The purpose of this RFE is to improve the ability to debug situations where the VM exits unexpectedly via one of the vm_abort/vm_exit code paths. In some cases we don't dump core where arguably we should.
1. By default, dump core for any abort/C Heap OOM *except* "simple" failures during startup (vm_exit_during_initialization).
2. Revisit jni_FatalError: os::abort(false) (see comments).
3. Consider VM options to change the default behaviors (may not be needed.)
4. Ensure dumping core implies creating hs_err log as well (this may already be true in all cases.)