External sdcards are accessed through /mnt/media_rw,
so access() each touched file for sdcardfs to update
its metadata.
Java handles single object renames directly so that
is the only place where it is needed.
Bug: 77849654
Test: use mtp with emulated sdcard
Change-Id: Ie460398010f1fe74d8084808a6333b121674362c
To make surround sound setting more clear to users, we are changing
surround setting from ALWAYS to MANUAL. With MANUAL, users could enable
surround formats according to their need.
Bug: 67479735
Test: Try creating AudioTrack with enable/disable surround formats.
Change-Id: I79e610f6c43d7003daf13da3bee4e03ba9b6ea87
Due to permissions changes, we now need to access
the underlying filesystem of removable devices in
order to get write access.
Add internalPath to StorageVolume, and have VolumeInfo
set the field on creation.
Bug: 77849654
Test: Can write to emulated sdcard through MTP
Change-Id: I63302ecf2dd2600a1c9f3f6ab106c3695654cbaa
Some sniffers in MediaExtractor might issue reads past the end of
the file. This results in IOException in our MediaDataSource. The
exception itself is not fatal, but need to make sure MediaDataSource
state is kept in sync otherwise the other extractors can't continue.
Bug: 79497983
Test: test the ExifInterface on the attached file in bug, as well as
selected HEIF files from other devices, Exif should be extracted
without error in the log.
Change-Id: I5c8f597b8ef14288e3465670b202adc93c09e6ea
Current limitation is that we cannot distinguish between key events
from Window#injectInputEvent() and key presses from physical devices.
Bug: 79185725
Test: Manually verified with the test app, and Cts
Change-Id: I1c3da4a963238512eec8b57fbc2d825c7436b140
AudioService:
* Call setBtScoActiveDevice and setBluetoothScoOn both in AudioService's
broadcast receiver so that these two methods must be triggerred in
the same sequence as ACTIVE_DEVICE_CHANGED and AUDIO_STATE_CHANGED
intents are sent and we no longer need to handle race condition by
synchronously checking active device in setBluetoothScoOn
* Default sco audio mode when no headset is active should be virtual
voice call, as many HFP devices do not accept SCO audio without an
ongoing call
* Synchronize checkScoAudioState() method with mScoClients
* Add helper functions connectBluetoothScoHelper and
disconnectBluetoothScoHelper to call various SCO setup and tear down
methods based on sco audio mode
* Try raw, virtual call, and voice recognition mode when disconnecting
externally started SCO
* Add new sco state SCO_STATE_DEACTIVATING to allow back to back calling
of startBluetoothSco and stopBluetoothSco
Audio Manager:
* Modified AudioManager logic so that removed devices callback is called
before newly added devices
BluetoothHeadset:
* Modified BluetoothHeadset so that start and stop SCO using virtual
voice call no longer need a parameter and will use active device by
default
* Modified documentation around various sco mangement APIs to match
their expected behaviors
Bug: 76114959
Test: VoIP calls sanity test cases
Change-Id: Id50db88f4ff36069b0f392c81dd9d90c24cd2206
A corrupted image file may create two problems.
1. A corrupted IFD pointer may point to an IFD that has already
been read, thus creating an infinite loop and a stack overflow.
2. A corrupted IFD offset value may have a negative value, thus
prompting a random reading of the file and creating an infinite
loop.
This CL addresses these issues.
Bug: 63800695
Test: Run cts (ExifInterfaceTest)
Change-Id: I706a0c3eae6af8301af69407333ea88e5681df3c
Source of deadlock between PlayerBase.mLock and
PlaybackActivityMonitor.mPlayerLock:
android.media.MediaPlayer.release()
> android.media.PlayerBase.baseRelease()
> synchronized (mLock)
> com.android.server.audio.PlaybackActivityMonitor.releasePlayer()
> synchronized(mPlayerLock)
and:
com.android.server.audio.PlaybackActivityMonitor.unmutePlayersForCall()
> synchronized (mPlayerLock)
> android.media.PlayerProxy.setVolume()
> android.media.PlayerBase$IPlayerWrapper.setVolume()
> android.media.PlayerBase.baseSetVolume()
> synchronized (mLock)
playerSetVolume()
Since system_server can have its own players, the calls to
AudioService from PlayerBase can be synchronous, hence the
deadlock.
The fix consists in never holding the lock in PlayerBase
while calling into AudioService.
Refactor the playstate update into a method used for
start / stop / pause.
Bug: 72294559
Test: see bug
Change-Id: I6451aa3bf19a0365472ba007b116a9e6151ab33e
Use the heif embedded thumbnail as long as it's not
too small to avoid decoding the full image and downscale.
bug: 74395267
Test: CTS MediaMetadataRetriever test; manual test of thumbnail
extraction by browsing new folders containing heif files in
Downloads app.
Change-Id: I5b2be0521452376153a773cb7671fefe7f9bc439
For microphone enumeration APIs, it could fail due to hal v2,
which is WAI. We shouldn't log as error. Using warning instead.
Bug: 64038649
Test: build
Change-Id: I1a64b8537e27e712a09f122d4763e0e5a345b08f
