The /<partition>/etc/aconfig_flags.textproto files are no longer needed
(all clients have migrated to use /<partition>/etc/aconfig_flags.pb).
Stop generating them.
Also update how an empty file is generated in case there are no flags
for a given partition. Previously
echo "" > file
was used, which resulted in a file containing a newline, which caused
`printflags` to error out. Change the command to
echo -n > file
to create an empty file.
Bug: 302452801
Bug: 304278614
Test: m nothing
Change-Id: I2f7d23df4e888046651b4155b75c5006d0b5e1d7
Bug: 304814040
Test: CI, unit test,
b build build/make/tools/aconfig:aconfig.test.cpp
b test build/make/tools/aconfig:AconfigJavaHostTest
Change-Id: Ie9063a6c0bd08f96e0593bad8cb731d826b25fea
GKI kernel modules are required to be signed, but curretnly we strip all
16K kernel modules. Use SYSTEM_DLKM_MODULE_PATTERNS to determine which
modules are GKI, and skip stripping for these kernel modules.
Test: boot using 16K boot image
Bug: 303725887
Change-Id: I4cd71b39ceb75650d03744220ec4ddf4aeeda871
so that it's triggered even when APEX list shrinks.
(-d explain will print "command line changed for ...")
To repro:
- Pick one VAPEX with vintf fragment.
- Copy the fragment to /vendor/etc/vintf/manifest (DEVICE_MANIFEST_FILE)
- `m check-vintf-all` should fail
- Remove the APEX from PRODUCT_PACKAGES
- `m check-vintf-all` should pass
Bug: 304769672
Test: m check-vintf-all
(see above)
Change-Id: Iae9da91105faebfb8a5127fbb02a414b9ffdca10
Since there's only one, and it appears early, we can stop when we find
it rather than trawl through megabytes of localized strings we don't
need to see.
Bug: https://github.com/onetrueawk/awk/issues/199#issuecomment-1756885708
Test: treehugger
Change-Id: If4d7c63da7df76c23fca417a4b727189f79e3ce6
abort() seems friendlier, but in practice it puts a couple of irrelevant
libc.so frames on the top of the stack trace and clobbers all the
registers that might have helped remotely debug a sanitizer failure.
Let's switch to trapping directly in the sanitized code instead.
Bug: http://b/298741930
Test: treehugger
Change-Id: I4fca63cb42bfba16a934c1fa88b08697faf3be41
Revert submission 2761821-suppress-hiddenapi-check
Reason for revert: have some typo - break next build
Reverted changes: /q/submissionid:2761821-suppress-hiddenapi-check
Change-Id: I24f91b77f76cf0950d407aee3dedc30412213d00
Relanding https://r.android.com/2773148 in a different way, since
removing REL from ALL_CODENAMES prevents -next builds from booting
(see the comment for the code that relies on this behavior).
This reverts commit d637190deb.
Bug: http://b/303087088
Test: m ndk # with barbet-next-userdebug in goog/main
Change-Id: I8f750645f79c3cf48a85879c36e1549df47eee97
The codename of the current release is replaced by "REL" when the
build is configured as a release rather than a preview. For example,
PLATFORM_VERSION_CODENAME.UpsideDownCake will be "REL" rather than
UpsideDownCake in a -next target when the upcoming release is
UpsideDownCake. "REL" shouldn't really be treated as a codename
though. It's a placeholder to indicate that the build is a release and
so doesn't really have a codename. The list of all codenames ends up
in ro.build.version.all_codenames, and also ends up feeding the logic
for stub generation in soong, neither of which are places that should
include REL.
Bug: http://b/303087088
Test: m ndk # with barbet-next-userdebug in goog/main
Change-Id: I97972439e1f06c09f0ed59e7e2191c489a214e88
avbtool by default generates a random salt everytime, this makes builds
less reproducible. Use sha256 checksum of kernel image as the hex to
make the build reproducible.
Test: th
Bug: 293313353
Change-Id: I959b3dee77654098ab9fde475f11eaee8d40c790
The staging directory isn't used after the module packaging zip is
built. Delete the staging directory to reduce disk usage without
affecting incremental build performance.
Bug: 302573555
Change-Id: I377c60714f78760d0b6c2588935c12e10a9ed224
Test: treehugger
