Explanation of the Thread-Local Handshakes

Question

As part of the putative JDK 10 this JEP 312: Thread-Local Handshakes was proposed.

I've tried to grasp its description, but I am still not confident that I got the idea properly.

Is it essentially an attempt to reanimate something similar to the deprecated Thread.destroy()?

How could this interface theoretically look like?

Answer 1

After researching JEP: 312 Thread-Local-Handshakes and reading the excellent comments to this question, here is a quick summary.

• It is a performance enhancement which cannot be used directly by developers:

  A handshake mechanism is proposed as an optimisation of the Hotspot safepoints mechanism.

  The former allows threads in a safepoint to continue executing immediately after performing a regulatory operation (the handshake callback).

  The latter forces all threads to block on a safepoint in order for critical operations to take place (such as YoungGen GC or the STW parts of a Full GC). The threads are released only after the critical operation has completed its execution.

• This JEP is not related to thread termination APIs such as Thread.destroy(). Since this subject has been raised, I'll add a few articles on thread termination in the references below.

OP Commentators:

JimmyB , Andy Turner , skomisa , Alexander Bollaert

References:

• Task Management history for JEP: 312 (JDK-8185640)
• OpenJDK discussion on JDK-8185640
• Safepoints in Hotspot JVM
• Java GC safepoints
• How to kill a Java Thread
• Shutting down threads cleanly