Article View: comp.security.misc
Article #383Re: Covert channel
From: John F. Haugh II
Date: Sat, 15 Aug 1992 04:32
Date: Sat, 15 Aug 1992 04:32
44 lines
2973 bytes
2973 bytes
In article <1992Aug14.190446.11355@hpcvusn.cv.hp.com>, mark@hpcvxmsr.cv.hp.com (Mark Rowe) writes: > Disagree, any channel can be a covert channel regardless of whether or > not it is normally thought of as a communication mechanism. A trojan > horse that acquires classified information formulates it into an email > messages where the information is imbedded in seemingly harmless text > and then uses the system mail system to send it out, is certainly > using a standard communication mechanism. What makes this a "covert" > channel is that the message is hidden or disguised or masquerading as > something else as opposed to an overt channel where the information is > obvious and undisguised (or disguised in some understood manner such > as encrypting). I'd argue that if the mechanism is designed to transmit data (for example, a system mailer) that the mechanism had better know the sensitivity of all of the data in the message. A trojan horse that collects classified data will be executing with a MAC label that dominates the data that has been collected. The data that the trojan horse outputs will all have a MAC label that dominates the trojan horse process which creates the output data. Suppose for a moment that the trojan horse creates a mail message with bogus headers. Either that entire message, or the data in the bogus headers, must be labeled appropriate. The mailer would have to be broken to take a message with a MAC label and give that data to a process with some other MAC label which does not dominate the MAC label of the message itself. I had this discussion back when I was starting to learn about MAC. My question was, what is to prevent a user from putting the confidential data in the filename and then doing an "ls" of the directory? The answer was that you can only write a filename into a directory if the directory has the appropriate label. I think this applies here as well. You can only put that sensitive data into the header if the entire message, or that field, has the appropriate label. And you can only read that data if your process' label dominates the message's label. What you are describing sounds more like busted MAC than a covert channel. -- John F. Haugh II | Life's Little Instruction Book: Ma Bell: (512) 251-2151 | "#138. Learn Spanish. In a few years, UUCP: ...!cs.utexas.edu!rpp386!jfh | more than 35% of all Americans will speak Domain: jfh@rpp386.cactus.org | it as their first language."
Technical Details
Message-ID:
<21378@rpp386.lonestar.org>Path:
arc1-ca.pugleaf.net!sparky!uunet!cs.utexas.edu!chinacat!rpp386!jfhReferences:
<71269183717599@juju.mcs.anl.gov> <61350002@otter.hpl.hp.com> <1992Aug14.190446.11355@hpcvusn.cv.hp.com>Show Full Headers
Complete Article Headers
Path: arc1-ca.pugleaf.net!sparky!uunet!cs.utexas.edu!chinacat!rpp386!jfh From: jfh@rpp386.lonestar.org (John F. Haugh II) Newsgroups: comp.security.misc Subject: Re: Covert channel Message-ID: <21378@rpp386.lonestar.org> Date: 15 Aug 92 04:32:44 GMT References: <71269183717599@juju.mcs.anl.gov> <61350002@otter.hpl.hp.com> <1992Aug14.190446.11355@hpcvusn.cv.hp.com> Organization: River Parishes Programming, Austin TX Lines: 44