On Mon, 10 Jun 2024 13:48:08 GMT, Roger Riggs <rri...@openjdk.org> wrote:
>> I don't have a test or counter example and don't have a detailed >> understanding of the blocking needs of Hotspot. > > Another reason to retain the checking of GetThreadInterruptEvent is to be > belt and suspenders against the Java code changing and opening up a potential > error. At the moment, the reaper thread is encapsulated and not likely to get > an interrupt, but that might not always be the case and someone changing the > Java code might be unaware of the native code details. $.02 I did some research, and now I'm pretty confident that WaitForSingleObject is the right option: - the thread involved here is a daemon thread, i.e. it doesn't prevent JVM exit - the Unix version of `waitForProcessExit0` ignores interrupts and signals, it only ends when the wait succeeds or fails - I didn't observe any problems with exiting the JVM while waiting on a process handle - it's not clear to me what action we would take if the thread was cancelled; completing a CompletableFuture with an InterruptedException doesn't feel right. ------------- PR Review Comment: https://git.openjdk.org/jdk/pull/19586#discussion_r1633522031