Whamcloud - gitweb
b=18751 Move prng.c to libcfs
[fs/lustre-release.git] / lustre / obdclass / lu_object.c
index f7efd24..8a86fa5 100644 (file)
@@ -26,7 +26,7 @@
  * GPL HEADER END
  */
 /*
- * Copyright  2008 Sun Microsystems, Inc. All rights reserved
+ * Copyright (c) 2007, 2010, Oracle and/or its affiliates. All rights reserved.
  * Use is subject to license terms.
  */
 /*
@@ -82,13 +82,14 @@ void lu_object_put(const struct lu_env *env, struct lu_object *o)
         site = o->lo_dev->ld_site;
         orig = o;
         kill_it = 0;
-        write_lock(&site->ls_guard);
-        if (atomic_dec_and_test(&top->loh_ref)) {
+        cfs_write_lock(&site->ls_guard);
+        if (cfs_atomic_dec_and_test(&top->loh_ref)) {
                 /*
                  * When last reference is released, iterate over object
                  * layers, and notify them that object is no longer busy.
                  */
-                list_for_each_entry_reverse(o, &top->loh_layers, lo_linkage) {
+                cfs_list_for_each_entry_reverse(o, &top->loh_layers,
+                                                lo_linkage) {
                         if (o->lo_ops->loo_object_release != NULL)
                                 o->lo_ops->loo_object_release(env, o);
                 }
@@ -107,13 +108,19 @@ void lu_object_put(const struct lu_env *env, struct lu_object *o)
                          * object lookup is possible and we can safely destroy
                          * object below.
                          */
-                        hlist_del_init(&top->loh_hash);
-                        list_del_init(&top->loh_lru);
+                        cfs_hlist_del_init(&top->loh_hash);
+                        cfs_list_del_init(&top->loh_lru);
                         -- site->ls_total;
                         kill_it = 1;
                 }
+        } else if (lu_object_is_dying(top)) {
+                /*
+                 * somebody may be waiting for this, currently only used
+                 * for cl_object, see cl_object_put_last().
+                 */
+                cfs_waitq_broadcast(&site->ls_marche_funebre);
         }
-        write_unlock(&site->ls_guard);
+        cfs_write_unlock(&site->ls_guard);
         if (kill_it)
                 /*
                  * Object was already removed from hash and lru above, can
@@ -123,19 +130,20 @@ void lu_object_put(const struct lu_env *env, struct lu_object *o)
 }
 EXPORT_SYMBOL(lu_object_put);
 
-/*
+/**
  * Allocate new object.
  *
  * This follows object creation protocol, described in the comment within
  * struct lu_device_operations definition.
  */
 static struct lu_object *lu_object_alloc(const struct lu_env *env,
-                                         struct lu_site *s,
-                                         const struct lu_fid *f)
+                                         struct lu_device *dev,
+                                         const struct lu_fid *f,
+                                         const struct lu_object_conf *conf)
 {
         struct lu_object *scan;
         struct lu_object *top;
-        struct list_head *layers;
+        cfs_list_t *layers;
         int clean;
         int result;
         ENTRY;
@@ -144,8 +152,7 @@ static struct lu_object *lu_object_alloc(const struct lu_env *env,
          * Create top-level object slice. This will also create
          * lu_object_header.
          */
-        top = s->ls_top_dev->ld_ops->ldo_object_alloc(env,
-                                                      NULL, s->ls_top_dev);
+        top = dev->ld_ops->ldo_object_alloc(env, NULL, dev);
         if (top == NULL)
                 RETURN(ERR_PTR(-ENOMEM));
         /*
@@ -161,12 +168,12 @@ static struct lu_object *lu_object_alloc(const struct lu_env *env,
                  * object slices are created.
                  */
                 clean = 1;
-                list_for_each_entry(scan, layers, lo_linkage) {
+                cfs_list_for_each_entry(scan, layers, lo_linkage) {
                         if (scan->lo_flags & LU_OBJECT_ALLOCATED)
                                 continue;
                         clean = 0;
                         scan->lo_header = top->lo_header;
-                        result = scan->lo_ops->loo_object_init(env, scan);
+                        result = scan->lo_ops->loo_object_init(env, scan, conf);
                         if (result != 0) {
                                 lu_object_free(env, top);
                                 RETURN(ERR_PTR(result));
@@ -175,7 +182,7 @@ static struct lu_object *lu_object_alloc(const struct lu_env *env,
                 }
         } while (!clean);
 
-        list_for_each_entry_reverse(scan, layers, lo_linkage) {
+        cfs_list_for_each_entry_reverse(scan, layers, lo_linkage) {
                 if (scan->lo_ops->loo_object_start != NULL) {
                         result = scan->lo_ops->loo_object_start(env, scan);
                         if (result != 0) {
@@ -194,17 +201,17 @@ static struct lu_object *lu_object_alloc(const struct lu_env *env,
  */
 static void lu_object_free(const struct lu_env *env, struct lu_object *o)
 {
-        struct list_head splice;
-        struct lu_object *scan;
-        struct lu_site          *site;
-        struct list_head        *layers;
+        cfs_list_t            splice;
+        struct lu_object     *scan;
+        struct lu_site       *site;
+        cfs_list_t           *layers;
 
         site   = o->lo_dev->ld_site;
         layers = &o->lo_header->loh_layers;
         /*
          * First call ->loo_object_delete() method to release all resources.
          */
-        list_for_each_entry_reverse(scan, layers, lo_linkage) {
+        cfs_list_for_each_entry_reverse(scan, layers, lo_linkage) {
                 if (scan->lo_ops->loo_object_delete != NULL)
                         scan->lo_ops->loo_object_delete(env, scan);
         }
@@ -216,15 +223,15 @@ static void lu_object_free(const struct lu_env *env, struct lu_object *o)
          * top-level slice.
          */
         CFS_INIT_LIST_HEAD(&splice);
-        list_splice_init(layers, &splice);
-        while (!list_empty(&splice)) {
+        cfs_list_splice_init(layers, &splice);
+        while (!cfs_list_empty(&splice)) {
                 /*
                  * Free layers in bottom-to-top order, so that object header
                  * lives as long as possible and ->loo_object_free() methods
                  * can look at its contents.
                  */
                 o = container_of0(splice.prev, struct lu_object, lo_linkage);
-                list_del_init(&o->lo_linkage);
+                cfs_list_del_init(&o->lo_linkage);
                 LASSERT(o->lo_ops->loo_object_free != NULL);
                 o->lo_ops->loo_object_free(env, o);
         }
@@ -236,7 +243,7 @@ static void lu_object_free(const struct lu_env *env, struct lu_object *o)
  */
 int lu_site_purge(const struct lu_env *env, struct lu_site *s, int nr)
 {
-        struct list_head         dispose;
+        cfs_list_t               dispose;
         struct lu_object_header *h;
         struct lu_object_header *temp;
 
@@ -245,8 +252,8 @@ int lu_site_purge(const struct lu_env *env, struct lu_site *s, int nr)
          * Under LRU list lock, scan LRU list and move unreferenced objects to
          * the dispose list, removing them from LRU and hash table.
          */
-        write_lock(&s->ls_guard);
-        list_for_each_entry_safe(h, temp, &s->ls_lru, loh_lru) {
+        cfs_write_lock(&s->ls_guard);
+        cfs_list_for_each_entry_safe(h, temp, &s->ls_lru, loh_lru) {
                 /*
                  * Objects are sorted in lru order, and "busy" objects (ones
                  * with h->loh_ref > 0) naturally tend to live near hot end
@@ -259,21 +266,21 @@ int lu_site_purge(const struct lu_env *env, struct lu_site *s, int nr)
                  */
                 if (nr-- == 0)
                         break;
-                if (atomic_read(&h->loh_ref) > 0)
+                if (cfs_atomic_read(&h->loh_ref) > 0)
                         continue;
-                hlist_del_init(&h->loh_hash);
-                list_move(&h->loh_lru, &dispose);
+                cfs_hlist_del_init(&h->loh_hash);
+                cfs_list_move(&h->loh_lru, &dispose);
                 s->ls_total --;
         }
-        write_unlock(&s->ls_guard);
+        cfs_write_unlock(&s->ls_guard);
         /*
          * Free everything on the dispose list. This is safe against races due
          * to the reasons described in lu_object_put().
          */
-        while (!list_empty(&dispose)) {
+        while (!cfs_list_empty(&dispose)) {
                 h = container_of0(dispose.next,
                                  struct lu_object_header, loh_lru);
-                list_del_init(&h->loh_lru);
+                cfs_list_del_init(&h->loh_lru);
                 lu_object_free(env, lu_object_top(h));
                 s->ls_stats.s_lru_purged ++;
         }
@@ -336,7 +343,7 @@ int lu_cdebug_printer(const struct lu_env *env,
         struct lu_cdebug_data       *key;
         int used;
         int complete;
-       va_list args;
+        va_list args;
 
         va_start(args, format);
 
@@ -351,9 +358,10 @@ int lu_cdebug_printer(const struct lu_env *env,
         vsnprintf(key->lck_area + used,
                   ARRAY_SIZE(key->lck_area) - used, format, args);
         if (complete) {
-                libcfs_debug_msg(NULL, info->lpi_subsys, info->lpi_mask,
-                                 (char *)info->lpi_file, info->lpi_fn,
-                                 info->lpi_line, "%s", key->lck_area);
+                if (cfs_cdebug_show(info->lpi_mask, info->lpi_subsys))
+                        libcfs_debug_msg(NULL, info->lpi_subsys, info->lpi_mask,
+                                         (char *)info->lpi_file, info->lpi_fn,
+                                         info->lpi_line, "%s", key->lck_area);
                 key->lck_area[0] = 0;
         }
         va_end(args);
@@ -361,23 +369,25 @@ int lu_cdebug_printer(const struct lu_env *env,
 }
 EXPORT_SYMBOL(lu_cdebug_printer);
 
-/*
+/**
  * Print object header.
  */
-static void lu_object_header_print(const struct lu_env *env,
-                                   void *cookie, lu_printer_t printer,
-                                   const struct lu_object_header *hdr)
+void lu_object_header_print(const struct lu_env *env, void *cookie,
+                            lu_printer_t printer,
+                            const struct lu_object_header *hdr)
 {
         (*printer)(env, cookie, "header@%p[%#lx, %d, "DFID"%s%s%s]",
-                   hdr, hdr->loh_flags, atomic_read(&hdr->loh_ref),
+                   hdr, hdr->loh_flags, cfs_atomic_read(&hdr->loh_ref),
                    PFID(&hdr->loh_fid),
-                   hlist_unhashed(&hdr->loh_hash) ? "" : " hash",
-                   list_empty(&hdr->loh_lru) ? "" : " lru",
+                   cfs_hlist_unhashed(&hdr->loh_hash) ? "" : " hash",
+                   cfs_list_empty((cfs_list_t *)&hdr->loh_lru) ? \
+                   "" : " lru",
                    hdr->loh_attr & LOHA_EXISTS ? " exist":"");
 }
+EXPORT_SYMBOL(lu_object_header_print);
 
-/*
- * Print human readable representation of the @o to the @printer.
+/**
+ * Print human readable representation of the \a o to the \a printer.
  */
 void lu_object_print(const struct lu_env *env, void *cookie,
                      lu_printer_t printer, const struct lu_object *o)
@@ -388,21 +398,24 @@ void lu_object_print(const struct lu_env *env, void *cookie,
 
         top = o->lo_header;
         lu_object_header_print(env, cookie, printer, top);
-        (*printer)(env, cookie, "\n");
-        list_for_each_entry(o, &top->loh_layers, lo_linkage) {
+        (*printer)(env, cookie, "\n");
+        cfs_list_for_each_entry(o, &top->loh_layers, lo_linkage) {
                 depth = o->lo_depth + 4;
-                LASSERT(o->lo_ops->loo_object_print != NULL);
+
                 /*
-                 * print `.' @depth times.
+                 * print `.' \a depth times followed by type name and address
                  */
-                (*printer)(env, cookie, "%*.*s", depth, depth, ruler);
-                o->lo_ops->loo_object_print(env, cookie, printer, o);
+                (*printer)(env, cookie, "%*.*s%s@%p", depth, depth, ruler,
+                           o->lo_dev->ld_type->ldt_name, o);
+                if (o->lo_ops->loo_object_print != NULL)
+                        o->lo_ops->loo_object_print(env, cookie, printer, o);
                 (*printer)(env, cookie, "\n");
         }
+        (*printer)(env, cookie, "} header@%p\n", top);
 }
 EXPORT_SYMBOL(lu_object_print);
 
-/*
+/**
  * Check object consistency.
  */
 int lu_object_invariant(const struct lu_object *o)
@@ -410,7 +423,7 @@ int lu_object_invariant(const struct lu_object *o)
         struct lu_object_header *top;
 
         top = o->lo_header;
-        list_for_each_entry(o, &top->loh_layers, lo_linkage) {
+        cfs_list_for_each_entry(o, &top->loh_layers, lo_linkage) {
                 if (o->lo_ops->loo_object_invariant != NULL &&
                     !o->lo_ops->loo_object_invariant(o))
                         return 0;
@@ -420,18 +433,32 @@ int lu_object_invariant(const struct lu_object *o)
 EXPORT_SYMBOL(lu_object_invariant);
 
 static struct lu_object *htable_lookup(struct lu_site *s,
-                                       const struct hlist_head *bucket,
-                                       const struct lu_fid *f)
+                                       const cfs_hlist_head_t *bucket,
+                                       const struct lu_fid *f,
+                                       cfs_waitlink_t *waiter)
 {
         struct lu_object_header *h;
-        struct hlist_node *scan;
+        cfs_hlist_node_t *scan;
 
-        hlist_for_each_entry(h, scan, bucket, loh_hash) {
+        cfs_hlist_for_each_entry(h, scan, bucket, loh_hash) {
                 s->ls_stats.s_cache_check ++;
-                if (likely(lu_fid_eq(&h->loh_fid, f) &&
-                           !lu_object_is_dying(h))) {
+                if (likely(lu_fid_eq(&h->loh_fid, f))) {
+                        if (unlikely(lu_object_is_dying(h))) {
+                                /*
+                                 * Lookup found an object being destroyed;
+                                 * this object cannot be returned (to assure
+                                 * that references to dying objects are
+                                 * eventually drained), and moreover, lookup
+                                 * has to wait until object is freed.
+                                 */
+                                cfs_waitlink_init(waiter);
+                                cfs_waitq_add(&s->ls_marche_funebre, waiter);
+                                cfs_set_current_state(CFS_TASK_UNINT);
+                                s->ls_stats.s_cache_death_race ++;
+                                return ERR_PTR(-EAGAIN);
+                        }
                         /* bump reference count... */
-                        if (atomic_add_return(1, &h->loh_ref) == 1)
+                        if (cfs_atomic_add_return(1, &h->loh_ref) == 1)
                                 ++ s->ls_busy;
                         /* and move to the head of the LRU */
                         /*
@@ -451,20 +478,35 @@ static __u32 fid_hash(const struct lu_fid *f, int bits)
 {
         /* all objects with same id and different versions will belong to same
          * collisions list. */
-        return hash_long(fid_flatten(f), bits);
+        return cfs_hash_long(fid_flatten(f), bits);
 }
 
-/*
- * Search cache for an object with the fid @f. If such object is found, return
- * it. Otherwise, create new object, insert it into cache and return it. In
- * any case, additional reference is acquired on the returned object.
+/**
+ * Search cache for an object with the fid \a f. If such object is found,
+ * return it. Otherwise, create new object, insert it into cache and return
+ * it. In any case, additional reference is acquired on the returned object.
  */
 struct lu_object *lu_object_find(const struct lu_env *env,
-                                 struct lu_site *s, const struct lu_fid *f)
+                                 struct lu_device *dev, const struct lu_fid *f,
+                                 const struct lu_object_conf *conf)
+{
+        return lu_object_find_at(env, dev->ld_site->ls_top_dev, f, conf);
+}
+EXPORT_SYMBOL(lu_object_find);
+
+/**
+ * Core logic of lu_object_find*() functions.
+ */
+static struct lu_object *lu_object_find_try(const struct lu_env *env,
+                                            struct lu_device *dev,
+                                            const struct lu_fid *f,
+                                            const struct lu_object_conf *conf,
+                                            cfs_waitlink_t *waiter)
 {
-        struct lu_object     *o;
-        struct lu_object     *shadow;
-        struct hlist_head *bucket;
+        struct lu_site        *s;
+        struct lu_object      *o;
+        struct lu_object      *shadow;
+        cfs_hlist_head_t      *bucket;
 
         /*
          * This uses standard index maintenance protocol:
@@ -478,13 +520,17 @@ struct lu_object *lu_object_find(const struct lu_env *env,
          *       object just allocated.
          *     - unlock index;
          *     - return object.
+         *
+         * If dying object is found during index search, add @waiter to the
+         * site wait-queue and return ERR_PTR(-EAGAIN).
          */
 
+        s = dev->ld_site;
         bucket = s->ls_hash + fid_hash(f, s->ls_hash_bits);
 
-        read_lock(&s->ls_guard);
-        o = htable_lookup(s, bucket, f);
-        read_unlock(&s->ls_guard);
+        cfs_read_lock(&s->ls_guard);
+        o = htable_lookup(s, bucket, f, waiter);
+        cfs_read_unlock(&s->ls_guard);
 
         if (o != NULL)
                 return o;
@@ -493,43 +539,128 @@ struct lu_object *lu_object_find(const struct lu_env *env,
          * Allocate new object. This may result in rather complicated
          * operations, including fld queries, inode loading, etc.
          */
-        o = lu_object_alloc(env, s, f);
+        o = lu_object_alloc(env, dev, f, conf);
         if (unlikely(IS_ERR(o)))
                 return o;
 
         LASSERT(lu_fid_eq(lu_object_fid(o), f));
 
-        write_lock(&s->ls_guard);
-        shadow = htable_lookup(s, bucket, f);
+        cfs_write_lock(&s->ls_guard);
+        shadow = htable_lookup(s, bucket, f, waiter);
         if (likely(shadow == NULL)) {
-                hlist_add_head(&o->lo_header->loh_hash, bucket);
-                list_add_tail(&o->lo_header->loh_lru, &s->ls_lru);
+                cfs_hlist_add_head(&o->lo_header->loh_hash, bucket);
+                cfs_list_add_tail(&o->lo_header->loh_lru, &s->ls_lru);
                 ++ s->ls_busy;
                 ++ s->ls_total;
                 shadow = o;
                 o = NULL;
         } else
                 s->ls_stats.s_cache_race ++;
-        write_unlock(&s->ls_guard);
+        cfs_write_unlock(&s->ls_guard);
         if (o != NULL)
                 lu_object_free(env, o);
         return shadow;
 }
-EXPORT_SYMBOL(lu_object_find);
 
-/*
+/**
+ * Much like lu_object_find(), but top level device of object is specifically
+ * \a dev rather than top level device of the site. This interface allows
+ * objects of different "stacking" to be created within the same site.
+ */
+struct lu_object *lu_object_find_at(const struct lu_env *env,
+                                    struct lu_device *dev,
+                                    const struct lu_fid *f,
+                                    const struct lu_object_conf *conf)
+{
+        struct lu_object *obj;
+        cfs_waitlink_t    wait;
+
+        while (1) {
+                obj = lu_object_find_try(env, dev, f, conf, &wait);
+                if (obj == ERR_PTR(-EAGAIN)) {
+                        /*
+                         * lu_object_find_try() already added waiter into the
+                         * wait queue.
+                         */
+                        cfs_waitq_wait(&wait, CFS_TASK_UNINT);
+                        cfs_waitq_del(&dev->ld_site->ls_marche_funebre, &wait);
+                } else
+                        break;
+        }
+        return obj;
+}
+EXPORT_SYMBOL(lu_object_find_at);
+
+/**
+ * Find object with given fid, and return its slice belonging to given device.
+ */
+struct lu_object *lu_object_find_slice(const struct lu_env *env,
+                                       struct lu_device *dev,
+                                       const struct lu_fid *f,
+                                       const struct lu_object_conf *conf)
+{
+        struct lu_object *top;
+        struct lu_object *obj;
+
+        top = lu_object_find(env, dev, f, conf);
+        if (!IS_ERR(top)) {
+                obj = lu_object_locate(top->lo_header, dev->ld_type);
+                if (obj == NULL)
+                        lu_object_put(env, top);
+        } else
+                obj = top;
+        return obj;
+}
+EXPORT_SYMBOL(lu_object_find_slice);
+
+/**
+ * Global list of all device types.
+ */
+static CFS_LIST_HEAD(lu_device_types);
+
+int lu_device_type_init(struct lu_device_type *ldt)
+{
+        int result;
+
+        CFS_INIT_LIST_HEAD(&ldt->ldt_linkage);
+        result = ldt->ldt_ops->ldto_init(ldt);
+        if (result == 0)
+                cfs_list_add(&ldt->ldt_linkage, &lu_device_types);
+        return result;
+}
+EXPORT_SYMBOL(lu_device_type_init);
+
+void lu_device_type_fini(struct lu_device_type *ldt)
+{
+        cfs_list_del_init(&ldt->ldt_linkage);
+        ldt->ldt_ops->ldto_fini(ldt);
+}
+EXPORT_SYMBOL(lu_device_type_fini);
+
+void lu_types_stop(void)
+{
+        struct lu_device_type *ldt;
+
+        cfs_list_for_each_entry(ldt, &lu_device_types, ldt_linkage) {
+                if (ldt->ldt_device_nr == 0)
+                        ldt->ldt_ops->ldto_stop(ldt);
+        }
+}
+EXPORT_SYMBOL(lu_types_stop);
+
+/**
  * Global list of all sites on this node
  */
 static CFS_LIST_HEAD(lu_sites);
-static DECLARE_MUTEX(lu_sites_guard);
+static CFS_DECLARE_MUTEX(lu_sites_guard);
 
-/*
+/**
  * Global environment used by site shrinker.
  */
 static struct lu_env lu_shrink_env;
 
-/*
- * Print all objects in @s.
+/**
+ * Print all objects in \a s.
  */
 void lu_site_print(const struct lu_env *env, struct lu_site *s, void *cookie,
                    lu_printer_t printer)
@@ -538,12 +669,12 @@ void lu_site_print(const struct lu_env *env, struct lu_site *s, void *cookie,
 
         for (i = 0; i < s->ls_hash_size; ++i) {
                 struct lu_object_header *h;
-                struct hlist_node       *scan;
+                cfs_hlist_node_t        *scan;
 
-                read_lock(&s->ls_guard);
-                hlist_for_each_entry(h, scan, &s->ls_hash[i], loh_hash) {
+                cfs_read_lock(&s->ls_guard);
+                cfs_hlist_for_each_entry(h, scan, &s->ls_hash[i], loh_hash) {
 
-                        if (!list_empty(&h->loh_layers)) {
+                        if (!cfs_list_empty(&h->loh_layers)) {
                                 const struct lu_object *obj;
 
                                 obj = lu_object_top(h);
@@ -551,7 +682,7 @@ void lu_site_print(const struct lu_env *env, struct lu_site *s, void *cookie,
                         } else
                                 lu_object_header_print(env, cookie, printer, h);
                 }
-                read_unlock(&s->ls_guard);
+                cfs_read_unlock(&s->ls_guard);
         }
 }
 EXPORT_SYMBOL(lu_site_print);
@@ -560,7 +691,7 @@ enum {
         LU_CACHE_PERCENT   = 20,
 };
 
-/*
+/**
  * Return desired hash table order.
  */
 static int lu_htable_order(void)
@@ -575,7 +706,7 @@ static int lu_htable_order(void)
          *
          * Size of lu_object is (arbitrary) taken as 1K (together with inode).
          */
-        cache_size = num_physpages;
+        cache_size = cfs_num_physpages;
 
 #if BITS_PER_LONG == 32
         /* limit hashtable size for lowmem systems to low RAM */
@@ -592,8 +723,10 @@ static int lu_htable_order(void)
         return bits;
 }
 
-/*
- * Initialize site @s, with @d as the top level device.
+static cfs_lock_class_key_t lu_site_guard_class;
+
+/**
+ * Initialize site \a s, with \a d as the top level device.
  */
 int lu_site_init(struct lu_site *s, struct lu_device *top)
 {
@@ -603,12 +736,15 @@ int lu_site_init(struct lu_site *s, struct lu_device *top)
         ENTRY;
 
         memset(s, 0, sizeof *s);
-        rwlock_init(&s->ls_guard);
+        cfs_rwlock_init(&s->ls_guard);
+        cfs_lockdep_set_class(&s->ls_guard, &lu_site_guard_class);
         CFS_INIT_LIST_HEAD(&s->ls_lru);
         CFS_INIT_LIST_HEAD(&s->ls_linkage);
+        cfs_waitq_init(&s->ls_marche_funebre);
         s->ls_top_dev = top;
         top->ld_site = s;
         lu_device_get(top);
+        lu_ref_add(&top->ld_reference, "site-top", s);
 
         for (bits = lu_htable_order(), size = 1 << bits;
              (s->ls_hash =
@@ -625,101 +761,114 @@ int lu_site_init(struct lu_site *s, struct lu_device *top)
         s->ls_hash_mask = size - 1;
 
         for (i = 0; i < size; i++)
-                INIT_HLIST_HEAD(&s->ls_hash[i]);
+                CFS_INIT_HLIST_HEAD(&s->ls_hash[i]);
 
         RETURN(0);
 }
 EXPORT_SYMBOL(lu_site_init);
 
-/*
- * Finalize @s and release its resources.
+/**
+ * Finalize \a s and release its resources.
  */
 void lu_site_fini(struct lu_site *s)
 {
-        LASSERT(list_empty(&s->ls_lru));
+        LASSERT(cfs_list_empty(&s->ls_lru));
         LASSERT(s->ls_total == 0);
 
-        down(&lu_sites_guard);
-        list_del_init(&s->ls_linkage);
-        up(&lu_sites_guard);
+        cfs_down(&lu_sites_guard);
+        cfs_list_del_init(&s->ls_linkage);
+        cfs_up(&lu_sites_guard);
 
         if (s->ls_hash != NULL) {
                 int i;
                 for (i = 0; i < s->ls_hash_size; i++)
-                        LASSERT(hlist_empty(&s->ls_hash[i]));
+                        LASSERT(cfs_hlist_empty(&s->ls_hash[i]));
                 cfs_free_large(s->ls_hash);
                 s->ls_hash = NULL;
         }
         if (s->ls_top_dev != NULL) {
                 s->ls_top_dev->ld_site = NULL;
+                lu_ref_del(&s->ls_top_dev->ld_reference, "site-top", s);
                 lu_device_put(s->ls_top_dev);
                 s->ls_top_dev = NULL;
         }
 }
 EXPORT_SYMBOL(lu_site_fini);
 
-/*
+/**
  * Called when initialization of stack for this site is completed.
  */
 int lu_site_init_finish(struct lu_site *s)
 {
         int result;
-        down(&lu_sites_guard);
+        cfs_down(&lu_sites_guard);
         result = lu_context_refill(&lu_shrink_env.le_ctx);
         if (result == 0)
-                list_add(&s->ls_linkage, &lu_sites);
-        up(&lu_sites_guard);
+                cfs_list_add(&s->ls_linkage, &lu_sites);
+        cfs_up(&lu_sites_guard);
         return result;
 }
 EXPORT_SYMBOL(lu_site_init_finish);
 
-/*
- * Acquire additional reference on device @d
+/**
+ * Acquire additional reference on device \a d
  */
 void lu_device_get(struct lu_device *d)
 {
-        atomic_inc(&d->ld_ref);
+        cfs_atomic_inc(&d->ld_ref);
 }
 EXPORT_SYMBOL(lu_device_get);
 
-/*
- * Release reference on device @d.
+/**
+ * Release reference on device \a d.
  */
 void lu_device_put(struct lu_device *d)
 {
-        atomic_dec(&d->ld_ref);
+        LASSERT(cfs_atomic_read(&d->ld_ref) > 0);
+        cfs_atomic_dec(&d->ld_ref);
 }
 EXPORT_SYMBOL(lu_device_put);
 
-/*
- * Initialize device @d of type @t.
+/**
+ * Initialize device \a d of type \a t.
  */
 int lu_device_init(struct lu_device *d, struct lu_device_type *t)
 {
+        if (t->ldt_device_nr++ == 0 && t->ldt_ops->ldto_start != NULL)
+                t->ldt_ops->ldto_start(t);
         memset(d, 0, sizeof *d);
-        atomic_set(&d->ld_ref, 0);
+        cfs_atomic_set(&d->ld_ref, 0);
         d->ld_type = t;
+        lu_ref_init(&d->ld_reference);
         return 0;
 }
 EXPORT_SYMBOL(lu_device_init);
 
-/*
- * Finalize device @d.
+/**
+ * Finalize device \a d.
  */
 void lu_device_fini(struct lu_device *d)
 {
-        if (d->ld_obd != NULL)
-                /* finish lprocfs */
-                lprocfs_obd_cleanup(d->ld_obd);
+        struct lu_device_type *t;
 
-        LASSERTF(atomic_read(&d->ld_ref) == 0,
-                 "Refcount is %u\n", atomic_read(&d->ld_ref));
+        t = d->ld_type;
+        if (d->ld_obd != NULL) {
+                d->ld_obd->obd_lu_dev = NULL;
+                d->ld_obd = NULL;
+        }
+
+        lu_ref_fini(&d->ld_reference);
+        LASSERTF(cfs_atomic_read(&d->ld_ref) == 0,
+                 "Refcount is %u\n", cfs_atomic_read(&d->ld_ref));
+        LASSERT(t->ldt_device_nr > 0);
+        if (--t->ldt_device_nr == 0 && t->ldt_ops->ldto_stop != NULL)
+                t->ldt_ops->ldto_stop(t);
 }
 EXPORT_SYMBOL(lu_device_fini);
 
-/*
- * Initialize object @o that is part of compound object @h and was created by
- * device @d.
+/**
+ * Initialize object \a o that is part of compound object \a h and was created
+ * by device \a d.
  */
 int lu_object_init(struct lu_object *o,
                    struct lu_object_header *h, struct lu_device *d)
@@ -728,34 +877,39 @@ int lu_object_init(struct lu_object *o,
         o->lo_header = h;
         o->lo_dev    = d;
         lu_device_get(d);
+        o->lo_dev_ref = lu_ref_add(&d->ld_reference, "lu_object", o);
         CFS_INIT_LIST_HEAD(&o->lo_linkage);
         return 0;
 }
 EXPORT_SYMBOL(lu_object_init);
 
-/*
+/**
  * Finalize object and release its resources.
  */
 void lu_object_fini(struct lu_object *o)
 {
-        LASSERT(list_empty(&o->lo_linkage));
+        struct lu_device *dev = o->lo_dev;
 
-        if (o->lo_dev != NULL) {
-                lu_device_put(o->lo_dev);
+        LASSERT(cfs_list_empty(&o->lo_linkage));
+
+        if (dev != NULL) {
+                lu_ref_del_at(&dev->ld_reference,
+                              o->lo_dev_ref , "lu_object", o);
+                lu_device_put(dev);
                 o->lo_dev = NULL;
         }
 }
 EXPORT_SYMBOL(lu_object_fini);
 
-/*
- * Add object @o as first layer of compound object @h
+/**
+ * Add object \a o as first layer of compound object \a h
  *
  * This is typically called by the ->ldo_object_alloc() method of top-level
  * device.
  */
 void lu_object_add_top(struct lu_object_header *h, struct lu_object *o)
 {
-        list_move(&o->lo_linkage, &h->loh_layers);
+        cfs_list_move(&o->lo_linkage, &h->loh_layers);
 }
 EXPORT_SYMBOL(lu_object_add_top);
 
@@ -767,7 +921,7 @@ EXPORT_SYMBOL(lu_object_add_top);
  */
 void lu_object_add(struct lu_object *before, struct lu_object *o)
 {
-        list_move(&o->lo_linkage, &before->lo_linkage);
+        cfs_list_move(&o->lo_linkage, &before->lo_linkage);
 }
 EXPORT_SYMBOL(lu_object_add);
 
@@ -777,35 +931,37 @@ EXPORT_SYMBOL(lu_object_add);
 int lu_object_header_init(struct lu_object_header *h)
 {
         memset(h, 0, sizeof *h);
-        atomic_set(&h->loh_ref, 1);
-        INIT_HLIST_NODE(&h->loh_hash);
+        cfs_atomic_set(&h->loh_ref, 1);
+        CFS_INIT_HLIST_NODE(&h->loh_hash);
         CFS_INIT_LIST_HEAD(&h->loh_lru);
         CFS_INIT_LIST_HEAD(&h->loh_layers);
+        lu_ref_init(&h->loh_reference);
         return 0;
 }
 EXPORT_SYMBOL(lu_object_header_init);
 
-/*
+/**
  * Finalize compound object.
  */
 void lu_object_header_fini(struct lu_object_header *h)
 {
-        LASSERT(list_empty(&h->loh_layers));
-        LASSERT(list_empty(&h->loh_lru));
-        LASSERT(hlist_unhashed(&h->loh_hash));
+        LASSERT(cfs_list_empty(&h->loh_layers));
+        LASSERT(cfs_list_empty(&h->loh_lru));
+        LASSERT(cfs_hlist_unhashed(&h->loh_hash));
+        lu_ref_fini(&h->loh_reference);
 }
 EXPORT_SYMBOL(lu_object_header_fini);
 
-/*
+/**
  * Given a compound object, find its slice, corresponding to the device type
- * @dtype.
+ * \a dtype.
  */
 struct lu_object *lu_object_locate(struct lu_object_header *h,
-                                   struct lu_device_type *dtype)
+                                   const struct lu_device_type *dtype)
 {
         struct lu_object *o;
 
-        list_for_each_entry(o, &h->loh_layers, lo_linkage) {
+        cfs_list_for_each_entry(o, &h->loh_layers, lo_linkage) {
                 if (o->lo_dev->ld_type == dtype)
                         return o;
         }
@@ -815,9 +971,9 @@ EXPORT_SYMBOL(lu_object_locate);
 
 
 
-/*
+/**
  * Finalize and free devices in the device stack.
- * 
+ *
  * Finalize device stack by purging object cache, and calling
  * lu_device_type_operations::ldto_device_fini() and
  * lu_device_type_operations::ldto_device_free() on all devices in the stack.
@@ -831,13 +987,14 @@ void lu_stack_fini(const struct lu_env *env, struct lu_device *top)
         lu_site_purge(env, site, ~0);
         for (scan = top; scan != NULL; scan = next) {
                 next = scan->ld_type->ldt_ops->ldto_device_fini(env, scan);
+                lu_ref_del(&scan->ld_reference, "lu-stack", &lu_site_init);
                 lu_device_put(scan);
         }
 
         /* purge again. */
         lu_site_purge(env, site, ~0);
 
-        if (!list_empty(&site->ls_lru) || site->ls_total != 0) {
+        if (!cfs_list_empty(&site->ls_lru) || site->ls_total != 0) {
                 /*
                  * Uh-oh, objects still exist.
                  */
@@ -852,24 +1009,34 @@ void lu_stack_fini(const struct lu_env *env, struct lu_device *top)
 
                 next = ldt->ldt_ops->ldto_device_free(env, scan);
                 type = ldt->ldt_obd_type;
-                type->typ_refcnt--;
-                class_put_type(type);
+                if (type != NULL) {
+                        type->typ_refcnt--;
+                        class_put_type(type);
+                }
         }
 }
 EXPORT_SYMBOL(lu_stack_fini);
 
 enum {
-        /*
+        /**
          * Maximal number of tld slots.
          */
-        LU_CONTEXT_KEY_NR = 16
+        LU_CONTEXT_KEY_NR = 32
 };
 
 static struct lu_context_key *lu_keys[LU_CONTEXT_KEY_NR] = { NULL, };
 
-static spinlock_t lu_keys_guard = SPIN_LOCK_UNLOCKED;
+static cfs_spinlock_t lu_keys_guard = CFS_SPIN_LOCK_UNLOCKED;
 
-/*
+/**
+ * Global counter incremented whenever key is registered, unregistered,
+ * revived or quiesced. This is used to void unnecessary calls to
+ * lu_context_refill(). No locking is provided, as initialization and shutdown
+ * are supposed to be externally serialized.
+ */
+static unsigned key_set_version = 0;
+
+/**
  * Register new key.
  */
 int lu_context_key_register(struct lu_context_key *key)
@@ -883,76 +1050,210 @@ int lu_context_key_register(struct lu_context_key *key)
         LASSERT(key->lct_owner != NULL);
 
         result = -ENFILE;
-        spin_lock(&lu_keys_guard);
+        cfs_spin_lock(&lu_keys_guard);
         for (i = 0; i < ARRAY_SIZE(lu_keys); ++i) {
                 if (lu_keys[i] == NULL) {
                         key->lct_index = i;
-                        atomic_set(&key->lct_used, 1);
+                        cfs_atomic_set(&key->lct_used, 1);
                         lu_keys[i] = key;
+                        lu_ref_init(&key->lct_reference);
                         result = 0;
+                        ++key_set_version;
                         break;
                 }
         }
-        spin_unlock(&lu_keys_guard);
+        cfs_spin_unlock(&lu_keys_guard);
         return result;
 }
 EXPORT_SYMBOL(lu_context_key_register);
 
 static void key_fini(struct lu_context *ctx, int index)
 {
-        if (ctx->lc_value[index] != NULL) {
+        if (ctx->lc_value != NULL && ctx->lc_value[index] != NULL) {
                 struct lu_context_key *key;
 
                 key = lu_keys[index];
                 LASSERT(key != NULL);
                 LASSERT(key->lct_fini != NULL);
-                LASSERT(atomic_read(&key->lct_used) > 1);
+                LASSERT(cfs_atomic_read(&key->lct_used) > 1);
 
                 key->lct_fini(ctx, key, ctx->lc_value[index]);
-                atomic_dec(&key->lct_used);
+                lu_ref_del(&key->lct_reference, "ctx", ctx);
+                cfs_atomic_dec(&key->lct_used);
                 LASSERT(key->lct_owner != NULL);
                 if (!(ctx->lc_tags & LCT_NOREF)) {
-                        LASSERT(module_refcount(key->lct_owner) > 0);
-                        module_put(key->lct_owner);
+                        LASSERT(cfs_module_refcount(key->lct_owner) > 0);
+                        cfs_module_put(key->lct_owner);
                 }
                 ctx->lc_value[index] = NULL;
         }
 }
 
-/*
+/**
  * Deregister key.
  */
 void lu_context_key_degister(struct lu_context_key *key)
 {
-        LASSERT(atomic_read(&key->lct_used) >= 1);
-        LASSERT(0 <= key->lct_index && key->lct_index < ARRAY_SIZE(lu_keys));
+        LASSERT(cfs_atomic_read(&key->lct_used) >= 1);
+        LINVRNT(0 <= key->lct_index && key->lct_index < ARRAY_SIZE(lu_keys));
+
+        lu_context_key_quiesce(key);
 
+        ++key_set_version;
+        cfs_spin_lock(&lu_keys_guard);
         key_fini(&lu_shrink_env.le_ctx, key->lct_index);
+        if (lu_keys[key->lct_index]) {
+                lu_keys[key->lct_index] = NULL;
+                lu_ref_fini(&key->lct_reference);
+        }
+        cfs_spin_unlock(&lu_keys_guard);
 
-        if (atomic_read(&key->lct_used) > 1)
-                CERROR("key has instances.\n");
-        spin_lock(&lu_keys_guard);
-        lu_keys[key->lct_index] = NULL;
-        spin_unlock(&lu_keys_guard);
+        LASSERTF(cfs_atomic_read(&key->lct_used) == 1,
+                 "key has instances: %d\n",
+                 cfs_atomic_read(&key->lct_used));
 }
 EXPORT_SYMBOL(lu_context_key_degister);
 
-/*
- * Return value associated with key @key in context @ctx.
+/**
+ * Register a number of keys. This has to be called after all keys have been
+ * initialized by a call to LU_CONTEXT_KEY_INIT().
+ */
+int lu_context_key_register_many(struct lu_context_key *k, ...)
+{
+        struct lu_context_key *key = k;
+        va_list args;
+        int result;
+
+        va_start(args, k);
+        do {
+                result = lu_context_key_register(key);
+                if (result)
+                        break;
+                key = va_arg(args, struct lu_context_key *);
+        } while (key != NULL);
+        va_end(args);
+
+        if (result != 0) {
+                va_start(args, k);
+                while (k != key) {
+                        lu_context_key_degister(k);
+                        k = va_arg(args, struct lu_context_key *);
+                }
+                va_end(args);
+        }
+
+        return result;
+}
+EXPORT_SYMBOL(lu_context_key_register_many);
+
+/**
+ * De-register a number of keys. This is a dual to
+ * lu_context_key_register_many().
+ */
+void lu_context_key_degister_many(struct lu_context_key *k, ...)
+{
+        va_list args;
+
+        va_start(args, k);
+        do {
+                lu_context_key_degister(k);
+                k = va_arg(args, struct lu_context_key*);
+        } while (k != NULL);
+        va_end(args);
+}
+EXPORT_SYMBOL(lu_context_key_degister_many);
+
+/**
+ * Revive a number of keys.
+ */
+void lu_context_key_revive_many(struct lu_context_key *k, ...)
+{
+        va_list args;
+
+        va_start(args, k);
+        do {
+                lu_context_key_revive(k);
+                k = va_arg(args, struct lu_context_key*);
+        } while (k != NULL);
+        va_end(args);
+}
+EXPORT_SYMBOL(lu_context_key_revive_many);
+
+/**
+ * Quiescent a number of keys.
+ */
+void lu_context_key_quiesce_many(struct lu_context_key *k, ...)
+{
+        va_list args;
+
+        va_start(args, k);
+        do {
+                lu_context_key_quiesce(k);
+                k = va_arg(args, struct lu_context_key*);
+        } while (k != NULL);
+        va_end(args);
+}
+EXPORT_SYMBOL(lu_context_key_quiesce_many);
+
+/**
+ * Return value associated with key \a key in context \a ctx.
  */
 void *lu_context_key_get(const struct lu_context *ctx,
-                         struct lu_context_key *key)
+                         const struct lu_context_key *key)
 {
-        LASSERT(ctx->lc_state == LCS_ENTERED);
-        LASSERT(0 <= key->lct_index && key->lct_index < ARRAY_SIZE(lu_keys));
+        LINVRNT(ctx->lc_state == LCS_ENTERED);
+        LINVRNT(0 <= key->lct_index && key->lct_index < ARRAY_SIZE(lu_keys));
+        LASSERT(lu_keys[key->lct_index] == key);
         return ctx->lc_value[key->lct_index];
 }
 EXPORT_SYMBOL(lu_context_key_get);
 
+/**
+ * List of remembered contexts. XXX document me.
+ */
+static CFS_LIST_HEAD(lu_context_remembered);
+
+/**
+ * Destroy \a key in all remembered contexts. This is used to destroy key
+ * values in "shared" contexts (like service threads), when a module owning
+ * the key is about to be unloaded.
+ */
+void lu_context_key_quiesce(struct lu_context_key *key)
+{
+        struct lu_context *ctx;
+        extern unsigned cl_env_cache_purge(unsigned nr);
+
+        if (!(key->lct_tags & LCT_QUIESCENT)) {
+                /*
+                 * XXX layering violation.
+                 */
+                cl_env_cache_purge(~0);
+                key->lct_tags |= LCT_QUIESCENT;
+                /*
+                 * XXX memory barrier has to go here.
+                 */
+                cfs_spin_lock(&lu_keys_guard);
+                cfs_list_for_each_entry(ctx, &lu_context_remembered,
+                                        lc_remember)
+                        key_fini(ctx, key->lct_index);
+                cfs_spin_unlock(&lu_keys_guard);
+                ++key_set_version;
+        }
+}
+EXPORT_SYMBOL(lu_context_key_quiesce);
+
+void lu_context_key_revive(struct lu_context_key *key)
+{
+        key->lct_tags &= ~LCT_QUIESCENT;
+        ++key_set_version;
+}
+EXPORT_SYMBOL(lu_context_key_revive);
+
 static void keys_fini(struct lu_context *ctx)
 {
         int i;
 
+        cfs_spin_lock(&lu_keys_guard);
         if (ctx->lc_value != NULL) {
                 for (i = 0; i < ARRAY_SIZE(lu_keys); ++i)
                         key_fini(ctx, i);
@@ -960,9 +1261,10 @@ static void keys_fini(struct lu_context *ctx)
                          ARRAY_SIZE(lu_keys) * sizeof ctx->lc_value[0]);
                 ctx->lc_value = NULL;
         }
+        cfs_spin_unlock(&lu_keys_guard);
 }
 
-static int keys_fill(const struct lu_context *ctx)
+static int keys_fill(struct lu_context *ctx)
 {
         int i;
 
@@ -970,22 +1272,37 @@ static int keys_fill(const struct lu_context *ctx)
                 struct lu_context_key *key;
 
                 key = lu_keys[i];
-                if (ctx->lc_value[i] == NULL &&
-                    key != NULL && key->lct_tags & ctx->lc_tags) {
+                if (ctx->lc_value[i] == NULL && key != NULL &&
+                    (key->lct_tags & ctx->lc_tags) &&
+                    /*
+                     * Don't create values for a LCT_QUIESCENT key, as this
+                     * will pin module owning a key.
+                     */
+                    !(key->lct_tags & LCT_QUIESCENT)) {
                         void *value;
 
-                        LASSERT(key->lct_init != NULL);
-                        LASSERT(key->lct_index == i);
+                        LINVRNT(key->lct_init != NULL);
+                        LINVRNT(key->lct_index == i);
 
                         value = key->lct_init(ctx, key);
                         if (unlikely(IS_ERR(value)))
                                 return PTR_ERR(value);
+
                         LASSERT(key->lct_owner != NULL);
                         if (!(ctx->lc_tags & LCT_NOREF))
-                                try_module_get(key->lct_owner);
-                        atomic_inc(&key->lct_used);
+                                cfs_try_module_get(key->lct_owner);
+                        lu_ref_add_atomic(&key->lct_reference, "ctx", ctx);
+                        cfs_atomic_inc(&key->lct_used);
+                        /*
+                         * This is the only place in the code, where an
+                         * element of ctx->lc_value[] array is set to non-NULL
+                         * value.
+                         */
                         ctx->lc_value[i] = value;
+                        if (key->lct_exit != NULL)
+                                ctx->lc_tags |= LCT_HAS_EXIT;
                 }
+                ctx->lc_version = key_set_version;
         }
         return 0;
 }
@@ -1005,7 +1322,7 @@ static int keys_init(struct lu_context *ctx)
         return result;
 }
 
-/*
+/**
  * Initialize context data-structure. Create values for all keys.
  */
 int lu_context_init(struct lu_context *ctx, __u32 tags)
@@ -1013,41 +1330,50 @@ int lu_context_init(struct lu_context *ctx, __u32 tags)
         memset(ctx, 0, sizeof *ctx);
         ctx->lc_state = LCS_INITIALIZED;
         ctx->lc_tags = tags;
+        if (tags & LCT_REMEMBER) {
+                cfs_spin_lock(&lu_keys_guard);
+                cfs_list_add(&ctx->lc_remember, &lu_context_remembered);
+                cfs_spin_unlock(&lu_keys_guard);
+        } else
+                CFS_INIT_LIST_HEAD(&ctx->lc_remember);
         return keys_init(ctx);
 }
 EXPORT_SYMBOL(lu_context_init);
 
-/*
+/**
  * Finalize context data-structure. Destroy key values.
  */
 void lu_context_fini(struct lu_context *ctx)
 {
-        LASSERT(ctx->lc_state == LCS_INITIALIZED || ctx->lc_state == LCS_LEFT);
+        LINVRNT(ctx->lc_state == LCS_INITIALIZED || ctx->lc_state == LCS_LEFT);
         ctx->lc_state = LCS_FINALIZED;
         keys_fini(ctx);
+        cfs_spin_lock(&lu_keys_guard);
+        cfs_list_del_init(&ctx->lc_remember);
+        cfs_spin_unlock(&lu_keys_guard);
 }
 EXPORT_SYMBOL(lu_context_fini);
 
-/*
+/**
  * Called before entering context.
  */
 void lu_context_enter(struct lu_context *ctx)
 {
-        LASSERT(ctx->lc_state == LCS_INITIALIZED || ctx->lc_state == LCS_LEFT);
+        LINVRNT(ctx->lc_state == LCS_INITIALIZED || ctx->lc_state == LCS_LEFT);
         ctx->lc_state = LCS_ENTERED;
 }
 EXPORT_SYMBOL(lu_context_enter);
 
-/*
- * Called after exiting from @ctx
+/**
+ * Called after exiting from \a ctx
  */
 void lu_context_exit(struct lu_context *ctx)
 {
         int i;
 
-        LASSERT(ctx->lc_state == LCS_ENTERED);
+        LINVRNT(ctx->lc_state == LCS_ENTERED);
         ctx->lc_state = LCS_LEFT;
-        if (ctx->lc_value != NULL) {
+        if (ctx->lc_tags & LCT_HAS_EXIT && ctx->lc_value != NULL) {
                 for (i = 0; i < ARRAY_SIZE(lu_keys); ++i) {
                         if (ctx->lc_value[i] != NULL) {
                                 struct lu_context_key *key;
@@ -1063,41 +1389,27 @@ void lu_context_exit(struct lu_context *ctx)
 }
 EXPORT_SYMBOL(lu_context_exit);
 
-/*
+/**
  * Allocate for context all missing keys that were registered after context
  * creation.
  */
-int lu_context_refill(const struct lu_context *ctx)
+int lu_context_refill(struct lu_context *ctx)
 {
-        LASSERT(ctx->lc_value != NULL);
-        return keys_fill(ctx);
+        LINVRNT(ctx->lc_value != NULL);
+        return ctx->lc_version == key_set_version ? 0 : keys_fill(ctx);
 }
 EXPORT_SYMBOL(lu_context_refill);
 
-static int lu_env_setup(struct lu_env *env, struct lu_context *ses,
-                        __u32 tags, int noref)
+int lu_env_init(struct lu_env *env, __u32 tags)
 {
         int result;
 
-        LASSERT(ergo(!noref, !(tags & LCT_NOREF)));
-
-        env->le_ses = ses;
+        env->le_ses = NULL;
         result = lu_context_init(&env->le_ctx, tags);
         if (likely(result == 0))
                 lu_context_enter(&env->le_ctx);
         return result;
 }
-
-static int lu_env_init_noref(struct lu_env *env, struct lu_context *ses,
-                             __u32 tags)
-{
-        return lu_env_setup(env, ses, tags, 1);
-}
-
-int lu_env_init(struct lu_env *env, struct lu_context *ses, __u32 tags)
-{
-        return lu_env_setup(env, ses, tags, 0);
-}
 EXPORT_SYMBOL(lu_env_init);
 
 void lu_env_fini(struct lu_env *env)
@@ -1108,6 +1420,20 @@ void lu_env_fini(struct lu_env *env)
 }
 EXPORT_SYMBOL(lu_env_fini);
 
+int lu_env_refill(struct lu_env *env)
+{
+        int result;
+
+        result = lu_context_refill(&env->le_ctx);
+        if (result == 0 && env->le_ses != NULL)
+                result = lu_context_refill(env->le_ses);
+        return result;
+}
+EXPORT_SYMBOL(lu_env_refill);
+
+static struct cfs_shrinker *lu_site_shrinker = NULL;
+
+#ifdef __KERNEL__
 static int lu_cache_shrink(int nr, unsigned int gfp_mask)
 {
         struct lu_site *s;
@@ -1116,74 +1442,172 @@ static int lu_cache_shrink(int nr, unsigned int gfp_mask)
         int remain = nr;
         CFS_LIST_HEAD(splice);
 
-        if (nr != 0 && !(gfp_mask & __GFP_FS))
-                return -1;
+        if (nr != 0) {
+                if (!(gfp_mask & __GFP_FS))
+                        return -1;
+                CDEBUG(D_INODE, "Shrink %d objects\n", nr);
+        }
 
-        down(&lu_sites_guard);
-        list_for_each_entry_safe(s, tmp, &lu_sites, ls_linkage) {
+        cfs_down(&lu_sites_guard);
+        cfs_list_for_each_entry_safe(s, tmp, &lu_sites, ls_linkage) {
                 if (nr != 0) {
                         remain = lu_site_purge(&lu_shrink_env, s, remain);
                         /*
                          * Move just shrunk site to the tail of site list to
                          * assure shrinking fairness.
                          */
-                        list_move_tail(&s->ls_linkage, &splice);
+                        cfs_list_move_tail(&s->ls_linkage, &splice);
                 }
-                read_lock(&s->ls_guard);
+                cfs_read_lock(&s->ls_guard);
                 cached += s->ls_total - s->ls_busy;
-                read_unlock(&s->ls_guard);
-                if (remain <= 0)
+                cfs_read_unlock(&s->ls_guard);
+                if (nr && remain <= 0)
                         break;
         }
-        list_splice(&splice, lu_sites.prev);
-        up(&lu_sites_guard);
+        cfs_list_splice(&splice, lu_sites.prev);
+        cfs_up(&lu_sites_guard);
+
+        cached = (cached / 100) * sysctl_vfs_cache_pressure;
+        if (nr == 0)
+                CDEBUG(D_INODE, "%d objects cached\n", cached);
         return cached;
 }
 
-static struct shrinker *lu_site_shrinker = NULL;
-
 /*
+ * Debugging stuff.
+ */
+
+/**
+ * Environment to be used in debugger, contains all tags.
+ */
+struct lu_env lu_debugging_env;
+
+/**
+ * Debugging printer function using printk().
+ */
+int lu_printk_printer(const struct lu_env *env,
+                      void *unused, const char *format, ...)
+{
+        va_list args;
+
+        va_start(args, format);
+        vprintk(format, args);
+        va_end(args);
+        return 0;
+}
+
+void lu_debugging_setup(void)
+{
+        lu_env_init(&lu_debugging_env, ~0);
+}
+
+void lu_context_keys_dump(void)
+{
+        int i;
+
+        for (i = 0; i < ARRAY_SIZE(lu_keys); ++i) {
+                struct lu_context_key *key;
+
+                key = lu_keys[i];
+                if (key != NULL) {
+                        CERROR("[%i]: %p %x (%p,%p,%p) %i %i \"%s\"@%p\n",
+                               i, key, key->lct_tags,
+                               key->lct_init, key->lct_fini, key->lct_exit,
+                               key->lct_index, cfs_atomic_read(&key->lct_used),
+                               key->lct_owner ? key->lct_owner->name : "",
+                               key->lct_owner);
+                        lu_ref_print(&key->lct_reference);
+                }
+        }
+}
+EXPORT_SYMBOL(lu_context_keys_dump);
+#else  /* !__KERNEL__ */
+static int lu_cache_shrink(int nr, unsigned int gfp_mask)
+{
+        return 0;
+}
+#endif /* __KERNEL__ */
+
+int  cl_global_init(void);
+void cl_global_fini(void);
+int  lu_ref_global_init(void);
+void lu_ref_global_fini(void);
+
+int dt_global_init(void);
+void dt_global_fini(void);
+
+int llo_global_init(void);
+void llo_global_fini(void);
+
+/**
  * Initialization of global lu_* data.
  */
 int lu_global_init(void)
 {
         int result;
 
+        CDEBUG(D_CONSOLE, "Lustre LU module (%p).\n", &lu_keys);
+
+        result = lu_ref_global_init();
+        if (result != 0)
+                return result;
+
         LU_CONTEXT_KEY_INIT(&lu_global_key);
         result = lu_context_key_register(&lu_global_key);
-        if (result == 0) {
-                /*
-                 * At this level, we don't know what tags are needed, so
-                 * allocate them conservatively. This should not be too bad,
-                 * because this environment is global.
-                 */
-                down(&lu_sites_guard);
-                result = lu_env_init_noref(&lu_shrink_env, NULL, LCT_SHRINKER);
-                up(&lu_sites_guard);
-                if (result == 0) {
-                        /*
-                         * seeks estimation: 3 seeks to read a record from oi,
-                         * one to read inode, one for ea. Unfortunately
-                         * setting this high value results in lu_object/inode
-                         * cache consuming all the memory.
-                         */
-                        lu_site_shrinker = set_shrinker(DEFAULT_SEEKS,
-                                                        lu_cache_shrink);
-                        if (result == 0)
-                                result = lu_time_global_init();
-                }
-        }
+        if (result != 0)
+                return result;
+        /*
+         * At this level, we don't know what tags are needed, so allocate them
+         * conservatively. This should not be too bad, because this
+         * environment is global.
+         */
+        cfs_down(&lu_sites_guard);
+        result = lu_env_init(&lu_shrink_env, LCT_SHRINKER);
+        cfs_up(&lu_sites_guard);
+        if (result != 0)
+                return result;
+
+        /*
+         * seeks estimation: 3 seeks to read a record from oi, one to read
+         * inode, one for ea. Unfortunately setting this high value results in
+         * lu_object/inode cache consuming all the memory.
+         */
+        lu_site_shrinker = cfs_set_shrinker(CFS_DEFAULT_SEEKS, lu_cache_shrink);
+        if (lu_site_shrinker == NULL)
+                return -ENOMEM;
+
+        result = lu_time_global_init();
+        if (result)
+                GOTO(out, result);
+
+#ifdef __KERNEL__
+        result = dt_global_init();
+        if (result)
+                GOTO(out, result);
+
+        result = llo_global_init();
+        if (result)
+                GOTO(out, result);
+#endif
+        result = cl_global_init();
+out:
+
         return result;
 }
 
-/*
+/**
  * Dual to lu_global_init().
  */
 void lu_global_fini(void)
 {
+        cl_global_fini();
+#ifdef __KERNEL__
+        llo_global_fini();
+        dt_global_fini();
+#endif
         lu_time_global_fini();
         if (lu_site_shrinker != NULL) {
-                remove_shrinker(lu_site_shrinker);
+                cfs_remove_shrinker(lu_site_shrinker);
                 lu_site_shrinker = NULL;
         }
 
@@ -1193,9 +1617,11 @@ void lu_global_fini(void)
          * Tear shrinker environment down _after_ de-registering
          * lu_global_key, because the latter has a value in the former.
          */
-        down(&lu_sites_guard);
+        cfs_down(&lu_sites_guard);
         lu_env_fini(&lu_shrink_env);
-        up(&lu_sites_guard);
+        cfs_up(&lu_sites_guard);
+
+        lu_ref_global_fini();
 }
 
 struct lu_buf LU_BUF_NULL = {
@@ -1204,84 +1630,79 @@ struct lu_buf LU_BUF_NULL = {
 };
 EXPORT_SYMBOL(LU_BUF_NULL);
 
-/*
- * XXX: Functions below logically belong to fid module, but they are used by
- * dt_store_open(). Put them here until better place is found.
+/**
+ * Output site statistical counters into a buffer. Suitable for
+ * lprocfs_rd_*()-style functions.
  */
-
-void fid_pack(struct lu_fid_pack *pack, const struct lu_fid *fid,
-              struct lu_fid *befider)
+int lu_site_stats_print(const struct lu_site *s, char *page, int count)
 {
-        int recsize;
-        __u64 seq;
-        __u32 oid;
-
-        seq = fid_seq(fid);
-        oid = fid_oid(fid);
-
-        /*
-         * Two cases: compact 6 bytes representation for a common case, and
-         * full 17 byte representation for "unusual" fid.
-         */
+        int i;
+        int populated;
 
         /*
-         * Check that usual case is really usual.
+         * How many hash buckets are not-empty? Don't bother with locks: it's
+         * an estimation anyway.
          */
-        CLASSERT(LUSTRE_SEQ_MAX_WIDTH < 0xffffull);
-
-        if (fid_is_igif(fid) ||
-            seq > 0xffffffull || oid > 0xffff || fid_ver(fid) != 0) {
-                fid_cpu_to_be(befider, fid);
-                recsize = sizeof *befider;
-        } else {
-                unsigned char *small_befider;
-
-                small_befider = (char *)befider;
-
-                small_befider[0] = seq >> 16;
-                small_befider[1] = seq >> 8;
-                small_befider[2] = seq;
-
-                small_befider[3] = oid >> 8;
-                small_befider[4] = oid;
-
-                recsize = 5;
-        }
-        memcpy(pack->fp_area, befider, recsize);
-        pack->fp_len = recsize + 1;
+        for (i = 0, populated = 0; i < s->ls_hash_size; i++)
+                populated += !cfs_hlist_empty(&s->ls_hash[i]);
+
+        return snprintf(page, count, "%d %d %d/%d %d %d %d %d %d %d %d\n",
+                        s->ls_total,
+                        s->ls_busy,
+                        populated,
+                        s->ls_hash_size,
+                        s->ls_stats.s_created,
+                        s->ls_stats.s_cache_hit,
+                        s->ls_stats.s_cache_miss,
+                        s->ls_stats.s_cache_check,
+                        s->ls_stats.s_cache_race,
+                        s->ls_stats.s_cache_death_race,
+                        s->ls_stats.s_lru_purged);
 }
-EXPORT_SYMBOL(fid_pack);
+EXPORT_SYMBOL(lu_site_stats_print);
+
+const char *lu_time_names[LU_TIME_NR] = {
+        [LU_TIME_FIND_LOOKUP] = "find_lookup",
+        [LU_TIME_FIND_ALLOC]  = "find_alloc",
+        [LU_TIME_FIND_INSERT] = "find_insert"
+};
+EXPORT_SYMBOL(lu_time_names);
 
-int fid_unpack(const struct lu_fid_pack *pack, struct lu_fid *fid)
+/**
+ * Helper function to initialize a number of kmem slab caches at once.
+ */
+int lu_kmem_init(struct lu_kmem_descr *caches)
 {
         int result;
 
-        result = 0;
-        switch (pack->fp_len) {
-        case sizeof *fid + 1:
-                memcpy(fid, pack->fp_area, sizeof *fid);
-                fid_be_to_cpu(fid, fid);
-                break;
-        case 6: {
-                const unsigned char *area;
-
-                area = pack->fp_area;
-                fid->f_seq = (area[0] << 16) | (area[1] << 8) | area[2];
-                fid->f_oid = (area[3] << 8) | area[4];
-                fid->f_ver = 0;
-                break;
-        }
-        default:
-                CERROR("Unexpected packed fid size: %d\n", pack->fp_len);
-                result = -EIO;
+        for (result = 0; caches->ckd_cache != NULL; ++caches) {
+                *caches->ckd_cache = cfs_mem_cache_create(caches->ckd_name,
+                                                          caches->ckd_size,
+                                                          0, 0);
+                if (*caches->ckd_cache == NULL) {
+                        result = -ENOMEM;
+                        break;
+                }
         }
         return result;
 }
-EXPORT_SYMBOL(fid_unpack);
+EXPORT_SYMBOL(lu_kmem_init);
 
-const char *lu_time_names[LU_TIME_NR] = {
-        [LU_TIME_FIND_LOOKUP] = "find_lookup",
-        [LU_TIME_FIND_ALLOC]  = "find_alloc",
-        [LU_TIME_FIND_INSERT] = "find_insert"
-};
-EXPORT_SYMBOL(lu_time_names);
+/**
+ * Helper function to finalize a number of kmem slab cached at once. Dual to
+ * lu_kmem_init().
+ */
+void lu_kmem_fini(struct lu_kmem_descr *caches)
+{
+        int rc;
+
+        for (; caches->ckd_cache != NULL; ++caches) {
+                if (*caches->ckd_cache != NULL) {
+                        rc = cfs_mem_cache_destroy(*caches->ckd_cache);
+                        LASSERTF(rc == 0, "couldn't destroy %s slab\n",
+                                 caches->ckd_name);
+                        *caches->ckd_cache = NULL;
+                }
+        }
+}
+EXPORT_SYMBOL(lu_kmem_fini);