in6_addr *)
					     x->props.saddr.a6))
				continue;
			break;
720
		}
721 722 723 724 725 726 727 728 729 730 731

		xfrm_state_hold(x);
		return x;
	}

	return NULL;
}

static inline struct xfrm_state *
__xfrm_state_locate(struct xfrm_state *x, int use_spi, int family)
{
732 733
	struct net *net = xs_net(x);

734
	if (use_spi)
735
		return __xfrm_state_lookup(net, &x->id.daddr, x->id.spi,
736 737
					   x->id.proto, family);
	else
738
		return __xfrm_state_lookup_byaddr(net, &x->id.daddr,
739 740 741 742
						  &x->props.saddr,
						  x->id.proto, family);
}

743
static void xfrm_hash_grow_check(struct net *net, int have_hash_collision)
744 745
{
	if (have_hash_collision &&
746 747 748
	    (net->xfrm.state_hmask + 1) < xfrm_state_hashmax &&
	    net->xfrm.state_num > net->xfrm.state_hmask)
		schedule_work(&net->xfrm.state_hash_work);
749 750
}

L
Linus Torvalds 已提交
751
struct xfrm_state *
752
xfrm_state_find(xfrm_address_t *daddr, xfrm_address_t *saddr,
L
Linus Torvalds 已提交
753 754 755 756
		struct flowi *fl, struct xfrm_tmpl *tmpl,
		struct xfrm_policy *pol, int *err,
		unsigned short family)
{
757
	struct net *net = xp_net(pol);
758
	unsigned int h;
759
	struct hlist_node *entry;
760
	struct xfrm_state *x, *x0, *to_put;
L
Linus Torvalds 已提交
761 762 763
	int acquire_in_progress = 0;
	int error = 0;
	struct xfrm_state *best = NULL;
764

765 766
	to_put = NULL;

L
Linus Torvalds 已提交
767
	spin_lock_bh(&xfrm_state_lock);
768 769
	h = xfrm_dst_hash(net, daddr, saddr, tmpl->reqid, family);
	hlist_for_each_entry(x, entry, net->xfrm.state_bydst+h, bydst) {
L
Linus Torvalds 已提交
770 771
		if (x->props.family == family &&
		    x->props.reqid == tmpl->reqid &&
772
		    !(x->props.flags & XFRM_STATE_WILDRECV) &&
L
Linus Torvalds 已提交
773 774 775 776 777 778 779 780 781 782 783 784 785 786 787 788 789 790
		    xfrm_state_addr_check(x, daddr, saddr, family) &&
		    tmpl->mode == x->props.mode &&
		    tmpl->id.proto == x->id.proto &&
		    (tmpl->id.spi == x->id.spi || !tmpl->id.spi)) {
			/* Resolution logic:
			   1. There is a valid state with matching selector.
			      Done.
			   2. Valid state with inappropriate selector. Skip.

			   Entering area of "sysdeps".

			   3. If state is not valid, selector is temporary,
			      it selects only session which triggered
			      previous resolution. Key manager will do
			      something to install a state with proper
			      selector.
			 */
			if (x->km.state == XFRM_STATE_VALID) {
791
				if ((x->sel.family && !xfrm_selector_match(&x->sel, fl, x->sel.family)) ||
792
				    !security_xfrm_state_pol_flow_match(x, pol, fl))
L
Linus Torvalds 已提交
793 794 795 796 797 798 799 800 801 802
					continue;
				if (!best ||
				    best->km.dying > x->km.dying ||
				    (best->km.dying == x->km.dying &&
				     best->curlft.add_time < x->curlft.add_time))
					best = x;
			} else if (x->km.state == XFRM_STATE_ACQ) {
				acquire_in_progress = 1;
			} else if (x->km.state == XFRM_STATE_ERROR ||
				   x->km.state == XFRM_STATE_EXPIRED) {
803
				if (xfrm_selector_match(&x->sel, fl, x->sel.family) &&
804
				    security_xfrm_state_pol_flow_match(x, pol, fl))
L
Linus Torvalds 已提交
805 806 807 808 809 810 811
					error = -ESRCH;
			}
		}
	}

	x = best;
	if (!x && !error && !acquire_in_progress) {
812
		if (tmpl->id.spi &&
813
		    (x0 = __xfrm_state_lookup(net, daddr, tmpl->id.spi,
814
					      tmpl->id.proto, family)) != NULL) {
815
			to_put = x0;
L
Linus Torvalds 已提交
816 817 818
			error = -EEXIST;
			goto out;
		}
819
		x = xfrm_state_alloc(net);
L
Linus Torvalds 已提交
820 821 822 823 824 825 826 827
		if (x == NULL) {
			error = -ENOMEM;
			goto out;
		}
		/* Initialize temporary selector matching only
		 * to current session. */
		xfrm_init_tempsel(x, fl, tmpl, daddr, saddr, family);

828 829 830
		error = security_xfrm_state_alloc_acquire(x, pol->security, fl->secid);
		if (error) {
			x->km.state = XFRM_STATE_DEAD;
831
			to_put = x;
832 833 834 835
			x = NULL;
			goto out;
		}

L
Linus Torvalds 已提交
836 837
		if (km_query(x, tmpl, pol) == 0) {
			x->km.state = XFRM_STATE_ACQ;
838 839 840 841
			list_add(&x->km.all, &net->xfrm.state_all);
			hlist_add_head(&x->bydst, net->xfrm.state_bydst+h);
			h = xfrm_src_hash(net, daddr, saddr, family);
			hlist_add_head(&x->bysrc, net->xfrm.state_bysrc+h);
L
Linus Torvalds 已提交
842
			if (x->id.spi) {
843 844
				h = xfrm_spi_hash(net, &x->id.daddr, x->id.spi, x->id.proto, family);
				hlist_add_head(&x->byspi, net->xfrm.state_byspi+h);
L
Linus Torvalds 已提交
845
			}
A
Alexey Dobriyan 已提交
846 847
			x->lft.hard_add_expires_seconds = net->xfrm.sysctl_acq_expires;
			x->timer.expires = jiffies + net->xfrm.sysctl_acq_expires*HZ;
L
Linus Torvalds 已提交
848
			add_timer(&x->timer);
849 850
			net->xfrm.state_num++;
			xfrm_hash_grow_check(net, x->bydst.next != NULL);
L
Linus Torvalds 已提交
851 852
		} else {
			x->km.state = XFRM_STATE_DEAD;
853
			to_put = x;
L
Linus Torvalds 已提交
854 855 856 857 858 859 860 861 862 863
			x = NULL;
			error = -ESRCH;
		}
	}
out:
	if (x)
		xfrm_state_hold(x);
	else
		*err = acquire_in_progress ? -EAGAIN : error;
	spin_unlock_bh(&xfrm_state_lock);
864 865
	if (to_put)
		xfrm_state_put(to_put);
L
Linus Torvalds 已提交
866 867 868
	return x;
}

869
struct xfrm_state *
870 871
xfrm_stateonly_find(struct net *net,
		    xfrm_address_t *daddr, xfrm_address_t *saddr,
872 873
		    unsigned short family, u8 mode, u8 proto, u32 reqid)
{
874
	unsigned int h;
875 876 877 878
	struct xfrm_state *rx = NULL, *x = NULL;
	struct hlist_node *entry;

	spin_lock(&xfrm_state_lock);
879 880
	h = xfrm_dst_hash(net, daddr, saddr, reqid, family);
	hlist_for_each_entry(x, entry, net->xfrm.state_bydst+h, bydst) {
881 882 883 884 885 886 887 888 889 890 891 892 893 894 895 896 897 898 899 900 901
		if (x->props.family == family &&
		    x->props.reqid == reqid &&
		    !(x->props.flags & XFRM_STATE_WILDRECV) &&
		    xfrm_state_addr_check(x, daddr, saddr, family) &&
		    mode == x->props.mode &&
		    proto == x->id.proto &&
		    x->km.state == XFRM_STATE_VALID) {
			rx = x;
			break;
		}
	}

	if (rx)
		xfrm_state_hold(rx);
	spin_unlock(&xfrm_state_lock);


	return rx;
}
EXPORT_SYMBOL(xfrm_stateonly_find);

L
Linus Torvalds 已提交
902 903
static void __xfrm_state_insert(struct xfrm_state *x)
{
904
	struct net *net = xs_net(x);
905
	unsigned int h;
L
Linus Torvalds 已提交
906

907 908
	x->genid = ++xfrm_state_genid;

909
	list_add(&x->km.all, &net->xfrm.state_all);
910

911
	h = xfrm_dst_hash(net, &x->id.daddr, &x->props.saddr,
912
			  x->props.reqid, x->props.family);
913
	hlist_add_head(&x->bydst, net->xfrm.state_bydst+h);
L
Linus Torvalds 已提交
914

915 916
	h = xfrm_src_hash(net, &x->id.daddr, &x->props.saddr, x->props.family);
	hlist_add_head(&x->bysrc, net->xfrm.state_bysrc+h);
L
Linus Torvalds 已提交
917

918
	if (x->id.spi) {
919
		h = xfrm_spi_hash(net, &x->id.daddr, x->id.spi, x->id.proto,
920 921
				  x->props.family);

922
		hlist_add_head(&x->byspi, net->xfrm.state_byspi+h);
923 924
	}

925 926 927
	mod_timer(&x->timer, jiffies + HZ);
	if (x->replay_maxage)
		mod_timer(&x->rtimer, jiffies + x->replay_maxage);
928

929
	wake_up(&net->xfrm.km_waitq);
930

931
	net->xfrm.state_num++;
932

933
	xfrm_hash_grow_check(net, x->bydst.next != NULL);
L
Linus Torvalds 已提交
934 935
}

936 937 938
/* xfrm_state_lock is held */
static void __xfrm_state_bump_genids(struct xfrm_state *xnew)
{
939
	struct net *net = xs_net(xnew);
940 941 942 943 944 945
	unsigned short family = xnew->props.family;
	u32 reqid = xnew->props.reqid;
	struct xfrm_state *x;
	struct hlist_node *entry;
	unsigned int h;

946 947
	h = xfrm_dst_hash(net, &xnew->id.daddr, &xnew->props.saddr, reqid, family);
	hlist_for_each_entry(x, entry, net->xfrm.state_bydst+h, bydst) {
948 949
		if (x->props.family	== family &&
		    x->props.reqid	== reqid &&
950 951
		    !xfrm_addr_cmp(&x->id.daddr, &xnew->id.daddr, family) &&
		    !xfrm_addr_cmp(&x->props.saddr, &xnew->props.saddr, family))
952 953 954 955
			x->genid = xfrm_state_genid;
	}
}

L
Linus Torvalds 已提交
956 957 958
void xfrm_state_insert(struct xfrm_state *x)
{
	spin_lock_bh(&xfrm_state_lock);
959
	__xfrm_state_bump_genids(x);
L
Linus Torvalds 已提交
960 961 962 963 964
	__xfrm_state_insert(x);
	spin_unlock_bh(&xfrm_state_lock);
}
EXPORT_SYMBOL(xfrm_state_insert);

965
/* xfrm_state_lock is held */
966
static struct xfrm_state *__find_acq_core(struct net *net, unsigned short family, u8 mode, u32 reqid, u8 proto, xfrm_address_t *daddr, xfrm_address_t *saddr, int create)
967
{
968
	unsigned int h = xfrm_dst_hash(net, daddr, saddr, reqid, family);
969
	struct hlist_node *entry;
970 971
	struct xfrm_state *x;

972
	hlist_for_each_entry(x, entry, net->xfrm.state_bydst+h, bydst) {
973 974 975 976
		if (x->props.reqid  != reqid ||
		    x->props.mode   != mode ||
		    x->props.family != family ||
		    x->km.state     != XFRM_STATE_ACQ ||
977 978
		    x->id.spi       != 0 ||
		    x->id.proto	    != proto)
979 980 981 982 983 984 985 986 987 988 989 990 991 992 993 994
			continue;

		switch (family) {
		case AF_INET:
			if (x->id.daddr.a4    != daddr->a4 ||
			    x->props.saddr.a4 != saddr->a4)
				continue;
			break;
		case AF_INET6:
			if (!ipv6_addr_equal((struct in6_addr *)x->id.daddr.a6,
					     (struct in6_addr *)daddr) ||
			    !ipv6_addr_equal((struct in6_addr *)
					     x->props.saddr.a6,
					     (struct in6_addr *)saddr))
				continue;
			break;
995
		}
996 997 998 999 1000 1001 1002 1003

		xfrm_state_hold(x);
		return x;
	}

	if (!create)
		return NULL;

1004
	x = xfrm_state_alloc(net);
1005 1006 1007 1008 1009 1010 1011 1012 1013 1014 1015 1016 1017 1018 1019 1020 1021 1022 1023 1024 1025 1026 1027
	if (likely(x)) {
		switch (family) {
		case AF_INET:
			x->sel.daddr.a4 = daddr->a4;
			x->sel.saddr.a4 = saddr->a4;
			x->sel.prefixlen_d = 32;
			x->sel.prefixlen_s = 32;
			x->props.saddr.a4 = saddr->a4;
			x->id.daddr.a4 = daddr->a4;
			break;

		case AF_INET6:
			ipv6_addr_copy((struct in6_addr *)x->sel.daddr.a6,
				       (struct in6_addr *)daddr);
			ipv6_addr_copy((struct in6_addr *)x->sel.saddr.a6,
				       (struct in6_addr *)saddr);
			x->sel.prefixlen_d = 128;
			x->sel.prefixlen_s = 128;
			ipv6_addr_copy((struct in6_addr *)x->props.saddr.a6,
				       (struct in6_addr *)saddr);
			ipv6_addr_copy((struct in6_addr *)x->id.daddr.a6,
				       (struct in6_addr *)daddr);
			break;
1028
		}
1029 1030 1031 1032 1033 1034

		x->km.state = XFRM_STATE_ACQ;
		x->id.proto = proto;
		x->props.family = family;
		x->props.mode = mode;
		x->props.reqid = reqid;
A
Alexey Dobriyan 已提交
1035
		x->lft.hard_add_expires_seconds = net->xfrm.sysctl_acq_expires;
1036
		xfrm_state_hold(x);
A
Alexey Dobriyan 已提交
1037
		x->timer.expires = jiffies + net->xfrm.sysctl_acq_expires*HZ;
1038
		add_timer(&x->timer);
1039 1040 1041 1042
		list_add(&x->km.all, &net->xfrm.state_all);
		hlist_add_head(&x->bydst, net->xfrm.state_bydst+h);
		h = xfrm_src_hash(net, daddr, saddr, family);
		hlist_add_head(&x->bysrc, net->xfrm.state_bysrc+h);
1043

1044
		net->xfrm.state_num++;
1045

1046
		xfrm_hash_grow_check(net, x->bydst.next != NULL);
1047 1048 1049 1050 1051
	}

	return x;
}

1052
static struct xfrm_state *__xfrm_find_acq_byseq(struct net *net, u32 seq);
L
Linus Torvalds 已提交
1053 1054 1055

int xfrm_state_add(struct xfrm_state *x)
{
1056
	struct net *net = xs_net(x);
1057
	struct xfrm_state *x1, *to_put;
L
Linus Torvalds 已提交
1058 1059
	int family;
	int err;
1060
	int use_spi = xfrm_id_proto_match(x->id.proto, IPSEC_PROTO_ANY);
L
Linus Torvalds 已提交
1061 1062 1063

	family = x->props.family;

1064 1065
	to_put = NULL;

L
Linus Torvalds 已提交
1066 1067
	spin_lock_bh(&xfrm_state_lock);

1068
	x1 = __xfrm_state_locate(x, use_spi, family);
L
Linus Torvalds 已提交
1069
	if (x1) {
1070
		to_put = x1;
L
Linus Torvalds 已提交
1071 1072 1073 1074 1075
		x1 = NULL;
		err = -EEXIST;
		goto out;
	}

1076
	if (use_spi && x->km.seq) {
1077
		x1 = __xfrm_find_acq_byseq(net, x->km.seq);
1078 1079
		if (x1 && ((x1->id.proto != x->id.proto) ||
		    xfrm_addr_cmp(&x1->id.daddr, &x->id.daddr, family))) {
1080
			to_put = x1;
L
Linus Torvalds 已提交
1081 1082 1083 1084
			x1 = NULL;
		}
	}

1085
	if (use_spi && !x1)
1086
		x1 = __find_acq_core(net, family, x->props.mode, x->props.reqid,
1087 1088
				     x->id.proto,
				     &x->id.daddr, &x->props.saddr, 0);
L
Linus Torvalds 已提交
1089

1090
	__xfrm_state_bump_genids(x);
L
Linus Torvalds 已提交
1091 1092 1093 1094 1095 1096 1097 1098 1099 1100 1101
	__xfrm_state_insert(x);
	err = 0;

out:
	spin_unlock_bh(&xfrm_state_lock);

	if (x1) {
		xfrm_state_delete(x1);
		xfrm_state_put(x1);
	}

1102 1103 1104
	if (to_put)
		xfrm_state_put(to_put);

L
Linus Torvalds 已提交
1105 1106 1107 1108
	return err;
}
EXPORT_SYMBOL(xfrm_state_add);

1109
#ifdef CONFIG_XFRM_MIGRATE
1110
static struct xfrm_state *xfrm_state_clone(struct xfrm_state *orig, int *errp)
1111
{
1112
	struct net *net = xs_net(orig);
1113
	int err = -ENOMEM;
1114
	struct xfrm_state *x = xfrm_state_alloc(net);
1115 1116 1117 1118 1119 1120 1121 1122 1123 1124 1125 1126 1127 1128 1129 1130 1131 1132 1133 1134 1135 1136 1137 1138 1139 1140 1141 1142 1143 1144 1145 1146 1147
	if (!x)
		goto error;

	memcpy(&x->id, &orig->id, sizeof(x->id));
	memcpy(&x->sel, &orig->sel, sizeof(x->sel));
	memcpy(&x->lft, &orig->lft, sizeof(x->lft));
	x->props.mode = orig->props.mode;
	x->props.replay_window = orig->props.replay_window;
	x->props.reqid = orig->props.reqid;
	x->props.family = orig->props.family;
	x->props.saddr = orig->props.saddr;

	if (orig->aalg) {
		x->aalg = xfrm_algo_clone(orig->aalg);
		if (!x->aalg)
			goto error;
	}
	x->props.aalgo = orig->props.aalgo;

	if (orig->ealg) {
		x->ealg = xfrm_algo_clone(orig->ealg);
		if (!x->ealg)
			goto error;
	}
	x->props.ealgo = orig->props.ealgo;

	if (orig->calg) {
		x->calg = xfrm_algo_clone(orig->calg);
		if (!x->calg)
			goto error;
	}
	x->props.calgo = orig->props.calgo;

1148
	if (orig->encap) {
1149 1150 1151 1152 1153 1154 1155 1156 1157 1158 1159 1160 1161 1162 1163 1164 1165 1166 1167 1168 1169 1170 1171 1172 1173 1174 1175 1176 1177 1178 1179 1180 1181 1182 1183 1184 1185 1186 1187 1188 1189 1190 1191 1192 1193 1194
		x->encap = kmemdup(orig->encap, sizeof(*x->encap), GFP_KERNEL);
		if (!x->encap)
			goto error;
	}

	if (orig->coaddr) {
		x->coaddr = kmemdup(orig->coaddr, sizeof(*x->coaddr),
				    GFP_KERNEL);
		if (!x->coaddr)
			goto error;
	}

	err = xfrm_init_state(x);
	if (err)
		goto error;

	x->props.flags = orig->props.flags;

	x->curlft.add_time = orig->curlft.add_time;
	x->km.state = orig->km.state;
	x->km.seq = orig->km.seq;

	return x;

 error:
	if (errp)
		*errp = err;
	if (x) {
		kfree(x->aalg);
		kfree(x->ealg);
		kfree(x->calg);
		kfree(x->encap);
		kfree(x->coaddr);
	}
	kfree(x);
	return NULL;
}

/* xfrm_state_lock is held */
struct xfrm_state * xfrm_migrate_state_find(struct xfrm_migrate *m)
{
	unsigned int h;
	struct xfrm_state *x;
	struct hlist_node *entry;

	if (m->reqid) {
1195
		h = xfrm_dst_hash(&init_net, &m->old_daddr, &m->old_saddr,
1196
				  m->reqid, m->old_family);
1197
		hlist_for_each_entry(x, entry, init_net.xfrm.state_bydst+h, bydst) {
1198 1199 1200 1201 1202 1203 1204 1205 1206 1207 1208 1209 1210 1211
			if (x->props.mode != m->mode ||
			    x->id.proto != m->proto)
				continue;
			if (m->reqid && x->props.reqid != m->reqid)
				continue;
			if (xfrm_addr_cmp(&x->id.daddr, &m->old_daddr,
					  m->old_family) ||
			    xfrm_addr_cmp(&x->props.saddr, &m->old_saddr,
					  m->old_family))
				continue;
			xfrm_state_hold(x);
			return x;
		}
	} else {
1212
		h = xfrm_src_hash(&init_net, &m->old_daddr, &m->old_saddr,
1213
				  m->old_family);
1214
		hlist_for_each_entry(x, entry, init_net.xfrm.state_bysrc+h, bysrc) {
1215 1216 1217 1218 1219 1220 1221 1222 1223 1224 1225 1226 1227
			if (x->props.mode != m->mode ||
			    x->id.proto != m->proto)
				continue;
			if (xfrm_addr_cmp(&x->id.daddr, &m->old_daddr,
					  m->old_family) ||
			    xfrm_addr_cmp(&x->props.saddr, &m->old_saddr,
					  m->old_family))
				continue;
			xfrm_state_hold(x);
			return x;
		}
	}

1228
	return NULL;
1229 1230 1231 1232 1233 1234 1235 1236 1237 1238 1239 1240 1241 1242 1243 1244 1245 1246 1247 1248 1249 1250 1251 1252 1253 1254 1255 1256 1257 1258 1259 1260 1261 1262
}
EXPORT_SYMBOL(xfrm_migrate_state_find);

struct xfrm_state * xfrm_state_migrate(struct xfrm_state *x,
				       struct xfrm_migrate *m)
{
	struct xfrm_state *xc;
	int err;

	xc = xfrm_state_clone(x, &err);
	if (!xc)
		return NULL;

	memcpy(&xc->id.daddr, &m->new_daddr, sizeof(xc->id.daddr));
	memcpy(&xc->props.saddr, &m->new_saddr, sizeof(xc->props.saddr));

	/* add state */
	if (!xfrm_addr_cmp(&x->id.daddr, &m->new_daddr, m->new_family)) {
		/* a care is needed when the destination address of the
		   state is to be updated as it is a part of triplet */
		xfrm_state_insert(xc);
	} else {
		if ((err = xfrm_state_add(xc)) < 0)
			goto error;
	}

	return xc;
error:
	kfree(xc);
	return NULL;
}
EXPORT_SYMBOL(xfrm_state_migrate);
#endif

L
Linus Torvalds 已提交
1263 1264
int xfrm_state_update(struct xfrm_state *x)
{
1265
	struct xfrm_state *x1, *to_put;
L
Linus Torvalds 已提交
1266
	int err;
1267
	int use_spi = xfrm_id_proto_match(x->id.proto, IPSEC_PROTO_ANY);
L
Linus Torvalds 已提交
1268

1269 1270
	to_put = NULL;

L
Linus Torvalds 已提交
1271
	spin_lock_bh(&xfrm_state_lock);
1272
	x1 = __xfrm_state_locate(x, use_spi, x->props.family);
L
Linus Torvalds 已提交
1273 1274 1275 1276 1277 1278

	err = -ESRCH;
	if (!x1)
		goto out;

	if (xfrm_state_kern(x1)) {
1279
		to_put = x1;
L
Linus Torvalds 已提交
1280 1281 1282 1283 1284 1285 1286 1287 1288 1289 1290 1291 1292
		err = -EEXIST;
		goto out;
	}

	if (x1->km.state == XFRM_STATE_ACQ) {
		__xfrm_state_insert(x);
		x = NULL;
	}
	err = 0;

out:
	spin_unlock_bh(&xfrm_state_lock);

1293 1294 1295
	if (to_put)
		xfrm_state_put(to_put);

L
Linus Torvalds 已提交
1296 1297 1298 1299 1300 1301 1302 1303 1304 1305 1306 1307 1308 1309
	if (err)
		return err;

	if (!x) {
		xfrm_state_delete(x1);
		xfrm_state_put(x1);
		return 0;
	}

	err = -EINVAL;
	spin_lock_bh(&x1->lock);
	if (likely(x1->km.state == XFRM_STATE_VALID)) {
		if (x->encap && x1->encap)
			memcpy(x1->encap, x->encap, sizeof(*x1->encap));
1310 1311 1312 1313 1314
		if (x->coaddr && x1->coaddr) {
			memcpy(x1->coaddr, x->coaddr, sizeof(*x1->coaddr));
		}
		if (!use_spi && memcmp(&x1->sel, &x->sel, sizeof(x1->sel)))
			memcpy(&x1->sel, &x->sel, sizeof(x1->sel));
L
Linus Torvalds 已提交
1315 1316 1317
		memcpy(&x1->lft, &x->lft, sizeof(x1->lft));
		x1->km.dying = 0;

1318
		mod_timer(&x1->timer, jiffies + HZ);
L
Linus Torvalds 已提交
1319 1320 1321 1322 1323 1324 1325 1326 1327 1328 1329 1330 1331 1332 1333 1334
		if (x1->curlft.use_time)
			xfrm_state_check_expire(x1);

		err = 0;
	}
	spin_unlock_bh(&x1->lock);

	xfrm_state_put(x1);

	return err;
}
EXPORT_SYMBOL(xfrm_state_update);

int xfrm_state_check_expire(struct xfrm_state *x)
{
	if (!x->curlft.use_time)
1335
		x->curlft.use_time = get_seconds();
L
Linus Torvalds 已提交
1336 1337 1338 1339 1340 1341

	if (x->km.state != XFRM_STATE_VALID)
		return -EINVAL;

	if (x->curlft.bytes >= x->lft.hard_byte_limit ||
	    x->curlft.packets >= x->lft.hard_packet_limit) {
1342
		x->km.state = XFRM_STATE_EXPIRED;
1343
		mod_timer(&x->timer, jiffies);
L
Linus Torvalds 已提交
1344 1345 1346 1347 1348
		return -EINVAL;
	}

	if (!x->km.dying &&
	    (x->curlft.bytes >= x->lft.soft_byte_limit ||
1349 1350
	     x->curlft.packets >= x->lft.soft_packet_limit)) {
		x->km.dying = 1;
1351
		km_state_expired(x, 0, 0);
1352
	}
L
Linus Torvalds 已提交
1353 1354 1355 1356 1357
	return 0;
}
EXPORT_SYMBOL(xfrm_state_check_expire);

struct xfrm_state *
1358
xfrm_state_lookup(struct net *net, xfrm_address_t *daddr, __be32 spi, u8 proto,
L
Linus Torvalds 已提交
1359 1360 1361 1362 1363
		  unsigned short family)
{
	struct xfrm_state *x;

	spin_lock_bh(&xfrm_state_lock);
1364
	x = __xfrm_state_lookup(net, daddr, spi, proto, family);
L
Linus Torvalds 已提交
1365 1366 1367 1368 1369 1370
	spin_unlock_bh(&xfrm_state_lock);
	return x;
}
EXPORT_SYMBOL(xfrm_state_lookup);

struct xfrm_state *
1371 1372
xfrm_state_lookup_byaddr(struct net *net,
			 xfrm_address_t *daddr, xfrm_address_t *saddr,
1373 1374 1375 1376 1377
			 u8 proto, unsigned short family)
{
	struct xfrm_state *x;

	spin_lock_bh(&xfrm_state_lock);
1378
	x = __xfrm_state_lookup_byaddr(net, daddr, saddr, proto, family);
1379 1380 1381 1382 1383 1384
	spin_unlock_bh(&xfrm_state_lock);
	return x;
}
EXPORT_SYMBOL(xfrm_state_lookup_byaddr);

struct xfrm_state *
1385
xfrm_find_acq(struct net *net, u8 mode, u32 reqid, u8 proto,
1386
	      xfrm_address_t *daddr, xfrm_address_t *saddr,
L
Linus Torvalds 已提交
1387 1388 1389 1390 1391
	      int create, unsigned short family)
{
	struct xfrm_state *x;

	spin_lock_bh(&xfrm_state_lock);
1392
	x = __find_acq_core(net, family, mode, reqid, proto, daddr, saddr, create);
L
Linus Torvalds 已提交
1393
	spin_unlock_bh(&xfrm_state_lock);
1394

L
Linus Torvalds 已提交
1395 1396 1397 1398
	return x;
}
EXPORT_SYMBOL(xfrm_find_acq);

1399 1400 1401 1402 1403 1404 1405 1406 1407 1408 1409 1410 1411 1412 1413 1414 1415 1416 1417 1418 1419 1420 1421 1422 1423 1424 1425 1426 1427 1428 1429 1430 1431 1432 1433 1434 1435 1436
#ifdef CONFIG_XFRM_SUB_POLICY
int
xfrm_tmpl_sort(struct xfrm_tmpl **dst, struct xfrm_tmpl **src, int n,
	       unsigned short family)
{
	int err = 0;
	struct xfrm_state_afinfo *afinfo = xfrm_state_get_afinfo(family);
	if (!afinfo)
		return -EAFNOSUPPORT;

	spin_lock_bh(&xfrm_state_lock);
	if (afinfo->tmpl_sort)
		err = afinfo->tmpl_sort(dst, src, n);
	spin_unlock_bh(&xfrm_state_lock);
	xfrm_state_put_afinfo(afinfo);
	return err;
}
EXPORT_SYMBOL(xfrm_tmpl_sort);

int
xfrm_state_sort(struct xfrm_state **dst, struct xfrm_state **src, int n,
		unsigned short family)
{
	int err = 0;
	struct xfrm_state_afinfo *afinfo = xfrm_state_get_afinfo(family);
	if (!afinfo)
		return -EAFNOSUPPORT;

	spin_lock_bh(&xfrm_state_lock);
	if (afinfo->state_sort)
		err = afinfo->state_sort(dst, src, n);
	spin_unlock_bh(&xfrm_state_lock);
	xfrm_state_put_afinfo(afinfo);
	return err;
}
EXPORT_SYMBOL(xfrm_state_sort);
#endif

L
Linus Torvalds 已提交
1437 1438
/* Silly enough, but I'm lazy to build resolution list */

1439
static struct xfrm_state *__xfrm_find_acq_byseq(struct net *net, u32 seq)
L
Linus Torvalds 已提交
1440 1441 1442
{
	int i;

1443
	for (i = 0; i <= net->xfrm.state_hmask; i++) {
1444 1445 1446
		struct hlist_node *entry;
		struct xfrm_state *x;

1447
		hlist_for_each_entry(x, entry, net->xfrm.state_bydst+i, bydst) {
1448 1449
			if (x->km.seq == seq &&
			    x->km.state == XFRM_STATE_ACQ) {
L
Linus Torvalds 已提交
1450 1451 1452 1453 1454 1455 1456 1457
				xfrm_state_hold(x);
				return x;
			}
		}
	}
	return NULL;
}

1458
struct xfrm_state *xfrm_find_acq_byseq(struct net *net, u32 seq)
L
Linus Torvalds 已提交
1459 1460 1461 1462
{
	struct xfrm_state *x;

	spin_lock_bh(&xfrm_state_lock);
1463
	x = __xfrm_find_acq_byseq(net, seq);
L
Linus Torvalds 已提交
1464 1465 1466 1467 1468 1469 1470 1471 1472 1473 1474 1475 1476 1477 1478 1479 1480 1481
	spin_unlock_bh(&xfrm_state_lock);
	return x;
}
EXPORT_SYMBOL(xfrm_find_acq_byseq);

u32 xfrm_get_acqseq(void)
{
	u32 res;
	static u32 acqseq;
	static DEFINE_SPINLOCK(acqseq_lock);

	spin_lock_bh(&acqseq_lock);
	res = (++acqseq ? : ++acqseq);
	spin_unlock_bh(&acqseq_lock);
	return res;
}
EXPORT_SYMBOL(xfrm_get_acqseq);

1482
int xfrm_alloc_spi(struct xfrm_state *x, u32 low, u32 high)
L
Linus Torvalds 已提交
1483
{
1484
	struct net *net = xs_net(x);
1485
	unsigned int h;
L
Linus Torvalds 已提交
1486
	struct xfrm_state *x0;
1487 1488 1489
	int err = -ENOENT;
	__be32 minspi = htonl(low);
	__be32 maxspi = htonl(high);
L
Linus Torvalds 已提交
1490

1491 1492 1493 1494 1495
	spin_lock_bh(&x->lock);
	if (x->km.state == XFRM_STATE_DEAD)
		goto unlock;

	err = 0;
L
Linus Torvalds 已提交
1496
	if (x->id.spi)
1497 1498 1499
		goto unlock;

	err = -ENOENT;
L
Linus Torvalds 已提交
1500 1501

	if (minspi == maxspi) {
1502
		x0 = xfrm_state_lookup(net, &x->id.daddr, minspi, x->id.proto, x->props.family);
L
Linus Torvalds 已提交
1503 1504
		if (x0) {
			xfrm_state_put(x0);
1505
			goto unlock;
L
Linus Torvalds 已提交
1506 1507 1508 1509
		}
		x->id.spi = minspi;
	} else {
		u32 spi = 0;
A
Al Viro 已提交
1510 1511
		for (h=0; h<high-low+1; h++) {
			spi = low + net_random()%(high-low+1);
1512
			x0 = xfrm_state_lookup(net, &x->id.daddr, htonl(spi), x->id.proto, x->props.family);
L
Linus Torvalds 已提交
1513 1514 1515 1516 1517 1518 1519 1520 1521
			if (x0 == NULL) {
				x->id.spi = htonl(spi);
				break;
			}
			xfrm_state_put(x0);
		}
	}
	if (x->id.spi) {
		spin_lock_bh(&xfrm_state_lock);
1522 1523
		h = xfrm_spi_hash(net, &x->id.daddr, x->id.spi, x->id.proto, x->props.family);
		hlist_add_head(&x->byspi, net->xfrm.state_byspi+h);
L
Linus Torvalds 已提交
1524
		spin_unlock_bh(&xfrm_state_lock);
1525 1526

		err = 0;
L
Linus Torvalds 已提交
1527
	}
1528 1529 1530 1531 1532

unlock:
	spin_unlock_bh(&x->lock);

	return err;
L
Linus Torvalds 已提交
1533 1534 1535
}
EXPORT_SYMBOL(xfrm_alloc_spi);

1536
int xfrm_state_walk(struct net *net, struct xfrm_state_walk *walk,
1537
		    int (*func)(struct xfrm_state *, int, void*),
L
Linus Torvalds 已提交
1538 1539
		    void *data)
{
H
Herbert Xu 已提交
1540 1541
	struct xfrm_state *state;
	struct xfrm_state_walk *x;
L
Linus Torvalds 已提交
1542 1543
	int err = 0;

H
Herbert Xu 已提交
1544
	if (walk->seq != 0 && list_empty(&walk->all))
1545 1546
		return 0;

L
Linus Torvalds 已提交
1547
	spin_lock_bh(&xfrm_state_lock);
H
Herbert Xu 已提交
1548
	if (list_empty(&walk->all))
1549
		x = list_first_entry(&net->xfrm.state_all, struct xfrm_state_walk, all);
H
Herbert Xu 已提交
1550 1551
	else
		x = list_entry(&walk->all, struct xfrm_state_walk, all);
1552
	list_for_each_entry_from(x, &net->xfrm.state_all, all) {
H
Herbert Xu 已提交
1553
		if (x->state == XFRM_STATE_DEAD)
1554
			continue;
H
Herbert Xu 已提交
1555 1556
		state = container_of(x, struct xfrm_state, km);
		if (!xfrm_id_proto_match(state->id.proto, walk->proto))
1557
			continue;
H
Herbert Xu 已提交
1558 1559 1560 1561
		err = func(state, walk->seq, data);
		if (err) {
			list_move_tail(&walk->all, &x->all);
			goto out;
L
Linus Torvalds 已提交
1562
		}
H
Herbert Xu 已提交
1563
		walk->seq++;
L
Linus Torvalds 已提交
1564
	}
H
Herbert Xu 已提交
1565
	if (walk->seq == 0) {
L
Linus Torvalds 已提交
1566 1567 1568
		err = -ENOENT;
		goto out;
	}
H
Herbert Xu 已提交
1569
	list_del_init(&walk->all);
L
Linus Torvalds 已提交
1570 1571 1572 1573 1574 1575
out:
	spin_unlock_bh(&xfrm_state_lock);
	return err;
}
EXPORT_SYMBOL(xfrm_state_walk);

H
Herbert Xu 已提交
1576 1577
void xfrm_state_walk_init(struct xfrm_state_walk *walk, u8 proto)
{
H
Herbert Xu 已提交
1578
	INIT_LIST_HEAD(&walk->all);
H
Herbert Xu 已提交
1579
	walk->proto = proto;
H
Herbert Xu 已提交
1580 1581
	walk->state = XFRM_STATE_DEAD;
	walk->seq = 0;
H
Herbert Xu 已提交
1582 1583 1584
}
EXPORT_SYMBOL(xfrm_state_walk_init);

1585 1586
void xfrm_state_walk_done(struct xfrm_state_walk *walk)
{
H
Herbert Xu 已提交
1587
	if (list_empty(&walk->all))
H
Herbert Xu 已提交
1588 1589
		return;

H
Herbert Xu 已提交
1590 1591 1592
	spin_lock_bh(&xfrm_state_lock);
	list_del(&walk->all);
	spin_lock_bh(&xfrm_state_lock);
1593 1594 1595
}
EXPORT_SYMBOL(xfrm_state_walk_done);

1596 1597 1598 1599 1600 1601 1602 1603 1604 1605 1606 1607 1608 1609 1610 1611 1612 1613

void xfrm_replay_notify(struct xfrm_state *x, int event)
{
	struct km_event c;
	/* we send notify messages in case
	 *  1. we updated on of the sequence numbers, and the seqno difference
	 *     is at least x->replay_maxdiff, in this case we also update the
	 *     timeout of our timer function
	 *  2. if x->replay_maxage has elapsed since last update,
	 *     and there were changes
	 *
	 *  The state structure must be locked!
	 */

	switch (event) {
	case XFRM_REPLAY_UPDATE:
		if (x->replay_maxdiff &&
		    (x->replay.seq - x->preplay.seq < x->replay_maxdiff) &&
J
Jamal Hadi Salim 已提交
1614 1615 1616 1617 1618 1619
		    (x->replay.oseq - x->preplay.oseq < x->replay_maxdiff)) {
			if (x->xflags & XFRM_TIME_DEFER)
				event = XFRM_REPLAY_TIMEOUT;
			else
				return;
		}
1620 1621 1622 1623 1624 1625

		break;

	case XFRM_REPLAY_TIMEOUT:
		if ((x->replay.seq == x->preplay.seq) &&
		    (x->replay.bitmap == x->preplay.bitmap) &&
J
Jamal Hadi Salim 已提交
1626 1627
		    (x->replay.oseq == x->preplay.oseq)) {
			x->xflags |= XFRM_TIME_DEFER;
1628
			return;
J
Jamal Hadi Salim 已提交
1629
		}
1630 1631 1632 1633 1634 1635 1636 1637 1638 1639

		break;
	}

	memcpy(&x->preplay, &x->replay, sizeof(struct xfrm_replay_state));
	c.event = XFRM_MSG_NEWAE;
	c.data.aevent = event;
	km_state_notify(x, &c);

	if (x->replay_maxage &&
1640
	    !mod_timer(&x->rtimer, jiffies + x->replay_maxage))
J
Jamal Hadi Salim 已提交
1641
		x->xflags &= ~XFRM_TIME_DEFER;
1642 1643 1644 1645 1646 1647 1648 1649
}

static void xfrm_replay_timer_handler(unsigned long data)
{
	struct xfrm_state *x = (struct xfrm_state*)data;

	spin_lock(&x->lock);

J
Jamal Hadi Salim 已提交
1650
	if (x->km.state == XFRM_STATE_VALID) {
1651
		if (xfrm_aevent_is_on(xs_net(x)))
J
Jamal Hadi Salim 已提交
1652 1653 1654 1655
			xfrm_replay_notify(x, XFRM_REPLAY_TIMEOUT);
		else
			x->xflags |= XFRM_TIME_DEFER;
	}
1656 1657 1658 1659

	spin_unlock(&x->lock);
}

P
Paul Moore 已提交
1660 1661
int xfrm_replay_check(struct xfrm_state *x,
		      struct sk_buff *skb, __be32 net_seq)
L
Linus Torvalds 已提交
1662 1663
{
	u32 diff;
1664
	u32 seq = ntohl(net_seq);
L
Linus Torvalds 已提交
1665 1666

	if (unlikely(seq == 0))
P
Paul Moore 已提交
1667
		goto err;
L
Linus Torvalds 已提交
1668 1669 1670 1671 1672

	if (likely(seq > x->replay.seq))
		return 0;

	diff = x->replay.seq - seq;
1673 1674
	if (diff >= min_t(unsigned int, x->props.replay_window,
			  sizeof(x->replay.bitmap) * 8)) {
L
Linus Torvalds 已提交
1675
		x->stats.replay_window++;
P
Paul Moore 已提交
1676
		goto err;
L
Linus Torvalds 已提交
1677 1678 1679 1680
	}

	if (x->replay.bitmap & (1U << diff)) {
		x->stats.replay++;
P
Paul Moore 已提交
1681
		goto err;
L
Linus Torvalds 已提交
1682 1683
	}
	return 0;
P
Paul Moore 已提交
1684 1685 1686 1687

err:
	xfrm_audit_state_replay(x, skb, net_seq);
	return -EINVAL;
L
Linus Torvalds 已提交
1688 1689
}

1690
void xfrm_replay_advance(struct xfrm_state *x, __be32 net_seq)
L
Linus Torvalds 已提交
1691 1692
{
	u32 diff;
1693
	u32 seq = ntohl(net_seq);
L
Linus Torvalds 已提交
1694 1695 1696 1697 1698 1699 1700 1701 1702 1703 1704 1705

	if (seq > x->replay.seq) {
		diff = seq - x->replay.seq;
		if (diff < x->props.replay_window)
			x->replay.bitmap = ((x->replay.bitmap) << diff) | 1;
		else
			x->replay.bitmap = 1;
		x->replay.seq = seq;
	} else {
		diff = x->replay.seq - seq;
		x->replay.bitmap |= (1U << diff);
	}
1706

1707
	if (xfrm_aevent_is_on(xs_net(x)))
1708
		xfrm_replay_notify(x, XFRM_REPLAY_UPDATE);
L
Linus Torvalds 已提交
1709 1710
}

1711
static LIST_HEAD(xfrm_km_list);
L
Linus Torvalds 已提交
1712 1713
static DEFINE_RWLOCK(xfrm_km_lock);

1714
void km_policy_notify(struct xfrm_policy *xp, int dir, struct km_event *c)
L
Linus Torvalds 已提交
1715 1716 1717
{
	struct xfrm_mgr *km;

1718 1719 1720 1721 1722 1723
	read_lock(&xfrm_km_lock);
	list_for_each_entry(km, &xfrm_km_list, list)
		if (km->notify_policy)
			km->notify_policy(xp, dir, c);
	read_unlock(&xfrm_km_lock);
}
L
Linus Torvalds 已提交
1724

1725 1726 1727
void km_state_notify(struct xfrm_state *x, struct km_event *c)
{
	struct xfrm_mgr *km;
L
Linus Torvalds 已提交
1728 1729
	read_lock(&xfrm_km_lock);
	list_for_each_entry(km, &xfrm_km_list, list)
1730 1731
		if (km->notify)
			km->notify(x, c);
L
Linus Torvalds 已提交
1732
	read_unlock(&xfrm_km_lock);
1733 1734 1735 1736 1737
}

EXPORT_SYMBOL(km_policy_notify);
EXPORT_SYMBOL(km_state_notify);

1738
void km_state_expired(struct xfrm_state *x, int hard, u32 pid)
1739
{
1740
	struct net *net = xs_net(x);
1741 1742
	struct km_event c;

1743
	c.data.hard = hard;
1744
	c.pid = pid;
1745
	c.event = XFRM_MSG_EXPIRE;
1746
	km_state_notify(x, &c);
L
Linus Torvalds 已提交
1747 1748

	if (hard)
1749
		wake_up(&net->xfrm.km_waitq);
L
Linus Torvalds 已提交
1750 1751
}

1752
EXPORT_SYMBOL(km_state_expired);
1753 1754 1755 1756
/*
 * We send to all registered managers regardless of failure
 * We are happy with one success
*/
1757
int km_query(struct xfrm_state *x, struct xfrm_tmpl *t, struct xfrm_policy *pol)
L
Linus Torvalds 已提交
1758
{
1759
	int err = -EINVAL, acqret;
L
Linus Torvalds 已提交
1760 1761 1762 1763
	struct xfrm_mgr *km;

	read_lock(&xfrm_km_lock);
	list_for_each_entry(km, &xfrm_km_list, list) {
1764 1765 1766
		acqret = km->acquire(x, t, pol, XFRM_POLICY_OUT);
		if (!acqret)
			err = acqret;
L
Linus Torvalds 已提交
1767 1768 1769 1770
	}
	read_unlock(&xfrm_km_lock);
	return err;
}
1771
EXPORT_SYMBOL(km_query);
L
Linus Torvalds 已提交
1772

A
Al Viro 已提交
1773
int km_new_mapping(struct xfrm_state *x, xfrm_address_t *ipaddr, __be16 sport)
L
Linus Torvalds 已提交
1774 1775 1776 1777 1778 1779 1780 1781 1782 1783 1784 1785 1786 1787 1788 1789
{
	int err = -EINVAL;
	struct xfrm_mgr *km;

	read_lock(&xfrm_km_lock);
	list_for_each_entry(km, &xfrm_km_list, list) {
		if (km->new_mapping)
			err = km->new_mapping(x, ipaddr, sport);
		if (!err)
			break;
	}
	read_unlock(&xfrm_km_lock);
	return err;
}
EXPORT_SYMBOL(km_new_mapping);

1790
void km_policy_expired(struct xfrm_policy *pol, int dir, int hard, u32 pid)
L
Linus Torvalds 已提交
1791
{
1792
	struct net *net = xp_net(pol);
1793
	struct km_event c;
L
Linus Torvalds 已提交
1794

1795
	c.data.hard = hard;
1796
	c.pid = pid;
1797
	c.event = XFRM_MSG_POLEXPIRE;
1798
	km_policy_notify(pol, dir, &c);
L
Linus Torvalds 已提交
1799 1800

	if (hard)
1801
		wake_up(&net->xfrm.km_waitq);
L
Linus Torvalds 已提交
1802
}
1803
EXPORT_SYMBOL(km_policy_expired);
L
Linus Torvalds 已提交
1804

1805
#ifdef CONFIG_XFRM_MIGRATE
1806
int km_migrate(struct xfrm_selector *sel, u8 dir, u8 type,
1807 1808
	       struct xfrm_migrate *m, int num_migrate,
	       struct xfrm_kmaddress *k)
1809 1810 1811 1812 1813 1814 1815 1816
{
	int err = -EINVAL;
	int ret;
	struct xfrm_mgr *km;

	read_lock(&xfrm_km_lock);
	list_for_each_entry(km, &xfrm_km_list, list) {
		if (km->migrate) {
1817
			ret = km->migrate(sel, dir, type, m, num_migrate, k);
1818 1819 1820 1821 1822 1823 1824 1825
			if (!ret)
				err = ret;
		}
	}
	read_unlock(&xfrm_km_lock);
	return err;
}
EXPORT_SYMBOL(km_migrate);
1826
#endif
1827

1828
int km_report(struct net *net, u8 proto, struct xfrm_selector *sel, xfrm_address_t *addr)
1829 1830 1831 1832 1833 1834 1835 1836
{
	int err = -EINVAL;
	int ret;
	struct xfrm_mgr *km;

	read_lock(&xfrm_km_lock);
	list_for_each_entry(km, &xfrm_km_list, list) {
		if (km->report) {
1837
			ret = km->report(net, proto, sel, addr);
1838 1839 1840 1841 1842 1843 1844 1845 1846
			if (!ret)
				err = ret;
		}
	}
	read_unlock(&xfrm_km_lock);
	return err;
}
EXPORT_SYMBOL(km_report);

L
Linus Torvalds 已提交
1847 1848 1849 1850 1851 1852 1853 1854 1855 1856 1857 1858 1859 1860 1861 1862 1863 1864 1865 1866 1867
int xfrm_user_policy(struct sock *sk, int optname, u8 __user *optval, int optlen)
{
	int err;
	u8 *data;
	struct xfrm_mgr *km;
	struct xfrm_policy *pol = NULL;

	if (optlen <= 0 || optlen > PAGE_SIZE)
		return -EMSGSIZE;

	data = kmalloc(optlen, GFP_KERNEL);
	if (!data)
		return -ENOMEM;

	err = -EFAULT;
	if (copy_from_user(data, optval, optlen))
		goto out;

	err = -EINVAL;
	read_lock(&xfrm_km_lock);
	list_for_each_entry(km, &xfrm_km_list, list) {
1868
		pol = km->compile_policy(sk, optname, data,
L
Linus Torvalds 已提交
1869 1870 1871 1872 1873 1874 1875 1876 1877 1878 1879 1880 1881 1882 1883 1884 1885 1886 1887 1888 1889 1890 1891 1892 1893 1894 1895 1896 1897 1898 1899 1900 1901 1902 1903 1904 1905 1906 1907 1908 1909 1910 1911
					 optlen, &err);
		if (err >= 0)
			break;
	}
	read_unlock(&xfrm_km_lock);

	if (err >= 0) {
		xfrm_sk_policy_insert(sk, err, pol);
		xfrm_pol_put(pol);
		err = 0;
	}

out:
	kfree(data);
	return err;
}
EXPORT_SYMBOL(xfrm_user_policy);

int xfrm_register_km(struct xfrm_mgr *km)
{
	write_lock_bh(&xfrm_km_lock);
	list_add_tail(&km->list, &xfrm_km_list);
	write_unlock_bh(&xfrm_km_lock);
	return 0;
}
EXPORT_SYMBOL(xfrm_register_km);

int xfrm_unregister_km(struct xfrm_mgr *km)
{
	write_lock_bh(&xfrm_km_lock);
	list_del(&km->list);
	write_unlock_bh(&xfrm_km_lock);
	return 0;
}
EXPORT_SYMBOL(xfrm_unregister_km);

int xfrm_state_register_afinfo(struct xfrm_state_afinfo *afinfo)
{
	int err = 0;
	if (unlikely(afinfo == NULL))
		return -EINVAL;
	if (unlikely(afinfo->family >= NPROTO))
		return -EAFNOSUPPORT;
1912
	write_lock_bh(&xfrm_state_afinfo_lock);
L
Linus Torvalds 已提交
1913 1914
	if (unlikely(xfrm_state_afinfo[afinfo->family] != NULL))
		err = -ENOBUFS;
1915
	else
L
Linus Torvalds 已提交
1916
		xfrm_state_afinfo[afinfo->family] = afinfo;
1917
	write_unlock_bh(&xfrm_state_afinfo_lock);
L
Linus Torvalds 已提交
1918 1919 1920 1921 1922 1923 1924 1925 1926 1927 1928
	return err;
}
EXPORT_SYMBOL(xfrm_state_register_afinfo);

int xfrm_state_unregister_afinfo(struct xfrm_state_afinfo *afinfo)
{
	int err = 0;
	if (unlikely(afinfo == NULL))
		return -EINVAL;
	if (unlikely(afinfo->family >= NPROTO))
		return -EAFNOSUPPORT;
1929
	write_lock_bh(&xfrm_state_afinfo_lock);
L
Linus Torvalds 已提交
1930 1931 1932
	if (likely(xfrm_state_afinfo[afinfo->family] != NULL)) {
		if (unlikely(xfrm_state_afinfo[afinfo->family] != afinfo))
			err = -EINVAL;
1933
		else
L
Linus Torvalds 已提交
1934 1935
			xfrm_state_afinfo[afinfo->family] = NULL;
	}
1936
	write_unlock_bh(&xfrm_state_afinfo_lock);
L
Linus Torvalds 已提交
1937 1938 1939 1940
	return err;
}
EXPORT_SYMBOL(xfrm_state_unregister_afinfo);

1941
static struct xfrm_state_afinfo *xfrm_state_get_afinfo(unsigned int family)
L
Linus Torvalds 已提交
1942 1943 1944 1945 1946 1947
{
	struct xfrm_state_afinfo *afinfo;
	if (unlikely(family >= NPROTO))
		return NULL;
	read_lock(&xfrm_state_afinfo_lock);
	afinfo = xfrm_state_afinfo[family];
1948 1949
	if (unlikely(!afinfo))
		read_unlock(&xfrm_state_afinfo_lock);
L
Linus Torvalds 已提交
1950 1951 1952
	return afinfo;
}

1953
static void xfrm_state_put_afinfo(struct xfrm_state_afinfo *afinfo)
1954
	__releases(xfrm_state_afinfo_lock)
L
Linus Torvalds 已提交
1955
{
1956
	read_unlock(&xfrm_state_afinfo_lock);
L
Linus Torvalds 已提交
1957 1958 1959 1960 1961 1962 1963 1964 1965 1966 1967 1968 1969 1970 1971 1972 1973 1974 1975
}

/* Temporarily located here until net/xfrm/xfrm_tunnel.c is created */
void xfrm_state_delete_tunnel(struct xfrm_state *x)
{
	if (x->tunnel) {
		struct xfrm_state *t = x->tunnel;

		if (atomic_read(&t->tunnel_users) == 2)
			xfrm_state_delete(t);
		atomic_dec(&t->tunnel_users);
		xfrm_state_put(t);
		x->tunnel = NULL;
	}
}
EXPORT_SYMBOL(xfrm_state_delete_tunnel);

int xfrm_state_mtu(struct xfrm_state *x, int mtu)
{
1976
	int res;
L
Linus Torvalds 已提交
1977

1978 1979 1980 1981 1982
	spin_lock_bh(&x->lock);
	if (x->km.state == XFRM_STATE_VALID &&
	    x->type && x->type->get_mtu)
		res = x->type->get_mtu(x, mtu);
	else
1983
		res = mtu - x->props.header_len;
1984
	spin_unlock_bh(&x->lock);
L
Linus Torvalds 已提交
1985 1986 1987
	return res;
}

H
Herbert Xu 已提交
1988 1989
int xfrm_init_state(struct xfrm_state *x)
{
1990
	struct xfrm_state_afinfo *afinfo;
1991
	struct xfrm_mode *inner_mode;
1992
	int family = x->props.family;
H
Herbert Xu 已提交
1993 1994
	int err;

1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009
	err = -EAFNOSUPPORT;
	afinfo = xfrm_state_get_afinfo(family);
	if (!afinfo)
		goto error;

	err = 0;
	if (afinfo->init_flags)
		err = afinfo->init_flags(x);

	xfrm_state_put_afinfo(afinfo);

	if (err)
		goto error;

	err = -EPROTONOSUPPORT;
2010

2011 2012 2013 2014 2015 2016 2017 2018 2019 2020 2021 2022 2023 2024
	if (x->sel.family != AF_UNSPEC) {
		inner_mode = xfrm_get_mode(x->props.mode, x->sel.family);
		if (inner_mode == NULL)
			goto error;

		if (!(inner_mode->flags & XFRM_MODE_FLAG_TUNNEL) &&
		    family != x->sel.family) {
			xfrm_put_mode(inner_mode);
			goto error;
		}

		x->inner_mode = inner_mode;
	} else {
		struct xfrm_mode *inner_mode_iaf;
2025
		int iafamily = AF_INET;
2026

2027
		inner_mode = xfrm_get_mode(x->props.mode, x->props.family);
2028 2029 2030 2031 2032 2033 2034
		if (inner_mode == NULL)
			goto error;

		if (!(inner_mode->flags & XFRM_MODE_FLAG_TUNNEL)) {
			xfrm_put_mode(inner_mode);
			goto error;
		}
2035
		x->inner_mode = inner_mode;
2036

2037 2038
		if (x->props.family == AF_INET)
			iafamily = AF_INET6;
2039

2040 2041 2042 2043 2044 2045
		inner_mode_iaf = xfrm_get_mode(x->props.mode, iafamily);
		if (inner_mode_iaf) {
			if (inner_mode_iaf->flags & XFRM_MODE_FLAG_TUNNEL)
				x->inner_mode_iaf = inner_mode_iaf;
			else
				xfrm_put_mode(inner_mode_iaf);
2046 2047
		}
	}
2048

2049
	x->type = xfrm_get_type(x->id.proto, family);
H
Herbert Xu 已提交
2050 2051 2052 2053 2054 2055 2056
	if (x->type == NULL)
		goto error;

	err = x->type->init_state(x);
	if (err)
		goto error;

2057 2058
	x->outer_mode = xfrm_get_mode(x->props.mode, family);
	if (x->outer_mode == NULL)
2059 2060
		goto error;

H
Herbert Xu 已提交
2061 2062 2063 2064 2065 2066 2067
	x->km.state = XFRM_STATE_VALID;

error:
	return err;
}

EXPORT_SYMBOL(xfrm_init_state);
2068

2069
int __net_init xfrm_state_init(struct net *net)
L
Linus Torvalds 已提交
2070
{
2071 2072
	unsigned int sz;

2073 2074
	INIT_LIST_HEAD(&net->xfrm.state_all);

2075 2076
	sz = sizeof(struct hlist_head) * 8;

2077 2078 2079
	net->xfrm.state_bydst = xfrm_hash_alloc(sz);
	if (!net->xfrm.state_bydst)
		goto out_bydst;
2080 2081 2082
	net->xfrm.state_bysrc = xfrm_hash_alloc(sz);
	if (!net->xfrm.state_bysrc)
		goto out_bysrc;
2083 2084 2085
	net->xfrm.state_byspi = xfrm_hash_alloc(sz);
	if (!net->xfrm.state_byspi)
		goto out_byspi;
2086
	net->xfrm.state_hmask = ((sz / sizeof(struct hlist_head)) - 1);
L
Linus Torvalds 已提交
2087

2088
	net->xfrm.state_num = 0;
2089
	INIT_WORK(&net->xfrm.state_hash_work, xfrm_hash_resize);
2090
	INIT_HLIST_HEAD(&net->xfrm.state_gc_list);
2091
	INIT_WORK(&net->xfrm.state_gc_work, xfrm_state_gc_task);
A
Alexey Dobriyan 已提交
2092
	init_waitqueue_head(&net->xfrm.km_waitq);
2093
	return 0;
2094

2095 2096
out_byspi:
	xfrm_hash_free(net->xfrm.state_bysrc, sz);
2097 2098
out_bysrc:
	xfrm_hash_free(net->xfrm.state_bydst, sz);
2099 2100
out_bydst:
	return -ENOMEM;
2101 2102 2103 2104
}

void xfrm_state_fini(struct net *net)
{
2105
	struct xfrm_audit audit_info;
2106 2107
	unsigned int sz;

2108 2109 2110 2111 2112 2113 2114
	flush_work(&net->xfrm.state_hash_work);
	audit_info.loginuid = -1;
	audit_info.sessionid = -1;
	audit_info.secid = 0;
	xfrm_state_flush(net, IPSEC_PROTO_ANY, &audit_info);
	flush_work(&net->xfrm.state_gc_work);

2115
	WARN_ON(!list_empty(&net->xfrm.state_all));
2116

2117
	sz = (net->xfrm.state_hmask + 1) * sizeof(struct hlist_head);
2118 2119
	WARN_ON(!hlist_empty(net->xfrm.state_byspi));
	xfrm_hash_free(net->xfrm.state_byspi, sz);
2120 2121
	WARN_ON(!hlist_empty(net->xfrm.state_bysrc));
	xfrm_hash_free(net->xfrm.state_bysrc, sz);
2122 2123
	WARN_ON(!hlist_empty(net->xfrm.state_bydst));
	xfrm_hash_free(net->xfrm.state_bydst, sz);
L
Linus Torvalds 已提交
2124 2125
}

J
Joy Latten 已提交
2126
#ifdef CONFIG_AUDITSYSCALL
I
Ilpo Järvinen 已提交
2127 2128
static void xfrm_audit_helper_sainfo(struct xfrm_state *x,
				     struct audit_buffer *audit_buf)
J
Joy Latten 已提交
2129
{
P
Paul Moore 已提交
2130 2131 2132 2133
	struct xfrm_sec_ctx *ctx = x->security;
	u32 spi = ntohl(x->id.spi);

	if (ctx)
J
Joy Latten 已提交
2134
		audit_log_format(audit_buf, " sec_alg=%u sec_doi=%u sec_obj=%s",
P
Paul Moore 已提交
2135
				 ctx->ctx_alg, ctx->ctx_doi, ctx->ctx_str);
J
Joy Latten 已提交
2136 2137 2138

	switch(x->props.family) {
	case AF_INET:
H
Harvey Harrison 已提交
2139 2140
		audit_log_format(audit_buf, " src=%pI4 dst=%pI4",
				 &x->props.saddr.a4, &x->id.daddr.a4);
J
Joy Latten 已提交
2141 2142
		break;
	case AF_INET6:
H
Harvey Harrison 已提交
2143
		audit_log_format(audit_buf, " src=%pI6 dst=%pI6",
2144
				 x->props.saddr.a6, x->id.daddr.a6);
J
Joy Latten 已提交
2145 2146
		break;
	}
P
Paul Moore 已提交
2147 2148

	audit_log_format(audit_buf, " spi=%u(0x%x)", spi, spi);
J
Joy Latten 已提交
2149 2150
}

I
Ilpo Järvinen 已提交
2151 2152
static void xfrm_audit_helper_pktinfo(struct sk_buff *skb, u16 family,
				      struct audit_buffer *audit_buf)
P
Paul Moore 已提交
2153 2154 2155 2156 2157 2158 2159
{
	struct iphdr *iph4;
	struct ipv6hdr *iph6;

	switch (family) {
	case AF_INET:
		iph4 = ip_hdr(skb);
H
Harvey Harrison 已提交
2160 2161
		audit_log_format(audit_buf, " src=%pI4 dst=%pI4",
				 &iph4->saddr, &iph4->daddr);
P
Paul Moore 已提交
2162 2163 2164 2165
		break;
	case AF_INET6:
		iph6 = ipv6_hdr(skb);
		audit_log_format(audit_buf,
H
Harvey Harrison 已提交
2166
				 " src=%pI6 dst=%pI6 flowlbl=0x%x%02x%02x",
2167
				 &iph6->saddr,&iph6->daddr,
P
Paul Moore 已提交
2168 2169 2170 2171 2172 2173 2174
				 iph6->flow_lbl[0] & 0x0f,
				 iph6->flow_lbl[1],
				 iph6->flow_lbl[2]);
		break;
	}
}

P
Paul Moore 已提交
2175
void xfrm_audit_state_add(struct xfrm_state *x, int result,
2176
			  uid_t auid, u32 sessionid, u32 secid)
J
Joy Latten 已提交
2177 2178 2179
{
	struct audit_buffer *audit_buf;

P
Paul Moore 已提交
2180
	audit_buf = xfrm_audit_start("SAD-add");
J
Joy Latten 已提交
2181 2182
	if (audit_buf == NULL)
		return;
2183
	xfrm_audit_helper_usrinfo(auid, sessionid, secid, audit_buf);
P
Paul Moore 已提交
2184 2185
	xfrm_audit_helper_sainfo(x, audit_buf);
	audit_log_format(audit_buf, " res=%u", result);
J
Joy Latten 已提交
2186 2187 2188 2189
	audit_log_end(audit_buf);
}
EXPORT_SYMBOL_GPL(xfrm_audit_state_add);

P
Paul Moore 已提交
2190
void xfrm_audit_state_delete(struct xfrm_state *x, int result,
2191
			     uid_t auid, u32 sessionid, u32 secid)
J
Joy Latten 已提交
2192 2193 2194
{
	struct audit_buffer *audit_buf;

P
Paul Moore 已提交
2195
	audit_buf = xfrm_audit_start("SAD-delete");
J
Joy Latten 已提交
2196 2197
	if (audit_buf == NULL)
		return;
2198
	xfrm_audit_helper_usrinfo(auid, sessionid, secid, audit_buf);
P
Paul Moore 已提交
2199 2200
	xfrm_audit_helper_sainfo(x, audit_buf);
	audit_log_format(audit_buf, " res=%u", result);
J
Joy Latten 已提交
2201 2202 2203
	audit_log_end(audit_buf);
}
EXPORT_SYMBOL_GPL(xfrm_audit_state_delete);
P
Paul Moore 已提交
2204 2205 2206 2207 2208 2209 2210 2211 2212 2213 2214 2215 2216 2217 2218 2219 2220 2221 2222 2223 2224 2225 2226 2227 2228 2229 2230 2231 2232 2233 2234 2235 2236 2237 2238 2239 2240 2241 2242 2243 2244 2245 2246 2247 2248 2249 2250 2251 2252 2253 2254 2255 2256 2257 2258 2259 2260 2261 2262 2263 2264 2265 2266 2267 2268 2269 2270 2271 2272 2273 2274 2275 2276 2277 2278 2279 2280 2281 2282 2283 2284 2285 2286

void xfrm_audit_state_replay_overflow(struct xfrm_state *x,
				      struct sk_buff *skb)
{
	struct audit_buffer *audit_buf;
	u32 spi;

	audit_buf = xfrm_audit_start("SA-replay-overflow");
	if (audit_buf == NULL)
		return;
	xfrm_audit_helper_pktinfo(skb, x->props.family, audit_buf);
	/* don't record the sequence number because it's inherent in this kind
	 * of audit message */
	spi = ntohl(x->id.spi);
	audit_log_format(audit_buf, " spi=%u(0x%x)", spi, spi);
	audit_log_end(audit_buf);
}
EXPORT_SYMBOL_GPL(xfrm_audit_state_replay_overflow);

static void xfrm_audit_state_replay(struct xfrm_state *x,
			     struct sk_buff *skb, __be32 net_seq)
{
	struct audit_buffer *audit_buf;
	u32 spi;

	audit_buf = xfrm_audit_start("SA-replayed-pkt");
	if (audit_buf == NULL)
		return;
	xfrm_audit_helper_pktinfo(skb, x->props.family, audit_buf);
	spi = ntohl(x->id.spi);
	audit_log_format(audit_buf, " spi=%u(0x%x) seqno=%u",
			 spi, spi, ntohl(net_seq));
	audit_log_end(audit_buf);
}

void xfrm_audit_state_notfound_simple(struct sk_buff *skb, u16 family)
{
	struct audit_buffer *audit_buf;

	audit_buf = xfrm_audit_start("SA-notfound");
	if (audit_buf == NULL)
		return;
	xfrm_audit_helper_pktinfo(skb, family, audit_buf);
	audit_log_end(audit_buf);
}
EXPORT_SYMBOL_GPL(xfrm_audit_state_notfound_simple);

void xfrm_audit_state_notfound(struct sk_buff *skb, u16 family,
			       __be32 net_spi, __be32 net_seq)
{
	struct audit_buffer *audit_buf;
	u32 spi;

	audit_buf = xfrm_audit_start("SA-notfound");
	if (audit_buf == NULL)
		return;
	xfrm_audit_helper_pktinfo(skb, family, audit_buf);
	spi = ntohl(net_spi);
	audit_log_format(audit_buf, " spi=%u(0x%x) seqno=%u",
			 spi, spi, ntohl(net_seq));
	audit_log_end(audit_buf);
}
EXPORT_SYMBOL_GPL(xfrm_audit_state_notfound);

void xfrm_audit_state_icvfail(struct xfrm_state *x,
			      struct sk_buff *skb, u8 proto)
{
	struct audit_buffer *audit_buf;
	__be32 net_spi;
	__be32 net_seq;

	audit_buf = xfrm_audit_start("SA-icv-failure");
	if (audit_buf == NULL)
		return;
	xfrm_audit_helper_pktinfo(skb, x->props.family, audit_buf);
	if (xfrm_parse_spi(skb, proto, &net_spi, &net_seq) == 0) {
		u32 spi = ntohl(net_spi);
		audit_log_format(audit_buf, " spi=%u(0x%x) seqno=%u",
				 spi, spi, ntohl(net_seq));
	}
	audit_log_end(audit_buf);
}
EXPORT_SYMBOL_GPL(xfrm_audit_state_icvfail);
J
Joy Latten 已提交
2287
#endif /* CONFIG_AUDITSYSCALL */