CancelledProducerConsumerLoops.java FAILED in JDK8 on linux-arm

[adamfarley]

Failing Test Info
Test Name: java/util/concurrent/BlockingQueue/CancelledProducerConsumerLoops.java
Test Duration: 135.5 seconds
Machine it fails on: test-docker-ubuntu2404-armv7-9 and test-docker-ubuntu2404-armv7-1
Machines it passes on (if intermittent): test-docker-ubuntu2404-armv7-10

Jenkins Test Job Info
Jenkins Test Job Name: Test_openjdk8_hs_sanity.openjdk_arm_linux_testList_1
Jenkins Test Job URL: https://ci.adoptium.net/job/Test_openjdk8_hs_sanity.openjdk_arm_linux_testList_1/16
TRSS link for the build (if available): https://trss.adoptium.net/allTestsInfo?buildId=696122c94a4969006d287b54&limit=1

Deep History Link from TRSS (if available)

https://trss.adoptium.net/deepHistory?testId=696123984a4969006d287d5e

** Rerun in Grinder Link (if available) or Steps to Reproduce**

Run JDK8 CancelledProducerConsumerLoops.java in grinder.

Java Version output (from console output)

13:02:25  openjdk version "1.8.0_482-beta"
13:02:25  OpenJDK Runtime Environment (Temurin)(build 1.8.0_482-beta-202601242115-b08)
13:02:25  OpenJDK Client VM (Temurin)(build 25.482-b08, mixed mode)

Relevant Console Output (showing exceptions/errors/crashes)

13:06:11  STDERR:
13:06:11  java.util.concurrent.ExecutionException: java.lang.NullPointerException
13:06:11  	at java.util.concurrent.FutureTask.report(FutureTask.java:122)
13:06:11  	at java.util.concurrent.FutureTask.get(FutureTask.java:192)
13:06:11  	at CancelledProducerConsumerLoops.oneRun(CancelledProducerConsumerLoops.java:101)
13:06:11  	at CancelledProducerConsumerLoops.oneTest(CancelledProducerConsumerLoops.java:125)
13:06:11  	at CancelledProducerConsumerLoops.main(CancelledProducerConsumerLoops.java:65)
13:06:11  	at sun.reflect.NativeMethodAccessorImpl.invoke0(Native Method)
13:06:11  	at sun.reflect.NativeMethodAccessorImpl.invoke(NativeMethodAccessorImpl.java:62)
13:06:11  	at sun.reflect.DelegatingMethodAccessorImpl.invoke(DelegatingMethodAccessorImpl.java:43)
13:06:11  	at java.lang.reflect.Method.invoke(Method.java:498)
13:06:11  	at com.sun.javatest.regtest.agent.MainActionHelper$AgentVMRunnable.run(MainActionHelper.java:298)
13:06:11  	at java.lang.Thread.run(Thread.java:750)
13:06:11  Caused by: java.lang.NullPointerException
13:06:11  	at java.util.concurrent.LinkedBlockingQueue.dequeue(LinkedBlockingQueue.java:213)
13:06:11  	at java.util.concurrent.LinkedBlockingQueue.poll(LinkedBlockingQueue.java:469)
13:06:11  	at CancelledProducerConsumerLoops$Consumer.call(CancelledProducerConsumerLoops.java:175)
13:06:11  	at CancelledProducerConsumerLoops$Consumer.call(CancelledProducerConsumerLoops.java:165)
13:06:11  	at java.util.concurrent.FutureTask.run(FutureTask.java:266)
13:06:11  	at java.util.concurrent.ThreadPoolExecutor.runWorker(ThreadPoolExecutor.java:1149)
13:06:11  	at java.util.concurrent.ThreadPoolExecutor$Worker.run(ThreadPoolExecutor.java:624)
13:06:11  	... 1 more

Attach or link to relevant artifacts

CancelledProducerConsumerLoops.jtr.xml

[adamfarley]

Another Dequeue NPE can be seen while running java/util/concurrent/ConcurrentQueues/ConcurrentQueueLoops.java on the same JDK version and platform.

13:56:59  STDERR:
13:56:59  java.lang.NullPointerException
13:56:59  	at java.util.concurrent.LinkedBlockingQueue.dequeue(LinkedBlockingQueue.java:216)
13:56:59  	at java.util.concurrent.LinkedBlockingQueue.poll(LinkedBlockingQueue.java:491)
13:56:59  	at ConcurrentQueueLoops$Stage.call(ConcurrentQueueLoops.java:123)
13:56:59  	at ConcurrentQueueLoops$Stage.call(ConcurrentQueueLoops.java:105)
13:56:59  	at java.util.concurrent.FutureTask.run(FutureTask.java:266)
13:56:59  	at java.util.concurrent.ThreadPoolExecutor.runWorker(ThreadPoolExecutor.java:1149)
13:56:59  	at java.util.concurrent.ThreadPoolExecutor$Worker.run(ThreadPoolExecutor.java:624)
13:56:59  	at java.lang.Thread.run(Thread.java:750)

However, this has only failed 10% of the time in a single grinder on test-docker-ubuntu2404-armv7-9. Pipeline jobs that ran this test on the same nodes that commonly fail CancelledProducerConsumerLoops.java appear to pass, though this could just be intermittent.

For comparison, CancelledProducerConsumerLoops.java appears to fail 50% of the time.