Cannot generate baseline profile for bundle [270587281]

Fixed

Bug

Status Update

No update yet.

Description

mh...@google.com

created issue #1

Feb 23, 2023 07:54PM

Component used: not sure what this means but BaselineProfileRule from macrobenchmark library

Version used: latest nightly snapshot from androidx.dev (1.2.0-SNAPSHOT)

Devices/Android versions reproduced on: Android P on go/acid

The macrobenchmark library fails to get a profile when the apk under test is actually a bundle with multiple splits. We at chrome use bundles with multiple splits (actually most of our code lives in splits rather than in the base module). When I try to generate a profile I get this in the output of the instrumentation:

java.lang.IllegalStateException: Expected no stderr from profman --dump-classes-and-methods --profile-file=/data/misc/profiles/ref/org.chromium.chrome/primary.prof --apk=/data/app/org.chromium.chrome-1CGJyRx56jh5RwfbZL4cQA==/base.apk
package:/data/app/org.chromium.chrome-1CGJyRx56jh5RwfbZL4cQA==/split_cablev2_authenticator.apk
package:/data/app/org.chromium.chrome-1CGJyRx56jh5RwfbZL4cQA==/split_chrome.apk, saw /storage/emulated/0/Android/media/org.chromium.chrome.startup.benchmark.tests/temporaryScript387305115831681219.tmp[2]: package:/data/app/org.chromium.chrome-1CGJyRx56jh5RwfbZL4cQA==/split_cablev2_authenticator.apk: inaccessible or not found
/storage/emulated/0/Android/media/org.chromium.chrome.startup.benchmark.tests/temporaryScript387305115831681219.tmp[4]: package:/data/app/org.chromium.chrome-1CGJyRx56jh5RwfbZL4cQA==/split_chrome.apk: inaccessible or not found

	at androidx.benchmark.Shell.executeScriptCaptureStdout$default(chromium-ChromeStartupBenchmark.apk-default-1:53)
	at androidx.benchmark.macro.BaselineProfilesKt.extractProfileRooted(chromium-ChromeStartupBenchmark.apk-default-1:100)
	at androidx.benchmark.macro.junit4.BaselineProfileRule.collectBaselineProfile(chromium-ChromeStartupBenchmark.apk-default-1:22)
	at org.chromium.chrome.browser.perf.StartupBenchmarkTest.testGenerateBaselineProfile(chromium-ChromeStartupBenchmark.apk-default-1:53)

The cause of this is this line here expects only a single path to return but for bundles, you get multiple ones (one for each split). eg:

generic_x86:/ # pm path org.chromium.chrome
package:/data/app/org.chromium.chrome-1CGJyRx56jh5RwfbZL4cQA==/base.apk
package:/data/app/org.chromium.chrome-1CGJyRx56jh5RwfbZL4cQA==/split_cablev2_authenticator.apk
package:/data/app/org.chromium.chrome-1CGJyRx56jh5RwfbZL4cQA==/split_chrome.apk

The rest of the code in BaselineProfiles.kt also expects only the one profile. But there are usually multiple (one for each split).

generic_x86:/ # ls /data/misc/profiles/ref/org.chromium.chrome/                                                                      
cablev2_authenticator.split.prof chrome.split.prof primary.prof

The documentation for baseline profiles mentions both APKs and Bundles so I assume bundles should be supported?

Comments

mh...@google.com <mh...@google.com> #2Feb 23, 2023 08:05PM

A couple of questions:

1. Have you saw crash in real device or only in simulators?
2. Do you use dynamic feature for language ID?

ra...@google.com <ra...@google.com> #3Feb 23, 2023 09:21PM

Tested on Android 12 Emulator with custom executor, but cannot repro this issue.

cc...@google.com <cc...@google.com> #4Feb 28, 2023 12:59AM

Accepted by cc...@google.com.

Second crash in the description is from a real device. Experienced it myself on two different Xiaomi phones, plus lots of crashes from users in the Google Play console.
Dynamic features are not used in the application.

As a wild guess, I have downgraded build tools from 31.0.0 to 30.0.3, compileSdk from 31 to 30, and moved all work with Language ID to the service in a separate process (just to be sure that crash can kill secondary process instead of main). This combination is in beta for 2 days by now and I don't see any SIGSEGV crashes.

ap...@google.com <ap...@google.com> #5Feb 28, 2023 09:07PM

Marked as fixed.

Hmm, I feel the crash might be something related to separate/secondary process.

I also changed compileSdk and targetSDK to 31 but still cannot repro this issue.

na...@google.com <na...@google.com> #6Mar 8, 2023 07:51PM

On the contrary, there was no separate process before, when crashes started.

In the new build (with the aforementioned changes) I can see SIGSEGV crash, but only one instead of dozens and it has a bit different backtrace:

signal 11 (SIGSEGV), code 1 (SEGV_MAPERR)
liblanguage_id_jni.so (offset 0x11e000)

backtrace:
  #00  pc 000000000003c7c0  /data/app/azagroup.reedy-mF7zTu2bv_ELlbFArwNgqA==/split_config.arm64_v8a.apk!lib/arm64-v8a/liblanguage_id_jni.so (offset 0x11e000)
  #00  pc 000000000003b960  /data/app/azagroup.reedy-mF7zTu2bv_ELlbFArwNgqA==/split_config.arm64_v8a.apk!lib/arm64-v8a/liblanguage_id_jni.so (offset 0x11e000)
  #00  pc 000000000003bb48  /data/app/azagroup.reedy-mF7zTu2bv_ELlbFArwNgqA==/split_config.arm64_v8a.apk!lib/arm64-v8a/liblanguage_id_jni.so (offset 0x11e000)
  #00  pc 000000000003bafc  /data/app/azagroup.reedy-mF7zTu2bv_ELlbFArwNgqA==/split_config.arm64_v8a.apk!lib/arm64-v8a/liblanguage_id_jni.so (offset 0x11e000)
  #00  pc 0000000000036c98  /data/app/azagroup.reedy-mF7zTu2bv_ELlbFArwNgqA==/split_config.arm64_v8a.apk!lib/arm64-v8a/liblanguage_id_jni.so (offset 0x11e000)
  #00  pc 0000000000032714  /data/app/azagroup.reedy-mF7zTu2bv_ELlbFArwNgqA==/split_config.arm64_v8a.apk!lib/arm64-v8a/liblanguage_id_jni.so (offset 0x11e000)
  #00  pc 0000000000031cac  /data/app/azagroup.reedy-mF7zTu2bv_ELlbFArwNgqA==/split_config.arm64_v8a.apk!lib/arm64-v8a/liblanguage_id_jni.so (offset 0x11e000)
  #00  pc 0000000000057438  /data/app/azagroup.reedy-mF7zTu2bv_ELlbFArwNgqA==/oat/arm64/base.odex (offset 0x57000)

Issue 270587281