Tony Nguyen wrote: > We're handling universal addresses from rpcbind(3NSL) so the address > should have correct format.
Though the connection is local and to a daemon under our control, the daemon can certainly malfunction and emit malformed addresses. And as a general rule, errors in one process -- especially one accessed over a network connection -- should never result in corrupting the address space of another. It's fine if the other process fails, but it must be an intentional, well-defined failure. > However, I agree with and Dave that we should better handle malformed > input. I've made the change to make sure that we don't walk past the > input address. Good. Dave
