| Age | Commit message (Collapse) | Author |
|---|
|
Change-Id: Ibaead8e08cea509610f644779a3e09de6fc44634
Flag: android.app.background_install_control_callback_api
Bug: 374120984
Test: atest
|
|
When adding a dependencies, if the variants don't match, but the
dependency only has 1 variant anyways, soong will always use that
variant. This makes it hard to add new variants to soong, because the
1-variant fallback stops being used and you start getting missing
variant errors. Make changes to bp files such that all dependencies
correctly specify the variant to use.
Bug: 372091092
Flag: EXEMPT refactor
Test: m nothing
Change-Id: I305d9a549fc132c3d14f56eae1abbac1b9defd10
|
|
Test: atest BackgroundInstallControlServiceHostTest BackgroundInstallControlServiceTest BinaryTransparencyHostTest
Bug: 296060433
API-Coverage-Bug: 319762342
Change-Id: I88c0d664ebe4b26e3f5bb131c9468c8876a286d4
|
|
This reverts commit 9ba42db72ded7711f9625fe0165a8dfbeffa2aad.
Reason for revert: Seems suspicious for many new system crashes/reboots, b/319579008, b/319529606, b/319567367 and others
Change-Id: Ia4d7770e9f82e52b3f450333b4a7f525cb0c9dc9
|
|
and added permission restriction to service APIs.
Deferring push mechanism for SystemAPI until new permissions are added.
OWNERS file will be updated separately in AOSP.
Resubmit after reversion of b/313009440, added permission provisioning in
BinaryTransparencyHostTest
Bug: 296060433
Test: atest BackgroundInstallControlServiceHostTest
BackgroundInstallControlServiceTest
BackgroundInstallControlCallbackHelperTest
BinaryTransparencyHostTest
GtsBackgroundInstallControlManagerTestCases
Change-Id: Ie7ec065a606b892990bcb74ad21c6d8fed3b285d
|
|
Bug: 255714762
Test: presubmit
Change-Id: I7e29732e1e77e37cfb491521d14d2f3eb07deb36
|
|
I'm removing ApkVerityTestApp soon, so switch the placeholder app here
to another (instead of maintaining our own copy) from
frameworks/base/tests/FeatureSplit/.
Bug: 277916185
Test: atest BinaryTransparencyHostTest
Change-Id: Iac0a9bc52221f626fe0eeeadcfebc3ad96d356a3
|
|
The test collects expected APEX package names from listing /apex by a
shell command. The shell command also tries to remove the noise
(normally an APEX has two entires, see below).
$ ls -d /apex/*/
/apex/com.android.adbd/
/apex/com.android.adbd@340815002/
...
/apex/com.google.mainline.primary.libs@340716000/
/apex/sharedlibs/
The test used to deduplicate by filtering out '@'. But apparently an
(DLCA) APEX only has one entry with '@'.
With this change, the shell command removes '@\d+' then deduplicate the
strings.
Bug: 288551133
Test: BinaryTransparencyHostTest#testCollectAllApexInfo
Change-Id: I4bc3e8226dd7790aa276ac5dfad7371c60046384
|
|
@Presubmit: since it's now running there
@LargeTest: the the test can easily run more than a MediumTest's budget
Bug: 264296226
Test: atest BinaryTransparencyHostTest
Change-Id: I0a77893baa18c9bfed1bbe65d91e7d66c5b444ba
|
|
measurePackage is changed into measureApk, which would hint that the
method deals with a file instead of a "package" in the package manager's
term.
collectAppInfo is added to measure an app package, which may include
split.
Test
* Add a test that leverages debug.transparency.bg-install-apps of BICs.
* Copy BaseInstallMultiple from another test in frameworks/base to
support split install.
Misc
* Delete getApexInfo and corresponding unit tests, since it's now
coverred through collectAllApexInfo in integration test. This also
makes it easy for refactoring without having to keeping the Bundle
works as the return type.
Bug: 264296226
Test: atest BinaryTransparencyServiceTest BinaryTransparencyHostTest
Change-Id: Iaa4118dfa8605acda313dbcc2466ae96a60b4721
|
|
* Run uninstallRebootlessApex (which reboots the device, thus slow) only
necessarily in test cases.
* Use EasterEgg from the device, not a local build. Surprisingly this
app is signed with the platform key, so a different build can have
different key thus may fail to install.
* Increase waiting time to 30 seconds. For some reason that scheduling
may not happen very soon, even on real device.
* Always cancel the pending job before starting a test. For example,
this can cancel a job scheduled at boot, so that we can correctly
verify that a package install actually triggers a job scheduling.
* Add a simply check to uniqueness of expected APEX list.
Bug: 265244016
Test: atest BinaryTransparencyServiceTest BinaryTransparencyHostTest
Change-Id: I4b7bf232be2ee74b6c0129bc9938c680f1835619
|
|
... of rebootless APEX and preload update installation.
Changes made to the actual service:
* DO_BINARY_MEASUREMENTS_JOB_ID is now a constant, like most of jobs in
the framework. This makes it easy to use from the host.
* Replace the local state of sScheduled with a query to JobScheduler,
which is the source of truth. I've messed up the state during test
and manual interaction, and this is supposed to make it more
resilient.
Bug: 265244016
Test: atest BinaryTransparencyHostTest
Test: atest BinaryTransparencyServiceTest
Change-Id: I17872bbded0b5c5c44e6fbd1c669f1eb00f3333e
|
|
A new host-side test "BinaryTransparencyHostTest" is added to drive the
actual test on device. The host can be used to set up the testing
environment (e.g. update a preload, or later install MBA), then ask the
helper app "BinaryTransparencyTestApp" to verify returned value of
private APIs in the main service.
Since the APIs are private (@hide, because we're not ready to make
anything public yet), the (bundled) helper app needs to be compiled with
`platform_apis: true`.
Plus, since the service is not accessible to apps by sepolicy, the test
has to run with SELinux disabled. An alternative is to add a
userdebug_or_eng rule in sepolicy, but it's better to avoid adding rule
for one specific test, and isolate the setup here (in AndroidTest.xml).
The test is not yet added to TEST_MAPPING/presubmit. Once it runs
successfully for a few runs in postsubmit, we should graduate it to
presubmit.
Test: atest BinaryTransparencyHostTest
Test: atest BinaryTransparencyServiceTest
Bug: 265244016
Change-Id: Ibd87f4cf6a2ae989ddfa8eaf9494cff5d34005ed
|