Background: A regression was introduced when something about the MediaHeaderView's visibility was changed, which caused that view to be counted when showing notifications but not when counting how many should be shown. That regression caused b/188000471 which was fixed by ag/I93b4b6039eb812619787744f893fe1924e1afc4d. However, that fix only addressed half the problem, as it resulted in the media player always being shown, regardless of the hard-coded notification max.
This change brings the logic of these two methods in line with each other, so that the number of notifications shown *includes* the media player, but ensures that a zero-heignt media player is not treated as a notification (similar to how any GONE views are ignored).
Fixes: 189102584
Test: manual
Change-Id: Ic7447941593f08ed504f3870af5a72d2628a043e
This file was written on the assumption that bindService was
synchronous, which it isn't. This change adds a CountDownLatch to force
the class to wait for the binding to finish.
Bug: 190222116
Test: atest RemoteProvisionerUnitTests
Change-Id: I917a61da612f21f9a0f783bea5d24270d4e1db42
- SplashscreenContentDrawer.ColorCache needs INTERACT_ACROSS_USERS_FULL
to register the broadcast receiver.
Bug: 190516351
Test: Build Andorid Auto OS and check if CarLauncher is showing.
Change-Id: Ib0700a0f2ba1c31ffc60199a3aa8019961464efe
Instead of separately querying light ids and light infos, just provide a
list of light infos. They already contain the light ids.
This way, we also avoid leaving uninitialized array elements when the
lookup fails.
In the InputReader code, a lock is obtained when the light information
is being read, but that memory could change right after the call
completes.
Bug: 190126442
Test: atest inputflinger_tests libinput_tests
Change-Id: I16ef37dce96d7077a8634d565b573fc26f1ebb5d
Ensure sensors are cancelled when ending an AuthSession to prevent stale operations from blocking new sessions.
Fix: 190213740
Test: atest com.android.systemui.biometrics com.android.server.biometrics
Change-Id: I53687b069ef130ae4f4b6e8d64f3431a4485c5af
When we fail to create an SkSurface from a dequeued
buffer we then call cancelBuffer on that buffer. However,
we don't update our tracking to say the buffer is no
longer dequeued with and doesn't own the fence any more.
I don't think this is the main issue causing the crash in
the attached bug because I don't think we're seeing the
error in the logs for this chunk of code. However this
possibly related issue was found while tracking down our
use of the dequeue_fence fd.
Test: manual building and running of phone
Bug: 187240173
Change-Id: Icb4099eeea5be6aedd5376e07a6e3454f5d1d1e3
KeyguardClockSwitch will be recreated by
NotificationPanelViewController for a variety of reasons, including
onDensityOrFontScaleChanged(). When this happens, the large/small
clock state can get out of sync with the initial state. Set the clock
state to null on creation to avoid this.
Fixes: 188390632
Test: manual, it's a race condition that can be hard to trigger
Change-Id: I49cda9afd5c708b05aba56a40d29c1cefe847de4
It'll help to debug content provider related performance issues.
Bug: 190416935
Test: Manual - Record perfetto trace & verify auth name is logged.
Test: CtsContentTestCases:android.content.cts
Test: FrameworksCoreTests:android.content
Change-Id: Ifaa1c58135e6aa2a46ecbba92a9266e7d29d5421
Added a mechanism for underlying data service to report
service temporarily unavailable. In this case, frameworks
uses the configured retry timer to retry establishing data
connection again.
Fix: 184122463
Test: atest DataConnectionTest
Change-Id: I57705b720b6438d8f9f0076f83decc1c17782f6b
