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
We rely on a weird, GNU specific behaviour of objcopy:
$ aarch64-linux-android-objcopy -w libgcc.a stripped.a --strip-unneeded-symbol=*
$ llvm-nm --defined-only stripped.a | grep multi
0000000000000000 T __multi3
$ aarch64-linux-android-objcopy -w libgcc.a stripped.a --strip-unneeded-symbol=* --keep-symbol=FOO
$ llvm-nm --defined-only stripped.a | grep multi
llvm-objcopy has a different, more consistent behaviour that does not
mark exported symbols as unneeded when "--keep-symbol" flag exists.
However this behaviour will leave most symbols not deleted in the
libgcc_stripped library.
Revert the change while I'm implementing an alternative solution.
Test: presubmit
Bug: 142585047
Change-Id: I175e811f8a1f6afc6558267fc54e159ad5e12acf
Also fix an issue with deps file introduced in ab5e5142. Now correctly
depends on llvm-readelf insteaf of GNU readelf.
Test: presubmit
Change-Id: Id287e2844626c07917348a62820f8b5edbb22c3d
We had a typo in the original checkin. We need two hypens before
"remove-section" for llvm-strip, just like we have for strip.
Bug: 140721128
Test: Clean build of Go device
Change-Id: Ifb83dc4e6ad8f8070d9ced678f9147d6a040aee7
llvm-objcopy now implements '--regex' option, which allows it to replace
GNU objcopy for creating libgcc_stripped archieve.
Test: build
Bug: 131631155
Change-Id: Ie1733ac8e9b3765f7652a562406dcedb4735cc1a
sysprop_library now checks the API stability itself, cutting dependency
on java_sdk_library. Under the directory {module_dir}/api,
{module_name}-current.txt and {module_name}-latest.txt hold API
signatures.
When sysprop_library is built, or a user run "m {module_name}-check-api"
command, API check is performed. First, current.txt must have exactly
same signature with built sysprop_library module. Second, current.txt
must be compatible with latest.txt.
Build system emits a handy error message to generate/update those API
files, in case of missing or mismatching. Also, a script file for
freezing API files is introduced.
Bug: 131637873
Test: 1) m && boot blueline
Test: 2) m {sysprop_library} performs API check
Test: 3) manual test for check-api, freezing api
Change-Id: I9d25f5dc64299e666527ca8e23d7233966901c4e
Merged-In: I9d25f5dc64299e666527ca8e23d7233966901c4e
Merged-In: Ib7ad4f17e82c90da5ef3f80e2ab88c0b53112c60
(cherry picked from commit 093f0eb133)
This tool is used to modify apex_manifest.json when building apex
module.
Here's the usage:
usage: jsonmodify [-h] [-o output] [-v path value] [-s path value]
[-r path] [-a path [value ...]]
[input]
positional arguments:
input JSON file
optional arguments:
-h, --help show this help message and exit
-o output, --out output
write result to a file. If omitted, print to stdout
-v path value, --value path value
set value of the key specified by path. If path
doesn't exist, creates new one.
-s path value, --replace path value
replace value of the key specified by path. If path
doesn't exist, no op.
-r path, --remove path
remove the key specified by path. If path doesn't
exist, no op.
-a path [value ...], --append_list path [value ...]
append values to the list specified by path. If path
doesn't exist, creates new list for it.
Bug: 138695532
Test: m jsonmodify
Test: echo {} | jsonmodify -v name hello -a list.nested a b c
{
"name": "hello",
"list": {
"nested": [
"a",
"b",
"c"
]
}
}
Change-Id: I2cd043c614b3ad2306a0c27ccee302633c6d2525
If sort_bss_symbols_by_size is true, a shared library is built twice.
The first build generates an unsorted output file, which is used to
generate the symbol ordering file. The output of the second build is
a shared library with its bss symbols sorted by their size.
With this, the only user of symbol_ordering_file, libc, is migrated to
use the new property, so we remove symbol_ordering_file support as well.
Bug: 135754984
Test: Build and check the resulting libc.so has its bss symbols sorted.
Change-Id: I5c892b44d82eb99cbc070cfa2c680be3087f3364
Make the jar wrapper script correctly handle quoted arguments with
spaces in them. Also allow JVM arguments in the form -J-XX, not just
-JXX.
Test: m checkbuild
Change-Id: Iec5105bc390f2a12c6a4cda7f76d37585c39f520
clang-format tries to automatically detect the Cpp standard used,
particularly when determining to put spaces in template definitions:
Cpp03: std::vector<std::pair<bool, bool> >
vs
Cpp11: std::vector<std::pair<bool, bool>>
It doesn't always get this correct as seen in
https://android-review.googlesource.com/c/platform/system/core/+/973463/8
but since we know that all code within Android is C++17, we can safely
assume it should be formatted with the newer standard.
Test: clang-format doesn't break the above change.
Change-Id: I7d1d709690b7bca6da3863cc9a58e53eaec751b0
As of [1], the Google format, which we default to, uses IncludeBlocks:
Regroup, which sorts all includes with <> and without .h into the top
group reserved for libc headers, e.g.:
+#include <android-base/properties.h>
+#include <android-base/strings.h>
#include <ctype.h>
#include <stdlib.h>
#include <sys/system_properties.h>
#include <iostream>
-#include <android-base/properties.h>
-#include <android-base/strings.h>
This is accurate to the Google C++ style guide, which states that
other library headers should be included with "", however we have a
strong tendency to use <> for these headers for Android, so we must
revert this change.
Note, clang-format allows regexes for creating the blocks as well,
however, short of including all possible libc headers, I don't think
there's a way to differentiate between <sys/wait.h> and
<other_lib/header.h>.
[1]:
62e3198c4f (diff-b689ebff2e51e675b346a88289e74542)
Test: clang-format doesn't do the above.
Change-Id: Ifef605ffd6b4c8a36512377644e64ddd26b68a21
The new package manager behavior requires packages without source code
to have an application element with hasCode attribute set to false in
their manifest. With this change, Soong can now automatically insert one
for codeless apps.
Test: app_test.go, manifest_fixer_test.py
Fixes: 124375490
Change-Id: Ied89a8d07c63805ab910859a4f7c45fc1c60bb73
All users have been removed.
Test: treehugger
Change-Id: I35e55f3ce79c8e68585be120b5dec810563c7a83
Merged-In: I35e55f3ce79c8e68585be120b5dec810563c7a83
Add a tool that can check that the <uses-library> tags in an
AndroidManifest.xml file match a list provided by the build.
Bug: 132357300
Test: manifest_check_test
Change-Id: If15abf792282bef677469595e80f19923b87ab62
In preparation for making it use multiple files so that some
implementation can be shared with a new tool.
Also make manifest_fixer_test a python_test_host, and add it
to TEST_MAPPING.
Test: m checkbuild
Change-Id: Iaae177efcb978187b126d90359b82cdfd3176857
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
The jars containing the merged header classes from transitive
static dependencies were stripping all of META-INF/. Kotlin
puts metadata in META-INF that is used to map the package to
the static class that contains the package-level functions.
This also exposed that the metadata in META-INF is always
called "name.kotlin_module", so a library that contains
kotlin files with a dependency that also contains kotlin files
would shade the metadata file. Use a unique name instead.
Fixes: 131709692
Test: m checkbuild
Change-Id: I5cd276e563206e37c4c0d90fe9f346e9396f88c0
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
There are no remaining users.
Also mark BUILD_BROKEN_ANDROIDMK_EXPORTS and BUILD_BROKEN_ENG_DEBUG_TAGS
as deprecated in scripts/build_broken_logs.go
Test: treehugger
Change-Id: If7892bef1b9001f12a99565f886b395cf1985e70
Some people apparently still talk to the network during their build.
Allow this temporarily with a BUILD_BROKEN_USES_NETWORK check.
Bug: 129992021
Test: attempt to talk to the network during the build with and without
this flag.
Change-Id: Ifb967c656aa24c4599e7232d0f1b5a303b5bac52
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
The setup_go_workspace_for_soong now allows the reverse operation
of unbinding the directories of the previous operation. Also,
refactored the script to easily add more directories to bind/unbind
in the future, catch failures on running in subshell commands and
a cleaner way to find the repo top directory.
Bug: b/129407866
Test: Manually tested by running the script on binding and unbinding
the directories. Tested both darwin and linux OS.
Change-Id: I7c93230aeab819ab5747e990e95aa26077071d9e
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
Notice file for an APEX is created by merging notice files for the
modules included in it (plus the notice file for the APEX itself if
specified).
Notice files having the same content are not duplicated; it is emitted
only once.
Bug: 128701495
Test: m (apex_test is amended)
Test: m and inspect $(PRODUCT_OUT)/obj/NOTICE.txt to check there are
license entries for /system/apex/*.apex files
Change-Id: I169d91038291a6c71615de97cf5b03174afab5d4
Some non-interpretable error messages has been shown in case of missing
current api files. Instead of that messy errors, This clearly tells
users what is the problem and how to fix it.
Bug: 126259114
Test: 1) try to build a sysprop_library and a java_sdk_library without
creating txt files under api/, and see it fails.
Test: 2) Copy and paste the emitted command, and try to build again.
Change-Id: I986781cc710475647c2b996b0f1c51a6f50c0ba2
Only put uncompressed native libs in an APK if the min_sdk_version
supports it (>= 23, Marshmallow), and set
android:extractNativeLibs="false" in the AndroidManifest.xml so
that the platform won't extract them anyways.
Bug: 117618214
Test: m checkbuild
Change-Id: I760017e48bf3c6b618aabde0982df45995765d48
And add a helper script that can help parse the output from
build_test.bash
Test: check for BUILD_BROKEN_ENG_DEBUG_TAGS in soong.log
Test: go run ../build/soong/scripts/build_broken_logs.go *
Change-Id: Idd0fc8b59770dcdbe44eeba262558708a9497f96
... except for Darwin Mach-O, as it is not supported by llvm-strip.
Test: m checkbuild
Bug: 119221035
Change-Id: I021637b6dd3530bff1f563f2ec7c2168e1083b7e
Bug: http://b/122481018
The '-keep' option of llvm-strip has changed to '-keep-section'. Update
scripts/strip.sh accordingly.
Test: Tested on various targets in AOSP and internal branch. Googlers
see http://go/clang-r349610-testing
Change-Id: Ibfc97ecdcdc305c97a4169cead1e1e3cddda9c4c
Draft APIs are available to the platform and to CTS to allow
developers to iterate on an API, but hidden from the NDK artifacts to
avoid releasing the API until it is ready.
Test: Mark binder_ndk headers and library as drafts, make checkbuild,
build-ndk-prebuilts.sh, verify missing from NDK artifact.
Bug: http://b/120091134
Change-Id: I8685e92bdaaea581e17fe98e7a2bfb9388f9f132
The last change should have changed this from taking a boolean to a
simple switch.
Test: build a package with LOCAL_PREFER_INTEGRITY=true
Bug: 112037137
Change-Id: I5ce3b1455404b5c8e781496d26c0431db52a170f
Yi Kong