First, this cancels the existing pip animation if it is
told to exit. Without this, an exit animation that
starts before an enter animation finishes will have its
state clobbered by the exit's finishTransition call.
Next, make pip-enter transitions queue-up so that they
don't interleave before other queued transitions (very
common during CTS).
Bug: 231150615
Test: atest PinnedStackTests
Change-Id: I7f162c0b1f24ab845770b805c8fae2252814945e
If an internal system window wants to drop focus from an embedded
window, requestFocusTransfer doesn't need to be called and instead we
can directly call setFocusedWindow. This fixes the case where transfer
focus fails if the old focused window loses visibility by the time the
transfer request arrives. The transfer won't be allowed because the old
window isn't focused anymore so we can't honor the transfer request.
This fix is fine for internal system windows because there's no security
issue with transferring focus from embedded to something WMS computes.
However, there's still a race condition for cases where apps want to
transfer focus from embedded back to host when they are setting
visibility on the embedded window since the embedded window can become
invisible before the transfer goes through.
Test: Pip Menu focus lost
Fixes: 230851770
Change-Id: I09db0bbdf4db6eeaffa30275233811b13ea31132
Merged-In: I09db0bbdf4db6eeaffa30275233811b13ea31132
If an internal system window wants to drop focus from an embedded
window, requestFocusTransfer doesn't need to be called and instead we
can directly call setFocusedWindow. This fixes the case where transfer
focus fails if the old focused window loses visibility by the time the
transfer request arrives. The transfer won't be allowed because the old
window isn't focused anymore so we can't honor the transfer request.
This fix is fine for internal system windows because there's no security
issue with transferring focus from embedded to something WMS computes.
However, there's still a race condition for cases where apps want to
transfer focus from embedded back to host when they are setting
visibility on the embedded window since the embedded window can become
invisible before the transfer goes through.
Test: Pip Menu focus lost
Fixes: 230851770
Change-Id: I09db0bbdf4db6eeaffa30275233811b13ea31132
This fixes conflict between app transition and recents animation in the
following scenario.
1) App transition animation finishes after app closing animation, which
is controlled by recents, finishes.
During the app closing animation, recents makes the closing app surface
invisible, but app transition animation overrides it to visible again.
This causes a flicker.
2) App transition starts during recents animation.
This can happen when a user launches an activity, and immediately after
that, swipes up the screen to close the app.
While recents is running, we assume animation on tasks is controlled by
recents, and visibility is commited without animation after recents
animation finishes. However starting app transition during recents
breaks this assumption, which ends up with playing one more unexpected
closing animation (so users see closing animation twice).
Bug: 223499269
Test: atest AppTransitionTest + manual
1. Launch Gmail app
2. Click icon on the bottom tab (e.g. Chat)
3. Swipe up from the bottom (immediately after step 2)
4. Verify closing animation only plays once
Change-Id: Id0a8b472b9a3d7cf5b55852de83cbd50b985b834
Reference RCU has 2 models with different numbers of keys.
And some keys have multiple HID key IDs used by different
partners.
This change lists all possible HID key values of reference RCU.
Change-Id: I9c3c93c17701f39b415989ccf066744fb314d29c
bug: 229692045
test: manually tested on tm-dev build
When the activity exits PiP and is reparented to the original Task, the
organizer should handle it as a new launch.
Bug: 225371112
Test: atest WmTests:TaskFragmentTest
Test: atest WmTests:TaskFragmentOrganizerControllerTest
Change-Id: Ia7e32e995a35e167b0d76e701c972b269ae068cc
In general this removes commit b2e3780. Because since rotation
is a part of configuration, it is enough to redraw according to
the configuration change.
Otherwise because forceRelayout is also used for syncRedraw,
that will cause to drop pre-allocated buffers and increase the
time of the initial frames for reallocation.
Bug: 229366100
Test: ActivityRecordTests#testLandscapeSeascapeRotationByApp
Test: Rotate 180 degree with various apps and no upside down
afterimage remains on the screen.
Change-Id: I3f03f7c8674ddc4a8c0b9c0b0528816114075732
The "core apps only" mode of system_server is no longer used, and
onlyCore is hard-coded to false. For details, see http://ag/17685636.
Remove the corresponding obsolete code from WindowManagerService.
Bug: 208476087
Change-Id: Icec64be737709b63c14f0d2a73c9abf6894ce87e
The transaction was not applied anymore after the refactor and the
end of animation callback was not correctly set.
Bug: 228202811
Test: com.android.server.wm.BackNavigationControllerTests#backNavInfo_HomeWhenBackToLauncher
Change-Id: Id4635f2988a23ad7e214fb37739ba8dc2c394ee4
Main motivation is to store a back callback's exact priority value in WM. This is required by the IME migration (ag/17076160) to compare the priority levels of IME window callback and focused window callback in BackNavigationController.
This also consolidates the WindowState#mSystemOnBackInvokedCallback and WindowState#mApplicationOnBackInvokedCallback fields into one field, as tracking two fields for one callback was error prone. We had to remember to clear the application / system field when the other field is set, and failing to do so has resulted in bugs such as b/222675481.
Bug: 224856664
Test: atest BackNavigationControllerTest
Test: atest WindowOnBackInvokedDispatcherTest
Test: m -j and test back behavior throughout the system on apps that
opted in and out.
Change-Id: Ic57113610d934f33d2c9ca4cef59f39a9b87e832
MANAGE_CLOUDSEARCH is and will be wanted by many applications. privileged/role should be able to cover most use cases.
Bug: 227041245
Test: atest
Change-Id: Idfb85f9e181968df935c4f27490bae2babadd0d9
- Better logging
- Remove method form OnBackInvokedDispatcher interface (but not the
implementation)
- Add test for ag/17211289
Bug: 222675481
Test: BackNavigationControllerTests#testUnregisterCallbacksWithSystemCallback
Change-Id: I6a3ae30da7e29dddda116acd8fa0c9af4b9faa6e
There might be other leaf tasks beside the top leaf task are not
occluded. Like the case when split screen is activated, both top leaf
tasks in each side are not occluded. This makes sure to include them
into remote animation target.
Fix: 227307495
Test: pass existing tests
Test: enter split, quick switch back and forth won't crash
Change-Id: Ic8af56b1a9e029dc4f1c2c57ab59071aab2924b4
