* changes:
  Preserve qsFullScreen update in NSSL
  Perform NLVM combines on the bg thread

* changes:
  Move barState to PanelExpansionInteractor
  Move isTracking to PanelExpansionInteractor

This was removed initially with the FooterViewRefactor, but we should
keep it in case it could cause any issues. I was hoping it would fix
b/329432313 but unfortunately it doesn't.

Bug: 293167744
Test: existing tests should pass
Flag: ACONFIG com.android.systemui.notifications_footer_view_refactor STAGING
Change-Id: Ia6a3506f05b8d45b35c3ef2160c9d34c96dbfdb5

It seems like although the flows themselves don't emit too often (thanks
to distinctUntilChanged), the combines still do quite a bit of work that
should ideally happen in the background. The biggest offender here is
the footer visibility, as it combines a lot of flows that can emit
relatively often.

Also reducing the amount of emissions in the shadeExpansion sub-flows
using distinctUntilChanged. Removed qsExpansion as it's redundant since
we're already checking qsFullScreen, which is only true when qsExpansion
is 1.

Bug: 326018643
Bug: 324483341
Test: v2/android-crystalball-eng/health/microbench/systemui/main/systemui-notification-4-jank-suite
Test: v2/android-crystalball-eng/health/microbench/systemui/systemui-notification-onmeasure-jank-suite
Flag: NONE

Change-Id: Ifc470bd1bbf9142c1114d7bdfb92449be5f754ce

This CL removes the isSpecified lambda from Element.computeValue. This
is not necessary anymore since http://aosp/2978852, which was pulled to
our prebuilts in http://ag/26476625.

Bug: 316901148
Test: atest PlatformComposeSceneTransitionLayoutTests
Flag: N/A
Change-Id: I668ee3694f17d610f5e151e560aa2012bd73d618

* changes:
  Restore direct display children to display layer
  Always use a DisplayArea for WindowingLayer

When performing a tilt-to-wake gesture on a Wear device, if the offload session fails, the policy is POLICY_DOZE but the screen state is STATE_ON.

The doze scale factor shouldn't be applied if the screen is on.

The initial doze brightness or the default doze brightness shouldn't be used if the screen is on.

Bug: 328490718
Bug: 328797653
Test: adb logcat | grep -i "brightnessEvent"
Test: atest AutomaticBrightnessControllerTest
Test: atest AutomaticBrightnessStrategyTest
Test: atest DisplayPowerControllerTest
Change-Id: Ic223bb73a5f5ded28c8be3a8e73c1d494be78a29

Previous change was blocking boot in HSUM. This fix adds a check to skip  Home launch delay in this systems.

Flag: ACONFIG enable_home_delay TRUNKFOOD
Bug: 326619801
Bug: 326841974
Test: v2/aaos-engprod/device_boot_test_aae_toolbox
Change-Id: I3d869541587b12d94c688c888a407f6e463914c8

It should be optional for the app window to be invisible
before turning to visible.

Bug: 329929112
Test: OpenActivityEmbeddingSecondaryToSplitTest

Change-Id: Ib82f24afc014884d32de6694fa0d9b9279e00936

Flag: ACONFIG com.android.systemui.scene_container DEVELOPMENT
Test: Updated and ran affected test
Bug: 303267342
Change-Id: I060272e61bd11377c0716515bff0a2fea2bd3d8f

Flag: ACONFIG com.android.systemui.scene_container DEVELOPMENT
Test: Updated and ran affected test
Bug: 303267342
Change-Id: I50679b666e30894ef1fe9729233f655e3b7a2a7c

This undoes I78d4832b0ff7623684fc22fe1f464a02a1d91477.
Because now a display always has a real DisplayArea to provide
windowing layer (created by DisplayAreaPolicyBuilder).

Old:
 RootWrapper (DisplayContent#mSurfaceControl)
   > WindowingLayer (pure layer, not a WindowContainer)
     >> Child of display

 The surface of direct child under display uses WindowingLayer
 as parent.

Now:
 Display (DisplayContent#mSurfaceControl)
   > WindowingLayer (DisplayArea)
     >> Child of WindowingLayer

 The surface of direct child under display uses display as parent.
 Because the surface hierarchy is consistent with window hierarchy.

Bug: 326975721
Bug: 266243524
Test: AccessibilityOverlayTest#testA11yServiceShowsOverlayOnVirtualDisplay_shouldAppear
Change-Id: I7a1d09eb566e29e532d56627a9955692b55a5682

This fixes the case if FEATURE_WINDOWED_MAGNIFICATION is supported:
 Correct hierarchy
  Display 0
   > WindowedMagnification:0:31 (WindowingLayer)
   > HideDisplayCutout:32:35
   > Leaf:36:36
   > Overlays

 Wrong hierarchy after calling migrateToNewSurfaceControl
 (The DisplayArea 32~36 should not belong to 0~31)
  Display 0
   > WindowedMagnification:0:31 (WindowingLayer)
     >> HideDisplayCutout:32:35
     >> Leaf:36:36
   > Overlays

There are various assumptions that the real parent surface should be
used as the target of reparent except display. Such as onParentChanged
and animation leash management. So it might not be appropriate to be a
general override in WindowContainer.

To reduce confusing of WindowingLayer, this change makes the layer
name consistent with its purpose and window hierarchy. There will
be a real DisplayArea for WindowingLayer, so the exception case of
relationship between surface and container is gone.

Before:
 RootWrapper (DisplayContent#mSurfaceControl)
  > Display 1 (WindowingLayer)
    >> TaskDisplayArea
  > Overlays

After:
 Display 1 (DisplayContent#mSurfaceControl)
  > WindowingLayer
    >> TaskDisplayArea
  > Overlays

Bug: 326975721
Test: atest DisplayContentTests#testValidWindowingLayer
            DisplayAreaPolicyBuilderTest
Test: Kill systemui (which triggers setOrganizer ->
      sendDisplayAreaVanished -> migrateToNewSurfaceControl)
      and check surface hierarchy.

Change-Id: Id4c2723a67738827be81ffb6fd4f47cf3a094634

Original change: https://android-review.googlesource.com/c/platform/frameworks/base/+/2990318



Change-Id: Ib61c1ec7dd467deea50db74b04cfceea30fdb1b1
Signed-off-by: Automerger Merge Worker <android-build-automerger-merge-worker@system.gserviceaccount.com>

Original change: https://android-review.googlesource.com/c/platform/frameworks/base/+/2990318



Change-Id: I37ab3afec228ee1c9115dd9a245d8ef1b67b6a85
Signed-off-by: Automerger Merge Worker <android-build-automerger-merge-worker@system.gserviceaccount.com>

* changes:
  Exclude unnecessary window early in DisplayWindowsObserver
  Update window info by attributes in an earlier stage

Previously, some windows are excluded from the final list of
accessibility windows in populateReportedWindowLocked().
This is done after we compute touchable region to decide which windows
are touchable and to be exposed.
Thus, there are some corner cases where an unwanted window joins in
touchable region computation and occludes an important window,
resulting in missing window info.

With this change, unnecessary windows are excluded when we compute
touchable regions.

This is a retry of abandoned CL [1] with a cleaner approach behind
the computeWindowChangesOnA11y flag.

[1] Iacff5b7e67a9e7e989671dc85ad884b0e9f2f2f0

Bug: 240885392
Bug: 254581919
Test: android.autofillservice.cts.servicebehavior.SettingsIntentTest
Test: android.autofillservice.cts.inline.InlineTooltipTest
Test: CtsAccessibilityServiceTestCases CtsAccessibilityTestCases AccessibilityWindowManagerWithAccessibilityWindowTest
Change-Id: Id4b24284db40feaa744c9a5598d0067566fa3963

This is a pure refactoring behind computeWindowChangesOnA11y flag.
Doing this will simplify code flow after the flag is fully enabled.

Bug: 322444245
Test: CtsAccessibilityServiceTestCases CtsAccessibilityTestCases AccessibilityWindowManagerTest
Change-Id: I32514c709da04b50398fc5edc7dc4e7de172882a