Previously, the key pair generation would error out even if we
successfully provisioned attestation keys. Instead, we should retry
key generation after the GenerateRkpKeyService reports an OK status.
Bug: 231495834
Test: RemoteProvisionerUnitTests
Change-Id: I049294cbc7119de55b5de02499bf4609d4c6de5d
Implement support for the X25519 key agreement functionality.
Similar to Ed25519, two new classes are added:
* AndroidKeyStoreXDHPrivateKey
* AndroidKeyStoreXDHPublicKey
The private key class is simply a handle to the KeyMint key.
The public key class implements XECPublicKey, the interface
needed for using this key in a platform-backed key agreement.
Because of Conscrypt API boundaries, the functionality of Conscrypt's
OpenSSLX25519PublicKey is duplicated here - namely, matching the
prefix of the encoded key.
Bug: 194359292
Test: atest android.keystore.cts.Curve25519Test
Change-Id: Ifc12be528ab544fd6909bb0dd6224a0a4dd400c6
Wire Ed25519 signing into Keystore. This consists of registering a
provider for Ed25519.
Ed25519 has its own digest scheme, so the caller should specify "none"
as the digest scheme, and that's the tag that's going to be passed into
KeyMint.
However, unlike other uses of the "NONE" digest scheme, the input to the
signature algorithm should not be truncated.
Bug: 194359292
Test: atest android.keystore.cts.Curve25519Test
Change-Id: Icce4f7f2f8fa10081a9c6beff4813c2d91756469
Implement support for Ed25519 signing keys in Android Keystore.
Because Conscrypt does not yet handle those keys, the Keystore classes
implement EdECPublicKey directly and parse the keys.
Specifically, AndroidKeyStoreEdECPublicKey can take an encoded X.509 key
specification, validate the encoding is of an Ed25519 key, then parse
the oddity and Y point on the curve.
RFC8032 describes EdDSA signature scheme, particularly Ed25519.
RFC8410, Section 3, defines the OID for Ed25519 keys (1.3.101.112).
RFC8410, Section 4, describes the encoding of the public key.
Bug: 195309719
Bug: 194359292
Bug: 214203951
Test: atest android.security.keystore2.AndroidKeyStoreEdECPublicKeyTest
Change-Id: I07b793cbd5029630768368ad4a863bbc1c828ced
This change adds some integers to the AIDL interface in order to convey
status back to the caller of generateKey(). This will inform the caller
as to whether or not the errors that may occur during provisioning are
permanent, and if not, what to do with the transient error.
Bug: 227306369
Test: RemoteProvisionerUnitTests
Change-Id: I9202358a102b0fb0a104525632a005acb7355840
On systems that rely solely on remotely-provisioned keys (RKP),
the attestation keys may run out or be unavailable for attesting
a newly-generated key. This could happen when:
* the device first connects to the Internet
* The device had all the keys used and:
** It hadn't yet completed obtaining new ones.
** The RKP server declines to issue new keys.
In these cases, the caller must be informed that their key generation
request failed (likely temporarily), and that they should retry it.
The retry policy returned tells the caller when to re-try.
Bug: 227306369
Test: atest android.keystore.cts.KeyStoreExceptionTest
Merged-In: Ief30a3ab97da95b68d172e725c38acbefab92fa9
Change-Id: I0b2619fcbcb3ac4d94ed85f3ce5934e015c0828c
Ensure that the user gets an accurate error message when they try to
generate Curve 25519 keys according to JEP 324
(https://openjdk.java.net/jeps/324).
Android Keystore requires every key to have a name, so it is not
possible to generate a key using NamedParameterSpec only (with a
KeyPairGenerator).
Support this and throw an exception to the caller indicating how they
_can_ generate keys with this curve.
Bug: 222440855
Bug: 195309719
Bug: 194359292
Test: atest android.keystore.cts.KeyFactoryTest android.keystore.cts.Curve25519Test
Test: atest CtsLibcoreTestCases:libcore.java.security.ProviderTest
Change-Id: I5aa163f177507906c6482d079eb6cb55d93accf7
Do not register Curve 25519 algorithms as key factories, until we fix
the registration to be in compliance with JEP 324.
Bug: 222440855
Bug: 222194540
Test: atest android.keystore.cts.Curve25519Test android.keystore.cts.KeyFactoryTest#testAlgorithmList CtsLibcoreTestCases:libcore.java.security.ProviderTest#test_Provider_getServices
Merged-In: Ibd53070a890955affaff5e4e7213892afd423db7
Change-Id: I11b3574aeff54b3eb8bf496c4c14aa1338629ce5
Add support for Curve 25519 in the public API.
This requires upgrading the keymint dependency to V2.
Note that this CL only passes tha tags to Keystore,
but does not yet let the caller use the generated keys
because of missing Conscrypt classes.
Bug: 194359292
Test: atest android.keystore.cts.Curve25519Test
Change-Id: I15223abec34b72c857e26fcc47d8ecf08c1f8c8d
The default value for canUnlockViaBiometrics, which determines if we
are able to show a bio prompt, is true. However, if there are 0
biometric authenticator IDs, then it's impossible for the user to
satisfy a bio prompt. In this case, we should set canUnlockViaBiometrics
to false.
The loop that is normally expected to invert canUnlockViaBiometrics was
never run in the case of 0 bio authenticator ids, so we mistakenly
let the crypto init operation succeed when we should have blocked it.
Bug: 188864794
Test: Manual, using sample app that displays a biometric prompt.
Change-Id: Ib95b0564aa098157718b8d4a45b11baa69dad71b
This fixes the contract between equals and hashCode in
AndroidKeystorePublicKey. The previous fix made only a reference
comparisson between certificate blobs. In this patch java.util.Arrays is
used to compare and compute the hash of the array.
Bug: 196118021
Test: See following CL.
Change-Id: I2b8b7e740fb377de39fd21f763e15cb00024b2fc
Fix a NullPointerException when trying to insert SecretKey that already
exists.
Bug: 202146009
Test: atest android.keystore.cts.AndroidKeyStoreTest#testKeyStore_SetKeyEntry_ReplacedWithSameGeneratedSecretKey
Change-Id: If3a4bd6677ab3173c5c1a7c921ba567b7981662b
The code was doing a reference compare, not object value comparison,
resulting in failures in the KeyStore setEntry API.
Test: CtsKeystoreTestCases:android.keystore.cts.AndroidKeyStoreTest
Fixes: 197138784
Change-Id: I2c5e47283eed5694951869e9ea3853364ddef9d1
This fixes the contract between equals and hashCode in
AndroidKeystoreKey and AndroidKeystorePublicKey.
Bug: 196118021
Test: N/A
Change-Id: I3f7e6d72d53c7051c13daeb5aa6ce1abf4eb0cc5
The KeyMint spec requires the specification of the EC_CURVE tag when
generating an EC key. This patch adds the correct curve tag parameter to
the parameter list.
Test: CtsVerifier Protected confirmation test.
Bug: 192908276
Merged-In: I2e7dd4868abda85d244e73592ff12d688f5c21fc
Change-Id: I2e7dd4868abda85d244e73592ff12d688f5c21fc
This file was written on the assumption that bindService was
synchronous, which it isn't. This change adds a CountDownLatch to force
the class to wait for the binding to finish. If the relevant key
generation service is not present on the system, then this
functionality will just silently be skipped over.
Bug: 190222116
Test: atest RemoteProvisionerUnitTests
Change-Id: Ie34997a08aa743642c66a20c4b756cd47bff4af1
Merged-In: Ie34997a08aa743642c66a20c4b756cd47bff4af1
This file was written on the assumption that bindService was
synchronous, which it isn't. This change adds a CountDownLatch to force
the class to wait for the binding to finish.
Bug: 190222116
Test: atest RemoteProvisionerUnitTests
Change-Id: I917a61da612f21f9a0f783bea5d24270d4e1db42
Merged-In: I917a61da612f21f9a0f783bea5d24270d4e1db42
getUniqueAliases may return a null if an error occurred. This would lead
to a NPE in engineAliases.
This patch makes getUniqueAliases return an empty HashSet instead.
Test: atest KeystoreTests
Change-Id: I387d90ea851a8b9c18bb2b20d1a0bfc1ab76c99f
AndroidKeyStoreCipherSpiBase.engineDoFinal may get called with a null
input argument. In the case where we forward the operation to the
default provider doFinal() needs to be called instead of
doFinal(byte[], int, int).
Bug: 183913233
Test: atest android.keystore.cts.CipherTest#testEncryptsAndDecryptsUsingCipherStreams
Change-Id: Ia3afaf281be7c8e5493ac8e4155a7aa02d1d37f0
As a part of internal libcore API cleanup some of the functions
previously exposed are getting removed from public surface.
Math#randomLongInternal is a wrapper around java.util.Random and has no
specific implications so its usages are get refactored.
Bug: 154796679
Test: m droid
Change-Id: I29e0e9307fbaf9c1ac018b83014efb2d3dd74479
This change adds an AIDL interface which the RemoteProvisioner app
implements that allows the keystore SPI to inform the app when an
attestation key may have been used, and when the underlying attestation
key pool is totally empty. The former is a non-blocking call, and the
latter blocks until completion.
Since the latter involves network, there are timeouts involved on the
app side to ensure that the blocking call doesn't hang indefinitely if
there's no network.
Test: atest CtsKeystoreTestCases && atest RemoteProvisionerUnitTests
Change-Id: Ie49e37659c96ce5c1626d1b99a4a7ccc62028156
The legacy provider is obsolete now that all calling code has
been fixed.
Bug: 183100147
Bug: 183093711
Bug: 171305684
Test: N/A
Merged-In: I0d71d3c9cdd586a508827eb26120c872cb8643ea
Change-Id: I0d71d3c9cdd586a508827eb26120c872cb8643ea
KeyChain supports device id attestation through KeyGenParameterSpec now.
No need to call attest key individually. Also calling attest key
individually is no longer supported by Keystore 2.0 and KeyMint.
Also isBoundKeyAlgorithm returns true.
Test: atest FrameworksServicesTests:DevicePolicyManagerTest
Bug: 171305387
Merged-In: I759fe245b48fe435153fded2c74c9ae99634c146
Change-Id: I759fe245b48fe435153fded2c74c9ae99634c146
AndroidKeyStoreKeyPairGeneratorSpi used the Uid as namespace which is
wrong, and ParcelableKeyGenParameterSpec inadvertently used the Uid as
namespace specifier during conversion.
Bug: 160623310
Test: com.android.keychain.tests.BasicKeyChainServiceTest#testGenerateKeyPairErrorsOnBadUid
Change-Id: I84b4c69c639e42922449e00a3708cef89b82f63e
This allows apps to request that AndroidKeyStore generate attestation
keys that can be used to sign attestations of other keys that the app
generates or imports.
Bug: 163606833
Test: atest CtsKeystoreTests
Change-Id: I943a6922271cbe909cb3a9d67021663b5646aa70
Now that attestation and generation of keys occurs in the same step, the
device ID attestation parameters need to be passed into the
KeyPairGeneratorSpi. This change shifts functionality that was
previously in AttestationUtils into KeyGenParameterSpec and the
keystore2 KeyPairGeneratorSpi. The API changes should be gated to
Platform APIs and hidden from less privileged components.
Test: atest cts/tests/tests/keystore/src/android/keystore/cts/KeyGenParameterSpecTest.java
Bug: 177369988
Change-Id: Iafbc1661583bdf61da644b2c0838b9024018ee82
* Add missing purpose to HMAC operations.
* Pass correct key descriptor to wrapped key import.
* Corrected error message on Import wrapped key failure due to missing
wrapping key.
* Do not bail out on missing attestation challenge when device
properties are requested.
* Fix AndroidKeyStoreKey.equals(). It is sufficient to compare the
unique key ID.
Test: Keystore CTS tests.
Change-Id: Id3b93b18486e3a818ae8e91be6344294ec592e68
Bug: b/174140443
Test: atest CtsKeystoreTestCases
The new CTS tests for this feature is introduced in aosp/1556464
Change-Id: I9620c4a3e5d2c10ed8a50d494e63eb2fb19dabef
Merged-In: I9620c4a3e5d2c10ed8a50d494e63eb2fb19dabef
