<html>
  <head>
    <meta content="text/html; charset=utf-8" http-equiv="Content-Type">
  </head>
  <body bgcolor="#FFFFFF" text="#000000">
    To the best of my recollection, the STARTTLS commit was a result of
    this "bug" :<a
      href="https://bugzilla.cyrusimap.org/show_bug.cgi?id=2980"><a class="moz-txt-link-freetext" href="https://bugzilla.cyrusimap.org/show_bug.cgi?id=2980">https://bugzilla.cyrusimap.org/show_bug.cgi?id=2980</a></a><br>
    <br>
    I don't know if having STARTTLS advertised on a pre-auth'd
    connection was causing issues or just annoyed the guy.  If we
    re-enable it, we should definitely do some testing to make sure that
    that it doesn't break the LMTP client code in used by lmtpproxy.<br>
    <br>
    <br>
    <div class="moz-cite-prefix">On 06/14/2016 07:55 PM, ellie timoney
      via Cyrus-devel wrote:<br>
    </div>
    <blockquote
cite="mid:1465948523.3941552.637870905.61AB4F09@webmail.messagingengine.com"
      type="cite">
      <title></title>
      <blockquote>
        <div>I use "lmtpd -a" listen on a NIC interface and receive lmtp
          request from a remote postfix instance. Now preauth works, but
          mail data was transfered without encryption.<br>
        </div>
      </blockquote>
      <div> </div>
      <div>Ah, I see what you mean.  I didn't know that -a option was
        there.<br>
      </div>
      <div> </div>
      <blockquote>
        <div dir="ltr">
          <div>I guess the commit your mentioned disabled startssl
            because the author think we just need ssl to protect PLAIN
            Password auth request.. <br>
          </div>
        </div>
      </blockquote>
      <div> </div>
      <div>My guess would be an assumption that no-one would send LMTP
        traffic over the internet (that's what SMTP is for).  If one
        expects LMTP traffic (and especially pre-authed LMTP traffic) to
        be within a single server, or at most between servers sitting in
        nearby racks over a private network, then this all makes sense.<br>
      </div>
      <div> </div>
      <div>Ken, do you want to chime in here?  The disabling STARTTLS
        commit (b93e6be) and the one to add the -a option (a501222) were
        both yours.  I'm hoping to get a clearer understanding of the
        intent.<br>
      </div>
      <div> </div>
      <div>For what it's worth, lmtpd(8) man page says:<br>
      </div>
      <div> </div>
      <blockquote>
        <div>       -a     Preauthorize connections initiated on an
          internet socket, instead  of  requiring  LMTP  AUTH.<br>
        </div>
        <div>              This should only be used for connections
          coming from trusted hosts.<br>
        </div>
      </blockquote>
      <div> </div>
      <div>Maybe this could be expanded: "from trusted hosts, over
        trusted networks".<br>
      </div>
      <div> </div>
      <blockquote>
        <div dir="ltr">
          <div>Personally, I think all mail data should be encrypted in
            internet transfer.<br>
          </div>
        </div>
      </blockquote>
      <div> </div>
      <div>This kind of sounds like the answer might be "don't use -a
        when your listen address is reachable over the internet"...<br>
      </div>
      <div> </div>
      <div>My own inclination is to re-enable STARTTLS, but I'd like to
        better understand why it was disabled before I do so.<br>
      </div>
      <div> </div>
      <div>On Tue, Jun 14, 2016, at 07:16 PM, qyb wrote:<br>
      </div>
      <blockquote type="cite">
        <div dir="ltr">
          <div>I use "lmtpd -a" listen on a NIC interface and receive
            lmtp request from a remote postfix instance. Now preauth
            works, but mail data was transfered without encryption.<br>
          </div>
          <div> </div>
          <div>I guess the commit your mentioned disabled startssl
            because the author think we just need ssl to protect PLAIN
            Password auth request.. Personally, I think all mail data
            should be encrypted in internet transfer.<br>
          </div>
        </div>
        <div>
          <div> </div>
          <div defang_data-gmailquote="yes">
            <div>On Tue, Jun 14, 2016 at 9:25 AM, ellie timoney via
              Cyrus-devel <span dir="ltr"><<a moz-do-not-send="true"
                  href="mailto:cyrus-devel@lists.andrew.cmu.edu">cyrus-devel@lists.andrew.cmu.edu</a>></span>
              wrote:<br>
            </div>
            <blockquote
              style="margin-top:0px;margin-right:0px;margin-bottom:0px;margin-left:0.8ex;border-left-width:1px;border-left-color:rgb(204,
              204, 204);border-left-style:solid;padding-left:1ex;"
              defang_data-gmailquote="yes">
              <div><br>
              </div>
              <div>
                <div> </div>
                <div><span>On Wed, Jun 1, 2016, at 03:28 AM, qyb via
                    Cyrus-devel wrote:</span><br>
                </div>
                <blockquote type="cite">
                  <div dir="ltr">
                    <div><span>I noticed that cyrus disable TLS on
                        preauth'd connection.</span><br>
                    </div>
                    <div>
                      <div> <br>
                      </div>
                      <div><span>Authentication info(plain password...)
                          need TLS protection. And I think that RFC822
                          text also need TLS.</span><br>
                      </div>
                    </div>
                  </div>
                </blockquote>
                <div> <br>
                </div>
                <div> </div>
                <div>Can you expand on this a bit?<br>
                </div>
                <div> <br>
                </div>
                <div>As far as I understand, connections are only ever
                  preauth'd when they come in via UNIX-domain sockets,
                  which are inherently local.  What are you trying to
                  protect, and from whom?<br>
                </div>
                <div> <br>
                </div>
                <div>For what it's worth, it looks like STARTTLS used to
                  work (at least to some degree) for preauth'd LMTP, but
                  was explicitly disabled in 2001 by this commit:<br>
                </div>
                <div><a moz-do-not-send="true"
href="https://cgit.cyrus.foundation/cyrus-imapd/commit/?id=b93e6be5b19362f9e295b40ceb81b702d73de6bb">https://cgit.cyrus.foundation/cyrus-imapd/commit/?id=b93e6be5b19362f9e295b40ceb81b702d73de6bb</a><br>
                </div>
                <div>So I guess you might be able to re-enable it by
                  doing the inverse of that, though I'm not really
                  seeing the point?<br>
                </div>
              </div>
            </blockquote>
          </div>
        </div>
      </blockquote>
      <div> </div>
    </blockquote>
    <br>
    <pre class="moz-signature" cols="72">-- 
Kenneth Murchison
Principal Systems Software Engineer
Carnegie Mellon University
</pre>
  </body>
</html>