RFR: 8292591: Experimentally add back barrier-less Java thread transitions

[David Holmes]

On Thu, 1 Sep 2022 16:47:58 GMT, Robbin Ehn <rehn at openjdk.org> wrote:

> Please consider, only implemented on x64/aarch64 linux/windows.
> 
> On my box calling clock_gettime via JNI goes from 35ns to 28ns when enabled.
> 
> Passes t1-7 with option forced on, also passes t1-4 as is in this PR.

Can you explain the operation and placement of the system memory barriers please. It is not obvious how the elided membars are replaced with the new mechanism. "barrier-less" seems a bit of a misnomer as we're really replacing small barriers with a sledgehammer barrier in fewer places.

And does this really do enough to address the JNI overhead issue versus a true transition-less mechanism?

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

PR: https://git.openjdk.org/jdk/pull/10123