Thursday, May 19, 2011

CEIC 2011 Recap

After leaving a cold and rainy 50 degrees and arriving in Orlando to a warm, sunny 80 degrees, I was immediately in a better mood. The Royal Pacific venue is awesome. It's located at Universal Studios, has nice rooms and great restaurants. Registration was quick and painless with no long DefCon style lines. I was surprised a bit though that 1100 people were here as I thought the con would be a little smaller. However it doesn't feel as crowded as some others I've been to. They did mention that the amount of attendees has doubled since 2009.

I first attended an Encase Forensic v7 Preview workshop to outline what is being released in June. They have FINALLY added true multi-core, multi-threading to take advantage of good hardware. Some highlights include all modules like the ProTools Suite are now included in the base product and more noteworthy native processing for iOS, RIM, Android, and WinPhone6. There is also a new evidence format (EX01) and shiny new frontend for opening cases and adding evidence. The new image format also supports AES256 encryption, so the use of encrypted hard drives may be a thing of the past. The case processor has now been integrated and allows for templates to be created scripting much of what you want to preprocess like mounting compound files. They are also breaking out a new product called Evidence Processor which will allow you to distribute the load to multiple machines and merge them back into a single case. Overall it looks to be a winner and should help them compete better against FTK.

Next I attended "Memory Analysis and Malware Triage" by David Nardoni and two guys from General Dynamics. This was a pretty basic presentation probably more worthwhile 3 years ago or to someone who had never done memory analysis. It included a lab using Memoryze analyzing an rbot sample. They covered the key indicators to look for in a memory capture and what they can reveal. They also mentioned a tool called FingerPrint from HBGary.

To wrap up Day 1, I attended “What’s new in Windows Forensics” by John Marsh. This was mostly a review of what has been out for awhile now. Since I don't examine Win7 and 2008 systems on a regular basis, most of it isn't applicable for me until it becomes more mainstream in the corporate environment. There was the usual stuff on mining UsrJrnl and TxF transaction journals. He also mentioned that last access times are disabled by default now. The most interesting segment was on the registry. You now have to check for two different registries based on privilege level to capture all the details (UsrClass.dat). There are also transaction logs for the registry which will be huge for malware investigations. Some of the attendees from the UK also talked about a Woanware which has some nice tools. Finally we covered mounting and sharing out the volume shadow copy using vss admin. VSS makes a restore point every 7 days, prior to patching, and whenever it installs an unsigned application.

This was followed by a nice welcome reception by Guidance at the lagoon with food and drinks.

Day number 2 started off with a great keynote by Eric O’Neil. As a fan of the movie Breach, I was thrilled to see this talk. He talked a lot about his experiences busting Robert Hansen, which was awesome to hear first hand anecdotal stories. He mentioned about how scared he really was when he stole Hansen's palm pilot and had to sprint back to the room because forensics took too long imaging it along with the memory card and he couldn't figure out which bag pocket he took it out of originally. He also said the best thing he ever learned from Hansen was on day one. He said the spy is always in the worst position. This means he is always the one who will suffer the consequences if caught and constantly looking over his shoulder. O’Neil also believed that while Hansen may have started spying originally to make money for his family, he wasn't greedy and told the Russians to stop giving him so much money and keep it under 10K per drop. He ultimately thought Hansen kept spying because the Russians made him feel like he mattered and was a success while he was loathed by his peers at the FBI. Aside from Hansen, he also touched on some other interesting topics. He said while on travel he always puts up the do not disturb sign on his room and then sets traps in the room. Of course, some one always enters his room looking for "stuff". He also covered some common things he is seeing on the many corporate espionage cases he has worked. From dumpster diving and posing as contracted shredding companies to well-placed interns and phony shell companies the environment is ruthless. He reiterated an idea all should be familiar with. If you have something cool, somebody wants to steal it.

Next I attended "Android? Encase Does.." by Andy Spruill. I liked this lab because we got to walk though analyzing evidence files from a Sprint Evo 4g. So Android is leveraging YAFFS with a FAT formatted sdcard typically. Google has pushed hard for developers to always write their application data to the sdcard, however this isn't always the case. The two options for acquiring included rooting the phone and usb to usb debugging. The former allows you to see much more of the file system, however its way more intrusive. The main location for application data is in /Android/data. You should always process the sdcard first as many tools accessing the built-in flash will modify the timestamps on the sdcard. Once processed you can export location data to a KML file and view it in Google Earth for an awesome tracking visual. Its good to become familiar with SQLite and SQLiteBrowser as all the applications use it. Also of interest, the navigation app records the turn by turn direction as wav files that can be retrieved to show where the target may have been driving at a certain time. Spruill suggest that your practice rooting Android phones as it is quickly becoming an essential skill.

From there I listened to Rob Lee's Super Timeline presentation. The session basically walked through building an accurate timeline using SIFT tools (regtime, fls, log2timeline, etc). He noted that FAT will stay in local time regardless of what time zone you are in. He also mentioned that NTFS will keep from 8-12 timestamps (STDInfo, FNInfo, SFNInfo). MFT Examiner a tool from the UK. He also mentioned that while it still has lots of legitimate hits, looking for all zeroes in the nanoseconds field is a decent indicator of timestomping manipulation.

At this point, I couldn't fathom sitting through an Enscripting 101 session, so I got on the waitlist for “Revealing Intent with Windows 7 Artifacts” by Alissa Torres from Northrop Grumman. She was a great presenter and had people engaged the whole time with her HappyCubes. There are two types of Win7 jumplists: AutoDestination for Users and CustomDestination for Apps. You can mount the compressed files to gain further details. The .search-ms connectors have lots of metadata and can be exported as xml to find more user activity. Federated searches(.osdx) is also a new feature in Win7 that allows you to search a bunch of predefined sources including websites and network shares. Libraries are another artifact which is groups of files from different locations kept in a single container. StickyNotes also can contain some user attribution. TZworks and Nirsoft provide good shell bag parsers. Yaru is a nice tool for finding deleted regkeys. If you delete a directory in Win7, there will only be an $I file, not an $R file. DMThumbs is a good parser for the new thumbs.db format in Win7.

After this there was a cool happy hour in the Exhibit hall followed by a great party by Mandiant at the Wantilan Luau. They gave out t-shirts and had an awesome open bar.

I kicked off Day 3 with Simon Key's presentation “File Identification and Recovery Using Block-Based Hash Analysis”. I must confess I was not properly caffeinated so it took me awhile to get into this. I originally learned about this about 3 years ago when attending training by Guidance. Simon has made tremendous improvements in the quality and usability of the enscript. It’s help function has a nice explanation of how to use it. First it’s a good idea to close all your mounted compound files as that may give you errors when running the enscript. If you are doing multiple files always use the hash list. He also mentioned a common mistake is to think it’s found parts of your file when it's only sectors of all x00s or xFFs. The intelligent tail analysis function does take a lot of time, but it helps you when the last block of your file is missing and you don’t want to keep hashing the same block over and over. Simon walked us through 3 different demos which were great. VLC actually played a partial recovery with only 8% of the sectors. He also showed me a new feature of his enscript that he calls Block-Based File Identification aka FuzzyHashing aka ssdeep. If you know the structure of your file, for example Word docs are set up in 64byte blocks, you can find varied versions of the file. Make sure to check process all data with current files, so you don't waste time on deleted data. Overall I enjoyed this one quite a bit, it was a nice refresher to the subject.

Next I skipped out on the Mock Court Trial presentation to get into Rob Lee's session on web browser analysis. I'm glad I did, it was packed as usual. There was a heavy focus on the new stuff in IE8/9 and then at the very end on Firefox. Most of the material is stuff from his SANS 408 course. First off, when you see file:// in the index.dat file, it doesn't necessarily mean it was opened in the browser, but more likely through local file clicking. There is no such thing as last-modified in the index.dat so if you see that they don't match is most likely means your tool isn't functioning correctly. WebHistorian and NetAnalysis have been updated to fix this. DOM storage is a great place to look as most of the app preferences are stored here. Session recovery also has some very good evidence like clear text passwords; however there isn't a lot of automated parsing yet. MiTech makes his recommended Structured Storage Viewer. Suggestsites.dat also can give you a clue as to what the suspect was doing even if they have cleared their browsing history. Internet Evidence Finder(IEF) is another favorite tool for carving evidence out of memory and disk, however the timestamps aren't found for memory. It's worth noting that the pagefile can often move artifacts back into ram after reboots. The infamous Chewbacca defense is often used to debunk evidence by saying they have to prove something isn't possible. Flash Cookies have become huge over the last few years as they don't expire, are browser independent, and aren't cleared automatically. Rob recommended reading the WSJ article series on web privacy. A nice trick for recovering files is to make a file with the same name and in the exact location and then use the recover last version function to restore from VSS. The Firefox sessionstore.js is in clear text and you can use FirefoxSessionStoreExtractor from woanware to parse. He believes the privacy mode in Firefox is superior to IE as it overwrites instead of deleting. Another cool indicator is when an exactly an hour of history is missing showing they used the clear last hour option.

After a steak lunch, most of the attendees were in a food coma for Litchfield's session “Database Breach Investigations Made Practical”. He is really one of the few who are creating DB specific tools in this sector which is awesome. Apparently he doesn't have to work anymore either since he sold his company back in 2007, but he is still giving back to the community. He said that Oracle is harder to triage because there are better native tools for MSSQL and MySQL. He started off by outlining all the different db artifacts that can be used. He mentioned that if you ever see the Java Wrapper Class in the DB ObjCode, which is typically in ram, this is a sure sign of an intrusion because the code isn't used anymore. LogMiner was a tool he recommended. BlockSize for Oracle is 8192, ID 10 is the user table, and ID 18 is the obj table. Also checkout databasesecurity.com. Some of his standalone cmd line tools include filter, dumpaction, and orablock. Another GUI tool is DataBlockExaminer for Oracle, which will show you deleted rows in red.

To wrap up Day 3, I went to "The Art of Mobile device Malware and How to Detect and Defend Against it" by Roy Hu. Who knew but apparently Accenture has some good talent in the mobile security space. They are seeing quite a bit of non-targeted information stealers and banking targeted malware. They also have found that while remote wipe is recommended it often leaves artifacts behind on the phone. They expect to see Near Field Communication (NFC) take off more in the US the way it has in Asia and the EU. They mentioned briefly that the mobile variant of Zeus dubbed Zitmo. The second half of the presentation was a techincal dive on DroidDream, given that name because it was only active a night when the owner of the phone was most likely asleep and charging their phone. DroidDream used XOR encryption and leveraged Exploid for 2.1 or less and RageAgainstTheCage for 2.2 or less to root the phone. This has all been patched in Android 2.3. They mentioned that you should use Lookout AV for your Android phone, however there are also trojaned clones of it so beware. Also, Lookout had a nice presentation at DefCon18 which is recommended. One of their favorite Mobile Device Management (MDM) Tools is by Good Technology because it actually uses its own encryption and separates out corporate data and personal data. I spoke with them afterwards and said there wasn't any good anomaly detection today for malware on cell phones that they are aware of and your basically stuck reviewing logs of installed apps and having to compare that to osint feeds. I asked them specifically about malware targeting specific companies and they didn't have any examples of that.

On Wednesday morning, I attended “iOS Forensics and Encase” by Sean Morrissey. He recommends having a small charger for use inside a faraday bag to extend battery life and avoid the phone locking. He said its best practice to use a MacOSX workstaton and its native tools for analysis. He likes the PList Editor from the development tools, however you need the XCode3 instead of the newer XCode4, which removed some functionality. MacForensicLab is one of his favorite data carvers. He also likes the iPod Robot Plist editor for windows platforms. He said the forensic community has known about the GPS log data that Apple kept since iOS3, but kept quite on it to avoid notice. Since it went public, what use to provide up to a year of GPS data, is now going to be only 7 days and probably encrypted with iOS5. Another favorite is MSAB XRY, which does a complete physical dump. And also ZRT for doing automated screenshots. He thinks about 80% of what you need you can get from logical dumps and that usually is enough. Physical dumps are going to mostly contain fragmented data that you have to manually parse out. His favorite acquisition tools are FTS iXam and AccessData's MPE+. He has verified these tools by using HFS debug and tracking the incremental writing of CatlogID's. Encase has no native HFS+ support yet, however you can use a hexeditor to change the file header from HX. You can also change a raw dd file to a dmg extension and Mac's will mount it. Apple devices always use local time and he likes TimeLord for analysis. C4All.CA has nice tools (C4M and C4P). Binary Plist Finder & Parser are also good tools. The AT&T sim card only contains the last 10 calls and provider data and it is being phased out to use built in hardware in the future. The Encase Neutrino product doesn't parse out as much data from iOS as some of the other products like CellBrite.

To close this con out I attended "Encase and Flasher Box HEX Dumping Analysis" by John Thackray. He is British born but a Kiwi by choice in his own words. When it comes to cell phones there is no one single product that is going to get you everything you want. Extraction is the proper term as a true bit-by-bit copy isn't really possible. A flasher box should be a last resort as it can sometimes destroy evidence and even brick the phone. Test devices are essential. Check out the forum phoneforensics.com. All the firmware data you want is on the chipset and has nothing to do with the sim card. Flasher boxes have their code updated frequently so you need to update weekly. The process is very fast however the data you get back is highly fragmented. Locate all the maintenance codes as they are best way to get the phone to spit out make/model/version information and also perform other options. PM Records (Permanent Memory) are 0-999. Absolute Records are the memory offsets used to create binary dumps. He really showed us how the unlocking process worked on a Nokia which was cool to see. It was basically knowing where the IMEI and Security codes are kept. To find out a new structure when a new phone comes out the process is best down with KDiff3. By using a control phone and comparing hex dumps you can track when a text is sent or a call is made and see where the changes are put. The phone numbers themselves are usually stored as a reverse nibble. Timestamps are often different for sent messages and received messages. Hexaminer is a great tool for creating searchable 7-bit hex terms from known ascii text. Next we worked through a lab on a Samsung phone and manually recovered SMS text messages from raw hex. Someone from the crowd recommended a tool called Alibi(SMS Edit) for modifying sms text messages. Thackray said TigerText is another one that sends text messages and immediately deletes them. He also gave us a tool on CD called LiveExaminer.

So now that my first CEIC is in the books I have to say I was very impressed. The venue in Orlando is awesome and everything was well run, except for a minor lighting snafu. It wasn't overcrowded like Blackhat/Defcon, the food was good, and it was easy to talk to the presenters. I think the only drawback is that a good chunk of the crowd isn't very technical, think e-Discovery legal people and new to forensics cops. I would also say it’s better to register sooner rather than later to make sure to get into your favorite sessions. In the future I would like to see a more dedicated advanced technical lab-based track for people that have been doing digital forensics for awhile. Being able to work through some evidence is much more appealing than just a pure lecture. I will definitely come back to the Orlando location on the odd years as this was a great experience for me.

Best Presenter - Rob Lee - I think he was the most polished of the speakers and talked about things I wanted to hear. You can tell he is someone who like to know why something works and not just get the output of a tool.

Most Fun Presentation - Alissa Torres - She kept the crowd smiling the entire time and her enthusiasm was infectious.

Best Presentation - John Thackray - This shed light on area of phone forensics that most people don't have a lot of experience. It delivered exactly what I was looking to hear and probably taught me the most of all the presentations.

Best Vendor - NetWitness by a nose - Mandiant had a great party, but I liked the NetWitness booth the most as they took the time to really show me the product and give me the details I was looking for in regards to capabilities and deployment scenarios. Their tools show great promise for being able to process things in bulk off the wire.

1 comment:

  1. HI!
    nice Blog...Great to know...
    gps devices garmin is an excellent discovery of its kind!!

    ReplyDelete