On Tue, 24 Feb 2026 02:08:25 GMT, Serguei Spitsyn <[email protected]> wrote:

>> The function `JvmtiThreadState::process_pending_interp_only()` should 
>> protect the check with `state->is_pending_interp_only_mode()` and call to 
>> `JvmtiEventController::enter_interp_only_mode(state)` with the 
>> `JvmtiThreadState_lock`. Some level of optimization with a check of 
>> `seen_interp_only_mode()` is used as the code path of the 
>> `process_pending_interp_only()` is hot. Then the `seen_interp_only_mode()` 
>> has to be set somewhat earlier when the `is_pending_interp_only_mode()` is 
>> set.
>> This issue was discovered when in the work on the PR update of 
>> https://bugs.openjdk.org/browse/JDK-8373367 .
>> 
>> Testing:
>>  - TBD: mach5 tiers 1-5
>
> Serguei Spitsyn has updated the pull request incrementally with one 
> additional commit since the last revision:
> 
>   review: convert volatile bool to Atomic<bool>

src/hotspot/share/prims/jvmtiThreadState.hpp line 207:

> 205:   // Return true if any thread has entered interp_only_mode at any point 
> during the JVMs execution.
> 206:   static bool seen_interp_only_mode() {
> 207:     return _seen_interp_only_mode.load_relaxed();

Not sure `load_relaxed` is enough here.
I think we need `load_acquire`/`release_store`

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

PR Review Comment: https://git.openjdk.org/jdk/pull/29800#discussion_r2855087177

Reply via email to