The audio tests use a static cache of some HAL objects for
performance reasons (speed up by 10x on Pixel).
Those object are destroy during the test environment tear-down.
This tear-down was destroying the objects in a FIFO instead of LIFO
order. Thus the DeficesFactory was destroyed before the Device it
created.
Bug: 79889318
Test: vts-tradefed run commandAndExit vts --module VtsHalAudioV2_0Target
check that the device destructor is called before the
devicesFactory one.
Change-Id: I1b4345158139ba14a8779a9508f7ebdc41129d1d
Signed-off-by: Kevin Rocard <krocard@google.com>
Declaring the interfaces allows HIDL hwservicemanager to start
the HAL service in case it has crashed before audioserver
has called 'getService'.
Bug: 80227481
Test: use the repro case from the bug
Change-Id: Ibb46beb714b3985ab373d56d92da1105cfda28c6
VTS tests can only depend on NDK shared libraries,
they are pushed and run on older desert version.
In theory the VTS tests should be built against the NDK,
but this is not possible due to all the system dependencies.
Bug: 78871270
Test: compile
Change-Id: I48fbffd0316efc151917071c54aa19e533c615df
Signed-off-by: Kevin Rocard <krocard@google.com>
The TTYMode enum numeric value was converted to a char* instead of its
literal value and vice versa.
Instead convert it to/from its literal value.
Eg: TTYMode::FULL -> "FULL" instead of "3"
"FULL" -> 0 as atoi conversion would fail
Test: atest VtsHalAudioV4_0TargetTest
Bug: 71486871
Change-Id: I29bbf6bf3b5532269afcc5d39ea10eff2871bdea
Signed-off-by: Kevin Rocard <krocard@google.com>
Signed-off-by: Jungyee Yoo <jungyee.yoo@lge.corp-partner.google.com>
Previously, the setter of optional properties were not tested
if the getter was not supported.
This was an issue as the framework rarely gets, most
often only sets as it consider having the ownership of those properties.
Thus implementation tend to only support set.
As a result optional setter were not tested although called by the
framework.
For example the default impl setTtyMode has a bug that was not detected
because the Pixel HAL does not support getTtyMode.
This patch makes sure to always test getter and setter.
Test: atest VtsHalAudioV4_0TargetTest
Bug: 71486871
Change-Id: Ib0b03fb6eb77cc0428a33eb463166dd7c339bd3d
Signed-off-by: Kevin Rocard <krocard@google.com>
This factorization had not been ported from the AOSP patch.
The code is functionally identical.
Test: compile
Bug: 69010523
Change-Id: Ied3a657d7c219b580eb32377789096f6b2f6dc19
Signed-off-by: Kevin Rocard <krocard@google.com>
Test was calling getSupportedSamplingRate instead of channel mask.
Bug: 77307068
Test: atest VtsHalAudioV4_0TargetTest
Change-Id: Ib6e9d017793edfc95853cab0c2955b8c801bc66f
Signed-off-by: Kevin Rocard <krocard@google.com>
Thus the test was always incorrectly failing.
Also improve the test error messages.
Bug: 77307068
Test: VtsHalAudioV4_0TargetTest
Change-Id: I95db94ed99f7ca32af35422e36a95084e72279d8
Signed-off-by: Kevin Rocard <krocard@google.com>
Legacy implementation through getParameter can not return a status_t.
This is problematic for
- getSupportedChannelMasks(format)
- getSupportedSampleRate(format)
as they should be able to return NOT_SUPPORTED if they do not support
the provided format.
In that case, allow the legacy implementation to return an empty string
that will be converted to NOT_SUPPORTED.
Test: atest VtsHalAudioV4_0TargetTest
Bug: 77307068
Change-Id: I78c37caf059885e3d33e6a308876dbc0e3ef7145
Signed-off-by: Kevin Rocard <krocard@google.com>
This does not change the test logic or constraints.
It only improves the error messages.
Bug: 77307068
Test: atest VtsHalAudioV4_0TargetTest
Change-Id: I6c79ddd014d4ab8aba4f1d0b918888f83dfb63ad
Signed-off-by: Kevin Rocard <krocard@google.com>
The rest of the API (*::analyseStatus) returns NOT_SUPPORTED
when the legacy API returns -ENOSYS.
setParameter legacy -> treble shim did not follow this
conversion due to the legacy API stating that for get_paramers,
-ENOSYS should be returned if
"the implementation does not accept a parameter change while the
output is active but the parameter is acceptable otherwise",
aka INVALID_STATE.
Thus setParameter shim used to return
- OK for OK
- INVALID_STATE for -ENOSYS
- INVALID_ARGUMENTS for everything else
This leads to several problems:
- an implementation of the legacy API can not report NOT_SUPPORTED
- it is inconsistent with the rest of the status_t conversion methods
- shim methods implemented over getParameter can not distinguish
between different failures as required by the .hal documentation
Most importantly, on the system side, the Result is transformed to a
status_t again but without any special logic for methods wrapping
getParameter in the shim.
This can not be changed as the system can not know which methods
are implemented with a legacy wrapper under the Treble API boundary.
Thus setParam now converts status_t to Result in the same way
as all the other shim methods.
This patch is the second half of I41204c0807d2bd4675e941771cbc9a43d7d14855
that was reverted due to a merge conflict.
Bug: 72873273
Bug: 69811500
Bug: 69010523
Test: playback and record for media and voice call
Original-Change-Id: I41204c0807d2bd4675e941771cbc9a43d7d14855
Change-Id: I41328afce56ce31d4a26159ca2d4b16d14cce05b
Signed-off-by: Kevin Rocard <krocard@google.com>
This patch does not have any functional change.
It refactors Util.h to make it version dependant.
This is needed in the next patch to add a helper
that uses HIDL types.
This patch is the fixed first half of
I41204c0807d2bd4675e941771cbc9a43d7d14855
that was reverted due to a merge conflict.
Bug: 72873273
Bug: 69811500
Bug: 69010523
Test: playback and record for media and voice call
Original-Change-Id: I41204c0807d2bd4675e941771cbc9a43d7d14855
Change-Id: Ie543519fa0dca62486b75f42bb0c8a93d6cad3c3
Signed-off-by: Kevin Rocard <krocard@google.com>
A debugDump was used previously because the debug did not exist.
Now that the framework uses debug(), the HAL must also do so.
Bug: 78307161
Test: adb shell dumpsys media.audio_flinger
Change-Id: Ic5841d5284d742c572629249b80aa1de841983c7
Signed-off-by: Kevin Rocard <krocard@google.com>
The rest of the API (*::analyseStatus) returns NOT_SUPPORTED
when the legacy API returns -ENOSYS.
setParameter legacy -> treble shim did not follow this
conversion due to the legacy API stating that for get_paramers,
-ENOSYS should be returned if
"the implementation does not accept a parameter change while the
output is active but the parameter is acceptable otherwise",
aka INVALID_STATE.
Thus setParameter shim used to return
- OK for OK
- INVALID_STATE for -ENOSYS
- INVALID_ARGUMENTS for everything else
This leads to several problems:
- an implementation of the legacy API can not report NOT_SUPPORTED
- is inconsistent with the rest of the status_t conversion methods
- shim methods implemented over getParameter can not distinguish
failures error required by the .hal documentation
Most importantly, on the system side, the Result is transformed to a
status_t again but without any special logic for methods wrapping
getParameter in the shim.
This can not be changed as the system can not know which methods
are implemented with a legacy wrapper under the Treble API boundary.
Thus setParam now converts status_t to Result in the same way
as all the other shim methods.
Bug: 72873273
Bug: 69811500
Bug: 69010523
Test: playback and record for media and voice call
Change-Id: I41204c0807d2bd4675e941771cbc9a43d7d14855
Merged-In: I41204c0807d2bd4675e941771cbc9a43d7d14855
Signed-off-by: Kevin Rocard <krocard@google.com>
This was breaking Treble interface consistency test.
Bug: 77853088
Test: play music effect work
Change-Id: Ieb54f9c154e346286da353d0c3cdc770574c9bfe
Signed-off-by: Kevin Rocard <krocard@google.com>
- The ALL devices should not appear in the XSD as they are helpers for
enum value validity checks.
- Add the V4 values HEARING AIDS, ECHO_CANCELLER, BLUETOOTH_BLE.
- Sort the values in the same order as the enum values in
audio-base.h for easier comparison.
Bug: 77307068
Test: xmllint -xinclude --schema \
hardware/interfaces/audio/4.0/config/audio_policy_configuration.xsd \
out/target/product/walleye/vendor/etc/audio_policy_configuration.xml
Signed-off-by: Kevin Rocard <krocard@google.com>
Change-Id: I8b970d21eb0b3f256d6866be9ef935d12283076e
The metadata API was not present in the legacy HAL which lots of
developer still use.
Test: Check the method is correctly called on each track change
Bug: 38184704
Bug: 69623109
Change-Id: If56b4ee3739194e49808cf24d728c791b0588f9f
Signed-off-by: Kevin Rocard <krocard@google.com>
The XML format has not changed. Symlink the V4 XSD to the V2.
Bug: 76165942
Bug: 38184704
Test: vts-tradefed run commandAndExit vts --module VtsHalAudioEffectV4_0Target
Change-Id: I25a61c125c461c6c5b89d53142dcf887d6c94347
Signed-off-by: Kevin Rocard <krocard@google.com>
Add conversions between hal and framework.
Bug: 64038649
Bug: 75041465
Test: Manual Testing and Cts test
Change-Id: I88a292f0509e09467750affbe84350ed82cdd500
The xsd did not have its version in its name, thus the V4 and V2 were
overwriting each other.
Bug: 38184704
Test: vts-tradefed run commandAndExit vts --module VtsHalAudioV4_0Target
Change-Id: Id7e733729a9afcdcc582f88a5527b6eb1c7283bf
Signed-off-by: Kevin Rocard <krocard@google.com>
They were supposed to be added by the change with id
I2f488462614114e4c82149460b030585981ae15e
but actually were not.
Also, rename OUT_TOP_CENTER_... to OUT_TOP_SIDE_...
according to Dean's suggestion.
Bug: 38184704
Bug: 63901775
Test: make
Change-Id: Ib263b06e4f3aa84525c2255f7a52f9bb39dbebcd
