RFR: 8341708: Optimize safepoint poll encoding with smaller poll data offset [v2]

[Aleksey Shipilev]

On Thu, 10 Oct 2024 12:49:28 GMT, Quan Anh Mai <qamai at openjdk.org> wrote:

>> I tried to avoid this kind of coupling between `Thread` and `JavaThread`, but I don't mind go that way as well. This is would also avoid the awkwardness of having `JavaThread::polling_data_*()` accessors all around our code calling essentially the superclass accessor otherwise. See new commit. I am re-running tests now.
>
> My thought is that it would make everything else see `poll_data` being a member of `JavaThread`, which seems more appropriate, while the fact that it is actually a member of `Thread` is merely an implementation detail.

Right. I modeled this after nmethod barriers, which also (?) technically belong to Java threads only. But there is no need to spend more time on this :)

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

PR Review Comment: https://git.openjdk.org/jdk/pull/21405#discussion_r1795413682