RE: On the ultimate futility of server-based mail scanning

From: Richard M. Smith (rmsat_private)
Date: Tue Mar 05 2002 - 18:30:58 PST

  • Next message: Michael Bacarella: "Re: efingerd remote buffer overflow and a dangerous feature"

    No security system works 100% of the time.  However server-based
    scanning of email attachments probably works better than many other
    options like relying on end-users to not run unsafe attachments.  I vote
    that we keep it.
    
    Richard
    
    -----Original Message-----
    From: David F. Skoll [mailto:dfsat_private] 
    Sent: Monday, March 04, 2002 5:07 PM
    To: bugtraqat_private
    Subject: On the ultimate futility of server-based mail scanning
    
    
    Several postings on Bugtraq recently talked about DoS attacks against
    server-based mail-scanners.  Compress four gigabytes of zeros and
    debilitate mail scanners which uncompress .gz files, for example.
    
    Several mail scanners try to be clever and examine .zip files, .tar.gz
    files, .arc files, etc. to look inside for viruses.
    
    This is ultimately futile.
    
    I gave one scenario:
    
    (cat small_x86_jmp_code; \
     dd if=/dev/zero bs=1k count=400k; \
     cat virus_payload) | gzip > virus.attach.gz
    
    This DoS's virus-scanners which do not limit scanning-size, and sneaks
    past those which do.
    
    There's an even better method, and one which is very amenable to
    social-engineering:
    
    "HEY!  NUDE pictures of Pamela Anderson in the attachment nudie.zip.
    Just  unzip and then run pam.exe.  Oh, heh, heh, heh -- to keep your
    boss from  seeing this, we've password-protected the zip file.  The
    unzip password  is z3kr3t.  Enjoy!"
    
    Zip encryption is pathetic.  But I don't think anyone's seriously
    suggesting server-based scanners should brute-force encrypted zip files
    to check for viruses, or perform AI analysis of messages to extract
    passwords.
    
    Ultimately, the responsibility falls on the MUA and the end-user's OS
    vendor.  We either put secure end-user software onto the desktop, or we
    admit defeat.
    
    --
    David F. Skoll
    
    Roaring Penguin Software Inc. | http://www.roaringpenguin.com GPG
    fingerprint: C523 771C 3710 0F54 B2D2 4B0D C6EF 6991 34AB 95BA GPG
    public key:  http://www.roaringpenguin.com/dskoll-key-2002.txt ID:
    34AB95BA
    



    This archive was generated by hypermail 2b30 : Wed Mar 06 2002 - 14:52:27 PST