This commit was manufactured by cvs2svn to create branch 'vserver'.
[linux-2.6.git] / security / keys / keyring.c
diff --git a/security/keys/keyring.c b/security/keys/keyring.c
new file mode 100644 (file)
index 0000000..98d4b0b
--- /dev/null
@@ -0,0 +1,895 @@
+/* keyring.c: keyring handling
+ *
+ * Copyright (C) 2004 Red Hat, Inc. All Rights Reserved.
+ * Written by David Howells (dhowells@redhat.com)
+ *
+ * This program is free software; you can redistribute it and/or
+ * modify it under the terms of the GNU General Public License
+ * as published by the Free Software Foundation; either version
+ * 2 of the License, or (at your option) any later version.
+ */
+
+#include <linux/module.h>
+#include <linux/init.h>
+#include <linux/sched.h>
+#include <linux/slab.h>
+#include <linux/seq_file.h>
+#include <linux/err.h>
+#include <asm/uaccess.h>
+#include "internal.h"
+
+/*
+ * when plumbing the depths of the key tree, this sets a hard limit set on how
+ * deep we're willing to go
+ */
+#define KEYRING_SEARCH_MAX_DEPTH 6
+
+/*
+ * we keep all named keyrings in a hash to speed looking them up
+ */
+#define KEYRING_NAME_HASH_SIZE (1 << 5)
+
+static struct list_head        keyring_name_hash[KEYRING_NAME_HASH_SIZE];
+static rwlock_t                keyring_name_lock = RW_LOCK_UNLOCKED;
+
+static inline unsigned keyring_hash(const char *desc)
+{
+       unsigned bucket = 0;
+
+       for (; *desc; desc++)
+               bucket += (unsigned char) *desc;
+
+       return bucket & (KEYRING_NAME_HASH_SIZE - 1);
+}
+
+/*
+ * the keyring type definition
+ */
+static int keyring_instantiate(struct key *keyring,
+                              const void *data, size_t datalen);
+static int keyring_duplicate(struct key *keyring, const struct key *source);
+static int keyring_match(const struct key *keyring, const void *criterion);
+static void keyring_destroy(struct key *keyring);
+static void keyring_describe(const struct key *keyring, struct seq_file *m);
+static long keyring_read(const struct key *keyring,
+                        char __user *buffer, size_t buflen);
+
+struct key_type key_type_keyring = {
+       .name           = "keyring",
+       .def_datalen    = sizeof(struct keyring_list),
+       .instantiate    = keyring_instantiate,
+       .duplicate      = keyring_duplicate,
+       .match          = keyring_match,
+       .destroy        = keyring_destroy,
+       .describe       = keyring_describe,
+       .read           = keyring_read,
+};
+
+/*
+ * semaphore to serialise link/link calls to prevent two link calls in parallel
+ * introducing a cycle
+ */
+DECLARE_RWSEM(keyring_serialise_link_sem);
+
+/*****************************************************************************/
+/*
+ * publish the name of a keyring so that it can be found by name (if it has
+ * one)
+ */
+void keyring_publish_name(struct key *keyring)
+{
+       int bucket;
+
+       if (keyring->description) {
+               bucket = keyring_hash(keyring->description);
+
+               write_lock(&keyring_name_lock);
+
+               if (!keyring_name_hash[bucket].next)
+                       INIT_LIST_HEAD(&keyring_name_hash[bucket]);
+
+               list_add_tail(&keyring->type_data.link,
+                             &keyring_name_hash[bucket]);
+
+               write_unlock(&keyring_name_lock);
+       }
+
+} /* end keyring_publish_name() */
+
+/*****************************************************************************/
+/*
+ * initialise a keyring
+ * - we object if we were given any data
+ */
+static int keyring_instantiate(struct key *keyring,
+                              const void *data, size_t datalen)
+{
+       int ret;
+
+       ret = -EINVAL;
+       if (datalen == 0) {
+               /* make the keyring available by name if it has one */
+               keyring_publish_name(keyring);
+               ret = 0;
+       }
+
+       return ret;
+
+} /* end keyring_instantiate() */
+
+/*****************************************************************************/
+/*
+ * duplicate the list of subscribed keys from a source keyring into this one
+ */
+static int keyring_duplicate(struct key *keyring, const struct key *source)
+{
+       struct keyring_list *sklist, *klist;
+       unsigned max;
+       size_t size;
+       int loop, ret;
+
+       const unsigned limit =
+               (PAGE_SIZE - sizeof(*klist)) / sizeof(struct key);
+
+       ret = 0;
+       sklist = source->payload.subscriptions;
+
+       if (sklist && sklist->nkeys > 0) {
+               max = sklist->nkeys;
+               BUG_ON(max > limit);
+
+               max = (max + 3) & ~3;
+               if (max > limit)
+                       max = limit;
+
+               ret = -ENOMEM;
+               size = sizeof(*klist) + sizeof(struct key) * max;
+               klist = kmalloc(size, GFP_KERNEL);
+               if (!klist)
+                       goto error;
+
+               klist->maxkeys = max;
+               klist->nkeys = sklist->nkeys;
+               memcpy(klist->keys,
+                      sklist->keys,
+                      sklist->nkeys * sizeof(struct key));
+
+               for (loop = klist->nkeys - 1; loop >= 0; loop--)
+                       atomic_inc(&klist->keys[loop]->usage);
+
+               keyring->payload.subscriptions = klist;
+               ret = 0;
+       }
+
+ error:
+       return ret;
+
+} /* end keyring_duplicate() */
+
+/*****************************************************************************/
+/*
+ * match keyrings on their name
+ */
+static int keyring_match(const struct key *keyring, const void *description)
+{
+       return keyring->description &&
+               strcmp(keyring->description, description) == 0;
+
+} /* end keyring_match() */
+
+/*****************************************************************************/
+/*
+ * dispose of the data dangling from the corpse of a keyring
+ */
+static void keyring_destroy(struct key *keyring)
+{
+       struct keyring_list *klist;
+       int loop;
+
+       if (keyring->description) {
+               write_lock(&keyring_name_lock);
+               list_del(&keyring->type_data.link);
+               write_unlock(&keyring_name_lock);
+       }
+
+       klist = keyring->payload.subscriptions;
+       if (klist) {
+               for (loop = klist->nkeys - 1; loop >= 0; loop--)
+                       key_put(klist->keys[loop]);
+               kfree(klist);
+       }
+
+} /* end keyring_destroy() */
+
+/*****************************************************************************/
+/*
+ * describe the keyring
+ */
+static void keyring_describe(const struct key *keyring, struct seq_file *m)
+{
+       struct keyring_list *klist;
+
+       if (keyring->description) {
+               seq_puts(m, keyring->description);
+       }
+       else {
+               seq_puts(m, "[anon]");
+       }
+
+       klist = keyring->payload.subscriptions;
+       if (klist)
+               seq_printf(m, ": %u/%u", klist->nkeys, klist->maxkeys);
+       else
+               seq_puts(m, ": empty");
+
+} /* end keyring_describe() */
+
+/*****************************************************************************/
+/*
+ * read a list of key IDs from the keyring's contents
+ */
+static long keyring_read(const struct key *keyring,
+                        char __user *buffer, size_t buflen)
+{
+       struct keyring_list *klist;
+       struct key *key;
+       size_t qty, tmp;
+       int loop, ret;
+
+       ret = 0;
+       klist = keyring->payload.subscriptions;
+
+       if (klist) {
+               /* calculate how much data we could return */
+               qty = klist->nkeys * sizeof(key_serial_t);
+
+               if (buffer && buflen > 0) {
+                       if (buflen > qty)
+                               buflen = qty;
+
+                       /* copy the IDs of the subscribed keys into the
+                        * buffer */
+                       ret = -EFAULT;
+
+                       for (loop = 0; loop < klist->nkeys; loop++) {
+                               key = klist->keys[loop];
+
+                               tmp = sizeof(key_serial_t);
+                               if (tmp > buflen)
+                                       tmp = buflen;
+
+                               if (copy_to_user(buffer,
+                                                &key->serial,
+                                                tmp) != 0)
+                                       goto error;
+
+                               buflen -= tmp;
+                               if (buflen == 0)
+                                       break;
+                               buffer += tmp;
+                       }
+               }
+
+               ret = qty;
+       }
+
+ error:
+       return ret;
+
+} /* end keyring_read() */
+
+/*****************************************************************************/
+/*
+ * allocate a keyring and link into the destination keyring
+ */
+struct key *keyring_alloc(const char *description, uid_t uid, gid_t gid,
+                         int not_in_quota, struct key *dest)
+{
+       struct key *keyring;
+       int ret;
+
+       keyring = key_alloc(&key_type_keyring, description,
+                           uid, gid, KEY_USR_ALL, not_in_quota);
+
+       if (!IS_ERR(keyring)) {
+               ret = key_instantiate_and_link(keyring, NULL, 0, dest);
+               if (ret < 0) {
+                       key_put(keyring);
+                       keyring = ERR_PTR(ret);
+               }
+       }
+
+       return keyring;
+
+} /* end keyring_alloc() */
+
+/*****************************************************************************/
+/*
+ * search the supplied keyring tree for a key that matches the criterion
+ * - perform a breadth-then-depth search up to the prescribed limit
+ * - we only find keys on which we have search permission
+ * - we use the supplied match function to see if the description (or other
+ *   feature of interest) matches
+ * - we readlock the keyrings as we search down the tree
+ * - we return -EAGAIN if we didn't find any matching key
+ * - we return -ENOKEY if we only found negative matching keys
+ */
+struct key *keyring_search_aux(struct key *keyring,
+                              struct key_type *type,
+                              const void *description,
+                              key_match_func_t match)
+{
+       struct {
+               struct key *keyring;
+               int kix;
+       } stack[KEYRING_SEARCH_MAX_DEPTH];
+
+       struct keyring_list *keylist;
+       struct timespec now;
+       struct key *key;
+       long err;
+       int sp, psp, kix;
+
+       key_check(keyring);
+
+       /* top keyring must have search permission to begin the search */
+       key = ERR_PTR(-EACCES);
+       if (!key_permission(keyring, KEY_SEARCH))
+               goto error;
+
+       key = ERR_PTR(-ENOTDIR);
+       if (keyring->type != &key_type_keyring)
+               goto error;
+
+       now = current_kernel_time();
+       err = -EAGAIN;
+       sp = 0;
+
+       /* start processing a new keyring */
+ descend:
+       read_lock(&keyring->lock);
+       if (keyring->flags & KEY_FLAG_REVOKED)
+               goto not_this_keyring;
+
+       keylist = keyring->payload.subscriptions;
+       if (!keylist)
+               goto not_this_keyring;
+
+       /* iterate through the keys in this keyring first */
+       for (kix = 0; kix < keylist->nkeys; kix++) {
+               key = keylist->keys[kix];
+
+               /* ignore keys not of this type */
+               if (key->type != type)
+                       continue;
+
+               /* skip revoked keys and expired keys */
+               if (key->flags & KEY_FLAG_REVOKED)
+                       continue;
+
+               if (key->expiry && now.tv_sec >= key->expiry)
+                       continue;
+
+               /* keys that don't match */
+               if (!match(key, description))
+                       continue;
+
+               /* key must have search permissions */
+               if (!key_permission(key, KEY_SEARCH))
+                       continue;
+
+               /* we set a different error code if we find a negative key */
+               if (key->flags & KEY_FLAG_NEGATIVE) {
+                       err = -ENOKEY;
+                       continue;
+               }
+
+               goto found;
+       }
+
+       /* search through the keyrings nested in this one */
+       kix = 0;
+ ascend:
+       while (kix < keylist->nkeys) {
+               key = keylist->keys[kix];
+               if (key->type != &key_type_keyring)
+                       goto next;
+
+               /* recursively search nested keyrings
+                * - only search keyrings for which we have search permission
+                */
+               if (sp >= KEYRING_SEARCH_MAX_DEPTH)
+                       goto next;
+
+               if (!key_permission(key, KEY_SEARCH))
+                       goto next;
+
+               /* evade loops in the keyring tree */
+               for (psp = 0; psp < sp; psp++)
+                       if (stack[psp].keyring == keyring)
+                               goto next;
+
+               /* stack the current position */
+               stack[sp].keyring = keyring;
+               stack[sp].kix = kix;
+               sp++;
+
+               /* begin again with the new keyring */
+               keyring = key;
+               goto descend;
+
+       next:
+               kix++;
+       }
+
+       /* the keyring we're looking at was disqualified or didn't contain a
+        * matching key */
+ not_this_keyring:
+       read_unlock(&keyring->lock);
+
+       if (sp > 0) {
+               /* resume the processing of a keyring higher up in the tree */
+               sp--;
+               keyring = stack[sp].keyring;
+               keylist = keyring->payload.subscriptions;
+               kix = stack[sp].kix + 1;
+               goto ascend;
+       }
+
+       key = ERR_PTR(err);
+       goto error;
+
+       /* we found a viable match */
+ found:
+       atomic_inc(&key->usage);
+       read_unlock(&keyring->lock);
+
+       /* unwind the keyring stack */
+       while (sp > 0) {
+               sp--;
+               read_unlock(&stack[sp].keyring->lock);
+       }
+
+       key_check(key);
+ error:
+       return key;
+
+} /* end keyring_search_aux() */
+
+/*****************************************************************************/
+/*
+ * search the supplied keyring tree for a key that matches the criterion
+ * - perform a breadth-then-depth search up to the prescribed limit
+ * - we only find keys on which we have search permission
+ * - we readlock the keyrings as we search down the tree
+ * - we return -EAGAIN if we didn't find any matching key
+ * - we return -ENOKEY if we only found negative matching keys
+ */
+struct key *keyring_search(struct key *keyring,
+                          struct key_type *type,
+                          const char *description)
+{
+       return keyring_search_aux(keyring, type, description, type->match);
+
+} /* end keyring_search() */
+
+EXPORT_SYMBOL(keyring_search);
+
+/*****************************************************************************/
+/*
+ * search the given keyring only (no recursion)
+ * - keyring must be locked by caller
+ */
+struct key *__keyring_search_one(struct key *keyring,
+                                const struct key_type *ktype,
+                                const char *description,
+                                key_perm_t perm)
+{
+       struct keyring_list *klist;
+       struct key *key;
+       int loop;
+
+       klist = keyring->payload.subscriptions;
+       if (klist) {
+               for (loop = 0; loop < klist->nkeys; loop++) {
+                       key = klist->keys[loop];
+
+                       if (key->type == ktype &&
+                           key->type->match(key, description) &&
+                           key_permission(key, perm) &&
+                           !(key->flags & KEY_FLAG_REVOKED)
+                           )
+                               goto found;
+               }
+       }
+
+       key = ERR_PTR(-ENOKEY);
+       goto error;
+
+ found:
+       atomic_inc(&key->usage);
+ error:
+       return key;
+
+} /* end __keyring_search_one() */
+
+/*****************************************************************************/
+/*
+ * find a keyring with the specified name
+ * - all named keyrings are searched
+ * - only find keyrings with search permission for the process
+ * - only find keyrings with a serial number greater than the one specified
+ */
+struct key *find_keyring_by_name(const char *name, key_serial_t bound)
+{
+       struct key *keyring;
+       int bucket;
+
+       keyring = ERR_PTR(-EINVAL);
+       if (!name)
+               goto error;
+
+       bucket = keyring_hash(name);
+
+       read_lock(&keyring_name_lock);
+
+       if (keyring_name_hash[bucket].next) {
+               /* search this hash bucket for a keyring with a matching name
+                * that's readable and that hasn't been revoked */
+               list_for_each_entry(keyring,
+                                   &keyring_name_hash[bucket],
+                                   type_data.link
+                                   ) {
+                       if (keyring->flags & KEY_FLAG_REVOKED)
+                               continue;
+
+                       if (strcmp(keyring->description, name) != 0)
+                               continue;
+
+                       if (!key_permission(keyring, KEY_SEARCH))
+                               continue;
+
+                       /* found a potential candidate, but we still need to
+                        * check the serial number */
+                       if (keyring->serial <= bound)
+                               continue;
+
+                       /* we've got a match */
+                       atomic_inc(&keyring->usage);
+                       read_unlock(&keyring_name_lock);
+                       goto error;
+               }
+       }
+
+       read_unlock(&keyring_name_lock);
+       keyring = ERR_PTR(-ENOKEY);
+
+ error:
+       return keyring;
+
+} /* end find_keyring_by_name() */
+
+/*****************************************************************************/
+/*
+ * see if a cycle will will be created by inserting acyclic tree B in acyclic
+ * tree A at the topmost level (ie: as a direct child of A)
+ * - since we are adding B to A at the top level, checking for cycles should
+ *   just be a matter of seeing if node A is somewhere in tree B
+ */
+static int keyring_detect_cycle(struct key *A, struct key *B)
+{
+       struct {
+               struct key *subtree;
+               int kix;
+       } stack[KEYRING_SEARCH_MAX_DEPTH];
+
+       struct keyring_list *keylist;
+       struct key *subtree, *key;
+       int sp, kix, ret;
+
+       ret = -EDEADLK;
+       if (A == B)
+               goto error;
+
+       subtree = B;
+       sp = 0;
+
+       /* start processing a new keyring */
+ descend:
+       read_lock(&subtree->lock);
+       if (subtree->flags & KEY_FLAG_REVOKED)
+               goto not_this_keyring;
+
+       keylist = subtree->payload.subscriptions;
+       if (!keylist)
+               goto not_this_keyring;
+       kix = 0;
+
+ ascend:
+       /* iterate through the remaining keys in this keyring */
+       for (; kix < keylist->nkeys; kix++) {
+               key = keylist->keys[kix];
+
+               if (key == A)
+                       goto cycle_detected;
+
+               /* recursively check nested keyrings */
+               if (key->type == &key_type_keyring) {
+                       if (sp >= KEYRING_SEARCH_MAX_DEPTH)
+                               goto too_deep;
+
+                       /* stack the current position */
+                       stack[sp].subtree = subtree;
+                       stack[sp].kix = kix;
+                       sp++;
+
+                       /* begin again with the new keyring */
+                       subtree = key;
+                       goto descend;
+               }
+       }
+
+       /* the keyring we're looking at was disqualified or didn't contain a
+        * matching key */
+ not_this_keyring:
+       read_unlock(&subtree->lock);
+
+       if (sp > 0) {
+               /* resume the checking of a keyring higher up in the tree */
+               sp--;
+               subtree = stack[sp].subtree;
+               keylist = subtree->payload.subscriptions;
+               kix = stack[sp].kix + 1;
+               goto ascend;
+       }
+
+       ret = 0; /* no cycles detected */
+
+ error:
+       return ret;
+
+ too_deep:
+       ret = -ELOOP;
+       goto error_unwind;
+ cycle_detected:
+       ret = -EDEADLK;
+ error_unwind:
+       read_unlock(&subtree->lock);
+
+       /* unwind the keyring stack */
+       while (sp > 0) {
+               sp--;
+               read_unlock(&stack[sp].subtree->lock);
+       }
+
+       goto error;
+
+} /* end keyring_detect_cycle() */
+
+/*****************************************************************************/
+/*
+ * link a key into to a keyring
+ * - must be called with the keyring's semaphore held
+ */
+int __key_link(struct key *keyring, struct key *key)
+{
+       struct keyring_list *klist, *nklist;
+       unsigned max;
+       size_t size;
+       int ret;
+
+       ret = -EKEYREVOKED;
+       if (keyring->flags & KEY_FLAG_REVOKED)
+               goto error;
+
+       ret = -ENOTDIR;
+       if (keyring->type != &key_type_keyring)
+               goto error;
+
+       /* serialise link/link calls to prevent parallel calls causing a
+        * cycle when applied to two keyring in opposite orders */
+       down_write(&keyring_serialise_link_sem);
+
+       /* check that we aren't going to create a cycle adding one keyring to
+        * another */
+       if (key->type == &key_type_keyring) {
+               ret = keyring_detect_cycle(keyring, key);
+               if (ret < 0)
+                       goto error2;
+       }
+
+       /* check that we aren't going to overrun the user's quota */
+       ret = key_payload_reserve(keyring,
+                                 keyring->datalen + KEYQUOTA_LINK_BYTES);
+       if (ret < 0)
+               goto error2;
+
+       klist = keyring->payload.subscriptions;
+
+       if (klist && klist->nkeys < klist->maxkeys) {
+               /* there's sufficient slack space to add directly */
+               atomic_inc(&key->usage);
+
+               write_lock(&keyring->lock);
+               klist->keys[klist->nkeys++] = key;
+               write_unlock(&keyring->lock);
+
+               ret = 0;
+       }
+       else {
+               /* grow the key list */
+               max = 4;
+               if (klist)
+                       max += klist->maxkeys;
+
+               ret = -ENFILE;
+               size = sizeof(*klist) + sizeof(*key) * max;
+               if (size > PAGE_SIZE)
+                       goto error3;
+
+               ret = -ENOMEM;
+               nklist = kmalloc(size, GFP_KERNEL);
+               if (!nklist)
+                       goto error3;
+               nklist->maxkeys = max;
+               nklist->nkeys = 0;
+
+               if (klist) {
+                       nklist->nkeys = klist->nkeys;
+                       memcpy(nklist->keys,
+                              klist->keys,
+                              sizeof(struct key *) * klist->nkeys);
+               }
+
+               /* add the key into the new space */
+               atomic_inc(&key->usage);
+
+               write_lock(&keyring->lock);
+               keyring->payload.subscriptions = nklist;
+               nklist->keys[nklist->nkeys++] = key;
+               write_unlock(&keyring->lock);
+
+               /* dispose of the old keyring list */
+               kfree(klist);
+
+               ret = 0;
+       }
+
+ error2:
+       up_write(&keyring_serialise_link_sem);
+ error:
+       return ret;
+
+ error3:
+       /* undo the quota changes */
+       key_payload_reserve(keyring,
+                           keyring->datalen - KEYQUOTA_LINK_BYTES);
+       goto error2;
+
+} /* end __key_link() */
+
+/*****************************************************************************/
+/*
+ * link a key to a keyring
+ */
+int key_link(struct key *keyring, struct key *key)
+{
+       int ret;
+
+       key_check(keyring);
+       key_check(key);
+
+       down_write(&keyring->sem);
+       ret = __key_link(keyring, key);
+       up_write(&keyring->sem);
+
+       return ret;
+
+} /* end key_link() */
+
+EXPORT_SYMBOL(key_link);
+
+/*****************************************************************************/
+/*
+ * unlink the first link to a key from a keyring
+ */
+int key_unlink(struct key *keyring, struct key *key)
+{
+       struct keyring_list *klist;
+       int loop, ret;
+
+       key_check(keyring);
+       key_check(key);
+
+       ret = -ENOTDIR;
+       if (keyring->type != &key_type_keyring)
+               goto error;
+
+       down_write(&keyring->sem);
+
+       klist = keyring->payload.subscriptions;
+       if (klist) {
+               /* search the keyring for the key */
+               for (loop = 0; loop < klist->nkeys; loop++)
+                       if (klist->keys[loop] == key)
+                               goto key_is_present;
+       }
+
+       up_write(&keyring->sem);
+       ret = -ENOENT;
+       goto error;
+
+ key_is_present:
+       /* adjust the user's quota */
+       key_payload_reserve(keyring,
+                           keyring->datalen - KEYQUOTA_LINK_BYTES);
+
+       /* shuffle down the key pointers
+        * - it might be worth shrinking the allocated memory, but that runs
+        *   the risk of ENOMEM as we would have to copy
+        */
+       write_lock(&keyring->lock);
+
+       klist->nkeys--;
+       if (loop < klist->nkeys)
+               memcpy(&klist->keys[loop],
+                      &klist->keys[loop + 1],
+                      (klist->nkeys - loop) * sizeof(struct key *));
+
+       write_unlock(&keyring->lock);
+
+       up_write(&keyring->sem);
+       key_put(key);
+       ret = 0;
+
+ error:
+       return ret;
+
+} /* end key_unlink() */
+
+EXPORT_SYMBOL(key_unlink);
+
+/*****************************************************************************/
+/*
+ * clear the specified process keyring
+ * - implements keyctl(KEYCTL_CLEAR)
+ */
+int keyring_clear(struct key *keyring)
+{
+       struct keyring_list *klist;
+       int loop, ret;
+
+       ret = -ENOTDIR;
+       if (keyring->type == &key_type_keyring) {
+               /* detach the pointer block with the locks held */
+               down_write(&keyring->sem);
+
+               klist = keyring->payload.subscriptions;
+               if (klist) {
+                       /* adjust the quota */
+                       key_payload_reserve(keyring,
+                                           sizeof(struct keyring_list));
+
+                       write_lock(&keyring->lock);
+                       keyring->payload.subscriptions = NULL;
+                       write_unlock(&keyring->lock);
+               }
+
+               up_write(&keyring->sem);
+
+               /* free the keys after the locks have been dropped */
+               if (klist) {
+                       for (loop = klist->nkeys - 1; loop >= 0; loop--)
+                               key_put(klist->keys[loop]);
+
+                       kfree(klist);
+               }
+
+               ret = 0;
+       }
+
+       return ret;
+
+} /* end keyring_clear() */
+
+EXPORT_SYMBOL(keyring_clear);