[v3.4.2] AsyncLogger thread is stuck on 100% CPU on lmax queue polling

[shloim]

Describe the bug
Some low probability race condition causes lmax to loop indefinitely on Thread.yield

To Reproduce
This was found on a large production system with very low probability. Using log4j with the following setting may cause it:
-DLog4jContextSelector=org.apache.logging.log4j.core.async.AsyncLoggerContextSelector
-Dlog4j2.enable.threadlocals=true
-Dlog4j2.enable.direct.encoders=true
-XX:CompileCommand=dontinline,org.apache.logging.log4j.core.async.perftest.NoOpIdleStrategy::idle

Expected behavior
Thread should continue to process events

Desktop (please complete the following information):

• OS: Centos 8
• Version 5.19.0-0
• JVM Version OpenJDK 64-Bit Server VM GraalVM LIBGRAAL 20.3.3 (build 11.0.12+5-jvmci-20.3-b06)

Additional context
It would seem that the queue datastructure becomes corrupted and cas=uses the polling thread to loop infinitely with the following stacktrace:

"Log4j2-TF-1-AsyncLogger[AsyncContext@46f5f779]-1" #18 daemon prio=5 os_prio=0 cpu=4404755.82ms elapsed=4733.22s tid=0x00007fed7ade9800 nid=0xcd runnable  [0x00007fed7bffd000]
   java.lang.Thread.State: RUNNABLE
        at java.lang.Thread.yield([email protected]/Native Method)
        at java.util.concurrent.locks.AbstractQueuedSynchronizer.transferAfterCancelledWait([email protected]/AbstractQueuedSynchronizer.java:1752)
        at java.util.concurrent.locks.AbstractQueuedSynchronizer$ConditionObject.awaitNanos([email protected]/AbstractQueuedSynchronizer.java:2119)
        at com.lmax.disruptor.TimeoutBlockingWaitStrategy.waitFor(TimeoutBlockingWaitStrategy.java:38)
        at com.lmax.disruptor.ProcessingSequenceBarrier.waitFor(ProcessingSequenceBarrier.java:56)
        at com.lmax.disruptor.BatchEventProcessor.processEvents(BatchEventProcessor.java:159)
        at com.lmax.disruptor.BatchEventProcessor.run(BatchEventProcessor.java:125)
        at java.lang.Thread.run([email protected]/Thread.java:829)