See accompanying frameworks/native commit
"SurfaceFlinger: Add parent-less relative layering" for a full explanation.
Test: Manual of bug repro steps. Plus tests for new SurfaceControl functionality included in frameworks/native.
Bug: 36693738
Change-Id: Ic54598117c1f44a206d33f03d0cc463fbef43fcc
* changes:
Fix broken CTS tests.
Draw a default focus highlight if needed.
Add a public API defaultFocusHighlightEnabled.
Detect unhandled keyboard focused state.
A more targeted version of "ViewRootImpl: Fix child lifetime".
In this version we avoid hijacking the window visibility callback
and introduce a new path.
Our contract has been that at any time after returning from
onStop, the WindowManager may destroy the clients Surface. We
see in setWindowStopped, we set the stopped state on the hardware
renderer in order to prevent further use of the Surface. However
there is nothing synchronously dispatching this change to child views
and so SurfaceView is not necessarily informed to release it's Surface
in time. Typically SurfaceView will be informed through the
onWindowVisibilityChanged callback. The signal producing this
is emitted from SystemServer prior to onStop but has to pass
through the clients handler thread. It seems this has always been
broken, and we have always had intermittent reports of EGL_BAD_ALLOC
scenarios. I speculate that elevation of WindowManager scheduling during
app close scenarios is perhaps making this more severe, as it seems to
ensure the WindowManger will win the race to destroy the Surface before
the client handler thread process onWindowVisibilityChanged (unless
the FIFO timeout is surpassed).
Test: Put Chrome in PiP. Turn screen off. No Crash!
Bug: 36561071
Change-Id: I9233ba8151c7f577b1d1044afc0c2ac3a65be4b4
The default focus highlight should be necessary when the view is
attached to window.
Test: cts-tradefed run singleCommand cts --skip-device-info
--skip-preconditions --abi armeabi-v7a -m CtsViewTestCases -t
android.view.cts.ViewTest
Test: cts-tradefed run singleCommand cts --skip-device-info
--skip-preconditions --abi armeabi-v7a -m CtsWidgetTestCases -t
android.widget.cts.MediaControllerTest
Bug: 37082416
Change-Id: I5d6b0c00b5ba03c8f62e4f71be9336823a73134b
This commit draws a default highlight for a focused
View if it's detected to have no state_focused defined
and its useDefaultFocusHighlight attribute is true.
When we detect a default highlight is needed, we show it
on top of the view to. Once we detect that it's no longer
needed, we remove it.
Test: Check that views without a focused_state in its
state spec have a default highlight when they get focused.
Bug: 35096940
Change-Id: Ifbe4bb9e1297d98845314e24d8b758f14e5987a9
This commit adds a public API to turn on/off default
focus highlights. When a focused View is detected to
have no state_focused defined, we can use this API to
decide wether we should show a default highlight for it.
Test: cts-tradefed run singleCommand cts --skip-device-info
--skip-preconditions --abi armeabi-v7a -m CtsViewTestCases
-t android.view.cts.View_DefaultFocusHighlightTest
Bug: 35096940
Change-Id: I76b45d6bf5761641a0ed7f4d0b04cb325ed72b52
We now have a software feature for autofill which can be used
by partners to disable it on low-end devices or form factors
for which autofill doesn't make sense.
bug:35956220
Test: manual (requires a custom build)
Change-Id: I6c06462ed9ca3ae93331700dce38a8c08dfd0722
This reverts commit ba29dbc3b3.
The original commit causes apps such as phone and clock to not wake the
device.
Fixes: 37197491
Change-Id: I1b45cafbcbf12991d8e1e716e4d1751cd077c556
Also:
- Give the session an integer ID as the activityToken is not stable over
restarts of the activity
- Verify that session is only accessed by one UID
- stabilize AccessibilityViewIds over activity lifecycle at least for
the IDs we can do that. This required to split the ID namespace in
"per-app" and "per-activity" views. Only the later ones can be
restored.
- Do not end session when app is killed (as it can be restarted)
Bug: 35484143
Fixes: 36392498
Test: cts-tradefed run cts-dev -m CtsAutoFillServiceTestCases --test=android.autofillservice.cts.SessionLifecycleTest
cts-tradefed run cts-dev -m CtsAutoFillServiceTestCases
Change-Id: I229acc1b3ce35fb57262da7d7466b5d4328b49d4
AdapterViews (and other dynamic layouts) inflate views which
can lead to multiple views with the same id. The original
user-specified focus logic assumed that id's were unique causing
focus loops. This now uses the full "insideOut" id-search
instead of an id->view map.
Bug: 32647147
Test: FocusFinderTest still passes. Added #testDuplicateId
specifically for this scenario
Change-Id: Iaed98438f5ad70c866dd72e699453eab0ac0cf58
As far as I can tell this has always been broken. We've always had
intermittent reports of buffer-queue-abandoned as well (a recent bug
came with some reports from N). During onStop SurfaceView relies on
onWindowVisibilityChanged, to trigger a visibility change. At this
point SurfaceView will emit the SurfaceDestroyed callback in order to
stop the client from further use of the Surface. The contract we've
been using with ViewRootImpl is at any point following
Activity.performStop returning the WindowManager was free to destroy
the Surfaces. This is why in setWindowsStopped we destroy the hardware
resources for the ViewRoot. However we aren't dispatching anything to
the SurfaceView. The WindowManager will send an app visibility
notification, but that would go through the handler. This means by the
time we return from Stop, there is no guarantee that the
onWindowVisibilityChanged callbacks have been invoked at all. It
seemed most sensible to dispatch the visibility callbacks directly. We
also ensure that getHostVisibility will return false after this point,
so that performTraversals will not reverse our visibility request if
it occurs again prior to the window visibility notification from the
WindowManager. We also guard against emitting a second window
visibility changed callback in the traversals. I don't know at this
point what value the window visibility notification provides but I
don't feel excited about removing it in this CL at this point in
the development cycle.
Test: Put Chrome in PiP. Turn screen off. No Crash!
Bug: 36561071
Change-Id: Id1673561b2299d477b2761b3ac6afa14eabbf7fb
Fix a bug where a malformed Parceled representation
of an AccessibilityNodeInfo could be used to mess with
Bundles as they get reparceled.
Bug: 36491278
Test: Verified that POC no longer works, a11y cts still passes.
Change-Id: I10f24747e3ab87d77cd1deba56db4526e3aa5441
(cherry picked from commit 687bb44b43)
This partially reverts CL I0f6b130a7df57557e40b52a7b7ac00be965a17c3.
Fixes: 34133139
Test: Confirm that unit test passed.
adb install out/target/product/.../data/app/FrameworksCoreTests/FrameworksCoreTests.apk
adb shell am instrument -w \
-e class android.view.inputmethod.InputMethodInfoTest \
com.android.frameworks.coretests/android.support.test.runner.AndroidJUnitRunner
Change-Id: I2b94263b2a0bf330bf9973567fea2d35176754d7
1) Switch AccessibilityButtonChooserActivity to non-exported
2) Update AccessibilityButtonChooserActivity's intent action to
be CTS-compliant
3) Switches to correct permission to validate a click of the
accessibility button originated from the status bar service
Bug: 36866328
Bug: 36976440
Test: Manual / Created app to validate launch protection
Change-Id: Ifcd5e8dbbe683dc99bb900a5870a9ba374fbd5d4
It happened when views provided an Accessibility node that had children.
Bug: 37078783
Fixes: 37009008
Test: LoginActivityTest#testAutoFillWhenViewHasChildAccessibilityNodes
Change-Id: Ie694094afe9129b933096a385eb867baf4550e7f
Also clear this overlay once the data changes. Some views call
notifyValueUpdated even if the value did not change. Hence remeber the
data that was autofilled and check if the data really changed.
Make the autofilledDrawable publicly accessible so that virtual views
can use the same UI.
As the activity is not aware that the views were autofilled the state
has to be explicitly persisted over activity lifecycle events as part
of the view. The lifecyle does not work yet as the AutofillIds are not
stable over activity lifecycle events.
Fixes: 34946006
Test: Autofilled and changed all supported views.
cts-tradefed run cts-dev -m CtsUiRenderingTestCases
Change-Id: I58b105bb221f1b6466dd136dfe5062d3babb5aa8
Moving tests for accessibility manager into the a11y
directory and getting them to use some of the newer
test infrastructure. Minor tweaks to
AccessibilityManager to make it testable.
Deleting the AccessibilityManagerService tests entirely.
This class is difficult to unit test with the current
infrastructure, but is covered in CTS tests to a large
degree. I wasn't able to get these working quickly, and
they rely on timeouts that make them run much slower than
the other tests. They also throw exceptions when their
mock accessibility services receive accessibility events,
which causes the system to blacklist those services, causing
the tests to fail differently until the system is reset.
I think it's better to focus on writing new, more robust
tests than on rewriting these tests to pass.
Bug: 36614219
Test: A11y unit tests now work.
Change-Id: I8effe44f22110d55ebb700dc46f59b7e7a1aa946
(cherry picked from commit 9213d23178)
We need PRIVATE_FLAG_NO_MOVE_ANIMATION to get consistent
results from client initiated window movement in CTS.
Bug: 36865695
Test: Annotations only.
Change-Id: I7850ce3920ba1e89c7759bf9bc25f25731f568a7
In order to clear the measure cache, we need to requestLayout
when updating the child layout params. To see why, consider the case of
a Frame or Linear layout which will measure different heights
depending on the (top/left/right/bottom)Margin parameters of it's
childrens layout params. Now imagine the following sequence of events:
1. We request a layout on the FrameLayout
2. We measure the FrameLayout and place a value in the cache.
3. Now we update the margin parameters on one of the frame layouts
children. Because the parent already has a layout requested
we don't call parent.requestLayout (see View.java#requestLayout),
and thus the parent measure cache isn't cleared.
4. Now we measure the frame layout again and we incorrectly
used the cached value.
Calling to requestLayout when the child layout params
change clears the cache properly. If the child didn't
call request layout from it's own relayout, it must mean that
a layout was already pending (step 1 in the sequence),
and so no more work should be triggered besides clearing the cache.
Bug: 33095565
Bug: 33308065
Bug: 34388764
Test: Manual case in bugs.
Change-Id: I9148f32530588e4dc859297f9658f506b38e72f0
There are two separate overrides for display metrics in DisplayManager
and WindowManager:
- In DM - LogicalDisplay#mOverrideDisplayInfo, in most cases not null.
- In WM - DisplayContent#mBaseDisplayWidth/Height/Density, different
from #mInitialDisplayWidth/Height/Density values when some metrics are
forced.
When display was resized its windows weren't updated because of
two problems: old LogicaDisplay#mOverrideDisplayInfo was preventing
WM from detecting the change and override (base) display metrics were
never updated by resize.
When display size changes:
- Before this CL:
DM receives DISPLAY_CHANGED event, it updates internal values and
WM is notified about them with a message. In most cases there is an
override obtained from WM and WM doesn't get new values from
LogicalDisplay#getDisplayInfoLocked().
- With this CL:
WM will requests real updated values from DM without any overrides
and will decide whether to apply them or not: if there is no override
in WM - it will apply values from WM, otherwise it will keep the
override. Also it will always update initial display metrics if there
is a real change detected.
Bug: 35258051
Bug: 34164473
Bug: 36518752
Test: android.server.cts.ActivityManagerDisplayTests
Test: #testDisplayResize
Test: #testForceDisplayMetrics
Change-Id: I2495c27797f11f9aaee4ea06648a8ccd29ac5b62
