> I understand the current implementation. But unfortunately that it > might not work in the new implementation - as there are persistence > request comes both from libdladm and the kernel, we will use a daemon > to do the persistence processing but the daemon won't know about the > boundary of each subcommand invocation.
That seems like an implementation detail. For example, each request could include the rootdir and libdladm requests would set the rootdir to the value set by the given application. -- meem
