am: c65d4b728c -s ours
am skip reason: change_id Ia64238f30620d9906aa50c2469121c2653875e8d with SHA1 5dc581aef2 is in history
Change-Id: If188f2ed89de4ac8f0366b0f0b9ea979e90575ba
- This is preferable to canceling the animation, which the user just
interprets as the nav gesture not working, and may fix some cases
where the user may interpret a swipe up as not working if the wallpaper
is slow to draw.
Bug: 135536076
Bug: 134091263
Test: atest RecentsAnimationTests
Test: atest RecentsAnimationControllerTests
Change-Id: Icf9ed20971a701a254383c65aa1811a99680ba51
Allow muting BT SCO volume only from explicit mute request and not
when stream volume index is set to 0. This is to match BT requirement
that the valid SCO volume range is 0 to 15 and let the headset deal with
actual attenuation for index 0.
If explicitly muted via adjustStreamVolume() from an app with permission
MODIFY_PHONE_STATE, then we set the audio policy stream volume to 0 and
effectively mute RX path.
Also change default value of Settings.System.MUTE_STREAMS_AFFECTED
to include AudioManager.STREAM_BLUETOOTH_SCO.
Bug: 111195322
Test: test volume changes in call over BT SCO
Change-Id: I3b6560bbd1bcbb01a29aa746f23bae74329a1fa7
This reverts commit 0999f93e4a.
Reason for revert: There is a better choice (ag/8051966) than adding the permission to resolve b/130827484
Bug: 130827484
Change-Id: I1b8fd74a173d4b0ef981e51f7e0a9c5f84d5f416
The AppOpsModeChangeWatchers call back even if the mode was set to
already set mode. PermissionPolicyService listens to the same app-ops it
changes. I.e. it used to change the mode to default, which then causes a
call-back which made set it to default again, etc...
This exhausts a thread until the app-op is finally set to non-default.
The fix is to prevent the loop by checking if the app-op is already the
correct mode before setting it.
Fixes: 135674928
Test: atest IncidentReportListenerTest CollectorHostsideLibTest on
cuttlefish and crosshatch
Change-Id: Ic65945e814957ac59495b3da221fabb0f3d42b66
This CL changes the default behavior for trust agents to extend unlock
mode.
The Q droidfood phase allowed trust agent behavior to switch between
active unlock (legacy behavior) and extend unlock modes --- where the
extend unlock mode would prevent a trust agent (such as Smart Lock)
from actively unlocking a locked phone, instead only allowing it to
extend the duration that an already unlocked phone can continue to
stay in the unlocked state. In preparation for Q proper, the developer
options that controlled this toggle are being removed, and extend
unlock mode is the only allowed mode of behavior for trust agents.
Bug: 120871688
Test: Trust agents (like Smart Lock) only extend unlock on Q
Merged-In: Ifad45f90e0f2b5c3d180a74a97a89d45d75cb02a
Change-Id: Ifad45f90e0f2b5c3d180a74a97a89d45d75cb02a
Compiled methods don't get any interpreter activity modifying their
hotness counters. This CL compiles more methods to prevented
Original set of profile is the updated profile with revised thresholds
for image classes. Here we used a 3% threshold.
Additionally, the CL adds imgdiag diagnosed methods that have dirty JIT
counters shortly after boot on the following apps:
system_server
com.android.systemui
com.android.phone
com.android.nfc
com.google.android.networkstack
com.android.bluetooth
com.google.android.apps.nexuslauncher
com.google.android.gms.persistent
com.google.android.as
com.google.process.gservices
com.google.android.gms
com.google.android.inputmethod.latin
Before:
Boot oat size: 32,800,596 bytes
Persistent app .art PSS: 19115K
Persistent app .art Private Dirty: 15356K
System boot .art PSS: 60948K
After:
Boot oat size: 37261124 bytes
Persistent app .art PSS: 12610K
Persistent app .art Private Dirty: 8740K
System boot .art PSS: 50735K
Private dirty savings for persistent apps: 7MB
Net RAM savings: 5MB (10MB dirty .art PSS, 5MB oat size incrase)
Bug: 134776403
Test: Reboot and inspect RAM usage
Change-Id: Ia1c59b4be341a28de02b7d20aedcb6b69815c652
