On Fri, 17 May 2024 05:52:36 GMT, David Holmes <dhol...@openjdk.org> wrote:
> I have to wonder whether this solution will potentially cause problems > because the code will now block for safepoints. We could fallback to > `Thread::name()` if the current thread is in-native. Thanks for feedback. Here is the update. I've updated the safe_get_thread_name() to not change thread state. In "jvmtiEnter.xsl" functions the thread name is s read once before the transition happened and re-used then. So I updated the tracing to 're-read' if the transition to VM happened to update the thread name once it became known. ------------- PR Comment: https://git.openjdk.org/jdk/pull/19275#issuecomment-2118004152
