Previously the recovery binary was configured to be installed to the
system.img and then got copied to recovery.img in the recovery.img's
build rule.
With this change, a module, such as the recovery binary, can configure
itself to be installed directly to the recovery.img, just like how other
modules get installed to system.img.
Bug: 19667686
Change-Id: I46b0b4a95cf078a68999db9c0f6635d6a3f5cd86
The extra system space is needed for mips64r6/mips32r6 quick
mode images and the extra userdata is needed to run ART tests.
Change-Id: I96dc1553d950dcf046b57feff10a369e9155bd4d
Previously when a file in LOCAL_SRC_FILES starts with "../", the object
file may escape out of the module's intermediate directory, because we
insert the source file's path (but not with LOCAL_PATH) to the object
file's path. Even worse when two object files escape to the same destination
and cause conflict.
This change fixes the issue by removing the "../" inside the object
files' paths. To do that, we have to set up the compilation rules for
those files one by one, instead of using the one-for-all static
pattern rules.
Bug: 19641115
Change-Id: I19f3c48ece3244fa14acb2caa609deea710840d3
By default we build both 64-bit and 32-bit odex files for a Java library.
With this change:
- For system server jars (PRODUCT_SYSTEM_SERVER_JARS), we build only
64-bit odex;
- A library can opt to build only 64-bit odex with
"LOCAL_MULTILIB := first".
Bug: 19650934
Change-Id: Ic0b7fd381396ed276e6129f883881c5c41c6e154
On non-arm architectures, there is no TARGET_CPU_VARIANT set. For x86,
art assumes that the CPU variant is actually the TARGET_ARCH_VARIANT.
Therefore, if no TARGET_CPU_VARIANT is set, use the TARGET_ARCH_VARIANT.
If TARGET_ARCH_VARIANT is not set, then use default as the value.
Change-Id: I17fc9e3ac7412462103b8f0b287fce106298b741
Expectation is that classes in pre-compiled apps should be structurally
sound and not cause a hard failure.
Bug: 19606409
Change-Id: Idc9da4d4c6bd259555671c657d3414a97940717f
We should put the NDK library path before
$(TARGET_OUT_INTERMEDIATE_LIBRARIES), so that we link against the NDK
libc/libm, instead of the platform libc/libm in
$(TARGET_OUT_INTERMEDIATE_LIBRARIES), which may still being written out,
because we don't have dependency on them.
Bug: 19613709
Change-Id: I26a8b272e38b7436bca3324246b21cd71349662b
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
Ying Wang