Found a couple of responsive issues and fixing them.
1. Numpad key digits had a min size of 32 sp. This would look off when
we maximize the display size and go into half folded on a foldable
device. I set the text to autosize with a max size of 32sp.
2. Userswitcher anchor was squished in password security method on
tablet. I removed some clipping in order to fix this.
Fixes: 261062809
Test: Open password view with max display and normal display sizes for
tablet.
Test: Half fold on foldable for pin view in bouncer.
Change-Id: I78fbbb231c35118bef65660b7ca5562d77dad16b
If a permission has the "pre23" flag, and an app is upgrading past api
23, then we should not assume that a "development" permission remains
granted
Fixes: 259458532
Test: atest RevokeSawPermissionTest
Change-Id: I214396f455c5ed9e8bac2e50b1525b86475c81c7
Turns off handling of `-e mobile ...` commands from
`DemoModeController` if the new status bar mobile icon pipeline is
enabled.
This will be a requirement to transfer demo mode to the new pipeline
Bug: 249790009
Test: manual; enable demo mode and flag `606`, and send mobile demo
commands
Change-Id: I6beaccc56d2c2b816c2983c3822f0e15b022efe5
This sets config_show_sidefps_hint_on_bouncer to true as default, and removes the large screen override.
Bug: 260195934
Test: Current tests && verified manually
Change-Id: Icc92a0915f229e716e653e742173181caad050da
Binder.getCallingUidOrThrow throws if the caller is NOT:
1) Called from within a binder transaction OR
2) Called from within a
Binder.clearCallingIdentity/restoreCallingIdentity block
Number 2 was added in I162db933f9e52cd6f9f46796bda11ad6216d3d66
to allow us to migrate from Binder.getCallingUid to it. There are
many use cases where callers are not in a Binder transaction, but
have set the calling UID "explicitly" by calling clearCallingIdentity.
In order to further ease the migration from Binder.getCallingUid to
Binder.getCallingUidOrThrow, this hidden API behaves similarly,
but logs at WTF level instead of throwing IllegalStateException.
The motivation is to first migrate to the WTF version, so that
system_server will log and continue. The goal is to only use this
in the system (hence @hide). It will provide early signals as to the
feasibility of a given migration.
For example, its desirable to migrate the permission checking in code
generated by the @EnforcePermission annotation (via the AIDL compiler).
This call site is widely used, and itself subject to mass migrations to
@EnforcePermission instead of manual permission enforcement.
Bug: 252975769
Test: TH
Change-Id: I11ea3fdf889689d76211506e81420e694238ee68
Removes javadoc that appears in places where it is not considered as such
by javac / metalava.
Test: compiles
Change-Id: I701d7dfe17db67e4735e0ded4ff1c3e0ddb3bfa2
Instead of delaying 500ms
Test: atest KeyguardUpdateMonitorTest
Test: Attempt to authenticate with SideFPS right
after pressing the power button => observe FP can
authenticate
Fixes: 259172472
Change-Id: I7f07c549628a21a92ddfac0d0679d0f054ad1dda
This is a partial alternative solution of [1] to avoid black background
(no wallpaper target) when turning on screen with a translucent activity
which is occluding keyguard while AOD is disabled.
[1]: Ifc3ce28424fe87b1e0a0a2af07bdad559575ebec
Now instead of always recomputing wallpaper target after committing
visibility, the update hint only needs to be set when there is a
translucent occluding activity.
Also use usingShellTransitions to skip some legacy animation checks
that shell transition doesn't care.
Bug: 260059642
Test: Enable shell transition. Disable AOD
KeyguardTests#testDialogShowWhenLockedActivity
KeyguardTests#testTranslucentShowWhenLockedActivity
Test: Transition latency of OpenAppMicrobenchmark
Change-Id: Ic3c704f0579690c51f7188a26c5256dc3e74e1f7
If a display is disabled, do not share it outside of
DisplayManagerService.
This will help by not creating any unnecessary updates for displays that
are disabled, and also not informing Applications of displays that they
are unable to use.
Also fixing a concurrency issue with DMS.handleLogicalDisplaySwapped
which became apparent with this CL. Deferring the runnable with a
Handler.post() caused the display-state-OFF update to happen after a newer
change to turn the display ON. This caused the device to startup into a
black screen. Removed the unnecessary call (since the displays are now
made to both go into the "transition" state) to update the state and
removed a lot of the redundant code with handleLogicalDisplayChange()
which was called at the end of the method anyway.
Bug: 221071695
Test: atest com.server.android.display
Test: Manually verify display layouts still successfully change, and
also confirm logs no longer hand excessive add/remove
display notifications.
Change-Id: I003fd91567630a447cf030b6fb83e5cb1041bcb5
Merged-In: I003fd91567630a447cf030b6fb83e5cb1041bcb5
in order to share surface effects between launcher and sysui, the
package has been moved to the AnimationLib, which is visible to launcher
as well.
fyi the corresponding tests remain in the SystemUiLib module.
technically they should be moved to AnimationLib together, but those
tests are dependant on SystemUI testing frameworks (e.g. SysuiTestCase,
FakeExecutor, etc.). because of this, some of the variables are marked
as VisibleForTesting.
Bug: 249536060
Test: MultiRippleControllerTest MultiRippleViewTest RippleAnimationTest RippleViewTest TurbulenceNoiseControllerTest TurbulenceNoiseViewTest
Change-Id: I37cdc7cf9064d4d8f02afe72ad16b1cc3df95b6c
