Path: utzoo!utgpu!news-server.csri.toronto.edu!bonnie.concordia.ca!uunet!bellcore!epic!karn From: karn@epic.bellcore.com (Phil R. Karn) Newsgroups: comp.org.eff.talk Subject: Re: stealing passwords is easy! Message-ID: <1991Jun12.191551.7844@bellcore.bellcore.com> Date: 12 Jun 91 19:15:51 GMT References: <14906.28501E22@fidogate.FIDONET.ORG> Sender: usenet@bellcore.bellcore.com (Poster of News) Reply-To: karn@thumper.bellcore.com Organization: Packet Communications Research Group (Bellcore) Lines: 28 In article <14906.28501E22@fidogate.FIDONET.ORG>, david.turrell@f111.n125.z1.FIDONET.ORG (david turrell) writes: |> I remember a WYLBUR system whose teletypes echoed the password. After typing a |> "mask" of G's on top of W's on top of M's. Then people left the printout lying |> around; didn't even bother to throw it away when they left. |> Yes, back at Cornell in the middle 70's I demonstrated a similar attack against VM/CMS passwords. As a typical IBM mainframe system, VM/CMS operated in half duplex (virtual 2741 mode), so it had no way of directing a terminal to suppress echo as a password was typed in. So a series of overstruck characters were printed as part of the prompt in an attempt to mask the password. I'm not sure now, but I think the characters were S, M and *. Printing terminals, particularly the dot-matrix kind (Decwriters and Silent 700-class thermal printers) were of course the standard in those days. I noticed that a few dots in the matrix were left unhit by the mask characters in the font used by at least one dot matrix printer (the NCR terminal in my room, used by many other members of my fraternity, including one with system programmer privileges). Given a "masked" password from the trash I discovered I could easily narrow down the possible letters for each position within the password by examining the matrix dot positions left clear by the password mask. Since the passwords were only four characters long to begin with, it didn't take long to determine the right one by trial and error. Phil