* Kernel config requirements are built as
separate rules in kernel/configs. All kernel requirements
(including minlts and configs) are represented there.
* device framework compatibility matrix and final
framework compatibility matrix need special environment
variables, so they are not generic enough to be written
in the soong rule.
See compatibility_matrices/Android.bp for sample usages.
Test: builds and compare output before change
Change-Id: I53bf88362314f96487ca243c9292570b7931f24b
These two fields are still needed for some carriers. Should
not be removed in 1.3. Reverted back to what we have in 1.0.
Test: Telephony sanity tests
Bug: 73659459
Change-Id: I33e7b9b0cb26b56fc3c0e011557657136cb38a6c
Would the legacy HAL return -ENOENT while retrieving an effect,
the enumeration has to be aborted, and the effects queried so far
need to be returned. A missing 'break' was causing returning
no effects at all.
Bug: 117567478
Test: make
Change-Id: I7cc05004f88b39272d06d2ed33a0693f8e1e3d04
Merged-In: I7cc05004f88b39272d06d2ed33a0693f8e1e3d04
Until the underlying bug is fixed, do not produce a sanitized
binary.
Bug: 117506164
Bug: 117580200
Test: m && SANITIZE_TARGET=address m
Change-Id: I4e6d1206512513c687f7ae065d2974c55c617b61
Divide BBOX_TRANSFORM op into 2 ops
- AXIS_ALIGNED_BBOX_TRANSFORM
- ROTATED_BBOX_TRANSFORM
Rotated bounding boxes use different tensor format than axis-aligned
bounding boxes, and it would be less confusing if they were represented
by a separate operator code.
Bug: 113562630
Test: NeuralNetworksTest_static
Test: VtsHalNeuralnetworksV1_xTargetTest with sample driver
Change-Id: Ie08f2e0d0da77f6750766a394969653478d054d5
Merged-In: Ie08f2e0d0da77f6750766a394969653478d054d5
(cherry picked from commit d2bae1c268)
* changes:
Update OWNERS
Fix indexing bug in NNAPI
Attach a bug to syncing docs
Reserve enums for new NNAPI ops.
Add new op PAD_V2
Add new ops ARGMAX and ARGMIN
toString is broken in some cases because of recursive definitions. This
exercises that case.
Bug: 117342189
Test: hidl_test
Test: m android.hardware.tests.baz@1.0 android.hardware.tests.baz-V1.0-java
(this runs hidl-gen on this for both Java and C++ output)
Test: manually inspected output
Change-Id: Ic09ac133345f2d83d6f83d3d0af1f63e7e3250ee
1. Deprecated the fields 'mvnoType', 'mvnoMatchData', 'maxConnsTime',
and 'maxConns'.
2. Added a new flag 'preferred' indicating if this data profile
is preferred for default data connection setup.
3. Move modemCognative flag from setupDataCall and setInitialAttachApn
into the struct DataProfileInfo and rename it to 'persistent'.
4. Removed isRoaming flag in setupDataCall, setInitialAttachApn, and
setDataProfile.
Test: Telephony sanity tests
Bug: 73659459
Change-Id: Ia28715e85755b47a1ee870b5c90e5505a7fd8c4a
pointer needs a fixup (to abort) since they can't be used in remote
transactions, so removing it from the union here.
Bug: 117322722
Test: hidl_test
Change-Id: Ie582f6b79a8176ca1585bf4b1f43ae9d5d97fa92
The test was assuming Stream::getFrameCount() would return the
AudioConfig::frameCount used to create the stream, similarly
to the sampleRate or the channelMask.
Nevertheless this is not the case as the AudioConfig::frameCount
field is not to used by Stream::openOutputStream (nor Input) as
the buffer size is chosen by the HAL, not the framework.
Thus the test can not assume the frameCount chosen by the HAL and
must not test for a particular value.
Test: atest VtsHalAudioV4_0TargetTest
Bug: 116169974
Change-Id: Ibbf0825f540957e01e7c512d851cb9628991f4c4
Signed-off-by: Kevin Rocard <krocard@google.com>
registerCallback() and getHealthInfo() unintentionally broadcast
health info to all callbacks, which has a performance impact.
* registerCallback() still invokes the new callback immediately
* getHealthInfo() does not call any callbacks at all.
Test: VTS test call getHealthInfo does not invoke update()
Bug: 117167903
Change-Id: Ida99fdd73831e747fbf2d65089c7c0e7661fe7c4
