For ZFS pools, we may have to probe around more than we want to in order to figure out where the requested pool guid lives. The first set of loaddev will succeed, then further set will be blocked by the hook we install.
For these, we really just don't need the hooks to run. The currdev hook just validates the currdev passed in (presumably expecting user-specified currdev) and loaddev makes it immutable, but we don't know that loaddev is correct until after we've found the pool.