>> This reinserts setting common pool parallelism inadvertently clobbered in JDK-8259800
> This looks okay to me, I just wonder if we should have a test to the value of getCommonPoolParallelism.
We once had a test, but now that Runtime.availableProcessors (the only means of checking) can return varying values, we can't include in tck tests. The error here wasn't caught because we didn't run the usual performance tests (that catch these problems immediately) because I was focused only on JDK-8259800. Sorry.