Record LOCAL_ACONFIG_FILES when LOCAL_SOONG_MODULE_INFO_JSON is defined.
Bug: b/308625757
Test: out/soong/module-info-${TARGET_PRODUCT}.json is unchanged
Test: fixes some aconfig flags.
Change-Id: I7ca9d2723e38ec5588bc6ada8769320a0b63bcb5
This change adds isOptimizationEnabled method in FakeFeatureFlagsImpl.
This method can help test to decide whether the code using flag is
optimized.
Bug: 309406391
Test: atest aconfig.test aconfig.test.java
Change-Id: I7e7109ac06ab44f9b8dee56c76be9479f60a9148
If the kernel version isn't available to the build system and
PRODUCT_ENABLE_UFFD_GC isn't explicitly set, there will be a build
warning. To address the warning and make the behavior more explicit on
GSI builds, we need this change.
Bug: 321751629
Test: lunch aosp_x86_64-trunk_staging-userdebug && m dist
Change-Id: I5b73890235827929ea3bce5a07260d6c67a00a27
(Re-land of Ie882fccd864920289e48366e99a4ebd67e784d0d)
We no longer build GKIs from the platform tree.
These build commands were neither used nor maintained anymore, so clean
them up to reduce maintenance effort.
Keep the command line options as no-op, so existing scripts that still
specifies the deprecated options don't break.
Bug: 229701033
Test: presubmit
Change-Id: I0f7d05562dbc3eed29e902d6dc9a0f2e4083aaa6
Currently, EXT4 image building is done in a 2 pass fashion:
First pass: build ext4 image with small margin/headroom, check the
number of block used in the output image
Second pass: Build ext4 image again with an estimated block count from
first pass
The two pass implementation is done to keep image size at a
minimum. Recently some failures in first pass invocation is observed due
to image being undersized. Since we will reduce the image size in 2nd
pass, it's OK to increase the image size estimate during first pass. Add
a headroom 10% of total file size to fix build failures.
Test: cherry pick r.android.com/q/topic:%22aosp_cf_system_x86_64%22 and
m aosp_cf_system_x86_64
Bug: 321003625
Change-Id: I254310c67a08b98b05d2c858d4ab59a48112a07b
Window Extensions is not supported on TV. Adding it to TV GSI causes
some CTS failure.
Bug: 321898211
Test: make TV GSI, check artifacts
Test: atest CtsWindowManagerJetpackTestCases on TV GSI
Change-Id: Ice6602c7b06c2e09e557423426abaa94da55c836
Move aconfig storage file definition into its own crate, as flag read
lib will need it as well. The new crate is placed under
aconfig_storage_file dir.
Bug: b/321077378
Test: atest aconfig.test, Cargo test at repo root level
Change-Id: I91e8ada2c9bab479ed81cf03824042f3fa460be5
Looks like modules can now be missing "shared_libs" and "static_libs".
Use default values of empty sets where applicable, or check for the
existence of the keys before looking for them.
Test: python find_static_candidates.py --module audioserver
Bug: 322190716
Change-Id: I8be9bd2c1562975db57ffbf89d57ff2767f0b1f1
Previously, aconfig repo is the root directory of aconfig binary crate,
but it also hosts printflags crate inside, and there is no cargo support
for printflags binary crate. In addition, with more aconfig development,
more crates are being added to this repo. Thus this repo should be
configured as a Cargo workspace with multiple crates rather than a
single crate.
Note the top level Cargo.toml file specifies the crates this workspace
carries:
(1) aconfig_protos: the proto library crate that will be used by many other
crates such as aconfig binary crate and printflags binary crate
(2) aconfig: the aconfig binary crate
(3) printflags: the printflags binary crate
(1) aconfig_protos crate setup:
Inside aconfig_protos dir we set up the aconfig_protos crate, the
previously src/proto.rs is now aconfig_protos/src/lib.rs, the build.rs
is carried over to this crate.
(2) aconfig binary crate setup:
Notice its Cargo.toml file claims package dependency on aconfig_protos
crate. It no longer carries proto related module and build.rs file.
(3) printflags binary crate setup:
Similary, notice that in its Cargo.toml file, it claims package
dependency on aconfig_protos crate.
With this setup, we can Cargo build/test each crate individually when
inside a specific crate dir. But we can also run Cargo build/test at
repo root level, which will build/test all the crates in this workplace.
This is the structuring cl. The next cl is to move storage modules into
its own library crate. This storage file library crate will be used by
both aconfig binary crate as well as flag read library crate (to be
created as another new crate here).
Bug: b/321984352
Test: top and individual crate dir level Cargo build/test, m each
individual targets
Change-Id: I75833f4997f7ee554ff6c1557df9ac87f62b2732
Merged-In: I75833f4997f7ee554ff6c1557df9ac87f62b2732
Allow Cow version to be overwritten by device specific makefiles. This
change is needed so we can begin enableing virtual_ab_cow_version := 3
on cuttlefish
Test: m
Change-Id: Idd314dd24362a6e5d905db16ac260a24bbb7ae45
The current logic to determine if a build is pre or postsubmit is the
existence of the change_info file but this file also exists in
postsubmit, causing the build to break. Use the build ID as the
heuristic instead which should be more reliable.
Test: abtd run
Bug: 314171817
Change-Id: I9871eee71c0574583b528135dadd42e3f485a780
Previously, META/apex_info.pb contained only /system/apex apexes. Now,
it has all apexes from all possible partitions.
The main purpose of this file is to caculate the decompressed apex size
when applying OTA. Hence it should have all apexes, not just system
apexes.
Bug: 320228659
Test: m dist # check META/apex_info.pb
Change-Id: I3428dc502e4fe3336d1fc5ca941f1fbc332985cd
With Kernel caching enabled for files in Cog, we no longer need this workaround. This also makes clean builds after make clean simpler (no need to rerun source again after m clean).
Bug: b/289391270
TESTED: Ran ~5 full builds without workaround and none of them stalled
NOTE FOR REVIEWERS - original patch and result patch are not identical.
PLEASE REVIEW CAREFULLY.
Diffs between the patches:
# This function moves the reclient binaries into a directory that exists in a
> -# non-cog part of the overall filesystem. This is to workaround the problem
> -# described in b/289391270.
> -function _copy_reclient_binaries_from_cog() {
> - if [[ "${OUT_DIR}" == "" ]]; then
> - OUT_DIR="out"
> - fi
> - local RECLIENT_VERSION=`readlink prebuilts/remoteexecution-client/live`
> -
> - local NONCOG_RECLIENT_BIN_DIR_BASE="${OUT_DIR}/.reclient"
> - local NONCOG_RECLIENT_BIN_DIR="${NONCOG_RECLIENT_BIN_DIR_BASE}/${RECLIENT_VERSION}"
> -
> - # Create the non cog directory and setup live symlink.
> - mkdir -p ${NONCOG_RECLIENT_BIN_DIR}
> -
> - if [ `ls ${NONCOG_RECLIENT_BIN_DIR} | wc -l` -lt 8 ]; then
> - # Not all binaries exist, copy them from the Cog directory.
> - local TOP=$(gettop)
> - cp ${TOP}/prebuilts/remoteexecution-client/live/* ${NONCOG_RECLIENT_BIN_DIR}
> - fi
> -
> - ln -sfn ${RECLIENT_VERSION} ${NONCOG_RECLIENT_BIN_DIR_BASE}/live
> - export RBE_DIR="${NONCOG_RECLIENT_BIN_DIR_BASE}/live"
> -}
> -
> - _copy_reclient_binaries_from_cog
Original patch:
diff --git a/cogsetup.sh b/cogsetup.sh
old mode 100644
new mode 100644
--- a/cogsetup.sh
+++ b/cogsetup.sh
@@ -34,31 +34,6 @@
mkdir -p ${DEFAULT_OUTPUT_DIR}
ln -s ${DEFAULT_OUTPUT_DIR} `pwd`/out
}
-
-# This function moves the reclient binaries into a directory that exists in a
-# non-cog part of the overall filesystem. This is to workaround the problem
-# described in b/289391270.
-function _copy_reclient_binaries_from_cog() {
- if [[ "${OUT_DIR}" == "" ]]; then
- OUT_DIR="out"
- fi
- local RECLIENT_VERSION=`readlink prebuilts/remoteexecution-client/live`
-
- local NONCOG_RECLIENT_BIN_DIR_BASE="${OUT_DIR}/.reclient"
- local NONCOG_RECLIENT_BIN_DIR="${NONCOG_RECLIENT_BIN_DIR_BASE}/${RECLIENT_VERSION}"
-
- # Create the non cog directory and setup live symlink.
- mkdir -p ${NONCOG_RECLIENT_BIN_DIR}
-
- if [ `ls ${NONCOG_RECLIENT_BIN_DIR} | wc -l` -lt 8 ]; then
- # Not all binaries exist, copy them from the Cog directory.
- local TOP=$(gettop)
- cp ${TOP}/prebuilts/remoteexecution-clie
[[[Original patch trimmed due to size. Decoded string size: 1502. Decoded string SHA1: f5a824559440a15a8ccb76dbf414facf5d578da9.]]]
Result patch:
diff --git a/cogsetup.sh b/cogsetup.sh
index 3005d58..6439af0 100644
--- a/cogsetup.sh
+++ b/cogsetup.sh
@@ -35,31 +35,6 @@
ln -s ${DEFAULT_OUTPUT_DIR} `pwd`/out
}
-# This function moves the reclient binaries into a directory that exists in a
-# non-cog part of the overall filesystem. This is to workaround the problem
-# described in b/289391270.
-function _copy_reclient_binaries_from_cog() {
- if [[ "${OUT_DIR}" == "" ]]; then
- OUT_DIR="out"
- fi
- local RECLIENT_VERSION=`readlink prebuilts/remoteexecution-client/live`
-
- local NONCOG_RECLIENT_BIN_DIR_BASE="${OUT_DIR}/.reclient"
- local NONCOG_RECLIENT_BIN_DIR="${NONCOG_RECLIENT_BIN_DIR_BASE}/${RECLIENT_VERSION}"
-
- # Create the non cog directory and setup live symlink.
- mkdir -p ${NONCOG_RECLIENT_BIN_DIR}
-
- if [ `ls ${NONCOG_RECLIENT_BIN_DIR} | wc -l` -lt 8 ]; then
- # Not all binaries exist, copy them from the Cog directory.
- local TOP=$(gettop)
- cp ${TOP}/prebuilts/remoteexecution-client/live/* ${NONCOG_RECLIENT_BIN_DIR}
[[[Result patch trimmed due to size. Decoded string size: 1500. Decoded string SHA1: 75090baaf2a80305e8faea0e882e7edf257a2420.]]]
Change-Id: Id9963c3f1de7d3d4151a2e84fe52cf9d2f66ba10
RELEASE_BOARD_API_LEVEL_FROZEN must not be overriden by a board
config value. Use RELEASE_BOARD_API_LEVEL_FROZEN directly to build
configs.
Bug: 295269182
Test: TH
Change-Id: I60ad4cf1a9f377619fef98d6f9ef4d04d3395991
An old python zipfile hack prevented zipfile module from decoding 64 bit
sizes correctly, remove the legacy hack to fix.
Test: check_target_files_signatures -v 5GB_target_files.zip
Bug: 319367048
Change-Id: I376c7b68f549ddf88680280d604548d1849cdfe2
Based on the build system feedback, it is better to create storage file
individually per each aconfig command invocation in a makefile. Thus
updating the create-storage command to create one file at a time based
on passed in file selection.
Bug: b/321243743
Test: atest aconfig.test and manually invoke create-storage command
Change-Id: I53625c08e44d3ece4476835827c6b7d28ddd5a1b
Because `system.img` is a dependency of `fontchain_lint`, and
`system.img` is currently not generated inside `Soong`, it is not
possible to directly convert `fontchain_lint` to `Android.bp`.
Move it to build/core/task first.
Bug: 319050958
Test: m fontchain_lint
Change-Id: I2c318f6f51107cd80f0c70531e23f73569b0ec83
Revert submission 1433935-symlink_outputs
Reason for revert: symlink_outputs was added so bazel could run ninja files, but we abanoned that approach in roboleaf, and then roboleaf was cancelled entirely. Remove this feature so we're more compatible with upstream ninja / n2.
Reverted changes: /q/submissionid:1433935-symlink_outputs
Change-Id: I5a779695689148ed32037f7f82887c458a3587cd
