- Jun 03, 2024
-
-
vignesh ramanathan authored
BUG: b/329826852 Test: Manual Change-Id: Ie07952be4c0e88236e79c21b3f07537beeaf91a8 Change-Id: I774db150e103c445c972e1f5e63ee73218d83392
-
- May 29, 2024
-
-
sandeepbandaru authored
Bug: 342548962 Change-Id: Ic43da26a50115a8d6340d33cc66d81a756aab1ee
-
Idriss Juhoor authored
There is an issue when the AccessibilityManagerService receives the onSomePackageChange callback before the UserSwitch signal. The service would be marked as initialized, but it would not load the services when the switch user signal would be later received. This patch ensures the onSomePackageChanged is handled only after the service is initialized. Bug: 340927041 Test: atest AccessibilityEndToEndTest Test: AccessibilityManagerServiceTest in presubmit Change-Id: I99fba0d08cb0ca5f2e24c104d1a7735109955683
-
Piyush Mehrotra authored
A previous change (ag/26748034) added this call for BMM logging. getPackageInfo is a relatively heavy binder call, which triggered performance regression alerts (b/342082681 and b/342273383). The additional getPackageInfo call resulted in 5-10% increase in latency for package commit operation. For the purposes of BMM logging in restoreAtInstall flow, we just need the name of the package. Fields like version code are already logged in subsequent BMM logging during restore. Bug: 342082681 Test: Verified that logging is correct Change-Id: I8025789b2c696eac1b7738a1ae332c67aea8ce5f
-
Hyoungho Choi authored
StorageManager.FLAG_INCLUDE_RECENT flag is used to get volumes mounted within a week. Unfortunately, however, it doesn't work as we expected. It makes to return volumes used a long time ago. Bug: 342962382 Test: Manual (cherry picked from https://partner-android-review.googlesource.com/q/commit:31a78ecd8cbfa234b8f46c475717e7e6030aa8a4) Merged-In: I4394ad3b04d89a55ecb5ab3e569571784f4c619e Change-Id: I4394ad3b04d89a55ecb5ab3e569571784f4c619e
-
Ivan Chiang authored
Test: atest CtsPackageManagerTestCases:PackageManagerShellCommandMultiUserTest Test: atest CtsPackageManagerTestCases:PackageManagerTest Bug: 297517712 (cherry picked from https://googleplex-android-review.googlesource.com/q/commit:81eb9f8294645684ce1fad39d5d4a00ef11736e4) Merged-In: I271a3526ea4555249e3a2797605269257330e0e9 Change-Id: I271a3526ea4555249e3a2797605269257330e0e9
-
wilsonshih authored
Modify the set mComposed order to preventing from recursive call. Bug: 343030535 Test: atest BackNavigationControllerTests Change-Id: Iaab6685a42c3416aa2f7a6d7d2db2669ba776cef
-
Chris Li authored
Bug: 339380439 Test: refactor, pass existing tests Flag: com.android.window.flags.insets_control_changed_item Change-Id: If95ba8f513f9c8d09a73f67b88cf840920e893ed
-
Josh Hou authored
The delegate selector is a relatively new addition to backup and restore. If there is no backup in the source device, an attribute not found exception may occur when parsing the data during the restore process. Default value is provided for cases where the attribute is not found. Bug: 343180989 Test: Manual test Change-Id: Idb11da383a7a2387ebb3c6e3bbad873deb81a898
-
Jiaming Liu authored
This change ensures that the decor surface change happens after the transition is finished. Before this change, the decor surface was moved before the transition was sent to Shell, which caused a flicker when finishing dragging because the decor surface was unboosted but the TaskFragments bounds were not updated and the snapshots were still displayed. Bug: 327067596 Test: atest TransitionTests Change-Id: I3286023767992ec762bc9459d3f6b2a746209930
-
- May 28, 2024
-
-
Vlad Popa authored
Flag: EXEMPT bugfix Test: kill audio server and check interface Bug: 342526946 Change-Id: Ic7ce3f9fa19c923fa69bc7cec79d1728ab68e75e Merged-In: Ic7ce3f9fa19c923fa69bc7cec79d1728ab68e75e
-
Vlad Popa authored
Flag: EXEMPT bugfix Test: kill audio server and check interface Bug: 342526946 Change-Id: Ic7ce3f9fa19c923fa69bc7cec79d1728ab68e75e
-
qinyige1 authored
For debugging layer leak. Bug: 339138626 Test: Manual Change-Id: I8500831ec0ab1aad77eb2241447becb94157a469
-
Julia Reynolds authored
If a user activates a mode, the app cannot deactivate a mode until it activates a mode. Like rule 'snoozing', but with the opposite states. Flag: android.app.modes_ui Bug: 333420153 Test: ConditionProvidersTest Change-Id: I57c7db08d854cd66954136fbcdc9da0d9cd575d2
-
Guojing Yuan authored
associations Fix: 338005013 Test: CTS Change-Id: I6d11ce553f3f190bb65d716b23eada1efc34e171
-
Santiago Seifert authored
The goal being its reuse in provider service proxy. This CL also renames the pending transfer initiation data holder to cover regular transfers. Bug: b/319651986 Test: Presubmit (non-functional refactor) Flag: EXEMPT refactor Change-Id: I2b78f36447765853c02a94a926913c6e8e096bc7
-
Vlad Popa authored
Test: build Bug: 341854556 Change-Id: I20522e9f53545e0983d4c3c78390525a3fe081f0
-
Marzia Favaro authored
Task#getDimmer is now depending on isTranslucentForTransition since ag/27102803, therefore stub this method too Bug: 343147794 Test: TransitionTests#testTransientLaunch Change-Id: I5484f7b3c1ae5090ee005603d7f5291daf0352c6
-
Jean-Michel Trivi authored
Statically import permissions used more than once. Remove some permission-related unused methods. Bug: 341995844 Test: make Change-Id: I56fa0b3b4c5be55c2f21424499b35c8fc551514c
-
dakinola authored
Found an issue with screen recording, where if you open an app to be recorded, but close the initial activity before recording starts, the recording fails because we expect to identify the task to the record by looking for an activity with matching launch cookie. As a backup, we also supply the task id (when possible) into the projection and try to find the opened task that way. Fix: 333779790 Flag: None Test: manual testing by attempting to screenrecord Google Photos while on the trash acvitity, then navigating back to the home activity before recording starts Test: atest FrameworksCoreTests:ContentRecordingSessionTest Test: atest FrameworksServicesTests:MediaProjectionManagerServiceTest Test: atest WmTests:WindowManagerServiceTests Test: atest RecordingServiceTest Change-Id: Ied4c8f117ce39288971d293dbcafa022711bf26a
-
Oli Thompson authored
Bug: 298661886 Test: manual Change-Id: Ib2b5c4d7306fcf4e921898797e9fba1d76e50c66
-
wilsonshih authored
Core won't ask to draw snapshot window if the aspect ratio isn't match. Add a flag to verify if it is safe to delete. Flag: com.android.window.flags.draw_snapshot_aspect_ratio_match Bug: 341020277 Test: presubmit Change-Id: I3e2b3435ba2ffe60c90c22ed563435e0c1e781d7
-
wilsonshih authored
That jumping is because the WindowFrame of top fullscreen window isn't align with app main window. In core there only check app main window's cutout for layout, so for snapshot starting window, it should use app's main window's cutout instead of top opaque window. Another reason there doesn't need to use getTopFullscreenOpaqueWindow is because core won't ask to create snapshot starting window if the size/orientation of snapshot isn't align with the window configuration, so the SnapshotDrawerUtils will not needs to draw background and bars, since the aspectRatioMismatch should always return false. Bug: 341020277 Test: manual, create a second window for an activity, verify snapshot starting window won't jump when hot launch the app from background. Change-Id: I4c6b0ccacc79730552cbf9b4928b5c842f4feb1e
-
Yunfan Chen authored
Moving the size override after other compat overrides to make sure the letterbox overrides can be applied correctly. The fields overridden by letterbox doesn't need extra override regarding the stable configuration as all insets are already in consideration. Besides, check shouldCreateCompatDisplayInsets() before applying the override to avoid inconsistency when the compat display insets is not yet created. This is a partial cherry-pick contains the critical part of the change. Bug: 332468995 Bug: 340970172 Bug: 336687188 Test: See reproduce steps in b/332468995 Test: CtsWindowManagerDeviceWindow:WindowMetricsActivityTests Test: CtsWindowManagerSdk25TestCases:AspectRatioSdk25Tests Test: v2/android-crystalball-eng/health/microbench/startup/hscapps/compile-speed-profile/open-maps Merged-In: Iee7ae3ea7e483e6f80fa6ef4659627d620653081 Change-Id: If2e2e7f0eb504f2468f262b195f3a1cd28820423
-
Yohei Yukawa authored
This is a follow up CL to my previous CL [1], which introduced a tricky bug where InputMethodManagerService may look to forget certain IMEs when an IME calls IMM#setAdditionalInputMethodSubtypes() API. The issue is that when an IME X calls InputMethodManagerService#setAdditionalInputMethodSubtypes(), IMMS stops recogning IMEs that are not visible from IME X. This happens because InputMethodSettings for the calling user will be replaced with packageManager.queryIntentServices( new Intent(InputMethod.SERVICE_INTERFACE), PackageManager.GET_META_DATA | PackageManager.MATCH_DIRECT_BOOT_AUTO | PackageManager.MATCH_DISABLED_UNTIL_USED_COMPONENTS); where package visibility filtering will be applied with IME X's identity rather than system_server's identity. With this commit we start calling packageManager#queryIntentServices() with system_server's identity as it used to be before my commit [1]. See also the corresponding regression test [2]. [1]: If7d9dc5636f1bf04f02a18c84d55681914e82014 43198777 [2]: I25245a48088c8dd75affc1e68e174a8f8af0d189 Bug: 343108534 Fix: 335281466 Fix: 339928613 Fix: 340806169 Fix: 342105635 Fix: 339912139 Test: atest CtsInputMethodTestCases:InputMethodSubtypeEndToEndTest Change-Id: I3ea4e233202f8b77b8162a743f1ffedff9abb4af
-
Seigo Nonaka authored
Bug: 331717791 Test: atest UpdatableFontDirTest Test: Step 1: Build AOSP image and install Test: Step 2: Install font family. Test: Step 3: Remove certificate manually by adb shell. Test: Step 4: Reboot the device Test: Step 5: Install font family. Change-Id: I0373a2e66b7a6a51168094772afb36d8905b6979 Merged-In: I0373a2e66b7a6a51168094772afb36d8905b6979 (cherry picked from commit c273a7e8)
-
Riddle Hsu authored
Simulation: Task A starts(NEW_TASK) Task B, finishes Task B starts(NEW_TASK) Task C Task B calls moveTaskToBack in onPause Task C (declare as translucent and sleep 200ms in onCreate to delay transition) starts(NEW_TASK) Task B -> attempt to move task B to front finishes Because the initial open transition may be done until the last step, then moveTaskToBack for B will run after the last step. Then expect B should be on top but the result is on home screen because A and C are finished, and B is moved to back. This change is kinda a workaround to mitigate the case of deferred move-to-back. Because it is regular to have closing targets in an open transition, the moving task can run with the collecting transition directly. Fix: 325645759 Test: atest TransitionTests#testDeferredMoveTaskToBack Merged-In: I46d8e26110ebefea74e5dd06c85c003b12693416 Change-Id: I46d8e26110ebefea74e5dd06c85c003b12693416 (cherry picked from commit 7079ba9a)
-
pajacechen authored
Test: atest com.android.server.power.batterysaver.BatterySaverStateMachineTest --iterations 10 Test: http://ab/I99400010281271439 Fix: 340357106 Flag: NA Change-Id: Ibc4e903dac33931e0f1f83dd372b2cd56d09fde7
-
- May 27, 2024
-
-
Li Li authored
Currently it's hard to debug why a binder transaction fails and who's causing the issue. To help debugging this kind of issues, dump the stack of the sender when the transaction fails. Bug: 331005349 Test: logcat Change-Id: I1a032784932ebe10af5f355ec1364a5d68831f2b
-
Kevin Liu authored
As strict mode is in place, the default strict_mode is now set to true. We are encourage teams to write tests in strict_mode to make tests Bivalent (able to run on device and devicelessly) For more info on strict mode: go/roboStrictMode Flag: NONE Test: atest SystemUiRoboTests, SpaRoboRNGTests, SettingsLibRoboTests, FrameworksServicesRoboTests, BackupFrameworksServicesRoboTests, InputRoboRNGTests Bug: 334089788 Change-Id: I9827910496af67eab33efddb7c32d8e40b497c13
-
Matthew Sedam authored
Bug: N/A Change-Id: I753a4f5b7f5050aa6483a1f413e83ee202c42ea0 Test: Presubmits
-
Rupesh Bansal authored
We have historically preserved the lux buffers over a configurable long horizon values since the clock uptime. However, the time when android is sleeping is not accounted for in this, which is why the lux buffers can go incorrect. Changing this to uptime, and also using the exact time at which a sensor event was emitted to maintain the bufffer Test: atest AutomaticBrightnessControllerTest Bug: 341219242 Change-Id: I6b5c5b2c8754ddcd97249fcaf89b1575a865d4d8
-
Oleg Blinnikov authored
Bug: 332711269 Bug: 338927759 Test: atest ExternalDisplayPolicyTest DisplayManagerServiceTest LogicalDisplayMapperTest Change-Id: Iad0c5b0ea19a8f1f14642109639fd435b0e43bc4 (cherry picked from commit ab71a6c1)
-
Sandro Montanari authored
Flags that were reverted and resubmitted are not supported by the Gantry infra. The recommended approach in these cases is to instead rename the flag. Test: atest SdkSandboxAuditTests SdkSandboxNextTests Bug: 342991282 Change-Id: I26d28abcc1b75c2a3f53cf9391a403ea9c5b27a2
-
Riddle Hsu authored
The case happens when a pip is entering in previous rotation and then display updates. The transition type can be either TRANSIT_CHANGE or TRANSIT_PIP: PipTask CHANGE sb=Rect(0, 0 - 1080, 2340) eb=Rect(0, 0 - 2340, 1080) r=0->3 Display CHANGE sb=Rect(0, 0 - 1080, 2340) eb=Rect(0, 0 - 2340, 1080) r=0->3 The destination pip bounds should be calculated from (2340, 1080). Because a display snapshot should have covered the screen, the pip should go to the end state immediately and only animate the display rotation animation. This also merges [1] and [2]. [1]: I110d1c11f3d3fdcfb83698e5cf1ec4efb062bd10 [2]: Ia9c78105f4a9782c156744e6cb38681f265955a0 Bug: 340367710 Test: Enable auto rotation and home rotation. Swipe up (do not release touch) an auto pip activity in portrait. Rotate the device to landscape. Continue the swipe up to enter pip. The display should show a rotation animation and the pip can show on correct landscape position, Merged-In: Ia1e5e8c7edb8c2a078a66b587d5d26a495166de2 Change-Id: Ia1e5e8c7edb8c2a078a66b587d5d26a495166de2 (cherry picked from commit 6373912c)
-
- May 26, 2024
-
-
Beatrice Marchegiani authored
Add a check in TarBackupReader to enable the restore of packages in the V_To_U_Allowlist in the V to U downgrade scenario Bug: 341732443 Test: atest -v TarBackupReaderTest, manual test (details: https://b.corp.google.com/issues/341732443#comment18) Change-Id: I12a1f272b17a12abb532a98c0d93f0a9854fcb51
-
- May 25, 2024
-
-
Jean-Michel Trivi authored
Implement Object.hashCode() method. Allow AudioVolumeGroup.getAudioVolumeGroups() to be called by holders of MODIFY_AUDIO_SETTINGS_PRIVILEGED, like the other volume group APIs. Bug: 340879593 Test: atest AudioManagerTest#testVolumeGroupHashCode Change-Id: Ie4898d013eef9352bb5b9fa7ce7e09fca095356e
-
- May 24, 2024
-
-
Paul Colta authored
Instead of throwing an exception, stop the actions on device type mismatch. Test: atest com.android.server.hdmi Bug: 341680401 Change-Id: I89d30446a5b5cb9f464fd376af83896e319bf897
-
Jing Ji authored
Fix a watchdog bite with it. Bug: 342548620 Test: Manual Change-Id: I31faaffd5573d6eb4dda8f005ae50a17df76cbc5
-
Julia Reynolds authored
Test: NotificationManagerServiceTest Flag: android.app.remove_remote_views Bug: 342602960 Change-Id: I7c1973ddcf4410d4acc638288630e597c61298a1
-