BruCON 2015

(My) BruCON 2015 notes (5)

Adrian Citu

Here are my quick notes from the BruCON 2015 conference. All the slides can be found here.

This is the last ticket about the BruCOn 2015 so, it contains the presentations for which my notes are not so good 🙂

Creating REAL Threat Intelligence with EvernotebruCon

goal:

  • experiment to generate threat intelligence with Evernote.
  • use Evernote as intelligence repository.

background:

  • before buying new commercial solution
    • try quick and dirty solution in house
    • invest in people &process first, then Products.

Use Evernote as a GTD(Getting Things Done ??)-Based Task Mgmt System.

  • treat  Evernote like  a  Database
    • Notebook  ==  Table
    • Note  ==  Free  Form  Record
    • Nested  Notebooks
    • Hierarchical  Tags

Looking Forward – Finding the right balance for INFOSEC

Some words about the infosec in the past and today:

  • in 1999 the security community was a small community
  • things start to change now BUT
    the ratios security/it people it’s very low

Some words about the security breaches:

  • 99% of the breaches is are due to basic things; BUT the companies are focusing on much complicated
    attacks but are forgetting the easy to fill gaps.
  • end-users are still the weakest link; for fishing only one click it’s enough to get in inside the it infrastructure
  • profiling is difficult BUT targeting the user is muck more easier.

Some words about the security industry:

  • security industry is too technology centric.
  • we just hope that the technology will solve the problems magically.
  • technology it takes over talent.

Some ideas for the defence:

  • Disabling local administrator accounts, or randomizing.
  • Rotating domain admin account passwords.
  • Disallow PowerShell execution for normal users.
  • Disallowing executables to be run through TEMP and other directories.
  • Network segmentation of user workstations.
  • Focus on detection capabilities over anything.

I am the Cavalry

The Cavalry is a organization that is focused on issues where computer security intersect public safety and human life. The areas of focus for The Cavalry are medical devices, automobiles, home electronics and public infrastructure.

How to influence people (a pen test like approach):

  • recon
  • empathizing (replaces finding vulnerabilities)
  • enabling changes (replaces exploitation)

Recon

known the official structure and the non-official one.

unofficial structure

  • who is liked
  • trusted influencers

Empathizing

  • understand the stakeholders
  • studies can give you a hint about the way of thinking
  • motivation, career ambitions
  • how the stakeholders make decisions
  • learn how to speak the stakeholders language – cross domain issues

Enabling changes

  • work the system;
  • be adaptive if it didn’t worked
  • riding waves, news
  • speak their language

(My) BruCON 2015 notes (4)

Adrian Citu

Here are my quick notes from the BruCON 2015 conference. All the slides can be found here.

The malware is just code so, as any other code it is possible (in theory) to analyze/reverse engineer it manually.

The triage is one of the functions of the incident response program and must answer the following three questions regarded to a specific input:

  • is the input malicious ?
  • if yes, what is exploiting ?
  • are we exposed ?

Triage is not malware analysis and should be quick and efficient. The triage workflow:

  • passive analysis.
  • first interaction and download.
    • some malware are crafted to be able to interact with the initial URL only limited number of times
    • some malware could profile your browser, check the browser version, platform, or use the user agent script to decide if the exploit can be executed or not.
    • some tools:useragentstring.com (to check your user agent), onlinecurl.com (on-line version of curl, copy paste a url and you get back the response), hurl.it (idem as previous one).
  • web component analysis.
    • once you have the web component (which is typically an html page + JavaScript) you could try to analyze it.
    • use jsBeautify.org to try to have something human readable in case the code is obfuscated.
    • try to use the browser debugger, eventually change JavaScript eval expressions.
  • exploit analysis.
    • can use showmycode.com to understand the exploit; it is capable to decompile Java, Flash, .Net, PHP
    • sometimes you can blindly search the metasploit exploit template library
  • payload extraction.
  • payload analysis.
    • can submit the file/s to VirusTotal or malwr (virtualized Cuckoo instances).
    • malwr can give you infos about the registry keys created, network traffic.
  • build IOCs (Indication Of Compromise).
    • collection of indicators which can be used to describe a compromised system.

This was an workshop, so the participants had to play with some of the tools. Here is the quick workflow that i followed:

start from a url -> use the onlinecurl.com to get the response (initial interaction) -> saved the response on a file and used to browser debugger to understand what the component is doing (web component analysis)-> get from the JavaScript another url that contains a link to a Java .class file -> use it showmycode.com to decompile the class file (exploit analysis)-> write some Java code to decode parts of the exploit and execute it on ideone.com (payload extraction)-> …time over :(.

(My) BruCON 2015 notes (3)

Adrian Citu

Here are my quick notes from the BruCON 2015 conference. All the slides can be found here.

Intrusion detection on Linux and OSX with osquery

osquery goals:

  • explore your operation system using sqlbruCon
  • host visibility is motivated by intrusion detection
  • wanted 1 binary; 6MB with no dependencies
  • done in C++
  • created and used by facebook (installed on 60 000 osx)

Same osquery query examples:

  • select pid, name, uid from processes where [constraints]
  • the sql syntax is the SQL lite  syntax.
  • no Windows version; linux and osx only.
  • a lot of cool examples: the list of events that happened on the host (files changed, usb sticks inserted).

A second binary that is presented is called osqueryd. osqueryd is a daemon; it uses a json-config config file to set options and define a schedule. The goal of osqueryd is to execute queries peridically and then send the result to an ELK (ElasticSearch+LogStash+Kibana) system.

The queries used by Facebook are packaged in different files; these packages can be also downloaded free of charge.

Kernel Tales: Security Testing of aarch64 Android Kernels

A kernel contains code from various sources; hardware manufacturers, software companies (google, samsung).
An Android kernel is almost a Linux kerner, so the attack surface is through syscalls;
The way to corrupt the kernel: the user space tries to write in the kernel space by using syscalls or even better use the ioctls to replace the syscall.

The overall idea is to fuzzy the parameters passed to ioctls. The fuzzer used (dronity) is based on the trinity fuzzer; dronity is not yet available (it will be soon on github).

The demo contained the following items:

  • dronity fuzzing a Android development board.
  • dronity fuzzing a real Android tablet.
  • dronity fuzzing an aarch64 kernel on qemu.
  • dronity fuzzing an aarch64 kernel with gdb attached (on qemu) in order to intercept the exception thrown by the kernel.

(My) BruCON 2015 notes (2)

Adrian Citu

Here are my quick notes from the BruCON 2015 conference.All the slides can be found here.

cve-search a free software to collect search and analyze common vulnerabilities and exposures in software

Some of the goals of the tool: bruCon

  • do vulnerability search on off-line local search.
  • fast lookup of vulnerabilities on different criteria.
  • allow localized classification of vulnerabilities; localization geographically or from the business perspective.
  •  build new tools based on local database of software and hardware vulnerabilities.

Some of the components/features of the tool:

  • db_updater.py
    • the goal of the script is to fetch vulnerabilities from different data-sources.
    • the data sources used are: NIST NVD, D2SEC, Microsoft Bulletin, vFeed.
    • the scrript can be extended to easily fetch other sources.
  • MongoDB
    • is the DB used to store the info.
  • search.py
    • used to search for vulnerabilities in the DB.
  • web interface
    • you can see, search
    • color scheme for vulnerabilities
  • you can use your own tagging system to weight the critical software/vendors in your constituency.
  • you can use statistics using external tools/languages like R
ex: searh.py -p oracle.....| jq -r '.cvss' | Rscript .........

Problems that face the application:

  • some vendors do not publish vulnerabilities information in a parsable way.
  • some vendors do not support the CPE naming convention.

Software using cve-search:

  • CVE-Portal
  • CVE-Scan
  • NorthernSec Vulnerability-Management (still under development)

Roadmap and future:

  • add vulnerabilities data sources from software and hardware vendors
  • expand cve-search to include vulnerabilities whitout CVE assignment.

(My) BruCON 2015 notes (1)

Adrian Citu

Here are my quick notes from the BruCON 2015 conference.All the slides can be found here.

Nightmares of a Pentester bruCon

The goal is to present some (basic) that the operational security teams (a.k.a Blue Team) can use in order to make the life harder to the pen-testers.

Some numbers about the security breaches :

  • Verizon report: Quantify the impact of a data breach with new data from the 2015 DBIR
  • top 3 industries that are under attack: public sector, it sector, financial services
  • 70% of the attacks are targeting a second victim; so the first victim is a step stone to reach the target
  • 90% off the incidents have as root cause the peoples (the weakest link).

Good security programs are built in and not bolt on.

External defenses tips and tricks:

Don’t talk to strangers:

  • implement blocks from all the emerging cyber-threats lists.
  • reject specific user agent strings.
  • ban the port scanning.

If you are going to talk be sure to know who is it

  • disable smtp/verify/validation.
  • analyze the certificates and (external) domain age.
  • use SPF for the email validation
  • use DNS analysis
  • don’t forward DNS

Internal defenses tips and tricks:

Your internal network is an hostile environment; treat it as such

  • monitor inside more than outside.
  • segmentation of all servers from users.
  • never use VPN pools; always tie a user to a specific IP address.
  • remove your network default route.
  • intercept all http/s requests.

Users have the ability to use the company resources

  • white-list the approved and managed software.
  • disallow local admin privileges.
  • users should only be allowed to go to categorized sites; any other traffic must be denied.
  • host based firewals, ids and behavioral analysis.
  • scans all hosts for vulnerabilities on a regular basis.
  • randomize all the local admin passwords.

Servers have specific purposes

  • do not install workstation software.
  • manage updates centrally.
  • segment the servers.
  • standards images should have no additional services installed.
  • do not allow the use of local account to log in remotely.