Re: Commercial stack fragility (Was RE: Cert Advisory 2002-03 and HP JetDirect)

From: Andrew M Hoerter (amhat_private)
Date: Fri Mar 01 2002 - 10:38:35 PST

  • Next message: Brian McWilliams: "Re: the dangers of disclosing vulnerabilities when the guilty party is ignorant of industry standards"

    On Wed, 27 February 2002 A.D., Brewis, Mark wrote:
    
    > Quite often these are commercial, off the peg TCP/IP stacks.  I have seen
    > some dreadful examples, both in terms of fragility and of TCP sequence
    > number generation.  I've seen sequential, sequential based on standard
    > increments, and repeating sequences.
    >
    > [...]
    > 
    > Compromise a network via the printers and you will have a network managers
    > attention.  The only problem lies in the paucity of solutions available to
    > correct the issue.
    
    Although it won't guard against attacks from within, one excellent
    solution to this problem is an appropriately designed firewall.  The
    latest release of OpenBSD[1] contains a new packet filter (`pf') which 
    can help protect buggy TCP stacks.  Two features will be of interest:
    
    *  The 'modulate state' directive, which causes a highly random initial
       sequence number to be substituted for those supplied by a less
       vigilant stack.
    
    *  The 'scrub' directive, which causes full fragment reassembly and 
       other packet normalization to take place before delivery to possibly
       fragile stacks.
    
    [1] http://www.openbsd.org/
    
    -- 
    "Everyone may openly covet everyone else's property, as long as he 
    appeals to democracy; and everyone may act on his desire for another 
    man's property, provided that he finds entrance into government."
           -- Hans-Hermann Hoppe
    



    This archive was generated by hypermail 2b30 : Fri Mar 01 2002 - 15:31:19 PST