It doesn't make sense to specify a DEVICE_MANIFEST_FILE but the
build not including it. The conditional definition of the module
makes it problematic to have it listed in the PRODUCT_PACKAGES
of a generic mk.
Test: build mainline_arm64 + mainline_system_arm64
Change-Id: I40e435ce114263ceabbd373d3d719318c592fc13
Stop using armv7-a without neon for 32-bit unbundled apps, and update
generic_x86_arm to match the cuttlefish configuration that includes
neon.
Test: treehugger
Change-Id: Ieb6a2106655803a8ca609907c12168e628ee1b85
boot_profile_jars.zip is used to process the boot image profile (by
filtering out data that does not belong to the boot classpath).
We only need the jars that are actually preloaded, and not the ones which
are updatable.
Test: launch taimen_nopreload-userdebug; m
Bug: 123313750
Change-Id: I630f78dda3d9b38aac04b0f364b25fbcda4e26a3
To solve AIDEGen issue: DocumentsUI have no dependency libcore.
Bug: 122557164
Test: 1. m `build/soong/soong_ui.bash --dumpvar-mode PRODUCT_OUT`/module-info.json
2. open $ANDROID_PRODUCT_OUT/module-info.json
"out/target/common/obj/JAVA_LIBRARIES/core-lambda-stubs_intermediates/classes-header.jar",
"out/target/common/obj/JAVA_LIBRARIES/core.platform.api.stubs_intermediates/classes-header.jar",
"out/target/common/obj/JAVA_LIBRARIES/ext_intermediates/classes-header.jar",
"out/target/common/obj/JAVA_LIBRARIES/framework_intermediates/classes-header.jar"
will be added to DocumentsUI's dependencies.
Change-Id: I6b20c315f3367ff2a016d305ec6fe2a8de620a14
Bug: 111136242
Test: When BOARD_PREBUILT_DTB_DIR is set correctly,
generated $OUT/boot.img contains the DTB image.
Change-Id: I282e31b04cc60383377b9e9b54f8fe64a8140242
We currently pass the boot dex files installed on /system, but these
may be eg stripped. That's currently OK as dex2oat is looking
at the dex files in boot.oat, but dex2oat could look at them for
various reasons.
Test: m
Change-Id: I85915c979cf94ec29751b10a0d8a5957b6cea956
Support signing system_other.img but shouldn't include it into the
top-level vbmeta.img. system_other verifiation will not be included
in /vbmeta chains and will be done separately.
Bug: 112103720
Test: avbtool info_image --image $OUT/system_other.img
Test: avbtool info_image --image $OUT/vbmeta.img, checks 'system_other' is NOT included.
Test: Checks $OUT/obj/PACKAGING/system_other_intermediates/system_other_image_info.txt
See the following:
avb_system_other_hashtree_enable=true
avb_system_other_add_hashtree_footer_args=--rollback_index 1551744000
avb_system_other_key_path=external/avb/test/data/testkey_rsa4096.pem
avb_system_other_algorithm=SHA256_RSA4096
Change-Id: Ia152aaab1387dcf556a42222adb39ea76881263a
Uses a new BUILD_BROKEN_ENG_DEBUG_TAGS variable that defaults to true
until we mark all of the broken targets, then we'll switch it to false.
These are fairly trivial to fix, so I don't expect to keep this flag
around for long.
Test: build_test on aosp and internal master; check logs
Change-Id: I6176fec265e78aecff965b7cf1636f831f68a140
The instructions to look up for classes.dex don't work when looking
at different jar versions (aligned, unaligned, etc).
Test: m
Change-Id: I5f178ca0d00497edbab892722a5ee84f0c47d0a0
The REMOVE_OAHL_FROM_BCP flag is always set to true and is no longer
necessary so this change removes any uses of the flag.
Bug: 18027885
Test: cts/tests/signature/runSignatureTest.sh
Change-Id: Ied8c3315b29094ad6acee539520af3e4c9e79839
TARGET_BUILD_APPS sets TURBINE_ENABLED := false, which triggers
an issue where a dependency is added on classes-header.jar but
there is no rule to create it. Skip creating the dependency
when TURBINE_ENABLED is false for now.
Bug: 123183568
Test: m TARGET_PRODUCT=mainline_modules_arm UNBUNDLED_BUILD_SDKS_FROM_SOURCE=true TARGET_BUILD_APPS=com.android.media
Change-Id: I8ec3e244b8f9618c58f4ac57329a5a07f01dc797
This commit adds `$(SOONG_OUT_DIR)/ndk/%` to artifact whitelist so that
when NDK stub libraries won't be blocked by artifact path requirements.
This commit does not break the integrity of artifact path requirements
because they are not installed to devices.
Bug: 123053270
Bug: 119084334
Test: lunch aosp_crosshatch-userdebug
Change-Id: Ieea76a8126269f53b0a58bd069b449a9f9b1263a
This commit introduces a prebuilt ELF binaries checker. The checker
will check:
1. Whether all DT_NEEDED shared libraries are specified in
`shared_libs` (Android.bp) or `LOCAL_SHARED_LIBRARIES` (Android.mk).
2. Whether all undefined symbols in the prebuilt binary can be resolved
to defined symbols exported by its dependencies.
This ensures that prebuilt binaries won't silently become ABI
incompatible.
To check the prebuilt binaries, all of the dependencies must be
specified in `shared_libs` (Android.bp) or `LOCAL_SHARED_LIBRARIES`
(Android.mk).
If your prebuilt binaries cannot be checked for some reason, you may add
the following property to Android.bp:
check_elf_files: false,
Or, add the following setting to Android.mk:
LOCAL_CHECK_ELF_FILES := false
Bug: 119084334
Test: CHECK_ELF_FILES=true make check-elf-files
Change-Id: I523d3083f22fd4053c096d26f61f8375800281c8
Pass @CorePlatformApi stubs to the `hiddenapi` singleton rule which
generate a CSV of all public or domain API flags.
Bug: 119068555
Test: m appcompat
Test: dexlayout <core-oj jar> | grep CORE-PLATFORM-API
Change-Id: I597ee330835e564d61cfac44497e762b552b901c
`hiddenapi` is being refactored to work with a single CSV file as
opposued to a multitude of text files (one per flag). This patch
changes the singleton rule for listing public/private APIs from
stubs to expect a CSV as an output.
Bug: 119068555
Test: compiles, hiddenapi-flags.csv unchanged
Change-Id: If56c7ab8a4c3bbd33e447ef0dfd33f2c303a937c
The docs zips were being built for make docs dist but not for
make docs. Add them to ALL_DOCS and to the $(LOCAL_MODULE)
phony rule.
Bug: 123068677
Test: m docs
Test: m ds-docs
Change-Id: I32634c05e392f7a5483c81d59de35ca9bc4d4ad0
These two places are checking for ifdef(SANITIZE_TARGET) but what
they are really looking for is the second stage of ASan build.
Fix the checks so that they do not apply to HWASan.
I have not seen any change in behavior, but there are some new files
under /system in hwasan build that were not there before, and things
keep working in general, so this feels like the right move.
Bug: 112438058
Test: none; SANITIZE_TARGET=hwaddress keeps working
Change-Id: I4544f408263b908be6ef4a47dd2b5c937e0c1f33
These centralize the decisions on whether to build certain images or
not, and allow the product definition to override that choice.
There are a few use cases here:
* For GSI-like cases, we only want to build the system image. This
didn't really change, but it's somewhat simpler to configure, and
easier to understand the build logic.
* On the opposite side, when you're planning on using a GSI, the device
specific build can only build the vendor images (or some other set).
* Some cases (Fuchsia, etc) don't want to build any images, as they'll
be distributing the build artifacts in their own packaging.
I suspect in the future, TARGET_BUILD_APPS may be able to be refactored
into the third use case.
Test: treehugger
Test: Create a product definition that includes nothing, try to build it.
Test: compare build-aosp_crosshatch.ninja and build-crosshatch.ninja before/after
Change-Id: I685ab841be3718d3dd7052c28ccd764bb6f1991a
Recent move of hiddenapi rules to soong sets the dependencies of
the hiddenapi_flags.csv rule in dex_preopt.mk but doesn't check
for existence of "frameworks/base" as the hiddenapi-generate-csv
rule in frameworks/base. Solve this by removing the dependency
injection and using the SOONG_HIDDENAPI_FLAGS variable directly
in frameworks/base.
This is necessary because go/art-build creates a fake
hiddenapi_flags.csv and if there is a dependency injection rule,
a phony target is created and all dependencies of flags.csv need
to be satisfied.
Test: master-art compiles
Change-Id: I13ef5e7ec167f9372a5629fe1894c7acd4091d29
Define non_system_module in base_rules.mk when the module is product, vendor,
product_services or proprietary module.
Bug: 74699609
Test: m -j
Change-Id: Iebdd3ec8c643be3d9877b7a96cbdd83465fa3878
So that we can pass this into MAINDEXCLASSES.
Fixes: 122988865
Test: OUT_DIR=$HOME/out mmma frameworks/base/core/tests/hosttests/test-apps
Change-Id: I955d3de46d348e3573a6773f69149e7cabab8904
The strip rule does not need to wait for the boot image to finish,
create a separate DEXPREOPT_STRIP_DEPS list that only contains
tools used by GenerateStripRule in build/soong/dexpreopt/dexpreopt.go.
Bug: 123038675
Test: m checkbuild
Change-Id: I0814c8d99781533f5778e14547d3e9671e06317f
Perform hiddenapi CSV generation and dex encoding for Soong modules
in Soong. This fixes an issue where dexpreopting was happening on
a different jar than was being installed.
Bug: 122856783
Test: m checkbuild
Change-Id: I24a235c63ff62fed7e1af9f2fd17e55b0c1598a5
Simplify importing Soong modules into Make by always copying
LOCAL_PREBUILT_MODULE_FILE to LOCAL_BUILT_MODULE, which lets
Soong pick the final output file instead of trying to figure
out which of the various intermediate files should be the
output file.
Also fixes an issue where hostdex modules could attempt to
copy to both $(common_javalib.jar) and $(LOCAL_BUILT_MODULE),
which are the same file for host java modules.
Test: set compile_dex:true on core.platform.api.stubs. no warnings
of target overriding
Test: m checkbuild
Change-Id: I2e089012436fe8649db82a673d446d1c5a73a731
The impl lib name of java_sdk_library is changed to {module_name}.jar
instead of {module_name}.impl.jar
Bug: 110404779
Test: m -j
Change-Id: I275cd40e3682640b96425e56698cea770888cbb1
Merged-In: I275cd40e3682640b96425e56698cea770888cbb1
(cherry picked from commit 0372f189ac)
Appcompat runs for app defined in Android.mk, but not in Android.bp
So make appcompat run for app from soong(Android.bp)
Bug: 110073830
Bug: 122026042
Test: m out/target/product/$(get_build_var TARGET_DEVICE)/appcompat.zip
Change-Id: I73832410ec5850222b04086c2a377146886ba035
This enables execute-only memory (XOM) layouts to be used by default in
the build system. As of now, there's only support for ARM64 devices, so
this only affects those. Since userland XOM requires kernel support,
devices without the necessary support will continue to allow pages
marked execute-only to be read and they should be unaffected by this.
Bug: 77958880
Test: Device with and without kernel support boot.
Test: Binary throws segfault when reading from XOM on supported device.
Test: Debugger works and stack traces are still generated correctly.
Test: Teamfooding, stable during regular usage.
Change-Id: I8db4235c8e60eea6d701bdf3c43b79a06d6b01cd
Simplify importing Soong modules into Make by always copying
LOCAL_PREBUILT_MODULE_FILE to LOCAL_BUILT_MODULE, which lets
Soong pick the final output file instead of trying to figure
out which of the various intermediate files should be the
output file.
Also fixes an issue where hostdex modules could attempt to
copy to both $(common_javalib.jar) and $(LOCAL_BUILT_MODULE),
which are the same file for host java modules.
Test: set compile_dex:true on core.platform.api.stubs. no warnings
of target overriding
Test: m checkbuild
Change-Id: Id84d499a0869961be5a906d8472b75b9c843b4b4
This way if a BoardConfig.mk configures a specific image to exist (so it
doesn't end up as a folder on /system), but does not configure for it to
be created (like the device targets on AOSP that use a prebuilt), we
won't unnecessarily trigger the build system to build the contents.
Test: `m` before and after, comparing file lists
Test: check treehugger builds before/after
Change-Id: If0e4b958b3dfaa02771a5da70f970379635f904e
toybox xargs doesnt have --no-run-if-empty:
xargs: Unknown option no-run-if-empty (see "xargs --help")
But we shouldn't need it anyway.
Test: make -j32 showcommands dist TARGET_BUILD_APPS='Launcher3 Launcher3Tests' EMMA_INSTRUMENT_STATIC=true TARGET_BUILD_VARIANT=userdebug
Change-Id: I72f680c27b58a42b9fb9de1a0bc4187cabcfb516
Only dex jars should be installed, make it an error to not provide
a dex jar for an device installable java module.
Bug: 122831369
Test: m checkbuild
Change-Id: I99c5738e6053fbaa3dd22406f50b7a002555702c
The old code calls check-sum-of-partition-sizes within a foreach
loop, causing syntax error:
... fi partition_size_list=...
Add a semicolon after fi to fix this.
Test: build with multiple groups, each group with a partition in it
Change-Id: Ic4387408f3efc1744871619c300546765c678a61
This reverts commit 3387e65686.
Reason for revert: Topic likely causing build failure on aosp_x86_64-eng in aosp_master:
"build/make/core/base_rules.mk:271: error: art/build/apex: MODULE.TARGET.SHARED_LIBRARIES.com.android.runtime.host.libart-compiler already defined by art/build/apex."
Change-Id: I2033828e7a13062b72d1d66a7f289cf5bb12b776
But keep it as part of the boot class path and deliver the
dex files uncompressed in /system/framework/conscrypt.jar.
Test: Pixel 2 XL boots.
Bug: 119868597
Change-Id: I41b96a586401d120998be5a111331935030922c6
Since resource check has been done in Soong, we do not need that
here.
Bug: 118390303
Test: treehugger
Change-Id: Ib9d0862c65c6dc6e7fd7563995a9a335b3185786
Signed-off-by: Yoshisato Yanagisawa <yyanagisawa@google.com>
$(PRODUCT_OUT)/apex is where shared libraries in APEXes get installed
into. The path can be considered as a fake path as the shared libraries
are installed there just to have symbol files for them under
$(PRODUCT_OUT)/symbols/apex for debugging purpose. Shared libraries in
APEXes are made available to the path at runtime by mounting APEXes via
apexd. The files under $(PRODUCT_OUT)/apex are never compiled into a
file system image (e.g. system.img)
Bug: 120846816
Test: build/soong/build_test.bash -dist -products mainline_system_arm64
Change-Id: I3727091214f08f945559fa5f0f1450e74c1a5c9b
This fixes the WITH_TIDY flakes that we've been seeing, as we've been
relying on a particular order of target-specific variable inheritance.
Change-Id: I68c8a56620c29a41e45f26e5ea81a5313490c1cf
Fixes: 122637600
Test: WITH_TIDY=true m libhellojni_jni
I'll be adding more intermediate targets in this file, so move all of
the uses closer to the end.
Bug: 122637600
Test: build-aosp_arm64.ninja is the same before and after
Change-Id: Ib8a588e570cb2a61bab75118db6e480d6afceaa8
Since the code to start compiler proxy is implemented in soong now
(https://android-review.googlesource.com/c/platform/build/soong/+/839293).
We do not need to do that in goma.mk.
Bug: 118390303
Test: After staring goma client, executed followings:
Test: USE_GOMA=true make -j 4
Change-Id: If3da9908483e183f539c0d6ef9dc5826200b0d6b
Signed-off-by: Yoshisato Yanagisawa <yyanagisawa@google.com>
Dump the list of APKs that aren't located at system partition and signed
with system certificate.
And when enforcement option is enabled, it makes build error if there is
the apk that satisfies the condition above.
Bug: 74699609
Test: m -j
Test: m out/target/product/$(get_build_var TARGET_DEVICE)/certificate_violation_modules.txt
Change-Id: I23c41f2665dd97abac3e77d1c82d81ff91b894eb
vtsc just ignores the -d argument and apparently can't write a depfile.
So for now, just stop assuming it produces one.
Bug: 121058584
Bug: 120496070
Test: NINJA_ARGS="-w missingdepfile=err" m libvts_interfacespecification
Change-Id: Ie1483fbeead9f710d8a6636df07ce44bbc4a6e2d
It is a list of <module_name>:<manifest_name> pairs. When the module
name of an APK or an APEX matches with <module_name>, then its app
manifest name is overridden to <manifest_name>.
<module_name> and <manifest_name> can be patterns as in
com.android.%:com.mycompany.android.%.release
Test: m with PRODUCT_MANIFEST_PACKAGE_NAME_OVERRIDES for
1) an APK in Android.mk
2) an APK in ANdroid.bp
3) an APEX
and check that manifest names are modified as specified
Change-Id: Ic09e059ea7b7ea99a50a1cf423a25aa587ef4466
Data Binding only works as part of tools. It also only builds there
(in studio-master-dev branch). This CL removes references to the data
binding checkout which will be followed by removal of data binding
from master manifest.
Bug: 122478832
Change-Id: I338cc1aa10a5ad39114bc0ed14a529365016e9a4
strip.sh uses $2.d to write the depfile, while this had been passing in
$@.tmp in as $2. Instead, move the file first, then use strip.sh with
the real output file so that we can find the depfile.
Also don't specify a depfile (or other dependencies) when
LOCAL_CERTIFICATE==PRESIGNED, as we'll skip all of that in the rule in
that case.
Bug: 121058584
Test: NINJA_ARGS="-w missingdepfile=true" m
Change-Id: I1323e775e3a6ce5a06a19d59f3fb07495f116f34
We used to build retrofit full OTA package as long as
PRODUCT_RETROFIT_DYNAMIC_PARTITIONS is true. This doesn't work with AOSP
targets that have the flag set but without any available vendor image.
This CL detects such a condition and uses a separate flag to guard the
retrofit OTA building as well as the split super images generation.
Bug: 120852744
Test: `m -j dist` on blueline (w/ vendor images) and aosp_blueline (w/o
vendor images).
Change-Id: I65726f24f8fc546be6802941a6a06590a3804c16
It was referring to a non-existant macro paramater resulting in empty
output.
Test: m out/target/common/obj/PACKAGING/hiddenapi-greylist.csv
& then check output.
(cherry picked from commit 8626c8ea95)
Change-Id: Ib48042ecca320c4d7c3a27959e41e5c55099b27a
Merged-In: Ib48042ecca320c4d7c3a27959e41e5c55099b27a
Caught by https://android-review.googlesource.com/861510
Bug: 121058584
Test: m out/target/common/obj/APPS/Gallery2_intermediates/renderscript/rs.srcjar
Test: NINJA_ARGS="-t deps out/target/common/obj/APPS/Gallery2_intermediates/renderscript/rs.srcjar" m
Change-Id: I92a803963f4431e8506b1f2517a7186d16667709
Overlooked the transitive dependency on this lib in my previous CL that
packs libfec.so.
Bug: 122280472
Test: Build `otatools.zip`. Unzip and check that
`ldd bin/delta_generator` no longer shows any missing lib.
Change-Id: I7b633ea870f05b86ee7760c736dac7c74dee5dc2
This allows us to have vendor-specific product makefiles for building
unbundled artifacts. The makefile could be used to specify path to the
keystore used to sign the artifacts for the vendor.
Bug: 121224311
Test: m
Change-Id: I260bfddc5fc438fc0986b39d9010087bf758c7ca
delta_generator now has a dependency on libfec.so [1], for generating
A/B incrementals.
[1] https://android-review.googlesource.com/c/platform/system/update_engine/+/854154
Bug: 122280472
Test: `m -j otatools-package` with aosp_marlin-userdebug. Check that
`otatools.zip` contains libfec.so.
Change-Id: Ia6642bb2c7ccb0f89ad3079bdf65e4dbffee13f5
These replace $USER and `hostname`, which will soon become stable values
due to sandboxing the build on Linux.
Bug: 122270019
Test: check build.prop
Change-Id: I7493e45a2a2defbdba5d82596cb91d68480f0187
Apicheck seems to only be used for signature file
conversion and diffing at this point.
Test: make dist
Not Yet Merged In Ib8c09ba7f85fbb48526132ca8cc85856d63ca407
Change-Id: I5ca70a019ef4f7828f0a566f7745fc2dbf673b2e
This is just a hash of the current.txt and system-current.txt right now,
though it may expand to include other API surfaces in the future.
Once prebuilts/sdk/current is populated with api_fingerprint.txt files,
we can use those for unbundled builds, but until then, just use the
PLATFORM_PREVIEW_SDK_VERSION as a placeholder.
MD5 was just the most convenient, since we don't have a sha tool that we
can use on Mac currently. I'm hoping we can get a toybox-based tool in
the future that standardizes that, but we aren't currently using sha*sum
from toybox on Linux, much less on Darwin yet.
Test: m dist out/dist/api_fingerprint.txt
Test: m out/target/product/.../system/build.prop
Change-Id: If69f270560d05135cb81a9bb2d1b208ea78f86df
Native tests (BUILD_NATIVE_TEST) use their own MODULE_CLASS.
Check for it when selecting the linker for ASAN.
Test: build a native test, readelf -l <test> | grep linker
Change-Id: I34ca8c443c792bdf8b4b1fa812806c56f13a72d0
Bug: http://b/116873221
The coverage runtime (libclang_rt.profile) needs stderr, which is
exported by bionic only after API level 23.
Test: Build with coverage on globally
Change-Id: I1bdfbbf248d1743d529e9f40f0b52da92bb20ef7
It's desirable that the background texts display in the center of the
screen. So we will add the flag "--center_alignment" when generating
these images.
Bug: 121280655
Test: check and run the graphic tests
Change-Id: I6ba73ba7ef130288e950f95748e0e3fe28aa0827
unzip prints a warning and exits with code 1 when a zip file is empty.
Pipe the output of unzip through grep -v to hide the warning,
using PIPESTATUS to preserve the exit code of unzip, and then
filter out exit codes that are not 0 or 1.
Test: m
Change-Id: If2cf725fc3315d5996d5a7190288df46d84a59e2
If an AOSP target is built with dynamic partitions support but without
vendor.img available at build time, don't write 'vendor' to
dynamic_partitions_info.txt that will be used for building OTA packages.
We can't remove the partition from BoardConfig file, because the name
will be still needed in other places, such as generating the partition
metadata (super_empty.img). Otherwise manually flashing vendor.img at a
later point would require additional steps (e.g. `fastboot
create-logical-partition`).
Bug: 120852744
Test: `m -j dist` with and without vendor projects.
Change-Id: Ia1c3ed5bbea0255f1908958140dac97a1597586d
Mainline modules are tightly coupled to the platform, and should
build against the current SDK from source and not prebuilts. Add
a flag UNBUNDLED_BUILD_SDKS_FROM_SOURCE to specify that a
TARGET_BUILD_APPS build should build the current SDK instead of
using the prebuilts.
Bug: 121194841
Bug: 121231426
Test: no change to out/build-aosp_sailfish.ninja
Test: forrest unbundled build
Test: forrest master apps build
Test: forrest mainline modules build
Change-Id: I9ebc08745409a817d831817cb282aba1de6d81bb
It is useful to have the exact dex2oat invocation used to generate
artifacts for some tests. This adds the --write-invocation-to= flag to
the build to save this information.
Test: m droid
Bug: 119332327
Change-Id: Ie17ac73c790b0c55ef5569f03d11e7f03b5ef035
Building $OUT/obj/JAVA_LIBRARIES/core-tests_intermediates/javalib.jar
is no longer updating the common intermediates:
out/target/common/obj/JAVA_LIBRARIES/core-tests_intermediates/javalib.jar
out/target/common/obj/JAVA_LIBRARIES/core-tests_intermediates/classes.jar
after I25163e91886cea6941afa25cdb529ed053278dcb. Add the missing
dependency on common javalib.jar, which already depends on common
classes.jar.
Bug: 119412419
Bug: 121194944
Test: rm out/target/common/obj/JAVA_LIBRARIES/core-tests_intermediates/classes.jar &&
m core-tests &&
ls out/target/common/obj/JAVA_LIBRARIES/core-tests_intermediates/classes.jar
Change-Id: Id463d159e46c82656c364fdb2dbe53f33ee1a103
Don't create the install rules for boot.art when WITH_DEXPREOPT
is not set to true, which will ensure there is no dependency on
boot.art in a normal build, which matches the behavior before
I25163e91886cea6941afa25cdb529ed053278dcb.
This should fix the Mac build, which never sets WITH_DEXPREOPT
and does not support running dex2oat.
Bug: 119412419
Bug: 121130576
Bug: 121183382
Test: lunch aosp_sailfish-eng && m WITH_DEXPREOPT=false
files does not depend on out/target/product/sailfish/dex_bootjars/system/framework/arm/boot.art
Change-Id: I39e96b1f35f6808d3007cf1785ec3db89e0decb6
Move the dexpreopting logic into Soong. Make modules will be
dexpreopted by executing the Soong logic in the standalone
dexpreopt_gen binary, which will generate scripts that will
perform dexpreopting for each module. Export global configuration
as JSON to $OUT/dexpreopt.config, which will be used by
dexpreopt_gen and Soong, and per-module JSON configuration that
will be used by dexpreopt_gen.
This relands I59b20c931ee3e5a8d35eb30da4148691c5095502,
I39d580999947ee54cfefe875b57a028be5333bd7,
Ie7daa94e107d53eff075ca58dbe721bd9d7fc8c2 and
Ica006a007d112c232311435aaac0c0e476232b67, with a minor
update to match the changes made to dexpreopt_gen arguments
and a fix to correctly keep dexpreopt disabled on mac builds.
Bug: 119412419
Bug: 120273280
Test: no differences to dexpreopt outputs on aosp_sailfish system/,
only expected changes to dexpreopt outputs on system_other
(.vdex files for privileged Soong modules no longer incorrectly
contain .dex contents).
Change-Id: I25163e91886cea6941afa25cdb529ed053278dcb
Ie7daa94e107d53eff075ca58dbe721bd9d7fc8c2 attempted to pre-create
the directories that unzip would unzip into, but incorrectly
created them in the top of the source tree instead of in
$(PRODUCT_OUT). Fix the directory location and add a cleanspec
to clean up the incorrect directories in the source tree.
Bug: 119412419
Test: m correctly cleans up incorrect directories in source tree
Test: extract unzip command from out/verbose.log.gz and add bash -x
Change-Id: Ica006a007d112c232311435aaac0c0e476232b67
unzip sometimes fails with:
checkdir error: cannot create out/target/product/.../system/framework/oat
File exists
I think this happens when two unzips run in parallel, see that the
parent directory is missing, race to create it, and then one of them
treats the EEXIST when creating it as an error instead of continuing.
Work around this by creating the directories with mkdir -p before
running unzip.
Test: m installclean && m
Bug: 119412419
Change-Id: Ie7daa94e107d53eff075ca58dbe721bd9d7fc8c2
uncompress-prebuilt-embedded-jni-libs uses zip2zip, add a dependency
to the rule that calls it.
Test: treehugger
Change-Id: I39d580999947ee54cfefe875b57a028be5333bd7
build_super_image.py calls `lpmake` and expects that (i.e.
HOST_OUT_EXECUTABLES) in PATH. However, we have to explicitly set up
that inside the build.
Bug: 120553014
Test: Set OUR_DIR and build blueline-userdebug.
Change-Id: I000f9d069902a902c9504a2ba31dba6dbc6917a2
BOARD_USES_<X>IMAGE is not equivalent with that image actually
existing. Since these targets depend on the INSTALLED_<X>IMAGE
variables being defined, gate the conditionals on that instead.
This logic is used in other places in Makefile
Bug: 120974093
Test: make
Change-Id: I6766042c801bf7fddc4b84671dbe5d3a157aea6b
Move the dexpreopting logic into Soong. Make modules will be
dexpreopted by executing the Soong logic in the standalone
dexpreopt_gen binary, which will generate scripts that will
perform dexpreopting for each module. Export global configuration
as JSON to $OUT/dexpreopt.config, which will be used by
dexpreopt_gen and Soong, and per-module JSON configuration that
will be used by dexpreopt_gen.
Bug: 119412419
Bug: 120273280
Test: no differences to dexpreopt outputs on aosp_sailfish system/,
only expected changes to dexpreopt outputs on system_other
(.vdex files for privileged Soong modules no longer incorrectly
contain .dex contents).
Change-Id: I59b20c931ee3e5a8d35eb30da4148691c5095502
Add a LOCAL_SANITIZE_NO_RECOVER variable that allows specifying which
sanitizers running in diagnostics mode shouldn't recover. This can help
debugging as we test enabling sanitizers in new libraries since it'll
cause tombstones to be generated along with the diagnostics information.
Bug: 80195448
Bug: 110791537
Test: Compiled test module with this flag, checked compiler command.
Test: Test module crashed, tombstone contained diagnostics information.
Change-Id: I441b9c873e54bf6404325f4d0ac59835350c2889
Generate the images for recovery's wipe data menu. And use the lossless
tool zopflipng to compress the generated image. This saves space
in the boot image at the cost of the increase of build time. In practice,
it adds about ~20 seconds to build the bootimage with 32 threads.
zopflipng generally runs slower. Even with a predefined filters
strategy, it runs about 3 times slower than pngcrush, e.g. ~40s vs 15s
to compress a 1.8M png file. However, zopflipng also gives a better
compression: ~1.4M vs 1.6M for the same image. So I guess the time
consumption is a price worth paying.
Bug: 74397117
Test: build the bootimage and check the resource images
Change-Id: I3e71f8bedd3c183a3374b1a9464bf05500e1aa61
We've already added new APIs in Q, so we shouldn't be claiming a
value of "0", which signals a production platform release.
Bug: 120797950
Test: manual
Change-Id: I6c9063ac4ace79b3ee6b1c67ccdf76f4063ad165
The final APK may contain compressed or uncompressed dex files,
and this affects dex2oat compilation.
Test: build sailfish, test that SystemUI.vdex and Settings.vdex
don't have the dex code.
Change-Id: Ic032f61ed67dcdf342f6eaef71d805b89d6fc99a
We used to manually generate the background text images for recovery. That
process requires opening various emulators and takes a long time. Now that
the host Java tool is ready to generate these images, we can embed the
generation step into the build of the recovery image.
Since there's currently no make variable that defines screen width of the
device, we put the commonly used values for pixels here, i.e. 1440 for
xxxhdpi and 1080 for xxhdpi. Devices with other dpi can still use this
feature if they define the screen width in their config files.
Test: build bootimage with screen with set/unset
Change-Id: Ic8389a527e6e7194de58a9c50763e841e886652b
Same as soong. This needs do match, otherwise, for example, CFI may be
disable in a static library in soong, and left enabled in a shared
library in make; that would not work as CFI only supports DSO granularity.
Bug: 120508119, 112709969
Change-Id: I00d6b1c9c373bcb6804c135407c6eeae88b375b6
Test: hwasan build of master branch boots
This adds support for overlays to be placed in the product and
product_services partitions.
LOCAL_VENDOR_MODULE := true #/vendor/overlay
LOCAL_ODM_MODULE := true #/vendor/odm/overlay
LOCAL_PRODUCT_MODULE := true #/system/product/overlay
LOCAL_PRODUCT_SERVICES_MODULE := true #
/system/product_services/overlay
Bug: 110869880
Test: built overlay with different partitions and verified correct apk
location
Change-Id: I21f594e784a18cf59f4ff03c00e30b463ec85008
* New clang compiler makes some integer santizers enabling
implicit-integer-sign-change, but Android code does not
boot with this new sanitizer yet.
Bug: 119329758
Test: build and boot with new clang compiler
Change-Id: Ic80cde49d3ef51277fbe2a0aa8c1b8f2f8bfd80c
Use the new script to build super_empty and super split images. No more
transformation to lpmake_args.
Test: build target_files_package for retrofit device
Change-Id: Id5f6bd607654ca869bcdf58d86b7ae300e3927eb
Use the script to build it to avoid having duplicate logic in
build-superimage-target-args and build_super_image.py
Test: build it
Change-Id: I82f74d56d744c5f5b8dd38d41fedd264b042bc7a
For non-retrofit (launch) devices, super.img is used for factory, so
source images should be from target_files.
In this change, build-superimage-target procedure is converted to a
more flexible script so that it can be built.
Bug: 119322123
Test: build target files for device launch with dynamic partitions
Change-Id: I6ee0cc3e145357dfc74be248f81f5f8f4e51fc5c
Bug: http://b/117796718
ucrt is either preinstalled in newer Windows or can be installed in
older Windows, whereas a dependence on msvcr110 doesn't work on Windows
without additional dependencies.
Test: Run generated binaries on Windows VM (7, 8, and 10) without
installing any extra dependencies.
Change-Id: If5a5ffaa9d822ca05b4b54896937f224103f33d7
Combine existing hiddenapi-* txt files into a single CSV containing
a list of all class members and their corresponding flags.
Test: m, phone boots
Change-Id: I67f1b5f331b262e9b681d9a3c78c9e4cbb5ea082
ODM can already be built with dynamic sizes (in that
build_image.py builds it correctly). Add it to the list of
valid dynamic partitions.
Test: builds
Change-Id: Ib5d1348530d2120e3af7dbe422ac3ebe6149b4fb
Fixes: 120047099
Pass --avoid-storing-invocation to dex2oat, which will remove the
command line from the final .odex files. This will help make the
.odex files reproducible across devices and build system changes,
simplifying comparisons.
Bug: 119412419
Test: m checkbuild
Change-Id: I46cedadab9a87addc0f9e35f7bbe578beae1df60
This was only needed for pushing first stage init to the recovery
image. This isn't actually needed however, as the recovery image can
be fully shared and can run full init as long as there is a symlink
from /init to /system/bin/init, so that is added instead.
Bug: 80395578
Test: boot to recovery via this symlink
Change-Id: I8910543525a841401b209bfd98af30c5a0dc2688
AArch64 execute-only memory is only supported when using lld as the
linker. There's still a few modules which don't use lld, so in those
cases we need to disable this option.
Bug: 77958880
Test: Module with LOCAL_USE_CLANG_LLD false builds without XOM
Test: Module without LOCAL_USE_CLANG_LLD defined builds with XOM
Change-Id: I5dfe3cd1e1cac2b3ead13912af9f7fe896d507e9
Only use one consecutive tab in make rules and use spaces
for any additional indentation. This should generally prevent
tabs from appearing on the cmdline.
This cl is the result of 3 regex replaces:
\\\n\t\t([^\t]) -> \\\n\t \1
\\\n\t\t\t([^\t]) -> \\\n\t \1
\\\n\t\t\t\t([^\t]) -> \\\n\t \1
Test: make mainline_system_arm64
Change-Id: Ic3f72deed35e39e4b1a4b492db65aeabc5bef0d5
Like TARGET_SYSTEM_PROP, but is merged into /product/build.prop.
Also change the formatting of the output of both of these macros
slightly, so make the start and end clearer.
Bug: 119911662
Test: migrate downstream products to use this
Change-Id: Iddae7aac3c51b5706f3fb2690c5ca5a98840ad34
Adds build system support for generating AArch64 binaries with
execute-only memory layouts via a new LOCAL_XOM property. Also adds
support for an ENABLE_XOM build flag for global builds.
Bug: 77958880
Test: make -j ENABLE_XOM=true
Change-Id: I6af9e3615d0a9fdff802eae50e6ad94311ec8046
When PRODUCT_RETROFIT_DYNAMIC_PARTITIONS is empty, collapse-pairs will
write this line to build.prop:
ro.boot.dynamic_partitions=ro.boot.dynamic_partitions_retrofit=
... which is not what we want.
Test: build without dynamic partitions
Change-Id: I04c3c75fad20604657814b7f9cce2cc0635627ee
Test: with this is true, build
`verified_assembled_system_matrix.xml`;
file includes kernel requirements
Change-Id: I833b4fefabefb2f0457171055afe5722981e3589
Treehugger Robot