fscrypt: allow unprivileged users to add/remove keys for v2 policies

Allow the FS_IOC_ADD_ENCRYPTION_KEY and FS_IOC_REMOVE_ENCRYPTION_KEY
ioctls to be used by non-root users to add and remove encryption keys
from the filesystem-level crypto keyrings, subject to limitations.

Motivation: while privileged fscrypt key management is sufficient for
some users (e.g. Android and Chromium OS, where a privileged process
manages all keys), the old API by design also allows non-root users to
set up and use encrypted directories, and we don't want to regress on
that.  Especially, we don't want to force users to continue using the
old API, running into the visibility mismatch between files and keyrings
and being unable to "lock" encrypted directories.

Intuitively, the ioctls have to be privileged since they manipulate
filesystem-level state.  However, it's actually safe to make them
unprivileged if we very carefully enforce some specific limitations.

First, each key must be identified by a cryptographic hash so that a
user can't add the wrong key for another user's files.  For v2
encryption policies, we use the key_identifier for this.  v1 policies
don't have this, so managing keys for them remains privileged.

Second, each key a user adds is charged to their quota for the keyrings
service.  Thus, a user can't exhaust memory by adding a huge number of
keys.  By default each non-root user is allowed up to 200 keys; this can
be changed using the existing sysctl 'kernel.keys.maxkeys'.

Third, if multiple users add the same key, we keep track of those users
of the key (of which there remains a single copy), and won't really
remove the key, i.e. "lock" the encrypted files, until all those users
have removed it.  This prevents denial of service attacks that would be
possible under simpler schemes, such allowing the first user who added a
key to remove it -- since that could be a malicious user who has
compromised the key.  Of course, encryption keys should be kept secret,
but the idea is that using encryption should never be *less* secure than
not using encryption, even if your key was compromised.

We tolerate that a user will be unable to really remove a key, i.e.
unable to "lock" their encrypted files, if another user has added the
same key.  But in a sense, this is actually a good thing because it will
avoid providing a false notion of security where a key appears to have
been removed when actually it's still in memory, available to any
attacker who compromises the operating system kernel.

Reviewed-by: Theodore Ts'o <tytso@mit.edu>
Signed-off-by: Eric Biggers <ebiggers@google.com>

1 files changed, 27 insertions, 4 deletions

diff --git a/fs/crypto/fscrypt_private.h b/fs/crypto/fscrypt_private.h
index c89e37d38e42..d0e238234234 100644
--- a/fs/crypto/fscrypt_private.h
+++ b/fs/crypto/fscrypt_private.h
@@ -335,9 +335,16 @@ struct fscrypt_master_key {
 	 * FS_IOC_REMOVE_ENCRYPTION_KEY can be retried, or
 	 * FS_IOC_ADD_ENCRYPTION_KEY can add the secret again.
 	 *
-	 * Locking: protected by key->sem.
+	 * Locking: protected by key->sem (outer) and mk_secret_sem (inner).
+	 * The reason for two locks is that key->sem also protects modifying
+	 * mk_users, which ranks it above the semaphore for the keyring key
+	 * type, which is in turn above page faults (via keyring_read).  But
+	 * sometimes filesystems call fscrypt_get_encryption_info() from within
+	 * a transaction, which ranks it below page faults.  So we need a
+	 * separate lock which protects mk_secret but not also mk_users.
 	 */
 	struct fscrypt_master_key_secret	mk_secret;
+	struct rw_semaphore			mk_secret_sem;
 
 	/*
 	 * For v1 policy keys: an arbitrary key descriptor which was assigned by
@@ -348,6 +355,22 @@ struct fscrypt_master_key {
 	struct fscrypt_key_specifier		mk_spec;
 
 	/*
+	 * Keyring which contains a key of type 'key_type_fscrypt_user' for each
+	 * user who has added this key.  Normally each key will be added by just
+	 * one user, but it's possible that multiple users share a key, and in
+	 * that case we need to keep track of those users so that one user can't
+	 * remove the key before the others want it removed too.
+	 *
+	 * This is NULL for v1 policy keys; those can only be added by root.
+	 *
+	 * Locking: in addition to this keyrings own semaphore, this is
+	 * protected by the master key's key->sem, so we can do atomic
+	 * search+insert.  It can also be searched without taking any locks, but
+	 * in that case the returned key may have already been removed.
+	 */
+	struct key		*mk_users;
+
+	/*
 	 * Length of ->mk_decrypted_inodes, plus one if mk_secret is present.
 	 * Once this goes to 0, the master key is removed from ->s_master_keys.
 	 * The 'struct fscrypt_master_key' will continue to live as long as the
@@ -374,9 +397,9 @@ is_master_key_secret_present(const struct fscrypt_master_key_secret *secret)
 	/*
 	 * The READ_ONCE() is only necessary for fscrypt_drop_inode() and
 	 * fscrypt_key_describe().  These run in atomic context, so they can't
-	 * take key->sem and thus 'secret' can change concurrently which would
-	 * be a data race.  But they only need to know whether the secret *was*
-	 * present at the time of check, so READ_ONCE() suffices.
+	 * take ->mk_secret_sem and thus 'secret' can change concurrently which
+	 * would be a data race.  But they only need to know whether the secret
+	 * *was* present at the time of check, so READ_ONCE() suffices.
 	 */
 	return READ_ONCE(secret->size) != 0;
 }