xfrm_state.c 53.4 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"

27 28 29
struct sock *xfrm_nl;
EXPORT_SYMBOL(xfrm_nl);

30
u32 sysctl_xfrm_aevent_etime __read_mostly = XFRM_AE_ETIME;
31 32
EXPORT_SYMBOL(sysctl_xfrm_aevent_etime);

33
u32 sysctl_xfrm_aevent_rseqth __read_mostly = XFRM_AE_SEQT_SIZE;
34 35
EXPORT_SYMBOL(sysctl_xfrm_aevent_rseqth);

36 37
u32 sysctl_xfrm_acq_expires __read_mostly = 30;

L
Linus Torvalds 已提交
38 39 40
/* Each xfrm_state may be linked to two tables:

   1. Hash table by (spi,daddr,ah/esp) to find SA by SPI. (input,ctl)
41
   2. Hash table by (daddr,family,reqid) to find what SAs exist for given
L
Linus Torvalds 已提交
42 43 44 45 46
      destination/tunnel endpoint. (output)
 */

static DEFINE_SPINLOCK(xfrm_state_lock);

47
static unsigned int xfrm_state_hashmax __read_mostly = 1 * 1024 * 1024;
48
static unsigned int xfrm_state_genid;
49

50 51 52
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
53 54 55 56 57 58 59
#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 */

60 61
static inline unsigned int xfrm_dst_hash(struct net *net,
					 xfrm_address_t *daddr,
62 63
					 xfrm_address_t *saddr,
					 u32 reqid,
64
					 unsigned short family)
65
{
66
	return __xfrm_dst_hash(daddr, saddr, reqid, family, net->xfrm.state_hmask);
67 68
}

69 70
static inline unsigned int xfrm_src_hash(struct net *net,
					 xfrm_address_t *daddr,
71
					 xfrm_address_t *saddr,
72
					 unsigned short family)
73
{
74
	return __xfrm_src_hash(daddr, saddr, family, net->xfrm.state_hmask);
75 76 77
}

static inline unsigned int
78
xfrm_spi_hash(struct net *net, xfrm_address_t *daddr, __be32 spi, u8 proto, unsigned short family)
79
{
80
	return __xfrm_spi_hash(daddr, spi, proto, family, net->xfrm.state_hmask);
81 82 83 84 85 86 87 88 89 90 91 92 93 94
}

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;

95 96 97
		h = __xfrm_dst_hash(&x->id.daddr, &x->props.saddr,
				    x->props.reqid, x->props.family,
				    nhashmask);
98 99
		hlist_add_head(&x->bydst, ndsttable+h);

100 101
		h = __xfrm_src_hash(&x->id.daddr, &x->props.saddr,
				    x->props.family,
102 103 104
				    nhashmask);
		hlist_add_head(&x->bysrc, nsrctable+h);

105 106 107 108 109 110
		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);
		}
111 112 113
	}
}

114
static unsigned long xfrm_hash_new_size(unsigned int state_hmask)
115
{
116
	return ((state_hmask + 1) << 1) * sizeof(struct hlist_head);
117 118 119 120
}

static DEFINE_MUTEX(hash_resize_mutex);

121
static void xfrm_hash_resize(struct work_struct *work)
122
{
123
	struct net *net = container_of(work, struct net, xfrm.state_hash_work);
124 125 126 127 128 129 130
	struct hlist_head *ndst, *nsrc, *nspi, *odst, *osrc, *ospi;
	unsigned long nsize, osize;
	unsigned int nhashmask, ohashmask;
	int i;

	mutex_lock(&hash_resize_mutex);

131
	nsize = xfrm_hash_new_size(net->xfrm.state_hmask);
132
	ndst = xfrm_hash_alloc(nsize);
133 134
	if (!ndst)
		goto out_unlock;
135
	nsrc = xfrm_hash_alloc(nsize);
136
	if (!nsrc) {
137
		xfrm_hash_free(ndst, nsize);
138 139
		goto out_unlock;
	}
140
	nspi = xfrm_hash_alloc(nsize);
141
	if (!nspi) {
142 143
		xfrm_hash_free(ndst, nsize);
		xfrm_hash_free(nsrc, nsize);
144 145 146 147 148 149
		goto out_unlock;
	}

	spin_lock_bh(&xfrm_state_lock);

	nhashmask = (nsize / sizeof(struct hlist_head)) - 1U;
150 151
	for (i = net->xfrm.state_hmask; i >= 0; i--)
		xfrm_hash_transfer(net->xfrm.state_bydst+i, ndst, nsrc, nspi,
152 153
				   nhashmask);

154 155 156 157
	odst = net->xfrm.state_bydst;
	osrc = net->xfrm.state_bysrc;
	ospi = net->xfrm.state_byspi;
	ohashmask = net->xfrm.state_hmask;
158

159 160 161 162
	net->xfrm.state_bydst = ndst;
	net->xfrm.state_bysrc = nsrc;
	net->xfrm.state_byspi = nspi;
	net->xfrm.state_hmask = nhashmask;
163 164 165 166

	spin_unlock_bh(&xfrm_state_lock);

	osize = (ohashmask + 1) * sizeof(struct hlist_head);
167 168 169
	xfrm_hash_free(odst, osize);
	xfrm_hash_free(osrc, osize);
	xfrm_hash_free(ospi, osize);
170 171 172 173 174

out_unlock:
	mutex_unlock(&hash_resize_mutex);
}

L
Linus Torvalds 已提交
175 176 177 178 179
static DEFINE_RWLOCK(xfrm_state_afinfo_lock);
static struct xfrm_state_afinfo *xfrm_state_afinfo[NPROTO];

static DEFINE_SPINLOCK(xfrm_state_gc_lock);

180
int __xfrm_state_delete(struct xfrm_state *x);
L
Linus Torvalds 已提交
181

182
int km_query(struct xfrm_state *x, struct xfrm_tmpl *t, struct xfrm_policy *pol);
183
void km_state_expired(struct xfrm_state *x, int hard, u32 pid);
L
Linus Torvalds 已提交
184

185 186 187 188 189 190 191 192 193 194 195 196 197
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)
198
	__releases(xfrm_state_afinfo_lock)
199 200 201 202
{
	write_unlock_bh(&xfrm_state_afinfo_lock);
}

203
int xfrm_register_type(const struct xfrm_type *type, unsigned short family)
204 205
{
	struct xfrm_state_afinfo *afinfo = xfrm_state_lock_afinfo(family);
206
	const struct xfrm_type **typemap;
207 208 209 210 211 212 213 214 215 216 217 218 219 220 221
	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);

222
int xfrm_unregister_type(const struct xfrm_type *type, unsigned short family)
223 224
{
	struct xfrm_state_afinfo *afinfo = xfrm_state_lock_afinfo(family);
225
	const struct xfrm_type **typemap;
226 227 228 229 230 231 232 233 234 235 236 237 238 239 240
	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);

241
static const struct xfrm_type *xfrm_get_type(u8 proto, unsigned short family)
242 243
{
	struct xfrm_state_afinfo *afinfo;
244 245
	const struct xfrm_type **typemap;
	const struct xfrm_type *type;
246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267
	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;
}

268
static void xfrm_put_type(const struct xfrm_type *type)
269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287
{
	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;
288 289
	if (modemap[mode->encap])
		goto out;
290

291 292 293 294 295 296 297 298 299
	err = -ENOENT;
	if (!try_module_get(afinfo->owner))
		goto out;

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

out:
300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 317 318 319 320 321
	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;
322
		module_put(mode->afinfo->owner);
323 324 325 326 327 328 329 330 331 332 333 334 335 336 337 338 339 340 341 342 343 344 345 346 347 348 349 350 351 352 353 354 355 356 357 358 359 360 361 362 363
		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
364 365
static void xfrm_state_gc_destroy(struct xfrm_state *x)
{
366 367
	del_timer_sync(&x->timer);
	del_timer_sync(&x->rtimer);
J
Jesper Juhl 已提交
368 369 370 371
	kfree(x->aalg);
	kfree(x->ealg);
	kfree(x->calg);
	kfree(x->encap);
372
	kfree(x->coaddr);
373 374
	if (x->inner_mode)
		xfrm_put_mode(x->inner_mode);
375 376
	if (x->inner_mode_iaf)
		xfrm_put_mode(x->inner_mode_iaf);
377 378
	if (x->outer_mode)
		xfrm_put_mode(x->outer_mode);
L
Linus Torvalds 已提交
379 380 381 382
	if (x->type) {
		x->type->destructor(x);
		xfrm_put_type(x->type);
	}
383
	security_xfrm_state_free(x);
L
Linus Torvalds 已提交
384 385 386
	kfree(x);
}

387
static void xfrm_state_gc_task(struct work_struct *work)
L
Linus Torvalds 已提交
388
{
389
	struct net *net = container_of(work, struct net, xfrm.state_gc_work);
H
Herbert Xu 已提交
390 391 392
	struct xfrm_state *x;
	struct hlist_node *entry, *tmp;
	struct hlist_head gc_list;
L
Linus Torvalds 已提交
393 394

	spin_lock_bh(&xfrm_state_gc_lock);
395
	hlist_move_list(&net->xfrm.state_gc_list, &gc_list);
L
Linus Torvalds 已提交
396 397
	spin_unlock_bh(&xfrm_state_gc_lock);

H
Herbert Xu 已提交
398
	hlist_for_each_entry_safe(x, entry, tmp, &gc_list, gclist)
L
Linus Torvalds 已提交
399
		xfrm_state_gc_destroy(x);
400

A
Alexey Dobriyan 已提交
401
	wake_up(&net->xfrm.km_waitq);
L
Linus Torvalds 已提交
402 403 404 405 406 407 408
}

static inline unsigned long make_jiffies(long secs)
{
	if (secs >= (MAX_SCHEDULE_TIMEOUT-1)/HZ)
		return MAX_SCHEDULE_TIMEOUT-1;
	else
409
		return secs*HZ;
L
Linus Torvalds 已提交
410 411 412 413 414
}

static void xfrm_timer_handler(unsigned long data)
{
	struct xfrm_state *x = (struct xfrm_state*)data;
415
	struct net *net = xs_net(x);
416
	unsigned long now = get_seconds();
L
Linus Torvalds 已提交
417 418
	long next = LONG_MAX;
	int warn = 0;
J
Joy Latten 已提交
419
	int err = 0;
L
Linus Torvalds 已提交
420 421 422 423 424 425 426 427 428 429 430 431 432 433 434 435 436 437 438 439 440 441 442 443 444 445 446 447 448 449 450 451 452 453 454 455 456 457 458 459 460

	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;
	}

461
	x->km.dying = warn;
L
Linus Torvalds 已提交
462
	if (warn)
463
		km_state_expired(x, 0, 0);
L
Linus Torvalds 已提交
464
resched:
465 466 467
	if (next != LONG_MAX)
		mod_timer(&x->timer, jiffies + make_jiffies(next));

L
Linus Torvalds 已提交
468 469 470 471 472
	goto out;

expired:
	if (x->km.state == XFRM_STATE_ACQ && x->id.spi == 0) {
		x->km.state = XFRM_STATE_EXPIRED;
473
		wake_up(&net->xfrm.km_waitq);
L
Linus Torvalds 已提交
474 475 476
		next = 2;
		goto resched;
	}
J
Joy Latten 已提交
477 478 479

	err = __xfrm_state_delete(x);
	if (!err && x->id.spi)
480
		km_state_expired(x, 1, 0);
L
Linus Torvalds 已提交
481

J
Joy Latten 已提交
482
	xfrm_audit_state_delete(x, err ? 0 : 1,
483 484
				audit_get_loginuid(current),
				audit_get_sessionid(current), 0);
J
Joy Latten 已提交
485

L
Linus Torvalds 已提交
486 487 488 489
out:
	spin_unlock(&x->lock);
}

490 491
static void xfrm_replay_timer_handler(unsigned long data);

492
struct xfrm_state *xfrm_state_alloc(struct net *net)
L
Linus Torvalds 已提交
493 494 495
{
	struct xfrm_state *x;

496
	x = kzalloc(sizeof(struct xfrm_state), GFP_ATOMIC);
L
Linus Torvalds 已提交
497 498

	if (x) {
499
		write_pnet(&x->xs_net, net);
L
Linus Torvalds 已提交
500 501
		atomic_set(&x->refcnt, 1);
		atomic_set(&x->tunnel_users, 0);
H
Herbert Xu 已提交
502
		INIT_LIST_HEAD(&x->km.all);
503 504 505
		INIT_HLIST_NODE(&x->bydst);
		INIT_HLIST_NODE(&x->bysrc);
		INIT_HLIST_NODE(&x->byspi);
506 507 508
		setup_timer(&x->timer, xfrm_timer_handler, (unsigned long)x);
		setup_timer(&x->rtimer, xfrm_replay_timer_handler,
				(unsigned long)x);
509
		x->curlft.add_time = get_seconds();
L
Linus Torvalds 已提交
510 511 512 513
		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;
514 515
		x->replay_maxage = 0;
		x->replay_maxdiff = 0;
516 517
		x->inner_mode = NULL;
		x->inner_mode_iaf = NULL;
L
Linus Torvalds 已提交
518 519 520 521 522 523 524 525
		spin_lock_init(&x->lock);
	}
	return x;
}
EXPORT_SYMBOL(xfrm_state_alloc);

void __xfrm_state_destroy(struct xfrm_state *x)
{
526 527
	struct net *net = xs_net(x);

528
	WARN_ON(x->km.state != XFRM_STATE_DEAD);
L
Linus Torvalds 已提交
529 530

	spin_lock_bh(&xfrm_state_gc_lock);
531
	hlist_add_head(&x->gclist, &net->xfrm.state_gc_list);
L
Linus Torvalds 已提交
532
	spin_unlock_bh(&xfrm_state_gc_lock);
533
	schedule_work(&net->xfrm.state_gc_work);
L
Linus Torvalds 已提交
534 535 536
}
EXPORT_SYMBOL(__xfrm_state_destroy);

537
int __xfrm_state_delete(struct xfrm_state *x)
L
Linus Torvalds 已提交
538
{
539
	struct net *net = xs_net(x);
540 541
	int err = -ESRCH;

L
Linus Torvalds 已提交
542 543 544
	if (x->km.state != XFRM_STATE_DEAD) {
		x->km.state = XFRM_STATE_DEAD;
		spin_lock(&xfrm_state_lock);
H
Herbert Xu 已提交
545
		list_del(&x->km.all);
546 547
		hlist_del(&x->bydst);
		hlist_del(&x->bysrc);
548
		if (x->id.spi)
549
			hlist_del(&x->byspi);
550
		net->xfrm.state_num--;
L
Linus Torvalds 已提交
551 552 553 554 555 556
		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.
		 */
557
		xfrm_state_put(x);
558
		err = 0;
L
Linus Torvalds 已提交
559
	}
560 561

	return err;
L
Linus Torvalds 已提交
562
}
563
EXPORT_SYMBOL(__xfrm_state_delete);
L
Linus Torvalds 已提交
564

565
int xfrm_state_delete(struct xfrm_state *x)
L
Linus Torvalds 已提交
566
{
567 568
	int err;

L
Linus Torvalds 已提交
569
	spin_lock_bh(&x->lock);
570
	err = __xfrm_state_delete(x);
L
Linus Torvalds 已提交
571
	spin_unlock_bh(&x->lock);
572 573

	return err;
L
Linus Torvalds 已提交
574 575 576
}
EXPORT_SYMBOL(xfrm_state_delete);

577 578 579
#ifdef CONFIG_SECURITY_NETWORK_XFRM
static inline int
xfrm_state_flush_secctx_check(u8 proto, struct xfrm_audit *audit_info)
L
Linus Torvalds 已提交
580
{
581 582
	int i, err = 0;

583
	for (i = 0; i <= init_net.xfrm.state_hmask; i++) {
584 585 586
		struct hlist_node *entry;
		struct xfrm_state *x;

587
		hlist_for_each_entry(x, entry, init_net.xfrm.state_bydst+i, bydst) {
588 589
			if (xfrm_id_proto_match(x->id.proto, proto) &&
			   (err = security_xfrm_state_delete(x)) != 0) {
J
Joy Latten 已提交
590 591
				xfrm_audit_state_delete(x, 0,
							audit_info->loginuid,
592
							audit_info->sessionid,
J
Joy Latten 已提交
593
							audit_info->secid);
594 595 596 597 598 599 600 601 602 603 604 605 606 607 608 609 610 611
				return err;
			}
		}
	}

	return err;
}
#else
static inline int
xfrm_state_flush_secctx_check(u8 proto, struct xfrm_audit *audit_info)
{
	return 0;
}
#endif

int xfrm_state_flush(u8 proto, struct xfrm_audit *audit_info)
{
	int i, err = 0;
L
Linus Torvalds 已提交
612 613

	spin_lock_bh(&xfrm_state_lock);
614 615 616 617
	err = xfrm_state_flush_secctx_check(proto, audit_info);
	if (err)
		goto out;

618
	for (i = 0; i <= init_net.xfrm.state_hmask; i++) {
619 620
		struct hlist_node *entry;
		struct xfrm_state *x;
L
Linus Torvalds 已提交
621
restart:
622
		hlist_for_each_entry(x, entry, init_net.xfrm.state_bydst+i, bydst) {
L
Linus Torvalds 已提交
623
			if (!xfrm_state_kern(x) &&
624
			    xfrm_id_proto_match(x->id.proto, proto)) {
L
Linus Torvalds 已提交
625 626 627
				xfrm_state_hold(x);
				spin_unlock_bh(&xfrm_state_lock);

J
Joy Latten 已提交
628
				err = xfrm_state_delete(x);
J
Joy Latten 已提交
629 630
				xfrm_audit_state_delete(x, err ? 0 : 1,
							audit_info->loginuid,
631
							audit_info->sessionid,
J
Joy Latten 已提交
632
							audit_info->secid);
L
Linus Torvalds 已提交
633 634 635 636 637 638 639
				xfrm_state_put(x);

				spin_lock_bh(&xfrm_state_lock);
				goto restart;
			}
		}
	}
640 641 642
	err = 0;

out:
L
Linus Torvalds 已提交
643
	spin_unlock_bh(&xfrm_state_lock);
A
Alexey Dobriyan 已提交
644
	wake_up(&init_net.xfrm.km_waitq);
645
	return err;
L
Linus Torvalds 已提交
646 647 648
}
EXPORT_SYMBOL(xfrm_state_flush);

649
void xfrm_sad_getinfo(struct xfrmk_sadinfo *si)
J
Jamal Hadi Salim 已提交
650 651
{
	spin_lock_bh(&xfrm_state_lock);
652
	si->sadcnt = init_net.xfrm.state_num;
653
	si->sadhcnt = init_net.xfrm.state_hmask;
J
Jamal Hadi Salim 已提交
654 655 656 657 658
	si->sadhmcnt = xfrm_state_hashmax;
	spin_unlock_bh(&xfrm_state_lock);
}
EXPORT_SYMBOL(xfrm_sad_getinfo);

L
Linus Torvalds 已提交
659 660 661 662 663 664 665 666 667 668 669 670 671 672
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;
}

673
static struct xfrm_state *__xfrm_state_lookup(xfrm_address_t *daddr, __be32 spi, u8 proto, unsigned short family)
674
{
675
	unsigned int h = xfrm_spi_hash(&init_net, daddr, spi, proto, family);
676
	struct xfrm_state *x;
677
	struct hlist_node *entry;
678

679
	hlist_for_each_entry(x, entry, init_net.xfrm.state_byspi+h, byspi) {
680 681 682 683 684 685 686 687 688 689 690 691 692 693 694 695
		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;
696
		}
697 698 699 700 701 702 703 704 705 706

		xfrm_state_hold(x);
		return x;
	}

	return NULL;
}

static struct xfrm_state *__xfrm_state_lookup_byaddr(xfrm_address_t *daddr, xfrm_address_t *saddr, u8 proto, unsigned short family)
{
707
	unsigned int h = xfrm_src_hash(&init_net, daddr, saddr, family);
708
	struct xfrm_state *x;
709
	struct hlist_node *entry;
710

711
	hlist_for_each_entry(x, entry, init_net.xfrm.state_bysrc+h, bysrc) {
712 713 714 715 716 717 718 719 720 721 722 723 724 725 726 727 728 729 730
		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;
731
		}
732 733 734 735 736 737 738 739 740 741 742 743 744 745 746 747 748 749 750 751

		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)
{
	if (use_spi)
		return __xfrm_state_lookup(&x->id.daddr, x->id.spi,
					   x->id.proto, family);
	else
		return __xfrm_state_lookup_byaddr(&x->id.daddr,
						  &x->props.saddr,
						  x->id.proto, family);
}

752
static void xfrm_hash_grow_check(struct net *net, int have_hash_collision)
753 754
{
	if (have_hash_collision &&
755 756 757
	    (net->xfrm.state_hmask + 1) < xfrm_state_hashmax &&
	    net->xfrm.state_num > net->xfrm.state_hmask)
		schedule_work(&net->xfrm.state_hash_work);
758 759
}

L
Linus Torvalds 已提交
760
struct xfrm_state *
761
xfrm_state_find(xfrm_address_t *daddr, xfrm_address_t *saddr,
L
Linus Torvalds 已提交
762 763 764 765
		struct flowi *fl, struct xfrm_tmpl *tmpl,
		struct xfrm_policy *pol, int *err,
		unsigned short family)
{
766
	unsigned int h;
767
	struct hlist_node *entry;
768
	struct xfrm_state *x, *x0, *to_put;
L
Linus Torvalds 已提交
769 770 771
	int acquire_in_progress = 0;
	int error = 0;
	struct xfrm_state *best = NULL;
772

773 774
	to_put = NULL;

L
Linus Torvalds 已提交
775
	spin_lock_bh(&xfrm_state_lock);
776
	h = xfrm_dst_hash(&init_net, daddr, saddr, tmpl->reqid, family);
777
	hlist_for_each_entry(x, entry, init_net.xfrm.state_bydst+h, bydst) {
L
Linus Torvalds 已提交
778 779
		if (x->props.family == family &&
		    x->props.reqid == tmpl->reqid &&
780
		    !(x->props.flags & XFRM_STATE_WILDRECV) &&
L
Linus Torvalds 已提交
781 782 783 784 785 786 787 788 789 790 791 792 793 794 795 796 797 798
		    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) {
799
				if ((x->sel.family && !xfrm_selector_match(&x->sel, fl, x->sel.family)) ||
800
				    !security_xfrm_state_pol_flow_match(x, pol, fl))
L
Linus Torvalds 已提交
801 802 803 804 805 806 807 808 809 810
					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) {
811
				if (xfrm_selector_match(&x->sel, fl, x->sel.family) &&
812
				    security_xfrm_state_pol_flow_match(x, pol, fl))
L
Linus Torvalds 已提交
813 814 815 816 817 818 819
					error = -ESRCH;
			}
		}
	}

	x = best;
	if (!x && !error && !acquire_in_progress) {
820
		if (tmpl->id.spi &&
821 822
		    (x0 = __xfrm_state_lookup(daddr, tmpl->id.spi,
					      tmpl->id.proto, family)) != NULL) {
823
			to_put = x0;
L
Linus Torvalds 已提交
824 825 826
			error = -EEXIST;
			goto out;
		}
827
		x = xfrm_state_alloc(&init_net);
L
Linus Torvalds 已提交
828 829 830 831 832 833 834 835
		if (x == NULL) {
			error = -ENOMEM;
			goto out;
		}
		/* Initialize temporary selector matching only
		 * to current session. */
		xfrm_init_tempsel(x, fl, tmpl, daddr, saddr, family);

836 837 838
		error = security_xfrm_state_alloc_acquire(x, pol->security, fl->secid);
		if (error) {
			x->km.state = XFRM_STATE_DEAD;
839
			to_put = x;
840 841 842 843
			x = NULL;
			goto out;
		}

L
Linus Torvalds 已提交
844 845
		if (km_query(x, tmpl, pol) == 0) {
			x->km.state = XFRM_STATE_ACQ;
846
			list_add(&x->km.all, &init_net.xfrm.state_all);
847
			hlist_add_head(&x->bydst, init_net.xfrm.state_bydst+h);
848
			h = xfrm_src_hash(&init_net, daddr, saddr, family);
849
			hlist_add_head(&x->bysrc, init_net.xfrm.state_bysrc+h);
L
Linus Torvalds 已提交
850
			if (x->id.spi) {
851
				h = xfrm_spi_hash(&init_net, &x->id.daddr, x->id.spi, x->id.proto, family);
852
				hlist_add_head(&x->byspi, init_net.xfrm.state_byspi+h);
L
Linus Torvalds 已提交
853
			}
854 855
			x->lft.hard_add_expires_seconds = sysctl_xfrm_acq_expires;
			x->timer.expires = jiffies + sysctl_xfrm_acq_expires*HZ;
L
Linus Torvalds 已提交
856
			add_timer(&x->timer);
857
			init_net.xfrm.state_num++;
858
			xfrm_hash_grow_check(&init_net, x->bydst.next != NULL);
L
Linus Torvalds 已提交
859 860
		} else {
			x->km.state = XFRM_STATE_DEAD;
861
			to_put = x;
L
Linus Torvalds 已提交
862 863 864 865 866 867 868 869 870 871
			x = NULL;
			error = -ESRCH;
		}
	}
out:
	if (x)
		xfrm_state_hold(x);
	else
		*err = acquire_in_progress ? -EAGAIN : error;
	spin_unlock_bh(&xfrm_state_lock);
872 873
	if (to_put)
		xfrm_state_put(to_put);
L
Linus Torvalds 已提交
874 875 876
	return x;
}

877 878 879 880
struct xfrm_state *
xfrm_stateonly_find(xfrm_address_t *daddr, xfrm_address_t *saddr,
		    unsigned short family, u8 mode, u8 proto, u32 reqid)
{
881
	unsigned int h;
882 883 884 885
	struct xfrm_state *rx = NULL, *x = NULL;
	struct hlist_node *entry;

	spin_lock(&xfrm_state_lock);
886
	h = xfrm_dst_hash(&init_net, daddr, saddr, reqid, family);
887
	hlist_for_each_entry(x, entry, init_net.xfrm.state_bydst+h, bydst) {
888 889 890 891 892 893 894 895 896 897 898 899 900 901 902 903 904 905 906 907 908
		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
909 910
static void __xfrm_state_insert(struct xfrm_state *x)
{
911
	struct net *net = xs_net(x);
912
	unsigned int h;
L
Linus Torvalds 已提交
913

914 915
	x->genid = ++xfrm_state_genid;

916
	list_add(&x->km.all, &net->xfrm.state_all);
917

918
	h = xfrm_dst_hash(net, &x->id.daddr, &x->props.saddr,
919
			  x->props.reqid, x->props.family);
920
	hlist_add_head(&x->bydst, net->xfrm.state_bydst+h);
L
Linus Torvalds 已提交
921

922 923
	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
924

925
	if (x->id.spi) {
926
		h = xfrm_spi_hash(net, &x->id.daddr, x->id.spi, x->id.proto,
927 928
				  x->props.family);

929
		hlist_add_head(&x->byspi, net->xfrm.state_byspi+h);
930 931
	}

932 933 934
	mod_timer(&x->timer, jiffies + HZ);
	if (x->replay_maxage)
		mod_timer(&x->rtimer, jiffies + x->replay_maxage);
935

936
	wake_up(&net->xfrm.km_waitq);
937

938
	net->xfrm.state_num++;
939

940
	xfrm_hash_grow_check(net, x->bydst.next != NULL);
L
Linus Torvalds 已提交
941 942
}

943 944 945
/* xfrm_state_lock is held */
static void __xfrm_state_bump_genids(struct xfrm_state *xnew)
{
946
	struct net *net = xs_net(xnew);
947 948 949 950 951 952
	unsigned short family = xnew->props.family;
	u32 reqid = xnew->props.reqid;
	struct xfrm_state *x;
	struct hlist_node *entry;
	unsigned int h;

953 954
	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) {
955 956
		if (x->props.family	== family &&
		    x->props.reqid	== reqid &&
957 958
		    !xfrm_addr_cmp(&x->id.daddr, &xnew->id.daddr, family) &&
		    !xfrm_addr_cmp(&x->props.saddr, &xnew->props.saddr, family))
959 960 961 962
			x->genid = xfrm_state_genid;
	}
}

L
Linus Torvalds 已提交
963 964 965
void xfrm_state_insert(struct xfrm_state *x)
{
	spin_lock_bh(&xfrm_state_lock);
966
	__xfrm_state_bump_genids(x);
L
Linus Torvalds 已提交
967 968 969 970 971
	__xfrm_state_insert(x);
	spin_unlock_bh(&xfrm_state_lock);
}
EXPORT_SYMBOL(xfrm_state_insert);

972 973 974
/* xfrm_state_lock is held */
static struct xfrm_state *__find_acq_core(unsigned short family, u8 mode, u32 reqid, u8 proto, xfrm_address_t *daddr, xfrm_address_t *saddr, int create)
{
975
	unsigned int h = xfrm_dst_hash(&init_net, daddr, saddr, reqid, family);
976
	struct hlist_node *entry;
977 978
	struct xfrm_state *x;

979
	hlist_for_each_entry(x, entry, init_net.xfrm.state_bydst+h, bydst) {
980 981 982 983
		if (x->props.reqid  != reqid ||
		    x->props.mode   != mode ||
		    x->props.family != family ||
		    x->km.state     != XFRM_STATE_ACQ ||
984 985
		    x->id.spi       != 0 ||
		    x->id.proto	    != proto)
986 987 988 989 990 991 992 993 994 995 996 997 998 999 1000 1001
			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;
1002
		}
1003 1004 1005 1006 1007 1008 1009 1010

		xfrm_state_hold(x);
		return x;
	}

	if (!create)
		return NULL;

1011
	x = xfrm_state_alloc(&init_net);
1012 1013 1014 1015 1016 1017 1018 1019 1020 1021 1022 1023 1024 1025 1026 1027 1028 1029 1030 1031 1032 1033 1034
	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;
1035
		}
1036 1037 1038 1039 1040 1041

		x->km.state = XFRM_STATE_ACQ;
		x->id.proto = proto;
		x->props.family = family;
		x->props.mode = mode;
		x->props.reqid = reqid;
1042
		x->lft.hard_add_expires_seconds = sysctl_xfrm_acq_expires;
1043
		xfrm_state_hold(x);
1044
		x->timer.expires = jiffies + sysctl_xfrm_acq_expires*HZ;
1045
		add_timer(&x->timer);
1046
		list_add(&x->km.all, &init_net.xfrm.state_all);
1047
		hlist_add_head(&x->bydst, init_net.xfrm.state_bydst+h);
1048
		h = xfrm_src_hash(&init_net, daddr, saddr, family);
1049
		hlist_add_head(&x->bysrc, init_net.xfrm.state_bysrc+h);
1050

1051
		init_net.xfrm.state_num++;
1052

1053
		xfrm_hash_grow_check(&init_net, x->bydst.next != NULL);
1054 1055 1056 1057 1058
	}

	return x;
}

L
Linus Torvalds 已提交
1059 1060 1061 1062
static struct xfrm_state *__xfrm_find_acq_byseq(u32 seq);

int xfrm_state_add(struct xfrm_state *x)
{
1063
	struct xfrm_state *x1, *to_put;
L
Linus Torvalds 已提交
1064 1065
	int family;
	int err;
1066
	int use_spi = xfrm_id_proto_match(x->id.proto, IPSEC_PROTO_ANY);
L
Linus Torvalds 已提交
1067 1068 1069

	family = x->props.family;

1070 1071
	to_put = NULL;

L
Linus Torvalds 已提交
1072 1073
	spin_lock_bh(&xfrm_state_lock);

1074
	x1 = __xfrm_state_locate(x, use_spi, family);
L
Linus Torvalds 已提交
1075
	if (x1) {
1076
		to_put = x1;
L
Linus Torvalds 已提交
1077 1078 1079 1080 1081
		x1 = NULL;
		err = -EEXIST;
		goto out;
	}

1082
	if (use_spi && x->km.seq) {
L
Linus Torvalds 已提交
1083
		x1 = __xfrm_find_acq_byseq(x->km.seq);
1084 1085
		if (x1 && ((x1->id.proto != x->id.proto) ||
		    xfrm_addr_cmp(&x1->id.daddr, &x->id.daddr, family))) {
1086
			to_put = x1;
L
Linus Torvalds 已提交
1087 1088 1089 1090
			x1 = NULL;
		}
	}

1091
	if (use_spi && !x1)
1092 1093 1094
		x1 = __find_acq_core(family, x->props.mode, x->props.reqid,
				     x->id.proto,
				     &x->id.daddr, &x->props.saddr, 0);
L
Linus Torvalds 已提交
1095

1096
	__xfrm_state_bump_genids(x);
L
Linus Torvalds 已提交
1097 1098 1099 1100 1101 1102 1103 1104 1105 1106 1107
	__xfrm_state_insert(x);
	err = 0;

out:
	spin_unlock_bh(&xfrm_state_lock);

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

1108 1109 1110
	if (to_put)
		xfrm_state_put(to_put);

L
Linus Torvalds 已提交
1111 1112 1113 1114
	return err;
}
EXPORT_SYMBOL(xfrm_state_add);

1115
#ifdef CONFIG_XFRM_MIGRATE
1116
static struct xfrm_state *xfrm_state_clone(struct xfrm_state *orig, int *errp)
1117
{
1118
	struct net *net = xs_net(orig);
1119
	int err = -ENOMEM;
1120
	struct xfrm_state *x = xfrm_state_alloc(net);
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 1148 1149 1150 1151 1152 1153
	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;

1154
	if (orig->encap) {
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 1195 1196 1197 1198 1199 1200
		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) {
1201
		h = xfrm_dst_hash(&init_net, &m->old_daddr, &m->old_saddr,
1202
				  m->reqid, m->old_family);
1203
		hlist_for_each_entry(x, entry, init_net.xfrm.state_bydst+h, bydst) {
1204 1205 1206 1207 1208 1209 1210 1211 1212 1213 1214 1215 1216 1217
			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 {
1218
		h = xfrm_src_hash(&init_net, &m->old_daddr, &m->old_saddr,
1219
				  m->old_family);
1220
		hlist_for_each_entry(x, entry, init_net.xfrm.state_bysrc+h, bysrc) {
1221 1222 1223 1224 1225 1226 1227 1228 1229 1230 1231 1232 1233
			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;
		}
	}

1234
	return NULL;
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 1263 1264 1265 1266 1267 1268
}
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
1269 1270
int xfrm_state_update(struct xfrm_state *x)
{
1271
	struct xfrm_state *x1, *to_put;
L
Linus Torvalds 已提交
1272
	int err;
1273
	int use_spi = xfrm_id_proto_match(x->id.proto, IPSEC_PROTO_ANY);
L
Linus Torvalds 已提交
1274

1275 1276
	to_put = NULL;

L
Linus Torvalds 已提交
1277
	spin_lock_bh(&xfrm_state_lock);
1278
	x1 = __xfrm_state_locate(x, use_spi, x->props.family);
L
Linus Torvalds 已提交
1279 1280 1281 1282 1283 1284

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

	if (xfrm_state_kern(x1)) {
1285
		to_put = x1;
L
Linus Torvalds 已提交
1286 1287 1288 1289 1290 1291 1292 1293 1294 1295 1296 1297 1298
		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);

1299 1300 1301
	if (to_put)
		xfrm_state_put(to_put);

L
Linus Torvalds 已提交
1302 1303 1304 1305 1306 1307 1308 1309 1310 1311 1312 1313 1314 1315
	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));
1316 1317 1318 1319 1320
		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
1321 1322 1323
		memcpy(&x1->lft, &x->lft, sizeof(x1->lft));
		x1->km.dying = 0;

1324
		mod_timer(&x1->timer, jiffies + HZ);
L
Linus Torvalds 已提交
1325 1326 1327 1328 1329 1330 1331 1332 1333 1334 1335 1336 1337 1338 1339 1340
		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)
1341
		x->curlft.use_time = get_seconds();
L
Linus Torvalds 已提交
1342 1343 1344 1345 1346 1347

	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) {
1348
		x->km.state = XFRM_STATE_EXPIRED;
1349
		mod_timer(&x->timer, jiffies);
L
Linus Torvalds 已提交
1350 1351 1352 1353 1354
		return -EINVAL;
	}

	if (!x->km.dying &&
	    (x->curlft.bytes >= x->lft.soft_byte_limit ||
1355 1356
	     x->curlft.packets >= x->lft.soft_packet_limit)) {
		x->km.dying = 1;
1357
		km_state_expired(x, 0, 0);
1358
	}
L
Linus Torvalds 已提交
1359 1360 1361 1362 1363
	return 0;
}
EXPORT_SYMBOL(xfrm_state_check_expire);

struct xfrm_state *
1364
xfrm_state_lookup(xfrm_address_t *daddr, __be32 spi, u8 proto,
L
Linus Torvalds 已提交
1365 1366 1367 1368 1369
		  unsigned short family)
{
	struct xfrm_state *x;

	spin_lock_bh(&xfrm_state_lock);
1370
	x = __xfrm_state_lookup(daddr, spi, proto, family);
L
Linus Torvalds 已提交
1371 1372 1373 1374 1375 1376
	spin_unlock_bh(&xfrm_state_lock);
	return x;
}
EXPORT_SYMBOL(xfrm_state_lookup);

struct xfrm_state *
1377 1378 1379 1380 1381 1382
xfrm_state_lookup_byaddr(xfrm_address_t *daddr, xfrm_address_t *saddr,
			 u8 proto, unsigned short family)
{
	struct xfrm_state *x;

	spin_lock_bh(&xfrm_state_lock);
1383
	x = __xfrm_state_lookup_byaddr(daddr, saddr, proto, family);
1384 1385 1386 1387 1388 1389
	spin_unlock_bh(&xfrm_state_lock);
	return x;
}
EXPORT_SYMBOL(xfrm_state_lookup_byaddr);

struct xfrm_state *
1390 1391
xfrm_find_acq(u8 mode, u32 reqid, u8 proto,
	      xfrm_address_t *daddr, xfrm_address_t *saddr,
L
Linus Torvalds 已提交
1392 1393 1394 1395 1396
	      int create, unsigned short family)
{
	struct xfrm_state *x;

	spin_lock_bh(&xfrm_state_lock);
1397
	x = __find_acq_core(family, mode, reqid, proto, daddr, saddr, create);
L
Linus Torvalds 已提交
1398
	spin_unlock_bh(&xfrm_state_lock);
1399

L
Linus Torvalds 已提交
1400 1401 1402 1403
	return x;
}
EXPORT_SYMBOL(xfrm_find_acq);

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 1437 1438 1439 1440 1441
#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
1442 1443 1444 1445 1446 1447
/* Silly enough, but I'm lazy to build resolution list */

static struct xfrm_state *__xfrm_find_acq_byseq(u32 seq)
{
	int i;

1448
	for (i = 0; i <= init_net.xfrm.state_hmask; i++) {
1449 1450 1451
		struct hlist_node *entry;
		struct xfrm_state *x;

1452
		hlist_for_each_entry(x, entry, init_net.xfrm.state_bydst+i, bydst) {
1453 1454
			if (x->km.seq == seq &&
			    x->km.state == XFRM_STATE_ACQ) {
L
Linus Torvalds 已提交
1455 1456 1457 1458 1459 1460 1461 1462 1463 1464 1465 1466 1467 1468 1469 1470 1471 1472 1473 1474 1475 1476 1477 1478 1479 1480 1481 1482 1483 1484 1485 1486
				xfrm_state_hold(x);
				return x;
			}
		}
	}
	return NULL;
}

struct xfrm_state *xfrm_find_acq_byseq(u32 seq)
{
	struct xfrm_state *x;

	spin_lock_bh(&xfrm_state_lock);
	x = __xfrm_find_acq_byseq(seq);
	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);

1487
int xfrm_alloc_spi(struct xfrm_state *x, u32 low, u32 high)
L
Linus Torvalds 已提交
1488
{
1489
	unsigned int h;
L
Linus Torvalds 已提交
1490
	struct xfrm_state *x0;
1491 1492 1493
	int err = -ENOENT;
	__be32 minspi = htonl(low);
	__be32 maxspi = htonl(high);
L
Linus Torvalds 已提交
1494

1495 1496 1497 1498 1499
	spin_lock_bh(&x->lock);
	if (x->km.state == XFRM_STATE_DEAD)
		goto unlock;

	err = 0;
L
Linus Torvalds 已提交
1500
	if (x->id.spi)
1501 1502 1503
		goto unlock;

	err = -ENOENT;
L
Linus Torvalds 已提交
1504 1505 1506 1507 1508

	if (minspi == maxspi) {
		x0 = xfrm_state_lookup(&x->id.daddr, minspi, x->id.proto, x->props.family);
		if (x0) {
			xfrm_state_put(x0);
1509
			goto unlock;
L
Linus Torvalds 已提交
1510 1511 1512 1513
		}
		x->id.spi = minspi;
	} else {
		u32 spi = 0;
A
Al Viro 已提交
1514 1515
		for (h=0; h<high-low+1; h++) {
			spi = low + net_random()%(high-low+1);
L
Linus Torvalds 已提交
1516 1517 1518 1519 1520 1521 1522 1523 1524 1525
			x0 = xfrm_state_lookup(&x->id.daddr, htonl(spi), x->id.proto, x->props.family);
			if (x0 == NULL) {
				x->id.spi = htonl(spi);
				break;
			}
			xfrm_state_put(x0);
		}
	}
	if (x->id.spi) {
		spin_lock_bh(&xfrm_state_lock);
1526
		h = xfrm_spi_hash(&init_net, &x->id.daddr, x->id.spi, x->id.proto, x->props.family);
1527
		hlist_add_head(&x->byspi, init_net.xfrm.state_byspi+h);
L
Linus Torvalds 已提交
1528
		spin_unlock_bh(&xfrm_state_lock);
1529 1530

		err = 0;
L
Linus Torvalds 已提交
1531
	}
1532 1533 1534 1535 1536

unlock:
	spin_unlock_bh(&x->lock);

	return err;
L
Linus Torvalds 已提交
1537 1538 1539
}
EXPORT_SYMBOL(xfrm_alloc_spi);

1540 1541
int xfrm_state_walk(struct xfrm_state_walk *walk,
		    int (*func)(struct xfrm_state *, int, void*),
L
Linus Torvalds 已提交
1542 1543
		    void *data)
{
H
Herbert Xu 已提交
1544 1545
	struct xfrm_state *state;
	struct xfrm_state_walk *x;
L
Linus Torvalds 已提交
1546 1547
	int err = 0;

H
Herbert Xu 已提交
1548
	if (walk->seq != 0 && list_empty(&walk->all))
1549 1550
		return 0;

L
Linus Torvalds 已提交
1551
	spin_lock_bh(&xfrm_state_lock);
H
Herbert Xu 已提交
1552
	if (list_empty(&walk->all))
1553
		x = list_first_entry(&init_net.xfrm.state_all, struct xfrm_state_walk, all);
H
Herbert Xu 已提交
1554 1555
	else
		x = list_entry(&walk->all, struct xfrm_state_walk, all);
1556
	list_for_each_entry_from(x, &init_net.xfrm.state_all, all) {
H
Herbert Xu 已提交
1557
		if (x->state == XFRM_STATE_DEAD)
1558
			continue;
H
Herbert Xu 已提交
1559 1560
		state = container_of(x, struct xfrm_state, km);
		if (!xfrm_id_proto_match(state->id.proto, walk->proto))
1561
			continue;
H
Herbert Xu 已提交
1562 1563 1564 1565
		err = func(state, walk->seq, data);
		if (err) {
			list_move_tail(&walk->all, &x->all);
			goto out;
L
Linus Torvalds 已提交
1566
		}
H
Herbert Xu 已提交
1567
		walk->seq++;
L
Linus Torvalds 已提交
1568
	}
H
Herbert Xu 已提交
1569
	if (walk->seq == 0) {
L
Linus Torvalds 已提交
1570 1571 1572
		err = -ENOENT;
		goto out;
	}
H
Herbert Xu 已提交
1573
	list_del_init(&walk->all);
L
Linus Torvalds 已提交
1574 1575 1576 1577 1578 1579
out:
	spin_unlock_bh(&xfrm_state_lock);
	return err;
}
EXPORT_SYMBOL(xfrm_state_walk);

H
Herbert Xu 已提交
1580 1581
void xfrm_state_walk_init(struct xfrm_state_walk *walk, u8 proto)
{
H
Herbert Xu 已提交
1582
	INIT_LIST_HEAD(&walk->all);
H
Herbert Xu 已提交
1583
	walk->proto = proto;
H
Herbert Xu 已提交
1584 1585
	walk->state = XFRM_STATE_DEAD;
	walk->seq = 0;
H
Herbert Xu 已提交
1586 1587 1588
}
EXPORT_SYMBOL(xfrm_state_walk_init);

1589 1590
void xfrm_state_walk_done(struct xfrm_state_walk *walk)
{
H
Herbert Xu 已提交
1591
	if (list_empty(&walk->all))
H
Herbert Xu 已提交
1592 1593
		return;

H
Herbert Xu 已提交
1594 1595 1596
	spin_lock_bh(&xfrm_state_lock);
	list_del(&walk->all);
	spin_lock_bh(&xfrm_state_lock);
1597 1598 1599
}
EXPORT_SYMBOL(xfrm_state_walk_done);

1600 1601 1602 1603 1604 1605 1606 1607 1608 1609 1610 1611 1612 1613 1614 1615 1616 1617

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
1618 1619 1620 1621 1622 1623
		    (x->replay.oseq - x->preplay.oseq < x->replay_maxdiff)) {
			if (x->xflags & XFRM_TIME_DEFER)
				event = XFRM_REPLAY_TIMEOUT;
			else
				return;
		}
1624 1625 1626 1627 1628 1629

		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
1630 1631
		    (x->replay.oseq == x->preplay.oseq)) {
			x->xflags |= XFRM_TIME_DEFER;
1632
			return;
J
Jamal Hadi Salim 已提交
1633
		}
1634 1635 1636 1637 1638 1639 1640 1641 1642 1643

		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 &&
1644
	    !mod_timer(&x->rtimer, jiffies + x->replay_maxage))
J
Jamal Hadi Salim 已提交
1645
		x->xflags &= ~XFRM_TIME_DEFER;
1646 1647 1648 1649 1650 1651 1652 1653
}

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
1654 1655 1656 1657 1658 1659
	if (x->km.state == XFRM_STATE_VALID) {
		if (xfrm_aevent_is_on())
			xfrm_replay_notify(x, XFRM_REPLAY_TIMEOUT);
		else
			x->xflags |= XFRM_TIME_DEFER;
	}
1660 1661 1662 1663

	spin_unlock(&x->lock);
}

P
Paul Moore 已提交
1664 1665
int xfrm_replay_check(struct xfrm_state *x,
		      struct sk_buff *skb, __be32 net_seq)
L
Linus Torvalds 已提交
1666 1667
{
	u32 diff;
1668
	u32 seq = ntohl(net_seq);
L
Linus Torvalds 已提交
1669 1670

	if (unlikely(seq == 0))
P
Paul Moore 已提交
1671
		goto err;
L
Linus Torvalds 已提交
1672 1673 1674 1675 1676

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

	diff = x->replay.seq - seq;
1677 1678
	if (diff >= min_t(unsigned int, x->props.replay_window,
			  sizeof(x->replay.bitmap) * 8)) {
L
Linus Torvalds 已提交
1679
		x->stats.replay_window++;
P
Paul Moore 已提交
1680
		goto err;
L
Linus Torvalds 已提交
1681 1682 1683 1684
	}

	if (x->replay.bitmap & (1U << diff)) {
		x->stats.replay++;
P
Paul Moore 已提交
1685
		goto err;
L
Linus Torvalds 已提交
1686 1687
	}
	return 0;
P
Paul Moore 已提交
1688 1689 1690 1691

err:
	xfrm_audit_state_replay(x, skb, net_seq);
	return -EINVAL;
L
Linus Torvalds 已提交
1692 1693
}

1694
void xfrm_replay_advance(struct xfrm_state *x, __be32 net_seq)
L
Linus Torvalds 已提交
1695 1696
{
	u32 diff;
1697
	u32 seq = ntohl(net_seq);
L
Linus Torvalds 已提交
1698 1699 1700 1701 1702 1703 1704 1705 1706 1707 1708 1709

	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);
	}
1710 1711 1712

	if (xfrm_aevent_is_on())
		xfrm_replay_notify(x, XFRM_REPLAY_UPDATE);
L
Linus Torvalds 已提交
1713 1714
}

1715
static LIST_HEAD(xfrm_km_list);
L
Linus Torvalds 已提交
1716 1717
static DEFINE_RWLOCK(xfrm_km_lock);

1718
void km_policy_notify(struct xfrm_policy *xp, int dir, struct km_event *c)
L
Linus Torvalds 已提交
1719 1720 1721
{
	struct xfrm_mgr *km;

1722 1723 1724 1725 1726 1727
	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
1728

1729 1730 1731
void km_state_notify(struct xfrm_state *x, struct km_event *c)
{
	struct xfrm_mgr *km;
L
Linus Torvalds 已提交
1732 1733
	read_lock(&xfrm_km_lock);
	list_for_each_entry(km, &xfrm_km_list, list)
1734 1735
		if (km->notify)
			km->notify(x, c);
L
Linus Torvalds 已提交
1736
	read_unlock(&xfrm_km_lock);
1737 1738 1739 1740 1741
}

EXPORT_SYMBOL(km_policy_notify);
EXPORT_SYMBOL(km_state_notify);

1742
void km_state_expired(struct xfrm_state *x, int hard, u32 pid)
1743
{
1744
	struct net *net = xs_net(x);
1745 1746
	struct km_event c;

1747
	c.data.hard = hard;
1748
	c.pid = pid;
1749
	c.event = XFRM_MSG_EXPIRE;
1750
	km_state_notify(x, &c);
L
Linus Torvalds 已提交
1751 1752

	if (hard)
1753
		wake_up(&net->xfrm.km_waitq);
L
Linus Torvalds 已提交
1754 1755
}

1756
EXPORT_SYMBOL(km_state_expired);
1757 1758 1759 1760
/*
 * We send to all registered managers regardless of failure
 * We are happy with one success
*/
1761
int km_query(struct xfrm_state *x, struct xfrm_tmpl *t, struct xfrm_policy *pol)
L
Linus Torvalds 已提交
1762
{
1763
	int err = -EINVAL, acqret;
L
Linus Torvalds 已提交
1764 1765 1766 1767
	struct xfrm_mgr *km;

	read_lock(&xfrm_km_lock);
	list_for_each_entry(km, &xfrm_km_list, list) {
1768 1769 1770
		acqret = km->acquire(x, t, pol, XFRM_POLICY_OUT);
		if (!acqret)
			err = acqret;
L
Linus Torvalds 已提交
1771 1772 1773 1774
	}
	read_unlock(&xfrm_km_lock);
	return err;
}
1775
EXPORT_SYMBOL(km_query);
L
Linus Torvalds 已提交
1776

A
Al Viro 已提交
1777
int km_new_mapping(struct xfrm_state *x, xfrm_address_t *ipaddr, __be16 sport)
L
Linus Torvalds 已提交
1778 1779 1780 1781 1782 1783 1784 1785 1786 1787 1788 1789 1790 1791 1792 1793
{
	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);

1794
void km_policy_expired(struct xfrm_policy *pol, int dir, int hard, u32 pid)
L
Linus Torvalds 已提交
1795
{
1796
	struct net *net = xp_net(pol);
1797
	struct km_event c;
L
Linus Torvalds 已提交
1798

1799
	c.data.hard = hard;
1800
	c.pid = pid;
1801
	c.event = XFRM_MSG_POLEXPIRE;
1802
	km_policy_notify(pol, dir, &c);
L
Linus Torvalds 已提交
1803 1804

	if (hard)
1805
		wake_up(&net->xfrm.km_waitq);
L
Linus Torvalds 已提交
1806
}
1807
EXPORT_SYMBOL(km_policy_expired);
L
Linus Torvalds 已提交
1808

1809
#ifdef CONFIG_XFRM_MIGRATE
1810
int km_migrate(struct xfrm_selector *sel, u8 dir, u8 type,
1811 1812
	       struct xfrm_migrate *m, int num_migrate,
	       struct xfrm_kmaddress *k)
1813 1814 1815 1816 1817 1818 1819 1820
{
	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) {
1821
			ret = km->migrate(sel, dir, type, m, num_migrate, k);
1822 1823 1824 1825 1826 1827 1828 1829
			if (!ret)
				err = ret;
		}
	}
	read_unlock(&xfrm_km_lock);
	return err;
}
EXPORT_SYMBOL(km_migrate);
1830
#endif
1831

1832 1833 1834 1835 1836 1837 1838 1839 1840 1841 1842 1843 1844 1845 1846 1847 1848 1849 1850
int km_report(u8 proto, struct xfrm_selector *sel, xfrm_address_t *addr)
{
	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) {
			ret = km->report(proto, sel, addr);
			if (!ret)
				err = ret;
		}
	}
	read_unlock(&xfrm_km_lock);
	return err;
}
EXPORT_SYMBOL(km_report);

L
Linus Torvalds 已提交
1851 1852 1853 1854 1855 1856 1857 1858 1859 1860 1861 1862 1863 1864 1865 1866 1867 1868 1869 1870 1871
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) {
1872
		pol = km->compile_policy(sk, optname, data,
L
Linus Torvalds 已提交
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 1912 1913 1914 1915
					 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;
1916
	write_lock_bh(&xfrm_state_afinfo_lock);
L
Linus Torvalds 已提交
1917 1918
	if (unlikely(xfrm_state_afinfo[afinfo->family] != NULL))
		err = -ENOBUFS;
1919
	else
L
Linus Torvalds 已提交
1920
		xfrm_state_afinfo[afinfo->family] = afinfo;
1921
	write_unlock_bh(&xfrm_state_afinfo_lock);
L
Linus Torvalds 已提交
1922 1923 1924 1925 1926 1927 1928 1929 1930 1931 1932
	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;
1933
	write_lock_bh(&xfrm_state_afinfo_lock);
L
Linus Torvalds 已提交
1934 1935 1936
	if (likely(xfrm_state_afinfo[afinfo->family] != NULL)) {
		if (unlikely(xfrm_state_afinfo[afinfo->family] != afinfo))
			err = -EINVAL;
1937
		else
L
Linus Torvalds 已提交
1938 1939
			xfrm_state_afinfo[afinfo->family] = NULL;
	}
1940
	write_unlock_bh(&xfrm_state_afinfo_lock);
L
Linus Torvalds 已提交
1941 1942 1943 1944
	return err;
}
EXPORT_SYMBOL(xfrm_state_unregister_afinfo);

1945
static struct xfrm_state_afinfo *xfrm_state_get_afinfo(unsigned int family)
L
Linus Torvalds 已提交
1946 1947 1948 1949 1950 1951
{
	struct xfrm_state_afinfo *afinfo;
	if (unlikely(family >= NPROTO))
		return NULL;
	read_lock(&xfrm_state_afinfo_lock);
	afinfo = xfrm_state_afinfo[family];
1952 1953
	if (unlikely(!afinfo))
		read_unlock(&xfrm_state_afinfo_lock);
L
Linus Torvalds 已提交
1954 1955 1956
	return afinfo;
}

1957
static void xfrm_state_put_afinfo(struct xfrm_state_afinfo *afinfo)
1958
	__releases(xfrm_state_afinfo_lock)
L
Linus Torvalds 已提交
1959
{
1960
	read_unlock(&xfrm_state_afinfo_lock);
L
Linus Torvalds 已提交
1961 1962 1963 1964 1965 1966 1967 1968 1969 1970 1971 1972 1973 1974 1975 1976 1977 1978 1979
}

/* 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)
{
1980
	int res;
L
Linus Torvalds 已提交
1981

1982 1983 1984 1985 1986
	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
1987
		res = mtu - x->props.header_len;
1988
	spin_unlock_bh(&x->lock);
L
Linus Torvalds 已提交
1989 1990 1991
	return res;
}

H
Herbert Xu 已提交
1992 1993
int xfrm_init_state(struct xfrm_state *x)
{
1994
	struct xfrm_state_afinfo *afinfo;
1995
	struct xfrm_mode *inner_mode;
1996
	int family = x->props.family;
H
Herbert Xu 已提交
1997 1998
	int err;

1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013
	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;
2014

2015 2016 2017 2018 2019 2020 2021 2022 2023 2024 2025 2026 2027 2028 2029 2030 2031 2032 2033 2034 2035 2036 2037 2038 2039 2040 2041 2042 2043 2044 2045 2046 2047 2048 2049 2050 2051 2052 2053 2054 2055
	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;

		inner_mode = xfrm_get_mode(x->props.mode, AF_INET);
		if (inner_mode == NULL)
			goto error;

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

		inner_mode_iaf = xfrm_get_mode(x->props.mode, AF_INET6);
		if (inner_mode_iaf == NULL)
			goto error;

		if (!(inner_mode_iaf->flags & XFRM_MODE_FLAG_TUNNEL)) {
			xfrm_put_mode(inner_mode_iaf);
			goto error;
		}

		if (x->props.family == AF_INET) {
			x->inner_mode = inner_mode;
			x->inner_mode_iaf = inner_mode_iaf;
		} else {
			x->inner_mode = inner_mode_iaf;
			x->inner_mode_iaf = inner_mode;
		}
	}
2056

2057
	x->type = xfrm_get_type(x->id.proto, family);
H
Herbert Xu 已提交
2058 2059 2060 2061 2062 2063 2064
	if (x->type == NULL)
		goto error;

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

2065 2066
	x->outer_mode = xfrm_get_mode(x->props.mode, family);
	if (x->outer_mode == NULL)
2067 2068
		goto error;

H
Herbert Xu 已提交
2069 2070 2071 2072 2073 2074 2075
	x->km.state = XFRM_STATE_VALID;

error:
	return err;
}

EXPORT_SYMBOL(xfrm_init_state);
2076

2077
int __net_init xfrm_state_init(struct net *net)
L
Linus Torvalds 已提交
2078
{
2079 2080
	unsigned int sz;

2081 2082
	INIT_LIST_HEAD(&net->xfrm.state_all);

2083 2084
	sz = sizeof(struct hlist_head) * 8;

2085 2086 2087
	net->xfrm.state_bydst = xfrm_hash_alloc(sz);
	if (!net->xfrm.state_bydst)
		goto out_bydst;
2088 2089 2090
	net->xfrm.state_bysrc = xfrm_hash_alloc(sz);
	if (!net->xfrm.state_bysrc)
		goto out_bysrc;
2091 2092 2093
	net->xfrm.state_byspi = xfrm_hash_alloc(sz);
	if (!net->xfrm.state_byspi)
		goto out_byspi;
2094
	net->xfrm.state_hmask = ((sz / sizeof(struct hlist_head)) - 1);
L
Linus Torvalds 已提交
2095

2096
	net->xfrm.state_num = 0;
2097
	INIT_WORK(&net->xfrm.state_hash_work, xfrm_hash_resize);
2098
	INIT_HLIST_HEAD(&net->xfrm.state_gc_list);
2099
	INIT_WORK(&net->xfrm.state_gc_work, xfrm_state_gc_task);
A
Alexey Dobriyan 已提交
2100
	init_waitqueue_head(&net->xfrm.km_waitq);
2101
	return 0;
2102

2103 2104
out_byspi:
	xfrm_hash_free(net->xfrm.state_bysrc, sz);
2105 2106
out_bysrc:
	xfrm_hash_free(net->xfrm.state_bydst, sz);
2107 2108
out_bydst:
	return -ENOMEM;
2109 2110 2111 2112
}

void xfrm_state_fini(struct net *net)
{
2113 2114
	unsigned int sz;

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 */