Saves about 300us for each AccessibilityEvent sent, which
translates to 1 ms in gfx-avg-frame-time in the
QuickSettingsPull.
Test: Ran a11y cts and unit tests. Ran systrace and the
QuickSettingsPull to measure jank improvement.
Change-Id: I6d219b0e82c21047e50d402d930d8a5cc93fa6ad
Not needed anymore \o/. Also fixes a flicker when transitioning
between two activities that both set FLAG_LIGHT_STATUS that was
somehow introduced recently.
Test:
- Start an light status bar activity from another activity that
already has light status bar, ensure there is no flicker.
- Open IME and make sure the content gets resized like before.
Change-Id: Ie9c9e1cd40f909c449d36ae436250063af20539e
Test: Enable in SysUI tuner, drag PIP. This is only experimental to help
figure out what UX we want to keep.
Change-Id: I0d6f2f0c5909d6a76aae4a8fb84c5076f6996fdd
Also speed up RenderNode creation:
Use finalizer() instead of NativeAllocationRegistry, this
shaves ~3us off of creation currently
Avoid instanceof, instead have SurfaceView explicitly ask
for updates.
Remove unused method call.
Test: ran benchmarks
Change-Id: I3117fdf72313a4e6a9965baca9f2a8b855c19b34
Switched presentation feature to use new window TYPE_PRESENTATION
and also add its own window token to the display the presentation
is running on. This is needed as window manager no longer allows
tokens to have windows on multiple displays.
Bug: 32566372
Test: Presentation mode works.
Change-Id: I9c2998311b65640743b8e23ec4f10bf1ffbfd785
Window tokens can now only be on one display, so we now require clients
that want to add/remove window tokens to specify the display they would
like the token to be created on. This simplifies the token handling code
in WM and will be useful moving forward for clients that want to add
windows to external displays.
Test: Existing tests pass
Change-Id: I6b2d8d58a913b3624f1a9a7bebbb99315613f103
* changes:
The big keyguard transition refactor (6/n)
The big keyguard transition refactor (5/n)
The big keyguard transition refactor (4/n)
The big keyguard transition refactor (3/n)
The big keyguard transition refactor (2/n)
The big keyguard transition refactor (1/n)
Nuke KeyguardScrim
Test: Kill SystemUI while lockscreen is showing, make sure nothing
is visible when being killed.
Bug: 32057734
Change-Id: I9f8d1e5a0e0f968460d8170627a849623c6a7245
Notify activity manager when dreaming showing state changed so
KeyguardController can update the occluded state when the device
is dreaming.
Test: Set dreaming while charging, wait until screen times out,
make sure that dream is occluding Keyguard.
Bug: 32057734
Change-Id: Ied6f485d9b4a1526cb4cd5f0701f86b1ea05830a
Test: Manual, ensure that the accessibility focus is still visible on
a moved window (enable talkback, focus a view, then move the
window).
Change-Id: Ifd67afe355cac2f1df01ddd16f960282f4d36132
The heart of this change are two things:
1) Instead of using the force hide mechanism to hide windows behind
Keyguard, we actually make the activities invisible in activity manager.
2) When Keyguard is going away, we change the visibilities in activity
manager and run an app transition.
At the very core we move the responsibility of hiding activities to
ActivityStack, which checks whether Keyguard is showing, and then
hides all non-show-when-locked activities. For that, we need to check
whether any window of an activity has SHOW_WHEN_LOCKED set. We
introduce a callback from WM -> AM in case these Keyguard flags have
changed.
Furthermore, we decide whether to occlude Keyguard in KeyguardController,
which just checks whether the top activity has SHOW_WHEN_LOCKED set. When
this state changes, we prepare an occlude/unocclude app transition, and
in PWM we just inform the Keyguard about the animation so SysUI can play
along this animations in a mostly synchronized manner.
Since we now use an app transition when unlocking the phone, we get
lockscreen launch animations for free - window manager automatically
waits until the activity is drawn, or directly executes the transition
if there is nothing to animate. Thus, we can remove all the infrastructure
around "waitingForActivityDrawn".
The logic to show/hide non-app windows is moved to policy, and we add the
ability to run animations on non-app windows when executing an app
transition.
Test:
1) runtest frameworks-services -c com.android.server.wm.AppTransitionTests
2) Manually test unlocking Keyguard:
2a) Without security
2b) With security
2c) With security but trusted
2d) Portrait while activity behind is in landscape
3) Test launching things from Keyguard
3a) Without security
3b) With security
3c) Launch camera without security
3d) Launch camera with security
3e) Launch camera with securtiy and trusted
3f) Launch voice affordance
4) Set no notifications on lockscreen, drag down, make sure you get
the correct animation
5) Test clicking "emergency" on bouncer
5b) Test "Emergency info" on emergency dialer
5c) Test clicking edit button on emergency info, should show pattern on
Keyguard
Bug: 32057734
Change-Id: Icada03cca74d6a612c1f988845f4d4f601087558
- Creating a PinnedStackController to keep track of the state of the PIP
to prevent changes in the system (ie. IME showing) and user interaction
from clobbering each other.
- Refactoring calls in AM into WM/controller
Test: android.server.cts.ActivityManagerPinnedStackTests
Change-Id: Ie59dfd45d5c54764ba69a589b3b8148845e92cc3
Signed-off-by: Winson Chung <winsonc@google.com>
Test: docs only, no test apart from verifying that it builds
Bug: #32158219 clean up InputConnection.commitContent() javadocs
Change-Id: I94ca33a34dfaece8db912367d2290db67a4a038a
- Ensure we use the right display size when calculating PIP bounds.
- Also update interface to take the display id.
Test: android.server.cts.ActivityManagerPinnedStackTests
Test: #testPinnedStackDefaultBounds
Test: #testPinnedStackMovementBounds
Change-Id: I01fd8ba6dee212c29a9a092673ee8f7843e41af6
Now display-specific settings, such as dimensions and orientation,
are stored in display override config. For default display it is
mirroring the global config. Each time when global config is updated,
override of the default display should be updated too and vice versa.
Test: Existing and manual tests still pass.
Change-Id: Ic6c2190092d328820f314a05bed43c875db18170
For N the Camera used ROTATION_ANIMATION_JUMPCUT
which was reinterpreted as seamless rotation if possible.
However this ended up being non ideal in app transition
scenarios where the rotation animation replaces the app
transition. Here we also can't seamlessly rotate, so the
window just suddenly appears with jumpcut. In fact we want
crossfade as a fallback animation and this is what distinguishes
ROTATION_ANIMATION_JUMPCUT and SEAMLESS. The problem was patched
over in N by having double-tap gestures fire camera intents
with a SEAMLESS ActivityOptions.rotationAnimationHint,
but it was too late to un-hide for consumption by apps. While here,
clean up some documentation that doesn't really parse.
Test: Manual, try the RotationAnimation activity in ApiDemos sample.
Bug: 28838855
Bug: 28823590
Change-Id: I17b5e9fa967df8ade0943337cd7b6c3956d3a499
on a per RenderNode basis. With Skia renderer we
see 30% speed improvement in Invalidate Tree UI Jank test,
when SkiaDisplayList objects are reused.
Test: manually built and run on angler-eng.
Change-Id: Ie4ec50ddb2015150e3ec678dde7ebed0c8d90067
All draw* calls in Canvas are regular JNI
All draw* calls in DisplayListCanvas are FastNative
Unifies Canvas JNI on nMethodName naming
CanvasPerf results before:
INSTRUMENTATION_STATUS: basicViewGroupDraw_min=12492
INSTRUMENTATION_STATUS: recordSimpleBitmapView_min=13912
and after:
INSTRUMENTATION_STATUS: basicViewGroupDraw_min=11945
INSTRUMENTATION_STATUS: recordSimpleBitmapView_min=13318
Test: refactor, makes & boots
Change-Id: I06000df1d125e17d60c6498865be7a7638a4a13e
1. Taking into account scroll state
2. Drawing focus also for views with edges.
Bug: 32099818
Test: Manually ensure that that button draws focus
Change-Id: I085db4deb1561d8b480e4012018d48bc83a7f81f
- This CL provides the framework for manipulating the pinned stack using
an input policy (to be determined later) provided by the SystemUI.
Test: android.server.cts.ActivityManagerPinnedStackTests
Test: #testNonTappablePipActivity
Change-Id: I025c41fff26ed05a35d68e59f10330680ed11ea8
With current approach there is a high chance of making a programming
error and altering the state of singleton Configuration.EMPTY object,
because previously configuration was usually set by link to it.
This CL changes the behavior to create Configuration member objects
only once and alter their state afterwards.
Test: Manual and existing tests still pass.
Change-Id: Ie267c2fbf555f137670b7bd2cd251b817c69bc8e
NOTE: Linear blending is currently disabled in this CL as the
feature is still a work in progress
Android currently performs all blending (any kind of linear math
on colors really) on gamma-encoded colors. Since Android assumes
that the default color space is sRGB, all bitmaps and colors
are encoded with the sRGB Opto-Electronic Conversion Function
(OECF, which can be approximated with a power function). Since
the power curve is not linear, our linear math is incorrect.
The result is that we generate colors that tend to be too dark;
this affects blending but also anti-aliasing, gradients, blurs,
etc.
The solution is to convert gamma-encoded colors back to linear
space before doing any math on them, using the sRGB Electo-Optical
Conversion Function (EOCF). This is achieved in different
ways in different parts of the pipeline:
- Using hardware conversions when sampling from OpenGL textures
or writing into OpenGL frame buffers
- Using software conversion functions, to translate app-supplied
colors to and from sRGB
- Using Skia's color spaces
Any type of processing on colors must roughly ollow these steps:
[sRGB input]->EOCF->[linear data]->[processing]->OECF->[sRGB output]
For the sRGB color space, the conversion functions are defined as
follows:
OECF(linear) :=
linear <= 0.0031308 ? linear * 12.92 : (pow(linear, 1/2.4) * 1.055) - 0.055
EOCF(srgb) :=
srgb <= 0.04045 ? srgb / 12.92 : pow((srgb + 0.055) / 1.055, 2.4)
The EOCF is simply the reciprocal of the OECF.
While it is highly recommended to use the exact sRGB conversion
functions everywhere possible, it is sometimes useful or beneficial
to rely on approximations:
- pow(x,2.2) and pow(x,1/2.2)
- x^2 and sqrt(x)
The latter is particularly useful in fragment shaders (for instance
to apply dithering in sRGB space), especially if the sqrt() can be
replaced with an inversesqrt().
Here is a fairly exhaustive list of modifications implemented
in this CL:
- Set TARGET_ENABLE_LINEAR_BLENDING := false in BoardConfig.mk
to disable linear blending. This is only for GLES 2.0 GPUs
with no hardware sRGB support. This flag is currently assumed
to be false (see note above)
- sRGB writes are disabled when entering a functor (WebView).
This will need to be fixed at some point
- Skia bitmaps are created with the sRGB color space
- Bitmaps using a 565 config are expanded to 888
- Linear blending is disabled when entering a functor
- External textures are not properly sampled (see below)
- Gradients are interpolated in linear space
- Texture-based dithering was replaced with analytical dithering
- Dithering is done in the quantization color space, which is
why we must do EOCF(OECF(color)+dither)
- Text is now gamma corrected differently depending on the luminance
of the source pixel. The asumption is that a bright pixel will be
blended on a dark background and the other way around. The source
alpha is gamma corrected to thicken dark on bright and thin
bright on dark to match the intended design of fonts. This also
matches the behavior of popular design/drawing applications
- Removed the asset atlas. It did not contain anything useful and
could not be sampled in sRGB without a yet-to-be-defined GL
extension
- The last column of color matrices is converted to linear space
because its value are added to linear colors
Missing features:
- Resource qualifier?
- Regeneration of goldeng images for automated tests
- Handle alpha8/grey8 properly
- Disable sRGB write for layers with external textures
Test: Manual testing while work in progress
Bug: 29940137
Change-Id: I6a07b15ab49b554377cd33a36b6d9971a15e9a0b
Make sure ACTION_HOVER_EXIT is never sent to a
ViewGroup more than once.
While processing ACTION_HOVER_MOVE during exiting
a hover target never send it a ACTION_HOVER_ENTER.
Test: android.view.cts.HoverTest
Bug: 32071138
Bug: 32071098
Change-Id: Ibc1809137907176437a3a5a4ea14b00f1c629b58
Just the basic cleanup first, will follow-up with a more
comprehensive update later
Test: builds & boots, refactor no behavior change
Change-Id: Ie3a7144b5fefdfda085e2ab50da8b0e627a1ecb8
