This reverts commit 758140ceae.
Reason for revert: modifying glob results no longer corrupts cache after aosp/1557261
Change-Id: Id94f1f5835a93e83dc9f465627913b91908d3560
This reverts commit c2aff9e9fd.
Reason for revert: modifying glob results no longer corrupts cache after aosp/1557261
Change-Id: I682b92ad08c82ac4fa5482497cb587bd0223d21b
CC libraries generate TOC files which contain the list of exported
symbols. By depending on the TOC file instead of the .so, changes to
shared library dependencies will not result in rebuilding Rust
dependencies as long as the exported symbols remain unchanged. This
should improve incremental build times during development.
This also includes a minor fix where exported linkObjects should be
deduplicated to avoid the same object being included many times.
Bug: 173619911
Test: m libkeystore2; modify a bionic file; m libkeystore2
doesn't rebuild the rust target.
Change-Id: I6383217c125bf8dd7125a5e013a78754cac4edf2
Previously Make class loader contexts used a simplified one-level map.
Now they use the same tree representation as in Soong. The difference
between Make and Soong represenataions is insubstantial and caused only
by the language differences between JSON and Go.
Bug: 132357300
Test: lunch aosp_cf_x86_phone-userdebug && m
Change-Id: Ia273afb7f026dbe3b25d4327d55143657b026b98
This CL creates the framework necessary for generating
BazelTargetModules from regular Soong Android modules.
BazelTargetModules are code-generated into Bazel targets in BUILD files.
See the follow-up CL for examples of creating filegroup/genrule
BazelTargetModules.
Test: GENERATE_BAZEL_FILES=true m nothing # creates out/soong/bp2build
with no BUILD files, because there are no BazelTargetModules in the
module graph.
Change-Id: I33a96365bd439043b13af6db9e439592e9983188
Historically we've only had to add one major release a year to the Mac
SDK allowlist, but this year the versioning scheme changed and 11.1 is a
minor update to 11 (instead of 10.x.1 like in years past). Unfortunately
the Xcode tools don't let you just specify '11' like they did with
'10.x', so this is more complicated to fix in a scalable way.
Test: build with xcode 12.3 and macOS SDK 11.1
Change-Id: I2e88e9289727261389a863e7147a0b146ec5148a
Each variable can specify a conditions_default for properties to be used
when the variable is not set, not set to a true value (for bools), or is
set to a value that is not present in the module (for strings).
Test: m nothing
Test: go test soong tests
Change-Id: I76ec026da2369b407f0f530f77760f530e7958fc
Add gen_ndk_usedby_apex.sh script to generate NDK API list used by Mainlain modules when modules get build.
Test: TARGET_BUILD_APPS=com.android.adbd m dist apps_only
Change-Id: Ib1b2f0dd2f9ae85b1545c6cc5bb4c5bbdfac1c15
Missed a license kind needed for the large-scale-change. It didn't turn
up in testing until all the --override and --filter flags were
determined.
See: http://go/android-license-checking-v2-lsc
Bug: 151953481
Bug: 151177513
Bug: 67772237
Change-Id: Ic6b5f11717e88504e09d4817e9952d3ed1132b93
The following warnings have been turned on, and turned on as errors:
-Wbool-operation
-Wimplicit-int-float-conversion
-Wint-in-bool-context
-Wstring-compare
-Wstring-plus-int
-Wxor-used-as-pow
Test: no extra tests necessary
Change-Id: Ic8ab26de9e5323e9b81ad79eef787abfb264848c
The main difference between a singleton and a singleton module that
affects this change is that the latter requires a module to be defined
in order for the singleton to be instantiated. This change defines the
module in both the tests and the standard build.
Bug: 177892522
Test: m droid
Change-Id: Id6669835cf3a2ab7359fa53047eb9042992a9362
Previously, the createGlobalSoongConfig() function was explicitly
prevented from being used in tests because it would fail. However, it
turns out that is no longer the case and it does now work.
That allows the following changes to be made:
* Tests no longer need to use GlobalSoongConfigForTests() to
prepopulate the cache.
* GlobalSoongConfigForTests() is only needed in the dexpreopt
package.
Bug: 177892522
Test: m nothing
Change-Id: Ifcbb1a44254c5d2d10c1d02ab23227488d1d1ed1
Insulate tests that exercise code in the java package from having to
register the build components provided by the java package by providing
a single function that registers them all. This follows the pattern
currently used in the cc and rust packages.
This change is in preparation for switching the dex_bootjars singleton
from a singleton, which does not require a module definition in order
to be instantiated, to a singleton module which does. That will require
adding a module definition into java.GatherRequiredDepsForTest() and
this change ensures that the required components will have been
registered in every test.
Bug: 177892522
Test: m nothing
Change-Id: I6475db8240894947dd07c89a940a3e4f201aa598
cc_test without sanitize:memtag_heap acts as if it has implicit
sanitize{memtag_heap:true, diag:{memtag_heap:true}}. This is unaffected
by SANITIZE_TARGET.
Refactor the test to cover all permutations.
Change memtag include lists to act similar to SANITIZE_TARGET_DIAG: the
the sync include list upgrades async targets to sync, unless diag is
explicitly set to false in the target definition.
Bug: b/135772972
Test: cc_test.go
Change-Id: I6a969f2f5804cd5f47fc4e93a20e3b99ea5fa111
Calling reflect.ValueOf for every variable for every module was
using ~3 seconds of CPU time on my AOSP builds.
Test: m checkbuild
Change-Id: Idf459ad8ddf5e07f6c0df0e58e2442aaa6ab3342
Add some instructions for what to do.
Bug: 177806245
Test: m nothing (with java_sdk_library with no tracking files)
Change-Id: I286018f4d9105d5ade1e76cddb8670d0d097d77d
The vendor snapshot header globs were repeated 9 times, once for each
possible header suffix. Globbing is not a fast operation, so this
took 9.68s of CPU time in my AOSP builds. Replace it instead with a
single glob, followed by filtering down to the desired header
extensions.
Test: m checkbuild
Change-Id: I9c8f8cdad8cbffc0a42217659a48dd9d676e4e4c
Windows prefer32 was implemented in cc, which caused a mismatch
between cc and other packages like python. This manifested as
missing variant when a python_test_host with compile_multilib: "first"
tried to depend on a cc_binary_host, which implicitly has multilib
"first".
Fixes: 177904375
Test: m checkbuild
Change-Id: Iaaac16ae171c06d90d04d7cac11789d3f39b8d99
Dan Willemsen