nfsd: nfsd should drop CAP_MKNOD for non-root
[safe/jmp/linux-2.6] / net / xfrm / xfrm_state.c
index f7c0951..62a5425 100644 (file)
 #include <linux/ipsec.h>
 #include <linux/module.h>
 #include <linux/cache.h>
+#include <linux/audit.h>
 #include <asm/uaccess.h>
 
 #include "xfrm_hash.h"
 
-struct sock *xfrm_nl;
-EXPORT_SYMBOL(xfrm_nl);
-
-u32 sysctl_xfrm_aevent_etime __read_mostly = XFRM_AE_ETIME;
-EXPORT_SYMBOL(sysctl_xfrm_aevent_etime);
-
-u32 sysctl_xfrm_aevent_rseqth __read_mostly = XFRM_AE_SEQT_SIZE;
-EXPORT_SYMBOL(sysctl_xfrm_aevent_rseqth);
-
-u32 sysctl_xfrm_acq_expires __read_mostly = 30;
-
 /* Each xfrm_state may be linked to two tables:
 
    1. Hash table by (spi,daddr,ah/esp) to find SA by SPI. (input,ctl)
@@ -43,42 +33,40 @@ u32 sysctl_xfrm_acq_expires __read_mostly = 30;
 
 static DEFINE_SPINLOCK(xfrm_state_lock);
 
-/* Hash table to find appropriate SA towards given target (endpoint
- * of tunnel or destination of transport mode) allowed by selector.
- *
- * Main use is finding SA after policy selected tunnel or transport mode.
- * Also, it can be used by ah/esp icmp error handler to find offending SA.
- */
-static struct hlist_head *xfrm_state_bydst __read_mostly;
-static struct hlist_head *xfrm_state_bysrc __read_mostly;
-static struct hlist_head *xfrm_state_byspi __read_mostly;
-static unsigned int xfrm_state_hmask __read_mostly;
 static unsigned int xfrm_state_hashmax __read_mostly = 1 * 1024 * 1024;
-static unsigned int xfrm_state_num;
 static unsigned int xfrm_state_genid;
 
 static struct xfrm_state_afinfo *xfrm_state_get_afinfo(unsigned int family);
 static void xfrm_state_put_afinfo(struct xfrm_state_afinfo *afinfo);
 
-static inline unsigned int xfrm_dst_hash(xfrm_address_t *daddr,
+#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 */
+
+static inline unsigned int xfrm_dst_hash(struct net *net,
+                                        xfrm_address_t *daddr,
                                         xfrm_address_t *saddr,
                                         u32 reqid,
                                         unsigned short family)
 {
-       return __xfrm_dst_hash(daddr, saddr, reqid, family, xfrm_state_hmask);
+       return __xfrm_dst_hash(daddr, saddr, reqid, family, net->xfrm.state_hmask);
 }
 
-static inline unsigned int xfrm_src_hash(xfrm_address_t *daddr,
+static inline unsigned int xfrm_src_hash(struct net *net,
+                                        xfrm_address_t *daddr,
                                         xfrm_address_t *saddr,
                                         unsigned short family)
 {
-       return __xfrm_src_hash(daddr, saddr, family, xfrm_state_hmask);
+       return __xfrm_src_hash(daddr, saddr, family, net->xfrm.state_hmask);
 }
 
 static inline unsigned int
-xfrm_spi_hash(xfrm_address_t *daddr, __be32 spi, u8 proto, unsigned short family)
+xfrm_spi_hash(struct net *net, xfrm_address_t *daddr, __be32 spi, u8 proto, unsigned short family)
 {
-       return __xfrm_spi_hash(daddr, spi, proto, family, xfrm_state_hmask);
+       return __xfrm_spi_hash(daddr, spi, proto, family, net->xfrm.state_hmask);
 }
 
 static void xfrm_hash_transfer(struct hlist_head *list,
@@ -112,16 +100,16 @@ static void xfrm_hash_transfer(struct hlist_head *list,
        }
 }
 
-static unsigned long xfrm_hash_new_size(void)
+static unsigned long xfrm_hash_new_size(unsigned int state_hmask)
 {
-       return ((xfrm_state_hmask + 1) << 1) *
-               sizeof(struct hlist_head);
+       return ((state_hmask + 1) << 1) * sizeof(struct hlist_head);
 }
 
 static DEFINE_MUTEX(hash_resize_mutex);
 
-static void xfrm_hash_resize(struct work_struct *__unused)
+static void xfrm_hash_resize(struct work_struct *work)
 {
+       struct net *net = container_of(work, struct net, xfrm.state_hash_work);
        struct hlist_head *ndst, *nsrc, *nspi, *odst, *osrc, *ospi;
        unsigned long nsize, osize;
        unsigned int nhashmask, ohashmask;
@@ -129,7 +117,7 @@ static void xfrm_hash_resize(struct work_struct *__unused)
 
        mutex_lock(&hash_resize_mutex);
 
-       nsize = xfrm_hash_new_size();
+       nsize = xfrm_hash_new_size(net->xfrm.state_hmask);
        ndst = xfrm_hash_alloc(nsize);
        if (!ndst)
                goto out_unlock;
@@ -148,19 +136,19 @@ static void xfrm_hash_resize(struct work_struct *__unused)
        spin_lock_bh(&xfrm_state_lock);
 
        nhashmask = (nsize / sizeof(struct hlist_head)) - 1U;
-       for (i = xfrm_state_hmask; i >= 0; i--)
-               xfrm_hash_transfer(xfrm_state_bydst+i, ndst, nsrc, nspi,
+       for (i = net->xfrm.state_hmask; i >= 0; i--)
+               xfrm_hash_transfer(net->xfrm.state_bydst+i, ndst, nsrc, nspi,
                                   nhashmask);
 
-       odst = xfrm_state_bydst;
-       osrc = xfrm_state_bysrc;
-       ospi = xfrm_state_byspi;
-       ohashmask = xfrm_state_hmask;
+       odst = net->xfrm.state_bydst;
+       osrc = net->xfrm.state_bysrc;
+       ospi = net->xfrm.state_byspi;
+       ohashmask = net->xfrm.state_hmask;
 
-       xfrm_state_bydst = ndst;
-       xfrm_state_bysrc = nsrc;
-       xfrm_state_byspi = nspi;
-       xfrm_state_hmask = nhashmask;
+       net->xfrm.state_bydst = ndst;
+       net->xfrm.state_bysrc = nsrc;
+       net->xfrm.state_byspi = nspi;
+       net->xfrm.state_hmask = nhashmask;
 
        spin_unlock_bh(&xfrm_state_lock);
 
@@ -173,16 +161,9 @@ out_unlock:
        mutex_unlock(&hash_resize_mutex);
 }
 
-static DECLARE_WORK(xfrm_hash_work, xfrm_hash_resize);
-
-DECLARE_WAIT_QUEUE_HEAD(km_waitq);
-EXPORT_SYMBOL(km_waitq);
-
 static DEFINE_RWLOCK(xfrm_state_afinfo_lock);
 static struct xfrm_state_afinfo *xfrm_state_afinfo[NPROTO];
 
-static struct work_struct xfrm_state_gc_work;
-static HLIST_HEAD(xfrm_state_gc_list);
 static DEFINE_SPINLOCK(xfrm_state_gc_lock);
 
 int __xfrm_state_delete(struct xfrm_state *x);
@@ -203,14 +184,15 @@ static struct xfrm_state_afinfo *xfrm_state_lock_afinfo(unsigned int family)
 }
 
 static void xfrm_state_unlock_afinfo(struct xfrm_state_afinfo *afinfo)
+       __releases(xfrm_state_afinfo_lock)
 {
        write_unlock_bh(&xfrm_state_afinfo_lock);
 }
 
-int xfrm_register_type(struct xfrm_type *type, unsigned short family)
+int xfrm_register_type(const struct xfrm_type *type, unsigned short family)
 {
        struct xfrm_state_afinfo *afinfo = xfrm_state_lock_afinfo(family);
-       struct xfrm_type **typemap;
+       const struct xfrm_type **typemap;
        int err = 0;
 
        if (unlikely(afinfo == NULL))
@@ -226,10 +208,10 @@ int xfrm_register_type(struct xfrm_type *type, unsigned short family)
 }
 EXPORT_SYMBOL(xfrm_register_type);
 
-int xfrm_unregister_type(struct xfrm_type *type, unsigned short family)
+int xfrm_unregister_type(const struct xfrm_type *type, unsigned short family)
 {
        struct xfrm_state_afinfo *afinfo = xfrm_state_lock_afinfo(family);
-       struct xfrm_type **typemap;
+       const struct xfrm_type **typemap;
        int err = 0;
 
        if (unlikely(afinfo == NULL))
@@ -245,11 +227,11 @@ int xfrm_unregister_type(struct xfrm_type *type, unsigned short family)
 }
 EXPORT_SYMBOL(xfrm_unregister_type);
 
-static struct xfrm_type *xfrm_get_type(u8 proto, unsigned short family)
+static const struct xfrm_type *xfrm_get_type(u8 proto, unsigned short family)
 {
        struct xfrm_state_afinfo *afinfo;
-       struct xfrm_type **typemap;
-       struct xfrm_type *type;
+       const struct xfrm_type **typemap;
+       const struct xfrm_type *type;
        int modload_attempted = 0;
 
 retry:
@@ -272,7 +254,7 @@ retry:
        return type;
 }
 
-static void xfrm_put_type(struct xfrm_type *type)
+static void xfrm_put_type(const struct xfrm_type *type)
 {
        module_put(type->owner);
 }
@@ -379,6 +361,8 @@ static void xfrm_state_gc_destroy(struct xfrm_state *x)
        kfree(x->coaddr);
        if (x->inner_mode)
                xfrm_put_mode(x->inner_mode);
+       if (x->inner_mode_iaf)
+               xfrm_put_mode(x->inner_mode_iaf);
        if (x->outer_mode)
                xfrm_put_mode(x->outer_mode);
        if (x->type) {
@@ -389,21 +373,21 @@ static void xfrm_state_gc_destroy(struct xfrm_state *x)
        kfree(x);
 }
 
-static void xfrm_state_gc_task(struct work_struct *data)
+static void xfrm_state_gc_task(struct work_struct *work)
 {
+       struct net *net = container_of(work, struct net, xfrm.state_gc_work);
        struct xfrm_state *x;
        struct hlist_node *entry, *tmp;
        struct hlist_head gc_list;
 
        spin_lock_bh(&xfrm_state_gc_lock);
-       gc_list.first = xfrm_state_gc_list.first;
-       INIT_HLIST_HEAD(&xfrm_state_gc_list);
+       hlist_move_list(&net->xfrm.state_gc_list, &gc_list);
        spin_unlock_bh(&xfrm_state_gc_lock);
 
-       hlist_for_each_entry_safe(x, entry, tmp, &gc_list, bydst)
+       hlist_for_each_entry_safe(x, entry, tmp, &gc_list, gclist)
                xfrm_state_gc_destroy(x);
 
-       wake_up(&km_waitq);
+       wake_up(&net->xfrm.km_waitq);
 }
 
 static inline unsigned long make_jiffies(long secs)
@@ -417,6 +401,7 @@ static inline unsigned long make_jiffies(long secs)
 static void xfrm_timer_handler(unsigned long data)
 {
        struct xfrm_state *x = (struct xfrm_state*)data;
+       struct net *net = xs_net(x);
        unsigned long now = get_seconds();
        long next = LONG_MAX;
        int warn = 0;
@@ -474,7 +459,7 @@ resched:
 expired:
        if (x->km.state == XFRM_STATE_ACQ && x->id.spi == 0) {
                x->km.state = XFRM_STATE_EXPIRED;
-               wake_up(&km_waitq);
+               wake_up(&net->xfrm.km_waitq);
                next = 2;
                goto resched;
        }
@@ -484,7 +469,8 @@ expired:
                km_state_expired(x, 1, 0);
 
        xfrm_audit_state_delete(x, err ? 0 : 1,
-                               audit_get_loginuid(current->audit_context), 0);
+                               audit_get_loginuid(current),
+                               audit_get_sessionid(current), 0);
 
 out:
        spin_unlock(&x->lock);
@@ -492,15 +478,17 @@ out:
 
 static void xfrm_replay_timer_handler(unsigned long data);
 
-struct xfrm_state *xfrm_state_alloc(void)
+struct xfrm_state *xfrm_state_alloc(struct net *net)
 {
        struct xfrm_state *x;
 
        x = kzalloc(sizeof(struct xfrm_state), GFP_ATOMIC);
 
        if (x) {
+               write_pnet(&x->xs_net, net);
                atomic_set(&x->refcnt, 1);
                atomic_set(&x->tunnel_users, 0);
+               INIT_LIST_HEAD(&x->km.all);
                INIT_HLIST_NODE(&x->bydst);
                INIT_HLIST_NODE(&x->bysrc);
                INIT_HLIST_NODE(&x->byspi);
@@ -514,6 +502,8 @@ struct xfrm_state *xfrm_state_alloc(void)
                x->lft.hard_packet_limit = XFRM_INF;
                x->replay_maxage = 0;
                x->replay_maxdiff = 0;
+               x->inner_mode = NULL;
+               x->inner_mode_iaf = NULL;
                spin_lock_init(&x->lock);
        }
        return x;
@@ -522,27 +512,31 @@ EXPORT_SYMBOL(xfrm_state_alloc);
 
 void __xfrm_state_destroy(struct xfrm_state *x)
 {
-       BUG_TRAP(x->km.state == XFRM_STATE_DEAD);
+       struct net *net = xs_net(x);
+
+       WARN_ON(x->km.state != XFRM_STATE_DEAD);
 
        spin_lock_bh(&xfrm_state_gc_lock);
-       hlist_add_head(&x->bydst, &xfrm_state_gc_list);
+       hlist_add_head(&x->gclist, &net->xfrm.state_gc_list);
        spin_unlock_bh(&xfrm_state_gc_lock);
-       schedule_work(&xfrm_state_gc_work);
+       schedule_work(&net->xfrm.state_gc_work);
 }
 EXPORT_SYMBOL(__xfrm_state_destroy);
 
 int __xfrm_state_delete(struct xfrm_state *x)
 {
+       struct net *net = xs_net(x);
        int err = -ESRCH;
 
        if (x->km.state != XFRM_STATE_DEAD) {
                x->km.state = XFRM_STATE_DEAD;
                spin_lock(&xfrm_state_lock);
+               list_del(&x->km.all);
                hlist_del(&x->bydst);
                hlist_del(&x->bysrc);
                if (x->id.spi)
                        hlist_del(&x->byspi);
-               xfrm_state_num--;
+               net->xfrm.state_num--;
                spin_unlock(&xfrm_state_lock);
 
                /* All xfrm_state objects are created by xfrm_state_alloc.
@@ -571,19 +565,20 @@ EXPORT_SYMBOL(xfrm_state_delete);
 
 #ifdef CONFIG_SECURITY_NETWORK_XFRM
 static inline int
-xfrm_state_flush_secctx_check(u8 proto, struct xfrm_audit *audit_info)
+xfrm_state_flush_secctx_check(struct net *net, u8 proto, struct xfrm_audit *audit_info)
 {
        int i, err = 0;
 
-       for (i = 0; i <= xfrm_state_hmask; i++) {
+       for (i = 0; i <= net->xfrm.state_hmask; i++) {
                struct hlist_node *entry;
                struct xfrm_state *x;
 
-               hlist_for_each_entry(x, entry, xfrm_state_bydst+i, bydst) {
+               hlist_for_each_entry(x, entry, net->xfrm.state_bydst+i, bydst) {
                        if (xfrm_id_proto_match(x->id.proto, proto) &&
                           (err = security_xfrm_state_delete(x)) != 0) {
                                xfrm_audit_state_delete(x, 0,
                                                        audit_info->loginuid,
+                                                       audit_info->sessionid,
                                                        audit_info->secid);
                                return err;
                        }
@@ -594,26 +589,26 @@ xfrm_state_flush_secctx_check(u8 proto, struct xfrm_audit *audit_info)
 }
 #else
 static inline int
-xfrm_state_flush_secctx_check(u8 proto, struct xfrm_audit *audit_info)
+xfrm_state_flush_secctx_check(struct net *net, u8 proto, struct xfrm_audit *audit_info)
 {
        return 0;
 }
 #endif
 
-int xfrm_state_flush(u8 proto, struct xfrm_audit *audit_info)
+int xfrm_state_flush(struct net *net, u8 proto, struct xfrm_audit *audit_info)
 {
        int i, err = 0;
 
        spin_lock_bh(&xfrm_state_lock);
-       err = xfrm_state_flush_secctx_check(proto, audit_info);
+       err = xfrm_state_flush_secctx_check(net, proto, audit_info);
        if (err)
                goto out;
 
-       for (i = 0; i <= xfrm_state_hmask; i++) {
+       for (i = 0; i <= net->xfrm.state_hmask; i++) {
                struct hlist_node *entry;
                struct xfrm_state *x;
 restart:
-               hlist_for_each_entry(x, entry, xfrm_state_bydst+i, bydst) {
+               hlist_for_each_entry(x, entry, net->xfrm.state_bydst+i, bydst) {
                        if (!xfrm_state_kern(x) &&
                            xfrm_id_proto_match(x->id.proto, proto)) {
                                xfrm_state_hold(x);
@@ -622,6 +617,7 @@ restart:
                                err = xfrm_state_delete(x);
                                xfrm_audit_state_delete(x, err ? 0 : 1,
                                                        audit_info->loginuid,
+                                                       audit_info->sessionid,
                                                        audit_info->secid);
                                xfrm_state_put(x);
 
@@ -634,7 +630,7 @@ restart:
 
 out:
        spin_unlock_bh(&xfrm_state_lock);
-       wake_up(&km_waitq);
+       wake_up(&net->xfrm.km_waitq);
        return err;
 }
 EXPORT_SYMBOL(xfrm_state_flush);
@@ -642,8 +638,8 @@ EXPORT_SYMBOL(xfrm_state_flush);
 void xfrm_sad_getinfo(struct xfrmk_sadinfo *si)
 {
        spin_lock_bh(&xfrm_state_lock);
-       si->sadcnt = xfrm_state_num;
-       si->sadhcnt = xfrm_state_hmask;
+       si->sadcnt = init_net.xfrm.state_num;
+       si->sadhcnt = init_net.xfrm.state_hmask;
        si->sadhmcnt = xfrm_state_hashmax;
        spin_unlock_bh(&xfrm_state_lock);
 }
@@ -663,13 +659,13 @@ xfrm_init_tempsel(struct xfrm_state *x, struct flowi *fl,
        return 0;
 }
 
-static struct xfrm_state *__xfrm_state_lookup(xfrm_address_t *daddr, __be32 spi, u8 proto, unsigned short family)
+static struct xfrm_state *__xfrm_state_lookup(struct net *net, xfrm_address_t *daddr, __be32 spi, u8 proto, unsigned short family)
 {
-       unsigned int h = xfrm_spi_hash(daddr, spi, proto, family);
+       unsigned int h = xfrm_spi_hash(net, daddr, spi, proto, family);
        struct xfrm_state *x;
        struct hlist_node *entry;
 
-       hlist_for_each_entry(x, entry, xfrm_state_byspi+h, byspi) {
+       hlist_for_each_entry(x, entry, net->xfrm.state_byspi+h, byspi) {
                if (x->props.family != family ||
                    x->id.spi       != spi ||
                    x->id.proto     != proto)
@@ -695,13 +691,13 @@ static struct xfrm_state *__xfrm_state_lookup(xfrm_address_t *daddr, __be32 spi,
        return NULL;
 }
 
-static struct xfrm_state *__xfrm_state_lookup_byaddr(xfrm_address_t *daddr, xfrm_address_t *saddr, u8 proto, unsigned short family)
+static struct xfrm_state *__xfrm_state_lookup_byaddr(struct net *net, xfrm_address_t *daddr, xfrm_address_t *saddr, u8 proto, unsigned short family)
 {
-       unsigned int h = xfrm_src_hash(daddr, saddr, family);
+       unsigned int h = xfrm_src_hash(net, daddr, saddr, family);
        struct xfrm_state *x;
        struct hlist_node *entry;
 
-       hlist_for_each_entry(x, entry, xfrm_state_bysrc+h, bysrc) {
+       hlist_for_each_entry(x, entry, net->xfrm.state_bysrc+h, bysrc) {
                if (x->props.family != family ||
                    x->id.proto     != proto)
                        continue;
@@ -733,21 +729,61 @@ static struct xfrm_state *__xfrm_state_lookup_byaddr(xfrm_address_t *daddr, xfrm
 static inline struct xfrm_state *
 __xfrm_state_locate(struct xfrm_state *x, int use_spi, int family)
 {
+       struct net *net = xs_net(x);
+
        if (use_spi)
-               return __xfrm_state_lookup(&x->id.daddr, x->id.spi,
+               return __xfrm_state_lookup(net, &x->id.daddr, x->id.spi,
                                           x->id.proto, family);
        else
-               return __xfrm_state_lookup_byaddr(&x->id.daddr,
+               return __xfrm_state_lookup_byaddr(net, &x->id.daddr,
                                                  &x->props.saddr,
                                                  x->id.proto, family);
 }
 
-static void xfrm_hash_grow_check(int have_hash_collision)
+static void xfrm_hash_grow_check(struct net *net, int have_hash_collision)
 {
        if (have_hash_collision &&
-           (xfrm_state_hmask + 1) < xfrm_state_hashmax &&
-           xfrm_state_num > xfrm_state_hmask)
-               schedule_work(&xfrm_hash_work);
+           (net->xfrm.state_hmask + 1) < xfrm_state_hashmax &&
+           net->xfrm.state_num > net->xfrm.state_hmask)
+               schedule_work(&net->xfrm.state_hash_work);
+}
+
+static void xfrm_state_look_at(struct xfrm_policy *pol, struct xfrm_state *x,
+                              struct flowi *fl, unsigned short family,
+                              xfrm_address_t *daddr, xfrm_address_t *saddr,
+                              struct xfrm_state **best, int *acq_in_progress,
+                              int *error)
+{
+       /* 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) {
+               if ((x->sel.family &&
+                    !xfrm_selector_match(&x->sel, fl, x->sel.family)) ||
+                   !security_xfrm_state_pol_flow_match(x, pol, fl))
+                       return;
+
+               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) {
+               *acq_in_progress = 1;
+       } else if (x->km.state == XFRM_STATE_ERROR ||
+                  x->km.state == XFRM_STATE_EXPIRED) {
+               if (xfrm_selector_match(&x->sel, fl, x->sel.family) &&
+                   security_xfrm_state_pol_flow_match(x, pol, fl))
+                       *error = -ESRCH;
+       }
 }
 
 struct xfrm_state *
@@ -756,66 +792,57 @@ xfrm_state_find(xfrm_address_t *daddr, xfrm_address_t *saddr,
                struct xfrm_policy *pol, int *err,
                unsigned short family)
 {
+       static xfrm_address_t saddr_wildcard = { };
+       struct net *net = xp_net(pol);
        unsigned int h;
        struct hlist_node *entry;
-       struct xfrm_state *x, *x0;
+       struct xfrm_state *x, *x0, *to_put;
        int acquire_in_progress = 0;
        int error = 0;
        struct xfrm_state *best = NULL;
 
+       to_put = NULL;
+
        spin_lock_bh(&xfrm_state_lock);
-       h = xfrm_dst_hash(daddr, saddr, tmpl->reqid, family);
-       hlist_for_each_entry(x, entry, xfrm_state_bydst+h, bydst) {
+       h = xfrm_dst_hash(net, daddr, saddr, tmpl->reqid, family);
+       hlist_for_each_entry(x, entry, net->xfrm.state_bydst+h, bydst) {
                if (x->props.family == family &&
                    x->props.reqid == tmpl->reqid &&
                    !(x->props.flags & XFRM_STATE_WILDRECV) &&
                    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) {
-                               if (!xfrm_selector_match(&x->sel, fl, x->sel.family) ||
-                                   !security_xfrm_state_pol_flow_match(x, pol, fl))
-                                       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) {
-                               if (xfrm_selector_match(&x->sel, fl, x->sel.family) &&
-                                   security_xfrm_state_pol_flow_match(x, pol, fl))
-                                       error = -ESRCH;
-                       }
-               }
+                   (tmpl->id.spi == x->id.spi || !tmpl->id.spi))
+                       xfrm_state_look_at(pol, x, fl, family, daddr, saddr,
+                                          &best, &acquire_in_progress, &error);
        }
+       if (best)
+               goto found;
 
+       h = xfrm_dst_hash(net, daddr, &saddr_wildcard, tmpl->reqid, family);
+       hlist_for_each_entry(x, entry, net->xfrm.state_bydst+h, bydst) {
+               if (x->props.family == family &&
+                   x->props.reqid == tmpl->reqid &&
+                   !(x->props.flags & XFRM_STATE_WILDRECV) &&
+                   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))
+                       xfrm_state_look_at(pol, x, fl, family, daddr, saddr,
+                                          &best, &acquire_in_progress, &error);
+       }
+
+found:
        x = best;
        if (!x && !error && !acquire_in_progress) {
                if (tmpl->id.spi &&
-                   (x0 = __xfrm_state_lookup(daddr, tmpl->id.spi,
+                   (x0 = __xfrm_state_lookup(net, daddr, tmpl->id.spi,
                                              tmpl->id.proto, family)) != NULL) {
-                       xfrm_state_put(x0);
+                       to_put = x0;
                        error = -EEXIST;
                        goto out;
                }
-               x = xfrm_state_alloc();
+               x = xfrm_state_alloc(net);
                if (x == NULL) {
                        error = -ENOMEM;
                        goto out;
@@ -827,28 +854,29 @@ xfrm_state_find(xfrm_address_t *daddr, xfrm_address_t *saddr,
                error = security_xfrm_state_alloc_acquire(x, pol->security, fl->secid);
                if (error) {
                        x->km.state = XFRM_STATE_DEAD;
-                       xfrm_state_put(x);
+                       to_put = x;
                        x = NULL;
                        goto out;
                }
 
                if (km_query(x, tmpl, pol) == 0) {
                        x->km.state = XFRM_STATE_ACQ;
-                       hlist_add_head(&x->bydst, xfrm_state_bydst+h);
-                       h = xfrm_src_hash(daddr, saddr, family);
-                       hlist_add_head(&x->bysrc, xfrm_state_bysrc+h);
+                       list_add(&x->km.all, &net->xfrm.state_all);
+                       hlist_add_head(&x->bydst, net->xfrm.state_bydst+h);
+                       h = xfrm_src_hash(net, daddr, saddr, family);
+                       hlist_add_head(&x->bysrc, net->xfrm.state_bysrc+h);
                        if (x->id.spi) {
-                               h = xfrm_spi_hash(&x->id.daddr, x->id.spi, x->id.proto, family);
-                               hlist_add_head(&x->byspi, xfrm_state_byspi+h);
+                               h = xfrm_spi_hash(net, &x->id.daddr, x->id.spi, x->id.proto, family);
+                               hlist_add_head(&x->byspi, net->xfrm.state_byspi+h);
                        }
-                       x->lft.hard_add_expires_seconds = sysctl_xfrm_acq_expires;
-                       x->timer.expires = jiffies + sysctl_xfrm_acq_expires*HZ;
+                       x->lft.hard_add_expires_seconds = net->xfrm.sysctl_acq_expires;
+                       x->timer.expires = jiffies + net->xfrm.sysctl_acq_expires*HZ;
                        add_timer(&x->timer);
-                       xfrm_state_num++;
-                       xfrm_hash_grow_check(x->bydst.next != NULL);
+                       net->xfrm.state_num++;
+                       xfrm_hash_grow_check(net, x->bydst.next != NULL);
                } else {
                        x->km.state = XFRM_STATE_DEAD;
-                       xfrm_state_put(x);
+                       to_put = x;
                        x = NULL;
                        error = -ESRCH;
                }
@@ -859,11 +887,14 @@ out:
        else
                *err = acquire_in_progress ? -EAGAIN : error;
        spin_unlock_bh(&xfrm_state_lock);
+       if (to_put)
+               xfrm_state_put(to_put);
        return x;
 }
 
 struct xfrm_state *
-xfrm_stateonly_find(xfrm_address_t *daddr, xfrm_address_t *saddr,
+xfrm_stateonly_find(struct net *net,
+                   xfrm_address_t *daddr, xfrm_address_t *saddr,
                    unsigned short family, u8 mode, u8 proto, u32 reqid)
 {
        unsigned int h;
@@ -871,8 +902,8 @@ xfrm_stateonly_find(xfrm_address_t *daddr, xfrm_address_t *saddr,
        struct hlist_node *entry;
 
        spin_lock(&xfrm_state_lock);
-       h = xfrm_dst_hash(daddr, saddr, reqid, family);
-       hlist_for_each_entry(x, entry, xfrm_state_bydst+h, bydst) {
+       h = xfrm_dst_hash(net, daddr, saddr, reqid, family);
+       hlist_for_each_entry(x, entry, net->xfrm.state_bydst+h, bydst) {
                if (x->props.family == family &&
                    x->props.reqid == reqid &&
                    !(x->props.flags & XFRM_STATE_WILDRECV) &&
@@ -896,46 +927,50 @@ EXPORT_SYMBOL(xfrm_stateonly_find);
 
 static void __xfrm_state_insert(struct xfrm_state *x)
 {
+       struct net *net = xs_net(x);
        unsigned int h;
 
        x->genid = ++xfrm_state_genid;
 
-       h = xfrm_dst_hash(&x->id.daddr, &x->props.saddr,
+       list_add(&x->km.all, &net->xfrm.state_all);
+
+       h = xfrm_dst_hash(net, &x->id.daddr, &x->props.saddr,
                          x->props.reqid, x->props.family);
-       hlist_add_head(&x->bydst, xfrm_state_bydst+h);
+       hlist_add_head(&x->bydst, net->xfrm.state_bydst+h);
 
-       h = xfrm_src_hash(&x->id.daddr, &x->props.saddr, x->props.family);
-       hlist_add_head(&x->bysrc, xfrm_state_bysrc+h);
+       h = xfrm_src_hash(net, &x->id.daddr, &x->props.saddr, x->props.family);
+       hlist_add_head(&x->bysrc, net->xfrm.state_bysrc+h);
 
        if (x->id.spi) {
-               h = xfrm_spi_hash(&x->id.daddr, x->id.spi, x->id.proto,
+               h = xfrm_spi_hash(net, &x->id.daddr, x->id.spi, x->id.proto,
                                  x->props.family);
 
-               hlist_add_head(&x->byspi, xfrm_state_byspi+h);
+               hlist_add_head(&x->byspi, net->xfrm.state_byspi+h);
        }
 
        mod_timer(&x->timer, jiffies + HZ);
        if (x->replay_maxage)
                mod_timer(&x->rtimer, jiffies + x->replay_maxage);
 
-       wake_up(&km_waitq);
+       wake_up(&net->xfrm.km_waitq);
 
-       xfrm_state_num++;
+       net->xfrm.state_num++;
 
-       xfrm_hash_grow_check(x->bydst.next != NULL);
+       xfrm_hash_grow_check(net, x->bydst.next != NULL);
 }
 
 /* xfrm_state_lock is held */
 static void __xfrm_state_bump_genids(struct xfrm_state *xnew)
 {
+       struct net *net = xs_net(xnew);
        unsigned short family = xnew->props.family;
        u32 reqid = xnew->props.reqid;
        struct xfrm_state *x;
        struct hlist_node *entry;
        unsigned int h;
 
-       h = xfrm_dst_hash(&xnew->id.daddr, &xnew->props.saddr, reqid, family);
-       hlist_for_each_entry(x, entry, xfrm_state_bydst+h, bydst) {
+       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) {
                if (x->props.family     == family &&
                    x->props.reqid      == reqid &&
                    !xfrm_addr_cmp(&x->id.daddr, &xnew->id.daddr, family) &&
@@ -954,13 +989,13 @@ void xfrm_state_insert(struct xfrm_state *x)
 EXPORT_SYMBOL(xfrm_state_insert);
 
 /* 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)
+static struct xfrm_state *__find_acq_core(struct net *net, unsigned short family, u8 mode, u32 reqid, u8 proto, xfrm_address_t *daddr, xfrm_address_t *saddr, int create)
 {
-       unsigned int h = xfrm_dst_hash(daddr, saddr, reqid, family);
+       unsigned int h = xfrm_dst_hash(net, daddr, saddr, reqid, family);
        struct hlist_node *entry;
        struct xfrm_state *x;
 
-       hlist_for_each_entry(x, entry, xfrm_state_bydst+h, bydst) {
+       hlist_for_each_entry(x, entry, net->xfrm.state_bydst+h, bydst) {
                if (x->props.reqid  != reqid ||
                    x->props.mode   != mode ||
                    x->props.family != family ||
@@ -992,7 +1027,7 @@ static struct xfrm_state *__find_acq_core(unsigned short family, u8 mode, u32 re
        if (!create)
                return NULL;
 
-       x = xfrm_state_alloc();
+       x = xfrm_state_alloc(net);
        if (likely(x)) {
                switch (family) {
                case AF_INET:
@@ -1023,54 +1058,58 @@ static struct xfrm_state *__find_acq_core(unsigned short family, u8 mode, u32 re
                x->props.family = family;
                x->props.mode = mode;
                x->props.reqid = reqid;
-               x->lft.hard_add_expires_seconds = sysctl_xfrm_acq_expires;
+               x->lft.hard_add_expires_seconds = net->xfrm.sysctl_acq_expires;
                xfrm_state_hold(x);
-               x->timer.expires = jiffies + sysctl_xfrm_acq_expires*HZ;
+               x->timer.expires = jiffies + net->xfrm.sysctl_acq_expires*HZ;
                add_timer(&x->timer);
-               hlist_add_head(&x->bydst, xfrm_state_bydst+h);
-               h = xfrm_src_hash(daddr, saddr, family);
-               hlist_add_head(&x->bysrc, xfrm_state_bysrc+h);
+               list_add(&x->km.all, &net->xfrm.state_all);
+               hlist_add_head(&x->bydst, net->xfrm.state_bydst+h);
+               h = xfrm_src_hash(net, daddr, saddr, family);
+               hlist_add_head(&x->bysrc, net->xfrm.state_bysrc+h);
 
-               xfrm_state_num++;
+               net->xfrm.state_num++;
 
-               xfrm_hash_grow_check(x->bydst.next != NULL);
+               xfrm_hash_grow_check(net, x->bydst.next != NULL);
        }
 
        return x;
 }
 
-static struct xfrm_state *__xfrm_find_acq_byseq(u32 seq);
+static struct xfrm_state *__xfrm_find_acq_byseq(struct net *net, u32 seq);
 
 int xfrm_state_add(struct xfrm_state *x)
 {
-       struct xfrm_state *x1;
+       struct net *net = xs_net(x);
+       struct xfrm_state *x1, *to_put;
        int family;
        int err;
        int use_spi = xfrm_id_proto_match(x->id.proto, IPSEC_PROTO_ANY);
 
        family = x->props.family;
 
+       to_put = NULL;
+
        spin_lock_bh(&xfrm_state_lock);
 
        x1 = __xfrm_state_locate(x, use_spi, family);
        if (x1) {
-               xfrm_state_put(x1);
+               to_put = x1;
                x1 = NULL;
                err = -EEXIST;
                goto out;
        }
 
        if (use_spi && x->km.seq) {
-               x1 = __xfrm_find_acq_byseq(x->km.seq);
+               x1 = __xfrm_find_acq_byseq(net, x->km.seq);
                if (x1 && ((x1->id.proto != x->id.proto) ||
                    xfrm_addr_cmp(&x1->id.daddr, &x->id.daddr, family))) {
-                       xfrm_state_put(x1);
+                       to_put = x1;
                        x1 = NULL;
                }
        }
 
        if (use_spi && !x1)
-               x1 = __find_acq_core(family, x->props.mode, x->props.reqid,
+               x1 = __find_acq_core(net, family, x->props.mode, x->props.reqid,
                                     x->id.proto,
                                     &x->id.daddr, &x->props.saddr, 0);
 
@@ -1086,15 +1125,19 @@ out:
                xfrm_state_put(x1);
        }
 
+       if (to_put)
+               xfrm_state_put(to_put);
+
        return err;
 }
 EXPORT_SYMBOL(xfrm_state_add);
 
 #ifdef CONFIG_XFRM_MIGRATE
-struct xfrm_state *xfrm_state_clone(struct xfrm_state *orig, int *errp)
+static struct xfrm_state *xfrm_state_clone(struct xfrm_state *orig, int *errp)
 {
+       struct net *net = xs_net(orig);
        int err = -ENOMEM;
-       struct xfrm_state *x = xfrm_state_alloc();
+       struct xfrm_state *x = xfrm_state_alloc(net);
        if (!x)
                goto error;
 
@@ -1166,7 +1209,6 @@ struct xfrm_state *xfrm_state_clone(struct xfrm_state *orig, int *errp)
        kfree(x);
        return NULL;
 }
-EXPORT_SYMBOL(xfrm_state_clone);
 
 /* xfrm_state_lock is held */
 struct xfrm_state * xfrm_migrate_state_find(struct xfrm_migrate *m)
@@ -1176,9 +1218,9 @@ struct xfrm_state * xfrm_migrate_state_find(struct xfrm_migrate *m)
        struct hlist_node *entry;
 
        if (m->reqid) {
-               h = xfrm_dst_hash(&m->old_daddr, &m->old_saddr,
+               h = xfrm_dst_hash(&init_net, &m->old_daddr, &m->old_saddr,
                                  m->reqid, m->old_family);
-               hlist_for_each_entry(x, entry, xfrm_state_bydst+h, bydst) {
+               hlist_for_each_entry(x, entry, init_net.xfrm.state_bydst+h, bydst) {
                        if (x->props.mode != m->mode ||
                            x->id.proto != m->proto)
                                continue;
@@ -1193,9 +1235,9 @@ struct xfrm_state * xfrm_migrate_state_find(struct xfrm_migrate *m)
                        return x;
                }
        } else {
-               h = xfrm_src_hash(&m->old_daddr, &m->old_saddr,
+               h = xfrm_src_hash(&init_net, &m->old_daddr, &m->old_saddr,
                                  m->old_family);
-               hlist_for_each_entry(x, entry, xfrm_state_bysrc+h, bysrc) {
+               hlist_for_each_entry(x, entry, init_net.xfrm.state_bysrc+h, bysrc) {
                        if (x->props.mode != m->mode ||
                            x->id.proto != m->proto)
                                continue;
@@ -1246,10 +1288,12 @@ EXPORT_SYMBOL(xfrm_state_migrate);
 
 int xfrm_state_update(struct xfrm_state *x)
 {
-       struct xfrm_state *x1;
+       struct xfrm_state *x1, *to_put;
        int err;
        int use_spi = xfrm_id_proto_match(x->id.proto, IPSEC_PROTO_ANY);
 
+       to_put = NULL;
+
        spin_lock_bh(&xfrm_state_lock);
        x1 = __xfrm_state_locate(x, use_spi, x->props.family);
 
@@ -1258,7 +1302,7 @@ int xfrm_state_update(struct xfrm_state *x)
                goto out;
 
        if (xfrm_state_kern(x1)) {
-               xfrm_state_put(x1);
+               to_put = x1;
                err = -EEXIST;
                goto out;
        }
@@ -1272,6 +1316,9 @@ int xfrm_state_update(struct xfrm_state *x)
 out:
        spin_unlock_bh(&xfrm_state_lock);
 
+       if (to_put)
+               xfrm_state_put(to_put);
+
        if (err)
                return err;
 
@@ -1334,40 +1381,41 @@ int xfrm_state_check_expire(struct xfrm_state *x)
 EXPORT_SYMBOL(xfrm_state_check_expire);
 
 struct xfrm_state *
-xfrm_state_lookup(xfrm_address_t *daddr, __be32 spi, u8 proto,
+xfrm_state_lookup(struct net *net, xfrm_address_t *daddr, __be32 spi, u8 proto,
                  unsigned short family)
 {
        struct xfrm_state *x;
 
        spin_lock_bh(&xfrm_state_lock);
-       x = __xfrm_state_lookup(daddr, spi, proto, family);
+       x = __xfrm_state_lookup(net, daddr, spi, proto, family);
        spin_unlock_bh(&xfrm_state_lock);
        return x;
 }
 EXPORT_SYMBOL(xfrm_state_lookup);
 
 struct xfrm_state *
-xfrm_state_lookup_byaddr(xfrm_address_t *daddr, xfrm_address_t *saddr,
+xfrm_state_lookup_byaddr(struct net *net,
+                        xfrm_address_t *daddr, xfrm_address_t *saddr,
                         u8 proto, unsigned short family)
 {
        struct xfrm_state *x;
 
        spin_lock_bh(&xfrm_state_lock);
-       x = __xfrm_state_lookup_byaddr(daddr, saddr, proto, family);
+       x = __xfrm_state_lookup_byaddr(net, daddr, saddr, proto, family);
        spin_unlock_bh(&xfrm_state_lock);
        return x;
 }
 EXPORT_SYMBOL(xfrm_state_lookup_byaddr);
 
 struct xfrm_state *
-xfrm_find_acq(u8 mode, u32 reqid, u8 proto,
+xfrm_find_acq(struct net *net, u8 mode, u32 reqid, u8 proto,
              xfrm_address_t *daddr, xfrm_address_t *saddr,
              int create, unsigned short family)
 {
        struct xfrm_state *x;
 
        spin_lock_bh(&xfrm_state_lock);
-       x = __find_acq_core(family, mode, reqid, proto, daddr, saddr, create);
+       x = __find_acq_core(net, family, mode, reqid, proto, daddr, saddr, create);
        spin_unlock_bh(&xfrm_state_lock);
 
        return x;
@@ -1414,15 +1462,15 @@ EXPORT_SYMBOL(xfrm_state_sort);
 
 /* Silly enough, but I'm lazy to build resolution list */
 
-static struct xfrm_state *__xfrm_find_acq_byseq(u32 seq)
+static struct xfrm_state *__xfrm_find_acq_byseq(struct net *net, u32 seq)
 {
        int i;
 
-       for (i = 0; i <= xfrm_state_hmask; i++) {
+       for (i = 0; i <= net->xfrm.state_hmask; i++) {
                struct hlist_node *entry;
                struct xfrm_state *x;
 
-               hlist_for_each_entry(x, entry, xfrm_state_bydst+i, bydst) {
+               hlist_for_each_entry(x, entry, net->xfrm.state_bydst+i, bydst) {
                        if (x->km.seq == seq &&
                            x->km.state == XFRM_STATE_ACQ) {
                                xfrm_state_hold(x);
@@ -1433,12 +1481,12 @@ static struct xfrm_state *__xfrm_find_acq_byseq(u32 seq)
        return NULL;
 }
 
-struct xfrm_state *xfrm_find_acq_byseq(u32 seq)
+struct xfrm_state *xfrm_find_acq_byseq(struct net *net, u32 seq)
 {
        struct xfrm_state *x;
 
        spin_lock_bh(&xfrm_state_lock);
-       x = __xfrm_find_acq_byseq(seq);
+       x = __xfrm_find_acq_byseq(net, seq);
        spin_unlock_bh(&xfrm_state_lock);
        return x;
 }
@@ -1459,6 +1507,7 @@ EXPORT_SYMBOL(xfrm_get_acqseq);
 
 int xfrm_alloc_spi(struct xfrm_state *x, u32 low, u32 high)
 {
+       struct net *net = xs_net(x);
        unsigned int h;
        struct xfrm_state *x0;
        int err = -ENOENT;
@@ -1476,7 +1525,7 @@ int xfrm_alloc_spi(struct xfrm_state *x, u32 low, u32 high)
        err = -ENOENT;
 
        if (minspi == maxspi) {
-               x0 = xfrm_state_lookup(&x->id.daddr, minspi, x->id.proto, x->props.family);
+               x0 = xfrm_state_lookup(net, &x->id.daddr, minspi, x->id.proto, x->props.family);
                if (x0) {
                        xfrm_state_put(x0);
                        goto unlock;
@@ -1486,7 +1535,7 @@ int xfrm_alloc_spi(struct xfrm_state *x, u32 low, u32 high)
                u32 spi = 0;
                for (h=0; h<high-low+1; h++) {
                        spi = low + net_random()%(high-low+1);
-                       x0 = xfrm_state_lookup(&x->id.daddr, htonl(spi), x->id.proto, x->props.family);
+                       x0 = xfrm_state_lookup(net, &x->id.daddr, htonl(spi), x->id.proto, x->props.family);
                        if (x0 == NULL) {
                                x->id.spi = htonl(spi);
                                break;
@@ -1496,8 +1545,8 @@ int xfrm_alloc_spi(struct xfrm_state *x, u32 low, u32 high)
        }
        if (x->id.spi) {
                spin_lock_bh(&xfrm_state_lock);
-               h = xfrm_spi_hash(&x->id.daddr, x->id.spi, x->id.proto, x->props.family);
-               hlist_add_head(&x->byspi, xfrm_state_byspi+h);
+               h = xfrm_spi_hash(net, &x->id.daddr, x->id.spi, x->id.proto, x->props.family);
+               hlist_add_head(&x->byspi, net->xfrm.state_byspi+h);
                spin_unlock_bh(&xfrm_state_lock);
 
                err = 0;
@@ -1510,40 +1559,66 @@ unlock:
 }
 EXPORT_SYMBOL(xfrm_alloc_spi);
 
-int xfrm_state_walk(u8 proto, int (*func)(struct xfrm_state *, int, void*),
+int xfrm_state_walk(struct net *net, struct xfrm_state_walk *walk,
+                   int (*func)(struct xfrm_state *, int, void*),
                    void *data)
 {
-       int i;
-       struct xfrm_state *x, *last = NULL;
-       struct hlist_node *entry;
-       int count = 0;
+       struct xfrm_state *state;
+       struct xfrm_state_walk *x;
        int err = 0;
 
+       if (walk->seq != 0 && list_empty(&walk->all))
+               return 0;
+
        spin_lock_bh(&xfrm_state_lock);
-       for (i = 0; i <= xfrm_state_hmask; i++) {
-               hlist_for_each_entry(x, entry, xfrm_state_bydst+i, bydst) {
-                       if (!xfrm_id_proto_match(x->id.proto, proto))
-                               continue;
-                       if (last) {
-                               err = func(last, count, data);
-                               if (err)
-                                       goto out;
-                       }
-                       last = x;
-                       count++;
+       if (list_empty(&walk->all))
+               x = list_first_entry(&net->xfrm.state_all, struct xfrm_state_walk, all);
+       else
+               x = list_entry(&walk->all, struct xfrm_state_walk, all);
+       list_for_each_entry_from(x, &net->xfrm.state_all, all) {
+               if (x->state == XFRM_STATE_DEAD)
+                       continue;
+               state = container_of(x, struct xfrm_state, km);
+               if (!xfrm_id_proto_match(state->id.proto, walk->proto))
+                       continue;
+               err = func(state, walk->seq, data);
+               if (err) {
+                       list_move_tail(&walk->all, &x->all);
+                       goto out;
                }
+               walk->seq++;
        }
-       if (count == 0) {
+       if (walk->seq == 0) {
                err = -ENOENT;
                goto out;
        }
-       err = func(last, 0, data);
+       list_del_init(&walk->all);
 out:
        spin_unlock_bh(&xfrm_state_lock);
        return err;
 }
 EXPORT_SYMBOL(xfrm_state_walk);
 
+void xfrm_state_walk_init(struct xfrm_state_walk *walk, u8 proto)
+{
+       INIT_LIST_HEAD(&walk->all);
+       walk->proto = proto;
+       walk->state = XFRM_STATE_DEAD;
+       walk->seq = 0;
+}
+EXPORT_SYMBOL(xfrm_state_walk_init);
+
+void xfrm_state_walk_done(struct xfrm_state_walk *walk)
+{
+       if (list_empty(&walk->all))
+               return;
+
+       spin_lock_bh(&xfrm_state_lock);
+       list_del(&walk->all);
+       spin_lock_bh(&xfrm_state_lock);
+}
+EXPORT_SYMBOL(xfrm_state_walk_done);
+
 
 void xfrm_replay_notify(struct xfrm_state *x, int event)
 {
@@ -1599,7 +1674,7 @@ static void xfrm_replay_timer_handler(unsigned long data)
        spin_lock(&x->lock);
 
        if (x->km.state == XFRM_STATE_VALID) {
-               if (xfrm_aevent_is_on())
+               if (xfrm_aevent_is_on(xs_net(x)))
                        xfrm_replay_notify(x, XFRM_REPLAY_TIMEOUT);
                else
                        x->xflags |= XFRM_TIME_DEFER;
@@ -1608,13 +1683,14 @@ static void xfrm_replay_timer_handler(unsigned long data)
        spin_unlock(&x->lock);
 }
 
-int xfrm_replay_check(struct xfrm_state *x, __be32 net_seq)
+int xfrm_replay_check(struct xfrm_state *x,
+                     struct sk_buff *skb, __be32 net_seq)
 {
        u32 diff;
        u32 seq = ntohl(net_seq);
 
        if (unlikely(seq == 0))
-               return -EINVAL;
+               goto err;
 
        if (likely(seq > x->replay.seq))
                return 0;
@@ -1623,16 +1699,19 @@ int xfrm_replay_check(struct xfrm_state *x, __be32 net_seq)
        if (diff >= min_t(unsigned int, x->props.replay_window,
                          sizeof(x->replay.bitmap) * 8)) {
                x->stats.replay_window++;
-               return -EINVAL;
+               goto err;
        }
 
        if (x->replay.bitmap & (1U << diff)) {
                x->stats.replay++;
-               return -EINVAL;
+               goto err;
        }
        return 0;
+
+err:
+       xfrm_audit_state_replay(x, skb, net_seq);
+       return -EINVAL;
 }
-EXPORT_SYMBOL(xfrm_replay_check);
 
 void xfrm_replay_advance(struct xfrm_state *x, __be32 net_seq)
 {
@@ -1651,10 +1730,9 @@ void xfrm_replay_advance(struct xfrm_state *x, __be32 net_seq)
                x->replay.bitmap |= (1U << diff);
        }
 
-       if (xfrm_aevent_is_on())
+       if (xfrm_aevent_is_on(xs_net(x)))
                xfrm_replay_notify(x, XFRM_REPLAY_UPDATE);
 }
-EXPORT_SYMBOL(xfrm_replay_advance);
 
 static LIST_HEAD(xfrm_km_list);
 static DEFINE_RWLOCK(xfrm_km_lock);
@@ -1685,6 +1763,7 @@ EXPORT_SYMBOL(km_state_notify);
 
 void km_state_expired(struct xfrm_state *x, int hard, u32 pid)
 {
+       struct net *net = xs_net(x);
        struct km_event c;
 
        c.data.hard = hard;
@@ -1693,7 +1772,7 @@ void km_state_expired(struct xfrm_state *x, int hard, u32 pid)
        km_state_notify(x, &c);
 
        if (hard)
-               wake_up(&km_waitq);
+               wake_up(&net->xfrm.km_waitq);
 }
 
 EXPORT_SYMBOL(km_state_expired);
@@ -1736,6 +1815,7 @@ EXPORT_SYMBOL(km_new_mapping);
 
 void km_policy_expired(struct xfrm_policy *pol, int dir, int hard, u32 pid)
 {
+       struct net *net = xp_net(pol);
        struct km_event c;
 
        c.data.hard = hard;
@@ -1744,13 +1824,14 @@ void km_policy_expired(struct xfrm_policy *pol, int dir, int hard, u32 pid)
        km_policy_notify(pol, dir, &c);
 
        if (hard)
-               wake_up(&km_waitq);
+               wake_up(&net->xfrm.km_waitq);
 }
 EXPORT_SYMBOL(km_policy_expired);
 
 #ifdef CONFIG_XFRM_MIGRATE
 int km_migrate(struct xfrm_selector *sel, u8 dir, u8 type,
-              struct xfrm_migrate *m, int num_migrate)
+              struct xfrm_migrate *m, int num_migrate,
+              struct xfrm_kmaddress *k)
 {
        int err = -EINVAL;
        int ret;
@@ -1759,7 +1840,7 @@ int km_migrate(struct xfrm_selector *sel, u8 dir, u8 type,
        read_lock(&xfrm_km_lock);
        list_for_each_entry(km, &xfrm_km_list, list) {
                if (km->migrate) {
-                       ret = km->migrate(sel, dir, type, m, num_migrate);
+                       ret = km->migrate(sel, dir, type, m, num_migrate, k);
                        if (!ret)
                                err = ret;
                }
@@ -1770,7 +1851,7 @@ int km_migrate(struct xfrm_selector *sel, u8 dir, u8 type,
 EXPORT_SYMBOL(km_migrate);
 #endif
 
-int km_report(u8 proto, struct xfrm_selector *sel, xfrm_address_t *addr)
+int km_report(struct net *net, u8 proto, struct xfrm_selector *sel, xfrm_address_t *addr)
 {
        int err = -EINVAL;
        int ret;
@@ -1779,7 +1860,7 @@ int km_report(u8 proto, struct xfrm_selector *sel, xfrm_address_t *addr)
        read_lock(&xfrm_km_lock);
        list_for_each_entry(km, &xfrm_km_list, list) {
                if (km->report) {
-                       ret = km->report(proto, sel, addr);
+                       ret = km->report(net, proto, sel, addr);
                        if (!ret)
                                err = ret;
                }
@@ -1896,6 +1977,7 @@ static struct xfrm_state_afinfo *xfrm_state_get_afinfo(unsigned int family)
 }
 
 static void xfrm_state_put_afinfo(struct xfrm_state_afinfo *afinfo)
+       __releases(xfrm_state_afinfo_lock)
 {
        read_unlock(&xfrm_state_afinfo_lock);
 }
@@ -1932,6 +2014,7 @@ int xfrm_state_mtu(struct xfrm_state *x, int mtu)
 int xfrm_init_state(struct xfrm_state *x)
 {
        struct xfrm_state_afinfo *afinfo;
+       struct xfrm_mode *inner_mode;
        int family = x->props.family;
        int err;
 
@@ -1950,13 +2033,44 @@ int xfrm_init_state(struct xfrm_state *x)
                goto error;
 
        err = -EPROTONOSUPPORT;
-       x->inner_mode = xfrm_get_mode(x->props.mode, x->sel.family);
-       if (x->inner_mode == NULL)
-               goto error;
 
-       if (!(x->inner_mode->flags & XFRM_MODE_FLAG_TUNNEL) &&
-           family != x->sel.family)
-               goto error;
+       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;
+               int iafamily = AF_INET;
+
+               inner_mode = xfrm_get_mode(x->props.mode, x->props.family);
+               if (inner_mode == NULL)
+                       goto error;
+
+               if (!(inner_mode->flags & XFRM_MODE_FLAG_TUNNEL)) {
+                       xfrm_put_mode(inner_mode);
+                       goto error;
+               }
+               x->inner_mode = inner_mode;
+
+               if (x->props.family == AF_INET)
+                       iafamily = AF_INET6;
+
+               inner_mode_iaf = xfrm_get_mode(x->props.mode, iafamily);
+               if (inner_mode_iaf) {
+                       if (inner_mode_iaf->flags & XFRM_MODE_FLAG_TUNNEL)
+                               x->inner_mode_iaf = inner_mode_iaf;
+                       else
+                               xfrm_put_mode(inner_mode_iaf);
+               }
+       }
 
        x->type = xfrm_get_type(x->id.proto, family);
        if (x->type == NULL)
@@ -1978,90 +2092,222 @@ error:
 
 EXPORT_SYMBOL(xfrm_init_state);
 
-void __init xfrm_state_init(void)
+int __net_init xfrm_state_init(struct net *net)
 {
        unsigned int sz;
 
+       INIT_LIST_HEAD(&net->xfrm.state_all);
+
        sz = sizeof(struct hlist_head) * 8;
 
-       xfrm_state_bydst = xfrm_hash_alloc(sz);
-       xfrm_state_bysrc = xfrm_hash_alloc(sz);
-       xfrm_state_byspi = xfrm_hash_alloc(sz);
-       if (!xfrm_state_bydst || !xfrm_state_bysrc || !xfrm_state_byspi)
-               panic("XFRM: Cannot allocate bydst/bysrc/byspi hashes.");
-       xfrm_state_hmask = ((sz / sizeof(struct hlist_head)) - 1);
+       net->xfrm.state_bydst = xfrm_hash_alloc(sz);
+       if (!net->xfrm.state_bydst)
+               goto out_bydst;
+       net->xfrm.state_bysrc = xfrm_hash_alloc(sz);
+       if (!net->xfrm.state_bysrc)
+               goto out_bysrc;
+       net->xfrm.state_byspi = xfrm_hash_alloc(sz);
+       if (!net->xfrm.state_byspi)
+               goto out_byspi;
+       net->xfrm.state_hmask = ((sz / sizeof(struct hlist_head)) - 1);
+
+       net->xfrm.state_num = 0;
+       INIT_WORK(&net->xfrm.state_hash_work, xfrm_hash_resize);
+       INIT_HLIST_HEAD(&net->xfrm.state_gc_list);
+       INIT_WORK(&net->xfrm.state_gc_work, xfrm_state_gc_task);
+       init_waitqueue_head(&net->xfrm.km_waitq);
+       return 0;
+
+out_byspi:
+       xfrm_hash_free(net->xfrm.state_bysrc, sz);
+out_bysrc:
+       xfrm_hash_free(net->xfrm.state_bydst, sz);
+out_bydst:
+       return -ENOMEM;
+}
+
+void xfrm_state_fini(struct net *net)
+{
+       struct xfrm_audit audit_info;
+       unsigned int sz;
+
+       flush_work(&net->xfrm.state_hash_work);
+       audit_info.loginuid = -1;
+       audit_info.sessionid = -1;
+       audit_info.secid = 0;
+       xfrm_state_flush(net, IPSEC_PROTO_ANY, &audit_info);
+       flush_work(&net->xfrm.state_gc_work);
 
-       INIT_WORK(&xfrm_state_gc_work, xfrm_state_gc_task);
+       WARN_ON(!list_empty(&net->xfrm.state_all));
+
+       sz = (net->xfrm.state_hmask + 1) * sizeof(struct hlist_head);
+       WARN_ON(!hlist_empty(net->xfrm.state_byspi));
+       xfrm_hash_free(net->xfrm.state_byspi, sz);
+       WARN_ON(!hlist_empty(net->xfrm.state_bysrc));
+       xfrm_hash_free(net->xfrm.state_bysrc, sz);
+       WARN_ON(!hlist_empty(net->xfrm.state_bydst));
+       xfrm_hash_free(net->xfrm.state_bydst, sz);
 }
 
 #ifdef CONFIG_AUDITSYSCALL
-static inline void xfrm_audit_common_stateinfo(struct xfrm_state *x,
-                                              struct audit_buffer *audit_buf)
+static void xfrm_audit_helper_sainfo(struct xfrm_state *x,
+                                    struct audit_buffer *audit_buf)
 {
-       if (x->security)
+       struct xfrm_sec_ctx *ctx = x->security;
+       u32 spi = ntohl(x->id.spi);
+
+       if (ctx)
                audit_log_format(audit_buf, " sec_alg=%u sec_doi=%u sec_obj=%s",
-                                x->security->ctx_alg, x->security->ctx_doi,
-                                x->security->ctx_str);
+                                ctx->ctx_alg, ctx->ctx_doi, ctx->ctx_str);
 
        switch(x->props.family) {
        case AF_INET:
-               audit_log_format(audit_buf, " src=%u.%u.%u.%u dst=%u.%u.%u.%u",
-                                NIPQUAD(x->props.saddr.a4),
-                                NIPQUAD(x->id.daddr.a4));
+               audit_log_format(audit_buf, " src=%pI4 dst=%pI4",
+                                &x->props.saddr.a4, &x->id.daddr.a4);
                break;
        case AF_INET6:
-               {
-                       struct in6_addr saddr6, daddr6;
-
-                       memcpy(&saddr6, x->props.saddr.a6,
-                               sizeof(struct in6_addr));
-                       memcpy(&daddr6, x->id.daddr.a6,
-                               sizeof(struct in6_addr));
-                       audit_log_format(audit_buf,
-                                        " src=" NIP6_FMT " dst=" NIP6_FMT,
-                                        NIP6(saddr6), NIP6(daddr6));
-               }
+               audit_log_format(audit_buf, " src=%pI6 dst=%pI6",
+                                x->props.saddr.a6, x->id.daddr.a6);
                break;
        }
+
+       audit_log_format(audit_buf, " spi=%u(0x%x)", spi, spi);
 }
 
-void
-xfrm_audit_state_add(struct xfrm_state *x, int result, u32 auid, u32 sid)
+static void xfrm_audit_helper_pktinfo(struct sk_buff *skb, u16 family,
+                                     struct audit_buffer *audit_buf)
+{
+       struct iphdr *iph4;
+       struct ipv6hdr *iph6;
+
+       switch (family) {
+       case AF_INET:
+               iph4 = ip_hdr(skb);
+               audit_log_format(audit_buf, " src=%pI4 dst=%pI4",
+                                &iph4->saddr, &iph4->daddr);
+               break;
+       case AF_INET6:
+               iph6 = ipv6_hdr(skb);
+               audit_log_format(audit_buf,
+                                " src=%pI6 dst=%pI6 flowlbl=0x%x%02x%02x",
+                                &iph6->saddr,&iph6->daddr,
+                                iph6->flow_lbl[0] & 0x0f,
+                                iph6->flow_lbl[1],
+                                iph6->flow_lbl[2]);
+               break;
+       }
+}
+
+void xfrm_audit_state_add(struct xfrm_state *x, int result,
+                         uid_t auid, u32 sessionid, u32 secid)
+{
+       struct audit_buffer *audit_buf;
+
+       audit_buf = xfrm_audit_start("SAD-add");
+       if (audit_buf == NULL)
+               return;
+       xfrm_audit_helper_usrinfo(auid, sessionid, secid, audit_buf);
+       xfrm_audit_helper_sainfo(x, audit_buf);
+       audit_log_format(audit_buf, " res=%u", result);
+       audit_log_end(audit_buf);
+}
+EXPORT_SYMBOL_GPL(xfrm_audit_state_add);
+
+void xfrm_audit_state_delete(struct xfrm_state *x, int result,
+                            uid_t auid, u32 sessionid, u32 secid)
 {
        struct audit_buffer *audit_buf;
-       u32 spi;
-       extern int audit_enabled;
 
-       if (audit_enabled == 0)
+       audit_buf = xfrm_audit_start("SAD-delete");
+       if (audit_buf == NULL)
                return;
-       audit_buf = xfrm_audit_start(auid, sid);
+       xfrm_audit_helper_usrinfo(auid, sessionid, secid, audit_buf);
+       xfrm_audit_helper_sainfo(x, audit_buf);
+       audit_log_format(audit_buf, " res=%u", result);
+       audit_log_end(audit_buf);
+}
+EXPORT_SYMBOL_GPL(xfrm_audit_state_delete);
+
+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;
-       audit_log_format(audit_buf, " op=SAD-add res=%u",result);
-       xfrm_audit_common_stateinfo(x, audit_buf);
+       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_add);
+EXPORT_SYMBOL_GPL(xfrm_audit_state_replay_overflow);
 
-void
-xfrm_audit_state_delete(struct xfrm_state *x, int result, u32 auid, u32 sid)
+static void xfrm_audit_state_replay(struct xfrm_state *x,
+                            struct sk_buff *skb, __be32 net_seq)
 {
        struct audit_buffer *audit_buf;
        u32 spi;
-       extern int audit_enabled;
 
-       if (audit_enabled == 0)
-               return;
-       audit_buf = xfrm_audit_start(auid, sid);
+       audit_buf = xfrm_audit_start("SA-replayed-pkt");
        if (audit_buf == NULL)
                return;
-       audit_log_format(audit_buf, " op=SAD-delete res=%u",result);
-       xfrm_audit_common_stateinfo(x, audit_buf);
+       xfrm_audit_helper_pktinfo(skb, x->props.family, audit_buf);
        spi = ntohl(x->id.spi);
-       audit_log_format(audit_buf, " spi=%u(0x%x)", spi, 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_delete);
+
+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);
 #endif /* CONFIG_AUDITSYSCALL */