This allows targets to opt-in to running R8 without the -ignorewarnings
flag. Ignored warnings can cause code to be optimized incorrectly, so
this makes the build safer for such targets.
Ideally ignore_warnings should default to false, but all targets need to
be fixed for that to happen. As a first step, provide a mechanism for
individual targets to ensure they do not introduce regressions.
Bug: 229727645
Bug: 180878971
Bug: 226127213
Bug: 239990030
Test: m (runs dex_test)
Change-Id: Ic0eef29598c1ee47e958da8a5048d9696165a235
This reverts commit 4e445be558.
Reason for revert: GtsExoPlayerTestCases now uses multidex.
Test: m + manual builds for test_suites_x86_64_coverage
Change-Id: I608b3b79137b216f546446e1c4f89f4542572581
Bug: 192032291
android_test defaults to using R8, but with shrinking, optimization and
obfuscation disabled, eliminating most of the benefits of R8. Instead,
use D8 by default, improving build performance and avoiding any other
issues that may arise in test-specific code related to whole-program R8
execution. An initial audit shows that android_test targets that *do*
enable shrinking or optimization also explicitly opt in to R8.
A follow-up CL will do the same for android_test_helper_app, but that
requires some additional auditing of downstream targets.
Bug: 192032291
Test: m + presubmit
Change-Id: I5b14a0986dde210f241a77c3a93daacf9e53d667
Standalone D8 invocations tend to be shorter-lived than R8 invocations
that involve deeper layers of bytecode optimization. Introduce a
separate set of JVM flags to pass to D8 invocations which limit the
runtime JIT tiering, reducing overall compile times and significantly
reducing total CPU consumption for D8 build workloads.
Representative incremental build times with this change:
* real 6m1.388s
* user 63m32.412s
and without:
* real 6m30.624s
* user 79m12.626s
Bug: 205303544
Test: time m
Change-Id: Ic568a4316325c6a4e0da41ffebbb07f0c5f7e8e3
Use the new --deps-file argument to collect a deps file from r8, which
will correctly track files included by proguard flags files.
Fixes: 221956283
Test: manual
Change-Id: I92341948311c2c994cc0ae032dc19e09b2e80550
r8 gets the static libs in the program jar, it shouldn't also get
them as library jars. Keep a separate classpath for dexing that
includes libs but not static_libs.
Bug: 222468116
Test: m checkbuild
Test: TestD8
Test: TestR8
Change-Id: Icca3393f496cbcadcc633f3b156761e6c145f975
R8 will parse proguard files out of resources in injars. Use the jar
with resources instead of the classes jar so that R8 can see the
proguard files.
R8 fails when an input jar contains dex files, so filter out dex files
from the input jar before passing it to r8.
This relands Ibb870ee9c70470336f542a3b7542dab86716dbf8 with a fix to
move the temporary input jar outside the output directory. Putting it
in the output directory interacted poorly with RBE in local comparison
mode, which incorrectly deleted the output directory.
Bug: 195558228
Test: m checkbuild
Change-Id: Id4b091c3bd72d4c2904883e8793ec4b9499e4b78
This reverts commit 29c294b2cf.
The deadlock in R8 was fixed, and the timeout causes problems when
RBE is enabled and the R8 action gets stuck in a queue for 30
minutes.
Test: checkbuild
Change-Id: Ia0c3826aea3288e9db3bdbd4955e1e16c0a38fe3
R8 will parse proguard files out of resources in injars. Use the jar
with resources instead of the classes jar so that R8 can see the
proguard files.
R8 fails when an input jar contains dex files, so filter out dex files
from the input jar before passing it to r8.
Bug: 195558228
Test: m checkbuild
Change-Id: Ibb870ee9c70470336f542a3b7542dab86716dbf8
R8 processes are sometimes hanging on the build servers. Wrap R8 with
run_with_timeout to dump the stacks with jstack and kill the process
after 30 minutes. Switch from running with the r8-compat-proguard
shell script to running the jar directly so that jstack gets the pid
of the java process.
Bug: 181095653
Test: m checkbuild
Test: m NetworkStackNextIntegrationTests
Test: m USE_RBE=true RBE_R8=true RBE_R8_EXEC_STRATEGY=remote NetworkStackNextIntegrationTests
Change-Id: If6996bd8eb39c7a8453d79e825004339c009ade2
SdkSpec.Version was an int type. Now it becomes ApiLevel type which
is a better abstraction of the version (or api level).
Bug: 1655587
Test: m
Change-Id: I4d67b9b9eae45f653b6af4f5b73da9e091b3dfab
... in preparation for making the handling of sdk versions consistent
across java and cc modules.
Bug: 175678607
Test: m
Change-Id: I598f0454bce9b7320621022115412fbe97403945
Remove the references to the android package in remotexec so that
the android package can reference the remoteexec package. This
will allow RuleBuilder to integrate directly with remoteexec.
Bug: 182612695
Test: m checkbuild
Change-Id: I15be5ef126d8aacbd605518638f341daf6f31bb3
Export information about java dependencies through a Provider
instead of accessing the module directly.
Test: java_test.go
Test: no changes to build.ninja
Change-Id: Ifc5d566bf6f6ebc0ad399e948effaa1ef6a22876
In order to allow the hiddenapi code which expects its inputs to be
ModuleOutPath to consume the output from the deapexer module type (used
by prebuilt_apex) this change converts the hiddenapi from using
ModuleOutPath to use OutputPath instead.
This is part of a general cleanup to remove ModuleOutPath that is
separated out in order to avoid having the hiddenapi work dependent
upon and possibly delayed by issues with that general cleanup.
Bug: 178361284
Bug: 179124768
Test: m droid
Change-Id: I890f775bf5e33528dbaa62b95fa4198185ff8bf8
proguard_usage.zip contained an empty unused.txt file when
the unused.txt file wasn't specified as an output of the r8
rule.
Test: m USE_RBE=true RBE_R8=true out/soong/.intermediates/frameworks/base/packages/Shell/Shell/android_common/proguard_usage.zip
Change-Id: Ie6a7c8b2581fe21d643e675143d6cc0e2ce80456
Use the -printusage flag in R8 to output a list of the unused
methods. Some of the files can be large (2MB for DocumentsUI,
87MB for all of AOSP), so immediately zip them and remove the
originals. The zipped files will be merged and disted.
Bug: 151857441
Test: m TARGET_BUILD_APPS=DocumentsUI dist
Change-Id: I780e84e80eba7fe4d4fa15fec0f461890afd900b
We want to support a compile_dex property for java_import. This splits
dex-related properties into a dexer struct which can be embedded in
relevant modules.
Test: m
Test: soong tests
Bug: 160455085
Change-Id: If56a51dac43f630d49483a36db29cd50e9ccd529
Test: Ran a build with `RBE_JAR="true" RBE_JAR_EXEC_STRATEGY="remote"
RBE_ZIP="true" RBE_ZIP_EXEC_STRATEGY="remote" ... use_rbe m` and that
succeeded.
Bug: b/156765207
Change-Id: Ia17b532fbb282be414adf879914870082dd0841b
ART tests need to explicitly control the compression.
Bug: 157239179
Test: "atest ArtGtestsTarget" with http://go/aog/1302773
Change-Id: I4490aa0a61fa71f830feb6a46dffd8d20fccfc66
Merged-In: I4490aa0a61fa71f830feb6a46dffd8d20fccfc66
This CL adds RBE support to javac, r8, and d8 rules which is only
enabled if respective environment variables are set.
Test: an aosp_crosshatch build with and without the new variables.
Change-Id: Ic82f3627944f6a5ee7b9f3228170c2709b1bfcb8
The value format that sdk_version (and min_sdk_version, etc.) can have
has consistently evolved and is quite complicated. Furthermore, with the
Mainline module effort, we are expected to have more sdk_versions like
'module-app-current', 'module-lib-current', etc.
The goal of this change is to abstract the various sdk versions, which
are currently represented in string and is parsed in various places,
into a type called sdkSpec, so that adding new sdk veresions becomes
easier than before.
The sdk_version string is now parsed in only one place 'SdkSpecFrom', in
which it is converted into the sdkSpec struct. The struct type provides
several methods that again converts sdkSpec into context-specific
information such as the effective version number, etc.
Bug: 146757305
Bug: 147879031
Test: m
Change-Id: I252f3706544f00ea71c61c23460f07561dd28ab0
We were passing "--classpath foo.jar --classpath bar.jar" to turbine,
it now wants (and may always have expected?)
"--classpath foo.jar bar.jar".
Test: m checkbuild
Change-Id: Ib7e0ca64a34e236110f7b785d6a0fb51ed75567e
Some module types (`android_test`, etc.) set `optimize.enabled` by
default. If such module happens to have `defaults` attribute which
clears `optimize.enabled`, the latter value is ignored.
Fixes: 129858282
Test: unit tests in java_test.go,
`atest CtsExtendedMockingTestCases` succeeds with aog/936802
reverted (that is, with cts/test/mocking converted to Android.bp)
Change-Id: Ib8e3a0ab0bd489d70ed07f626082aeae31c45e7c
Pass -XX:+TieredCompilation -XX:TieredStopAtLevel=1 to the JVM to
reduce the time spent JITing, which in local experiments reduced
the CPU time of R8 while increasing the wall time, and reduced the
total number of runnable threads.
Without the new flags:
$ time out/soong/host/linux-x86/bin/r8-compat-proguard -injars out/target/common/obj/APPS/Dialer_intermediates/classes.jar ...
real 0m30.090s
user 4m35.337s
sys 0m12.853s
With the new flags:
$ time out/soong/host/linux-x86/bin/r8-compat-proguard -JXX:+TieredCompilation -JXX:TieredStopAtLevel=1 -injars out/target/common/obj/APPS/Dialer_intermediates/classes.jar ...
real 0m35.674s
user 2m45.135s
sys 0m9.531s
The increase in wall time should be offset by the better parallelism from
the reduced CPU time.
Bug: 119270658
Test: m checkbuild
Change-Id: I8fc78b5ef1f8c6f51fd97a29461b96ae2b35cada
This can happen if Android application has only a manifest file and no
java code.
Fixes: 124483155
Test: treehugger with Android.mk converted to Android.bp in
platforms/frameworks/base/tests/privapp-permnissions
Change-Id: I45f95680b62a7a4bda0e7ab156f3df24b0dad70b
Also preserve the alignment with hiddenapi.
Test: conscrypt apex has its conscrypt.jar zipaligned.
Change-Id: I43cc0eadb9548c3d8055e12c077fb7bbe1935fe6
Enable the kotlin-annotation-processing plugin and pass annotation
processors to it.
Bug: 122251693
Test: m checkbuild
Test: TestKapt in kotlin_test.go
Change-Id: I841df454beaaa7edd263eea714ca0d958a03c9de
Privileged APKs need to store their dex files uncompressed so they
can be verified and mapped directly out of the APK.
Also track whether the module will be dexpreopted or not in order
to determine if the dex file should be stripped before signing.
Test: SystemUI.apk contains an uncompressed dex file
Change-Id: I4dca86c7f8778595882405b34adcf2a7bae03c67
The value from instrumentation_for should not go to
aapt2 link --rename-instrumentation-target-package, that should
be the equivalent of LOCAL_MANIFEST_INSTRUMENTATION_FOR. That
property is never used in Make, so it is left unimplemented in
Soong.
Add the module listed in instrumentation_for as a shared library
for javac, but don't import its resources.
Bug: 117804211
Test: m checkbuild
Change-Id: I7f035dc0ecb964a3ca391ae1ca2b87cb0f6a7cec
Using a directory to zip the output of d8 and r8 was causing the
empty classes.dex.jar file in the output directory to get
zipped in. Use a file glob instead so that it only picks up the
classes*.dex files even if classes.dex.jar already exists.
Test: m checkbuild
Change-Id: I1f3c1cbc23439fc52756aad027ab8cac10213d1d
d8 and r8 were missing dependencies on jars passed as library
arguments. This normally worked because the input jar is likely
to also depend on the same jars. When the classpath was added
to d8 it started to fail on libraries that have no code but
include other static libraries.
Bug: 116840360
Test: rm out/soong/.intermediates/external/nist-sip/nist-sip/android_common/turbine-combined/nist-sip.jar && m out/soong/.intermediates/frameworks/base/ext/android_common/dex/ext.jar
Change-Id: Ie642f4a343b949778d8ee6dcc4511f3fd7633be1
d8 requires the full classpath if --min-api requires default method
or try-with-resources desugaring. Always pass the full classpath,
since it will already have been built to generate the input jar.
Bug: 116243313
Test: m checkbuild
Change-Id: Ib0c18873880697b7cf942e0548dd2c53ee5ad23c
The change https://r8-review.googlesource.com/c/r8/+/26482 made R8
process data resources by default. This is not expected for the
Android Platform build
Test: m -j PRODUCT-aosp_x86-userdebug core-oj
Change-Id: Idb17e39dae524dacf4b258c9f27be86e3179667f
R8 no longer outputs an empty map file when minification
(obfuscation) is turned off.
Test: m -j PRODUCT-aosp_x86-eng
Change-Id: I317e703597368dc8c16751bcc8d0e43f985614fa