jimingham wrote: We'd have to make the change and see to be sure, but what I would expect is that if any other thread tried to call Kill while whoever is the currently the primary process listener is handling the stop, it would immediately fail because the process is not stopped. Right now, the Kill is sometimes allowed because the public state is "stopped" even though we haven't actually finished handling the stop event. So I think what I was suggesting will fix the original problem.
The race Pavel is pointing out also seems real to me, but that should be fixed at the source, rather than after the fact. Process::HijackProcessEvents should check the previous "primary event listener" to see if it has any events and if so they should be transferred to the Hijack Listener. We should also do the same thing in RestoreProcessEvents. That way we'll never drop an event when shuffling primary listeners. https://github.com/llvm/llvm-project/pull/144919 _______________________________________________ lldb-commits mailing list lldb-commits@lists.llvm.org https://lists.llvm.org/cgi-bin/mailman/listinfo/lldb-commits
