Log message when a client tries to connect to an invalid hostname is incorrect

[rransom]

In connection_ap_handshake_rewrite_and_attach:

  addresstype = parse_extended_hostname(socks->address,
                         remapped_to_exit || options->AllowDotExit);

  if (addresstype == BAD_HOSTNAME) {
    log_warn(LD_APP, "Invalid onion hostname %s; rejecting",
             safe_str_client(socks->address));

parse_extended_hostname also returns BAD_HOSTNAME for .exit hostnames when AllowDotExit is off.

Also, parse_extended_hostname's documentation comment does not mention BAD_HOSTNAME.

[warms0x]

Per my discussion in #tor, it would seem the appropriate course of action is to update parse_extended_hostname's documentation. Since parse_extended_hostname actually invokes a log_warn() about the .exit notation being disable, it doesn't make much sense for connection_ap_handshake_rewrite_and_attach to do it as well.

It might make sense for parse_extended_hostname to return a new constant like DISABLED_HOSTNAME which would prevent the log_warn() in the description

[nickm]

Fix in branch bug3325 in my public repo; please review; should be straightforward.

[warms0x]

I've grabbed the latest refs from your public repo (presuming you meant the one under git.torproject.org) and I did not see a 'bug3325' branch at all.

[Sebastian]

Nick means ​https://gitweb.torproject.org/nickm/tor.git/shortlog/refs/heads/bug3325 (to get it,
git clone ​git://git.torproject.org/tor && cd tor && git remote add nickm ​git://git.torproject.org/nickm/tor && git fetch nickm && git checkout bug3325 )

[warms0x]

My mistake, I misread the comment assuming that rransom had made it.

That said, the change looks sane to me and seems to cooperate (for whatever my opinion is worth)

[nickm]

Thanks for the review; merged to master for inclusion in 0.2.3.x.