On Tue, 5 May 2026 07:34:28 GMT, David Holmes <[email protected]> wrote:

>> Since the suspender can read this concurrently I think we should use atomic 
>> store when writing per the style guide.
>
> Yes you are right. Thanks

I note that `_jni_active_critical` does not use `Atomic::store`. 

Also note we are piggy-backing on the state changes to the thread - if we see 
that the thread is handshake-safe then we must be guaranteed to see any update 
to the count that happened whilst the thread was handshake-unsafe. If we look 
at the count prior to that (claim_handshake) then we may see zero incorrectly 
but if we get passed `can_process_handshake` then we know examination of the 
count in `get_op` must return the correct value.

But yes I can still use `Atomic::store` per the style guide.

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

PR Review Comment: https://git.openjdk.org/jdk/pull/30936#discussion_r3199501935

Reply via email to