OSDN Git Service

android-x86/build.git
9 years agoMerge "Explain __USE_MINGW_ANSI_STDIO."
Elliott Hughes [Fri, 15 May 2015 20:52:55 +0000 (20:52 +0000)]
Merge "Explain __USE_MINGW_ANSI_STDIO."

9 years agoMerge "Update device to use set_prop() macro"
Nick Kralevich [Fri, 15 May 2015 19:26:56 +0000 (19:26 +0000)]
Merge "Update device to use set_prop() macro"

9 years agoMerge "Label /dev/goldfish_pipe as qemu_device"
Nick Kralevich [Fri, 15 May 2015 13:00:25 +0000 (13:00 +0000)]
Merge "Label /dev/goldfish_pipe as qemu_device"

9 years agoMerge "Add tzdatacheck to image"
Neil Fuller [Fri, 15 May 2015 09:09:44 +0000 (09:09 +0000)]
Merge "Add tzdatacheck to image"

9 years agoLabel /dev/goldfish_pipe as qemu_device
Yu Ning [Thu, 14 May 2015 16:00:35 +0000 (00:00 +0800)]
Label /dev/goldfish_pipe as qemu_device

In goldfish kernel 3.10, qemu_pipe has been renamed to goldfish_pipe.
However, in the emulator's SELinux policy, there is no specific security
context assigned to /dev/goldfish_pipe, and the one inherited from /dev
(u:object_r:device:s0) prevents various processes (qemud, qemu-props,
etc.) from reading and writing goldfish_pipe. Consequently, the classic
x86_64 emulator will not boot if GPU emulation is enabled ("-gpu host"),
and does not render the UI correctly if launched with "-gpu off".

Fix this issue by assigning /dev/goldfish_pipe the same security context
as /dev/qemu_pipe.

This CL also benefits the new ("ranchu") emulator, where all supported
ABIs (arm64, mips64, x86 and x86_64) use 3.10-based kernels. Without
this fix, the new emulator boots and works, but there are avc denials
related to goldfish_pipe.

Last but not least, it is now possible to boot the classic x86 emulator
with a 3.10-based kernel instead of the current 3.4-based one, without
disabling SELinux.

Change-Id: Iad979c0ee9d0a410be12b83ac1bef9476b50a6dc
Signed-off-by: Yu Ning <yu.ning@intel.com>
9 years agoExplain __USE_MINGW_ANSI_STDIO.
Elliott Hughes [Fri, 15 May 2015 03:55:49 +0000 (20:55 -0700)]
Explain __USE_MINGW_ANSI_STDIO.

Change-Id: I6900a7cddfb5aa4365c3e0f6f6f0fa9203c0fe43

9 years agoMerge "[MIPS] Temporarily disable Mips shared-textrel errors"
Ying Wang [Thu, 14 May 2015 16:11:58 +0000 (16:11 +0000)]
Merge "[MIPS] Temporarily disable Mips shared-textrel errors"

9 years agoMerge "Remove unused #include <selinux/android.h>"
Jeff Vander Stoep [Thu, 14 May 2015 15:27:50 +0000 (15:27 +0000)]
Merge "Remove unused #include <selinux/android.h>"

9 years agoAdd tzdatacheck to image
Neil Fuller [Wed, 13 May 2015 16:47:29 +0000 (17:47 +0100)]
Add tzdatacheck to image

Bug: 21110439
Bug: 19941636
(cherry picked from commit 7d18a684177064a74dafbba8cc2f270f911936bf)

Change-Id: I29ab303d2695ba1cadf70c2d2f4440d34a8cede9

9 years agoRemove unused #include <selinux/android.h>
Jeff Vander Stoep [Wed, 13 May 2015 21:24:04 +0000 (14:24 -0700)]
Remove unused #include <selinux/android.h>

Change-Id: Iebc59784c7318edb1c1f686e040ed581259bdd5a

9 years agoMerge "Revert "Revert "Turn on _FORTIFY_SOURCE=2 for the host."""
Elliott Hughes [Wed, 13 May 2015 21:12:21 +0000 (21:12 +0000)]
Merge "Revert "Revert "Turn on _FORTIFY_SOURCE=2 for the host."""

9 years agoMerge "Add libbenchmark's new dependency on libbase."
Elliott Hughes [Wed, 13 May 2015 20:53:32 +0000 (20:53 +0000)]
Merge "Add libbenchmark's new dependency on libbase."

9 years agoAdd libbenchmark's new dependency on libbase.
Elliott Hughes [Wed, 13 May 2015 20:47:30 +0000 (13:47 -0700)]
Add libbenchmark's new dependency on libbase.

Change-Id: I6f9700b3e51a3b681d3d91b88103eed31764f61e

9 years agoRevert "Revert "Turn on _FORTIFY_SOURCE=2 for the host.""
Elliott Hughes [Wed, 13 May 2015 20:15:19 +0000 (20:15 +0000)]
Revert "Revert "Turn on _FORTIFY_SOURCE=2 for the host.""

This reverts commit e27b214fc248d8927c42c5529d5c9fe75e2591aa.

Change-Id: I3d739c0594dbe4ecd62c5b07064ada7de6cdaadb

9 years agoMerge "Revert "Turn on _FORTIFY_SOURCE=2 for the host.""
Elliott Hughes [Wed, 13 May 2015 19:33:17 +0000 (19:33 +0000)]
Merge "Revert "Turn on _FORTIFY_SOURCE=2 for the host.""

9 years agoRevert "Turn on _FORTIFY_SOURCE=2 for the host."
Elliott Hughes [Wed, 13 May 2015 19:32:59 +0000 (19:32 +0000)]
Revert "Turn on _FORTIFY_SOURCE=2 for the host."

This reverts commit 224806d00ea1fbafde1adfd6fdf26693170f2648.

Change-Id: I654157ec4be7ff95356ab4a25b27c44762793e74

9 years agoMerge "Turn on _FORTIFY_SOURCE=2 for the host."
Elliott Hughes [Wed, 13 May 2015 19:10:37 +0000 (19:10 +0000)]
Merge "Turn on _FORTIFY_SOURCE=2 for the host."

9 years agoMerge "Add support for clobbered blocks"
Tao Bao [Wed, 13 May 2015 19:03:07 +0000 (19:03 +0000)]
Merge "Add support for clobbered blocks"

9 years agoMerge "Add bootsigner path as an option"
Tao Bao [Wed, 13 May 2015 18:42:16 +0000 (18:42 +0000)]
Merge "Add bootsigner path as an option"

9 years agoMerge "Remove libc++ from prepackaged shared libs in cts."
dcashman [Wed, 13 May 2015 18:19:42 +0000 (18:19 +0000)]
Merge "Remove libc++ from prepackaged shared libs in cts."

9 years agoAdd support for clobbered blocks
Tao Bao [Tue, 12 May 2015 18:42:31 +0000 (11:42 -0700)]
Add support for clobbered blocks

In ext4 filesystems, some blocks might be changed even being mounted
R/O, such as the superblock (block 0). We need to exclude such blocks
from integrity verification. Plus such blocks should always be
written to the target by copying instead of patching.

Bug: http://b/20939131
Change-Id: I991169ec307dfb231b2fe8908a0668595ecb2060

9 years ago[MIPS] Temporarily disable Mips shared-textrel errors
Duane Sand [Wed, 29 Apr 2015 23:20:51 +0000 (16:20 -0700)]
[MIPS] Temporarily disable Mips shared-textrel errors

Clang++ for Mips and Mips64 generates read-only exception unwind
tables that trigger DT_TEXTREL warnings at link time.  Until Clang
is fixed, ignore those performance warnings instead of failing the build.

With this patch, Mips clang++ can be (optionally) used when building
libdeqp.

NDK's Mips llvm has been using an alternate temporary fix, marking the
.gcc_exception_table section as read-write for Mips only:
   https://android-review.googlesource.com/#/c/119660/

A permanent fix using a read-only exception table is pending upstream:
   http://reviews.llvm.org/D9669

Change-Id: Ie0cd7da398acbe45dbe39adc251e7fd5b5ca1445

9 years agoAdd bootsigner path as an option
Baligh Uddin [Thu, 20 Nov 2014 17:52:05 +0000 (09:52 -0800)]
Add bootsigner path as an option

Change-Id: I3d804b7937aafea8274b06edab097a6dca8f04cd
(cherry picked from commit e204868f1f2bf2c6a85cf09edb52c06d593584c0)

9 years agoRemove libc++ from prepackaged shared libs in cts.
dcashman [Tue, 12 May 2015 18:31:09 +0000 (11:31 -0700)]
Remove libc++ from prepackaged shared libs in cts.

Commit 28acbeab18f6083299c07f9ebe769d22e49f8107 removed the dependency of
sepolicy-analyze on libc++, eliminating the only consumer of the library for the
cts host-side tests.  Remove the library since it is no longer needed but leave
the ability to add other shared libs in the future.

Bug: 19566396
Change-Id: I4fbfa44ce9f099ad058fd0630ac48749e389e3e1

9 years agoTurn on _FORTIFY_SOURCE=2 for the host.
Elliott Hughes [Wed, 13 May 2015 04:41:52 +0000 (21:41 -0700)]
Turn on _FORTIFY_SOURCE=2 for the host.

Bug: http://b/20558757
Change-Id: Ic0c173d25dc6b2fa9c3122a523af5f37bc1b8c55

9 years agoUpdate device to use set_prop() macro
William Roberts [Mon, 11 May 2015 17:23:59 +0000 (10:23 -0700)]
Update device to use set_prop() macro

Change-Id: I630ba0178439c935d08062892990d43a3cc1239e
Signed-off-by: William Roberts <william.c.roberts@linux.intel.com>
9 years agoMerge "Revert "Revert "Enable linker -fix-cortex-a53-843419"""
Andrew Hsieh [Wed, 13 May 2015 01:08:49 +0000 (01:08 +0000)]
Merge "Revert "Revert "Enable linker -fix-cortex-a53-843419"""

9 years agoMerge "Fix using libc++_static in dynamic binaries."
Dan Albert [Tue, 12 May 2015 22:00:32 +0000 (22:00 +0000)]
Merge "Fix using libc++_static in dynamic binaries."

9 years agoFix using libc++_static in dynamic binaries.
Dan Albert [Tue, 12 May 2015 18:00:31 +0000 (11:00 -0700)]
Fix using libc++_static in dynamic binaries.

Change-Id: I89bacfbf270d2d1bc1b9b2fa6951b355a67b808c

9 years agoRevert "Revert "Enable linker -fix-cortex-a53-843419""
Andrew Hsieh [Mon, 11 May 2015 06:03:07 +0000 (06:03 +0000)]
Revert "Revert "Enable linker -fix-cortex-a53-843419""

This reverts commit 32e1689684682aadf1a5d5af523ccafae5907572.

Now that prebuilts/gcc/darwin-x86/aarch64/aarch64-linux-android-4.9 is also updated

Change-Id: Id95c1c4cc651c434461655e62b8f23afd56b53e4

9 years agoMerge "Wrap zipfile.write(), writestr() and close()"
Tao Bao [Fri, 8 May 2015 22:52:03 +0000 (22:52 +0000)]
Merge "Wrap zipfile.write(), writestr() and close()"

9 years agoWrap zipfile.write(), writestr() and close()
Tao Bao [Wed, 1 Apr 2015 18:21:55 +0000 (11:21 -0700)]
Wrap zipfile.write(), writestr() and close()

In order to work around the zip 2GiB limit, we need to wrap the related
functions in zipfile. Calls to those functions should always be replaced
with calls to the wrappers instead.

Bug: 18015246
Change-Id: Ice494371ca6654e88ded2ae0eb680f51082effcb

9 years agoMerge "Add task to print transitive dependencies and their license files."
Ying Wang [Thu, 7 May 2015 22:43:56 +0000 (22:43 +0000)]
Merge "Add task to print transitive dependencies and their license files."

9 years agoAdd task to print transitive dependencies and their license files.
Ying Wang [Thu, 7 May 2015 19:08:53 +0000 (12:08 -0700)]
Add task to print transitive dependencies and their license files.

Print modules and their transitive dependencies with license files.
To invoke, run
"make deps-license PROJ_PATH=<proj-path-patterns> DEP_PATH=<dep-path-patterns>".
PROJ_PATH restricts the paths of the source modules;
DEP_PATH restricts the paths of the dependency modules.
Both can be makefile patterns supported by makefile function $(filter).
Example:
  $ make deps-license packages/app/% external/%
  prints all modules in packages/app/ with their dpendencies in external/.
The printout lines look like "<module_name> :: <module_paths> :: <license_files>".

Bug: 20823995
Change-Id: I06b66e85ff56c8628bffa3d948085ed45870100f

9 years agoMerge "Do not pack relocations for executables"
Dmitriy Ivanov [Thu, 7 May 2015 21:09:48 +0000 (21:09 +0000)]
Merge "Do not pack relocations for executables"

9 years agoDo not pack relocations for executables
Dmitriy Ivanov [Wed, 29 Apr 2015 19:13:37 +0000 (12:13 -0700)]
Do not pack relocations for executables

Bug: http://b/20665974
Change-Id: Ibc13b5d6bd05dfbc7ff8475068fe7363f58e7e67

9 years agoMerge "Build: Update Mips64 generic build for ART"
Andreas Gampe [Thu, 7 May 2015 16:02:16 +0000 (16:02 +0000)]
Merge "Build: Update Mips64 generic build for ART"

9 years agoBuild: Update Mips64 generic build for ART
Andreas Gampe [Thu, 7 May 2015 03:56:33 +0000 (20:56 -0700)]
Build: Update Mips64 generic build for ART

For ART testing, we need:

1) A larger userdata partition. A lot of files end up there as it
is multi-arch.

2) Don't strip prebuilts. Technically we only care about core-libart,
but this is the best high-level change that doesn't impact other
files.

Change-Id: Ic36bfcf80ba50a602752ca0a3031dda89a0f3051

9 years agoMerge "Remove stlport from global package list."
Dan Albert [Wed, 6 May 2015 21:15:57 +0000 (21:15 +0000)]
Merge "Remove stlport from global package list."

9 years agoRemove stlport from global package list.
Dan Albert [Wed, 6 May 2015 16:49:10 +0000 (09:49 -0700)]
Remove stlport from global package list.

This is now whitelisted per device.

Bug: http://b/15193147
Change-Id: I7fcd0891242fb552a17753b151cf431a398bbc44

9 years agoMerge "Disallow using stlport."
Dan Albert [Wed, 6 May 2015 19:53:39 +0000 (19:53 +0000)]
Merge "Disallow using stlport."

9 years agoMerge "Cleanup libc++ configuration."
Dan Albert [Wed, 6 May 2015 19:51:05 +0000 (19:51 +0000)]
Merge "Cleanup libc++ configuration."

9 years agoDisallow using stlport.
Dan Albert [Wed, 6 May 2015 17:08:32 +0000 (10:08 -0700)]
Disallow using stlport.

The only remaining users of stlport are vendor blobs. Prevent any new
users from using it.

Bug: http://b/15193147
Change-Id: I577a16c8c52e2c7d939c3b5026e18ad90e4b9f26

9 years agoCleanup libc++ configuration.
Dan Albert [Wed, 6 May 2015 17:08:11 +0000 (10:08 -0700)]
Cleanup libc++ configuration.

Change-Id: I2162f2a19c7f13cfbd4a2530a783a4e671d7cdb1

9 years agoMerge "Make Windows a non-multilib target."
Dan Albert [Tue, 5 May 2015 23:38:59 +0000 (23:38 +0000)]
Merge "Make Windows a non-multilib target."

9 years agoMake Windows a non-multilib target.
Dan Albert [Tue, 5 May 2015 01:17:52 +0000 (18:17 -0700)]
Make Windows a non-multilib target.

We don't have a toolchain for 64-bit windows.

This allows running `USE_MINGW=1 mm` in a directory that has a host
module with LOCAL_MULTILIB := both.

Change-Id: I31f981b38fb80b0d6582bab0a4bd580a3c654c91

9 years agoMerge "Select the correct target STL with USE_MINGW."
Dan Albert [Tue, 5 May 2015 21:40:01 +0000 (21:40 +0000)]
Merge "Select the correct target STL with USE_MINGW."

9 years agoMerge "Fix JDK check for USE_MINGW=1."
Dan Albert [Tue, 5 May 2015 18:45:54 +0000 (18:45 +0000)]
Merge "Fix JDK check for USE_MINGW=1."

9 years agoFix JDK check for USE_MINGW=1.
Dan Albert [Tue, 5 May 2015 18:41:52 +0000 (11:41 -0700)]
Fix JDK check for USE_MINGW=1.

Change-Id: I3bdd9258e8870fc00a5f1f74682a949dc487d851

9 years agoMerge "Fix mips build"
Dmitriy Ivanov [Tue, 5 May 2015 01:16:56 +0000 (01:16 +0000)]
Merge "Fix mips build"

9 years agoFix mips build
Dmitriy Ivanov [Tue, 5 May 2015 01:10:47 +0000 (18:10 -0700)]
Fix mips build

Change-Id: Ifbb2df36c173511c118f34cfa39bb01ca7f7d380

9 years agoMerge "Upgrade x86 gcc to 4.9 and disable devirtualization."
Chih-Hung Hsieh [Mon, 4 May 2015 23:36:50 +0000 (23:36 +0000)]
Merge "Upgrade x86 gcc to 4.9 and disable devirtualization."

9 years agoSelect the correct target STL with USE_MINGW.
Dan Albert [Mon, 4 May 2015 23:34:19 +0000 (16:34 -0700)]
Select the correct target STL with USE_MINGW.

The previous check tested against USE_MINGW even for target builds.
Hadn't been a problem because people don't typically set USE_MINGW
directly.

Change-Id: I90fe0ea890c44917eb29dd02d7c7f76c19e7fbd6

9 years agoMerge "Fix prebuilts for target builds with USE_MINGW=1."
Dan Albert [Mon, 4 May 2015 22:46:57 +0000 (22:46 +0000)]
Merge "Fix prebuilts for target builds with USE_MINGW=1."

9 years agoFix prebuilts for target builds with USE_MINGW=1.
Dan Albert [Mon, 4 May 2015 19:44:44 +0000 (12:44 -0700)]
Fix prebuilts for target builds with USE_MINGW=1.

USE_MINGW=1 mm didn't work in directories that contained target modules
because the build system would use the Windows locations and extensions
when trying to find the host GCC prebuilts. Windows is the target OS,
not the OS we're building from.

Change-Id: Ic994fed15388d0c7d393f71ba28fe7afdc659f5c

9 years agoUpgrade x86 gcc to 4.9 and disable devirtualization.
Chih-Hung Hsieh [Fri, 1 May 2015 21:53:59 +0000 (14:53 -0700)]
Upgrade x86 gcc to 4.9 and disable devirtualization.

Also filter out gcc-only flags -Wno-clobbered and -fno-devirtualize
when compiled with clang/llvm.

BUG: 19872411
Change-Id: I6de57583be04da607f569df65e93531787dbb789

9 years agoMerge "Pack module before making symbolic version"
Dmitriy Ivanov [Mon, 4 May 2015 20:15:46 +0000 (20:15 +0000)]
Merge "Pack module before making symbolic version"

9 years agoPack module before making symbolic version
Dmitriy Ivanov [Sat, 2 May 2015 01:12:29 +0000 (18:12 -0700)]
Pack module before making symbolic version

  We need PT_LOAD segments to match for the gdb sake.
  If we pack module after stripping symbolic version
  PT_LOAD differ from actual ones; this confuses gdb.

Bug: http://b/20687795
Change-Id: If7b1ffcda918d0cc47051a30ca1202007ed62403
(cherry picked from commit 258b29cf7652ae170a8c7907fa7361cb30542cd1)

9 years agoMerge "common.py: Add support for squashfs"
Mohamad Ayyash [Fri, 1 May 2015 22:58:08 +0000 (22:58 +0000)]
Merge "common.py: Add support for squashfs"

9 years agocommon.py: Add support for squashfs
Mohamad Ayyash [Fri, 1 May 2015 22:39:36 +0000 (15:39 -0700)]
common.py: Add support for squashfs

Change-Id: Ia40f8c6307d0213d43207aee91c4c8a5b5ad6f1e
Signed-off-by: Mohamad Ayyash <mkayyash@google.com>
9 years agoMerge "Revert "Revert "Pack relocations for mips64"""
Dmitriy Ivanov [Thu, 30 Apr 2015 06:10:19 +0000 (06:10 +0000)]
Merge "Revert "Revert "Pack relocations for mips64"""

9 years agoRevert "Revert "Pack relocations for mips64""
Dmitriy Ivanov [Wed, 29 Apr 2015 22:34:21 +0000 (15:34 -0700)]
Revert "Revert "Pack relocations for mips64""

This reverts commit 6399dcce598c2ac54bd87bc754028513405c049f.

Bug: http://b/20658994

9 years agoMerge "Revert "Revert "Pack relocations for mips"""
Dmitriy Ivanov [Wed, 29 Apr 2015 22:43:32 +0000 (22:43 +0000)]
Merge "Revert "Revert "Pack relocations for mips"""

9 years agoRevert "Revert "Pack relocations for mips""
Dmitriy Ivanov [Wed, 29 Apr 2015 21:50:05 +0000 (14:50 -0700)]
Revert "Revert "Pack relocations for mips""

This reverts commit d05fd6fc821db3e09c988370c6ac24c3354232f7.

Bug: http://b/20658994
Change-Id: Iccbd4846b310306ed11b87f4566b8aaadfa63ebb

9 years agoMerge "Revert "Pack relocations for mips""
Dimitry Ivanov [Wed, 29 Apr 2015 19:11:05 +0000 (19:11 +0000)]
Merge "Revert "Pack relocations for mips""

9 years agoMerge "Define BOARD_SEPOLICY_DIRS for 64-bit emulators."
Nick Kralevich [Wed, 29 Apr 2015 19:07:00 +0000 (19:07 +0000)]
Merge "Define BOARD_SEPOLICY_DIRS for 64-bit emulators."

9 years agoMerge "Revert "Pack relocations for mips64""
Dimitry Ivanov [Wed, 29 Apr 2015 19:03:56 +0000 (19:03 +0000)]
Merge "Revert "Pack relocations for mips64""

9 years agoRevert "Pack relocations for mips64"
Dimitry Ivanov [Wed, 29 Apr 2015 18:59:05 +0000 (18:59 +0000)]
Revert "Pack relocations for mips64"

This reverts commit 8bd30fe3cb82e50caab50e4091c5977e5e24db0e.

Bug: http://b/20658994
Change-Id: I6c31094559a12b824344802bf3993f91aee79a26

9 years agoRevert "Pack relocations for mips"
Dimitry Ivanov [Wed, 29 Apr 2015 18:57:54 +0000 (18:57 +0000)]
Revert "Pack relocations for mips"

This reverts commit 9dc1a7359f2351e18f08d69ea725f884541b2f2f.

Bug: http://b/20658994
Change-Id: Ib10c764a7d06db823ee2fccfa47240e0a639d4bc

9 years agoDefine BOARD_SEPOLICY_DIRS for 64-bit emulators.
Stephen Smalley [Wed, 29 Apr 2015 13:55:08 +0000 (09:55 -0400)]
Define BOARD_SEPOLICY_DIRS for 64-bit emulators.

Define BOARD_SEPOLICY_DIRS for the arm64, mips64, and x86_64
emulator targets.  As a first cut, simply inherit from the
existing policy directories used for generic and generic_x86.
We may need further board-specific policy added for these targets
but testing will require first enabling SELinux in the relevant
kernel configs.

Change-Id: I7b4459b32298698fc2908cbbdd0e3afadbe5ac24
Signed-off-by: Stephen Smalley <sds@tycho.nsa.gov>
9 years agoMerge "Add float support to binary event log."
Jeff Brown [Wed, 29 Apr 2015 01:42:45 +0000 (01:42 +0000)]
Merge "Add float support to binary event log."

9 years agoAdd float support to binary event log.
Jeff Brown [Tue, 28 Apr 2015 19:46:22 +0000 (12:46 -0700)]
Add float support to binary event log.

Bug: 20664753
Change-Id: Iab232b1f172094919839d57f15644dfb2eaa1425

9 years agoMerge "Don't pack ASAN executables."
Evgeniy Stepanov [Tue, 28 Apr 2015 22:06:29 +0000 (22:06 +0000)]
Merge "Don't pack ASAN executables."

9 years agoDon't pack ASAN executables.
Dan Albert [Tue, 28 Apr 2015 21:59:11 +0000 (14:59 -0700)]
Don't pack ASAN executables.

The relocation packer is causing the kernel to load the executable
overlapping ASAN's shadow space.

Bug: http://b/20665974
Change-Id: Ifc5914f4fbed5f4f00ed1c795d01cf2fcb849cfe

9 years agoMerge "--no-as-needed needs -Wl."
Dan Albert [Tue, 28 Apr 2015 21:56:44 +0000 (21:56 +0000)]
Merge "--no-as-needed needs -Wl."

9 years ago--no-as-needed needs -Wl.
Dan Albert [Tue, 28 Apr 2015 21:55:50 +0000 (14:55 -0700)]
--no-as-needed needs -Wl.

Not sure why my checkbuild passed.

Change-Id: Iead84121daaaa32c5dd1f0712e9b7caaffd58352

9 years agoMerge "Make asan more closely match clang behavior."
Dan Albert [Tue, 28 Apr 2015 21:18:57 +0000 (21:18 +0000)]
Merge "Make asan more closely match clang behavior."

9 years agoMake asan more closely match clang behavior.
Dan Albert [Tue, 28 Apr 2015 20:23:36 +0000 (13:23 -0700)]
Make asan more closely match clang behavior.

Always link libm with asan. Hasn't been a problem before because ASAN
was only ever used when libc++ was as well, which already links libm.

Pass --no-as-needed for host modules.

These aren't needed for the target builds because the target uses the
shared RTL.

Change-Id: I5d6a3c2dd557b5231be2b7711df6438693753a88

9 years agoMerge "Fix bad variable names."
Dan Albert [Tue, 28 Apr 2015 18:32:47 +0000 (18:32 +0000)]
Merge "Fix bad variable names."

9 years agoFix bad variable names.
Dan Albert [Tue, 28 Apr 2015 18:26:45 +0000 (11:26 -0700)]
Fix bad variable names.

My working tree was still dirty when I verified the original change.

Change-Id: I3e6fa6908b809c016231b398c5d30d8cce3b87ab

9 years agoMerge "Fix AddressSanitizer link order and multilib setup."
Evgenii Stepanov [Mon, 27 Apr 2015 22:46:40 +0000 (22:46 +0000)]
Merge "Fix AddressSanitizer link order and multilib setup."

9 years agoFix AddressSanitizer link order and multilib setup.
Evgenii Stepanov [Fri, 24 Apr 2015 23:34:47 +0000 (16:34 -0700)]
Fix AddressSanitizer link order and multilib setup.

ASan runtime library (when using dynamic linking) must be the first
dependency of the main executable to achieve correct symbol
interposition. This matches how the clang driver works.

In multilib setup, ASan-RT name depends on the target arch:
  /system/lib/libclang_rt.asan-arm-android.so
  /system/lib64/libclang_rt.asan-arm64-android.so

We also set RPATH to /system/lib/asan or /system/lib64/asan
to have a place for ASan-only versions of system libraries.

Change-Id: I5c0cdb89e5e08a1950eb276e406da9f31a6e52dd

9 years agoMerge "Add LOCAL_SANITIZE_RECOVER."
Dan Albert [Mon, 27 Apr 2015 19:53:01 +0000 (19:53 +0000)]
Merge "Add LOCAL_SANITIZE_RECOVER."

9 years agoAdd LOCAL_SANITIZE_RECOVER.
Dan Albert [Fri, 17 Apr 2015 01:08:44 +0000 (18:08 -0700)]
Add LOCAL_SANITIZE_RECOVER.

This is needed for projects that have known ubsan issues that we can't
fix right away (perhaps because it's an upstream project that we're
diligent about keeping in sync with upsteam).

Also make the normal ubsan use -fno-sanitize-recover=all by default.

Change-Id: I1b0f3309792f32dbd08c18816d7306e76c8d7c30

9 years agoMerge "Fix a trivial comment typo."
Elliott Hughes [Sat, 25 Apr 2015 18:04:21 +0000 (18:04 +0000)]
Merge "Fix a trivial comment typo."

9 years agoFix a trivial comment typo.
Elliott Hughes [Sat, 25 Apr 2015 18:02:40 +0000 (11:02 -0700)]
Fix a trivial comment typo.

Bug: https://code.google.com/p/android/issues/detail?id=170069
Change-Id: Iacd308893561a991971b191d4d10a78ca82753d4

9 years agoMerge "Check partition hash before resorting to block_image_verify"
Sami Tolvanen [Sat, 25 Apr 2015 08:00:01 +0000 (08:00 +0000)]
Merge "Check partition hash before resorting to block_image_verify"

9 years agoMerge "host compiler: enable compiler hardening flags"
Nick Kralevich [Sat, 25 Apr 2015 04:35:46 +0000 (04:35 +0000)]
Merge "host compiler: enable compiler hardening flags"

9 years agohost compiler: enable compiler hardening flags
Nick Kralevich [Sat, 25 Apr 2015 04:26:08 +0000 (21:26 -0700)]
host compiler: enable compiler hardening flags

Enable the following compiler hardening flags:

* -Wl,-z,relro
* -Wl,-z,now
* -fstack-protector

relro / bind_now make the relro region read-only after linking,
preventing certain attacks against ELF data structures.

stack-protector adds stack canaries, which can detect exploits
which overwrite parts of the stack.

Explicitly not added in this change is FORTIFY_SOURCE=2. Adding that
option turns on glibc's warn_unused_result attributes. This generates
a huge number of new compile time warnings, and for the multiple
makefiles which have -Werror in them, turns those warnings into errors.
I'm not able to fix all the errors right away.

Bug: 20558757
Change-Id: I86791177c6695f5325233d9dd9a5dd3ccc2b1a2f

9 years agoMerge "Pack relocations for mips64"
Dmitriy Ivanov [Fri, 24 Apr 2015 23:51:28 +0000 (23:51 +0000)]
Merge "Pack relocations for mips64"

9 years agoPack relocations for mips64
Dmitriy Ivanov [Fri, 24 Apr 2015 23:32:38 +0000 (16:32 -0700)]
Pack relocations for mips64

Bug: http://b/18051137
Change-Id: I063c5d6cf007658c0efe9cb1d65ed8026b3e7dae

9 years agoMerge "Do not pack relocations for prebuilts by default"
Dmitriy Ivanov [Fri, 24 Apr 2015 22:56:08 +0000 (22:56 +0000)]
Merge "Do not pack relocations for prebuilts by default"

9 years agoDo not pack relocations for prebuilts by default
Dmitriy Ivanov [Fri, 24 Apr 2015 22:33:14 +0000 (15:33 -0700)]
Do not pack relocations for prebuilts by default

 Do not pack relocations for prebuilds unless
 LOCAL_PACK_MODULE_RELOCATIONS is true

Bug: http://b/20537715
Bug: http://b/18051137
Change-Id: Iddef3b09741da6ae10d73c98103b868cc7695d38

9 years agoMerge "Pack relocations for mips"
Dmitriy Ivanov [Fri, 24 Apr 2015 19:45:22 +0000 (19:45 +0000)]
Merge "Pack relocations for mips"

9 years agoPack relocations for mips
Dmitriy Ivanov [Fri, 24 Apr 2015 19:20:02 +0000 (12:20 -0700)]
Pack relocations for mips

Bug: http://b/18051137
Change-Id: Id5e0adfb55273f23e1c99b4eecb318c6b8d19989

9 years agoMerge "Pack relocations for x86"
Dmitriy Ivanov [Fri, 24 Apr 2015 17:44:09 +0000 (17:44 +0000)]
Merge "Pack relocations for x86"

9 years agoPack relocations for x86
Dmitriy Ivanov [Fri, 24 Apr 2015 01:43:05 +0000 (18:43 -0700)]
Pack relocations for x86

Bug: http://b/18051137
Change-Id: I3f316343e9bbd2037a9583d5b6fb41a06be153cd

9 years agoCheck partition hash before resorting to block_image_verify
Sami Tolvanen [Fri, 24 Apr 2015 10:54:01 +0000 (11:54 +0100)]
Check partition hash before resorting to block_image_verify

It's not necessary to call block_image_verify for the most common
case where we attempt an update for the first time and the target
partition has expected contents. It is sufficient to confirm that
the target partition is intact.

Computing a hash is much faster than simulating the update, which
means this change should make the update speed for v3 much closer
to v2 for the common case. If the update was interrupted, this is
going to make resuming somewhat slower, of course. Hopefully this
should be rather rare.

Bug: 20309033
Change-Id: I04792f3681bf2cd5d9a4e1c41eb883c7f26c6f58

9 years agoMerge "Disable packed relocations for NDK builds"
Dmitriy Ivanov [Fri, 24 Apr 2015 04:48:43 +0000 (04:48 +0000)]
Merge "Disable packed relocations for NDK builds"

9 years agoDisable packed relocations for NDK builds
Dmitriy Ivanov [Fri, 24 Apr 2015 01:48:52 +0000 (18:48 -0700)]
Disable packed relocations for NDK builds

Bug: http://b/20537715
Change-Id: I1289a88d5eb694980b026b695cb25edd8ba76ed5