We should not be relying on the HAL service to add CREATION_TIME to
keys. It was always intended to be an optional tag that could be
added by keystore, or maybe the caller of keystore. One widespread
Keymaster implementation started adding it (somewhat erroneously) if
it wasn't provided, and it appears that this implementation's behavior
became assumed to be the required behavior.
Test: VtsHalKeymasterV4_0TargetTest
Change-Id: I34267c4e1f59fd8ee5f898f8c746a7b49f4d74a5
Without the setting of TAG_MAC_LENGTH, the test fails due to
MISSING_MAC_PURPOSE.
Bug: 145626599
Test: VtsHalKeymasterV4_0TargetTest
Change-Id: Ic58411b86e07dfeeb78211abf9271ee995beabc9
This is updating the public exponent being checked from 3 to 65537. The
test was updated to use the latter public exponent without the check
also being updated
Test: atest VtsHalKeymasterV4_0TargetTest
Change-Id: I55fa2e67c94f0c52a1d65644f16b5f703a661e00
This test should will flag builds running as eng or userdebug that
report back the device is locked during development. This will also
catch the case where the device is a user build but reporting that it
isn't locked. This should help to avoid instances in the future where
userdebug builds report a locked device in the VBMeta information.
This patch also does a little bit of cleanup of the surrounding VBMeta
checking code.
Test: atest VtsHalKeymasterV4_0TargetTest
Change-Id: I3b387ade5eeee6a68b9ff307e503417d264ecbfe
The Keymaster 4.0 HAL specifies that the public exponent for RSA is F4
(2^16+1). There were a few tests still using 3 as the exponent. This
patch updates those incorrect exponents accordingly.
Bug: 143404829
Test: atest VtsHalKeymasterV4_0TargetTest
Change-Id: Ibc82a8a912bc5926bcdd544e0370e4185a888c0d
This tests passing a large input to finish. This should either succeed
or fail with the right error code.
Test: Run new VTS test
Change-Id: Ic4ef90adc6274317796bbe752f95fc9efa5fdb07
This test will check that the length of the attestation application id
field will be properly encoded in valid DER ASN.1 in cases where the
length is long enough to require extra bytes to encode. In those cases,
the encoding of that field should include:
-A byte to specify how many bytes are required to enumerate the length
-The bytes required to enumerate the length
-The actual data that follows
Bug: 142674020
Test: atest keymaster_hidl_hal_test
Change-Id: I6d162efa4c8c6e0922989e234d0377caf3c1758e
Per Keymaster 4.0 spec, TEE and StrongBox implementations are only
required to support HMAC keys between 64 and 512 bits in length.
StrongBox implementations additionally must not support anything larger
than 512 bits. The tests removed in this CL specified key sizes larger
than 512 bits.
Bug: 143404829
Test: m VtsHalKeymasterV4_0TargetTest && adb sync data && \
adb shell data/nativetest64/VtsHalKeymasterV4_0TargetTest/VtsHalKeymasterV4_0TargetTest
Change-Id: I96ee3a20b981c288d88366f536b9924f907268f3
This test checks that length metadata for the ASN.1 encoding of
attestation application ids are correct. It generates an app id that
will have a length between 127 and 256, which should create an encoding
that requires two bytes of length metadata - one byte to specify how many
bytes are needed for the length, and one byte for the length.
Some implementations of keymaster only use one byte in this case, which
will fail on strict ASN.1 parsers.
Bug: 142674020
Test: m VtsHalKeymasterV4_0TargetTest && adb sync data \
&& adb shell data/nativetest64/VtsHalKeymasterV4_0TargetTest/VtsHalKeymasterV4_0TargetTest
Change-Id: I7dfc38a09247eb3cb237f33a202044668d15cbca
NullOr now stores the references a pointers internally. This fixes UB
where the internal reference was initalized by dereferencing nullptr.
Test: Compiles
Bug: 121390225
Change-Id: I2073e5aeac401309aa63b08e05db3c467fab6b69
1. AES operation attempted with unauthorized purpose.
2. AES-GCM encryption performed with different nonces, should
generate different ciphertexts.
3. AES-GCM encryption decryption round trip with delays between
begin and update and finish.
Bug: 133258003
Test: VtsHalKeymasterV4_0TargetTest
Change-Id: Ia8b4b4b317ecff51b18e64dfa3b84bf77475812d
Replace libcrypto with libcrypto_static, which can be protected through
visibility to ensure only modules that don't affect FIPS certification
can use it.
Bug: 141248879
Test: m checkbuild
Change-Id: I8685cb06d15f3425eeb96d998ffda54c82dcd387
BUG: b/139689895
TEST: Added VTS tests to keymaster_hidl_hal_test.cpp
TEST: Ran on emulator against soft keymaster::v4_0::ng
Change-Id: I6c682cafee65cf7ea426bd03865bf868586efc62
Since these were combined into libhidlbase.
Bug: 135686713
Test: build only (libhwbinder/libhidltransport are empty)
Change-Id: I075670b64eebbbbd6a6ae0e84ad51bf1c6f5ba36
Due to changes in implementation between keymaster 3.0 and 4.0, rollback
resistance is now specified by the caller. This patch addresses that
inconsistency to make sure rollback resistance is properly tested. If
rollback resistance is supported by the hardware, then it will now be
tested.
Test: atest VtsHalKeymasterV4_0TargetTest
Change-Id: I21e8d1e66932ddfad2d42ce8a43591431f3ff284
GSI images do not have AVB verification enabled and therefore lack
several properties the keymaster HAL test depended on. Selectively
disable those parts of the test that would fail with AVB verification
disabled. Also disable date format checks under GSI. When invoked from
GSI the TEE-backed keymaster doesn't use the correct date format.
Bug: 130843899
Test: VtsHalKeymasterV4_0TargetTest
Exempt-From-Owner-Approval: change only affects VTS-on-GSI behavior
Change-Id: Idaafb7b515c41290c766a8132f35d498ca15f48a
The TEE keymaster has been seen to be almost a minute out of sync with
the host clock during attestation. Increase the leniency window to two
minutes.
Bug: 134408892
Bug: 134408367
Test: VtsHalKeymasterV4_0TargetTest
Change-Id: Ic256a939dcd7e7b108099cfcf237cacde8dde059
Object derived from RefBase must be owned by sp rather then other smart
pointer implementations.
Bug: 79474587
Change-Id: I866f67e1cb091efb3026450d50a410b5985539b6
This does two things:
- makes sure that HALs configured as lazy HALs will be retrieved
- will detect bad manifest entries earlier
Bug: 131703193
Test: boot
Change-Id: I82e10f49367b097023eb31797c877c15eedb5e00
In Keymaster 3, both INVALID_INPUT_LENGTH and INVALID_ARGUMENT were
acceptable for oversized messages. Keymaster 4 VTS requires that
INVALID_ARGUMENT be returned, but the spec has no such restriction. This
loosens VTS to allow either INVALID_INPUT_LENGTH or INVALID_ARGUMENT in
this case.
Bug: 129297054
Test: atest VtsHalKeymasterV4_0TargetTest Pixel 3, Trusty tests
The spec requires that SHA1 not be allowed for wrapped keys and that
only SHA_2_256 be used. Unfortunately, the previous VTS required SHA1
support. This patch takes the middle ground by requiring SHA_2_256 be
supported for importWrappedKey, but not disallowing it from supporting
SHA1.
This makes it possible for a spec compliant keymaster to pass VTS
while not disqualifying shipped devices.
Bug: 129291873
Test: atest VtsHalKeymasterV4_0TargetTest:ImportWrappedKeyTest, Trusty
Change-Id: I6c3a9182b51f2e7a46173d5bfc34d3c3264d954f
This test verifies that verification tokens with different time stamps do
not have the same MAC. This may not guarantee that the MAC is computed
correctly but it catches implementation that do not include the time
stamp in the mac.
It also checks that the MAC changes when both time stamp and challenge
changes.
Test: yes it is
Bug: 131859731
Bug: 132288466
Bug: 132287277
Change-Id: I85aa1d873eff46df7a66fc69bd61a031e6e6fbe0
The BOOT_PATCHLEVEL value is allowed to have 00 in the days position
according to the keymaster specification. This test's comment already
suggests that it's allowed, so update the expectation to match.
Test: VtsHalKeymasterV4_0TargetTest
Bug: 130843899
Change-Id: Ib43da43b2e0398b48fb59710bf4066f2641de2eb
The verified boot hash in the attestation record is a binary blob, while
the property read from the system is a hex-encoded value. Convert the
boot hash from the attestation record into hex before comparing.
Test: VtsHalKeymasterV4_0TargetTest
Bug: 130843899
Change-Id: I6f6e0da71501d741dd8b27d0778e1854af17ace6
hidl-generated makefiles are now generated such that bpfmt(file) == file.
Bug: 67417008
Test: enable bpfmt hook
Change-Id: I1f69d292bc23a7cc293a66110cb02d597e1019ad
Keymaster VTS test coverage on 4.0 was incomplete. This significantly
expands the coverage of the spec. The bugs listed are errors found that
these tests will cover, but are not indicative of the complete set of
things tested.
Test: atest VtsHalKeymasterV4_0TargetTest
Bug: 79953279
Bug: 119553313
Bug: 119541233
Bug: 119396995
Bug: 119542230
Bug: 119549128
Bug: 119549677
Bug: 122184852
Bug: 122261372
Change-Id: I42d78091b48398597bbebe1d9c91b806494ddf4c
(cherry picked from commit 8c0edf6c84)
Keymaster VTS test coverage on 4.0 was incomplete. This significantly
expands the coverage of the spec. The bugs listed are errors found that
these tests will cover, but are not indicative of the complete set of
things tested.
Test: atest VtsHalKeymasterV4_0TargetTest
Bug: 79953279
Bug: 119553313
Bug: 119541233
Bug: 119396995
Bug: 119542230
Bug: 119549128
Bug: 119549677
Bug: 122184852
Bug: 122261372
Change-Id: I42d78091b48398597bbebe1d9c91b806494ddf4c
Test importing of an Elliptic Curve P-256 key, encoded using the RFC5915
specification (which requires the curve OID in key in addition to the
wrapper) and the same key encoded using SEC1 (which allows omitting the
OID if it's known from the wrapper).
Test: atest VtsHalKeymasterV4_0TargetTest ImportKeyTest
Bug: 124437839
Bug: 127799174
Bug: 129398850
Change-Id: I5f5df86e55a758ed739403d830baa5c7308813a3
Merged-In: I5f5df86e55a758ed739403d830baa5c7308813a3
Test importing of an Elliptic Curve P-256 key, encoded using the RFC5915
specification (which requires the curve OID in key in addition to the
wrapper) and the same key encoded using SEC1 (which allows omitting the
OID if it's known from the wrapper).
Test: atest VtsHalKeymasterV4_0TargetTest ImportKeyTest
Bug: 124437839
Bug: 127799174
Change-Id: I5f5df86e55a758ed739403d830baa5c7308813a3
operator< on hidl_vec<uint8_t> violates strict weak ordering in the case
that one oparand is shorter that the other and the shorter is a prefix
of the longer.
if x and y are incomparable, i.e., neither x < y nor y < x and
y and z are incomparable, i.e., neither y < z nor z < y, then
x and z must be incomparable.
As for the current implementation the first two statements are true but
the third is not given the following example input:
x:="aa", y:="a", z:="ab".
This patch fixes the issue by defining a < b if a is a prefix of b.
As this relation is used in a std::sort algorithm which demands strict
weak ordering this bug leads to undefined behavior.
Change-Id: I4961bb35e2fd4f5fcf561ec0c7c536f81830aab8
