On Mon, 20 May 2024 04:25:49 GMT, David Holmes <dhol...@openjdk.org> wrote:
> Okay but how does this `dbgRawMonitor` lock prevent the GC from clearing a > jobject? The JNI ref. It's actually a global ref (I said local ref earlier). As long as any setters of DebugRawMonitor::ownerThread (setting the thread or clearing it) grab dbgRawMonitor first, and the current thread accessing ownerThread does the same, then you know that whatever ownerThread points to is either NULL, the current thread, or some other thread that is still alive and is actually holding the monitor that the DebugRawMonitor is encapsulating. As long as the current thread continues to hold dbgRawMonitor, no other thread will be able to change ownerThread. This means that if some other thread currently owns the monitor that the DebugRawMonitor is encapsulating, then it will first block on dbgRawMonitor before trying to release it and clear ownerThread. ------------- PR Comment: https://git.openjdk.org/jdk/pull/19044#issuecomment-2119650743
