I don't think there is any reason. The original code had H_LOCK until the end of CheckHost so it's just paranoia on my part. H_LOCK after the rpc will be the next thing to go if no one can tell me why it's there.