no_default_compiler_flags is only used by the crt* modules,
is unnecessary, and causes problems when necessary flags like
-no-canonical-prefixes are not passed. Remove the property.
Use useVndk() instead of noDefaultCompilerFlags() to determine
if adding libc as a dependency is necessary and won't cause a
circular dependency.
Bug: 68719465
Test: m checkbuild
Change-Id: Iea1a082dc701dfeab211049a22f7066257347b80
This reverts commit 33c252c2f1.
I have a fix to the crashes that this CL set was causing, and have
uploaded it as a patchset to this revert.
This also contains a partial fix that was initially submitted
separately as
https://android-review.googlesource.com/#/c/platform/build/soong/+/524295/
Bug: 30227045
Test: ./art/test/testrunner/run_build_test_target.py -j110 art-asan #no
build errors
Test: m -j50 ASAN_OPTIONS=detect_leaks=0 SANITIZE_HOST=address # no
build errors
Change-Id: I3e53549fa03413d35d9a952f04de1e7629e1f06d
This reverts commit d4b484b070.
Rationale: second in group of commits that left aosp_x86_64 not
building. (See https://android-build.googleplex.com/builds/
submitted/4426589/aosp_x86_64-eng/latest/logs/build_error.log)
Bug: 30227045
Test: builds
Change-Id: I38ab5284c614d6ee68e7359219bd75c7d50131be
This CL makes multiples changes in preparation for platform-wide CFI.
(a) Adds a second -version-script=... to the command line
when building components that use a version script. This ensures that
__cfi_check is also exported, and allows CFI to be enabled for these
components.
(b) Adds both topdown and bottom up mutators for CFI to help propagate
dependencies correctly for components that may need CFI disabled.
(c) Fixes an issue with the mutators to correctly apply settings to
both generated variants
(d) Fixes issues when components have more than a single visibility
flag.
Bug: 30227045
Test: SANITIZE_TARGET=cfi m -j40 # dependencies are correctly built
# with/without CFI
Change-Id: I44793cc03bcbcdaa957cc49c7240b87d7c9db327
Added export_llndk_headers properties to llndk_library module. And a new
module type llndk_headers is added. This is to enable an LLNDK library
to reexport other LLNDK headers.
Bug: 65395259
Test: do the following
// frameworks/native/libs/arect/Android.bp
llndk_headers {
name: "libarect_vendor_headers",
export_include_dirs: ["include"],
}
// frameworks/native/libs/nativewindow/Android.bp
llndk_library {
name: "libnativewindow",
....
export_llndk_headers: ["libarect_vendor_headers"],
}
check that
-Iframeworks/native/libs/arect/include is in LOCAL_EXPORT_CFLAGS of
libnativewindow.vendor in out/soong/Android-<product>.mk
Change-Id: If1650414b2967f2042f4ebe2b593ed3f3ea45d3a
Also adds checks that the dependencies are android.Modules and
are not disabled.
Test: m checkbuild
Change-Id: I05e945f38915d49cd3c0ab72a86576949bc7eff2
Move firstUniqueElements to android.FirstUniqueStrings,
lastUniqueElements to android.LastUniqueStrings, and lastUniquePaths
to android.LastUniquePaths.
Test: m checkbuild
Change-Id: Ieac840405126c7f8f98afb4a4ef35c01a18fe7fb
This uses knowledge of transitive dependencies to reorder
linker command line arguments such that if module A depends
on module B, then module A is automatically listed before
module B in the linker command line.
This should mostly remove the need for Android.bp files to
list all of their static dependencies in link order
Bug: 66260943
Test: reorder the entries of static_libs in an Android.bp and see that linking still succeeds
Change-Id: I20f851ab9f2f30031254e4f30023b6140d15d6c3
linkageMutator removes srcs property of the shared variant of a lib in
order to reuse *.o files compiled for the static variant also to the
shared variant.
However, this causes problem when vendor-specific srcs are specified in
target: {vendor: {srcs: ["..."]}}. For example, let's assume
cc_library {
name: "libfoo",
srcs: ["foo.c"],
target: {
vendor: {
srcs: ["bar.c"],
},
},
}
Then,
static_vendor: inputs = foo.o, bar.o
shared_vendor: inputs = foo.o (from static_vendor), bar.o (from
static_vendor), bar.o
So, bar.o is included twice and this causes multiple symbol definition
error.
In order to handle the problem, vendor mutator is applied before the
linkage mutator and the vendor-specific srcs are squashed in the vendor
mutator.
Bug: 67731122
Test: build
Test: cc_test.go
Change-Id: I2a5390295dddfc41260e9b6f02746908cdf47228
When a lib is explicitly marked as `vendor_available: false`, then it
can't be directly depended by a vendor lib which is installed to /vendor
partition. This is to hide some VNDK libs (including llndk) from vendors
so that platform owners can have a freedom of modifying their ABI
without breaking vendors.
In addition, the list of the private libs are exported to the make world
as VNDK_PRIVATE_LIBRARIES.
Also, fixed a bug that allowed a vndk lib to link against to vendor
library (or vendor variant of a system lib) if the lib is prebuilt.
Bug: 64730695
Bug: 64994918
Test: Add `vendor_available: false` to libft2 and libcompiler_rt.
Add the libs to shared_libs property of a vendor library in soong
(i.e. libnbaio_mono). The build fails with the error message.
Change-Id: Iab575db96bb4f6739a592f3fa0a75124296bea0c
Bug: http://b/65598278
The profile runtime depends on libgcc for some symbols (only under some
circumstances - armv5, ndk r14, static executables). Since Android
build passes -nostdlib and adds libgcc manually, the profile runtime
gets passed to the linker later than libgcc.
Instead, explicitly add the profile runtime to the linker command (and
pass one other flag added by the clang driver to the link).
Test: Build a library with profile instrumentation that otherwise fails
instrumented build.
Change-Id: I24b34cebd2c3bb6a540f8f4c465ace1be4eb90f3
Signed-off-by: Pirama Arumuga Nainar <pirama@google.com>
This reverts commit 2370af0e23.
Reason for revert: New Build Breakage: aosp-master/aosp_arm64_ab-userdebug @ 4376965
Change-Id: Ibe4b819c4292457c454bf42e6d94fba3071ec04b
Bug: http://b/63768402
Bug: http://b/65598278
Add support for the 'pgo' property to specify how a module is processed
under PGO. A sample property is below:
pgo: {
instrumentation: true, // could be "sampling: true" when supported
profile_file: "pgo_simple.profdata",
benchmarks: ["pgo_simple"],
}
1. Runtime profiles can be gathered using "sampling" or
"instrumentation". Sampling is not supported initially.
2. If 'toolchain/pgo-profiles' project is found,
'toolchain/pgo-profiles/${profile_file}' is passed to the compiler and
linker when building this module.
3. If ANDROID_PGO_INSTRUMENT environment variable is set, and includes a
benchmark in the 'benchmarks' list, appropriate flags (for e.g.
-fprofile-generate for instrumentation) are passed to the compiler and
linker when building this module.
Test: Add example modules that specify the pgo property and verify
appropriate flags and dependencies in the Ninja file. Some
tests/examples are in https://android-review.googlesource.com/474805
Change-Id: I6242e0c904497a115e367dea6927ba1c4b906355
The linux kernel requires that the ELF interpreter (runtime linker)
that's referenced by PT_INTERP be either an absolute path, or a relative
path from the current working directory. We'd prefer a relative path
from the binary, similarly to how we handle looking up shared libraries,
but that's not supported.
Instead, extract the load sections from the runtime linker ELF binary
and embed them into each host bionic binary, omitting the PT_INTERP
declaration. The kernel will treat it as a static binary, and we'll use
a special entry point (linker_wrapper) to fix up the arguments passed by
the kernel before jumping to the embedded linker. From the linker's
point of view, it looks like the kernel loaded the linker like normal.
Bug: 31559095
Test: Enable host bionic,
out/soong/host/linux_bionic-x86/nativetest64/libdemangle_test/libdemangle_test
Change-Id: I8d0aea9790b5e86fcc3ea6e2d00cfa33907e2853
cc_genrule is the same as a normal genrule, but can depend on other cc
modules (like cc_object).
Test: mmma external/minijail
Change-Id: I8df87665c7bdc76ce89c92755c054f967a818e57
So that we don't get confused when using :<module> in srcs to depend on
a module that could also be a HostBinTool.
Test: m -j
Change-Id: Ia3b1c26826e70f84c6dc5ff78c95dd11d76901b6
Currently, abi is dumped for platform variants of system libraries.
Dump them for vendor variants since they are the ones which need to be abi
stable on security updates. This also ties abi dumping to BOARD_VNDK_VERSION.
Test: For libfoo:
1) Added a source file with a dummy function for the core
variant.
2) Excluded the source file from the vendor variant.
3) BOARD_VNDK_VERSION=current mm -j64 produces libfoo.so.lsdump with no
dummy_function since the source file was not included in the vendor
variant.
Test: Inspected build.ninja and confirmed that all of a library's abi
dump dependencies (.sdump files are dependencies of the final .lsdump files) are
from vendor variants.
Change-Id: Ie0bf91fcd81606c131845d9872261166b5db72aa
launcher.
For Python2, we bundle embedded launcher as bootstrapper within every
.par file. This feature is only enabled for linux_x86_64 for now. We
provide a user flag: hermetic_enabled within bp file. By default, Pyhon2
still use classic bootstrapping way to construct .par file and relys on
host interpreter. Once embedded_launcher is enabled, launcher will be
used to bootstrap .par file and execute user program.
For Python3, the launcher will be ready soon, and for now it still relys
on classic bootstrapping.
Test: Real example is used to test.
Bug: b/63018041
Change-Id: I28deba413d8ad3af407595e46f77d663e79a3705
All the stats below are for a hikey960-eng build on AOSP master.
Number of order-only inputs in out/soong/build.ninja:
Before: 2847162
After: 606508
Size of: Before After
out/soong/build.ninja 572MB 233MB -59%
out/soong/Android-hikey960.mk 15MB 9MB -40%
out/build-hikey960.ninja 367MB 358MB -2%
Ninja time in `m nothing`:
Before: 6.9s
After: 5.0s
Soong generation time:
Before: 11.6s
After: 6.6s
Test: m nothing
Test: wrote script that counted inputs, only duplicates were removed
Test: treehugger
Change-Id: I6b7c5ef5b1395014b7bf5fd0b8112d42bee127bf
header-abi-diff has switched to recursive diffing of types and can
diff more accurately if types included in re-exported shared library
headers are included as well.
Bug: 62060883
Test: make -j64, android.hardware.light@2.0.so.lsdump contains structs
from libcutils as well (libcutils' headers are re-exported by
android.hardware.light@2.0)
Change-Id: I814819e4f7258b4b380350fe8ad0ccf8dbd5ce5c
Right now this is hard coded in two places (here and in
core/binary.mk). Keep it in one place so it's easier to change.
Test: make checkbuild
Bug: None
Change-Id: I2a2c784d4c667b326f871e6144db92753c16c85f
Enabling the lto property for a module builds that module and all static
dependencies with LTO.
LTO (link-time optimization) allows the compiler to optimize and
generate code for the entire module at link time, rather than
per-compilation unit. LTO is required for Clang CFI and other
whole-program optimization techniques. LTO also allows cross-compilation
unit optimizations that should result in faster and smaller code, at the
expense of additional compilation time.
Test: make -j12 libc with lto: true for libc
Change-Id: Ib8baefedf60e02701d44673a7c473e0845730101
LL-NDK, VNDK-core, VNDK-SP libraries are exported to make as
SOONG_LLNDK_LIBRARIES, SOONG_VNDK_CORE_LIBRARIES, and
SOONG_VNDK_SAMEPROCESS_LIBRARIES. This can be used to auto-generate
ld.config.txt from a template.
Bug: 64013660
Test: BOARD_VNDK_VERSION=current m -j successful
Test: check out/soong/make_vars*.mk and look for SOONG_*_LIBRARIES
Merged-In: I0f4c5d05d9cd28c3fc9fdcca6ce0e6eaeaacbe8d
Change-Id: I0f4c5d05d9cd28c3fc9fdcca6ce0e6eaeaacbe8d
When the lib is vendor-only, then .vendor suffix is not added.
Furthermore, this change correctly adds .vendor suffix even to the names
listed in LOCAL_SHARED_LIBRARIES so that we don't need to add the suffix
in the make world.
This also allows us to use the original name (without the .vendor
suffix) of the vendor-only modules in make (e.g. in PRODUCT_PACKAGES or
as a make target).
Bug: 37480243
Test: BOARD_VNDK_VERSION=current m -j <name> is successful, where <name>
is one of the vendor-only libraries in Soong. (i.e.
android.hardware.renderscript@1.0-impl)
Test: m -j does not break anything
Change-Id: I203e546ff941878a40c5e7cfbb9f70b617df272d
'vndk' tag must be parsed in cc_defaults as well.
Test: build with a module that has 'vndk' tag in cc_defaults.
Change-Id: I21a5a831dbf8cae00be98c00f4bb98504da53c8d
InitDefaultsModule and InitDefaultableModule were requiring the
module to be passed in as an argument twice, once as a
android.Module and once as an android.DefaultableMOdule. Use
a type assertion instead.
Test: builds
Change-Id: I57d721e68a4f016e02c17a678af5f2ba25a9a430
If BOARD_VNDK_VERSION is set, and a module is set to
`vendor_available: true` it is installed in /system and /vendor.
However, if the module is a VNDK library, it must be
installed at `/system/${LIB}/vndk` instead of /vendor/${LIB}.
For those modules, need following to set.
vendor_available: true,
vndk: {
enabled: true,
support_system_process: true,
},
`support_system_process` is optional to define.
If it is defined to true, the module is regarded as vndk-sp.
link-type check for VNDK modules is added to make sure that VNDK
modules only link to other VNDK shared libraries or LL-NDKs.
move the ABI checks to VNDK from all of vendor_available.
Bug: 38304436
Test: attempt to compile with BOARD_VNDK_VERSION:=current
Test: Use `vendor_available_vndk: true` for VNDK modules and compile
with BOARD_VNDK_VERSION:=current
Change-Id: I409268e0b7f05a9d01697bf9f9f4726b5aac631f
If a vendor (or vendor_available) lib has a dependency to a llndk
library, the name of the llndk library is modified to <name>.llndk.
There was a bug that this name modification is done only for shared_libs
property, but not for export_shared_lib_headers property. As a result,
the headers of the llndk was not re-exported. Fix the bug by modifying
export_shared_lib_headers property as well.
Bug: 63178444
Test: build
Test: look into out/soong/Android-<target>.mk. libutils.vendor has
system/core/liblog/include_vndk in its LOCAL_EXPORT_CFLAGS.
Change-Id: I6fcb1036cdb81ba167ead815b436a5cada283ede
Change module factories from returning a blueprint.Module and a list
of property structs to returning an android.Module, which holds the
list of property structs.
Test: build.ninja identical except for Factory: comment lines
Change-Id: Ica1d823f009db812c518f271a386fbff39c9766f
Add a list of flags which are not understood by clang LibTooling tools
and filter them out of the Cflags the tools are invoked with.
Test: In frameworks/av, make libmedia vendor_available (this invokes
header-abi-dumper on this module), mm -j64.
Bug: 62447349
Change-Id: I46f017212b89f4331145c999103d0ed44da0abaf
We now add export_static_lib_headers, export_generated_headers to the
filters while dumping the abi of a library using header-abi-dumper
(through -I<dir> additions to the invocation of header-abi-dumper and
header-abi-linker)
Also add support for zipped reference source based abi dumps.
Test: mm -j64 in hardware/interfaces/nfc/default/1.0 produces
android.hardware.nfc@1.0.so.lsdump with abi filtered out using generated
headers.
Test: Copied the linked abi dumps produced by mm -j64 in bionic/libc to
prebuilts/abi-dumps/ndk and gzipped them. Ran mm -j64 again in
bionic/libc and verified header-abi-diff getting invoked.
Bug: 32750600
Change-Id: I26210af908c87a6143e39fa25f50307acb68a387
Soong command lines have gotten very long due to hidl modules
reexporting lots of libraries. Dedup the include dir flags.
Test: m -j checkbuild
Change-Id: I6ada1251012da42344e2c00ae66001a649023d2c
The deps have to be exported alongside the headers to make sure
the headers are in place when a module that depends on the shared
library builds.
Bug: 37555583
Test: use protos in a cc_library
Change-Id: I8f5b8d7937868458c1cf02b990a348aa395634da
hasSrcExt is used to determine extra flags and dependencies when
generated files are used. If the generated files are being handled
in a static library whose objects are reused in a shared library,
the flags and dependencies still need to apply. Instead of clearing
the source files in the shared library, move them to an OriginalSrcs
property, and check that in hasSrcExt along with the Srcs property.
Also pass extra exported include directories from the static
library to the shared library.
Bug: 37555583
Test: use protos in a cc_library
Change-Id: I709779ec03b66b220b7bd58a1f6f0b9b5067d955
header-abi-dumper: dumps abi exported by source files for Vndk.
header-abi-linker: links abi dumps produced by header-abi-dumper.
header-abi-diff: compares linked dumps.
Test: mm -j64 showcommands > make_log in bionic/libc.
This produced linked dumps in out/soong/.intermediates.
Copied these dumps to
prebuilts/abi-dumps/ndk/current/arm64/source-based/.
Changed the abi and re-ran mm -j64 showcommands > make_log
confirmed that the build reported compatibility breakge without
actually failing (advisory mode).
Change-Id: Iccad6908fe68a80f47230751671d156893b96ead
Nothing changes if BOARD_VNDK_VERSION isn't set.
When the VNDK is enabled (BOARD_VNDK_VERSION in Make), this will split
/system and /vendor modules into two different variant spaces that can't
link to each other. There are a few interfaces between the two variant
spaces:
The `llndk_library` stubs will be available in the /vendor variant, but
won't be installed, so at runtime the /system variant will be used.
Setting `vendor_available: true` will split a module into both variants.
The /system (or "core") variant will compile just like today. The
/vendor ("vendor") variant will compile against everything else in the
vendor space (so LL-NDK instead of libc/liblog/etc). There will be two
copies of these libraries installed onto the final device.
Since the available runtime interfaces for vendor modules may be
reduced, and your dependencies may not expose their private interfaces,
we allow the vendor variants to reduce their compilation set, and export
a different set of headers:
cc_library {
name: "libfoo",
srcs: ["common.cpp", "private_impl.cpp"],
export_include_dirs: ["include"],
target: {
vendor: {
export_include_dirs: ["include_vndk"],
exclude_srcs: ["private_impl.cpp"],
srcs: ["vendor_only.cpp"],
},
},
}
So the "core" variant would compile with both "common.cpp" and
"private_impl.cpp", and export "include".
The "vendor" variant would compile "common.cpp" and "vendor_only.cpp",
and export "include_vndk".
Bug: 36426473
Bug: 36079834
Test: out/soong/build.ninja, out/soong/Android- only changes due to _core addition and
.llndk -> .vendor
Test: attempt to compile with BOARD_VNDK_VERSION:=current
Change-Id: Idef28764043bf6c33dc0d2e7e2026c38867ff769
Soong can rename these automatically for internal use, so don't require
users to set these.
Bug: 33241851
Test: Android-aosp_arm.mk is the same before/after
Test: build.ninja is the same before/after
Test: build-aosp_arm.ninja is the same before/after
Merged-In: Ia92356a0ec079f908fd49812231228046783e50d
Change-Id: Ia92356a0ec079f908fd49812231228046783e50d
The vendor image will have more than just proprietary modules in it
under Treble, so let's stop marking open source code as proprietary just
to move it to vendor.
Bug: 36452052
Bug: 37134596
Test: compare build.ninja before/after, no changes.
Test: Set vendor: true, ensure it works.
Change-Id: I44b0ec7007d0e311bdcbd44b238b1ef2d05cc6ff
Instead of using the NDK headers and libraries, add LL-NDK specific
headers and library stubs for VNDK users. This allows us to provide an
expanded liblog interface.
Test: aosp_arm; m -j
Test: Enable BOARD_VNDK_VERSION on aosp_arm; m -j
Test: Inspect out/soong/build.ninja before/after (w/o vndk)
Change-Id: Ic85f07fa10c695b5baab10c41f5e0ad38700bf3d
This CL moves the location of ASAN-ified libraries on disk in the
following manner:
/data/lib* --> /data/asan/system/lib*
/data/vendor/* --> /data/asan/vendor/*
There are a couple of advantages to this, including better isolation
from other components, and more transparent linker renaming and
SELinux policies.
Bug: 36574794
Bug: 36674745
Test: m -j40 && SANITIZE_TARGET="address" m -j40 and the device
boots. All sanitized libraries are correctly located in /data/asan/*.
Change-Id: I06bf459260ec451d4495a02562f640ad622f34c2
Include order should be module includes, dependency exported includes,
and then global includes. Module includes and global includes are
computed during compileFlags, but dependency exported includes are
not handled until later. Move the global includes into a new
flags variable so that the dependency includes can be appended
to the module includes.
Test: m -j native
Change-Id: Ifc3894f0a898a070d6da8eed4f4b9e8cc0cd2523
Instead of having a module define `use_vndk: true`, assume that we're
building with the VNDK if we're installed on the vendor partition, and
BOARD_VNDK_VERSION==current. This now matches our behavior in Make.
Once BOARD_VNDK_VERSION!=current, we'll need to disable modules that
need to otherwise compile against the VNDK, since we can only compile
against the current VNDK.
Test: build.ninja for aosp_arm is the same before/after
Test: Ensure there are no boards that set BOARD_VNDK_VERSION
Change-Id: If937fa7bdb119648137af52daebadf486163484b
This was corrupting the nonvariants list with the variants list whenever
it hit the else case, causing a missing variant issue, since we need to
use the variant when linking against lib*.ndk.
Test: aosp_arm build.ninja is the same before/after
Test: https://android-review.googlesource.com/c/355163/, set
BOARD_VNDK_VERSION := current
Change-Id: I047937e838464670d0902506871d2203ec8cc17d
This reverts commit 4c48f724e1. It's
causing test failures, warnings and complaints, so backing it out and
we'll resolve those before putting it back in.
Bug: 33681361
Test: m -j
Colin Cross