This is determined by:
- a product config flag
- the vendor API level
It is then passed to the device as a system property
"ro.dalvik.vm.enable_uffd_gc".
This change is a no-op change. It doesn't enable userfaultfd GC by
default. OVERRIDE_ENABLE_UFFD_GC=default can be passed to the build
system to enable userfaultfd GC for testing purposes.
Bug: 242553398
Test: -
1. lunch aosp_redfin-userdebug
2. OVERRIDE_ENABLE_UFFD_GC=default build/soong/soong_ui.bash --dumpvars-mode --vars=ENABLE_UFFD_GC
3. See "false" in the output
Test: -
1. lunch aosp_oriole-userdebug
2. OVERRIDE_ENABLE_UFFD_GC=default build/soong/soong_ui.bash --dumpvars-mode --vars=ENABLE_UFFD_GC
3. See "true" in the output
Test: -
1. lunch aosp_redfin-userdebug
2. OVERRIDE_ENABLE_UFFD_GC=true build/soong/soong_ui.bash --dumpvars-mode --vars=ENABLE_UFFD_GC
3. See "true" in the output
Test: -
1. lunch aosp_oriole-userdebug
2. OVERRIDE_ENABLE_UFFD_GC=false build/soong/soong_ui.bash --dumpvars-mode --vars=ENABLE_UFFD_GC
3. See "false" in the output
Change-Id: Ifd6e6cddb502315912ff949619a5b526ae0d73ff
The change in the packages/modules/Virtualization in the same topic
takes care for defining the right variant of the APEX to be installed:
* Devices with AVF support will get the full com.android.virt APEX
* Devices without AVF support will get an almost empty version that only
contains an app that defines AVF-related permissions, and soon the BCP
fragment jar.
Bug: 243512044
Test: build & flash oriole
Test: build & flash redfin
Change-Id: I2c38e03529a77dd2820212812433741725115ce0
There are still devices that set BUILD_BROKEN_USES_BUILD_HOST_* and
build host modules in Make. Add basic support for building against
libc_musl in Make when USE_HOST_MUSL is set.
Bug: 258535366
Test: build a host tool defined in Android.mk file in internal branch
Change-Id: I9150be749bbeaac2ac5a33b2cf915004aa3033df
Currently, the Art benchmark service is depending on the boot.zip file
to filter out boot image profiles based on the existing jars in this zip
file, and to expand profiling to apex system server jars and standalone
jars we need to export these jars in another zip file.
Bug: 241823638
Test: Checked the generated `system_server.zip` using `m droidcore dist`
Test: Checked with aosp and internal targets
Change-Id: I477dcfd3381bd24bdc50aabb773b38f8840f0ce0
Currently, the Art benchmark service is depending on the boot.zip file
to filter out boot image profiles based on the existing jars in this zip
file, and to expand profiling to apex system server jars and standalone
jars we need to export these jars in another zip file.
Bug: 241823638
Test: Check the generated `system_server.zip` using `m droidcore dist`
Change-Id: Ief2c92588b5e4f6ccd5163ece8d241bc473b22ad
Previously, the file is generated only if there's interesting files. The
condition isn't really desirable in build rule evaluation, and makes it
a bit harder to handle in the consuming code.
Bug: 256793394
Test: atest ComposHostTestCases odsign_e2e_tests
Change-Id: I87b257b6ce6ab0ba565378c885adf35c1d25235b
Fix build error when packages/modules is not existed.
Bug: 256565230
Test: make dist, check otatools.zip
Change-Id: I4153b143e677e35f95a3d47663cb8efcde7abdbf
Generate a all_modules.txt file on every build that
can be used for autocompleting.
Because this adds time to every non-incremental build,
I profiled it by using `$(shell date +"%s.%3N")` before
and after, and disovered it only adds about 0.01 seconds
to the build. And of course it won't be rerun if there
weren't build changes.
Bug: 255867758
Test: Manually
Change-Id: I1d3f1008611fd542884322b27b19ea2fa1d341fe
This reverts commit 6137ffa5de.
Reason for revert: Roll-forward with fix for codename SDKs (PS2)
Test: m sdk dist in all 3 variants of branches
Change-Id: Iffdf650a43aa050d70e01f73d78cf622cb206fc6
Also, create a python_binary_host so that it runs using the hermetic
python toolchain.
Test: m check-elf-files
Bug: 203436762
Change-Id: I964342a27f6b6c9dcdbbe910d4dc6c9ec00c2213
build_test is only run on presubmit in internal master
for cls that change build/make and a few other repos,
but notably not build/soong. Add this cl to a topic
with a build/soong change we'd like to test with build_test.
Bug: 220375749
Test: Presubmits
Change-Id: I3e2f59e5f341cce486368dac9bf2f28f265e53fd
Instead of wrongly using the key that was used to sign pvmfw, we should
have been using the key that pvmfw will use to verify payloads it
receives so fix that.
Bug: 255575669
Test: -
Change-Id: I68a87a092c1aa4a60f4fbe60c5ddf82c4e34c38a
This reverts commit 8b22209184.
Reason for revert: Doesn't match install path for pre-release SDKs
Change-Id: Ic1ce28e57692d27d08d33dee5b7403bf1bf1b655
There are no usages of LOCAL_JAVA_LAYERS_FILE,
and java-layers.py was written in python 2.
Bug: 203436762
Test: Presubmits
Change-Id: I1be0f2547757a853e20811f28529479487037e94
The build system previously named the platforms subdir by
PLATFORM_VERSION, which is 13 for android 13, aka T, so the subdir
name would be android-13.
Android studio installs the SDK into a subdir identified by the api
level, so android-33 for android T, and also the extension api level
for extension SDKs, e.g. android-33-ext4.
Update the build system subdir name to match the android studio install
path, in order to avoid confusion and make it simpler to just unzip
a built SDK into the platforms dir, avoiding the need to provide a
package.xml file for studio to install it correctly.
Bug: 254023524
Test: m sdk sdk_repo dist, verify subdir is android-33 in:
$ zipinfo -1 out/dist/android-sdk_eng.hansson_linux-x86.zip \
| cut -f 1-3 -d '/' | grep platforms | sort | uniq
Change-Id: I1d3a7fd4c480859c28cdcbcb8c0cf2800ac4a1d7
