Break out the checks to see if a focus loss is handled by the
framework in a separate method, and reorder the if statements
for code clarity.
Annotate PlayerFocusEnforcer methods for nullability, add
javadoc
Test: atest AudioFocusTest
Change-Id: I9aaab804c78a2e2627ddbd7b9bd41fdec47d2b57
Merged-In: I9aaab804c78a2e2627ddbd7b9bd41fdec47d2b57
* changes:
AudioService: fix disconnect/connect of A2DP device
audioservice: re send BT_SCO parameter to HAL when audioserver restarts
audioservice: fix hasMediaDynamicPolicy() for loopback and render policies
Due to the expected BECOMING_NOISY behavior associated
with a device disconnection, the disconnection is handled
asynchronously after a fixed delay. This delay caused an
inversion of commands in the processing order of the
disconnection of a device closely followed by connection
of the same device.
The fix consists in:
- overriding the equals() operator for BtDeviceConnectionInfo
so messages for a given device in the message queue
can be checked / removed.
- when AudioDeviceBroker receives a command for A2DP
connection or disconnection, remove all upcoming connection
and disconnection commands in the queue for this device
(see postBluetoothA2dpDeviceConnectionStateSuppressNoisyIntent)
- remove AudioDeviceBroker.handleSetA2dpSinkConnectionState, which
was only used in BtHelper.onA2dpProfileConnected() with
a CONNECTED state, and have this method perform a regular device
connection (just like when coming from AM->AS).
- in AudioDeviceInventory.onSetA2dpSinkConnectionState(), support
receiving a connection event for an already connected device,
to support codec changes.
This change also includes modifications to the classes involved
in the device connection to make them support mocking/spying
to reproduce the bug conditions (see AudioDeviceBrokerTest).
Bug: 134932649
Test: atest AudioDeviceBrokerTest
Change-Id: If2b3b41409c77467a181a2f9b42310db9b9de8c5
Merged-In: If2b3b41409c77467a181a2f9b42310db9b9de8c5
Fix omission in AudioDeviceBroker.onAudioServerDied() where parameter
"BT_SCO=on|off" must be send to HAL according to current forced usage
for communication.
Bug: 135512789
Test: killaudio server during call over BT SCO
Change-Id: I1039748db5591321869f70c3dc589fbc62d41c31
Merged-In: I1039748db5591321869f70c3dc589fbc62d41c31
Exclude mixes with LOOPBACK + RENDER flags when looking for dynamic
policies matching media usage. As this is only used in the context
of sending becoming noisy intent we want to send the intent when playback
capture is active.
Bug: 137055231
Test: enable live caption and disconnect headset while music is playing
Change-Id: Ib3cd38f58c2ff78a2f2f13c5c22b637f9701e345
Merged-In: Ib3cd38f58c2ff78a2f2f13c5c22b637f9701e345
Use a dedicated thread for all SoundPool and MediaPlayer
operations. This ensures that the client (AudioService) doesn't
get blocked or delayed. This also removes the need to use
locking as all access to data happens on the same thread.
Simplify the implementation of the mapping between UI effect IDs
and SoundPool sample IDs.
Add event logging for sound effects loading.
Bug: 135763139
Test: normal scenario
media volume responsiveness with stuck SoundPool
behavior when declared sound assets are missing
Change-Id: Ie5a1684a2cabe05ff53270ad97cab161a87a59f9
Merged-In: Ie5a1684a2cabe05ff53270ad97cab161a87a59f9
Move the code related to sound effects loading and playback
into a separate class SoundEffectsHelper.
Bug: 135763139
Test: make and run phone
Change-Id: I30a7d897d740b02943c68924074f14fae3c3eff8
Merged-In: I30a7d897d740b02943c68924074f14fae3c3eff8
sendKeyEvent API always forwards the key control to the TV.
But for volume keys, we want it to be forwarded to the audio system if
system audio mode is on and if we are not a audio system itself.
sendVolumeKeyEvent can handle this logic.
Test: manual
Bug: 129116245
Change-Id: I6b81072ab49c092fac65cd542c19e611fc939cea
Merged-In: I6b81072ab49c092fac65cd542c19e611fc939cea
(cherry picked from commit ada40f59b63fe613448278a7cfbb7055146f8ccd)
The libcore.timezone APIs may form the basis for some
new SystemApis. Before starting that process the API
surface is being rationalized to establish the core
use cases.
Test: Treehugger
Bug: 139091367
Change-Id: I6c8b791524d16223c414612913b22f4d70c27d61
Merged-In: I6c8b791524d16223c414612913b22f4d70c27d61
(cherry picked from commit 400efa36ef)
This reverts commit cfe38cdb1c.
The CL 8f712189dfc02285573337e2b4ab17678011db14 in libcore project avoids
the need for adding the new paths in LoadedApk. With the CL, a classloader
does not give up even when loader.findLibrary() has failed due to some
paths are not added to it. Instead, the classloader converts the given
libname into a filename (e.g. foo -> libfoo.so) then calls dlopen()
with the filename. The classloader reports failure only when the dlopen()
call has failed. Therefore, manually adding these paths to the classloader
is no longer needed and thus removed.
Bug: 109720125
Test: System.loadLibrary("adsprpc") is successful in Pixel (because
libadsprpc.so is in Pixel's vendor public lib list)
Test: atest cts/tests/tests/jni
Merged-In: I1970eba7e732728699042a36b89571915ec81451
(cherry picked from commit 37131e1ee6)
Change-Id: I1970eba7e732728699042a36b89571915ec81451
Clogged queue means UI and other commands can not reach
the service and will cause volume controls to malfunction.
Example dump snippet:
Message handler (watch for unhandled messages):
Handler (com.android.server.audio.AudioService$AudioHandler) {a904698} @ 61825
Looper (AudioService, tid 118) {6808ef1}
Message 0: { when=-52s689ms what=25 ... }
Message 1: { when=-52s8ms what=8 ... }
Message 2: { when=-52s4ms what=8 ... }
...
Message 35: { when=-2s436ms what=0 ... }
(Total messages: 36, polling=false, quitting=false)
This indicates that the command queue is not being processed
because the handler got stuck. Checking stacks of the system_server
can help in figuring out the reason.
Bug: 135763139
Test: adb shell dumpsys audio
Change-Id: I02009720668af218edb38d5a5948a5964b72c685
Merged-In: I02009720668af218edb38d5a5948a5964b72c685
Do not show volume panel on TVs when system audio mode is ON.
Bug: 130524216
Test: make
Change-Id: Id8730455ee79f0ea3fa4c23e0a8b6bd2791eb2d3
Merged-In: Id8730455ee79f0ea3fa4c23e0a8b6bd2791eb2d3
(cherry picked from commit db961d82edee13b7f2a1e83f451c4f20430f8fce)
Added the property profilesystemserver in the RUNTIME_NATIVE_BOOT
namespace. This property is overrides the system one if it is
present.
Bug: 139883463
Test: set the property manually and verify that system server is started
Test: with profiling
(cherry picked from commit 7b31c74ddb)
Merged-In: Ifd69530e52a717a381b3f91b15a74329614906f2
Change-Id: I00594949b845a75152c7ab3c94aa84a55b072776
Some Zygote code that was pushed to AOSP master is different from the
pushed Q release. In particular, a call to set the process name for a
Zygote child was missing, causing an app zygote test to fail.
Bug: 139535125
Test: atest android.app.cts.ServiceTest#testAppZygoteServices
Change-Id: I53b0bc0116f1573cb1e52a998e10346bd5601f67
Merged-In: I2bce277ff8f2de4614e19d5385fe6712b076f9c9
Code that caused it to fail was fixed.
Bug: 139535125
Test: atest android.app.cts.ServiceTest#testAppZygoteServices
Change-Id: I5f8eed38d4a8a56e750a6bb7fa7d397d9cd4c13d
Merged-In: I2bce277ff8f2de4614e19d5385fe6712b076f9c9
Exempt-From-Owner-Approval: Re-enabling fixed TEST_MAPPING tests
for AOSP after Q push.
According to the original documentation, the functions in Process.java
should return -1 when the process doesn't exist,this should be detected
by Process.readProcLine() when it fails to open the /proc/<pid>/status
file. Unfortunately readProcLine() zeroes-out the output array before
attempting to open the file, so the expected return value -1 is changed
to be 0 when the process doesn't exist. It's inconsistent with the
interface description definition, so correct it.
Test: try to use Process.getUidForPid(int pid) api to read a non-existent
process pid's uid, it will return -1 instead of 0.
Merged-In: Ic2ac7defea8b15dd65f907d04e168a1283a4f621
Change-Id: Ic2ac7defea8b15dd65f907d04e168a1283a4f621
Signed-off-by: qiwang <wangqi_a@xiaomi.com>
Resolved issue 118612296.
On any notification error, the NMS silently cancels the notification,
including foreground service notifications. Thus, an app could pass in a
garbage notification deliberately and start a foreground service silently.
This patch resolved this issue by judging the notification's flag, and if
it is a foreground notification, still crash the app as previous platforms,
and if it is a normal notification, don't crash the app.
Background: In 3ad4cdd1, which was merged into Android 9 release, the crash
behaviour is removed. But it is an important rule that foreground services
guaranteed to show an ongoing notification.
Test: Run the sample apk provided in the issue, it's main thread received
a RemoteServiceException: Bad notification posted from package...
as intended behaviour.
Bug: 118612296
Merged-In: I94d9de50bb03c33666471e3dbd9c721e9278f7cb
Change-Id: Ice6faab055e2dacd3d2d12803fcf51e13d3b0f21
Geo-Fencing Maximum Wait Time a device shall allow to determines it
location meeting operator policy. If the device is unable to determine
its position meeting operator policy within the GeoFencing Maximum Wait Time,
it shall present the alert to the user and discontinue further positioning
determination for the alert.
Bug: 139022202
Test: manual
Merged-In: If3d8d7915f2d86da20c63e99bee01f30e3ea89c0
Change-Id: If3d8d7915f2d86da20c63e99bee01f30e3ea89c0
(cherry picked from commit 02d6ac0f20)
The existing LayoutCompilerTest only covered compiled layouts generated from
text XML files. It did not cover layouts that were extracted from the APK. This
is an important area to cover.
This test works by calling `pm compile --compile-layouts` on the test APK, and
then directly loads the `compiled_view.dex` file. From there, it pulls out the
compiled layouts and executes their inflater. This makes sure the generated DEX
files validate and execute without error.
Bug: 111895153
Merged-In: Ifc6719c2f438474b5474def02422ef68eee800b2
Change-Id: I7024069bb4d77a81871366f259fb34ae54a55aac
Readding the @deprecated javadoc to Capabilities as it is valid
to @hide a superclass of a @SystemApi.
Bug: 138990145
Test: manual
Change-Id: I1749aa3f09b7782fd0d665238be4c288a33536ea
This timing issue is happened when test case already verified
NetworkMonitor#setAcceptPartialConnectivity() before
NetworkAgent send the CONNECTED state to ConnectivityService.
This test case expects that the
NetworkMonitor#setAcceptPartialConnectivity() will be invoked
when network state is CONNECTED in
ConnectivityService#updateNetworkInfo().
NetworkAgent communicates with ConnectivityService by
AsyncChannel. Before they connect to each other, the thread of
NetworkAgent is keeping idle, so waitForIdle cannot handle this
kind of case. Although they have connected to each other, but
they don't communicate with each other, waitForIdle() still
cannot handle this case. It means waitForIdle() can only handle
the case that NetworkAgent already connected to
ConnectivityService and NetworkAgent is communicating with
ConnectivityService. So, that's why waitForIdle() sometimes
doesn't work in this case, and the most simplest and right way
to fix this case is to wait the available callback then verify
the method.
Bug: 140306320
Test: atest FrameworksNetTests: \
ConnectivityServiceTest#testPartialConnectivity \
--generate-new-metrics 500
Change-Id: I64bfecbfe3c5776bd5c347a821f9127e79539c48
