The behaviour is semantically identical, however,
the tool additionally sorts the symbols by address,
compresses frame unwind information more efficiently,
and improves random-accessibility for lazy decompression.
Overall, the changes balance and the output size is same,
however, libunwindstack can access the data much faster
while using less memory (due to the lazy decompression).
It will also enable further improvements in the future.
Bug: 110133331
Test: ART unwinding tests, run prefetto on the device.
Change-Id: Id48f9fe67fb67fcf2b90cc3b217b71bb8f5147ca
Remove the global list of vendor public library modules used to rewrite
dependencies from the vendor module to the stubs for system modules,
and replace it with building the stubs directly in the system variant
of the vendor module.
Bug: 178231622
Test: vendor_public_library_test.go
Change-Id: I826e69ffd507d7e85fa3d4d85b5157428c642143
There are others still to look at, but this set doesn't seem to break
anything for me.
Bug: http://b/147452927
Test: treehugger
Change-Id: I794684e23eee1326deb34bb4d5ed5831a47b8fa3
Storing ndkKnownLibs prevents multiple tests from running in parallel
as one may be writing to the list while another is reading from it.
Store it in the config so each test has its own copy.
Test: go test -race ./apex
Change-Id: Iba57c9494012c9e0ae9e5ffaa63b9b2bd2c77492
This reverts commit 2019658b86.
Reason for revert: broke builds where makefiles were using M4 without depending upon it
Change-Id: I18304246e7a8a8f8c2620dd2c65ed77acd256be2
This allows setting per-module make variables earlier in the build,
particularly for prebuilt_build_tool users like LEX/M4/BISON. I moved
filegroup over because it's a simpler common interface, but it doesn't
strictly need it.
With this, the last user of the hardcoded cc.m4Cmd variable is gone.
Test: Inspect out/soong/make_vars-*.mk, out/soong/late-*.mk
Test: treehugger
Change-Id: I195b688131feac0c100c338a0749368aa5d50f4f
This doesn't need to be manually maintained. It briefly did need to be
during the transition from the old prebuilts, but that's long gone.
Test: treehugger
Bug: http://b/113547923
Change-Id: If05633f3cf622ab39e560a3dfcc88f3eb50406bf
Exclude vendor/ and hardware/ from the manually written binder interface
whitelist.
Bug: 136279235
Test: interfaces in excluded directories are not checked against whitelist.
Change-Id: I0d640e23489b37d0c0787d5fca6bcdab10034109
This is a follow-up CL of I9a24f6975bd4b226a94f61a13d43857dcdce6b88
Generated files are passed to make via MakeVars:
- SOONG_LLNDK_LIBRARIES_FILE
- SOONG_VNDKCORE_LIBRARIES_FILE
- SOONG_VNDKSP_LIBRARIES_FILE
- SOONG_VNDKPRIVATE_LIBRARIES_FILE
- SOONG_VNDKCOREVARIANT_LIBRARIES_FILE
- SOONG_VNDK_LIBRARIES_FILE
Previously filenames were "guessed" from module names. Now VndkMutator
stores filenames as well and generate those files with them.
Bug: 142963962
Bug: 141450808
Test: m && device boots && TH
Change-Id: I0c248b707188f904df60ead50059fefe90bfd17f
Clang derives the value of __ANDROID_API__ from the triple these days. In a
future version of clang I plan to start making the behaviour of the HWASAN pass
dependent on the API level in the triple, so it's going to need to be accurate.
Test: walleye-userdebug boots
Change-Id: Ie5e36b5c8f6dcda084cc12b1160abbdf94765174
Export LLNDK_MOVED_TO_APEX_LIBRARIES which is used by make
to filter out LLNDK libs that been migrated to an APEX.
It's used by the ld.config.txt generaton, and to filter out
the vndk_package dependencies.
Test: Build&flashed crosshatch image
Bug: 137320025
Change-Id: I1449dad79edf36a628fa1ac15c35f163dbd8ee2a
We already link the shared prebuilt from the toolchain. This is
redundant and the wrong version.
Test: make checkbuild
Bug: http://b/74067984
Change-Id: I4b5d23514158eebfb4eea2aa6761e076131c3224
When HWASAN is enabled, the runtime is conceptually part of Bionic (and
mutually depends on it), so it needs to be treated in the same way as the
Bionic libs.
Now there are only two copies of the runtime: the one in
/system/lib64/bootstrap (which won't be used by ordinary processes) and the
one in the runtime APEX.
This reduces the size of the HWASAN system image and fixes an issue where
multiple copies of the HWASAN runtime were being loaded into 64-bit binaries in
APEXes because the linker namespace for the binary is different from the one
for its dependent libraries outside of APEXes. HWASAN only supports loading
one copy of the runtime per process, so this was causing such binaries to
crash on startup.
Change-Id: I228896e193a035e6dfba9f6e28d0b2e12fc163ea
Looks like the reason for it existing has been fixed. It should probably
just be removed.
Bug: None
Test: WITH_TIDY=1 m
Change-Id: I770b2fec4ac44f265ff31731c9c0bd4da14d5b0f
When no-vendor-variant VNDK is enabled, the vendor variant of VNDK
libraries are not installed. Since not all VNDK libraries will be
ready for this, we keep a list of library names in cc/vndk.go to
indicate which libraries must have their vendor variants always
installed regardless of whether no-vendor-variant VNDK is enabled.
Also add --remove-build-id option to the strip script to facilitate
the check of functional identity of the two variants.
Bug: 119423884
Test: Add a dummy VNDK library and build with
TARGET_VNDK_USE_CORE_VARIANT := true, with the corresponding
build/make change.
Change-Id: Ieb1589488690e1cef1e310669a8b47a8b8759dac
Expose all of SingletonContext to makeVarsContext, and then export
the subset of it that is used through MakeVarsContext.SingletonContext,
plus what is necessary for PathContext, directly through
MakeVarsContext.
Test: m checkbuild
Change-Id: Ie00f36e577fe110b6fa03b901da489d8547773c6
Add an opaque OnceKey type and use it for all calls to Once in
build/soong. A future patch will convert the arguments to
Once* to OnceKey once users outside build/soong have been updated.
Test: onceper_test.go
Change-Id: Ifcb338e6e603e804e507203c9508d30ffb2df966
Make no longer supports building for Windows, they must be in Android.bp
files now.
Bug: 122618577
Test: compare build-aosp_arm.ninja before/after
Change-Id: Ie70d8519850da5b1bdddf73788f984eb3f4f0900
When building with two OSes of the same class (Host, etc) enabled,
independently produce variants of each module for each OS, instead of
making `compile_multilib: "first"` select just one of the OSes.
My use case this is for Linux + LinuxBionic, which are both considered
Host class, but this would also applied if we wanted to generate rules
for Linux+Darwin, or some combination of firmware bits (either in
addition to Android in the Device Class, or multiple firmwares in a new
Class)
Bug: 31559095
Test: enable host bionic, see a linux(_glibc) and linux_bionic
definition for binaries
Change-Id: I55e7f046758b3dc2a31da2c4b20274ec68ece0a0
Scudo is now compatible with the -fsanitize-minimal-runtime, and offers a new
dynamic library that doesn't bundle UBSan.
This patch adds support for this new library in Soong, preferring it over the
full one, unless a UBSan or diagnostic dependency is found.
Test: aosp compiled with m -j
Test: local test enabling Scudo for tombstoned
Change-Id: I17794131db148b33f8a8710ac43302cadf1af314
Remove more GCC-specific flags, TARGET_(arm|thumb)_CFLAGS now contains
pre-filtered clang flags, and GLOBAL_CLANG_CFLAGS_NO_OVERRIDE now
contains all flags instead of just the clang additional flags.
Test: m
Change-Id: I61e4c5a92113d4ff3b95261783269dc1d226c57a
This variable is a space separated string of clang-tidy flags to be passed
to clang-tidy before any other system required extra flags.
Note that when this flag or local tidy_flags is defined,
the default -header-filter flag is suppressed.
Test: make with WITH_TIDY=1 WITH_TIDY_FLAGS="-extra-arg=-DABCD1=1 -extra-arg=-DABCD2=2"
Bug: 32668284
Change-Id: If7bd31c65404ef7fe6c3499d51f0f209a704efd9
It allows use of sanitizer preprocessor macros (like __has_feature())
in assembly files.
Bug: 112438058
Test: SANITIZE_TARGET=hwaddress
Change-Id: Id3fed5492ab9473cbb13cd6c0485302412e8b1de
We need to keep support in Soong until toolchain_library no longer uses
GCC.
Test: out/build-aosp_arm64.ninja is the same before/after
Test: build_test on downstream branches
Change-Id: Ib78bcd6d5544afc3a3164f65fa72a3f08b2acf29
* PATH_TO_CLANG_TIDY_SHELL is ${config.ClangTidyShellPath},
wich is "build/soong/scripts/clang-tidy.sh"
Bug: 110538415
Test: build with WITH_TIDY=1
Change-Id: I27d7be095c998583fa9ad4d705aa1332bd018140
* Some flags are accepted by clang but not clang-tidy.
They could cause the diagnostic-unused-command-line-argument warning.
Flag -flto messed up the -I flags.
* Add clang-diagnostic-unused-command-line-argument to
default clang-tidy checks.
* Move CLANG_TIDY_UNKNOWN_CFLAGS to build/make/core/clang/tidy.mk.
Bug: 111850071
Bug: 111885396
Test: build with WITH_TIDY=1
Change-Id: Iabeeb27715acf83ef6aafe3e77206b9a01a0d889
This is a new sanitizer similar to ASan, but with a few differences
from the build system perspective:
* Only runs on AArch64.
* Supports static binaries.
* Bionic itself will be built with HWASan.
* Does not have any "if a library is sanitized than the binary must
be, too" requirements unlike ASan. Even better, individual static
libraries can be freely sanitized or not. We propagate "nosanitize"
from binaries to static libraries anyway, because otherwise there
is no good way to disable hwasan on a binary as a whole.
Same a CFI, we export a list of sanitized static libraries to make.
In fact, we export separate lists for vendor and regular libraries,
because it is possible for one to be sanitized without the other
(i.e. there can be mylib.hwasan.vendor w/o mylib.hwasan or vice
versa).
Bug: 112438058, 112709969
Test: manual, part of a bigger patch set
Change-Id: Ie4fdeb522ac03cf9684526882e84dfee3807b6a7
Scudo is a hardened usermode allocator that is part of LLVM's compiler-rt
project (home of the Sanitizers). clang allows for -fsanitize=scudo as a
possible command line option to link the shared Scudo library to a binary.
This patch add Scudo as a potential sanitize option. Scudo is not compatible
with ASan and TSan and will be disabled if either is enabled.
Bug: 72112048
Test: aosp compiled with m -j
Test: local experiment with scudo: true to ensure that a test target
(tombstoned) could be linked with scudo.
Change-Id: I76bb6c60891d4782f6665a112c4c2bf7c31645da
* Pass LLVM_{OBJCOPY,STRIP} to makefile rules.
* strip.sh is used only in soong-only mode.
It will use llvm-strip and and llvm-objcopy
when --use-llvm-strip is given.
Keep flags of strip.sh in alphabetic order.
* Tested build/soong/scripts/build-ndk-prebuilts.sh
by adding "UseClangLld": true, to soong.variables file.
Bug: 80093681
Test: build/soong/scripts/build-ndk-prebuilts.sh
Change-Id: I612267304eea434c7a33cc086b27b577d5f64094
Use the host platform "strip", and also expose "nm" and "otool" used for
toc.
Test: `m PRODUCT-sdk-sdk` on Mac
Change-Id: I7a6d127f479d3e563385ead46bc3a4b49b99776b
atree was using the host's strip tool during sdk generation, so expose
the prebuilt for use instead.
`id` was used with a test running minijail0 during the build
`sha1sum` is used by development/build/tools/mk_sdk_repo_xml.sh
`rmdir` was being used to remove some tmp files
Test: m PRODUCT-sdk_arm64-sdk dist sdk_repo
Test: Run other builds through forrest
Change-Id: I49c756a3e9497c0bbd9030754e6aa193a2d989a4
* USE_CLANG_LLD is unedefined in current builds.
* When USE_CLANG_LLD is defined to 'true' or '1',
use clang's lld instead of ld or ld.gold.
* When lld is enabled:
* ld-only flags are not passed to 'lld'.
* location_packer is disabled.
* Use new lld's --pack-dyn-relocs=android.
* When lld does not work:
* In Android.mk files use LOCAL_USE_CLANG_LLD := false.
* In Android.bp files use use_clang_lld: false.
* Only arm, arm64, x86, and x86_64_devices have LLD flags;
all other hosts and targets do not call lld yet.
Bug: 73768157
Test: make checkbuild and boot
Change-Id: I06b8a1e868a600997a7e70fe05c299d751d23d5f
An upcoming change will stop exporting ProductVariables from Config, so
switch to using existing accessor functions, and add more when they're
missing.
Bug: 76168832
Test: out/soong/build.ninja is identical
Change-Id: Ie0135bdbd2df3258ef3ddb53e5f8fc00aa9b97f7
Normally, when building with VNDK, platform modules are not allowed to
link against vendor libraries, because the ABI of the vendor libraries
are not guaranteed to be stable and may differ across multiple vendor
images.
However, the vendor public libraries are the exceptions. Vendor public
libraries are vendor libraries that are exposed to 3rd party apps and
listed in /vendor/etc/public.libraries.txt. Since they are intended to
be exposed to public, their ABI stability is guaranteed (by definition,
though it is up to the vendor to actually guarantee it).
This change provides a way to make a vendor lib as public by defining a
module of type 'vendor_public_library' with a map file that enumerates
public symbols that are publicized:
cc_library {
name: "libvendor",
proprietary: true,
...
}
vendor_public_library {
name: "libvendor",
symbol_file: "libvendor.map.txt",
}
This defines a stub library module named libvendor.vendorpublic from the
map file. `shared_libs: ["libvendor"]` is redirected to the stub library
when it is from the outside of the vendor partition.
Bug: 74275385
Test: m -j
Test: cc_test.go passes
Change-Id: I5bed94d7c4282b777632ab2f0fb63c203ee313ba
* This is a workaround of llvm LTO bug.
* Keep using current NDK include directory path.
NDKTriple is default ClangTriple but kept as GccTriple for arm device.
Bug: 72619014
Test: make checkbuild
Change-Id: I5dc63c99760325c60bc2da98fd6a3125cef7267d
Elliott Hughes