frameworks-base

Fork

Add minimize & pip buttons to freeform windows

Thanks to @farmerbb for assistance and @rogerdott for the graphics

Change-Id: I8ddad6eab27f52574ee1c10e8006612fb46a83c5

Set forceDesktopMode to true

android_view_ThreadedRenderer: prevent Null Pointer Exception in createBitmap

To avoid Playstore crashes with nouveau and Hardware Bitmap,
a check is added to bitmap returned by Bitmap::createFrom(buffer)
inspired by similar behavior in Bitmap.cpp Bitmap_createHardwareBitmap()

Fixes the following crash:

*** *** *** *** *** *** *** *** *** *** *** *** *** *** *** ***
Build fingerprint: 'Android-x86/android_x86/x86:8.1.0/OPM8.181005.003/uten01131829:userdebug/test-keys'
Revision: '0'
ABI: 'x86'
pid: 4644, tid: 4644, name: android.vending >>> com.android.vending <<<
signal 11 (SIGSEGV), code 1 (SEGV_MAPERR), fault addr 0xc
Cause: null pointer dereference
eax 89195380 ebx b1d36290 ecx 00000000 edx ae783074
esi 00000000 edi ae7a9380
xcs 00000073 xds 0000007b xes 0000007b xfs 0000003b xss 0000007b
eip b1c4f7c8 ebp bff902c8 esp bff9028c flags 00010246

backtrace:
#00 pc 000fd7c8 /system/lib/libandroid_runtime.so (android::bitmap::createBitmap(_JNIEnv*, android::Bitmap*, int, _jbyteArray*, _jobject*, int)+104)
#01 pc 000d08ca /system/lib/libandroid_runtime.so (android::android_view_ThreadedRenderer_createHardwareBitmapFromRenderNode(_JNIEnv*, _jobject*, long long, int, int)+1002)

pm: ignore restorecon failure

On the 9p filesystem, restorecon won't work. It causes apk can't be
installed. Just ignore the errors to workaround it.

Modify color inversion matrix to swap Red and Blue colors

Instead of color inversion it will produce Red and Blue colors swap,
in order to correct displayed colors for R4xx and other old drivers
still based on KMS API

ABIPicker: match package name by patterns

PRC apps from different markets usually have different package names.
This change tries to match them by patterns.

To test it, install the Implosion apps from Wandoujia and CoolMarket
and run them OK.

Fix the memory leak bug introduced by PRC compatibility feature

Fix the memory leak bug introduced by PRC compatibility feature,
introduced by https://android.intel.com/#/c/471490/

Change-Id: Iaf9bd21afa17f3a81ab700c63ae7c0bb0851a594
Tracked-On: https://jira01.devtools.intel.com/browse/OAM-27449
Signed-off-by: jgu21 <jinghui.gu@intel.com>
Reviewed-on: https://android.intel.com:443/489353

Fix KW issues on PRC compatibility

Fix KW issues on PRC compatibility,introduced by below commit.
https://android.intel.com/#/c/484542/

Change-Id: Ib10899655dd2a18569387dbafbe33856acf5df59
Tracked-On: https://jira01.devtools.intel.com/browse/OAM-26904
Signed-off-by: jgu21 <jinghui.gu@intel.com>
Reviewed-on: https://android.intel.com:443/487804

Reduce the overhead of the PRC compatibility package feature during system bootup

The PRC compatibility package introduced too heavy overhead
in system bootup. This commit aims to reduce the overhead
in system bootup and improve the performance of PRC
compatibility package.

NOTE: The format of ThirdPartySO has been changed to improve
the performance. From now on, the lib name should be trimmed
as below if want to add into the list. For "libabc_v1_2_3.so",
add "abc_v" into the list, that is, the version information
at the tail of lib name should be removed.

Change-Id: Ic374e363d3d31f9bd69be839b33b1bd65950ef61
Tracked-On:https://jira01.devtools.intel.com/browse/OAM-25819
Signed-off-by: xiajiang <xia1.jiang@intel.com>
Reviewed-on: https://android.intel.com:443/484542

Enable the PRC compatibility package feature for PRC market

In PRC market, some APKs are packed in non-standard way, that is,
the x86(_64) libraries aren't workable although they're found in the APK.
This patch intends to relieve the impact from below 2 defects,
which is based-on the heuristic algorithm.

NOTE: To enable this feature, set "PRC_COMPATIBILITY_PACKAGE := true"
in device BoardConfig.mk before enabling houdini.

1. Missing x86(_64) libraries: The x86(_64) libraries are ported and
existing in the APK. But it's incomplete, and not workable.
2. Mixed arm(64) libraries: Several libraries are existing in lib/x86(_64)/
directory in the APK. But some of them are arm(64) libraries,
instead of x86(_64) ones.

All of above always cause APP crash if installed x86(_64) libraries
by PackageManager.

This patch aims to improve PackageManager to figure out these defects
and install arm(64) libraries to run the APP with houdini support.
The basic idea is to compare x86(_64) libraries with arm(64) alternatives
to determine which one should be workable.

To customize it for specific APPs, 3 lists are provided under
/system/vendor/etc/misc/ on the device.

1. /system/vendor/etc/misc/.OEMWhiteList:
This patch will be disabled for the APPs containing in the OEMWhiteList
2. /system/vendor/etc/misc/.OEMBlackList:
The APP is enforced to install the arm(64) libraries if existed.
3. /system/vendor/etc/misc/.ThirdPartySO:
This is another list which contains the names of all 3rd-party libraries,
which will impact on the final decision of APP installation.

Change-Id: I2613d9ebc8fe012c801f4a38fc5dede413f15a91
Tracked-On: https://jira01.devtools.intel.com/browse/OAM-20470
Signed-off-by: xiajiang <xia1.jiang@intel.com>
Reviewed-on: https://android.intel.com:443/471490

fd_utils: fix DEBUG mode booting issue (pie-x86)

Porting to pie-x86 codebase of oreo-86 commit
c283ba107a ("fd_utils: fix DEBUG mode booting issue")

Original commit message:

Remove the "/android/" prefix before the white list checking.

Suggested by Chris Vandomelen <chris@sightworks.com>.

Camera: ignore exceptions from getSupportedPreviewFpsRange

input: simulate long press properly

The original implementation of long press sends two ACTION_DOWN events
which would be interpreted as a double tap. For example, sending
a long press POWER key will launch the camera:

11-21 16:27:37.320 2223 2223 I Input : injectKeyEvent: KeyEvent { action=ACTION_DOWN, keyCode=KEYCODE_POWER, scanCode=0, metaState=0, flags=0x0, repeatCount=0, eventTime=39641, downTime=39641, deviceId=-1, source=0x101 }
11-21 16:27:37.321 2223 2223 I Input : injectKeyEvent: KeyEvent { action=ACTION_DOWN, keyCode=KEYCODE_POWER, scanCode=0, metaState=0, flags=0x80, repeatCount=1, eventTime=39641, downTime=39641, deviceId=-1, source=0x101 }
11-21 16:27:37.322 1411 1565 I GestureLauncherService: Power button double tap gesture detected, launching camera. Interval=0ms
11-21 16:27:37.322 2223 2223 I Input : injectKeyEvent: KeyEvent { action=ACTION_UP, keyCode=KEYCODE_POWER, scanCode=0, metaState=0, flags=0x0, repeatCount=0, eventTime=39641, downTime=39641, deviceId=-1, source=0x101 }

This is unexpected and incorrect.

Just simulate the long press by delaying ACTION_UP one second.

DiskInfo: support CD/DVD type

Detect HDMI audio by switch events

If detect SW_VIDEOOUT_INSERT, assume HDMI is plugged.

Support save local time to RTC

Make it work for dual boot with Windows, we could set property
persist.rtc_local_time to 1, so that RemixOS and Windows both
use local time in rtc.

NO_REF_TASK
Tested:
1) set time in android, reboot to bios, make sure the time in bois is
local time
2) reboot back to android, the current time should be correct local time
(this need another change)

Change-Id: Id2c0ce150fb9320b132ad8bdd83b38dcae1070db

Map keycode 120 to APP_SWITCH

Define the unused keycode 120 to be APP_SWITCH.

InputReader: read the pointercal from TSCalibration2

Watch the pointercal and reload it on changed.

Make "windows" key to act as "home" key

Issue: AXIA-1893
Change-Id: I7b2b19f7e34ec8a1867e8e6ce522e65133e5267f
Signed-off-by: juntingwang <Junting.Wang@windriver.com>

camera: fix NullPointerException

GLSurfaceView: Be less picky about EGLConfig alpha sizes

EGLChooseConfig returns a "best match" set of visuals meeting or
exceeding the required r/g/b/a component depths. But GLSurfaceView
oddly requires that the returned visual be an exact match. Add to
that that the (rarely used outside of CTS) default request specifies
zero alpha bits and that not all drivers expose a zero-alpha
EGLConfig, and the default configuration will fail needlessly.

It's not incorrect to have alpha bits you didn't request: the only way
to produce divergent behavior is for a fragment shader to write out
explicit alpha values (into the channel it didn't want to begin with!)
with values other than 1.0 and then rely on them being ignored and
treated as 1.0.

For: AXIA-1448
Change-Id: I2f64995d7b9de1ae082aa47822af525390102083
Signed-off-by: Andy Ross <andy.ross@windriver.com>

import YuvToEncoder R3 patches, BZ 18528, 19092, 19696

BZ: 47824

Please refer http://umgbugzilla.sh.intel.com:41006/show_bug.cgi?id=19092 and
external/jpeg/libjpeg.doc

The MCU height is max_v_samp_factor = 2 DCT rows so you must pass at least 16
scanlines on each call to jpeg_write_raw_data(), which is to say 16 actual
sample rows of Y and 8 each of Cb and Cr.

The original implement of Yuv420SpToJpegEncoder::compress didn't add padding to the frame buffer
when height and width aren't aligned with 16 pixel. It will cause illegal memory violation and core dump.

Category: aosp improvement
Domain: Video.Media-jpeg
Origin: Internal
Upstream: Yes

Change-Id: Ibcf14230d616e2d440ace244bb420723b5c01dc2
Orig-Change-Id: Ic1b7494b98ee9c1997b226d58abd034b1dcb18f6
Signed-off-by: Tong, Bo <box.tong@intel.com>
Signed-off-by: Zhao Liang <leo.zhao@intel.com>

Merge cherrypicks of [10484934, 10484473, 10472987, 10481068] into qt-qpr2-release

Change-Id: I3ba00e16f09289a9368e55b59b0a527bc55bb3f3

DO NOT MERGE: Revert Move startInput for WINDOW_FOCUS_GAIN to background thread

Reason for revert:
Caused an unexpected regression Bug 144174015

Bug: 139806621
Bug: 144103599
Fix: 144174015
Test: Manually verified Bug 144174015 disappeared as follows
1. Open Gmail then start composing an email
2. Swipe up the home button to recents then re-launch Gmail
3. Do the step 2 several times.
4. Make sure that you can still type something on Gmail.
Change-Id: I9265f01ed2f6e4aca7728d278f06ceea5633dac5
(cherry picked from commit 344858dd9c5616a0de16fa0ac72e0d384e2406c4)

DO NOT MERGE: Revert: Freeup lock when IME is set inactive and unbound

Reason for revert:
Caused an unexpected regression Bug 144174015

Bug: 139806621
Bug: 144103599
Fix: 144174015
Test: Manually verified Bug 144174015 disappeared as follows
1. Open Gmail then start composing an email
2. Swipe up the home button to recents then re-launch Gmail
3. Do the step 2 several times.
4. Make sure that you can still type something on Gmail.
Change-Id: I04a77afea17f9d3eb05017fa00313fad4e48cd5c
(cherry picked from commit 9494c9dbb73b0ce237cb2f64fba90434b1b09c09)

Merge cherrypicks of [10460465, 10460521, 10460352, 10458819] into qt-qpr2-release

Change-Id: Iceb99c57cf8adeb4ffa9fa79fbc5f8246058e2cf

Allow settingIgnored for DBH request if inEmergency

Bug: 150232136

Test: on device
Change-Id: Ia987418a591d716b787d406d725338a8563a55dd
(cherry picked from commit 6d19cef854bdb382507daefae2a5956400a255d5)

Merge cherrypicks of [10297641, 10298303, 10298494, 10298514, 10297196, 10298164, 10297642, 10297643, 10296046, 10296047, 10298245, 10298246, 10297197, 10297198, 10298495, 10298496] into qt-qpr2-release

Change-Id: I1f8470eec0bf64e28290002664f6e5a6c8d733dc

Fix serialization issue of ExternalVibration

Remove excessive serialization of Audio Attributes
Bug: 140417434
Test: atest ExternalVibrationTest#testSerialization

Change-Id: Ib7ceaed875889126a53f874eec64fab4817e48d1
(cherry picked from commit b1a33a8b4fd4b1603c0465a904be29f0c4a07e64)

Revoke 'always' web handler status when not autoverifying

If an app has previously used autoVerify to make claims about its status
re handling web navigation intents, but is updated such that it no
longer makes those claims, step down its "official handler" status as
though it had never invoked autoVerify in the first place.

Bug: 146204120
Test: manual: as described in bug; observe policy before/after via
'adb shell dumpsys package d'
Test: atest CtsOsHostTestCases
Change-Id: I58502d1b32d793aba9aa772fa2ad5ac38acca48a
Merged-In: I58502d1b32d793aba9aa772fa2ad5ac38acca48a
(cherry picked from commit d2a71cc4b8f11688f85f33507b75d00041c14852)