- Don't use reverse-JNI to access metadata ptr, pass it as an arg instead
- Use @FastNative since the calls are short and bounded in time
Performance improvement:
- On a 10-second trace of camera app running on sargo, percentage of
time used in CameraMetadataNative.get went from 4.36% to 3.77%, a 15%
reduction in time.
Test: atest CtsCameraTestCases
Bug: 150214459
Change-Id: I28d9428beaa7eada6292e24fe6ca1dbd9c2ff153
Previously, the client won't send the modified insets state to window
manager if the dispatched state and the local state are the same. The
following case can make the requested insets state at the server side
stale:
1. Window A requests an insets source to be invisible while having
the control.
2. Window A loses the control.
3. The new control target requests the insets source to be visible.
4. Window A receives the new insets state from server.
5. Window A also requests the insets source to be visible while not
having the control.
6. Window A gain the control.
7. Window A won't send the new requested state (visible) to window
manager because the local state and the dispatched state are the
same.
8. Window manager keeps assuming that window A is requesting the insets
source to be invisible which is incorrect.
This CL stores what insets state is sent to server, and updates/sends
the state to server while gaining the control if the requested state is
changed.
Fix: 146964271
Test: Manual.
Change-Id: I6ee533d9316c769faef539bfb980197c351ee8d1
We now infer the PIP/MW mode change from the new configuration.
Note also that both
- Activity#isInPictureInPictureMode
- Activity#isInMultiWindowMode
infer the current state from the configuration rather than querying
against the WM.
Also in this CL:
- When in removePinnedStackInSurfaceTransaction, keep the pinned stack
hidden till the windowing mode is set to fullscreen, this is to surpress
the attempt to set the activities to be started in reparenting
- When in ActivityRecord#shouldBeVisible, should take account the force
hidden flag, which is not actually in use before
Bug: 144097203
Bug: 142282126
Bug: 138329093
Test: atest ActivityLifecyclePipTests \
ActivityLifecycleSplitScreenTests \
ActivityLifecycleTopResumedStateTests \
PinnedStackTests \
SplitScreenTests \
ActivityTaskManagerServiceTests \
RecentsAnimationTest \
AssistantStackTests \
StartActivityTests \
ActivityVisibilityTests \
MultiDisplaySecurityTests \
MultiDisplaySystemDecorationTests
Change-Id: Ibe032b5e50ba5c6d6bc44ebb54d07ac974ebe656
This commit adds support for validating and filtering IPsec algorithms.
Without a public API exposing IKEv2 algorithms (and their respective
public APIs), the allowedAlgorithms can only filter the proposals for
IPsec (Child) SA algorithms.
Additionally, this removes the HMAC_SHA1 from the IKE SA's integrity
algorithm proposals due to insecurity
Bug: 153701879
Test: FrameworksNetTests passing, new tests added
Change-Id: I7e61a1612692db275b751330af5bacbf86836a8c
Merged-In: I7e61a1612692db275b751330af5bacbf86836a8c
(cherry picked from commit 94e1c08a9a)
attach() may throw various kinds of runtime exceptions, and since there
is no way to represent them in attach() other than returning null (it
is not supposed to throw), we should catch them here and log them.
Bug: 153332815
Change-Id: Icbb7643c3abf6c6b0edc9d1eb9a74c38c731acdf
An earlier CL with benchmarks has shown that sending strings as UTF-8
is 50% faster for US-ASCII strings, and still 68% faster for complex
strings referencing higher Unicode planes. (So an improvement in
both cases!)
Since code across the OS still makes heavy assumptions about Parcel
strings typically being UTF-16, we need to carefully migrate
Parcelables by hand, which is what this CLs begins doing.
Bug: 154436100
Test: manual
Change-Id: I9a675473e0ce3a4a5c5e305dd851b40bb1560e1c
This is something clients will need if they register to listen to
DisplayAreas for several features. This will help distinguish the
DisplayAreaInfos.
Test: Builds
Bug: 152114574
Change-Id: If1d14aee070388274afb6a43bc1c631d79fb4b38
Before, CTS test used WindowManager#shouldShowSystemDecors() to set show
system decorations after overlay display launched, which wouldn't enable
system decorations without relaunching. Now, we can set those flag
before display launch similar to what VirtualDisplayHelper does.
Bug: 153624902
Bug: 154158053
Test: atest CtsWindowManagerDeviceTestCases:MultiDisplayClientTests
Test: atest CtsWindowManagerDeviceTestCases:MultiDisplaySystemDecorationTests
Change-Id: I83c21a74c722a23ac77141305a08d9f67c5bbcf0
If the WindowContainer was revoked from a registered organizer, the
client could still call getLeash to system server and control the leash
for the WindowContainer. Instead, pass the leash back to the client in
onTaskAppeared and onDisplayAreaAppeared. Once the WindowContainer is
revoked from the client, the leash will reference the old
WindowContainer SurfaceControl and will not be able to control the
WindowContainer anymore.
Test: Split screen
Test: DisplayAreaOrganizerTest
Test: WindowOrganizerTest
Bug: 154558563
Change-Id: I1f6eb987a2a3fecfef912a3009ee52989c85ff4b
Fixes an issue where the WindowInsetsController was only available after
the first traversal.
Bug: 155280356
Test: atest WindowInsetsControllerTests
Change-Id: I225efcaa31e97da9e8ebc6bd7d24c433c8e80826
Such that all it's goodness can also be used in context when
ViewRootImpl isn't available, like the SystemUI controller used
for Car and Split
Test: InsetsControllerTest
Fixes: 154631128
Change-Id: I54a3f8a34810472d9273e4627a7811b7abd0863f
Update the APIs documentation and add test cases
Bug: 152316540
Test: atest InlinePresentationStyleUtilsTest
Change-Id: Icff5d5b36fecd94ceb7db5446eefa16b4b1aac3b
This is a temporary failure, does not fail the session, but requires
caller to re-commit. E.g. there are connectivity issues which can be
fixed later.
Bug: b/153874006
Test: atest PackageManagerShellCommandTest PackageManagerShellCommandIncrementalTest IncrementalServiceTest
Change-Id: I02791a2963130dbecb510c4a7cafcf04f6245761
