[JDK20/21] Add support for JVMTI Local Variable functions

[babsingh]

Since JDK20, all JVMTI Local Variable functions are expected to return
JVMTI_ERROR_THREAD_NOT_SUSPENDED if the thread was not suspended and
was not the current thread.

Also, there is no need to halt and resume the thread for inspection
since these functions expect the thread to be suspended.

Related: #17711

[babsingh]

Timeline to illustrate change in JVMTI spec behaviour

JDK8 - https://docs.oracle.com/javase/8/docs/platform/jvmti/jvmti.html#local
JDK19 - https://download.java.net/java/early_access/jdk19/docs/specs/jvmti.html#local

• The below error code was added to all Local Variable functions in JDK20.
• JVMTI_ERROR_THREAD_NOT_SUSPENDED: Thread was not suspended and was not the current thread.

JDK20 - https://docs.oracle.com/en/java/javase/20/docs/specs/jvmti.html#local
JDK21 - https://download.java.net/java/early_access/jdk21/docs/specs/jvmti.html#local

[babsingh]

@gacholio Requesting your review.

[gacholio]

This does not address being suspended by java. In the past, suspension by java or JVMTI was indistiguishable (making the whole thing extremely unreliable). Please verify that the two are still identical for normal threads, which would require changes to this PR.

[gacholio]

Are suspend and resume even legal in java in the latest releases?

[babsingh]

Are suspend and resume even legal in java in the latest releases?

Since JDK20, java.lang.Thread.resume and java.lang.Thread.suspend throw UnsupportedOperationException for both normal and virtual threads.

This does not address being suspended by java. In the past, suspension by java or JVMTI was indistiguishable (making the whole thing extremely unreliable). Please verify that the two are still identical for normal threads, which would require changes to this PR.

In JDK20+, suspend and resume are only possible through JVMTI; the Java methods just return UnsupportedOperationException.

[gacholio]

jenkins test sanity zlinux jdk8,jdk21

[gacholio]

There's rather a lot of failures in the JDK21 testing.

[pshipton]

I only see one failure in testJitserverArguments_0 due to "Connection refused", which seems unrelated to this change.

[gacholio]

Yes, I'm misreading the output.