RFR: 8319447: Improve performance of delayed task handling

[Viktor Klang]

On Tue, 25 Feb 2025 22:41:07 GMT, Doug Lea <dl at openjdk.org> wrote:

>>> @sunmisc You are right that it would be nice if there were a way to efficiently use getAndSet here because a failed reference CAS hits slow paths that vary across GCs. But all of the ways I know to do this are much worse.
>> 
>> After a few days of benchmarks, I realized that you would be absolutely right. Although I thought if we separate the head (for deleting) and the tail (for inserting) there would be less contention. 
>> Even the fact that we can only modify the head (delete) in one thread without volatile does not help.
>> Perhaps I have made a mistake somewhere in the implementation
>
> @sunmisc Thanks for independently trying alternatives. We both had reasons to suspect that other mechanics might work out as well or better, but none seem to.

@DougLea Final review completed! I've added a few more comments and suggestions but after that I think this is ready to go. Thanks for the great work on this, Doug! /cc @AlanBateman

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

PR Comment: https://git.openjdk.org/jdk/pull/23702#issuecomment-2754373403