I'm going to be adding some methods to this interface, give it
a name based on how it's going to be used, not based on what methods
it contains.
Bug: 323382414
Test: m nothing --no-skip-soong-tests
Change-Id: I9bba04ba756c4dbe00625e2d04af81e78a11cae9
It's been available for over a year, lets progress the python3 migration
by removing it.
Bug: 203436762
Test: Presubmits
Change-Id: I46a6477ec65e98b7e6df5cc1359f676dcbd0f08e
aosp/3255916 didn't add all of the required infrastructure.
Refactor it to be more like x86_64 so it requires less
boilerplate code.
Bug: 364383722
Test: presubmit
Change-Id: Idc5279ffc5fe964bea9f98873c01468944d90091
There are two versions of aosp extservices - one for Tplus and one for
Sminus. When prebuilts are active, the `contents` validation of both
variants of extservices bootclasspath fragments should be disabled to
account for BCP jar skew.
This CL adds ApexAvailableName to ApexInfo to achive that. If an
api_domain matching the ApexAvailableName has been selected in
`all_apex_contributions`, the source apex variants will be disabled.
Bug: 361771739
Test: lunch cf_x86_64_phone-next-userdebug
Test: m nothing with ag/28851886
Change-Id: I505a7ad8295c258e2eeef7f183669f7a75a1c88e
I'm rewriting how globs work in soong to make them compatible with
hash-based ninja implementations. As part of this, I'm removing this
unnecessary usage of globs to make them simpler.
In this case, since we're already globbing the directory at analysis
time and causing soong to rerun, we can ensure the aapt2 action also
reruns by putting a hash of the glob results onto the command line.
Bug: 364749114
Test: m framework-res, touch frameworks/base/core/res/assets/test1.txt, m framework-res rebuilt framework-res.apk, rm frameworks/base/core/res/assets/test1.txt, m framework-res rebuilt framework-res.apk again
Change-Id: I4f666367a9a0fd0dfa42dc51ef3a788a02b41747
This commit adds the libbpf_prog soong module which is heavily based on
the bpf soong module. Due to issues encountered and concerns around
compatibility and testability of mainlined bpf modules, support for
libbpf is added as a separate module.
Specifically, when changing the bpf module from
InitAndroidArchModule(... android.MultilibCommon) to MultilibFirst, the
soong apex bpfs functionality breaks. We have not identified a good
mechanism to fix this, and as we would then be creating architecture
specific bpf program binarys, additional testing and infrastructure
would be required.
It should be noted that libbpf_prog should not be used for features
which target 32bit kernel architectures at this point, as we do not
have a mechanism to identify the kernel arch at build time. I.E. 32bit
userspace could be on either 32bit or 64bit kernel arch, and we require
arch specific includes (vmlinux.h). Therefore, we default to 32bit
userspace on 64bit kernel to match the precedence set with the legacy
bpf programs.
Example use:
libbpf_prog {
name: "xyz.o",
srcs: ["xyz.c"],
arch: {
arm: {
local_include_dirs: ["vmlinux/arm64"],
},
arm64: {
local_include_dirs: ["vmlinux/arm64"],
},
x86: {
local_include_dirs: ["vmlinux/x86_64"],
},
x86_64: {
local_include_dirs: ["vmlinux/x86_64"],
},
},
}
Bug: 359646531
Test: manual
Change-Id: Ie58515d70abee061470cf4bb803228e00d496ac3
Signed-off-by: Neill Kapron <nkapron@google.com>
Revert submission 3235656-revert-3195743-compatibility_matrix-KLJOLWYIMW
Reason for revert: The root cause of the break is already be merged. Related changes: ag/29139095, ag/29139608
Reverted changes: /q/submissionid:3235656-revert-3195743-compatibility_matrix-KLJOLWYIMW
Change-Id: Ie23bdd343bebbd077553dc25f0bfd13539daf57a
Skip combining jars into turbine-combined, combined, and withres jars
and instead collect transitive jars to use in the classpath and to
produce the final dexed jar.
This reduces the size of a `m checkbuild` in git_main by 11%, from
1300 KiB to 1154 KiB. It may also improve caching and reduce uplink
network bandwidth when building with RBE, as now the classpath inputs
to rules are themselves outputs of previous rules and so already in
the RBE CAS.
The downside is that the classpath inputs to each rule are now much
longer, increasing the Soong ninja file size 11%, from 4.6 GiB to
5.1 GiB. This could be mitigated in the future by supporting something
like depsets in the generated ninja file to reduce duplication.
Bug: 308016794
Test: TestSimple, TestKotlin, TestClasspath
Flag: build.RELEASE_USE_TRANSITIVE_JARS_IN_CLASSPATH
Change-Id: I2b7b4261375494370da70f98597c8719f1d561cf
* changes:
ndk_library: limit exports
Reland "Truely re-export export_header_libs from ndk_library"
ndk_library depends on the correct arch variant of cc_library_headers
So that the build can't access extra information unintentionally.
Particuarly ANDROID_BUILD_TOP is dangerous.
In the future PATH should be locked down as well.
Bug: 307824623
Test: Added a all_genrules target and built that
Change-Id: I88bb0efb0a82529a1c85875a53cf20c8384d07fe
The NDK should stand alone, and the libbinder_ndk case
is created because libbinder_ndk in the NDK used to
contain extra C++ headers, but these were moved into
the SDK to be next to AIDL. Since many modules depend
on these headers, exports are only allowed for this case.
Bug: 357711733
Test: build with neverallow breaking, for instance:
error: frameworks/wilhelm/Android.bp:56:1: module "libOpenSLES.ndk" variant "android_x86_64_silvermont_sdk_shared_21": violates neverallow requirements. Not allowed:
module types: ["ndk_library"]
properties matching: "Export_header_libs" matches: .is-set
EXCEPT in dirs: ["frameworks/native/libs/binder/ndk/"]
Change-Id: I9a32d3cb7f644fb5dbc1e8072894f2f585b2cd6d
Pass the name of Soong's checkbuild target for each module to
Make so that it can depend on it from the main checkbuild rule.
This will give better control over which files get built, allowing
checkbuild to skip the jar combining step when transitive classpath
jars are enabled. The per-module checkbuild targets are passed to
make instead of added directly as a dependency of checkbuild in order
to maintain the existing behavior of only building modules exposed
to make in checkbuild.
Also tweak the existing calls to CheckbuildFile and add
InstallFileWithoutCheckbuild to match the files that are
in the Make-based checkbuild.
Bug: 308016794
Test: m checkbuild
Change-Id: Ic5140819381d58f4d00f23a7a12447950c4cf268
This change adds `overrides` property to all module types. It is used
to prevent another module (or modules) from being installed or packaged.
Bug: 330141242
Test: go test ./...
Change-Id: I4f05c603f0c5dbb699d00327882c7498472b59de
At ToT, platform variants of modules are hidden from make, unless the module
has `//apex_available:platform` in its Android.bp file. However, there
are special cases which always require the platform variant to be
visible to make - e.g. bootstrap bionic libraries.
`markPlatformAvailability` handles these special cases. This CL updates
the subsequent apexTransitionMutator to hide the platform variant when
module.NotAvailableForPlatform is true in addition to the existing
module type specific `AvailableFor` check
Test: no diff in file_list.txt
Bug: 281077552
Change-Id: Ie9d7341e206276ff96d3d64fec21b8f5dcfd6ceb
