Re: Systems compromised with ShellBOT perl script - part 2

From: Harry de Grote (rik.bobbaers_at_cc.kuleuven.ac.be)
Date: 10/20/04

  • Next message: Stephen J. Smoogen: "Re: Systems compromised with ShellBOT perl script - part 2"
    To: incidents@securityfocus.com
    Date: Wed, 20 Oct 2004 17:59:57 +0200
    
    

    Op Wednesday 20 October 2004 07:04, security@kemhosting.com sgreifde:
    > This thread is a couple months old, but I'm having issues with this hack,
    > found it in the archives and thought it'd be helpful if I 'resusitated' it.
    > See bottom of email for rest of thread.
    >
    > Today, hackers used the ShellBOT perl script to bring down Apache and start
    > up their IRC listener. They (somehow) copied it into /tmp and executed it.
    > This confuses me because I have my /tmp directory mounted
    > rw,noexec,nosuid. Does Perl somehow bypass this?

    try doing this in your no-exec /rmp: /lib/ld-linux.so.2 /bin/bash
    (should work if you have a 2.4 kernel, not in 2.6 anymore)

    thats just 1 way to bypass the noexec flag

    > While the script was running, I ran lsof and found that it had recursively
    > accessed all my (virtual host) httpd logs (probably in an attempt to delete
    > it's tracks = the reason I can't see how they copied the script into /tmp)
    > which are owned by root. this is also confusing since the process the
    > script spawned was owned by user apache.
    >
    > Some info on my box:
    > Redhat ES kernel 2.4.21-9.0.1.ELsmp
    > httpd-2.0.46-32.ent
    > php-4.3.2-11.ent
    >
    > Anyone have any ideas on how this can happen? Mainly the executing of a
    > script on a noexec mount! Obviously I'm not a guru, so it's probably
    > something simple - so please, share!

    there are , as you can see easy ways to bypass that... :)

    -- 
    harry
    aka Rik Bobbaers
    K.U.Leuven - LUDIT             -=- Tel: +32 485 52 71 50
    Rik.Bobbaers@cc.kuleuven.ac.be -=- http://harry.ulyssis.org
    "\x41\x20\x63\x6f\x6d\x70\x75\x74\x65\x72\x20\x77\x69\x74\x68\x6f\x75\x74\x20"
    "\x57\x69\x6e\x64\x6f\x77\x73\x20\x69\x73\x20\x6c\x69\x6b\x65\x20\x61\x20\x66"
    "\x69\x73\x68\x20\x77\x69\x74\x68\x6f\x75\x74\x20\x61\x20\x62\x69\x63\x79\x63"
    "\x6c\x65\x0a\x00"
    

  • Next message: Stephen J. Smoogen: "Re: Systems compromised with ShellBOT perl script - part 2"

    Relevant Pages

    • Re: cx88 totally fried in 2.6.15-rcX -was- Re: HD3000 - no NTSC via tuner
      ... On Monday 28 November 2005 20:17, Michael Krufky wrote: ... as it would apply to a working 2.6.14.x kernel. ... that I comment/uncomment stuff in the buildit (thats another script I ... message by Gene Heskett are: ...
      (Linux-Kernel)
    • Re: 2.6.22-rc5 regression
      ... Bisecting: 128 revisions left to test after this ... of the fact that I use the same directory tree to build the kernel ... [torvalds@woody linux]$ git bisect bad v2.6.22-rc5 ... -# Simple script to generate a deb package for a Linux kernel. ...
      (Linux-Kernel)
    • Re: When was /dev/cua* depreciated?
      ... but that first script edits startup scripts itself! ... Actually the kernel module bit is the best quality by far, ... assumed that any shipped software will be entirely useless and must be ...
      (uk.comp.os.linux)
    • Re: [TOOL] kprobestest : Kprobe stress test tool
      ... This tool list up all symbols in the kernel via /proc/kallsyms, ... Finally, the script sorts all 'passed' symbols into 'tested', 'untested', ... each culprit in these lists. ... Hardware name: Deskpro EN Series ...
      (Linux-Kernel)
    • Re: **Caution-External**: Re: USB hotplug woes
      ... script before it has actually done any of the volume discovery. ... The goal is that if someone where to stick in an USB flash drive, ... Further more I tried the 2.6 kernel and the machine just went nuts only ... Do you want the script to autorun whe the usb device is pluged in? ...
      (Debian-User)