Rather than passing pointers in to get output from the functions, return
values from getPosition() and getBounds().
Bug: 245989146
Bug: 21566609
Test: Build, presubmit
Change-Id: Ie5b44a433c94b6c455486922f6894ccde1f1b127
PointerController is the source of truth of the mouse cursor position.
When the VirtualDevice APIs queried the cursor position, we were
previously getting the position from WM. WM's values only update once
system_server's global monitor has processed the input event, which
leads us into a race condition.
Remove the race consition by querying the mouse cursor position directly
from PointerController.
Bug: 266687189
Test: atest VirtualMouseTest
Change-Id: Id0e0e20a0d547f969d3a27d43fdfdca34d0fa7c0
PointerIcon.TYPE_NOT_SPECIFIED will be used for requesting default
pointer.
For the type, MouseCursorController will decide whether to show mouse or
stylus icon, based on the active source.
Need complement: the resource for default stylus icon should be added.
Test: Manual Test(hover pointer on handwriting area)
Bug: b/215436642
Change-Id: I6a337cb69bea57427f676e561900802270d206ae
Update local variable names in PointerController so they match the
general code style.
Bug: 254277939
Test: compiles
Change-Id: I7f71e12d48cf76ec8ff8fbaa435c18f9c2277096
Just a simple dump to start, but start pushing more PointerController
information into bugreports.
Bug: 254277939
Test: manually inspect `dumpsys input`
Change-Id: I7ca637aa8a8f1cc1188eddcf694f7c4aea7d5821
Previously, when InputManagerService requests for PointerController to
change the pointer display, there was no way to know when the request
was completed or whether it succeeded. This could lead to a few issues:
- WM's MousePositionTracker's coordinates would not be updated until the
next mouse event was generated, meaning the position would be out of
sync.
- The creation of a virtual mouse device moves the pointer to a specific
displayId. In order to test this behavior, we would need to sleep in
the test code to wait for the system to update the pointer display and
position, resulting in generally flaky tests.
Here, we add a way to synchonize changes to the pointer display so that
InputMangerService can know the current pointer display with certainty.
PointerController, which is updated in the InputReader thread, is the
source of truth of the pointer display. We add a policy call to notify
IMS when the pointer display changes.
When the pointer display is changed, the cursor position on the updated
display is also updated so that the VirtualMouse#getCursorPosition() API
is synchronized to the pointer display change.
Bug: 216792538
Test: atest FrameworksServicesTests:InputManagerServiceTests
Test: atest PointerIconTest
Change-Id: I578fd1aba9335e2e078d749321e55a6d05299f3b
A strong pointer to PointerController::DisplayInfoListener is given away
when the listener is registered, so PC cannot guarantee that the
listener is destroyed when it is destroyed. This means the listener can
outlive PC, so there is a race condition between PC's destruction and an
update to the listener.
While it could be argued that it is the caller's responsibility to
ensure that the listener is not updated after it is unregistered, there
is no way to guarantee that using strong pointers. Here, we can be
defensive and protect against this case.
Bug: 212672261
Test: atest libinputservice_test
Change-Id: I358a725980cc8c7d6ad0483a9b2a8b8715a03424
02b0545d02
Changes since the first time it landed:
- Unregister the WindowInfosLisntener in PointerController's destructor.
Bug: 188939842
Bug: 144544464
Bug: 206817973
Test: forrest run - CtsHardwareTestsCases
Change-Id: I92a3f128545e73c85d2a5079ee914e2f890c4308
Revert "Change PointerController to display space"
Revert submission 16194643-pointer-controller-in-display-space
Reason for revert: b/206817973
Reverted Changes:
I764c070ad:Change PointerController to display space
I5e9e19c36:Change PointerController to display space
Change-Id: I615d343968b818f498e905bab7963106b4e62651
PointerController used to work in the logical display space, so
TouchInputMapper and CursorInputMapper would need to transform the
coordinates before interacting with it.
This CL makes PointerController work in the display space. It will
transform incoming and outgoing coordinates to stay in the display
space using the DisplayInfo provided by SurfaceFlinger. Using info
provided by SF also means that there will be better synchonization
between the pointers and display changes like rotation.
Bug: 188939842
Bug: 144544464
Test: manual: ensure mouse and touch spots work in different display
orientations and sizes set using "adb shell wm size"
Change-Id: Ic2e05f06c70f4aaf5c104af9c9723e48c545de05
Change-Id: I5e9e19c3678766985ca2193cfe045a11f812fa2b
Modified current animation logic to use callbacks from the controllers
to further clean and modularize code.
Test: Pixel 3XL device, atest PointerController_test, compile
Change-Id: I1073bd78687cca491663c0349751dab4b30aa8e2
Abstracted logic that applies to all pointer types into general
PointerController class and moved implementation of logic specific
to the mouse cursor and touch spots to MouseCursorController and
TouchSpotController, respectively.
Test: Pixel 3XL device, atest PointerController_test, compile
Change-Id: Ia5825c37ca75951cc8bcd7d5102c986bd957e69f
This change makes it so that PointerController does not ask its Policy
to load any resources for any displays until a DisplayViewport is set,
and verifies this with unit tests.
Bug: 145699789
Bug: 146385350
Test: atest libinputservice_test
Change-Id: I2e48e7ac4700e6f9fdf939a7bd0e6639b051ade6
Merged-In: I2e48e7ac4700e6f9fdf939a7bd0e6639b051ade6
When show taps enabled, the presentation would be
PRESENTATION_SPOT for drawing the spots. That would
ignore loadPointerIcon when reload resources.
Always load default icon when reload resources.
Bug: 115395392
Test: Enable show taps and use mouse,
enabled large mouse pointer manually.
Change-Id: I74f2bf33c9046e8eaba7d284ca16366da6e3d221
Currently the PointerController would be associated with the top most
freeform or external display if enable desktop mode.
But for presenting spots when enable showTouches from develop options,
they should be shown on the corresponding display where user touched.
Also changed Vector to std::vector.
Test: atest inputflinger_tests
Test: Use device support multi-display and touch screen (like mojave).
Enable showTaps from develop options and check if show tap spots.
Bug: 120815589
Change-Id: I0854fa97813aabf005dccb8ec6fbda10106126e9
When mouse pointer changes displays, it should also reload the
icon from new resources. Otherwise, if the densities of the
previous and new displays are different, the size of the pointer
sprite will look too small or too large.
- Add getDisplayContext to get the corresponding Context by displayId.
- Cache system pointer icons per display, clear if display removed.
- Fix icon moved to default when not resetting out of task bound.
Bug: 113559891
Test: Enable mouse pointer on default display, move to other screen
Change-Id: Ic42d0ec32d9c979281e13c83b9e8b57134fd4f0d
* Add explicit to conversion constructors/operators
Bug: 28341362
Test: make with WITH_TIDY=1 DEFAULT_GLOBAL_TIDY_CHECKS=-*,google-explicit-constructor
Change-Id: I31d45e4c6ff9fa69801a14c341f2da66baf565a0
Limit mouse boundary to one display viewport.
Currently if desktop mode is enable, selects the topmost freeform display,
or find first external display if there is no freeform display.
Return the default display if desktop mode is not enable.
- Add getPointerDisplayId to find the preferred display.
- Limit update viewports from InputReader.
- obtainPointerController would also find the associated display id.
- Reload cursor resources if resolution or display changed.
Bug: 113559891
Test: atest inputfliger_tests
Test: ActivityView test app
Change-Id: Iffd01e27f5010ccfb0481a028658b7290ca6316d
Limit mouse boundary to one display viewport.
Currently if desktop mode is enable, selects the topmost freeform display,
or find first external display if there is no freeform display.
Return the default display if desktop mode is not enable.
- Add getPointerDisplayId to find the preferred display.
- Limit update viewports from InputReader. Also implements updatePointerDisplay
to find and associate with the specified display viewport.
- Reload cursor resources if resolution or display changed.
Bug: 113559891
Test: atest inputfliger_tests
Test: ActivityView test app
Change-Id: Ief3410a517be457dfe6f1180342bb1dc88833635
These pragma statements used to guard against compiler warnings in Sk..
code. They are no longer necessary.
Test: make
Change-Id: I9f7f88276edb546ba08c7344c142eb487a04f978
The loop as constructed causes i to overflow twice when i = 0 on integer
overflow sanitized builds.
runtime error: unsigned integer overflow: 0 - 1 cannot be represented
in type 'size_t' (aka 'unsigned long')
runtime error: unsigned integer overflow: 18446744073709551615 + 1
cannot be represented in type 'size_t' (aka 'unsigned long')
This refactors the loop to avoid the overflows.
Bug: 30969751
Test: Compiles, device boots, enabled pointer location overlay.
Change-Id: I844bb3b84b1f536c50d06fb489fcc22590d4aa98
Currently PointerController starts listening to display events
immediately (in its constructor) and never explicitly removes
the callback. The reference dangling from the looper
prevents the PointerController instance from being deleted
when all the clients have released their references.
As a result, when USB or BT mouse is disconnected,
the mouse stays frozen on screen and only goes away
after a 15 sec inactivity timeout.
This change introduces an intermediary LooperCallback
which holds only a weak reference to PointerController.
The pointer now disappears immediately upon mouse
disconnect.
Bug: 30824220
Change-Id: I5f7208dbfa381b3e21f248cc0da402f307faa184
The underlying implementation needs to be completely rethought. If a
process crashed while you were in pointer capture mode, you were
pretty much stuck in it. If the mouse happened to move outside of
your bounds right before you called the API, you'd never actually get
an event (whatever it was hovering over would). There's no easy way
for the system to tell you when you enter or exit this mode because
it doesn't actually track who the current request is from.
These are all solvable, but not in the N time frame. Maybe next time.
Bug: 26830970
Change-Id: I899649594c7ca8db5962fcdaa71c5b21d3aa42c3
The underlying implementation needs to be completely rethought. If a
process crashed while you were in pointer capture mode, you were
pretty much stuck in it. If the mouse happened to move outside of
your bounds right before you called the API, you'd never actually get
an event (whatever it was hovering over would). There's no easy way
for the system to tell you when you enter or exit this mode because
it doesn't actually track who the current request is from.
These are all solvable, but not in the N time frame. Maybe next time.
Bug: 26830970
Change-Id: I03efd63c499b86dc278491ca3284566c1965581f
This patch does the following things:
- add new graphic assets of large icons for accessibility needs
(imported from ChromeOS data)
- add the logic to choose the set of normal icons or accessibility icons
- make InputManagerService to observe the settings change, so that
it can reload the new resources
This patch misses non-1x graphic though, because ChromeOS doesn't
have such data.
Bug: 25778857
Change-Id: Ia5f95d47f50b3f6eea555c3af8069bc6bae0b400
Right now, it only supports I-beam on EditText, but further
rules will come in the future.
The png files for the icons are from chromium.
Bug: 24180385
Change-Id: I8de4ec8a5412b4830c08aa232c5083841c5c751c
Refactored the directory structure so that services can be optionally
excluded. This is step 1. Will be followed by another change that makes
it possible to remove services from the build.
Change-Id: Ideacedfd34b5e213217ad3ff4ebb21c4a8e73f85
