Likes Likes:  0
Resultaten 1 tot 2 van de 2
Geen
  1. #1
    E.Kellinis
    IE Certificate Stealing (Phising) bug
    Gast
    n/a Berichten
    Berichten zijn liked



    Thread Starter

    IE Certificate Stealing (Phising) bug

    #########################################
    Application: Internet Explorer
    Vendors: http://www.microsoft.com
    Version: 6.0.2800
    Platforms: Windows
    Bug: Certificate Stealing (Phising)
    Risk: Medium
    Exploitation: Remote with browser
    Date: 30 Apr 2004
    Author: Emmanouel Kellinis
    e-mail: me@cipher(dot)org(dot)uk
    web: http://www.cipher.org.uk
    List : BugTraq(SecurityFocus)
    #########################################


    =======
    Product
    =======
    A popular Web browser, created by Microsoft,
    used to view pages on the World Wide Web.

    ===
    Bug
    ===

    In Internet Explorer enables someone to use an
    ssl certificate in a website which belongs to
    someone else.A combination Refresh and OnUnload
    on BODY tag. This bug can be used in
    Phising scams.



    Lets say that we want to use example.com
    certificate.


    We point with REFRESH Meta Tag that website.

    < meta http-equiv="REFRESH"
    content="0;url=https://www.example.com/"
    >


    Then inside our BODY tag we use onUnload to inform
    the webbrowser what to do when it will unload that
    webpage (using the window.location method).

    < BODY onUnload='window.location=""' >

    The result of that will be,
    the browser will inform us if we want to use the
    certificate of example.com (If we trust that party
    we will say yes)Then the contents of the protected
    webpage will be downloaded to our website using our
    domain name .

    We have something like that on the URL field of IE.
    Address : http://www.ourdomainname/FakeSSL.html
    then we have the contents of the index page of example.com
    in the browsing area and the SSL lock (right corner).

    *Remember that we are in the fake website all that time.


    If inside the index page links and forms use virtual
    pointers to directories or files
    (e.g. images/ or form/submit.php) we can use the trust
    of the visitor and steal information.
    (e.g. via submit forms).

    NOTE: the lock in the right corner doesnt work, if you
    click it says "this type of document does not have security
    certificate" which mean that this lock shouldnt be there ,
    in case that a visitor will check the certificate only in
    the the popup window at the begining of the session and after
    loading the webpage will not check the validity using the right
    corner lock, then we fake the certificate.

    =====================
    Proof Of Concept Code
    =====================

    < html>
    < head>
    < title>Your Page Title</title>
    < meta http-equiv="REFRESH"
    content="0;url=https://www.example.com/">

    < META HTTP-EQUIV="Content-Type" CONTENT="text/html;">

    < /HEAD>
    < BODY onUnload='window.location=""'>

    < /BODY>
    < /HTML>

    ====
    FIX
    ====
    Do not use virtual directories , instead use the real path or url
    Refresh access to the root directory



    ================================================== =======
    *PK:http://www.cipher.org.uk/files/pgp/c...public.key.txt
    ================================================== =======

  2. #2
    Michael Wojcik
    IE Certificate Stealing (Phising) bug
    Gast
    n/a Berichten
    Berichten zijn liked



    Thread Starter

    RE: IE Certificate Stealing (Phising) bug

    > From: E.Kellinis [mailto:me@cipher.org.uk]
    > Sent: Friday, April 30, 2004 11:09 AM
    >
    > If inside the index page links and forms use virtual
    > pointers to directories or files
    > (e.g. images/ or form/submit.php) we can use the trust
    > of the visitor and steal information.


    Those aren't called "virtual pointers to directories or files". They're
    "relative URLs". It's worth pointing this out because your recommended fix:

    > Do not use virtual directories , instead use the real path or url
    > Refresh access to the root directory


    doesn't make sense as written. "virtual directories" in this context
    commonly refers to URL path segments that are mapped by the HTTP server to
    filesystem entities with different names. Avoiding them doesn't mitigate
    this attack. What you want to recommend here (presumably - I haven't tested
    it myself) is using relative URLs as reference values in secure HTML pages.

    --
    Michael Wojcik
    Principal Software Systems Developer, Micro Focus

Webhostingtalk.nl

Contact

  • Rokin 113-115
  • 1012 KP, Amsterdam
  • Nederland
  • Contact
© Copyright 2001-2026 Webhostingtalk.nl.
Web Statistics