OSDN Git Service

android-x86/build.git
8 years agoMerge "Clean generated source dir if .proto/.rs is gonee."
Ying Wang [Wed, 2 Mar 2016 19:01:28 +0000 (19:01 +0000)]
Merge "Clean generated source dir if .proto/.rs is gonee."

8 years agoMerge "Support arch specific variants of LOCAL_STRIP_MODULE/LOCAL_PACK_MODULE_RELOCAT...
Ying Wang [Wed, 2 Mar 2016 19:01:24 +0000 (19:01 +0000)]
Merge "Support arch specific variants of LOCAL_STRIP_MODULE/LOCAL_PACK_MODULE_RELOCATIONS"

8 years agoClean generated source dir if .proto/.rs is gonee.
Ying Wang [Wed, 2 Mar 2016 06:05:25 +0000 (22:05 -0800)]
Clean generated source dir if .proto/.rs is gonee.

Similar to how we clean the generated source dir if .aidl becomes
obsolete.

Bug: 26688474
Change-Id: I66c5b6d6513259b5f4f4bac84f384f4bf953d82a
(cherry-pick from commit d5ffec938344defd437d8e315d6a3376b1100a5a)

8 years agoSupport arch specific variants of LOCAL_STRIP_MODULE/LOCAL_PACK_MODULE_RELOCATIONS
Ying Wang [Wed, 2 Mar 2016 04:14:52 +0000 (20:14 -0800)]
Support arch specific variants of LOCAL_STRIP_MODULE/LOCAL_PACK_MODULE_RELOCATIONS

So that we can disable only one in multilib modules.

Bug: 27442756
Change-Id: I4ca379fac997f9165c47cb93d34bf1f483f5a241
(cherry-pick from commit b6eb8ddbde11de0bc59cfedf7adb91a83c27d279)

8 years agoMerge "Ensure prebuilt executables are executable"
Dan Willemsen [Wed, 2 Mar 2016 18:12:06 +0000 (18:12 +0000)]
Merge "Ensure prebuilt executables are executable"

8 years agoMerge "Use Jack to check build"
Yohann Roussel [Wed, 2 Mar 2016 08:46:24 +0000 (08:46 +0000)]
Merge "Use Jack to check build"

8 years agoEnsure prebuilt executables are executable
Dan Willemsen [Wed, 2 Mar 2016 06:04:57 +0000 (22:04 -0800)]
Ensure prebuilt executables are executable

Use 'chmod +x' to make sure that they're executable, even if the source
file is not.

Change-Id: Id155f6b4ff188f5c909fd6237fd64183af264f79

8 years agoMerge changes I12810c1b,I62287a80,Ied96fdc3
Dan Willemsen [Tue, 1 Mar 2016 23:18:27 +0000 (23:18 +0000)]
Merge changes I12810c1b,I62287a80,Ied96fdc3

* changes:
  Remove LOCAL_ACP_UNAVAILABLE
  Remove unused dependencies on $(ACP)
  Replace some uses of acp with cp

8 years agoMerge "Adding a brillo_tests task."
Ari Hausman-Cohen [Tue, 1 Mar 2016 22:38:07 +0000 (22:38 +0000)]
Merge "Adding a brillo_tests task."

8 years agoRemove LOCAL_ACP_UNAVAILABLE
Dan Willemsen [Mon, 29 Feb 2016 21:20:29 +0000 (13:20 -0800)]
Remove LOCAL_ACP_UNAVAILABLE

Now that copy-file-to-target doesn't use acp, nothing in the acp build
path uses acp, so we don't need to special case it to prevent loops.

Change-Id: I12810c1b064d0c03135a80077a76bc4c9cc18b24

8 years agoRemove unused dependencies on $(ACP)
Dan Willemsen [Tue, 1 Mar 2016 01:52:39 +0000 (17:52 -0800)]
Remove unused dependencies on $(ACP)

Most of these are calling to copy-file-to-target or similar, which no
longer use $(ACP).

Change-Id: I62287a80c577c34df587b74e70055c2f56050ce7

8 years agoReplace some uses of acp with cp
Dan Willemsen [Mon, 29 Feb 2016 20:45:18 +0000 (12:45 -0800)]
Replace some uses of acp with cp

ACP was originally created to overcome differences with cp between
Linux, Darwin, and Windows. We've since dropped Windows as a build host,
and don't use features like '-u' anymore.

For most of our current usecases, 'acp' is identical to 'cp' except that
it splits 'cp -p' into 'acp -p' for mode and ownership, and 'acp -t' for
timestamps. On Linux, this could be specified using
'--preserve=mode,ownership', but Darwin doesn't have this.

Since we're removing the destination file before copying, 'cp' already
preserves the mode (modulated by the umask). So the only extra thing
that gets preserved with 'acp -p' is ownership, which we should not care
about in the build system. (In many cases we shouldn't be preserving
mode either, so that readonly source trees can actually be marked
readonly, but that will be a future change)

Change-Id: Ied96fdc303ac5c774347c07363daec8b6dfb22e4

8 years agoMerge "Remove unused config variables"
Dan Willemsen [Tue, 1 Mar 2016 18:49:37 +0000 (18:49 +0000)]
Merge "Remove unused config variables"

8 years agoMerge "Remove destination before copying file"
Dan Willemsen [Tue, 1 Mar 2016 18:20:06 +0000 (18:20 +0000)]
Merge "Remove destination before copying file"

8 years agoUse Jack to check build
Yohann Roussel [Tue, 18 Aug 2015 17:18:18 +0000 (19:18 +0200)]
Use Jack to check build

Run jack with no outputs as a quick check for compilation errors and use
a timestamp to store that check was made.

Bug: 19069325

(cherry picked from commit 43084d9f49a994a7eaebd3c94758d0ebf4c74b6e)

Change-Id: I9b84b503b28cfdfa245f91da0061ee3a79386b28

8 years agoRemove unused config variables
Dan Willemsen [Tue, 1 Mar 2016 09:14:03 +0000 (01:14 -0800)]
Remove unused config variables

These either don't exist or are not used.

Change-Id: I219e4691337223357797fae47ce112df8428bdfa

8 years agoMerge "Remove special cases in installclean"
Dan Willemsen [Tue, 1 Mar 2016 07:04:40 +0000 (07:04 +0000)]
Merge "Remove special cases in installclean"

8 years agoRemove destination before copying file
Dan Willemsen [Mon, 29 Feb 2016 20:45:18 +0000 (12:45 -0800)]
Remove destination before copying file

If there's a symlink as the destination to one of these macros,
currently we'll write to the destination of that symlink instead of
overwriting the symlink. We've run into this a few times when a module
is added to replace a symlink that used to exist via
LOCAL_POST_INSTALL_CMD. These have required manual discovery, and
additions to CleanSpec.mk files:
http://android-review.googlesource.com/143334

Use `rm -f` for single-file targets to remove the destination before
copying. On Linux, `cp --remove-destination` can work, but is not
supported by Darwin or acp.

There may still be problems with dependencies when symlinks are
involved, since ninja will use the destination of the symlink to check
whether it is up to date. But at least with this change, if any
dependency gets regenerated, we'll properly reset the file.

Change-Id: I6d3ac0bd9ced5e21a0ff9dad0eaff012a7bc9c75

8 years agoMerge "sandybridge setup: no AES_NI, AVX, MOVBE"
Olof Johansson [Mon, 29 Feb 2016 23:27:04 +0000 (23:27 +0000)]
Merge "sandybridge setup: no AES_NI, AVX, MOVBE"

8 years agoRemove special cases in installclean
Dan Willemsen [Mon, 29 Feb 2016 23:15:16 +0000 (15:15 -0800)]
Remove special cases in installclean

When switch build types or products, it's no longer necessary to special
case these projects that change their command line based on the build
type or product. Ninja keeps track of the command line last used to
create a file, and will mark it as dirty if the new command line is
different.

Change-Id: I905ff9599eae2952bddc05e7328f77f0849be20a

8 years agoMerge "Remove USE_NINJA=false"
Dan Willemsen [Mon, 29 Feb 2016 22:16:14 +0000 (22:16 +0000)]
Merge "Remove USE_NINJA=false"

8 years agoAdding a brillo_tests task.
Ari Hausman-Cohen [Mon, 29 Feb 2016 21:08:49 +0000 (13:08 -0800)]
Adding a brillo_tests task.

Places whitelisted brillo tests and the whitelist itself
in a zip when run.

BUG: 27385399
Change-Id: I93c2ea8cc521292a6de811bb47bc87a727edd21f
TEST: manual make dist brillo_tests, confirmed desired files were in zip.

8 years agoMerge "delete recovery policy on changing build types"
Nick Kralevich [Mon, 29 Feb 2016 16:44:37 +0000 (16:44 +0000)]
Merge "delete recovery policy on changing build types"

8 years agoMerge "Check LEGACY_USE_JAVA7 before using 1.8."
Nicolas Geoffray [Mon, 29 Feb 2016 12:27:58 +0000 (12:27 +0000)]
Merge "Check LEGACY_USE_JAVA7 before using 1.8."

8 years agoCheck LEGACY_USE_JAVA7 before using 1.8.
Nicolas Geoffray [Mon, 29 Feb 2016 12:07:00 +0000 (12:07 +0000)]
Check LEGACY_USE_JAVA7 before using 1.8.

Fixes current bots not updated to Java8 yet.

Change-Id: I61f5a3f0de6ca0ee9eca64f0ee08454f93e59704

8 years agoMerge "Link everything in libpthread.a to ckati"
Shinichiro Hamaji [Mon, 29 Feb 2016 11:04:36 +0000 (11:04 +0000)]
Merge "Link everything in libpthread.a to ckati"

8 years agoMerge "Change the default for LOCAL_JAVA_LANGUAGE_VERSION from 1.7 to 1.8"
Neil Fuller [Mon, 29 Feb 2016 10:58:09 +0000 (10:58 +0000)]
Merge "Change the default for LOCAL_JAVA_LANGUAGE_VERSION from 1.7 to 1.8"

8 years agodelete recovery policy on changing build types
Nick Kralevich [Sat, 27 Feb 2016 18:41:41 +0000 (10:41 -0800)]
delete recovery policy on changing build types

When the build type changes (for example, from "shamu-userdebug"
to "shamu-user"), the build system doesn't delete all files
and start over. Rather, build artifacts from the old build type
are reused for the new build type.

This is problematic for the recovery SELinux policy, which differs
between build types. Reusing a userdebug policy on a user build
is inappropriate and could lead to security bugs.

Force the deletion of the recovery SELinux policy when changing
build types, so it can be properly regenerated. This is consistent
with how we treat the normal SELinux policy (see commit
a8b3d54101eccb9950651103c199edf0ce2520f7).

Change-Id: I4ebafe3712dc121644828f6538865061aad58cc0

8 years agoRemove USE_NINJA=false
Dan Willemsen [Wed, 6 Jan 2016 22:28:36 +0000 (14:28 -0800)]
Remove USE_NINJA=false

And everything special-cased on that. Add a warning if USE_NINJA is
set to let users know that it no longer changes anything.

Change-Id: Ib8739151fe26ea6bf8f76b7ac2b8f4097dab0b47

8 years agoMerge "Make multiple rules to generate a file an error"
Dan Willemsen [Sat, 27 Feb 2016 05:12:28 +0000 (05:12 +0000)]
Merge "Make multiple rules to generate a file an error"

8 years agoMerge "releasetools: Add --oem_no_mount option."
Tao Bao [Sat, 27 Feb 2016 00:08:39 +0000 (00:08 +0000)]
Merge "releasetools: Add --oem_no_mount option."

8 years agoreleasetools: Add --oem_no_mount option.
Tao Bao [Fri, 26 Feb 2016 03:49:55 +0000 (19:49 -0800)]
releasetools: Add --oem_no_mount option.

We may have devices with OEM-specific properties but without an OEM
partition (e.g. the properties might be set by init based on hardware
SKUs). For such devices, we supply --oem_no_mount to skip mounting the
OEM partition in the updater-script. The option is only meaningful when
-o (--oem_settings) is specified.

Bug: 27359929
Change-Id: Ic08396e478a82be4188e980e704b33b4f704a8d7

8 years agoChange the default for LOCAL_JAVA_LANGUAGE_VERSION from 1.7 to 1.8
Neil Fuller [Tue, 16 Feb 2016 13:04:51 +0000 (13:04 +0000)]
Change the default for LOCAL_JAVA_LANGUAGE_VERSION from 1.7 to 1.8

Modify the compiler flags for Jack and javac.

This has the following effects:
1) Generally, some of the type inference rules changed.
2) javac: bytecode is generated with the v52 major version (not v51)
3) jack: Java 8 language features are supported.

The javac / dx toolchain does not support Java 8 language features.

Bug: 26753820
(cherry picked from commit fda1ace26116a6677cc77c92c24e5259817fb86e)

Change-Id: I07769de473775d95b13feb38c0eb37086eb120f7

8 years agoLink everything in libpthread.a to ckati
Shinichiro Hamaji [Fri, 26 Feb 2016 07:58:13 +0000 (16:58 +0900)]
Link everything in libpthread.a to ckati

With this workaround, we can delete a lot of unnecessary code
in ckati: https://github.com/google/kati/commit/8380fb8525adfb5

Change-Id: I3de1bf4cd6f2a65ee9322e8504459621871612fc

8 years agoMerge "releasetools: Support generating downgrade incremental OTAs."
Tao Bao [Fri, 26 Feb 2016 00:39:09 +0000 (00:39 +0000)]
Merge "releasetools: Support generating downgrade incremental OTAs."

8 years agoMerge "Add a global define for non-external code."
Dan Albert [Thu, 25 Feb 2016 22:17:42 +0000 (22:17 +0000)]
Merge "Add a global define for non-external code."

8 years agoAdd a global define for non-external code.
Dan Albert [Sat, 30 Jan 2016 00:51:51 +0000 (16:51 -0800)]
Add a global define for non-external code.

There are some code policies we want to enforce more strictly, but
it's hard to do so for third party code because we have to either
carry the diff burden or upstream the patch, and in the latter case
the turnaround time for fixes can be problematic, and sometimes
upstream won't accept changes (sometimes people just need to win the
obfuscated C contest).

We define ANDROID_STRICT for any code that we expect to be able to
make these policy fixes as we change policies.

Change-Id: I15faf62cec1932dd859a082f66942b2606d0ff45

8 years agoMerge "Don't ignore LOCAL_INIT_RC if LOCAL_INIT_RC_32/64 is used."
Ying Wang [Thu, 25 Feb 2016 18:28:57 +0000 (18:28 +0000)]
Merge "Don't ignore LOCAL_INIT_RC if LOCAL_INIT_RC_32/64 is used."

8 years agoMerge "Use new clang scan-build script."
Chih-hung Hsieh [Thu, 25 Feb 2016 17:16:54 +0000 (17:16 +0000)]
Merge "Use new clang scan-build script."

8 years agoUse new clang scan-build script.
Chih-Hung Hsieh [Thu, 25 Feb 2016 00:42:15 +0000 (16:42 -0800)]
Use new clang scan-build script.

The actual clang compiler called for static analysis is decided
by build/core/binary.mk, not the one given to --use-analyzer.

BUG: 13287788
Change-Id: I58105c20b56ce17ddf329a275c750d14284d1e25

8 years agoDon't ignore LOCAL_INIT_RC if LOCAL_INIT_RC_32/64 is used.
Ying Wang [Thu, 25 Feb 2016 01:34:52 +0000 (17:34 -0800)]
Don't ignore LOCAL_INIT_RC if LOCAL_INIT_RC_32/64 is used.

It's a valid situation for all three of LOCAL_INIT_RC, LOCAL_INIT_RC_32,
and LOCAL_INIT_RC_64 to be used.

Bug: 26773181
Change-Id: If9661f93b1823279075fc3d55195f7a939e01b6f

8 years agoMerge "Mark the cortex-{a7,a53,a53.57} chips as supporting LPAE"
Alex Light [Wed, 24 Feb 2016 23:26:02 +0000 (23:26 +0000)]
Merge "Mark the cortex-{a7,a53,a53.57} chips as supporting LPAE"

8 years agoMark the cortex-{a7,a53,a53.57} chips as supporting LPAE
Alex Light [Wed, 24 Feb 2016 21:08:58 +0000 (13:08 -0800)]
Mark the cortex-{a7,a53,a53.57} chips as supporting LPAE

Bug: 27340895
Bug: 27324856

Change-Id: Id3994c6d334c8bb673fc3849550f591136a8dd6a

8 years agoMerge "Update to use the latest clang-2629532."
Stephen Hines [Wed, 24 Feb 2016 21:11:13 +0000 (21:11 +0000)]
Merge "Update to use the latest clang-2629532."

8 years agoreleasetools: Support generating downgrade incremental OTAs.
Tao Bao [Tue, 23 Feb 2016 19:38:39 +0000 (11:38 -0800)]
releasetools: Support generating downgrade incremental OTAs.

Add --downgrade flag to ota_from_target_files.py script. It allows
generating an incremental OTA that updates from a newer build to an
older one (based on timestamp comparison). "post-timestamp" line in the
metadata file will be replaced by "ota-downgrade=yes". A data wipe will
always be enforced, so "ota-wipe=yes" will also be included in the
metadata file.

Bug: 26883782
Change-Id: Iaa05f662d948b7ab632a9fbb7051cc3f8bf68c21

8 years agoUpdate to use the latest clang-2629532.
Stephen Hines [Wed, 24 Feb 2016 05:58:06 +0000 (21:58 -0800)]
Update to use the latest clang-2629532.

Bug: http://b/25642296
Change-Id: Ib6c94bc57f7145c1cc9b14338126132fc4648509

8 years agosandybridge setup: no AES_NI, AVX, MOVBE
Olof Johansson [Wed, 24 Feb 2016 02:21:42 +0000 (18:21 -0800)]
sandybridge setup: no AES_NI, AVX, MOVBE

Sandy Bridge actually doesn't have all of these options. For example AVX is only
available on the higher-end SKUs (not on Celeron G550).

Change-Id: Ib595a9a6b464626d0c88525c6aaa4d69176645cc

8 years agoMerge "Warn more clearly when there are duplicate copied headers"
Dan Willemsen [Tue, 23 Feb 2016 23:46:25 +0000 (23:46 +0000)]
Merge "Warn more clearly when there are duplicate copied headers"

8 years agoMake multiple rules to generate a file an error
Colin Cross [Wed, 23 Sep 2015 20:52:22 +0000 (13:52 -0700)]
Make multiple rules to generate a file an error

Pass -w dupbuild=err to ninja to make defining multiple rules to build a
file an error instead of a warning.  Proceeding with the build would
result in undefined behavior, and nobody notices the warning.

Change-Id: Iadac88f8835121a8685bff835acba638100bb654

8 years agoWarn more clearly when there are duplicate copied headers
Dan Willemsen [Tue, 23 Feb 2016 21:40:07 +0000 (13:40 -0800)]
Warn more clearly when there are duplicate copied headers

When more than one makefile tries to copy a header to the same
destination, the warning is not clear, and hard to track down and assign
blame:

build/core/copy_headers.mk:15: warning: ignoring old commands for target `out/target/product/bullhead/obj/include/qcom/display/copybit.h'

With this change, the same behavior is kept, but the warning message is
more descriptive, and contains the offending Android.mk files:

build/core/Makefile:54: Duplicate header copy: out/target/product/bullhead/obj/include/qcom/display/copybit.h
build/core/Makefile:54: Defined in: hardware/qcom/display/msm8994/libcopybit/Android.mk hardware/qcom/display/msm8994/libcopybit/Android.mk

In this case, a $(CLEAR_VARS) is missing, so the same Android.mk file is
copying the same headers twice.

Bug: 27302058
Change-Id: Icf8f580ae71a78741db21c1d8f3213424459e637

8 years agoMerge "Make sure my_src_jar is set up properly for host prebuilt jar"
Ying Wang [Tue, 23 Feb 2016 18:06:01 +0000 (18:06 +0000)]
Merge "Make sure my_src_jar is set up properly for host prebuilt jar"

8 years agoMerge "Add profman to the build scripts"
Calin Juravle [Tue, 23 Feb 2016 14:07:40 +0000 (14:07 +0000)]
Merge "Add profman to the build scripts"

8 years agoAdd profman to the build scripts
Calin Juravle [Fri, 19 Feb 2016 16:54:20 +0000 (16:54 +0000)]
Add profman to the build scripts

Bug: 26719109
Bug: 26563023

(cherry picked from commit aab43390a050a4eeb2914a3186994ca8b6a5c4e6)

Change-Id: Idc49872eebb283436fcdd02fe095e8c8826aff44

8 years agoMake sure my_src_jar is set up properly for host prebuilt jar
Ying Wang [Mon, 22 Feb 2016 23:54:27 +0000 (15:54 -0800)]
Make sure my_src_jar is set up properly for host prebuilt jar

Make sure my_src_jar is set up properly for host prebuilt jar when we
need to generate host .jack for host dalvik java libraries.

Change-Id: If85e27147cdc6e6a7a154c1cf308f9d0a71ff068

8 years agoMerge "clang now supports -fdebug-prefix-map=."
Elliott Hughes [Sat, 20 Feb 2016 00:27:25 +0000 (00:27 +0000)]
Merge "clang now supports -fdebug-prefix-map=."

8 years agoclang now supports -fdebug-prefix-map=.
Elliott Hughes [Fri, 19 Feb 2016 21:34:57 +0000 (13:34 -0800)]
clang now supports -fdebug-prefix-map=.

Change-Id: Id0569831f09845871d8e6990c310516bea5a2510

8 years agoMerge "Switch to production constant for v2 signing scheme ID."
Alex Klyubin [Wed, 17 Feb 2016 16:41:14 +0000 (16:41 +0000)]
Merge "Switch to production constant for v2 signing scheme ID."

8 years agoMerge "Add support for testing obfuscated app with Jack"
Yohann Roussel [Wed, 17 Feb 2016 16:05:18 +0000 (16:05 +0000)]
Merge "Add support for testing obfuscated app with Jack"

8 years agoAdd support for testing obfuscated app with Jack
Yohann Roussel [Tue, 3 Nov 2015 17:33:21 +0000 (18:33 +0100)]
Add support for testing obfuscated app with Jack

Bug: 17481317

Change-Id: I80cb4e3991e858a39e7415fb7f1896d6992c950f

8 years agoMerge "Include Brillo's cannonical .clang-format file."
Alex Deymo [Tue, 16 Feb 2016 22:32:06 +0000 (22:32 +0000)]
Merge "Include Brillo's cannonical .clang-format file."

8 years agoInclude Brillo's cannonical .clang-format file.
Alex Deymo [Wed, 10 Feb 2016 18:11:05 +0000 (10:11 -0800)]
Include Brillo's cannonical .clang-format file.

Many Brillo repositories need to share a common .clang-format file to
ensure the same formatting rules. This patch moves the .clang-format
file already in system/core/metricsd to a common location to be shared
by other Brillo repositories.

Bug: 27121653
TEST=symlink from system/update_engine, ran clang-format.

Change-Id: Ie04a5a9cf54b9cc24f180fe3896501db4d883a64

8 years agoMerge "Clear logd intermediates"
Sami Tolvanen [Tue, 16 Feb 2016 21:52:13 +0000 (21:52 +0000)]
Merge "Clear logd intermediates"

8 years agoMerge "Define a class for renderscript bitcode files"
Dan Willemsen [Tue, 16 Feb 2016 21:23:36 +0000 (21:23 +0000)]
Merge "Define a class for renderscript bitcode files"

8 years agoMerge "makeparallel: unset MAKEFLAGS and MAKELEVEL"
Colin Cross [Tue, 16 Feb 2016 21:08:59 +0000 (21:08 +0000)]
Merge "makeparallel: unset MAKEFLAGS and MAKELEVEL"

8 years agoSwitch to production constant for v2 signing scheme ID.
Alex Klyubin [Tue, 16 Feb 2016 20:42:37 +0000 (12:42 -0800)]
Switch to production constant for v2 signing scheme ID.

Bug: 25794543
Change-Id: I0e06f31ea3117c6865d18f853f99304964b323b4

8 years agoMerge "Add new package: java.util.function"
Neil Fuller [Tue, 16 Feb 2016 13:15:29 +0000 (13:15 +0000)]
Merge "Add new package: java.util.function"

8 years agoAdd new package: java.util.function
Neil Fuller [Wed, 27 Jan 2016 15:57:20 +0000 (15:57 +0000)]
Add new package: java.util.function

Bug: 26814204
(cherry-picked from commit 9429a880a347de14ecc4971cb9fe25b52fab7950)

Change-Id: Ibd4f7a985ddcdb7564a79326ce846257882e294f

8 years agoDefine a class for renderscript bitcode files
Dan Willemsen [Sat, 13 Feb 2016 08:19:40 +0000 (00:19 -0800)]
Define a class for renderscript bitcode files

These have been using SHARED_LIBRARIES, but aren't elf shared libraries.
Continue installing them to /system/lib[64], but do not apply any other
normal shared library logic to them.

Change-Id: I3055ff86bb7b116c7107c41578ed6f0f304b1cf1

8 years agoMerge "Remove debug statement"
Dan Willemsen [Fri, 12 Feb 2016 23:41:37 +0000 (23:41 +0000)]
Merge "Remove debug statement"

8 years agoMerge "Use newest clang static analyzers."
Chih-hung Hsieh [Fri, 12 Feb 2016 17:33:40 +0000 (17:33 +0000)]
Merge "Use newest clang static analyzers."

8 years agoUse newest clang static analyzers.
Chih-Hung Hsieh [Tue, 2 Feb 2016 19:30:39 +0000 (11:30 -0800)]
Use newest clang static analyzers.

* When WITH_STATIC_ANALYZER is set and non-zero, and clang compiler is used,
  call new clang ccc-analyzer or c++-analyzer.
* Otherwise, if WITH_SYNTAX_CHECK is set and non-zero,
  call compiler with -fsyntax-only.
* Replace "--sysroot=path" with "--sysroot path", to work with ccc-analyzer.
* ccc-analyzer executes the original compilation command to generate
  object files before calling clang with --analyze to do static analysis.
* When clang is called with --analyze, macro __clang_analyzer__ is defined.

BUG: 13287788

Change-Id: I5edb25b52998d871385dd000778db2ce83224078

8 years agoMerge "fs_config: error on multiple specifications of android_filesystem_config.h"
Dan Willemsen [Thu, 11 Feb 2016 21:01:18 +0000 (21:01 +0000)]
Merge "fs_config: error on multiple specifications of android_filesystem_config.h"

8 years agoMerge "Makefile: Add the missing library in otatools.zip."
Tao Bao [Thu, 11 Feb 2016 18:51:04 +0000 (18:51 +0000)]
Merge "Makefile: Add the missing library in otatools.zip."

8 years agoMakefile: Add the missing library in otatools.zip.
Tao Bao [Thu, 11 Feb 2016 17:50:41 +0000 (09:50 -0800)]
Makefile: Add the missing library in otatools.zip.

With the change in [1], delta_generator now needs
libprotobuf-cpp-lite.so instead of libprotobuf-cpp-lite-rtti.so to
generate A/B payloads.

[1] commit ab5bd668f6be600a8cceb8772e426c0aa902a5e1

Bug: 27145830
Change-Id: Ib9a93bf0fbe7fa44fc5fb94668d17fa1a2e07b05
(cherry picked from commit fdd2693b650d24d98dd694508724b901f50f278f)

8 years agofs_config: error on multiple specifications of android_filesystem_config.h
William Roberts [Thu, 11 Feb 2016 18:35:07 +0000 (10:35 -0800)]
fs_config: error on multiple specifications of android_filesystem_config.h

Multiple files added to TARGET_ANDROID_FILESYSTEM_CONFIG_H will build,
however the tooling will only respect the first header found. The
sanity check in the makefile will only check the first path found for
proper naming convention. One can end up with a successful build that
does not respect all files in the TARGET_ANDROID_FILESYSTEM_CONFIG_H.

Check that one and only one item is set to
TARGET_ANDROID_FILESYSTEM_CONFIG_H variable.

Output:
*** Multiple fs_config files specified, see "device/lge/bullhead/two/android_filesystem_config.h device/lge/bullhead/one/android_filesystem_config.h"..

Change-Id: I70a43f44713b273cb9a78f54ff16854b9e11af4a
Signed-off-by: William Roberts <william.c.roberts@intel.com>
8 years agoMerge "Remove usages of jill.jar, use jack instead"
Sebastien Hertz [Thu, 11 Feb 2016 08:48:44 +0000 (08:48 +0000)]
Merge "Remove usages of jill.jar, use jack instead"

8 years agoClear logd intermediates
Sami Tolvanen [Wed, 10 Feb 2016 18:41:26 +0000 (10:41 -0800)]
Clear logd intermediates

Needed by changes from
  Idcdc5bff133f13c1267f0ec0a75cc8cf1ddbda0d

Bug: 26902605
Change-Id: Id3ca7889ede30b54b7af73dd50653ca1a20d59aa

8 years agoRemove debug statement
Dan Willemsen [Wed, 10 Feb 2016 22:39:27 +0000 (14:39 -0800)]
Remove debug statement

Change-Id: Ic8b6082a61e2fb1b09b2b254312d3dc7ef59d987

8 years agoMerge "Sign APKs using SHA-256 instead of SHA-1 when possible."
Alex Klyubin [Wed, 10 Feb 2016 21:35:05 +0000 (21:35 +0000)]
Merge "Sign APKs using SHA-256 instead of SHA-1 when possible."

8 years agoMerge "Fix metadata location when file system doesn't span the partition"
Sami Tolvanen [Wed, 10 Feb 2016 18:50:17 +0000 (18:50 +0000)]
Merge "Fix metadata location when file system doesn't span the partition"

8 years agoMerge "Tell Soong about HOST_CROSS_*"
Dan Willemsen [Wed, 10 Feb 2016 18:13:22 +0000 (18:13 +0000)]
Merge "Tell Soong about HOST_CROSS_*"

8 years agoSign APKs using SHA-256 instead of SHA-1 when possible.
Alex Klyubin [Wed, 13 Jan 2016 18:32:47 +0000 (10:32 -0800)]
Sign APKs using SHA-256 instead of SHA-1 when possible.

This changes the build system to provide the signapk tool with the
minSdkVersion of the APK being signed. signapk in turn will then use
SHA-256 instead of SHA-1 if minSdkVersion is 18 (JB MR2) or higher
(see c2c49ed0c13846f7f96249c7419971dfcddc9215).

To avoid increasing incremental OTA update package sizes for already
released platforms, release build scripts disable the above logic when
signing target files ZIPs for pre-N platforms.

Bug: 25643280

Change-Id: I3f2faaf49c6fa392ffbf1ee9f30de476f9f73231

8 years agoFix metadata location when file system doesn't span the partition
Sami Tolvanen [Tue, 9 Feb 2016 20:28:58 +0000 (12:28 -0800)]
Fix metadata location when file system doesn't span the partition

Pad the sparse image with a zero fill chunk to correctly position
verity and FEC metadata at the end of the partition.

Bug: 27073791
Change-Id: I9f70d579a42e5007d50e9c02a98a608d2815f0ed

8 years agoRemove usages of jill.jar, use jack instead
Yohann Roussel [Tue, 19 Jan 2016 15:08:34 +0000 (16:08 +0100)]
Remove usages of jill.jar, use jack instead

Bug: 27113650
(cherry picked from commit d0bb7bdf5a62fd1cd0469e0b5d6b54feaecb3cdd)

Change-Id: Ie9be0d5c7e43ea857411431bb79db5cbb476484f

8 years agoMerge "releasetools: Add two new entries to metadata."
Tao Bao [Wed, 10 Feb 2016 17:05:47 +0000 (17:05 +0000)]
Merge "releasetools: Add two new entries to metadata."

8 years agoMerge "improve performance of blockimgdiff"
Doug Zongker [Wed, 10 Feb 2016 16:52:10 +0000 (16:52 +0000)]
Merge "improve performance of blockimgdiff"

8 years agoimprove performance of blockimgdiff
Doug Zongker [Tue, 9 Feb 2016 16:28:09 +0000 (08:28 -0800)]
improve performance of blockimgdiff

Substantially improves the performance of some BlockImageDiff steps on
large images.  GenerateDigraph and AssertSequenceGood are now linear in
the number of blocks in the target image rather than quadratic in the
size of the partition.  FindVertexSequence tracks the score of each
vertex, the set of sources and sinks, and maintains a heap for finding
the best vertex to use for the greedy step, instead of repeatedly
re-scanning the set of remaining vertexes.

Change-Id: Ifeea5053d80ef4f06a5aa239c9a1797dd5e47841

8 years agoMerge "Use Jan 1 2009 as timestamp in OTA update ZIPs."
Alex Klyubin [Wed, 10 Feb 2016 01:08:12 +0000 (01:08 +0000)]
Merge "Use Jan 1 2009 as timestamp in OTA update ZIPs."

8 years agoTell Soong about HOST_CROSS_*
Dan Willemsen [Tue, 9 Feb 2016 22:55:13 +0000 (14:55 -0800)]
Tell Soong about HOST_CROSS_*

This configures Soong to build windows binaries, which requires support
for 64-bit windows binaries in BUILD_PREBUILT for USE_SOONG=true.

module_arch_supported.mk did not support 64-bit being the secondary
architecture when evaluating multilib conditionals. All other uses of
HOST_*_IS_64_BIT already check the proper version.

Change-Id: Iff664733e6991f4adbe8ddd620b091bbb55d1d86

8 years agoMerge "Allow java targets to support 1.8 source / target"
Neil Fuller [Tue, 9 Feb 2016 23:01:40 +0000 (23:01 +0000)]
Merge "Allow java targets to support 1.8 source / target"

8 years agoUse Jan 1 2009 as timestamp in OTA update ZIPs.
Alex Klyubin [Tue, 9 Feb 2016 21:09:57 +0000 (13:09 -0800)]
Use Jan 1 2009 as timestamp in OTA update ZIPs.

This is a follow-up to 6c41036bcf35fe39162b50d27533f0f3bfab3028 where
I forgot to update a section of OTA update ZIP code.

Bug: 26864066
Change-Id: Idbcde71d6377a16807e41c999120eeddd5b4d8a4

8 years agoMerge "Add 64-bit windows cross-compiles"
Dan Willemsen [Tue, 9 Feb 2016 20:54:10 +0000 (20:54 +0000)]
Merge "Add 64-bit windows cross-compiles"

8 years agoMerge "Disable java when JAVA_NOT_REQUIRED is set"
Dan Willemsen [Tue, 9 Feb 2016 05:21:39 +0000 (05:21 +0000)]
Merge "Disable java when JAVA_NOT_REQUIRED is set"

8 years agoDisable java when JAVA_NOT_REQUIRED is set
Dan Willemsen [Sat, 6 Feb 2016 02:24:17 +0000 (18:24 -0800)]
Disable java when JAVA_NOT_REQUIRED is set

Change-Id: Id3056cde93302be20000bf7af33aeed0fa8fdf52

8 years agoMerge "Revert "Do not diff .odex files""
Dan Willemsen [Mon, 8 Feb 2016 22:40:33 +0000 (22:40 +0000)]
Merge "Revert "Do not diff .odex files""

8 years agoRevert "Do not diff .odex files"
Dan Willemsen [Mon, 8 Feb 2016 20:48:48 +0000 (20:48 +0000)]
Revert "Do not diff .odex files"

The bug has been fixed, .odex files are deterministic.

This reverts commit c4438d3d29389d0df0b744c339d67f4a673be96c.

Bug: 26956807
Change-Id: Iaa42dfba5cf79ee64ac20cf4f1f3a4598f4275f4

8 years agoreleasetools: Add two new entries to metadata.
Tao Bao [Thu, 4 Feb 2016 22:26:02 +0000 (14:26 -0800)]
releasetools: Add two new entries to metadata.

Add "ota-required-cache" into the metadata file in an OTA package,
which shows the minimum free space on /cache to apply the update.

Add "ota-type" into the metadata file, which shows the OTA type for
this package (i.e. one of FILE, BLOCK and AB).

Also add the cache free space check into updater-script when generating
block-based incremental OTAs (we only had such lines for file-based
incrementals before).

Bug: 26731903
Change-Id: Id6ff0fc4cdfb1443636b0b3800b0f8bddb5bb1d0

8 years agoAdd 64-bit windows cross-compiles
Dan Willemsen [Sat, 6 Feb 2016 00:20:19 +0000 (16:20 -0800)]
Add 64-bit windows cross-compiles

This is mostly the same as the existing 2ND_HOST / HOST_CROSS support.

The interesting thing I did here was make x86 the 'first' architecture,
and x86_64 the second. This way LOCAL_MULTILIB := first defaults to
32-bit windows modules.

  windows-x86/bin   <- defaults to 32-bit executables
  windows-x86/lib   <- 32-bit libraries, like before
  windows-x86/lib64 <- 64-bit libraries
  windows-x86/obj   <- 32-bit intermediates
  windows-x86/obj64 <- 64-bit intermediates

Then modules are registered with the names:

  host_cross_liblog    <- 32-bit, like before
  host_cross_liblog_64 <- 64-bit

Bug: 26957718
Change-Id: I9f119411acb43e973ec1e6bca3c1dc291c91556c

8 years agoMerge "ota_from_target_files: Include the payload properties."
Alex Deymo [Fri, 5 Feb 2016 23:11:18 +0000 (23:11 +0000)]
Merge "ota_from_target_files: Include the payload properties."

8 years agoMerge "Fix unused src tracking for generated .o files"
Dan Willemsen [Fri, 5 Feb 2016 18:12:20 +0000 (18:12 +0000)]
Merge "Fix unused src tracking for generated .o files"