The Mac linker doesn't understand --start-group, which is needed for
properly linking libgcc/libc statically. It isn't needed for dynamic
executables though, so use that driver behavior where appropriate.
Bug: 19567451
Change-Id: Ifeb03bea55bc87561c64ddafdb99f664fef0f00e
BUILD_HOST_static=1 forces all host binaries to be statically linked.
Since -nodefaultlibs was passed (to disable libstdc++), libgcc wasn't
being passed. This change emulates the driver's behavior.
Also fix default STL selection for BUILD_HOST_static.
Bug: 19567451
Change-Id: I617aab782d40ac76ca5a7d9dddf0f202a4e3a69b
http://llvm.org/bugs/show_bug.cgi?id=15086,
llvm tail call optimization is wrong for x86.
For Android/x86 to use SSE* instructions safely, stack should be 16-byte
aligned before JNI function call, which isn't true for all x86 device,
so -mstackrealign should be the default.
BUG: 19234330
Change-Id: I4c6676366788772dbe64fd7f0dd33b3ed5c9b80e
Previously if user has a directory with name dummy in the root of the
source tree, "zip -qd package.apk dummy" fails with:
"zip error: Nothing to do!".
This change mitigates the error.
Change-Id: I642e3bf0378e5b9911a068ecb72f795b3e92f1fe
- Add a new flag to zipalign (-p) that page aligns shared
libraries (zip entries ending with ".so") in the archive.
- Add a new build variable LOCAL_PAGE_ALIGN_SHARED_LIBRARIES
to turn on this behaviour in zipalign.
- Add a new LOCAL_JNI_SHARED_LIBRARIES_ZIP_OPTIONS to control
zip behaviour.
Bug: 8076853
Bug: 19330157
Co-Authored-By: Simon Baldwin <simonb@google.com>
Co-Authored-By: Dimitry Ivanov <dimitry@google.com>
Change-Id: I1aa2c039bb2a590ae72f256acc9ba5401c2c59b1
These CFLAGS and CPPFLAGS will always be added last, and are
controlled by the build system. This way we can add warnings that
users are not allowed to disable.
Change-Id: Id71f69249078f62ca2687ecbf764aff0fd3a1c1b
Implicit function declarations are much more dangerous on LP64 because
sizeof(int) != sizeof(void*), so any function that returns a pointer will
lose its top bits, leading to relatively hard to debug crashes.
Change-Id: Ia05beffb949ca747833c2e12c40daf896f7a60a5
Implicit function declarations are much more dangerous on LP64 because
sizeof(int) != sizeof(void*), so any function that returns a pointer will
lose its top bits, leading to relatively hard to debug crashes.
Change-Id: Ib2570326dde1b1cde09c63c9b4c973d741528ee2
We don't support platform builds under CYGWIN and can safely
assume that all windows host builds are for MinGW32.
HAVE_WINSOCK2_H is currently unused. The only references to it
are in code that define it themselves or in application code that
by definition isn't compiled for windows.
Change-Id: I236d6e070df93f713cb716bb8a461c9c1d32a180
With the long PLATFORM_VERSION, if you have also a long device/product
name and/or a long username, ro.build.fingerprint can easily exceed the
system property's length limit of 92 bytes.
This mitigates the risk a little bit.
Bug: 15593959
Change-Id: I9e316025227b30aa6887c9ce0065364fdbb9fc8f
We don't use EXT2 for bootimage anymore.
This will allow getting rid of external/genext2fs
Bug: 19282953
Change-Id: Ie56528c7877036d2501a09888a4420637f9cd647
When incremental builds only affect the system partition,
the build.prop file for other partitions might go out of
date.
Add a dependency on system/build.prop s.t. when
it's rebuilt,
the fingerprints are updated on the other
partition's build.prop's.
Bug:19058575
Change-Id: Iea7b761e0800c402eb4ce1154e8f6af5a17f7b06
The transitive symbol resolving causes build breakage when a binary
linked against NDK has indirect dependency on platform libraries, or
vice versa.
Change-Id: Idbfe76079c1db14eb1156d0f396e13d26f0ff582
Normally the build function initialize-package-file will delete all
class files and all directory entries from JAR files, but sometimes
external projects (eg. ICU4J) depend on having directory entries in
their JAR files.
This change adds the flag LOCAL_DONT_DELETE_JAR_DIRS (analogous to the
flag LOCAL_DONT_DELETE_JAR_META_INF) which when set will skip deletion
of directory entries in initialize-package-file.
Change-Id: I4464b947b7528fca23925affa95e4071915f04d4
Note that this doesn't play nicely with acov out of the box. Clang
apparently generates .gcno files that aren't compatible with gcov-4.8.
This can be solved by installing gcc-4.6 and invoking lcov with
`--gcov-tool /usr/bin/gcov-4.6`.
http://stackoverflow.com/questions/17758126/clang-code-coverage-invalid-output
Change-Id: I79547e1c579fa79db47ff07d5e90c42cedbd5cbb
Don't remember why I didn't enable this for the host when I made the
first pass, but it works just fine.
Change-Id: I0892c0bc353bf8b60b432ba9f69f97281177d41d
Clang driver needs -static flag, not -Bstatic,
to produce statically linked modules.
However, -static is not added if LDFLAGS contains -shared,
which is used in bionic/linker to create a shared object with static libraries.
BUG: 18008984
Change-Id: I75c3e24973ee77170285ec4c8e7aacc345722685
So the build system regenerates import_includes when you modify
Android.mk to add a new dependency library.
Change-Id: Ic92b097b659bb68a9065e1d66da59e0dc7e2836a
Before this change, Java resources are added as a separate step
(add-java-resources-to-package) after dex is run, so jarjar isn't run on
the resource files.
With this change, we add Java resources immediately after we call javac,
so jarjar is run on the resource files (the module's own resource, as
well as resources carried by static Java libraries).
When we generate the final apk/jar, we use the jarjar'ed jar as the
inital pacakge file, with class files and empty folders removed.
(cherry-picked from commit 140274707e)
Bug: 18837479
Change-Id: I15ecf282bfb65fd53dd03fbd03dd4c71927c186a
Mohamad Ayyash