Add VTS tests that just exercise a single role, sink or source.
Use the AuthGraph core library to provide the implementation of the
other role. This means that the tests are best written in Rust.
Put the test code into a library, so that other HALs which include
AuthGraph as a component can exercise the AuthGraph parts in their own
VTS tests.
Test: VtsAidlAuthGraphRoleTest
Bug: 284470121
Change-Id: I73f6aa277b41cc728587d707d7a6f82f0d18e08f
Use the core library's new service implementation, which wraps a
channel to the TA.
In this nonsecure case, the TA is local in-process, so use the core
library's AuthGraphTa, and implement the SerializedChannel as just
a direct invocation of the TA.
Move this code into a _nonsecure library, so the main.rs just has
the code needed to start the executable and register the service.
Test: VtsAidlAuthGraphSessionTest
Bug: 284470121
Change-Id: I738d3876872a8cd248f0ebec708676d1173b6e37
getMessage is often empty. getDescription will always include at least
the status code.
Test: atest VtsHalRemotelyProvisionedComponentTargetTest
Change-Id: Id401184a3c3642188e26fa2ec03f19515f89ae4f
The authgraph_hal library should only depend on libauthgraph_wire, not
on libauthgraph_core, so adjust error processing accordingly.
Test: VtsAidlAuthGraphSessionTest
Bug: 284470121
Change-Id: I48056db6ceeab409d0f165f8e051809129643c6f
KeyMint should be able to handle challenges up to 128 bytes.
Document this (and the expected error code) in Tag.aidl.
Bug: 307714384
Test: m
Change-Id: Id099dd055c81b10d777effdf364395f84d8b35c6
Add an implementation for the AuthGraph key exchange HAL, to allow
testing and policy compliance.
In real use, IAuthGraphKeyExchange instances are expected to be
retrieved from some other HAL, allowing the AuthGraph instance to be
specifically correlated with the component that uses it.
Bug: 284470121
Bug: 291228560
Test: hal_implementation_test
Test: VtsAidlAuthGraphSessionTest
Change-Id: Ib064292d60bead663af7721fd1406f2a9b5d8ecd
Introduce the AIDL methods for key exchange and the CDDLs for
the relevant input/output types that are CBOR encoded.
Use a standard comment format in CDDL files to indicate where additional
structure definitions can be found, in the hope that one day tooling
will be available to automatically check schema syntax.
Add authgraph to compatibility matrix
This CL merges the CLs: aosp/2750484, aosp/2750485, aosp/2750486 by
drysdale@google.com
Co-authored-by: David Drysdale <drysdale@google.com>
Bug: 284470121
Test: validate schemas with https://cddl.anweiss.tech/
Test: VtsAidlAuthGraphSessionTest (with nonsecure default impl)
Change-Id: I2793666ede028f8abe91569587c09dcac21c67e0
Revert submission 2812435-revert-2778549-expose-avf-rkp-hal-GTFGLMUUKQ
Reason for revert: This change relands the topic
https://r.android.com/q/topic:%22expose-avf-rkp-hal%22
This SELinux denial has been fixed in system/sepolicy
Reverted changes: /q/submissionid:2812435-revert-2778549-expose-avf-rkp-hal-GTFGLMUUKQ
Change-Id: Ia1aed212bf8de4b81239997cc1d1ca665cc456eb
This cl adjusts IRPC VTS tests so that it can handle the IRPC AVF
impl in an environment where the IRPC operations are not supported.
Bug: 299257581
Test: atest VtsHalRemotelyProvisionedComponentTargetTest
Change-Id: Iae96076c20e360e6a4aca5518ca8df21920f6978
Since there were no VTS test to strictly check RSA_OAEP_MGF_DIGEST,
there are released devices with Keymint which do not include this tag in
key characteristics, hence these test fails on such Keymint and UDC
Android framework.
Hence version check is added before asserting MGF digest checks.
Bug: 297306437
Test: atest VtsAidlKeyMintTargetTest
Change-Id: I43054f8dbbd46de53deef5e6771c736e770280e0
Negative test cases should not expect UNKNOWN_ERROR from Keymint since
the exact cause of failure is known.
In general, we should avoid UNKNOWN_ERROR because it makes error
attribution difficult.
To avoid adding retroactive requirements KM implementation, relax the
check to expect any error.
Bug: 298194325
Test: VtsAidlKeyMintTargetTest
Change-Id: I136fb6d36ae92c9e3722ffefe9a067d3515dcbf9
std::unique consolidates the unique nonces at the start of the `nonces`
vector, but it doesn't modify nonces.size(), so these tests weren't
actually verifying that the nonces were unique. Add a vector::erase
call to shrink the vector.
After upgrading libc++, std::unique is [[nodiscard]] and this bug is
a compiler error.
Bug: 175635923
Test: treehugger
Test: m VtsAidlSharedSecretTargetTest VtsHalKeymasterV4_0TargetTest
Change-Id: I7fd8c40a3920bf3a8988c8065503c78ba36dc742
Check the VSR of the device to select the DICE validation rules that
will be appropriate to use for VTS.
Test: TH
Change-Id: Iff19debd1e442a0b318da1a4d8a08d470efba0ae
The original change to add this test didn't make it into the Android 13
version of the VTS test, so the version gate needs to be updated to be
v3+
Bug: 292318194
Test: VtsAidlKeyMintTargetTest --gtest_filter="*EcdsaMissingCurve*"
Change-Id: I94bf816688e57c7c04893a23cf0399129de94229
Allow for devices that claim to need external timestamps, but don't.
Test: VtsAidlKeyMintTargetTest
Bug: 300211206
Change-Id: Ie450d9969c337d5274502f3600e14c0b481e8b34
Extract CDDL from MacedPublicKey.aidl and DeviceInfo.aidl.
Make -70000 optional in `PublicKey`.
Make CDDL compliant with RFC8610:
- Move `bstr` contents for generated data (e.g. `ECDSA(..)`) to comments.
- Fix alternatives in CBOR-encoded bstr instances.
Add more RFC references and unify reference style.
Try to use a standard format to indicate that external files are needed
for the complete CDDL schema. (Hopefully in the future we might have
tooling that uses these to check the schemas.)
Test: Compile with docs.rs/cddl-cat and https://cddl.anweiss.tech/
Change-Id: Ie1d6c39f1cc0314e312df9205b570b89178ebd16
Earlier, attestation properties didn't match on GSI images, hence
EcdsaAttestationIdTags VTS test case was skipped on GSI images.
Recently attestation properties reading priority changed as
ro.product.*_for_attestation -> ro.product.vendor.* -> ro.product.*
that means on GSI images ro.product.vendor.* properties could be used
and hence attestation should work. Incase ro.product.vendor.* properties
are not same as provisioned values to KM. They should be set as
ro.product.*_for_attestation on base build.
Bug: 298586194
Test: atest VtsAidlKeyMintTargetTest:PerInstance/NewKeyGenerationTest#EcdsaAttestationIdTags/0_android_hardware_security_keymint_IKeyMintDevice_default
Change-Id: Ie945bd8f7060e0e768daf9681d121ea5f170a6e1
This solution was adopted from Cuttlefish's host side Keymint
implementation: I22bde00aed311c6774f83acc08a2c21e6e75141f.
Bug: 296983430
Test: Tested with Cuttlefish that the logs are present in logcat.
Change-Id: I942b0200bb164a2a865b255c6f26d628cbd345a4
Move CDDL out of comments and into files, per-method
generateCertificateRequest.cddl
generateCertificateRequestV2.cddl
This makes it easier to read and it means tools can parse it.
Test: Treehugger
Change-Id: I9b71b094d128a5a4566a4d352aaa11d1c9b595ff
On top of checking that the patch level are a UINT, also check that they
follow the YYYYMM or YYYYMMDD format in the CSR v3 as is required by the
server validation logic. This check is not applied in the factory as the
value might not yet be correctly provisioned.
Bug: 269813991
Test: atest VtsHalRemotelyProvisionedComponentTargetTest
Change-Id: I5c62ba176dae390ea0a387bba6cb975226e3873a
It turns out we had a bug (b/263844771) in how RKP support was
detected, and that was fixed. However, due to this bug, some S chipests
shipped without RKP support which is now required by the tests.
This change drops the RKP requirement from S chipsets. There should be
no new S chipsets, so this effectively grandfathers in the previous
ones that were skipped by the RKP VTS tests.
T+ tests (both VTS and other suites) will verify that RKP support is
there, so there is no gap introduced by this change.
Bug: 297139913
Test: VtsAidlKeyMintTargetTest
Change-Id: I387e5f058ada698747aac103c1745682291f2d1c
The test case for an auth-per-operation HAT with an invalid HMAC
is wrong -- it is re-using the previous HAT, which fails for a
different reason (has an old challenge).
Fix the test to use the HAT that's wrong in the intended way.
Bug: 297333975
Test: VtsAidlKeyMintTargetTest
Change-Id: I15fe9b0c1b53452df0f67dd44534fdb80a6c2a9c
Remove the inline definition of the Android Profile for DICE and instead
reference the definition that exists alongside the Open Profile for DICE
and is now the source of truth for the profile.
Test: n/a
Change-Id: Ia71a674234be13542ad0ce4db0b764e8ee0c7a62
Update TimeoutAuthenticationMultiSid test to support
generateKey for Strongbox implementations without
factory attestation.
Bug: 293211157
Test: run vts -m VtsAidlKeyMintTarget
Change-Id: I27bf08d2fd2d9e0217a90ee8ccb789adfd9d5f7f
The invalid value used for the second IMEI attestation test is
potentially wrong in two ways:
- It doesn't match the provisioned value.
- It's not a valid IMEI, not least because it is longer than 16 bytes.
Make the test value shorter so the second failure doesn't apply and
the test can reliably expect CANNOT_ATTEST_IDS.
Bug: 292959871
Test: VtsAidlKeyMintTargetTest
Change-Id: If8c6b9e08b48e6caf5c767578e1ac43964214619
continuing the execution of the test.
If generateKey fails and execution continues then it leads to issues
while verifying the attest records and causing the crash.
Test: atest VtsAidlKeyMintTargetTest
Bug: 292300030
Change-Id: I66bd650423e9e5bbbfe8411a1455c4ea5846f1ff
Specify that DICE-based RKP implementations may also allow a ROM
extension to manage the UDS public key.
Test: The words are semantically parseable
Change-Id: I8f9c6efb01fc76318220cf1bc4a0eb3a3ad42f87
Some of the SE Javacard OS implementations does not fulfill all of the
expectations as to when cipher text is to be generated. However, the
implementations are compliant with Javacard Specifications. This patch
skips the known VTS failures till U.
Bug: 290850651
Test: run vts -m VtsAidlKeyMintTarget
Change-Id: Id3746adc7385b41f4b4b2ebc8e7583972769134d
Removed the check to skip the attest-id tests on GSI, modified the
attest-id tests to support this.
Bug: 290643623
Test: atest VtsAidlKeyMintTargetTest
Change-Id: Id79d7fb4c70ed94ed76bc57f3d66ce47e9b67b48
When deliberately testing invalid ID attestation, use the helper
function (which checks the error return code is correct) in one more
place.
Test: VtsAidlKeyMintTargetTest
Bug: 286733800
Change-Id: I6ea5bd7ee19b3b172330117bfde1b16745debba7
Avoid the ADD_FAILURE at the end if attestion ID failure uses an allowed
return code.
Test: VtsAidlKeyMintTargetTest
Bug: 286733800
Change-Id: I0dcac312ac4516a078b2742721e3a19074da52b1
Updated VTS tests to verify mgf-digests in key characteristics of
RSA-OAEP keys. Added new tests to import RSA-OAEP keys with
mgf-digests and verified imported key characteristics.
Bug: 279721313
Test: atest VtsAidlKeyMintTargetTest
Change-Id: I06474a85c9e77fded264031ff5636f2c35bee6b4
Update the default KeyMint version to v3.
Note this affects the pure software implementation of KeyMint that is
not used for anything that tests currently run against.
Bug: 275982952
Test: m (that it builds)
Change-Id: I6ab10329af590bd2a045710dfff47c6e78740464
Generalize the existing helper function to allow more variants.
Remove a couple of pointless invocations of the existing helper.
Bug: 286733800
Test: VtsAidlKeyMintTargetTest
Change-Id: Ic01c53cbe79f55c2d403a66acbfd04029395c287
If some check in a VTS test case fails, the test function may exit early
and not call CheckedDeleteKey(&some_keyblob), thus "leaking" a key blob.
This isn't normally an issue, but if the key blob happens to use a
feature that uses some secure storage (e.g. ROLLBACK_RESISTANCE or
USAGE_COUNT_LIMIT=1) then this may leak some scarse resource.
To avoid the chance of this, use an RAII holder to ensure that
manually-managed keyblobs (i.e. key blobs that are not held in the
key_blob_ member of the base test class) are always deleted.
Bug: 262212842
Test: VtsAidlKeyMintTargetTest
Change-Id: Ie8806095e249870484b9875eb660070607f339a3
Update the RKP readme to match contemporary philosophy about the design.
This includes replacing discussion if the obsolete term `BCC` with a
description of the Android Profile for DICE.
The privacy concerns are relaxed to match updates to the HAL which
remove the superencryption of the DICE chain.
Test: n/a
Fix: 281755202
Change-Id: I3a6fd2cd12599c5843b5dce0044eb16c2afbffa2
It should definitely be the case that a different SPL triggers key
requires upgrade, but the converse isn't true -- if no SPL change, it's
OK for the device to request upgrade anyhow.
Bug: 281604435
Change-Id: Ic03ce51fb4b18ff669595ab430f9fccd1da48997
Align with the Open Profile for DICE by requiring that the configuration
hash be included because the configuration input is a hash of the
specified configuration desscriptor.
Test: n/a
Change-Id: I9d2ef560dc8e6f567b5b8d1a244f5138c45ae420
Introduce a field to the configuration descriptor that provides a
standard semantically-defined version number rather than the
vendor-defined component version which acts more like a build ID.
Test: n/a
Bug: 282205139
Change-Id: Idb0c991ab12ae75687236f2489e639e4422a0225
Only specify the requirements for `normal` DICE mode and allow vendors
to choose the non-normal mode that fits their need per the ope-dice
specification.
Add a note that RKP required `normal` mode in the DICE chain in order to
trust the device.
Test: n/a
Bug: 263144485
Change-Id: Iaaa3799c53234de61a51ebc855822b93ab3e5bb8
The Open Profile for DICE give possible guidelines on the requirements
for the DICE mode but Android needs those to be strictly specified.
Fix: 263144485
Test: n/a
Change-Id: Ia5fc937654504199cabf4709f1c15484242e0161
Added a check to make sure IMEI is not "null".
Bug: 281676499
Test: atest VtsAidlKeyMintTargetTest
Change-Id: Ia1569a30412d633eee4d4de8cd00dea077d1c23d
In the second case out of the two cases of authorization enforcement
described for update(), it seems like the challenge is expected in
the timestamp token.
Test: N/A
Change-Id: I33e1b84bf8218335665b31ca144b3b4ecb342328
Strongbox may not support 1024 bit key size for RSA.
So in NoUserConfirmation test updated the key size to
2048 so that the test works for both TEE and Strongbox.
Bug: 280117495
Test: run VtsAidlKeyMintTarget
Change-Id: I32bb28001aca9b69eedb1bd3d0bcff43052d06e4
Updated the BootLoaderStateTest for strongbox implementations which
do not support factory attestation.
Test: vts -m VtsAidlKeyMintTarget
Change-Id: I8fe176a18fc0b9e2b2d0b012b7b63124d15c9e2f
It's already documented that IRPC v3 doesn't make use of test mode keys
however VTS still required support for their generation. Fix this and
simplify implementation of the v3 HAL by expecting an error in all cases
that the deprecated test mode keys are seen.
IRPC v3 also fully deprecated the EEK meaning a v3 implementation must
unconditionally report CURVE_NONE for supportedEekCurve.
The VTS tests are enhanced with contextual version constants rather than
reusing constants with seemingly unrelated names.
Bug: 278013975
Test: atest VtsHalRemotelyProvisionedComponentTargetTest
Change-Id: I5709a0b1cd77eb28e677f64bb781fad58d91570a
A bug in the Trusty HAL service caused it to replace MGF1 digest tags
with Tag::INVALID. This tests that MGF1 tags are returned properly in
the MGF1 success test, and verifies that Tag::INVALID is never
returned by any test.
Bug: 278157584
Test: adb shell /data/nativetest/VtsAidlKeyMintTargetTest/VtsAidlKeyMintTargetTest
Change-Id: I5d391310795c99f37acf3c48310c127a7a31fac3
No tests are instantiated if KeyMint is present on the the device.
Explicitly allow that.
Bug: 277975776
Test: VtsAidlKeyMintTargetTest
Change-Id: I88f1c0a81f36d198dabcb1420b62a00bacdbb6e7
