RFR: 8373367: interp-only mechanism fails to work for carrier threads in a corner case [v9]

[Alex Menkov]

On Wed, 25 Feb 2026 23:44:09 GMT, Serguei Spitsyn <sspitsyn at openjdk.org> wrote:

>> src/hotspot/share/prims/jvmtiEventController.cpp line 381:
>> 
>>> 379:   if (target == nullptr ||                                                         // an unmounted virtual thread
>>> 380:       JvmtiEnvBase::is_thread_carrying_vthread(target, state->get_thread_oop())) { // a vthread carrying thread
>>> 381:     return;  // EnterInterpOnlyModeClosure will be executed right after mount.
>> 
>> The condition is not clear to me. How we do the deoptimization for mounted vthread?
>
> Thank you for checking.  It can be done in two different ways. The `JvmtiThreadState` of a mounted virtual thread will be used to enter `interp_only_mode` for mounted virtual thread. The `state->get_thread()` has to be non `null` in both cases. 
> One way is when virtual thread is currently mounted. The chain of calls is: 
>  `SetEventNotificationMode()` =>
>  `JvmtiEventControllerPrivate::set_user_enabled()` =>
>  `JvmtiEventControllerPrivate::recompute_enabled()` =>
>  `JvmtiEventControllerPrivate::recompute_thread_enabled() =>
>   `JvmtiEventControllerPrivate::enter_interp_only_mode()` => 
>    . . . =>
>    `EnterInterpOnlyModeClosure::do_thread()`
> 
> Another way is at a `mount` transition point. The chain of calls is: 
>   `JavaThread::rebind_to_jvmti_thread_state_of()` =>
>  `JvmtiThreadState::process_pending_interp_only()` =>
>  `JvmtiEventController::enter_interp_only_mode()` => 
>  `EnterInterpOnlyModeClosure::do_thread()`

Ok, I see now. Usage of `is_thread_carrying_vthread` is not obvious.
 I suggest to explain it a bit, something like
// There are 2 cases when deoptimization is postponed:
// 1. Unmonted virtual thread - EnterInterpOnlyModeClosure will be executed right after mount;
// 2. Carrier thread with mounted virtual thread - EnterInterpOnlyModeClosure will be executed right after unmount.

And update or remove the comment on return (it tells only about mount).

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

PR Review Comment: https://git.openjdk.org/jdk/pull/29436#discussion_r2856453611