6 * Kazunori MIYAZAWA @USAGI
7 * Kunihiro Ishiguro <kunihiro@ipinfusion.com>
9 * YOSHIFUJI Hideaki @USAGI
10 * Split up af-specific functions
11 * Derek Atkins <derek@ihtfp.com>
12 * Add UDP Encapsulation
16 #include <linux/workqueue.h>
18 #include <linux/pfkeyv2.h>
19 #include <linux/ipsec.h>
20 #include <linux/module.h>
21 #include <linux/cache.h>
22 #include <linux/audit.h>
23 #include <asm/uaccess.h>
24 #include <linux/ktime.h>
25 #include <linux/slab.h>
26 #include <linux/interrupt.h>
27 #include <linux/kernel.h>
29 #include "xfrm_hash.h"
31 /* Each xfrm_state may be linked to two tables:
33 1. Hash table by (spi,daddr,ah/esp) to find SA by SPI. (input,ctl)
34 2. Hash table by (daddr,family,reqid) to find what SAs exist for given
35 destination/tunnel endpoint. (output)
38 static DEFINE_SPINLOCK(xfrm_state_lock);
40 static unsigned int xfrm_state_hashmax __read_mostly = 1 * 1024 * 1024;
41 static unsigned int xfrm_state_genid;
43 static struct xfrm_state_afinfo *xfrm_state_get_afinfo(unsigned int family);
44 static void xfrm_state_put_afinfo(struct xfrm_state_afinfo *afinfo);
46 #ifdef CONFIG_AUDITSYSCALL
47 static void xfrm_audit_state_replay(struct xfrm_state *x,
48 struct sk_buff *skb, __be32 net_seq);
50 #define xfrm_audit_state_replay(x, s, sq) do { ; } while (0)
51 #endif /* CONFIG_AUDITSYSCALL */
53 static inline unsigned int xfrm_dst_hash(struct net *net,
54 xfrm_address_t *daddr,
55 xfrm_address_t *saddr,
57 unsigned short family)
59 return __xfrm_dst_hash(daddr, saddr, reqid, family, net->xfrm.state_hmask);
62 static inline unsigned int xfrm_src_hash(struct net *net,
63 xfrm_address_t *daddr,
64 xfrm_address_t *saddr,
65 unsigned short family)
67 return __xfrm_src_hash(daddr, saddr, family, net->xfrm.state_hmask);
70 static inline unsigned int
71 xfrm_spi_hash(struct net *net, xfrm_address_t *daddr, __be32 spi, u8 proto, unsigned short family)
73 return __xfrm_spi_hash(daddr, spi, proto, family, net->xfrm.state_hmask);
76 static void xfrm_hash_transfer(struct hlist_head *list,
77 struct hlist_head *ndsttable,
78 struct hlist_head *nsrctable,
79 struct hlist_head *nspitable,
80 unsigned int nhashmask)
82 struct hlist_node *entry, *tmp;
85 hlist_for_each_entry_safe(x, entry, tmp, list, bydst) {
88 h = __xfrm_dst_hash(&x->id.daddr, &x->props.saddr,
89 x->props.reqid, x->props.family,
91 hlist_add_head(&x->bydst, ndsttable+h);
93 h = __xfrm_src_hash(&x->id.daddr, &x->props.saddr,
96 hlist_add_head(&x->bysrc, nsrctable+h);
99 h = __xfrm_spi_hash(&x->id.daddr, x->id.spi,
100 x->id.proto, x->props.family,
102 hlist_add_head(&x->byspi, nspitable+h);
107 static unsigned long xfrm_hash_new_size(unsigned int state_hmask)
109 return ((state_hmask + 1) << 1) * sizeof(struct hlist_head);
112 static DEFINE_MUTEX(hash_resize_mutex);
114 static void xfrm_hash_resize(struct work_struct *work)
116 struct net *net = container_of(work, struct net, xfrm.state_hash_work);
117 struct hlist_head *ndst, *nsrc, *nspi, *odst, *osrc, *ospi;
118 unsigned long nsize, osize;
119 unsigned int nhashmask, ohashmask;
122 mutex_lock(&hash_resize_mutex);
124 nsize = xfrm_hash_new_size(net->xfrm.state_hmask);
125 ndst = xfrm_hash_alloc(nsize);
128 nsrc = xfrm_hash_alloc(nsize);
130 xfrm_hash_free(ndst, nsize);
133 nspi = xfrm_hash_alloc(nsize);
135 xfrm_hash_free(ndst, nsize);
136 xfrm_hash_free(nsrc, nsize);
140 spin_lock_bh(&xfrm_state_lock);
142 nhashmask = (nsize / sizeof(struct hlist_head)) - 1U;
143 for (i = net->xfrm.state_hmask; i >= 0; i--)
144 xfrm_hash_transfer(net->xfrm.state_bydst+i, ndst, nsrc, nspi,
147 odst = net->xfrm.state_bydst;
148 osrc = net->xfrm.state_bysrc;
149 ospi = net->xfrm.state_byspi;
150 ohashmask = net->xfrm.state_hmask;
152 net->xfrm.state_bydst = ndst;
153 net->xfrm.state_bysrc = nsrc;
154 net->xfrm.state_byspi = nspi;
155 net->xfrm.state_hmask = nhashmask;
157 spin_unlock_bh(&xfrm_state_lock);
159 osize = (ohashmask + 1) * sizeof(struct hlist_head);
160 xfrm_hash_free(odst, osize);
161 xfrm_hash_free(osrc, osize);
162 xfrm_hash_free(ospi, osize);
165 mutex_unlock(&hash_resize_mutex);
168 static DEFINE_RWLOCK(xfrm_state_afinfo_lock);
169 static struct xfrm_state_afinfo *xfrm_state_afinfo[NPROTO];
171 static DEFINE_SPINLOCK(xfrm_state_gc_lock);
173 int __xfrm_state_delete(struct xfrm_state *x);
175 int km_query(struct xfrm_state *x, struct xfrm_tmpl *t, struct xfrm_policy *pol);
176 void km_state_expired(struct xfrm_state *x, int hard, u32 pid);
178 static struct xfrm_state_afinfo *xfrm_state_lock_afinfo(unsigned int family)
180 struct xfrm_state_afinfo *afinfo;
181 if (unlikely(family >= NPROTO))
183 write_lock_bh(&xfrm_state_afinfo_lock);
184 afinfo = xfrm_state_afinfo[family];
185 if (unlikely(!afinfo))
186 write_unlock_bh(&xfrm_state_afinfo_lock);
190 static void xfrm_state_unlock_afinfo(struct xfrm_state_afinfo *afinfo)
191 __releases(xfrm_state_afinfo_lock)
193 write_unlock_bh(&xfrm_state_afinfo_lock);
196 int xfrm_register_type(const struct xfrm_type *type, unsigned short family)
198 struct xfrm_state_afinfo *afinfo = xfrm_state_lock_afinfo(family);
199 const struct xfrm_type **typemap;
202 if (unlikely(afinfo == NULL))
203 return -EAFNOSUPPORT;
204 typemap = afinfo->type_map;
206 if (likely(typemap[type->proto] == NULL))
207 typemap[type->proto] = type;
210 xfrm_state_unlock_afinfo(afinfo);
213 EXPORT_SYMBOL(xfrm_register_type);
215 int xfrm_unregister_type(const struct xfrm_type *type, unsigned short family)
217 struct xfrm_state_afinfo *afinfo = xfrm_state_lock_afinfo(family);
218 const struct xfrm_type **typemap;
221 if (unlikely(afinfo == NULL))
222 return -EAFNOSUPPORT;
223 typemap = afinfo->type_map;
225 if (unlikely(typemap[type->proto] != type))
228 typemap[type->proto] = NULL;
229 xfrm_state_unlock_afinfo(afinfo);
232 EXPORT_SYMBOL(xfrm_unregister_type);
234 static const struct xfrm_type *xfrm_get_type(u8 proto, unsigned short family)
236 struct xfrm_state_afinfo *afinfo;
237 const struct xfrm_type **typemap;
238 const struct xfrm_type *type;
239 int modload_attempted = 0;
242 afinfo = xfrm_state_get_afinfo(family);
243 if (unlikely(afinfo == NULL))
245 typemap = afinfo->type_map;
247 type = typemap[proto];
248 if (unlikely(type && !try_module_get(type->owner)))
250 if (!type && !modload_attempted) {
251 xfrm_state_put_afinfo(afinfo);
252 request_module("xfrm-type-%d-%d", family, proto);
253 modload_attempted = 1;
257 xfrm_state_put_afinfo(afinfo);
261 static void xfrm_put_type(const struct xfrm_type *type)
263 module_put(type->owner);
266 int xfrm_register_mode(struct xfrm_mode *mode, int family)
268 struct xfrm_state_afinfo *afinfo;
269 struct xfrm_mode **modemap;
272 if (unlikely(mode->encap >= XFRM_MODE_MAX))
275 afinfo = xfrm_state_lock_afinfo(family);
276 if (unlikely(afinfo == NULL))
277 return -EAFNOSUPPORT;
280 modemap = afinfo->mode_map;
281 if (modemap[mode->encap])
285 if (!try_module_get(afinfo->owner))
288 mode->afinfo = afinfo;
289 modemap[mode->encap] = mode;
293 xfrm_state_unlock_afinfo(afinfo);
296 EXPORT_SYMBOL(xfrm_register_mode);
298 int xfrm_unregister_mode(struct xfrm_mode *mode, int family)
300 struct xfrm_state_afinfo *afinfo;
301 struct xfrm_mode **modemap;
304 if (unlikely(mode->encap >= XFRM_MODE_MAX))
307 afinfo = xfrm_state_lock_afinfo(family);
308 if (unlikely(afinfo == NULL))
309 return -EAFNOSUPPORT;
312 modemap = afinfo->mode_map;
313 if (likely(modemap[mode->encap] == mode)) {
314 modemap[mode->encap] = NULL;
315 module_put(mode->afinfo->owner);
319 xfrm_state_unlock_afinfo(afinfo);
322 EXPORT_SYMBOL(xfrm_unregister_mode);
324 static struct xfrm_mode *xfrm_get_mode(unsigned int encap, int family)
326 struct xfrm_state_afinfo *afinfo;
327 struct xfrm_mode *mode;
328 int modload_attempted = 0;
330 if (unlikely(encap >= XFRM_MODE_MAX))
334 afinfo = xfrm_state_get_afinfo(family);
335 if (unlikely(afinfo == NULL))
338 mode = afinfo->mode_map[encap];
339 if (unlikely(mode && !try_module_get(mode->owner)))
341 if (!mode && !modload_attempted) {
342 xfrm_state_put_afinfo(afinfo);
343 request_module("xfrm-mode-%d-%d", family, encap);
344 modload_attempted = 1;
348 xfrm_state_put_afinfo(afinfo);
352 static void xfrm_put_mode(struct xfrm_mode *mode)
354 module_put(mode->owner);
357 static void xfrm_state_gc_destroy(struct xfrm_state *x)
359 tasklet_hrtimer_cancel(&x->mtimer);
360 del_timer_sync(&x->rtimer);
367 xfrm_put_mode(x->inner_mode);
368 if (x->inner_mode_iaf)
369 xfrm_put_mode(x->inner_mode_iaf);
371 xfrm_put_mode(x->outer_mode);
373 x->type->destructor(x);
374 xfrm_put_type(x->type);
376 security_xfrm_state_free(x);
380 static void xfrm_state_gc_task(struct work_struct *work)
382 struct net *net = container_of(work, struct net, xfrm.state_gc_work);
383 struct xfrm_state *x;
384 struct hlist_node *entry, *tmp;
385 struct hlist_head gc_list;
387 spin_lock_bh(&xfrm_state_gc_lock);
388 hlist_move_list(&net->xfrm.state_gc_list, &gc_list);
389 spin_unlock_bh(&xfrm_state_gc_lock);
391 hlist_for_each_entry_safe(x, entry, tmp, &gc_list, gclist)
392 xfrm_state_gc_destroy(x);
394 wake_up(&net->xfrm.km_waitq);
397 static inline unsigned long make_jiffies(long secs)
399 if (secs >= (MAX_SCHEDULE_TIMEOUT-1)/HZ)
400 return MAX_SCHEDULE_TIMEOUT-1;
405 static enum hrtimer_restart xfrm_timer_handler(struct hrtimer * me)
407 struct tasklet_hrtimer *thr = container_of(me, struct tasklet_hrtimer, timer);
408 struct xfrm_state *x = container_of(thr, struct xfrm_state, mtimer);
409 struct net *net = xs_net(x);
410 unsigned long now = get_seconds();
411 long next = LONG_MAX;
416 if (x->km.state == XFRM_STATE_DEAD)
418 if (x->km.state == XFRM_STATE_EXPIRED)
420 if (x->lft.hard_add_expires_seconds) {
421 long tmo = x->lft.hard_add_expires_seconds +
422 x->curlft.add_time - now;
428 if (x->lft.hard_use_expires_seconds) {
429 long tmo = x->lft.hard_use_expires_seconds +
430 (x->curlft.use_time ? : now) - now;
438 if (x->lft.soft_add_expires_seconds) {
439 long tmo = x->lft.soft_add_expires_seconds +
440 x->curlft.add_time - now;
446 if (x->lft.soft_use_expires_seconds) {
447 long tmo = x->lft.soft_use_expires_seconds +
448 (x->curlft.use_time ? : now) - now;
457 km_state_expired(x, 0, 0);
459 if (next != LONG_MAX){
460 tasklet_hrtimer_start(&x->mtimer, ktime_set(next, 0), HRTIMER_MODE_REL);
466 if (x->km.state == XFRM_STATE_ACQ && x->id.spi == 0) {
467 x->km.state = XFRM_STATE_EXPIRED;
468 wake_up(&net->xfrm.km_waitq);
473 err = __xfrm_state_delete(x);
474 if (!err && x->id.spi)
475 km_state_expired(x, 1, 0);
477 xfrm_audit_state_delete(x, err ? 0 : 1,
478 audit_get_loginuid(current),
479 audit_get_sessionid(current), 0);
482 spin_unlock(&x->lock);
483 return HRTIMER_NORESTART;
486 static void xfrm_replay_timer_handler(unsigned long data);
488 struct xfrm_state *xfrm_state_alloc(struct net *net)
490 struct xfrm_state *x;
492 x = kzalloc(sizeof(struct xfrm_state), GFP_ATOMIC);
495 write_pnet(&x->xs_net, net);
496 atomic_set(&x->refcnt, 1);
497 atomic_set(&x->tunnel_users, 0);
498 INIT_LIST_HEAD(&x->km.all);
499 INIT_HLIST_NODE(&x->bydst);
500 INIT_HLIST_NODE(&x->bysrc);
501 INIT_HLIST_NODE(&x->byspi);
502 tasklet_hrtimer_init(&x->mtimer, xfrm_timer_handler, CLOCK_REALTIME, HRTIMER_MODE_ABS);
503 setup_timer(&x->rtimer, xfrm_replay_timer_handler,
505 x->curlft.add_time = get_seconds();
506 x->lft.soft_byte_limit = XFRM_INF;
507 x->lft.soft_packet_limit = XFRM_INF;
508 x->lft.hard_byte_limit = XFRM_INF;
509 x->lft.hard_packet_limit = XFRM_INF;
510 x->replay_maxage = 0;
511 x->replay_maxdiff = 0;
512 x->inner_mode = NULL;
513 x->inner_mode_iaf = NULL;
514 spin_lock_init(&x->lock);
518 EXPORT_SYMBOL(xfrm_state_alloc);
520 void __xfrm_state_destroy(struct xfrm_state *x)
522 struct net *net = xs_net(x);
524 WARN_ON(x->km.state != XFRM_STATE_DEAD);
526 spin_lock_bh(&xfrm_state_gc_lock);
527 hlist_add_head(&x->gclist, &net->xfrm.state_gc_list);
528 spin_unlock_bh(&xfrm_state_gc_lock);
529 schedule_work(&net->xfrm.state_gc_work);
531 EXPORT_SYMBOL(__xfrm_state_destroy);
533 int __xfrm_state_delete(struct xfrm_state *x)
535 struct net *net = xs_net(x);
538 if (x->km.state != XFRM_STATE_DEAD) {
539 x->km.state = XFRM_STATE_DEAD;
540 spin_lock(&xfrm_state_lock);
541 list_del(&x->km.all);
542 hlist_del(&x->bydst);
543 hlist_del(&x->bysrc);
545 hlist_del(&x->byspi);
546 net->xfrm.state_num--;
547 spin_unlock(&xfrm_state_lock);
549 /* All xfrm_state objects are created by xfrm_state_alloc.
550 * The xfrm_state_alloc call gives a reference, and that
551 * is what we are dropping here.
559 EXPORT_SYMBOL(__xfrm_state_delete);
561 int xfrm_state_delete(struct xfrm_state *x)
565 spin_lock_bh(&x->lock);
566 err = __xfrm_state_delete(x);
567 spin_unlock_bh(&x->lock);
571 EXPORT_SYMBOL(xfrm_state_delete);
573 #ifdef CONFIG_SECURITY_NETWORK_XFRM
575 xfrm_state_flush_secctx_check(struct net *net, u8 proto, struct xfrm_audit *audit_info)
579 for (i = 0; i <= net->xfrm.state_hmask; i++) {
580 struct hlist_node *entry;
581 struct xfrm_state *x;
583 hlist_for_each_entry(x, entry, net->xfrm.state_bydst+i, bydst) {
584 if (xfrm_id_proto_match(x->id.proto, proto) &&
585 (err = security_xfrm_state_delete(x)) != 0) {
586 xfrm_audit_state_delete(x, 0,
587 audit_info->loginuid,
588 audit_info->sessionid,
599 xfrm_state_flush_secctx_check(struct net *net, u8 proto, struct xfrm_audit *audit_info)
605 int xfrm_state_flush(struct net *net, u8 proto, struct xfrm_audit *audit_info)
607 int i, err = 0, cnt = 0;
609 spin_lock_bh(&xfrm_state_lock);
610 err = xfrm_state_flush_secctx_check(net, proto, audit_info);
615 for (i = 0; i <= net->xfrm.state_hmask; i++) {
616 struct hlist_node *entry;
617 struct xfrm_state *x;
619 hlist_for_each_entry(x, entry, net->xfrm.state_bydst+i, bydst) {
620 if (!xfrm_state_kern(x) &&
621 xfrm_id_proto_match(x->id.proto, proto)) {
623 spin_unlock_bh(&xfrm_state_lock);
625 err = xfrm_state_delete(x);
626 xfrm_audit_state_delete(x, err ? 0 : 1,
627 audit_info->loginuid,
628 audit_info->sessionid,
634 spin_lock_bh(&xfrm_state_lock);
643 spin_unlock_bh(&xfrm_state_lock);
644 wake_up(&net->xfrm.km_waitq);
647 EXPORT_SYMBOL(xfrm_state_flush);
649 void xfrm_sad_getinfo(struct net *net, struct xfrmk_sadinfo *si)
651 spin_lock_bh(&xfrm_state_lock);
652 si->sadcnt = net->xfrm.state_num;
653 si->sadhcnt = net->xfrm.state_hmask;
654 si->sadhmcnt = xfrm_state_hashmax;
655 spin_unlock_bh(&xfrm_state_lock);
657 EXPORT_SYMBOL(xfrm_sad_getinfo);
660 xfrm_init_tempsel(struct xfrm_state *x, struct flowi *fl,
661 struct xfrm_tmpl *tmpl,
662 xfrm_address_t *daddr, xfrm_address_t *saddr,
663 unsigned short family)
665 struct xfrm_state_afinfo *afinfo = xfrm_state_get_afinfo(family);
668 afinfo->init_tempsel(x, fl, tmpl, daddr, saddr);
669 xfrm_state_put_afinfo(afinfo);
673 static struct xfrm_state *__xfrm_state_lookup(struct net *net, u32 mark, xfrm_address_t *daddr, __be32 spi, u8 proto, unsigned short family)
675 unsigned int h = xfrm_spi_hash(net, daddr, spi, proto, family);
676 struct xfrm_state *x;
677 struct hlist_node *entry;
679 hlist_for_each_entry(x, entry, net->xfrm.state_byspi+h, byspi) {
680 if (x->props.family != family ||
682 x->id.proto != proto ||
683 xfrm_addr_cmp(&x->id.daddr, daddr, family))
686 if ((mark & x->mark.m) != x->mark.v)
695 static struct xfrm_state *__xfrm_state_lookup_byaddr(struct net *net, u32 mark, xfrm_address_t *daddr, xfrm_address_t *saddr, u8 proto, unsigned short family)
697 unsigned int h = xfrm_src_hash(net, daddr, saddr, family);
698 struct xfrm_state *x;
699 struct hlist_node *entry;
701 hlist_for_each_entry(x, entry, net->xfrm.state_bysrc+h, bysrc) {
702 if (x->props.family != family ||
703 x->id.proto != proto ||
704 xfrm_addr_cmp(&x->id.daddr, daddr, family) ||
705 xfrm_addr_cmp(&x->props.saddr, saddr, family))
708 if ((mark & x->mark.m) != x->mark.v)
717 static inline struct xfrm_state *
718 __xfrm_state_locate(struct xfrm_state *x, int use_spi, int family)
720 struct net *net = xs_net(x);
721 u32 mark = x->mark.v & x->mark.m;
724 return __xfrm_state_lookup(net, mark, &x->id.daddr,
725 x->id.spi, x->id.proto, family);
727 return __xfrm_state_lookup_byaddr(net, mark,
730 x->id.proto, family);
733 static void xfrm_hash_grow_check(struct net *net, int have_hash_collision)
735 if (have_hash_collision &&
736 (net->xfrm.state_hmask + 1) < xfrm_state_hashmax &&
737 net->xfrm.state_num > net->xfrm.state_hmask)
738 schedule_work(&net->xfrm.state_hash_work);
741 static void xfrm_state_look_at(struct xfrm_policy *pol, struct xfrm_state *x,
742 struct flowi *fl, unsigned short family,
743 xfrm_address_t *daddr, xfrm_address_t *saddr,
744 struct xfrm_state **best, int *acq_in_progress,
748 * 1. There is a valid state with matching selector. Done.
749 * 2. Valid state with inappropriate selector. Skip.
751 * Entering area of "sysdeps".
753 * 3. If state is not valid, selector is temporary, it selects
754 * only session which triggered previous resolution. Key
755 * manager will do something to install a state with proper
758 if (x->km.state == XFRM_STATE_VALID) {
759 if ((x->sel.family &&
760 !xfrm_selector_match(&x->sel, fl, x->sel.family)) ||
761 !security_xfrm_state_pol_flow_match(x, pol, fl))
765 (*best)->km.dying > x->km.dying ||
766 ((*best)->km.dying == x->km.dying &&
767 (*best)->curlft.add_time < x->curlft.add_time))
769 } else if (x->km.state == XFRM_STATE_ACQ) {
770 *acq_in_progress = 1;
771 } else if (x->km.state == XFRM_STATE_ERROR ||
772 x->km.state == XFRM_STATE_EXPIRED) {
773 if (xfrm_selector_match(&x->sel, fl, x->sel.family) &&
774 security_xfrm_state_pol_flow_match(x, pol, fl))
780 xfrm_state_find(xfrm_address_t *daddr, xfrm_address_t *saddr,
781 struct flowi *fl, struct xfrm_tmpl *tmpl,
782 struct xfrm_policy *pol, int *err,
783 unsigned short family)
785 static xfrm_address_t saddr_wildcard = { };
786 struct net *net = xp_net(pol);
787 unsigned int h, h_wildcard;
788 struct hlist_node *entry;
789 struct xfrm_state *x, *x0, *to_put;
790 int acquire_in_progress = 0;
792 struct xfrm_state *best = NULL;
793 u32 mark = pol->mark.v & pol->mark.m;
797 spin_lock_bh(&xfrm_state_lock);
798 h = xfrm_dst_hash(net, daddr, saddr, tmpl->reqid, family);
799 hlist_for_each_entry(x, entry, net->xfrm.state_bydst+h, bydst) {
800 if (x->props.family == family &&
801 x->props.reqid == tmpl->reqid &&
802 (mark & x->mark.m) == x->mark.v &&
803 !(x->props.flags & XFRM_STATE_WILDRECV) &&
804 xfrm_state_addr_check(x, daddr, saddr, family) &&
805 tmpl->mode == x->props.mode &&
806 tmpl->id.proto == x->id.proto &&
807 (tmpl->id.spi == x->id.spi || !tmpl->id.spi))
808 xfrm_state_look_at(pol, x, fl, family, daddr, saddr,
809 &best, &acquire_in_progress, &error);
814 h_wildcard = xfrm_dst_hash(net, daddr, &saddr_wildcard, tmpl->reqid, family);
815 hlist_for_each_entry(x, entry, net->xfrm.state_bydst+h_wildcard, bydst) {
816 if (x->props.family == family &&
817 x->props.reqid == tmpl->reqid &&
818 (mark & x->mark.m) == x->mark.v &&
819 !(x->props.flags & XFRM_STATE_WILDRECV) &&
820 xfrm_state_addr_check(x, daddr, saddr, family) &&
821 tmpl->mode == x->props.mode &&
822 tmpl->id.proto == x->id.proto &&
823 (tmpl->id.spi == x->id.spi || !tmpl->id.spi))
824 xfrm_state_look_at(pol, x, fl, family, daddr, saddr,
825 &best, &acquire_in_progress, &error);
830 if (!x && !error && !acquire_in_progress) {
832 (x0 = __xfrm_state_lookup(net, mark, daddr, tmpl->id.spi,
833 tmpl->id.proto, family)) != NULL) {
838 x = xfrm_state_alloc(net);
843 /* Initialize temporary selector matching only
844 * to current session. */
845 xfrm_init_tempsel(x, fl, tmpl, daddr, saddr, family);
846 memcpy(&x->mark, &pol->mark, sizeof(x->mark));
848 error = security_xfrm_state_alloc_acquire(x, pol->security, fl->secid);
850 x->km.state = XFRM_STATE_DEAD;
856 if (km_query(x, tmpl, pol) == 0) {
857 x->km.state = XFRM_STATE_ACQ;
858 list_add(&x->km.all, &net->xfrm.state_all);
859 hlist_add_head(&x->bydst, net->xfrm.state_bydst+h);
860 h = xfrm_src_hash(net, daddr, saddr, family);
861 hlist_add_head(&x->bysrc, net->xfrm.state_bysrc+h);
863 h = xfrm_spi_hash(net, &x->id.daddr, x->id.spi, x->id.proto, family);
864 hlist_add_head(&x->byspi, net->xfrm.state_byspi+h);
866 x->lft.hard_add_expires_seconds = net->xfrm.sysctl_acq_expires;
867 tasklet_hrtimer_start(&x->mtimer, ktime_set(net->xfrm.sysctl_acq_expires, 0), HRTIMER_MODE_REL);
868 net->xfrm.state_num++;
869 xfrm_hash_grow_check(net, x->bydst.next != NULL);
871 x->km.state = XFRM_STATE_DEAD;
881 *err = acquire_in_progress ? -EAGAIN : error;
882 spin_unlock_bh(&xfrm_state_lock);
884 xfrm_state_put(to_put);
889 xfrm_stateonly_find(struct net *net, u32 mark,
890 xfrm_address_t *daddr, xfrm_address_t *saddr,
891 unsigned short family, u8 mode, u8 proto, u32 reqid)
894 struct xfrm_state *rx = NULL, *x = NULL;
895 struct hlist_node *entry;
897 spin_lock(&xfrm_state_lock);
898 h = xfrm_dst_hash(net, daddr, saddr, reqid, family);
899 hlist_for_each_entry(x, entry, net->xfrm.state_bydst+h, bydst) {
900 if (x->props.family == family &&
901 x->props.reqid == reqid &&
902 (mark & x->mark.m) == x->mark.v &&
903 !(x->props.flags & XFRM_STATE_WILDRECV) &&
904 xfrm_state_addr_check(x, daddr, saddr, family) &&
905 mode == x->props.mode &&
906 proto == x->id.proto &&
907 x->km.state == XFRM_STATE_VALID) {
915 spin_unlock(&xfrm_state_lock);
920 EXPORT_SYMBOL(xfrm_stateonly_find);
922 static void __xfrm_state_insert(struct xfrm_state *x)
924 struct net *net = xs_net(x);
927 x->genid = ++xfrm_state_genid;
929 list_add(&x->km.all, &net->xfrm.state_all);
931 h = xfrm_dst_hash(net, &x->id.daddr, &x->props.saddr,
932 x->props.reqid, x->props.family);
933 hlist_add_head(&x->bydst, net->xfrm.state_bydst+h);
935 h = xfrm_src_hash(net, &x->id.daddr, &x->props.saddr, x->props.family);
936 hlist_add_head(&x->bysrc, net->xfrm.state_bysrc+h);
939 h = xfrm_spi_hash(net, &x->id.daddr, x->id.spi, x->id.proto,
942 hlist_add_head(&x->byspi, net->xfrm.state_byspi+h);
945 tasklet_hrtimer_start(&x->mtimer, ktime_set(1, 0), HRTIMER_MODE_REL);
946 if (x->replay_maxage)
947 mod_timer(&x->rtimer, jiffies + x->replay_maxage);
949 wake_up(&net->xfrm.km_waitq);
951 net->xfrm.state_num++;
953 xfrm_hash_grow_check(net, x->bydst.next != NULL);
956 /* xfrm_state_lock is held */
957 static void __xfrm_state_bump_genids(struct xfrm_state *xnew)
959 struct net *net = xs_net(xnew);
960 unsigned short family = xnew->props.family;
961 u32 reqid = xnew->props.reqid;
962 struct xfrm_state *x;
963 struct hlist_node *entry;
965 u32 mark = xnew->mark.v & xnew->mark.m;
967 h = xfrm_dst_hash(net, &xnew->id.daddr, &xnew->props.saddr, reqid, family);
968 hlist_for_each_entry(x, entry, net->xfrm.state_bydst+h, bydst) {
969 if (x->props.family == family &&
970 x->props.reqid == reqid &&
971 (mark & x->mark.m) == x->mark.v &&
972 !xfrm_addr_cmp(&x->id.daddr, &xnew->id.daddr, family) &&
973 !xfrm_addr_cmp(&x->props.saddr, &xnew->props.saddr, family))
974 x->genid = xfrm_state_genid;
978 void xfrm_state_insert(struct xfrm_state *x)
980 spin_lock_bh(&xfrm_state_lock);
981 __xfrm_state_bump_genids(x);
982 __xfrm_state_insert(x);
983 spin_unlock_bh(&xfrm_state_lock);
985 EXPORT_SYMBOL(xfrm_state_insert);
987 /* xfrm_state_lock is held */
988 static struct xfrm_state *__find_acq_core(struct net *net, struct xfrm_mark *m, unsigned short family, u8 mode, u32 reqid, u8 proto, xfrm_address_t *daddr, xfrm_address_t *saddr, int create)
990 unsigned int h = xfrm_dst_hash(net, daddr, saddr, reqid, family);
991 struct hlist_node *entry;
992 struct xfrm_state *x;
993 u32 mark = m->v & m->m;
995 hlist_for_each_entry(x, entry, net->xfrm.state_bydst+h, bydst) {
996 if (x->props.reqid != reqid ||
997 x->props.mode != mode ||
998 x->props.family != family ||
999 x->km.state != XFRM_STATE_ACQ ||
1001 x->id.proto != proto ||
1002 (mark & x->mark.m) != x->mark.v ||
1003 xfrm_addr_cmp(&x->id.daddr, daddr, family) ||
1004 xfrm_addr_cmp(&x->props.saddr, saddr, family))
1014 x = xfrm_state_alloc(net);
1018 x->sel.daddr.a4 = daddr->a4;
1019 x->sel.saddr.a4 = saddr->a4;
1020 x->sel.prefixlen_d = 32;
1021 x->sel.prefixlen_s = 32;
1022 x->props.saddr.a4 = saddr->a4;
1023 x->id.daddr.a4 = daddr->a4;
1027 ipv6_addr_copy((struct in6_addr *)x->sel.daddr.a6,
1028 (struct in6_addr *)daddr);
1029 ipv6_addr_copy((struct in6_addr *)x->sel.saddr.a6,
1030 (struct in6_addr *)saddr);
1031 x->sel.prefixlen_d = 128;
1032 x->sel.prefixlen_s = 128;
1033 ipv6_addr_copy((struct in6_addr *)x->props.saddr.a6,
1034 (struct in6_addr *)saddr);
1035 ipv6_addr_copy((struct in6_addr *)x->id.daddr.a6,
1036 (struct in6_addr *)daddr);
1040 x->km.state = XFRM_STATE_ACQ;
1041 x->id.proto = proto;
1042 x->props.family = family;
1043 x->props.mode = mode;
1044 x->props.reqid = reqid;
1047 x->lft.hard_add_expires_seconds = net->xfrm.sysctl_acq_expires;
1049 tasklet_hrtimer_start(&x->mtimer, ktime_set(net->xfrm.sysctl_acq_expires, 0), HRTIMER_MODE_REL);
1050 list_add(&x->km.all, &net->xfrm.state_all);
1051 hlist_add_head(&x->bydst, net->xfrm.state_bydst+h);
1052 h = xfrm_src_hash(net, daddr, saddr, family);
1053 hlist_add_head(&x->bysrc, net->xfrm.state_bysrc+h);
1055 net->xfrm.state_num++;
1057 xfrm_hash_grow_check(net, x->bydst.next != NULL);
1063 static struct xfrm_state *__xfrm_find_acq_byseq(struct net *net, u32 mark, u32 seq);
1065 int xfrm_state_add(struct xfrm_state *x)
1067 struct net *net = xs_net(x);
1068 struct xfrm_state *x1, *to_put;
1071 u32 mark = x->mark.v & x->mark.m;
1072 int use_spi = xfrm_id_proto_match(x->id.proto, IPSEC_PROTO_ANY);
1074 family = x->props.family;
1078 spin_lock_bh(&xfrm_state_lock);
1080 x1 = __xfrm_state_locate(x, use_spi, family);
1088 if (use_spi && x->km.seq) {
1089 x1 = __xfrm_find_acq_byseq(net, mark, x->km.seq);
1090 if (x1 && ((x1->id.proto != x->id.proto) ||
1091 xfrm_addr_cmp(&x1->id.daddr, &x->id.daddr, family))) {
1098 x1 = __find_acq_core(net, &x->mark, family, x->props.mode,
1099 x->props.reqid, x->id.proto,
1100 &x->id.daddr, &x->props.saddr, 0);
1102 __xfrm_state_bump_genids(x);
1103 __xfrm_state_insert(x);
1107 spin_unlock_bh(&xfrm_state_lock);
1110 xfrm_state_delete(x1);
1115 xfrm_state_put(to_put);
1119 EXPORT_SYMBOL(xfrm_state_add);
1121 #ifdef CONFIG_XFRM_MIGRATE
1122 static struct xfrm_state *xfrm_state_clone(struct xfrm_state *orig, int *errp)
1124 struct net *net = xs_net(orig);
1126 struct xfrm_state *x = xfrm_state_alloc(net);
1130 memcpy(&x->id, &orig->id, sizeof(x->id));
1131 memcpy(&x->sel, &orig->sel, sizeof(x->sel));
1132 memcpy(&x->lft, &orig->lft, sizeof(x->lft));
1133 x->props.mode = orig->props.mode;
1134 x->props.replay_window = orig->props.replay_window;
1135 x->props.reqid = orig->props.reqid;
1136 x->props.family = orig->props.family;
1137 x->props.saddr = orig->props.saddr;
1140 x->aalg = xfrm_algo_auth_clone(orig->aalg);
1144 x->props.aalgo = orig->props.aalgo;
1147 x->ealg = xfrm_algo_clone(orig->ealg);
1151 x->props.ealgo = orig->props.ealgo;
1154 x->calg = xfrm_algo_clone(orig->calg);
1158 x->props.calgo = orig->props.calgo;
1161 x->encap = kmemdup(orig->encap, sizeof(*x->encap), GFP_KERNEL);
1167 x->coaddr = kmemdup(orig->coaddr, sizeof(*x->coaddr),
1173 memcpy(&x->mark, &orig->mark, sizeof(x->mark));
1175 err = xfrm_init_state(x);
1179 x->props.flags = orig->props.flags;
1181 x->curlft.add_time = orig->curlft.add_time;
1182 x->km.state = orig->km.state;
1183 x->km.seq = orig->km.seq;
1195 /* xfrm_state_lock is held */
1196 struct xfrm_state * xfrm_migrate_state_find(struct xfrm_migrate *m)
1199 struct xfrm_state *x;
1200 struct hlist_node *entry;
1203 h = xfrm_dst_hash(&init_net, &m->old_daddr, &m->old_saddr,
1204 m->reqid, m->old_family);
1205 hlist_for_each_entry(x, entry, init_net.xfrm.state_bydst+h, bydst) {
1206 if (x->props.mode != m->mode ||
1207 x->id.proto != m->proto)
1209 if (m->reqid && x->props.reqid != m->reqid)
1211 if (xfrm_addr_cmp(&x->id.daddr, &m->old_daddr,
1213 xfrm_addr_cmp(&x->props.saddr, &m->old_saddr,
1220 h = xfrm_src_hash(&init_net, &m->old_daddr, &m->old_saddr,
1222 hlist_for_each_entry(x, entry, init_net.xfrm.state_bysrc+h, bysrc) {
1223 if (x->props.mode != m->mode ||
1224 x->id.proto != m->proto)
1226 if (xfrm_addr_cmp(&x->id.daddr, &m->old_daddr,
1228 xfrm_addr_cmp(&x->props.saddr, &m->old_saddr,
1238 EXPORT_SYMBOL(xfrm_migrate_state_find);
1240 struct xfrm_state * xfrm_state_migrate(struct xfrm_state *x,
1241 struct xfrm_migrate *m)
1243 struct xfrm_state *xc;
1246 xc = xfrm_state_clone(x, &err);
1250 memcpy(&xc->id.daddr, &m->new_daddr, sizeof(xc->id.daddr));
1251 memcpy(&xc->props.saddr, &m->new_saddr, sizeof(xc->props.saddr));
1254 if (!xfrm_addr_cmp(&x->id.daddr, &m->new_daddr, m->new_family)) {
1255 /* a care is needed when the destination address of the
1256 state is to be updated as it is a part of triplet */
1257 xfrm_state_insert(xc);
1259 if ((err = xfrm_state_add(xc)) < 0)
1268 EXPORT_SYMBOL(xfrm_state_migrate);
1271 int xfrm_state_update(struct xfrm_state *x)
1273 struct xfrm_state *x1, *to_put;
1275 int use_spi = xfrm_id_proto_match(x->id.proto, IPSEC_PROTO_ANY);
1279 spin_lock_bh(&xfrm_state_lock);
1280 x1 = __xfrm_state_locate(x, use_spi, x->props.family);
1286 if (xfrm_state_kern(x1)) {
1292 if (x1->km.state == XFRM_STATE_ACQ) {
1293 __xfrm_state_insert(x);
1299 spin_unlock_bh(&xfrm_state_lock);
1302 xfrm_state_put(to_put);
1308 xfrm_state_delete(x1);
1314 spin_lock_bh(&x1->lock);
1315 if (likely(x1->km.state == XFRM_STATE_VALID)) {
1316 if (x->encap && x1->encap)
1317 memcpy(x1->encap, x->encap, sizeof(*x1->encap));
1318 if (x->coaddr && x1->coaddr) {
1319 memcpy(x1->coaddr, x->coaddr, sizeof(*x1->coaddr));
1321 if (!use_spi && memcmp(&x1->sel, &x->sel, sizeof(x1->sel)))
1322 memcpy(&x1->sel, &x->sel, sizeof(x1->sel));
1323 memcpy(&x1->lft, &x->lft, sizeof(x1->lft));
1326 tasklet_hrtimer_start(&x1->mtimer, ktime_set(1, 0), HRTIMER_MODE_REL);
1327 if (x1->curlft.use_time)
1328 xfrm_state_check_expire(x1);
1332 spin_unlock_bh(&x1->lock);
1338 EXPORT_SYMBOL(xfrm_state_update);
1340 int xfrm_state_check_expire(struct xfrm_state *x)
1342 if (!x->curlft.use_time)
1343 x->curlft.use_time = get_seconds();
1345 if (x->km.state != XFRM_STATE_VALID)
1348 if (x->curlft.bytes >= x->lft.hard_byte_limit ||
1349 x->curlft.packets >= x->lft.hard_packet_limit) {
1350 x->km.state = XFRM_STATE_EXPIRED;
1351 tasklet_hrtimer_start(&x->mtimer, ktime_set(0,0), HRTIMER_MODE_REL);
1356 (x->curlft.bytes >= x->lft.soft_byte_limit ||
1357 x->curlft.packets >= x->lft.soft_packet_limit)) {
1359 km_state_expired(x, 0, 0);
1363 EXPORT_SYMBOL(xfrm_state_check_expire);
1366 xfrm_state_lookup(struct net *net, u32 mark, xfrm_address_t *daddr, __be32 spi,
1367 u8 proto, unsigned short family)
1369 struct xfrm_state *x;
1371 spin_lock_bh(&xfrm_state_lock);
1372 x = __xfrm_state_lookup(net, mark, daddr, spi, proto, family);
1373 spin_unlock_bh(&xfrm_state_lock);
1376 EXPORT_SYMBOL(xfrm_state_lookup);
1379 xfrm_state_lookup_byaddr(struct net *net, u32 mark,
1380 xfrm_address_t *daddr, xfrm_address_t *saddr,
1381 u8 proto, unsigned short family)
1383 struct xfrm_state *x;
1385 spin_lock_bh(&xfrm_state_lock);
1386 x = __xfrm_state_lookup_byaddr(net, mark, daddr, saddr, proto, family);
1387 spin_unlock_bh(&xfrm_state_lock);
1390 EXPORT_SYMBOL(xfrm_state_lookup_byaddr);
1393 xfrm_find_acq(struct net *net, struct xfrm_mark *mark, u8 mode, u32 reqid, u8 proto,
1394 xfrm_address_t *daddr, xfrm_address_t *saddr,
1395 int create, unsigned short family)
1397 struct xfrm_state *x;
1399 spin_lock_bh(&xfrm_state_lock);
1400 x = __find_acq_core(net, mark, family, mode, reqid, proto, daddr, saddr, create);
1401 spin_unlock_bh(&xfrm_state_lock);
1405 EXPORT_SYMBOL(xfrm_find_acq);
1407 #ifdef CONFIG_XFRM_SUB_POLICY
1409 xfrm_tmpl_sort(struct xfrm_tmpl **dst, struct xfrm_tmpl **src, int n,
1410 unsigned short family)
1413 struct xfrm_state_afinfo *afinfo = xfrm_state_get_afinfo(family);
1415 return -EAFNOSUPPORT;
1417 spin_lock_bh(&xfrm_state_lock);
1418 if (afinfo->tmpl_sort)
1419 err = afinfo->tmpl_sort(dst, src, n);
1420 spin_unlock_bh(&xfrm_state_lock);
1421 xfrm_state_put_afinfo(afinfo);
1424 EXPORT_SYMBOL(xfrm_tmpl_sort);
1427 xfrm_state_sort(struct xfrm_state **dst, struct xfrm_state **src, int n,
1428 unsigned short family)
1431 struct xfrm_state_afinfo *afinfo = xfrm_state_get_afinfo(family);
1433 return -EAFNOSUPPORT;
1435 spin_lock_bh(&xfrm_state_lock);
1436 if (afinfo->state_sort)
1437 err = afinfo->state_sort(dst, src, n);
1438 spin_unlock_bh(&xfrm_state_lock);
1439 xfrm_state_put_afinfo(afinfo);
1442 EXPORT_SYMBOL(xfrm_state_sort);
1445 /* Silly enough, but I'm lazy to build resolution list */
1447 static struct xfrm_state *__xfrm_find_acq_byseq(struct net *net, u32 mark, u32 seq)
1451 for (i = 0; i <= net->xfrm.state_hmask; i++) {
1452 struct hlist_node *entry;
1453 struct xfrm_state *x;
1455 hlist_for_each_entry(x, entry, net->xfrm.state_bydst+i, bydst) {
1456 if (x->km.seq == seq &&
1457 (mark & x->mark.m) == x->mark.v &&
1458 x->km.state == XFRM_STATE_ACQ) {
1467 struct xfrm_state *xfrm_find_acq_byseq(struct net *net, u32 mark, u32 seq)
1469 struct xfrm_state *x;
1471 spin_lock_bh(&xfrm_state_lock);
1472 x = __xfrm_find_acq_byseq(net, mark, seq);
1473 spin_unlock_bh(&xfrm_state_lock);
1476 EXPORT_SYMBOL(xfrm_find_acq_byseq);
1478 u32 xfrm_get_acqseq(void)
1481 static atomic_t acqseq;
1484 res = atomic_inc_return(&acqseq);
1489 EXPORT_SYMBOL(xfrm_get_acqseq);
1491 int xfrm_alloc_spi(struct xfrm_state *x, u32 low, u32 high)
1493 struct net *net = xs_net(x);
1495 struct xfrm_state *x0;
1497 __be32 minspi = htonl(low);
1498 __be32 maxspi = htonl(high);
1499 u32 mark = x->mark.v & x->mark.m;
1501 spin_lock_bh(&x->lock);
1502 if (x->km.state == XFRM_STATE_DEAD)
1511 if (minspi == maxspi) {
1512 x0 = xfrm_state_lookup(net, mark, &x->id.daddr, minspi, x->id.proto, x->props.family);
1520 for (h=0; h<high-low+1; h++) {
1521 spi = low + net_random()%(high-low+1);
1522 x0 = xfrm_state_lookup(net, mark, &x->id.daddr, htonl(spi), x->id.proto, x->props.family);
1524 x->id.spi = htonl(spi);
1531 spin_lock_bh(&xfrm_state_lock);
1532 h = xfrm_spi_hash(net, &x->id.daddr, x->id.spi, x->id.proto, x->props.family);
1533 hlist_add_head(&x->byspi, net->xfrm.state_byspi+h);
1534 spin_unlock_bh(&xfrm_state_lock);
1540 spin_unlock_bh(&x->lock);
1544 EXPORT_SYMBOL(xfrm_alloc_spi);
1546 int xfrm_state_walk(struct net *net, struct xfrm_state_walk *walk,
1547 int (*func)(struct xfrm_state *, int, void*),
1550 struct xfrm_state *state;
1551 struct xfrm_state_walk *x;
1554 if (walk->seq != 0 && list_empty(&walk->all))
1557 spin_lock_bh(&xfrm_state_lock);
1558 if (list_empty(&walk->all))
1559 x = list_first_entry(&net->xfrm.state_all, struct xfrm_state_walk, all);
1561 x = list_entry(&walk->all, struct xfrm_state_walk, all);
1562 list_for_each_entry_from(x, &net->xfrm.state_all, all) {
1563 if (x->state == XFRM_STATE_DEAD)
1565 state = container_of(x, struct xfrm_state, km);
1566 if (!xfrm_id_proto_match(state->id.proto, walk->proto))
1568 err = func(state, walk->seq, data);
1570 list_move_tail(&walk->all, &x->all);
1575 if (walk->seq == 0) {
1579 list_del_init(&walk->all);
1581 spin_unlock_bh(&xfrm_state_lock);
1584 EXPORT_SYMBOL(xfrm_state_walk);
1586 void xfrm_state_walk_init(struct xfrm_state_walk *walk, u8 proto)
1588 INIT_LIST_HEAD(&walk->all);
1589 walk->proto = proto;
1590 walk->state = XFRM_STATE_DEAD;
1593 EXPORT_SYMBOL(xfrm_state_walk_init);
1595 void xfrm_state_walk_done(struct xfrm_state_walk *walk)
1597 if (list_empty(&walk->all))
1600 spin_lock_bh(&xfrm_state_lock);
1601 list_del(&walk->all);
1602 spin_unlock_bh(&xfrm_state_lock);
1604 EXPORT_SYMBOL(xfrm_state_walk_done);
1607 void xfrm_replay_notify(struct xfrm_state *x, int event)
1610 /* we send notify messages in case
1611 * 1. we updated on of the sequence numbers, and the seqno difference
1612 * is at least x->replay_maxdiff, in this case we also update the
1613 * timeout of our timer function
1614 * 2. if x->replay_maxage has elapsed since last update,
1615 * and there were changes
1617 * The state structure must be locked!
1621 case XFRM_REPLAY_UPDATE:
1622 if (x->replay_maxdiff &&
1623 (x->replay.seq - x->preplay.seq < x->replay_maxdiff) &&
1624 (x->replay.oseq - x->preplay.oseq < x->replay_maxdiff)) {
1625 if (x->xflags & XFRM_TIME_DEFER)
1626 event = XFRM_REPLAY_TIMEOUT;
1633 case XFRM_REPLAY_TIMEOUT:
1634 if ((x->replay.seq == x->preplay.seq) &&
1635 (x->replay.bitmap == x->preplay.bitmap) &&
1636 (x->replay.oseq == x->preplay.oseq)) {
1637 x->xflags |= XFRM_TIME_DEFER;
1644 memcpy(&x->preplay, &x->replay, sizeof(struct xfrm_replay_state));
1645 c.event = XFRM_MSG_NEWAE;
1646 c.data.aevent = event;
1647 km_state_notify(x, &c);
1649 if (x->replay_maxage &&
1650 !mod_timer(&x->rtimer, jiffies + x->replay_maxage))
1651 x->xflags &= ~XFRM_TIME_DEFER;
1654 static void xfrm_replay_timer_handler(unsigned long data)
1656 struct xfrm_state *x = (struct xfrm_state*)data;
1658 spin_lock(&x->lock);
1660 if (x->km.state == XFRM_STATE_VALID) {
1661 if (xfrm_aevent_is_on(xs_net(x)))
1662 xfrm_replay_notify(x, XFRM_REPLAY_TIMEOUT);
1664 x->xflags |= XFRM_TIME_DEFER;
1667 spin_unlock(&x->lock);
1670 int xfrm_replay_check(struct xfrm_state *x,
1671 struct sk_buff *skb, __be32 net_seq)
1674 u32 seq = ntohl(net_seq);
1676 if (unlikely(seq == 0))
1679 if (likely(seq > x->replay.seq))
1682 diff = x->replay.seq - seq;
1683 if (diff >= min_t(unsigned int, x->props.replay_window,
1684 sizeof(x->replay.bitmap) * 8)) {
1685 x->stats.replay_window++;
1689 if (x->replay.bitmap & (1U << diff)) {
1696 xfrm_audit_state_replay(x, skb, net_seq);
1700 void xfrm_replay_advance(struct xfrm_state *x, __be32 net_seq)
1703 u32 seq = ntohl(net_seq);
1705 if (seq > x->replay.seq) {
1706 diff = seq - x->replay.seq;
1707 if (diff < x->props.replay_window)
1708 x->replay.bitmap = ((x->replay.bitmap) << diff) | 1;
1710 x->replay.bitmap = 1;
1711 x->replay.seq = seq;
1713 diff = x->replay.seq - seq;
1714 x->replay.bitmap |= (1U << diff);
1717 if (xfrm_aevent_is_on(xs_net(x)))
1718 xfrm_replay_notify(x, XFRM_REPLAY_UPDATE);
1721 static LIST_HEAD(xfrm_km_list);
1722 static DEFINE_RWLOCK(xfrm_km_lock);
1724 void km_policy_notify(struct xfrm_policy *xp, int dir, struct km_event *c)
1726 struct xfrm_mgr *km;
1728 read_lock(&xfrm_km_lock);
1729 list_for_each_entry(km, &xfrm_km_list, list)
1730 if (km->notify_policy)
1731 km->notify_policy(xp, dir, c);
1732 read_unlock(&xfrm_km_lock);
1735 void km_state_notify(struct xfrm_state *x, struct km_event *c)
1737 struct xfrm_mgr *km;
1738 read_lock(&xfrm_km_lock);
1739 list_for_each_entry(km, &xfrm_km_list, list)
1742 read_unlock(&xfrm_km_lock);
1745 EXPORT_SYMBOL(km_policy_notify);
1746 EXPORT_SYMBOL(km_state_notify);
1748 void km_state_expired(struct xfrm_state *x, int hard, u32 pid)
1750 struct net *net = xs_net(x);
1755 c.event = XFRM_MSG_EXPIRE;
1756 km_state_notify(x, &c);
1759 wake_up(&net->xfrm.km_waitq);
1762 EXPORT_SYMBOL(km_state_expired);
1764 * We send to all registered managers regardless of failure
1765 * We are happy with one success
1767 int km_query(struct xfrm_state *x, struct xfrm_tmpl *t, struct xfrm_policy *pol)
1769 int err = -EINVAL, acqret;
1770 struct xfrm_mgr *km;
1772 read_lock(&xfrm_km_lock);
1773 list_for_each_entry(km, &xfrm_km_list, list) {
1774 acqret = km->acquire(x, t, pol, XFRM_POLICY_OUT);
1778 read_unlock(&xfrm_km_lock);
1781 EXPORT_SYMBOL(km_query);
1783 int km_new_mapping(struct xfrm_state *x, xfrm_address_t *ipaddr, __be16 sport)
1786 struct xfrm_mgr *km;
1788 read_lock(&xfrm_km_lock);
1789 list_for_each_entry(km, &xfrm_km_list, list) {
1790 if (km->new_mapping)
1791 err = km->new_mapping(x, ipaddr, sport);
1795 read_unlock(&xfrm_km_lock);
1798 EXPORT_SYMBOL(km_new_mapping);
1800 void km_policy_expired(struct xfrm_policy *pol, int dir, int hard, u32 pid)
1802 struct net *net = xp_net(pol);
1807 c.event = XFRM_MSG_POLEXPIRE;
1808 km_policy_notify(pol, dir, &c);
1811 wake_up(&net->xfrm.km_waitq);
1813 EXPORT_SYMBOL(km_policy_expired);
1815 #ifdef CONFIG_XFRM_MIGRATE
1816 int km_migrate(struct xfrm_selector *sel, u8 dir, u8 type,
1817 struct xfrm_migrate *m, int num_migrate,
1818 struct xfrm_kmaddress *k)
1822 struct xfrm_mgr *km;
1824 read_lock(&xfrm_km_lock);
1825 list_for_each_entry(km, &xfrm_km_list, list) {
1827 ret = km->migrate(sel, dir, type, m, num_migrate, k);
1832 read_unlock(&xfrm_km_lock);
1835 EXPORT_SYMBOL(km_migrate);
1838 int km_report(struct net *net, u8 proto, struct xfrm_selector *sel, xfrm_address_t *addr)
1842 struct xfrm_mgr *km;
1844 read_lock(&xfrm_km_lock);
1845 list_for_each_entry(km, &xfrm_km_list, list) {
1847 ret = km->report(net, proto, sel, addr);
1852 read_unlock(&xfrm_km_lock);
1855 EXPORT_SYMBOL(km_report);
1857 int xfrm_user_policy(struct sock *sk, int optname, u8 __user *optval, int optlen)
1861 struct xfrm_mgr *km;
1862 struct xfrm_policy *pol = NULL;
1864 if (optlen <= 0 || optlen > PAGE_SIZE)
1867 data = kmalloc(optlen, GFP_KERNEL);
1872 if (copy_from_user(data, optval, optlen))
1876 read_lock(&xfrm_km_lock);
1877 list_for_each_entry(km, &xfrm_km_list, list) {
1878 pol = km->compile_policy(sk, optname, data,
1883 read_unlock(&xfrm_km_lock);
1886 xfrm_sk_policy_insert(sk, err, pol);
1895 EXPORT_SYMBOL(xfrm_user_policy);
1897 int xfrm_register_km(struct xfrm_mgr *km)
1899 write_lock_bh(&xfrm_km_lock);
1900 list_add_tail(&km->list, &xfrm_km_list);
1901 write_unlock_bh(&xfrm_km_lock);
1904 EXPORT_SYMBOL(xfrm_register_km);
1906 int xfrm_unregister_km(struct xfrm_mgr *km)
1908 write_lock_bh(&xfrm_km_lock);
1909 list_del(&km->list);
1910 write_unlock_bh(&xfrm_km_lock);
1913 EXPORT_SYMBOL(xfrm_unregister_km);
1915 int xfrm_state_register_afinfo(struct xfrm_state_afinfo *afinfo)
1918 if (unlikely(afinfo == NULL))
1920 if (unlikely(afinfo->family >= NPROTO))
1921 return -EAFNOSUPPORT;
1922 write_lock_bh(&xfrm_state_afinfo_lock);
1923 if (unlikely(xfrm_state_afinfo[afinfo->family] != NULL))
1926 xfrm_state_afinfo[afinfo->family] = afinfo;
1927 write_unlock_bh(&xfrm_state_afinfo_lock);
1930 EXPORT_SYMBOL(xfrm_state_register_afinfo);
1932 int xfrm_state_unregister_afinfo(struct xfrm_state_afinfo *afinfo)
1935 if (unlikely(afinfo == NULL))
1937 if (unlikely(afinfo->family >= NPROTO))
1938 return -EAFNOSUPPORT;
1939 write_lock_bh(&xfrm_state_afinfo_lock);
1940 if (likely(xfrm_state_afinfo[afinfo->family] != NULL)) {
1941 if (unlikely(xfrm_state_afinfo[afinfo->family] != afinfo))
1944 xfrm_state_afinfo[afinfo->family] = NULL;
1946 write_unlock_bh(&xfrm_state_afinfo_lock);
1949 EXPORT_SYMBOL(xfrm_state_unregister_afinfo);
1951 static struct xfrm_state_afinfo *xfrm_state_get_afinfo(unsigned int family)
1953 struct xfrm_state_afinfo *afinfo;
1954 if (unlikely(family >= NPROTO))
1956 read_lock(&xfrm_state_afinfo_lock);
1957 afinfo = xfrm_state_afinfo[family];
1958 if (unlikely(!afinfo))
1959 read_unlock(&xfrm_state_afinfo_lock);
1963 static void xfrm_state_put_afinfo(struct xfrm_state_afinfo *afinfo)
1964 __releases(xfrm_state_afinfo_lock)
1966 read_unlock(&xfrm_state_afinfo_lock);
1969 /* Temporarily located here until net/xfrm/xfrm_tunnel.c is created */
1970 void xfrm_state_delete_tunnel(struct xfrm_state *x)
1973 struct xfrm_state *t = x->tunnel;
1975 if (atomic_read(&t->tunnel_users) == 2)
1976 xfrm_state_delete(t);
1977 atomic_dec(&t->tunnel_users);
1982 EXPORT_SYMBOL(xfrm_state_delete_tunnel);
1984 int xfrm_state_mtu(struct xfrm_state *x, int mtu)
1988 spin_lock_bh(&x->lock);
1989 if (x->km.state == XFRM_STATE_VALID &&
1990 x->type && x->type->get_mtu)
1991 res = x->type->get_mtu(x, mtu);
1993 res = mtu - x->props.header_len;
1994 spin_unlock_bh(&x->lock);
1998 int xfrm_init_state(struct xfrm_state *x)
2000 struct xfrm_state_afinfo *afinfo;
2001 struct xfrm_mode *inner_mode;
2002 int family = x->props.family;
2005 err = -EAFNOSUPPORT;
2006 afinfo = xfrm_state_get_afinfo(family);
2011 if (afinfo->init_flags)
2012 err = afinfo->init_flags(x);
2014 xfrm_state_put_afinfo(afinfo);
2019 err = -EPROTONOSUPPORT;
2021 if (x->sel.family != AF_UNSPEC) {
2022 inner_mode = xfrm_get_mode(x->props.mode, x->sel.family);
2023 if (inner_mode == NULL)
2026 if (!(inner_mode->flags & XFRM_MODE_FLAG_TUNNEL) &&
2027 family != x->sel.family) {
2028 xfrm_put_mode(inner_mode);
2032 x->inner_mode = inner_mode;
2034 struct xfrm_mode *inner_mode_iaf;
2035 int iafamily = AF_INET;
2037 inner_mode = xfrm_get_mode(x->props.mode, x->props.family);
2038 if (inner_mode == NULL)
2041 if (!(inner_mode->flags & XFRM_MODE_FLAG_TUNNEL)) {
2042 xfrm_put_mode(inner_mode);
2045 x->inner_mode = inner_mode;
2047 if (x->props.family == AF_INET)
2048 iafamily = AF_INET6;
2050 inner_mode_iaf = xfrm_get_mode(x->props.mode, iafamily);
2051 if (inner_mode_iaf) {
2052 if (inner_mode_iaf->flags & XFRM_MODE_FLAG_TUNNEL)
2053 x->inner_mode_iaf = inner_mode_iaf;
2055 xfrm_put_mode(inner_mode_iaf);
2059 x->type = xfrm_get_type(x->id.proto, family);
2060 if (x->type == NULL)
2063 err = x->type->init_state(x);
2067 x->outer_mode = xfrm_get_mode(x->props.mode, family);
2068 if (x->outer_mode == NULL)
2071 x->km.state = XFRM_STATE_VALID;
2077 EXPORT_SYMBOL(xfrm_init_state);
2079 int __net_init xfrm_state_init(struct net *net)
2083 INIT_LIST_HEAD(&net->xfrm.state_all);
2085 sz = sizeof(struct hlist_head) * 8;
2087 net->xfrm.state_bydst = xfrm_hash_alloc(sz);
2088 if (!net->xfrm.state_bydst)
2090 net->xfrm.state_bysrc = xfrm_hash_alloc(sz);
2091 if (!net->xfrm.state_bysrc)
2093 net->xfrm.state_byspi = xfrm_hash_alloc(sz);
2094 if (!net->xfrm.state_byspi)
2096 net->xfrm.state_hmask = ((sz / sizeof(struct hlist_head)) - 1);
2098 net->xfrm.state_num = 0;
2099 INIT_WORK(&net->xfrm.state_hash_work, xfrm_hash_resize);
2100 INIT_HLIST_HEAD(&net->xfrm.state_gc_list);
2101 INIT_WORK(&net->xfrm.state_gc_work, xfrm_state_gc_task);
2102 init_waitqueue_head(&net->xfrm.km_waitq);
2106 xfrm_hash_free(net->xfrm.state_bysrc, sz);
2108 xfrm_hash_free(net->xfrm.state_bydst, sz);
2113 void xfrm_state_fini(struct net *net)
2115 struct xfrm_audit audit_info;
2118 flush_work(&net->xfrm.state_hash_work);
2119 audit_info.loginuid = -1;
2120 audit_info.sessionid = -1;
2121 audit_info.secid = 0;
2122 xfrm_state_flush(net, IPSEC_PROTO_ANY, &audit_info);
2123 flush_work(&net->xfrm.state_gc_work);
2125 WARN_ON(!list_empty(&net->xfrm.state_all));
2127 sz = (net->xfrm.state_hmask + 1) * sizeof(struct hlist_head);
2128 WARN_ON(!hlist_empty(net->xfrm.state_byspi));
2129 xfrm_hash_free(net->xfrm.state_byspi, sz);
2130 WARN_ON(!hlist_empty(net->xfrm.state_bysrc));
2131 xfrm_hash_free(net->xfrm.state_bysrc, sz);
2132 WARN_ON(!hlist_empty(net->xfrm.state_bydst));
2133 xfrm_hash_free(net->xfrm.state_bydst, sz);
2136 #ifdef CONFIG_AUDITSYSCALL
2137 static void xfrm_audit_helper_sainfo(struct xfrm_state *x,
2138 struct audit_buffer *audit_buf)
2140 struct xfrm_sec_ctx *ctx = x->security;
2141 u32 spi = ntohl(x->id.spi);
2144 audit_log_format(audit_buf, " sec_alg=%u sec_doi=%u sec_obj=%s",
2145 ctx->ctx_alg, ctx->ctx_doi, ctx->ctx_str);
2147 switch(x->props.family) {
2149 audit_log_format(audit_buf, " src=%pI4 dst=%pI4",
2150 &x->props.saddr.a4, &x->id.daddr.a4);
2153 audit_log_format(audit_buf, " src=%pI6 dst=%pI6",
2154 x->props.saddr.a6, x->id.daddr.a6);
2158 audit_log_format(audit_buf, " spi=%u(0x%x)", spi, spi);
2161 static void xfrm_audit_helper_pktinfo(struct sk_buff *skb, u16 family,
2162 struct audit_buffer *audit_buf)
2165 struct ipv6hdr *iph6;
2170 audit_log_format(audit_buf, " src=%pI4 dst=%pI4",
2171 &iph4->saddr, &iph4->daddr);
2174 iph6 = ipv6_hdr(skb);
2175 audit_log_format(audit_buf,
2176 " src=%pI6 dst=%pI6 flowlbl=0x%x%02x%02x",
2177 &iph6->saddr,&iph6->daddr,
2178 iph6->flow_lbl[0] & 0x0f,
2185 void xfrm_audit_state_add(struct xfrm_state *x, int result,
2186 uid_t auid, u32 sessionid, u32 secid)
2188 struct audit_buffer *audit_buf;
2190 audit_buf = xfrm_audit_start("SAD-add");
2191 if (audit_buf == NULL)
2193 xfrm_audit_helper_usrinfo(auid, sessionid, secid, audit_buf);
2194 xfrm_audit_helper_sainfo(x, audit_buf);
2195 audit_log_format(audit_buf, " res=%u", result);
2196 audit_log_end(audit_buf);
2198 EXPORT_SYMBOL_GPL(xfrm_audit_state_add);
2200 void xfrm_audit_state_delete(struct xfrm_state *x, int result,
2201 uid_t auid, u32 sessionid, u32 secid)
2203 struct audit_buffer *audit_buf;
2205 audit_buf = xfrm_audit_start("SAD-delete");
2206 if (audit_buf == NULL)
2208 xfrm_audit_helper_usrinfo(auid, sessionid, secid, audit_buf);
2209 xfrm_audit_helper_sainfo(x, audit_buf);
2210 audit_log_format(audit_buf, " res=%u", result);
2211 audit_log_end(audit_buf);
2213 EXPORT_SYMBOL_GPL(xfrm_audit_state_delete);
2215 void xfrm_audit_state_replay_overflow(struct xfrm_state *x,
2216 struct sk_buff *skb)
2218 struct audit_buffer *audit_buf;
2221 audit_buf = xfrm_audit_start("SA-replay-overflow");
2222 if (audit_buf == NULL)
2224 xfrm_audit_helper_pktinfo(skb, x->props.family, audit_buf);
2225 /* don't record the sequence number because it's inherent in this kind
2226 * of audit message */
2227 spi = ntohl(x->id.spi);
2228 audit_log_format(audit_buf, " spi=%u(0x%x)", spi, spi);
2229 audit_log_end(audit_buf);
2231 EXPORT_SYMBOL_GPL(xfrm_audit_state_replay_overflow);
2233 static void xfrm_audit_state_replay(struct xfrm_state *x,
2234 struct sk_buff *skb, __be32 net_seq)
2236 struct audit_buffer *audit_buf;
2239 audit_buf = xfrm_audit_start("SA-replayed-pkt");
2240 if (audit_buf == NULL)
2242 xfrm_audit_helper_pktinfo(skb, x->props.family, audit_buf);
2243 spi = ntohl(x->id.spi);
2244 audit_log_format(audit_buf, " spi=%u(0x%x) seqno=%u",
2245 spi, spi, ntohl(net_seq));
2246 audit_log_end(audit_buf);
2249 void xfrm_audit_state_notfound_simple(struct sk_buff *skb, u16 family)
2251 struct audit_buffer *audit_buf;
2253 audit_buf = xfrm_audit_start("SA-notfound");
2254 if (audit_buf == NULL)
2256 xfrm_audit_helper_pktinfo(skb, family, audit_buf);
2257 audit_log_end(audit_buf);
2259 EXPORT_SYMBOL_GPL(xfrm_audit_state_notfound_simple);
2261 void xfrm_audit_state_notfound(struct sk_buff *skb, u16 family,
2262 __be32 net_spi, __be32 net_seq)
2264 struct audit_buffer *audit_buf;
2267 audit_buf = xfrm_audit_start("SA-notfound");
2268 if (audit_buf == NULL)
2270 xfrm_audit_helper_pktinfo(skb, family, audit_buf);
2271 spi = ntohl(net_spi);
2272 audit_log_format(audit_buf, " spi=%u(0x%x) seqno=%u",
2273 spi, spi, ntohl(net_seq));
2274 audit_log_end(audit_buf);
2276 EXPORT_SYMBOL_GPL(xfrm_audit_state_notfound);
2278 void xfrm_audit_state_icvfail(struct xfrm_state *x,
2279 struct sk_buff *skb, u8 proto)
2281 struct audit_buffer *audit_buf;
2285 audit_buf = xfrm_audit_start("SA-icv-failure");
2286 if (audit_buf == NULL)
2288 xfrm_audit_helper_pktinfo(skb, x->props.family, audit_buf);
2289 if (xfrm_parse_spi(skb, proto, &net_spi, &net_seq) == 0) {
2290 u32 spi = ntohl(net_spi);
2291 audit_log_format(audit_buf, " spi=%u(0x%x) seqno=%u",
2292 spi, spi, ntohl(net_seq));
2294 audit_log_end(audit_buf);
2296 EXPORT_SYMBOL_GPL(xfrm_audit_state_icvfail);
2297 #endif /* CONFIG_AUDITSYSCALL */