noOverrideExternalGlobalCflags is a new set of flags intended to be
added at the end of the command line for 3rd party projects like
external/, hardware/, vendor/, and more. Our previous flags for external
projects occur way too early on the command line, leading to issues with
use for suppressing diagnostics. Note that support for this variable in
build/make is currently unimplemented, as there are no projects that
would depend on it. It could be added in the future, if it turns out to
be useful.
This change initially applies `-Wno-unused-but-set-variable` and
`-Wno-unused-but-set-parameter` to external projects because these
upstream projects aren't easily fixed, and the risk of actual bugs is
low.
Bug: http://b/197240255
Test: Build and check warnings
Change-Id: I26f56c5c52725dddb70dd8130ad61270eac7a9aa
This is configured from Make by setting up Darwin+Arm64 as a HOST_CROSS
target (which is largely true, as binaries can't be executed on X86_64
machines). On the Soong side, it's a bit blurier, as we don't current
have any other users that are the same OS but not natively executable
(Linux/musl is executable, Windows/x86 is a different OS).
Instead of requiring cc_binary/etc to become multi-architecture-aware
and using something like common_first/MultiTarget, this defaults all
non-multi-architecture-aware modules to be built with both
architectures. It then adds a dependency with the
DarwinUniversalVariantTag so that supporting modules can get the outputs
of the other variant.
Cc uses that dependency tag to run lipo on shared libraries and binaries
so that the output of the x86_64 variant is actually a fat binary
including both architectures.
Bug: 203607969
Test: build sdk-repo targets on a Mac
Change-Id: Icbddb0a177c0ba19d3e0d11f8cf568e0d1ea3245
* The obj-* targets only call clang or clang++ to compile
C/C++ files, like the tidy-* targets only call clang-tidy.
* Add (obj|tidy)-dir_os and module_os-(obj|tidy) phony targets
to include only targets of the specified OS.
* Add (obj|tidy)-dir_os_subset and module_os_subset-(obj|tidy) phony targets
to include only a subset of that OS target group.
* Most clang-tidy runs produce the same warnings for all variants
of an OS. The *_subset targets reduce runs over multiple variants.
* The obj-*subset targets are useful for a quick test to compile
at least one variant of every C/C++ file for each OS.
* The (obj|tidy)-soong phony target is changed to include only
(obj|tidy)-top_level_directory targets instead of ALL
module-(obj|tidy) targets in all subdirectories.
* For aosp_arm64-eng make tidy-* targets; count NINJA commands:
tidy-soong 381K
tidy-soong_subset 106K
tidy-soong_android 294K
tidy-soong_android_subset 73K
tidy-soong_linux_glibc 84K
tidy-soong_windows 12K
tidy-bionic 12K
tidy-bionic_subset 2.4K
tidy-bionic_android 11.5K
tidy-bionic_android_subset 2.1K
tidy-bionic_linux_glibc 1K
tidy-device_android 7K
tidy-device_android_subset 5K
tidy-hardware_android 3K
tidy-hardware_android_subset 9K
libfec_rs-tidy 33
libfec_rs_subset-tidy 9
libfec_rs_android-tidy 18
libfec_rs_android_subset-tidy 3
* For aosp_arm64-eng make obj-* targets; count NINJA commands:
obj-soong 399K
obj-soong_subset 112K
obj-soong_android 305K
obj-soong_android_subset 75K
obj-soong_linux_glibc 90K
obj-soong_linux_glibc_subset 38K
obj-soong_windows 12K
obj-bionic 15K
obj-bionic_subset 3K
obj-bionic_android 14K
obj-bionic_android_subset 2.3K
obj-bionic_linux_glibc 1.3K
obj-system-core 6K
obj-system-core_subset 3K
obj-frameworks-base 16K
obj-frameworks-base_subset 6K
libfec_rs-obj 33
libfec_rs_subset-obj 9
libfec_rs_android-obj 18
libfec_rs_android_subset-obj 3
Test: NINJA_ARGS="-n" WITH_TIDY=1 make some_obj_tidy_target;
compare output commands from various phony targets
Bug: 199169329
Bug: 202769240
Change-Id: I186c45dc07c5884888c1063dfc09cf212ffb0ebf
* shared libraries and binaries can depend on NDK share libraries
Bug: 208310402
Test: WITH_TIDY=1 make tidy-soong
Test: WITH_TIDY=1 make droid
Change-Id: I88ca8555c62d5ca49e36ca089749722166afec30
Write toc files that list the exported symbols so dependents are
only rebuilt if the exported symbols change.
This exports the CC function TransformSharedObjectToToc, and also
removes an unused arg from its signature.
Bug: 178185435
Test: New Soong test passes.
Test: m <toc file path>
Change-Id: I7ab69bf7e7f32f25eb4c7ca9d18d877dac1511db
We're now packaging llvm-windres (aosp/1870074) which can be used as a
substitute for GNU binutils' windres, but we've also removed the last
consumer of .rc files in aosp/1867048 (external/mdnsresponder/).
In the worst case, we can bring back support for .rc files but instead
process them with llvm-windres.
Bug: 203794880
Test: lunch aosp_cf_x86_64_phone-userdebug && m
Change-Id: I30a873537e004b46a87191a932d076a89463c873
* RBE adds implicit dependent files as inputs
and creates too long commands.
* Adds .tidy files as validation paths.
* Change type of tidyFiles to WritablePaths.
Bug: 203577539
Test: lunch aosp_cf_arm64_phone-userdebug; USE_RBE=true WITH_TIDY=1 make
Test: USE_RBE=true WITH_TIDY=1 mm in external/skia
Change-Id: I5fc9d6f8c9c58abcd3721ca9fbae1bc785e4e609
Used as a supplement to C/C++ srcs to disable
clang-tidy for selected srcs, when a library
contains many files in srcs and only some of them
are too large to compile with clang-tidy.
Test: WITH_TIDY=1 TIDY_TIMEOUT=90 make tidy-soong
Bug: 198098397
Change-Id: Ib32eb0e46ddbc717999797717bfd8c57e182ee88
* After this change, make libtinyxml2-tidy should generate
correct out/.../libtinyxml2.tidy.d file.
Touching one of the dependent files, e.g., tinyxml2.h,
should invoke clang-tidy but not to recompile tinyxml2.o.
* Note that linking libraries still depend on .tidy targets,
so make libtinyxml2 will call clang-tidy, regnerate
tinyxml2.tidy.d, tinyxml2.tidy, tinyxml2.o.d, tinyxml2.o.
* A clang-tidy rule uses the same $cFlags for both clang and clang-tidy.
To share it and avoid over long command lines, we use modern shell
array variables and cannot use old bourne shell.
Bug: 199169329
Test: WITH_TIDY=1 make
Test: WITH_TIDY=1 make libtinyxml2-tidy | grep tinyxml2
Test: touch external/tinyxml2/tinyxml2.h; make libtinyxml2-tidy | grep tinyxml2
Change-Id: I6175add58d7313ee50c9308b78c9290a60770052
* In builder.go, share common flags in a module.
* This replaces the sharing of cflags/cppflags/asflags in cc.go.
* A unit test in apex_test.go now fails and is commented out.
It is a failing test hidden by old optimization in cc.go.
* In module.go, expand the reference variable $someflags<n>,
or ${someflags<n>} to keep many existing unit tests work as is.
* The build.ninja size was reduced from 8.1GB to 6.2GB,
for aosp_arm64-eng WITH_TIDY=1 USE_RBE=true,
and from 7.5GB to 5.6GB when USE_RBE is 0.
Content of build.ninja is also more readable and searchable.
Read/write build.ninja times are also reduced,
depending on disk I/O speed.
Test: make WITH_TIDY=1
Change-Id: I17f96adf4844136d52e5d40f57a19d9e290162b7
* Tested with a clang-tidy wrapper that dumps TIDY_TIMEOUT.
Test: make WITH_TIDY=1 TIDY_TIMEOUT=42
Bug: 199451930
Change-Id: I86580225bad7487da786db2ba62c09bd99d70c67
* When WITH_TIDY=1, these targets allow quick check of C/C++
source code with clang-tidy, without building C/C++ binaries.
* For each module with tidy rules, add a module-tidy target, e.g.,
libart-tidy, libartd-tidy, bionic-benchmarks-tidy, libnativehelper-tidy, etc.
* Add a tidy-soong phony target that depends on all module-tidy targets.
* For each directory X/Y add a tidy-X-Y phony target that depends
on all *-tidy targets in X/Y and tidy-X-Y-Z for all X/Y/Z directories,
e.g., tidy-bionic, tidy-bionic-benchmarks, tidy-libnativehelper, etc.
* Only soong modules are collected for now.
Tidy rules in .mk files will be collected later.
* Some comment lines reformatted by gofmt.
Test: WITH_TIDY=1 make <some_module>-tidy tidy-<some_directory>
Test: WITH_TIDY=1 make tidy-soong
Bug: 199169329
Change-Id: I45aef3875f70288a8e070761e5f083dbbdfa6e94
We now default to lld for platform builds, removing all need for this
property. For more details, see "Effecient archive file handling" in
https://lld.llvm.org/NewLLD.html#key-concepts
Bug: 189475744
Test: Manual build succeeds
Change-Id: If1104d68b13de8c7afab35c1741a68f64394b448
This commit adds `--root-dir . --root-dir $OUT_DIR:out` to the commands
of ABI tools. If the user sets non-default $OUT_DIR, the tools write
"out" instead of the actual $OUT_DIR to the ABI dumps.
Test: export OUT_DIR=~/aosp_out ; \
make libstagefright_bufferqueue_helper.vendor
Bug: 192157880
Change-Id: I4e2d1b252273106fb6a3ecfa1727d5c91ba03444
Rename the ninja variables that have an extraneous Clang in the name,
since there are no longer any non-Clang variables.
Bug: 68947919
Test: no change to command lines used to build mmma bionic
Change-Id: I081d5e1d423a66d55faece6cb1d265cd5e504d5f
Host bionic will use multiple files for CrtBegin, make CrtBegin and
CrtEnd Paths instead of OptionalPath.
Test: go test ./build/soong/cc/...
Change-Id: Ie2e954cd87808a903617696da443009f6173e312
The host bionic bootstrapping no longer needs an injected symbol.
Replace host_bionic_inject with host_bionic_verify that validates
the resulting binary, and add it as a validation dependency of the
binary.
Test: build and run host bionic binary
Change-Id: I3e303d2a164b6eef851bdc8075e6ee456c05b0a8
Improve the dependency accuracy by adding an order-only dependency
on the shared library alongside the existing depenency on the
table of contents file. The dependency won't cause the module
to be rebuilt when the the shared library changes, but will allow
anything using the ninja graph for sandboxing to know the dependency
exists.
Test: manual
Change-Id: I3accbcabee62fa0ad3eb3d1aaedc5a4bffe27308
The --use-gnu-strip option was removed from the shell script already,
and it no longer uses the $CROSS_COMPILE environment variable.
Bug: http://b/185257607
Bug: http://b/147452927
Test: treehugger
Change-Id: If9f7b75d52c1e9cd167d73eeba8f77054ffd63a7
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
rewrapper supports a comma separate list of rsp files, replace
REParams.RSPFile with REParmas.RSPFiles.
Test: remoteexec_test.go
Change-Id: I7850c071c23d368d6fad4480dd527d146c13c6d3
With clang r377782, llvm binutils implements all the necessary
funtionaility for strip.sh. We can finally get rid of all the fallback
GNU binutils usage.
Test: m
Bug: 141010852
Bug: 135627985
Change-Id: I110f6028dab7f599decf59a5cb1b927b35e11857
Remove the references to the android package in remotexec so that
the android package can reference the remoteexec package. This
will allow RuleBuilder to integrate directly with remoteexec.
Bug: 182612695
Test: m checkbuild
Change-Id: I15be5ef126d8aacbd605518638f341daf6f31bb3
When soong runs the header-abi-dumper remotely, it must provide a few
dynamic libraries as toolchain inputs, or the action will fail
unconditionally on the remote execution system.
This patch allows disabling fallback for the header abi dumper
actions, i.e., this now works with remote execution:
export RBE_ABI_DUMPER_EXEC_STRATEGY=remote
Test: ran against a remote execution server; actions no longer fail remotely
Change-Id: I0c48484c03d0923dae63004fea6632704b317e95
Revert submission 1461902-debug-compilation-dir
Reason for revert: "-Xclang" isn't being uniformly respected everywhere. For example, in ".S" compilations, when I pass `-Xclang -fdebug-compilation-dir=.", the assembler seems to be ignoring it and then inserting the `pwd` into the command, however when I pass "-fdebug-compilation-dir=.", it strips out the path to the current working directory.
This indicates that we need to update re-client's input processor so that we can pass -fdebug-compilation-dir=. without "-Xclang" and then remove `PWD` setting.
I'll followup with a patch to add -fdebug-compilation-dir.
Reverted Changes:
Ib0f271e55:Add -fdebug-compilation-dir option
Ifa0592af5:Remove env-var-allowlist
Change-Id: I12fa0bf2fb4975ac3d1bffaf1358331548abc339
The main use of this flag is to be to make both the debug info
and the build command line be independent of the compilation dir.
See: https://reviews.llvm.org/D63387 & https://blog.llvm.org/2019/11/deterministic-builds-with-clang-and-lld.html
Removed -fdebug-prefix-map and PWD variable being set in the compile
command line - these were workarounds we originally did to ensure that
the compile output is independent of the working directory and are no
longer needed now that we have -fdebug-compilation-dir.
Test: Ran "m
out/target/product/crosshatch/obj/SHARED_LIBRARIES/libexif.vendor_intermediates/same_vndk_variants.timestamp"
which fails on an RBE hardened stack without this change and passes with
this change.
Bug: b/169266636
Change-Id: Ib0f271e55f6cf0c89a30e65d3a0b25fe3f3a8f48
The cc stripping logic can be reused for Rust. Export the Stripper
structure for that purpose. Extract the strip-related flags from
builderFlags into StripFlags. Add the method flagsToStripFlags
(similarly to flagsToBuilderFlags).
Add the helper method disableStripping on libraryDecorator.
Test: m
Bug: 153430439
Change-Id: I11aef1abb8d498a4c1672500a7398279edf7f548
To support module specific lexer flags, this follows the same strategy
as the yacc flags:
- add LexProperties to the BaseCompilerProperties
- propagate those flags to the generator generation (i.e. genLex)
- add a placeholder for custom flags
- replace the placeholder with the concatenated flags
This might not support escaping very well, but I figured that this is a
very edge case. Support for escaping etc. could be added later on.
Bug: 159682555
Signed-off-by: Matthias Maennich <maennich@google.com>
Change-Id: I31a3b783bb05213fe1621191031952b41b318103
We used this when doing cache-only for header-abi-dumper to prevent
cross-branch cache hits since abi-dumper had the absolute path of the files
in its output. Since header-abi-dumper no longer outputs absolute paths,
we no longer need to prevent cross-branch cache-hits.
Bug: b/162045672
Change-Id: Iecc3ad9614cda6202cc5f8a9d538f806a79bdf3b
Add a -r argument to soong_zip that reads a list of files from a file
like the -l argument but treats it as a Ninja rsp file with escaping.
Replace the -l arguments in Soong that are using rsp files with -r.
Fixes: 162435077
Test: TestReadRespFile, TestZip
Change-Id: I4605312e99406ab1bd0c37af9c5ad212393f0403
Test: Ran with the following command and it succeeded
RBE_CLANG_TIDY="true" RBE_CLANG_TIDY_EXEC_STRATEGY="remote" RBE_CXX_EXEC_STRATEGY="remote_local_fallback" RBE_METALAVA="true" RBE_METALAVA_EXEC_STRATEGY="local" RBE_ABI_LINKER="true" RBE_ABI_LINKER_EXEC_STRATEGY="remote_local_fallback" RBE_CXX_LINKS="true" RBE_CXX_LINKS_EXEC_STRATEGY="remote" use_rbe m out/soong/.intermediates/external/android-clat/clatd/android_arm_armv7-a-neon/obj/external/android-clat/icmp.tidy
Bug: b/157147559
Change-Id: I110b6157fc090abd14ac32330fc59a3d76cdfa82
Use the -Z profile-emit flag to ensure that rust gcda files have
/proc/self/cwd/ appended to them similar to cc modules. This makes sure
our gcda output is consistent no matter what language the binary is
written in.
Bug: 156482307
Test: gcda outputs to the new path.
Change-Id: Ife4d55f885c7e33dffa66f7436bc2bf8b5916586
objects.
Necessary to make whole_static_libs work with
cc_prebuilt_library_static since it doesn't propagate the list of
object files.
Test: Build & boot
Test: m libsigchain && \
ar t out/soong/.intermediates/art/sigchainlib/libsigchain/android_arm64_armv8-a_cortex-a73_static/libsigchain.a
(Check that the list is sigchain.o followed by async_safe_log.o, both
in a normal build and in one where async_safe is a prebuilt static
lib.)
Bug: 154248570
Change-Id: Iaada8490ce713c13804b5771ad606f4a27e72a2f
This CL adds a remoteexec package that allows adding a configurable RBE
prefix to the template.
Test: built aosp crosshatch userdebug with and without RBE_CXX_LINKS.
Change-Id: Ica920c3d7f79f2996210b9cbd448126451c1707c
This commit adds a header ABI checker option to check all APIs that can
be found. Without this option, the checker by default only checks the
class/struct/enum that are directly or indirectly referred by one of the
exported symbols. With `check_all_apis: true,`, the checker will check
all class/struct/enum.
Bug: 141709599
Test: Add `header_abi_checker: { check_all_apis: true, }` to a library
and see breakage if I change some enum.
Change-Id: I61f90e07b60a6752fc6be4398420c1ad1291102f
The "ccNoDeps" rule was introduced to separate out ".s" compilations
from other ".S" and ".c*" compilations. The ".s" compilation does not
produce a dependency file and does not support running a preprocessor
through it.
However, it does have ".include" directives, that do NOT take a macro,
but still does the equivalent of a "#include". The compilation of ".s"
assembly files also do NOT produce a dependency file.
Because they don't produce a dependency file AND because RBE's input
processor does not yet support finding dependencies for these files, I'm
making this change to not prepend rewrapper for these commands.
Test: Tested by running a build with this change and ensuring the ".s"
compilations aren't being sent to rewrapper.
Change-Id: I60bb14ff92596e4992e9f675bdc199f1440d4327
Otherwise binaries under:
prebuilts/gcc/linux-x86/host/x86_64-w64-mingw32-4.8/bin/
are being invoked.
https://sourceware.org/binutils/docs-2.33.1/binutils/windres.html
states:
When windres reads an rc file, it runs it through the C preprocessor first.
Use the `--preprocessor` flag to specify the exact command to run. The
args to the preprocessor get wiped out when setting that flag, so
`-E -xc-header -DRC_INVOKED` must be re-passed, else clang will error as
it tries to invoke ld.
Once AOSP packages llvm-rc, aosp/1206346 can be rebased on top, so this
patch should be a short lived change.
Bug: 147295872
Test: <remove mingw gcc> && lunch aosp_x86-eng && \
cd external/mdnsresponder && mm
Change-Id: Ic0e6b75e7e964ee9f9757e534cddd16438604c34
Suggested-by: Pirama Arumuga Nainar <pirama@google.com>
Signed-off-by: Nick Desaulniers <ndesaulniers@google.com>
This CL adds RBE support to javac, r8, and d8 rules which is only
enabled if respective environment variables are set.
Test: an aosp_crosshatch build with and without the new variables.
Change-Id: Ic82f3627944f6a5ee7b9f3228170c2709b1bfcb8
Native compiler flags are currently applied in approximately:
global cflags
local cflags
local include dirs
global include dirs
global conlyflags
local conlyflags
global cppflags
local cppflags
This means that a flag that is enabled in the global cppflags
cannot be disabled in the local cflags, and an Android.bp author
must know to disable it in the local cppflags.
The previous CL split the global and local flags into separate
variables. Rearrange the order that the variables are applied
to be:
global cflags
global conlyflags
global cppflags
local cflags
local include dirs
local conlyflags
local cppflags
global include dirs
Bug: 143713277
Test: m native
Change-Id: I171524ab40096a636a8e549e1e4bc3347ef9f97a
Native compiler flags are currently applied in approximately:
global cflags
local cflags
local include dirs
global include dirs
global conlyflags
local conlyflags
global cppflags
local cppflags
This means that a flag that is enabled in the global cppflags
cannot be disabled in the local cflags, and an Android.bp author
must know to disable it in the local cppflags. A better order
would be:
global cflags
global conlyflags
global cppflags
local cflags
local conlyflags
local cppflags
local include dirs
global include dirs
We are mixing both the global and local cflags into a single
variable, and similar for conlyflags and cppflags, which
prevents reordering them. This CL prepares to reorder them
by splitting the global and local cflags into separate variables.
Bug: 143713277
Test: m native
Change-Id: Ic55a8c3516c331dc5f2af9d00e59ceca9d3e6c15
Previous solution by using objcopy uses a quirky behaviour of the GNU
objcopy and there is no equivalent option in llvm-objcopy.
Instead of removing symbols, extract and repack libgcc to only include
required objects.
Bug: 142585047
Test: presubmit
Change-Id: I58af74c18838f797e481da38c3265f0624fddf99
Remove the distinction between pctx.StaticRule and
pctx.AndroidStaticRule so that all of the local rules correctly
get assigned to the localPool. Also put Module and Singleton
rules into the localPool.
Test: compare out/soong/build.ninja
Change-Id: Id2bb38eff3c7209340fe55bc9006f00bd3661d81
Android builds by default put artifacts into out/ subdirectory of
the source tree, causing the extractor to record their names as
relative. The indexer considers such files as sources, which is wrong.
Fortunately, the extractor can be fed a set of filename rewriting
rules (see build/tools/vnames.json).
Also, undo previous unsuccessful attempt use to absolute path for the
output directory to distinguish between source code and artifacts.
Bug: 141385476
Test: run the build, inspect compilation units of the kzip file
Change-Id: I89ec3aed8fd14f43ea6e0b226d54f643346f6125
The P flag is supported as of llvm r354044, add back the flag to reduce
the size of intermediate archieve files. This does not affect the final
binaries.
Test: built
Bug: 71618641
Change-Id: I017780e4dcaa31c7fbe10b5e7482db1bba83e716
Usually, ".S" files are processes with the c preprocessor, and ".s"
files are not, so they don't have any dependency information, since it
is generated by the preprocessor.
But with the -xassembler-with-cpp flag, ".s" files are processed with
the preprocessor, so we should ask for dependency information from them.
Test: NINJA_ARGS="-t deps out/soong/.intermediates/external/sonivox/arm-wt-22k/libsonivox/android_arm_armv7-a-neon_core_static/obj/external/sonivox/arm-wt-22k/lib_src/ARM-E_filter_gnu.o" m
Test: treehugger
Change-Id: Iee7baeebc2b205b5a2f33e7c1705ea4a5b4fc95a
* changes:
Introduce inject_bssl_hash library property.
BoringSSL FIPS build - introduce extraLibFlags and use for STL libs.
Allow linker scripts when building objects.
Allow .o files as srcs.
Rationale: On non-bionic, stl.go currently adds system libraries to
ldFlags, this causes problems for partialLd rules. However adding the
same libraries to libFlags breaks some existing modules due to symbol
conflicts as the system libraries are linked before some module code.
Introduced a general mechanism for adding libraries to be linked
last rather than making this STL-specific.
Bug: 134581881
Bug: 137267623
Test: TH
Change-Id: I779f28c6586b3fea85cc6299b686e4fde95262d3
We've been getting these dependencies transitively through the
dependency on the object itself (which is a workaround for the lack of
dep file support in these tools). But for remote builds to work, we need
to know about these dependencies like any other object compilation.
For regular builds, this increases the size of the ninja file by a few
tens of megabytes (~1-2%).
WITH_TIDY builds were already larger (~40-50%), but are now about 90%
larger than a normal build.
Test: treehugger
Change-Id: Icdb4ca3d4d08d5706593d96d5c627149fa14fed8
This commit adds tags, such as NDK, VNDK-core, and PLATFORM, to
LSDUMP_PATHS. The script updating the reference ABI dumps uses the tags
to determine the directories where the dumps should be created.
Test: make findlsdumps
Bug: 133176785
Change-Id: I8540286238cf0ec55c65e1c4f60cb9c12e5e57a1
Use Kythe (https://kythe.io) to build cross reference for the Android
source code. ~generate the input for it during the build. This is done
on demand: if XREF_CORPUS environment variable is set, build emits a
Ninja rule to generate Kythe input for each compilation rule. It
also emits two consolidation rules (`xref_cxx` and `xref_java`),
that depend on all Kythe input generation rules for C++ and Java.
The value of the XREF_CORPUS environment variable is recorded in the
generated files and thus passed to Kythe. For the AOSP master branch it is
`android.googlesource.com/platform/superproject`, so the command to build
all input for Kythe on that branch is:
```
XREF_CORPUS=android.googlesource.com/platform/superproject m xref_cxx xref_java
```
Each Kythe input generation rule generates a single file with .kzip
extension. Individual .kzip files have a lot of common information, so
there will be a post-build consolidation step run to combine them.
The consolidated .kzip file is then passed to Kythe backend.
The tools to generate .kzip files are provided by Kythe (it calls them
'extractors'). We are going to build them in toolbuilding branches
(clang-tools and build-tools) and check them in as binaries into master
and other PDK branches:
For C++, `prebuilts/clang-tools/linux-x86/bin/cxx_extractor`
for Java, `prebuilts/build-tools/common/framework/javac_extractor.jar`
Bug: 121267023
Test: 1) When XREF_CORPUS is set, build generates Ninja rules to create
.kzip files; 2) When XREF_CORPUS is set, building
`xref_cxx`/`xref_java` creates .kzip files; 3) Unless XREF_CORPUS is
set, build generates the same Ninja rules as before
Change-Id: If957b35d7abc82dbfbb3665980e7c34afe7c789e
Clang will soon start using relocations that are currently unsupported by
the version of ld.bfd that Android uses when HWASAN is enabled. This will
cause partialLd to fail since the clang driver defaults to ld.bfd.
Switch to using LLD, since it will support those relocations.
Change-Id: I0964f53dc63773f3b52bb377f863df9c39961a07
This reverts commit 6d8c0a50bc.
The switchover to lld appears to be making adb crash on exit for unclear
reasons. Revert it as a quick fix pending further investigation.
Bug: http://b/134613180
Test: adb.exe under windbg
Change-Id: Ibdf5981c3c6828b3684974b7c30a7c96449c64c5
Failures in strip.sh are still occurring with darwinStripPool set to
10, try 5.
Bug: 132822437
Test: none
Change-Id: I2df247c89b6d8f6e741d4e17057b06ff9ec72b40
New strip option named keep_symbols_and_debug_frame, that will keep the
symbols and the .debug_frame. This is meant for use by libc.so only on
arm32. Other libraries might want to use it to keep better unwinding
information on device.
Bug: 132992102
Test: Built libc.so with this option and verified that it contains
Test: the .debug_frame section.
Change-Id: I823a28199dec8316e8b26fe31ff9f17e6b11d406
strip.sh can use a file descriptor per .o file when run on .a files,
which can hit the system file descriptor limit on darwin. This
causes failures when manay variants of libgcc_stripped are built
simultaneously. Put all strip rules on darwin into a pool that
limits them to 10 concurrent processes, which will limit the file
descriptor usage to ~7500.
Fixes: 132822437
Test: no mention of darwinStripPool in out/soong/build.ninja on linux
Test: m libgcc_stripped on darwin
Change-Id: I3d4fbbd8d44d2e9059a79df113ab95336ec2c658
We use libgcc as fallback for symbols not present in libclang_rt
builtins, however we didn't know what exact symbols were being used,
some may not be intended to fallback.
Create libgcc_stripped, which only contains unwind symbols from libgcc.
Bug: 29275768
Test: bionic-unit-tests
Change-Id: I5b349fa6138e51663bf3b67109b880b4356da8e8
This reverts commit 61166dc047.
One difference from the earlier change is that import libraries are now
using the '.lib' extension instead of '.a' to prevent clash with
AdbWinApi.a.
Bug: http://b/110800681
The following flags that the binutils linkers support are not
available in lld for Windows:
-soname
--no-undefined
-rpath
Windows also uses "import libraries", which are stub libraries used only
for linking. The binutils linkers accepted a DLL and treated them as an
import library. But lld issues the following error:
lld-link: error: ...DLL: bad file type. Did you specify a DLL instead
of an import library?
To resolve this, pass '-out-implib=libFoo.lib' to lld when linking
libFoo.dll to get lld to generate an import library. Add libFoo.lib as
an implicit output to the 'ld' build rule.
Rewrite the shared libraries when building a library/binary to use the
import library instead of the DLL. As a side-effect, this also uses the
newly-created AdbWinApi.lib that's alongside
development/host/windows/prebuilt/usb/AdbWinApi.dll
Test: Run Windows tests (go/android-llvm-windows-testing) and check
absence of regressions. Also check that the following commands pass:
$ adb.exe devices
$ fastboot.exe devices
Change-Id: I34e07d345e0207086ac8e8ea12525d8c322b20fd
