1. Change from onStopUser to onSwitchUser since onStopUser will actually
never be called.
2. Add some checks in the cancelAttentionCheck()
Bug: 111939367
Test: manually
Change-Id: I155a3acb20ef7fcad693a0a319b097cf50a474d1
With my previous CL [1], IMMS#calledFromValidUserLocked() now always
gets called with the same parameter. Let's simplify it.
This is a mechanical refactoring. There should be no behavior change.
[1]: Ieb94ed14698ee2669adc5fb2ea0e93c7d0e09e8b
716897ce2e
Bug: 34886274
Test: atest CtsInputMethodTestCases CtsInputMethodServiceHostTestCases
Change-Id: I571db5a648ddf8b35caad3d3b64549b7cc067d14
Plugins might be initialized before the StatusBar service is ready.
This means that #setForcePluginOpen might be called before #add.
Test: atest StatusBarWindowControllerTest
Change-Id: I99a220521272a2348ecece4c4581e7c82fa79490
Fixes: 123051520
libmedia2_jni.so was kept in /system/lib to make MediaPlayer2
work before we have native_loader change. Now, it is not needed
anymore since we use the libmedia2_jni.so in media apex.
Test: build & boot & atest MediaPlayer2Test
Bug: 112766913
Change-Id: I305bd4a4fad4b9528d22bc180ebcf3a9725840f0
To know from what UI location a notification was expanded we here pass
the location of the notification through
onNotificationExpansionChanged().
Bug: 120767764
Test: atest SystemUITests
Test: atest NotificationManagerServiceTest
Change-Id: I20612f5a88cf987f434392ae7a004d3ee3098998
From device boot until BOOT_COMPLETED, log the dispatch latency of a
broadcast, which is the duration between it being enqueued and it being
dispatched, to get an idea of how long the queue is blocked for.
Each time a broadcast is dispatched, we check mLogLatencyMetrics, which
defaults to true when the queue is initialized, and log to statsd.
When the queue is empty, we check if BOOT_COMPLETED is done. If it is,
we stop logging by setting mLogLatencyMetrics to false, since it means
that we have finished all broadcasts that
would have been blocked by BOOT_COMPLETED.
Bug: 120794810
Test: flash, reboot, use stasd_testdrive to verify logging
Change-Id: I40c14744e5247412c0a03032fd5621706758a324
This is in preparation for adding more logic related to docked state.
Bug: 122301289
Test: Used adb to switch between custom clock faces.
Change-Id: I0a13ef4e5894df1280a04c5a9df8bbed8054557b
If we have multi-package sessions, we check whether they contain APEX
packages and proceed with the pre-reboot verification of the whole
session. Session is then applied by apexd on reboot, and the state is
recovered and updated by StagingManager as soon as the sessions are
restored.
Bug: 118865310
Test: adb install-multi-package file1.apex file2.apex; verified sessions
are written correctly, restored correctly on stop/start, applied
correctly at reboot and their state is updated.
Test: atest apex_e2e_tests (to make sure we didn't regress)
Change-Id: Ia31d5badace016074d158e31c574fbad81576984
The existing code didn't restore the child sessions, since the
XmlPullParser object was advanced within the
PackageInstallerSession#readFromXml code (in order to read grant
permissions).
Slightly refactor the code to make it more self-contained and testable,
and adding a test to make sure that the most important bits of a
session, or of a group of sessions, are dumped and restored properly.
Bug: 118865310
Bug: 109941548
Test: atest FrameworksServicesTests:PackageInstallerSessionTest;
verified that install_sessions.xml is correctly written and restored in
presence of multi-group sessions.
Change-Id: I5aef61f64e7223844f11661ee068dd3c1e112611
When executing rollbacks, we now set the SessionParams installerPackageName
to the targetPackage installer instead of using the default which is the
calling package.
Test: atest RollbackTest with Ic85464583f7c6b0778d853a87670d250a42c33de in
succession works.
Bug: 112431924
Change-Id: Id115c2c0d6838d81213dfc0d186ef30e13d6f879
