The system is often willing to clear cached data to make room for
incoming installs, so use StorageManager.getAllocatableBytes() when
making "does it fit?" style decisions.
Add new INSTALL_ALLOCATE_AGGRESSIVE flag, which will flow through
to use StorageManager.FLAG_ALLOCATE_AGGRESSIVE when making allocation
related requests. (This can be used by installers to indicate
packages that are critical to system health or security.
Test: runtest -x frameworks/base/core/tests/coretests/src/android/content/pm/PackageHelperTests.java
Bug: 36131437
Change-Id: If8118762fd1ca1f497d2cdd1787bdb3c9759dcc0
There is some flakiness in View#onConfigurationChanged callback -
if ViewRootImpl receives config update earlier than ActivityThread,
it may not detect the configuration change and skip inner updates.
Also now ViewRootImpl assumes that it receives the global config as
a param, but instead it gets merged config from WM. This means that
ViewRootImpl#sConfigCallbacks was sending incorrect values to the
recipients.
This CL switches to sending global and override configuration to the
client separately. Also in case if there is a corresponding activity,
it first updates it and waits for update callback to ViewRootImpl.
This way global config and override config for activity will always
be set first and resources will be updated before inner state of
ViewRootImpl is updated.
Bug: 35870157
Bug: 34164473
Test: android.server.cts.ActivityManagerDisplayTests
Test: testOnMovedToDisplayCallback
Change-Id: Ic9e7541cf25ecfac6ec90e48f7efb0ece91f657e
Apps can listen for the LOCALE_CHANGED broadcast to
change names if needed.
Fixes: 35336590
Test: runtest systemui-notification, cts, manual
Change-Id: I62d2778f61e99200692755a6f268828a374837e6
Root Cause: systemServer's class path is not set
after set wrap.system_server property, and restart system_server,
it shows a java.lang.RuntimeException:
Missing class when invoking static main com.android.server.SystemServer
Solution: Correctly pass and parse a passed classpath.
Bug: 34692265
Test: adb root && adb shell stop && adb shell setprop wrap.system_server logwrapper && adb shell start
Change-Id: Ia6707dc05fa627af6cc28360d26b894487a6eff1
- Adds a scrim that fades in as the PiP is dragged past the bottom movement
bounds.
- Tweaks the values for minimizing the PiP and dismissing the PiP
- Fixes an issue where the PiP could be minimized in the bottom corners
of the screen
Test: Manual - drag PiP to bottom of screen to dismiss or swipe down on
PiP when it's at the bottom of the screen
Bug: 35358768
Bug: 35358628
Change-Id: Ib4ad5ed7094fef76ea979b46ea7e03d746625d00
Also fix a bug that doesn't trigger directory specific update for
removal.
Test: Manual tests.
Bug: 31001477
Change-Id: Ibae8a7aa5c72c82dc5e3cb4951fbc91ea6456964
(cherry picked from commit c574ec40640d83dfbbb0f022103c6dfa774bca40)
Please see commit 3082eb7c72 for an
explanation of this change.
This capability is not used by system_server.
Bug: 34951864
Test: code compiles, device boots, no selinux errors ever reported.
Change-Id: I4242b1abaa8679b9bfa0d31a1df565b46b7b3cc3
Commit ba0370eef3 added a call to
preloadOpenGL(), The nyc-mr2-dev cherrypick of that same commit
(b3a1417663) was improperly handled
by the automerger and added yet another call to preloadOpenGL().
Remove the duplicate call.
Test: code compiles.
Change-Id: Ie2a1c73155197506ee9571f2c05988c77a750281
Displays can be rotated separately and rotation watcher clients
are only interested in rotation of some specific display. This CL
adds displayId to rotation watchers and only informs them about
changes on their display.
Bug: 34242678
Test: Manual and debug.
Change-Id: If0f03804da0392c2b14a4e7c2d6a06068ad8760b
This allows to avoid A11yManager -> A11yManagerService IPC, when there's no
subscribers to a given event
Test: steps:
- Enable A11yManager.DEBUG
- Navigate through a few random activities
- In logcat, ensure log messages are present, notifying that certain events
won't be dispatched
Change-Id: Ia019fb66053f10095b3651407d09de8e89cdd227
Messaging style is now not dropping previous messages
anymore if they don't fit completely. Instead we are
truncating it.
Test: send various text messages, observe nice layout
Fixes: 35944700
Change-Id: I0a08b06bf7624ee62e81bfd384721c8458c9c9d4
Detaching a Counter from a TimeBase means you can't use it anymore.
Only newly constructed Counters are automatically attached to a TimeBase.
Bug: 34200689
Test: make FrameworkCoreTests && adb install ... && adb shell am instrument -w -e class com.android.internal.os.BatteryStatsSensorTest com.android.frameworks.coretests
Change-Id: I9309000d5625aa6fe61a3c05f135e5828137d8ce
This fixes the unexpected priority 112 of the daemon threads
(eg. HeapTaskDaemon). The problem was that when the zygote main
thread's priority is reset, it directly calls setpriority() and fails
to update the priority value in java.lang.Thread, which in turn causes
any threads created by the thread to unexpectedly inherit the boosted
priority. Calling java.lang.Thread.setPriority instead fixes.
(cherry picked from commit 1e3db871e5)
Bug: 35801778
Bug: 28866384
Test: angler master userdebug boots and thread priorities checked.
Change-Id: I68a6ed7244a9067acc2749feca7f88422bf44b02
Exposing actions from the PIP InputConsumer to accessibility,
stripping all actions from a covered PIP app, and adding the
InputConsumer's actions on the PIP app's root view.
We were also using an "undefined" accessibility ID to mean
three different things: a root view, a host view of a virtual
view hierarchy, and a truly undefined view. I've introduced
new values for cases where the id could be defined.
Also gathering all window IDs into one place to reduce the
chance of collisions.
Bug: 34773134
Test: In progress. Current cts passes.
Change-Id: I97269741a292cf406272bf02359c76c396f84640
By supporting multiple filters per one request we should be able to cover
multiple kinds of use cases such as:
- Letting the user select from a list of devices of more then one medium
type (e.g. Bluetooth and BLE)
- Allowing to provide multiple criteria for any field (e.g. filtering by
more than one service UUID)
Bug: 30932767
Test: Provide multiple filters and ensure that devices matching either are
shown in the list to choose from.
Ensure wifi SSIDs are shown in the list if wifi filter is provided
Change-Id: I0a978787551a1ee5750ec5544b241d3bbfed5a7c
* changes:
AOD: Add wakelock for charging text while dozing
AOD: Show charging status on AOD1
AOD: Fix default for always-on setting
Revert "Revert "AOD: Refactor always on configuration""
Change Ic35bd3c04d150fd3eb85d76db0043880b31a011f mistakenly
flipped the default to on. Change it back to off.
Test: runtest -x frameworks/base/packages/SystemUI/tests/src/com/android/systemui/doze/DozeConfigurationTest.java
Fixes: 36027947
Change-Id: If1619c10f8cc269f2dd852e166ac959e6c9c590a
Phone portrait: restrict movement to left / right edges
Phone landscape: restrict movement to top / bottom edges
Tablet: allow PIP on all edges regardless of orientation
Test: manual - have PIP try to move along top / bottom edge
in phone portrait, and left / right edge in phone landscape
Fixes: 35358644
Change-Id: Ie50ae220782936942d1bcd8d9251d7275832a47f
Begin migration to non-static methods to allow easier mocking.
New clients should use the non-static methods.
Old clients that want to unit test metrics
should move to the new interface.
Bug: 35138327
Test: runtest --path frameworks/base/packages/SystemUI/tests/src/com/android/systemui/statusbar/phone/StatusBarTest.java
Change-Id: I168f3787ee50ddde8aa2d42c05b2b816d1d3a30e
Here we have SurfaceView bypass the WindowManager and speak
directly to SurfaceFlinger using child surfaces. We also
implement some logic in the WM to handle child surfaces
in various Surface replacement scenarios.
For those following along in the revert Saga, this
also includes the follow up CLs to the original CL.
- Surface inset calculation
- Animation fixes.
The error causing the revert was an incorrect JNI signature
around deferTransactionUntilSurface. I've noted it inline.
Bug: 28858420
Bug: 31518219
Bug: 34888808
Bug: 35588318
Bug: 35396882
Test: Existing tests still pass (except for the ones that don't and will be deleted).
Change-Id: Ie56b6f7ab16f32d7fc459b8eba26594337ad55de
By supporting multiple filters per one request we should be able to cover
multiple kinds of use cases such as:
- Letting the user select from a list of devices of more then one medium
type (e.g. Bluetooth and BLE)
- Allowing to provide multiple criteria for any field (e.g. filtering by
more than one service UUID)
Bug: 30932767
Test: Provide multiple filters and ensure that devices matching either are
shown in the list to choose from.
Ensure wifi SSIDs are shown in the list if wifi filter is provided
Change-Id: I6621da388e2bf4ed97c5af2692629a321d0b63c7
This fixes the unexpected priority 112 of the daemon threads
(eg. HeapTaskDaemon). The problem was that when the zygote main
thread's priority is reset, it directly calls setpriority() and fails
to update the priority value in java.lang.Thread, which in turn causes
any threads created by the thread to unexpectedly inherit the boosted
priority. Calling java.lang.Thread.setPriority instead fixes.
Bug: 35801778
Bug: 28866384
Test: angler master userdebug boots and thread priorities checked.
Change-Id: I68a6ed7244a9067acc2749feca7f88422bf44b02
- fs_mgr reports fs_stat int value during bootup when mounting
an ext4 partition.
- This includes information like:
if previous shutdown was clean
if e2fsck has fixed any fs issue
if any mounting or quota related operations has failed.
- Each information is set as flag. Flag definitions are avaiable from
FsStatFlags in system/core/fs_mgr/fs_mgr.cpp
- log ro.boottime.init.mount_all.[early|default|late] which represent
time duration for each mount stage. Depending on device's config, some of
these stages may not exist. So do not log if value is 0, which means not existing.
bug: 32246772
bug: 35949600
bug: 35329915
Test: manual, check log after boot-up.
Change-Id: I4ceca88776119d88d5352d6793be122e82688279
