We are splitting android.hidl.memory into android.hidl.allocator and
android.hidl.memory to reflect the fact that we have two separate
interfaces which are served over different transports.
Bug: 35327976
Test: hidl_test, device boots with allocator
Merged-In: I36b1554d20f89ecd60a836c04f788ac83e0a1f5c
Change-Id: I36b1554d20f89ecd60a836c04f788ac83e0a1f5c
Adds the -fsanitize-blacklist option for CFI, using the built in
blacklist at external/compiler-rt/lib/cfi/cfi_blacklist.txt.
Also refactors the CFI cflags and ldflags into ../soong/cc/makevars.go
to ensure they're consistent across Soong and make projects.
Bug: 30227045
Test: ENABLE_CFI=true m -j40 builds and boots.
Test: The blacklist prevents runtime errors that otherwise occur.
Change-Id: I0c2801ed459a3b9adeb37daff3ca212564801259
The init.rc path selection uses the same partition_tag detection to
select an installation partition. So make sure that we always set
partition_tag even if we've been given an explicit module path.
Bug: 35314904
Test: set proprietary: true to Soong module with an init rc, verify
that was the only change to build-aosp_arm64.ninja
Change-Id: Ic6ffd7a5177959db96a34a24896b242ee5b09e41
Instead of having test tools search the OUT_DIR for the
testcases folders (of which there can be multiple different
targets), let's have these variables become part of the runtime
environment after a developer has run lunch.
Bug: 35322559
Test: . build/envsetup.sh ; lunch ; printenv | grep TESTCASES
Change-Id: Ic4addc2b9727ff717b2e595b9be224650e2f1b1d
Remove more installation directories. This gets called in between each
incremental build on our build servers.
Bug: 35267243
Test: (m -j; m -j installclean; m -j) ensure no large recompiles
Change-Id: Ied6bea5cad05c1f5e98b1bd1a30fb28ff09df23b
We check the needed stash size in ReviseStashSize(), and may not
generate a stash command if it would exceed the max allowed size. This
CL fixes a bug when skipping a stash operation: we shouldn't update the
'stashes' map if a stash command won't be generated.
Bug: 35313668
Test: Successfully generate the package that was failing due to the bug.
Change-Id: If0a3a5fadda9b4a4edad66a2a5826b5f978400ae
Fix an issue with incremental builds and the new common
testcase directories.
Bug: None
Test: Removed https://android-review.googlesource.com/#/c/234478/
then ran `m cts-platform-version-check` then reapplied the
change to verify the failure. Then applied this new change
to verify the problem was resolved.
Change-Id: I77529021e852af35571fb99537d87100010b47af
Enable iff WITH_DEXPREOPT_APP_IMAGE is true.
Test: Clean + build with the flag on and off. Tested Sailfish
booting.
Bug: 34927277
(cherry picked from commit e8ab2a71d7)
Change-Id: I63f98a85c69d99746b11bb6e85c23cf5e48910cf
Soong (or any other prebuilt) can provide a .gcnodir file using
LOCAL_PREBUILT_COVERAGE_ARCHIVE. Just like today, if one isn't provided
for static libraries, an empty one will be created for Make modules to
use.
Also fixes a problem I noticed including whole static libs having
coverage into static libs. It was trying to write to /WHOLE since
PRIVATE_INTERMEDIATES_DIR was not set.
Bug: 32749731
Test: See build/soong change for soong-side tests
Test: Enabled coverage of a soong static lib, then include it in a make
static lib with LOCAL_WHOLE_STATIC_LIBRARIES and ensure that the any
shared libraries using the make static lib get the embedded coverage
information.
Change-Id: I32762f099c9757074fec922ee4822f819c9ceaf5
When specifying multiple arguments to make, we were getting messages
about nothing to be done:
m -j USE_SOONG_UI=true checkbuild tests cts blueprint_tools
...
make: Nothing to be done for `tests'.
make: Nothing to be done for `cts'.
make: Nothing to be done for `blueprint_tools'.
Test: m -j USE_SOONG_UI=true blueprint_tools libc adb
Change-Id: Id448bf0e09736fcd92ba058aec7ce556f6f778a3
This CL disables CFI for 32-bit ARM processes, which is broken due to
a compiler error in the most recent version of clang.
Bug: 35157333
Test: ENABLE_CFI=true m -j40 does not enable CFI for 32-bit processes
Change-Id: I6a6e1d14c8365da1056b4cbccad11ef2f15c70b2
Instead of dumping all *TS testcases into the common testcase
directory directly, place them all in their own subdirectory.
Bug: 35153177
Test: `make cts -j` and checked the contents of both:
out/host/linux-x86/testcases/ &
out/host/linux-x86/cts/android-cts/testcases/
Change-Id: I22a768dba58d40be1e61885534fb813e7ff08c99
Aggregates the different *TS testcases into two common folders,
one under each the host and product hierarchies.
Bug: 34715822
Test: `make cts -j` and checked the contents of both:
out/host/linux-x86/testcases/ &
out/host/linux-x86/cts/android-cts/testcases/
Change-Id: I7d5201998ddcb011ecc687c272a0d0a36da248b7
We already support generating downgrade OTAs for non-A/B devices (with
mandatory data wipe), but we have missed the --downgrade flag in A/B OTA
path.
This CL factors out the function that writes the downgrade metadata, and
fixes the path for generating A/B OTAs.
Bug: 35094540
Test: Generate incrementals with --downgrade for A/B and non-A/B OTAs.
Change-Id: I30b9bf83e69e8aba3be666507681b555db6ab743
This reverts commit 01fc887a07.
A number of build breakages occurred which were not caused by presubmit checks.
Change-Id: I1351dba594c8a8b210416f75a67c68554d5b9372
Commit f1a13180db intended to remove the
verity blocks from care_map.txt, but it added new code without removing
the old one. This leads to a malformed care_map.txt and causes
update_verifier failure.
Bug: 34391662
Test: 'm -j dist' gives a TF.zip with 4-line META/care_map.txt (as
opposed to a 6-line file).
Change-Id: I7ff1aa525795c4b049af54c1755b0f0ea84f7e0e
For streaming OTAs, we will also need the info in the metadata entry
(META-INF/com/android/metadata). Compute and pack its offset/length
values into 'ota-streaming-property-files'.
Bug: 34986195
Test: Create an OTA package and check the offset/length values.
Change-Id: Id150700f2bc9bff02467cda9fe8927c8a374412a
Creating static archives is often a multi-command process due to
adding whole static libraires or hitting command line length limits.
If one of the intermediate commands fails, the output file may
already exist. Unlike make, ninja has no option to delete output
files on failed builds, instead assuming all build commands will
produce their output file atomically
(https://github.com/ninja-build/ninja/issues/1135).
Change the static library rules to generate to a temporary file
that is then atomically moved into place as the output file.
Test: m -j checkbuild tests cts
Change-Id: I4faf269f0c8e313c738154870a5aa0b4774a72bc
Steven Moreland