SUBST problem

From: Dave Tarbatt - ACS (D.A.Tarbattat_private)
Date: Tue Nov 30 1999 - 01:19:30 PST

  • Next message: Mixter: "serious Qpopper 3.0 vulnerability"

    I've not seen this mentioned on any of the security mailing lists (NTBUGTRAQ,
    BUGTRAQ etc.) and I cannot find reference to it by searching microsoft.com.
    I've e-mailed Microsoft about it on the 13th October but apart from a reply
    saying "we'll look into it", no reply as yet.
    
    *** The problem:
    Tested with NT4WS SP3 and SP5.
    SUBSTed drives are persistent between different logged on users. Users can be
    misled into saving data somewhere other than where they first thought, running
    trojaned executables etc.
    
    *** To recreate (typical example):
    An ordinary user logs onto the NT workstation and maps a drive to a
    subdirectory:
    
    SUBST M: C:\TEMP
    
    They log off.
    
    A second user logs onto the same workstation. The SUBSTed drive is still in
    effect. Their profile defines that M: be their home directory, mapped to
    \\SERVER\USERNAME$. It doesn't get connected and there is no error message.
    The user saves their documents to what they believe to be their home drive
    (M:) but in actual fact they end up in C:\TEMP.
    
    They log off.
    
    The first user comes back and reads their saved documents from C:\TEMP. There
    are many other possible exploits that this could be used for.
    
    *** Workaround/fix:
    None known. You could delete %WINDIR%\SYSTEM32\SUBST.EXE but someone could
    always just run their own version from a floppy, network drive or whatever.
    If you reboot the machine every time before you log in, SUBSTed drives are
    removed. Maybe practical on workstations where you have EXEs run from network
    drives at login, not too practical on servers (but if you can't trust the
    people who have access to them anyway....)
    
    Dave,
    http://redirect.to/null/
    PGP fingerprint: AE23 A19C 3E5E 74F4 2193  4BB3 E154 54AF 1350 F4FC
    



    This archive was generated by hypermail 2b30 : Fri Apr 13 2001 - 15:14:58 PDT