Re: application whitelisting (was RE: Active Directory Question)

dlimanov_at_sct.com
Date: 01/14/04

  • Next message: Kayne Ian (Softlab): "RE: application whitelisting (was RE: Active Directory Question)"
    To: jlacour@zonelabs.com
    Date: Wed, 14 Jan 2004 14:36:49 -0500
    
    

    Some Host Intrusion Prevention Systems (Okena, for example) offer very
    granular control of what user can and can not do. Application
    installation is one of the things you can control and grant or deny on
    group-membership basis, using HIPS products.
    Probably a better (if more expensive) way then AD.
    My $0.02.

    Dimitri

    |---------+---------------------------->
    | | "John LaCour" |
    | | <jlacour@zonelabs|
    | | .com> |
    | | |
    | | 01/14/2004 12:15 |
    | | PM |
    | | |
    |---------+---------------------------->
    >--------------------------------------------------------------------------------------------------------------|
      | |
      | To: "Kayne Ian (Softlab)" <Ian.Kayne@softlab.co.uk>, <focus-ms@securityfocus.com> |
      | cc: |
      | Subject: application whitelisting (was RE: Active Directory Question) |
    >--------------------------------------------------------------------------------------------------------------|

    -----BEGIN PGP SIGNED MESSAGE-----
    Hash: SHA1

    > From: Kayne Ian (Softlab) [mailto:Ian.Kayne@softlab.co.uk]
    >
    > A better way (for example) would be to write an app that
    > hooks kernel calls to load a process, then compare a checksum
    > of the process in question to a "whitelist" of allowed
    > application checksums - if a match is found, the call is
    > allowed. If not, the call is denied. Bear in mind that you
    > need to checksum the loaded process, not the exe file on disk
    > otherwise any packer (UPX etc) would effectively allow a bad
    > app to slip by. That somewhat raises the skill required to bypass
    > it.

    This is generally what ZoneAlarm, ZoneAlarm Pro, and Integrity
    products do. Other Endpoint Security/Desktop Firewall software
    do similar things as well.

    An md5-like hash of the application is saved (in a protected file)
    along with the network access permissions associated with that
    application.

    UPX cannot be used to defeat this*. If you have a malicious program
    that has a hash not on your whitelist, UPX-ing it isn't going to
    chance that.

    The most significant risk to this approach is people having bad
    policies
    about what is whitelisted or what whitelisted programs are permitted
    to do.

    *Ok, there is some really small possibility of a hash collision.

    - -John
    - --
    John LaCour
    Zone Labs Security Services

    -----BEGIN PGP SIGNATURE-----
    Version: PGP 8.0.2

    iQA/AwUBQAV5IaeZbSyAsADEEQKvjgCgkTQQlKJfK6BgkTdmBIY9ENd87UYAn0s2
    R+sEGGThZ/GckW+VBAReHj3L
    =+GpG
    -----END PGP SIGNATURE-----

    ---------------------------------------------------------------------------

    ---------------------------------------------------------------------------

    ---------------------------------------------------------------------------
    ---------------------------------------------------------------------------


  • Next message: Kayne Ian (Softlab): "RE: application whitelisting (was RE: Active Directory Question)"

    Relevant Pages

    • application whitelisting (was RE: Active Directory Question)
      ... > hooks kernel calls to load a process, then compare a checksum ... > app to slip by. ... An md5-like hash of the application is saved ... If you have a malicious program ...
      (Focus-Microsoft)
    • Re: About Hsiehs hash: initial value? 64 bit?
      ... when implementing Paul Hsieh's hash for incremental updates? ... Well if you just want to use it as a checksum, ... CRCs of course, have a much longer history as checksums. ...
      (comp.programming)
    • Re: Byte to byte compare, duplicate file finder/killer
      ... cryptographically strong hash, simply because the CRC ... By using a secure hash instead of CRC, the actual byte-to-byte compare ... checksum (files with identical CRCs or checksums are trivial to ...
      (comp.programming)
    • Re: Performance ideas - comparing first N chars of 2 TEXT columns
      ... You should never count on CHECKSUM or BINARY_CHECKSUM to return a unique ... CHECKSUM hash. ... > Both text fields actually contain XML, ... The XML does stretch beyond 8192 chars, ...
      (microsoft.public.sqlserver.programming)
    • Re: Any known fast hash algorithms ?
      ... What I meant with that the principle is not that different from CRC, ... different from hash and checksum has allways been a little bit abstract. ... Looking at your reply I would say that you call a function a hash when it ... still end up with the best possible distribution of those bits. ...
      (borland.public.delphi.language.basm)