cc modules have complex image variant settings. To avoid setting
wrong partitions to modules, add a test to check the install
partition.
Bug: 184885453
Test: m nothing
Change-Id: I080661335f2da7281d6e6a0d2fedd1fcecb4b2cb
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
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
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
Test behavior was changed a while ago so that tests by default ignore
non-existent source paths (unless they explicitly check for/rely on
them). Prior to that CheckSnapshot() could detect when files were
missing from the snapshot but it no longer can.
This change disallows non-existent source files in all the sdk tests
which means that they are disallowed when processing the snapshots as
they use the same preparers as were used to process the sources.
This caused a test failure which has been temporarily ignored and has
a TODO and bug associated with it.
Bug: 183184375
Test: m nothing
Change-Id: I969d8515d20ef5ae515f2b5f93d8ed4e4f8ede75
Background: `min_sdk_version` of a crt object is the oldest SDK version
that the crt object supports. When it's set to for example 16, Soong
creates a number of versioned variants of the crt object starting from
16 to the latest version. The variant for version X is provided to NDK
clients having `min_sdk_version` set to X.
Problem: all versioned variants of a crt object were built with `-target
<arch>-linux-android16`. Therefore they all have been with `#define __ANDROID_API__
16`. This is because the mutated variants still have the same
min_sdk_version property and the clang triple follows min_sdk_version,
not sdk_version. This is too conservative and against our intention to
provide the latest crt object that matches with the min_sdk_version of
the client.
In the other hand, the platform(non-sdk) variant of the crt object
doesn't have such a problem. min_sdk_version is completely ignored.
However, this is a bug and will be fixed by aosp/1640364. As a side
effect of the fixing, the platform variant will begin to built with a
very old __ANDROID_API__ which unnecessarily turns off the new platform
features like the TLS segment over-alignment.
This change fixes the problems:
* For the versioned variants of crt objects, we set both
`min_sdk_version` and `sdk_versio` to the version that the variant is
created for.
* For the platform variant of crt objects, `min_sdk_version` is force
reset to "current".
Bug: 183191008
Test: m
Change-Id: I8c9d0fcea816de8cd1532dac4a47eee4f726c037
Background:
When compiling cc_* modules, min_sdk_version determines the "version"
part of the clang triple: -target <arch>-linux-android<version>. The
version part is used to make sure that calls to the APIs that are added
after the version are guarded with a runtime check (i.e.
__builtin_available).
Previously, min_sdk_version was used as the version part only when the
cc_* module is unbundled (e.g. built for an APEX or built with SDK). In
other words, min_sdk_version has been ignored for the platform variants.
They were built with the version number 10000.
This was problematic for Mainline module tests. Since they are neither
part of an APEX nor built with SDK (because they need to have access to
some of the module-only APIs), they are built with the version number
10000. As a side effect, __builtin_available macro are expanded to 1 at
build time - because 10000 is higher than any API versions. When such a
test built in the latest platform source tree runs on a device running
an old platform, it tries to call an API that might not be available on
the platform and experience a crash, due to the lack of the runtime
check.
This change fixes the problem by using min_sdk_version as the "version"
part of the clang triple, regardless of the module's variant. Then
__builtin_available macro in the tests doesn't expand to 1, but to a
function call to the libclang_rt.builtin library which checks the system
property ro.build.version.sdk.
Bug: N/A
Test: run resolv_stress_test
Change-Id: I88f64c5a35f1b5276c3350e177b116932011a940
__ANDROID_VNDK__ is defined for the modules that are able to use the
VNDK libraries. As both product and vendor variants define
__ANDROID_VNDK__, we don't know if a module is built for vendor or
product on build time.
__ANDROID_VENDOR__ and __ANDROID_PRODUCT__ macros can be used to
specify the image-variant-dependent codes.
Bug: 180646847
Test: m nothing
Change-Id: Id6c3e1e3d47deaf3684c0c02964718658cf2fec5
Restructures the cc package test setup code to create FixturePreparer
instances for setting up a test fixture and converts some tests to
use it.
The goal with this change is not to switch all the cc tests over to
directly using the new model but instead to ensure that the majority of
the cc tests run with the new model, to allow existing tests to easily
switch to the new model when needed and to allow dependent packages to
be switched to the new model.
Bug: 181070625
Test: m nothing
Change-Id: I00415f10fb44c1b9e78e1317e7f50bb61984d3a4
Previously, a cc library that included .aidl, .proto and/or .sysprop
files and exported headers generated from at least one of those types
would actually export generated headers from all of them.
While headers generated from .sysprop files are always exported those
generated from .aidl or .proto should only be exported when explicitly
requested.
This change treats them separately as expected. It has the potential
to break the build as it could reduce the set of headers exported and
so a dependent module that needed those would break. The fix in that
case is to simply add one (or both) of the following to the module
that previously exported those headers:
aidl: {
export_aidl_headers: true,
}
proto: {
export_proto_headers: true,
}
Bug: 180712399
Test: m droid
Change-Id: I488182e27dd423d261443612f98d5c112dd3ef8f
Add a ctx parameter to AndroidMkExtraEntriesFunc to allow them to
access providers.
Test: m checkbuild
Change-Id: Id6becc1e425c3c3d8519248f8c0ce80777fac7cc
Merged-In: Id6becc1e425c3c3d8519248f8c0ce80777fac7cc
Improves the test coverage for exporting of generated headers. These
tests highlight the bug described in b/180712399.
Follow up changes will refactor and fix bugs in this code. Adding the
tests separately help ensure that those changes do not inadvertantly
change the behavior.
Test: m nothing
Bug: 180712399
Change-Id: I0225b0cf53259071edb99a94be5014ed0e019bde
When using the directed vendor snapshot, the build fails because soong
fails to find some variants of the modules that have been excluded
from VENDOR_SNAPSHOT_MODULES, even though those modules are not going
to be used by the build.
The solution implemented here is to generate fake versions of those
modules (empty files) and include them in the generated Android.bp,
so that soong finds the modules, even though trying to use them would
fail.
Bug: 171821997
Bug: 179275601
Test: source build/envsetup.sh
Test: m -j nothing
Change-Id: Ibd3e963ab3e5504c0ac817f7cabbd241bf47a5cb
When a module has both 'product_specific: true' and
'vendor_available: true', the module does not have core variant but
have product and vendor variant modules. As the module generated
multiple variants, the names of the modules have the variant name
suffix. However, it did not create the module with the base module
name because it does not have core variant.
As the base module, in this case, is the product variant, use the
base module name for the product variant without adding the
'.product' suffix to it.
Bug: 161913599
Test: m hidl_test_product@2.0-adapter-helper
Change-Id: I904b43744579a6e89cecbc2e529122eaa8b14cae
The module names for vendor and product variants have the image
variant suffix to avoid conflict with the core modules. It requires
updating the module names in the dependency tree with the suffixes.
We had a hidden bug that used the original module properties to
update the names of its dependent modules.
Also, it must cover the product variants modules.
Test: updated cc_test.go and build
Change-Id: I6b4ea062d13c8fac1e699138d44376e52e0d7852
Move the vendor snapshot tests into vendor_snapshot_test.go to
reduce the size of cc_test.go
Bug: 177098205
Test: go test ./build/soong/cc/...
Change-Id: Ib1120f986fc0111a9487fce00539a4c5ae7e2e87
*_available will not provide if a module is VNDK-private or not.
Fixed it by using IsVndkPrivate() with private() interface.
By allowing exclude_from_vendor_snapshot for vendor_available
modules, available() interface is no longer required. Tests are
updated for exclude_from_vendor_snapshot on vendor_available modules.
Bug: 175768895
Test: m nothing
Change-Id: I9ff85256170d59b378a4c354d5c007cfbeacabc5
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
Vendors can now generate only needed modules by setting the following
Makefile variables:
- DIRECTED_VENDOR_SNAPSHOT: set to true
- VENDOR_SNAPSHOT_MODULES: list of snapshot candidates
e.g.
DIRECTED_VENDOR_SNAPSHOT := true
VENDOR_SNAPSHOT_MODULES := toybox_vendor sh_vendor libbase libcutils ...
Bug: 157967325
Test: m dist vendor-snapshot after setting those in BoardConfig.mk
Change-Id: I6515a43d9358d94483d7c7fa2b066f9dd457f6aa
The check no longer tracks private dependencies like stubImplDepTag and
staticVariantTag. It also doesn't do the check for double_loadable
libraries that are not depended on by a LLNDK library.
Bug: N/A
Test: m
Test: add 'double_loadable: true, gen_trace: true` to an aidl_interface
module. The build doesn't break.
Change-Id: Iccd1a9d445a48d03c373708ba1bdd34b9a7f152d
Move the global lists of VNDK modules into boolean properties on the
modules themselves, and use the new SingletonModule functionality
to visit all modules and collect the list of modules with the properties
set.
Bug: 176904285
Test: all soong tests
Test: m checkbuild
Test: compare Soong outputs
Change-Id: Icf7e2f8f190a517d30f0780e98762bc0084ddb24
Memtag_heap adds an ELF note that enables MTE heap tagging in
bionic/scudo. Ignored on non-executables. With diagnostic
(diag:{memtag_heap:true}) enables the SYNC mode, otherwise - ASYNC mode.
Memtag_heap defaults to set (with diag) on cc_test targets, unset
otherwise. Ignored on non MTE-compatible hardware.
Bug: b/135772972
Test: soong tests
Change-Id: I88fd0f159e609e17bd13487749980a1ba02cb91c
The LLNDK vendor variants need to exist even when the VNDK is not
being used in order for the next patch to list them once the global
maps are removed.
Bug: 176904285
Test: m checkbuild
Change-Id: Ib29ede455d5b6a4b7d3f4685db8fba6d32025314
In case of VNDK, 'vendor_available: false' had a special meaning that
defines VNDK-private libraries. It is not trivial because not
defining a boolean property means 'false' normally. To avoid the
confusion replace it with the 'vndk.private: true' for VNDK-private
libraries and 'private: true' for LLNDK-private libraries.
All VNDK libraries must define 'vendor_available: true' and may have
'vndk.private: true' if they are VNDK-private.
With this change '(vendor|product)_available: false' is the same as
not defining the property.
LLNDK-private must define 'private: true' instead of
'vendor_available: false'.
Bug: 175768895
Test: build
Change-Id: I57fbca351be317257d95027f3cdcdbbe537eab23
A fake vendor snapshot is a vendor snapshot whose prebuilt binaries and
captured headers are all empty. It's much faster to be built than the
real vendor snapshot, so users can exploit the fake vendor snapshot to
reduce the size of vendor snapshot they need, by installing the fake
snapshot and then inspecting the ninja dependencies.
Bug: 157967325
Test: m dist vendor-fake-snapshot
Change-Id: I5e16e8dbbf9dd5e753cdd471ca73d06984a6cb2c
Replace the vndk_libraries_txt module type with llndk_libraries_txt,
etc. in preparation for making it a new SingletonModule, which will
only work with a single instance of the module type.
Bug: 176904285
Test: m checkbuild
Change-Id: Ie010a9eeee8f5849201aa4ab4eb9b2e7a9cd7d5b
Justin Yun