Article View: vmsnet.mail.pmdf
Article #685RE: Did a rewritten "From:" make a message appear forged?
From: Ned Freed
Date: Thu, 17 Dec 1992 18:27
Date: Thu, 17 Dec 1992 18:27
102 lines
5943 bytes
5943 bytes
> One of our users (ECH) got a message that said in the VMS MAIL "From:" line: > From: IN%"hannon@ac.dal.ca" 16-DEC-1992 11:55:18.99 > and replied to that message. When user "hannon" got the reply she was > surprised, because she hadn't sent the original message. She was worried that > someone was forging mail. Although it is trivial to forge mail, this does not look like a case of forged mail. > We got ECH to send us the original message he replied to. The headers, > including VMS MAIL's, looked like this: > From: IN%"hannon@ac.dal.ca" 16-DEC-1992 11:55:18.99 > To: IN%"ECH@ac.dal.ca" > CC: > Subj: Beer compatible solder > Return-path: <hannon@sparc3303.cacd.cr.rockwell.com> > Received: from sparc3303.cacd.cr.rockwell.c (sparc3303.cacd.cr.rockwell.com) by > AC.DAL.CA (PMDF #2545 ) id <01GSDREERFN4001Z65@AC.DAL.CA>; Wed, > 16 Dec 1992 11:28:35 -0400 > Received: from pc234sun2601.cacd.cr.rockwell.com ([131.198.60.234]) by > sparc3303.cacd.cr.rockwell.com (4.0/SMI-4.0) id AA15863; Wed, > 16 Dec 92 09:27:56 CST > Date: 16 Dec 1992 09:25:20 +0000 > From: hannon@ac.dal.ca (<hannon@sparc3303.cacd.cr.rockwell.com>) > Subject: Beer compatible solder > To: ECH@ac.dal.ca > Message-id: <QB2F4A61@pc234sun2601> > Content-transfer-encoding: 7BIT > My theory, based on the domain name "pc234sun2601.cacd.cr.rockwell.com" in > one of the "Received:" lines, is that a badly written PC mail program wrote > the original "From:" line like > From: hannon (<hannon@sparc3303.cacd.cr.rockwell.com>) > and that PMDF rewrote "hannon" to "hannon@ac.dal.ca". Is this possible? Yes, this is precisely what happened. > If so, is there any way to prevent PMDF from doing this? Not really. When PMDF receives mail with headers containing local-parts and no domain names it can do one of four things: it can reject the message because the header is invalid, it can do nothing, it can add the name of the local host to the address, or it can add the name of the host it received the mail from to the address. None of these choices is appropriate for this particular message. Rejecting messages like this would cause you to lose about 15% of your mail. You have no idea (because PMDF insulates you from it) just how common the no-host-name-in-address problem is, and how often PMDF is called upon to fix broken headers. I don't think you or any other customer site would appreciate bouncing a significant fraction of your mail. PMDF therefore doesn't even implement this as an option. Delivering mail, bad address or no, is the top priority. Doing nothing is never appropriate either. It is imperative that PMDF do something to make these headers legal, since there are plenty of mail systems out there that will not accept messages with illegal headers. PMDF doesn't want to be placed in a position where it cannot send the message forward or backward. This option is therefore unworkable in practice. Adding the local host name is often the appropriate thing to do, because SMTP is often used by remote mail clients to submit messages. And often as not these clients leave the host name out of the From: address, since they assume that the SMTP server will tack on its own name. Failure to do this breaks PMDF's ability to support things like Eudora, Maelstrom, Popper, and many other PC/Mac remote mail clients. PMDF therefore implements this option as a default since it is far and away the most likely "correct" action to take. PMDF also implements the "add the name of the sending system" option. This is done because sendmail often leaves out its local host name on Cc: addresses. We therefore provide this as a per-channel option called "remotehost". This can be used in lieu of fixing broken sendmails at sites where this problem is particularly acute. It is not a reasonable default action to take, however. I suppose we could implement a fifth option, which would dredge the host out of the oldest Received: header. However, I suspect that this action isn't appropriate even in this case. Unless I miss my guess, the problem here is that this message originated with a remote mail client and its SMTP relay didn't do the right thing to fix up the message headers. The proper domain name to add is therefore unknown; it could be pc234sun2601.cacd.cr.rockwell.com, it could be sparc3303.cacd.cr.rockwell.com (but not sparc3303.cacd.cr.rockwell.c, which is what the remotehost option would produce), or it could be something else that's completely different. There's just no way to be sure. > I can see it might be > difficult, given that mapping my theoretical "From:" line to VMS MAIL might > produce ``From: IN%"hannon"'' which is not much better than what was done in > the first place. Correct again. This is one of those cases where there's definitely a problem but there's nothing that you can do about it. All the proposed cures are either unworkable in practice or don't solve the problem. My suggestion would be to contact the postmaster at this remote site and ask them to fix their broken mailer that is emitting these bogons. Ned
Technical Details
Message-ID:
<01GSF5N0EBNM94E282@SIGURD.INNOSOFT.COM>Path:
arc1-ca.pugleaf.net!sparky!uunet!usc!sdd.hp.com!ux1.cso.uiuc.edu!moe.ksu.ksu.edu!mccall!ipmdf-newsgate!listShow Full Headers
Complete Article Headers
Path: arc1-ca.pugleaf.net!sparky!uunet!usc!sdd.hp.com!ux1.cso.uiuc.edu!moe.ksu.ksu.edu!mccall!ipmdf-newsgate!list Newsgroups: vmsnet.mail.pmdf Subject: RE: Did a rewritten "From:" make a message appear forged? Message-ID: <01GSF5N0EBNM94E282@SIGURD.INNOSOFT.COM> From: Ned Freed <ned@sigurd.innosoft.com> Date: 17 Dec 1992 11:27:07 -0700 (PDT) Organization: The Internet Return-Path: <epmdf@YMIR.CLAREMONT.EDU> Resent-Date: 17 Dec 1992 11:27:07 -0700 (PDT) Resent-From: epmdf@YMIR.CLAREMONT.EDU Errors-To: epmdf@YMIR.CLAREMONT.EDU Resent-Message-ID: <01GSF5OFQAGY90MXYZ@YMIR.CLAREMONT.EDU> X-Vms-To: IN%"AE@AC.DAL.CA" X-Vms-Cc: IPMDF Mime-Version: 1.0 Content-Type: TEXT/PLAIN; CHARSET=US-ASCII Content-Transfer-Encoding: 7BIT Lines: 102