* changes:
merge_target_files: Checks VINTF metadata of resulting target files
Check VINTF compatibility at OTA generation time.
releasetools: add tests for check_target_files_vintf
Add a script to check VINTF compat of target files package.
Instead of relying on META/{system,vendor}_{manifest,matrix}.xml
and blindly copy compatibility.zip from target files package to
OTA package, do a static check on the input target files package
before generating the OTA package.
META/{system,vendor}_{manifest,matrix} does not contain ODM manifest
fragments, which is incorrect.
Also, the on-device check of compatibility.zip uses a old libvintf
binary on the device, which may not understand the incoming VINTF
metadata. This change removes the on-device check. Hence, it removes
the requirement of forwards compatibility of libvintf.
This behavior can be skipped with --skip-compatibility-check.
Test: build OTA package
Bug: 139300422
Bug: 131425279
Change-Id: I7fb93be9eb73f578fc05a182c6c9d1f073db2800
Instead of checking META/{system,vendor}_{manifest,matrix}.xml (which
is error-prone because ODM SKU-specific manifests are not checked),
this script read the target files package, remaps its directory
structure so that checkvintf understands it, and check VINTF
compatibility.
Also, put it in otatools.zip.
Test: run it on an extracted target files package
Bug: 131425279
Change-Id: I06036f9a8d7242d4bc11524028be40e780c508e8
Core variant of some hidl interfaces must be included in
mainline_system explicitly because they have dependency with external
modules.
Bug: 138284120
Test: build mainline_system_arm64
Change-Id: I1893bb807429dbcacac3db3dcab2a830456775ea
Support for rust modules built in Soong.
Bug: 136189233
Test: Test module builds.
Test: crosvm builds.
Change-Id: Idf2e4a4f760cbcee1d35fe6567ef8e12d4855094
This saves another 5-15% of makefile eval time (depending on
target/branch).
I'm not sure if deps-license is still used at all, but we'll only save
the information for it if it's actually going to run now.
Since the default module tag is "optional", we really don't need to save
a list of most modules in another variable. I don't see anything that
really needs to see the list of optional modules.
Test: treehugger
Change-Id: Iaf54e9932fbe8e61f10bd0c9336fc4fdefea27ab
It turns out that this single line was responsible for 28-44% of the
time we spent parsing makefiles. So instead of filtering across
$(ALL_MODULES), check to see if a variable that we always would set is
set.
Test: treehugger
Test: check kati-build eval time before/after
Change-Id: I946c2efd060181a4ae9edd3409c53d9b49a58285
This was showing up in the top-ten longest makefiles during the product
config and cleanspec kati runs sometimes, even though it's really never
used. So instead of running the sometimes-expensive find command, skip it
unless an AUX OS was specified.
At some point we should remove this entirely, but that's a larger change
that doesn't need to happen yet.
Test: build-aosp_arm.ninja is the same before/after
Change-Id: I0ea10193d2f07e1717224e9020a7defaf6191c05
Bug: none
Test: m -j FrameworksCorePackageManagerTests
confirm FrameworksCorePackageManagerTests.config doesn't have the
placeholder.
Change-Id: Ie16720fcf74490fddd1e282f8526e7161d6d10ce
avb_*_key_path are usually relative file paths. If these files cannot be
located from the current directory, try prepending OPTIONS.search_path.
Bug: 139836256
Test: python -m unittest test_common
Test: python -m unittest test_add_img_to_target_files
Change-Id: I020b257f458a1f5762691d0499feaf4027e37126
The avb_pubkey may not be present, if the apex was initially
unsigned or generated from a bundle. In this case, running
sign_apex to generate a signed apex binary would result in an error.
This fix checks for presense of avbpubkey before attempting the
deletion
BUG: 139994107
Change-Id: I3cb2e88a11ad8797e38ba5fb98c96a4ec4135fc8
No longer speed compile, we will rely on JIT and compilation during
boot to ensure performance.
Removed system ROM instrumentations since this will be done
dynamically.
Fixed renamed profilebootclasspath property.
Test: lunch taimen_nopreload-userdebug
Exempt-From-Owner-Approval: cherry-pick
Bug: 139883463
(cherry picked from commit 6f8fda4937)
Merged-In: I97a3a25c3094f84e130e02692b02832795ee6ef1
Change-Id: Id3e9b30a60d00ad3593f1587aa4b04ef7afe1fec
Add Tuner Hal 1.0 entry to configure file for building
Test: Manual
bug: 135709729
Change-Id: I6c8fa9382c64dc86f719fdfb17726fa694b6abb7
Merged-In: I6c8fa9382c64dc86f719fdfb17726fa694b6abb7
After the split up of prebuilt_internal.mk, the logic for running
host-init-verifier on init.rc files was misplaced. This change moves
it back to the right location.
Test: see that a mistake in rootdir/init.rc causes the build to fail
Change-Id: I60e503cc48639f4f11c47e6636f140672fde61b4
Include the misc_info.txt in "dist_files" so we can recreate a new
mixed super image (system and vendor images from different builds)
later in the test pipeline.
The misc_info.txt for products with BOARD_SUPER_IMAGE_IN_UPDATE_PACKAGE
set to true is removed from their update packages (essentially, reverts
aosp/976069) since the super_misc_info.txt created here can be used.
Bug: 139404128
Test: $ lunch aosp_x86-userdebug
$ m dist
$ diff $OUT_DIR/dist/super_misc_info.txt $OUT/misc_info.txt
Change-Id: Iad1ff981124a19933b0b575bc99941e0fc3813a7
Without this, we can end up packaging log files from previous CTS runs in
the zip file. If the names of those log files contain whitespace characters,
it can result in a build failure.
Change-Id: Id96d1915a03b6a715acfdc212c9b2d6f28045baa
