Hack.lu Logo

Hack.lu 2017 Wrap-Up Day 2

As said yesterday, the second day started very (too?) early… The winner of the first slot was Aaron Zauner who talked about pseudo-random numbers generators. The complete title of the talk was “Because ‘User Random’ isn’t everything: a deep dive into CSPRGNs in Operating Systems & Programming Languages”. He started with an overview of random numbers generators and why we need them. They are used almost everywhere even in the Bash shell where you can use ${RANDOM}.  CSPRNG is also known as RNG or “Random Number Generator”. It is implemented at operating system level via /dev/urandom on Linux on RtlGenRandom() on Windows but also in programming languages. And sometimes, with security issues like CVE-2017-11671 (GCC fails to generate incorrect code for RDRAND/RDSEED. /dev/random & /dev/urandom devices are using really old code! (fro mid-90’s). According to Aaron, it was a pure luck if no major incident arises in the last years. And today? Aaron explained what changed with the kernel 4.2. Then he switched to the different language and how they are implementing random numbers generators. He covered Ruby, Node.js and Erlang. All of them did not implement proper random number generators but he also explained what changed to improve this feature. I was a little bit afraid of the talk at 8AM but it was nice and easy to understand for a crypto talk.

The next talk was “Keynterceptor: Press any key to continue” by Niels van Dijkhuizen. Attacks via HID USB devices are not new. Niels reviewed a timeline with all the well-known attacks from 2005 with the KeyHost USB logger until 207 with the BashBunny. The main problems with those attacks: they need an unlocked computer, some social engineer skills and an Internet connection (most of the time). They are products to protect against these attacks. Basically, they act as a USB firewall: USBProxy, USBGuest, GoodDog, DuckHunt, etc. Those products are Windows tools, for Linux, have a look at GRSecurity. Then Niels explains how own solution which gets rid of all the previous constraints: his implants is inline between the keyboard and the host. It must also have notions of real)time. The rogue device clones itself as a classic HID device (“HP Elite USB Keyboard”) and also adds random delays to fake a real human typing on a keyboard. This allows bypassing the DuckHunt tool. Niels makes a demonstration of his tool. It comes with another device called the “Companion” which has a 3G/4G module that connects to the Keynterceptor via a 433Mhz connection. A nice demo was broadcasted and his devices were available during the coffee break. This is a very nice tool for red teams…

Then, Clement Rouault, Thomas Imbert presented a view into ALPC-RPC.The idea of the talk: how to abuse the UAC feature in Microsoft Windows.They were curious about this feature. How to trigger the UAC manually? Via RPC! A very nice tool to investigate RPC interface is RpcView. Then, they switched to ALPC: what is it and how does ir work. It is a client/server solution. Clients connect to a port and exchange messages that have two parts: the PORT_MESSAGE header and APLC_MESSAGE_ATTRIBUTES. They explained in details how they reverse-engineering the messages and, of course, they discovered vulnerabilities. They were able to build a full RPC client in Python and, with the help of fuzzing techniques, they found bugs: NULL dereference, out-of-bounds access, logic bugs, etc. Based on their research, one CVE was created: CVE-2017-11783.

After the coffee break, a very special talk was scheduled: “The untold stories of Hackers in Detention”. Two hackers came on stage to tell how they were arrested and put in jail. It was a very interesting talk. They explained their personal stories how they were arrested, how it happened (interviews, etc). Also gave some advice: How to behave in jail, what to do and not do, the administrative tasks, etc. This was not recorded and, to respect them, no further details will be provided.

The second keynote was assigned to Ange Albertini: “Infosec and failure”. Ange’s presentation are always a good surprise. You never know how he will design his slides.As he said, his talk is not about “funny” failures. Infosec is typically about winning. The keynote was a suite of facts that prove us that we usually fail to provide good infosec services and pieces of advice, also in the way we communicate to other people. Ange likes retro-gaming and made several comparisons between the gaming and infosec industries. According to him, we should have some retropwning events to play and learn from old exploits. According to Ange, an Infosec crash is coming like the video game industry in 1983 and a new cycle is coming. It was a great keynote with plenty of real facts that we should take care of! Lean, improve, share, don’t be shy, be proactive.

After the lunch, I skipped the second session of lightning talks and got back for “Sigma – Generic Signatures for Log Events” by Thomas Patzke. Let’s talk with logs… When the talk started, my first feeling was “What? Another talk about logs?” but, in fact, it was interesting. The idea behind Sigma is that everybody is looking for a nice way to detect threats but all solutions have different features and syntax. Some example of threats are:

  • Authentication and accounts (large amount of failed logins, lateral movement, etc.)
  • Process execution (exec from an unusual location, unknown process relationship, evil hashes, etc…
  • Windows events

The problem we are facing: there is a lack of standardised format. Here comes Sigma. The goal of this tool is to write use case in YAML files that contain all the details to detect a security issue. Thomas gave some examples like detecting Mimikatz or webshells.

Sigma comes with a generator tool that can generate queries for multiple tools: Splunk, Elasticsearch or Logpoint. This is more complex than expected because field names are different, there are inconsistent file names, etc. In my opinion, Sigma could be useful to write use cases in total independence of any SIEM solution. It is still an ongoing project and, good news, recent versions of MISP can integrate Sigma. A field has been added and a tool exists to generate Sigma rules from MISP data.

The next talk was “SMT Solvers in the IT Security – deobfuscating binary code with logic” by Thaís Moreira Hamasaki. She started with an introduction to CLP or “Constraint Logic Programming”. Applications in infosec can be useful like malware de-obfuscation. Thais explained how to perform malware analysis using CLP. I did not follow more about this talk that was too theoretical for me.

Then, we came back to more practical stuff with Omar Eissa who presented “Network Automation is not your Safe Haven: Protocol Analysis and Vulnerabilities of Autonomic Network”. Omar is working for ERNW and they always provide good content. This time they tested the protocol used by Cisco to provision new routers. The goal is to make a router ready for use in a few minutes without any configuration: the Cisco Autonomic network. It’s a proprietary protocol developed by Cisco. Omar explained how this protocol is working and then how to abuse it. They found several vulnerabilities

  • CVE-2017-6664: There is no way to protect against malicious nodes within the network
  • CVE-2017-6665 : Possible to reset of the secure channel
  • CVE-2017-3849: registrar crash
  • CVE-2017-3850: DeathKiss – crash with 1 IPv6 packet
The talk had many demos that demonstrated the vulnerabilities above. A very nice talk.

The next speaker was Frank Denis who presented “API design for cryptography”. The idea of the talk started with a simple Google query: “How to encrypt stuff in C”. Frank found plenty of awful replies with many examples that you should never use. Crypto is hard to design but also hard to use. He reviewed several issues in the current crypto libraries then presented libhydrogen which is a library developed to solve all the issues introduced by the other libraries. Crypto is not easy to use and developer don’t read the documentation, they just expect some pieces of code that they can copy/paste. The library presented by Frank is called libhyrogen. You can find the source code here.

Then, Okhin came on stage to give an overview of the encryption VS the law in France. The title of his talk was “WTFRance”. He explained the status of the French law against encryption and tools. Basically, many political people would like to get rid of encryption to better fight crime. It was interesting to learn that France leads the fight against crypto and then push ideas at EU level. Note that he also mentioned several politician names that are “against” encryption.

The next talk was my preferred for this second day: “In Soviet Russia, Vulnerability Finds You” presented by Inbar Raz. Inbar is a regular speaker at hack.lu and proposes always entertaining presentations! This time he came with several examples of interesting he found “by mistake”. Indeed, sometimes, you find interesting stuff by accident. Inbar game several examples like an issue on a taxi reservation website, the security of an airport in Poland or fighting against bots via the Tinder application. For each example, a status was given. It’s sad to see that some of them were unresolved for a while! An excellent talk, I like it!

The last slot was assigned to Jelena Milosevic. Jelena is a nurse but she has also a passion for infosec. Based on her job, she learns interesting stuff from healthcare environments. Her talk was a compilation of mistakes, facts and advice for hospitals and health-related services. We all know that those environments are usually very badly protected. It was, once again, proven by Jelena.

The day ended with the social event and the classic Powerpoint karaoke. Tomorrow, it will start again at 08AM with a very interesting talk…

Leave a Reply

Your email address will not be published. Required fields are marked *

This site uses Akismet to reduce spam. Learn how your comment data is processed.