Request: make "setAppPackageName" optional, as the APK files are already provided [157783076]

Infeasible

Feature Request

[AOSP] assigned

Status Update

No update yet.

Description

lb...@gmail.com

created issue #1

May 30, 2020 11:07AM

Currently, if you wish to install an app using the Android framework, with the INSTALL_REASON_USER parameter, you also have to specify the package name.

That's even though you already send the APK files to be installed, and the OS already finds the package name anyway during install procedure (to check if the app is already installed, for example).

Please allow to avoid this parameter.

I'm talking about this:

https://developer.android.com/reference/android/content/pm/PackageInstaller.SessionParams#setAppPackageName(java.lang.String)

I ask it to be optional and still let ` params.setInstallReason(PackageManager.INSTALL_REASON_USER)` work well, as it gets the package name anyway.

Comments

am...@google.com <am...@google.com> Jun 1, 2020 03:21AM

Assigned to am...@google.com.

am...@google.com <am...@google.com> #2Jun 1, 2020 06:56AM

Saw this as well on Mokey running recent tot master build.

First benchmark only shows jit happening during the allocation phase

Benchmark in the middle of the suite shows it happening during measurement as well, before method tracing and allocation counting.

I verified during the run that the platform still considers the app compiled:

> adb shell dumpsys package dexopt | grep -A1 androidx.compose.ui.benchmark.test
  [androidx.compose.ui.benchmark.test]
    path: /data/app/~~E9h-5MtFh1GXjXCq3kQo8g==/androidx.compose.ui.benchmark.test-jOZKJEy6zBL_SpFBJqlisA==/base.apk
      arm64: [status=speed] [reason=cmdline] [primary-abi]
        [location is /data/app/~~E9h-5MtFh1GXjXCq3kQo8g==/androidx.compose.ui.benchmark.test-jOZKJEy6zBL_SpFBJqlisA==/oat/arm64/base.odex]

Disabling method tracing works around this issue, but we don't seem to see it in CI, on older platform versions (wembley 30, sargo 31): https://screenshot.googleplex.com/57EEmDd8MBNvdwE

My guess is that when turning off method tracing, something fails to be cleaned up, and we don't get to keep using the odex/art files from the speed compilation.

am...@google.com <am...@google.com> #3Jun 4, 2020 12:12PM

Status: Won't Fix (Infeasible)

Appears to be an ART regression starting in API 34

Bramble, API 33

google/bramble/bramble:13/TQ3A.230705.001/10216780:userdebug/dev-keys

112,994   ns    Trace    Method Trace    ModifiersBenchmark.baseReuse[Modifier_10x]
129,934   ns    Trace    Method Trace    ModifiersBenchmark.fullReuse[Modifier_10x]

Bramble, API 34

google/bramble/bramble:14/UP1A.231005.007/10754064:userdebug/dev-keys

 77,300   ns    Trace    Method Trace    ModifiersBenchmark.baseReuse[Modifier_10x]
395,151   ns    Trace    Method Trace    ModifiersBenchmark.fullReuse[Modifier_10x]

Will disable method tracing by default on API 34, and file a bug to ART.

lb...@gmail.com <lb...@gmail.com> #4Jun 4, 2020 01:03PM

Filed art bug here: b/303686344

Issue 157783076