RFR: 8311981: Test gc/stringdedup/TestStringDeduplicationAgeThreshold.java#ZGenerational timed out [v2]

[Patricio Chilano Mateo]

On Mon, 14 Aug 2023 01:04:14 GMT, David Holmes <dholmes at openjdk.org> wrote:

>> nit typo: s/musn't/mustn't/
>> 
>> which is preferred according to urbandictionary.com.
>> 
>> I also don't understand the last sentence. More accurately, I understand the first
>> phrase, but I don't understand the need for the second phrase.
>
> Earlier when I mistakenly thought the no-arg version was this version with default args, I was concerned about its use in the `HandshakeState` destructor, but reasoned that it was safe as there couldn't be a real pending op at that time to cause contention on the lock - hence the comment in the main code. I left the comment just to add some information on when during a thread's lifetime we could hit this problem - to make it easier to reason about. I can remove it if it is causing confusion.

I see. The only thing which might be confusing is what exactly "termination" phase includes. Because when we are in Threads::remove() with a status of _thread_exiting we can still fail to acquire the lock here while trying to grab the Threads_lock.

-------------

PR Review Comment: https://git.openjdk.org/jdk/pull/15240#discussion_r1293701812