Insecure Password Authentication in Yahoo! Messenger

From: vuln-dev (vuln-devat_private)
Date: Wed Nov 21 2001 - 19:11:12 PST

  • Next message: 3APA3A: "Re: Information Leak Bug Discovered in Netscape Mail!"

    Hi fellow users! 
    
    The members of GOBBLES Research have discovered a flaw in the authentication 
    protocol for Yahoo! Messenger.  Attached to this letter is a full disclosure 
    advisory + discussion of the vulnerability.  At this time there is no "fix" 
    available. 
    
    Thanks for all your support! 
    
    The GOBBLES Research Team
    http://www.bugtraq.org
    
    ++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++ 
    ++++++++++++++GOBBLES+SECURITY+RESEARCH+TEAM+INCORPORATED+++++++++++++++++ 
            ALERT! ALERT! YAHOO MESSENGER REPLAY ATTACK! ALERT! ALERT! 
    ++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++ 
    
    "Overt control... always has a finite life, because in the end there will be
    a challenge and rebellion against it. Covert control... can go on forever,
    because you don't rebel against something you don't know exists."
    
    oooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooo
    
    PRODUCT
    *******
    
    program: Yahoo! Messenger YMSG protocol
    website: http://messenger.yahoo.com/
    
    BACKGROUND
    **********
    
    GOBBLES was sitting down in lab with Yahoo! Messenger client running
    chatting with beautiful Susie who on he friends list on Yahoo. GOBBLES
    trying finish read Bruce Schneier _Applied Cryptography_ but Susie keep
    buzzing like crazy about music and pets (she say she has nice pussy) and
    then suddenly she say she writing some SPARC assembly codes and GOBBLES just
    go crazy hehe. Anyway after GOBBLES wake up to see she still talk about
    silly gossip, he click 'ignore' button and begin determined to break
    Messenger protocol with he new crypto knowledge.
    
    GOBBLES discovered very inane move made by Yahoo engineers in apparent
    effort to stop cleartext transmission of password during initial login
    packet exchange. Without full disclosure this hole not brought to public
    attention and make Yahoo more secure, so how can idiot say that full
    disclosure only secure a job?
    
    It wrong saying that only people who defend full disclosure are they who
    capitalize on it because not like GOBBLES acquire sports car working at
    GOBBLES BUGTRAQ SECURITY INC. When aleph1 leave r00t hacking group to be
    security professional he not profit like everyone say, so why everyone have
    to give these people hard time if they become security professional and
    contribute earth-shattering vulnerability like GOBBLES does in order to
    improve security and liberate humanity from darkside malicious cracker out
    there in wild?
    
    Revealing symlink vulnerability in Bruce Piano Tuner v0.7 help admin protect
    system from cracker out there and make cracker lose root. GOBBLES owes
    securityfocus website great deal because they provide helpful information
    that help me with cracking incident last year. This cracker he come on one
    of my system using team bugtraq penetrator exploit in bad way because
    GOBBLES was asleep and couldn't apply patch before cracker gank it from
    bugtraq. Anyway, cracker not do any damage, but he thought it his base. With
    patch information from securityfocus, GOBBLES was able fix security hole and
    lock him out :P
    
    TECHNICAL DETAILS
    *****************
    
    GOBBLES look at Dug Song dsniff msgsnarf.c and observe that YMSG application
    header have following format:
    
    struct ymsg {
        unsigned char version[8];
        unsigned short length;
        unsigned short type;
        unsigned char unknown1[4];
        unsigned char unknown2[4];
    }
    
    GOBBLES look at libyahoo code from sourceforge and observe:
    
    /* Service constants */
    #define YAHOO_SERVICE_LOGON             1
    
    GOBBLES doing this on FreeBSD machine so he get patch for like...
    
    FreeBSD-SA-01:48                                           Security Advisory
                                                                    FreeBSD,
    Inc.
    
    Topic:          tcpdump contains remote buffer overflow
    
    Category:       core
    Module:         tcpdump
    Announced:      2001-07-17
    Credits:        Nick Cleaton <nickat_private>
    Affects:        All releases of FreeBSD 4.x prior to 4.4,
                    FreeBSD 4.3-STABLE prior to the correction date
                    FreeBSD 3.x is unaffected.
    Corrected:      2001-07-09
    Vendor status:  Patch released
    FreeBSD only:   NO
    
    ... this so while he run tcpdump next he won't be compromised by darkside
    malicious cracker out there in wild or penetrator typing in wrong host to
    Vetescan stealth vulnerability assessment tool.
    
    GOBBLES then run tcpdump and using YMSG structure above as reference he find
    first packet with service type constant 0x01. This LOGIN packet. GOBBLES
    notice it use MCF/MD5 encryption of user password with crypt(3) salt like
    $1$_2S43d5f$. Encrypted password is sent over the wire in LOGIN packet. User
    is immediately authenticated. 
    
    MAJOR VULNERABILITY #1 IS THAT ATTACKER MAY SNIFF THIS AND BRUTE FORCE USER
    PASSWORD AKIN TO RECENT TEAM BUGTRAQ POP3 'APOP PASSWORDS AT RISK'
    ADVISORY:
    
    http://archives.neohapsis.com/archives/bugtraq/2001-07/0163.html
    
    MAJOR VULNERABILITY #2 IS THAT ATTACKER WHO CAPTURES LOGIN PACKET CAN USE
    SIMPLE REPLAY ATTACK WITH PASSWORD HASH INJECTED IN PACKET SENT BY CUSTOM
    CLIENT. 
    
    So only defense afforded by this dumb scheme over sending password
    unencrypted is protection against password reuse attack:
    
    "Sociological Ghasche Tree Model And Its 
     Ramifications Concerning Password Reuse 
     On Worldwide Computer Networks - USENIX 1999"  
    
    http://www.sec-lab.cs.qwe.edu/professors/Floyd/whitepapers/1999/reuse.pdf
    
    FIX
    ***
    
    GOBBLES suggest Yahoo engineer use brain and come up with challenge-response
    mechanism. They set bad example for commercial security research community.
    They should pull finger out of bum and focus on protecting user instead of
    just pulling in money from job. Here $0.2 -- go buy clue.
    
    DEMONSTRATION
    *************
    
    As you see above, easy to observe packets with tcpdump. Ethereal not seem to
    have YMSG protocol support yet, so SANS professional at sans.org might need
    make $50,000/hour other way by using different tool for analysis instead of
    ethereal. Come with most operating system is tcpdump or snoop. Can also get
    sweet tool ngrep from route[GUARDENT] site:
    
    http://www.packetfactory.net/
    
    Or maybe you be creative and roll your own with Michael Schiffman libnet
    socket wrapper. It easy reproduce, no hurry, just take your time and see
    what you come up with.
    
    GREETS
    ******
    
    dianora / evil angelica, tsk, snow, carolyn meinel, john vranesevich, steve
    gibson, kimble, knightmare, emmanuel goldstein, box.sk, @stake,
    securityfocus, sans.org, blackhat.com, defcon.org, 2600.com, #phrack@efnet,
    #hackphreak@undernet, bugtraq (thanks aleph1 and david ahmad for devoting
    your time to a great list), ntbugtraq (russel the love muscle ;D), cert.org,
    paul vixie, vesselin bontchev, reese witherspoon, kirsten dunst, katie
    holmes, aleister crowley, manly p hall, franz bardon, dennis ritchie,
    nietzsche, w. richard stevens, radiohead, george michael, larry wall,
    beethoven, francis bacon, bruce willis, bruce schneier, alan turing, john
    von neumann, donald knuth, michael abrash, robert sedgewick, and all our
    friends and family.
    
    
    GOBBLES SECURITY
    http://www.bugtraq.org/
    



    This archive was generated by hypermail 2b30 : Wed Nov 21 2001 - 19:41:50 PST