On Tue, 3 Nov 2026 12:08:48 GMT, Markus Grönlund <[email protected]> wrote:
> Please see https://bugs.openjdk.org/browse/JDK-8343177 Thank you for the pointer, it clarifies the problem of thread IDs. I am fine with keeping double call if it was a deliberate design choice. In that case, I believe the fix proposed here is correct and aligned with the intent of JDK-8343177. Alternatively, I can remove `if (jt->is_attaching_via_jni())` check and always initialize JfrCPUTimeThreadSampling in `_thread_in_vm` state only (as shown in the inline comment). Do you have any preference? ------------- PR Comment: https://git.openjdk.org/jdk/pull/33678#issuecomment-6621948563
