ForkJoinPool blocks thread instead of helping

ADDITIONAL SYSTEM INFORMATION :
All OS
Java 17.0.2 + 18 + 19 ea

A DESCRIPTION OF THE PROBLEM :
When an ForkJoinTask (e.g. Arrays.parallelSort) is invoked from a thread in a ForkJoinPool it blocks the thread instead of helping execute subtasks.


REGRESSION : Last worked in version 16

STEPS TO FOLLOW TO REPRODUCE THE PROBLEM :
Case 1) Try sorting an array inside a ForkJoinPool with parallelism 1:
ForkJoinPool fjp = new ForkJoinPool(1);
String[] objs = Stream.generate(() -> "").limit(10_000).toArray(String[]::new);
fjp.invoke(ForkJoinTask.adapt(() -> Arrays.parallelSort(objs)));

Case 2)
  Sort an array inside a ForkJoinPool with multiple threads

EXPECTED VERSUS ACTUAL BEHAVIOR :
EXPECTED -
Arrays gets sorted using all available threads
ACTUAL -
Case 1)
  Sorting hangs indefinitely

Case 2)
  Sorting does not use all threads (or deadlocks of multiple parallel sorts are run simultaneously)


---------- BEGIN SOURCE ----------
import static org.junit.jupiter.api.Assertions.*;

import java.util.Arrays;
import java.util.Comparator;
import java.util.Map;
import java.util.Map.Entry;
import java.util.concurrent.ConcurrentHashMap;
import java.util.concurrent.ForkJoinPool;
import java.util.concurrent.ForkJoinTask;
import java.util.concurrent.atomic.AtomicInteger;
import java.util.stream.IntStream;
import java.util.stream.Stream;

import org.junit.jupiter.api.Test;

class ForkJoinThreadCountTest {

private static final int TIMEOUT = 2000;

@Test
void test1Thread() {
ForkJoinPool fjp = new ForkJoinPool(1);
Thread killThread = new Thread(() -> {
try {
Thread.sleep(TIMEOUT);
fjp.shutdownNow();
System.out.println("Stopping thread pool");
} catch (InterruptedException e) {
// Expected
}
});
String[] objs = Stream.generate(() -> "").limit(10_000).toArray(String[]::new);

killThread.start();
fjp.invoke(ForkJoinTask.adapt(() -> Arrays.parallelSort(objs)));
killThread.interrupt();
}

@Test
void test2Threads() {
ForkJoinPool fjp = new ForkJoinPool(2);
Map<Thread, AtomicInteger> comparisonCounts = new ConcurrentHashMap<>();
Comparator<String> comp = (a,b) -> {
comparisonCounts.computeIfAbsent(Thread.currentThread(), t -> new AtomicInteger()).incrementAndGet();
return 0;
};
String[] objs = IntStream.range(0, 10_000_000).<String> mapToObj(i -> "").toArray(String[]::new);
// Expected approximately same workload on both threads
fjp.invoke(ForkJoinTask.adapt(() -> Arrays.parallelSort(objs, comp)));

for (Entry<Thread, AtomicInteger> entry : comparisonCounts.entrySet()) {
int comparisonsOnThread = entry.getValue().get();
if (comparisonsOnThread < 500_000) {
fail("Work not distributed evenly: " + comparisonCounts);
}
}
}

@Test
void testCommonForkJoinPool() throws Exception {
Thread testThread = new Thread( () -> {
int parallelism = ForkJoinPool.getCommonPoolParallelism();
IntStream.range(0, 2*parallelism).parallel().forEach(i -> Stream.generate(() -> "").limit(10_000).parallel().sorted().toArray() );
});
testThread.start();
testThread.join(TIMEOUT);
if (testThread.isAlive()) {
fail("Sorting did not terminate");
}
}
}
---------- END SOURCE ----------

FREQUENCY : always