Every time we create a credential, contact the Provisioner app and tell
it that a key was generated. This may not strictly be true, but the
provisioner has heuristics to ensure that it only contacts the backend
if necessary. So, at most, we're spinning a few extra cycles whenever
a new credential is created (which is a rare occurence) to ensure that
we have RKP keys available for future requests.
Test: CtsIdentityTestCases
Fixes: 224771551
Change-Id: I6dd20635e6933842a95242e6d0cbfb9bf8c8f734
This new PresentationSession interface enables an application to do a
multi-document presentation, something which isn't possible with the
existing API. As a practical example of this consider presenting both
your Mobile Driving License and your Vaccination Certificate in a single
transaction.
Also update the documentation for IdentityCredential to clarify that
the same AuthKey is used for multiple getEntries() calls on the same
credential.
Also deprecate existing IdentityCredential.getEntries() method and
related methods and classes.
Bug: 197965513
Test: New CTS tests and new screen in CtsVerifier
Change-Id: I74534969143882552407917a82f44d43da12711c
* changes:
Move framework java filegroups into subdirectories
Partial cp of "Move Tuner resource updating from Tuner java into Tuner client"
Partial cp of "Initial boilerplate for an updatable graphics jar"
All the java code used to build the framework jar and run metalava
was previously defined in the toplevel Android.bp files. Move these
into the subdirs where the source actually lives.
This simplifies the rules themselves (no path and needless prefix) and
declutters the top level Android.bp.
Test: m
Change-Id: I97086e309eacb879d16facb8497d9940fa5ddaf6
- Add PackageManager system features (with versions) for the normal
and direct access store
- Deprecate IdentityCredentialStore.deleteCredentialByName() and add
IdentityCredential.delete() as a replacement.
- Add IdentityCredential.proveOwnership()
- Add IdentityCredential.update()
- Add docs for ProofOfBinding CBOR in X.509 extension of certificate
for AuthenticationKey
- Add IdentityCredential.setAllowUsingExpiredKeys()
- Add version of IdentityCredential.storeStaticAuthenticationData()
which takes a an expiration date. Deprecate the old variant of
this method.
Bug: 170146643
Test: atest android.security.identity.cts
Change-Id: I39a0ed65ed6efaa424ada7a9495e3b1da67cf452
Implement Enrollment-Specific ID, which is calculated using fixed device
identifiers, as well as the provisioning package and the Organization
Identifier set by the Device Policy Controller.
Test: atest FrameworksServicesTests:EnterpriseSpecificIdCalculatorTest
Test: atest com.android.cts.devicepolicy.MixedDeviceOwnerTest#testEnrollmentSpecificIdCorrectCalculation com.android.cts.devicepolicy.MixedManagedProfileOwnerTest#testEnrollmentSpecificIdCorrectCalculation com.android.cts.devicepolicy.MixedDeviceOwnerTest#testEnrollmentSpecificIdEmptyAndMultipleSet com.android.cts.devicepolicy.MixedManagedProfileOwnerTest#testEnrollmentSpecificIdEmptyAndMultipleSet
Bug: 168627890
Change-Id: I8b24efa6b8c82d6181f2b20bc8880ddeb6caa4c5
Key derivation for session encryption and MACing now involves mixing
in SessionTranscriptBytes. Update docs to reflect this.
Also, the standard changed such that instead of DeviceAuthentication
being MACed or signed, it's instead DeviceAuthenticationBytes which is
defined as #6.24(bstr .cbor DeviceAuthentication). The same also for
ReaderAuthentication, now ReaderAuthenticationBytes is the CBOR which
is signed by the reader.
Also make a note that the encryptMessageToReader() and
decryptMessageFromReader() should NOT be used and applications should
instead implement these themselves. This is because we don't have the
SessionTranscript available and it's way too late to start adding
public API now. For the next Android version these methods will be
deprecated. Realistically this shouldn't be a problem because
applications are expected to use the Jetpack anyway.
Bug: 159482543
Test: atest android.security.identity.cts
Change-Id: I380a973a0cc78f1206fd7a33d0bd4896a0b16c6d
This change contains no actual syntactical or semantic changes, just
clarifications on the inputs and outputs.
Test: N/A
Bug: 151082886
Change-Id: Ic7797aa53d292abdeb779cb55b404f8a433bce79
The DIS version of 18013-5 now specifically says
The first encryption with a key shall use a counter value of 1. For each
following encryption the counter value shall be increased by 1.
in section '9.2.1.4 Mechanism". The previous version said
The counter value is an unsigned integer, which starts at 0 for both
the mDL and the mDL Reader. For each encryption the counter value shall
be increased by 1.
which for some strange reason was interpreted by someone to mean that
counters should start at 1.
Update our implementation to use 1 as now called for by the standard.
Bug: 111446262
Test: atest android.security.identity.cts
Change-Id: I09d1216713d57b54036e4f9aa6677dfa5713133c
Having this method return null is the expected and documented behavior
when either the IC HAL or credstore isn't available.
Test: atest android.security.identity.cts (with credstore not running)
Bug: 148495024
Change-Id: Ifa17c58a84057499b1aeb8404959d5c0badfe52a
The Identity Credential APIs provides an interface to a secure store
for user identity documents. These APIs are deliberately fairly
general and abstract. To the extent possible, specification of the
message formats and semantics of communication with credential
verification devices and Issuing Authorities (IAs) is out of scope for
these APIs.
The Identity Credential APIs rely on user authentication to protect
data elements in credentials which is implemented through
auth-tokens. This CL contains changes to CryptoObject to allow this.
Bug: 111446262
Test: CtsIdentityTestCases
Change-Id: I48f21a561b762d86c9ca8d229962782572412f47
