Host gtests that don't use pthreads themselves need to be linked
against libpthread because gtest itself uses pthreads to ensure
thread-safety. (It can't know in advance whether the code under
test is going to spin up threads, so it has to be pessimistic.)
Change-Id: I14fb1098c163d7fe6ae9a9a0314acead4ec9383d
arm and arm64 have distinct headers, but mips and mip64, and x86 and x86_64
use headers that work for both widths. So where arm/arm64 need to handle the
second architecture case specially, all we need to do for the others is
hard-code the name. (x86_64.mk already hard-codes x86; we need to change
x86.mk for the case where we're building the 32-bit binaries for a mixed
system. mips64.mk doesn't exist yet, but when it does, it'll hard-code
just plain "mips" too.)
Change-Id: Ia6b9f77b4eb2c78729b454045875c409e0ea8197
To ease the transition between toolchains, allow a target to specify
a list of cflags that the toolchain does not support. These will be
filtered out of the cflags provided by the module.
Add TARGET_GLOBAL_UNSUPPORTED_CFLAGS := -fstack-protector for the
aarch64 toolchain, it does not yet suport -fstack-protector.
Change-Id: I168d0c6f131326fad305ec86fad46e6a3e03295a
This should never have been on the default include path.
The NDK statically links its own libthread_db, so I'm removing
bionic's unused copy from devices.
Bug: 11882807
Change-Id: I49a67fe0902cc4bc178360f6c993959774d74e3a
* commit '1acb1b64b9be43503fa2a9949858b8bb36417e32':
add new gen/ directory for generated sources
warn on LOCAL_MODULE_PATH in multiarch shared libraries
Support LOCAL_MODULE_RELATIVE_PATH
Allow modules to generate source into $OUT/gen, which will then
be copied into $OUT/obj and $OUT/obj_$(TARGET_2ND_ARCH) as
necessary. This allows a single build rule invocation that includes
generated source to build for the first and second architectures.
Modules will need to change calls to local-intermediates-dir into
local-generated-sources-dir.
Change-Id: I62504bad9454b3d9fde7b84ab9f0a487a2ecf0bf
Using LOCAL_MODULE_PATH in a shared library module while building a
multiarch target will cause build rules for both architectures to install
into the same path. Warn and suggest LOCAL_MODULE_RELATIVE_PATH.
Change-Id: I16208ccada6d43a26a342af35096f49d8df26e81
Most users of LOCAL_MODULE_PATH are setting a subdirectory of the
normal install path, for example to install HALs into system/lib/hw.
This is problematic for multiarch builds, where the install location
depends on the arch. Allow modules to specify LOCAL_MODULE_RELATIVE_PATH.
HALs will generally use:
LOCAL_MODULE_RELATIVE_PATH := hw
Change-Id: I4e4ceec61d026bbe74ba604554c06104bde42e5e
We have no rigorous way to detect whether the code referenced for this
mechanism changes in between builds. So declare this as a phony target
so that it is rebuilt every time.
Change-Id: Ie1670d96895719d406e7b0defd53ba3cd224dd83
Signed-off-by: Andrew Boie <andrew.p.boie@intel.com>
Sometimes it is useful to be able to tell ota_from_target_files
to not sign the output zip file. For instance, the private
release key may not be available when ota_from_target_files
is executed; similarly the release tools may not be available
or executable where the private key is stored.
This change adds an option, '--no_signing', to simply output the
unsigned OTA zip file, instead of spuriously signing it with the
test key even though the zip file would need to be re-signed later
with a different key.
Change-Id: I1f3c4dc8ffa35ce85478f848b147aff3d40fe283
* commit '7c7f28e71044ca04d368dec626faba37fe74818d':
Add 2nd arch directories for apps
Set up rules to build prebuilts for TARGET_2ND_ARCH
Set up rules to build packages for TARGET_2ND_ARCH
Ying Wang