ADFA-3863 Fix LayoutInflaterTest hang that wedged Jacoco/SonarQube CI#1371
Open
hal-eisen-adfa wants to merge 1 commit into
Open
ADFA-3863 Fix LayoutInflaterTest hang that wedged Jacoco/SonarQube CI#1371hal-eisen-adfa wants to merge 1 commit into
hal-eisen-adfa wants to merge 1 commit into
Conversation
The analyze.yml workflow was timing out at the 180-minute job limit because :xml-inflater:testV8DebugUnitTest hung waiting on a Tooling API child JVM that never exited. Add defense in depth so a future hang fails in minutes, not hours, and fix the underlying race. Safety nets: - ToolingApiTestLauncher: bound cancelCurrentBuild()/shutdown() RPCs with 30s timeouts and fall back to proc.destroyForcibly() if the child JVM does not exit. The child can no longer leak regardless of RPC behavior. - LayoutInflaterTest: @rule Timeout caps each test at 2 minutes. - build.gradle.kts: project-wide Test task timeout of 10 minutes as a final backstop. - analyze.yml: per-step timeout of 60 minutes on "Build and analyze" so a wedge cannot burn the full 180-minute job budget. Root cause: - XmlInflaterTest.initIfNeeded used a non-atomic get/set pair on AtomicBoolean. Replace with compareAndSet, and reset the flag on failure so a failed init does not silently poison subsequent tests.
Contributor
📝 WalkthroughRelease NotesCI Reliability Improvements
Bug Fix
Risk Notes
WalkthroughThe PR hardens the CI pipeline against indefinite test hangs by introducing timeout protection at four independent levels: workflow step timeout, Gradle task timeout, test launcher cleanup timeout, and individual test method timeout. Additionally, XmlInflaterTest initialization is refactored to use atomic compare-and-set semantics, allowing retries when setup fails instead of permanently blocking. ChangesLayered Test Timeout Protection
Estimated code review effort🎯 3 (Moderate) | ⏱️ ~20 minutes Suggested reviewers
Poem
🚥 Pre-merge checks | ✅ 4 | ❌ 1❌ Failed checks (1 warning)
✅ Passed checks (4 passed)
✏️ Tip: You can configure your own custom pre-merge checks in the settings. ✨ Finishing Touches📝 Generate docstrings
🧪 Generate unit tests (beta)
Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out. Comment |
![coderabbitai[bot]](https://avatars.githubusercontent.com/in/347564?s=60&v=4){kind=small}
Contributor
There was a problem hiding this comment.
Actionable comments posted: 1
🧹 Nitpick comments (1)
.github/workflows/analyze.yml (1)
95-100: ⚡ Quick winApply the same backstop to the earlier Gradle step.
Build and analyzeis now bounded, butAssemble V8 Debugis still an unbounded./gradlewinvocation. A hang there would still sit until the 180-minute job timeout, so the workflow only has partial protection right now.Suggested change
- name: Assemble V8 Debug + timeout-minutes: 60 run: | echo "gradle_time_start=$(date +%s)" >> $GITHUB_ENV flox activate -d flox/base -- ./gradlew :app:assembleV8Debug --no-daemon echo "gradle_time_end=$(date +%s)" >> $GITHUB_ENV🤖 Prompt for AI Agents
Verify each finding against current code. Fix only still-valid issues, skip the rest with a brief reason, keep changes minimal, and validate. In @.github/workflows/analyze.yml around lines 95 - 100, The earlier unbounded Gradle step "Assemble V8 Debug" needs the same timeout and env backstop as the "Build and analyze" step; update the "Assemble V8 Debug" job/step to include timeout-minutes: 60 and the GRADLE_OPTS and SONAR_TOKEN env entries (or at least GRADLE_OPTS="-Xmx10g -XX:MaxMetaspaceSize=512m") and keep the existing run command that invokes ./gradlew (the Assemble V8 Debug step) so the Gradle invocation is bounded and uses the same memory/metaspace settings as the Build and analyze step.
🤖 Prompt for all review comments with AI agents
Verify each finding against current code. Fix only still-valid issues, skip the
rest with a brief reason, keep changes minimal, and validate.
Inline comments:
In
`@xml-inflater/src/test/java/com/itsaky/androidide/inflater/XmlInflaterTest.kt`:
- Around line 40-44: The current init.compareAndSet(false, true) lets
non-winning callers return immediately and use the inflater before setup
finishes; change the init logic so losers block until initialization completes
(success or failure) instead of returning. Replace the AtomicBoolean init with a
state machine or synchronization primitive (e.g., an enum/AtomicInteger states
like UNINITIALIZED/INITIALIZING/READY/FAILED plus wait/notify or a
CountDownLatch) and update the initialization block and all callers to: the
winning thread performs setup and sets the terminal state (READY or FAILED),
then notifies/writes the latch, while non-winning threads wait for that terminal
state and then proceed or throw based on success/failure; ensure you update the
same init usage shown in XmlInflaterTest.kt (the init variable and the init
compareAndSet block spanning the original lines and the related 46-55 section).
---
Nitpick comments:
In @.github/workflows/analyze.yml:
- Around line 95-100: The earlier unbounded Gradle step "Assemble V8 Debug"
needs the same timeout and env backstop as the "Build and analyze" step; update
the "Assemble V8 Debug" job/step to include timeout-minutes: 60 and the
GRADLE_OPTS and SONAR_TOKEN env entries (or at least GRADLE_OPTS="-Xmx10g
-XX:MaxMetaspaceSize=512m") and keep the existing run command that invokes
./gradlew (the Assemble V8 Debug step) so the Gradle invocation is bounded and
uses the same memory/metaspace settings as the Build and analyze step.
🪄 Autofix (Beta)
Fix all unresolved CodeRabbit comments on this PR:
- Push a commit to this branch (recommended)
- Create a new PR with the fixes
ℹ️ Review info
⚙️ Run configuration
Configuration used: Organization UI
Review profile: CHILL
Plan: Pro
Run ID: c366db36-381c-4eff-b6c9-e2a814a48bc9
📒 Files selected for processing (5)
.github/workflows/analyze.ymlbuild.gradle.ktstesting/tooling/src/main/java/com/itsaky/androidide/testing/tooling/ToolingApiTestLauncher.ktxml-inflater/src/test/java/com/itsaky/androidide/inflater/LayoutInflaterTest.ktxml-inflater/src/test/java/com/itsaky/androidide/inflater/XmlInflaterTest.kt
Comment on lines
+40
to
44
| // Atomic claim of the init slot. If another caller already won the race, | ||
| // they are responsible for completing setup; we just return. | ||
| if (!init.compareAndSet(false, true)) { | ||
| return | ||
| } |
Contributor
There was a problem hiding this comment.
Non-winning callers can continue before initialization finishes.
compareAndSet prevents duplicate init, but losers return immediately and may use the inflater before setup completes. This creates a race/flaky path when tests run in parallel. Use a wait/notify (or latch/state-machine) so non-winning callers block until init reaches success/failure.
Suggested fix (state machine + wait/notify)
object XmlInflaterTest {
- private val init = AtomicBoolean(false)
+ private enum class InitState { UNINITIALIZED, INITIALIZING, INITIALIZED }
+ private val initLock = Object()
+ `@Volatile` private var initState = InitState.UNINITIALIZED
internal val activity by lazy { Robolectric.buildActivity(AppCompatActivity::class.java).get() }
fun initIfNeeded() {
- // Atomic claim of the init slot. If another caller already won the race,
- // they are responsible for completing setup; we just return.
- if (!init.compareAndSet(false, true)) {
- return
- }
+ synchronized(initLock) {
+ while (initState == InitState.INITIALIZING) initLock.wait()
+ if (initState == InitState.INITIALIZED) return
+ initState = InitState.INITIALIZING
+ }
try {
ToolingApiTestLauncher.launchServer {
assertThat(result is InitializeResult.Success).isTrue()
runBlocking { IProjectManager.getInstance().setup(gradleBuild.get()) }
}
+ synchronized(initLock) {
+ initState = InitState.INITIALIZED
+ initLock.notifyAll()
+ }
} catch (error: Throwable) {
- // Roll back so subsequent tests can retry rather than silently skipping setup.
- init.set(false)
+ synchronized(initLock) {
+ initState = InitState.UNINITIALIZED
+ initLock.notifyAll()
+ }
throw error
}
}
}Also applies to: 46-55
🤖 Prompt for AI Agents
Verify each finding against current code. Fix only still-valid issues, skip the
rest with a brief reason, keep changes minimal, and validate.
In `@xml-inflater/src/test/java/com/itsaky/androidide/inflater/XmlInflaterTest.kt`
around lines 40 - 44, The current init.compareAndSet(false, true) lets
non-winning callers return immediately and use the inflater before setup
finishes; change the init logic so losers block until initialization completes
(success or failure) instead of returning. Replace the AtomicBoolean init with a
state machine or synchronization primitive (e.g., an enum/AtomicInteger states
like UNINITIALIZED/INITIALIZING/READY/FAILED plus wait/notify or a
CountDownLatch) and update the initialization block and all callers to: the
winning thread performs setup and sets the terminal state (READY or FAILED),
then notifies/writes the latch, while non-winning threads wait for that terminal
state and then proceed or throw based on success/failure; ensure you update the
same init usage shown in XmlInflaterTest.kt (the init variable and the init
compareAndSet block spanning the original lines and the related 46-55 section).
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
The analyze.yml workflow was timing out at the 180-minute job limit because :xml-inflater:testV8DebugUnitTest hung waiting on a Tooling API child JVM that never exited. Add defense in depth so a future hang fails in minutes, not hours, and fix the underlying race.
Safety nets:
Root cause: