As part of the work to modularize the hiddenAPI processing the
generation of the monolithic hidden API index file needs to be moved
to the platform_bootclasspath module type. Doing that broke the
TestBootDexJarsFromSourcesAndPrebuilts tests which checks the inputs to
the rule that creates that file. Fixing that required added a
platform_bootclasspath module to the test fixture for those tests which
highlighted an issue with the prebuilt_apex module.
Previously, when the prebuilt_apex created apex variants it would use
its own name as the apex variant name, even when that name included the
prebuilt_ prefix. That broke the platform_bootclasspath logic as it was
looking for apex variants for "myapex" but the only ones available were
"prebuilt_myapex".
This change ensures that it always uses the unprefixed name and fixes
the TestNoUpdatableJarsInBootImage to match. This also adds some
improved error reporting in platform_bootclasspath which helped debug
this problem.
Bug: 177892522
Test: m nothing
Change-Id: I3e88b5cec767f77dcc0e94b3ae38b499d07eadf0
Boot jars, updatable boot jars and art apex jars are part of two
separate but related configuration objects, the main Config struct
(actually the nested productVariables struct) and the dexpreopt
specific GlobalConfig. The fields in both are initialized from the same
data in the make config files but handled separately.
Previously each test that used one of the configuration objects would
generally just initialize the one it used. That would make the test
sensitive to the specific configuration object that was used. A
refactoring that change the code from using one configuration object to
the other would cause the test to fail.
Also, some tests would inadvertently create invalid configurations by
setting ArtApexJars without also setting BootJars. While the ability to
create invalid configurations is useful (and there are some tests that
exist to verify the behavior in that case) most tests should not be
using them.
This change simplifies the configuration of the tests and improves
their realism by:
1. Providing a new FixtureConfigureBootJars method that takes a set of
boot jars and sets ArtApexJars, and BootJars in the
dexpreopt.GlobalConfig and BootJars in the product variables too.
2. Providing a new FixtureConfigureUpdatableBootJars method that takes
a set of boot jars and sets UpdatableBootJars in both the
dexpreopt.GlobalConfig and productVariables.
3. Migrating existing tests to use these new methods.
Some tests still use the dexpreopt.FixtureSet...Jars() methods directly,
generally to create invalid configurations.
Bug: 177892522
Test: m nothing
Change-Id: I4d8f0b9762cfcc7ae6383bef08563d7c3fa13955
Adds the fragments property to the platform_bootclasspath to allow the
fragments that contribute to it to be specified.
Bug: 177892522
Test: m nothing
Change-Id: I14f83d9336f6984442c7315cc86dfdd0a0fd2d20
Adds a FinalDeps mutator to add dependencies from the
platform_bootclasspath to the configured boot jars which can be from
either the platform or any apex. It adds dependencies for every
configured boot jar, whether in ArtApexJars, BootJars or
UpdatableBootJars.
At the moment the dependencies are only used for testing purposes
but following changes will make more use of them.
Bug: 177892522
Test: m nothing
Change-Id: I981305bf45bc20539a3d36987252f490e2b885cc
This change fixes a bug that shared lib dependencies of an rlib is not
installed to the APEX even when the rlib is part of the APEX.
Bug: N/A
Test: m
Change-Id: I88fe461584499839d8018d6b4292374592e7562b
Previously, SdkSpec was constructed only from the user string. It didn't
make use of the Config struct where information about the latest stable
SDK version, etc. is recorded. As a result, the build system couldn't
check if the sdk version "current" is referring to the in-development
(i.e. not-yet-frozen) SDK version or the latest stable version.
"current" was always assumed to be in-development (IsPreview() returns
true) even when Platform_sdk_final == true.
As the first step for fixing that, this change requires
android.EarlyModuleContext to be passed when constructing SdkSpec from
the user string.
In the following changes, "current" will be mapped to either
FutureApiLevel (10000) or one of the FinalApiLevels() depending on
whether the platform SDK was finalized or not.
Bug: 175678607
Test: m
Change-Id: Ifea12ebf147ecccf12e7266dd382819806571543
The property init_rc accepts paths and references to other modules;
however, none of that is passed onto make, resulting in errors if a
non-local path is used.
Test: m
Bug: 184567830
Change-Id: Idbbf9de66c5182784d055e1cd64bcef34a7dbd79
This follows the same pattern as is used by the prebuilt_apex in order
to make files encapsulated with a prebuilt .apex file available for use
by the rest of the build. It separates the extraction of the
appropriate apex from the zip supplied to apex_set into a separate
module type whose output is used by the apex_set and will (in a follow
up change) be used by the deapexer module too.
Bug: 181267622
Test: m droid
Change-Id: Icd13ce1f21845494e875d3e092abd4a9d2efca57
This dedups the apex selection (which was previously being done by both
the apex.Prebuilt module type and the Deapexer module type) by
separating it out into its own module.
The apex.Prebuilt module retrieves the selected apex from its
selected_apex property which is simply initialized with
":<selector-module-name>".
The Deapexer module retrieves the selected apex it should use from its
src property which is initialized in the same way. That makes it easy
for the Deapexer module to be reused by the apex_set in a similar way.
Bug: 181267622
Test: m droid
Change-Id: I90cfb55698d35a97dcf28b95afcb1f58584bc90c
Previously, the private deapexer module type was treated as a prebuilt
as that was the quickest way to implement it. However, there is no
reason why it should be as there is no corresponding source module type
which it might replace.
This change removes the prebuilt support to simplify the module type in
preparation for future work. As part of that it extracts the logic for
retrieving a single path from a PrebuiltSrcsSupplier to ensure
consistent behavior.
Bug: 181267622
Test: m droid
Change-Id: I1b2bb49601c13a3fff99a0026c235eaa70d2e612
Previously, if an appropriate src property was not specified it would
return "" which resolves to the top level directory. This change causes
it to report an error.
Bug: 181267622
Test: m droid
Change-Id: Ia5be324a0eff18e43b352d71c6768c8767986053
When `test_for` dependencies are added from libraries to APEX modules,
they can be created from the platform variants of the libraries, since
those are used for building tests. Hence we need an alias from the
platform variant of the APEX module to have a target for those
dependencies.
This is only necessary for libraries that are split by the APEX
mutator, i.e. is a member of some APEX. Normally that's not the case
for test libraries, but there may be exceptions (read
com.android.art.testing).
This relands https://r.android.com/1654679 after decoupling it from the
topic that caused b/184239856.
Test: m nothing
Bug: 183882457
Change-Id: If643c75ce9bc25fa01ad9d1e3ba8e1d060d03bb2
All version names will go through ApiLevelFromUser which triggers an
error when the name is not a valid one.
Bug: 175678607
Test: m
Change-Id: Id33bf64085603914d45ad7942cb8908a4734493f
Remove Android-targeting gcc toolchain libraries -- libgcc,
libgcc_stripped, libatomic, and libgcov. Also remove libunwind_llvm,
which is replaced with a libunwind toolchain prebuilt.
The __atomic_* library functions are now part of the compiler-rt
builtins library:
https://android-review.googlesource.com/c/toolchain/llvm_android/+/1625025
Bug: http://b/153025717
Test: treehugger
Change-Id: I971d0a4a49f1aaeb3546e80b6d94208277a171ac
... in preparation for making the handling of sdk versions consistent
across java and cc modules.
Bug: 175678607
Test: m
Change-Id: I598f0454bce9b7320621022115412fbe97403945
This reverts commit 6d15d63556.
Reason for revert: Broke git_sc-mainline-prod on test_suites_x86_64
Fixes: 184305592
Change-Id: I4b2e2675e0dd9e2e84966f545a52f45d5b731bab
Revert submission 1658000
Reason for revert: Breaks full-eng build: b/184239856
Reverted Changes:
I4f8ead785:Avoid internal APEX stubs for libsigchain and clea...
I68affdf69:Allow dependencies from platform variants to APEX ...
I54b33784e:Rename libdexfile_external_static to libdexfile_st...
Id68ae9438:libdexfile_external is being replaced by libdexfil...
I12ac84eb4:libdexfile_external is replaced by libdexfile.
If05dbffc8:Rename libdexfile_external_static to libdexfile_st...
Ia011fa3a8:Merge libdexfile_external into libdexfile.
Change-Id: If494dc5385042a4620a76a9eadc1613ae0eb1655
So far, when a library `libfoo` has `stubs.versions: ["10", "11"]`, then
`shared_libs: ["libfoo"]` is linked to the version 11 of the stub.
This requires the author of `libfoo` to manually update the property
whenever a new version is introduced. Otherwise, clients are not able
to use the newly added APIs because the latest stub is for an old
version.
This change eliminates the need for manual updating. "current" version
is always implicitly added to `stubs.versions`. It is added even when
nothing is set on the property, if `stubs.symbol_file` is set. i.e.
```
cc_library {
name: "libfoo",
stubs: {
symbol_file: "libfoo.map.txt",
// no versions: [...] needed
},
}
cc_library {
name: "a_client",
shared_libs: ["libfoo"],
apex_available: ["myapex"],
min_sdk_version: "29",
}
apex {
name: "myapex",
native_shared_libraries: ["a_client"],
min_sdk_version: "29",
}
```
`a_client` links to the "current" stub of `libfoo` that has all symbols
shown in the map file.
Note that, above doesn't mean that the client has unlimited access to
APIs that are introduced even after the min_sdk_version of the client
(29 in this example). The use of such APIs still has to be guarded with
`__builtin_available` check.
Bug: N/A
Test: m
Change-Id: I70bb1600c18e74d36c6b24c3569d2149f02aaf96
Adds PrepareForTestWithNeverallowRules to make it easy to test
neverallow rules. Avoid exporting any unnecessary neverallow related
methods from the android package.
Bug: 181070625
Test: m nothing
Change-Id: Idfc6955cb23f1a4d1790be7879388154b03f3980
Deprecated the method to try and prevent any other uses being added.
Bug: 183650682
Test: m nothing
Change-Id: Ia6f43851e5a00c9d96af780e3bd21e03175e1a2f
Fixes the few tests that break due to this and which cannot easily be
separated into their own changes.
Bug: 183650682
Test: m nothing
Change-Id: Ia2f31213a1f114a78e66a81d89279ecde9f4c465
When `test_for` dependencies are added from libraries to APEX modules,
they can be created from the platform variants of the libraries, since
those are used for building tests. Hence we need an alias from the
platform variant of the APEX module to have a target for those
dependencies.
This is only necessary for libraries that are split by the APEX
mutator, i.e. is a member of some APEX. Normally that's not the case
for test libraries, but there may be exceptions (read
com.android.art.testing).
Test: m nothing
Bug: 183882457
Change-Id: I68affdf69d7ec05c0ee8730e8ec04d7cb9e0e44a
* changes:
Make test_for arch variant.
Don't use APEX stubs between internal libs in the same APEX when building test_for modules.
Add FilterListPred.
Don't panic on "go test" invocations from the command line.
