Having a better look, the patch from redhat does not help with the issue we are seeing where a passed call-back function seems to have one fewer arguments than what the function receiving the call-back expects. Scary. For now I'm moving on, but I'm inclined to simply side-step this issue with -fpermissive.