Re: More on Trojan Horse -- discussion of "Trojan Horse" terminology

2004-04-09 Thread Onno Benschop
On Fri, 2004-04-09 at 19:21, James Devenish wrote:
> In message <[EMAIL PROTECTED]>
> on Fri, Apr 09, 2004 at 06:32:24PM +1000, Onno Benschop wrote:
> Yes (though I only tried this under Linux/Pentium).
> 
> > *and* code executes that does something else,
> 
> I haven't tried this under Mac OS X.
> 
> > > From examining this file, I see that it contains a GEO "general
> > > encapsulated object" that itself encapsulates a PowerPC PEF header (Mac
> > > OS 9 executable) with filename "virus.mp3" -- the same as the existing
> > > file. Perhaps iTunes extracts the GEO, overwriting the original
> > > virus.mp3 in the process?
> > 
> > If that is the case, we're talking about an iTunes exploit, not a
> > Trojan.
> 
> Not necessarily. iTunes would not be executing the code itself -- it
> might merely be following a liberal interpretation of ID3 (eek!). While
> this would be a misfeature in iTunes, the malicious binary would
> actually be launched by the Finder.

Yeah, except that iTunes is the one making the .mp3 into an executable.


> Thus, if ID3 provides a sanctioned
> way to initiate the extraction of arbitrary files hidden within music, I
> would think it to be an example of the Trojan Horse phenomenon.

Yup, but my understanding of the ID3 definition is that it contains
meta-data like artist, name, album etc. There should be no "extraction"
required to get this stuff out. Merely from this byte, or from this
delimiter to this delimiter is the name of the artist.

I wouldn't have thought that the ID3 definition had any means of
packaging anything - unless iTunes decided that it would be cool to say
that from this byte to this byte is a compressed image that can be
extracted as a separate file, in which case the guy who thought of that
is a moron.


> It would
> be possible, for example, for a cracker to insert malicious code into
> other people's audio files as part of website defacement.

And it would be pretty subtle too, evil...


>  Although
> Trojans are by their definition (hmm...what definition?)

Here are some :-)


"The Collaborative International Dictionary of English v.0.48"
Trojan horse Tro"jan horse`, n. from the incident described
   in Homer's Iliad.
   1. (Classical mythology) a large hollow wooden horse built by
  Greek soldiers besieging Troy during the Trojan War, and
  left as a "gift" when they pretended to abandon their
  seige. It was taken into the city by the Trojans, and
  Greek soldiers concealed inside came out and opened the
  gates to the city, enabling the capture of the city by the
  Greeks.
  RP + PJC

   2. Hence, any thing or person which appears harmless but is
  designed to destroy or attack from within. It may
  sometimes refer to a group; -- see also fifth column.
  RP + PJC

   3. (Computers) A computer program designed to evade the
  security precautions within a computer system and perform
  illicit operations, or to do malicious damage, and often
  designed to look like a different kind of program, such as
  a game, archiver, or directory lister. This term is not
  applied to a program that replicates itself, such as a
  virus.
  RP + PJC

"The Collaborative International Dictionary of English v.0.48"
fifth column fifth` col"umn, n. from a statement during the
   Spanish Civil War (1936) that the Falange had four columns of
   soldiers marching on the city, and a fifth column "already
   there" (i.e. sympathizers inside the Republican lines).
   1. a group of persons inside the battle lines of a territory
  engaged in a conflict, who secretly sympathize with the
  enemy, and who engage in espionage or sabotage; --
  sometimes also referred to as a trojan horse.
  RP

   2. Hence, any faction of persons within a group who secretly
  sympathize with an enemy, especially those who engage in
  activities harmful to the group; an enemy in one's midst;
  a group of traitors.
  RP

"WordNet (r) 2.0 (August 2003)"
Trojan horse
n 1: a subversive group that supports the enemy and engages in
 espionage or sabotage; an enemy in your midst syn:
 fifth column, Trojan horse
2: a program that appears desirable but actually contains
   something harmful; "the contents of a trojan can be a
   virus or a worm"; "when he downloaded the free game it
   turned out to be a trojan horse" syn: trojan
3: a large hollow wooden figure of a horse (filled with Greek
   soldiers) left by the Greeks outside Troy during the
   Trojan War syn: Trojan Horse, Wooden Horse

"The Free On-line Dictionary of Computing (19 Sep 2003)"
Trojan horse

(Coined by
   MIT-hacker-turned-NSA-spook Dan Edwards) A malicious,
   security-breaking program that is disguised as something
   benign, such as a directory lister, archiver, game, or (in one
   notorious 1990 case on the Mac) a program to find and destroy
   viruses!  A Trojan horse is similar to a back door.

 

Re: More on Trojan Horse -- discussion of "Trojan Horse" terminology

2004-04-09 Thread Craig Ringer
On Fri, 2004-04-09 at 17:21, James Devenish wrote:

> I suspect this problem does not lie with the OS or the user. It's either
> with ID3 or iTunes (most likely: iTunes' overzealous honouring of ID3).

iTunes does support storing album covers or somesuch, doesn't it?
Perhaps it supports storing images in ID3 data, and that support is a
little _too_ flexible?

Craig Ringer



Re: More on Trojan Horse -- discussion of "Trojan Horse" terminology

2004-04-09 Thread James Devenish
In message <[EMAIL PROTECTED]>
on Fri, Apr 09, 2004 at 06:32:24PM +1000, Onno Benschop wrote:
> To make sure here, we're talking about a valid stuffit archive file that
> has a compressed file that when extracted is identified as an mp3?

Yes (I tried this under Mac OS X).

> Now if you can actually play the mp3 and music happens

Yes (though I only tried this under Linux/Pentium).

> *and* code executes that does something else,

I haven't tried this under Mac OS X.

> > From examining this file, I see that it contains a GEO "general
> > encapsulated object" that itself encapsulates a PowerPC PEF header (Mac
> > OS 9 executable) with filename "virus.mp3" -- the same as the existing
> > file. Perhaps iTunes extracts the GEO, overwriting the original
> > virus.mp3 in the process?
> 
> If that is the case, we're talking about an iTunes exploit, not a
> Trojan.

Not necessarily. iTunes would not be executing the code itself -- it
might merely be following a liberal interpretation of ID3 (eek!). While
this would be a misfeature in iTunes, the malicious binary would
actually be launched by the Finder. Thus, if ID3 provides a sanctioned
way to initiate the extraction of arbitrary files hidden within music, I
would think it to be an example of the Trojan Horse phenomenon. It would
be possible, for example, for a cracker to insert malicious code into
other people's audio files as part of website defacement. Although
Trojans are by their definition (hmm...what definition?) a social
engineering exploit, a pure social engineering exploit would not need to
involve the concealment of an executable payload.

I did think at first that it must be iTunes-specific, because iTunes is
the default player for MP3 files. (Intego hasn't provided sufficient
details.) However, it might be a common vulnerability amongst audio
players that interpret ID3 headers. On most UNIX systems, however, you'd
also need to set the "executable" permission -- something that probably
can't be conveyed via ID3.

I suspect that it's unlikely that a malicious exploit would be a virus
-- more likely a worm.

> So now we're opening the same file twice?

Not by the sound of what I've read on the web. (The double-open
procedure was merely my own speculation.)

> I realise I'm arguing semantics here, but in this world I believe that
> this is important, because the difference determines where the fix lies
> - the User, the OS or iTunes.

I suspect this problem does not lie with the OS or the user. It's either
with ID3 or iTunes (most likely: iTunes' overzealous honouring of ID3).

> If I were you and you didn't have a completely separate machine that you
> would be prepared to sacrifice, I wouldn't even have gone as far as you
> state you have...

I didn't believe that a StuffIt Expander, `vim` or `file` exploit was
involved. However, you are correct that I don't consider my test machine
'entirely sacrificial'. I did think twice before using StuffIt Expander,
and perhaps I shouldn't have carried through with it. However, I cannot
see any evidence that any files were modified as a result of
'unstuffing' the file (apart from com.stuffit.Expander.plist, which I
have now removed), nor can I see any suspicious processes.

> > Get a Mac! Oops ;-)
> I did - two years or so later - a Mac 512ED, which served me well for
> four years when I sold it just before the LC came out.

Ah, yes, I think the 512K was my mainstay during primary school.