From: nick at virus-l.demon.co.uk (Nick FitzGerald)
Subject: Mydoom

"Ferris, Robin" <R.Ferris@...ier.ac.uk> wrote:

> Does any one know what the size of the attachment is when is comes in as a
> zip file?

Yes and no.

Or, more helpfully, it is not a fixed size.

The size of the .ZIP depends on the length of the randomly selected 
filename that the sending instance of Mydoom chooses for the copy 
inside the .ZIP.  That filename is included twice in the .ZIP -- once 
in the header at the beginning of the stream of packed data for the 
file and once in the "central directory" at the end of the .ZIP.

...

BTW, if anyone is filtering on file size, you will almost certainly 
miss some copies of this (and other malware too).  I don't get large 
enough volumes of these things in my personal Email to do a suitable 
analysis, but I've seen stats from places like MessageLabs (perhaps 
Alex is reading and can post something on this??) showing interesting 
file size distributions for various of these monolithic replicators; 
passage through the Email system is not necessarily kind (aka "bit 
perfect") to them.

Now add the cases where the new self-mailer gets infected with a 
parasitic PE infector that expands an existing PE section, or adds one 
to the file.  This is then seen by a not yet updated scanner that 
happily disinfects the self-mailer of the  parasitic virus, but not 
knowing the new self-mailer leaves it to continue on its travels.  (Why 
any system admin would be so brain-dead as to _want_ to allow any kind 
of attachment known to have come from a "probably infected" machine 
in/out of their network is entirely beyond reason anyway...)  As many 
parasitic viruses cannot be "perfectly" removed (in the sense that the 
infected host cannot be rendered back into a bit-perfect replica of its 
pre-infection self) and disinfected files can even be left a different 
size from their original state, the copy of the self-mailer that 
spreads after such an encounter with an imperfect disinfector will be a 
physically different file, so simple hash-like detection and file size 
detection will fail for some samples.

By way of example, in the approx 40 Mydoom samples I have received as 
the result of its natural replication and spread in the last 12 hours 
or so, all but one decode to the commonly cited 22,528 byte file size.  
The odd one out has, for some unknown (but vaguely imaginable) reason 
"picked up" two extra bytes -- a pair of 0xFF characters added to the 
end of the file.  Because of the way Mydoom spreads, such non-fatal 
modifications to its .EXE file will be reproduced in future copies of 
the virus should this modified sample successfully replicate on some 
other potential victim machine.


-- 
Nick FitzGerald
Computer Virus Consulting Ltd.
Ph/FAX: +64 3 3529854

Powered by blists - more mailing lists