Add JVMTI_EVENT_MONITOR_WAIT/ED event trigger for virtual threads

[fengxue-IS]

Fixes: #21400

[fengxue-IS]

@babsingh please take a look

[fengxue-IS]

local testing of the failed test case passed on amac
personal build in progress

[babsingh]

The top commit in https://github.com/babsingh/openj9/commits/monitorWaitHook has my code review; some of which @keithc-ca has already mentioned above.

Let's not add fixes 21400 in the commit message since the test would need to be excluded before the issue is closed.

#21402 should also pass with these changes, but it still fails.

[fengxue-IS]

Let's not add fixes 21400 in the commit message since the test would need to be excluded before the issue is closed.

#21402 should also pass with these changes, but it still fails.

#21402 is still failing because the hook is suppressed by jvmtiHook.c:3846:shouldPostEvent(). I will look into where the the flag for suppression finishes and if we need to re-order the hook trigger point.

[fengxue-IS]

I think I've found the root cause here, the call to enter/exitVThreadTransitionCritical in preparePinnedVirtualThreadForMount/Unmount isn't accompanied by the matching call to VM_VMHelpers::virtualThreadHideFrames which cause some jvmti events to be suppressed during hook dispatch.

In order to fix this issue, we must ensure that all event hooks which should be posted for vthreads must occur before or after the critical region. so that the stack/thread data is in a walkable state.

I will re-work this PR to correctly handle the framehide and critical transition region.

[babsingh]

@fengxue-IS There is a merge conflict, which needs to be resolved.

[babsingh]

Changes LGTM. Passing to @keithc-ca for review/merge.

[fengxue-IS]

@keithc-ca are you okay with this PR or is there anything you would suggest changing?

[keithc-ca]

Please rebase to avoid implicit merges with other changes (e.g. #21611).

[keithc-ca]

Jenkins test sanity zlinux jdk24

[keithc-ca]

I was awaiting the results from https://openj9-jenkins.osuosl.org/job/Grinder/4266 which failed.

[babsingh]

I was awaiting the results from https://openj9-jenkins.osuosl.org/job/Grinder/4266 which failed.

This is a separate issue unrelated to this PR. The test is RI specific; it will need to be updated for OpenJ9. @fengxue-IS has this task in his backlog; he will be creating extension repo PRs to fix it.

09:55:57  MonitorWaited event:
09:55:57  	thread: 0x13e6f40, object: 0x13e6f08, timed_out: false
09:55:57  Thread: 0x13e6f40, name: Debuggee Thread, state(5):  ALIVE RUNNABLE, attrs: virtual daemon
09:55:57  JVMTI Stack Trace for thread Debuggee Thread: frame count: 7
09:55:57   0: java/lang/Object: wait(JI)V
09:55:57   1: java/lang/Object: wait(J)V
09:55:57   2: monitorwaited01Task: run()V
09:55:57   3: java/lang/Thread: runWith(Ljava/lang/Object;Ljava/lang/Runnable;)V
09:55:57   4: java/lang/VirtualThread: run(Ljava/lang/Runnable;)V
09:55:57   5: java/lang/VirtualThread$VThreadContinuation$1: run()V
09:55:57   6: jdk/internal/vm/Continuation: enter(Ljdk/internal/vm/Continuation;)V
09:55:57  
09:55:57  Expected 8 methods in the stack but found 7

[keithc-ca]

That test should have been updated and verified to pass before this was merged.

[babsingh]

That test should have been updated and verified to pass before this was merged.

The test is excluded. It doesn't make a difference. Fixes can go in asynchronously since the test is excluded.

[keithc-ca]

Yes, the test is excluded, but this claimed to fix the problem, but does not.

[babsingh]

It does fix the issue documented in #21400 by adding support for the JVMTI events. The failure in #21590 (comment) is a separate issue tracked under #21402.