OSDN Git Service

android-x86/build.git
7 years agoMerge "Allow goldfish-setup to take wake locks"
Treehugger Robot [Fri, 6 Jan 2017 19:57:52 +0000 (19:57 +0000)]
Merge "Allow goldfish-setup to take wake locks"

7 years agoAllow goldfish-setup to take wake locks
Lingfeng Yang [Fri, 6 Jan 2017 18:12:08 +0000 (10:12 -0800)]
Allow goldfish-setup to take wake locks

Change-Id: I6a5db8586168e47e974a82afd2e75fd858aa5750

7 years agoMerge "Build split file_contexts."
Treehugger Robot [Fri, 6 Jan 2017 17:23:45 +0000 (17:23 +0000)]
Merge "Build split file_contexts."

7 years agoMerge "TARGET_RECOVERY_UPDATER_LIBS should not be used with AB updates"
Dan Willemsen [Thu, 5 Jan 2017 22:42:06 +0000 (22:42 +0000)]
Merge "TARGET_RECOVERY_UPDATER_LIBS should not be used with AB updates"

7 years agoBuild split file_contexts.
dcashman [Mon, 12 Dec 2016 18:34:47 +0000 (10:34 -0800)]
Build split file_contexts.

Bug: 31363362
Test: Device uses new file_contexts files for labeling decisions.
Change-Id: I33c3297a2ec806f4b9f5dd25f3dd20c4b1001641
Signed-off-by: Sandeep Patil <sspatil@google.com>
7 years agoTARGET_RECOVERY_UPDATER_LIBS should not be used with AB updates
Dan Willemsen [Thu, 5 Jan 2017 01:16:09 +0000 (17:16 -0800)]
TARGET_RECOVERY_UPDATER_LIBS should not be used with AB updates

Also mark these variables as readonly so that later Android.mk cannot
change them.

Bug: 33607778
Test: build/tools/kati_all_products.sh -j12
Change-Id: I5f9a4df02bfdfd56549728f074bbb3fc251edb3d

7 years agoMerge changes from topic 'fsconfig2-fixups'
Elliott Hughes [Thu, 5 Jan 2017 00:26:04 +0000 (00:26 +0000)]
Merge changes from topic 'fsconfig2-fixups'

* changes:
  fsconfig: fix pylint warning
  fsconfig: fix style formating with yapf

7 years agoMerge changes from topic 'build-vts-zip'
Dan Willemsen [Wed, 4 Jan 2017 22:58:53 +0000 (22:58 +0000)]
Merge changes from topic 'build-vts-zip'

* changes:
  delete vts.mk (which is moved to test/vts)
  include VTS packaging task files

7 years agodelete vts.mk (which is moved to test/vts)
Keun Soo Yim [Fri, 2 Sep 2016 18:43:43 +0000 (11:43 -0700)]
delete vts.mk (which is moved to test/vts)

Test: m -j vts
Change-Id: I3ab29e59b6f77f3ac92bbfe7a6f7a8f36b2ba3bd
(cherry picked from commit a942dc8433c4e02bdfe0313471bcac8fef318fde)

7 years agoinclude VTS packaging task files
Keun Soo Yim [Wed, 31 Aug 2016 16:20:51 +0000 (09:20 -0700)]
include VTS packaging task files

Test: m -j vts
Change-Id: I8737412bf4e7b5c2db32325105e7c588949b959d
(cherry picked from commit 199a710f73d9522b599ff75b8dc8ea54e7fcf6e0)

7 years agofsconfig: fix pylint warning
William Roberts [Wed, 14 Dec 2016 03:17:07 +0000 (19:17 -0800)]
fsconfig: fix pylint warning

Fix pylint warning:
W:292,47: Cell variable lineno defined in loop (cell-var-from-loop)

For the closure function that prints the error message, we actually
want the value of modified lineno within the closure.

Test: run unit tests

Change-Id: I7d08184563d347c826ca1af7558fb8e378f151e1
Signed-off-by: William Roberts <william.c.roberts@intel.com>
7 years agofsconfig: fix style formating with yapf
William Roberts [Wed, 14 Dec 2016 03:12:35 +0000 (19:12 -0800)]
fsconfig: fix style formating with yapf

A few changes got introduced that yapf re-styles. To make it easy,
keep this file formated to:

yapf -i --style=google fs_config_generator.py

Test: run unit tests

Change-Id: I3160b92d7fd07a4d315dde59f1d20fe4c7587da1
Signed-off-by: William Roberts <william.c.roberts@intel.com>
7 years agoMerge "Define TARGET_OUT_VENDOR_APPS_PRIVILEGED for 2nd arch"
Treehugger Robot [Wed, 4 Jan 2017 02:44:04 +0000 (02:44 +0000)]
Merge "Define TARGET_OUT_VENDOR_APPS_PRIVILEGED for 2nd arch"

7 years agoMerge "fs_config: Fix oem range parsing"
Elliott Hughes [Wed, 4 Jan 2017 00:37:58 +0000 (00:37 +0000)]
Merge "fs_config: Fix oem range parsing"

7 years agoMerge "blockimgdiff.py: Clean up stash id computation in BBOTA v3+."
Tao Bao [Tue, 3 Jan 2017 23:44:41 +0000 (23:44 +0000)]
Merge "blockimgdiff.py: Clean up stash id computation in BBOTA v3+."

7 years agoMerge "Remove option for non-pic dex preopt."
Treehugger Robot [Tue, 3 Jan 2017 09:39:38 +0000 (09:39 +0000)]
Merge "Remove option for non-pic dex preopt."

7 years agofs_config: Fix oem range parsing
Johan Redestig [Tue, 3 Jan 2017 08:36:47 +0000 (09:36 +0100)]
fs_config: Fix oem range parsing

Addresses the following build error:
  "value" not in valid range [], got: 2901 for: "AID_ABC" file:\
  "device/somc/abc/config.fs"

that happens when parsing device/somc/abc/config.fs:
  [AID_ABC]
  value: 2901

The problem is that the AID.PREFIX is 'AID_' so the _OEM_RANGE
expression is searching for AID__OEM_RESERVED, which never
exists. Drop the extra '_'.

Change-Id: I8a1556731e14e49cef66e3e91121e4be23a308e0
Test: ./build/tools/fs_config/fs_config_generator.py fsconfig --aid-header=system/core/include/private/android_filesystem_config.h device/somc/abc/config.fs

7 years agoblockimgdiff.py: Clean up stash id computation in BBOTA v3+.
Tao Bao [Wed, 28 Dec 2016 17:14:39 +0000 (09:14 -0800)]
blockimgdiff.py: Clean up stash id computation in BBOTA v3+.

Only BBOTA v2 needs to maintain a pool of available 'stash slot id'.
BBOTA v3+ uses the hash of the stashed blocks as the slot id, which
doesn't need the id pool anymore.

Bug: 33694544
Test: Generate v2 and v4 incrementals w/ and w/o the CL. They produce
      the same packages respectively.
Change-Id: I8121af5b6b1bee98c3639d54a00b06fd12e378e8

7 years agoMerge "Revert "Revert "Merge changes from topic 'fsconfig-2'"""
Elliott Hughes [Tue, 27 Dec 2016 20:51:06 +0000 (20:51 +0000)]
Merge "Revert "Revert "Merge changes from topic 'fsconfig-2'"""

7 years agoMerge "Update pathmap.mk for renderscript support library migration"
Treehugger Robot [Fri, 23 Dec 2016 03:02:54 +0000 (03:02 +0000)]
Merge "Update pathmap.mk for renderscript support library migration"

7 years agoUpdate pathmap.mk for renderscript support library migration
Alan Viverette [Tue, 26 Jul 2016 17:06:37 +0000 (13:06 -0400)]
Update pathmap.mk for renderscript support library migration

Bug: 30255901
Change-Id: I213baaf4c76a7e5b8d20168624b74eb61582d215
(cherry picked from commit 49f03349c818de16be3a0c74f6dc07c85b94b7b0)

7 years agoMerge "Remove --ignore_dirty=$(OUT_DIR)/% from Kati flags"
Dan Willemsen [Thu, 22 Dec 2016 23:27:37 +0000 (23:27 +0000)]
Merge "Remove --ignore_dirty=$(OUT_DIR)/% from Kati flags"

7 years agoRemove --ignore_dirty=$(OUT_DIR)/% from Kati flags
Dan Willemsen [Sat, 10 Sep 2016 06:18:22 +0000 (23:18 -0700)]
Remove --ignore_dirty=$(OUT_DIR)/% from Kati flags

We really should be re-running Kati when a file in here changes, since
it may actually affect the result. The general problem we've run into is
when there are two kati output files that are clean according to the
current rules, but if they actually ran we'd force an installclean.

This also had the side-effect of not running shell commands that touched
$OUT_DIR, like writing files using $(shell echo ... >out/...), now those
may be run.

This will cause Kati to re-read the makefiles more often -- twice when a
CleanSpec.mk changes, twice when switching products on the same device.
Basically every time we read the old state, do something about it, and
write out the new state, we need to rerun the next time so we aren't
continuously cleaning up what changed.

Bug: 30947985
Test: Manual, switching between configurations; enable --regen_debug
Change-Id: I888f9f8295f318f8214b2c79525cf94f457a5a8b

7 years agoMerge "Fix incremental builds with AAPT2"
Dan Willemsen [Thu, 22 Dec 2016 21:28:24 +0000 (21:28 +0000)]
Merge "Fix incremental builds with AAPT2"

7 years agoMerge "embedded.mk: +android.hidl.memory@1.0-service"
Treehugger Robot [Thu, 22 Dec 2016 21:20:36 +0000 (21:20 +0000)]
Merge "embedded.mk: +android.hidl.memory@1.0-service"

7 years agoFix incremental builds with AAPT2
Dan Willemsen [Thu, 22 Dec 2016 21:17:17 +0000 (13:17 -0800)]
Fix incremental builds with AAPT2

When we removed the src/ directory, the R.stamp file was removed -- with
AAPT, this triggers src/android/Manifest.java to be regenerated. But
AAPT2 only uses the R.stamp file as a timestamp to find the
Manifest.java file, not to create it.

So add a new CleanSpec.mk to fix the builds by removing the target that
AAPT2 uses to generate the src/android/Manifest.java file.

Test: None
Change-Id: If56f326ad701f7f92aa7c8aeda57e92d9c62cff9

7 years agoMerge "Support non-gtest NATIVE_TESTS"
Dan Willemsen [Thu, 22 Dec 2016 20:29:41 +0000 (20:29 +0000)]
Merge "Support non-gtest NATIVE_TESTS"

7 years agoMerge "Remove generated java source post clean logic"
Treehugger Robot [Thu, 22 Dec 2016 20:09:13 +0000 (20:09 +0000)]
Merge "Remove generated java source post clean logic"

7 years agoembedded.mk: +android.hidl.memory@1.0-service
Steven Moreland [Thu, 15 Dec 2016 20:30:23 +0000 (12:30 -0800)]
embedded.mk: +android.hidl.memory@1.0-service

This is the service that provides shared memory for hidl/treble
processes.

Bug: 32185232
Test: builds
Change-Id: I79162a781daad7aa704f4ee071fef0bbdea59a18

7 years agoRemove generated java source post clean logic
Dan Willemsen [Thu, 22 Dec 2016 01:37:00 +0000 (17:37 -0800)]
Remove generated java source post clean logic

Just move the proto and renderscript sources into their own directories
-- the entire directory is wiped if the file list changes since the
command line will change. So we can just enable/disable pulling sources
from those directories based on whether there were files in the list or
not.

Bug: 30947985
Test: m -j java
Test: Remove a proto file from a java lib, ensure the generated source
      is not included.
Change-Id: If7529979de6fa62a651933a3a974f47b033851d6

7 years agoMerge "Revert "Revert "Extends VNDK support in Make by setting LOCAL_USE_VNDK for...
Vijay Venkatraman [Thu, 22 Dec 2016 05:54:02 +0000 (05:54 +0000)]
Merge "Revert "Revert "Extends VNDK support in Make by setting LOCAL_USE_VNDK for modules going into vendor partition, except when they already have LOCAL_SDK_VERSION set."""

7 years agoRevert "Revert "Extends VNDK support in Make by setting LOCAL_USE_VNDK for modules...
Vijay Venkatraman [Tue, 20 Dec 2016 17:50:21 +0000 (17:50 +0000)]
Revert "Revert "Extends VNDK support in Make by setting LOCAL_USE_VNDK for modules going into vendor partition, except when they already have LOCAL_SDK_VERSION set.""

This reverts commit 823172e209ab92dbae6dd5a4ef750f73dbfc5aa8.

Test: build.ninja for any target doesnt change, unless
BOARD_VNDK_VERSION is turned on in BoardConfig.mk
Change-Id: I4e1baaf5a2251de20f71a47096ce53e23d7e2d80

7 years agoSupport non-gtest NATIVE_TESTS
Dan Willemsen [Thu, 22 Dec 2016 04:27:06 +0000 (20:27 -0800)]
Support non-gtest NATIVE_TESTS

We've had this in Soong for a while, but never supported it in Make.

Bug: 33819328
Test: Change LOCAL_GTEST on a module, check m -j ... showcommands
Change-Id: Id8f4b9908c82c05d9a986a38945b33bac578e991

7 years agoMerge "HOST_OUT_EXECUTABLES path is not needed in misc_info file"
Tao Bao [Wed, 21 Dec 2016 06:08:23 +0000 (06:08 +0000)]
Merge "HOST_OUT_EXECUTABLES path is not needed in misc_info file"

7 years agoMerge "otatools: added missing shared library for make_ext4fs"
Tao Bao [Wed, 21 Dec 2016 04:30:26 +0000 (04:30 +0000)]
Merge "otatools: added missing shared library for make_ext4fs"

7 years agoMerge "logpersist: introduce split to logcat and logpersist domains"
Treehugger Robot [Tue, 20 Dec 2016 22:08:41 +0000 (22:08 +0000)]
Merge "logpersist: introduce split to logcat and logpersist domains"

7 years agologpersist: introduce split to logcat and logpersist domains
Mark Salyzyn [Wed, 26 Oct 2016 21:19:32 +0000 (14:19 -0700)]
logpersist: introduce split to logcat and logpersist domains

logcat_exec transition to logpersist from init with perms

Test: compile
Bug: 30566487
Change-Id: I0cea55f71be011e921841ef9d3e1283f11606f36

7 years agoMerge "Add ART_USE_READ_BARRIER to product variables."
Hiroshi Yamauchi [Tue, 20 Dec 2016 18:56:19 +0000 (18:56 +0000)]
Merge "Add ART_USE_READ_BARRIER to product variables."

7 years agoHOST_OUT_EXECUTABLES path is not needed in misc_info file
Vikram Dattu [Fri, 16 Dec 2016 08:45:19 +0000 (09:45 +0100)]
HOST_OUT_EXECUTABLES path is not needed in misc_info file

HOST_OUT_EXECUTABLES is already added to the PATH variable,
so it is not needed to add the path info for binaries in
misc_info.txt and <partition>_image_info.txt.

Earlier the mkuserimg item in the build_image dictionary is
hardcoded to "mkuserimg.sh", but now it is customized for
mkuserimg.sh and mkuserimg_mke2fs.sh, and maintained in
dictionary "ext_mkuserimg=$(MKEXTUSERIMG)" in misc_info.txt
and <partition>_image_info.txt, where it is used in the
build_image script while creating the images.

The problem here is the value for this key is set to build
path of the file mkuserimg file
$(HOST_OUT_EXECUTABLES)/mkuserimg.sh,
i.e. out/host/linux_x86/bin/mkuserimg.sh,
there by standalone signing the images using otatools is
not working as the executables are packed in bin folder.

Test: tools/releasetools/sign_target_files_apks
    -p <extracted ota-tools.zip folder>
    --extra_signapk_args=-f /etc/opt/cert_data.dat
    -v
    --replace_verity_private_key ~/build/target/product/security/verity
    --replace_verity_public_key ~/build/target/product/security/verity.x509.pem
    -k <key maping>
    <input target files zip>
    <output target files zip>

Change-Id: I57af1025ec38f3794f779c49faa0bf965afc6a5d

7 years agootatools: added missing shared library for make_ext4fs
Vikram Dattu [Fri, 16 Dec 2016 15:33:03 +0000 (16:33 +0100)]
otatools: added missing shared library for make_ext4fs

Added the missing shared library libpcre2.so for make_ext4fs
in the otatools.zip.
Standalone signing the images with otatools zip is not working
due to missing library for make_ext4fs.

Test: tools/releasetools/sign_target_files_apks
    -p <extracted ota-tools.zip folder>
    --extra_signapk_args=-f /etc/opt/cert_data.dat
    -v
    --replace_verity_private_key ~/build/target/product/security/verity
    --replace_verity_public_key ~/build/target/product/security/verity.x509.pem
    -k <key maping>
    <input target files zip>
    <output target files zip>

Change-Id: I4fc71a1c710f3075bb3bdb90d915538526004ee6

7 years agoAdd ART_USE_READ_BARRIER to product variables.
Hiroshi Yamauchi [Fri, 16 Dec 2016 00:00:41 +0000 (16:00 -0800)]
Add ART_USE_READ_BARRIER to product variables.

Bug: 33492005
Test: top-level build for angler.
Change-Id: I12aa2e18487c16423223f110b6dfb0d6dff1e7e8

7 years agoMerge "blockimgdiff.py: Add a test case for ReviseStashSize()."
Tao Bao [Mon, 19 Dec 2016 21:15:17 +0000 (21:15 +0000)]
Merge "blockimgdiff.py: Add a test case for ReviseStashSize()."

7 years agoblockimgdiff.py: Add a test case for ReviseStashSize().
Tao Bao [Mon, 19 Dec 2016 19:01:38 +0000 (11:01 -0800)]
blockimgdiff.py: Add a test case for ReviseStashSize().

Bug: 33687949
Test: python -m unittest test_blockimgdiff
Change-Id: I31e42648d2c0e60b941908c098ce5ccc967861ca

7 years agoRemove option for non-pic dex preopt.
Richard Uhler [Mon, 5 Dec 2016 11:47:02 +0000 (11:47 +0000)]
Remove option for non-pic dex preopt.

Test: make checkbuild, aosp_bullhead-userdebug boots.
Bug: 33192586

Change-Id: I4dd179fac33d777fb10cc85a99ff68c3b0d0e03d

7 years agoRevert "Revert "Merge changes from topic 'fsconfig-2'""
Elliott Hughes [Tue, 13 Dec 2016 23:37:07 +0000 (23:37 +0000)]
Revert "Revert "Merge changes from topic 'fsconfig-2'""

This reverts commit fad4b4b715be25e874829345a14219716fc8c157.

Incorporating the following fixes:

1.

fsconfig: fix fs_config_* build for discovered headers

When android_file system_config.h is picked up from the device
directory, neither TARGET_FS_CONFIG_GEN or TARGET_ANDROID_FILESYSTEM_CONFIG_H
are specified. Thus, the build is not generating the required fs_config_files
and fs_config_dirs.

Test: Ensure that make fs_config_files works, and produces the same output as before
      Build the system image and mount it as a loop back and dump the file system
      capabilities with getcap. Verify that output to the supplied
      android_file system_config.h

From the loopback of the system.img mount, from CWD system/bin:
$ getcap *
cnss-daemon = cap_net_bind_service+ep
hostapd = cap_net_admin,cap_net_raw+ep
imsdatadaemon = cap_net_bind_service+ep
ims_rtp_daemon = cap_net_bind_service+ep
logd = cap_setgid,cap_audit_control,cap_syslog+ep
mm-qcamera-daemon = cap_sys_nice+ep
pm-service = cap_net_bind_service+ep
run-as = cap_setgid,cap_setuid+ep
surfaceflinger = cap_sys_nice+ep
webview_zygote32 = cap_setgid,cap_setuid,cap_setpcap+ep
webview_zygote64 = cap_setgid,cap_setuid,cap_setpcap+ep

Compared to the android_filesystem_config.h:
    { 00700, AID_CAMERA,    AID_SHELL,     (1ULL << CAP_SYS_NICE), "system/bin/mm-qcamera-daemon" },
    { 00755, AID_SYSTEM,    AID_SYSTEM,    (1ULL << CAP_NET_BIND_SERVICE), "system/bin/pm-service" },
    { 00755, AID_SYSTEM,    AID_SYSTEM,    (1ULL << CAP_NET_BIND_SERVICE), "system/bin/imsdatadaemon" },
    { 00755, AID_SYSTEM,    AID_RADIO,     (1ULL << CAP_NET_BIND_SERVICE), "system/bin/ims_rtp_daemon" },
    { 00755, AID_SYSTEM,    AID_SYSTEM,    (1ULL << CAP_NET_BIND_SERVICE), "system/bin/cnss-daemon"},

2.

fsconfig: fix error message for duplicate AID

Fixes:
    raise ValueError('Duplicate aid value "%u" for %s' % value,
TypeError: %u format: a number is required, not str

and

    raise ValueError('Duplicate aid value "%s" for %s' % value,
TypeError: not enough arguments for format string

3.

fsconfig: add test for duplicate ranges

Add a test for duplicate range detection.

4.

fsconfig: skip AID_APP, AID_USER and all ranges

Do not output AID_APP, AID_USER and ranges. A range
is defined as ending with AID_ and ending in _START or
_END.

5.

fsconfig: test for skip AID_APP, AID_USER and all ranges

Test against AIDs that caused the bionic tests to fail.

Change-Id: I95569a9ccc83bd3231f8a6f395532cc2de316bd2
Signed-off-by: William Roberts <william.c.roberts@intel.com>
7 years agoMerge "Revert "Add build dependencies for split sepolicy files.""
Daniel Cashman [Sat, 17 Dec 2016 00:59:42 +0000 (00:59 +0000)]
Merge "Revert "Add build dependencies for split sepolicy files.""

7 years agoRevert "Add build dependencies for split sepolicy files."
Daniel Cashman [Sat, 17 Dec 2016 00:53:52 +0000 (00:53 +0000)]
Revert "Add build dependencies for split sepolicy files."

This reverts commit 086189145caa9b154d02aa3a888a8815bb60a0d5.

Change-Id: Idfd210d1f16eefcedda6154b0361611f53c5158f

7 years agoMerge "Add build dependencies for split sepolicy files."
Treehugger Robot [Sat, 17 Dec 2016 00:17:12 +0000 (00:17 +0000)]
Merge "Add build dependencies for split sepolicy files."

7 years agoMerge "releasetools: Fix the computation in ReviseStashSize()."
Tao Bao [Fri, 16 Dec 2016 23:57:40 +0000 (23:57 +0000)]
Merge "releasetools: Fix the computation in ReviseStashSize()."

7 years agoAdd build dependencies for split sepolicy files.
dcashman [Mon, 28 Nov 2016 15:19:35 +0000 (07:19 -0800)]
Add build dependencies for split sepolicy files.

Add build targets for split sepolicy files so they'll appear in the root dir for
on-device compilation.  nonplat_sepolicy will eventually be removed as it should
be provided by a different partition.  Also replace sepolicy.recovery with the
appropriate split components.

Bug: 31363362
Test: Policy builds on-device and boots.
Change-Id: I017dabe6940c3cd20de6c00bb5253274d5a9269b

7 years agoMerge "Add default NetworkRecommendation to core.mk"
Treehugger Robot [Fri, 16 Dec 2016 20:23:47 +0000 (20:23 +0000)]
Merge "Add default NetworkRecommendation to core.mk"

7 years agoreleasetools: Fix the computation in ReviseStashSize().
Tao Bao [Fri, 16 Dec 2016 19:13:55 +0000 (11:13 -0800)]
releasetools: Fix the computation in ReviseStashSize().

We compute the max stashed_blocks in ReviseStashSize(), prior to calling
WriteTransfers(), to avoid running out of space due to stashing.

There is a bug when computing the to-be-freed stashed blocks, where we
wrongly free the space _before_ executing the transfer command. This leads
to a script failure where the max stash size violates the max allowed
size in WriteTransfers().

Note that this bug doesn't affect already generated packages. It's only
an underestimate in ReviseStashSize(). The check in WriteTransfers() has
been correct to ensure the max stash size.

Bug: 33687949
Test: Successfully generated incremental OTA which failed previously.
Change-Id: I4f4f043c6f521fce81ca5286e6156f22d99bf7f7

7 years agoAdd default NetworkRecommendation to core.mk
Joe LaPenna [Thu, 8 Dec 2016 00:06:43 +0000 (16:06 -0800)]
Add default NetworkRecommendation to core.mk

This makes the apk avaiable on all devices and can be overridden on a
per-product basis using LOCAL_OVERRIDES_PACKAGES.

BUG: 33224286
Test: Verified with lunch aosp_bullhead-userdebug; make

Change-Id: Ib345193caa4431b6405294bd8759d93349335814

7 years agoMerge "aapt no longer needed for signing APKs"
Alex Klyubin [Fri, 16 Dec 2016 17:21:13 +0000 (17:21 +0000)]
Merge "aapt no longer needed for signing APKs"

7 years agoMerge "Remove unused get-package-min-sdk-version-int"
Alex Klyubin [Fri, 16 Dec 2016 17:20:23 +0000 (17:20 +0000)]
Merge "Remove unused get-package-min-sdk-version-int"

7 years agoMerge "Revert "Extends VNDK support in Make by setting LOCAL_USE_VNDK for modules...
Vijay Venkatraman [Fri, 16 Dec 2016 02:11:11 +0000 (02:11 +0000)]
Merge "Revert "Extends VNDK support in Make by setting LOCAL_USE_VNDK for modules going into vendor partition, except when they already have LOCAL_SDK_VERSION set.""

7 years agoRevert "Extends VNDK support in Make by setting LOCAL_USE_VNDK for modules going...
Vijay Venkatraman [Fri, 16 Dec 2016 02:08:18 +0000 (02:08 +0000)]
Revert "Extends VNDK support in Make by setting LOCAL_USE_VNDK for modules going into vendor partition, except when they already have LOCAL_SDK_VERSION set."

This reverts commit 2b6c204c73b97147ec00484904ee6f4bc003cb15.

Change-Id: I60a042f2ed0c767ec3a45c31ee1b99d43891efaa

7 years agoRemove unused get-package-min-sdk-version-int
Alex Klyubin [Thu, 15 Dec 2016 20:53:58 +0000 (12:53 -0800)]
Remove unused get-package-min-sdk-version-int

The only user of get-package-min-sdk-version-int was signapk. signapk
no longer needs to be provided with the APK's minSdkVersion though.

(cherry picked from commit e185da21ca07006c5ced90e1577c0682566428de)
Test: make clean && make

Change-Id: I6867a004aec0f3752cbbc99cc30e02ca5404b3c4

7 years agoaapt no longer needed for signing APKs
Alex Klyubin [Thu, 15 Dec 2016 20:51:25 +0000 (12:51 -0800)]
aapt no longer needed for signing APKs

signapk now auto-detects the APK's minSdkVersion, removing the need to
run aapt dump badging during APK signing.

Test: make clean && make
Change-Id: Ifc74292a9900443d053e437b50b540c8d0b33dbd

7 years agoMerge "Extends VNDK support in Make by setting LOCAL_USE_VNDK for modules going into...
Vijay Venkatraman [Thu, 15 Dec 2016 21:26:16 +0000 (21:26 +0000)]
Merge "Extends VNDK support in Make by setting LOCAL_USE_VNDK for modules going into vendor partition, except when they already have LOCAL_SDK_VERSION set."

7 years agoMerge "Split mac_permissions.xml into plat and non-plat components."
Treehugger Robot [Thu, 15 Dec 2016 20:13:28 +0000 (20:13 +0000)]
Merge "Split mac_permissions.xml into plat and non-plat components."

7 years agoSplit mac_permissions.xml into plat and non-plat components.
dcashman [Wed, 14 Dec 2016 21:48:48 +0000 (13:48 -0800)]
Split mac_permissions.xml into plat and non-plat components.

Bug: 31363362
Test: Bullhead and Sailfish both build and boot w/out new denials.
Change-Id: Id7e29ec09510741c4e925231401a6de81704fbf0

7 years agoMerge "Remove variables exported by Soong"
Treehugger Robot [Thu, 15 Dec 2016 07:19:03 +0000 (07:19 +0000)]
Merge "Remove variables exported by Soong"

7 years agoMerge "Don't clean generated java sources when AIDL list changes"
Treehugger Robot [Thu, 15 Dec 2016 07:01:12 +0000 (07:01 +0000)]
Merge "Don't clean generated java sources when AIDL list changes"

7 years agoDon't clean generated java sources when AIDL list changes
Dan Willemsen [Thu, 15 Dec 2016 03:53:54 +0000 (19:53 -0800)]
Don't clean generated java sources when AIDL list changes

AIDL doesn't generate into the intermediate src directory, we pass exact
aidl->java files to the compiler. So there is no need to clean the
intermediate src directory when the aidl file list changes for a module.

Test: None, just code inspection
Change-Id: I01feff7cc399ac5b88b83333a1ac86928d0a81e6

7 years agoMerge "Identify non-local EXPORT_C_INCLUDE_DIRS in soong_to_convert"
Dan Willemsen [Thu, 15 Dec 2016 00:30:03 +0000 (00:30 +0000)]
Merge "Identify non-local EXPORT_C_INCLUDE_DIRS in soong_to_convert"

7 years agoRemove variables exported by Soong
Dan Willemsen [Wed, 14 Dec 2016 23:54:01 +0000 (15:54 -0800)]
Remove variables exported by Soong

LLVM_RELEASE_VERSION is unused from envsetup.sh and during product
config loading.  LLVM_PREBUILTS_VERSION is still used in envsetup.sh, so
needs to stay.

Delay reading core/clang/config.mk until after Soong's variables are
read so that we don't need a copy of LLVM_PREBUILTS_BASE.

Test: Compare build-aosp_flounder.ninja
Change-Id: I85b5836a2230838c8f3766fb19c6108fc5c9ca50

7 years agoMerge "Faster auto-detection of APK's minSdkVersion"
Alex Klyubin [Wed, 14 Dec 2016 23:42:13 +0000 (23:42 +0000)]
Merge "Faster auto-detection of APK's minSdkVersion"

7 years agoFaster auto-detection of APK's minSdkVersion
Alex Klyubin [Tue, 13 Dec 2016 02:19:17 +0000 (18:19 -0800)]
Faster auto-detection of APK's minSdkVersion

Prior to this change, when signing APKs, the build system invoked
'aapt dump badging' on each APK, to detect the value to pass into
signapk as --min-sdk-version. Now that signapk uses the apksig
library, it can auto-detect that value on its own, thus avoiding the
need to invoke 'aapt dump badging' and thus speeding up the build
process.

The semantics of signapk's --min-sdk-version flag is changed by this
commit from having the default value of 0 to having the default value
of "auto-detect from APK".

P.S. The get-package-min-sdk-version-int is not removed from
core/definitions.mk in this commnit, because this function is used in
another project's .mk file and thus that .mk file needs to be modified
first.

Test: rm -Rf out/ && make
Change-Id: I0972fcf0abbde9cbf6794e6c05c743c77c8a78f9

7 years agoMerge "releasetools: Add prefix when dumping fingerprints."
Treehugger Robot [Wed, 14 Dec 2016 20:53:02 +0000 (20:53 +0000)]
Merge "releasetools: Add prefix when dumping fingerprints."

7 years agoreleasetools: Add prefix when dumping fingerprints.
Tao Bao [Wed, 14 Dec 2016 19:53:38 +0000 (11:53 -0800)]
releasetools: Add prefix when dumping fingerprints.

We used to dump "Source: <fingerprint>" in update logs. The "Source: "
prefix was unintentionally dropped out.

Test: Check the generated incremental BBOTA script.
Change-Id: I4de62333aa38e3fb09a76df0e769b62af48e0313

7 years agoMerge "Implement LOCAL_TEST_DATA to ship data with tests"
Treehugger Robot [Wed, 14 Dec 2016 03:52:31 +0000 (03:52 +0000)]
Merge "Implement LOCAL_TEST_DATA to ship data with tests"

7 years agoIdentify non-local EXPORT_C_INCLUDE_DIRS in soong_to_convert
Dan Willemsen [Wed, 14 Dec 2016 02:28:33 +0000 (18:28 -0800)]
Identify non-local EXPORT_C_INCLUDE_DIRS in soong_to_convert

Soong only supports exporting include directories under the current
directory. So bring non-local directories up as a potential problem in
soong_to_convert.txt

Test: m -j $OUT/soong_to_convert.txt, inspect
Change-Id: I7a15b92e10a1d8b8d3496c6f0529a0d0824f301e

7 years agoImplement LOCAL_TEST_DATA to ship data with tests
Dan Willemsen [Sat, 10 Dec 2016 00:05:09 +0000 (16:05 -0800)]
Implement LOCAL_TEST_DATA to ship data with tests

This can be used to ship source data as test artifacts next to native
tests. It works for both local builds and the test bundles using
package_modules.mk.

You just specify a file list relative to the local directory, and those
files will be copied next to the executable under
/data/nativetest*/<module>/...:

  LOCAL_MODULE := mytest
  LOCAL_TEST_DATA := data/file1 file2

  /data/nativetest/mytest/mytest
  /data/nativetest/mytest/data/file1
  /data/nativetest/mytest/file2

If the data is in another directory, you may also specify a different
prefix for the source files:

  LOCAL_TEST_DATA := external/skia:resources/f.xml

  /data/nativetest/skia_test/resources/f.xml

And there's a new convenience macro to find a list of files in this
format:

  LOCAL_TEST_DATA := $(call find-test-data-in-subdirs,external/skia,"*.xml",resources)

I'll expand this to native benchmarks and fuzz tests in a later change,
since they don't have their own module classes yet.

Bug: 30564705
Test: m -j minikin_tests; ls $OUT/data/nativetest*/minikin_tests
Test: m -j continuous_native_tests dist; zipinfo -1 out/dist/*continuous_native_tests*.zip
Change-Id: Ic76a7b62e7f567f259c4ab1510ee97d26600ba9a

7 years agoMerge "Sort and reorganize clear_vars.mk to make merges easier"
Dan Willemsen [Tue, 13 Dec 2016 22:22:58 +0000 (22:22 +0000)]
Merge "Sort and reorganize clear_vars.mk to make merges easier"

7 years agoSort and reorganize clear_vars.mk to make merges easier
Dan Willemsen [Tue, 13 Dec 2016 00:52:57 +0000 (16:52 -0800)]
Sort and reorganize clear_vars.mk to make merges easier

Moved common variables from other sections back to the common section,
then sort each section.

Test: build-aosp_angler.ninja is identical before/after
Change-Id: Ibb75935205aa80aaa392c358f8d0599ba624f90d
Merged-In: Ibb75935205aa80aaa392c358f8d0599ba624f90d

7 years agoMerge "Use legacy-test instead of core-junit"
Paul Duffin [Tue, 13 Dec 2016 09:34:17 +0000 (09:34 +0000)]
Merge "Use legacy-test instead of core-junit"

7 years agoMerge "Revert "Merge changes from topic 'fsconfig-2'""
Elliott Hughes [Tue, 13 Dec 2016 04:23:48 +0000 (04:23 +0000)]
Merge "Revert "Merge changes from topic 'fsconfig-2'""

7 years agoRevert "Merge changes from topic 'fsconfig-2'"
Elliott Hughes [Tue, 13 Dec 2016 01:28:44 +0000 (17:28 -0800)]
Revert "Merge changes from topic 'fsconfig-2'"

This reverts commit 1b2de51881e0ff76e2ad9d3d0405fc7185b0f86f, reversing
changes made to 77ea32f378651a3585e245eb964dab979d191310.

7 years agoMerge "Fix PDK dupbuild issues"
Treehugger Robot [Sat, 10 Dec 2016 07:14:11 +0000 (07:14 +0000)]
Merge "Fix PDK dupbuild issues"

7 years agoFix PDK dupbuild issues
Dan Willemsen [Sat, 10 Dec 2016 05:15:41 +0000 (21:15 -0800)]
Fix PDK dupbuild issues

The PDK uses pattern rules in order to install files from the PDK. When
those files already have build rules, the explicit rules override the
pattern rules, and everything works. But because Make (and Kati) doesn't
attempt to clean the file paths, if one of the rules has a redundant /,
we'll export two ninja rules, and ninja will error out with a dupbuild
error.

The PDK pattern rules are clean, but the explicit notice file creation
was not, it was always adding a double // in between NOTICE_FILES/src
and the module path.

Some modules were also setting a LOCAL_MODULE_PATH with a trailing /,
which is redundant, and also hits the above problem. Instead of fixing
all of the modules, just strip a trailing / from my_module_path.

Bug: 33451638
Test: Build with a PDK
Change-Id: Iff3e98fd191ea90626b9b89f179537e8a75f5ef2

7 years agoMerge changes from topic 'fsconfig-2'
Elliott Hughes [Sat, 10 Dec 2016 00:13:04 +0000 (00:13 +0000)]
Merge changes from topic 'fsconfig-2'

* changes:
  fs_config: add unit tests
  fs_config: drop fs_config_files/dirs PRODUCT_PACKAGES requirement
  fs_config: add group to build
  fs_config: introduce group generator
  fs_config: add passwd to build
  fs_config: introduce passwd generator
  fs_config: generate friendly in AID class
  fs_config: limit characters for AID_<name> sections
  fs_config: generate oem AID header file
  fs_config: android_id header generator
  fs_config: support parsing android_filesystem_config.h
  fs_config: modularize fs_config_generator

7 years agoMerge "Add more module targets for host, host-cross, host32, host64"
Treehugger Robot [Fri, 9 Dec 2016 22:54:59 +0000 (22:54 +0000)]
Merge "Add more module targets for host, host-cross, host32, host64"

7 years agoAdd more module targets for host, host-cross, host32, host64
Colin Cross [Fri, 9 Dec 2016 21:08:27 +0000 (13:08 -0800)]
Add more module targets for host, host-cross, host32, host64

Previously, clang-host would build the clang module for 32-bit and
64-bit for the host and for windows.  Make clang-host only compile
for host, and add clang-host-cross for windows.  Also add targets
that only build 32-bit or 64-bit: clang-host32, clang-host64,
clang-host-cross32, clang-host-cross64, clang-target32, and
clang-target64.

Test: inspect build.ninja
Change-Id: I7676f9497dfc852f2a0255dda8da06c88eec0db8

7 years agoUse legacy-test instead of core-junit
Paul Duffin [Fri, 9 Dec 2016 15:21:45 +0000 (15:21 +0000)]
Use legacy-test instead of core-junit

The core-junit library has been replaced with the legacy-test
library. The latter included everything that was in core-junit
plus some android.test classes that were in frameworks.

Bug: 30188076
Test: make checkbuild
Change-Id: Id2ba827705dbd2c27a1183e1153b03c11a8a4074

7 years agoMerge "Pass eng product varible to soong"
Colin Cross [Thu, 8 Dec 2016 23:42:16 +0000 (23:42 +0000)]
Merge "Pass eng product varible to soong"

7 years agoPass eng product varible to soong
Colin Cross [Thu, 8 Dec 2016 18:05:55 +0000 (10:05 -0800)]
Pass eng product varible to soong

Test: mmma -j frameworks/compile/libbcc
Change-Id: I79a903528948b728f730385a9de5e431617ff257

7 years agoMerge "Build: Skip JNI lib symlink in second-stage build"
Treehugger Robot [Wed, 7 Dec 2016 17:35:53 +0000 (17:35 +0000)]
Merge "Build: Skip JNI lib symlink in second-stage build"

7 years agoMerge "build_image: add support for specifying number of inodes"
Pat Tjin [Wed, 7 Dec 2016 07:07:25 +0000 (07:07 +0000)]
Merge "build_image: add support for specifying number of inodes"

7 years agobuild_image: add support for specifying number of inodes
Patrick Tjin [Thu, 20 Oct 2016 17:58:12 +0000 (10:58 -0700)]
build_image: add support for specifying number of inodes

Add support for specifying number of inodes when creating
system, vendor, oem partitions.  These are all read-only
and have no use for extra inodes.  Removing extra inodes
saves a lot of space.

Bug: 32246383
Change-Id: I13f1d4614b64a4abc752c42a1c65d3d151481c21
(cherry picked from commit b59eca358603a4174d4f7c9b70749e7cd0e6f3fe)

7 years agoMerge "custom_image: fix build break"
Treehugger Robot [Wed, 7 Dec 2016 03:20:12 +0000 (03:20 +0000)]
Merge "custom_image: fix build break"

7 years agoBuild: Skip JNI lib symlink in second-stage build
Andreas Gampe [Wed, 7 Dec 2016 01:56:29 +0000 (17:56 -0800)]
Build: Skip JNI lib symlink in second-stage build

When creating a sanitized image, skip creating JNI library
symlinks in the second round.

Bug: 33279120
Test: m && m SANITIZE_TARGET=address & ls -l $OUT/system/app/NfcNci/lib/*
Change-Id: Ib5eace9a49eb8b693603ba5cc59e392d575c44e3

7 years agoMerge "Disable Jack's NNode cache"
Benoit Lamarche [Tue, 6 Dec 2016 15:54:45 +0000 (15:54 +0000)]
Merge "Disable Jack's NNode cache"
am: 3c731228ba

Change-Id: Iea59e0c4cd0fa11277913d70a9de698565ae13fc

7 years agoMerge "Disable Jack's NNode cache"
Benoit Lamarche [Tue, 6 Dec 2016 15:47:34 +0000 (15:47 +0000)]
Merge "Disable Jack's NNode cache"

7 years agoDisable Jack's NNode cache
Benoit Lamarche [Tue, 6 Dec 2016 14:00:03 +0000 (15:00 +0100)]
Disable Jack's NNode cache

Test: N/A
Bug: 33367537
Change-Id: Idf7a518dfcbf5665917465b1ae2f0e09f4532688

7 years agoresolve merge conflicts of e04c941 to stage-aosp-master
Benoit Lamarche [Tue, 6 Dec 2016 10:32:41 +0000 (11:32 +0100)]
resolve merge conflicts of e04c941 to stage-aosp-master

Change-Id: I92ca7c5a10532220031dc799ad8900c9b20dc6fe

7 years agoMerge "Set platform min sdk for Jack to o-b1"
Benoit Lamarche [Tue, 6 Dec 2016 10:16:25 +0000 (10:16 +0000)]
Merge "Set platform min sdk for Jack to o-b1"

7 years agocustom_image: fix build break
bowgotsai [Tue, 6 Dec 2016 09:30:44 +0000 (17:30 +0800)]
custom_image: fix build break

Fix build break caused by
https://android-review.googlesource.com/#/c/305575/.

Bug: 33358681
Test: make custom_images

Change-Id: I1ba7cf61491c8593b20bdac58e984f1928697b8c

7 years agoMerge "Tell Soong about TARGET_COPY_OUT_VENDOR"
Dan Willemsen [Tue, 6 Dec 2016 03:32:05 +0000 (03:32 +0000)]
Merge "Tell Soong about TARGET_COPY_OUT_VENDOR"
am: 47b0322534

Change-Id: I55ac26073caf34173f6c051f87e10b651c23589b

7 years agoMerge "Tell Soong about TARGET_COPY_OUT_VENDOR"
Treehugger Robot [Tue, 6 Dec 2016 03:27:00 +0000 (03:27 +0000)]
Merge "Tell Soong about TARGET_COPY_OUT_VENDOR"